{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 8.997899649941656,
  "eval_steps": 536,
  "global_step": 19278,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004667444574095683,
      "grad_norm": 4.125,
      "learning_rate": 2e-05,
      "loss": 2.581,
      "step": 1
    },
    {
      "epoch": 0.0004667444574095683,
      "eval_loss": 2.3898825645446777,
      "eval_runtime": 93.0875,
      "eval_samples_per_second": 19.38,
      "eval_steps_per_second": 2.428,
      "step": 1
    },
    {
      "epoch": 0.0009334889148191366,
      "grad_norm": 4.15625,
      "learning_rate": 4e-05,
      "loss": 2.5575,
      "step": 2
    },
    {
      "epoch": 0.0014002333722287048,
      "grad_norm": 4.5625,
      "learning_rate": 6e-05,
      "loss": 2.4575,
      "step": 3
    },
    {
      "epoch": 0.0018669778296382731,
      "grad_norm": 3.015625,
      "learning_rate": 8e-05,
      "loss": 2.1761,
      "step": 4
    },
    {
      "epoch": 0.002333722287047841,
      "grad_norm": 2.171875,
      "learning_rate": 0.0001,
      "loss": 2.0576,
      "step": 5
    },
    {
      "epoch": 0.0028004667444574095,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00012,
      "loss": 1.7047,
      "step": 6
    },
    {
      "epoch": 0.003267211201866978,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00014,
      "loss": 1.8031,
      "step": 7
    },
    {
      "epoch": 0.0037339556592765463,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00016,
      "loss": 1.5404,
      "step": 8
    },
    {
      "epoch": 0.004200700116686115,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018,
      "loss": 1.5804,
      "step": 9
    },
    {
      "epoch": 0.004667444574095682,
      "grad_norm": 0.75,
      "learning_rate": 0.0002,
      "loss": 1.4808,
      "step": 10
    },
    {
      "epoch": 0.005134189031505251,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019999999892344455,
      "loss": 1.4612,
      "step": 11
    },
    {
      "epoch": 0.005600933488914819,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019999999569377815,
      "loss": 1.4376,
      "step": 12
    },
    {
      "epoch": 0.006067677946324387,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019999999031100084,
      "loss": 1.4622,
      "step": 13
    },
    {
      "epoch": 0.006534422403733956,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019999998277511286,
      "loss": 1.3237,
      "step": 14
    },
    {
      "epoch": 0.007001166861143524,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019999997308611426,
      "loss": 1.4007,
      "step": 15
    },
    {
      "epoch": 0.007467911318553093,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001999999612440053,
      "loss": 1.349,
      "step": 16
    },
    {
      "epoch": 0.007934655775962661,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019999994724878624,
      "loss": 1.2734,
      "step": 17
    },
    {
      "epoch": 0.00840140023337223,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019999993110045734,
      "loss": 1.3,
      "step": 18
    },
    {
      "epoch": 0.008868144690781798,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019999991279901898,
      "loss": 1.387,
      "step": 19
    },
    {
      "epoch": 0.009334889148191364,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019999989234447153,
      "loss": 1.2229,
      "step": 20
    },
    {
      "epoch": 0.009801633605600933,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019999986973681547,
      "loss": 1.1438,
      "step": 21
    },
    {
      "epoch": 0.010268378063010501,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019999984497605125,
      "loss": 1.3672,
      "step": 22
    },
    {
      "epoch": 0.01073512252042007,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019999981806217943,
      "loss": 1.2266,
      "step": 23
    },
    {
      "epoch": 0.011201866977829638,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019999978899520055,
      "loss": 1.1961,
      "step": 24
    },
    {
      "epoch": 0.011668611435239206,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019999975777511527,
      "loss": 1.2069,
      "step": 25
    },
    {
      "epoch": 0.012135355892648775,
      "grad_norm": 0.625,
      "learning_rate": 0.00019999972440192426,
      "loss": 1.0946,
      "step": 26
    },
    {
      "epoch": 0.012602100350058343,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019999968887562822,
      "loss": 1.2584,
      "step": 27
    },
    {
      "epoch": 0.013068844807467912,
      "grad_norm": 0.625,
      "learning_rate": 0.00019999965119622796,
      "loss": 1.1032,
      "step": 28
    },
    {
      "epoch": 0.01353558926487748,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019999961136372423,
      "loss": 0.9899,
      "step": 29
    },
    {
      "epoch": 0.014002333722287048,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001999995693781179,
      "loss": 1.1128,
      "step": 30
    },
    {
      "epoch": 0.014469078179696617,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019999952523940993,
      "loss": 1.2303,
      "step": 31
    },
    {
      "epoch": 0.014935822637106185,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001999994789476012,
      "loss": 1.0485,
      "step": 32
    },
    {
      "epoch": 0.015402567094515752,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019999943050269273,
      "loss": 0.9912,
      "step": 33
    },
    {
      "epoch": 0.015869311551925322,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001999993799046856,
      "loss": 1.1316,
      "step": 34
    },
    {
      "epoch": 0.01633605600933489,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019999932715358086,
      "loss": 1.1705,
      "step": 35
    },
    {
      "epoch": 0.01680280046674446,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019999927224937966,
      "loss": 1.0543,
      "step": 36
    },
    {
      "epoch": 0.017269544924154025,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019999921519208317,
      "loss": 1.1632,
      "step": 37
    },
    {
      "epoch": 0.017736289381563596,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019999915598169264,
      "loss": 1.086,
      "step": 38
    },
    {
      "epoch": 0.018203033838973162,
      "grad_norm": 0.625,
      "learning_rate": 0.0001999990946182093,
      "loss": 1.2955,
      "step": 39
    },
    {
      "epoch": 0.01866977829638273,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019999903110163453,
      "loss": 1.1241,
      "step": 40
    },
    {
      "epoch": 0.0191365227537923,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019999896543196964,
      "loss": 1.1581,
      "step": 41
    },
    {
      "epoch": 0.019603267211201866,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001999988976092161,
      "loss": 1.0129,
      "step": 42
    },
    {
      "epoch": 0.020070011668611436,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019999882763337533,
      "loss": 1.258,
      "step": 43
    },
    {
      "epoch": 0.020536756126021002,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019999875550444886,
      "loss": 1.1702,
      "step": 44
    },
    {
      "epoch": 0.021003500583430573,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019999868122243825,
      "loss": 1.1404,
      "step": 45
    },
    {
      "epoch": 0.02147024504084014,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019999860478734505,
      "loss": 0.9923,
      "step": 46
    },
    {
      "epoch": 0.02193698949824971,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019999852619917094,
      "loss": 1.0718,
      "step": 47
    },
    {
      "epoch": 0.022403733955659276,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019999844545791763,
      "loss": 1.2165,
      "step": 48
    },
    {
      "epoch": 0.022870478413068846,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019999836256358685,
      "loss": 0.9337,
      "step": 49
    },
    {
      "epoch": 0.023337222870478413,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019999827751618032,
      "loss": 1.067,
      "step": 50
    },
    {
      "epoch": 0.023803967327887983,
      "grad_norm": 0.625,
      "learning_rate": 0.00019999819031569995,
      "loss": 1.0405,
      "step": 51
    },
    {
      "epoch": 0.02427071178529755,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001999981009621476,
      "loss": 1.1471,
      "step": 52
    },
    {
      "epoch": 0.024737456242707116,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001999980094555252,
      "loss": 1.0263,
      "step": 53
    },
    {
      "epoch": 0.025204200700116686,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019999791579583472,
      "loss": 1.0821,
      "step": 54
    },
    {
      "epoch": 0.025670945157526253,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019999781998307812,
      "loss": 0.9545,
      "step": 55
    },
    {
      "epoch": 0.026137689614935823,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019999772201725752,
      "loss": 1.1339,
      "step": 56
    },
    {
      "epoch": 0.02660443407234539,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019999762189837502,
      "loss": 1.1229,
      "step": 57
    },
    {
      "epoch": 0.02707117852975496,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019999751962643277,
      "loss": 1.1392,
      "step": 58
    },
    {
      "epoch": 0.027537922987164527,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019999741520143298,
      "loss": 1.0464,
      "step": 59
    },
    {
      "epoch": 0.028004667444574097,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001999973086233779,
      "loss": 0.9927,
      "step": 60
    },
    {
      "epoch": 0.028471411901983663,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001999971998922698,
      "loss": 1.0365,
      "step": 61
    },
    {
      "epoch": 0.028938156359393234,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019999708900811104,
      "loss": 1.0833,
      "step": 62
    },
    {
      "epoch": 0.0294049008168028,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019999697597090398,
      "loss": 1.0158,
      "step": 63
    },
    {
      "epoch": 0.02987164527421237,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019999686078065113,
      "loss": 1.1899,
      "step": 64
    },
    {
      "epoch": 0.030338389731621937,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001999967434373549,
      "loss": 1.0692,
      "step": 65
    },
    {
      "epoch": 0.030805134189031504,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001999966239410178,
      "loss": 1.0199,
      "step": 66
    },
    {
      "epoch": 0.03127187864644107,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001999965022916425,
      "loss": 0.9763,
      "step": 67
    },
    {
      "epoch": 0.031738623103850644,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001999963784892315,
      "loss": 1.002,
      "step": 68
    },
    {
      "epoch": 0.03220536756126021,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019999625253378756,
      "loss": 1.1227,
      "step": 69
    },
    {
      "epoch": 0.03267211201866978,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019999612442531332,
      "loss": 1.0688,
      "step": 70
    },
    {
      "epoch": 0.033138856476079344,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019999599416381162,
      "loss": 1.0535,
      "step": 71
    },
    {
      "epoch": 0.03360560093348892,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001999958617492852,
      "loss": 1.2168,
      "step": 72
    },
    {
      "epoch": 0.034072345390898484,
      "grad_norm": 0.625,
      "learning_rate": 0.0001999957271817369,
      "loss": 1.0747,
      "step": 73
    },
    {
      "epoch": 0.03453908984830805,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019999559046116966,
      "loss": 1.0446,
      "step": 74
    },
    {
      "epoch": 0.03500583430571762,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019999545158758643,
      "loss": 1.0294,
      "step": 75
    },
    {
      "epoch": 0.03547257876312719,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019999531056099019,
      "loss": 1.0653,
      "step": 76
    },
    {
      "epoch": 0.03593932322053676,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019999516738138394,
      "loss": 0.9742,
      "step": 77
    },
    {
      "epoch": 0.036406067677946324,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019999502204877078,
      "loss": 0.9658,
      "step": 78
    },
    {
      "epoch": 0.03687281213535589,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019999487456315385,
      "loss": 1.1283,
      "step": 79
    },
    {
      "epoch": 0.03733955659276546,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019999472492453636,
      "loss": 1.0825,
      "step": 80
    },
    {
      "epoch": 0.03780630105017503,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001999945731329215,
      "loss": 1.1282,
      "step": 81
    },
    {
      "epoch": 0.0382730455075846,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001999944191883125,
      "loss": 1.0473,
      "step": 82
    },
    {
      "epoch": 0.038739789964994165,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019999426309071273,
      "loss": 1.0766,
      "step": 83
    },
    {
      "epoch": 0.03920653442240373,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019999410484012554,
      "loss": 1.0673,
      "step": 84
    },
    {
      "epoch": 0.039673278879813305,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019999394443655428,
      "loss": 0.9415,
      "step": 85
    },
    {
      "epoch": 0.04014002333722287,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019999378188000248,
      "loss": 1.0481,
      "step": 86
    },
    {
      "epoch": 0.04060676779463244,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019999361717047362,
      "loss": 0.9316,
      "step": 87
    },
    {
      "epoch": 0.041073512252042005,
      "grad_norm": 0.625,
      "learning_rate": 0.00019999345030797123,
      "loss": 1.0445,
      "step": 88
    },
    {
      "epoch": 0.04154025670945158,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001999932812924989,
      "loss": 0.9591,
      "step": 89
    },
    {
      "epoch": 0.042007001166861145,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019999311012406027,
      "loss": 1.0437,
      "step": 90
    },
    {
      "epoch": 0.04247374562427071,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019999293680265907,
      "loss": 1.0453,
      "step": 91
    },
    {
      "epoch": 0.04294049008168028,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019999276132829898,
      "loss": 1.1028,
      "step": 92
    },
    {
      "epoch": 0.043407234539089845,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001999925837009838,
      "loss": 0.9184,
      "step": 93
    },
    {
      "epoch": 0.04387397899649942,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019999240392071731,
      "loss": 1.1142,
      "step": 94
    },
    {
      "epoch": 0.044340723453908985,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019999222198750349,
      "loss": 1.0435,
      "step": 95
    },
    {
      "epoch": 0.04480746791131855,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019999203790134614,
      "loss": 0.9208,
      "step": 96
    },
    {
      "epoch": 0.04527421236872812,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001999918516622493,
      "loss": 1.0377,
      "step": 97
    },
    {
      "epoch": 0.04574095682613769,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001999916632702169,
      "loss": 1.0346,
      "step": 98
    },
    {
      "epoch": 0.04620770128354726,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001999914727252531,
      "loss": 0.8648,
      "step": 99
    },
    {
      "epoch": 0.046674445740956826,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019999128002736192,
      "loss": 0.9425,
      "step": 100
    },
    {
      "epoch": 0.04714119019836639,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019999108517654754,
      "loss": 0.9718,
      "step": 101
    },
    {
      "epoch": 0.047607934655775966,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019999088817281417,
      "loss": 0.9913,
      "step": 102
    },
    {
      "epoch": 0.04807467911318553,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019999068901616602,
      "loss": 1.0457,
      "step": 103
    },
    {
      "epoch": 0.0485414235705951,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001999904877066074,
      "loss": 0.9852,
      "step": 104
    },
    {
      "epoch": 0.049008168028004666,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019999028424414267,
      "loss": 1.0041,
      "step": 105
    },
    {
      "epoch": 0.04947491248541423,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019999007862877614,
      "loss": 1.0784,
      "step": 106
    },
    {
      "epoch": 0.049941656942823806,
      "grad_norm": 0.5,
      "learning_rate": 0.00019998987086051226,
      "loss": 1.0024,
      "step": 107
    },
    {
      "epoch": 0.05040840140023337,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019998966093935557,
      "loss": 0.9716,
      "step": 108
    },
    {
      "epoch": 0.05087514585764294,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001999894488653105,
      "loss": 1.0776,
      "step": 109
    },
    {
      "epoch": 0.051341890315052506,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001999892346383817,
      "loss": 0.9575,
      "step": 110
    },
    {
      "epoch": 0.05180863477246208,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001999890182585737,
      "loss": 0.9859,
      "step": 111
    },
    {
      "epoch": 0.052275379229871646,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019998879972589118,
      "loss": 1.065,
      "step": 112
    },
    {
      "epoch": 0.05274212368728121,
      "grad_norm": 0.478515625,
      "learning_rate": 0.0001999885790403389,
      "loss": 0.9501,
      "step": 113
    },
    {
      "epoch": 0.05320886814469078,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019998835620192157,
      "loss": 1.0011,
      "step": 114
    },
    {
      "epoch": 0.05367561260210035,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019998813121064398,
      "loss": 1.0163,
      "step": 115
    },
    {
      "epoch": 0.05414235705950992,
      "grad_norm": 0.515625,
      "learning_rate": 0.000199987904066511,
      "loss": 0.9972,
      "step": 116
    },
    {
      "epoch": 0.05460910151691949,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019998767476952747,
      "loss": 0.9394,
      "step": 117
    },
    {
      "epoch": 0.05507584597432905,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001999874433196984,
      "loss": 0.9621,
      "step": 118
    },
    {
      "epoch": 0.05554259043173862,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019998720971702873,
      "loss": 0.9837,
      "step": 119
    },
    {
      "epoch": 0.056009334889148193,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019998697396152347,
      "loss": 1.1347,
      "step": 120
    },
    {
      "epoch": 0.05647607934655776,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019998673605318778,
      "loss": 0.902,
      "step": 121
    },
    {
      "epoch": 0.05694282380396733,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001999864959920267,
      "loss": 1.0605,
      "step": 122
    },
    {
      "epoch": 0.05740956826137689,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001999862537780454,
      "loss": 1.0244,
      "step": 123
    },
    {
      "epoch": 0.05787631271878647,
      "grad_norm": 0.625,
      "learning_rate": 0.00019998600941124916,
      "loss": 1.2095,
      "step": 124
    },
    {
      "epoch": 0.058343057176196034,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019998576289164316,
      "loss": 0.9162,
      "step": 125
    },
    {
      "epoch": 0.0588098016336056,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019998551421923278,
      "loss": 1.0131,
      "step": 126
    },
    {
      "epoch": 0.05927654609101517,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019998526339402332,
      "loss": 0.9307,
      "step": 127
    },
    {
      "epoch": 0.05974329054842474,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019998501041602022,
      "loss": 0.8748,
      "step": 128
    },
    {
      "epoch": 0.06021003500583431,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019998475528522892,
      "loss": 1.088,
      "step": 129
    },
    {
      "epoch": 0.060676779463243874,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001999844980016549,
      "loss": 1.1424,
      "step": 130
    },
    {
      "epoch": 0.06114352392065344,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001999842385653037,
      "loss": 0.9336,
      "step": 131
    },
    {
      "epoch": 0.06161026837806301,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019998397697618088,
      "loss": 1.0417,
      "step": 132
    },
    {
      "epoch": 0.06207701283547258,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019998371323429212,
      "loss": 0.9003,
      "step": 133
    },
    {
      "epoch": 0.06254375729288214,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001999834473396431,
      "loss": 1.0819,
      "step": 134
    },
    {
      "epoch": 0.06301050175029171,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001999831792922395,
      "loss": 0.9295,
      "step": 135
    },
    {
      "epoch": 0.06347724620770129,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001999829090920871,
      "loss": 0.9579,
      "step": 136
    },
    {
      "epoch": 0.06394399066511085,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019998263673919175,
      "loss": 0.9949,
      "step": 137
    },
    {
      "epoch": 0.06441073512252042,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019998236223355933,
      "loss": 0.9855,
      "step": 138
    },
    {
      "epoch": 0.06487747957993,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019998208557519572,
      "loss": 1.0609,
      "step": 139
    },
    {
      "epoch": 0.06534422403733955,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019998180676410684,
      "loss": 1.0431,
      "step": 140
    },
    {
      "epoch": 0.06581096849474913,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019998152580029874,
      "loss": 0.9369,
      "step": 141
    },
    {
      "epoch": 0.06627771295215869,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019998124268377746,
      "loss": 0.8901,
      "step": 142
    },
    {
      "epoch": 0.06674445740956826,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019998095741454908,
      "loss": 0.9857,
      "step": 143
    },
    {
      "epoch": 0.06721120186697783,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019998066999261975,
      "loss": 0.924,
      "step": 144
    },
    {
      "epoch": 0.0676779463243874,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019998038041799568,
      "loss": 1.0101,
      "step": 145
    },
    {
      "epoch": 0.06814469078179697,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019998008869068313,
      "loss": 1.0248,
      "step": 146
    },
    {
      "epoch": 0.06861143523920653,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001999797948106883,
      "loss": 0.9731,
      "step": 147
    },
    {
      "epoch": 0.0690781796966161,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019997949877801756,
      "loss": 0.8834,
      "step": 148
    },
    {
      "epoch": 0.06954492415402568,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019997920059267727,
      "loss": 0.9599,
      "step": 149
    },
    {
      "epoch": 0.07001166861143523,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019997890025467388,
      "loss": 1.0223,
      "step": 150
    },
    {
      "epoch": 0.07047841306884481,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019997859776401384,
      "loss": 0.92,
      "step": 151
    },
    {
      "epoch": 0.07094515752625438,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019997829312070367,
      "loss": 0.9074,
      "step": 152
    },
    {
      "epoch": 0.07141190198366394,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001999779863247499,
      "loss": 1.0465,
      "step": 153
    },
    {
      "epoch": 0.07187864644107352,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019997767737615915,
      "loss": 0.8792,
      "step": 154
    },
    {
      "epoch": 0.07234539089848308,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019997736627493811,
      "loss": 0.9099,
      "step": 155
    },
    {
      "epoch": 0.07281213535589265,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019997705302109343,
      "loss": 1.0126,
      "step": 156
    },
    {
      "epoch": 0.07327887981330222,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019997673761463185,
      "loss": 0.9207,
      "step": 157
    },
    {
      "epoch": 0.07374562427071178,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019997642005556022,
      "loss": 0.9632,
      "step": 158
    },
    {
      "epoch": 0.07421236872812136,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001999761003438853,
      "loss": 0.8723,
      "step": 159
    },
    {
      "epoch": 0.07467911318553092,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019997577847961407,
      "loss": 0.8775,
      "step": 160
    },
    {
      "epoch": 0.07514585764294049,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019997545446275337,
      "loss": 0.8926,
      "step": 161
    },
    {
      "epoch": 0.07561260210035006,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019997512829331022,
      "loss": 1.0073,
      "step": 162
    },
    {
      "epoch": 0.07607934655775962,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019997479997129164,
      "loss": 0.9234,
      "step": 163
    },
    {
      "epoch": 0.0765460910151692,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019997446949670468,
      "loss": 1.0193,
      "step": 164
    },
    {
      "epoch": 0.07701283547257877,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00019997413686955647,
      "loss": 0.8789,
      "step": 165
    },
    {
      "epoch": 0.07747957992998833,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019997380208985416,
      "loss": 0.9006,
      "step": 166
    },
    {
      "epoch": 0.0779463243873979,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019997346515760497,
      "loss": 1.0905,
      "step": 167
    },
    {
      "epoch": 0.07841306884480746,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019997312607281619,
      "loss": 1.023,
      "step": 168
    },
    {
      "epoch": 0.07887981330221704,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019997278483549504,
      "loss": 1.0344,
      "step": 169
    },
    {
      "epoch": 0.07934655775962661,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00019997244144564892,
      "loss": 0.9687,
      "step": 170
    },
    {
      "epoch": 0.07981330221703617,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019997209590328523,
      "loss": 0.9775,
      "step": 171
    },
    {
      "epoch": 0.08028004667444574,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001999717482084114,
      "loss": 0.9761,
      "step": 172
    },
    {
      "epoch": 0.0807467911318553,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019997139836103487,
      "loss": 0.9741,
      "step": 173
    },
    {
      "epoch": 0.08121353558926488,
      "grad_norm": 0.5,
      "learning_rate": 0.00019997104636116324,
      "loss": 0.9491,
      "step": 174
    },
    {
      "epoch": 0.08168028004667445,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00019997069220880405,
      "loss": 0.9257,
      "step": 175
    },
    {
      "epoch": 0.08214702450408401,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019997033590396493,
      "loss": 0.9054,
      "step": 176
    },
    {
      "epoch": 0.08261376896149358,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001999699774466536,
      "loss": 1.0,
      "step": 177
    },
    {
      "epoch": 0.08308051341890316,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019996961683687773,
      "loss": 1.0593,
      "step": 178
    },
    {
      "epoch": 0.08354725787631272,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019996925407464507,
      "loss": 0.949,
      "step": 179
    },
    {
      "epoch": 0.08401400233372229,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019996888915996345,
      "loss": 0.9735,
      "step": 180
    },
    {
      "epoch": 0.08448074679113185,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019996852209284074,
      "loss": 0.9375,
      "step": 181
    },
    {
      "epoch": 0.08494749124854142,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019996815287328483,
      "loss": 0.9325,
      "step": 182
    },
    {
      "epoch": 0.085414235705951,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019996778150130367,
      "loss": 0.8899,
      "step": 183
    },
    {
      "epoch": 0.08588098016336056,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019996740797690525,
      "loss": 1.0411,
      "step": 184
    },
    {
      "epoch": 0.08634772462077013,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019996703230009763,
      "loss": 0.962,
      "step": 185
    },
    {
      "epoch": 0.08681446907817969,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001999666544708889,
      "loss": 0.8849,
      "step": 186
    },
    {
      "epoch": 0.08728121353558926,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019996627448928716,
      "loss": 0.9275,
      "step": 187
    },
    {
      "epoch": 0.08774795799299884,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019996589235530063,
      "loss": 1.0102,
      "step": 188
    },
    {
      "epoch": 0.0882147024504084,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019996550806893754,
      "loss": 0.971,
      "step": 189
    },
    {
      "epoch": 0.08868144690781797,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0001999651216302061,
      "loss": 0.8787,
      "step": 190
    },
    {
      "epoch": 0.08914819136522754,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019996473303911472,
      "loss": 1.032,
      "step": 191
    },
    {
      "epoch": 0.0896149358226371,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019996434229567172,
      "loss": 0.9348,
      "step": 192
    },
    {
      "epoch": 0.09008168028004668,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001999639493998855,
      "loss": 0.9619,
      "step": 193
    },
    {
      "epoch": 0.09054842473745624,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019996355435176455,
      "loss": 0.8818,
      "step": 194
    },
    {
      "epoch": 0.09101516919486581,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019996315715131737,
      "loss": 0.9511,
      "step": 195
    },
    {
      "epoch": 0.09148191365227538,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019996275779855249,
      "loss": 0.9715,
      "step": 196
    },
    {
      "epoch": 0.09194865810968494,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019996235629347853,
      "loss": 0.9739,
      "step": 197
    },
    {
      "epoch": 0.09241540256709452,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001999619526361041,
      "loss": 0.8739,
      "step": 198
    },
    {
      "epoch": 0.09288214702450408,
      "grad_norm": 0.5,
      "learning_rate": 0.00019996154682643796,
      "loss": 0.9189,
      "step": 199
    },
    {
      "epoch": 0.09334889148191365,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019996113886448883,
      "loss": 0.9125,
      "step": 200
    },
    {
      "epoch": 0.09381563593932322,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019996072875026543,
      "loss": 0.8406,
      "step": 201
    },
    {
      "epoch": 0.09428238039673278,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019996031648377663,
      "loss": 0.8843,
      "step": 202
    },
    {
      "epoch": 0.09474912485414236,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019995990206503133,
      "loss": 1.0226,
      "step": 203
    },
    {
      "epoch": 0.09521586931155193,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001999594854940384,
      "loss": 1.0359,
      "step": 204
    },
    {
      "epoch": 0.09568261376896149,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019995906677080686,
      "loss": 0.974,
      "step": 205
    },
    {
      "epoch": 0.09614935822637106,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001999586458953457,
      "loss": 0.8993,
      "step": 206
    },
    {
      "epoch": 0.09661610268378062,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019995822286766402,
      "loss": 0.9165,
      "step": 207
    },
    {
      "epoch": 0.0970828471411902,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019995779768777085,
      "loss": 0.8766,
      "step": 208
    },
    {
      "epoch": 0.09754959159859977,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019995737035567542,
      "loss": 0.9438,
      "step": 209
    },
    {
      "epoch": 0.09801633605600933,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019995694087138687,
      "loss": 0.9538,
      "step": 210
    },
    {
      "epoch": 0.0984830805134189,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00019995650923491454,
      "loss": 0.8677,
      "step": 211
    },
    {
      "epoch": 0.09894982497082846,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001999560754462676,
      "loss": 0.9725,
      "step": 212
    },
    {
      "epoch": 0.09941656942823804,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001999556395054555,
      "loss": 0.9733,
      "step": 213
    },
    {
      "epoch": 0.09988331388564761,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019995520141248752,
      "loss": 0.9579,
      "step": 214
    },
    {
      "epoch": 0.10035005834305717,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001999547611673732,
      "loss": 0.9401,
      "step": 215
    },
    {
      "epoch": 0.10081680280046675,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019995431877012197,
      "loss": 1.0415,
      "step": 216
    },
    {
      "epoch": 0.10128354725787632,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019995387422074332,
      "loss": 1.0081,
      "step": 217
    },
    {
      "epoch": 0.10175029171528588,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001999534275192469,
      "loss": 0.8797,
      "step": 218
    },
    {
      "epoch": 0.10221703617269545,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019995297866564227,
      "loss": 0.9835,
      "step": 219
    },
    {
      "epoch": 0.10268378063010501,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019995252765993908,
      "loss": 0.918,
      "step": 220
    },
    {
      "epoch": 0.10315052508751459,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001999520745021471,
      "loss": 0.824,
      "step": 221
    },
    {
      "epoch": 0.10361726954492416,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019995161919227606,
      "loss": 0.8907,
      "step": 222
    },
    {
      "epoch": 0.10408401400233372,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019995116173033575,
      "loss": 0.9877,
      "step": 223
    },
    {
      "epoch": 0.10455075845974329,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019995070211633603,
      "loss": 0.9514,
      "step": 224
    },
    {
      "epoch": 0.10501750291715285,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00019995024035028677,
      "loss": 0.7733,
      "step": 225
    },
    {
      "epoch": 0.10548424737456243,
      "grad_norm": 0.625,
      "learning_rate": 0.000199949776432198,
      "loss": 0.9043,
      "step": 226
    },
    {
      "epoch": 0.105950991831972,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001999493103620796,
      "loss": 0.7945,
      "step": 227
    },
    {
      "epoch": 0.10641773628938156,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019994884213994166,
      "loss": 0.9175,
      "step": 228
    },
    {
      "epoch": 0.10688448074679113,
      "grad_norm": 0.5,
      "learning_rate": 0.00019994837176579426,
      "loss": 0.8977,
      "step": 229
    },
    {
      "epoch": 0.1073512252042007,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019994789923964752,
      "loss": 0.9744,
      "step": 230
    },
    {
      "epoch": 0.10781796966161027,
      "grad_norm": 0.5,
      "learning_rate": 0.00019994742456151164,
      "loss": 0.9331,
      "step": 231
    },
    {
      "epoch": 0.10828471411901984,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001999469477313968,
      "loss": 1.0236,
      "step": 232
    },
    {
      "epoch": 0.1087514585764294,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019994646874931324,
      "loss": 0.9392,
      "step": 233
    },
    {
      "epoch": 0.10921820303383897,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019994598761527137,
      "loss": 1.0308,
      "step": 234
    },
    {
      "epoch": 0.10968494749124855,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019994550432928145,
      "loss": 0.9339,
      "step": 235
    },
    {
      "epoch": 0.1101516919486581,
      "grad_norm": 0.5078125,
      "learning_rate": 0.000199945018891354,
      "loss": 0.9856,
      "step": 236
    },
    {
      "epoch": 0.11061843640606768,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019994453130149932,
      "loss": 0.754,
      "step": 237
    },
    {
      "epoch": 0.11108518086347724,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019994404155972804,
      "loss": 0.8992,
      "step": 238
    },
    {
      "epoch": 0.11155192532088681,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019994354966605064,
      "loss": 1.0772,
      "step": 239
    },
    {
      "epoch": 0.11201866977829639,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001999430556204777,
      "loss": 0.9704,
      "step": 240
    },
    {
      "epoch": 0.11248541423570595,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019994255942301994,
      "loss": 1.0512,
      "step": 241
    },
    {
      "epoch": 0.11295215869311552,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019994206107368795,
      "loss": 0.7931,
      "step": 242
    },
    {
      "epoch": 0.1134189031505251,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019994156057249253,
      "loss": 0.7867,
      "step": 243
    },
    {
      "epoch": 0.11388564760793465,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001999410579194444,
      "loss": 0.9791,
      "step": 244
    },
    {
      "epoch": 0.11435239206534423,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001999405531145544,
      "loss": 0.79,
      "step": 245
    },
    {
      "epoch": 0.11481913652275379,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019994004615783343,
      "loss": 0.8584,
      "step": 246
    },
    {
      "epoch": 0.11528588098016336,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019993953704929235,
      "loss": 0.8858,
      "step": 247
    },
    {
      "epoch": 0.11575262543757293,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019993902578894216,
      "loss": 0.9514,
      "step": 248
    },
    {
      "epoch": 0.1162193698949825,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001999385123767939,
      "loss": 0.9232,
      "step": 249
    },
    {
      "epoch": 0.11668611435239207,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001999379968128585,
      "loss": 1.0322,
      "step": 250
    },
    {
      "epoch": 0.11715285880980163,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019993747909714723,
      "loss": 0.9013,
      "step": 251
    },
    {
      "epoch": 0.1176196032672112,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001999369592296711,
      "loss": 0.8958,
      "step": 252
    },
    {
      "epoch": 0.11808634772462077,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019993643721044139,
      "loss": 0.9049,
      "step": 253
    },
    {
      "epoch": 0.11855309218203033,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001999359130394693,
      "loss": 0.8687,
      "step": 254
    },
    {
      "epoch": 0.11901983663943991,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001999353867167661,
      "loss": 0.9216,
      "step": 255
    },
    {
      "epoch": 0.11948658109684948,
      "grad_norm": 0.5,
      "learning_rate": 0.00019993485824234316,
      "loss": 0.9323,
      "step": 256
    },
    {
      "epoch": 0.11995332555425904,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019993432761621186,
      "loss": 0.9418,
      "step": 257
    },
    {
      "epoch": 0.12042007001166861,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019993379483838357,
      "loss": 0.8182,
      "step": 258
    },
    {
      "epoch": 0.12088681446907817,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019993325990886983,
      "loss": 0.8715,
      "step": 259
    },
    {
      "epoch": 0.12135355892648775,
      "grad_norm": 0.453125,
      "learning_rate": 0.00019993272282768212,
      "loss": 0.7782,
      "step": 260
    },
    {
      "epoch": 0.12182030338389732,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019993218359483196,
      "loss": 0.8634,
      "step": 261
    },
    {
      "epoch": 0.12228704784130688,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019993164221033106,
      "loss": 0.7912,
      "step": 262
    },
    {
      "epoch": 0.12275379229871645,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00019993109867419103,
      "loss": 0.8787,
      "step": 263
    },
    {
      "epoch": 0.12322053675612601,
      "grad_norm": 0.5,
      "learning_rate": 0.00019993055298642357,
      "loss": 0.8288,
      "step": 264
    },
    {
      "epoch": 0.12368728121353559,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019993000514704042,
      "loss": 0.8709,
      "step": 265
    },
    {
      "epoch": 0.12415402567094516,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019992945515605337,
      "loss": 0.8269,
      "step": 266
    },
    {
      "epoch": 0.12462077012835472,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019992890301347429,
      "loss": 0.9573,
      "step": 267
    },
    {
      "epoch": 0.12508751458576428,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019992834871931505,
      "loss": 0.9278,
      "step": 268
    },
    {
      "epoch": 0.12555425904317385,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001999277922735876,
      "loss": 0.8542,
      "step": 269
    },
    {
      "epoch": 0.12602100350058343,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019992723367630392,
      "loss": 0.8666,
      "step": 270
    },
    {
      "epoch": 0.126487747957993,
      "grad_norm": 0.486328125,
      "learning_rate": 0.000199926672927476,
      "loss": 0.8194,
      "step": 271
    },
    {
      "epoch": 0.12695449241540258,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019992611002711595,
      "loss": 0.925,
      "step": 272
    },
    {
      "epoch": 0.12742123687281215,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001999255449752359,
      "loss": 0.8912,
      "step": 273
    },
    {
      "epoch": 0.1278879813302217,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019992497777184798,
      "loss": 0.8968,
      "step": 274
    },
    {
      "epoch": 0.12835472578763127,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019992440841696442,
      "loss": 0.9033,
      "step": 275
    },
    {
      "epoch": 0.12882147024504084,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019992383691059748,
      "loss": 1.0323,
      "step": 276
    },
    {
      "epoch": 0.12928821470245042,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001999232632527595,
      "loss": 0.8634,
      "step": 277
    },
    {
      "epoch": 0.12975495915986,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019992268744346275,
      "loss": 0.9324,
      "step": 278
    },
    {
      "epoch": 0.13022170361726954,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019992210948271968,
      "loss": 0.9209,
      "step": 279
    },
    {
      "epoch": 0.1306884480746791,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019992152937054272,
      "loss": 0.9396,
      "step": 280
    },
    {
      "epoch": 0.13115519253208868,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001999209471069444,
      "loss": 0.9412,
      "step": 281
    },
    {
      "epoch": 0.13162193698949826,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001999203626919372,
      "loss": 0.8208,
      "step": 282
    },
    {
      "epoch": 0.13208868144690783,
      "grad_norm": 0.5,
      "learning_rate": 0.00019991977612553372,
      "loss": 0.8783,
      "step": 283
    },
    {
      "epoch": 0.13255542590431738,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001999191874077466,
      "loss": 0.8393,
      "step": 284
    },
    {
      "epoch": 0.13302217036172695,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019991859653858854,
      "loss": 1.0623,
      "step": 285
    },
    {
      "epoch": 0.13348891481913652,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019991800351807222,
      "loss": 1.002,
      "step": 286
    },
    {
      "epoch": 0.1339556592765461,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019991740834621042,
      "loss": 0.8858,
      "step": 287
    },
    {
      "epoch": 0.13442240373395567,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019991681102301596,
      "loss": 0.8736,
      "step": 288
    },
    {
      "epoch": 0.13488914819136522,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019991621154850169,
      "loss": 0.8057,
      "step": 289
    },
    {
      "epoch": 0.1353558926487748,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019991560992268053,
      "loss": 0.9185,
      "step": 290
    },
    {
      "epoch": 0.13582263710618436,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019991500614556545,
      "loss": 0.8324,
      "step": 291
    },
    {
      "epoch": 0.13628938156359394,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001999144002171694,
      "loss": 0.9509,
      "step": 292
    },
    {
      "epoch": 0.1367561260210035,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019991379213750546,
      "loss": 0.7803,
      "step": 293
    },
    {
      "epoch": 0.13722287047841306,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019991318190658674,
      "loss": 0.9608,
      "step": 294
    },
    {
      "epoch": 0.13768961493582263,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019991256952442635,
      "loss": 0.8689,
      "step": 295
    },
    {
      "epoch": 0.1381563593932322,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001999119549910375,
      "loss": 0.8479,
      "step": 296
    },
    {
      "epoch": 0.13862310385064178,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019991133830643337,
      "loss": 0.8892,
      "step": 297
    },
    {
      "epoch": 0.13908984830805135,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019991071947062728,
      "loss": 0.8905,
      "step": 298
    },
    {
      "epoch": 0.13955659276546092,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019991009848363257,
      "loss": 0.9355,
      "step": 299
    },
    {
      "epoch": 0.14002333722287047,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019990947534546258,
      "loss": 0.871,
      "step": 300
    },
    {
      "epoch": 0.14049008168028004,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001999088500561307,
      "loss": 0.8625,
      "step": 301
    },
    {
      "epoch": 0.14095682613768962,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019990822261565047,
      "loss": 0.9275,
      "step": 302
    },
    {
      "epoch": 0.1414235705950992,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019990759302403535,
      "loss": 0.8776,
      "step": 303
    },
    {
      "epoch": 0.14189031505250876,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019990696128129888,
      "loss": 0.843,
      "step": 304
    },
    {
      "epoch": 0.1423570595099183,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001999063273874547,
      "loss": 1.0007,
      "step": 305
    },
    {
      "epoch": 0.14282380396732788,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019990569134251644,
      "loss": 0.9292,
      "step": 306
    },
    {
      "epoch": 0.14329054842473746,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019990505314649781,
      "loss": 0.8729,
      "step": 307
    },
    {
      "epoch": 0.14375729288214703,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019990441279941253,
      "loss": 0.9855,
      "step": 308
    },
    {
      "epoch": 0.1442240373395566,
      "grad_norm": 0.5,
      "learning_rate": 0.0001999037703012744,
      "loss": 0.9412,
      "step": 309
    },
    {
      "epoch": 0.14469078179696615,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019990312565209726,
      "loss": 1.0191,
      "step": 310
    },
    {
      "epoch": 0.14515752625437572,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00019990247885189494,
      "loss": 0.7305,
      "step": 311
    },
    {
      "epoch": 0.1456242707117853,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019990182990068146,
      "loss": 0.8149,
      "step": 312
    },
    {
      "epoch": 0.14609101516919487,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001999011787984707,
      "loss": 0.885,
      "step": 313
    },
    {
      "epoch": 0.14655775962660444,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019990052554527674,
      "loss": 0.9428,
      "step": 314
    },
    {
      "epoch": 0.147024504084014,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019989987014111359,
      "loss": 0.9553,
      "step": 315
    },
    {
      "epoch": 0.14749124854142356,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019989921258599546,
      "loss": 0.9251,
      "step": 316
    },
    {
      "epoch": 0.14795799299883314,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019989855287993636,
      "loss": 0.8842,
      "step": 317
    },
    {
      "epoch": 0.1484247374562427,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001998978910229506,
      "loss": 0.8508,
      "step": 318
    },
    {
      "epoch": 0.14889148191365228,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019989722701505244,
      "loss": 0.8295,
      "step": 319
    },
    {
      "epoch": 0.14935822637106183,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00019989656085625613,
      "loss": 0.8384,
      "step": 320
    },
    {
      "epoch": 0.1498249708284714,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00019989589254657605,
      "loss": 0.7348,
      "step": 321
    },
    {
      "epoch": 0.15029171528588098,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019989522208602653,
      "loss": 0.9145,
      "step": 322
    },
    {
      "epoch": 0.15075845974329055,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001998945494746221,
      "loss": 0.9695,
      "step": 323
    },
    {
      "epoch": 0.15122520420070013,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019989387471237713,
      "loss": 0.9439,
      "step": 324
    },
    {
      "epoch": 0.1516919486581097,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001998931977993062,
      "loss": 0.9206,
      "step": 325
    },
    {
      "epoch": 0.15215869311551924,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001998925187354239,
      "loss": 0.813,
      "step": 326
    },
    {
      "epoch": 0.15262543757292882,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019989183752074484,
      "loss": 1.0323,
      "step": 327
    },
    {
      "epoch": 0.1530921820303384,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001998911541552837,
      "loss": 0.7331,
      "step": 328
    },
    {
      "epoch": 0.15355892648774797,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019989046863905517,
      "loss": 0.8732,
      "step": 329
    },
    {
      "epoch": 0.15402567094515754,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019988978097207403,
      "loss": 0.9539,
      "step": 330
    },
    {
      "epoch": 0.15449241540256708,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019988909115435508,
      "loss": 0.9257,
      "step": 331
    },
    {
      "epoch": 0.15495915985997666,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019988839918591315,
      "loss": 0.7343,
      "step": 332
    },
    {
      "epoch": 0.15542590431738623,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019988770506676316,
      "loss": 0.8209,
      "step": 333
    },
    {
      "epoch": 0.1558926487747958,
      "grad_norm": 0.5,
      "learning_rate": 0.00019988700879692005,
      "loss": 0.8491,
      "step": 334
    },
    {
      "epoch": 0.15635939323220538,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00019988631037639882,
      "loss": 0.8388,
      "step": 335
    },
    {
      "epoch": 0.15682613768961493,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001998856098052145,
      "loss": 0.837,
      "step": 336
    },
    {
      "epoch": 0.1572928821470245,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001998849070833822,
      "loss": 0.8748,
      "step": 337
    },
    {
      "epoch": 0.15775962660443407,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019988420221091698,
      "loss": 0.8861,
      "step": 338
    },
    {
      "epoch": 0.15822637106184365,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019988349518783408,
      "loss": 0.8853,
      "step": 339
    },
    {
      "epoch": 0.15869311551925322,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019988278601414873,
      "loss": 0.8224,
      "step": 340
    },
    {
      "epoch": 0.15915985997666277,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019988207468987613,
      "loss": 0.7861,
      "step": 341
    },
    {
      "epoch": 0.15962660443407234,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001998813612150317,
      "loss": 0.8256,
      "step": 342
    },
    {
      "epoch": 0.1600933488914819,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001998806455896307,
      "loss": 0.7111,
      "step": 343
    },
    {
      "epoch": 0.16056009334889149,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019987992781368862,
      "loss": 1.1183,
      "step": 344
    },
    {
      "epoch": 0.16102683780630106,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019987920788722084,
      "loss": 0.8174,
      "step": 345
    },
    {
      "epoch": 0.1614935822637106,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019987848581024288,
      "loss": 0.916,
      "step": 346
    },
    {
      "epoch": 0.16196032672112018,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019987776158277032,
      "loss": 0.7234,
      "step": 347
    },
    {
      "epoch": 0.16242707117852975,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019987703520481874,
      "loss": 0.9876,
      "step": 348
    },
    {
      "epoch": 0.16289381563593933,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001998763066764038,
      "loss": 1.0152,
      "step": 349
    },
    {
      "epoch": 0.1633605600933489,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019987557599754113,
      "loss": 0.9657,
      "step": 350
    },
    {
      "epoch": 0.16382730455075847,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019987484316824652,
      "loss": 0.8207,
      "step": 351
    },
    {
      "epoch": 0.16429404900816802,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019987410818853572,
      "loss": 0.9195,
      "step": 352
    },
    {
      "epoch": 0.1647607934655776,
      "grad_norm": 0.5,
      "learning_rate": 0.00019987337105842453,
      "loss": 0.9246,
      "step": 353
    },
    {
      "epoch": 0.16522753792298717,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019987263177792888,
      "loss": 0.8218,
      "step": 354
    },
    {
      "epoch": 0.16569428238039674,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019987189034706464,
      "loss": 0.8671,
      "step": 355
    },
    {
      "epoch": 0.1661610268378063,
      "grad_norm": 0.5,
      "learning_rate": 0.0001998711467658478,
      "loss": 0.8414,
      "step": 356
    },
    {
      "epoch": 0.16662777129521586,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019987040103429436,
      "loss": 0.8736,
      "step": 357
    },
    {
      "epoch": 0.16709451575262543,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00019986965315242037,
      "loss": 0.7668,
      "step": 358
    },
    {
      "epoch": 0.167561260210035,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019986890312024197,
      "loss": 0.8707,
      "step": 359
    },
    {
      "epoch": 0.16802800466744458,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00019986815093777524,
      "loss": 0.708,
      "step": 360
    },
    {
      "epoch": 0.16849474912485415,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019986739660503644,
      "loss": 0.8333,
      "step": 361
    },
    {
      "epoch": 0.1689614935822637,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019986664012204176,
      "loss": 0.8402,
      "step": 362
    },
    {
      "epoch": 0.16942823803967327,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019986588148880754,
      "loss": 0.8683,
      "step": 363
    },
    {
      "epoch": 0.16989498249708285,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001998651207053501,
      "loss": 1.028,
      "step": 364
    },
    {
      "epoch": 0.17036172695449242,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019986435777168577,
      "loss": 0.8522,
      "step": 365
    },
    {
      "epoch": 0.170828471411902,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019986359268783104,
      "loss": 0.8638,
      "step": 366
    },
    {
      "epoch": 0.17129521586931154,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019986282545380235,
      "loss": 0.7876,
      "step": 367
    },
    {
      "epoch": 0.1717619603267211,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019986205606961624,
      "loss": 0.7971,
      "step": 368
    },
    {
      "epoch": 0.1722287047841307,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019986128453528925,
      "loss": 0.8204,
      "step": 369
    },
    {
      "epoch": 0.17269544924154026,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000199860510850838,
      "loss": 0.8482,
      "step": 370
    },
    {
      "epoch": 0.17316219369894983,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019985973501627916,
      "loss": 0.9228,
      "step": 371
    },
    {
      "epoch": 0.17362893815635938,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019985895703162945,
      "loss": 1.0125,
      "step": 372
    },
    {
      "epoch": 0.17409568261376895,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019985817689690557,
      "loss": 0.9471,
      "step": 373
    },
    {
      "epoch": 0.17456242707117853,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019985739461212437,
      "loss": 0.8284,
      "step": 374
    },
    {
      "epoch": 0.1750291715285881,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019985661017730267,
      "loss": 0.8444,
      "step": 375
    },
    {
      "epoch": 0.17549591598599767,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019985582359245734,
      "loss": 1.0169,
      "step": 376
    },
    {
      "epoch": 0.17596266044340725,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019985503485760536,
      "loss": 0.8166,
      "step": 377
    },
    {
      "epoch": 0.1764294049008168,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019985424397276366,
      "loss": 0.7685,
      "step": 378
    },
    {
      "epoch": 0.17689614935822637,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001998534509379493,
      "loss": 0.8422,
      "step": 379
    },
    {
      "epoch": 0.17736289381563594,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019985265575317936,
      "loss": 0.8165,
      "step": 380
    },
    {
      "epoch": 0.17782963827304551,
      "grad_norm": 0.46875,
      "learning_rate": 0.00019985185841847098,
      "loss": 0.7526,
      "step": 381
    },
    {
      "epoch": 0.1782963827304551,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019985105893384128,
      "loss": 0.9166,
      "step": 382
    },
    {
      "epoch": 0.17876312718786463,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00019985025729930752,
      "loss": 0.7758,
      "step": 383
    },
    {
      "epoch": 0.1792298716452742,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019984945351488688,
      "loss": 0.7748,
      "step": 384
    },
    {
      "epoch": 0.17969661610268378,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019984864758059676,
      "loss": 0.9146,
      "step": 385
    },
    {
      "epoch": 0.18016336056009336,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019984783949645446,
      "loss": 0.7409,
      "step": 386
    },
    {
      "epoch": 0.18063010501750293,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001998470292624774,
      "loss": 0.88,
      "step": 387
    },
    {
      "epoch": 0.18109684947491247,
      "grad_norm": 0.578125,
      "learning_rate": 0.000199846216878683,
      "loss": 1.0229,
      "step": 388
    },
    {
      "epoch": 0.18156359393232205,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00019984540234508879,
      "loss": 0.8394,
      "step": 389
    },
    {
      "epoch": 0.18203033838973162,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001998445856617123,
      "loss": 0.8287,
      "step": 390
    },
    {
      "epoch": 0.1824970828471412,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019984376682857108,
      "loss": 0.8967,
      "step": 391
    },
    {
      "epoch": 0.18296382730455077,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019984294584568277,
      "loss": 0.8842,
      "step": 392
    },
    {
      "epoch": 0.18343057176196031,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001998421227130651,
      "loss": 0.7919,
      "step": 393
    },
    {
      "epoch": 0.1838973162193699,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019984129743073567,
      "loss": 0.8026,
      "step": 394
    },
    {
      "epoch": 0.18436406067677946,
      "grad_norm": 0.5,
      "learning_rate": 0.0001998404699987124,
      "loss": 0.8767,
      "step": 395
    },
    {
      "epoch": 0.18483080513418904,
      "grad_norm": 0.5,
      "learning_rate": 0.000199839640417013,
      "loss": 0.8463,
      "step": 396
    },
    {
      "epoch": 0.1852975495915986,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019983880868565542,
      "loss": 0.9201,
      "step": 397
    },
    {
      "epoch": 0.18576429404900816,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019983797480465748,
      "loss": 0.782,
      "step": 398
    },
    {
      "epoch": 0.18623103850641773,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019983713877403715,
      "loss": 0.8767,
      "step": 399
    },
    {
      "epoch": 0.1866977829638273,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019983630059381248,
      "loss": 0.9837,
      "step": 400
    },
    {
      "epoch": 0.18716452742123688,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019983546026400149,
      "loss": 0.7993,
      "step": 401
    },
    {
      "epoch": 0.18763127187864645,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019983461778462226,
      "loss": 0.874,
      "step": 402
    },
    {
      "epoch": 0.18809801633605602,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019983377315569293,
      "loss": 0.9171,
      "step": 403
    },
    {
      "epoch": 0.18856476079346557,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019983292637723173,
      "loss": 0.832,
      "step": 404
    },
    {
      "epoch": 0.18903150525087514,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019983207744925683,
      "loss": 0.821,
      "step": 405
    },
    {
      "epoch": 0.18949824970828472,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019983122637178656,
      "loss": 0.9555,
      "step": 406
    },
    {
      "epoch": 0.1899649941656943,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019983037314483918,
      "loss": 0.9288,
      "step": 407
    },
    {
      "epoch": 0.19043173862310386,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019982951776843316,
      "loss": 0.8557,
      "step": 408
    },
    {
      "epoch": 0.1908984830805134,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019982866024258682,
      "loss": 0.8824,
      "step": 409
    },
    {
      "epoch": 0.19136522753792298,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019982780056731868,
      "loss": 0.8061,
      "step": 410
    },
    {
      "epoch": 0.19183197199533256,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001998269387426472,
      "loss": 0.9368,
      "step": 411
    },
    {
      "epoch": 0.19229871645274213,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019982607476859098,
      "loss": 0.8888,
      "step": 412
    },
    {
      "epoch": 0.1927654609101517,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019982520864516865,
      "loss": 0.9224,
      "step": 413
    },
    {
      "epoch": 0.19323220536756125,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019982434037239878,
      "loss": 0.8144,
      "step": 414
    },
    {
      "epoch": 0.19369894982497082,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019982346995030015,
      "loss": 0.8414,
      "step": 415
    },
    {
      "epoch": 0.1941656942823804,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019982259737889142,
      "loss": 0.8562,
      "step": 416
    },
    {
      "epoch": 0.19463243873978997,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001998217226581914,
      "loss": 0.7829,
      "step": 417
    },
    {
      "epoch": 0.19509918319719954,
      "grad_norm": 0.5,
      "learning_rate": 0.00019982084578821897,
      "loss": 0.8173,
      "step": 418
    },
    {
      "epoch": 0.1955659276546091,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019981996676899296,
      "loss": 0.8386,
      "step": 419
    },
    {
      "epoch": 0.19603267211201866,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019981908560053234,
      "loss": 0.9298,
      "step": 420
    },
    {
      "epoch": 0.19649941656942824,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019981820228285598,
      "loss": 0.9126,
      "step": 421
    },
    {
      "epoch": 0.1969661610268378,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00019981731681598306,
      "loss": 0.6823,
      "step": 422
    },
    {
      "epoch": 0.19743290548424738,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019981642919993252,
      "loss": 0.7723,
      "step": 423
    },
    {
      "epoch": 0.19789964994165693,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001998155394347235,
      "loss": 0.8236,
      "step": 424
    },
    {
      "epoch": 0.1983663943990665,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019981464752037522,
      "loss": 0.7825,
      "step": 425
    },
    {
      "epoch": 0.19883313885647608,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019981375345690678,
      "loss": 0.9205,
      "step": 426
    },
    {
      "epoch": 0.19929988331388565,
      "grad_norm": 0.5,
      "learning_rate": 0.0001998128572443375,
      "loss": 0.9017,
      "step": 427
    },
    {
      "epoch": 0.19976662777129522,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019981195888268667,
      "loss": 0.8884,
      "step": 428
    },
    {
      "epoch": 0.2002333722287048,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019981105837197364,
      "loss": 0.8075,
      "step": 429
    },
    {
      "epoch": 0.20070011668611434,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019981015571221772,
      "loss": 0.9266,
      "step": 430
    },
    {
      "epoch": 0.20116686114352392,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019980925090343844,
      "loss": 0.8949,
      "step": 431
    },
    {
      "epoch": 0.2016336056009335,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00019980834394565524,
      "loss": 0.7104,
      "step": 432
    },
    {
      "epoch": 0.20210035005834306,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001998074348388877,
      "loss": 0.8944,
      "step": 433
    },
    {
      "epoch": 0.20256709451575264,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001998065235831553,
      "loss": 0.7776,
      "step": 434
    },
    {
      "epoch": 0.20303383897316218,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001998056101784777,
      "loss": 0.8335,
      "step": 435
    },
    {
      "epoch": 0.20350058343057176,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019980469462487461,
      "loss": 0.8517,
      "step": 436
    },
    {
      "epoch": 0.20396732788798133,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019980377692236569,
      "loss": 0.9135,
      "step": 437
    },
    {
      "epoch": 0.2044340723453909,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019980285707097074,
      "loss": 0.8213,
      "step": 438
    },
    {
      "epoch": 0.20490081680280048,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019980193507070952,
      "loss": 0.8519,
      "step": 439
    },
    {
      "epoch": 0.20536756126021002,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019980101092160194,
      "loss": 0.9009,
      "step": 440
    },
    {
      "epoch": 0.2058343057176196,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019980008462366783,
      "loss": 1.0034,
      "step": 441
    },
    {
      "epoch": 0.20630105017502917,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001997991561769272,
      "loss": 0.8987,
      "step": 442
    },
    {
      "epoch": 0.20676779463243875,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019979822558139997,
      "loss": 0.9159,
      "step": 443
    },
    {
      "epoch": 0.20723453908984832,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00019979729283710622,
      "loss": 0.779,
      "step": 444
    },
    {
      "epoch": 0.20770128354725786,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019979635794406607,
      "loss": 0.7709,
      "step": 445
    },
    {
      "epoch": 0.20816802800466744,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019979542090229957,
      "loss": 0.9629,
      "step": 446
    },
    {
      "epoch": 0.208634772462077,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019979448171182694,
      "loss": 0.9416,
      "step": 447
    },
    {
      "epoch": 0.20910151691948659,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001997935403726684,
      "loss": 0.8382,
      "step": 448
    },
    {
      "epoch": 0.20956826137689616,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001997925968848442,
      "loss": 0.8457,
      "step": 449
    },
    {
      "epoch": 0.2100350058343057,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019979165124837467,
      "loss": 0.7761,
      "step": 450
    },
    {
      "epoch": 0.21050175029171528,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001997907034632802,
      "loss": 0.8188,
      "step": 451
    },
    {
      "epoch": 0.21096849474912485,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019978975352958115,
      "loss": 0.8768,
      "step": 452
    },
    {
      "epoch": 0.21143523920653443,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019978880144729794,
      "loss": 1.0063,
      "step": 453
    },
    {
      "epoch": 0.211901983663944,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019978784721645117,
      "loss": 0.8357,
      "step": 454
    },
    {
      "epoch": 0.21236872812135357,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001997868908370613,
      "loss": 0.8385,
      "step": 455
    },
    {
      "epoch": 0.21283547257876312,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000199785932309149,
      "loss": 0.7696,
      "step": 456
    },
    {
      "epoch": 0.2133022170361727,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019978497163273482,
      "loss": 0.8369,
      "step": 457
    },
    {
      "epoch": 0.21376896149358227,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001997840088078395,
      "loss": 1.0143,
      "step": 458
    },
    {
      "epoch": 0.21423570595099184,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001997830438344838,
      "loss": 0.7826,
      "step": 459
    },
    {
      "epoch": 0.2147024504084014,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019978207671268839,
      "loss": 0.9136,
      "step": 460
    },
    {
      "epoch": 0.21516919486581096,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001997811074424742,
      "loss": 0.845,
      "step": 461
    },
    {
      "epoch": 0.21563593932322053,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019978013602386203,
      "loss": 0.9591,
      "step": 462
    },
    {
      "epoch": 0.2161026837806301,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00019977916245687282,
      "loss": 0.818,
      "step": 463
    },
    {
      "epoch": 0.21656942823803968,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00019977818674152755,
      "loss": 0.7291,
      "step": 464
    },
    {
      "epoch": 0.21703617269544925,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019977720887784722,
      "loss": 0.7468,
      "step": 465
    },
    {
      "epoch": 0.2175029171528588,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019977622886585287,
      "loss": 0.7187,
      "step": 466
    },
    {
      "epoch": 0.21796966161026837,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019977524670556563,
      "loss": 0.8585,
      "step": 467
    },
    {
      "epoch": 0.21843640606767795,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001997742623970066,
      "loss": 0.7945,
      "step": 468
    },
    {
      "epoch": 0.21890315052508752,
      "grad_norm": 0.515625,
      "learning_rate": 0.000199773275940197,
      "loss": 0.8108,
      "step": 469
    },
    {
      "epoch": 0.2193698949824971,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001997722873351581,
      "loss": 0.7664,
      "step": 470
    },
    {
      "epoch": 0.21983663943990664,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019977129658191108,
      "loss": 0.8279,
      "step": 471
    },
    {
      "epoch": 0.2203033838973162,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019977030368047743,
      "loss": 0.9911,
      "step": 472
    },
    {
      "epoch": 0.2207701283547258,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019976930863087843,
      "loss": 0.8442,
      "step": 473
    },
    {
      "epoch": 0.22123687281213536,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019976831143313548,
      "loss": 0.8565,
      "step": 474
    },
    {
      "epoch": 0.22170361726954493,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019976731208727012,
      "loss": 0.7531,
      "step": 475
    },
    {
      "epoch": 0.22217036172695448,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019976631059330383,
      "loss": 0.7538,
      "step": 476
    },
    {
      "epoch": 0.22263710618436405,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019976530695125816,
      "loss": 0.9925,
      "step": 477
    },
    {
      "epoch": 0.22310385064177363,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019976430116115475,
      "loss": 0.8312,
      "step": 478
    },
    {
      "epoch": 0.2235705950991832,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019976329322301525,
      "loss": 0.856,
      "step": 479
    },
    {
      "epoch": 0.22403733955659277,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019976228313686137,
      "loss": 0.964,
      "step": 480
    },
    {
      "epoch": 0.22450408401400235,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019976127090271483,
      "loss": 0.8208,
      "step": 481
    },
    {
      "epoch": 0.2249708284714119,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019976025652059746,
      "loss": 0.8248,
      "step": 482
    },
    {
      "epoch": 0.22543757292882147,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019975923999053107,
      "loss": 0.7991,
      "step": 483
    },
    {
      "epoch": 0.22590431738623104,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019975822131253757,
      "loss": 0.715,
      "step": 484
    },
    {
      "epoch": 0.22637106184364061,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019975720048663885,
      "loss": 0.7423,
      "step": 485
    },
    {
      "epoch": 0.2268378063010502,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019975617751285694,
      "loss": 0.8253,
      "step": 486
    },
    {
      "epoch": 0.22730455075845973,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019975515239121383,
      "loss": 0.7883,
      "step": 487
    },
    {
      "epoch": 0.2277712952158693,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019975412512173165,
      "loss": 0.7187,
      "step": 488
    },
    {
      "epoch": 0.22823803967327888,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019975309570443244,
      "loss": 0.8448,
      "step": 489
    },
    {
      "epoch": 0.22870478413068845,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001997520641393384,
      "loss": 0.8407,
      "step": 490
    },
    {
      "epoch": 0.22917152858809803,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019975103042647177,
      "loss": 0.9334,
      "step": 491
    },
    {
      "epoch": 0.22963827304550757,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019974999456585474,
      "loss": 0.7595,
      "step": 492
    },
    {
      "epoch": 0.23010501750291715,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001997489565575097,
      "loss": 0.8393,
      "step": 493
    },
    {
      "epoch": 0.23057176196032672,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019974791640145893,
      "loss": 0.7795,
      "step": 494
    },
    {
      "epoch": 0.2310385064177363,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019974687409772483,
      "loss": 0.8779,
      "step": 495
    },
    {
      "epoch": 0.23150525087514587,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001997458296463299,
      "loss": 0.7768,
      "step": 496
    },
    {
      "epoch": 0.23197199533255541,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019974478304729654,
      "loss": 0.8143,
      "step": 497
    },
    {
      "epoch": 0.232438739789965,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019974373430064736,
      "loss": 0.8656,
      "step": 498
    },
    {
      "epoch": 0.23290548424737456,
      "grad_norm": 0.625,
      "learning_rate": 0.00019974268340640488,
      "loss": 0.9543,
      "step": 499
    },
    {
      "epoch": 0.23337222870478413,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019974163036459182,
      "loss": 0.9469,
      "step": 500
    },
    {
      "epoch": 0.2338389731621937,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019974057517523078,
      "loss": 0.8112,
      "step": 501
    },
    {
      "epoch": 0.23430571761960325,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019973951783834446,
      "loss": 0.6788,
      "step": 502
    },
    {
      "epoch": 0.23477246207701283,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019973845835395565,
      "loss": 0.8324,
      "step": 503
    },
    {
      "epoch": 0.2352392065344224,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001997373967220872,
      "loss": 0.9217,
      "step": 504
    },
    {
      "epoch": 0.23570595099183198,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019973633294276196,
      "loss": 0.8814,
      "step": 505
    },
    {
      "epoch": 0.23617269544924155,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019973526701600277,
      "loss": 0.766,
      "step": 506
    },
    {
      "epoch": 0.23663943990665112,
      "grad_norm": 0.5,
      "learning_rate": 0.00019973419894183262,
      "loss": 0.7433,
      "step": 507
    },
    {
      "epoch": 0.23710618436406067,
      "grad_norm": 0.5,
      "learning_rate": 0.00019973312872027452,
      "loss": 0.7257,
      "step": 508
    },
    {
      "epoch": 0.23757292882147024,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019973205635135153,
      "loss": 0.7668,
      "step": 509
    },
    {
      "epoch": 0.23803967327887982,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019973098183508666,
      "loss": 0.8133,
      "step": 510
    },
    {
      "epoch": 0.2385064177362894,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019972990517150315,
      "loss": 0.7507,
      "step": 511
    },
    {
      "epoch": 0.23897316219369896,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019972882636062412,
      "loss": 0.7349,
      "step": 512
    },
    {
      "epoch": 0.2394399066511085,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019972774540247276,
      "loss": 0.8308,
      "step": 513
    },
    {
      "epoch": 0.23990665110851808,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019972666229707244,
      "loss": 0.7733,
      "step": 514
    },
    {
      "epoch": 0.24037339556592766,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019972557704444642,
      "loss": 0.8146,
      "step": 515
    },
    {
      "epoch": 0.24084014002333723,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019972448964461806,
      "loss": 0.9102,
      "step": 516
    },
    {
      "epoch": 0.2413068844807468,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019972340009761084,
      "loss": 0.7334,
      "step": 517
    },
    {
      "epoch": 0.24177362893815635,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00019972230840344812,
      "loss": 0.7908,
      "step": 518
    },
    {
      "epoch": 0.24224037339556592,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019972121456215348,
      "loss": 0.7758,
      "step": 519
    },
    {
      "epoch": 0.2427071178529755,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019972011857375045,
      "loss": 0.8219,
      "step": 520
    },
    {
      "epoch": 0.24317386231038507,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019971902043826263,
      "loss": 0.8146,
      "step": 521
    },
    {
      "epoch": 0.24364060676779464,
      "grad_norm": 0.625,
      "learning_rate": 0.00019971792015571364,
      "loss": 0.9178,
      "step": 522
    },
    {
      "epoch": 0.2441073512252042,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001997168177261272,
      "loss": 0.8689,
      "step": 523
    },
    {
      "epoch": 0.24457409568261376,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019971571314952703,
      "loss": 0.756,
      "step": 524
    },
    {
      "epoch": 0.24504084014002334,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019971460642593692,
      "loss": 0.8528,
      "step": 525
    },
    {
      "epoch": 0.2455075845974329,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001997134975553807,
      "loss": 0.8385,
      "step": 526
    },
    {
      "epoch": 0.24597432905484248,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00019971238653788227,
      "loss": 0.7451,
      "step": 527
    },
    {
      "epoch": 0.24644107351225203,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019971127337346548,
      "loss": 0.8731,
      "step": 528
    },
    {
      "epoch": 0.2469078179696616,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019971015806215437,
      "loss": 0.7527,
      "step": 529
    },
    {
      "epoch": 0.24737456242707118,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019970904060397294,
      "loss": 0.9084,
      "step": 530
    },
    {
      "epoch": 0.24784130688448075,
      "grad_norm": 0.47265625,
      "learning_rate": 0.0001997079209989452,
      "loss": 0.6927,
      "step": 531
    },
    {
      "epoch": 0.24830805134189032,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019970679924709533,
      "loss": 0.8436,
      "step": 532
    },
    {
      "epoch": 0.2487747957992999,
      "grad_norm": 0.5,
      "learning_rate": 0.0001997056753484474,
      "loss": 0.7204,
      "step": 533
    },
    {
      "epoch": 0.24924154025670944,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001997045493030257,
      "loss": 0.7987,
      "step": 534
    },
    {
      "epoch": 0.24970828471411902,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001997034211108544,
      "loss": 0.7463,
      "step": 535
    },
    {
      "epoch": 0.25017502917152856,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019970229077195787,
      "loss": 0.7439,
      "step": 536
    },
    {
      "epoch": 0.25017502917152856,
      "eval_loss": 0.9957383871078491,
      "eval_runtime": 96.4538,
      "eval_samples_per_second": 18.703,
      "eval_steps_per_second": 2.343,
      "step": 536
    },
    {
      "epoch": 0.25064177362893814,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019970115828636033,
      "loss": 0.8223,
      "step": 537
    },
    {
      "epoch": 0.2511085180863477,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019970002365408626,
      "loss": 0.8557,
      "step": 538
    },
    {
      "epoch": 0.2515752625437573,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019969888687516006,
      "loss": 0.8198,
      "step": 539
    },
    {
      "epoch": 0.25204200700116686,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019969774794960623,
      "loss": 0.7595,
      "step": 540
    },
    {
      "epoch": 0.25250875145857643,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019969660687744925,
      "loss": 0.8335,
      "step": 541
    },
    {
      "epoch": 0.252975495915986,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001996954636587137,
      "loss": 0.8795,
      "step": 542
    },
    {
      "epoch": 0.2534422403733956,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019969431829342419,
      "loss": 0.8579,
      "step": 543
    },
    {
      "epoch": 0.25390898483080515,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019969317078160544,
      "loss": 0.7966,
      "step": 544
    },
    {
      "epoch": 0.2543757292882147,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001996920211232821,
      "loss": 0.9128,
      "step": 545
    },
    {
      "epoch": 0.2548424737456243,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019969086931847892,
      "loss": 0.7504,
      "step": 546
    },
    {
      "epoch": 0.2553092182030338,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019968971536722072,
      "loss": 0.7172,
      "step": 547
    },
    {
      "epoch": 0.2557759626604434,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00019968855926953233,
      "loss": 0.7656,
      "step": 548
    },
    {
      "epoch": 0.25624270711785296,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019968740102543865,
      "loss": 0.7943,
      "step": 549
    },
    {
      "epoch": 0.25670945157526254,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019968624063496464,
      "loss": 0.7665,
      "step": 550
    },
    {
      "epoch": 0.2571761960326721,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00019968507809813526,
      "loss": 0.748,
      "step": 551
    },
    {
      "epoch": 0.2576429404900817,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019968391341497555,
      "loss": 0.8829,
      "step": 552
    },
    {
      "epoch": 0.25810968494749126,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019968274658551058,
      "loss": 0.7753,
      "step": 553
    },
    {
      "epoch": 0.25857642940490083,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019968157760976544,
      "loss": 0.8457,
      "step": 554
    },
    {
      "epoch": 0.2590431738623104,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019968040648776542,
      "loss": 0.8644,
      "step": 555
    },
    {
      "epoch": 0.25950991831972,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019967923321953556,
      "loss": 0.827,
      "step": 556
    },
    {
      "epoch": 0.2599766627771295,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019967805780510126,
      "loss": 0.7913,
      "step": 557
    },
    {
      "epoch": 0.26044340723453907,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019967688024448776,
      "loss": 0.771,
      "step": 558
    },
    {
      "epoch": 0.26091015169194864,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00019967570053772046,
      "loss": 0.7563,
      "step": 559
    },
    {
      "epoch": 0.2613768961493582,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019967451868482475,
      "loss": 0.8177,
      "step": 560
    },
    {
      "epoch": 0.2618436406067678,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019967333468582602,
      "loss": 0.8346,
      "step": 561
    },
    {
      "epoch": 0.26231038506417736,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019967214854074982,
      "loss": 0.8214,
      "step": 562
    },
    {
      "epoch": 0.26277712952158694,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019967096024962167,
      "loss": 0.9256,
      "step": 563
    },
    {
      "epoch": 0.2632438739789965,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019966976981246716,
      "loss": 0.7873,
      "step": 564
    },
    {
      "epoch": 0.2637106184364061,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00019966857722931192,
      "loss": 0.7267,
      "step": 565
    },
    {
      "epoch": 0.26417736289381566,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019966738250018162,
      "loss": 0.9068,
      "step": 566
    },
    {
      "epoch": 0.2646441073512252,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000199666185625102,
      "loss": 0.7013,
      "step": 567
    },
    {
      "epoch": 0.26511085180863475,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019966498660409885,
      "loss": 0.724,
      "step": 568
    },
    {
      "epoch": 0.2655775962660443,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019966378543719794,
      "loss": 0.9003,
      "step": 569
    },
    {
      "epoch": 0.2660443407234539,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019966258212442513,
      "loss": 0.7738,
      "step": 570
    },
    {
      "epoch": 0.26651108518086347,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019966137666580636,
      "loss": 0.7163,
      "step": 571
    },
    {
      "epoch": 0.26697782963827305,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019966016906136758,
      "loss": 0.8256,
      "step": 572
    },
    {
      "epoch": 0.2674445740956826,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00019965895931113478,
      "loss": 0.6719,
      "step": 573
    },
    {
      "epoch": 0.2679113185530922,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019965774741513402,
      "loss": 0.8816,
      "step": 574
    },
    {
      "epoch": 0.26837806301050177,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001996565333733914,
      "loss": 0.7635,
      "step": 575
    },
    {
      "epoch": 0.26884480746791134,
      "grad_norm": 0.53125,
      "learning_rate": 0.000199655317185933,
      "loss": 0.8282,
      "step": 576
    },
    {
      "epoch": 0.2693115519253209,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019965409885278507,
      "loss": 0.7184,
      "step": 577
    },
    {
      "epoch": 0.26977829638273043,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00019965287837397385,
      "loss": 0.7891,
      "step": 578
    },
    {
      "epoch": 0.27024504084014,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019965165574952555,
      "loss": 0.8856,
      "step": 579
    },
    {
      "epoch": 0.2707117852975496,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019965043097946658,
      "loss": 0.7419,
      "step": 580
    },
    {
      "epoch": 0.27117852975495915,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00019964920406382322,
      "loss": 0.6847,
      "step": 581
    },
    {
      "epoch": 0.2716452742123687,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019964797500262197,
      "loss": 0.719,
      "step": 582
    },
    {
      "epoch": 0.2721120186697783,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019964674379588927,
      "loss": 0.8563,
      "step": 583
    },
    {
      "epoch": 0.2725787631271879,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019964551044365157,
      "loss": 0.8969,
      "step": 584
    },
    {
      "epoch": 0.27304550758459745,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019964427494593553,
      "loss": 0.8194,
      "step": 585
    },
    {
      "epoch": 0.273512252042007,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019964303730276764,
      "loss": 0.8075,
      "step": 586
    },
    {
      "epoch": 0.2739789964994166,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00019964179751417462,
      "loss": 0.7459,
      "step": 587
    },
    {
      "epoch": 0.2744457409568261,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019964055558018313,
      "loss": 0.7756,
      "step": 588
    },
    {
      "epoch": 0.2749124854142357,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019963931150081995,
      "loss": 0.8687,
      "step": 589
    },
    {
      "epoch": 0.27537922987164526,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019963806527611183,
      "loss": 0.785,
      "step": 590
    },
    {
      "epoch": 0.27584597432905483,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001996368169060856,
      "loss": 0.8049,
      "step": 591
    },
    {
      "epoch": 0.2763127187864644,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019963556639076817,
      "loss": 0.7738,
      "step": 592
    },
    {
      "epoch": 0.276779463243874,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019963431373018646,
      "loss": 0.8263,
      "step": 593
    },
    {
      "epoch": 0.27724620770128355,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019963305892436742,
      "loss": 0.6784,
      "step": 594
    },
    {
      "epoch": 0.2777129521586931,
      "grad_norm": 0.515625,
      "learning_rate": 0.00019963180197333805,
      "loss": 0.7123,
      "step": 595
    },
    {
      "epoch": 0.2781796966161027,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019963054287712545,
      "loss": 0.7788,
      "step": 596
    },
    {
      "epoch": 0.2786464410735123,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019962928163575676,
      "loss": 0.8286,
      "step": 597
    },
    {
      "epoch": 0.27911318553092185,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019962801824925904,
      "loss": 0.7915,
      "step": 598
    },
    {
      "epoch": 0.27957992998833137,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001996267527176596,
      "loss": 0.7893,
      "step": 599
    },
    {
      "epoch": 0.28004667444574094,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001996254850409856,
      "loss": 0.9239,
      "step": 600
    },
    {
      "epoch": 0.2805134189031505,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001996242152192644,
      "loss": 0.8571,
      "step": 601
    },
    {
      "epoch": 0.2809801633605601,
      "grad_norm": 0.484375,
      "learning_rate": 0.00019962294325252328,
      "loss": 0.7047,
      "step": 602
    },
    {
      "epoch": 0.28144690781796966,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019962166914078968,
      "loss": 0.6812,
      "step": 603
    },
    {
      "epoch": 0.28191365227537923,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000199620392884091,
      "loss": 0.7156,
      "step": 604
    },
    {
      "epoch": 0.2823803967327888,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019961911448245474,
      "loss": 0.76,
      "step": 605
    },
    {
      "epoch": 0.2828471411901984,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019961783393590842,
      "loss": 0.7147,
      "step": 606
    },
    {
      "epoch": 0.28331388564760795,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001996165512444796,
      "loss": 0.7387,
      "step": 607
    },
    {
      "epoch": 0.28378063010501753,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019961526640819587,
      "loss": 0.715,
      "step": 608
    },
    {
      "epoch": 0.28424737456242705,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000199613979427085,
      "loss": 0.8067,
      "step": 609
    },
    {
      "epoch": 0.2847141190198366,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019961269030117457,
      "loss": 0.808,
      "step": 610
    },
    {
      "epoch": 0.2851808634772462,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019961139903049242,
      "loss": 0.7604,
      "step": 611
    },
    {
      "epoch": 0.28564760793465577,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019961010561506633,
      "loss": 0.7635,
      "step": 612
    },
    {
      "epoch": 0.28611435239206534,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019960881005492414,
      "loss": 0.7656,
      "step": 613
    },
    {
      "epoch": 0.2865810968494749,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019960751235009375,
      "loss": 0.7342,
      "step": 614
    },
    {
      "epoch": 0.2870478413068845,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019960621250060312,
      "loss": 0.8269,
      "step": 615
    },
    {
      "epoch": 0.28751458576429406,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001996049105064802,
      "loss": 0.7303,
      "step": 616
    },
    {
      "epoch": 0.28798133022170364,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019960360636775304,
      "loss": 0.71,
      "step": 617
    },
    {
      "epoch": 0.2884480746791132,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019960230008444973,
      "loss": 0.9302,
      "step": 618
    },
    {
      "epoch": 0.2889148191365227,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001996009916565984,
      "loss": 0.7775,
      "step": 619
    },
    {
      "epoch": 0.2893815635939323,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001995996810842272,
      "loss": 0.8243,
      "step": 620
    },
    {
      "epoch": 0.2898483080513419,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019959836836736433,
      "loss": 0.7412,
      "step": 621
    },
    {
      "epoch": 0.29031505250875145,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00019959705350603812,
      "loss": 0.65,
      "step": 622
    },
    {
      "epoch": 0.290781796966161,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019959573650027687,
      "loss": 0.77,
      "step": 623
    },
    {
      "epoch": 0.2912485414235706,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019959441735010883,
      "loss": 0.7805,
      "step": 624
    },
    {
      "epoch": 0.29171528588098017,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019959309605556256,
      "loss": 0.7656,
      "step": 625
    },
    {
      "epoch": 0.29218203033838974,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019959177261666638,
      "loss": 0.8074,
      "step": 626
    },
    {
      "epoch": 0.2926487747957993,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019959044703344886,
      "loss": 0.8004,
      "step": 627
    },
    {
      "epoch": 0.2931155192532089,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019958911930593854,
      "loss": 0.7589,
      "step": 628
    },
    {
      "epoch": 0.29358226371061846,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019958778943416394,
      "loss": 0.7345,
      "step": 629
    },
    {
      "epoch": 0.294049008168028,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019958645741815375,
      "loss": 0.7486,
      "step": 630
    },
    {
      "epoch": 0.29451575262543755,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019958512325793666,
      "loss": 0.7811,
      "step": 631
    },
    {
      "epoch": 0.29498249708284713,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019958378695354134,
      "loss": 0.713,
      "step": 632
    },
    {
      "epoch": 0.2954492415402567,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019958244850499664,
      "loss": 0.7191,
      "step": 633
    },
    {
      "epoch": 0.2959159859976663,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001995811079123313,
      "loss": 0.8741,
      "step": 634
    },
    {
      "epoch": 0.29638273045507585,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019957976517557424,
      "loss": 0.7818,
      "step": 635
    },
    {
      "epoch": 0.2968494749124854,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019957842029475433,
      "loss": 0.6942,
      "step": 636
    },
    {
      "epoch": 0.297316219369895,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019957707326990058,
      "loss": 0.7754,
      "step": 637
    },
    {
      "epoch": 0.29778296382730457,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019957572410104192,
      "loss": 0.7771,
      "step": 638
    },
    {
      "epoch": 0.29824970828471414,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019957437278820747,
      "loss": 0.74,
      "step": 639
    },
    {
      "epoch": 0.29871645274212366,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019957301933142628,
      "loss": 0.7339,
      "step": 640
    },
    {
      "epoch": 0.29918319719953324,
      "grad_norm": 0.625,
      "learning_rate": 0.0001995716637307275,
      "loss": 0.874,
      "step": 641
    },
    {
      "epoch": 0.2996499416569428,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019957030598614033,
      "loss": 0.8613,
      "step": 642
    },
    {
      "epoch": 0.3001166861143524,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019956894609769401,
      "loss": 0.7766,
      "step": 643
    },
    {
      "epoch": 0.30058343057176196,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019956758406541777,
      "loss": 0.8244,
      "step": 644
    },
    {
      "epoch": 0.30105017502917153,
      "grad_norm": 0.546875,
      "learning_rate": 0.000199566219889341,
      "loss": 0.804,
      "step": 645
    },
    {
      "epoch": 0.3015169194865811,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019956485356949306,
      "loss": 0.8804,
      "step": 646
    },
    {
      "epoch": 0.3019836639439907,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019956348510590335,
      "loss": 0.7264,
      "step": 647
    },
    {
      "epoch": 0.30245040840140025,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019956211449860133,
      "loss": 0.9112,
      "step": 648
    },
    {
      "epoch": 0.3029171528588098,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001995607417476165,
      "loss": 0.6976,
      "step": 649
    },
    {
      "epoch": 0.3033838973162194,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019955936685297845,
      "loss": 0.685,
      "step": 650
    },
    {
      "epoch": 0.3038506417736289,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019955798981471676,
      "loss": 0.8505,
      "step": 651
    },
    {
      "epoch": 0.3043173862310385,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001995566106328611,
      "loss": 0.8052,
      "step": 652
    },
    {
      "epoch": 0.30478413068844806,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019955522930744114,
      "loss": 0.7442,
      "step": 653
    },
    {
      "epoch": 0.30525087514585764,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019955384583848665,
      "loss": 0.7123,
      "step": 654
    },
    {
      "epoch": 0.3057176196032672,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019955246022602738,
      "loss": 0.7227,
      "step": 655
    },
    {
      "epoch": 0.3061843640606768,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019955107247009323,
      "loss": 0.724,
      "step": 656
    },
    {
      "epoch": 0.30665110851808636,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000199549682570714,
      "loss": 0.7437,
      "step": 657
    },
    {
      "epoch": 0.30711785297549593,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019954829052791967,
      "loss": 0.7436,
      "step": 658
    },
    {
      "epoch": 0.3075845974329055,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001995468963417402,
      "loss": 0.7072,
      "step": 659
    },
    {
      "epoch": 0.3080513418903151,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019954550001220555,
      "loss": 0.9228,
      "step": 660
    },
    {
      "epoch": 0.3085180863477246,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019954410153934587,
      "loss": 0.8562,
      "step": 661
    },
    {
      "epoch": 0.30898483080513417,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019954270092319124,
      "loss": 0.7602,
      "step": 662
    },
    {
      "epoch": 0.30945157526254374,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019954129816377182,
      "loss": 0.7952,
      "step": 663
    },
    {
      "epoch": 0.3099183197199533,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019953989326111779,
      "loss": 0.7912,
      "step": 664
    },
    {
      "epoch": 0.3103850641773629,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00019953848621525943,
      "loss": 0.6111,
      "step": 665
    },
    {
      "epoch": 0.31085180863477246,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019953707702622704,
      "loss": 0.8098,
      "step": 666
    },
    {
      "epoch": 0.31131855309218204,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019953566569405094,
      "loss": 0.8148,
      "step": 667
    },
    {
      "epoch": 0.3117852975495916,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001995342522187615,
      "loss": 0.7623,
      "step": 668
    },
    {
      "epoch": 0.3122520420070012,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019953283660038916,
      "loss": 0.7721,
      "step": 669
    },
    {
      "epoch": 0.31271878646441076,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019953141883896446,
      "loss": 0.711,
      "step": 670
    },
    {
      "epoch": 0.3131855309218203,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019952999893451786,
      "loss": 0.6888,
      "step": 671
    },
    {
      "epoch": 0.31365227537922985,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001995285768870799,
      "loss": 0.7593,
      "step": 672
    },
    {
      "epoch": 0.3141190198366394,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019952715269668132,
      "loss": 0.7696,
      "step": 673
    },
    {
      "epoch": 0.314585764294049,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001995257263633527,
      "loss": 0.7612,
      "step": 674
    },
    {
      "epoch": 0.31505250875145857,
      "grad_norm": 0.53125,
      "learning_rate": 0.00019952429788712474,
      "loss": 0.6959,
      "step": 675
    },
    {
      "epoch": 0.31551925320886814,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019952286726802827,
      "loss": 0.7807,
      "step": 676
    },
    {
      "epoch": 0.3159859976662777,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019952143450609403,
      "loss": 0.7659,
      "step": 677
    },
    {
      "epoch": 0.3164527421236873,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00019951999960135286,
      "loss": 0.69,
      "step": 678
    },
    {
      "epoch": 0.31691948658109687,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019951856255383566,
      "loss": 0.742,
      "step": 679
    },
    {
      "epoch": 0.31738623103850644,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00019951712336357345,
      "loss": 0.7583,
      "step": 680
    },
    {
      "epoch": 0.317852975495916,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001995156820305971,
      "loss": 0.7252,
      "step": 681
    },
    {
      "epoch": 0.31831971995332553,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019951423855493774,
      "loss": 0.746,
      "step": 682
    },
    {
      "epoch": 0.3187864644107351,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001995127929366264,
      "loss": 0.7351,
      "step": 683
    },
    {
      "epoch": 0.3192532088681447,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019951134517569421,
      "loss": 0.6398,
      "step": 684
    },
    {
      "epoch": 0.31971995332555425,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00019950989527217237,
      "loss": 0.5848,
      "step": 685
    },
    {
      "epoch": 0.3201866977829638,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019950844322609205,
      "loss": 0.6832,
      "step": 686
    },
    {
      "epoch": 0.3206534422403734,
      "grad_norm": 0.75,
      "learning_rate": 0.00019950698903748454,
      "loss": 0.8341,
      "step": 687
    },
    {
      "epoch": 0.32112018669778297,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001995055327063812,
      "loss": 0.6931,
      "step": 688
    },
    {
      "epoch": 0.32158693115519255,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001995040742328133,
      "loss": 0.7179,
      "step": 689
    },
    {
      "epoch": 0.3220536756126021,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019950261361681228,
      "loss": 0.7363,
      "step": 690
    },
    {
      "epoch": 0.3225204200700117,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019950115085840957,
      "loss": 0.7609,
      "step": 691
    },
    {
      "epoch": 0.3229871645274212,
      "grad_norm": 0.625,
      "learning_rate": 0.0001994996859576367,
      "loss": 0.776,
      "step": 692
    },
    {
      "epoch": 0.3234539089848308,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019949821891452523,
      "loss": 0.7253,
      "step": 693
    },
    {
      "epoch": 0.32392065344224036,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019949674972910667,
      "loss": 0.7126,
      "step": 694
    },
    {
      "epoch": 0.32438739789964993,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001994952784014127,
      "loss": 0.7645,
      "step": 695
    },
    {
      "epoch": 0.3248541423570595,
      "grad_norm": 0.60546875,
      "learning_rate": 0.000199493804931475,
      "loss": 0.7993,
      "step": 696
    },
    {
      "epoch": 0.3253208868144691,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001994923293193253,
      "loss": 0.6651,
      "step": 697
    },
    {
      "epoch": 0.32578763127187865,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019949085156499533,
      "loss": 0.8008,
      "step": 698
    },
    {
      "epoch": 0.3262543757292882,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019948937166851697,
      "loss": 0.7117,
      "step": 699
    },
    {
      "epoch": 0.3267211201866978,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019948788962992204,
      "loss": 0.7917,
      "step": 700
    },
    {
      "epoch": 0.3271878646441074,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019948640544924245,
      "loss": 0.6921,
      "step": 701
    },
    {
      "epoch": 0.32765460910151695,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001994849191265102,
      "loss": 0.7896,
      "step": 702
    },
    {
      "epoch": 0.32812135355892647,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019948343066175723,
      "loss": 0.7888,
      "step": 703
    },
    {
      "epoch": 0.32858809801633604,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019948194005501562,
      "loss": 0.7162,
      "step": 704
    },
    {
      "epoch": 0.3290548424737456,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001994804473063175,
      "loss": 0.7838,
      "step": 705
    },
    {
      "epoch": 0.3295215869311552,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019947895241569495,
      "loss": 0.7563,
      "step": 706
    },
    {
      "epoch": 0.32998833138856476,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00019947745538318018,
      "loss": 0.556,
      "step": 707
    },
    {
      "epoch": 0.33045507584597433,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019947595620880542,
      "loss": 0.6932,
      "step": 708
    },
    {
      "epoch": 0.3309218203033839,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019947445489260296,
      "loss": 0.8337,
      "step": 709
    },
    {
      "epoch": 0.3313885647607935,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001994729514346051,
      "loss": 0.7472,
      "step": 710
    },
    {
      "epoch": 0.33185530921820305,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019947144583484425,
      "loss": 0.7849,
      "step": 711
    },
    {
      "epoch": 0.3323220536756126,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001994699380933528,
      "loss": 0.8254,
      "step": 712
    },
    {
      "epoch": 0.33278879813302215,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019946842821016322,
      "loss": 0.7681,
      "step": 713
    },
    {
      "epoch": 0.3332555425904317,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00019946691618530804,
      "loss": 0.7437,
      "step": 714
    },
    {
      "epoch": 0.3337222870478413,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019946540201881975,
      "loss": 0.783,
      "step": 715
    },
    {
      "epoch": 0.33418903150525087,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019946388571073103,
      "loss": 0.7149,
      "step": 716
    },
    {
      "epoch": 0.33465577596266044,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00019946236726107447,
      "loss": 0.6135,
      "step": 717
    },
    {
      "epoch": 0.33512252042007,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001994608466698828,
      "loss": 0.705,
      "step": 718
    },
    {
      "epoch": 0.3355892648774796,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019945932393718873,
      "loss": 0.7772,
      "step": 719
    },
    {
      "epoch": 0.33605600933488916,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019945779906302508,
      "loss": 0.7351,
      "step": 720
    },
    {
      "epoch": 0.33652275379229873,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019945627204742466,
      "loss": 0.8505,
      "step": 721
    },
    {
      "epoch": 0.3369894982497083,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019945474289042035,
      "loss": 0.7008,
      "step": 722
    },
    {
      "epoch": 0.3374562427071178,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019945321159204512,
      "loss": 0.7036,
      "step": 723
    },
    {
      "epoch": 0.3379229871645274,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019945167815233186,
      "loss": 0.8273,
      "step": 724
    },
    {
      "epoch": 0.338389731621937,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019945014257131364,
      "loss": 0.6616,
      "step": 725
    },
    {
      "epoch": 0.33885647607934655,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019944860484902348,
      "loss": 0.6831,
      "step": 726
    },
    {
      "epoch": 0.3393232205367561,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00019944706498549453,
      "loss": 0.7227,
      "step": 727
    },
    {
      "epoch": 0.3397899649941657,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00019944552298075996,
      "loss": 0.7427,
      "step": 728
    },
    {
      "epoch": 0.34025670945157527,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001994439788348529,
      "loss": 0.8514,
      "step": 729
    },
    {
      "epoch": 0.34072345390898484,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001994424325478067,
      "loss": 0.6433,
      "step": 730
    },
    {
      "epoch": 0.3411901983663944,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019944088411965452,
      "loss": 0.78,
      "step": 731
    },
    {
      "epoch": 0.341656942823804,
      "grad_norm": 0.625,
      "learning_rate": 0.00019943933355042983,
      "loss": 0.8557,
      "step": 732
    },
    {
      "epoch": 0.34212368728121356,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019943778084016594,
      "loss": 0.7974,
      "step": 733
    },
    {
      "epoch": 0.3425904317386231,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019943622598889632,
      "loss": 0.7358,
      "step": 734
    },
    {
      "epoch": 0.34305717619603265,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019943466899665437,
      "loss": 0.8237,
      "step": 735
    },
    {
      "epoch": 0.3435239206534422,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019943310986347373,
      "loss": 0.6492,
      "step": 736
    },
    {
      "epoch": 0.3439906651108518,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001994315485893879,
      "loss": 0.7173,
      "step": 737
    },
    {
      "epoch": 0.3444574095682614,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019942998517443047,
      "loss": 0.8168,
      "step": 738
    },
    {
      "epoch": 0.34492415402567095,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001994284196186352,
      "loss": 0.6831,
      "step": 739
    },
    {
      "epoch": 0.3453908984830805,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001994268519220357,
      "loss": 0.6027,
      "step": 740
    },
    {
      "epoch": 0.3458576429404901,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019942528208466577,
      "loss": 0.6888,
      "step": 741
    },
    {
      "epoch": 0.34632438739789967,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001994237101065592,
      "loss": 0.771,
      "step": 742
    },
    {
      "epoch": 0.34679113185530924,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019942213598774983,
      "loss": 0.6988,
      "step": 743
    },
    {
      "epoch": 0.34725787631271876,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019942055972827158,
      "loss": 0.7537,
      "step": 744
    },
    {
      "epoch": 0.34772462077012833,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019941898132815836,
      "loss": 0.6787,
      "step": 745
    },
    {
      "epoch": 0.3481913652275379,
      "grad_norm": 0.75,
      "learning_rate": 0.00019941740078744419,
      "loss": 0.8621,
      "step": 746
    },
    {
      "epoch": 0.3486581096849475,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019941581810616304,
      "loss": 0.7556,
      "step": 747
    },
    {
      "epoch": 0.34912485414235706,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019941423328434903,
      "loss": 0.6841,
      "step": 748
    },
    {
      "epoch": 0.34959159859976663,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019941264632203628,
      "loss": 0.7186,
      "step": 749
    },
    {
      "epoch": 0.3500583430571762,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019941105721925896,
      "loss": 0.7302,
      "step": 750
    },
    {
      "epoch": 0.3505250875145858,
      "grad_norm": 0.625,
      "learning_rate": 0.00019940946597605126,
      "loss": 0.794,
      "step": 751
    },
    {
      "epoch": 0.35099183197199535,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001994078725924475,
      "loss": 0.724,
      "step": 752
    },
    {
      "epoch": 0.3514585764294049,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001994062770684819,
      "loss": 0.6629,
      "step": 753
    },
    {
      "epoch": 0.3519253208868145,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001994046794041889,
      "loss": 0.7525,
      "step": 754
    },
    {
      "epoch": 0.352392065344224,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019940307959960283,
      "loss": 0.7591,
      "step": 755
    },
    {
      "epoch": 0.3528588098016336,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019940147765475818,
      "loss": 0.8073,
      "step": 756
    },
    {
      "epoch": 0.35332555425904316,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019939987356968947,
      "loss": 0.8765,
      "step": 757
    },
    {
      "epoch": 0.35379229871645274,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019939826734443114,
      "loss": 0.6879,
      "step": 758
    },
    {
      "epoch": 0.3542590431738623,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019939665897901785,
      "loss": 0.7468,
      "step": 759
    },
    {
      "epoch": 0.3547257876312719,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001993950484734842,
      "loss": 0.7073,
      "step": 760
    },
    {
      "epoch": 0.35519253208868146,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019939343582786487,
      "loss": 0.7915,
      "step": 761
    },
    {
      "epoch": 0.35565927654609103,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019939182104219462,
      "loss": 0.7578,
      "step": 762
    },
    {
      "epoch": 0.3561260210035006,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019939020411650815,
      "loss": 0.7697,
      "step": 763
    },
    {
      "epoch": 0.3565927654609102,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019938858505084032,
      "loss": 0.7542,
      "step": 764
    },
    {
      "epoch": 0.3570595099183197,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019938696384522598,
      "loss": 0.8836,
      "step": 765
    },
    {
      "epoch": 0.35752625437572927,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019938534049970005,
      "loss": 0.7789,
      "step": 766
    },
    {
      "epoch": 0.35799299883313884,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019938371501429747,
      "loss": 0.6261,
      "step": 767
    },
    {
      "epoch": 0.3584597432905484,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019938208738905325,
      "loss": 0.7938,
      "step": 768
    },
    {
      "epoch": 0.358926487747958,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019938045762400236,
      "loss": 0.659,
      "step": 769
    },
    {
      "epoch": 0.35939323220536756,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019937882571918002,
      "loss": 0.6463,
      "step": 770
    },
    {
      "epoch": 0.35985997666277714,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019937719167462126,
      "loss": 0.8015,
      "step": 771
    },
    {
      "epoch": 0.3603267211201867,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00019937555549036132,
      "loss": 0.6138,
      "step": 772
    },
    {
      "epoch": 0.3607934655775963,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001993739171664354,
      "loss": 0.7856,
      "step": 773
    },
    {
      "epoch": 0.36126021003500586,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001993722767028788,
      "loss": 0.7126,
      "step": 774
    },
    {
      "epoch": 0.3617269544924154,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019937063409972684,
      "loss": 0.8082,
      "step": 775
    },
    {
      "epoch": 0.36219369894982495,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00019936898935701486,
      "loss": 0.7173,
      "step": 776
    },
    {
      "epoch": 0.3626604434072345,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019936734247477826,
      "loss": 0.6554,
      "step": 777
    },
    {
      "epoch": 0.3631271878646441,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019936569345305256,
      "loss": 0.6865,
      "step": 778
    },
    {
      "epoch": 0.36359393232205367,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019936404229187323,
      "loss": 0.7052,
      "step": 779
    },
    {
      "epoch": 0.36406067677946324,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001993623889912758,
      "loss": 0.8104,
      "step": 780
    },
    {
      "epoch": 0.3645274212368728,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001993607335512959,
      "loss": 0.6942,
      "step": 781
    },
    {
      "epoch": 0.3649941656942824,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019935907597196917,
      "loss": 0.638,
      "step": 782
    },
    {
      "epoch": 0.36546091015169196,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019935741625333128,
      "loss": 0.8416,
      "step": 783
    },
    {
      "epoch": 0.36592765460910154,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000199355754395418,
      "loss": 0.7029,
      "step": 784
    },
    {
      "epoch": 0.3663943990665111,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001993540903982651,
      "loss": 0.8415,
      "step": 785
    },
    {
      "epoch": 0.36686114352392063,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019935242426190837,
      "loss": 0.7105,
      "step": 786
    },
    {
      "epoch": 0.3673278879813302,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019935075598638373,
      "loss": 0.8521,
      "step": 787
    },
    {
      "epoch": 0.3677946324387398,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019934908557172704,
      "loss": 0.796,
      "step": 788
    },
    {
      "epoch": 0.36826137689614935,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019934741301797434,
      "loss": 0.6624,
      "step": 789
    },
    {
      "epoch": 0.3687281213535589,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019934573832516164,
      "loss": 0.7359,
      "step": 790
    },
    {
      "epoch": 0.3691948658109685,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019934406149332495,
      "loss": 0.5849,
      "step": 791
    },
    {
      "epoch": 0.36966161026837807,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019934238252250036,
      "loss": 0.7446,
      "step": 792
    },
    {
      "epoch": 0.37012835472578764,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019934070141272407,
      "loss": 0.6786,
      "step": 793
    },
    {
      "epoch": 0.3705950991831972,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019933901816403227,
      "loss": 0.7116,
      "step": 794
    },
    {
      "epoch": 0.3710618436406068,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019933733277646118,
      "loss": 0.7427,
      "step": 795
    },
    {
      "epoch": 0.3715285880980163,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001993356452500471,
      "loss": 0.6392,
      "step": 796
    },
    {
      "epoch": 0.3719953325554259,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019933395558482638,
      "loss": 0.7705,
      "step": 797
    },
    {
      "epoch": 0.37246207701283546,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019933226378083537,
      "loss": 0.7715,
      "step": 798
    },
    {
      "epoch": 0.37292882147024503,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019933056983811049,
      "loss": 0.6703,
      "step": 799
    },
    {
      "epoch": 0.3733955659276546,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019932887375668826,
      "loss": 0.6387,
      "step": 800
    },
    {
      "epoch": 0.3738623103850642,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019932717553660515,
      "loss": 0.7242,
      "step": 801
    },
    {
      "epoch": 0.37432905484247375,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019932547517789777,
      "loss": 0.7549,
      "step": 802
    },
    {
      "epoch": 0.3747957992998833,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019932377268060266,
      "loss": 0.7823,
      "step": 803
    },
    {
      "epoch": 0.3752625437572929,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019932206804475658,
      "loss": 0.689,
      "step": 804
    },
    {
      "epoch": 0.3757292882147025,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019932036127039615,
      "loss": 0.7811,
      "step": 805
    },
    {
      "epoch": 0.37619603267211205,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00019931865235755814,
      "loss": 0.5949,
      "step": 806
    },
    {
      "epoch": 0.37666277712952156,
      "grad_norm": 0.546875,
      "learning_rate": 0.00019931694130627934,
      "loss": 0.6188,
      "step": 807
    },
    {
      "epoch": 0.37712952158693114,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019931522811659661,
      "loss": 0.6804,
      "step": 808
    },
    {
      "epoch": 0.3775962660443407,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019931351278854685,
      "loss": 0.6732,
      "step": 809
    },
    {
      "epoch": 0.3780630105017503,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019931179532216696,
      "loss": 0.6789,
      "step": 810
    },
    {
      "epoch": 0.37852975495915986,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001993100757174939,
      "loss": 0.6879,
      "step": 811
    },
    {
      "epoch": 0.37899649941656943,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019930835397456473,
      "loss": 0.6632,
      "step": 812
    },
    {
      "epoch": 0.379463243873979,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019930663009341652,
      "loss": 0.57,
      "step": 813
    },
    {
      "epoch": 0.3799299883313886,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001993049040740864,
      "loss": 0.8101,
      "step": 814
    },
    {
      "epoch": 0.38039673278879815,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001993031759166115,
      "loss": 0.7654,
      "step": 815
    },
    {
      "epoch": 0.3808634772462077,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019930144562102904,
      "loss": 0.7981,
      "step": 816
    },
    {
      "epoch": 0.38133022170361724,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001992997131873763,
      "loss": 0.7095,
      "step": 817
    },
    {
      "epoch": 0.3817969661610268,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019929797861569052,
      "loss": 0.7036,
      "step": 818
    },
    {
      "epoch": 0.3822637106184364,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001992962419060091,
      "loss": 0.7372,
      "step": 819
    },
    {
      "epoch": 0.38273045507584597,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019929450305836946,
      "loss": 0.7175,
      "step": 820
    },
    {
      "epoch": 0.38319719953325554,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00019929276207280894,
      "loss": 0.5719,
      "step": 821
    },
    {
      "epoch": 0.3836639439906651,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00019929101894936512,
      "loss": 0.6493,
      "step": 822
    },
    {
      "epoch": 0.3841306884480747,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001992892736880755,
      "loss": 0.7137,
      "step": 823
    },
    {
      "epoch": 0.38459743290548426,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019928752628897765,
      "loss": 0.8295,
      "step": 824
    },
    {
      "epoch": 0.38506417736289383,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001992857767521092,
      "loss": 0.7852,
      "step": 825
    },
    {
      "epoch": 0.3855309218203034,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00019928402507750782,
      "loss": 0.6636,
      "step": 826
    },
    {
      "epoch": 0.3859976662777129,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001992822712652112,
      "loss": 0.6628,
      "step": 827
    },
    {
      "epoch": 0.3864644107351225,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019928051531525713,
      "loss": 0.754,
      "step": 828
    },
    {
      "epoch": 0.3869311551925321,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019927875722768344,
      "loss": 0.699,
      "step": 829
    },
    {
      "epoch": 0.38739789964994165,
      "grad_norm": 0.625,
      "learning_rate": 0.00019927699700252792,
      "loss": 0.7438,
      "step": 830
    },
    {
      "epoch": 0.3878646441073512,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019927523463982851,
      "loss": 0.6645,
      "step": 831
    },
    {
      "epoch": 0.3883313885647608,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001992734701396232,
      "loss": 0.7631,
      "step": 832
    },
    {
      "epoch": 0.38879813302217037,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001992717035019499,
      "loss": 0.734,
      "step": 833
    },
    {
      "epoch": 0.38926487747957994,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019926993472684664,
      "loss": 0.7883,
      "step": 834
    },
    {
      "epoch": 0.3897316219369895,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019926816381435157,
      "loss": 0.6986,
      "step": 835
    },
    {
      "epoch": 0.3901983663943991,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019926639076450279,
      "loss": 0.7749,
      "step": 836
    },
    {
      "epoch": 0.39066511085180866,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001992646155773385,
      "loss": 0.7334,
      "step": 837
    },
    {
      "epoch": 0.3911318553092182,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019926283825289688,
      "loss": 0.6619,
      "step": 838
    },
    {
      "epoch": 0.39159859976662775,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019926105879121623,
      "loss": 0.7252,
      "step": 839
    },
    {
      "epoch": 0.3920653442240373,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019925927719233487,
      "loss": 0.7034,
      "step": 840
    },
    {
      "epoch": 0.3925320886814469,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001992574934562911,
      "loss": 0.9161,
      "step": 841
    },
    {
      "epoch": 0.3929988331388565,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001992557075831234,
      "loss": 0.907,
      "step": 842
    },
    {
      "epoch": 0.39346557759626605,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019925391957287017,
      "loss": 0.7525,
      "step": 843
    },
    {
      "epoch": 0.3939323220536756,
      "grad_norm": 0.625,
      "learning_rate": 0.00019925212942556995,
      "loss": 0.7048,
      "step": 844
    },
    {
      "epoch": 0.3943990665110852,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019925033714126124,
      "loss": 0.6646,
      "step": 845
    },
    {
      "epoch": 0.39486581096849477,
      "grad_norm": 0.5625,
      "learning_rate": 0.00019924854271998266,
      "loss": 0.6329,
      "step": 846
    },
    {
      "epoch": 0.39533255542590434,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019924674616177281,
      "loss": 0.7151,
      "step": 847
    },
    {
      "epoch": 0.39579929988331386,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019924494746667045,
      "loss": 0.7011,
      "step": 848
    },
    {
      "epoch": 0.39626604434072343,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001992431466347142,
      "loss": 0.7077,
      "step": 849
    },
    {
      "epoch": 0.396732788798133,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019924134366594294,
      "loss": 0.7014,
      "step": 850
    },
    {
      "epoch": 0.3971995332555426,
      "grad_norm": 0.75,
      "learning_rate": 0.00019923953856039543,
      "loss": 0.7786,
      "step": 851
    },
    {
      "epoch": 0.39766627771295215,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019923773131811054,
      "loss": 0.7086,
      "step": 852
    },
    {
      "epoch": 0.39813302217036173,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001992359219391272,
      "loss": 0.6755,
      "step": 853
    },
    {
      "epoch": 0.3985997666277713,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019923411042348433,
      "loss": 0.7066,
      "step": 854
    },
    {
      "epoch": 0.3990665110851809,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019923229677122097,
      "loss": 0.8076,
      "step": 855
    },
    {
      "epoch": 0.39953325554259045,
      "grad_norm": 0.75,
      "learning_rate": 0.0001992304809823762,
      "loss": 0.7037,
      "step": 856
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.59375,
      "learning_rate": 0.000199228663056989,
      "loss": 0.6435,
      "step": 857
    },
    {
      "epoch": 0.4004667444574096,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019922684299509863,
      "loss": 0.7999,
      "step": 858
    },
    {
      "epoch": 0.4009334889148191,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019922502079674424,
      "loss": 0.6508,
      "step": 859
    },
    {
      "epoch": 0.4014002333722287,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019922319646196505,
      "loss": 0.768,
      "step": 860
    },
    {
      "epoch": 0.40186697782963826,
      "grad_norm": 0.75,
      "learning_rate": 0.00019922136999080036,
      "loss": 0.788,
      "step": 861
    },
    {
      "epoch": 0.40233372228704783,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019921954138328948,
      "loss": 0.7233,
      "step": 862
    },
    {
      "epoch": 0.4028004667444574,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001992177106394718,
      "loss": 0.7035,
      "step": 863
    },
    {
      "epoch": 0.403267211201867,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001992158777593867,
      "loss": 0.7538,
      "step": 864
    },
    {
      "epoch": 0.40373395565927656,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019921404274307372,
      "loss": 0.5887,
      "step": 865
    },
    {
      "epoch": 0.40420070011668613,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019921220559057225,
      "loss": 0.9041,
      "step": 866
    },
    {
      "epoch": 0.4046674445740957,
      "grad_norm": 0.72265625,
      "learning_rate": 0.000199210366301922,
      "loss": 0.7397,
      "step": 867
    },
    {
      "epoch": 0.4051341890315053,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001992085248771624,
      "loss": 0.7871,
      "step": 868
    },
    {
      "epoch": 0.4056009334889148,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019920668131633322,
      "loss": 0.6565,
      "step": 869
    },
    {
      "epoch": 0.40606767794632437,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019920483561947414,
      "loss": 0.7352,
      "step": 870
    },
    {
      "epoch": 0.40653442240373394,
      "grad_norm": 0.625,
      "learning_rate": 0.00019920298778662487,
      "loss": 0.6845,
      "step": 871
    },
    {
      "epoch": 0.4070011668611435,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001992011378178252,
      "loss": 0.691,
      "step": 872
    },
    {
      "epoch": 0.4074679113185531,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019919928571311495,
      "loss": 0.6989,
      "step": 873
    },
    {
      "epoch": 0.40793465577596266,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019919743147253405,
      "loss": 0.683,
      "step": 874
    },
    {
      "epoch": 0.40840140023337224,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019919557509612236,
      "loss": 0.5746,
      "step": 875
    },
    {
      "epoch": 0.4088681446907818,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001991937165839199,
      "loss": 0.7188,
      "step": 876
    },
    {
      "epoch": 0.4093348891481914,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00019919185593596662,
      "loss": 0.5876,
      "step": 877
    },
    {
      "epoch": 0.40980163360560096,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001991899931523027,
      "loss": 0.7175,
      "step": 878
    },
    {
      "epoch": 0.4102683780630105,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019918812823296813,
      "loss": 0.6975,
      "step": 879
    },
    {
      "epoch": 0.41073512252042005,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001991862611780031,
      "loss": 0.674,
      "step": 880
    },
    {
      "epoch": 0.4112018669778296,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019918439198744783,
      "loss": 0.6851,
      "step": 881
    },
    {
      "epoch": 0.4116686114352392,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019918252066134255,
      "loss": 0.7126,
      "step": 882
    },
    {
      "epoch": 0.41213535589264877,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019918064719972757,
      "loss": 0.6733,
      "step": 883
    },
    {
      "epoch": 0.41260210035005834,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001991787716026432,
      "loss": 0.5945,
      "step": 884
    },
    {
      "epoch": 0.4130688448074679,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019917689387012984,
      "loss": 0.7451,
      "step": 885
    },
    {
      "epoch": 0.4135355892648775,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001991750140022279,
      "loss": 0.8363,
      "step": 886
    },
    {
      "epoch": 0.41400233372228706,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019917313199897788,
      "loss": 0.6196,
      "step": 887
    },
    {
      "epoch": 0.41446907817969664,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001991712478604203,
      "loss": 0.6753,
      "step": 888
    },
    {
      "epoch": 0.4149358226371062,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019916936158659572,
      "loss": 0.7073,
      "step": 889
    },
    {
      "epoch": 0.41540256709451573,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001991674731775448,
      "loss": 0.6567,
      "step": 890
    },
    {
      "epoch": 0.4158693115519253,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001991655826333081,
      "loss": 0.73,
      "step": 891
    },
    {
      "epoch": 0.4163360560093349,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019916368995392638,
      "loss": 0.7652,
      "step": 892
    },
    {
      "epoch": 0.41680280046674445,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001991617951394404,
      "loss": 0.654,
      "step": 893
    },
    {
      "epoch": 0.417269544924154,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019915989818989095,
      "loss": 0.6455,
      "step": 894
    },
    {
      "epoch": 0.4177362893815636,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00019915799910531887,
      "loss": 0.6531,
      "step": 895
    },
    {
      "epoch": 0.41820303383897317,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019915609788576506,
      "loss": 0.6891,
      "step": 896
    },
    {
      "epoch": 0.41866977829638274,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019915419453127046,
      "loss": 0.5585,
      "step": 897
    },
    {
      "epoch": 0.4191365227537923,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019915228904187598,
      "loss": 0.6863,
      "step": 898
    },
    {
      "epoch": 0.4196032672112019,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019915038141762276,
      "loss": 0.7216,
      "step": 899
    },
    {
      "epoch": 0.4200700116686114,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019914847165855183,
      "loss": 0.6804,
      "step": 900
    },
    {
      "epoch": 0.420536756126021,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00019914655976470424,
      "loss": 0.5742,
      "step": 901
    },
    {
      "epoch": 0.42100350058343056,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00019914464573612126,
      "loss": 0.584,
      "step": 902
    },
    {
      "epoch": 0.42147024504084013,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019914272957284403,
      "loss": 0.6384,
      "step": 903
    },
    {
      "epoch": 0.4219369894982497,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019914081127491383,
      "loss": 0.6122,
      "step": 904
    },
    {
      "epoch": 0.4224037339556593,
      "grad_norm": 0.6953125,
      "learning_rate": 0.000199138890842372,
      "loss": 0.642,
      "step": 905
    },
    {
      "epoch": 0.42287047841306885,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019913696827525982,
      "loss": 0.637,
      "step": 906
    },
    {
      "epoch": 0.4233372228704784,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019913504357361875,
      "loss": 0.6106,
      "step": 907
    },
    {
      "epoch": 0.423803967327888,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001991331167374902,
      "loss": 0.7521,
      "step": 908
    },
    {
      "epoch": 0.42427071178529757,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019913118776691558,
      "loss": 0.7924,
      "step": 909
    },
    {
      "epoch": 0.42473745624270715,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001991292566619366,
      "loss": 0.6847,
      "step": 910
    },
    {
      "epoch": 0.42520420070011666,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019912732342259467,
      "loss": 0.5853,
      "step": 911
    },
    {
      "epoch": 0.42567094515752624,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001991253880489315,
      "loss": 0.7638,
      "step": 912
    },
    {
      "epoch": 0.4261376896149358,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019912345054098876,
      "loss": 0.7482,
      "step": 913
    },
    {
      "epoch": 0.4266044340723454,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019912151089880812,
      "loss": 0.6919,
      "step": 914
    },
    {
      "epoch": 0.42707117852975496,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001991195691224314,
      "loss": 0.6083,
      "step": 915
    },
    {
      "epoch": 0.42753792298716453,
      "grad_norm": 0.625,
      "learning_rate": 0.00019911762521190038,
      "loss": 0.6483,
      "step": 916
    },
    {
      "epoch": 0.4280046674445741,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019911567916725692,
      "loss": 0.6598,
      "step": 917
    },
    {
      "epoch": 0.4284714119019837,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001991137309885429,
      "loss": 0.6888,
      "step": 918
    },
    {
      "epoch": 0.42893815635939325,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019911178067580026,
      "loss": 0.7007,
      "step": 919
    },
    {
      "epoch": 0.4294049008168028,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019910982822907107,
      "loss": 0.6844,
      "step": 920
    },
    {
      "epoch": 0.42987164527421234,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001991078736483973,
      "loss": 0.8368,
      "step": 921
    },
    {
      "epoch": 0.4303383897316219,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019910591693382104,
      "loss": 0.7621,
      "step": 922
    },
    {
      "epoch": 0.4308051341890315,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019910395808538442,
      "loss": 0.6159,
      "step": 923
    },
    {
      "epoch": 0.43127187864644106,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00019910199710312964,
      "loss": 0.6782,
      "step": 924
    },
    {
      "epoch": 0.43173862310385064,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00019910003398709888,
      "loss": 0.6109,
      "step": 925
    },
    {
      "epoch": 0.4322053675612602,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019909806873733447,
      "loss": 0.6611,
      "step": 926
    },
    {
      "epoch": 0.4326721120186698,
      "grad_norm": 0.625,
      "learning_rate": 0.00019909610135387867,
      "loss": 0.6061,
      "step": 927
    },
    {
      "epoch": 0.43313885647607936,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019909413183677387,
      "loss": 0.6799,
      "step": 928
    },
    {
      "epoch": 0.43360560093348893,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019909216018606244,
      "loss": 0.7896,
      "step": 929
    },
    {
      "epoch": 0.4340723453908985,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019909018640178688,
      "loss": 0.6194,
      "step": 930
    },
    {
      "epoch": 0.434539089848308,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019908821048398966,
      "loss": 0.6702,
      "step": 931
    },
    {
      "epoch": 0.4350058343057176,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001990862324327133,
      "loss": 0.7371,
      "step": 932
    },
    {
      "epoch": 0.43547257876312717,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019908425224800043,
      "loss": 0.6706,
      "step": 933
    },
    {
      "epoch": 0.43593932322053675,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001990822699298937,
      "loss": 0.5697,
      "step": 934
    },
    {
      "epoch": 0.4364060676779463,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019908028547843576,
      "loss": 0.6944,
      "step": 935
    },
    {
      "epoch": 0.4368728121353559,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019907829889366933,
      "loss": 0.6388,
      "step": 936
    },
    {
      "epoch": 0.43733955659276547,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019907631017563723,
      "loss": 0.7356,
      "step": 937
    },
    {
      "epoch": 0.43780630105017504,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001990743193243822,
      "loss": 0.6149,
      "step": 938
    },
    {
      "epoch": 0.4382730455075846,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019907232633994715,
      "loss": 0.7016,
      "step": 939
    },
    {
      "epoch": 0.4387397899649942,
      "grad_norm": 0.6328125,
      "learning_rate": 0.000199070331222375,
      "loss": 0.6267,
      "step": 940
    },
    {
      "epoch": 0.43920653442240376,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001990683339717087,
      "loss": 0.6472,
      "step": 941
    },
    {
      "epoch": 0.4396732788798133,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019906633458799126,
      "loss": 0.7162,
      "step": 942
    },
    {
      "epoch": 0.44014002333722285,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00019906433307126573,
      "loss": 0.5808,
      "step": 943
    },
    {
      "epoch": 0.4406067677946324,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019906232942157515,
      "loss": 0.7619,
      "step": 944
    },
    {
      "epoch": 0.441073512252042,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019906032363896275,
      "loss": 0.6483,
      "step": 945
    },
    {
      "epoch": 0.4415402567094516,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019905831572347165,
      "loss": 0.6812,
      "step": 946
    },
    {
      "epoch": 0.44200700116686115,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019905630567514513,
      "loss": 0.6345,
      "step": 947
    },
    {
      "epoch": 0.4424737456242707,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001990542934940264,
      "loss": 0.7125,
      "step": 948
    },
    {
      "epoch": 0.4429404900816803,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019905227918015888,
      "loss": 0.6523,
      "step": 949
    },
    {
      "epoch": 0.44340723453908987,
      "grad_norm": 0.671875,
      "learning_rate": 0.00019905026273358588,
      "loss": 0.6574,
      "step": 950
    },
    {
      "epoch": 0.44387397899649944,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001990482441543508,
      "loss": 0.611,
      "step": 951
    },
    {
      "epoch": 0.44434072345390896,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019904622344249712,
      "loss": 0.6512,
      "step": 952
    },
    {
      "epoch": 0.44480746791131853,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019904420059806838,
      "loss": 0.7811,
      "step": 953
    },
    {
      "epoch": 0.4452742123687281,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001990421756211081,
      "loss": 0.6213,
      "step": 954
    },
    {
      "epoch": 0.4457409568261377,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019904014851165988,
      "loss": 0.7291,
      "step": 955
    },
    {
      "epoch": 0.44620770128354725,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019903811926976742,
      "loss": 0.7575,
      "step": 956
    },
    {
      "epoch": 0.4466744457409568,
      "grad_norm": 0.59375,
      "learning_rate": 0.00019903608789547433,
      "loss": 0.6126,
      "step": 957
    },
    {
      "epoch": 0.4471411901983664,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001990340543888244,
      "loss": 0.8007,
      "step": 958
    },
    {
      "epoch": 0.447607934655776,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00019903201874986137,
      "loss": 0.636,
      "step": 959
    },
    {
      "epoch": 0.44807467911318555,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019902998097862912,
      "loss": 0.7971,
      "step": 960
    },
    {
      "epoch": 0.4485414235705951,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019902794107517152,
      "loss": 0.5808,
      "step": 961
    },
    {
      "epoch": 0.4490081680280047,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019902589903953244,
      "loss": 0.6067,
      "step": 962
    },
    {
      "epoch": 0.4494749124854142,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001990238548717559,
      "loss": 0.7232,
      "step": 963
    },
    {
      "epoch": 0.4499416569428238,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019902180857188587,
      "loss": 0.6778,
      "step": 964
    },
    {
      "epoch": 0.45040840140023336,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019901976013996647,
      "loss": 0.6538,
      "step": 965
    },
    {
      "epoch": 0.45087514585764293,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019901770957604179,
      "loss": 0.63,
      "step": 966
    },
    {
      "epoch": 0.4513418903150525,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001990156568801559,
      "loss": 0.5631,
      "step": 967
    },
    {
      "epoch": 0.4518086347724621,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019901360205235312,
      "loss": 0.6583,
      "step": 968
    },
    {
      "epoch": 0.45227537922987165,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001990115450926776,
      "loss": 0.6951,
      "step": 969
    },
    {
      "epoch": 0.45274212368728123,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019900948600117365,
      "loss": 0.8222,
      "step": 970
    },
    {
      "epoch": 0.4532088681446908,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001990074247778856,
      "loss": 0.6175,
      "step": 971
    },
    {
      "epoch": 0.4536756126021004,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019900536142285788,
      "loss": 0.7423,
      "step": 972
    },
    {
      "epoch": 0.4541423570595099,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00019900329593613484,
      "loss": 0.6104,
      "step": 973
    },
    {
      "epoch": 0.45460910151691947,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019900122831776105,
      "loss": 0.6204,
      "step": 974
    },
    {
      "epoch": 0.45507584597432904,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019899915856778092,
      "loss": 0.65,
      "step": 975
    },
    {
      "epoch": 0.4555425904317386,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001989970866862391,
      "loss": 0.7242,
      "step": 976
    },
    {
      "epoch": 0.4560093348891482,
      "grad_norm": 0.578125,
      "learning_rate": 0.00019899501267318015,
      "loss": 0.613,
      "step": 977
    },
    {
      "epoch": 0.45647607934655776,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00019899293652864873,
      "loss": 0.6448,
      "step": 978
    },
    {
      "epoch": 0.45694282380396734,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019899085825268956,
      "loss": 0.6587,
      "step": 979
    },
    {
      "epoch": 0.4574095682613769,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019898877784534736,
      "loss": 0.6648,
      "step": 980
    },
    {
      "epoch": 0.4578763127187865,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000198986695306667,
      "loss": 0.624,
      "step": 981
    },
    {
      "epoch": 0.45834305717619606,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001989846106366932,
      "loss": 0.6782,
      "step": 982
    },
    {
      "epoch": 0.4588098016336056,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00019898252383547093,
      "loss": 0.6257,
      "step": 983
    },
    {
      "epoch": 0.45927654609101515,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001989804349030451,
      "loss": 0.6897,
      "step": 984
    },
    {
      "epoch": 0.4597432905484247,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001989783438394607,
      "loss": 0.6195,
      "step": 985
    },
    {
      "epoch": 0.4602100350058343,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019897625064476272,
      "loss": 0.6427,
      "step": 986
    },
    {
      "epoch": 0.46067677946324387,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019897415531899623,
      "loss": 0.6818,
      "step": 987
    },
    {
      "epoch": 0.46114352392065344,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001989720578622064,
      "loss": 0.6442,
      "step": 988
    },
    {
      "epoch": 0.461610268378063,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019896995827443833,
      "loss": 0.5443,
      "step": 989
    },
    {
      "epoch": 0.4620770128354726,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019896785655573726,
      "loss": 0.7202,
      "step": 990
    },
    {
      "epoch": 0.46254375729288216,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019896575270614844,
      "loss": 0.7193,
      "step": 991
    },
    {
      "epoch": 0.46301050175029174,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019896364672571714,
      "loss": 0.692,
      "step": 992
    },
    {
      "epoch": 0.4634772462077013,
      "grad_norm": 0.75,
      "learning_rate": 0.00019896153861448875,
      "loss": 0.7826,
      "step": 993
    },
    {
      "epoch": 0.46394399066511083,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001989594283725086,
      "loss": 0.5714,
      "step": 994
    },
    {
      "epoch": 0.4644107351225204,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001989573159998222,
      "loss": 0.7018,
      "step": 995
    },
    {
      "epoch": 0.46487747957993,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019895520149647495,
      "loss": 0.6572,
      "step": 996
    },
    {
      "epoch": 0.46534422403733955,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019895308486251242,
      "loss": 0.7123,
      "step": 997
    },
    {
      "epoch": 0.4658109684947491,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019895096609798022,
      "loss": 0.5941,
      "step": 998
    },
    {
      "epoch": 0.4662777129521587,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001989488452029239,
      "loss": 0.5684,
      "step": 999
    },
    {
      "epoch": 0.46674445740956827,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019894672217738915,
      "loss": 0.6617,
      "step": 1000
    },
    {
      "epoch": 0.46721120186697784,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019894459702142175,
      "loss": 0.7052,
      "step": 1001
    },
    {
      "epoch": 0.4676779463243874,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019894246973506735,
      "loss": 0.6782,
      "step": 1002
    },
    {
      "epoch": 0.468144690781797,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00019894034031837178,
      "loss": 0.6076,
      "step": 1003
    },
    {
      "epoch": 0.4686114352392065,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019893820877138095,
      "loss": 0.6817,
      "step": 1004
    },
    {
      "epoch": 0.4690781796966161,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019893607509414068,
      "loss": 0.6389,
      "step": 1005
    },
    {
      "epoch": 0.46954492415402566,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019893393928669694,
      "loss": 0.6889,
      "step": 1006
    },
    {
      "epoch": 0.47001166861143523,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019893180134909573,
      "loss": 0.4746,
      "step": 1007
    },
    {
      "epoch": 0.4704784130688448,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001989296612813831,
      "loss": 0.6132,
      "step": 1008
    },
    {
      "epoch": 0.4709451575262544,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019892751908360504,
      "loss": 0.6687,
      "step": 1009
    },
    {
      "epoch": 0.47141190198366395,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019892537475580774,
      "loss": 0.562,
      "step": 1010
    },
    {
      "epoch": 0.4718786464410735,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001989232282980374,
      "loss": 0.6405,
      "step": 1011
    },
    {
      "epoch": 0.4723453908984831,
      "grad_norm": 0.625,
      "learning_rate": 0.00019892107971034018,
      "loss": 0.5299,
      "step": 1012
    },
    {
      "epoch": 0.47281213535589267,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019891892899276235,
      "loss": 0.6425,
      "step": 1013
    },
    {
      "epoch": 0.47327887981330224,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019891677614535024,
      "loss": 0.7187,
      "step": 1014
    },
    {
      "epoch": 0.47374562427071176,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001989146211681502,
      "loss": 0.6603,
      "step": 1015
    },
    {
      "epoch": 0.47421236872812134,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019891246406120858,
      "loss": 0.6571,
      "step": 1016
    },
    {
      "epoch": 0.4746791131855309,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001989103048245719,
      "loss": 0.6117,
      "step": 1017
    },
    {
      "epoch": 0.4751458576429405,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001989081434582866,
      "loss": 0.6603,
      "step": 1018
    },
    {
      "epoch": 0.47561260210035006,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001989059799623992,
      "loss": 0.6893,
      "step": 1019
    },
    {
      "epoch": 0.47607934655775963,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019890381433695638,
      "loss": 0.68,
      "step": 1020
    },
    {
      "epoch": 0.4765460910151692,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019890164658200465,
      "loss": 0.66,
      "step": 1021
    },
    {
      "epoch": 0.4770128354725788,
      "grad_norm": 0.640625,
      "learning_rate": 0.00019889947669759074,
      "loss": 0.6105,
      "step": 1022
    },
    {
      "epoch": 0.47747957992998835,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001988973046837614,
      "loss": 0.681,
      "step": 1023
    },
    {
      "epoch": 0.4779463243873979,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019889513054056332,
      "loss": 0.6251,
      "step": 1024
    },
    {
      "epoch": 0.47841306884480744,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019889295426804338,
      "loss": 0.6463,
      "step": 1025
    },
    {
      "epoch": 0.478879813302217,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001988907758662484,
      "loss": 0.7035,
      "step": 1026
    },
    {
      "epoch": 0.4793465577596266,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001988885953352253,
      "loss": 0.6776,
      "step": 1027
    },
    {
      "epoch": 0.47981330221703616,
      "grad_norm": 0.75,
      "learning_rate": 0.000198886412675021,
      "loss": 0.6199,
      "step": 1028
    },
    {
      "epoch": 0.48028004667444574,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019888422788568258,
      "loss": 0.6544,
      "step": 1029
    },
    {
      "epoch": 0.4807467911318553,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019888204096725696,
      "loss": 0.7016,
      "step": 1030
    },
    {
      "epoch": 0.4812135355892649,
      "grad_norm": 0.75,
      "learning_rate": 0.0001988798519197913,
      "loss": 0.587,
      "step": 1031
    },
    {
      "epoch": 0.48168028004667446,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019887766074333274,
      "loss": 0.5552,
      "step": 1032
    },
    {
      "epoch": 0.48214702450408403,
      "grad_norm": 0.875,
      "learning_rate": 0.00019887546743792842,
      "loss": 0.7311,
      "step": 1033
    },
    {
      "epoch": 0.4826137689614936,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001988732720036256,
      "loss": 0.6107,
      "step": 1034
    },
    {
      "epoch": 0.4830805134189031,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019887107444047151,
      "loss": 0.5597,
      "step": 1035
    },
    {
      "epoch": 0.4835472578763127,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019886887474851347,
      "loss": 0.6131,
      "step": 1036
    },
    {
      "epoch": 0.48401400233372227,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001988666729277989,
      "loss": 0.6671,
      "step": 1037
    },
    {
      "epoch": 0.48448074679113184,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019886446897837514,
      "loss": 0.7221,
      "step": 1038
    },
    {
      "epoch": 0.4849474912485414,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019886226290028968,
      "loss": 0.7206,
      "step": 1039
    },
    {
      "epoch": 0.485414235705951,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019886005469359002,
      "loss": 0.6285,
      "step": 1040
    },
    {
      "epoch": 0.48588098016336057,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001988578443583237,
      "loss": 0.617,
      "step": 1041
    },
    {
      "epoch": 0.48634772462077014,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019885563189453827,
      "loss": 0.5879,
      "step": 1042
    },
    {
      "epoch": 0.4868144690781797,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019885341730228144,
      "loss": 0.6629,
      "step": 1043
    },
    {
      "epoch": 0.4872812135355893,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019885120058160083,
      "loss": 0.5798,
      "step": 1044
    },
    {
      "epoch": 0.48774795799299886,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001988489817325442,
      "loss": 0.7046,
      "step": 1045
    },
    {
      "epoch": 0.4882147024504084,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019884676075515936,
      "loss": 0.7055,
      "step": 1046
    },
    {
      "epoch": 0.48868144690781795,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000198844537649494,
      "loss": 0.6404,
      "step": 1047
    },
    {
      "epoch": 0.4891481913652275,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019884231241559618,
      "loss": 0.6823,
      "step": 1048
    },
    {
      "epoch": 0.4896149358226371,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019884008505351367,
      "loss": 0.4446,
      "step": 1049
    },
    {
      "epoch": 0.49008168028004667,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001988378555632944,
      "loss": 0.5704,
      "step": 1050
    },
    {
      "epoch": 0.49054842473745625,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001988356239449865,
      "loss": 0.6071,
      "step": 1051
    },
    {
      "epoch": 0.4910151691948658,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019883339019863793,
      "loss": 0.673,
      "step": 1052
    },
    {
      "epoch": 0.4914819136522754,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019883115432429684,
      "loss": 0.6836,
      "step": 1053
    },
    {
      "epoch": 0.49194865810968497,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019882891632201134,
      "loss": 0.487,
      "step": 1054
    },
    {
      "epoch": 0.49241540256709454,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001988266761918296,
      "loss": 0.6445,
      "step": 1055
    },
    {
      "epoch": 0.49288214702450406,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019882443393379987,
      "loss": 0.4923,
      "step": 1056
    },
    {
      "epoch": 0.49334889148191363,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019882218954797045,
      "loss": 0.5783,
      "step": 1057
    },
    {
      "epoch": 0.4938156359393232,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019881994303438963,
      "loss": 0.687,
      "step": 1058
    },
    {
      "epoch": 0.4942823803967328,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001988176943931058,
      "loss": 0.6454,
      "step": 1059
    },
    {
      "epoch": 0.49474912485414235,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019881544362416735,
      "loss": 0.677,
      "step": 1060
    },
    {
      "epoch": 0.4952158693115519,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019881319072762276,
      "loss": 0.6048,
      "step": 1061
    },
    {
      "epoch": 0.4956826137689615,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019881093570352055,
      "loss": 0.6567,
      "step": 1062
    },
    {
      "epoch": 0.4961493582263711,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019880867855190927,
      "loss": 0.7238,
      "step": 1063
    },
    {
      "epoch": 0.49661610268378065,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019880641927283752,
      "loss": 0.6651,
      "step": 1064
    },
    {
      "epoch": 0.4970828471411902,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019880415786635393,
      "loss": 0.6228,
      "step": 1065
    },
    {
      "epoch": 0.4975495915985998,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001988018943325072,
      "loss": 0.6091,
      "step": 1066
    },
    {
      "epoch": 0.4980163360560093,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019879962867134604,
      "loss": 0.6545,
      "step": 1067
    },
    {
      "epoch": 0.4984830805134189,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019879736088291924,
      "loss": 0.8203,
      "step": 1068
    },
    {
      "epoch": 0.49894982497082846,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001987950909672757,
      "loss": 0.6723,
      "step": 1069
    },
    {
      "epoch": 0.49941656942823803,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001987928189244642,
      "loss": 0.6822,
      "step": 1070
    },
    {
      "epoch": 0.4998833138856476,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001987905447545337,
      "loss": 0.5764,
      "step": 1071
    },
    {
      "epoch": 0.5003500583430571,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019878826845753315,
      "loss": 0.6364,
      "step": 1072
    },
    {
      "epoch": 0.5003500583430571,
      "eval_loss": 1.0250318050384521,
      "eval_runtime": 94.4783,
      "eval_samples_per_second": 19.094,
      "eval_steps_per_second": 2.392,
      "step": 1072
    },
    {
      "epoch": 0.5008168028004667,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019878599003351157,
      "loss": 0.5955,
      "step": 1073
    },
    {
      "epoch": 0.5012835472578763,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019878370948251805,
      "loss": 0.4872,
      "step": 1074
    },
    {
      "epoch": 0.5017502917152858,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019878142680460162,
      "loss": 0.5654,
      "step": 1075
    },
    {
      "epoch": 0.5022170361726954,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019877914199981152,
      "loss": 0.5411,
      "step": 1076
    },
    {
      "epoch": 0.502683780630105,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019877685506819687,
      "loss": 0.5922,
      "step": 1077
    },
    {
      "epoch": 0.5031505250875146,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019877456600980693,
      "loss": 0.6406,
      "step": 1078
    },
    {
      "epoch": 0.5036172695449241,
      "grad_norm": 0.765625,
      "learning_rate": 0.000198772274824691,
      "loss": 0.611,
      "step": 1079
    },
    {
      "epoch": 0.5040840140023337,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019876998151289842,
      "loss": 0.6001,
      "step": 1080
    },
    {
      "epoch": 0.5045507584597433,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019876768607447854,
      "loss": 0.5799,
      "step": 1081
    },
    {
      "epoch": 0.5050175029171529,
      "grad_norm": 0.875,
      "learning_rate": 0.00019876538850948077,
      "loss": 0.7036,
      "step": 1082
    },
    {
      "epoch": 0.5054842473745624,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001987630888179546,
      "loss": 0.6494,
      "step": 1083
    },
    {
      "epoch": 0.505950991831972,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019876078699994959,
      "loss": 0.5809,
      "step": 1084
    },
    {
      "epoch": 0.5064177362893816,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019875848305551524,
      "loss": 0.5221,
      "step": 1085
    },
    {
      "epoch": 0.5068844807467912,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019875617698470114,
      "loss": 0.6165,
      "step": 1086
    },
    {
      "epoch": 0.5073512252042007,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019875386878755703,
      "loss": 0.6447,
      "step": 1087
    },
    {
      "epoch": 0.5078179696616103,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019875155846413255,
      "loss": 0.6486,
      "step": 1088
    },
    {
      "epoch": 0.5082847141190199,
      "grad_norm": 0.75,
      "learning_rate": 0.00019874924601447744,
      "loss": 0.6323,
      "step": 1089
    },
    {
      "epoch": 0.5087514585764294,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019874693143864146,
      "loss": 0.5859,
      "step": 1090
    },
    {
      "epoch": 0.509218203033839,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019874461473667454,
      "loss": 0.6746,
      "step": 1091
    },
    {
      "epoch": 0.5096849474912486,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019874229590862648,
      "loss": 0.6366,
      "step": 1092
    },
    {
      "epoch": 0.5101516919486581,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019873997495454724,
      "loss": 0.5875,
      "step": 1093
    },
    {
      "epoch": 0.5106184364060676,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019873765187448678,
      "loss": 0.6922,
      "step": 1094
    },
    {
      "epoch": 0.5110851808634772,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019873532666849514,
      "loss": 0.6177,
      "step": 1095
    },
    {
      "epoch": 0.5115519253208868,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019873299933662235,
      "loss": 0.6782,
      "step": 1096
    },
    {
      "epoch": 0.5120186697782964,
      "grad_norm": 0.75,
      "learning_rate": 0.00019873066987891855,
      "loss": 0.5991,
      "step": 1097
    },
    {
      "epoch": 0.5124854142357059,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019872833829543386,
      "loss": 0.5489,
      "step": 1098
    },
    {
      "epoch": 0.5129521586931155,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019872600458621854,
      "loss": 0.5954,
      "step": 1099
    },
    {
      "epoch": 0.5134189031505251,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019872366875132278,
      "loss": 0.6374,
      "step": 1100
    },
    {
      "epoch": 0.5138856476079346,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019872133079079692,
      "loss": 0.5936,
      "step": 1101
    },
    {
      "epoch": 0.5143523920653442,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019871899070469123,
      "loss": 0.6048,
      "step": 1102
    },
    {
      "epoch": 0.5148191365227538,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019871664849305617,
      "loss": 0.6304,
      "step": 1103
    },
    {
      "epoch": 0.5152858809801634,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019871430415594214,
      "loss": 0.6339,
      "step": 1104
    },
    {
      "epoch": 0.5157526254375729,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019871195769339962,
      "loss": 0.6001,
      "step": 1105
    },
    {
      "epoch": 0.5162193698949825,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019870960910547912,
      "loss": 0.5366,
      "step": 1106
    },
    {
      "epoch": 0.5166861143523921,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001987072583922312,
      "loss": 0.7068,
      "step": 1107
    },
    {
      "epoch": 0.5171528588098017,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001987049055537065,
      "loss": 0.5369,
      "step": 1108
    },
    {
      "epoch": 0.5176196032672112,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019870255058995572,
      "loss": 0.5677,
      "step": 1109
    },
    {
      "epoch": 0.5180863477246208,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019870019350102944,
      "loss": 0.5989,
      "step": 1110
    },
    {
      "epoch": 0.5185530921820304,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019869783428697852,
      "loss": 0.5877,
      "step": 1111
    },
    {
      "epoch": 0.51901983663944,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001986954729478537,
      "loss": 0.5889,
      "step": 1112
    },
    {
      "epoch": 0.5194865810968494,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019869310948370585,
      "loss": 0.6184,
      "step": 1113
    },
    {
      "epoch": 0.519953325554259,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019869074389458586,
      "loss": 0.5686,
      "step": 1114
    },
    {
      "epoch": 0.5204200700116686,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00019868837618054465,
      "loss": 0.5511,
      "step": 1115
    },
    {
      "epoch": 0.5208868144690781,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001986860063416332,
      "loss": 0.694,
      "step": 1116
    },
    {
      "epoch": 0.5213535589264877,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019868363437790257,
      "loss": 0.5902,
      "step": 1117
    },
    {
      "epoch": 0.5218203033838973,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019868126028940375,
      "loss": 0.6503,
      "step": 1118
    },
    {
      "epoch": 0.5222870478413069,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019867888407618792,
      "loss": 0.6597,
      "step": 1119
    },
    {
      "epoch": 0.5227537922987164,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019867650573830625,
      "loss": 0.6554,
      "step": 1120
    },
    {
      "epoch": 0.523220536756126,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019867412527580994,
      "loss": 0.65,
      "step": 1121
    },
    {
      "epoch": 0.5236872812135356,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001986717426887502,
      "loss": 0.5183,
      "step": 1122
    },
    {
      "epoch": 0.5241540256709452,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001986693579771784,
      "loss": 0.5503,
      "step": 1123
    },
    {
      "epoch": 0.5246207701283547,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019866697114114578,
      "loss": 0.6149,
      "step": 1124
    },
    {
      "epoch": 0.5250875145857643,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019866458218070385,
      "loss": 0.6891,
      "step": 1125
    },
    {
      "epoch": 0.5255542590431739,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019866219109590397,
      "loss": 0.5487,
      "step": 1126
    },
    {
      "epoch": 0.5260210035005835,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019865979788679768,
      "loss": 0.5028,
      "step": 1127
    },
    {
      "epoch": 0.526487747957993,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019865740255343645,
      "loss": 0.6483,
      "step": 1128
    },
    {
      "epoch": 0.5269544924154026,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001986550050958719,
      "loss": 0.5406,
      "step": 1129
    },
    {
      "epoch": 0.5274212368728122,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019865260551415564,
      "loss": 0.6311,
      "step": 1130
    },
    {
      "epoch": 0.5278879813302217,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019865020380833927,
      "loss": 0.5721,
      "step": 1131
    },
    {
      "epoch": 0.5283547257876313,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019864779997847462,
      "loss": 0.5387,
      "step": 1132
    },
    {
      "epoch": 0.5288214702450409,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019864539402461335,
      "loss": 0.6383,
      "step": 1133
    },
    {
      "epoch": 0.5292882147024504,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001986429859468073,
      "loss": 0.6705,
      "step": 1134
    },
    {
      "epoch": 0.5297549591598599,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019864057574510834,
      "loss": 0.5996,
      "step": 1135
    },
    {
      "epoch": 0.5302217036172695,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001986381634195683,
      "loss": 0.7158,
      "step": 1136
    },
    {
      "epoch": 0.5306884480746791,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001986357489702392,
      "loss": 0.584,
      "step": 1137
    },
    {
      "epoch": 0.5311551925320886,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019863333239717297,
      "loss": 0.6267,
      "step": 1138
    },
    {
      "epoch": 0.5316219369894982,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019863091370042165,
      "loss": 0.631,
      "step": 1139
    },
    {
      "epoch": 0.5320886814469078,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019862849288003736,
      "loss": 0.5929,
      "step": 1140
    },
    {
      "epoch": 0.5325554259043174,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019862606993607213,
      "loss": 0.5792,
      "step": 1141
    },
    {
      "epoch": 0.5330221703617269,
      "grad_norm": 0.65625,
      "learning_rate": 0.00019862364486857823,
      "loss": 0.4699,
      "step": 1142
    },
    {
      "epoch": 0.5334889148191365,
      "grad_norm": 0.75,
      "learning_rate": 0.00019862121767760783,
      "loss": 0.6793,
      "step": 1143
    },
    {
      "epoch": 0.5339556592765461,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019861878836321316,
      "loss": 0.6824,
      "step": 1144
    },
    {
      "epoch": 0.5344224037339557,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019861635692544657,
      "loss": 0.5772,
      "step": 1145
    },
    {
      "epoch": 0.5348891481913652,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001986139233643604,
      "loss": 0.6505,
      "step": 1146
    },
    {
      "epoch": 0.5353558926487748,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019861148768000707,
      "loss": 0.5209,
      "step": 1147
    },
    {
      "epoch": 0.5358226371061844,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019860904987243898,
      "loss": 0.6217,
      "step": 1148
    },
    {
      "epoch": 0.536289381563594,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019860660994170862,
      "loss": 0.6072,
      "step": 1149
    },
    {
      "epoch": 0.5367561260210035,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001986041678878686,
      "loss": 0.6351,
      "step": 1150
    },
    {
      "epoch": 0.5372228704784131,
      "grad_norm": 0.609375,
      "learning_rate": 0.00019860172371097138,
      "loss": 0.4966,
      "step": 1151
    },
    {
      "epoch": 0.5376896149358227,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001985992774110697,
      "loss": 0.6,
      "step": 1152
    },
    {
      "epoch": 0.5381563593932323,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019859682898821612,
      "loss": 0.5285,
      "step": 1153
    },
    {
      "epoch": 0.5386231038506418,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00019859437844246347,
      "loss": 0.565,
      "step": 1154
    },
    {
      "epoch": 0.5390898483080513,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019859192577386444,
      "loss": 0.6159,
      "step": 1155
    },
    {
      "epoch": 0.5395565927654609,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019858947098247185,
      "loss": 0.6855,
      "step": 1156
    },
    {
      "epoch": 0.5400233372228704,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019858701406833858,
      "loss": 0.6716,
      "step": 1157
    },
    {
      "epoch": 0.54049008168028,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019858455503151749,
      "loss": 0.6272,
      "step": 1158
    },
    {
      "epoch": 0.5409568261376896,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019858209387206154,
      "loss": 0.5411,
      "step": 1159
    },
    {
      "epoch": 0.5414235705950992,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019857963059002377,
      "loss": 0.5304,
      "step": 1160
    },
    {
      "epoch": 0.5418903150525087,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019857716518545716,
      "loss": 0.6087,
      "step": 1161
    },
    {
      "epoch": 0.5423570595099183,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019857469765841482,
      "loss": 0.6002,
      "step": 1162
    },
    {
      "epoch": 0.5428238039673279,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019857222800894982,
      "loss": 0.6597,
      "step": 1163
    },
    {
      "epoch": 0.5432905484247375,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00019856975623711544,
      "loss": 0.5125,
      "step": 1164
    },
    {
      "epoch": 0.543757292882147,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019856728234296482,
      "loss": 0.6435,
      "step": 1165
    },
    {
      "epoch": 0.5442240373395566,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019856480632655127,
      "loss": 0.5924,
      "step": 1166
    },
    {
      "epoch": 0.5446907817969662,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00019856232818792806,
      "loss": 0.6101,
      "step": 1167
    },
    {
      "epoch": 0.5451575262543757,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019855984792714856,
      "loss": 0.5965,
      "step": 1168
    },
    {
      "epoch": 0.5456242707117853,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001985573655442662,
      "loss": 0.5914,
      "step": 1169
    },
    {
      "epoch": 0.5460910151691949,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001985548810393344,
      "loss": 0.5135,
      "step": 1170
    },
    {
      "epoch": 0.5465577596266045,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019855239441240666,
      "loss": 0.3963,
      "step": 1171
    },
    {
      "epoch": 0.547024504084014,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019854990566353653,
      "loss": 0.6302,
      "step": 1172
    },
    {
      "epoch": 0.5474912485414236,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019854741479277758,
      "loss": 0.6625,
      "step": 1173
    },
    {
      "epoch": 0.5479579929988332,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019854492180018345,
      "loss": 0.4819,
      "step": 1174
    },
    {
      "epoch": 0.5484247374562428,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001985424266858078,
      "loss": 0.6229,
      "step": 1175
    },
    {
      "epoch": 0.5488914819136522,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001985399294497044,
      "loss": 0.5842,
      "step": 1176
    },
    {
      "epoch": 0.5493582263710618,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019853743009192698,
      "loss": 0.6479,
      "step": 1177
    },
    {
      "epoch": 0.5498249708284714,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019853492861252937,
      "loss": 0.4992,
      "step": 1178
    },
    {
      "epoch": 0.550291715285881,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001985324250115654,
      "loss": 0.586,
      "step": 1179
    },
    {
      "epoch": 0.5507584597432905,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000198529919289089,
      "loss": 0.6077,
      "step": 1180
    },
    {
      "epoch": 0.5512252042007001,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001985274114451541,
      "loss": 0.5771,
      "step": 1181
    },
    {
      "epoch": 0.5516919486581097,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019852490147981473,
      "loss": 0.6446,
      "step": 1182
    },
    {
      "epoch": 0.5521586931155192,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019852238939312492,
      "loss": 0.6201,
      "step": 1183
    },
    {
      "epoch": 0.5526254375729288,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019851987518513874,
      "loss": 0.5849,
      "step": 1184
    },
    {
      "epoch": 0.5530921820303384,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019851735885591034,
      "loss": 0.5069,
      "step": 1185
    },
    {
      "epoch": 0.553558926487748,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001985148404054939,
      "loss": 0.5743,
      "step": 1186
    },
    {
      "epoch": 0.5540256709451575,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019851231983394367,
      "loss": 0.6359,
      "step": 1187
    },
    {
      "epoch": 0.5544924154025671,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019850979714131385,
      "loss": 0.6098,
      "step": 1188
    },
    {
      "epoch": 0.5549591598599767,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019850727232765883,
      "loss": 0.5488,
      "step": 1189
    },
    {
      "epoch": 0.5554259043173863,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019850474539303295,
      "loss": 0.5868,
      "step": 1190
    },
    {
      "epoch": 0.5558926487747958,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019850221633749057,
      "loss": 0.5783,
      "step": 1191
    },
    {
      "epoch": 0.5563593932322054,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019849968516108622,
      "loss": 0.5875,
      "step": 1192
    },
    {
      "epoch": 0.556826137689615,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019849715186387435,
      "loss": 0.6392,
      "step": 1193
    },
    {
      "epoch": 0.5572928821470245,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019849461644590947,
      "loss": 0.6698,
      "step": 1194
    },
    {
      "epoch": 0.5577596266044341,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019849207890724628,
      "loss": 0.5326,
      "step": 1195
    },
    {
      "epoch": 0.5582263710618437,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019848953924793931,
      "loss": 0.5651,
      "step": 1196
    },
    {
      "epoch": 0.5586931155192532,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001984869974680433,
      "loss": 0.7025,
      "step": 1197
    },
    {
      "epoch": 0.5591598599766627,
      "grad_norm": 0.703125,
      "learning_rate": 0.00019848445356761296,
      "loss": 0.5442,
      "step": 1198
    },
    {
      "epoch": 0.5596266044340723,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019848190754670307,
      "loss": 0.617,
      "step": 1199
    },
    {
      "epoch": 0.5600933488914819,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00019847935940536844,
      "loss": 0.6329,
      "step": 1200
    },
    {
      "epoch": 0.5605600933488915,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019847680914366397,
      "loss": 0.5358,
      "step": 1201
    },
    {
      "epoch": 0.561026837806301,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001984742567616445,
      "loss": 0.6069,
      "step": 1202
    },
    {
      "epoch": 0.5614935822637106,
      "grad_norm": 0.671875,
      "learning_rate": 0.000198471702259365,
      "loss": 0.6207,
      "step": 1203
    },
    {
      "epoch": 0.5619603267211202,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00019846914563688054,
      "loss": 0.5826,
      "step": 1204
    },
    {
      "epoch": 0.5624270711785297,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001984665868942461,
      "loss": 0.5606,
      "step": 1205
    },
    {
      "epoch": 0.5628938156359393,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001984640260315168,
      "loss": 0.537,
      "step": 1206
    },
    {
      "epoch": 0.5633605600933489,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019846146304874777,
      "loss": 0.6786,
      "step": 1207
    },
    {
      "epoch": 0.5638273045507585,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019845889794599416,
      "loss": 0.4961,
      "step": 1208
    },
    {
      "epoch": 0.564294049008168,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001984563307233113,
      "loss": 0.5691,
      "step": 1209
    },
    {
      "epoch": 0.5647607934655776,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019845376138075436,
      "loss": 0.5507,
      "step": 1210
    },
    {
      "epoch": 0.5652275379229872,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001984511899183787,
      "loss": 0.5836,
      "step": 1211
    },
    {
      "epoch": 0.5656942823803968,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001984486163362397,
      "loss": 0.64,
      "step": 1212
    },
    {
      "epoch": 0.5661610268378063,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019844604063439273,
      "loss": 0.6291,
      "step": 1213
    },
    {
      "epoch": 0.5666277712952159,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001984434628128933,
      "loss": 0.6231,
      "step": 1214
    },
    {
      "epoch": 0.5670945157526255,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019844088287179691,
      "loss": 0.4757,
      "step": 1215
    },
    {
      "epoch": 0.5675612602100351,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019843830081115908,
      "loss": 0.6567,
      "step": 1216
    },
    {
      "epoch": 0.5680280046674445,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019843571663103542,
      "loss": 0.6016,
      "step": 1217
    },
    {
      "epoch": 0.5684947491248541,
      "grad_norm": 0.75,
      "learning_rate": 0.00019843313033148156,
      "loss": 0.4611,
      "step": 1218
    },
    {
      "epoch": 0.5689614935822637,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019843054191255317,
      "loss": 0.6219,
      "step": 1219
    },
    {
      "epoch": 0.5694282380396732,
      "grad_norm": 0.78125,
      "learning_rate": 0.000198427951374306,
      "loss": 0.5978,
      "step": 1220
    },
    {
      "epoch": 0.5698949824970828,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019842535871679586,
      "loss": 0.5723,
      "step": 1221
    },
    {
      "epoch": 0.5703617269544924,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019842276394007856,
      "loss": 0.5891,
      "step": 1222
    },
    {
      "epoch": 0.570828471411902,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019842016704420989,
      "loss": 0.5347,
      "step": 1223
    },
    {
      "epoch": 0.5712952158693115,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019841756802924584,
      "loss": 0.5659,
      "step": 1224
    },
    {
      "epoch": 0.5717619603267211,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019841496689524237,
      "loss": 0.552,
      "step": 1225
    },
    {
      "epoch": 0.5722287047841307,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019841236364225546,
      "loss": 0.5454,
      "step": 1226
    },
    {
      "epoch": 0.5726954492415403,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019840975827034121,
      "loss": 0.5268,
      "step": 1227
    },
    {
      "epoch": 0.5731621936989498,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019840715077955562,
      "loss": 0.6266,
      "step": 1228
    },
    {
      "epoch": 0.5736289381563594,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019840454116995494,
      "loss": 0.5017,
      "step": 1229
    },
    {
      "epoch": 0.574095682613769,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019840192944159526,
      "loss": 0.5586,
      "step": 1230
    },
    {
      "epoch": 0.5745624270711785,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019839931559453286,
      "loss": 0.508,
      "step": 1231
    },
    {
      "epoch": 0.5750291715285881,
      "grad_norm": 0.75,
      "learning_rate": 0.00019839669962882405,
      "loss": 0.4471,
      "step": 1232
    },
    {
      "epoch": 0.5754959159859977,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019839408154452513,
      "loss": 0.6899,
      "step": 1233
    },
    {
      "epoch": 0.5759626604434073,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019839146134169243,
      "loss": 0.553,
      "step": 1234
    },
    {
      "epoch": 0.5764294049008168,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019838883902038244,
      "loss": 0.4862,
      "step": 1235
    },
    {
      "epoch": 0.5768961493582264,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019838621458065155,
      "loss": 0.4894,
      "step": 1236
    },
    {
      "epoch": 0.577362893815636,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001983835880225563,
      "loss": 0.4977,
      "step": 1237
    },
    {
      "epoch": 0.5778296382730455,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019838095934615323,
      "loss": 0.6053,
      "step": 1238
    },
    {
      "epoch": 0.578296382730455,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019837832855149897,
      "loss": 0.6234,
      "step": 1239
    },
    {
      "epoch": 0.5787631271878646,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019837569563865018,
      "loss": 0.5609,
      "step": 1240
    },
    {
      "epoch": 0.5792298716452742,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019837306060766345,
      "loss": 0.5618,
      "step": 1241
    },
    {
      "epoch": 0.5796966161026837,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001983704234585956,
      "loss": 0.6144,
      "step": 1242
    },
    {
      "epoch": 0.5801633605600933,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001983677841915034,
      "loss": 0.5141,
      "step": 1243
    },
    {
      "epoch": 0.5806301050175029,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019836514280644366,
      "loss": 0.7074,
      "step": 1244
    },
    {
      "epoch": 0.5810968494749125,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019836249930347326,
      "loss": 0.568,
      "step": 1245
    },
    {
      "epoch": 0.581563593932322,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019835985368264912,
      "loss": 0.5369,
      "step": 1246
    },
    {
      "epoch": 0.5820303383897316,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001983572059440282,
      "loss": 0.5687,
      "step": 1247
    },
    {
      "epoch": 0.5824970828471412,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019835455608766746,
      "loss": 0.5049,
      "step": 1248
    },
    {
      "epoch": 0.5829638273045508,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019835190411362407,
      "loss": 0.504,
      "step": 1249
    },
    {
      "epoch": 0.5834305717619603,
      "grad_norm": 0.828125,
      "learning_rate": 0.000198349250021955,
      "loss": 0.5745,
      "step": 1250
    },
    {
      "epoch": 0.5838973162193699,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001983465938127175,
      "loss": 0.5861,
      "step": 1251
    },
    {
      "epoch": 0.5843640606767795,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00019834393548596869,
      "loss": 0.4657,
      "step": 1252
    },
    {
      "epoch": 0.5848308051341891,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019834127504176583,
      "loss": 0.6865,
      "step": 1253
    },
    {
      "epoch": 0.5852975495915986,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019833861248016622,
      "loss": 0.6339,
      "step": 1254
    },
    {
      "epoch": 0.5857642940490082,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019833594780122718,
      "loss": 0.6107,
      "step": 1255
    },
    {
      "epoch": 0.5862310385064178,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019833328100500608,
      "loss": 0.5271,
      "step": 1256
    },
    {
      "epoch": 0.5866977829638274,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019833061209156033,
      "loss": 0.5909,
      "step": 1257
    },
    {
      "epoch": 0.5871645274212369,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019832794106094742,
      "loss": 0.5531,
      "step": 1258
    },
    {
      "epoch": 0.5876312718786464,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019832526791322484,
      "loss": 0.5723,
      "step": 1259
    },
    {
      "epoch": 0.588098016336056,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019832259264845013,
      "loss": 0.6176,
      "step": 1260
    },
    {
      "epoch": 0.5885647607934655,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001983199152666809,
      "loss": 0.5575,
      "step": 1261
    },
    {
      "epoch": 0.5890315052508751,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019831723576797482,
      "loss": 0.5004,
      "step": 1262
    },
    {
      "epoch": 0.5894982497082847,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001983145541523896,
      "loss": 0.4624,
      "step": 1263
    },
    {
      "epoch": 0.5899649941656943,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001983118704199829,
      "loss": 0.6366,
      "step": 1264
    },
    {
      "epoch": 0.5904317386231038,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019830918457081257,
      "loss": 0.5971,
      "step": 1265
    },
    {
      "epoch": 0.5908984830805134,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019830649660493643,
      "loss": 0.5568,
      "step": 1266
    },
    {
      "epoch": 0.591365227537923,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019830380652241232,
      "loss": 0.6277,
      "step": 1267
    },
    {
      "epoch": 0.5918319719953326,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019830111432329818,
      "loss": 0.5998,
      "step": 1268
    },
    {
      "epoch": 0.5922987164527421,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019829842000765205,
      "loss": 0.4945,
      "step": 1269
    },
    {
      "epoch": 0.5927654609101517,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001982957235755318,
      "loss": 0.5901,
      "step": 1270
    },
    {
      "epoch": 0.5932322053675613,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019829302502699558,
      "loss": 0.5444,
      "step": 1271
    },
    {
      "epoch": 0.5936989498249708,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001982903243621015,
      "loss": 0.4735,
      "step": 1272
    },
    {
      "epoch": 0.5941656942823804,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019828762158090766,
      "loss": 0.6055,
      "step": 1273
    },
    {
      "epoch": 0.59463243873979,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019828491668347225,
      "loss": 0.5696,
      "step": 1274
    },
    {
      "epoch": 0.5950991831971996,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019828220966985357,
      "loss": 0.5917,
      "step": 1275
    },
    {
      "epoch": 0.5955659276546091,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019827950054010987,
      "loss": 0.5807,
      "step": 1276
    },
    {
      "epoch": 0.5960326721120187,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019827678929429946,
      "loss": 0.5153,
      "step": 1277
    },
    {
      "epoch": 0.5964994165694283,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019827407593248075,
      "loss": 0.5854,
      "step": 1278
    },
    {
      "epoch": 0.5969661610268379,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019827136045471214,
      "loss": 0.5567,
      "step": 1279
    },
    {
      "epoch": 0.5974329054842473,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001982686428610521,
      "loss": 0.424,
      "step": 1280
    },
    {
      "epoch": 0.5978996499416569,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019826592315155917,
      "loss": 0.5982,
      "step": 1281
    },
    {
      "epoch": 0.5983663943990665,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019826320132629188,
      "loss": 0.4398,
      "step": 1282
    },
    {
      "epoch": 0.598833138856476,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019826047738530882,
      "loss": 0.5948,
      "step": 1283
    },
    {
      "epoch": 0.5992998833138856,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019825775132866868,
      "loss": 0.5216,
      "step": 1284
    },
    {
      "epoch": 0.5997666277712952,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019825502315643013,
      "loss": 0.5817,
      "step": 1285
    },
    {
      "epoch": 0.6002333722287048,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019825229286865194,
      "loss": 0.5673,
      "step": 1286
    },
    {
      "epoch": 0.6007001166861143,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019824956046539285,
      "loss": 0.5603,
      "step": 1287
    },
    {
      "epoch": 0.6011668611435239,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001982468259467117,
      "loss": 0.6177,
      "step": 1288
    },
    {
      "epoch": 0.6016336056009335,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001982440893126674,
      "loss": 0.5323,
      "step": 1289
    },
    {
      "epoch": 0.6021003500583431,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019824135056331884,
      "loss": 0.6421,
      "step": 1290
    },
    {
      "epoch": 0.6025670945157526,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019823860969872501,
      "loss": 0.5715,
      "step": 1291
    },
    {
      "epoch": 0.6030338389731622,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001982358667189449,
      "loss": 0.4738,
      "step": 1292
    },
    {
      "epoch": 0.6035005834305718,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019823312162403763,
      "loss": 0.5069,
      "step": 1293
    },
    {
      "epoch": 0.6039673278879814,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001982303744140622,
      "loss": 0.5627,
      "step": 1294
    },
    {
      "epoch": 0.6044340723453909,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019822762508907787,
      "loss": 0.6167,
      "step": 1295
    },
    {
      "epoch": 0.6049008168028005,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019822487364914374,
      "loss": 0.4521,
      "step": 1296
    },
    {
      "epoch": 0.6053675612602101,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019822212009431913,
      "loss": 0.6713,
      "step": 1297
    },
    {
      "epoch": 0.6058343057176196,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001982193644246633,
      "loss": 0.5946,
      "step": 1298
    },
    {
      "epoch": 0.6063010501750292,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019821660664023555,
      "loss": 0.5451,
      "step": 1299
    },
    {
      "epoch": 0.6067677946324388,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001982138467410953,
      "loss": 0.5154,
      "step": 1300
    },
    {
      "epoch": 0.6072345390898483,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019821108472730197,
      "loss": 0.5641,
      "step": 1301
    },
    {
      "epoch": 0.6077012835472578,
      "grad_norm": 0.875,
      "learning_rate": 0.00019820832059891502,
      "loss": 0.5768,
      "step": 1302
    },
    {
      "epoch": 0.6081680280046674,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019820555435599393,
      "loss": 0.5741,
      "step": 1303
    },
    {
      "epoch": 0.608634772462077,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00019820278599859832,
      "loss": 0.5361,
      "step": 1304
    },
    {
      "epoch": 0.6091015169194866,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019820001552678777,
      "loss": 0.6414,
      "step": 1305
    },
    {
      "epoch": 0.6095682613768961,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00019819724294062194,
      "loss": 0.5521,
      "step": 1306
    },
    {
      "epoch": 0.6100350058343057,
      "grad_norm": 0.875,
      "learning_rate": 0.0001981944682401605,
      "loss": 0.5643,
      "step": 1307
    },
    {
      "epoch": 0.6105017502917153,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001981916914254632,
      "loss": 0.4732,
      "step": 1308
    },
    {
      "epoch": 0.6109684947491248,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019818891249658987,
      "loss": 0.5662,
      "step": 1309
    },
    {
      "epoch": 0.6114352392065344,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001981861314536003,
      "loss": 0.524,
      "step": 1310
    },
    {
      "epoch": 0.611901983663944,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019818334829655435,
      "loss": 0.5412,
      "step": 1311
    },
    {
      "epoch": 0.6123687281213536,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000198180563025512,
      "loss": 0.6389,
      "step": 1312
    },
    {
      "epoch": 0.6128354725787631,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001981777756405332,
      "loss": 0.4883,
      "step": 1313
    },
    {
      "epoch": 0.6133022170361727,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019817498614167796,
      "loss": 0.4852,
      "step": 1314
    },
    {
      "epoch": 0.6137689614935823,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019817219452900636,
      "loss": 0.5479,
      "step": 1315
    },
    {
      "epoch": 0.6142357059509919,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019816940080257843,
      "loss": 0.5989,
      "step": 1316
    },
    {
      "epoch": 0.6147024504084014,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019816660496245443,
      "loss": 0.5139,
      "step": 1317
    },
    {
      "epoch": 0.615169194865811,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001981638070086945,
      "loss": 0.4746,
      "step": 1318
    },
    {
      "epoch": 0.6156359393232206,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019816100694135887,
      "loss": 0.5103,
      "step": 1319
    },
    {
      "epoch": 0.6161026837806302,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019815820476050785,
      "loss": 0.4566,
      "step": 1320
    },
    {
      "epoch": 0.6165694282380396,
      "grad_norm": 1.0,
      "learning_rate": 0.0001981554004662018,
      "loss": 0.5635,
      "step": 1321
    },
    {
      "epoch": 0.6170361726954492,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019815259405850104,
      "loss": 0.5734,
      "step": 1322
    },
    {
      "epoch": 0.6175029171528588,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019814978553746605,
      "loss": 0.6637,
      "step": 1323
    },
    {
      "epoch": 0.6179696616102683,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001981469749031573,
      "loss": 0.5198,
      "step": 1324
    },
    {
      "epoch": 0.6184364060676779,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019814416215563525,
      "loss": 0.5453,
      "step": 1325
    },
    {
      "epoch": 0.6189031505250875,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001981413472949605,
      "loss": 0.4322,
      "step": 1326
    },
    {
      "epoch": 0.6193698949824971,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019813853032119366,
      "loss": 0.5662,
      "step": 1327
    },
    {
      "epoch": 0.6198366394399066,
      "grad_norm": 0.875,
      "learning_rate": 0.00019813571123439535,
      "loss": 0.5567,
      "step": 1328
    },
    {
      "epoch": 0.6203033838973162,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019813289003462633,
      "loss": 0.527,
      "step": 1329
    },
    {
      "epoch": 0.6207701283547258,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019813006672194728,
      "loss": 0.5942,
      "step": 1330
    },
    {
      "epoch": 0.6212368728121354,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019812724129641903,
      "loss": 0.6001,
      "step": 1331
    },
    {
      "epoch": 0.6217036172695449,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001981244137581024,
      "loss": 0.5326,
      "step": 1332
    },
    {
      "epoch": 0.6221703617269545,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001981215841070583,
      "loss": 0.6015,
      "step": 1333
    },
    {
      "epoch": 0.6226371061843641,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019811875234334756,
      "loss": 0.4905,
      "step": 1334
    },
    {
      "epoch": 0.6231038506417736,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019811591846703128,
      "loss": 0.5234,
      "step": 1335
    },
    {
      "epoch": 0.6235705950991832,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019811308247817038,
      "loss": 0.5209,
      "step": 1336
    },
    {
      "epoch": 0.6240373395565928,
      "grad_norm": 0.75,
      "learning_rate": 0.00019811024437682594,
      "loss": 0.5024,
      "step": 1337
    },
    {
      "epoch": 0.6245040840140024,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019810740416305913,
      "loss": 0.549,
      "step": 1338
    },
    {
      "epoch": 0.6249708284714119,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019810456183693104,
      "loss": 0.6548,
      "step": 1339
    },
    {
      "epoch": 0.6254375729288215,
      "grad_norm": 0.875,
      "learning_rate": 0.00019810171739850286,
      "loss": 0.4852,
      "step": 1340
    },
    {
      "epoch": 0.6259043173862311,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001980988708478359,
      "loss": 0.4537,
      "step": 1341
    },
    {
      "epoch": 0.6263710618436406,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019809602218499137,
      "loss": 0.6128,
      "step": 1342
    },
    {
      "epoch": 0.6268378063010501,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019809317141003066,
      "loss": 0.5842,
      "step": 1343
    },
    {
      "epoch": 0.6273045507584597,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019809031852301514,
      "loss": 0.5787,
      "step": 1344
    },
    {
      "epoch": 0.6277712952158693,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019808746352400624,
      "loss": 0.5334,
      "step": 1345
    },
    {
      "epoch": 0.6282380396732788,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001980846064130654,
      "loss": 0.4776,
      "step": 1346
    },
    {
      "epoch": 0.6287047841306884,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019808174719025413,
      "loss": 0.3992,
      "step": 1347
    },
    {
      "epoch": 0.629171528588098,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019807888585563404,
      "loss": 0.4611,
      "step": 1348
    },
    {
      "epoch": 0.6296382730455076,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00019807602240926674,
      "loss": 0.4907,
      "step": 1349
    },
    {
      "epoch": 0.6301050175029171,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019807315685121385,
      "loss": 0.5409,
      "step": 1350
    },
    {
      "epoch": 0.6305717619603267,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019807028918153707,
      "loss": 0.4067,
      "step": 1351
    },
    {
      "epoch": 0.6310385064177363,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019806741940029817,
      "loss": 0.4746,
      "step": 1352
    },
    {
      "epoch": 0.6315052508751459,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001980645475075589,
      "loss": 0.5216,
      "step": 1353
    },
    {
      "epoch": 0.6319719953325554,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019806167350338113,
      "loss": 0.5447,
      "step": 1354
    },
    {
      "epoch": 0.632438739789965,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019805879738782673,
      "loss": 0.4504,
      "step": 1355
    },
    {
      "epoch": 0.6329054842473746,
      "grad_norm": 1.0,
      "learning_rate": 0.00019805591916095762,
      "loss": 0.6486,
      "step": 1356
    },
    {
      "epoch": 0.6333722287047842,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019805303882283577,
      "loss": 0.5157,
      "step": 1357
    },
    {
      "epoch": 0.6338389731621937,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019805015637352322,
      "loss": 0.5648,
      "step": 1358
    },
    {
      "epoch": 0.6343057176196033,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000198047271813082,
      "loss": 0.5111,
      "step": 1359
    },
    {
      "epoch": 0.6347724620770129,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001980443851415742,
      "loss": 0.5553,
      "step": 1360
    },
    {
      "epoch": 0.6352392065344225,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019804149635906207,
      "loss": 0.5729,
      "step": 1361
    },
    {
      "epoch": 0.635705950991832,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001980386054656077,
      "loss": 0.5166,
      "step": 1362
    },
    {
      "epoch": 0.6361726954492415,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001980357124612734,
      "loss": 0.4623,
      "step": 1363
    },
    {
      "epoch": 0.6366394399066511,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019803281734612145,
      "loss": 0.5095,
      "step": 1364
    },
    {
      "epoch": 0.6371061843640606,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019802992012021414,
      "loss": 0.4926,
      "step": 1365
    },
    {
      "epoch": 0.6375729288214702,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001980270207836139,
      "loss": 0.5714,
      "step": 1366
    },
    {
      "epoch": 0.6380396732788798,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019802411933638313,
      "loss": 0.4658,
      "step": 1367
    },
    {
      "epoch": 0.6385064177362894,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019802121577858434,
      "loss": 0.5335,
      "step": 1368
    },
    {
      "epoch": 0.6389731621936989,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019801831011027998,
      "loss": 0.5037,
      "step": 1369
    },
    {
      "epoch": 0.6394399066511085,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019801540233153268,
      "loss": 0.5258,
      "step": 1370
    },
    {
      "epoch": 0.6399066511085181,
      "grad_norm": 0.875,
      "learning_rate": 0.000198012492442405,
      "loss": 0.5598,
      "step": 1371
    },
    {
      "epoch": 0.6403733955659276,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019800958044295964,
      "loss": 0.551,
      "step": 1372
    },
    {
      "epoch": 0.6408401400233372,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019800666633325924,
      "loss": 0.6314,
      "step": 1373
    },
    {
      "epoch": 0.6413068844807468,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019800375011336656,
      "loss": 0.4747,
      "step": 1374
    },
    {
      "epoch": 0.6417736289381564,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019800083178334442,
      "loss": 0.4619,
      "step": 1375
    },
    {
      "epoch": 0.6422403733955659,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019799791134325566,
      "loss": 0.495,
      "step": 1376
    },
    {
      "epoch": 0.6427071178529755,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001979949887931631,
      "loss": 0.5978,
      "step": 1377
    },
    {
      "epoch": 0.6431738623103851,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019799206413312974,
      "loss": 0.6492,
      "step": 1378
    },
    {
      "epoch": 0.6436406067677947,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001979891373632185,
      "loss": 0.4621,
      "step": 1379
    },
    {
      "epoch": 0.6441073512252042,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001979862084834924,
      "loss": 0.5321,
      "step": 1380
    },
    {
      "epoch": 0.6445740956826138,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019798327749401452,
      "loss": 0.5326,
      "step": 1381
    },
    {
      "epoch": 0.6450408401400234,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019798034439484798,
      "loss": 0.5767,
      "step": 1382
    },
    {
      "epoch": 0.645507584597433,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00019797740918605587,
      "loss": 0.4531,
      "step": 1383
    },
    {
      "epoch": 0.6459743290548424,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019797447186770147,
      "loss": 0.4831,
      "step": 1384
    },
    {
      "epoch": 0.646441073512252,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019797153243984798,
      "loss": 0.5255,
      "step": 1385
    },
    {
      "epoch": 0.6469078179696616,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001979685909025587,
      "loss": 0.4433,
      "step": 1386
    },
    {
      "epoch": 0.6473745624270711,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019796564725589697,
      "loss": 0.5535,
      "step": 1387
    },
    {
      "epoch": 0.6478413068844807,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019796270149992614,
      "loss": 0.4754,
      "step": 1388
    },
    {
      "epoch": 0.6483080513418903,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019795975363470964,
      "loss": 0.4592,
      "step": 1389
    },
    {
      "epoch": 0.6487747957992999,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019795680366031096,
      "loss": 0.5216,
      "step": 1390
    },
    {
      "epoch": 0.6492415402567094,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019795385157679365,
      "loss": 0.5227,
      "step": 1391
    },
    {
      "epoch": 0.649708284714119,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001979508973842212,
      "loss": 0.4431,
      "step": 1392
    },
    {
      "epoch": 0.6501750291715286,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019794794108265727,
      "loss": 0.5014,
      "step": 1393
    },
    {
      "epoch": 0.6506417736289382,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019794498267216548,
      "loss": 0.6202,
      "step": 1394
    },
    {
      "epoch": 0.6511085180863477,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019794202215280956,
      "loss": 0.5228,
      "step": 1395
    },
    {
      "epoch": 0.6515752625437573,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00019793905952465322,
      "loss": 0.5027,
      "step": 1396
    },
    {
      "epoch": 0.6520420070011669,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019793609478776025,
      "loss": 0.5725,
      "step": 1397
    },
    {
      "epoch": 0.6525087514585765,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001979331279421945,
      "loss": 0.4994,
      "step": 1398
    },
    {
      "epoch": 0.652975495915986,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019793015898801985,
      "loss": 0.425,
      "step": 1399
    },
    {
      "epoch": 0.6534422403733956,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019792718792530024,
      "loss": 0.5431,
      "step": 1400
    },
    {
      "epoch": 0.6539089848308052,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001979242147540996,
      "loss": 0.5034,
      "step": 1401
    },
    {
      "epoch": 0.6543757292882147,
      "grad_norm": 0.97265625,
      "learning_rate": 0.000197921239474482,
      "loss": 0.5028,
      "step": 1402
    },
    {
      "epoch": 0.6548424737456243,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019791826208651143,
      "loss": 0.5547,
      "step": 1403
    },
    {
      "epoch": 0.6553092182030339,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019791528259025207,
      "loss": 0.6541,
      "step": 1404
    },
    {
      "epoch": 0.6557759626604434,
      "grad_norm": 0.859375,
      "learning_rate": 0.000197912300985768,
      "loss": 0.5839,
      "step": 1405
    },
    {
      "epoch": 0.6562427071178529,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019790931727312348,
      "loss": 0.5855,
      "step": 1406
    },
    {
      "epoch": 0.6567094515752625,
      "grad_norm": 0.75,
      "learning_rate": 0.0001979063314523827,
      "loss": 0.5087,
      "step": 1407
    },
    {
      "epoch": 0.6571761960326721,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019790334352361,
      "loss": 0.4872,
      "step": 1408
    },
    {
      "epoch": 0.6576429404900817,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001979003534868697,
      "loss": 0.478,
      "step": 1409
    },
    {
      "epoch": 0.6581096849474912,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019789736134222615,
      "loss": 0.5371,
      "step": 1410
    },
    {
      "epoch": 0.6585764294049008,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019789436708974377,
      "loss": 0.523,
      "step": 1411
    },
    {
      "epoch": 0.6590431738623104,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001978913707294871,
      "loss": 0.6405,
      "step": 1412
    },
    {
      "epoch": 0.65950991831972,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019788837226152056,
      "loss": 0.5186,
      "step": 1413
    },
    {
      "epoch": 0.6599766627771295,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001978853716859088,
      "loss": 0.5361,
      "step": 1414
    },
    {
      "epoch": 0.6604434072345391,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019788236900271635,
      "loss": 0.4894,
      "step": 1415
    },
    {
      "epoch": 0.6609101516919487,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001978793642120079,
      "loss": 0.5222,
      "step": 1416
    },
    {
      "epoch": 0.6613768961493582,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019787635731384815,
      "loss": 0.4927,
      "step": 1417
    },
    {
      "epoch": 0.6618436406067678,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001978733483083018,
      "loss": 0.4338,
      "step": 1418
    },
    {
      "epoch": 0.6623103850641774,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001978703371954337,
      "loss": 0.4584,
      "step": 1419
    },
    {
      "epoch": 0.662777129521587,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019786732397530864,
      "loss": 0.5202,
      "step": 1420
    },
    {
      "epoch": 0.6632438739789965,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001978643086479915,
      "loss": 0.5412,
      "step": 1421
    },
    {
      "epoch": 0.6637106184364061,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019786129121354722,
      "loss": 0.5155,
      "step": 1422
    },
    {
      "epoch": 0.6641773628938157,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019785827167204077,
      "loss": 0.551,
      "step": 1423
    },
    {
      "epoch": 0.6646441073512253,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019785525002353715,
      "loss": 0.5832,
      "step": 1424
    },
    {
      "epoch": 0.6651108518086347,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019785222626810144,
      "loss": 0.4761,
      "step": 1425
    },
    {
      "epoch": 0.6655775962660443,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001978492004057987,
      "loss": 0.4961,
      "step": 1426
    },
    {
      "epoch": 0.6660443407234539,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001978461724366941,
      "loss": 0.4185,
      "step": 1427
    },
    {
      "epoch": 0.6665110851808634,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019784314236085288,
      "loss": 0.4806,
      "step": 1428
    },
    {
      "epoch": 0.666977829638273,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019784011017834024,
      "loss": 0.4095,
      "step": 1429
    },
    {
      "epoch": 0.6674445740956826,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019783707588922144,
      "loss": 0.4009,
      "step": 1430
    },
    {
      "epoch": 0.6679113185530922,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019783403949356188,
      "loss": 0.5247,
      "step": 1431
    },
    {
      "epoch": 0.6683780630105017,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001978310009914269,
      "loss": 0.5265,
      "step": 1432
    },
    {
      "epoch": 0.6688448074679113,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019782796038288192,
      "loss": 0.4718,
      "step": 1433
    },
    {
      "epoch": 0.6693115519253209,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001978249176679924,
      "loss": 0.5344,
      "step": 1434
    },
    {
      "epoch": 0.6697782963827305,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001978218728468239,
      "loss": 0.4821,
      "step": 1435
    },
    {
      "epoch": 0.67024504084014,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019781882591944188,
      "loss": 0.3894,
      "step": 1436
    },
    {
      "epoch": 0.6707117852975496,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019781577688591205,
      "loss": 0.4959,
      "step": 1437
    },
    {
      "epoch": 0.6711785297549592,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019781272574629998,
      "loss": 0.6066,
      "step": 1438
    },
    {
      "epoch": 0.6716452742123687,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001978096725006714,
      "loss": 0.4791,
      "step": 1439
    },
    {
      "epoch": 0.6721120186697783,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001978066171490921,
      "loss": 0.4337,
      "step": 1440
    },
    {
      "epoch": 0.6725787631271879,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001978035596916278,
      "loss": 0.4816,
      "step": 1441
    },
    {
      "epoch": 0.6730455075845975,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019780050012834432,
      "loss": 0.444,
      "step": 1442
    },
    {
      "epoch": 0.673512252042007,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001977974384593076,
      "loss": 0.5353,
      "step": 1443
    },
    {
      "epoch": 0.6739789964994166,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019779437468458346,
      "loss": 0.4419,
      "step": 1444
    },
    {
      "epoch": 0.6744457409568262,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019779130880423796,
      "loss": 0.5248,
      "step": 1445
    },
    {
      "epoch": 0.6749124854142357,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001977882408183371,
      "loss": 0.4598,
      "step": 1446
    },
    {
      "epoch": 0.6753792298716452,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019778517072694687,
      "loss": 0.5538,
      "step": 1447
    },
    {
      "epoch": 0.6758459743290548,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019778209853013349,
      "loss": 0.5071,
      "step": 1448
    },
    {
      "epoch": 0.6763127187864644,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019777902422796296,
      "loss": 0.5546,
      "step": 1449
    },
    {
      "epoch": 0.676779463243874,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001977759478205016,
      "loss": 0.4922,
      "step": 1450
    },
    {
      "epoch": 0.6772462077012835,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019777286930781557,
      "loss": 0.5289,
      "step": 1451
    },
    {
      "epoch": 0.6777129521586931,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001977697886899712,
      "loss": 0.4832,
      "step": 1452
    },
    {
      "epoch": 0.6781796966161027,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019776670596703482,
      "loss": 0.4692,
      "step": 1453
    },
    {
      "epoch": 0.6786464410735122,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019776362113907276,
      "loss": 0.5184,
      "step": 1454
    },
    {
      "epoch": 0.6791131855309218,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001977605342061515,
      "loss": 0.5701,
      "step": 1455
    },
    {
      "epoch": 0.6795799299883314,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001977574451683374,
      "loss": 0.553,
      "step": 1456
    },
    {
      "epoch": 0.680046674445741,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001977543540256971,
      "loss": 0.4372,
      "step": 1457
    },
    {
      "epoch": 0.6805134189031505,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001977512607782971,
      "loss": 0.451,
      "step": 1458
    },
    {
      "epoch": 0.6809801633605601,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000197748165426204,
      "loss": 0.5201,
      "step": 1459
    },
    {
      "epoch": 0.6814469078179697,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001977450679694844,
      "loss": 0.541,
      "step": 1460
    },
    {
      "epoch": 0.6819136522753793,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019774196840820506,
      "loss": 0.5176,
      "step": 1461
    },
    {
      "epoch": 0.6823803967327888,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001977388667424327,
      "loss": 0.591,
      "step": 1462
    },
    {
      "epoch": 0.6828471411901984,
      "grad_norm": 0.875,
      "learning_rate": 0.0001977357629722341,
      "loss": 0.5705,
      "step": 1463
    },
    {
      "epoch": 0.683313885647608,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019773265709767607,
      "loss": 0.405,
      "step": 1464
    },
    {
      "epoch": 0.6837806301050176,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001977295491188255,
      "loss": 0.4849,
      "step": 1465
    },
    {
      "epoch": 0.6842473745624271,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001977264390357493,
      "loss": 0.5683,
      "step": 1466
    },
    {
      "epoch": 0.6847141190198366,
      "grad_norm": 0.734375,
      "learning_rate": 0.00019772332684851444,
      "loss": 0.4424,
      "step": 1467
    },
    {
      "epoch": 0.6851808634772462,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019772021255718797,
      "loss": 0.5637,
      "step": 1468
    },
    {
      "epoch": 0.6856476079346557,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019771709616183685,
      "loss": 0.6238,
      "step": 1469
    },
    {
      "epoch": 0.6861143523920653,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019771397766252827,
      "loss": 0.5114,
      "step": 1470
    },
    {
      "epoch": 0.6865810968494749,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001977108570593293,
      "loss": 0.5068,
      "step": 1471
    },
    {
      "epoch": 0.6870478413068845,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019770773435230716,
      "loss": 0.5077,
      "step": 1472
    },
    {
      "epoch": 0.687514585764294,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019770460954152908,
      "loss": 0.596,
      "step": 1473
    },
    {
      "epoch": 0.6879813302217036,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019770148262706242,
      "loss": 0.4074,
      "step": 1474
    },
    {
      "epoch": 0.6884480746791132,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001976983536089744,
      "loss": 0.4277,
      "step": 1475
    },
    {
      "epoch": 0.6889148191365227,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019769522248733241,
      "loss": 0.4967,
      "step": 1476
    },
    {
      "epoch": 0.6893815635939323,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001976920892622039,
      "loss": 0.4541,
      "step": 1477
    },
    {
      "epoch": 0.6898483080513419,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001976889539336563,
      "loss": 0.4746,
      "step": 1478
    },
    {
      "epoch": 0.6903150525087515,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019768581650175717,
      "loss": 0.5254,
      "step": 1479
    },
    {
      "epoch": 0.690781796966161,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019768267696657402,
      "loss": 0.4079,
      "step": 1480
    },
    {
      "epoch": 0.6912485414235706,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019767953532817442,
      "loss": 0.4263,
      "step": 1481
    },
    {
      "epoch": 0.6917152858809802,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019767639158662606,
      "loss": 0.4549,
      "step": 1482
    },
    {
      "epoch": 0.6921820303383898,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019767324574199665,
      "loss": 0.4927,
      "step": 1483
    },
    {
      "epoch": 0.6926487747957993,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019767009779435383,
      "loss": 0.5501,
      "step": 1484
    },
    {
      "epoch": 0.6931155192532089,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001976669477437655,
      "loss": 0.5278,
      "step": 1485
    },
    {
      "epoch": 0.6935822637106185,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001976637955902994,
      "loss": 0.5078,
      "step": 1486
    },
    {
      "epoch": 0.6940490081680281,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001976606413340234,
      "loss": 0.5094,
      "step": 1487
    },
    {
      "epoch": 0.6945157526254375,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019765748497500545,
      "loss": 0.5271,
      "step": 1488
    },
    {
      "epoch": 0.6949824970828471,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001976543265133135,
      "loss": 0.4738,
      "step": 1489
    },
    {
      "epoch": 0.6954492415402567,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019765116594901554,
      "loss": 0.5253,
      "step": 1490
    },
    {
      "epoch": 0.6959159859976662,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019764800328217962,
      "loss": 0.5159,
      "step": 1491
    },
    {
      "epoch": 0.6963827304550758,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019764483851287387,
      "loss": 0.5285,
      "step": 1492
    },
    {
      "epoch": 0.6968494749124854,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001976416716411664,
      "loss": 0.4709,
      "step": 1493
    },
    {
      "epoch": 0.697316219369895,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00019763850266712543,
      "loss": 0.3935,
      "step": 1494
    },
    {
      "epoch": 0.6977829638273045,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019763533159081913,
      "loss": 0.5108,
      "step": 1495
    },
    {
      "epoch": 0.6982497082847141,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001976321584123158,
      "loss": 0.4132,
      "step": 1496
    },
    {
      "epoch": 0.6987164527421237,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001976289831316838,
      "loss": 0.5321,
      "step": 1497
    },
    {
      "epoch": 0.6991831971995333,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019762580574899147,
      "loss": 0.5543,
      "step": 1498
    },
    {
      "epoch": 0.6996499416569428,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019762262626430723,
      "loss": 0.4831,
      "step": 1499
    },
    {
      "epoch": 0.7001166861143524,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019761944467769956,
      "loss": 0.4984,
      "step": 1500
    },
    {
      "epoch": 0.700583430571762,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019761626098923685,
      "loss": 0.4496,
      "step": 1501
    },
    {
      "epoch": 0.7010501750291716,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019761307519898782,
      "loss": 0.541,
      "step": 1502
    },
    {
      "epoch": 0.7015169194865811,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019760988730702095,
      "loss": 0.4288,
      "step": 1503
    },
    {
      "epoch": 0.7019836639439907,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001976066973134049,
      "loss": 0.4098,
      "step": 1504
    },
    {
      "epoch": 0.7024504084014003,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019760350521820836,
      "loss": 0.4283,
      "step": 1505
    },
    {
      "epoch": 0.7029171528588098,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019760031102150006,
      "loss": 0.4578,
      "step": 1506
    },
    {
      "epoch": 0.7033838973162194,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019759711472334875,
      "loss": 0.602,
      "step": 1507
    },
    {
      "epoch": 0.703850641773629,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001975939163238233,
      "loss": 0.4655,
      "step": 1508
    },
    {
      "epoch": 0.7043173862310385,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019759071582299255,
      "loss": 0.5103,
      "step": 1509
    },
    {
      "epoch": 0.704784130688448,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001975875132209254,
      "loss": 0.572,
      "step": 1510
    },
    {
      "epoch": 0.7052508751458576,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019758430851769086,
      "loss": 0.5164,
      "step": 1511
    },
    {
      "epoch": 0.7057176196032672,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019758110171335785,
      "loss": 0.435,
      "step": 1512
    },
    {
      "epoch": 0.7061843640606768,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019757789280799544,
      "loss": 0.3969,
      "step": 1513
    },
    {
      "epoch": 0.7066511085180863,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00019757468180167274,
      "loss": 0.3589,
      "step": 1514
    },
    {
      "epoch": 0.7071178529754959,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019757146869445888,
      "loss": 0.5392,
      "step": 1515
    },
    {
      "epoch": 0.7075845974329055,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00019756825348642306,
      "loss": 0.3913,
      "step": 1516
    },
    {
      "epoch": 0.708051341890315,
      "grad_norm": 1.0,
      "learning_rate": 0.00019756503617763448,
      "loss": 0.4911,
      "step": 1517
    },
    {
      "epoch": 0.7085180863477246,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019756181676816242,
      "loss": 0.5328,
      "step": 1518
    },
    {
      "epoch": 0.7089848308051342,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019755859525807617,
      "loss": 0.5472,
      "step": 1519
    },
    {
      "epoch": 0.7094515752625438,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019755537164744516,
      "loss": 0.5794,
      "step": 1520
    },
    {
      "epoch": 0.7099183197199533,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019755214593633873,
      "loss": 0.547,
      "step": 1521
    },
    {
      "epoch": 0.7103850641773629,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019754891812482636,
      "loss": 0.4875,
      "step": 1522
    },
    {
      "epoch": 0.7108518086347725,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019754568821297758,
      "loss": 0.5612,
      "step": 1523
    },
    {
      "epoch": 0.7113185530921821,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001975424562008619,
      "loss": 0.4908,
      "step": 1524
    },
    {
      "epoch": 0.7117852975495916,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019753922208854888,
      "loss": 0.4195,
      "step": 1525
    },
    {
      "epoch": 0.7122520420070012,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019753598587610822,
      "loss": 0.493,
      "step": 1526
    },
    {
      "epoch": 0.7127187864644108,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019753274756360956,
      "loss": 0.4796,
      "step": 1527
    },
    {
      "epoch": 0.7131855309218204,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001975295071511226,
      "loss": 0.5781,
      "step": 1528
    },
    {
      "epoch": 0.7136522753792298,
      "grad_norm": 0.765625,
      "learning_rate": 0.00019752626463871717,
      "loss": 0.3916,
      "step": 1529
    },
    {
      "epoch": 0.7141190198366394,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019752302002646306,
      "loss": 0.557,
      "step": 1530
    },
    {
      "epoch": 0.714585764294049,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019751977331443013,
      "loss": 0.4787,
      "step": 1531
    },
    {
      "epoch": 0.7150525087514585,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019751652450268825,
      "loss": 0.4198,
      "step": 1532
    },
    {
      "epoch": 0.7155192532088681,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019751327359130742,
      "loss": 0.478,
      "step": 1533
    },
    {
      "epoch": 0.7159859976662777,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019751002058035765,
      "loss": 0.5439,
      "step": 1534
    },
    {
      "epoch": 0.7164527421236873,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00019750676546990888,
      "loss": 0.4628,
      "step": 1535
    },
    {
      "epoch": 0.7169194865810968,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001975035082600313,
      "loss": 0.5295,
      "step": 1536
    },
    {
      "epoch": 0.7173862310385064,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019750024895079499,
      "loss": 0.4579,
      "step": 1537
    },
    {
      "epoch": 0.717852975495916,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019749698754227018,
      "loss": 0.4844,
      "step": 1538
    },
    {
      "epoch": 0.7183197199533256,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019749372403452702,
      "loss": 0.4228,
      "step": 1539
    },
    {
      "epoch": 0.7187864644107351,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001974904584276358,
      "loss": 0.5105,
      "step": 1540
    },
    {
      "epoch": 0.7192532088681447,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019748719072166685,
      "loss": 0.5078,
      "step": 1541
    },
    {
      "epoch": 0.7197199533255543,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019748392091669054,
      "loss": 0.4045,
      "step": 1542
    },
    {
      "epoch": 0.7201866977829638,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019748064901277724,
      "loss": 0.3985,
      "step": 1543
    },
    {
      "epoch": 0.7206534422403734,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019747737500999741,
      "loss": 0.4223,
      "step": 1544
    },
    {
      "epoch": 0.721120186697783,
      "grad_norm": 0.75,
      "learning_rate": 0.00019747409890842155,
      "loss": 0.3918,
      "step": 1545
    },
    {
      "epoch": 0.7215869311551926,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019747082070812016,
      "loss": 0.4111,
      "step": 1546
    },
    {
      "epoch": 0.7220536756126021,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001974675404091639,
      "loss": 0.4737,
      "step": 1547
    },
    {
      "epoch": 0.7225204200700117,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00019746425801162333,
      "loss": 0.474,
      "step": 1548
    },
    {
      "epoch": 0.7229871645274213,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019746097351556915,
      "loss": 0.5133,
      "step": 1549
    },
    {
      "epoch": 0.7234539089848308,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019745768692107208,
      "loss": 0.4541,
      "step": 1550
    },
    {
      "epoch": 0.7239206534422403,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001974543982282029,
      "loss": 0.477,
      "step": 1551
    },
    {
      "epoch": 0.7243873978996499,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019745110743703235,
      "loss": 0.5925,
      "step": 1552
    },
    {
      "epoch": 0.7248541423570595,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019744781454763134,
      "loss": 0.4749,
      "step": 1553
    },
    {
      "epoch": 0.725320886814469,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001974445195600708,
      "loss": 0.4109,
      "step": 1554
    },
    {
      "epoch": 0.7257876312718786,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001974412224744216,
      "loss": 0.4807,
      "step": 1555
    },
    {
      "epoch": 0.7262543757292882,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001974379232907548,
      "loss": 0.525,
      "step": 1556
    },
    {
      "epoch": 0.7267211201866978,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019743462200914137,
      "loss": 0.463,
      "step": 1557
    },
    {
      "epoch": 0.7271878646441073,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00019743131862965245,
      "loss": 0.3741,
      "step": 1558
    },
    {
      "epoch": 0.7276546091015169,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001974280131523591,
      "loss": 0.492,
      "step": 1559
    },
    {
      "epoch": 0.7281213535589265,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019742470557733256,
      "loss": 0.5732,
      "step": 1560
    },
    {
      "epoch": 0.7285880980163361,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019742139590464404,
      "loss": 0.3877,
      "step": 1561
    },
    {
      "epoch": 0.7290548424737456,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019741808413436474,
      "loss": 0.3923,
      "step": 1562
    },
    {
      "epoch": 0.7295215869311552,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019741477026656598,
      "loss": 0.4943,
      "step": 1563
    },
    {
      "epoch": 0.7299883313885648,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019741145430131915,
      "loss": 0.4176,
      "step": 1564
    },
    {
      "epoch": 0.7304550758459744,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019740813623869566,
      "loss": 0.4785,
      "step": 1565
    },
    {
      "epoch": 0.7309218203033839,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019740481607876688,
      "loss": 0.4781,
      "step": 1566
    },
    {
      "epoch": 0.7313885647607935,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019740149382160437,
      "loss": 0.4586,
      "step": 1567
    },
    {
      "epoch": 0.7318553092182031,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019739816946727962,
      "loss": 0.4585,
      "step": 1568
    },
    {
      "epoch": 0.7323220536756126,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001973948430158642,
      "loss": 0.4897,
      "step": 1569
    },
    {
      "epoch": 0.7327887981330222,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00019739151446742975,
      "loss": 0.3999,
      "step": 1570
    },
    {
      "epoch": 0.7332555425904317,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019738818382204794,
      "loss": 0.4348,
      "step": 1571
    },
    {
      "epoch": 0.7337222870478413,
      "grad_norm": 1.0,
      "learning_rate": 0.00019738485107979048,
      "loss": 0.4347,
      "step": 1572
    },
    {
      "epoch": 0.7341890315052508,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019738151624072915,
      "loss": 0.4766,
      "step": 1573
    },
    {
      "epoch": 0.7346557759626604,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001973781793049357,
      "loss": 0.4679,
      "step": 1574
    },
    {
      "epoch": 0.73512252042007,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019737484027248202,
      "loss": 0.5101,
      "step": 1575
    },
    {
      "epoch": 0.7355892648774796,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019737149914343996,
      "loss": 0.4748,
      "step": 1576
    },
    {
      "epoch": 0.7360560093348891,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019736815591788154,
      "loss": 0.4756,
      "step": 1577
    },
    {
      "epoch": 0.7365227537922987,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019736481059587864,
      "loss": 0.5661,
      "step": 1578
    },
    {
      "epoch": 0.7369894982497083,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019736146317750336,
      "loss": 0.4576,
      "step": 1579
    },
    {
      "epoch": 0.7374562427071178,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019735811366282778,
      "loss": 0.4464,
      "step": 1580
    },
    {
      "epoch": 0.7379229871645274,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019735476205192396,
      "loss": 0.5366,
      "step": 1581
    },
    {
      "epoch": 0.738389731621937,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001973514083448641,
      "loss": 0.4377,
      "step": 1582
    },
    {
      "epoch": 0.7388564760793466,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001973480525417204,
      "loss": 0.4715,
      "step": 1583
    },
    {
      "epoch": 0.7393232205367561,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019734469464256513,
      "loss": 0.5448,
      "step": 1584
    },
    {
      "epoch": 0.7397899649941657,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001973413346474706,
      "loss": 0.4515,
      "step": 1585
    },
    {
      "epoch": 0.7402567094515753,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019733797255650908,
      "loss": 0.3848,
      "step": 1586
    },
    {
      "epoch": 0.7407234539089849,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019733460836975306,
      "loss": 0.4414,
      "step": 1587
    },
    {
      "epoch": 0.7411901983663944,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019733124208727492,
      "loss": 0.4362,
      "step": 1588
    },
    {
      "epoch": 0.741656942823804,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019732787370914712,
      "loss": 0.4298,
      "step": 1589
    },
    {
      "epoch": 0.7421236872812136,
      "grad_norm": 0.875,
      "learning_rate": 0.00019732450323544222,
      "loss": 0.3198,
      "step": 1590
    },
    {
      "epoch": 0.7425904317386232,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001973211306662328,
      "loss": 0.4627,
      "step": 1591
    },
    {
      "epoch": 0.7430571761960326,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019731775600159143,
      "loss": 0.4042,
      "step": 1592
    },
    {
      "epoch": 0.7435239206534422,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019731437924159085,
      "loss": 0.4888,
      "step": 1593
    },
    {
      "epoch": 0.7439906651108518,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019731100038630366,
      "loss": 0.4372,
      "step": 1594
    },
    {
      "epoch": 0.7444574095682613,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001973076194358027,
      "loss": 0.5551,
      "step": 1595
    },
    {
      "epoch": 0.7449241540256709,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019730423639016068,
      "loss": 0.5257,
      "step": 1596
    },
    {
      "epoch": 0.7453908984830805,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001973008512494505,
      "loss": 0.5248,
      "step": 1597
    },
    {
      "epoch": 0.7458576429404901,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001972974640137451,
      "loss": 0.456,
      "step": 1598
    },
    {
      "epoch": 0.7463243873978996,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001972940746831173,
      "loss": 0.4842,
      "step": 1599
    },
    {
      "epoch": 0.7467911318553092,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001972906832576401,
      "loss": 0.4286,
      "step": 1600
    },
    {
      "epoch": 0.7472578763127188,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019728728973738657,
      "loss": 0.4016,
      "step": 1601
    },
    {
      "epoch": 0.7477246207701284,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019728389412242977,
      "loss": 0.5307,
      "step": 1602
    },
    {
      "epoch": 0.7481913652275379,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019728049641284275,
      "loss": 0.5228,
      "step": 1603
    },
    {
      "epoch": 0.7486581096849475,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001972770966086987,
      "loss": 0.3824,
      "step": 1604
    },
    {
      "epoch": 0.7491248541423571,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001972736947100709,
      "loss": 0.452,
      "step": 1605
    },
    {
      "epoch": 0.7495915985997667,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019727029071703247,
      "loss": 0.4048,
      "step": 1606
    },
    {
      "epoch": 0.7500583430571762,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019726688462965678,
      "loss": 0.5633,
      "step": 1607
    },
    {
      "epoch": 0.7505250875145858,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001972634764480171,
      "loss": 0.4046,
      "step": 1608
    },
    {
      "epoch": 0.7505250875145858,
      "eval_loss": 1.0971609354019165,
      "eval_runtime": 94.2121,
      "eval_samples_per_second": 19.148,
      "eval_steps_per_second": 2.399,
      "step": 1608
    },
    {
      "epoch": 0.7509918319719954,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001972600661721869,
      "loss": 0.4298,
      "step": 1609
    },
    {
      "epoch": 0.751458576429405,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019725665380223957,
      "loss": 0.3477,
      "step": 1610
    },
    {
      "epoch": 0.7519253208868145,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019725323933824856,
      "loss": 0.3404,
      "step": 1611
    },
    {
      "epoch": 0.7523920653442241,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019724982278028738,
      "loss": 0.4663,
      "step": 1612
    },
    {
      "epoch": 0.7528588098016336,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019724640412842966,
      "loss": 0.53,
      "step": 1613
    },
    {
      "epoch": 0.7533255542590431,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001972429833827489,
      "loss": 0.3965,
      "step": 1614
    },
    {
      "epoch": 0.7537922987164527,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019723956054331886,
      "loss": 0.4167,
      "step": 1615
    },
    {
      "epoch": 0.7542590431738623,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019723613561021318,
      "loss": 0.5089,
      "step": 1616
    },
    {
      "epoch": 0.7547257876312718,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019723270858350563,
      "loss": 0.3663,
      "step": 1617
    },
    {
      "epoch": 0.7551925320886814,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019722927946326995,
      "loss": 0.4208,
      "step": 1618
    },
    {
      "epoch": 0.755659276546091,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019722584824958007,
      "loss": 0.473,
      "step": 1619
    },
    {
      "epoch": 0.7561260210035006,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019722241494250976,
      "loss": 0.3862,
      "step": 1620
    },
    {
      "epoch": 0.7565927654609101,
      "grad_norm": 0.96875,
      "learning_rate": 0.000197218979542133,
      "loss": 0.4701,
      "step": 1621
    },
    {
      "epoch": 0.7570595099183197,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001972155420485237,
      "loss": 0.4606,
      "step": 1622
    },
    {
      "epoch": 0.7575262543757293,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019721210246175595,
      "loss": 0.4542,
      "step": 1623
    },
    {
      "epoch": 0.7579929988331389,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001972086607819038,
      "loss": 0.5826,
      "step": 1624
    },
    {
      "epoch": 0.7584597432905484,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001972052170090413,
      "loss": 0.4587,
      "step": 1625
    },
    {
      "epoch": 0.758926487747958,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019720177114324262,
      "loss": 0.3716,
      "step": 1626
    },
    {
      "epoch": 0.7593932322053676,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019719832318458198,
      "loss": 0.4412,
      "step": 1627
    },
    {
      "epoch": 0.7598599766627772,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001971948731331336,
      "loss": 0.526,
      "step": 1628
    },
    {
      "epoch": 0.7603267211201867,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019719142098897173,
      "loss": 0.4982,
      "step": 1629
    },
    {
      "epoch": 0.7607934655775963,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019718796675217073,
      "loss": 0.489,
      "step": 1630
    },
    {
      "epoch": 0.7612602100350059,
      "grad_norm": 0.80859375,
      "learning_rate": 0.000197184510422805,
      "loss": 0.4201,
      "step": 1631
    },
    {
      "epoch": 0.7617269544924155,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019718105200094895,
      "loss": 0.3797,
      "step": 1632
    },
    {
      "epoch": 0.7621936989498249,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019717759148667698,
      "loss": 0.4649,
      "step": 1633
    },
    {
      "epoch": 0.7626604434072345,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019717412888006365,
      "loss": 0.5103,
      "step": 1634
    },
    {
      "epoch": 0.7631271878646441,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019717066418118356,
      "loss": 0.4454,
      "step": 1635
    },
    {
      "epoch": 0.7635939323220536,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001971671973901112,
      "loss": 0.3624,
      "step": 1636
    },
    {
      "epoch": 0.7640606767794632,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001971637285069213,
      "loss": 0.379,
      "step": 1637
    },
    {
      "epoch": 0.7645274212368728,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001971602575316885,
      "loss": 0.3831,
      "step": 1638
    },
    {
      "epoch": 0.7649941656942824,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001971567844644876,
      "loss": 0.4298,
      "step": 1639
    },
    {
      "epoch": 0.7654609101516919,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001971533093053933,
      "loss": 0.4778,
      "step": 1640
    },
    {
      "epoch": 0.7659276546091015,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019714983205448044,
      "loss": 0.4568,
      "step": 1641
    },
    {
      "epoch": 0.7663943990665111,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019714635271182393,
      "loss": 0.5514,
      "step": 1642
    },
    {
      "epoch": 0.7668611435239207,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019714287127749867,
      "loss": 0.4077,
      "step": 1643
    },
    {
      "epoch": 0.7673278879813302,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001971393877515796,
      "loss": 0.4071,
      "step": 1644
    },
    {
      "epoch": 0.7677946324387398,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019713590213414172,
      "loss": 0.4482,
      "step": 1645
    },
    {
      "epoch": 0.7682613768961494,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019713241442526012,
      "loss": 0.4094,
      "step": 1646
    },
    {
      "epoch": 0.768728121353559,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019712892462500983,
      "loss": 0.4571,
      "step": 1647
    },
    {
      "epoch": 0.7691948658109685,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019712543273346604,
      "loss": 0.419,
      "step": 1648
    },
    {
      "epoch": 0.7696616102683781,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019712193875070396,
      "loss": 0.4194,
      "step": 1649
    },
    {
      "epoch": 0.7701283547257877,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019711844267679873,
      "loss": 0.4515,
      "step": 1650
    },
    {
      "epoch": 0.7705950991831972,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001971149445118257,
      "loss": 0.4023,
      "step": 1651
    },
    {
      "epoch": 0.7710618436406068,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019711144425586016,
      "loss": 0.5049,
      "step": 1652
    },
    {
      "epoch": 0.7715285880980164,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019710794190897746,
      "loss": 0.3869,
      "step": 1653
    },
    {
      "epoch": 0.7719953325554259,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019710443747125305,
      "loss": 0.4698,
      "step": 1654
    },
    {
      "epoch": 0.7724620770128354,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019710093094276235,
      "loss": 0.3737,
      "step": 1655
    },
    {
      "epoch": 0.772928821470245,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019709742232358086,
      "loss": 0.4001,
      "step": 1656
    },
    {
      "epoch": 0.7733955659276546,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019709391161378415,
      "loss": 0.4949,
      "step": 1657
    },
    {
      "epoch": 0.7738623103850641,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019709039881344781,
      "loss": 0.3657,
      "step": 1658
    },
    {
      "epoch": 0.7743290548424737,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019708688392264744,
      "loss": 0.4652,
      "step": 1659
    },
    {
      "epoch": 0.7747957992998833,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019708336694145874,
      "loss": 0.4408,
      "step": 1660
    },
    {
      "epoch": 0.7752625437572929,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019707984786995745,
      "loss": 0.3676,
      "step": 1661
    },
    {
      "epoch": 0.7757292882147024,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001970763267082193,
      "loss": 0.3739,
      "step": 1662
    },
    {
      "epoch": 0.776196032672112,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019707280345632015,
      "loss": 0.4358,
      "step": 1663
    },
    {
      "epoch": 0.7766627771295216,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001970692781143358,
      "loss": 0.3696,
      "step": 1664
    },
    {
      "epoch": 0.7771295215869312,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019706575068234225,
      "loss": 0.3856,
      "step": 1665
    },
    {
      "epoch": 0.7775962660443407,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019706222116041536,
      "loss": 0.4888,
      "step": 1666
    },
    {
      "epoch": 0.7780630105017503,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019705868954863115,
      "loss": 0.4437,
      "step": 1667
    },
    {
      "epoch": 0.7785297549591599,
      "grad_norm": 0.875,
      "learning_rate": 0.00019705515584706568,
      "loss": 0.4813,
      "step": 1668
    },
    {
      "epoch": 0.7789964994165695,
      "grad_norm": 1.015625,
      "learning_rate": 0.000197051620055795,
      "loss": 0.4835,
      "step": 1669
    },
    {
      "epoch": 0.779463243873979,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019704808217489527,
      "loss": 0.4474,
      "step": 1670
    },
    {
      "epoch": 0.7799299883313886,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019704454220444267,
      "loss": 0.4925,
      "step": 1671
    },
    {
      "epoch": 0.7803967327887982,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019704100014451337,
      "loss": 0.3988,
      "step": 1672
    },
    {
      "epoch": 0.7808634772462077,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001970374559951837,
      "loss": 0.3584,
      "step": 1673
    },
    {
      "epoch": 0.7813302217036173,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019703390975652993,
      "loss": 0.4052,
      "step": 1674
    },
    {
      "epoch": 0.7817969661610268,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001970303614286284,
      "loss": 0.4827,
      "step": 1675
    },
    {
      "epoch": 0.7822637106184364,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019702681101155557,
      "loss": 0.4047,
      "step": 1676
    },
    {
      "epoch": 0.7827304550758459,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001970232585053878,
      "loss": 0.4237,
      "step": 1677
    },
    {
      "epoch": 0.7831971995332555,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019701970391020166,
      "loss": 0.4886,
      "step": 1678
    },
    {
      "epoch": 0.7836639439906651,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019701614722607364,
      "loss": 0.4002,
      "step": 1679
    },
    {
      "epoch": 0.7841306884480747,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019701258845308036,
      "loss": 0.3882,
      "step": 1680
    },
    {
      "epoch": 0.7845974329054842,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019700902759129836,
      "loss": 0.4207,
      "step": 1681
    },
    {
      "epoch": 0.7850641773628938,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001970054646408044,
      "loss": 0.3959,
      "step": 1682
    },
    {
      "epoch": 0.7855309218203034,
      "grad_norm": 0.875,
      "learning_rate": 0.00019700189960167514,
      "loss": 0.4273,
      "step": 1683
    },
    {
      "epoch": 0.785997666277713,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019699833247398735,
      "loss": 0.3282,
      "step": 1684
    },
    {
      "epoch": 0.7864644107351225,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019699476325781785,
      "loss": 0.4213,
      "step": 1685
    },
    {
      "epoch": 0.7869311551925321,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019699119195324346,
      "loss": 0.4118,
      "step": 1686
    },
    {
      "epoch": 0.7873978996499417,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001969876185603411,
      "loss": 0.3371,
      "step": 1687
    },
    {
      "epoch": 0.7878646441073512,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019698404307918774,
      "loss": 0.4357,
      "step": 1688
    },
    {
      "epoch": 0.7883313885647608,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001969804655098603,
      "loss": 0.4939,
      "step": 1689
    },
    {
      "epoch": 0.7887981330221704,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019697688585243583,
      "loss": 0.3687,
      "step": 1690
    },
    {
      "epoch": 0.78926487747958,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001969733041069914,
      "loss": 0.4065,
      "step": 1691
    },
    {
      "epoch": 0.7897316219369895,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019696972027360418,
      "loss": 0.3539,
      "step": 1692
    },
    {
      "epoch": 0.7901983663943991,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019696613435235126,
      "loss": 0.4587,
      "step": 1693
    },
    {
      "epoch": 0.7906651108518087,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019696254634330988,
      "loss": 0.4311,
      "step": 1694
    },
    {
      "epoch": 0.7911318553092183,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019695895624655731,
      "loss": 0.3925,
      "step": 1695
    },
    {
      "epoch": 0.7915985997666277,
      "grad_norm": 0.75,
      "learning_rate": 0.00019695536406217083,
      "loss": 0.3238,
      "step": 1696
    },
    {
      "epoch": 0.7920653442240373,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001969517697902278,
      "loss": 0.3983,
      "step": 1697
    },
    {
      "epoch": 0.7925320886814469,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001969481734308056,
      "loss": 0.3313,
      "step": 1698
    },
    {
      "epoch": 0.7929988331388564,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019694457498398163,
      "loss": 0.4683,
      "step": 1699
    },
    {
      "epoch": 0.793465577596266,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001969409744498334,
      "loss": 0.3512,
      "step": 1700
    },
    {
      "epoch": 0.7939323220536756,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019693737182843846,
      "loss": 0.4548,
      "step": 1701
    },
    {
      "epoch": 0.7943990665110852,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019693376711987432,
      "loss": 0.4215,
      "step": 1702
    },
    {
      "epoch": 0.7948658109684947,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019693016032421864,
      "loss": 0.5173,
      "step": 1703
    },
    {
      "epoch": 0.7953325554259043,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019692655144154907,
      "loss": 0.4404,
      "step": 1704
    },
    {
      "epoch": 0.7957992998833139,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001969229404719433,
      "loss": 0.4476,
      "step": 1705
    },
    {
      "epoch": 0.7962660443407235,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019691932741547908,
      "loss": 0.397,
      "step": 1706
    },
    {
      "epoch": 0.796732788798133,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001969157122722342,
      "loss": 0.4198,
      "step": 1707
    },
    {
      "epoch": 0.7971995332555426,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019691209504228653,
      "loss": 0.5072,
      "step": 1708
    },
    {
      "epoch": 0.7976662777129522,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001969084757257139,
      "loss": 0.4204,
      "step": 1709
    },
    {
      "epoch": 0.7981330221703618,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001969048543225943,
      "loss": 0.3841,
      "step": 1710
    },
    {
      "epoch": 0.7985997666277713,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019690123083300563,
      "loss": 0.4777,
      "step": 1711
    },
    {
      "epoch": 0.7990665110851809,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019689760525702594,
      "loss": 0.4967,
      "step": 1712
    },
    {
      "epoch": 0.7995332555425905,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00019689397759473334,
      "loss": 0.365,
      "step": 1713
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019689034784620585,
      "loss": 0.5012,
      "step": 1714
    },
    {
      "epoch": 0.8004667444574096,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001968867160115217,
      "loss": 0.4253,
      "step": 1715
    },
    {
      "epoch": 0.8009334889148192,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019688308209075907,
      "loss": 0.3764,
      "step": 1716
    },
    {
      "epoch": 0.8014002333722287,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019687944608399617,
      "loss": 0.4565,
      "step": 1717
    },
    {
      "epoch": 0.8018669778296382,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001968758079913113,
      "loss": 0.363,
      "step": 1718
    },
    {
      "epoch": 0.8023337222870478,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001968721678127828,
      "loss": 0.5048,
      "step": 1719
    },
    {
      "epoch": 0.8028004667444574,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019686852554848906,
      "loss": 0.4447,
      "step": 1720
    },
    {
      "epoch": 0.803267211201867,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001968648811985085,
      "loss": 0.4441,
      "step": 1721
    },
    {
      "epoch": 0.8037339556592765,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019686123476291955,
      "loss": 0.3931,
      "step": 1722
    },
    {
      "epoch": 0.8042007001166861,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019685758624180073,
      "loss": 0.4077,
      "step": 1723
    },
    {
      "epoch": 0.8046674445740957,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019685393563523065,
      "loss": 0.458,
      "step": 1724
    },
    {
      "epoch": 0.8051341890315052,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019685028294328785,
      "loss": 0.4118,
      "step": 1725
    },
    {
      "epoch": 0.8056009334889148,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019684662816605103,
      "loss": 0.4255,
      "step": 1726
    },
    {
      "epoch": 0.8060676779463244,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001968429713035988,
      "loss": 0.4846,
      "step": 1727
    },
    {
      "epoch": 0.806534422403734,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019683931235600998,
      "loss": 0.4882,
      "step": 1728
    },
    {
      "epoch": 0.8070011668611435,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019683565132336335,
      "loss": 0.4408,
      "step": 1729
    },
    {
      "epoch": 0.8074679113185531,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019683198820573768,
      "loss": 0.4586,
      "step": 1730
    },
    {
      "epoch": 0.8079346557759627,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019682832300321184,
      "loss": 0.3996,
      "step": 1731
    },
    {
      "epoch": 0.8084014002333723,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019682465571586478,
      "loss": 0.3814,
      "step": 1732
    },
    {
      "epoch": 0.8088681446907818,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019682098634377548,
      "loss": 0.2916,
      "step": 1733
    },
    {
      "epoch": 0.8093348891481914,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019681731488702293,
      "loss": 0.4154,
      "step": 1734
    },
    {
      "epoch": 0.809801633605601,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019681364134568611,
      "loss": 0.3865,
      "step": 1735
    },
    {
      "epoch": 0.8102683780630106,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019680996571984425,
      "loss": 0.3899,
      "step": 1736
    },
    {
      "epoch": 0.81073512252042,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019680628800957637,
      "loss": 0.3522,
      "step": 1737
    },
    {
      "epoch": 0.8112018669778296,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001968026082149617,
      "loss": 0.3967,
      "step": 1738
    },
    {
      "epoch": 0.8116686114352392,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001967989263360795,
      "loss": 0.4102,
      "step": 1739
    },
    {
      "epoch": 0.8121353558926487,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019679524237300897,
      "loss": 0.4287,
      "step": 1740
    },
    {
      "epoch": 0.8126021003500583,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019679155632582954,
      "loss": 0.4413,
      "step": 1741
    },
    {
      "epoch": 0.8130688448074679,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019678786819462041,
      "loss": 0.5034,
      "step": 1742
    },
    {
      "epoch": 0.8135355892648775,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019678417797946117,
      "loss": 0.4361,
      "step": 1743
    },
    {
      "epoch": 0.814002333722287,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001967804856804312,
      "loss": 0.3384,
      "step": 1744
    },
    {
      "epoch": 0.8144690781796966,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019677679129760994,
      "loss": 0.3776,
      "step": 1745
    },
    {
      "epoch": 0.8149358226371062,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019677309483107704,
      "loss": 0.4965,
      "step": 1746
    },
    {
      "epoch": 0.8154025670945158,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019676939628091202,
      "loss": 0.4591,
      "step": 1747
    },
    {
      "epoch": 0.8158693115519253,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001967656956471945,
      "loss": 0.4155,
      "step": 1748
    },
    {
      "epoch": 0.8163360560093349,
      "grad_norm": 0.71875,
      "learning_rate": 0.00019676199293000424,
      "loss": 0.3471,
      "step": 1749
    },
    {
      "epoch": 0.8168028004667445,
      "grad_norm": 0.875,
      "learning_rate": 0.00019675828812942088,
      "loss": 0.4256,
      "step": 1750
    },
    {
      "epoch": 0.817269544924154,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019675458124552423,
      "loss": 0.4129,
      "step": 1751
    },
    {
      "epoch": 0.8177362893815636,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019675087227839409,
      "loss": 0.3291,
      "step": 1752
    },
    {
      "epoch": 0.8182030338389732,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001967471612281103,
      "loss": 0.3793,
      "step": 1753
    },
    {
      "epoch": 0.8186697782963828,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019674344809475285,
      "loss": 0.4082,
      "step": 1754
    },
    {
      "epoch": 0.8191365227537923,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019673973287840158,
      "loss": 0.3314,
      "step": 1755
    },
    {
      "epoch": 0.8196032672112019,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001967360155791365,
      "loss": 0.4393,
      "step": 1756
    },
    {
      "epoch": 0.8200700116686115,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019673229619703774,
      "loss": 0.533,
      "step": 1757
    },
    {
      "epoch": 0.820536756126021,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019672857473218526,
      "loss": 0.4376,
      "step": 1758
    },
    {
      "epoch": 0.8210035005834305,
      "grad_norm": 1.0,
      "learning_rate": 0.00019672485118465928,
      "loss": 0.387,
      "step": 1759
    },
    {
      "epoch": 0.8214702450408401,
      "grad_norm": 0.875,
      "learning_rate": 0.00019672112555453993,
      "loss": 0.4112,
      "step": 1760
    },
    {
      "epoch": 0.8219369894982497,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001967173978419074,
      "loss": 0.4057,
      "step": 1761
    },
    {
      "epoch": 0.8224037339556592,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019671366804684205,
      "loss": 0.4317,
      "step": 1762
    },
    {
      "epoch": 0.8228704784130688,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019670993616942406,
      "loss": 0.4131,
      "step": 1763
    },
    {
      "epoch": 0.8233372228704784,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019670620220973387,
      "loss": 0.4334,
      "step": 1764
    },
    {
      "epoch": 0.823803967327888,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001967024661678518,
      "loss": 0.3953,
      "step": 1765
    },
    {
      "epoch": 0.8242707117852975,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00019669872804385841,
      "loss": 0.3453,
      "step": 1766
    },
    {
      "epoch": 0.8247374562427071,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001966949878378341,
      "loss": 0.4282,
      "step": 1767
    },
    {
      "epoch": 0.8252042007001167,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001966912455498594,
      "loss": 0.4296,
      "step": 1768
    },
    {
      "epoch": 0.8256709451575263,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001966875011800149,
      "loss": 0.3504,
      "step": 1769
    },
    {
      "epoch": 0.8261376896149358,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019668375472838126,
      "loss": 0.4587,
      "step": 1770
    },
    {
      "epoch": 0.8266044340723454,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019668000619503905,
      "loss": 0.5271,
      "step": 1771
    },
    {
      "epoch": 0.827071178529755,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019667625558006908,
      "loss": 0.4177,
      "step": 1772
    },
    {
      "epoch": 0.8275379229871646,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019667250288355205,
      "loss": 0.5163,
      "step": 1773
    },
    {
      "epoch": 0.8280046674445741,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019666874810556878,
      "loss": 0.3354,
      "step": 1774
    },
    {
      "epoch": 0.8284714119019837,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001966649912462001,
      "loss": 0.3692,
      "step": 1775
    },
    {
      "epoch": 0.8289381563593933,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019666123230552689,
      "loss": 0.4289,
      "step": 1776
    },
    {
      "epoch": 0.8294049008168028,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019665747128363013,
      "loss": 0.446,
      "step": 1777
    },
    {
      "epoch": 0.8298716452742124,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019665370818059078,
      "loss": 0.3976,
      "step": 1778
    },
    {
      "epoch": 0.8303383897316219,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019664994299648983,
      "loss": 0.417,
      "step": 1779
    },
    {
      "epoch": 0.8308051341890315,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019664617573140835,
      "loss": 0.3878,
      "step": 1780
    },
    {
      "epoch": 0.831271878646441,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019664240638542753,
      "loss": 0.4081,
      "step": 1781
    },
    {
      "epoch": 0.8317386231038506,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001966386349586284,
      "loss": 0.4148,
      "step": 1782
    },
    {
      "epoch": 0.8322053675612602,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019663486145109232,
      "loss": 0.5447,
      "step": 1783
    },
    {
      "epoch": 0.8326721120186698,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001966310858629004,
      "loss": 0.3902,
      "step": 1784
    },
    {
      "epoch": 0.8331388564760793,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019662730819413403,
      "loss": 0.3607,
      "step": 1785
    },
    {
      "epoch": 0.8336056009334889,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019662352844487447,
      "loss": 0.3457,
      "step": 1786
    },
    {
      "epoch": 0.8340723453908985,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019661974661520313,
      "loss": 0.356,
      "step": 1787
    },
    {
      "epoch": 0.834539089848308,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019661596270520147,
      "loss": 0.3302,
      "step": 1788
    },
    {
      "epoch": 0.8350058343057176,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019661217671495092,
      "loss": 0.3902,
      "step": 1789
    },
    {
      "epoch": 0.8354725787631272,
      "grad_norm": 1.3046875,
      "learning_rate": 0.000196608388644533,
      "loss": 0.4734,
      "step": 1790
    },
    {
      "epoch": 0.8359393232205368,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019660459849402934,
      "loss": 0.388,
      "step": 1791
    },
    {
      "epoch": 0.8364060676779463,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019660080626352145,
      "loss": 0.3567,
      "step": 1792
    },
    {
      "epoch": 0.8368728121353559,
      "grad_norm": 0.875,
      "learning_rate": 0.000196597011953091,
      "loss": 0.3428,
      "step": 1793
    },
    {
      "epoch": 0.8373395565927655,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019659321556281972,
      "loss": 0.4481,
      "step": 1794
    },
    {
      "epoch": 0.8378063010501751,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019658941709278934,
      "loss": 0.4069,
      "step": 1795
    },
    {
      "epoch": 0.8382730455075846,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001965856165430817,
      "loss": 0.3826,
      "step": 1796
    },
    {
      "epoch": 0.8387397899649942,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00019658181391377848,
      "loss": 0.2763,
      "step": 1797
    },
    {
      "epoch": 0.8392065344224038,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001965780092049617,
      "loss": 0.3585,
      "step": 1798
    },
    {
      "epoch": 0.8396732788798134,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001965742024167132,
      "loss": 0.3959,
      "step": 1799
    },
    {
      "epoch": 0.8401400233372228,
      "grad_norm": 1.046875,
      "learning_rate": 0.000196570393549115,
      "loss": 0.4705,
      "step": 1800
    },
    {
      "epoch": 0.8406067677946324,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019656658260224904,
      "loss": 0.4451,
      "step": 1801
    },
    {
      "epoch": 0.841073512252042,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019656276957619742,
      "loss": 0.4259,
      "step": 1802
    },
    {
      "epoch": 0.8415402567094515,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019655895447104227,
      "loss": 0.4195,
      "step": 1803
    },
    {
      "epoch": 0.8420070011668611,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001965551372868657,
      "loss": 0.3651,
      "step": 1804
    },
    {
      "epoch": 0.8424737456242707,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019655131802374985,
      "loss": 0.3945,
      "step": 1805
    },
    {
      "epoch": 0.8429404900816803,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019654749668177702,
      "loss": 0.3596,
      "step": 1806
    },
    {
      "epoch": 0.8434072345390898,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019654367326102948,
      "loss": 0.3782,
      "step": 1807
    },
    {
      "epoch": 0.8438739789964994,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019653984776158952,
      "loss": 0.3764,
      "step": 1808
    },
    {
      "epoch": 0.844340723453909,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019653602018353952,
      "loss": 0.3736,
      "step": 1809
    },
    {
      "epoch": 0.8448074679113186,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001965321905269619,
      "loss": 0.325,
      "step": 1810
    },
    {
      "epoch": 0.8452742123687281,
      "grad_norm": 0.6875,
      "learning_rate": 0.00019652835879193912,
      "loss": 0.3127,
      "step": 1811
    },
    {
      "epoch": 0.8457409568261377,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019652452497855366,
      "loss": 0.4186,
      "step": 1812
    },
    {
      "epoch": 0.8462077012835473,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019652068908688813,
      "loss": 0.4679,
      "step": 1813
    },
    {
      "epoch": 0.8466744457409568,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000196516851117025,
      "loss": 0.3121,
      "step": 1814
    },
    {
      "epoch": 0.8471411901983664,
      "grad_norm": 1.2265625,
      "learning_rate": 0.000196513011069047,
      "loss": 0.3522,
      "step": 1815
    },
    {
      "epoch": 0.847607934655776,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019650916894303682,
      "loss": 0.3794,
      "step": 1816
    },
    {
      "epoch": 0.8480746791131856,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019650532473907716,
      "loss": 0.4495,
      "step": 1817
    },
    {
      "epoch": 0.8485414235705951,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019650147845725076,
      "loss": 0.3639,
      "step": 1818
    },
    {
      "epoch": 0.8490081680280047,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019649763009764044,
      "loss": 0.3879,
      "step": 1819
    },
    {
      "epoch": 0.8494749124854143,
      "grad_norm": 0.78125,
      "learning_rate": 0.00019649377966032909,
      "loss": 0.3394,
      "step": 1820
    },
    {
      "epoch": 0.8499416569428238,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019648992714539962,
      "loss": 0.3928,
      "step": 1821
    },
    {
      "epoch": 0.8504084014002333,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019648607255293493,
      "loss": 0.3423,
      "step": 1822
    },
    {
      "epoch": 0.8508751458576429,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001964822158830181,
      "loss": 0.4007,
      "step": 1823
    },
    {
      "epoch": 0.8513418903150525,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019647835713573208,
      "loss": 0.3958,
      "step": 1824
    },
    {
      "epoch": 0.851808634772462,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019647449631115998,
      "loss": 0.3731,
      "step": 1825
    },
    {
      "epoch": 0.8522753792298716,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019647063340938491,
      "loss": 0.324,
      "step": 1826
    },
    {
      "epoch": 0.8527421236872812,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019646676843049012,
      "loss": 0.4611,
      "step": 1827
    },
    {
      "epoch": 0.8532088681446908,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019646290137455874,
      "loss": 0.4185,
      "step": 1828
    },
    {
      "epoch": 0.8536756126021003,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019645903224167408,
      "loss": 0.3925,
      "step": 1829
    },
    {
      "epoch": 0.8541423570595099,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019645516103191944,
      "loss": 0.3905,
      "step": 1830
    },
    {
      "epoch": 0.8546091015169195,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019645128774537816,
      "loss": 0.3788,
      "step": 1831
    },
    {
      "epoch": 0.8550758459743291,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019644741238213363,
      "loss": 0.4176,
      "step": 1832
    },
    {
      "epoch": 0.8555425904317386,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001964435349422693,
      "loss": 0.3912,
      "step": 1833
    },
    {
      "epoch": 0.8560093348891482,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019643965542586863,
      "loss": 0.4296,
      "step": 1834
    },
    {
      "epoch": 0.8564760793465578,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019643577383301523,
      "loss": 0.4221,
      "step": 1835
    },
    {
      "epoch": 0.8569428238039674,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019643189016379262,
      "loss": 0.3477,
      "step": 1836
    },
    {
      "epoch": 0.8574095682613769,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019642800441828438,
      "loss": 0.3722,
      "step": 1837
    },
    {
      "epoch": 0.8578763127187865,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019642411659657425,
      "loss": 0.4089,
      "step": 1838
    },
    {
      "epoch": 0.8583430571761961,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001964202266987459,
      "loss": 0.3811,
      "step": 1839
    },
    {
      "epoch": 0.8588098016336057,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019641633472488308,
      "loss": 0.3723,
      "step": 1840
    },
    {
      "epoch": 0.8592765460910151,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001964124406750696,
      "loss": 0.3523,
      "step": 1841
    },
    {
      "epoch": 0.8597432905484247,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019640854454938927,
      "loss": 0.3542,
      "step": 1842
    },
    {
      "epoch": 0.8602100350058343,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019640464634792607,
      "loss": 0.3185,
      "step": 1843
    },
    {
      "epoch": 0.8606767794632438,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019640074607076385,
      "loss": 0.3849,
      "step": 1844
    },
    {
      "epoch": 0.8611435239206534,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019639684371798656,
      "loss": 0.4132,
      "step": 1845
    },
    {
      "epoch": 0.861610268378063,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019639293928967836,
      "loss": 0.4004,
      "step": 1846
    },
    {
      "epoch": 0.8620770128354726,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019638903278592317,
      "loss": 0.3093,
      "step": 1847
    },
    {
      "epoch": 0.8625437572928821,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019638512420680514,
      "loss": 0.4005,
      "step": 1848
    },
    {
      "epoch": 0.8630105017502917,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019638121355240848,
      "loss": 0.3725,
      "step": 1849
    },
    {
      "epoch": 0.8634772462077013,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019637730082281736,
      "loss": 0.3596,
      "step": 1850
    },
    {
      "epoch": 0.8639439906651109,
      "grad_norm": 0.859375,
      "learning_rate": 0.000196373386018116,
      "loss": 0.3901,
      "step": 1851
    },
    {
      "epoch": 0.8644107351225204,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001963694691383887,
      "loss": 0.3743,
      "step": 1852
    },
    {
      "epoch": 0.86487747957993,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019636555018371984,
      "loss": 0.3376,
      "step": 1853
    },
    {
      "epoch": 0.8653442240373396,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019636162915419374,
      "loss": 0.4384,
      "step": 1854
    },
    {
      "epoch": 0.8658109684947491,
      "grad_norm": 1.0,
      "learning_rate": 0.00019635770604989485,
      "loss": 0.3543,
      "step": 1855
    },
    {
      "epoch": 0.8662777129521587,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019635378087090764,
      "loss": 0.4132,
      "step": 1856
    },
    {
      "epoch": 0.8667444574095683,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019634985361731663,
      "loss": 0.4052,
      "step": 1857
    },
    {
      "epoch": 0.8672112018669779,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019634592428920637,
      "loss": 0.4537,
      "step": 1858
    },
    {
      "epoch": 0.8676779463243874,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00019634199288666147,
      "loss": 0.4343,
      "step": 1859
    },
    {
      "epoch": 0.868144690781797,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019633805940976655,
      "loss": 0.3985,
      "step": 1860
    },
    {
      "epoch": 0.8686114352392066,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019633412385860634,
      "loss": 0.3521,
      "step": 1861
    },
    {
      "epoch": 0.869078179696616,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019633018623326556,
      "loss": 0.3505,
      "step": 1862
    },
    {
      "epoch": 0.8695449241540256,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019632624653382897,
      "loss": 0.3098,
      "step": 1863
    },
    {
      "epoch": 0.8700116686114352,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001963223047603814,
      "loss": 0.5098,
      "step": 1864
    },
    {
      "epoch": 0.8704784130688448,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001963183609130078,
      "loss": 0.4233,
      "step": 1865
    },
    {
      "epoch": 0.8709451575262543,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019631441499179297,
      "loss": 0.3872,
      "step": 1866
    },
    {
      "epoch": 0.8714119019836639,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019631046699682193,
      "loss": 0.3203,
      "step": 1867
    },
    {
      "epoch": 0.8718786464410735,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001963065169281797,
      "loss": 0.3691,
      "step": 1868
    },
    {
      "epoch": 0.8723453908984831,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019630256478595128,
      "loss": 0.429,
      "step": 1869
    },
    {
      "epoch": 0.8728121353558926,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019629861057022182,
      "loss": 0.3601,
      "step": 1870
    },
    {
      "epoch": 0.8732788798133022,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001962946542810764,
      "loss": 0.4536,
      "step": 1871
    },
    {
      "epoch": 0.8737456242707118,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019629069591860026,
      "loss": 0.3148,
      "step": 1872
    },
    {
      "epoch": 0.8742123687281214,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019628673548287856,
      "loss": 0.3374,
      "step": 1873
    },
    {
      "epoch": 0.8746791131855309,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019628277297399667,
      "loss": 0.4128,
      "step": 1874
    },
    {
      "epoch": 0.8751458576429405,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001962788083920398,
      "loss": 0.4096,
      "step": 1875
    },
    {
      "epoch": 0.8756126021003501,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019627484173709338,
      "loss": 0.3788,
      "step": 1876
    },
    {
      "epoch": 0.8760793465577597,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001962708730092428,
      "loss": 0.2991,
      "step": 1877
    },
    {
      "epoch": 0.8765460910151692,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001962669022085735,
      "loss": 0.4103,
      "step": 1878
    },
    {
      "epoch": 0.8770128354725788,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000196262929335171,
      "loss": 0.338,
      "step": 1879
    },
    {
      "epoch": 0.8774795799299884,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001962589543891208,
      "loss": 0.2937,
      "step": 1880
    },
    {
      "epoch": 0.877946324387398,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019625497737050856,
      "loss": 0.3802,
      "step": 1881
    },
    {
      "epoch": 0.8784130688448075,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019625099827941985,
      "loss": 0.3681,
      "step": 1882
    },
    {
      "epoch": 0.878879813302217,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019624701711594032,
      "loss": 0.3591,
      "step": 1883
    },
    {
      "epoch": 0.8793465577596266,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019624303388015576,
      "loss": 0.3714,
      "step": 1884
    },
    {
      "epoch": 0.8798133022170361,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019623904857215185,
      "loss": 0.3315,
      "step": 1885
    },
    {
      "epoch": 0.8802800466744457,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019623506119201446,
      "loss": 0.4366,
      "step": 1886
    },
    {
      "epoch": 0.8807467911318553,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019623107173982946,
      "loss": 0.3618,
      "step": 1887
    },
    {
      "epoch": 0.8812135355892649,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001962270802156827,
      "loss": 0.431,
      "step": 1888
    },
    {
      "epoch": 0.8816802800466744,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001962230866196601,
      "loss": 0.3255,
      "step": 1889
    },
    {
      "epoch": 0.882147024504084,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019621909095184775,
      "loss": 0.3385,
      "step": 1890
    },
    {
      "epoch": 0.8826137689614936,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001962150932123316,
      "loss": 0.3374,
      "step": 1891
    },
    {
      "epoch": 0.8830805134189031,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019621109340119767,
      "loss": 0.4671,
      "step": 1892
    },
    {
      "epoch": 0.8835472578763127,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001962070915185322,
      "loss": 0.4194,
      "step": 1893
    },
    {
      "epoch": 0.8840140023337223,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001962030875644213,
      "loss": 0.4178,
      "step": 1894
    },
    {
      "epoch": 0.8844807467911319,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019619908153895122,
      "loss": 0.366,
      "step": 1895
    },
    {
      "epoch": 0.8849474912485414,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019619507344220815,
      "loss": 0.3652,
      "step": 1896
    },
    {
      "epoch": 0.885414235705951,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019619106327427844,
      "loss": 0.3465,
      "step": 1897
    },
    {
      "epoch": 0.8858809801633606,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019618705103524836,
      "loss": 0.3645,
      "step": 1898
    },
    {
      "epoch": 0.8863477246207702,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001961830367252044,
      "loss": 0.3429,
      "step": 1899
    },
    {
      "epoch": 0.8868144690781797,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019617902034423294,
      "loss": 0.3952,
      "step": 1900
    },
    {
      "epoch": 0.8872812135355893,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019617500189242045,
      "loss": 0.3707,
      "step": 1901
    },
    {
      "epoch": 0.8877479579929989,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019617098136985345,
      "loss": 0.4329,
      "step": 1902
    },
    {
      "epoch": 0.8882147024504085,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019616695877661851,
      "loss": 0.3569,
      "step": 1903
    },
    {
      "epoch": 0.8886814469078179,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019616293411280227,
      "loss": 0.3814,
      "step": 1904
    },
    {
      "epoch": 0.8891481913652275,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019615890737849137,
      "loss": 0.3459,
      "step": 1905
    },
    {
      "epoch": 0.8896149358226371,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019615487857377247,
      "loss": 0.394,
      "step": 1906
    },
    {
      "epoch": 0.8900816802800466,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001961508476987324,
      "loss": 0.3676,
      "step": 1907
    },
    {
      "epoch": 0.8905484247374562,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019614681475345783,
      "loss": 0.3543,
      "step": 1908
    },
    {
      "epoch": 0.8910151691948658,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019614277973803573,
      "loss": 0.3679,
      "step": 1909
    },
    {
      "epoch": 0.8914819136522754,
      "grad_norm": 1.0,
      "learning_rate": 0.00019613874265255287,
      "loss": 0.369,
      "step": 1910
    },
    {
      "epoch": 0.8919486581096849,
      "grad_norm": 1.0,
      "learning_rate": 0.0001961347034970962,
      "loss": 0.3688,
      "step": 1911
    },
    {
      "epoch": 0.8924154025670945,
      "grad_norm": 0.875,
      "learning_rate": 0.00019613066227175275,
      "loss": 0.2986,
      "step": 1912
    },
    {
      "epoch": 0.8928821470245041,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019612661897660947,
      "loss": 0.3513,
      "step": 1913
    },
    {
      "epoch": 0.8933488914819137,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019612257361175338,
      "loss": 0.3795,
      "step": 1914
    },
    {
      "epoch": 0.8938156359393232,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019611852617727168,
      "loss": 0.3989,
      "step": 1915
    },
    {
      "epoch": 0.8942823803967328,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019611447667325143,
      "loss": 0.325,
      "step": 1916
    },
    {
      "epoch": 0.8947491248541424,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001961104250997799,
      "loss": 0.2928,
      "step": 1917
    },
    {
      "epoch": 0.895215869311552,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019610637145694426,
      "loss": 0.2861,
      "step": 1918
    },
    {
      "epoch": 0.8956826137689615,
      "grad_norm": 1.0,
      "learning_rate": 0.0001961023157448318,
      "loss": 0.3887,
      "step": 1919
    },
    {
      "epoch": 0.8961493582263711,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019609825796352987,
      "loss": 0.3222,
      "step": 1920
    },
    {
      "epoch": 0.8966161026837807,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019609419811312582,
      "loss": 0.3636,
      "step": 1921
    },
    {
      "epoch": 0.8970828471411902,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019609013619370706,
      "loss": 0.3066,
      "step": 1922
    },
    {
      "epoch": 0.8975495915985998,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019608607220536107,
      "loss": 0.4214,
      "step": 1923
    },
    {
      "epoch": 0.8980163360560094,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019608200614817532,
      "loss": 0.4406,
      "step": 1924
    },
    {
      "epoch": 0.8984830805134189,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019607793802223737,
      "loss": 0.3549,
      "step": 1925
    },
    {
      "epoch": 0.8989498249708284,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019607386782763482,
      "loss": 0.3276,
      "step": 1926
    },
    {
      "epoch": 0.899416569428238,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001960697955644553,
      "loss": 0.2895,
      "step": 1927
    },
    {
      "epoch": 0.8998833138856476,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001960657212327865,
      "loss": 0.3163,
      "step": 1928
    },
    {
      "epoch": 0.9003500583430571,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019606164483271612,
      "loss": 0.4171,
      "step": 1929
    },
    {
      "epoch": 0.9008168028004667,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00019605756636433195,
      "loss": 0.3123,
      "step": 1930
    },
    {
      "epoch": 0.9012835472578763,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001960534858277218,
      "loss": 0.3176,
      "step": 1931
    },
    {
      "epoch": 0.9017502917152859,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019604940322297352,
      "loss": 0.2874,
      "step": 1932
    },
    {
      "epoch": 0.9022170361726954,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019604531855017503,
      "loss": 0.3031,
      "step": 1933
    },
    {
      "epoch": 0.902683780630105,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019604123180941424,
      "loss": 0.3409,
      "step": 1934
    },
    {
      "epoch": 0.9031505250875146,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001960371430007792,
      "loss": 0.3479,
      "step": 1935
    },
    {
      "epoch": 0.9036172695449242,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00019603305212435792,
      "loss": 0.3327,
      "step": 1936
    },
    {
      "epoch": 0.9040840140023337,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019602895918023846,
      "loss": 0.3258,
      "step": 1937
    },
    {
      "epoch": 0.9045507584597433,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019602486416850896,
      "loss": 0.4291,
      "step": 1938
    },
    {
      "epoch": 0.9050175029171529,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019602076708925763,
      "loss": 0.275,
      "step": 1939
    },
    {
      "epoch": 0.9054842473745625,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019601666794257262,
      "loss": 0.3257,
      "step": 1940
    },
    {
      "epoch": 0.905950991831972,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001960125667285422,
      "loss": 0.3307,
      "step": 1941
    },
    {
      "epoch": 0.9064177362893816,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019600846344725473,
      "loss": 0.3245,
      "step": 1942
    },
    {
      "epoch": 0.9068844807467912,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019600435809879847,
      "loss": 0.3343,
      "step": 1943
    },
    {
      "epoch": 0.9073512252042008,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019600025068326188,
      "loss": 0.3594,
      "step": 1944
    },
    {
      "epoch": 0.9078179696616102,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001959961412007334,
      "loss": 0.257,
      "step": 1945
    },
    {
      "epoch": 0.9082847141190198,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019599202965130146,
      "loss": 0.2809,
      "step": 1946
    },
    {
      "epoch": 0.9087514585764294,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019598791603505465,
      "loss": 0.346,
      "step": 1947
    },
    {
      "epoch": 0.9092182030338389,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019598380035208148,
      "loss": 0.3983,
      "step": 1948
    },
    {
      "epoch": 0.9096849474912485,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001959796826024706,
      "loss": 0.4167,
      "step": 1949
    },
    {
      "epoch": 0.9101516919486581,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001959755627863107,
      "loss": 0.3533,
      "step": 1950
    },
    {
      "epoch": 0.9106184364060677,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001959714409036904,
      "loss": 0.2936,
      "step": 1951
    },
    {
      "epoch": 0.9110851808634772,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001959673169546985,
      "loss": 0.38,
      "step": 1952
    },
    {
      "epoch": 0.9115519253208868,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019596319093942376,
      "loss": 0.3085,
      "step": 1953
    },
    {
      "epoch": 0.9120186697782964,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001959590628579551,
      "loss": 0.3357,
      "step": 1954
    },
    {
      "epoch": 0.912485414235706,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00019595493271038134,
      "loss": 0.2537,
      "step": 1955
    },
    {
      "epoch": 0.9129521586931155,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001959508004967914,
      "loss": 0.3756,
      "step": 1956
    },
    {
      "epoch": 0.9134189031505251,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019594666621727426,
      "loss": 0.3032,
      "step": 1957
    },
    {
      "epoch": 0.9138856476079347,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019594252987191894,
      "loss": 0.2906,
      "step": 1958
    },
    {
      "epoch": 0.9143523920653442,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019593839146081452,
      "loss": 0.3394,
      "step": 1959
    },
    {
      "epoch": 0.9148191365227538,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001959342509840501,
      "loss": 0.3655,
      "step": 1960
    },
    {
      "epoch": 0.9152858809801634,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001959301084417148,
      "loss": 0.3079,
      "step": 1961
    },
    {
      "epoch": 0.915752625437573,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019592596383389782,
      "loss": 0.3948,
      "step": 1962
    },
    {
      "epoch": 0.9162193698949825,
      "grad_norm": 1.59375,
      "learning_rate": 0.0001959218171606884,
      "loss": 0.3659,
      "step": 1963
    },
    {
      "epoch": 0.9166861143523921,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019591766842217583,
      "loss": 0.3954,
      "step": 1964
    },
    {
      "epoch": 0.9171528588098017,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019591351761844944,
      "loss": 0.2578,
      "step": 1965
    },
    {
      "epoch": 0.9176196032672111,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019590936474959863,
      "loss": 0.253,
      "step": 1966
    },
    {
      "epoch": 0.9180863477246207,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019590520981571275,
      "loss": 0.3245,
      "step": 1967
    },
    {
      "epoch": 0.9185530921820303,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001959010528168813,
      "loss": 0.3099,
      "step": 1968
    },
    {
      "epoch": 0.9190198366394399,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001958968937531938,
      "loss": 0.39,
      "step": 1969
    },
    {
      "epoch": 0.9194865810968494,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019589273262473974,
      "loss": 0.4934,
      "step": 1970
    },
    {
      "epoch": 0.919953325554259,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001958885694316088,
      "loss": 0.2898,
      "step": 1971
    },
    {
      "epoch": 0.9204200700116686,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019588440417389052,
      "loss": 0.3874,
      "step": 1972
    },
    {
      "epoch": 0.9208868144690782,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019588023685167464,
      "loss": 0.3158,
      "step": 1973
    },
    {
      "epoch": 0.9213535589264877,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019587606746505093,
      "loss": 0.3268,
      "step": 1974
    },
    {
      "epoch": 0.9218203033838973,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019587189601410904,
      "loss": 0.3633,
      "step": 1975
    },
    {
      "epoch": 0.9222870478413069,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001958677224989389,
      "loss": 0.3194,
      "step": 1976
    },
    {
      "epoch": 0.9227537922987165,
      "grad_norm": 1.125,
      "learning_rate": 0.0001958635469196303,
      "loss": 0.3088,
      "step": 1977
    },
    {
      "epoch": 0.923220536756126,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001958593692762732,
      "loss": 0.2646,
      "step": 1978
    },
    {
      "epoch": 0.9236872812135356,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019585518956895752,
      "loss": 0.3572,
      "step": 1979
    },
    {
      "epoch": 0.9241540256709452,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00019585100779777326,
      "loss": 0.2942,
      "step": 1980
    },
    {
      "epoch": 0.9246207701283548,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019584682396281045,
      "loss": 0.3331,
      "step": 1981
    },
    {
      "epoch": 0.9250875145857643,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019584263806415915,
      "loss": 0.3521,
      "step": 1982
    },
    {
      "epoch": 0.9255542590431739,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019583845010190954,
      "loss": 0.3404,
      "step": 1983
    },
    {
      "epoch": 0.9260210035005835,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019583426007615176,
      "loss": 0.2842,
      "step": 1984
    },
    {
      "epoch": 0.926487747957993,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000195830067986976,
      "loss": 0.3825,
      "step": 1985
    },
    {
      "epoch": 0.9269544924154026,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019582587383447258,
      "loss": 0.3745,
      "step": 1986
    },
    {
      "epoch": 0.9274212368728121,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001958216776187318,
      "loss": 0.3401,
      "step": 1987
    },
    {
      "epoch": 0.9278879813302217,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019581747933984395,
      "loss": 0.3232,
      "step": 1988
    },
    {
      "epoch": 0.9283547257876312,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019581327899789947,
      "loss": 0.3038,
      "step": 1989
    },
    {
      "epoch": 0.9288214702450408,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019580907659298878,
      "loss": 0.3631,
      "step": 1990
    },
    {
      "epoch": 0.9292882147024504,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019580487212520238,
      "loss": 0.3289,
      "step": 1991
    },
    {
      "epoch": 0.92975495915986,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019580066559463078,
      "loss": 0.3263,
      "step": 1992
    },
    {
      "epoch": 0.9302217036172695,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019579645700136452,
      "loss": 0.3576,
      "step": 1993
    },
    {
      "epoch": 0.9306884480746791,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019579224634549432,
      "loss": 0.3478,
      "step": 1994
    },
    {
      "epoch": 0.9311551925320887,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019578803362711074,
      "loss": 0.4191,
      "step": 1995
    },
    {
      "epoch": 0.9316219369894982,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001957838188463045,
      "loss": 0.3105,
      "step": 1996
    },
    {
      "epoch": 0.9320886814469078,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001957796020031664,
      "loss": 0.401,
      "step": 1997
    },
    {
      "epoch": 0.9325554259043174,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001957753830977872,
      "loss": 0.3612,
      "step": 1998
    },
    {
      "epoch": 0.933022170361727,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001957711621302577,
      "loss": 0.2855,
      "step": 1999
    },
    {
      "epoch": 0.9334889148191365,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019576693910066883,
      "loss": 0.3532,
      "step": 2000
    },
    {
      "epoch": 0.9339556592765461,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019576271400911153,
      "loss": 0.3911,
      "step": 2001
    },
    {
      "epoch": 0.9344224037339557,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019575848685567674,
      "loss": 0.3296,
      "step": 2002
    },
    {
      "epoch": 0.9348891481913653,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019575425764045546,
      "loss": 0.3503,
      "step": 2003
    },
    {
      "epoch": 0.9353558926487748,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019575002636353878,
      "loss": 0.3773,
      "step": 2004
    },
    {
      "epoch": 0.9358226371061844,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019574579302501782,
      "loss": 0.3565,
      "step": 2005
    },
    {
      "epoch": 0.936289381563594,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019574155762498369,
      "loss": 0.3327,
      "step": 2006
    },
    {
      "epoch": 0.9367561260210036,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019573732016352758,
      "loss": 0.3242,
      "step": 2007
    },
    {
      "epoch": 0.937222870478413,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019573308064074077,
      "loss": 0.3316,
      "step": 2008
    },
    {
      "epoch": 0.9376896149358226,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001957288390567145,
      "loss": 0.339,
      "step": 2009
    },
    {
      "epoch": 0.9381563593932322,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019572459541154013,
      "loss": 0.2443,
      "step": 2010
    },
    {
      "epoch": 0.9386231038506417,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019572034970530896,
      "loss": 0.3403,
      "step": 2011
    },
    {
      "epoch": 0.9390898483080513,
      "grad_norm": 1.28125,
      "learning_rate": 0.00019571610193811254,
      "loss": 0.4092,
      "step": 2012
    },
    {
      "epoch": 0.9395565927654609,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001957118521100422,
      "loss": 0.3658,
      "step": 2013
    },
    {
      "epoch": 0.9400233372228705,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019570760022118947,
      "loss": 0.3343,
      "step": 2014
    },
    {
      "epoch": 0.94049008168028,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019570334627164594,
      "loss": 0.33,
      "step": 2015
    },
    {
      "epoch": 0.9409568261376896,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019569909026150318,
      "loss": 0.3458,
      "step": 2016
    },
    {
      "epoch": 0.9414235705950992,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019569483219085285,
      "loss": 0.3078,
      "step": 2017
    },
    {
      "epoch": 0.9418903150525088,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019569057205978658,
      "loss": 0.2643,
      "step": 2018
    },
    {
      "epoch": 0.9423570595099183,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019568630986839614,
      "loss": 0.2847,
      "step": 2019
    },
    {
      "epoch": 0.9428238039673279,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019568204561677325,
      "loss": 0.2386,
      "step": 2020
    },
    {
      "epoch": 0.9432905484247375,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001956777793050098,
      "loss": 0.3289,
      "step": 2021
    },
    {
      "epoch": 0.943757292882147,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019567351093319757,
      "loss": 0.3323,
      "step": 2022
    },
    {
      "epoch": 0.9442240373395566,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019566924050142852,
      "loss": 0.3465,
      "step": 2023
    },
    {
      "epoch": 0.9446907817969662,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019566496800979457,
      "loss": 0.2917,
      "step": 2024
    },
    {
      "epoch": 0.9451575262543758,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001956606934583877,
      "loss": 0.3368,
      "step": 2025
    },
    {
      "epoch": 0.9456242707117853,
      "grad_norm": 0.875,
      "learning_rate": 0.00019565641684729998,
      "loss": 0.271,
      "step": 2026
    },
    {
      "epoch": 0.9460910151691949,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019565213817662348,
      "loss": 0.2888,
      "step": 2027
    },
    {
      "epoch": 0.9465577596266045,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001956478574464503,
      "loss": 0.3175,
      "step": 2028
    },
    {
      "epoch": 0.947024504084014,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019564357465687263,
      "loss": 0.3547,
      "step": 2029
    },
    {
      "epoch": 0.9474912485414235,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019563928980798268,
      "loss": 0.2719,
      "step": 2030
    },
    {
      "epoch": 0.9479579929988331,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019563500289987272,
      "loss": 0.3066,
      "step": 2031
    },
    {
      "epoch": 0.9484247374562427,
      "grad_norm": 0.83984375,
      "learning_rate": 0.000195630713932635,
      "loss": 0.2879,
      "step": 2032
    },
    {
      "epoch": 0.9488914819136522,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019562642290636193,
      "loss": 0.4229,
      "step": 2033
    },
    {
      "epoch": 0.9493582263710618,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001956221298211459,
      "loss": 0.3117,
      "step": 2034
    },
    {
      "epoch": 0.9498249708284714,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001956178346770793,
      "loss": 0.3059,
      "step": 2035
    },
    {
      "epoch": 0.950291715285881,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001956135374742546,
      "loss": 0.4111,
      "step": 2036
    },
    {
      "epoch": 0.9507584597432905,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001956092382127644,
      "loss": 0.3413,
      "step": 2037
    },
    {
      "epoch": 0.9512252042007001,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001956049368927012,
      "loss": 0.389,
      "step": 2038
    },
    {
      "epoch": 0.9516919486581097,
      "grad_norm": 1.296875,
      "learning_rate": 0.00019560063351415762,
      "loss": 0.4335,
      "step": 2039
    },
    {
      "epoch": 0.9521586931155193,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019559632807722633,
      "loss": 0.3232,
      "step": 2040
    },
    {
      "epoch": 0.9526254375729288,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019559202058200005,
      "loss": 0.3001,
      "step": 2041
    },
    {
      "epoch": 0.9530921820303384,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019558771102857152,
      "loss": 0.3469,
      "step": 2042
    },
    {
      "epoch": 0.953558926487748,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001955833994170335,
      "loss": 0.3511,
      "step": 2043
    },
    {
      "epoch": 0.9540256709451576,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019557908574747884,
      "loss": 0.3203,
      "step": 2044
    },
    {
      "epoch": 0.9544924154025671,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001955747700200004,
      "loss": 0.4013,
      "step": 2045
    },
    {
      "epoch": 0.9549591598599767,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019557045223469113,
      "loss": 0.4068,
      "step": 2046
    },
    {
      "epoch": 0.9554259043173863,
      "grad_norm": 1.0390625,
      "learning_rate": 0.000195566132391644,
      "loss": 0.3385,
      "step": 2047
    },
    {
      "epoch": 0.9558926487747959,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019556181049095196,
      "loss": 0.3085,
      "step": 2048
    },
    {
      "epoch": 0.9563593932322053,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00019555748653270815,
      "loss": 0.3056,
      "step": 2049
    },
    {
      "epoch": 0.9568261376896149,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019555316051700564,
      "loss": 0.2945,
      "step": 2050
    },
    {
      "epoch": 0.9572928821470245,
      "grad_norm": 1.0,
      "learning_rate": 0.00019554883244393754,
      "loss": 0.2822,
      "step": 2051
    },
    {
      "epoch": 0.957759626604434,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001955445023135971,
      "loss": 0.32,
      "step": 2052
    },
    {
      "epoch": 0.9582263710618436,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001955401701260775,
      "loss": 0.2972,
      "step": 2053
    },
    {
      "epoch": 0.9586931155192532,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019553583588147202,
      "loss": 0.3058,
      "step": 2054
    },
    {
      "epoch": 0.9591598599766628,
      "grad_norm": 1.015625,
      "learning_rate": 0.000195531499579874,
      "loss": 0.3197,
      "step": 2055
    },
    {
      "epoch": 0.9596266044340723,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001955271612213768,
      "loss": 0.2878,
      "step": 2056
    },
    {
      "epoch": 0.9600933488914819,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019552282080607385,
      "loss": 0.2634,
      "step": 2057
    },
    {
      "epoch": 0.9605600933488915,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019551847833405855,
      "loss": 0.2922,
      "step": 2058
    },
    {
      "epoch": 0.961026837806301,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019551413380542446,
      "loss": 0.272,
      "step": 2059
    },
    {
      "epoch": 0.9614935822637106,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001955097872202651,
      "loss": 0.3444,
      "step": 2060
    },
    {
      "epoch": 0.9619603267211202,
      "grad_norm": 1.1953125,
      "learning_rate": 0.000195505438578674,
      "loss": 0.4296,
      "step": 2061
    },
    {
      "epoch": 0.9624270711785298,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001955010878807449,
      "loss": 0.376,
      "step": 2062
    },
    {
      "epoch": 0.9628938156359393,
      "grad_norm": 1.0,
      "learning_rate": 0.00019549673512657138,
      "loss": 0.3584,
      "step": 2063
    },
    {
      "epoch": 0.9633605600933489,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001954923803162472,
      "loss": 0.3325,
      "step": 2064
    },
    {
      "epoch": 0.9638273045507585,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019548802344986616,
      "loss": 0.3009,
      "step": 2065
    },
    {
      "epoch": 0.9642940490081681,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019548366452752197,
      "loss": 0.3958,
      "step": 2066
    },
    {
      "epoch": 0.9647607934655776,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001954793035493086,
      "loss": 0.3345,
      "step": 2067
    },
    {
      "epoch": 0.9652275379229872,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019547494051531987,
      "loss": 0.3474,
      "step": 2068
    },
    {
      "epoch": 0.9656942823803968,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019547057542564973,
      "loss": 0.3191,
      "step": 2069
    },
    {
      "epoch": 0.9661610268378062,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019546620828039216,
      "loss": 0.3123,
      "step": 2070
    },
    {
      "epoch": 0.9666277712952158,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001954618390796412,
      "loss": 0.299,
      "step": 2071
    },
    {
      "epoch": 0.9670945157526254,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019545746782349097,
      "loss": 0.4014,
      "step": 2072
    },
    {
      "epoch": 0.967561260210035,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019545309451203554,
      "loss": 0.3896,
      "step": 2073
    },
    {
      "epoch": 0.9680280046674445,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019544871914536907,
      "loss": 0.3294,
      "step": 2074
    },
    {
      "epoch": 0.9684947491248541,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019544434172358576,
      "loss": 0.3798,
      "step": 2075
    },
    {
      "epoch": 0.9689614935822637,
      "grad_norm": 0.875,
      "learning_rate": 0.00019543996224677987,
      "loss": 0.2892,
      "step": 2076
    },
    {
      "epoch": 0.9694282380396733,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019543558071504573,
      "loss": 0.3419,
      "step": 2077
    },
    {
      "epoch": 0.9698949824970828,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019543119712847764,
      "loss": 0.3206,
      "step": 2078
    },
    {
      "epoch": 0.9703617269544924,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019542681148716998,
      "loss": 0.3337,
      "step": 2079
    },
    {
      "epoch": 0.970828471411902,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019542242379121718,
      "loss": 0.3223,
      "step": 2080
    },
    {
      "epoch": 0.9712952158693116,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019541803404071376,
      "loss": 0.2642,
      "step": 2081
    },
    {
      "epoch": 0.9717619603267211,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019541364223575416,
      "loss": 0.3075,
      "step": 2082
    },
    {
      "epoch": 0.9722287047841307,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019540924837643299,
      "loss": 0.2677,
      "step": 2083
    },
    {
      "epoch": 0.9726954492415403,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019540485246284485,
      "loss": 0.3498,
      "step": 2084
    },
    {
      "epoch": 0.9731621936989499,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001954004544950844,
      "loss": 0.3377,
      "step": 2085
    },
    {
      "epoch": 0.9736289381563594,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001953960544732463,
      "loss": 0.343,
      "step": 2086
    },
    {
      "epoch": 0.974095682613769,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001953916523974253,
      "loss": 0.2828,
      "step": 2087
    },
    {
      "epoch": 0.9745624270711786,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001953872482677162,
      "loss": 0.305,
      "step": 2088
    },
    {
      "epoch": 0.9750291715285881,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019538284208421377,
      "loss": 0.2998,
      "step": 2089
    },
    {
      "epoch": 0.9754959159859977,
      "grad_norm": 0.953125,
      "learning_rate": 0.000195378433847013,
      "loss": 0.3505,
      "step": 2090
    },
    {
      "epoch": 0.9759626604434072,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019537402355620866,
      "loss": 0.333,
      "step": 2091
    },
    {
      "epoch": 0.9764294049008168,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019536961121189577,
      "loss": 0.3503,
      "step": 2092
    },
    {
      "epoch": 0.9768961493582263,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019536519681416937,
      "loss": 0.3627,
      "step": 2093
    },
    {
      "epoch": 0.9773628938156359,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019536078036312446,
      "loss": 0.2954,
      "step": 2094
    },
    {
      "epoch": 0.9778296382730455,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019535636185885613,
      "loss": 0.2887,
      "step": 2095
    },
    {
      "epoch": 0.978296382730455,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019535194130145954,
      "loss": 0.2876,
      "step": 2096
    },
    {
      "epoch": 0.9787631271878646,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019534751869102986,
      "loss": 0.3709,
      "step": 2097
    },
    {
      "epoch": 0.9792298716452742,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019534309402766234,
      "loss": 0.2436,
      "step": 2098
    },
    {
      "epoch": 0.9796966161026838,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019533866731145217,
      "loss": 0.2366,
      "step": 2099
    },
    {
      "epoch": 0.9801633605600933,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019533423854249473,
      "loss": 0.2663,
      "step": 2100
    },
    {
      "epoch": 0.9806301050175029,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019532980772088537,
      "loss": 0.3062,
      "step": 2101
    },
    {
      "epoch": 0.9810968494749125,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001953253748467195,
      "loss": 0.3535,
      "step": 2102
    },
    {
      "epoch": 0.9815635939323221,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019532093992009252,
      "loss": 0.387,
      "step": 2103
    },
    {
      "epoch": 0.9820303383897316,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019531650294109994,
      "loss": 0.3244,
      "step": 2104
    },
    {
      "epoch": 0.9824970828471412,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019531206390983728,
      "loss": 0.2721,
      "step": 2105
    },
    {
      "epoch": 0.9829638273045508,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019530762282640016,
      "loss": 0.3763,
      "step": 2106
    },
    {
      "epoch": 0.9834305717619604,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019530317969088416,
      "loss": 0.2861,
      "step": 2107
    },
    {
      "epoch": 0.9838973162193699,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00019529873450338498,
      "loss": 0.2865,
      "step": 2108
    },
    {
      "epoch": 0.9843640606767795,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001952942872639983,
      "loss": 0.3348,
      "step": 2109
    },
    {
      "epoch": 0.9848308051341891,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019528983797281987,
      "loss": 0.3812,
      "step": 2110
    },
    {
      "epoch": 0.9852975495915987,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019528538662994552,
      "loss": 0.2858,
      "step": 2111
    },
    {
      "epoch": 0.9857642940490081,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019528093323547107,
      "loss": 0.3019,
      "step": 2112
    },
    {
      "epoch": 0.9862310385064177,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019527647778949244,
      "loss": 0.294,
      "step": 2113
    },
    {
      "epoch": 0.9866977829638273,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001952720202921055,
      "loss": 0.3028,
      "step": 2114
    },
    {
      "epoch": 0.9871645274212368,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019526756074340624,
      "loss": 0.2922,
      "step": 2115
    },
    {
      "epoch": 0.9876312718786464,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019526309914349072,
      "loss": 0.3568,
      "step": 2116
    },
    {
      "epoch": 0.988098016336056,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019525863549245498,
      "loss": 0.2498,
      "step": 2117
    },
    {
      "epoch": 0.9885647607934656,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019525416979039513,
      "loss": 0.3667,
      "step": 2118
    },
    {
      "epoch": 0.9890315052508751,
      "grad_norm": 1.0,
      "learning_rate": 0.00019524970203740728,
      "loss": 0.2874,
      "step": 2119
    },
    {
      "epoch": 0.9894982497082847,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019524523223358773,
      "loss": 0.3162,
      "step": 2120
    },
    {
      "epoch": 0.9899649941656943,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001952407603790326,
      "loss": 0.3,
      "step": 2121
    },
    {
      "epoch": 0.9904317386231039,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019523628647383825,
      "loss": 0.3815,
      "step": 2122
    },
    {
      "epoch": 0.9908984830805134,
      "grad_norm": 1.0625,
      "learning_rate": 0.000195231810518101,
      "loss": 0.3099,
      "step": 2123
    },
    {
      "epoch": 0.991365227537923,
      "grad_norm": 1.125,
      "learning_rate": 0.00019522733251191716,
      "loss": 0.3147,
      "step": 2124
    },
    {
      "epoch": 0.9918319719953326,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019522285245538324,
      "loss": 0.2891,
      "step": 2125
    },
    {
      "epoch": 0.9922987164527421,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019521837034859564,
      "loss": 0.3103,
      "step": 2126
    },
    {
      "epoch": 0.9927654609101517,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019521388619165087,
      "loss": 0.268,
      "step": 2127
    },
    {
      "epoch": 0.9932322053675613,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001952093999846455,
      "loss": 0.4177,
      "step": 2128
    },
    {
      "epoch": 0.9936989498249709,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001952049117276761,
      "loss": 0.3214,
      "step": 2129
    },
    {
      "epoch": 0.9941656942823804,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019520042142083933,
      "loss": 0.287,
      "step": 2130
    },
    {
      "epoch": 0.99463243873979,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019519592906423188,
      "loss": 0.2723,
      "step": 2131
    },
    {
      "epoch": 0.9950991831971996,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019519143465795042,
      "loss": 0.3087,
      "step": 2132
    },
    {
      "epoch": 0.995565927654609,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019518693820209175,
      "loss": 0.3426,
      "step": 2133
    },
    {
      "epoch": 0.9960326721120186,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001951824396967527,
      "loss": 0.3383,
      "step": 2134
    },
    {
      "epoch": 0.9964994165694282,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019517793914203014,
      "loss": 0.3288,
      "step": 2135
    },
    {
      "epoch": 0.9969661610268378,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019517343653802094,
      "loss": 0.2814,
      "step": 2136
    },
    {
      "epoch": 0.9974329054842473,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019516893188482203,
      "loss": 0.2673,
      "step": 2137
    },
    {
      "epoch": 0.9978996499416569,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019516442518253046,
      "loss": 0.2672,
      "step": 2138
    },
    {
      "epoch": 0.9983663943990665,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001951599164312432,
      "loss": 0.2843,
      "step": 2139
    },
    {
      "epoch": 0.9988331388564761,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019515540563105739,
      "loss": 0.2489,
      "step": 2140
    },
    {
      "epoch": 0.9992998833138856,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001951508927820701,
      "loss": 0.3219,
      "step": 2141
    },
    {
      "epoch": 0.9997666277712952,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001951463778843785,
      "loss": 0.2532,
      "step": 2142
    },
    {
      "epoch": 1.0002333722287047,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019514186093807983,
      "loss": 0.2466,
      "step": 2143
    },
    {
      "epoch": 1.0007001166861142,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019513734194327134,
      "loss": 0.2551,
      "step": 2144
    },
    {
      "epoch": 1.0007001166861142,
      "eval_loss": 1.2306029796600342,
      "eval_runtime": 93.8413,
      "eval_samples_per_second": 19.224,
      "eval_steps_per_second": 2.408,
      "step": 2144
    },
    {
      "epoch": 1.0011668611435238,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019513282090005032,
      "loss": 0.2502,
      "step": 2145
    },
    {
      "epoch": 1.0016336056009334,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001951282978085141,
      "loss": 0.2808,
      "step": 2146
    },
    {
      "epoch": 1.002100350058343,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001951237726687601,
      "loss": 0.2566,
      "step": 2147
    },
    {
      "epoch": 1.0025670945157525,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019511924548088573,
      "loss": 0.2137,
      "step": 2148
    },
    {
      "epoch": 1.0030338389731621,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019511471624498845,
      "loss": 0.2158,
      "step": 2149
    },
    {
      "epoch": 1.0035005834305717,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001951101849611658,
      "loss": 0.224,
      "step": 2150
    },
    {
      "epoch": 1.0039673278879813,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019510565162951537,
      "loss": 0.2626,
      "step": 2151
    },
    {
      "epoch": 1.0044340723453908,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019510111625013472,
      "loss": 0.2251,
      "step": 2152
    },
    {
      "epoch": 1.0049008168028004,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001950965788231215,
      "loss": 0.23,
      "step": 2153
    },
    {
      "epoch": 1.00536756126021,
      "grad_norm": 1.0,
      "learning_rate": 0.00019509203934857348,
      "loss": 0.2859,
      "step": 2154
    },
    {
      "epoch": 1.0058343057176196,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001950874978265883,
      "loss": 0.2624,
      "step": 2155
    },
    {
      "epoch": 1.0063010501750291,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019508295425726382,
      "loss": 0.2305,
      "step": 2156
    },
    {
      "epoch": 1.0067677946324387,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001950784086406978,
      "loss": 0.2771,
      "step": 2157
    },
    {
      "epoch": 1.0072345390898483,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001950738609769882,
      "loss": 0.2428,
      "step": 2158
    },
    {
      "epoch": 1.0077012835472579,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019506931126623284,
      "loss": 0.2298,
      "step": 2159
    },
    {
      "epoch": 1.0081680280046674,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019506475950852977,
      "loss": 0.2872,
      "step": 2160
    },
    {
      "epoch": 1.008634772462077,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001950602057039769,
      "loss": 0.2608,
      "step": 2161
    },
    {
      "epoch": 1.0091015169194866,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019505564985267237,
      "loss": 0.2544,
      "step": 2162
    },
    {
      "epoch": 1.0095682613768961,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019505109195471423,
      "loss": 0.234,
      "step": 2163
    },
    {
      "epoch": 1.0100350058343057,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001950465320102006,
      "loss": 0.182,
      "step": 2164
    },
    {
      "epoch": 1.0105017502917153,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019504197001922967,
      "loss": 0.2491,
      "step": 2165
    },
    {
      "epoch": 1.0109684947491249,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019503740598189972,
      "loss": 0.1935,
      "step": 2166
    },
    {
      "epoch": 1.0114352392065344,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019503283989830895,
      "loss": 0.2332,
      "step": 2167
    },
    {
      "epoch": 1.011901983663944,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001950282717685557,
      "loss": 0.2799,
      "step": 2168
    },
    {
      "epoch": 1.0123687281213536,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001950237015927383,
      "loss": 0.2629,
      "step": 2169
    },
    {
      "epoch": 1.0128354725787632,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019501912937095522,
      "loss": 0.2301,
      "step": 2170
    },
    {
      "epoch": 1.0133022170361727,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019501455510330482,
      "loss": 0.214,
      "step": 2171
    },
    {
      "epoch": 1.0137689614935823,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019500997878988562,
      "loss": 0.1599,
      "step": 2172
    },
    {
      "epoch": 1.0142357059509919,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001950054004307962,
      "loss": 0.2481,
      "step": 2173
    },
    {
      "epoch": 1.0147024504084015,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019500082002613507,
      "loss": 0.2762,
      "step": 2174
    },
    {
      "epoch": 1.015169194865811,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001949962375760009,
      "loss": 0.2261,
      "step": 2175
    },
    {
      "epoch": 1.0156359393232206,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019499165308049232,
      "loss": 0.2385,
      "step": 2176
    },
    {
      "epoch": 1.0161026837806302,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019498706653970805,
      "loss": 0.2564,
      "step": 2177
    },
    {
      "epoch": 1.0165694282380398,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019498247795374688,
      "loss": 0.2645,
      "step": 2178
    },
    {
      "epoch": 1.0170361726954493,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019497788732270754,
      "loss": 0.2569,
      "step": 2179
    },
    {
      "epoch": 1.017502917152859,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001949732946466889,
      "loss": 0.2739,
      "step": 2180
    },
    {
      "epoch": 1.0179696616102685,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001949686999257899,
      "loss": 0.2604,
      "step": 2181
    },
    {
      "epoch": 1.018436406067678,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019496410316010936,
      "loss": 0.2632,
      "step": 2182
    },
    {
      "epoch": 1.0189031505250876,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019495950434974635,
      "loss": 0.2007,
      "step": 2183
    },
    {
      "epoch": 1.0193698949824972,
      "grad_norm": 1.28125,
      "learning_rate": 0.00019495490349479987,
      "loss": 0.2974,
      "step": 2184
    },
    {
      "epoch": 1.0198366394399065,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019495030059536893,
      "loss": 0.2166,
      "step": 2185
    },
    {
      "epoch": 1.0203033838973161,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019494569565155266,
      "loss": 0.3114,
      "step": 2186
    },
    {
      "epoch": 1.0207701283547257,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019494108866345024,
      "loss": 0.1917,
      "step": 2187
    },
    {
      "epoch": 1.0212368728121353,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019493647963116084,
      "loss": 0.2037,
      "step": 2188
    },
    {
      "epoch": 1.0217036172695448,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019493186855478373,
      "loss": 0.2702,
      "step": 2189
    },
    {
      "epoch": 1.0221703617269544,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001949272554344181,
      "loss": 0.3081,
      "step": 2190
    },
    {
      "epoch": 1.022637106184364,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019492264027016337,
      "loss": 0.2116,
      "step": 2191
    },
    {
      "epoch": 1.0231038506417736,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019491802306211887,
      "loss": 0.2118,
      "step": 2192
    },
    {
      "epoch": 1.0235705950991831,
      "grad_norm": 1.09375,
      "learning_rate": 0.000194913403810384,
      "loss": 0.2531,
      "step": 2193
    },
    {
      "epoch": 1.0240373395565927,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001949087825150583,
      "loss": 0.3199,
      "step": 2194
    },
    {
      "epoch": 1.0245040840140023,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019490415917624115,
      "loss": 0.1776,
      "step": 2195
    },
    {
      "epoch": 1.0249708284714119,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019489953379403217,
      "loss": 0.2414,
      "step": 2196
    },
    {
      "epoch": 1.0254375729288214,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019489490636853094,
      "loss": 0.3021,
      "step": 2197
    },
    {
      "epoch": 1.025904317386231,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001948902768998371,
      "loss": 0.2418,
      "step": 2198
    },
    {
      "epoch": 1.0263710618436406,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001948856453880503,
      "loss": 0.2139,
      "step": 2199
    },
    {
      "epoch": 1.0268378063010501,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019488101183327028,
      "loss": 0.2284,
      "step": 2200
    },
    {
      "epoch": 1.0273045507584597,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001948763762355968,
      "loss": 0.2503,
      "step": 2201
    },
    {
      "epoch": 1.0277712952158693,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001948717385951297,
      "loss": 0.2955,
      "step": 2202
    },
    {
      "epoch": 1.0282380396732789,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019486709891196875,
      "loss": 0.2613,
      "step": 2203
    },
    {
      "epoch": 1.0287047841306884,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019486245718621395,
      "loss": 0.2269,
      "step": 2204
    },
    {
      "epoch": 1.029171528588098,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001948578134179652,
      "loss": 0.256,
      "step": 2205
    },
    {
      "epoch": 1.0296382730455076,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019485316760732243,
      "loss": 0.226,
      "step": 2206
    },
    {
      "epoch": 1.0301050175029172,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00019484851975438576,
      "loss": 0.2082,
      "step": 2207
    },
    {
      "epoch": 1.0305717619603267,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019484386985925523,
      "loss": 0.2892,
      "step": 2208
    },
    {
      "epoch": 1.0310385064177363,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019483921792203093,
      "loss": 0.2331,
      "step": 2209
    },
    {
      "epoch": 1.0315052508751459,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019483456394281303,
      "loss": 0.2458,
      "step": 2210
    },
    {
      "epoch": 1.0319719953325555,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019482990792170178,
      "loss": 0.2395,
      "step": 2211
    },
    {
      "epoch": 1.032438739789965,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019482524985879737,
      "loss": 0.2352,
      "step": 2212
    },
    {
      "epoch": 1.0329054842473746,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019482058975420015,
      "loss": 0.2235,
      "step": 2213
    },
    {
      "epoch": 1.0333722287047842,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001948159276080104,
      "loss": 0.2207,
      "step": 2214
    },
    {
      "epoch": 1.0338389731621938,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019481126342032855,
      "loss": 0.2723,
      "step": 2215
    },
    {
      "epoch": 1.0343057176196033,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019480659719125498,
      "loss": 0.302,
      "step": 2216
    },
    {
      "epoch": 1.034772462077013,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001948019289208902,
      "loss": 0.218,
      "step": 2217
    },
    {
      "epoch": 1.0352392065344225,
      "grad_norm": 0.875,
      "learning_rate": 0.00019479725860933468,
      "loss": 0.1877,
      "step": 2218
    },
    {
      "epoch": 1.035705950991832,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019479258625668906,
      "loss": 0.2687,
      "step": 2219
    },
    {
      "epoch": 1.0361726954492416,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019478791186305382,
      "loss": 0.2537,
      "step": 2220
    },
    {
      "epoch": 1.0366394399066512,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019478323542852972,
      "loss": 0.2508,
      "step": 2221
    },
    {
      "epoch": 1.0371061843640608,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019477855695321735,
      "loss": 0.2425,
      "step": 2222
    },
    {
      "epoch": 1.0375729288214703,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001947738764372175,
      "loss": 0.2167,
      "step": 2223
    },
    {
      "epoch": 1.03803967327888,
      "grad_norm": 1.203125,
      "learning_rate": 0.00019476919388063097,
      "loss": 0.2439,
      "step": 2224
    },
    {
      "epoch": 1.0385064177362895,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019476450928355852,
      "loss": 0.2648,
      "step": 2225
    },
    {
      "epoch": 1.0389731621936988,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019475982264610106,
      "loss": 0.2187,
      "step": 2226
    },
    {
      "epoch": 1.0394399066511084,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019475513396835948,
      "loss": 0.265,
      "step": 2227
    },
    {
      "epoch": 1.039906651108518,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001947504432504347,
      "loss": 0.25,
      "step": 2228
    },
    {
      "epoch": 1.0403733955659276,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019474575049242777,
      "loss": 0.251,
      "step": 2229
    },
    {
      "epoch": 1.0408401400233371,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019474105569443974,
      "loss": 0.25,
      "step": 2230
    },
    {
      "epoch": 1.0413068844807467,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019473635885657162,
      "loss": 0.1806,
      "step": 2231
    },
    {
      "epoch": 1.0417736289381563,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001947316599789246,
      "loss": 0.2547,
      "step": 2232
    },
    {
      "epoch": 1.0422403733955659,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019472695906159982,
      "loss": 0.1936,
      "step": 2233
    },
    {
      "epoch": 1.0427071178529754,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019472225610469855,
      "loss": 0.2443,
      "step": 2234
    },
    {
      "epoch": 1.043173862310385,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019471755110832197,
      "loss": 0.2192,
      "step": 2235
    },
    {
      "epoch": 1.0436406067677946,
      "grad_norm": 1.25,
      "learning_rate": 0.00019471284407257146,
      "loss": 0.2388,
      "step": 2236
    },
    {
      "epoch": 1.0441073512252041,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019470813499754828,
      "loss": 0.2427,
      "step": 2237
    },
    {
      "epoch": 1.0445740956826137,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001947034238833539,
      "loss": 0.2716,
      "step": 2238
    },
    {
      "epoch": 1.0450408401400233,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00019469871073008973,
      "loss": 0.2475,
      "step": 2239
    },
    {
      "epoch": 1.0455075845974329,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019469399553785724,
      "loss": 0.2376,
      "step": 2240
    },
    {
      "epoch": 1.0459743290548424,
      "grad_norm": 0.9296875,
      "learning_rate": 0.000194689278306758,
      "loss": 0.2361,
      "step": 2241
    },
    {
      "epoch": 1.046441073512252,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001946845590368935,
      "loss": 0.2503,
      "step": 2242
    },
    {
      "epoch": 1.0469078179696616,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019467983772836537,
      "loss": 0.2045,
      "step": 2243
    },
    {
      "epoch": 1.0473745624270712,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019467511438127531,
      "loss": 0.2466,
      "step": 2244
    },
    {
      "epoch": 1.0478413068844807,
      "grad_norm": 0.96875,
      "learning_rate": 0.000194670388995725,
      "loss": 0.2657,
      "step": 2245
    },
    {
      "epoch": 1.0483080513418903,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001946656615718162,
      "loss": 0.2298,
      "step": 2246
    },
    {
      "epoch": 1.0487747957992999,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019466093210965062,
      "loss": 0.2114,
      "step": 2247
    },
    {
      "epoch": 1.0492415402567095,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019465620060933017,
      "loss": 0.2674,
      "step": 2248
    },
    {
      "epoch": 1.049708284714119,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001946514670709567,
      "loss": 0.2503,
      "step": 2249
    },
    {
      "epoch": 1.0501750291715286,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019464673149463213,
      "loss": 0.218,
      "step": 2250
    },
    {
      "epoch": 1.0506417736289382,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001946419938804584,
      "loss": 0.2836,
      "step": 2251
    },
    {
      "epoch": 1.0511085180863478,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019463725422853753,
      "loss": 0.1677,
      "step": 2252
    },
    {
      "epoch": 1.0515752625437573,
      "grad_norm": 1.0,
      "learning_rate": 0.00019463251253897158,
      "loss": 0.2379,
      "step": 2253
    },
    {
      "epoch": 1.052042007001167,
      "grad_norm": 1.0,
      "learning_rate": 0.00019462776881186267,
      "loss": 0.1799,
      "step": 2254
    },
    {
      "epoch": 1.0525087514585765,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001946230230473129,
      "loss": 0.2155,
      "step": 2255
    },
    {
      "epoch": 1.052975495915986,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001946182752454244,
      "loss": 0.2522,
      "step": 2256
    },
    {
      "epoch": 1.0534422403733956,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019461352540629952,
      "loss": 0.2253,
      "step": 2257
    },
    {
      "epoch": 1.0539089848308052,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019460877353004043,
      "loss": 0.2448,
      "step": 2258
    },
    {
      "epoch": 1.0543757292882148,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001946040196167495,
      "loss": 0.2603,
      "step": 2259
    },
    {
      "epoch": 1.0548424737456243,
      "grad_norm": 1.296875,
      "learning_rate": 0.00019459926366652904,
      "loss": 0.2377,
      "step": 2260
    },
    {
      "epoch": 1.055309218203034,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019459450567948147,
      "loss": 0.2353,
      "step": 2261
    },
    {
      "epoch": 1.0557759626604435,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019458974565570925,
      "loss": 0.2976,
      "step": 2262
    },
    {
      "epoch": 1.056242707117853,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019458498359531485,
      "loss": 0.2382,
      "step": 2263
    },
    {
      "epoch": 1.0567094515752626,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019458021949840082,
      "loss": 0.2588,
      "step": 2264
    },
    {
      "epoch": 1.0571761960326722,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019457545336506972,
      "loss": 0.182,
      "step": 2265
    },
    {
      "epoch": 1.0576429404900818,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001945706851954242,
      "loss": 0.2257,
      "step": 2266
    },
    {
      "epoch": 1.0581096849474911,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00019456591498956684,
      "loss": 0.1687,
      "step": 2267
    },
    {
      "epoch": 1.0585764294049007,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019456114274760044,
      "loss": 0.2213,
      "step": 2268
    },
    {
      "epoch": 1.0590431738623103,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019455636846962772,
      "loss": 0.2771,
      "step": 2269
    },
    {
      "epoch": 1.0595099183197199,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019455159215575146,
      "loss": 0.2325,
      "step": 2270
    },
    {
      "epoch": 1.0599766627771294,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019454681380607456,
      "loss": 0.2455,
      "step": 2271
    },
    {
      "epoch": 1.060443407234539,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019454203342069982,
      "loss": 0.1583,
      "step": 2272
    },
    {
      "epoch": 1.0609101516919486,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019453725099973022,
      "loss": 0.2193,
      "step": 2273
    },
    {
      "epoch": 1.0613768961493582,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001945324665432687,
      "loss": 0.2357,
      "step": 2274
    },
    {
      "epoch": 1.0618436406067677,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001945276800514183,
      "loss": 0.2292,
      "step": 2275
    },
    {
      "epoch": 1.0623103850641773,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019452289152428206,
      "loss": 0.2365,
      "step": 2276
    },
    {
      "epoch": 1.0627771295215869,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001945181009619631,
      "loss": 0.2065,
      "step": 2277
    },
    {
      "epoch": 1.0632438739789964,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019451330836456457,
      "loss": 0.2211,
      "step": 2278
    },
    {
      "epoch": 1.063710618436406,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001945085137321896,
      "loss": 0.1949,
      "step": 2279
    },
    {
      "epoch": 1.0641773628938156,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019450371706494153,
      "loss": 0.2269,
      "step": 2280
    },
    {
      "epoch": 1.0646441073512252,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019449891836292353,
      "loss": 0.2181,
      "step": 2281
    },
    {
      "epoch": 1.0651108518086347,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000194494117626239,
      "loss": 0.1948,
      "step": 2282
    },
    {
      "epoch": 1.0655775962660443,
      "grad_norm": 1.203125,
      "learning_rate": 0.00019448931485499126,
      "loss": 0.2224,
      "step": 2283
    },
    {
      "epoch": 1.0660443407234539,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001944845100492837,
      "loss": 0.2516,
      "step": 2284
    },
    {
      "epoch": 1.0665110851808635,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019447970320921986,
      "loss": 0.1979,
      "step": 2285
    },
    {
      "epoch": 1.066977829638273,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019447489433490315,
      "loss": 0.1938,
      "step": 2286
    },
    {
      "epoch": 1.0674445740956826,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019447008342643717,
      "loss": 0.2753,
      "step": 2287
    },
    {
      "epoch": 1.0679113185530922,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019446527048392546,
      "loss": 0.2355,
      "step": 2288
    },
    {
      "epoch": 1.0683780630105018,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001944604555074717,
      "loss": 0.1796,
      "step": 2289
    },
    {
      "epoch": 1.0688448074679113,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019445563849717948,
      "loss": 0.1926,
      "step": 2290
    },
    {
      "epoch": 1.069311551925321,
      "grad_norm": 1.0,
      "learning_rate": 0.0001944508194531526,
      "loss": 0.24,
      "step": 2291
    },
    {
      "epoch": 1.0697782963827305,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019444599837549476,
      "loss": 0.2633,
      "step": 2292
    },
    {
      "epoch": 1.07024504084014,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019444117526430977,
      "loss": 0.2094,
      "step": 2293
    },
    {
      "epoch": 1.0707117852975496,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019443635011970152,
      "loss": 0.2457,
      "step": 2294
    },
    {
      "epoch": 1.0711785297549592,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019443152294177385,
      "loss": 0.1909,
      "step": 2295
    },
    {
      "epoch": 1.0716452742123688,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001944266937306307,
      "loss": 0.2116,
      "step": 2296
    },
    {
      "epoch": 1.0721120186697783,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019442186248637612,
      "loss": 0.1813,
      "step": 2297
    },
    {
      "epoch": 1.072578763127188,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019441702920911406,
      "loss": 0.2442,
      "step": 2298
    },
    {
      "epoch": 1.0730455075845975,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019441219389894858,
      "loss": 0.2369,
      "step": 2299
    },
    {
      "epoch": 1.073512252042007,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019440735655598386,
      "loss": 0.2274,
      "step": 2300
    },
    {
      "epoch": 1.0739789964994166,
      "grad_norm": 1.3515625,
      "learning_rate": 0.00019440251718032397,
      "loss": 0.326,
      "step": 2301
    },
    {
      "epoch": 1.0744457409568262,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019439767577207314,
      "loss": 0.1723,
      "step": 2302
    },
    {
      "epoch": 1.0749124854142358,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019439283233133563,
      "loss": 0.2544,
      "step": 2303
    },
    {
      "epoch": 1.0753792298716454,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001943879868582157,
      "loss": 0.2578,
      "step": 2304
    },
    {
      "epoch": 1.075845974329055,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019438313935281768,
      "loss": 0.2702,
      "step": 2305
    },
    {
      "epoch": 1.0763127187864645,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019437828981524598,
      "loss": 0.2418,
      "step": 2306
    },
    {
      "epoch": 1.076779463243874,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019437343824560497,
      "loss": 0.2026,
      "step": 2307
    },
    {
      "epoch": 1.0772462077012834,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001943685846439991,
      "loss": 0.3111,
      "step": 2308
    },
    {
      "epoch": 1.0777129521586932,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019436372901053295,
      "loss": 0.2151,
      "step": 2309
    },
    {
      "epoch": 1.0781796966161026,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019435887134531098,
      "loss": 0.1683,
      "step": 2310
    },
    {
      "epoch": 1.0786464410735122,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019435401164843782,
      "loss": 0.2121,
      "step": 2311
    },
    {
      "epoch": 1.0791131855309217,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019434914992001812,
      "loss": 0.2417,
      "step": 2312
    },
    {
      "epoch": 1.0795799299883313,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019434428616015652,
      "loss": 0.1975,
      "step": 2313
    },
    {
      "epoch": 1.0800466744457409,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001943394203689578,
      "loss": 0.2332,
      "step": 2314
    },
    {
      "epoch": 1.0805134189031504,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019433455254652663,
      "loss": 0.2676,
      "step": 2315
    },
    {
      "epoch": 1.08098016336056,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001943296826929679,
      "loss": 0.257,
      "step": 2316
    },
    {
      "epoch": 1.0814469078179696,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019432481080838642,
      "loss": 0.2414,
      "step": 2317
    },
    {
      "epoch": 1.0819136522753792,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019431993689288714,
      "loss": 0.2656,
      "step": 2318
    },
    {
      "epoch": 1.0823803967327887,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019431506094657496,
      "loss": 0.2405,
      "step": 2319
    },
    {
      "epoch": 1.0828471411901983,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001943101829695549,
      "loss": 0.2192,
      "step": 2320
    },
    {
      "epoch": 1.0833138856476079,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019430530296193194,
      "loss": 0.2129,
      "step": 2321
    },
    {
      "epoch": 1.0837806301050175,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019430042092381117,
      "loss": 0.223,
      "step": 2322
    },
    {
      "epoch": 1.084247374562427,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019429553685529772,
      "loss": 0.2025,
      "step": 2323
    },
    {
      "epoch": 1.0847141190198366,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019429065075649672,
      "loss": 0.2192,
      "step": 2324
    },
    {
      "epoch": 1.0851808634772462,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001942857626275134,
      "loss": 0.299,
      "step": 2325
    },
    {
      "epoch": 1.0856476079346558,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019428087246845302,
      "loss": 0.2877,
      "step": 2326
    },
    {
      "epoch": 1.0861143523920653,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019427598027942083,
      "loss": 0.2732,
      "step": 2327
    },
    {
      "epoch": 1.086581096849475,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019427108606052216,
      "loss": 0.1935,
      "step": 2328
    },
    {
      "epoch": 1.0870478413068845,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019426618981186242,
      "loss": 0.3053,
      "step": 2329
    },
    {
      "epoch": 1.087514585764294,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019426129153354708,
      "loss": 0.1771,
      "step": 2330
    },
    {
      "epoch": 1.0879813302217036,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019425639122568148,
      "loss": 0.2274,
      "step": 2331
    },
    {
      "epoch": 1.0884480746791132,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001942514888883712,
      "loss": 0.2305,
      "step": 2332
    },
    {
      "epoch": 1.0889148191365228,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019424658452172185,
      "loss": 0.1955,
      "step": 2333
    },
    {
      "epoch": 1.0893815635939323,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019424167812583893,
      "loss": 0.1955,
      "step": 2334
    },
    {
      "epoch": 1.089848308051342,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001942367697008281,
      "loss": 0.2281,
      "step": 2335
    },
    {
      "epoch": 1.0903150525087515,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001942318592467951,
      "loss": 0.2311,
      "step": 2336
    },
    {
      "epoch": 1.090781796966161,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00019422694676384558,
      "loss": 0.3157,
      "step": 2337
    },
    {
      "epoch": 1.0912485414235706,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019422203225208539,
      "loss": 0.2205,
      "step": 2338
    },
    {
      "epoch": 1.0917152858809802,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019421711571162027,
      "loss": 0.2418,
      "step": 2339
    },
    {
      "epoch": 1.0921820303383898,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001942121971425561,
      "loss": 0.1829,
      "step": 2340
    },
    {
      "epoch": 1.0926487747957994,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019420727654499885,
      "loss": 0.2171,
      "step": 2341
    },
    {
      "epoch": 1.093115519253209,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019420235391905436,
      "loss": 0.2278,
      "step": 2342
    },
    {
      "epoch": 1.0935822637106185,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019419742926482872,
      "loss": 0.2366,
      "step": 2343
    },
    {
      "epoch": 1.094049008168028,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019419250258242786,
      "loss": 0.1932,
      "step": 2344
    },
    {
      "epoch": 1.0945157526254377,
      "grad_norm": 1.203125,
      "learning_rate": 0.00019418757387195795,
      "loss": 0.3093,
      "step": 2345
    },
    {
      "epoch": 1.0949824970828472,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019418264313352505,
      "loss": 0.2753,
      "step": 2346
    },
    {
      "epoch": 1.0954492415402568,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019417771036723535,
      "loss": 0.2706,
      "step": 2347
    },
    {
      "epoch": 1.0959159859976664,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019417277557319507,
      "loss": 0.2216,
      "step": 2348
    },
    {
      "epoch": 1.096382730455076,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019416783875151044,
      "loss": 0.2515,
      "step": 2349
    },
    {
      "epoch": 1.0968494749124855,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019416289990228775,
      "loss": 0.2861,
      "step": 2350
    },
    {
      "epoch": 1.0973162193698949,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019415795902563335,
      "loss": 0.2058,
      "step": 2351
    },
    {
      "epoch": 1.0977829638273044,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019415301612165365,
      "loss": 0.2403,
      "step": 2352
    },
    {
      "epoch": 1.098249708284714,
      "grad_norm": 0.9375,
      "learning_rate": 0.000194148071190455,
      "loss": 0.2202,
      "step": 2353
    },
    {
      "epoch": 1.0987164527421236,
      "grad_norm": 1.34375,
      "learning_rate": 0.00019414312423214397,
      "loss": 0.3434,
      "step": 2354
    },
    {
      "epoch": 1.0991831971995332,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019413817524682698,
      "loss": 0.304,
      "step": 2355
    },
    {
      "epoch": 1.0996499416569427,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019413322423461065,
      "loss": 0.2471,
      "step": 2356
    },
    {
      "epoch": 1.1001166861143523,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019412827119560152,
      "loss": 0.2006,
      "step": 2357
    },
    {
      "epoch": 1.100583430571762,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019412331612990634,
      "loss": 0.1803,
      "step": 2358
    },
    {
      "epoch": 1.1010501750291715,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001941183590376317,
      "loss": 0.212,
      "step": 2359
    },
    {
      "epoch": 1.101516919486581,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019411339991888435,
      "loss": 0.2127,
      "step": 2360
    },
    {
      "epoch": 1.1019836639439906,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001941084387737711,
      "loss": 0.3057,
      "step": 2361
    },
    {
      "epoch": 1.1024504084014002,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001941034756023987,
      "loss": 0.2501,
      "step": 2362
    },
    {
      "epoch": 1.1029171528588098,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019409851040487412,
      "loss": 0.2064,
      "step": 2363
    },
    {
      "epoch": 1.1033838973162193,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001940935431813042,
      "loss": 0.1881,
      "step": 2364
    },
    {
      "epoch": 1.103850641773629,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019408857393179586,
      "loss": 0.2226,
      "step": 2365
    },
    {
      "epoch": 1.1043173862310385,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019408360265645618,
      "loss": 0.1776,
      "step": 2366
    },
    {
      "epoch": 1.104784130688448,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001940786293553921,
      "loss": 0.2592,
      "step": 2367
    },
    {
      "epoch": 1.1052508751458576,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019407365402871078,
      "loss": 0.2413,
      "step": 2368
    },
    {
      "epoch": 1.1057176196032672,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001940686766765193,
      "loss": 0.242,
      "step": 2369
    },
    {
      "epoch": 1.1061843640606768,
      "grad_norm": 1.0,
      "learning_rate": 0.00019406369729892483,
      "loss": 0.2174,
      "step": 2370
    },
    {
      "epoch": 1.1066511085180863,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001940587158960346,
      "loss": 0.235,
      "step": 2371
    },
    {
      "epoch": 1.107117852975496,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019405373246795588,
      "loss": 0.1654,
      "step": 2372
    },
    {
      "epoch": 1.1075845974329055,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019404874701479594,
      "loss": 0.2066,
      "step": 2373
    },
    {
      "epoch": 1.108051341890315,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001940437595366621,
      "loss": 0.1897,
      "step": 2374
    },
    {
      "epoch": 1.1085180863477246,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001940387700336618,
      "loss": 0.2093,
      "step": 2375
    },
    {
      "epoch": 1.1089848308051342,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019403377850590243,
      "loss": 0.1885,
      "step": 2376
    },
    {
      "epoch": 1.1094515752625438,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019402878495349145,
      "loss": 0.2773,
      "step": 2377
    },
    {
      "epoch": 1.1099183197199534,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019402378937653645,
      "loss": 0.2894,
      "step": 2378
    },
    {
      "epoch": 1.110385064177363,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019401879177514492,
      "loss": 0.16,
      "step": 2379
    },
    {
      "epoch": 1.1108518086347725,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001940137921494245,
      "loss": 0.2041,
      "step": 2380
    },
    {
      "epoch": 1.111318553092182,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001940087904994828,
      "loss": 0.1897,
      "step": 2381
    },
    {
      "epoch": 1.1117852975495917,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00019400378682542756,
      "loss": 0.2953,
      "step": 2382
    },
    {
      "epoch": 1.1122520420070012,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019399878112736648,
      "loss": 0.2477,
      "step": 2383
    },
    {
      "epoch": 1.1127187864644108,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019399377340540737,
      "loss": 0.1695,
      "step": 2384
    },
    {
      "epoch": 1.1131855309218204,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019398876365965802,
      "loss": 0.2543,
      "step": 2385
    },
    {
      "epoch": 1.11365227537923,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019398375189022627,
      "loss": 0.1865,
      "step": 2386
    },
    {
      "epoch": 1.1141190198366395,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019397873809722012,
      "loss": 0.2478,
      "step": 2387
    },
    {
      "epoch": 1.114585764294049,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00019397372228074743,
      "loss": 0.2818,
      "step": 2388
    },
    {
      "epoch": 1.1150525087514587,
      "grad_norm": 0.84375,
      "learning_rate": 0.00019396870444091623,
      "loss": 0.2183,
      "step": 2389
    },
    {
      "epoch": 1.1155192532088682,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001939636845778346,
      "loss": 0.165,
      "step": 2390
    },
    {
      "epoch": 1.1159859976662778,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019395866269161056,
      "loss": 0.2518,
      "step": 2391
    },
    {
      "epoch": 1.1164527421236872,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019395363878235225,
      "loss": 0.2346,
      "step": 2392
    },
    {
      "epoch": 1.116919486581097,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019394861285016788,
      "loss": 0.2032,
      "step": 2393
    },
    {
      "epoch": 1.1173862310385063,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019394358489516564,
      "loss": 0.2578,
      "step": 2394
    },
    {
      "epoch": 1.117852975495916,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019393855491745376,
      "loss": 0.2035,
      "step": 2395
    },
    {
      "epoch": 1.1183197199533255,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001939335229171406,
      "loss": 0.2096,
      "step": 2396
    },
    {
      "epoch": 1.118786464410735,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019392848889433447,
      "loss": 0.229,
      "step": 2397
    },
    {
      "epoch": 1.1192532088681446,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019392345284914372,
      "loss": 0.1959,
      "step": 2398
    },
    {
      "epoch": 1.1197199533255542,
      "grad_norm": 1.125,
      "learning_rate": 0.00019391841478167681,
      "loss": 0.2616,
      "step": 2399
    },
    {
      "epoch": 1.1201866977829638,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001939133746920423,
      "loss": 0.262,
      "step": 2400
    },
    {
      "epoch": 1.1206534422403733,
      "grad_norm": 1.234375,
      "learning_rate": 0.00019390833258034858,
      "loss": 0.2233,
      "step": 2401
    },
    {
      "epoch": 1.121120186697783,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019390328844670426,
      "loss": 0.1906,
      "step": 2402
    },
    {
      "epoch": 1.1215869311551925,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019389824229121798,
      "loss": 0.2365,
      "step": 2403
    },
    {
      "epoch": 1.122053675612602,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019389319411399836,
      "loss": 0.1846,
      "step": 2404
    },
    {
      "epoch": 1.1225204200700116,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001938881439151541,
      "loss": 0.2078,
      "step": 2405
    },
    {
      "epoch": 1.1229871645274212,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019388309169479388,
      "loss": 0.214,
      "step": 2406
    },
    {
      "epoch": 1.1234539089848308,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019387803745302658,
      "loss": 0.2262,
      "step": 2407
    },
    {
      "epoch": 1.1239206534422403,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019387298118996095,
      "loss": 0.2706,
      "step": 2408
    },
    {
      "epoch": 1.12438739789965,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001938679229057059,
      "loss": 0.2475,
      "step": 2409
    },
    {
      "epoch": 1.1248541423570595,
      "grad_norm": 1.0,
      "learning_rate": 0.00019386286260037033,
      "loss": 0.2066,
      "step": 2410
    },
    {
      "epoch": 1.125320886814469,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019385780027406316,
      "loss": 0.217,
      "step": 2411
    },
    {
      "epoch": 1.1257876312718786,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001938527359268934,
      "loss": 0.1894,
      "step": 2412
    },
    {
      "epoch": 1.1262543757292882,
      "grad_norm": 0.859375,
      "learning_rate": 0.00019384766955897015,
      "loss": 0.1968,
      "step": 2413
    },
    {
      "epoch": 1.1267211201866978,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019384260117040242,
      "loss": 0.2198,
      "step": 2414
    },
    {
      "epoch": 1.1271878646441074,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019383753076129937,
      "loss": 0.2411,
      "step": 2415
    },
    {
      "epoch": 1.127654609101517,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019383245833177014,
      "loss": 0.2902,
      "step": 2416
    },
    {
      "epoch": 1.1281213535589265,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000193827383881924,
      "loss": 0.1835,
      "step": 2417
    },
    {
      "epoch": 1.128588098016336,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019382230741187023,
      "loss": 0.1948,
      "step": 2418
    },
    {
      "epoch": 1.1290548424737457,
      "grad_norm": 0.90625,
      "learning_rate": 0.000193817228921718,
      "loss": 0.2108,
      "step": 2419
    },
    {
      "epoch": 1.1295215869311552,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019381214841157678,
      "loss": 0.2795,
      "step": 2420
    },
    {
      "epoch": 1.1299883313885648,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019380706588155593,
      "loss": 0.2193,
      "step": 2421
    },
    {
      "epoch": 1.1304550758459744,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019380198133176486,
      "loss": 0.2879,
      "step": 2422
    },
    {
      "epoch": 1.130921820303384,
      "grad_norm": 1.296875,
      "learning_rate": 0.00019379689476231303,
      "loss": 0.2675,
      "step": 2423
    },
    {
      "epoch": 1.1313885647607935,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019379180617331004,
      "loss": 0.2809,
      "step": 2424
    },
    {
      "epoch": 1.131855309218203,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019378671556486533,
      "loss": 0.2362,
      "step": 2425
    },
    {
      "epoch": 1.1323220536756127,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019378162293708862,
      "loss": 0.2254,
      "step": 2426
    },
    {
      "epoch": 1.1327887981330222,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019377652829008954,
      "loss": 0.2095,
      "step": 2427
    },
    {
      "epoch": 1.1332555425904318,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019377143162397774,
      "loss": 0.2136,
      "step": 2428
    },
    {
      "epoch": 1.1337222870478414,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019376633293886295,
      "loss": 0.2394,
      "step": 2429
    },
    {
      "epoch": 1.134189031505251,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000193761232234855,
      "loss": 0.2411,
      "step": 2430
    },
    {
      "epoch": 1.1346557759626605,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019375612951206365,
      "loss": 0.2059,
      "step": 2431
    },
    {
      "epoch": 1.1351225204200701,
      "grad_norm": 1.25,
      "learning_rate": 0.00019375102477059886,
      "loss": 0.3163,
      "step": 2432
    },
    {
      "epoch": 1.1355892648774795,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019374591801057047,
      "loss": 0.206,
      "step": 2433
    },
    {
      "epoch": 1.1360560093348893,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019374080923208843,
      "loss": 0.161,
      "step": 2434
    },
    {
      "epoch": 1.1365227537922986,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019373569843526278,
      "loss": 0.2216,
      "step": 2435
    },
    {
      "epoch": 1.1369894982497084,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019373058562020354,
      "loss": 0.1693,
      "step": 2436
    },
    {
      "epoch": 1.1374562427071178,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001937254707870208,
      "loss": 0.1914,
      "step": 2437
    },
    {
      "epoch": 1.1379229871645273,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001937203539358247,
      "loss": 0.2169,
      "step": 2438
    },
    {
      "epoch": 1.138389731621937,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019371523506672537,
      "loss": 0.201,
      "step": 2439
    },
    {
      "epoch": 1.1388564760793465,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019371011417983306,
      "loss": 0.2133,
      "step": 2440
    },
    {
      "epoch": 1.139323220536756,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019370499127525798,
      "loss": 0.1873,
      "step": 2441
    },
    {
      "epoch": 1.1397899649941656,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019369986635311054,
      "loss": 0.1905,
      "step": 2442
    },
    {
      "epoch": 1.1402567094515752,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019369473941350095,
      "loss": 0.2438,
      "step": 2443
    },
    {
      "epoch": 1.1407234539089848,
      "grad_norm": 1.296875,
      "learning_rate": 0.0001936896104565397,
      "loss": 0.2276,
      "step": 2444
    },
    {
      "epoch": 1.1411901983663943,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001936844794823372,
      "loss": 0.2276,
      "step": 2445
    },
    {
      "epoch": 1.141656942823804,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019367934649100387,
      "loss": 0.2201,
      "step": 2446
    },
    {
      "epoch": 1.1421236872812135,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001936742114826503,
      "loss": 0.2041,
      "step": 2447
    },
    {
      "epoch": 1.142590431738623,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019366907445738703,
      "loss": 0.24,
      "step": 2448
    },
    {
      "epoch": 1.1430571761960326,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019366393541532463,
      "loss": 0.2643,
      "step": 2449
    },
    {
      "epoch": 1.1435239206534422,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001936587943565738,
      "loss": 0.1913,
      "step": 2450
    },
    {
      "epoch": 1.1439906651108518,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001936536512812452,
      "loss": 0.2067,
      "step": 2451
    },
    {
      "epoch": 1.1444574095682614,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019364850618944958,
      "loss": 0.2144,
      "step": 2452
    },
    {
      "epoch": 1.144924154025671,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019364335908129773,
      "loss": 0.2385,
      "step": 2453
    },
    {
      "epoch": 1.1453908984830805,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019363820995690046,
      "loss": 0.2412,
      "step": 2454
    },
    {
      "epoch": 1.14585764294049,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019363305881636863,
      "loss": 0.2939,
      "step": 2455
    },
    {
      "epoch": 1.1463243873978997,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00019362790565981316,
      "loss": 0.1874,
      "step": 2456
    },
    {
      "epoch": 1.1467911318553092,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019362275048734497,
      "loss": 0.2217,
      "step": 2457
    },
    {
      "epoch": 1.1472578763127188,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019361759329907508,
      "loss": 0.2098,
      "step": 2458
    },
    {
      "epoch": 1.1477246207701284,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001936124340951146,
      "loss": 0.2123,
      "step": 2459
    },
    {
      "epoch": 1.148191365227538,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001936072728755745,
      "loss": 0.2159,
      "step": 2460
    },
    {
      "epoch": 1.1486581096849475,
      "grad_norm": 0.9453125,
      "learning_rate": 0.000193602109640566,
      "loss": 0.1896,
      "step": 2461
    },
    {
      "epoch": 1.149124854142357,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019359694439020016,
      "loss": 0.1636,
      "step": 2462
    },
    {
      "epoch": 1.1495915985997667,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019359177712458832,
      "loss": 0.2027,
      "step": 2463
    },
    {
      "epoch": 1.1500583430571762,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019358660784384163,
      "loss": 0.1809,
      "step": 2464
    },
    {
      "epoch": 1.1505250875145858,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019358143654807147,
      "loss": 0.1893,
      "step": 2465
    },
    {
      "epoch": 1.1509918319719954,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019357626323738913,
      "loss": 0.2289,
      "step": 2466
    },
    {
      "epoch": 1.151458576429405,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019357108791190602,
      "loss": 0.2143,
      "step": 2467
    },
    {
      "epoch": 1.1519253208868145,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019356591057173355,
      "loss": 0.2221,
      "step": 2468
    },
    {
      "epoch": 1.1523920653442241,
      "grad_norm": 1.0,
      "learning_rate": 0.0001935607312169833,
      "loss": 0.2413,
      "step": 2469
    },
    {
      "epoch": 1.1528588098016337,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019355554984776661,
      "loss": 0.2145,
      "step": 2470
    },
    {
      "epoch": 1.1533255542590433,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019355036646419516,
      "loss": 0.2249,
      "step": 2471
    },
    {
      "epoch": 1.1537922987164528,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019354518106638052,
      "loss": 0.2432,
      "step": 2472
    },
    {
      "epoch": 1.1542590431738624,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019353999365443435,
      "loss": 0.2494,
      "step": 2473
    },
    {
      "epoch": 1.1547257876312718,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001935348042284683,
      "loss": 0.2807,
      "step": 2474
    },
    {
      "epoch": 1.1551925320886816,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019352961278859416,
      "loss": 0.1944,
      "step": 2475
    },
    {
      "epoch": 1.155659276546091,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019352441933492368,
      "loss": 0.2025,
      "step": 2476
    },
    {
      "epoch": 1.1561260210035007,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001935192238675687,
      "loss": 0.2445,
      "step": 2477
    },
    {
      "epoch": 1.15659276546091,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019351402638664107,
      "loss": 0.2177,
      "step": 2478
    },
    {
      "epoch": 1.1570595099183196,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019350882689225267,
      "loss": 0.241,
      "step": 2479
    },
    {
      "epoch": 1.1575262543757292,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001935036253845155,
      "loss": 0.2241,
      "step": 2480
    },
    {
      "epoch": 1.1579929988331388,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019349842186354153,
      "loss": 0.1815,
      "step": 2481
    },
    {
      "epoch": 1.1584597432905483,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019349321632944276,
      "loss": 0.2616,
      "step": 2482
    },
    {
      "epoch": 1.158926487747958,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019348800878233138,
      "loss": 0.1701,
      "step": 2483
    },
    {
      "epoch": 1.1593932322053675,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019348279922231938,
      "loss": 0.1838,
      "step": 2484
    },
    {
      "epoch": 1.159859976662777,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019347758764951903,
      "loss": 0.236,
      "step": 2485
    },
    {
      "epoch": 1.1603267211201866,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019347237406404247,
      "loss": 0.2314,
      "step": 2486
    },
    {
      "epoch": 1.1607934655775962,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019346715846600202,
      "loss": 0.2304,
      "step": 2487
    },
    {
      "epoch": 1.1612602100350058,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019346194085550993,
      "loss": 0.2471,
      "step": 2488
    },
    {
      "epoch": 1.1617269544924154,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019345672123267856,
      "loss": 0.1926,
      "step": 2489
    },
    {
      "epoch": 1.162193698949825,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001934514995976203,
      "loss": 0.2159,
      "step": 2490
    },
    {
      "epoch": 1.1626604434072345,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019344627595044753,
      "loss": 0.2628,
      "step": 2491
    },
    {
      "epoch": 1.163127187864644,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001934410502912728,
      "loss": 0.176,
      "step": 2492
    },
    {
      "epoch": 1.1635939323220537,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019343582262020857,
      "loss": 0.1886,
      "step": 2493
    },
    {
      "epoch": 1.1640606767794632,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001934305929373674,
      "loss": 0.1654,
      "step": 2494
    },
    {
      "epoch": 1.1645274212368728,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001934253612428619,
      "loss": 0.2338,
      "step": 2495
    },
    {
      "epoch": 1.1649941656942824,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019342012753680474,
      "loss": 0.2454,
      "step": 2496
    },
    {
      "epoch": 1.165460910151692,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019341489181930858,
      "loss": 0.2252,
      "step": 2497
    },
    {
      "epoch": 1.1659276546091015,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001934096540904861,
      "loss": 0.2087,
      "step": 2498
    },
    {
      "epoch": 1.166394399066511,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019340441435045018,
      "loss": 0.2019,
      "step": 2499
    },
    {
      "epoch": 1.1668611435239207,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019339917259931358,
      "loss": 0.1595,
      "step": 2500
    },
    {
      "epoch": 1.1673278879813302,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019339392883718914,
      "loss": 0.1916,
      "step": 2501
    },
    {
      "epoch": 1.1677946324387398,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019338868306418983,
      "loss": 0.23,
      "step": 2502
    },
    {
      "epoch": 1.1682613768961494,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001933834352804285,
      "loss": 0.19,
      "step": 2503
    },
    {
      "epoch": 1.168728121353559,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019337818548601825,
      "loss": 0.2698,
      "step": 2504
    },
    {
      "epoch": 1.1691948658109685,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019337293368107202,
      "loss": 0.2835,
      "step": 2505
    },
    {
      "epoch": 1.1696616102683781,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019336767986570297,
      "loss": 0.2395,
      "step": 2506
    },
    {
      "epoch": 1.1701283547257877,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019336242404002416,
      "loss": 0.2379,
      "step": 2507
    },
    {
      "epoch": 1.1705950991831973,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019335716620414876,
      "loss": 0.1639,
      "step": 2508
    },
    {
      "epoch": 1.1710618436406068,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019335190635819002,
      "loss": 0.2091,
      "step": 2509
    },
    {
      "epoch": 1.1715285880980164,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019334664450226113,
      "loss": 0.2841,
      "step": 2510
    },
    {
      "epoch": 1.171995332555426,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019334138063647544,
      "loss": 0.1889,
      "step": 2511
    },
    {
      "epoch": 1.1724620770128356,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019333611476094625,
      "loss": 0.1703,
      "step": 2512
    },
    {
      "epoch": 1.1729288214702451,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019333084687578697,
      "loss": 0.2213,
      "step": 2513
    },
    {
      "epoch": 1.1733955659276547,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019332557698111094,
      "loss": 0.1575,
      "step": 2514
    },
    {
      "epoch": 1.173862310385064,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019332030507703175,
      "loss": 0.2017,
      "step": 2515
    },
    {
      "epoch": 1.1743290548424739,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019331503116366285,
      "loss": 0.2111,
      "step": 2516
    },
    {
      "epoch": 1.1747957992998832,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001933097552411178,
      "loss": 0.1965,
      "step": 2517
    },
    {
      "epoch": 1.175262543757293,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019330447730951016,
      "loss": 0.2294,
      "step": 2518
    },
    {
      "epoch": 1.1757292882147024,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001932991973689536,
      "loss": 0.225,
      "step": 2519
    },
    {
      "epoch": 1.176196032672112,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001932939154195618,
      "loss": 0.2125,
      "step": 2520
    },
    {
      "epoch": 1.1766627771295215,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019328863146144852,
      "loss": 0.2299,
      "step": 2521
    },
    {
      "epoch": 1.177129521586931,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019328334549472749,
      "loss": 0.1957,
      "step": 2522
    },
    {
      "epoch": 1.1775962660443406,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019327805751951252,
      "loss": 0.2401,
      "step": 2523
    },
    {
      "epoch": 1.1780630105017502,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019327276753591747,
      "loss": 0.1759,
      "step": 2524
    },
    {
      "epoch": 1.1785297549591598,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019326747554405626,
      "loss": 0.2082,
      "step": 2525
    },
    {
      "epoch": 1.1789964994165694,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019326218154404283,
      "loss": 0.2176,
      "step": 2526
    },
    {
      "epoch": 1.179463243873979,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019325688553599115,
      "loss": 0.18,
      "step": 2527
    },
    {
      "epoch": 1.1799299883313885,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019325158752001525,
      "loss": 0.2154,
      "step": 2528
    },
    {
      "epoch": 1.180396732788798,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019324628749622922,
      "loss": 0.203,
      "step": 2529
    },
    {
      "epoch": 1.1808634772462077,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00019324098546474714,
      "loss": 0.177,
      "step": 2530
    },
    {
      "epoch": 1.1813302217036172,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019323568142568318,
      "loss": 0.1831,
      "step": 2531
    },
    {
      "epoch": 1.1817969661610268,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019323037537915158,
      "loss": 0.1953,
      "step": 2532
    },
    {
      "epoch": 1.1822637106184364,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019322506732526656,
      "loss": 0.185,
      "step": 2533
    },
    {
      "epoch": 1.182730455075846,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019321975726414238,
      "loss": 0.1687,
      "step": 2534
    },
    {
      "epoch": 1.1831971995332555,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001932144451958934,
      "loss": 0.2447,
      "step": 2535
    },
    {
      "epoch": 1.183663943990665,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000193209131120634,
      "loss": 0.1618,
      "step": 2536
    },
    {
      "epoch": 1.1841306884480747,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001932038150384786,
      "loss": 0.1916,
      "step": 2537
    },
    {
      "epoch": 1.1845974329054842,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019319849694954163,
      "loss": 0.1777,
      "step": 2538
    },
    {
      "epoch": 1.1850641773628938,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019319317685393764,
      "loss": 0.1951,
      "step": 2539
    },
    {
      "epoch": 1.1855309218203034,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019318785475178115,
      "loss": 0.1766,
      "step": 2540
    },
    {
      "epoch": 1.185997666277713,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019318253064318675,
      "loss": 0.1843,
      "step": 2541
    },
    {
      "epoch": 1.1864644107351225,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019317720452826907,
      "loss": 0.2082,
      "step": 2542
    },
    {
      "epoch": 1.1869311551925321,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019317187640714278,
      "loss": 0.1468,
      "step": 2543
    },
    {
      "epoch": 1.1873978996499417,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019316654627992262,
      "loss": 0.2385,
      "step": 2544
    },
    {
      "epoch": 1.1878646441073513,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019316121414672337,
      "loss": 0.2404,
      "step": 2545
    },
    {
      "epoch": 1.1883313885647608,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019315588000765982,
      "loss": 0.2167,
      "step": 2546
    },
    {
      "epoch": 1.1887981330221704,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001931505438628468,
      "loss": 0.1818,
      "step": 2547
    },
    {
      "epoch": 1.18926487747958,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001931452057123992,
      "loss": 0.2769,
      "step": 2548
    },
    {
      "epoch": 1.1897316219369896,
      "grad_norm": 0.8984375,
      "learning_rate": 0.000193139865556432,
      "loss": 0.1953,
      "step": 2549
    },
    {
      "epoch": 1.1901983663943991,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019313452339506015,
      "loss": 0.1683,
      "step": 2550
    },
    {
      "epoch": 1.1906651108518087,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019312917922839867,
      "loss": 0.178,
      "step": 2551
    },
    {
      "epoch": 1.1911318553092183,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019312383305656263,
      "loss": 0.1987,
      "step": 2552
    },
    {
      "epoch": 1.1915985997666279,
      "grad_norm": 1.125,
      "learning_rate": 0.00019311848487966714,
      "loss": 0.2476,
      "step": 2553
    },
    {
      "epoch": 1.1920653442240374,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019311313469782737,
      "loss": 0.238,
      "step": 2554
    },
    {
      "epoch": 1.192532088681447,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001931077825111585,
      "loss": 0.1729,
      "step": 2555
    },
    {
      "epoch": 1.1929988331388564,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019310242831977573,
      "loss": 0.1716,
      "step": 2556
    },
    {
      "epoch": 1.1934655775962661,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019309707212379442,
      "loss": 0.2436,
      "step": 2557
    },
    {
      "epoch": 1.1939323220536755,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019309171392332982,
      "loss": 0.1739,
      "step": 2558
    },
    {
      "epoch": 1.1943990665110853,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001930863537184974,
      "loss": 0.1955,
      "step": 2559
    },
    {
      "epoch": 1.1948658109684946,
      "grad_norm": 1.125,
      "learning_rate": 0.00019308099150941243,
      "loss": 0.2132,
      "step": 2560
    },
    {
      "epoch": 1.1953325554259044,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019307562729619046,
      "loss": 0.2133,
      "step": 2561
    },
    {
      "epoch": 1.1957992998833138,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000193070261078947,
      "loss": 0.2074,
      "step": 2562
    },
    {
      "epoch": 1.1962660443407234,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001930648928577975,
      "loss": 0.2746,
      "step": 2563
    },
    {
      "epoch": 1.196732788798133,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019305952263285763,
      "loss": 0.2192,
      "step": 2564
    },
    {
      "epoch": 1.1971995332555425,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000193054150404243,
      "loss": 0.2045,
      "step": 2565
    },
    {
      "epoch": 1.197666277712952,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019304877617206922,
      "loss": 0.1731,
      "step": 2566
    },
    {
      "epoch": 1.1981330221703617,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001930433999364521,
      "loss": 0.1934,
      "step": 2567
    },
    {
      "epoch": 1.1985997666277712,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019303802169750732,
      "loss": 0.1758,
      "step": 2568
    },
    {
      "epoch": 1.1990665110851808,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001930326414553507,
      "loss": 0.1535,
      "step": 2569
    },
    {
      "epoch": 1.1995332555425904,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001930272592100981,
      "loss": 0.2302,
      "step": 2570
    },
    {
      "epoch": 1.2,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001930218749618654,
      "loss": 0.2061,
      "step": 2571
    },
    {
      "epoch": 1.2004667444574095,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001930164887107685,
      "loss": 0.245,
      "step": 2572
    },
    {
      "epoch": 1.200933488914819,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019301110045692342,
      "loss": 0.2025,
      "step": 2573
    },
    {
      "epoch": 1.2014002333722287,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001930057102004461,
      "loss": 0.1918,
      "step": 2574
    },
    {
      "epoch": 1.2018669778296382,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001930003179414527,
      "loss": 0.1863,
      "step": 2575
    },
    {
      "epoch": 1.2023337222870478,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019299492368005927,
      "loss": 0.1803,
      "step": 2576
    },
    {
      "epoch": 1.2028004667444574,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019298952741638194,
      "loss": 0.1946,
      "step": 2577
    },
    {
      "epoch": 1.203267211201867,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019298412915053688,
      "loss": 0.2358,
      "step": 2578
    },
    {
      "epoch": 1.2037339556592765,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001929787288826404,
      "loss": 0.1976,
      "step": 2579
    },
    {
      "epoch": 1.2042007001166861,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001929733266128087,
      "loss": 0.1949,
      "step": 2580
    },
    {
      "epoch": 1.2046674445740957,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019296792234115812,
      "loss": 0.2932,
      "step": 2581
    },
    {
      "epoch": 1.2051341890315053,
      "grad_norm": 1.125,
      "learning_rate": 0.00019296251606780505,
      "loss": 0.25,
      "step": 2582
    },
    {
      "epoch": 1.2056009334889148,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019295710779286584,
      "loss": 0.1869,
      "step": 2583
    },
    {
      "epoch": 1.2060676779463244,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019295169751645696,
      "loss": 0.1854,
      "step": 2584
    },
    {
      "epoch": 1.206534422403734,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001929462852386949,
      "loss": 0.1981,
      "step": 2585
    },
    {
      "epoch": 1.2070011668611436,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019294087095969622,
      "loss": 0.1741,
      "step": 2586
    },
    {
      "epoch": 1.2074679113185531,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019293545467957744,
      "loss": 0.2241,
      "step": 2587
    },
    {
      "epoch": 1.2079346557759627,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019293003639845522,
      "loss": 0.1876,
      "step": 2588
    },
    {
      "epoch": 1.2084014002333723,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001929246161164462,
      "loss": 0.273,
      "step": 2589
    },
    {
      "epoch": 1.2088681446907819,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001929191938336671,
      "loss": 0.2252,
      "step": 2590
    },
    {
      "epoch": 1.2093348891481914,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019291376955023467,
      "loss": 0.1794,
      "step": 2591
    },
    {
      "epoch": 1.209801633605601,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001929083432662657,
      "loss": 0.2024,
      "step": 2592
    },
    {
      "epoch": 1.2102683780630106,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019290291498187697,
      "loss": 0.2628,
      "step": 2593
    },
    {
      "epoch": 1.2107351225204201,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019289748469718542,
      "loss": 0.2353,
      "step": 2594
    },
    {
      "epoch": 1.2112018669778297,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019289205241230798,
      "loss": 0.2071,
      "step": 2595
    },
    {
      "epoch": 1.2116686114352393,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019288661812736155,
      "loss": 0.2128,
      "step": 2596
    },
    {
      "epoch": 1.2121353558926489,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001928811818424632,
      "loss": 0.1791,
      "step": 2597
    },
    {
      "epoch": 1.2126021003500584,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019287574355772994,
      "loss": 0.2009,
      "step": 2598
    },
    {
      "epoch": 1.2130688448074678,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019287030327327887,
      "loss": 0.1913,
      "step": 2599
    },
    {
      "epoch": 1.2135355892648776,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019286486098922713,
      "loss": 0.1723,
      "step": 2600
    },
    {
      "epoch": 1.214002333722287,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019285941670569193,
      "loss": 0.2285,
      "step": 2601
    },
    {
      "epoch": 1.2144690781796967,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019285397042279043,
      "loss": 0.1893,
      "step": 2602
    },
    {
      "epoch": 1.214935822637106,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019284852214063994,
      "loss": 0.2155,
      "step": 2603
    },
    {
      "epoch": 1.2154025670945157,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019284307185935773,
      "loss": 0.2071,
      "step": 2604
    },
    {
      "epoch": 1.2158693115519252,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019283761957906116,
      "loss": 0.2217,
      "step": 2605
    },
    {
      "epoch": 1.2163360560093348,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019283216529986764,
      "loss": 0.2731,
      "step": 2606
    },
    {
      "epoch": 1.2168028004667444,
      "grad_norm": 0.828125,
      "learning_rate": 0.00019282670902189464,
      "loss": 0.1293,
      "step": 2607
    },
    {
      "epoch": 1.217269544924154,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019282125074525957,
      "loss": 0.2126,
      "step": 2608
    },
    {
      "epoch": 1.2177362893815635,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00019281579047008,
      "loss": 0.244,
      "step": 2609
    },
    {
      "epoch": 1.218203033838973,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001928103281964735,
      "loss": 0.1617,
      "step": 2610
    },
    {
      "epoch": 1.2186697782963827,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001928048639245576,
      "loss": 0.2016,
      "step": 2611
    },
    {
      "epoch": 1.2191365227537923,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019279939765445007,
      "loss": 0.2403,
      "step": 2612
    },
    {
      "epoch": 1.2196032672112018,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019279392938626851,
      "loss": 0.1594,
      "step": 2613
    },
    {
      "epoch": 1.2200700116686114,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001927884591201307,
      "loss": 0.1849,
      "step": 2614
    },
    {
      "epoch": 1.220536756126021,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019278298685615442,
      "loss": 0.1904,
      "step": 2615
    },
    {
      "epoch": 1.2210035005834305,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019277751259445748,
      "loss": 0.2165,
      "step": 2616
    },
    {
      "epoch": 1.2214702450408401,
      "grad_norm": 0.890625,
      "learning_rate": 0.00019277203633515777,
      "loss": 0.1595,
      "step": 2617
    },
    {
      "epoch": 1.2219369894982497,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019276655807837315,
      "loss": 0.2092,
      "step": 2618
    },
    {
      "epoch": 1.2224037339556593,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019276107782422166,
      "loss": 0.1396,
      "step": 2619
    },
    {
      "epoch": 1.2228704784130688,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019275559557282122,
      "loss": 0.1837,
      "step": 2620
    },
    {
      "epoch": 1.2233372228704784,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019275011132428987,
      "loss": 0.1566,
      "step": 2621
    },
    {
      "epoch": 1.223803967327888,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019274462507874575,
      "loss": 0.1832,
      "step": 2622
    },
    {
      "epoch": 1.2242707117852976,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019273913683630692,
      "loss": 0.1714,
      "step": 2623
    },
    {
      "epoch": 1.2247374562427071,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019273364659709162,
      "loss": 0.1839,
      "step": 2624
    },
    {
      "epoch": 1.2252042007001167,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019272815436121798,
      "loss": 0.2136,
      "step": 2625
    },
    {
      "epoch": 1.2256709451575263,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001927226601288043,
      "loss": 0.1723,
      "step": 2626
    },
    {
      "epoch": 1.2261376896149359,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019271716389996886,
      "loss": 0.1657,
      "step": 2627
    },
    {
      "epoch": 1.2266044340723454,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019271166567483004,
      "loss": 0.2068,
      "step": 2628
    },
    {
      "epoch": 1.227071178529755,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019270616545350618,
      "loss": 0.2109,
      "step": 2629
    },
    {
      "epoch": 1.2275379229871646,
      "grad_norm": 0.875,
      "learning_rate": 0.00019270066323611573,
      "loss": 0.1497,
      "step": 2630
    },
    {
      "epoch": 1.2280046674445741,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019269515902277715,
      "loss": 0.1843,
      "step": 2631
    },
    {
      "epoch": 1.2284714119019837,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019268965281360893,
      "loss": 0.2663,
      "step": 2632
    },
    {
      "epoch": 1.2289381563593933,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019268414460872963,
      "loss": 0.1725,
      "step": 2633
    },
    {
      "epoch": 1.2294049008168029,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001926786344082579,
      "loss": 0.2056,
      "step": 2634
    },
    {
      "epoch": 1.2298716452742124,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001926731222123123,
      "loss": 0.3282,
      "step": 2635
    },
    {
      "epoch": 1.230338389731622,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001926676080210116,
      "loss": 0.1815,
      "step": 2636
    },
    {
      "epoch": 1.2308051341890316,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00019266209183447445,
      "loss": 0.1998,
      "step": 2637
    },
    {
      "epoch": 1.2312718786464412,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019265657365281967,
      "loss": 0.1926,
      "step": 2638
    },
    {
      "epoch": 1.2317386231038507,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019265105347616602,
      "loss": 0.2383,
      "step": 2639
    },
    {
      "epoch": 1.23220536756126,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019264553130463242,
      "loss": 0.1994,
      "step": 2640
    },
    {
      "epoch": 1.2326721120186699,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019264000713833777,
      "loss": 0.1897,
      "step": 2641
    },
    {
      "epoch": 1.2331388564760792,
      "grad_norm": 1.125,
      "learning_rate": 0.00019263448097740096,
      "loss": 0.2974,
      "step": 2642
    },
    {
      "epoch": 1.233605600933489,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019262895282194098,
      "loss": 0.2032,
      "step": 2643
    },
    {
      "epoch": 1.2340723453908984,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019262342267207687,
      "loss": 0.1618,
      "step": 2644
    },
    {
      "epoch": 1.234539089848308,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001926178905279277,
      "loss": 0.2134,
      "step": 2645
    },
    {
      "epoch": 1.2350058343057175,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019261235638961258,
      "loss": 0.2682,
      "step": 2646
    },
    {
      "epoch": 1.235472578763127,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001926068202572507,
      "loss": 0.2019,
      "step": 2647
    },
    {
      "epoch": 1.2359393232205367,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019260128213096123,
      "loss": 0.1589,
      "step": 2648
    },
    {
      "epoch": 1.2364060676779463,
      "grad_norm": 1.125,
      "learning_rate": 0.0001925957420108634,
      "loss": 0.1959,
      "step": 2649
    },
    {
      "epoch": 1.2368728121353558,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019259019989707648,
      "loss": 0.191,
      "step": 2650
    },
    {
      "epoch": 1.2373395565927654,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019258465578971988,
      "loss": 0.1931,
      "step": 2651
    },
    {
      "epoch": 1.237806301050175,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001925791096889129,
      "loss": 0.2238,
      "step": 2652
    },
    {
      "epoch": 1.2382730455075845,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019257356159477494,
      "loss": 0.1554,
      "step": 2653
    },
    {
      "epoch": 1.2387397899649941,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001925680115074255,
      "loss": 0.1976,
      "step": 2654
    },
    {
      "epoch": 1.2392065344224037,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019256245942698408,
      "loss": 0.2239,
      "step": 2655
    },
    {
      "epoch": 1.2396732788798133,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001925569053535702,
      "loss": 0.1902,
      "step": 2656
    },
    {
      "epoch": 1.2401400233372228,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019255134928730346,
      "loss": 0.1886,
      "step": 2657
    },
    {
      "epoch": 1.2406067677946324,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019254579122830346,
      "loss": 0.2132,
      "step": 2658
    },
    {
      "epoch": 1.241073512252042,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001925402311766899,
      "loss": 0.1374,
      "step": 2659
    },
    {
      "epoch": 1.2415402567094516,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019253466913258252,
      "loss": 0.2233,
      "step": 2660
    },
    {
      "epoch": 1.2420070011668611,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019252910509610102,
      "loss": 0.1861,
      "step": 2661
    },
    {
      "epoch": 1.2424737456242707,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019252353906736524,
      "loss": 0.1634,
      "step": 2662
    },
    {
      "epoch": 1.2429404900816803,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019251797104649502,
      "loss": 0.2075,
      "step": 2663
    },
    {
      "epoch": 1.2434072345390899,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019251240103361017,
      "loss": 0.1733,
      "step": 2664
    },
    {
      "epoch": 1.2438739789964994,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019250682902883072,
      "loss": 0.1776,
      "step": 2665
    },
    {
      "epoch": 1.244340723453909,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001925012550322766,
      "loss": 0.2666,
      "step": 2666
    },
    {
      "epoch": 1.2448074679113186,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019249567904406785,
      "loss": 0.1493,
      "step": 2667
    },
    {
      "epoch": 1.2452742123687282,
      "grad_norm": 1.0,
      "learning_rate": 0.00019249010106432447,
      "loss": 0.1723,
      "step": 2668
    },
    {
      "epoch": 1.2457409568261377,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019248452109316662,
      "loss": 0.2069,
      "step": 2669
    },
    {
      "epoch": 1.2462077012835473,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001924789391307144,
      "loss": 0.1732,
      "step": 2670
    },
    {
      "epoch": 1.2466744457409569,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019247335517708802,
      "loss": 0.178,
      "step": 2671
    },
    {
      "epoch": 1.2471411901983664,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001924677692324077,
      "loss": 0.1849,
      "step": 2672
    },
    {
      "epoch": 1.247607934655776,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00019246218129679373,
      "loss": 0.2077,
      "step": 2673
    },
    {
      "epoch": 1.2480746791131856,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001924565913703664,
      "loss": 0.2684,
      "step": 2674
    },
    {
      "epoch": 1.2485414235705952,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019245099945324604,
      "loss": 0.2062,
      "step": 2675
    },
    {
      "epoch": 1.2490081680280047,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019244540554555313,
      "loss": 0.2145,
      "step": 2676
    },
    {
      "epoch": 1.2494749124854143,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019243980964740804,
      "loss": 0.2253,
      "step": 2677
    },
    {
      "epoch": 1.2499416569428239,
      "grad_norm": 1.3359375,
      "learning_rate": 0.0001924342117589313,
      "loss": 0.258,
      "step": 2678
    },
    {
      "epoch": 1.2504084014002332,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001924286118802434,
      "loss": 0.1971,
      "step": 2679
    },
    {
      "epoch": 1.250875145857643,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019242301001146496,
      "loss": 0.1894,
      "step": 2680
    },
    {
      "epoch": 1.250875145857643,
      "eval_loss": 1.2540687322616577,
      "eval_runtime": 94.4185,
      "eval_samples_per_second": 19.106,
      "eval_steps_per_second": 2.394,
      "step": 2680
    },
    {
      "epoch": 1.2513418903150524,
      "grad_norm": 0.875,
      "learning_rate": 0.00019241740615271659,
      "loss": 0.1938,
      "step": 2681
    },
    {
      "epoch": 1.2518086347724622,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001924118003041189,
      "loss": 0.155,
      "step": 2682
    },
    {
      "epoch": 1.2522753792298715,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001924061924657926,
      "loss": 0.1516,
      "step": 2683
    },
    {
      "epoch": 1.2527421236872813,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001924005826378585,
      "loss": 0.1383,
      "step": 2684
    },
    {
      "epoch": 1.2532088681446907,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019239497082043728,
      "loss": 0.2517,
      "step": 2685
    },
    {
      "epoch": 1.2536756126021005,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019238935701364988,
      "loss": 0.1693,
      "step": 2686
    },
    {
      "epoch": 1.2541423570595098,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019238374121761703,
      "loss": 0.2126,
      "step": 2687
    },
    {
      "epoch": 1.2546091015169196,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001923781234324598,
      "loss": 0.193,
      "step": 2688
    },
    {
      "epoch": 1.255075845974329,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019237250365829908,
      "loss": 0.1817,
      "step": 2689
    },
    {
      "epoch": 1.2555425904317385,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019236688189525585,
      "loss": 0.2106,
      "step": 2690
    },
    {
      "epoch": 1.2560093348891481,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019236125814345116,
      "loss": 0.1914,
      "step": 2691
    },
    {
      "epoch": 1.2564760793465577,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001923556324030061,
      "loss": 0.1825,
      "step": 2692
    },
    {
      "epoch": 1.2569428238039673,
      "grad_norm": 1.0,
      "learning_rate": 0.00019235000467404185,
      "loss": 0.153,
      "step": 2693
    },
    {
      "epoch": 1.2574095682613768,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019234437495667948,
      "loss": 0.2263,
      "step": 2694
    },
    {
      "epoch": 1.2578763127187864,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001923387432510403,
      "loss": 0.2134,
      "step": 2695
    },
    {
      "epoch": 1.258343057176196,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001923331095572455,
      "loss": 0.1849,
      "step": 2696
    },
    {
      "epoch": 1.2588098016336056,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001923274738754164,
      "loss": 0.1565,
      "step": 2697
    },
    {
      "epoch": 1.2592765460910151,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019232183620567436,
      "loss": 0.2124,
      "step": 2698
    },
    {
      "epoch": 1.2597432905484247,
      "grad_norm": 1.0,
      "learning_rate": 0.00019231619654814077,
      "loss": 0.1587,
      "step": 2699
    },
    {
      "epoch": 1.2602100350058343,
      "grad_norm": 1.03125,
      "learning_rate": 0.000192310554902937,
      "loss": 0.1961,
      "step": 2700
    },
    {
      "epoch": 1.2606767794632439,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019230491127018463,
      "loss": 0.2365,
      "step": 2701
    },
    {
      "epoch": 1.2611435239206534,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00019229926565000506,
      "loss": 0.1328,
      "step": 2702
    },
    {
      "epoch": 1.261610268378063,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001922936180425199,
      "loss": 0.1529,
      "step": 2703
    },
    {
      "epoch": 1.2620770128354726,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019228796844785072,
      "loss": 0.1803,
      "step": 2704
    },
    {
      "epoch": 1.2625437572928822,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019228231686611923,
      "loss": 0.1732,
      "step": 2705
    },
    {
      "epoch": 1.2630105017502917,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019227666329744704,
      "loss": 0.1984,
      "step": 2706
    },
    {
      "epoch": 1.2634772462077013,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001922710077419559,
      "loss": 0.2114,
      "step": 2707
    },
    {
      "epoch": 1.2639439906651109,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019226535019976763,
      "loss": 0.1533,
      "step": 2708
    },
    {
      "epoch": 1.2644107351225204,
      "grad_norm": 1.125,
      "learning_rate": 0.00019225969067100397,
      "loss": 0.1754,
      "step": 2709
    },
    {
      "epoch": 1.26487747957993,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019225402915578677,
      "loss": 0.2632,
      "step": 2710
    },
    {
      "epoch": 1.2653442240373396,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019224836565423801,
      "loss": 0.1809,
      "step": 2711
    },
    {
      "epoch": 1.2658109684947492,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019224270016647958,
      "loss": 0.1861,
      "step": 2712
    },
    {
      "epoch": 1.2662777129521587,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019223703269263345,
      "loss": 0.1918,
      "step": 2713
    },
    {
      "epoch": 1.2667444574095683,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001922313632328217,
      "loss": 0.1902,
      "step": 2714
    },
    {
      "epoch": 1.2672112018669779,
      "grad_norm": 1.0,
      "learning_rate": 0.00019222569178716634,
      "loss": 0.1934,
      "step": 2715
    },
    {
      "epoch": 1.2676779463243875,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019222001835578954,
      "loss": 0.1607,
      "step": 2716
    },
    {
      "epoch": 1.268144690781797,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019221434293881337,
      "loss": 0.1728,
      "step": 2717
    },
    {
      "epoch": 1.2686114352392066,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019220866553636015,
      "loss": 0.1794,
      "step": 2718
    },
    {
      "epoch": 1.2690781796966162,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00019220298614855202,
      "loss": 0.1652,
      "step": 2719
    },
    {
      "epoch": 1.2695449241540255,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00019219730477551127,
      "loss": 0.1125,
      "step": 2720
    },
    {
      "epoch": 1.2700116686114353,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019219162141736028,
      "loss": 0.2013,
      "step": 2721
    },
    {
      "epoch": 1.2704784130688447,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001921859360742214,
      "loss": 0.1761,
      "step": 2722
    },
    {
      "epoch": 1.2709451575262545,
      "grad_norm": 1.203125,
      "learning_rate": 0.00019218024874621703,
      "loss": 0.2777,
      "step": 2723
    },
    {
      "epoch": 1.2714119019836638,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019217455943346964,
      "loss": 0.1714,
      "step": 2724
    },
    {
      "epoch": 1.2718786464410736,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001921688681361017,
      "loss": 0.2285,
      "step": 2725
    },
    {
      "epoch": 1.272345390898483,
      "grad_norm": 1.0,
      "learning_rate": 0.00019216317485423577,
      "loss": 0.1795,
      "step": 2726
    },
    {
      "epoch": 1.2728121353558928,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019215747958799444,
      "loss": 0.1628,
      "step": 2727
    },
    {
      "epoch": 1.2732788798133021,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001921517823375003,
      "loss": 0.2006,
      "step": 2728
    },
    {
      "epoch": 1.273745624270712,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019214608310287606,
      "loss": 0.1517,
      "step": 2729
    },
    {
      "epoch": 1.2742123687281213,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001921403818842444,
      "loss": 0.183,
      "step": 2730
    },
    {
      "epoch": 1.2746791131855308,
      "grad_norm": 1.0,
      "learning_rate": 0.00019213467868172809,
      "loss": 0.189,
      "step": 2731
    },
    {
      "epoch": 1.2751458576429404,
      "grad_norm": 0.8125,
      "learning_rate": 0.00019212897349544993,
      "loss": 0.1537,
      "step": 2732
    },
    {
      "epoch": 1.27561260210035,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019212326632553274,
      "loss": 0.1602,
      "step": 2733
    },
    {
      "epoch": 1.2760793465577596,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019211755717209942,
      "loss": 0.1194,
      "step": 2734
    },
    {
      "epoch": 1.2765460910151691,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001921118460352729,
      "loss": 0.1566,
      "step": 2735
    },
    {
      "epoch": 1.2770128354725787,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019210613291517615,
      "loss": 0.1927,
      "step": 2736
    },
    {
      "epoch": 1.2774795799299883,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001921004178119321,
      "loss": 0.196,
      "step": 2737
    },
    {
      "epoch": 1.2779463243873979,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019209470072566393,
      "loss": 0.2736,
      "step": 2738
    },
    {
      "epoch": 1.2784130688448074,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019208898165649463,
      "loss": 0.2075,
      "step": 2739
    },
    {
      "epoch": 1.278879813302217,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019208326060454742,
      "loss": 0.1691,
      "step": 2740
    },
    {
      "epoch": 1.2793465577596266,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019207753756994542,
      "loss": 0.1904,
      "step": 2741
    },
    {
      "epoch": 1.2798133022170362,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019207181255281186,
      "loss": 0.1766,
      "step": 2742
    },
    {
      "epoch": 1.2802800466744457,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019206608555327005,
      "loss": 0.2311,
      "step": 2743
    },
    {
      "epoch": 1.2807467911318553,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019206035657144324,
      "loss": 0.1986,
      "step": 2744
    },
    {
      "epoch": 1.2812135355892649,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019205462560745483,
      "loss": 0.2101,
      "step": 2745
    },
    {
      "epoch": 1.2816802800466744,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001920488926614282,
      "loss": 0.1721,
      "step": 2746
    },
    {
      "epoch": 1.282147024504084,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019204315773348676,
      "loss": 0.1847,
      "step": 2747
    },
    {
      "epoch": 1.2826137689614936,
      "grad_norm": 1.1171875,
      "learning_rate": 0.000192037420823754,
      "loss": 0.1453,
      "step": 2748
    },
    {
      "epoch": 1.2830805134189032,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001920316819323535,
      "loss": 0.1495,
      "step": 2749
    },
    {
      "epoch": 1.2835472578763127,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00019202594105940875,
      "loss": 0.2295,
      "step": 2750
    },
    {
      "epoch": 1.2840140023337223,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019202019820504336,
      "loss": 0.1825,
      "step": 2751
    },
    {
      "epoch": 1.2844807467911319,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019201445336938105,
      "loss": 0.178,
      "step": 2752
    },
    {
      "epoch": 1.2849474912485415,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019200870655254544,
      "loss": 0.14,
      "step": 2753
    },
    {
      "epoch": 1.285414235705951,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001920029577546603,
      "loss": 0.1509,
      "step": 2754
    },
    {
      "epoch": 1.2858809801633606,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001919972069758494,
      "loss": 0.2064,
      "step": 2755
    },
    {
      "epoch": 1.2863477246207702,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019199145421623656,
      "loss": 0.192,
      "step": 2756
    },
    {
      "epoch": 1.2868144690781798,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019198569947594566,
      "loss": 0.1495,
      "step": 2757
    },
    {
      "epoch": 1.2872812135355893,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019197994275510055,
      "loss": 0.1765,
      "step": 2758
    },
    {
      "epoch": 1.287747957992999,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019197418405382527,
      "loss": 0.1473,
      "step": 2759
    },
    {
      "epoch": 1.2882147024504085,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019196842337224372,
      "loss": 0.175,
      "step": 2760
    },
    {
      "epoch": 1.288681446907818,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019196266071047998,
      "loss": 0.1873,
      "step": 2761
    },
    {
      "epoch": 1.2891481913652276,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019195689606865812,
      "loss": 0.169,
      "step": 2762
    },
    {
      "epoch": 1.289614935822637,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019195112944690225,
      "loss": 0.1428,
      "step": 2763
    },
    {
      "epoch": 1.2900816802800468,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019194536084533658,
      "loss": 0.1059,
      "step": 2764
    },
    {
      "epoch": 1.2905484247374561,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019193959026408524,
      "loss": 0.1372,
      "step": 2765
    },
    {
      "epoch": 1.291015169194866,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019193381770327253,
      "loss": 0.2027,
      "step": 2766
    },
    {
      "epoch": 1.2914819136522753,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019192804316302273,
      "loss": 0.2615,
      "step": 2767
    },
    {
      "epoch": 1.291948658109685,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019192226664346018,
      "loss": 0.1939,
      "step": 2768
    },
    {
      "epoch": 1.2924154025670944,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001919164881447092,
      "loss": 0.2044,
      "step": 2769
    },
    {
      "epoch": 1.2928821470245042,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019191070766689427,
      "loss": 0.1697,
      "step": 2770
    },
    {
      "epoch": 1.2933488914819136,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001919049252101398,
      "loss": 0.1932,
      "step": 2771
    },
    {
      "epoch": 1.2938156359393231,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019189914077457037,
      "loss": 0.1852,
      "step": 2772
    },
    {
      "epoch": 1.2942823803967327,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019189335436031044,
      "loss": 0.2019,
      "step": 2773
    },
    {
      "epoch": 1.2947491248541423,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00019188756596748462,
      "loss": 0.1645,
      "step": 2774
    },
    {
      "epoch": 1.2952158693115519,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001918817755962176,
      "loss": 0.1413,
      "step": 2775
    },
    {
      "epoch": 1.2956826137689614,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019187598324663395,
      "loss": 0.1621,
      "step": 2776
    },
    {
      "epoch": 1.296149358226371,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019187018891885847,
      "loss": 0.1889,
      "step": 2777
    },
    {
      "epoch": 1.2966161026837806,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001918643926130159,
      "loss": 0.2034,
      "step": 2778
    },
    {
      "epoch": 1.2970828471411902,
      "grad_norm": 1.0,
      "learning_rate": 0.00019185859432923103,
      "loss": 0.195,
      "step": 2779
    },
    {
      "epoch": 1.2975495915985997,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001918527940676287,
      "loss": 0.2068,
      "step": 2780
    },
    {
      "epoch": 1.2980163360560093,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019184699182833378,
      "loss": 0.2034,
      "step": 2781
    },
    {
      "epoch": 1.2984830805134189,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019184118761147125,
      "loss": 0.2015,
      "step": 2782
    },
    {
      "epoch": 1.2989498249708284,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019183538141716603,
      "loss": 0.1963,
      "step": 2783
    },
    {
      "epoch": 1.299416569428238,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019182957324554315,
      "loss": 0.2142,
      "step": 2784
    },
    {
      "epoch": 1.2998833138856476,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019182376309672772,
      "loss": 0.1844,
      "step": 2785
    },
    {
      "epoch": 1.3003500583430572,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019181795097084474,
      "loss": 0.1997,
      "step": 2786
    },
    {
      "epoch": 1.3008168028004667,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001918121368680194,
      "loss": 0.1859,
      "step": 2787
    },
    {
      "epoch": 1.3012835472578763,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019180632078837694,
      "loss": 0.1309,
      "step": 2788
    },
    {
      "epoch": 1.301750291715286,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019180050273204252,
      "loss": 0.1696,
      "step": 2789
    },
    {
      "epoch": 1.3022170361726955,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019179468269914137,
      "loss": 0.1662,
      "step": 2790
    },
    {
      "epoch": 1.302683780630105,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001917888606897989,
      "loss": 0.1974,
      "step": 2791
    },
    {
      "epoch": 1.3031505250875146,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001917830367041404,
      "loss": 0.1475,
      "step": 2792
    },
    {
      "epoch": 1.3036172695449242,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019177721074229126,
      "loss": 0.2007,
      "step": 2793
    },
    {
      "epoch": 1.3040840140023338,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000191771382804377,
      "loss": 0.1717,
      "step": 2794
    },
    {
      "epoch": 1.3045507584597433,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019176555289052301,
      "loss": 0.1578,
      "step": 2795
    },
    {
      "epoch": 1.305017502917153,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019175972100085486,
      "loss": 0.1874,
      "step": 2796
    },
    {
      "epoch": 1.3054842473745625,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019175388713549808,
      "loss": 0.1612,
      "step": 2797
    },
    {
      "epoch": 1.305950991831972,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019174805129457836,
      "loss": 0.2134,
      "step": 2798
    },
    {
      "epoch": 1.3064177362893816,
      "grad_norm": 1.0,
      "learning_rate": 0.00019174221347822125,
      "loss": 0.1829,
      "step": 2799
    },
    {
      "epoch": 1.3068844807467912,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019173637368655254,
      "loss": 0.1626,
      "step": 2800
    },
    {
      "epoch": 1.3073512252042008,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019173053191969789,
      "loss": 0.188,
      "step": 2801
    },
    {
      "epoch": 1.3078179696616103,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019172468817778312,
      "loss": 0.1982,
      "step": 2802
    },
    {
      "epoch": 1.30828471411902,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019171884246093403,
      "loss": 0.1656,
      "step": 2803
    },
    {
      "epoch": 1.3087514585764293,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001917129947692765,
      "loss": 0.2287,
      "step": 2804
    },
    {
      "epoch": 1.309218203033839,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019170714510293645,
      "loss": 0.2813,
      "step": 2805
    },
    {
      "epoch": 1.3096849474912484,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001917012934620398,
      "loss": 0.131,
      "step": 2806
    },
    {
      "epoch": 1.3101516919486582,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019169543984671257,
      "loss": 0.1602,
      "step": 2807
    },
    {
      "epoch": 1.3106184364060676,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001916895842570808,
      "loss": 0.1949,
      "step": 2808
    },
    {
      "epoch": 1.3110851808634774,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001916837266932705,
      "loss": 0.1493,
      "step": 2809
    },
    {
      "epoch": 1.3115519253208867,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00019167786715540788,
      "loss": 0.1412,
      "step": 2810
    },
    {
      "epoch": 1.3120186697782965,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019167200564361905,
      "loss": 0.1374,
      "step": 2811
    },
    {
      "epoch": 1.3124854142357059,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019166614215803022,
      "loss": 0.2021,
      "step": 2812
    },
    {
      "epoch": 1.3129521586931157,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019166027669876768,
      "loss": 0.1328,
      "step": 2813
    },
    {
      "epoch": 1.313418903150525,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019165440926595764,
      "loss": 0.1806,
      "step": 2814
    },
    {
      "epoch": 1.3138856476079346,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001916485398597265,
      "loss": 0.211,
      "step": 2815
    },
    {
      "epoch": 1.3143523920653442,
      "grad_norm": 1.125,
      "learning_rate": 0.0001916426684802006,
      "loss": 0.1906,
      "step": 2816
    },
    {
      "epoch": 1.3148191365227537,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019163679512750634,
      "loss": 0.1799,
      "step": 2817
    },
    {
      "epoch": 1.3152858809801633,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019163091980177024,
      "loss": 0.1887,
      "step": 2818
    },
    {
      "epoch": 1.3157526254375729,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001916250425031188,
      "loss": 0.1652,
      "step": 2819
    },
    {
      "epoch": 1.3162193698949824,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019161916323167847,
      "loss": 0.1971,
      "step": 2820
    },
    {
      "epoch": 1.316686114352392,
      "grad_norm": 0.796875,
      "learning_rate": 0.00019161328198757596,
      "loss": 0.1169,
      "step": 2821
    },
    {
      "epoch": 1.3171528588098016,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019160739877093782,
      "loss": 0.1541,
      "step": 2822
    },
    {
      "epoch": 1.3176196032672112,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001916015135818907,
      "loss": 0.1941,
      "step": 2823
    },
    {
      "epoch": 1.3180863477246207,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019159562642056144,
      "loss": 0.1527,
      "step": 2824
    },
    {
      "epoch": 1.3185530921820303,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019158973728707667,
      "loss": 0.2008,
      "step": 2825
    },
    {
      "epoch": 1.31901983663944,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00019158384618156324,
      "loss": 0.1558,
      "step": 2826
    },
    {
      "epoch": 1.3194865810968495,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000191577953104148,
      "loss": 0.164,
      "step": 2827
    },
    {
      "epoch": 1.319953325554259,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001915720580549578,
      "loss": 0.1704,
      "step": 2828
    },
    {
      "epoch": 1.3204200700116686,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001915661610341196,
      "loss": 0.1499,
      "step": 2829
    },
    {
      "epoch": 1.3208868144690782,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019156026204176038,
      "loss": 0.2331,
      "step": 2830
    },
    {
      "epoch": 1.3213535589264878,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001915543610780071,
      "loss": 0.1358,
      "step": 2831
    },
    {
      "epoch": 1.3218203033838973,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019154845814298686,
      "loss": 0.165,
      "step": 2832
    },
    {
      "epoch": 1.322287047841307,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019154255323682672,
      "loss": 0.1723,
      "step": 2833
    },
    {
      "epoch": 1.3227537922987165,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019153664635965386,
      "loss": 0.2014,
      "step": 2834
    },
    {
      "epoch": 1.323220536756126,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019153073751159545,
      "loss": 0.1423,
      "step": 2835
    },
    {
      "epoch": 1.3236872812135356,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001915248266927787,
      "loss": 0.1277,
      "step": 2836
    },
    {
      "epoch": 1.3241540256709452,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019151891390333087,
      "loss": 0.115,
      "step": 2837
    },
    {
      "epoch": 1.3246207701283548,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019151299914337927,
      "loss": 0.1544,
      "step": 2838
    },
    {
      "epoch": 1.3250875145857643,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001915070824130513,
      "loss": 0.1786,
      "step": 2839
    },
    {
      "epoch": 1.325554259043174,
      "grad_norm": 1.2265625,
      "learning_rate": 0.0001915011637124743,
      "loss": 0.1805,
      "step": 2840
    },
    {
      "epoch": 1.3260210035005835,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019149524304177572,
      "loss": 0.1552,
      "step": 2841
    },
    {
      "epoch": 1.326487747957993,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019148932040108306,
      "loss": 0.1509,
      "step": 2842
    },
    {
      "epoch": 1.3269544924154026,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001914833957905238,
      "loss": 0.171,
      "step": 2843
    },
    {
      "epoch": 1.3274212368728122,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019147746921022553,
      "loss": 0.1814,
      "step": 2844
    },
    {
      "epoch": 1.3278879813302216,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019147154066031586,
      "loss": 0.1858,
      "step": 2845
    },
    {
      "epoch": 1.3283547257876314,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019146561014092242,
      "loss": 0.1424,
      "step": 2846
    },
    {
      "epoch": 1.3288214702450407,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019145967765217293,
      "loss": 0.1424,
      "step": 2847
    },
    {
      "epoch": 1.3292882147024505,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019145374319419508,
      "loss": 0.1789,
      "step": 2848
    },
    {
      "epoch": 1.3297549591598599,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019144780676711668,
      "loss": 0.1707,
      "step": 2849
    },
    {
      "epoch": 1.3302217036172697,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019144186837106552,
      "loss": 0.195,
      "step": 2850
    },
    {
      "epoch": 1.330688448074679,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019143592800616952,
      "loss": 0.2097,
      "step": 2851
    },
    {
      "epoch": 1.3311551925320888,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001914299856725565,
      "loss": 0.1411,
      "step": 2852
    },
    {
      "epoch": 1.3316219369894982,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019142404137035448,
      "loss": 0.1315,
      "step": 2853
    },
    {
      "epoch": 1.332088681446908,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019141809509969136,
      "loss": 0.2245,
      "step": 2854
    },
    {
      "epoch": 1.3325554259043173,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019141214686069525,
      "loss": 0.1475,
      "step": 2855
    },
    {
      "epoch": 1.3330221703617269,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019140619665349422,
      "loss": 0.1923,
      "step": 2856
    },
    {
      "epoch": 1.3334889148191365,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019140024447821633,
      "loss": 0.1726,
      "step": 2857
    },
    {
      "epoch": 1.333955659276546,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019139429033498978,
      "loss": 0.1477,
      "step": 2858
    },
    {
      "epoch": 1.3344224037339556,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019138833422394276,
      "loss": 0.2183,
      "step": 2859
    },
    {
      "epoch": 1.3348891481913652,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019138237614520348,
      "loss": 0.1677,
      "step": 2860
    },
    {
      "epoch": 1.3353558926487747,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019137641609890028,
      "loss": 0.2039,
      "step": 2861
    },
    {
      "epoch": 1.3358226371061843,
      "grad_norm": 1.3046875,
      "learning_rate": 0.00019137045408516143,
      "loss": 0.2137,
      "step": 2862
    },
    {
      "epoch": 1.336289381563594,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00019136449010411535,
      "loss": 0.2071,
      "step": 2863
    },
    {
      "epoch": 1.3367561260210035,
      "grad_norm": 1.265625,
      "learning_rate": 0.00019135852415589044,
      "loss": 0.1962,
      "step": 2864
    },
    {
      "epoch": 1.337222870478413,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001913525562406151,
      "loss": 0.1582,
      "step": 2865
    },
    {
      "epoch": 1.3376896149358226,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019134658635841788,
      "loss": 0.1801,
      "step": 2866
    },
    {
      "epoch": 1.3381563593932322,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019134061450942733,
      "loss": 0.1776,
      "step": 2867
    },
    {
      "epoch": 1.3386231038506418,
      "grad_norm": 0.953125,
      "learning_rate": 0.00019133464069377197,
      "loss": 0.1394,
      "step": 2868
    },
    {
      "epoch": 1.3390898483080513,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001913286649115805,
      "loss": 0.2525,
      "step": 2869
    },
    {
      "epoch": 1.339556592765461,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001913226871629815,
      "loss": 0.1332,
      "step": 2870
    },
    {
      "epoch": 1.3400233372228705,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019131670744810376,
      "loss": 0.194,
      "step": 2871
    },
    {
      "epoch": 1.34049008168028,
      "grad_norm": 0.984375,
      "learning_rate": 0.00019131072576707595,
      "loss": 0.1408,
      "step": 2872
    },
    {
      "epoch": 1.3409568261376896,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001913047421200269,
      "loss": 0.2097,
      "step": 2873
    },
    {
      "epoch": 1.3414235705950992,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00019129875650708547,
      "loss": 0.1535,
      "step": 2874
    },
    {
      "epoch": 1.3418903150525088,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00019129276892838055,
      "loss": 0.1313,
      "step": 2875
    },
    {
      "epoch": 1.3423570595099183,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00019128677938404099,
      "loss": 0.1595,
      "step": 2876
    },
    {
      "epoch": 1.342823803967328,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001912807878741958,
      "loss": 0.1708,
      "step": 2877
    },
    {
      "epoch": 1.3432905484247375,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019127479439897394,
      "loss": 0.1833,
      "step": 2878
    },
    {
      "epoch": 1.343757292882147,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001912687989585045,
      "loss": 0.2133,
      "step": 2879
    },
    {
      "epoch": 1.3442240373395566,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019126280155291653,
      "loss": 0.1526,
      "step": 2880
    },
    {
      "epoch": 1.3446907817969662,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019125680218233924,
      "loss": 0.1742,
      "step": 2881
    },
    {
      "epoch": 1.3451575262543758,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001912508008469017,
      "loss": 0.1903,
      "step": 2882
    },
    {
      "epoch": 1.3456242707117854,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019124479754673322,
      "loss": 0.201,
      "step": 2883
    },
    {
      "epoch": 1.346091015169195,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019123879228196295,
      "loss": 0.1898,
      "step": 2884
    },
    {
      "epoch": 1.3465577596266045,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001912327850527203,
      "loss": 0.1923,
      "step": 2885
    },
    {
      "epoch": 1.347024504084014,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019122677585913452,
      "loss": 0.1444,
      "step": 2886
    },
    {
      "epoch": 1.3474912485414237,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001912207647013351,
      "loss": 0.1765,
      "step": 2887
    },
    {
      "epoch": 1.347957992998833,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00019121475157945133,
      "loss": 0.1321,
      "step": 2888
    },
    {
      "epoch": 1.3484247374562428,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019120873649361282,
      "loss": 0.2113,
      "step": 2889
    },
    {
      "epoch": 1.3488914819136522,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019120271944394896,
      "loss": 0.171,
      "step": 2890
    },
    {
      "epoch": 1.349358226371062,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019119670043058943,
      "loss": 0.1867,
      "step": 2891
    },
    {
      "epoch": 1.3498249708284713,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019119067945366368,
      "loss": 0.1535,
      "step": 2892
    },
    {
      "epoch": 1.350291715285881,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019118465651330146,
      "loss": 0.1381,
      "step": 2893
    },
    {
      "epoch": 1.3507584597432905,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001911786316096324,
      "loss": 0.1791,
      "step": 2894
    },
    {
      "epoch": 1.3512252042007002,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019117260474278623,
      "loss": 0.1802,
      "step": 2895
    },
    {
      "epoch": 1.3516919486581096,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019116657591289278,
      "loss": 0.1122,
      "step": 2896
    },
    {
      "epoch": 1.3521586931155192,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00019116054512008176,
      "loss": 0.1794,
      "step": 2897
    },
    {
      "epoch": 1.3526254375729287,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019115451236448305,
      "loss": 0.1581,
      "step": 2898
    },
    {
      "epoch": 1.3530921820303383,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00019114847764622653,
      "loss": 0.1433,
      "step": 2899
    },
    {
      "epoch": 1.353558926487748,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019114244096544221,
      "loss": 0.1771,
      "step": 2900
    },
    {
      "epoch": 1.3540256709451575,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019113640232226,
      "loss": 0.2008,
      "step": 2901
    },
    {
      "epoch": 1.354492415402567,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001911303617168099,
      "loss": 0.2152,
      "step": 2902
    },
    {
      "epoch": 1.3549591598599766,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019112431914922203,
      "loss": 0.1475,
      "step": 2903
    },
    {
      "epoch": 1.3554259043173862,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019111827461962643,
      "loss": 0.1559,
      "step": 2904
    },
    {
      "epoch": 1.3558926487747958,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001911122281281533,
      "loss": 0.148,
      "step": 2905
    },
    {
      "epoch": 1.3563593932322053,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001911061796749328,
      "loss": 0.1458,
      "step": 2906
    },
    {
      "epoch": 1.356826137689615,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001911001292600952,
      "loss": 0.1826,
      "step": 2907
    },
    {
      "epoch": 1.3572928821470245,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019109407688377072,
      "loss": 0.1566,
      "step": 2908
    },
    {
      "epoch": 1.357759626604434,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019108802254608968,
      "loss": 0.2108,
      "step": 2909
    },
    {
      "epoch": 1.3582263710618436,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019108196624718245,
      "loss": 0.194,
      "step": 2910
    },
    {
      "epoch": 1.3586931155192532,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019107590798717947,
      "loss": 0.1409,
      "step": 2911
    },
    {
      "epoch": 1.3591598599766628,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001910698477662111,
      "loss": 0.1518,
      "step": 2912
    },
    {
      "epoch": 1.3596266044340724,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019106378558440788,
      "loss": 0.1766,
      "step": 2913
    },
    {
      "epoch": 1.360093348891482,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019105772144190032,
      "loss": 0.1622,
      "step": 2914
    },
    {
      "epoch": 1.3605600933488915,
      "grad_norm": 1.125,
      "learning_rate": 0.000191051655338819,
      "loss": 0.2193,
      "step": 2915
    },
    {
      "epoch": 1.361026837806301,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019104558727529446,
      "loss": 0.2111,
      "step": 2916
    },
    {
      "epoch": 1.3614935822637106,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00019103951725145746,
      "loss": 0.1782,
      "step": 2917
    },
    {
      "epoch": 1.3619603267211202,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00019103344526743863,
      "loss": 0.1424,
      "step": 2918
    },
    {
      "epoch": 1.3624270711785298,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019102737132336875,
      "loss": 0.2199,
      "step": 2919
    },
    {
      "epoch": 1.3628938156359394,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001910212954193785,
      "loss": 0.1626,
      "step": 2920
    },
    {
      "epoch": 1.363360560093349,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019101521755559883,
      "loss": 0.1582,
      "step": 2921
    },
    {
      "epoch": 1.3638273045507585,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00019100913773216052,
      "loss": 0.1316,
      "step": 2922
    },
    {
      "epoch": 1.364294049008168,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001910030559491945,
      "loss": 0.1846,
      "step": 2923
    },
    {
      "epoch": 1.3647607934655777,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00019099697220683172,
      "loss": 0.1509,
      "step": 2924
    },
    {
      "epoch": 1.3652275379229872,
      "grad_norm": 0.921875,
      "learning_rate": 0.00019099088650520316,
      "loss": 0.1865,
      "step": 2925
    },
    {
      "epoch": 1.3656942823803968,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019098479884443984,
      "loss": 0.2182,
      "step": 2926
    },
    {
      "epoch": 1.3661610268378064,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001909787092246729,
      "loss": 0.1925,
      "step": 2927
    },
    {
      "epoch": 1.366627771295216,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019097261764603339,
      "loss": 0.139,
      "step": 2928
    },
    {
      "epoch": 1.3670945157526253,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019096652410865246,
      "loss": 0.1821,
      "step": 2929
    },
    {
      "epoch": 1.367561260210035,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019096042861266134,
      "loss": 0.1595,
      "step": 2930
    },
    {
      "epoch": 1.3680280046674445,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001909543311581913,
      "loss": 0.1841,
      "step": 2931
    },
    {
      "epoch": 1.3684947491248542,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019094823174537357,
      "loss": 0.192,
      "step": 2932
    },
    {
      "epoch": 1.3689614935822636,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00019094213037433951,
      "loss": 0.1815,
      "step": 2933
    },
    {
      "epoch": 1.3694282380396734,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019093602704522048,
      "loss": 0.1609,
      "step": 2934
    },
    {
      "epoch": 1.3698949824970827,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001909299217581479,
      "loss": 0.1293,
      "step": 2935
    },
    {
      "epoch": 1.3703617269544925,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019092381451325322,
      "loss": 0.138,
      "step": 2936
    },
    {
      "epoch": 1.370828471411902,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001909177053106679,
      "loss": 0.1629,
      "step": 2937
    },
    {
      "epoch": 1.3712952158693117,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00019091159415052353,
      "loss": 0.127,
      "step": 2938
    },
    {
      "epoch": 1.371761960326721,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019090548103295167,
      "loss": 0.1893,
      "step": 2939
    },
    {
      "epoch": 1.3722287047841306,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019089936595808395,
      "loss": 0.1606,
      "step": 2940
    },
    {
      "epoch": 1.3726954492415402,
      "grad_norm": 1.0703125,
      "learning_rate": 0.000190893248926052,
      "loss": 0.1806,
      "step": 2941
    },
    {
      "epoch": 1.3731621936989498,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019088712993698757,
      "loss": 0.1469,
      "step": 2942
    },
    {
      "epoch": 1.3736289381563593,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001908810089910224,
      "loss": 0.1875,
      "step": 2943
    },
    {
      "epoch": 1.374095682613769,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019087488608828825,
      "loss": 0.1216,
      "step": 2944
    },
    {
      "epoch": 1.3745624270711785,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00019086876122891702,
      "loss": 0.1995,
      "step": 2945
    },
    {
      "epoch": 1.375029171528588,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001908626344130405,
      "loss": 0.1312,
      "step": 2946
    },
    {
      "epoch": 1.3754959159859976,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019085650564079065,
      "loss": 0.1448,
      "step": 2947
    },
    {
      "epoch": 1.3759626604434072,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019085037491229945,
      "loss": 0.1716,
      "step": 2948
    },
    {
      "epoch": 1.3764294049008168,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00019084424222769885,
      "loss": 0.25,
      "step": 2949
    },
    {
      "epoch": 1.3768961493582264,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019083810758712094,
      "loss": 0.1126,
      "step": 2950
    },
    {
      "epoch": 1.377362893815636,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019083197099069776,
      "loss": 0.1775,
      "step": 2951
    },
    {
      "epoch": 1.3778296382730455,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019082583243856149,
      "loss": 0.2081,
      "step": 2952
    },
    {
      "epoch": 1.378296382730455,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019081969193084425,
      "loss": 0.2033,
      "step": 2953
    },
    {
      "epoch": 1.3787631271878646,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001908135494676783,
      "loss": 0.1293,
      "step": 2954
    },
    {
      "epoch": 1.3792298716452742,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00019080740504919588,
      "loss": 0.1306,
      "step": 2955
    },
    {
      "epoch": 1.3796966161026838,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019080125867552926,
      "loss": 0.167,
      "step": 2956
    },
    {
      "epoch": 1.3801633605600934,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019079511034681082,
      "loss": 0.1967,
      "step": 2957
    },
    {
      "epoch": 1.380630105017503,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019078896006317287,
      "loss": 0.1147,
      "step": 2958
    },
    {
      "epoch": 1.3810968494749125,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00019078280782474792,
      "loss": 0.1934,
      "step": 2959
    },
    {
      "epoch": 1.381563593932322,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019077665363166838,
      "loss": 0.1559,
      "step": 2960
    },
    {
      "epoch": 1.3820303383897317,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019077049748406674,
      "loss": 0.1676,
      "step": 2961
    },
    {
      "epoch": 1.3824970828471412,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001907643393820756,
      "loss": 0.1637,
      "step": 2962
    },
    {
      "epoch": 1.3829638273045508,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001907581793258275,
      "loss": 0.1129,
      "step": 2963
    },
    {
      "epoch": 1.3834305717619604,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019075201731545515,
      "loss": 0.163,
      "step": 2964
    },
    {
      "epoch": 1.38389731621937,
      "grad_norm": 1.125,
      "learning_rate": 0.00019074585335109116,
      "loss": 0.1903,
      "step": 2965
    },
    {
      "epoch": 1.3843640606767795,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00019073968743286824,
      "loss": 0.1699,
      "step": 2966
    },
    {
      "epoch": 1.384830805134189,
      "grad_norm": 1.125,
      "learning_rate": 0.0001907335195609192,
      "loss": 0.203,
      "step": 2967
    },
    {
      "epoch": 1.3852975495915987,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001907273497353768,
      "loss": 0.1403,
      "step": 2968
    },
    {
      "epoch": 1.3857642940490082,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019072117795637393,
      "loss": 0.1947,
      "step": 2969
    },
    {
      "epoch": 1.3862310385064176,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019071500422404342,
      "loss": 0.1703,
      "step": 2970
    },
    {
      "epoch": 1.3866977829638274,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019070882853851817,
      "loss": 0.1536,
      "step": 2971
    },
    {
      "epoch": 1.3871645274212367,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019070265089993124,
      "loss": 0.1394,
      "step": 2972
    },
    {
      "epoch": 1.3876312718786465,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019069647130841563,
      "loss": 0.1803,
      "step": 2973
    },
    {
      "epoch": 1.388098016336056,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001906902897641043,
      "loss": 0.206,
      "step": 2974
    },
    {
      "epoch": 1.3885647607934657,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019068410626713048,
      "loss": 0.1582,
      "step": 2975
    },
    {
      "epoch": 1.389031505250875,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019067792081762716,
      "loss": 0.1771,
      "step": 2976
    },
    {
      "epoch": 1.3894982497082848,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019067173341572765,
      "loss": 0.155,
      "step": 2977
    },
    {
      "epoch": 1.3899649941656942,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001906655440615651,
      "loss": 0.1885,
      "step": 2978
    },
    {
      "epoch": 1.390431738623104,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00019065935275527278,
      "loss": 0.1256,
      "step": 2979
    },
    {
      "epoch": 1.3908984830805133,
      "grad_norm": 0.90625,
      "learning_rate": 0.00019065315949698403,
      "loss": 0.1455,
      "step": 2980
    },
    {
      "epoch": 1.391365227537923,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00019064696428683218,
      "loss": 0.1925,
      "step": 2981
    },
    {
      "epoch": 1.3918319719953325,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001906407671249506,
      "loss": 0.106,
      "step": 2982
    },
    {
      "epoch": 1.392298716452742,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019063456801147273,
      "loss": 0.1871,
      "step": 2983
    },
    {
      "epoch": 1.3927654609101516,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019062836694653203,
      "loss": 0.1697,
      "step": 2984
    },
    {
      "epoch": 1.3932322053675612,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019062216393026207,
      "loss": 0.1737,
      "step": 2985
    },
    {
      "epoch": 1.3936989498249708,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019061595896279637,
      "loss": 0.1855,
      "step": 2986
    },
    {
      "epoch": 1.3941656942823804,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001906097520442685,
      "loss": 0.1218,
      "step": 2987
    },
    {
      "epoch": 1.39463243873979,
      "grad_norm": 1.0,
      "learning_rate": 0.00019060354317481217,
      "loss": 0.1566,
      "step": 2988
    },
    {
      "epoch": 1.3950991831971995,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019059733235456103,
      "loss": 0.1784,
      "step": 2989
    },
    {
      "epoch": 1.395565927654609,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001905911195836488,
      "loss": 0.1398,
      "step": 2990
    },
    {
      "epoch": 1.3960326721120186,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019058490486220927,
      "loss": 0.1513,
      "step": 2991
    },
    {
      "epoch": 1.3964994165694282,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001905786881903762,
      "loss": 0.1531,
      "step": 2992
    },
    {
      "epoch": 1.3969661610268378,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019057246956828347,
      "loss": 0.1282,
      "step": 2993
    },
    {
      "epoch": 1.3974329054842474,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019056624899606498,
      "loss": 0.179,
      "step": 2994
    },
    {
      "epoch": 1.397899649941657,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001905600264738547,
      "loss": 0.2017,
      "step": 2995
    },
    {
      "epoch": 1.3983663943990665,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00019055380200178653,
      "loss": 0.1518,
      "step": 2996
    },
    {
      "epoch": 1.398833138856476,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019054757557999456,
      "loss": 0.1581,
      "step": 2997
    },
    {
      "epoch": 1.3992998833138857,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019054134720861282,
      "loss": 0.1928,
      "step": 2998
    },
    {
      "epoch": 1.3997666277712952,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019053511688777542,
      "loss": 0.1199,
      "step": 2999
    },
    {
      "epoch": 1.4002333722287048,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001905288846176165,
      "loss": 0.1489,
      "step": 3000
    },
    {
      "epoch": 1.4007001166861144,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019052265039827022,
      "loss": 0.1539,
      "step": 3001
    },
    {
      "epoch": 1.401166861143524,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019051641422987087,
      "loss": 0.1273,
      "step": 3002
    },
    {
      "epoch": 1.4016336056009335,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019051017611255267,
      "loss": 0.1701,
      "step": 3003
    },
    {
      "epoch": 1.402100350058343,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00019050393604645,
      "loss": 0.1478,
      "step": 3004
    },
    {
      "epoch": 1.4025670945157527,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019049769403169716,
      "loss": 0.1715,
      "step": 3005
    },
    {
      "epoch": 1.4030338389731623,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019049145006842853,
      "loss": 0.1369,
      "step": 3006
    },
    {
      "epoch": 1.4035005834305718,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00019048520415677856,
      "loss": 0.1648,
      "step": 3007
    },
    {
      "epoch": 1.4039673278879814,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001904789562968818,
      "loss": 0.1156,
      "step": 3008
    },
    {
      "epoch": 1.404434072345391,
      "grad_norm": 5.40625,
      "learning_rate": 0.0001904727064888727,
      "loss": 0.172,
      "step": 3009
    },
    {
      "epoch": 1.4049008168028005,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00019046645473288587,
      "loss": 0.113,
      "step": 3010
    },
    {
      "epoch": 1.40536756126021,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019046020102905589,
      "loss": 0.1608,
      "step": 3011
    },
    {
      "epoch": 1.4058343057176197,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00019045394537751739,
      "loss": 0.1178,
      "step": 3012
    },
    {
      "epoch": 1.406301050175029,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00019044768777840506,
      "loss": 0.1492,
      "step": 3013
    },
    {
      "epoch": 1.4067677946324388,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00019044142823185372,
      "loss": 0.1734,
      "step": 3014
    },
    {
      "epoch": 1.4072345390898482,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019043516673799806,
      "loss": 0.1534,
      "step": 3015
    },
    {
      "epoch": 1.407701283547258,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019042890329697292,
      "loss": 0.1573,
      "step": 3016
    },
    {
      "epoch": 1.4081680280046673,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019042263790891318,
      "loss": 0.1439,
      "step": 3017
    },
    {
      "epoch": 1.4086347724620771,
      "grad_norm": 1.171875,
      "learning_rate": 0.00019041637057395369,
      "loss": 0.1518,
      "step": 3018
    },
    {
      "epoch": 1.4091015169194865,
      "grad_norm": 1.046875,
      "learning_rate": 0.00019041010129222942,
      "loss": 0.1678,
      "step": 3019
    },
    {
      "epoch": 1.4095682613768963,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001904038300638754,
      "loss": 0.1809,
      "step": 3020
    },
    {
      "epoch": 1.4100350058343056,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00019039755688902656,
      "loss": 0.1796,
      "step": 3021
    },
    {
      "epoch": 1.4105017502917152,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019039128176781805,
      "loss": 0.1495,
      "step": 3022
    },
    {
      "epoch": 1.4109684947491248,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019038500470038494,
      "loss": 0.1861,
      "step": 3023
    },
    {
      "epoch": 1.4114352392065344,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001903787256868624,
      "loss": 0.2207,
      "step": 3024
    },
    {
      "epoch": 1.411901983663944,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00019037244472738563,
      "loss": 0.1352,
      "step": 3025
    },
    {
      "epoch": 1.4123687281213535,
      "grad_norm": 1.125,
      "learning_rate": 0.00019036616182208985,
      "loss": 0.1424,
      "step": 3026
    },
    {
      "epoch": 1.412835472578763,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019035987697111034,
      "loss": 0.1688,
      "step": 3027
    },
    {
      "epoch": 1.4133022170361726,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00019035359017458242,
      "loss": 0.1725,
      "step": 3028
    },
    {
      "epoch": 1.4137689614935822,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001903473014326415,
      "loss": 0.1478,
      "step": 3029
    },
    {
      "epoch": 1.4142357059509918,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019034101074542285,
      "loss": 0.1358,
      "step": 3030
    },
    {
      "epoch": 1.4147024504084014,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00019033471811306205,
      "loss": 0.1742,
      "step": 3031
    },
    {
      "epoch": 1.415169194865811,
      "grad_norm": 1.09375,
      "learning_rate": 0.00019032842353569455,
      "loss": 0.1807,
      "step": 3032
    },
    {
      "epoch": 1.4156359393232205,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019032212701345586,
      "loss": 0.1495,
      "step": 3033
    },
    {
      "epoch": 1.41610268378063,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019031582854648157,
      "loss": 0.2042,
      "step": 3034
    },
    {
      "epoch": 1.4165694282380397,
      "grad_norm": 0.875,
      "learning_rate": 0.00019030952813490725,
      "loss": 0.118,
      "step": 3035
    },
    {
      "epoch": 1.4170361726954492,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019030322577886862,
      "loss": 0.1751,
      "step": 3036
    },
    {
      "epoch": 1.4175029171528588,
      "grad_norm": 1.03125,
      "learning_rate": 0.00019029692147850137,
      "loss": 0.1581,
      "step": 3037
    },
    {
      "epoch": 1.4179696616102684,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019029061523394115,
      "loss": 0.1508,
      "step": 3038
    },
    {
      "epoch": 1.418436406067678,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019028430704532388,
      "loss": 0.137,
      "step": 3039
    },
    {
      "epoch": 1.4189031505250875,
      "grad_norm": 1.125,
      "learning_rate": 0.00019027799691278524,
      "loss": 0.1701,
      "step": 3040
    },
    {
      "epoch": 1.419369894982497,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019027168483646122,
      "loss": 0.1359,
      "step": 3041
    },
    {
      "epoch": 1.4198366394399067,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001902653708164876,
      "loss": 0.1769,
      "step": 3042
    },
    {
      "epoch": 1.4203033838973163,
      "grad_norm": 1.1875,
      "learning_rate": 0.00019025905485300047,
      "loss": 0.1742,
      "step": 3043
    },
    {
      "epoch": 1.4207701283547258,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019025273694613567,
      "loss": 0.1614,
      "step": 3044
    },
    {
      "epoch": 1.4212368728121354,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00019024641709602938,
      "loss": 0.1322,
      "step": 3045
    },
    {
      "epoch": 1.421703617269545,
      "grad_norm": 1.0625,
      "learning_rate": 0.00019024009530281755,
      "loss": 0.1956,
      "step": 3046
    },
    {
      "epoch": 1.4221703617269545,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00019023377156663637,
      "loss": 0.139,
      "step": 3047
    },
    {
      "epoch": 1.4226371061843641,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00019022744588762194,
      "loss": 0.1984,
      "step": 3048
    },
    {
      "epoch": 1.4231038506417737,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001902211182659105,
      "loss": 0.1606,
      "step": 3049
    },
    {
      "epoch": 1.4235705950991833,
      "grad_norm": 1.015625,
      "learning_rate": 0.00019021478870163828,
      "loss": 0.1197,
      "step": 3050
    },
    {
      "epoch": 1.4240373395565928,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001902084571949416,
      "loss": 0.1593,
      "step": 3051
    },
    {
      "epoch": 1.4245040840140024,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00019020212374595673,
      "loss": 0.1109,
      "step": 3052
    },
    {
      "epoch": 1.424970828471412,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00019019578835482003,
      "loss": 0.1716,
      "step": 3053
    },
    {
      "epoch": 1.4254375729288213,
      "grad_norm": 1.109375,
      "learning_rate": 0.00019018945102166796,
      "loss": 0.1986,
      "step": 3054
    },
    {
      "epoch": 1.4259043173862311,
      "grad_norm": 0.96875,
      "learning_rate": 0.00019018311174663695,
      "loss": 0.1271,
      "step": 3055
    },
    {
      "epoch": 1.4263710618436405,
      "grad_norm": 1.140625,
      "learning_rate": 0.00019017677052986346,
      "loss": 0.1247,
      "step": 3056
    },
    {
      "epoch": 1.4268378063010503,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00019017042737148407,
      "loss": 0.1392,
      "step": 3057
    },
    {
      "epoch": 1.4273045507584596,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00019016408227163532,
      "loss": 0.1545,
      "step": 3058
    },
    {
      "epoch": 1.4277712952158694,
      "grad_norm": 0.9375,
      "learning_rate": 0.00019015773523045382,
      "loss": 0.1272,
      "step": 3059
    },
    {
      "epoch": 1.4282380396732788,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00019015138624807628,
      "loss": 0.1074,
      "step": 3060
    },
    {
      "epoch": 1.4287047841306886,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00019014503532463936,
      "loss": 0.147,
      "step": 3061
    },
    {
      "epoch": 1.429171528588098,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00019013868246027982,
      "loss": 0.1672,
      "step": 3062
    },
    {
      "epoch": 1.4296382730455075,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00019013232765513442,
      "loss": 0.1496,
      "step": 3063
    },
    {
      "epoch": 1.430105017502917,
      "grad_norm": 1.15625,
      "learning_rate": 0.00019012597090934,
      "loss": 0.1918,
      "step": 3064
    },
    {
      "epoch": 1.4305717619603266,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00019011961222303346,
      "loss": 0.1492,
      "step": 3065
    },
    {
      "epoch": 1.4310385064177362,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00019011325159635164,
      "loss": 0.0934,
      "step": 3066
    },
    {
      "epoch": 1.4315052508751458,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019010688902943153,
      "loss": 0.1387,
      "step": 3067
    },
    {
      "epoch": 1.4319719953325554,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00019010052452241016,
      "loss": 0.1545,
      "step": 3068
    },
    {
      "epoch": 1.432438739789965,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001900941580754245,
      "loss": 0.1493,
      "step": 3069
    },
    {
      "epoch": 1.4329054842473745,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019008778968861167,
      "loss": 0.1726,
      "step": 3070
    },
    {
      "epoch": 1.433372228704784,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00019008141936210877,
      "loss": 0.1225,
      "step": 3071
    },
    {
      "epoch": 1.4338389731621937,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019007504709605297,
      "loss": 0.1669,
      "step": 3072
    },
    {
      "epoch": 1.4343057176196032,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019006867289058147,
      "loss": 0.1362,
      "step": 3073
    },
    {
      "epoch": 1.4347724620770128,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00019006229674583148,
      "loss": 0.1464,
      "step": 3074
    },
    {
      "epoch": 1.4352392065344224,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00019005591866194034,
      "loss": 0.2105,
      "step": 3075
    },
    {
      "epoch": 1.435705950991832,
      "grad_norm": 1.078125,
      "learning_rate": 0.00019004953863904536,
      "loss": 0.1486,
      "step": 3076
    },
    {
      "epoch": 1.4361726954492415,
      "grad_norm": 1.21875,
      "learning_rate": 0.00019004315667728387,
      "loss": 0.1906,
      "step": 3077
    },
    {
      "epoch": 1.436639439906651,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00019003677277679333,
      "loss": 0.1404,
      "step": 3078
    },
    {
      "epoch": 1.4371061843640607,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00019003038693771117,
      "loss": 0.1547,
      "step": 3079
    },
    {
      "epoch": 1.4375729288214703,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001900239991601749,
      "loss": 0.1294,
      "step": 3080
    },
    {
      "epoch": 1.4380396732788798,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00019001760944432206,
      "loss": 0.1501,
      "step": 3081
    },
    {
      "epoch": 1.4385064177362894,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001900112177902902,
      "loss": 0.1499,
      "step": 3082
    },
    {
      "epoch": 1.438973162193699,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00019000482419821694,
      "loss": 0.128,
      "step": 3083
    },
    {
      "epoch": 1.4394399066511085,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018999842866823994,
      "loss": 0.1374,
      "step": 3084
    },
    {
      "epoch": 1.4399066511085181,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018999203120049693,
      "loss": 0.141,
      "step": 3085
    },
    {
      "epoch": 1.4403733955659277,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018998563179512562,
      "loss": 0.1561,
      "step": 3086
    },
    {
      "epoch": 1.4408401400233373,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018997923045226386,
      "loss": 0.1551,
      "step": 3087
    },
    {
      "epoch": 1.4413068844807468,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018997282717204942,
      "loss": 0.1664,
      "step": 3088
    },
    {
      "epoch": 1.4417736289381564,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001899664219546202,
      "loss": 0.1504,
      "step": 3089
    },
    {
      "epoch": 1.442240373395566,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00018996001480011405,
      "loss": 0.1611,
      "step": 3090
    },
    {
      "epoch": 1.4427071178529756,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018995360570866901,
      "loss": 0.1828,
      "step": 3091
    },
    {
      "epoch": 1.4431738623103851,
      "grad_norm": 1.1796875,
      "learning_rate": 0.000189947194680423,
      "loss": 0.1149,
      "step": 3092
    },
    {
      "epoch": 1.4436406067677947,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001899407817155141,
      "loss": 0.1535,
      "step": 3093
    },
    {
      "epoch": 1.4441073512252043,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001899343668140804,
      "loss": 0.1718,
      "step": 3094
    },
    {
      "epoch": 1.4445740956826136,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018992794997625998,
      "loss": 0.1655,
      "step": 3095
    },
    {
      "epoch": 1.4450408401400234,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000189921531202191,
      "loss": 0.1724,
      "step": 3096
    },
    {
      "epoch": 1.4455075845974328,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018991511049201173,
      "loss": 0.1395,
      "step": 3097
    },
    {
      "epoch": 1.4459743290548426,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018990868784586035,
      "loss": 0.1292,
      "step": 3098
    },
    {
      "epoch": 1.446441073512252,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018990226326387518,
      "loss": 0.1569,
      "step": 3099
    },
    {
      "epoch": 1.4469078179696617,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018989583674619448,
      "loss": 0.1384,
      "step": 3100
    },
    {
      "epoch": 1.447374562427071,
      "grad_norm": 1.7265625,
      "learning_rate": 0.0001898894082929567,
      "loss": 0.183,
      "step": 3101
    },
    {
      "epoch": 1.4478413068844809,
      "grad_norm": 1.3984375,
      "learning_rate": 0.00018988297790430028,
      "loss": 0.1465,
      "step": 3102
    },
    {
      "epoch": 1.4483080513418902,
      "grad_norm": 1.2890625,
      "learning_rate": 0.00018987654558036353,
      "loss": 0.2297,
      "step": 3103
    },
    {
      "epoch": 1.4487747957993,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001898701113212851,
      "loss": 0.1831,
      "step": 3104
    },
    {
      "epoch": 1.4492415402567094,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018986367512720343,
      "loss": 0.1119,
      "step": 3105
    },
    {
      "epoch": 1.449708284714119,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018985723699825716,
      "loss": 0.1256,
      "step": 3106
    },
    {
      "epoch": 1.4501750291715285,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018985079693458485,
      "loss": 0.163,
      "step": 3107
    },
    {
      "epoch": 1.450641773628938,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018984435493632517,
      "loss": 0.1152,
      "step": 3108
    },
    {
      "epoch": 1.4511085180863477,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001898379110036169,
      "loss": 0.1016,
      "step": 3109
    },
    {
      "epoch": 1.4515752625437572,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018983146513659868,
      "loss": 0.1613,
      "step": 3110
    },
    {
      "epoch": 1.4520420070011668,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018982501733540939,
      "loss": 0.1604,
      "step": 3111
    },
    {
      "epoch": 1.4525087514585764,
      "grad_norm": 1.0,
      "learning_rate": 0.0001898185676001878,
      "loss": 0.1646,
      "step": 3112
    },
    {
      "epoch": 1.452975495915986,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018981211593107274,
      "loss": 0.1519,
      "step": 3113
    },
    {
      "epoch": 1.4534422403733955,
      "grad_norm": 1.25,
      "learning_rate": 0.00018980566232820324,
      "loss": 0.1538,
      "step": 3114
    },
    {
      "epoch": 1.453908984830805,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018979920679171815,
      "loss": 0.1319,
      "step": 3115
    },
    {
      "epoch": 1.4543757292882147,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018979274932175654,
      "loss": 0.1692,
      "step": 3116
    },
    {
      "epoch": 1.4548424737456243,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001897862899184574,
      "loss": 0.1353,
      "step": 3117
    },
    {
      "epoch": 1.4553092182030338,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00018977982858195983,
      "loss": 0.1443,
      "step": 3118
    },
    {
      "epoch": 1.4557759626604434,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018977336531240287,
      "loss": 0.1473,
      "step": 3119
    },
    {
      "epoch": 1.456242707117853,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018976690010992582,
      "loss": 0.1397,
      "step": 3120
    },
    {
      "epoch": 1.4567094515752625,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018976043297466778,
      "loss": 0.1336,
      "step": 3121
    },
    {
      "epoch": 1.4571761960326721,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018975396390676807,
      "loss": 0.1897,
      "step": 3122
    },
    {
      "epoch": 1.4576429404900817,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018974749290636588,
      "loss": 0.1445,
      "step": 3123
    },
    {
      "epoch": 1.4581096849474913,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018974101997360063,
      "loss": 0.1276,
      "step": 3124
    },
    {
      "epoch": 1.4585764294049008,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018973454510861165,
      "loss": 0.1202,
      "step": 3125
    },
    {
      "epoch": 1.4590431738623104,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018972806831153832,
      "loss": 0.1314,
      "step": 3126
    },
    {
      "epoch": 1.45950991831972,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018972158958252013,
      "loss": 0.1111,
      "step": 3127
    },
    {
      "epoch": 1.4599766627771296,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018971510892169658,
      "loss": 0.2049,
      "step": 3128
    },
    {
      "epoch": 1.4604434072345391,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018970862632920721,
      "loss": 0.1286,
      "step": 3129
    },
    {
      "epoch": 1.4609101516919487,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018970214180519154,
      "loss": 0.1733,
      "step": 3130
    },
    {
      "epoch": 1.4613768961493583,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018969565534978928,
      "loss": 0.1491,
      "step": 3131
    },
    {
      "epoch": 1.4618436406067679,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018968916696314,
      "loss": 0.174,
      "step": 3132
    },
    {
      "epoch": 1.4623103850641774,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018968267664538346,
      "loss": 0.0946,
      "step": 3133
    },
    {
      "epoch": 1.462777129521587,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001896761843966594,
      "loss": 0.1489,
      "step": 3134
    },
    {
      "epoch": 1.4632438739789966,
      "grad_norm": 1.25,
      "learning_rate": 0.00018966969021710757,
      "loss": 0.1995,
      "step": 3135
    },
    {
      "epoch": 1.463710618436406,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018966319410686783,
      "loss": 0.1606,
      "step": 3136
    },
    {
      "epoch": 1.4641773628938157,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018965669606608006,
      "loss": 0.1621,
      "step": 3137
    },
    {
      "epoch": 1.464644107351225,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018965019609488412,
      "loss": 0.1626,
      "step": 3138
    },
    {
      "epoch": 1.4651108518086349,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018964369419341998,
      "loss": 0.1293,
      "step": 3139
    },
    {
      "epoch": 1.4655775962660442,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018963719036182766,
      "loss": 0.1151,
      "step": 3140
    },
    {
      "epoch": 1.466044340723454,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018963068460024716,
      "loss": 0.156,
      "step": 3141
    },
    {
      "epoch": 1.4665110851808634,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001896241769088186,
      "loss": 0.1482,
      "step": 3142
    },
    {
      "epoch": 1.4669778296382732,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018961766728768206,
      "loss": 0.1445,
      "step": 3143
    },
    {
      "epoch": 1.4674445740956825,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018961115573697771,
      "loss": 0.1386,
      "step": 3144
    },
    {
      "epoch": 1.4679113185530923,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018960464225684575,
      "loss": 0.1313,
      "step": 3145
    },
    {
      "epoch": 1.4683780630105017,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001895981268474264,
      "loss": 0.1217,
      "step": 3146
    },
    {
      "epoch": 1.4688448074679112,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018959160950886004,
      "loss": 0.1294,
      "step": 3147
    },
    {
      "epoch": 1.4693115519253208,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001895850902412869,
      "loss": 0.118,
      "step": 3148
    },
    {
      "epoch": 1.4697782963827304,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018957856904484733,
      "loss": 0.1218,
      "step": 3149
    },
    {
      "epoch": 1.47024504084014,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001895720459196818,
      "loss": 0.1905,
      "step": 3150
    },
    {
      "epoch": 1.4707117852975495,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018956552086593073,
      "loss": 0.1845,
      "step": 3151
    },
    {
      "epoch": 1.471178529754959,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018955899388373463,
      "loss": 0.147,
      "step": 3152
    },
    {
      "epoch": 1.4716452742123687,
      "grad_norm": 1.0625,
      "learning_rate": 0.000189552464973234,
      "loss": 0.1541,
      "step": 3153
    },
    {
      "epoch": 1.4721120186697783,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018954593413456948,
      "loss": 0.1369,
      "step": 3154
    },
    {
      "epoch": 1.4725787631271878,
      "grad_norm": 1.125,
      "learning_rate": 0.00018953940136788163,
      "loss": 0.1947,
      "step": 3155
    },
    {
      "epoch": 1.4730455075845974,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018953286667331112,
      "loss": 0.1823,
      "step": 3156
    },
    {
      "epoch": 1.473512252042007,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018952633005099866,
      "loss": 0.1745,
      "step": 3157
    },
    {
      "epoch": 1.4739789964994165,
      "grad_norm": 1.0,
      "learning_rate": 0.00018951979150108494,
      "loss": 0.1631,
      "step": 3158
    },
    {
      "epoch": 1.4744457409568261,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00018951325102371084,
      "loss": 0.1122,
      "step": 3159
    },
    {
      "epoch": 1.4749124854142357,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018950670861901713,
      "loss": 0.148,
      "step": 3160
    },
    {
      "epoch": 1.4753792298716453,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018950016428714462,
      "loss": 0.1399,
      "step": 3161
    },
    {
      "epoch": 1.4758459743290548,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001894936180282343,
      "loss": 0.1515,
      "step": 3162
    },
    {
      "epoch": 1.4763127187864644,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001894870698424271,
      "loss": 0.1228,
      "step": 3163
    },
    {
      "epoch": 1.476779463243874,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000189480519729864,
      "loss": 0.1121,
      "step": 3164
    },
    {
      "epoch": 1.4772462077012836,
      "grad_norm": 1.234375,
      "learning_rate": 0.00018947396769068603,
      "loss": 0.2029,
      "step": 3165
    },
    {
      "epoch": 1.4777129521586931,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018946741372503426,
      "loss": 0.1247,
      "step": 3166
    },
    {
      "epoch": 1.4781796966161027,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018946085783304976,
      "loss": 0.1842,
      "step": 3167
    },
    {
      "epoch": 1.4786464410735123,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001894543000148738,
      "loss": 0.1275,
      "step": 3168
    },
    {
      "epoch": 1.4791131855309219,
      "grad_norm": 1.0,
      "learning_rate": 0.00018944774027064747,
      "loss": 0.1679,
      "step": 3169
    },
    {
      "epoch": 1.4795799299883314,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018944117860051206,
      "loss": 0.2096,
      "step": 3170
    },
    {
      "epoch": 1.480046674445741,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018943461500460883,
      "loss": 0.1202,
      "step": 3171
    },
    {
      "epoch": 1.4805134189031506,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018942804948307912,
      "loss": 0.1437,
      "step": 3172
    },
    {
      "epoch": 1.4809801633605602,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018942148203606428,
      "loss": 0.133,
      "step": 3173
    },
    {
      "epoch": 1.4814469078179697,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018941491266370573,
      "loss": 0.1761,
      "step": 3174
    },
    {
      "epoch": 1.4819136522753793,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001894083413661449,
      "loss": 0.1547,
      "step": 3175
    },
    {
      "epoch": 1.4823803967327889,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018940176814352327,
      "loss": 0.1261,
      "step": 3176
    },
    {
      "epoch": 1.4828471411901984,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018939519299598237,
      "loss": 0.1571,
      "step": 3177
    },
    {
      "epoch": 1.483313885647608,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001893886159236638,
      "loss": 0.0881,
      "step": 3178
    },
    {
      "epoch": 1.4837806301050174,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001893820369267091,
      "loss": 0.1149,
      "step": 3179
    },
    {
      "epoch": 1.4842473745624272,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018937545600526004,
      "loss": 0.2139,
      "step": 3180
    },
    {
      "epoch": 1.4847141190198365,
      "grad_norm": 1.25,
      "learning_rate": 0.0001893688731594582,
      "loss": 0.1234,
      "step": 3181
    },
    {
      "epoch": 1.4851808634772463,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018936228838944537,
      "loss": 0.1424,
      "step": 3182
    },
    {
      "epoch": 1.4856476079346557,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018935570169536337,
      "loss": 0.1479,
      "step": 3183
    },
    {
      "epoch": 1.4861143523920655,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001893491130773539,
      "loss": 0.158,
      "step": 3184
    },
    {
      "epoch": 1.4865810968494748,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018934252253555896,
      "loss": 0.1294,
      "step": 3185
    },
    {
      "epoch": 1.4870478413068846,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018933593007012034,
      "loss": 0.1346,
      "step": 3186
    },
    {
      "epoch": 1.487514585764294,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018932933568118005,
      "loss": 0.108,
      "step": 3187
    },
    {
      "epoch": 1.4879813302217035,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018932273936888002,
      "loss": 0.1251,
      "step": 3188
    },
    {
      "epoch": 1.488448074679113,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018931614113336232,
      "loss": 0.1276,
      "step": 3189
    },
    {
      "epoch": 1.4889148191365227,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018930954097476902,
      "loss": 0.0895,
      "step": 3190
    },
    {
      "epoch": 1.4893815635939323,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001893029388932422,
      "loss": 0.16,
      "step": 3191
    },
    {
      "epoch": 1.4898483080513418,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018929633488892403,
      "loss": 0.1235,
      "step": 3192
    },
    {
      "epoch": 1.4903150525087514,
      "grad_norm": 1.453125,
      "learning_rate": 0.00018928972896195667,
      "loss": 0.1876,
      "step": 3193
    },
    {
      "epoch": 1.490781796966161,
      "grad_norm": 1.203125,
      "learning_rate": 0.0001892831211124824,
      "loss": 0.1804,
      "step": 3194
    },
    {
      "epoch": 1.4912485414235706,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001892765113406435,
      "loss": 0.1482,
      "step": 3195
    },
    {
      "epoch": 1.4917152858809801,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018926989964658216,
      "loss": 0.1183,
      "step": 3196
    },
    {
      "epoch": 1.4921820303383897,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018926328603044093,
      "loss": 0.1475,
      "step": 3197
    },
    {
      "epoch": 1.4926487747957993,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018925667049236208,
      "loss": 0.177,
      "step": 3198
    },
    {
      "epoch": 1.4931155192532088,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018925005303248806,
      "loss": 0.12,
      "step": 3199
    },
    {
      "epoch": 1.4935822637106184,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001892434336509614,
      "loss": 0.1514,
      "step": 3200
    },
    {
      "epoch": 1.494049008168028,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001892368123479246,
      "loss": 0.1076,
      "step": 3201
    },
    {
      "epoch": 1.4945157526254376,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001892301891235202,
      "loss": 0.1415,
      "step": 3202
    },
    {
      "epoch": 1.4949824970828471,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00018922356397789084,
      "loss": 0.1415,
      "step": 3203
    },
    {
      "epoch": 1.4954492415402567,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018921693691117914,
      "loss": 0.1238,
      "step": 3204
    },
    {
      "epoch": 1.4959159859976663,
      "grad_norm": 1.0,
      "learning_rate": 0.0001892103079235278,
      "loss": 0.1272,
      "step": 3205
    },
    {
      "epoch": 1.4963827304550759,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018920367701507956,
      "loss": 0.1685,
      "step": 3206
    },
    {
      "epoch": 1.4968494749124854,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018919704418597717,
      "loss": 0.1443,
      "step": 3207
    },
    {
      "epoch": 1.497316219369895,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018919040943636347,
      "loss": 0.108,
      "step": 3208
    },
    {
      "epoch": 1.4977829638273046,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018918377276638126,
      "loss": 0.1196,
      "step": 3209
    },
    {
      "epoch": 1.4982497082847142,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018917713417617352,
      "loss": 0.1534,
      "step": 3210
    },
    {
      "epoch": 1.4987164527421237,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001891704936658831,
      "loss": 0.1454,
      "step": 3211
    },
    {
      "epoch": 1.4991831971995333,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018916385123565304,
      "loss": 0.1383,
      "step": 3212
    },
    {
      "epoch": 1.4996499416569429,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018915720688562632,
      "loss": 0.1727,
      "step": 3213
    },
    {
      "epoch": 1.5001166861143522,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018915056061594602,
      "loss": 0.1441,
      "step": 3214
    },
    {
      "epoch": 1.500583430571762,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018914391242675523,
      "loss": 0.1295,
      "step": 3215
    },
    {
      "epoch": 1.5010501750291714,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018913726231819711,
      "loss": 0.1015,
      "step": 3216
    },
    {
      "epoch": 1.5010501750291714,
      "eval_loss": 1.3733340501785278,
      "eval_runtime": 93.5915,
      "eval_samples_per_second": 19.275,
      "eval_steps_per_second": 2.415,
      "step": 3216
    },
    {
      "epoch": 1.5015169194865812,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001891306102904148,
      "loss": 0.1458,
      "step": 3217
    },
    {
      "epoch": 1.5019836639439905,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001891239563435516,
      "loss": 0.1363,
      "step": 3218
    },
    {
      "epoch": 1.5024504084014003,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001891173004777507,
      "loss": 0.1664,
      "step": 3219
    },
    {
      "epoch": 1.5029171528588097,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001891106426931555,
      "loss": 0.1577,
      "step": 3220
    },
    {
      "epoch": 1.5033838973162195,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018910398298990923,
      "loss": 0.154,
      "step": 3221
    },
    {
      "epoch": 1.5038506417736288,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018909732136815536,
      "loss": 0.1448,
      "step": 3222
    },
    {
      "epoch": 1.5043173862310386,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018909065782803732,
      "loss": 0.1127,
      "step": 3223
    },
    {
      "epoch": 1.504784130688448,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001890839923696986,
      "loss": 0.1458,
      "step": 3224
    },
    {
      "epoch": 1.5052508751458578,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00018907732499328264,
      "loss": 0.1454,
      "step": 3225
    },
    {
      "epoch": 1.505717619603267,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00018907065569893303,
      "loss": 0.1018,
      "step": 3226
    },
    {
      "epoch": 1.506184364060677,
      "grad_norm": 0.765625,
      "learning_rate": 0.00018906398448679339,
      "loss": 0.0956,
      "step": 3227
    },
    {
      "epoch": 1.5066511085180863,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018905731135700736,
      "loss": 0.1371,
      "step": 3228
    },
    {
      "epoch": 1.507117852975496,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001890506363097186,
      "loss": 0.0939,
      "step": 3229
    },
    {
      "epoch": 1.5075845974329054,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018904395934507082,
      "loss": 0.1551,
      "step": 3230
    },
    {
      "epoch": 1.5080513418903152,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001890372804632078,
      "loss": 0.1197,
      "step": 3231
    },
    {
      "epoch": 1.5085180863477246,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018903059966427337,
      "loss": 0.1274,
      "step": 3232
    },
    {
      "epoch": 1.5089848308051343,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018902391694841132,
      "loss": 0.1136,
      "step": 3233
    },
    {
      "epoch": 1.5094515752625437,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001890172323157656,
      "loss": 0.1279,
      "step": 3234
    },
    {
      "epoch": 1.5099183197199533,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018901054576648006,
      "loss": 0.1235,
      "step": 3235
    },
    {
      "epoch": 1.5103850641773628,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018900385730069873,
      "loss": 0.1796,
      "step": 3236
    },
    {
      "epoch": 1.5108518086347724,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001889971669185656,
      "loss": 0.1142,
      "step": 3237
    },
    {
      "epoch": 1.511318553092182,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018899047462022473,
      "loss": 0.1602,
      "step": 3238
    },
    {
      "epoch": 1.5117852975495916,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001889837804058202,
      "loss": 0.1228,
      "step": 3239
    },
    {
      "epoch": 1.5122520420070011,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018897708427549614,
      "loss": 0.1259,
      "step": 3240
    },
    {
      "epoch": 1.5127187864644107,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00018897038622939675,
      "loss": 0.1411,
      "step": 3241
    },
    {
      "epoch": 1.5131855309218203,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001889636862676662,
      "loss": 0.1498,
      "step": 3242
    },
    {
      "epoch": 1.5136522753792299,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018895698439044882,
      "loss": 0.1294,
      "step": 3243
    },
    {
      "epoch": 1.5141190198366394,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018895028059788885,
      "loss": 0.1396,
      "step": 3244
    },
    {
      "epoch": 1.514585764294049,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018894357489013063,
      "loss": 0.1169,
      "step": 3245
    },
    {
      "epoch": 1.5150525087514586,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001889368672673186,
      "loss": 0.1623,
      "step": 3246
    },
    {
      "epoch": 1.5155192532088682,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001889301577295971,
      "loss": 0.1028,
      "step": 3247
    },
    {
      "epoch": 1.5159859976662777,
      "grad_norm": 1.125,
      "learning_rate": 0.00018892344627711064,
      "loss": 0.1329,
      "step": 3248
    },
    {
      "epoch": 1.5164527421236873,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001889167329100037,
      "loss": 0.1264,
      "step": 3249
    },
    {
      "epoch": 1.5169194865810969,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00018891001762842086,
      "loss": 0.119,
      "step": 3250
    },
    {
      "epoch": 1.5173862310385065,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001889033004325067,
      "loss": 0.1207,
      "step": 3251
    },
    {
      "epoch": 1.517852975495916,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018889658132240585,
      "loss": 0.1311,
      "step": 3252
    },
    {
      "epoch": 1.5183197199533256,
      "grad_norm": 1.03125,
      "learning_rate": 0.000188889860298263,
      "loss": 0.129,
      "step": 3253
    },
    {
      "epoch": 1.5187864644107352,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001888831373602228,
      "loss": 0.1369,
      "step": 3254
    },
    {
      "epoch": 1.5192532088681447,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018887641250843,
      "loss": 0.1071,
      "step": 3255
    },
    {
      "epoch": 1.5197199533255543,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018886968574302948,
      "loss": 0.1715,
      "step": 3256
    },
    {
      "epoch": 1.5201866977829637,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018886295706416602,
      "loss": 0.1573,
      "step": 3257
    },
    {
      "epoch": 1.5206534422403735,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018885622647198447,
      "loss": 0.1125,
      "step": 3258
    },
    {
      "epoch": 1.5211201866977828,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018884949396662977,
      "loss": 0.1202,
      "step": 3259
    },
    {
      "epoch": 1.5215869311551926,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00018884275954824694,
      "loss": 0.1054,
      "step": 3260
    },
    {
      "epoch": 1.522053675612602,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001888360232169809,
      "loss": 0.1554,
      "step": 3261
    },
    {
      "epoch": 1.5225204200700118,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001888292849729767,
      "loss": 0.1216,
      "step": 3262
    },
    {
      "epoch": 1.5229871645274211,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018882254481637943,
      "loss": 0.1349,
      "step": 3263
    },
    {
      "epoch": 1.523453908984831,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001888158027473342,
      "loss": 0.0976,
      "step": 3264
    },
    {
      "epoch": 1.5239206534422403,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018880905876598626,
      "loss": 0.1516,
      "step": 3265
    },
    {
      "epoch": 1.52438739789965,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001888023128724807,
      "loss": 0.1504,
      "step": 3266
    },
    {
      "epoch": 1.5248541423570594,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018879556506696282,
      "loss": 0.1137,
      "step": 3267
    },
    {
      "epoch": 1.5253208868144692,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018878881534957793,
      "loss": 0.1903,
      "step": 3268
    },
    {
      "epoch": 1.5257876312718786,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001887820637204713,
      "loss": 0.134,
      "step": 3269
    },
    {
      "epoch": 1.5262543757292883,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018877531017978834,
      "loss": 0.1233,
      "step": 3270
    },
    {
      "epoch": 1.5267211201866977,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001887685547276745,
      "loss": 0.1489,
      "step": 3271
    },
    {
      "epoch": 1.5271878646441075,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018876179736427516,
      "loss": 0.1175,
      "step": 3272
    },
    {
      "epoch": 1.5276546091015168,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018875503808973582,
      "loss": 0.1036,
      "step": 3273
    },
    {
      "epoch": 1.5281213535589266,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018874827690420202,
      "loss": 0.1046,
      "step": 3274
    },
    {
      "epoch": 1.528588098016336,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001887415138078194,
      "loss": 0.1361,
      "step": 3275
    },
    {
      "epoch": 1.5290548424737458,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001887347488007335,
      "loss": 0.1369,
      "step": 3276
    },
    {
      "epoch": 1.5295215869311551,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018872798188309,
      "loss": 0.129,
      "step": 3277
    },
    {
      "epoch": 1.5299883313885647,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018872121305503462,
      "loss": 0.1371,
      "step": 3278
    },
    {
      "epoch": 1.5304550758459743,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018871444231671308,
      "loss": 0.1104,
      "step": 3279
    },
    {
      "epoch": 1.5309218203033839,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018870766966827115,
      "loss": 0.1004,
      "step": 3280
    },
    {
      "epoch": 1.5313885647607934,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018870089510985468,
      "loss": 0.1918,
      "step": 3281
    },
    {
      "epoch": 1.531855309218203,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018869411864160952,
      "loss": 0.1191,
      "step": 3282
    },
    {
      "epoch": 1.5323220536756126,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001886873402636816,
      "loss": 0.145,
      "step": 3283
    },
    {
      "epoch": 1.5327887981330222,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018868055997621683,
      "loss": 0.1245,
      "step": 3284
    },
    {
      "epoch": 1.5332555425904317,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001886737777793612,
      "loss": 0.1225,
      "step": 3285
    },
    {
      "epoch": 1.5337222870478413,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018866699367326074,
      "loss": 0.1479,
      "step": 3286
    },
    {
      "epoch": 1.5341890315052509,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018866020765806156,
      "loss": 0.1415,
      "step": 3287
    },
    {
      "epoch": 1.5346557759626605,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018865341973390974,
      "loss": 0.1354,
      "step": 3288
    },
    {
      "epoch": 1.53512252042007,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001886466299009514,
      "loss": 0.1438,
      "step": 3289
    },
    {
      "epoch": 1.5355892648774796,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018863983815933279,
      "loss": 0.1533,
      "step": 3290
    },
    {
      "epoch": 1.5360560093348892,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018863304450920013,
      "loss": 0.1384,
      "step": 3291
    },
    {
      "epoch": 1.5365227537922987,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018862624895069963,
      "loss": 0.1299,
      "step": 3292
    },
    {
      "epoch": 1.5369894982497083,
      "grad_norm": 1.0,
      "learning_rate": 0.00018861945148397768,
      "loss": 0.1345,
      "step": 3293
    },
    {
      "epoch": 1.537456242707118,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018861265210918062,
      "loss": 0.1423,
      "step": 3294
    },
    {
      "epoch": 1.5379229871645275,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00018860585082645486,
      "loss": 0.1002,
      "step": 3295
    },
    {
      "epoch": 1.538389731621937,
      "grad_norm": 1.1484375,
      "learning_rate": 0.0001885990476359468,
      "loss": 0.1648,
      "step": 3296
    },
    {
      "epoch": 1.5388564760793466,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018859224253780292,
      "loss": 0.1421,
      "step": 3297
    },
    {
      "epoch": 1.539323220536756,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001885854355321698,
      "loss": 0.1265,
      "step": 3298
    },
    {
      "epoch": 1.5397899649941658,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018857862661919396,
      "loss": 0.2223,
      "step": 3299
    },
    {
      "epoch": 1.5402567094515751,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018857181579902199,
      "loss": 0.1319,
      "step": 3300
    },
    {
      "epoch": 1.540723453908985,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001885650030718006,
      "loss": 0.1148,
      "step": 3301
    },
    {
      "epoch": 1.5411901983663943,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001885581884376764,
      "loss": 0.1427,
      "step": 3302
    },
    {
      "epoch": 1.541656942823804,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018855137189679612,
      "loss": 0.1124,
      "step": 3303
    },
    {
      "epoch": 1.5421236872812134,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018854455344930658,
      "loss": 0.1084,
      "step": 3304
    },
    {
      "epoch": 1.5425904317386232,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018853773309535456,
      "loss": 0.133,
      "step": 3305
    },
    {
      "epoch": 1.5430571761960326,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018853091083508692,
      "loss": 0.1619,
      "step": 3306
    },
    {
      "epoch": 1.5435239206534423,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018852408666865053,
      "loss": 0.1362,
      "step": 3307
    },
    {
      "epoch": 1.5439906651108517,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018851726059619233,
      "loss": 0.1626,
      "step": 3308
    },
    {
      "epoch": 1.5444574095682615,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001885104326178593,
      "loss": 0.0871,
      "step": 3309
    },
    {
      "epoch": 1.5449241540256708,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018850360273379847,
      "loss": 0.1492,
      "step": 3310
    },
    {
      "epoch": 1.5453908984830806,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018849677094415687,
      "loss": 0.1047,
      "step": 3311
    },
    {
      "epoch": 1.54585764294049,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018848993724908159,
      "loss": 0.1165,
      "step": 3312
    },
    {
      "epoch": 1.5463243873978998,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001884831016487198,
      "loss": 0.1252,
      "step": 3313
    },
    {
      "epoch": 1.5467911318553091,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018847626414321864,
      "loss": 0.1311,
      "step": 3314
    },
    {
      "epoch": 1.547257876312719,
      "grad_norm": 0.828125,
      "learning_rate": 0.00018846942473272536,
      "loss": 0.1077,
      "step": 3315
    },
    {
      "epoch": 1.5477246207701283,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001884625834173872,
      "loss": 0.0914,
      "step": 3316
    },
    {
      "epoch": 1.548191365227538,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018845574019735146,
      "loss": 0.1385,
      "step": 3317
    },
    {
      "epoch": 1.5486581096849474,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018844889507276552,
      "loss": 0.0971,
      "step": 3318
    },
    {
      "epoch": 1.549124854142357,
      "grad_norm": 1.125,
      "learning_rate": 0.0001884420480437767,
      "loss": 0.1524,
      "step": 3319
    },
    {
      "epoch": 1.5495915985997666,
      "grad_norm": 0.796875,
      "learning_rate": 0.00018843519911053246,
      "loss": 0.1031,
      "step": 3320
    },
    {
      "epoch": 1.5500583430571762,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018842834827318027,
      "loss": 0.116,
      "step": 3321
    },
    {
      "epoch": 1.5505250875145857,
      "grad_norm": 1.34375,
      "learning_rate": 0.00018842149553186763,
      "loss": 0.1601,
      "step": 3322
    },
    {
      "epoch": 1.5509918319719953,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001884146408867421,
      "loss": 0.1146,
      "step": 3323
    },
    {
      "epoch": 1.5514585764294049,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00018840778433795122,
      "loss": 0.1566,
      "step": 3324
    },
    {
      "epoch": 1.5519253208868145,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018840092588564268,
      "loss": 0.1393,
      "step": 3325
    },
    {
      "epoch": 1.552392065344224,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018839406552996412,
      "loss": 0.1206,
      "step": 3326
    },
    {
      "epoch": 1.5528588098016336,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018838720327106324,
      "loss": 0.1275,
      "step": 3327
    },
    {
      "epoch": 1.5533255542590432,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018838033910908782,
      "loss": 0.141,
      "step": 3328
    },
    {
      "epoch": 1.5537922987164527,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018837347304418564,
      "loss": 0.1265,
      "step": 3329
    },
    {
      "epoch": 1.5542590431738623,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018836660507650453,
      "loss": 0.118,
      "step": 3330
    },
    {
      "epoch": 1.554725787631272,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018835973520619237,
      "loss": 0.1141,
      "step": 3331
    },
    {
      "epoch": 1.5551925320886815,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018835286343339707,
      "loss": 0.173,
      "step": 3332
    },
    {
      "epoch": 1.555659276546091,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00018834598975826658,
      "loss": 0.1402,
      "step": 3333
    },
    {
      "epoch": 1.5561260210035006,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018833911418094892,
      "loss": 0.144,
      "step": 3334
    },
    {
      "epoch": 1.5565927654609102,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018833223670159212,
      "loss": 0.1109,
      "step": 3335
    },
    {
      "epoch": 1.5570595099183198,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018832535732034428,
      "loss": 0.1168,
      "step": 3336
    },
    {
      "epoch": 1.5575262543757293,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018831847603735345,
      "loss": 0.1735,
      "step": 3337
    },
    {
      "epoch": 1.557992998833139,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018831159285276786,
      "loss": 0.1464,
      "step": 3338
    },
    {
      "epoch": 1.5584597432905483,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018830470776673569,
      "loss": 0.1243,
      "step": 3339
    },
    {
      "epoch": 1.558926487747958,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018829782077940522,
      "loss": 0.1271,
      "step": 3340
    },
    {
      "epoch": 1.5593932322053674,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018829093189092467,
      "loss": 0.1185,
      "step": 3341
    },
    {
      "epoch": 1.5598599766627772,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001882840411014424,
      "loss": 0.171,
      "step": 3342
    },
    {
      "epoch": 1.5603267211201866,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018827714841110677,
      "loss": 0.1358,
      "step": 3343
    },
    {
      "epoch": 1.5607934655775964,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00018827025382006621,
      "loss": 0.1083,
      "step": 3344
    },
    {
      "epoch": 1.5612602100350057,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018826335732846912,
      "loss": 0.1927,
      "step": 3345
    },
    {
      "epoch": 1.5617269544924155,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000188256458936464,
      "loss": 0.1223,
      "step": 3346
    },
    {
      "epoch": 1.5621936989498248,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018824955864419943,
      "loss": 0.1362,
      "step": 3347
    },
    {
      "epoch": 1.5626604434072346,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018824265645182395,
      "loss": 0.1321,
      "step": 3348
    },
    {
      "epoch": 1.563127187864644,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018823575235948616,
      "loss": 0.0993,
      "step": 3349
    },
    {
      "epoch": 1.5635939323220538,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018822884636733472,
      "loss": 0.1136,
      "step": 3350
    },
    {
      "epoch": 1.5640606767794631,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018822193847551835,
      "loss": 0.1354,
      "step": 3351
    },
    {
      "epoch": 1.564527421236873,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018821502868418574,
      "loss": 0.1226,
      "step": 3352
    },
    {
      "epoch": 1.5649941656942823,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018820811699348568,
      "loss": 0.1318,
      "step": 3353
    },
    {
      "epoch": 1.565460910151692,
      "grad_norm": 0.953125,
      "learning_rate": 0.000188201203403567,
      "loss": 0.1255,
      "step": 3354
    },
    {
      "epoch": 1.5659276546091014,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018819428791457853,
      "loss": 0.1237,
      "step": 3355
    },
    {
      "epoch": 1.5663943990665112,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001881873705266692,
      "loss": 0.1004,
      "step": 3356
    },
    {
      "epoch": 1.5668611435239206,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018818045123998795,
      "loss": 0.121,
      "step": 3357
    },
    {
      "epoch": 1.5673278879813304,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018817353005468374,
      "loss": 0.1035,
      "step": 3358
    },
    {
      "epoch": 1.5677946324387397,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001881666069709056,
      "loss": 0.1324,
      "step": 3359
    },
    {
      "epoch": 1.5682613768961493,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001881596819888026,
      "loss": 0.1788,
      "step": 3360
    },
    {
      "epoch": 1.5687281213535589,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001881527551085238,
      "loss": 0.1213,
      "step": 3361
    },
    {
      "epoch": 1.5691948658109685,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001881458263302184,
      "loss": 0.1127,
      "step": 3362
    },
    {
      "epoch": 1.569661610268378,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018813889565403558,
      "loss": 0.1785,
      "step": 3363
    },
    {
      "epoch": 1.5701283547257876,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001881319630801245,
      "loss": 0.0996,
      "step": 3364
    },
    {
      "epoch": 1.5705950991831972,
      "grad_norm": 1.0,
      "learning_rate": 0.0001881250286086345,
      "loss": 0.1759,
      "step": 3365
    },
    {
      "epoch": 1.5710618436406067,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018811809223971484,
      "loss": 0.1573,
      "step": 3366
    },
    {
      "epoch": 1.5715285880980163,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018811115397351488,
      "loss": 0.1181,
      "step": 3367
    },
    {
      "epoch": 1.571995332555426,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018810421381018402,
      "loss": 0.1182,
      "step": 3368
    },
    {
      "epoch": 1.5724620770128355,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001880972717498717,
      "loss": 0.1302,
      "step": 3369
    },
    {
      "epoch": 1.572928821470245,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018809032779272737,
      "loss": 0.1246,
      "step": 3370
    },
    {
      "epoch": 1.5733955659276546,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001880833819389005,
      "loss": 0.1094,
      "step": 3371
    },
    {
      "epoch": 1.5738623103850642,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00018807643418854077,
      "loss": 0.0892,
      "step": 3372
    },
    {
      "epoch": 1.5743290548424738,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018806948454179763,
      "loss": 0.1304,
      "step": 3373
    },
    {
      "epoch": 1.5747957992998833,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001880625329988208,
      "loss": 0.1277,
      "step": 3374
    },
    {
      "epoch": 1.575262543757293,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00018805557955975993,
      "loss": 0.0976,
      "step": 3375
    },
    {
      "epoch": 1.5757292882147025,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001880486242247647,
      "loss": 0.1364,
      "step": 3376
    },
    {
      "epoch": 1.576196032672112,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001880416669939849,
      "loss": 0.114,
      "step": 3377
    },
    {
      "epoch": 1.5766627771295216,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018803470786757034,
      "loss": 0.1136,
      "step": 3378
    },
    {
      "epoch": 1.5771295215869312,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018802774684567087,
      "loss": 0.1024,
      "step": 3379
    },
    {
      "epoch": 1.5775962660443408,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001880207839284363,
      "loss": 0.1124,
      "step": 3380
    },
    {
      "epoch": 1.5780630105017504,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001880138191160166,
      "loss": 0.1393,
      "step": 3381
    },
    {
      "epoch": 1.5785297549591597,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018800685240856173,
      "loss": 0.1104,
      "step": 3382
    },
    {
      "epoch": 1.5789964994165695,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018799988380622168,
      "loss": 0.1332,
      "step": 3383
    },
    {
      "epoch": 1.5794632438739789,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018799291330914652,
      "loss": 0.1461,
      "step": 3384
    },
    {
      "epoch": 1.5799299883313886,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018798594091748628,
      "loss": 0.1598,
      "step": 3385
    },
    {
      "epoch": 1.580396732788798,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001879789666313911,
      "loss": 0.1022,
      "step": 3386
    },
    {
      "epoch": 1.5808634772462078,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018797199045101116,
      "loss": 0.1063,
      "step": 3387
    },
    {
      "epoch": 1.5813302217036171,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001879650123764967,
      "loss": 0.1121,
      "step": 3388
    },
    {
      "epoch": 1.581796966161027,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018795803240799787,
      "loss": 0.1178,
      "step": 3389
    },
    {
      "epoch": 1.5822637106184363,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018795105054566501,
      "loss": 0.0857,
      "step": 3390
    },
    {
      "epoch": 1.582730455075846,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018794406678964852,
      "loss": 0.1119,
      "step": 3391
    },
    {
      "epoch": 1.5831971995332554,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018793708114009863,
      "loss": 0.137,
      "step": 3392
    },
    {
      "epoch": 1.5836639439906652,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018793009359716582,
      "loss": 0.1298,
      "step": 3393
    },
    {
      "epoch": 1.5841306884480746,
      "grad_norm": 1.0,
      "learning_rate": 0.00018792310416100057,
      "loss": 0.1173,
      "step": 3394
    },
    {
      "epoch": 1.5845974329054844,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001879161128317533,
      "loss": 0.1103,
      "step": 3395
    },
    {
      "epoch": 1.5850641773628937,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018790911960957457,
      "loss": 0.1438,
      "step": 3396
    },
    {
      "epoch": 1.5855309218203035,
      "grad_norm": 1.171875,
      "learning_rate": 0.000187902124494615,
      "loss": 0.1464,
      "step": 3397
    },
    {
      "epoch": 1.5859976662777129,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018789512748702513,
      "loss": 0.1318,
      "step": 3398
    },
    {
      "epoch": 1.5864644107351227,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018788812858695563,
      "loss": 0.1228,
      "step": 3399
    },
    {
      "epoch": 1.586931155192532,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018788112779455726,
      "loss": 0.1111,
      "step": 3400
    },
    {
      "epoch": 1.5873978996499416,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018787412510998065,
      "loss": 0.1652,
      "step": 3401
    },
    {
      "epoch": 1.5878646441073512,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018786712053337664,
      "loss": 0.1257,
      "step": 3402
    },
    {
      "epoch": 1.5883313885647607,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018786011406489604,
      "loss": 0.1074,
      "step": 3403
    },
    {
      "epoch": 1.5887981330221703,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018785310570468972,
      "loss": 0.132,
      "step": 3404
    },
    {
      "epoch": 1.58926487747958,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018784609545290855,
      "loss": 0.1805,
      "step": 3405
    },
    {
      "epoch": 1.5897316219369895,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018783908330970346,
      "loss": 0.1233,
      "step": 3406
    },
    {
      "epoch": 1.590198366394399,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018783206927522548,
      "loss": 0.1518,
      "step": 3407
    },
    {
      "epoch": 1.5906651108518086,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018782505334962558,
      "loss": 0.1431,
      "step": 3408
    },
    {
      "epoch": 1.5911318553092182,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018781803553305482,
      "loss": 0.1385,
      "step": 3409
    },
    {
      "epoch": 1.5915985997666278,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018781101582566434,
      "loss": 0.1273,
      "step": 3410
    },
    {
      "epoch": 1.5920653442240373,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00018780399422760525,
      "loss": 0.1133,
      "step": 3411
    },
    {
      "epoch": 1.592532088681447,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018779697073902877,
      "loss": 0.1172,
      "step": 3412
    },
    {
      "epoch": 1.5929988331388565,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00018778994536008609,
      "loss": 0.0939,
      "step": 3413
    },
    {
      "epoch": 1.593465577596266,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018778291809092842,
      "loss": 0.0979,
      "step": 3414
    },
    {
      "epoch": 1.5939323220536756,
      "grad_norm": 1.296875,
      "learning_rate": 0.00018777588893170718,
      "loss": 0.1596,
      "step": 3415
    },
    {
      "epoch": 1.5943990665110852,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018776885788257368,
      "loss": 0.1494,
      "step": 3416
    },
    {
      "epoch": 1.5948658109684948,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018776182494367928,
      "loss": 0.142,
      "step": 3417
    },
    {
      "epoch": 1.5953325554259044,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018775479011517537,
      "loss": 0.1347,
      "step": 3418
    },
    {
      "epoch": 1.595799299883314,
      "grad_norm": 0.875,
      "learning_rate": 0.0001877477533972135,
      "loss": 0.1227,
      "step": 3419
    },
    {
      "epoch": 1.5962660443407235,
      "grad_norm": 1.296875,
      "learning_rate": 0.00018774071478994513,
      "loss": 0.1524,
      "step": 3420
    },
    {
      "epoch": 1.596732788798133,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018773367429352186,
      "loss": 0.1503,
      "step": 3421
    },
    {
      "epoch": 1.5971995332555426,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001877266319080952,
      "loss": 0.1236,
      "step": 3422
    },
    {
      "epoch": 1.597666277712952,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018771958763381683,
      "loss": 0.1544,
      "step": 3423
    },
    {
      "epoch": 1.5981330221703618,
      "grad_norm": 1.125,
      "learning_rate": 0.00018771254147083842,
      "loss": 0.1376,
      "step": 3424
    },
    {
      "epoch": 1.5985997666277711,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018770549341931165,
      "loss": 0.1075,
      "step": 3425
    },
    {
      "epoch": 1.599066511085181,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001876984434793883,
      "loss": 0.1232,
      "step": 3426
    },
    {
      "epoch": 1.5995332555425903,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00018769139165122017,
      "loss": 0.1288,
      "step": 3427
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018768433793495907,
      "loss": 0.1331,
      "step": 3428
    },
    {
      "epoch": 1.6004667444574094,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001876772823307569,
      "loss": 0.0756,
      "step": 3429
    },
    {
      "epoch": 1.6009334889148192,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018767022483876558,
      "loss": 0.1273,
      "step": 3430
    },
    {
      "epoch": 1.6014002333722286,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018766316545913703,
      "loss": 0.1452,
      "step": 3431
    },
    {
      "epoch": 1.6018669778296384,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018765610419202325,
      "loss": 0.1117,
      "step": 3432
    },
    {
      "epoch": 1.6023337222870477,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018764904103757627,
      "loss": 0.1139,
      "step": 3433
    },
    {
      "epoch": 1.6028004667444575,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018764197599594823,
      "loss": 0.1248,
      "step": 3434
    },
    {
      "epoch": 1.6032672112018669,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018763490906729122,
      "loss": 0.1329,
      "step": 3435
    },
    {
      "epoch": 1.6037339556592767,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018762784025175736,
      "loss": 0.1108,
      "step": 3436
    },
    {
      "epoch": 1.604200700116686,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018762076954949888,
      "loss": 0.1271,
      "step": 3437
    },
    {
      "epoch": 1.6046674445740958,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018761369696066797,
      "loss": 0.09,
      "step": 3438
    },
    {
      "epoch": 1.6051341890315052,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018760662248541702,
      "loss": 0.1286,
      "step": 3439
    },
    {
      "epoch": 1.605600933488915,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018759954612389824,
      "loss": 0.1328,
      "step": 3440
    },
    {
      "epoch": 1.6060676779463243,
      "grad_norm": 1.0,
      "learning_rate": 0.0001875924678762641,
      "loss": 0.1163,
      "step": 3441
    },
    {
      "epoch": 1.6065344224037341,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018758538774266687,
      "loss": 0.127,
      "step": 3442
    },
    {
      "epoch": 1.6070011668611435,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018757830572325909,
      "loss": 0.118,
      "step": 3443
    },
    {
      "epoch": 1.607467911318553,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018757122181819325,
      "loss": 0.1063,
      "step": 3444
    },
    {
      "epoch": 1.6079346557759626,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018756413602762182,
      "loss": 0.1236,
      "step": 3445
    },
    {
      "epoch": 1.6084014002333722,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018755704835169738,
      "loss": 0.131,
      "step": 3446
    },
    {
      "epoch": 1.6088681446907818,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018754995879057253,
      "loss": 0.1402,
      "step": 3447
    },
    {
      "epoch": 1.6093348891481913,
      "grad_norm": 1.25,
      "learning_rate": 0.00018754286734439996,
      "loss": 0.1812,
      "step": 3448
    },
    {
      "epoch": 1.609801633605601,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018753577401333228,
      "loss": 0.1159,
      "step": 3449
    },
    {
      "epoch": 1.6102683780630105,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018752867879752233,
      "loss": 0.1026,
      "step": 3450
    },
    {
      "epoch": 1.61073512252042,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018752158169712274,
      "loss": 0.1228,
      "step": 3451
    },
    {
      "epoch": 1.6112018669778296,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018751448271228644,
      "loss": 0.118,
      "step": 3452
    },
    {
      "epoch": 1.6116686114352392,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001875073818431662,
      "loss": 0.1111,
      "step": 3453
    },
    {
      "epoch": 1.6121353558926488,
      "grad_norm": 1.0,
      "learning_rate": 0.00018750027908991497,
      "loss": 0.121,
      "step": 3454
    },
    {
      "epoch": 1.6126021003500584,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018749317445268562,
      "loss": 0.0994,
      "step": 3455
    },
    {
      "epoch": 1.613068844807468,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018748606793163116,
      "loss": 0.1224,
      "step": 3456
    },
    {
      "epoch": 1.6135355892648775,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018747895952690458,
      "loss": 0.089,
      "step": 3457
    },
    {
      "epoch": 1.614002333722287,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018747184923865895,
      "loss": 0.1202,
      "step": 3458
    },
    {
      "epoch": 1.6144690781796966,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018746473706704735,
      "loss": 0.1146,
      "step": 3459
    },
    {
      "epoch": 1.6149358226371062,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018745762301222292,
      "loss": 0.13,
      "step": 3460
    },
    {
      "epoch": 1.6154025670945158,
      "grad_norm": 0.875,
      "learning_rate": 0.00018745050707433882,
      "loss": 0.1004,
      "step": 3461
    },
    {
      "epoch": 1.6158693115519254,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018744338925354829,
      "loss": 0.1203,
      "step": 3462
    },
    {
      "epoch": 1.616336056009335,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018743626955000458,
      "loss": 0.1421,
      "step": 3463
    },
    {
      "epoch": 1.6168028004667443,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018742914796386094,
      "loss": 0.1404,
      "step": 3464
    },
    {
      "epoch": 1.617269544924154,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018742202449527077,
      "loss": 0.103,
      "step": 3465
    },
    {
      "epoch": 1.6177362893815634,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001874148991443874,
      "loss": 0.1081,
      "step": 3466
    },
    {
      "epoch": 1.6182030338389732,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018740777191136426,
      "loss": 0.1117,
      "step": 3467
    },
    {
      "epoch": 1.6186697782963826,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018740064279635483,
      "loss": 0.1245,
      "step": 3468
    },
    {
      "epoch": 1.6191365227537924,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018739351179951257,
      "loss": 0.1549,
      "step": 3469
    },
    {
      "epoch": 1.6196032672112017,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018738637892099104,
      "loss": 0.1213,
      "step": 3470
    },
    {
      "epoch": 1.6200700116686115,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018737924416094382,
      "loss": 0.1481,
      "step": 3471
    },
    {
      "epoch": 1.6205367561260209,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018737210751952453,
      "loss": 0.1079,
      "step": 3472
    },
    {
      "epoch": 1.6210035005834307,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001873649689968868,
      "loss": 0.0891,
      "step": 3473
    },
    {
      "epoch": 1.62147024504084,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018735782859318438,
      "loss": 0.1318,
      "step": 3474
    },
    {
      "epoch": 1.6219369894982498,
      "grad_norm": 1.0,
      "learning_rate": 0.00018735068630857096,
      "loss": 0.13,
      "step": 3475
    },
    {
      "epoch": 1.6224037339556592,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018734354214320037,
      "loss": 0.1157,
      "step": 3476
    },
    {
      "epoch": 1.622870478413069,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018733639609722639,
      "loss": 0.1035,
      "step": 3477
    },
    {
      "epoch": 1.6233372228704783,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001873292481708029,
      "loss": 0.0999,
      "step": 3478
    },
    {
      "epoch": 1.6238039673278881,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001873220983640838,
      "loss": 0.1551,
      "step": 3479
    },
    {
      "epoch": 1.6242707117852975,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018731494667722308,
      "loss": 0.1131,
      "step": 3480
    },
    {
      "epoch": 1.6247374562427073,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018730779311037466,
      "loss": 0.1192,
      "step": 3481
    },
    {
      "epoch": 1.6252042007001166,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018730063766369255,
      "loss": 0.1023,
      "step": 3482
    },
    {
      "epoch": 1.6256709451575264,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018729348033733086,
      "loss": 0.0854,
      "step": 3483
    },
    {
      "epoch": 1.6261376896149358,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018728632113144373,
      "loss": 0.1187,
      "step": 3484
    },
    {
      "epoch": 1.6266044340723453,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018727916004618523,
      "loss": 0.0961,
      "step": 3485
    },
    {
      "epoch": 1.627071178529755,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018727199708170957,
      "loss": 0.093,
      "step": 3486
    },
    {
      "epoch": 1.6275379229871645,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018726483223817098,
      "loss": 0.1379,
      "step": 3487
    },
    {
      "epoch": 1.628004667444574,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018725766551572375,
      "loss": 0.0947,
      "step": 3488
    },
    {
      "epoch": 1.6284714119019836,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018725049691452214,
      "loss": 0.1178,
      "step": 3489
    },
    {
      "epoch": 1.6289381563593932,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018724332643472056,
      "loss": 0.1235,
      "step": 3490
    },
    {
      "epoch": 1.6294049008168028,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018723615407647334,
      "loss": 0.1087,
      "step": 3491
    },
    {
      "epoch": 1.6298716452742124,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018722897983993491,
      "loss": 0.1111,
      "step": 3492
    },
    {
      "epoch": 1.630338389731622,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018722180372525982,
      "loss": 0.0701,
      "step": 3493
    },
    {
      "epoch": 1.6308051341890315,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018721462573260248,
      "loss": 0.1055,
      "step": 3494
    },
    {
      "epoch": 1.631271878646441,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001872074458621175,
      "loss": 0.1324,
      "step": 3495
    },
    {
      "epoch": 1.6317386231038506,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018720026411395946,
      "loss": 0.1787,
      "step": 3496
    },
    {
      "epoch": 1.6322053675612602,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018719308048828294,
      "loss": 0.099,
      "step": 3497
    },
    {
      "epoch": 1.6326721120186698,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001871858949852427,
      "loss": 0.1315,
      "step": 3498
    },
    {
      "epoch": 1.6331388564760794,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018717870760499337,
      "loss": 0.0942,
      "step": 3499
    },
    {
      "epoch": 1.633605600933489,
      "grad_norm": 0.8125,
      "learning_rate": 0.00018717151834768978,
      "loss": 0.1061,
      "step": 3500
    },
    {
      "epoch": 1.6340723453908985,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018716432721348664,
      "loss": 0.1113,
      "step": 3501
    },
    {
      "epoch": 1.634539089848308,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018715713420253884,
      "loss": 0.1471,
      "step": 3502
    },
    {
      "epoch": 1.6350058343057177,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018714993931500128,
      "loss": 0.1036,
      "step": 3503
    },
    {
      "epoch": 1.6354725787631272,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018714274255102878,
      "loss": 0.1013,
      "step": 3504
    },
    {
      "epoch": 1.6359393232205366,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018713554391077634,
      "loss": 0.1105,
      "step": 3505
    },
    {
      "epoch": 1.6364060676779464,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018712834339439896,
      "loss": 0.1498,
      "step": 3506
    },
    {
      "epoch": 1.6368728121353557,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018712114100205175,
      "loss": 0.1235,
      "step": 3507
    },
    {
      "epoch": 1.6373395565927655,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018711393673388963,
      "loss": 0.0877,
      "step": 3508
    },
    {
      "epoch": 1.6378063010501749,
      "grad_norm": 0.828125,
      "learning_rate": 0.00018710673059006783,
      "loss": 0.0917,
      "step": 3509
    },
    {
      "epoch": 1.6382730455075847,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018709952257074149,
      "loss": 0.1262,
      "step": 3510
    },
    {
      "epoch": 1.638739789964994,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018709231267606576,
      "loss": 0.1259,
      "step": 3511
    },
    {
      "epoch": 1.6392065344224038,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018708510090619592,
      "loss": 0.1114,
      "step": 3512
    },
    {
      "epoch": 1.6396732788798132,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018707788726128724,
      "loss": 0.1255,
      "step": 3513
    },
    {
      "epoch": 1.640140023337223,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018707067174149498,
      "loss": 0.1133,
      "step": 3514
    },
    {
      "epoch": 1.6406067677946323,
      "grad_norm": 1.3125,
      "learning_rate": 0.0001870634543469746,
      "loss": 0.0997,
      "step": 3515
    },
    {
      "epoch": 1.6410735122520421,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018705623507788146,
      "loss": 0.1056,
      "step": 3516
    },
    {
      "epoch": 1.6415402567094515,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018704901393437097,
      "loss": 0.102,
      "step": 3517
    },
    {
      "epoch": 1.6420070011668613,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018704179091659864,
      "loss": 0.126,
      "step": 3518
    },
    {
      "epoch": 1.6424737456242706,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018703456602471997,
      "loss": 0.1861,
      "step": 3519
    },
    {
      "epoch": 1.6429404900816804,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018702733925889052,
      "loss": 0.152,
      "step": 3520
    },
    {
      "epoch": 1.6434072345390898,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001870201106192659,
      "loss": 0.1251,
      "step": 3521
    },
    {
      "epoch": 1.6438739789964996,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018701288010600177,
      "loss": 0.1242,
      "step": 3522
    },
    {
      "epoch": 1.644340723453909,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018700564771925376,
      "loss": 0.0818,
      "step": 3523
    },
    {
      "epoch": 1.6448074679113187,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018699841345917764,
      "loss": 0.1194,
      "step": 3524
    },
    {
      "epoch": 1.645274212368728,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018699117732592915,
      "loss": 0.0872,
      "step": 3525
    },
    {
      "epoch": 1.6457409568261376,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018698393931966407,
      "loss": 0.1181,
      "step": 3526
    },
    {
      "epoch": 1.6462077012835472,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018697669944053828,
      "loss": 0.1007,
      "step": 3527
    },
    {
      "epoch": 1.6466744457409568,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018696945768870764,
      "loss": 0.1032,
      "step": 3528
    },
    {
      "epoch": 1.6471411901983664,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018696221406432813,
      "loss": 0.107,
      "step": 3529
    },
    {
      "epoch": 1.647607934655776,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001869549685675556,
      "loss": 0.1077,
      "step": 3530
    },
    {
      "epoch": 1.6480746791131855,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00018694772119854617,
      "loss": 0.1684,
      "step": 3531
    },
    {
      "epoch": 1.648541423570595,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00018694047195745578,
      "loss": 0.0765,
      "step": 3532
    },
    {
      "epoch": 1.6490081680280047,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001869332208444406,
      "loss": 0.0813,
      "step": 3533
    },
    {
      "epoch": 1.6494749124854142,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018692596785965674,
      "loss": 0.1289,
      "step": 3534
    },
    {
      "epoch": 1.6499416569428238,
      "grad_norm": 0.875,
      "learning_rate": 0.0001869187130032603,
      "loss": 0.0869,
      "step": 3535
    },
    {
      "epoch": 1.6504084014002334,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018691145627540757,
      "loss": 0.0939,
      "step": 3536
    },
    {
      "epoch": 1.650875145857643,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001869041976762547,
      "loss": 0.1406,
      "step": 3537
    },
    {
      "epoch": 1.6513418903150525,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018689693720595808,
      "loss": 0.1091,
      "step": 3538
    },
    {
      "epoch": 1.651808634772462,
      "grad_norm": 1.1328125,
      "learning_rate": 0.000186889674864674,
      "loss": 0.1147,
      "step": 3539
    },
    {
      "epoch": 1.6522753792298717,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018688241065255878,
      "loss": 0.1437,
      "step": 3540
    },
    {
      "epoch": 1.6527421236872812,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00018687514456976886,
      "loss": 0.0942,
      "step": 3541
    },
    {
      "epoch": 1.6532088681446908,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018686787661646072,
      "loss": 0.1087,
      "step": 3542
    },
    {
      "epoch": 1.6536756126021004,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001868606067927908,
      "loss": 0.1341,
      "step": 3543
    },
    {
      "epoch": 1.65414235705951,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018685333509891564,
      "loss": 0.1126,
      "step": 3544
    },
    {
      "epoch": 1.6546091015169195,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001868460615349918,
      "loss": 0.1407,
      "step": 3545
    },
    {
      "epoch": 1.655075845974329,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018683878610117592,
      "loss": 0.1398,
      "step": 3546
    },
    {
      "epoch": 1.6555425904317387,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001868315087976246,
      "loss": 0.0965,
      "step": 3547
    },
    {
      "epoch": 1.656009334889148,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001868242296244946,
      "loss": 0.0853,
      "step": 3548
    },
    {
      "epoch": 1.6564760793465578,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018681694858194256,
      "loss": 0.0973,
      "step": 3549
    },
    {
      "epoch": 1.6569428238039672,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001868096656701253,
      "loss": 0.0998,
      "step": 3550
    },
    {
      "epoch": 1.657409568261377,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018680238088919962,
      "loss": 0.1244,
      "step": 3551
    },
    {
      "epoch": 1.6578763127187863,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001867950942393224,
      "loss": 0.1234,
      "step": 3552
    },
    {
      "epoch": 1.6583430571761961,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018678780572065046,
      "loss": 0.1044,
      "step": 3553
    },
    {
      "epoch": 1.6588098016336055,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018678051533334078,
      "loss": 0.1419,
      "step": 3554
    },
    {
      "epoch": 1.6592765460910153,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00018677322307755032,
      "loss": 0.1454,
      "step": 3555
    },
    {
      "epoch": 1.6597432905484246,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001867659289534361,
      "loss": 0.0975,
      "step": 3556
    },
    {
      "epoch": 1.6602100350058344,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00018675863296115514,
      "loss": 0.0928,
      "step": 3557
    },
    {
      "epoch": 1.6606767794632438,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018675133510086457,
      "loss": 0.1004,
      "step": 3558
    },
    {
      "epoch": 1.6611435239206536,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018674403537272152,
      "loss": 0.1357,
      "step": 3559
    },
    {
      "epoch": 1.661610268378063,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018673673377688312,
      "loss": 0.0829,
      "step": 3560
    },
    {
      "epoch": 1.6620770128354727,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001867294303135066,
      "loss": 0.1016,
      "step": 3561
    },
    {
      "epoch": 1.662543757292882,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018672212498274923,
      "loss": 0.1184,
      "step": 3562
    },
    {
      "epoch": 1.6630105017502919,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001867148177847683,
      "loss": 0.0973,
      "step": 3563
    },
    {
      "epoch": 1.6634772462077012,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018670750871972112,
      "loss": 0.1038,
      "step": 3564
    },
    {
      "epoch": 1.663943990665111,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018670019778776507,
      "loss": 0.1071,
      "step": 3565
    },
    {
      "epoch": 1.6644107351225204,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018669288498905756,
      "loss": 0.0918,
      "step": 3566
    },
    {
      "epoch": 1.6648774795799302,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018668557032375607,
      "loss": 0.1077,
      "step": 3567
    },
    {
      "epoch": 1.6653442240373395,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018667825379201807,
      "loss": 0.1125,
      "step": 3568
    },
    {
      "epoch": 1.665810968494749,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018667093539400108,
      "loss": 0.0886,
      "step": 3569
    },
    {
      "epoch": 1.6662777129521587,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018666361512986268,
      "loss": 0.1095,
      "step": 3570
    },
    {
      "epoch": 1.6667444574095682,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018665629299976052,
      "loss": 0.1078,
      "step": 3571
    },
    {
      "epoch": 1.6672112018669778,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001866489690038522,
      "loss": 0.0694,
      "step": 3572
    },
    {
      "epoch": 1.6676779463243874,
      "grad_norm": 1.0,
      "learning_rate": 0.00018664164314229545,
      "loss": 0.1241,
      "step": 3573
    },
    {
      "epoch": 1.668144690781797,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018663431541524799,
      "loss": 0.0936,
      "step": 3574
    },
    {
      "epoch": 1.6686114352392065,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018662698582286758,
      "loss": 0.1189,
      "step": 3575
    },
    {
      "epoch": 1.669078179696616,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018661965436531208,
      "loss": 0.0966,
      "step": 3576
    },
    {
      "epoch": 1.6695449241540257,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001866123210427393,
      "loss": 0.1019,
      "step": 3577
    },
    {
      "epoch": 1.6700116686114352,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018660498585530716,
      "loss": 0.0984,
      "step": 3578
    },
    {
      "epoch": 1.6704784130688448,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018659764880317358,
      "loss": 0.0938,
      "step": 3579
    },
    {
      "epoch": 1.6709451575262544,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018659030988649652,
      "loss": 0.0961,
      "step": 3580
    },
    {
      "epoch": 1.671411901983664,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018658296910543405,
      "loss": 0.0865,
      "step": 3581
    },
    {
      "epoch": 1.6718786464410735,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018657562646014414,
      "loss": 0.0997,
      "step": 3582
    },
    {
      "epoch": 1.672345390898483,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018656828195078498,
      "loss": 0.0884,
      "step": 3583
    },
    {
      "epoch": 1.6728121353558927,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018656093557751463,
      "loss": 0.135,
      "step": 3584
    },
    {
      "epoch": 1.6732788798133023,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018655358734049132,
      "loss": 0.1157,
      "step": 3585
    },
    {
      "epoch": 1.6737456242707118,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018654623723987325,
      "loss": 0.1034,
      "step": 3586
    },
    {
      "epoch": 1.6742123687281214,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018653888527581867,
      "loss": 0.1191,
      "step": 3587
    },
    {
      "epoch": 1.674679113185531,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018653153144848584,
      "loss": 0.1135,
      "step": 3588
    },
    {
      "epoch": 1.6751458576429403,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00018652417575803316,
      "loss": 0.1553,
      "step": 3589
    },
    {
      "epoch": 1.6756126021003501,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018651681820461895,
      "loss": 0.1416,
      "step": 3590
    },
    {
      "epoch": 1.6760793465577595,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018650945878840168,
      "loss": 0.1154,
      "step": 3591
    },
    {
      "epoch": 1.6765460910151693,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018650209750953973,
      "loss": 0.0871,
      "step": 3592
    },
    {
      "epoch": 1.6770128354725786,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018649473436819166,
      "loss": 0.112,
      "step": 3593
    },
    {
      "epoch": 1.6774795799299884,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018648736936451604,
      "loss": 0.108,
      "step": 3594
    },
    {
      "epoch": 1.6779463243873978,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018648000249867134,
      "loss": 0.1284,
      "step": 3595
    },
    {
      "epoch": 1.6784130688448076,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018647263377081626,
      "loss": 0.1165,
      "step": 3596
    },
    {
      "epoch": 1.678879813302217,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018646526318110947,
      "loss": 0.0901,
      "step": 3597
    },
    {
      "epoch": 1.6793465577596267,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018645789072970956,
      "loss": 0.096,
      "step": 3598
    },
    {
      "epoch": 1.679813302217036,
      "grad_norm": 0.828125,
      "learning_rate": 0.00018645051641677537,
      "loss": 0.0874,
      "step": 3599
    },
    {
      "epoch": 1.6802800466744459,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018644314024246565,
      "loss": 0.1144,
      "step": 3600
    },
    {
      "epoch": 1.6807467911318552,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001864357622069392,
      "loss": 0.0933,
      "step": 3601
    },
    {
      "epoch": 1.681213535589265,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018642838231035488,
      "loss": 0.0965,
      "step": 3602
    },
    {
      "epoch": 1.6816802800466744,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018642100055287161,
      "loss": 0.0946,
      "step": 3603
    },
    {
      "epoch": 1.6821470245040842,
      "grad_norm": 2.3125,
      "learning_rate": 0.0001864136169346483,
      "loss": 0.1412,
      "step": 3604
    },
    {
      "epoch": 1.6826137689614935,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018640623145584396,
      "loss": 0.0988,
      "step": 3605
    },
    {
      "epoch": 1.6830805134189033,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018639884411661756,
      "loss": 0.1067,
      "step": 3606
    },
    {
      "epoch": 1.6835472578763127,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001863914549171282,
      "loss": 0.0917,
      "step": 3607
    },
    {
      "epoch": 1.6840140023337224,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018638406385753495,
      "loss": 0.1246,
      "step": 3608
    },
    {
      "epoch": 1.6844807467911318,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018637667093799698,
      "loss": 0.124,
      "step": 3609
    },
    {
      "epoch": 1.6849474912485414,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018636927615867343,
      "loss": 0.1136,
      "step": 3610
    },
    {
      "epoch": 1.685414235705951,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018636187951972356,
      "loss": 0.1301,
      "step": 3611
    },
    {
      "epoch": 1.6858809801633605,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018635448102130657,
      "loss": 0.0894,
      "step": 3612
    },
    {
      "epoch": 1.68634772462077,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001863470806635818,
      "loss": 0.1527,
      "step": 3613
    },
    {
      "epoch": 1.6868144690781797,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001863396784467086,
      "loss": 0.0736,
      "step": 3614
    },
    {
      "epoch": 1.6872812135355892,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018633227437084632,
      "loss": 0.1101,
      "step": 3615
    },
    {
      "epoch": 1.6877479579929988,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001863248684361544,
      "loss": 0.115,
      "step": 3616
    },
    {
      "epoch": 1.6882147024504084,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018631746064279225,
      "loss": 0.0881,
      "step": 3617
    },
    {
      "epoch": 1.688681446907818,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018631005099091944,
      "loss": 0.1193,
      "step": 3618
    },
    {
      "epoch": 1.6891481913652275,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018630263948069543,
      "loss": 0.152,
      "step": 3619
    },
    {
      "epoch": 1.689614935822637,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018629522611227986,
      "loss": 0.1064,
      "step": 3620
    },
    {
      "epoch": 1.6900816802800467,
      "grad_norm": 0.8125,
      "learning_rate": 0.00018628781088583232,
      "loss": 0.0914,
      "step": 3621
    },
    {
      "epoch": 1.6905484247374563,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001862803938015125,
      "loss": 0.1054,
      "step": 3622
    },
    {
      "epoch": 1.6910151691948658,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018627297485948,
      "loss": 0.0978,
      "step": 3623
    },
    {
      "epoch": 1.6914819136522754,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018626555405989466,
      "loss": 0.1098,
      "step": 3624
    },
    {
      "epoch": 1.691948658109685,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001862581314029163,
      "loss": 0.0979,
      "step": 3625
    },
    {
      "epoch": 1.6924154025670946,
      "grad_norm": 0.765625,
      "learning_rate": 0.00018625070688870458,
      "loss": 0.0754,
      "step": 3626
    },
    {
      "epoch": 1.6928821470245041,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018624328051741946,
      "loss": 0.0875,
      "step": 3627
    },
    {
      "epoch": 1.6933488914819137,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018623585228922084,
      "loss": 0.112,
      "step": 3628
    },
    {
      "epoch": 1.6938156359393233,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00018622842220426865,
      "loss": 0.1529,
      "step": 3629
    },
    {
      "epoch": 1.6942823803967326,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018622099026272283,
      "loss": 0.084,
      "step": 3630
    },
    {
      "epoch": 1.6947491248541424,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018621355646474347,
      "loss": 0.1358,
      "step": 3631
    },
    {
      "epoch": 1.6952158693115518,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001862061208104905,
      "loss": 0.1399,
      "step": 3632
    },
    {
      "epoch": 1.6956826137689616,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001861986833001242,
      "loss": 0.0766,
      "step": 3633
    },
    {
      "epoch": 1.696149358226371,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018619124393380455,
      "loss": 0.1123,
      "step": 3634
    },
    {
      "epoch": 1.6966161026837807,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018618380271169177,
      "loss": 0.1094,
      "step": 3635
    },
    {
      "epoch": 1.69708284714119,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00018617635963394613,
      "loss": 0.068,
      "step": 3636
    },
    {
      "epoch": 1.6975495915985999,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018616891470072785,
      "loss": 0.1002,
      "step": 3637
    },
    {
      "epoch": 1.6980163360560092,
      "grad_norm": 1.234375,
      "learning_rate": 0.00018616146791219725,
      "loss": 0.1179,
      "step": 3638
    },
    {
      "epoch": 1.698483080513419,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018615401926851463,
      "loss": 0.1017,
      "step": 3639
    },
    {
      "epoch": 1.6989498249708284,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018614656876984035,
      "loss": 0.1034,
      "step": 3640
    },
    {
      "epoch": 1.6994165694282382,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001861391164163349,
      "loss": 0.1159,
      "step": 3641
    },
    {
      "epoch": 1.6998833138856475,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018613166220815872,
      "loss": 0.1267,
      "step": 3642
    },
    {
      "epoch": 1.7003500583430573,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018612420614547227,
      "loss": 0.0969,
      "step": 3643
    },
    {
      "epoch": 1.7008168028004667,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00018611674822843608,
      "loss": 0.1118,
      "step": 3644
    },
    {
      "epoch": 1.7012835472578764,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018610928845721078,
      "loss": 0.0793,
      "step": 3645
    },
    {
      "epoch": 1.7017502917152858,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018610182683195695,
      "loss": 0.114,
      "step": 3646
    },
    {
      "epoch": 1.7022170361726956,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018609436335283524,
      "loss": 0.1234,
      "step": 3647
    },
    {
      "epoch": 1.702683780630105,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018608689802000639,
      "loss": 0.086,
      "step": 3648
    },
    {
      "epoch": 1.7031505250875147,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001860794308336311,
      "loss": 0.1074,
      "step": 3649
    },
    {
      "epoch": 1.703617269544924,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018607196179387016,
      "loss": 0.1074,
      "step": 3650
    },
    {
      "epoch": 1.7040840140023337,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001860644909008844,
      "loss": 0.1649,
      "step": 3651
    },
    {
      "epoch": 1.7045507584597432,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018605701815483462,
      "loss": 0.1171,
      "step": 3652
    },
    {
      "epoch": 1.7050175029171528,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001860495435558818,
      "loss": 0.1151,
      "step": 3653
    },
    {
      "epoch": 1.7054842473745624,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018604206710418682,
      "loss": 0.1371,
      "step": 3654
    },
    {
      "epoch": 1.705950991831972,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018603458879991064,
      "loss": 0.1308,
      "step": 3655
    },
    {
      "epoch": 1.7064177362893815,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018602710864321434,
      "loss": 0.1181,
      "step": 3656
    },
    {
      "epoch": 1.706884480746791,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018601962663425894,
      "loss": 0.0837,
      "step": 3657
    },
    {
      "epoch": 1.7073512252042007,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018601214277320552,
      "loss": 0.1602,
      "step": 3658
    },
    {
      "epoch": 1.7078179696616103,
      "grad_norm": 1.0,
      "learning_rate": 0.00018600465706021524,
      "loss": 0.1126,
      "step": 3659
    },
    {
      "epoch": 1.7082847141190198,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001859971694954493,
      "loss": 0.1395,
      "step": 3660
    },
    {
      "epoch": 1.7087514585764294,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018598968007906886,
      "loss": 0.0905,
      "step": 3661
    },
    {
      "epoch": 1.709218203033839,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018598218881123518,
      "loss": 0.1079,
      "step": 3662
    },
    {
      "epoch": 1.7096849474912486,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00018597469569210962,
      "loss": 0.1126,
      "step": 3663
    },
    {
      "epoch": 1.7101516919486581,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018596720072185346,
      "loss": 0.1175,
      "step": 3664
    },
    {
      "epoch": 1.7106184364060677,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018595970390062805,
      "loss": 0.0989,
      "step": 3665
    },
    {
      "epoch": 1.7110851808634773,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018595220522859486,
      "loss": 0.1003,
      "step": 3666
    },
    {
      "epoch": 1.7115519253208868,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018594470470591535,
      "loss": 0.1117,
      "step": 3667
    },
    {
      "epoch": 1.7120186697782964,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00018593720233275096,
      "loss": 0.1645,
      "step": 3668
    },
    {
      "epoch": 1.712485414235706,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018592969810926326,
      "loss": 0.1166,
      "step": 3669
    },
    {
      "epoch": 1.7129521586931156,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018592219203561382,
      "loss": 0.0914,
      "step": 3670
    },
    {
      "epoch": 1.7134189031505251,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018591468411196426,
      "loss": 0.1366,
      "step": 3671
    },
    {
      "epoch": 1.7138856476079347,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018590717433847624,
      "loss": 0.1388,
      "step": 3672
    },
    {
      "epoch": 1.714352392065344,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00018589966271531142,
      "loss": 0.0812,
      "step": 3673
    },
    {
      "epoch": 1.7148191365227539,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018589214924263157,
      "loss": 0.1273,
      "step": 3674
    },
    {
      "epoch": 1.7152858809801632,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018588463392059843,
      "loss": 0.1298,
      "step": 3675
    },
    {
      "epoch": 1.715752625437573,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018587711674937385,
      "loss": 0.1253,
      "step": 3676
    },
    {
      "epoch": 1.7162193698949824,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018586959772911964,
      "loss": 0.0957,
      "step": 3677
    },
    {
      "epoch": 1.7166861143523922,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001858620768599977,
      "loss": 0.0978,
      "step": 3678
    },
    {
      "epoch": 1.7171528588098015,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018585455414217003,
      "loss": 0.0942,
      "step": 3679
    },
    {
      "epoch": 1.7176196032672113,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018584702957579854,
      "loss": 0.101,
      "step": 3680
    },
    {
      "epoch": 1.7180863477246207,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018583950316104523,
      "loss": 0.1299,
      "step": 3681
    },
    {
      "epoch": 1.7185530921820305,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018583197489807223,
      "loss": 0.087,
      "step": 3682
    },
    {
      "epoch": 1.7190198366394398,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018582444478704152,
      "loss": 0.0805,
      "step": 3683
    },
    {
      "epoch": 1.7194865810968496,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001858169128281153,
      "loss": 0.0869,
      "step": 3684
    },
    {
      "epoch": 1.719953325554259,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018580937902145576,
      "loss": 0.0941,
      "step": 3685
    },
    {
      "epoch": 1.7204200700116687,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018580184336722505,
      "loss": 0.0925,
      "step": 3686
    },
    {
      "epoch": 1.720886814469078,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001857943058655855,
      "loss": 0.0949,
      "step": 3687
    },
    {
      "epoch": 1.721353558926488,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001857867665166993,
      "loss": 0.1011,
      "step": 3688
    },
    {
      "epoch": 1.7218203033838972,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018577922532072886,
      "loss": 0.0884,
      "step": 3689
    },
    {
      "epoch": 1.722287047841307,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018577168227783654,
      "loss": 0.1166,
      "step": 3690
    },
    {
      "epoch": 1.7227537922987164,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001857641373881847,
      "loss": 0.0932,
      "step": 3691
    },
    {
      "epoch": 1.7232205367561262,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018575659065193584,
      "loss": 0.0996,
      "step": 3692
    },
    {
      "epoch": 1.7236872812135355,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018574904206925243,
      "loss": 0.1152,
      "step": 3693
    },
    {
      "epoch": 1.7241540256709451,
      "grad_norm": 1.0,
      "learning_rate": 0.00018574149164029703,
      "loss": 0.0994,
      "step": 3694
    },
    {
      "epoch": 1.7246207701283547,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018573393936523216,
      "loss": 0.1292,
      "step": 3695
    },
    {
      "epoch": 1.7250875145857643,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018572638524422047,
      "loss": 0.1135,
      "step": 3696
    },
    {
      "epoch": 1.7255542590431738,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018571882927742456,
      "loss": 0.1447,
      "step": 3697
    },
    {
      "epoch": 1.7260210035005834,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018571127146500718,
      "loss": 0.1479,
      "step": 3698
    },
    {
      "epoch": 1.726487747957993,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000185703711807131,
      "loss": 0.102,
      "step": 3699
    },
    {
      "epoch": 1.7269544924154026,
      "grad_norm": 1.0,
      "learning_rate": 0.00018569615030395885,
      "loss": 0.1398,
      "step": 3700
    },
    {
      "epoch": 1.7274212368728121,
      "grad_norm": 0.75,
      "learning_rate": 0.00018568858695565345,
      "loss": 0.0915,
      "step": 3701
    },
    {
      "epoch": 1.7278879813302217,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018568102176237775,
      "loss": 0.1029,
      "step": 3702
    },
    {
      "epoch": 1.7283547257876313,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018567345472429456,
      "loss": 0.1234,
      "step": 3703
    },
    {
      "epoch": 1.7288214702450408,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018566588584156683,
      "loss": 0.157,
      "step": 3704
    },
    {
      "epoch": 1.7292882147024504,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018565831511435757,
      "loss": 0.1155,
      "step": 3705
    },
    {
      "epoch": 1.72975495915986,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001856507425428297,
      "loss": 0.0976,
      "step": 3706
    },
    {
      "epoch": 1.7302217036172696,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018564316812714636,
      "loss": 0.0977,
      "step": 3707
    },
    {
      "epoch": 1.7306884480746791,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018563559186747056,
      "loss": 0.1068,
      "step": 3708
    },
    {
      "epoch": 1.7311551925320887,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018562801376396544,
      "loss": 0.0898,
      "step": 3709
    },
    {
      "epoch": 1.7316219369894983,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018562043381679418,
      "loss": 0.0868,
      "step": 3710
    },
    {
      "epoch": 1.7320886814469079,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018561285202611997,
      "loss": 0.1146,
      "step": 3711
    },
    {
      "epoch": 1.7325554259043174,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018560526839210608,
      "loss": 0.1063,
      "step": 3712
    },
    {
      "epoch": 1.733022170361727,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001855976829149158,
      "loss": 0.0681,
      "step": 3713
    },
    {
      "epoch": 1.7334889148191364,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018559009559471242,
      "loss": 0.1153,
      "step": 3714
    },
    {
      "epoch": 1.7339556592765462,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018558250643165933,
      "loss": 0.1109,
      "step": 3715
    },
    {
      "epoch": 1.7344224037339555,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001855749154259199,
      "loss": 0.0992,
      "step": 3716
    },
    {
      "epoch": 1.7348891481913653,
      "grad_norm": 1.0,
      "learning_rate": 0.0001855673225776576,
      "loss": 0.1373,
      "step": 3717
    },
    {
      "epoch": 1.7353558926487747,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018555972788703593,
      "loss": 0.1075,
      "step": 3718
    },
    {
      "epoch": 1.7358226371061845,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018555213135421834,
      "loss": 0.0706,
      "step": 3719
    },
    {
      "epoch": 1.7362893815635938,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00018554453297936848,
      "loss": 0.0837,
      "step": 3720
    },
    {
      "epoch": 1.7367561260210036,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001855369327626499,
      "loss": 0.1385,
      "step": 3721
    },
    {
      "epoch": 1.737222870478413,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018552933070422624,
      "loss": 0.0956,
      "step": 3722
    },
    {
      "epoch": 1.7376896149358227,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018552172680426124,
      "loss": 0.1038,
      "step": 3723
    },
    {
      "epoch": 1.738156359393232,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018551412106291852,
      "loss": 0.1031,
      "step": 3724
    },
    {
      "epoch": 1.738623103850642,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018550651348036193,
      "loss": 0.087,
      "step": 3725
    },
    {
      "epoch": 1.7390898483080512,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00018549890405675522,
      "loss": 0.0707,
      "step": 3726
    },
    {
      "epoch": 1.739556592765461,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018549129279226223,
      "loss": 0.1369,
      "step": 3727
    },
    {
      "epoch": 1.7400233372228704,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018548367968704687,
      "loss": 0.1269,
      "step": 3728
    },
    {
      "epoch": 1.7404900816802802,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018547606474127304,
      "loss": 0.0739,
      "step": 3729
    },
    {
      "epoch": 1.7409568261376895,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018546844795510467,
      "loss": 0.1033,
      "step": 3730
    },
    {
      "epoch": 1.7414235705950993,
      "grad_norm": 1.0,
      "learning_rate": 0.0001854608293287058,
      "loss": 0.1084,
      "step": 3731
    },
    {
      "epoch": 1.7418903150525087,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001854532088622405,
      "loss": 0.1093,
      "step": 3732
    },
    {
      "epoch": 1.7423570595099185,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018544558655587276,
      "loss": 0.1117,
      "step": 3733
    },
    {
      "epoch": 1.7428238039673278,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018543796240976673,
      "loss": 0.1261,
      "step": 3734
    },
    {
      "epoch": 1.7432905484247374,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018543033642408658,
      "loss": 0.1143,
      "step": 3735
    },
    {
      "epoch": 1.743757292882147,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018542270859899654,
      "loss": 0.0934,
      "step": 3736
    },
    {
      "epoch": 1.7442240373395566,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018541507893466075,
      "loss": 0.0842,
      "step": 3737
    },
    {
      "epoch": 1.7446907817969661,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018540744743124357,
      "loss": 0.1123,
      "step": 3738
    },
    {
      "epoch": 1.7451575262543757,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018539981408890929,
      "loss": 0.1094,
      "step": 3739
    },
    {
      "epoch": 1.7456242707117853,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018539217890782222,
      "loss": 0.1185,
      "step": 3740
    },
    {
      "epoch": 1.7460910151691948,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00018538454188814685,
      "loss": 0.0744,
      "step": 3741
    },
    {
      "epoch": 1.7465577596266044,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001853769030300475,
      "loss": 0.0983,
      "step": 3742
    },
    {
      "epoch": 1.747024504084014,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018536926233368874,
      "loss": 0.0922,
      "step": 3743
    },
    {
      "epoch": 1.7474912485414236,
      "grad_norm": 0.875,
      "learning_rate": 0.00018536161979923504,
      "loss": 0.0842,
      "step": 3744
    },
    {
      "epoch": 1.7479579929988331,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018535397542685094,
      "loss": 0.0974,
      "step": 3745
    },
    {
      "epoch": 1.7484247374562427,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018534632921670107,
      "loss": 0.0907,
      "step": 3746
    },
    {
      "epoch": 1.7488914819136523,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018533868116895,
      "loss": 0.1031,
      "step": 3747
    },
    {
      "epoch": 1.7493582263710619,
      "grad_norm": 1.125,
      "learning_rate": 0.00018533103128376248,
      "loss": 0.1227,
      "step": 3748
    },
    {
      "epoch": 1.7498249708284714,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001853233795613032,
      "loss": 0.0832,
      "step": 3749
    },
    {
      "epoch": 1.750291715285881,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001853157260017368,
      "loss": 0.0964,
      "step": 3750
    },
    {
      "epoch": 1.7507584597432906,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018530807060522823,
      "loss": 0.0779,
      "step": 3751
    },
    {
      "epoch": 1.7512252042007002,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018530041337194226,
      "loss": 0.1441,
      "step": 3752
    },
    {
      "epoch": 1.7512252042007002,
      "eval_loss": 1.46182119846344,
      "eval_runtime": 94.8522,
      "eval_samples_per_second": 19.019,
      "eval_steps_per_second": 2.383,
      "step": 3752
    },
    {
      "epoch": 1.7516919486581097,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001852927543020437,
      "loss": 0.1097,
      "step": 3753
    },
    {
      "epoch": 1.7521586931155193,
      "grad_norm": 1.25,
      "learning_rate": 0.00018528509339569755,
      "loss": 0.1364,
      "step": 3754
    },
    {
      "epoch": 1.7526254375729287,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018527743065306867,
      "loss": 0.1132,
      "step": 3755
    },
    {
      "epoch": 1.7530921820303385,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018526976607432212,
      "loss": 0.1266,
      "step": 3756
    },
    {
      "epoch": 1.7535589264877478,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001852620996596229,
      "loss": 0.1122,
      "step": 3757
    },
    {
      "epoch": 1.7540256709451576,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018525443140913605,
      "loss": 0.1017,
      "step": 3758
    },
    {
      "epoch": 1.754492415402567,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018524676132302672,
      "loss": 0.0835,
      "step": 3759
    },
    {
      "epoch": 1.7549591598599767,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018523908940146005,
      "loss": 0.1003,
      "step": 3760
    },
    {
      "epoch": 1.755425904317386,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018523141564460115,
      "loss": 0.1446,
      "step": 3761
    },
    {
      "epoch": 1.755892648774796,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001852237400526154,
      "loss": 0.1253,
      "step": 3762
    },
    {
      "epoch": 1.7563593932322052,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018521606262566787,
      "loss": 0.1101,
      "step": 3763
    },
    {
      "epoch": 1.756826137689615,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018520838336392404,
      "loss": 0.0999,
      "step": 3764
    },
    {
      "epoch": 1.7572928821470244,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018520070226754913,
      "loss": 0.0991,
      "step": 3765
    },
    {
      "epoch": 1.7577596266044342,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018519301933670859,
      "loss": 0.1192,
      "step": 3766
    },
    {
      "epoch": 1.7582263710618435,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001851853345715678,
      "loss": 0.1076,
      "step": 3767
    },
    {
      "epoch": 1.7586931155192533,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018517764797229224,
      "loss": 0.0971,
      "step": 3768
    },
    {
      "epoch": 1.7591598599766627,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018516995953904742,
      "loss": 0.0882,
      "step": 3769
    },
    {
      "epoch": 1.7596266044340725,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001851622692719989,
      "loss": 0.1058,
      "step": 3770
    },
    {
      "epoch": 1.7600933488914818,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018515457717131222,
      "loss": 0.1063,
      "step": 3771
    },
    {
      "epoch": 1.7605600933488916,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000185146883237153,
      "loss": 0.0974,
      "step": 3772
    },
    {
      "epoch": 1.761026837806301,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018513918746968693,
      "loss": 0.0916,
      "step": 3773
    },
    {
      "epoch": 1.7614935822637108,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00018513148986907968,
      "loss": 0.1313,
      "step": 3774
    },
    {
      "epoch": 1.7619603267211201,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018512379043549699,
      "loss": 0.0916,
      "step": 3775
    },
    {
      "epoch": 1.7624270711785297,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018511608916910464,
      "loss": 0.0721,
      "step": 3776
    },
    {
      "epoch": 1.7628938156359393,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018510838607006848,
      "loss": 0.0969,
      "step": 3777
    },
    {
      "epoch": 1.7633605600933488,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018510068113855433,
      "loss": 0.0896,
      "step": 3778
    },
    {
      "epoch": 1.7638273045507584,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001850929743747281,
      "loss": 0.1078,
      "step": 3779
    },
    {
      "epoch": 1.764294049008168,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001850852657787557,
      "loss": 0.1194,
      "step": 3780
    },
    {
      "epoch": 1.7647607934655776,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018507755535080312,
      "loss": 0.114,
      "step": 3781
    },
    {
      "epoch": 1.7652275379229871,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018506984309103638,
      "loss": 0.0993,
      "step": 3782
    },
    {
      "epoch": 1.7656942823803967,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018506212899962156,
      "loss": 0.1162,
      "step": 3783
    },
    {
      "epoch": 1.7661610268378063,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018505441307672472,
      "loss": 0.0636,
      "step": 3784
    },
    {
      "epoch": 1.7666277712952159,
      "grad_norm": 0.734375,
      "learning_rate": 0.000185046695322512,
      "loss": 0.0706,
      "step": 3785
    },
    {
      "epoch": 1.7670945157526254,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018503897573714952,
      "loss": 0.104,
      "step": 3786
    },
    {
      "epoch": 1.767561260210035,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00018503125432080356,
      "loss": 0.0986,
      "step": 3787
    },
    {
      "epoch": 1.7680280046674446,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001850235310736404,
      "loss": 0.0874,
      "step": 3788
    },
    {
      "epoch": 1.7684947491248542,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018501580599582622,
      "loss": 0.0729,
      "step": 3789
    },
    {
      "epoch": 1.7689614935822637,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018500807908752745,
      "loss": 0.0752,
      "step": 3790
    },
    {
      "epoch": 1.7694282380396733,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001850003503489104,
      "loss": 0.1197,
      "step": 3791
    },
    {
      "epoch": 1.7698949824970829,
      "grad_norm": 1.0,
      "learning_rate": 0.0001849926197801415,
      "loss": 0.0857,
      "step": 3792
    },
    {
      "epoch": 1.7703617269544925,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001849848873813872,
      "loss": 0.0802,
      "step": 3793
    },
    {
      "epoch": 1.770828471411902,
      "grad_norm": 1.3671875,
      "learning_rate": 0.00018497715315281396,
      "loss": 0.1092,
      "step": 3794
    },
    {
      "epoch": 1.7712952158693116,
      "grad_norm": 1.1953125,
      "learning_rate": 0.00018496941709458836,
      "loss": 0.0813,
      "step": 3795
    },
    {
      "epoch": 1.7717619603267212,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001849616792068769,
      "loss": 0.0848,
      "step": 3796
    },
    {
      "epoch": 1.7722287047841307,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018495393948984626,
      "loss": 0.1235,
      "step": 3797
    },
    {
      "epoch": 1.77269544924154,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018494619794366305,
      "loss": 0.083,
      "step": 3798
    },
    {
      "epoch": 1.77316219369895,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018493845456849387,
      "loss": 0.0859,
      "step": 3799
    },
    {
      "epoch": 1.7736289381563592,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018493070936450562,
      "loss": 0.1075,
      "step": 3800
    },
    {
      "epoch": 1.774095682613769,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001849229623318649,
      "loss": 0.1053,
      "step": 3801
    },
    {
      "epoch": 1.7745624270711784,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018491521347073858,
      "loss": 0.0763,
      "step": 3802
    },
    {
      "epoch": 1.7750291715285882,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018490746278129353,
      "loss": 0.0979,
      "step": 3803
    },
    {
      "epoch": 1.7754959159859975,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001848997102636966,
      "loss": 0.0945,
      "step": 3804
    },
    {
      "epoch": 1.7759626604434073,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001848919559181147,
      "loss": 0.0933,
      "step": 3805
    },
    {
      "epoch": 1.7764294049008167,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018488419974471481,
      "loss": 0.1016,
      "step": 3806
    },
    {
      "epoch": 1.7768961493582265,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001848764417436639,
      "loss": 0.1144,
      "step": 3807
    },
    {
      "epoch": 1.7773628938156358,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018486868191512902,
      "loss": 0.107,
      "step": 3808
    },
    {
      "epoch": 1.7778296382730456,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00018486092025927727,
      "loss": 0.0887,
      "step": 3809
    },
    {
      "epoch": 1.778296382730455,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018485315677627576,
      "loss": 0.1196,
      "step": 3810
    },
    {
      "epoch": 1.7787631271878648,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018484539146629162,
      "loss": 0.1247,
      "step": 3811
    },
    {
      "epoch": 1.7792298716452741,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018483762432949208,
      "loss": 0.0837,
      "step": 3812
    },
    {
      "epoch": 1.779696616102684,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00018482985536604437,
      "loss": 0.1015,
      "step": 3813
    },
    {
      "epoch": 1.7801633605600933,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001848220845761157,
      "loss": 0.1258,
      "step": 3814
    },
    {
      "epoch": 1.780630105017503,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018481431195987348,
      "loss": 0.0834,
      "step": 3815
    },
    {
      "epoch": 1.7810968494749124,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018480653751748504,
      "loss": 0.093,
      "step": 3816
    },
    {
      "epoch": 1.781563593932322,
      "grad_norm": 1.0,
      "learning_rate": 0.0001847987612491177,
      "loss": 0.1321,
      "step": 3817
    },
    {
      "epoch": 1.7820303383897316,
      "grad_norm": 1.015625,
      "learning_rate": 0.000184790983154939,
      "loss": 0.098,
      "step": 3818
    },
    {
      "epoch": 1.7824970828471411,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00018478320323511634,
      "loss": 0.0821,
      "step": 3819
    },
    {
      "epoch": 1.7829638273045507,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018477542148981723,
      "loss": 0.1082,
      "step": 3820
    },
    {
      "epoch": 1.7834305717619603,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018476763791920922,
      "loss": 0.0917,
      "step": 3821
    },
    {
      "epoch": 1.7838973162193699,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018475985252345995,
      "loss": 0.0959,
      "step": 3822
    },
    {
      "epoch": 1.7843640606767794,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018475206530273698,
      "loss": 0.1518,
      "step": 3823
    },
    {
      "epoch": 1.784830805134189,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018474427625720803,
      "loss": 0.0786,
      "step": 3824
    },
    {
      "epoch": 1.7852975495915986,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018473648538704077,
      "loss": 0.0987,
      "step": 3825
    },
    {
      "epoch": 1.7857642940490082,
      "grad_norm": 1.1484375,
      "learning_rate": 0.000184728692692403,
      "loss": 0.1259,
      "step": 3826
    },
    {
      "epoch": 1.7862310385064177,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018472089817346242,
      "loss": 0.1071,
      "step": 3827
    },
    {
      "epoch": 1.7866977829638273,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018471310183038696,
      "loss": 0.1308,
      "step": 3828
    },
    {
      "epoch": 1.7871645274212369,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018470530366334438,
      "loss": 0.1065,
      "step": 3829
    },
    {
      "epoch": 1.7876312718786465,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018469750367250263,
      "loss": 0.1253,
      "step": 3830
    },
    {
      "epoch": 1.788098016336056,
      "grad_norm": 1.0,
      "learning_rate": 0.00018468970185802968,
      "loss": 0.1057,
      "step": 3831
    },
    {
      "epoch": 1.7885647607934656,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00018468189822009348,
      "loss": 0.0777,
      "step": 3832
    },
    {
      "epoch": 1.7890315052508752,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018467409275886202,
      "loss": 0.1072,
      "step": 3833
    },
    {
      "epoch": 1.7894982497082847,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001846662854745034,
      "loss": 0.1175,
      "step": 3834
    },
    {
      "epoch": 1.7899649941656943,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018465847636718575,
      "loss": 0.1134,
      "step": 3835
    },
    {
      "epoch": 1.790431738623104,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00018465066543707717,
      "loss": 0.0662,
      "step": 3836
    },
    {
      "epoch": 1.7908984830805135,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018464285268434582,
      "loss": 0.0944,
      "step": 3837
    },
    {
      "epoch": 1.791365227537923,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00018463503810915995,
      "loss": 0.0891,
      "step": 3838
    },
    {
      "epoch": 1.7918319719953324,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018462722171168783,
      "loss": 0.1148,
      "step": 3839
    },
    {
      "epoch": 1.7922987164527422,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001846194034920977,
      "loss": 0.0893,
      "step": 3840
    },
    {
      "epoch": 1.7927654609101515,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001846115834505579,
      "loss": 0.1159,
      "step": 3841
    },
    {
      "epoch": 1.7932322053675613,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018460376158723689,
      "loss": 0.0933,
      "step": 3842
    },
    {
      "epoch": 1.7936989498249707,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018459593790230299,
      "loss": 0.0755,
      "step": 3843
    },
    {
      "epoch": 1.7941656942823805,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001845881123959247,
      "loss": 0.0827,
      "step": 3844
    },
    {
      "epoch": 1.7946324387397898,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00018458028506827047,
      "loss": 0.0745,
      "step": 3845
    },
    {
      "epoch": 1.7950991831971996,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018457245591950886,
      "loss": 0.0844,
      "step": 3846
    },
    {
      "epoch": 1.795565927654609,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00018456462494980847,
      "loss": 0.0886,
      "step": 3847
    },
    {
      "epoch": 1.7960326721120188,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018455679215933788,
      "loss": 0.0838,
      "step": 3848
    },
    {
      "epoch": 1.7964994165694281,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018454895754826573,
      "loss": 0.073,
      "step": 3849
    },
    {
      "epoch": 1.796966161026838,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001845411211167607,
      "loss": 0.0955,
      "step": 3850
    },
    {
      "epoch": 1.7974329054842473,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018453328286499154,
      "loss": 0.0768,
      "step": 3851
    },
    {
      "epoch": 1.797899649941657,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00018452544279312703,
      "loss": 0.0859,
      "step": 3852
    },
    {
      "epoch": 1.7983663943990664,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018451760090133591,
      "loss": 0.0676,
      "step": 3853
    },
    {
      "epoch": 1.7988331388564762,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001845097571897871,
      "loss": 0.1309,
      "step": 3854
    },
    {
      "epoch": 1.7992998833138856,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018450191165864945,
      "loss": 0.0865,
      "step": 3855
    },
    {
      "epoch": 1.7997666277712954,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018449406430809186,
      "loss": 0.0926,
      "step": 3856
    },
    {
      "epoch": 1.8002333722287047,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018448621513828334,
      "loss": 0.0857,
      "step": 3857
    },
    {
      "epoch": 1.8007001166861145,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018447836414939287,
      "loss": 0.0829,
      "step": 3858
    },
    {
      "epoch": 1.8011668611435239,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001844705113415895,
      "loss": 0.0753,
      "step": 3859
    },
    {
      "epoch": 1.8016336056009334,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018446265671504227,
      "loss": 0.0852,
      "step": 3860
    },
    {
      "epoch": 1.802100350058343,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018445480026992037,
      "loss": 0.0915,
      "step": 3861
    },
    {
      "epoch": 1.8025670945157526,
      "grad_norm": 1.0,
      "learning_rate": 0.00018444694200639287,
      "loss": 0.0848,
      "step": 3862
    },
    {
      "epoch": 1.8030338389731622,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018443908192462905,
      "loss": 0.1146,
      "step": 3863
    },
    {
      "epoch": 1.8035005834305717,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001844312200247981,
      "loss": 0.087,
      "step": 3864
    },
    {
      "epoch": 1.8039673278879813,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001844233563070693,
      "loss": 0.1719,
      "step": 3865
    },
    {
      "epoch": 1.8044340723453909,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018441549077161194,
      "loss": 0.0931,
      "step": 3866
    },
    {
      "epoch": 1.8049008168028005,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00018440762341859543,
      "loss": 0.077,
      "step": 3867
    },
    {
      "epoch": 1.80536756126021,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018439975424818913,
      "loss": 0.1099,
      "step": 3868
    },
    {
      "epoch": 1.8058343057176196,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018439188326056249,
      "loss": 0.1242,
      "step": 3869
    },
    {
      "epoch": 1.8063010501750292,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018438401045588497,
      "loss": 0.1104,
      "step": 3870
    },
    {
      "epoch": 1.8067677946324388,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018437613583432606,
      "loss": 0.0989,
      "step": 3871
    },
    {
      "epoch": 1.8072345390898483,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018436825939605536,
      "loss": 0.1336,
      "step": 3872
    },
    {
      "epoch": 1.807701283547258,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018436038114124238,
      "loss": 0.0844,
      "step": 3873
    },
    {
      "epoch": 1.8081680280046675,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018435250107005682,
      "loss": 0.1055,
      "step": 3874
    },
    {
      "epoch": 1.808634772462077,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018434461918266832,
      "loss": 0.1219,
      "step": 3875
    },
    {
      "epoch": 1.8091015169194866,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001843367354792466,
      "loss": 0.108,
      "step": 3876
    },
    {
      "epoch": 1.8095682613768962,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001843288499599614,
      "loss": 0.0936,
      "step": 3877
    },
    {
      "epoch": 1.8100350058343058,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001843209626249825,
      "loss": 0.0775,
      "step": 3878
    },
    {
      "epoch": 1.8105017502917153,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018431307347447967,
      "loss": 0.1005,
      "step": 3879
    },
    {
      "epoch": 1.8109684947491247,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018430518250862283,
      "loss": 0.1253,
      "step": 3880
    },
    {
      "epoch": 1.8114352392065345,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018429728972758192,
      "loss": 0.086,
      "step": 3881
    },
    {
      "epoch": 1.8119019836639438,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018428939513152678,
      "loss": 0.1013,
      "step": 3882
    },
    {
      "epoch": 1.8123687281213536,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018428149872062746,
      "loss": 0.0956,
      "step": 3883
    },
    {
      "epoch": 1.812835472578763,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018427360049505398,
      "loss": 0.1251,
      "step": 3884
    },
    {
      "epoch": 1.8133022170361728,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018426570045497638,
      "loss": 0.095,
      "step": 3885
    },
    {
      "epoch": 1.8137689614935821,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018425779860056473,
      "loss": 0.0983,
      "step": 3886
    },
    {
      "epoch": 1.814235705950992,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001842498949319892,
      "loss": 0.1059,
      "step": 3887
    },
    {
      "epoch": 1.8147024504084013,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00018424198944941994,
      "loss": 0.0814,
      "step": 3888
    },
    {
      "epoch": 1.815169194865811,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001842340821530272,
      "loss": 0.1074,
      "step": 3889
    },
    {
      "epoch": 1.8156359393232204,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001842261730429812,
      "loss": 0.0931,
      "step": 3890
    },
    {
      "epoch": 1.8161026837806302,
      "grad_norm": 0.96875,
      "learning_rate": 0.00018421826211945225,
      "loss": 0.1113,
      "step": 3891
    },
    {
      "epoch": 1.8165694282380396,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018421034938261068,
      "loss": 0.0825,
      "step": 3892
    },
    {
      "epoch": 1.8170361726954494,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018420243483262684,
      "loss": 0.1021,
      "step": 3893
    },
    {
      "epoch": 1.8175029171528587,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018419451846967119,
      "loss": 0.1404,
      "step": 3894
    },
    {
      "epoch": 1.8179696616102685,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001841866002939141,
      "loss": 0.095,
      "step": 3895
    },
    {
      "epoch": 1.8184364060676779,
      "grad_norm": 1.21875,
      "learning_rate": 0.0001841786803055261,
      "loss": 0.0921,
      "step": 3896
    },
    {
      "epoch": 1.8189031505250877,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018417075850467773,
      "loss": 0.0765,
      "step": 3897
    },
    {
      "epoch": 1.819369894982497,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018416283489153953,
      "loss": 0.0953,
      "step": 3898
    },
    {
      "epoch": 1.8198366394399068,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018415490946628214,
      "loss": 0.0896,
      "step": 3899
    },
    {
      "epoch": 1.8203033838973162,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018414698222907617,
      "loss": 0.0912,
      "step": 3900
    },
    {
      "epoch": 1.8207701283547257,
      "grad_norm": 1.0,
      "learning_rate": 0.0001841390531800923,
      "loss": 0.1125,
      "step": 3901
    },
    {
      "epoch": 1.8212368728121353,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018413112231950124,
      "loss": 0.115,
      "step": 3902
    },
    {
      "epoch": 1.8217036172695449,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00018412318964747382,
      "loss": 0.0856,
      "step": 3903
    },
    {
      "epoch": 1.8221703617269545,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018411525516418077,
      "loss": 0.1201,
      "step": 3904
    },
    {
      "epoch": 1.822637106184364,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001841073188697929,
      "loss": 0.1108,
      "step": 3905
    },
    {
      "epoch": 1.8231038506417736,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001840993807644812,
      "loss": 0.0961,
      "step": 3906
    },
    {
      "epoch": 1.8235705950991832,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001840914408484165,
      "loss": 0.0753,
      "step": 3907
    },
    {
      "epoch": 1.8240373395565928,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001840834991217698,
      "loss": 0.1569,
      "step": 3908
    },
    {
      "epoch": 1.8245040840140023,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018407555558471204,
      "loss": 0.1007,
      "step": 3909
    },
    {
      "epoch": 1.824970828471412,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018406761023741428,
      "loss": 0.0774,
      "step": 3910
    },
    {
      "epoch": 1.8254375729288215,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018405966308004764,
      "loss": 0.101,
      "step": 3911
    },
    {
      "epoch": 1.825904317386231,
      "grad_norm": 0.875,
      "learning_rate": 0.00018405171411278314,
      "loss": 0.0845,
      "step": 3912
    },
    {
      "epoch": 1.8263710618436406,
      "grad_norm": 0.921875,
      "learning_rate": 0.000184043763335792,
      "loss": 0.0975,
      "step": 3913
    },
    {
      "epoch": 1.8268378063010502,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001840358107492454,
      "loss": 0.085,
      "step": 3914
    },
    {
      "epoch": 1.8273045507584598,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00018402785635331453,
      "loss": 0.0631,
      "step": 3915
    },
    {
      "epoch": 1.8277712952158693,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001840199001481707,
      "loss": 0.1053,
      "step": 3916
    },
    {
      "epoch": 1.828238039673279,
      "grad_norm": 1.0,
      "learning_rate": 0.00018401194213398518,
      "loss": 0.117,
      "step": 3917
    },
    {
      "epoch": 1.8287047841306885,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018400398231092934,
      "loss": 0.0769,
      "step": 3918
    },
    {
      "epoch": 1.829171528588098,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018399602067917456,
      "loss": 0.0819,
      "step": 3919
    },
    {
      "epoch": 1.8296382730455076,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018398805723889228,
      "loss": 0.0954,
      "step": 3920
    },
    {
      "epoch": 1.830105017502917,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001839800919902539,
      "loss": 0.0844,
      "step": 3921
    },
    {
      "epoch": 1.8305717619603268,
      "grad_norm": 1.0,
      "learning_rate": 0.000183972124933431,
      "loss": 0.0982,
      "step": 3922
    },
    {
      "epoch": 1.8310385064177361,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018396415606859506,
      "loss": 0.0809,
      "step": 3923
    },
    {
      "epoch": 1.831505250875146,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001839561853959177,
      "loss": 0.1076,
      "step": 3924
    },
    {
      "epoch": 1.8319719953325553,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001839482129155705,
      "loss": 0.08,
      "step": 3925
    },
    {
      "epoch": 1.832438739789965,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00018394023862772514,
      "loss": 0.1377,
      "step": 3926
    },
    {
      "epoch": 1.8329054842473744,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001839322625325533,
      "loss": 0.0973,
      "step": 3927
    },
    {
      "epoch": 1.8333722287047842,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018392428463022672,
      "loss": 0.0952,
      "step": 3928
    },
    {
      "epoch": 1.8338389731621936,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001839163049209172,
      "loss": 0.0782,
      "step": 3929
    },
    {
      "epoch": 1.8343057176196034,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001839083234047965,
      "loss": 0.0784,
      "step": 3930
    },
    {
      "epoch": 1.8347724620770127,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001839003400820365,
      "loss": 0.1112,
      "step": 3931
    },
    {
      "epoch": 1.8352392065344225,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001838923549528091,
      "loss": 0.0735,
      "step": 3932
    },
    {
      "epoch": 1.8357059509918319,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018388436801728623,
      "loss": 0.0975,
      "step": 3933
    },
    {
      "epoch": 1.8361726954492417,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00018387637927563984,
      "loss": 0.0765,
      "step": 3934
    },
    {
      "epoch": 1.836639439906651,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018386838872804195,
      "loss": 0.1003,
      "step": 3935
    },
    {
      "epoch": 1.8371061843640608,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018386039637466455,
      "loss": 0.1046,
      "step": 3936
    },
    {
      "epoch": 1.8375729288214702,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018385240221567984,
      "loss": 0.1162,
      "step": 3937
    },
    {
      "epoch": 1.83803967327888,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018384440625125983,
      "loss": 0.0915,
      "step": 3938
    },
    {
      "epoch": 1.8385064177362893,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018383640848157674,
      "loss": 0.0986,
      "step": 3939
    },
    {
      "epoch": 1.838973162193699,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018382840890680276,
      "loss": 0.0927,
      "step": 3940
    },
    {
      "epoch": 1.8394399066511085,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001838204075271101,
      "loss": 0.1065,
      "step": 3941
    },
    {
      "epoch": 1.839906651108518,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018381240434267107,
      "loss": 0.0728,
      "step": 3942
    },
    {
      "epoch": 1.8403733955659276,
      "grad_norm": 1.109375,
      "learning_rate": 0.000183804399353658,
      "loss": 0.1295,
      "step": 3943
    },
    {
      "epoch": 1.8408401400233372,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018379639256024322,
      "loss": 0.1083,
      "step": 3944
    },
    {
      "epoch": 1.8413068844807468,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018378838396259915,
      "loss": 0.0928,
      "step": 3945
    },
    {
      "epoch": 1.8417736289381563,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018378037356089823,
      "loss": 0.12,
      "step": 3946
    },
    {
      "epoch": 1.842240373395566,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018377236135531288,
      "loss": 0.1009,
      "step": 3947
    },
    {
      "epoch": 1.8427071178529755,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018376434734601564,
      "loss": 0.0881,
      "step": 3948
    },
    {
      "epoch": 1.843173862310385,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001837563315331791,
      "loss": 0.0904,
      "step": 3949
    },
    {
      "epoch": 1.8436406067677946,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018374831391697574,
      "loss": 0.1047,
      "step": 3950
    },
    {
      "epoch": 1.8441073512252042,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018374029449757831,
      "loss": 0.0852,
      "step": 3951
    },
    {
      "epoch": 1.8445740956826138,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001837322732751594,
      "loss": 0.1025,
      "step": 3952
    },
    {
      "epoch": 1.8450408401400233,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001837242502498918,
      "loss": 0.089,
      "step": 3953
    },
    {
      "epoch": 1.845507584597433,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018371622542194814,
      "loss": 0.088,
      "step": 3954
    },
    {
      "epoch": 1.8459743290548425,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018370819879150128,
      "loss": 0.0758,
      "step": 3955
    },
    {
      "epoch": 1.846441073512252,
      "grad_norm": 1.0,
      "learning_rate": 0.00018370017035872406,
      "loss": 0.1093,
      "step": 3956
    },
    {
      "epoch": 1.8469078179696616,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018369214012378926,
      "loss": 0.0886,
      "step": 3957
    },
    {
      "epoch": 1.8473745624270712,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00018368410808686982,
      "loss": 0.0771,
      "step": 3958
    },
    {
      "epoch": 1.8478413068844808,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018367607424813872,
      "loss": 0.1009,
      "step": 3959
    },
    {
      "epoch": 1.8483080513418904,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001836680386077689,
      "loss": 0.0932,
      "step": 3960
    },
    {
      "epoch": 1.8487747957993,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018366000116593333,
      "loss": 0.1112,
      "step": 3961
    },
    {
      "epoch": 1.8492415402567095,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018365196192280517,
      "loss": 0.0976,
      "step": 3962
    },
    {
      "epoch": 1.849708284714119,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00018364392087855743,
      "loss": 0.0749,
      "step": 3963
    },
    {
      "epoch": 1.8501750291715284,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018363587803336327,
      "loss": 0.0939,
      "step": 3964
    },
    {
      "epoch": 1.8506417736289382,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018362783338739583,
      "loss": 0.0762,
      "step": 3965
    },
    {
      "epoch": 1.8511085180863476,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001836197869408284,
      "loss": 0.0985,
      "step": 3966
    },
    {
      "epoch": 1.8515752625437574,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018361173869383416,
      "loss": 0.089,
      "step": 3967
    },
    {
      "epoch": 1.8520420070011667,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001836036886465864,
      "loss": 0.0871,
      "step": 3968
    },
    {
      "epoch": 1.8525087514585765,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018359563679925847,
      "loss": 0.1358,
      "step": 3969
    },
    {
      "epoch": 1.8529754959159859,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018358758315202377,
      "loss": 0.1076,
      "step": 3970
    },
    {
      "epoch": 1.8534422403733957,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018357952770505564,
      "loss": 0.0901,
      "step": 3971
    },
    {
      "epoch": 1.853908984830805,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018357147045852754,
      "loss": 0.1258,
      "step": 3972
    },
    {
      "epoch": 1.8543757292882148,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018356341141261294,
      "loss": 0.0738,
      "step": 3973
    },
    {
      "epoch": 1.8548424737456242,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001835553505674854,
      "loss": 0.0652,
      "step": 3974
    },
    {
      "epoch": 1.855309218203034,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018354728792331843,
      "loss": 0.0766,
      "step": 3975
    },
    {
      "epoch": 1.8557759626604433,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001835392234802857,
      "loss": 0.1044,
      "step": 3976
    },
    {
      "epoch": 1.856242707117853,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00018353115723856073,
      "loss": 0.091,
      "step": 3977
    },
    {
      "epoch": 1.8567094515752625,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018352308919831732,
      "loss": 0.0725,
      "step": 3978
    },
    {
      "epoch": 1.8571761960326723,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018351501935972913,
      "loss": 0.0873,
      "step": 3979
    },
    {
      "epoch": 1.8576429404900816,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018350694772296987,
      "loss": 0.0922,
      "step": 3980
    },
    {
      "epoch": 1.8581096849474914,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018349887428821342,
      "loss": 0.0926,
      "step": 3981
    },
    {
      "epoch": 1.8585764294049008,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018349079905563356,
      "loss": 0.0881,
      "step": 3982
    },
    {
      "epoch": 1.8590431738623105,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018348272202540412,
      "loss": 0.0834,
      "step": 3983
    },
    {
      "epoch": 1.85950991831972,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018347464319769908,
      "loss": 0.0707,
      "step": 3984
    },
    {
      "epoch": 1.8599766627771295,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018346656257269234,
      "loss": 0.1147,
      "step": 3985
    },
    {
      "epoch": 1.860443407234539,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018345848015055794,
      "loss": 0.096,
      "step": 3986
    },
    {
      "epoch": 1.8609101516919486,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00018345039593146983,
      "loss": 0.0786,
      "step": 3987
    },
    {
      "epoch": 1.8613768961493582,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001834423099156021,
      "loss": 0.0858,
      "step": 3988
    },
    {
      "epoch": 1.8618436406067678,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018343422210312884,
      "loss": 0.11,
      "step": 3989
    },
    {
      "epoch": 1.8623103850641773,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018342613249422424,
      "loss": 0.0772,
      "step": 3990
    },
    {
      "epoch": 1.862777129521587,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00018341804108906243,
      "loss": 0.1147,
      "step": 3991
    },
    {
      "epoch": 1.8632438739789965,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018340994788781768,
      "loss": 0.0997,
      "step": 3992
    },
    {
      "epoch": 1.863710618436406,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018340185289066416,
      "loss": 0.1001,
      "step": 3993
    },
    {
      "epoch": 1.8641773628938156,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018339375609777622,
      "loss": 0.0802,
      "step": 3994
    },
    {
      "epoch": 1.8646441073512252,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001833856575093282,
      "loss": 0.1102,
      "step": 3995
    },
    {
      "epoch": 1.8651108518086348,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018337755712549445,
      "loss": 0.0937,
      "step": 3996
    },
    {
      "epoch": 1.8655775962660444,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018336945494644937,
      "loss": 0.0858,
      "step": 3997
    },
    {
      "epoch": 1.866044340723454,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018336135097236743,
      "loss": 0.0986,
      "step": 3998
    },
    {
      "epoch": 1.8665110851808635,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018335324520342314,
      "loss": 0.0963,
      "step": 3999
    },
    {
      "epoch": 1.866977829638273,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018334513763979096,
      "loss": 0.0862,
      "step": 4000
    },
    {
      "epoch": 1.8674445740956827,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018333702828164552,
      "loss": 0.0949,
      "step": 4001
    },
    {
      "epoch": 1.8679113185530922,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001833289171291614,
      "loss": 0.0908,
      "step": 4002
    },
    {
      "epoch": 1.8683780630105018,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018332080418251322,
      "loss": 0.0792,
      "step": 4003
    },
    {
      "epoch": 1.8688448074679114,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018331268944187567,
      "loss": 0.0812,
      "step": 4004
    },
    {
      "epoch": 1.8693115519253207,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001833045729074235,
      "loss": 0.0725,
      "step": 4005
    },
    {
      "epoch": 1.8697782963827305,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018329645457933145,
      "loss": 0.1034,
      "step": 4006
    },
    {
      "epoch": 1.8702450408401399,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001832883344577743,
      "loss": 0.0946,
      "step": 4007
    },
    {
      "epoch": 1.8707117852975497,
      "grad_norm": 0.875,
      "learning_rate": 0.00018328021254292694,
      "loss": 0.0705,
      "step": 4008
    },
    {
      "epoch": 1.871178529754959,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018327208883496415,
      "loss": 0.0667,
      "step": 4009
    },
    {
      "epoch": 1.8716452742123688,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018326396333406093,
      "loss": 0.0912,
      "step": 4010
    },
    {
      "epoch": 1.8721120186697782,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018325583604039217,
      "loss": 0.0882,
      "step": 4011
    },
    {
      "epoch": 1.872578763127188,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001832477069541329,
      "loss": 0.086,
      "step": 4012
    },
    {
      "epoch": 1.8730455075845973,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00018323957607545811,
      "loss": 0.1507,
      "step": 4013
    },
    {
      "epoch": 1.873512252042007,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018323144340454294,
      "loss": 0.1024,
      "step": 4014
    },
    {
      "epoch": 1.8739789964994165,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018322330894156243,
      "loss": 0.0976,
      "step": 4015
    },
    {
      "epoch": 1.8744457409568263,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018321517268669171,
      "loss": 0.0949,
      "step": 4016
    },
    {
      "epoch": 1.8749124854142356,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018320703464010603,
      "loss": 0.0932,
      "step": 4017
    },
    {
      "epoch": 1.8753792298716454,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018319889480198053,
      "loss": 0.0949,
      "step": 4018
    },
    {
      "epoch": 1.8758459743290548,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018319075317249056,
      "loss": 0.0962,
      "step": 4019
    },
    {
      "epoch": 1.8763127187864646,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018318260975181138,
      "loss": 0.1067,
      "step": 4020
    },
    {
      "epoch": 1.876779463243874,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001831744645401183,
      "loss": 0.0781,
      "step": 4021
    },
    {
      "epoch": 1.8772462077012837,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018316631753758668,
      "loss": 0.1055,
      "step": 4022
    },
    {
      "epoch": 1.877712952158693,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000183158168744392,
      "loss": 0.0984,
      "step": 4023
    },
    {
      "epoch": 1.8781796966161028,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018315001816070968,
      "loss": 0.0858,
      "step": 4024
    },
    {
      "epoch": 1.8786464410735122,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018314186578671522,
      "loss": 0.102,
      "step": 4025
    },
    {
      "epoch": 1.8791131855309218,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001831337116225841,
      "loss": 0.1047,
      "step": 4026
    },
    {
      "epoch": 1.8795799299883313,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018312555566849198,
      "loss": 0.0963,
      "step": 4027
    },
    {
      "epoch": 1.880046674445741,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018311739792461443,
      "loss": 0.0886,
      "step": 4028
    },
    {
      "epoch": 1.8805134189031505,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018310923839112703,
      "loss": 0.1028,
      "step": 4029
    },
    {
      "epoch": 1.88098016336056,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018310107706820554,
      "loss": 0.0868,
      "step": 4030
    },
    {
      "epoch": 1.8814469078179696,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018309291395602565,
      "loss": 0.0849,
      "step": 4031
    },
    {
      "epoch": 1.8819136522753792,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018308474905476317,
      "loss": 0.1204,
      "step": 4032
    },
    {
      "epoch": 1.8823803967327888,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001830765823645938,
      "loss": 0.0652,
      "step": 4033
    },
    {
      "epoch": 1.8828471411901984,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018306841388569348,
      "loss": 0.0834,
      "step": 4034
    },
    {
      "epoch": 1.883313885647608,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018306024361823805,
      "loss": 0.1016,
      "step": 4035
    },
    {
      "epoch": 1.8837806301050175,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001830520715624034,
      "loss": 0.0864,
      "step": 4036
    },
    {
      "epoch": 1.884247374562427,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018304389771836544,
      "loss": 0.1059,
      "step": 4037
    },
    {
      "epoch": 1.8847141190198367,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001830357220863003,
      "loss": 0.0983,
      "step": 4038
    },
    {
      "epoch": 1.8851808634772462,
      "grad_norm": 1.125,
      "learning_rate": 0.00018302754466638393,
      "loss": 0.0979,
      "step": 4039
    },
    {
      "epoch": 1.8856476079346558,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018301936545879236,
      "loss": 0.0833,
      "step": 4040
    },
    {
      "epoch": 1.8861143523920654,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001830111844637018,
      "loss": 0.0704,
      "step": 4041
    },
    {
      "epoch": 1.886581096849475,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018300300168128827,
      "loss": 0.0793,
      "step": 4042
    },
    {
      "epoch": 1.8870478413068845,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018299481711172807,
      "loss": 0.0891,
      "step": 4043
    },
    {
      "epoch": 1.887514585764294,
      "grad_norm": 1.0,
      "learning_rate": 0.00018298663075519734,
      "loss": 0.098,
      "step": 4044
    },
    {
      "epoch": 1.8879813302217037,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001829784426118724,
      "loss": 0.0844,
      "step": 4045
    },
    {
      "epoch": 1.888448074679113,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00018297025268192952,
      "loss": 0.0646,
      "step": 4046
    },
    {
      "epoch": 1.8889148191365228,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018296206096554503,
      "loss": 0.0711,
      "step": 4047
    },
    {
      "epoch": 1.8893815635939322,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018295386746289532,
      "loss": 0.0719,
      "step": 4048
    },
    {
      "epoch": 1.889848308051342,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001829456721741568,
      "loss": 0.0847,
      "step": 4049
    },
    {
      "epoch": 1.8903150525087513,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018293747509950597,
      "loss": 0.1078,
      "step": 4050
    },
    {
      "epoch": 1.890781796966161,
      "grad_norm": 0.875,
      "learning_rate": 0.00018292927623911922,
      "loss": 0.0889,
      "step": 4051
    },
    {
      "epoch": 1.8912485414235705,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018292107559317317,
      "loss": 0.0975,
      "step": 4052
    },
    {
      "epoch": 1.8917152858809803,
      "grad_norm": 1.21875,
      "learning_rate": 0.00018291287316184437,
      "loss": 0.107,
      "step": 4053
    },
    {
      "epoch": 1.8921820303383896,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018290466894530942,
      "loss": 0.0885,
      "step": 4054
    },
    {
      "epoch": 1.8926487747957994,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018289646294374494,
      "loss": 0.0757,
      "step": 4055
    },
    {
      "epoch": 1.8931155192532088,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018288825515732767,
      "loss": 0.1002,
      "step": 4056
    },
    {
      "epoch": 1.8935822637106186,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018288004558623426,
      "loss": 0.068,
      "step": 4057
    },
    {
      "epoch": 1.894049008168028,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018287183423064155,
      "loss": 0.0816,
      "step": 4058
    },
    {
      "epoch": 1.8945157526254377,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018286362109072628,
      "loss": 0.0808,
      "step": 4059
    },
    {
      "epoch": 1.894982497082847,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001828554061666653,
      "loss": 0.0911,
      "step": 4060
    },
    {
      "epoch": 1.8954492415402568,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001828471894586355,
      "loss": 0.0783,
      "step": 4061
    },
    {
      "epoch": 1.8959159859976662,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018283897096681381,
      "loss": 0.0902,
      "step": 4062
    },
    {
      "epoch": 1.896382730455076,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018283075069137715,
      "loss": 0.1027,
      "step": 4063
    },
    {
      "epoch": 1.8968494749124853,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018282252863250254,
      "loss": 0.0606,
      "step": 4064
    },
    {
      "epoch": 1.8973162193698951,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018281430479036694,
      "loss": 0.0868,
      "step": 4065
    },
    {
      "epoch": 1.8977829638273045,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018280607916514754,
      "loss": 0.0761,
      "step": 4066
    },
    {
      "epoch": 1.898249708284714,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018279785175702136,
      "loss": 0.0858,
      "step": 4067
    },
    {
      "epoch": 1.8987164527421236,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018278962256616557,
      "loss": 0.0881,
      "step": 4068
    },
    {
      "epoch": 1.8991831971995332,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018278139159275732,
      "loss": 0.0749,
      "step": 4069
    },
    {
      "epoch": 1.8996499416569428,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018277315883697388,
      "loss": 0.1185,
      "step": 4070
    },
    {
      "epoch": 1.9001166861143524,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018276492429899251,
      "loss": 0.1023,
      "step": 4071
    },
    {
      "epoch": 1.900583430571762,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018275668797899046,
      "loss": 0.1018,
      "step": 4072
    },
    {
      "epoch": 1.9010501750291715,
      "grad_norm": 1.140625,
      "learning_rate": 0.0001827484498771451,
      "loss": 0.1336,
      "step": 4073
    },
    {
      "epoch": 1.901516919486581,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001827402099936338,
      "loss": 0.1202,
      "step": 4074
    },
    {
      "epoch": 1.9019836639439907,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018273196832863402,
      "loss": 0.084,
      "step": 4075
    },
    {
      "epoch": 1.9024504084014002,
      "grad_norm": 0.796875,
      "learning_rate": 0.00018272372488232314,
      "loss": 0.0699,
      "step": 4076
    },
    {
      "epoch": 1.9029171528588098,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018271547965487868,
      "loss": 0.0838,
      "step": 4077
    },
    {
      "epoch": 1.9033838973162194,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018270723264647817,
      "loss": 0.0821,
      "step": 4078
    },
    {
      "epoch": 1.903850641773629,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00018269898385729917,
      "loss": 0.1036,
      "step": 4079
    },
    {
      "epoch": 1.9043173862310385,
      "grad_norm": 1.234375,
      "learning_rate": 0.0001826907332875193,
      "loss": 0.1143,
      "step": 4080
    },
    {
      "epoch": 1.904784130688448,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018268248093731621,
      "loss": 0.0667,
      "step": 4081
    },
    {
      "epoch": 1.9052508751458577,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00018267422680686755,
      "loss": 0.0753,
      "step": 4082
    },
    {
      "epoch": 1.9057176196032672,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018266597089635104,
      "loss": 0.1102,
      "step": 4083
    },
    {
      "epoch": 1.9061843640606768,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001826577132059445,
      "loss": 0.078,
      "step": 4084
    },
    {
      "epoch": 1.9066511085180864,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018264945373582564,
      "loss": 0.0794,
      "step": 4085
    },
    {
      "epoch": 1.907117852975496,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018264119248617236,
      "loss": 0.0902,
      "step": 4086
    },
    {
      "epoch": 1.9075845974329055,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018263292945716253,
      "loss": 0.089,
      "step": 4087
    },
    {
      "epoch": 1.9080513418903151,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018262466464897403,
      "loss": 0.0848,
      "step": 4088
    },
    {
      "epoch": 1.9085180863477245,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018261639806178482,
      "loss": 0.1133,
      "step": 4089
    },
    {
      "epoch": 1.9089848308051343,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001826081296957729,
      "loss": 0.103,
      "step": 4090
    },
    {
      "epoch": 1.9094515752625436,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001825998595511163,
      "loss": 0.0935,
      "step": 4091
    },
    {
      "epoch": 1.9099183197199534,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018259158762799307,
      "loss": 0.1049,
      "step": 4092
    },
    {
      "epoch": 1.9103850641773628,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018258331392658131,
      "loss": 0.0908,
      "step": 4093
    },
    {
      "epoch": 1.9108518086347726,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001825750384470592,
      "loss": 0.0798,
      "step": 4094
    },
    {
      "epoch": 1.911318553092182,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018256676118960486,
      "loss": 0.0939,
      "step": 4095
    },
    {
      "epoch": 1.9117852975495917,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001825584821543966,
      "loss": 0.0742,
      "step": 4096
    },
    {
      "epoch": 1.912252042007001,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018255020134161256,
      "loss": 0.0862,
      "step": 4097
    },
    {
      "epoch": 1.9127187864644108,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018254191875143112,
      "loss": 0.1023,
      "step": 4098
    },
    {
      "epoch": 1.9131855309218202,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018253363438403057,
      "loss": 0.1377,
      "step": 4099
    },
    {
      "epoch": 1.91365227537923,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00018252534823958932,
      "loss": 0.0719,
      "step": 4100
    },
    {
      "epoch": 1.9141190198366393,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018251706031828575,
      "loss": 0.0723,
      "step": 4101
    },
    {
      "epoch": 1.9145857642940491,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018250877062029832,
      "loss": 0.0995,
      "step": 4102
    },
    {
      "epoch": 1.9150525087514585,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018250047914580552,
      "loss": 0.0839,
      "step": 4103
    },
    {
      "epoch": 1.9155192532088683,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018249218589498587,
      "loss": 0.1177,
      "step": 4104
    },
    {
      "epoch": 1.9159859976662776,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018248389086801789,
      "loss": 0.0751,
      "step": 4105
    },
    {
      "epoch": 1.9164527421236874,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018247559406508028,
      "loss": 0.1018,
      "step": 4106
    },
    {
      "epoch": 1.9169194865810968,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001824672954863516,
      "loss": 0.0562,
      "step": 4107
    },
    {
      "epoch": 1.9173862310385066,
      "grad_norm": 0.8125,
      "learning_rate": 0.00018245899513201052,
      "loss": 0.0697,
      "step": 4108
    },
    {
      "epoch": 1.917852975495916,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018245069300223583,
      "loss": 0.0764,
      "step": 4109
    },
    {
      "epoch": 1.9183197199533255,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001824423890972062,
      "loss": 0.0572,
      "step": 4110
    },
    {
      "epoch": 1.918786464410735,
      "grad_norm": 1.125,
      "learning_rate": 0.0001824340834171005,
      "loss": 0.104,
      "step": 4111
    },
    {
      "epoch": 1.9192532088681447,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018242577596209748,
      "loss": 0.0671,
      "step": 4112
    },
    {
      "epoch": 1.9197199533255542,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018241746673237607,
      "loss": 0.0794,
      "step": 4113
    },
    {
      "epoch": 1.9201866977829638,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018240915572811516,
      "loss": 0.0773,
      "step": 4114
    },
    {
      "epoch": 1.9206534422403734,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018240084294949372,
      "loss": 0.0691,
      "step": 4115
    },
    {
      "epoch": 1.921120186697783,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018239252839669066,
      "loss": 0.1084,
      "step": 4116
    },
    {
      "epoch": 1.9215869311551925,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018238421206988504,
      "loss": 0.0819,
      "step": 4117
    },
    {
      "epoch": 1.922053675612602,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018237589396925596,
      "loss": 0.0693,
      "step": 4118
    },
    {
      "epoch": 1.9225204200700117,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018236757409498247,
      "loss": 0.0651,
      "step": 4119
    },
    {
      "epoch": 1.9229871645274212,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018235925244724374,
      "loss": 0.1047,
      "step": 4120
    },
    {
      "epoch": 1.9234539089848308,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001823509290262189,
      "loss": 0.0896,
      "step": 4121
    },
    {
      "epoch": 1.9239206534422404,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018234260383208722,
      "loss": 0.1091,
      "step": 4122
    },
    {
      "epoch": 1.92438739789965,
      "grad_norm": 1.0,
      "learning_rate": 0.0001823342768650279,
      "loss": 0.1292,
      "step": 4123
    },
    {
      "epoch": 1.9248541423570595,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00018232594812522026,
      "loss": 0.0618,
      "step": 4124
    },
    {
      "epoch": 1.9253208868144691,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001823176176128436,
      "loss": 0.0719,
      "step": 4125
    },
    {
      "epoch": 1.9257876312718787,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00018230928532807732,
      "loss": 0.0695,
      "step": 4126
    },
    {
      "epoch": 1.9262543757292883,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018230095127110078,
      "loss": 0.0776,
      "step": 4127
    },
    {
      "epoch": 1.9267211201866978,
      "grad_norm": 1.046875,
      "learning_rate": 0.00018229261544209344,
      "loss": 0.0894,
      "step": 4128
    },
    {
      "epoch": 1.9271878646441074,
      "grad_norm": 1.1640625,
      "learning_rate": 0.0001822842778412348,
      "loss": 0.0983,
      "step": 4129
    },
    {
      "epoch": 1.9276546091015168,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018227593846870436,
      "loss": 0.0768,
      "step": 4130
    },
    {
      "epoch": 1.9281213535589266,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00018226759732468167,
      "loss": 0.0697,
      "step": 4131
    },
    {
      "epoch": 1.928588098016336,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018225925440934634,
      "loss": 0.0761,
      "step": 4132
    },
    {
      "epoch": 1.9290548424737457,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000182250909722878,
      "loss": 0.0693,
      "step": 4133
    },
    {
      "epoch": 1.929521586931155,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001822425632654563,
      "loss": 0.072,
      "step": 4134
    },
    {
      "epoch": 1.9299883313885648,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00018223421503726095,
      "loss": 0.0908,
      "step": 4135
    },
    {
      "epoch": 1.9304550758459742,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018222586503847173,
      "loss": 0.0725,
      "step": 4136
    },
    {
      "epoch": 1.930921820303384,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001822175132692684,
      "loss": 0.0838,
      "step": 4137
    },
    {
      "epoch": 1.9313885647607933,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001822091597298308,
      "loss": 0.0739,
      "step": 4138
    },
    {
      "epoch": 1.9318553092182031,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018220080442033878,
      "loss": 0.0837,
      "step": 4139
    },
    {
      "epoch": 1.9323220536756125,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018219244734097223,
      "loss": 0.0847,
      "step": 4140
    },
    {
      "epoch": 1.9327887981330223,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001821840884919111,
      "loss": 0.0651,
      "step": 4141
    },
    {
      "epoch": 1.9332555425904316,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00018217572787333533,
      "loss": 0.0774,
      "step": 4142
    },
    {
      "epoch": 1.9337222870478414,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018216736548542503,
      "loss": 0.0731,
      "step": 4143
    },
    {
      "epoch": 1.9341890315052508,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001821590013283601,
      "loss": 0.0934,
      "step": 4144
    },
    {
      "epoch": 1.9346557759626606,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018215063540232075,
      "loss": 0.0751,
      "step": 4145
    },
    {
      "epoch": 1.93512252042007,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001821422677074871,
      "loss": 0.0859,
      "step": 4146
    },
    {
      "epoch": 1.9355892648774797,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018213389824403925,
      "loss": 0.0798,
      "step": 4147
    },
    {
      "epoch": 1.936056009334889,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00018212552701215746,
      "loss": 0.0547,
      "step": 4148
    },
    {
      "epoch": 1.9365227537922989,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018211715401202194,
      "loss": 0.1104,
      "step": 4149
    },
    {
      "epoch": 1.9369894982497082,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00018210877924381303,
      "loss": 0.0684,
      "step": 4150
    },
    {
      "epoch": 1.9374562427071178,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00018210040270771094,
      "loss": 0.1049,
      "step": 4151
    },
    {
      "epoch": 1.9379229871645274,
      "grad_norm": 1.171875,
      "learning_rate": 0.00018209202440389613,
      "loss": 0.1035,
      "step": 4152
    },
    {
      "epoch": 1.938389731621937,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018208364433254892,
      "loss": 0.0814,
      "step": 4153
    },
    {
      "epoch": 1.9388564760793465,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018207526249384978,
      "loss": 0.1017,
      "step": 4154
    },
    {
      "epoch": 1.939323220536756,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001820668788879792,
      "loss": 0.0965,
      "step": 4155
    },
    {
      "epoch": 1.9397899649941657,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018205849351511763,
      "loss": 0.0881,
      "step": 4156
    },
    {
      "epoch": 1.9402567094515752,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018205010637544565,
      "loss": 0.091,
      "step": 4157
    },
    {
      "epoch": 1.9407234539089848,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018204171746914386,
      "loss": 0.0787,
      "step": 4158
    },
    {
      "epoch": 1.9411901983663944,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018203332679639285,
      "loss": 0.0678,
      "step": 4159
    },
    {
      "epoch": 1.941656942823804,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018202493435737335,
      "loss": 0.0941,
      "step": 4160
    },
    {
      "epoch": 1.9421236872812135,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00018201654015226595,
      "loss": 0.0945,
      "step": 4161
    },
    {
      "epoch": 1.9425904317386231,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018200814418125146,
      "loss": 0.0889,
      "step": 4162
    },
    {
      "epoch": 1.9430571761960327,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018199974644451066,
      "loss": 0.0574,
      "step": 4163
    },
    {
      "epoch": 1.9435239206534423,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001819913469422243,
      "loss": 0.0929,
      "step": 4164
    },
    {
      "epoch": 1.9439906651108518,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018198294567457328,
      "loss": 0.0886,
      "step": 4165
    },
    {
      "epoch": 1.9444574095682614,
      "grad_norm": 1.125,
      "learning_rate": 0.00018197454264173849,
      "loss": 0.1094,
      "step": 4166
    },
    {
      "epoch": 1.944924154025671,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00018196613784390083,
      "loss": 0.079,
      "step": 4167
    },
    {
      "epoch": 1.9453908984830806,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018195773128124127,
      "loss": 0.1143,
      "step": 4168
    },
    {
      "epoch": 1.9458576429404901,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018194932295394084,
      "loss": 0.0636,
      "step": 4169
    },
    {
      "epoch": 1.9463243873978997,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018194091286218055,
      "loss": 0.0794,
      "step": 4170
    },
    {
      "epoch": 1.946791131855309,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018193250100614149,
      "loss": 0.0788,
      "step": 4171
    },
    {
      "epoch": 1.9472578763127188,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018192408738600482,
      "loss": 0.0931,
      "step": 4172
    },
    {
      "epoch": 1.9477246207701282,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001819156720019516,
      "loss": 0.0968,
      "step": 4173
    },
    {
      "epoch": 1.948191365227538,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00018190725485416311,
      "loss": 0.101,
      "step": 4174
    },
    {
      "epoch": 1.9486581096849473,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001818988359428205,
      "loss": 0.1104,
      "step": 4175
    },
    {
      "epoch": 1.9491248541423571,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00018189041526810512,
      "loss": 0.0716,
      "step": 4176
    },
    {
      "epoch": 1.9495915985997665,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018188199283019824,
      "loss": 0.0836,
      "step": 4177
    },
    {
      "epoch": 1.9500583430571763,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018187356862928123,
      "loss": 0.0889,
      "step": 4178
    },
    {
      "epoch": 1.9505250875145856,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018186514266553539,
      "loss": 0.0851,
      "step": 4179
    },
    {
      "epoch": 1.9509918319719954,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00018185671493914222,
      "loss": 0.1038,
      "step": 4180
    },
    {
      "epoch": 1.9514585764294048,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00018184828545028317,
      "loss": 0.0971,
      "step": 4181
    },
    {
      "epoch": 1.9519253208868146,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001818398541991397,
      "loss": 0.0587,
      "step": 4182
    },
    {
      "epoch": 1.952392065344224,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001818314211858934,
      "loss": 0.0897,
      "step": 4183
    },
    {
      "epoch": 1.9528588098016337,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018182298641072577,
      "loss": 0.1065,
      "step": 4184
    },
    {
      "epoch": 1.953325554259043,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018181454987381847,
      "loss": 0.09,
      "step": 4185
    },
    {
      "epoch": 1.9537922987164529,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00018180611157535314,
      "loss": 0.0958,
      "step": 4186
    },
    {
      "epoch": 1.9542590431738622,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018179767151551147,
      "loss": 0.0759,
      "step": 4187
    },
    {
      "epoch": 1.954725787631272,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018178922969447515,
      "loss": 0.095,
      "step": 4188
    },
    {
      "epoch": 1.9551925320886814,
      "grad_norm": 1.0859375,
      "learning_rate": 0.000181780786112426,
      "loss": 0.0901,
      "step": 4189
    },
    {
      "epoch": 1.9556592765460912,
      "grad_norm": 0.875,
      "learning_rate": 0.0001817723407695458,
      "loss": 0.0602,
      "step": 4190
    },
    {
      "epoch": 1.9561260210035005,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00018176389366601634,
      "loss": 0.0799,
      "step": 4191
    },
    {
      "epoch": 1.95659276546091,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018175544480201951,
      "loss": 0.093,
      "step": 4192
    },
    {
      "epoch": 1.9570595099183197,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018174699417773728,
      "loss": 0.1169,
      "step": 4193
    },
    {
      "epoch": 1.9575262543757292,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00018173854179335156,
      "loss": 0.0505,
      "step": 4194
    },
    {
      "epoch": 1.9579929988331388,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018173008764904436,
      "loss": 0.0881,
      "step": 4195
    },
    {
      "epoch": 1.9584597432905484,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018172163174499768,
      "loss": 0.0902,
      "step": 4196
    },
    {
      "epoch": 1.958926487747958,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001817131740813936,
      "loss": 0.0953,
      "step": 4197
    },
    {
      "epoch": 1.9593932322053675,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001817047146584142,
      "loss": 0.0762,
      "step": 4198
    },
    {
      "epoch": 1.9598599766627771,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018169625347624167,
      "loss": 0.0983,
      "step": 4199
    },
    {
      "epoch": 1.9603267211201867,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001816877905350581,
      "loss": 0.0838,
      "step": 4200
    },
    {
      "epoch": 1.9607934655775963,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018167932583504584,
      "loss": 0.092,
      "step": 4201
    },
    {
      "epoch": 1.9612602100350058,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018167085937638704,
      "loss": 0.0689,
      "step": 4202
    },
    {
      "epoch": 1.9617269544924154,
      "grad_norm": 1.046875,
      "learning_rate": 0.000181662391159264,
      "loss": 0.0765,
      "step": 4203
    },
    {
      "epoch": 1.962193698949825,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001816539211838591,
      "loss": 0.0768,
      "step": 4204
    },
    {
      "epoch": 1.9626604434072346,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001816454494503547,
      "loss": 0.0654,
      "step": 4205
    },
    {
      "epoch": 1.9631271878646441,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018163697595893313,
      "loss": 0.0901,
      "step": 4206
    },
    {
      "epoch": 1.9635939323220537,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018162850070977692,
      "loss": 0.0733,
      "step": 4207
    },
    {
      "epoch": 1.9640606767794633,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018162002370306852,
      "loss": 0.0774,
      "step": 4208
    },
    {
      "epoch": 1.9645274212368729,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018161154493899045,
      "loss": 0.1083,
      "step": 4209
    },
    {
      "epoch": 1.9649941656942824,
      "grad_norm": 1.140625,
      "learning_rate": 0.00018160306441772527,
      "loss": 0.1082,
      "step": 4210
    },
    {
      "epoch": 1.965460910151692,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00018159458213945557,
      "loss": 0.0535,
      "step": 4211
    },
    {
      "epoch": 1.9659276546091016,
      "grad_norm": 1.234375,
      "learning_rate": 0.000181586098104364,
      "loss": 0.1158,
      "step": 4212
    },
    {
      "epoch": 1.9663943990665111,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001815776123126332,
      "loss": 0.0836,
      "step": 4213
    },
    {
      "epoch": 1.9668611435239205,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001815691247644459,
      "loss": 0.0827,
      "step": 4214
    },
    {
      "epoch": 1.9673278879813303,
      "grad_norm": 0.828125,
      "learning_rate": 0.00018156063545998486,
      "loss": 0.0723,
      "step": 4215
    },
    {
      "epoch": 1.9677946324387396,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001815521443994328,
      "loss": 0.0845,
      "step": 4216
    },
    {
      "epoch": 1.9682613768961494,
      "grad_norm": 1.015625,
      "learning_rate": 0.00018154365158297263,
      "loss": 0.0961,
      "step": 4217
    },
    {
      "epoch": 1.9687281213535588,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018153515701078713,
      "loss": 0.09,
      "step": 4218
    },
    {
      "epoch": 1.9691948658109686,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018152666068305926,
      "loss": 0.0853,
      "step": 4219
    },
    {
      "epoch": 1.969661610268378,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00018151816259997193,
      "loss": 0.0621,
      "step": 4220
    },
    {
      "epoch": 1.9701283547257877,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018150966276170812,
      "loss": 0.0702,
      "step": 4221
    },
    {
      "epoch": 1.970595099183197,
      "grad_norm": 0.734375,
      "learning_rate": 0.00018150116116845083,
      "loss": 0.0612,
      "step": 4222
    },
    {
      "epoch": 1.9710618436406069,
      "grad_norm": 1.28125,
      "learning_rate": 0.0001814926578203831,
      "loss": 0.0979,
      "step": 4223
    },
    {
      "epoch": 1.9715285880980162,
      "grad_norm": 0.9609375,
      "learning_rate": 0.000181484152717688,
      "loss": 0.0918,
      "step": 4224
    },
    {
      "epoch": 1.971995332555426,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00018147564586054875,
      "loss": 0.0999,
      "step": 4225
    },
    {
      "epoch": 1.9724620770128354,
      "grad_norm": 0.84375,
      "learning_rate": 0.00018146713724914842,
      "loss": 0.0872,
      "step": 4226
    },
    {
      "epoch": 1.9729288214702452,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001814586268836702,
      "loss": 0.0709,
      "step": 4227
    },
    {
      "epoch": 1.9733955659276545,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001814501147642974,
      "loss": 0.082,
      "step": 4228
    },
    {
      "epoch": 1.9738623103850643,
      "grad_norm": 1.2734375,
      "learning_rate": 0.00018144160089121324,
      "loss": 0.0824,
      "step": 4229
    },
    {
      "epoch": 1.9743290548424737,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018143308526460107,
      "loss": 0.0707,
      "step": 4230
    },
    {
      "epoch": 1.9747957992998835,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00018142456788464425,
      "loss": 0.1082,
      "step": 4231
    },
    {
      "epoch": 1.9752625437572928,
      "grad_norm": 1.0703125,
      "learning_rate": 0.0001814160487515261,
      "loss": 0.0805,
      "step": 4232
    },
    {
      "epoch": 1.9757292882147024,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001814075278654301,
      "loss": 0.0834,
      "step": 4233
    },
    {
      "epoch": 1.976196032672112,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001813990052265397,
      "loss": 0.0764,
      "step": 4234
    },
    {
      "epoch": 1.9766627771295215,
      "grad_norm": 0.875,
      "learning_rate": 0.0001813904808350384,
      "loss": 0.0634,
      "step": 4235
    },
    {
      "epoch": 1.9771295215869311,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018138195469110974,
      "loss": 0.0815,
      "step": 4236
    },
    {
      "epoch": 1.9775962660443407,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001813734267949373,
      "loss": 0.1021,
      "step": 4237
    },
    {
      "epoch": 1.9780630105017503,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001813648971467047,
      "loss": 0.075,
      "step": 4238
    },
    {
      "epoch": 1.9785297549591598,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00018135636574659558,
      "loss": 0.0619,
      "step": 4239
    },
    {
      "epoch": 1.9789964994165694,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018134783259479367,
      "loss": 0.1008,
      "step": 4240
    },
    {
      "epoch": 1.979463243873979,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00018133929769148264,
      "loss": 0.0957,
      "step": 4241
    },
    {
      "epoch": 1.9799299883313886,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00018133076103684628,
      "loss": 0.0961,
      "step": 4242
    },
    {
      "epoch": 1.9803967327887981,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018132222263106843,
      "loss": 0.077,
      "step": 4243
    },
    {
      "epoch": 1.9808634772462077,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018131368247433286,
      "loss": 0.0892,
      "step": 4244
    },
    {
      "epoch": 1.9813302217036173,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001813051405668235,
      "loss": 0.0996,
      "step": 4245
    },
    {
      "epoch": 1.9817969661610269,
      "grad_norm": 0.953125,
      "learning_rate": 0.00018129659690872425,
      "loss": 0.1232,
      "step": 4246
    },
    {
      "epoch": 1.9822637106184364,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018128805150021906,
      "loss": 0.088,
      "step": 4247
    },
    {
      "epoch": 1.982730455075846,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00018127950434149197,
      "loss": 0.0817,
      "step": 4248
    },
    {
      "epoch": 1.9831971995332556,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018127095543272694,
      "loss": 0.0796,
      "step": 4249
    },
    {
      "epoch": 1.9836639439906651,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018126240477410804,
      "loss": 0.0885,
      "step": 4250
    },
    {
      "epoch": 1.9841306884480747,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018125385236581945,
      "loss": 0.0828,
      "step": 4251
    },
    {
      "epoch": 1.9845974329054843,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018124529820804524,
      "loss": 0.087,
      "step": 4252
    },
    {
      "epoch": 1.9850641773628939,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018123674230096964,
      "loss": 0.0706,
      "step": 4253
    },
    {
      "epoch": 1.9855309218203034,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00018122818464477684,
      "loss": 0.0871,
      "step": 4254
    },
    {
      "epoch": 1.9859976662777128,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001812196252396511,
      "loss": 0.0876,
      "step": 4255
    },
    {
      "epoch": 1.9864644107351226,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001812110640857767,
      "loss": 0.0693,
      "step": 4256
    },
    {
      "epoch": 1.986931155192532,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00018120250118333799,
      "loss": 0.0652,
      "step": 4257
    },
    {
      "epoch": 1.9873978996499417,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00018119393653251935,
      "loss": 0.0922,
      "step": 4258
    },
    {
      "epoch": 1.987864644107351,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018118537013350517,
      "loss": 0.1167,
      "step": 4259
    },
    {
      "epoch": 1.9883313885647609,
      "grad_norm": 1.0,
      "learning_rate": 0.00018117680198647987,
      "loss": 0.0897,
      "step": 4260
    },
    {
      "epoch": 1.9887981330221702,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018116823209162798,
      "loss": 0.0802,
      "step": 4261
    },
    {
      "epoch": 1.98926487747958,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00018115966044913396,
      "loss": 0.0863,
      "step": 4262
    },
    {
      "epoch": 1.9897316219369894,
      "grad_norm": 0.859375,
      "learning_rate": 0.00018115108705918243,
      "loss": 0.0687,
      "step": 4263
    },
    {
      "epoch": 1.9901983663943992,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018114251192195794,
      "loss": 0.0796,
      "step": 4264
    },
    {
      "epoch": 1.9906651108518085,
      "grad_norm": 0.984375,
      "learning_rate": 0.00018113393503764512,
      "loss": 0.0867,
      "step": 4265
    },
    {
      "epoch": 1.9911318553092183,
      "grad_norm": 0.765625,
      "learning_rate": 0.00018112535640642873,
      "loss": 0.0629,
      "step": 4266
    },
    {
      "epoch": 1.9915985997666277,
      "grad_norm": 1.25,
      "learning_rate": 0.00018111677602849336,
      "loss": 0.1159,
      "step": 4267
    },
    {
      "epoch": 1.9920653442240375,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018110819390402378,
      "loss": 0.0746,
      "step": 4268
    },
    {
      "epoch": 1.9925320886814468,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00018109961003320483,
      "loss": 0.0988,
      "step": 4269
    },
    {
      "epoch": 1.9929988331388566,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001810910244162213,
      "loss": 0.0916,
      "step": 4270
    },
    {
      "epoch": 1.993465577596266,
      "grad_norm": 1.015625,
      "learning_rate": 0.000181082437053258,
      "loss": 0.1245,
      "step": 4271
    },
    {
      "epoch": 1.9939323220536758,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001810738479444999,
      "loss": 0.0735,
      "step": 4272
    },
    {
      "epoch": 1.9943990665110851,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001810652570901319,
      "loss": 0.0764,
      "step": 4273
    },
    {
      "epoch": 1.994865810968495,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018105666449033897,
      "loss": 0.0993,
      "step": 4274
    },
    {
      "epoch": 1.9953325554259043,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018104807014530613,
      "loss": 0.07,
      "step": 4275
    },
    {
      "epoch": 1.9957992998833138,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001810394740552184,
      "loss": 0.084,
      "step": 4276
    },
    {
      "epoch": 1.9962660443407234,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018103087622026086,
      "loss": 0.086,
      "step": 4277
    },
    {
      "epoch": 1.996732788798133,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018102227664061867,
      "loss": 0.1095,
      "step": 4278
    },
    {
      "epoch": 1.9971995332555426,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018101367531647697,
      "loss": 0.0952,
      "step": 4279
    },
    {
      "epoch": 1.9976662777129521,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00018100507224802097,
      "loss": 0.0988,
      "step": 4280
    },
    {
      "epoch": 1.9981330221703617,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00018099646743543584,
      "loss": 0.0571,
      "step": 4281
    },
    {
      "epoch": 1.9985997666277713,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00018098786087890697,
      "loss": 0.0668,
      "step": 4282
    },
    {
      "epoch": 1.9990665110851809,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018097925257861954,
      "loss": 0.0848,
      "step": 4283
    },
    {
      "epoch": 1.9995332555425904,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018097064253475898,
      "loss": 0.059,
      "step": 4284
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.8125,
      "learning_rate": 0.00018096203074751066,
      "loss": 0.0623,
      "step": 4285
    },
    {
      "epoch": 2.0004667444574094,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00018095341721705995,
      "loss": 0.052,
      "step": 4286
    },
    {
      "epoch": 2.000933488914819,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001809448019435924,
      "loss": 0.0554,
      "step": 4287
    },
    {
      "epoch": 2.0014002333722285,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018093618492729337,
      "loss": 0.0604,
      "step": 4288
    },
    {
      "epoch": 2.0014002333722285,
      "eval_loss": 1.522891879081726,
      "eval_runtime": 94.7932,
      "eval_samples_per_second": 19.031,
      "eval_steps_per_second": 2.384,
      "step": 4288
    },
    {
      "epoch": 2.0018669778296383,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018092756616834853,
      "loss": 0.0776,
      "step": 4289
    },
    {
      "epoch": 2.0023337222870476,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001809189456669434,
      "loss": 0.0731,
      "step": 4290
    },
    {
      "epoch": 2.0028004667444574,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001809103234232636,
      "loss": 0.0816,
      "step": 4291
    },
    {
      "epoch": 2.003267211201867,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018090169943749476,
      "loss": 0.0853,
      "step": 4292
    },
    {
      "epoch": 2.0037339556592766,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00018089307370982254,
      "loss": 0.0745,
      "step": 4293
    },
    {
      "epoch": 2.004200700116686,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018088444624043273,
      "loss": 0.0552,
      "step": 4294
    },
    {
      "epoch": 2.0046674445740957,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018087581702951101,
      "loss": 0.0632,
      "step": 4295
    },
    {
      "epoch": 2.005134189031505,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018086718607724324,
      "loss": 0.0516,
      "step": 4296
    },
    {
      "epoch": 2.005600933488915,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018085855338381522,
      "loss": 0.0699,
      "step": 4297
    },
    {
      "epoch": 2.0060676779463242,
      "grad_norm": 0.640625,
      "learning_rate": 0.00018084991894941286,
      "loss": 0.0563,
      "step": 4298
    },
    {
      "epoch": 2.006534422403734,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00018084128277422203,
      "loss": 0.0566,
      "step": 4299
    },
    {
      "epoch": 2.0070011668611434,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001808326448584287,
      "loss": 0.0758,
      "step": 4300
    },
    {
      "epoch": 2.007467911318553,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018082400520221878,
      "loss": 0.0844,
      "step": 4301
    },
    {
      "epoch": 2.0079346557759625,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018081536380577839,
      "loss": 0.084,
      "step": 4302
    },
    {
      "epoch": 2.0084014002333723,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00018080672066929357,
      "loss": 0.0732,
      "step": 4303
    },
    {
      "epoch": 2.0088681446907817,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00018079807579295037,
      "loss": 0.0606,
      "step": 4304
    },
    {
      "epoch": 2.0093348891481915,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018078942917693495,
      "loss": 0.0716,
      "step": 4305
    },
    {
      "epoch": 2.009801633605601,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001807807808214335,
      "loss": 0.063,
      "step": 4306
    },
    {
      "epoch": 2.0102683780630106,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001807721307266322,
      "loss": 0.0634,
      "step": 4307
    },
    {
      "epoch": 2.01073512252042,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00018076347889271733,
      "loss": 0.0626,
      "step": 4308
    },
    {
      "epoch": 2.0112018669778298,
      "grad_norm": 1.1875,
      "learning_rate": 0.00018075482531987512,
      "loss": 0.0827,
      "step": 4309
    },
    {
      "epoch": 2.011668611435239,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018074617000829192,
      "loss": 0.086,
      "step": 4310
    },
    {
      "epoch": 2.012135355892649,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001807375129581541,
      "loss": 0.0842,
      "step": 4311
    },
    {
      "epoch": 2.0126021003500583,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018072885416964805,
      "loss": 0.0728,
      "step": 4312
    },
    {
      "epoch": 2.013068844807468,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001807201936429602,
      "loss": 0.059,
      "step": 4313
    },
    {
      "epoch": 2.0135355892648774,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000180711531378277,
      "loss": 0.0724,
      "step": 4314
    },
    {
      "epoch": 2.014002333722287,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018070286737578503,
      "loss": 0.071,
      "step": 4315
    },
    {
      "epoch": 2.0144690781796966,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018069420163567074,
      "loss": 0.0494,
      "step": 4316
    },
    {
      "epoch": 2.0149358226371064,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018068553415812077,
      "loss": 0.0641,
      "step": 4317
    },
    {
      "epoch": 2.0154025670945157,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00018067686494332175,
      "loss": 0.08,
      "step": 4318
    },
    {
      "epoch": 2.0158693115519255,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018066819399146028,
      "loss": 0.0682,
      "step": 4319
    },
    {
      "epoch": 2.016336056009335,
      "grad_norm": 1.125,
      "learning_rate": 0.00018065952130272308,
      "loss": 0.0832,
      "step": 4320
    },
    {
      "epoch": 2.0168028004667446,
      "grad_norm": 1.03125,
      "learning_rate": 0.00018065084687729694,
      "loss": 0.0583,
      "step": 4321
    },
    {
      "epoch": 2.017269544924154,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00018064217071536853,
      "loss": 0.0675,
      "step": 4322
    },
    {
      "epoch": 2.017736289381564,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00018063349281712476,
      "loss": 0.0584,
      "step": 4323
    },
    {
      "epoch": 2.018203033838973,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001806248131827524,
      "loss": 0.0732,
      "step": 4324
    },
    {
      "epoch": 2.018669778296383,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00018061613181243836,
      "loss": 0.0533,
      "step": 4325
    },
    {
      "epoch": 2.0191365227537923,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018060744870636956,
      "loss": 0.0696,
      "step": 4326
    },
    {
      "epoch": 2.0196032672112016,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018059876386473294,
      "loss": 0.068,
      "step": 4327
    },
    {
      "epoch": 2.0200700116686114,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018059007728771554,
      "loss": 0.0796,
      "step": 4328
    },
    {
      "epoch": 2.020536756126021,
      "grad_norm": 0.75,
      "learning_rate": 0.0001805813889755043,
      "loss": 0.0561,
      "step": 4329
    },
    {
      "epoch": 2.0210035005834306,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001805726989282864,
      "loss": 0.0654,
      "step": 4330
    },
    {
      "epoch": 2.02147024504084,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001805640071462489,
      "loss": 0.0584,
      "step": 4331
    },
    {
      "epoch": 2.0219369894982497,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018055531362957893,
      "loss": 0.0767,
      "step": 4332
    },
    {
      "epoch": 2.022403733955659,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00018054661837846366,
      "loss": 0.1072,
      "step": 4333
    },
    {
      "epoch": 2.022870478413069,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00018053792139309036,
      "loss": 0.0618,
      "step": 4334
    },
    {
      "epoch": 2.0233372228704782,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00018052922267364622,
      "loss": 0.0615,
      "step": 4335
    },
    {
      "epoch": 2.023803967327888,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001805205222203186,
      "loss": 0.0671,
      "step": 4336
    },
    {
      "epoch": 2.0242707117852974,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00018051182003329477,
      "loss": 0.0593,
      "step": 4337
    },
    {
      "epoch": 2.024737456242707,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001805031161127621,
      "loss": 0.0884,
      "step": 4338
    },
    {
      "epoch": 2.0252042007001165,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018049441045890808,
      "loss": 0.0682,
      "step": 4339
    },
    {
      "epoch": 2.0256709451575263,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00018048570307192008,
      "loss": 0.1023,
      "step": 4340
    },
    {
      "epoch": 2.0261376896149357,
      "grad_norm": 0.875,
      "learning_rate": 0.00018047699395198558,
      "loss": 0.0566,
      "step": 4341
    },
    {
      "epoch": 2.0266044340723455,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001804682830992921,
      "loss": 0.0865,
      "step": 4342
    },
    {
      "epoch": 2.027071178529755,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001804595705140272,
      "loss": 0.0753,
      "step": 4343
    },
    {
      "epoch": 2.0275379229871646,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018045085619637852,
      "loss": 0.0643,
      "step": 4344
    },
    {
      "epoch": 2.028004667444574,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001804421401465336,
      "loss": 0.0732,
      "step": 4345
    },
    {
      "epoch": 2.0284714119019838,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00018043342236468015,
      "loss": 0.0574,
      "step": 4346
    },
    {
      "epoch": 2.028938156359393,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001804247028510059,
      "loss": 0.0836,
      "step": 4347
    },
    {
      "epoch": 2.029404900816803,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00018041598160569854,
      "loss": 0.056,
      "step": 4348
    },
    {
      "epoch": 2.0298716452742123,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001804072586289459,
      "loss": 0.0625,
      "step": 4349
    },
    {
      "epoch": 2.030338389731622,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00018039853392093575,
      "loss": 0.0649,
      "step": 4350
    },
    {
      "epoch": 2.0308051341890314,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00018038980748185597,
      "loss": 0.0631,
      "step": 4351
    },
    {
      "epoch": 2.031271878646441,
      "grad_norm": 1.109375,
      "learning_rate": 0.00018038107931189442,
      "loss": 0.0648,
      "step": 4352
    },
    {
      "epoch": 2.0317386231038506,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00018037234941123908,
      "loss": 0.0654,
      "step": 4353
    },
    {
      "epoch": 2.0322053675612604,
      "grad_norm": 1.09375,
      "learning_rate": 0.00018036361778007786,
      "loss": 0.0791,
      "step": 4354
    },
    {
      "epoch": 2.0326721120186697,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001803548844185988,
      "loss": 0.0736,
      "step": 4355
    },
    {
      "epoch": 2.0331388564760795,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001803461493269899,
      "loss": 0.0625,
      "step": 4356
    },
    {
      "epoch": 2.033605600933489,
      "grad_norm": 0.9375,
      "learning_rate": 0.00018033741250543926,
      "loss": 0.0681,
      "step": 4357
    },
    {
      "epoch": 2.0340723453908987,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000180328673954135,
      "loss": 0.0689,
      "step": 4358
    },
    {
      "epoch": 2.034539089848308,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001803199336732653,
      "loss": 0.0518,
      "step": 4359
    },
    {
      "epoch": 2.035005834305718,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018031119166301828,
      "loss": 0.0589,
      "step": 4360
    },
    {
      "epoch": 2.035472578763127,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001803024479235822,
      "loss": 0.0689,
      "step": 4361
    },
    {
      "epoch": 2.035939323220537,
      "grad_norm": 0.890625,
      "learning_rate": 0.00018029370245514528,
      "loss": 0.0615,
      "step": 4362
    },
    {
      "epoch": 2.0364060676779463,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00018028495525789593,
      "loss": 0.0612,
      "step": 4363
    },
    {
      "epoch": 2.036872812135356,
      "grad_norm": 1.15625,
      "learning_rate": 0.00018027620633202237,
      "loss": 0.0575,
      "step": 4364
    },
    {
      "epoch": 2.0373395565927654,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00018026745567771302,
      "loss": 0.0756,
      "step": 4365
    },
    {
      "epoch": 2.0378063010501752,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018025870329515633,
      "loss": 0.0783,
      "step": 4366
    },
    {
      "epoch": 2.0382730455075846,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00018024994918454067,
      "loss": 0.0733,
      "step": 4367
    },
    {
      "epoch": 2.0387397899649944,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018024119334605456,
      "loss": 0.0764,
      "step": 4368
    },
    {
      "epoch": 2.0392065344224037,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00018023243577988655,
      "loss": 0.0627,
      "step": 4369
    },
    {
      "epoch": 2.039673278879813,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001802236764862252,
      "loss": 0.0738,
      "step": 4370
    },
    {
      "epoch": 2.040140023337223,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00018021491546525904,
      "loss": 0.0753,
      "step": 4371
    },
    {
      "epoch": 2.0406067677946322,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00018020615271717677,
      "loss": 0.0584,
      "step": 4372
    },
    {
      "epoch": 2.041073512252042,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00018019738824216704,
      "loss": 0.0709,
      "step": 4373
    },
    {
      "epoch": 2.0415402567094514,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00018018862204041854,
      "loss": 0.0764,
      "step": 4374
    },
    {
      "epoch": 2.042007001166861,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00018017985411212007,
      "loss": 0.071,
      "step": 4375
    },
    {
      "epoch": 2.0424737456242705,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018017108445746036,
      "loss": 0.0855,
      "step": 4376
    },
    {
      "epoch": 2.0429404900816803,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00018016231307662828,
      "loss": 0.0788,
      "step": 4377
    },
    {
      "epoch": 2.0434072345390897,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00018015353996981263,
      "loss": 0.0695,
      "step": 4378
    },
    {
      "epoch": 2.0438739789964995,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00018014476513720232,
      "loss": 0.0579,
      "step": 4379
    },
    {
      "epoch": 2.044340723453909,
      "grad_norm": 0.90625,
      "learning_rate": 0.00018013598857898633,
      "loss": 0.0821,
      "step": 4380
    },
    {
      "epoch": 2.0448074679113186,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00018012721029535356,
      "loss": 0.0692,
      "step": 4381
    },
    {
      "epoch": 2.045274212368728,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00018011843028649307,
      "loss": 0.095,
      "step": 4382
    },
    {
      "epoch": 2.0457409568261378,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001801096485525939,
      "loss": 0.0649,
      "step": 4383
    },
    {
      "epoch": 2.046207701283547,
      "grad_norm": 1.078125,
      "learning_rate": 0.00018010086509384507,
      "loss": 0.1107,
      "step": 4384
    },
    {
      "epoch": 2.046674445740957,
      "grad_norm": 1.203125,
      "learning_rate": 0.00018009207991043577,
      "loss": 0.0925,
      "step": 4385
    },
    {
      "epoch": 2.0471411901983663,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001800832930025551,
      "loss": 0.0677,
      "step": 4386
    },
    {
      "epoch": 2.047607934655776,
      "grad_norm": 1.0625,
      "learning_rate": 0.00018007450437039233,
      "loss": 0.0864,
      "step": 4387
    },
    {
      "epoch": 2.0480746791131854,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00018006571401413658,
      "loss": 0.0746,
      "step": 4388
    },
    {
      "epoch": 2.048541423570595,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001800569219339772,
      "loss": 0.0929,
      "step": 4389
    },
    {
      "epoch": 2.0490081680280046,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00018004812813010346,
      "loss": 0.0677,
      "step": 4390
    },
    {
      "epoch": 2.0494749124854144,
      "grad_norm": 0.625,
      "learning_rate": 0.00018003933260270474,
      "loss": 0.0574,
      "step": 4391
    },
    {
      "epoch": 2.0499416569428237,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00018003053535197032,
      "loss": 0.0456,
      "step": 4392
    },
    {
      "epoch": 2.0504084014002335,
      "grad_norm": 0.921875,
      "learning_rate": 0.00018002173637808973,
      "loss": 0.0551,
      "step": 4393
    },
    {
      "epoch": 2.050875145857643,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00018001293568125232,
      "loss": 0.0578,
      "step": 4394
    },
    {
      "epoch": 2.0513418903150527,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001800041332616477,
      "loss": 0.0738,
      "step": 4395
    },
    {
      "epoch": 2.051808634772462,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017999532911946528,
      "loss": 0.076,
      "step": 4396
    },
    {
      "epoch": 2.052275379229872,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017998652325489463,
      "loss": 0.0621,
      "step": 4397
    },
    {
      "epoch": 2.052742123687281,
      "grad_norm": 1.234375,
      "learning_rate": 0.00017997771566812546,
      "loss": 0.1016,
      "step": 4398
    },
    {
      "epoch": 2.053208868144691,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017996890635934732,
      "loss": 0.0671,
      "step": 4399
    },
    {
      "epoch": 2.0536756126021003,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017996009532874986,
      "loss": 0.0559,
      "step": 4400
    },
    {
      "epoch": 2.05414235705951,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017995128257652286,
      "loss": 0.0744,
      "step": 4401
    },
    {
      "epoch": 2.0546091015169194,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017994246810285603,
      "loss": 0.078,
      "step": 4402
    },
    {
      "epoch": 2.0550758459743292,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00017993365190793918,
      "loss": 0.0713,
      "step": 4403
    },
    {
      "epoch": 2.0555425904317386,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00017992483399196209,
      "loss": 0.0557,
      "step": 4404
    },
    {
      "epoch": 2.0560093348891484,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017991601435511466,
      "loss": 0.0583,
      "step": 4405
    },
    {
      "epoch": 2.0564760793465577,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017990719299758678,
      "loss": 0.0711,
      "step": 4406
    },
    {
      "epoch": 2.0569428238039675,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017989836991956837,
      "loss": 0.0694,
      "step": 4407
    },
    {
      "epoch": 2.057409568261377,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001798895451212494,
      "loss": 0.0755,
      "step": 4408
    },
    {
      "epoch": 2.0578763127187867,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001798807186028199,
      "loss": 0.0723,
      "step": 4409
    },
    {
      "epoch": 2.058343057176196,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017987189036446988,
      "loss": 0.0637,
      "step": 4410
    },
    {
      "epoch": 2.0588098016336054,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00017986306040638944,
      "loss": 0.0504,
      "step": 4411
    },
    {
      "epoch": 2.059276546091015,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017985422872876872,
      "loss": 0.0584,
      "step": 4412
    },
    {
      "epoch": 2.0597432905484245,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017984539533179785,
      "loss": 0.0513,
      "step": 4413
    },
    {
      "epoch": 2.0602100350058343,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017983656021566705,
      "loss": 0.0639,
      "step": 4414
    },
    {
      "epoch": 2.0606767794632437,
      "grad_norm": 1.25,
      "learning_rate": 0.0001798277233805665,
      "loss": 0.0955,
      "step": 4415
    },
    {
      "epoch": 2.0611435239206535,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017981888482668647,
      "loss": 0.0705,
      "step": 4416
    },
    {
      "epoch": 2.061610268378063,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00017981004455421733,
      "loss": 0.0599,
      "step": 4417
    },
    {
      "epoch": 2.0620770128354726,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017980120256334936,
      "loss": 0.0549,
      "step": 4418
    },
    {
      "epoch": 2.062543757292882,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017979235885427297,
      "loss": 0.0569,
      "step": 4419
    },
    {
      "epoch": 2.0630105017502918,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017978351342717855,
      "loss": 0.0763,
      "step": 4420
    },
    {
      "epoch": 2.063477246207701,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017977466628225658,
      "loss": 0.0591,
      "step": 4421
    },
    {
      "epoch": 2.063943990665111,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017976581741969752,
      "loss": 0.0544,
      "step": 4422
    },
    {
      "epoch": 2.0644107351225203,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001797569668396919,
      "loss": 0.0568,
      "step": 4423
    },
    {
      "epoch": 2.06487747957993,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001797481145424303,
      "loss": 0.0705,
      "step": 4424
    },
    {
      "epoch": 2.0653442240373394,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00017973926052810332,
      "loss": 0.0501,
      "step": 4425
    },
    {
      "epoch": 2.065810968494749,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001797304047969016,
      "loss": 0.0526,
      "step": 4426
    },
    {
      "epoch": 2.0662777129521586,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017972154734901578,
      "loss": 0.0778,
      "step": 4427
    },
    {
      "epoch": 2.0667444574095684,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001797126881846366,
      "loss": 0.0461,
      "step": 4428
    },
    {
      "epoch": 2.0672112018669777,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001797038273039548,
      "loss": 0.0759,
      "step": 4429
    },
    {
      "epoch": 2.0676779463243875,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017969496470716118,
      "loss": 0.0768,
      "step": 4430
    },
    {
      "epoch": 2.068144690781797,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017968610039444655,
      "loss": 0.0707,
      "step": 4431
    },
    {
      "epoch": 2.0686114352392067,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017967723436600175,
      "loss": 0.0611,
      "step": 4432
    },
    {
      "epoch": 2.069078179696616,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017966836662201767,
      "loss": 0.0644,
      "step": 4433
    },
    {
      "epoch": 2.069544924154026,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001796594971626853,
      "loss": 0.0685,
      "step": 4434
    },
    {
      "epoch": 2.070011668611435,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00017965062598819555,
      "loss": 0.0478,
      "step": 4435
    },
    {
      "epoch": 2.070478413068845,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00017964175309873945,
      "loss": 0.031,
      "step": 4436
    },
    {
      "epoch": 2.0709451575262543,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017963287849450806,
      "loss": 0.0548,
      "step": 4437
    },
    {
      "epoch": 2.071411901983664,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017962400217569243,
      "loss": 0.0704,
      "step": 4438
    },
    {
      "epoch": 2.0718786464410734,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017961512414248367,
      "loss": 0.0901,
      "step": 4439
    },
    {
      "epoch": 2.0723453908984832,
      "grad_norm": 1.046875,
      "learning_rate": 0.000179606244395073,
      "loss": 0.0903,
      "step": 4440
    },
    {
      "epoch": 2.0728121353558926,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001795973629336515,
      "loss": 0.05,
      "step": 4441
    },
    {
      "epoch": 2.0732788798133024,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017958847975841052,
      "loss": 0.0608,
      "step": 4442
    },
    {
      "epoch": 2.0737456242707117,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00017957959486954124,
      "loss": 0.0677,
      "step": 4443
    },
    {
      "epoch": 2.0742123687281215,
      "grad_norm": 1.125,
      "learning_rate": 0.00017957070826723497,
      "loss": 0.0845,
      "step": 4444
    },
    {
      "epoch": 2.074679113185531,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017956181995168309,
      "loss": 0.0949,
      "step": 4445
    },
    {
      "epoch": 2.0751458576429407,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017955292992307694,
      "loss": 0.0797,
      "step": 4446
    },
    {
      "epoch": 2.07561260210035,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017954403818160795,
      "loss": 0.0853,
      "step": 4447
    },
    {
      "epoch": 2.07607934655776,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017953514472746757,
      "loss": 0.0513,
      "step": 4448
    },
    {
      "epoch": 2.076546091015169,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017952624956084728,
      "loss": 0.0724,
      "step": 4449
    },
    {
      "epoch": 2.077012835472579,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00017951735268193855,
      "loss": 0.0639,
      "step": 4450
    },
    {
      "epoch": 2.0774795799299883,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017950845409093303,
      "loss": 0.0689,
      "step": 4451
    },
    {
      "epoch": 2.0779463243873977,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001794995537880223,
      "loss": 0.0832,
      "step": 4452
    },
    {
      "epoch": 2.0784130688448075,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017949065177339795,
      "loss": 0.0693,
      "step": 4453
    },
    {
      "epoch": 2.078879813302217,
      "grad_norm": 0.875,
      "learning_rate": 0.00017948174804725166,
      "loss": 0.085,
      "step": 4454
    },
    {
      "epoch": 2.0793465577596266,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017947284260977516,
      "loss": 0.068,
      "step": 4455
    },
    {
      "epoch": 2.079813302217036,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001794639354611602,
      "loss": 0.0594,
      "step": 4456
    },
    {
      "epoch": 2.0802800466744458,
      "grad_norm": 1.0,
      "learning_rate": 0.0001794550266015985,
      "loss": 0.0765,
      "step": 4457
    },
    {
      "epoch": 2.080746791131855,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017944611603128196,
      "loss": 0.0755,
      "step": 4458
    },
    {
      "epoch": 2.081213535589265,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017943720375040236,
      "loss": 0.0592,
      "step": 4459
    },
    {
      "epoch": 2.0816802800466743,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017942828975915168,
      "loss": 0.0869,
      "step": 4460
    },
    {
      "epoch": 2.082147024504084,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017941937405772173,
      "loss": 0.0612,
      "step": 4461
    },
    {
      "epoch": 2.0826137689614934,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017941045664630458,
      "loss": 0.0683,
      "step": 4462
    },
    {
      "epoch": 2.083080513418903,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00017940153752509218,
      "loss": 0.0675,
      "step": 4463
    },
    {
      "epoch": 2.0835472578763126,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017939261669427657,
      "loss": 0.0892,
      "step": 4464
    },
    {
      "epoch": 2.0840140023337224,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017938369415404983,
      "loss": 0.0719,
      "step": 4465
    },
    {
      "epoch": 2.0844807467911317,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001793747699046041,
      "loss": 0.0644,
      "step": 4466
    },
    {
      "epoch": 2.0849474912485415,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001793658439461315,
      "loss": 0.1054,
      "step": 4467
    },
    {
      "epoch": 2.085414235705951,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00017935691627882423,
      "loss": 0.0778,
      "step": 4468
    },
    {
      "epoch": 2.0858809801633607,
      "grad_norm": 0.75,
      "learning_rate": 0.00017934798690287447,
      "loss": 0.0525,
      "step": 4469
    },
    {
      "epoch": 2.08634772462077,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00017933905581847453,
      "loss": 0.0557,
      "step": 4470
    },
    {
      "epoch": 2.08681446907818,
      "grad_norm": 1.171875,
      "learning_rate": 0.0001793301230258167,
      "loss": 0.0679,
      "step": 4471
    },
    {
      "epoch": 2.087281213535589,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001793211885250933,
      "loss": 0.0613,
      "step": 4472
    },
    {
      "epoch": 2.087747957992999,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001793122523164967,
      "loss": 0.057,
      "step": 4473
    },
    {
      "epoch": 2.0882147024504083,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00017930331440021933,
      "loss": 0.0802,
      "step": 4474
    },
    {
      "epoch": 2.088681446907818,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017929437477645359,
      "loss": 0.0512,
      "step": 4475
    },
    {
      "epoch": 2.0891481913652274,
      "grad_norm": 1.296875,
      "learning_rate": 0.000179285433445392,
      "loss": 0.0852,
      "step": 4476
    },
    {
      "epoch": 2.0896149358226372,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017927649040722703,
      "loss": 0.071,
      "step": 4477
    },
    {
      "epoch": 2.0900816802800466,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017926754566215128,
      "loss": 0.0868,
      "step": 4478
    },
    {
      "epoch": 2.0905484247374564,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017925859921035731,
      "loss": 0.0749,
      "step": 4479
    },
    {
      "epoch": 2.0910151691948657,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017924965105203778,
      "loss": 0.1042,
      "step": 4480
    },
    {
      "epoch": 2.0914819136522755,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001792407011873853,
      "loss": 0.0717,
      "step": 4481
    },
    {
      "epoch": 2.091948658109685,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00017923174961659264,
      "loss": 0.0601,
      "step": 4482
    },
    {
      "epoch": 2.0924154025670947,
      "grad_norm": 1.1328125,
      "learning_rate": 0.0001792227963398525,
      "loss": 0.0658,
      "step": 4483
    },
    {
      "epoch": 2.092882147024504,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017921384135735762,
      "loss": 0.063,
      "step": 4484
    },
    {
      "epoch": 2.093348891481914,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001792048846693009,
      "loss": 0.0619,
      "step": 4485
    },
    {
      "epoch": 2.093815635939323,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00017919592627587512,
      "loss": 0.0707,
      "step": 4486
    },
    {
      "epoch": 2.094282380396733,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017918696617727316,
      "loss": 0.0667,
      "step": 4487
    },
    {
      "epoch": 2.0947491248541423,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00017917800437368793,
      "loss": 0.0597,
      "step": 4488
    },
    {
      "epoch": 2.095215869311552,
      "grad_norm": 0.875,
      "learning_rate": 0.00017916904086531248,
      "loss": 0.0694,
      "step": 4489
    },
    {
      "epoch": 2.0956826137689615,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001791600756523397,
      "loss": 0.0564,
      "step": 4490
    },
    {
      "epoch": 2.0961493582263713,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017915110873496266,
      "loss": 0.0927,
      "step": 4491
    },
    {
      "epoch": 2.0966161026837806,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017914214011337443,
      "loss": 0.078,
      "step": 4492
    },
    {
      "epoch": 2.0970828471411904,
      "grad_norm": 1.0,
      "learning_rate": 0.00017913316978776815,
      "loss": 0.0675,
      "step": 4493
    },
    {
      "epoch": 2.0975495915985998,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017912419775833688,
      "loss": 0.091,
      "step": 4494
    },
    {
      "epoch": 2.098016336056009,
      "grad_norm": 0.71875,
      "learning_rate": 0.00017911522402527387,
      "loss": 0.056,
      "step": 4495
    },
    {
      "epoch": 2.098483080513419,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017910624858877232,
      "loss": 0.0753,
      "step": 4496
    },
    {
      "epoch": 2.0989498249708283,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017909727144902544,
      "loss": 0.0578,
      "step": 4497
    },
    {
      "epoch": 2.099416569428238,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017908829260622654,
      "loss": 0.0778,
      "step": 4498
    },
    {
      "epoch": 2.0998833138856474,
      "grad_norm": 0.98828125,
      "learning_rate": 0.000179079312060569,
      "loss": 0.0818,
      "step": 4499
    },
    {
      "epoch": 2.100350058343057,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00017907032981224607,
      "loss": 0.059,
      "step": 4500
    },
    {
      "epoch": 2.1008168028004666,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001790613458614512,
      "loss": 0.0717,
      "step": 4501
    },
    {
      "epoch": 2.1012835472578764,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00017905236020837787,
      "loss": 0.0831,
      "step": 4502
    },
    {
      "epoch": 2.1017502917152857,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00017904337285321953,
      "loss": 0.0487,
      "step": 4503
    },
    {
      "epoch": 2.1022170361726955,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017903438379616966,
      "loss": 0.0707,
      "step": 4504
    },
    {
      "epoch": 2.102683780630105,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001790253930374218,
      "loss": 0.0525,
      "step": 4505
    },
    {
      "epoch": 2.1031505250875147,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017901640057716955,
      "loss": 0.0576,
      "step": 4506
    },
    {
      "epoch": 2.103617269544924,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017900740641560653,
      "loss": 0.0779,
      "step": 4507
    },
    {
      "epoch": 2.104084014002334,
      "grad_norm": 1.0,
      "learning_rate": 0.0001789984105529264,
      "loss": 0.091,
      "step": 4508
    },
    {
      "epoch": 2.104550758459743,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017898941298932284,
      "loss": 0.0523,
      "step": 4509
    },
    {
      "epoch": 2.105017502917153,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00017898041372498957,
      "loss": 0.065,
      "step": 4510
    },
    {
      "epoch": 2.1054842473745623,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017897141276012034,
      "loss": 0.0629,
      "step": 4511
    },
    {
      "epoch": 2.105950991831972,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017896241009490903,
      "loss": 0.0722,
      "step": 4512
    },
    {
      "epoch": 2.1064177362893814,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017895340572954938,
      "loss": 0.0972,
      "step": 4513
    },
    {
      "epoch": 2.1068844807467912,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001789443996642353,
      "loss": 0.0612,
      "step": 4514
    },
    {
      "epoch": 2.1073512252042006,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001789353918991607,
      "loss": 0.0498,
      "step": 4515
    },
    {
      "epoch": 2.1078179696616104,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001789263824345196,
      "loss": 0.0894,
      "step": 4516
    },
    {
      "epoch": 2.1082847141190197,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017891737127050586,
      "loss": 0.0646,
      "step": 4517
    },
    {
      "epoch": 2.1087514585764295,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00017890835840731354,
      "loss": 0.0541,
      "step": 4518
    },
    {
      "epoch": 2.109218203033839,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017889934384513674,
      "loss": 0.0691,
      "step": 4519
    },
    {
      "epoch": 2.1096849474912487,
      "grad_norm": 0.75,
      "learning_rate": 0.0001788903275841695,
      "loss": 0.057,
      "step": 4520
    },
    {
      "epoch": 2.110151691948658,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017888130962460602,
      "loss": 0.073,
      "step": 4521
    },
    {
      "epoch": 2.110618436406068,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017887228996664038,
      "loss": 0.0663,
      "step": 4522
    },
    {
      "epoch": 2.111085180863477,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017886326861046685,
      "loss": 0.0855,
      "step": 4523
    },
    {
      "epoch": 2.111551925320887,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017885424555627965,
      "loss": 0.0676,
      "step": 4524
    },
    {
      "epoch": 2.1120186697782963,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017884522080427304,
      "loss": 0.0622,
      "step": 4525
    },
    {
      "epoch": 2.112485414235706,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017883619435464136,
      "loss": 0.0557,
      "step": 4526
    },
    {
      "epoch": 2.1129521586931155,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017882716620757892,
      "loss": 0.0922,
      "step": 4527
    },
    {
      "epoch": 2.1134189031505253,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017881813636328015,
      "loss": 0.093,
      "step": 4528
    },
    {
      "epoch": 2.1138856476079346,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00017880910482193945,
      "loss": 0.0529,
      "step": 4529
    },
    {
      "epoch": 2.1143523920653444,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001788000715837513,
      "loss": 0.0731,
      "step": 4530
    },
    {
      "epoch": 2.1148191365227538,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017879103664891015,
      "loss": 0.0627,
      "step": 4531
    },
    {
      "epoch": 2.1152858809801636,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017878200001761058,
      "loss": 0.0703,
      "step": 4532
    },
    {
      "epoch": 2.115752625437573,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017877296169004712,
      "loss": 0.0686,
      "step": 4533
    },
    {
      "epoch": 2.1162193698949823,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00017876392166641444,
      "loss": 0.0878,
      "step": 4534
    },
    {
      "epoch": 2.116686114352392,
      "grad_norm": 1.1953125,
      "learning_rate": 0.0001787548799469071,
      "loss": 0.0672,
      "step": 4535
    },
    {
      "epoch": 2.1171528588098014,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00017874583653171983,
      "loss": 0.0677,
      "step": 4536
    },
    {
      "epoch": 2.117619603267211,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001787367914210473,
      "loss": 0.046,
      "step": 4537
    },
    {
      "epoch": 2.1180863477246206,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001787277446150843,
      "loss": 0.0609,
      "step": 4538
    },
    {
      "epoch": 2.1185530921820304,
      "grad_norm": 1.0,
      "learning_rate": 0.00017871869611402563,
      "loss": 0.0849,
      "step": 4539
    },
    {
      "epoch": 2.1190198366394397,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001787096459180661,
      "loss": 0.0755,
      "step": 4540
    },
    {
      "epoch": 2.1194865810968495,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017870059402740055,
      "loss": 0.0688,
      "step": 4541
    },
    {
      "epoch": 2.119953325554259,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017869154044222385,
      "loss": 0.0831,
      "step": 4542
    },
    {
      "epoch": 2.1204200700116687,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00017868248516273103,
      "loss": 0.0521,
      "step": 4543
    },
    {
      "epoch": 2.120886814469078,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017867342818911696,
      "loss": 0.0742,
      "step": 4544
    },
    {
      "epoch": 2.121353558926488,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00017866436952157676,
      "loss": 0.061,
      "step": 4545
    },
    {
      "epoch": 2.121820303383897,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017865530916030535,
      "loss": 0.1028,
      "step": 4546
    },
    {
      "epoch": 2.122287047841307,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017864624710549787,
      "loss": 0.0861,
      "step": 4547
    },
    {
      "epoch": 2.1227537922987163,
      "grad_norm": 0.765625,
      "learning_rate": 0.00017863718335734942,
      "loss": 0.0612,
      "step": 4548
    },
    {
      "epoch": 2.123220536756126,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017862811791605521,
      "loss": 0.0827,
      "step": 4549
    },
    {
      "epoch": 2.1236872812135354,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00017861905078181032,
      "loss": 0.0656,
      "step": 4550
    },
    {
      "epoch": 2.1241540256709452,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017860998195481006,
      "loss": 0.0789,
      "step": 4551
    },
    {
      "epoch": 2.1246207701283546,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017860091143524968,
      "loss": 0.06,
      "step": 4552
    },
    {
      "epoch": 2.1250875145857644,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017859183922332449,
      "loss": 0.0789,
      "step": 4553
    },
    {
      "epoch": 2.1255542590431737,
      "grad_norm": 0.765625,
      "learning_rate": 0.00017858276531922978,
      "loss": 0.0633,
      "step": 4554
    },
    {
      "epoch": 2.1260210035005835,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001785736897231609,
      "loss": 0.0846,
      "step": 4555
    },
    {
      "epoch": 2.126487747957993,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00017856461243531338,
      "loss": 0.0627,
      "step": 4556
    },
    {
      "epoch": 2.1269544924154027,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017855553345588255,
      "loss": 0.0899,
      "step": 4557
    },
    {
      "epoch": 2.127421236872812,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001785464527850639,
      "loss": 0.0807,
      "step": 4558
    },
    {
      "epoch": 2.127887981330222,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000178537370423053,
      "loss": 0.0471,
      "step": 4559
    },
    {
      "epoch": 2.128354725787631,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001785282863700454,
      "loss": 0.076,
      "step": 4560
    },
    {
      "epoch": 2.128821470245041,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017851920062623663,
      "loss": 0.0566,
      "step": 4561
    },
    {
      "epoch": 2.1292882147024503,
      "grad_norm": 1.4453125,
      "learning_rate": 0.00017851011319182237,
      "loss": 0.0747,
      "step": 4562
    },
    {
      "epoch": 2.12975495915986,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00017850102406699822,
      "loss": 0.0638,
      "step": 4563
    },
    {
      "epoch": 2.1302217036172695,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017849193325195997,
      "loss": 0.0554,
      "step": 4564
    },
    {
      "epoch": 2.1306884480746793,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001784828407469033,
      "loss": 0.0674,
      "step": 4565
    },
    {
      "epoch": 2.1311551925320886,
      "grad_norm": 1.0078125,
      "learning_rate": 0.000178473746552024,
      "loss": 0.063,
      "step": 4566
    },
    {
      "epoch": 2.1316219369894984,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017846465066751784,
      "loss": 0.083,
      "step": 4567
    },
    {
      "epoch": 2.1320886814469078,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001784555530935807,
      "loss": 0.0883,
      "step": 4568
    },
    {
      "epoch": 2.1325554259043176,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017844645383040847,
      "loss": 0.0666,
      "step": 4569
    },
    {
      "epoch": 2.133022170361727,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00017843735287819703,
      "loss": 0.0615,
      "step": 4570
    },
    {
      "epoch": 2.1334889148191367,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017842825023714237,
      "loss": 0.0876,
      "step": 4571
    },
    {
      "epoch": 2.133955659276546,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017841914590744045,
      "loss": 0.0783,
      "step": 4572
    },
    {
      "epoch": 2.134422403733956,
      "grad_norm": 0.65625,
      "learning_rate": 0.00017841003988928734,
      "loss": 0.0512,
      "step": 4573
    },
    {
      "epoch": 2.134889148191365,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017840093218287906,
      "loss": 0.0682,
      "step": 4574
    },
    {
      "epoch": 2.135355892648775,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017839182278841172,
      "loss": 0.0586,
      "step": 4575
    },
    {
      "epoch": 2.1358226371061844,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017838271170608145,
      "loss": 0.0543,
      "step": 4576
    },
    {
      "epoch": 2.1362893815635937,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00017837359893608444,
      "loss": 0.0688,
      "step": 4577
    },
    {
      "epoch": 2.1367561260210035,
      "grad_norm": 0.875,
      "learning_rate": 0.0001783644844786169,
      "loss": 0.0599,
      "step": 4578
    },
    {
      "epoch": 2.137222870478413,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017835536833387506,
      "loss": 0.0692,
      "step": 4579
    },
    {
      "epoch": 2.1376896149358227,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017834625050205518,
      "loss": 0.0741,
      "step": 4580
    },
    {
      "epoch": 2.138156359393232,
      "grad_norm": 1.265625,
      "learning_rate": 0.0001783371309833536,
      "loss": 0.0774,
      "step": 4581
    },
    {
      "epoch": 2.138623103850642,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017832800977796672,
      "loss": 0.0689,
      "step": 4582
    },
    {
      "epoch": 2.139089848308051,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001783188868860908,
      "loss": 0.067,
      "step": 4583
    },
    {
      "epoch": 2.139556592765461,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001783097623079224,
      "loss": 0.0534,
      "step": 4584
    },
    {
      "epoch": 2.1400233372228703,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017830063604365795,
      "loss": 0.0845,
      "step": 4585
    },
    {
      "epoch": 2.14049008168028,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017829150809349392,
      "loss": 0.0485,
      "step": 4586
    },
    {
      "epoch": 2.1409568261376895,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017828237845762682,
      "loss": 0.0609,
      "step": 4587
    },
    {
      "epoch": 2.1414235705950992,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017827324713625328,
      "loss": 0.0728,
      "step": 4588
    },
    {
      "epoch": 2.1418903150525086,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001782641141295699,
      "loss": 0.0679,
      "step": 4589
    },
    {
      "epoch": 2.1423570595099184,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001782549794377733,
      "loss": 0.0542,
      "step": 4590
    },
    {
      "epoch": 2.1428238039673277,
      "grad_norm": 1.125,
      "learning_rate": 0.00017824584306106014,
      "loss": 0.0984,
      "step": 4591
    },
    {
      "epoch": 2.1432905484247375,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001782367049996272,
      "loss": 0.08,
      "step": 4592
    },
    {
      "epoch": 2.143757292882147,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017822756525367117,
      "loss": 0.0871,
      "step": 4593
    },
    {
      "epoch": 2.1442240373395567,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017821842382338888,
      "loss": 0.0731,
      "step": 4594
    },
    {
      "epoch": 2.144690781796966,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017820928070897711,
      "loss": 0.0692,
      "step": 4595
    },
    {
      "epoch": 2.145157526254376,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00017820013591063279,
      "loss": 0.0516,
      "step": 4596
    },
    {
      "epoch": 2.145624270711785,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017819098942855276,
      "loss": 0.0621,
      "step": 4597
    },
    {
      "epoch": 2.146091015169195,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017818184126293396,
      "loss": 0.0631,
      "step": 4598
    },
    {
      "epoch": 2.1465577596266043,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017817269141397343,
      "loss": 0.0625,
      "step": 4599
    },
    {
      "epoch": 2.147024504084014,
      "grad_norm": 0.71875,
      "learning_rate": 0.00017816353988186803,
      "loss": 0.0544,
      "step": 4600
    },
    {
      "epoch": 2.1474912485414235,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00017815438666681497,
      "loss": 0.0782,
      "step": 4601
    },
    {
      "epoch": 2.1479579929988333,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017814523176901123,
      "loss": 0.071,
      "step": 4602
    },
    {
      "epoch": 2.1484247374562426,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001781360751886539,
      "loss": 0.0744,
      "step": 4603
    },
    {
      "epoch": 2.1488914819136524,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017812691692594025,
      "loss": 0.0638,
      "step": 4604
    },
    {
      "epoch": 2.1493582263710618,
      "grad_norm": 1.0,
      "learning_rate": 0.00017811775698106736,
      "loss": 0.07,
      "step": 4605
    },
    {
      "epoch": 2.1498249708284716,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017810859535423248,
      "loss": 0.0821,
      "step": 4606
    },
    {
      "epoch": 2.150291715285881,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001780994320456329,
      "loss": 0.0792,
      "step": 4607
    },
    {
      "epoch": 2.1507584597432907,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017809026705546587,
      "loss": 0.0543,
      "step": 4608
    },
    {
      "epoch": 2.1512252042007,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017808110038392878,
      "loss": 0.0518,
      "step": 4609
    },
    {
      "epoch": 2.15169194865811,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017807193203121896,
      "loss": 0.0611,
      "step": 4610
    },
    {
      "epoch": 2.152158693115519,
      "grad_norm": 0.875,
      "learning_rate": 0.0001780627619975338,
      "loss": 0.0546,
      "step": 4611
    },
    {
      "epoch": 2.152625437572929,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00017805359028307075,
      "loss": 0.0616,
      "step": 4612
    },
    {
      "epoch": 2.1530921820303384,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017804441688802736,
      "loss": 0.0617,
      "step": 4613
    },
    {
      "epoch": 2.153558926487748,
      "grad_norm": 0.78125,
      "learning_rate": 0.00017803524181260103,
      "loss": 0.0556,
      "step": 4614
    },
    {
      "epoch": 2.1540256709451575,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017802606505698938,
      "loss": 0.0586,
      "step": 4615
    },
    {
      "epoch": 2.154492415402567,
      "grad_norm": 1.109375,
      "learning_rate": 0.00017801688662138996,
      "loss": 0.0714,
      "step": 4616
    },
    {
      "epoch": 2.1549591598599767,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017800770650600045,
      "loss": 0.0543,
      "step": 4617
    },
    {
      "epoch": 2.1554259043173865,
      "grad_norm": 1.140625,
      "learning_rate": 0.00017799852471101844,
      "loss": 0.0905,
      "step": 4618
    },
    {
      "epoch": 2.155892648774796,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017798934123664165,
      "loss": 0.0719,
      "step": 4619
    },
    {
      "epoch": 2.156359393232205,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017798015608306783,
      "loss": 0.0869,
      "step": 4620
    },
    {
      "epoch": 2.156826137689615,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001779709692504947,
      "loss": 0.0647,
      "step": 4621
    },
    {
      "epoch": 2.1572928821470243,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001779617807391201,
      "loss": 0.0561,
      "step": 4622
    },
    {
      "epoch": 2.157759626604434,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001779525905491419,
      "loss": 0.0704,
      "step": 4623
    },
    {
      "epoch": 2.1582263710618435,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017794339868075788,
      "loss": 0.0782,
      "step": 4624
    },
    {
      "epoch": 2.1586931155192532,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00017793420513416604,
      "loss": 0.0498,
      "step": 4625
    },
    {
      "epoch": 2.1591598599766626,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00017792500990956427,
      "loss": 0.0554,
      "step": 4626
    },
    {
      "epoch": 2.1596266044340724,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001779158130071506,
      "loss": 0.0732,
      "step": 4627
    },
    {
      "epoch": 2.1600933488914817,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017790661442712303,
      "loss": 0.0903,
      "step": 4628
    },
    {
      "epoch": 2.1605600933488915,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001778974141696796,
      "loss": 0.0543,
      "step": 4629
    },
    {
      "epoch": 2.161026837806301,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017788821223501842,
      "loss": 0.0865,
      "step": 4630
    },
    {
      "epoch": 2.1614935822637107,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001778790086233376,
      "loss": 0.0642,
      "step": 4631
    },
    {
      "epoch": 2.16196032672112,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017786980333483533,
      "loss": 0.0625,
      "step": 4632
    },
    {
      "epoch": 2.16242707117853,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001778605963697098,
      "loss": 0.079,
      "step": 4633
    },
    {
      "epoch": 2.162893815635939,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001778513877281592,
      "loss": 0.0616,
      "step": 4634
    },
    {
      "epoch": 2.163360560093349,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001778421774103819,
      "loss": 0.0646,
      "step": 4635
    },
    {
      "epoch": 2.1638273045507583,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017783296541657614,
      "loss": 0.063,
      "step": 4636
    },
    {
      "epoch": 2.164294049008168,
      "grad_norm": 1.1875,
      "learning_rate": 0.00017782375174694027,
      "loss": 0.0823,
      "step": 4637
    },
    {
      "epoch": 2.1647607934655775,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00017781453640167266,
      "loss": 0.0579,
      "step": 4638
    },
    {
      "epoch": 2.1652275379229873,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017780531938097177,
      "loss": 0.0613,
      "step": 4639
    },
    {
      "epoch": 2.1656942823803966,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017779610068503602,
      "loss": 0.0591,
      "step": 4640
    },
    {
      "epoch": 2.1661610268378064,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001777868803140639,
      "loss": 0.0794,
      "step": 4641
    },
    {
      "epoch": 2.1666277712952158,
      "grad_norm": 1.453125,
      "learning_rate": 0.00017777765826825395,
      "loss": 0.0986,
      "step": 4642
    },
    {
      "epoch": 2.1670945157526256,
      "grad_norm": 1.078125,
      "learning_rate": 0.0001777684345478047,
      "loss": 0.0694,
      "step": 4643
    },
    {
      "epoch": 2.167561260210035,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017775920915291479,
      "loss": 0.0819,
      "step": 4644
    },
    {
      "epoch": 2.1680280046674447,
      "grad_norm": 1.125,
      "learning_rate": 0.0001777499820837828,
      "loss": 0.0942,
      "step": 4645
    },
    {
      "epoch": 2.168494749124854,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017774075334060747,
      "loss": 0.0686,
      "step": 4646
    },
    {
      "epoch": 2.168961493582264,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017773152292358744,
      "loss": 0.0739,
      "step": 4647
    },
    {
      "epoch": 2.169428238039673,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017772229083292152,
      "loss": 0.0874,
      "step": 4648
    },
    {
      "epoch": 2.169894982497083,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00017771305706880842,
      "loss": 0.0595,
      "step": 4649
    },
    {
      "epoch": 2.1703617269544924,
      "grad_norm": 0.703125,
      "learning_rate": 0.00017770382163144696,
      "loss": 0.0619,
      "step": 4650
    },
    {
      "epoch": 2.170828471411902,
      "grad_norm": 0.875,
      "learning_rate": 0.00017769458452103603,
      "loss": 0.0678,
      "step": 4651
    },
    {
      "epoch": 2.1712952158693115,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00017768534573777447,
      "loss": 0.0697,
      "step": 4652
    },
    {
      "epoch": 2.1717619603267213,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017767610528186123,
      "loss": 0.087,
      "step": 4653
    },
    {
      "epoch": 2.1722287047841307,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017766686315349526,
      "loss": 0.0703,
      "step": 4654
    },
    {
      "epoch": 2.1726954492415405,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017765761935287558,
      "loss": 0.0591,
      "step": 4655
    },
    {
      "epoch": 2.17316219369895,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017764837388020114,
      "loss": 0.0816,
      "step": 4656
    },
    {
      "epoch": 2.1736289381563596,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017763912673567108,
      "loss": 0.062,
      "step": 4657
    },
    {
      "epoch": 2.174095682613769,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017762987791948448,
      "loss": 0.0592,
      "step": 4658
    },
    {
      "epoch": 2.1745624270711783,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001776206274318405,
      "loss": 0.0635,
      "step": 4659
    },
    {
      "epoch": 2.175029171528588,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017761137527293826,
      "loss": 0.0692,
      "step": 4660
    },
    {
      "epoch": 2.175495915985998,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000177602121442977,
      "loss": 0.0748,
      "step": 4661
    },
    {
      "epoch": 2.1759626604434072,
      "grad_norm": 0.9765625,
      "learning_rate": 0.000177592865942156,
      "loss": 0.078,
      "step": 4662
    },
    {
      "epoch": 2.1764294049008166,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017758360877067445,
      "loss": 0.0777,
      "step": 4663
    },
    {
      "epoch": 2.1768961493582264,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017757434992873177,
      "loss": 0.079,
      "step": 4664
    },
    {
      "epoch": 2.1773628938156357,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017756508941652727,
      "loss": 0.0707,
      "step": 4665
    },
    {
      "epoch": 2.1778296382730455,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001775558272342603,
      "loss": 0.0787,
      "step": 4666
    },
    {
      "epoch": 2.178296382730455,
      "grad_norm": 1.0,
      "learning_rate": 0.00017754656338213033,
      "loss": 0.081,
      "step": 4667
    },
    {
      "epoch": 2.1787631271878647,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001775372978603368,
      "loss": 0.0725,
      "step": 4668
    },
    {
      "epoch": 2.179229871645274,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017752803066907924,
      "loss": 0.0772,
      "step": 4669
    },
    {
      "epoch": 2.179696616102684,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017751876180855715,
      "loss": 0.0715,
      "step": 4670
    },
    {
      "epoch": 2.180163360560093,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017750949127897015,
      "loss": 0.0564,
      "step": 4671
    },
    {
      "epoch": 2.180630105017503,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017750021908051776,
      "loss": 0.0508,
      "step": 4672
    },
    {
      "epoch": 2.1810968494749123,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017749094521339967,
      "loss": 0.0586,
      "step": 4673
    },
    {
      "epoch": 2.181563593932322,
      "grad_norm": 1.0,
      "learning_rate": 0.00017748166967781554,
      "loss": 0.0726,
      "step": 4674
    },
    {
      "epoch": 2.1820303383897315,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001774723924739651,
      "loss": 0.0767,
      "step": 4675
    },
    {
      "epoch": 2.1824970828471413,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017746311360204812,
      "loss": 0.0755,
      "step": 4676
    },
    {
      "epoch": 2.1829638273045506,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017745383306226432,
      "loss": 0.0698,
      "step": 4677
    },
    {
      "epoch": 2.1834305717619604,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001774445508548136,
      "loss": 0.0685,
      "step": 4678
    },
    {
      "epoch": 2.1838973162193698,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00017743526697989575,
      "loss": 0.0559,
      "step": 4679
    },
    {
      "epoch": 2.1843640606767796,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017742598143771069,
      "loss": 0.0881,
      "step": 4680
    },
    {
      "epoch": 2.184830805134189,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017741669422845834,
      "loss": 0.0559,
      "step": 4681
    },
    {
      "epoch": 2.1852975495915987,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017740740535233867,
      "loss": 0.0571,
      "step": 4682
    },
    {
      "epoch": 2.185764294049008,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017739811480955167,
      "loss": 0.0719,
      "step": 4683
    },
    {
      "epoch": 2.186231038506418,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001773888226002974,
      "loss": 0.0798,
      "step": 4684
    },
    {
      "epoch": 2.186697782963827,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001773795287247759,
      "loss": 0.0565,
      "step": 4685
    },
    {
      "epoch": 2.187164527421237,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001773702331831873,
      "loss": 0.0727,
      "step": 4686
    },
    {
      "epoch": 2.1876312718786464,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017736093597573173,
      "loss": 0.0766,
      "step": 4687
    },
    {
      "epoch": 2.188098016336056,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017735163710260937,
      "loss": 0.09,
      "step": 4688
    },
    {
      "epoch": 2.1885647607934655,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017734233656402045,
      "loss": 0.0596,
      "step": 4689
    },
    {
      "epoch": 2.1890315052508753,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017733303436016522,
      "loss": 0.0582,
      "step": 4690
    },
    {
      "epoch": 2.1894982497082847,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00017732373049124392,
      "loss": 0.0592,
      "step": 4691
    },
    {
      "epoch": 2.1899649941656945,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00017731442495745695,
      "loss": 0.061,
      "step": 4692
    },
    {
      "epoch": 2.190431738623104,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001773051177590046,
      "loss": 0.0582,
      "step": 4693
    },
    {
      "epoch": 2.1908984830805136,
      "grad_norm": 1.21875,
      "learning_rate": 0.00017729580889608735,
      "loss": 0.079,
      "step": 4694
    },
    {
      "epoch": 2.191365227537923,
      "grad_norm": 1.15625,
      "learning_rate": 0.00017728649836890552,
      "loss": 0.0805,
      "step": 4695
    },
    {
      "epoch": 2.1918319719953328,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017727718617765965,
      "loss": 0.0726,
      "step": 4696
    },
    {
      "epoch": 2.192298716452742,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001772678723225502,
      "loss": 0.0529,
      "step": 4697
    },
    {
      "epoch": 2.192765460910152,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017725855680377774,
      "loss": 0.0599,
      "step": 4698
    },
    {
      "epoch": 2.1932322053675612,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017724923962154283,
      "loss": 0.074,
      "step": 4699
    },
    {
      "epoch": 2.193698949824971,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001772399207760461,
      "loss": 0.0543,
      "step": 4700
    },
    {
      "epoch": 2.1941656942823804,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017723060026748818,
      "loss": 0.0805,
      "step": 4701
    },
    {
      "epoch": 2.1946324387397897,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017722127809606972,
      "loss": 0.0729,
      "step": 4702
    },
    {
      "epoch": 2.1950991831971995,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00017721195426199146,
      "loss": 0.058,
      "step": 4703
    },
    {
      "epoch": 2.195565927654609,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017720262876545417,
      "loss": 0.0431,
      "step": 4704
    },
    {
      "epoch": 2.1960326721120187,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017719330160665862,
      "loss": 0.0594,
      "step": 4705
    },
    {
      "epoch": 2.196499416569428,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017718397278580566,
      "loss": 0.0827,
      "step": 4706
    },
    {
      "epoch": 2.196966161026838,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001771746423030961,
      "loss": 0.0601,
      "step": 4707
    },
    {
      "epoch": 2.197432905484247,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017716531015873088,
      "loss": 0.0682,
      "step": 4708
    },
    {
      "epoch": 2.197899649941657,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017715597635291092,
      "loss": 0.0793,
      "step": 4709
    },
    {
      "epoch": 2.1983663943990663,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017714664088583717,
      "loss": 0.0753,
      "step": 4710
    },
    {
      "epoch": 2.198833138856476,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00017713730375771064,
      "loss": 0.0843,
      "step": 4711
    },
    {
      "epoch": 2.1992998833138855,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001771279649687324,
      "loss": 0.0633,
      "step": 4712
    },
    {
      "epoch": 2.1997666277712953,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017711862451910349,
      "loss": 0.058,
      "step": 4713
    },
    {
      "epoch": 2.2002333722287046,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000177109282409025,
      "loss": 0.0503,
      "step": 4714
    },
    {
      "epoch": 2.2007001166861144,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017709993863869816,
      "loss": 0.0848,
      "step": 4715
    },
    {
      "epoch": 2.201166861143524,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001770905932083241,
      "loss": 0.0694,
      "step": 4716
    },
    {
      "epoch": 2.2016336056009336,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017708124611810397,
      "loss": 0.0789,
      "step": 4717
    },
    {
      "epoch": 2.202100350058343,
      "grad_norm": 0.875,
      "learning_rate": 0.00017707189736823912,
      "loss": 0.0701,
      "step": 4718
    },
    {
      "epoch": 2.2025670945157527,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001770625469589308,
      "loss": 0.0745,
      "step": 4719
    },
    {
      "epoch": 2.203033838973162,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017705319489038038,
      "loss": 0.0614,
      "step": 4720
    },
    {
      "epoch": 2.203500583430572,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017704384116278913,
      "loss": 0.0598,
      "step": 4721
    },
    {
      "epoch": 2.203967327887981,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017703448577635852,
      "loss": 0.0584,
      "step": 4722
    },
    {
      "epoch": 2.204434072345391,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017702512873128996,
      "loss": 0.1029,
      "step": 4723
    },
    {
      "epoch": 2.2049008168028004,
      "grad_norm": 1.15625,
      "learning_rate": 0.00017701577002778493,
      "loss": 0.1049,
      "step": 4724
    },
    {
      "epoch": 2.20536756126021,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001770064096660449,
      "loss": 0.0675,
      "step": 4725
    },
    {
      "epoch": 2.2058343057176195,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00017699704764627145,
      "loss": 0.0507,
      "step": 4726
    },
    {
      "epoch": 2.2063010501750293,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001769876839686661,
      "loss": 0.0743,
      "step": 4727
    },
    {
      "epoch": 2.2067677946324387,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017697831863343054,
      "loss": 0.0539,
      "step": 4728
    },
    {
      "epoch": 2.2072345390898485,
      "grad_norm": 1.1875,
      "learning_rate": 0.00017696895164076636,
      "loss": 0.0995,
      "step": 4729
    },
    {
      "epoch": 2.207701283547258,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017695958299087522,
      "loss": 0.0713,
      "step": 4730
    },
    {
      "epoch": 2.2081680280046676,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017695021268395887,
      "loss": 0.0583,
      "step": 4731
    },
    {
      "epoch": 2.208634772462077,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001769408407202191,
      "loss": 0.0666,
      "step": 4732
    },
    {
      "epoch": 2.2091015169194868,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00017693146709985764,
      "loss": 0.0686,
      "step": 4733
    },
    {
      "epoch": 2.209568261376896,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017692209182307636,
      "loss": 0.0764,
      "step": 4734
    },
    {
      "epoch": 2.210035005834306,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017691271489007706,
      "loss": 0.0585,
      "step": 4735
    },
    {
      "epoch": 2.2105017502917153,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017690333630106172,
      "loss": 0.0648,
      "step": 4736
    },
    {
      "epoch": 2.210968494749125,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001768939560562322,
      "loss": 0.0557,
      "step": 4737
    },
    {
      "epoch": 2.2114352392065344,
      "grad_norm": 0.71875,
      "learning_rate": 0.00017688457415579047,
      "loss": 0.0638,
      "step": 4738
    },
    {
      "epoch": 2.211901983663944,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017687519059993859,
      "loss": 0.0834,
      "step": 4739
    },
    {
      "epoch": 2.2123687281213535,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017686580538887854,
      "loss": 0.0429,
      "step": 4740
    },
    {
      "epoch": 2.212835472578763,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001768564185228124,
      "loss": 0.0687,
      "step": 4741
    },
    {
      "epoch": 2.2133022170361727,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001768470300019423,
      "loss": 0.064,
      "step": 4742
    },
    {
      "epoch": 2.2137689614935825,
      "grad_norm": 1.3046875,
      "learning_rate": 0.0001768376398264704,
      "loss": 0.1207,
      "step": 4743
    },
    {
      "epoch": 2.214235705950992,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017682824799659884,
      "loss": 0.0918,
      "step": 4744
    },
    {
      "epoch": 2.214702450408401,
      "grad_norm": 1.0,
      "learning_rate": 0.0001768188545125299,
      "loss": 0.0701,
      "step": 4745
    },
    {
      "epoch": 2.215169194865811,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017680945937446576,
      "loss": 0.0792,
      "step": 4746
    },
    {
      "epoch": 2.2156359393232203,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017680006258260873,
      "loss": 0.0812,
      "step": 4747
    },
    {
      "epoch": 2.21610268378063,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00017679066413716117,
      "loss": 0.0739,
      "step": 4748
    },
    {
      "epoch": 2.2165694282380395,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017678126403832537,
      "loss": 0.0663,
      "step": 4749
    },
    {
      "epoch": 2.2170361726954493,
      "grad_norm": 0.625,
      "learning_rate": 0.0001767718622863038,
      "loss": 0.0511,
      "step": 4750
    },
    {
      "epoch": 2.2175029171528586,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00017676245888129886,
      "loss": 0.0409,
      "step": 4751
    },
    {
      "epoch": 2.2179696616102684,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017675305382351298,
      "loss": 0.0858,
      "step": 4752
    },
    {
      "epoch": 2.218436406067678,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017674364711314872,
      "loss": 0.1016,
      "step": 4753
    },
    {
      "epoch": 2.2189031505250876,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00017673423875040855,
      "loss": 0.0584,
      "step": 4754
    },
    {
      "epoch": 2.219369894982497,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017672482873549514,
      "loss": 0.0598,
      "step": 4755
    },
    {
      "epoch": 2.2198366394399067,
      "grad_norm": 0.91015625,
      "learning_rate": 0.000176715417068611,
      "loss": 0.0678,
      "step": 4756
    },
    {
      "epoch": 2.220303383897316,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017670600374995884,
      "loss": 0.089,
      "step": 4757
    },
    {
      "epoch": 2.220770128354726,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017669658877974127,
      "loss": 0.0709,
      "step": 4758
    },
    {
      "epoch": 2.221236872812135,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017668717215816107,
      "loss": 0.0777,
      "step": 4759
    },
    {
      "epoch": 2.221703617269545,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017667775388542097,
      "loss": 0.0882,
      "step": 4760
    },
    {
      "epoch": 2.2221703617269544,
      "grad_norm": 0.78125,
      "learning_rate": 0.00017666833396172375,
      "loss": 0.0506,
      "step": 4761
    },
    {
      "epoch": 2.222637106184364,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00017665891238727224,
      "loss": 0.0665,
      "step": 4762
    },
    {
      "epoch": 2.2231038506417735,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017664948916226928,
      "loss": 0.0668,
      "step": 4763
    },
    {
      "epoch": 2.2235705950991833,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001766400642869178,
      "loss": 0.0756,
      "step": 4764
    },
    {
      "epoch": 2.2240373395565927,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017663063776142068,
      "loss": 0.0526,
      "step": 4765
    },
    {
      "epoch": 2.2245040840140025,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017662120958598088,
      "loss": 0.0661,
      "step": 4766
    },
    {
      "epoch": 2.224970828471412,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017661177976080147,
      "loss": 0.0568,
      "step": 4767
    },
    {
      "epoch": 2.2254375729288216,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017660234828608545,
      "loss": 0.0605,
      "step": 4768
    },
    {
      "epoch": 2.225904317386231,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017659291516203585,
      "loss": 0.1026,
      "step": 4769
    },
    {
      "epoch": 2.2263710618436408,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001765834803888558,
      "loss": 0.0448,
      "step": 4770
    },
    {
      "epoch": 2.22683780630105,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017657404396674843,
      "loss": 0.0786,
      "step": 4771
    },
    {
      "epoch": 2.22730455075846,
      "grad_norm": 1.0,
      "learning_rate": 0.000176564605895917,
      "loss": 0.0824,
      "step": 4772
    },
    {
      "epoch": 2.2277712952158693,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001765551661765646,
      "loss": 0.061,
      "step": 4773
    },
    {
      "epoch": 2.228238039673279,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017654572480889456,
      "loss": 0.0719,
      "step": 4774
    },
    {
      "epoch": 2.2287047841306884,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001765362817931101,
      "loss": 0.0641,
      "step": 4775
    },
    {
      "epoch": 2.229171528588098,
      "grad_norm": 1.2109375,
      "learning_rate": 0.0001765268371294146,
      "loss": 0.0858,
      "step": 4776
    },
    {
      "epoch": 2.2296382730455075,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001765173908180114,
      "loss": 0.0845,
      "step": 4777
    },
    {
      "epoch": 2.2301050175029173,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017650794285910387,
      "loss": 0.0723,
      "step": 4778
    },
    {
      "epoch": 2.2305717619603267,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017649849325289543,
      "loss": 0.0564,
      "step": 4779
    },
    {
      "epoch": 2.2310385064177365,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00017648904199958957,
      "loss": 0.0452,
      "step": 4780
    },
    {
      "epoch": 2.231505250875146,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017647958909938974,
      "loss": 0.054,
      "step": 4781
    },
    {
      "epoch": 2.2319719953325556,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001764701345524995,
      "loss": 0.091,
      "step": 4782
    },
    {
      "epoch": 2.232438739789965,
      "grad_norm": 1.34375,
      "learning_rate": 0.00017646067835912246,
      "loss": 0.0929,
      "step": 4783
    },
    {
      "epoch": 2.2329054842473743,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017645122051946217,
      "loss": 0.0595,
      "step": 4784
    },
    {
      "epoch": 2.233372228704784,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001764417610337223,
      "loss": 0.0548,
      "step": 4785
    },
    {
      "epoch": 2.233838973162194,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017643229990210645,
      "loss": 0.0655,
      "step": 4786
    },
    {
      "epoch": 2.2343057176196033,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017642283712481838,
      "loss": 0.0813,
      "step": 4787
    },
    {
      "epoch": 2.2347724620770126,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017641337270206186,
      "loss": 0.0828,
      "step": 4788
    },
    {
      "epoch": 2.2352392065344224,
      "grad_norm": 0.875,
      "learning_rate": 0.00017640390663404065,
      "loss": 0.0826,
      "step": 4789
    },
    {
      "epoch": 2.235705950991832,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017639443892095854,
      "loss": 0.0788,
      "step": 4790
    },
    {
      "epoch": 2.2361726954492416,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017638496956301939,
      "loss": 0.0624,
      "step": 4791
    },
    {
      "epoch": 2.236639439906651,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017637549856042712,
      "loss": 0.06,
      "step": 4792
    },
    {
      "epoch": 2.2371061843640607,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001763660259133856,
      "loss": 0.0427,
      "step": 4793
    },
    {
      "epoch": 2.23757292882147,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017635655162209885,
      "loss": 0.0628,
      "step": 4794
    },
    {
      "epoch": 2.23803967327888,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001763470756867708,
      "loss": 0.0826,
      "step": 4795
    },
    {
      "epoch": 2.2385064177362892,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001763375981076055,
      "loss": 0.06,
      "step": 4796
    },
    {
      "epoch": 2.238973162193699,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017632811888480704,
      "loss": 0.0784,
      "step": 4797
    },
    {
      "epoch": 2.2394399066511084,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00017631863801857947,
      "loss": 0.0758,
      "step": 4798
    },
    {
      "epoch": 2.239906651108518,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00017630915550912693,
      "loss": 0.0627,
      "step": 4799
    },
    {
      "epoch": 2.2403733955659275,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017629967135665363,
      "loss": 0.0688,
      "step": 4800
    },
    {
      "epoch": 2.2408401400233373,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00017629018556136375,
      "loss": 0.0804,
      "step": 4801
    },
    {
      "epoch": 2.2413068844807467,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017628069812346153,
      "loss": 0.0629,
      "step": 4802
    },
    {
      "epoch": 2.2417736289381565,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017627120904315123,
      "loss": 0.0745,
      "step": 4803
    },
    {
      "epoch": 2.242240373395566,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017626171832063717,
      "loss": 0.0806,
      "step": 4804
    },
    {
      "epoch": 2.2427071178529756,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017625222595612372,
      "loss": 0.0692,
      "step": 4805
    },
    {
      "epoch": 2.243173862310385,
      "grad_norm": 1.0,
      "learning_rate": 0.00017624273194981522,
      "loss": 0.0625,
      "step": 4806
    },
    {
      "epoch": 2.2436406067677948,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00017623323630191614,
      "loss": 0.0571,
      "step": 4807
    },
    {
      "epoch": 2.244107351225204,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00017622373901263089,
      "loss": 0.0628,
      "step": 4808
    },
    {
      "epoch": 2.244574095682614,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00017621424008216395,
      "loss": 0.0488,
      "step": 4809
    },
    {
      "epoch": 2.2450408401400233,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001762047395107199,
      "loss": 0.0782,
      "step": 4810
    },
    {
      "epoch": 2.245507584597433,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017619523729850322,
      "loss": 0.0695,
      "step": 4811
    },
    {
      "epoch": 2.2459743290548424,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017618573344571853,
      "loss": 0.0449,
      "step": 4812
    },
    {
      "epoch": 2.246441073512252,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001761762279525705,
      "loss": 0.0525,
      "step": 4813
    },
    {
      "epoch": 2.2469078179696615,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017616672081926374,
      "loss": 0.064,
      "step": 4814
    },
    {
      "epoch": 2.2473745624270713,
      "grad_norm": 0.87890625,
      "learning_rate": 0.000176157212046003,
      "loss": 0.0796,
      "step": 4815
    },
    {
      "epoch": 2.2478413068844807,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00017614770163299298,
      "loss": 0.0865,
      "step": 4816
    },
    {
      "epoch": 2.2483080513418905,
      "grad_norm": 1.125,
      "learning_rate": 0.00017613818958043845,
      "loss": 0.0688,
      "step": 4817
    },
    {
      "epoch": 2.2487747957993,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001761286758885442,
      "loss": 0.0775,
      "step": 4818
    },
    {
      "epoch": 2.2492415402567096,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017611916055751512,
      "loss": 0.088,
      "step": 4819
    },
    {
      "epoch": 2.249708284714119,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017610964358755603,
      "loss": 0.0549,
      "step": 4820
    },
    {
      "epoch": 2.250175029171529,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001761001249788719,
      "loss": 0.074,
      "step": 4821
    },
    {
      "epoch": 2.250641773628938,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017609060473166763,
      "loss": 0.0741,
      "step": 4822
    },
    {
      "epoch": 2.2511085180863475,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017608108284614823,
      "loss": 0.0601,
      "step": 4823
    },
    {
      "epoch": 2.2515752625437573,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001760715593225187,
      "loss": 0.058,
      "step": 4824
    },
    {
      "epoch": 2.2515752625437573,
      "eval_loss": 1.5635347366333008,
      "eval_runtime": 93.6209,
      "eval_samples_per_second": 19.269,
      "eval_steps_per_second": 2.414,
      "step": 4824
    },
    {
      "epoch": 2.252042007001167,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017606203416098407,
      "loss": 0.0809,
      "step": 4825
    },
    {
      "epoch": 2.2525087514585764,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017605250736174948,
      "loss": 0.0669,
      "step": 4826
    },
    {
      "epoch": 2.252975495915986,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00017604297892502,
      "loss": 0.0493,
      "step": 4827
    },
    {
      "epoch": 2.2534422403733956,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017603344885100083,
      "loss": 0.0595,
      "step": 4828
    },
    {
      "epoch": 2.2539089848308054,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017602391713989715,
      "loss": 0.0675,
      "step": 4829
    },
    {
      "epoch": 2.2543757292882147,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017601438379191416,
      "loss": 0.0799,
      "step": 4830
    },
    {
      "epoch": 2.254842473745624,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017600484880725715,
      "loss": 0.0541,
      "step": 4831
    },
    {
      "epoch": 2.255309218203034,
      "grad_norm": 0.75,
      "learning_rate": 0.00017599531218613144,
      "loss": 0.0591,
      "step": 4832
    },
    {
      "epoch": 2.2557759626604432,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00017598577392874233,
      "loss": 0.067,
      "step": 4833
    },
    {
      "epoch": 2.256242707117853,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017597623403529518,
      "loss": 0.0776,
      "step": 4834
    },
    {
      "epoch": 2.2567094515752624,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017596669250599548,
      "loss": 0.0815,
      "step": 4835
    },
    {
      "epoch": 2.257176196032672,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017595714934104856,
      "loss": 0.0591,
      "step": 4836
    },
    {
      "epoch": 2.2576429404900815,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017594760454065994,
      "loss": 0.0697,
      "step": 4837
    },
    {
      "epoch": 2.2581096849474913,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001759380581050351,
      "loss": 0.0655,
      "step": 4838
    },
    {
      "epoch": 2.2585764294049007,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001759285100343796,
      "loss": 0.0654,
      "step": 4839
    },
    {
      "epoch": 2.2590431738623105,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017591896032889912,
      "loss": 0.065,
      "step": 4840
    },
    {
      "epoch": 2.25950991831972,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017590940898879915,
      "loss": 0.0683,
      "step": 4841
    },
    {
      "epoch": 2.2599766627771296,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017589985601428537,
      "loss": 0.0576,
      "step": 4842
    },
    {
      "epoch": 2.260443407234539,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017589030140556348,
      "loss": 0.0592,
      "step": 4843
    },
    {
      "epoch": 2.2609101516919488,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001758807451628392,
      "loss": 0.0665,
      "step": 4844
    },
    {
      "epoch": 2.261376896149358,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001758711872863183,
      "loss": 0.0417,
      "step": 4845
    },
    {
      "epoch": 2.261843640606768,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017586162777620654,
      "loss": 0.0583,
      "step": 4846
    },
    {
      "epoch": 2.2623103850641773,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017585206663270976,
      "loss": 0.0634,
      "step": 4847
    },
    {
      "epoch": 2.262777129521587,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017584250385603386,
      "loss": 0.0734,
      "step": 4848
    },
    {
      "epoch": 2.2632438739789964,
      "grad_norm": 0.875,
      "learning_rate": 0.00017583293944638465,
      "loss": 0.0724,
      "step": 4849
    },
    {
      "epoch": 2.263710618436406,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017582337340396815,
      "loss": 0.0662,
      "step": 4850
    },
    {
      "epoch": 2.2641773628938155,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001758138057289903,
      "loss": 0.0698,
      "step": 4851
    },
    {
      "epoch": 2.2646441073512253,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001758042364216571,
      "loss": 0.0836,
      "step": 4852
    },
    {
      "epoch": 2.2651108518086347,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017579466548217453,
      "loss": 0.0705,
      "step": 4853
    },
    {
      "epoch": 2.2655775962660445,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017578509291074875,
      "loss": 0.102,
      "step": 4854
    },
    {
      "epoch": 2.266044340723454,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00017577551870758584,
      "loss": 0.0641,
      "step": 4855
    },
    {
      "epoch": 2.2665110851808636,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00017576594287289194,
      "loss": 0.0436,
      "step": 4856
    },
    {
      "epoch": 2.266977829638273,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017575636540687325,
      "loss": 0.0743,
      "step": 4857
    },
    {
      "epoch": 2.267444574095683,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00017574678630973592,
      "loss": 0.0564,
      "step": 4858
    },
    {
      "epoch": 2.267911318553092,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017573720558168625,
      "loss": 0.0598,
      "step": 4859
    },
    {
      "epoch": 2.268378063010502,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017572762322293052,
      "loss": 0.0585,
      "step": 4860
    },
    {
      "epoch": 2.2688448074679113,
      "grad_norm": 1.359375,
      "learning_rate": 0.000175718039233675,
      "loss": 0.0732,
      "step": 4861
    },
    {
      "epoch": 2.269311551925321,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017570845361412614,
      "loss": 0.0591,
      "step": 4862
    },
    {
      "epoch": 2.2697782963827304,
      "grad_norm": 1.0,
      "learning_rate": 0.00017569886636449025,
      "loss": 0.0722,
      "step": 4863
    },
    {
      "epoch": 2.2702450408401402,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017568927748497378,
      "loss": 0.0791,
      "step": 4864
    },
    {
      "epoch": 2.2707117852975496,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017567968697578315,
      "loss": 0.0794,
      "step": 4865
    },
    {
      "epoch": 2.271178529754959,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017567009483712495,
      "loss": 0.0691,
      "step": 4866
    },
    {
      "epoch": 2.2716452742123687,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017566050106920564,
      "loss": 0.0768,
      "step": 4867
    },
    {
      "epoch": 2.2721120186697785,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001756509056722318,
      "loss": 0.06,
      "step": 4868
    },
    {
      "epoch": 2.272578763127188,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017564130864641,
      "loss": 0.0805,
      "step": 4869
    },
    {
      "epoch": 2.2730455075845972,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001756317099919469,
      "loss": 0.0593,
      "step": 4870
    },
    {
      "epoch": 2.273512252042007,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017562210970904917,
      "loss": 0.0667,
      "step": 4871
    },
    {
      "epoch": 2.273978996499417,
      "grad_norm": 1.59375,
      "learning_rate": 0.00017561250779792354,
      "loss": 0.0704,
      "step": 4872
    },
    {
      "epoch": 2.274445740956826,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001756029042587767,
      "loss": 0.0618,
      "step": 4873
    },
    {
      "epoch": 2.2749124854142355,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017559329909181543,
      "loss": 0.0759,
      "step": 4874
    },
    {
      "epoch": 2.2753792298716453,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00017558369229724662,
      "loss": 0.0527,
      "step": 4875
    },
    {
      "epoch": 2.2758459743290547,
      "grad_norm": 1.078125,
      "learning_rate": 0.000175574083875277,
      "loss": 0.078,
      "step": 4876
    },
    {
      "epoch": 2.2763127187864645,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001755644738261135,
      "loss": 0.064,
      "step": 4877
    },
    {
      "epoch": 2.276779463243874,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017555486214996307,
      "loss": 0.0562,
      "step": 4878
    },
    {
      "epoch": 2.2772462077012836,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001755452488470326,
      "loss": 0.066,
      "step": 4879
    },
    {
      "epoch": 2.277712952158693,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001755356339175291,
      "loss": 0.0479,
      "step": 4880
    },
    {
      "epoch": 2.2781796966161028,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017552601736165962,
      "loss": 0.0657,
      "step": 4881
    },
    {
      "epoch": 2.278646441073512,
      "grad_norm": 0.765625,
      "learning_rate": 0.00017551639917963118,
      "loss": 0.0553,
      "step": 4882
    },
    {
      "epoch": 2.279113185530922,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001755067793716509,
      "loss": 0.0437,
      "step": 4883
    },
    {
      "epoch": 2.2795799299883313,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017549715793792586,
      "loss": 0.0557,
      "step": 4884
    },
    {
      "epoch": 2.280046674445741,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017548753487866322,
      "loss": 0.0562,
      "step": 4885
    },
    {
      "epoch": 2.2805134189031504,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00017547791019407023,
      "loss": 0.071,
      "step": 4886
    },
    {
      "epoch": 2.28098016336056,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001754682838843541,
      "loss": 0.0607,
      "step": 4887
    },
    {
      "epoch": 2.2814469078179695,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017545865594972206,
      "loss": 0.0804,
      "step": 4888
    },
    {
      "epoch": 2.2819136522753793,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017544902639038146,
      "loss": 0.0837,
      "step": 4889
    },
    {
      "epoch": 2.2823803967327887,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017543939520653958,
      "loss": 0.0737,
      "step": 4890
    },
    {
      "epoch": 2.2828471411901985,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017542976239840385,
      "loss": 0.0535,
      "step": 4891
    },
    {
      "epoch": 2.283313885647608,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017542012796618162,
      "loss": 0.0501,
      "step": 4892
    },
    {
      "epoch": 2.2837806301050176,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017541049191008035,
      "loss": 0.0655,
      "step": 4893
    },
    {
      "epoch": 2.284247374562427,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00017540085423030754,
      "loss": 0.0638,
      "step": 4894
    },
    {
      "epoch": 2.284714119019837,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017539121492707066,
      "loss": 0.0828,
      "step": 4895
    },
    {
      "epoch": 2.285180863477246,
      "grad_norm": 1.109375,
      "learning_rate": 0.0001753815740005773,
      "loss": 0.0566,
      "step": 4896
    },
    {
      "epoch": 2.285647607934656,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017537193145103502,
      "loss": 0.0707,
      "step": 4897
    },
    {
      "epoch": 2.2861143523920653,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001753622872786514,
      "loss": 0.0678,
      "step": 4898
    },
    {
      "epoch": 2.286581096849475,
      "grad_norm": 1.171875,
      "learning_rate": 0.00017535264148363413,
      "loss": 0.0711,
      "step": 4899
    },
    {
      "epoch": 2.2870478413068844,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017534299406619087,
      "loss": 0.0602,
      "step": 4900
    },
    {
      "epoch": 2.2875145857642942,
      "grad_norm": 1.15625,
      "learning_rate": 0.00017533334502652936,
      "loss": 0.1016,
      "step": 4901
    },
    {
      "epoch": 2.2879813302217036,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017532369436485738,
      "loss": 0.0684,
      "step": 4902
    },
    {
      "epoch": 2.2884480746791134,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00017531404208138266,
      "loss": 0.0589,
      "step": 4903
    },
    {
      "epoch": 2.2889148191365227,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017530438817631305,
      "loss": 0.0791,
      "step": 4904
    },
    {
      "epoch": 2.289381563593932,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001752947326498564,
      "loss": 0.0499,
      "step": 4905
    },
    {
      "epoch": 2.289848308051342,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017528507550222062,
      "loss": 0.0692,
      "step": 4906
    },
    {
      "epoch": 2.2903150525087517,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017527541673361365,
      "loss": 0.0804,
      "step": 4907
    },
    {
      "epoch": 2.290781796966161,
      "grad_norm": 1.0,
      "learning_rate": 0.00017526575634424342,
      "loss": 0.0877,
      "step": 4908
    },
    {
      "epoch": 2.2912485414235704,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000175256094334318,
      "loss": 0.0727,
      "step": 4909
    },
    {
      "epoch": 2.29171528588098,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001752464307040453,
      "loss": 0.1109,
      "step": 4910
    },
    {
      "epoch": 2.29218203033839,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001752367654536335,
      "loss": 0.0728,
      "step": 4911
    },
    {
      "epoch": 2.2926487747957993,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017522709858329064,
      "loss": 0.0808,
      "step": 4912
    },
    {
      "epoch": 2.2931155192532087,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017521743009322492,
      "loss": 0.0586,
      "step": 4913
    },
    {
      "epoch": 2.2935822637106185,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017520775998364443,
      "loss": 0.0507,
      "step": 4914
    },
    {
      "epoch": 2.294049008168028,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017519808825475742,
      "loss": 0.0688,
      "step": 4915
    },
    {
      "epoch": 2.2945157526254376,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001751884149067722,
      "loss": 0.0668,
      "step": 4916
    },
    {
      "epoch": 2.294982497082847,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017517873993989693,
      "loss": 0.0555,
      "step": 4917
    },
    {
      "epoch": 2.2954492415402568,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017516906335434,
      "loss": 0.0454,
      "step": 4918
    },
    {
      "epoch": 2.295915985997666,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001751593851503097,
      "loss": 0.0634,
      "step": 4919
    },
    {
      "epoch": 2.296382730455076,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017514970532801449,
      "loss": 0.078,
      "step": 4920
    },
    {
      "epoch": 2.2968494749124853,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00017514002388766274,
      "loss": 0.061,
      "step": 4921
    },
    {
      "epoch": 2.297316219369895,
      "grad_norm": 0.65625,
      "learning_rate": 0.00017513034082946284,
      "loss": 0.0489,
      "step": 4922
    },
    {
      "epoch": 2.2977829638273044,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00017512065615362342,
      "loss": 0.0471,
      "step": 4923
    },
    {
      "epoch": 2.298249708284714,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017511096986035289,
      "loss": 0.0434,
      "step": 4924
    },
    {
      "epoch": 2.2987164527421236,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017510128194985985,
      "loss": 0.0898,
      "step": 4925
    },
    {
      "epoch": 2.2991831971995333,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001750915924223529,
      "loss": 0.0698,
      "step": 4926
    },
    {
      "epoch": 2.2996499416569427,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017508190127804062,
      "loss": 0.057,
      "step": 4927
    },
    {
      "epoch": 2.3001166861143525,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017507220851713172,
      "loss": 0.0852,
      "step": 4928
    },
    {
      "epoch": 2.300583430571762,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017506251413983487,
      "loss": 0.0789,
      "step": 4929
    },
    {
      "epoch": 2.3010501750291716,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001750528181463588,
      "loss": 0.0465,
      "step": 4930
    },
    {
      "epoch": 2.301516919486581,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001750431205369123,
      "loss": 0.0677,
      "step": 4931
    },
    {
      "epoch": 2.301983663943991,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017503342131170414,
      "loss": 0.0764,
      "step": 4932
    },
    {
      "epoch": 2.3024504084014,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017502372047094317,
      "loss": 0.0678,
      "step": 4933
    },
    {
      "epoch": 2.30291715285881,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017501401801483827,
      "loss": 0.0591,
      "step": 4934
    },
    {
      "epoch": 2.3033838973162193,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00017500431394359834,
      "loss": 0.0648,
      "step": 4935
    },
    {
      "epoch": 2.303850641773629,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001749946082574323,
      "loss": 0.0651,
      "step": 4936
    },
    {
      "epoch": 2.3043173862310384,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001749849009565491,
      "loss": 0.063,
      "step": 4937
    },
    {
      "epoch": 2.3047841306884482,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00017497519204115784,
      "loss": 0.0575,
      "step": 4938
    },
    {
      "epoch": 2.3052508751458576,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001749654815114675,
      "loss": 0.0621,
      "step": 4939
    },
    {
      "epoch": 2.3057176196032674,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017495576936768715,
      "loss": 0.059,
      "step": 4940
    },
    {
      "epoch": 2.3061843640606767,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001749460556100259,
      "loss": 0.0519,
      "step": 4941
    },
    {
      "epoch": 2.3066511085180865,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017493634023869295,
      "loss": 0.0845,
      "step": 4942
    },
    {
      "epoch": 2.307117852975496,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017492662325389744,
      "loss": 0.0715,
      "step": 4943
    },
    {
      "epoch": 2.3075845974329057,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017491690465584864,
      "loss": 0.0704,
      "step": 4944
    },
    {
      "epoch": 2.308051341890315,
      "grad_norm": 1.0234375,
      "learning_rate": 0.0001749071844447557,
      "loss": 0.0697,
      "step": 4945
    },
    {
      "epoch": 2.308518086347725,
      "grad_norm": 0.69921875,
      "learning_rate": 0.000174897462620828,
      "loss": 0.0464,
      "step": 4946
    },
    {
      "epoch": 2.308984830805134,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017488773918427478,
      "loss": 0.0753,
      "step": 4947
    },
    {
      "epoch": 2.3094515752625435,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001748780141353055,
      "loss": 0.0576,
      "step": 4948
    },
    {
      "epoch": 2.3099183197199533,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017486828747412952,
      "loss": 0.0532,
      "step": 4949
    },
    {
      "epoch": 2.310385064177363,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017485855920095617,
      "loss": 0.0725,
      "step": 4950
    },
    {
      "epoch": 2.3108518086347725,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017484882931599502,
      "loss": 0.0755,
      "step": 4951
    },
    {
      "epoch": 2.311318553092182,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017483909781945556,
      "loss": 0.0814,
      "step": 4952
    },
    {
      "epoch": 2.3117852975495916,
      "grad_norm": 0.875,
      "learning_rate": 0.00017482936471154722,
      "loss": 0.0486,
      "step": 4953
    },
    {
      "epoch": 2.3122520420070014,
      "grad_norm": 1.125,
      "learning_rate": 0.0001748196299924797,
      "loss": 0.0677,
      "step": 4954
    },
    {
      "epoch": 2.3127187864644108,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001748098936624625,
      "loss": 0.0618,
      "step": 4955
    },
    {
      "epoch": 2.31318553092182,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001748001557217053,
      "loss": 0.0468,
      "step": 4956
    },
    {
      "epoch": 2.31365227537923,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00017479041617041773,
      "loss": 0.0624,
      "step": 4957
    },
    {
      "epoch": 2.3141190198366393,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017478067500880954,
      "loss": 0.0443,
      "step": 4958
    },
    {
      "epoch": 2.314585764294049,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017477093223709043,
      "loss": 0.0737,
      "step": 4959
    },
    {
      "epoch": 2.3150525087514584,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001747611878554702,
      "loss": 0.0557,
      "step": 4960
    },
    {
      "epoch": 2.315519253208868,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017475144186415866,
      "loss": 0.0698,
      "step": 4961
    },
    {
      "epoch": 2.3159859976662776,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017474169426336561,
      "loss": 0.0588,
      "step": 4962
    },
    {
      "epoch": 2.3164527421236873,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017473194505330096,
      "loss": 0.0722,
      "step": 4963
    },
    {
      "epoch": 2.3169194865810967,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017472219423417463,
      "loss": 0.073,
      "step": 4964
    },
    {
      "epoch": 2.3173862310385065,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017471244180619654,
      "loss": 0.0689,
      "step": 4965
    },
    {
      "epoch": 2.317852975495916,
      "grad_norm": 1.359375,
      "learning_rate": 0.00017470268776957672,
      "loss": 0.0701,
      "step": 4966
    },
    {
      "epoch": 2.3183197199533256,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001746929321245251,
      "loss": 0.0614,
      "step": 4967
    },
    {
      "epoch": 2.318786464410735,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001746831748712518,
      "loss": 0.0426,
      "step": 4968
    },
    {
      "epoch": 2.319253208868145,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017467341600996687,
      "loss": 0.0652,
      "step": 4969
    },
    {
      "epoch": 2.319719953325554,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017466365554088048,
      "loss": 0.0676,
      "step": 4970
    },
    {
      "epoch": 2.320186697782964,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017465389346420272,
      "loss": 0.0744,
      "step": 4971
    },
    {
      "epoch": 2.3206534422403733,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001746441297801438,
      "loss": 0.0684,
      "step": 4972
    },
    {
      "epoch": 2.321120186697783,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00017463436448891394,
      "loss": 0.0744,
      "step": 4973
    },
    {
      "epoch": 2.3215869311551924,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001746245975907234,
      "loss": 0.0625,
      "step": 4974
    },
    {
      "epoch": 2.3220536756126022,
      "grad_norm": 1.328125,
      "learning_rate": 0.0001746148290857825,
      "loss": 0.0814,
      "step": 4975
    },
    {
      "epoch": 2.3225204200700116,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017460505897430153,
      "loss": 0.0615,
      "step": 4976
    },
    {
      "epoch": 2.3229871645274214,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017459528725649088,
      "loss": 0.0731,
      "step": 4977
    },
    {
      "epoch": 2.3234539089848307,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017458551393256092,
      "loss": 0.0604,
      "step": 4978
    },
    {
      "epoch": 2.3239206534422405,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00017457573900272206,
      "loss": 0.0694,
      "step": 4979
    },
    {
      "epoch": 2.32438739789965,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00017456596246718484,
      "loss": 0.0633,
      "step": 4980
    },
    {
      "epoch": 2.3248541423570597,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001745561843261597,
      "loss": 0.0546,
      "step": 4981
    },
    {
      "epoch": 2.325320886814469,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00017454640457985717,
      "loss": 0.0699,
      "step": 4982
    },
    {
      "epoch": 2.325787631271879,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017453662322848787,
      "loss": 0.0515,
      "step": 4983
    },
    {
      "epoch": 2.326254375729288,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017452684027226235,
      "loss": 0.0703,
      "step": 4984
    },
    {
      "epoch": 2.326721120186698,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001745170557113913,
      "loss": 0.0875,
      "step": 4985
    },
    {
      "epoch": 2.3271878646441073,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017450726954608534,
      "loss": 0.0633,
      "step": 4986
    },
    {
      "epoch": 2.327654609101517,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001744974817765552,
      "loss": 0.0737,
      "step": 4987
    },
    {
      "epoch": 2.3281213535589265,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001744876924030116,
      "loss": 0.0511,
      "step": 4988
    },
    {
      "epoch": 2.3285880980163363,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001744779014256654,
      "loss": 0.057,
      "step": 4989
    },
    {
      "epoch": 2.3290548424737456,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001744681088447273,
      "loss": 0.0808,
      "step": 4990
    },
    {
      "epoch": 2.329521586931155,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001744583146604082,
      "loss": 0.0541,
      "step": 4991
    },
    {
      "epoch": 2.3299883313885648,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017444851887291898,
      "loss": 0.067,
      "step": 4992
    },
    {
      "epoch": 2.3304550758459746,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017443872148247054,
      "loss": 0.0655,
      "step": 4993
    },
    {
      "epoch": 2.330921820303384,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017442892248927382,
      "loss": 0.0767,
      "step": 4994
    },
    {
      "epoch": 2.3313885647607933,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00017441912189353982,
      "loss": 0.0718,
      "step": 4995
    },
    {
      "epoch": 2.331855309218203,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001744093196954796,
      "loss": 0.0595,
      "step": 4996
    },
    {
      "epoch": 2.332322053675613,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017439951589530412,
      "loss": 0.0642,
      "step": 4997
    },
    {
      "epoch": 2.332788798133022,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017438971049322453,
      "loss": 0.0548,
      "step": 4998
    },
    {
      "epoch": 2.3332555425904316,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017437990348945193,
      "loss": 0.0617,
      "step": 4999
    },
    {
      "epoch": 2.3337222870478413,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001743700948841975,
      "loss": 0.0531,
      "step": 5000
    },
    {
      "epoch": 2.3341890315052507,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001743602846776724,
      "loss": 0.0751,
      "step": 5001
    },
    {
      "epoch": 2.3346557759626605,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00017435047287008788,
      "loss": 0.0622,
      "step": 5002
    },
    {
      "epoch": 2.33512252042007,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001743406594616552,
      "loss": 0.0522,
      "step": 5003
    },
    {
      "epoch": 2.3355892648774796,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001743308444525856,
      "loss": 0.0577,
      "step": 5004
    },
    {
      "epoch": 2.336056009334889,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017432102784309045,
      "loss": 0.0766,
      "step": 5005
    },
    {
      "epoch": 2.336522753792299,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017431120963338112,
      "loss": 0.0755,
      "step": 5006
    },
    {
      "epoch": 2.336989498249708,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000174301389823669,
      "loss": 0.0381,
      "step": 5007
    },
    {
      "epoch": 2.337456242707118,
      "grad_norm": 1.125,
      "learning_rate": 0.00017429156841416554,
      "loss": 0.0924,
      "step": 5008
    },
    {
      "epoch": 2.3379229871645273,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00017428174540508215,
      "loss": 0.0777,
      "step": 5009
    },
    {
      "epoch": 2.338389731621937,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017427192079663036,
      "loss": 0.0619,
      "step": 5010
    },
    {
      "epoch": 2.3388564760793464,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017426209458902173,
      "loss": 0.0617,
      "step": 5011
    },
    {
      "epoch": 2.3393232205367562,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017425226678246782,
      "loss": 0.0696,
      "step": 5012
    },
    {
      "epoch": 2.3397899649941656,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001742424373771802,
      "loss": 0.073,
      "step": 5013
    },
    {
      "epoch": 2.3402567094515754,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017423260637337054,
      "loss": 0.0499,
      "step": 5014
    },
    {
      "epoch": 2.3407234539089847,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001742227737712505,
      "loss": 0.0555,
      "step": 5015
    },
    {
      "epoch": 2.3411901983663945,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017421293957103177,
      "loss": 0.0577,
      "step": 5016
    },
    {
      "epoch": 2.341656942823804,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017420310377292615,
      "loss": 0.0793,
      "step": 5017
    },
    {
      "epoch": 2.3421236872812137,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017419326637714533,
      "loss": 0.0549,
      "step": 5018
    },
    {
      "epoch": 2.342590431738623,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001741834273839012,
      "loss": 0.0561,
      "step": 5019
    },
    {
      "epoch": 2.343057176196033,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00017417358679340557,
      "loss": 0.0695,
      "step": 5020
    },
    {
      "epoch": 2.343523920653442,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001741637446058703,
      "loss": 0.076,
      "step": 5021
    },
    {
      "epoch": 2.343990665110852,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017415390082150737,
      "loss": 0.0607,
      "step": 5022
    },
    {
      "epoch": 2.3444574095682613,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001741440554405286,
      "loss": 0.053,
      "step": 5023
    },
    {
      "epoch": 2.344924154025671,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001741342084631461,
      "loss": 0.0567,
      "step": 5024
    },
    {
      "epoch": 2.3453908984830805,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017412435988957184,
      "loss": 0.093,
      "step": 5025
    },
    {
      "epoch": 2.3458576429404903,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017411450972001786,
      "loss": 0.0688,
      "step": 5026
    },
    {
      "epoch": 2.3463243873978996,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017410465795469627,
      "loss": 0.0707,
      "step": 5027
    },
    {
      "epoch": 2.3467911318553094,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017409480459381916,
      "loss": 0.0372,
      "step": 5028
    },
    {
      "epoch": 2.3472578763127188,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001740849496375987,
      "loss": 0.0673,
      "step": 5029
    },
    {
      "epoch": 2.347724620770128,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017407509308624706,
      "loss": 0.048,
      "step": 5030
    },
    {
      "epoch": 2.348191365227538,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017406523493997646,
      "loss": 0.0541,
      "step": 5031
    },
    {
      "epoch": 2.3486581096849477,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00017405537519899923,
      "loss": 0.0556,
      "step": 5032
    },
    {
      "epoch": 2.349124854142357,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017404551386352755,
      "loss": 0.0687,
      "step": 5033
    },
    {
      "epoch": 2.3495915985997664,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00017403565093377385,
      "loss": 0.0867,
      "step": 5034
    },
    {
      "epoch": 2.350058343057176,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017402578640995044,
      "loss": 0.0683,
      "step": 5035
    },
    {
      "epoch": 2.350525087514586,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017401592029226968,
      "loss": 0.0778,
      "step": 5036
    },
    {
      "epoch": 2.3509918319719953,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00017400605258094406,
      "loss": 0.0531,
      "step": 5037
    },
    {
      "epoch": 2.3514585764294047,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017399618327618602,
      "loss": 0.0854,
      "step": 5038
    },
    {
      "epoch": 2.3519253208868145,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017398631237820804,
      "loss": 0.0561,
      "step": 5039
    },
    {
      "epoch": 2.352392065344224,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001739764398872227,
      "loss": 0.0808,
      "step": 5040
    },
    {
      "epoch": 2.3528588098016336,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001739665658034425,
      "loss": 0.0491,
      "step": 5041
    },
    {
      "epoch": 2.353325554259043,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017395669012708006,
      "loss": 0.0517,
      "step": 5042
    },
    {
      "epoch": 2.353792298716453,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017394681285834803,
      "loss": 0.0766,
      "step": 5043
    },
    {
      "epoch": 2.354259043173862,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017393693399745907,
      "loss": 0.0597,
      "step": 5044
    },
    {
      "epoch": 2.354725787631272,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017392705354462592,
      "loss": 0.0619,
      "step": 5045
    },
    {
      "epoch": 2.3551925320886813,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00017391717150006127,
      "loss": 0.0767,
      "step": 5046
    },
    {
      "epoch": 2.355659276546091,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001739072878639779,
      "loss": 0.0636,
      "step": 5047
    },
    {
      "epoch": 2.3561260210035004,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001738974026365886,
      "loss": 0.0682,
      "step": 5048
    },
    {
      "epoch": 2.3565927654609102,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017388751581810626,
      "loss": 0.0598,
      "step": 5049
    },
    {
      "epoch": 2.3570595099183196,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017387762740874368,
      "loss": 0.0481,
      "step": 5050
    },
    {
      "epoch": 2.3575262543757294,
      "grad_norm": 0.765625,
      "learning_rate": 0.00017386773740871386,
      "loss": 0.0542,
      "step": 5051
    },
    {
      "epoch": 2.3579929988331387,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001738578458182297,
      "loss": 0.0792,
      "step": 5052
    },
    {
      "epoch": 2.3584597432905485,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017384795263750412,
      "loss": 0.0597,
      "step": 5053
    },
    {
      "epoch": 2.358926487747958,
      "grad_norm": 0.75,
      "learning_rate": 0.00017383805786675024,
      "loss": 0.0552,
      "step": 5054
    },
    {
      "epoch": 2.3593932322053677,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00017382816150618102,
      "loss": 0.0595,
      "step": 5055
    },
    {
      "epoch": 2.359859976662777,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001738182635560096,
      "loss": 0.0501,
      "step": 5056
    },
    {
      "epoch": 2.360326721120187,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017380836401644902,
      "loss": 0.0703,
      "step": 5057
    },
    {
      "epoch": 2.360793465577596,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001737984628877125,
      "loss": 0.0593,
      "step": 5058
    },
    {
      "epoch": 2.361260210035006,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017378856017001318,
      "loss": 0.0548,
      "step": 5059
    },
    {
      "epoch": 2.3617269544924153,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001737786558635643,
      "loss": 0.0701,
      "step": 5060
    },
    {
      "epoch": 2.362193698949825,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00017376874996857908,
      "loss": 0.0574,
      "step": 5061
    },
    {
      "epoch": 2.3626604434072345,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017375884248527085,
      "loss": 0.0768,
      "step": 5062
    },
    {
      "epoch": 2.3631271878646443,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001737489334138529,
      "loss": 0.0676,
      "step": 5063
    },
    {
      "epoch": 2.3635939323220536,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017373902275453858,
      "loss": 0.0857,
      "step": 5064
    },
    {
      "epoch": 2.3640606767794634,
      "grad_norm": 1.21875,
      "learning_rate": 0.00017372911050754129,
      "loss": 0.0779,
      "step": 5065
    },
    {
      "epoch": 2.3645274212368728,
      "grad_norm": 1.1875,
      "learning_rate": 0.00017371919667307447,
      "loss": 0.0882,
      "step": 5066
    },
    {
      "epoch": 2.3649941656942826,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00017370928125135147,
      "loss": 0.0524,
      "step": 5067
    },
    {
      "epoch": 2.365460910151692,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00017369936424258594,
      "loss": 0.062,
      "step": 5068
    },
    {
      "epoch": 2.3659276546091017,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001736894456469913,
      "loss": 0.0674,
      "step": 5069
    },
    {
      "epoch": 2.366394399066511,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017367952546478116,
      "loss": 0.0483,
      "step": 5070
    },
    {
      "epoch": 2.366861143523921,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00017366960369616906,
      "loss": 0.066,
      "step": 5071
    },
    {
      "epoch": 2.36732788798133,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001736596803413687,
      "loss": 0.0806,
      "step": 5072
    },
    {
      "epoch": 2.3677946324387396,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017364975540059362,
      "loss": 0.0697,
      "step": 5073
    },
    {
      "epoch": 2.3682613768961494,
      "grad_norm": 0.875,
      "learning_rate": 0.00017363982887405764,
      "loss": 0.0699,
      "step": 5074
    },
    {
      "epoch": 2.368728121353559,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017362990076197444,
      "loss": 0.078,
      "step": 5075
    },
    {
      "epoch": 2.3691948658109685,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017361997106455775,
      "loss": 0.0772,
      "step": 5076
    },
    {
      "epoch": 2.369661610268378,
      "grad_norm": 0.75,
      "learning_rate": 0.00017361003978202144,
      "loss": 0.0547,
      "step": 5077
    },
    {
      "epoch": 2.3701283547257876,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017360010691457929,
      "loss": 0.0753,
      "step": 5078
    },
    {
      "epoch": 2.3705950991831974,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017359017246244517,
      "loss": 0.0772,
      "step": 5079
    },
    {
      "epoch": 2.371061843640607,
      "grad_norm": 1.109375,
      "learning_rate": 0.000173580236425833,
      "loss": 0.0711,
      "step": 5080
    },
    {
      "epoch": 2.371528588098016,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017357029880495667,
      "loss": 0.071,
      "step": 5081
    },
    {
      "epoch": 2.371995332555426,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001735603596000302,
      "loss": 0.0707,
      "step": 5082
    },
    {
      "epoch": 2.3724620770128353,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017355041881126758,
      "loss": 0.0621,
      "step": 5083
    },
    {
      "epoch": 2.372928821470245,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017354047643888283,
      "loss": 0.0743,
      "step": 5084
    },
    {
      "epoch": 2.3733955659276544,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017353053248309004,
      "loss": 0.061,
      "step": 5085
    },
    {
      "epoch": 2.3738623103850642,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017352058694410327,
      "loss": 0.0734,
      "step": 5086
    },
    {
      "epoch": 2.3743290548424736,
      "grad_norm": 0.765625,
      "learning_rate": 0.00017351063982213672,
      "loss": 0.051,
      "step": 5087
    },
    {
      "epoch": 2.3747957992998834,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001735006911174045,
      "loss": 0.0562,
      "step": 5088
    },
    {
      "epoch": 2.3752625437572927,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017349074083012088,
      "loss": 0.0486,
      "step": 5089
    },
    {
      "epoch": 2.3757292882147025,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00017348078896050007,
      "loss": 0.047,
      "step": 5090
    },
    {
      "epoch": 2.376196032672112,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017347083550875634,
      "loss": 0.0503,
      "step": 5091
    },
    {
      "epoch": 2.3766627771295217,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017346088047510397,
      "loss": 0.0653,
      "step": 5092
    },
    {
      "epoch": 2.377129521586931,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017345092385975737,
      "loss": 0.0854,
      "step": 5093
    },
    {
      "epoch": 2.377596266044341,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001734409656629309,
      "loss": 0.0438,
      "step": 5094
    },
    {
      "epoch": 2.37806301050175,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017343100588483896,
      "loss": 0.0605,
      "step": 5095
    },
    {
      "epoch": 2.37852975495916,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017342104452569595,
      "loss": 0.0906,
      "step": 5096
    },
    {
      "epoch": 2.3789964994165693,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001734110815857164,
      "loss": 0.0755,
      "step": 5097
    },
    {
      "epoch": 2.379463243873979,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017340111706511482,
      "loss": 0.0733,
      "step": 5098
    },
    {
      "epoch": 2.3799299883313885,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00017339115096410575,
      "loss": 0.0433,
      "step": 5099
    },
    {
      "epoch": 2.3803967327887983,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017338118328290379,
      "loss": 0.0597,
      "step": 5100
    },
    {
      "epoch": 2.3808634772462076,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017337121402172348,
      "loss": 0.0605,
      "step": 5101
    },
    {
      "epoch": 2.3813302217036174,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017336124318077957,
      "loss": 0.0667,
      "step": 5102
    },
    {
      "epoch": 2.3817969661610268,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001733512707602867,
      "loss": 0.063,
      "step": 5103
    },
    {
      "epoch": 2.3822637106184366,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00017334129676045958,
      "loss": 0.0605,
      "step": 5104
    },
    {
      "epoch": 2.382730455075846,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017333132118151296,
      "loss": 0.0549,
      "step": 5105
    },
    {
      "epoch": 2.3831971995332557,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017332134402366165,
      "loss": 0.0773,
      "step": 5106
    },
    {
      "epoch": 2.383663943990665,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00017331136528712042,
      "loss": 0.0809,
      "step": 5107
    },
    {
      "epoch": 2.384130688448075,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017330138497210416,
      "loss": 0.0544,
      "step": 5108
    },
    {
      "epoch": 2.384597432905484,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017329140307882778,
      "loss": 0.0378,
      "step": 5109
    },
    {
      "epoch": 2.385064177362894,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017328141960750617,
      "loss": 0.0397,
      "step": 5110
    },
    {
      "epoch": 2.3855309218203034,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001732714345583543,
      "loss": 0.0594,
      "step": 5111
    },
    {
      "epoch": 2.3859976662777127,
      "grad_norm": 0.95703125,
      "learning_rate": 0.0001732614479315871,
      "loss": 0.0635,
      "step": 5112
    },
    {
      "epoch": 2.3864644107351225,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00017325145972741965,
      "loss": 0.0561,
      "step": 5113
    },
    {
      "epoch": 2.3869311551925323,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017324146994606704,
      "loss": 0.0838,
      "step": 5114
    },
    {
      "epoch": 2.3873978996499416,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017323147858774429,
      "loss": 0.0708,
      "step": 5115
    },
    {
      "epoch": 2.387864644107351,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017322148565266656,
      "loss": 0.0626,
      "step": 5116
    },
    {
      "epoch": 2.388331388564761,
      "grad_norm": 0.94921875,
      "learning_rate": 0.000173211491141049,
      "loss": 0.0554,
      "step": 5117
    },
    {
      "epoch": 2.3887981330221706,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017320149505310682,
      "loss": 0.0612,
      "step": 5118
    },
    {
      "epoch": 2.38926487747958,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001731914973890552,
      "loss": 0.0426,
      "step": 5119
    },
    {
      "epoch": 2.3897316219369893,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017318149814910945,
      "loss": 0.0415,
      "step": 5120
    },
    {
      "epoch": 2.390198366394399,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017317149733348485,
      "loss": 0.0777,
      "step": 5121
    },
    {
      "epoch": 2.390665110851809,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017316149494239674,
      "loss": 0.0652,
      "step": 5122
    },
    {
      "epoch": 2.3911318553092182,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00017315149097606045,
      "loss": 0.0517,
      "step": 5123
    },
    {
      "epoch": 2.3915985997666276,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017314148543469141,
      "loss": 0.0641,
      "step": 5124
    },
    {
      "epoch": 2.3920653442240374,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017313147831850504,
      "loss": 0.0715,
      "step": 5125
    },
    {
      "epoch": 2.3925320886814467,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017312146962771677,
      "loss": 0.0625,
      "step": 5126
    },
    {
      "epoch": 2.3929988331388565,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017311145936254216,
      "loss": 0.0556,
      "step": 5127
    },
    {
      "epoch": 2.393465577596266,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017310144752319672,
      "loss": 0.0652,
      "step": 5128
    },
    {
      "epoch": 2.3939323220536757,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017309143410989598,
      "loss": 0.0532,
      "step": 5129
    },
    {
      "epoch": 2.394399066511085,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001730814191228556,
      "loss": 0.0804,
      "step": 5130
    },
    {
      "epoch": 2.394865810968495,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00017307140256229116,
      "loss": 0.0577,
      "step": 5131
    },
    {
      "epoch": 2.395332555425904,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017306138442841834,
      "loss": 0.0444,
      "step": 5132
    },
    {
      "epoch": 2.395799299883314,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017305136472145285,
      "loss": 0.053,
      "step": 5133
    },
    {
      "epoch": 2.3962660443407233,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017304134344161044,
      "loss": 0.0567,
      "step": 5134
    },
    {
      "epoch": 2.396732788798133,
      "grad_norm": 0.78125,
      "learning_rate": 0.00017303132058910686,
      "loss": 0.0552,
      "step": 5135
    },
    {
      "epoch": 2.3971995332555425,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017302129616415793,
      "loss": 0.0589,
      "step": 5136
    },
    {
      "epoch": 2.3976662777129523,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017301127016697943,
      "loss": 0.0658,
      "step": 5137
    },
    {
      "epoch": 2.3981330221703616,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00017300124259778734,
      "loss": 0.0571,
      "step": 5138
    },
    {
      "epoch": 2.3985997666277714,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017299121345679746,
      "loss": 0.0735,
      "step": 5139
    },
    {
      "epoch": 2.3990665110851808,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017298118274422576,
      "loss": 0.0606,
      "step": 5140
    },
    {
      "epoch": 2.3995332555425906,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00017297115046028828,
      "loss": 0.052,
      "step": 5141
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.0,
      "learning_rate": 0.00017296111660520088,
      "loss": 0.0621,
      "step": 5142
    },
    {
      "epoch": 2.4004667444574097,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00017295108117917977,
      "loss": 0.0533,
      "step": 5143
    },
    {
      "epoch": 2.400933488914819,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00017294104418244088,
      "loss": 0.0673,
      "step": 5144
    },
    {
      "epoch": 2.401400233372229,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017293100561520042,
      "loss": 0.0688,
      "step": 5145
    },
    {
      "epoch": 2.401866977829638,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001729209654776745,
      "loss": 0.0574,
      "step": 5146
    },
    {
      "epoch": 2.402333722287048,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017291092377007923,
      "loss": 0.0686,
      "step": 5147
    },
    {
      "epoch": 2.4028004667444574,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001729008804926309,
      "loss": 0.0545,
      "step": 5148
    },
    {
      "epoch": 2.403267211201867,
      "grad_norm": 0.875,
      "learning_rate": 0.00017289083564554577,
      "loss": 0.0679,
      "step": 5149
    },
    {
      "epoch": 2.4037339556592765,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017288078922904002,
      "loss": 0.057,
      "step": 5150
    },
    {
      "epoch": 2.4042007001166863,
      "grad_norm": 0.984375,
      "learning_rate": 0.00017287074124333005,
      "loss": 0.0794,
      "step": 5151
    },
    {
      "epoch": 2.4046674445740956,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017286069168863215,
      "loss": 0.0641,
      "step": 5152
    },
    {
      "epoch": 2.4051341890315054,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017285064056516273,
      "loss": 0.0615,
      "step": 5153
    },
    {
      "epoch": 2.405600933488915,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001728405878731382,
      "loss": 0.048,
      "step": 5154
    },
    {
      "epoch": 2.406067677946324,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017283053361277494,
      "loss": 0.0531,
      "step": 5155
    },
    {
      "epoch": 2.406534422403734,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017282047778428952,
      "loss": 0.1063,
      "step": 5156
    },
    {
      "epoch": 2.4070011668611437,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00017281042038789842,
      "loss": 0.0697,
      "step": 5157
    },
    {
      "epoch": 2.407467911318553,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001728003614238182,
      "loss": 0.0538,
      "step": 5158
    },
    {
      "epoch": 2.4079346557759624,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017279030089226543,
      "loss": 0.0786,
      "step": 5159
    },
    {
      "epoch": 2.4084014002333722,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017278023879345667,
      "loss": 0.06,
      "step": 5160
    },
    {
      "epoch": 2.408868144690782,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017277017512760864,
      "loss": 0.0492,
      "step": 5161
    },
    {
      "epoch": 2.4093348891481914,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017276010989493803,
      "loss": 0.06,
      "step": 5162
    },
    {
      "epoch": 2.4098016336056007,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001727500430956615,
      "loss": 0.0706,
      "step": 5163
    },
    {
      "epoch": 2.4102683780630105,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017273997472999581,
      "loss": 0.0553,
      "step": 5164
    },
    {
      "epoch": 2.41073512252042,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017272990479815778,
      "loss": 0.0594,
      "step": 5165
    },
    {
      "epoch": 2.4112018669778297,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001727198333003642,
      "loss": 0.077,
      "step": 5166
    },
    {
      "epoch": 2.411668611435239,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00017270976023683193,
      "loss": 0.0717,
      "step": 5167
    },
    {
      "epoch": 2.412135355892649,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017269968560777783,
      "loss": 0.0601,
      "step": 5168
    },
    {
      "epoch": 2.412602100350058,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017268960941341886,
      "loss": 0.0523,
      "step": 5169
    },
    {
      "epoch": 2.413068844807468,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017267953165397195,
      "loss": 0.0774,
      "step": 5170
    },
    {
      "epoch": 2.4135355892648773,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00017266945232965406,
      "loss": 0.0443,
      "step": 5171
    },
    {
      "epoch": 2.414002333722287,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017265937144068225,
      "loss": 0.0898,
      "step": 5172
    },
    {
      "epoch": 2.4144690781796965,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017264928898727354,
      "loss": 0.0599,
      "step": 5173
    },
    {
      "epoch": 2.4149358226371063,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017263920496964506,
      "loss": 0.0671,
      "step": 5174
    },
    {
      "epoch": 2.4154025670945156,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017262911938801387,
      "loss": 0.0822,
      "step": 5175
    },
    {
      "epoch": 2.4158693115519254,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017261903224259718,
      "loss": 0.0729,
      "step": 5176
    },
    {
      "epoch": 2.4163360560093348,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017260894353361215,
      "loss": 0.0586,
      "step": 5177
    },
    {
      "epoch": 2.4168028004667446,
      "grad_norm": 0.80078125,
      "learning_rate": 0.000172598853261276,
      "loss": 0.0551,
      "step": 5178
    },
    {
      "epoch": 2.417269544924154,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017258876142580597,
      "loss": 0.0615,
      "step": 5179
    },
    {
      "epoch": 2.4177362893815637,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017257866802741943,
      "loss": 0.0617,
      "step": 5180
    },
    {
      "epoch": 2.418203033838973,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00017256857306633357,
      "loss": 0.0567,
      "step": 5181
    },
    {
      "epoch": 2.418669778296383,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00017255847654276584,
      "loss": 0.0519,
      "step": 5182
    },
    {
      "epoch": 2.419136522753792,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001725483784569336,
      "loss": 0.0684,
      "step": 5183
    },
    {
      "epoch": 2.419603267211202,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001725382788090543,
      "loss": 0.05,
      "step": 5184
    },
    {
      "epoch": 2.4200700116686114,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017252817759934536,
      "loss": 0.074,
      "step": 5185
    },
    {
      "epoch": 2.420536756126021,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001725180748280243,
      "loss": 0.0489,
      "step": 5186
    },
    {
      "epoch": 2.4210035005834305,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00017250797049530862,
      "loss": 0.0472,
      "step": 5187
    },
    {
      "epoch": 2.4214702450408403,
      "grad_norm": 1.140625,
      "learning_rate": 0.00017249786460141588,
      "loss": 0.0988,
      "step": 5188
    },
    {
      "epoch": 2.4219369894982496,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017248775714656366,
      "loss": 0.057,
      "step": 5189
    },
    {
      "epoch": 2.4224037339556594,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017247764813096964,
      "loss": 0.0608,
      "step": 5190
    },
    {
      "epoch": 2.422870478413069,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017246753755485142,
      "loss": 0.0675,
      "step": 5191
    },
    {
      "epoch": 2.4233372228704786,
      "grad_norm": 0.71875,
      "learning_rate": 0.00017245742541842672,
      "loss": 0.0563,
      "step": 5192
    },
    {
      "epoch": 2.423803967327888,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017244731172191325,
      "loss": 0.0689,
      "step": 5193
    },
    {
      "epoch": 2.4242707117852977,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00017243719646552877,
      "loss": 0.0448,
      "step": 5194
    },
    {
      "epoch": 2.424737456242707,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017242707964949108,
      "loss": 0.0676,
      "step": 5195
    },
    {
      "epoch": 2.425204200700117,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00017241696127401804,
      "loss": 0.0757,
      "step": 5196
    },
    {
      "epoch": 2.4256709451575262,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017240684133932742,
      "loss": 0.0551,
      "step": 5197
    },
    {
      "epoch": 2.4261376896149356,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017239671984563718,
      "loss": 0.0635,
      "step": 5198
    },
    {
      "epoch": 2.4266044340723454,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017238659679316527,
      "loss": 0.0745,
      "step": 5199
    },
    {
      "epoch": 2.427071178529755,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017237647218212962,
      "loss": 0.0703,
      "step": 5200
    },
    {
      "epoch": 2.4275379229871645,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017236634601274818,
      "loss": 0.0778,
      "step": 5201
    },
    {
      "epoch": 2.428004667444574,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017235621828523903,
      "loss": 0.0621,
      "step": 5202
    },
    {
      "epoch": 2.4284714119019837,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00017234608899982024,
      "loss": 0.0669,
      "step": 5203
    },
    {
      "epoch": 2.4289381563593935,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001723359581567099,
      "loss": 0.0411,
      "step": 5204
    },
    {
      "epoch": 2.429404900816803,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00017232582575612608,
      "loss": 0.0672,
      "step": 5205
    },
    {
      "epoch": 2.429871645274212,
      "grad_norm": 0.9140625,
      "learning_rate": 0.000172315691798287,
      "loss": 0.0751,
      "step": 5206
    },
    {
      "epoch": 2.430338389731622,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017230555628341084,
      "loss": 0.0859,
      "step": 5207
    },
    {
      "epoch": 2.4308051341890313,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017229541921171584,
      "loss": 0.0555,
      "step": 5208
    },
    {
      "epoch": 2.431271878646441,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017228528058342026,
      "loss": 0.0553,
      "step": 5209
    },
    {
      "epoch": 2.4317386231038505,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00017227514039874238,
      "loss": 0.0888,
      "step": 5210
    },
    {
      "epoch": 2.4322053675612603,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017226499865790048,
      "loss": 0.0775,
      "step": 5211
    },
    {
      "epoch": 2.4326721120186696,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017225485536111304,
      "loss": 0.0569,
      "step": 5212
    },
    {
      "epoch": 2.4331388564760794,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017224471050859838,
      "loss": 0.0583,
      "step": 5213
    },
    {
      "epoch": 2.4336056009334888,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017223456410057496,
      "loss": 0.0659,
      "step": 5214
    },
    {
      "epoch": 2.4340723453908986,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017222441613726122,
      "loss": 0.0598,
      "step": 5215
    },
    {
      "epoch": 2.434539089848308,
      "grad_norm": 0.71875,
      "learning_rate": 0.00017221426661887565,
      "loss": 0.043,
      "step": 5216
    },
    {
      "epoch": 2.4350058343057177,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00017220411554563683,
      "loss": 0.0484,
      "step": 5217
    },
    {
      "epoch": 2.435472578763127,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00017219396291776328,
      "loss": 0.0558,
      "step": 5218
    },
    {
      "epoch": 2.435939323220537,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001721838087354736,
      "loss": 0.0521,
      "step": 5219
    },
    {
      "epoch": 2.436406067677946,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001721736529989864,
      "loss": 0.0544,
      "step": 5220
    },
    {
      "epoch": 2.436872812135356,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001721634957085204,
      "loss": 0.0556,
      "step": 5221
    },
    {
      "epoch": 2.4373395565927654,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001721533368642943,
      "loss": 0.0515,
      "step": 5222
    },
    {
      "epoch": 2.437806301050175,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00017214317646652676,
      "loss": 0.0671,
      "step": 5223
    },
    {
      "epoch": 2.4382730455075845,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001721330145154366,
      "loss": 0.0594,
      "step": 5224
    },
    {
      "epoch": 2.4387397899649943,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017212285101124257,
      "loss": 0.0571,
      "step": 5225
    },
    {
      "epoch": 2.4392065344224036,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017211268595416362,
      "loss": 0.0404,
      "step": 5226
    },
    {
      "epoch": 2.4396732788798134,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017210251934441845,
      "loss": 0.0658,
      "step": 5227
    },
    {
      "epoch": 2.440140023337223,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017209235118222606,
      "loss": 0.0516,
      "step": 5228
    },
    {
      "epoch": 2.4406067677946326,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001720821814678054,
      "loss": 0.0503,
      "step": 5229
    },
    {
      "epoch": 2.441073512252042,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017207201020137536,
      "loss": 0.0556,
      "step": 5230
    },
    {
      "epoch": 2.4415402567094517,
      "grad_norm": 1.140625,
      "learning_rate": 0.00017206183738315497,
      "loss": 0.0598,
      "step": 5231
    },
    {
      "epoch": 2.442007001166861,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017205166301336328,
      "loss": 0.0403,
      "step": 5232
    },
    {
      "epoch": 2.442473745624271,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017204148709221934,
      "loss": 0.0492,
      "step": 5233
    },
    {
      "epoch": 2.4429404900816802,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00017203130961994224,
      "loss": 0.0676,
      "step": 5234
    },
    {
      "epoch": 2.44340723453909,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017202113059675114,
      "loss": 0.075,
      "step": 5235
    },
    {
      "epoch": 2.4438739789964994,
      "grad_norm": 0.6875,
      "learning_rate": 0.00017201095002286517,
      "loss": 0.0471,
      "step": 5236
    },
    {
      "epoch": 2.4443407234539087,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017200076789850358,
      "loss": 0.0461,
      "step": 5237
    },
    {
      "epoch": 2.4448074679113185,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017199058422388556,
      "loss": 0.072,
      "step": 5238
    },
    {
      "epoch": 2.4452742123687283,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001719803989992304,
      "loss": 0.0578,
      "step": 5239
    },
    {
      "epoch": 2.4457409568261377,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017197021222475735,
      "loss": 0.0646,
      "step": 5240
    },
    {
      "epoch": 2.446207701283547,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017196002390068582,
      "loss": 0.0614,
      "step": 5241
    },
    {
      "epoch": 2.446674445740957,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001719498340272351,
      "loss": 0.0543,
      "step": 5242
    },
    {
      "epoch": 2.4471411901983666,
      "grad_norm": 1.03125,
      "learning_rate": 0.00017193964260462462,
      "loss": 0.0681,
      "step": 5243
    },
    {
      "epoch": 2.447607934655776,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017192944963307386,
      "loss": 0.0637,
      "step": 5244
    },
    {
      "epoch": 2.4480746791131853,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001719192551128022,
      "loss": 0.0578,
      "step": 5245
    },
    {
      "epoch": 2.448541423570595,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001719090590440292,
      "loss": 0.0662,
      "step": 5246
    },
    {
      "epoch": 2.449008168028005,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00017189886142697435,
      "loss": 0.0578,
      "step": 5247
    },
    {
      "epoch": 2.4494749124854143,
      "grad_norm": 1.1328125,
      "learning_rate": 0.00017188866226185726,
      "loss": 0.0735,
      "step": 5248
    },
    {
      "epoch": 2.4499416569428236,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017187846154889753,
      "loss": 0.0614,
      "step": 5249
    },
    {
      "epoch": 2.4504084014002334,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00017186825928831474,
      "loss": 0.0793,
      "step": 5250
    },
    {
      "epoch": 2.4508751458576428,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001718580554803286,
      "loss": 0.0592,
      "step": 5251
    },
    {
      "epoch": 2.4513418903150526,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001718478501251588,
      "loss": 0.047,
      "step": 5252
    },
    {
      "epoch": 2.451808634772462,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017183764322302507,
      "loss": 0.059,
      "step": 5253
    },
    {
      "epoch": 2.4522753792298717,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017182743477414718,
      "loss": 0.0442,
      "step": 5254
    },
    {
      "epoch": 2.452742123687281,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001718172247787449,
      "loss": 0.0611,
      "step": 5255
    },
    {
      "epoch": 2.453208868144691,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00017180701323703812,
      "loss": 0.0488,
      "step": 5256
    },
    {
      "epoch": 2.4536756126021,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017179680014924666,
      "loss": 0.0557,
      "step": 5257
    },
    {
      "epoch": 2.45414235705951,
      "grad_norm": 1.3125,
      "learning_rate": 0.00017178658551559044,
      "loss": 0.0713,
      "step": 5258
    },
    {
      "epoch": 2.4546091015169194,
      "grad_norm": 1.046875,
      "learning_rate": 0.00017177636933628937,
      "loss": 0.0774,
      "step": 5259
    },
    {
      "epoch": 2.455075845974329,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00017176615161156345,
      "loss": 0.0461,
      "step": 5260
    },
    {
      "epoch": 2.4555425904317385,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00017175593234163266,
      "loss": 0.0747,
      "step": 5261
    },
    {
      "epoch": 2.4560093348891483,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017174571152671702,
      "loss": 0.0515,
      "step": 5262
    },
    {
      "epoch": 2.4564760793465577,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00017173548916703658,
      "loss": 0.0812,
      "step": 5263
    },
    {
      "epoch": 2.4569428238039674,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001717252652628115,
      "loss": 0.0336,
      "step": 5264
    },
    {
      "epoch": 2.457409568261377,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00017171503981426188,
      "loss": 0.0541,
      "step": 5265
    },
    {
      "epoch": 2.4578763127187866,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017170481282160788,
      "loss": 0.066,
      "step": 5266
    },
    {
      "epoch": 2.458343057176196,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00017169458428506972,
      "loss": 0.0436,
      "step": 5267
    },
    {
      "epoch": 2.4588098016336057,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017168435420486758,
      "loss": 0.065,
      "step": 5268
    },
    {
      "epoch": 2.459276546091015,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017167412258122178,
      "loss": 0.072,
      "step": 5269
    },
    {
      "epoch": 2.459743290548425,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001716638894143526,
      "loss": 0.0574,
      "step": 5270
    },
    {
      "epoch": 2.4602100350058342,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00017165365470448035,
      "loss": 0.0639,
      "step": 5271
    },
    {
      "epoch": 2.460676779463244,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001716434184518254,
      "loss": 0.0708,
      "step": 5272
    },
    {
      "epoch": 2.4611435239206534,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017163318065660822,
      "loss": 0.0587,
      "step": 5273
    },
    {
      "epoch": 2.461610268378063,
      "grad_norm": 0.96875,
      "learning_rate": 0.00017162294131904912,
      "loss": 0.0595,
      "step": 5274
    },
    {
      "epoch": 2.4620770128354725,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00017161270043936867,
      "loss": 0.074,
      "step": 5275
    },
    {
      "epoch": 2.4625437572928823,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001716024580177873,
      "loss": 0.0533,
      "step": 5276
    },
    {
      "epoch": 2.4630105017502917,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017159221405452555,
      "loss": 0.0574,
      "step": 5277
    },
    {
      "epoch": 2.4634772462077015,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00017158196854980405,
      "loss": 0.0439,
      "step": 5278
    },
    {
      "epoch": 2.463943990665111,
      "grad_norm": 0.875,
      "learning_rate": 0.0001715717215038433,
      "loss": 0.0696,
      "step": 5279
    },
    {
      "epoch": 2.46441073512252,
      "grad_norm": 0.90234375,
      "learning_rate": 0.000171561472916864,
      "loss": 0.0699,
      "step": 5280
    },
    {
      "epoch": 2.46487747957993,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00017155122278908675,
      "loss": 0.0797,
      "step": 5281
    },
    {
      "epoch": 2.4653442240373398,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001715409711207323,
      "loss": 0.0491,
      "step": 5282
    },
    {
      "epoch": 2.465810968494749,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001715307179120214,
      "loss": 0.0368,
      "step": 5283
    },
    {
      "epoch": 2.4662777129521585,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017152046316317476,
      "loss": 0.0576,
      "step": 5284
    },
    {
      "epoch": 2.4667444574095683,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001715102068744132,
      "loss": 0.0485,
      "step": 5285
    },
    {
      "epoch": 2.467211201866978,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017149994904595755,
      "loss": 0.0546,
      "step": 5286
    },
    {
      "epoch": 2.4676779463243874,
      "grad_norm": 1.15625,
      "learning_rate": 0.00017148968967802862,
      "loss": 0.0798,
      "step": 5287
    },
    {
      "epoch": 2.4681446907817968,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001714794287708474,
      "loss": 0.0587,
      "step": 5288
    },
    {
      "epoch": 2.4686114352392066,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017146916632463475,
      "loss": 0.0747,
      "step": 5289
    },
    {
      "epoch": 2.469078179696616,
      "grad_norm": 1.1640625,
      "learning_rate": 0.00017145890233961165,
      "loss": 0.0692,
      "step": 5290
    },
    {
      "epoch": 2.4695449241540257,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00017144863681599915,
      "loss": 0.0383,
      "step": 5291
    },
    {
      "epoch": 2.470011668611435,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017143836975401814,
      "loss": 0.0607,
      "step": 5292
    },
    {
      "epoch": 2.470478413068845,
      "grad_norm": 1.0,
      "learning_rate": 0.00017142810115388984,
      "loss": 0.0635,
      "step": 5293
    },
    {
      "epoch": 2.470945157526254,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017141783101583524,
      "loss": 0.0599,
      "step": 5294
    },
    {
      "epoch": 2.471411901983664,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001714075593400755,
      "loss": 0.0519,
      "step": 5295
    },
    {
      "epoch": 2.4718786464410734,
      "grad_norm": 0.921875,
      "learning_rate": 0.00017139728612683178,
      "loss": 0.0519,
      "step": 5296
    },
    {
      "epoch": 2.472345390898483,
      "grad_norm": 0.71875,
      "learning_rate": 0.00017138701137632526,
      "loss": 0.0421,
      "step": 5297
    },
    {
      "epoch": 2.4728121353558925,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001713767350887772,
      "loss": 0.0489,
      "step": 5298
    },
    {
      "epoch": 2.4732788798133023,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017136645726440882,
      "loss": 0.0646,
      "step": 5299
    },
    {
      "epoch": 2.4737456242707117,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017135617790344144,
      "loss": 0.0674,
      "step": 5300
    },
    {
      "epoch": 2.4742123687281214,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017134589700609637,
      "loss": 0.049,
      "step": 5301
    },
    {
      "epoch": 2.474679113185531,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017133561457259502,
      "loss": 0.0606,
      "step": 5302
    },
    {
      "epoch": 2.4751458576429406,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001713253306031587,
      "loss": 0.0674,
      "step": 5303
    },
    {
      "epoch": 2.47561260210035,
      "grad_norm": 0.875,
      "learning_rate": 0.0001713150450980089,
      "loss": 0.0571,
      "step": 5304
    },
    {
      "epoch": 2.4760793465577597,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017130475805736704,
      "loss": 0.0726,
      "step": 5305
    },
    {
      "epoch": 2.476546091015169,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001712944694814546,
      "loss": 0.0803,
      "step": 5306
    },
    {
      "epoch": 2.477012835472579,
      "grad_norm": 0.953125,
      "learning_rate": 0.00017128417937049316,
      "loss": 0.0617,
      "step": 5307
    },
    {
      "epoch": 2.4774795799299882,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017127388772470426,
      "loss": 0.066,
      "step": 5308
    },
    {
      "epoch": 2.477946324387398,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017126359454430946,
      "loss": 0.0775,
      "step": 5309
    },
    {
      "epoch": 2.4784130688448074,
      "grad_norm": 1.109375,
      "learning_rate": 0.00017125329982953038,
      "loss": 0.1242,
      "step": 5310
    },
    {
      "epoch": 2.478879813302217,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017124300358058875,
      "loss": 0.0517,
      "step": 5311
    },
    {
      "epoch": 2.4793465577596265,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017123270579770617,
      "loss": 0.0606,
      "step": 5312
    },
    {
      "epoch": 2.4798133022170363,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001712224064811044,
      "loss": 0.0798,
      "step": 5313
    },
    {
      "epoch": 2.4802800466744457,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001712121056310052,
      "loss": 0.0731,
      "step": 5314
    },
    {
      "epoch": 2.4807467911318555,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017120180324763036,
      "loss": 0.0734,
      "step": 5315
    },
    {
      "epoch": 2.481213535589265,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001711914993312017,
      "loss": 0.0571,
      "step": 5316
    },
    {
      "epoch": 2.4816802800466746,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00017118119388194103,
      "loss": 0.0629,
      "step": 5317
    },
    {
      "epoch": 2.482147024504084,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017117088690007031,
      "loss": 0.0505,
      "step": 5318
    },
    {
      "epoch": 2.4826137689614938,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017116057838581145,
      "loss": 0.0716,
      "step": 5319
    },
    {
      "epoch": 2.483080513418903,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017115026833938637,
      "loss": 0.08,
      "step": 5320
    },
    {
      "epoch": 2.483547257876313,
      "grad_norm": 0.765625,
      "learning_rate": 0.00017113995676101707,
      "loss": 0.0535,
      "step": 5321
    },
    {
      "epoch": 2.4840140023337223,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017112964365092553,
      "loss": 0.0641,
      "step": 5322
    },
    {
      "epoch": 2.4844807467911316,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001711193290093339,
      "loss": 0.0483,
      "step": 5323
    },
    {
      "epoch": 2.4849474912485414,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001711090128364642,
      "loss": 0.0472,
      "step": 5324
    },
    {
      "epoch": 2.485414235705951,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017109869513253853,
      "loss": 0.0444,
      "step": 5325
    },
    {
      "epoch": 2.4858809801633606,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001710883758977791,
      "loss": 0.0615,
      "step": 5326
    },
    {
      "epoch": 2.48634772462077,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00017107805513240802,
      "loss": 0.0579,
      "step": 5327
    },
    {
      "epoch": 2.4868144690781797,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00017106773283664757,
      "loss": 0.0578,
      "step": 5328
    },
    {
      "epoch": 2.4872812135355895,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00017105740901071997,
      "loss": 0.0749,
      "step": 5329
    },
    {
      "epoch": 2.487747957992999,
      "grad_norm": 0.90625,
      "learning_rate": 0.00017104708365484755,
      "loss": 0.0622,
      "step": 5330
    },
    {
      "epoch": 2.488214702450408,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017103675676925258,
      "loss": 0.0419,
      "step": 5331
    },
    {
      "epoch": 2.488681446907818,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001710264283541574,
      "loss": 0.078,
      "step": 5332
    },
    {
      "epoch": 2.4891481913652274,
      "grad_norm": 1.0,
      "learning_rate": 0.00017101609840978443,
      "loss": 0.0649,
      "step": 5333
    },
    {
      "epoch": 2.489614935822637,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017100576693635606,
      "loss": 0.0548,
      "step": 5334
    },
    {
      "epoch": 2.4900816802800465,
      "grad_norm": 1.203125,
      "learning_rate": 0.00017099543393409475,
      "loss": 0.0763,
      "step": 5335
    },
    {
      "epoch": 2.4905484247374563,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00017098509940322295,
      "loss": 0.077,
      "step": 5336
    },
    {
      "epoch": 2.4910151691948657,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017097476334396324,
      "loss": 0.051,
      "step": 5337
    },
    {
      "epoch": 2.4914819136522754,
      "grad_norm": 1.390625,
      "learning_rate": 0.0001709644257565381,
      "loss": 0.0761,
      "step": 5338
    },
    {
      "epoch": 2.491948658109685,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00017095408664117017,
      "loss": 0.0751,
      "step": 5339
    },
    {
      "epoch": 2.4924154025670946,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017094374599808197,
      "loss": 0.0466,
      "step": 5340
    },
    {
      "epoch": 2.492882147024504,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017093340382749625,
      "loss": 0.0616,
      "step": 5341
    },
    {
      "epoch": 2.4933488914819137,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017092306012963566,
      "loss": 0.0534,
      "step": 5342
    },
    {
      "epoch": 2.493815635939323,
      "grad_norm": 0.8125,
      "learning_rate": 0.00017091271490472286,
      "loss": 0.0509,
      "step": 5343
    },
    {
      "epoch": 2.494282380396733,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001709023681529806,
      "loss": 0.0704,
      "step": 5344
    },
    {
      "epoch": 2.4947491248541422,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017089201987463177,
      "loss": 0.0707,
      "step": 5345
    },
    {
      "epoch": 2.495215869311552,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00017088167006989907,
      "loss": 0.0777,
      "step": 5346
    },
    {
      "epoch": 2.4956826137689614,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00017087131873900533,
      "loss": 0.0589,
      "step": 5347
    },
    {
      "epoch": 2.496149358226371,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017086096588217348,
      "loss": 0.0703,
      "step": 5348
    },
    {
      "epoch": 2.4966161026837805,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001708506114996264,
      "loss": 0.0638,
      "step": 5349
    },
    {
      "epoch": 2.4970828471411903,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017084025559158708,
      "loss": 0.0695,
      "step": 5350
    },
    {
      "epoch": 2.4975495915985997,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017082989815827846,
      "loss": 0.0515,
      "step": 5351
    },
    {
      "epoch": 2.4980163360560095,
      "grad_norm": 0.93359375,
      "learning_rate": 0.0001708195391999235,
      "loss": 0.0563,
      "step": 5352
    },
    {
      "epoch": 2.498483080513419,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001708091787167453,
      "loss": 0.0621,
      "step": 5353
    },
    {
      "epoch": 2.4989498249708286,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00017079881670896695,
      "loss": 0.0546,
      "step": 5354
    },
    {
      "epoch": 2.499416569428238,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001707884531768115,
      "loss": 0.0555,
      "step": 5355
    },
    {
      "epoch": 2.4998833138856478,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00017077808812050212,
      "loss": 0.0513,
      "step": 5356
    },
    {
      "epoch": 2.500350058343057,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00017076772154026197,
      "loss": 0.0603,
      "step": 5357
    },
    {
      "epoch": 2.5008168028004665,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017075735343631426,
      "loss": 0.0488,
      "step": 5358
    },
    {
      "epoch": 2.5012835472578763,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001707469838088822,
      "loss": 0.0732,
      "step": 5359
    },
    {
      "epoch": 2.501750291715286,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017073661265818913,
      "loss": 0.0669,
      "step": 5360
    },
    {
      "epoch": 2.501750291715286,
      "eval_loss": 1.6183894872665405,
      "eval_runtime": 95.4633,
      "eval_samples_per_second": 18.897,
      "eval_steps_per_second": 2.367,
      "step": 5360
    },
    {
      "epoch": 2.5022170361726954,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00017072623998445828,
      "loss": 0.0597,
      "step": 5361
    },
    {
      "epoch": 2.5026837806301048,
      "grad_norm": 0.94140625,
      "learning_rate": 0.000170715865787913,
      "loss": 0.0476,
      "step": 5362
    },
    {
      "epoch": 2.5031505250875146,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017070549006877668,
      "loss": 0.0541,
      "step": 5363
    },
    {
      "epoch": 2.5036172695449244,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001706951128272727,
      "loss": 0.0869,
      "step": 5364
    },
    {
      "epoch": 2.5040840140023337,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00017068473406362452,
      "loss": 0.0514,
      "step": 5365
    },
    {
      "epoch": 2.504550758459743,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001706743537780556,
      "loss": 0.0578,
      "step": 5366
    },
    {
      "epoch": 2.505017502917153,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00017066397197078938,
      "loss": 0.0502,
      "step": 5367
    },
    {
      "epoch": 2.5054842473745627,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001706535886420495,
      "loss": 0.0677,
      "step": 5368
    },
    {
      "epoch": 2.505950991831972,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00017064320379205943,
      "loss": 0.0499,
      "step": 5369
    },
    {
      "epoch": 2.5064177362893814,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001706328174210428,
      "loss": 0.0562,
      "step": 5370
    },
    {
      "epoch": 2.506884480746791,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00017062242952922325,
      "loss": 0.0736,
      "step": 5371
    },
    {
      "epoch": 2.507351225204201,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017061204011682444,
      "loss": 0.0964,
      "step": 5372
    },
    {
      "epoch": 2.5078179696616103,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00017060164918407007,
      "loss": 0.0778,
      "step": 5373
    },
    {
      "epoch": 2.5082847141190197,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001705912567311838,
      "loss": 0.063,
      "step": 5374
    },
    {
      "epoch": 2.5087514585764294,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017058086275838953,
      "loss": 0.0528,
      "step": 5375
    },
    {
      "epoch": 2.5092182030338392,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00017057046726591096,
      "loss": 0.0455,
      "step": 5376
    },
    {
      "epoch": 2.5096849474912486,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001705600702539719,
      "loss": 0.0612,
      "step": 5377
    },
    {
      "epoch": 2.510151691948658,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017054967172279627,
      "loss": 0.0565,
      "step": 5378
    },
    {
      "epoch": 2.5106184364060677,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00017053927167260787,
      "loss": 0.0726,
      "step": 5379
    },
    {
      "epoch": 2.511085180863477,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017052887010363075,
      "loss": 0.0542,
      "step": 5380
    },
    {
      "epoch": 2.511551925320887,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001705184670160888,
      "loss": 0.0625,
      "step": 5381
    },
    {
      "epoch": 2.5120186697782962,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017050806241020595,
      "loss": 0.062,
      "step": 5382
    },
    {
      "epoch": 2.512485414235706,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00017049765628620635,
      "loss": 0.0724,
      "step": 5383
    },
    {
      "epoch": 2.5129521586931154,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00017048724864431397,
      "loss": 0.0536,
      "step": 5384
    },
    {
      "epoch": 2.513418903150525,
      "grad_norm": 0.65625,
      "learning_rate": 0.00017047683948475292,
      "loss": 0.0437,
      "step": 5385
    },
    {
      "epoch": 2.5138856476079345,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00017046642880774732,
      "loss": 0.0483,
      "step": 5386
    },
    {
      "epoch": 2.5143523920653443,
      "grad_norm": 1.015625,
      "learning_rate": 0.00017045601661352133,
      "loss": 0.0785,
      "step": 5387
    },
    {
      "epoch": 2.5148191365227537,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00017044560290229913,
      "loss": 0.0582,
      "step": 5388
    },
    {
      "epoch": 2.5152858809801635,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017043518767430492,
      "loss": 0.0527,
      "step": 5389
    },
    {
      "epoch": 2.515752625437573,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000170424770929763,
      "loss": 0.0409,
      "step": 5390
    },
    {
      "epoch": 2.5162193698949826,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00017041435266889764,
      "loss": 0.061,
      "step": 5391
    },
    {
      "epoch": 2.516686114352392,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00017040393289193307,
      "loss": 0.0566,
      "step": 5392
    },
    {
      "epoch": 2.5171528588098018,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00017039351159909376,
      "loss": 0.0598,
      "step": 5393
    },
    {
      "epoch": 2.517619603267211,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00017038308879060402,
      "loss": 0.0447,
      "step": 5394
    },
    {
      "epoch": 2.518086347724621,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001703726644666883,
      "loss": 0.0649,
      "step": 5395
    },
    {
      "epoch": 2.5185530921820303,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00017036223862757107,
      "loss": 0.0762,
      "step": 5396
    },
    {
      "epoch": 2.51901983663944,
      "grad_norm": 1.0,
      "learning_rate": 0.00017035181127347674,
      "loss": 0.0653,
      "step": 5397
    },
    {
      "epoch": 2.5194865810968494,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001703413824046298,
      "loss": 0.0448,
      "step": 5398
    },
    {
      "epoch": 2.519953325554259,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00017033095202125497,
      "loss": 0.0515,
      "step": 5399
    },
    {
      "epoch": 2.5204200700116686,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00017032052012357665,
      "loss": 0.0471,
      "step": 5400
    },
    {
      "epoch": 2.520886814469078,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001703100867118195,
      "loss": 0.0592,
      "step": 5401
    },
    {
      "epoch": 2.5213535589264877,
      "grad_norm": 1.0,
      "learning_rate": 0.00017029965178620818,
      "loss": 0.0593,
      "step": 5402
    },
    {
      "epoch": 2.5218203033838975,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00017028921534696735,
      "loss": 0.0473,
      "step": 5403
    },
    {
      "epoch": 2.522287047841307,
      "grad_norm": 0.734375,
      "learning_rate": 0.00017027877739432174,
      "loss": 0.0544,
      "step": 5404
    },
    {
      "epoch": 2.522753792298716,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001702683379284961,
      "loss": 0.0652,
      "step": 5405
    },
    {
      "epoch": 2.523220536756126,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001702578969497151,
      "loss": 0.0762,
      "step": 5406
    },
    {
      "epoch": 2.523687281213536,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00017024745445820368,
      "loss": 0.0533,
      "step": 5407
    },
    {
      "epoch": 2.524154025670945,
      "grad_norm": 0.84375,
      "learning_rate": 0.00017023701045418662,
      "loss": 0.0449,
      "step": 5408
    },
    {
      "epoch": 2.5246207701283545,
      "grad_norm": 1.0625,
      "learning_rate": 0.00017022656493788883,
      "loss": 0.0713,
      "step": 5409
    },
    {
      "epoch": 2.5250875145857643,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00017021611790953513,
      "loss": 0.0558,
      "step": 5410
    },
    {
      "epoch": 2.525554259043174,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017020566936935052,
      "loss": 0.0548,
      "step": 5411
    },
    {
      "epoch": 2.5260210035005835,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00017019521931755996,
      "loss": 0.0438,
      "step": 5412
    },
    {
      "epoch": 2.526487747957993,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00017018476775438845,
      "loss": 0.0497,
      "step": 5413
    },
    {
      "epoch": 2.5269544924154026,
      "grad_norm": 0.98046875,
      "learning_rate": 0.000170174314680061,
      "loss": 0.0596,
      "step": 5414
    },
    {
      "epoch": 2.5274212368728124,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001701638600948027,
      "loss": 0.0527,
      "step": 5415
    },
    {
      "epoch": 2.5278879813302217,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00017015340399883865,
      "loss": 0.064,
      "step": 5416
    },
    {
      "epoch": 2.528354725787631,
      "grad_norm": 0.859375,
      "learning_rate": 0.00017014294639239398,
      "loss": 0.053,
      "step": 5417
    },
    {
      "epoch": 2.528821470245041,
      "grad_norm": 0.890625,
      "learning_rate": 0.00017013248727569384,
      "loss": 0.05,
      "step": 5418
    },
    {
      "epoch": 2.5292882147024502,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00017012202664896343,
      "loss": 0.0737,
      "step": 5419
    },
    {
      "epoch": 2.52975495915986,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000170111564512428,
      "loss": 0.0449,
      "step": 5420
    },
    {
      "epoch": 2.5302217036172694,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00017010110086631278,
      "loss": 0.0986,
      "step": 5421
    },
    {
      "epoch": 2.530688448074679,
      "grad_norm": 0.9375,
      "learning_rate": 0.00017009063571084312,
      "loss": 0.0601,
      "step": 5422
    },
    {
      "epoch": 2.5311551925320885,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00017008016904624426,
      "loss": 0.0453,
      "step": 5423
    },
    {
      "epoch": 2.5316219369894983,
      "grad_norm": 1.09375,
      "learning_rate": 0.00017006970087274164,
      "loss": 0.0693,
      "step": 5424
    },
    {
      "epoch": 2.5320886814469077,
      "grad_norm": 1.078125,
      "learning_rate": 0.00017005923119056058,
      "loss": 0.0729,
      "step": 5425
    },
    {
      "epoch": 2.5325554259043175,
      "grad_norm": 0.796875,
      "learning_rate": 0.00017004875999992658,
      "loss": 0.0606,
      "step": 5426
    },
    {
      "epoch": 2.533022170361727,
      "grad_norm": 0.703125,
      "learning_rate": 0.00017003828730106503,
      "loss": 0.0489,
      "step": 5427
    },
    {
      "epoch": 2.5334889148191366,
      "grad_norm": 0.828125,
      "learning_rate": 0.00017002781309420146,
      "loss": 0.0502,
      "step": 5428
    },
    {
      "epoch": 2.533955659276546,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00017001733737956136,
      "loss": 0.0506,
      "step": 5429
    },
    {
      "epoch": 2.5344224037339558,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001700068601573703,
      "loss": 0.0495,
      "step": 5430
    },
    {
      "epoch": 2.534889148191365,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016999638142785388,
      "loss": 0.0657,
      "step": 5431
    },
    {
      "epoch": 2.535355892648775,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001699859011912377,
      "loss": 0.0513,
      "step": 5432
    },
    {
      "epoch": 2.5358226371061843,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016997541944774742,
      "loss": 0.0751,
      "step": 5433
    },
    {
      "epoch": 2.536289381563594,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016996493619760868,
      "loss": 0.0746,
      "step": 5434
    },
    {
      "epoch": 2.5367561260210034,
      "grad_norm": 0.75,
      "learning_rate": 0.0001699544514410473,
      "loss": 0.0354,
      "step": 5435
    },
    {
      "epoch": 2.537222870478413,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00016994396517828895,
      "loss": 0.0535,
      "step": 5436
    },
    {
      "epoch": 2.5376896149358226,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001699334774095594,
      "loss": 0.0517,
      "step": 5437
    },
    {
      "epoch": 2.5381563593932324,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001699229881350845,
      "loss": 0.1052,
      "step": 5438
    },
    {
      "epoch": 2.5386231038506417,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001699124973550901,
      "loss": 0.07,
      "step": 5439
    },
    {
      "epoch": 2.539089848308051,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00016990200506980204,
      "loss": 0.0776,
      "step": 5440
    },
    {
      "epoch": 2.539556592765461,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00016989151127944627,
      "loss": 0.0444,
      "step": 5441
    },
    {
      "epoch": 2.5400233372228707,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001698810159842487,
      "loss": 0.0634,
      "step": 5442
    },
    {
      "epoch": 2.54049008168028,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016987051918443532,
      "loss": 0.0574,
      "step": 5443
    },
    {
      "epoch": 2.5409568261376894,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00016986002088023217,
      "loss": 0.0511,
      "step": 5444
    },
    {
      "epoch": 2.541423570595099,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016984952107186523,
      "loss": 0.0408,
      "step": 5445
    },
    {
      "epoch": 2.541890315052509,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001698390197595606,
      "loss": 0.0474,
      "step": 5446
    },
    {
      "epoch": 2.5423570595099183,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00016982851694354444,
      "loss": 0.0418,
      "step": 5447
    },
    {
      "epoch": 2.5428238039673277,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016981801262404278,
      "loss": 0.0436,
      "step": 5448
    },
    {
      "epoch": 2.5432905484247375,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016980750680128186,
      "loss": 0.0376,
      "step": 5449
    },
    {
      "epoch": 2.5437572928821472,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016979699947548786,
      "loss": 0.0393,
      "step": 5450
    },
    {
      "epoch": 2.5442240373395566,
      "grad_norm": 0.984375,
      "learning_rate": 0.00016978649064688703,
      "loss": 0.0654,
      "step": 5451
    },
    {
      "epoch": 2.544690781796966,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001697759803157056,
      "loss": 0.0638,
      "step": 5452
    },
    {
      "epoch": 2.5451575262543757,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001697654684821699,
      "loss": 0.0596,
      "step": 5453
    },
    {
      "epoch": 2.5456242707117855,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001697549551465063,
      "loss": 0.0632,
      "step": 5454
    },
    {
      "epoch": 2.546091015169195,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00016974444030894107,
      "loss": 0.0561,
      "step": 5455
    },
    {
      "epoch": 2.5465577596266042,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00016973392396970073,
      "loss": 0.0481,
      "step": 5456
    },
    {
      "epoch": 2.547024504084014,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00016972340612901157,
      "loss": 0.0493,
      "step": 5457
    },
    {
      "epoch": 2.547491248541424,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001697128867871001,
      "loss": 0.0634,
      "step": 5458
    },
    {
      "epoch": 2.547957992998833,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001697023659441929,
      "loss": 0.0622,
      "step": 5459
    },
    {
      "epoch": 2.5484247374562425,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00016969184360051637,
      "loss": 0.0632,
      "step": 5460
    },
    {
      "epoch": 2.5488914819136523,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00016968131975629715,
      "loss": 0.0747,
      "step": 5461
    },
    {
      "epoch": 2.5493582263710617,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001696707944117618,
      "loss": 0.0786,
      "step": 5462
    },
    {
      "epoch": 2.5498249708284715,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001696602675671369,
      "loss": 0.0578,
      "step": 5463
    },
    {
      "epoch": 2.550291715285881,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00016964973922264923,
      "loss": 0.0517,
      "step": 5464
    },
    {
      "epoch": 2.5507584597432906,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016963920937852533,
      "loss": 0.036,
      "step": 5465
    },
    {
      "epoch": 2.5512252042007,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00016962867803499204,
      "loss": 0.0769,
      "step": 5466
    },
    {
      "epoch": 2.5516919486581098,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016961814519227602,
      "loss": 0.0492,
      "step": 5467
    },
    {
      "epoch": 2.552158693115519,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00016960761085060407,
      "loss": 0.0852,
      "step": 5468
    },
    {
      "epoch": 2.552625437572929,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00016959707501020308,
      "loss": 0.0621,
      "step": 5469
    },
    {
      "epoch": 2.5530921820303383,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016958653767129982,
      "loss": 0.0583,
      "step": 5470
    },
    {
      "epoch": 2.553558926487748,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016957599883412122,
      "loss": 0.0645,
      "step": 5471
    },
    {
      "epoch": 2.5540256709451574,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016956545849889415,
      "loss": 0.0459,
      "step": 5472
    },
    {
      "epoch": 2.554492415402567,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001695549166658456,
      "loss": 0.0635,
      "step": 5473
    },
    {
      "epoch": 2.5549591598599766,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016954437333520245,
      "loss": 0.0535,
      "step": 5474
    },
    {
      "epoch": 2.5554259043173864,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016953382850719184,
      "loss": 0.0501,
      "step": 5475
    },
    {
      "epoch": 2.5558926487747957,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00016952328218204074,
      "loss": 0.0592,
      "step": 5476
    },
    {
      "epoch": 2.5563593932322055,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00016951273435997623,
      "loss": 0.0401,
      "step": 5477
    },
    {
      "epoch": 2.556826137689615,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00016950218504122543,
      "loss": 0.0611,
      "step": 5478
    },
    {
      "epoch": 2.5572928821470247,
      "grad_norm": 1.09375,
      "learning_rate": 0.00016949163422601548,
      "loss": 0.0682,
      "step": 5479
    },
    {
      "epoch": 2.557759626604434,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001694810819145735,
      "loss": 0.0571,
      "step": 5480
    },
    {
      "epoch": 2.558226371061844,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001694705281071268,
      "loss": 0.0647,
      "step": 5481
    },
    {
      "epoch": 2.558693115519253,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00016945997280390254,
      "loss": 0.0597,
      "step": 5482
    },
    {
      "epoch": 2.5591598599766625,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016944941600512795,
      "loss": 0.0458,
      "step": 5483
    },
    {
      "epoch": 2.5596266044340723,
      "grad_norm": 0.78125,
      "learning_rate": 0.00016943885771103043,
      "loss": 0.043,
      "step": 5484
    },
    {
      "epoch": 2.560093348891482,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016942829792183724,
      "loss": 0.0406,
      "step": 5485
    },
    {
      "epoch": 2.5605600933488915,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016941773663777576,
      "loss": 0.0407,
      "step": 5486
    },
    {
      "epoch": 2.561026837806301,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001694071738590734,
      "loss": 0.0463,
      "step": 5487
    },
    {
      "epoch": 2.5614935822637106,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016939660958595755,
      "loss": 0.0508,
      "step": 5488
    },
    {
      "epoch": 2.5619603267211204,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016938604381865573,
      "loss": 0.0464,
      "step": 5489
    },
    {
      "epoch": 2.5624270711785297,
      "grad_norm": 1.15625,
      "learning_rate": 0.0001693754765573954,
      "loss": 0.0681,
      "step": 5490
    },
    {
      "epoch": 2.562893815635939,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016936490780240408,
      "loss": 0.071,
      "step": 5491
    },
    {
      "epoch": 2.563360560093349,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016935433755390934,
      "loss": 0.0464,
      "step": 5492
    },
    {
      "epoch": 2.5638273045507587,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016934376581213874,
      "loss": 0.0541,
      "step": 5493
    },
    {
      "epoch": 2.564294049008168,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016933319257731995,
      "loss": 0.0426,
      "step": 5494
    },
    {
      "epoch": 2.5647607934655774,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00016932261784968059,
      "loss": 0.0799,
      "step": 5495
    },
    {
      "epoch": 2.565227537922987,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00016931204162944834,
      "loss": 0.0464,
      "step": 5496
    },
    {
      "epoch": 2.565694282380397,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001693014639168509,
      "loss": 0.0782,
      "step": 5497
    },
    {
      "epoch": 2.5661610268378063,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00016929088471211612,
      "loss": 0.0494,
      "step": 5498
    },
    {
      "epoch": 2.5666277712952157,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00016928030401547166,
      "loss": 0.0563,
      "step": 5499
    },
    {
      "epoch": 2.5670945157526255,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016926972182714541,
      "loss": 0.0721,
      "step": 5500
    },
    {
      "epoch": 2.5675612602100353,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001692591381473652,
      "loss": 0.0731,
      "step": 5501
    },
    {
      "epoch": 2.5680280046674446,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001692485529763589,
      "loss": 0.0701,
      "step": 5502
    },
    {
      "epoch": 2.568494749124854,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001692379663143544,
      "loss": 0.0583,
      "step": 5503
    },
    {
      "epoch": 2.5689614935822638,
      "grad_norm": 0.78125,
      "learning_rate": 0.00016922737816157967,
      "loss": 0.0553,
      "step": 5504
    },
    {
      "epoch": 2.569428238039673,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001692167885182627,
      "loss": 0.0582,
      "step": 5505
    },
    {
      "epoch": 2.569894982497083,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016920619738463146,
      "loss": 0.0639,
      "step": 5506
    },
    {
      "epoch": 2.5703617269544923,
      "grad_norm": 1.015625,
      "learning_rate": 0.000169195604760914,
      "loss": 0.054,
      "step": 5507
    },
    {
      "epoch": 2.570828471411902,
      "grad_norm": 0.875,
      "learning_rate": 0.0001691850106473384,
      "loss": 0.056,
      "step": 5508
    },
    {
      "epoch": 2.5712952158693114,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001691744150441328,
      "loss": 0.0521,
      "step": 5509
    },
    {
      "epoch": 2.571761960326721,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016916381795152523,
      "loss": 0.0626,
      "step": 5510
    },
    {
      "epoch": 2.5722287047841306,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016915321936974395,
      "loss": 0.0431,
      "step": 5511
    },
    {
      "epoch": 2.5726954492415404,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016914261929901713,
      "loss": 0.0575,
      "step": 5512
    },
    {
      "epoch": 2.5731621936989497,
      "grad_norm": 0.890625,
      "learning_rate": 0.000169132017739573,
      "loss": 0.0661,
      "step": 5513
    },
    {
      "epoch": 2.5736289381563595,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00016912141469163983,
      "loss": 0.0508,
      "step": 5514
    },
    {
      "epoch": 2.574095682613769,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001691108101554459,
      "loss": 0.0503,
      "step": 5515
    },
    {
      "epoch": 2.5745624270711787,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016910020413121951,
      "loss": 0.0529,
      "step": 5516
    },
    {
      "epoch": 2.575029171528588,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016908959661918912,
      "loss": 0.057,
      "step": 5517
    },
    {
      "epoch": 2.575495915985998,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016907898761958306,
      "loss": 0.0704,
      "step": 5518
    },
    {
      "epoch": 2.575962660443407,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001690683771326297,
      "loss": 0.045,
      "step": 5519
    },
    {
      "epoch": 2.576429404900817,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001690577651585576,
      "loss": 0.0682,
      "step": 5520
    },
    {
      "epoch": 2.5768961493582263,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00016904715169759518,
      "loss": 0.0641,
      "step": 5521
    },
    {
      "epoch": 2.577362893815636,
      "grad_norm": 0.75,
      "learning_rate": 0.00016903653674997095,
      "loss": 0.055,
      "step": 5522
    },
    {
      "epoch": 2.5778296382730455,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001690259203159135,
      "loss": 0.0384,
      "step": 5523
    },
    {
      "epoch": 2.5782963827304552,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016901530239565141,
      "loss": 0.0505,
      "step": 5524
    },
    {
      "epoch": 2.5787631271878646,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016900468298941329,
      "loss": 0.0509,
      "step": 5525
    },
    {
      "epoch": 2.579229871645274,
      "grad_norm": 1.078125,
      "learning_rate": 0.00016899406209742773,
      "loss": 0.0864,
      "step": 5526
    },
    {
      "epoch": 2.5796966161026837,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016898343971992352,
      "loss": 0.0656,
      "step": 5527
    },
    {
      "epoch": 2.5801633605600935,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001689728158571293,
      "loss": 0.0556,
      "step": 5528
    },
    {
      "epoch": 2.580630105017503,
      "grad_norm": 1.125,
      "learning_rate": 0.00016896219050927384,
      "loss": 0.0975,
      "step": 5529
    },
    {
      "epoch": 2.5810968494749122,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016895156367658588,
      "loss": 0.0481,
      "step": 5530
    },
    {
      "epoch": 2.581563593932322,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016894093535929424,
      "loss": 0.0585,
      "step": 5531
    },
    {
      "epoch": 2.582030338389732,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001689303055576278,
      "loss": 0.0459,
      "step": 5532
    },
    {
      "epoch": 2.582497082847141,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001689196742718154,
      "loss": 0.0762,
      "step": 5533
    },
    {
      "epoch": 2.5829638273045505,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016890904150208594,
      "loss": 0.0603,
      "step": 5534
    },
    {
      "epoch": 2.5834305717619603,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00016889840724866834,
      "loss": 0.0687,
      "step": 5535
    },
    {
      "epoch": 2.58389731621937,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00016888777151179163,
      "loss": 0.0744,
      "step": 5536
    },
    {
      "epoch": 2.5843640606767795,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00016887713429168474,
      "loss": 0.0427,
      "step": 5537
    },
    {
      "epoch": 2.584830805134189,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016886649558857673,
      "loss": 0.0479,
      "step": 5538
    },
    {
      "epoch": 2.5852975495915986,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00016885585540269666,
      "loss": 0.0735,
      "step": 5539
    },
    {
      "epoch": 2.5857642940490084,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00016884521373427362,
      "loss": 0.0324,
      "step": 5540
    },
    {
      "epoch": 2.586231038506418,
      "grad_norm": 0.75,
      "learning_rate": 0.00016883457058353675,
      "loss": 0.0406,
      "step": 5541
    },
    {
      "epoch": 2.586697782963827,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016882392595071524,
      "loss": 0.0535,
      "step": 5542
    },
    {
      "epoch": 2.587164527421237,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001688132798360382,
      "loss": 0.0467,
      "step": 5543
    },
    {
      "epoch": 2.5876312718786463,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016880263223973492,
      "loss": 0.059,
      "step": 5544
    },
    {
      "epoch": 2.588098016336056,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001687919831620346,
      "loss": 0.0478,
      "step": 5545
    },
    {
      "epoch": 2.5885647607934654,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016878133260316656,
      "loss": 0.0525,
      "step": 5546
    },
    {
      "epoch": 2.589031505250875,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016877068056336014,
      "loss": 0.0479,
      "step": 5547
    },
    {
      "epoch": 2.5894982497082846,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016876002704284464,
      "loss": 0.0465,
      "step": 5548
    },
    {
      "epoch": 2.5899649941656944,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001687493720418495,
      "loss": 0.0413,
      "step": 5549
    },
    {
      "epoch": 2.5904317386231037,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016873871556060405,
      "loss": 0.0558,
      "step": 5550
    },
    {
      "epoch": 2.5908984830805135,
      "grad_norm": 1.0625,
      "learning_rate": 0.00016872805759933783,
      "loss": 0.0537,
      "step": 5551
    },
    {
      "epoch": 2.591365227537923,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016871739815828025,
      "loss": 0.0491,
      "step": 5552
    },
    {
      "epoch": 2.5918319719953327,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00016870673723766086,
      "loss": 0.0765,
      "step": 5553
    },
    {
      "epoch": 2.592298716452742,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00016869607483770916,
      "loss": 0.0742,
      "step": 5554
    },
    {
      "epoch": 2.592765460910152,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001686854109586548,
      "loss": 0.0516,
      "step": 5555
    },
    {
      "epoch": 2.593232205367561,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001686747456007273,
      "loss": 0.0404,
      "step": 5556
    },
    {
      "epoch": 2.593698949824971,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016866407876415634,
      "loss": 0.0438,
      "step": 5557
    },
    {
      "epoch": 2.5941656942823803,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00016865341044917156,
      "loss": 0.0518,
      "step": 5558
    },
    {
      "epoch": 2.59463243873979,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001686427406560027,
      "loss": 0.0517,
      "step": 5559
    },
    {
      "epoch": 2.5950991831971995,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016863206938487946,
      "loss": 0.0848,
      "step": 5560
    },
    {
      "epoch": 2.5955659276546093,
      "grad_norm": 1.0,
      "learning_rate": 0.00016862139663603162,
      "loss": 0.0647,
      "step": 5561
    },
    {
      "epoch": 2.5960326721120186,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00016861072240968898,
      "loss": 0.0661,
      "step": 5562
    },
    {
      "epoch": 2.5964994165694284,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00016860004670608136,
      "loss": 0.0573,
      "step": 5563
    },
    {
      "epoch": 2.5969661610268377,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016858936952543862,
      "loss": 0.0658,
      "step": 5564
    },
    {
      "epoch": 2.597432905484247,
      "grad_norm": 0.859375,
      "learning_rate": 0.00016857869086799068,
      "loss": 0.0493,
      "step": 5565
    },
    {
      "epoch": 2.597899649941657,
      "grad_norm": 0.75,
      "learning_rate": 0.00016856801073396742,
      "loss": 0.041,
      "step": 5566
    },
    {
      "epoch": 2.5983663943990667,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001685573291235988,
      "loss": 0.0359,
      "step": 5567
    },
    {
      "epoch": 2.598833138856476,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016854664603711483,
      "loss": 0.0695,
      "step": 5568
    },
    {
      "epoch": 2.5992998833138854,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00016853596147474548,
      "loss": 0.0479,
      "step": 5569
    },
    {
      "epoch": 2.599766627771295,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016852527543672087,
      "loss": 0.0447,
      "step": 5570
    },
    {
      "epoch": 2.600233372228705,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016851458792327105,
      "loss": 0.0576,
      "step": 5571
    },
    {
      "epoch": 2.6007001166861143,
      "grad_norm": 1.09375,
      "learning_rate": 0.00016850389893462611,
      "loss": 0.043,
      "step": 5572
    },
    {
      "epoch": 2.6011668611435237,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016849320847101625,
      "loss": 0.0493,
      "step": 5573
    },
    {
      "epoch": 2.6016336056009335,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001684825165326716,
      "loss": 0.0745,
      "step": 5574
    },
    {
      "epoch": 2.6021003500583433,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00016847182311982242,
      "loss": 0.0524,
      "step": 5575
    },
    {
      "epoch": 2.6025670945157526,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00016846112823269889,
      "loss": 0.0504,
      "step": 5576
    },
    {
      "epoch": 2.603033838973162,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001684504318715313,
      "loss": 0.0562,
      "step": 5577
    },
    {
      "epoch": 2.603500583430572,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016843973403654999,
      "loss": 0.0722,
      "step": 5578
    },
    {
      "epoch": 2.6039673278879816,
      "grad_norm": 0.90625,
      "learning_rate": 0.00016842903472798524,
      "loss": 0.0619,
      "step": 5579
    },
    {
      "epoch": 2.604434072345391,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016841833394606748,
      "loss": 0.0539,
      "step": 5580
    },
    {
      "epoch": 2.6049008168028003,
      "grad_norm": 1.171875,
      "learning_rate": 0.00016840763169102704,
      "loss": 0.0876,
      "step": 5581
    },
    {
      "epoch": 2.60536756126021,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001683969279630944,
      "loss": 0.0363,
      "step": 5582
    },
    {
      "epoch": 2.60583430571762,
      "grad_norm": 0.625,
      "learning_rate": 0.00016838622276250004,
      "loss": 0.0375,
      "step": 5583
    },
    {
      "epoch": 2.606301050175029,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016837551608947442,
      "loss": 0.0531,
      "step": 5584
    },
    {
      "epoch": 2.6067677946324386,
      "grad_norm": 1.078125,
      "learning_rate": 0.00016836480794424807,
      "loss": 0.0647,
      "step": 5585
    },
    {
      "epoch": 2.6072345390898484,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016835409832705156,
      "loss": 0.0512,
      "step": 5586
    },
    {
      "epoch": 2.6077012835472577,
      "grad_norm": 0.78125,
      "learning_rate": 0.00016834338723811546,
      "loss": 0.0531,
      "step": 5587
    },
    {
      "epoch": 2.6081680280046675,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001683326746776704,
      "loss": 0.0416,
      "step": 5588
    },
    {
      "epoch": 2.608634772462077,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016832196064594704,
      "loss": 0.0612,
      "step": 5589
    },
    {
      "epoch": 2.6091015169194867,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016831124514317609,
      "loss": 0.0523,
      "step": 5590
    },
    {
      "epoch": 2.609568261376896,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016830052816958822,
      "loss": 0.0341,
      "step": 5591
    },
    {
      "epoch": 2.610035005834306,
      "grad_norm": 1.0078125,
      "learning_rate": 0.0001682898097254142,
      "loss": 0.0494,
      "step": 5592
    },
    {
      "epoch": 2.610501750291715,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001682790898108848,
      "loss": 0.0479,
      "step": 5593
    },
    {
      "epoch": 2.610968494749125,
      "grad_norm": 1.09375,
      "learning_rate": 0.00016826836842623086,
      "loss": 0.0773,
      "step": 5594
    },
    {
      "epoch": 2.6114352392065343,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00016825764557168319,
      "loss": 0.0798,
      "step": 5595
    },
    {
      "epoch": 2.611901983663944,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001682469212474727,
      "loss": 0.0541,
      "step": 5596
    },
    {
      "epoch": 2.6123687281213535,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00016823619545383022,
      "loss": 0.0473,
      "step": 5597
    },
    {
      "epoch": 2.6128354725787633,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016822546819098678,
      "loss": 0.0523,
      "step": 5598
    },
    {
      "epoch": 2.6133022170361726,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001682147394591733,
      "loss": 0.0437,
      "step": 5599
    },
    {
      "epoch": 2.6137689614935824,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00016820400925862078,
      "loss": 0.0536,
      "step": 5600
    },
    {
      "epoch": 2.6142357059509918,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00016819327758956033,
      "loss": 0.0529,
      "step": 5601
    },
    {
      "epoch": 2.6147024504084015,
      "grad_norm": 0.71875,
      "learning_rate": 0.00016818254445222287,
      "loss": 0.0571,
      "step": 5602
    },
    {
      "epoch": 2.615169194865811,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00016817180984683963,
      "loss": 0.05,
      "step": 5603
    },
    {
      "epoch": 2.6156359393232207,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00016816107377364166,
      "loss": 0.0584,
      "step": 5604
    },
    {
      "epoch": 2.61610268378063,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00016815033623286015,
      "loss": 0.0662,
      "step": 5605
    },
    {
      "epoch": 2.61656942823804,
      "grad_norm": 1.0546875,
      "learning_rate": 0.0001681395972247263,
      "loss": 0.0488,
      "step": 5606
    },
    {
      "epoch": 2.617036172695449,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001681288567494713,
      "loss": 0.055,
      "step": 5607
    },
    {
      "epoch": 2.6175029171528585,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00016811811480732644,
      "loss": 0.0619,
      "step": 5608
    },
    {
      "epoch": 2.6179696616102683,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00016810737139852301,
      "loss": 0.0577,
      "step": 5609
    },
    {
      "epoch": 2.618436406067678,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016809662652329227,
      "loss": 0.0583,
      "step": 5610
    },
    {
      "epoch": 2.6189031505250875,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016808588018186563,
      "loss": 0.0553,
      "step": 5611
    },
    {
      "epoch": 2.619369894982497,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016807513237447443,
      "loss": 0.0528,
      "step": 5612
    },
    {
      "epoch": 2.6198366394399066,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00016806438310135012,
      "loss": 0.0631,
      "step": 5613
    },
    {
      "epoch": 2.6203033838973164,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001680536323627241,
      "loss": 0.0513,
      "step": 5614
    },
    {
      "epoch": 2.620770128354726,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001680428801588279,
      "loss": 0.0514,
      "step": 5615
    },
    {
      "epoch": 2.621236872812135,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016803212648989298,
      "loss": 0.0407,
      "step": 5616
    },
    {
      "epoch": 2.621703617269545,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016802137135615088,
      "loss": 0.0743,
      "step": 5617
    },
    {
      "epoch": 2.6221703617269547,
      "grad_norm": 0.75,
      "learning_rate": 0.00016801061475783317,
      "loss": 0.0522,
      "step": 5618
    },
    {
      "epoch": 2.622637106184364,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001679998566951715,
      "loss": 0.0501,
      "step": 5619
    },
    {
      "epoch": 2.6231038506417734,
      "grad_norm": 0.78125,
      "learning_rate": 0.00016798909716839747,
      "loss": 0.0578,
      "step": 5620
    },
    {
      "epoch": 2.6235705950991832,
      "grad_norm": 0.984375,
      "learning_rate": 0.00016797833617774273,
      "loss": 0.0497,
      "step": 5621
    },
    {
      "epoch": 2.624037339556593,
      "grad_norm": 1.09375,
      "learning_rate": 0.00016796757372343897,
      "loss": 0.0655,
      "step": 5622
    },
    {
      "epoch": 2.6245040840140024,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00016795680980571796,
      "loss": 0.0566,
      "step": 5623
    },
    {
      "epoch": 2.6249708284714117,
      "grad_norm": 1.0,
      "learning_rate": 0.0001679460444248114,
      "loss": 0.061,
      "step": 5624
    },
    {
      "epoch": 2.6254375729288215,
      "grad_norm": 0.71875,
      "learning_rate": 0.00016793527758095117,
      "loss": 0.039,
      "step": 5625
    },
    {
      "epoch": 2.6259043173862313,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000167924509274369,
      "loss": 0.0488,
      "step": 5626
    },
    {
      "epoch": 2.6263710618436407,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016791373950529682,
      "loss": 0.0535,
      "step": 5627
    },
    {
      "epoch": 2.62683780630105,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016790296827396642,
      "loss": 0.0408,
      "step": 5628
    },
    {
      "epoch": 2.62730455075846,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001678921955806098,
      "loss": 0.0597,
      "step": 5629
    },
    {
      "epoch": 2.627771295215869,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016788142142545892,
      "loss": 0.0509,
      "step": 5630
    },
    {
      "epoch": 2.628238039673279,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016787064580874565,
      "loss": 0.0426,
      "step": 5631
    },
    {
      "epoch": 2.6287047841306883,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016785986873070211,
      "loss": 0.0474,
      "step": 5632
    },
    {
      "epoch": 2.629171528588098,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00016784909019156033,
      "loss": 0.054,
      "step": 5633
    },
    {
      "epoch": 2.6296382730455075,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00016783831019155234,
      "loss": 0.0596,
      "step": 5634
    },
    {
      "epoch": 2.6301050175029173,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016782752873091027,
      "loss": 0.0486,
      "step": 5635
    },
    {
      "epoch": 2.6305717619603266,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001678167458098662,
      "loss": 0.0428,
      "step": 5636
    },
    {
      "epoch": 2.6310385064177364,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016780596142865239,
      "loss": 0.0666,
      "step": 5637
    },
    {
      "epoch": 2.6315052508751458,
      "grad_norm": 1.0234375,
      "learning_rate": 0.000167795175587501,
      "loss": 0.0663,
      "step": 5638
    },
    {
      "epoch": 2.6319719953325555,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016778438828664424,
      "loss": 0.0506,
      "step": 5639
    },
    {
      "epoch": 2.632438739789965,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001677735995263144,
      "loss": 0.0609,
      "step": 5640
    },
    {
      "epoch": 2.6329054842473747,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016776280930674378,
      "loss": 0.0436,
      "step": 5641
    },
    {
      "epoch": 2.633372228704784,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001677520176281647,
      "loss": 0.0505,
      "step": 5642
    },
    {
      "epoch": 2.633838973162194,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001677412244908095,
      "loss": 0.037,
      "step": 5643
    },
    {
      "epoch": 2.634305717619603,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016773042989491054,
      "loss": 0.0604,
      "step": 5644
    },
    {
      "epoch": 2.634772462077013,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00016771963384070034,
      "loss": 0.0522,
      "step": 5645
    },
    {
      "epoch": 2.6352392065344223,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001677088363284112,
      "loss": 0.0385,
      "step": 5646
    },
    {
      "epoch": 2.635705950991832,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016769803735827577,
      "loss": 0.0416,
      "step": 5647
    },
    {
      "epoch": 2.6361726954492415,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00016768723693052646,
      "loss": 0.0568,
      "step": 5648
    },
    {
      "epoch": 2.6366394399066513,
      "grad_norm": 1.03125,
      "learning_rate": 0.00016767643504539582,
      "loss": 0.0544,
      "step": 5649
    },
    {
      "epoch": 2.6371061843640606,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016766563170311647,
      "loss": 0.0461,
      "step": 5650
    },
    {
      "epoch": 2.63757292882147,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00016765482690392095,
      "loss": 0.0475,
      "step": 5651
    },
    {
      "epoch": 2.63803967327888,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000167644020648042,
      "loss": 0.0556,
      "step": 5652
    },
    {
      "epoch": 2.6385064177362896,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00016763321293571218,
      "loss": 0.0586,
      "step": 5653
    },
    {
      "epoch": 2.638973162193699,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016762240376716426,
      "loss": 0.0636,
      "step": 5654
    },
    {
      "epoch": 2.6394399066511083,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016761159314263103,
      "loss": 0.0398,
      "step": 5655
    },
    {
      "epoch": 2.639906651108518,
      "grad_norm": 1.03125,
      "learning_rate": 0.0001676007810623451,
      "loss": 0.0799,
      "step": 5656
    },
    {
      "epoch": 2.640373395565928,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016758996752653938,
      "loss": 0.0457,
      "step": 5657
    },
    {
      "epoch": 2.6408401400233372,
      "grad_norm": 1.140625,
      "learning_rate": 0.00016757915253544665,
      "loss": 0.0559,
      "step": 5658
    },
    {
      "epoch": 2.6413068844807466,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016756833608929977,
      "loss": 0.0372,
      "step": 5659
    },
    {
      "epoch": 2.6417736289381564,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016755751818833166,
      "loss": 0.0565,
      "step": 5660
    },
    {
      "epoch": 2.642240373395566,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00016754669883277528,
      "loss": 0.0545,
      "step": 5661
    },
    {
      "epoch": 2.6427071178529755,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016753587802286347,
      "loss": 0.0548,
      "step": 5662
    },
    {
      "epoch": 2.643173862310385,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001675250557588293,
      "loss": 0.0694,
      "step": 5663
    },
    {
      "epoch": 2.6436406067677947,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016751423204090576,
      "loss": 0.042,
      "step": 5664
    },
    {
      "epoch": 2.6441073512252045,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00016750340686932588,
      "loss": 0.0512,
      "step": 5665
    },
    {
      "epoch": 2.644574095682614,
      "grad_norm": 0.75,
      "learning_rate": 0.00016749258024432274,
      "loss": 0.0467,
      "step": 5666
    },
    {
      "epoch": 2.645040840140023,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016748175216612952,
      "loss": 0.0437,
      "step": 5667
    },
    {
      "epoch": 2.645507584597433,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00016747092263497926,
      "loss": 0.0539,
      "step": 5668
    },
    {
      "epoch": 2.6459743290548423,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016746009165110516,
      "loss": 0.0425,
      "step": 5669
    },
    {
      "epoch": 2.646441073512252,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00016744925921474048,
      "loss": 0.0646,
      "step": 5670
    },
    {
      "epoch": 2.6469078179696615,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001674384253261184,
      "loss": 0.0577,
      "step": 5671
    },
    {
      "epoch": 2.6473745624270713,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016742758998547217,
      "loss": 0.0598,
      "step": 5672
    },
    {
      "epoch": 2.6478413068844806,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00016741675319303516,
      "loss": 0.0481,
      "step": 5673
    },
    {
      "epoch": 2.6483080513418904,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001674059149490406,
      "loss": 0.0531,
      "step": 5674
    },
    {
      "epoch": 2.6487747957992998,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016739507525372193,
      "loss": 0.0508,
      "step": 5675
    },
    {
      "epoch": 2.6492415402567095,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016738423410731255,
      "loss": 0.0438,
      "step": 5676
    },
    {
      "epoch": 2.649708284714119,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001673733915100458,
      "loss": 0.0473,
      "step": 5677
    },
    {
      "epoch": 2.6501750291715287,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016736254746215517,
      "loss": 0.0415,
      "step": 5678
    },
    {
      "epoch": 2.650641773628938,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00016735170196387417,
      "loss": 0.0516,
      "step": 5679
    },
    {
      "epoch": 2.651108518086348,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00016734085501543633,
      "loss": 0.054,
      "step": 5680
    },
    {
      "epoch": 2.651575262543757,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016733000661707515,
      "loss": 0.0548,
      "step": 5681
    },
    {
      "epoch": 2.652042007001167,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001673191567690242,
      "loss": 0.0469,
      "step": 5682
    },
    {
      "epoch": 2.6525087514585763,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016730830547151713,
      "loss": 0.0546,
      "step": 5683
    },
    {
      "epoch": 2.652975495915986,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00016729745272478757,
      "loss": 0.0503,
      "step": 5684
    },
    {
      "epoch": 2.6534422403733955,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00016728659852906916,
      "loss": 0.0428,
      "step": 5685
    },
    {
      "epoch": 2.6539089848308053,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00016727574288459565,
      "loss": 0.055,
      "step": 5686
    },
    {
      "epoch": 2.6543757292882146,
      "grad_norm": 1.15625,
      "learning_rate": 0.00016726488579160074,
      "loss": 0.0827,
      "step": 5687
    },
    {
      "epoch": 2.6548424737456244,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001672540272503182,
      "loss": 0.045,
      "step": 5688
    },
    {
      "epoch": 2.655309218203034,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016724316726098186,
      "loss": 0.0717,
      "step": 5689
    },
    {
      "epoch": 2.655775962660443,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001672323058238255,
      "loss": 0.0647,
      "step": 5690
    },
    {
      "epoch": 2.656242707117853,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016722144293908299,
      "loss": 0.0372,
      "step": 5691
    },
    {
      "epoch": 2.6567094515752627,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016721057860698827,
      "loss": 0.0396,
      "step": 5692
    },
    {
      "epoch": 2.657176196032672,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016719971282777517,
      "loss": 0.0545,
      "step": 5693
    },
    {
      "epoch": 2.6576429404900814,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016718884560167774,
      "loss": 0.0577,
      "step": 5694
    },
    {
      "epoch": 2.6581096849474912,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016717797692892988,
      "loss": 0.0476,
      "step": 5695
    },
    {
      "epoch": 2.658576429404901,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016716710680976567,
      "loss": 0.0425,
      "step": 5696
    },
    {
      "epoch": 2.6590431738623104,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016715623524441912,
      "loss": 0.049,
      "step": 5697
    },
    {
      "epoch": 2.6595099183197197,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001671453622331243,
      "loss": 0.0497,
      "step": 5698
    },
    {
      "epoch": 2.6599766627771295,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016713448777611536,
      "loss": 0.0522,
      "step": 5699
    },
    {
      "epoch": 2.6604434072345393,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001671236118736264,
      "loss": 0.0471,
      "step": 5700
    },
    {
      "epoch": 2.6609101516919487,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016711273452589158,
      "loss": 0.0405,
      "step": 5701
    },
    {
      "epoch": 2.661376896149358,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016710185573314514,
      "loss": 0.0461,
      "step": 5702
    },
    {
      "epoch": 2.661843640606768,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016709097549562127,
      "loss": 0.0399,
      "step": 5703
    },
    {
      "epoch": 2.6623103850641776,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001670800938135543,
      "loss": 0.0649,
      "step": 5704
    },
    {
      "epoch": 2.662777129521587,
      "grad_norm": 0.84375,
      "learning_rate": 0.00016706921068717847,
      "loss": 0.0526,
      "step": 5705
    },
    {
      "epoch": 2.6632438739789963,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001670583261167281,
      "loss": 0.0473,
      "step": 5706
    },
    {
      "epoch": 2.663710618436406,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001670474401024376,
      "loss": 0.0576,
      "step": 5707
    },
    {
      "epoch": 2.664177362893816,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001670365526445413,
      "loss": 0.0505,
      "step": 5708
    },
    {
      "epoch": 2.6646441073512253,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00016702566374327368,
      "loss": 0.0539,
      "step": 5709
    },
    {
      "epoch": 2.6651108518086346,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001670147733988691,
      "loss": 0.044,
      "step": 5710
    },
    {
      "epoch": 2.6655775962660444,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016700388161156211,
      "loss": 0.0464,
      "step": 5711
    },
    {
      "epoch": 2.6660443407234538,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00016699298838158722,
      "loss": 0.0493,
      "step": 5712
    },
    {
      "epoch": 2.6665110851808635,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016698209370917893,
      "loss": 0.0371,
      "step": 5713
    },
    {
      "epoch": 2.666977829638273,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001669711975945719,
      "loss": 0.0581,
      "step": 5714
    },
    {
      "epoch": 2.6674445740956827,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00016696030003800062,
      "loss": 0.0541,
      "step": 5715
    },
    {
      "epoch": 2.667911318553092,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016694940103969982,
      "loss": 0.0564,
      "step": 5716
    },
    {
      "epoch": 2.668378063010502,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016693850059990413,
      "loss": 0.0488,
      "step": 5717
    },
    {
      "epoch": 2.668844807467911,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016692759871884824,
      "loss": 0.0421,
      "step": 5718
    },
    {
      "epoch": 2.669311551925321,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00016691669539676692,
      "loss": 0.068,
      "step": 5719
    },
    {
      "epoch": 2.6697782963827303,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00016690579063389488,
      "loss": 0.0595,
      "step": 5720
    },
    {
      "epoch": 2.67024504084014,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016689488443046694,
      "loss": 0.0441,
      "step": 5721
    },
    {
      "epoch": 2.6707117852975495,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001668839767867179,
      "loss": 0.0446,
      "step": 5722
    },
    {
      "epoch": 2.6711785297549593,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00016687306770288266,
      "loss": 0.0533,
      "step": 5723
    },
    {
      "epoch": 2.6716452742123686,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00016686215717919605,
      "loss": 0.0333,
      "step": 5724
    },
    {
      "epoch": 2.6721120186697784,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00016685124521589306,
      "loss": 0.0501,
      "step": 5725
    },
    {
      "epoch": 2.672578763127188,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016684033181320852,
      "loss": 0.0442,
      "step": 5726
    },
    {
      "epoch": 2.6730455075845976,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00016682941697137752,
      "loss": 0.0605,
      "step": 5727
    },
    {
      "epoch": 2.673512252042007,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000166818500690635,
      "loss": 0.0471,
      "step": 5728
    },
    {
      "epoch": 2.6739789964994167,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00016680758297121602,
      "loss": 0.0395,
      "step": 5729
    },
    {
      "epoch": 2.674445740956826,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001667966638133557,
      "loss": 0.0513,
      "step": 5730
    },
    {
      "epoch": 2.674912485414236,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016678574321728906,
      "loss": 0.0516,
      "step": 5731
    },
    {
      "epoch": 2.6753792298716452,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016677482118325127,
      "loss": 0.0427,
      "step": 5732
    },
    {
      "epoch": 2.6758459743290546,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001667638977114775,
      "loss": 0.0477,
      "step": 5733
    },
    {
      "epoch": 2.6763127187864644,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016675297280220293,
      "loss": 0.0477,
      "step": 5734
    },
    {
      "epoch": 2.676779463243874,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001667420464556628,
      "loss": 0.079,
      "step": 5735
    },
    {
      "epoch": 2.6772462077012835,
      "grad_norm": 1.0,
      "learning_rate": 0.00016673111867209235,
      "loss": 0.0713,
      "step": 5736
    },
    {
      "epoch": 2.677712952158693,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016672018945172688,
      "loss": 0.0524,
      "step": 5737
    },
    {
      "epoch": 2.6781796966161027,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016670925879480172,
      "loss": 0.0535,
      "step": 5738
    },
    {
      "epoch": 2.6786464410735125,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00016669832670155219,
      "loss": 0.0532,
      "step": 5739
    },
    {
      "epoch": 2.679113185530922,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016668739317221367,
      "loss": 0.0505,
      "step": 5740
    },
    {
      "epoch": 2.679579929988331,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016667645820702158,
      "loss": 0.0438,
      "step": 5741
    },
    {
      "epoch": 2.680046674445741,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001666655218062114,
      "loss": 0.0443,
      "step": 5742
    },
    {
      "epoch": 2.6805134189031508,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00016665458397001852,
      "loss": 0.0495,
      "step": 5743
    },
    {
      "epoch": 2.68098016336056,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001666436446986785,
      "loss": 0.0585,
      "step": 5744
    },
    {
      "epoch": 2.6814469078179695,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001666327039924269,
      "loss": 0.0527,
      "step": 5745
    },
    {
      "epoch": 2.6819136522753793,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00016662176185149925,
      "loss": 0.0338,
      "step": 5746
    },
    {
      "epoch": 2.682380396732789,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016661081827613113,
      "loss": 0.0394,
      "step": 5747
    },
    {
      "epoch": 2.6828471411901984,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001665998732665582,
      "loss": 0.0535,
      "step": 5748
    },
    {
      "epoch": 2.6833138856476078,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001665889268230161,
      "loss": 0.0548,
      "step": 5749
    },
    {
      "epoch": 2.6837806301050176,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016657797894574053,
      "loss": 0.0402,
      "step": 5750
    },
    {
      "epoch": 2.6842473745624273,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001665670296349672,
      "loss": 0.0547,
      "step": 5751
    },
    {
      "epoch": 2.6847141190198367,
      "grad_norm": 1.078125,
      "learning_rate": 0.00016655607889093184,
      "loss": 0.0799,
      "step": 5752
    },
    {
      "epoch": 2.685180863477246,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016654512671387028,
      "loss": 0.0456,
      "step": 5753
    },
    {
      "epoch": 2.685647607934656,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016653417310401834,
      "loss": 0.0551,
      "step": 5754
    },
    {
      "epoch": 2.686114352392065,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016652321806161178,
      "loss": 0.0471,
      "step": 5755
    },
    {
      "epoch": 2.686581096849475,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016651226158688655,
      "loss": 0.0532,
      "step": 5756
    },
    {
      "epoch": 2.6870478413068843,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016650130368007853,
      "loss": 0.0598,
      "step": 5757
    },
    {
      "epoch": 2.687514585764294,
      "grad_norm": 0.75,
      "learning_rate": 0.00016649034434142366,
      "loss": 0.0464,
      "step": 5758
    },
    {
      "epoch": 2.6879813302217035,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001664793835711579,
      "loss": 0.0778,
      "step": 5759
    },
    {
      "epoch": 2.6884480746791133,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00016646842136951723,
      "loss": 0.0546,
      "step": 5760
    },
    {
      "epoch": 2.6889148191365226,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001664574577367377,
      "loss": 0.0566,
      "step": 5761
    },
    {
      "epoch": 2.6893815635939324,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00016644649267305543,
      "loss": 0.0671,
      "step": 5762
    },
    {
      "epoch": 2.689848308051342,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001664355261787064,
      "loss": 0.0551,
      "step": 5763
    },
    {
      "epoch": 2.6903150525087516,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001664245582539268,
      "loss": 0.0494,
      "step": 5764
    },
    {
      "epoch": 2.690781796966161,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016641358889895274,
      "loss": 0.0678,
      "step": 5765
    },
    {
      "epoch": 2.6912485414235707,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001664026181140204,
      "loss": 0.0334,
      "step": 5766
    },
    {
      "epoch": 2.69171528588098,
      "grad_norm": 1.375,
      "learning_rate": 0.00016639164589936608,
      "loss": 0.0505,
      "step": 5767
    },
    {
      "epoch": 2.69218203033839,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016638067225522592,
      "loss": 0.0348,
      "step": 5768
    },
    {
      "epoch": 2.6926487747957992,
      "grad_norm": 0.75,
      "learning_rate": 0.00016636969718183625,
      "loss": 0.0335,
      "step": 5769
    },
    {
      "epoch": 2.693115519253209,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016635872067943333,
      "loss": 0.0531,
      "step": 5770
    },
    {
      "epoch": 2.6935822637106184,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016634774274825357,
      "loss": 0.0604,
      "step": 5771
    },
    {
      "epoch": 2.694049008168028,
      "grad_norm": 0.84375,
      "learning_rate": 0.00016633676338853327,
      "loss": 0.052,
      "step": 5772
    },
    {
      "epoch": 2.6945157526254375,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016632578260050888,
      "loss": 0.0385,
      "step": 5773
    },
    {
      "epoch": 2.6949824970828473,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001663148003844168,
      "loss": 0.0454,
      "step": 5774
    },
    {
      "epoch": 2.6954492415402567,
      "grad_norm": 0.875,
      "learning_rate": 0.00016630381674049346,
      "loss": 0.045,
      "step": 5775
    },
    {
      "epoch": 2.695915985997666,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001662928316689754,
      "loss": 0.0376,
      "step": 5776
    },
    {
      "epoch": 2.696382730455076,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001662818451700991,
      "loss": 0.0439,
      "step": 5777
    },
    {
      "epoch": 2.6968494749124856,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016627085724410114,
      "loss": 0.0473,
      "step": 5778
    },
    {
      "epoch": 2.697316219369895,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016625986789121813,
      "loss": 0.0504,
      "step": 5779
    },
    {
      "epoch": 2.6977829638273043,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016624887711168658,
      "loss": 0.0417,
      "step": 5780
    },
    {
      "epoch": 2.698249708284714,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001662378849057433,
      "loss": 0.0686,
      "step": 5781
    },
    {
      "epoch": 2.698716452742124,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016622689127362479,
      "loss": 0.0366,
      "step": 5782
    },
    {
      "epoch": 2.6991831971995333,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016621589621556784,
      "loss": 0.0496,
      "step": 5783
    },
    {
      "epoch": 2.6996499416569426,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016620489973180917,
      "loss": 0.0482,
      "step": 5784
    },
    {
      "epoch": 2.7001166861143524,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016619390182258557,
      "loss": 0.05,
      "step": 5785
    },
    {
      "epoch": 2.700583430571762,
      "grad_norm": 0.90625,
      "learning_rate": 0.00016618290248813383,
      "loss": 0.0519,
      "step": 5786
    },
    {
      "epoch": 2.7010501750291716,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00016617190172869077,
      "loss": 0.0533,
      "step": 5787
    },
    {
      "epoch": 2.701516919486581,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001661608995444932,
      "loss": 0.0616,
      "step": 5788
    },
    {
      "epoch": 2.7019836639439907,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001661498959357781,
      "loss": 0.063,
      "step": 5789
    },
    {
      "epoch": 2.7024504084014005,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016613889090278235,
      "loss": 0.0356,
      "step": 5790
    },
    {
      "epoch": 2.70291715285881,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016612788444574287,
      "loss": 0.0591,
      "step": 5791
    },
    {
      "epoch": 2.703383897316219,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016611687656489668,
      "loss": 0.052,
      "step": 5792
    },
    {
      "epoch": 2.703850641773629,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016610586726048077,
      "loss": 0.0435,
      "step": 5793
    },
    {
      "epoch": 2.7043173862310383,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001660948565327322,
      "loss": 0.0324,
      "step": 5794
    },
    {
      "epoch": 2.704784130688448,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016608384438188803,
      "loss": 0.0559,
      "step": 5795
    },
    {
      "epoch": 2.7052508751458575,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001660728308081854,
      "loss": 0.0503,
      "step": 5796
    },
    {
      "epoch": 2.7057176196032673,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016606181581186138,
      "loss": 0.0484,
      "step": 5797
    },
    {
      "epoch": 2.7061843640606766,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016605079939315318,
      "loss": 0.0469,
      "step": 5798
    },
    {
      "epoch": 2.7066511085180864,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00016603978155229796,
      "loss": 0.0489,
      "step": 5799
    },
    {
      "epoch": 2.707117852975496,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000166028762289533,
      "loss": 0.0556,
      "step": 5800
    },
    {
      "epoch": 2.7075845974329056,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016601774160509552,
      "loss": 0.0399,
      "step": 5801
    },
    {
      "epoch": 2.708051341890315,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016600671949922285,
      "loss": 0.0524,
      "step": 5802
    },
    {
      "epoch": 2.7085180863477247,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016599569597215222,
      "loss": 0.0522,
      "step": 5803
    },
    {
      "epoch": 2.708984830805134,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00016598467102412102,
      "loss": 0.0424,
      "step": 5804
    },
    {
      "epoch": 2.709451575262544,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016597364465536672,
      "loss": 0.049,
      "step": 5805
    },
    {
      "epoch": 2.7099183197199532,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001659626168661266,
      "loss": 0.0536,
      "step": 5806
    },
    {
      "epoch": 2.710385064177363,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00016595158765663815,
      "loss": 0.0551,
      "step": 5807
    },
    {
      "epoch": 2.7108518086347724,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00016594055702713885,
      "loss": 0.0518,
      "step": 5808
    },
    {
      "epoch": 2.711318553092182,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001659295249778662,
      "loss": 0.0638,
      "step": 5809
    },
    {
      "epoch": 2.7117852975495915,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001659184915090577,
      "loss": 0.0435,
      "step": 5810
    },
    {
      "epoch": 2.7122520420070013,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016590745662095098,
      "loss": 0.042,
      "step": 5811
    },
    {
      "epoch": 2.7127187864644107,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016589642031378357,
      "loss": 0.0487,
      "step": 5812
    },
    {
      "epoch": 2.7131855309218205,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016588538258779313,
      "loss": 0.0489,
      "step": 5813
    },
    {
      "epoch": 2.71365227537923,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001658743434432173,
      "loss": 0.0535,
      "step": 5814
    },
    {
      "epoch": 2.714119019836639,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016586330288029374,
      "loss": 0.0571,
      "step": 5815
    },
    {
      "epoch": 2.714585764294049,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016585226089926023,
      "loss": 0.0726,
      "step": 5816
    },
    {
      "epoch": 2.7150525087514588,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016584121750035444,
      "loss": 0.0604,
      "step": 5817
    },
    {
      "epoch": 2.715519253208868,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016583017268381422,
      "loss": 0.0666,
      "step": 5818
    },
    {
      "epoch": 2.7159859976662775,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001658191264498773,
      "loss": 0.059,
      "step": 5819
    },
    {
      "epoch": 2.7164527421236873,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001658080787987816,
      "loss": 0.0526,
      "step": 5820
    },
    {
      "epoch": 2.716919486581097,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016579702973076494,
      "loss": 0.0761,
      "step": 5821
    },
    {
      "epoch": 2.7173862310385064,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016578597924606526,
      "loss": 0.0583,
      "step": 5822
    },
    {
      "epoch": 2.7178529754959158,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001657749273449204,
      "loss": 0.0348,
      "step": 5823
    },
    {
      "epoch": 2.7183197199533256,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001657638740275684,
      "loss": 0.0494,
      "step": 5824
    },
    {
      "epoch": 2.7187864644107353,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016575281929424719,
      "loss": 0.0459,
      "step": 5825
    },
    {
      "epoch": 2.7192532088681447,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016574176314519486,
      "loss": 0.0478,
      "step": 5826
    },
    {
      "epoch": 2.719719953325554,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00016573070558064944,
      "loss": 0.0453,
      "step": 5827
    },
    {
      "epoch": 2.720186697782964,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000165719646600849,
      "loss": 0.0349,
      "step": 5828
    },
    {
      "epoch": 2.7206534422403736,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016570858620603164,
      "loss": 0.047,
      "step": 5829
    },
    {
      "epoch": 2.721120186697783,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001656975243964355,
      "loss": 0.0609,
      "step": 5830
    },
    {
      "epoch": 2.7215869311551923,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001656864611722988,
      "loss": 0.0445,
      "step": 5831
    },
    {
      "epoch": 2.722053675612602,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016567539653385968,
      "loss": 0.0531,
      "step": 5832
    },
    {
      "epoch": 2.722520420070012,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001656643304813564,
      "loss": 0.0448,
      "step": 5833
    },
    {
      "epoch": 2.7229871645274213,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016565326301502727,
      "loss": 0.0535,
      "step": 5834
    },
    {
      "epoch": 2.7234539089848306,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00016564219413511052,
      "loss": 0.0662,
      "step": 5835
    },
    {
      "epoch": 2.7239206534422404,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00016563112384184448,
      "loss": 0.0363,
      "step": 5836
    },
    {
      "epoch": 2.72438739789965,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016562005213546755,
      "loss": 0.0469,
      "step": 5837
    },
    {
      "epoch": 2.7248541423570596,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001656089790162181,
      "loss": 0.0446,
      "step": 5838
    },
    {
      "epoch": 2.725320886814469,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001655979044843345,
      "loss": 0.0493,
      "step": 5839
    },
    {
      "epoch": 2.7257876312718787,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00016558682854005526,
      "loss": 0.0512,
      "step": 5840
    },
    {
      "epoch": 2.726254375729288,
      "grad_norm": 0.9765625,
      "learning_rate": 0.0001655757511836188,
      "loss": 0.0723,
      "step": 5841
    },
    {
      "epoch": 2.726721120186698,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001655646724152637,
      "loss": 0.0542,
      "step": 5842
    },
    {
      "epoch": 2.7271878646441072,
      "grad_norm": 1.109375,
      "learning_rate": 0.00016555359223522845,
      "loss": 0.0687,
      "step": 5843
    },
    {
      "epoch": 2.727654609101517,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001655425106437516,
      "loss": 0.0544,
      "step": 5844
    },
    {
      "epoch": 2.7281213535589264,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016553142764107176,
      "loss": 0.0425,
      "step": 5845
    },
    {
      "epoch": 2.728588098016336,
      "grad_norm": 0.75,
      "learning_rate": 0.00016552034322742763,
      "loss": 0.0508,
      "step": 5846
    },
    {
      "epoch": 2.7290548424737455,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016550925740305782,
      "loss": 0.054,
      "step": 5847
    },
    {
      "epoch": 2.7295215869311553,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00016549817016820096,
      "loss": 0.041,
      "step": 5848
    },
    {
      "epoch": 2.7299883313885647,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016548708152309583,
      "loss": 0.0398,
      "step": 5849
    },
    {
      "epoch": 2.7304550758459745,
      "grad_norm": 0.75,
      "learning_rate": 0.0001654759914679812,
      "loss": 0.0488,
      "step": 5850
    },
    {
      "epoch": 2.730921820303384,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016546490000309585,
      "loss": 0.0554,
      "step": 5851
    },
    {
      "epoch": 2.7313885647607936,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016545380712867854,
      "loss": 0.0413,
      "step": 5852
    },
    {
      "epoch": 2.731855309218203,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016544271284496813,
      "loss": 0.0532,
      "step": 5853
    },
    {
      "epoch": 2.7323220536756128,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001654316171522035,
      "loss": 0.0444,
      "step": 5854
    },
    {
      "epoch": 2.732788798133022,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00016542052005062358,
      "loss": 0.0483,
      "step": 5855
    },
    {
      "epoch": 2.733255542590432,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001654094215404673,
      "loss": 0.0491,
      "step": 5856
    },
    {
      "epoch": 2.7337222870478413,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00016539832162197355,
      "loss": 0.0706,
      "step": 5857
    },
    {
      "epoch": 2.7341890315052506,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016538722029538138,
      "loss": 0.0531,
      "step": 5858
    },
    {
      "epoch": 2.7346557759626604,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001653761175609298,
      "loss": 0.0848,
      "step": 5859
    },
    {
      "epoch": 2.73512252042007,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016536501341885793,
      "loss": 0.0406,
      "step": 5860
    },
    {
      "epoch": 2.7355892648774796,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00016535390786940476,
      "loss": 0.0549,
      "step": 5861
    },
    {
      "epoch": 2.736056009334889,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016534280091280945,
      "loss": 0.0554,
      "step": 5862
    },
    {
      "epoch": 2.7365227537922987,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00016533169254931115,
      "loss": 0.0373,
      "step": 5863
    },
    {
      "epoch": 2.7369894982497085,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016532058277914896,
      "loss": 0.0404,
      "step": 5864
    },
    {
      "epoch": 2.737456242707118,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016530947160256222,
      "loss": 0.0335,
      "step": 5865
    },
    {
      "epoch": 2.737922987164527,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016529835901979005,
      "loss": 0.049,
      "step": 5866
    },
    {
      "epoch": 2.738389731621937,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016528724503107179,
      "loss": 0.0412,
      "step": 5867
    },
    {
      "epoch": 2.738856476079347,
      "grad_norm": 0.9609375,
      "learning_rate": 0.0001652761296366467,
      "loss": 0.0575,
      "step": 5868
    },
    {
      "epoch": 2.739323220536756,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001652650128367541,
      "loss": 0.0444,
      "step": 5869
    },
    {
      "epoch": 2.7397899649941655,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00016525389463163337,
      "loss": 0.0774,
      "step": 5870
    },
    {
      "epoch": 2.7402567094515753,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001652427750215239,
      "loss": 0.04,
      "step": 5871
    },
    {
      "epoch": 2.740723453908985,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00016523165400666504,
      "loss": 0.0324,
      "step": 5872
    },
    {
      "epoch": 2.7411901983663944,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016522053158729637,
      "loss": 0.0384,
      "step": 5873
    },
    {
      "epoch": 2.741656942823804,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016520940776365724,
      "loss": 0.0476,
      "step": 5874
    },
    {
      "epoch": 2.7421236872812136,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016519828253598723,
      "loss": 0.0562,
      "step": 5875
    },
    {
      "epoch": 2.7425904317386234,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016518715590452585,
      "loss": 0.0413,
      "step": 5876
    },
    {
      "epoch": 2.7430571761960327,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001651760278695127,
      "loss": 0.0547,
      "step": 5877
    },
    {
      "epoch": 2.743523920653442,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016516489843118734,
      "loss": 0.0596,
      "step": 5878
    },
    {
      "epoch": 2.743990665110852,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001651537675897894,
      "loss": 0.0338,
      "step": 5879
    },
    {
      "epoch": 2.7444574095682612,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00016514263534555857,
      "loss": 0.0569,
      "step": 5880
    },
    {
      "epoch": 2.744924154025671,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016513150169873453,
      "loss": 0.0572,
      "step": 5881
    },
    {
      "epoch": 2.7453908984830804,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016512036664955698,
      "loss": 0.0499,
      "step": 5882
    },
    {
      "epoch": 2.74585764294049,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001651092301982657,
      "loss": 0.0608,
      "step": 5883
    },
    {
      "epoch": 2.7463243873978995,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016509809234510042,
      "loss": 0.0564,
      "step": 5884
    },
    {
      "epoch": 2.7467911318553093,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016508695309030102,
      "loss": 0.0456,
      "step": 5885
    },
    {
      "epoch": 2.7472578763127187,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001650758124341073,
      "loss": 0.0406,
      "step": 5886
    },
    {
      "epoch": 2.7477246207701285,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016506467037675913,
      "loss": 0.0447,
      "step": 5887
    },
    {
      "epoch": 2.748191365227538,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00016505352691849643,
      "loss": 0.0414,
      "step": 5888
    },
    {
      "epoch": 2.7486581096849476,
      "grad_norm": 0.671875,
      "learning_rate": 0.00016504238205955908,
      "loss": 0.034,
      "step": 5889
    },
    {
      "epoch": 2.749124854142357,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001650312358001871,
      "loss": 0.0575,
      "step": 5890
    },
    {
      "epoch": 2.7495915985997668,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016502008814062044,
      "loss": 0.0603,
      "step": 5891
    },
    {
      "epoch": 2.750058343057176,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00016500893908109917,
      "loss": 0.0479,
      "step": 5892
    },
    {
      "epoch": 2.750525087514586,
      "grad_norm": 0.75,
      "learning_rate": 0.0001649977886218633,
      "loss": 0.0382,
      "step": 5893
    },
    {
      "epoch": 2.7509918319719953,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016498663676315294,
      "loss": 0.0496,
      "step": 5894
    },
    {
      "epoch": 2.751458576429405,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001649754835052082,
      "loss": 0.0374,
      "step": 5895
    },
    {
      "epoch": 2.7519253208868144,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016496432884826915,
      "loss": 0.0604,
      "step": 5896
    },
    {
      "epoch": 2.7519253208868144,
      "eval_loss": 1.6690407991409302,
      "eval_runtime": 94.473,
      "eval_samples_per_second": 19.095,
      "eval_steps_per_second": 2.392,
      "step": 5896
    },
    {
      "epoch": 2.752392065344224,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016495317279257608,
      "loss": 0.0607,
      "step": 5897
    },
    {
      "epoch": 2.7528588098016336,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001649420153383691,
      "loss": 0.0687,
      "step": 5898
    },
    {
      "epoch": 2.7533255542590434,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00016493085648588846,
      "loss": 0.0367,
      "step": 5899
    },
    {
      "epoch": 2.7537922987164527,
      "grad_norm": 0.9453125,
      "learning_rate": 0.00016491969623537447,
      "loss": 0.0696,
      "step": 5900
    },
    {
      "epoch": 2.754259043173862,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001649085345870674,
      "loss": 0.0457,
      "step": 5901
    },
    {
      "epoch": 2.754725787631272,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00016489737154120753,
      "loss": 0.0609,
      "step": 5902
    },
    {
      "epoch": 2.7551925320886816,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016488620709803527,
      "loss": 0.0587,
      "step": 5903
    },
    {
      "epoch": 2.755659276546091,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00016487504125779098,
      "loss": 0.0489,
      "step": 5904
    },
    {
      "epoch": 2.7561260210035003,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016486387402071503,
      "loss": 0.0575,
      "step": 5905
    },
    {
      "epoch": 2.75659276546091,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00016485270538704795,
      "loss": 0.0476,
      "step": 5906
    },
    {
      "epoch": 2.75705950991832,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00016484153535703015,
      "loss": 0.0516,
      "step": 5907
    },
    {
      "epoch": 2.7575262543757293,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00016483036393090216,
      "loss": 0.0407,
      "step": 5908
    },
    {
      "epoch": 2.7579929988331386,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001648191911089045,
      "loss": 0.0505,
      "step": 5909
    },
    {
      "epoch": 2.7584597432905484,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016480801689127775,
      "loss": 0.0542,
      "step": 5910
    },
    {
      "epoch": 2.7589264877479582,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00016479684127826248,
      "loss": 0.0603,
      "step": 5911
    },
    {
      "epoch": 2.7593932322053676,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001647856642700993,
      "loss": 0.0573,
      "step": 5912
    },
    {
      "epoch": 2.759859976662777,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001647744858670289,
      "loss": 0.043,
      "step": 5913
    },
    {
      "epoch": 2.7603267211201867,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016476330606929196,
      "loss": 0.0533,
      "step": 5914
    },
    {
      "epoch": 2.7607934655775965,
      "grad_norm": 0.75,
      "learning_rate": 0.0001647521248771292,
      "loss": 0.0417,
      "step": 5915
    },
    {
      "epoch": 2.761260210035006,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001647409422907813,
      "loss": 0.0409,
      "step": 5916
    },
    {
      "epoch": 2.7617269544924152,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016472975831048912,
      "loss": 0.0498,
      "step": 5917
    },
    {
      "epoch": 2.762193698949825,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00016471857293649342,
      "loss": 0.0475,
      "step": 5918
    },
    {
      "epoch": 2.7626604434072344,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016470738616903503,
      "loss": 0.08,
      "step": 5919
    },
    {
      "epoch": 2.763127187864644,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016469619800835484,
      "loss": 0.043,
      "step": 5920
    },
    {
      "epoch": 2.7635939323220535,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016468500845469368,
      "loss": 0.0513,
      "step": 5921
    },
    {
      "epoch": 2.7640606767794633,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016467381750829256,
      "loss": 0.0445,
      "step": 5922
    },
    {
      "epoch": 2.7645274212368727,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00016466262516939237,
      "loss": 0.037,
      "step": 5923
    },
    {
      "epoch": 2.7649941656942825,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016465143143823413,
      "loss": 0.0442,
      "step": 5924
    },
    {
      "epoch": 2.765460910151692,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016464023631505885,
      "loss": 0.0421,
      "step": 5925
    },
    {
      "epoch": 2.7659276546091016,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00016462903980010756,
      "loss": 0.0462,
      "step": 5926
    },
    {
      "epoch": 2.766394399066511,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00016461784189362132,
      "loss": 0.0454,
      "step": 5927
    },
    {
      "epoch": 2.7668611435239208,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016460664259584123,
      "loss": 0.0459,
      "step": 5928
    },
    {
      "epoch": 2.76732788798133,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00016459544190700848,
      "loss": 0.0352,
      "step": 5929
    },
    {
      "epoch": 2.76779463243874,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00016458423982736417,
      "loss": 0.046,
      "step": 5930
    },
    {
      "epoch": 2.7682613768961493,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016457303635714955,
      "loss": 0.0544,
      "step": 5931
    },
    {
      "epoch": 2.768728121353559,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001645618314966058,
      "loss": 0.0621,
      "step": 5932
    },
    {
      "epoch": 2.7691948658109684,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00016455062524597418,
      "loss": 0.0486,
      "step": 5933
    },
    {
      "epoch": 2.769661610268378,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00016453941760549595,
      "loss": 0.0445,
      "step": 5934
    },
    {
      "epoch": 2.7701283547257876,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001645282085754125,
      "loss": 0.0393,
      "step": 5935
    },
    {
      "epoch": 2.7705950991831974,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001645169981559651,
      "loss": 0.0632,
      "step": 5936
    },
    {
      "epoch": 2.7710618436406067,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016450578634739514,
      "loss": 0.0792,
      "step": 5937
    },
    {
      "epoch": 2.7715285880980165,
      "grad_norm": 0.796875,
      "learning_rate": 0.00016449457314994406,
      "loss": 0.046,
      "step": 5938
    },
    {
      "epoch": 2.771995332555426,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016448335856385321,
      "loss": 0.0462,
      "step": 5939
    },
    {
      "epoch": 2.772462077012835,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016447214258936415,
      "loss": 0.0438,
      "step": 5940
    },
    {
      "epoch": 2.772928821470245,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016446092522671833,
      "loss": 0.0479,
      "step": 5941
    },
    {
      "epoch": 2.773395565927655,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016444970647615725,
      "loss": 0.042,
      "step": 5942
    },
    {
      "epoch": 2.773862310385064,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016443848633792248,
      "loss": 0.0386,
      "step": 5943
    },
    {
      "epoch": 2.7743290548424735,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001644272648122556,
      "loss": 0.0447,
      "step": 5944
    },
    {
      "epoch": 2.7747957992998833,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016441604189939822,
      "loss": 0.054,
      "step": 5945
    },
    {
      "epoch": 2.775262543757293,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000164404817599592,
      "loss": 0.0507,
      "step": 5946
    },
    {
      "epoch": 2.7757292882147024,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016439359191307862,
      "loss": 0.0482,
      "step": 5947
    },
    {
      "epoch": 2.776196032672112,
      "grad_norm": 0.75,
      "learning_rate": 0.00016438236484009973,
      "loss": 0.0437,
      "step": 5948
    },
    {
      "epoch": 2.7766627771295216,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001643711363808971,
      "loss": 0.0553,
      "step": 5949
    },
    {
      "epoch": 2.7771295215869314,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016435990653571248,
      "loss": 0.0476,
      "step": 5950
    },
    {
      "epoch": 2.7775962660443407,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016434867530478768,
      "loss": 0.0452,
      "step": 5951
    },
    {
      "epoch": 2.77806301050175,
      "grad_norm": 0.84375,
      "learning_rate": 0.00016433744268836445,
      "loss": 0.046,
      "step": 5952
    },
    {
      "epoch": 2.77852975495916,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016432620868668475,
      "loss": 0.0622,
      "step": 5953
    },
    {
      "epoch": 2.7789964994165697,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00016431497329999042,
      "loss": 0.0381,
      "step": 5954
    },
    {
      "epoch": 2.779463243873979,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016430373652852329,
      "loss": 0.0402,
      "step": 5955
    },
    {
      "epoch": 2.7799299883313884,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001642924983725254,
      "loss": 0.0563,
      "step": 5956
    },
    {
      "epoch": 2.780396732788798,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00016428125883223872,
      "loss": 0.0556,
      "step": 5957
    },
    {
      "epoch": 2.780863477246208,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016427001790790516,
      "loss": 0.0433,
      "step": 5958
    },
    {
      "epoch": 2.7813302217036173,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016425877559976684,
      "loss": 0.0408,
      "step": 5959
    },
    {
      "epoch": 2.7817969661610267,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016424753190806578,
      "loss": 0.0279,
      "step": 5960
    },
    {
      "epoch": 2.7822637106184365,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016423628683304408,
      "loss": 0.0431,
      "step": 5961
    },
    {
      "epoch": 2.782730455075846,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00016422504037494383,
      "loss": 0.0491,
      "step": 5962
    },
    {
      "epoch": 2.7831971995332556,
      "grad_norm": 0.71875,
      "learning_rate": 0.00016421379253400722,
      "loss": 0.0346,
      "step": 5963
    },
    {
      "epoch": 2.783663943990665,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001642025433104764,
      "loss": 0.0416,
      "step": 5964
    },
    {
      "epoch": 2.7841306884480748,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001641912927045936,
      "loss": 0.0521,
      "step": 5965
    },
    {
      "epoch": 2.784597432905484,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016418004071660103,
      "loss": 0.0472,
      "step": 5966
    },
    {
      "epoch": 2.785064177362894,
      "grad_norm": 0.90625,
      "learning_rate": 0.000164168787346741,
      "loss": 0.0544,
      "step": 5967
    },
    {
      "epoch": 2.7855309218203033,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00016415753259525577,
      "loss": 0.0585,
      "step": 5968
    },
    {
      "epoch": 2.785997666277713,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016414627646238767,
      "loss": 0.0417,
      "step": 5969
    },
    {
      "epoch": 2.7864644107351224,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001641350189483791,
      "loss": 0.0487,
      "step": 5970
    },
    {
      "epoch": 2.786931155192532,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016412376005347238,
      "loss": 0.0492,
      "step": 5971
    },
    {
      "epoch": 2.7873978996499416,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016411249977790998,
      "loss": 0.0459,
      "step": 5972
    },
    {
      "epoch": 2.7878646441073514,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016410123812193427,
      "loss": 0.051,
      "step": 5973
    },
    {
      "epoch": 2.7883313885647607,
      "grad_norm": 0.859375,
      "learning_rate": 0.00016408997508578785,
      "loss": 0.0437,
      "step": 5974
    },
    {
      "epoch": 2.7887981330221705,
      "grad_norm": 1.0859375,
      "learning_rate": 0.0001640787106697131,
      "loss": 0.0582,
      "step": 5975
    },
    {
      "epoch": 2.78926487747958,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00016406744487395267,
      "loss": 0.0427,
      "step": 5976
    },
    {
      "epoch": 2.7897316219369896,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016405617769874904,
      "loss": 0.0564,
      "step": 5977
    },
    {
      "epoch": 2.790198366394399,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00016404490914434484,
      "loss": 0.0509,
      "step": 5978
    },
    {
      "epoch": 2.790665110851809,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00016403363921098268,
      "loss": 0.0337,
      "step": 5979
    },
    {
      "epoch": 2.791131855309218,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001640223678989052,
      "loss": 0.0478,
      "step": 5980
    },
    {
      "epoch": 2.791598599766628,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016401109520835516,
      "loss": 0.0484,
      "step": 5981
    },
    {
      "epoch": 2.7920653442240373,
      "grad_norm": 1.0,
      "learning_rate": 0.00016399982113957516,
      "loss": 0.0569,
      "step": 5982
    },
    {
      "epoch": 2.7925320886814466,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016398854569280803,
      "loss": 0.0718,
      "step": 5983
    },
    {
      "epoch": 2.7929988331388564,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016397726886829652,
      "loss": 0.0609,
      "step": 5984
    },
    {
      "epoch": 2.7934655775962662,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016396599066628343,
      "loss": 0.0578,
      "step": 5985
    },
    {
      "epoch": 2.7939323220536756,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016395471108701158,
      "loss": 0.0606,
      "step": 5986
    },
    {
      "epoch": 2.794399066511085,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016394343013072382,
      "loss": 0.0394,
      "step": 5987
    },
    {
      "epoch": 2.7948658109684947,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001639321477976631,
      "loss": 0.0475,
      "step": 5988
    },
    {
      "epoch": 2.7953325554259045,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016392086408807227,
      "loss": 0.051,
      "step": 5989
    },
    {
      "epoch": 2.795799299883314,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001639095790021943,
      "loss": 0.0525,
      "step": 5990
    },
    {
      "epoch": 2.7962660443407232,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00016389829254027225,
      "loss": 0.0578,
      "step": 5991
    },
    {
      "epoch": 2.796732788798133,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016388700470254902,
      "loss": 0.0523,
      "step": 5992
    },
    {
      "epoch": 2.797199533255543,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001638757154892677,
      "loss": 0.0327,
      "step": 5993
    },
    {
      "epoch": 2.797666277712952,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016386442490067137,
      "loss": 0.045,
      "step": 5994
    },
    {
      "epoch": 2.7981330221703615,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001638531329370031,
      "loss": 0.0533,
      "step": 5995
    },
    {
      "epoch": 2.7985997666277713,
      "grad_norm": 1.2421875,
      "learning_rate": 0.00016384183959850605,
      "loss": 0.0584,
      "step": 5996
    },
    {
      "epoch": 2.799066511085181,
      "grad_norm": 0.796875,
      "learning_rate": 0.00016383054488542338,
      "loss": 0.0549,
      "step": 5997
    },
    {
      "epoch": 2.7995332555425905,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00016381924879799818,
      "loss": 0.0359,
      "step": 5998
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016380795133647382,
      "loss": 0.0486,
      "step": 5999
    },
    {
      "epoch": 2.8004667444574096,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016379665250109343,
      "loss": 0.0416,
      "step": 6000
    },
    {
      "epoch": 2.8009334889148194,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00016378535229210034,
      "loss": 0.0338,
      "step": 6001
    },
    {
      "epoch": 2.8014002333722288,
      "grad_norm": 0.98046875,
      "learning_rate": 0.00016377405070973784,
      "loss": 0.0697,
      "step": 6002
    },
    {
      "epoch": 2.801866977829638,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016376274775424928,
      "loss": 0.0559,
      "step": 6003
    },
    {
      "epoch": 2.802333722287048,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016375144342587798,
      "loss": 0.0439,
      "step": 6004
    },
    {
      "epoch": 2.8028004667444573,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001637401377248674,
      "loss": 0.0528,
      "step": 6005
    },
    {
      "epoch": 2.803267211201867,
      "grad_norm": 0.875,
      "learning_rate": 0.00016372883065146092,
      "loss": 0.0523,
      "step": 6006
    },
    {
      "epoch": 2.8037339556592764,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016371752220590202,
      "loss": 0.0368,
      "step": 6007
    },
    {
      "epoch": 2.804200700116686,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016370621238843419,
      "loss": 0.0493,
      "step": 6008
    },
    {
      "epoch": 2.8046674445740956,
      "grad_norm": 0.75,
      "learning_rate": 0.00016369490119930088,
      "loss": 0.0391,
      "step": 6009
    },
    {
      "epoch": 2.8051341890315054,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001636835886387457,
      "loss": 0.047,
      "step": 6010
    },
    {
      "epoch": 2.8056009334889147,
      "grad_norm": 0.8515625,
      "learning_rate": 0.0001636722747070122,
      "loss": 0.0456,
      "step": 6011
    },
    {
      "epoch": 2.8060676779463245,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000163660959404344,
      "loss": 0.0423,
      "step": 6012
    },
    {
      "epoch": 2.806534422403734,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016364964273098468,
      "loss": 0.0373,
      "step": 6013
    },
    {
      "epoch": 2.8070011668611436,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00016363832468717794,
      "loss": 0.039,
      "step": 6014
    },
    {
      "epoch": 2.807467911318553,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00016362700527316747,
      "loss": 0.0485,
      "step": 6015
    },
    {
      "epoch": 2.807934655775963,
      "grad_norm": 0.625,
      "learning_rate": 0.00016361568448919694,
      "loss": 0.0353,
      "step": 6016
    },
    {
      "epoch": 2.808401400233372,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00016360436233551018,
      "loss": 0.0362,
      "step": 6017
    },
    {
      "epoch": 2.808868144690782,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016359303881235092,
      "loss": 0.0504,
      "step": 6018
    },
    {
      "epoch": 2.8093348891481913,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000163581713919963,
      "loss": 0.0524,
      "step": 6019
    },
    {
      "epoch": 2.809801633605601,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001635703876585902,
      "loss": 0.0462,
      "step": 6020
    },
    {
      "epoch": 2.8102683780630104,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016355906002847644,
      "loss": 0.0519,
      "step": 6021
    },
    {
      "epoch": 2.81073512252042,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001635477310298656,
      "loss": 0.0506,
      "step": 6022
    },
    {
      "epoch": 2.8112018669778296,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00016353640066300157,
      "loss": 0.0497,
      "step": 6023
    },
    {
      "epoch": 2.8116686114352394,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016352506892812837,
      "loss": 0.0428,
      "step": 6024
    },
    {
      "epoch": 2.8121353558926487,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016351373582548995,
      "loss": 0.058,
      "step": 6025
    },
    {
      "epoch": 2.812602100350058,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001635024013553303,
      "loss": 0.0295,
      "step": 6026
    },
    {
      "epoch": 2.813068844807468,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016349106551789353,
      "loss": 0.0485,
      "step": 6027
    },
    {
      "epoch": 2.8135355892648777,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001634797283134237,
      "loss": 0.0401,
      "step": 6028
    },
    {
      "epoch": 2.814002333722287,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001634683897421648,
      "loss": 0.0537,
      "step": 6029
    },
    {
      "epoch": 2.8144690781796964,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016345704980436115,
      "loss": 0.0559,
      "step": 6030
    },
    {
      "epoch": 2.814935822637106,
      "grad_norm": 0.890625,
      "learning_rate": 0.00016344570850025675,
      "loss": 0.0449,
      "step": 6031
    },
    {
      "epoch": 2.815402567094516,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00016343436583009582,
      "loss": 0.0477,
      "step": 6032
    },
    {
      "epoch": 2.8158693115519253,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016342302179412268,
      "loss": 0.0419,
      "step": 6033
    },
    {
      "epoch": 2.8163360560093347,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016341167639258148,
      "loss": 0.0339,
      "step": 6034
    },
    {
      "epoch": 2.8168028004667445,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016340032962571655,
      "loss": 0.0535,
      "step": 6035
    },
    {
      "epoch": 2.8172695449241543,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00016338898149377212,
      "loss": 0.0428,
      "step": 6036
    },
    {
      "epoch": 2.8177362893815636,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00016337763199699264,
      "loss": 0.0565,
      "step": 6037
    },
    {
      "epoch": 2.818203033838973,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016336628113562243,
      "loss": 0.0521,
      "step": 6038
    },
    {
      "epoch": 2.8186697782963828,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016335492890990586,
      "loss": 0.0466,
      "step": 6039
    },
    {
      "epoch": 2.8191365227537926,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001633435753200874,
      "loss": 0.0342,
      "step": 6040
    },
    {
      "epoch": 2.819603267211202,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00016333222036641148,
      "loss": 0.0322,
      "step": 6041
    },
    {
      "epoch": 2.8200700116686113,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016332086404912254,
      "loss": 0.0379,
      "step": 6042
    },
    {
      "epoch": 2.820536756126021,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001633095063684652,
      "loss": 0.0434,
      "step": 6043
    },
    {
      "epoch": 2.8210035005834304,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016329814732468393,
      "loss": 0.0434,
      "step": 6044
    },
    {
      "epoch": 2.82147024504084,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001632867869180233,
      "loss": 0.0313,
      "step": 6045
    },
    {
      "epoch": 2.8219369894982496,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00016327542514872793,
      "loss": 0.0633,
      "step": 6046
    },
    {
      "epoch": 2.8224037339556594,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016326406201704248,
      "loss": 0.0593,
      "step": 6047
    },
    {
      "epoch": 2.8228704784130687,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016325269752321154,
      "loss": 0.0357,
      "step": 6048
    },
    {
      "epoch": 2.8233372228704785,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016324133166747986,
      "loss": 0.0378,
      "step": 6049
    },
    {
      "epoch": 2.823803967327888,
      "grad_norm": 0.734375,
      "learning_rate": 0.00016322996445009214,
      "loss": 0.0421,
      "step": 6050
    },
    {
      "epoch": 2.8242707117852976,
      "grad_norm": 0.625,
      "learning_rate": 0.00016321859587129317,
      "loss": 0.029,
      "step": 6051
    },
    {
      "epoch": 2.824737456242707,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001632072259313276,
      "loss": 0.0517,
      "step": 6052
    },
    {
      "epoch": 2.825204200700117,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001631958546304404,
      "loss": 0.0329,
      "step": 6053
    },
    {
      "epoch": 2.825670945157526,
      "grad_norm": 0.6875,
      "learning_rate": 0.00016318448196887632,
      "loss": 0.0418,
      "step": 6054
    },
    {
      "epoch": 2.826137689614936,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016317310794688024,
      "loss": 0.0446,
      "step": 6055
    },
    {
      "epoch": 2.8266044340723453,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00016316173256469702,
      "loss": 0.0426,
      "step": 6056
    },
    {
      "epoch": 2.827071178529755,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016315035582257167,
      "loss": 0.0565,
      "step": 6057
    },
    {
      "epoch": 2.8275379229871644,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00016313897772074904,
      "loss": 0.061,
      "step": 6058
    },
    {
      "epoch": 2.8280046674445742,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001631275982594742,
      "loss": 0.0561,
      "step": 6059
    },
    {
      "epoch": 2.8284714119019836,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00016311621743899212,
      "loss": 0.0382,
      "step": 6060
    },
    {
      "epoch": 2.8289381563593934,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016310483525954785,
      "loss": 0.0282,
      "step": 6061
    },
    {
      "epoch": 2.8294049008168027,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00016309345172138646,
      "loss": 0.0656,
      "step": 6062
    },
    {
      "epoch": 2.8298716452742125,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00016308206682475306,
      "loss": 0.0407,
      "step": 6063
    },
    {
      "epoch": 2.830338389731622,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016307068056989275,
      "loss": 0.0541,
      "step": 6064
    },
    {
      "epoch": 2.8308051341890312,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016305929295705075,
      "loss": 0.0409,
      "step": 6065
    },
    {
      "epoch": 2.831271878646441,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00016304790398647217,
      "loss": 0.0506,
      "step": 6066
    },
    {
      "epoch": 2.831738623103851,
      "grad_norm": 0.75,
      "learning_rate": 0.00016303651365840227,
      "loss": 0.0417,
      "step": 6067
    },
    {
      "epoch": 2.83220536756126,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001630251219730863,
      "loss": 0.0372,
      "step": 6068
    },
    {
      "epoch": 2.8326721120186695,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016301372893076952,
      "loss": 0.0485,
      "step": 6069
    },
    {
      "epoch": 2.8331388564760793,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00016300233453169725,
      "loss": 0.0398,
      "step": 6070
    },
    {
      "epoch": 2.833605600933489,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001629909387761148,
      "loss": 0.0493,
      "step": 6071
    },
    {
      "epoch": 2.8340723453908985,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016297954166426756,
      "loss": 0.0467,
      "step": 6072
    },
    {
      "epoch": 2.834539089848308,
      "grad_norm": 0.90234375,
      "learning_rate": 0.0001629681431964009,
      "loss": 0.0519,
      "step": 6073
    },
    {
      "epoch": 2.8350058343057176,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016295674337276026,
      "loss": 0.0494,
      "step": 6074
    },
    {
      "epoch": 2.8354725787631274,
      "grad_norm": 1.09375,
      "learning_rate": 0.0001629453421935911,
      "loss": 0.0501,
      "step": 6075
    },
    {
      "epoch": 2.8359393232205368,
      "grad_norm": 0.84375,
      "learning_rate": 0.00016293393965913886,
      "loss": 0.042,
      "step": 6076
    },
    {
      "epoch": 2.836406067677946,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00016292253576964907,
      "loss": 0.04,
      "step": 6077
    },
    {
      "epoch": 2.836872812135356,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016291113052536727,
      "loss": 0.04,
      "step": 6078
    },
    {
      "epoch": 2.8373395565927657,
      "grad_norm": 0.96875,
      "learning_rate": 0.00016289972392653906,
      "loss": 0.0555,
      "step": 6079
    },
    {
      "epoch": 2.837806301050175,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00016288831597340998,
      "loss": 0.0645,
      "step": 6080
    },
    {
      "epoch": 2.8382730455075844,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001628769066662257,
      "loss": 0.0597,
      "step": 6081
    },
    {
      "epoch": 2.838739789964994,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00016286549600523183,
      "loss": 0.0455,
      "step": 6082
    },
    {
      "epoch": 2.839206534422404,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016285408399067407,
      "loss": 0.0363,
      "step": 6083
    },
    {
      "epoch": 2.8396732788798134,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001628426706227982,
      "loss": 0.0447,
      "step": 6084
    },
    {
      "epoch": 2.8401400233372227,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016283125590184984,
      "loss": 0.039,
      "step": 6085
    },
    {
      "epoch": 2.8406067677946325,
      "grad_norm": 1.125,
      "learning_rate": 0.00016281983982807487,
      "loss": 0.0575,
      "step": 6086
    },
    {
      "epoch": 2.841073512252042,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016280842240171906,
      "loss": 0.0467,
      "step": 6087
    },
    {
      "epoch": 2.8415402567094517,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00016279700362302818,
      "loss": 0.0634,
      "step": 6088
    },
    {
      "epoch": 2.842007001166861,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016278558349224818,
      "loss": 0.0458,
      "step": 6089
    },
    {
      "epoch": 2.842473745624271,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001627741620096249,
      "loss": 0.0663,
      "step": 6090
    },
    {
      "epoch": 2.84294049008168,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00016276273917540423,
      "loss": 0.045,
      "step": 6091
    },
    {
      "epoch": 2.84340723453909,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001627513149898322,
      "loss": 0.0505,
      "step": 6092
    },
    {
      "epoch": 2.8438739789964993,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001627398894531547,
      "loss": 0.0537,
      "step": 6093
    },
    {
      "epoch": 2.844340723453909,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016272846256561779,
      "loss": 0.0461,
      "step": 6094
    },
    {
      "epoch": 2.8448074679113184,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016271703432746746,
      "loss": 0.0429,
      "step": 6095
    },
    {
      "epoch": 2.8452742123687282,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001627056047389498,
      "loss": 0.0422,
      "step": 6096
    },
    {
      "epoch": 2.8457409568261376,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00016269417380031088,
      "loss": 0.0587,
      "step": 6097
    },
    {
      "epoch": 2.8462077012835474,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00016268274151179688,
      "loss": 0.0315,
      "step": 6098
    },
    {
      "epoch": 2.8466744457409567,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016267130787365387,
      "loss": 0.0375,
      "step": 6099
    },
    {
      "epoch": 2.8471411901983665,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001626598728861281,
      "loss": 0.0541,
      "step": 6100
    },
    {
      "epoch": 2.847607934655776,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00016264843654946573,
      "loss": 0.0509,
      "step": 6101
    },
    {
      "epoch": 2.8480746791131857,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00016263699886391302,
      "loss": 0.0442,
      "step": 6102
    },
    {
      "epoch": 2.848541423570595,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001626255598297162,
      "loss": 0.0413,
      "step": 6103
    },
    {
      "epoch": 2.849008168028005,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016261411944712163,
      "loss": 0.0432,
      "step": 6104
    },
    {
      "epoch": 2.849474912485414,
      "grad_norm": 0.984375,
      "learning_rate": 0.00016260267771637556,
      "loss": 0.0496,
      "step": 6105
    },
    {
      "epoch": 2.849941656942824,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016259123463772439,
      "loss": 0.0519,
      "step": 6106
    },
    {
      "epoch": 2.8504084014002333,
      "grad_norm": 0.84375,
      "learning_rate": 0.00016257979021141447,
      "loss": 0.0419,
      "step": 6107
    },
    {
      "epoch": 2.8508751458576427,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001625683444376923,
      "loss": 0.0363,
      "step": 6108
    },
    {
      "epoch": 2.8513418903150525,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001625568973168042,
      "loss": 0.0282,
      "step": 6109
    },
    {
      "epoch": 2.8518086347724623,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016254544884899673,
      "loss": 0.0449,
      "step": 6110
    },
    {
      "epoch": 2.8522753792298716,
      "grad_norm": 0.671875,
      "learning_rate": 0.00016253399903451632,
      "loss": 0.0342,
      "step": 6111
    },
    {
      "epoch": 2.852742123687281,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00016252254787360953,
      "loss": 0.0421,
      "step": 6112
    },
    {
      "epoch": 2.8532088681446908,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001625110953665229,
      "loss": 0.042,
      "step": 6113
    },
    {
      "epoch": 2.8536756126021006,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00016249964151350305,
      "loss": 0.0431,
      "step": 6114
    },
    {
      "epoch": 2.85414235705951,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001624881863147966,
      "loss": 0.0433,
      "step": 6115
    },
    {
      "epoch": 2.8546091015169193,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016247672977065013,
      "loss": 0.0379,
      "step": 6116
    },
    {
      "epoch": 2.855075845974329,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016246527188131036,
      "loss": 0.0548,
      "step": 6117
    },
    {
      "epoch": 2.855542590431739,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00016245381264702398,
      "loss": 0.0384,
      "step": 6118
    },
    {
      "epoch": 2.856009334889148,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00016244235206803773,
      "loss": 0.0688,
      "step": 6119
    },
    {
      "epoch": 2.8564760793465576,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00016243089014459833,
      "loss": 0.0643,
      "step": 6120
    },
    {
      "epoch": 2.8569428238039674,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016241942687695261,
      "loss": 0.0369,
      "step": 6121
    },
    {
      "epoch": 2.857409568261377,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001624079622653474,
      "loss": 0.0323,
      "step": 6122
    },
    {
      "epoch": 2.8578763127187865,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00016239649631002946,
      "loss": 0.0331,
      "step": 6123
    },
    {
      "epoch": 2.858343057176196,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001623850290112458,
      "loss": 0.0314,
      "step": 6124
    },
    {
      "epoch": 2.8588098016336057,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001623735603692432,
      "loss": 0.0503,
      "step": 6125
    },
    {
      "epoch": 2.859276546091015,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016236209038426863,
      "loss": 0.0458,
      "step": 6126
    },
    {
      "epoch": 2.859743290548425,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001623506190565691,
      "loss": 0.0375,
      "step": 6127
    },
    {
      "epoch": 2.860210035005834,
      "grad_norm": 0.78125,
      "learning_rate": 0.00016233914638639154,
      "loss": 0.0614,
      "step": 6128
    },
    {
      "epoch": 2.860676779463244,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00016232767237398302,
      "loss": 0.0401,
      "step": 6129
    },
    {
      "epoch": 2.8611435239206533,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001623161970195905,
      "loss": 0.0561,
      "step": 6130
    },
    {
      "epoch": 2.861610268378063,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00016230472032346117,
      "loss": 0.0326,
      "step": 6131
    },
    {
      "epoch": 2.8620770128354724,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001622932422858421,
      "loss": 0.0353,
      "step": 6132
    },
    {
      "epoch": 2.8625437572928822,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016228176290698033,
      "loss": 0.0363,
      "step": 6133
    },
    {
      "epoch": 2.8630105017502916,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016227028218712317,
      "loss": 0.0501,
      "step": 6134
    },
    {
      "epoch": 2.8634772462077014,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016225880012651773,
      "loss": 0.0486,
      "step": 6135
    },
    {
      "epoch": 2.8639439906651107,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016224731672541122,
      "loss": 0.0364,
      "step": 6136
    },
    {
      "epoch": 2.8644107351225205,
      "grad_norm": 0.671875,
      "learning_rate": 0.00016223583198405096,
      "loss": 0.0288,
      "step": 6137
    },
    {
      "epoch": 2.86487747957993,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016222434590268416,
      "loss": 0.0416,
      "step": 6138
    },
    {
      "epoch": 2.8653442240373397,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00016221285848155815,
      "loss": 0.038,
      "step": 6139
    },
    {
      "epoch": 2.865810968494749,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016220136972092028,
      "loss": 0.0407,
      "step": 6140
    },
    {
      "epoch": 2.866277712952159,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001621898796210179,
      "loss": 0.0362,
      "step": 6141
    },
    {
      "epoch": 2.866744457409568,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001621783881820984,
      "loss": 0.0569,
      "step": 6142
    },
    {
      "epoch": 2.867211201866978,
      "grad_norm": 0.8125,
      "learning_rate": 0.00016216689540440925,
      "loss": 0.0462,
      "step": 6143
    },
    {
      "epoch": 2.8676779463243873,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00016215540128819783,
      "loss": 0.0415,
      "step": 6144
    },
    {
      "epoch": 2.868144690781797,
      "grad_norm": 0.875,
      "learning_rate": 0.00016214390583371165,
      "loss": 0.0441,
      "step": 6145
    },
    {
      "epoch": 2.8686114352392065,
      "grad_norm": 0.515625,
      "learning_rate": 0.00016213240904119824,
      "loss": 0.0371,
      "step": 6146
    },
    {
      "epoch": 2.869078179696616,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016212091091090516,
      "loss": 0.0334,
      "step": 6147
    },
    {
      "epoch": 2.8695449241540256,
      "grad_norm": 1.015625,
      "learning_rate": 0.0001621094114430799,
      "loss": 0.0472,
      "step": 6148
    },
    {
      "epoch": 2.8700116686114354,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00016209791063797013,
      "loss": 0.0454,
      "step": 6149
    },
    {
      "epoch": 2.8704784130688448,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016208640849582344,
      "loss": 0.0429,
      "step": 6150
    },
    {
      "epoch": 2.870945157526254,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016207490501688747,
      "loss": 0.0396,
      "step": 6151
    },
    {
      "epoch": 2.871411901983664,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00016206340020140993,
      "loss": 0.034,
      "step": 6152
    },
    {
      "epoch": 2.8718786464410737,
      "grad_norm": 1.2578125,
      "learning_rate": 0.00016205189404963852,
      "loss": 0.0622,
      "step": 6153
    },
    {
      "epoch": 2.872345390898483,
      "grad_norm": 1.0,
      "learning_rate": 0.000162040386561821,
      "loss": 0.0352,
      "step": 6154
    },
    {
      "epoch": 2.8728121353558924,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00016202887773820513,
      "loss": 0.0431,
      "step": 6155
    },
    {
      "epoch": 2.873278879813302,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016201736757903866,
      "loss": 0.0403,
      "step": 6156
    },
    {
      "epoch": 2.873745624270712,
      "grad_norm": 0.84375,
      "learning_rate": 0.00016200585608456948,
      "loss": 0.0498,
      "step": 6157
    },
    {
      "epoch": 2.8742123687281214,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00016199434325504542,
      "loss": 0.0361,
      "step": 6158
    },
    {
      "epoch": 2.8746791131855307,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016198282909071437,
      "loss": 0.0565,
      "step": 6159
    },
    {
      "epoch": 2.8751458576429405,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016197131359182423,
      "loss": 0.0547,
      "step": 6160
    },
    {
      "epoch": 2.8756126021003503,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016195979675862296,
      "loss": 0.0428,
      "step": 6161
    },
    {
      "epoch": 2.8760793465577597,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001619482785913585,
      "loss": 0.0468,
      "step": 6162
    },
    {
      "epoch": 2.876546091015169,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001619367590902789,
      "loss": 0.0389,
      "step": 6163
    },
    {
      "epoch": 2.877012835472579,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016192523825563215,
      "loss": 0.0536,
      "step": 6164
    },
    {
      "epoch": 2.8774795799299886,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001619137160876663,
      "loss": 0.0507,
      "step": 6165
    },
    {
      "epoch": 2.877946324387398,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016190219258662943,
      "loss": 0.0324,
      "step": 6166
    },
    {
      "epoch": 2.8784130688448073,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001618906677527697,
      "loss": 0.0392,
      "step": 6167
    },
    {
      "epoch": 2.878879813302217,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00016187914158633524,
      "loss": 0.0391,
      "step": 6168
    },
    {
      "epoch": 2.8793465577596264,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001618676140875742,
      "loss": 0.0396,
      "step": 6169
    },
    {
      "epoch": 2.8798133022170362,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00016185608525673474,
      "loss": 0.0319,
      "step": 6170
    },
    {
      "epoch": 2.8802800466744456,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00016184455509406519,
      "loss": 0.0355,
      "step": 6171
    },
    {
      "epoch": 2.8807467911318554,
      "grad_norm": 0.75,
      "learning_rate": 0.0001618330235998137,
      "loss": 0.0338,
      "step": 6172
    },
    {
      "epoch": 2.8812135355892647,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016182149077422864,
      "loss": 0.0485,
      "step": 6173
    },
    {
      "epoch": 2.8816802800466745,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001618099566175583,
      "loss": 0.0652,
      "step": 6174
    },
    {
      "epoch": 2.882147024504084,
      "grad_norm": 0.8828125,
      "learning_rate": 0.000161798421130051,
      "loss": 0.0497,
      "step": 6175
    },
    {
      "epoch": 2.8826137689614937,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001617868843119551,
      "loss": 0.0559,
      "step": 6176
    },
    {
      "epoch": 2.883080513418903,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00016177534616351906,
      "loss": 0.0521,
      "step": 6177
    },
    {
      "epoch": 2.883547257876313,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00016176380668499128,
      "loss": 0.0382,
      "step": 6178
    },
    {
      "epoch": 2.884014002333722,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00016175226587662018,
      "loss": 0.0471,
      "step": 6179
    },
    {
      "epoch": 2.884480746791132,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001617407237386543,
      "loss": 0.0428,
      "step": 6180
    },
    {
      "epoch": 2.8849474912485413,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00016172918027134215,
      "loss": 0.0465,
      "step": 6181
    },
    {
      "epoch": 2.885414235705951,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00016171763547493222,
      "loss": 0.0374,
      "step": 6182
    },
    {
      "epoch": 2.8858809801633605,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00016170608934967318,
      "loss": 0.0425,
      "step": 6183
    },
    {
      "epoch": 2.8863477246207703,
      "grad_norm": 0.90625,
      "learning_rate": 0.00016169454189581353,
      "loss": 0.0369,
      "step": 6184
    },
    {
      "epoch": 2.8868144690781796,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00016168299311360195,
      "loss": 0.0342,
      "step": 6185
    },
    {
      "epoch": 2.8872812135355894,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00016167144300328708,
      "loss": 0.0401,
      "step": 6186
    },
    {
      "epoch": 2.8877479579929988,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00016165989156511764,
      "loss": 0.0451,
      "step": 6187
    },
    {
      "epoch": 2.8882147024504086,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001616483387993423,
      "loss": 0.0262,
      "step": 6188
    },
    {
      "epoch": 2.888681446907818,
      "grad_norm": 0.640625,
      "learning_rate": 0.00016163678470620986,
      "loss": 0.0326,
      "step": 6189
    },
    {
      "epoch": 2.8891481913652273,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016162522928596902,
      "loss": 0.0466,
      "step": 6190
    },
    {
      "epoch": 2.889614935822637,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016161367253886863,
      "loss": 0.0364,
      "step": 6191
    },
    {
      "epoch": 2.890081680280047,
      "grad_norm": 0.90625,
      "learning_rate": 0.00016160211446515752,
      "loss": 0.0482,
      "step": 6192
    },
    {
      "epoch": 2.890548424737456,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001615905550650845,
      "loss": 0.0333,
      "step": 6193
    },
    {
      "epoch": 2.8910151691948656,
      "grad_norm": 1.046875,
      "learning_rate": 0.00016157899433889855,
      "loss": 0.0413,
      "step": 6194
    },
    {
      "epoch": 2.8914819136522754,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00016156743228684848,
      "loss": 0.0415,
      "step": 6195
    },
    {
      "epoch": 2.891948658109685,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00016155586890918331,
      "loss": 0.0472,
      "step": 6196
    },
    {
      "epoch": 2.8924154025670945,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016154430420615198,
      "loss": 0.0375,
      "step": 6197
    },
    {
      "epoch": 2.892882147024504,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001615327381780035,
      "loss": 0.0394,
      "step": 6198
    },
    {
      "epoch": 2.8933488914819137,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001615211708249869,
      "loss": 0.0421,
      "step": 6199
    },
    {
      "epoch": 2.8938156359393234,
      "grad_norm": 0.75,
      "learning_rate": 0.00016150960214735122,
      "loss": 0.0425,
      "step": 6200
    },
    {
      "epoch": 2.894282380396733,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00016149803214534557,
      "loss": 0.0313,
      "step": 6201
    },
    {
      "epoch": 2.894749124854142,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016148646081921904,
      "loss": 0.0403,
      "step": 6202
    },
    {
      "epoch": 2.895215869311552,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016147488816922078,
      "loss": 0.0382,
      "step": 6203
    },
    {
      "epoch": 2.8956826137689617,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001614633141956,
      "loss": 0.0291,
      "step": 6204
    },
    {
      "epoch": 2.896149358226371,
      "grad_norm": 0.625,
      "learning_rate": 0.00016145173889860583,
      "loss": 0.0258,
      "step": 6205
    },
    {
      "epoch": 2.8966161026837804,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00016144016227848758,
      "loss": 0.0326,
      "step": 6206
    },
    {
      "epoch": 2.8970828471411902,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00016142858433549443,
      "loss": 0.035,
      "step": 6207
    },
    {
      "epoch": 2.8975495915986,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00016141700506987574,
      "loss": 0.0234,
      "step": 6208
    },
    {
      "epoch": 2.8980163360560094,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001614054244818808,
      "loss": 0.0505,
      "step": 6209
    },
    {
      "epoch": 2.8984830805134187,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00016139384257175888,
      "loss": 0.0347,
      "step": 6210
    },
    {
      "epoch": 2.8989498249708285,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016138225933975945,
      "loss": 0.0362,
      "step": 6211
    },
    {
      "epoch": 2.899416569428238,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00016137067478613187,
      "loss": 0.0453,
      "step": 6212
    },
    {
      "epoch": 2.8998833138856477,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00016135908891112556,
      "loss": 0.0367,
      "step": 6213
    },
    {
      "epoch": 2.900350058343057,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00016134750171499,
      "loss": 0.0464,
      "step": 6214
    },
    {
      "epoch": 2.900816802800467,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00016133591319797468,
      "loss": 0.0433,
      "step": 6215
    },
    {
      "epoch": 2.901283547257876,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016132432336032906,
      "loss": 0.048,
      "step": 6216
    },
    {
      "epoch": 2.901750291715286,
      "grad_norm": 0.796875,
      "learning_rate": 0.00016131273220230274,
      "loss": 0.0439,
      "step": 6217
    },
    {
      "epoch": 2.9022170361726953,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00016130113972414524,
      "loss": 0.0373,
      "step": 6218
    },
    {
      "epoch": 2.902683780630105,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00016128954592610622,
      "loss": 0.0444,
      "step": 6219
    },
    {
      "epoch": 2.9031505250875145,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016127795080843525,
      "loss": 0.0362,
      "step": 6220
    },
    {
      "epoch": 2.9036172695449243,
      "grad_norm": 0.65234375,
      "learning_rate": 0.000161266354371382,
      "loss": 0.0312,
      "step": 6221
    },
    {
      "epoch": 2.9040840140023336,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001612547566151962,
      "loss": 0.0326,
      "step": 6222
    },
    {
      "epoch": 2.9045507584597434,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001612431575401275,
      "loss": 0.0331,
      "step": 6223
    },
    {
      "epoch": 2.9050175029171528,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016123155714642566,
      "loss": 0.0567,
      "step": 6224
    },
    {
      "epoch": 2.9054842473745626,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001612199554343405,
      "loss": 0.0475,
      "step": 6225
    },
    {
      "epoch": 2.905950991831972,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00016120835240412173,
      "loss": 0.0377,
      "step": 6226
    },
    {
      "epoch": 2.9064177362893817,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00016119674805601926,
      "loss": 0.0335,
      "step": 6227
    },
    {
      "epoch": 2.906884480746791,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00016118514239028286,
      "loss": 0.0427,
      "step": 6228
    },
    {
      "epoch": 2.907351225204201,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001611735354071625,
      "loss": 0.0457,
      "step": 6229
    },
    {
      "epoch": 2.90781796966161,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016116192710690803,
      "loss": 0.0472,
      "step": 6230
    },
    {
      "epoch": 2.90828471411902,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001611503174897694,
      "loss": 0.0339,
      "step": 6231
    },
    {
      "epoch": 2.9087514585764294,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001611387065559966,
      "loss": 0.0485,
      "step": 6232
    },
    {
      "epoch": 2.9092182030338387,
      "grad_norm": 0.921875,
      "learning_rate": 0.00016112709430583962,
      "loss": 0.0457,
      "step": 6233
    },
    {
      "epoch": 2.9096849474912485,
      "grad_norm": 0.65625,
      "learning_rate": 0.00016111548073954846,
      "loss": 0.0335,
      "step": 6234
    },
    {
      "epoch": 2.9101516919486583,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016110386585737322,
      "loss": 0.046,
      "step": 6235
    },
    {
      "epoch": 2.9106184364060677,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00016109224965956393,
      "loss": 0.0432,
      "step": 6236
    },
    {
      "epoch": 2.911085180863477,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001610806321463707,
      "loss": 0.031,
      "step": 6237
    },
    {
      "epoch": 2.911551925320887,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00016106901331804373,
      "loss": 0.0492,
      "step": 6238
    },
    {
      "epoch": 2.9120186697782966,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00016105739317483312,
      "loss": 0.0518,
      "step": 6239
    },
    {
      "epoch": 2.912485414235706,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001610457717169891,
      "loss": 0.0398,
      "step": 6240
    },
    {
      "epoch": 2.9129521586931153,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001610341489447619,
      "loss": 0.0396,
      "step": 6241
    },
    {
      "epoch": 2.913418903150525,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016102252485840173,
      "loss": 0.0465,
      "step": 6242
    },
    {
      "epoch": 2.913885647607935,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001610108994581589,
      "loss": 0.0551,
      "step": 6243
    },
    {
      "epoch": 2.9143523920653442,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016099927274428372,
      "loss": 0.059,
      "step": 6244
    },
    {
      "epoch": 2.9148191365227536,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00016098764471702653,
      "loss": 0.0506,
      "step": 6245
    },
    {
      "epoch": 2.9152858809801634,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00016097601537663764,
      "loss": 0.0356,
      "step": 6246
    },
    {
      "epoch": 2.915752625437573,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001609643847233675,
      "loss": 0.0336,
      "step": 6247
    },
    {
      "epoch": 2.9162193698949825,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016095275275746654,
      "loss": 0.0559,
      "step": 6248
    },
    {
      "epoch": 2.916686114352392,
      "grad_norm": 0.875,
      "learning_rate": 0.00016094111947918517,
      "loss": 0.0441,
      "step": 6249
    },
    {
      "epoch": 2.9171528588098017,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016092948488877388,
      "loss": 0.0342,
      "step": 6250
    },
    {
      "epoch": 2.917619603267211,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001609178489864832,
      "loss": 0.0345,
      "step": 6251
    },
    {
      "epoch": 2.918086347724621,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016090621177256362,
      "loss": 0.0549,
      "step": 6252
    },
    {
      "epoch": 2.91855309218203,
      "grad_norm": 0.9140625,
      "learning_rate": 0.0001608945732472657,
      "loss": 0.0459,
      "step": 6253
    },
    {
      "epoch": 2.91901983663944,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001608829334108401,
      "loss": 0.0393,
      "step": 6254
    },
    {
      "epoch": 2.9194865810968493,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016087129226353736,
      "loss": 0.0513,
      "step": 6255
    },
    {
      "epoch": 2.919953325554259,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00016085964980560818,
      "loss": 0.0358,
      "step": 6256
    },
    {
      "epoch": 2.9204200700116685,
      "grad_norm": 1.1796875,
      "learning_rate": 0.0001608480060373032,
      "loss": 0.0471,
      "step": 6257
    },
    {
      "epoch": 2.9208868144690783,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00016083636095887315,
      "loss": 0.0364,
      "step": 6258
    },
    {
      "epoch": 2.9213535589264876,
      "grad_norm": 0.953125,
      "learning_rate": 0.00016082471457056876,
      "loss": 0.0448,
      "step": 6259
    },
    {
      "epoch": 2.9218203033838974,
      "grad_norm": 0.859375,
      "learning_rate": 0.00016081306687264077,
      "loss": 0.0618,
      "step": 6260
    },
    {
      "epoch": 2.9222870478413068,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00016080141786533995,
      "loss": 0.0443,
      "step": 6261
    },
    {
      "epoch": 2.9227537922987166,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00016078976754891716,
      "loss": 0.0355,
      "step": 6262
    },
    {
      "epoch": 2.923220536756126,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00016077811592362325,
      "loss": 0.0477,
      "step": 6263
    },
    {
      "epoch": 2.9236872812135357,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00016076646298970903,
      "loss": 0.0445,
      "step": 6264
    },
    {
      "epoch": 2.924154025670945,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001607548087474255,
      "loss": 0.0447,
      "step": 6265
    },
    {
      "epoch": 2.924620770128355,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00016074315319702346,
      "loss": 0.0392,
      "step": 6266
    },
    {
      "epoch": 2.925087514585764,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00016073149633875398,
      "loss": 0.0396,
      "step": 6267
    },
    {
      "epoch": 2.925554259043174,
      "grad_norm": 0.640625,
      "learning_rate": 0.00016071983817286798,
      "loss": 0.0324,
      "step": 6268
    },
    {
      "epoch": 2.9260210035005834,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001607081786996165,
      "loss": 0.048,
      "step": 6269
    },
    {
      "epoch": 2.926487747957993,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00016069651791925059,
      "loss": 0.0465,
      "step": 6270
    },
    {
      "epoch": 2.9269544924154025,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001606848558320213,
      "loss": 0.0492,
      "step": 6271
    },
    {
      "epoch": 2.927421236872812,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00016067319243817972,
      "loss": 0.0228,
      "step": 6272
    },
    {
      "epoch": 2.9278879813302217,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000160661527737977,
      "loss": 0.0409,
      "step": 6273
    },
    {
      "epoch": 2.9283547257876315,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016064986173166426,
      "loss": 0.0319,
      "step": 6274
    },
    {
      "epoch": 2.928821470245041,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00016063819441949271,
      "loss": 0.0407,
      "step": 6275
    },
    {
      "epoch": 2.92928821470245,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001606265258017136,
      "loss": 0.0447,
      "step": 6276
    },
    {
      "epoch": 2.92975495915986,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00016061485587857806,
      "loss": 0.0401,
      "step": 6277
    },
    {
      "epoch": 2.9302217036172697,
      "grad_norm": 0.9375,
      "learning_rate": 0.00016060318465033747,
      "loss": 0.0515,
      "step": 6278
    },
    {
      "epoch": 2.930688448074679,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00016059151211724303,
      "loss": 0.0621,
      "step": 6279
    },
    {
      "epoch": 2.9311551925320884,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00016057983827954612,
      "loss": 0.0515,
      "step": 6280
    },
    {
      "epoch": 2.9316219369894982,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001605681631374981,
      "loss": 0.0612,
      "step": 6281
    },
    {
      "epoch": 2.932088681446908,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001605564866913503,
      "loss": 0.0538,
      "step": 6282
    },
    {
      "epoch": 2.9325554259043174,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001605448089413542,
      "loss": 0.0487,
      "step": 6283
    },
    {
      "epoch": 2.9330221703617267,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00016053312988776117,
      "loss": 0.0381,
      "step": 6284
    },
    {
      "epoch": 2.9334889148191365,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00016052144953082267,
      "loss": 0.0286,
      "step": 6285
    },
    {
      "epoch": 2.9339556592765463,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001605097678707902,
      "loss": 0.039,
      "step": 6286
    },
    {
      "epoch": 2.9344224037339557,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001604980849079153,
      "loss": 0.0393,
      "step": 6287
    },
    {
      "epoch": 2.934889148191365,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00016048640064244952,
      "loss": 0.0372,
      "step": 6288
    },
    {
      "epoch": 2.935355892648775,
      "grad_norm": 0.640625,
      "learning_rate": 0.00016047471507464443,
      "loss": 0.0378,
      "step": 6289
    },
    {
      "epoch": 2.9358226371061846,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016046302820475164,
      "loss": 0.0501,
      "step": 6290
    },
    {
      "epoch": 2.936289381563594,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00016045134003302272,
      "loss": 0.0395,
      "step": 6291
    },
    {
      "epoch": 2.9367561260210033,
      "grad_norm": 0.75,
      "learning_rate": 0.0001604396505597094,
      "loss": 0.0455,
      "step": 6292
    },
    {
      "epoch": 2.937222870478413,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00016042795978506334,
      "loss": 0.0247,
      "step": 6293
    },
    {
      "epoch": 2.9376896149358225,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00016041626770933628,
      "loss": 0.0381,
      "step": 6294
    },
    {
      "epoch": 2.9381563593932323,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001604045743327799,
      "loss": 0.0556,
      "step": 6295
    },
    {
      "epoch": 2.9386231038506416,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00016039287965564607,
      "loss": 0.0535,
      "step": 6296
    },
    {
      "epoch": 2.9390898483080514,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001603811836781865,
      "loss": 0.0481,
      "step": 6297
    },
    {
      "epoch": 2.9395565927654608,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00016036948640065307,
      "loss": 0.0367,
      "step": 6298
    },
    {
      "epoch": 2.9400233372228706,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001603577878232976,
      "loss": 0.0484,
      "step": 6299
    },
    {
      "epoch": 2.94049008168028,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00016034608794637199,
      "loss": 0.0342,
      "step": 6300
    },
    {
      "epoch": 2.9409568261376897,
      "grad_norm": 0.703125,
      "learning_rate": 0.00016033438677012817,
      "loss": 0.0362,
      "step": 6301
    },
    {
      "epoch": 2.941423570595099,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00016032268429481804,
      "loss": 0.0472,
      "step": 6302
    },
    {
      "epoch": 2.941890315052509,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001603109805206936,
      "loss": 0.035,
      "step": 6303
    },
    {
      "epoch": 2.942357059509918,
      "grad_norm": 1.015625,
      "learning_rate": 0.00016029927544800683,
      "loss": 0.0471,
      "step": 6304
    },
    {
      "epoch": 2.942823803967328,
      "grad_norm": 0.75,
      "learning_rate": 0.00016028756907700975,
      "loss": 0.0452,
      "step": 6305
    },
    {
      "epoch": 2.9432905484247374,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016027586140795444,
      "loss": 0.0305,
      "step": 6306
    },
    {
      "epoch": 2.943757292882147,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00016026415244109292,
      "loss": 0.0422,
      "step": 6307
    },
    {
      "epoch": 2.9442240373395565,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00016025244217667738,
      "loss": 0.0515,
      "step": 6308
    },
    {
      "epoch": 2.9446907817969663,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00016024073061495986,
      "loss": 0.045,
      "step": 6309
    },
    {
      "epoch": 2.9451575262543757,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001602290177561926,
      "loss": 0.0386,
      "step": 6310
    },
    {
      "epoch": 2.9456242707117855,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00016021730360062778,
      "loss": 0.0391,
      "step": 6311
    },
    {
      "epoch": 2.946091015169195,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00016020558814851756,
      "loss": 0.0462,
      "step": 6312
    },
    {
      "epoch": 2.9465577596266046,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00016019387140011425,
      "loss": 0.0324,
      "step": 6313
    },
    {
      "epoch": 2.947024504084014,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00016018215335567007,
      "loss": 0.0353,
      "step": 6314
    },
    {
      "epoch": 2.9474912485414233,
      "grad_norm": 0.828125,
      "learning_rate": 0.00016017043401543742,
      "loss": 0.0374,
      "step": 6315
    },
    {
      "epoch": 2.947957992998833,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00016015871337966852,
      "loss": 0.032,
      "step": 6316
    },
    {
      "epoch": 2.948424737456243,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001601469914486158,
      "loss": 0.0499,
      "step": 6317
    },
    {
      "epoch": 2.9488914819136522,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016013526822253159,
      "loss": 0.041,
      "step": 6318
    },
    {
      "epoch": 2.9493582263710616,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00016012354370166832,
      "loss": 0.0414,
      "step": 6319
    },
    {
      "epoch": 2.9498249708284714,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00016011181788627848,
      "loss": 0.0444,
      "step": 6320
    },
    {
      "epoch": 2.950291715285881,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00016010009077661449,
      "loss": 0.0432,
      "step": 6321
    },
    {
      "epoch": 2.9507584597432905,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00016008836237292888,
      "loss": 0.0254,
      "step": 6322
    },
    {
      "epoch": 2.9512252042007,
      "grad_norm": 0.78125,
      "learning_rate": 0.00016007663267547415,
      "loss": 0.0288,
      "step": 6323
    },
    {
      "epoch": 2.9516919486581097,
      "grad_norm": 0.765625,
      "learning_rate": 0.00016006490168450285,
      "loss": 0.0448,
      "step": 6324
    },
    {
      "epoch": 2.9521586931155195,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001600531694002676,
      "loss": 0.05,
      "step": 6325
    },
    {
      "epoch": 2.952625437572929,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00016004143582302096,
      "loss": 0.0324,
      "step": 6326
    },
    {
      "epoch": 2.953092182030338,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001600297009530156,
      "loss": 0.0331,
      "step": 6327
    },
    {
      "epoch": 2.953558926487748,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00016001796479050417,
      "loss": 0.0556,
      "step": 6328
    },
    {
      "epoch": 2.9540256709451578,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00016000622733573934,
      "loss": 0.0485,
      "step": 6329
    },
    {
      "epoch": 2.954492415402567,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001599944885889739,
      "loss": 0.0312,
      "step": 6330
    },
    {
      "epoch": 2.9549591598599765,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015998274855046058,
      "loss": 0.0481,
      "step": 6331
    },
    {
      "epoch": 2.9554259043173863,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001599710072204521,
      "loss": 0.0259,
      "step": 6332
    },
    {
      "epoch": 2.955892648774796,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015995926459920126,
      "loss": 0.0483,
      "step": 6333
    },
    {
      "epoch": 2.9563593932322054,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015994752068696096,
      "loss": 0.0487,
      "step": 6334
    },
    {
      "epoch": 2.9568261376896148,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00015993577548398404,
      "loss": 0.049,
      "step": 6335
    },
    {
      "epoch": 2.9572928821470246,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015992402899052334,
      "loss": 0.0377,
      "step": 6336
    },
    {
      "epoch": 2.957759626604434,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015991228120683184,
      "loss": 0.0303,
      "step": 6337
    },
    {
      "epoch": 2.9582263710618437,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00015990053213316244,
      "loss": 0.0465,
      "step": 6338
    },
    {
      "epoch": 2.958693115519253,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015988878176976813,
      "loss": 0.0297,
      "step": 6339
    },
    {
      "epoch": 2.959159859976663,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015987703011690187,
      "loss": 0.0374,
      "step": 6340
    },
    {
      "epoch": 2.959626604434072,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015986527717481674,
      "loss": 0.0391,
      "step": 6341
    },
    {
      "epoch": 2.960093348891482,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015985352294376577,
      "loss": 0.0393,
      "step": 6342
    },
    {
      "epoch": 2.9605600933488914,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015984176742400203,
      "loss": 0.0409,
      "step": 6343
    },
    {
      "epoch": 2.961026837806301,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015983001061577868,
      "loss": 0.0458,
      "step": 6344
    },
    {
      "epoch": 2.9614935822637105,
      "grad_norm": 1.109375,
      "learning_rate": 0.00015981825251934877,
      "loss": 0.0505,
      "step": 6345
    },
    {
      "epoch": 2.9619603267211203,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015980649313496553,
      "loss": 0.0355,
      "step": 6346
    },
    {
      "epoch": 2.9624270711785297,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015979473246288214,
      "loss": 0.0356,
      "step": 6347
    },
    {
      "epoch": 2.9628938156359395,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00015978297050335183,
      "loss": 0.0476,
      "step": 6348
    },
    {
      "epoch": 2.963360560093349,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001597712072566278,
      "loss": 0.0451,
      "step": 6349
    },
    {
      "epoch": 2.9638273045507586,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001597594427229634,
      "loss": 0.0349,
      "step": 6350
    },
    {
      "epoch": 2.964294049008168,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015974767690261188,
      "loss": 0.0389,
      "step": 6351
    },
    {
      "epoch": 2.9647607934655777,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015973590979582658,
      "loss": 0.036,
      "step": 6352
    },
    {
      "epoch": 2.965227537922987,
      "grad_norm": 0.90625,
      "learning_rate": 0.00015972414140286086,
      "loss": 0.0485,
      "step": 6353
    },
    {
      "epoch": 2.965694282380397,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00015971237172396813,
      "loss": 0.0428,
      "step": 6354
    },
    {
      "epoch": 2.9661610268378062,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015970060075940179,
      "loss": 0.0364,
      "step": 6355
    },
    {
      "epoch": 2.966627771295216,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00015968882850941524,
      "loss": 0.0164,
      "step": 6356
    },
    {
      "epoch": 2.9670945157526254,
      "grad_norm": 0.8359375,
      "learning_rate": 0.000159677054974262,
      "loss": 0.0385,
      "step": 6357
    },
    {
      "epoch": 2.9675612602100347,
      "grad_norm": 0.875,
      "learning_rate": 0.00015966528015419557,
      "loss": 0.047,
      "step": 6358
    },
    {
      "epoch": 2.9680280046674445,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015965350404946945,
      "loss": 0.0409,
      "step": 6359
    },
    {
      "epoch": 2.9684947491248543,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001596417266603372,
      "loss": 0.0471,
      "step": 6360
    },
    {
      "epoch": 2.9689614935822637,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001596299479870524,
      "loss": 0.0387,
      "step": 6361
    },
    {
      "epoch": 2.969428238039673,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00015961816802986865,
      "loss": 0.0549,
      "step": 6362
    },
    {
      "epoch": 2.969894982497083,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015960638678903962,
      "loss": 0.0474,
      "step": 6363
    },
    {
      "epoch": 2.9703617269544926,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015959460426481892,
      "loss": 0.0386,
      "step": 6364
    },
    {
      "epoch": 2.970828471411902,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015958282045746026,
      "loss": 0.0283,
      "step": 6365
    },
    {
      "epoch": 2.9712952158693113,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001595710353672174,
      "loss": 0.0505,
      "step": 6366
    },
    {
      "epoch": 2.971761960326721,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015955924899434406,
      "loss": 0.0603,
      "step": 6367
    },
    {
      "epoch": 2.972228704784131,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00015954746133909397,
      "loss": 0.0466,
      "step": 6368
    },
    {
      "epoch": 2.9726954492415403,
      "grad_norm": 0.79296875,
      "learning_rate": 0.000159535672401721,
      "loss": 0.0427,
      "step": 6369
    },
    {
      "epoch": 2.9731621936989496,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015952388218247893,
      "loss": 0.0315,
      "step": 6370
    },
    {
      "epoch": 2.9736289381563594,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015951209068162166,
      "loss": 0.0326,
      "step": 6371
    },
    {
      "epoch": 2.974095682613769,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015950029789940298,
      "loss": 0.0318,
      "step": 6372
    },
    {
      "epoch": 2.9745624270711786,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015948850383607695,
      "loss": 0.0409,
      "step": 6373
    },
    {
      "epoch": 2.975029171528588,
      "grad_norm": 0.75,
      "learning_rate": 0.00015947670849189737,
      "loss": 0.0404,
      "step": 6374
    },
    {
      "epoch": 2.9754959159859977,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001594649118671183,
      "loss": 0.0433,
      "step": 6375
    },
    {
      "epoch": 2.975962660443407,
      "grad_norm": 0.80078125,
      "learning_rate": 0.0001594531139619937,
      "loss": 0.0553,
      "step": 6376
    },
    {
      "epoch": 2.976429404900817,
      "grad_norm": 0.890625,
      "learning_rate": 0.00015944131477677758,
      "loss": 0.0416,
      "step": 6377
    },
    {
      "epoch": 2.976896149358226,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00015942951431172398,
      "loss": 0.0253,
      "step": 6378
    },
    {
      "epoch": 2.977362893815636,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015941771256708704,
      "loss": 0.0282,
      "step": 6379
    },
    {
      "epoch": 2.9778296382730454,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001594059095431208,
      "loss": 0.0562,
      "step": 6380
    },
    {
      "epoch": 2.978296382730455,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00015939410524007944,
      "loss": 0.0335,
      "step": 6381
    },
    {
      "epoch": 2.9787631271878645,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015938229965821704,
      "loss": 0.0421,
      "step": 6382
    },
    {
      "epoch": 2.9792298716452743,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001593704927977879,
      "loss": 0.0546,
      "step": 6383
    },
    {
      "epoch": 2.9796966161026837,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015935868465904616,
      "loss": 0.0417,
      "step": 6384
    },
    {
      "epoch": 2.9801633605600935,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00015934687524224607,
      "loss": 0.0323,
      "step": 6385
    },
    {
      "epoch": 2.980630105017503,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015933506454764192,
      "loss": 0.0424,
      "step": 6386
    },
    {
      "epoch": 2.9810968494749126,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015932325257548798,
      "loss": 0.048,
      "step": 6387
    },
    {
      "epoch": 2.981563593932322,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001593114393260386,
      "loss": 0.035,
      "step": 6388
    },
    {
      "epoch": 2.9820303383897317,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00015929962479954815,
      "loss": 0.0538,
      "step": 6389
    },
    {
      "epoch": 2.982497082847141,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015928780899627097,
      "loss": 0.0348,
      "step": 6390
    },
    {
      "epoch": 2.982963827304551,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00015927599191646144,
      "loss": 0.042,
      "step": 6391
    },
    {
      "epoch": 2.9834305717619602,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015926417356037408,
      "loss": 0.0442,
      "step": 6392
    },
    {
      "epoch": 2.98389731621937,
      "grad_norm": 0.75,
      "learning_rate": 0.0001592523539282633,
      "loss": 0.0461,
      "step": 6393
    },
    {
      "epoch": 2.9843640606767794,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015924053302038358,
      "loss": 0.0442,
      "step": 6394
    },
    {
      "epoch": 2.984830805134189,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015922871083698948,
      "loss": 0.0298,
      "step": 6395
    },
    {
      "epoch": 2.9852975495915985,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001592168873783355,
      "loss": 0.0611,
      "step": 6396
    },
    {
      "epoch": 2.985764294049008,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015920506264467625,
      "loss": 0.0385,
      "step": 6397
    },
    {
      "epoch": 2.9862310385064177,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015919323663626632,
      "loss": 0.0307,
      "step": 6398
    },
    {
      "epoch": 2.9866977829638275,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001591814093533603,
      "loss": 0.0416,
      "step": 6399
    },
    {
      "epoch": 2.987164527421237,
      "grad_norm": 0.953125,
      "learning_rate": 0.0001591695807962129,
      "loss": 0.0446,
      "step": 6400
    },
    {
      "epoch": 2.987631271878646,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015915775096507874,
      "loss": 0.0431,
      "step": 6401
    },
    {
      "epoch": 2.988098016336056,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015914591986021263,
      "loss": 0.0493,
      "step": 6402
    },
    {
      "epoch": 2.988564760793466,
      "grad_norm": 0.75,
      "learning_rate": 0.0001591340874818692,
      "loss": 0.0398,
      "step": 6403
    },
    {
      "epoch": 2.989031505250875,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00015912225383030325,
      "loss": 0.0386,
      "step": 6404
    },
    {
      "epoch": 2.9894982497082845,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001591104189057696,
      "loss": 0.0306,
      "step": 6405
    },
    {
      "epoch": 2.9899649941656943,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015909858270852302,
      "loss": 0.044,
      "step": 6406
    },
    {
      "epoch": 2.990431738623104,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001590867452388184,
      "loss": 0.0242,
      "step": 6407
    },
    {
      "epoch": 2.9908984830805134,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001590749064969106,
      "loss": 0.0474,
      "step": 6408
    },
    {
      "epoch": 2.9913652275379228,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001590630664830545,
      "loss": 0.0436,
      "step": 6409
    },
    {
      "epoch": 2.9918319719953326,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015905122519750507,
      "loss": 0.0323,
      "step": 6410
    },
    {
      "epoch": 2.9922987164527424,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001590393826405172,
      "loss": 0.0384,
      "step": 6411
    },
    {
      "epoch": 2.9927654609101517,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015902753881234598,
      "loss": 0.0256,
      "step": 6412
    },
    {
      "epoch": 2.993232205367561,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001590156937132463,
      "loss": 0.0312,
      "step": 6413
    },
    {
      "epoch": 2.993698949824971,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00015900384734347328,
      "loss": 0.0376,
      "step": 6414
    },
    {
      "epoch": 2.9941656942823807,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00015899199970328194,
      "loss": 0.0414,
      "step": 6415
    },
    {
      "epoch": 2.99463243873979,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015898015079292746,
      "loss": 0.0431,
      "step": 6416
    },
    {
      "epoch": 2.9950991831971994,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015896830061266483,
      "loss": 0.0403,
      "step": 6417
    },
    {
      "epoch": 2.995565927654609,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015895644916274929,
      "loss": 0.0296,
      "step": 6418
    },
    {
      "epoch": 2.9960326721120185,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00015894459644343596,
      "loss": 0.0495,
      "step": 6419
    },
    {
      "epoch": 2.9964994165694283,
      "grad_norm": 1.0390625,
      "learning_rate": 0.0001589327424549801,
      "loss": 0.0718,
      "step": 6420
    },
    {
      "epoch": 2.9969661610268377,
      "grad_norm": 0.75,
      "learning_rate": 0.0001589208871976369,
      "loss": 0.0417,
      "step": 6421
    },
    {
      "epoch": 2.9974329054842475,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015890903067166157,
      "loss": 0.048,
      "step": 6422
    },
    {
      "epoch": 2.997899649941657,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015889717287730952,
      "loss": 0.0491,
      "step": 6423
    },
    {
      "epoch": 2.9983663943990666,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015888531381483598,
      "loss": 0.0301,
      "step": 6424
    },
    {
      "epoch": 2.998833138856476,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015887345348449626,
      "loss": 0.0449,
      "step": 6425
    },
    {
      "epoch": 2.9992998833138858,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015886159188654585,
      "loss": 0.0429,
      "step": 6426
    },
    {
      "epoch": 2.999766627771295,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015884972902124,
      "loss": 0.0344,
      "step": 6427
    },
    {
      "epoch": 3.000233372228705,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015883786488883416,
      "loss": 0.0389,
      "step": 6428
    },
    {
      "epoch": 3.0007001166861142,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015882599948958384,
      "loss": 0.0385,
      "step": 6429
    },
    {
      "epoch": 3.001166861143524,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00015881413282374448,
      "loss": 0.0257,
      "step": 6430
    },
    {
      "epoch": 3.0016336056009334,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015880226489157158,
      "loss": 0.0439,
      "step": 6431
    },
    {
      "epoch": 3.002100350058343,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015879039569332068,
      "loss": 0.0352,
      "step": 6432
    },
    {
      "epoch": 3.002100350058343,
      "eval_loss": 1.6984552145004272,
      "eval_runtime": 56.6516,
      "eval_samples_per_second": 31.844,
      "eval_steps_per_second": 3.989,
      "step": 6432
    },
    {
      "epoch": 3.0025670945157525,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015877852522924732,
      "loss": 0.0355,
      "step": 6433
    },
    {
      "epoch": 3.0030338389731623,
      "grad_norm": 0.75,
      "learning_rate": 0.00015876665349960712,
      "loss": 0.0413,
      "step": 6434
    },
    {
      "epoch": 3.0035005834305717,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015875478050465564,
      "loss": 0.0377,
      "step": 6435
    },
    {
      "epoch": 3.0039673278879815,
      "grad_norm": 0.625,
      "learning_rate": 0.00015874290624464857,
      "loss": 0.0338,
      "step": 6436
    },
    {
      "epoch": 3.004434072345391,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015873103071984153,
      "loss": 0.032,
      "step": 6437
    },
    {
      "epoch": 3.0049008168028006,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015871915393049024,
      "loss": 0.0349,
      "step": 6438
    },
    {
      "epoch": 3.00536756126021,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015870727587685043,
      "loss": 0.0259,
      "step": 6439
    },
    {
      "epoch": 3.00583430571762,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001586953965591778,
      "loss": 0.0251,
      "step": 6440
    },
    {
      "epoch": 3.006301050175029,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00015868351597772822,
      "loss": 0.0528,
      "step": 6441
    },
    {
      "epoch": 3.006767794632439,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015867163413275738,
      "loss": 0.0298,
      "step": 6442
    },
    {
      "epoch": 3.0072345390898483,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015865975102452118,
      "loss": 0.04,
      "step": 6443
    },
    {
      "epoch": 3.007701283547258,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015864786665327544,
      "loss": 0.028,
      "step": 6444
    },
    {
      "epoch": 3.0081680280046674,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015863598101927608,
      "loss": 0.0247,
      "step": 6445
    },
    {
      "epoch": 3.0086347724620772,
      "grad_norm": 0.86328125,
      "learning_rate": 0.000158624094122779,
      "loss": 0.0329,
      "step": 6446
    },
    {
      "epoch": 3.0091015169194866,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001586122059640401,
      "loss": 0.0345,
      "step": 6447
    },
    {
      "epoch": 3.0095682613768964,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015860031654331538,
      "loss": 0.0317,
      "step": 6448
    },
    {
      "epoch": 3.0100350058343057,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015858842586086085,
      "loss": 0.0252,
      "step": 6449
    },
    {
      "epoch": 3.010501750291715,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001585765339169325,
      "loss": 0.0352,
      "step": 6450
    },
    {
      "epoch": 3.010968494749125,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015856464071178637,
      "loss": 0.0356,
      "step": 6451
    },
    {
      "epoch": 3.011435239206534,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015855274624567854,
      "loss": 0.0306,
      "step": 6452
    },
    {
      "epoch": 3.011901983663944,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00015854085051886512,
      "loss": 0.0237,
      "step": 6453
    },
    {
      "epoch": 3.0123687281213534,
      "grad_norm": 1.359375,
      "learning_rate": 0.00015852895353160224,
      "loss": 0.0491,
      "step": 6454
    },
    {
      "epoch": 3.012835472578763,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015851705528414605,
      "loss": 0.0448,
      "step": 6455
    },
    {
      "epoch": 3.0133022170361725,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015850515577675275,
      "loss": 0.0415,
      "step": 6456
    },
    {
      "epoch": 3.0137689614935823,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015849325500967852,
      "loss": 0.0258,
      "step": 6457
    },
    {
      "epoch": 3.0142357059509917,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001584813529831796,
      "loss": 0.0319,
      "step": 6458
    },
    {
      "epoch": 3.0147024504084015,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015846944969751228,
      "loss": 0.0236,
      "step": 6459
    },
    {
      "epoch": 3.015169194865811,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015845754515293284,
      "loss": 0.0324,
      "step": 6460
    },
    {
      "epoch": 3.0156359393232206,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001584456393496976,
      "loss": 0.0339,
      "step": 6461
    },
    {
      "epoch": 3.01610268378063,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015843373228806288,
      "loss": 0.0329,
      "step": 6462
    },
    {
      "epoch": 3.0165694282380398,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015842182396828505,
      "loss": 0.0302,
      "step": 6463
    },
    {
      "epoch": 3.017036172695449,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015840991439062055,
      "loss": 0.0253,
      "step": 6464
    },
    {
      "epoch": 3.017502917152859,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001583980035553258,
      "loss": 0.0436,
      "step": 6465
    },
    {
      "epoch": 3.0179696616102683,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00015838609146265722,
      "loss": 0.0263,
      "step": 6466
    },
    {
      "epoch": 3.018436406067678,
      "grad_norm": 0.890625,
      "learning_rate": 0.00015837417811287132,
      "loss": 0.0462,
      "step": 6467
    },
    {
      "epoch": 3.0189031505250874,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015836226350622457,
      "loss": 0.0424,
      "step": 6468
    },
    {
      "epoch": 3.019369894982497,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015835034764297357,
      "loss": 0.0291,
      "step": 6469
    },
    {
      "epoch": 3.0198366394399065,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015833843052337482,
      "loss": 0.0348,
      "step": 6470
    },
    {
      "epoch": 3.0203033838973163,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015832651214768497,
      "loss": 0.031,
      "step": 6471
    },
    {
      "epoch": 3.0207701283547257,
      "grad_norm": 0.5625,
      "learning_rate": 0.00015831459251616057,
      "loss": 0.0251,
      "step": 6472
    },
    {
      "epoch": 3.0212368728121355,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001583026716290583,
      "loss": 0.0294,
      "step": 6473
    },
    {
      "epoch": 3.021703617269545,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00015829074948663484,
      "loss": 0.0346,
      "step": 6474
    },
    {
      "epoch": 3.0221703617269546,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015827882608914683,
      "loss": 0.0346,
      "step": 6475
    },
    {
      "epoch": 3.022637106184364,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015826690143685106,
      "loss": 0.0376,
      "step": 6476
    },
    {
      "epoch": 3.023103850641774,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015825497553000422,
      "loss": 0.0276,
      "step": 6477
    },
    {
      "epoch": 3.023570595099183,
      "grad_norm": 1.1875,
      "learning_rate": 0.0001582430483688632,
      "loss": 0.0324,
      "step": 6478
    },
    {
      "epoch": 3.024037339556593,
      "grad_norm": 0.75,
      "learning_rate": 0.00015823111995368465,
      "loss": 0.0325,
      "step": 6479
    },
    {
      "epoch": 3.0245040840140023,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001582191902847255,
      "loss": 0.0307,
      "step": 6480
    },
    {
      "epoch": 3.024970828471412,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001582072593622426,
      "loss": 0.0273,
      "step": 6481
    },
    {
      "epoch": 3.0254375729288214,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001581953271864928,
      "loss": 0.0226,
      "step": 6482
    },
    {
      "epoch": 3.0259043173862312,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015818339375773304,
      "loss": 0.0284,
      "step": 6483
    },
    {
      "epoch": 3.0263710618436406,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001581714590762203,
      "loss": 0.0378,
      "step": 6484
    },
    {
      "epoch": 3.0268378063010504,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015815952314221146,
      "loss": 0.0286,
      "step": 6485
    },
    {
      "epoch": 3.0273045507584597,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015814758595596358,
      "loss": 0.02,
      "step": 6486
    },
    {
      "epoch": 3.0277712952158695,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001581356475177336,
      "loss": 0.0397,
      "step": 6487
    },
    {
      "epoch": 3.028238039673279,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015812370782777873,
      "loss": 0.0346,
      "step": 6488
    },
    {
      "epoch": 3.0287047841306887,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015811176688635588,
      "loss": 0.0311,
      "step": 6489
    },
    {
      "epoch": 3.029171528588098,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001580998246937222,
      "loss": 0.0233,
      "step": 6490
    },
    {
      "epoch": 3.0296382730455074,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015808788125013486,
      "loss": 0.0241,
      "step": 6491
    },
    {
      "epoch": 3.030105017502917,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015807593655585095,
      "loss": 0.0367,
      "step": 6492
    },
    {
      "epoch": 3.0305717619603265,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015806399061112772,
      "loss": 0.0278,
      "step": 6493
    },
    {
      "epoch": 3.0310385064177363,
      "grad_norm": 0.59375,
      "learning_rate": 0.00015805204341622236,
      "loss": 0.0311,
      "step": 6494
    },
    {
      "epoch": 3.0315052508751457,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001580400949713921,
      "loss": 0.0253,
      "step": 6495
    },
    {
      "epoch": 3.0319719953325555,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015802814527689416,
      "loss": 0.0313,
      "step": 6496
    },
    {
      "epoch": 3.032438739789965,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015801619433298587,
      "loss": 0.0267,
      "step": 6497
    },
    {
      "epoch": 3.0329054842473746,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001580042421399246,
      "loss": 0.0408,
      "step": 6498
    },
    {
      "epoch": 3.033372228704784,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015799228869796758,
      "loss": 0.039,
      "step": 6499
    },
    {
      "epoch": 3.0338389731621938,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00015798033400737227,
      "loss": 0.0667,
      "step": 6500
    },
    {
      "epoch": 3.034305717619603,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015796837806839604,
      "loss": 0.0361,
      "step": 6501
    },
    {
      "epoch": 3.034772462077013,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001579564208812963,
      "loss": 0.0272,
      "step": 6502
    },
    {
      "epoch": 3.0352392065344223,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015794446244633053,
      "loss": 0.0399,
      "step": 6503
    },
    {
      "epoch": 3.035705950991832,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001579325027637562,
      "loss": 0.0434,
      "step": 6504
    },
    {
      "epoch": 3.0361726954492414,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015792054183383077,
      "loss": 0.0328,
      "step": 6505
    },
    {
      "epoch": 3.036639439906651,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015790857965681184,
      "loss": 0.031,
      "step": 6506
    },
    {
      "epoch": 3.0371061843640605,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015789661623295692,
      "loss": 0.0359,
      "step": 6507
    },
    {
      "epoch": 3.0375729288214703,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015788465156252364,
      "loss": 0.0356,
      "step": 6508
    },
    {
      "epoch": 3.0380396732788797,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015787268564576955,
      "loss": 0.0352,
      "step": 6509
    },
    {
      "epoch": 3.0385064177362895,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015786071848295237,
      "loss": 0.0266,
      "step": 6510
    },
    {
      "epoch": 3.038973162193699,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001578487500743297,
      "loss": 0.036,
      "step": 6511
    },
    {
      "epoch": 3.0394399066511086,
      "grad_norm": 0.984375,
      "learning_rate": 0.00015783678042015925,
      "loss": 0.0465,
      "step": 6512
    },
    {
      "epoch": 3.039906651108518,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00015782480952069878,
      "loss": 0.0358,
      "step": 6513
    },
    {
      "epoch": 3.040373395565928,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015781283737620599,
      "loss": 0.0343,
      "step": 6514
    },
    {
      "epoch": 3.040840140023337,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015780086398693865,
      "loss": 0.0302,
      "step": 6515
    },
    {
      "epoch": 3.041306884480747,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015778888935315458,
      "loss": 0.0388,
      "step": 6516
    },
    {
      "epoch": 3.0417736289381563,
      "grad_norm": 0.75,
      "learning_rate": 0.00015777691347511162,
      "loss": 0.0356,
      "step": 6517
    },
    {
      "epoch": 3.042240373395566,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015776493635306758,
      "loss": 0.0294,
      "step": 6518
    },
    {
      "epoch": 3.0427071178529754,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001577529579872804,
      "loss": 0.0322,
      "step": 6519
    },
    {
      "epoch": 3.0431738623103852,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00015774097837800794,
      "loss": 0.0398,
      "step": 6520
    },
    {
      "epoch": 3.0436406067677946,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015772899752550815,
      "loss": 0.0395,
      "step": 6521
    },
    {
      "epoch": 3.0441073512252044,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015771701543003904,
      "loss": 0.0322,
      "step": 6522
    },
    {
      "epoch": 3.0445740956826137,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001577050320918585,
      "loss": 0.0267,
      "step": 6523
    },
    {
      "epoch": 3.0450408401400235,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001576930475112246,
      "loss": 0.0235,
      "step": 6524
    },
    {
      "epoch": 3.045507584597433,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015768106168839537,
      "loss": 0.0221,
      "step": 6525
    },
    {
      "epoch": 3.0459743290548427,
      "grad_norm": 0.99609375,
      "learning_rate": 0.0001576690746236289,
      "loss": 0.0489,
      "step": 6526
    },
    {
      "epoch": 3.046441073512252,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015765708631718326,
      "loss": 0.034,
      "step": 6527
    },
    {
      "epoch": 3.046907817969662,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015764509676931656,
      "loss": 0.0399,
      "step": 6528
    },
    {
      "epoch": 3.047374562427071,
      "grad_norm": 0.74609375,
      "learning_rate": 0.000157633105980287,
      "loss": 0.0326,
      "step": 6529
    },
    {
      "epoch": 3.047841306884481,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001576211139503527,
      "loss": 0.0295,
      "step": 6530
    },
    {
      "epoch": 3.0483080513418903,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015760912067977188,
      "loss": 0.0352,
      "step": 6531
    },
    {
      "epoch": 3.0487747957993,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001575971261688028,
      "loss": 0.0436,
      "step": 6532
    },
    {
      "epoch": 3.0492415402567095,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00015758513041770367,
      "loss": 0.0388,
      "step": 6533
    },
    {
      "epoch": 3.049708284714119,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001575731334267328,
      "loss": 0.0333,
      "step": 6534
    },
    {
      "epoch": 3.0501750291715286,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015756113519614848,
      "loss": 0.0358,
      "step": 6535
    },
    {
      "epoch": 3.050641773628938,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015754913572620907,
      "loss": 0.0464,
      "step": 6536
    },
    {
      "epoch": 3.0511085180863478,
      "grad_norm": 0.625,
      "learning_rate": 0.00015753713501717292,
      "loss": 0.0251,
      "step": 6537
    },
    {
      "epoch": 3.051575262543757,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015752513306929839,
      "loss": 0.038,
      "step": 6538
    },
    {
      "epoch": 3.052042007001167,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015751312988284392,
      "loss": 0.0416,
      "step": 6539
    },
    {
      "epoch": 3.0525087514585763,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00015750112545806797,
      "loss": 0.0425,
      "step": 6540
    },
    {
      "epoch": 3.052975495915986,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000157489119795229,
      "loss": 0.0455,
      "step": 6541
    },
    {
      "epoch": 3.0534422403733954,
      "grad_norm": 0.90625,
      "learning_rate": 0.00015747711289458547,
      "loss": 0.0404,
      "step": 6542
    },
    {
      "epoch": 3.053908984830805,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015746510475639594,
      "loss": 0.0385,
      "step": 6543
    },
    {
      "epoch": 3.0543757292882145,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015745309538091893,
      "loss": 0.0349,
      "step": 6544
    },
    {
      "epoch": 3.0548424737456243,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015744108476841305,
      "loss": 0.0338,
      "step": 6545
    },
    {
      "epoch": 3.0553092182030337,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015742907291913687,
      "loss": 0.029,
      "step": 6546
    },
    {
      "epoch": 3.0557759626604435,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015741705983334905,
      "loss": 0.0283,
      "step": 6547
    },
    {
      "epoch": 3.056242707117853,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001574050455113082,
      "loss": 0.0291,
      "step": 6548
    },
    {
      "epoch": 3.0567094515752626,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00015739302995327305,
      "loss": 0.0321,
      "step": 6549
    },
    {
      "epoch": 3.057176196032672,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00015738101315950227,
      "loss": 0.0307,
      "step": 6550
    },
    {
      "epoch": 3.057642940490082,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001573689951302546,
      "loss": 0.0275,
      "step": 6551
    },
    {
      "epoch": 3.058109684947491,
      "grad_norm": 0.96875,
      "learning_rate": 0.00015735697586578885,
      "loss": 0.0413,
      "step": 6552
    },
    {
      "epoch": 3.058576429404901,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015734495536636375,
      "loss": 0.0341,
      "step": 6553
    },
    {
      "epoch": 3.0590431738623103,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015733293363223813,
      "loss": 0.0334,
      "step": 6554
    },
    {
      "epoch": 3.05950991831972,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00015732091066367085,
      "loss": 0.0612,
      "step": 6555
    },
    {
      "epoch": 3.0599766627771294,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015730888646092077,
      "loss": 0.0426,
      "step": 6556
    },
    {
      "epoch": 3.0604434072345392,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015729686102424677,
      "loss": 0.0302,
      "step": 6557
    },
    {
      "epoch": 3.0609101516919486,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00015728483435390776,
      "loss": 0.0351,
      "step": 6558
    },
    {
      "epoch": 3.0613768961493584,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015727280645016275,
      "loss": 0.0359,
      "step": 6559
    },
    {
      "epoch": 3.0618436406067677,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015726077731327066,
      "loss": 0.033,
      "step": 6560
    },
    {
      "epoch": 3.0623103850641775,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015724874694349045,
      "loss": 0.0295,
      "step": 6561
    },
    {
      "epoch": 3.062777129521587,
      "grad_norm": 0.5625,
      "learning_rate": 0.00015723671534108123,
      "loss": 0.0346,
      "step": 6562
    },
    {
      "epoch": 3.0632438739789967,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015722468250630204,
      "loss": 0.0348,
      "step": 6563
    },
    {
      "epoch": 3.063710618436406,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015721264843941195,
      "loss": 0.0331,
      "step": 6564
    },
    {
      "epoch": 3.064177362893816,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00015720061314067007,
      "loss": 0.0474,
      "step": 6565
    },
    {
      "epoch": 3.064644107351225,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00015718857661033551,
      "loss": 0.0193,
      "step": 6566
    },
    {
      "epoch": 3.065110851808635,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001571765388486674,
      "loss": 0.0254,
      "step": 6567
    },
    {
      "epoch": 3.0655775962660443,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015716449985592502,
      "loss": 0.0327,
      "step": 6568
    },
    {
      "epoch": 3.066044340723454,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015715245963236752,
      "loss": 0.0299,
      "step": 6569
    },
    {
      "epoch": 3.0665110851808635,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015714041817825418,
      "loss": 0.032,
      "step": 6570
    },
    {
      "epoch": 3.0669778296382733,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001571283754938442,
      "loss": 0.0343,
      "step": 6571
    },
    {
      "epoch": 3.0674445740956826,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015711633157939692,
      "loss": 0.0488,
      "step": 6572
    },
    {
      "epoch": 3.067911318553092,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015710428643517162,
      "loss": 0.0354,
      "step": 6573
    },
    {
      "epoch": 3.0683780630105018,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015709224006142772,
      "loss": 0.033,
      "step": 6574
    },
    {
      "epoch": 3.068844807467911,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015708019245842453,
      "loss": 0.0313,
      "step": 6575
    },
    {
      "epoch": 3.069311551925321,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015706814362642147,
      "loss": 0.0283,
      "step": 6576
    },
    {
      "epoch": 3.0697782963827303,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015705609356567797,
      "loss": 0.0339,
      "step": 6577
    },
    {
      "epoch": 3.07024504084014,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015704404227645345,
      "loss": 0.0308,
      "step": 6578
    },
    {
      "epoch": 3.0707117852975494,
      "grad_norm": 0.875,
      "learning_rate": 0.0001570319897590074,
      "loss": 0.0275,
      "step": 6579
    },
    {
      "epoch": 3.071178529754959,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015701993601359936,
      "loss": 0.0316,
      "step": 6580
    },
    {
      "epoch": 3.0716452742123685,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00015700788104048882,
      "loss": 0.02,
      "step": 6581
    },
    {
      "epoch": 3.0721120186697783,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015699582483993534,
      "loss": 0.0315,
      "step": 6582
    },
    {
      "epoch": 3.0725787631271877,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00015698376741219854,
      "loss": 0.0459,
      "step": 6583
    },
    {
      "epoch": 3.0730455075845975,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015697170875753798,
      "loss": 0.0312,
      "step": 6584
    },
    {
      "epoch": 3.073512252042007,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015695964887621334,
      "loss": 0.0312,
      "step": 6585
    },
    {
      "epoch": 3.0739789964994166,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015694758776848425,
      "loss": 0.0299,
      "step": 6586
    },
    {
      "epoch": 3.074445740956826,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001569355254346104,
      "loss": 0.0335,
      "step": 6587
    },
    {
      "epoch": 3.074912485414236,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001569234618748515,
      "loss": 0.0365,
      "step": 6588
    },
    {
      "epoch": 3.075379229871645,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015691139708946735,
      "loss": 0.0455,
      "step": 6589
    },
    {
      "epoch": 3.075845974329055,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015689933107871767,
      "loss": 0.0441,
      "step": 6590
    },
    {
      "epoch": 3.0763127187864643,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00015688726384286222,
      "loss": 0.0511,
      "step": 6591
    },
    {
      "epoch": 3.076779463243874,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015687519538216092,
      "loss": 0.0299,
      "step": 6592
    },
    {
      "epoch": 3.0772462077012834,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015686312569687352,
      "loss": 0.0271,
      "step": 6593
    },
    {
      "epoch": 3.0777129521586932,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015685105478725997,
      "loss": 0.0254,
      "step": 6594
    },
    {
      "epoch": 3.0781796966161026,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001568389826535801,
      "loss": 0.0287,
      "step": 6595
    },
    {
      "epoch": 3.0786464410735124,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00015682690929609387,
      "loss": 0.0428,
      "step": 6596
    },
    {
      "epoch": 3.0791131855309217,
      "grad_norm": 0.75,
      "learning_rate": 0.00015681483471506125,
      "loss": 0.034,
      "step": 6597
    },
    {
      "epoch": 3.0795799299883315,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015680275891074219,
      "loss": 0.0479,
      "step": 6598
    },
    {
      "epoch": 3.080046674445741,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001567906818833967,
      "loss": 0.0206,
      "step": 6599
    },
    {
      "epoch": 3.0805134189031507,
      "grad_norm": 0.890625,
      "learning_rate": 0.00015677860363328482,
      "loss": 0.0438,
      "step": 6600
    },
    {
      "epoch": 3.08098016336056,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001567665241606666,
      "loss": 0.0271,
      "step": 6601
    },
    {
      "epoch": 3.08144690781797,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015675444346580216,
      "loss": 0.0398,
      "step": 6602
    },
    {
      "epoch": 3.081913652275379,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015674236154895156,
      "loss": 0.0232,
      "step": 6603
    },
    {
      "epoch": 3.082380396732789,
      "grad_norm": 1.3125,
      "learning_rate": 0.00015673027841037497,
      "loss": 0.0515,
      "step": 6604
    },
    {
      "epoch": 3.0828471411901983,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015671819405033255,
      "loss": 0.032,
      "step": 6605
    },
    {
      "epoch": 3.083313885647608,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015670610846908444,
      "loss": 0.0335,
      "step": 6606
    },
    {
      "epoch": 3.0837806301050175,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015669402166689097,
      "loss": 0.0334,
      "step": 6607
    },
    {
      "epoch": 3.0842473745624273,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015668193364401224,
      "loss": 0.0344,
      "step": 6608
    },
    {
      "epoch": 3.0847141190198366,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015666984440070863,
      "loss": 0.0283,
      "step": 6609
    },
    {
      "epoch": 3.0851808634772464,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015665775393724037,
      "loss": 0.0319,
      "step": 6610
    },
    {
      "epoch": 3.0856476079346558,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001566456622538678,
      "loss": 0.044,
      "step": 6611
    },
    {
      "epoch": 3.0861143523920656,
      "grad_norm": 0.75,
      "learning_rate": 0.0001566335693508513,
      "loss": 0.0278,
      "step": 6612
    },
    {
      "epoch": 3.086581096849475,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001566214752284512,
      "loss": 0.0374,
      "step": 6613
    },
    {
      "epoch": 3.0870478413068847,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015660937988692795,
      "loss": 0.0303,
      "step": 6614
    },
    {
      "epoch": 3.087514585764294,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00015659728332654188,
      "loss": 0.0309,
      "step": 6615
    },
    {
      "epoch": 3.0879813302217034,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015658518554755355,
      "loss": 0.0331,
      "step": 6616
    },
    {
      "epoch": 3.088448074679113,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001565730865502234,
      "loss": 0.0287,
      "step": 6617
    },
    {
      "epoch": 3.0889148191365225,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001565609863348119,
      "loss": 0.0288,
      "step": 6618
    },
    {
      "epoch": 3.0893815635939323,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00015654888490157962,
      "loss": 0.0265,
      "step": 6619
    },
    {
      "epoch": 3.0898483080513417,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001565367822507871,
      "loss": 0.0389,
      "step": 6620
    },
    {
      "epoch": 3.0903150525087515,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015652467838269494,
      "loss": 0.0287,
      "step": 6621
    },
    {
      "epoch": 3.090781796966161,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015651257329756373,
      "loss": 0.0319,
      "step": 6622
    },
    {
      "epoch": 3.0912485414235706,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001565004669956541,
      "loss": 0.0398,
      "step": 6623
    },
    {
      "epoch": 3.09171528588098,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015648835947722675,
      "loss": 0.0382,
      "step": 6624
    },
    {
      "epoch": 3.09218203033839,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015647625074254235,
      "loss": 0.0365,
      "step": 6625
    },
    {
      "epoch": 3.092648774795799,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015646414079186158,
      "loss": 0.0333,
      "step": 6626
    },
    {
      "epoch": 3.093115519253209,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015645202962544523,
      "loss": 0.0256,
      "step": 6627
    },
    {
      "epoch": 3.0935822637106183,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015643991724355406,
      "loss": 0.0376,
      "step": 6628
    },
    {
      "epoch": 3.094049008168028,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001564278036464488,
      "loss": 0.0252,
      "step": 6629
    },
    {
      "epoch": 3.0945157526254374,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001564156888343904,
      "loss": 0.0288,
      "step": 6630
    },
    {
      "epoch": 3.0949824970828472,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015640357280763956,
      "loss": 0.0363,
      "step": 6631
    },
    {
      "epoch": 3.0954492415402566,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015639145556645725,
      "loss": 0.0361,
      "step": 6632
    },
    {
      "epoch": 3.0959159859976664,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015637933711110433,
      "loss": 0.0332,
      "step": 6633
    },
    {
      "epoch": 3.0963827304550757,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015636721744184172,
      "loss": 0.0299,
      "step": 6634
    },
    {
      "epoch": 3.0968494749124855,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015635509655893037,
      "loss": 0.03,
      "step": 6635
    },
    {
      "epoch": 3.097316219369895,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015634297446263126,
      "loss": 0.038,
      "step": 6636
    },
    {
      "epoch": 3.0977829638273047,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015633085115320541,
      "loss": 0.0375,
      "step": 6637
    },
    {
      "epoch": 3.098249708284714,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015631872663091385,
      "loss": 0.0338,
      "step": 6638
    },
    {
      "epoch": 3.098716452742124,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015630660089601759,
      "loss": 0.0359,
      "step": 6639
    },
    {
      "epoch": 3.099183197199533,
      "grad_norm": 0.546875,
      "learning_rate": 0.00015629447394877776,
      "loss": 0.0157,
      "step": 6640
    },
    {
      "epoch": 3.099649941656943,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015628234578945547,
      "loss": 0.0459,
      "step": 6641
    },
    {
      "epoch": 3.1001166861143523,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015627021641831176,
      "loss": 0.0333,
      "step": 6642
    },
    {
      "epoch": 3.100583430571762,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001562580858356079,
      "loss": 0.0376,
      "step": 6643
    },
    {
      "epoch": 3.1010501750291715,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015624595404160503,
      "loss": 0.0347,
      "step": 6644
    },
    {
      "epoch": 3.1015169194865813,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00015623382103656437,
      "loss": 0.0256,
      "step": 6645
    },
    {
      "epoch": 3.1019836639439906,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015622168682074715,
      "loss": 0.0416,
      "step": 6646
    },
    {
      "epoch": 3.1024504084014004,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015620955139441462,
      "loss": 0.0339,
      "step": 6647
    },
    {
      "epoch": 3.1029171528588098,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015619741475782808,
      "loss": 0.025,
      "step": 6648
    },
    {
      "epoch": 3.1033838973162196,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015618527691124885,
      "loss": 0.0334,
      "step": 6649
    },
    {
      "epoch": 3.103850641773629,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015617313785493828,
      "loss": 0.0373,
      "step": 6650
    },
    {
      "epoch": 3.1043173862310387,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00015616099758915773,
      "loss": 0.0164,
      "step": 6651
    },
    {
      "epoch": 3.104784130688448,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001561488561141686,
      "loss": 0.0266,
      "step": 6652
    },
    {
      "epoch": 3.105250875145858,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00015613671343023223,
      "loss": 0.0293,
      "step": 6653
    },
    {
      "epoch": 3.105717619603267,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00015612456953761021,
      "loss": 0.0309,
      "step": 6654
    },
    {
      "epoch": 3.106184364060677,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001561124244365639,
      "loss": 0.0382,
      "step": 6655
    },
    {
      "epoch": 3.1066511085180863,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015610027812735486,
      "loss": 0.0322,
      "step": 6656
    },
    {
      "epoch": 3.107117852975496,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001560881306102446,
      "loss": 0.0414,
      "step": 6657
    },
    {
      "epoch": 3.1075845974329055,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001560759818854946,
      "loss": 0.0245,
      "step": 6658
    },
    {
      "epoch": 3.108051341890315,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015606383195336654,
      "loss": 0.0294,
      "step": 6659
    },
    {
      "epoch": 3.1085180863477246,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015605168081412197,
      "loss": 0.0353,
      "step": 6660
    },
    {
      "epoch": 3.108984830805134,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001560395284680225,
      "loss": 0.0309,
      "step": 6661
    },
    {
      "epoch": 3.109451575262544,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001560273749153298,
      "loss": 0.0273,
      "step": 6662
    },
    {
      "epoch": 3.109918319719953,
      "grad_norm": 0.625,
      "learning_rate": 0.00015601522015630558,
      "loss": 0.0238,
      "step": 6663
    },
    {
      "epoch": 3.110385064177363,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001560030641912115,
      "loss": 0.036,
      "step": 6664
    },
    {
      "epoch": 3.1108518086347723,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015599090702030934,
      "loss": 0.0277,
      "step": 6665
    },
    {
      "epoch": 3.111318553092182,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015597874864386083,
      "loss": 0.0278,
      "step": 6666
    },
    {
      "epoch": 3.1117852975495914,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00015596658906212772,
      "loss": 0.0519,
      "step": 6667
    },
    {
      "epoch": 3.1122520420070012,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015595442827537185,
      "loss": 0.0307,
      "step": 6668
    },
    {
      "epoch": 3.1127187864644106,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015594226628385508,
      "loss": 0.0342,
      "step": 6669
    },
    {
      "epoch": 3.1131855309218204,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015593010308783924,
      "loss": 0.0335,
      "step": 6670
    },
    {
      "epoch": 3.1136522753792297,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015591793868758623,
      "loss": 0.0347,
      "step": 6671
    },
    {
      "epoch": 3.1141190198366395,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015590577308335796,
      "loss": 0.0336,
      "step": 6672
    },
    {
      "epoch": 3.114585764294049,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015589360627541634,
      "loss": 0.0363,
      "step": 6673
    },
    {
      "epoch": 3.1150525087514587,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001558814382640234,
      "loss": 0.0223,
      "step": 6674
    },
    {
      "epoch": 3.115519253208868,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00015586926904944106,
      "loss": 0.0359,
      "step": 6675
    },
    {
      "epoch": 3.115985997666278,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001558570986319314,
      "loss": 0.0303,
      "step": 6676
    },
    {
      "epoch": 3.116452742123687,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001558449270117564,
      "loss": 0.0362,
      "step": 6677
    },
    {
      "epoch": 3.116919486581097,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015583275418917819,
      "loss": 0.0416,
      "step": 6678
    },
    {
      "epoch": 3.1173862310385063,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015582058016445883,
      "loss": 0.0282,
      "step": 6679
    },
    {
      "epoch": 3.117852975495916,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001558084049378604,
      "loss": 0.0311,
      "step": 6680
    },
    {
      "epoch": 3.1183197199533255,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015579622850964514,
      "loss": 0.0397,
      "step": 6681
    },
    {
      "epoch": 3.1187864644107353,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015578405088007515,
      "loss": 0.0284,
      "step": 6682
    },
    {
      "epoch": 3.1192532088681446,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015577187204941266,
      "loss": 0.0404,
      "step": 6683
    },
    {
      "epoch": 3.1197199533255544,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015575969201791988,
      "loss": 0.038,
      "step": 6684
    },
    {
      "epoch": 3.1201866977829638,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015574751078585908,
      "loss": 0.0269,
      "step": 6685
    },
    {
      "epoch": 3.1206534422403736,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001557353283534925,
      "loss": 0.0368,
      "step": 6686
    },
    {
      "epoch": 3.121120186697783,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015572314472108248,
      "loss": 0.0308,
      "step": 6687
    },
    {
      "epoch": 3.1215869311551927,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015571095988889134,
      "loss": 0.0352,
      "step": 6688
    },
    {
      "epoch": 3.122053675612602,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001556987738571814,
      "loss": 0.0278,
      "step": 6689
    },
    {
      "epoch": 3.122520420070012,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015568658662621507,
      "loss": 0.0451,
      "step": 6690
    },
    {
      "epoch": 3.122987164527421,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015567439819625474,
      "loss": 0.0338,
      "step": 6691
    },
    {
      "epoch": 3.123453908984831,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001556622085675628,
      "loss": 0.0347,
      "step": 6692
    },
    {
      "epoch": 3.1239206534422403,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001556500177404018,
      "loss": 0.0306,
      "step": 6693
    },
    {
      "epoch": 3.12438739789965,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00015563782571503416,
      "loss": 0.0239,
      "step": 6694
    },
    {
      "epoch": 3.1248541423570595,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001556256324917224,
      "loss": 0.0364,
      "step": 6695
    },
    {
      "epoch": 3.1253208868144693,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015561343807072908,
      "loss": 0.0284,
      "step": 6696
    },
    {
      "epoch": 3.1257876312718786,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00015560124245231672,
      "loss": 0.039,
      "step": 6697
    },
    {
      "epoch": 3.126254375729288,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001555890456367479,
      "loss": 0.0365,
      "step": 6698
    },
    {
      "epoch": 3.126721120186698,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00015557684762428528,
      "loss": 0.0231,
      "step": 6699
    },
    {
      "epoch": 3.1271878646441076,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015556464841519146,
      "loss": 0.0264,
      "step": 6700
    },
    {
      "epoch": 3.127654609101517,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001555524480097291,
      "loss": 0.0271,
      "step": 6701
    },
    {
      "epoch": 3.1281213535589263,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015554024640816088,
      "loss": 0.0257,
      "step": 6702
    },
    {
      "epoch": 3.128588098016336,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00015552804361074955,
      "loss": 0.0268,
      "step": 6703
    },
    {
      "epoch": 3.1290548424737454,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00015551583961775784,
      "loss": 0.0238,
      "step": 6704
    },
    {
      "epoch": 3.1295215869311552,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001555036344294485,
      "loss": 0.0269,
      "step": 6705
    },
    {
      "epoch": 3.1299883313885646,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015549142804608431,
      "loss": 0.0393,
      "step": 6706
    },
    {
      "epoch": 3.1304550758459744,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015547922046792813,
      "loss": 0.0293,
      "step": 6707
    },
    {
      "epoch": 3.1309218203033837,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001554670116952428,
      "loss": 0.0255,
      "step": 6708
    },
    {
      "epoch": 3.1313885647607935,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015545480172829112,
      "loss": 0.0377,
      "step": 6709
    },
    {
      "epoch": 3.131855309218203,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015544259056733605,
      "loss": 0.0366,
      "step": 6710
    },
    {
      "epoch": 3.1323220536756127,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015543037821264048,
      "loss": 0.0331,
      "step": 6711
    },
    {
      "epoch": 3.132788798133022,
      "grad_norm": 0.75,
      "learning_rate": 0.00015541816466446736,
      "loss": 0.0361,
      "step": 6712
    },
    {
      "epoch": 3.133255542590432,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001554059499230797,
      "loss": 0.0194,
      "step": 6713
    },
    {
      "epoch": 3.133722287047841,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00015539373398874045,
      "loss": 0.0339,
      "step": 6714
    },
    {
      "epoch": 3.134189031505251,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015538151686171264,
      "loss": 0.0265,
      "step": 6715
    },
    {
      "epoch": 3.1346557759626603,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015536929854225934,
      "loss": 0.0374,
      "step": 6716
    },
    {
      "epoch": 3.13512252042007,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001553570790306436,
      "loss": 0.0345,
      "step": 6717
    },
    {
      "epoch": 3.1355892648774795,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001553448583271285,
      "loss": 0.0403,
      "step": 6718
    },
    {
      "epoch": 3.1360560093348893,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015533263643197724,
      "loss": 0.0448,
      "step": 6719
    },
    {
      "epoch": 3.1365227537922986,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015532041334545291,
      "loss": 0.0251,
      "step": 6720
    },
    {
      "epoch": 3.1369894982497084,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001553081890678187,
      "loss": 0.0312,
      "step": 6721
    },
    {
      "epoch": 3.1374562427071178,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001552959635993378,
      "loss": 0.0316,
      "step": 6722
    },
    {
      "epoch": 3.1379229871645276,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015528373694027344,
      "loss": 0.025,
      "step": 6723
    },
    {
      "epoch": 3.138389731621937,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015527150909088895,
      "loss": 0.0287,
      "step": 6724
    },
    {
      "epoch": 3.1388564760793467,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015525928005144748,
      "loss": 0.0252,
      "step": 6725
    },
    {
      "epoch": 3.139323220536756,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00015524704982221242,
      "loss": 0.0214,
      "step": 6726
    },
    {
      "epoch": 3.139789964994166,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001552348184034471,
      "loss": 0.0411,
      "step": 6727
    },
    {
      "epoch": 3.140256709451575,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00015522258579541483,
      "loss": 0.0365,
      "step": 6728
    },
    {
      "epoch": 3.140723453908985,
      "grad_norm": 0.7734375,
      "learning_rate": 0.000155210351998379,
      "loss": 0.0366,
      "step": 6729
    },
    {
      "epoch": 3.1411901983663943,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015519811701260309,
      "loss": 0.0309,
      "step": 6730
    },
    {
      "epoch": 3.141656942823804,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015518588083835043,
      "loss": 0.0284,
      "step": 6731
    },
    {
      "epoch": 3.1421236872812135,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015517364347588454,
      "loss": 0.0408,
      "step": 6732
    },
    {
      "epoch": 3.1425904317386233,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001551614049254689,
      "loss": 0.0385,
      "step": 6733
    },
    {
      "epoch": 3.1430571761960326,
      "grad_norm": 0.77734375,
      "learning_rate": 0.000155149165187367,
      "loss": 0.0287,
      "step": 6734
    },
    {
      "epoch": 3.1435239206534424,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015513692426184238,
      "loss": 0.0343,
      "step": 6735
    },
    {
      "epoch": 3.143990665110852,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015512468214915864,
      "loss": 0.0261,
      "step": 6736
    },
    {
      "epoch": 3.144457409568261,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001551124388495793,
      "loss": 0.0337,
      "step": 6737
    },
    {
      "epoch": 3.144924154025671,
      "grad_norm": 0.546875,
      "learning_rate": 0.000155100194363368,
      "loss": 0.0308,
      "step": 6738
    },
    {
      "epoch": 3.1453908984830807,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001550879486907884,
      "loss": 0.0213,
      "step": 6739
    },
    {
      "epoch": 3.14585764294049,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001550757018321041,
      "loss": 0.0352,
      "step": 6740
    },
    {
      "epoch": 3.1463243873978994,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015506345378757885,
      "loss": 0.0257,
      "step": 6741
    },
    {
      "epoch": 3.1467911318553092,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001550512045574764,
      "loss": 0.03,
      "step": 6742
    },
    {
      "epoch": 3.1472578763127186,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015503895414206037,
      "loss": 0.0299,
      "step": 6743
    },
    {
      "epoch": 3.1477246207701284,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015502670254159462,
      "loss": 0.0261,
      "step": 6744
    },
    {
      "epoch": 3.1481913652275377,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001550144497563429,
      "loss": 0.0365,
      "step": 6745
    },
    {
      "epoch": 3.1486581096849475,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015500219578656904,
      "loss": 0.0284,
      "step": 6746
    },
    {
      "epoch": 3.149124854142357,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00015498994063253688,
      "loss": 0.0327,
      "step": 6747
    },
    {
      "epoch": 3.1495915985997667,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015497768429451026,
      "loss": 0.0284,
      "step": 6748
    },
    {
      "epoch": 3.150058343057176,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015496542677275317,
      "loss": 0.0391,
      "step": 6749
    },
    {
      "epoch": 3.150525087514586,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001549531680675294,
      "loss": 0.0267,
      "step": 6750
    },
    {
      "epoch": 3.150991831971995,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015494090817910294,
      "loss": 0.0377,
      "step": 6751
    },
    {
      "epoch": 3.151458576429405,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001549286471077378,
      "loss": 0.0311,
      "step": 6752
    },
    {
      "epoch": 3.1519253208868143,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015491638485369793,
      "loss": 0.0336,
      "step": 6753
    },
    {
      "epoch": 3.152392065344224,
      "grad_norm": 0.875,
      "learning_rate": 0.00015490412141724738,
      "loss": 0.0467,
      "step": 6754
    },
    {
      "epoch": 3.1528588098016335,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015489185679865014,
      "loss": 0.0297,
      "step": 6755
    },
    {
      "epoch": 3.1533255542590433,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00015487959099817034,
      "loss": 0.0499,
      "step": 6756
    },
    {
      "epoch": 3.1537922987164526,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015486732401607208,
      "loss": 0.0281,
      "step": 6757
    },
    {
      "epoch": 3.1542590431738624,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00015485505585261943,
      "loss": 0.0369,
      "step": 6758
    },
    {
      "epoch": 3.1547257876312718,
      "grad_norm": 0.90625,
      "learning_rate": 0.00015484278650807658,
      "loss": 0.0334,
      "step": 6759
    },
    {
      "epoch": 3.1551925320886816,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00015483051598270768,
      "loss": 0.0387,
      "step": 6760
    },
    {
      "epoch": 3.155659276546091,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015481824427677693,
      "loss": 0.0272,
      "step": 6761
    },
    {
      "epoch": 3.1561260210035007,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00015480597139054858,
      "loss": 0.0167,
      "step": 6762
    },
    {
      "epoch": 3.15659276546091,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015479369732428685,
      "loss": 0.0432,
      "step": 6763
    },
    {
      "epoch": 3.15705950991832,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015478142207825602,
      "loss": 0.0337,
      "step": 6764
    },
    {
      "epoch": 3.157526254375729,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015476914565272038,
      "loss": 0.026,
      "step": 6765
    },
    {
      "epoch": 3.157992998833139,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00015475686804794427,
      "loss": 0.0403,
      "step": 6766
    },
    {
      "epoch": 3.1584597432905483,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015474458926419203,
      "loss": 0.0349,
      "step": 6767
    },
    {
      "epoch": 3.158926487747958,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00015473230930172806,
      "loss": 0.0383,
      "step": 6768
    },
    {
      "epoch": 3.1593932322053675,
      "grad_norm": 0.5625,
      "learning_rate": 0.00015472002816081673,
      "loss": 0.0248,
      "step": 6769
    },
    {
      "epoch": 3.1598599766627773,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001547077458417225,
      "loss": 0.0309,
      "step": 6770
    },
    {
      "epoch": 3.1603267211201866,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001546954623447098,
      "loss": 0.0288,
      "step": 6771
    },
    {
      "epoch": 3.1607934655775964,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001546831776700431,
      "loss": 0.0323,
      "step": 6772
    },
    {
      "epoch": 3.161260210035006,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00015467089181798687,
      "loss": 0.0519,
      "step": 6773
    },
    {
      "epoch": 3.1617269544924156,
      "grad_norm": 0.765625,
      "learning_rate": 0.00015465860478880577,
      "loss": 0.0367,
      "step": 6774
    },
    {
      "epoch": 3.162193698949825,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015464631658276422,
      "loss": 0.0387,
      "step": 6775
    },
    {
      "epoch": 3.1626604434072347,
      "grad_norm": 1.2109375,
      "learning_rate": 0.00015463402720012682,
      "loss": 0.031,
      "step": 6776
    },
    {
      "epoch": 3.163127187864644,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001546217366411582,
      "loss": 0.0364,
      "step": 6777
    },
    {
      "epoch": 3.163593932322054,
      "grad_norm": 0.875,
      "learning_rate": 0.000154609444906123,
      "loss": 0.0436,
      "step": 6778
    },
    {
      "epoch": 3.1640606767794632,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015459715199528588,
      "loss": 0.0266,
      "step": 6779
    },
    {
      "epoch": 3.1645274212368726,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015458485790891148,
      "loss": 0.0434,
      "step": 6780
    },
    {
      "epoch": 3.1649941656942824,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015457256264726454,
      "loss": 0.0232,
      "step": 6781
    },
    {
      "epoch": 3.165460910151692,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015456026621060977,
      "loss": 0.0419,
      "step": 6782
    },
    {
      "epoch": 3.1659276546091015,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015454796859921195,
      "loss": 0.0324,
      "step": 6783
    },
    {
      "epoch": 3.166394399066511,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015453566981333582,
      "loss": 0.032,
      "step": 6784
    },
    {
      "epoch": 3.1668611435239207,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015452336985324623,
      "loss": 0.0438,
      "step": 6785
    },
    {
      "epoch": 3.16732788798133,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015451106871920795,
      "loss": 0.0325,
      "step": 6786
    },
    {
      "epoch": 3.16779463243874,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001544987664114859,
      "loss": 0.0282,
      "step": 6787
    },
    {
      "epoch": 3.168261376896149,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00015448646293034498,
      "loss": 0.0178,
      "step": 6788
    },
    {
      "epoch": 3.168728121353559,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015447415827605002,
      "loss": 0.0393,
      "step": 6789
    },
    {
      "epoch": 3.1691948658109683,
      "grad_norm": 0.76171875,
      "learning_rate": 0.000154461852448866,
      "loss": 0.03,
      "step": 6790
    },
    {
      "epoch": 3.169661610268378,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015444954544905788,
      "loss": 0.0284,
      "step": 6791
    },
    {
      "epoch": 3.1701283547257875,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001544372372768906,
      "loss": 0.0293,
      "step": 6792
    },
    {
      "epoch": 3.1705950991831973,
      "grad_norm": 0.5625,
      "learning_rate": 0.00015442492793262927,
      "loss": 0.0236,
      "step": 6793
    },
    {
      "epoch": 3.1710618436406066,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015441261741653878,
      "loss": 0.0371,
      "step": 6794
    },
    {
      "epoch": 3.1715285880980164,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001544003057288843,
      "loss": 0.036,
      "step": 6795
    },
    {
      "epoch": 3.1719953325554258,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015438799286993088,
      "loss": 0.0239,
      "step": 6796
    },
    {
      "epoch": 3.1724620770128356,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015437567883994362,
      "loss": 0.0343,
      "step": 6797
    },
    {
      "epoch": 3.172928821470245,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001543633636391877,
      "loss": 0.0254,
      "step": 6798
    },
    {
      "epoch": 3.1733955659276547,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001543510472679282,
      "loss": 0.0278,
      "step": 6799
    },
    {
      "epoch": 3.173862310385064,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015433872972643036,
      "loss": 0.0295,
      "step": 6800
    },
    {
      "epoch": 3.174329054842474,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015432641101495936,
      "loss": 0.0273,
      "step": 6801
    },
    {
      "epoch": 3.174795799299883,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001543140911337805,
      "loss": 0.0285,
      "step": 6802
    },
    {
      "epoch": 3.175262543757293,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015430177008315897,
      "loss": 0.0242,
      "step": 6803
    },
    {
      "epoch": 3.1757292882147024,
      "grad_norm": 0.75,
      "learning_rate": 0.00015428944786336009,
      "loss": 0.0359,
      "step": 6804
    },
    {
      "epoch": 3.176196032672112,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015427712447464915,
      "loss": 0.0373,
      "step": 6805
    },
    {
      "epoch": 3.1766627771295215,
      "grad_norm": 1.3828125,
      "learning_rate": 0.00015426479991729152,
      "loss": 0.0595,
      "step": 6806
    },
    {
      "epoch": 3.1771295215869313,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015425247419155252,
      "loss": 0.0335,
      "step": 6807
    },
    {
      "epoch": 3.1775962660443406,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015424014729769754,
      "loss": 0.0501,
      "step": 6808
    },
    {
      "epoch": 3.1780630105017504,
      "grad_norm": 1.0,
      "learning_rate": 0.00015422781923599206,
      "loss": 0.0329,
      "step": 6809
    },
    {
      "epoch": 3.17852975495916,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015421549000670145,
      "loss": 0.0261,
      "step": 6810
    },
    {
      "epoch": 3.1789964994165696,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015420315961009115,
      "loss": 0.0137,
      "step": 6811
    },
    {
      "epoch": 3.179463243873979,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015419082804642671,
      "loss": 0.0358,
      "step": 6812
    },
    {
      "epoch": 3.1799299883313887,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015417849531597358,
      "loss": 0.0205,
      "step": 6813
    },
    {
      "epoch": 3.180396732788798,
      "grad_norm": 1.1015625,
      "learning_rate": 0.0001541661614189974,
      "loss": 0.0326,
      "step": 6814
    },
    {
      "epoch": 3.180863477246208,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015415382635576363,
      "loss": 0.0282,
      "step": 6815
    },
    {
      "epoch": 3.1813302217036172,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00015414149012653788,
      "loss": 0.0382,
      "step": 6816
    },
    {
      "epoch": 3.181796966161027,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001541291527315858,
      "loss": 0.0233,
      "step": 6817
    },
    {
      "epoch": 3.1822637106184364,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00015411681417117298,
      "loss": 0.0328,
      "step": 6818
    },
    {
      "epoch": 3.182730455075846,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00015410447444556515,
      "loss": 0.0306,
      "step": 6819
    },
    {
      "epoch": 3.1831971995332555,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015409213355502792,
      "loss": 0.0413,
      "step": 6820
    },
    {
      "epoch": 3.1836639439906653,
      "grad_norm": 0.625,
      "learning_rate": 0.00015407979149982705,
      "loss": 0.0275,
      "step": 6821
    },
    {
      "epoch": 3.1841306884480747,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001540674482802283,
      "loss": 0.034,
      "step": 6822
    },
    {
      "epoch": 3.184597432905484,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015405510389649734,
      "loss": 0.0259,
      "step": 6823
    },
    {
      "epoch": 3.185064177362894,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00015404275834890002,
      "loss": 0.0306,
      "step": 6824
    },
    {
      "epoch": 3.1855309218203036,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015403041163770218,
      "loss": 0.0345,
      "step": 6825
    },
    {
      "epoch": 3.185997666277713,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015401806376316962,
      "loss": 0.0353,
      "step": 6826
    },
    {
      "epoch": 3.1864644107351223,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015400571472556823,
      "loss": 0.0363,
      "step": 6827
    },
    {
      "epoch": 3.186931155192532,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015399336452516385,
      "loss": 0.0273,
      "step": 6828
    },
    {
      "epoch": 3.1873978996499415,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00015398101316222245,
      "loss": 0.0369,
      "step": 6829
    },
    {
      "epoch": 3.1878646441073513,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00015396866063700995,
      "loss": 0.0256,
      "step": 6830
    },
    {
      "epoch": 3.1883313885647606,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015395630694979229,
      "loss": 0.0363,
      "step": 6831
    },
    {
      "epoch": 3.1887981330221704,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015394395210083545,
      "loss": 0.0331,
      "step": 6832
    },
    {
      "epoch": 3.1892648774795798,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00015393159609040552,
      "loss": 0.0205,
      "step": 6833
    },
    {
      "epoch": 3.1897316219369896,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00015391923891876848,
      "loss": 0.029,
      "step": 6834
    },
    {
      "epoch": 3.190198366394399,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015390688058619038,
      "loss": 0.037,
      "step": 6835
    },
    {
      "epoch": 3.1906651108518087,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015389452109293737,
      "loss": 0.0297,
      "step": 6836
    },
    {
      "epoch": 3.191131855309218,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015388216043927546,
      "loss": 0.038,
      "step": 6837
    },
    {
      "epoch": 3.191598599766628,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015386979862547088,
      "loss": 0.0257,
      "step": 6838
    },
    {
      "epoch": 3.192065344224037,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001538574356517898,
      "loss": 0.0301,
      "step": 6839
    },
    {
      "epoch": 3.192532088681447,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00015384507151849832,
      "loss": 0.0238,
      "step": 6840
    },
    {
      "epoch": 3.1929988331388564,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015383270622586274,
      "loss": 0.0422,
      "step": 6841
    },
    {
      "epoch": 3.193465577596266,
      "grad_norm": 0.484375,
      "learning_rate": 0.00015382033977414923,
      "loss": 0.0194,
      "step": 6842
    },
    {
      "epoch": 3.1939323220536755,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015380797216362412,
      "loss": 0.0233,
      "step": 6843
    },
    {
      "epoch": 3.1943990665110853,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001537956033945537,
      "loss": 0.029,
      "step": 6844
    },
    {
      "epoch": 3.1948658109684946,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001537832334672042,
      "loss": 0.0354,
      "step": 6845
    },
    {
      "epoch": 3.1953325554259044,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015377086238184203,
      "loss": 0.0281,
      "step": 6846
    },
    {
      "epoch": 3.195799299883314,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015375849013873354,
      "loss": 0.0361,
      "step": 6847
    },
    {
      "epoch": 3.1962660443407236,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001537461167381451,
      "loss": 0.0397,
      "step": 6848
    },
    {
      "epoch": 3.196732788798133,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015373374218034314,
      "loss": 0.0293,
      "step": 6849
    },
    {
      "epoch": 3.1971995332555427,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001537213664655941,
      "loss": 0.041,
      "step": 6850
    },
    {
      "epoch": 3.197666277712952,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001537089895941644,
      "loss": 0.0336,
      "step": 6851
    },
    {
      "epoch": 3.198133022170362,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00015369661156632063,
      "loss": 0.0281,
      "step": 6852
    },
    {
      "epoch": 3.1985997666277712,
      "grad_norm": 0.92578125,
      "learning_rate": 0.0001536842323823292,
      "loss": 0.0322,
      "step": 6853
    },
    {
      "epoch": 3.199066511085181,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015367185204245666,
      "loss": 0.0276,
      "step": 6854
    },
    {
      "epoch": 3.1995332555425904,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001536594705469696,
      "loss": 0.0249,
      "step": 6855
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015364708789613465,
      "loss": 0.0242,
      "step": 6856
    },
    {
      "epoch": 3.2004667444574095,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015363470409021833,
      "loss": 0.0406,
      "step": 6857
    },
    {
      "epoch": 3.2009334889148193,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015362231912948734,
      "loss": 0.0315,
      "step": 6858
    },
    {
      "epoch": 3.2014002333722287,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001536099330142083,
      "loss": 0.0415,
      "step": 6859
    },
    {
      "epoch": 3.2018669778296385,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015359754574464796,
      "loss": 0.0292,
      "step": 6860
    },
    {
      "epoch": 3.202333722287048,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015358515732107297,
      "loss": 0.0357,
      "step": 6861
    },
    {
      "epoch": 3.202800466744457,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001535727677437501,
      "loss": 0.0252,
      "step": 6862
    },
    {
      "epoch": 3.203267211201867,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015356037701294608,
      "loss": 0.0335,
      "step": 6863
    },
    {
      "epoch": 3.2037339556592768,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001535479851289277,
      "loss": 0.0319,
      "step": 6864
    },
    {
      "epoch": 3.204200700116686,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015353559209196183,
      "loss": 0.028,
      "step": 6865
    },
    {
      "epoch": 3.2046674445740955,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015352319790231525,
      "loss": 0.0204,
      "step": 6866
    },
    {
      "epoch": 3.2051341890315053,
      "grad_norm": 0.625,
      "learning_rate": 0.00015351080256025484,
      "loss": 0.0162,
      "step": 6867
    },
    {
      "epoch": 3.2056009334889146,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015349840606604747,
      "loss": 0.0387,
      "step": 6868
    },
    {
      "epoch": 3.2060676779463244,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015348600841996003,
      "loss": 0.0337,
      "step": 6869
    },
    {
      "epoch": 3.2065344224037338,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015347360962225955,
      "loss": 0.0235,
      "step": 6870
    },
    {
      "epoch": 3.2070011668611436,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015346120967321287,
      "loss": 0.0335,
      "step": 6871
    },
    {
      "epoch": 3.207467911318553,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015344880857308705,
      "loss": 0.0297,
      "step": 6872
    },
    {
      "epoch": 3.2079346557759627,
      "grad_norm": 0.86328125,
      "learning_rate": 0.0001534364063221491,
      "loss": 0.0371,
      "step": 6873
    },
    {
      "epoch": 3.208401400233372,
      "grad_norm": 0.54296875,
      "learning_rate": 0.000153424002920666,
      "loss": 0.0158,
      "step": 6874
    },
    {
      "epoch": 3.208868144690782,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015341159836890485,
      "loss": 0.0283,
      "step": 6875
    },
    {
      "epoch": 3.209334889148191,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001533991926671327,
      "loss": 0.0248,
      "step": 6876
    },
    {
      "epoch": 3.209801633605601,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015338678581561674,
      "loss": 0.0312,
      "step": 6877
    },
    {
      "epoch": 3.2102683780630104,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015337437781462405,
      "loss": 0.0212,
      "step": 6878
    },
    {
      "epoch": 3.21073512252042,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015336196866442176,
      "loss": 0.035,
      "step": 6879
    },
    {
      "epoch": 3.2112018669778295,
      "grad_norm": 0.640625,
      "learning_rate": 0.00015334955836527708,
      "loss": 0.0265,
      "step": 6880
    },
    {
      "epoch": 3.2116686114352393,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00015333714691745723,
      "loss": 0.0396,
      "step": 6881
    },
    {
      "epoch": 3.2121353558926486,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001533247343212294,
      "loss": 0.0231,
      "step": 6882
    },
    {
      "epoch": 3.2126021003500584,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015331232057686086,
      "loss": 0.0342,
      "step": 6883
    },
    {
      "epoch": 3.213068844807468,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00015329990568461896,
      "loss": 0.0318,
      "step": 6884
    },
    {
      "epoch": 3.2135355892648776,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00015328748964477093,
      "loss": 0.0194,
      "step": 6885
    },
    {
      "epoch": 3.214002333722287,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015327507245758412,
      "loss": 0.0247,
      "step": 6886
    },
    {
      "epoch": 3.2144690781796967,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00015326265412332588,
      "loss": 0.042,
      "step": 6887
    },
    {
      "epoch": 3.214935822637106,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015325023464226364,
      "loss": 0.0366,
      "step": 6888
    },
    {
      "epoch": 3.215402567094516,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015323781401466473,
      "loss": 0.0373,
      "step": 6889
    },
    {
      "epoch": 3.2158693115519252,
      "grad_norm": 0.625,
      "learning_rate": 0.00015322539224079662,
      "loss": 0.0317,
      "step": 6890
    },
    {
      "epoch": 3.216336056009335,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015321296932092676,
      "loss": 0.0341,
      "step": 6891
    },
    {
      "epoch": 3.2168028004667444,
      "grad_norm": 0.9140625,
      "learning_rate": 0.00015320054525532265,
      "loss": 0.0417,
      "step": 6892
    },
    {
      "epoch": 3.217269544924154,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00015318812004425176,
      "loss": 0.0361,
      "step": 6893
    },
    {
      "epoch": 3.2177362893815635,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00015317569368798162,
      "loss": 0.0278,
      "step": 6894
    },
    {
      "epoch": 3.2182030338389733,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001531632661867798,
      "loss": 0.027,
      "step": 6895
    },
    {
      "epoch": 3.2186697782963827,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00015315083754091388,
      "loss": 0.0283,
      "step": 6896
    },
    {
      "epoch": 3.2191365227537925,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015313840775065148,
      "loss": 0.0295,
      "step": 6897
    },
    {
      "epoch": 3.219603267211202,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015312597681626014,
      "loss": 0.0293,
      "step": 6898
    },
    {
      "epoch": 3.2200700116686116,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015311354473800763,
      "loss": 0.0386,
      "step": 6899
    },
    {
      "epoch": 3.220536756126021,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015310111151616156,
      "loss": 0.0213,
      "step": 6900
    },
    {
      "epoch": 3.2210035005834308,
      "grad_norm": 0.75,
      "learning_rate": 0.00015308867715098961,
      "loss": 0.0274,
      "step": 6901
    },
    {
      "epoch": 3.22147024504084,
      "grad_norm": 1.015625,
      "learning_rate": 0.00015307624164275958,
      "loss": 0.0396,
      "step": 6902
    },
    {
      "epoch": 3.22193698949825,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015306380499173917,
      "loss": 0.0266,
      "step": 6903
    },
    {
      "epoch": 3.2224037339556593,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015305136719819616,
      "loss": 0.0273,
      "step": 6904
    },
    {
      "epoch": 3.2228704784130686,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015303892826239835,
      "loss": 0.0277,
      "step": 6905
    },
    {
      "epoch": 3.2233372228704784,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001530264881846136,
      "loss": 0.0257,
      "step": 6906
    },
    {
      "epoch": 3.223803967327888,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015301404696510968,
      "loss": 0.0344,
      "step": 6907
    },
    {
      "epoch": 3.2242707117852976,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00015300160460415452,
      "loss": 0.0362,
      "step": 6908
    },
    {
      "epoch": 3.224737456242707,
      "grad_norm": 0.703125,
      "learning_rate": 0.000152989161102016,
      "loss": 0.0317,
      "step": 6909
    },
    {
      "epoch": 3.2252042007001167,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00015297671645896204,
      "loss": 0.0481,
      "step": 6910
    },
    {
      "epoch": 3.225670945157526,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00015296427067526061,
      "loss": 0.0233,
      "step": 6911
    },
    {
      "epoch": 3.226137689614936,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001529518237511797,
      "loss": 0.046,
      "step": 6912
    },
    {
      "epoch": 3.226604434072345,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015293937568698724,
      "loss": 0.0379,
      "step": 6913
    },
    {
      "epoch": 3.227071178529755,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001529269264829513,
      "loss": 0.0376,
      "step": 6914
    },
    {
      "epoch": 3.2275379229871644,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00015291447613933988,
      "loss": 0.0304,
      "step": 6915
    },
    {
      "epoch": 3.228004667444574,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001529020246564211,
      "loss": 0.0286,
      "step": 6916
    },
    {
      "epoch": 3.2284714119019835,
      "grad_norm": 0.62109375,
      "learning_rate": 0.000152889572034463,
      "loss": 0.0346,
      "step": 6917
    },
    {
      "epoch": 3.2289381563593933,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015287711827373376,
      "loss": 0.0358,
      "step": 6918
    },
    {
      "epoch": 3.2294049008168026,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015286466337450152,
      "loss": 0.0332,
      "step": 6919
    },
    {
      "epoch": 3.2298716452742124,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001528522073370344,
      "loss": 0.0418,
      "step": 6920
    },
    {
      "epoch": 3.230338389731622,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001528397501616006,
      "loss": 0.0257,
      "step": 6921
    },
    {
      "epoch": 3.2308051341890316,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015282729184846837,
      "loss": 0.0307,
      "step": 6922
    },
    {
      "epoch": 3.231271878646441,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015281483239790595,
      "loss": 0.0431,
      "step": 6923
    },
    {
      "epoch": 3.2317386231038507,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015280237181018155,
      "loss": 0.0453,
      "step": 6924
    },
    {
      "epoch": 3.23220536756126,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015278991008556351,
      "loss": 0.0278,
      "step": 6925
    },
    {
      "epoch": 3.23267211201867,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015277744722432018,
      "loss": 0.0315,
      "step": 6926
    },
    {
      "epoch": 3.2331388564760792,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015276498322671983,
      "loss": 0.0278,
      "step": 6927
    },
    {
      "epoch": 3.233605600933489,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015275251809303083,
      "loss": 0.015,
      "step": 6928
    },
    {
      "epoch": 3.2340723453908984,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00015274005182352158,
      "loss": 0.0215,
      "step": 6929
    },
    {
      "epoch": 3.234539089848308,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015272758441846053,
      "loss": 0.0322,
      "step": 6930
    },
    {
      "epoch": 3.2350058343057175,
      "grad_norm": 1.2265625,
      "learning_rate": 0.00015271511587811604,
      "loss": 0.0346,
      "step": 6931
    },
    {
      "epoch": 3.2354725787631273,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015270264620275664,
      "loss": 0.0522,
      "step": 6932
    },
    {
      "epoch": 3.2359393232205367,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001526901753926508,
      "loss": 0.0267,
      "step": 6933
    },
    {
      "epoch": 3.2364060676779465,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015267770344806704,
      "loss": 0.0327,
      "step": 6934
    },
    {
      "epoch": 3.236872812135356,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015266523036927386,
      "loss": 0.0286,
      "step": 6935
    },
    {
      "epoch": 3.2373395565927656,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001526527561565398,
      "loss": 0.0189,
      "step": 6936
    },
    {
      "epoch": 3.237806301050175,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015264028081013353,
      "loss": 0.0333,
      "step": 6937
    },
    {
      "epoch": 3.2382730455075848,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015262780433032358,
      "loss": 0.027,
      "step": 6938
    },
    {
      "epoch": 3.238739789964994,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00015261532671737862,
      "loss": 0.0246,
      "step": 6939
    },
    {
      "epoch": 3.239206534422404,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001526028479715673,
      "loss": 0.037,
      "step": 6940
    },
    {
      "epoch": 3.2396732788798133,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001525903680931583,
      "loss": 0.0217,
      "step": 6941
    },
    {
      "epoch": 3.240140023337223,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015257788708242034,
      "loss": 0.0461,
      "step": 6942
    },
    {
      "epoch": 3.2406067677946324,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015256540493962213,
      "loss": 0.0309,
      "step": 6943
    },
    {
      "epoch": 3.241073512252042,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001525529216650324,
      "loss": 0.0403,
      "step": 6944
    },
    {
      "epoch": 3.2415402567094516,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015254043725892,
      "loss": 0.0324,
      "step": 6945
    },
    {
      "epoch": 3.2420070011668614,
      "grad_norm": 0.734375,
      "learning_rate": 0.00015252795172155366,
      "loss": 0.0301,
      "step": 6946
    },
    {
      "epoch": 3.2424737456242707,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015251546505320223,
      "loss": 0.0283,
      "step": 6947
    },
    {
      "epoch": 3.24294049008168,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015250297725413457,
      "loss": 0.0245,
      "step": 6948
    },
    {
      "epoch": 3.24340723453909,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015249048832461962,
      "loss": 0.0313,
      "step": 6949
    },
    {
      "epoch": 3.2438739789964997,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015247799826492614,
      "loss": 0.0339,
      "step": 6950
    },
    {
      "epoch": 3.244340723453909,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015246550707532318,
      "loss": 0.0282,
      "step": 6951
    },
    {
      "epoch": 3.2448074679113184,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015245301475607962,
      "loss": 0.04,
      "step": 6952
    },
    {
      "epoch": 3.245274212368728,
      "grad_norm": 0.875,
      "learning_rate": 0.00015244052130746448,
      "loss": 0.0445,
      "step": 6953
    },
    {
      "epoch": 3.2457409568261375,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001524280267297467,
      "loss": 0.0306,
      "step": 6954
    },
    {
      "epoch": 3.2462077012835473,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015241553102319534,
      "loss": 0.0358,
      "step": 6955
    },
    {
      "epoch": 3.2466744457409566,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015240303418807946,
      "loss": 0.0289,
      "step": 6956
    },
    {
      "epoch": 3.2471411901983664,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015239053622466807,
      "loss": 0.0464,
      "step": 6957
    },
    {
      "epoch": 3.247607934655776,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015237803713323035,
      "loss": 0.0442,
      "step": 6958
    },
    {
      "epoch": 3.2480746791131856,
      "grad_norm": 0.75,
      "learning_rate": 0.00015236553691403538,
      "loss": 0.0414,
      "step": 6959
    },
    {
      "epoch": 3.248541423570595,
      "grad_norm": 1.0625,
      "learning_rate": 0.00015235303556735227,
      "loss": 0.0383,
      "step": 6960
    },
    {
      "epoch": 3.2490081680280047,
      "grad_norm": 0.8125,
      "learning_rate": 0.00015234053309345024,
      "loss": 0.0406,
      "step": 6961
    },
    {
      "epoch": 3.249474912485414,
      "grad_norm": 1.0703125,
      "learning_rate": 0.00015232802949259844,
      "loss": 0.0595,
      "step": 6962
    },
    {
      "epoch": 3.249941656942824,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015231552476506613,
      "loss": 0.0276,
      "step": 6963
    },
    {
      "epoch": 3.2504084014002332,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001523030189111225,
      "loss": 0.0523,
      "step": 6964
    },
    {
      "epoch": 3.250875145857643,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015229051193103686,
      "loss": 0.0475,
      "step": 6965
    },
    {
      "epoch": 3.2513418903150524,
      "grad_norm": 0.578125,
      "learning_rate": 0.00015227800382507849,
      "loss": 0.0348,
      "step": 6966
    },
    {
      "epoch": 3.251808634772462,
      "grad_norm": 0.875,
      "learning_rate": 0.00015226549459351665,
      "loss": 0.0337,
      "step": 6967
    },
    {
      "epoch": 3.2522753792298715,
      "grad_norm": 1.015625,
      "learning_rate": 0.00015225298423662074,
      "loss": 0.0296,
      "step": 6968
    },
    {
      "epoch": 3.2522753792298715,
      "eval_loss": 1.7366282939910889,
      "eval_runtime": 56.0103,
      "eval_samples_per_second": 32.208,
      "eval_steps_per_second": 4.035,
      "step": 6968
    },
    {
      "epoch": 3.2527421236872813,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00015224047275466015,
      "loss": 0.0262,
      "step": 6969
    },
    {
      "epoch": 3.2532088681446907,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00015222796014790417,
      "loss": 0.0315,
      "step": 6970
    },
    {
      "epoch": 3.2536756126021005,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015221544641662226,
      "loss": 0.0311,
      "step": 6971
    },
    {
      "epoch": 3.25414235705951,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00015220293156108388,
      "loss": 0.0401,
      "step": 6972
    },
    {
      "epoch": 3.2546091015169196,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00015219041558155847,
      "loss": 0.0163,
      "step": 6973
    },
    {
      "epoch": 3.255075845974329,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015217789847831546,
      "loss": 0.0422,
      "step": 6974
    },
    {
      "epoch": 3.2555425904317388,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00015216538025162444,
      "loss": 0.021,
      "step": 6975
    },
    {
      "epoch": 3.256009334889148,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001521528609017549,
      "loss": 0.0266,
      "step": 6976
    },
    {
      "epoch": 3.256476079346558,
      "grad_norm": 0.90625,
      "learning_rate": 0.0001521403404289764,
      "loss": 0.0387,
      "step": 6977
    },
    {
      "epoch": 3.2569428238039673,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015212781883355852,
      "loss": 0.0241,
      "step": 6978
    },
    {
      "epoch": 3.257409568261377,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015211529611577088,
      "loss": 0.0251,
      "step": 6979
    },
    {
      "epoch": 3.2578763127187864,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00015210277227588304,
      "loss": 0.0424,
      "step": 6980
    },
    {
      "epoch": 3.258343057176196,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015209024731416474,
      "loss": 0.0256,
      "step": 6981
    },
    {
      "epoch": 3.2588098016336056,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00015207772123088562,
      "loss": 0.0347,
      "step": 6982
    },
    {
      "epoch": 3.2592765460910154,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015206519402631536,
      "loss": 0.0416,
      "step": 6983
    },
    {
      "epoch": 3.2597432905484247,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00015205266570072375,
      "loss": 0.0308,
      "step": 6984
    },
    {
      "epoch": 3.2602100350058345,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00015204013625438046,
      "loss": 0.0311,
      "step": 6985
    },
    {
      "epoch": 3.260676779463244,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001520276056875553,
      "loss": 0.0338,
      "step": 6986
    },
    {
      "epoch": 3.261143523920653,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015201507400051803,
      "loss": 0.0314,
      "step": 6987
    },
    {
      "epoch": 3.261610268378063,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001520025411935386,
      "loss": 0.0347,
      "step": 6988
    },
    {
      "epoch": 3.262077012835473,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015199000726688668,
      "loss": 0.0276,
      "step": 6989
    },
    {
      "epoch": 3.262543757292882,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00015197747222083223,
      "loss": 0.0302,
      "step": 6990
    },
    {
      "epoch": 3.2630105017502915,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00015196493605564514,
      "loss": 0.0159,
      "step": 6991
    },
    {
      "epoch": 3.2634772462077013,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00015195239877159535,
      "loss": 0.0249,
      "step": 6992
    },
    {
      "epoch": 3.263943990665111,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015193986036895276,
      "loss": 0.0362,
      "step": 6993
    },
    {
      "epoch": 3.2644107351225204,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001519273208479873,
      "loss": 0.0237,
      "step": 6994
    },
    {
      "epoch": 3.26487747957993,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00015191478020896906,
      "loss": 0.0255,
      "step": 6995
    },
    {
      "epoch": 3.2653442240373396,
      "grad_norm": 1.0625,
      "learning_rate": 0.000151902238452168,
      "loss": 0.036,
      "step": 6996
    },
    {
      "epoch": 3.265810968494749,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001518896955778541,
      "loss": 0.023,
      "step": 6997
    },
    {
      "epoch": 3.2662777129521587,
      "grad_norm": 0.75,
      "learning_rate": 0.00015187715158629757,
      "loss": 0.0209,
      "step": 6998
    },
    {
      "epoch": 3.266744457409568,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015186460647776835,
      "loss": 0.0379,
      "step": 6999
    },
    {
      "epoch": 3.267211201866978,
      "grad_norm": 0.98046875,
      "learning_rate": 0.0001518520602525366,
      "loss": 0.0356,
      "step": 7000
    },
    {
      "epoch": 3.2676779463243872,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015183951291087251,
      "loss": 0.0324,
      "step": 7001
    },
    {
      "epoch": 3.268144690781797,
      "grad_norm": 0.875,
      "learning_rate": 0.00015182696445304616,
      "loss": 0.0398,
      "step": 7002
    },
    {
      "epoch": 3.2686114352392064,
      "grad_norm": 1.046875,
      "learning_rate": 0.00015181441487932776,
      "loss": 0.045,
      "step": 7003
    },
    {
      "epoch": 3.269078179696616,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00015180186418998755,
      "loss": 0.0343,
      "step": 7004
    },
    {
      "epoch": 3.2695449241540255,
      "grad_norm": 0.6875,
      "learning_rate": 0.00015178931238529568,
      "loss": 0.0357,
      "step": 7005
    },
    {
      "epoch": 3.2700116686114353,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00015177675946552249,
      "loss": 0.0275,
      "step": 7006
    },
    {
      "epoch": 3.2704784130688447,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001517642054309382,
      "loss": 0.0335,
      "step": 7007
    },
    {
      "epoch": 3.2709451575262545,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015175165028181314,
      "loss": 0.0318,
      "step": 7008
    },
    {
      "epoch": 3.271411901983664,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015173909401841766,
      "loss": 0.0196,
      "step": 7009
    },
    {
      "epoch": 3.2718786464410736,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00015172653664102202,
      "loss": 0.0236,
      "step": 7010
    },
    {
      "epoch": 3.272345390898483,
      "grad_norm": 1.1171875,
      "learning_rate": 0.0001517139781498967,
      "loss": 0.0324,
      "step": 7011
    },
    {
      "epoch": 3.2728121353558928,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00015170141854531203,
      "loss": 0.0258,
      "step": 7012
    },
    {
      "epoch": 3.273278879813302,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00015168885782753844,
      "loss": 0.0381,
      "step": 7013
    },
    {
      "epoch": 3.273745624270712,
      "grad_norm": 0.796875,
      "learning_rate": 0.00015167629599684642,
      "loss": 0.0285,
      "step": 7014
    },
    {
      "epoch": 3.2742123687281213,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015166373305350642,
      "loss": 0.0387,
      "step": 7015
    },
    {
      "epoch": 3.274679113185531,
      "grad_norm": 0.94921875,
      "learning_rate": 0.0001516511689977889,
      "loss": 0.0371,
      "step": 7016
    },
    {
      "epoch": 3.2751458576429404,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001516386038299644,
      "loss": 0.0233,
      "step": 7017
    },
    {
      "epoch": 3.27561260210035,
      "grad_norm": 0.53125,
      "learning_rate": 0.00015162603755030353,
      "loss": 0.0198,
      "step": 7018
    },
    {
      "epoch": 3.2760793465577596,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015161347015907675,
      "loss": 0.0155,
      "step": 7019
    },
    {
      "epoch": 3.2765460910151694,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015160090165655468,
      "loss": 0.0354,
      "step": 7020
    },
    {
      "epoch": 3.2770128354725787,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00015158833204300794,
      "loss": 0.0443,
      "step": 7021
    },
    {
      "epoch": 3.2774795799299885,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015157576131870717,
      "loss": 0.0237,
      "step": 7022
    },
    {
      "epoch": 3.277946324387398,
      "grad_norm": 0.75,
      "learning_rate": 0.00015156318948392303,
      "loss": 0.035,
      "step": 7023
    },
    {
      "epoch": 3.2784130688448077,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015155061653892625,
      "loss": 0.0261,
      "step": 7024
    },
    {
      "epoch": 3.278879813302217,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001515380424839875,
      "loss": 0.03,
      "step": 7025
    },
    {
      "epoch": 3.2793465577596264,
      "grad_norm": 0.78125,
      "learning_rate": 0.00015152546731937746,
      "loss": 0.0318,
      "step": 7026
    },
    {
      "epoch": 3.279813302217036,
      "grad_norm": 0.73828125,
      "learning_rate": 0.000151512891045367,
      "loss": 0.0337,
      "step": 7027
    },
    {
      "epoch": 3.280280046674446,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015150031366222678,
      "loss": 0.0209,
      "step": 7028
    },
    {
      "epoch": 3.2807467911318553,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001514877351702277,
      "loss": 0.0352,
      "step": 7029
    },
    {
      "epoch": 3.2812135355892647,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015147515556964057,
      "loss": 0.0307,
      "step": 7030
    },
    {
      "epoch": 3.2816802800466744,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001514625748607362,
      "loss": 0.0341,
      "step": 7031
    },
    {
      "epoch": 3.2821470245040842,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001514499930437855,
      "loss": 0.0353,
      "step": 7032
    },
    {
      "epoch": 3.2826137689614936,
      "grad_norm": 0.609375,
      "learning_rate": 0.00015143741011905938,
      "loss": 0.0248,
      "step": 7033
    },
    {
      "epoch": 3.283080513418903,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015142482608682874,
      "loss": 0.0242,
      "step": 7034
    },
    {
      "epoch": 3.2835472578763127,
      "grad_norm": 0.96484375,
      "learning_rate": 0.00015141224094736454,
      "loss": 0.0451,
      "step": 7035
    },
    {
      "epoch": 3.284014002333722,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015139965470093773,
      "loss": 0.0267,
      "step": 7036
    },
    {
      "epoch": 3.284480746791132,
      "grad_norm": 0.953125,
      "learning_rate": 0.00015138706734781936,
      "loss": 0.032,
      "step": 7037
    },
    {
      "epoch": 3.2849474912485412,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001513744788882804,
      "loss": 0.0481,
      "step": 7038
    },
    {
      "epoch": 3.285414235705951,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001513618893225919,
      "loss": 0.0302,
      "step": 7039
    },
    {
      "epoch": 3.2858809801633604,
      "grad_norm": 0.703125,
      "learning_rate": 0.00015134929865102495,
      "loss": 0.0214,
      "step": 7040
    },
    {
      "epoch": 3.28634772462077,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001513367068738506,
      "loss": 0.0295,
      "step": 7041
    },
    {
      "epoch": 3.2868144690781795,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00015132411399133998,
      "loss": 0.0301,
      "step": 7042
    },
    {
      "epoch": 3.2872812135355893,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001513115200037643,
      "loss": 0.0339,
      "step": 7043
    },
    {
      "epoch": 3.2877479579929987,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001512989249113946,
      "loss": 0.0479,
      "step": 7044
    },
    {
      "epoch": 3.2882147024504085,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001512863287145022,
      "loss": 0.0164,
      "step": 7045
    },
    {
      "epoch": 3.288681446907818,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015127373141335818,
      "loss": 0.0247,
      "step": 7046
    },
    {
      "epoch": 3.2891481913652276,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015126113300823385,
      "loss": 0.0256,
      "step": 7047
    },
    {
      "epoch": 3.289614935822637,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00015124853349940049,
      "loss": 0.0409,
      "step": 7048
    },
    {
      "epoch": 3.2900816802800468,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001512359328871293,
      "loss": 0.0415,
      "step": 7049
    },
    {
      "epoch": 3.290548424737456,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00015122333117169166,
      "loss": 0.0287,
      "step": 7050
    },
    {
      "epoch": 3.291015169194866,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00015121072835335884,
      "loss": 0.0338,
      "step": 7051
    },
    {
      "epoch": 3.2914819136522753,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001511981244324022,
      "loss": 0.0326,
      "step": 7052
    },
    {
      "epoch": 3.291948658109685,
      "grad_norm": 0.5625,
      "learning_rate": 0.00015118551940909315,
      "loss": 0.022,
      "step": 7053
    },
    {
      "epoch": 3.2924154025670944,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015117291328370312,
      "loss": 0.0303,
      "step": 7054
    },
    {
      "epoch": 3.292882147024504,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015116030605650345,
      "loss": 0.0399,
      "step": 7055
    },
    {
      "epoch": 3.2933488914819136,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015114769772776561,
      "loss": 0.0422,
      "step": 7056
    },
    {
      "epoch": 3.2938156359393234,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015113508829776113,
      "loss": 0.0304,
      "step": 7057
    },
    {
      "epoch": 3.2942823803967327,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00015112247776676144,
      "loss": 0.0242,
      "step": 7058
    },
    {
      "epoch": 3.2947491248541425,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001511098661350381,
      "loss": 0.0304,
      "step": 7059
    },
    {
      "epoch": 3.295215869311552,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001510972534028626,
      "loss": 0.0368,
      "step": 7060
    },
    {
      "epoch": 3.2956826137689617,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015108463957050656,
      "loss": 0.0419,
      "step": 7061
    },
    {
      "epoch": 3.296149358226371,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015107202463824154,
      "loss": 0.0333,
      "step": 7062
    },
    {
      "epoch": 3.296616102683781,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015105940860633919,
      "loss": 0.0328,
      "step": 7063
    },
    {
      "epoch": 3.29708284714119,
      "grad_norm": 0.84375,
      "learning_rate": 0.00015104679147507111,
      "loss": 0.0358,
      "step": 7064
    },
    {
      "epoch": 3.2975495915986,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00015103417324470897,
      "loss": 0.0395,
      "step": 7065
    },
    {
      "epoch": 3.2980163360560093,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00015102155391552447,
      "loss": 0.0316,
      "step": 7066
    },
    {
      "epoch": 3.298483080513419,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001510089334877893,
      "loss": 0.0259,
      "step": 7067
    },
    {
      "epoch": 3.2989498249708284,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00015099631196177518,
      "loss": 0.0347,
      "step": 7068
    },
    {
      "epoch": 3.299416569428238,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00015098368933775386,
      "loss": 0.021,
      "step": 7069
    },
    {
      "epoch": 3.2998833138856476,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001509710656159972,
      "loss": 0.0288,
      "step": 7070
    },
    {
      "epoch": 3.3003500583430574,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001509584407967769,
      "loss": 0.025,
      "step": 7071
    },
    {
      "epoch": 3.3008168028004667,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00015094581488036486,
      "loss": 0.0276,
      "step": 7072
    },
    {
      "epoch": 3.301283547257876,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001509331878670329,
      "loss": 0.0225,
      "step": 7073
    },
    {
      "epoch": 3.301750291715286,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00015092055975705288,
      "loss": 0.0215,
      "step": 7074
    },
    {
      "epoch": 3.3022170361726957,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00015090793055069668,
      "loss": 0.0332,
      "step": 7075
    },
    {
      "epoch": 3.302683780630105,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015089530024823628,
      "loss": 0.025,
      "step": 7076
    },
    {
      "epoch": 3.3031505250875144,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001508826688499436,
      "loss": 0.0323,
      "step": 7077
    },
    {
      "epoch": 3.303617269544924,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001508700363560906,
      "loss": 0.0316,
      "step": 7078
    },
    {
      "epoch": 3.3040840140023335,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00015085740276694928,
      "loss": 0.036,
      "step": 7079
    },
    {
      "epoch": 3.3045507584597433,
      "grad_norm": 0.671875,
      "learning_rate": 0.00015084476808279167,
      "loss": 0.0261,
      "step": 7080
    },
    {
      "epoch": 3.3050175029171527,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00015083213230388976,
      "loss": 0.0391,
      "step": 7081
    },
    {
      "epoch": 3.3054842473745625,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015081949543051566,
      "loss": 0.0501,
      "step": 7082
    },
    {
      "epoch": 3.305950991831972,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00015080685746294144,
      "loss": 0.0271,
      "step": 7083
    },
    {
      "epoch": 3.3064177362893816,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001507942184014392,
      "loss": 0.0306,
      "step": 7084
    },
    {
      "epoch": 3.306884480746791,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00015078157824628105,
      "loss": 0.0285,
      "step": 7085
    },
    {
      "epoch": 3.3073512252042008,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00015076893699773924,
      "loss": 0.0465,
      "step": 7086
    },
    {
      "epoch": 3.30781796966161,
      "grad_norm": 0.65625,
      "learning_rate": 0.00015075629465608582,
      "loss": 0.0242,
      "step": 7087
    },
    {
      "epoch": 3.30828471411902,
      "grad_norm": 0.71875,
      "learning_rate": 0.00015074365122159312,
      "loss": 0.0315,
      "step": 7088
    },
    {
      "epoch": 3.3087514585764293,
      "grad_norm": 1.09375,
      "learning_rate": 0.00015073100669453328,
      "loss": 0.0468,
      "step": 7089
    },
    {
      "epoch": 3.309218203033839,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001507183610751786,
      "loss": 0.0313,
      "step": 7090
    },
    {
      "epoch": 3.3096849474912484,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015070571436380127,
      "loss": 0.0272,
      "step": 7091
    },
    {
      "epoch": 3.310151691948658,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001506930665606737,
      "loss": 0.0496,
      "step": 7092
    },
    {
      "epoch": 3.3106184364060676,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00015068041766606817,
      "loss": 0.0423,
      "step": 7093
    },
    {
      "epoch": 3.3110851808634774,
      "grad_norm": 0.796875,
      "learning_rate": 0.000150667767680257,
      "loss": 0.0338,
      "step": 7094
    },
    {
      "epoch": 3.3115519253208867,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00015065511660351263,
      "loss": 0.0336,
      "step": 7095
    },
    {
      "epoch": 3.3120186697782965,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015064246443610731,
      "loss": 0.0401,
      "step": 7096
    },
    {
      "epoch": 3.312485414235706,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001506298111783136,
      "loss": 0.0361,
      "step": 7097
    },
    {
      "epoch": 3.3129521586931157,
      "grad_norm": 0.59375,
      "learning_rate": 0.00015061715683040383,
      "loss": 0.0283,
      "step": 7098
    },
    {
      "epoch": 3.313418903150525,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00015060450139265055,
      "loss": 0.0341,
      "step": 7099
    },
    {
      "epoch": 3.313885647607935,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00015059184486532621,
      "loss": 0.0276,
      "step": 7100
    },
    {
      "epoch": 3.314352392065344,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00015057918724870327,
      "loss": 0.0343,
      "step": 7101
    },
    {
      "epoch": 3.314819136522754,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00015056652854305438,
      "loss": 0.0361,
      "step": 7102
    },
    {
      "epoch": 3.3152858809801633,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00015055386874865198,
      "loss": 0.0297,
      "step": 7103
    },
    {
      "epoch": 3.315752625437573,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001505412078657687,
      "loss": 0.036,
      "step": 7104
    },
    {
      "epoch": 3.3162193698949824,
      "grad_norm": 1.171875,
      "learning_rate": 0.00015052854589467715,
      "loss": 0.0428,
      "step": 7105
    },
    {
      "epoch": 3.3166861143523922,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00015051588283564998,
      "loss": 0.0469,
      "step": 7106
    },
    {
      "epoch": 3.3171528588098016,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00015050321868895973,
      "loss": 0.0359,
      "step": 7107
    },
    {
      "epoch": 3.3176196032672114,
      "grad_norm": 0.828125,
      "learning_rate": 0.00015049055345487918,
      "loss": 0.0471,
      "step": 7108
    },
    {
      "epoch": 3.3180863477246207,
      "grad_norm": 0.578125,
      "learning_rate": 0.000150477887133681,
      "loss": 0.0212,
      "step": 7109
    },
    {
      "epoch": 3.3185530921820305,
      "grad_norm": 0.859375,
      "learning_rate": 0.00015046521972563793,
      "loss": 0.0431,
      "step": 7110
    },
    {
      "epoch": 3.31901983663944,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00015045255123102264,
      "loss": 0.0301,
      "step": 7111
    },
    {
      "epoch": 3.3194865810968492,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00015043988165010797,
      "loss": 0.023,
      "step": 7112
    },
    {
      "epoch": 3.319953325554259,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001504272109831667,
      "loss": 0.0261,
      "step": 7113
    },
    {
      "epoch": 3.320420070011669,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00015041453923047155,
      "loss": 0.0274,
      "step": 7114
    },
    {
      "epoch": 3.320886814469078,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001504018663922955,
      "loss": 0.0252,
      "step": 7115
    },
    {
      "epoch": 3.3213535589264875,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00015038919246891132,
      "loss": 0.0482,
      "step": 7116
    },
    {
      "epoch": 3.3218203033838973,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015037651746059195,
      "loss": 0.0515,
      "step": 7117
    },
    {
      "epoch": 3.322287047841307,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00015036384136761022,
      "loss": 0.0357,
      "step": 7118
    },
    {
      "epoch": 3.3227537922987165,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001503511641902392,
      "loss": 0.0317,
      "step": 7119
    },
    {
      "epoch": 3.323220536756126,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00015033848592875163,
      "loss": 0.0267,
      "step": 7120
    },
    {
      "epoch": 3.3236872812135356,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015032580658342067,
      "loss": 0.0381,
      "step": 7121
    },
    {
      "epoch": 3.324154025670945,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00015031312615451924,
      "loss": 0.0208,
      "step": 7122
    },
    {
      "epoch": 3.3246207701283548,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001503004446423204,
      "loss": 0.0318,
      "step": 7123
    },
    {
      "epoch": 3.325087514585764,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00015028776204709714,
      "loss": 0.0295,
      "step": 7124
    },
    {
      "epoch": 3.325554259043174,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001502750783691226,
      "loss": 0.0282,
      "step": 7125
    },
    {
      "epoch": 3.3260210035005833,
      "grad_norm": 1.3203125,
      "learning_rate": 0.00015026239360866982,
      "loss": 0.0336,
      "step": 7126
    },
    {
      "epoch": 3.326487747957993,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015024970776601198,
      "loss": 0.0348,
      "step": 7127
    },
    {
      "epoch": 3.3269544924154024,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00015023702084142215,
      "loss": 0.0277,
      "step": 7128
    },
    {
      "epoch": 3.327421236872812,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00015022433283517353,
      "loss": 0.0278,
      "step": 7129
    },
    {
      "epoch": 3.3278879813302216,
      "grad_norm": 1.0859375,
      "learning_rate": 0.00015021164374753925,
      "loss": 0.0514,
      "step": 7130
    },
    {
      "epoch": 3.3283547257876314,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001501989535787926,
      "loss": 0.0316,
      "step": 7131
    },
    {
      "epoch": 3.3288214702450407,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00015018626232920682,
      "loss": 0.033,
      "step": 7132
    },
    {
      "epoch": 3.3292882147024505,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001501735699990551,
      "loss": 0.0205,
      "step": 7133
    },
    {
      "epoch": 3.32975495915986,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00015016087658861077,
      "loss": 0.037,
      "step": 7134
    },
    {
      "epoch": 3.3302217036172697,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00015014818209814707,
      "loss": 0.0342,
      "step": 7135
    },
    {
      "epoch": 3.330688448074679,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001501354865279374,
      "loss": 0.0372,
      "step": 7136
    },
    {
      "epoch": 3.331155192532089,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00015012278987825507,
      "loss": 0.0317,
      "step": 7137
    },
    {
      "epoch": 3.331621936989498,
      "grad_norm": 0.90625,
      "learning_rate": 0.00015011009214937346,
      "loss": 0.0449,
      "step": 7138
    },
    {
      "epoch": 3.332088681446908,
      "grad_norm": 0.8671875,
      "learning_rate": 0.000150097393341566,
      "loss": 0.0375,
      "step": 7139
    },
    {
      "epoch": 3.3325554259043173,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00015008469345510604,
      "loss": 0.0256,
      "step": 7140
    },
    {
      "epoch": 3.333022170361727,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00015007199249026705,
      "loss": 0.0331,
      "step": 7141
    },
    {
      "epoch": 3.3334889148191365,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00015005929044732255,
      "loss": 0.0303,
      "step": 7142
    },
    {
      "epoch": 3.3339556592765462,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00015004658732654594,
      "loss": 0.0279,
      "step": 7143
    },
    {
      "epoch": 3.3344224037339556,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001500338831282108,
      "loss": 0.036,
      "step": 7144
    },
    {
      "epoch": 3.3348891481913654,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00015002117785259067,
      "loss": 0.0525,
      "step": 7145
    },
    {
      "epoch": 3.3353558926487747,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000150008471499959,
      "loss": 0.0474,
      "step": 7146
    },
    {
      "epoch": 3.3358226371061845,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014999576407058953,
      "loss": 0.0345,
      "step": 7147
    },
    {
      "epoch": 3.336289381563594,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014998305556475575,
      "loss": 0.0224,
      "step": 7148
    },
    {
      "epoch": 3.3367561260210037,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014997034598273137,
      "loss": 0.0256,
      "step": 7149
    },
    {
      "epoch": 3.337222870478413,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014995763532478994,
      "loss": 0.0325,
      "step": 7150
    },
    {
      "epoch": 3.3376896149358224,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014994492359120522,
      "loss": 0.0308,
      "step": 7151
    },
    {
      "epoch": 3.338156359393232,
      "grad_norm": 1.0625,
      "learning_rate": 0.0001499322107822509,
      "loss": 0.0308,
      "step": 7152
    },
    {
      "epoch": 3.338623103850642,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014991949689820063,
      "loss": 0.0328,
      "step": 7153
    },
    {
      "epoch": 3.3390898483080513,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014990678193932825,
      "loss": 0.0311,
      "step": 7154
    },
    {
      "epoch": 3.3395565927654607,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014989406590590747,
      "loss": 0.0299,
      "step": 7155
    },
    {
      "epoch": 3.3400233372228705,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014988134879821208,
      "loss": 0.0435,
      "step": 7156
    },
    {
      "epoch": 3.3404900816802803,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014986863061651592,
      "loss": 0.0333,
      "step": 7157
    },
    {
      "epoch": 3.3409568261376896,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00014985591136109283,
      "loss": 0.0269,
      "step": 7158
    },
    {
      "epoch": 3.341423570595099,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001498431910322166,
      "loss": 0.0375,
      "step": 7159
    },
    {
      "epoch": 3.3418903150525088,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001498304696301612,
      "loss": 0.0334,
      "step": 7160
    },
    {
      "epoch": 3.342357059509918,
      "grad_norm": 0.828125,
      "learning_rate": 0.0001498177471552005,
      "loss": 0.034,
      "step": 7161
    },
    {
      "epoch": 3.342823803967328,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014980502360760844,
      "loss": 0.0325,
      "step": 7162
    },
    {
      "epoch": 3.3432905484247373,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014979229898765895,
      "loss": 0.035,
      "step": 7163
    },
    {
      "epoch": 3.343757292882147,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000149779573295626,
      "loss": 0.0293,
      "step": 7164
    },
    {
      "epoch": 3.3442240373395564,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014976684653178363,
      "loss": 0.0422,
      "step": 7165
    },
    {
      "epoch": 3.344690781796966,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014975411869640586,
      "loss": 0.0319,
      "step": 7166
    },
    {
      "epoch": 3.3451575262543756,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014974138978976668,
      "loss": 0.0332,
      "step": 7167
    },
    {
      "epoch": 3.3456242707117854,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001497286598121402,
      "loss": 0.0261,
      "step": 7168
    },
    {
      "epoch": 3.3460910151691947,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014971592876380049,
      "loss": 0.0232,
      "step": 7169
    },
    {
      "epoch": 3.3465577596266045,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014970319664502166,
      "loss": 0.0311,
      "step": 7170
    },
    {
      "epoch": 3.347024504084014,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014969046345607786,
      "loss": 0.0272,
      "step": 7171
    },
    {
      "epoch": 3.3474912485414237,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014967772919724327,
      "loss": 0.0274,
      "step": 7172
    },
    {
      "epoch": 3.347957992998833,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014966499386879204,
      "loss": 0.0427,
      "step": 7173
    },
    {
      "epoch": 3.348424737456243,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014965225747099842,
      "loss": 0.0311,
      "step": 7174
    },
    {
      "epoch": 3.348891481913652,
      "grad_norm": 0.96875,
      "learning_rate": 0.00014963952000413653,
      "loss": 0.037,
      "step": 7175
    },
    {
      "epoch": 3.349358226371062,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014962678146848078,
      "loss": 0.033,
      "step": 7176
    },
    {
      "epoch": 3.3498249708284713,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014961404186430532,
      "loss": 0.0417,
      "step": 7177
    },
    {
      "epoch": 3.350291715285881,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014960130119188452,
      "loss": 0.041,
      "step": 7178
    },
    {
      "epoch": 3.3507584597432905,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00014958855945149262,
      "loss": 0.0283,
      "step": 7179
    },
    {
      "epoch": 3.3512252042007002,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014957581664340407,
      "loss": 0.0326,
      "step": 7180
    },
    {
      "epoch": 3.3516919486581096,
      "grad_norm": 0.98828125,
      "learning_rate": 0.00014956307276789313,
      "loss": 0.0472,
      "step": 7181
    },
    {
      "epoch": 3.3521586931155194,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014955032782523427,
      "loss": 0.033,
      "step": 7182
    },
    {
      "epoch": 3.3526254375729287,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001495375818157019,
      "loss": 0.0449,
      "step": 7183
    },
    {
      "epoch": 3.3530921820303385,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001495248347395704,
      "loss": 0.0349,
      "step": 7184
    },
    {
      "epoch": 3.353558926487748,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014951208659711427,
      "loss": 0.0295,
      "step": 7185
    },
    {
      "epoch": 3.3540256709451577,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014949933738860795,
      "loss": 0.035,
      "step": 7186
    },
    {
      "epoch": 3.354492415402567,
      "grad_norm": 0.9375,
      "learning_rate": 0.000149486587114326,
      "loss": 0.0436,
      "step": 7187
    },
    {
      "epoch": 3.354959159859977,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014947383577454295,
      "loss": 0.0416,
      "step": 7188
    },
    {
      "epoch": 3.355425904317386,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001494610833695333,
      "loss": 0.0351,
      "step": 7189
    },
    {
      "epoch": 3.355892648774796,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014944832989957166,
      "loss": 0.0395,
      "step": 7190
    },
    {
      "epoch": 3.3563593932322053,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001494355753649326,
      "loss": 0.0316,
      "step": 7191
    },
    {
      "epoch": 3.356826137689615,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014942281976589074,
      "loss": 0.0324,
      "step": 7192
    },
    {
      "epoch": 3.3572928821470245,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014941006310272074,
      "loss": 0.0355,
      "step": 7193
    },
    {
      "epoch": 3.357759626604434,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00014939730537569728,
      "loss": 0.0436,
      "step": 7194
    },
    {
      "epoch": 3.3582263710618436,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014938454658509502,
      "loss": 0.0291,
      "step": 7195
    },
    {
      "epoch": 3.3586931155192534,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001493717867311887,
      "loss": 0.0261,
      "step": 7196
    },
    {
      "epoch": 3.3591598599766628,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014935902581425297,
      "loss": 0.0295,
      "step": 7197
    },
    {
      "epoch": 3.359626604434072,
      "grad_norm": 0.578125,
      "learning_rate": 0.00014934626383456272,
      "loss": 0.0338,
      "step": 7198
    },
    {
      "epoch": 3.360093348891482,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014933350079239258,
      "loss": 0.0352,
      "step": 7199
    },
    {
      "epoch": 3.3605600933488917,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001493207366880175,
      "loss": 0.0391,
      "step": 7200
    },
    {
      "epoch": 3.361026837806301,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014930797152171223,
      "loss": 0.0455,
      "step": 7201
    },
    {
      "epoch": 3.3614935822637104,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001492952052937516,
      "loss": 0.0289,
      "step": 7202
    },
    {
      "epoch": 3.36196032672112,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001492824380044105,
      "loss": 0.0328,
      "step": 7203
    },
    {
      "epoch": 3.3624270711785296,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014926966965396384,
      "loss": 0.0247,
      "step": 7204
    },
    {
      "epoch": 3.3628938156359394,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014925690024268653,
      "loss": 0.0416,
      "step": 7205
    },
    {
      "epoch": 3.3633605600933487,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014924412977085352,
      "loss": 0.0416,
      "step": 7206
    },
    {
      "epoch": 3.3638273045507585,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014923135823873975,
      "loss": 0.0203,
      "step": 7207
    },
    {
      "epoch": 3.364294049008168,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001492185856466202,
      "loss": 0.0343,
      "step": 7208
    },
    {
      "epoch": 3.3647607934655777,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014920581199476992,
      "loss": 0.0339,
      "step": 7209
    },
    {
      "epoch": 3.365227537922987,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00014919303728346388,
      "loss": 0.0399,
      "step": 7210
    },
    {
      "epoch": 3.365694282380397,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014918026151297719,
      "loss": 0.038,
      "step": 7211
    },
    {
      "epoch": 3.366161026837806,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014916748468358487,
      "loss": 0.0406,
      "step": 7212
    },
    {
      "epoch": 3.366627771295216,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001491547067955621,
      "loss": 0.0268,
      "step": 7213
    },
    {
      "epoch": 3.3670945157526253,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014914192784918392,
      "loss": 0.0333,
      "step": 7214
    },
    {
      "epoch": 3.367561260210035,
      "grad_norm": 1.109375,
      "learning_rate": 0.00014912914784472555,
      "loss": 0.0474,
      "step": 7215
    },
    {
      "epoch": 3.3680280046674445,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014911636678246207,
      "loss": 0.0369,
      "step": 7216
    },
    {
      "epoch": 3.3684947491248542,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014910358466266877,
      "loss": 0.0247,
      "step": 7217
    },
    {
      "epoch": 3.3689614935822636,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014909080148562076,
      "loss": 0.0264,
      "step": 7218
    },
    {
      "epoch": 3.3694282380396734,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014907801725159333,
      "loss": 0.0411,
      "step": 7219
    },
    {
      "epoch": 3.3698949824970827,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014906523196086177,
      "loss": 0.0332,
      "step": 7220
    },
    {
      "epoch": 3.3703617269544925,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001490524456137013,
      "loss": 0.0194,
      "step": 7221
    },
    {
      "epoch": 3.370828471411902,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014903965821038724,
      "loss": 0.0257,
      "step": 7222
    },
    {
      "epoch": 3.3712952158693117,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014902686975119497,
      "loss": 0.028,
      "step": 7223
    },
    {
      "epoch": 3.371761960326721,
      "grad_norm": 0.94921875,
      "learning_rate": 0.00014901408023639977,
      "loss": 0.0419,
      "step": 7224
    },
    {
      "epoch": 3.372228704784131,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014900128966627705,
      "loss": 0.0321,
      "step": 7225
    },
    {
      "epoch": 3.37269544924154,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014898849804110216,
      "loss": 0.0304,
      "step": 7226
    },
    {
      "epoch": 3.37316219369895,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014897570536115062,
      "loss": 0.0533,
      "step": 7227
    },
    {
      "epoch": 3.3736289381563593,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014896291162669776,
      "loss": 0.0246,
      "step": 7228
    },
    {
      "epoch": 3.374095682613769,
      "grad_norm": 0.91015625,
      "learning_rate": 0.0001489501168380191,
      "loss": 0.0288,
      "step": 7229
    },
    {
      "epoch": 3.3745624270711785,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001489373209953901,
      "loss": 0.0303,
      "step": 7230
    },
    {
      "epoch": 3.3750291715285883,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001489245240990863,
      "loss": 0.0348,
      "step": 7231
    },
    {
      "epoch": 3.3754959159859976,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001489117261493832,
      "loss": 0.0286,
      "step": 7232
    },
    {
      "epoch": 3.3759626604434074,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014889892714655637,
      "loss": 0.0225,
      "step": 7233
    },
    {
      "epoch": 3.3764294049008168,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001488861270908814,
      "loss": 0.0322,
      "step": 7234
    },
    {
      "epoch": 3.3768961493582266,
      "grad_norm": 0.5625,
      "learning_rate": 0.00014887332598263384,
      "loss": 0.0206,
      "step": 7235
    },
    {
      "epoch": 3.377362893815636,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014886052382208936,
      "loss": 0.0356,
      "step": 7236
    },
    {
      "epoch": 3.3778296382730453,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014884772060952362,
      "loss": 0.0264,
      "step": 7237
    },
    {
      "epoch": 3.378296382730455,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001488349163452122,
      "loss": 0.0268,
      "step": 7238
    },
    {
      "epoch": 3.378763127187865,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001488221110294309,
      "loss": 0.0221,
      "step": 7239
    },
    {
      "epoch": 3.379229871645274,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014880930466245538,
      "loss": 0.0253,
      "step": 7240
    },
    {
      "epoch": 3.3796966161026836,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014879649724456134,
      "loss": 0.0284,
      "step": 7241
    },
    {
      "epoch": 3.3801633605600934,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001487836887760246,
      "loss": 0.0419,
      "step": 7242
    },
    {
      "epoch": 3.380630105017503,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00014877087925712092,
      "loss": 0.0329,
      "step": 7243
    },
    {
      "epoch": 3.3810968494749125,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014875806868812607,
      "loss": 0.0286,
      "step": 7244
    },
    {
      "epoch": 3.381563593932322,
      "grad_norm": 1.4140625,
      "learning_rate": 0.00014874525706931593,
      "loss": 0.0379,
      "step": 7245
    },
    {
      "epoch": 3.3820303383897317,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001487324444009663,
      "loss": 0.0315,
      "step": 7246
    },
    {
      "epoch": 3.382497082847141,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001487196306833531,
      "loss": 0.0195,
      "step": 7247
    },
    {
      "epoch": 3.382963827304551,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014870681591675215,
      "loss": 0.0263,
      "step": 7248
    },
    {
      "epoch": 3.38343057176196,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014869400010143947,
      "loss": 0.0264,
      "step": 7249
    },
    {
      "epoch": 3.38389731621937,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001486811832376909,
      "loss": 0.0385,
      "step": 7250
    },
    {
      "epoch": 3.3843640606767793,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014866836532578242,
      "loss": 0.0305,
      "step": 7251
    },
    {
      "epoch": 3.384830805134189,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014865554636599008,
      "loss": 0.0298,
      "step": 7252
    },
    {
      "epoch": 3.3852975495915985,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00014864272635858986,
      "loss": 0.0428,
      "step": 7253
    },
    {
      "epoch": 3.3857642940490082,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001486299053038577,
      "loss": 0.0352,
      "step": 7254
    },
    {
      "epoch": 3.3862310385064176,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014861708320206978,
      "loss": 0.0229,
      "step": 7255
    },
    {
      "epoch": 3.3866977829638274,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014860426005350213,
      "loss": 0.0377,
      "step": 7256
    },
    {
      "epoch": 3.3871645274212367,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00014859143585843078,
      "loss": 0.0521,
      "step": 7257
    },
    {
      "epoch": 3.3876312718786465,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014857861061713193,
      "loss": 0.0307,
      "step": 7258
    },
    {
      "epoch": 3.388098016336056,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001485657843298817,
      "loss": 0.024,
      "step": 7259
    },
    {
      "epoch": 3.3885647607934657,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014855295699695625,
      "loss": 0.0396,
      "step": 7260
    },
    {
      "epoch": 3.389031505250875,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014854012861863173,
      "loss": 0.0291,
      "step": 7261
    },
    {
      "epoch": 3.389498249708285,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001485272991951844,
      "loss": 0.0416,
      "step": 7262
    },
    {
      "epoch": 3.389964994165694,
      "grad_norm": 0.90625,
      "learning_rate": 0.00014851446872689053,
      "loss": 0.0435,
      "step": 7263
    },
    {
      "epoch": 3.390431738623104,
      "grad_norm": 0.53125,
      "learning_rate": 0.00014850163721402625,
      "loss": 0.0344,
      "step": 7264
    },
    {
      "epoch": 3.3908984830805133,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014848880465686797,
      "loss": 0.0311,
      "step": 7265
    },
    {
      "epoch": 3.391365227537923,
      "grad_norm": 1.015625,
      "learning_rate": 0.00014847597105569187,
      "loss": 0.0384,
      "step": 7266
    },
    {
      "epoch": 3.3918319719953325,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014846313641077435,
      "loss": 0.0343,
      "step": 7267
    },
    {
      "epoch": 3.3922987164527423,
      "grad_norm": 1.0625,
      "learning_rate": 0.00014845030072239173,
      "loss": 0.0346,
      "step": 7268
    },
    {
      "epoch": 3.3927654609101516,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001484374639908204,
      "loss": 0.0293,
      "step": 7269
    },
    {
      "epoch": 3.3932322053675614,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014842462621633668,
      "loss": 0.0295,
      "step": 7270
    },
    {
      "epoch": 3.393698949824971,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00014841178739921708,
      "loss": 0.0446,
      "step": 7271
    },
    {
      "epoch": 3.3941656942823806,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014839894753973794,
      "loss": 0.0301,
      "step": 7272
    },
    {
      "epoch": 3.39463243873979,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001483861066381758,
      "loss": 0.0268,
      "step": 7273
    },
    {
      "epoch": 3.3950991831971997,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014837326469480707,
      "loss": 0.0342,
      "step": 7274
    },
    {
      "epoch": 3.395565927654609,
      "grad_norm": 0.87109375,
      "learning_rate": 0.0001483604217099083,
      "loss": 0.0461,
      "step": 7275
    },
    {
      "epoch": 3.3960326721120184,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000148347577683756,
      "loss": 0.0505,
      "step": 7276
    },
    {
      "epoch": 3.396499416569428,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014833473261662667,
      "loss": 0.0315,
      "step": 7277
    },
    {
      "epoch": 3.396966161026838,
      "grad_norm": 1.078125,
      "learning_rate": 0.00014832188650879696,
      "loss": 0.0514,
      "step": 7278
    },
    {
      "epoch": 3.3974329054842474,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014830903936054342,
      "loss": 0.0358,
      "step": 7279
    },
    {
      "epoch": 3.3978996499416567,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014829619117214261,
      "loss": 0.0346,
      "step": 7280
    },
    {
      "epoch": 3.3983663943990665,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014828334194387125,
      "loss": 0.0324,
      "step": 7281
    },
    {
      "epoch": 3.3988331388564763,
      "grad_norm": 0.88671875,
      "learning_rate": 0.000148270491676006,
      "loss": 0.0396,
      "step": 7282
    },
    {
      "epoch": 3.3992998833138857,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014825764036882348,
      "loss": 0.0275,
      "step": 7283
    },
    {
      "epoch": 3.399766627771295,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001482447880226004,
      "loss": 0.0194,
      "step": 7284
    },
    {
      "epoch": 3.400233372228705,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014823193463761353,
      "loss": 0.0273,
      "step": 7285
    },
    {
      "epoch": 3.400700116686114,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014821908021413958,
      "loss": 0.0326,
      "step": 7286
    },
    {
      "epoch": 3.401166861143524,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00014820622475245535,
      "loss": 0.0363,
      "step": 7287
    },
    {
      "epoch": 3.4016336056009333,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014819336825283762,
      "loss": 0.0364,
      "step": 7288
    },
    {
      "epoch": 3.402100350058343,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001481805107155632,
      "loss": 0.0541,
      "step": 7289
    },
    {
      "epoch": 3.4025670945157525,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001481676521409089,
      "loss": 0.0294,
      "step": 7290
    },
    {
      "epoch": 3.4030338389731623,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001481547925291516,
      "loss": 0.0273,
      "step": 7291
    },
    {
      "epoch": 3.4035005834305716,
      "grad_norm": 0.90625,
      "learning_rate": 0.00014814193188056823,
      "loss": 0.0299,
      "step": 7292
    },
    {
      "epoch": 3.4039673278879814,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001481290701954356,
      "loss": 0.0345,
      "step": 7293
    },
    {
      "epoch": 3.4044340723453907,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00014811620747403072,
      "loss": 0.0268,
      "step": 7294
    },
    {
      "epoch": 3.4049008168028005,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014810334371663053,
      "loss": 0.033,
      "step": 7295
    },
    {
      "epoch": 3.40536756126021,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014809047892351196,
      "loss": 0.0192,
      "step": 7296
    },
    {
      "epoch": 3.4058343057176197,
      "grad_norm": 0.67578125,
      "learning_rate": 0.000148077613094952,
      "loss": 0.0236,
      "step": 7297
    },
    {
      "epoch": 3.406301050175029,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014806474623122771,
      "loss": 0.0464,
      "step": 7298
    },
    {
      "epoch": 3.406767794632439,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001480518783326161,
      "loss": 0.0298,
      "step": 7299
    },
    {
      "epoch": 3.407234539089848,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001480390093993942,
      "loss": 0.0318,
      "step": 7300
    },
    {
      "epoch": 3.407701283547258,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014802613943183917,
      "loss": 0.0336,
      "step": 7301
    },
    {
      "epoch": 3.4081680280046673,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014801326843022806,
      "loss": 0.0299,
      "step": 7302
    },
    {
      "epoch": 3.408634772462077,
      "grad_norm": 0.6875,
      "learning_rate": 0.000148000396394838,
      "loss": 0.0354,
      "step": 7303
    },
    {
      "epoch": 3.4091015169194865,
      "grad_norm": 0.984375,
      "learning_rate": 0.00014798752332594617,
      "loss": 0.0511,
      "step": 7304
    },
    {
      "epoch": 3.4095682613768963,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014797464922382972,
      "loss": 0.0262,
      "step": 7305
    },
    {
      "epoch": 3.4100350058343056,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014796177408876583,
      "loss": 0.0307,
      "step": 7306
    },
    {
      "epoch": 3.4105017502917154,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001479488979210317,
      "loss": 0.0289,
      "step": 7307
    },
    {
      "epoch": 3.410968494749125,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014793602072090463,
      "loss": 0.025,
      "step": 7308
    },
    {
      "epoch": 3.4114352392065346,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014792314248866185,
      "loss": 0.0401,
      "step": 7309
    },
    {
      "epoch": 3.411901983663944,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014791026322458065,
      "loss": 0.0243,
      "step": 7310
    },
    {
      "epoch": 3.4123687281213537,
      "grad_norm": 0.859375,
      "learning_rate": 0.00014789738292893834,
      "loss": 0.0216,
      "step": 7311
    },
    {
      "epoch": 3.412835472578763,
      "grad_norm": 1.2421875,
      "learning_rate": 0.0001478845016020122,
      "loss": 0.0509,
      "step": 7312
    },
    {
      "epoch": 3.413302217036173,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001478716192440796,
      "loss": 0.0284,
      "step": 7313
    },
    {
      "epoch": 3.413768961493582,
      "grad_norm": 1.0,
      "learning_rate": 0.00014785873585541795,
      "loss": 0.0341,
      "step": 7314
    },
    {
      "epoch": 3.414235705950992,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014784585143630462,
      "loss": 0.0295,
      "step": 7315
    },
    {
      "epoch": 3.4147024504084014,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014783296598701697,
      "loss": 0.0225,
      "step": 7316
    },
    {
      "epoch": 3.415169194865811,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014782007950783253,
      "loss": 0.0328,
      "step": 7317
    },
    {
      "epoch": 3.4156359393232205,
      "grad_norm": 0.87109375,
      "learning_rate": 0.00014780719199902875,
      "loss": 0.0391,
      "step": 7318
    },
    {
      "epoch": 3.41610268378063,
      "grad_norm": 0.53125,
      "learning_rate": 0.00014779430346088302,
      "loss": 0.0259,
      "step": 7319
    },
    {
      "epoch": 3.4165694282380397,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014778141389367298,
      "loss": 0.0234,
      "step": 7320
    },
    {
      "epoch": 3.4170361726954495,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014776852329767602,
      "loss": 0.0256,
      "step": 7321
    },
    {
      "epoch": 3.417502917152859,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014775563167316978,
      "loss": 0.0409,
      "step": 7322
    },
    {
      "epoch": 3.417969661610268,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00014774273902043181,
      "loss": 0.0205,
      "step": 7323
    },
    {
      "epoch": 3.418436406067678,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014772984533973968,
      "loss": 0.026,
      "step": 7324
    },
    {
      "epoch": 3.4189031505250878,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014771695063137106,
      "loss": 0.0265,
      "step": 7325
    },
    {
      "epoch": 3.419369894982497,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001477040548956035,
      "loss": 0.0424,
      "step": 7326
    },
    {
      "epoch": 3.4198366394399065,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00014769115813271475,
      "loss": 0.0344,
      "step": 7327
    },
    {
      "epoch": 3.4203033838973163,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014767826034298243,
      "loss": 0.0229,
      "step": 7328
    },
    {
      "epoch": 3.4207701283547256,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00014766536152668427,
      "loss": 0.0151,
      "step": 7329
    },
    {
      "epoch": 3.4212368728121354,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014765246168409798,
      "loss": 0.0336,
      "step": 7330
    },
    {
      "epoch": 3.4217036172695448,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014763956081550135,
      "loss": 0.0369,
      "step": 7331
    },
    {
      "epoch": 3.4221703617269545,
      "grad_norm": 0.81640625,
      "learning_rate": 0.0001476266589211721,
      "loss": 0.0337,
      "step": 7332
    },
    {
      "epoch": 3.422637106184364,
      "grad_norm": 0.51171875,
      "learning_rate": 0.000147613756001388,
      "loss": 0.0177,
      "step": 7333
    },
    {
      "epoch": 3.4231038506417737,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014760085205642697,
      "loss": 0.0338,
      "step": 7334
    },
    {
      "epoch": 3.423570595099183,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00014758794708656675,
      "loss": 0.0393,
      "step": 7335
    },
    {
      "epoch": 3.424037339556593,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014757504109208524,
      "loss": 0.0284,
      "step": 7336
    },
    {
      "epoch": 3.424504084014002,
      "grad_norm": 0.984375,
      "learning_rate": 0.00014756213407326031,
      "loss": 0.0333,
      "step": 7337
    },
    {
      "epoch": 3.424970828471412,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014754922603036987,
      "loss": 0.0511,
      "step": 7338
    },
    {
      "epoch": 3.4254375729288213,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014753631696369182,
      "loss": 0.0309,
      "step": 7339
    },
    {
      "epoch": 3.425904317386231,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014752340687350412,
      "loss": 0.0431,
      "step": 7340
    },
    {
      "epoch": 3.4263710618436405,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00014751049576008476,
      "loss": 0.0204,
      "step": 7341
    },
    {
      "epoch": 3.4268378063010503,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014749758362371166,
      "loss": 0.0245,
      "step": 7342
    },
    {
      "epoch": 3.4273045507584596,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00014748467046466295,
      "loss": 0.0153,
      "step": 7343
    },
    {
      "epoch": 3.4277712952158694,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014747175628321656,
      "loss": 0.0294,
      "step": 7344
    },
    {
      "epoch": 3.428238039673279,
      "grad_norm": 0.98828125,
      "learning_rate": 0.0001474588410796506,
      "loss": 0.0494,
      "step": 7345
    },
    {
      "epoch": 3.4287047841306886,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014744592485424314,
      "loss": 0.0261,
      "step": 7346
    },
    {
      "epoch": 3.429171528588098,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014743300760727227,
      "loss": 0.029,
      "step": 7347
    },
    {
      "epoch": 3.4296382730455077,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014742008933901613,
      "loss": 0.0324,
      "step": 7348
    },
    {
      "epoch": 3.430105017502917,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014740717004975285,
      "loss": 0.0316,
      "step": 7349
    },
    {
      "epoch": 3.430571761960327,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001473942497397606,
      "loss": 0.0302,
      "step": 7350
    },
    {
      "epoch": 3.4310385064177362,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014738132840931757,
      "loss": 0.0252,
      "step": 7351
    },
    {
      "epoch": 3.431505250875146,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014736840605870192,
      "loss": 0.0292,
      "step": 7352
    },
    {
      "epoch": 3.4319719953325554,
      "grad_norm": 0.78125,
      "learning_rate": 0.000147355482688192,
      "loss": 0.0281,
      "step": 7353
    },
    {
      "epoch": 3.432438739789965,
      "grad_norm": 1.0078125,
      "learning_rate": 0.00014734255829806597,
      "loss": 0.0449,
      "step": 7354
    },
    {
      "epoch": 3.4329054842473745,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014732963288860212,
      "loss": 0.0243,
      "step": 7355
    },
    {
      "epoch": 3.4333722287047843,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014731670646007876,
      "loss": 0.0301,
      "step": 7356
    },
    {
      "epoch": 3.4338389731621937,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00014730377901277423,
      "loss": 0.0427,
      "step": 7357
    },
    {
      "epoch": 3.4343057176196035,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014729085054696685,
      "loss": 0.0207,
      "step": 7358
    },
    {
      "epoch": 3.434772462077013,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014727792106293498,
      "loss": 0.0311,
      "step": 7359
    },
    {
      "epoch": 3.4352392065344226,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00014726499056095702,
      "loss": 0.021,
      "step": 7360
    },
    {
      "epoch": 3.435705950991832,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001472520590413114,
      "loss": 0.0287,
      "step": 7361
    },
    {
      "epoch": 3.4361726954492413,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014723912650427647,
      "loss": 0.0325,
      "step": 7362
    },
    {
      "epoch": 3.436639439906651,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014722619295013076,
      "loss": 0.0354,
      "step": 7363
    },
    {
      "epoch": 3.437106184364061,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001472132583791527,
      "loss": 0.0291,
      "step": 7364
    },
    {
      "epoch": 3.4375729288214703,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014720032279162083,
      "loss": 0.0305,
      "step": 7365
    },
    {
      "epoch": 3.4380396732788796,
      "grad_norm": 0.73046875,
      "learning_rate": 0.0001471873861878136,
      "loss": 0.02,
      "step": 7366
    },
    {
      "epoch": 3.4385064177362894,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014717444856800964,
      "loss": 0.0366,
      "step": 7367
    },
    {
      "epoch": 3.438973162193699,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001471615099324874,
      "loss": 0.0296,
      "step": 7368
    },
    {
      "epoch": 3.4394399066511085,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00014714857028152558,
      "loss": 0.0256,
      "step": 7369
    },
    {
      "epoch": 3.439906651108518,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014713562961540274,
      "loss": 0.0252,
      "step": 7370
    },
    {
      "epoch": 3.4403733955659277,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00014712268793439745,
      "loss": 0.0226,
      "step": 7371
    },
    {
      "epoch": 3.440840140023337,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001471097452387884,
      "loss": 0.0541,
      "step": 7372
    },
    {
      "epoch": 3.441306884480747,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001470968015288543,
      "loss": 0.0298,
      "step": 7373
    },
    {
      "epoch": 3.441773628938156,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001470838568048738,
      "loss": 0.0368,
      "step": 7374
    },
    {
      "epoch": 3.442240373395566,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014707091106712562,
      "loss": 0.021,
      "step": 7375
    },
    {
      "epoch": 3.4427071178529753,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001470579643158885,
      "loss": 0.0263,
      "step": 7376
    },
    {
      "epoch": 3.443173862310385,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001470450165514412,
      "loss": 0.0201,
      "step": 7377
    },
    {
      "epoch": 3.4436406067677945,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014703206777406247,
      "loss": 0.0254,
      "step": 7378
    },
    {
      "epoch": 3.4441073512252043,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014701911798403115,
      "loss": 0.0253,
      "step": 7379
    },
    {
      "epoch": 3.4445740956826136,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014700616718162608,
      "loss": 0.0306,
      "step": 7380
    },
    {
      "epoch": 3.4450408401400234,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014699321536712606,
      "loss": 0.0236,
      "step": 7381
    },
    {
      "epoch": 3.445507584597433,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014698026254080995,
      "loss": 0.0247,
      "step": 7382
    },
    {
      "epoch": 3.4459743290548426,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014696730870295672,
      "loss": 0.0331,
      "step": 7383
    },
    {
      "epoch": 3.446441073512252,
      "grad_norm": 0.9921875,
      "learning_rate": 0.00014695435385384515,
      "loss": 0.0304,
      "step": 7384
    },
    {
      "epoch": 3.4469078179696617,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014694139799375431,
      "loss": 0.019,
      "step": 7385
    },
    {
      "epoch": 3.447374562427071,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014692844112296305,
      "loss": 0.0284,
      "step": 7386
    },
    {
      "epoch": 3.447841306884481,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001469154832417504,
      "loss": 0.0395,
      "step": 7387
    },
    {
      "epoch": 3.4483080513418902,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00014690252435039532,
      "loss": 0.0254,
      "step": 7388
    },
    {
      "epoch": 3.4487747957993,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001468895644491769,
      "loss": 0.0433,
      "step": 7389
    },
    {
      "epoch": 3.4492415402567094,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014687660353837406,
      "loss": 0.0264,
      "step": 7390
    },
    {
      "epoch": 3.449708284714119,
      "grad_norm": 1.125,
      "learning_rate": 0.00014686364161826598,
      "loss": 0.0552,
      "step": 7391
    },
    {
      "epoch": 3.4501750291715285,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001468506786891317,
      "loss": 0.0311,
      "step": 7392
    },
    {
      "epoch": 3.4506417736289383,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014683771475125032,
      "loss": 0.0316,
      "step": 7393
    },
    {
      "epoch": 3.4511085180863477,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014682474980490096,
      "loss": 0.0219,
      "step": 7394
    },
    {
      "epoch": 3.4515752625437575,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001468117838503628,
      "loss": 0.0333,
      "step": 7395
    },
    {
      "epoch": 3.452042007001167,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000146798816887915,
      "loss": 0.019,
      "step": 7396
    },
    {
      "epoch": 3.4525087514585766,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014678584891783674,
      "loss": 0.042,
      "step": 7397
    },
    {
      "epoch": 3.452975495915986,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014677287994040722,
      "loss": 0.0276,
      "step": 7398
    },
    {
      "epoch": 3.4534422403733958,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014675990995590573,
      "loss": 0.0363,
      "step": 7399
    },
    {
      "epoch": 3.453908984830805,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001467469389646115,
      "loss": 0.0273,
      "step": 7400
    },
    {
      "epoch": 3.4543757292882145,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001467339669668038,
      "loss": 0.0204,
      "step": 7401
    },
    {
      "epoch": 3.4548424737456243,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014672099396276193,
      "loss": 0.0227,
      "step": 7402
    },
    {
      "epoch": 3.455309218203034,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014670801995276525,
      "loss": 0.0317,
      "step": 7403
    },
    {
      "epoch": 3.4557759626604434,
      "grad_norm": 0.75,
      "learning_rate": 0.00014669504493709305,
      "loss": 0.0276,
      "step": 7404
    },
    {
      "epoch": 3.4562427071178528,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00014668206891602472,
      "loss": 0.0309,
      "step": 7405
    },
    {
      "epoch": 3.4567094515752625,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001466690918898397,
      "loss": 0.0227,
      "step": 7406
    },
    {
      "epoch": 3.4571761960326723,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014665611385881729,
      "loss": 0.0399,
      "step": 7407
    },
    {
      "epoch": 3.4576429404900817,
      "grad_norm": 0.734375,
      "learning_rate": 0.000146643134823237,
      "loss": 0.0275,
      "step": 7408
    },
    {
      "epoch": 3.458109684947491,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001466301547833783,
      "loss": 0.0289,
      "step": 7409
    },
    {
      "epoch": 3.458576429404901,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001466171737395206,
      "loss": 0.0298,
      "step": 7410
    },
    {
      "epoch": 3.45904317386231,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001466041916919434,
      "loss": 0.0261,
      "step": 7411
    },
    {
      "epoch": 3.45950991831972,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001465912086409263,
      "loss": 0.0308,
      "step": 7412
    },
    {
      "epoch": 3.4599766627771293,
      "grad_norm": 0.609375,
      "learning_rate": 0.00014657822458674874,
      "loss": 0.0219,
      "step": 7413
    },
    {
      "epoch": 3.460443407234539,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014656523952969033,
      "loss": 0.0301,
      "step": 7414
    },
    {
      "epoch": 3.4609101516919485,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014655225347003068,
      "loss": 0.031,
      "step": 7415
    },
    {
      "epoch": 3.4613768961493583,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014653926640804935,
      "loss": 0.0244,
      "step": 7416
    },
    {
      "epoch": 3.4618436406067676,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014652627834402596,
      "loss": 0.0208,
      "step": 7417
    },
    {
      "epoch": 3.4623103850641774,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014651328927824017,
      "loss": 0.0357,
      "step": 7418
    },
    {
      "epoch": 3.462777129521587,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00014650029921097168,
      "loss": 0.046,
      "step": 7419
    },
    {
      "epoch": 3.4632438739789966,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014648730814250015,
      "loss": 0.027,
      "step": 7420
    },
    {
      "epoch": 3.463710618436406,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014647431607310528,
      "loss": 0.0221,
      "step": 7421
    },
    {
      "epoch": 3.4641773628938157,
      "grad_norm": 0.85546875,
      "learning_rate": 0.00014646132300306683,
      "loss": 0.0341,
      "step": 7422
    },
    {
      "epoch": 3.464644107351225,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014644832893266459,
      "loss": 0.0365,
      "step": 7423
    },
    {
      "epoch": 3.465110851808635,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014643533386217824,
      "loss": 0.0262,
      "step": 7424
    },
    {
      "epoch": 3.4655775962660442,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014642233779188764,
      "loss": 0.0362,
      "step": 7425
    },
    {
      "epoch": 3.466044340723454,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014640934072207259,
      "loss": 0.0257,
      "step": 7426
    },
    {
      "epoch": 3.4665110851808634,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00014639634265301295,
      "loss": 0.0383,
      "step": 7427
    },
    {
      "epoch": 3.466977829638273,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014638334358498857,
      "loss": 0.0212,
      "step": 7428
    },
    {
      "epoch": 3.4674445740956825,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00014637034351827936,
      "loss": 0.0544,
      "step": 7429
    },
    {
      "epoch": 3.4679113185530923,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014635734245316523,
      "loss": 0.0361,
      "step": 7430
    },
    {
      "epoch": 3.4683780630105017,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014634434038992604,
      "loss": 0.0291,
      "step": 7431
    },
    {
      "epoch": 3.4688448074679115,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014633133732884179,
      "loss": 0.0276,
      "step": 7432
    },
    {
      "epoch": 3.469311551925321,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001463183332701924,
      "loss": 0.0399,
      "step": 7433
    },
    {
      "epoch": 3.4697782963827306,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00014630532821425797,
      "loss": 0.0182,
      "step": 7434
    },
    {
      "epoch": 3.47024504084014,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014629232216131842,
      "loss": 0.0288,
      "step": 7435
    },
    {
      "epoch": 3.4707117852975498,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014627931511165384,
      "loss": 0.0255,
      "step": 7436
    },
    {
      "epoch": 3.471178529754959,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001462663070655442,
      "loss": 0.0227,
      "step": 7437
    },
    {
      "epoch": 3.471645274212369,
      "grad_norm": 0.89453125,
      "learning_rate": 0.0001462532980232697,
      "loss": 0.0342,
      "step": 7438
    },
    {
      "epoch": 3.4721120186697783,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014624028798511032,
      "loss": 0.032,
      "step": 7439
    },
    {
      "epoch": 3.472578763127188,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00014622727695134626,
      "loss": 0.0178,
      "step": 7440
    },
    {
      "epoch": 3.4730455075845974,
      "grad_norm": 0.484375,
      "learning_rate": 0.00014621426492225764,
      "loss": 0.0194,
      "step": 7441
    },
    {
      "epoch": 3.473512252042007,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001462012518981246,
      "loss": 0.0348,
      "step": 7442
    },
    {
      "epoch": 3.4739789964994165,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001461882378792274,
      "loss": 0.0328,
      "step": 7443
    },
    {
      "epoch": 3.474445740956826,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00014617522286584613,
      "loss": 0.0284,
      "step": 7444
    },
    {
      "epoch": 3.4749124854142357,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001461622068582611,
      "loss": 0.0234,
      "step": 7445
    },
    {
      "epoch": 3.4753792298716455,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014614918985675254,
      "loss": 0.0325,
      "step": 7446
    },
    {
      "epoch": 3.475845974329055,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014613617186160071,
      "loss": 0.0359,
      "step": 7447
    },
    {
      "epoch": 3.476312718786464,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014612315287308595,
      "loss": 0.0314,
      "step": 7448
    },
    {
      "epoch": 3.476779463243874,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001461101328914885,
      "loss": 0.0236,
      "step": 7449
    },
    {
      "epoch": 3.477246207701284,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001460971119170887,
      "loss": 0.0249,
      "step": 7450
    },
    {
      "epoch": 3.477712952158693,
      "grad_norm": 0.8203125,
      "learning_rate": 0.000146084089950167,
      "loss": 0.0318,
      "step": 7451
    },
    {
      "epoch": 3.4781796966161025,
      "grad_norm": 0.421875,
      "learning_rate": 0.00014607106699100365,
      "loss": 0.0182,
      "step": 7452
    },
    {
      "epoch": 3.4786464410735123,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014605804303987916,
      "loss": 0.0338,
      "step": 7453
    },
    {
      "epoch": 3.4791131855309216,
      "grad_norm": 0.83984375,
      "learning_rate": 0.0001460450180970739,
      "loss": 0.031,
      "step": 7454
    },
    {
      "epoch": 3.4795799299883314,
      "grad_norm": 1.046875,
      "learning_rate": 0.0001460319921628683,
      "loss": 0.035,
      "step": 7455
    },
    {
      "epoch": 3.480046674445741,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014601896523754282,
      "loss": 0.0326,
      "step": 7456
    },
    {
      "epoch": 3.4805134189031506,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014600593732137798,
      "loss": 0.0326,
      "step": 7457
    },
    {
      "epoch": 3.48098016336056,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014599290841465424,
      "loss": 0.0271,
      "step": 7458
    },
    {
      "epoch": 3.4814469078179697,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014597987851765215,
      "loss": 0.0326,
      "step": 7459
    },
    {
      "epoch": 3.481913652275379,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001459668476306523,
      "loss": 0.0387,
      "step": 7460
    },
    {
      "epoch": 3.482380396732789,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014595381575393518,
      "loss": 0.0335,
      "step": 7461
    },
    {
      "epoch": 3.4828471411901982,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00014594078288778146,
      "loss": 0.0436,
      "step": 7462
    },
    {
      "epoch": 3.483313885647608,
      "grad_norm": 0.5,
      "learning_rate": 0.00014592774903247167,
      "loss": 0.0149,
      "step": 7463
    },
    {
      "epoch": 3.4837806301050174,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001459147141882865,
      "loss": 0.0345,
      "step": 7464
    },
    {
      "epoch": 3.484247374562427,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001459016783555066,
      "loss": 0.0318,
      "step": 7465
    },
    {
      "epoch": 3.4847141190198365,
      "grad_norm": 0.6640625,
      "learning_rate": 0.0001458886415344126,
      "loss": 0.0323,
      "step": 7466
    },
    {
      "epoch": 3.4851808634772463,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001458756037252853,
      "loss": 0.0335,
      "step": 7467
    },
    {
      "epoch": 3.4856476079346557,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014586256492840532,
      "loss": 0.0239,
      "step": 7468
    },
    {
      "epoch": 3.4861143523920655,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014584952514405344,
      "loss": 0.0222,
      "step": 7469
    },
    {
      "epoch": 3.486581096849475,
      "grad_norm": 0.95703125,
      "learning_rate": 0.00014583648437251035,
      "loss": 0.0337,
      "step": 7470
    },
    {
      "epoch": 3.4870478413068846,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014582344261405695,
      "loss": 0.0331,
      "step": 7471
    },
    {
      "epoch": 3.487514585764294,
      "grad_norm": 0.6171875,
      "learning_rate": 0.000145810399868974,
      "loss": 0.023,
      "step": 7472
    },
    {
      "epoch": 3.4879813302217038,
      "grad_norm": 0.75,
      "learning_rate": 0.00014579735613754228,
      "loss": 0.0327,
      "step": 7473
    },
    {
      "epoch": 3.488448074679113,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001457843114200427,
      "loss": 0.0232,
      "step": 7474
    },
    {
      "epoch": 3.488914819136523,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014577126571675608,
      "loss": 0.0292,
      "step": 7475
    },
    {
      "epoch": 3.4893815635939323,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001457582190279633,
      "loss": 0.0265,
      "step": 7476
    },
    {
      "epoch": 3.489848308051342,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00014574517135394538,
      "loss": 0.0134,
      "step": 7477
    },
    {
      "epoch": 3.4903150525087514,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014573212269498311,
      "loss": 0.0288,
      "step": 7478
    },
    {
      "epoch": 3.490781796966161,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001457190730513575,
      "loss": 0.0334,
      "step": 7479
    },
    {
      "epoch": 3.4912485414235706,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001457060224233495,
      "loss": 0.0252,
      "step": 7480
    },
    {
      "epoch": 3.4917152858809803,
      "grad_norm": 1.0546875,
      "learning_rate": 0.00014569297081124013,
      "loss": 0.0461,
      "step": 7481
    },
    {
      "epoch": 3.4921820303383897,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001456799182153104,
      "loss": 0.0323,
      "step": 7482
    },
    {
      "epoch": 3.4926487747957995,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014566686463584138,
      "loss": 0.0266,
      "step": 7483
    },
    {
      "epoch": 3.493115519253209,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001456538100731141,
      "loss": 0.0288,
      "step": 7484
    },
    {
      "epoch": 3.4935822637106186,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014564075452740962,
      "loss": 0.02,
      "step": 7485
    },
    {
      "epoch": 3.494049008168028,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014562769799900902,
      "loss": 0.0273,
      "step": 7486
    },
    {
      "epoch": 3.4945157526254373,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014561464048819348,
      "loss": 0.0263,
      "step": 7487
    },
    {
      "epoch": 3.494982497082847,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014560158199524413,
      "loss": 0.0291,
      "step": 7488
    },
    {
      "epoch": 3.495449241540257,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014558852252044213,
      "loss": 0.0281,
      "step": 7489
    },
    {
      "epoch": 3.4959159859976663,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014557546206406865,
      "loss": 0.0245,
      "step": 7490
    },
    {
      "epoch": 3.4963827304550756,
      "grad_norm": 0.921875,
      "learning_rate": 0.0001455624006264049,
      "loss": 0.0385,
      "step": 7491
    },
    {
      "epoch": 3.4968494749124854,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001455493382077321,
      "loss": 0.0242,
      "step": 7492
    },
    {
      "epoch": 3.4973162193698952,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014553627480833154,
      "loss": 0.0236,
      "step": 7493
    },
    {
      "epoch": 3.4977829638273046,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014552321042848445,
      "loss": 0.0248,
      "step": 7494
    },
    {
      "epoch": 3.498249708284714,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014551014506847214,
      "loss": 0.0303,
      "step": 7495
    },
    {
      "epoch": 3.4987164527421237,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014549707872857587,
      "loss": 0.0388,
      "step": 7496
    },
    {
      "epoch": 3.499183197199533,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014548401140907706,
      "loss": 0.0213,
      "step": 7497
    },
    {
      "epoch": 3.499649941656943,
      "grad_norm": 0.71875,
      "learning_rate": 0.000145470943110257,
      "loss": 0.0293,
      "step": 7498
    },
    {
      "epoch": 3.5001166861143522,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00014545787383239705,
      "loss": 0.019,
      "step": 7499
    },
    {
      "epoch": 3.500583430571762,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014544480357577873,
      "loss": 0.0295,
      "step": 7500
    },
    {
      "epoch": 3.5010501750291714,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001454317323406833,
      "loss": 0.0212,
      "step": 7501
    },
    {
      "epoch": 3.501516919486581,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014541866012739228,
      "loss": 0.0343,
      "step": 7502
    },
    {
      "epoch": 3.5019836639439905,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014540558693618713,
      "loss": 0.0312,
      "step": 7503
    },
    {
      "epoch": 3.5024504084014003,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001453925127673493,
      "loss": 0.0262,
      "step": 7504
    },
    {
      "epoch": 3.5024504084014003,
      "eval_loss": 1.7928109169006348,
      "eval_runtime": 58.9301,
      "eval_samples_per_second": 30.613,
      "eval_steps_per_second": 3.835,
      "step": 7504
    },
    {
      "epoch": 3.5029171528588097,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001453794376211603,
      "loss": 0.0311,
      "step": 7505
    },
    {
      "epoch": 3.5033838973162195,
      "grad_norm": 0.5,
      "learning_rate": 0.00014536636149790167,
      "loss": 0.0155,
      "step": 7506
    },
    {
      "epoch": 3.503850641773629,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014535328439785493,
      "loss": 0.029,
      "step": 7507
    },
    {
      "epoch": 3.5043173862310386,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014534020632130166,
      "loss": 0.0221,
      "step": 7508
    },
    {
      "epoch": 3.504784130688448,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014532712726852344,
      "loss": 0.0353,
      "step": 7509
    },
    {
      "epoch": 3.5052508751458578,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001453140472398019,
      "loss": 0.023,
      "step": 7510
    },
    {
      "epoch": 3.505717619603267,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014530096623541862,
      "loss": 0.0211,
      "step": 7511
    },
    {
      "epoch": 3.506184364060677,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014528788425565526,
      "loss": 0.0299,
      "step": 7512
    },
    {
      "epoch": 3.5066511085180863,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014527480130079355,
      "loss": 0.0202,
      "step": 7513
    },
    {
      "epoch": 3.507117852975496,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001452617173711151,
      "loss": 0.0229,
      "step": 7514
    },
    {
      "epoch": 3.5075845974329054,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014524863246690166,
      "loss": 0.0286,
      "step": 7515
    },
    {
      "epoch": 3.508051341890315,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00014523554658843498,
      "loss": 0.0414,
      "step": 7516
    },
    {
      "epoch": 3.5085180863477246,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001452224597359968,
      "loss": 0.0301,
      "step": 7517
    },
    {
      "epoch": 3.5089848308051343,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014520937190986883,
      "loss": 0.0356,
      "step": 7518
    },
    {
      "epoch": 3.5094515752625437,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014519628311033297,
      "loss": 0.02,
      "step": 7519
    },
    {
      "epoch": 3.5099183197199535,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000145183193337671,
      "loss": 0.0299,
      "step": 7520
    },
    {
      "epoch": 3.510385064177363,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014517010259216474,
      "loss": 0.0228,
      "step": 7521
    },
    {
      "epoch": 3.510851808634772,
      "grad_norm": 0.7109375,
      "learning_rate": 0.000145157010874096,
      "loss": 0.0177,
      "step": 7522
    },
    {
      "epoch": 3.511318553092182,
      "grad_norm": 1.046875,
      "learning_rate": 0.00014514391818374676,
      "loss": 0.0455,
      "step": 7523
    },
    {
      "epoch": 3.511785297549592,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014513082452139887,
      "loss": 0.0426,
      "step": 7524
    },
    {
      "epoch": 3.512252042007001,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014511772988733422,
      "loss": 0.0281,
      "step": 7525
    },
    {
      "epoch": 3.5127187864644105,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014510463428183486,
      "loss": 0.0152,
      "step": 7526
    },
    {
      "epoch": 3.5131855309218203,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014509153770518265,
      "loss": 0.0457,
      "step": 7527
    },
    {
      "epoch": 3.51365227537923,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014507844015765957,
      "loss": 0.0162,
      "step": 7528
    },
    {
      "epoch": 3.5141190198366394,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014506534163954767,
      "loss": 0.0182,
      "step": 7529
    },
    {
      "epoch": 3.514585764294049,
      "grad_norm": 0.8046875,
      "learning_rate": 0.000145052242151129,
      "loss": 0.0266,
      "step": 7530
    },
    {
      "epoch": 3.5150525087514586,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014503914169268552,
      "loss": 0.0282,
      "step": 7531
    },
    {
      "epoch": 3.5155192532088684,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00014502604026449934,
      "loss": 0.0182,
      "step": 7532
    },
    {
      "epoch": 3.5159859976662777,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001450129378668526,
      "loss": 0.038,
      "step": 7533
    },
    {
      "epoch": 3.516452742123687,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00014499983450002736,
      "loss": 0.0241,
      "step": 7534
    },
    {
      "epoch": 3.516919486581097,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014498673016430572,
      "loss": 0.0284,
      "step": 7535
    },
    {
      "epoch": 3.5173862310385067,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001449736248599699,
      "loss": 0.0178,
      "step": 7536
    },
    {
      "epoch": 3.517852975495916,
      "grad_norm": 1.3359375,
      "learning_rate": 0.00014496051858730203,
      "loss": 0.0385,
      "step": 7537
    },
    {
      "epoch": 3.5183197199533254,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001449474113465843,
      "loss": 0.0447,
      "step": 7538
    },
    {
      "epoch": 3.518786464410735,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014493430313809895,
      "loss": 0.0234,
      "step": 7539
    },
    {
      "epoch": 3.519253208868145,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014492119396212818,
      "loss": 0.0229,
      "step": 7540
    },
    {
      "epoch": 3.5197199533255543,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014490808381895426,
      "loss": 0.0223,
      "step": 7541
    },
    {
      "epoch": 3.5201866977829637,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001448949727088595,
      "loss": 0.0269,
      "step": 7542
    },
    {
      "epoch": 3.5206534422403735,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014488186063212614,
      "loss": 0.0293,
      "step": 7543
    },
    {
      "epoch": 3.521120186697783,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014486874758903656,
      "loss": 0.0301,
      "step": 7544
    },
    {
      "epoch": 3.5215869311551926,
      "grad_norm": 0.7265625,
      "learning_rate": 0.000144855633579873,
      "loss": 0.0226,
      "step": 7545
    },
    {
      "epoch": 3.522053675612602,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014484251860491791,
      "loss": 0.0388,
      "step": 7546
    },
    {
      "epoch": 3.5225204200700118,
      "grad_norm": 0.578125,
      "learning_rate": 0.00014482940266445367,
      "loss": 0.0151,
      "step": 7547
    },
    {
      "epoch": 3.522987164527421,
      "grad_norm": 0.8203125,
      "learning_rate": 0.0001448162857587626,
      "loss": 0.0248,
      "step": 7548
    },
    {
      "epoch": 3.523453908984831,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014480316788812722,
      "loss": 0.0203,
      "step": 7549
    },
    {
      "epoch": 3.5239206534422403,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00014479004905282991,
      "loss": 0.021,
      "step": 7550
    },
    {
      "epoch": 3.52438739789965,
      "grad_norm": 0.87890625,
      "learning_rate": 0.0001447769292531531,
      "loss": 0.0394,
      "step": 7551
    },
    {
      "epoch": 3.5248541423570594,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001447638084893794,
      "loss": 0.0218,
      "step": 7552
    },
    {
      "epoch": 3.525320886814469,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001447506867617912,
      "loss": 0.0152,
      "step": 7553
    },
    {
      "epoch": 3.5257876312718786,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014473756407067104,
      "loss": 0.0245,
      "step": 7554
    },
    {
      "epoch": 3.5262543757292883,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014472444041630154,
      "loss": 0.0237,
      "step": 7555
    },
    {
      "epoch": 3.5267211201866977,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014471131579896516,
      "loss": 0.026,
      "step": 7556
    },
    {
      "epoch": 3.5271878646441075,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014469819021894456,
      "loss": 0.0335,
      "step": 7557
    },
    {
      "epoch": 3.527654609101517,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014468506367652236,
      "loss": 0.0219,
      "step": 7558
    },
    {
      "epoch": 3.5281213535589266,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014467193617198113,
      "loss": 0.0249,
      "step": 7559
    },
    {
      "epoch": 3.528588098016336,
      "grad_norm": 0.546875,
      "learning_rate": 0.00014465880770560358,
      "loss": 0.0142,
      "step": 7560
    },
    {
      "epoch": 3.529054842473746,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014464567827767233,
      "loss": 0.0303,
      "step": 7561
    },
    {
      "epoch": 3.529521586931155,
      "grad_norm": 0.84375,
      "learning_rate": 0.0001446325478884701,
      "loss": 0.0366,
      "step": 7562
    },
    {
      "epoch": 3.529988331388565,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00014461941653827958,
      "loss": 0.0149,
      "step": 7563
    },
    {
      "epoch": 3.5304550758459743,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014460628422738352,
      "loss": 0.0175,
      "step": 7564
    },
    {
      "epoch": 3.5309218203033836,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014459315095606467,
      "loss": 0.0238,
      "step": 7565
    },
    {
      "epoch": 3.5313885647607934,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001445800167246058,
      "loss": 0.0333,
      "step": 7566
    },
    {
      "epoch": 3.5318553092182032,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001445668815332897,
      "loss": 0.0273,
      "step": 7567
    },
    {
      "epoch": 3.5323220536756126,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001445537453823992,
      "loss": 0.023,
      "step": 7568
    },
    {
      "epoch": 3.532788798133022,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014454060827221715,
      "loss": 0.0316,
      "step": 7569
    },
    {
      "epoch": 3.5332555425904317,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014452747020302637,
      "loss": 0.0328,
      "step": 7570
    },
    {
      "epoch": 3.5337222870478415,
      "grad_norm": 0.578125,
      "learning_rate": 0.00014451433117510973,
      "loss": 0.024,
      "step": 7571
    },
    {
      "epoch": 3.534189031505251,
      "grad_norm": 2.078125,
      "learning_rate": 0.0001445011911887502,
      "loss": 0.0451,
      "step": 7572
    },
    {
      "epoch": 3.5346557759626602,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014448805024423059,
      "loss": 0.0202,
      "step": 7573
    },
    {
      "epoch": 3.53512252042007,
      "grad_norm": 0.625,
      "learning_rate": 0.00014447490834183395,
      "loss": 0.0252,
      "step": 7574
    },
    {
      "epoch": 3.53558926487748,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001444617654818432,
      "loss": 0.0333,
      "step": 7575
    },
    {
      "epoch": 3.536056009334889,
      "grad_norm": 0.546875,
      "learning_rate": 0.00014444862166454125,
      "loss": 0.0201,
      "step": 7576
    },
    {
      "epoch": 3.5365227537922985,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014443547689021118,
      "loss": 0.0261,
      "step": 7577
    },
    {
      "epoch": 3.5369894982497083,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014442233115913598,
      "loss": 0.0286,
      "step": 7578
    },
    {
      "epoch": 3.537456242707118,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014440918447159871,
      "loss": 0.0273,
      "step": 7579
    },
    {
      "epoch": 3.5379229871645275,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014439603682788244,
      "loss": 0.0401,
      "step": 7580
    },
    {
      "epoch": 3.538389731621937,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014438288822827023,
      "loss": 0.0271,
      "step": 7581
    },
    {
      "epoch": 3.5388564760793466,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001443697386730452,
      "loss": 0.021,
      "step": 7582
    },
    {
      "epoch": 3.539323220536756,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014435658816249047,
      "loss": 0.0247,
      "step": 7583
    },
    {
      "epoch": 3.5397899649941658,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00014434343669688916,
      "loss": 0.0252,
      "step": 7584
    },
    {
      "epoch": 3.540256709451575,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014433028427652446,
      "loss": 0.0309,
      "step": 7585
    },
    {
      "epoch": 3.540723453908985,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014431713090167959,
      "loss": 0.0333,
      "step": 7586
    },
    {
      "epoch": 3.5411901983663943,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014430397657263767,
      "loss": 0.0334,
      "step": 7587
    },
    {
      "epoch": 3.541656942823804,
      "grad_norm": 0.70703125,
      "learning_rate": 0.000144290821289682,
      "loss": 0.0283,
      "step": 7588
    },
    {
      "epoch": 3.5421236872812134,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00014427766505309585,
      "loss": 0.0259,
      "step": 7589
    },
    {
      "epoch": 3.542590431738623,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00014426450786316237,
      "loss": 0.0172,
      "step": 7590
    },
    {
      "epoch": 3.5430571761960326,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014425134972016496,
      "loss": 0.0261,
      "step": 7591
    },
    {
      "epoch": 3.5435239206534423,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001442381906243869,
      "loss": 0.0186,
      "step": 7592
    },
    {
      "epoch": 3.5439906651108517,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001442250305761115,
      "loss": 0.0284,
      "step": 7593
    },
    {
      "epoch": 3.5444574095682615,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001442118695756221,
      "loss": 0.027,
      "step": 7594
    },
    {
      "epoch": 3.544924154025671,
      "grad_norm": 0.609375,
      "learning_rate": 0.00014419870762320214,
      "loss": 0.018,
      "step": 7595
    },
    {
      "epoch": 3.5453908984830806,
      "grad_norm": 0.5625,
      "learning_rate": 0.00014418554471913494,
      "loss": 0.0314,
      "step": 7596
    },
    {
      "epoch": 3.54585764294049,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014417238086370397,
      "loss": 0.0243,
      "step": 7597
    },
    {
      "epoch": 3.5463243873979,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001441592160571926,
      "loss": 0.0294,
      "step": 7598
    },
    {
      "epoch": 3.546791131855309,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014414605029988432,
      "loss": 0.0241,
      "step": 7599
    },
    {
      "epoch": 3.547257876312719,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00014413288359206258,
      "loss": 0.0143,
      "step": 7600
    },
    {
      "epoch": 3.5477246207701283,
      "grad_norm": 0.97265625,
      "learning_rate": 0.0001441197159340109,
      "loss": 0.0208,
      "step": 7601
    },
    {
      "epoch": 3.548191365227538,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014410654732601282,
      "loss": 0.0182,
      "step": 7602
    },
    {
      "epoch": 3.5486581096849474,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014409337776835176,
      "loss": 0.0247,
      "step": 7603
    },
    {
      "epoch": 3.549124854142357,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014408020726131143,
      "loss": 0.0298,
      "step": 7604
    },
    {
      "epoch": 3.5495915985997666,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001440670358051753,
      "loss": 0.022,
      "step": 7605
    },
    {
      "epoch": 3.5500583430571764,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014405386340022696,
      "loss": 0.0193,
      "step": 7606
    },
    {
      "epoch": 3.5505250875145857,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001440406900467501,
      "loss": 0.0183,
      "step": 7607
    },
    {
      "epoch": 3.550991831971995,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014402751574502833,
      "loss": 0.0197,
      "step": 7608
    },
    {
      "epoch": 3.551458576429405,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001440143404953453,
      "loss": 0.0279,
      "step": 7609
    },
    {
      "epoch": 3.5519253208868147,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014400116429798466,
      "loss": 0.0292,
      "step": 7610
    },
    {
      "epoch": 3.552392065344224,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014398798715323013,
      "loss": 0.0235,
      "step": 7611
    },
    {
      "epoch": 3.5528588098016334,
      "grad_norm": 0.91796875,
      "learning_rate": 0.00014397480906136546,
      "loss": 0.0454,
      "step": 7612
    },
    {
      "epoch": 3.553325554259043,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014396163002267432,
      "loss": 0.0258,
      "step": 7613
    },
    {
      "epoch": 3.553792298716453,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014394845003744053,
      "loss": 0.0327,
      "step": 7614
    },
    {
      "epoch": 3.5542590431738623,
      "grad_norm": 1.03125,
      "learning_rate": 0.00014393526910594785,
      "loss": 0.0373,
      "step": 7615
    },
    {
      "epoch": 3.5547257876312717,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014392208722848008,
      "loss": 0.0298,
      "step": 7616
    },
    {
      "epoch": 3.5551925320886815,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014390890440532105,
      "loss": 0.019,
      "step": 7617
    },
    {
      "epoch": 3.5556592765460913,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001438957206367546,
      "loss": 0.0375,
      "step": 7618
    },
    {
      "epoch": 3.5561260210035006,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00014388253592306454,
      "loss": 0.0162,
      "step": 7619
    },
    {
      "epoch": 3.55659276546091,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001438693502645348,
      "loss": 0.0157,
      "step": 7620
    },
    {
      "epoch": 3.5570595099183198,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014385616366144931,
      "loss": 0.0206,
      "step": 7621
    },
    {
      "epoch": 3.5575262543757296,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00014384297611409194,
      "loss": 0.0226,
      "step": 7622
    },
    {
      "epoch": 3.557992998833139,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00014382978762274664,
      "loss": 0.0214,
      "step": 7623
    },
    {
      "epoch": 3.5584597432905483,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014381659818769738,
      "loss": 0.0252,
      "step": 7624
    },
    {
      "epoch": 3.558926487747958,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014380340780922816,
      "loss": 0.0221,
      "step": 7625
    },
    {
      "epoch": 3.5593932322053674,
      "grad_norm": 0.91015625,
      "learning_rate": 0.00014379021648762297,
      "loss": 0.0252,
      "step": 7626
    },
    {
      "epoch": 3.559859976662777,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014377702422316583,
      "loss": 0.0336,
      "step": 7627
    },
    {
      "epoch": 3.5603267211201866,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001437638310161408,
      "loss": 0.0125,
      "step": 7628
    },
    {
      "epoch": 3.5607934655775964,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00014375063686683187,
      "loss": 0.0376,
      "step": 7629
    },
    {
      "epoch": 3.5612602100350057,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014373744177552324,
      "loss": 0.0295,
      "step": 7630
    },
    {
      "epoch": 3.5617269544924155,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00014372424574249896,
      "loss": 0.0346,
      "step": 7631
    },
    {
      "epoch": 3.562193698949825,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014371104876804313,
      "loss": 0.0277,
      "step": 7632
    },
    {
      "epoch": 3.5626604434072346,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014369785085243994,
      "loss": 0.0292,
      "step": 7633
    },
    {
      "epoch": 3.563127187864644,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014368465199597356,
      "loss": 0.0207,
      "step": 7634
    },
    {
      "epoch": 3.563593932322054,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00014367145219892812,
      "loss": 0.0237,
      "step": 7635
    },
    {
      "epoch": 3.564060676779463,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014365825146158787,
      "loss": 0.0153,
      "step": 7636
    },
    {
      "epoch": 3.564527421236873,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014364504978423703,
      "loss": 0.0218,
      "step": 7637
    },
    {
      "epoch": 3.5649941656942823,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00014363184716715985,
      "loss": 0.0222,
      "step": 7638
    },
    {
      "epoch": 3.565460910151692,
      "grad_norm": 0.9375,
      "learning_rate": 0.00014361864361064056,
      "loss": 0.0316,
      "step": 7639
    },
    {
      "epoch": 3.5659276546091014,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014360543911496348,
      "loss": 0.029,
      "step": 7640
    },
    {
      "epoch": 3.5663943990665112,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00014359223368041297,
      "loss": 0.0272,
      "step": 7641
    },
    {
      "epoch": 3.5668611435239206,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00014357902730727327,
      "loss": 0.0366,
      "step": 7642
    },
    {
      "epoch": 3.5673278879813304,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014356581999582874,
      "loss": 0.0205,
      "step": 7643
    },
    {
      "epoch": 3.5677946324387397,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001435526117463638,
      "loss": 0.0216,
      "step": 7644
    },
    {
      "epoch": 3.5682613768961495,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014353940255916277,
      "loss": 0.0286,
      "step": 7645
    },
    {
      "epoch": 3.568728121353559,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014352619243451013,
      "loss": 0.0354,
      "step": 7646
    },
    {
      "epoch": 3.5691948658109682,
      "grad_norm": 0.8984375,
      "learning_rate": 0.00014351298137269027,
      "loss": 0.0377,
      "step": 7647
    },
    {
      "epoch": 3.569661610268378,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00014349976937398764,
      "loss": 0.0194,
      "step": 7648
    },
    {
      "epoch": 3.570128354725788,
      "grad_norm": 1.0390625,
      "learning_rate": 0.00014348655643868667,
      "loss": 0.0248,
      "step": 7649
    },
    {
      "epoch": 3.570595099183197,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014347334256707193,
      "loss": 0.0293,
      "step": 7650
    },
    {
      "epoch": 3.5710618436406065,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001434601277594279,
      "loss": 0.019,
      "step": 7651
    },
    {
      "epoch": 3.5715285880980163,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014344691201603906,
      "loss": 0.0238,
      "step": 7652
    },
    {
      "epoch": 3.571995332555426,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014343369533719003,
      "loss": 0.0217,
      "step": 7653
    },
    {
      "epoch": 3.5724620770128355,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014342047772316537,
      "loss": 0.0251,
      "step": 7654
    },
    {
      "epoch": 3.572928821470245,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001434072591742496,
      "loss": 0.0231,
      "step": 7655
    },
    {
      "epoch": 3.5733955659276546,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001433940396907274,
      "loss": 0.0253,
      "step": 7656
    },
    {
      "epoch": 3.5738623103850644,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001433808192728834,
      "loss": 0.0285,
      "step": 7657
    },
    {
      "epoch": 3.5743290548424738,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001433675979210022,
      "loss": 0.0216,
      "step": 7658
    },
    {
      "epoch": 3.574795799299883,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014335437563536848,
      "loss": 0.0197,
      "step": 7659
    },
    {
      "epoch": 3.575262543757293,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00014334115241626698,
      "loss": 0.0297,
      "step": 7660
    },
    {
      "epoch": 3.5757292882147027,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014332792826398237,
      "loss": 0.0307,
      "step": 7661
    },
    {
      "epoch": 3.576196032672112,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001433147031787994,
      "loss": 0.0462,
      "step": 7662
    },
    {
      "epoch": 3.5766627771295214,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014330147716100282,
      "loss": 0.0286,
      "step": 7663
    },
    {
      "epoch": 3.577129521586931,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014328825021087737,
      "loss": 0.0333,
      "step": 7664
    },
    {
      "epoch": 3.577596266044341,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00014327502232870787,
      "loss": 0.0168,
      "step": 7665
    },
    {
      "epoch": 3.5780630105017504,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00014326179351477914,
      "loss": 0.0206,
      "step": 7666
    },
    {
      "epoch": 3.5785297549591597,
      "grad_norm": 0.84375,
      "learning_rate": 0.000143248563769376,
      "loss": 0.0362,
      "step": 7667
    },
    {
      "epoch": 3.5789964994165695,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014323533309278328,
      "loss": 0.0223,
      "step": 7668
    },
    {
      "epoch": 3.579463243873979,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014322210148528586,
      "loss": 0.0261,
      "step": 7669
    },
    {
      "epoch": 3.5799299883313886,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014320886894716867,
      "loss": 0.0268,
      "step": 7670
    },
    {
      "epoch": 3.580396732788798,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014319563547871656,
      "loss": 0.0244,
      "step": 7671
    },
    {
      "epoch": 3.580863477246208,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014318240108021453,
      "loss": 0.0227,
      "step": 7672
    },
    {
      "epoch": 3.581330221703617,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014316916575194744,
      "loss": 0.0289,
      "step": 7673
    },
    {
      "epoch": 3.581796966161027,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001431559294942004,
      "loss": 0.0243,
      "step": 7674
    },
    {
      "epoch": 3.5822637106184363,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001431426923072582,
      "loss": 0.0215,
      "step": 7675
    },
    {
      "epoch": 3.582730455075846,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00014312945419140603,
      "loss": 0.037,
      "step": 7676
    },
    {
      "epoch": 3.5831971995332554,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014311621514692888,
      "loss": 0.0298,
      "step": 7677
    },
    {
      "epoch": 3.5836639439906652,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014310297517411172,
      "loss": 0.0289,
      "step": 7678
    },
    {
      "epoch": 3.5841306884480746,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014308973427323973,
      "loss": 0.0288,
      "step": 7679
    },
    {
      "epoch": 3.5845974329054844,
      "grad_norm": 0.8125,
      "learning_rate": 0.00014307649244459794,
      "loss": 0.0272,
      "step": 7680
    },
    {
      "epoch": 3.5850641773628937,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014306324968847145,
      "loss": 0.0191,
      "step": 7681
    },
    {
      "epoch": 3.5855309218203035,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014305000600514544,
      "loss": 0.0236,
      "step": 7682
    },
    {
      "epoch": 3.585997666277713,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00014303676139490503,
      "loss": 0.0161,
      "step": 7683
    },
    {
      "epoch": 3.5864644107351227,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001430235158580354,
      "loss": 0.0239,
      "step": 7684
    },
    {
      "epoch": 3.586931155192532,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014301026939482175,
      "loss": 0.0241,
      "step": 7685
    },
    {
      "epoch": 3.5873978996499414,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014299702200554925,
      "loss": 0.0214,
      "step": 7686
    },
    {
      "epoch": 3.587864644107351,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014298377369050317,
      "loss": 0.0242,
      "step": 7687
    },
    {
      "epoch": 3.588331388564761,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014297052444996875,
      "loss": 0.0196,
      "step": 7688
    },
    {
      "epoch": 3.5887981330221703,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014295727428423127,
      "loss": 0.0139,
      "step": 7689
    },
    {
      "epoch": 3.5892648774795797,
      "grad_norm": 0.73046875,
      "learning_rate": 0.000142944023193576,
      "loss": 0.0255,
      "step": 7690
    },
    {
      "epoch": 3.5897316219369895,
      "grad_norm": 0.85546875,
      "learning_rate": 0.0001429307711782883,
      "loss": 0.0286,
      "step": 7691
    },
    {
      "epoch": 3.5901983663943993,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00014291751823865343,
      "loss": 0.0306,
      "step": 7692
    },
    {
      "epoch": 3.5906651108518086,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00014290426437495678,
      "loss": 0.0226,
      "step": 7693
    },
    {
      "epoch": 3.591131855309218,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00014289100958748372,
      "loss": 0.016,
      "step": 7694
    },
    {
      "epoch": 3.5915985997666278,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014287775387651964,
      "loss": 0.0194,
      "step": 7695
    },
    {
      "epoch": 3.5920653442240376,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00014286449724234995,
      "loss": 0.0281,
      "step": 7696
    },
    {
      "epoch": 3.592532088681447,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014285123968526005,
      "loss": 0.0235,
      "step": 7697
    },
    {
      "epoch": 3.5929988331388563,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014283798120553543,
      "loss": 0.0281,
      "step": 7698
    },
    {
      "epoch": 3.593465577596266,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00014282472180346154,
      "loss": 0.0198,
      "step": 7699
    },
    {
      "epoch": 3.593932322053676,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001428114614793239,
      "loss": 0.0186,
      "step": 7700
    },
    {
      "epoch": 3.594399066511085,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014279820023340792,
      "loss": 0.0294,
      "step": 7701
    },
    {
      "epoch": 3.5948658109684946,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014278493806599928,
      "loss": 0.0253,
      "step": 7702
    },
    {
      "epoch": 3.5953325554259044,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014277167497738345,
      "loss": 0.0175,
      "step": 7703
    },
    {
      "epoch": 3.595799299883314,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00014275841096784595,
      "loss": 0.014,
      "step": 7704
    },
    {
      "epoch": 3.5962660443407235,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014274514603767246,
      "loss": 0.0276,
      "step": 7705
    },
    {
      "epoch": 3.596732788798133,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014273188018714853,
      "loss": 0.0191,
      "step": 7706
    },
    {
      "epoch": 3.5971995332555426,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014271861341655983,
      "loss": 0.0283,
      "step": 7707
    },
    {
      "epoch": 3.597666277712952,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00014270534572619197,
      "loss": 0.0228,
      "step": 7708
    },
    {
      "epoch": 3.598133022170362,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00014269207711633064,
      "loss": 0.0277,
      "step": 7709
    },
    {
      "epoch": 3.598599766627771,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001426788075872615,
      "loss": 0.0293,
      "step": 7710
    },
    {
      "epoch": 3.599066511085181,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001426655371392703,
      "loss": 0.0305,
      "step": 7711
    },
    {
      "epoch": 3.5995332555425903,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014265226577264277,
      "loss": 0.0267,
      "step": 7712
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.8046875,
      "learning_rate": 0.0001426389934876646,
      "loss": 0.0281,
      "step": 7713
    },
    {
      "epoch": 3.6004667444574094,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001426257202846216,
      "loss": 0.0246,
      "step": 7714
    },
    {
      "epoch": 3.6009334889148192,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014261244616379956,
      "loss": 0.0295,
      "step": 7715
    },
    {
      "epoch": 3.6014002333722286,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00014259917112548426,
      "loss": 0.026,
      "step": 7716
    },
    {
      "epoch": 3.6018669778296384,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00014258589516996154,
      "loss": 0.0239,
      "step": 7717
    },
    {
      "epoch": 3.6023337222870477,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014257261829751726,
      "loss": 0.0197,
      "step": 7718
    },
    {
      "epoch": 3.6028004667444575,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014255934050843728,
      "loss": 0.0212,
      "step": 7719
    },
    {
      "epoch": 3.603267211201867,
      "grad_norm": 0.578125,
      "learning_rate": 0.00014254606180300744,
      "loss": 0.0209,
      "step": 7720
    },
    {
      "epoch": 3.6037339556592767,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00014253278218151374,
      "loss": 0.0207,
      "step": 7721
    },
    {
      "epoch": 3.604200700116686,
      "grad_norm": 0.84765625,
      "learning_rate": 0.000142519501644242,
      "loss": 0.0331,
      "step": 7722
    },
    {
      "epoch": 3.604667444574096,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014250622019147826,
      "loss": 0.0179,
      "step": 7723
    },
    {
      "epoch": 3.605134189031505,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001424929378235084,
      "loss": 0.0234,
      "step": 7724
    },
    {
      "epoch": 3.605600933488915,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00014247965454061844,
      "loss": 0.0178,
      "step": 7725
    },
    {
      "epoch": 3.6060676779463243,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001424663703430944,
      "loss": 0.0259,
      "step": 7726
    },
    {
      "epoch": 3.606534422403734,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00014245308523122227,
      "loss": 0.0134,
      "step": 7727
    },
    {
      "epoch": 3.6070011668611435,
      "grad_norm": 0.703125,
      "learning_rate": 0.00014243979920528814,
      "loss": 0.0236,
      "step": 7728
    },
    {
      "epoch": 3.607467911318553,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014242651226557804,
      "loss": 0.0202,
      "step": 7729
    },
    {
      "epoch": 3.6079346557759626,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014241322441237804,
      "loss": 0.0243,
      "step": 7730
    },
    {
      "epoch": 3.6084014002333724,
      "grad_norm": 0.46875,
      "learning_rate": 0.00014239993564597428,
      "loss": 0.0095,
      "step": 7731
    },
    {
      "epoch": 3.6088681446907818,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00014238664596665285,
      "loss": 0.0182,
      "step": 7732
    },
    {
      "epoch": 3.609334889148191,
      "grad_norm": 0.796875,
      "learning_rate": 0.0001423733553746999,
      "loss": 0.0291,
      "step": 7733
    },
    {
      "epoch": 3.609801633605601,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001423600638704016,
      "loss": 0.0193,
      "step": 7734
    },
    {
      "epoch": 3.6102683780630107,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00014234677145404413,
      "loss": 0.0167,
      "step": 7735
    },
    {
      "epoch": 3.61073512252042,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001423334781259137,
      "loss": 0.0222,
      "step": 7736
    },
    {
      "epoch": 3.6112018669778294,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014232018388629647,
      "loss": 0.0173,
      "step": 7737
    },
    {
      "epoch": 3.611668611435239,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00014230688873547877,
      "loss": 0.0213,
      "step": 7738
    },
    {
      "epoch": 3.612135355892649,
      "grad_norm": 0.765625,
      "learning_rate": 0.0001422935926737468,
      "loss": 0.0198,
      "step": 7739
    },
    {
      "epoch": 3.6126021003500584,
      "grad_norm": 0.83984375,
      "learning_rate": 0.00014228029570138685,
      "loss": 0.0279,
      "step": 7740
    },
    {
      "epoch": 3.6130688448074677,
      "grad_norm": 0.546875,
      "learning_rate": 0.00014226699781868523,
      "loss": 0.0133,
      "step": 7741
    },
    {
      "epoch": 3.6135355892648775,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00014225369902592827,
      "loss": 0.0163,
      "step": 7742
    },
    {
      "epoch": 3.6140023337222873,
      "grad_norm": 0.78125,
      "learning_rate": 0.00014224039932340225,
      "loss": 0.0308,
      "step": 7743
    },
    {
      "epoch": 3.6144690781796966,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00014222709871139358,
      "loss": 0.0175,
      "step": 7744
    },
    {
      "epoch": 3.614935822637106,
      "grad_norm": 0.5625,
      "learning_rate": 0.00014221379719018865,
      "loss": 0.0178,
      "step": 7745
    },
    {
      "epoch": 3.615402567094516,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014220049476007377,
      "loss": 0.0154,
      "step": 7746
    },
    {
      "epoch": 3.6158693115519256,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001421871914213355,
      "loss": 0.0223,
      "step": 7747
    },
    {
      "epoch": 3.616336056009335,
      "grad_norm": 1.0625,
      "learning_rate": 0.00014217388717426013,
      "loss": 0.04,
      "step": 7748
    },
    {
      "epoch": 3.6168028004667443,
      "grad_norm": 0.5,
      "learning_rate": 0.00014216058201913418,
      "loss": 0.0147,
      "step": 7749
    },
    {
      "epoch": 3.617269544924154,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001421472759562441,
      "loss": 0.0245,
      "step": 7750
    },
    {
      "epoch": 3.6177362893815634,
      "grad_norm": 0.609375,
      "learning_rate": 0.00014213396898587644,
      "loss": 0.0206,
      "step": 7751
    },
    {
      "epoch": 3.6182030338389732,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001421206611083177,
      "loss": 0.0255,
      "step": 7752
    },
    {
      "epoch": 3.6186697782963826,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00014210735232385435,
      "loss": 0.0126,
      "step": 7753
    },
    {
      "epoch": 3.6191365227537924,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014209404263277302,
      "loss": 0.0209,
      "step": 7754
    },
    {
      "epoch": 3.6196032672112017,
      "grad_norm": 0.828125,
      "learning_rate": 0.00014208073203536022,
      "loss": 0.0267,
      "step": 7755
    },
    {
      "epoch": 3.6200700116686115,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00014206742053190257,
      "loss": 0.0443,
      "step": 7756
    },
    {
      "epoch": 3.620536756126021,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001420541081226867,
      "loss": 0.0222,
      "step": 7757
    },
    {
      "epoch": 3.6210035005834307,
      "grad_norm": 0.53125,
      "learning_rate": 0.00014204079480799924,
      "loss": 0.018,
      "step": 7758
    },
    {
      "epoch": 3.62147024504084,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00014202748058812678,
      "loss": 0.0208,
      "step": 7759
    },
    {
      "epoch": 3.62193698949825,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014201416546335603,
      "loss": 0.0233,
      "step": 7760
    },
    {
      "epoch": 3.622403733955659,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00014200084943397372,
      "loss": 0.0187,
      "step": 7761
    },
    {
      "epoch": 3.622870478413069,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001419875325002665,
      "loss": 0.021,
      "step": 7762
    },
    {
      "epoch": 3.6233372228704783,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001419742146625211,
      "loss": 0.0188,
      "step": 7763
    },
    {
      "epoch": 3.623803967327888,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00014196089592102436,
      "loss": 0.0208,
      "step": 7764
    },
    {
      "epoch": 3.6242707117852975,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001419475762760629,
      "loss": 0.0246,
      "step": 7765
    },
    {
      "epoch": 3.6247374562427073,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001419342557279236,
      "loss": 0.0121,
      "step": 7766
    },
    {
      "epoch": 3.6252042007001166,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00014192093427689326,
      "loss": 0.0224,
      "step": 7767
    },
    {
      "epoch": 3.6256709451575264,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001419076119232587,
      "loss": 0.0238,
      "step": 7768
    },
    {
      "epoch": 3.6261376896149358,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001418942886673067,
      "loss": 0.0124,
      "step": 7769
    },
    {
      "epoch": 3.6266044340723456,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014188096450932423,
      "loss": 0.03,
      "step": 7770
    },
    {
      "epoch": 3.627071178529755,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014186763944959815,
      "loss": 0.0278,
      "step": 7771
    },
    {
      "epoch": 3.6275379229871643,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001418543134884153,
      "loss": 0.0164,
      "step": 7772
    },
    {
      "epoch": 3.628004667444574,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014184098662606265,
      "loss": 0.037,
      "step": 7773
    },
    {
      "epoch": 3.628471411901984,
      "grad_norm": 0.71875,
      "learning_rate": 0.00014182765886282717,
      "loss": 0.0176,
      "step": 7774
    },
    {
      "epoch": 3.628938156359393,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014181433019899576,
      "loss": 0.0206,
      "step": 7775
    },
    {
      "epoch": 3.6294049008168026,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001418010006348554,
      "loss": 0.0181,
      "step": 7776
    },
    {
      "epoch": 3.6298716452742124,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014178767017069315,
      "loss": 0.015,
      "step": 7777
    },
    {
      "epoch": 3.630338389731622,
      "grad_norm": 0.703125,
      "learning_rate": 0.000141774338806796,
      "loss": 0.0206,
      "step": 7778
    },
    {
      "epoch": 3.6308051341890315,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00014176100654345093,
      "loss": 0.0162,
      "step": 7779
    },
    {
      "epoch": 3.631271878646441,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001417476733809451,
      "loss": 0.0217,
      "step": 7780
    },
    {
      "epoch": 3.6317386231038506,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00014173433931956555,
      "loss": 0.0241,
      "step": 7781
    },
    {
      "epoch": 3.6322053675612604,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00014172100435959936,
      "loss": 0.0221,
      "step": 7782
    },
    {
      "epoch": 3.63267211201867,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00014170766850133364,
      "loss": 0.0153,
      "step": 7783
    },
    {
      "epoch": 3.633138856476079,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00014169433174505554,
      "loss": 0.0146,
      "step": 7784
    },
    {
      "epoch": 3.633605600933489,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014168099409105227,
      "loss": 0.0265,
      "step": 7785
    },
    {
      "epoch": 3.6340723453908987,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001416676555396109,
      "loss": 0.0224,
      "step": 7786
    },
    {
      "epoch": 3.634539089848308,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014165431609101868,
      "loss": 0.0212,
      "step": 7787
    },
    {
      "epoch": 3.6350058343057174,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014164097574556286,
      "loss": 0.021,
      "step": 7788
    },
    {
      "epoch": 3.6354725787631272,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00014162763450353057,
      "loss": 0.0296,
      "step": 7789
    },
    {
      "epoch": 3.6359393232205366,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00014161429236520917,
      "loss": 0.022,
      "step": 7790
    },
    {
      "epoch": 3.6364060676779464,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00014160094933088585,
      "loss": 0.0209,
      "step": 7791
    },
    {
      "epoch": 3.6368728121353557,
      "grad_norm": 0.765625,
      "learning_rate": 0.00014158760540084795,
      "loss": 0.0265,
      "step": 7792
    },
    {
      "epoch": 3.6373395565927655,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014157426057538277,
      "loss": 0.0264,
      "step": 7793
    },
    {
      "epoch": 3.637806301050175,
      "grad_norm": 0.75,
      "learning_rate": 0.00014156091485477762,
      "loss": 0.0231,
      "step": 7794
    },
    {
      "epoch": 3.6382730455075847,
      "grad_norm": 0.34375,
      "learning_rate": 0.00014154756823931986,
      "loss": 0.0125,
      "step": 7795
    },
    {
      "epoch": 3.638739789964994,
      "grad_norm": 1.09375,
      "learning_rate": 0.00014153422072929686,
      "loss": 0.0383,
      "step": 7796
    },
    {
      "epoch": 3.639206534422404,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014152087232499602,
      "loss": 0.0251,
      "step": 7797
    },
    {
      "epoch": 3.639673278879813,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001415075230267047,
      "loss": 0.0225,
      "step": 7798
    },
    {
      "epoch": 3.640140023337223,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001414941728347104,
      "loss": 0.0235,
      "step": 7799
    },
    {
      "epoch": 3.6406067677946323,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014148082174930048,
      "loss": 0.0297,
      "step": 7800
    },
    {
      "epoch": 3.641073512252042,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014146746977076246,
      "loss": 0.0199,
      "step": 7801
    },
    {
      "epoch": 3.6415402567094515,
      "grad_norm": 0.625,
      "learning_rate": 0.0001414541168993838,
      "loss": 0.0236,
      "step": 7802
    },
    {
      "epoch": 3.6420070011668613,
      "grad_norm": 0.419921875,
      "learning_rate": 0.000141440763135452,
      "loss": 0.0112,
      "step": 7803
    },
    {
      "epoch": 3.6424737456242706,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001414274084792546,
      "loss": 0.0182,
      "step": 7804
    },
    {
      "epoch": 3.6429404900816804,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014141405293107914,
      "loss": 0.0336,
      "step": 7805
    },
    {
      "epoch": 3.6434072345390898,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014140069649121317,
      "loss": 0.0256,
      "step": 7806
    },
    {
      "epoch": 3.6438739789964996,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00014138733915994428,
      "loss": 0.0129,
      "step": 7807
    },
    {
      "epoch": 3.644340723453909,
      "grad_norm": 0.75,
      "learning_rate": 0.00014137398093756005,
      "loss": 0.0322,
      "step": 7808
    },
    {
      "epoch": 3.6448074679113187,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014136062182434808,
      "loss": 0.0277,
      "step": 7809
    },
    {
      "epoch": 3.645274212368728,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014134726182059605,
      "loss": 0.0218,
      "step": 7810
    },
    {
      "epoch": 3.6457409568261374,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001413339009265916,
      "loss": 0.0207,
      "step": 7811
    },
    {
      "epoch": 3.646207701283547,
      "grad_norm": 0.59765625,
      "learning_rate": 0.0001413205391426224,
      "loss": 0.0234,
      "step": 7812
    },
    {
      "epoch": 3.646674445740957,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00014130717646897614,
      "loss": 0.0224,
      "step": 7813
    },
    {
      "epoch": 3.6471411901983664,
      "grad_norm": 0.734375,
      "learning_rate": 0.00014129381290594058,
      "loss": 0.0245,
      "step": 7814
    },
    {
      "epoch": 3.6476079346557757,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00014128044845380334,
      "loss": 0.0208,
      "step": 7815
    },
    {
      "epoch": 3.6480746791131855,
      "grad_norm": 0.88671875,
      "learning_rate": 0.0001412670831128523,
      "loss": 0.0234,
      "step": 7816
    },
    {
      "epoch": 3.6485414235705953,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00014125371688337516,
      "loss": 0.027,
      "step": 7817
    },
    {
      "epoch": 3.6490081680280047,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001412403497656597,
      "loss": 0.0286,
      "step": 7818
    },
    {
      "epoch": 3.649474912485414,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00014122698175999377,
      "loss": 0.0286,
      "step": 7819
    },
    {
      "epoch": 3.649941656942824,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001412136128666652,
      "loss": 0.0241,
      "step": 7820
    },
    {
      "epoch": 3.6504084014002336,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001412002430859618,
      "loss": 0.0217,
      "step": 7821
    },
    {
      "epoch": 3.650875145857643,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00014118687241817144,
      "loss": 0.0178,
      "step": 7822
    },
    {
      "epoch": 3.6513418903150523,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00014117350086358203,
      "loss": 0.0235,
      "step": 7823
    },
    {
      "epoch": 3.651808634772462,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001411601284224815,
      "loss": 0.0192,
      "step": 7824
    },
    {
      "epoch": 3.652275379229872,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00014114675509515766,
      "loss": 0.0265,
      "step": 7825
    },
    {
      "epoch": 3.6527421236872812,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001411333808818986,
      "loss": 0.0211,
      "step": 7826
    },
    {
      "epoch": 3.6532088681446906,
      "grad_norm": 1.15625,
      "learning_rate": 0.00014112000578299217,
      "loss": 0.027,
      "step": 7827
    },
    {
      "epoch": 3.6536756126021004,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001411066297987264,
      "loss": 0.0204,
      "step": 7828
    },
    {
      "epoch": 3.65414235705951,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001410932529293893,
      "loss": 0.0226,
      "step": 7829
    },
    {
      "epoch": 3.6546091015169195,
      "grad_norm": 0.796875,
      "learning_rate": 0.00014107987517526885,
      "loss": 0.0223,
      "step": 7830
    },
    {
      "epoch": 3.655075845974329,
      "grad_norm": 0.953125,
      "learning_rate": 0.00014106649653665312,
      "loss": 0.0308,
      "step": 7831
    },
    {
      "epoch": 3.6555425904317387,
      "grad_norm": 0.625,
      "learning_rate": 0.00014105311701383014,
      "loss": 0.0141,
      "step": 7832
    },
    {
      "epoch": 3.656009334889148,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014103973660708805,
      "loss": 0.0196,
      "step": 7833
    },
    {
      "epoch": 3.656476079346558,
      "grad_norm": 0.640625,
      "learning_rate": 0.00014102635531671484,
      "loss": 0.018,
      "step": 7834
    },
    {
      "epoch": 3.656942823803967,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00014101297314299868,
      "loss": 0.0257,
      "step": 7835
    },
    {
      "epoch": 3.657409568261377,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00014099959008622776,
      "loss": 0.0127,
      "step": 7836
    },
    {
      "epoch": 3.6578763127187863,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014098620614669017,
      "loss": 0.0186,
      "step": 7837
    },
    {
      "epoch": 3.658343057176196,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00014097282132467402,
      "loss": 0.0218,
      "step": 7838
    },
    {
      "epoch": 3.6588098016336055,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00014095943562046764,
      "loss": 0.0124,
      "step": 7839
    },
    {
      "epoch": 3.6592765460910153,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001409460490343591,
      "loss": 0.0186,
      "step": 7840
    },
    {
      "epoch": 3.6597432905484246,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001409326615666367,
      "loss": 0.0294,
      "step": 7841
    },
    {
      "epoch": 3.6602100350058344,
      "grad_norm": 0.5625,
      "learning_rate": 0.00014091927321758873,
      "loss": 0.0151,
      "step": 7842
    },
    {
      "epoch": 3.6606767794632438,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001409058839875034,
      "loss": 0.0316,
      "step": 7843
    },
    {
      "epoch": 3.6611435239206536,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014089249387666896,
      "loss": 0.0248,
      "step": 7844
    },
    {
      "epoch": 3.661610268378063,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014087910288537378,
      "loss": 0.028,
      "step": 7845
    },
    {
      "epoch": 3.6620770128354727,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00014086571101390617,
      "loss": 0.0172,
      "step": 7846
    },
    {
      "epoch": 3.662543757292882,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00014085231826255445,
      "loss": 0.017,
      "step": 7847
    },
    {
      "epoch": 3.663010501750292,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00014083892463160697,
      "loss": 0.021,
      "step": 7848
    },
    {
      "epoch": 3.663477246207701,
      "grad_norm": 0.609375,
      "learning_rate": 0.00014082553012135216,
      "loss": 0.0181,
      "step": 7849
    },
    {
      "epoch": 3.663943990665111,
      "grad_norm": 0.96875,
      "learning_rate": 0.0001408121347320784,
      "loss": 0.0269,
      "step": 7850
    },
    {
      "epoch": 3.6644107351225204,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00014079873846407405,
      "loss": 0.0235,
      "step": 7851
    },
    {
      "epoch": 3.66487747957993,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001407853413176276,
      "loss": 0.0224,
      "step": 7852
    },
    {
      "epoch": 3.6653442240373395,
      "grad_norm": 0.578125,
      "learning_rate": 0.00014077194329302755,
      "loss": 0.0244,
      "step": 7853
    },
    {
      "epoch": 3.665810968494749,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014075854439056228,
      "loss": 0.0209,
      "step": 7854
    },
    {
      "epoch": 3.6662777129521587,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00014074514461052034,
      "loss": 0.0238,
      "step": 7855
    },
    {
      "epoch": 3.6667444574095684,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00014073174395319024,
      "loss": 0.0166,
      "step": 7856
    },
    {
      "epoch": 3.667211201866978,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001407183424188605,
      "loss": 0.0225,
      "step": 7857
    },
    {
      "epoch": 3.667677946324387,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014070494000781967,
      "loss": 0.02,
      "step": 7858
    },
    {
      "epoch": 3.668144690781797,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00014069153672035628,
      "loss": 0.0243,
      "step": 7859
    },
    {
      "epoch": 3.6686114352392067,
      "grad_norm": 0.5390625,
      "learning_rate": 0.000140678132556759,
      "loss": 0.0147,
      "step": 7860
    },
    {
      "epoch": 3.669078179696616,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001406647275173164,
      "loss": 0.0196,
      "step": 7861
    },
    {
      "epoch": 3.6695449241540254,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001406513216023171,
      "loss": 0.023,
      "step": 7862
    },
    {
      "epoch": 3.6700116686114352,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001406379148120497,
      "loss": 0.0251,
      "step": 7863
    },
    {
      "epoch": 3.670478413068845,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001406245071468029,
      "loss": 0.0264,
      "step": 7864
    },
    {
      "epoch": 3.6709451575262544,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014061109860686544,
      "loss": 0.0274,
      "step": 7865
    },
    {
      "epoch": 3.6714119019836637,
      "grad_norm": 0.671875,
      "learning_rate": 0.00014059768919252595,
      "loss": 0.0312,
      "step": 7866
    },
    {
      "epoch": 3.6718786464410735,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00014058427890407312,
      "loss": 0.0184,
      "step": 7867
    },
    {
      "epoch": 3.6723453908984833,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001405708677417958,
      "loss": 0.0221,
      "step": 7868
    },
    {
      "epoch": 3.6728121353558927,
      "grad_norm": 0.93359375,
      "learning_rate": 0.00014055745570598264,
      "loss": 0.0448,
      "step": 7869
    },
    {
      "epoch": 3.673278879813302,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001405440427969225,
      "loss": 0.0198,
      "step": 7870
    },
    {
      "epoch": 3.673745624270712,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014053062901490409,
      "loss": 0.026,
      "step": 7871
    },
    {
      "epoch": 3.6742123687281216,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001405172143602163,
      "loss": 0.0142,
      "step": 7872
    },
    {
      "epoch": 3.674679113185531,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00014050379883314793,
      "loss": 0.0276,
      "step": 7873
    },
    {
      "epoch": 3.6751458576429403,
      "grad_norm": 0.890625,
      "learning_rate": 0.0001404903824339878,
      "loss": 0.0269,
      "step": 7874
    },
    {
      "epoch": 3.67561260210035,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00014047696516302483,
      "loss": 0.0176,
      "step": 7875
    },
    {
      "epoch": 3.6760793465577595,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001404635470205479,
      "loss": 0.0332,
      "step": 7876
    },
    {
      "epoch": 3.6765460910151693,
      "grad_norm": 0.83203125,
      "learning_rate": 0.0001404501280068459,
      "loss": 0.0219,
      "step": 7877
    },
    {
      "epoch": 3.6770128354725786,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00014043670812220778,
      "loss": 0.0266,
      "step": 7878
    },
    {
      "epoch": 3.6774795799299884,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00014042328736692245,
      "loss": 0.0249,
      "step": 7879
    },
    {
      "epoch": 3.6779463243873978,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001404098657412789,
      "loss": 0.0265,
      "step": 7880
    },
    {
      "epoch": 3.6784130688448076,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00014039644324556608,
      "loss": 0.0246,
      "step": 7881
    },
    {
      "epoch": 3.678879813302217,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00014038301988007304,
      "loss": 0.0235,
      "step": 7882
    },
    {
      "epoch": 3.6793465577596267,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00014036959564508878,
      "loss": 0.0197,
      "step": 7883
    },
    {
      "epoch": 3.679813302217036,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00014035617054090232,
      "loss": 0.0151,
      "step": 7884
    },
    {
      "epoch": 3.680280046674446,
      "grad_norm": 0.6875,
      "learning_rate": 0.00014034274456780276,
      "loss": 0.0223,
      "step": 7885
    },
    {
      "epoch": 3.680746791131855,
      "grad_norm": 0.65625,
      "learning_rate": 0.00014032931772607913,
      "loss": 0.0244,
      "step": 7886
    },
    {
      "epoch": 3.681213535589265,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00014031589001602053,
      "loss": 0.0197,
      "step": 7887
    },
    {
      "epoch": 3.6816802800466744,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001403024614379161,
      "loss": 0.0239,
      "step": 7888
    },
    {
      "epoch": 3.682147024504084,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00014028903199205496,
      "loss": 0.0136,
      "step": 7889
    },
    {
      "epoch": 3.6826137689614935,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014027560167872625,
      "loss": 0.0217,
      "step": 7890
    },
    {
      "epoch": 3.6830805134189033,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00014026217049821916,
      "loss": 0.0264,
      "step": 7891
    },
    {
      "epoch": 3.6835472578763127,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00014024873845082286,
      "loss": 0.0151,
      "step": 7892
    },
    {
      "epoch": 3.6840140023337224,
      "grad_norm": 0.75,
      "learning_rate": 0.00014023530553682655,
      "loss": 0.0322,
      "step": 7893
    },
    {
      "epoch": 3.684480746791132,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001402218717565195,
      "loss": 0.022,
      "step": 7894
    },
    {
      "epoch": 3.6849474912485416,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001402084371101909,
      "loss": 0.0227,
      "step": 7895
    },
    {
      "epoch": 3.685414235705951,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00014019500159813007,
      "loss": 0.0183,
      "step": 7896
    },
    {
      "epoch": 3.6858809801633603,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014018156522062624,
      "loss": 0.0203,
      "step": 7897
    },
    {
      "epoch": 3.68634772462077,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00014016812797796872,
      "loss": 0.0183,
      "step": 7898
    },
    {
      "epoch": 3.68681446907818,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00014015468987044685,
      "loss": 0.0269,
      "step": 7899
    },
    {
      "epoch": 3.6872812135355892,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00014014125089834993,
      "loss": 0.0275,
      "step": 7900
    },
    {
      "epoch": 3.6877479579929986,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00014012781106196738,
      "loss": 0.0213,
      "step": 7901
    },
    {
      "epoch": 3.6882147024504084,
      "grad_norm": 0.94140625,
      "learning_rate": 0.0001401143703615885,
      "loss": 0.0242,
      "step": 7902
    },
    {
      "epoch": 3.688681446907818,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001401009287975027,
      "loss": 0.021,
      "step": 7903
    },
    {
      "epoch": 3.6891481913652275,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00014008748636999945,
      "loss": 0.0182,
      "step": 7904
    },
    {
      "epoch": 3.689614935822637,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00014007404307936816,
      "loss": 0.0105,
      "step": 7905
    },
    {
      "epoch": 3.6900816802800467,
      "grad_norm": 0.5625,
      "learning_rate": 0.00014006059892589825,
      "loss": 0.0176,
      "step": 7906
    },
    {
      "epoch": 3.6905484247374565,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00014004715390987917,
      "loss": 0.0262,
      "step": 7907
    },
    {
      "epoch": 3.691015169194866,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00014003370803160046,
      "loss": 0.016,
      "step": 7908
    },
    {
      "epoch": 3.691481913652275,
      "grad_norm": 0.59375,
      "learning_rate": 0.00014002026129135157,
      "loss": 0.025,
      "step": 7909
    },
    {
      "epoch": 3.691948658109685,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00014000681368942205,
      "loss": 0.0309,
      "step": 7910
    },
    {
      "epoch": 3.6924154025670948,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001399933652261015,
      "loss": 0.0183,
      "step": 7911
    },
    {
      "epoch": 3.692882147024504,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013997991590167935,
      "loss": 0.026,
      "step": 7912
    },
    {
      "epoch": 3.6933488914819135,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013996646571644527,
      "loss": 0.0246,
      "step": 7913
    },
    {
      "epoch": 3.6938156359393233,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013995301467068885,
      "loss": 0.0301,
      "step": 7914
    },
    {
      "epoch": 3.6942823803967326,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013993956276469973,
      "loss": 0.0267,
      "step": 7915
    },
    {
      "epoch": 3.6947491248541424,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013992610999876746,
      "loss": 0.0173,
      "step": 7916
    },
    {
      "epoch": 3.6952158693115518,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013991265637318176,
      "loss": 0.0199,
      "step": 7917
    },
    {
      "epoch": 3.6956826137689616,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00013989920188823228,
      "loss": 0.0178,
      "step": 7918
    },
    {
      "epoch": 3.696149358226371,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013988574654420871,
      "loss": 0.0266,
      "step": 7919
    },
    {
      "epoch": 3.6966161026837807,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013987229034140077,
      "loss": 0.0183,
      "step": 7920
    },
    {
      "epoch": 3.69708284714119,
      "grad_norm": 0.9453125,
      "learning_rate": 0.0001398588332800982,
      "loss": 0.0347,
      "step": 7921
    },
    {
      "epoch": 3.6975495915986,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00013984537536059071,
      "loss": 0.0323,
      "step": 7922
    },
    {
      "epoch": 3.698016336056009,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013983191658316804,
      "loss": 0.0212,
      "step": 7923
    },
    {
      "epoch": 3.698483080513419,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013981845694812007,
      "loss": 0.0248,
      "step": 7924
    },
    {
      "epoch": 3.6989498249708284,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013980499645573652,
      "loss": 0.0252,
      "step": 7925
    },
    {
      "epoch": 3.699416569428238,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013979153510630722,
      "loss": 0.0219,
      "step": 7926
    },
    {
      "epoch": 3.6998833138856475,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013977807290012204,
      "loss": 0.0178,
      "step": 7927
    },
    {
      "epoch": 3.7003500583430573,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001397646098374708,
      "loss": 0.028,
      "step": 7928
    },
    {
      "epoch": 3.7008168028004667,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001397511459186434,
      "loss": 0.0212,
      "step": 7929
    },
    {
      "epoch": 3.7012835472578764,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013973768114392972,
      "loss": 0.0192,
      "step": 7930
    },
    {
      "epoch": 3.701750291715286,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00013972421551361968,
      "loss": 0.0145,
      "step": 7931
    },
    {
      "epoch": 3.7022170361726956,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001397107490280032,
      "loss": 0.0183,
      "step": 7932
    },
    {
      "epoch": 3.702683780630105,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013969728168737025,
      "loss": 0.0181,
      "step": 7933
    },
    {
      "epoch": 3.7031505250875147,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013968381349201077,
      "loss": 0.0221,
      "step": 7934
    },
    {
      "epoch": 3.703617269544924,
      "grad_norm": 0.84375,
      "learning_rate": 0.00013967034444221474,
      "loss": 0.0263,
      "step": 7935
    },
    {
      "epoch": 3.7040840140023334,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001396568745382722,
      "loss": 0.0295,
      "step": 7936
    },
    {
      "epoch": 3.7045507584597432,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013964340378047315,
      "loss": 0.0174,
      "step": 7937
    },
    {
      "epoch": 3.705017502917153,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001396299321691076,
      "loss": 0.0286,
      "step": 7938
    },
    {
      "epoch": 3.7054842473745624,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013961645970446567,
      "loss": 0.0186,
      "step": 7939
    },
    {
      "epoch": 3.7059509918319717,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001396029863868374,
      "loss": 0.0202,
      "step": 7940
    },
    {
      "epoch": 3.7064177362893815,
      "grad_norm": 0.65234375,
      "learning_rate": 0.0001395895122165129,
      "loss": 0.0192,
      "step": 7941
    },
    {
      "epoch": 3.7068844807467913,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013957603719378224,
      "loss": 0.0258,
      "step": 7942
    },
    {
      "epoch": 3.7073512252042007,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00013956256131893564,
      "loss": 0.0225,
      "step": 7943
    },
    {
      "epoch": 3.70781796966161,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001395490845922632,
      "loss": 0.0236,
      "step": 7944
    },
    {
      "epoch": 3.70828471411902,
      "grad_norm": 0.75,
      "learning_rate": 0.00013953560701405505,
      "loss": 0.0327,
      "step": 7945
    },
    {
      "epoch": 3.7087514585764296,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013952212858460143,
      "loss": 0.0344,
      "step": 7946
    },
    {
      "epoch": 3.709218203033839,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013950864930419255,
      "loss": 0.0198,
      "step": 7947
    },
    {
      "epoch": 3.7096849474912483,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013949516917311858,
      "loss": 0.0144,
      "step": 7948
    },
    {
      "epoch": 3.710151691948658,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001394816881916698,
      "loss": 0.0179,
      "step": 7949
    },
    {
      "epoch": 3.710618436406068,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013946820636013649,
      "loss": 0.017,
      "step": 7950
    },
    {
      "epoch": 3.7110851808634773,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001394547236788089,
      "loss": 0.0202,
      "step": 7951
    },
    {
      "epoch": 3.7115519253208866,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00013944124014797732,
      "loss": 0.0184,
      "step": 7952
    },
    {
      "epoch": 3.7120186697782964,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001394277557679321,
      "loss": 0.0203,
      "step": 7953
    },
    {
      "epoch": 3.712485414235706,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013941427053896358,
      "loss": 0.0183,
      "step": 7954
    },
    {
      "epoch": 3.7129521586931156,
      "grad_norm": 0.9296875,
      "learning_rate": 0.00013940078446136203,
      "loss": 0.017,
      "step": 7955
    },
    {
      "epoch": 3.713418903150525,
      "grad_norm": 0.80859375,
      "learning_rate": 0.0001393872975354179,
      "loss": 0.028,
      "step": 7956
    },
    {
      "epoch": 3.7138856476079347,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013937380976142153,
      "loss": 0.0185,
      "step": 7957
    },
    {
      "epoch": 3.714352392065344,
      "grad_norm": 0.828125,
      "learning_rate": 0.00013936032113966339,
      "loss": 0.0277,
      "step": 7958
    },
    {
      "epoch": 3.714819136522754,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013934683167043382,
      "loss": 0.0284,
      "step": 7959
    },
    {
      "epoch": 3.715285880980163,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013933334135402334,
      "loss": 0.0246,
      "step": 7960
    },
    {
      "epoch": 3.715752625437573,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013931985019072237,
      "loss": 0.0144,
      "step": 7961
    },
    {
      "epoch": 3.7162193698949824,
      "grad_norm": 0.625,
      "learning_rate": 0.00013930635818082139,
      "loss": 0.0217,
      "step": 7962
    },
    {
      "epoch": 3.716686114352392,
      "grad_norm": 0.984375,
      "learning_rate": 0.0001392928653246109,
      "loss": 0.0235,
      "step": 7963
    },
    {
      "epoch": 3.7171528588098015,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013927937162238146,
      "loss": 0.0267,
      "step": 7964
    },
    {
      "epoch": 3.7176196032672113,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013926587707442352,
      "loss": 0.016,
      "step": 7965
    },
    {
      "epoch": 3.7180863477246207,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001392523816810277,
      "loss": 0.0212,
      "step": 7966
    },
    {
      "epoch": 3.7185530921820305,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00013923888544248458,
      "loss": 0.0175,
      "step": 7967
    },
    {
      "epoch": 3.71901983663944,
      "grad_norm": 0.91796875,
      "learning_rate": 0.0001392253883590847,
      "loss": 0.0244,
      "step": 7968
    },
    {
      "epoch": 3.7194865810968496,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013921189043111868,
      "loss": 0.0238,
      "step": 7969
    },
    {
      "epoch": 3.719953325554259,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013919839165887717,
      "loss": 0.0221,
      "step": 7970
    },
    {
      "epoch": 3.7204200700116687,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013918489204265078,
      "loss": 0.0191,
      "step": 7971
    },
    {
      "epoch": 3.720886814469078,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001391713915827302,
      "loss": 0.034,
      "step": 7972
    },
    {
      "epoch": 3.721353558926488,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013915789027940607,
      "loss": 0.0148,
      "step": 7973
    },
    {
      "epoch": 3.7218203033838972,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013914438813296913,
      "loss": 0.0215,
      "step": 7974
    },
    {
      "epoch": 3.722287047841307,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001391308851437101,
      "loss": 0.0189,
      "step": 7975
    },
    {
      "epoch": 3.7227537922987164,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013911738131191966,
      "loss": 0.0212,
      "step": 7976
    },
    {
      "epoch": 3.723220536756126,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013910387663788865,
      "loss": 0.0205,
      "step": 7977
    },
    {
      "epoch": 3.7236872812135355,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013909037112190773,
      "loss": 0.0117,
      "step": 7978
    },
    {
      "epoch": 3.724154025670945,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001390768647642678,
      "loss": 0.027,
      "step": 7979
    },
    {
      "epoch": 3.7246207701283547,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001390633575652596,
      "loss": 0.0168,
      "step": 7980
    },
    {
      "epoch": 3.7250875145857645,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013904984952517398,
      "loss": 0.0153,
      "step": 7981
    },
    {
      "epoch": 3.725554259043174,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013903634064430176,
      "loss": 0.0154,
      "step": 7982
    },
    {
      "epoch": 3.726021003500583,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001390228309229338,
      "loss": 0.0171,
      "step": 7983
    },
    {
      "epoch": 3.726487747957993,
      "grad_norm": 0.7578125,
      "learning_rate": 0.000139009320361361,
      "loss": 0.0274,
      "step": 7984
    },
    {
      "epoch": 3.7269544924154028,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001389958089598743,
      "loss": 0.0265,
      "step": 7985
    },
    {
      "epoch": 3.727421236872812,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013898229671876447,
      "loss": 0.014,
      "step": 7986
    },
    {
      "epoch": 3.7278879813302215,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001389687836383226,
      "loss": 0.0167,
      "step": 7987
    },
    {
      "epoch": 3.7283547257876313,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001389552697188396,
      "loss": 0.02,
      "step": 7988
    },
    {
      "epoch": 3.728821470245041,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013894175496060638,
      "loss": 0.0176,
      "step": 7989
    },
    {
      "epoch": 3.7292882147024504,
      "grad_norm": 0.6875,
      "learning_rate": 0.000138928239363914,
      "loss": 0.0272,
      "step": 7990
    },
    {
      "epoch": 3.7297549591598598,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013891472292905342,
      "loss": 0.0263,
      "step": 7991
    },
    {
      "epoch": 3.7302217036172696,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013890120565631566,
      "loss": 0.0192,
      "step": 7992
    },
    {
      "epoch": 3.7306884480746794,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013888768754599182,
      "loss": 0.0245,
      "step": 7993
    },
    {
      "epoch": 3.7311551925320887,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001388741685983729,
      "loss": 0.028,
      "step": 7994
    },
    {
      "epoch": 3.731621936989498,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013886064881375,
      "loss": 0.009,
      "step": 7995
    },
    {
      "epoch": 3.732088681446908,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013884712819241422,
      "loss": 0.0182,
      "step": 7996
    },
    {
      "epoch": 3.7325554259043177,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013883360673465665,
      "loss": 0.0167,
      "step": 7997
    },
    {
      "epoch": 3.733022170361727,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013882008444076843,
      "loss": 0.0224,
      "step": 7998
    },
    {
      "epoch": 3.7334889148191364,
      "grad_norm": 0.75,
      "learning_rate": 0.00013880656131104075,
      "loss": 0.0229,
      "step": 7999
    },
    {
      "epoch": 3.733955659276546,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013879303734576473,
      "loss": 0.0193,
      "step": 8000
    },
    {
      "epoch": 3.7344224037339555,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001387795125452316,
      "loss": 0.0173,
      "step": 8001
    },
    {
      "epoch": 3.7348891481913653,
      "grad_norm": 0.578125,
      "learning_rate": 0.00013876598690973252,
      "loss": 0.0156,
      "step": 8002
    },
    {
      "epoch": 3.7353558926487747,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013875246043955876,
      "loss": 0.0244,
      "step": 8003
    },
    {
      "epoch": 3.7358226371061845,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001387389331350015,
      "loss": 0.0148,
      "step": 8004
    },
    {
      "epoch": 3.736289381563594,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013872540499635202,
      "loss": 0.0214,
      "step": 8005
    },
    {
      "epoch": 3.7367561260210036,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00013871187602390163,
      "loss": 0.0085,
      "step": 8006
    },
    {
      "epoch": 3.737222870478413,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013869834621794158,
      "loss": 0.0252,
      "step": 8007
    },
    {
      "epoch": 3.7376896149358227,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013868481557876322,
      "loss": 0.0203,
      "step": 8008
    },
    {
      "epoch": 3.738156359393232,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013867128410665784,
      "loss": 0.018,
      "step": 8009
    },
    {
      "epoch": 3.738623103850642,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001386577518019168,
      "loss": 0.0195,
      "step": 8010
    },
    {
      "epoch": 3.7390898483080512,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001386442186648315,
      "loss": 0.0267,
      "step": 8011
    },
    {
      "epoch": 3.739556592765461,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013863068469569326,
      "loss": 0.0301,
      "step": 8012
    },
    {
      "epoch": 3.7400233372228704,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013861714989479352,
      "loss": 0.0283,
      "step": 8013
    },
    {
      "epoch": 3.74049008168028,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013860361426242372,
      "loss": 0.017,
      "step": 8014
    },
    {
      "epoch": 3.7409568261376895,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013859007779887524,
      "loss": 0.0235,
      "step": 8015
    },
    {
      "epoch": 3.7414235705950993,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001385765405044396,
      "loss": 0.0186,
      "step": 8016
    },
    {
      "epoch": 3.7418903150525087,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00013856300237940822,
      "loss": 0.0179,
      "step": 8017
    },
    {
      "epoch": 3.7423570595099185,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013854946342407265,
      "loss": 0.0192,
      "step": 8018
    },
    {
      "epoch": 3.742823803967328,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013853592363872433,
      "loss": 0.0157,
      "step": 8019
    },
    {
      "epoch": 3.7432905484247376,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013852238302365484,
      "loss": 0.0202,
      "step": 8020
    },
    {
      "epoch": 3.743757292882147,
      "grad_norm": 0.875,
      "learning_rate": 0.00013850884157915567,
      "loss": 0.0282,
      "step": 8021
    },
    {
      "epoch": 3.7442240373395563,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013849529930551842,
      "loss": 0.0174,
      "step": 8022
    },
    {
      "epoch": 3.744690781796966,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013848175620303468,
      "loss": 0.0285,
      "step": 8023
    },
    {
      "epoch": 3.745157526254376,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013846821227199603,
      "loss": 0.0231,
      "step": 8024
    },
    {
      "epoch": 3.7456242707117853,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013845466751269409,
      "loss": 0.0166,
      "step": 8025
    },
    {
      "epoch": 3.7460910151691946,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001384411219254205,
      "loss": 0.0306,
      "step": 8026
    },
    {
      "epoch": 3.7465577596266044,
      "grad_norm": 0.625,
      "learning_rate": 0.00013842757551046691,
      "loss": 0.0194,
      "step": 8027
    },
    {
      "epoch": 3.747024504084014,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00013841402826812496,
      "loss": 0.0118,
      "step": 8028
    },
    {
      "epoch": 3.7474912485414236,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013840048019868637,
      "loss": 0.0211,
      "step": 8029
    },
    {
      "epoch": 3.747957992998833,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013838693130244282,
      "loss": 0.0214,
      "step": 8030
    },
    {
      "epoch": 3.7484247374562427,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00013837338157968607,
      "loss": 0.0331,
      "step": 8031
    },
    {
      "epoch": 3.7488914819136525,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013835983103070783,
      "loss": 0.0166,
      "step": 8032
    },
    {
      "epoch": 3.749358226371062,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013834627965579984,
      "loss": 0.0215,
      "step": 8033
    },
    {
      "epoch": 3.749824970828471,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013833272745525393,
      "loss": 0.0163,
      "step": 8034
    },
    {
      "epoch": 3.750291715285881,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013831917442936188,
      "loss": 0.0261,
      "step": 8035
    },
    {
      "epoch": 3.750758459743291,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001383056205784155,
      "loss": 0.0263,
      "step": 8036
    },
    {
      "epoch": 3.7512252042007,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013829206590270656,
      "loss": 0.0181,
      "step": 8037
    },
    {
      "epoch": 3.7516919486581095,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00013827851040252697,
      "loss": 0.0179,
      "step": 8038
    },
    {
      "epoch": 3.7521586931155193,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001382649540781686,
      "loss": 0.0168,
      "step": 8039
    },
    {
      "epoch": 3.7526254375729287,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00013825139692992327,
      "loss": 0.0214,
      "step": 8040
    },
    {
      "epoch": 3.7526254375729287,
      "eval_loss": 1.835245966911316,
      "eval_runtime": 56.4633,
      "eval_samples_per_second": 31.95,
      "eval_steps_per_second": 4.003,
      "step": 8040
    },
    {
      "epoch": 3.7530921820303385,
      "grad_norm": 0.484375,
      "learning_rate": 0.00013823783895808298,
      "loss": 0.0152,
      "step": 8041
    },
    {
      "epoch": 3.753558926487748,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013822428016293956,
      "loss": 0.0214,
      "step": 8042
    },
    {
      "epoch": 3.7540256709451576,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000138210720544785,
      "loss": 0.0086,
      "step": 8043
    },
    {
      "epoch": 3.754492415402567,
      "grad_norm": 0.4921875,
      "learning_rate": 0.0001381971601039112,
      "loss": 0.0161,
      "step": 8044
    },
    {
      "epoch": 3.7549591598599767,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013818359884061019,
      "loss": 0.0157,
      "step": 8045
    },
    {
      "epoch": 3.755425904317386,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013817003675517392,
      "loss": 0.0177,
      "step": 8046
    },
    {
      "epoch": 3.755892648774796,
      "grad_norm": 0.96484375,
      "learning_rate": 0.0001381564738478944,
      "loss": 0.0166,
      "step": 8047
    },
    {
      "epoch": 3.7563593932322052,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013814291011906367,
      "loss": 0.0166,
      "step": 8048
    },
    {
      "epoch": 3.756826137689615,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013812934556897376,
      "loss": 0.013,
      "step": 8049
    },
    {
      "epoch": 3.7572928821470244,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013811578019791673,
      "loss": 0.0159,
      "step": 8050
    },
    {
      "epoch": 3.757759626604434,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013810221400618468,
      "loss": 0.0256,
      "step": 8051
    },
    {
      "epoch": 3.7582263710618435,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001380886469940697,
      "loss": 0.0203,
      "step": 8052
    },
    {
      "epoch": 3.7586931155192533,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013807507916186387,
      "loss": 0.0184,
      "step": 8053
    },
    {
      "epoch": 3.7591598599766627,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013806151050985933,
      "loss": 0.0258,
      "step": 8054
    },
    {
      "epoch": 3.7596266044340725,
      "grad_norm": 0.828125,
      "learning_rate": 0.00013804794103834827,
      "loss": 0.022,
      "step": 8055
    },
    {
      "epoch": 3.760093348891482,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013803437074762277,
      "loss": 0.0174,
      "step": 8056
    },
    {
      "epoch": 3.7605600933488916,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001380207996379751,
      "loss": 0.0211,
      "step": 8057
    },
    {
      "epoch": 3.761026837806301,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013800722770969745,
      "loss": 0.0237,
      "step": 8058
    },
    {
      "epoch": 3.7614935822637108,
      "grad_norm": 0.6484375,
      "learning_rate": 0.000137993654963082,
      "loss": 0.02,
      "step": 8059
    },
    {
      "epoch": 3.76196032672112,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013798008139842097,
      "loss": 0.0201,
      "step": 8060
    },
    {
      "epoch": 3.7624270711785295,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013796650701600668,
      "loss": 0.0186,
      "step": 8061
    },
    {
      "epoch": 3.7628938156359393,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00013795293181613139,
      "loss": 0.0189,
      "step": 8062
    },
    {
      "epoch": 3.763360560093349,
      "grad_norm": 0.75,
      "learning_rate": 0.00013793935579908733,
      "loss": 0.0229,
      "step": 8063
    },
    {
      "epoch": 3.7638273045507584,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013792577896516687,
      "loss": 0.0129,
      "step": 8064
    },
    {
      "epoch": 3.7642940490081678,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001379122013146623,
      "loss": 0.0179,
      "step": 8065
    },
    {
      "epoch": 3.7647607934655776,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013789862284786596,
      "loss": 0.0212,
      "step": 8066
    },
    {
      "epoch": 3.7652275379229874,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00013788504356507025,
      "loss": 0.019,
      "step": 8067
    },
    {
      "epoch": 3.7656942823803967,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001378714634665675,
      "loss": 0.0159,
      "step": 8068
    },
    {
      "epoch": 3.766161026837806,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013785788255265012,
      "loss": 0.026,
      "step": 8069
    },
    {
      "epoch": 3.766627771295216,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013784430082361052,
      "loss": 0.0195,
      "step": 8070
    },
    {
      "epoch": 3.7670945157526257,
      "grad_norm": 0.75,
      "learning_rate": 0.00013783071827974113,
      "loss": 0.0234,
      "step": 8071
    },
    {
      "epoch": 3.767561260210035,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013781713492133445,
      "loss": 0.0185,
      "step": 8072
    },
    {
      "epoch": 3.7680280046674444,
      "grad_norm": 0.84375,
      "learning_rate": 0.00013780355074868283,
      "loss": 0.0249,
      "step": 8073
    },
    {
      "epoch": 3.768494749124854,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013778996576207887,
      "loss": 0.0276,
      "step": 8074
    },
    {
      "epoch": 3.768961493582264,
      "grad_norm": 0.9609375,
      "learning_rate": 0.00013777637996181504,
      "loss": 0.0296,
      "step": 8075
    },
    {
      "epoch": 3.7694282380396733,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013776279334818376,
      "loss": 0.0201,
      "step": 8076
    },
    {
      "epoch": 3.7698949824970827,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013774920592147768,
      "loss": 0.0137,
      "step": 8077
    },
    {
      "epoch": 3.7703617269544925,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001377356176819893,
      "loss": 0.0129,
      "step": 8078
    },
    {
      "epoch": 3.7708284714119022,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001377220286300112,
      "loss": 0.0251,
      "step": 8079
    },
    {
      "epoch": 3.7712952158693116,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013770843876583598,
      "loss": 0.0164,
      "step": 8080
    },
    {
      "epoch": 3.771761960326721,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013769484808975625,
      "loss": 0.0127,
      "step": 8081
    },
    {
      "epoch": 3.7722287047841307,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001376812566020646,
      "loss": 0.0205,
      "step": 8082
    },
    {
      "epoch": 3.77269544924154,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013766766430305368,
      "loss": 0.0215,
      "step": 8083
    },
    {
      "epoch": 3.77316219369895,
      "grad_norm": 0.94140625,
      "learning_rate": 0.00013765407119301618,
      "loss": 0.0359,
      "step": 8084
    },
    {
      "epoch": 3.7736289381563592,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013764047727224475,
      "loss": 0.0231,
      "step": 8085
    },
    {
      "epoch": 3.774095682613769,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013762688254103205,
      "loss": 0.0314,
      "step": 8086
    },
    {
      "epoch": 3.7745624270711784,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013761328699967083,
      "loss": 0.0268,
      "step": 8087
    },
    {
      "epoch": 3.775029171528588,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013759969064845384,
      "loss": 0.0163,
      "step": 8088
    },
    {
      "epoch": 3.7754959159859975,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00013758609348767372,
      "loss": 0.0162,
      "step": 8089
    },
    {
      "epoch": 3.7759626604434073,
      "grad_norm": 0.71875,
      "learning_rate": 0.00013757249551762336,
      "loss": 0.024,
      "step": 8090
    },
    {
      "epoch": 3.7764294049008167,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001375588967385955,
      "loss": 0.0357,
      "step": 8091
    },
    {
      "epoch": 3.7768961493582265,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013754529715088288,
      "loss": 0.0251,
      "step": 8092
    },
    {
      "epoch": 3.777362893815636,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001375316967547784,
      "loss": 0.0157,
      "step": 8093
    },
    {
      "epoch": 3.7778296382730456,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001375180955505748,
      "loss": 0.0355,
      "step": 8094
    },
    {
      "epoch": 3.778296382730455,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013750449353856501,
      "loss": 0.0169,
      "step": 8095
    },
    {
      "epoch": 3.778763127187865,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00013749089071904183,
      "loss": 0.0159,
      "step": 8096
    },
    {
      "epoch": 3.779229871645274,
      "grad_norm": 0.62109375,
      "learning_rate": 0.0001374772870922982,
      "loss": 0.0176,
      "step": 8097
    },
    {
      "epoch": 3.779696616102684,
      "grad_norm": 0.68359375,
      "learning_rate": 0.000137463682658627,
      "loss": 0.0233,
      "step": 8098
    },
    {
      "epoch": 3.7801633605600933,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00013745007741832114,
      "loss": 0.0094,
      "step": 8099
    },
    {
      "epoch": 3.780630105017503,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00013743647137167355,
      "loss": 0.0181,
      "step": 8100
    },
    {
      "epoch": 3.7810968494749124,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001374228645189772,
      "loss": 0.0138,
      "step": 8101
    },
    {
      "epoch": 3.781563593932322,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013740925686052505,
      "loss": 0.0323,
      "step": 8102
    },
    {
      "epoch": 3.7820303383897316,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013739564839661012,
      "loss": 0.0134,
      "step": 8103
    },
    {
      "epoch": 3.782497082847141,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013738203912752538,
      "loss": 0.0136,
      "step": 8104
    },
    {
      "epoch": 3.7829638273045507,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013736842905356385,
      "loss": 0.017,
      "step": 8105
    },
    {
      "epoch": 3.7834305717619605,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013735481817501855,
      "loss": 0.021,
      "step": 8106
    },
    {
      "epoch": 3.78389731621937,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013734120649218262,
      "loss": 0.0198,
      "step": 8107
    },
    {
      "epoch": 3.784364060676779,
      "grad_norm": 0.578125,
      "learning_rate": 0.00013732759400534908,
      "loss": 0.0183,
      "step": 8108
    },
    {
      "epoch": 3.784830805134189,
      "grad_norm": 0.53125,
      "learning_rate": 0.000137313980714811,
      "loss": 0.0147,
      "step": 8109
    },
    {
      "epoch": 3.785297549591599,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013730036662086152,
      "loss": 0.0276,
      "step": 8110
    },
    {
      "epoch": 3.785764294049008,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013728675172379378,
      "loss": 0.0169,
      "step": 8111
    },
    {
      "epoch": 3.7862310385064175,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001372731360239009,
      "loss": 0.0155,
      "step": 8112
    },
    {
      "epoch": 3.7866977829638273,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013725951952147603,
      "loss": 0.0138,
      "step": 8113
    },
    {
      "epoch": 3.787164527421237,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013724590221681236,
      "loss": 0.0208,
      "step": 8114
    },
    {
      "epoch": 3.7876312718786465,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001372322841102031,
      "loss": 0.0129,
      "step": 8115
    },
    {
      "epoch": 3.788098016336056,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013721866520194148,
      "loss": 0.0189,
      "step": 8116
    },
    {
      "epoch": 3.7885647607934656,
      "grad_norm": 0.578125,
      "learning_rate": 0.00013720504549232068,
      "loss": 0.0138,
      "step": 8117
    },
    {
      "epoch": 3.7890315052508754,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013719142498163394,
      "loss": 0.0188,
      "step": 8118
    },
    {
      "epoch": 3.7894982497082847,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013717780367017462,
      "loss": 0.029,
      "step": 8119
    },
    {
      "epoch": 3.789964994165694,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001371641815582359,
      "loss": 0.0073,
      "step": 8120
    },
    {
      "epoch": 3.790431738623104,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00013715055864611111,
      "loss": 0.0079,
      "step": 8121
    },
    {
      "epoch": 3.7908984830805137,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013713693493409357,
      "loss": 0.0148,
      "step": 8122
    },
    {
      "epoch": 3.791365227537923,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00013712331042247666,
      "loss": 0.0216,
      "step": 8123
    },
    {
      "epoch": 3.7918319719953324,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00013710968511155368,
      "loss": 0.0111,
      "step": 8124
    },
    {
      "epoch": 3.792298716452742,
      "grad_norm": 0.82421875,
      "learning_rate": 0.00013709605900161795,
      "loss": 0.0159,
      "step": 8125
    },
    {
      "epoch": 3.7927654609101515,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013708243209296293,
      "loss": 0.0248,
      "step": 8126
    },
    {
      "epoch": 3.7932322053675613,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013706880438588205,
      "loss": 0.0185,
      "step": 8127
    },
    {
      "epoch": 3.7936989498249707,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013705517588066864,
      "loss": 0.0218,
      "step": 8128
    },
    {
      "epoch": 3.7941656942823805,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001370415465776162,
      "loss": 0.013,
      "step": 8129
    },
    {
      "epoch": 3.79463243873979,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00013702791647701812,
      "loss": 0.0154,
      "step": 8130
    },
    {
      "epoch": 3.7950991831971996,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013701428557916795,
      "loss": 0.0171,
      "step": 8131
    },
    {
      "epoch": 3.795565927654609,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013700065388435917,
      "loss": 0.0169,
      "step": 8132
    },
    {
      "epoch": 3.796032672112019,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013698702139288522,
      "loss": 0.0209,
      "step": 8133
    },
    {
      "epoch": 3.796499416569428,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013697338810503967,
      "loss": 0.0176,
      "step": 8134
    },
    {
      "epoch": 3.796966161026838,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013695975402111605,
      "loss": 0.0172,
      "step": 8135
    },
    {
      "epoch": 3.7974329054842473,
      "grad_norm": 0.484375,
      "learning_rate": 0.00013694611914140793,
      "loss": 0.0145,
      "step": 8136
    },
    {
      "epoch": 3.797899649941657,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013693248346620886,
      "loss": 0.0185,
      "step": 8137
    },
    {
      "epoch": 3.7983663943990664,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00013691884699581244,
      "loss": 0.0106,
      "step": 8138
    },
    {
      "epoch": 3.798833138856476,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001369052097305123,
      "loss": 0.0158,
      "step": 8139
    },
    {
      "epoch": 3.7992998833138856,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013689157167060204,
      "loss": 0.0188,
      "step": 8140
    },
    {
      "epoch": 3.7997666277712954,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013687793281637532,
      "loss": 0.0212,
      "step": 8141
    },
    {
      "epoch": 3.8002333722287047,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013686429316812576,
      "loss": 0.0137,
      "step": 8142
    },
    {
      "epoch": 3.8007001166861145,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001368506527261471,
      "loss": 0.0146,
      "step": 8143
    },
    {
      "epoch": 3.801166861143524,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013683701149073298,
      "loss": 0.0099,
      "step": 8144
    },
    {
      "epoch": 3.8016336056009337,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00013682336946217713,
      "loss": 0.0131,
      "step": 8145
    },
    {
      "epoch": 3.802100350058343,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001368097266407733,
      "loss": 0.0216,
      "step": 8146
    },
    {
      "epoch": 3.8025670945157524,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001367960830268152,
      "loss": 0.015,
      "step": 8147
    },
    {
      "epoch": 3.803033838973162,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001367824386205966,
      "loss": 0.0198,
      "step": 8148
    },
    {
      "epoch": 3.803500583430572,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013676879342241132,
      "loss": 0.019,
      "step": 8149
    },
    {
      "epoch": 3.8039673278879813,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00013675514743255308,
      "loss": 0.0224,
      "step": 8150
    },
    {
      "epoch": 3.8044340723453907,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013674150065131574,
      "loss": 0.0144,
      "step": 8151
    },
    {
      "epoch": 3.8049008168028005,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013672785307899314,
      "loss": 0.0179,
      "step": 8152
    },
    {
      "epoch": 3.8053675612602103,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013671420471587912,
      "loss": 0.0164,
      "step": 8153
    },
    {
      "epoch": 3.8058343057176196,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001367005555622675,
      "loss": 0.029,
      "step": 8154
    },
    {
      "epoch": 3.806301050175029,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013668690561845223,
      "loss": 0.0167,
      "step": 8155
    },
    {
      "epoch": 3.8067677946324388,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013667325488472717,
      "loss": 0.0174,
      "step": 8156
    },
    {
      "epoch": 3.8072345390898485,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013665960336138624,
      "loss": 0.0276,
      "step": 8157
    },
    {
      "epoch": 3.807701283547258,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013664595104872338,
      "loss": 0.0115,
      "step": 8158
    },
    {
      "epoch": 3.8081680280046672,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013663229794703253,
      "loss": 0.0178,
      "step": 8159
    },
    {
      "epoch": 3.808634772462077,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001366186440566077,
      "loss": 0.0279,
      "step": 8160
    },
    {
      "epoch": 3.809101516919487,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001366049893777428,
      "loss": 0.0161,
      "step": 8161
    },
    {
      "epoch": 3.809568261376896,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00013659133391073184,
      "loss": 0.0136,
      "step": 8162
    },
    {
      "epoch": 3.8100350058343055,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001365776776558689,
      "loss": 0.0248,
      "step": 8163
    },
    {
      "epoch": 3.8105017502917153,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013656402061344792,
      "loss": 0.0285,
      "step": 8164
    },
    {
      "epoch": 3.8109684947491247,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013655036278376309,
      "loss": 0.0146,
      "step": 8165
    },
    {
      "epoch": 3.8114352392065345,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013653670416710837,
      "loss": 0.0201,
      "step": 8166
    },
    {
      "epoch": 3.811901983663944,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013652304476377783,
      "loss": 0.0155,
      "step": 8167
    },
    {
      "epoch": 3.8123687281213536,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013650938457406564,
      "loss": 0.0167,
      "step": 8168
    },
    {
      "epoch": 3.812835472578763,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013649572359826587,
      "loss": 0.0148,
      "step": 8169
    },
    {
      "epoch": 3.813302217036173,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013648206183667272,
      "loss": 0.0117,
      "step": 8170
    },
    {
      "epoch": 3.813768961493582,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001364683992895803,
      "loss": 0.0141,
      "step": 8171
    },
    {
      "epoch": 3.814235705950992,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013645473595728276,
      "loss": 0.0209,
      "step": 8172
    },
    {
      "epoch": 3.8147024504084013,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013644107184007428,
      "loss": 0.0154,
      "step": 8173
    },
    {
      "epoch": 3.815169194865811,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013642740693824913,
      "loss": 0.0166,
      "step": 8174
    },
    {
      "epoch": 3.8156359393232204,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00013641374125210153,
      "loss": 0.0123,
      "step": 8175
    },
    {
      "epoch": 3.8161026837806302,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013640007478192562,
      "loss": 0.0121,
      "step": 8176
    },
    {
      "epoch": 3.8165694282380396,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001363864075280157,
      "loss": 0.0109,
      "step": 8177
    },
    {
      "epoch": 3.8170361726954494,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00013637273949066608,
      "loss": 0.0105,
      "step": 8178
    },
    {
      "epoch": 3.8175029171528587,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013635907067017104,
      "loss": 0.0124,
      "step": 8179
    },
    {
      "epoch": 3.8179696616102685,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00013634540106682483,
      "loss": 0.019,
      "step": 8180
    },
    {
      "epoch": 3.818436406067678,
      "grad_norm": 0.97265625,
      "learning_rate": 0.00013633173068092183,
      "loss": 0.022,
      "step": 8181
    },
    {
      "epoch": 3.8189031505250877,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013631805951275637,
      "loss": 0.017,
      "step": 8182
    },
    {
      "epoch": 3.819369894982497,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013630438756262277,
      "loss": 0.0145,
      "step": 8183
    },
    {
      "epoch": 3.819836639439907,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00013629071483081543,
      "loss": 0.013,
      "step": 8184
    },
    {
      "epoch": 3.820303383897316,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00013627704131762876,
      "loss": 0.0255,
      "step": 8185
    },
    {
      "epoch": 3.8207701283547255,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013626336702335712,
      "loss": 0.0425,
      "step": 8186
    },
    {
      "epoch": 3.8212368728121353,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013624969194829493,
      "loss": 0.0193,
      "step": 8187
    },
    {
      "epoch": 3.821703617269545,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00013623601609273668,
      "loss": 0.0128,
      "step": 8188
    },
    {
      "epoch": 3.8221703617269545,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001362223394569768,
      "loss": 0.0207,
      "step": 8189
    },
    {
      "epoch": 3.822637106184364,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00013620866204130976,
      "loss": 0.0238,
      "step": 8190
    },
    {
      "epoch": 3.8231038506417736,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013619498384603005,
      "loss": 0.014,
      "step": 8191
    },
    {
      "epoch": 3.8235705950991834,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013618130487143217,
      "loss": 0.0117,
      "step": 8192
    },
    {
      "epoch": 3.8240373395565928,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00013616762511781068,
      "loss": 0.0154,
      "step": 8193
    },
    {
      "epoch": 3.824504084014002,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013615394458546006,
      "loss": 0.0196,
      "step": 8194
    },
    {
      "epoch": 3.824970828471412,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00013614026327467494,
      "loss": 0.0092,
      "step": 8195
    },
    {
      "epoch": 3.8254375729288217,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013612658118574979,
      "loss": 0.024,
      "step": 8196
    },
    {
      "epoch": 3.825904317386231,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00013611289831897932,
      "loss": 0.0199,
      "step": 8197
    },
    {
      "epoch": 3.8263710618436404,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00013609921467465808,
      "loss": 0.0191,
      "step": 8198
    },
    {
      "epoch": 3.82683780630105,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00013608553025308066,
      "loss": 0.0113,
      "step": 8199
    },
    {
      "epoch": 3.82730455075846,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013607184505454173,
      "loss": 0.0138,
      "step": 8200
    },
    {
      "epoch": 3.8277712952158693,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013605815907933598,
      "loss": 0.0191,
      "step": 8201
    },
    {
      "epoch": 3.8282380396732787,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013604447232775808,
      "loss": 0.0149,
      "step": 8202
    },
    {
      "epoch": 3.8287047841306885,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00013603078480010265,
      "loss": 0.0108,
      "step": 8203
    },
    {
      "epoch": 3.8291715285880983,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00013601709649666447,
      "loss": 0.0162,
      "step": 8204
    },
    {
      "epoch": 3.8296382730455076,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013600340741773824,
      "loss": 0.021,
      "step": 8205
    },
    {
      "epoch": 3.830105017502917,
      "grad_norm": 0.7109375,
      "learning_rate": 0.0001359897175636187,
      "loss": 0.0152,
      "step": 8206
    },
    {
      "epoch": 3.830571761960327,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001359760269346006,
      "loss": 0.0191,
      "step": 8207
    },
    {
      "epoch": 3.831038506417736,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013596233553097876,
      "loss": 0.0154,
      "step": 8208
    },
    {
      "epoch": 3.831505250875146,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001359486433530479,
      "loss": 0.0244,
      "step": 8209
    },
    {
      "epoch": 3.8319719953325553,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00013593495040110283,
      "loss": 0.014,
      "step": 8210
    },
    {
      "epoch": 3.832438739789965,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001359212566754385,
      "loss": 0.0131,
      "step": 8211
    },
    {
      "epoch": 3.8329054842473744,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001359075621763496,
      "loss": 0.0135,
      "step": 8212
    },
    {
      "epoch": 3.8333722287047842,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013589386690413105,
      "loss": 0.0198,
      "step": 8213
    },
    {
      "epoch": 3.8338389731621936,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001358801708590777,
      "loss": 0.0122,
      "step": 8214
    },
    {
      "epoch": 3.8343057176196034,
      "grad_norm": 0.82421875,
      "learning_rate": 0.0001358664740414845,
      "loss": 0.0282,
      "step": 8215
    },
    {
      "epoch": 3.8347724620770127,
      "grad_norm": 0.423828125,
      "learning_rate": 0.0001358527764516463,
      "loss": 0.0154,
      "step": 8216
    },
    {
      "epoch": 3.8352392065344225,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013583907808985805,
      "loss": 0.0204,
      "step": 8217
    },
    {
      "epoch": 3.835705950991832,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001358253789564147,
      "loss": 0.0223,
      "step": 8218
    },
    {
      "epoch": 3.8361726954492417,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001358116790516112,
      "loss": 0.0162,
      "step": 8219
    },
    {
      "epoch": 3.836639439906651,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013579797837574248,
      "loss": 0.0192,
      "step": 8220
    },
    {
      "epoch": 3.837106184364061,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013578427692910358,
      "loss": 0.0163,
      "step": 8221
    },
    {
      "epoch": 3.83757292882147,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001357705747119895,
      "loss": 0.0163,
      "step": 8222
    },
    {
      "epoch": 3.83803967327888,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013575687172469526,
      "loss": 0.0141,
      "step": 8223
    },
    {
      "epoch": 3.8385064177362893,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001357431679675159,
      "loss": 0.0137,
      "step": 8224
    },
    {
      "epoch": 3.838973162193699,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013572946344074646,
      "loss": 0.0146,
      "step": 8225
    },
    {
      "epoch": 3.8394399066511085,
      "grad_norm": 0.458984375,
      "learning_rate": 0.000135715758144682,
      "loss": 0.0103,
      "step": 8226
    },
    {
      "epoch": 3.8399066511085183,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013570205207961771,
      "loss": 0.0221,
      "step": 8227
    },
    {
      "epoch": 3.8403733955659276,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013568834524584859,
      "loss": 0.0187,
      "step": 8228
    },
    {
      "epoch": 3.840840140023337,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001356746376436698,
      "loss": 0.0165,
      "step": 8229
    },
    {
      "epoch": 3.8413068844807468,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013566092927337647,
      "loss": 0.0149,
      "step": 8230
    },
    {
      "epoch": 3.8417736289381565,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013564722013526377,
      "loss": 0.0157,
      "step": 8231
    },
    {
      "epoch": 3.842240373395566,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013563351022962688,
      "loss": 0.0181,
      "step": 8232
    },
    {
      "epoch": 3.8427071178529753,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013561979955676099,
      "loss": 0.0149,
      "step": 8233
    },
    {
      "epoch": 3.843173862310385,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001356060881169613,
      "loss": 0.0165,
      "step": 8234
    },
    {
      "epoch": 3.843640606767795,
      "grad_norm": 0.66796875,
      "learning_rate": 0.000135592375910523,
      "loss": 0.0134,
      "step": 8235
    },
    {
      "epoch": 3.844107351225204,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013557866293774135,
      "loss": 0.0173,
      "step": 8236
    },
    {
      "epoch": 3.8445740956826135,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013556494919891163,
      "loss": 0.0147,
      "step": 8237
    },
    {
      "epoch": 3.8450408401400233,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001355512346943291,
      "loss": 0.0093,
      "step": 8238
    },
    {
      "epoch": 3.845507584597433,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013553751942428904,
      "loss": 0.0189,
      "step": 8239
    },
    {
      "epoch": 3.8459743290548425,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013552380338908675,
      "loss": 0.0166,
      "step": 8240
    },
    {
      "epoch": 3.846441073512252,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013551008658901754,
      "loss": 0.0236,
      "step": 8241
    },
    {
      "epoch": 3.8469078179696616,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013549636902437682,
      "loss": 0.0214,
      "step": 8242
    },
    {
      "epoch": 3.8473745624270714,
      "grad_norm": 0.671875,
      "learning_rate": 0.00013548265069545984,
      "loss": 0.0159,
      "step": 8243
    },
    {
      "epoch": 3.847841306884481,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013546893160256205,
      "loss": 0.0185,
      "step": 8244
    },
    {
      "epoch": 3.84830805134189,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013545521174597878,
      "loss": 0.0213,
      "step": 8245
    },
    {
      "epoch": 3.8487747957993,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013544149112600548,
      "loss": 0.0175,
      "step": 8246
    },
    {
      "epoch": 3.8492415402567097,
      "grad_norm": 0.75,
      "learning_rate": 0.00013542776974293758,
      "loss": 0.0191,
      "step": 8247
    },
    {
      "epoch": 3.849708284714119,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013541404759707045,
      "loss": 0.0151,
      "step": 8248
    },
    {
      "epoch": 3.8501750291715284,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013540032468869958,
      "loss": 0.0192,
      "step": 8249
    },
    {
      "epoch": 3.8506417736289382,
      "grad_norm": 0.6875,
      "learning_rate": 0.00013538660101812048,
      "loss": 0.0314,
      "step": 8250
    },
    {
      "epoch": 3.8511085180863476,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013537287658562858,
      "loss": 0.0158,
      "step": 8251
    },
    {
      "epoch": 3.8515752625437574,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00013535915139151937,
      "loss": 0.0157,
      "step": 8252
    },
    {
      "epoch": 3.8520420070011667,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00013534542543608844,
      "loss": 0.0316,
      "step": 8253
    },
    {
      "epoch": 3.8525087514585765,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013533169871963126,
      "loss": 0.0172,
      "step": 8254
    },
    {
      "epoch": 3.852975495915986,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001353179712424434,
      "loss": 0.0156,
      "step": 8255
    },
    {
      "epoch": 3.8534422403733957,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013530424300482044,
      "loss": 0.0124,
      "step": 8256
    },
    {
      "epoch": 3.853908984830805,
      "grad_norm": 0.64453125,
      "learning_rate": 0.000135290514007058,
      "loss": 0.0212,
      "step": 8257
    },
    {
      "epoch": 3.854375729288215,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001352767842494516,
      "loss": 0.0095,
      "step": 8258
    },
    {
      "epoch": 3.854842473745624,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013526305373229687,
      "loss": 0.0252,
      "step": 8259
    },
    {
      "epoch": 3.855309218203034,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013524932245588954,
      "loss": 0.0215,
      "step": 8260
    },
    {
      "epoch": 3.8557759626604433,
      "grad_norm": 0.78125,
      "learning_rate": 0.00013523559042052515,
      "loss": 0.0237,
      "step": 8261
    },
    {
      "epoch": 3.856242707117853,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013522185762649937,
      "loss": 0.0183,
      "step": 8262
    },
    {
      "epoch": 3.8567094515752625,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013520812407410795,
      "loss": 0.0209,
      "step": 8263
    },
    {
      "epoch": 3.8571761960326723,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013519438976364656,
      "loss": 0.0159,
      "step": 8264
    },
    {
      "epoch": 3.8576429404900816,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00013518065469541086,
      "loss": 0.0209,
      "step": 8265
    },
    {
      "epoch": 3.8581096849474914,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001351669188696967,
      "loss": 0.0097,
      "step": 8266
    },
    {
      "epoch": 3.8585764294049008,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013515318228679975,
      "loss": 0.0182,
      "step": 8267
    },
    {
      "epoch": 3.8590431738623105,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013513944494701572,
      "loss": 0.0196,
      "step": 8268
    },
    {
      "epoch": 3.85950991831972,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001351257068506405,
      "loss": 0.0201,
      "step": 8269
    },
    {
      "epoch": 3.8599766627771297,
      "grad_norm": 0.8125,
      "learning_rate": 0.00013511196799796983,
      "loss": 0.0329,
      "step": 8270
    },
    {
      "epoch": 3.860443407234539,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013509822838929952,
      "loss": 0.0202,
      "step": 8271
    },
    {
      "epoch": 3.8609101516919484,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013508448802492543,
      "loss": 0.0136,
      "step": 8272
    },
    {
      "epoch": 3.861376896149358,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013507074690514339,
      "loss": 0.0146,
      "step": 8273
    },
    {
      "epoch": 3.861843640606768,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013505700503024925,
      "loss": 0.0187,
      "step": 8274
    },
    {
      "epoch": 3.8623103850641773,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013504326240053888,
      "loss": 0.0199,
      "step": 8275
    },
    {
      "epoch": 3.8627771295215867,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001350295190163082,
      "loss": 0.0137,
      "step": 8276
    },
    {
      "epoch": 3.8632438739789965,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013501577487785312,
      "loss": 0.0163,
      "step": 8277
    },
    {
      "epoch": 3.8637106184364063,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013500202998546954,
      "loss": 0.0213,
      "step": 8278
    },
    {
      "epoch": 3.8641773628938156,
      "grad_norm": 1.1171875,
      "learning_rate": 0.00013498828433945343,
      "loss": 0.0264,
      "step": 8279
    },
    {
      "epoch": 3.864644107351225,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013497453794010072,
      "loss": 0.0136,
      "step": 8280
    },
    {
      "epoch": 3.865110851808635,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013496079078770742,
      "loss": 0.0162,
      "step": 8281
    },
    {
      "epoch": 3.8655775962660446,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001349470428825695,
      "loss": 0.0114,
      "step": 8282
    },
    {
      "epoch": 3.866044340723454,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00013493329422498298,
      "loss": 0.017,
      "step": 8283
    },
    {
      "epoch": 3.8665110851808633,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013491954481524386,
      "loss": 0.0134,
      "step": 8284
    },
    {
      "epoch": 3.866977829638273,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013490579465364822,
      "loss": 0.0196,
      "step": 8285
    },
    {
      "epoch": 3.867444574095683,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00013489204374049206,
      "loss": 0.0164,
      "step": 8286
    },
    {
      "epoch": 3.8679113185530922,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00013487829207607153,
      "loss": 0.0233,
      "step": 8287
    },
    {
      "epoch": 3.8683780630105016,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00013486453966068263,
      "loss": 0.0123,
      "step": 8288
    },
    {
      "epoch": 3.8688448074679114,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013485078649462152,
      "loss": 0.0147,
      "step": 8289
    },
    {
      "epoch": 3.8693115519253207,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00013483703257818433,
      "loss": 0.0108,
      "step": 8290
    },
    {
      "epoch": 3.8697782963827305,
      "grad_norm": 0.75,
      "learning_rate": 0.00013482327791166716,
      "loss": 0.0187,
      "step": 8291
    },
    {
      "epoch": 3.87024504084014,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013480952249536618,
      "loss": 0.019,
      "step": 8292
    },
    {
      "epoch": 3.8707117852975497,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013479576632957757,
      "loss": 0.0214,
      "step": 8293
    },
    {
      "epoch": 3.871178529754959,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013478200941459748,
      "loss": 0.0194,
      "step": 8294
    },
    {
      "epoch": 3.871645274212369,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013476825175072214,
      "loss": 0.0157,
      "step": 8295
    },
    {
      "epoch": 3.872112018669778,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013475449333824778,
      "loss": 0.0132,
      "step": 8296
    },
    {
      "epoch": 3.872578763127188,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001347407341774706,
      "loss": 0.018,
      "step": 8297
    },
    {
      "epoch": 3.8730455075845973,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013472697426868688,
      "loss": 0.0166,
      "step": 8298
    },
    {
      "epoch": 3.873512252042007,
      "grad_norm": 0.734375,
      "learning_rate": 0.00013471321361219288,
      "loss": 0.0217,
      "step": 8299
    },
    {
      "epoch": 3.8739789964994165,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013469945220828488,
      "loss": 0.0242,
      "step": 8300
    },
    {
      "epoch": 3.8744457409568263,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00013468569005725914,
      "loss": 0.0111,
      "step": 8301
    },
    {
      "epoch": 3.8749124854142356,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013467192715941204,
      "loss": 0.0101,
      "step": 8302
    },
    {
      "epoch": 3.8753792298716454,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00013465816351503987,
      "loss": 0.0126,
      "step": 8303
    },
    {
      "epoch": 3.8758459743290548,
      "grad_norm": 0.59375,
      "learning_rate": 0.000134644399124439,
      "loss": 0.0147,
      "step": 8304
    },
    {
      "epoch": 3.8763127187864646,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00013463063398790574,
      "loss": 0.0084,
      "step": 8305
    },
    {
      "epoch": 3.876779463243874,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013461686810573655,
      "loss": 0.0131,
      "step": 8306
    },
    {
      "epoch": 3.8772462077012837,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013460310147822777,
      "loss": 0.0125,
      "step": 8307
    },
    {
      "epoch": 3.877712952158693,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013458933410567578,
      "loss": 0.0107,
      "step": 8308
    },
    {
      "epoch": 3.878179696616103,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001345755659883771,
      "loss": 0.0222,
      "step": 8309
    },
    {
      "epoch": 3.878646441073512,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001345617971266281,
      "loss": 0.0163,
      "step": 8310
    },
    {
      "epoch": 3.8791131855309215,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013454802752072527,
      "loss": 0.0193,
      "step": 8311
    },
    {
      "epoch": 3.8795799299883313,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001345342571709651,
      "loss": 0.0157,
      "step": 8312
    },
    {
      "epoch": 3.880046674445741,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013452048607764402,
      "loss": 0.0203,
      "step": 8313
    },
    {
      "epoch": 3.8805134189031505,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013450671424105858,
      "loss": 0.0155,
      "step": 8314
    },
    {
      "epoch": 3.88098016336056,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001344929416615053,
      "loss": 0.0119,
      "step": 8315
    },
    {
      "epoch": 3.8814469078179696,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001344791683392807,
      "loss": 0.0104,
      "step": 8316
    },
    {
      "epoch": 3.8819136522753794,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013446539427468138,
      "loss": 0.0208,
      "step": 8317
    },
    {
      "epoch": 3.882380396732789,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013445161946800386,
      "loss": 0.0122,
      "step": 8318
    },
    {
      "epoch": 3.882847141190198,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013443784391954476,
      "loss": 0.0155,
      "step": 8319
    },
    {
      "epoch": 3.883313885647608,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013442406762960064,
      "loss": 0.0146,
      "step": 8320
    },
    {
      "epoch": 3.8837806301050177,
      "grad_norm": 0.76171875,
      "learning_rate": 0.0001344102905984682,
      "loss": 0.023,
      "step": 8321
    },
    {
      "epoch": 3.884247374562427,
      "grad_norm": 0.578125,
      "learning_rate": 0.000134396512826444,
      "loss": 0.0185,
      "step": 8322
    },
    {
      "epoch": 3.8847141190198364,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001343827343138247,
      "loss": 0.0092,
      "step": 8323
    },
    {
      "epoch": 3.8851808634772462,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013436895506090699,
      "loss": 0.0136,
      "step": 8324
    },
    {
      "epoch": 3.885647607934656,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013435517506798757,
      "loss": 0.0122,
      "step": 8325
    },
    {
      "epoch": 3.8861143523920654,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0001343413943353631,
      "loss": 0.0092,
      "step": 8326
    },
    {
      "epoch": 3.8865810968494747,
      "grad_norm": 0.65625,
      "learning_rate": 0.0001343276128633303,
      "loss": 0.0245,
      "step": 8327
    },
    {
      "epoch": 3.8870478413068845,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001343138306521859,
      "loss": 0.0163,
      "step": 8328
    },
    {
      "epoch": 3.8875145857642943,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00013430004770222665,
      "loss": 0.0152,
      "step": 8329
    },
    {
      "epoch": 3.8879813302217037,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013428626401374932,
      "loss": 0.0122,
      "step": 8330
    },
    {
      "epoch": 3.888448074679113,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013427247958705071,
      "loss": 0.017,
      "step": 8331
    },
    {
      "epoch": 3.888914819136523,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00013425869442242757,
      "loss": 0.0171,
      "step": 8332
    },
    {
      "epoch": 3.889381563593932,
      "grad_norm": 0.625,
      "learning_rate": 0.00013424490852017676,
      "loss": 0.0222,
      "step": 8333
    },
    {
      "epoch": 3.889848308051342,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013423112188059504,
      "loss": 0.0174,
      "step": 8334
    },
    {
      "epoch": 3.8903150525087513,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001342173345039793,
      "loss": 0.0091,
      "step": 8335
    },
    {
      "epoch": 3.890781796966161,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013420354639062635,
      "loss": 0.0189,
      "step": 8336
    },
    {
      "epoch": 3.8912485414235705,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013418975754083312,
      "loss": 0.0137,
      "step": 8337
    },
    {
      "epoch": 3.8917152858809803,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013417596795489652,
      "loss": 0.0095,
      "step": 8338
    },
    {
      "epoch": 3.8921820303383896,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013416217763311336,
      "loss": 0.0138,
      "step": 8339
    },
    {
      "epoch": 3.8926487747957994,
      "grad_norm": 0.71875,
      "learning_rate": 0.0001341483865757806,
      "loss": 0.0214,
      "step": 8340
    },
    {
      "epoch": 3.8931155192532088,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00013413459478319522,
      "loss": 0.0118,
      "step": 8341
    },
    {
      "epoch": 3.8935822637106186,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00013412080225565413,
      "loss": 0.0226,
      "step": 8342
    },
    {
      "epoch": 3.894049008168028,
      "grad_norm": 0.734375,
      "learning_rate": 0.0001341070089934543,
      "loss": 0.014,
      "step": 8343
    },
    {
      "epoch": 3.8945157526254377,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013409321499689274,
      "loss": 0.0128,
      "step": 8344
    },
    {
      "epoch": 3.894982497082847,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013407942026626642,
      "loss": 0.0142,
      "step": 8345
    },
    {
      "epoch": 3.895449241540257,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001340656248018724,
      "loss": 0.0144,
      "step": 8346
    },
    {
      "epoch": 3.895915985997666,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013405182860400764,
      "loss": 0.0181,
      "step": 8347
    },
    {
      "epoch": 3.896382730455076,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00013403803167296927,
      "loss": 0.0168,
      "step": 8348
    },
    {
      "epoch": 3.8968494749124853,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00013402423400905427,
      "loss": 0.0156,
      "step": 8349
    },
    {
      "epoch": 3.897316219369895,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001340104356125598,
      "loss": 0.0128,
      "step": 8350
    },
    {
      "epoch": 3.8977829638273045,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00013399663648378288,
      "loss": 0.0193,
      "step": 8351
    },
    {
      "epoch": 3.8982497082847143,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001339828366230207,
      "loss": 0.0123,
      "step": 8352
    },
    {
      "epoch": 3.8987164527421236,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001339690360305703,
      "loss": 0.0151,
      "step": 8353
    },
    {
      "epoch": 3.899183197199533,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013395523470672888,
      "loss": 0.0172,
      "step": 8354
    },
    {
      "epoch": 3.899649941656943,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001339414326517936,
      "loss": 0.0092,
      "step": 8355
    },
    {
      "epoch": 3.9001166861143526,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001339276298660616,
      "loss": 0.0193,
      "step": 8356
    },
    {
      "epoch": 3.900583430571762,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013391382634983008,
      "loss": 0.0206,
      "step": 8357
    },
    {
      "epoch": 3.9010501750291713,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013390002210339626,
      "loss": 0.0206,
      "step": 8358
    },
    {
      "epoch": 3.901516919486581,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013388621712705735,
      "loss": 0.0133,
      "step": 8359
    },
    {
      "epoch": 3.901983663943991,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013387241142111058,
      "loss": 0.0206,
      "step": 8360
    },
    {
      "epoch": 3.9024504084014002,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013385860498585323,
      "loss": 0.0154,
      "step": 8361
    },
    {
      "epoch": 3.9029171528588096,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013384479782158255,
      "loss": 0.0107,
      "step": 8362
    },
    {
      "epoch": 3.9033838973162194,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013383098992859579,
      "loss": 0.0177,
      "step": 8363
    },
    {
      "epoch": 3.903850641773629,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001338171813071903,
      "loss": 0.0092,
      "step": 8364
    },
    {
      "epoch": 3.9043173862310385,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001338033719576634,
      "loss": 0.0103,
      "step": 8365
    },
    {
      "epoch": 3.904784130688448,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00013378956188031234,
      "loss": 0.0228,
      "step": 8366
    },
    {
      "epoch": 3.9052508751458577,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013377575107543455,
      "loss": 0.0199,
      "step": 8367
    },
    {
      "epoch": 3.9057176196032675,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001337619395433274,
      "loss": 0.0201,
      "step": 8368
    },
    {
      "epoch": 3.906184364060677,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00013374812728428817,
      "loss": 0.022,
      "step": 8369
    },
    {
      "epoch": 3.906651108518086,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013373431429861436,
      "loss": 0.0135,
      "step": 8370
    },
    {
      "epoch": 3.907117852975496,
      "grad_norm": 0.58203125,
      "learning_rate": 0.0001337205005866033,
      "loss": 0.0154,
      "step": 8371
    },
    {
      "epoch": 3.9075845974329058,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013370668614855248,
      "loss": 0.0217,
      "step": 8372
    },
    {
      "epoch": 3.908051341890315,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013369287098475928,
      "loss": 0.0177,
      "step": 8373
    },
    {
      "epoch": 3.9085180863477245,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001336790550955212,
      "loss": 0.0167,
      "step": 8374
    },
    {
      "epoch": 3.9089848308051343,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00013366523848113573,
      "loss": 0.0143,
      "step": 8375
    },
    {
      "epoch": 3.9094515752625436,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00013365142114190025,
      "loss": 0.0146,
      "step": 8376
    },
    {
      "epoch": 3.9099183197199534,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001336376030781124,
      "loss": 0.0103,
      "step": 8377
    },
    {
      "epoch": 3.9103850641773628,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001336237842900696,
      "loss": 0.0143,
      "step": 8378
    },
    {
      "epoch": 3.9108518086347726,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00013360996477806942,
      "loss": 0.0123,
      "step": 8379
    },
    {
      "epoch": 3.911318553092182,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013359614454240944,
      "loss": 0.0203,
      "step": 8380
    },
    {
      "epoch": 3.9117852975495917,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00013358232358338715,
      "loss": 0.0104,
      "step": 8381
    },
    {
      "epoch": 3.912252042007001,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001335685019013002,
      "loss": 0.0224,
      "step": 8382
    },
    {
      "epoch": 3.912718786464411,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00013355467949644615,
      "loss": 0.0166,
      "step": 8383
    },
    {
      "epoch": 3.91318553092182,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001335408563691226,
      "loss": 0.0148,
      "step": 8384
    },
    {
      "epoch": 3.91365227537923,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013352703251962725,
      "loss": 0.0202,
      "step": 8385
    },
    {
      "epoch": 3.9141190198366393,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00013351320794825766,
      "loss": 0.0136,
      "step": 8386
    },
    {
      "epoch": 3.914585764294049,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013349938265531152,
      "loss": 0.0211,
      "step": 8387
    },
    {
      "epoch": 3.9150525087514585,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001334855566410865,
      "loss": 0.0239,
      "step": 8388
    },
    {
      "epoch": 3.9155192532088683,
      "grad_norm": 0.8828125,
      "learning_rate": 0.0001334717299058803,
      "loss": 0.0209,
      "step": 8389
    },
    {
      "epoch": 3.9159859976662776,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013345790244999062,
      "loss": 0.015,
      "step": 8390
    },
    {
      "epoch": 3.9164527421236874,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013344407427371517,
      "loss": 0.0067,
      "step": 8391
    },
    {
      "epoch": 3.916919486581097,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00013343024537735169,
      "loss": 0.0098,
      "step": 8392
    },
    {
      "epoch": 3.9173862310385066,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013341641576119796,
      "loss": 0.017,
      "step": 8393
    },
    {
      "epoch": 3.917852975495916,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001334025854255517,
      "loss": 0.0122,
      "step": 8394
    },
    {
      "epoch": 3.9183197199533257,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001333887543707107,
      "loss": 0.0168,
      "step": 8395
    },
    {
      "epoch": 3.918786464410735,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001333749225969728,
      "loss": 0.0097,
      "step": 8396
    },
    {
      "epoch": 3.9192532088681444,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013336109010463577,
      "loss": 0.0169,
      "step": 8397
    },
    {
      "epoch": 3.9197199533255542,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00013334725689399747,
      "loss": 0.0274,
      "step": 8398
    },
    {
      "epoch": 3.920186697782964,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001333334229653557,
      "loss": 0.0134,
      "step": 8399
    },
    {
      "epoch": 3.9206534422403734,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013331958831900838,
      "loss": 0.02,
      "step": 8400
    },
    {
      "epoch": 3.9211201866977827,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00013330575295525336,
      "loss": 0.0079,
      "step": 8401
    },
    {
      "epoch": 3.9215869311551925,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00013329191687438851,
      "loss": 0.0203,
      "step": 8402
    },
    {
      "epoch": 3.9220536756126023,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013327808007671178,
      "loss": 0.0109,
      "step": 8403
    },
    {
      "epoch": 3.9225204200700117,
      "grad_norm": 0.66015625,
      "learning_rate": 0.000133264242562521,
      "loss": 0.016,
      "step": 8404
    },
    {
      "epoch": 3.922987164527421,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001332504043321142,
      "loss": 0.0119,
      "step": 8405
    },
    {
      "epoch": 3.923453908984831,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013323656538578933,
      "loss": 0.0148,
      "step": 8406
    },
    {
      "epoch": 3.9239206534422406,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001332227257238443,
      "loss": 0.0086,
      "step": 8407
    },
    {
      "epoch": 3.92438739789965,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013320888534657717,
      "loss": 0.019,
      "step": 8408
    },
    {
      "epoch": 3.9248541423570593,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00013319504425428585,
      "loss": 0.0081,
      "step": 8409
    },
    {
      "epoch": 3.925320886814469,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001331812024472684,
      "loss": 0.0157,
      "step": 8410
    },
    {
      "epoch": 3.925787631271879,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013316735992582282,
      "loss": 0.0116,
      "step": 8411
    },
    {
      "epoch": 3.9262543757292883,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00013315351669024726,
      "loss": 0.0093,
      "step": 8412
    },
    {
      "epoch": 3.9267211201866976,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00013313967274083965,
      "loss": 0.0153,
      "step": 8413
    },
    {
      "epoch": 3.9271878646441074,
      "grad_norm": 0.8984375,
      "learning_rate": 0.0001331258280778981,
      "loss": 0.0222,
      "step": 8414
    },
    {
      "epoch": 3.9276546091015168,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00013311198270172074,
      "loss": 0.0207,
      "step": 8415
    },
    {
      "epoch": 3.9281213535589266,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00013309813661260567,
      "loss": 0.0067,
      "step": 8416
    },
    {
      "epoch": 3.928588098016336,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00013308428981085096,
      "loss": 0.0081,
      "step": 8417
    },
    {
      "epoch": 3.9290548424737457,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001330704422967548,
      "loss": 0.0149,
      "step": 8418
    },
    {
      "epoch": 3.929521586931155,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001330565940706153,
      "loss": 0.0187,
      "step": 8419
    },
    {
      "epoch": 3.929988331388565,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001330427451327307,
      "loss": 0.0128,
      "step": 8420
    },
    {
      "epoch": 3.930455075845974,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00013302889548339914,
      "loss": 0.0082,
      "step": 8421
    },
    {
      "epoch": 3.930921820303384,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013301504512291875,
      "loss": 0.0115,
      "step": 8422
    },
    {
      "epoch": 3.9313885647607933,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013300119405158786,
      "loss": 0.013,
      "step": 8423
    },
    {
      "epoch": 3.931855309218203,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013298734226970465,
      "loss": 0.014,
      "step": 8424
    },
    {
      "epoch": 3.9323220536756125,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013297348977756731,
      "loss": 0.0112,
      "step": 8425
    },
    {
      "epoch": 3.9327887981330223,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013295963657547418,
      "loss": 0.0096,
      "step": 8426
    },
    {
      "epoch": 3.9332555425904316,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00013294578266372353,
      "loss": 0.0219,
      "step": 8427
    },
    {
      "epoch": 3.9337222870478414,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013293192804261357,
      "loss": 0.0153,
      "step": 8428
    },
    {
      "epoch": 3.934189031505251,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013291807271244272,
      "loss": 0.0095,
      "step": 8429
    },
    {
      "epoch": 3.9346557759626606,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013290421667350923,
      "loss": 0.0138,
      "step": 8430
    },
    {
      "epoch": 3.93512252042007,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013289035992611144,
      "loss": 0.0098,
      "step": 8431
    },
    {
      "epoch": 3.9355892648774797,
      "grad_norm": 0.703125,
      "learning_rate": 0.00013287650247054765,
      "loss": 0.0165,
      "step": 8432
    },
    {
      "epoch": 3.936056009334889,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013286264430711637,
      "loss": 0.0121,
      "step": 8433
    },
    {
      "epoch": 3.936522753792299,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00013284878543611589,
      "loss": 0.0228,
      "step": 8434
    },
    {
      "epoch": 3.9369894982497082,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013283492585784457,
      "loss": 0.0148,
      "step": 8435
    },
    {
      "epoch": 3.9374562427071176,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00013282106557260088,
      "loss": 0.016,
      "step": 8436
    },
    {
      "epoch": 3.9379229871645274,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013280720458068324,
      "loss": 0.0113,
      "step": 8437
    },
    {
      "epoch": 3.938389731621937,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013279334288239007,
      "loss": 0.0154,
      "step": 8438
    },
    {
      "epoch": 3.9388564760793465,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013277948047801984,
      "loss": 0.0176,
      "step": 8439
    },
    {
      "epoch": 3.939323220536756,
      "grad_norm": 0.84375,
      "learning_rate": 0.00013276561736787107,
      "loss": 0.0229,
      "step": 8440
    },
    {
      "epoch": 3.9397899649941657,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00013275175355224213,
      "loss": 0.0103,
      "step": 8441
    },
    {
      "epoch": 3.9402567094515755,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00013273788903143166,
      "loss": 0.0351,
      "step": 8442
    },
    {
      "epoch": 3.940723453908985,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001327240238057381,
      "loss": 0.0092,
      "step": 8443
    },
    {
      "epoch": 3.941190198366394,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013271015787545996,
      "loss": 0.0102,
      "step": 8444
    },
    {
      "epoch": 3.941656942823804,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00013269629124089587,
      "loss": 0.0202,
      "step": 8445
    },
    {
      "epoch": 3.9421236872812138,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00013268242390234437,
      "loss": 0.012,
      "step": 8446
    },
    {
      "epoch": 3.942590431738623,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00013266855586010397,
      "loss": 0.0066,
      "step": 8447
    },
    {
      "epoch": 3.9430571761960325,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013265468711447335,
      "loss": 0.0109,
      "step": 8448
    },
    {
      "epoch": 3.9435239206534423,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00013264081766575108,
      "loss": 0.0104,
      "step": 8449
    },
    {
      "epoch": 3.943990665110852,
      "grad_norm": 0.318359375,
      "learning_rate": 0.0001326269475142358,
      "loss": 0.0081,
      "step": 8450
    },
    {
      "epoch": 3.9444574095682614,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001326130766602261,
      "loss": 0.0131,
      "step": 8451
    },
    {
      "epoch": 3.9449241540256708,
      "grad_norm": 0.546875,
      "learning_rate": 0.00013259920510402072,
      "loss": 0.009,
      "step": 8452
    },
    {
      "epoch": 3.9453908984830806,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013258533284591826,
      "loss": 0.0197,
      "step": 8453
    },
    {
      "epoch": 3.9458576429404904,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001325714598862174,
      "loss": 0.0131,
      "step": 8454
    },
    {
      "epoch": 3.9463243873978997,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00013255758622521694,
      "loss": 0.0127,
      "step": 8455
    },
    {
      "epoch": 3.946791131855309,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00013254371186321548,
      "loss": 0.0184,
      "step": 8456
    },
    {
      "epoch": 3.947257876312719,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00013252983680051182,
      "loss": 0.0183,
      "step": 8457
    },
    {
      "epoch": 3.947724620770128,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00013251596103740467,
      "loss": 0.0222,
      "step": 8458
    },
    {
      "epoch": 3.948191365227538,
      "grad_norm": 0.6953125,
      "learning_rate": 0.0001325020845741928,
      "loss": 0.0162,
      "step": 8459
    },
    {
      "epoch": 3.9486581096849473,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013248820741117498,
      "loss": 0.0151,
      "step": 8460
    },
    {
      "epoch": 3.949124854142357,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013247432954865005,
      "loss": 0.0141,
      "step": 8461
    },
    {
      "epoch": 3.9495915985997665,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013246045098691672,
      "loss": 0.0133,
      "step": 8462
    },
    {
      "epoch": 3.9500583430571763,
      "grad_norm": 0.84765625,
      "learning_rate": 0.0001324465717262739,
      "loss": 0.0164,
      "step": 8463
    },
    {
      "epoch": 3.9505250875145856,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013243269176702036,
      "loss": 0.0142,
      "step": 8464
    },
    {
      "epoch": 3.9509918319719954,
      "grad_norm": 0.494140625,
      "learning_rate": 0.000132418811109455,
      "loss": 0.0106,
      "step": 8465
    },
    {
      "epoch": 3.951458576429405,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00013240492975387667,
      "loss": 0.013,
      "step": 8466
    },
    {
      "epoch": 3.9519253208868146,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013239104770058425,
      "loss": 0.0112,
      "step": 8467
    },
    {
      "epoch": 3.952392065344224,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013237716494987663,
      "loss": 0.0181,
      "step": 8468
    },
    {
      "epoch": 3.9528588098016337,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001323632815020527,
      "loss": 0.0127,
      "step": 8469
    },
    {
      "epoch": 3.953325554259043,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013234939735741146,
      "loss": 0.019,
      "step": 8470
    },
    {
      "epoch": 3.953792298716453,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00013233551251625177,
      "loss": 0.027,
      "step": 8471
    },
    {
      "epoch": 3.9542590431738622,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001323216269788726,
      "loss": 0.0181,
      "step": 8472
    },
    {
      "epoch": 3.954725787631272,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013230774074557297,
      "loss": 0.0088,
      "step": 8473
    },
    {
      "epoch": 3.9551925320886814,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00013229385381665184,
      "loss": 0.0128,
      "step": 8474
    },
    {
      "epoch": 3.955659276546091,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013227996619240818,
      "loss": 0.0125,
      "step": 8475
    },
    {
      "epoch": 3.9561260210035005,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013226607787314107,
      "loss": 0.0149,
      "step": 8476
    },
    {
      "epoch": 3.9565927654609103,
      "grad_norm": 0.796875,
      "learning_rate": 0.00013225218885914948,
      "loss": 0.0219,
      "step": 8477
    },
    {
      "epoch": 3.9570595099183197,
      "grad_norm": 0.67578125,
      "learning_rate": 0.0001322382991507325,
      "loss": 0.0198,
      "step": 8478
    },
    {
      "epoch": 3.957526254375729,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013222440874818914,
      "loss": 0.0142,
      "step": 8479
    },
    {
      "epoch": 3.957992998833139,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001322105176518185,
      "loss": 0.023,
      "step": 8480
    },
    {
      "epoch": 3.9584597432905486,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013219662586191972,
      "loss": 0.0095,
      "step": 8481
    },
    {
      "epoch": 3.958926487747958,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00013218273337879186,
      "loss": 0.0105,
      "step": 8482
    },
    {
      "epoch": 3.9593932322053673,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000132168840202734,
      "loss": 0.0214,
      "step": 8483
    },
    {
      "epoch": 3.959859976662777,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013215494633404536,
      "loss": 0.0215,
      "step": 8484
    },
    {
      "epoch": 3.960326721120187,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00013214105177302503,
      "loss": 0.0197,
      "step": 8485
    },
    {
      "epoch": 3.9607934655775963,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013212715651997217,
      "loss": 0.0151,
      "step": 8486
    },
    {
      "epoch": 3.9612602100350056,
      "grad_norm": 0.4375,
      "learning_rate": 0.000132113260575186,
      "loss": 0.0123,
      "step": 8487
    },
    {
      "epoch": 3.9617269544924154,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00013209936393896574,
      "loss": 0.0156,
      "step": 8488
    },
    {
      "epoch": 3.962193698949825,
      "grad_norm": 0.74609375,
      "learning_rate": 0.0001320854666116105,
      "loss": 0.0209,
      "step": 8489
    },
    {
      "epoch": 3.9626604434072346,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001320715685934196,
      "loss": 0.0119,
      "step": 8490
    },
    {
      "epoch": 3.963127187864644,
      "grad_norm": 0.8359375,
      "learning_rate": 0.00013205766988469225,
      "loss": 0.0206,
      "step": 8491
    },
    {
      "epoch": 3.9635939323220537,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013204377048572766,
      "loss": 0.0102,
      "step": 8492
    },
    {
      "epoch": 3.9640606767794635,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013202987039682515,
      "loss": 0.016,
      "step": 8493
    },
    {
      "epoch": 3.964527421236873,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00013201596961828402,
      "loss": 0.0159,
      "step": 8494
    },
    {
      "epoch": 3.964994165694282,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013200206815040355,
      "loss": 0.0159,
      "step": 8495
    },
    {
      "epoch": 3.965460910151692,
      "grad_norm": 0.51953125,
      "learning_rate": 0.000131988165993483,
      "loss": 0.0111,
      "step": 8496
    },
    {
      "epoch": 3.965927654609102,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013197426314782176,
      "loss": 0.0163,
      "step": 8497
    },
    {
      "epoch": 3.966394399066511,
      "grad_norm": 0.5625,
      "learning_rate": 0.00013196035961371918,
      "loss": 0.0165,
      "step": 8498
    },
    {
      "epoch": 3.9668611435239205,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00013194645539147458,
      "loss": 0.0233,
      "step": 8499
    },
    {
      "epoch": 3.9673278879813303,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00013193255048138736,
      "loss": 0.0085,
      "step": 8500
    },
    {
      "epoch": 3.9677946324387396,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00013191864488375688,
      "loss": 0.0133,
      "step": 8501
    },
    {
      "epoch": 3.9682613768961494,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00013190473859888256,
      "loss": 0.0197,
      "step": 8502
    },
    {
      "epoch": 3.968728121353559,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00013189083162706383,
      "loss": 0.021,
      "step": 8503
    },
    {
      "epoch": 3.9691948658109686,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00013187692396860015,
      "loss": 0.0156,
      "step": 8504
    },
    {
      "epoch": 3.969661610268378,
      "grad_norm": 0.859375,
      "learning_rate": 0.0001318630156237909,
      "loss": 0.0165,
      "step": 8505
    },
    {
      "epoch": 3.9701283547257877,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00013184910659293553,
      "loss": 0.0115,
      "step": 8506
    },
    {
      "epoch": 3.970595099183197,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001318351968763336,
      "loss": 0.0149,
      "step": 8507
    },
    {
      "epoch": 3.971061843640607,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00013182128647428457,
      "loss": 0.0135,
      "step": 8508
    },
    {
      "epoch": 3.9715285880980162,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001318073753870879,
      "loss": 0.0131,
      "step": 8509
    },
    {
      "epoch": 3.971995332555426,
      "grad_norm": 1.015625,
      "learning_rate": 0.00013179346361504317,
      "loss": 0.0212,
      "step": 8510
    },
    {
      "epoch": 3.9724620770128354,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013177955115844988,
      "loss": 0.0212,
      "step": 8511
    },
    {
      "epoch": 3.972928821470245,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013176563801760763,
      "loss": 0.0152,
      "step": 8512
    },
    {
      "epoch": 3.9733955659276545,
      "grad_norm": 0.421875,
      "learning_rate": 0.0001317517241928159,
      "loss": 0.0172,
      "step": 8513
    },
    {
      "epoch": 3.9738623103850643,
      "grad_norm": 0.515625,
      "learning_rate": 0.00013173780968437436,
      "loss": 0.0126,
      "step": 8514
    },
    {
      "epoch": 3.9743290548424737,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00013172389449258253,
      "loss": 0.0169,
      "step": 8515
    },
    {
      "epoch": 3.9747957992998835,
      "grad_norm": 0.578125,
      "learning_rate": 0.00013170997861774005,
      "loss": 0.0187,
      "step": 8516
    },
    {
      "epoch": 3.975262543757293,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00013169606206014654,
      "loss": 0.008,
      "step": 8517
    },
    {
      "epoch": 3.975729288214702,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013168214482010167,
      "loss": 0.016,
      "step": 8518
    },
    {
      "epoch": 3.976196032672112,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00013166822689790507,
      "loss": 0.0131,
      "step": 8519
    },
    {
      "epoch": 3.9766627771295218,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013165430829385644,
      "loss": 0.021,
      "step": 8520
    },
    {
      "epoch": 3.977129521586931,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00013164038900825538,
      "loss": 0.0199,
      "step": 8521
    },
    {
      "epoch": 3.9775962660443405,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013162646904140167,
      "loss": 0.0099,
      "step": 8522
    },
    {
      "epoch": 3.9780630105017503,
      "grad_norm": 0.53125,
      "learning_rate": 0.00013161254839359497,
      "loss": 0.0179,
      "step": 8523
    },
    {
      "epoch": 3.97852975495916,
      "grad_norm": 0.625,
      "learning_rate": 0.00013159862706513503,
      "loss": 0.0141,
      "step": 8524
    },
    {
      "epoch": 3.9789964994165694,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00013158470505632161,
      "loss": 0.0122,
      "step": 8525
    },
    {
      "epoch": 3.9794632438739788,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00013157078236745442,
      "loss": 0.0122,
      "step": 8526
    },
    {
      "epoch": 3.9799299883313886,
      "grad_norm": 0.609375,
      "learning_rate": 0.00013155685899883329,
      "loss": 0.0164,
      "step": 8527
    },
    {
      "epoch": 3.9803967327887984,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00013154293495075794,
      "loss": 0.0131,
      "step": 8528
    },
    {
      "epoch": 3.9808634772462077,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001315290102235282,
      "loss": 0.0135,
      "step": 8529
    },
    {
      "epoch": 3.981330221703617,
      "grad_norm": 0.76953125,
      "learning_rate": 0.0001315150848174439,
      "loss": 0.02,
      "step": 8530
    },
    {
      "epoch": 3.981796966161027,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001315011587328049,
      "loss": 0.0092,
      "step": 8531
    },
    {
      "epoch": 3.9822637106184366,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00013148723196991095,
      "loss": 0.0114,
      "step": 8532
    },
    {
      "epoch": 3.982730455075846,
      "grad_norm": 0.7578125,
      "learning_rate": 0.00013147330452906197,
      "loss": 0.0229,
      "step": 8533
    },
    {
      "epoch": 3.9831971995332554,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00013145937641055784,
      "loss": 0.0102,
      "step": 8534
    },
    {
      "epoch": 3.983663943990665,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00013144544761469843,
      "loss": 0.0143,
      "step": 8535
    },
    {
      "epoch": 3.984130688448075,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001314315181417836,
      "loss": 0.02,
      "step": 8536
    },
    {
      "epoch": 3.9845974329054843,
      "grad_norm": 0.25390625,
      "learning_rate": 0.00013141758799211338,
      "loss": 0.0117,
      "step": 8537
    },
    {
      "epoch": 3.9850641773628936,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00013140365716598758,
      "loss": 0.0158,
      "step": 8538
    },
    {
      "epoch": 3.9855309218203034,
      "grad_norm": 0.55078125,
      "learning_rate": 0.0001313897256637062,
      "loss": 0.0107,
      "step": 8539
    },
    {
      "epoch": 3.985997666277713,
      "grad_norm": 0.46875,
      "learning_rate": 0.0001313757934855692,
      "loss": 0.0118,
      "step": 8540
    },
    {
      "epoch": 3.9864644107351226,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001313618606318766,
      "loss": 0.009,
      "step": 8541
    },
    {
      "epoch": 3.986931155192532,
      "grad_norm": 0.8515625,
      "learning_rate": 0.00013134792710292828,
      "loss": 0.0123,
      "step": 8542
    },
    {
      "epoch": 3.9873978996499417,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00013133399289902436,
      "loss": 0.0061,
      "step": 8543
    },
    {
      "epoch": 3.987864644107351,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013132005802046477,
      "loss": 0.0093,
      "step": 8544
    },
    {
      "epoch": 3.988331388564761,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00013130612246754962,
      "loss": 0.0096,
      "step": 8545
    },
    {
      "epoch": 3.9887981330221702,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00013129218624057887,
      "loss": 0.0154,
      "step": 8546
    },
    {
      "epoch": 3.98926487747958,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013127824933985264,
      "loss": 0.0166,
      "step": 8547
    },
    {
      "epoch": 3.9897316219369894,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00013126431176567103,
      "loss": 0.0104,
      "step": 8548
    },
    {
      "epoch": 3.990198366394399,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00013125037351833407,
      "loss": 0.0172,
      "step": 8549
    },
    {
      "epoch": 3.9906651108518085,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001312364345981419,
      "loss": 0.0134,
      "step": 8550
    },
    {
      "epoch": 3.9911318553092183,
      "grad_norm": 0.65625,
      "learning_rate": 0.00013122249500539468,
      "loss": 0.0148,
      "step": 8551
    },
    {
      "epoch": 3.9915985997666277,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001312085547403924,
      "loss": 0.0118,
      "step": 8552
    },
    {
      "epoch": 3.9920653442240375,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00013119461380343544,
      "loss": 0.013,
      "step": 8553
    },
    {
      "epoch": 3.992532088681447,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00013118067219482378,
      "loss": 0.0104,
      "step": 8554
    },
    {
      "epoch": 3.9929988331388566,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001311667299148576,
      "loss": 0.0138,
      "step": 8555
    },
    {
      "epoch": 3.993465577596266,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001311527869638372,
      "loss": 0.0142,
      "step": 8556
    },
    {
      "epoch": 3.9939323220536758,
      "grad_norm": 0.625,
      "learning_rate": 0.00013113884334206273,
      "loss": 0.011,
      "step": 8557
    },
    {
      "epoch": 3.994399066511085,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013112489904983438,
      "loss": 0.0159,
      "step": 8558
    },
    {
      "epoch": 3.994865810968495,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00013111095408745244,
      "loss": 0.013,
      "step": 8559
    },
    {
      "epoch": 3.9953325554259043,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00013109700845521715,
      "loss": 0.0115,
      "step": 8560
    },
    {
      "epoch": 3.9957992998833136,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00013108306215342877,
      "loss": 0.0231,
      "step": 8561
    },
    {
      "epoch": 3.9962660443407234,
      "grad_norm": 0.640625,
      "learning_rate": 0.00013106911518238754,
      "loss": 0.015,
      "step": 8562
    },
    {
      "epoch": 3.996732788798133,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001310551675423938,
      "loss": 0.0244,
      "step": 8563
    },
    {
      "epoch": 3.9971995332555426,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001310412192337479,
      "loss": 0.009,
      "step": 8564
    },
    {
      "epoch": 3.997666277712952,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00013102727025675,
      "loss": 0.0139,
      "step": 8565
    },
    {
      "epoch": 3.9981330221703617,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00013101332061170063,
      "loss": 0.0125,
      "step": 8566
    },
    {
      "epoch": 3.9985997666277715,
      "grad_norm": 0.625,
      "learning_rate": 0.00013099937029890005,
      "loss": 0.0176,
      "step": 8567
    },
    {
      "epoch": 3.999066511085181,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00013098541931864862,
      "loss": 0.0176,
      "step": 8568
    },
    {
      "epoch": 3.99953325554259,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00013097146767124672,
      "loss": 0.0079,
      "step": 8569
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00013095751535699478,
      "loss": 0.0104,
      "step": 8570
    },
    {
      "epoch": 4.00046674445741,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00013094356237619318,
      "loss": 0.0116,
      "step": 8571
    },
    {
      "epoch": 4.000933488914819,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00013092960872914232,
      "loss": 0.0063,
      "step": 8572
    },
    {
      "epoch": 4.0014002333722285,
      "grad_norm": 0.482421875,
      "learning_rate": 0.0001309156544161427,
      "loss": 0.0091,
      "step": 8573
    },
    {
      "epoch": 4.001866977829638,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00013090169943749476,
      "loss": 0.0108,
      "step": 8574
    },
    {
      "epoch": 4.002333722287048,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001308877437934989,
      "loss": 0.0154,
      "step": 8575
    },
    {
      "epoch": 4.002800466744457,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00013087378748445567,
      "loss": 0.0134,
      "step": 8576
    },
    {
      "epoch": 4.002800466744457,
      "eval_loss": 1.9588065147399902,
      "eval_runtime": 55.9281,
      "eval_samples_per_second": 32.256,
      "eval_steps_per_second": 4.041,
      "step": 8576
    },
    {
      "epoch": 4.003267211201867,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013085983051066554,
      "loss": 0.0153,
      "step": 8577
    },
    {
      "epoch": 4.003733955659277,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00013084587287242902,
      "loss": 0.0075,
      "step": 8578
    },
    {
      "epoch": 4.004200700116686,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00013083191457004663,
      "loss": 0.0146,
      "step": 8579
    },
    {
      "epoch": 4.004667444574095,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001308179556038189,
      "loss": 0.0105,
      "step": 8580
    },
    {
      "epoch": 4.005134189031505,
      "grad_norm": 0.70703125,
      "learning_rate": 0.0001308039959740464,
      "loss": 0.0206,
      "step": 8581
    },
    {
      "epoch": 4.005600933488915,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001307900356810297,
      "loss": 0.0049,
      "step": 8582
    },
    {
      "epoch": 4.006067677946325,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00013077607472506938,
      "loss": 0.0154,
      "step": 8583
    },
    {
      "epoch": 4.006534422403734,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00013076211310646599,
      "loss": 0.0071,
      "step": 8584
    },
    {
      "epoch": 4.007001166861143,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013074815082552015,
      "loss": 0.0124,
      "step": 8585
    },
    {
      "epoch": 4.007467911318553,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00013073418788253256,
      "loss": 0.0116,
      "step": 8586
    },
    {
      "epoch": 4.007934655775963,
      "grad_norm": 0.380859375,
      "learning_rate": 0.0001307202242778038,
      "loss": 0.0046,
      "step": 8587
    },
    {
      "epoch": 4.008401400233372,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001307062600116345,
      "loss": 0.0066,
      "step": 8588
    },
    {
      "epoch": 4.008868144690782,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00013069229508432537,
      "loss": 0.0205,
      "step": 8589
    },
    {
      "epoch": 4.0093348891481915,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001306783294961771,
      "loss": 0.0109,
      "step": 8590
    },
    {
      "epoch": 4.009801633605601,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001306643632474903,
      "loss": 0.007,
      "step": 8591
    },
    {
      "epoch": 4.01026837806301,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00013065039633856579,
      "loss": 0.0132,
      "step": 8592
    },
    {
      "epoch": 4.01073512252042,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001306364287697042,
      "loss": 0.0073,
      "step": 8593
    },
    {
      "epoch": 4.01120186697783,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00013062246054120632,
      "loss": 0.0086,
      "step": 8594
    },
    {
      "epoch": 4.01166861143524,
      "grad_norm": 0.59375,
      "learning_rate": 0.00013060849165337287,
      "loss": 0.0124,
      "step": 8595
    },
    {
      "epoch": 4.0121353558926485,
      "grad_norm": 0.5,
      "learning_rate": 0.00013059452210650467,
      "loss": 0.0092,
      "step": 8596
    },
    {
      "epoch": 4.012602100350058,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00013058055190090246,
      "loss": 0.0086,
      "step": 8597
    },
    {
      "epoch": 4.013068844807468,
      "grad_norm": 0.546875,
      "learning_rate": 0.000130566581036867,
      "loss": 0.0099,
      "step": 8598
    },
    {
      "epoch": 4.013535589264878,
      "grad_norm": 0.421875,
      "learning_rate": 0.00013055260951469916,
      "loss": 0.0106,
      "step": 8599
    },
    {
      "epoch": 4.014002333722287,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00013053863733469978,
      "loss": 0.0102,
      "step": 8600
    },
    {
      "epoch": 4.014469078179697,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001305246644971696,
      "loss": 0.0108,
      "step": 8601
    },
    {
      "epoch": 4.014935822637106,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00013051069100240956,
      "loss": 0.0038,
      "step": 8602
    },
    {
      "epoch": 4.015402567094516,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00013049671685072046,
      "loss": 0.0073,
      "step": 8603
    },
    {
      "epoch": 4.015869311551925,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001304827420424032,
      "loss": 0.0077,
      "step": 8604
    },
    {
      "epoch": 4.016336056009335,
      "grad_norm": 0.5,
      "learning_rate": 0.00013046876657775872,
      "loss": 0.0104,
      "step": 8605
    },
    {
      "epoch": 4.016802800466745,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001304547904570879,
      "loss": 0.0232,
      "step": 8606
    },
    {
      "epoch": 4.0172695449241544,
      "grad_norm": 0.21875,
      "learning_rate": 0.0001304408136806916,
      "loss": 0.0029,
      "step": 8607
    },
    {
      "epoch": 4.017736289381563,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00013042683624887085,
      "loss": 0.0099,
      "step": 8608
    },
    {
      "epoch": 4.018203033838973,
      "grad_norm": 0.328125,
      "learning_rate": 0.00013041285816192653,
      "loss": 0.0059,
      "step": 8609
    },
    {
      "epoch": 4.018669778296383,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00013039887942015965,
      "loss": 0.0104,
      "step": 8610
    },
    {
      "epoch": 4.019136522753793,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00013038490002387115,
      "loss": 0.0083,
      "step": 8611
    },
    {
      "epoch": 4.019603267211202,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00013037091997336204,
      "loss": 0.0106,
      "step": 8612
    },
    {
      "epoch": 4.020070011668611,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00013035693926893334,
      "loss": 0.0092,
      "step": 8613
    },
    {
      "epoch": 4.020536756126021,
      "grad_norm": 0.5859375,
      "learning_rate": 0.000130342957910886,
      "loss": 0.0135,
      "step": 8614
    },
    {
      "epoch": 4.02100350058343,
      "grad_norm": 0.66796875,
      "learning_rate": 0.0001303289758995212,
      "loss": 0.014,
      "step": 8615
    },
    {
      "epoch": 4.02147024504084,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00013031499323513986,
      "loss": 0.0092,
      "step": 8616
    },
    {
      "epoch": 4.02193698949825,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00013030100991804307,
      "loss": 0.0099,
      "step": 8617
    },
    {
      "epoch": 4.0224037339556595,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001302870259485319,
      "loss": 0.0095,
      "step": 8618
    },
    {
      "epoch": 4.022870478413068,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001302730413269075,
      "loss": 0.0083,
      "step": 8619
    },
    {
      "epoch": 4.023337222870478,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00013025905605347092,
      "loss": 0.0142,
      "step": 8620
    },
    {
      "epoch": 4.023803967327888,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00013024507012852327,
      "loss": 0.0087,
      "step": 8621
    },
    {
      "epoch": 4.024270711785298,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00013023108355236573,
      "loss": 0.0081,
      "step": 8622
    },
    {
      "epoch": 4.024737456242707,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00013021709632529943,
      "loss": 0.0041,
      "step": 8623
    },
    {
      "epoch": 4.0252042007001165,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00013020310844762543,
      "loss": 0.0079,
      "step": 8624
    },
    {
      "epoch": 4.025670945157526,
      "grad_norm": 0.296875,
      "learning_rate": 0.0001301891199196451,
      "loss": 0.0187,
      "step": 8625
    },
    {
      "epoch": 4.026137689614936,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001301751307416595,
      "loss": 0.0051,
      "step": 8626
    },
    {
      "epoch": 4.026604434072345,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001301611409139698,
      "loss": 0.0111,
      "step": 8627
    },
    {
      "epoch": 4.027071178529755,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00013014715043687734,
      "loss": 0.0062,
      "step": 8628
    },
    {
      "epoch": 4.027537922987165,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00013013315931068324,
      "loss": 0.0068,
      "step": 8629
    },
    {
      "epoch": 4.028004667444574,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001301191675356888,
      "loss": 0.0062,
      "step": 8630
    },
    {
      "epoch": 4.028471411901983,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00013010517511219527,
      "loss": 0.0134,
      "step": 8631
    },
    {
      "epoch": 4.028938156359393,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001300911820405039,
      "loss": 0.0115,
      "step": 8632
    },
    {
      "epoch": 4.029404900816803,
      "grad_norm": 0.453125,
      "learning_rate": 0.000130077188320916,
      "loss": 0.007,
      "step": 8633
    },
    {
      "epoch": 4.029871645274213,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00013006319395373285,
      "loss": 0.0074,
      "step": 8634
    },
    {
      "epoch": 4.030338389731622,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00013004919893925583,
      "loss": 0.01,
      "step": 8635
    },
    {
      "epoch": 4.030805134189031,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00013003520327778618,
      "loss": 0.0094,
      "step": 8636
    },
    {
      "epoch": 4.031271878646441,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00013002120696962524,
      "loss": 0.0073,
      "step": 8637
    },
    {
      "epoch": 4.031738623103851,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00013000721001507447,
      "loss": 0.0058,
      "step": 8638
    },
    {
      "epoch": 4.03220536756126,
      "grad_norm": 0.328125,
      "learning_rate": 0.00012999321241443515,
      "loss": 0.0078,
      "step": 8639
    },
    {
      "epoch": 4.03267211201867,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001299792141680087,
      "loss": 0.0093,
      "step": 8640
    },
    {
      "epoch": 4.0331388564760795,
      "grad_norm": 0.546875,
      "learning_rate": 0.0001299652152760965,
      "loss": 0.0109,
      "step": 8641
    },
    {
      "epoch": 4.033605600933489,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00012995121573899996,
      "loss": 0.0065,
      "step": 8642
    },
    {
      "epoch": 4.034072345390898,
      "grad_norm": 0.158203125,
      "learning_rate": 0.00012993721555702053,
      "loss": 0.0022,
      "step": 8643
    },
    {
      "epoch": 4.034539089848308,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00012992321473045962,
      "loss": 0.0064,
      "step": 8644
    },
    {
      "epoch": 4.035005834305718,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0001299092132596187,
      "loss": 0.0072,
      "step": 8645
    },
    {
      "epoch": 4.035472578763128,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00012989521114479922,
      "loss": 0.0082,
      "step": 8646
    },
    {
      "epoch": 4.0359393232205365,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001298812083863027,
      "loss": 0.008,
      "step": 8647
    },
    {
      "epoch": 4.036406067677946,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001298672049844306,
      "loss": 0.0156,
      "step": 8648
    },
    {
      "epoch": 4.036872812135356,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00012985320093948446,
      "loss": 0.0111,
      "step": 8649
    },
    {
      "epoch": 4.037339556592766,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012983919625176574,
      "loss": 0.0095,
      "step": 8650
    },
    {
      "epoch": 4.037806301050175,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012982519092157605,
      "loss": 0.011,
      "step": 8651
    },
    {
      "epoch": 4.038273045507585,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00012981118494921688,
      "loss": 0.0093,
      "step": 8652
    },
    {
      "epoch": 4.038739789964994,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001297971783349898,
      "loss": 0.0077,
      "step": 8653
    },
    {
      "epoch": 4.039206534422403,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00012978317107919645,
      "loss": 0.0169,
      "step": 8654
    },
    {
      "epoch": 4.039673278879813,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00012976916318213839,
      "loss": 0.0119,
      "step": 8655
    },
    {
      "epoch": 4.040140023337223,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00012975515464411718,
      "loss": 0.0139,
      "step": 8656
    },
    {
      "epoch": 4.040606767794633,
      "grad_norm": 0.86328125,
      "learning_rate": 0.00012974114546543451,
      "loss": 0.0143,
      "step": 8657
    },
    {
      "epoch": 4.041073512252042,
      "grad_norm": 0.5625,
      "learning_rate": 0.000129727135646392,
      "loss": 0.0121,
      "step": 8658
    },
    {
      "epoch": 4.041540256709451,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00012971312518729125,
      "loss": 0.0075,
      "step": 8659
    },
    {
      "epoch": 4.042007001166861,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012969911408843392,
      "loss": 0.0133,
      "step": 8660
    },
    {
      "epoch": 4.042473745624271,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012968510235012172,
      "loss": 0.0089,
      "step": 8661
    },
    {
      "epoch": 4.04294049008168,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00012967108997265635,
      "loss": 0.0063,
      "step": 8662
    },
    {
      "epoch": 4.04340723453909,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00012965707695633948,
      "loss": 0.0106,
      "step": 8663
    },
    {
      "epoch": 4.0438739789964995,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00012964306330147287,
      "loss": 0.0065,
      "step": 8664
    },
    {
      "epoch": 4.044340723453909,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00012962904900835816,
      "loss": 0.0128,
      "step": 8665
    },
    {
      "epoch": 4.044807467911318,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00012961503407729722,
      "loss": 0.0079,
      "step": 8666
    },
    {
      "epoch": 4.045274212368728,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012960101850859168,
      "loss": 0.0112,
      "step": 8667
    },
    {
      "epoch": 4.045740956826138,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001295870023025434,
      "loss": 0.0096,
      "step": 8668
    },
    {
      "epoch": 4.046207701283548,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012957298545945414,
      "loss": 0.0045,
      "step": 8669
    },
    {
      "epoch": 4.0466744457409565,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00012955896797962562,
      "loss": 0.0078,
      "step": 8670
    },
    {
      "epoch": 4.047141190198366,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001295449498633598,
      "loss": 0.0117,
      "step": 8671
    },
    {
      "epoch": 4.047607934655776,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00012953093111095842,
      "loss": 0.009,
      "step": 8672
    },
    {
      "epoch": 4.048074679113186,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012951691172272334,
      "loss": 0.01,
      "step": 8673
    },
    {
      "epoch": 4.048541423570595,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012950289169895638,
      "loss": 0.0086,
      "step": 8674
    },
    {
      "epoch": 4.049008168028005,
      "grad_norm": 0.193359375,
      "learning_rate": 0.0001294888710399594,
      "loss": 0.0029,
      "step": 8675
    },
    {
      "epoch": 4.049474912485414,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012947484974603433,
      "loss": 0.0082,
      "step": 8676
    },
    {
      "epoch": 4.049941656942824,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00012946082781748304,
      "loss": 0.0051,
      "step": 8677
    },
    {
      "epoch": 4.050408401400233,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00012944680525460745,
      "loss": 0.0105,
      "step": 8678
    },
    {
      "epoch": 4.050875145857643,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012943278205770946,
      "loss": 0.011,
      "step": 8679
    },
    {
      "epoch": 4.051341890315053,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012941875822709102,
      "loss": 0.0105,
      "step": 8680
    },
    {
      "epoch": 4.0518086347724624,
      "grad_norm": 0.1728515625,
      "learning_rate": 0.0001294047337630541,
      "loss": 0.0057,
      "step": 8681
    },
    {
      "epoch": 4.052275379229871,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012939070866590057,
      "loss": 0.01,
      "step": 8682
    },
    {
      "epoch": 4.052742123687281,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012937668293593253,
      "loss": 0.0084,
      "step": 8683
    },
    {
      "epoch": 4.053208868144691,
      "grad_norm": 0.421875,
      "learning_rate": 0.0001293626565734519,
      "loss": 0.0118,
      "step": 8684
    },
    {
      "epoch": 4.053675612602101,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001293486295787607,
      "loss": 0.0092,
      "step": 8685
    },
    {
      "epoch": 4.05414235705951,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001293346019521609,
      "loss": 0.0112,
      "step": 8686
    },
    {
      "epoch": 4.0546091015169194,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00012932057369395463,
      "loss": 0.0179,
      "step": 8687
    },
    {
      "epoch": 4.055075845974329,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012930654480444387,
      "loss": 0.0121,
      "step": 8688
    },
    {
      "epoch": 4.055542590431739,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00012929251528393066,
      "loss": 0.015,
      "step": 8689
    },
    {
      "epoch": 4.056009334889148,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001292784851327171,
      "loss": 0.0065,
      "step": 8690
    },
    {
      "epoch": 4.056476079346558,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012926445435110527,
      "loss": 0.0132,
      "step": 8691
    },
    {
      "epoch": 4.0569428238039675,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012925042293939726,
      "loss": 0.0123,
      "step": 8692
    },
    {
      "epoch": 4.057409568261377,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001292363908978952,
      "loss": 0.011,
      "step": 8693
    },
    {
      "epoch": 4.057876312718786,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00012922235822690123,
      "loss": 0.0045,
      "step": 8694
    },
    {
      "epoch": 4.058343057176196,
      "grad_norm": 0.4375,
      "learning_rate": 0.00012920832492671744,
      "loss": 0.0064,
      "step": 8695
    },
    {
      "epoch": 4.058809801633606,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00012919429099764602,
      "loss": 0.0111,
      "step": 8696
    },
    {
      "epoch": 4.059276546091015,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012918025643998912,
      "loss": 0.0107,
      "step": 8697
    },
    {
      "epoch": 4.0597432905484245,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00012916622125404889,
      "loss": 0.0063,
      "step": 8698
    },
    {
      "epoch": 4.060210035005834,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00012915218544012758,
      "loss": 0.0052,
      "step": 8699
    },
    {
      "epoch": 4.060676779463244,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001291381489985274,
      "loss": 0.0051,
      "step": 8700
    },
    {
      "epoch": 4.061143523920653,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00012912411192955053,
      "loss": 0.0056,
      "step": 8701
    },
    {
      "epoch": 4.061610268378063,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001291100742334992,
      "loss": 0.0054,
      "step": 8702
    },
    {
      "epoch": 4.062077012835473,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012909603591067565,
      "loss": 0.0065,
      "step": 8703
    },
    {
      "epoch": 4.062543757292882,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001290819969613822,
      "loss": 0.0062,
      "step": 8704
    },
    {
      "epoch": 4.063010501750291,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012906795738592107,
      "loss": 0.0089,
      "step": 8705
    },
    {
      "epoch": 4.063477246207701,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001290539171845946,
      "loss": 0.0112,
      "step": 8706
    },
    {
      "epoch": 4.063943990665111,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00012903987635770503,
      "loss": 0.0077,
      "step": 8707
    },
    {
      "epoch": 4.064410735122521,
      "grad_norm": 0.396484375,
      "learning_rate": 0.0001290258349055547,
      "loss": 0.0069,
      "step": 8708
    },
    {
      "epoch": 4.06487747957993,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00012901179282844594,
      "loss": 0.0035,
      "step": 8709
    },
    {
      "epoch": 4.065344224037339,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001289977501266811,
      "loss": 0.0093,
      "step": 8710
    },
    {
      "epoch": 4.065810968494749,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012898370680056252,
      "loss": 0.0138,
      "step": 8711
    },
    {
      "epoch": 4.066277712952159,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00012896966285039257,
      "loss": 0.0054,
      "step": 8712
    },
    {
      "epoch": 4.066744457409568,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00012895561827647365,
      "loss": 0.0126,
      "step": 8713
    },
    {
      "epoch": 4.067211201866978,
      "grad_norm": 0.453125,
      "learning_rate": 0.00012894157307910813,
      "loss": 0.0053,
      "step": 8714
    },
    {
      "epoch": 4.0676779463243875,
      "grad_norm": 0.421875,
      "learning_rate": 0.00012892752725859842,
      "loss": 0.0054,
      "step": 8715
    },
    {
      "epoch": 4.068144690781797,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00012891348081524697,
      "loss": 0.0085,
      "step": 8716
    },
    {
      "epoch": 4.068611435239206,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001288994337493562,
      "loss": 0.0087,
      "step": 8717
    },
    {
      "epoch": 4.069078179696616,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00012888538606122856,
      "loss": 0.0067,
      "step": 8718
    },
    {
      "epoch": 4.069544924154026,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001288713377511665,
      "loss": 0.0049,
      "step": 8719
    },
    {
      "epoch": 4.070011668611436,
      "grad_norm": 0.375,
      "learning_rate": 0.00012885728881947253,
      "loss": 0.0051,
      "step": 8720
    },
    {
      "epoch": 4.0704784130688445,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012884323926644906,
      "loss": 0.0062,
      "step": 8721
    },
    {
      "epoch": 4.070945157526254,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001288291890923987,
      "loss": 0.007,
      "step": 8722
    },
    {
      "epoch": 4.071411901983664,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012881513829762387,
      "loss": 0.0059,
      "step": 8723
    },
    {
      "epoch": 4.071878646441074,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00012880108688242717,
      "loss": 0.0107,
      "step": 8724
    },
    {
      "epoch": 4.072345390898483,
      "grad_norm": 0.375,
      "learning_rate": 0.0001287870348471111,
      "loss": 0.0069,
      "step": 8725
    },
    {
      "epoch": 4.072812135355893,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012877298219197825,
      "loss": 0.0154,
      "step": 8726
    },
    {
      "epoch": 4.073278879813302,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00012875892891733113,
      "loss": 0.0066,
      "step": 8727
    },
    {
      "epoch": 4.073745624270712,
      "grad_norm": 0.453125,
      "learning_rate": 0.00012874487502347238,
      "loss": 0.0112,
      "step": 8728
    },
    {
      "epoch": 4.074212368728121,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001287308205107046,
      "loss": 0.0095,
      "step": 8729
    },
    {
      "epoch": 4.074679113185531,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00012871676537933038,
      "loss": 0.0082,
      "step": 8730
    },
    {
      "epoch": 4.075145857642941,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00012870270962965228,
      "loss": 0.0152,
      "step": 8731
    },
    {
      "epoch": 4.0756126021003505,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00012868865326197307,
      "loss": 0.0048,
      "step": 8732
    },
    {
      "epoch": 4.076079346557759,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012867459627659527,
      "loss": 0.0059,
      "step": 8733
    },
    {
      "epoch": 4.076546091015169,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012866053867382164,
      "loss": 0.0093,
      "step": 8734
    },
    {
      "epoch": 4.077012835472579,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00012864648045395475,
      "loss": 0.0126,
      "step": 8735
    },
    {
      "epoch": 4.077479579929989,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00012863242161729741,
      "loss": 0.013,
      "step": 8736
    },
    {
      "epoch": 4.077946324387398,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00012861836216415226,
      "loss": 0.007,
      "step": 8737
    },
    {
      "epoch": 4.0784130688448075,
      "grad_norm": 0.546875,
      "learning_rate": 0.000128604302094822,
      "loss": 0.0119,
      "step": 8738
    },
    {
      "epoch": 4.078879813302217,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00012859024140960938,
      "loss": 0.0058,
      "step": 8739
    },
    {
      "epoch": 4.079346557759626,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012857618010881717,
      "loss": 0.0109,
      "step": 8740
    },
    {
      "epoch": 4.079813302217036,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00012856211819274805,
      "loss": 0.0089,
      "step": 8741
    },
    {
      "epoch": 4.080280046674446,
      "grad_norm": 0.53125,
      "learning_rate": 0.00012854805566170485,
      "loss": 0.0108,
      "step": 8742
    },
    {
      "epoch": 4.080746791131856,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00012853399251599033,
      "loss": 0.0132,
      "step": 8743
    },
    {
      "epoch": 4.0812135355892645,
      "grad_norm": 0.53125,
      "learning_rate": 0.00012851992875590734,
      "loss": 0.0055,
      "step": 8744
    },
    {
      "epoch": 4.081680280046674,
      "grad_norm": 0.6484375,
      "learning_rate": 0.0001285058643817586,
      "loss": 0.0143,
      "step": 8745
    },
    {
      "epoch": 4.082147024504084,
      "grad_norm": 0.376953125,
      "learning_rate": 0.000128491799393847,
      "loss": 0.0101,
      "step": 8746
    },
    {
      "epoch": 4.082613768961494,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012847773379247532,
      "loss": 0.0112,
      "step": 8747
    },
    {
      "epoch": 4.083080513418903,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00012846366757794642,
      "loss": 0.0049,
      "step": 8748
    },
    {
      "epoch": 4.083547257876313,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00012844960075056322,
      "loss": 0.0041,
      "step": 8749
    },
    {
      "epoch": 4.084014002333722,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012843553331062855,
      "loss": 0.014,
      "step": 8750
    },
    {
      "epoch": 4.084480746791132,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00012842146525844522,
      "loss": 0.0063,
      "step": 8751
    },
    {
      "epoch": 4.084947491248541,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012840739659431629,
      "loss": 0.0066,
      "step": 8752
    },
    {
      "epoch": 4.085414235705951,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00012839332731854456,
      "loss": 0.0098,
      "step": 8753
    },
    {
      "epoch": 4.085880980163361,
      "grad_norm": 0.48046875,
      "learning_rate": 0.000128379257431433,
      "loss": 0.0059,
      "step": 8754
    },
    {
      "epoch": 4.0863477246207704,
      "grad_norm": 0.205078125,
      "learning_rate": 0.00012836518693328455,
      "loss": 0.003,
      "step": 8755
    },
    {
      "epoch": 4.086814469078179,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012835111582440214,
      "loss": 0.0112,
      "step": 8756
    },
    {
      "epoch": 4.087281213535589,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012833704410508874,
      "loss": 0.0055,
      "step": 8757
    },
    {
      "epoch": 4.087747957992999,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012832297177564733,
      "loss": 0.0095,
      "step": 8758
    },
    {
      "epoch": 4.088214702450409,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00012830889883638094,
      "loss": 0.0066,
      "step": 8759
    },
    {
      "epoch": 4.088681446907818,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012829482528759257,
      "loss": 0.0158,
      "step": 8760
    },
    {
      "epoch": 4.0891481913652274,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00012828075112958516,
      "loss": 0.0059,
      "step": 8761
    },
    {
      "epoch": 4.089614935822637,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00012826667636266185,
      "loss": 0.0066,
      "step": 8762
    },
    {
      "epoch": 4.090081680280047,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001282526009871256,
      "loss": 0.0084,
      "step": 8763
    },
    {
      "epoch": 4.090548424737456,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012823852500327954,
      "loss": 0.0101,
      "step": 8764
    },
    {
      "epoch": 4.091015169194866,
      "grad_norm": 0.30859375,
      "learning_rate": 0.0001282244484114267,
      "loss": 0.0077,
      "step": 8765
    },
    {
      "epoch": 4.0914819136522755,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012821037121187013,
      "loss": 0.0133,
      "step": 8766
    },
    {
      "epoch": 4.091948658109685,
      "grad_norm": 0.466796875,
      "learning_rate": 0.000128196293404913,
      "loss": 0.0092,
      "step": 8767
    },
    {
      "epoch": 4.092415402567094,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001281822149908584,
      "loss": 0.0165,
      "step": 8768
    },
    {
      "epoch": 4.092882147024504,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012816813597000943,
      "loss": 0.0082,
      "step": 8769
    },
    {
      "epoch": 4.093348891481914,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012815405634266922,
      "loss": 0.008,
      "step": 8770
    },
    {
      "epoch": 4.093815635939324,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00012813997610914096,
      "loss": 0.009,
      "step": 8771
    },
    {
      "epoch": 4.0942823803967325,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00012812589526972777,
      "loss": 0.0078,
      "step": 8772
    },
    {
      "epoch": 4.094749124854142,
      "grad_norm": 0.279296875,
      "learning_rate": 0.0001281118138247329,
      "loss": 0.0033,
      "step": 8773
    },
    {
      "epoch": 4.095215869311552,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012809773177445942,
      "loss": 0.0059,
      "step": 8774
    },
    {
      "epoch": 4.095682613768962,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00012808364911921064,
      "loss": 0.0086,
      "step": 8775
    },
    {
      "epoch": 4.096149358226371,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00012806956585928976,
      "loss": 0.0057,
      "step": 8776
    },
    {
      "epoch": 4.096616102683781,
      "grad_norm": 0.34375,
      "learning_rate": 0.00012805548199499992,
      "loss": 0.0055,
      "step": 8777
    },
    {
      "epoch": 4.09708284714119,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00012804139752664446,
      "loss": 0.0034,
      "step": 8778
    },
    {
      "epoch": 4.0975495915986,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001280273124545266,
      "loss": 0.0075,
      "step": 8779
    },
    {
      "epoch": 4.098016336056009,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001280132267789496,
      "loss": 0.0125,
      "step": 8780
    },
    {
      "epoch": 4.098483080513419,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012799914050021672,
      "loss": 0.0105,
      "step": 8781
    },
    {
      "epoch": 4.098949824970829,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00012798505361863135,
      "loss": 0.008,
      "step": 8782
    },
    {
      "epoch": 4.099416569428238,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001279709661344967,
      "loss": 0.0098,
      "step": 8783
    },
    {
      "epoch": 4.099883313885647,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00012795687804811605,
      "loss": 0.0058,
      "step": 8784
    },
    {
      "epoch": 4.100350058343057,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00012794278935979284,
      "loss": 0.0071,
      "step": 8785
    },
    {
      "epoch": 4.100816802800467,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001279287000698304,
      "loss": 0.0098,
      "step": 8786
    },
    {
      "epoch": 4.101283547257876,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00012791461017853202,
      "loss": 0.0061,
      "step": 8787
    },
    {
      "epoch": 4.101750291715286,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00012790051968620113,
      "loss": 0.012,
      "step": 8788
    },
    {
      "epoch": 4.1022170361726955,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001278864285931411,
      "loss": 0.0086,
      "step": 8789
    },
    {
      "epoch": 4.102683780630105,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001278723368996553,
      "loss": 0.0106,
      "step": 8790
    },
    {
      "epoch": 4.103150525087514,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00012785824460604718,
      "loss": 0.0062,
      "step": 8791
    },
    {
      "epoch": 4.103617269544924,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012784415171262017,
      "loss": 0.0104,
      "step": 8792
    },
    {
      "epoch": 4.104084014002334,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00012783005821967764,
      "loss": 0.0087,
      "step": 8793
    },
    {
      "epoch": 4.104550758459744,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00012781596412752308,
      "loss": 0.0072,
      "step": 8794
    },
    {
      "epoch": 4.1050175029171525,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012780186943645995,
      "loss": 0.0109,
      "step": 8795
    },
    {
      "epoch": 4.105484247374562,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00012778777414679173,
      "loss": 0.0057,
      "step": 8796
    },
    {
      "epoch": 4.105950991831972,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001277736782588219,
      "loss": 0.0094,
      "step": 8797
    },
    {
      "epoch": 4.106417736289382,
      "grad_norm": 0.453125,
      "learning_rate": 0.00012775958177285398,
      "loss": 0.0047,
      "step": 8798
    },
    {
      "epoch": 4.106884480746791,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012774548468919144,
      "loss": 0.0061,
      "step": 8799
    },
    {
      "epoch": 4.107351225204201,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00012773138700813784,
      "loss": 0.0081,
      "step": 8800
    },
    {
      "epoch": 4.10781796966161,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0001277172887299967,
      "loss": 0.0062,
      "step": 8801
    },
    {
      "epoch": 4.10828471411902,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001277031898550716,
      "loss": 0.0039,
      "step": 8802
    },
    {
      "epoch": 4.108751458576429,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012768909038366608,
      "loss": 0.0126,
      "step": 8803
    },
    {
      "epoch": 4.109218203033839,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00012767499031608373,
      "loss": 0.006,
      "step": 8804
    },
    {
      "epoch": 4.109684947491249,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00012766088965262813,
      "loss": 0.0071,
      "step": 8805
    },
    {
      "epoch": 4.1101516919486585,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012764678839360285,
      "loss": 0.012,
      "step": 8806
    },
    {
      "epoch": 4.110618436406067,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00012763268653931157,
      "loss": 0.0174,
      "step": 8807
    },
    {
      "epoch": 4.111085180863477,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001276185840900579,
      "loss": 0.0044,
      "step": 8808
    },
    {
      "epoch": 4.111551925320887,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00012760448104614546,
      "loss": 0.0053,
      "step": 8809
    },
    {
      "epoch": 4.112018669778297,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001275903774078779,
      "loss": 0.0056,
      "step": 8810
    },
    {
      "epoch": 4.112485414235706,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00012757627317555893,
      "loss": 0.0129,
      "step": 8811
    },
    {
      "epoch": 4.1129521586931155,
      "grad_norm": 0.4765625,
      "learning_rate": 0.0001275621683494922,
      "loss": 0.0103,
      "step": 8812
    },
    {
      "epoch": 4.113418903150525,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00012754806292998135,
      "loss": 0.005,
      "step": 8813
    },
    {
      "epoch": 4.113885647607935,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00012753395691733022,
      "loss": 0.0142,
      "step": 8814
    },
    {
      "epoch": 4.114352392065344,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001275198503118424,
      "loss": 0.0044,
      "step": 8815
    },
    {
      "epoch": 4.114819136522754,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00012750574311382168,
      "loss": 0.0068,
      "step": 8816
    },
    {
      "epoch": 4.115285880980164,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00012749163532357183,
      "loss": 0.007,
      "step": 8817
    },
    {
      "epoch": 4.115752625437573,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012747752694139652,
      "loss": 0.0161,
      "step": 8818
    },
    {
      "epoch": 4.116219369894982,
      "grad_norm": 0.73828125,
      "learning_rate": 0.00012746341796759957,
      "loss": 0.0089,
      "step": 8819
    },
    {
      "epoch": 4.116686114352392,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00012744930840248478,
      "loss": 0.0046,
      "step": 8820
    },
    {
      "epoch": 4.117152858809802,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.00012743519824635592,
      "loss": 0.0059,
      "step": 8821
    },
    {
      "epoch": 4.117619603267211,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00012742108749951679,
      "loss": 0.005,
      "step": 8822
    },
    {
      "epoch": 4.118086347724621,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0001274069761622712,
      "loss": 0.0077,
      "step": 8823
    },
    {
      "epoch": 4.11855309218203,
      "grad_norm": 0.578125,
      "learning_rate": 0.00012739286423492304,
      "loss": 0.0138,
      "step": 8824
    },
    {
      "epoch": 4.11901983663944,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00012737875171777615,
      "loss": 0.0063,
      "step": 8825
    },
    {
      "epoch": 4.119486581096849,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012736463861113428,
      "loss": 0.0126,
      "step": 8826
    },
    {
      "epoch": 4.119953325554259,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00012735052491530144,
      "loss": 0.0043,
      "step": 8827
    },
    {
      "epoch": 4.120420070011669,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00012733641063058143,
      "loss": 0.0066,
      "step": 8828
    },
    {
      "epoch": 4.1208868144690785,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00012732229575727818,
      "loss": 0.0092,
      "step": 8829
    },
    {
      "epoch": 4.121353558926487,
      "grad_norm": 0.421875,
      "learning_rate": 0.00012730818029569558,
      "loss": 0.0074,
      "step": 8830
    },
    {
      "epoch": 4.121820303383897,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00012729406424613754,
      "loss": 0.0078,
      "step": 8831
    },
    {
      "epoch": 4.122287047841307,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012727994760890806,
      "loss": 0.0082,
      "step": 8832
    },
    {
      "epoch": 4.122753792298717,
      "grad_norm": 0.27734375,
      "learning_rate": 0.000127265830384311,
      "loss": 0.0042,
      "step": 8833
    },
    {
      "epoch": 4.123220536756126,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00012725171257265037,
      "loss": 0.0057,
      "step": 8834
    },
    {
      "epoch": 4.1236872812135354,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00012723759417423012,
      "loss": 0.0064,
      "step": 8835
    },
    {
      "epoch": 4.124154025670945,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00012722347518935427,
      "loss": 0.0058,
      "step": 8836
    },
    {
      "epoch": 4.124620770128355,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00012720935561832679,
      "loss": 0.0046,
      "step": 8837
    },
    {
      "epoch": 4.125087514585764,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001271952354614517,
      "loss": 0.0093,
      "step": 8838
    },
    {
      "epoch": 4.125554259043174,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012718111471903297,
      "loss": 0.007,
      "step": 8839
    },
    {
      "epoch": 4.1260210035005835,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012716699339137476,
      "loss": 0.0168,
      "step": 8840
    },
    {
      "epoch": 4.126487747957993,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012715287147878102,
      "loss": 0.0103,
      "step": 8841
    },
    {
      "epoch": 4.126954492415402,
      "grad_norm": 0.427734375,
      "learning_rate": 0.0001271387489815558,
      "loss": 0.0087,
      "step": 8842
    },
    {
      "epoch": 4.127421236872812,
      "grad_norm": 0.79296875,
      "learning_rate": 0.0001271246259000032,
      "loss": 0.0152,
      "step": 8843
    },
    {
      "epoch": 4.127887981330222,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00012711050223442733,
      "loss": 0.0143,
      "step": 8844
    },
    {
      "epoch": 4.128354725787632,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00012709637798513225,
      "loss": 0.0098,
      "step": 8845
    },
    {
      "epoch": 4.1288214702450405,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001270822531524221,
      "loss": 0.0124,
      "step": 8846
    },
    {
      "epoch": 4.12928821470245,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012706812773660098,
      "loss": 0.0063,
      "step": 8847
    },
    {
      "epoch": 4.12975495915986,
      "grad_norm": 0.359375,
      "learning_rate": 0.00012705400173797305,
      "loss": 0.0045,
      "step": 8848
    },
    {
      "epoch": 4.13022170361727,
      "grad_norm": 0.578125,
      "learning_rate": 0.00012703987515684244,
      "loss": 0.0086,
      "step": 8849
    },
    {
      "epoch": 4.130688448074679,
      "grad_norm": 1.171875,
      "learning_rate": 0.00012702574799351333,
      "loss": 0.0146,
      "step": 8850
    },
    {
      "epoch": 4.131155192532089,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012701162024828988,
      "loss": 0.0111,
      "step": 8851
    },
    {
      "epoch": 4.131621936989498,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012699749192147624,
      "loss": 0.0105,
      "step": 8852
    },
    {
      "epoch": 4.132088681446908,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.00012698336301337668,
      "loss": 0.0031,
      "step": 8853
    },
    {
      "epoch": 4.132555425904317,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012696923352429538,
      "loss": 0.0098,
      "step": 8854
    },
    {
      "epoch": 4.133022170361727,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00012695510345453652,
      "loss": 0.0169,
      "step": 8855
    },
    {
      "epoch": 4.133488914819137,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001269409728044044,
      "loss": 0.0116,
      "step": 8856
    },
    {
      "epoch": 4.1339556592765465,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00012692684157420325,
      "loss": 0.006,
      "step": 8857
    },
    {
      "epoch": 4.134422403733955,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001269127097642373,
      "loss": 0.0073,
      "step": 8858
    },
    {
      "epoch": 4.134889148191365,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00012689857737481088,
      "loss": 0.012,
      "step": 8859
    },
    {
      "epoch": 4.135355892648775,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012688444440622825,
      "loss": 0.0092,
      "step": 8860
    },
    {
      "epoch": 4.135822637106184,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012687031085879369,
      "loss": 0.0106,
      "step": 8861
    },
    {
      "epoch": 4.136289381563594,
      "grad_norm": 0.3359375,
      "learning_rate": 0.0001268561767328115,
      "loss": 0.0058,
      "step": 8862
    },
    {
      "epoch": 4.1367561260210035,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00012684204202858605,
      "loss": 0.0052,
      "step": 8863
    },
    {
      "epoch": 4.137222870478413,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00012682790674642166,
      "loss": 0.0058,
      "step": 8864
    },
    {
      "epoch": 4.137689614935822,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00012681377088662266,
      "loss": 0.0092,
      "step": 8865
    },
    {
      "epoch": 4.138156359393232,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012679963444949343,
      "loss": 0.012,
      "step": 8866
    },
    {
      "epoch": 4.138623103850642,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012678549743533834,
      "loss": 0.0079,
      "step": 8867
    },
    {
      "epoch": 4.139089848308052,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00012677135984446173,
      "loss": 0.0102,
      "step": 8868
    },
    {
      "epoch": 4.1395565927654605,
      "grad_norm": 0.71875,
      "learning_rate": 0.00012675722167716807,
      "loss": 0.0084,
      "step": 8869
    },
    {
      "epoch": 4.14002333722287,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00012674308293376176,
      "loss": 0.0035,
      "step": 8870
    },
    {
      "epoch": 4.14049008168028,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00012672894361454717,
      "loss": 0.0064,
      "step": 8871
    },
    {
      "epoch": 4.14095682613769,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00012671480371982877,
      "loss": 0.0071,
      "step": 8872
    },
    {
      "epoch": 4.141423570595099,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00012670066324991104,
      "loss": 0.0077,
      "step": 8873
    },
    {
      "epoch": 4.141890315052509,
      "grad_norm": 0.53125,
      "learning_rate": 0.00012668652220509838,
      "loss": 0.0073,
      "step": 8874
    },
    {
      "epoch": 4.142357059509918,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001266723805856953,
      "loss": 0.0107,
      "step": 8875
    },
    {
      "epoch": 4.142823803967328,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00012665823839200626,
      "loss": 0.0092,
      "step": 8876
    },
    {
      "epoch": 4.143290548424737,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00012664409562433576,
      "loss": 0.0146,
      "step": 8877
    },
    {
      "epoch": 4.143757292882147,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00012662995228298831,
      "loss": 0.0052,
      "step": 8878
    },
    {
      "epoch": 4.144224037339557,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012661580836826846,
      "loss": 0.0045,
      "step": 8879
    },
    {
      "epoch": 4.1446907817969665,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00012660166388048073,
      "loss": 0.0071,
      "step": 8880
    },
    {
      "epoch": 4.145157526254375,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00012658751881992967,
      "loss": 0.0094,
      "step": 8881
    },
    {
      "epoch": 4.145624270711785,
      "grad_norm": 0.5,
      "learning_rate": 0.0001265733731869198,
      "loss": 0.0123,
      "step": 8882
    },
    {
      "epoch": 4.146091015169195,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00012655922698175576,
      "loss": 0.0063,
      "step": 8883
    },
    {
      "epoch": 4.146557759626605,
      "grad_norm": 0.92578125,
      "learning_rate": 0.00012654508020474206,
      "loss": 0.017,
      "step": 8884
    },
    {
      "epoch": 4.147024504084014,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012653093285618333,
      "loss": 0.0096,
      "step": 8885
    },
    {
      "epoch": 4.1474912485414235,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00012651678493638422,
      "loss": 0.0039,
      "step": 8886
    },
    {
      "epoch": 4.147957992998833,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001265026364456493,
      "loss": 0.0059,
      "step": 8887
    },
    {
      "epoch": 4.148424737456243,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00012648848738428317,
      "loss": 0.0067,
      "step": 8888
    },
    {
      "epoch": 4.148891481913652,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012647433775259055,
      "loss": 0.0097,
      "step": 8889
    },
    {
      "epoch": 4.149358226371062,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012646018755087607,
      "loss": 0.0085,
      "step": 8890
    },
    {
      "epoch": 4.149824970828472,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00012644603677944437,
      "loss": 0.0085,
      "step": 8891
    },
    {
      "epoch": 4.150291715285881,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001264318854386002,
      "loss": 0.0061,
      "step": 8892
    },
    {
      "epoch": 4.15075845974329,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00012641773352864817,
      "loss": 0.0066,
      "step": 8893
    },
    {
      "epoch": 4.1512252042007,
      "grad_norm": 0.4375,
      "learning_rate": 0.00012640358104989302,
      "loss": 0.0075,
      "step": 8894
    },
    {
      "epoch": 4.15169194865811,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001263894280026395,
      "loss": 0.0086,
      "step": 8895
    },
    {
      "epoch": 4.15215869311552,
      "grad_norm": 0.296875,
      "learning_rate": 0.0001263752743871923,
      "loss": 0.0052,
      "step": 8896
    },
    {
      "epoch": 4.152625437572929,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00012636112020385616,
      "loss": 0.0092,
      "step": 8897
    },
    {
      "epoch": 4.153092182030338,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00012634696545293589,
      "loss": 0.0113,
      "step": 8898
    },
    {
      "epoch": 4.153558926487748,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001263328101347362,
      "loss": 0.0091,
      "step": 8899
    },
    {
      "epoch": 4.154025670945158,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001263186542495619,
      "loss": 0.0081,
      "step": 8900
    },
    {
      "epoch": 4.154492415402567,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00012630449779771775,
      "loss": 0.0089,
      "step": 8901
    },
    {
      "epoch": 4.154959159859977,
      "grad_norm": 0.439453125,
      "learning_rate": 0.0001262903407795086,
      "loss": 0.0085,
      "step": 8902
    },
    {
      "epoch": 4.1554259043173865,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00012627618319523926,
      "loss": 0.0118,
      "step": 8903
    },
    {
      "epoch": 4.155892648774795,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012626202504521452,
      "loss": 0.0062,
      "step": 8904
    },
    {
      "epoch": 4.156359393232205,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012624786632973924,
      "loss": 0.0107,
      "step": 8905
    },
    {
      "epoch": 4.156826137689615,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012623370704911826,
      "loss": 0.0203,
      "step": 8906
    },
    {
      "epoch": 4.157292882147025,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00012621954720365652,
      "loss": 0.0103,
      "step": 8907
    },
    {
      "epoch": 4.157759626604434,
      "grad_norm": 0.265625,
      "learning_rate": 0.0001262053867936588,
      "loss": 0.0109,
      "step": 8908
    },
    {
      "epoch": 4.1582263710618435,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012619122581943002,
      "loss": 0.0055,
      "step": 8909
    },
    {
      "epoch": 4.158693115519253,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001261770642812751,
      "loss": 0.006,
      "step": 8910
    },
    {
      "epoch": 4.159159859976663,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00012616290217949894,
      "loss": 0.0103,
      "step": 8911
    },
    {
      "epoch": 4.159626604434072,
      "grad_norm": 0.5,
      "learning_rate": 0.00012614873951440647,
      "loss": 0.0067,
      "step": 8912
    },
    {
      "epoch": 4.160093348891482,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012613457628630264,
      "loss": 0.0072,
      "step": 8913
    },
    {
      "epoch": 4.1605600933488915,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012612041249549235,
      "loss": 0.0096,
      "step": 8914
    },
    {
      "epoch": 4.161026837806301,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001261062481422806,
      "loss": 0.0073,
      "step": 8915
    },
    {
      "epoch": 4.16149358226371,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0001260920832269724,
      "loss": 0.0051,
      "step": 8916
    },
    {
      "epoch": 4.16196032672112,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012607791774987264,
      "loss": 0.0093,
      "step": 8917
    },
    {
      "epoch": 4.16242707117853,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012606375171128641,
      "loss": 0.0071,
      "step": 8918
    },
    {
      "epoch": 4.16289381563594,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001260495851115187,
      "loss": 0.0089,
      "step": 8919
    },
    {
      "epoch": 4.1633605600933485,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001260354179508745,
      "loss": 0.0106,
      "step": 8920
    },
    {
      "epoch": 4.163827304550758,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00012602125022965888,
      "loss": 0.009,
      "step": 8921
    },
    {
      "epoch": 4.164294049008168,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012600708194817687,
      "loss": 0.0063,
      "step": 8922
    },
    {
      "epoch": 4.164760793465578,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012599291310673354,
      "loss": 0.011,
      "step": 8923
    },
    {
      "epoch": 4.165227537922987,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012597874370563392,
      "loss": 0.0126,
      "step": 8924
    },
    {
      "epoch": 4.165694282380397,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00012596457374518315,
      "loss": 0.0043,
      "step": 8925
    },
    {
      "epoch": 4.166161026837806,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001259504032256863,
      "loss": 0.0093,
      "step": 8926
    },
    {
      "epoch": 4.166627771295216,
      "grad_norm": 0.625,
      "learning_rate": 0.00012593623214744842,
      "loss": 0.0123,
      "step": 8927
    },
    {
      "epoch": 4.167094515752625,
      "grad_norm": 0.546875,
      "learning_rate": 0.00012592206051077474,
      "loss": 0.0077,
      "step": 8928
    },
    {
      "epoch": 4.167561260210035,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012590788831597037,
      "loss": 0.0078,
      "step": 8929
    },
    {
      "epoch": 4.168028004667445,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00012589371556334035,
      "loss": 0.0175,
      "step": 8930
    },
    {
      "epoch": 4.1684947491248545,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00012587954225318993,
      "loss": 0.0075,
      "step": 8931
    },
    {
      "epoch": 4.168961493582263,
      "grad_norm": 0.578125,
      "learning_rate": 0.00012586536838582425,
      "loss": 0.009,
      "step": 8932
    },
    {
      "epoch": 4.169428238039673,
      "grad_norm": 0.51953125,
      "learning_rate": 0.0001258511939615485,
      "loss": 0.0119,
      "step": 8933
    },
    {
      "epoch": 4.169894982497083,
      "grad_norm": 0.72265625,
      "learning_rate": 0.00012583701898066785,
      "loss": 0.0116,
      "step": 8934
    },
    {
      "epoch": 4.170361726954493,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00012582284344348754,
      "loss": 0.0106,
      "step": 8935
    },
    {
      "epoch": 4.170828471411902,
      "grad_norm": 0.267578125,
      "learning_rate": 0.00012580866735031272,
      "loss": 0.0036,
      "step": 8936
    },
    {
      "epoch": 4.1712952158693115,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001257944907014487,
      "loss": 0.0097,
      "step": 8937
    },
    {
      "epoch": 4.171761960326721,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00012578031349720064,
      "loss": 0.0101,
      "step": 8938
    },
    {
      "epoch": 4.172228704784131,
      "grad_norm": 0.46875,
      "learning_rate": 0.00012576613573787382,
      "loss": 0.0087,
      "step": 8939
    },
    {
      "epoch": 4.17269544924154,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012575195742377355,
      "loss": 0.0145,
      "step": 8940
    },
    {
      "epoch": 4.17316219369895,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012573777855520504,
      "loss": 0.0088,
      "step": 8941
    },
    {
      "epoch": 4.17362893815636,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001257235991324736,
      "loss": 0.0085,
      "step": 8942
    },
    {
      "epoch": 4.174095682613769,
      "grad_norm": 0.421875,
      "learning_rate": 0.00012570941915588453,
      "loss": 0.0157,
      "step": 8943
    },
    {
      "epoch": 4.174562427071178,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00012569523862574316,
      "loss": 0.0107,
      "step": 8944
    },
    {
      "epoch": 4.175029171528588,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00012568105754235477,
      "loss": 0.0095,
      "step": 8945
    },
    {
      "epoch": 4.175495915985998,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00012566687590602474,
      "loss": 0.0056,
      "step": 8946
    },
    {
      "epoch": 4.175962660443407,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012565269371705835,
      "loss": 0.008,
      "step": 8947
    },
    {
      "epoch": 4.176429404900817,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00012563851097576102,
      "loss": 0.0091,
      "step": 8948
    },
    {
      "epoch": 4.176896149358226,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00012562432768243814,
      "loss": 0.0078,
      "step": 8949
    },
    {
      "epoch": 4.177362893815636,
      "grad_norm": 0.400390625,
      "learning_rate": 0.000125610143837395,
      "loss": 0.0082,
      "step": 8950
    },
    {
      "epoch": 4.177829638273045,
      "grad_norm": 0.6875,
      "learning_rate": 0.0001255959594409371,
      "loss": 0.0127,
      "step": 8951
    },
    {
      "epoch": 4.178296382730455,
      "grad_norm": 0.484375,
      "learning_rate": 0.00012558177449336976,
      "loss": 0.0134,
      "step": 8952
    },
    {
      "epoch": 4.178763127187865,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001255675889949984,
      "loss": 0.0053,
      "step": 8953
    },
    {
      "epoch": 4.1792298716452745,
      "grad_norm": 0.40625,
      "learning_rate": 0.00012555340294612854,
      "loss": 0.0065,
      "step": 8954
    },
    {
      "epoch": 4.179696616102683,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00012553921634706557,
      "loss": 0.0052,
      "step": 8955
    },
    {
      "epoch": 4.180163360560093,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001255250291981149,
      "loss": 0.0086,
      "step": 8956
    },
    {
      "epoch": 4.180630105017503,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012551084149958202,
      "loss": 0.0096,
      "step": 8957
    },
    {
      "epoch": 4.181096849474913,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012549665325177248,
      "loss": 0.0077,
      "step": 8958
    },
    {
      "epoch": 4.181563593932322,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012548246445499167,
      "loss": 0.0071,
      "step": 8959
    },
    {
      "epoch": 4.1820303383897315,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00012546827510954512,
      "loss": 0.0048,
      "step": 8960
    },
    {
      "epoch": 4.182497082847141,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00012545408521573832,
      "loss": 0.0044,
      "step": 8961
    },
    {
      "epoch": 4.182963827304551,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00012543989477387688,
      "loss": 0.0055,
      "step": 8962
    },
    {
      "epoch": 4.18343057176196,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00012542570378426624,
      "loss": 0.0054,
      "step": 8963
    },
    {
      "epoch": 4.18389731621937,
      "grad_norm": 0.75390625,
      "learning_rate": 0.000125411512247212,
      "loss": 0.0101,
      "step": 8964
    },
    {
      "epoch": 4.18436406067678,
      "grad_norm": 0.287109375,
      "learning_rate": 0.00012539732016301973,
      "loss": 0.0047,
      "step": 8965
    },
    {
      "epoch": 4.184830805134189,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00012538312753199494,
      "loss": 0.0044,
      "step": 8966
    },
    {
      "epoch": 4.185297549591598,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00012536893435444326,
      "loss": 0.0053,
      "step": 8967
    },
    {
      "epoch": 4.185764294049008,
      "grad_norm": 0.625,
      "learning_rate": 0.0001253547406306703,
      "loss": 0.0118,
      "step": 8968
    },
    {
      "epoch": 4.186231038506418,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001253405463609816,
      "loss": 0.014,
      "step": 8969
    },
    {
      "epoch": 4.186697782963828,
      "grad_norm": 0.640625,
      "learning_rate": 0.00012532635154568288,
      "loss": 0.0125,
      "step": 8970
    },
    {
      "epoch": 4.187164527421237,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001253121561850797,
      "loss": 0.0132,
      "step": 8971
    },
    {
      "epoch": 4.187631271878646,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00012529796027947767,
      "loss": 0.0085,
      "step": 8972
    },
    {
      "epoch": 4.188098016336056,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00012528376382918248,
      "loss": 0.0036,
      "step": 8973
    },
    {
      "epoch": 4.188564760793466,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00012526956683449984,
      "loss": 0.0081,
      "step": 8974
    },
    {
      "epoch": 4.189031505250875,
      "grad_norm": 0.390625,
      "learning_rate": 0.00012525536929573538,
      "loss": 0.0066,
      "step": 8975
    },
    {
      "epoch": 4.189498249708285,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00012524117121319477,
      "loss": 0.0034,
      "step": 8976
    },
    {
      "epoch": 4.1899649941656945,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001252269725871838,
      "loss": 0.0054,
      "step": 8977
    },
    {
      "epoch": 4.190431738623104,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012521277341800808,
      "loss": 0.0082,
      "step": 8978
    },
    {
      "epoch": 4.190898483080513,
      "grad_norm": 0.484375,
      "learning_rate": 0.00012519857370597336,
      "loss": 0.0096,
      "step": 8979
    },
    {
      "epoch": 4.191365227537923,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00012518437345138543,
      "loss": 0.0052,
      "step": 8980
    },
    {
      "epoch": 4.191831971995333,
      "grad_norm": 0.53125,
      "learning_rate": 0.00012517017265454997,
      "loss": 0.0062,
      "step": 8981
    },
    {
      "epoch": 4.1922987164527425,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001251559713157728,
      "loss": 0.0099,
      "step": 8982
    },
    {
      "epoch": 4.1927654609101515,
      "grad_norm": 0.56640625,
      "learning_rate": 0.0001251417694353596,
      "loss": 0.0125,
      "step": 8983
    },
    {
      "epoch": 4.193232205367561,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00012512756701361626,
      "loss": 0.0078,
      "step": 8984
    },
    {
      "epoch": 4.193698949824971,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001251133640508485,
      "loss": 0.0124,
      "step": 8985
    },
    {
      "epoch": 4.194165694282381,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00012509916054736215,
      "loss": 0.0112,
      "step": 8986
    },
    {
      "epoch": 4.19463243873979,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012508495650346302,
      "loss": 0.009,
      "step": 8987
    },
    {
      "epoch": 4.1950991831971995,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00012507075191945695,
      "loss": 0.0093,
      "step": 8988
    },
    {
      "epoch": 4.195565927654609,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00012505654679564976,
      "loss": 0.0167,
      "step": 8989
    },
    {
      "epoch": 4.196032672112018,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012504234113234735,
      "loss": 0.015,
      "step": 8990
    },
    {
      "epoch": 4.196499416569428,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00012502813492985552,
      "loss": 0.0073,
      "step": 8991
    },
    {
      "epoch": 4.196966161026838,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00012501392818848017,
      "loss": 0.0085,
      "step": 8992
    },
    {
      "epoch": 4.197432905484248,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00012499972090852722,
      "loss": 0.0077,
      "step": 8993
    },
    {
      "epoch": 4.1978996499416565,
      "grad_norm": 0.359375,
      "learning_rate": 0.00012498551309030254,
      "loss": 0.0067,
      "step": 8994
    },
    {
      "epoch": 4.198366394399066,
      "grad_norm": 0.34765625,
      "learning_rate": 0.000124971304734112,
      "loss": 0.0043,
      "step": 8995
    },
    {
      "epoch": 4.198833138856476,
      "grad_norm": 0.5625,
      "learning_rate": 0.00012495709584026162,
      "loss": 0.0069,
      "step": 8996
    },
    {
      "epoch": 4.199299883313886,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012494288640905725,
      "loss": 0.0127,
      "step": 8997
    },
    {
      "epoch": 4.199766627771295,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012492867644080485,
      "loss": 0.0143,
      "step": 8998
    },
    {
      "epoch": 4.200233372228705,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001249144659358104,
      "loss": 0.0076,
      "step": 8999
    },
    {
      "epoch": 4.200700116686114,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012490025489437985,
      "loss": 0.0073,
      "step": 9000
    },
    {
      "epoch": 4.201166861143524,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001248860433168192,
      "loss": 0.009,
      "step": 9001
    },
    {
      "epoch": 4.201633605600933,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001248718312034344,
      "loss": 0.0112,
      "step": 9002
    },
    {
      "epoch": 4.202100350058343,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00012485761855453152,
      "loss": 0.0079,
      "step": 9003
    },
    {
      "epoch": 4.202567094515753,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001248434053704165,
      "loss": 0.0061,
      "step": 9004
    },
    {
      "epoch": 4.2030338389731625,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001248291916513954,
      "loss": 0.0081,
      "step": 9005
    },
    {
      "epoch": 4.203500583430571,
      "grad_norm": 0.75,
      "learning_rate": 0.00012481497739777423,
      "loss": 0.0092,
      "step": 9006
    },
    {
      "epoch": 4.203967327887981,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012480076260985913,
      "loss": 0.0136,
      "step": 9007
    },
    {
      "epoch": 4.204434072345391,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00012478654728795605,
      "loss": 0.0079,
      "step": 9008
    },
    {
      "epoch": 4.204900816802801,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00012477233143237112,
      "loss": 0.0067,
      "step": 9009
    },
    {
      "epoch": 4.20536756126021,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012475811504341039,
      "loss": 0.0089,
      "step": 9010
    },
    {
      "epoch": 4.2058343057176195,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00012474389812138002,
      "loss": 0.0048,
      "step": 9011
    },
    {
      "epoch": 4.206301050175029,
      "grad_norm": 0.275390625,
      "learning_rate": 0.000124729680666586,
      "loss": 0.004,
      "step": 9012
    },
    {
      "epoch": 4.206767794632439,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00012471546267933456,
      "loss": 0.0059,
      "step": 9013
    },
    {
      "epoch": 4.207234539089848,
      "grad_norm": 0.80078125,
      "learning_rate": 0.00012470124415993179,
      "loss": 0.0181,
      "step": 9014
    },
    {
      "epoch": 4.207701283547258,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001246870251086838,
      "loss": 0.0078,
      "step": 9015
    },
    {
      "epoch": 4.208168028004668,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001246728055258968,
      "loss": 0.0072,
      "step": 9016
    },
    {
      "epoch": 4.208634772462077,
      "grad_norm": 0.435546875,
      "learning_rate": 0.0001246585854118769,
      "loss": 0.0064,
      "step": 9017
    },
    {
      "epoch": 4.209101516919486,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001246443647669303,
      "loss": 0.0042,
      "step": 9018
    },
    {
      "epoch": 4.209568261376896,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00012463014359136317,
      "loss": 0.0096,
      "step": 9019
    },
    {
      "epoch": 4.210035005834306,
      "grad_norm": 0.375,
      "learning_rate": 0.00012461592188548176,
      "loss": 0.0056,
      "step": 9020
    },
    {
      "epoch": 4.210501750291716,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001246016996495922,
      "loss": 0.0053,
      "step": 9021
    },
    {
      "epoch": 4.210968494749125,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00012458747688400077,
      "loss": 0.0069,
      "step": 9022
    },
    {
      "epoch": 4.211435239206534,
      "grad_norm": 0.41796875,
      "learning_rate": 0.0001245732535890137,
      "loss": 0.0057,
      "step": 9023
    },
    {
      "epoch": 4.211901983663944,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00012455902976493718,
      "loss": 0.0045,
      "step": 9024
    },
    {
      "epoch": 4.212368728121354,
      "grad_norm": 0.34765625,
      "learning_rate": 0.0001245448054120775,
      "loss": 0.0049,
      "step": 9025
    },
    {
      "epoch": 4.212835472578763,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00012453058053074094,
      "loss": 0.0107,
      "step": 9026
    },
    {
      "epoch": 4.213302217036173,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001245163551212338,
      "loss": 0.0077,
      "step": 9027
    },
    {
      "epoch": 4.2137689614935825,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00012450212918386225,
      "loss": 0.0092,
      "step": 9028
    },
    {
      "epoch": 4.214235705950992,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012448790271893273,
      "loss": 0.0129,
      "step": 9029
    },
    {
      "epoch": 4.214702450408401,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00012447367572675152,
      "loss": 0.008,
      "step": 9030
    },
    {
      "epoch": 4.215169194865811,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00012445944820762487,
      "loss": 0.0068,
      "step": 9031
    },
    {
      "epoch": 4.215635939323221,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00012444522016185917,
      "loss": 0.0096,
      "step": 9032
    },
    {
      "epoch": 4.21610268378063,
      "grad_norm": 0.3203125,
      "learning_rate": 0.0001244309915897608,
      "loss": 0.0058,
      "step": 9033
    },
    {
      "epoch": 4.2165694282380395,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00012441676249163602,
      "loss": 0.0037,
      "step": 9034
    },
    {
      "epoch": 4.217036172695449,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00012440253286779125,
      "loss": 0.0065,
      "step": 9035
    },
    {
      "epoch": 4.217502917152859,
      "grad_norm": 0.28515625,
      "learning_rate": 0.0001243883027185329,
      "loss": 0.0042,
      "step": 9036
    },
    {
      "epoch": 4.217969661610268,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00012437407204416733,
      "loss": 0.0127,
      "step": 9037
    },
    {
      "epoch": 4.218436406067678,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00012435984084500093,
      "loss": 0.0056,
      "step": 9038
    },
    {
      "epoch": 4.218903150525088,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012434560912134013,
      "loss": 0.0186,
      "step": 9039
    },
    {
      "epoch": 4.219369894982497,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00012433137687349138,
      "loss": 0.0081,
      "step": 9040
    },
    {
      "epoch": 4.219836639439906,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.00012431714410176108,
      "loss": 0.0026,
      "step": 9041
    },
    {
      "epoch": 4.220303383897316,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001243029108064557,
      "loss": 0.0055,
      "step": 9042
    },
    {
      "epoch": 4.220770128354726,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00012428867698788165,
      "loss": 0.0083,
      "step": 9043
    },
    {
      "epoch": 4.221236872812136,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00012427444264634545,
      "loss": 0.0112,
      "step": 9044
    },
    {
      "epoch": 4.221703617269545,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00012426020778215358,
      "loss": 0.0121,
      "step": 9045
    },
    {
      "epoch": 4.222170361726954,
      "grad_norm": 0.6015625,
      "learning_rate": 0.0001242459723956125,
      "loss": 0.0108,
      "step": 9046
    },
    {
      "epoch": 4.222637106184364,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00012423173648702876,
      "loss": 0.0063,
      "step": 9047
    },
    {
      "epoch": 4.223103850641774,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001242175000567088,
      "loss": 0.0092,
      "step": 9048
    },
    {
      "epoch": 4.223570595099183,
      "grad_norm": 0.5,
      "learning_rate": 0.00012420326310495924,
      "loss": 0.0091,
      "step": 9049
    },
    {
      "epoch": 4.224037339556593,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00012418902563208657,
      "loss": 0.0095,
      "step": 9050
    },
    {
      "epoch": 4.2245040840140025,
      "grad_norm": 0.431640625,
      "learning_rate": 0.0001241747876383973,
      "loss": 0.006,
      "step": 9051
    },
    {
      "epoch": 4.224970828471412,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012416054912419805,
      "loss": 0.0122,
      "step": 9052
    },
    {
      "epoch": 4.225437572928821,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001241463100897954,
      "loss": 0.0065,
      "step": 9053
    },
    {
      "epoch": 4.225904317386231,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012413207053549584,
      "loss": 0.0073,
      "step": 9054
    },
    {
      "epoch": 4.226371061843641,
      "grad_norm": 0.7421875,
      "learning_rate": 0.0001241178304616061,
      "loss": 0.0185,
      "step": 9055
    },
    {
      "epoch": 4.2268378063010505,
      "grad_norm": 0.30078125,
      "learning_rate": 0.0001241035898684327,
      "loss": 0.0059,
      "step": 9056
    },
    {
      "epoch": 4.2273045507584595,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012408934875628223,
      "loss": 0.0071,
      "step": 9057
    },
    {
      "epoch": 4.227771295215869,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00012407510712546135,
      "loss": 0.0061,
      "step": 9058
    },
    {
      "epoch": 4.228238039673279,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012406086497627675,
      "loss": 0.0142,
      "step": 9059
    },
    {
      "epoch": 4.228704784130689,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00012404662230903498,
      "loss": 0.0055,
      "step": 9060
    },
    {
      "epoch": 4.229171528588098,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00012403237912404276,
      "loss": 0.0146,
      "step": 9061
    },
    {
      "epoch": 4.2296382730455075,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012401813542160675,
      "loss": 0.0133,
      "step": 9062
    },
    {
      "epoch": 4.230105017502917,
      "grad_norm": 0.7890625,
      "learning_rate": 0.0001240038912020337,
      "loss": 0.0175,
      "step": 9063
    },
    {
      "epoch": 4.230571761960327,
      "grad_norm": 0.3203125,
      "learning_rate": 0.00012398964646563013,
      "loss": 0.0042,
      "step": 9064
    },
    {
      "epoch": 4.231038506417736,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012397540121270294,
      "loss": 0.0049,
      "step": 9065
    },
    {
      "epoch": 4.231505250875146,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00012396115544355873,
      "loss": 0.0082,
      "step": 9066
    },
    {
      "epoch": 4.231971995332556,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00012394690915850425,
      "loss": 0.0118,
      "step": 9067
    },
    {
      "epoch": 4.2324387397899645,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012393266235784627,
      "loss": 0.0066,
      "step": 9068
    },
    {
      "epoch": 4.232905484247374,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001239184150418915,
      "loss": 0.0064,
      "step": 9069
    },
    {
      "epoch": 4.233372228704784,
      "grad_norm": 0.53125,
      "learning_rate": 0.00012390416721094673,
      "loss": 0.0082,
      "step": 9070
    },
    {
      "epoch": 4.233838973162194,
      "grad_norm": 0.375,
      "learning_rate": 0.00012388991886531868,
      "loss": 0.0063,
      "step": 9071
    },
    {
      "epoch": 4.234305717619603,
      "grad_norm": 0.298828125,
      "learning_rate": 0.0001238756700053142,
      "loss": 0.0055,
      "step": 9072
    },
    {
      "epoch": 4.234772462077013,
      "grad_norm": 0.484375,
      "learning_rate": 0.00012386142063124008,
      "loss": 0.0118,
      "step": 9073
    },
    {
      "epoch": 4.235239206534422,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012384717074340307,
      "loss": 0.0139,
      "step": 9074
    },
    {
      "epoch": 4.235705950991832,
      "grad_norm": 0.5,
      "learning_rate": 0.00012383292034211,
      "loss": 0.0054,
      "step": 9075
    },
    {
      "epoch": 4.236172695449241,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012381866942766774,
      "loss": 0.0068,
      "step": 9076
    },
    {
      "epoch": 4.236639439906651,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001238044180003831,
      "loss": 0.0081,
      "step": 9077
    },
    {
      "epoch": 4.237106184364061,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001237901660605629,
      "loss": 0.0085,
      "step": 9078
    },
    {
      "epoch": 4.2375729288214705,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012377591360851408,
      "loss": 0.0174,
      "step": 9079
    },
    {
      "epoch": 4.238039673278879,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012376166064454342,
      "loss": 0.0105,
      "step": 9080
    },
    {
      "epoch": 4.238506417736289,
      "grad_norm": 0.28125,
      "learning_rate": 0.00012374740716895784,
      "loss": 0.0041,
      "step": 9081
    },
    {
      "epoch": 4.238973162193699,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00012373315318206426,
      "loss": 0.0039,
      "step": 9082
    },
    {
      "epoch": 4.239439906651109,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00012371889868416952,
      "loss": 0.0046,
      "step": 9083
    },
    {
      "epoch": 4.239906651108518,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00012370464367558062,
      "loss": 0.0062,
      "step": 9084
    },
    {
      "epoch": 4.2403733955659275,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001236903881566044,
      "loss": 0.0122,
      "step": 9085
    },
    {
      "epoch": 4.240840140023337,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001236761321275479,
      "loss": 0.0138,
      "step": 9086
    },
    {
      "epoch": 4.241306884480747,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00012366187558871794,
      "loss": 0.006,
      "step": 9087
    },
    {
      "epoch": 4.241773628938156,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00012364761854042158,
      "loss": 0.0058,
      "step": 9088
    },
    {
      "epoch": 4.242240373395566,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00012363336098296574,
      "loss": 0.0066,
      "step": 9089
    },
    {
      "epoch": 4.242707117852976,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00012361910291665742,
      "loss": 0.0097,
      "step": 9090
    },
    {
      "epoch": 4.243173862310385,
      "grad_norm": 0.375,
      "learning_rate": 0.0001236048443418036,
      "loss": 0.0072,
      "step": 9091
    },
    {
      "epoch": 4.243640606767794,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00012359058525871131,
      "loss": 0.0063,
      "step": 9092
    },
    {
      "epoch": 4.244107351225204,
      "grad_norm": 0.5625,
      "learning_rate": 0.00012357632566768753,
      "loss": 0.0065,
      "step": 9093
    },
    {
      "epoch": 4.244574095682614,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00012356206556903931,
      "loss": 0.007,
      "step": 9094
    },
    {
      "epoch": 4.245040840140024,
      "grad_norm": 0.34375,
      "learning_rate": 0.00012354780496307368,
      "loss": 0.0064,
      "step": 9095
    },
    {
      "epoch": 4.245507584597433,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00012353354385009769,
      "loss": 0.0074,
      "step": 9096
    },
    {
      "epoch": 4.245974329054842,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012351928223041836,
      "loss": 0.021,
      "step": 9097
    },
    {
      "epoch": 4.246441073512252,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001235050201043428,
      "loss": 0.0119,
      "step": 9098
    },
    {
      "epoch": 4.246907817969662,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012349075747217808,
      "loss": 0.0057,
      "step": 9099
    },
    {
      "epoch": 4.247374562427071,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012347649433423129,
      "loss": 0.0072,
      "step": 9100
    },
    {
      "epoch": 4.247841306884481,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00012346223069080952,
      "loss": 0.0061,
      "step": 9101
    },
    {
      "epoch": 4.2483080513418905,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001234479665422199,
      "loss": 0.0047,
      "step": 9102
    },
    {
      "epoch": 4.2487747957993,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012343370188876954,
      "loss": 0.0106,
      "step": 9103
    },
    {
      "epoch": 4.249241540256709,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012341943673076558,
      "loss": 0.0035,
      "step": 9104
    },
    {
      "epoch": 4.249708284714119,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001234051710685152,
      "loss": 0.008,
      "step": 9105
    },
    {
      "epoch": 4.250175029171529,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00012339090490232546,
      "loss": 0.0118,
      "step": 9106
    },
    {
      "epoch": 4.250641773628939,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00012337663823250362,
      "loss": 0.0081,
      "step": 9107
    },
    {
      "epoch": 4.2511085180863475,
      "grad_norm": 0.26171875,
      "learning_rate": 0.0001233623710593568,
      "loss": 0.0044,
      "step": 9108
    },
    {
      "epoch": 4.251575262543757,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00012334810338319224,
      "loss": 0.0042,
      "step": 9109
    },
    {
      "epoch": 4.252042007001167,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00012333383520431708,
      "loss": 0.0053,
      "step": 9110
    },
    {
      "epoch": 4.252508751458576,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00012331956652303857,
      "loss": 0.0042,
      "step": 9111
    },
    {
      "epoch": 4.252975495915986,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00012330529733966392,
      "loss": 0.0108,
      "step": 9112
    },
    {
      "epoch": 4.252975495915986,
      "eval_loss": 1.994627833366394,
      "eval_runtime": 55.4701,
      "eval_samples_per_second": 32.522,
      "eval_steps_per_second": 4.074,
      "step": 9112
    },
    {
      "epoch": 4.253442240373396,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012329102765450034,
      "loss": 0.0114,
      "step": 9113
    },
    {
      "epoch": 4.253908984830805,
      "grad_norm": 0.474609375,
      "learning_rate": 0.0001232767574678551,
      "loss": 0.0087,
      "step": 9114
    },
    {
      "epoch": 4.254375729288215,
      "grad_norm": 0.421875,
      "learning_rate": 0.00012326248678003548,
      "loss": 0.0071,
      "step": 9115
    },
    {
      "epoch": 4.254842473745624,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00012324821559134866,
      "loss": 0.0048,
      "step": 9116
    },
    {
      "epoch": 4.255309218203034,
      "grad_norm": 0.27734375,
      "learning_rate": 0.000123233943902102,
      "loss": 0.0049,
      "step": 9117
    },
    {
      "epoch": 4.255775962660444,
      "grad_norm": 0.359375,
      "learning_rate": 0.00012321967171260276,
      "loss": 0.0066,
      "step": 9118
    },
    {
      "epoch": 4.256242707117853,
      "grad_norm": 0.9375,
      "learning_rate": 0.0001232053990231582,
      "loss": 0.0143,
      "step": 9119
    },
    {
      "epoch": 4.256709451575262,
      "grad_norm": 0.267578125,
      "learning_rate": 0.00012319112583407564,
      "loss": 0.0034,
      "step": 9120
    },
    {
      "epoch": 4.257176196032672,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00012317685214566242,
      "loss": 0.0146,
      "step": 9121
    },
    {
      "epoch": 4.257642940490082,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001231625779582259,
      "loss": 0.009,
      "step": 9122
    },
    {
      "epoch": 4.258109684947491,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00012314830327207331,
      "loss": 0.0072,
      "step": 9123
    },
    {
      "epoch": 4.258576429404901,
      "grad_norm": 0.291015625,
      "learning_rate": 0.0001231340280875121,
      "loss": 0.0145,
      "step": 9124
    },
    {
      "epoch": 4.2590431738623105,
      "grad_norm": 0.421875,
      "learning_rate": 0.00012311975240484962,
      "loss": 0.0093,
      "step": 9125
    },
    {
      "epoch": 4.25950991831972,
      "grad_norm": 0.28515625,
      "learning_rate": 0.00012310547622439318,
      "loss": 0.0031,
      "step": 9126
    },
    {
      "epoch": 4.259976662777129,
      "grad_norm": 0.734375,
      "learning_rate": 0.00012309119954645022,
      "loss": 0.0175,
      "step": 9127
    },
    {
      "epoch": 4.260443407234539,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00012307692237132812,
      "loss": 0.011,
      "step": 9128
    },
    {
      "epoch": 4.260910151691949,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00012306264469933426,
      "loss": 0.0061,
      "step": 9129
    },
    {
      "epoch": 4.2613768961493586,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00012304836653077609,
      "loss": 0.0158,
      "step": 9130
    },
    {
      "epoch": 4.2618436406067675,
      "grad_norm": 0.26171875,
      "learning_rate": 0.00012303408786596104,
      "loss": 0.0068,
      "step": 9131
    },
    {
      "epoch": 4.262310385064177,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012301980870519646,
      "loss": 0.0114,
      "step": 9132
    },
    {
      "epoch": 4.262777129521587,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001230055290487899,
      "loss": 0.0067,
      "step": 9133
    },
    {
      "epoch": 4.263243873978997,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00012299124889704873,
      "loss": 0.0114,
      "step": 9134
    },
    {
      "epoch": 4.263710618436406,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001229769682502805,
      "loss": 0.0118,
      "step": 9135
    },
    {
      "epoch": 4.2641773628938155,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00012296268710879264,
      "loss": 0.0073,
      "step": 9136
    },
    {
      "epoch": 4.264644107351225,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012294840547289262,
      "loss": 0.0098,
      "step": 9137
    },
    {
      "epoch": 4.265110851808635,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00012293412334288804,
      "loss": 0.0163,
      "step": 9138
    },
    {
      "epoch": 4.265577596266044,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012291984071908626,
      "loss": 0.0064,
      "step": 9139
    },
    {
      "epoch": 4.266044340723454,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00012290555760179495,
      "loss": 0.0067,
      "step": 9140
    },
    {
      "epoch": 4.266511085180864,
      "grad_norm": 0.267578125,
      "learning_rate": 0.00012289127399132152,
      "loss": 0.0056,
      "step": 9141
    },
    {
      "epoch": 4.266977829638273,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012287698988797357,
      "loss": 0.0076,
      "step": 9142
    },
    {
      "epoch": 4.267444574095682,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00012286270529205865,
      "loss": 0.0097,
      "step": 9143
    },
    {
      "epoch": 4.267911318553092,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012284842020388433,
      "loss": 0.0051,
      "step": 9144
    },
    {
      "epoch": 4.268378063010502,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00012283413462375816,
      "loss": 0.0098,
      "step": 9145
    },
    {
      "epoch": 4.268844807467912,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012281984855198774,
      "loss": 0.0149,
      "step": 9146
    },
    {
      "epoch": 4.269311551925321,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00012280556198888067,
      "loss": 0.0061,
      "step": 9147
    },
    {
      "epoch": 4.26977829638273,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00012279127493474454,
      "loss": 0.0081,
      "step": 9148
    },
    {
      "epoch": 4.27024504084014,
      "grad_norm": 0.2314453125,
      "learning_rate": 0.00012277698738988695,
      "loss": 0.0044,
      "step": 9149
    },
    {
      "epoch": 4.27071178529755,
      "grad_norm": 0.2294921875,
      "learning_rate": 0.00012276269935461562,
      "loss": 0.0046,
      "step": 9150
    },
    {
      "epoch": 4.271178529754959,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00012274841082923806,
      "loss": 0.0073,
      "step": 9151
    },
    {
      "epoch": 4.271645274212369,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00012273412181406198,
      "loss": 0.009,
      "step": 9152
    },
    {
      "epoch": 4.2721120186697785,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00012271983230939504,
      "loss": 0.0045,
      "step": 9153
    },
    {
      "epoch": 4.272578763127187,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00012270554231554493,
      "loss": 0.0136,
      "step": 9154
    },
    {
      "epoch": 4.273045507584597,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00012269125183281924,
      "loss": 0.006,
      "step": 9155
    },
    {
      "epoch": 4.273512252042007,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001226769608615258,
      "loss": 0.0073,
      "step": 9156
    },
    {
      "epoch": 4.273978996499417,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00012266266940197217,
      "loss": 0.0073,
      "step": 9157
    },
    {
      "epoch": 4.274445740956826,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00012264837745446612,
      "loss": 0.0073,
      "step": 9158
    },
    {
      "epoch": 4.2749124854142355,
      "grad_norm": 0.5,
      "learning_rate": 0.0001226340850193154,
      "loss": 0.0105,
      "step": 9159
    },
    {
      "epoch": 4.275379229871645,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00012261979209682772,
      "loss": 0.0141,
      "step": 9160
    },
    {
      "epoch": 4.275845974329055,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00012260549868731082,
      "loss": 0.0099,
      "step": 9161
    },
    {
      "epoch": 4.276312718786464,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001225912047910724,
      "loss": 0.0046,
      "step": 9162
    },
    {
      "epoch": 4.276779463243874,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00012257691040842034,
      "loss": 0.0076,
      "step": 9163
    },
    {
      "epoch": 4.277246207701284,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00012256261553966234,
      "loss": 0.01,
      "step": 9164
    },
    {
      "epoch": 4.277712952158693,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00012254832018510615,
      "loss": 0.0082,
      "step": 9165
    },
    {
      "epoch": 4.278179696616102,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012253402434505965,
      "loss": 0.0149,
      "step": 9166
    },
    {
      "epoch": 4.278646441073512,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00012251972801983062,
      "loss": 0.0113,
      "step": 9167
    },
    {
      "epoch": 4.279113185530922,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00012250543120972683,
      "loss": 0.0062,
      "step": 9168
    },
    {
      "epoch": 4.279579929988332,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00012249113391505612,
      "loss": 0.0156,
      "step": 9169
    },
    {
      "epoch": 4.280046674445741,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00012247683613612637,
      "loss": 0.0067,
      "step": 9170
    },
    {
      "epoch": 4.28051341890315,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001224625378732454,
      "loss": 0.0059,
      "step": 9171
    },
    {
      "epoch": 4.28098016336056,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00012244823912672102,
      "loss": 0.0042,
      "step": 9172
    },
    {
      "epoch": 4.28144690781797,
      "grad_norm": 0.8671875,
      "learning_rate": 0.0001224339398968612,
      "loss": 0.0078,
      "step": 9173
    },
    {
      "epoch": 4.281913652275379,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00012241964018397375,
      "loss": 0.0062,
      "step": 9174
    },
    {
      "epoch": 4.282380396732789,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012240533998836658,
      "loss": 0.0114,
      "step": 9175
    },
    {
      "epoch": 4.2828471411901985,
      "grad_norm": 0.546875,
      "learning_rate": 0.00012239103931034758,
      "loss": 0.0097,
      "step": 9176
    },
    {
      "epoch": 4.283313885647608,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00012237673815022467,
      "loss": 0.0062,
      "step": 9177
    },
    {
      "epoch": 4.283780630105017,
      "grad_norm": 0.3046875,
      "learning_rate": 0.0001223624365083057,
      "loss": 0.0048,
      "step": 9178
    },
    {
      "epoch": 4.284247374562427,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001223481343848987,
      "loss": 0.0083,
      "step": 9179
    },
    {
      "epoch": 4.284714119019837,
      "grad_norm": 0.470703125,
      "learning_rate": 0.0001223338317803116,
      "loss": 0.0051,
      "step": 9180
    },
    {
      "epoch": 4.285180863477247,
      "grad_norm": 0.703125,
      "learning_rate": 0.0001223195286948523,
      "loss": 0.0077,
      "step": 9181
    },
    {
      "epoch": 4.2856476079346555,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00012230522512882876,
      "loss": 0.0105,
      "step": 9182
    },
    {
      "epoch": 4.286114352392065,
      "grad_norm": 0.447265625,
      "learning_rate": 0.000122290921082549,
      "loss": 0.0072,
      "step": 9183
    },
    {
      "epoch": 4.286581096849475,
      "grad_norm": 0.6875,
      "learning_rate": 0.00012227661655632096,
      "loss": 0.0124,
      "step": 9184
    },
    {
      "epoch": 4.287047841306885,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00012226231155045263,
      "loss": 0.005,
      "step": 9185
    },
    {
      "epoch": 4.287514585764294,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00012224800606525207,
      "loss": 0.0034,
      "step": 9186
    },
    {
      "epoch": 4.287981330221704,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00012223370010102725,
      "loss": 0.0092,
      "step": 9187
    },
    {
      "epoch": 4.288448074679113,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012221939365808618,
      "loss": 0.0117,
      "step": 9188
    },
    {
      "epoch": 4.288914819136522,
      "grad_norm": 0.2578125,
      "learning_rate": 0.00012220508673673695,
      "loss": 0.0028,
      "step": 9189
    },
    {
      "epoch": 4.289381563593932,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012219077933728754,
      "loss": 0.0087,
      "step": 9190
    },
    {
      "epoch": 4.289848308051342,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00012217647146004602,
      "loss": 0.0089,
      "step": 9191
    },
    {
      "epoch": 4.290315052508752,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012216216310532047,
      "loss": 0.0135,
      "step": 9192
    },
    {
      "epoch": 4.2907817969661615,
      "grad_norm": 0.703125,
      "learning_rate": 0.00012214785427341898,
      "loss": 0.0079,
      "step": 9193
    },
    {
      "epoch": 4.29124854142357,
      "grad_norm": 0.49609375,
      "learning_rate": 0.0001221335449646496,
      "loss": 0.0073,
      "step": 9194
    },
    {
      "epoch": 4.29171528588098,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001221192351793204,
      "loss": 0.0095,
      "step": 9195
    },
    {
      "epoch": 4.29218203033839,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00012210492491773956,
      "loss": 0.0058,
      "step": 9196
    },
    {
      "epoch": 4.292648774795799,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012209061418021517,
      "loss": 0.0057,
      "step": 9197
    },
    {
      "epoch": 4.293115519253209,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012207630296705535,
      "loss": 0.0086,
      "step": 9198
    },
    {
      "epoch": 4.2935822637106185,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012206199127856823,
      "loss": 0.0084,
      "step": 9199
    },
    {
      "epoch": 4.294049008168028,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00012204767911506196,
      "loss": 0.0054,
      "step": 9200
    },
    {
      "epoch": 4.294515752625437,
      "grad_norm": 0.3046875,
      "learning_rate": 0.00012203336647684467,
      "loss": 0.0049,
      "step": 9201
    },
    {
      "epoch": 4.294982497082847,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012201905336422456,
      "loss": 0.0058,
      "step": 9202
    },
    {
      "epoch": 4.295449241540257,
      "grad_norm": 0.625,
      "learning_rate": 0.00012200473977750983,
      "loss": 0.0101,
      "step": 9203
    },
    {
      "epoch": 4.2959159859976666,
      "grad_norm": 0.353515625,
      "learning_rate": 0.0001219904257170086,
      "loss": 0.0046,
      "step": 9204
    },
    {
      "epoch": 4.2963827304550755,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00012197611118302915,
      "loss": 0.0065,
      "step": 9205
    },
    {
      "epoch": 4.296849474912485,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00012196179617587961,
      "loss": 0.0051,
      "step": 9206
    },
    {
      "epoch": 4.297316219369895,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001219474806958682,
      "loss": 0.0087,
      "step": 9207
    },
    {
      "epoch": 4.297782963827305,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00012193316474330325,
      "loss": 0.0043,
      "step": 9208
    },
    {
      "epoch": 4.298249708284714,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00012191884831849289,
      "loss": 0.0102,
      "step": 9209
    },
    {
      "epoch": 4.2987164527421236,
      "grad_norm": 0.609375,
      "learning_rate": 0.00012190453142174543,
      "loss": 0.0167,
      "step": 9210
    },
    {
      "epoch": 4.299183197199533,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00012189021405336906,
      "loss": 0.0068,
      "step": 9211
    },
    {
      "epoch": 4.299649941656943,
      "grad_norm": 0.5,
      "learning_rate": 0.00012187589621367213,
      "loss": 0.009,
      "step": 9212
    },
    {
      "epoch": 4.300116686114352,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012186157790296289,
      "loss": 0.0122,
      "step": 9213
    },
    {
      "epoch": 4.300583430571762,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012184725912154959,
      "loss": 0.01,
      "step": 9214
    },
    {
      "epoch": 4.301050175029172,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00012183293986974058,
      "loss": 0.0087,
      "step": 9215
    },
    {
      "epoch": 4.301516919486581,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00012181862014784415,
      "loss": 0.0124,
      "step": 9216
    },
    {
      "epoch": 4.30198366394399,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001218042999561686,
      "loss": 0.0105,
      "step": 9217
    },
    {
      "epoch": 4.3024504084014,
      "grad_norm": 0.640625,
      "learning_rate": 0.0001217899792950223,
      "loss": 0.011,
      "step": 9218
    },
    {
      "epoch": 4.30291715285881,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012177565816471357,
      "loss": 0.0051,
      "step": 9219
    },
    {
      "epoch": 4.30338389731622,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00012176133656555078,
      "loss": 0.0085,
      "step": 9220
    },
    {
      "epoch": 4.303850641773629,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012174701449784223,
      "loss": 0.0069,
      "step": 9221
    },
    {
      "epoch": 4.304317386231038,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00012173269196189636,
      "loss": 0.008,
      "step": 9222
    },
    {
      "epoch": 4.304784130688448,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012171836895802149,
      "loss": 0.0115,
      "step": 9223
    },
    {
      "epoch": 4.305250875145858,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012170404548652606,
      "loss": 0.0114,
      "step": 9224
    },
    {
      "epoch": 4.305717619603267,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012168972154771844,
      "loss": 0.0113,
      "step": 9225
    },
    {
      "epoch": 4.306184364060677,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00012167539714190705,
      "loss": 0.0048,
      "step": 9226
    },
    {
      "epoch": 4.3066511085180865,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00012166107226940032,
      "loss": 0.0058,
      "step": 9227
    },
    {
      "epoch": 4.307117852975496,
      "grad_norm": 0.234375,
      "learning_rate": 0.00012164674693050666,
      "loss": 0.0033,
      "step": 9228
    },
    {
      "epoch": 4.307584597432905,
      "grad_norm": 0.5,
      "learning_rate": 0.00012163242112553453,
      "loss": 0.0068,
      "step": 9229
    },
    {
      "epoch": 4.308051341890315,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00012161809485479234,
      "loss": 0.0093,
      "step": 9230
    },
    {
      "epoch": 4.308518086347725,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012160376811858863,
      "loss": 0.0128,
      "step": 9231
    },
    {
      "epoch": 4.308984830805134,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001215894409172318,
      "loss": 0.0043,
      "step": 9232
    },
    {
      "epoch": 4.3094515752625435,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00012157511325103033,
      "loss": 0.0108,
      "step": 9233
    },
    {
      "epoch": 4.309918319719953,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00012156078512029276,
      "loss": 0.0095,
      "step": 9234
    },
    {
      "epoch": 4.310385064177363,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012154645652532757,
      "loss": 0.0125,
      "step": 9235
    },
    {
      "epoch": 4.310851808634773,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00012153212746644327,
      "loss": 0.0113,
      "step": 9236
    },
    {
      "epoch": 4.311318553092182,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00012151779794394835,
      "loss": 0.0087,
      "step": 9237
    },
    {
      "epoch": 4.311785297549592,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00012150346795815137,
      "loss": 0.0091,
      "step": 9238
    },
    {
      "epoch": 4.312252042007001,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00012148913750936089,
      "loss": 0.006,
      "step": 9239
    },
    {
      "epoch": 4.31271878646441,
      "grad_norm": 0.953125,
      "learning_rate": 0.00012147480659788542,
      "loss": 0.0075,
      "step": 9240
    },
    {
      "epoch": 4.31318553092182,
      "grad_norm": 0.484375,
      "learning_rate": 0.00012146047522403354,
      "loss": 0.0072,
      "step": 9241
    },
    {
      "epoch": 4.31365227537923,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00012144614338811382,
      "loss": 0.0111,
      "step": 9242
    },
    {
      "epoch": 4.31411901983664,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00012143181109043484,
      "loss": 0.0058,
      "step": 9243
    },
    {
      "epoch": 4.314585764294049,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00012141747833130517,
      "loss": 0.0092,
      "step": 9244
    },
    {
      "epoch": 4.315052508751458,
      "grad_norm": 0.53125,
      "learning_rate": 0.00012140314511103344,
      "loss": 0.009,
      "step": 9245
    },
    {
      "epoch": 4.315519253208868,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012138881142992824,
      "loss": 0.0105,
      "step": 9246
    },
    {
      "epoch": 4.315985997666278,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00012137447728829823,
      "loss": 0.0128,
      "step": 9247
    },
    {
      "epoch": 4.316452742123687,
      "grad_norm": 0.59765625,
      "learning_rate": 0.000121360142686452,
      "loss": 0.0087,
      "step": 9248
    },
    {
      "epoch": 4.316919486581097,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012134580762469818,
      "loss": 0.0092,
      "step": 9249
    },
    {
      "epoch": 4.3173862310385065,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00012133147210334544,
      "loss": 0.0086,
      "step": 9250
    },
    {
      "epoch": 4.317852975495916,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00012131713612270243,
      "loss": 0.0075,
      "step": 9251
    },
    {
      "epoch": 4.318319719953325,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012130279968307785,
      "loss": 0.0112,
      "step": 9252
    },
    {
      "epoch": 4.318786464410735,
      "grad_norm": 0.671875,
      "learning_rate": 0.00012128846278478033,
      "loss": 0.0176,
      "step": 9253
    },
    {
      "epoch": 4.319253208868145,
      "grad_norm": 1.0234375,
      "learning_rate": 0.00012127412542811861,
      "loss": 0.0233,
      "step": 9254
    },
    {
      "epoch": 4.319719953325555,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00012125978761340135,
      "loss": 0.0091,
      "step": 9255
    },
    {
      "epoch": 4.3201866977829635,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00012124544934093727,
      "loss": 0.0049,
      "step": 9256
    },
    {
      "epoch": 4.320653442240373,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00012123111061103512,
      "loss": 0.0083,
      "step": 9257
    },
    {
      "epoch": 4.321120186697783,
      "grad_norm": 0.484375,
      "learning_rate": 0.00012121677142400359,
      "loss": 0.017,
      "step": 9258
    },
    {
      "epoch": 4.321586931155193,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012120243178015142,
      "loss": 0.01,
      "step": 9259
    },
    {
      "epoch": 4.322053675612602,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00012118809167978737,
      "loss": 0.0111,
      "step": 9260
    },
    {
      "epoch": 4.322520420070012,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00012117375112322021,
      "loss": 0.0051,
      "step": 9261
    },
    {
      "epoch": 4.322987164527421,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001211594101107587,
      "loss": 0.0118,
      "step": 9262
    },
    {
      "epoch": 4.323453908984831,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00012114506864271158,
      "loss": 0.0124,
      "step": 9263
    },
    {
      "epoch": 4.32392065344224,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001211307267193877,
      "loss": 0.0085,
      "step": 9264
    },
    {
      "epoch": 4.32438739789965,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00012111638434109584,
      "loss": 0.0117,
      "step": 9265
    },
    {
      "epoch": 4.32485414235706,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00012110204150814477,
      "loss": 0.0105,
      "step": 9266
    },
    {
      "epoch": 4.3253208868144695,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00012108769822084334,
      "loss": 0.0106,
      "step": 9267
    },
    {
      "epoch": 4.325787631271878,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00012107335447950037,
      "loss": 0.0154,
      "step": 9268
    },
    {
      "epoch": 4.326254375729288,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00012105901028442473,
      "loss": 0.0056,
      "step": 9269
    },
    {
      "epoch": 4.326721120186698,
      "grad_norm": 0.53515625,
      "learning_rate": 0.0001210446656359252,
      "loss": 0.0092,
      "step": 9270
    },
    {
      "epoch": 4.327187864644108,
      "grad_norm": 0.5,
      "learning_rate": 0.00012103032053431067,
      "loss": 0.0112,
      "step": 9271
    },
    {
      "epoch": 4.327654609101517,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00012101597497988999,
      "loss": 0.0123,
      "step": 9272
    },
    {
      "epoch": 4.3281213535589265,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00012100162897297207,
      "loss": 0.015,
      "step": 9273
    },
    {
      "epoch": 4.328588098016336,
      "grad_norm": 0.44140625,
      "learning_rate": 0.0001209872825138658,
      "loss": 0.0097,
      "step": 9274
    },
    {
      "epoch": 4.329054842473745,
      "grad_norm": 0.578125,
      "learning_rate": 0.00012097293560288003,
      "loss": 0.0127,
      "step": 9275
    },
    {
      "epoch": 4.329521586931155,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00012095858824032367,
      "loss": 0.0152,
      "step": 9276
    },
    {
      "epoch": 4.329988331388565,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00012094424042650568,
      "loss": 0.0155,
      "step": 9277
    },
    {
      "epoch": 4.3304550758459746,
      "grad_norm": 0.25,
      "learning_rate": 0.00012092989216173493,
      "loss": 0.0036,
      "step": 9278
    },
    {
      "epoch": 4.330921820303384,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00012091554344632036,
      "loss": 0.0047,
      "step": 9279
    },
    {
      "epoch": 4.331388564760793,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00012090119428057099,
      "loss": 0.0065,
      "step": 9280
    },
    {
      "epoch": 4.331855309218203,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00012088684466479569,
      "loss": 0.009,
      "step": 9281
    },
    {
      "epoch": 4.332322053675613,
      "grad_norm": 0.375,
      "learning_rate": 0.00012087249459930343,
      "loss": 0.0085,
      "step": 9282
    },
    {
      "epoch": 4.332788798133022,
      "grad_norm": 0.453125,
      "learning_rate": 0.00012085814408440324,
      "loss": 0.008,
      "step": 9283
    },
    {
      "epoch": 4.3332555425904316,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00012084379312040406,
      "loss": 0.0127,
      "step": 9284
    },
    {
      "epoch": 4.333722287047841,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00012082944170761488,
      "loss": 0.0155,
      "step": 9285
    },
    {
      "epoch": 4.334189031505251,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00012081508984634468,
      "loss": 0.0133,
      "step": 9286
    },
    {
      "epoch": 4.33465577596266,
      "grad_norm": 0.46875,
      "learning_rate": 0.00012080073753690253,
      "loss": 0.0082,
      "step": 9287
    },
    {
      "epoch": 4.33512252042007,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012078638477959743,
      "loss": 0.009,
      "step": 9288
    },
    {
      "epoch": 4.33558926487748,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00012077203157473838,
      "loss": 0.0066,
      "step": 9289
    },
    {
      "epoch": 4.336056009334889,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00012075767792263446,
      "loss": 0.0052,
      "step": 9290
    },
    {
      "epoch": 4.336522753792298,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001207433238235947,
      "loss": 0.0125,
      "step": 9291
    },
    {
      "epoch": 4.336989498249708,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00012072896927792818,
      "loss": 0.0163,
      "step": 9292
    },
    {
      "epoch": 4.337456242707118,
      "grad_norm": 0.357421875,
      "learning_rate": 0.0001207146142859439,
      "loss": 0.0065,
      "step": 9293
    },
    {
      "epoch": 4.337922987164528,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00012070025884795107,
      "loss": 0.0086,
      "step": 9294
    },
    {
      "epoch": 4.338389731621937,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012068590296425864,
      "loss": 0.0098,
      "step": 9295
    },
    {
      "epoch": 4.338856476079346,
      "grad_norm": 0.625,
      "learning_rate": 0.00012067154663517577,
      "loss": 0.008,
      "step": 9296
    },
    {
      "epoch": 4.339323220536756,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0001206571898610116,
      "loss": 0.0038,
      "step": 9297
    },
    {
      "epoch": 4.339789964994166,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00012064283264207516,
      "loss": 0.0073,
      "step": 9298
    },
    {
      "epoch": 4.340256709451575,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00012062847497867566,
      "loss": 0.0118,
      "step": 9299
    },
    {
      "epoch": 4.340723453908985,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00012061411687112223,
      "loss": 0.0074,
      "step": 9300
    },
    {
      "epoch": 4.3411901983663945,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00012059975831972397,
      "loss": 0.007,
      "step": 9301
    },
    {
      "epoch": 4.341656942823804,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00012058539932479003,
      "loss": 0.0089,
      "step": 9302
    },
    {
      "epoch": 4.342123687281213,
      "grad_norm": 0.375,
      "learning_rate": 0.00012057103988662964,
      "loss": 0.0056,
      "step": 9303
    },
    {
      "epoch": 4.342590431738623,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00012055668000555194,
      "loss": 0.0102,
      "step": 9304
    },
    {
      "epoch": 4.343057176196033,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00012054231968186606,
      "loss": 0.0052,
      "step": 9305
    },
    {
      "epoch": 4.343523920653443,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00012052795891588129,
      "loss": 0.0102,
      "step": 9306
    },
    {
      "epoch": 4.3439906651108515,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001205135977079068,
      "loss": 0.0128,
      "step": 9307
    },
    {
      "epoch": 4.344457409568261,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012049923605825178,
      "loss": 0.0083,
      "step": 9308
    },
    {
      "epoch": 4.344924154025671,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00012048487396722543,
      "loss": 0.0063,
      "step": 9309
    },
    {
      "epoch": 4.345390898483081,
      "grad_norm": 0.421875,
      "learning_rate": 0.00012047051143513706,
      "loss": 0.0125,
      "step": 9310
    },
    {
      "epoch": 4.34585764294049,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00012045614846229584,
      "loss": 0.005,
      "step": 9311
    },
    {
      "epoch": 4.3463243873979,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00012044178504901105,
      "loss": 0.0072,
      "step": 9312
    },
    {
      "epoch": 4.346791131855309,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00012042742119559194,
      "loss": 0.0129,
      "step": 9313
    },
    {
      "epoch": 4.347257876312719,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00012041305690234781,
      "loss": 0.0143,
      "step": 9314
    },
    {
      "epoch": 4.347724620770128,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00012039869216958788,
      "loss": 0.0089,
      "step": 9315
    },
    {
      "epoch": 4.348191365227538,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00012038432699762152,
      "loss": 0.0103,
      "step": 9316
    },
    {
      "epoch": 4.348658109684948,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00012036996138675796,
      "loss": 0.0061,
      "step": 9317
    },
    {
      "epoch": 4.349124854142357,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00012035559533730653,
      "loss": 0.0091,
      "step": 9318
    },
    {
      "epoch": 4.349591598599766,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00012034122884957657,
      "loss": 0.0046,
      "step": 9319
    },
    {
      "epoch": 4.350058343057176,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00012032686192387736,
      "loss": 0.0068,
      "step": 9320
    },
    {
      "epoch": 4.350525087514586,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00012031249456051825,
      "loss": 0.0096,
      "step": 9321
    },
    {
      "epoch": 4.350991831971996,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001202981267598086,
      "loss": 0.0072,
      "step": 9322
    },
    {
      "epoch": 4.351458576429405,
      "grad_norm": 0.25,
      "learning_rate": 0.0001202837585220578,
      "loss": 0.0043,
      "step": 9323
    },
    {
      "epoch": 4.3519253208868145,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00012026938984757514,
      "loss": 0.0038,
      "step": 9324
    },
    {
      "epoch": 4.352392065344224,
      "grad_norm": 0.40625,
      "learning_rate": 0.00012025502073667001,
      "loss": 0.0069,
      "step": 9325
    },
    {
      "epoch": 4.352858809801633,
      "grad_norm": 0.65625,
      "learning_rate": 0.00012024065118965184,
      "loss": 0.0125,
      "step": 9326
    },
    {
      "epoch": 4.353325554259043,
      "grad_norm": 0.75,
      "learning_rate": 0.00012022628120683,
      "loss": 0.0127,
      "step": 9327
    },
    {
      "epoch": 4.353792298716453,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012021191078851384,
      "loss": 0.0074,
      "step": 9328
    },
    {
      "epoch": 4.354259043173863,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00012019753993501284,
      "loss": 0.0038,
      "step": 9329
    },
    {
      "epoch": 4.3547257876312715,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00012018316864663641,
      "loss": 0.0181,
      "step": 9330
    },
    {
      "epoch": 4.355192532088681,
      "grad_norm": 0.4375,
      "learning_rate": 0.00012016879692369394,
      "loss": 0.0078,
      "step": 9331
    },
    {
      "epoch": 4.355659276546091,
      "grad_norm": 0.4453125,
      "learning_rate": 0.0001201544247664949,
      "loss": 0.0061,
      "step": 9332
    },
    {
      "epoch": 4.356126021003501,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00012014005217534874,
      "loss": 0.005,
      "step": 9333
    },
    {
      "epoch": 4.35659276546091,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00012012567915056492,
      "loss": 0.0086,
      "step": 9334
    },
    {
      "epoch": 4.35705950991832,
      "grad_norm": 0.515625,
      "learning_rate": 0.00012011130569245288,
      "loss": 0.0073,
      "step": 9335
    },
    {
      "epoch": 4.357526254375729,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00012009693180132213,
      "loss": 0.0067,
      "step": 9336
    },
    {
      "epoch": 4.357992998833139,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00012008255747748214,
      "loss": 0.0099,
      "step": 9337
    },
    {
      "epoch": 4.358459743290548,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00012006818272124238,
      "loss": 0.0051,
      "step": 9338
    },
    {
      "epoch": 4.358926487747958,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001200538075329124,
      "loss": 0.012,
      "step": 9339
    },
    {
      "epoch": 4.359393232205368,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00012003943191280171,
      "loss": 0.0036,
      "step": 9340
    },
    {
      "epoch": 4.3598599766627775,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00012002505586121977,
      "loss": 0.0159,
      "step": 9341
    },
    {
      "epoch": 4.360326721120186,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00012001067937847623,
      "loss": 0.0078,
      "step": 9342
    },
    {
      "epoch": 4.360793465577596,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011999630246488051,
      "loss": 0.0142,
      "step": 9343
    },
    {
      "epoch": 4.361260210035006,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00011998192512074222,
      "loss": 0.0043,
      "step": 9344
    },
    {
      "epoch": 4.361726954492416,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00011996754734637092,
      "loss": 0.0074,
      "step": 9345
    },
    {
      "epoch": 4.362193698949825,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011995316914207617,
      "loss": 0.0068,
      "step": 9346
    },
    {
      "epoch": 4.3626604434072345,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00011993879050816751,
      "loss": 0.0203,
      "step": 9347
    },
    {
      "epoch": 4.363127187864644,
      "grad_norm": 0.400390625,
      "learning_rate": 0.0001199244114449546,
      "loss": 0.0065,
      "step": 9348
    },
    {
      "epoch": 4.363593932322054,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011991003195274704,
      "loss": 0.0097,
      "step": 9349
    },
    {
      "epoch": 4.364060676779463,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00011989565203185437,
      "loss": 0.0105,
      "step": 9350
    },
    {
      "epoch": 4.364527421236873,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011988127168258623,
      "loss": 0.0089,
      "step": 9351
    },
    {
      "epoch": 4.364994165694283,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00011986689090525225,
      "loss": 0.0076,
      "step": 9352
    },
    {
      "epoch": 4.365460910151692,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00011985250970016209,
      "loss": 0.0049,
      "step": 9353
    },
    {
      "epoch": 4.365927654609101,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00011983812806762534,
      "loss": 0.0102,
      "step": 9354
    },
    {
      "epoch": 4.366394399066511,
      "grad_norm": 0.40234375,
      "learning_rate": 0.0001198237460079517,
      "loss": 0.0118,
      "step": 9355
    },
    {
      "epoch": 4.366861143523921,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001198093635214508,
      "loss": 0.0076,
      "step": 9356
    },
    {
      "epoch": 4.367327887981331,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011979498060843234,
      "loss": 0.0137,
      "step": 9357
    },
    {
      "epoch": 4.3677946324387396,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00011978059726920596,
      "loss": 0.0075,
      "step": 9358
    },
    {
      "epoch": 4.368261376896149,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00011976621350408142,
      "loss": 0.0048,
      "step": 9359
    },
    {
      "epoch": 4.368728121353559,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00011975182931336834,
      "loss": 0.0135,
      "step": 9360
    },
    {
      "epoch": 4.369194865810968,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00011973744469737644,
      "loss": 0.0079,
      "step": 9361
    },
    {
      "epoch": 4.369661610268378,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011972305965641549,
      "loss": 0.011,
      "step": 9362
    },
    {
      "epoch": 4.370128354725788,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00011970867419079519,
      "loss": 0.0076,
      "step": 9363
    },
    {
      "epoch": 4.370595099183197,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00011969428830082523,
      "loss": 0.0085,
      "step": 9364
    },
    {
      "epoch": 4.371061843640607,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011967990198681544,
      "loss": 0.013,
      "step": 9365
    },
    {
      "epoch": 4.371528588098016,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001196655152490755,
      "loss": 0.007,
      "step": 9366
    },
    {
      "epoch": 4.371995332555426,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00011965112808791519,
      "loss": 0.0064,
      "step": 9367
    },
    {
      "epoch": 4.372462077012836,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00011963674050364429,
      "loss": 0.0081,
      "step": 9368
    },
    {
      "epoch": 4.372928821470245,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00011962235249657258,
      "loss": 0.0061,
      "step": 9369
    },
    {
      "epoch": 4.373395565927654,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00011960796406700985,
      "loss": 0.0074,
      "step": 9370
    },
    {
      "epoch": 4.373862310385064,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00011959357521526591,
      "loss": 0.0067,
      "step": 9371
    },
    {
      "epoch": 4.374329054842474,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00011957918594165055,
      "loss": 0.0074,
      "step": 9372
    },
    {
      "epoch": 4.374795799299883,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011956479624647359,
      "loss": 0.007,
      "step": 9373
    },
    {
      "epoch": 4.375262543757293,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00011955040613004484,
      "loss": 0.0135,
      "step": 9374
    },
    {
      "epoch": 4.3757292882147025,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011953601559267418,
      "loss": 0.0096,
      "step": 9375
    },
    {
      "epoch": 4.376196032672112,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00011952162463467144,
      "loss": 0.0056,
      "step": 9376
    },
    {
      "epoch": 4.376662777129521,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00011950723325634643,
      "loss": 0.0071,
      "step": 9377
    },
    {
      "epoch": 4.377129521586931,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011949284145800905,
      "loss": 0.01,
      "step": 9378
    },
    {
      "epoch": 4.377596266044341,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011947844923996918,
      "loss": 0.0176,
      "step": 9379
    },
    {
      "epoch": 4.378063010501751,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011946405660253668,
      "loss": 0.0184,
      "step": 9380
    },
    {
      "epoch": 4.3785297549591595,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011944966354602145,
      "loss": 0.0091,
      "step": 9381
    },
    {
      "epoch": 4.378996499416569,
      "grad_norm": 0.328125,
      "learning_rate": 0.00011943527007073338,
      "loss": 0.0038,
      "step": 9382
    },
    {
      "epoch": 4.379463243873979,
      "grad_norm": 0.25,
      "learning_rate": 0.00011942087617698239,
      "loss": 0.0036,
      "step": 9383
    },
    {
      "epoch": 4.379929988331389,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011940648186507837,
      "loss": 0.0129,
      "step": 9384
    },
    {
      "epoch": 4.380396732788798,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011939208713533127,
      "loss": 0.0115,
      "step": 9385
    },
    {
      "epoch": 4.380863477246208,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011937769198805104,
      "loss": 0.0108,
      "step": 9386
    },
    {
      "epoch": 4.381330221703617,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011936329642354755,
      "loss": 0.0062,
      "step": 9387
    },
    {
      "epoch": 4.381796966161027,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011934890044213085,
      "loss": 0.0197,
      "step": 9388
    },
    {
      "epoch": 4.382263710618436,
      "grad_norm": 0.4375,
      "learning_rate": 0.00011933450404411089,
      "loss": 0.0074,
      "step": 9389
    },
    {
      "epoch": 4.382730455075846,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011932010722979754,
      "loss": 0.01,
      "step": 9390
    },
    {
      "epoch": 4.383197199533256,
      "grad_norm": 0.2109375,
      "learning_rate": 0.00011930570999950085,
      "loss": 0.0024,
      "step": 9391
    },
    {
      "epoch": 4.3836639439906655,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00011929131235353084,
      "loss": 0.0103,
      "step": 9392
    },
    {
      "epoch": 4.384130688448074,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00011927691429219743,
      "loss": 0.0067,
      "step": 9393
    },
    {
      "epoch": 4.384597432905484,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011926251581581069,
      "loss": 0.0046,
      "step": 9394
    },
    {
      "epoch": 4.385064177362894,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011924811692468061,
      "loss": 0.0081,
      "step": 9395
    },
    {
      "epoch": 4.385530921820304,
      "grad_norm": 0.294921875,
      "learning_rate": 0.00011923371761911721,
      "loss": 0.0033,
      "step": 9396
    },
    {
      "epoch": 4.385997666277713,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011921931789943053,
      "loss": 0.0098,
      "step": 9397
    },
    {
      "epoch": 4.3864644107351225,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00011920491776593064,
      "loss": 0.0049,
      "step": 9398
    },
    {
      "epoch": 4.386931155192532,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00011919051721892753,
      "loss": 0.0099,
      "step": 9399
    },
    {
      "epoch": 4.387397899649942,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011917611625873128,
      "loss": 0.0095,
      "step": 9400
    },
    {
      "epoch": 4.387864644107351,
      "grad_norm": 0.44921875,
      "learning_rate": 0.000119161714885652,
      "loss": 0.0079,
      "step": 9401
    },
    {
      "epoch": 4.388331388564761,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011914731309999973,
      "loss": 0.0083,
      "step": 9402
    },
    {
      "epoch": 4.388798133022171,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00011913291090208457,
      "loss": 0.0055,
      "step": 9403
    },
    {
      "epoch": 4.3892648774795795,
      "grad_norm": 0.5078125,
      "learning_rate": 0.0001191185082922166,
      "loss": 0.0074,
      "step": 9404
    },
    {
      "epoch": 4.389731621936989,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011910410527070595,
      "loss": 0.0127,
      "step": 9405
    },
    {
      "epoch": 4.390198366394399,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00011908970183786267,
      "loss": 0.0124,
      "step": 9406
    },
    {
      "epoch": 4.390665110851809,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011907529799399697,
      "loss": 0.0081,
      "step": 9407
    },
    {
      "epoch": 4.391131855309218,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00011906089373941895,
      "loss": 0.0126,
      "step": 9408
    },
    {
      "epoch": 4.391598599766628,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011904648907443871,
      "loss": 0.0201,
      "step": 9409
    },
    {
      "epoch": 4.392065344224037,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00011903208399936644,
      "loss": 0.0062,
      "step": 9410
    },
    {
      "epoch": 4.392532088681447,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011901767851451229,
      "loss": 0.0129,
      "step": 9411
    },
    {
      "epoch": 4.392998833138856,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011900327262018642,
      "loss": 0.0143,
      "step": 9412
    },
    {
      "epoch": 4.393465577596266,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011898886631669901,
      "loss": 0.0087,
      "step": 9413
    },
    {
      "epoch": 4.393932322053676,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00011897445960436024,
      "loss": 0.0069,
      "step": 9414
    },
    {
      "epoch": 4.3943990665110855,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001189600524834803,
      "loss": 0.009,
      "step": 9415
    },
    {
      "epoch": 4.394865810968494,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00011894564495436938,
      "loss": 0.0075,
      "step": 9416
    },
    {
      "epoch": 4.395332555425904,
      "grad_norm": 0.220703125,
      "learning_rate": 0.00011893123701733773,
      "loss": 0.0036,
      "step": 9417
    },
    {
      "epoch": 4.395799299883314,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011891682867269551,
      "loss": 0.0083,
      "step": 9418
    },
    {
      "epoch": 4.396266044340724,
      "grad_norm": 0.271484375,
      "learning_rate": 0.00011890241992075299,
      "loss": 0.0044,
      "step": 9419
    },
    {
      "epoch": 4.396732788798133,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011888801076182041,
      "loss": 0.0131,
      "step": 9420
    },
    {
      "epoch": 4.3971995332555425,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011887360119620798,
      "loss": 0.007,
      "step": 9421
    },
    {
      "epoch": 4.397666277712952,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011885919122422602,
      "loss": 0.0054,
      "step": 9422
    },
    {
      "epoch": 4.398133022170362,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00011884478084618468,
      "loss": 0.0077,
      "step": 9423
    },
    {
      "epoch": 4.398599766627771,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00011883037006239434,
      "loss": 0.0096,
      "step": 9424
    },
    {
      "epoch": 4.399066511085181,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00011881595887316526,
      "loss": 0.0061,
      "step": 9425
    },
    {
      "epoch": 4.399533255542591,
      "grad_norm": 0.453125,
      "learning_rate": 0.00011880154727880766,
      "loss": 0.0107,
      "step": 9426
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011878713527963192,
      "loss": 0.0037,
      "step": 9427
    },
    {
      "epoch": 4.400466744457409,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011877272287594832,
      "loss": 0.008,
      "step": 9428
    },
    {
      "epoch": 4.400933488914819,
      "grad_norm": 0.3125,
      "learning_rate": 0.00011875831006806712,
      "loss": 0.0063,
      "step": 9429
    },
    {
      "epoch": 4.401400233372229,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011874389685629874,
      "loss": 0.0137,
      "step": 9430
    },
    {
      "epoch": 4.401866977829639,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011872948324095346,
      "loss": 0.0128,
      "step": 9431
    },
    {
      "epoch": 4.402333722287048,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011871506922234161,
      "loss": 0.0079,
      "step": 9432
    },
    {
      "epoch": 4.402800466744457,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011870065480077355,
      "loss": 0.0105,
      "step": 9433
    },
    {
      "epoch": 4.403267211201867,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011868623997655966,
      "loss": 0.0113,
      "step": 9434
    },
    {
      "epoch": 4.403733955659277,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00011867182475001026,
      "loss": 0.0045,
      "step": 9435
    },
    {
      "epoch": 4.404200700116686,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00011865740912143577,
      "loss": 0.0067,
      "step": 9436
    },
    {
      "epoch": 4.404667444574096,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011864299309114661,
      "loss": 0.0054,
      "step": 9437
    },
    {
      "epoch": 4.405134189031505,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00011862857665945309,
      "loss": 0.009,
      "step": 9438
    },
    {
      "epoch": 4.405600933488914,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00011861415982666561,
      "loss": 0.0023,
      "step": 9439
    },
    {
      "epoch": 4.406067677946324,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00011859974259309465,
      "loss": 0.0089,
      "step": 9440
    },
    {
      "epoch": 4.406534422403734,
      "grad_norm": 0.4296875,
      "learning_rate": 0.0001185853249590506,
      "loss": 0.006,
      "step": 9441
    },
    {
      "epoch": 4.407001166861144,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00011857090692484384,
      "loss": 0.0171,
      "step": 9442
    },
    {
      "epoch": 4.4074679113185535,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00011855648849078491,
      "loss": 0.0127,
      "step": 9443
    },
    {
      "epoch": 4.407934655775962,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00011854206965718417,
      "loss": 0.007,
      "step": 9444
    },
    {
      "epoch": 4.408401400233372,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00011852765042435207,
      "loss": 0.01,
      "step": 9445
    },
    {
      "epoch": 4.408868144690782,
      "grad_norm": 0.625,
      "learning_rate": 0.00011851323079259912,
      "loss": 0.0118,
      "step": 9446
    },
    {
      "epoch": 4.409334889148191,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00011849881076223579,
      "loss": 0.0073,
      "step": 9447
    },
    {
      "epoch": 4.409801633605601,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011848439033357252,
      "loss": 0.0065,
      "step": 9448
    },
    {
      "epoch": 4.4102683780630105,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011846996950691982,
      "loss": 0.0035,
      "step": 9449
    },
    {
      "epoch": 4.41073512252042,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00011845554828258818,
      "loss": 0.0049,
      "step": 9450
    },
    {
      "epoch": 4.411201866977829,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00011844112666088812,
      "loss": 0.0047,
      "step": 9451
    },
    {
      "epoch": 4.411668611435239,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011842670464213012,
      "loss": 0.0066,
      "step": 9452
    },
    {
      "epoch": 4.412135355892649,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00011841228222662475,
      "loss": 0.0066,
      "step": 9453
    },
    {
      "epoch": 4.412602100350059,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011839785941468251,
      "loss": 0.0133,
      "step": 9454
    },
    {
      "epoch": 4.4130688448074675,
      "grad_norm": 0.87890625,
      "learning_rate": 0.00011838343620661392,
      "loss": 0.0101,
      "step": 9455
    },
    {
      "epoch": 4.413535589264877,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011836901260272957,
      "loss": 0.0108,
      "step": 9456
    },
    {
      "epoch": 4.414002333722287,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011835458860334002,
      "loss": 0.0068,
      "step": 9457
    },
    {
      "epoch": 4.414469078179697,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00011834016420875577,
      "loss": 0.0091,
      "step": 9458
    },
    {
      "epoch": 4.414935822637106,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011832573941928747,
      "loss": 0.0088,
      "step": 9459
    },
    {
      "epoch": 4.415402567094516,
      "grad_norm": 0.26953125,
      "learning_rate": 0.00011831131423524564,
      "loss": 0.0052,
      "step": 9460
    },
    {
      "epoch": 4.415869311551925,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011829688865694093,
      "loss": 0.0119,
      "step": 9461
    },
    {
      "epoch": 4.416336056009335,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011828246268468388,
      "loss": 0.0052,
      "step": 9462
    },
    {
      "epoch": 4.416802800466744,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011826803631878514,
      "loss": 0.0107,
      "step": 9463
    },
    {
      "epoch": 4.417269544924154,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00011825360955955532,
      "loss": 0.0076,
      "step": 9464
    },
    {
      "epoch": 4.417736289381564,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00011823918240730498,
      "loss": 0.0044,
      "step": 9465
    },
    {
      "epoch": 4.4182030338389735,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00011822475486234486,
      "loss": 0.0039,
      "step": 9466
    },
    {
      "epoch": 4.418669778296382,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011821032692498553,
      "loss": 0.0145,
      "step": 9467
    },
    {
      "epoch": 4.419136522753792,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011819589859553764,
      "loss": 0.0092,
      "step": 9468
    },
    {
      "epoch": 4.419603267211202,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00011818146987431189,
      "loss": 0.005,
      "step": 9469
    },
    {
      "epoch": 4.420070011668612,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001181670407616189,
      "loss": 0.0107,
      "step": 9470
    },
    {
      "epoch": 4.420536756126021,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00011815261125776938,
      "loss": 0.0049,
      "step": 9471
    },
    {
      "epoch": 4.4210035005834305,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00011813818136307397,
      "loss": 0.0146,
      "step": 9472
    },
    {
      "epoch": 4.42147024504084,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00011812375107784344,
      "loss": 0.0094,
      "step": 9473
    },
    {
      "epoch": 4.42193698949825,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011810932040238842,
      "loss": 0.0108,
      "step": 9474
    },
    {
      "epoch": 4.422403733955659,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00011809488933701962,
      "loss": 0.0034,
      "step": 9475
    },
    {
      "epoch": 4.422870478413069,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001180804578820478,
      "loss": 0.0055,
      "step": 9476
    },
    {
      "epoch": 4.423337222870479,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011806602603778365,
      "loss": 0.0109,
      "step": 9477
    },
    {
      "epoch": 4.423803967327888,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001180515938045379,
      "loss": 0.013,
      "step": 9478
    },
    {
      "epoch": 4.424270711785297,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00011803716118262134,
      "loss": 0.0111,
      "step": 9479
    },
    {
      "epoch": 4.424737456242707,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00011802272817234468,
      "loss": 0.0078,
      "step": 9480
    },
    {
      "epoch": 4.425204200700117,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00011800829477401867,
      "loss": 0.0091,
      "step": 9481
    },
    {
      "epoch": 4.425670945157526,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011799386098795409,
      "loss": 0.0133,
      "step": 9482
    },
    {
      "epoch": 4.426137689614936,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011797942681446172,
      "loss": 0.0097,
      "step": 9483
    },
    {
      "epoch": 4.426604434072345,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011796499225385236,
      "loss": 0.0096,
      "step": 9484
    },
    {
      "epoch": 4.427071178529755,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00011795055730643678,
      "loss": 0.0059,
      "step": 9485
    },
    {
      "epoch": 4.427537922987165,
      "grad_norm": 0.484375,
      "learning_rate": 0.00011793612197252579,
      "loss": 0.0105,
      "step": 9486
    },
    {
      "epoch": 4.428004667444574,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011792168625243019,
      "loss": 0.0039,
      "step": 9487
    },
    {
      "epoch": 4.428471411901984,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011790725014646078,
      "loss": 0.0149,
      "step": 9488
    },
    {
      "epoch": 4.4289381563593935,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00011789281365492842,
      "loss": 0.0082,
      "step": 9489
    },
    {
      "epoch": 4.429404900816802,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00011787837677814393,
      "loss": 0.0064,
      "step": 9490
    },
    {
      "epoch": 4.429871645274212,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011786393951641815,
      "loss": 0.0147,
      "step": 9491
    },
    {
      "epoch": 4.430338389731622,
      "grad_norm": 0.2431640625,
      "learning_rate": 0.00011784950187006192,
      "loss": 0.0042,
      "step": 9492
    },
    {
      "epoch": 4.430805134189032,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011783506383938611,
      "loss": 0.0052,
      "step": 9493
    },
    {
      "epoch": 4.431271878646441,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001178206254247016,
      "loss": 0.0088,
      "step": 9494
    },
    {
      "epoch": 4.4317386231038505,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00011780618662631926,
      "loss": 0.0071,
      "step": 9495
    },
    {
      "epoch": 4.43220536756126,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011779174744454998,
      "loss": 0.0085,
      "step": 9496
    },
    {
      "epoch": 4.43267211201867,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011777730787970461,
      "loss": 0.0095,
      "step": 9497
    },
    {
      "epoch": 4.433138856476079,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011776286793209406,
      "loss": 0.0154,
      "step": 9498
    },
    {
      "epoch": 4.433605600933489,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011774842760202926,
      "loss": 0.013,
      "step": 9499
    },
    {
      "epoch": 4.434072345390899,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011773398688982116,
      "loss": 0.013,
      "step": 9500
    },
    {
      "epoch": 4.434539089848308,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011771954579578062,
      "loss": 0.0151,
      "step": 9501
    },
    {
      "epoch": 4.435005834305717,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001177051043202186,
      "loss": 0.0091,
      "step": 9502
    },
    {
      "epoch": 4.435472578763127,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00011769066246344604,
      "loss": 0.011,
      "step": 9503
    },
    {
      "epoch": 4.435939323220537,
      "grad_norm": 0.375,
      "learning_rate": 0.00011767622022577389,
      "loss": 0.0118,
      "step": 9504
    },
    {
      "epoch": 4.436406067677947,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011766177760751311,
      "loss": 0.0067,
      "step": 9505
    },
    {
      "epoch": 4.436872812135356,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00011764733460897466,
      "loss": 0.0103,
      "step": 9506
    },
    {
      "epoch": 4.437339556592765,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011763289123046953,
      "loss": 0.0097,
      "step": 9507
    },
    {
      "epoch": 4.437806301050175,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00011761844747230866,
      "loss": 0.0064,
      "step": 9508
    },
    {
      "epoch": 4.438273045507585,
      "grad_norm": 0.33203125,
      "learning_rate": 0.0001176040033348031,
      "loss": 0.0078,
      "step": 9509
    },
    {
      "epoch": 4.438739789964994,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011758955881826379,
      "loss": 0.0105,
      "step": 9510
    },
    {
      "epoch": 4.439206534422404,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00011757511392300178,
      "loss": 0.0082,
      "step": 9511
    },
    {
      "epoch": 4.4396732788798134,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00011756066864932806,
      "loss": 0.0153,
      "step": 9512
    },
    {
      "epoch": 4.440140023337223,
      "grad_norm": 0.8671875,
      "learning_rate": 0.00011754622299755367,
      "loss": 0.012,
      "step": 9513
    },
    {
      "epoch": 4.440606767794632,
      "grad_norm": 0.2255859375,
      "learning_rate": 0.0001175317769679896,
      "loss": 0.0041,
      "step": 9514
    },
    {
      "epoch": 4.441073512252042,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00011751733056094693,
      "loss": 0.0089,
      "step": 9515
    },
    {
      "epoch": 4.441540256709452,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00011750288377673673,
      "loss": 0.007,
      "step": 9516
    },
    {
      "epoch": 4.4420070011668615,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011748843661566997,
      "loss": 0.0125,
      "step": 9517
    },
    {
      "epoch": 4.44247374562427,
      "grad_norm": 0.494140625,
      "learning_rate": 0.0001174739890780578,
      "loss": 0.0111,
      "step": 9518
    },
    {
      "epoch": 4.44294049008168,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011745954116421125,
      "loss": 0.0063,
      "step": 9519
    },
    {
      "epoch": 4.44340723453909,
      "grad_norm": 1.140625,
      "learning_rate": 0.00011744509287444141,
      "loss": 0.0316,
      "step": 9520
    },
    {
      "epoch": 4.4438739789965,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00011743064420905935,
      "loss": 0.0063,
      "step": 9521
    },
    {
      "epoch": 4.444340723453909,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00011741619516837621,
      "loss": 0.0084,
      "step": 9522
    },
    {
      "epoch": 4.4448074679113185,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011740174575270306,
      "loss": 0.0113,
      "step": 9523
    },
    {
      "epoch": 4.445274212368728,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00011738729596235099,
      "loss": 0.0061,
      "step": 9524
    },
    {
      "epoch": 4.445740956826137,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011737284579763117,
      "loss": 0.0096,
      "step": 9525
    },
    {
      "epoch": 4.446207701283547,
      "grad_norm": 0.4375,
      "learning_rate": 0.00011735839525885471,
      "loss": 0.009,
      "step": 9526
    },
    {
      "epoch": 4.446674445740957,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011734394434633272,
      "loss": 0.0065,
      "step": 9527
    },
    {
      "epoch": 4.447141190198367,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00011732949306037636,
      "loss": 0.0158,
      "step": 9528
    },
    {
      "epoch": 4.447607934655776,
      "grad_norm": 0.46875,
      "learning_rate": 0.00011731504140129682,
      "loss": 0.0077,
      "step": 9529
    },
    {
      "epoch": 4.448074679113185,
      "grad_norm": 0.68359375,
      "learning_rate": 0.0001173005893694052,
      "loss": 0.0098,
      "step": 9530
    },
    {
      "epoch": 4.448541423570595,
      "grad_norm": 0.33984375,
      "learning_rate": 0.00011728613696501267,
      "loss": 0.0061,
      "step": 9531
    },
    {
      "epoch": 4.449008168028005,
      "grad_norm": 0.484375,
      "learning_rate": 0.00011727168418843049,
      "loss": 0.0095,
      "step": 9532
    },
    {
      "epoch": 4.449474912485414,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00011725723103996975,
      "loss": 0.0069,
      "step": 9533
    },
    {
      "epoch": 4.449941656942824,
      "grad_norm": 0.61328125,
      "learning_rate": 0.0001172427775199417,
      "loss": 0.0105,
      "step": 9534
    },
    {
      "epoch": 4.450408401400233,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00011722832362865749,
      "loss": 0.0051,
      "step": 9535
    },
    {
      "epoch": 4.450875145857643,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011721386936642839,
      "loss": 0.0112,
      "step": 9536
    },
    {
      "epoch": 4.451341890315052,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00011719941473356557,
      "loss": 0.0085,
      "step": 9537
    },
    {
      "epoch": 4.451808634772462,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00011718495973038026,
      "loss": 0.0118,
      "step": 9538
    },
    {
      "epoch": 4.452275379229872,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011717050435718371,
      "loss": 0.0074,
      "step": 9539
    },
    {
      "epoch": 4.4527421236872815,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011715604861428713,
      "loss": 0.0074,
      "step": 9540
    },
    {
      "epoch": 4.45320886814469,
      "grad_norm": 0.484375,
      "learning_rate": 0.0001171415925020018,
      "loss": 0.0142,
      "step": 9541
    },
    {
      "epoch": 4.4536756126021,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00011712713602063898,
      "loss": 0.007,
      "step": 9542
    },
    {
      "epoch": 4.45414235705951,
      "grad_norm": 0.384765625,
      "learning_rate": 0.0001171126791705099,
      "loss": 0.0067,
      "step": 9543
    },
    {
      "epoch": 4.45460910151692,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011709822195192586,
      "loss": 0.0114,
      "step": 9544
    },
    {
      "epoch": 4.455075845974329,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00011708376436519812,
      "loss": 0.0087,
      "step": 9545
    },
    {
      "epoch": 4.4555425904317385,
      "grad_norm": 0.494140625,
      "learning_rate": 0.000117069306410638,
      "loss": 0.0072,
      "step": 9546
    },
    {
      "epoch": 4.456009334889148,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011705484808855676,
      "loss": 0.0109,
      "step": 9547
    },
    {
      "epoch": 4.456476079346558,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00011704038939926571,
      "loss": 0.0074,
      "step": 9548
    },
    {
      "epoch": 4.456942823803967,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011702593034307619,
      "loss": 0.0186,
      "step": 9549
    },
    {
      "epoch": 4.457409568261377,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00011701147092029947,
      "loss": 0.0054,
      "step": 9550
    },
    {
      "epoch": 4.457876312718787,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011699701113124693,
      "loss": 0.0104,
      "step": 9551
    },
    {
      "epoch": 4.458343057176196,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011698255097622988,
      "loss": 0.0099,
      "step": 9552
    },
    {
      "epoch": 4.458809801633605,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011696809045555965,
      "loss": 0.0113,
      "step": 9553
    },
    {
      "epoch": 4.459276546091015,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011695362956954761,
      "loss": 0.0108,
      "step": 9554
    },
    {
      "epoch": 4.459743290548425,
      "grad_norm": 0.859375,
      "learning_rate": 0.00011693916831850513,
      "loss": 0.0131,
      "step": 9555
    },
    {
      "epoch": 4.460210035005835,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011692470670274356,
      "loss": 0.008,
      "step": 9556
    },
    {
      "epoch": 4.460676779463244,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011691024472257428,
      "loss": 0.0081,
      "step": 9557
    },
    {
      "epoch": 4.461143523920653,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011689578237830868,
      "loss": 0.0141,
      "step": 9558
    },
    {
      "epoch": 4.461610268378063,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00011688131967025809,
      "loss": 0.0074,
      "step": 9559
    },
    {
      "epoch": 4.462077012835473,
      "grad_norm": 0.271484375,
      "learning_rate": 0.000116866856598734,
      "loss": 0.0035,
      "step": 9560
    },
    {
      "epoch": 4.462543757292882,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00011685239316404774,
      "loss": 0.0084,
      "step": 9561
    },
    {
      "epoch": 4.463010501750292,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00011683792936651078,
      "loss": 0.0045,
      "step": 9562
    },
    {
      "epoch": 4.4634772462077015,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00011682346520643451,
      "loss": 0.0077,
      "step": 9563
    },
    {
      "epoch": 4.463943990665111,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011680900068413039,
      "loss": 0.0073,
      "step": 9564
    },
    {
      "epoch": 4.46441073512252,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00011679453579990982,
      "loss": 0.009,
      "step": 9565
    },
    {
      "epoch": 4.46487747957993,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00011678007055408424,
      "loss": 0.0029,
      "step": 9566
    },
    {
      "epoch": 4.46534422403734,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00011676560494696513,
      "loss": 0.0078,
      "step": 9567
    },
    {
      "epoch": 4.465810968494749,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00011675113897886399,
      "loss": 0.0095,
      "step": 9568
    },
    {
      "epoch": 4.4662777129521585,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011673667265009218,
      "loss": 0.0108,
      "step": 9569
    },
    {
      "epoch": 4.466744457409568,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00011672220596096126,
      "loss": 0.0062,
      "step": 9570
    },
    {
      "epoch": 4.467211201866978,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001167077389117827,
      "loss": 0.0182,
      "step": 9571
    },
    {
      "epoch": 4.467677946324388,
      "grad_norm": 1.3828125,
      "learning_rate": 0.000116693271502868,
      "loss": 0.0293,
      "step": 9572
    },
    {
      "epoch": 4.468144690781797,
      "grad_norm": 0.7734375,
      "learning_rate": 0.0001166788037345286,
      "loss": 0.0156,
      "step": 9573
    },
    {
      "epoch": 4.468611435239207,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00011666433560707609,
      "loss": 0.0038,
      "step": 9574
    },
    {
      "epoch": 4.469078179696616,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011664986712082194,
      "loss": 0.0077,
      "step": 9575
    },
    {
      "epoch": 4.469544924154025,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00011663539827607763,
      "loss": 0.0067,
      "step": 9576
    },
    {
      "epoch": 4.470011668611435,
      "grad_norm": 0.3671875,
      "learning_rate": 0.00011662092907315479,
      "loss": 0.0097,
      "step": 9577
    },
    {
      "epoch": 4.470478413068845,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011660645951236491,
      "loss": 0.0094,
      "step": 9578
    },
    {
      "epoch": 4.470945157526255,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011659198959401949,
      "loss": 0.0141,
      "step": 9579
    },
    {
      "epoch": 4.471411901983664,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011657751931843017,
      "loss": 0.0096,
      "step": 9580
    },
    {
      "epoch": 4.471878646441073,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00011656304868590844,
      "loss": 0.014,
      "step": 9581
    },
    {
      "epoch": 4.472345390898483,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011654857769676588,
      "loss": 0.0123,
      "step": 9582
    },
    {
      "epoch": 4.472812135355893,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011653410635131413,
      "loss": 0.0109,
      "step": 9583
    },
    {
      "epoch": 4.473278879813302,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00011651963464986468,
      "loss": 0.0062,
      "step": 9584
    },
    {
      "epoch": 4.473745624270712,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011650516259272917,
      "loss": 0.0172,
      "step": 9585
    },
    {
      "epoch": 4.4742123687281214,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011649069018021921,
      "loss": 0.021,
      "step": 9586
    },
    {
      "epoch": 4.474679113185531,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001164762174126464,
      "loss": 0.007,
      "step": 9587
    },
    {
      "epoch": 4.47514585764294,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011646174429032235,
      "loss": 0.0077,
      "step": 9588
    },
    {
      "epoch": 4.47561260210035,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00011644727081355866,
      "loss": 0.0061,
      "step": 9589
    },
    {
      "epoch": 4.47607934655776,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011643279698266698,
      "loss": 0.0141,
      "step": 9590
    },
    {
      "epoch": 4.4765460910151695,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011641832279795897,
      "loss": 0.015,
      "step": 9591
    },
    {
      "epoch": 4.4770128354725784,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011640384825974623,
      "loss": 0.0085,
      "step": 9592
    },
    {
      "epoch": 4.477479579929988,
      "grad_norm": 0.40625,
      "learning_rate": 0.00011638937336834046,
      "loss": 0.0085,
      "step": 9593
    },
    {
      "epoch": 4.477946324387398,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00011637489812405328,
      "loss": 0.0057,
      "step": 9594
    },
    {
      "epoch": 4.478413068844808,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00011636042252719641,
      "loss": 0.0063,
      "step": 9595
    },
    {
      "epoch": 4.478879813302217,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00011634594657808145,
      "loss": 0.0055,
      "step": 9596
    },
    {
      "epoch": 4.4793465577596265,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011633147027702013,
      "loss": 0.0073,
      "step": 9597
    },
    {
      "epoch": 4.479813302217036,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00011631699362432415,
      "loss": 0.0042,
      "step": 9598
    },
    {
      "epoch": 4.480280046674446,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011630251662030518,
      "loss": 0.01,
      "step": 9599
    },
    {
      "epoch": 4.480746791131855,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011628803926527495,
      "loss": 0.0064,
      "step": 9600
    },
    {
      "epoch": 4.481213535589265,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011627356155954516,
      "loss": 0.0093,
      "step": 9601
    },
    {
      "epoch": 4.481680280046675,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001162590835034275,
      "loss": 0.0109,
      "step": 9602
    },
    {
      "epoch": 4.482147024504084,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011624460509723379,
      "loss": 0.0074,
      "step": 9603
    },
    {
      "epoch": 4.482613768961493,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001162301263412757,
      "loss": 0.0084,
      "step": 9604
    },
    {
      "epoch": 4.483080513418903,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011621564723586496,
      "loss": 0.0106,
      "step": 9605
    },
    {
      "epoch": 4.483547257876313,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00011620116778131333,
      "loss": 0.0065,
      "step": 9606
    },
    {
      "epoch": 4.484014002333723,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00011618668797793261,
      "loss": 0.0115,
      "step": 9607
    },
    {
      "epoch": 4.484480746791132,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011617220782603452,
      "loss": 0.0057,
      "step": 9608
    },
    {
      "epoch": 4.484947491248541,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00011615772732593087,
      "loss": 0.011,
      "step": 9609
    },
    {
      "epoch": 4.485414235705951,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00011614324647793342,
      "loss": 0.0094,
      "step": 9610
    },
    {
      "epoch": 4.48588098016336,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00011612876528235396,
      "loss": 0.0077,
      "step": 9611
    },
    {
      "epoch": 4.48634772462077,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011611428373950426,
      "loss": 0.0137,
      "step": 9612
    },
    {
      "epoch": 4.48681446907818,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001160998018496962,
      "loss": 0.0117,
      "step": 9613
    },
    {
      "epoch": 4.4872812135355895,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011608531961324152,
      "loss": 0.0095,
      "step": 9614
    },
    {
      "epoch": 4.487747957992999,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00011607083703045205,
      "loss": 0.0052,
      "step": 9615
    },
    {
      "epoch": 4.488214702450408,
      "grad_norm": 0.84375,
      "learning_rate": 0.00011605635410163964,
      "loss": 0.012,
      "step": 9616
    },
    {
      "epoch": 4.488681446907818,
      "grad_norm": 0.484375,
      "learning_rate": 0.0001160418708271161,
      "loss": 0.0053,
      "step": 9617
    },
    {
      "epoch": 4.489148191365228,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011602738720719329,
      "loss": 0.0161,
      "step": 9618
    },
    {
      "epoch": 4.489614935822637,
      "grad_norm": 0.365234375,
      "learning_rate": 0.00011601290324218306,
      "loss": 0.008,
      "step": 9619
    },
    {
      "epoch": 4.4900816802800465,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00011599841893239725,
      "loss": 0.0093,
      "step": 9620
    },
    {
      "epoch": 4.490548424737456,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00011598393427814773,
      "loss": 0.0101,
      "step": 9621
    },
    {
      "epoch": 4.491015169194866,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00011596944927974635,
      "loss": 0.0199,
      "step": 9622
    },
    {
      "epoch": 4.491481913652275,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011595496393750505,
      "loss": 0.0092,
      "step": 9623
    },
    {
      "epoch": 4.491948658109685,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00011594047825173566,
      "loss": 0.0061,
      "step": 9624
    },
    {
      "epoch": 4.492415402567095,
      "grad_norm": 0.5,
      "learning_rate": 0.00011592599222275007,
      "loss": 0.0098,
      "step": 9625
    },
    {
      "epoch": 4.492882147024504,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011591150585086022,
      "loss": 0.0154,
      "step": 9626
    },
    {
      "epoch": 4.493348891481913,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011589701913637798,
      "loss": 0.0116,
      "step": 9627
    },
    {
      "epoch": 4.493815635939323,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00011588253207961528,
      "loss": 0.0076,
      "step": 9628
    },
    {
      "epoch": 4.494282380396733,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00011586804468088406,
      "loss": 0.0119,
      "step": 9629
    },
    {
      "epoch": 4.494749124854143,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011585355694049622,
      "loss": 0.0117,
      "step": 9630
    },
    {
      "epoch": 4.495215869311552,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00011583906885876368,
      "loss": 0.0076,
      "step": 9631
    },
    {
      "epoch": 4.495682613768961,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00011582458043599844,
      "loss": 0.0042,
      "step": 9632
    },
    {
      "epoch": 4.496149358226371,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011581009167251244,
      "loss": 0.011,
      "step": 9633
    },
    {
      "epoch": 4.496616102683781,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00011579560256861759,
      "loss": 0.0111,
      "step": 9634
    },
    {
      "epoch": 4.49708284714119,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00011578111312462591,
      "loss": 0.0144,
      "step": 9635
    },
    {
      "epoch": 4.4975495915986,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00011576662334084937,
      "loss": 0.0121,
      "step": 9636
    },
    {
      "epoch": 4.4980163360560095,
      "grad_norm": 0.6875,
      "learning_rate": 0.00011575213321759993,
      "loss": 0.0112,
      "step": 9637
    },
    {
      "epoch": 4.498483080513419,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00011573764275518957,
      "loss": 0.0068,
      "step": 9638
    },
    {
      "epoch": 4.498949824970828,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001157231519539303,
      "loss": 0.007,
      "step": 9639
    },
    {
      "epoch": 4.499416569428238,
      "grad_norm": 1.15625,
      "learning_rate": 0.00011570866081413414,
      "loss": 0.0102,
      "step": 9640
    },
    {
      "epoch": 4.499883313885648,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00011569416933611305,
      "loss": 0.011,
      "step": 9641
    },
    {
      "epoch": 4.500350058343058,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00011567967752017912,
      "loss": 0.0085,
      "step": 9642
    },
    {
      "epoch": 4.5008168028004665,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011566518536664433,
      "loss": 0.0071,
      "step": 9643
    },
    {
      "epoch": 4.501283547257876,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00011565069287582069,
      "loss": 0.007,
      "step": 9644
    },
    {
      "epoch": 4.501750291715286,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011563620004802029,
      "loss": 0.0118,
      "step": 9645
    },
    {
      "epoch": 4.502217036172695,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011562170688355516,
      "loss": 0.0155,
      "step": 9646
    },
    {
      "epoch": 4.502683780630105,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011560721338273735,
      "loss": 0.009,
      "step": 9647
    },
    {
      "epoch": 4.503150525087515,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001155927195458789,
      "loss": 0.0112,
      "step": 9648
    },
    {
      "epoch": 4.503150525087515,
      "eval_loss": 1.984728455543518,
      "eval_runtime": 58.7541,
      "eval_samples_per_second": 30.704,
      "eval_steps_per_second": 3.847,
      "step": 9648
    },
    {
      "epoch": 4.503617269544924,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001155782253732919,
      "loss": 0.0062,
      "step": 9649
    },
    {
      "epoch": 4.504084014002334,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011556373086528845,
      "loss": 0.0122,
      "step": 9650
    },
    {
      "epoch": 4.504550758459743,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00011554923602218055,
      "loss": 0.0105,
      "step": 9651
    },
    {
      "epoch": 4.505017502917153,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011553474084428042,
      "loss": 0.0081,
      "step": 9652
    },
    {
      "epoch": 4.505484247374563,
      "grad_norm": 0.578125,
      "learning_rate": 0.00011552024533190005,
      "loss": 0.0088,
      "step": 9653
    },
    {
      "epoch": 4.505950991831972,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00011550574948535156,
      "loss": 0.0068,
      "step": 9654
    },
    {
      "epoch": 4.506417736289381,
      "grad_norm": 0.310546875,
      "learning_rate": 0.0001154912533049471,
      "loss": 0.0085,
      "step": 9655
    },
    {
      "epoch": 4.506884480746791,
      "grad_norm": 0.88671875,
      "learning_rate": 0.00011547675679099877,
      "loss": 0.0149,
      "step": 9656
    },
    {
      "epoch": 4.507351225204201,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00011546225994381867,
      "loss": 0.005,
      "step": 9657
    },
    {
      "epoch": 4.507817969661611,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00011544776276371898,
      "loss": 0.0057,
      "step": 9658
    },
    {
      "epoch": 4.50828471411902,
      "grad_norm": 0.84765625,
      "learning_rate": 0.00011543326525101182,
      "loss": 0.0252,
      "step": 9659
    },
    {
      "epoch": 4.5087514585764294,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011541876740600932,
      "loss": 0.0134,
      "step": 9660
    },
    {
      "epoch": 4.509218203033839,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00011540426922902366,
      "loss": 0.0151,
      "step": 9661
    },
    {
      "epoch": 4.509684947491248,
      "grad_norm": 0.4375,
      "learning_rate": 0.00011538977072036699,
      "loss": 0.0108,
      "step": 9662
    },
    {
      "epoch": 4.510151691948658,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00011537527188035148,
      "loss": 0.0055,
      "step": 9663
    },
    {
      "epoch": 4.510618436406068,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00011536077270928932,
      "loss": 0.0093,
      "step": 9664
    },
    {
      "epoch": 4.5110851808634775,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011534627320749264,
      "loss": 0.0206,
      "step": 9665
    },
    {
      "epoch": 4.5115519253208864,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011533177337527372,
      "loss": 0.0096,
      "step": 9666
    },
    {
      "epoch": 4.512018669778296,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00011531727321294468,
      "loss": 0.0048,
      "step": 9667
    },
    {
      "epoch": 4.512485414235706,
      "grad_norm": 0.625,
      "learning_rate": 0.00011530277272081775,
      "loss": 0.0117,
      "step": 9668
    },
    {
      "epoch": 4.512952158693116,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00011528827189920518,
      "loss": 0.0141,
      "step": 9669
    },
    {
      "epoch": 4.513418903150525,
      "grad_norm": 0.328125,
      "learning_rate": 0.00011527377074841914,
      "loss": 0.0055,
      "step": 9670
    },
    {
      "epoch": 4.5138856476079345,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00011525926926877185,
      "loss": 0.0078,
      "step": 9671
    },
    {
      "epoch": 4.514352392065344,
      "grad_norm": 0.796875,
      "learning_rate": 0.00011524476746057556,
      "loss": 0.012,
      "step": 9672
    },
    {
      "epoch": 4.514819136522754,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011523026532414252,
      "loss": 0.0128,
      "step": 9673
    },
    {
      "epoch": 4.515285880980163,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011521576285978497,
      "loss": 0.0184,
      "step": 9674
    },
    {
      "epoch": 4.515752625437573,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00011520126006781518,
      "loss": 0.0065,
      "step": 9675
    },
    {
      "epoch": 4.516219369894983,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001151867569485454,
      "loss": 0.0213,
      "step": 9676
    },
    {
      "epoch": 4.516686114352392,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00011517225350228785,
      "loss": 0.0096,
      "step": 9677
    },
    {
      "epoch": 4.517152858809801,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00011515774972935487,
      "loss": 0.0103,
      "step": 9678
    },
    {
      "epoch": 4.517619603267211,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011514324563005874,
      "loss": 0.0134,
      "step": 9679
    },
    {
      "epoch": 4.518086347724621,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00011512874120471169,
      "loss": 0.0058,
      "step": 9680
    },
    {
      "epoch": 4.518553092182031,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011511423645362608,
      "loss": 0.0111,
      "step": 9681
    },
    {
      "epoch": 4.51901983663944,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001150997313771142,
      "loss": 0.0093,
      "step": 9682
    },
    {
      "epoch": 4.519486581096849,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00011508522597548832,
      "loss": 0.0069,
      "step": 9683
    },
    {
      "epoch": 4.519953325554259,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011507072024906083,
      "loss": 0.0118,
      "step": 9684
    },
    {
      "epoch": 4.520420070011669,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00011505621419814398,
      "loss": 0.0078,
      "step": 9685
    },
    {
      "epoch": 4.520886814469078,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011504170782305014,
      "loss": 0.0104,
      "step": 9686
    },
    {
      "epoch": 4.521353558926488,
      "grad_norm": 0.546875,
      "learning_rate": 0.00011502720112409165,
      "loss": 0.011,
      "step": 9687
    },
    {
      "epoch": 4.5218203033838975,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011501269410158082,
      "loss": 0.0185,
      "step": 9688
    },
    {
      "epoch": 4.522287047841306,
      "grad_norm": 0.484375,
      "learning_rate": 0.00011499818675583005,
      "loss": 0.0096,
      "step": 9689
    },
    {
      "epoch": 4.522753792298716,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00011498367908715166,
      "loss": 0.0096,
      "step": 9690
    },
    {
      "epoch": 4.523220536756126,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00011496917109585808,
      "loss": 0.0074,
      "step": 9691
    },
    {
      "epoch": 4.523687281213536,
      "grad_norm": 0.5,
      "learning_rate": 0.0001149546627822616,
      "loss": 0.0078,
      "step": 9692
    },
    {
      "epoch": 4.524154025670946,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00011494015414667463,
      "loss": 0.0083,
      "step": 9693
    },
    {
      "epoch": 4.5246207701283545,
      "grad_norm": 1.015625,
      "learning_rate": 0.00011492564518940958,
      "loss": 0.0155,
      "step": 9694
    },
    {
      "epoch": 4.525087514585764,
      "grad_norm": 0.546875,
      "learning_rate": 0.00011491113591077885,
      "loss": 0.0081,
      "step": 9695
    },
    {
      "epoch": 4.525554259043174,
      "grad_norm": 0.3515625,
      "learning_rate": 0.00011489662631109479,
      "loss": 0.006,
      "step": 9696
    },
    {
      "epoch": 4.526021003500583,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00011488211639066984,
      "loss": 0.0062,
      "step": 9697
    },
    {
      "epoch": 4.526487747957993,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00011486760614981643,
      "loss": 0.0213,
      "step": 9698
    },
    {
      "epoch": 4.526954492415403,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00011485309558884696,
      "loss": 0.0065,
      "step": 9699
    },
    {
      "epoch": 4.527421236872812,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011483858470807384,
      "loss": 0.0086,
      "step": 9700
    },
    {
      "epoch": 4.527887981330222,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00011482407350780958,
      "loss": 0.0108,
      "step": 9701
    },
    {
      "epoch": 4.528354725787631,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00011480956198836652,
      "loss": 0.0066,
      "step": 9702
    },
    {
      "epoch": 4.528821470245041,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00011479505015005716,
      "loss": 0.0167,
      "step": 9703
    },
    {
      "epoch": 4.529288214702451,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00011478053799319398,
      "loss": 0.0132,
      "step": 9704
    },
    {
      "epoch": 4.52975495915986,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001147660255180894,
      "loss": 0.0096,
      "step": 9705
    },
    {
      "epoch": 4.530221703617269,
      "grad_norm": 0.451171875,
      "learning_rate": 0.0001147515127250559,
      "loss": 0.0079,
      "step": 9706
    },
    {
      "epoch": 4.530688448074679,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00011473699961440599,
      "loss": 0.0124,
      "step": 9707
    },
    {
      "epoch": 4.531155192532089,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00011472248618645212,
      "loss": 0.0113,
      "step": 9708
    },
    {
      "epoch": 4.531621936989498,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011470797244150675,
      "loss": 0.0084,
      "step": 9709
    },
    {
      "epoch": 4.532088681446908,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00011469345837988245,
      "loss": 0.0079,
      "step": 9710
    },
    {
      "epoch": 4.5325554259043175,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001146789440018917,
      "loss": 0.0096,
      "step": 9711
    },
    {
      "epoch": 4.533022170361727,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00011466442930784697,
      "loss": 0.0113,
      "step": 9712
    },
    {
      "epoch": 4.533488914819136,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001146499142980608,
      "loss": 0.0141,
      "step": 9713
    },
    {
      "epoch": 4.533955659276546,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011463539897284574,
      "loss": 0.0129,
      "step": 9714
    },
    {
      "epoch": 4.534422403733956,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00011462088333251432,
      "loss": 0.0105,
      "step": 9715
    },
    {
      "epoch": 4.534889148191366,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00011460636737737903,
      "loss": 0.0078,
      "step": 9716
    },
    {
      "epoch": 4.5353558926487745,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00011459185110775246,
      "loss": 0.0059,
      "step": 9717
    },
    {
      "epoch": 4.535822637106184,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00011457733452394714,
      "loss": 0.01,
      "step": 9718
    },
    {
      "epoch": 4.536289381563594,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011456281762627565,
      "loss": 0.0105,
      "step": 9719
    },
    {
      "epoch": 4.536756126021004,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011454830041505049,
      "loss": 0.0101,
      "step": 9720
    },
    {
      "epoch": 4.537222870478413,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011453378289058433,
      "loss": 0.0077,
      "step": 9721
    },
    {
      "epoch": 4.537689614935823,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00011451926505318969,
      "loss": 0.007,
      "step": 9722
    },
    {
      "epoch": 4.538156359393232,
      "grad_norm": 0.7265625,
      "learning_rate": 0.00011450474690317914,
      "loss": 0.0172,
      "step": 9723
    },
    {
      "epoch": 4.538623103850642,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011449022844086534,
      "loss": 0.0118,
      "step": 9724
    },
    {
      "epoch": 4.539089848308051,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011447570966656083,
      "loss": 0.0135,
      "step": 9725
    },
    {
      "epoch": 4.539556592765461,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001144611905805782,
      "loss": 0.0103,
      "step": 9726
    },
    {
      "epoch": 4.540023337222871,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011444667118323011,
      "loss": 0.0116,
      "step": 9727
    },
    {
      "epoch": 4.5404900816802805,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011443215147482917,
      "loss": 0.0088,
      "step": 9728
    },
    {
      "epoch": 4.540956826137689,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011441763145568796,
      "loss": 0.0082,
      "step": 9729
    },
    {
      "epoch": 4.541423570595099,
      "grad_norm": 0.4609375,
      "learning_rate": 0.0001144031111261192,
      "loss": 0.0065,
      "step": 9730
    },
    {
      "epoch": 4.541890315052509,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011438859048643544,
      "loss": 0.0094,
      "step": 9731
    },
    {
      "epoch": 4.542357059509918,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00011437406953694936,
      "loss": 0.0072,
      "step": 9732
    },
    {
      "epoch": 4.542823803967328,
      "grad_norm": 0.453125,
      "learning_rate": 0.00011435954827797363,
      "loss": 0.006,
      "step": 9733
    },
    {
      "epoch": 4.5432905484247375,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001143450267098209,
      "loss": 0.0119,
      "step": 9734
    },
    {
      "epoch": 4.543757292882147,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00011433050483280383,
      "loss": 0.0111,
      "step": 9735
    },
    {
      "epoch": 4.544224037339557,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011431598264723506,
      "loss": 0.0048,
      "step": 9736
    },
    {
      "epoch": 4.544690781796966,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011430146015342733,
      "loss": 0.0089,
      "step": 9737
    },
    {
      "epoch": 4.545157526254376,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00011428693735169329,
      "loss": 0.0101,
      "step": 9738
    },
    {
      "epoch": 4.5456242707117855,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001142724142423456,
      "loss": 0.0182,
      "step": 9739
    },
    {
      "epoch": 4.5460910151691944,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00011425789082569705,
      "loss": 0.0081,
      "step": 9740
    },
    {
      "epoch": 4.546557759626604,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00011424336710206027,
      "loss": 0.0105,
      "step": 9741
    },
    {
      "epoch": 4.547024504084014,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011422884307174797,
      "loss": 0.0121,
      "step": 9742
    },
    {
      "epoch": 4.547491248541424,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011421431873507294,
      "loss": 0.0124,
      "step": 9743
    },
    {
      "epoch": 4.547957992998834,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00011419979409234782,
      "loss": 0.006,
      "step": 9744
    },
    {
      "epoch": 4.5484247374562425,
      "grad_norm": 0.515625,
      "learning_rate": 0.0001141852691438854,
      "loss": 0.0084,
      "step": 9745
    },
    {
      "epoch": 4.548891481913652,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011417074388999837,
      "loss": 0.013,
      "step": 9746
    },
    {
      "epoch": 4.549358226371062,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011415621833099952,
      "loss": 0.014,
      "step": 9747
    },
    {
      "epoch": 4.549824970828471,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011414169246720158,
      "loss": 0.0112,
      "step": 9748
    },
    {
      "epoch": 4.550291715285881,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00011412716629891733,
      "loss": 0.0086,
      "step": 9749
    },
    {
      "epoch": 4.550758459743291,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001141126398264595,
      "loss": 0.009,
      "step": 9750
    },
    {
      "epoch": 4.5512252042007,
      "grad_norm": 0.453125,
      "learning_rate": 0.00011409811305014088,
      "loss": 0.0052,
      "step": 9751
    },
    {
      "epoch": 4.551691948658109,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011408358597027421,
      "loss": 0.0086,
      "step": 9752
    },
    {
      "epoch": 4.552158693115519,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011406905858717234,
      "loss": 0.0105,
      "step": 9753
    },
    {
      "epoch": 4.552625437572929,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011405453090114804,
      "loss": 0.0126,
      "step": 9754
    },
    {
      "epoch": 4.553092182030339,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011404000291251406,
      "loss": 0.0117,
      "step": 9755
    },
    {
      "epoch": 4.553558926487748,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00011402547462158327,
      "loss": 0.0067,
      "step": 9756
    },
    {
      "epoch": 4.554025670945157,
      "grad_norm": 0.625,
      "learning_rate": 0.00011401094602866844,
      "loss": 0.0101,
      "step": 9757
    },
    {
      "epoch": 4.554492415402567,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011399641713408237,
      "loss": 0.009,
      "step": 9758
    },
    {
      "epoch": 4.554959159859977,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011398188793813795,
      "loss": 0.0043,
      "step": 9759
    },
    {
      "epoch": 4.555425904317386,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011396735844114797,
      "loss": 0.0193,
      "step": 9760
    },
    {
      "epoch": 4.555892648774796,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00011395282864342525,
      "loss": 0.0125,
      "step": 9761
    },
    {
      "epoch": 4.5563593932322055,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011393829854528262,
      "loss": 0.0086,
      "step": 9762
    },
    {
      "epoch": 4.556826137689615,
      "grad_norm": 0.62890625,
      "learning_rate": 0.000113923768147033,
      "loss": 0.0139,
      "step": 9763
    },
    {
      "epoch": 4.557292882147024,
      "grad_norm": 0.703125,
      "learning_rate": 0.00011390923744898918,
      "loss": 0.0116,
      "step": 9764
    },
    {
      "epoch": 4.557759626604434,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011389470645146405,
      "loss": 0.0114,
      "step": 9765
    },
    {
      "epoch": 4.558226371061844,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011388017515477048,
      "loss": 0.009,
      "step": 9766
    },
    {
      "epoch": 4.558693115519253,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00011386564355922133,
      "loss": 0.0189,
      "step": 9767
    },
    {
      "epoch": 4.5591598599766625,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011385111166512951,
      "loss": 0.0082,
      "step": 9768
    },
    {
      "epoch": 4.559626604434072,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011383657947280786,
      "loss": 0.0073,
      "step": 9769
    },
    {
      "epoch": 4.560093348891482,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00011382204698256933,
      "loss": 0.0078,
      "step": 9770
    },
    {
      "epoch": 4.560560093348892,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00011380751419472677,
      "loss": 0.0106,
      "step": 9771
    },
    {
      "epoch": 4.561026837806301,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00011379298110959313,
      "loss": 0.0104,
      "step": 9772
    },
    {
      "epoch": 4.561493582263711,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00011377844772748129,
      "loss": 0.0144,
      "step": 9773
    },
    {
      "epoch": 4.56196032672112,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00011376391404870421,
      "loss": 0.0095,
      "step": 9774
    },
    {
      "epoch": 4.562427071178529,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011374938007357474,
      "loss": 0.0214,
      "step": 9775
    },
    {
      "epoch": 4.562893815635939,
      "grad_norm": 0.349609375,
      "learning_rate": 0.0001137348458024059,
      "loss": 0.0053,
      "step": 9776
    },
    {
      "epoch": 4.563360560093349,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00011372031123551056,
      "loss": 0.0052,
      "step": 9777
    },
    {
      "epoch": 4.563827304550759,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001137057763732017,
      "loss": 0.0075,
      "step": 9778
    },
    {
      "epoch": 4.5642940490081685,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011369124121579229,
      "loss": 0.0087,
      "step": 9779
    },
    {
      "epoch": 4.564760793465577,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00011367670576359526,
      "loss": 0.0078,
      "step": 9780
    },
    {
      "epoch": 4.565227537922987,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00011366217001692356,
      "loss": 0.0089,
      "step": 9781
    },
    {
      "epoch": 4.565694282380397,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00011364763397609022,
      "loss": 0.0035,
      "step": 9782
    },
    {
      "epoch": 4.566161026837806,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00011363309764140814,
      "loss": 0.008,
      "step": 9783
    },
    {
      "epoch": 4.566627771295216,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011361856101319037,
      "loss": 0.012,
      "step": 9784
    },
    {
      "epoch": 4.5670945157526255,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011360402409174984,
      "loss": 0.0074,
      "step": 9785
    },
    {
      "epoch": 4.567561260210035,
      "grad_norm": 0.28125,
      "learning_rate": 0.0001135894868773996,
      "loss": 0.0067,
      "step": 9786
    },
    {
      "epoch": 4.568028004667444,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011357494937045263,
      "loss": 0.0083,
      "step": 9787
    },
    {
      "epoch": 4.568494749124854,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00011356041157122193,
      "loss": 0.0057,
      "step": 9788
    },
    {
      "epoch": 4.568961493582264,
      "grad_norm": 0.625,
      "learning_rate": 0.00011354587348002053,
      "loss": 0.0094,
      "step": 9789
    },
    {
      "epoch": 4.569428238039674,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00011353133509716145,
      "loss": 0.0045,
      "step": 9790
    },
    {
      "epoch": 4.5698949824970825,
      "grad_norm": 0.52734375,
      "learning_rate": 0.0001135167964229577,
      "loss": 0.0089,
      "step": 9791
    },
    {
      "epoch": 4.570361726954492,
      "grad_norm": 0.625,
      "learning_rate": 0.00011350225745772232,
      "loss": 0.0102,
      "step": 9792
    },
    {
      "epoch": 4.570828471411902,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011348771820176838,
      "loss": 0.0063,
      "step": 9793
    },
    {
      "epoch": 4.571295215869312,
      "grad_norm": 0.6796875,
      "learning_rate": 0.0001134731786554089,
      "loss": 0.0146,
      "step": 9794
    },
    {
      "epoch": 4.571761960326721,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011345863881895689,
      "loss": 0.0152,
      "step": 9795
    },
    {
      "epoch": 4.572228704784131,
      "grad_norm": 0.34375,
      "learning_rate": 0.00011344409869272552,
      "loss": 0.0088,
      "step": 9796
    },
    {
      "epoch": 4.57269544924154,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00011342955827702774,
      "loss": 0.0073,
      "step": 9797
    },
    {
      "epoch": 4.57316219369895,
      "grad_norm": 0.341796875,
      "learning_rate": 0.0001134150175721767,
      "loss": 0.0087,
      "step": 9798
    },
    {
      "epoch": 4.573628938156359,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00011340047657848542,
      "loss": 0.0086,
      "step": 9799
    },
    {
      "epoch": 4.574095682613769,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011338593529626704,
      "loss": 0.0099,
      "step": 9800
    },
    {
      "epoch": 4.574562427071179,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00011337139372583458,
      "loss": 0.0083,
      "step": 9801
    },
    {
      "epoch": 4.5750291715285885,
      "grad_norm": 0.5,
      "learning_rate": 0.00011335685186750123,
      "loss": 0.0101,
      "step": 9802
    },
    {
      "epoch": 4.575495915985997,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00011334230972158,
      "loss": 0.008,
      "step": 9803
    },
    {
      "epoch": 4.575962660443407,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011332776728838403,
      "loss": 0.0088,
      "step": 9804
    },
    {
      "epoch": 4.576429404900817,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011331322456822649,
      "loss": 0.0086,
      "step": 9805
    },
    {
      "epoch": 4.576896149358227,
      "grad_norm": 0.455078125,
      "learning_rate": 0.0001132986815614204,
      "loss": 0.0114,
      "step": 9806
    },
    {
      "epoch": 4.577362893815636,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00011328413826827897,
      "loss": 0.0132,
      "step": 9807
    },
    {
      "epoch": 4.5778296382730455,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011326959468911531,
      "loss": 0.0202,
      "step": 9808
    },
    {
      "epoch": 4.578296382730455,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00011325505082424255,
      "loss": 0.0111,
      "step": 9809
    },
    {
      "epoch": 4.578763127187864,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011324050667397383,
      "loss": 0.013,
      "step": 9810
    },
    {
      "epoch": 4.579229871645274,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011322596223862229,
      "loss": 0.009,
      "step": 9811
    },
    {
      "epoch": 4.579696616102684,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00011321141751850115,
      "loss": 0.0094,
      "step": 9812
    },
    {
      "epoch": 4.5801633605600935,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00011319687251392353,
      "loss": 0.0067,
      "step": 9813
    },
    {
      "epoch": 4.580630105017503,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011318232722520257,
      "loss": 0.0082,
      "step": 9814
    },
    {
      "epoch": 4.581096849474912,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001131677816526515,
      "loss": 0.0113,
      "step": 9815
    },
    {
      "epoch": 4.581563593932322,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00011315323579658348,
      "loss": 0.0105,
      "step": 9816
    },
    {
      "epoch": 4.582030338389732,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011313868965731169,
      "loss": 0.0079,
      "step": 9817
    },
    {
      "epoch": 4.582497082847141,
      "grad_norm": 0.390625,
      "learning_rate": 0.00011312414323514935,
      "loss": 0.0053,
      "step": 9818
    },
    {
      "epoch": 4.5829638273045505,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00011310959653040967,
      "loss": 0.0083,
      "step": 9819
    },
    {
      "epoch": 4.58343057176196,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011309504954340577,
      "loss": 0.0114,
      "step": 9820
    },
    {
      "epoch": 4.58389731621937,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00011308050227445095,
      "loss": 0.0145,
      "step": 9821
    },
    {
      "epoch": 4.58436406067678,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00011306595472385842,
      "loss": 0.0123,
      "step": 9822
    },
    {
      "epoch": 4.584830805134189,
      "grad_norm": 0.578125,
      "learning_rate": 0.00011305140689194136,
      "loss": 0.0105,
      "step": 9823
    },
    {
      "epoch": 4.585297549591599,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011303685877901304,
      "loss": 0.0082,
      "step": 9824
    },
    {
      "epoch": 4.585764294049008,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001130223103853867,
      "loss": 0.0164,
      "step": 9825
    },
    {
      "epoch": 4.586231038506417,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011300776171137557,
      "loss": 0.0057,
      "step": 9826
    },
    {
      "epoch": 4.586697782963827,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011299321275729287,
      "loss": 0.009,
      "step": 9827
    },
    {
      "epoch": 4.587164527421237,
      "grad_norm": 0.369140625,
      "learning_rate": 0.0001129786635234519,
      "loss": 0.006,
      "step": 9828
    },
    {
      "epoch": 4.587631271878647,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00011296411401016589,
      "loss": 0.0046,
      "step": 9829
    },
    {
      "epoch": 4.588098016336056,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00011294956421774812,
      "loss": 0.0041,
      "step": 9830
    },
    {
      "epoch": 4.588564760793465,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00011293501414651189,
      "loss": 0.0057,
      "step": 9831
    },
    {
      "epoch": 4.589031505250875,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011292046379677044,
      "loss": 0.0088,
      "step": 9832
    },
    {
      "epoch": 4.589498249708285,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00011290591316883707,
      "loss": 0.008,
      "step": 9833
    },
    {
      "epoch": 4.589964994165694,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00011289136226302505,
      "loss": 0.0091,
      "step": 9834
    },
    {
      "epoch": 4.590431738623104,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011287681107964773,
      "loss": 0.0132,
      "step": 9835
    },
    {
      "epoch": 4.5908984830805135,
      "grad_norm": 0.72265625,
      "learning_rate": 0.0001128622596190184,
      "loss": 0.0232,
      "step": 9836
    },
    {
      "epoch": 4.591365227537923,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00011284770788145029,
      "loss": 0.0046,
      "step": 9837
    },
    {
      "epoch": 4.591831971995332,
      "grad_norm": 0.392578125,
      "learning_rate": 0.0001128331558672568,
      "loss": 0.0072,
      "step": 9838
    },
    {
      "epoch": 4.592298716452742,
      "grad_norm": 0.578125,
      "learning_rate": 0.00011281860357675124,
      "loss": 0.0122,
      "step": 9839
    },
    {
      "epoch": 4.592765460910152,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00011280405101024689,
      "loss": 0.0102,
      "step": 9840
    },
    {
      "epoch": 4.593232205367562,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00011278949816805715,
      "loss": 0.0053,
      "step": 9841
    },
    {
      "epoch": 4.5936989498249705,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001127749450504953,
      "loss": 0.0053,
      "step": 9842
    },
    {
      "epoch": 4.59416569428238,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011276039165787474,
      "loss": 0.0072,
      "step": 9843
    },
    {
      "epoch": 4.59463243873979,
      "grad_norm": 0.68359375,
      "learning_rate": 0.00011274583799050878,
      "loss": 0.0095,
      "step": 9844
    },
    {
      "epoch": 4.5950991831972,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00011273128404871076,
      "loss": 0.0099,
      "step": 9845
    },
    {
      "epoch": 4.595565927654609,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00011271672983279406,
      "loss": 0.0058,
      "step": 9846
    },
    {
      "epoch": 4.596032672112019,
      "grad_norm": 0.345703125,
      "learning_rate": 0.0001127021753430721,
      "loss": 0.0049,
      "step": 9847
    },
    {
      "epoch": 4.596499416569428,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001126876205798582,
      "loss": 0.0089,
      "step": 9848
    },
    {
      "epoch": 4.596966161026838,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00011267306554346575,
      "loss": 0.0059,
      "step": 9849
    },
    {
      "epoch": 4.597432905484247,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00011265851023420812,
      "loss": 0.0081,
      "step": 9850
    },
    {
      "epoch": 4.597899649941657,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00011264395465239874,
      "loss": 0.0059,
      "step": 9851
    },
    {
      "epoch": 4.598366394399067,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00011262939879835097,
      "loss": 0.009,
      "step": 9852
    },
    {
      "epoch": 4.598833138856476,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00011261484267237822,
      "loss": 0.0071,
      "step": 9853
    },
    {
      "epoch": 4.599299883313885,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00011260028627479395,
      "loss": 0.0042,
      "step": 9854
    },
    {
      "epoch": 4.599766627771295,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00011258572960591154,
      "loss": 0.0096,
      "step": 9855
    },
    {
      "epoch": 4.600233372228705,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011257117266604436,
      "loss": 0.0133,
      "step": 9856
    },
    {
      "epoch": 4.600700116686115,
      "grad_norm": 0.6171875,
      "learning_rate": 0.0001125566154555059,
      "loss": 0.0076,
      "step": 9857
    },
    {
      "epoch": 4.601166861143524,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00011254205797460962,
      "loss": 0.0056,
      "step": 9858
    },
    {
      "epoch": 4.6016336056009335,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011252750022366887,
      "loss": 0.0086,
      "step": 9859
    },
    {
      "epoch": 4.602100350058343,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00011251294220299715,
      "loss": 0.0071,
      "step": 9860
    },
    {
      "epoch": 4.602567094515752,
      "grad_norm": 0.365234375,
      "learning_rate": 0.0001124983839129079,
      "loss": 0.007,
      "step": 9861
    },
    {
      "epoch": 4.603033838973162,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001124838253537146,
      "loss": 0.006,
      "step": 9862
    },
    {
      "epoch": 4.603500583430572,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00011246926652573064,
      "loss": 0.0095,
      "step": 9863
    },
    {
      "epoch": 4.603967327887982,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00011245470742926957,
      "loss": 0.0073,
      "step": 9864
    },
    {
      "epoch": 4.604434072345391,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00011244014806464484,
      "loss": 0.0151,
      "step": 9865
    },
    {
      "epoch": 4.6049008168028,
      "grad_norm": 0.39453125,
      "learning_rate": 0.0001124255884321699,
      "loss": 0.0046,
      "step": 9866
    },
    {
      "epoch": 4.60536756126021,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001124110285321583,
      "loss": 0.0103,
      "step": 9867
    },
    {
      "epoch": 4.60583430571762,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00011239646836492346,
      "loss": 0.0056,
      "step": 9868
    },
    {
      "epoch": 4.606301050175029,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00011238190793077885,
      "loss": 0.0181,
      "step": 9869
    },
    {
      "epoch": 4.606767794632439,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00011236734723003807,
      "loss": 0.0075,
      "step": 9870
    },
    {
      "epoch": 4.607234539089848,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001123527862630146,
      "loss": 0.0066,
      "step": 9871
    },
    {
      "epoch": 4.607701283547258,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011233822503002189,
      "loss": 0.0093,
      "step": 9872
    },
    {
      "epoch": 4.608168028004667,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011232366353137354,
      "loss": 0.0106,
      "step": 9873
    },
    {
      "epoch": 4.608634772462077,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00011230910176738303,
      "loss": 0.0083,
      "step": 9874
    },
    {
      "epoch": 4.609101516919487,
      "grad_norm": 0.75390625,
      "learning_rate": 0.0001122945397383639,
      "loss": 0.0201,
      "step": 9875
    },
    {
      "epoch": 4.6095682613768965,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00011227997744462968,
      "loss": 0.0185,
      "step": 9876
    },
    {
      "epoch": 4.610035005834305,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00011226541488649393,
      "loss": 0.0051,
      "step": 9877
    },
    {
      "epoch": 4.610501750291715,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00011225085206427019,
      "loss": 0.0116,
      "step": 9878
    },
    {
      "epoch": 4.610968494749125,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000112236288978272,
      "loss": 0.0084,
      "step": 9879
    },
    {
      "epoch": 4.611435239206535,
      "grad_norm": 0.40625,
      "learning_rate": 0.00011222172562881297,
      "loss": 0.0063,
      "step": 9880
    },
    {
      "epoch": 4.611901983663944,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011220716201620661,
      "loss": 0.0083,
      "step": 9881
    },
    {
      "epoch": 4.6123687281213535,
      "grad_norm": 0.578125,
      "learning_rate": 0.00011219259814076646,
      "loss": 0.0113,
      "step": 9882
    },
    {
      "epoch": 4.612835472578763,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001121780340028062,
      "loss": 0.018,
      "step": 9883
    },
    {
      "epoch": 4.613302217036173,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00011216346960263934,
      "loss": 0.0072,
      "step": 9884
    },
    {
      "epoch": 4.613768961493582,
      "grad_norm": 0.734375,
      "learning_rate": 0.00011214890494057949,
      "loss": 0.0101,
      "step": 9885
    },
    {
      "epoch": 4.614235705950992,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011213434001694022,
      "loss": 0.0048,
      "step": 9886
    },
    {
      "epoch": 4.6147024504084015,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00011211977483203515,
      "loss": 0.0071,
      "step": 9887
    },
    {
      "epoch": 4.615169194865811,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001121052093861779,
      "loss": 0.0079,
      "step": 9888
    },
    {
      "epoch": 4.61563593932322,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00011209064367968203,
      "loss": 0.0087,
      "step": 9889
    },
    {
      "epoch": 4.61610268378063,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011207607771286121,
      "loss": 0.0128,
      "step": 9890
    },
    {
      "epoch": 4.61656942823804,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00011206151148602906,
      "loss": 0.0079,
      "step": 9891
    },
    {
      "epoch": 4.61703617269545,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00011204694499949915,
      "loss": 0.0078,
      "step": 9892
    },
    {
      "epoch": 4.6175029171528585,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00011203237825358515,
      "loss": 0.0091,
      "step": 9893
    },
    {
      "epoch": 4.617969661610268,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011201781124860071,
      "loss": 0.007,
      "step": 9894
    },
    {
      "epoch": 4.618436406067678,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011200324398485943,
      "loss": 0.0069,
      "step": 9895
    },
    {
      "epoch": 4.618903150525087,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00011198867646267502,
      "loss": 0.0045,
      "step": 9896
    },
    {
      "epoch": 4.619369894982497,
      "grad_norm": 0.45703125,
      "learning_rate": 0.0001119741086823611,
      "loss": 0.0051,
      "step": 9897
    },
    {
      "epoch": 4.619836639439907,
      "grad_norm": 0.275390625,
      "learning_rate": 0.00011195954064423133,
      "loss": 0.005,
      "step": 9898
    },
    {
      "epoch": 4.620303383897316,
      "grad_norm": 0.71875,
      "learning_rate": 0.00011194497234859938,
      "loss": 0.0101,
      "step": 9899
    },
    {
      "epoch": 4.620770128354726,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00011193040379577893,
      "loss": 0.015,
      "step": 9900
    },
    {
      "epoch": 4.621236872812135,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00011191583498608364,
      "loss": 0.0104,
      "step": 9901
    },
    {
      "epoch": 4.621703617269545,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011190126591982719,
      "loss": 0.0078,
      "step": 9902
    },
    {
      "epoch": 4.622170361726955,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00011188669659732329,
      "loss": 0.0174,
      "step": 9903
    },
    {
      "epoch": 4.622637106184364,
      "grad_norm": 0.765625,
      "learning_rate": 0.00011187212701888563,
      "loss": 0.0137,
      "step": 9904
    },
    {
      "epoch": 4.623103850641773,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00011185755718482788,
      "loss": 0.0084,
      "step": 9905
    },
    {
      "epoch": 4.623570595099183,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00011184298709546379,
      "loss": 0.0109,
      "step": 9906
    },
    {
      "epoch": 4.624037339556593,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00011182841675110707,
      "loss": 0.0076,
      "step": 9907
    },
    {
      "epoch": 4.624504084014003,
      "grad_norm": 0.5625,
      "learning_rate": 0.00011181384615207138,
      "loss": 0.0092,
      "step": 9908
    },
    {
      "epoch": 4.624970828471412,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011179927529867048,
      "loss": 0.0136,
      "step": 9909
    },
    {
      "epoch": 4.6254375729288215,
      "grad_norm": 0.58984375,
      "learning_rate": 0.0001117847041912181,
      "loss": 0.0069,
      "step": 9910
    },
    {
      "epoch": 4.625904317386231,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011177013283002795,
      "loss": 0.0089,
      "step": 9911
    },
    {
      "epoch": 4.62637106184364,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00011175556121541377,
      "loss": 0.0067,
      "step": 9912
    },
    {
      "epoch": 4.62683780630105,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011174098934768936,
      "loss": 0.0141,
      "step": 9913
    },
    {
      "epoch": 4.62730455075846,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011172641722716839,
      "loss": 0.008,
      "step": 9914
    },
    {
      "epoch": 4.62777129521587,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00011171184485416466,
      "loss": 0.0082,
      "step": 9915
    },
    {
      "epoch": 4.6282380396732785,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011169727222899191,
      "loss": 0.0144,
      "step": 9916
    },
    {
      "epoch": 4.628704784130688,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011168269935196392,
      "loss": 0.0053,
      "step": 9917
    },
    {
      "epoch": 4.629171528588098,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011166812622339442,
      "loss": 0.0154,
      "step": 9918
    },
    {
      "epoch": 4.629638273045508,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011165355284359725,
      "loss": 0.013,
      "step": 9919
    },
    {
      "epoch": 4.630105017502917,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00011163897921288615,
      "loss": 0.0089,
      "step": 9920
    },
    {
      "epoch": 4.630571761960327,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011162440533157488,
      "loss": 0.0114,
      "step": 9921
    },
    {
      "epoch": 4.631038506417736,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001116098311999773,
      "loss": 0.0066,
      "step": 9922
    },
    {
      "epoch": 4.631505250875146,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00011159525681840716,
      "loss": 0.0048,
      "step": 9923
    },
    {
      "epoch": 4.631971995332555,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001115806821871783,
      "loss": 0.01,
      "step": 9924
    },
    {
      "epoch": 4.632438739789965,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00011156610730660446,
      "loss": 0.0092,
      "step": 9925
    },
    {
      "epoch": 4.632905484247375,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00011155153217699955,
      "loss": 0.0078,
      "step": 9926
    },
    {
      "epoch": 4.6333722287047845,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001115369567986773,
      "loss": 0.0115,
      "step": 9927
    },
    {
      "epoch": 4.633838973162193,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00011152238117195155,
      "loss": 0.013,
      "step": 9928
    },
    {
      "epoch": 4.634305717619603,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00011150780529713619,
      "loss": 0.0202,
      "step": 9929
    },
    {
      "epoch": 4.634772462077013,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011149322917454501,
      "loss": 0.0139,
      "step": 9930
    },
    {
      "epoch": 4.635239206534423,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00011147865280449181,
      "loss": 0.0122,
      "step": 9931
    },
    {
      "epoch": 4.635705950991832,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00011146407618729053,
      "loss": 0.0098,
      "step": 9932
    },
    {
      "epoch": 4.6361726954492415,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011144949932325497,
      "loss": 0.0102,
      "step": 9933
    },
    {
      "epoch": 4.636639439906651,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011143492221269895,
      "loss": 0.009,
      "step": 9934
    },
    {
      "epoch": 4.637106184364061,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00011142034485593637,
      "loss": 0.0117,
      "step": 9935
    },
    {
      "epoch": 4.63757292882147,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011140576725328114,
      "loss": 0.0136,
      "step": 9936
    },
    {
      "epoch": 4.63803967327888,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00011139118940504704,
      "loss": 0.0064,
      "step": 9937
    },
    {
      "epoch": 4.63850641773629,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00011137661131154801,
      "loss": 0.0042,
      "step": 9938
    },
    {
      "epoch": 4.6389731621936985,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011136203297309795,
      "loss": 0.013,
      "step": 9939
    },
    {
      "epoch": 4.639439906651108,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00011134745439001069,
      "loss": 0.0044,
      "step": 9940
    },
    {
      "epoch": 4.639906651108518,
      "grad_norm": 0.341796875,
      "learning_rate": 0.00011133287556260011,
      "loss": 0.0044,
      "step": 9941
    },
    {
      "epoch": 4.640373395565928,
      "grad_norm": 0.302734375,
      "learning_rate": 0.00011131829649118018,
      "loss": 0.0055,
      "step": 9942
    },
    {
      "epoch": 4.640840140023338,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00011130371717606475,
      "loss": 0.0165,
      "step": 9943
    },
    {
      "epoch": 4.641306884480747,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00011128913761756776,
      "loss": 0.018,
      "step": 9944
    },
    {
      "epoch": 4.641773628938156,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011127455781600311,
      "loss": 0.0154,
      "step": 9945
    },
    {
      "epoch": 4.642240373395566,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00011125997777168473,
      "loss": 0.0051,
      "step": 9946
    },
    {
      "epoch": 4.642707117852975,
      "grad_norm": 0.65625,
      "learning_rate": 0.00011124539748492651,
      "loss": 0.013,
      "step": 9947
    },
    {
      "epoch": 4.643173862310385,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00011123081695604242,
      "loss": 0.0206,
      "step": 9948
    },
    {
      "epoch": 4.643640606767795,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011121623618534637,
      "loss": 0.01,
      "step": 9949
    },
    {
      "epoch": 4.6441073512252045,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011120165517315235,
      "loss": 0.0083,
      "step": 9950
    },
    {
      "epoch": 4.644574095682614,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00011118707391977421,
      "loss": 0.0097,
      "step": 9951
    },
    {
      "epoch": 4.645040840140023,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011117249242552599,
      "loss": 0.0075,
      "step": 9952
    },
    {
      "epoch": 4.645507584597433,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00011115791069072161,
      "loss": 0.0065,
      "step": 9953
    },
    {
      "epoch": 4.645974329054843,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00011114332871567502,
      "loss": 0.0228,
      "step": 9954
    },
    {
      "epoch": 4.646441073512252,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00011112874650070021,
      "loss": 0.0059,
      "step": 9955
    },
    {
      "epoch": 4.6469078179696615,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00011111416404611115,
      "loss": 0.0061,
      "step": 9956
    },
    {
      "epoch": 4.647374562427071,
      "grad_norm": 0.4375,
      "learning_rate": 0.0001110995813522218,
      "loss": 0.0067,
      "step": 9957
    },
    {
      "epoch": 4.647841306884481,
      "grad_norm": 0.59375,
      "learning_rate": 0.00011108499841934612,
      "loss": 0.0112,
      "step": 9958
    },
    {
      "epoch": 4.64830805134189,
      "grad_norm": 0.46875,
      "learning_rate": 0.00011107041524779815,
      "loss": 0.0058,
      "step": 9959
    },
    {
      "epoch": 4.6487747957993,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011105583183789185,
      "loss": 0.0104,
      "step": 9960
    },
    {
      "epoch": 4.6492415402567095,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00011104124818994121,
      "loss": 0.0092,
      "step": 9961
    },
    {
      "epoch": 4.649708284714119,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001110266643042603,
      "loss": 0.0096,
      "step": 9962
    },
    {
      "epoch": 4.650175029171528,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00011101208018116305,
      "loss": 0.0149,
      "step": 9963
    },
    {
      "epoch": 4.650641773628938,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00011099749582096346,
      "loss": 0.0096,
      "step": 9964
    },
    {
      "epoch": 4.651108518086348,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00011098291122397562,
      "loss": 0.0085,
      "step": 9965
    },
    {
      "epoch": 4.651575262543758,
      "grad_norm": 0.326171875,
      "learning_rate": 0.0001109683263905135,
      "loss": 0.0118,
      "step": 9966
    },
    {
      "epoch": 4.6520420070011665,
      "grad_norm": 0.78125,
      "learning_rate": 0.00011095374132089114,
      "loss": 0.0177,
      "step": 9967
    },
    {
      "epoch": 4.652508751458576,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00011093915601542258,
      "loss": 0.016,
      "step": 9968
    },
    {
      "epoch": 4.652975495915986,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011092457047442186,
      "loss": 0.0109,
      "step": 9969
    },
    {
      "epoch": 4.653442240373396,
      "grad_norm": 0.53515625,
      "learning_rate": 0.000110909984698203,
      "loss": 0.0098,
      "step": 9970
    },
    {
      "epoch": 4.653908984830805,
      "grad_norm": 0.609375,
      "learning_rate": 0.00011089539868708008,
      "loss": 0.0076,
      "step": 9971
    },
    {
      "epoch": 4.654375729288215,
      "grad_norm": 0.421875,
      "learning_rate": 0.00011088081244136715,
      "loss": 0.0054,
      "step": 9972
    },
    {
      "epoch": 4.654842473745624,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00011086622596137824,
      "loss": 0.0112,
      "step": 9973
    },
    {
      "epoch": 4.655309218203034,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011085163924742741,
      "loss": 0.0146,
      "step": 9974
    },
    {
      "epoch": 4.655775962660443,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00011083705229982877,
      "loss": 0.015,
      "step": 9975
    },
    {
      "epoch": 4.656242707117853,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011082246511889638,
      "loss": 0.0106,
      "step": 9976
    },
    {
      "epoch": 4.656709451575263,
      "grad_norm": 0.453125,
      "learning_rate": 0.00011080787770494427,
      "loss": 0.0069,
      "step": 9977
    },
    {
      "epoch": 4.6571761960326725,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00011079329005828657,
      "loss": 0.0084,
      "step": 9978
    },
    {
      "epoch": 4.657642940490081,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001107787021792374,
      "loss": 0.0096,
      "step": 9979
    },
    {
      "epoch": 4.658109684947491,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00011076411406811078,
      "loss": 0.0082,
      "step": 9980
    },
    {
      "epoch": 4.658576429404901,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00011074952572522084,
      "loss": 0.0128,
      "step": 9981
    },
    {
      "epoch": 4.65904317386231,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001107349371508817,
      "loss": 0.007,
      "step": 9982
    },
    {
      "epoch": 4.65950991831972,
      "grad_norm": 0.46875,
      "learning_rate": 0.00011072034834540745,
      "loss": 0.0132,
      "step": 9983
    },
    {
      "epoch": 4.6599766627771295,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00011070575930911217,
      "loss": 0.0078,
      "step": 9984
    },
    {
      "epoch": 4.660443407234539,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011069117004231007,
      "loss": 0.012,
      "step": 9985
    },
    {
      "epoch": 4.660910151691949,
      "grad_norm": 0.34375,
      "learning_rate": 0.0001106765805453152,
      "loss": 0.0081,
      "step": 9986
    },
    {
      "epoch": 4.661376896149358,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00011066199081844171,
      "loss": 0.0113,
      "step": 9987
    },
    {
      "epoch": 4.661843640606768,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00011064740086200372,
      "loss": 0.0058,
      "step": 9988
    },
    {
      "epoch": 4.662310385064178,
      "grad_norm": 0.8125,
      "learning_rate": 0.00011063281067631536,
      "loss": 0.0184,
      "step": 9989
    },
    {
      "epoch": 4.6627771295215865,
      "grad_norm": 0.78125,
      "learning_rate": 0.0001106182202616908,
      "loss": 0.0112,
      "step": 9990
    },
    {
      "epoch": 4.663243873978996,
      "grad_norm": 0.53125,
      "learning_rate": 0.00011060362961844417,
      "loss": 0.0116,
      "step": 9991
    },
    {
      "epoch": 4.663710618436406,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00011058903874688964,
      "loss": 0.008,
      "step": 9992
    },
    {
      "epoch": 4.664177362893816,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00011057444764734133,
      "loss": 0.005,
      "step": 9993
    },
    {
      "epoch": 4.664644107351226,
      "grad_norm": 0.515625,
      "learning_rate": 0.00011055985632011347,
      "loss": 0.0082,
      "step": 9994
    },
    {
      "epoch": 4.665110851808635,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00011054526476552016,
      "loss": 0.0089,
      "step": 9995
    },
    {
      "epoch": 4.665577596266044,
      "grad_norm": 0.77734375,
      "learning_rate": 0.0001105306729838756,
      "loss": 0.0111,
      "step": 9996
    },
    {
      "epoch": 4.666044340723454,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00011051608097549398,
      "loss": 0.0045,
      "step": 9997
    },
    {
      "epoch": 4.666511085180863,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00011050148874068948,
      "loss": 0.0127,
      "step": 9998
    },
    {
      "epoch": 4.666977829638273,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00011048689627977627,
      "loss": 0.0074,
      "step": 9999
    },
    {
      "epoch": 4.667444574095683,
      "grad_norm": 0.390625,
      "learning_rate": 0.0001104723035930685,
      "loss": 0.0055,
      "step": 10000
    },
    {
      "epoch": 4.6679113185530925,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011045771068088047,
      "loss": 0.0119,
      "step": 10001
    },
    {
      "epoch": 4.668378063010501,
      "grad_norm": 0.484375,
      "learning_rate": 0.00011044311754352632,
      "loss": 0.0083,
      "step": 10002
    },
    {
      "epoch": 4.668844807467911,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00011042852418132024,
      "loss": 0.0099,
      "step": 10003
    },
    {
      "epoch": 4.669311551925321,
      "grad_norm": 0.640625,
      "learning_rate": 0.00011041393059457647,
      "loss": 0.0132,
      "step": 10004
    },
    {
      "epoch": 4.669778296382731,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00011039933678360923,
      "loss": 0.0084,
      "step": 10005
    },
    {
      "epoch": 4.67024504084014,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011038474274873274,
      "loss": 0.0136,
      "step": 10006
    },
    {
      "epoch": 4.6707117852975495,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00011037014849026117,
      "loss": 0.0146,
      "step": 10007
    },
    {
      "epoch": 4.671178529754959,
      "grad_norm": 0.279296875,
      "learning_rate": 0.00011035555400850884,
      "loss": 0.0067,
      "step": 10008
    },
    {
      "epoch": 4.671645274212369,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00011034095930378993,
      "loss": 0.0177,
      "step": 10009
    },
    {
      "epoch": 4.672112018669778,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00011032636437641867,
      "loss": 0.0124,
      "step": 10010
    },
    {
      "epoch": 4.672578763127188,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00011031176922670938,
      "loss": 0.0062,
      "step": 10011
    },
    {
      "epoch": 4.673045507584598,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00011029717385497624,
      "loss": 0.007,
      "step": 10012
    },
    {
      "epoch": 4.673512252042007,
      "grad_norm": 0.5859375,
      "learning_rate": 0.0001102825782615335,
      "loss": 0.0113,
      "step": 10013
    },
    {
      "epoch": 4.673978996499416,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00011026798244669546,
      "loss": 0.0092,
      "step": 10014
    },
    {
      "epoch": 4.674445740956826,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00011025338641077636,
      "loss": 0.0114,
      "step": 10015
    },
    {
      "epoch": 4.674912485414236,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00011023879015409047,
      "loss": 0.0133,
      "step": 10016
    },
    {
      "epoch": 4.675379229871645,
      "grad_norm": 0.546875,
      "learning_rate": 0.00011022419367695207,
      "loss": 0.0146,
      "step": 10017
    },
    {
      "epoch": 4.675845974329055,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00011020959697967544,
      "loss": 0.0074,
      "step": 10018
    },
    {
      "epoch": 4.676312718786464,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00011019500006257483,
      "loss": 0.0112,
      "step": 10019
    },
    {
      "epoch": 4.676779463243874,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00011018040292596459,
      "loss": 0.0048,
      "step": 10020
    },
    {
      "epoch": 4.677246207701284,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00011016580557015899,
      "loss": 0.0086,
      "step": 10021
    },
    {
      "epoch": 4.677712952158693,
      "grad_norm": 0.62890625,
      "learning_rate": 0.0001101512079954723,
      "loss": 0.0173,
      "step": 10022
    },
    {
      "epoch": 4.678179696616103,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00011013661020221881,
      "loss": 0.0125,
      "step": 10023
    },
    {
      "epoch": 4.6786464410735125,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001101220121907129,
      "loss": 0.0154,
      "step": 10024
    },
    {
      "epoch": 4.679113185530921,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00011010741396126882,
      "loss": 0.0129,
      "step": 10025
    },
    {
      "epoch": 4.679579929988331,
      "grad_norm": 0.78515625,
      "learning_rate": 0.0001100928155142009,
      "loss": 0.0122,
      "step": 10026
    },
    {
      "epoch": 4.680046674445741,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00011007821684982347,
      "loss": 0.0145,
      "step": 10027
    },
    {
      "epoch": 4.680513418903151,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00011006361796845083,
      "loss": 0.0082,
      "step": 10028
    },
    {
      "epoch": 4.6809801633605606,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00011004901887039734,
      "loss": 0.0098,
      "step": 10029
    },
    {
      "epoch": 4.6814469078179695,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00011003441955597733,
      "loss": 0.0178,
      "step": 10030
    },
    {
      "epoch": 4.681913652275379,
      "grad_norm": 0.4375,
      "learning_rate": 0.00011001982002550515,
      "loss": 0.006,
      "step": 10031
    },
    {
      "epoch": 4.682380396732789,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00011000522027929512,
      "loss": 0.0103,
      "step": 10032
    },
    {
      "epoch": 4.682847141190198,
      "grad_norm": 0.255859375,
      "learning_rate": 0.00010999062031766156,
      "loss": 0.0032,
      "step": 10033
    },
    {
      "epoch": 4.683313885647608,
      "grad_norm": 0.390625,
      "learning_rate": 0.00010997602014091893,
      "loss": 0.0076,
      "step": 10034
    },
    {
      "epoch": 4.6837806301050176,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010996141974938145,
      "loss": 0.0099,
      "step": 10035
    },
    {
      "epoch": 4.684247374562427,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00010994681914336358,
      "loss": 0.011,
      "step": 10036
    },
    {
      "epoch": 4.684714119019836,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010993221832317966,
      "loss": 0.0131,
      "step": 10037
    },
    {
      "epoch": 4.685180863477246,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010991761728914405,
      "loss": 0.012,
      "step": 10038
    },
    {
      "epoch": 4.685647607934656,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00010990301604157114,
      "loss": 0.0117,
      "step": 10039
    },
    {
      "epoch": 4.686114352392066,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001098884145807753,
      "loss": 0.0128,
      "step": 10040
    },
    {
      "epoch": 4.6865810968494745,
      "grad_norm": 0.76953125,
      "learning_rate": 0.00010987381290707093,
      "loss": 0.0217,
      "step": 10041
    },
    {
      "epoch": 4.687047841306884,
      "grad_norm": 0.46484375,
      "learning_rate": 0.0001098592110207724,
      "loss": 0.0057,
      "step": 10042
    },
    {
      "epoch": 4.687514585764294,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010984460892219415,
      "loss": 0.0137,
      "step": 10043
    },
    {
      "epoch": 4.687981330221704,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010983000661165053,
      "loss": 0.0133,
      "step": 10044
    },
    {
      "epoch": 4.688448074679113,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00010981540408945593,
      "loss": 0.0113,
      "step": 10045
    },
    {
      "epoch": 4.688914819136523,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010980080135592483,
      "loss": 0.004,
      "step": 10046
    },
    {
      "epoch": 4.689381563593932,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00010978619841137158,
      "loss": 0.0157,
      "step": 10047
    },
    {
      "epoch": 4.689848308051342,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010977159525611065,
      "loss": 0.0101,
      "step": 10048
    },
    {
      "epoch": 4.690315052508751,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00010975699189045638,
      "loss": 0.008,
      "step": 10049
    },
    {
      "epoch": 4.690781796966161,
      "grad_norm": 0.515625,
      "learning_rate": 0.00010974238831472329,
      "loss": 0.0159,
      "step": 10050
    },
    {
      "epoch": 4.691248541423571,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00010972778452922576,
      "loss": 0.0109,
      "step": 10051
    },
    {
      "epoch": 4.6917152858809805,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00010971318053427824,
      "loss": 0.0055,
      "step": 10052
    },
    {
      "epoch": 4.692182030338389,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010969857633019516,
      "loss": 0.0129,
      "step": 10053
    },
    {
      "epoch": 4.692648774795799,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00010968397191729098,
      "loss": 0.0078,
      "step": 10054
    },
    {
      "epoch": 4.693115519253209,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010966936729588011,
      "loss": 0.0129,
      "step": 10055
    },
    {
      "epoch": 4.693582263710619,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010965476246627708,
      "loss": 0.0131,
      "step": 10056
    },
    {
      "epoch": 4.694049008168028,
      "grad_norm": 0.53125,
      "learning_rate": 0.00010964015742879628,
      "loss": 0.0097,
      "step": 10057
    },
    {
      "epoch": 4.6945157526254375,
      "grad_norm": 0.8359375,
      "learning_rate": 0.0001096255521837522,
      "loss": 0.0236,
      "step": 10058
    },
    {
      "epoch": 4.694982497082847,
      "grad_norm": 0.412109375,
      "learning_rate": 0.0001096109467314593,
      "loss": 0.0069,
      "step": 10059
    },
    {
      "epoch": 4.695449241540256,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00010959634107223206,
      "loss": 0.008,
      "step": 10060
    },
    {
      "epoch": 4.695915985997666,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010958173520638493,
      "loss": 0.0131,
      "step": 10061
    },
    {
      "epoch": 4.696382730455076,
      "grad_norm": 0.546875,
      "learning_rate": 0.00010956712913423242,
      "loss": 0.0077,
      "step": 10062
    },
    {
      "epoch": 4.696849474912486,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00010955252285608902,
      "loss": 0.0069,
      "step": 10063
    },
    {
      "epoch": 4.697316219369895,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00010953791637226918,
      "loss": 0.007,
      "step": 10064
    },
    {
      "epoch": 4.697782963827304,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00010952330968308742,
      "loss": 0.0084,
      "step": 10065
    },
    {
      "epoch": 4.698249708284714,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010950870278885823,
      "loss": 0.0074,
      "step": 10066
    },
    {
      "epoch": 4.698716452742124,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00010949409568989614,
      "loss": 0.0084,
      "step": 10067
    },
    {
      "epoch": 4.699183197199533,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00010947948838651561,
      "loss": 0.0127,
      "step": 10068
    },
    {
      "epoch": 4.699649941656943,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001094648808790312,
      "loss": 0.0068,
      "step": 10069
    },
    {
      "epoch": 4.700116686114352,
      "grad_norm": 0.66015625,
      "learning_rate": 0.0001094502731677574,
      "loss": 0.0178,
      "step": 10070
    },
    {
      "epoch": 4.700583430571762,
      "grad_norm": 0.5,
      "learning_rate": 0.00010943566525300875,
      "loss": 0.0104,
      "step": 10071
    },
    {
      "epoch": 4.701050175029172,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010942105713509972,
      "loss": 0.0102,
      "step": 10072
    },
    {
      "epoch": 4.701516919486581,
      "grad_norm": 0.447265625,
      "learning_rate": 0.0001094064488143449,
      "loss": 0.0064,
      "step": 10073
    },
    {
      "epoch": 4.701983663943991,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001093918402910588,
      "loss": 0.0062,
      "step": 10074
    },
    {
      "epoch": 4.7024504084014005,
      "grad_norm": 0.40625,
      "learning_rate": 0.00010937723156555595,
      "loss": 0.0051,
      "step": 10075
    },
    {
      "epoch": 4.702917152858809,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010936262263815089,
      "loss": 0.0109,
      "step": 10076
    },
    {
      "epoch": 4.703383897316219,
      "grad_norm": 0.8125,
      "learning_rate": 0.0001093480135091582,
      "loss": 0.0102,
      "step": 10077
    },
    {
      "epoch": 4.703850641773629,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001093334041788924,
      "loss": 0.0163,
      "step": 10078
    },
    {
      "epoch": 4.704317386231039,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00010931879464766806,
      "loss": 0.0083,
      "step": 10079
    },
    {
      "epoch": 4.704784130688448,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010930418491579972,
      "loss": 0.0121,
      "step": 10080
    },
    {
      "epoch": 4.7052508751458575,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00010928957498360195,
      "loss": 0.0061,
      "step": 10081
    },
    {
      "epoch": 4.705717619603267,
      "grad_norm": 1.890625,
      "learning_rate": 0.00010927496485138935,
      "loss": 0.0091,
      "step": 10082
    },
    {
      "epoch": 4.706184364060677,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010926035451947645,
      "loss": 0.0119,
      "step": 10083
    },
    {
      "epoch": 4.706651108518086,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010924574398817782,
      "loss": 0.0085,
      "step": 10084
    },
    {
      "epoch": 4.707117852975496,
      "grad_norm": 0.78515625,
      "learning_rate": 0.00010923113325780809,
      "loss": 0.0167,
      "step": 10085
    },
    {
      "epoch": 4.707584597432906,
      "grad_norm": 0.53125,
      "learning_rate": 0.00010921652232868182,
      "loss": 0.0155,
      "step": 10086
    },
    {
      "epoch": 4.708051341890315,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001092019112011136,
      "loss": 0.0083,
      "step": 10087
    },
    {
      "epoch": 4.708518086347724,
      "grad_norm": 0.71484375,
      "learning_rate": 0.000109187299875418,
      "loss": 0.018,
      "step": 10088
    },
    {
      "epoch": 4.708984830805134,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00010917268835190966,
      "loss": 0.0075,
      "step": 10089
    },
    {
      "epoch": 4.709451575262544,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00010915807663090316,
      "loss": 0.0094,
      "step": 10090
    },
    {
      "epoch": 4.709918319719954,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00010914346471271309,
      "loss": 0.011,
      "step": 10091
    },
    {
      "epoch": 4.710385064177363,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001091288525976541,
      "loss": 0.0092,
      "step": 10092
    },
    {
      "epoch": 4.710851808634772,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010911424028604078,
      "loss": 0.0097,
      "step": 10093
    },
    {
      "epoch": 4.711318553092182,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010909962777818774,
      "loss": 0.0117,
      "step": 10094
    },
    {
      "epoch": 4.711785297549592,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00010908501507440965,
      "loss": 0.011,
      "step": 10095
    },
    {
      "epoch": 4.712252042007001,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001090704021750211,
      "loss": 0.0161,
      "step": 10096
    },
    {
      "epoch": 4.712718786464411,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010905578908033671,
      "loss": 0.018,
      "step": 10097
    },
    {
      "epoch": 4.7131855309218205,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010904117579067112,
      "loss": 0.0092,
      "step": 10098
    },
    {
      "epoch": 4.71365227537923,
      "grad_norm": 0.59765625,
      "learning_rate": 0.000109026562306339,
      "loss": 0.0115,
      "step": 10099
    },
    {
      "epoch": 4.714119019836639,
      "grad_norm": 0.4765625,
      "learning_rate": 0.000109011948627655,
      "loss": 0.0103,
      "step": 10100
    },
    {
      "epoch": 4.714585764294049,
      "grad_norm": 0.5625,
      "learning_rate": 0.00010899733475493372,
      "loss": 0.0093,
      "step": 10101
    },
    {
      "epoch": 4.715052508751459,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00010898272068848985,
      "loss": 0.009,
      "step": 10102
    },
    {
      "epoch": 4.715519253208868,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00010896810642863804,
      "loss": 0.0075,
      "step": 10103
    },
    {
      "epoch": 4.7159859976662775,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010895349197569293,
      "loss": 0.0118,
      "step": 10104
    },
    {
      "epoch": 4.716452742123687,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00010893887732996922,
      "loss": 0.0069,
      "step": 10105
    },
    {
      "epoch": 4.716919486581097,
      "grad_norm": 0.99609375,
      "learning_rate": 0.00010892426249178158,
      "loss": 0.014,
      "step": 10106
    },
    {
      "epoch": 4.717386231038507,
      "grad_norm": 0.54296875,
      "learning_rate": 0.00010890964746144463,
      "loss": 0.0094,
      "step": 10107
    },
    {
      "epoch": 4.717852975495916,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010889503223927311,
      "loss": 0.0169,
      "step": 10108
    },
    {
      "epoch": 4.7183197199533256,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010888041682558167,
      "loss": 0.0144,
      "step": 10109
    },
    {
      "epoch": 4.718786464410735,
      "grad_norm": 0.57421875,
      "learning_rate": 0.000108865801220685,
      "loss": 0.0104,
      "step": 10110
    },
    {
      "epoch": 4.719253208868144,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00010885118542489779,
      "loss": 0.0104,
      "step": 10111
    },
    {
      "epoch": 4.719719953325554,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00010883656943853475,
      "loss": 0.0086,
      "step": 10112
    },
    {
      "epoch": 4.720186697782964,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010882195326191056,
      "loss": 0.017,
      "step": 10113
    },
    {
      "epoch": 4.720653442240374,
      "grad_norm": 0.3671875,
      "learning_rate": 0.0001088073368953399,
      "loss": 0.005,
      "step": 10114
    },
    {
      "epoch": 4.721120186697783,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010879272033913754,
      "loss": 0.014,
      "step": 10115
    },
    {
      "epoch": 4.721586931155192,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00010877810359361815,
      "loss": 0.0109,
      "step": 10116
    },
    {
      "epoch": 4.722053675612602,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00010876348665909643,
      "loss": 0.0084,
      "step": 10117
    },
    {
      "epoch": 4.722520420070012,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010874886953588714,
      "loss": 0.0113,
      "step": 10118
    },
    {
      "epoch": 4.722987164527421,
      "grad_norm": 0.5234375,
      "learning_rate": 0.000108734252224305,
      "loss": 0.0071,
      "step": 10119
    },
    {
      "epoch": 4.723453908984831,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010871963472466466,
      "loss": 0.0165,
      "step": 10120
    },
    {
      "epoch": 4.72392065344224,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00010870501703728098,
      "loss": 0.0067,
      "step": 10121
    },
    {
      "epoch": 4.72438739789965,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010869039916246859,
      "loss": 0.0154,
      "step": 10122
    },
    {
      "epoch": 4.724854142357059,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00010867578110054226,
      "loss": 0.0059,
      "step": 10123
    },
    {
      "epoch": 4.725320886814469,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010866116285181673,
      "loss": 0.0114,
      "step": 10124
    },
    {
      "epoch": 4.725787631271879,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010864654441660676,
      "loss": 0.014,
      "step": 10125
    },
    {
      "epoch": 4.7262543757292885,
      "grad_norm": 0.453125,
      "learning_rate": 0.00010863192579522711,
      "loss": 0.0106,
      "step": 10126
    },
    {
      "epoch": 4.726721120186697,
      "grad_norm": 0.466796875,
      "learning_rate": 0.0001086173069879925,
      "loss": 0.0084,
      "step": 10127
    },
    {
      "epoch": 4.727187864644107,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00010860268799521774,
      "loss": 0.0061,
      "step": 10128
    },
    {
      "epoch": 4.727654609101517,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00010858806881721752,
      "loss": 0.0098,
      "step": 10129
    },
    {
      "epoch": 4.728121353558927,
      "grad_norm": 0.318359375,
      "learning_rate": 0.00010857344945430669,
      "loss": 0.0081,
      "step": 10130
    },
    {
      "epoch": 4.728588098016336,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010855882990679998,
      "loss": 0.0094,
      "step": 10131
    },
    {
      "epoch": 4.7290548424737455,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00010854421017501216,
      "loss": 0.0044,
      "step": 10132
    },
    {
      "epoch": 4.729521586931155,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00010852959025925798,
      "loss": 0.0106,
      "step": 10133
    },
    {
      "epoch": 4.729988331388565,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00010851497015985229,
      "loss": 0.0075,
      "step": 10134
    },
    {
      "epoch": 4.730455075845974,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010850034987710982,
      "loss": 0.0102,
      "step": 10135
    },
    {
      "epoch": 4.730921820303384,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00010848572941134543,
      "loss": 0.0093,
      "step": 10136
    },
    {
      "epoch": 4.731388564760794,
      "grad_norm": 0.5,
      "learning_rate": 0.0001084711087628738,
      "loss": 0.0075,
      "step": 10137
    },
    {
      "epoch": 4.731855309218203,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00010845648793200987,
      "loss": 0.0051,
      "step": 10138
    },
    {
      "epoch": 4.732322053675612,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010844186691906835,
      "loss": 0.0151,
      "step": 10139
    },
    {
      "epoch": 4.732788798133022,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010842724572436404,
      "loss": 0.0106,
      "step": 10140
    },
    {
      "epoch": 4.733255542590432,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010841262434821182,
      "loss": 0.0111,
      "step": 10141
    },
    {
      "epoch": 4.733722287047842,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010839800279092644,
      "loss": 0.0097,
      "step": 10142
    },
    {
      "epoch": 4.734189031505251,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00010838338105282273,
      "loss": 0.0133,
      "step": 10143
    },
    {
      "epoch": 4.73465577596266,
      "grad_norm": 0.65625,
      "learning_rate": 0.00010836875913421556,
      "loss": 0.0096,
      "step": 10144
    },
    {
      "epoch": 4.73512252042007,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001083541370354197,
      "loss": 0.0104,
      "step": 10145
    },
    {
      "epoch": 4.735589264877479,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010833951475675001,
      "loss": 0.0112,
      "step": 10146
    },
    {
      "epoch": 4.736056009334889,
      "grad_norm": 0.388671875,
      "learning_rate": 0.0001083248922985213,
      "loss": 0.0071,
      "step": 10147
    },
    {
      "epoch": 4.736522753792299,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00010831026966104845,
      "loss": 0.0105,
      "step": 10148
    },
    {
      "epoch": 4.7369894982497085,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010829564684464627,
      "loss": 0.0071,
      "step": 10149
    },
    {
      "epoch": 4.737456242707118,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00010828102384962955,
      "loss": 0.0149,
      "step": 10150
    },
    {
      "epoch": 4.737922987164527,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010826640067631326,
      "loss": 0.0095,
      "step": 10151
    },
    {
      "epoch": 4.738389731621937,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010825177732501217,
      "loss": 0.0177,
      "step": 10152
    },
    {
      "epoch": 4.738856476079347,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010823715379604116,
      "loss": 0.0153,
      "step": 10153
    },
    {
      "epoch": 4.739323220536756,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00010822253008971512,
      "loss": 0.0062,
      "step": 10154
    },
    {
      "epoch": 4.7397899649941655,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010820790620634887,
      "loss": 0.0099,
      "step": 10155
    },
    {
      "epoch": 4.740256709451575,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010819328214625726,
      "loss": 0.0204,
      "step": 10156
    },
    {
      "epoch": 4.740723453908985,
      "grad_norm": 0.54296875,
      "learning_rate": 0.0001081786579097552,
      "loss": 0.0096,
      "step": 10157
    },
    {
      "epoch": 4.741190198366395,
      "grad_norm": 0.71484375,
      "learning_rate": 0.0001081640334971576,
      "loss": 0.0082,
      "step": 10158
    },
    {
      "epoch": 4.741656942823804,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010814940890877924,
      "loss": 0.0138,
      "step": 10159
    },
    {
      "epoch": 4.742123687281214,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00010813478414493512,
      "loss": 0.0085,
      "step": 10160
    },
    {
      "epoch": 4.742590431738623,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010812015920594006,
      "loss": 0.0111,
      "step": 10161
    },
    {
      "epoch": 4.743057176196032,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010810553409210895,
      "loss": 0.0186,
      "step": 10162
    },
    {
      "epoch": 4.743523920653442,
      "grad_norm": 0.50390625,
      "learning_rate": 0.0001080909088037567,
      "loss": 0.0085,
      "step": 10163
    },
    {
      "epoch": 4.743990665110852,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00010807628334119821,
      "loss": 0.0056,
      "step": 10164
    },
    {
      "epoch": 4.744457409568262,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010806165770474839,
      "loss": 0.019,
      "step": 10165
    },
    {
      "epoch": 4.744924154025671,
      "grad_norm": 0.5234375,
      "learning_rate": 0.0001080470318947221,
      "loss": 0.0083,
      "step": 10166
    },
    {
      "epoch": 4.74539089848308,
      "grad_norm": 0.48828125,
      "learning_rate": 0.0001080324059114343,
      "loss": 0.0088,
      "step": 10167
    },
    {
      "epoch": 4.74585764294049,
      "grad_norm": 1.1484375,
      "learning_rate": 0.00010801777975519991,
      "loss": 0.0155,
      "step": 10168
    },
    {
      "epoch": 4.7463243873979,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001080031534263338,
      "loss": 0.014,
      "step": 10169
    },
    {
      "epoch": 4.746791131855309,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00010798852692515093,
      "loss": 0.0107,
      "step": 10170
    },
    {
      "epoch": 4.747257876312719,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010797390025196623,
      "loss": 0.0115,
      "step": 10171
    },
    {
      "epoch": 4.7477246207701285,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001079592734070946,
      "loss": 0.0085,
      "step": 10172
    },
    {
      "epoch": 4.748191365227538,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00010794464639085096,
      "loss": 0.0067,
      "step": 10173
    },
    {
      "epoch": 4.748658109684947,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010793001920355028,
      "loss": 0.0096,
      "step": 10174
    },
    {
      "epoch": 4.749124854142357,
      "grad_norm": 0.498046875,
      "learning_rate": 0.0001079153918455075,
      "loss": 0.0117,
      "step": 10175
    },
    {
      "epoch": 4.749591598599767,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00010790076431703751,
      "loss": 0.0073,
      "step": 10176
    },
    {
      "epoch": 4.750058343057177,
      "grad_norm": 0.625,
      "learning_rate": 0.00010788613661845535,
      "loss": 0.01,
      "step": 10177
    },
    {
      "epoch": 4.7505250875145855,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00010787150875007592,
      "loss": 0.0048,
      "step": 10178
    },
    {
      "epoch": 4.750991831971995,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00010785688071221414,
      "loss": 0.0105,
      "step": 10179
    },
    {
      "epoch": 4.751458576429405,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00010784225250518503,
      "loss": 0.0041,
      "step": 10180
    },
    {
      "epoch": 4.751925320886815,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010782762412930349,
      "loss": 0.0107,
      "step": 10181
    },
    {
      "epoch": 4.752392065344224,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010781299558488452,
      "loss": 0.0112,
      "step": 10182
    },
    {
      "epoch": 4.7528588098016336,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010779836687224312,
      "loss": 0.0226,
      "step": 10183
    },
    {
      "epoch": 4.753325554259043,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00010778373799169421,
      "loss": 0.0107,
      "step": 10184
    },
    {
      "epoch": 4.753325554259043,
      "eval_loss": 1.990020513534546,
      "eval_runtime": 55.9398,
      "eval_samples_per_second": 32.249,
      "eval_steps_per_second": 4.04,
      "step": 10184
    },
    {
      "epoch": 4.753792298716453,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00010776910894355279,
      "loss": 0.0116,
      "step": 10185
    },
    {
      "epoch": 4.754259043173862,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010775447972813379,
      "loss": 0.0104,
      "step": 10186
    },
    {
      "epoch": 4.754725787631272,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00010773985034575229,
      "loss": 0.0157,
      "step": 10187
    },
    {
      "epoch": 4.755192532088682,
      "grad_norm": 0.796875,
      "learning_rate": 0.00010772522079672322,
      "loss": 0.0149,
      "step": 10188
    },
    {
      "epoch": 4.7556592765460906,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00010771059108136155,
      "loss": 0.007,
      "step": 10189
    },
    {
      "epoch": 4.7561260210035,
      "grad_norm": 0.4296875,
      "learning_rate": 0.00010769596119998231,
      "loss": 0.005,
      "step": 10190
    },
    {
      "epoch": 4.75659276546091,
      "grad_norm": 0.9765625,
      "learning_rate": 0.00010768133115290051,
      "loss": 0.0109,
      "step": 10191
    },
    {
      "epoch": 4.75705950991832,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00010766670094043107,
      "loss": 0.0154,
      "step": 10192
    },
    {
      "epoch": 4.75752625437573,
      "grad_norm": 0.6328125,
      "learning_rate": 0.0001076520705628891,
      "loss": 0.0126,
      "step": 10193
    },
    {
      "epoch": 4.757992998833139,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010763744002058957,
      "loss": 0.0097,
      "step": 10194
    },
    {
      "epoch": 4.758459743290548,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010762280931384747,
      "loss": 0.0089,
      "step": 10195
    },
    {
      "epoch": 4.758926487747958,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00010760817844297779,
      "loss": 0.0098,
      "step": 10196
    },
    {
      "epoch": 4.759393232205367,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010759354740829564,
      "loss": 0.0184,
      "step": 10197
    },
    {
      "epoch": 4.759859976662777,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00010757891621011597,
      "loss": 0.0116,
      "step": 10198
    },
    {
      "epoch": 4.760326721120187,
      "grad_norm": 0.462890625,
      "learning_rate": 0.0001075642848487538,
      "loss": 0.0102,
      "step": 10199
    },
    {
      "epoch": 4.7607934655775965,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010754965332452422,
      "loss": 0.0086,
      "step": 10200
    },
    {
      "epoch": 4.761260210035006,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010753502163774221,
      "loss": 0.012,
      "step": 10201
    },
    {
      "epoch": 4.761726954492415,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010752038978872282,
      "loss": 0.0113,
      "step": 10202
    },
    {
      "epoch": 4.762193698949825,
      "grad_norm": 0.5,
      "learning_rate": 0.00010750575777778114,
      "loss": 0.008,
      "step": 10203
    },
    {
      "epoch": 4.762660443407235,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010749112560523211,
      "loss": 0.0076,
      "step": 10204
    },
    {
      "epoch": 4.763127187864644,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00010747649327139084,
      "loss": 0.0092,
      "step": 10205
    },
    {
      "epoch": 4.7635939323220535,
      "grad_norm": 1.8125,
      "learning_rate": 0.00010746186077657241,
      "loss": 0.0072,
      "step": 10206
    },
    {
      "epoch": 4.764060676779463,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010744722812109183,
      "loss": 0.0096,
      "step": 10207
    },
    {
      "epoch": 4.764527421236873,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010743259530526413,
      "loss": 0.0151,
      "step": 10208
    },
    {
      "epoch": 4.764994165694282,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010741796232940441,
      "loss": 0.0187,
      "step": 10209
    },
    {
      "epoch": 4.765460910151692,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010740332919382776,
      "loss": 0.0171,
      "step": 10210
    },
    {
      "epoch": 4.765927654609102,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010738869589884922,
      "loss": 0.0081,
      "step": 10211
    },
    {
      "epoch": 4.766394399066511,
      "grad_norm": 1.0625,
      "learning_rate": 0.00010737406244478383,
      "loss": 0.0147,
      "step": 10212
    },
    {
      "epoch": 4.76686114352392,
      "grad_norm": 0.7734375,
      "learning_rate": 0.00010735942883194672,
      "loss": 0.0225,
      "step": 10213
    },
    {
      "epoch": 4.76732788798133,
      "grad_norm": 0.75,
      "learning_rate": 0.00010734479506065294,
      "loss": 0.0121,
      "step": 10214
    },
    {
      "epoch": 4.76779463243874,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010733016113121754,
      "loss": 0.0177,
      "step": 10215
    },
    {
      "epoch": 4.76826137689615,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010731552704395566,
      "loss": 0.0206,
      "step": 10216
    },
    {
      "epoch": 4.768728121353559,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00010730089279918238,
      "loss": 0.0074,
      "step": 10217
    },
    {
      "epoch": 4.769194865810968,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010728625839721275,
      "loss": 0.0112,
      "step": 10218
    },
    {
      "epoch": 4.769661610268378,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010727162383836193,
      "loss": 0.017,
      "step": 10219
    },
    {
      "epoch": 4.770128354725788,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010725698912294497,
      "loss": 0.0114,
      "step": 10220
    },
    {
      "epoch": 4.770595099183197,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00010724235425127698,
      "loss": 0.0134,
      "step": 10221
    },
    {
      "epoch": 4.771061843640607,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010722771922367303,
      "loss": 0.0113,
      "step": 10222
    },
    {
      "epoch": 4.7715285880980165,
      "grad_norm": 0.7265625,
      "learning_rate": 0.0001072130840404483,
      "loss": 0.0203,
      "step": 10223
    },
    {
      "epoch": 4.771995332555425,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010719844870191789,
      "loss": 0.0094,
      "step": 10224
    },
    {
      "epoch": 4.772462077012835,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010718381320839683,
      "loss": 0.0218,
      "step": 10225
    },
    {
      "epoch": 4.772928821470245,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010716917756020038,
      "loss": 0.0161,
      "step": 10226
    },
    {
      "epoch": 4.773395565927655,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010715454175764355,
      "loss": 0.0072,
      "step": 10227
    },
    {
      "epoch": 4.773862310385065,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010713990580104145,
      "loss": 0.0107,
      "step": 10228
    },
    {
      "epoch": 4.7743290548424735,
      "grad_norm": 0.44921875,
      "learning_rate": 0.0001071252696907093,
      "loss": 0.0076,
      "step": 10229
    },
    {
      "epoch": 4.774795799299883,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010711063342696219,
      "loss": 0.0136,
      "step": 10230
    },
    {
      "epoch": 4.775262543757293,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010709599701011523,
      "loss": 0.0111,
      "step": 10231
    },
    {
      "epoch": 4.775729288214702,
      "grad_norm": 0.55859375,
      "learning_rate": 0.0001070813604404836,
      "loss": 0.0118,
      "step": 10232
    },
    {
      "epoch": 4.776196032672112,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010706672371838242,
      "loss": 0.0121,
      "step": 10233
    },
    {
      "epoch": 4.776662777129522,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00010705208684412684,
      "loss": 0.008,
      "step": 10234
    },
    {
      "epoch": 4.777129521586931,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010703744981803201,
      "loss": 0.0185,
      "step": 10235
    },
    {
      "epoch": 4.777596266044341,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010702281264041307,
      "loss": 0.0107,
      "step": 10236
    },
    {
      "epoch": 4.77806301050175,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010700817531158518,
      "loss": 0.0105,
      "step": 10237
    },
    {
      "epoch": 4.77852975495916,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00010699353783186347,
      "loss": 0.0119,
      "step": 10238
    },
    {
      "epoch": 4.77899649941657,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00010697890020156319,
      "loss": 0.0067,
      "step": 10239
    },
    {
      "epoch": 4.779463243873979,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010696426242099941,
      "loss": 0.0111,
      "step": 10240
    },
    {
      "epoch": 4.779929988331388,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010694962449048733,
      "loss": 0.0174,
      "step": 10241
    },
    {
      "epoch": 4.780396732788798,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010693498641034214,
      "loss": 0.0109,
      "step": 10242
    },
    {
      "epoch": 4.780863477246208,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010692034818087901,
      "loss": 0.0106,
      "step": 10243
    },
    {
      "epoch": 4.781330221703618,
      "grad_norm": 0.373046875,
      "learning_rate": 0.0001069057098024131,
      "loss": 0.0062,
      "step": 10244
    },
    {
      "epoch": 4.781796966161027,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00010689107127525954,
      "loss": 0.0094,
      "step": 10245
    },
    {
      "epoch": 4.7822637106184365,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010687643259973363,
      "loss": 0.0184,
      "step": 10246
    },
    {
      "epoch": 4.782730455075846,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00010686179377615049,
      "loss": 0.0121,
      "step": 10247
    },
    {
      "epoch": 4.783197199533255,
      "grad_norm": 0.64453125,
      "learning_rate": 0.0001068471548048253,
      "loss": 0.0106,
      "step": 10248
    },
    {
      "epoch": 4.783663943990665,
      "grad_norm": 0.8046875,
      "learning_rate": 0.00010683251568607328,
      "loss": 0.02,
      "step": 10249
    },
    {
      "epoch": 4.784130688448075,
      "grad_norm": 0.7421875,
      "learning_rate": 0.00010681787642020964,
      "loss": 0.017,
      "step": 10250
    },
    {
      "epoch": 4.784597432905485,
      "grad_norm": 0.5625,
      "learning_rate": 0.0001068032370075495,
      "loss": 0.0127,
      "step": 10251
    },
    {
      "epoch": 4.7850641773628935,
      "grad_norm": 0.484375,
      "learning_rate": 0.00010678859744840816,
      "loss": 0.0065,
      "step": 10252
    },
    {
      "epoch": 4.785530921820303,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001067739577431008,
      "loss": 0.0096,
      "step": 10253
    },
    {
      "epoch": 4.785997666277713,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010675931789194256,
      "loss": 0.0121,
      "step": 10254
    },
    {
      "epoch": 4.786464410735123,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010674467789524873,
      "loss": 0.0087,
      "step": 10255
    },
    {
      "epoch": 4.786931155192532,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00010673003775333452,
      "loss": 0.0111,
      "step": 10256
    },
    {
      "epoch": 4.787397899649942,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010671539746651512,
      "loss": 0.008,
      "step": 10257
    },
    {
      "epoch": 4.787864644107351,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010670075703510574,
      "loss": 0.013,
      "step": 10258
    },
    {
      "epoch": 4.788331388564761,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010668611645942166,
      "loss": 0.0159,
      "step": 10259
    },
    {
      "epoch": 4.78879813302217,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010667147573977805,
      "loss": 0.0098,
      "step": 10260
    },
    {
      "epoch": 4.78926487747958,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010665683487649017,
      "loss": 0.0189,
      "step": 10261
    },
    {
      "epoch": 4.78973162193699,
      "grad_norm": 0.5,
      "learning_rate": 0.00010664219386987324,
      "loss": 0.0071,
      "step": 10262
    },
    {
      "epoch": 4.790198366394399,
      "grad_norm": 0.2890625,
      "learning_rate": 0.00010662755272024253,
      "loss": 0.0108,
      "step": 10263
    },
    {
      "epoch": 4.790665110851808,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010661291142791323,
      "loss": 0.0102,
      "step": 10264
    },
    {
      "epoch": 4.791131855309218,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00010659826999320063,
      "loss": 0.0079,
      "step": 10265
    },
    {
      "epoch": 4.791598599766628,
      "grad_norm": 0.83203125,
      "learning_rate": 0.00010658362841641995,
      "loss": 0.0112,
      "step": 10266
    },
    {
      "epoch": 4.792065344224037,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00010656898669788644,
      "loss": 0.0072,
      "step": 10267
    },
    {
      "epoch": 4.792532088681447,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010655434483791538,
      "loss": 0.0153,
      "step": 10268
    },
    {
      "epoch": 4.792998833138856,
      "grad_norm": 0.28125,
      "learning_rate": 0.00010653970283682198,
      "loss": 0.0058,
      "step": 10269
    },
    {
      "epoch": 4.793465577596266,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00010652506069492154,
      "loss": 0.0085,
      "step": 10270
    },
    {
      "epoch": 4.793932322053676,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00010651041841252928,
      "loss": 0.0084,
      "step": 10271
    },
    {
      "epoch": 4.794399066511085,
      "grad_norm": 0.43359375,
      "learning_rate": 0.0001064957759899605,
      "loss": 0.0081,
      "step": 10272
    },
    {
      "epoch": 4.794865810968495,
      "grad_norm": 0.5,
      "learning_rate": 0.00010648113342753047,
      "loss": 0.0094,
      "step": 10273
    },
    {
      "epoch": 4.7953325554259045,
      "grad_norm": 0.81640625,
      "learning_rate": 0.00010646649072555444,
      "loss": 0.0165,
      "step": 10274
    },
    {
      "epoch": 4.795799299883313,
      "grad_norm": 0.490234375,
      "learning_rate": 0.0001064518478843477,
      "loss": 0.0115,
      "step": 10275
    },
    {
      "epoch": 4.796266044340723,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00010643720490422549,
      "loss": 0.0096,
      "step": 10276
    },
    {
      "epoch": 4.796732788798133,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00010642256178550312,
      "loss": 0.0098,
      "step": 10277
    },
    {
      "epoch": 4.797199533255543,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00010640791852849587,
      "loss": 0.0051,
      "step": 10278
    },
    {
      "epoch": 4.797666277712953,
      "grad_norm": 0.34375,
      "learning_rate": 0.00010639327513351905,
      "loss": 0.0051,
      "step": 10279
    },
    {
      "epoch": 4.7981330221703615,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010637863160088787,
      "loss": 0.0116,
      "step": 10280
    },
    {
      "epoch": 4.798599766627771,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00010636398793091773,
      "loss": 0.0098,
      "step": 10281
    },
    {
      "epoch": 4.799066511085181,
      "grad_norm": 0.8125,
      "learning_rate": 0.00010634934412392384,
      "loss": 0.0155,
      "step": 10282
    },
    {
      "epoch": 4.79953325554259,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00010633470018022152,
      "loss": 0.0082,
      "step": 10283
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001063200561001261,
      "loss": 0.0076,
      "step": 10284
    },
    {
      "epoch": 4.80046674445741,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010630541188395284,
      "loss": 0.0143,
      "step": 10285
    },
    {
      "epoch": 4.800933488914819,
      "grad_norm": 0.48046875,
      "learning_rate": 0.0001062907675320171,
      "loss": 0.0099,
      "step": 10286
    },
    {
      "epoch": 4.801400233372228,
      "grad_norm": 0.458984375,
      "learning_rate": 0.0001062761230446341,
      "loss": 0.013,
      "step": 10287
    },
    {
      "epoch": 4.801866977829638,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010626147842211924,
      "loss": 0.0061,
      "step": 10288
    },
    {
      "epoch": 4.802333722287048,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010624683366478781,
      "loss": 0.0058,
      "step": 10289
    },
    {
      "epoch": 4.802800466744458,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00010623218877295511,
      "loss": 0.0082,
      "step": 10290
    },
    {
      "epoch": 4.803267211201867,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00010621754374693646,
      "loss": 0.004,
      "step": 10291
    },
    {
      "epoch": 4.803733955659276,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00010620289858704722,
      "loss": 0.0109,
      "step": 10292
    },
    {
      "epoch": 4.804200700116686,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00010618825329360267,
      "loss": 0.0087,
      "step": 10293
    },
    {
      "epoch": 4.804667444574096,
      "grad_norm": 0.80859375,
      "learning_rate": 0.00010617360786691817,
      "loss": 0.0183,
      "step": 10294
    },
    {
      "epoch": 4.805134189031505,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00010615896230730904,
      "loss": 0.0105,
      "step": 10295
    },
    {
      "epoch": 4.805600933488915,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00010614431661509064,
      "loss": 0.0056,
      "step": 10296
    },
    {
      "epoch": 4.8060676779463245,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010612967079057825,
      "loss": 0.0109,
      "step": 10297
    },
    {
      "epoch": 4.806534422403734,
      "grad_norm": 0.73046875,
      "learning_rate": 0.00010611502483408726,
      "loss": 0.0093,
      "step": 10298
    },
    {
      "epoch": 4.807001166861143,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010610037874593299,
      "loss": 0.012,
      "step": 10299
    },
    {
      "epoch": 4.807467911318553,
      "grad_norm": 0.5,
      "learning_rate": 0.00010608573252643081,
      "loss": 0.0073,
      "step": 10300
    },
    {
      "epoch": 4.807934655775963,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00010607108617589607,
      "loss": 0.0047,
      "step": 10301
    },
    {
      "epoch": 4.808401400233373,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001060564396946441,
      "loss": 0.0096,
      "step": 10302
    },
    {
      "epoch": 4.8088681446907815,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00010604179308299025,
      "loss": 0.0091,
      "step": 10303
    },
    {
      "epoch": 4.809334889148191,
      "grad_norm": 0.69140625,
      "learning_rate": 0.0001060271463412499,
      "loss": 0.0137,
      "step": 10304
    },
    {
      "epoch": 4.809801633605601,
      "grad_norm": 0.625,
      "learning_rate": 0.00010601249946973841,
      "loss": 0.0121,
      "step": 10305
    },
    {
      "epoch": 4.810268378063011,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00010599785246877112,
      "loss": 0.0099,
      "step": 10306
    },
    {
      "epoch": 4.81073512252042,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00010598320533866342,
      "loss": 0.0089,
      "step": 10307
    },
    {
      "epoch": 4.81120186697783,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00010596855807973067,
      "loss": 0.0058,
      "step": 10308
    },
    {
      "epoch": 4.811668611435239,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010595391069228824,
      "loss": 0.0151,
      "step": 10309
    },
    {
      "epoch": 4.812135355892648,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010593926317665151,
      "loss": 0.0055,
      "step": 10310
    },
    {
      "epoch": 4.812602100350058,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010592461553313584,
      "loss": 0.0168,
      "step": 10311
    },
    {
      "epoch": 4.813068844807468,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010590996776205664,
      "loss": 0.0097,
      "step": 10312
    },
    {
      "epoch": 4.813535589264878,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010589531986372927,
      "loss": 0.0105,
      "step": 10313
    },
    {
      "epoch": 4.8140023337222875,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010588067183846913,
      "loss": 0.008,
      "step": 10314
    },
    {
      "epoch": 4.814469078179696,
      "grad_norm": 0.416015625,
      "learning_rate": 0.0001058660236865916,
      "loss": 0.0119,
      "step": 10315
    },
    {
      "epoch": 4.814935822637106,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00010585137540841205,
      "loss": 0.0046,
      "step": 10316
    },
    {
      "epoch": 4.815402567094516,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010583672700424592,
      "loss": 0.0071,
      "step": 10317
    },
    {
      "epoch": 4.815869311551925,
      "grad_norm": 0.46875,
      "learning_rate": 0.00010582207847440858,
      "loss": 0.0079,
      "step": 10318
    },
    {
      "epoch": 4.816336056009335,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010580742981921543,
      "loss": 0.0132,
      "step": 10319
    },
    {
      "epoch": 4.8168028004667445,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00010579278103898184,
      "loss": 0.0093,
      "step": 10320
    },
    {
      "epoch": 4.817269544924154,
      "grad_norm": 0.625,
      "learning_rate": 0.00010577813213402332,
      "loss": 0.0093,
      "step": 10321
    },
    {
      "epoch": 4.817736289381564,
      "grad_norm": 0.859375,
      "learning_rate": 0.00010576348310465514,
      "loss": 0.0174,
      "step": 10322
    },
    {
      "epoch": 4.818203033838973,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00010574883395119279,
      "loss": 0.0066,
      "step": 10323
    },
    {
      "epoch": 4.818669778296383,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00010573418467395168,
      "loss": 0.0075,
      "step": 10324
    },
    {
      "epoch": 4.819136522753793,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001057195352732472,
      "loss": 0.0174,
      "step": 10325
    },
    {
      "epoch": 4.8196032672112015,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00010570488574939479,
      "loss": 0.0056,
      "step": 10326
    },
    {
      "epoch": 4.820070011668611,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00010569023610270985,
      "loss": 0.0102,
      "step": 10327
    },
    {
      "epoch": 4.820536756126021,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00010567558633350782,
      "loss": 0.0084,
      "step": 10328
    },
    {
      "epoch": 4.821003500583431,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00010566093644210409,
      "loss": 0.0095,
      "step": 10329
    },
    {
      "epoch": 4.82147024504084,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010564628642881415,
      "loss": 0.0095,
      "step": 10330
    },
    {
      "epoch": 4.82193698949825,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00010563163629395339,
      "loss": 0.0099,
      "step": 10331
    },
    {
      "epoch": 4.822403733955659,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010561698603783725,
      "loss": 0.0126,
      "step": 10332
    },
    {
      "epoch": 4.822870478413069,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010560233566078118,
      "loss": 0.0078,
      "step": 10333
    },
    {
      "epoch": 4.823337222870478,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010558768516310058,
      "loss": 0.0167,
      "step": 10334
    },
    {
      "epoch": 4.823803967327888,
      "grad_norm": 0.34765625,
      "learning_rate": 0.00010557303454511097,
      "loss": 0.0055,
      "step": 10335
    },
    {
      "epoch": 4.824270711785298,
      "grad_norm": 0.32421875,
      "learning_rate": 0.00010555838380712768,
      "loss": 0.0034,
      "step": 10336
    },
    {
      "epoch": 4.8247374562427074,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010554373294946625,
      "loss": 0.0192,
      "step": 10337
    },
    {
      "epoch": 4.825204200700116,
      "grad_norm": 0.486328125,
      "learning_rate": 0.00010552908197244211,
      "loss": 0.0131,
      "step": 10338
    },
    {
      "epoch": 4.825670945157526,
      "grad_norm": 0.75390625,
      "learning_rate": 0.00010551443087637067,
      "loss": 0.0159,
      "step": 10339
    },
    {
      "epoch": 4.826137689614936,
      "grad_norm": 0.515625,
      "learning_rate": 0.00010549977966156743,
      "loss": 0.013,
      "step": 10340
    },
    {
      "epoch": 4.826604434072346,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010548512832834785,
      "loss": 0.0089,
      "step": 10341
    },
    {
      "epoch": 4.827071178529755,
      "grad_norm": 0.53125,
      "learning_rate": 0.00010547047687702733,
      "loss": 0.0086,
      "step": 10342
    },
    {
      "epoch": 4.827537922987164,
      "grad_norm": 0.6640625,
      "learning_rate": 0.00010545582530792142,
      "loss": 0.011,
      "step": 10343
    },
    {
      "epoch": 4.828004667444574,
      "grad_norm": 0.43359375,
      "learning_rate": 0.00010544117362134551,
      "loss": 0.007,
      "step": 10344
    },
    {
      "epoch": 4.828471411901984,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010542652181761512,
      "loss": 0.017,
      "step": 10345
    },
    {
      "epoch": 4.828938156359393,
      "grad_norm": 0.5,
      "learning_rate": 0.00010541186989704568,
      "loss": 0.0085,
      "step": 10346
    },
    {
      "epoch": 4.829404900816803,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00010539721785995267,
      "loss": 0.0092,
      "step": 10347
    },
    {
      "epoch": 4.8298716452742125,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00010538256570665157,
      "loss": 0.0057,
      "step": 10348
    },
    {
      "epoch": 4.830338389731622,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00010536791343745787,
      "loss": 0.0136,
      "step": 10349
    },
    {
      "epoch": 4.830805134189031,
      "grad_norm": 0.5,
      "learning_rate": 0.00010535326105268701,
      "loss": 0.0124,
      "step": 10350
    },
    {
      "epoch": 4.831271878646441,
      "grad_norm": 0.53515625,
      "learning_rate": 0.00010533860855265454,
      "loss": 0.0203,
      "step": 10351
    },
    {
      "epoch": 4.831738623103851,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010532395593767588,
      "loss": 0.0083,
      "step": 10352
    },
    {
      "epoch": 4.83220536756126,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00010530930320806653,
      "loss": 0.0063,
      "step": 10353
    },
    {
      "epoch": 4.8326721120186695,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00010529465036414203,
      "loss": 0.0088,
      "step": 10354
    },
    {
      "epoch": 4.833138856476079,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010527999740621779,
      "loss": 0.0102,
      "step": 10355
    },
    {
      "epoch": 4.833605600933489,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010526534433460937,
      "loss": 0.0093,
      "step": 10356
    },
    {
      "epoch": 4.834072345390899,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00010525069114963225,
      "loss": 0.006,
      "step": 10357
    },
    {
      "epoch": 4.834539089848308,
      "grad_norm": 0.486328125,
      "learning_rate": 0.0001052360378516019,
      "loss": 0.0092,
      "step": 10358
    },
    {
      "epoch": 4.835005834305718,
      "grad_norm": 0.578125,
      "learning_rate": 0.00010522138444083386,
      "loss": 0.0099,
      "step": 10359
    },
    {
      "epoch": 4.835472578763127,
      "grad_norm": 0.71484375,
      "learning_rate": 0.00010520673091764364,
      "loss": 0.0104,
      "step": 10360
    },
    {
      "epoch": 4.835939323220536,
      "grad_norm": 0.73828125,
      "learning_rate": 0.0001051920772823467,
      "loss": 0.0144,
      "step": 10361
    },
    {
      "epoch": 4.836406067677946,
      "grad_norm": 0.3984375,
      "learning_rate": 0.0001051774235352586,
      "loss": 0.0037,
      "step": 10362
    },
    {
      "epoch": 4.836872812135356,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00010516276967669481,
      "loss": 0.0122,
      "step": 10363
    },
    {
      "epoch": 4.837339556592766,
      "grad_norm": 0.419921875,
      "learning_rate": 0.00010514811570697085,
      "loss": 0.0071,
      "step": 10364
    },
    {
      "epoch": 4.8378063010501755,
      "grad_norm": 0.34375,
      "learning_rate": 0.00010513346162640225,
      "loss": 0.0048,
      "step": 10365
    },
    {
      "epoch": 4.838273045507584,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010511880743530454,
      "loss": 0.0128,
      "step": 10366
    },
    {
      "epoch": 4.838739789964994,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00010510415313399323,
      "loss": 0.0084,
      "step": 10367
    },
    {
      "epoch": 4.839206534422404,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00010508949872278382,
      "loss": 0.0172,
      "step": 10368
    },
    {
      "epoch": 4.839673278879813,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00010507484420199187,
      "loss": 0.0079,
      "step": 10369
    },
    {
      "epoch": 4.840140023337223,
      "grad_norm": 0.490234375,
      "learning_rate": 0.00010506018957193291,
      "loss": 0.0054,
      "step": 10370
    },
    {
      "epoch": 4.8406067677946325,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00010504553483292241,
      "loss": 0.0053,
      "step": 10371
    },
    {
      "epoch": 4.841073512252042,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010503087998527596,
      "loss": 0.0094,
      "step": 10372
    },
    {
      "epoch": 4.841540256709451,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010501622502930912,
      "loss": 0.0126,
      "step": 10373
    },
    {
      "epoch": 4.842007001166861,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010500156996533737,
      "loss": 0.0073,
      "step": 10374
    },
    {
      "epoch": 4.842473745624271,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00010498691479367624,
      "loss": 0.0119,
      "step": 10375
    },
    {
      "epoch": 4.842940490081681,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010497225951464131,
      "loss": 0.0159,
      "step": 10376
    },
    {
      "epoch": 4.8434072345390895,
      "grad_norm": 0.2421875,
      "learning_rate": 0.00010495760412854816,
      "loss": 0.0027,
      "step": 10377
    },
    {
      "epoch": 4.843873978996499,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010494294863571223,
      "loss": 0.0106,
      "step": 10378
    },
    {
      "epoch": 4.844340723453909,
      "grad_norm": 0.4375,
      "learning_rate": 0.00010492829303644917,
      "loss": 0.0098,
      "step": 10379
    },
    {
      "epoch": 4.844807467911319,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00010491363733107449,
      "loss": 0.0076,
      "step": 10380
    },
    {
      "epoch": 4.845274212368728,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010489898151990374,
      "loss": 0.0106,
      "step": 10381
    },
    {
      "epoch": 4.845740956826138,
      "grad_norm": 0.63671875,
      "learning_rate": 0.00010488432560325247,
      "loss": 0.0134,
      "step": 10382
    },
    {
      "epoch": 4.846207701283547,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00010486966958143626,
      "loss": 0.0076,
      "step": 10383
    },
    {
      "epoch": 4.846674445740957,
      "grad_norm": 0.5625,
      "learning_rate": 0.00010485501345477069,
      "loss": 0.0078,
      "step": 10384
    },
    {
      "epoch": 4.847141190198366,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010484035722357126,
      "loss": 0.0086,
      "step": 10385
    },
    {
      "epoch": 4.847607934655776,
      "grad_norm": 0.6484375,
      "learning_rate": 0.00010482570088815357,
      "loss": 0.0076,
      "step": 10386
    },
    {
      "epoch": 4.848074679113186,
      "grad_norm": 0.671875,
      "learning_rate": 0.0001048110444488332,
      "loss": 0.0071,
      "step": 10387
    },
    {
      "epoch": 4.8485414235705955,
      "grad_norm": 0.7578125,
      "learning_rate": 0.0001047963879059257,
      "loss": 0.01,
      "step": 10388
    },
    {
      "epoch": 4.849008168028004,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010478173125974664,
      "loss": 0.0082,
      "step": 10389
    },
    {
      "epoch": 4.849474912485414,
      "grad_norm": 0.70703125,
      "learning_rate": 0.00010476707451061161,
      "loss": 0.0156,
      "step": 10390
    },
    {
      "epoch": 4.849941656942824,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00010475241765883616,
      "loss": 0.0096,
      "step": 10391
    },
    {
      "epoch": 4.850408401400234,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00010473776070473591,
      "loss": 0.0038,
      "step": 10392
    },
    {
      "epoch": 4.850875145857643,
      "grad_norm": 0.734375,
      "learning_rate": 0.00010472310364862643,
      "loss": 0.022,
      "step": 10393
    },
    {
      "epoch": 4.8513418903150525,
      "grad_norm": 0.6953125,
      "learning_rate": 0.00010470844649082324,
      "loss": 0.01,
      "step": 10394
    },
    {
      "epoch": 4.851808634772462,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00010469378923164199,
      "loss": 0.0121,
      "step": 10395
    },
    {
      "epoch": 4.852275379229871,
      "grad_norm": 0.3125,
      "learning_rate": 0.00010467913187139826,
      "loss": 0.0045,
      "step": 10396
    },
    {
      "epoch": 4.852742123687281,
      "grad_norm": 0.306640625,
      "learning_rate": 0.0001046644744104076,
      "loss": 0.0059,
      "step": 10397
    },
    {
      "epoch": 4.853208868144691,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010464981684898568,
      "loss": 0.009,
      "step": 10398
    },
    {
      "epoch": 4.853675612602101,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010463515918744802,
      "loss": 0.0106,
      "step": 10399
    },
    {
      "epoch": 4.85414235705951,
      "grad_norm": 0.79296875,
      "learning_rate": 0.00010462050142611024,
      "loss": 0.0151,
      "step": 10400
    },
    {
      "epoch": 4.854609101516919,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00010460584356528791,
      "loss": 0.0065,
      "step": 10401
    },
    {
      "epoch": 4.855075845974329,
      "grad_norm": 0.9296875,
      "learning_rate": 0.0001045911856052967,
      "loss": 0.0077,
      "step": 10402
    },
    {
      "epoch": 4.855542590431739,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00010457652754645216,
      "loss": 0.007,
      "step": 10403
    },
    {
      "epoch": 4.856009334889148,
      "grad_norm": 0.38671875,
      "learning_rate": 0.0001045618693890699,
      "loss": 0.0069,
      "step": 10404
    },
    {
      "epoch": 4.856476079346558,
      "grad_norm": 0.484375,
      "learning_rate": 0.0001045472111334655,
      "loss": 0.01,
      "step": 10405
    },
    {
      "epoch": 4.856942823803967,
      "grad_norm": 0.62109375,
      "learning_rate": 0.00010453255277995464,
      "loss": 0.0108,
      "step": 10406
    },
    {
      "epoch": 4.857409568261377,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00010451789432885287,
      "loss": 0.0076,
      "step": 10407
    },
    {
      "epoch": 4.857876312718787,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00010450323578047583,
      "loss": 0.006,
      "step": 10408
    },
    {
      "epoch": 4.858343057176196,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010448857713513914,
      "loss": 0.0072,
      "step": 10409
    },
    {
      "epoch": 4.858809801633606,
      "grad_norm": 0.57421875,
      "learning_rate": 0.00010447391839315838,
      "loss": 0.0125,
      "step": 10410
    },
    {
      "epoch": 4.8592765460910154,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00010445925955484918,
      "loss": 0.01,
      "step": 10411
    },
    {
      "epoch": 4.859743290548424,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010444460062052721,
      "loss": 0.0086,
      "step": 10412
    },
    {
      "epoch": 4.860210035005834,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00010442994159050805,
      "loss": 0.007,
      "step": 10413
    },
    {
      "epoch": 4.860676779463244,
      "grad_norm": 0.74609375,
      "learning_rate": 0.00010441528246510732,
      "loss": 0.0106,
      "step": 10414
    },
    {
      "epoch": 4.861143523920654,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010440062324464069,
      "loss": 0.0152,
      "step": 10415
    },
    {
      "epoch": 4.861610268378063,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010438596392942372,
      "loss": 0.0161,
      "step": 10416
    },
    {
      "epoch": 4.8620770128354724,
      "grad_norm": 0.51171875,
      "learning_rate": 0.0001043713045197721,
      "loss": 0.0072,
      "step": 10417
    },
    {
      "epoch": 4.862543757292882,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00010435664501600144,
      "loss": 0.0087,
      "step": 10418
    },
    {
      "epoch": 4.863010501750292,
      "grad_norm": 0.59375,
      "learning_rate": 0.0001043419854184274,
      "loss": 0.0135,
      "step": 10419
    },
    {
      "epoch": 4.863477246207701,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010432732572736556,
      "loss": 0.017,
      "step": 10420
    },
    {
      "epoch": 4.863943990665111,
      "grad_norm": 0.40625,
      "learning_rate": 0.0001043126659431316,
      "loss": 0.0078,
      "step": 10421
    },
    {
      "epoch": 4.8644107351225205,
      "grad_norm": 0.57421875,
      "learning_rate": 0.0001042980060660412,
      "loss": 0.0115,
      "step": 10422
    },
    {
      "epoch": 4.86487747957993,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00010428334609640993,
      "loss": 0.0086,
      "step": 10423
    },
    {
      "epoch": 4.865344224037339,
      "grad_norm": 0.35546875,
      "learning_rate": 0.00010426868603455344,
      "loss": 0.0078,
      "step": 10424
    },
    {
      "epoch": 4.865810968494749,
      "grad_norm": 0.8203125,
      "learning_rate": 0.00010425402588078743,
      "loss": 0.0125,
      "step": 10425
    },
    {
      "epoch": 4.866277712952159,
      "grad_norm": 0.625,
      "learning_rate": 0.00010423936563542754,
      "loss": 0.0103,
      "step": 10426
    },
    {
      "epoch": 4.866744457409569,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00010422470529878937,
      "loss": 0.0067,
      "step": 10427
    },
    {
      "epoch": 4.8672112018669775,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00010421004487118861,
      "loss": 0.0049,
      "step": 10428
    },
    {
      "epoch": 4.867677946324387,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010419538435294094,
      "loss": 0.0155,
      "step": 10429
    },
    {
      "epoch": 4.868144690781797,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00010418072374436196,
      "loss": 0.0059,
      "step": 10430
    },
    {
      "epoch": 4.868611435239207,
      "grad_norm": 0.35546875,
      "learning_rate": 0.0001041660630457674,
      "loss": 0.0074,
      "step": 10431
    },
    {
      "epoch": 4.869078179696616,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010415140225747284,
      "loss": 0.0109,
      "step": 10432
    },
    {
      "epoch": 4.869544924154026,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00010413674137979402,
      "loss": 0.0142,
      "step": 10433
    },
    {
      "epoch": 4.870011668611435,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00010412208041304652,
      "loss": 0.0055,
      "step": 10434
    },
    {
      "epoch": 4.870478413068845,
      "grad_norm": 0.828125,
      "learning_rate": 0.00010410741935754608,
      "loss": 0.0144,
      "step": 10435
    },
    {
      "epoch": 4.870945157526254,
      "grad_norm": 0.46875,
      "learning_rate": 0.00010409275821360836,
      "loss": 0.0104,
      "step": 10436
    },
    {
      "epoch": 4.871411901983664,
      "grad_norm": 0.47265625,
      "learning_rate": 0.000104078096981549,
      "loss": 0.008,
      "step": 10437
    },
    {
      "epoch": 4.871878646441074,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010406343566168369,
      "loss": 0.0093,
      "step": 10438
    },
    {
      "epoch": 4.872345390898483,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010404877425432811,
      "loss": 0.0087,
      "step": 10439
    },
    {
      "epoch": 4.872812135355892,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00010403411275979789,
      "loss": 0.0105,
      "step": 10440
    },
    {
      "epoch": 4.873278879813302,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00010401945117840881,
      "loss": 0.0098,
      "step": 10441
    },
    {
      "epoch": 4.873745624270712,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00010400478951047645,
      "loss": 0.0152,
      "step": 10442
    },
    {
      "epoch": 4.874212368728122,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010399012775631651,
      "loss": 0.0095,
      "step": 10443
    },
    {
      "epoch": 4.874679113185531,
      "grad_norm": 0.48046875,
      "learning_rate": 0.00010397546591624472,
      "loss": 0.0102,
      "step": 10444
    },
    {
      "epoch": 4.8751458576429405,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00010396080399057672,
      "loss": 0.0088,
      "step": 10445
    },
    {
      "epoch": 4.87561260210035,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010394614197962822,
      "loss": 0.0075,
      "step": 10446
    },
    {
      "epoch": 4.876079346557759,
      "grad_norm": 0.56640625,
      "learning_rate": 0.00010393147988371492,
      "loss": 0.0108,
      "step": 10447
    },
    {
      "epoch": 4.876546091015169,
      "grad_norm": 0.251953125,
      "learning_rate": 0.00010391681770315249,
      "loss": 0.0073,
      "step": 10448
    },
    {
      "epoch": 4.877012835472579,
      "grad_norm": 0.69140625,
      "learning_rate": 0.00010390215543825664,
      "loss": 0.0135,
      "step": 10449
    },
    {
      "epoch": 4.877479579929989,
      "grad_norm": 0.349609375,
      "learning_rate": 0.00010388749308934303,
      "loss": 0.0041,
      "step": 10450
    },
    {
      "epoch": 4.877946324387398,
      "grad_norm": 0.69921875,
      "learning_rate": 0.0001038728306567274,
      "loss": 0.0095,
      "step": 10451
    },
    {
      "epoch": 4.878413068844807,
      "grad_norm": 0.275390625,
      "learning_rate": 0.0001038581681407254,
      "loss": 0.004,
      "step": 10452
    },
    {
      "epoch": 4.878879813302217,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00010384350554165276,
      "loss": 0.0062,
      "step": 10453
    },
    {
      "epoch": 4.879346557759627,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00010382884285982521,
      "loss": 0.0083,
      "step": 10454
    },
    {
      "epoch": 4.879813302217036,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00010381418009555843,
      "loss": 0.0056,
      "step": 10455
    },
    {
      "epoch": 4.880280046674446,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010379951724916807,
      "loss": 0.0125,
      "step": 10456
    },
    {
      "epoch": 4.880746791131855,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00010378485432096993,
      "loss": 0.0089,
      "step": 10457
    },
    {
      "epoch": 4.881213535589265,
      "grad_norm": 0.60546875,
      "learning_rate": 0.0001037701913112797,
      "loss": 0.0165,
      "step": 10458
    },
    {
      "epoch": 4.881680280046674,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010375552822041303,
      "loss": 0.0053,
      "step": 10459
    },
    {
      "epoch": 4.882147024504084,
      "grad_norm": 0.337890625,
      "learning_rate": 0.00010374086504868566,
      "loss": 0.0076,
      "step": 10460
    },
    {
      "epoch": 4.882613768961494,
      "grad_norm": 0.38671875,
      "learning_rate": 0.00010372620179641335,
      "loss": 0.0059,
      "step": 10461
    },
    {
      "epoch": 4.8830805134189035,
      "grad_norm": 0.59375,
      "learning_rate": 0.00010371153846391178,
      "loss": 0.0088,
      "step": 10462
    },
    {
      "epoch": 4.883547257876312,
      "grad_norm": 0.625,
      "learning_rate": 0.00010369687505149661,
      "loss": 0.0106,
      "step": 10463
    },
    {
      "epoch": 4.884014002333722,
      "grad_norm": 0.470703125,
      "learning_rate": 0.00010368221155948368,
      "loss": 0.0057,
      "step": 10464
    },
    {
      "epoch": 4.884480746791132,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010366754798818865,
      "loss": 0.0056,
      "step": 10465
    },
    {
      "epoch": 4.884947491248542,
      "grad_norm": 0.451171875,
      "learning_rate": 0.00010365288433792722,
      "loss": 0.0059,
      "step": 10466
    },
    {
      "epoch": 4.885414235705951,
      "grad_norm": 0.8828125,
      "learning_rate": 0.00010363822060901513,
      "loss": 0.0171,
      "step": 10467
    },
    {
      "epoch": 4.8858809801633605,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00010362355680176814,
      "loss": 0.0101,
      "step": 10468
    },
    {
      "epoch": 4.88634772462077,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010360889291650193,
      "loss": 0.005,
      "step": 10469
    },
    {
      "epoch": 4.88681446907818,
      "grad_norm": 0.466796875,
      "learning_rate": 0.00010359422895353225,
      "loss": 0.0063,
      "step": 10470
    },
    {
      "epoch": 4.887281213535589,
      "grad_norm": 0.421875,
      "learning_rate": 0.00010357956491317484,
      "loss": 0.0064,
      "step": 10471
    },
    {
      "epoch": 4.887747957992999,
      "grad_norm": 0.61328125,
      "learning_rate": 0.00010356490079574541,
      "loss": 0.0111,
      "step": 10472
    },
    {
      "epoch": 4.888214702450409,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00010355023660155974,
      "loss": 0.0074,
      "step": 10473
    },
    {
      "epoch": 4.8886814469078175,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010353557233093355,
      "loss": 0.0037,
      "step": 10474
    },
    {
      "epoch": 4.889148191365227,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00010352090798418252,
      "loss": 0.0115,
      "step": 10475
    },
    {
      "epoch": 4.889614935822637,
      "grad_norm": 0.2353515625,
      "learning_rate": 0.00010350624356162244,
      "loss": 0.0036,
      "step": 10476
    },
    {
      "epoch": 4.890081680280047,
      "grad_norm": 0.46875,
      "learning_rate": 0.00010349157906356908,
      "loss": 0.0097,
      "step": 10477
    },
    {
      "epoch": 4.890548424737457,
      "grad_norm": 0.30859375,
      "learning_rate": 0.00010347691449033812,
      "loss": 0.0041,
      "step": 10478
    },
    {
      "epoch": 4.891015169194866,
      "grad_norm": 0.76171875,
      "learning_rate": 0.00010346224984224533,
      "loss": 0.0083,
      "step": 10479
    },
    {
      "epoch": 4.891481913652275,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00010344758511960648,
      "loss": 0.0074,
      "step": 10480
    },
    {
      "epoch": 4.891948658109685,
      "grad_norm": 0.373046875,
      "learning_rate": 0.00010343292032273728,
      "loss": 0.0076,
      "step": 10481
    },
    {
      "epoch": 4.892415402567094,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00010341825545195351,
      "loss": 0.0062,
      "step": 10482
    },
    {
      "epoch": 4.892882147024504,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010340359050757089,
      "loss": 0.0056,
      "step": 10483
    },
    {
      "epoch": 4.893348891481914,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00010338892548990522,
      "loss": 0.0067,
      "step": 10484
    },
    {
      "epoch": 4.8938156359393234,
      "grad_norm": 0.443359375,
      "learning_rate": 0.0001033742603992722,
      "loss": 0.0061,
      "step": 10485
    },
    {
      "epoch": 4.894282380396733,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00010335959523598761,
      "loss": 0.012,
      "step": 10486
    },
    {
      "epoch": 4.894749124854142,
      "grad_norm": 0.6015625,
      "learning_rate": 0.00010334493000036722,
      "loss": 0.0097,
      "step": 10487
    },
    {
      "epoch": 4.895215869311552,
      "grad_norm": 0.240234375,
      "learning_rate": 0.00010333026469272679,
      "loss": 0.008,
      "step": 10488
    },
    {
      "epoch": 4.895682613768962,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00010331559931338204,
      "loss": 0.0078,
      "step": 10489
    },
    {
      "epoch": 4.896149358226371,
      "grad_norm": 0.59765625,
      "learning_rate": 0.00010330093386264877,
      "loss": 0.0071,
      "step": 10490
    },
    {
      "epoch": 4.8966161026837804,
      "grad_norm": 0.45703125,
      "learning_rate": 0.00010328626834084272,
      "loss": 0.0048,
      "step": 10491
    },
    {
      "epoch": 4.89708284714119,
      "grad_norm": 0.62890625,
      "learning_rate": 0.00010327160274827965,
      "loss": 0.0099,
      "step": 10492
    },
    {
      "epoch": 4.8975495915986,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00010325693708527537,
      "loss": 0.0175,
      "step": 10493
    },
    {
      "epoch": 4.89801633605601,
      "grad_norm": 0.431640625,
      "learning_rate": 0.00010324227135214561,
      "loss": 0.0039,
      "step": 10494
    },
    {
      "epoch": 4.898483080513419,
      "grad_norm": 0.380859375,
      "learning_rate": 0.00010322760554920616,
      "loss": 0.0073,
      "step": 10495
    },
    {
      "epoch": 4.8989498249708285,
      "grad_norm": 0.7890625,
      "learning_rate": 0.00010321293967677277,
      "loss": 0.0126,
      "step": 10496
    },
    {
      "epoch": 4.899416569428238,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00010319827373516124,
      "loss": 0.0066,
      "step": 10497
    },
    {
      "epoch": 4.899883313885647,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00010318360772468734,
      "loss": 0.0112,
      "step": 10498
    },
    {
      "epoch": 4.900350058343057,
      "grad_norm": 0.287109375,
      "learning_rate": 0.0001031689416456668,
      "loss": 0.0042,
      "step": 10499
    },
    {
      "epoch": 4.900816802800467,
      "grad_norm": 0.90234375,
      "learning_rate": 0.00010315427549841543,
      "loss": 0.0086,
      "step": 10500
    },
    {
      "epoch": 4.901283547257877,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00010313960928324903,
      "loss": 0.0103,
      "step": 10501
    },
    {
      "epoch": 4.9017502917152855,
      "grad_norm": 0.1923828125,
      "learning_rate": 0.00010312494300048335,
      "loss": 0.0022,
      "step": 10502
    },
    {
      "epoch": 4.902217036172695,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00010311027665043418,
      "loss": 0.0077,
      "step": 10503
    },
    {
      "epoch": 4.902683780630105,
      "grad_norm": 0.60546875,
      "learning_rate": 0.00010309561023341729,
      "loss": 0.0094,
      "step": 10504
    },
    {
      "epoch": 4.903150525087515,
      "grad_norm": 0.53125,
      "learning_rate": 0.00010308094374974848,
      "loss": 0.0108,
      "step": 10505
    },
    {
      "epoch": 4.903617269544924,
      "grad_norm": 0.31640625,
      "learning_rate": 0.00010306627719974355,
      "loss": 0.0038,
      "step": 10506
    },
    {
      "epoch": 4.904084014002334,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00010305161058371826,
      "loss": 0.0061,
      "step": 10507
    },
    {
      "epoch": 4.904550758459743,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00010303694390198841,
      "loss": 0.0151,
      "step": 10508
    },
    {
      "epoch": 4.905017502917153,
      "grad_norm": 0.71875,
      "learning_rate": 0.00010302227715486978,
      "loss": 0.0153,
      "step": 10509
    },
    {
      "epoch": 4.905484247374562,
      "grad_norm": 0.5234375,
      "learning_rate": 0.00010300761034267818,
      "loss": 0.0063,
      "step": 10510
    },
    {
      "epoch": 4.905950991831972,
      "grad_norm": 0.609375,
      "learning_rate": 0.0001029929434657294,
      "loss": 0.0118,
      "step": 10511
    },
    {
      "epoch": 4.906417736289382,
      "grad_norm": 0.330078125,
      "learning_rate": 0.00010297827652433922,
      "loss": 0.0035,
      "step": 10512
    },
    {
      "epoch": 4.9068844807467915,
      "grad_norm": 0.455078125,
      "learning_rate": 0.00010296360951882344,
      "loss": 0.0075,
      "step": 10513
    },
    {
      "epoch": 4.9073512252042,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010294894244949788,
      "loss": 0.0125,
      "step": 10514
    },
    {
      "epoch": 4.90781796966161,
      "grad_norm": 1.1015625,
      "learning_rate": 0.00010293427531667827,
      "loss": 0.0091,
      "step": 10515
    },
    {
      "epoch": 4.90828471411902,
      "grad_norm": 0.89453125,
      "learning_rate": 0.00010291960812068049,
      "loss": 0.0061,
      "step": 10516
    },
    {
      "epoch": 4.908751458576429,
      "grad_norm": 0.51953125,
      "learning_rate": 0.00010290494086182031,
      "loss": 0.0052,
      "step": 10517
    },
    {
      "epoch": 4.909218203033839,
      "grad_norm": 0.49609375,
      "learning_rate": 0.00010289027354041349,
      "loss": 0.0086,
      "step": 10518
    },
    {
      "epoch": 4.9096849474912485,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00010287560615677592,
      "loss": 0.0053,
      "step": 10519
    },
    {
      "epoch": 4.910151691948658,
      "grad_norm": 0.345703125,
      "learning_rate": 0.00010286093871122335,
      "loss": 0.0062,
      "step": 10520
    },
    {
      "epoch": 4.910618436406068,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00010284627120407158,
      "loss": 0.0054,
      "step": 10521
    },
    {
      "epoch": 4.911085180863477,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00010283160363563644,
      "loss": 0.0051,
      "step": 10522
    },
    {
      "epoch": 4.911551925320887,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00010281693600623374,
      "loss": 0.0032,
      "step": 10523
    },
    {
      "epoch": 4.912018669778297,
      "grad_norm": 0.67578125,
      "learning_rate": 0.00010280226831617928,
      "loss": 0.015,
      "step": 10524
    },
    {
      "epoch": 4.9124854142357055,
      "grad_norm": 0.435546875,
      "learning_rate": 0.00010278760056578885,
      "loss": 0.0101,
      "step": 10525
    },
    {
      "epoch": 4.912952158693115,
      "grad_norm": 0.6328125,
      "learning_rate": 0.00010277293275537832,
      "loss": 0.0096,
      "step": 10526
    },
    {
      "epoch": 4.913418903150525,
      "grad_norm": 0.921875,
      "learning_rate": 0.00010275826488526346,
      "loss": 0.0145,
      "step": 10527
    },
    {
      "epoch": 4.913885647607935,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00010274359695576008,
      "loss": 0.0118,
      "step": 10528
    },
    {
      "epoch": 4.914352392065345,
      "grad_norm": 0.765625,
      "learning_rate": 0.00010272892896718403,
      "loss": 0.0125,
      "step": 10529
    },
    {
      "epoch": 4.914819136522754,
      "grad_norm": 0.77734375,
      "learning_rate": 0.00010271426091985112,
      "loss": 0.0134,
      "step": 10530
    },
    {
      "epoch": 4.915285880980163,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00010269959281407712,
      "loss": 0.0072,
      "step": 10531
    },
    {
      "epoch": 4.915752625437573,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00010268492465017794,
      "loss": 0.0077,
      "step": 10532
    },
    {
      "epoch": 4.916219369894982,
      "grad_norm": 0.4375,
      "learning_rate": 0.00010267025642846932,
      "loss": 0.0055,
      "step": 10533
    },
    {
      "epoch": 4.916686114352392,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00010265558814926715,
      "loss": 0.008,
      "step": 10534
    },
    {
      "epoch": 4.917152858809802,
      "grad_norm": 0.9921875,
      "learning_rate": 0.0001026409198128872,
      "loss": 0.013,
      "step": 10535
    },
    {
      "epoch": 4.9176196032672115,
      "grad_norm": 0.4140625,
      "learning_rate": 0.00010262625141964532,
      "loss": 0.007,
      "step": 10536
    },
    {
      "epoch": 4.91808634772462,
      "grad_norm": 0.58984375,
      "learning_rate": 0.00010261158296985734,
      "loss": 0.0082,
      "step": 10537
    },
    {
      "epoch": 4.91855309218203,
      "grad_norm": 0.640625,
      "learning_rate": 0.00010259691446383906,
      "loss": 0.0161,
      "step": 10538
    },
    {
      "epoch": 4.91901983663944,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00010258224590190634,
      "loss": 0.0054,
      "step": 10539
    },
    {
      "epoch": 4.91948658109685,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00010256757728437503,
      "loss": 0.0105,
      "step": 10540
    },
    {
      "epoch": 4.919953325554259,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00010255290861156089,
      "loss": 0.0112,
      "step": 10541
    },
    {
      "epoch": 4.9204200700116685,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010253823988377982,
      "loss": 0.0108,
      "step": 10542
    },
    {
      "epoch": 4.920886814469078,
      "grad_norm": 0.359375,
      "learning_rate": 0.0001025235711013476,
      "loss": 0.0041,
      "step": 10543
    },
    {
      "epoch": 4.921353558926488,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00010250890226458012,
      "loss": 0.0084,
      "step": 10544
    },
    {
      "epoch": 4.921820303383897,
      "grad_norm": 0.390625,
      "learning_rate": 0.00010249423337379318,
      "loss": 0.0077,
      "step": 10545
    },
    {
      "epoch": 4.922287047841307,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00010247956442930265,
      "loss": 0.0075,
      "step": 10546
    },
    {
      "epoch": 4.922753792298717,
      "grad_norm": 0.66015625,
      "learning_rate": 0.00010246489543142432,
      "loss": 0.0107,
      "step": 10547
    },
    {
      "epoch": 4.923220536756126,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00010245022638047406,
      "loss": 0.0034,
      "step": 10548
    },
    {
      "epoch": 4.923687281213535,
      "grad_norm": 0.5859375,
      "learning_rate": 0.00010243555727676772,
      "loss": 0.0087,
      "step": 10549
    },
    {
      "epoch": 4.924154025670945,
      "grad_norm": 0.48828125,
      "learning_rate": 0.00010242088812062111,
      "loss": 0.0071,
      "step": 10550
    },
    {
      "epoch": 4.924620770128355,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00010240621891235009,
      "loss": 0.0072,
      "step": 10551
    },
    {
      "epoch": 4.925087514585765,
      "grad_norm": 0.66796875,
      "learning_rate": 0.00010239154965227053,
      "loss": 0.0148,
      "step": 10552
    },
    {
      "epoch": 4.925554259043174,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00010237688034069826,
      "loss": 0.0074,
      "step": 10553
    },
    {
      "epoch": 4.926021003500583,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001023622109779491,
      "loss": 0.0133,
      "step": 10554
    },
    {
      "epoch": 4.926487747957993,
      "grad_norm": 0.42578125,
      "learning_rate": 0.00010234754156433892,
      "loss": 0.006,
      "step": 10555
    },
    {
      "epoch": 4.926954492415403,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010233287210018357,
      "loss": 0.0109,
      "step": 10556
    },
    {
      "epoch": 4.927421236872812,
      "grad_norm": 0.375,
      "learning_rate": 0.00010231820258579889,
      "loss": 0.0049,
      "step": 10557
    },
    {
      "epoch": 4.927887981330222,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00010230353302150073,
      "loss": 0.0047,
      "step": 10558
    },
    {
      "epoch": 4.9283547257876315,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00010228886340760495,
      "loss": 0.0139,
      "step": 10559
    },
    {
      "epoch": 4.92882147024504,
      "grad_norm": 0.515625,
      "learning_rate": 0.00010227419374442742,
      "loss": 0.0075,
      "step": 10560
    },
    {
      "epoch": 4.92928821470245,
      "grad_norm": 0.498046875,
      "learning_rate": 0.00010225952403228395,
      "loss": 0.0138,
      "step": 10561
    },
    {
      "epoch": 4.92975495915986,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010224485427149047,
      "loss": 0.0038,
      "step": 10562
    },
    {
      "epoch": 4.93022170361727,
      "grad_norm": 0.4375,
      "learning_rate": 0.00010223018446236274,
      "loss": 0.0079,
      "step": 10563
    },
    {
      "epoch": 4.9306884480746795,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00010221551460521666,
      "loss": 0.0057,
      "step": 10564
    },
    {
      "epoch": 4.9311551925320884,
      "grad_norm": 0.392578125,
      "learning_rate": 0.00010220084470036813,
      "loss": 0.0047,
      "step": 10565
    },
    {
      "epoch": 4.931621936989498,
      "grad_norm": 0.36328125,
      "learning_rate": 0.00010218617474813294,
      "loss": 0.004,
      "step": 10566
    },
    {
      "epoch": 4.932088681446908,
      "grad_norm": 0.458984375,
      "learning_rate": 0.00010217150474882697,
      "loss": 0.0069,
      "step": 10567
    },
    {
      "epoch": 4.932555425904317,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00010215683470276611,
      "loss": 0.0084,
      "step": 10568
    },
    {
      "epoch": 4.933022170361727,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00010214216461026623,
      "loss": 0.0063,
      "step": 10569
    },
    {
      "epoch": 4.9334889148191365,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00010212749447164314,
      "loss": 0.0058,
      "step": 10570
    },
    {
      "epoch": 4.933955659276546,
      "grad_norm": 0.55078125,
      "learning_rate": 0.00010211282428721273,
      "loss": 0.0076,
      "step": 10571
    },
    {
      "epoch": 4.934422403733956,
      "grad_norm": 0.28125,
      "learning_rate": 0.0001020981540572909,
      "loss": 0.0031,
      "step": 10572
    },
    {
      "epoch": 4.934889148191365,
      "grad_norm": 0.46484375,
      "learning_rate": 0.00010208348378219348,
      "loss": 0.0082,
      "step": 10573
    },
    {
      "epoch": 4.935355892648775,
      "grad_norm": 0.6796875,
      "learning_rate": 0.00010206881346223632,
      "loss": 0.0144,
      "step": 10574
    },
    {
      "epoch": 4.935822637106185,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00010205414309773532,
      "loss": 0.0034,
      "step": 10575
    },
    {
      "epoch": 4.9362893815635935,
      "grad_norm": 0.447265625,
      "learning_rate": 0.00010203947268900633,
      "loss": 0.0062,
      "step": 10576
    },
    {
      "epoch": 4.936756126021003,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00010202480223636524,
      "loss": 0.003,
      "step": 10577
    },
    {
      "epoch": 4.937222870478413,
      "grad_norm": 0.578125,
      "learning_rate": 0.0001020101317401279,
      "loss": 0.0105,
      "step": 10578
    },
    {
      "epoch": 4.937689614935823,
      "grad_norm": 0.47265625,
      "learning_rate": 0.00010199546120061022,
      "loss": 0.0062,
      "step": 10579
    },
    {
      "epoch": 4.938156359393232,
      "grad_norm": 1.1796875,
      "learning_rate": 0.00010198079061812802,
      "loss": 0.0131,
      "step": 10580
    },
    {
      "epoch": 4.938623103850642,
      "grad_norm": 0.4140625,
      "learning_rate": 0.0001019661199929972,
      "loss": 0.004,
      "step": 10581
    },
    {
      "epoch": 4.939089848308051,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00010195144932553366,
      "loss": 0.0101,
      "step": 10582
    },
    {
      "epoch": 4.939556592765461,
      "grad_norm": 0.4609375,
      "learning_rate": 0.00010193677861605325,
      "loss": 0.0075,
      "step": 10583
    },
    {
      "epoch": 4.94002333722287,
      "grad_norm": 0.384765625,
      "learning_rate": 0.00010192210786487184,
      "loss": 0.0058,
      "step": 10584
    },
    {
      "epoch": 4.94049008168028,
      "grad_norm": 0.5703125,
      "learning_rate": 0.0001019074370723053,
      "loss": 0.0118,
      "step": 10585
    },
    {
      "epoch": 4.94095682613769,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00010189276623866953,
      "loss": 0.0061,
      "step": 10586
    },
    {
      "epoch": 4.9414235705950995,
      "grad_norm": 0.5,
      "learning_rate": 0.00010187809536428041,
      "loss": 0.0052,
      "step": 10587
    },
    {
      "epoch": 4.941890315052508,
      "grad_norm": 0.53125,
      "learning_rate": 0.0001018634244494538,
      "loss": 0.0125,
      "step": 10588
    },
    {
      "epoch": 4.942357059509918,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00010184875349450562,
      "loss": 0.0067,
      "step": 10589
    },
    {
      "epoch": 4.942823803967328,
      "grad_norm": 0.388671875,
      "learning_rate": 0.00010183408249975169,
      "loss": 0.0049,
      "step": 10590
    },
    {
      "epoch": 4.943290548424738,
      "grad_norm": 0.5703125,
      "learning_rate": 0.00010181941146550796,
      "loss": 0.0091,
      "step": 10591
    },
    {
      "epoch": 4.943757292882147,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00010180474039209026,
      "loss": 0.0096,
      "step": 10592
    },
    {
      "epoch": 4.9442240373395565,
      "grad_norm": 0.326171875,
      "learning_rate": 0.00010179006927981451,
      "loss": 0.0032,
      "step": 10593
    },
    {
      "epoch": 4.944690781796966,
      "grad_norm": 0.27734375,
      "learning_rate": 0.00010177539812899657,
      "loss": 0.0037,
      "step": 10594
    },
    {
      "epoch": 4.945157526254376,
      "grad_norm": 0.6171875,
      "learning_rate": 0.00010176072693995235,
      "loss": 0.0103,
      "step": 10595
    },
    {
      "epoch": 4.945624270711785,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010174605571299772,
      "loss": 0.007,
      "step": 10596
    },
    {
      "epoch": 4.946091015169195,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010173138444844859,
      "loss": 0.0046,
      "step": 10597
    },
    {
      "epoch": 4.946557759626605,
      "grad_norm": 0.44921875,
      "learning_rate": 0.00010171671314662082,
      "loss": 0.0075,
      "step": 10598
    },
    {
      "epoch": 4.947024504084014,
      "grad_norm": 0.515625,
      "learning_rate": 0.00010170204180783032,
      "loss": 0.0077,
      "step": 10599
    },
    {
      "epoch": 4.947491248541423,
      "grad_norm": 0.333984375,
      "learning_rate": 0.00010168737043239293,
      "loss": 0.0038,
      "step": 10600
    },
    {
      "epoch": 4.947957992998833,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010167269902062461,
      "loss": 0.008,
      "step": 10601
    },
    {
      "epoch": 4.948424737456243,
      "grad_norm": 0.515625,
      "learning_rate": 0.00010165802757284124,
      "loss": 0.0078,
      "step": 10602
    },
    {
      "epoch": 4.948891481913652,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010164335608935869,
      "loss": 0.011,
      "step": 10603
    },
    {
      "epoch": 4.949358226371062,
      "grad_norm": 0.482421875,
      "learning_rate": 0.00010162868457049286,
      "loss": 0.0048,
      "step": 10604
    },
    {
      "epoch": 4.949824970828471,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00010161401301655964,
      "loss": 0.0066,
      "step": 10605
    },
    {
      "epoch": 4.950291715285881,
      "grad_norm": 0.162109375,
      "learning_rate": 0.00010159934142787493,
      "loss": 0.0022,
      "step": 10606
    },
    {
      "epoch": 4.950758459743291,
      "grad_norm": 0.390625,
      "learning_rate": 0.00010158466980475461,
      "loss": 0.0069,
      "step": 10607
    },
    {
      "epoch": 4.9512252042007,
      "grad_norm": 0.4453125,
      "learning_rate": 0.00010156999814751462,
      "loss": 0.0067,
      "step": 10608
    },
    {
      "epoch": 4.95169194865811,
      "grad_norm": 0.671875,
      "learning_rate": 0.00010155532645647081,
      "loss": 0.0098,
      "step": 10609
    },
    {
      "epoch": 4.9521586931155195,
      "grad_norm": 0.5546875,
      "learning_rate": 0.0001015406547319391,
      "loss": 0.0084,
      "step": 10610
    },
    {
      "epoch": 4.952625437572928,
      "grad_norm": 0.4921875,
      "learning_rate": 0.00010152598297423539,
      "loss": 0.0082,
      "step": 10611
    },
    {
      "epoch": 4.953092182030338,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010151131118367556,
      "loss": 0.0081,
      "step": 10612
    },
    {
      "epoch": 4.953558926487748,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010149663936057552,
      "loss": 0.0084,
      "step": 10613
    },
    {
      "epoch": 4.954025670945158,
      "grad_norm": 0.376953125,
      "learning_rate": 0.00010148196750525118,
      "loss": 0.0053,
      "step": 10614
    },
    {
      "epoch": 4.954492415402568,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00010146729561801844,
      "loss": 0.0086,
      "step": 10615
    },
    {
      "epoch": 4.9549591598599765,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010145262369919317,
      "loss": 0.0046,
      "step": 10616
    },
    {
      "epoch": 4.955425904317386,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00010143795174909135,
      "loss": 0.0052,
      "step": 10617
    },
    {
      "epoch": 4.955892648774796,
      "grad_norm": 0.28125,
      "learning_rate": 0.0001014232797680288,
      "loss": 0.0055,
      "step": 10618
    },
    {
      "epoch": 4.956359393232205,
      "grad_norm": 0.53125,
      "learning_rate": 0.00010140860775632147,
      "loss": 0.0071,
      "step": 10619
    },
    {
      "epoch": 4.956826137689615,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00010139393571428524,
      "loss": 0.0069,
      "step": 10620
    },
    {
      "epoch": 4.957292882147025,
      "grad_norm": 0.64453125,
      "learning_rate": 0.00010137926364223605,
      "loss": 0.0073,
      "step": 10621
    },
    {
      "epoch": 4.957759626604434,
      "grad_norm": 0.310546875,
      "learning_rate": 0.00010136459154048976,
      "loss": 0.0045,
      "step": 10622
    },
    {
      "epoch": 4.958226371061843,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00010134991940936229,
      "loss": 0.0118,
      "step": 10623
    },
    {
      "epoch": 4.958693115519253,
      "grad_norm": 0.390625,
      "learning_rate": 0.00010133524724916959,
      "loss": 0.0088,
      "step": 10624
    },
    {
      "epoch": 4.959159859976663,
      "grad_norm": 0.69921875,
      "learning_rate": 0.00010132057506022752,
      "loss": 0.0127,
      "step": 10625
    },
    {
      "epoch": 4.959626604434073,
      "grad_norm": 0.48828125,
      "learning_rate": 0.000101305902842852,
      "loss": 0.0056,
      "step": 10626
    },
    {
      "epoch": 4.960093348891482,
      "grad_norm": 0.3125,
      "learning_rate": 0.00010129123059735895,
      "loss": 0.0038,
      "step": 10627
    },
    {
      "epoch": 4.960560093348891,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010127655832406428,
      "loss": 0.01,
      "step": 10628
    },
    {
      "epoch": 4.961026837806301,
      "grad_norm": 0.439453125,
      "learning_rate": 0.00010126188602328385,
      "loss": 0.0087,
      "step": 10629
    },
    {
      "epoch": 4.961493582263711,
      "grad_norm": 0.494140625,
      "learning_rate": 0.00010124721369533367,
      "loss": 0.0084,
      "step": 10630
    },
    {
      "epoch": 4.96196032672112,
      "grad_norm": 0.3359375,
      "learning_rate": 0.00010123254134052957,
      "loss": 0.0036,
      "step": 10631
    },
    {
      "epoch": 4.96242707117853,
      "grad_norm": 0.404296875,
      "learning_rate": 0.0001012178689591875,
      "loss": 0.0037,
      "step": 10632
    },
    {
      "epoch": 4.9628938156359395,
      "grad_norm": 0.390625,
      "learning_rate": 0.00010120319655162335,
      "loss": 0.0055,
      "step": 10633
    },
    {
      "epoch": 4.963360560093349,
      "grad_norm": 0.314453125,
      "learning_rate": 0.00010118852411815305,
      "loss": 0.0041,
      "step": 10634
    },
    {
      "epoch": 4.963827304550758,
      "grad_norm": 0.453125,
      "learning_rate": 0.0001011738516590925,
      "loss": 0.0063,
      "step": 10635
    },
    {
      "epoch": 4.964294049008168,
      "grad_norm": 0.3828125,
      "learning_rate": 0.00010115917917475762,
      "loss": 0.0074,
      "step": 10636
    },
    {
      "epoch": 4.964760793465578,
      "grad_norm": 0.2470703125,
      "learning_rate": 0.00010114450666546434,
      "loss": 0.0038,
      "step": 10637
    },
    {
      "epoch": 4.9652275379229875,
      "grad_norm": 0.2421875,
      "learning_rate": 0.00010112983413152854,
      "loss": 0.0037,
      "step": 10638
    },
    {
      "epoch": 4.9656942823803965,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00010111516157326614,
      "loss": 0.0062,
      "step": 10639
    },
    {
      "epoch": 4.966161026837806,
      "grad_norm": 0.361328125,
      "learning_rate": 0.00010110048899099314,
      "loss": 0.004,
      "step": 10640
    },
    {
      "epoch": 4.966627771295216,
      "grad_norm": 0.283203125,
      "learning_rate": 0.00010108581638502535,
      "loss": 0.0046,
      "step": 10641
    },
    {
      "epoch": 4.967094515752626,
      "grad_norm": 0.63671875,
      "learning_rate": 0.0001010711437556787,
      "loss": 0.0077,
      "step": 10642
    },
    {
      "epoch": 4.967561260210035,
      "grad_norm": 0.3984375,
      "learning_rate": 0.00010105647110326919,
      "loss": 0.0086,
      "step": 10643
    },
    {
      "epoch": 4.9680280046674445,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00010104179842811266,
      "loss": 0.0079,
      "step": 10644
    },
    {
      "epoch": 4.968494749124854,
      "grad_norm": 0.462890625,
      "learning_rate": 0.00010102712573052504,
      "loss": 0.006,
      "step": 10645
    },
    {
      "epoch": 4.968961493582263,
      "grad_norm": 0.5390625,
      "learning_rate": 0.0001010124530108223,
      "loss": 0.0108,
      "step": 10646
    },
    {
      "epoch": 4.969428238039673,
      "grad_norm": 0.41015625,
      "learning_rate": 0.0001009977802693203,
      "loss": 0.0054,
      "step": 10647
    },
    {
      "epoch": 4.969894982497083,
      "grad_norm": 0.408203125,
      "learning_rate": 0.00010098310750633499,
      "loss": 0.0106,
      "step": 10648
    },
    {
      "epoch": 4.970361726954493,
      "grad_norm": 0.37890625,
      "learning_rate": 0.00010096843472218227,
      "loss": 0.0067,
      "step": 10649
    },
    {
      "epoch": 4.970828471411902,
      "grad_norm": 0.322265625,
      "learning_rate": 0.00010095376191717809,
      "loss": 0.0059,
      "step": 10650
    },
    {
      "epoch": 4.971295215869311,
      "grad_norm": 0.423828125,
      "learning_rate": 0.00010093908909163837,
      "loss": 0.0044,
      "step": 10651
    },
    {
      "epoch": 4.971761960326721,
      "grad_norm": 0.58203125,
      "learning_rate": 0.00010092441624587899,
      "loss": 0.0092,
      "step": 10652
    },
    {
      "epoch": 4.972228704784131,
      "grad_norm": 0.41015625,
      "learning_rate": 0.00010090974338021591,
      "loss": 0.0027,
      "step": 10653
    },
    {
      "epoch": 4.97269544924154,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00010089507049496505,
      "loss": 0.0051,
      "step": 10654
    },
    {
      "epoch": 4.97316219369895,
      "grad_norm": 0.4765625,
      "learning_rate": 0.00010088039759044235,
      "loss": 0.0059,
      "step": 10655
    },
    {
      "epoch": 4.973628938156359,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010086572466696369,
      "loss": 0.0079,
      "step": 10656
    },
    {
      "epoch": 4.974095682613769,
      "grad_norm": 0.609375,
      "learning_rate": 0.00010085105172484503,
      "loss": 0.0093,
      "step": 10657
    },
    {
      "epoch": 4.974562427071179,
      "grad_norm": 0.40234375,
      "learning_rate": 0.00010083637876440226,
      "loss": 0.0055,
      "step": 10658
    },
    {
      "epoch": 4.975029171528588,
      "grad_norm": 0.39453125,
      "learning_rate": 0.00010082170578595136,
      "loss": 0.0045,
      "step": 10659
    },
    {
      "epoch": 4.975495915985998,
      "grad_norm": 0.5546875,
      "learning_rate": 0.00010080703278980821,
      "loss": 0.0098,
      "step": 10660
    },
    {
      "epoch": 4.9759626604434075,
      "grad_norm": 0.474609375,
      "learning_rate": 0.00010079235977628874,
      "loss": 0.0053,
      "step": 10661
    },
    {
      "epoch": 4.976429404900816,
      "grad_norm": 0.34375,
      "learning_rate": 0.00010077768674570887,
      "loss": 0.006,
      "step": 10662
    },
    {
      "epoch": 4.976896149358226,
      "grad_norm": 0.390625,
      "learning_rate": 0.00010076301369838457,
      "loss": 0.0067,
      "step": 10663
    },
    {
      "epoch": 4.977362893815636,
      "grad_norm": 0.37109375,
      "learning_rate": 0.00010074834063463173,
      "loss": 0.0064,
      "step": 10664
    },
    {
      "epoch": 4.977829638273046,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00010073366755476628,
      "loss": 0.0042,
      "step": 10665
    },
    {
      "epoch": 4.978296382730455,
      "grad_norm": 0.232421875,
      "learning_rate": 0.00010071899445910416,
      "loss": 0.0027,
      "step": 10666
    },
    {
      "epoch": 4.9787631271878645,
      "grad_norm": 0.546875,
      "learning_rate": 0.00010070432134796128,
      "loss": 0.012,
      "step": 10667
    },
    {
      "epoch": 4.979229871645274,
      "grad_norm": 0.306640625,
      "learning_rate": 0.00010068964822165358,
      "loss": 0.0048,
      "step": 10668
    },
    {
      "epoch": 4.979696616102684,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00010067497508049701,
      "loss": 0.0098,
      "step": 10669
    },
    {
      "epoch": 4.980163360560093,
      "grad_norm": 0.443359375,
      "learning_rate": 0.00010066030192480746,
      "loss": 0.0041,
      "step": 10670
    },
    {
      "epoch": 4.980630105017503,
      "grad_norm": 0.4375,
      "learning_rate": 0.00010064562875490088,
      "loss": 0.0043,
      "step": 10671
    },
    {
      "epoch": 4.981096849474913,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00010063095557109319,
      "loss": 0.0056,
      "step": 10672
    },
    {
      "epoch": 4.981563593932322,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00010061628237370035,
      "loss": 0.0081,
      "step": 10673
    },
    {
      "epoch": 4.982030338389731,
      "grad_norm": 0.404296875,
      "learning_rate": 0.00010060160916303824,
      "loss": 0.0076,
      "step": 10674
    },
    {
      "epoch": 4.982497082847141,
      "grad_norm": 0.55859375,
      "learning_rate": 0.00010058693593942284,
      "loss": 0.0066,
      "step": 10675
    },
    {
      "epoch": 4.982963827304551,
      "grad_norm": 0.703125,
      "learning_rate": 0.00010057226270317003,
      "loss": 0.0071,
      "step": 10676
    },
    {
      "epoch": 4.983430571761961,
      "grad_norm": 0.5078125,
      "learning_rate": 0.00010055758945459582,
      "loss": 0.0099,
      "step": 10677
    },
    {
      "epoch": 4.98389731621937,
      "grad_norm": 0.50390625,
      "learning_rate": 0.00010054291619401604,
      "loss": 0.0049,
      "step": 10678
    },
    {
      "epoch": 4.984364060676779,
      "grad_norm": 0.2392578125,
      "learning_rate": 0.00010052824292174668,
      "loss": 0.0036,
      "step": 10679
    },
    {
      "epoch": 4.984830805134189,
      "grad_norm": 0.3828125,
      "learning_rate": 0.0001005135696381037,
      "loss": 0.0069,
      "step": 10680
    },
    {
      "epoch": 4.985297549591599,
      "grad_norm": 0.33203125,
      "learning_rate": 0.00010049889634340294,
      "loss": 0.0035,
      "step": 10681
    },
    {
      "epoch": 4.985764294049008,
      "grad_norm": 0.416015625,
      "learning_rate": 0.00010048422303796044,
      "loss": 0.0096,
      "step": 10682
    },
    {
      "epoch": 4.986231038506418,
      "grad_norm": 0.369140625,
      "learning_rate": 0.00010046954972209206,
      "loss": 0.0036,
      "step": 10683
    },
    {
      "epoch": 4.9866977829638275,
      "grad_norm": 0.5390625,
      "learning_rate": 0.00010045487639611373,
      "loss": 0.0096,
      "step": 10684
    },
    {
      "epoch": 4.987164527421237,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00010044020306034143,
      "loss": 0.0052,
      "step": 10685
    },
    {
      "epoch": 4.987631271878646,
      "grad_norm": 0.7109375,
      "learning_rate": 0.00010042552971509107,
      "loss": 0.006,
      "step": 10686
    },
    {
      "epoch": 4.988098016336056,
      "grad_norm": 0.408203125,
      "learning_rate": 0.0001004108563606786,
      "loss": 0.0056,
      "step": 10687
    },
    {
      "epoch": 4.988564760793466,
      "grad_norm": 0.16015625,
      "learning_rate": 0.00010039618299741989,
      "loss": 0.0019,
      "step": 10688
    },
    {
      "epoch": 4.989031505250875,
      "grad_norm": 0.30078125,
      "learning_rate": 0.00010038150962563095,
      "loss": 0.0104,
      "step": 10689
    },
    {
      "epoch": 4.9894982497082845,
      "grad_norm": 0.44140625,
      "learning_rate": 0.00010036683624562771,
      "loss": 0.0059,
      "step": 10690
    },
    {
      "epoch": 4.989964994165694,
      "grad_norm": 0.41796875,
      "learning_rate": 0.00010035216285772602,
      "loss": 0.0111,
      "step": 10691
    },
    {
      "epoch": 4.990431738623104,
      "grad_norm": 0.478515625,
      "learning_rate": 0.00010033748946224191,
      "loss": 0.0077,
      "step": 10692
    },
    {
      "epoch": 4.990898483080514,
      "grad_norm": 0.34375,
      "learning_rate": 0.00010032281605949127,
      "loss": 0.005,
      "step": 10693
    },
    {
      "epoch": 4.991365227537923,
      "grad_norm": 0.5,
      "learning_rate": 0.00010030814264979003,
      "loss": 0.0047,
      "step": 10694
    },
    {
      "epoch": 4.991831971995333,
      "grad_norm": 0.181640625,
      "learning_rate": 0.00010029346923345415,
      "loss": 0.0019,
      "step": 10695
    },
    {
      "epoch": 4.992298716452742,
      "grad_norm": 0.427734375,
      "learning_rate": 0.00010027879581079955,
      "loss": 0.0038,
      "step": 10696
    },
    {
      "epoch": 4.992765460910151,
      "grad_norm": 0.298828125,
      "learning_rate": 0.00010026412238214218,
      "loss": 0.0032,
      "step": 10697
    },
    {
      "epoch": 4.993232205367561,
      "grad_norm": 0.396484375,
      "learning_rate": 0.00010024944894779796,
      "loss": 0.0051,
      "step": 10698
    },
    {
      "epoch": 4.993698949824971,
      "grad_norm": 0.42578125,
      "learning_rate": 0.0001002347755080828,
      "loss": 0.0057,
      "step": 10699
    },
    {
      "epoch": 4.994165694282381,
      "grad_norm": 0.357421875,
      "learning_rate": 0.00010022010206331269,
      "loss": 0.0042,
      "step": 10700
    },
    {
      "epoch": 4.9946324387397905,
      "grad_norm": 0.1943359375,
      "learning_rate": 0.00010020542861380351,
      "loss": 0.0023,
      "step": 10701
    },
    {
      "epoch": 4.995099183197199,
      "grad_norm": 0.263671875,
      "learning_rate": 0.00010019075515987127,
      "loss": 0.0051,
      "step": 10702
    },
    {
      "epoch": 4.995565927654609,
      "grad_norm": 0.29296875,
      "learning_rate": 0.00010017608170183181,
      "loss": 0.0036,
      "step": 10703
    },
    {
      "epoch": 4.996032672112019,
      "grad_norm": 0.375,
      "learning_rate": 0.00010016140824000115,
      "loss": 0.0052,
      "step": 10704
    },
    {
      "epoch": 4.996499416569428,
      "grad_norm": 0.3125,
      "learning_rate": 0.00010014673477469519,
      "loss": 0.004,
      "step": 10705
    },
    {
      "epoch": 4.996966161026838,
      "grad_norm": 0.52734375,
      "learning_rate": 0.00010013206130622988,
      "loss": 0.0082,
      "step": 10706
    },
    {
      "epoch": 4.9974329054842475,
      "grad_norm": 0.65234375,
      "learning_rate": 0.00010011738783492108,
      "loss": 0.0072,
      "step": 10707
    },
    {
      "epoch": 4.997899649941657,
      "grad_norm": 0.412109375,
      "learning_rate": 0.00010010271436108485,
      "loss": 0.0078,
      "step": 10708
    },
    {
      "epoch": 4.998366394399066,
      "grad_norm": 0.51171875,
      "learning_rate": 0.00010008804088503705,
      "loss": 0.0052,
      "step": 10709
    },
    {
      "epoch": 4.998833138856476,
      "grad_norm": 0.400390625,
      "learning_rate": 0.00010007336740709362,
      "loss": 0.0027,
      "step": 10710
    },
    {
      "epoch": 4.999299883313886,
      "grad_norm": 0.375,
      "learning_rate": 0.00010005869392757053,
      "loss": 0.0029,
      "step": 10711
    },
    {
      "epoch": 4.9997666277712955,
      "grad_norm": 0.353515625,
      "learning_rate": 0.00010004402044678368,
      "loss": 0.0052,
      "step": 10712
    },
    {
      "epoch": 5.0002333722287045,
      "grad_norm": 0.2275390625,
      "learning_rate": 0.00010002934696504902,
      "loss": 0.0025,
      "step": 10713
    },
    {
      "epoch": 5.000700116686114,
      "grad_norm": 0.1669921875,
      "learning_rate": 0.00010001467348268247,
      "loss": 0.0013,
      "step": 10714
    },
    {
      "epoch": 5.001166861143524,
      "grad_norm": 0.36328125,
      "learning_rate": 0.0001,
      "loss": 0.0041,
      "step": 10715
    },
    {
      "epoch": 5.001633605600934,
      "grad_norm": 0.361328125,
      "learning_rate": 9.998532651731755e-05,
      "loss": 0.0046,
      "step": 10716
    },
    {
      "epoch": 5.002100350058343,
      "grad_norm": 0.318359375,
      "learning_rate": 9.9970653034951e-05,
      "loss": 0.0029,
      "step": 10717
    },
    {
      "epoch": 5.0025670945157525,
      "grad_norm": 0.2275390625,
      "learning_rate": 9.995597955321635e-05,
      "loss": 0.0041,
      "step": 10718
    },
    {
      "epoch": 5.003033838973162,
      "grad_norm": 0.41796875,
      "learning_rate": 9.994130607242951e-05,
      "loss": 0.0023,
      "step": 10719
    },
    {
      "epoch": 5.003500583430572,
      "grad_norm": 0.3359375,
      "learning_rate": 9.992663259290641e-05,
      "loss": 0.0052,
      "step": 10720
    },
    {
      "epoch": 5.003500583430572,
      "eval_loss": 2.0806281566619873,
      "eval_runtime": 57.6303,
      "eval_samples_per_second": 31.303,
      "eval_steps_per_second": 3.922,
      "step": 10720
    },
    {
      "epoch": 5.003967327887981,
      "grad_norm": 0.42578125,
      "learning_rate": 9.991195911496297e-05,
      "loss": 0.0092,
      "step": 10721
    },
    {
      "epoch": 5.004434072345391,
      "grad_norm": 0.46875,
      "learning_rate": 9.989728563891517e-05,
      "loss": 0.0072,
      "step": 10722
    },
    {
      "epoch": 5.004900816802801,
      "grad_norm": 0.58984375,
      "learning_rate": 9.988261216507894e-05,
      "loss": 0.0107,
      "step": 10723
    },
    {
      "epoch": 5.00536756126021,
      "grad_norm": 0.283203125,
      "learning_rate": 9.986793869377016e-05,
      "loss": 0.0038,
      "step": 10724
    },
    {
      "epoch": 5.005834305717619,
      "grad_norm": 0.251953125,
      "learning_rate": 9.985326522530482e-05,
      "loss": 0.0023,
      "step": 10725
    },
    {
      "epoch": 5.006301050175029,
      "grad_norm": 0.3671875,
      "learning_rate": 9.983859175999887e-05,
      "loss": 0.0035,
      "step": 10726
    },
    {
      "epoch": 5.006767794632439,
      "grad_norm": 0.294921875,
      "learning_rate": 9.982391829816817e-05,
      "loss": 0.0021,
      "step": 10727
    },
    {
      "epoch": 5.007234539089849,
      "grad_norm": 0.2470703125,
      "learning_rate": 9.980924484012874e-05,
      "loss": 0.0019,
      "step": 10728
    },
    {
      "epoch": 5.007701283547258,
      "grad_norm": 0.400390625,
      "learning_rate": 9.97945713861965e-05,
      "loss": 0.0064,
      "step": 10729
    },
    {
      "epoch": 5.008168028004667,
      "grad_norm": 0.1640625,
      "learning_rate": 9.977989793668733e-05,
      "loss": 0.0018,
      "step": 10730
    },
    {
      "epoch": 5.008634772462077,
      "grad_norm": 0.478515625,
      "learning_rate": 9.97652244919172e-05,
      "loss": 0.0088,
      "step": 10731
    },
    {
      "epoch": 5.009101516919487,
      "grad_norm": 0.33984375,
      "learning_rate": 9.975055105220205e-05,
      "loss": 0.0041,
      "step": 10732
    },
    {
      "epoch": 5.009568261376896,
      "grad_norm": 0.1494140625,
      "learning_rate": 9.973587761785783e-05,
      "loss": 0.0017,
      "step": 10733
    },
    {
      "epoch": 5.010035005834306,
      "grad_norm": 0.1904296875,
      "learning_rate": 9.972120418920044e-05,
      "loss": 0.0019,
      "step": 10734
    },
    {
      "epoch": 5.0105017502917155,
      "grad_norm": 0.166015625,
      "learning_rate": 9.970653076654585e-05,
      "loss": 0.0012,
      "step": 10735
    },
    {
      "epoch": 5.010968494749124,
      "grad_norm": 0.240234375,
      "learning_rate": 9.969185735020997e-05,
      "loss": 0.002,
      "step": 10736
    },
    {
      "epoch": 5.011435239206534,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.967718394050873e-05,
      "loss": 0.0018,
      "step": 10737
    },
    {
      "epoch": 5.011901983663944,
      "grad_norm": 0.5390625,
      "learning_rate": 9.966251053775809e-05,
      "loss": 0.0108,
      "step": 10738
    },
    {
      "epoch": 5.012368728121354,
      "grad_norm": 0.37890625,
      "learning_rate": 9.9647837142274e-05,
      "loss": 0.0037,
      "step": 10739
    },
    {
      "epoch": 5.012835472578763,
      "grad_norm": 0.330078125,
      "learning_rate": 9.963316375437235e-05,
      "loss": 0.0024,
      "step": 10740
    },
    {
      "epoch": 5.0133022170361725,
      "grad_norm": 0.4921875,
      "learning_rate": 9.961849037436909e-05,
      "loss": 0.0123,
      "step": 10741
    },
    {
      "epoch": 5.013768961493582,
      "grad_norm": 0.236328125,
      "learning_rate": 9.960381700258015e-05,
      "loss": 0.0043,
      "step": 10742
    },
    {
      "epoch": 5.014235705950992,
      "grad_norm": 0.21875,
      "learning_rate": 9.958914363932145e-05,
      "loss": 0.0015,
      "step": 10743
    },
    {
      "epoch": 5.014702450408401,
      "grad_norm": 0.5,
      "learning_rate": 9.957447028490898e-05,
      "loss": 0.0077,
      "step": 10744
    },
    {
      "epoch": 5.015169194865811,
      "grad_norm": 0.16796875,
      "learning_rate": 9.95597969396586e-05,
      "loss": 0.0034,
      "step": 10745
    },
    {
      "epoch": 5.015635939323221,
      "grad_norm": 0.310546875,
      "learning_rate": 9.954512360388629e-05,
      "loss": 0.0041,
      "step": 10746
    },
    {
      "epoch": 5.01610268378063,
      "grad_norm": 0.1484375,
      "learning_rate": 9.953045027790798e-05,
      "loss": 0.0016,
      "step": 10747
    },
    {
      "epoch": 5.016569428238039,
      "grad_norm": 0.42578125,
      "learning_rate": 9.951577696203961e-05,
      "loss": 0.0075,
      "step": 10748
    },
    {
      "epoch": 5.017036172695449,
      "grad_norm": 0.42578125,
      "learning_rate": 9.950110365659707e-05,
      "loss": 0.0042,
      "step": 10749
    },
    {
      "epoch": 5.017502917152859,
      "grad_norm": 0.458984375,
      "learning_rate": 9.948643036189634e-05,
      "loss": 0.0055,
      "step": 10750
    },
    {
      "epoch": 5.017969661610269,
      "grad_norm": 0.193359375,
      "learning_rate": 9.947175707825334e-05,
      "loss": 0.0022,
      "step": 10751
    },
    {
      "epoch": 5.018436406067678,
      "grad_norm": 0.2890625,
      "learning_rate": 9.9457083805984e-05,
      "loss": 0.0025,
      "step": 10752
    },
    {
      "epoch": 5.018903150525087,
      "grad_norm": 0.30078125,
      "learning_rate": 9.944241054540422e-05,
      "loss": 0.0026,
      "step": 10753
    },
    {
      "epoch": 5.019369894982497,
      "grad_norm": 0.5,
      "learning_rate": 9.942773729682998e-05,
      "loss": 0.0037,
      "step": 10754
    },
    {
      "epoch": 5.019836639439907,
      "grad_norm": 0.13671875,
      "learning_rate": 9.94130640605772e-05,
      "loss": 0.0014,
      "step": 10755
    },
    {
      "epoch": 5.020303383897316,
      "grad_norm": 0.328125,
      "learning_rate": 9.939839083696177e-05,
      "loss": 0.0038,
      "step": 10756
    },
    {
      "epoch": 5.020770128354726,
      "grad_norm": 0.404296875,
      "learning_rate": 9.938371762629967e-05,
      "loss": 0.0053,
      "step": 10757
    },
    {
      "epoch": 5.0212368728121355,
      "grad_norm": 0.28515625,
      "learning_rate": 9.936904442890682e-05,
      "loss": 0.0047,
      "step": 10758
    },
    {
      "epoch": 5.021703617269545,
      "grad_norm": 0.2578125,
      "learning_rate": 9.935437124509913e-05,
      "loss": 0.0026,
      "step": 10759
    },
    {
      "epoch": 5.022170361726954,
      "grad_norm": 0.251953125,
      "learning_rate": 9.933969807519256e-05,
      "loss": 0.0039,
      "step": 10760
    },
    {
      "epoch": 5.022637106184364,
      "grad_norm": 0.671875,
      "learning_rate": 9.932502491950301e-05,
      "loss": 0.0066,
      "step": 10761
    },
    {
      "epoch": 5.023103850641774,
      "grad_norm": 0.328125,
      "learning_rate": 9.931035177834644e-05,
      "loss": 0.007,
      "step": 10762
    },
    {
      "epoch": 5.023570595099184,
      "grad_norm": 0.24609375,
      "learning_rate": 9.929567865203873e-05,
      "loss": 0.0067,
      "step": 10763
    },
    {
      "epoch": 5.0240373395565925,
      "grad_norm": 0.322265625,
      "learning_rate": 9.928100554089587e-05,
      "loss": 0.0025,
      "step": 10764
    },
    {
      "epoch": 5.024504084014002,
      "grad_norm": 0.0986328125,
      "learning_rate": 9.926633244523376e-05,
      "loss": 0.0013,
      "step": 10765
    },
    {
      "epoch": 5.024970828471412,
      "grad_norm": 0.1650390625,
      "learning_rate": 9.925165936536828e-05,
      "loss": 0.0016,
      "step": 10766
    },
    {
      "epoch": 5.025437572928822,
      "grad_norm": 0.294921875,
      "learning_rate": 9.923698630161545e-05,
      "loss": 0.007,
      "step": 10767
    },
    {
      "epoch": 5.025904317386231,
      "grad_norm": 0.10791015625,
      "learning_rate": 9.922231325429115e-05,
      "loss": 0.0012,
      "step": 10768
    },
    {
      "epoch": 5.026371061843641,
      "grad_norm": 0.30078125,
      "learning_rate": 9.920764022371129e-05,
      "loss": 0.0022,
      "step": 10769
    },
    {
      "epoch": 5.02683780630105,
      "grad_norm": 0.283203125,
      "learning_rate": 9.919296721019183e-05,
      "loss": 0.0026,
      "step": 10770
    },
    {
      "epoch": 5.02730455075846,
      "grad_norm": 0.4140625,
      "learning_rate": 9.917829421404868e-05,
      "loss": 0.0056,
      "step": 10771
    },
    {
      "epoch": 5.027771295215869,
      "grad_norm": 0.462890625,
      "learning_rate": 9.916362123559775e-05,
      "loss": 0.0075,
      "step": 10772
    },
    {
      "epoch": 5.028238039673279,
      "grad_norm": 0.33984375,
      "learning_rate": 9.914894827515498e-05,
      "loss": 0.0021,
      "step": 10773
    },
    {
      "epoch": 5.028704784130689,
      "grad_norm": 0.37109375,
      "learning_rate": 9.913427533303634e-05,
      "loss": 0.0035,
      "step": 10774
    },
    {
      "epoch": 5.029171528588098,
      "grad_norm": 0.318359375,
      "learning_rate": 9.911960240955767e-05,
      "loss": 0.0024,
      "step": 10775
    },
    {
      "epoch": 5.029638273045507,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.910492950503494e-05,
      "loss": 0.0019,
      "step": 10776
    },
    {
      "epoch": 5.030105017502917,
      "grad_norm": 0.2314453125,
      "learning_rate": 9.90902566197841e-05,
      "loss": 0.0023,
      "step": 10777
    },
    {
      "epoch": 5.030571761960327,
      "grad_norm": 0.35546875,
      "learning_rate": 9.907558375412102e-05,
      "loss": 0.007,
      "step": 10778
    },
    {
      "epoch": 5.031038506417736,
      "grad_norm": 0.15234375,
      "learning_rate": 9.906091090836167e-05,
      "loss": 0.0015,
      "step": 10779
    },
    {
      "epoch": 5.031505250875146,
      "grad_norm": 0.33984375,
      "learning_rate": 9.904623808282192e-05,
      "loss": 0.0054,
      "step": 10780
    },
    {
      "epoch": 5.0319719953325555,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.903156527781774e-05,
      "loss": 0.0016,
      "step": 10781
    },
    {
      "epoch": 5.032438739789965,
      "grad_norm": 0.294921875,
      "learning_rate": 9.901689249366503e-05,
      "loss": 0.0053,
      "step": 10782
    },
    {
      "epoch": 5.032905484247374,
      "grad_norm": 0.353515625,
      "learning_rate": 9.900221973067971e-05,
      "loss": 0.0045,
      "step": 10783
    },
    {
      "epoch": 5.033372228704784,
      "grad_norm": 0.396484375,
      "learning_rate": 9.898754698917771e-05,
      "loss": 0.0075,
      "step": 10784
    },
    {
      "epoch": 5.033838973162194,
      "grad_norm": 0.546875,
      "learning_rate": 9.897287426947496e-05,
      "loss": 0.0084,
      "step": 10785
    },
    {
      "epoch": 5.0343057176196035,
      "grad_norm": 0.4296875,
      "learning_rate": 9.895820157188734e-05,
      "loss": 0.0048,
      "step": 10786
    },
    {
      "epoch": 5.0347724620770125,
      "grad_norm": 0.3125,
      "learning_rate": 9.894352889673086e-05,
      "loss": 0.0025,
      "step": 10787
    },
    {
      "epoch": 5.035239206534422,
      "grad_norm": 0.55859375,
      "learning_rate": 9.892885624432131e-05,
      "loss": 0.0039,
      "step": 10788
    },
    {
      "epoch": 5.035705950991832,
      "grad_norm": 0.6171875,
      "learning_rate": 9.89141836149747e-05,
      "loss": 0.0059,
      "step": 10789
    },
    {
      "epoch": 5.036172695449242,
      "grad_norm": 0.462890625,
      "learning_rate": 9.889951100900692e-05,
      "loss": 0.0105,
      "step": 10790
    },
    {
      "epoch": 5.036639439906651,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.888483842673388e-05,
      "loss": 0.0023,
      "step": 10791
    },
    {
      "epoch": 5.0371061843640605,
      "grad_norm": 0.2265625,
      "learning_rate": 9.88701658684715e-05,
      "loss": 0.0024,
      "step": 10792
    },
    {
      "epoch": 5.03757292882147,
      "grad_norm": 0.4375,
      "learning_rate": 9.885549333453571e-05,
      "loss": 0.0057,
      "step": 10793
    },
    {
      "epoch": 5.03803967327888,
      "grad_norm": 0.1416015625,
      "learning_rate": 9.884082082524243e-05,
      "loss": 0.0045,
      "step": 10794
    },
    {
      "epoch": 5.038506417736289,
      "grad_norm": 0.169921875,
      "learning_rate": 9.882614834090753e-05,
      "loss": 0.0027,
      "step": 10795
    },
    {
      "epoch": 5.038973162193699,
      "grad_norm": 0.50390625,
      "learning_rate": 9.881147588184698e-05,
      "loss": 0.0056,
      "step": 10796
    },
    {
      "epoch": 5.039439906651109,
      "grad_norm": 0.0849609375,
      "learning_rate": 9.879680344837668e-05,
      "loss": 0.0011,
      "step": 10797
    },
    {
      "epoch": 5.039906651108518,
      "grad_norm": 0.337890625,
      "learning_rate": 9.878213104081252e-05,
      "loss": 0.0029,
      "step": 10798
    },
    {
      "epoch": 5.040373395565927,
      "grad_norm": 0.34765625,
      "learning_rate": 9.876745865947044e-05,
      "loss": 0.0022,
      "step": 10799
    },
    {
      "epoch": 5.040840140023337,
      "grad_norm": 0.291015625,
      "learning_rate": 9.875278630466635e-05,
      "loss": 0.002,
      "step": 10800
    },
    {
      "epoch": 5.041306884480747,
      "grad_norm": 0.42578125,
      "learning_rate": 9.873811397671616e-05,
      "loss": 0.0065,
      "step": 10801
    },
    {
      "epoch": 5.041773628938157,
      "grad_norm": 0.369140625,
      "learning_rate": 9.872344167593574e-05,
      "loss": 0.0052,
      "step": 10802
    },
    {
      "epoch": 5.042240373395566,
      "grad_norm": 0.359375,
      "learning_rate": 9.870876940264106e-05,
      "loss": 0.0035,
      "step": 10803
    },
    {
      "epoch": 5.042707117852975,
      "grad_norm": 0.1748046875,
      "learning_rate": 9.869409715714802e-05,
      "loss": 0.0023,
      "step": 10804
    },
    {
      "epoch": 5.043173862310385,
      "grad_norm": 0.224609375,
      "learning_rate": 9.867942493977249e-05,
      "loss": 0.0016,
      "step": 10805
    },
    {
      "epoch": 5.043640606767795,
      "grad_norm": 0.3359375,
      "learning_rate": 9.866475275083043e-05,
      "loss": 0.004,
      "step": 10806
    },
    {
      "epoch": 5.044107351225204,
      "grad_norm": 0.482421875,
      "learning_rate": 9.865008059063773e-05,
      "loss": 0.0047,
      "step": 10807
    },
    {
      "epoch": 5.044574095682614,
      "grad_norm": 0.3046875,
      "learning_rate": 9.863540845951025e-05,
      "loss": 0.0049,
      "step": 10808
    },
    {
      "epoch": 5.0450408401400235,
      "grad_norm": 0.306640625,
      "learning_rate": 9.862073635776398e-05,
      "loss": 0.0068,
      "step": 10809
    },
    {
      "epoch": 5.045507584597433,
      "grad_norm": 0.259765625,
      "learning_rate": 9.860606428571478e-05,
      "loss": 0.0111,
      "step": 10810
    },
    {
      "epoch": 5.045974329054842,
      "grad_norm": 0.27734375,
      "learning_rate": 9.859139224367856e-05,
      "loss": 0.0025,
      "step": 10811
    },
    {
      "epoch": 5.046441073512252,
      "grad_norm": 0.29296875,
      "learning_rate": 9.85767202319712e-05,
      "loss": 0.0033,
      "step": 10812
    },
    {
      "epoch": 5.046907817969662,
      "grad_norm": 0.169921875,
      "learning_rate": 9.856204825090868e-05,
      "loss": 0.0016,
      "step": 10813
    },
    {
      "epoch": 5.047374562427072,
      "grad_norm": 0.28125,
      "learning_rate": 9.854737630080684e-05,
      "loss": 0.0024,
      "step": 10814
    },
    {
      "epoch": 5.0478413068844805,
      "grad_norm": 0.359375,
      "learning_rate": 9.853270438198157e-05,
      "loss": 0.0041,
      "step": 10815
    },
    {
      "epoch": 5.04830805134189,
      "grad_norm": 0.1494140625,
      "learning_rate": 9.851803249474883e-05,
      "loss": 0.0012,
      "step": 10816
    },
    {
      "epoch": 5.0487747957993,
      "grad_norm": 0.380859375,
      "learning_rate": 9.850336063942451e-05,
      "loss": 0.0025,
      "step": 10817
    },
    {
      "epoch": 5.049241540256709,
      "grad_norm": 0.2734375,
      "learning_rate": 9.848868881632446e-05,
      "loss": 0.0027,
      "step": 10818
    },
    {
      "epoch": 5.049708284714119,
      "grad_norm": 0.326171875,
      "learning_rate": 9.847401702576465e-05,
      "loss": 0.0031,
      "step": 10819
    },
    {
      "epoch": 5.050175029171529,
      "grad_norm": 0.267578125,
      "learning_rate": 9.845934526806093e-05,
      "loss": 0.0028,
      "step": 10820
    },
    {
      "epoch": 5.050641773628938,
      "grad_norm": 0.2109375,
      "learning_rate": 9.844467354352921e-05,
      "loss": 0.0025,
      "step": 10821
    },
    {
      "epoch": 5.051108518086347,
      "grad_norm": 0.466796875,
      "learning_rate": 9.84300018524854e-05,
      "loss": 0.0075,
      "step": 10822
    },
    {
      "epoch": 5.051575262543757,
      "grad_norm": 0.32421875,
      "learning_rate": 9.84153301952454e-05,
      "loss": 0.0027,
      "step": 10823
    },
    {
      "epoch": 5.052042007001167,
      "grad_norm": 0.458984375,
      "learning_rate": 9.840065857212509e-05,
      "loss": 0.0039,
      "step": 10824
    },
    {
      "epoch": 5.052508751458577,
      "grad_norm": 0.056640625,
      "learning_rate": 9.838598698344036e-05,
      "loss": 0.0007,
      "step": 10825
    },
    {
      "epoch": 5.052975495915986,
      "grad_norm": 0.330078125,
      "learning_rate": 9.837131542950715e-05,
      "loss": 0.0051,
      "step": 10826
    },
    {
      "epoch": 5.053442240373395,
      "grad_norm": 0.486328125,
      "learning_rate": 9.835664391064132e-05,
      "loss": 0.0077,
      "step": 10827
    },
    {
      "epoch": 5.053908984830805,
      "grad_norm": 0.10888671875,
      "learning_rate": 9.834197242715877e-05,
      "loss": 0.0014,
      "step": 10828
    },
    {
      "epoch": 5.054375729288215,
      "grad_norm": 0.10302734375,
      "learning_rate": 9.832730097937537e-05,
      "loss": 0.0012,
      "step": 10829
    },
    {
      "epoch": 5.054842473745624,
      "grad_norm": 0.251953125,
      "learning_rate": 9.831262956760706e-05,
      "loss": 0.0034,
      "step": 10830
    },
    {
      "epoch": 5.055309218203034,
      "grad_norm": 0.390625,
      "learning_rate": 9.82979581921697e-05,
      "loss": 0.0043,
      "step": 10831
    },
    {
      "epoch": 5.0557759626604435,
      "grad_norm": 0.248046875,
      "learning_rate": 9.828328685337919e-05,
      "loss": 0.0023,
      "step": 10832
    },
    {
      "epoch": 5.056242707117853,
      "grad_norm": 0.26953125,
      "learning_rate": 9.826861555155141e-05,
      "loss": 0.0026,
      "step": 10833
    },
    {
      "epoch": 5.056709451575262,
      "grad_norm": 0.41015625,
      "learning_rate": 9.825394428700228e-05,
      "loss": 0.0109,
      "step": 10834
    },
    {
      "epoch": 5.057176196032672,
      "grad_norm": 0.439453125,
      "learning_rate": 9.823927306004768e-05,
      "loss": 0.0044,
      "step": 10835
    },
    {
      "epoch": 5.057642940490082,
      "grad_norm": 0.1123046875,
      "learning_rate": 9.822460187100346e-05,
      "loss": 0.0011,
      "step": 10836
    },
    {
      "epoch": 5.058109684947492,
      "grad_norm": 0.400390625,
      "learning_rate": 9.820993072018551e-05,
      "loss": 0.0043,
      "step": 10837
    },
    {
      "epoch": 5.0585764294049005,
      "grad_norm": 0.2373046875,
      "learning_rate": 9.819525960790977e-05,
      "loss": 0.0021,
      "step": 10838
    },
    {
      "epoch": 5.05904317386231,
      "grad_norm": 0.408203125,
      "learning_rate": 9.81805885344921e-05,
      "loss": 0.0061,
      "step": 10839
    },
    {
      "epoch": 5.05950991831972,
      "grad_norm": 0.267578125,
      "learning_rate": 9.816591750024835e-05,
      "loss": 0.004,
      "step": 10840
    },
    {
      "epoch": 5.05997666277713,
      "grad_norm": 0.3125,
      "learning_rate": 9.815124650549442e-05,
      "loss": 0.003,
      "step": 10841
    },
    {
      "epoch": 5.060443407234539,
      "grad_norm": 0.3515625,
      "learning_rate": 9.813657555054622e-05,
      "loss": 0.004,
      "step": 10842
    },
    {
      "epoch": 5.060910151691949,
      "grad_norm": 0.390625,
      "learning_rate": 9.812190463571964e-05,
      "loss": 0.0028,
      "step": 10843
    },
    {
      "epoch": 5.061376896149358,
      "grad_norm": 0.203125,
      "learning_rate": 9.81072337613305e-05,
      "loss": 0.0021,
      "step": 10844
    },
    {
      "epoch": 5.061843640606768,
      "grad_norm": 0.1455078125,
      "learning_rate": 9.809256292769473e-05,
      "loss": 0.0048,
      "step": 10845
    },
    {
      "epoch": 5.062310385064177,
      "grad_norm": 0.271484375,
      "learning_rate": 9.807789213512821e-05,
      "loss": 0.0025,
      "step": 10846
    },
    {
      "epoch": 5.062777129521587,
      "grad_norm": 0.4375,
      "learning_rate": 9.806322138394678e-05,
      "loss": 0.0093,
      "step": 10847
    },
    {
      "epoch": 5.063243873978997,
      "grad_norm": 0.453125,
      "learning_rate": 9.804855067446636e-05,
      "loss": 0.0098,
      "step": 10848
    },
    {
      "epoch": 5.0637106184364065,
      "grad_norm": 0.33203125,
      "learning_rate": 9.803388000700282e-05,
      "loss": 0.0024,
      "step": 10849
    },
    {
      "epoch": 5.064177362893815,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.801920938187201e-05,
      "loss": 0.0018,
      "step": 10850
    },
    {
      "epoch": 5.064644107351225,
      "grad_norm": 0.197265625,
      "learning_rate": 9.800453879938981e-05,
      "loss": 0.0049,
      "step": 10851
    },
    {
      "epoch": 5.065110851808635,
      "grad_norm": 0.53515625,
      "learning_rate": 9.798986825987212e-05,
      "loss": 0.0108,
      "step": 10852
    },
    {
      "epoch": 5.065577596266045,
      "grad_norm": 0.29296875,
      "learning_rate": 9.79751977636348e-05,
      "loss": 0.003,
      "step": 10853
    },
    {
      "epoch": 5.066044340723454,
      "grad_norm": 0.44140625,
      "learning_rate": 9.796052731099369e-05,
      "loss": 0.0037,
      "step": 10854
    },
    {
      "epoch": 5.0665110851808635,
      "grad_norm": 0.49609375,
      "learning_rate": 9.79458569022647e-05,
      "loss": 0.004,
      "step": 10855
    },
    {
      "epoch": 5.066977829638273,
      "grad_norm": 0.458984375,
      "learning_rate": 9.793118653776372e-05,
      "loss": 0.0065,
      "step": 10856
    },
    {
      "epoch": 5.067444574095683,
      "grad_norm": 0.23828125,
      "learning_rate": 9.791651621780655e-05,
      "loss": 0.0099,
      "step": 10857
    },
    {
      "epoch": 5.067911318553092,
      "grad_norm": 0.435546875,
      "learning_rate": 9.790184594270912e-05,
      "loss": 0.0039,
      "step": 10858
    },
    {
      "epoch": 5.068378063010502,
      "grad_norm": 0.2412109375,
      "learning_rate": 9.788717571278728e-05,
      "loss": 0.0014,
      "step": 10859
    },
    {
      "epoch": 5.0688448074679116,
      "grad_norm": 0.20703125,
      "learning_rate": 9.78725055283569e-05,
      "loss": 0.0012,
      "step": 10860
    },
    {
      "epoch": 5.0693115519253205,
      "grad_norm": 0.248046875,
      "learning_rate": 9.785783538973378e-05,
      "loss": 0.0052,
      "step": 10861
    },
    {
      "epoch": 5.06977829638273,
      "grad_norm": 0.3828125,
      "learning_rate": 9.78431652972339e-05,
      "loss": 0.0046,
      "step": 10862
    },
    {
      "epoch": 5.07024504084014,
      "grad_norm": 0.400390625,
      "learning_rate": 9.782849525117305e-05,
      "loss": 0.0051,
      "step": 10863
    },
    {
      "epoch": 5.07071178529755,
      "grad_norm": 0.263671875,
      "learning_rate": 9.781382525186708e-05,
      "loss": 0.0041,
      "step": 10864
    },
    {
      "epoch": 5.071178529754959,
      "grad_norm": 0.259765625,
      "learning_rate": 9.779915529963191e-05,
      "loss": 0.0025,
      "step": 10865
    },
    {
      "epoch": 5.0716452742123685,
      "grad_norm": 0.283203125,
      "learning_rate": 9.778448539478335e-05,
      "loss": 0.0036,
      "step": 10866
    },
    {
      "epoch": 5.072112018669778,
      "grad_norm": 0.55078125,
      "learning_rate": 9.776981553763728e-05,
      "loss": 0.0061,
      "step": 10867
    },
    {
      "epoch": 5.072578763127188,
      "grad_norm": 0.12109375,
      "learning_rate": 9.775514572850957e-05,
      "loss": 0.0012,
      "step": 10868
    },
    {
      "epoch": 5.073045507584597,
      "grad_norm": 0.134765625,
      "learning_rate": 9.774047596771606e-05,
      "loss": 0.0012,
      "step": 10869
    },
    {
      "epoch": 5.073512252042007,
      "grad_norm": 0.1640625,
      "learning_rate": 9.77258062555726e-05,
      "loss": 0.0013,
      "step": 10870
    },
    {
      "epoch": 5.073978996499417,
      "grad_norm": 0.427734375,
      "learning_rate": 9.771113659239504e-05,
      "loss": 0.0045,
      "step": 10871
    },
    {
      "epoch": 5.074445740956826,
      "grad_norm": 0.306640625,
      "learning_rate": 9.769646697849927e-05,
      "loss": 0.0019,
      "step": 10872
    },
    {
      "epoch": 5.074912485414235,
      "grad_norm": 0.1552734375,
      "learning_rate": 9.768179741420113e-05,
      "loss": 0.0014,
      "step": 10873
    },
    {
      "epoch": 5.075379229871645,
      "grad_norm": 0.62890625,
      "learning_rate": 9.766712789981644e-05,
      "loss": 0.0027,
      "step": 10874
    },
    {
      "epoch": 5.075845974329055,
      "grad_norm": 0.40234375,
      "learning_rate": 9.765245843566109e-05,
      "loss": 0.0058,
      "step": 10875
    },
    {
      "epoch": 5.076312718786465,
      "grad_norm": 0.1484375,
      "learning_rate": 9.76377890220509e-05,
      "loss": 0.0015,
      "step": 10876
    },
    {
      "epoch": 5.076779463243874,
      "grad_norm": 0.2421875,
      "learning_rate": 9.762311965930174e-05,
      "loss": 0.0023,
      "step": 10877
    },
    {
      "epoch": 5.077246207701283,
      "grad_norm": 0.15234375,
      "learning_rate": 9.760845034772946e-05,
      "loss": 0.0017,
      "step": 10878
    },
    {
      "epoch": 5.077712952158693,
      "grad_norm": 0.2021484375,
      "learning_rate": 9.75937810876499e-05,
      "loss": 0.0048,
      "step": 10879
    },
    {
      "epoch": 5.078179696616103,
      "grad_norm": 0.447265625,
      "learning_rate": 9.75791118793789e-05,
      "loss": 0.0092,
      "step": 10880
    },
    {
      "epoch": 5.078646441073512,
      "grad_norm": 0.08984375,
      "learning_rate": 9.756444272323229e-05,
      "loss": 0.0012,
      "step": 10881
    },
    {
      "epoch": 5.079113185530922,
      "grad_norm": 0.2734375,
      "learning_rate": 9.754977361952594e-05,
      "loss": 0.0033,
      "step": 10882
    },
    {
      "epoch": 5.0795799299883315,
      "grad_norm": 0.4765625,
      "learning_rate": 9.753510456857572e-05,
      "loss": 0.0048,
      "step": 10883
    },
    {
      "epoch": 5.080046674445741,
      "grad_norm": 0.2275390625,
      "learning_rate": 9.75204355706974e-05,
      "loss": 0.0027,
      "step": 10884
    },
    {
      "epoch": 5.08051341890315,
      "grad_norm": 0.35546875,
      "learning_rate": 9.750576662620686e-05,
      "loss": 0.0036,
      "step": 10885
    },
    {
      "epoch": 5.08098016336056,
      "grad_norm": 0.404296875,
      "learning_rate": 9.749109773541991e-05,
      "loss": 0.0059,
      "step": 10886
    },
    {
      "epoch": 5.08144690781797,
      "grad_norm": 0.265625,
      "learning_rate": 9.747642889865243e-05,
      "loss": 0.0036,
      "step": 10887
    },
    {
      "epoch": 5.08191365227538,
      "grad_norm": 0.291015625,
      "learning_rate": 9.746176011622023e-05,
      "loss": 0.0032,
      "step": 10888
    },
    {
      "epoch": 5.0823803967327885,
      "grad_norm": 0.193359375,
      "learning_rate": 9.744709138843914e-05,
      "loss": 0.0018,
      "step": 10889
    },
    {
      "epoch": 5.082847141190198,
      "grad_norm": 0.41796875,
      "learning_rate": 9.743242271562502e-05,
      "loss": 0.0037,
      "step": 10890
    },
    {
      "epoch": 5.083313885647608,
      "grad_norm": 0.326171875,
      "learning_rate": 9.74177540980937e-05,
      "loss": 0.0025,
      "step": 10891
    },
    {
      "epoch": 5.083780630105018,
      "grad_norm": 0.1669921875,
      "learning_rate": 9.740308553616097e-05,
      "loss": 0.0027,
      "step": 10892
    },
    {
      "epoch": 5.084247374562427,
      "grad_norm": 0.26171875,
      "learning_rate": 9.73884170301427e-05,
      "loss": 0.0019,
      "step": 10893
    },
    {
      "epoch": 5.084714119019837,
      "grad_norm": 0.2451171875,
      "learning_rate": 9.73737485803547e-05,
      "loss": 0.0016,
      "step": 10894
    },
    {
      "epoch": 5.085180863477246,
      "grad_norm": 0.6328125,
      "learning_rate": 9.735908018711284e-05,
      "loss": 0.0088,
      "step": 10895
    },
    {
      "epoch": 5.085647607934656,
      "grad_norm": 0.369140625,
      "learning_rate": 9.734441185073286e-05,
      "loss": 0.006,
      "step": 10896
    },
    {
      "epoch": 5.086114352392065,
      "grad_norm": 0.1669921875,
      "learning_rate": 9.732974357153069e-05,
      "loss": 0.0045,
      "step": 10897
    },
    {
      "epoch": 5.086581096849475,
      "grad_norm": 0.18359375,
      "learning_rate": 9.731507534982208e-05,
      "loss": 0.0011,
      "step": 10898
    },
    {
      "epoch": 5.087047841306885,
      "grad_norm": 0.12890625,
      "learning_rate": 9.73004071859229e-05,
      "loss": 0.0013,
      "step": 10899
    },
    {
      "epoch": 5.0875145857642945,
      "grad_norm": 0.453125,
      "learning_rate": 9.72857390801489e-05,
      "loss": 0.0082,
      "step": 10900
    },
    {
      "epoch": 5.087981330221703,
      "grad_norm": 0.306640625,
      "learning_rate": 9.727107103281599e-05,
      "loss": 0.0026,
      "step": 10901
    },
    {
      "epoch": 5.088448074679113,
      "grad_norm": 0.150390625,
      "learning_rate": 9.725640304423994e-05,
      "loss": 0.0017,
      "step": 10902
    },
    {
      "epoch": 5.088914819136523,
      "grad_norm": 0.42578125,
      "learning_rate": 9.724173511473657e-05,
      "loss": 0.0044,
      "step": 10903
    },
    {
      "epoch": 5.089381563593932,
      "grad_norm": 0.28515625,
      "learning_rate": 9.72270672446217e-05,
      "loss": 0.0021,
      "step": 10904
    },
    {
      "epoch": 5.089848308051342,
      "grad_norm": 0.0888671875,
      "learning_rate": 9.721239943421116e-05,
      "loss": 0.0012,
      "step": 10905
    },
    {
      "epoch": 5.0903150525087515,
      "grad_norm": 0.53515625,
      "learning_rate": 9.719773168382073e-05,
      "loss": 0.005,
      "step": 10906
    },
    {
      "epoch": 5.090781796966161,
      "grad_norm": 0.40625,
      "learning_rate": 9.71830639937663e-05,
      "loss": 0.0021,
      "step": 10907
    },
    {
      "epoch": 5.09124854142357,
      "grad_norm": 0.1416015625,
      "learning_rate": 9.716839636436357e-05,
      "loss": 0.0011,
      "step": 10908
    },
    {
      "epoch": 5.09171528588098,
      "grad_norm": 0.365234375,
      "learning_rate": 9.715372879592844e-05,
      "loss": 0.0033,
      "step": 10909
    },
    {
      "epoch": 5.09218203033839,
      "grad_norm": 0.08642578125,
      "learning_rate": 9.713906128877667e-05,
      "loss": 0.001,
      "step": 10910
    },
    {
      "epoch": 5.0926487747958,
      "grad_norm": 0.3984375,
      "learning_rate": 9.712439384322409e-05,
      "loss": 0.0145,
      "step": 10911
    },
    {
      "epoch": 5.0931155192532085,
      "grad_norm": 0.232421875,
      "learning_rate": 9.710972645958652e-05,
      "loss": 0.0029,
      "step": 10912
    },
    {
      "epoch": 5.093582263710618,
      "grad_norm": 0.427734375,
      "learning_rate": 9.709505913817971e-05,
      "loss": 0.0035,
      "step": 10913
    },
    {
      "epoch": 5.094049008168028,
      "grad_norm": 0.10400390625,
      "learning_rate": 9.708039187931953e-05,
      "loss": 0.0013,
      "step": 10914
    },
    {
      "epoch": 5.094515752625438,
      "grad_norm": 0.267578125,
      "learning_rate": 9.706572468332174e-05,
      "loss": 0.0021,
      "step": 10915
    },
    {
      "epoch": 5.094982497082847,
      "grad_norm": 0.279296875,
      "learning_rate": 9.705105755050214e-05,
      "loss": 0.0055,
      "step": 10916
    },
    {
      "epoch": 5.095449241540257,
      "grad_norm": 0.1884765625,
      "learning_rate": 9.703639048117657e-05,
      "loss": 0.0019,
      "step": 10917
    },
    {
      "epoch": 5.095915985997666,
      "grad_norm": 0.1181640625,
      "learning_rate": 9.70217234756608e-05,
      "loss": 0.001,
      "step": 10918
    },
    {
      "epoch": 5.096382730455076,
      "grad_norm": 0.326171875,
      "learning_rate": 9.70070565342706e-05,
      "loss": 0.0041,
      "step": 10919
    },
    {
      "epoch": 5.096849474912485,
      "grad_norm": 0.185546875,
      "learning_rate": 9.699238965732181e-05,
      "loss": 0.0014,
      "step": 10920
    },
    {
      "epoch": 5.097316219369895,
      "grad_norm": 0.208984375,
      "learning_rate": 9.697772284513022e-05,
      "loss": 0.0053,
      "step": 10921
    },
    {
      "epoch": 5.097782963827305,
      "grad_norm": 0.146484375,
      "learning_rate": 9.69630560980116e-05,
      "loss": 0.0012,
      "step": 10922
    },
    {
      "epoch": 5.0982497082847145,
      "grad_norm": 0.3671875,
      "learning_rate": 9.694838941628174e-05,
      "loss": 0.0032,
      "step": 10923
    },
    {
      "epoch": 5.098716452742123,
      "grad_norm": 0.053466796875,
      "learning_rate": 9.693372280025646e-05,
      "loss": 0.0006,
      "step": 10924
    },
    {
      "epoch": 5.099183197199533,
      "grad_norm": 0.314453125,
      "learning_rate": 9.691905625025151e-05,
      "loss": 0.0045,
      "step": 10925
    },
    {
      "epoch": 5.099649941656943,
      "grad_norm": 0.080078125,
      "learning_rate": 9.69043897665827e-05,
      "loss": 0.0033,
      "step": 10926
    },
    {
      "epoch": 5.100116686114353,
      "grad_norm": 0.23046875,
      "learning_rate": 9.688972334956583e-05,
      "loss": 0.0014,
      "step": 10927
    },
    {
      "epoch": 5.100583430571762,
      "grad_norm": 0.51171875,
      "learning_rate": 9.687505699951666e-05,
      "loss": 0.0081,
      "step": 10928
    },
    {
      "epoch": 5.1010501750291715,
      "grad_norm": 0.2265625,
      "learning_rate": 9.686039071675099e-05,
      "loss": 0.0014,
      "step": 10929
    },
    {
      "epoch": 5.101516919486581,
      "grad_norm": 0.76953125,
      "learning_rate": 9.684572450158457e-05,
      "loss": 0.0036,
      "step": 10930
    },
    {
      "epoch": 5.101983663943991,
      "grad_norm": 0.41796875,
      "learning_rate": 9.683105835433326e-05,
      "loss": 0.0092,
      "step": 10931
    },
    {
      "epoch": 5.1024504084014,
      "grad_norm": 0.185546875,
      "learning_rate": 9.681639227531271e-05,
      "loss": 0.0095,
      "step": 10932
    },
    {
      "epoch": 5.10291715285881,
      "grad_norm": 0.33203125,
      "learning_rate": 9.680172626483879e-05,
      "loss": 0.0034,
      "step": 10933
    },
    {
      "epoch": 5.1033838973162196,
      "grad_norm": 0.1328125,
      "learning_rate": 9.678706032322727e-05,
      "loss": 0.005,
      "step": 10934
    },
    {
      "epoch": 5.103850641773629,
      "grad_norm": 0.314453125,
      "learning_rate": 9.677239445079387e-05,
      "loss": 0.0024,
      "step": 10935
    },
    {
      "epoch": 5.104317386231038,
      "grad_norm": 0.412109375,
      "learning_rate": 9.675772864785442e-05,
      "loss": 0.0074,
      "step": 10936
    },
    {
      "epoch": 5.104784130688448,
      "grad_norm": 0.27734375,
      "learning_rate": 9.674306291472466e-05,
      "loss": 0.0021,
      "step": 10937
    },
    {
      "epoch": 5.105250875145858,
      "grad_norm": 0.255859375,
      "learning_rate": 9.672839725172037e-05,
      "loss": 0.0029,
      "step": 10938
    },
    {
      "epoch": 5.105717619603268,
      "grad_norm": 0.25390625,
      "learning_rate": 9.671373165915733e-05,
      "loss": 0.002,
      "step": 10939
    },
    {
      "epoch": 5.1061843640606766,
      "grad_norm": 0.337890625,
      "learning_rate": 9.669906613735128e-05,
      "loss": 0.004,
      "step": 10940
    },
    {
      "epoch": 5.106651108518086,
      "grad_norm": 0.357421875,
      "learning_rate": 9.668440068661801e-05,
      "loss": 0.0028,
      "step": 10941
    },
    {
      "epoch": 5.107117852975496,
      "grad_norm": 0.267578125,
      "learning_rate": 9.666973530727326e-05,
      "loss": 0.005,
      "step": 10942
    },
    {
      "epoch": 5.107584597432905,
      "grad_norm": 0.2265625,
      "learning_rate": 9.66550699996328e-05,
      "loss": 0.0019,
      "step": 10943
    },
    {
      "epoch": 5.108051341890315,
      "grad_norm": 0.251953125,
      "learning_rate": 9.664040476401241e-05,
      "loss": 0.0019,
      "step": 10944
    },
    {
      "epoch": 5.108518086347725,
      "grad_norm": 0.11962890625,
      "learning_rate": 9.662573960072782e-05,
      "loss": 0.0012,
      "step": 10945
    },
    {
      "epoch": 5.108984830805134,
      "grad_norm": 0.154296875,
      "learning_rate": 9.661107451009482e-05,
      "loss": 0.0034,
      "step": 10946
    },
    {
      "epoch": 5.109451575262543,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.659640949242912e-05,
      "loss": 0.0017,
      "step": 10947
    },
    {
      "epoch": 5.109918319719953,
      "grad_norm": 0.4140625,
      "learning_rate": 9.658174454804651e-05,
      "loss": 0.0058,
      "step": 10948
    },
    {
      "epoch": 5.110385064177363,
      "grad_norm": 0.515625,
      "learning_rate": 9.656707967726272e-05,
      "loss": 0.0054,
      "step": 10949
    },
    {
      "epoch": 5.110851808634773,
      "grad_norm": 0.126953125,
      "learning_rate": 9.655241488039356e-05,
      "loss": 0.0037,
      "step": 10950
    },
    {
      "epoch": 5.111318553092182,
      "grad_norm": 0.451171875,
      "learning_rate": 9.65377501577547e-05,
      "loss": 0.0048,
      "step": 10951
    },
    {
      "epoch": 5.111785297549591,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.652308550966189e-05,
      "loss": 0.0015,
      "step": 10952
    },
    {
      "epoch": 5.112252042007001,
      "grad_norm": 0.201171875,
      "learning_rate": 9.650842093643095e-05,
      "loss": 0.002,
      "step": 10953
    },
    {
      "epoch": 5.112718786464411,
      "grad_norm": 0.4375,
      "learning_rate": 9.649375643837758e-05,
      "loss": 0.003,
      "step": 10954
    },
    {
      "epoch": 5.11318553092182,
      "grad_norm": 0.4375,
      "learning_rate": 9.64790920158175e-05,
      "loss": 0.0069,
      "step": 10955
    },
    {
      "epoch": 5.11365227537923,
      "grad_norm": 0.3046875,
      "learning_rate": 9.646442766906649e-05,
      "loss": 0.0045,
      "step": 10956
    },
    {
      "epoch": 5.1141190198366395,
      "grad_norm": 0.162109375,
      "learning_rate": 9.644976339844027e-05,
      "loss": 0.0012,
      "step": 10957
    },
    {
      "epoch": 5.114585764294049,
      "grad_norm": 0.24609375,
      "learning_rate": 9.64350992042546e-05,
      "loss": 0.0029,
      "step": 10958
    },
    {
      "epoch": 5.115052508751458,
      "grad_norm": 0.3828125,
      "learning_rate": 9.642043508682517e-05,
      "loss": 0.0052,
      "step": 10959
    },
    {
      "epoch": 5.115519253208868,
      "grad_norm": 0.455078125,
      "learning_rate": 9.640577104646776e-05,
      "loss": 0.0065,
      "step": 10960
    },
    {
      "epoch": 5.115985997666278,
      "grad_norm": 0.29296875,
      "learning_rate": 9.63911070834981e-05,
      "loss": 0.0023,
      "step": 10961
    },
    {
      "epoch": 5.116452742123688,
      "grad_norm": 0.5703125,
      "learning_rate": 9.637644319823189e-05,
      "loss": 0.0051,
      "step": 10962
    },
    {
      "epoch": 5.1169194865810965,
      "grad_norm": 1.109375,
      "learning_rate": 9.636177939098489e-05,
      "loss": 0.012,
      "step": 10963
    },
    {
      "epoch": 5.117386231038506,
      "grad_norm": 0.1103515625,
      "learning_rate": 9.63471156620728e-05,
      "loss": 0.0018,
      "step": 10964
    },
    {
      "epoch": 5.117852975495916,
      "grad_norm": 0.46484375,
      "learning_rate": 9.633245201181137e-05,
      "loss": 0.0143,
      "step": 10965
    },
    {
      "epoch": 5.118319719953326,
      "grad_norm": 0.326171875,
      "learning_rate": 9.631778844051632e-05,
      "loss": 0.0049,
      "step": 10966
    },
    {
      "epoch": 5.118786464410735,
      "grad_norm": 0.2060546875,
      "learning_rate": 9.630312494850337e-05,
      "loss": 0.002,
      "step": 10967
    },
    {
      "epoch": 5.119253208868145,
      "grad_norm": 0.359375,
      "learning_rate": 9.628846153608826e-05,
      "loss": 0.0032,
      "step": 10968
    },
    {
      "epoch": 5.119719953325554,
      "grad_norm": 0.140625,
      "learning_rate": 9.627379820358666e-05,
      "loss": 0.0013,
      "step": 10969
    },
    {
      "epoch": 5.120186697782964,
      "grad_norm": 0.220703125,
      "learning_rate": 9.625913495131434e-05,
      "loss": 0.0025,
      "step": 10970
    },
    {
      "epoch": 5.120653442240373,
      "grad_norm": 0.3125,
      "learning_rate": 9.624447177958699e-05,
      "loss": 0.0059,
      "step": 10971
    },
    {
      "epoch": 5.121120186697783,
      "grad_norm": 0.3359375,
      "learning_rate": 9.622980868872031e-05,
      "loss": 0.0068,
      "step": 10972
    },
    {
      "epoch": 5.121586931155193,
      "grad_norm": 0.287109375,
      "learning_rate": 9.621514567903006e-05,
      "loss": 0.0033,
      "step": 10973
    },
    {
      "epoch": 5.1220536756126025,
      "grad_norm": 0.40625,
      "learning_rate": 9.620048275083192e-05,
      "loss": 0.0073,
      "step": 10974
    },
    {
      "epoch": 5.122520420070011,
      "grad_norm": 0.1572265625,
      "learning_rate": 9.618581990444158e-05,
      "loss": 0.0015,
      "step": 10975
    },
    {
      "epoch": 5.122987164527421,
      "grad_norm": 0.2109375,
      "learning_rate": 9.617115714017479e-05,
      "loss": 0.0014,
      "step": 10976
    },
    {
      "epoch": 5.123453908984831,
      "grad_norm": 0.359375,
      "learning_rate": 9.615649445834724e-05,
      "loss": 0.0022,
      "step": 10977
    },
    {
      "epoch": 5.123920653442241,
      "grad_norm": 0.26953125,
      "learning_rate": 9.614183185927463e-05,
      "loss": 0.0024,
      "step": 10978
    },
    {
      "epoch": 5.12438739789965,
      "grad_norm": 0.1484375,
      "learning_rate": 9.612716934327266e-05,
      "loss": 0.0016,
      "step": 10979
    },
    {
      "epoch": 5.1248541423570595,
      "grad_norm": 0.5390625,
      "learning_rate": 9.611250691065704e-05,
      "loss": 0.0162,
      "step": 10980
    },
    {
      "epoch": 5.125320886814469,
      "grad_norm": 0.333984375,
      "learning_rate": 9.609784456174341e-05,
      "loss": 0.0046,
      "step": 10981
    },
    {
      "epoch": 5.125787631271878,
      "grad_norm": 0.369140625,
      "learning_rate": 9.608318229684755e-05,
      "loss": 0.0038,
      "step": 10982
    },
    {
      "epoch": 5.126254375729288,
      "grad_norm": 0.55859375,
      "learning_rate": 9.606852011628513e-05,
      "loss": 0.0049,
      "step": 10983
    },
    {
      "epoch": 5.126721120186698,
      "grad_norm": 0.1904296875,
      "learning_rate": 9.60538580203718e-05,
      "loss": 0.0022,
      "step": 10984
    },
    {
      "epoch": 5.127187864644108,
      "grad_norm": 0.51171875,
      "learning_rate": 9.60391960094233e-05,
      "loss": 0.0126,
      "step": 10985
    },
    {
      "epoch": 5.1276546091015165,
      "grad_norm": 0.298828125,
      "learning_rate": 9.602453408375533e-05,
      "loss": 0.0031,
      "step": 10986
    },
    {
      "epoch": 5.128121353558926,
      "grad_norm": 0.412109375,
      "learning_rate": 9.600987224368351e-05,
      "loss": 0.0028,
      "step": 10987
    },
    {
      "epoch": 5.128588098016336,
      "grad_norm": 0.310546875,
      "learning_rate": 9.599521048952359e-05,
      "loss": 0.0038,
      "step": 10988
    },
    {
      "epoch": 5.129054842473746,
      "grad_norm": 0.328125,
      "learning_rate": 9.598054882159124e-05,
      "loss": 0.009,
      "step": 10989
    },
    {
      "epoch": 5.129521586931155,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.596588724020213e-05,
      "loss": 0.0018,
      "step": 10990
    },
    {
      "epoch": 5.129988331388565,
      "grad_norm": 0.25390625,
      "learning_rate": 9.595122574567193e-05,
      "loss": 0.002,
      "step": 10991
    },
    {
      "epoch": 5.130455075845974,
      "grad_norm": 0.1328125,
      "learning_rate": 9.593656433831634e-05,
      "loss": 0.0012,
      "step": 10992
    },
    {
      "epoch": 5.130921820303384,
      "grad_norm": 0.328125,
      "learning_rate": 9.592190301845104e-05,
      "loss": 0.003,
      "step": 10993
    },
    {
      "epoch": 5.131388564760793,
      "grad_norm": 0.3203125,
      "learning_rate": 9.590724178639165e-05,
      "loss": 0.0027,
      "step": 10994
    },
    {
      "epoch": 5.131855309218203,
      "grad_norm": 0.375,
      "learning_rate": 9.589258064245393e-05,
      "loss": 0.0045,
      "step": 10995
    },
    {
      "epoch": 5.132322053675613,
      "grad_norm": 0.2734375,
      "learning_rate": 9.58779195869535e-05,
      "loss": 0.0077,
      "step": 10996
    },
    {
      "epoch": 5.1327887981330225,
      "grad_norm": 0.2119140625,
      "learning_rate": 9.586325862020601e-05,
      "loss": 0.0023,
      "step": 10997
    },
    {
      "epoch": 5.133255542590431,
      "grad_norm": 0.318359375,
      "learning_rate": 9.584859774252717e-05,
      "loss": 0.002,
      "step": 10998
    },
    {
      "epoch": 5.133722287047841,
      "grad_norm": 0.138671875,
      "learning_rate": 9.583393695423264e-05,
      "loss": 0.0023,
      "step": 10999
    },
    {
      "epoch": 5.134189031505251,
      "grad_norm": 0.11572265625,
      "learning_rate": 9.581927625563806e-05,
      "loss": 0.0011,
      "step": 11000
    },
    {
      "epoch": 5.134655775962661,
      "grad_norm": 0.236328125,
      "learning_rate": 9.580461564705908e-05,
      "loss": 0.0026,
      "step": 11001
    },
    {
      "epoch": 5.13512252042007,
      "grad_norm": 0.44140625,
      "learning_rate": 9.57899551288114e-05,
      "loss": 0.0032,
      "step": 11002
    },
    {
      "epoch": 5.1355892648774795,
      "grad_norm": 0.158203125,
      "learning_rate": 9.577529470121066e-05,
      "loss": 0.0016,
      "step": 11003
    },
    {
      "epoch": 5.136056009334889,
      "grad_norm": 0.1962890625,
      "learning_rate": 9.576063436457249e-05,
      "loss": 0.0018,
      "step": 11004
    },
    {
      "epoch": 5.136522753792299,
      "grad_norm": 0.0751953125,
      "learning_rate": 9.574597411921259e-05,
      "loss": 0.0011,
      "step": 11005
    },
    {
      "epoch": 5.136989498249708,
      "grad_norm": 0.3671875,
      "learning_rate": 9.573131396544657e-05,
      "loss": 0.0032,
      "step": 11006
    },
    {
      "epoch": 5.137456242707118,
      "grad_norm": 0.353515625,
      "learning_rate": 9.57166539035901e-05,
      "loss": 0.0021,
      "step": 11007
    },
    {
      "epoch": 5.1379229871645276,
      "grad_norm": 0.173828125,
      "learning_rate": 9.570199393395883e-05,
      "loss": 0.0017,
      "step": 11008
    },
    {
      "epoch": 5.138389731621937,
      "grad_norm": 0.14453125,
      "learning_rate": 9.568733405686841e-05,
      "loss": 0.0012,
      "step": 11009
    },
    {
      "epoch": 5.138856476079346,
      "grad_norm": 0.39453125,
      "learning_rate": 9.567267427263446e-05,
      "loss": 0.0022,
      "step": 11010
    },
    {
      "epoch": 5.139323220536756,
      "grad_norm": 0.1630859375,
      "learning_rate": 9.565801458157261e-05,
      "loss": 0.0032,
      "step": 11011
    },
    {
      "epoch": 5.139789964994166,
      "grad_norm": 0.38671875,
      "learning_rate": 9.564335498399857e-05,
      "loss": 0.0132,
      "step": 11012
    },
    {
      "epoch": 5.140256709451576,
      "grad_norm": 0.166015625,
      "learning_rate": 9.562869548022792e-05,
      "loss": 0.0017,
      "step": 11013
    },
    {
      "epoch": 5.1407234539089846,
      "grad_norm": 0.2255859375,
      "learning_rate": 9.561403607057628e-05,
      "loss": 0.0019,
      "step": 11014
    },
    {
      "epoch": 5.141190198366394,
      "grad_norm": 0.1474609375,
      "learning_rate": 9.559937675535932e-05,
      "loss": 0.0014,
      "step": 11015
    },
    {
      "epoch": 5.141656942823804,
      "grad_norm": 0.41796875,
      "learning_rate": 9.558471753489268e-05,
      "loss": 0.0024,
      "step": 11016
    },
    {
      "epoch": 5.142123687281214,
      "grad_norm": 0.080078125,
      "learning_rate": 9.557005840949196e-05,
      "loss": 0.0009,
      "step": 11017
    },
    {
      "epoch": 5.142590431738623,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.555539937947277e-05,
      "loss": 0.0014,
      "step": 11018
    },
    {
      "epoch": 5.143057176196033,
      "grad_norm": 0.22265625,
      "learning_rate": 9.55407404451508e-05,
      "loss": 0.0026,
      "step": 11019
    },
    {
      "epoch": 5.143523920653442,
      "grad_norm": 0.28125,
      "learning_rate": 9.552608160684164e-05,
      "loss": 0.0023,
      "step": 11020
    },
    {
      "epoch": 5.143990665110852,
      "grad_norm": 0.11083984375,
      "learning_rate": 9.551142286486087e-05,
      "loss": 0.0039,
      "step": 11021
    },
    {
      "epoch": 5.144457409568261,
      "grad_norm": 0.0673828125,
      "learning_rate": 9.549676421952417e-05,
      "loss": 0.0008,
      "step": 11022
    },
    {
      "epoch": 5.144924154025671,
      "grad_norm": 0.34765625,
      "learning_rate": 9.548210567114714e-05,
      "loss": 0.0021,
      "step": 11023
    },
    {
      "epoch": 5.145390898483081,
      "grad_norm": 0.37890625,
      "learning_rate": 9.546744722004535e-05,
      "loss": 0.0049,
      "step": 11024
    },
    {
      "epoch": 5.14585764294049,
      "grad_norm": 0.57421875,
      "learning_rate": 9.54527888665345e-05,
      "loss": 0.0092,
      "step": 11025
    },
    {
      "epoch": 5.146324387397899,
      "grad_norm": 0.234375,
      "learning_rate": 9.543813061093014e-05,
      "loss": 0.0019,
      "step": 11026
    },
    {
      "epoch": 5.146791131855309,
      "grad_norm": 0.419921875,
      "learning_rate": 9.54234724535479e-05,
      "loss": 0.0042,
      "step": 11027
    },
    {
      "epoch": 5.147257876312719,
      "grad_norm": 0.2578125,
      "learning_rate": 9.540881439470335e-05,
      "loss": 0.0026,
      "step": 11028
    },
    {
      "epoch": 5.147724620770128,
      "grad_norm": 0.1572265625,
      "learning_rate": 9.539415643471212e-05,
      "loss": 0.0012,
      "step": 11029
    },
    {
      "epoch": 5.148191365227538,
      "grad_norm": 0.1953125,
      "learning_rate": 9.53794985738898e-05,
      "loss": 0.0047,
      "step": 11030
    },
    {
      "epoch": 5.1486581096849475,
      "grad_norm": 0.55859375,
      "learning_rate": 9.536484081255202e-05,
      "loss": 0.0087,
      "step": 11031
    },
    {
      "epoch": 5.149124854142357,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.535018315101437e-05,
      "loss": 0.0054,
      "step": 11032
    },
    {
      "epoch": 5.149591598599766,
      "grad_norm": 0.4609375,
      "learning_rate": 9.53355255895924e-05,
      "loss": 0.0065,
      "step": 11033
    },
    {
      "epoch": 5.150058343057176,
      "grad_norm": 0.458984375,
      "learning_rate": 9.532086812860178e-05,
      "loss": 0.0038,
      "step": 11034
    },
    {
      "epoch": 5.150525087514586,
      "grad_norm": 0.10400390625,
      "learning_rate": 9.530621076835805e-05,
      "loss": 0.001,
      "step": 11035
    },
    {
      "epoch": 5.150991831971996,
      "grad_norm": 0.30078125,
      "learning_rate": 9.529155350917679e-05,
      "loss": 0.0019,
      "step": 11036
    },
    {
      "epoch": 5.1514585764294045,
      "grad_norm": 0.734375,
      "learning_rate": 9.527689635137362e-05,
      "loss": 0.004,
      "step": 11037
    },
    {
      "epoch": 5.151925320886814,
      "grad_norm": 0.1943359375,
      "learning_rate": 9.526223929526412e-05,
      "loss": 0.0023,
      "step": 11038
    },
    {
      "epoch": 5.152392065344224,
      "grad_norm": 0.412109375,
      "learning_rate": 9.524758234116386e-05,
      "loss": 0.0045,
      "step": 11039
    },
    {
      "epoch": 5.152858809801634,
      "grad_norm": 0.1962890625,
      "learning_rate": 9.523292548938841e-05,
      "loss": 0.0014,
      "step": 11040
    },
    {
      "epoch": 5.153325554259043,
      "grad_norm": 0.22265625,
      "learning_rate": 9.521826874025338e-05,
      "loss": 0.0023,
      "step": 11041
    },
    {
      "epoch": 5.153792298716453,
      "grad_norm": 0.279296875,
      "learning_rate": 9.520361209407434e-05,
      "loss": 0.0029,
      "step": 11042
    },
    {
      "epoch": 5.154259043173862,
      "grad_norm": 0.11767578125,
      "learning_rate": 9.518895555116681e-05,
      "loss": 0.0014,
      "step": 11043
    },
    {
      "epoch": 5.154725787631272,
      "grad_norm": 0.208984375,
      "learning_rate": 9.517429911184644e-05,
      "loss": 0.0016,
      "step": 11044
    },
    {
      "epoch": 5.155192532088681,
      "grad_norm": 0.294921875,
      "learning_rate": 9.515964277642876e-05,
      "loss": 0.0025,
      "step": 11045
    },
    {
      "epoch": 5.155659276546091,
      "grad_norm": 0.248046875,
      "learning_rate": 9.514498654522933e-05,
      "loss": 0.0018,
      "step": 11046
    },
    {
      "epoch": 5.156126021003501,
      "grad_norm": 0.30078125,
      "learning_rate": 9.513033041856374e-05,
      "loss": 0.003,
      "step": 11047
    },
    {
      "epoch": 5.1565927654609105,
      "grad_norm": 0.2890625,
      "learning_rate": 9.511567439674754e-05,
      "loss": 0.0054,
      "step": 11048
    },
    {
      "epoch": 5.157059509918319,
      "grad_norm": 0.326171875,
      "learning_rate": 9.510101848009629e-05,
      "loss": 0.0049,
      "step": 11049
    },
    {
      "epoch": 5.157526254375729,
      "grad_norm": 0.28125,
      "learning_rate": 9.508636266892553e-05,
      "loss": 0.0029,
      "step": 11050
    },
    {
      "epoch": 5.157992998833139,
      "grad_norm": 0.0693359375,
      "learning_rate": 9.507170696355085e-05,
      "loss": 0.0007,
      "step": 11051
    },
    {
      "epoch": 5.158459743290549,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.505705136428779e-05,
      "loss": 0.0013,
      "step": 11052
    },
    {
      "epoch": 5.158926487747958,
      "grad_norm": 0.2138671875,
      "learning_rate": 9.504239587145187e-05,
      "loss": 0.0015,
      "step": 11053
    },
    {
      "epoch": 5.1593932322053675,
      "grad_norm": 0.203125,
      "learning_rate": 9.50277404853587e-05,
      "loss": 0.0024,
      "step": 11054
    },
    {
      "epoch": 5.159859976662777,
      "grad_norm": 0.41015625,
      "learning_rate": 9.501308520632379e-05,
      "loss": 0.004,
      "step": 11055
    },
    {
      "epoch": 5.160326721120187,
      "grad_norm": 0.33203125,
      "learning_rate": 9.499843003466267e-05,
      "loss": 0.0025,
      "step": 11056
    },
    {
      "epoch": 5.160793465577596,
      "grad_norm": 0.265625,
      "learning_rate": 9.498377497069091e-05,
      "loss": 0.0045,
      "step": 11057
    },
    {
      "epoch": 5.161260210035006,
      "grad_norm": 0.15625,
      "learning_rate": 9.496912001472405e-05,
      "loss": 0.0007,
      "step": 11058
    },
    {
      "epoch": 5.161726954492416,
      "grad_norm": 0.2470703125,
      "learning_rate": 9.495446516707761e-05,
      "loss": 0.0023,
      "step": 11059
    },
    {
      "epoch": 5.162193698949825,
      "grad_norm": 0.333984375,
      "learning_rate": 9.493981042806712e-05,
      "loss": 0.0023,
      "step": 11060
    },
    {
      "epoch": 5.162660443407234,
      "grad_norm": 0.41796875,
      "learning_rate": 9.492515579800815e-05,
      "loss": 0.004,
      "step": 11061
    },
    {
      "epoch": 5.163127187864644,
      "grad_norm": 0.357421875,
      "learning_rate": 9.49105012772162e-05,
      "loss": 0.0033,
      "step": 11062
    },
    {
      "epoch": 5.163593932322054,
      "grad_norm": 0.1708984375,
      "learning_rate": 9.489584686600678e-05,
      "loss": 0.0017,
      "step": 11063
    },
    {
      "epoch": 5.164060676779464,
      "grad_norm": 0.494140625,
      "learning_rate": 9.488119256469546e-05,
      "loss": 0.0047,
      "step": 11064
    },
    {
      "epoch": 5.164527421236873,
      "grad_norm": 0.2734375,
      "learning_rate": 9.486653837359776e-05,
      "loss": 0.0045,
      "step": 11065
    },
    {
      "epoch": 5.164994165694282,
      "grad_norm": 0.171875,
      "learning_rate": 9.485188429302917e-05,
      "loss": 0.0015,
      "step": 11066
    },
    {
      "epoch": 5.165460910151692,
      "grad_norm": 0.37109375,
      "learning_rate": 9.48372303233052e-05,
      "loss": 0.0032,
      "step": 11067
    },
    {
      "epoch": 5.165927654609101,
      "grad_norm": 0.283203125,
      "learning_rate": 9.482257646474141e-05,
      "loss": 0.0019,
      "step": 11068
    },
    {
      "epoch": 5.166394399066511,
      "grad_norm": 0.2109375,
      "learning_rate": 9.48079227176533e-05,
      "loss": 0.0032,
      "step": 11069
    },
    {
      "epoch": 5.166861143523921,
      "grad_norm": 0.068359375,
      "learning_rate": 9.479326908235636e-05,
      "loss": 0.0009,
      "step": 11070
    },
    {
      "epoch": 5.1673278879813305,
      "grad_norm": 0.236328125,
      "learning_rate": 9.477861555916613e-05,
      "loss": 0.0022,
      "step": 11071
    },
    {
      "epoch": 5.167794632438739,
      "grad_norm": 0.263671875,
      "learning_rate": 9.47639621483981e-05,
      "loss": 0.0023,
      "step": 11072
    },
    {
      "epoch": 5.168261376896149,
      "grad_norm": 0.1884765625,
      "learning_rate": 9.474930885036775e-05,
      "loss": 0.0016,
      "step": 11073
    },
    {
      "epoch": 5.168728121353559,
      "grad_norm": 0.392578125,
      "learning_rate": 9.473465566539067e-05,
      "loss": 0.0035,
      "step": 11074
    },
    {
      "epoch": 5.169194865810969,
      "grad_norm": 0.37890625,
      "learning_rate": 9.472000259378223e-05,
      "loss": 0.0025,
      "step": 11075
    },
    {
      "epoch": 5.169661610268378,
      "grad_norm": 0.21875,
      "learning_rate": 9.470534963585802e-05,
      "loss": 0.0022,
      "step": 11076
    },
    {
      "epoch": 5.1701283547257875,
      "grad_norm": 0.2421875,
      "learning_rate": 9.46906967919335e-05,
      "loss": 0.0026,
      "step": 11077
    },
    {
      "epoch": 5.170595099183197,
      "grad_norm": 0.53515625,
      "learning_rate": 9.467604406232418e-05,
      "loss": 0.0025,
      "step": 11078
    },
    {
      "epoch": 5.171061843640607,
      "grad_norm": 0.1484375,
      "learning_rate": 9.46613914473455e-05,
      "loss": 0.0013,
      "step": 11079
    },
    {
      "epoch": 5.171528588098016,
      "grad_norm": 0.15625,
      "learning_rate": 9.464673894731302e-05,
      "loss": 0.0012,
      "step": 11080
    },
    {
      "epoch": 5.171995332555426,
      "grad_norm": 0.181640625,
      "learning_rate": 9.46320865625422e-05,
      "loss": 0.0013,
      "step": 11081
    },
    {
      "epoch": 5.172462077012836,
      "grad_norm": 0.259765625,
      "learning_rate": 9.461743429334847e-05,
      "loss": 0.0021,
      "step": 11082
    },
    {
      "epoch": 5.172928821470245,
      "grad_norm": 0.53125,
      "learning_rate": 9.460278214004738e-05,
      "loss": 0.0043,
      "step": 11083
    },
    {
      "epoch": 5.173395565927654,
      "grad_norm": 0.1318359375,
      "learning_rate": 9.458813010295438e-05,
      "loss": 0.001,
      "step": 11084
    },
    {
      "epoch": 5.173862310385064,
      "grad_norm": 0.2275390625,
      "learning_rate": 9.457347818238491e-05,
      "loss": 0.0017,
      "step": 11085
    },
    {
      "epoch": 5.174329054842474,
      "grad_norm": 0.35546875,
      "learning_rate": 9.455882637865451e-05,
      "loss": 0.0042,
      "step": 11086
    },
    {
      "epoch": 5.174795799299884,
      "grad_norm": 0.142578125,
      "learning_rate": 9.454417469207861e-05,
      "loss": 0.001,
      "step": 11087
    },
    {
      "epoch": 5.1752625437572926,
      "grad_norm": 0.2333984375,
      "learning_rate": 9.452952312297269e-05,
      "loss": 0.0058,
      "step": 11088
    },
    {
      "epoch": 5.175729288214702,
      "grad_norm": 0.1376953125,
      "learning_rate": 9.451487167165217e-05,
      "loss": 0.0012,
      "step": 11089
    },
    {
      "epoch": 5.176196032672112,
      "grad_norm": 0.28515625,
      "learning_rate": 9.450022033843259e-05,
      "loss": 0.0036,
      "step": 11090
    },
    {
      "epoch": 5.176662777129522,
      "grad_norm": 0.11474609375,
      "learning_rate": 9.448556912362936e-05,
      "loss": 0.0041,
      "step": 11091
    },
    {
      "epoch": 5.177129521586931,
      "grad_norm": 0.177734375,
      "learning_rate": 9.447091802755793e-05,
      "loss": 0.0013,
      "step": 11092
    },
    {
      "epoch": 5.177596266044341,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.445626705053378e-05,
      "loss": 0.001,
      "step": 11093
    },
    {
      "epoch": 5.17806301050175,
      "grad_norm": 0.052490234375,
      "learning_rate": 9.444161619287235e-05,
      "loss": 0.0007,
      "step": 11094
    },
    {
      "epoch": 5.17852975495916,
      "grad_norm": 0.3203125,
      "learning_rate": 9.442696545488907e-05,
      "loss": 0.0027,
      "step": 11095
    },
    {
      "epoch": 5.178996499416569,
      "grad_norm": 0.1669921875,
      "learning_rate": 9.441231483689943e-05,
      "loss": 0.002,
      "step": 11096
    },
    {
      "epoch": 5.179463243873979,
      "grad_norm": 0.3515625,
      "learning_rate": 9.439766433921885e-05,
      "loss": 0.0029,
      "step": 11097
    },
    {
      "epoch": 5.179929988331389,
      "grad_norm": 0.2470703125,
      "learning_rate": 9.438301396216279e-05,
      "loss": 0.0019,
      "step": 11098
    },
    {
      "epoch": 5.1803967327887985,
      "grad_norm": 0.2001953125,
      "learning_rate": 9.436836370604662e-05,
      "loss": 0.0017,
      "step": 11099
    },
    {
      "epoch": 5.180863477246207,
      "grad_norm": 0.10400390625,
      "learning_rate": 9.435371357118587e-05,
      "loss": 0.001,
      "step": 11100
    },
    {
      "epoch": 5.181330221703617,
      "grad_norm": 0.435546875,
      "learning_rate": 9.433906355789592e-05,
      "loss": 0.0034,
      "step": 11101
    },
    {
      "epoch": 5.181796966161027,
      "grad_norm": 0.10791015625,
      "learning_rate": 9.43244136664922e-05,
      "loss": 0.0011,
      "step": 11102
    },
    {
      "epoch": 5.182263710618437,
      "grad_norm": 0.181640625,
      "learning_rate": 9.430976389729019e-05,
      "loss": 0.0012,
      "step": 11103
    },
    {
      "epoch": 5.182730455075846,
      "grad_norm": 0.1201171875,
      "learning_rate": 9.429511425060524e-05,
      "loss": 0.0011,
      "step": 11104
    },
    {
      "epoch": 5.1831971995332555,
      "grad_norm": 0.515625,
      "learning_rate": 9.42804647267528e-05,
      "loss": 0.0167,
      "step": 11105
    },
    {
      "epoch": 5.183663943990665,
      "grad_norm": 0.71875,
      "learning_rate": 9.426581532604834e-05,
      "loss": 0.0071,
      "step": 11106
    },
    {
      "epoch": 5.184130688448075,
      "grad_norm": 0.466796875,
      "learning_rate": 9.425116604880722e-05,
      "loss": 0.0067,
      "step": 11107
    },
    {
      "epoch": 5.184597432905484,
      "grad_norm": 0.16796875,
      "learning_rate": 9.423651689534488e-05,
      "loss": 0.0016,
      "step": 11108
    },
    {
      "epoch": 5.185064177362894,
      "grad_norm": 0.18359375,
      "learning_rate": 9.42218678659767e-05,
      "loss": 0.0011,
      "step": 11109
    },
    {
      "epoch": 5.185530921820304,
      "grad_norm": 0.29296875,
      "learning_rate": 9.420721896101815e-05,
      "loss": 0.0047,
      "step": 11110
    },
    {
      "epoch": 5.1859976662777125,
      "grad_norm": 0.1298828125,
      "learning_rate": 9.41925701807846e-05,
      "loss": 0.001,
      "step": 11111
    },
    {
      "epoch": 5.186464410735122,
      "grad_norm": 0.171875,
      "learning_rate": 9.417792152559142e-05,
      "loss": 0.0013,
      "step": 11112
    },
    {
      "epoch": 5.186931155192532,
      "grad_norm": 0.259765625,
      "learning_rate": 9.416327299575408e-05,
      "loss": 0.0068,
      "step": 11113
    },
    {
      "epoch": 5.187397899649942,
      "grad_norm": 0.408203125,
      "learning_rate": 9.414862459158796e-05,
      "loss": 0.0053,
      "step": 11114
    },
    {
      "epoch": 5.187864644107351,
      "grad_norm": 0.2109375,
      "learning_rate": 9.41339763134084e-05,
      "loss": 0.0013,
      "step": 11115
    },
    {
      "epoch": 5.188331388564761,
      "grad_norm": 0.4140625,
      "learning_rate": 9.411932816153088e-05,
      "loss": 0.0061,
      "step": 11116
    },
    {
      "epoch": 5.18879813302217,
      "grad_norm": 0.48046875,
      "learning_rate": 9.410468013627073e-05,
      "loss": 0.0053,
      "step": 11117
    },
    {
      "epoch": 5.18926487747958,
      "grad_norm": 0.400390625,
      "learning_rate": 9.409003223794337e-05,
      "loss": 0.0038,
      "step": 11118
    },
    {
      "epoch": 5.189731621936989,
      "grad_norm": 0.1640625,
      "learning_rate": 9.407538446686416e-05,
      "loss": 0.0013,
      "step": 11119
    },
    {
      "epoch": 5.190198366394399,
      "grad_norm": 0.1162109375,
      "learning_rate": 9.40607368233485e-05,
      "loss": 0.0009,
      "step": 11120
    },
    {
      "epoch": 5.190665110851809,
      "grad_norm": 0.279296875,
      "learning_rate": 9.404608930771177e-05,
      "loss": 0.0027,
      "step": 11121
    },
    {
      "epoch": 5.1911318553092185,
      "grad_norm": 0.41796875,
      "learning_rate": 9.403144192026938e-05,
      "loss": 0.003,
      "step": 11122
    },
    {
      "epoch": 5.191598599766627,
      "grad_norm": 0.21875,
      "learning_rate": 9.401679466133663e-05,
      "loss": 0.0057,
      "step": 11123
    },
    {
      "epoch": 5.192065344224037,
      "grad_norm": 0.1865234375,
      "learning_rate": 9.400214753122892e-05,
      "loss": 0.0028,
      "step": 11124
    },
    {
      "epoch": 5.192532088681447,
      "grad_norm": 0.19921875,
      "learning_rate": 9.398750053026164e-05,
      "loss": 0.0018,
      "step": 11125
    },
    {
      "epoch": 5.192998833138857,
      "grad_norm": 0.0966796875,
      "learning_rate": 9.397285365875013e-05,
      "loss": 0.002,
      "step": 11126
    },
    {
      "epoch": 5.193465577596266,
      "grad_norm": 0.1220703125,
      "learning_rate": 9.395820691700977e-05,
      "loss": 0.0012,
      "step": 11127
    },
    {
      "epoch": 5.1939323220536755,
      "grad_norm": 0.07568359375,
      "learning_rate": 9.394356030535594e-05,
      "loss": 0.0009,
      "step": 11128
    },
    {
      "epoch": 5.194399066511085,
      "grad_norm": 0.07666015625,
      "learning_rate": 9.392891382410397e-05,
      "loss": 0.0008,
      "step": 11129
    },
    {
      "epoch": 5.194865810968495,
      "grad_norm": 0.51953125,
      "learning_rate": 9.391426747356922e-05,
      "loss": 0.0072,
      "step": 11130
    },
    {
      "epoch": 5.195332555425904,
      "grad_norm": 0.44921875,
      "learning_rate": 9.389962125406703e-05,
      "loss": 0.0057,
      "step": 11131
    },
    {
      "epoch": 5.195799299883314,
      "grad_norm": 0.0859375,
      "learning_rate": 9.388497516591277e-05,
      "loss": 0.0011,
      "step": 11132
    },
    {
      "epoch": 5.196266044340724,
      "grad_norm": 0.474609375,
      "learning_rate": 9.387032920942179e-05,
      "loss": 0.004,
      "step": 11133
    },
    {
      "epoch": 5.196732788798133,
      "grad_norm": 0.1376953125,
      "learning_rate": 9.38556833849094e-05,
      "loss": 0.001,
      "step": 11134
    },
    {
      "epoch": 5.197199533255542,
      "grad_norm": 0.1806640625,
      "learning_rate": 9.384103769269098e-05,
      "loss": 0.0035,
      "step": 11135
    },
    {
      "epoch": 5.197666277712952,
      "grad_norm": 0.484375,
      "learning_rate": 9.382639213308186e-05,
      "loss": 0.0043,
      "step": 11136
    },
    {
      "epoch": 5.198133022170362,
      "grad_norm": 0.19921875,
      "learning_rate": 9.381174670639736e-05,
      "loss": 0.0013,
      "step": 11137
    },
    {
      "epoch": 5.198599766627772,
      "grad_norm": 0.326171875,
      "learning_rate": 9.379710141295281e-05,
      "loss": 0.002,
      "step": 11138
    },
    {
      "epoch": 5.199066511085181,
      "grad_norm": 0.2275390625,
      "learning_rate": 9.378245625306356e-05,
      "loss": 0.0014,
      "step": 11139
    },
    {
      "epoch": 5.19953325554259,
      "grad_norm": 0.65625,
      "learning_rate": 9.376781122704492e-05,
      "loss": 0.0053,
      "step": 11140
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.248046875,
      "learning_rate": 9.37531663352122e-05,
      "loss": 0.0022,
      "step": 11141
    },
    {
      "epoch": 5.20046674445741,
      "grad_norm": 0.10986328125,
      "learning_rate": 9.373852157788077e-05,
      "loss": 0.0008,
      "step": 11142
    },
    {
      "epoch": 5.200933488914819,
      "grad_norm": 0.51953125,
      "learning_rate": 9.372387695536592e-05,
      "loss": 0.0062,
      "step": 11143
    },
    {
      "epoch": 5.201400233372229,
      "grad_norm": 0.390625,
      "learning_rate": 9.370923246798293e-05,
      "loss": 0.0069,
      "step": 11144
    },
    {
      "epoch": 5.2018669778296385,
      "grad_norm": 0.17578125,
      "learning_rate": 9.369458811604717e-05,
      "loss": 0.0046,
      "step": 11145
    },
    {
      "epoch": 5.202333722287048,
      "grad_norm": 0.1494140625,
      "learning_rate": 9.367994389987392e-05,
      "loss": 0.0011,
      "step": 11146
    },
    {
      "epoch": 5.202800466744457,
      "grad_norm": 0.26171875,
      "learning_rate": 9.366529981977849e-05,
      "loss": 0.0019,
      "step": 11147
    },
    {
      "epoch": 5.203267211201867,
      "grad_norm": 0.388671875,
      "learning_rate": 9.365065587607617e-05,
      "loss": 0.0076,
      "step": 11148
    },
    {
      "epoch": 5.203733955659277,
      "grad_norm": 0.5625,
      "learning_rate": 9.36360120690823e-05,
      "loss": 0.0041,
      "step": 11149
    },
    {
      "epoch": 5.204200700116687,
      "grad_norm": 0.443359375,
      "learning_rate": 9.362136839911214e-05,
      "loss": 0.0047,
      "step": 11150
    },
    {
      "epoch": 5.2046674445740955,
      "grad_norm": 0.1572265625,
      "learning_rate": 9.360672486648098e-05,
      "loss": 0.003,
      "step": 11151
    },
    {
      "epoch": 5.205134189031505,
      "grad_norm": 0.123046875,
      "learning_rate": 9.359208147150414e-05,
      "loss": 0.001,
      "step": 11152
    },
    {
      "epoch": 5.205600933488915,
      "grad_norm": 0.220703125,
      "learning_rate": 9.35774382144969e-05,
      "loss": 0.0026,
      "step": 11153
    },
    {
      "epoch": 5.206067677946324,
      "grad_norm": 0.09814453125,
      "learning_rate": 9.356279509577452e-05,
      "loss": 0.003,
      "step": 11154
    },
    {
      "epoch": 5.206534422403734,
      "grad_norm": 0.189453125,
      "learning_rate": 9.354815211565233e-05,
      "loss": 0.0015,
      "step": 11155
    },
    {
      "epoch": 5.207001166861144,
      "grad_norm": 0.1396484375,
      "learning_rate": 9.353350927444559e-05,
      "loss": 0.0011,
      "step": 11156
    },
    {
      "epoch": 5.207467911318553,
      "grad_norm": 0.1552734375,
      "learning_rate": 9.351886657246955e-05,
      "loss": 0.0013,
      "step": 11157
    },
    {
      "epoch": 5.207934655775962,
      "grad_norm": 0.31640625,
      "learning_rate": 9.350422401003949e-05,
      "loss": 0.0028,
      "step": 11158
    },
    {
      "epoch": 5.208401400233372,
      "grad_norm": 0.341796875,
      "learning_rate": 9.348958158747073e-05,
      "loss": 0.0031,
      "step": 11159
    },
    {
      "epoch": 5.208868144690782,
      "grad_norm": 0.37109375,
      "learning_rate": 9.347493930507848e-05,
      "loss": 0.0026,
      "step": 11160
    },
    {
      "epoch": 5.209334889148192,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.3460297163178e-05,
      "loss": 0.0019,
      "step": 11161
    },
    {
      "epoch": 5.209801633605601,
      "grad_norm": 0.12060546875,
      "learning_rate": 9.344565516208463e-05,
      "loss": 0.0012,
      "step": 11162
    },
    {
      "epoch": 5.21026837806301,
      "grad_norm": 0.443359375,
      "learning_rate": 9.343101330211355e-05,
      "loss": 0.003,
      "step": 11163
    },
    {
      "epoch": 5.21073512252042,
      "grad_norm": 0.134765625,
      "learning_rate": 9.341637158358004e-05,
      "loss": 0.0016,
      "step": 11164
    },
    {
      "epoch": 5.21120186697783,
      "grad_norm": 0.3515625,
      "learning_rate": 9.340173000679938e-05,
      "loss": 0.0029,
      "step": 11165
    },
    {
      "epoch": 5.211668611435239,
      "grad_norm": 0.107421875,
      "learning_rate": 9.338708857208676e-05,
      "loss": 0.0009,
      "step": 11166
    },
    {
      "epoch": 5.212135355892649,
      "grad_norm": 0.337890625,
      "learning_rate": 9.337244727975749e-05,
      "loss": 0.0044,
      "step": 11167
    },
    {
      "epoch": 5.212602100350058,
      "grad_norm": 0.1123046875,
      "learning_rate": 9.335780613012674e-05,
      "loss": 0.003,
      "step": 11168
    },
    {
      "epoch": 5.213068844807468,
      "grad_norm": 0.37109375,
      "learning_rate": 9.334316512350984e-05,
      "loss": 0.0025,
      "step": 11169
    },
    {
      "epoch": 5.213535589264877,
      "grad_norm": 0.388671875,
      "learning_rate": 9.332852426022198e-05,
      "loss": 0.0021,
      "step": 11170
    },
    {
      "epoch": 5.214002333722287,
      "grad_norm": 0.154296875,
      "learning_rate": 9.33138835405784e-05,
      "loss": 0.0013,
      "step": 11171
    },
    {
      "epoch": 5.214469078179697,
      "grad_norm": 0.271484375,
      "learning_rate": 9.32992429648943e-05,
      "loss": 0.0015,
      "step": 11172
    },
    {
      "epoch": 5.2149358226371065,
      "grad_norm": 0.16796875,
      "learning_rate": 9.328460253348492e-05,
      "loss": 0.0013,
      "step": 11173
    },
    {
      "epoch": 5.215402567094515,
      "grad_norm": 0.11669921875,
      "learning_rate": 9.326996224666552e-05,
      "loss": 0.0009,
      "step": 11174
    },
    {
      "epoch": 5.215869311551925,
      "grad_norm": 0.14453125,
      "learning_rate": 9.32553221047513e-05,
      "loss": 0.0032,
      "step": 11175
    },
    {
      "epoch": 5.216336056009335,
      "grad_norm": 0.224609375,
      "learning_rate": 9.324068210805746e-05,
      "loss": 0.0022,
      "step": 11176
    },
    {
      "epoch": 5.216802800466745,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.322604225689927e-05,
      "loss": 0.0013,
      "step": 11177
    },
    {
      "epoch": 5.217269544924154,
      "grad_norm": 0.431640625,
      "learning_rate": 9.321140255159187e-05,
      "loss": 0.0056,
      "step": 11178
    },
    {
      "epoch": 5.2177362893815635,
      "grad_norm": 0.20703125,
      "learning_rate": 9.319676299245053e-05,
      "loss": 0.0019,
      "step": 11179
    },
    {
      "epoch": 5.218203033838973,
      "grad_norm": 0.11865234375,
      "learning_rate": 9.31821235797904e-05,
      "loss": 0.0008,
      "step": 11180
    },
    {
      "epoch": 5.218669778296383,
      "grad_norm": 0.234375,
      "learning_rate": 9.316748431392674e-05,
      "loss": 0.0018,
      "step": 11181
    },
    {
      "epoch": 5.219136522753792,
      "grad_norm": 0.1298828125,
      "learning_rate": 9.315284519517473e-05,
      "loss": 0.0014,
      "step": 11182
    },
    {
      "epoch": 5.219603267211202,
      "grad_norm": 0.32421875,
      "learning_rate": 9.313820622384952e-05,
      "loss": 0.0018,
      "step": 11183
    },
    {
      "epoch": 5.220070011668612,
      "grad_norm": 0.1982421875,
      "learning_rate": 9.312356740026639e-05,
      "loss": 0.0043,
      "step": 11184
    },
    {
      "epoch": 5.220536756126021,
      "grad_norm": 0.361328125,
      "learning_rate": 9.310892872474047e-05,
      "loss": 0.0029,
      "step": 11185
    },
    {
      "epoch": 5.22100350058343,
      "grad_norm": 0.412109375,
      "learning_rate": 9.309429019758694e-05,
      "loss": 0.0034,
      "step": 11186
    },
    {
      "epoch": 5.22147024504084,
      "grad_norm": 0.71875,
      "learning_rate": 9.307965181912101e-05,
      "loss": 0.0286,
      "step": 11187
    },
    {
      "epoch": 5.22193698949825,
      "grad_norm": 0.267578125,
      "learning_rate": 9.306501358965787e-05,
      "loss": 0.0023,
      "step": 11188
    },
    {
      "epoch": 5.222403733955659,
      "grad_norm": 0.2392578125,
      "learning_rate": 9.305037550951268e-05,
      "loss": 0.0039,
      "step": 11189
    },
    {
      "epoch": 5.222870478413069,
      "grad_norm": 0.2890625,
      "learning_rate": 9.303573757900061e-05,
      "loss": 0.0059,
      "step": 11190
    },
    {
      "epoch": 5.223337222870478,
      "grad_norm": 0.224609375,
      "learning_rate": 9.302109979843684e-05,
      "loss": 0.0017,
      "step": 11191
    },
    {
      "epoch": 5.223803967327888,
      "grad_norm": 0.2392578125,
      "learning_rate": 9.300646216813654e-05,
      "loss": 0.0046,
      "step": 11192
    },
    {
      "epoch": 5.224270711785298,
      "grad_norm": 0.09814453125,
      "learning_rate": 9.299182468841484e-05,
      "loss": 0.0008,
      "step": 11193
    },
    {
      "epoch": 5.224737456242707,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.297718735958695e-05,
      "loss": 0.0014,
      "step": 11194
    },
    {
      "epoch": 5.225204200700117,
      "grad_norm": 0.640625,
      "learning_rate": 9.296255018196803e-05,
      "loss": 0.0054,
      "step": 11195
    },
    {
      "epoch": 5.2256709451575265,
      "grad_norm": 0.146484375,
      "learning_rate": 9.294791315587319e-05,
      "loss": 0.0011,
      "step": 11196
    },
    {
      "epoch": 5.226137689614935,
      "grad_norm": 0.255859375,
      "learning_rate": 9.293327628161759e-05,
      "loss": 0.0039,
      "step": 11197
    },
    {
      "epoch": 5.226604434072345,
      "grad_norm": 0.046630859375,
      "learning_rate": 9.291863955951642e-05,
      "loss": 0.0028,
      "step": 11198
    },
    {
      "epoch": 5.227071178529755,
      "grad_norm": 0.050537109375,
      "learning_rate": 9.290400298988479e-05,
      "loss": 0.0006,
      "step": 11199
    },
    {
      "epoch": 5.227537922987165,
      "grad_norm": 0.3984375,
      "learning_rate": 9.288936657303783e-05,
      "loss": 0.0051,
      "step": 11200
    },
    {
      "epoch": 5.228004667444574,
      "grad_norm": 0.1318359375,
      "learning_rate": 9.287473030929071e-05,
      "loss": 0.0029,
      "step": 11201
    },
    {
      "epoch": 5.2284714119019835,
      "grad_norm": 0.34765625,
      "learning_rate": 9.286009419895856e-05,
      "loss": 0.0031,
      "step": 11202
    },
    {
      "epoch": 5.228938156359393,
      "grad_norm": 0.345703125,
      "learning_rate": 9.284545824235649e-05,
      "loss": 0.0027,
      "step": 11203
    },
    {
      "epoch": 5.229404900816803,
      "grad_norm": 0.34375,
      "learning_rate": 9.283082243979964e-05,
      "loss": 0.0046,
      "step": 11204
    },
    {
      "epoch": 5.229871645274212,
      "grad_norm": 0.2392578125,
      "learning_rate": 9.281618679160317e-05,
      "loss": 0.003,
      "step": 11205
    },
    {
      "epoch": 5.230338389731622,
      "grad_norm": 0.380859375,
      "learning_rate": 9.280155129808214e-05,
      "loss": 0.0043,
      "step": 11206
    },
    {
      "epoch": 5.230805134189032,
      "grad_norm": 0.1337890625,
      "learning_rate": 9.278691595955168e-05,
      "loss": 0.0011,
      "step": 11207
    },
    {
      "epoch": 5.231271878646441,
      "grad_norm": 0.34765625,
      "learning_rate": 9.277228077632696e-05,
      "loss": 0.0028,
      "step": 11208
    },
    {
      "epoch": 5.23173862310385,
      "grad_norm": 0.21875,
      "learning_rate": 9.275764574872306e-05,
      "loss": 0.004,
      "step": 11209
    },
    {
      "epoch": 5.23220536756126,
      "grad_norm": 0.287109375,
      "learning_rate": 9.274301087705504e-05,
      "loss": 0.0025,
      "step": 11210
    },
    {
      "epoch": 5.23267211201867,
      "grad_norm": 0.06201171875,
      "learning_rate": 9.272837616163808e-05,
      "loss": 0.0008,
      "step": 11211
    },
    {
      "epoch": 5.23313885647608,
      "grad_norm": 0.0625,
      "learning_rate": 9.271374160278724e-05,
      "loss": 0.0007,
      "step": 11212
    },
    {
      "epoch": 5.233605600933489,
      "grad_norm": 0.185546875,
      "learning_rate": 9.26991072008176e-05,
      "loss": 0.0015,
      "step": 11213
    },
    {
      "epoch": 5.234072345390898,
      "grad_norm": 0.1416015625,
      "learning_rate": 9.268447295604432e-05,
      "loss": 0.0039,
      "step": 11214
    },
    {
      "epoch": 5.234539089848308,
      "grad_norm": 0.314453125,
      "learning_rate": 9.266983886878245e-05,
      "loss": 0.0025,
      "step": 11215
    },
    {
      "epoch": 5.235005834305718,
      "grad_norm": 0.35546875,
      "learning_rate": 9.265520493934708e-05,
      "loss": 0.0017,
      "step": 11216
    },
    {
      "epoch": 5.235472578763127,
      "grad_norm": 0.35546875,
      "learning_rate": 9.264057116805333e-05,
      "loss": 0.0028,
      "step": 11217
    },
    {
      "epoch": 5.235939323220537,
      "grad_norm": 0.224609375,
      "learning_rate": 9.262593755521622e-05,
      "loss": 0.0069,
      "step": 11218
    },
    {
      "epoch": 5.2364060676779465,
      "grad_norm": 0.099609375,
      "learning_rate": 9.261130410115082e-05,
      "loss": 0.001,
      "step": 11219
    },
    {
      "epoch": 5.236872812135356,
      "grad_norm": 0.2119140625,
      "learning_rate": 9.259667080617227e-05,
      "loss": 0.0093,
      "step": 11220
    },
    {
      "epoch": 5.237339556592765,
      "grad_norm": 0.2255859375,
      "learning_rate": 9.258203767059562e-05,
      "loss": 0.0015,
      "step": 11221
    },
    {
      "epoch": 5.237806301050175,
      "grad_norm": 0.49609375,
      "learning_rate": 9.25674046947359e-05,
      "loss": 0.0052,
      "step": 11222
    },
    {
      "epoch": 5.238273045507585,
      "grad_norm": 0.1416015625,
      "learning_rate": 9.255277187890823e-05,
      "loss": 0.0012,
      "step": 11223
    },
    {
      "epoch": 5.238739789964995,
      "grad_norm": 0.298828125,
      "learning_rate": 9.253813922342764e-05,
      "loss": 0.0039,
      "step": 11224
    },
    {
      "epoch": 5.2392065344224035,
      "grad_norm": 0.341796875,
      "learning_rate": 9.252350672860918e-05,
      "loss": 0.0036,
      "step": 11225
    },
    {
      "epoch": 5.239673278879813,
      "grad_norm": 0.054443359375,
      "learning_rate": 9.250887439476792e-05,
      "loss": 0.0007,
      "step": 11226
    },
    {
      "epoch": 5.240140023337223,
      "grad_norm": 0.4140625,
      "learning_rate": 9.249424222221893e-05,
      "loss": 0.0023,
      "step": 11227
    },
    {
      "epoch": 5.240606767794633,
      "grad_norm": 0.248046875,
      "learning_rate": 9.24796102112772e-05,
      "loss": 0.0021,
      "step": 11228
    },
    {
      "epoch": 5.241073512252042,
      "grad_norm": 0.27734375,
      "learning_rate": 9.24649783622578e-05,
      "loss": 0.0046,
      "step": 11229
    },
    {
      "epoch": 5.241540256709452,
      "grad_norm": 0.1142578125,
      "learning_rate": 9.245034667547581e-05,
      "loss": 0.0046,
      "step": 11230
    },
    {
      "epoch": 5.242007001166861,
      "grad_norm": 0.07421875,
      "learning_rate": 9.243571515124621e-05,
      "loss": 0.0008,
      "step": 11231
    },
    {
      "epoch": 5.24247374562427,
      "grad_norm": 0.357421875,
      "learning_rate": 9.242108378988405e-05,
      "loss": 0.0041,
      "step": 11232
    },
    {
      "epoch": 5.24294049008168,
      "grad_norm": 0.263671875,
      "learning_rate": 9.240645259170438e-05,
      "loss": 0.0034,
      "step": 11233
    },
    {
      "epoch": 5.24340723453909,
      "grad_norm": 0.5625,
      "learning_rate": 9.239182155702222e-05,
      "loss": 0.0066,
      "step": 11234
    },
    {
      "epoch": 5.2438739789965,
      "grad_norm": 0.0849609375,
      "learning_rate": 9.237719068615256e-05,
      "loss": 0.0008,
      "step": 11235
    },
    {
      "epoch": 5.244340723453909,
      "grad_norm": 0.06689453125,
      "learning_rate": 9.236255997941045e-05,
      "loss": 0.0007,
      "step": 11236
    },
    {
      "epoch": 5.244807467911318,
      "grad_norm": 0.038818359375,
      "learning_rate": 9.234792943711091e-05,
      "loss": 0.0006,
      "step": 11237
    },
    {
      "epoch": 5.245274212368728,
      "grad_norm": 0.19140625,
      "learning_rate": 9.233329905956894e-05,
      "loss": 0.0036,
      "step": 11238
    },
    {
      "epoch": 5.245740956826138,
      "grad_norm": 0.0625,
      "learning_rate": 9.231866884709952e-05,
      "loss": 0.0008,
      "step": 11239
    },
    {
      "epoch": 5.246207701283547,
      "grad_norm": 0.322265625,
      "learning_rate": 9.23040388000177e-05,
      "loss": 0.0021,
      "step": 11240
    },
    {
      "epoch": 5.246674445740957,
      "grad_norm": 0.349609375,
      "learning_rate": 9.228940891863848e-05,
      "loss": 0.0026,
      "step": 11241
    },
    {
      "epoch": 5.2471411901983664,
      "grad_norm": 0.4375,
      "learning_rate": 9.22747792032768e-05,
      "loss": 0.0065,
      "step": 11242
    },
    {
      "epoch": 5.247607934655776,
      "grad_norm": 0.046875,
      "learning_rate": 9.226014965424774e-05,
      "loss": 0.0006,
      "step": 11243
    },
    {
      "epoch": 5.248074679113185,
      "grad_norm": 0.1875,
      "learning_rate": 9.224552027186622e-05,
      "loss": 0.0012,
      "step": 11244
    },
    {
      "epoch": 5.248541423570595,
      "grad_norm": 0.169921875,
      "learning_rate": 9.223089105644724e-05,
      "loss": 0.0013,
      "step": 11245
    },
    {
      "epoch": 5.249008168028005,
      "grad_norm": 0.259765625,
      "learning_rate": 9.221626200830582e-05,
      "loss": 0.0022,
      "step": 11246
    },
    {
      "epoch": 5.2494749124854145,
      "grad_norm": 0.2421875,
      "learning_rate": 9.220163312775692e-05,
      "loss": 0.0015,
      "step": 11247
    },
    {
      "epoch": 5.249941656942823,
      "grad_norm": 0.37890625,
      "learning_rate": 9.21870044151155e-05,
      "loss": 0.0027,
      "step": 11248
    },
    {
      "epoch": 5.250408401400233,
      "grad_norm": 0.203125,
      "learning_rate": 9.217237587069653e-05,
      "loss": 0.0018,
      "step": 11249
    },
    {
      "epoch": 5.250875145857643,
      "grad_norm": 0.109375,
      "learning_rate": 9.2157747494815e-05,
      "loss": 0.0011,
      "step": 11250
    },
    {
      "epoch": 5.251341890315053,
      "grad_norm": 0.080078125,
      "learning_rate": 9.214311928778588e-05,
      "loss": 0.0008,
      "step": 11251
    },
    {
      "epoch": 5.251808634772462,
      "grad_norm": 0.34375,
      "learning_rate": 9.21284912499241e-05,
      "loss": 0.0045,
      "step": 11252
    },
    {
      "epoch": 5.2522753792298715,
      "grad_norm": 0.06005859375,
      "learning_rate": 9.211386338154465e-05,
      "loss": 0.0007,
      "step": 11253
    },
    {
      "epoch": 5.252742123687281,
      "grad_norm": 0.1591796875,
      "learning_rate": 9.209923568296249e-05,
      "loss": 0.0013,
      "step": 11254
    },
    {
      "epoch": 5.253208868144691,
      "grad_norm": 0.3984375,
      "learning_rate": 9.208460815449254e-05,
      "loss": 0.0044,
      "step": 11255
    },
    {
      "epoch": 5.2536756126021,
      "grad_norm": 0.083984375,
      "learning_rate": 9.206998079644973e-05,
      "loss": 0.0067,
      "step": 11256
    },
    {
      "epoch": 5.2536756126021,
      "eval_loss": 2.1444499492645264,
      "eval_runtime": 55.9501,
      "eval_samples_per_second": 32.243,
      "eval_steps_per_second": 4.039,
      "step": 11256
    },
    {
      "epoch": 5.25414235705951,
      "grad_norm": 0.06884765625,
      "learning_rate": 9.205535360914905e-05,
      "loss": 0.0007,
      "step": 11257
    },
    {
      "epoch": 5.25460910151692,
      "grad_norm": 0.05419921875,
      "learning_rate": 9.204072659290544e-05,
      "loss": 0.0006,
      "step": 11258
    },
    {
      "epoch": 5.255075845974329,
      "grad_norm": 0.2158203125,
      "learning_rate": 9.202609974803378e-05,
      "loss": 0.0009,
      "step": 11259
    },
    {
      "epoch": 5.255542590431738,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.201147307484907e-05,
      "loss": 0.0011,
      "step": 11260
    },
    {
      "epoch": 5.256009334889148,
      "grad_norm": 0.234375,
      "learning_rate": 9.199684657366621e-05,
      "loss": 0.0011,
      "step": 11261
    },
    {
      "epoch": 5.256476079346558,
      "grad_norm": 0.42578125,
      "learning_rate": 9.198222024480009e-05,
      "loss": 0.0044,
      "step": 11262
    },
    {
      "epoch": 5.256942823803968,
      "grad_norm": 0.30859375,
      "learning_rate": 9.196759408856569e-05,
      "loss": 0.0028,
      "step": 11263
    },
    {
      "epoch": 5.257409568261377,
      "grad_norm": 0.197265625,
      "learning_rate": 9.19529681052779e-05,
      "loss": 0.0011,
      "step": 11264
    },
    {
      "epoch": 5.257876312718786,
      "grad_norm": 0.0732421875,
      "learning_rate": 9.193834229525167e-05,
      "loss": 0.0008,
      "step": 11265
    },
    {
      "epoch": 5.258343057176196,
      "grad_norm": 0.208984375,
      "learning_rate": 9.192371665880184e-05,
      "loss": 0.0042,
      "step": 11266
    },
    {
      "epoch": 5.258809801633606,
      "grad_norm": 0.416015625,
      "learning_rate": 9.190909119624335e-05,
      "loss": 0.0031,
      "step": 11267
    },
    {
      "epoch": 5.259276546091015,
      "grad_norm": 0.1650390625,
      "learning_rate": 9.189446590789108e-05,
      "loss": 0.0013,
      "step": 11268
    },
    {
      "epoch": 5.259743290548425,
      "grad_norm": 0.287109375,
      "learning_rate": 9.187984079405999e-05,
      "loss": 0.0017,
      "step": 11269
    },
    {
      "epoch": 5.2602100350058345,
      "grad_norm": 0.181640625,
      "learning_rate": 9.186521585506492e-05,
      "loss": 0.0015,
      "step": 11270
    },
    {
      "epoch": 5.260676779463244,
      "grad_norm": 0.052490234375,
      "learning_rate": 9.185059109122077e-05,
      "loss": 0.0007,
      "step": 11271
    },
    {
      "epoch": 5.261143523920653,
      "grad_norm": 0.408203125,
      "learning_rate": 9.183596650284246e-05,
      "loss": 0.0027,
      "step": 11272
    },
    {
      "epoch": 5.261610268378063,
      "grad_norm": 0.2578125,
      "learning_rate": 9.182134209024483e-05,
      "loss": 0.0018,
      "step": 11273
    },
    {
      "epoch": 5.262077012835473,
      "grad_norm": 0.2216796875,
      "learning_rate": 9.180671785374278e-05,
      "loss": 0.0013,
      "step": 11274
    },
    {
      "epoch": 5.262543757292882,
      "grad_norm": 0.0810546875,
      "learning_rate": 9.17920937936512e-05,
      "loss": 0.0041,
      "step": 11275
    },
    {
      "epoch": 5.2630105017502915,
      "grad_norm": 0.11181640625,
      "learning_rate": 9.177746991028493e-05,
      "loss": 0.001,
      "step": 11276
    },
    {
      "epoch": 5.263477246207701,
      "grad_norm": 0.049560546875,
      "learning_rate": 9.176284620395887e-05,
      "loss": 0.0007,
      "step": 11277
    },
    {
      "epoch": 5.263943990665111,
      "grad_norm": 0.54296875,
      "learning_rate": 9.174822267498784e-05,
      "loss": 0.0023,
      "step": 11278
    },
    {
      "epoch": 5.264410735122521,
      "grad_norm": 0.2421875,
      "learning_rate": 9.173359932368677e-05,
      "loss": 0.0038,
      "step": 11279
    },
    {
      "epoch": 5.26487747957993,
      "grad_norm": 0.38671875,
      "learning_rate": 9.171897615037046e-05,
      "loss": 0.0026,
      "step": 11280
    },
    {
      "epoch": 5.26534422403734,
      "grad_norm": 0.275390625,
      "learning_rate": 9.170435315535377e-05,
      "loss": 0.0016,
      "step": 11281
    },
    {
      "epoch": 5.265810968494749,
      "grad_norm": 0.25,
      "learning_rate": 9.168973033895158e-05,
      "loss": 0.0018,
      "step": 11282
    },
    {
      "epoch": 5.266277712952158,
      "grad_norm": 0.423828125,
      "learning_rate": 9.167510770147872e-05,
      "loss": 0.0042,
      "step": 11283
    },
    {
      "epoch": 5.266744457409568,
      "grad_norm": 0.39453125,
      "learning_rate": 9.166048524325e-05,
      "loss": 0.0017,
      "step": 11284
    },
    {
      "epoch": 5.267211201866978,
      "grad_norm": 0.1337890625,
      "learning_rate": 9.164586296458031e-05,
      "loss": 0.0008,
      "step": 11285
    },
    {
      "epoch": 5.267677946324388,
      "grad_norm": 0.28515625,
      "learning_rate": 9.163124086578446e-05,
      "loss": 0.0014,
      "step": 11286
    },
    {
      "epoch": 5.268144690781797,
      "grad_norm": 0.30859375,
      "learning_rate": 9.161661894717728e-05,
      "loss": 0.0048,
      "step": 11287
    },
    {
      "epoch": 5.268611435239206,
      "grad_norm": 0.146484375,
      "learning_rate": 9.160199720907357e-05,
      "loss": 0.0012,
      "step": 11288
    },
    {
      "epoch": 5.269078179696616,
      "grad_norm": 0.1787109375,
      "learning_rate": 9.158737565178822e-05,
      "loss": 0.0018,
      "step": 11289
    },
    {
      "epoch": 5.269544924154026,
      "grad_norm": 0.08984375,
      "learning_rate": 9.157275427563598e-05,
      "loss": 0.0011,
      "step": 11290
    },
    {
      "epoch": 5.270011668611435,
      "grad_norm": 0.09716796875,
      "learning_rate": 9.155813308093167e-05,
      "loss": 0.0031,
      "step": 11291
    },
    {
      "epoch": 5.270478413068845,
      "grad_norm": 0.08349609375,
      "learning_rate": 9.154351206799016e-05,
      "loss": 0.0009,
      "step": 11292
    },
    {
      "epoch": 5.2709451575262545,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.152889123712621e-05,
      "loss": 0.0011,
      "step": 11293
    },
    {
      "epoch": 5.271411901983664,
      "grad_norm": 0.361328125,
      "learning_rate": 9.151427058865461e-05,
      "loss": 0.0022,
      "step": 11294
    },
    {
      "epoch": 5.271878646441073,
      "grad_norm": 0.2060546875,
      "learning_rate": 9.149965012289019e-05,
      "loss": 0.0017,
      "step": 11295
    },
    {
      "epoch": 5.272345390898483,
      "grad_norm": 0.2333984375,
      "learning_rate": 9.148502984014774e-05,
      "loss": 0.0023,
      "step": 11296
    },
    {
      "epoch": 5.272812135355893,
      "grad_norm": 0.10546875,
      "learning_rate": 9.147040974074204e-05,
      "loss": 0.0014,
      "step": 11297
    },
    {
      "epoch": 5.273278879813303,
      "grad_norm": 0.275390625,
      "learning_rate": 9.145578982498788e-05,
      "loss": 0.0023,
      "step": 11298
    },
    {
      "epoch": 5.2737456242707115,
      "grad_norm": 0.373046875,
      "learning_rate": 9.144117009320006e-05,
      "loss": 0.0028,
      "step": 11299
    },
    {
      "epoch": 5.274212368728121,
      "grad_norm": 0.080078125,
      "learning_rate": 9.142655054569333e-05,
      "loss": 0.0007,
      "step": 11300
    },
    {
      "epoch": 5.274679113185531,
      "grad_norm": 0.1220703125,
      "learning_rate": 9.141193118278248e-05,
      "loss": 0.0036,
      "step": 11301
    },
    {
      "epoch": 5.275145857642941,
      "grad_norm": 0.2177734375,
      "learning_rate": 9.139731200478227e-05,
      "loss": 0.002,
      "step": 11302
    },
    {
      "epoch": 5.27561260210035,
      "grad_norm": 0.2490234375,
      "learning_rate": 9.138269301200751e-05,
      "loss": 0.0018,
      "step": 11303
    },
    {
      "epoch": 5.27607934655776,
      "grad_norm": 0.1689453125,
      "learning_rate": 9.136807420477291e-05,
      "loss": 0.0015,
      "step": 11304
    },
    {
      "epoch": 5.276546091015169,
      "grad_norm": 0.375,
      "learning_rate": 9.135345558339323e-05,
      "loss": 0.0045,
      "step": 11305
    },
    {
      "epoch": 5.277012835472579,
      "grad_norm": 0.1259765625,
      "learning_rate": 9.133883714818328e-05,
      "loss": 0.0008,
      "step": 11306
    },
    {
      "epoch": 5.277479579929988,
      "grad_norm": 0.0703125,
      "learning_rate": 9.132421889945776e-05,
      "loss": 0.0029,
      "step": 11307
    },
    {
      "epoch": 5.277946324387398,
      "grad_norm": 0.1630859375,
      "learning_rate": 9.130960083753142e-05,
      "loss": 0.0008,
      "step": 11308
    },
    {
      "epoch": 5.278413068844808,
      "grad_norm": 0.26953125,
      "learning_rate": 9.129498296271903e-05,
      "loss": 0.0016,
      "step": 11309
    },
    {
      "epoch": 5.2788798133022174,
      "grad_norm": 0.2060546875,
      "learning_rate": 9.128036527533532e-05,
      "loss": 0.0012,
      "step": 11310
    },
    {
      "epoch": 5.279346557759626,
      "grad_norm": 0.16796875,
      "learning_rate": 9.1265747775695e-05,
      "loss": 0.0013,
      "step": 11311
    },
    {
      "epoch": 5.279813302217036,
      "grad_norm": 0.29296875,
      "learning_rate": 9.125113046411285e-05,
      "loss": 0.0019,
      "step": 11312
    },
    {
      "epoch": 5.280280046674446,
      "grad_norm": 0.146484375,
      "learning_rate": 9.123651334090359e-05,
      "loss": 0.0026,
      "step": 11313
    },
    {
      "epoch": 5.280746791131856,
      "grad_norm": 0.20703125,
      "learning_rate": 9.122189640638188e-05,
      "loss": 0.0017,
      "step": 11314
    },
    {
      "epoch": 5.281213535589265,
      "grad_norm": 0.10595703125,
      "learning_rate": 9.120727966086248e-05,
      "loss": 0.001,
      "step": 11315
    },
    {
      "epoch": 5.2816802800466744,
      "grad_norm": 0.1328125,
      "learning_rate": 9.119266310466012e-05,
      "loss": 0.0027,
      "step": 11316
    },
    {
      "epoch": 5.282147024504084,
      "grad_norm": 0.318359375,
      "learning_rate": 9.117804673808947e-05,
      "loss": 0.0013,
      "step": 11317
    },
    {
      "epoch": 5.282613768961493,
      "grad_norm": 0.341796875,
      "learning_rate": 9.116343056146529e-05,
      "loss": 0.0037,
      "step": 11318
    },
    {
      "epoch": 5.283080513418903,
      "grad_norm": 0.1484375,
      "learning_rate": 9.114881457510225e-05,
      "loss": 0.0011,
      "step": 11319
    },
    {
      "epoch": 5.283547257876313,
      "grad_norm": 0.373046875,
      "learning_rate": 9.113419877931503e-05,
      "loss": 0.0015,
      "step": 11320
    },
    {
      "epoch": 5.2840140023337225,
      "grad_norm": 0.0556640625,
      "learning_rate": 9.111958317441836e-05,
      "loss": 0.0006,
      "step": 11321
    },
    {
      "epoch": 5.2844807467911314,
      "grad_norm": 0.193359375,
      "learning_rate": 9.110496776072693e-05,
      "loss": 0.0023,
      "step": 11322
    },
    {
      "epoch": 5.284947491248541,
      "grad_norm": 0.052490234375,
      "learning_rate": 9.109035253855538e-05,
      "loss": 0.0006,
      "step": 11323
    },
    {
      "epoch": 5.285414235705951,
      "grad_norm": 0.435546875,
      "learning_rate": 9.107573750821847e-05,
      "loss": 0.005,
      "step": 11324
    },
    {
      "epoch": 5.285880980163361,
      "grad_norm": 0.40234375,
      "learning_rate": 9.10611226700308e-05,
      "loss": 0.0063,
      "step": 11325
    },
    {
      "epoch": 5.28634772462077,
      "grad_norm": 0.158203125,
      "learning_rate": 9.104650802430709e-05,
      "loss": 0.0025,
      "step": 11326
    },
    {
      "epoch": 5.2868144690781795,
      "grad_norm": 0.07177734375,
      "learning_rate": 9.103189357136198e-05,
      "loss": 0.0008,
      "step": 11327
    },
    {
      "epoch": 5.287281213535589,
      "grad_norm": 0.33984375,
      "learning_rate": 9.101727931151017e-05,
      "loss": 0.0015,
      "step": 11328
    },
    {
      "epoch": 5.287747957992999,
      "grad_norm": 0.2451171875,
      "learning_rate": 9.100266524506632e-05,
      "loss": 0.0015,
      "step": 11329
    },
    {
      "epoch": 5.288214702450408,
      "grad_norm": 0.224609375,
      "learning_rate": 9.098805137234502e-05,
      "loss": 0.0017,
      "step": 11330
    },
    {
      "epoch": 5.288681446907818,
      "grad_norm": 0.1376953125,
      "learning_rate": 9.097343769366102e-05,
      "loss": 0.001,
      "step": 11331
    },
    {
      "epoch": 5.289148191365228,
      "grad_norm": 0.08154296875,
      "learning_rate": 9.09588242093289e-05,
      "loss": 0.0032,
      "step": 11332
    },
    {
      "epoch": 5.289614935822637,
      "grad_norm": 0.1552734375,
      "learning_rate": 9.094421091966332e-05,
      "loss": 0.001,
      "step": 11333
    },
    {
      "epoch": 5.290081680280046,
      "grad_norm": 0.1494140625,
      "learning_rate": 9.092959782497893e-05,
      "loss": 0.0009,
      "step": 11334
    },
    {
      "epoch": 5.290548424737456,
      "grad_norm": 0.126953125,
      "learning_rate": 9.091498492559038e-05,
      "loss": 0.0009,
      "step": 11335
    },
    {
      "epoch": 5.291015169194866,
      "grad_norm": 0.1708984375,
      "learning_rate": 9.090037222181227e-05,
      "loss": 0.0012,
      "step": 11336
    },
    {
      "epoch": 5.291481913652276,
      "grad_norm": 0.431640625,
      "learning_rate": 9.088575971395923e-05,
      "loss": 0.0018,
      "step": 11337
    },
    {
      "epoch": 5.291948658109685,
      "grad_norm": 0.1923828125,
      "learning_rate": 9.087114740234591e-05,
      "loss": 0.0015,
      "step": 11338
    },
    {
      "epoch": 5.292415402567094,
      "grad_norm": 0.2080078125,
      "learning_rate": 9.085653528728692e-05,
      "loss": 0.0017,
      "step": 11339
    },
    {
      "epoch": 5.292882147024504,
      "grad_norm": 0.244140625,
      "learning_rate": 9.084192336909686e-05,
      "loss": 0.0018,
      "step": 11340
    },
    {
      "epoch": 5.293348891481914,
      "grad_norm": 0.57421875,
      "learning_rate": 9.082731164809037e-05,
      "loss": 0.0082,
      "step": 11341
    },
    {
      "epoch": 5.293815635939323,
      "grad_norm": 0.1416015625,
      "learning_rate": 9.081270012458202e-05,
      "loss": 0.0012,
      "step": 11342
    },
    {
      "epoch": 5.294282380396733,
      "grad_norm": 0.1337890625,
      "learning_rate": 9.079808879888641e-05,
      "loss": 0.0007,
      "step": 11343
    },
    {
      "epoch": 5.2947491248541425,
      "grad_norm": 0.08935546875,
      "learning_rate": 9.078347767131819e-05,
      "loss": 0.0008,
      "step": 11344
    },
    {
      "epoch": 5.295215869311552,
      "grad_norm": 0.2275390625,
      "learning_rate": 9.076886674219193e-05,
      "loss": 0.0019,
      "step": 11345
    },
    {
      "epoch": 5.295682613768961,
      "grad_norm": 0.0888671875,
      "learning_rate": 9.07542560118222e-05,
      "loss": 0.0037,
      "step": 11346
    },
    {
      "epoch": 5.296149358226371,
      "grad_norm": 0.162109375,
      "learning_rate": 9.073964548052358e-05,
      "loss": 0.0013,
      "step": 11347
    },
    {
      "epoch": 5.296616102683781,
      "grad_norm": 0.09326171875,
      "learning_rate": 9.072503514861068e-05,
      "loss": 0.0008,
      "step": 11348
    },
    {
      "epoch": 5.297082847141191,
      "grad_norm": 0.30078125,
      "learning_rate": 9.071042501639806e-05,
      "loss": 0.002,
      "step": 11349
    },
    {
      "epoch": 5.2975495915985995,
      "grad_norm": 0.047119140625,
      "learning_rate": 9.069581508420029e-05,
      "loss": 0.0006,
      "step": 11350
    },
    {
      "epoch": 5.298016336056009,
      "grad_norm": 0.34375,
      "learning_rate": 9.068120535233195e-05,
      "loss": 0.0028,
      "step": 11351
    },
    {
      "epoch": 5.298483080513419,
      "grad_norm": 0.1611328125,
      "learning_rate": 9.066659582110761e-05,
      "loss": 0.0008,
      "step": 11352
    },
    {
      "epoch": 5.298949824970828,
      "grad_norm": 0.038330078125,
      "learning_rate": 9.065198649084178e-05,
      "loss": 0.0005,
      "step": 11353
    },
    {
      "epoch": 5.299416569428238,
      "grad_norm": 0.1396484375,
      "learning_rate": 9.063737736184909e-05,
      "loss": 0.0011,
      "step": 11354
    },
    {
      "epoch": 5.299883313885648,
      "grad_norm": 0.30859375,
      "learning_rate": 9.062276843444406e-05,
      "loss": 0.002,
      "step": 11355
    },
    {
      "epoch": 5.300350058343057,
      "grad_norm": 0.07861328125,
      "learning_rate": 9.060815970894122e-05,
      "loss": 0.001,
      "step": 11356
    },
    {
      "epoch": 5.300816802800467,
      "grad_norm": 0.15234375,
      "learning_rate": 9.05935511856551e-05,
      "loss": 0.0009,
      "step": 11357
    },
    {
      "epoch": 5.301283547257876,
      "grad_norm": 0.58984375,
      "learning_rate": 9.057894286490029e-05,
      "loss": 0.0049,
      "step": 11358
    },
    {
      "epoch": 5.301750291715286,
      "grad_norm": 0.0269775390625,
      "learning_rate": 9.056433474699127e-05,
      "loss": 0.0005,
      "step": 11359
    },
    {
      "epoch": 5.302217036172696,
      "grad_norm": 0.28125,
      "learning_rate": 9.054972683224259e-05,
      "loss": 0.002,
      "step": 11360
    },
    {
      "epoch": 5.302683780630105,
      "grad_norm": 0.357421875,
      "learning_rate": 9.053511912096884e-05,
      "loss": 0.0105,
      "step": 11361
    },
    {
      "epoch": 5.303150525087514,
      "grad_norm": 0.11181640625,
      "learning_rate": 9.05205116134844e-05,
      "loss": 0.0009,
      "step": 11362
    },
    {
      "epoch": 5.303617269544924,
      "grad_norm": 0.2236328125,
      "learning_rate": 9.05059043101039e-05,
      "loss": 0.0046,
      "step": 11363
    },
    {
      "epoch": 5.304084014002334,
      "grad_norm": 0.232421875,
      "learning_rate": 9.04912972111418e-05,
      "loss": 0.0033,
      "step": 11364
    },
    {
      "epoch": 5.304550758459743,
      "grad_norm": 0.1884765625,
      "learning_rate": 9.047669031691262e-05,
      "loss": 0.001,
      "step": 11365
    },
    {
      "epoch": 5.305017502917153,
      "grad_norm": 0.19921875,
      "learning_rate": 9.046208362773087e-05,
      "loss": 0.0036,
      "step": 11366
    },
    {
      "epoch": 5.3054842473745625,
      "grad_norm": 0.0791015625,
      "learning_rate": 9.044747714391103e-05,
      "loss": 0.0008,
      "step": 11367
    },
    {
      "epoch": 5.305950991831972,
      "grad_norm": 0.1875,
      "learning_rate": 9.043287086576763e-05,
      "loss": 0.0017,
      "step": 11368
    },
    {
      "epoch": 5.306417736289381,
      "grad_norm": 0.5078125,
      "learning_rate": 9.041826479361511e-05,
      "loss": 0.0077,
      "step": 11369
    },
    {
      "epoch": 5.306884480746791,
      "grad_norm": 0.2255859375,
      "learning_rate": 9.040365892776799e-05,
      "loss": 0.0014,
      "step": 11370
    },
    {
      "epoch": 5.307351225204201,
      "grad_norm": 0.06396484375,
      "learning_rate": 9.038905326854075e-05,
      "loss": 0.0007,
      "step": 11371
    },
    {
      "epoch": 5.307817969661611,
      "grad_norm": 0.14453125,
      "learning_rate": 9.037444781624781e-05,
      "loss": 0.0011,
      "step": 11372
    },
    {
      "epoch": 5.3082847141190195,
      "grad_norm": 0.6796875,
      "learning_rate": 9.035984257120375e-05,
      "loss": 0.0057,
      "step": 11373
    },
    {
      "epoch": 5.308751458576429,
      "grad_norm": 0.07958984375,
      "learning_rate": 9.034523753372295e-05,
      "loss": 0.0009,
      "step": 11374
    },
    {
      "epoch": 5.309218203033839,
      "grad_norm": 0.1728515625,
      "learning_rate": 9.03306327041199e-05,
      "loss": 0.0013,
      "step": 11375
    },
    {
      "epoch": 5.309684947491249,
      "grad_norm": 0.173828125,
      "learning_rate": 9.031602808270903e-05,
      "loss": 0.0014,
      "step": 11376
    },
    {
      "epoch": 5.310151691948658,
      "grad_norm": 0.0927734375,
      "learning_rate": 9.030142366980486e-05,
      "loss": 0.001,
      "step": 11377
    },
    {
      "epoch": 5.310618436406068,
      "grad_norm": 0.072265625,
      "learning_rate": 9.02868194657218e-05,
      "loss": 0.0007,
      "step": 11378
    },
    {
      "epoch": 5.311085180863477,
      "grad_norm": 0.314453125,
      "learning_rate": 9.027221547077424e-05,
      "loss": 0.002,
      "step": 11379
    },
    {
      "epoch": 5.311551925320887,
      "grad_norm": 0.0703125,
      "learning_rate": 9.025761168527673e-05,
      "loss": 0.0036,
      "step": 11380
    },
    {
      "epoch": 5.312018669778296,
      "grad_norm": 0.3515625,
      "learning_rate": 9.024300810954363e-05,
      "loss": 0.0029,
      "step": 11381
    },
    {
      "epoch": 5.312485414235706,
      "grad_norm": 0.251953125,
      "learning_rate": 9.022840474388938e-05,
      "loss": 0.0015,
      "step": 11382
    },
    {
      "epoch": 5.312952158693116,
      "grad_norm": 0.1240234375,
      "learning_rate": 9.021380158862843e-05,
      "loss": 0.0008,
      "step": 11383
    },
    {
      "epoch": 5.3134189031505255,
      "grad_norm": 0.10595703125,
      "learning_rate": 9.01991986440752e-05,
      "loss": 0.001,
      "step": 11384
    },
    {
      "epoch": 5.313885647607934,
      "grad_norm": 0.12060546875,
      "learning_rate": 9.018459591054408e-05,
      "loss": 0.0032,
      "step": 11385
    },
    {
      "epoch": 5.314352392065344,
      "grad_norm": 0.0673828125,
      "learning_rate": 9.016999338834951e-05,
      "loss": 0.0007,
      "step": 11386
    },
    {
      "epoch": 5.314819136522754,
      "grad_norm": 0.06884765625,
      "learning_rate": 9.015539107780589e-05,
      "loss": 0.0007,
      "step": 11387
    },
    {
      "epoch": 5.315285880980164,
      "grad_norm": 0.29296875,
      "learning_rate": 9.014078897922761e-05,
      "loss": 0.0025,
      "step": 11388
    },
    {
      "epoch": 5.315752625437573,
      "grad_norm": 0.26171875,
      "learning_rate": 9.012618709292908e-05,
      "loss": 0.0043,
      "step": 11389
    },
    {
      "epoch": 5.3162193698949824,
      "grad_norm": 0.373046875,
      "learning_rate": 9.011158541922473e-05,
      "loss": 0.0025,
      "step": 11390
    },
    {
      "epoch": 5.316686114352392,
      "grad_norm": 0.30859375,
      "learning_rate": 9.00969839584289e-05,
      "loss": 0.0018,
      "step": 11391
    },
    {
      "epoch": 5.317152858809802,
      "grad_norm": 0.2490234375,
      "learning_rate": 9.008238271085596e-05,
      "loss": 0.0018,
      "step": 11392
    },
    {
      "epoch": 5.317619603267211,
      "grad_norm": 0.3203125,
      "learning_rate": 9.006778167682036e-05,
      "loss": 0.0052,
      "step": 11393
    },
    {
      "epoch": 5.318086347724621,
      "grad_norm": 0.1337890625,
      "learning_rate": 9.005318085663644e-05,
      "loss": 0.0008,
      "step": 11394
    },
    {
      "epoch": 5.3185530921820305,
      "grad_norm": 0.13671875,
      "learning_rate": 9.003858025061857e-05,
      "loss": 0.0011,
      "step": 11395
    },
    {
      "epoch": 5.3190198366394394,
      "grad_norm": 0.041748046875,
      "learning_rate": 9.00239798590811e-05,
      "loss": 0.0005,
      "step": 11396
    },
    {
      "epoch": 5.319486581096849,
      "grad_norm": 0.2099609375,
      "learning_rate": 9.000937968233844e-05,
      "loss": 0.0013,
      "step": 11397
    },
    {
      "epoch": 5.319953325554259,
      "grad_norm": 0.26953125,
      "learning_rate": 8.999477972070492e-05,
      "loss": 0.0027,
      "step": 11398
    },
    {
      "epoch": 5.320420070011669,
      "grad_norm": 0.0810546875,
      "learning_rate": 8.998017997449485e-05,
      "loss": 0.0028,
      "step": 11399
    },
    {
      "epoch": 5.320886814469079,
      "grad_norm": 0.373046875,
      "learning_rate": 8.996558044402266e-05,
      "loss": 0.004,
      "step": 11400
    },
    {
      "epoch": 5.3213535589264875,
      "grad_norm": 0.10546875,
      "learning_rate": 8.995098112960266e-05,
      "loss": 0.0013,
      "step": 11401
    },
    {
      "epoch": 5.321820303383897,
      "grad_norm": 0.275390625,
      "learning_rate": 8.993638203154916e-05,
      "loss": 0.0032,
      "step": 11402
    },
    {
      "epoch": 5.322287047841307,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.992178315017654e-05,
      "loss": 0.0006,
      "step": 11403
    },
    {
      "epoch": 5.322753792298716,
      "grad_norm": 0.208984375,
      "learning_rate": 8.99071844857991e-05,
      "loss": 0.0048,
      "step": 11404
    },
    {
      "epoch": 5.323220536756126,
      "grad_norm": 0.416015625,
      "learning_rate": 8.98925860387312e-05,
      "loss": 0.0073,
      "step": 11405
    },
    {
      "epoch": 5.323687281213536,
      "grad_norm": 0.0634765625,
      "learning_rate": 8.98779878092871e-05,
      "loss": 0.0006,
      "step": 11406
    },
    {
      "epoch": 5.324154025670945,
      "grad_norm": 0.11083984375,
      "learning_rate": 8.986338979778119e-05,
      "loss": 0.0009,
      "step": 11407
    },
    {
      "epoch": 5.324620770128354,
      "grad_norm": 0.1474609375,
      "learning_rate": 8.984879200452775e-05,
      "loss": 0.0009,
      "step": 11408
    },
    {
      "epoch": 5.325087514585764,
      "grad_norm": 0.11474609375,
      "learning_rate": 8.983419442984106e-05,
      "loss": 0.0009,
      "step": 11409
    },
    {
      "epoch": 5.325554259043174,
      "grad_norm": 0.197265625,
      "learning_rate": 8.981959707403545e-05,
      "loss": 0.0015,
      "step": 11410
    },
    {
      "epoch": 5.326021003500584,
      "grad_norm": 0.041259765625,
      "learning_rate": 8.980499993742519e-05,
      "loss": 0.0005,
      "step": 11411
    },
    {
      "epoch": 5.326487747957993,
      "grad_norm": 0.1455078125,
      "learning_rate": 8.979040302032461e-05,
      "loss": 0.0008,
      "step": 11412
    },
    {
      "epoch": 5.326954492415402,
      "grad_norm": 0.08935546875,
      "learning_rate": 8.977580632304799e-05,
      "loss": 0.0007,
      "step": 11413
    },
    {
      "epoch": 5.327421236872812,
      "grad_norm": 0.06884765625,
      "learning_rate": 8.976120984590957e-05,
      "loss": 0.0029,
      "step": 11414
    },
    {
      "epoch": 5.327887981330222,
      "grad_norm": 0.09423828125,
      "learning_rate": 8.974661358922368e-05,
      "loss": 0.0009,
      "step": 11415
    },
    {
      "epoch": 5.328354725787631,
      "grad_norm": 0.1044921875,
      "learning_rate": 8.973201755330458e-05,
      "loss": 0.0008,
      "step": 11416
    },
    {
      "epoch": 5.328821470245041,
      "grad_norm": 0.1943359375,
      "learning_rate": 8.971742173846655e-05,
      "loss": 0.0031,
      "step": 11417
    },
    {
      "epoch": 5.3292882147024505,
      "grad_norm": 0.19921875,
      "learning_rate": 8.97028261450238e-05,
      "loss": 0.0012,
      "step": 11418
    },
    {
      "epoch": 5.32975495915986,
      "grad_norm": 0.3046875,
      "learning_rate": 8.968823077329065e-05,
      "loss": 0.0022,
      "step": 11419
    },
    {
      "epoch": 5.330221703617269,
      "grad_norm": 0.236328125,
      "learning_rate": 8.967363562358134e-05,
      "loss": 0.0025,
      "step": 11420
    },
    {
      "epoch": 5.330688448074679,
      "grad_norm": 0.2421875,
      "learning_rate": 8.965904069621008e-05,
      "loss": 0.0035,
      "step": 11421
    },
    {
      "epoch": 5.331155192532089,
      "grad_norm": 0.32421875,
      "learning_rate": 8.964444599149119e-05,
      "loss": 0.0037,
      "step": 11422
    },
    {
      "epoch": 5.331621936989499,
      "grad_norm": 0.150390625,
      "learning_rate": 8.962985150973885e-05,
      "loss": 0.0011,
      "step": 11423
    },
    {
      "epoch": 5.3320886814469075,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.961525725126729e-05,
      "loss": 0.001,
      "step": 11424
    },
    {
      "epoch": 5.332555425904317,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.960066321639079e-05,
      "loss": 0.0026,
      "step": 11425
    },
    {
      "epoch": 5.333022170361727,
      "grad_norm": 0.22265625,
      "learning_rate": 8.958606940542355e-05,
      "loss": 0.0015,
      "step": 11426
    },
    {
      "epoch": 5.333488914819137,
      "grad_norm": 0.10986328125,
      "learning_rate": 8.95714758186798e-05,
      "loss": 0.0012,
      "step": 11427
    },
    {
      "epoch": 5.333955659276546,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.95568824564737e-05,
      "loss": 0.0012,
      "step": 11428
    },
    {
      "epoch": 5.334422403733956,
      "grad_norm": 0.255859375,
      "learning_rate": 8.954228931911956e-05,
      "loss": 0.002,
      "step": 11429
    },
    {
      "epoch": 5.334889148191365,
      "grad_norm": 0.2021484375,
      "learning_rate": 8.952769640693151e-05,
      "loss": 0.0015,
      "step": 11430
    },
    {
      "epoch": 5.335355892648775,
      "grad_norm": 0.07568359375,
      "learning_rate": 8.951310372022377e-05,
      "loss": 0.0007,
      "step": 11431
    },
    {
      "epoch": 5.335822637106184,
      "grad_norm": 0.412109375,
      "learning_rate": 8.949851125931055e-05,
      "loss": 0.0018,
      "step": 11432
    },
    {
      "epoch": 5.336289381563594,
      "grad_norm": 0.0908203125,
      "learning_rate": 8.948391902450603e-05,
      "loss": 0.0007,
      "step": 11433
    },
    {
      "epoch": 5.336756126021004,
      "grad_norm": 0.166015625,
      "learning_rate": 8.946932701612442e-05,
      "loss": 0.0012,
      "step": 11434
    },
    {
      "epoch": 5.3372228704784135,
      "grad_norm": 0.2890625,
      "learning_rate": 8.945473523447985e-05,
      "loss": 0.0017,
      "step": 11435
    },
    {
      "epoch": 5.337689614935822,
      "grad_norm": 0.03759765625,
      "learning_rate": 8.944014367988656e-05,
      "loss": 0.0004,
      "step": 11436
    },
    {
      "epoch": 5.338156359393232,
      "grad_norm": 0.052001953125,
      "learning_rate": 8.942555235265868e-05,
      "loss": 0.0005,
      "step": 11437
    },
    {
      "epoch": 5.338623103850642,
      "grad_norm": 0.384765625,
      "learning_rate": 8.941096125311038e-05,
      "loss": 0.0014,
      "step": 11438
    },
    {
      "epoch": 5.339089848308051,
      "grad_norm": 0.09765625,
      "learning_rate": 8.939637038155585e-05,
      "loss": 0.0007,
      "step": 11439
    },
    {
      "epoch": 5.339556592765461,
      "grad_norm": 0.251953125,
      "learning_rate": 8.938177973830923e-05,
      "loss": 0.0015,
      "step": 11440
    },
    {
      "epoch": 5.3400233372228705,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.936718932368466e-05,
      "loss": 0.0007,
      "step": 11441
    },
    {
      "epoch": 5.34049008168028,
      "grad_norm": 0.306640625,
      "learning_rate": 8.935259913799632e-05,
      "loss": 0.0091,
      "step": 11442
    },
    {
      "epoch": 5.34095682613769,
      "grad_norm": 0.482421875,
      "learning_rate": 8.933800918155832e-05,
      "loss": 0.0117,
      "step": 11443
    },
    {
      "epoch": 5.341423570595099,
      "grad_norm": 0.28125,
      "learning_rate": 8.932341945468484e-05,
      "loss": 0.0013,
      "step": 11444
    },
    {
      "epoch": 5.341890315052509,
      "grad_norm": 0.15625,
      "learning_rate": 8.930882995768993e-05,
      "loss": 0.0011,
      "step": 11445
    },
    {
      "epoch": 5.342357059509919,
      "grad_norm": 0.23046875,
      "learning_rate": 8.929424069088781e-05,
      "loss": 0.0012,
      "step": 11446
    },
    {
      "epoch": 5.3428238039673275,
      "grad_norm": 0.283203125,
      "learning_rate": 8.927965165459256e-05,
      "loss": 0.0017,
      "step": 11447
    },
    {
      "epoch": 5.343290548424737,
      "grad_norm": 0.13671875,
      "learning_rate": 8.926506284911829e-05,
      "loss": 0.0028,
      "step": 11448
    },
    {
      "epoch": 5.343757292882147,
      "grad_norm": 0.515625,
      "learning_rate": 8.925047427477916e-05,
      "loss": 0.0058,
      "step": 11449
    },
    {
      "epoch": 5.344224037339557,
      "grad_norm": 0.15234375,
      "learning_rate": 8.923588593188923e-05,
      "loss": 0.0011,
      "step": 11450
    },
    {
      "epoch": 5.344690781796966,
      "grad_norm": 0.322265625,
      "learning_rate": 8.92212978207626e-05,
      "loss": 0.0019,
      "step": 11451
    },
    {
      "epoch": 5.345157526254376,
      "grad_norm": 0.2578125,
      "learning_rate": 8.920670994171341e-05,
      "loss": 0.0019,
      "step": 11452
    },
    {
      "epoch": 5.345624270711785,
      "grad_norm": 0.310546875,
      "learning_rate": 8.919212229505574e-05,
      "loss": 0.0017,
      "step": 11453
    },
    {
      "epoch": 5.346091015169195,
      "grad_norm": 0.1796875,
      "learning_rate": 8.917753488110366e-05,
      "loss": 0.0044,
      "step": 11454
    },
    {
      "epoch": 5.346557759626604,
      "grad_norm": 0.248046875,
      "learning_rate": 8.916294770017123e-05,
      "loss": 0.0014,
      "step": 11455
    },
    {
      "epoch": 5.347024504084014,
      "grad_norm": 0.306640625,
      "learning_rate": 8.914836075257264e-05,
      "loss": 0.0049,
      "step": 11456
    },
    {
      "epoch": 5.347491248541424,
      "grad_norm": 0.27734375,
      "learning_rate": 8.913377403862181e-05,
      "loss": 0.0013,
      "step": 11457
    },
    {
      "epoch": 5.3479579929988335,
      "grad_norm": 0.2333984375,
      "learning_rate": 8.91191875586329e-05,
      "loss": 0.0012,
      "step": 11458
    },
    {
      "epoch": 5.348424737456242,
      "grad_norm": 0.134765625,
      "learning_rate": 8.910460131291997e-05,
      "loss": 0.0012,
      "step": 11459
    },
    {
      "epoch": 5.348891481913652,
      "grad_norm": 0.310546875,
      "learning_rate": 8.909001530179702e-05,
      "loss": 0.0057,
      "step": 11460
    },
    {
      "epoch": 5.349358226371062,
      "grad_norm": 0.2255859375,
      "learning_rate": 8.907542952557817e-05,
      "loss": 0.0024,
      "step": 11461
    },
    {
      "epoch": 5.349824970828472,
      "grad_norm": 0.126953125,
      "learning_rate": 8.906084398457746e-05,
      "loss": 0.0007,
      "step": 11462
    },
    {
      "epoch": 5.350291715285881,
      "grad_norm": 0.220703125,
      "learning_rate": 8.904625867910888e-05,
      "loss": 0.0017,
      "step": 11463
    },
    {
      "epoch": 5.3507584597432905,
      "grad_norm": 0.05517578125,
      "learning_rate": 8.903167360948653e-05,
      "loss": 0.0008,
      "step": 11464
    },
    {
      "epoch": 5.3512252042007,
      "grad_norm": 0.130859375,
      "learning_rate": 8.901708877602443e-05,
      "loss": 0.0028,
      "step": 11465
    },
    {
      "epoch": 5.35169194865811,
      "grad_norm": 0.095703125,
      "learning_rate": 8.900250417903657e-05,
      "loss": 0.0006,
      "step": 11466
    },
    {
      "epoch": 5.352158693115519,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.898791981883699e-05,
      "loss": 0.001,
      "step": 11467
    },
    {
      "epoch": 5.352625437572929,
      "grad_norm": 1.359375,
      "learning_rate": 8.897333569573974e-05,
      "loss": 0.0062,
      "step": 11468
    },
    {
      "epoch": 5.3530921820303385,
      "grad_norm": 0.06787109375,
      "learning_rate": 8.89587518100588e-05,
      "loss": 0.0005,
      "step": 11469
    },
    {
      "epoch": 5.353558926487748,
      "grad_norm": 0.05908203125,
      "learning_rate": 8.894416816210816e-05,
      "loss": 0.0005,
      "step": 11470
    },
    {
      "epoch": 5.354025670945157,
      "grad_norm": 0.26171875,
      "learning_rate": 8.892958475220186e-05,
      "loss": 0.0021,
      "step": 11471
    },
    {
      "epoch": 5.354492415402567,
      "grad_norm": 0.1484375,
      "learning_rate": 8.89150015806539e-05,
      "loss": 0.001,
      "step": 11472
    },
    {
      "epoch": 5.354959159859977,
      "grad_norm": 0.1669921875,
      "learning_rate": 8.890041864777823e-05,
      "loss": 0.0025,
      "step": 11473
    },
    {
      "epoch": 5.355425904317387,
      "grad_norm": 0.396484375,
      "learning_rate": 8.888583595388888e-05,
      "loss": 0.0035,
      "step": 11474
    },
    {
      "epoch": 5.3558926487747955,
      "grad_norm": 0.05078125,
      "learning_rate": 8.887125349929981e-05,
      "loss": 0.0008,
      "step": 11475
    },
    {
      "epoch": 5.356359393232205,
      "grad_norm": 0.302734375,
      "learning_rate": 8.885667128432499e-05,
      "loss": 0.0023,
      "step": 11476
    },
    {
      "epoch": 5.356826137689615,
      "grad_norm": 0.2314453125,
      "learning_rate": 8.88420893092784e-05,
      "loss": 0.0061,
      "step": 11477
    },
    {
      "epoch": 5.357292882147025,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.882750757447402e-05,
      "loss": 0.0014,
      "step": 11478
    },
    {
      "epoch": 5.357759626604434,
      "grad_norm": 0.22265625,
      "learning_rate": 8.88129260802258e-05,
      "loss": 0.0015,
      "step": 11479
    },
    {
      "epoch": 5.358226371061844,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.879834482684768e-05,
      "loss": 0.0041,
      "step": 11480
    },
    {
      "epoch": 5.358693115519253,
      "grad_norm": 0.203125,
      "learning_rate": 8.878376381465364e-05,
      "loss": 0.0026,
      "step": 11481
    },
    {
      "epoch": 5.359159859976662,
      "grad_norm": 0.427734375,
      "learning_rate": 8.87691830439576e-05,
      "loss": 0.003,
      "step": 11482
    },
    {
      "epoch": 5.359626604434072,
      "grad_norm": 0.318359375,
      "learning_rate": 8.875460251507352e-05,
      "loss": 0.0039,
      "step": 11483
    },
    {
      "epoch": 5.360093348891482,
      "grad_norm": 0.1015625,
      "learning_rate": 8.87400222283153e-05,
      "loss": 0.0041,
      "step": 11484
    },
    {
      "epoch": 5.360560093348892,
      "grad_norm": 0.07080078125,
      "learning_rate": 8.872544218399691e-05,
      "loss": 0.0006,
      "step": 11485
    },
    {
      "epoch": 5.3610268378063015,
      "grad_norm": 0.50390625,
      "learning_rate": 8.871086238243227e-05,
      "loss": 0.0034,
      "step": 11486
    },
    {
      "epoch": 5.36149358226371,
      "grad_norm": 0.1494140625,
      "learning_rate": 8.869628282393526e-05,
      "loss": 0.0012,
      "step": 11487
    },
    {
      "epoch": 5.36196032672112,
      "grad_norm": 0.060546875,
      "learning_rate": 8.868170350881984e-05,
      "loss": 0.0006,
      "step": 11488
    },
    {
      "epoch": 5.36242707117853,
      "grad_norm": 0.40625,
      "learning_rate": 8.86671244373999e-05,
      "loss": 0.0023,
      "step": 11489
    },
    {
      "epoch": 5.362893815635939,
      "grad_norm": 0.062255859375,
      "learning_rate": 8.865254560998935e-05,
      "loss": 0.0006,
      "step": 11490
    },
    {
      "epoch": 5.363360560093349,
      "grad_norm": 0.1640625,
      "learning_rate": 8.863796702690208e-05,
      "loss": 0.0009,
      "step": 11491
    },
    {
      "epoch": 5.3638273045507585,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.8623388688452e-05,
      "loss": 0.0014,
      "step": 11492
    },
    {
      "epoch": 5.364294049008168,
      "grad_norm": 0.13671875,
      "learning_rate": 8.860881059495297e-05,
      "loss": 0.001,
      "step": 11493
    },
    {
      "epoch": 5.364760793465577,
      "grad_norm": 0.296875,
      "learning_rate": 8.859423274671887e-05,
      "loss": 0.0022,
      "step": 11494
    },
    {
      "epoch": 5.365227537922987,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.857965514406361e-05,
      "loss": 0.0009,
      "step": 11495
    },
    {
      "epoch": 5.365694282380397,
      "grad_norm": 0.263671875,
      "learning_rate": 8.856507778730106e-05,
      "loss": 0.0016,
      "step": 11496
    },
    {
      "epoch": 5.366161026837807,
      "grad_norm": 0.30078125,
      "learning_rate": 8.855050067674504e-05,
      "loss": 0.0018,
      "step": 11497
    },
    {
      "epoch": 5.3666277712952155,
      "grad_norm": 0.486328125,
      "learning_rate": 8.853592381270947e-05,
      "loss": 0.0027,
      "step": 11498
    },
    {
      "epoch": 5.367094515752625,
      "grad_norm": 0.0693359375,
      "learning_rate": 8.852134719550817e-05,
      "loss": 0.0007,
      "step": 11499
    },
    {
      "epoch": 5.367561260210035,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.8506770825455e-05,
      "loss": 0.0009,
      "step": 11500
    },
    {
      "epoch": 5.368028004667445,
      "grad_norm": 0.1982421875,
      "learning_rate": 8.849219470286381e-05,
      "loss": 0.0014,
      "step": 11501
    },
    {
      "epoch": 5.368494749124854,
      "grad_norm": 0.044921875,
      "learning_rate": 8.847761882804844e-05,
      "loss": 0.0004,
      "step": 11502
    },
    {
      "epoch": 5.368961493582264,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.846304320132272e-05,
      "loss": 0.0011,
      "step": 11503
    },
    {
      "epoch": 5.369428238039673,
      "grad_norm": 0.038330078125,
      "learning_rate": 8.844846782300052e-05,
      "loss": 0.0005,
      "step": 11504
    },
    {
      "epoch": 5.369894982497083,
      "grad_norm": 0.134765625,
      "learning_rate": 8.843389269339558e-05,
      "loss": 0.0011,
      "step": 11505
    },
    {
      "epoch": 5.370361726954492,
      "grad_norm": 0.322265625,
      "learning_rate": 8.841931781282175e-05,
      "loss": 0.0016,
      "step": 11506
    },
    {
      "epoch": 5.370828471411902,
      "grad_norm": 0.10205078125,
      "learning_rate": 8.840474318159287e-05,
      "loss": 0.0044,
      "step": 11507
    },
    {
      "epoch": 5.371295215869312,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.839016880002275e-05,
      "loss": 0.0038,
      "step": 11508
    },
    {
      "epoch": 5.3717619603267215,
      "grad_norm": 0.3203125,
      "learning_rate": 8.837559466842515e-05,
      "loss": 0.0016,
      "step": 11509
    },
    {
      "epoch": 5.37222870478413,
      "grad_norm": 0.08447265625,
      "learning_rate": 8.83610207871139e-05,
      "loss": 0.0006,
      "step": 11510
    },
    {
      "epoch": 5.37269544924154,
      "grad_norm": 0.14453125,
      "learning_rate": 8.83464471564028e-05,
      "loss": 0.0007,
      "step": 11511
    },
    {
      "epoch": 5.37316219369895,
      "grad_norm": 0.103515625,
      "learning_rate": 8.83318737766056e-05,
      "loss": 0.0046,
      "step": 11512
    },
    {
      "epoch": 5.37362893815636,
      "grad_norm": 0.099609375,
      "learning_rate": 8.831730064803613e-05,
      "loss": 0.0007,
      "step": 11513
    },
    {
      "epoch": 5.374095682613769,
      "grad_norm": 0.09326171875,
      "learning_rate": 8.830272777100813e-05,
      "loss": 0.0032,
      "step": 11514
    },
    {
      "epoch": 5.3745624270711785,
      "grad_norm": 0.0966796875,
      "learning_rate": 8.828815514583538e-05,
      "loss": 0.001,
      "step": 11515
    },
    {
      "epoch": 5.375029171528588,
      "grad_norm": 0.185546875,
      "learning_rate": 8.827358277283163e-05,
      "loss": 0.0013,
      "step": 11516
    },
    {
      "epoch": 5.375495915985998,
      "grad_norm": 0.0791015625,
      "learning_rate": 8.825901065231067e-05,
      "loss": 0.0037,
      "step": 11517
    },
    {
      "epoch": 5.375962660443407,
      "grad_norm": 0.08984375,
      "learning_rate": 8.824443878458624e-05,
      "loss": 0.0006,
      "step": 11518
    },
    {
      "epoch": 5.376429404900817,
      "grad_norm": 0.251953125,
      "learning_rate": 8.822986716997206e-05,
      "loss": 0.0027,
      "step": 11519
    },
    {
      "epoch": 5.376896149358227,
      "grad_norm": 0.2060546875,
      "learning_rate": 8.821529580878192e-05,
      "loss": 0.0015,
      "step": 11520
    },
    {
      "epoch": 5.377362893815636,
      "grad_norm": 0.0751953125,
      "learning_rate": 8.820072470132955e-05,
      "loss": 0.0008,
      "step": 11521
    },
    {
      "epoch": 5.377829638273045,
      "grad_norm": 0.0306396484375,
      "learning_rate": 8.818615384792863e-05,
      "loss": 0.0004,
      "step": 11522
    },
    {
      "epoch": 5.378296382730455,
      "grad_norm": 0.087890625,
      "learning_rate": 8.817158324889297e-05,
      "loss": 0.0009,
      "step": 11523
    },
    {
      "epoch": 5.378763127187865,
      "grad_norm": 0.1796875,
      "learning_rate": 8.815701290453622e-05,
      "loss": 0.0034,
      "step": 11524
    },
    {
      "epoch": 5.379229871645274,
      "grad_norm": 0.1787109375,
      "learning_rate": 8.814244281517213e-05,
      "loss": 0.0013,
      "step": 11525
    },
    {
      "epoch": 5.379696616102684,
      "grad_norm": 0.146484375,
      "learning_rate": 8.812787298111439e-05,
      "loss": 0.0009,
      "step": 11526
    },
    {
      "epoch": 5.380163360560093,
      "grad_norm": 0.228515625,
      "learning_rate": 8.811330340267672e-05,
      "loss": 0.0014,
      "step": 11527
    },
    {
      "epoch": 5.380630105017503,
      "grad_norm": 0.142578125,
      "learning_rate": 8.809873408017284e-05,
      "loss": 0.001,
      "step": 11528
    },
    {
      "epoch": 5.381096849474912,
      "grad_norm": 0.193359375,
      "learning_rate": 8.808416501391638e-05,
      "loss": 0.0011,
      "step": 11529
    },
    {
      "epoch": 5.381563593932322,
      "grad_norm": 0.140625,
      "learning_rate": 8.806959620422111e-05,
      "loss": 0.0011,
      "step": 11530
    },
    {
      "epoch": 5.382030338389732,
      "grad_norm": 0.16796875,
      "learning_rate": 8.805502765140066e-05,
      "loss": 0.001,
      "step": 11531
    },
    {
      "epoch": 5.3824970828471415,
      "grad_norm": 0.0673828125,
      "learning_rate": 8.804045935576869e-05,
      "loss": 0.0006,
      "step": 11532
    },
    {
      "epoch": 5.38296382730455,
      "grad_norm": 0.1806640625,
      "learning_rate": 8.802589131763892e-05,
      "loss": 0.0024,
      "step": 11533
    },
    {
      "epoch": 5.38343057176196,
      "grad_norm": 0.06689453125,
      "learning_rate": 8.801132353732499e-05,
      "loss": 0.0006,
      "step": 11534
    },
    {
      "epoch": 5.38389731621937,
      "grad_norm": 0.06396484375,
      "learning_rate": 8.799675601514059e-05,
      "loss": 0.0006,
      "step": 11535
    },
    {
      "epoch": 5.38436406067678,
      "grad_norm": 0.10107421875,
      "learning_rate": 8.798218875139932e-05,
      "loss": 0.0008,
      "step": 11536
    },
    {
      "epoch": 5.384830805134189,
      "grad_norm": 0.0966796875,
      "learning_rate": 8.796762174641487e-05,
      "loss": 0.0007,
      "step": 11537
    },
    {
      "epoch": 5.3852975495915985,
      "grad_norm": 0.042236328125,
      "learning_rate": 8.795305500050089e-05,
      "loss": 0.002,
      "step": 11538
    },
    {
      "epoch": 5.385764294049008,
      "grad_norm": 0.26171875,
      "learning_rate": 8.793848851397096e-05,
      "loss": 0.0022,
      "step": 11539
    },
    {
      "epoch": 5.386231038506418,
      "grad_norm": 0.244140625,
      "learning_rate": 8.79239222871388e-05,
      "loss": 0.0035,
      "step": 11540
    },
    {
      "epoch": 5.386697782963827,
      "grad_norm": 0.0294189453125,
      "learning_rate": 8.790935632031797e-05,
      "loss": 0.0004,
      "step": 11541
    },
    {
      "epoch": 5.387164527421237,
      "grad_norm": 0.033447265625,
      "learning_rate": 8.789479061382213e-05,
      "loss": 0.0004,
      "step": 11542
    },
    {
      "epoch": 5.3876312718786465,
      "grad_norm": 0.1103515625,
      "learning_rate": 8.788022516796485e-05,
      "loss": 0.0008,
      "step": 11543
    },
    {
      "epoch": 5.388098016336056,
      "grad_norm": 0.07177734375,
      "learning_rate": 8.78656599830598e-05,
      "loss": 0.0007,
      "step": 11544
    },
    {
      "epoch": 5.388564760793465,
      "grad_norm": 0.435546875,
      "learning_rate": 8.785109505942052e-05,
      "loss": 0.005,
      "step": 11545
    },
    {
      "epoch": 5.389031505250875,
      "grad_norm": 0.8203125,
      "learning_rate": 8.783653039736067e-05,
      "loss": 0.0103,
      "step": 11546
    },
    {
      "epoch": 5.389498249708285,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.78219659971938e-05,
      "loss": 0.0009,
      "step": 11547
    },
    {
      "epoch": 5.389964994165695,
      "grad_norm": 0.10400390625,
      "learning_rate": 8.780740185923353e-05,
      "loss": 0.0007,
      "step": 11548
    },
    {
      "epoch": 5.3904317386231035,
      "grad_norm": 0.1484375,
      "learning_rate": 8.77928379837934e-05,
      "loss": 0.0029,
      "step": 11549
    },
    {
      "epoch": 5.390898483080513,
      "grad_norm": 0.0908203125,
      "learning_rate": 8.777827437118704e-05,
      "loss": 0.0007,
      "step": 11550
    },
    {
      "epoch": 5.391365227537923,
      "grad_norm": 0.0302734375,
      "learning_rate": 8.776371102172799e-05,
      "loss": 0.0004,
      "step": 11551
    },
    {
      "epoch": 5.391831971995333,
      "grad_norm": 0.083984375,
      "learning_rate": 8.774914793572983e-05,
      "loss": 0.0051,
      "step": 11552
    },
    {
      "epoch": 5.392298716452742,
      "grad_norm": 0.16015625,
      "learning_rate": 8.77345851135061e-05,
      "loss": 0.003,
      "step": 11553
    },
    {
      "epoch": 5.392765460910152,
      "grad_norm": 0.047607421875,
      "learning_rate": 8.772002255537037e-05,
      "loss": 0.0006,
      "step": 11554
    },
    {
      "epoch": 5.393232205367561,
      "grad_norm": 0.07373046875,
      "learning_rate": 8.770546026163612e-05,
      "loss": 0.0043,
      "step": 11555
    },
    {
      "epoch": 5.393698949824971,
      "grad_norm": 0.34375,
      "learning_rate": 8.769089823261701e-05,
      "loss": 0.0036,
      "step": 11556
    },
    {
      "epoch": 5.39416569428238,
      "grad_norm": 0.6484375,
      "learning_rate": 8.767633646862651e-05,
      "loss": 0.0044,
      "step": 11557
    },
    {
      "epoch": 5.39463243873979,
      "grad_norm": 0.052978515625,
      "learning_rate": 8.766177496997812e-05,
      "loss": 0.0007,
      "step": 11558
    },
    {
      "epoch": 5.3950991831972,
      "grad_norm": 0.22265625,
      "learning_rate": 8.764721373698545e-05,
      "loss": 0.0018,
      "step": 11559
    },
    {
      "epoch": 5.3955659276546095,
      "grad_norm": 0.2412109375,
      "learning_rate": 8.763265276996196e-05,
      "loss": 0.004,
      "step": 11560
    },
    {
      "epoch": 5.396032672112018,
      "grad_norm": 0.115234375,
      "learning_rate": 8.761809206922116e-05,
      "loss": 0.0008,
      "step": 11561
    },
    {
      "epoch": 5.396499416569428,
      "grad_norm": 0.208984375,
      "learning_rate": 8.76035316350766e-05,
      "loss": 0.0009,
      "step": 11562
    },
    {
      "epoch": 5.396966161026838,
      "grad_norm": 0.2734375,
      "learning_rate": 8.758897146784175e-05,
      "loss": 0.0025,
      "step": 11563
    },
    {
      "epoch": 5.397432905484248,
      "grad_norm": 0.287109375,
      "learning_rate": 8.757441156783012e-05,
      "loss": 0.0018,
      "step": 11564
    },
    {
      "epoch": 5.397899649941657,
      "grad_norm": 0.05859375,
      "learning_rate": 8.755985193535518e-05,
      "loss": 0.0006,
      "step": 11565
    },
    {
      "epoch": 5.3983663943990665,
      "grad_norm": 0.099609375,
      "learning_rate": 8.754529257073044e-05,
      "loss": 0.001,
      "step": 11566
    },
    {
      "epoch": 5.398833138856476,
      "grad_norm": 0.06982421875,
      "learning_rate": 8.753073347426937e-05,
      "loss": 0.0007,
      "step": 11567
    },
    {
      "epoch": 5.399299883313885,
      "grad_norm": 0.193359375,
      "learning_rate": 8.751617464628544e-05,
      "loss": 0.0015,
      "step": 11568
    },
    {
      "epoch": 5.399766627771295,
      "grad_norm": 0.2578125,
      "learning_rate": 8.750161608709212e-05,
      "loss": 0.0011,
      "step": 11569
    },
    {
      "epoch": 5.400233372228705,
      "grad_norm": 0.1640625,
      "learning_rate": 8.748705779700289e-05,
      "loss": 0.001,
      "step": 11570
    },
    {
      "epoch": 5.400700116686115,
      "grad_norm": 0.061767578125,
      "learning_rate": 8.747249977633114e-05,
      "loss": 0.0005,
      "step": 11571
    },
    {
      "epoch": 5.4011668611435235,
      "grad_norm": 0.0849609375,
      "learning_rate": 8.745794202539042e-05,
      "loss": 0.0006,
      "step": 11572
    },
    {
      "epoch": 5.401633605600933,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.744338454449412e-05,
      "loss": 0.0009,
      "step": 11573
    },
    {
      "epoch": 5.402100350058343,
      "grad_norm": 0.07080078125,
      "learning_rate": 8.742882733395566e-05,
      "loss": 0.001,
      "step": 11574
    },
    {
      "epoch": 5.402567094515753,
      "grad_norm": 0.37109375,
      "learning_rate": 8.74142703940885e-05,
      "loss": 0.0018,
      "step": 11575
    },
    {
      "epoch": 5.403033838973162,
      "grad_norm": 0.0703125,
      "learning_rate": 8.739971372520607e-05,
      "loss": 0.0006,
      "step": 11576
    },
    {
      "epoch": 5.403500583430572,
      "grad_norm": 0.0673828125,
      "learning_rate": 8.738515732762179e-05,
      "loss": 0.0008,
      "step": 11577
    },
    {
      "epoch": 5.403967327887981,
      "grad_norm": 0.234375,
      "learning_rate": 8.737060120164904e-05,
      "loss": 0.0014,
      "step": 11578
    },
    {
      "epoch": 5.404434072345391,
      "grad_norm": 0.07275390625,
      "learning_rate": 8.735604534760129e-05,
      "loss": 0.0009,
      "step": 11579
    },
    {
      "epoch": 5.4049008168028,
      "grad_norm": 0.053955078125,
      "learning_rate": 8.73414897657919e-05,
      "loss": 0.0005,
      "step": 11580
    },
    {
      "epoch": 5.40536756126021,
      "grad_norm": 0.2138671875,
      "learning_rate": 8.732693445653427e-05,
      "loss": 0.0032,
      "step": 11581
    },
    {
      "epoch": 5.40583430571762,
      "grad_norm": 0.06591796875,
      "learning_rate": 8.731237942014182e-05,
      "loss": 0.0024,
      "step": 11582
    },
    {
      "epoch": 5.4063010501750295,
      "grad_norm": 0.32421875,
      "learning_rate": 8.729782465692792e-05,
      "loss": 0.0053,
      "step": 11583
    },
    {
      "epoch": 5.406767794632438,
      "grad_norm": 0.244140625,
      "learning_rate": 8.728327016720595e-05,
      "loss": 0.0082,
      "step": 11584
    },
    {
      "epoch": 5.407234539089848,
      "grad_norm": 0.330078125,
      "learning_rate": 8.726871595128925e-05,
      "loss": 0.0055,
      "step": 11585
    },
    {
      "epoch": 5.407701283547258,
      "grad_norm": 0.0908203125,
      "learning_rate": 8.725416200949126e-05,
      "loss": 0.0008,
      "step": 11586
    },
    {
      "epoch": 5.408168028004668,
      "grad_norm": 0.263671875,
      "learning_rate": 8.723960834212529e-05,
      "loss": 0.0015,
      "step": 11587
    },
    {
      "epoch": 5.408634772462077,
      "grad_norm": 0.134765625,
      "learning_rate": 8.72250549495047e-05,
      "loss": 0.0012,
      "step": 11588
    },
    {
      "epoch": 5.4091015169194865,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.721050183194286e-05,
      "loss": 0.0008,
      "step": 11589
    },
    {
      "epoch": 5.409568261376896,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.719594898975311e-05,
      "loss": 0.0014,
      "step": 11590
    },
    {
      "epoch": 5.410035005834306,
      "grad_norm": 0.1787109375,
      "learning_rate": 8.718139642324876e-05,
      "loss": 0.0008,
      "step": 11591
    },
    {
      "epoch": 5.410501750291715,
      "grad_norm": 0.251953125,
      "learning_rate": 8.71668441327432e-05,
      "loss": 0.0014,
      "step": 11592
    },
    {
      "epoch": 5.410968494749125,
      "grad_norm": 0.322265625,
      "learning_rate": 8.715229211854972e-05,
      "loss": 0.0032,
      "step": 11593
    },
    {
      "epoch": 5.411435239206535,
      "grad_norm": 0.275390625,
      "learning_rate": 8.713774038098164e-05,
      "loss": 0.004,
      "step": 11594
    },
    {
      "epoch": 5.411901983663944,
      "grad_norm": 0.33984375,
      "learning_rate": 8.712318892035226e-05,
      "loss": 0.0021,
      "step": 11595
    },
    {
      "epoch": 5.412368728121353,
      "grad_norm": 0.90234375,
      "learning_rate": 8.710863773697493e-05,
      "loss": 0.0032,
      "step": 11596
    },
    {
      "epoch": 5.412835472578763,
      "grad_norm": 0.54296875,
      "learning_rate": 8.709408683116294e-05,
      "loss": 0.0068,
      "step": 11597
    },
    {
      "epoch": 5.413302217036173,
      "grad_norm": 0.11767578125,
      "learning_rate": 8.707953620322955e-05,
      "loss": 0.0008,
      "step": 11598
    },
    {
      "epoch": 5.413768961493583,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.706498585348811e-05,
      "loss": 0.0013,
      "step": 11599
    },
    {
      "epoch": 5.414235705950992,
      "grad_norm": 0.435546875,
      "learning_rate": 8.705043578225189e-05,
      "loss": 0.0036,
      "step": 11600
    },
    {
      "epoch": 5.414702450408401,
      "grad_norm": 0.2578125,
      "learning_rate": 8.703588598983415e-05,
      "loss": 0.0033,
      "step": 11601
    },
    {
      "epoch": 5.415169194865811,
      "grad_norm": 0.13671875,
      "learning_rate": 8.702133647654815e-05,
      "loss": 0.0011,
      "step": 11602
    },
    {
      "epoch": 5.41563593932322,
      "grad_norm": 0.11279296875,
      "learning_rate": 8.700678724270717e-05,
      "loss": 0.0009,
      "step": 11603
    },
    {
      "epoch": 5.41610268378063,
      "grad_norm": 0.482421875,
      "learning_rate": 8.699223828862448e-05,
      "loss": 0.0056,
      "step": 11604
    },
    {
      "epoch": 5.41656942823804,
      "grad_norm": 0.224609375,
      "learning_rate": 8.697768961461334e-05,
      "loss": 0.0019,
      "step": 11605
    },
    {
      "epoch": 5.4170361726954495,
      "grad_norm": 0.2294921875,
      "learning_rate": 8.6963141220987e-05,
      "loss": 0.0033,
      "step": 11606
    },
    {
      "epoch": 5.417502917152859,
      "grad_norm": 0.04150390625,
      "learning_rate": 8.694859310805867e-05,
      "loss": 0.0004,
      "step": 11607
    },
    {
      "epoch": 5.417969661610268,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.693404527614163e-05,
      "loss": 0.001,
      "step": 11608
    },
    {
      "epoch": 5.418436406067678,
      "grad_norm": 0.10595703125,
      "learning_rate": 8.691949772554908e-05,
      "loss": 0.0009,
      "step": 11609
    },
    {
      "epoch": 5.418903150525088,
      "grad_norm": 0.09521484375,
      "learning_rate": 8.690495045659425e-05,
      "loss": 0.0007,
      "step": 11610
    },
    {
      "epoch": 5.419369894982497,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.689040346959038e-05,
      "loss": 0.0018,
      "step": 11611
    },
    {
      "epoch": 5.4198366394399065,
      "grad_norm": 0.1904296875,
      "learning_rate": 8.687585676485069e-05,
      "loss": 0.0019,
      "step": 11612
    },
    {
      "epoch": 5.420303383897316,
      "grad_norm": 0.46875,
      "learning_rate": 8.686131034268834e-05,
      "loss": 0.0049,
      "step": 11613
    },
    {
      "epoch": 5.420770128354726,
      "grad_norm": 0.2392578125,
      "learning_rate": 8.684676420341653e-05,
      "loss": 0.0012,
      "step": 11614
    },
    {
      "epoch": 5.421236872812135,
      "grad_norm": 0.25,
      "learning_rate": 8.683221834734851e-05,
      "loss": 0.0075,
      "step": 11615
    },
    {
      "epoch": 5.421703617269545,
      "grad_norm": 0.291015625,
      "learning_rate": 8.681767277479745e-05,
      "loss": 0.0034,
      "step": 11616
    },
    {
      "epoch": 5.4221703617269545,
      "grad_norm": 0.19140625,
      "learning_rate": 8.68031274860765e-05,
      "loss": 0.0013,
      "step": 11617
    },
    {
      "epoch": 5.422637106184364,
      "grad_norm": 0.056640625,
      "learning_rate": 8.678858248149888e-05,
      "loss": 0.0008,
      "step": 11618
    },
    {
      "epoch": 5.423103850641773,
      "grad_norm": 0.0615234375,
      "learning_rate": 8.677403776137772e-05,
      "loss": 0.0006,
      "step": 11619
    },
    {
      "epoch": 5.423570595099183,
      "grad_norm": 0.16015625,
      "learning_rate": 8.675949332602619e-05,
      "loss": 0.0014,
      "step": 11620
    },
    {
      "epoch": 5.424037339556593,
      "grad_norm": 0.357421875,
      "learning_rate": 8.674494917575748e-05,
      "loss": 0.0103,
      "step": 11621
    },
    {
      "epoch": 5.424504084014003,
      "grad_norm": 0.1474609375,
      "learning_rate": 8.673040531088473e-05,
      "loss": 0.0012,
      "step": 11622
    },
    {
      "epoch": 5.4249708284714115,
      "grad_norm": 0.140625,
      "learning_rate": 8.671586173172105e-05,
      "loss": 0.001,
      "step": 11623
    },
    {
      "epoch": 5.425437572928821,
      "grad_norm": 0.06005859375,
      "learning_rate": 8.67013184385796e-05,
      "loss": 0.0007,
      "step": 11624
    },
    {
      "epoch": 5.425904317386231,
      "grad_norm": 0.1796875,
      "learning_rate": 8.668677543177355e-05,
      "loss": 0.0013,
      "step": 11625
    },
    {
      "epoch": 5.426371061843641,
      "grad_norm": 0.306640625,
      "learning_rate": 8.667223271161598e-05,
      "loss": 0.0022,
      "step": 11626
    },
    {
      "epoch": 5.42683780630105,
      "grad_norm": 0.28125,
      "learning_rate": 8.665769027842003e-05,
      "loss": 0.0021,
      "step": 11627
    },
    {
      "epoch": 5.42730455075846,
      "grad_norm": 0.466796875,
      "learning_rate": 8.664314813249881e-05,
      "loss": 0.0053,
      "step": 11628
    },
    {
      "epoch": 5.427771295215869,
      "grad_norm": 0.1591796875,
      "learning_rate": 8.662860627416543e-05,
      "loss": 0.0007,
      "step": 11629
    },
    {
      "epoch": 5.428238039673279,
      "grad_norm": 0.400390625,
      "learning_rate": 8.6614064703733e-05,
      "loss": 0.0024,
      "step": 11630
    },
    {
      "epoch": 5.428704784130688,
      "grad_norm": 0.037109375,
      "learning_rate": 8.659952342151459e-05,
      "loss": 0.0005,
      "step": 11631
    },
    {
      "epoch": 5.429171528588098,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.658498242782333e-05,
      "loss": 0.0011,
      "step": 11632
    },
    {
      "epoch": 5.429638273045508,
      "grad_norm": 0.2578125,
      "learning_rate": 8.657044172297228e-05,
      "loss": 0.0009,
      "step": 11633
    },
    {
      "epoch": 5.4301050175029175,
      "grad_norm": 0.0859375,
      "learning_rate": 8.65559013072745e-05,
      "loss": 0.0009,
      "step": 11634
    },
    {
      "epoch": 5.430571761960326,
      "grad_norm": 0.455078125,
      "learning_rate": 8.65413611810431e-05,
      "loss": 0.0082,
      "step": 11635
    },
    {
      "epoch": 5.431038506417736,
      "grad_norm": 0.279296875,
      "learning_rate": 8.652682134459114e-05,
      "loss": 0.0016,
      "step": 11636
    },
    {
      "epoch": 5.431505250875146,
      "grad_norm": 0.2109375,
      "learning_rate": 8.651228179823163e-05,
      "loss": 0.0048,
      "step": 11637
    },
    {
      "epoch": 5.431971995332556,
      "grad_norm": 0.2353515625,
      "learning_rate": 8.649774254227767e-05,
      "loss": 0.0032,
      "step": 11638
    },
    {
      "epoch": 5.432438739789965,
      "grad_norm": 0.3359375,
      "learning_rate": 8.648320357704231e-05,
      "loss": 0.0022,
      "step": 11639
    },
    {
      "epoch": 5.4329054842473745,
      "grad_norm": 0.458984375,
      "learning_rate": 8.646866490283854e-05,
      "loss": 0.0022,
      "step": 11640
    },
    {
      "epoch": 5.433372228704784,
      "grad_norm": 0.08154296875,
      "learning_rate": 8.645412651997947e-05,
      "loss": 0.0009,
      "step": 11641
    },
    {
      "epoch": 5.433838973162194,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.643958842877808e-05,
      "loss": 0.001,
      "step": 11642
    },
    {
      "epoch": 5.434305717619603,
      "grad_norm": 0.220703125,
      "learning_rate": 8.642505062954738e-05,
      "loss": 0.0016,
      "step": 11643
    },
    {
      "epoch": 5.434772462077013,
      "grad_norm": 0.1669921875,
      "learning_rate": 8.64105131226004e-05,
      "loss": 0.0041,
      "step": 11644
    },
    {
      "epoch": 5.435239206534423,
      "grad_norm": 0.279296875,
      "learning_rate": 8.639597590825014e-05,
      "loss": 0.0035,
      "step": 11645
    },
    {
      "epoch": 5.4357059509918315,
      "grad_norm": 0.2255859375,
      "learning_rate": 8.638143898680964e-05,
      "loss": 0.0018,
      "step": 11646
    },
    {
      "epoch": 5.436172695449241,
      "grad_norm": 0.0634765625,
      "learning_rate": 8.636690235859189e-05,
      "loss": 0.0005,
      "step": 11647
    },
    {
      "epoch": 5.436639439906651,
      "grad_norm": 0.142578125,
      "learning_rate": 8.635236602390983e-05,
      "loss": 0.0011,
      "step": 11648
    },
    {
      "epoch": 5.437106184364061,
      "grad_norm": 0.333984375,
      "learning_rate": 8.633782998307646e-05,
      "loss": 0.0018,
      "step": 11649
    },
    {
      "epoch": 5.437572928821471,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.632329423640478e-05,
      "loss": 0.0016,
      "step": 11650
    },
    {
      "epoch": 5.43803967327888,
      "grad_norm": 0.33203125,
      "learning_rate": 8.630875878420775e-05,
      "loss": 0.0027,
      "step": 11651
    },
    {
      "epoch": 5.438506417736289,
      "grad_norm": 0.27734375,
      "learning_rate": 8.629422362679831e-05,
      "loss": 0.011,
      "step": 11652
    },
    {
      "epoch": 5.438973162193699,
      "grad_norm": 0.25390625,
      "learning_rate": 8.627968876448947e-05,
      "loss": 0.0036,
      "step": 11653
    },
    {
      "epoch": 5.439439906651108,
      "grad_norm": 0.1474609375,
      "learning_rate": 8.626515419759416e-05,
      "loss": 0.0009,
      "step": 11654
    },
    {
      "epoch": 5.439906651108518,
      "grad_norm": 0.4921875,
      "learning_rate": 8.625061992642531e-05,
      "loss": 0.0028,
      "step": 11655
    },
    {
      "epoch": 5.440373395565928,
      "grad_norm": 0.34375,
      "learning_rate": 8.623608595129584e-05,
      "loss": 0.0027,
      "step": 11656
    },
    {
      "epoch": 5.4408401400233375,
      "grad_norm": 0.16015625,
      "learning_rate": 8.622155227251875e-05,
      "loss": 0.0026,
      "step": 11657
    },
    {
      "epoch": 5.441306884480746,
      "grad_norm": 0.1689453125,
      "learning_rate": 8.620701889040691e-05,
      "loss": 0.0012,
      "step": 11658
    },
    {
      "epoch": 5.441773628938156,
      "grad_norm": 0.271484375,
      "learning_rate": 8.619248580527324e-05,
      "loss": 0.0026,
      "step": 11659
    },
    {
      "epoch": 5.442240373395566,
      "grad_norm": 0.26953125,
      "learning_rate": 8.61779530174307e-05,
      "loss": 0.0032,
      "step": 11660
    },
    {
      "epoch": 5.442707117852976,
      "grad_norm": 0.11083984375,
      "learning_rate": 8.616342052719217e-05,
      "loss": 0.0011,
      "step": 11661
    },
    {
      "epoch": 5.443173862310385,
      "grad_norm": 0.09228515625,
      "learning_rate": 8.614888833487051e-05,
      "loss": 0.0007,
      "step": 11662
    },
    {
      "epoch": 5.4436406067677945,
      "grad_norm": 0.396484375,
      "learning_rate": 8.613435644077868e-05,
      "loss": 0.0038,
      "step": 11663
    },
    {
      "epoch": 5.444107351225204,
      "grad_norm": 0.10302734375,
      "learning_rate": 8.611982484522953e-05,
      "loss": 0.0028,
      "step": 11664
    },
    {
      "epoch": 5.444574095682614,
      "grad_norm": 0.16015625,
      "learning_rate": 8.610529354853599e-05,
      "loss": 0.0015,
      "step": 11665
    },
    {
      "epoch": 5.445040840140023,
      "grad_norm": 0.283203125,
      "learning_rate": 8.609076255101083e-05,
      "loss": 0.0015,
      "step": 11666
    },
    {
      "epoch": 5.445507584597433,
      "grad_norm": 0.216796875,
      "learning_rate": 8.607623185296702e-05,
      "loss": 0.0027,
      "step": 11667
    },
    {
      "epoch": 5.445974329054843,
      "grad_norm": 0.12060546875,
      "learning_rate": 8.606170145471739e-05,
      "loss": 0.001,
      "step": 11668
    },
    {
      "epoch": 5.446441073512252,
      "grad_norm": 0.169921875,
      "learning_rate": 8.604717135657479e-05,
      "loss": 0.0007,
      "step": 11669
    },
    {
      "epoch": 5.446907817969661,
      "grad_norm": 0.302734375,
      "learning_rate": 8.603264155885207e-05,
      "loss": 0.0017,
      "step": 11670
    },
    {
      "epoch": 5.447374562427071,
      "grad_norm": 0.361328125,
      "learning_rate": 8.601811206186207e-05,
      "loss": 0.0032,
      "step": 11671
    },
    {
      "epoch": 5.447841306884481,
      "grad_norm": 0.30859375,
      "learning_rate": 8.600358286591764e-05,
      "loss": 0.0031,
      "step": 11672
    },
    {
      "epoch": 5.448308051341891,
      "grad_norm": 0.048828125,
      "learning_rate": 8.598905397133159e-05,
      "loss": 0.0019,
      "step": 11673
    },
    {
      "epoch": 5.4487747957993,
      "grad_norm": 0.2890625,
      "learning_rate": 8.597452537841676e-05,
      "loss": 0.0024,
      "step": 11674
    },
    {
      "epoch": 5.449241540256709,
      "grad_norm": 0.1923828125,
      "learning_rate": 8.595999708748596e-05,
      "loss": 0.0014,
      "step": 11675
    },
    {
      "epoch": 5.449708284714119,
      "grad_norm": 0.154296875,
      "learning_rate": 8.594546909885198e-05,
      "loss": 0.0007,
      "step": 11676
    },
    {
      "epoch": 5.450175029171529,
      "grad_norm": 0.41796875,
      "learning_rate": 8.593094141282767e-05,
      "loss": 0.0059,
      "step": 11677
    },
    {
      "epoch": 5.450641773628938,
      "grad_norm": 0.044677734375,
      "learning_rate": 8.59164140297258e-05,
      "loss": 0.0006,
      "step": 11678
    },
    {
      "epoch": 5.451108518086348,
      "grad_norm": 0.2177734375,
      "learning_rate": 8.590188694985914e-05,
      "loss": 0.0016,
      "step": 11679
    },
    {
      "epoch": 5.4515752625437575,
      "grad_norm": 0.1943359375,
      "learning_rate": 8.588736017354051e-05,
      "loss": 0.0013,
      "step": 11680
    },
    {
      "epoch": 5.452042007001167,
      "grad_norm": 0.26171875,
      "learning_rate": 8.587283370108269e-05,
      "loss": 0.0011,
      "step": 11681
    },
    {
      "epoch": 5.452508751458576,
      "grad_norm": 0.07861328125,
      "learning_rate": 8.585830753279844e-05,
      "loss": 0.0009,
      "step": 11682
    },
    {
      "epoch": 5.452975495915986,
      "grad_norm": 0.07958984375,
      "learning_rate": 8.584378166900047e-05,
      "loss": 0.0006,
      "step": 11683
    },
    {
      "epoch": 5.453442240373396,
      "grad_norm": 0.3046875,
      "learning_rate": 8.582925611000164e-05,
      "loss": 0.0032,
      "step": 11684
    },
    {
      "epoch": 5.4539089848308056,
      "grad_norm": 0.0595703125,
      "learning_rate": 8.581473085611462e-05,
      "loss": 0.0005,
      "step": 11685
    },
    {
      "epoch": 5.4543757292882145,
      "grad_norm": 0.150390625,
      "learning_rate": 8.580020590765216e-05,
      "loss": 0.0008,
      "step": 11686
    },
    {
      "epoch": 5.454842473745624,
      "grad_norm": 0.1962890625,
      "learning_rate": 8.578568126492706e-05,
      "loss": 0.0053,
      "step": 11687
    },
    {
      "epoch": 5.455309218203034,
      "grad_norm": 0.28125,
      "learning_rate": 8.577115692825201e-05,
      "loss": 0.002,
      "step": 11688
    },
    {
      "epoch": 5.455775962660443,
      "grad_norm": 0.169921875,
      "learning_rate": 8.575663289793973e-05,
      "loss": 0.0012,
      "step": 11689
    },
    {
      "epoch": 5.456242707117853,
      "grad_norm": 0.1953125,
      "learning_rate": 8.574210917430296e-05,
      "loss": 0.0037,
      "step": 11690
    },
    {
      "epoch": 5.4567094515752625,
      "grad_norm": 0.0703125,
      "learning_rate": 8.572758575765438e-05,
      "loss": 0.0006,
      "step": 11691
    },
    {
      "epoch": 5.457176196032672,
      "grad_norm": 0.1962890625,
      "learning_rate": 8.571306264830673e-05,
      "loss": 0.0014,
      "step": 11692
    },
    {
      "epoch": 5.457642940490082,
      "grad_norm": 0.1943359375,
      "learning_rate": 8.569853984657268e-05,
      "loss": 0.0014,
      "step": 11693
    },
    {
      "epoch": 5.458109684947491,
      "grad_norm": 0.1435546875,
      "learning_rate": 8.568401735276495e-05,
      "loss": 0.0013,
      "step": 11694
    },
    {
      "epoch": 5.458576429404901,
      "grad_norm": 0.0830078125,
      "learning_rate": 8.566949516719622e-05,
      "loss": 0.0009,
      "step": 11695
    },
    {
      "epoch": 5.459043173862311,
      "grad_norm": 0.08740234375,
      "learning_rate": 8.565497329017914e-05,
      "loss": 0.0009,
      "step": 11696
    },
    {
      "epoch": 5.4595099183197195,
      "grad_norm": 0.3359375,
      "learning_rate": 8.564045172202642e-05,
      "loss": 0.0018,
      "step": 11697
    },
    {
      "epoch": 5.459976662777129,
      "grad_norm": 0.3046875,
      "learning_rate": 8.562593046305066e-05,
      "loss": 0.003,
      "step": 11698
    },
    {
      "epoch": 5.460443407234539,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.56114095135646e-05,
      "loss": 0.0067,
      "step": 11699
    },
    {
      "epoch": 5.460910151691949,
      "grad_norm": 0.10400390625,
      "learning_rate": 8.559688887388086e-05,
      "loss": 0.0007,
      "step": 11700
    },
    {
      "epoch": 5.461376896149358,
      "grad_norm": 0.28515625,
      "learning_rate": 8.558236854431205e-05,
      "loss": 0.0025,
      "step": 11701
    },
    {
      "epoch": 5.461843640606768,
      "grad_norm": 0.03125,
      "learning_rate": 8.556784852517088e-05,
      "loss": 0.0004,
      "step": 11702
    },
    {
      "epoch": 5.462310385064177,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.555332881676993e-05,
      "loss": 0.0038,
      "step": 11703
    },
    {
      "epoch": 5.462777129521587,
      "grad_norm": 0.248046875,
      "learning_rate": 8.553880941942185e-05,
      "loss": 0.0015,
      "step": 11704
    },
    {
      "epoch": 5.463243873978996,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.552429033343922e-05,
      "loss": 0.0008,
      "step": 11705
    },
    {
      "epoch": 5.463710618436406,
      "grad_norm": 0.1552734375,
      "learning_rate": 8.55097715591347e-05,
      "loss": 0.0011,
      "step": 11706
    },
    {
      "epoch": 5.464177362893816,
      "grad_norm": 0.2041015625,
      "learning_rate": 8.549525309682088e-05,
      "loss": 0.0087,
      "step": 11707
    },
    {
      "epoch": 5.4646441073512255,
      "grad_norm": 0.2578125,
      "learning_rate": 8.548073494681032e-05,
      "loss": 0.0018,
      "step": 11708
    },
    {
      "epoch": 5.465110851808634,
      "grad_norm": 0.236328125,
      "learning_rate": 8.546621710941569e-05,
      "loss": 0.0017,
      "step": 11709
    },
    {
      "epoch": 5.465577596266044,
      "grad_norm": 0.08740234375,
      "learning_rate": 8.545169958494952e-05,
      "loss": 0.0008,
      "step": 11710
    },
    {
      "epoch": 5.466044340723454,
      "grad_norm": 0.2392578125,
      "learning_rate": 8.543718237372437e-05,
      "loss": 0.003,
      "step": 11711
    },
    {
      "epoch": 5.466511085180864,
      "grad_norm": 0.474609375,
      "learning_rate": 8.542266547605288e-05,
      "loss": 0.0046,
      "step": 11712
    },
    {
      "epoch": 5.466977829638273,
      "grad_norm": 0.17578125,
      "learning_rate": 8.540814889224756e-05,
      "loss": 0.001,
      "step": 11713
    },
    {
      "epoch": 5.4674445740956825,
      "grad_norm": 0.050537109375,
      "learning_rate": 8.539363262262098e-05,
      "loss": 0.0023,
      "step": 11714
    },
    {
      "epoch": 5.467911318553092,
      "grad_norm": 0.10791015625,
      "learning_rate": 8.53791166674857e-05,
      "loss": 0.0008,
      "step": 11715
    },
    {
      "epoch": 5.468378063010502,
      "grad_norm": 0.1455078125,
      "learning_rate": 8.536460102715427e-05,
      "loss": 0.0026,
      "step": 11716
    },
    {
      "epoch": 5.468844807467911,
      "grad_norm": 0.212890625,
      "learning_rate": 8.535008570193921e-05,
      "loss": 0.0048,
      "step": 11717
    },
    {
      "epoch": 5.469311551925321,
      "grad_norm": 0.31640625,
      "learning_rate": 8.533557069215305e-05,
      "loss": 0.005,
      "step": 11718
    },
    {
      "epoch": 5.469778296382731,
      "grad_norm": 0.10693359375,
      "learning_rate": 8.532105599810834e-05,
      "loss": 0.0007,
      "step": 11719
    },
    {
      "epoch": 5.47024504084014,
      "grad_norm": 0.103515625,
      "learning_rate": 8.530654162011757e-05,
      "loss": 0.0009,
      "step": 11720
    },
    {
      "epoch": 5.470711785297549,
      "grad_norm": 0.1943359375,
      "learning_rate": 8.529202755849327e-05,
      "loss": 0.0015,
      "step": 11721
    },
    {
      "epoch": 5.471178529754959,
      "grad_norm": 0.0257568359375,
      "learning_rate": 8.527751381354792e-05,
      "loss": 0.0005,
      "step": 11722
    },
    {
      "epoch": 5.471645274212369,
      "grad_norm": 0.12255859375,
      "learning_rate": 8.526300038559405e-05,
      "loss": 0.0009,
      "step": 11723
    },
    {
      "epoch": 5.472112018669779,
      "grad_norm": 0.259765625,
      "learning_rate": 8.524848727494412e-05,
      "loss": 0.0015,
      "step": 11724
    },
    {
      "epoch": 5.472578763127188,
      "grad_norm": 0.1484375,
      "learning_rate": 8.523397448191061e-05,
      "loss": 0.0008,
      "step": 11725
    },
    {
      "epoch": 5.473045507584597,
      "grad_norm": 0.388671875,
      "learning_rate": 8.521946200680605e-05,
      "loss": 0.0077,
      "step": 11726
    },
    {
      "epoch": 5.473512252042007,
      "grad_norm": 0.037353515625,
      "learning_rate": 8.520494984994286e-05,
      "loss": 0.0005,
      "step": 11727
    },
    {
      "epoch": 5.473978996499417,
      "grad_norm": 0.02392578125,
      "learning_rate": 8.51904380116335e-05,
      "loss": 0.0005,
      "step": 11728
    },
    {
      "epoch": 5.474445740956826,
      "grad_norm": 0.0908203125,
      "learning_rate": 8.517592649219046e-05,
      "loss": 0.0006,
      "step": 11729
    },
    {
      "epoch": 5.474912485414236,
      "grad_norm": 0.12158203125,
      "learning_rate": 8.516141529192617e-05,
      "loss": 0.0036,
      "step": 11730
    },
    {
      "epoch": 5.4753792298716455,
      "grad_norm": 0.046630859375,
      "learning_rate": 8.514690441115307e-05,
      "loss": 0.0005,
      "step": 11731
    },
    {
      "epoch": 5.475845974329054,
      "grad_norm": 0.1044921875,
      "learning_rate": 8.513239385018357e-05,
      "loss": 0.0005,
      "step": 11732
    },
    {
      "epoch": 5.476312718786464,
      "grad_norm": 0.193359375,
      "learning_rate": 8.511788360933016e-05,
      "loss": 0.0058,
      "step": 11733
    },
    {
      "epoch": 5.476779463243874,
      "grad_norm": 0.11376953125,
      "learning_rate": 8.510337368890522e-05,
      "loss": 0.0009,
      "step": 11734
    },
    {
      "epoch": 5.477246207701284,
      "grad_norm": 0.043212890625,
      "learning_rate": 8.508886408922115e-05,
      "loss": 0.0005,
      "step": 11735
    },
    {
      "epoch": 5.477712952158694,
      "grad_norm": 0.0869140625,
      "learning_rate": 8.50743548105904e-05,
      "loss": 0.0008,
      "step": 11736
    },
    {
      "epoch": 5.4781796966161025,
      "grad_norm": 0.193359375,
      "learning_rate": 8.505984585332537e-05,
      "loss": 0.0018,
      "step": 11737
    },
    {
      "epoch": 5.478646441073512,
      "grad_norm": 0.07666015625,
      "learning_rate": 8.50453372177384e-05,
      "loss": 0.0006,
      "step": 11738
    },
    {
      "epoch": 5.479113185530922,
      "grad_norm": 0.056640625,
      "learning_rate": 8.503082890414193e-05,
      "loss": 0.0006,
      "step": 11739
    },
    {
      "epoch": 5.479579929988331,
      "grad_norm": 0.248046875,
      "learning_rate": 8.501632091284832e-05,
      "loss": 0.0012,
      "step": 11740
    },
    {
      "epoch": 5.480046674445741,
      "grad_norm": 0.06689453125,
      "learning_rate": 8.500181324416996e-05,
      "loss": 0.0008,
      "step": 11741
    },
    {
      "epoch": 5.480513418903151,
      "grad_norm": 0.038330078125,
      "learning_rate": 8.498730589841916e-05,
      "loss": 0.0004,
      "step": 11742
    },
    {
      "epoch": 5.48098016336056,
      "grad_norm": 0.21484375,
      "learning_rate": 8.49727988759084e-05,
      "loss": 0.0007,
      "step": 11743
    },
    {
      "epoch": 5.481446907817969,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.495829217694989e-05,
      "loss": 0.0042,
      "step": 11744
    },
    {
      "epoch": 5.481913652275379,
      "grad_norm": 0.2353515625,
      "learning_rate": 8.494378580185607e-05,
      "loss": 0.0041,
      "step": 11745
    },
    {
      "epoch": 5.482380396732789,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.492927975093922e-05,
      "loss": 0.001,
      "step": 11746
    },
    {
      "epoch": 5.482847141190199,
      "grad_norm": 0.58984375,
      "learning_rate": 8.491477402451169e-05,
      "loss": 0.0023,
      "step": 11747
    },
    {
      "epoch": 5.483313885647608,
      "grad_norm": 0.16796875,
      "learning_rate": 8.490026862288583e-05,
      "loss": 0.0018,
      "step": 11748
    },
    {
      "epoch": 5.483780630105017,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.488576354637395e-05,
      "loss": 0.0008,
      "step": 11749
    },
    {
      "epoch": 5.484247374562427,
      "grad_norm": 0.140625,
      "learning_rate": 8.487125879528832e-05,
      "loss": 0.0036,
      "step": 11750
    },
    {
      "epoch": 5.484714119019837,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.48567543699413e-05,
      "loss": 0.0008,
      "step": 11751
    },
    {
      "epoch": 5.485180863477246,
      "grad_norm": 0.0289306640625,
      "learning_rate": 8.484225027064516e-05,
      "loss": 0.0004,
      "step": 11752
    },
    {
      "epoch": 5.485647607934656,
      "grad_norm": 0.51171875,
      "learning_rate": 8.482774649771219e-05,
      "loss": 0.002,
      "step": 11753
    },
    {
      "epoch": 5.4861143523920655,
      "grad_norm": 0.09814453125,
      "learning_rate": 8.481324305145464e-05,
      "loss": 0.0034,
      "step": 11754
    },
    {
      "epoch": 5.486581096849475,
      "grad_norm": 0.08447265625,
      "learning_rate": 8.479873993218486e-05,
      "loss": 0.0008,
      "step": 11755
    },
    {
      "epoch": 5.487047841306884,
      "grad_norm": 0.166015625,
      "learning_rate": 8.478423714021506e-05,
      "loss": 0.001,
      "step": 11756
    },
    {
      "epoch": 5.487514585764294,
      "grad_norm": 0.059814453125,
      "learning_rate": 8.476973467585749e-05,
      "loss": 0.0007,
      "step": 11757
    },
    {
      "epoch": 5.487981330221704,
      "grad_norm": 0.376953125,
      "learning_rate": 8.475523253942445e-05,
      "loss": 0.0034,
      "step": 11758
    },
    {
      "epoch": 5.4884480746791136,
      "grad_norm": 0.10498046875,
      "learning_rate": 8.474073073122819e-05,
      "loss": 0.0011,
      "step": 11759
    },
    {
      "epoch": 5.4889148191365225,
      "grad_norm": 0.05029296875,
      "learning_rate": 8.47262292515809e-05,
      "loss": 0.0004,
      "step": 11760
    },
    {
      "epoch": 5.489381563593932,
      "grad_norm": 0.046875,
      "learning_rate": 8.471172810079485e-05,
      "loss": 0.0005,
      "step": 11761
    },
    {
      "epoch": 5.489848308051342,
      "grad_norm": 0.1416015625,
      "learning_rate": 8.469722727918226e-05,
      "loss": 0.0017,
      "step": 11762
    },
    {
      "epoch": 5.490315052508752,
      "grad_norm": 0.46484375,
      "learning_rate": 8.468272678705534e-05,
      "loss": 0.002,
      "step": 11763
    },
    {
      "epoch": 5.490781796966161,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.46682266247263e-05,
      "loss": 0.0016,
      "step": 11764
    },
    {
      "epoch": 5.4912485414235706,
      "grad_norm": 0.10498046875,
      "learning_rate": 8.465372679250737e-05,
      "loss": 0.0008,
      "step": 11765
    },
    {
      "epoch": 5.49171528588098,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.463922729071072e-05,
      "loss": 0.0049,
      "step": 11766
    },
    {
      "epoch": 5.49218203033839,
      "grad_norm": 0.287109375,
      "learning_rate": 8.462472811964854e-05,
      "loss": 0.0034,
      "step": 11767
    },
    {
      "epoch": 5.492648774795799,
      "grad_norm": 0.201171875,
      "learning_rate": 8.461022927963303e-05,
      "loss": 0.002,
      "step": 11768
    },
    {
      "epoch": 5.493115519253209,
      "grad_norm": 0.2890625,
      "learning_rate": 8.459573077097636e-05,
      "loss": 0.0018,
      "step": 11769
    },
    {
      "epoch": 5.493582263710619,
      "grad_norm": 0.053955078125,
      "learning_rate": 8.45812325939907e-05,
      "loss": 0.0006,
      "step": 11770
    },
    {
      "epoch": 5.494049008168028,
      "grad_norm": 0.26953125,
      "learning_rate": 8.45667347489882e-05,
      "loss": 0.001,
      "step": 11771
    },
    {
      "epoch": 5.494515752625437,
      "grad_norm": 0.10205078125,
      "learning_rate": 8.455223723628105e-05,
      "loss": 0.0009,
      "step": 11772
    },
    {
      "epoch": 5.494982497082847,
      "grad_norm": 0.283203125,
      "learning_rate": 8.453774005618136e-05,
      "loss": 0.0064,
      "step": 11773
    },
    {
      "epoch": 5.495449241540257,
      "grad_norm": 0.1650390625,
      "learning_rate": 8.452324320900125e-05,
      "loss": 0.0062,
      "step": 11774
    },
    {
      "epoch": 5.495915985997666,
      "grad_norm": 0.04541015625,
      "learning_rate": 8.450874669505291e-05,
      "loss": 0.0004,
      "step": 11775
    },
    {
      "epoch": 5.496382730455076,
      "grad_norm": 0.11669921875,
      "learning_rate": 8.449425051464846e-05,
      "loss": 0.0007,
      "step": 11776
    },
    {
      "epoch": 5.496849474912485,
      "grad_norm": 0.1748046875,
      "learning_rate": 8.447975466809997e-05,
      "loss": 0.0017,
      "step": 11777
    },
    {
      "epoch": 5.497316219369895,
      "grad_norm": 0.0478515625,
      "learning_rate": 8.44652591557196e-05,
      "loss": 0.0027,
      "step": 11778
    },
    {
      "epoch": 5.497782963827304,
      "grad_norm": 0.189453125,
      "learning_rate": 8.445076397781944e-05,
      "loss": 0.0025,
      "step": 11779
    },
    {
      "epoch": 5.498249708284714,
      "grad_norm": 0.28125,
      "learning_rate": 8.443626913471159e-05,
      "loss": 0.0025,
      "step": 11780
    },
    {
      "epoch": 5.498716452742124,
      "grad_norm": 0.20703125,
      "learning_rate": 8.44217746267081e-05,
      "loss": 0.0104,
      "step": 11781
    },
    {
      "epoch": 5.4991831971995335,
      "grad_norm": 0.3359375,
      "learning_rate": 8.44072804541211e-05,
      "loss": 0.0013,
      "step": 11782
    },
    {
      "epoch": 5.499649941656942,
      "grad_norm": 0.09228515625,
      "learning_rate": 8.439278661726267e-05,
      "loss": 0.0024,
      "step": 11783
    },
    {
      "epoch": 5.500116686114352,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.437829311644484e-05,
      "loss": 0.0023,
      "step": 11784
    },
    {
      "epoch": 5.500583430571762,
      "grad_norm": 0.1474609375,
      "learning_rate": 8.43637999519797e-05,
      "loss": 0.001,
      "step": 11785
    },
    {
      "epoch": 5.501050175029172,
      "grad_norm": 0.0517578125,
      "learning_rate": 8.434930712417931e-05,
      "loss": 0.0006,
      "step": 11786
    },
    {
      "epoch": 5.501516919486581,
      "grad_norm": 0.041748046875,
      "learning_rate": 8.433481463335567e-05,
      "loss": 0.0006,
      "step": 11787
    },
    {
      "epoch": 5.5019836639439905,
      "grad_norm": 0.064453125,
      "learning_rate": 8.432032247982087e-05,
      "loss": 0.0006,
      "step": 11788
    },
    {
      "epoch": 5.5024504084014,
      "grad_norm": 0.21875,
      "learning_rate": 8.430583066388695e-05,
      "loss": 0.0024,
      "step": 11789
    },
    {
      "epoch": 5.50291715285881,
      "grad_norm": 1.1015625,
      "learning_rate": 8.429133918586589e-05,
      "loss": 0.0047,
      "step": 11790
    },
    {
      "epoch": 5.503383897316219,
      "grad_norm": 0.333984375,
      "learning_rate": 8.427684804606975e-05,
      "loss": 0.0025,
      "step": 11791
    },
    {
      "epoch": 5.503850641773629,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.426235724481049e-05,
      "loss": 0.0053,
      "step": 11792
    },
    {
      "epoch": 5.503850641773629,
      "eval_loss": 2.22942852973938,
      "eval_runtime": 57.9891,
      "eval_samples_per_second": 31.109,
      "eval_steps_per_second": 3.897,
      "step": 11792
    },
    {
      "epoch": 5.504317386231039,
      "grad_norm": 0.058837890625,
      "learning_rate": 8.424786678240011e-05,
      "loss": 0.0006,
      "step": 11793
    },
    {
      "epoch": 5.504784130688448,
      "grad_norm": 0.328125,
      "learning_rate": 8.423337665915068e-05,
      "loss": 0.0018,
      "step": 11794
    },
    {
      "epoch": 5.505250875145857,
      "grad_norm": 0.3671875,
      "learning_rate": 8.421888687537412e-05,
      "loss": 0.0015,
      "step": 11795
    },
    {
      "epoch": 5.505717619603267,
      "grad_norm": 0.03369140625,
      "learning_rate": 8.420439743138242e-05,
      "loss": 0.0005,
      "step": 11796
    },
    {
      "epoch": 5.506184364060677,
      "grad_norm": 0.072265625,
      "learning_rate": 8.41899083274876e-05,
      "loss": 0.0008,
      "step": 11797
    },
    {
      "epoch": 5.506651108518087,
      "grad_norm": 0.1484375,
      "learning_rate": 8.417541956400158e-05,
      "loss": 0.0013,
      "step": 11798
    },
    {
      "epoch": 5.507117852975496,
      "grad_norm": 0.07861328125,
      "learning_rate": 8.416093114123633e-05,
      "loss": 0.0006,
      "step": 11799
    },
    {
      "epoch": 5.507584597432905,
      "grad_norm": 0.169921875,
      "learning_rate": 8.414644305950383e-05,
      "loss": 0.0078,
      "step": 11800
    },
    {
      "epoch": 5.508051341890315,
      "grad_norm": 0.16015625,
      "learning_rate": 8.4131955319116e-05,
      "loss": 0.0009,
      "step": 11801
    },
    {
      "epoch": 5.508518086347725,
      "grad_norm": 0.2470703125,
      "learning_rate": 8.411746792038477e-05,
      "loss": 0.0015,
      "step": 11802
    },
    {
      "epoch": 5.508984830805134,
      "grad_norm": 0.087890625,
      "learning_rate": 8.410298086362204e-05,
      "loss": 0.0007,
      "step": 11803
    },
    {
      "epoch": 5.509451575262544,
      "grad_norm": 0.2265625,
      "learning_rate": 8.408849414913982e-05,
      "loss": 0.0011,
      "step": 11804
    },
    {
      "epoch": 5.5099183197199535,
      "grad_norm": 0.2041015625,
      "learning_rate": 8.407400777724996e-05,
      "loss": 0.0013,
      "step": 11805
    },
    {
      "epoch": 5.510385064177363,
      "grad_norm": 0.11376953125,
      "learning_rate": 8.405952174826437e-05,
      "loss": 0.0008,
      "step": 11806
    },
    {
      "epoch": 5.510851808634772,
      "grad_norm": 0.31640625,
      "learning_rate": 8.404503606249498e-05,
      "loss": 0.0019,
      "step": 11807
    },
    {
      "epoch": 5.511318553092182,
      "grad_norm": 0.115234375,
      "learning_rate": 8.403055072025365e-05,
      "loss": 0.0007,
      "step": 11808
    },
    {
      "epoch": 5.511785297549592,
      "grad_norm": 0.302734375,
      "learning_rate": 8.401606572185228e-05,
      "loss": 0.0016,
      "step": 11809
    },
    {
      "epoch": 5.512252042007001,
      "grad_norm": 0.01953125,
      "learning_rate": 8.400158106760276e-05,
      "loss": 0.0003,
      "step": 11810
    },
    {
      "epoch": 5.5127187864644105,
      "grad_norm": 0.1455078125,
      "learning_rate": 8.398709675781696e-05,
      "loss": 0.0013,
      "step": 11811
    },
    {
      "epoch": 5.51318553092182,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.397261279280673e-05,
      "loss": 0.0061,
      "step": 11812
    },
    {
      "epoch": 5.51365227537923,
      "grad_norm": 0.103515625,
      "learning_rate": 8.395812917288391e-05,
      "loss": 0.0018,
      "step": 11813
    },
    {
      "epoch": 5.51411901983664,
      "grad_norm": 0.12890625,
      "learning_rate": 8.39436458983604e-05,
      "loss": 0.001,
      "step": 11814
    },
    {
      "epoch": 5.514585764294049,
      "grad_norm": 0.0810546875,
      "learning_rate": 8.392916296954797e-05,
      "loss": 0.0008,
      "step": 11815
    },
    {
      "epoch": 5.515052508751459,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.39146803867585e-05,
      "loss": 0.0011,
      "step": 11816
    },
    {
      "epoch": 5.515519253208868,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.390019815030383e-05,
      "loss": 0.0008,
      "step": 11817
    },
    {
      "epoch": 5.515985997666277,
      "grad_norm": 0.1494140625,
      "learning_rate": 8.388571626049575e-05,
      "loss": 0.0031,
      "step": 11818
    },
    {
      "epoch": 5.516452742123687,
      "grad_norm": 0.322265625,
      "learning_rate": 8.387123471764606e-05,
      "loss": 0.0041,
      "step": 11819
    },
    {
      "epoch": 5.516919486581097,
      "grad_norm": 0.2119140625,
      "learning_rate": 8.385675352206661e-05,
      "loss": 0.0011,
      "step": 11820
    },
    {
      "epoch": 5.517386231038507,
      "grad_norm": 0.384765625,
      "learning_rate": 8.384227267406916e-05,
      "loss": 0.0067,
      "step": 11821
    },
    {
      "epoch": 5.5178529754959165,
      "grad_norm": 0.259765625,
      "learning_rate": 8.38277921739655e-05,
      "loss": 0.0011,
      "step": 11822
    },
    {
      "epoch": 5.518319719953325,
      "grad_norm": 0.1123046875,
      "learning_rate": 8.381331202206741e-05,
      "loss": 0.0008,
      "step": 11823
    },
    {
      "epoch": 5.518786464410735,
      "grad_norm": 0.28125,
      "learning_rate": 8.379883221868668e-05,
      "loss": 0.002,
      "step": 11824
    },
    {
      "epoch": 5.519253208868145,
      "grad_norm": 0.13671875,
      "learning_rate": 8.378435276413508e-05,
      "loss": 0.0007,
      "step": 11825
    },
    {
      "epoch": 5.519719953325554,
      "grad_norm": 0.035400390625,
      "learning_rate": 8.376987365872432e-05,
      "loss": 0.0003,
      "step": 11826
    },
    {
      "epoch": 5.520186697782964,
      "grad_norm": 0.043701171875,
      "learning_rate": 8.375539490276622e-05,
      "loss": 0.0004,
      "step": 11827
    },
    {
      "epoch": 5.5206534422403735,
      "grad_norm": 0.71484375,
      "learning_rate": 8.37409164965725e-05,
      "loss": 0.0022,
      "step": 11828
    },
    {
      "epoch": 5.521120186697783,
      "grad_norm": 0.3515625,
      "learning_rate": 8.372643844045484e-05,
      "loss": 0.0063,
      "step": 11829
    },
    {
      "epoch": 5.521586931155192,
      "grad_norm": 0.1708984375,
      "learning_rate": 8.371196073472506e-05,
      "loss": 0.0011,
      "step": 11830
    },
    {
      "epoch": 5.522053675612602,
      "grad_norm": 0.69921875,
      "learning_rate": 8.369748337969483e-05,
      "loss": 0.0081,
      "step": 11831
    },
    {
      "epoch": 5.522520420070012,
      "grad_norm": 0.12890625,
      "learning_rate": 8.368300637567587e-05,
      "loss": 0.0008,
      "step": 11832
    },
    {
      "epoch": 5.5229871645274216,
      "grad_norm": 0.08642578125,
      "learning_rate": 8.366852972297988e-05,
      "loss": 0.0046,
      "step": 11833
    },
    {
      "epoch": 5.5234539089848305,
      "grad_norm": 0.31640625,
      "learning_rate": 8.365405342191856e-05,
      "loss": 0.0022,
      "step": 11834
    },
    {
      "epoch": 5.52392065344224,
      "grad_norm": 0.1904296875,
      "learning_rate": 8.363957747280361e-05,
      "loss": 0.0012,
      "step": 11835
    },
    {
      "epoch": 5.52438739789965,
      "grad_norm": 0.06689453125,
      "learning_rate": 8.362510187594671e-05,
      "loss": 0.0006,
      "step": 11836
    },
    {
      "epoch": 5.52485414235706,
      "grad_norm": 0.0546875,
      "learning_rate": 8.361062663165953e-05,
      "loss": 0.0021,
      "step": 11837
    },
    {
      "epoch": 5.525320886814469,
      "grad_norm": 0.4921875,
      "learning_rate": 8.359615174025378e-05,
      "loss": 0.005,
      "step": 11838
    },
    {
      "epoch": 5.5257876312718786,
      "grad_norm": 0.2294921875,
      "learning_rate": 8.358167720204106e-05,
      "loss": 0.0015,
      "step": 11839
    },
    {
      "epoch": 5.526254375729288,
      "grad_norm": 0.12451171875,
      "learning_rate": 8.356720301733306e-05,
      "loss": 0.001,
      "step": 11840
    },
    {
      "epoch": 5.526721120186698,
      "grad_norm": 0.31640625,
      "learning_rate": 8.355272918644137e-05,
      "loss": 0.0029,
      "step": 11841
    },
    {
      "epoch": 5.527187864644107,
      "grad_norm": 0.3046875,
      "learning_rate": 8.353825570967769e-05,
      "loss": 0.0019,
      "step": 11842
    },
    {
      "epoch": 5.527654609101517,
      "grad_norm": 0.1533203125,
      "learning_rate": 8.352378258735364e-05,
      "loss": 0.0014,
      "step": 11843
    },
    {
      "epoch": 5.528121353558927,
      "grad_norm": 0.26953125,
      "learning_rate": 8.350930981978082e-05,
      "loss": 0.0019,
      "step": 11844
    },
    {
      "epoch": 5.528588098016336,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.349483740727086e-05,
      "loss": 0.0009,
      "step": 11845
    },
    {
      "epoch": 5.529054842473745,
      "grad_norm": 0.0712890625,
      "learning_rate": 8.348036535013537e-05,
      "loss": 0.0005,
      "step": 11846
    },
    {
      "epoch": 5.529521586931155,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.346589364868594e-05,
      "loss": 0.0009,
      "step": 11847
    },
    {
      "epoch": 5.529988331388565,
      "grad_norm": 0.1279296875,
      "learning_rate": 8.345142230323413e-05,
      "loss": 0.0011,
      "step": 11848
    },
    {
      "epoch": 5.530455075845975,
      "grad_norm": 0.048095703125,
      "learning_rate": 8.34369513140916e-05,
      "loss": 0.0005,
      "step": 11849
    },
    {
      "epoch": 5.530921820303384,
      "grad_norm": 0.1416015625,
      "learning_rate": 8.342248068156988e-05,
      "loss": 0.0042,
      "step": 11850
    },
    {
      "epoch": 5.531388564760793,
      "grad_norm": 0.22265625,
      "learning_rate": 8.340801040598055e-05,
      "loss": 0.0026,
      "step": 11851
    },
    {
      "epoch": 5.531855309218203,
      "grad_norm": 0.2275390625,
      "learning_rate": 8.339354048763512e-05,
      "loss": 0.0015,
      "step": 11852
    },
    {
      "epoch": 5.532322053675612,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.337907092684522e-05,
      "loss": 0.0008,
      "step": 11853
    },
    {
      "epoch": 5.532788798133022,
      "grad_norm": 0.0810546875,
      "learning_rate": 8.336460172392238e-05,
      "loss": 0.0007,
      "step": 11854
    },
    {
      "epoch": 5.533255542590432,
      "grad_norm": 0.07421875,
      "learning_rate": 8.335013287917809e-05,
      "loss": 0.0005,
      "step": 11855
    },
    {
      "epoch": 5.5337222870478415,
      "grad_norm": 0.154296875,
      "learning_rate": 8.333566439292394e-05,
      "loss": 0.0007,
      "step": 11856
    },
    {
      "epoch": 5.534189031505251,
      "grad_norm": 0.06640625,
      "learning_rate": 8.332119626547142e-05,
      "loss": 0.0006,
      "step": 11857
    },
    {
      "epoch": 5.53465577596266,
      "grad_norm": 0.130859375,
      "learning_rate": 8.330672849713202e-05,
      "loss": 0.0021,
      "step": 11858
    },
    {
      "epoch": 5.53512252042007,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.329226108821732e-05,
      "loss": 0.0035,
      "step": 11859
    },
    {
      "epoch": 5.53558926487748,
      "grad_norm": 1.421875,
      "learning_rate": 8.327779403903875e-05,
      "loss": 0.0139,
      "step": 11860
    },
    {
      "epoch": 5.536056009334889,
      "grad_norm": 0.0673828125,
      "learning_rate": 8.326332734990785e-05,
      "loss": 0.0024,
      "step": 11861
    },
    {
      "epoch": 5.5365227537922985,
      "grad_norm": 0.1044921875,
      "learning_rate": 8.324886102113605e-05,
      "loss": 0.0007,
      "step": 11862
    },
    {
      "epoch": 5.536989498249708,
      "grad_norm": 0.06982421875,
      "learning_rate": 8.323439505303488e-05,
      "loss": 0.0006,
      "step": 11863
    },
    {
      "epoch": 5.537456242707118,
      "grad_norm": 0.10302734375,
      "learning_rate": 8.321992944591578e-05,
      "loss": 0.0009,
      "step": 11864
    },
    {
      "epoch": 5.537922987164528,
      "grad_norm": 0.053466796875,
      "learning_rate": 8.32054642000902e-05,
      "loss": 0.0005,
      "step": 11865
    },
    {
      "epoch": 5.538389731621937,
      "grad_norm": 0.1806640625,
      "learning_rate": 8.319099931586965e-05,
      "loss": 0.0008,
      "step": 11866
    },
    {
      "epoch": 5.538856476079347,
      "grad_norm": 0.053466796875,
      "learning_rate": 8.31765347935655e-05,
      "loss": 0.0004,
      "step": 11867
    },
    {
      "epoch": 5.539323220536756,
      "grad_norm": 0.0390625,
      "learning_rate": 8.316207063348923e-05,
      "loss": 0.0005,
      "step": 11868
    },
    {
      "epoch": 5.539789964994165,
      "grad_norm": 0.0830078125,
      "learning_rate": 8.314760683595228e-05,
      "loss": 0.0006,
      "step": 11869
    },
    {
      "epoch": 5.540256709451575,
      "grad_norm": 0.1484375,
      "learning_rate": 8.313314340126603e-05,
      "loss": 0.0015,
      "step": 11870
    },
    {
      "epoch": 5.540723453908985,
      "grad_norm": 0.154296875,
      "learning_rate": 8.311868032974194e-05,
      "loss": 0.001,
      "step": 11871
    },
    {
      "epoch": 5.541190198366395,
      "grad_norm": 0.06787109375,
      "learning_rate": 8.310421762169136e-05,
      "loss": 0.0006,
      "step": 11872
    },
    {
      "epoch": 5.541656942823804,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.308975527742574e-05,
      "loss": 0.0006,
      "step": 11873
    },
    {
      "epoch": 5.542123687281213,
      "grad_norm": 0.7890625,
      "learning_rate": 8.307529329725645e-05,
      "loss": 0.008,
      "step": 11874
    },
    {
      "epoch": 5.542590431738623,
      "grad_norm": 0.15625,
      "learning_rate": 8.306083168149487e-05,
      "loss": 0.0035,
      "step": 11875
    },
    {
      "epoch": 5.543057176196033,
      "grad_norm": 0.21484375,
      "learning_rate": 8.304637043045238e-05,
      "loss": 0.0017,
      "step": 11876
    },
    {
      "epoch": 5.543523920653442,
      "grad_norm": 0.14453125,
      "learning_rate": 8.303190954444036e-05,
      "loss": 0.0013,
      "step": 11877
    },
    {
      "epoch": 5.543990665110852,
      "grad_norm": 0.28515625,
      "learning_rate": 8.301744902377012e-05,
      "loss": 0.0011,
      "step": 11878
    },
    {
      "epoch": 5.5444574095682615,
      "grad_norm": 0.21875,
      "learning_rate": 8.300298886875308e-05,
      "loss": 0.0059,
      "step": 11879
    },
    {
      "epoch": 5.544924154025671,
      "grad_norm": 0.1982421875,
      "learning_rate": 8.298852907970053e-05,
      "loss": 0.0016,
      "step": 11880
    },
    {
      "epoch": 5.54539089848308,
      "grad_norm": 0.1962890625,
      "learning_rate": 8.297406965692383e-05,
      "loss": 0.0013,
      "step": 11881
    },
    {
      "epoch": 5.54585764294049,
      "grad_norm": 0.05126953125,
      "learning_rate": 8.295961060073428e-05,
      "loss": 0.0005,
      "step": 11882
    },
    {
      "epoch": 5.5463243873979,
      "grad_norm": 0.36328125,
      "learning_rate": 8.294515191144325e-05,
      "loss": 0.0032,
      "step": 11883
    },
    {
      "epoch": 5.54679113185531,
      "grad_norm": 0.07763671875,
      "learning_rate": 8.293069358936202e-05,
      "loss": 0.0008,
      "step": 11884
    },
    {
      "epoch": 5.5472578763127185,
      "grad_norm": 0.2080078125,
      "learning_rate": 8.291623563480186e-05,
      "loss": 0.0008,
      "step": 11885
    },
    {
      "epoch": 5.547724620770128,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.290177804807418e-05,
      "loss": 0.0013,
      "step": 11886
    },
    {
      "epoch": 5.548191365227538,
      "grad_norm": 0.055419921875,
      "learning_rate": 8.288732082949013e-05,
      "loss": 0.0006,
      "step": 11887
    },
    {
      "epoch": 5.548658109684947,
      "grad_norm": 0.140625,
      "learning_rate": 8.287286397936107e-05,
      "loss": 0.0031,
      "step": 11888
    },
    {
      "epoch": 5.549124854142357,
      "grad_norm": 0.06396484375,
      "learning_rate": 8.285840749799822e-05,
      "loss": 0.0006,
      "step": 11889
    },
    {
      "epoch": 5.549591598599767,
      "grad_norm": 0.16015625,
      "learning_rate": 8.284395138571289e-05,
      "loss": 0.0009,
      "step": 11890
    },
    {
      "epoch": 5.550058343057176,
      "grad_norm": 0.2265625,
      "learning_rate": 8.282949564281633e-05,
      "loss": 0.0021,
      "step": 11891
    },
    {
      "epoch": 5.550525087514586,
      "grad_norm": 0.10546875,
      "learning_rate": 8.281504026961978e-05,
      "loss": 0.001,
      "step": 11892
    },
    {
      "epoch": 5.550991831971995,
      "grad_norm": 0.341796875,
      "learning_rate": 8.280058526643448e-05,
      "loss": 0.002,
      "step": 11893
    },
    {
      "epoch": 5.551458576429405,
      "grad_norm": 0.349609375,
      "learning_rate": 8.278613063357164e-05,
      "loss": 0.0026,
      "step": 11894
    },
    {
      "epoch": 5.551925320886815,
      "grad_norm": 0.36328125,
      "learning_rate": 8.277167637134253e-05,
      "loss": 0.0082,
      "step": 11895
    },
    {
      "epoch": 5.552392065344224,
      "grad_norm": 0.27734375,
      "learning_rate": 8.275722248005835e-05,
      "loss": 0.0052,
      "step": 11896
    },
    {
      "epoch": 5.552858809801633,
      "grad_norm": 0.326171875,
      "learning_rate": 8.274276896003027e-05,
      "loss": 0.0014,
      "step": 11897
    },
    {
      "epoch": 5.553325554259043,
      "grad_norm": 0.08837890625,
      "learning_rate": 8.272831581156955e-05,
      "loss": 0.0033,
      "step": 11898
    },
    {
      "epoch": 5.553792298716453,
      "grad_norm": 0.298828125,
      "learning_rate": 8.271386303498734e-05,
      "loss": 0.0011,
      "step": 11899
    },
    {
      "epoch": 5.554259043173863,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.269941063059484e-05,
      "loss": 0.0013,
      "step": 11900
    },
    {
      "epoch": 5.554725787631272,
      "grad_norm": 0.0419921875,
      "learning_rate": 8.268495859870322e-05,
      "loss": 0.0004,
      "step": 11901
    },
    {
      "epoch": 5.5551925320886815,
      "grad_norm": 0.058837890625,
      "learning_rate": 8.267050693962365e-05,
      "loss": 0.0005,
      "step": 11902
    },
    {
      "epoch": 5.555659276546091,
      "grad_norm": 0.15625,
      "learning_rate": 8.265605565366732e-05,
      "loss": 0.0022,
      "step": 11903
    },
    {
      "epoch": 5.5561260210035,
      "grad_norm": 0.236328125,
      "learning_rate": 8.264160474114531e-05,
      "loss": 0.0043,
      "step": 11904
    },
    {
      "epoch": 5.55659276546091,
      "grad_norm": 0.255859375,
      "learning_rate": 8.262715420236884e-05,
      "loss": 0.0015,
      "step": 11905
    },
    {
      "epoch": 5.55705950991832,
      "grad_norm": 0.04150390625,
      "learning_rate": 8.261270403764902e-05,
      "loss": 0.0025,
      "step": 11906
    },
    {
      "epoch": 5.55752625437573,
      "grad_norm": 0.205078125,
      "learning_rate": 8.259825424729696e-05,
      "loss": 0.0061,
      "step": 11907
    },
    {
      "epoch": 5.557992998833139,
      "grad_norm": 0.36328125,
      "learning_rate": 8.25838048316238e-05,
      "loss": 0.002,
      "step": 11908
    },
    {
      "epoch": 5.558459743290548,
      "grad_norm": 0.09326171875,
      "learning_rate": 8.256935579094066e-05,
      "loss": 0.0008,
      "step": 11909
    },
    {
      "epoch": 5.558926487747958,
      "grad_norm": 0.07421875,
      "learning_rate": 8.255490712555862e-05,
      "loss": 0.0008,
      "step": 11910
    },
    {
      "epoch": 5.559393232205368,
      "grad_norm": 0.150390625,
      "learning_rate": 8.254045883578876e-05,
      "loss": 0.0012,
      "step": 11911
    },
    {
      "epoch": 5.559859976662777,
      "grad_norm": 0.33203125,
      "learning_rate": 8.252601092194223e-05,
      "loss": 0.0015,
      "step": 11912
    },
    {
      "epoch": 5.5603267211201866,
      "grad_norm": 0.060546875,
      "learning_rate": 8.251156338433004e-05,
      "loss": 0.0006,
      "step": 11913
    },
    {
      "epoch": 5.560793465577596,
      "grad_norm": 0.10205078125,
      "learning_rate": 8.24971162232633e-05,
      "loss": 0.0009,
      "step": 11914
    },
    {
      "epoch": 5.561260210035006,
      "grad_norm": 0.111328125,
      "learning_rate": 8.248266943905309e-05,
      "loss": 0.003,
      "step": 11915
    },
    {
      "epoch": 5.561726954492415,
      "grad_norm": 0.08154296875,
      "learning_rate": 8.246822303201043e-05,
      "loss": 0.0024,
      "step": 11916
    },
    {
      "epoch": 5.562193698949825,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.245377700244637e-05,
      "loss": 0.0015,
      "step": 11917
    },
    {
      "epoch": 5.562660443407235,
      "grad_norm": 0.171875,
      "learning_rate": 8.243933135067197e-05,
      "loss": 0.0009,
      "step": 11918
    },
    {
      "epoch": 5.563127187864644,
      "grad_norm": 0.0791015625,
      "learning_rate": 8.242488607699826e-05,
      "loss": 0.0008,
      "step": 11919
    },
    {
      "epoch": 5.563593932322053,
      "grad_norm": 0.1064453125,
      "learning_rate": 8.241044118173624e-05,
      "loss": 0.0008,
      "step": 11920
    },
    {
      "epoch": 5.564060676779463,
      "grad_norm": 0.060302734375,
      "learning_rate": 8.239599666519692e-05,
      "loss": 0.0005,
      "step": 11921
    },
    {
      "epoch": 5.564527421236873,
      "grad_norm": 0.039306640625,
      "learning_rate": 8.238155252769135e-05,
      "loss": 0.0004,
      "step": 11922
    },
    {
      "epoch": 5.564994165694283,
      "grad_norm": 0.076171875,
      "learning_rate": 8.23671087695305e-05,
      "loss": 0.0032,
      "step": 11923
    },
    {
      "epoch": 5.565460910151692,
      "grad_norm": 0.224609375,
      "learning_rate": 8.235266539102533e-05,
      "loss": 0.0012,
      "step": 11924
    },
    {
      "epoch": 5.565927654609101,
      "grad_norm": 0.10546875,
      "learning_rate": 8.23382223924869e-05,
      "loss": 0.0031,
      "step": 11925
    },
    {
      "epoch": 5.566394399066511,
      "grad_norm": 0.0478515625,
      "learning_rate": 8.232377977422612e-05,
      "loss": 0.0005,
      "step": 11926
    },
    {
      "epoch": 5.566861143523921,
      "grad_norm": 0.27734375,
      "learning_rate": 8.230933753655395e-05,
      "loss": 0.0021,
      "step": 11927
    },
    {
      "epoch": 5.56732788798133,
      "grad_norm": 0.0810546875,
      "learning_rate": 8.229489567978141e-05,
      "loss": 0.0006,
      "step": 11928
    },
    {
      "epoch": 5.56779463243874,
      "grad_norm": 0.05712890625,
      "learning_rate": 8.228045420421939e-05,
      "loss": 0.0006,
      "step": 11929
    },
    {
      "epoch": 5.5682613768961495,
      "grad_norm": 0.169921875,
      "learning_rate": 8.226601311017886e-05,
      "loss": 0.002,
      "step": 11930
    },
    {
      "epoch": 5.568728121353558,
      "grad_norm": 0.14453125,
      "learning_rate": 8.225157239797072e-05,
      "loss": 0.0011,
      "step": 11931
    },
    {
      "epoch": 5.569194865810968,
      "grad_norm": 0.1748046875,
      "learning_rate": 8.223713206790595e-05,
      "loss": 0.0009,
      "step": 11932
    },
    {
      "epoch": 5.569661610268378,
      "grad_norm": 0.150390625,
      "learning_rate": 8.222269212029542e-05,
      "loss": 0.0032,
      "step": 11933
    },
    {
      "epoch": 5.570128354725788,
      "grad_norm": 0.166015625,
      "learning_rate": 8.220825255545009e-05,
      "loss": 0.0048,
      "step": 11934
    },
    {
      "epoch": 5.570595099183198,
      "grad_norm": 0.12109375,
      "learning_rate": 8.219381337368079e-05,
      "loss": 0.0007,
      "step": 11935
    },
    {
      "epoch": 5.5710618436406065,
      "grad_norm": 0.048828125,
      "learning_rate": 8.217937457529843e-05,
      "loss": 0.0036,
      "step": 11936
    },
    {
      "epoch": 5.571528588098016,
      "grad_norm": 0.333984375,
      "learning_rate": 8.216493616061391e-05,
      "loss": 0.0024,
      "step": 11937
    },
    {
      "epoch": 5.571995332555426,
      "grad_norm": 0.16796875,
      "learning_rate": 8.215049812993813e-05,
      "loss": 0.0008,
      "step": 11938
    },
    {
      "epoch": 5.572462077012835,
      "grad_norm": 0.05029296875,
      "learning_rate": 8.213606048358189e-05,
      "loss": 0.0004,
      "step": 11939
    },
    {
      "epoch": 5.572928821470245,
      "grad_norm": 0.1875,
      "learning_rate": 8.212162322185612e-05,
      "loss": 0.0009,
      "step": 11940
    },
    {
      "epoch": 5.573395565927655,
      "grad_norm": 0.033935546875,
      "learning_rate": 8.210718634507163e-05,
      "loss": 0.0004,
      "step": 11941
    },
    {
      "epoch": 5.573862310385064,
      "grad_norm": 0.0284423828125,
      "learning_rate": 8.209274985353927e-05,
      "loss": 0.0004,
      "step": 11942
    },
    {
      "epoch": 5.574329054842474,
      "grad_norm": 0.07275390625,
      "learning_rate": 8.207831374756987e-05,
      "loss": 0.0009,
      "step": 11943
    },
    {
      "epoch": 5.574795799299883,
      "grad_norm": 0.08056640625,
      "learning_rate": 8.206387802747426e-05,
      "loss": 0.0037,
      "step": 11944
    },
    {
      "epoch": 5.575262543757293,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.204944269356326e-05,
      "loss": 0.0008,
      "step": 11945
    },
    {
      "epoch": 5.575729288214703,
      "grad_norm": 0.09765625,
      "learning_rate": 8.203500774614766e-05,
      "loss": 0.0039,
      "step": 11946
    },
    {
      "epoch": 5.576196032672112,
      "grad_norm": 0.28125,
      "learning_rate": 8.20205731855383e-05,
      "loss": 0.0014,
      "step": 11947
    },
    {
      "epoch": 5.576662777129521,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.200613901204595e-05,
      "loss": 0.0008,
      "step": 11948
    },
    {
      "epoch": 5.577129521586931,
      "grad_norm": 0.1376953125,
      "learning_rate": 8.199170522598135e-05,
      "loss": 0.0012,
      "step": 11949
    },
    {
      "epoch": 5.577596266044341,
      "grad_norm": 0.23046875,
      "learning_rate": 8.197727182765536e-05,
      "loss": 0.0012,
      "step": 11950
    },
    {
      "epoch": 5.57806301050175,
      "grad_norm": 0.052001953125,
      "learning_rate": 8.19628388173787e-05,
      "loss": 0.0007,
      "step": 11951
    },
    {
      "epoch": 5.57852975495916,
      "grad_norm": 0.08740234375,
      "learning_rate": 8.194840619546212e-05,
      "loss": 0.0007,
      "step": 11952
    },
    {
      "epoch": 5.5789964994165695,
      "grad_norm": 0.047119140625,
      "learning_rate": 8.193397396221637e-05,
      "loss": 0.0004,
      "step": 11953
    },
    {
      "epoch": 5.579463243873979,
      "grad_norm": 0.095703125,
      "learning_rate": 8.191954211795223e-05,
      "loss": 0.0009,
      "step": 11954
    },
    {
      "epoch": 5.579929988331388,
      "grad_norm": 0.283203125,
      "learning_rate": 8.190511066298042e-05,
      "loss": 0.0029,
      "step": 11955
    },
    {
      "epoch": 5.580396732788798,
      "grad_norm": 0.1376953125,
      "learning_rate": 8.18906795976116e-05,
      "loss": 0.001,
      "step": 11956
    },
    {
      "epoch": 5.580863477246208,
      "grad_norm": 0.06640625,
      "learning_rate": 8.187624892215658e-05,
      "loss": 0.0006,
      "step": 11957
    },
    {
      "epoch": 5.581330221703618,
      "grad_norm": 0.1943359375,
      "learning_rate": 8.186181863692604e-05,
      "loss": 0.0007,
      "step": 11958
    },
    {
      "epoch": 5.5817969661610265,
      "grad_norm": 0.23828125,
      "learning_rate": 8.184738874223066e-05,
      "loss": 0.0013,
      "step": 11959
    },
    {
      "epoch": 5.582263710618436,
      "grad_norm": 0.181640625,
      "learning_rate": 8.183295923838111e-05,
      "loss": 0.0056,
      "step": 11960
    },
    {
      "epoch": 5.582730455075846,
      "grad_norm": 0.0771484375,
      "learning_rate": 8.181853012568815e-05,
      "loss": 0.0008,
      "step": 11961
    },
    {
      "epoch": 5.583197199533256,
      "grad_norm": 0.1923828125,
      "learning_rate": 8.180410140446239e-05,
      "loss": 0.0012,
      "step": 11962
    },
    {
      "epoch": 5.583663943990665,
      "grad_norm": 0.056884765625,
      "learning_rate": 8.178967307501449e-05,
      "loss": 0.0027,
      "step": 11963
    },
    {
      "epoch": 5.584130688448075,
      "grad_norm": 0.1220703125,
      "learning_rate": 8.177524513765517e-05,
      "loss": 0.0034,
      "step": 11964
    },
    {
      "epoch": 5.584597432905484,
      "grad_norm": 0.1943359375,
      "learning_rate": 8.176081759269503e-05,
      "loss": 0.0008,
      "step": 11965
    },
    {
      "epoch": 5.585064177362894,
      "grad_norm": 0.298828125,
      "learning_rate": 8.17463904404447e-05,
      "loss": 0.0051,
      "step": 11966
    },
    {
      "epoch": 5.585530921820303,
      "grad_norm": 0.06201171875,
      "learning_rate": 8.173196368121487e-05,
      "loss": 0.0006,
      "step": 11967
    },
    {
      "epoch": 5.585997666277713,
      "grad_norm": 0.12158203125,
      "learning_rate": 8.171753731531615e-05,
      "loss": 0.0007,
      "step": 11968
    },
    {
      "epoch": 5.586464410735123,
      "grad_norm": 0.09130859375,
      "learning_rate": 8.170311134305911e-05,
      "loss": 0.0007,
      "step": 11969
    },
    {
      "epoch": 5.5869311551925325,
      "grad_norm": 0.0556640625,
      "learning_rate": 8.168868576475436e-05,
      "loss": 0.0039,
      "step": 11970
    },
    {
      "epoch": 5.587397899649941,
      "grad_norm": 0.10546875,
      "learning_rate": 8.167426058071254e-05,
      "loss": 0.0011,
      "step": 11971
    },
    {
      "epoch": 5.587864644107351,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.165983579124424e-05,
      "loss": 0.0012,
      "step": 11972
    },
    {
      "epoch": 5.588331388564761,
      "grad_norm": 0.038330078125,
      "learning_rate": 8.164541139665999e-05,
      "loss": 0.0004,
      "step": 11973
    },
    {
      "epoch": 5.58879813302217,
      "grad_norm": 0.158203125,
      "learning_rate": 8.163098739727042e-05,
      "loss": 0.0008,
      "step": 11974
    },
    {
      "epoch": 5.58926487747958,
      "grad_norm": 0.1806640625,
      "learning_rate": 8.161656379338608e-05,
      "loss": 0.0012,
      "step": 11975
    },
    {
      "epoch": 5.5897316219369895,
      "grad_norm": 0.20703125,
      "learning_rate": 8.160214058531749e-05,
      "loss": 0.001,
      "step": 11976
    },
    {
      "epoch": 5.590198366394399,
      "grad_norm": 0.1396484375,
      "learning_rate": 8.158771777337524e-05,
      "loss": 0.0009,
      "step": 11977
    },
    {
      "epoch": 5.590665110851809,
      "grad_norm": 0.2158203125,
      "learning_rate": 8.157329535786988e-05,
      "loss": 0.0081,
      "step": 11978
    },
    {
      "epoch": 5.591131855309218,
      "grad_norm": 0.1298828125,
      "learning_rate": 8.155887333911189e-05,
      "loss": 0.0007,
      "step": 11979
    },
    {
      "epoch": 5.591598599766628,
      "grad_norm": 0.05322265625,
      "learning_rate": 8.154445171741181e-05,
      "loss": 0.0007,
      "step": 11980
    },
    {
      "epoch": 5.592065344224038,
      "grad_norm": 0.04541015625,
      "learning_rate": 8.153003049308019e-05,
      "loss": 0.0004,
      "step": 11981
    },
    {
      "epoch": 5.5925320886814465,
      "grad_norm": 0.30859375,
      "learning_rate": 8.151560966642751e-05,
      "loss": 0.002,
      "step": 11982
    },
    {
      "epoch": 5.592998833138856,
      "grad_norm": 0.09130859375,
      "learning_rate": 8.150118923776426e-05,
      "loss": 0.0027,
      "step": 11983
    },
    {
      "epoch": 5.593465577596266,
      "grad_norm": 0.031982421875,
      "learning_rate": 8.14867692074009e-05,
      "loss": 0.0024,
      "step": 11984
    },
    {
      "epoch": 5.593932322053676,
      "grad_norm": 0.07275390625,
      "learning_rate": 8.147234957564794e-05,
      "loss": 0.0035,
      "step": 11985
    },
    {
      "epoch": 5.594399066511086,
      "grad_norm": 0.0556640625,
      "learning_rate": 8.145793034281588e-05,
      "loss": 0.0006,
      "step": 11986
    },
    {
      "epoch": 5.594865810968495,
      "grad_norm": 0.1416015625,
      "learning_rate": 8.144351150921514e-05,
      "loss": 0.0009,
      "step": 11987
    },
    {
      "epoch": 5.595332555425904,
      "grad_norm": 0.044921875,
      "learning_rate": 8.142909307515617e-05,
      "loss": 0.0027,
      "step": 11988
    },
    {
      "epoch": 5.595799299883314,
      "grad_norm": 0.052490234375,
      "learning_rate": 8.141467504094944e-05,
      "loss": 0.0006,
      "step": 11989
    },
    {
      "epoch": 5.596266044340723,
      "grad_norm": 0.11181640625,
      "learning_rate": 8.140025740690539e-05,
      "loss": 0.001,
      "step": 11990
    },
    {
      "epoch": 5.596732788798133,
      "grad_norm": 0.1025390625,
      "learning_rate": 8.138584017333442e-05,
      "loss": 0.0007,
      "step": 11991
    },
    {
      "epoch": 5.597199533255543,
      "grad_norm": 0.014892578125,
      "learning_rate": 8.137142334054696e-05,
      "loss": 0.0002,
      "step": 11992
    },
    {
      "epoch": 5.597666277712952,
      "grad_norm": 0.12890625,
      "learning_rate": 8.135700690885344e-05,
      "loss": 0.0007,
      "step": 11993
    },
    {
      "epoch": 5.598133022170361,
      "grad_norm": 0.10498046875,
      "learning_rate": 8.134259087856424e-05,
      "loss": 0.0008,
      "step": 11994
    },
    {
      "epoch": 5.598599766627771,
      "grad_norm": 0.04541015625,
      "learning_rate": 8.132817524998975e-05,
      "loss": 0.0025,
      "step": 11995
    },
    {
      "epoch": 5.599066511085181,
      "grad_norm": 0.30078125,
      "learning_rate": 8.131376002344038e-05,
      "loss": 0.0031,
      "step": 11996
    },
    {
      "epoch": 5.599533255542591,
      "grad_norm": 0.060546875,
      "learning_rate": 8.129934519922649e-05,
      "loss": 0.0006,
      "step": 11997
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.044677734375,
      "learning_rate": 8.128493077765841e-05,
      "loss": 0.0005,
      "step": 11998
    },
    {
      "epoch": 5.600466744457409,
      "grad_norm": 0.251953125,
      "learning_rate": 8.127051675904658e-05,
      "loss": 0.0017,
      "step": 11999
    },
    {
      "epoch": 5.600933488914819,
      "grad_norm": 0.326171875,
      "learning_rate": 8.125610314370129e-05,
      "loss": 0.0062,
      "step": 12000
    },
    {
      "epoch": 5.601400233372229,
      "grad_norm": 0.166015625,
      "learning_rate": 8.124168993193289e-05,
      "loss": 0.0021,
      "step": 12001
    },
    {
      "epoch": 5.601866977829638,
      "grad_norm": 0.21875,
      "learning_rate": 8.122727712405172e-05,
      "loss": 0.0011,
      "step": 12002
    },
    {
      "epoch": 5.602333722287048,
      "grad_norm": 0.1806640625,
      "learning_rate": 8.12128647203681e-05,
      "loss": 0.0008,
      "step": 12003
    },
    {
      "epoch": 5.6028004667444575,
      "grad_norm": 0.3984375,
      "learning_rate": 8.119845272119235e-05,
      "loss": 0.0013,
      "step": 12004
    },
    {
      "epoch": 5.603267211201867,
      "grad_norm": 0.205078125,
      "learning_rate": 8.118404112683478e-05,
      "loss": 0.0009,
      "step": 12005
    },
    {
      "epoch": 5.603733955659276,
      "grad_norm": 0.05810546875,
      "learning_rate": 8.116962993760567e-05,
      "loss": 0.0005,
      "step": 12006
    },
    {
      "epoch": 5.604200700116686,
      "grad_norm": 0.267578125,
      "learning_rate": 8.115521915381533e-05,
      "loss": 0.0013,
      "step": 12007
    },
    {
      "epoch": 5.604667444574096,
      "grad_norm": 0.4765625,
      "learning_rate": 8.114080877577401e-05,
      "loss": 0.0024,
      "step": 12008
    },
    {
      "epoch": 5.605134189031506,
      "grad_norm": 0.10888671875,
      "learning_rate": 8.112639880379203e-05,
      "loss": 0.0008,
      "step": 12009
    },
    {
      "epoch": 5.6056009334889145,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.111198923817962e-05,
      "loss": 0.0009,
      "step": 12010
    },
    {
      "epoch": 5.606067677946324,
      "grad_norm": 0.0673828125,
      "learning_rate": 8.109758007924703e-05,
      "loss": 0.0005,
      "step": 12011
    },
    {
      "epoch": 5.606534422403734,
      "grad_norm": 0.07958984375,
      "learning_rate": 8.10831713273045e-05,
      "loss": 0.004,
      "step": 12012
    },
    {
      "epoch": 5.607001166861144,
      "grad_norm": 0.45703125,
      "learning_rate": 8.106876298266231e-05,
      "loss": 0.0023,
      "step": 12013
    },
    {
      "epoch": 5.607467911318553,
      "grad_norm": 0.08544921875,
      "learning_rate": 8.105435504563063e-05,
      "loss": 0.0007,
      "step": 12014
    },
    {
      "epoch": 5.607934655775963,
      "grad_norm": 0.040771484375,
      "learning_rate": 8.103994751651971e-05,
      "loss": 0.0004,
      "step": 12015
    },
    {
      "epoch": 5.608401400233372,
      "grad_norm": 0.061279296875,
      "learning_rate": 8.102554039563978e-05,
      "loss": 0.0006,
      "step": 12016
    },
    {
      "epoch": 5.608868144690781,
      "grad_norm": 0.08154296875,
      "learning_rate": 8.1011133683301e-05,
      "loss": 0.0007,
      "step": 12017
    },
    {
      "epoch": 5.609334889148191,
      "grad_norm": 0.061767578125,
      "learning_rate": 8.09967273798136e-05,
      "loss": 0.0005,
      "step": 12018
    },
    {
      "epoch": 5.609801633605601,
      "grad_norm": 0.146484375,
      "learning_rate": 8.09823214854877e-05,
      "loss": 0.001,
      "step": 12019
    },
    {
      "epoch": 5.610268378063011,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.096791600063357e-05,
      "loss": 0.0011,
      "step": 12020
    },
    {
      "epoch": 5.6107351225204205,
      "grad_norm": 0.259765625,
      "learning_rate": 8.09535109255613e-05,
      "loss": 0.0014,
      "step": 12021
    },
    {
      "epoch": 5.611201866977829,
      "grad_norm": 0.055908203125,
      "learning_rate": 8.093910626058106e-05,
      "loss": 0.0005,
      "step": 12022
    },
    {
      "epoch": 5.611668611435239,
      "grad_norm": 0.33203125,
      "learning_rate": 8.092470200600301e-05,
      "loss": 0.0038,
      "step": 12023
    },
    {
      "epoch": 5.612135355892649,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.091029816213732e-05,
      "loss": 0.0029,
      "step": 12024
    },
    {
      "epoch": 5.612602100350058,
      "grad_norm": 0.51171875,
      "learning_rate": 8.089589472929406e-05,
      "loss": 0.0021,
      "step": 12025
    },
    {
      "epoch": 5.613068844807468,
      "grad_norm": 0.03173828125,
      "learning_rate": 8.08814917077834e-05,
      "loss": 0.0004,
      "step": 12026
    },
    {
      "epoch": 5.6135355892648775,
      "grad_norm": 0.0810546875,
      "learning_rate": 8.086708909791544e-05,
      "loss": 0.0007,
      "step": 12027
    },
    {
      "epoch": 5.614002333722287,
      "grad_norm": 0.1640625,
      "learning_rate": 8.085268690000029e-05,
      "loss": 0.0014,
      "step": 12028
    },
    {
      "epoch": 5.614469078179697,
      "grad_norm": 0.06396484375,
      "learning_rate": 8.0838285114348e-05,
      "loss": 0.0006,
      "step": 12029
    },
    {
      "epoch": 5.614935822637106,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.082388374126876e-05,
      "loss": 0.0011,
      "step": 12030
    },
    {
      "epoch": 5.615402567094516,
      "grad_norm": 0.279296875,
      "learning_rate": 8.08094827810725e-05,
      "loss": 0.0009,
      "step": 12031
    },
    {
      "epoch": 5.615869311551926,
      "grad_norm": 0.04150390625,
      "learning_rate": 8.079508223406942e-05,
      "loss": 0.0005,
      "step": 12032
    },
    {
      "epoch": 5.6163360560093345,
      "grad_norm": 0.049072265625,
      "learning_rate": 8.07806821005695e-05,
      "loss": 0.0005,
      "step": 12033
    },
    {
      "epoch": 5.616802800466744,
      "grad_norm": 0.341796875,
      "learning_rate": 8.076628238088282e-05,
      "loss": 0.0024,
      "step": 12034
    },
    {
      "epoch": 5.617269544924154,
      "grad_norm": 0.026123046875,
      "learning_rate": 8.075188307531943e-05,
      "loss": 0.0003,
      "step": 12035
    },
    {
      "epoch": 5.617736289381564,
      "grad_norm": 0.1181640625,
      "learning_rate": 8.073748418418935e-05,
      "loss": 0.0006,
      "step": 12036
    },
    {
      "epoch": 5.618203033838973,
      "grad_norm": 0.30078125,
      "learning_rate": 8.07230857078026e-05,
      "loss": 0.0021,
      "step": 12037
    },
    {
      "epoch": 5.618669778296383,
      "grad_norm": 0.29296875,
      "learning_rate": 8.07086876464692e-05,
      "loss": 0.0036,
      "step": 12038
    },
    {
      "epoch": 5.619136522753792,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.069429000049918e-05,
      "loss": 0.0012,
      "step": 12039
    },
    {
      "epoch": 5.619603267211202,
      "grad_norm": 0.03759765625,
      "learning_rate": 8.067989277020251e-05,
      "loss": 0.0004,
      "step": 12040
    },
    {
      "epoch": 5.620070011668611,
      "grad_norm": 0.0771484375,
      "learning_rate": 8.066549595588916e-05,
      "loss": 0.0015,
      "step": 12041
    },
    {
      "epoch": 5.620536756126021,
      "grad_norm": 0.28125,
      "learning_rate": 8.065109955786916e-05,
      "loss": 0.0033,
      "step": 12042
    },
    {
      "epoch": 5.621003500583431,
      "grad_norm": 0.2431640625,
      "learning_rate": 8.063670357645246e-05,
      "loss": 0.0012,
      "step": 12043
    },
    {
      "epoch": 5.6214702450408405,
      "grad_norm": 0.056640625,
      "learning_rate": 8.062230801194898e-05,
      "loss": 0.0006,
      "step": 12044
    },
    {
      "epoch": 5.621936989498249,
      "grad_norm": 0.083984375,
      "learning_rate": 8.060791286466874e-05,
      "loss": 0.0005,
      "step": 12045
    },
    {
      "epoch": 5.622403733955659,
      "grad_norm": 0.1806640625,
      "learning_rate": 8.059351813492165e-05,
      "loss": 0.0008,
      "step": 12046
    },
    {
      "epoch": 5.622870478413069,
      "grad_norm": 0.33984375,
      "learning_rate": 8.057912382301762e-05,
      "loss": 0.001,
      "step": 12047
    },
    {
      "epoch": 5.623337222870479,
      "grad_norm": 0.0703125,
      "learning_rate": 8.056472992926663e-05,
      "loss": 0.0008,
      "step": 12048
    },
    {
      "epoch": 5.623803967327888,
      "grad_norm": 0.10302734375,
      "learning_rate": 8.055033645397858e-05,
      "loss": 0.0006,
      "step": 12049
    },
    {
      "epoch": 5.6242707117852975,
      "grad_norm": 0.0223388671875,
      "learning_rate": 8.053594339746334e-05,
      "loss": 0.0004,
      "step": 12050
    },
    {
      "epoch": 5.624737456242707,
      "grad_norm": 0.236328125,
      "learning_rate": 8.052155076003083e-05,
      "loss": 0.0042,
      "step": 12051
    },
    {
      "epoch": 5.625204200700117,
      "grad_norm": 0.203125,
      "learning_rate": 8.050715854199096e-05,
      "loss": 0.0025,
      "step": 12052
    },
    {
      "epoch": 5.625670945157526,
      "grad_norm": 0.02734375,
      "learning_rate": 8.049276674365361e-05,
      "loss": 0.0004,
      "step": 12053
    },
    {
      "epoch": 5.626137689614936,
      "grad_norm": 0.283203125,
      "learning_rate": 8.047837536532858e-05,
      "loss": 0.003,
      "step": 12054
    },
    {
      "epoch": 5.626604434072346,
      "grad_norm": 0.0517578125,
      "learning_rate": 8.046398440732583e-05,
      "loss": 0.0005,
      "step": 12055
    },
    {
      "epoch": 5.627071178529755,
      "grad_norm": 0.0279541015625,
      "learning_rate": 8.044959386995518e-05,
      "loss": 0.0004,
      "step": 12056
    },
    {
      "epoch": 5.627537922987164,
      "grad_norm": 0.07568359375,
      "learning_rate": 8.043520375352642e-05,
      "loss": 0.0005,
      "step": 12057
    },
    {
      "epoch": 5.628004667444574,
      "grad_norm": 0.212890625,
      "learning_rate": 8.042081405834946e-05,
      "loss": 0.0008,
      "step": 12058
    },
    {
      "epoch": 5.628471411901984,
      "grad_norm": 0.06298828125,
      "learning_rate": 8.040642478473412e-05,
      "loss": 0.0059,
      "step": 12059
    },
    {
      "epoch": 5.628938156359393,
      "grad_norm": 0.056640625,
      "learning_rate": 8.039203593299017e-05,
      "loss": 0.0006,
      "step": 12060
    },
    {
      "epoch": 5.629404900816803,
      "grad_norm": 0.10498046875,
      "learning_rate": 8.037764750342743e-05,
      "loss": 0.0006,
      "step": 12061
    },
    {
      "epoch": 5.629871645274212,
      "grad_norm": 0.04296875,
      "learning_rate": 8.036325949635572e-05,
      "loss": 0.0004,
      "step": 12062
    },
    {
      "epoch": 5.630338389731622,
      "grad_norm": 0.059326171875,
      "learning_rate": 8.034887191208485e-05,
      "loss": 0.0005,
      "step": 12063
    },
    {
      "epoch": 5.630805134189032,
      "grad_norm": 0.04150390625,
      "learning_rate": 8.033448475092452e-05,
      "loss": 0.0004,
      "step": 12064
    },
    {
      "epoch": 5.631271878646441,
      "grad_norm": 0.034912109375,
      "learning_rate": 8.032009801318458e-05,
      "loss": 0.0004,
      "step": 12065
    },
    {
      "epoch": 5.631738623103851,
      "grad_norm": 0.2060546875,
      "learning_rate": 8.030571169917477e-05,
      "loss": 0.001,
      "step": 12066
    },
    {
      "epoch": 5.6322053675612604,
      "grad_norm": 0.130859375,
      "learning_rate": 8.029132580920482e-05,
      "loss": 0.0005,
      "step": 12067
    },
    {
      "epoch": 5.632672112018669,
      "grad_norm": 0.1875,
      "learning_rate": 8.027694034358451e-05,
      "loss": 0.0055,
      "step": 12068
    },
    {
      "epoch": 5.633138856476079,
      "grad_norm": 0.0289306640625,
      "learning_rate": 8.026255530262356e-05,
      "loss": 0.0004,
      "step": 12069
    },
    {
      "epoch": 5.633605600933489,
      "grad_norm": 0.068359375,
      "learning_rate": 8.024817068663168e-05,
      "loss": 0.0005,
      "step": 12070
    },
    {
      "epoch": 5.634072345390899,
      "grad_norm": 0.091796875,
      "learning_rate": 8.023378649591859e-05,
      "loss": 0.0006,
      "step": 12071
    },
    {
      "epoch": 5.6345390898483085,
      "grad_norm": 0.10009765625,
      "learning_rate": 8.021940273079402e-05,
      "loss": 0.0006,
      "step": 12072
    },
    {
      "epoch": 5.635005834305717,
      "grad_norm": 0.09228515625,
      "learning_rate": 8.020501939156767e-05,
      "loss": 0.0009,
      "step": 12073
    },
    {
      "epoch": 5.635472578763127,
      "grad_norm": 0.115234375,
      "learning_rate": 8.019063647854919e-05,
      "loss": 0.0007,
      "step": 12074
    },
    {
      "epoch": 5.635939323220537,
      "grad_norm": 0.1708984375,
      "learning_rate": 8.017625399204831e-05,
      "loss": 0.0009,
      "step": 12075
    },
    {
      "epoch": 5.636406067677946,
      "grad_norm": 0.0927734375,
      "learning_rate": 8.016187193237467e-05,
      "loss": 0.0006,
      "step": 12076
    },
    {
      "epoch": 5.636872812135356,
      "grad_norm": 0.158203125,
      "learning_rate": 8.014749029983794e-05,
      "loss": 0.0008,
      "step": 12077
    },
    {
      "epoch": 5.6373395565927655,
      "grad_norm": 0.21875,
      "learning_rate": 8.013310909474779e-05,
      "loss": 0.0008,
      "step": 12078
    },
    {
      "epoch": 5.637806301050175,
      "grad_norm": 0.07861328125,
      "learning_rate": 8.01187283174138e-05,
      "loss": 0.0008,
      "step": 12079
    },
    {
      "epoch": 5.638273045507584,
      "grad_norm": 0.08154296875,
      "learning_rate": 8.010434796814568e-05,
      "loss": 0.0006,
      "step": 12080
    },
    {
      "epoch": 5.638739789964994,
      "grad_norm": 0.08203125,
      "learning_rate": 8.0089968047253e-05,
      "loss": 0.0004,
      "step": 12081
    },
    {
      "epoch": 5.639206534422404,
      "grad_norm": 0.034912109375,
      "learning_rate": 8.007558855504542e-05,
      "loss": 0.0004,
      "step": 12082
    },
    {
      "epoch": 5.639673278879814,
      "grad_norm": 0.30859375,
      "learning_rate": 8.00612094918325e-05,
      "loss": 0.0082,
      "step": 12083
    },
    {
      "epoch": 5.6401400233372225,
      "grad_norm": 0.0869140625,
      "learning_rate": 8.004683085792388e-05,
      "loss": 0.0007,
      "step": 12084
    },
    {
      "epoch": 5.640606767794632,
      "grad_norm": 0.04931640625,
      "learning_rate": 8.003245265362913e-05,
      "loss": 0.0005,
      "step": 12085
    },
    {
      "epoch": 5.641073512252042,
      "grad_norm": 0.0205078125,
      "learning_rate": 8.001807487925782e-05,
      "loss": 0.0003,
      "step": 12086
    },
    {
      "epoch": 5.641540256709452,
      "grad_norm": 0.07958984375,
      "learning_rate": 8.000369753511952e-05,
      "loss": 0.0005,
      "step": 12087
    },
    {
      "epoch": 5.642007001166861,
      "grad_norm": 0.2265625,
      "learning_rate": 7.998932062152382e-05,
      "loss": 0.0011,
      "step": 12088
    },
    {
      "epoch": 5.642473745624271,
      "grad_norm": 0.06298828125,
      "learning_rate": 7.997494413878024e-05,
      "loss": 0.0004,
      "step": 12089
    },
    {
      "epoch": 5.64294049008168,
      "grad_norm": 0.11328125,
      "learning_rate": 7.996056808719832e-05,
      "loss": 0.0011,
      "step": 12090
    },
    {
      "epoch": 5.64340723453909,
      "grad_norm": 0.048828125,
      "learning_rate": 7.99461924670876e-05,
      "loss": 0.0004,
      "step": 12091
    },
    {
      "epoch": 5.643873978996499,
      "grad_norm": 0.201171875,
      "learning_rate": 7.993181727875765e-05,
      "loss": 0.0016,
      "step": 12092
    },
    {
      "epoch": 5.644340723453909,
      "grad_norm": 0.318359375,
      "learning_rate": 7.991744252251788e-05,
      "loss": 0.0054,
      "step": 12093
    },
    {
      "epoch": 5.644807467911319,
      "grad_norm": 0.1796875,
      "learning_rate": 7.99030681986779e-05,
      "loss": 0.001,
      "step": 12094
    },
    {
      "epoch": 5.6452742123687285,
      "grad_norm": 0.053466796875,
      "learning_rate": 7.988869430754714e-05,
      "loss": 0.0027,
      "step": 12095
    },
    {
      "epoch": 5.645740956826137,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.987432084943509e-05,
      "loss": 0.001,
      "step": 12096
    },
    {
      "epoch": 5.646207701283547,
      "grad_norm": 0.08837890625,
      "learning_rate": 7.985994782465127e-05,
      "loss": 0.0068,
      "step": 12097
    },
    {
      "epoch": 5.646674445740957,
      "grad_norm": 0.05712890625,
      "learning_rate": 7.984557523350512e-05,
      "loss": 0.0005,
      "step": 12098
    },
    {
      "epoch": 5.647141190198367,
      "grad_norm": 0.0810546875,
      "learning_rate": 7.98312030763061e-05,
      "loss": 0.0007,
      "step": 12099
    },
    {
      "epoch": 5.647607934655776,
      "grad_norm": 0.08837890625,
      "learning_rate": 7.981683135336362e-05,
      "loss": 0.0007,
      "step": 12100
    },
    {
      "epoch": 5.6480746791131855,
      "grad_norm": 0.1298828125,
      "learning_rate": 7.980246006498718e-05,
      "loss": 0.0007,
      "step": 12101
    },
    {
      "epoch": 5.648541423570595,
      "grad_norm": 0.25390625,
      "learning_rate": 7.97880892114862e-05,
      "loss": 0.0011,
      "step": 12102
    },
    {
      "epoch": 5.649008168028004,
      "grad_norm": 0.2314453125,
      "learning_rate": 7.977371879317004e-05,
      "loss": 0.001,
      "step": 12103
    },
    {
      "epoch": 5.649474912485414,
      "grad_norm": 0.1337890625,
      "learning_rate": 7.975934881034819e-05,
      "loss": 0.0008,
      "step": 12104
    },
    {
      "epoch": 5.649941656942824,
      "grad_norm": 0.05322265625,
      "learning_rate": 7.974497926333001e-05,
      "loss": 0.002,
      "step": 12105
    },
    {
      "epoch": 5.650408401400234,
      "grad_norm": 0.052734375,
      "learning_rate": 7.973061015242489e-05,
      "loss": 0.002,
      "step": 12106
    },
    {
      "epoch": 5.650875145857643,
      "grad_norm": 0.1201171875,
      "learning_rate": 7.971624147794224e-05,
      "loss": 0.0007,
      "step": 12107
    },
    {
      "epoch": 5.651341890315052,
      "grad_norm": 0.06884765625,
      "learning_rate": 7.970187324019141e-05,
      "loss": 0.0005,
      "step": 12108
    },
    {
      "epoch": 5.651808634772462,
      "grad_norm": 0.1806640625,
      "learning_rate": 7.968750543948177e-05,
      "loss": 0.0036,
      "step": 12109
    },
    {
      "epoch": 5.652275379229872,
      "grad_norm": 0.033447265625,
      "learning_rate": 7.967313807612266e-05,
      "loss": 0.0003,
      "step": 12110
    },
    {
      "epoch": 5.652742123687281,
      "grad_norm": 0.1884765625,
      "learning_rate": 7.965877115042347e-05,
      "loss": 0.0049,
      "step": 12111
    },
    {
      "epoch": 5.653208868144691,
      "grad_norm": 0.103515625,
      "learning_rate": 7.964440466269349e-05,
      "loss": 0.004,
      "step": 12112
    },
    {
      "epoch": 5.6536756126021,
      "grad_norm": 0.15625,
      "learning_rate": 7.963003861324205e-05,
      "loss": 0.001,
      "step": 12113
    },
    {
      "epoch": 5.65414235705951,
      "grad_norm": 0.017578125,
      "learning_rate": 7.961567300237849e-05,
      "loss": 0.0003,
      "step": 12114
    },
    {
      "epoch": 5.65460910151692,
      "grad_norm": 0.064453125,
      "learning_rate": 7.96013078304121e-05,
      "loss": 0.0026,
      "step": 12115
    },
    {
      "epoch": 5.655075845974329,
      "grad_norm": 0.08349609375,
      "learning_rate": 7.958694309765219e-05,
      "loss": 0.0022,
      "step": 12116
    },
    {
      "epoch": 5.655542590431739,
      "grad_norm": 0.04248046875,
      "learning_rate": 7.957257880440805e-05,
      "loss": 0.0004,
      "step": 12117
    },
    {
      "epoch": 5.6560093348891485,
      "grad_norm": 0.038818359375,
      "learning_rate": 7.955821495098896e-05,
      "loss": 0.0004,
      "step": 12118
    },
    {
      "epoch": 5.656476079346557,
      "grad_norm": 0.228515625,
      "learning_rate": 7.954385153770418e-05,
      "loss": 0.0011,
      "step": 12119
    },
    {
      "epoch": 5.656942823803967,
      "grad_norm": 0.57421875,
      "learning_rate": 7.952948856486295e-05,
      "loss": 0.0068,
      "step": 12120
    },
    {
      "epoch": 5.657409568261377,
      "grad_norm": 0.06396484375,
      "learning_rate": 7.951512603277456e-05,
      "loss": 0.0005,
      "step": 12121
    },
    {
      "epoch": 5.657876312718787,
      "grad_norm": 0.033203125,
      "learning_rate": 7.950076394174825e-05,
      "loss": 0.0004,
      "step": 12122
    },
    {
      "epoch": 5.658343057176196,
      "grad_norm": 0.1435546875,
      "learning_rate": 7.948640229209321e-05,
      "loss": 0.0056,
      "step": 12123
    },
    {
      "epoch": 5.6588098016336055,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.947204108411869e-05,
      "loss": 0.0022,
      "step": 12124
    },
    {
      "epoch": 5.659276546091015,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.945768031813395e-05,
      "loss": 0.0008,
      "step": 12125
    },
    {
      "epoch": 5.659743290548425,
      "grad_norm": 0.099609375,
      "learning_rate": 7.944331999444811e-05,
      "loss": 0.0006,
      "step": 12126
    },
    {
      "epoch": 5.660210035005834,
      "grad_norm": 0.279296875,
      "learning_rate": 7.94289601133704e-05,
      "loss": 0.0016,
      "step": 12127
    },
    {
      "epoch": 5.660676779463244,
      "grad_norm": 0.0888671875,
      "learning_rate": 7.941460067520998e-05,
      "loss": 0.0007,
      "step": 12128
    },
    {
      "epoch": 5.661143523920654,
      "grad_norm": 0.10400390625,
      "learning_rate": 7.940024168027607e-05,
      "loss": 0.0005,
      "step": 12129
    },
    {
      "epoch": 5.661610268378063,
      "grad_norm": 0.1796875,
      "learning_rate": 7.938588312887782e-05,
      "loss": 0.0031,
      "step": 12130
    },
    {
      "epoch": 5.662077012835472,
      "grad_norm": 0.0478515625,
      "learning_rate": 7.937152502132437e-05,
      "loss": 0.0006,
      "step": 12131
    },
    {
      "epoch": 5.662543757292882,
      "grad_norm": 0.0299072265625,
      "learning_rate": 7.935716735792486e-05,
      "loss": 0.0004,
      "step": 12132
    },
    {
      "epoch": 5.663010501750292,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.934281013898846e-05,
      "loss": 0.0026,
      "step": 12133
    },
    {
      "epoch": 5.663477246207702,
      "grad_norm": 0.060791015625,
      "learning_rate": 7.932845336482426e-05,
      "loss": 0.0004,
      "step": 12134
    },
    {
      "epoch": 5.663943990665111,
      "grad_norm": 0.4375,
      "learning_rate": 7.93140970357414e-05,
      "loss": 0.0027,
      "step": 12135
    },
    {
      "epoch": 5.66441073512252,
      "grad_norm": 0.203125,
      "learning_rate": 7.929974115204898e-05,
      "loss": 0.0008,
      "step": 12136
    },
    {
      "epoch": 5.66487747957993,
      "grad_norm": 0.0400390625,
      "learning_rate": 7.928538571405611e-05,
      "loss": 0.0004,
      "step": 12137
    },
    {
      "epoch": 5.665344224037339,
      "grad_norm": 0.51171875,
      "learning_rate": 7.927103072207185e-05,
      "loss": 0.0022,
      "step": 12138
    },
    {
      "epoch": 5.665810968494749,
      "grad_norm": 0.1435546875,
      "learning_rate": 7.925667617640532e-05,
      "loss": 0.0046,
      "step": 12139
    },
    {
      "epoch": 5.666277712952159,
      "grad_norm": 0.2041015625,
      "learning_rate": 7.924232207736556e-05,
      "loss": 0.0084,
      "step": 12140
    },
    {
      "epoch": 5.6667444574095684,
      "grad_norm": 0.103515625,
      "learning_rate": 7.922796842526165e-05,
      "loss": 0.0042,
      "step": 12141
    },
    {
      "epoch": 5.667211201866978,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.921361522040258e-05,
      "loss": 0.0007,
      "step": 12142
    },
    {
      "epoch": 5.667677946324387,
      "grad_norm": 0.11181640625,
      "learning_rate": 7.919926246309748e-05,
      "loss": 0.0007,
      "step": 12143
    },
    {
      "epoch": 5.668144690781797,
      "grad_norm": 0.1845703125,
      "learning_rate": 7.918491015365533e-05,
      "loss": 0.0009,
      "step": 12144
    },
    {
      "epoch": 5.668611435239207,
      "grad_norm": 0.1123046875,
      "learning_rate": 7.917055829238514e-05,
      "loss": 0.0007,
      "step": 12145
    },
    {
      "epoch": 5.669078179696616,
      "grad_norm": 0.1796875,
      "learning_rate": 7.915620687959597e-05,
      "loss": 0.0025,
      "step": 12146
    },
    {
      "epoch": 5.6695449241540254,
      "grad_norm": 0.05908203125,
      "learning_rate": 7.914185591559679e-05,
      "loss": 0.0039,
      "step": 12147
    },
    {
      "epoch": 5.670011668611435,
      "grad_norm": 0.047119140625,
      "learning_rate": 7.912750540069658e-05,
      "loss": 0.0024,
      "step": 12148
    },
    {
      "epoch": 5.670478413068845,
      "grad_norm": 0.1591796875,
      "learning_rate": 7.911315533520432e-05,
      "loss": 0.001,
      "step": 12149
    },
    {
      "epoch": 5.670945157526255,
      "grad_norm": 0.244140625,
      "learning_rate": 7.909880571942904e-05,
      "loss": 0.0042,
      "step": 12150
    },
    {
      "epoch": 5.671411901983664,
      "grad_norm": 0.01226806640625,
      "learning_rate": 7.908445655367964e-05,
      "loss": 0.0002,
      "step": 12151
    },
    {
      "epoch": 5.6718786464410735,
      "grad_norm": 0.109375,
      "learning_rate": 7.907010783826509e-05,
      "loss": 0.0005,
      "step": 12152
    },
    {
      "epoch": 5.672345390898483,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.905575957349436e-05,
      "loss": 0.0007,
      "step": 12153
    },
    {
      "epoch": 5.672812135355892,
      "grad_norm": 0.06005859375,
      "learning_rate": 7.904141175967635e-05,
      "loss": 0.0006,
      "step": 12154
    },
    {
      "epoch": 5.673278879813302,
      "grad_norm": 0.357421875,
      "learning_rate": 7.902706439712e-05,
      "loss": 0.0019,
      "step": 12155
    },
    {
      "epoch": 5.673745624270712,
      "grad_norm": 0.267578125,
      "learning_rate": 7.901271748613422e-05,
      "loss": 0.0038,
      "step": 12156
    },
    {
      "epoch": 5.674212368728122,
      "grad_norm": 0.0849609375,
      "learning_rate": 7.899837102702794e-05,
      "loss": 0.0006,
      "step": 12157
    },
    {
      "epoch": 5.674679113185531,
      "grad_norm": 0.07666015625,
      "learning_rate": 7.898402502011002e-05,
      "loss": 0.0007,
      "step": 12158
    },
    {
      "epoch": 5.67514585764294,
      "grad_norm": 0.330078125,
      "learning_rate": 7.896967946568936e-05,
      "loss": 0.0019,
      "step": 12159
    },
    {
      "epoch": 5.67561260210035,
      "grad_norm": 0.080078125,
      "learning_rate": 7.895533436407483e-05,
      "loss": 0.0053,
      "step": 12160
    },
    {
      "epoch": 5.67607934655776,
      "grad_norm": 0.173828125,
      "learning_rate": 7.894098971557531e-05,
      "loss": 0.0007,
      "step": 12161
    },
    {
      "epoch": 5.676546091015169,
      "grad_norm": 0.08935546875,
      "learning_rate": 7.892664552049963e-05,
      "loss": 0.0006,
      "step": 12162
    },
    {
      "epoch": 5.677012835472579,
      "grad_norm": 0.12353515625,
      "learning_rate": 7.891230177915666e-05,
      "loss": 0.0032,
      "step": 12163
    },
    {
      "epoch": 5.677479579929988,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.889795849185524e-05,
      "loss": 0.0031,
      "step": 12164
    },
    {
      "epoch": 5.677946324387398,
      "grad_norm": 0.04833984375,
      "learning_rate": 7.888361565890417e-05,
      "loss": 0.0005,
      "step": 12165
    },
    {
      "epoch": 5.678413068844807,
      "grad_norm": 0.123046875,
      "learning_rate": 7.886927328061228e-05,
      "loss": 0.0005,
      "step": 12166
    },
    {
      "epoch": 5.678879813302217,
      "grad_norm": 0.0289306640625,
      "learning_rate": 7.885493135728841e-05,
      "loss": 0.0004,
      "step": 12167
    },
    {
      "epoch": 5.679346557759627,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.884058988924132e-05,
      "loss": 0.0011,
      "step": 12168
    },
    {
      "epoch": 5.6798133022170365,
      "grad_norm": 0.25,
      "learning_rate": 7.882624887677979e-05,
      "loss": 0.0012,
      "step": 12169
    },
    {
      "epoch": 5.680280046674445,
      "grad_norm": 0.119140625,
      "learning_rate": 7.881190832021264e-05,
      "loss": 0.0029,
      "step": 12170
    },
    {
      "epoch": 5.680746791131855,
      "grad_norm": 0.02734375,
      "learning_rate": 7.87975682198486e-05,
      "loss": 0.0004,
      "step": 12171
    },
    {
      "epoch": 5.681213535589265,
      "grad_norm": 0.0267333984375,
      "learning_rate": 7.878322857599641e-05,
      "loss": 0.0003,
      "step": 12172
    },
    {
      "epoch": 5.681680280046675,
      "grad_norm": 0.0234375,
      "learning_rate": 7.876888938896494e-05,
      "loss": 0.0003,
      "step": 12173
    },
    {
      "epoch": 5.682147024504084,
      "grad_norm": 0.27734375,
      "learning_rate": 7.875455065906275e-05,
      "loss": 0.0011,
      "step": 12174
    },
    {
      "epoch": 5.6826137689614935,
      "grad_norm": 0.02294921875,
      "learning_rate": 7.874021238659869e-05,
      "loss": 0.0004,
      "step": 12175
    },
    {
      "epoch": 5.683080513418903,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.872587457188144e-05,
      "loss": 0.0004,
      "step": 12176
    },
    {
      "epoch": 5.683547257876313,
      "grad_norm": 0.193359375,
      "learning_rate": 7.87115372152197e-05,
      "loss": 0.0041,
      "step": 12177
    },
    {
      "epoch": 5.684014002333722,
      "grad_norm": 0.10888671875,
      "learning_rate": 7.86972003169222e-05,
      "loss": 0.0005,
      "step": 12178
    },
    {
      "epoch": 5.684480746791132,
      "grad_norm": 0.11083984375,
      "learning_rate": 7.868286387729762e-05,
      "loss": 0.0009,
      "step": 12179
    },
    {
      "epoch": 5.684947491248542,
      "grad_norm": 0.09765625,
      "learning_rate": 7.866852789665463e-05,
      "loss": 0.0009,
      "step": 12180
    },
    {
      "epoch": 5.6854142357059505,
      "grad_norm": 0.1328125,
      "learning_rate": 7.865419237530187e-05,
      "loss": 0.0006,
      "step": 12181
    },
    {
      "epoch": 5.68588098016336,
      "grad_norm": 0.10400390625,
      "learning_rate": 7.863985731354806e-05,
      "loss": 0.0006,
      "step": 12182
    },
    {
      "epoch": 5.68634772462077,
      "grad_norm": 0.1484375,
      "learning_rate": 7.862552271170183e-05,
      "loss": 0.0023,
      "step": 12183
    },
    {
      "epoch": 5.68681446907818,
      "grad_norm": 0.1103515625,
      "learning_rate": 7.861118857007177e-05,
      "loss": 0.0005,
      "step": 12184
    },
    {
      "epoch": 5.68728121353559,
      "grad_norm": 0.07958984375,
      "learning_rate": 7.859685488896659e-05,
      "loss": 0.0007,
      "step": 12185
    },
    {
      "epoch": 5.687747957992999,
      "grad_norm": 0.09375,
      "learning_rate": 7.858252166869487e-05,
      "loss": 0.0005,
      "step": 12186
    },
    {
      "epoch": 5.688214702450408,
      "grad_norm": 0.04443359375,
      "learning_rate": 7.85681889095652e-05,
      "loss": 0.0004,
      "step": 12187
    },
    {
      "epoch": 5.688681446907818,
      "grad_norm": 0.08154296875,
      "learning_rate": 7.85538566118862e-05,
      "loss": 0.0006,
      "step": 12188
    },
    {
      "epoch": 5.689148191365227,
      "grad_norm": 0.359375,
      "learning_rate": 7.853952477596648e-05,
      "loss": 0.0012,
      "step": 12189
    },
    {
      "epoch": 5.689614935822637,
      "grad_norm": 0.130859375,
      "learning_rate": 7.852519340211462e-05,
      "loss": 0.0033,
      "step": 12190
    },
    {
      "epoch": 5.690081680280047,
      "grad_norm": 0.056396484375,
      "learning_rate": 7.851086249063912e-05,
      "loss": 0.0029,
      "step": 12191
    },
    {
      "epoch": 5.6905484247374565,
      "grad_norm": 0.07177734375,
      "learning_rate": 7.849653204184864e-05,
      "loss": 0.0007,
      "step": 12192
    },
    {
      "epoch": 5.691015169194866,
      "grad_norm": 0.09375,
      "learning_rate": 7.848220205605168e-05,
      "loss": 0.0008,
      "step": 12193
    },
    {
      "epoch": 5.691481913652275,
      "grad_norm": 0.1025390625,
      "learning_rate": 7.846787253355675e-05,
      "loss": 0.0005,
      "step": 12194
    },
    {
      "epoch": 5.691948658109685,
      "grad_norm": 0.1826171875,
      "learning_rate": 7.845354347467244e-05,
      "loss": 0.004,
      "step": 12195
    },
    {
      "epoch": 5.692415402567095,
      "grad_norm": 0.1953125,
      "learning_rate": 7.843921487970725e-05,
      "loss": 0.0035,
      "step": 12196
    },
    {
      "epoch": 5.692882147024504,
      "grad_norm": 0.2109375,
      "learning_rate": 7.842488674896968e-05,
      "loss": 0.0012,
      "step": 12197
    },
    {
      "epoch": 5.6933488914819135,
      "grad_norm": 0.30078125,
      "learning_rate": 7.841055908276822e-05,
      "loss": 0.0015,
      "step": 12198
    },
    {
      "epoch": 5.693815635939323,
      "grad_norm": 0.060302734375,
      "learning_rate": 7.83962318814114e-05,
      "loss": 0.0004,
      "step": 12199
    },
    {
      "epoch": 5.694282380396733,
      "grad_norm": 0.0194091796875,
      "learning_rate": 7.838190514520767e-05,
      "loss": 0.0002,
      "step": 12200
    },
    {
      "epoch": 5.694749124854142,
      "grad_norm": 0.05712890625,
      "learning_rate": 7.836757887446548e-05,
      "loss": 0.0005,
      "step": 12201
    },
    {
      "epoch": 5.695215869311552,
      "grad_norm": 0.158203125,
      "learning_rate": 7.835325306949335e-05,
      "loss": 0.0008,
      "step": 12202
    },
    {
      "epoch": 5.695682613768962,
      "grad_norm": 0.0615234375,
      "learning_rate": 7.833892773059971e-05,
      "loss": 0.0005,
      "step": 12203
    },
    {
      "epoch": 5.696149358226371,
      "grad_norm": 0.1220703125,
      "learning_rate": 7.832460285809296e-05,
      "loss": 0.0013,
      "step": 12204
    },
    {
      "epoch": 5.69661610268378,
      "grad_norm": 0.130859375,
      "learning_rate": 7.831027845228158e-05,
      "loss": 0.0009,
      "step": 12205
    },
    {
      "epoch": 5.69708284714119,
      "grad_norm": 0.052001953125,
      "learning_rate": 7.829595451347397e-05,
      "loss": 0.0004,
      "step": 12206
    },
    {
      "epoch": 5.6975495915986,
      "grad_norm": 0.0595703125,
      "learning_rate": 7.828163104197854e-05,
      "loss": 0.0004,
      "step": 12207
    },
    {
      "epoch": 5.69801633605601,
      "grad_norm": 0.040283203125,
      "learning_rate": 7.826730803810367e-05,
      "loss": 0.0004,
      "step": 12208
    },
    {
      "epoch": 5.698483080513419,
      "grad_norm": 0.08642578125,
      "learning_rate": 7.825298550215778e-05,
      "loss": 0.0007,
      "step": 12209
    },
    {
      "epoch": 5.698949824970828,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.823866343444926e-05,
      "loss": 0.0033,
      "step": 12210
    },
    {
      "epoch": 5.699416569428238,
      "grad_norm": 0.037841796875,
      "learning_rate": 7.822434183528642e-05,
      "loss": 0.0004,
      "step": 12211
    },
    {
      "epoch": 5.699883313885648,
      "grad_norm": 0.09130859375,
      "learning_rate": 7.821002070497769e-05,
      "loss": 0.0006,
      "step": 12212
    },
    {
      "epoch": 5.700350058343057,
      "grad_norm": 0.142578125,
      "learning_rate": 7.81957000438314e-05,
      "loss": 0.001,
      "step": 12213
    },
    {
      "epoch": 5.700816802800467,
      "grad_norm": 0.12255859375,
      "learning_rate": 7.818137985215586e-05,
      "loss": 0.001,
      "step": 12214
    },
    {
      "epoch": 5.7012835472578764,
      "grad_norm": 0.134765625,
      "learning_rate": 7.816706013025943e-05,
      "loss": 0.0006,
      "step": 12215
    },
    {
      "epoch": 5.701750291715286,
      "grad_norm": 0.068359375,
      "learning_rate": 7.815274087845042e-05,
      "loss": 0.0006,
      "step": 12216
    },
    {
      "epoch": 5.702217036172695,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.813842209703715e-05,
      "loss": 0.0008,
      "step": 12217
    },
    {
      "epoch": 5.702683780630105,
      "grad_norm": 0.42578125,
      "learning_rate": 7.812410378632787e-05,
      "loss": 0.0034,
      "step": 12218
    },
    {
      "epoch": 5.703150525087515,
      "grad_norm": 0.0888671875,
      "learning_rate": 7.810978594663093e-05,
      "loss": 0.0004,
      "step": 12219
    },
    {
      "epoch": 5.7036172695449245,
      "grad_norm": 0.21875,
      "learning_rate": 7.80954685782546e-05,
      "loss": 0.0016,
      "step": 12220
    },
    {
      "epoch": 5.7040840140023334,
      "grad_norm": 0.359375,
      "learning_rate": 7.808115168150714e-05,
      "loss": 0.002,
      "step": 12221
    },
    {
      "epoch": 5.704550758459743,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.80668352566968e-05,
      "loss": 0.001,
      "step": 12222
    },
    {
      "epoch": 5.705017502917153,
      "grad_norm": 0.22265625,
      "learning_rate": 7.80525193041318e-05,
      "loss": 0.0022,
      "step": 12223
    },
    {
      "epoch": 5.705484247374562,
      "grad_norm": 0.04296875,
      "learning_rate": 7.803820382412044e-05,
      "loss": 0.0004,
      "step": 12224
    },
    {
      "epoch": 5.705950991831972,
      "grad_norm": 0.0291748046875,
      "learning_rate": 7.802388881697092e-05,
      "loss": 0.0003,
      "step": 12225
    },
    {
      "epoch": 5.7064177362893815,
      "grad_norm": 0.31640625,
      "learning_rate": 7.800957428299142e-05,
      "loss": 0.0019,
      "step": 12226
    },
    {
      "epoch": 5.706884480746791,
      "grad_norm": 0.040771484375,
      "learning_rate": 7.799526022249022e-05,
      "loss": 0.0003,
      "step": 12227
    },
    {
      "epoch": 5.707351225204201,
      "grad_norm": 0.181640625,
      "learning_rate": 7.798094663577546e-05,
      "loss": 0.001,
      "step": 12228
    },
    {
      "epoch": 5.70781796966161,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.796663352315537e-05,
      "loss": 0.0007,
      "step": 12229
    },
    {
      "epoch": 5.70828471411902,
      "grad_norm": 0.0185546875,
      "learning_rate": 7.795232088493809e-05,
      "loss": 0.0002,
      "step": 12230
    },
    {
      "epoch": 5.70875145857643,
      "grad_norm": 0.1025390625,
      "learning_rate": 7.793800872143182e-05,
      "loss": 0.0009,
      "step": 12231
    },
    {
      "epoch": 5.7092182030338385,
      "grad_norm": 0.049072265625,
      "learning_rate": 7.792369703294469e-05,
      "loss": 0.0005,
      "step": 12232
    },
    {
      "epoch": 5.709684947491248,
      "grad_norm": 0.0927734375,
      "learning_rate": 7.790938581978484e-05,
      "loss": 0.0004,
      "step": 12233
    },
    {
      "epoch": 5.710151691948658,
      "grad_norm": 0.126953125,
      "learning_rate": 7.789507508226045e-05,
      "loss": 0.0006,
      "step": 12234
    },
    {
      "epoch": 5.710618436406068,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.788076482067961e-05,
      "loss": 0.0005,
      "step": 12235
    },
    {
      "epoch": 5.711085180863478,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.786645503535044e-05,
      "loss": 0.0037,
      "step": 12236
    },
    {
      "epoch": 5.711551925320887,
      "grad_norm": 0.0595703125,
      "learning_rate": 7.785214572658106e-05,
      "loss": 0.0004,
      "step": 12237
    },
    {
      "epoch": 5.712018669778296,
      "grad_norm": 0.205078125,
      "learning_rate": 7.783783689467957e-05,
      "loss": 0.0025,
      "step": 12238
    },
    {
      "epoch": 5.712485414235706,
      "grad_norm": 0.033447265625,
      "learning_rate": 7.782352853995402e-05,
      "loss": 0.0003,
      "step": 12239
    },
    {
      "epoch": 5.712952158693115,
      "grad_norm": 0.1669921875,
      "learning_rate": 7.78092206627125e-05,
      "loss": 0.001,
      "step": 12240
    },
    {
      "epoch": 5.713418903150525,
      "grad_norm": 0.16796875,
      "learning_rate": 7.779491326326309e-05,
      "loss": 0.001,
      "step": 12241
    },
    {
      "epoch": 5.713885647607935,
      "grad_norm": 0.040283203125,
      "learning_rate": 7.778060634191384e-05,
      "loss": 0.0004,
      "step": 12242
    },
    {
      "epoch": 5.7143523920653445,
      "grad_norm": 0.1240234375,
      "learning_rate": 7.776629989897276e-05,
      "loss": 0.0006,
      "step": 12243
    },
    {
      "epoch": 5.714819136522753,
      "grad_norm": 0.05712890625,
      "learning_rate": 7.775199393474795e-05,
      "loss": 0.0004,
      "step": 12244
    },
    {
      "epoch": 5.715285880980163,
      "grad_norm": 0.037353515625,
      "learning_rate": 7.773768844954737e-05,
      "loss": 0.0003,
      "step": 12245
    },
    {
      "epoch": 5.715752625437573,
      "grad_norm": 0.07080078125,
      "learning_rate": 7.772338344367907e-05,
      "loss": 0.0004,
      "step": 12246
    },
    {
      "epoch": 5.716219369894983,
      "grad_norm": 0.0693359375,
      "learning_rate": 7.770907891745102e-05,
      "loss": 0.0005,
      "step": 12247
    },
    {
      "epoch": 5.716686114352392,
      "grad_norm": 0.0262451171875,
      "learning_rate": 7.769477487117127e-05,
      "loss": 0.0003,
      "step": 12248
    },
    {
      "epoch": 5.7171528588098015,
      "grad_norm": 0.17578125,
      "learning_rate": 7.768047130514776e-05,
      "loss": 0.0006,
      "step": 12249
    },
    {
      "epoch": 5.717619603267211,
      "grad_norm": 0.1328125,
      "learning_rate": 7.766616821968843e-05,
      "loss": 0.0028,
      "step": 12250
    },
    {
      "epoch": 5.718086347724621,
      "grad_norm": 0.083984375,
      "learning_rate": 7.76518656151013e-05,
      "loss": 0.0006,
      "step": 12251
    },
    {
      "epoch": 5.71855309218203,
      "grad_norm": 0.12451171875,
      "learning_rate": 7.763756349169431e-05,
      "loss": 0.0007,
      "step": 12252
    },
    {
      "epoch": 5.71901983663944,
      "grad_norm": 0.07080078125,
      "learning_rate": 7.762326184977537e-05,
      "loss": 0.0004,
      "step": 12253
    },
    {
      "epoch": 5.71948658109685,
      "grad_norm": 0.052001953125,
      "learning_rate": 7.760896068965244e-05,
      "loss": 0.0005,
      "step": 12254
    },
    {
      "epoch": 5.719953325554259,
      "grad_norm": 0.171875,
      "learning_rate": 7.759466001163343e-05,
      "loss": 0.0008,
      "step": 12255
    },
    {
      "epoch": 5.720420070011668,
      "grad_norm": 0.0634765625,
      "learning_rate": 7.758035981602626e-05,
      "loss": 0.0031,
      "step": 12256
    },
    {
      "epoch": 5.720886814469078,
      "grad_norm": 0.154296875,
      "learning_rate": 7.75660601031388e-05,
      "loss": 0.0009,
      "step": 12257
    },
    {
      "epoch": 5.721353558926488,
      "grad_norm": 0.056884765625,
      "learning_rate": 7.755176087327895e-05,
      "loss": 0.0005,
      "step": 12258
    },
    {
      "epoch": 5.721820303383898,
      "grad_norm": 0.263671875,
      "learning_rate": 7.753746212675463e-05,
      "loss": 0.001,
      "step": 12259
    },
    {
      "epoch": 5.722287047841307,
      "grad_norm": 0.083984375,
      "learning_rate": 7.752316386387362e-05,
      "loss": 0.0007,
      "step": 12260
    },
    {
      "epoch": 5.722753792298716,
      "grad_norm": 0.0179443359375,
      "learning_rate": 7.750886608494388e-05,
      "loss": 0.0003,
      "step": 12261
    },
    {
      "epoch": 5.723220536756126,
      "grad_norm": 0.0947265625,
      "learning_rate": 7.74945687902732e-05,
      "loss": 0.0005,
      "step": 12262
    },
    {
      "epoch": 5.723687281213536,
      "grad_norm": 0.044189453125,
      "learning_rate": 7.74802719801694e-05,
      "loss": 0.0005,
      "step": 12263
    },
    {
      "epoch": 5.724154025670945,
      "grad_norm": 0.0478515625,
      "learning_rate": 7.746597565494033e-05,
      "loss": 0.0004,
      "step": 12264
    },
    {
      "epoch": 5.724620770128355,
      "grad_norm": 0.053955078125,
      "learning_rate": 7.745167981489385e-05,
      "loss": 0.0005,
      "step": 12265
    },
    {
      "epoch": 5.7250875145857645,
      "grad_norm": 0.267578125,
      "learning_rate": 7.743738446033768e-05,
      "loss": 0.0016,
      "step": 12266
    },
    {
      "epoch": 5.725554259043173,
      "grad_norm": 0.1513671875,
      "learning_rate": 7.742308959157966e-05,
      "loss": 0.0005,
      "step": 12267
    },
    {
      "epoch": 5.726021003500583,
      "grad_norm": 0.1103515625,
      "learning_rate": 7.740879520892758e-05,
      "loss": 0.0049,
      "step": 12268
    },
    {
      "epoch": 5.726487747957993,
      "grad_norm": 0.2294921875,
      "learning_rate": 7.739450131268922e-05,
      "loss": 0.0035,
      "step": 12269
    },
    {
      "epoch": 5.726954492415403,
      "grad_norm": 0.236328125,
      "learning_rate": 7.738020790317233e-05,
      "loss": 0.0014,
      "step": 12270
    },
    {
      "epoch": 5.727421236872813,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.736591498068465e-05,
      "loss": 0.0007,
      "step": 12271
    },
    {
      "epoch": 5.7278879813302215,
      "grad_norm": 0.08447265625,
      "learning_rate": 7.735162254553391e-05,
      "loss": 0.0084,
      "step": 12272
    },
    {
      "epoch": 5.728354725787631,
      "grad_norm": 0.08154296875,
      "learning_rate": 7.733733059802788e-05,
      "loss": 0.0032,
      "step": 12273
    },
    {
      "epoch": 5.728821470245041,
      "grad_norm": 0.0791015625,
      "learning_rate": 7.732303913847427e-05,
      "loss": 0.0004,
      "step": 12274
    },
    {
      "epoch": 5.72928821470245,
      "grad_norm": 0.03466796875,
      "learning_rate": 7.730874816718078e-05,
      "loss": 0.0004,
      "step": 12275
    },
    {
      "epoch": 5.72975495915986,
      "grad_norm": 0.212890625,
      "learning_rate": 7.729445768445512e-05,
      "loss": 0.0008,
      "step": 12276
    },
    {
      "epoch": 5.73022170361727,
      "grad_norm": 0.375,
      "learning_rate": 7.728016769060499e-05,
      "loss": 0.0014,
      "step": 12277
    },
    {
      "epoch": 5.730688448074679,
      "grad_norm": 0.09423828125,
      "learning_rate": 7.726587818593806e-05,
      "loss": 0.0006,
      "step": 12278
    },
    {
      "epoch": 5.731155192532089,
      "grad_norm": 0.0206298828125,
      "learning_rate": 7.725158917076196e-05,
      "loss": 0.0005,
      "step": 12279
    },
    {
      "epoch": 5.731621936989498,
      "grad_norm": 0.171875,
      "learning_rate": 7.723730064538442e-05,
      "loss": 0.0012,
      "step": 12280
    },
    {
      "epoch": 5.732088681446908,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.722301261011306e-05,
      "loss": 0.001,
      "step": 12281
    },
    {
      "epoch": 5.732555425904318,
      "grad_norm": 0.020751953125,
      "learning_rate": 7.720872506525547e-05,
      "loss": 0.0003,
      "step": 12282
    },
    {
      "epoch": 5.733022170361727,
      "grad_norm": 0.00836181640625,
      "learning_rate": 7.719443801111935e-05,
      "loss": 0.0002,
      "step": 12283
    },
    {
      "epoch": 5.733488914819136,
      "grad_norm": 0.0242919921875,
      "learning_rate": 7.718015144801227e-05,
      "loss": 0.0003,
      "step": 12284
    },
    {
      "epoch": 5.733955659276546,
      "grad_norm": 0.05322265625,
      "learning_rate": 7.716586537624184e-05,
      "loss": 0.0005,
      "step": 12285
    },
    {
      "epoch": 5.734422403733956,
      "grad_norm": 0.07275390625,
      "learning_rate": 7.71515797961157e-05,
      "loss": 0.0006,
      "step": 12286
    },
    {
      "epoch": 5.734889148191365,
      "grad_norm": 0.0517578125,
      "learning_rate": 7.713729470794136e-05,
      "loss": 0.0003,
      "step": 12287
    },
    {
      "epoch": 5.735355892648775,
      "grad_norm": 0.205078125,
      "learning_rate": 7.712301011202645e-05,
      "loss": 0.0052,
      "step": 12288
    },
    {
      "epoch": 5.7358226371061845,
      "grad_norm": 0.034912109375,
      "learning_rate": 7.71087260086785e-05,
      "loss": 0.0004,
      "step": 12289
    },
    {
      "epoch": 5.736289381563594,
      "grad_norm": 0.064453125,
      "learning_rate": 7.70944423982051e-05,
      "loss": 0.0006,
      "step": 12290
    },
    {
      "epoch": 5.736756126021003,
      "grad_norm": 0.1396484375,
      "learning_rate": 7.708015928091375e-05,
      "loss": 0.0011,
      "step": 12291
    },
    {
      "epoch": 5.737222870478413,
      "grad_norm": 0.0233154296875,
      "learning_rate": 7.7065876657112e-05,
      "loss": 0.0003,
      "step": 12292
    },
    {
      "epoch": 5.737689614935823,
      "grad_norm": 0.03369140625,
      "learning_rate": 7.705159452710737e-05,
      "loss": 0.0004,
      "step": 12293
    },
    {
      "epoch": 5.7381563593932325,
      "grad_norm": 0.05615234375,
      "learning_rate": 7.70373128912074e-05,
      "loss": 0.0005,
      "step": 12294
    },
    {
      "epoch": 5.7386231038506414,
      "grad_norm": 0.083984375,
      "learning_rate": 7.702303174971952e-05,
      "loss": 0.0006,
      "step": 12295
    },
    {
      "epoch": 5.739089848308051,
      "grad_norm": 0.05029296875,
      "learning_rate": 7.700875110295128e-05,
      "loss": 0.0003,
      "step": 12296
    },
    {
      "epoch": 5.739556592765461,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.699447095121015e-05,
      "loss": 0.0026,
      "step": 12297
    },
    {
      "epoch": 5.740023337222871,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.698019129480358e-05,
      "loss": 0.0005,
      "step": 12298
    },
    {
      "epoch": 5.74049008168028,
      "grad_norm": 0.031494140625,
      "learning_rate": 7.696591213403901e-05,
      "loss": 0.0004,
      "step": 12299
    },
    {
      "epoch": 5.7409568261376895,
      "grad_norm": 0.1396484375,
      "learning_rate": 7.695163346922392e-05,
      "loss": 0.0027,
      "step": 12300
    },
    {
      "epoch": 5.741423570595099,
      "grad_norm": 0.1220703125,
      "learning_rate": 7.693735530066575e-05,
      "loss": 0.0007,
      "step": 12301
    },
    {
      "epoch": 5.741890315052509,
      "grad_norm": 0.32421875,
      "learning_rate": 7.692307762867188e-05,
      "loss": 0.0032,
      "step": 12302
    },
    {
      "epoch": 5.742357059509918,
      "grad_norm": 0.1259765625,
      "learning_rate": 7.690880045354977e-05,
      "loss": 0.0005,
      "step": 12303
    },
    {
      "epoch": 5.742823803967328,
      "grad_norm": 0.10888671875,
      "learning_rate": 7.689452377560682e-05,
      "loss": 0.0006,
      "step": 12304
    },
    {
      "epoch": 5.743290548424738,
      "grad_norm": 0.0712890625,
      "learning_rate": 7.688024759515038e-05,
      "loss": 0.0029,
      "step": 12305
    },
    {
      "epoch": 5.743757292882147,
      "grad_norm": 0.032958984375,
      "learning_rate": 7.686597191248789e-05,
      "loss": 0.0004,
      "step": 12306
    },
    {
      "epoch": 5.744224037339556,
      "grad_norm": 0.037841796875,
      "learning_rate": 7.685169672792668e-05,
      "loss": 0.0004,
      "step": 12307
    },
    {
      "epoch": 5.744690781796966,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.683742204177414e-05,
      "loss": 0.0032,
      "step": 12308
    },
    {
      "epoch": 5.745157526254376,
      "grad_norm": 0.0498046875,
      "learning_rate": 7.682314785433756e-05,
      "loss": 0.0003,
      "step": 12309
    },
    {
      "epoch": 5.745624270711785,
      "grad_norm": 0.0162353515625,
      "learning_rate": 7.680887416592435e-05,
      "loss": 0.0003,
      "step": 12310
    },
    {
      "epoch": 5.746091015169195,
      "grad_norm": 0.130859375,
      "learning_rate": 7.679460097684182e-05,
      "loss": 0.0012,
      "step": 12311
    },
    {
      "epoch": 5.746557759626604,
      "grad_norm": 0.0390625,
      "learning_rate": 7.678032828739725e-05,
      "loss": 0.0023,
      "step": 12312
    },
    {
      "epoch": 5.747024504084014,
      "grad_norm": 0.040771484375,
      "learning_rate": 7.676605609789799e-05,
      "loss": 0.0003,
      "step": 12313
    },
    {
      "epoch": 5.747491248541424,
      "grad_norm": 0.056640625,
      "learning_rate": 7.675178440865133e-05,
      "loss": 0.0032,
      "step": 12314
    },
    {
      "epoch": 5.747957992998833,
      "grad_norm": 0.06396484375,
      "learning_rate": 7.673751321996454e-05,
      "loss": 0.0035,
      "step": 12315
    },
    {
      "epoch": 5.748424737456243,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.672324253214492e-05,
      "loss": 0.0006,
      "step": 12316
    },
    {
      "epoch": 5.7488914819136525,
      "grad_norm": 0.03955078125,
      "learning_rate": 7.670897234549967e-05,
      "loss": 0.0004,
      "step": 12317
    },
    {
      "epoch": 5.749358226371061,
      "grad_norm": 0.09521484375,
      "learning_rate": 7.669470266033612e-05,
      "loss": 0.003,
      "step": 12318
    },
    {
      "epoch": 5.749824970828471,
      "grad_norm": 0.10986328125,
      "learning_rate": 7.668043347696148e-05,
      "loss": 0.0007,
      "step": 12319
    },
    {
      "epoch": 5.750291715285881,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.666616479568297e-05,
      "loss": 0.0021,
      "step": 12320
    },
    {
      "epoch": 5.750758459743291,
      "grad_norm": 0.0311279296875,
      "learning_rate": 7.66518966168078e-05,
      "loss": 0.0003,
      "step": 12321
    },
    {
      "epoch": 5.751225204200701,
      "grad_norm": 0.0289306640625,
      "learning_rate": 7.663762894064324e-05,
      "loss": 0.0004,
      "step": 12322
    },
    {
      "epoch": 5.7516919486581095,
      "grad_norm": 0.06005859375,
      "learning_rate": 7.662336176749643e-05,
      "loss": 0.003,
      "step": 12323
    },
    {
      "epoch": 5.752158693115519,
      "grad_norm": 0.018310546875,
      "learning_rate": 7.660909509767455e-05,
      "loss": 0.0003,
      "step": 12324
    },
    {
      "epoch": 5.752625437572929,
      "grad_norm": 0.0247802734375,
      "learning_rate": 7.659482893148485e-05,
      "loss": 0.0004,
      "step": 12325
    },
    {
      "epoch": 5.753092182030338,
      "grad_norm": 0.020263671875,
      "learning_rate": 7.658056326923444e-05,
      "loss": 0.0003,
      "step": 12326
    },
    {
      "epoch": 5.753558926487748,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.65662981112305e-05,
      "loss": 0.0004,
      "step": 12327
    },
    {
      "epoch": 5.754025670945158,
      "grad_norm": 0.318359375,
      "learning_rate": 7.655203345778012e-05,
      "loss": 0.0055,
      "step": 12328
    },
    {
      "epoch": 5.754025670945158,
      "eval_loss": 2.309670925140381,
      "eval_runtime": 55.4413,
      "eval_samples_per_second": 32.539,
      "eval_steps_per_second": 4.076,
      "step": 12328
    },
    {
      "epoch": 5.754492415402567,
      "grad_norm": 0.439453125,
      "learning_rate": 7.65377693091905e-05,
      "loss": 0.0036,
      "step": 12329
    },
    {
      "epoch": 5.754959159859976,
      "grad_norm": 0.018310546875,
      "learning_rate": 7.652350566576875e-05,
      "loss": 0.0003,
      "step": 12330
    },
    {
      "epoch": 5.755425904317386,
      "grad_norm": 0.0250244140625,
      "learning_rate": 7.650924252782194e-05,
      "loss": 0.0003,
      "step": 12331
    },
    {
      "epoch": 5.755892648774796,
      "grad_norm": 0.059814453125,
      "learning_rate": 7.649497989565723e-05,
      "loss": 0.0005,
      "step": 12332
    },
    {
      "epoch": 5.756359393232206,
      "grad_norm": 0.208984375,
      "learning_rate": 7.648071776958167e-05,
      "loss": 0.0006,
      "step": 12333
    },
    {
      "epoch": 5.756826137689615,
      "grad_norm": 0.11865234375,
      "learning_rate": 7.646645614990235e-05,
      "loss": 0.0006,
      "step": 12334
    },
    {
      "epoch": 5.757292882147024,
      "grad_norm": 0.031982421875,
      "learning_rate": 7.645219503692635e-05,
      "loss": 0.0003,
      "step": 12335
    },
    {
      "epoch": 5.757759626604434,
      "grad_norm": 0.057373046875,
      "learning_rate": 7.643793443096072e-05,
      "loss": 0.0004,
      "step": 12336
    },
    {
      "epoch": 5.758226371061844,
      "grad_norm": 0.046875,
      "learning_rate": 7.64236743323125e-05,
      "loss": 0.0005,
      "step": 12337
    },
    {
      "epoch": 5.758693115519253,
      "grad_norm": 0.049560546875,
      "learning_rate": 7.640941474128871e-05,
      "loss": 0.0004,
      "step": 12338
    },
    {
      "epoch": 5.759159859976663,
      "grad_norm": 0.07080078125,
      "learning_rate": 7.639515565819642e-05,
      "loss": 0.0007,
      "step": 12339
    },
    {
      "epoch": 5.7596266044340725,
      "grad_norm": 0.08349609375,
      "learning_rate": 7.638089708334261e-05,
      "loss": 0.0007,
      "step": 12340
    },
    {
      "epoch": 5.760093348891482,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.636663901703428e-05,
      "loss": 0.0006,
      "step": 12341
    },
    {
      "epoch": 5.760560093348891,
      "grad_norm": 0.037109375,
      "learning_rate": 7.635238145957845e-05,
      "loss": 0.0003,
      "step": 12342
    },
    {
      "epoch": 5.761026837806301,
      "grad_norm": 0.10546875,
      "learning_rate": 7.633812441128208e-05,
      "loss": 0.0005,
      "step": 12343
    },
    {
      "epoch": 5.761493582263711,
      "grad_norm": 0.06982421875,
      "learning_rate": 7.632386787245213e-05,
      "loss": 0.0006,
      "step": 12344
    },
    {
      "epoch": 5.761960326721121,
      "grad_norm": 0.17578125,
      "learning_rate": 7.63096118433956e-05,
      "loss": 0.001,
      "step": 12345
    },
    {
      "epoch": 5.7624270711785295,
      "grad_norm": 0.027099609375,
      "learning_rate": 7.62953563244194e-05,
      "loss": 0.0003,
      "step": 12346
    },
    {
      "epoch": 5.762893815635939,
      "grad_norm": 0.05712890625,
      "learning_rate": 7.628110131583049e-05,
      "loss": 0.0005,
      "step": 12347
    },
    {
      "epoch": 5.763360560093349,
      "grad_norm": 0.03515625,
      "learning_rate": 7.626684681793577e-05,
      "loss": 0.0004,
      "step": 12348
    },
    {
      "epoch": 5.763827304550759,
      "grad_norm": 0.140625,
      "learning_rate": 7.625259283104218e-05,
      "loss": 0.0006,
      "step": 12349
    },
    {
      "epoch": 5.764294049008168,
      "grad_norm": 0.08251953125,
      "learning_rate": 7.623833935545662e-05,
      "loss": 0.0031,
      "step": 12350
    },
    {
      "epoch": 5.764760793465578,
      "grad_norm": 0.0361328125,
      "learning_rate": 7.622408639148593e-05,
      "loss": 0.0005,
      "step": 12351
    },
    {
      "epoch": 5.765227537922987,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.620983393943709e-05,
      "loss": 0.0033,
      "step": 12352
    },
    {
      "epoch": 5.765694282380396,
      "grad_norm": 0.2177734375,
      "learning_rate": 7.619558199961691e-05,
      "loss": 0.005,
      "step": 12353
    },
    {
      "epoch": 5.766161026837806,
      "grad_norm": 0.0849609375,
      "learning_rate": 7.618133057233225e-05,
      "loss": 0.0006,
      "step": 12354
    },
    {
      "epoch": 5.766627771295216,
      "grad_norm": 0.09130859375,
      "learning_rate": 7.616707965788999e-05,
      "loss": 0.0005,
      "step": 12355
    },
    {
      "epoch": 5.767094515752626,
      "grad_norm": 0.0194091796875,
      "learning_rate": 7.615282925659694e-05,
      "loss": 0.0003,
      "step": 12356
    },
    {
      "epoch": 5.7675612602100355,
      "grad_norm": 0.09033203125,
      "learning_rate": 7.613857936875994e-05,
      "loss": 0.0039,
      "step": 12357
    },
    {
      "epoch": 5.768028004667444,
      "grad_norm": 0.0223388671875,
      "learning_rate": 7.612432999468577e-05,
      "loss": 0.0002,
      "step": 12358
    },
    {
      "epoch": 5.768494749124854,
      "grad_norm": 0.08837890625,
      "learning_rate": 7.61100811346813e-05,
      "loss": 0.0006,
      "step": 12359
    },
    {
      "epoch": 5.768961493582264,
      "grad_norm": 0.0250244140625,
      "learning_rate": 7.609583278905328e-05,
      "loss": 0.0003,
      "step": 12360
    },
    {
      "epoch": 5.769428238039673,
      "grad_norm": 0.037109375,
      "learning_rate": 7.608158495810849e-05,
      "loss": 0.0003,
      "step": 12361
    },
    {
      "epoch": 5.769894982497083,
      "grad_norm": 0.03466796875,
      "learning_rate": 7.606733764215373e-05,
      "loss": 0.0004,
      "step": 12362
    },
    {
      "epoch": 5.7703617269544925,
      "grad_norm": 0.0277099609375,
      "learning_rate": 7.605309084149574e-05,
      "loss": 0.0003,
      "step": 12363
    },
    {
      "epoch": 5.770828471411902,
      "grad_norm": 0.08203125,
      "learning_rate": 7.603884455644131e-05,
      "loss": 0.0005,
      "step": 12364
    },
    {
      "epoch": 5.771295215869312,
      "grad_norm": 0.2353515625,
      "learning_rate": 7.60245987872971e-05,
      "loss": 0.0011,
      "step": 12365
    },
    {
      "epoch": 5.771761960326721,
      "grad_norm": 0.03515625,
      "learning_rate": 7.601035353436987e-05,
      "loss": 0.0003,
      "step": 12366
    },
    {
      "epoch": 5.772228704784131,
      "grad_norm": 0.091796875,
      "learning_rate": 7.599610879796637e-05,
      "loss": 0.0026,
      "step": 12367
    },
    {
      "epoch": 5.7726954492415405,
      "grad_norm": 0.15625,
      "learning_rate": 7.598186457839327e-05,
      "loss": 0.0007,
      "step": 12368
    },
    {
      "epoch": 5.7731621936989495,
      "grad_norm": 0.06689453125,
      "learning_rate": 7.596762087595728e-05,
      "loss": 0.0004,
      "step": 12369
    },
    {
      "epoch": 5.773628938156359,
      "grad_norm": 0.021728515625,
      "learning_rate": 7.595337769096506e-05,
      "loss": 0.0003,
      "step": 12370
    },
    {
      "epoch": 5.774095682613769,
      "grad_norm": 0.0517578125,
      "learning_rate": 7.593913502372331e-05,
      "loss": 0.0004,
      "step": 12371
    },
    {
      "epoch": 5.774562427071179,
      "grad_norm": 0.1552734375,
      "learning_rate": 7.592489287453868e-05,
      "loss": 0.0008,
      "step": 12372
    },
    {
      "epoch": 5.775029171528588,
      "grad_norm": 0.357421875,
      "learning_rate": 7.591065124371782e-05,
      "loss": 0.0021,
      "step": 12373
    },
    {
      "epoch": 5.7754959159859975,
      "grad_norm": 0.03759765625,
      "learning_rate": 7.589641013156736e-05,
      "loss": 0.0004,
      "step": 12374
    },
    {
      "epoch": 5.775962660443407,
      "grad_norm": 0.09375,
      "learning_rate": 7.588216953839395e-05,
      "loss": 0.0005,
      "step": 12375
    },
    {
      "epoch": 5.776429404900817,
      "grad_norm": 0.053955078125,
      "learning_rate": 7.586792946450417e-05,
      "loss": 0.0005,
      "step": 12376
    },
    {
      "epoch": 5.776896149358226,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.585368991020463e-05,
      "loss": 0.0008,
      "step": 12377
    },
    {
      "epoch": 5.777362893815636,
      "grad_norm": 0.01336669921875,
      "learning_rate": 7.583945087580196e-05,
      "loss": 0.0003,
      "step": 12378
    },
    {
      "epoch": 5.777829638273046,
      "grad_norm": 0.0458984375,
      "learning_rate": 7.582521236160272e-05,
      "loss": 0.0022,
      "step": 12379
    },
    {
      "epoch": 5.778296382730455,
      "grad_norm": 0.0213623046875,
      "learning_rate": 7.581097436791347e-05,
      "loss": 0.0002,
      "step": 12380
    },
    {
      "epoch": 5.778763127187864,
      "grad_norm": 0.035400390625,
      "learning_rate": 7.579673689504079e-05,
      "loss": 0.0012,
      "step": 12381
    },
    {
      "epoch": 5.779229871645274,
      "grad_norm": 0.216796875,
      "learning_rate": 7.578249994329122e-05,
      "loss": 0.0006,
      "step": 12382
    },
    {
      "epoch": 5.779696616102684,
      "grad_norm": 0.06591796875,
      "learning_rate": 7.576826351297127e-05,
      "loss": 0.0004,
      "step": 12383
    },
    {
      "epoch": 5.780163360560094,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.575402760438752e-05,
      "loss": 0.0005,
      "step": 12384
    },
    {
      "epoch": 5.780630105017503,
      "grad_norm": 0.07861328125,
      "learning_rate": 7.573979221784645e-05,
      "loss": 0.0007,
      "step": 12385
    },
    {
      "epoch": 5.781096849474912,
      "grad_norm": 0.0400390625,
      "learning_rate": 7.572555735365457e-05,
      "loss": 0.0003,
      "step": 12386
    },
    {
      "epoch": 5.781563593932322,
      "grad_norm": 0.09228515625,
      "learning_rate": 7.571132301211836e-05,
      "loss": 0.0006,
      "step": 12387
    },
    {
      "epoch": 5.782030338389731,
      "grad_norm": 0.0113525390625,
      "learning_rate": 7.569708919354433e-05,
      "loss": 0.0002,
      "step": 12388
    },
    {
      "epoch": 5.782497082847141,
      "grad_norm": 0.041259765625,
      "learning_rate": 7.568285589823895e-05,
      "loss": 0.0003,
      "step": 12389
    },
    {
      "epoch": 5.782963827304551,
      "grad_norm": 0.0294189453125,
      "learning_rate": 7.566862312650862e-05,
      "loss": 0.0021,
      "step": 12390
    },
    {
      "epoch": 5.7834305717619605,
      "grad_norm": 0.1884765625,
      "learning_rate": 7.565439087865986e-05,
      "loss": 0.0006,
      "step": 12391
    },
    {
      "epoch": 5.78389731621937,
      "grad_norm": 0.16015625,
      "learning_rate": 7.564015915499908e-05,
      "loss": 0.0034,
      "step": 12392
    },
    {
      "epoch": 5.784364060676779,
      "grad_norm": 0.032470703125,
      "learning_rate": 7.562592795583268e-05,
      "loss": 0.0004,
      "step": 12393
    },
    {
      "epoch": 5.784830805134189,
      "grad_norm": 0.0478515625,
      "learning_rate": 7.561169728146711e-05,
      "loss": 0.0004,
      "step": 12394
    },
    {
      "epoch": 5.785297549591599,
      "grad_norm": 0.021484375,
      "learning_rate": 7.559746713220877e-05,
      "loss": 0.0004,
      "step": 12395
    },
    {
      "epoch": 5.785764294049008,
      "grad_norm": 0.037353515625,
      "learning_rate": 7.558323750836401e-05,
      "loss": 0.0003,
      "step": 12396
    },
    {
      "epoch": 5.7862310385064175,
      "grad_norm": 0.0390625,
      "learning_rate": 7.556900841023924e-05,
      "loss": 0.0004,
      "step": 12397
    },
    {
      "epoch": 5.786697782963827,
      "grad_norm": 0.123046875,
      "learning_rate": 7.555477983814083e-05,
      "loss": 0.0006,
      "step": 12398
    },
    {
      "epoch": 5.787164527421237,
      "grad_norm": 0.09521484375,
      "learning_rate": 7.554055179237515e-05,
      "loss": 0.0024,
      "step": 12399
    },
    {
      "epoch": 5.787631271878647,
      "grad_norm": 0.03662109375,
      "learning_rate": 7.55263242732485e-05,
      "loss": 0.0003,
      "step": 12400
    },
    {
      "epoch": 5.788098016336056,
      "grad_norm": 0.0172119140625,
      "learning_rate": 7.551209728106727e-05,
      "loss": 0.0002,
      "step": 12401
    },
    {
      "epoch": 5.788564760793466,
      "grad_norm": 0.0703125,
      "learning_rate": 7.549787081613774e-05,
      "loss": 0.0005,
      "step": 12402
    },
    {
      "epoch": 5.789031505250875,
      "grad_norm": 0.15625,
      "learning_rate": 7.548364487876623e-05,
      "loss": 0.0008,
      "step": 12403
    },
    {
      "epoch": 5.789498249708284,
      "grad_norm": 0.03564453125,
      "learning_rate": 7.546941946925906e-05,
      "loss": 0.0003,
      "step": 12404
    },
    {
      "epoch": 5.789964994165694,
      "grad_norm": 0.0303955078125,
      "learning_rate": 7.54551945879225e-05,
      "loss": 0.0023,
      "step": 12405
    },
    {
      "epoch": 5.790431738623104,
      "grad_norm": 0.06396484375,
      "learning_rate": 7.544097023506285e-05,
      "loss": 0.0004,
      "step": 12406
    },
    {
      "epoch": 5.790898483080514,
      "grad_norm": 0.072265625,
      "learning_rate": 7.542674641098633e-05,
      "loss": 0.0006,
      "step": 12407
    },
    {
      "epoch": 5.791365227537923,
      "grad_norm": 0.921875,
      "learning_rate": 7.541252311599925e-05,
      "loss": 0.0024,
      "step": 12408
    },
    {
      "epoch": 5.791831971995332,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.539830035040782e-05,
      "loss": 0.0006,
      "step": 12409
    },
    {
      "epoch": 5.792298716452742,
      "grad_norm": 0.0263671875,
      "learning_rate": 7.538407811451825e-05,
      "loss": 0.0003,
      "step": 12410
    },
    {
      "epoch": 5.792765460910152,
      "grad_norm": 0.11474609375,
      "learning_rate": 7.536985640863683e-05,
      "loss": 0.0004,
      "step": 12411
    },
    {
      "epoch": 5.793232205367561,
      "grad_norm": 0.04833984375,
      "learning_rate": 7.535563523306974e-05,
      "loss": 0.0004,
      "step": 12412
    },
    {
      "epoch": 5.793698949824971,
      "grad_norm": 0.27734375,
      "learning_rate": 7.534141458812315e-05,
      "loss": 0.0011,
      "step": 12413
    },
    {
      "epoch": 5.7941656942823805,
      "grad_norm": 0.166015625,
      "learning_rate": 7.532719447410326e-05,
      "loss": 0.0013,
      "step": 12414
    },
    {
      "epoch": 5.79463243873979,
      "grad_norm": 0.046875,
      "learning_rate": 7.531297489131623e-05,
      "loss": 0.0003,
      "step": 12415
    },
    {
      "epoch": 5.795099183197199,
      "grad_norm": 0.035400390625,
      "learning_rate": 7.529875584006826e-05,
      "loss": 0.0003,
      "step": 12416
    },
    {
      "epoch": 5.795565927654609,
      "grad_norm": 0.07421875,
      "learning_rate": 7.528453732066548e-05,
      "loss": 0.0005,
      "step": 12417
    },
    {
      "epoch": 5.796032672112019,
      "grad_norm": 0.181640625,
      "learning_rate": 7.527031933341404e-05,
      "loss": 0.0006,
      "step": 12418
    },
    {
      "epoch": 5.796499416569429,
      "grad_norm": 0.0859375,
      "learning_rate": 7.525610187862004e-05,
      "loss": 0.0005,
      "step": 12419
    },
    {
      "epoch": 5.7969661610268375,
      "grad_norm": 0.0390625,
      "learning_rate": 7.524188495658964e-05,
      "loss": 0.0005,
      "step": 12420
    },
    {
      "epoch": 5.797432905484247,
      "grad_norm": 0.3515625,
      "learning_rate": 7.522766856762893e-05,
      "loss": 0.0011,
      "step": 12421
    },
    {
      "epoch": 5.797899649941657,
      "grad_norm": 0.072265625,
      "learning_rate": 7.521345271204397e-05,
      "loss": 0.0033,
      "step": 12422
    },
    {
      "epoch": 5.798366394399067,
      "grad_norm": 0.0498046875,
      "learning_rate": 7.51992373901409e-05,
      "loss": 0.0027,
      "step": 12423
    },
    {
      "epoch": 5.798833138856476,
      "grad_norm": 0.0242919921875,
      "learning_rate": 7.518502260222577e-05,
      "loss": 0.0003,
      "step": 12424
    },
    {
      "epoch": 5.799299883313886,
      "grad_norm": 0.0284423828125,
      "learning_rate": 7.517080834860462e-05,
      "loss": 0.0003,
      "step": 12425
    },
    {
      "epoch": 5.799766627771295,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.515659462958354e-05,
      "loss": 0.0007,
      "step": 12426
    },
    {
      "epoch": 5.800233372228705,
      "grad_norm": 0.049072265625,
      "learning_rate": 7.514238144546852e-05,
      "loss": 0.0004,
      "step": 12427
    },
    {
      "epoch": 5.800700116686114,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.512816879656562e-05,
      "loss": 0.0039,
      "step": 12428
    },
    {
      "epoch": 5.801166861143524,
      "grad_norm": 0.109375,
      "learning_rate": 7.511395668318082e-05,
      "loss": 0.0005,
      "step": 12429
    },
    {
      "epoch": 5.801633605600934,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.509974510562017e-05,
      "loss": 0.0007,
      "step": 12430
    },
    {
      "epoch": 5.802100350058343,
      "grad_norm": 0.087890625,
      "learning_rate": 7.508553406418962e-05,
      "loss": 0.0007,
      "step": 12431
    },
    {
      "epoch": 5.802567094515752,
      "grad_norm": 0.0615234375,
      "learning_rate": 7.507132355919516e-05,
      "loss": 0.0043,
      "step": 12432
    },
    {
      "epoch": 5.803033838973162,
      "grad_norm": 0.041748046875,
      "learning_rate": 7.505711359094277e-05,
      "loss": 0.0003,
      "step": 12433
    },
    {
      "epoch": 5.803500583430572,
      "grad_norm": 0.0771484375,
      "learning_rate": 7.504290415973842e-05,
      "loss": 0.0005,
      "step": 12434
    },
    {
      "epoch": 5.803967327887982,
      "grad_norm": 0.068359375,
      "learning_rate": 7.502869526588802e-05,
      "loss": 0.0023,
      "step": 12435
    },
    {
      "epoch": 5.804434072345391,
      "grad_norm": 0.035888671875,
      "learning_rate": 7.50144869096975e-05,
      "loss": 0.0003,
      "step": 12436
    },
    {
      "epoch": 5.8049008168028005,
      "grad_norm": 0.111328125,
      "learning_rate": 7.50002790914728e-05,
      "loss": 0.0005,
      "step": 12437
    },
    {
      "epoch": 5.80536756126021,
      "grad_norm": 0.1513671875,
      "learning_rate": 7.498607181151984e-05,
      "loss": 0.0007,
      "step": 12438
    },
    {
      "epoch": 5.805834305717619,
      "grad_norm": 0.134765625,
      "learning_rate": 7.49718650701445e-05,
      "loss": 0.0029,
      "step": 12439
    },
    {
      "epoch": 5.806301050175029,
      "grad_norm": 0.032958984375,
      "learning_rate": 7.495765886765269e-05,
      "loss": 0.0004,
      "step": 12440
    },
    {
      "epoch": 5.806767794632439,
      "grad_norm": 0.66015625,
      "learning_rate": 7.494345320435026e-05,
      "loss": 0.0147,
      "step": 12441
    },
    {
      "epoch": 5.8072345390898485,
      "grad_norm": 0.03857421875,
      "learning_rate": 7.492924808054307e-05,
      "loss": 0.0006,
      "step": 12442
    },
    {
      "epoch": 5.807701283547258,
      "grad_norm": 0.125,
      "learning_rate": 7.491504349653699e-05,
      "loss": 0.0007,
      "step": 12443
    },
    {
      "epoch": 5.808168028004667,
      "grad_norm": 0.283203125,
      "learning_rate": 7.490083945263789e-05,
      "loss": 0.0013,
      "step": 12444
    },
    {
      "epoch": 5.808634772462077,
      "grad_norm": 0.1259765625,
      "learning_rate": 7.488663594915154e-05,
      "loss": 0.0062,
      "step": 12445
    },
    {
      "epoch": 5.809101516919487,
      "grad_norm": 0.193359375,
      "learning_rate": 7.487243298638375e-05,
      "loss": 0.0031,
      "step": 12446
    },
    {
      "epoch": 5.809568261376896,
      "grad_norm": 0.1669921875,
      "learning_rate": 7.485823056464041e-05,
      "loss": 0.0008,
      "step": 12447
    },
    {
      "epoch": 5.8100350058343055,
      "grad_norm": 0.1591796875,
      "learning_rate": 7.484402868422724e-05,
      "loss": 0.0008,
      "step": 12448
    },
    {
      "epoch": 5.810501750291715,
      "grad_norm": 0.04296875,
      "learning_rate": 7.482982734545003e-05,
      "loss": 0.0029,
      "step": 12449
    },
    {
      "epoch": 5.810968494749125,
      "grad_norm": 0.1298828125,
      "learning_rate": 7.481562654861458e-05,
      "loss": 0.0008,
      "step": 12450
    },
    {
      "epoch": 5.811435239206534,
      "grad_norm": 0.111328125,
      "learning_rate": 7.480142629402663e-05,
      "loss": 0.0006,
      "step": 12451
    },
    {
      "epoch": 5.811901983663944,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.478722658199193e-05,
      "loss": 0.0009,
      "step": 12452
    },
    {
      "epoch": 5.812368728121354,
      "grad_norm": 0.271484375,
      "learning_rate": 7.477302741281621e-05,
      "loss": 0.0056,
      "step": 12453
    },
    {
      "epoch": 5.812835472578763,
      "grad_norm": 0.27734375,
      "learning_rate": 7.475882878680521e-05,
      "loss": 0.0016,
      "step": 12454
    },
    {
      "epoch": 5.813302217036172,
      "grad_norm": 0.0498046875,
      "learning_rate": 7.474463070426463e-05,
      "loss": 0.0004,
      "step": 12455
    },
    {
      "epoch": 5.813768961493582,
      "grad_norm": 0.06640625,
      "learning_rate": 7.473043316550016e-05,
      "loss": 0.0031,
      "step": 12456
    },
    {
      "epoch": 5.814235705950992,
      "grad_norm": 0.0218505859375,
      "learning_rate": 7.471623617081752e-05,
      "loss": 0.0003,
      "step": 12457
    },
    {
      "epoch": 5.814702450408402,
      "grad_norm": 0.2734375,
      "learning_rate": 7.470203972052237e-05,
      "loss": 0.0008,
      "step": 12458
    },
    {
      "epoch": 5.815169194865811,
      "grad_norm": 0.0712890625,
      "learning_rate": 7.468784381492033e-05,
      "loss": 0.0006,
      "step": 12459
    },
    {
      "epoch": 5.81563593932322,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.467364845431717e-05,
      "loss": 0.0009,
      "step": 12460
    },
    {
      "epoch": 5.81610268378063,
      "grad_norm": 0.01611328125,
      "learning_rate": 7.46594536390184e-05,
      "loss": 0.0003,
      "step": 12461
    },
    {
      "epoch": 5.81656942823804,
      "grad_norm": 0.13671875,
      "learning_rate": 7.464525936932974e-05,
      "loss": 0.0007,
      "step": 12462
    },
    {
      "epoch": 5.817036172695449,
      "grad_norm": 0.05224609375,
      "learning_rate": 7.463106564555678e-05,
      "loss": 0.0006,
      "step": 12463
    },
    {
      "epoch": 5.817502917152859,
      "grad_norm": 0.0556640625,
      "learning_rate": 7.461687246800509e-05,
      "loss": 0.0004,
      "step": 12464
    },
    {
      "epoch": 5.8179696616102685,
      "grad_norm": 0.16015625,
      "learning_rate": 7.460267983698033e-05,
      "loss": 0.0008,
      "step": 12465
    },
    {
      "epoch": 5.818436406067678,
      "grad_norm": 0.1484375,
      "learning_rate": 7.458848775278804e-05,
      "loss": 0.0007,
      "step": 12466
    },
    {
      "epoch": 5.818903150525087,
      "grad_norm": 0.0196533203125,
      "learning_rate": 7.45742962157338e-05,
      "loss": 0.0003,
      "step": 12467
    },
    {
      "epoch": 5.819369894982497,
      "grad_norm": 0.0859375,
      "learning_rate": 7.456010522612316e-05,
      "loss": 0.0008,
      "step": 12468
    },
    {
      "epoch": 5.819836639439907,
      "grad_norm": 0.0400390625,
      "learning_rate": 7.45459147842617e-05,
      "loss": 0.002,
      "step": 12469
    },
    {
      "epoch": 5.820303383897317,
      "grad_norm": 0.166015625,
      "learning_rate": 7.453172489045495e-05,
      "loss": 0.0007,
      "step": 12470
    },
    {
      "epoch": 5.8207701283547255,
      "grad_norm": 0.0299072265625,
      "learning_rate": 7.451753554500838e-05,
      "loss": 0.0004,
      "step": 12471
    },
    {
      "epoch": 5.821236872812135,
      "grad_norm": 0.03662109375,
      "learning_rate": 7.450334674822757e-05,
      "loss": 0.0003,
      "step": 12472
    },
    {
      "epoch": 5.821703617269545,
      "grad_norm": 0.0908203125,
      "learning_rate": 7.448915850041799e-05,
      "loss": 0.0008,
      "step": 12473
    },
    {
      "epoch": 5.822170361726954,
      "grad_norm": 0.0262451171875,
      "learning_rate": 7.447497080188511e-05,
      "loss": 0.0003,
      "step": 12474
    },
    {
      "epoch": 5.822637106184364,
      "grad_norm": 0.11376953125,
      "learning_rate": 7.446078365293446e-05,
      "loss": 0.0005,
      "step": 12475
    },
    {
      "epoch": 5.823103850641774,
      "grad_norm": 0.037353515625,
      "learning_rate": 7.444659705387147e-05,
      "loss": 0.0004,
      "step": 12476
    },
    {
      "epoch": 5.823570595099183,
      "grad_norm": 0.30859375,
      "learning_rate": 7.44324110050016e-05,
      "loss": 0.0036,
      "step": 12477
    },
    {
      "epoch": 5.824037339556593,
      "grad_norm": 0.07080078125,
      "learning_rate": 7.441822550663027e-05,
      "loss": 0.0005,
      "step": 12478
    },
    {
      "epoch": 5.824504084014002,
      "grad_norm": 0.357421875,
      "learning_rate": 7.440404055906293e-05,
      "loss": 0.005,
      "step": 12479
    },
    {
      "epoch": 5.824970828471412,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.438985616260503e-05,
      "loss": 0.0018,
      "step": 12480
    },
    {
      "epoch": 5.825437572928822,
      "grad_norm": 0.04296875,
      "learning_rate": 7.437567231756188e-05,
      "loss": 0.0004,
      "step": 12481
    },
    {
      "epoch": 5.825904317386231,
      "grad_norm": 0.01953125,
      "learning_rate": 7.436148902423898e-05,
      "loss": 0.0003,
      "step": 12482
    },
    {
      "epoch": 5.82637106184364,
      "grad_norm": 0.044189453125,
      "learning_rate": 7.434730628294167e-05,
      "loss": 0.0003,
      "step": 12483
    },
    {
      "epoch": 5.82683780630105,
      "grad_norm": 0.068359375,
      "learning_rate": 7.43331240939753e-05,
      "loss": 0.0006,
      "step": 12484
    },
    {
      "epoch": 5.82730455075846,
      "grad_norm": 0.0703125,
      "learning_rate": 7.431894245764525e-05,
      "loss": 0.0005,
      "step": 12485
    },
    {
      "epoch": 5.82777129521587,
      "grad_norm": 0.05810546875,
      "learning_rate": 7.430476137425688e-05,
      "loss": 0.0036,
      "step": 12486
    },
    {
      "epoch": 5.828238039673279,
      "grad_norm": 0.2470703125,
      "learning_rate": 7.42905808441155e-05,
      "loss": 0.0015,
      "step": 12487
    },
    {
      "epoch": 5.8287047841306885,
      "grad_norm": 0.043212890625,
      "learning_rate": 7.427640086752641e-05,
      "loss": 0.0004,
      "step": 12488
    },
    {
      "epoch": 5.829171528588098,
      "grad_norm": 0.1875,
      "learning_rate": 7.426222144479497e-05,
      "loss": 0.0009,
      "step": 12489
    },
    {
      "epoch": 5.829638273045507,
      "grad_norm": 0.10302734375,
      "learning_rate": 7.424804257622648e-05,
      "loss": 0.0005,
      "step": 12490
    },
    {
      "epoch": 5.830105017502917,
      "grad_norm": 0.0291748046875,
      "learning_rate": 7.423386426212617e-05,
      "loss": 0.0003,
      "step": 12491
    },
    {
      "epoch": 5.830571761960327,
      "grad_norm": 0.0172119140625,
      "learning_rate": 7.421968650279938e-05,
      "loss": 0.0003,
      "step": 12492
    },
    {
      "epoch": 5.831038506417737,
      "grad_norm": 0.0240478515625,
      "learning_rate": 7.420550929855133e-05,
      "loss": 0.0003,
      "step": 12493
    },
    {
      "epoch": 5.8315052508751455,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.419133264968729e-05,
      "loss": 0.0004,
      "step": 12494
    },
    {
      "epoch": 5.831971995332555,
      "grad_norm": 0.0322265625,
      "learning_rate": 7.417715655651247e-05,
      "loss": 0.0004,
      "step": 12495
    },
    {
      "epoch": 5.832438739789965,
      "grad_norm": 0.443359375,
      "learning_rate": 7.416298101933216e-05,
      "loss": 0.0014,
      "step": 12496
    },
    {
      "epoch": 5.832905484247375,
      "grad_norm": 0.10693359375,
      "learning_rate": 7.414880603845152e-05,
      "loss": 0.0005,
      "step": 12497
    },
    {
      "epoch": 5.833372228704784,
      "grad_norm": 0.052490234375,
      "learning_rate": 7.413463161417575e-05,
      "loss": 0.0004,
      "step": 12498
    },
    {
      "epoch": 5.833838973162194,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.412045774681006e-05,
      "loss": 0.0006,
      "step": 12499
    },
    {
      "epoch": 5.834305717619603,
      "grad_norm": 0.03173828125,
      "learning_rate": 7.410628443665966e-05,
      "loss": 0.0003,
      "step": 12500
    },
    {
      "epoch": 5.834772462077013,
      "grad_norm": 0.013916015625,
      "learning_rate": 7.409211168402965e-05,
      "loss": 0.0003,
      "step": 12501
    },
    {
      "epoch": 5.835239206534422,
      "grad_norm": 0.205078125,
      "learning_rate": 7.407793948922523e-05,
      "loss": 0.0006,
      "step": 12502
    },
    {
      "epoch": 5.835705950991832,
      "grad_norm": 0.1728515625,
      "learning_rate": 7.406376785255156e-05,
      "loss": 0.0007,
      "step": 12503
    },
    {
      "epoch": 5.836172695449242,
      "grad_norm": 0.031982421875,
      "learning_rate": 7.404959677431373e-05,
      "loss": 0.0003,
      "step": 12504
    },
    {
      "epoch": 5.8366394399066515,
      "grad_norm": 0.039794921875,
      "learning_rate": 7.403542625481685e-05,
      "loss": 0.0004,
      "step": 12505
    },
    {
      "epoch": 5.83710618436406,
      "grad_norm": 0.031982421875,
      "learning_rate": 7.402125629436608e-05,
      "loss": 0.0003,
      "step": 12506
    },
    {
      "epoch": 5.83757292882147,
      "grad_norm": 0.055908203125,
      "learning_rate": 7.400708689326649e-05,
      "loss": 0.0024,
      "step": 12507
    },
    {
      "epoch": 5.83803967327888,
      "grad_norm": 0.0546875,
      "learning_rate": 7.399291805182317e-05,
      "loss": 0.0004,
      "step": 12508
    },
    {
      "epoch": 5.83850641773629,
      "grad_norm": 0.1484375,
      "learning_rate": 7.397874977034117e-05,
      "loss": 0.0005,
      "step": 12509
    },
    {
      "epoch": 5.838973162193699,
      "grad_norm": 0.62109375,
      "learning_rate": 7.396458204912552e-05,
      "loss": 0.0038,
      "step": 12510
    },
    {
      "epoch": 5.8394399066511085,
      "grad_norm": 0.1025390625,
      "learning_rate": 7.395041488848134e-05,
      "loss": 0.0006,
      "step": 12511
    },
    {
      "epoch": 5.839906651108518,
      "grad_norm": 0.08203125,
      "learning_rate": 7.393624828871362e-05,
      "loss": 0.0004,
      "step": 12512
    },
    {
      "epoch": 5.840373395565928,
      "grad_norm": 0.042724609375,
      "learning_rate": 7.392208225012738e-05,
      "loss": 0.0003,
      "step": 12513
    },
    {
      "epoch": 5.840840140023337,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.390791677302765e-05,
      "loss": 0.0029,
      "step": 12514
    },
    {
      "epoch": 5.841306884480747,
      "grad_norm": 0.28515625,
      "learning_rate": 7.389375185771943e-05,
      "loss": 0.0047,
      "step": 12515
    },
    {
      "epoch": 5.8417736289381565,
      "grad_norm": 0.06396484375,
      "learning_rate": 7.38795875045077e-05,
      "loss": 0.0005,
      "step": 12516
    },
    {
      "epoch": 5.8422403733955655,
      "grad_norm": 0.099609375,
      "learning_rate": 7.386542371369743e-05,
      "loss": 0.0005,
      "step": 12517
    },
    {
      "epoch": 5.842707117852975,
      "grad_norm": 0.03466796875,
      "learning_rate": 7.385126048559358e-05,
      "loss": 0.0003,
      "step": 12518
    },
    {
      "epoch": 5.843173862310385,
      "grad_norm": 0.2392578125,
      "learning_rate": 7.38370978205011e-05,
      "loss": 0.0019,
      "step": 12519
    },
    {
      "epoch": 5.843640606767795,
      "grad_norm": 0.1923828125,
      "learning_rate": 7.382293571872493e-05,
      "loss": 0.0011,
      "step": 12520
    },
    {
      "epoch": 5.844107351225205,
      "grad_norm": 0.0791015625,
      "learning_rate": 7.380877418057e-05,
      "loss": 0.0007,
      "step": 12521
    },
    {
      "epoch": 5.8445740956826135,
      "grad_norm": 0.0849609375,
      "learning_rate": 7.379461320634125e-05,
      "loss": 0.0034,
      "step": 12522
    },
    {
      "epoch": 5.845040840140023,
      "grad_norm": 0.056884765625,
      "learning_rate": 7.37804527963435e-05,
      "loss": 0.0007,
      "step": 12523
    },
    {
      "epoch": 5.845507584597433,
      "grad_norm": 0.318359375,
      "learning_rate": 7.376629295088173e-05,
      "loss": 0.0009,
      "step": 12524
    },
    {
      "epoch": 5.845974329054842,
      "grad_norm": 0.2265625,
      "learning_rate": 7.375213367026077e-05,
      "loss": 0.0029,
      "step": 12525
    },
    {
      "epoch": 5.846441073512252,
      "grad_norm": 0.283203125,
      "learning_rate": 7.373797495478551e-05,
      "loss": 0.0017,
      "step": 12526
    },
    {
      "epoch": 5.846907817969662,
      "grad_norm": 0.036865234375,
      "learning_rate": 7.372381680476076e-05,
      "loss": 0.0004,
      "step": 12527
    },
    {
      "epoch": 5.847374562427071,
      "grad_norm": 0.12353515625,
      "learning_rate": 7.37096592204914e-05,
      "loss": 0.0007,
      "step": 12528
    },
    {
      "epoch": 5.847841306884481,
      "grad_norm": 0.126953125,
      "learning_rate": 7.369550220228225e-05,
      "loss": 0.0076,
      "step": 12529
    },
    {
      "epoch": 5.84830805134189,
      "grad_norm": 0.15234375,
      "learning_rate": 7.368134575043811e-05,
      "loss": 0.0012,
      "step": 12530
    },
    {
      "epoch": 5.8487747957993,
      "grad_norm": 0.29296875,
      "learning_rate": 7.366718986526382e-05,
      "loss": 0.0043,
      "step": 12531
    },
    {
      "epoch": 5.84924154025671,
      "grad_norm": 0.0322265625,
      "learning_rate": 7.365303454706414e-05,
      "loss": 0.0022,
      "step": 12532
    },
    {
      "epoch": 5.849708284714119,
      "grad_norm": 0.04638671875,
      "learning_rate": 7.363887979614384e-05,
      "loss": 0.0003,
      "step": 12533
    },
    {
      "epoch": 5.850175029171528,
      "grad_norm": 0.248046875,
      "learning_rate": 7.362472561280773e-05,
      "loss": 0.0009,
      "step": 12534
    },
    {
      "epoch": 5.850641773628938,
      "grad_norm": 0.02392578125,
      "learning_rate": 7.361057199736055e-05,
      "loss": 0.0003,
      "step": 12535
    },
    {
      "epoch": 5.851108518086348,
      "grad_norm": 0.072265625,
      "learning_rate": 7.359641895010702e-05,
      "loss": 0.0032,
      "step": 12536
    },
    {
      "epoch": 5.851575262543757,
      "grad_norm": 0.09033203125,
      "learning_rate": 7.358226647135186e-05,
      "loss": 0.0005,
      "step": 12537
    },
    {
      "epoch": 5.852042007001167,
      "grad_norm": 0.0263671875,
      "learning_rate": 7.356811456139985e-05,
      "loss": 0.0004,
      "step": 12538
    },
    {
      "epoch": 5.8525087514585765,
      "grad_norm": 0.078125,
      "learning_rate": 7.355396322055565e-05,
      "loss": 0.0005,
      "step": 12539
    },
    {
      "epoch": 5.852975495915986,
      "grad_norm": 0.107421875,
      "learning_rate": 7.353981244912394e-05,
      "loss": 0.0007,
      "step": 12540
    },
    {
      "epoch": 5.853442240373395,
      "grad_norm": 0.27734375,
      "learning_rate": 7.352566224740946e-05,
      "loss": 0.002,
      "step": 12541
    },
    {
      "epoch": 5.853908984830805,
      "grad_norm": 0.049560546875,
      "learning_rate": 7.351151261571684e-05,
      "loss": 0.0004,
      "step": 12542
    },
    {
      "epoch": 5.854375729288215,
      "grad_norm": 0.1162109375,
      "learning_rate": 7.349736355435072e-05,
      "loss": 0.0006,
      "step": 12543
    },
    {
      "epoch": 5.854842473745625,
      "grad_norm": 0.0771484375,
      "learning_rate": 7.348321506361579e-05,
      "loss": 0.0005,
      "step": 12544
    },
    {
      "epoch": 5.8553092182030335,
      "grad_norm": 0.06787109375,
      "learning_rate": 7.346906714381665e-05,
      "loss": 0.0005,
      "step": 12545
    },
    {
      "epoch": 5.855775962660443,
      "grad_norm": 0.044677734375,
      "learning_rate": 7.345491979525795e-05,
      "loss": 0.0004,
      "step": 12546
    },
    {
      "epoch": 5.856242707117853,
      "grad_norm": 0.134765625,
      "learning_rate": 7.344077301824425e-05,
      "loss": 0.0005,
      "step": 12547
    },
    {
      "epoch": 5.856709451575263,
      "grad_norm": 0.1923828125,
      "learning_rate": 7.342662681308018e-05,
      "loss": 0.0008,
      "step": 12548
    },
    {
      "epoch": 5.857176196032672,
      "grad_norm": 0.038330078125,
      "learning_rate": 7.341248118007035e-05,
      "loss": 0.0004,
      "step": 12549
    },
    {
      "epoch": 5.857642940490082,
      "grad_norm": 0.12890625,
      "learning_rate": 7.339833611951925e-05,
      "loss": 0.0005,
      "step": 12550
    },
    {
      "epoch": 5.858109684947491,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.338419163173153e-05,
      "loss": 0.0074,
      "step": 12551
    },
    {
      "epoch": 5.858576429404901,
      "grad_norm": 0.2197265625,
      "learning_rate": 7.337004771701168e-05,
      "loss": 0.0036,
      "step": 12552
    },
    {
      "epoch": 5.85904317386231,
      "grad_norm": 0.251953125,
      "learning_rate": 7.335590437566426e-05,
      "loss": 0.0028,
      "step": 12553
    },
    {
      "epoch": 5.85950991831972,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.334176160799375e-05,
      "loss": 0.0011,
      "step": 12554
    },
    {
      "epoch": 5.85997666277713,
      "grad_norm": 0.115234375,
      "learning_rate": 7.332761941430474e-05,
      "loss": 0.0008,
      "step": 12555
    },
    {
      "epoch": 5.8604434072345395,
      "grad_norm": 0.0262451171875,
      "learning_rate": 7.331347779490165e-05,
      "loss": 0.0003,
      "step": 12556
    },
    {
      "epoch": 5.860910151691948,
      "grad_norm": 0.04248046875,
      "learning_rate": 7.3299336750089e-05,
      "loss": 0.0004,
      "step": 12557
    },
    {
      "epoch": 5.861376896149358,
      "grad_norm": 0.0625,
      "learning_rate": 7.328519628017125e-05,
      "loss": 0.0004,
      "step": 12558
    },
    {
      "epoch": 5.861843640606768,
      "grad_norm": 0.04443359375,
      "learning_rate": 7.327105638545285e-05,
      "loss": 0.0004,
      "step": 12559
    },
    {
      "epoch": 5.862310385064177,
      "grad_norm": 0.34765625,
      "learning_rate": 7.325691706623827e-05,
      "loss": 0.0049,
      "step": 12560
    },
    {
      "epoch": 5.862777129521587,
      "grad_norm": 0.439453125,
      "learning_rate": 7.324277832283195e-05,
      "loss": 0.002,
      "step": 12561
    },
    {
      "epoch": 5.8632438739789965,
      "grad_norm": 0.23046875,
      "learning_rate": 7.322864015553828e-05,
      "loss": 0.0076,
      "step": 12562
    },
    {
      "epoch": 5.863710618436406,
      "grad_norm": 0.1748046875,
      "learning_rate": 7.321450256466171e-05,
      "loss": 0.0012,
      "step": 12563
    },
    {
      "epoch": 5.864177362893816,
      "grad_norm": 0.0732421875,
      "learning_rate": 7.320036555050662e-05,
      "loss": 0.0003,
      "step": 12564
    },
    {
      "epoch": 5.864644107351225,
      "grad_norm": 0.0166015625,
      "learning_rate": 7.318622911337739e-05,
      "loss": 0.0003,
      "step": 12565
    },
    {
      "epoch": 5.865110851808635,
      "grad_norm": 0.08349609375,
      "learning_rate": 7.317209325357837e-05,
      "loss": 0.0037,
      "step": 12566
    },
    {
      "epoch": 5.865577596266045,
      "grad_norm": 0.0257568359375,
      "learning_rate": 7.315795797141398e-05,
      "loss": 0.0003,
      "step": 12567
    },
    {
      "epoch": 5.8660443407234535,
      "grad_norm": 0.2412109375,
      "learning_rate": 7.314382326718853e-05,
      "loss": 0.0016,
      "step": 12568
    },
    {
      "epoch": 5.866511085180863,
      "grad_norm": 0.017333984375,
      "learning_rate": 7.312968914120635e-05,
      "loss": 0.0002,
      "step": 12569
    },
    {
      "epoch": 5.866977829638273,
      "grad_norm": 0.0257568359375,
      "learning_rate": 7.311555559377179e-05,
      "loss": 0.0003,
      "step": 12570
    },
    {
      "epoch": 5.867444574095683,
      "grad_norm": 0.07177734375,
      "learning_rate": 7.310142262518915e-05,
      "loss": 0.0004,
      "step": 12571
    },
    {
      "epoch": 5.867911318553093,
      "grad_norm": 0.255859375,
      "learning_rate": 7.30872902357627e-05,
      "loss": 0.0031,
      "step": 12572
    },
    {
      "epoch": 5.868378063010502,
      "grad_norm": 0.138671875,
      "learning_rate": 7.307315842579676e-05,
      "loss": 0.0006,
      "step": 12573
    },
    {
      "epoch": 5.868844807467911,
      "grad_norm": 0.1435546875,
      "learning_rate": 7.305902719559562e-05,
      "loss": 0.0023,
      "step": 12574
    },
    {
      "epoch": 5.869311551925321,
      "grad_norm": 0.13671875,
      "learning_rate": 7.304489654546351e-05,
      "loss": 0.0006,
      "step": 12575
    },
    {
      "epoch": 5.86977829638273,
      "grad_norm": 0.0252685546875,
      "learning_rate": 7.303076647570465e-05,
      "loss": 0.0004,
      "step": 12576
    },
    {
      "epoch": 5.87024504084014,
      "grad_norm": 0.134765625,
      "learning_rate": 7.301663698662334e-05,
      "loss": 0.0007,
      "step": 12577
    },
    {
      "epoch": 5.87071178529755,
      "grad_norm": 0.1455078125,
      "learning_rate": 7.300250807852378e-05,
      "loss": 0.0049,
      "step": 12578
    },
    {
      "epoch": 5.8711785297549595,
      "grad_norm": 0.083984375,
      "learning_rate": 7.298837975171015e-05,
      "loss": 0.0005,
      "step": 12579
    },
    {
      "epoch": 5.871645274212368,
      "grad_norm": 0.181640625,
      "learning_rate": 7.297425200648669e-05,
      "loss": 0.0008,
      "step": 12580
    },
    {
      "epoch": 5.872112018669778,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.296012484315757e-05,
      "loss": 0.0005,
      "step": 12581
    },
    {
      "epoch": 5.872578763127188,
      "grad_norm": 0.09912109375,
      "learning_rate": 7.294599826202694e-05,
      "loss": 0.0009,
      "step": 12582
    },
    {
      "epoch": 5.873045507584598,
      "grad_norm": 0.0537109375,
      "learning_rate": 7.293187226339903e-05,
      "loss": 0.0005,
      "step": 12583
    },
    {
      "epoch": 5.873512252042007,
      "grad_norm": 0.08544921875,
      "learning_rate": 7.291774684757793e-05,
      "loss": 0.0005,
      "step": 12584
    },
    {
      "epoch": 5.8739789964994165,
      "grad_norm": 0.078125,
      "learning_rate": 7.290362201486778e-05,
      "loss": 0.0004,
      "step": 12585
    },
    {
      "epoch": 5.874445740956826,
      "grad_norm": 0.023681640625,
      "learning_rate": 7.288949776557268e-05,
      "loss": 0.0003,
      "step": 12586
    },
    {
      "epoch": 5.874912485414236,
      "grad_norm": 0.035888671875,
      "learning_rate": 7.287537409999681e-05,
      "loss": 0.0004,
      "step": 12587
    },
    {
      "epoch": 5.875379229871645,
      "grad_norm": 0.349609375,
      "learning_rate": 7.286125101844424e-05,
      "loss": 0.0039,
      "step": 12588
    },
    {
      "epoch": 5.875845974329055,
      "grad_norm": 0.06640625,
      "learning_rate": 7.2847128521219e-05,
      "loss": 0.0005,
      "step": 12589
    },
    {
      "epoch": 5.8763127187864646,
      "grad_norm": 0.03515625,
      "learning_rate": 7.283300660862525e-05,
      "loss": 0.0004,
      "step": 12590
    },
    {
      "epoch": 5.876779463243874,
      "grad_norm": 0.0303955078125,
      "learning_rate": 7.2818885280967e-05,
      "loss": 0.0003,
      "step": 12591
    },
    {
      "epoch": 5.877246207701283,
      "grad_norm": 0.02978515625,
      "learning_rate": 7.28047645385483e-05,
      "loss": 0.0003,
      "step": 12592
    },
    {
      "epoch": 5.877712952158693,
      "grad_norm": 0.0400390625,
      "learning_rate": 7.27906443816732e-05,
      "loss": 0.0005,
      "step": 12593
    },
    {
      "epoch": 5.878179696616103,
      "grad_norm": 0.040771484375,
      "learning_rate": 7.277652481064573e-05,
      "loss": 0.0003,
      "step": 12594
    },
    {
      "epoch": 5.878646441073512,
      "grad_norm": 0.0859375,
      "learning_rate": 7.276240582576987e-05,
      "loss": 0.0006,
      "step": 12595
    },
    {
      "epoch": 5.8791131855309215,
      "grad_norm": 0.076171875,
      "learning_rate": 7.274828742734962e-05,
      "loss": 0.0005,
      "step": 12596
    },
    {
      "epoch": 5.879579929988331,
      "grad_norm": 0.23046875,
      "learning_rate": 7.2734169615689e-05,
      "loss": 0.0022,
      "step": 12597
    },
    {
      "epoch": 5.880046674445741,
      "grad_norm": 0.22265625,
      "learning_rate": 7.272005239109196e-05,
      "loss": 0.0016,
      "step": 12598
    },
    {
      "epoch": 5.880513418903151,
      "grad_norm": 0.232421875,
      "learning_rate": 7.270593575386245e-05,
      "loss": 0.0013,
      "step": 12599
    },
    {
      "epoch": 5.88098016336056,
      "grad_norm": 0.046875,
      "learning_rate": 7.269181970430443e-05,
      "loss": 0.0005,
      "step": 12600
    },
    {
      "epoch": 5.88144690781797,
      "grad_norm": 0.033203125,
      "learning_rate": 7.267770424272183e-05,
      "loss": 0.0003,
      "step": 12601
    },
    {
      "epoch": 5.881913652275379,
      "grad_norm": 0.51953125,
      "learning_rate": 7.266358936941858e-05,
      "loss": 0.002,
      "step": 12602
    },
    {
      "epoch": 5.882380396732788,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.264947508469861e-05,
      "loss": 0.0004,
      "step": 12603
    },
    {
      "epoch": 5.882847141190198,
      "grad_norm": 0.10107421875,
      "learning_rate": 7.263536138886573e-05,
      "loss": 0.0006,
      "step": 12604
    },
    {
      "epoch": 5.883313885647608,
      "grad_norm": 0.12451171875,
      "learning_rate": 7.26212482822239e-05,
      "loss": 0.0006,
      "step": 12605
    },
    {
      "epoch": 5.883780630105018,
      "grad_norm": 0.0279541015625,
      "learning_rate": 7.260713576507699e-05,
      "loss": 0.0004,
      "step": 12606
    },
    {
      "epoch": 5.8842473745624275,
      "grad_norm": 0.055908203125,
      "learning_rate": 7.259302383772883e-05,
      "loss": 0.0005,
      "step": 12607
    },
    {
      "epoch": 5.884714119019836,
      "grad_norm": 0.0693359375,
      "learning_rate": 7.257891250048325e-05,
      "loss": 0.0007,
      "step": 12608
    },
    {
      "epoch": 5.885180863477246,
      "grad_norm": 0.1611328125,
      "learning_rate": 7.256480175364413e-05,
      "loss": 0.0009,
      "step": 12609
    },
    {
      "epoch": 5.885647607934656,
      "grad_norm": 0.028564453125,
      "learning_rate": 7.255069159751526e-05,
      "loss": 0.0004,
      "step": 12610
    },
    {
      "epoch": 5.886114352392065,
      "grad_norm": 0.07373046875,
      "learning_rate": 7.253658203240046e-05,
      "loss": 0.0004,
      "step": 12611
    },
    {
      "epoch": 5.886581096849475,
      "grad_norm": 0.032958984375,
      "learning_rate": 7.252247305860353e-05,
      "loss": 0.0004,
      "step": 12612
    },
    {
      "epoch": 5.8870478413068845,
      "grad_norm": 0.236328125,
      "learning_rate": 7.250836467642823e-05,
      "loss": 0.0008,
      "step": 12613
    },
    {
      "epoch": 5.887514585764294,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.249425688617835e-05,
      "loss": 0.0005,
      "step": 12614
    },
    {
      "epoch": 5.887981330221704,
      "grad_norm": 0.057373046875,
      "learning_rate": 7.248014968815762e-05,
      "loss": 0.0005,
      "step": 12615
    },
    {
      "epoch": 5.888448074679113,
      "grad_norm": 0.11865234375,
      "learning_rate": 7.24660430826698e-05,
      "loss": 0.0038,
      "step": 12616
    },
    {
      "epoch": 5.888914819136523,
      "grad_norm": 0.0206298828125,
      "learning_rate": 7.245193707001866e-05,
      "loss": 0.0003,
      "step": 12617
    },
    {
      "epoch": 5.889381563593933,
      "grad_norm": 0.038818359375,
      "learning_rate": 7.243783165050784e-05,
      "loss": 0.0003,
      "step": 12618
    },
    {
      "epoch": 5.8898483080513415,
      "grad_norm": 0.1298828125,
      "learning_rate": 7.24237268244411e-05,
      "loss": 0.001,
      "step": 12619
    },
    {
      "epoch": 5.890315052508751,
      "grad_norm": 0.08154296875,
      "learning_rate": 7.240962259212212e-05,
      "loss": 0.0035,
      "step": 12620
    },
    {
      "epoch": 5.890781796966161,
      "grad_norm": 0.04296875,
      "learning_rate": 7.239551895385455e-05,
      "loss": 0.0022,
      "step": 12621
    },
    {
      "epoch": 5.891248541423571,
      "grad_norm": 0.0118408203125,
      "learning_rate": 7.238141590994212e-05,
      "loss": 0.0002,
      "step": 12622
    },
    {
      "epoch": 5.89171528588098,
      "grad_norm": 0.0263671875,
      "learning_rate": 7.236731346068844e-05,
      "loss": 0.0003,
      "step": 12623
    },
    {
      "epoch": 5.89218203033839,
      "grad_norm": 0.018310546875,
      "learning_rate": 7.235321160639716e-05,
      "loss": 0.0003,
      "step": 12624
    },
    {
      "epoch": 5.892648774795799,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.23391103473719e-05,
      "loss": 0.0006,
      "step": 12625
    },
    {
      "epoch": 5.893115519253209,
      "grad_norm": 0.0595703125,
      "learning_rate": 7.232500968391629e-05,
      "loss": 0.0005,
      "step": 12626
    },
    {
      "epoch": 5.893582263710618,
      "grad_norm": 0.263671875,
      "learning_rate": 7.231090961633395e-05,
      "loss": 0.0008,
      "step": 12627
    },
    {
      "epoch": 5.894049008168028,
      "grad_norm": 0.019287109375,
      "learning_rate": 7.22968101449284e-05,
      "loss": 0.0003,
      "step": 12628
    },
    {
      "epoch": 5.894515752625438,
      "grad_norm": 0.0361328125,
      "learning_rate": 7.22827112700033e-05,
      "loss": 0.0003,
      "step": 12629
    },
    {
      "epoch": 5.8949824970828475,
      "grad_norm": 0.322265625,
      "learning_rate": 7.226861299186218e-05,
      "loss": 0.0035,
      "step": 12630
    },
    {
      "epoch": 5.895449241540256,
      "grad_norm": 0.054443359375,
      "learning_rate": 7.225451531080855e-05,
      "loss": 0.0004,
      "step": 12631
    },
    {
      "epoch": 5.895915985997666,
      "grad_norm": 0.5859375,
      "learning_rate": 7.224041822714604e-05,
      "loss": 0.0163,
      "step": 12632
    },
    {
      "epoch": 5.896382730455076,
      "grad_norm": 0.0458984375,
      "learning_rate": 7.222632174117812e-05,
      "loss": 0.0004,
      "step": 12633
    },
    {
      "epoch": 5.896849474912486,
      "grad_norm": 0.0400390625,
      "learning_rate": 7.221222585320828e-05,
      "loss": 0.0004,
      "step": 12634
    },
    {
      "epoch": 5.897316219369895,
      "grad_norm": 0.048828125,
      "learning_rate": 7.219813056354006e-05,
      "loss": 0.0046,
      "step": 12635
    },
    {
      "epoch": 5.8977829638273045,
      "grad_norm": 0.06787109375,
      "learning_rate": 7.218403587247694e-05,
      "loss": 0.0004,
      "step": 12636
    },
    {
      "epoch": 5.898249708284714,
      "grad_norm": 0.1591796875,
      "learning_rate": 7.21699417803224e-05,
      "loss": 0.0006,
      "step": 12637
    },
    {
      "epoch": 5.898716452742123,
      "grad_norm": 0.09619140625,
      "learning_rate": 7.215584828737985e-05,
      "loss": 0.0005,
      "step": 12638
    },
    {
      "epoch": 5.899183197199533,
      "grad_norm": 0.125,
      "learning_rate": 7.214175539395282e-05,
      "loss": 0.0034,
      "step": 12639
    },
    {
      "epoch": 5.899649941656943,
      "grad_norm": 0.1962890625,
      "learning_rate": 7.212766310034472e-05,
      "loss": 0.0025,
      "step": 12640
    },
    {
      "epoch": 5.900116686114353,
      "grad_norm": 0.06640625,
      "learning_rate": 7.21135714068589e-05,
      "loss": 0.0005,
      "step": 12641
    },
    {
      "epoch": 5.900583430571762,
      "grad_norm": 0.05419921875,
      "learning_rate": 7.209948031379887e-05,
      "loss": 0.0023,
      "step": 12642
    },
    {
      "epoch": 5.901050175029171,
      "grad_norm": 0.06298828125,
      "learning_rate": 7.208538982146799e-05,
      "loss": 0.0004,
      "step": 12643
    },
    {
      "epoch": 5.901516919486581,
      "grad_norm": 0.1953125,
      "learning_rate": 7.207129993016963e-05,
      "loss": 0.0008,
      "step": 12644
    },
    {
      "epoch": 5.901983663943991,
      "grad_norm": 0.2255859375,
      "learning_rate": 7.205721064020715e-05,
      "loss": 0.0039,
      "step": 12645
    },
    {
      "epoch": 5.9024504084014,
      "grad_norm": 0.06787109375,
      "learning_rate": 7.204312195188394e-05,
      "loss": 0.0004,
      "step": 12646
    },
    {
      "epoch": 5.90291715285881,
      "grad_norm": 0.203125,
      "learning_rate": 7.202903386550335e-05,
      "loss": 0.001,
      "step": 12647
    },
    {
      "epoch": 5.903383897316219,
      "grad_norm": 0.076171875,
      "learning_rate": 7.201494638136866e-05,
      "loss": 0.0007,
      "step": 12648
    },
    {
      "epoch": 5.903850641773629,
      "grad_norm": 0.018798828125,
      "learning_rate": 7.200085949978326e-05,
      "loss": 0.0003,
      "step": 12649
    },
    {
      "epoch": 5.904317386231039,
      "grad_norm": 0.1142578125,
      "learning_rate": 7.19867732210504e-05,
      "loss": 0.0005,
      "step": 12650
    },
    {
      "epoch": 5.904784130688448,
      "grad_norm": 0.185546875,
      "learning_rate": 7.197268754547344e-05,
      "loss": 0.0007,
      "step": 12651
    },
    {
      "epoch": 5.905250875145858,
      "grad_norm": 0.0291748046875,
      "learning_rate": 7.195860247335558e-05,
      "loss": 0.0003,
      "step": 12652
    },
    {
      "epoch": 5.9057176196032675,
      "grad_norm": 0.05517578125,
      "learning_rate": 7.19445180050001e-05,
      "loss": 0.0004,
      "step": 12653
    },
    {
      "epoch": 5.906184364060676,
      "grad_norm": 0.0113525390625,
      "learning_rate": 7.19304341407103e-05,
      "loss": 0.0002,
      "step": 12654
    },
    {
      "epoch": 5.906651108518086,
      "grad_norm": 0.07421875,
      "learning_rate": 7.191635088078938e-05,
      "loss": 0.0058,
      "step": 12655
    },
    {
      "epoch": 5.907117852975496,
      "grad_norm": 0.07080078125,
      "learning_rate": 7.19022682255406e-05,
      "loss": 0.0005,
      "step": 12656
    },
    {
      "epoch": 5.907584597432906,
      "grad_norm": 0.07666015625,
      "learning_rate": 7.188818617526715e-05,
      "loss": 0.0005,
      "step": 12657
    },
    {
      "epoch": 5.908051341890315,
      "grad_norm": 0.0281982421875,
      "learning_rate": 7.187410473027225e-05,
      "loss": 0.0003,
      "step": 12658
    },
    {
      "epoch": 5.9085180863477245,
      "grad_norm": 0.035888671875,
      "learning_rate": 7.186002389085908e-05,
      "loss": 0.0005,
      "step": 12659
    },
    {
      "epoch": 5.908984830805134,
      "grad_norm": 0.17578125,
      "learning_rate": 7.18459436573308e-05,
      "loss": 0.0032,
      "step": 12660
    },
    {
      "epoch": 5.909451575262544,
      "grad_norm": 0.11572265625,
      "learning_rate": 7.18318640299906e-05,
      "loss": 0.0006,
      "step": 12661
    },
    {
      "epoch": 5.909918319719953,
      "grad_norm": 0.0201416015625,
      "learning_rate": 7.181778500914164e-05,
      "loss": 0.0003,
      "step": 12662
    },
    {
      "epoch": 5.910385064177363,
      "grad_norm": 0.0634765625,
      "learning_rate": 7.180370659508702e-05,
      "loss": 0.0005,
      "step": 12663
    },
    {
      "epoch": 5.9108518086347726,
      "grad_norm": 0.032470703125,
      "learning_rate": 7.178962878812988e-05,
      "loss": 0.0004,
      "step": 12664
    },
    {
      "epoch": 5.911318553092182,
      "grad_norm": 0.189453125,
      "learning_rate": 7.177555158857334e-05,
      "loss": 0.0014,
      "step": 12665
    },
    {
      "epoch": 5.911785297549591,
      "grad_norm": 0.28515625,
      "learning_rate": 7.176147499672048e-05,
      "loss": 0.0028,
      "step": 12666
    },
    {
      "epoch": 5.912252042007001,
      "grad_norm": 0.3203125,
      "learning_rate": 7.17473990128744e-05,
      "loss": 0.0015,
      "step": 12667
    },
    {
      "epoch": 5.912718786464411,
      "grad_norm": 0.46875,
      "learning_rate": 7.173332363733817e-05,
      "loss": 0.0017,
      "step": 12668
    },
    {
      "epoch": 5.913185530921821,
      "grad_norm": 0.036865234375,
      "learning_rate": 7.171924887041485e-05,
      "loss": 0.0017,
      "step": 12669
    },
    {
      "epoch": 5.9136522753792296,
      "grad_norm": 0.0291748046875,
      "learning_rate": 7.170517471240746e-05,
      "loss": 0.0003,
      "step": 12670
    },
    {
      "epoch": 5.914119019836639,
      "grad_norm": 0.087890625,
      "learning_rate": 7.169110116361906e-05,
      "loss": 0.0005,
      "step": 12671
    },
    {
      "epoch": 5.914585764294049,
      "grad_norm": 0.0751953125,
      "learning_rate": 7.167702822435268e-05,
      "loss": 0.0042,
      "step": 12672
    },
    {
      "epoch": 5.915052508751459,
      "grad_norm": 0.1552734375,
      "learning_rate": 7.166295589491129e-05,
      "loss": 0.0006,
      "step": 12673
    },
    {
      "epoch": 5.915519253208868,
      "grad_norm": 0.03173828125,
      "learning_rate": 7.164888417559788e-05,
      "loss": 0.0003,
      "step": 12674
    },
    {
      "epoch": 5.915985997666278,
      "grad_norm": 0.197265625,
      "learning_rate": 7.163481306671549e-05,
      "loss": 0.0009,
      "step": 12675
    },
    {
      "epoch": 5.916452742123687,
      "grad_norm": 0.056640625,
      "learning_rate": 7.162074256856703e-05,
      "loss": 0.0007,
      "step": 12676
    },
    {
      "epoch": 5.916919486581097,
      "grad_norm": 0.04931640625,
      "learning_rate": 7.160667268145544e-05,
      "loss": 0.0004,
      "step": 12677
    },
    {
      "epoch": 5.917386231038506,
      "grad_norm": 0.2578125,
      "learning_rate": 7.159260340568372e-05,
      "loss": 0.0011,
      "step": 12678
    },
    {
      "epoch": 5.917852975495916,
      "grad_norm": 0.15234375,
      "learning_rate": 7.157853474155479e-05,
      "loss": 0.0006,
      "step": 12679
    },
    {
      "epoch": 5.918319719953326,
      "grad_norm": 0.05517578125,
      "learning_rate": 7.15644666893715e-05,
      "loss": 0.003,
      "step": 12680
    },
    {
      "epoch": 5.918786464410735,
      "grad_norm": 0.035400390625,
      "learning_rate": 7.15503992494368e-05,
      "loss": 0.0004,
      "step": 12681
    },
    {
      "epoch": 5.919253208868144,
      "grad_norm": 0.0306396484375,
      "learning_rate": 7.153633242205359e-05,
      "loss": 0.0004,
      "step": 12682
    },
    {
      "epoch": 5.919719953325554,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.152226620752472e-05,
      "loss": 0.001,
      "step": 12683
    },
    {
      "epoch": 5.920186697782964,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.150820060615302e-05,
      "loss": 0.0007,
      "step": 12684
    },
    {
      "epoch": 5.920653442240374,
      "grad_norm": 0.08154296875,
      "learning_rate": 7.149413561824141e-05,
      "loss": 0.0007,
      "step": 12685
    },
    {
      "epoch": 5.921120186697783,
      "grad_norm": 0.150390625,
      "learning_rate": 7.148007124409268e-05,
      "loss": 0.002,
      "step": 12686
    },
    {
      "epoch": 5.9215869311551925,
      "grad_norm": 0.0595703125,
      "learning_rate": 7.146600748400965e-05,
      "loss": 0.0005,
      "step": 12687
    },
    {
      "epoch": 5.922053675612602,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.145194433829515e-05,
      "loss": 0.0005,
      "step": 12688
    },
    {
      "epoch": 5.922520420070011,
      "grad_norm": 0.033935546875,
      "learning_rate": 7.143788180725196e-05,
      "loss": 0.0004,
      "step": 12689
    },
    {
      "epoch": 5.922987164527421,
      "grad_norm": 0.0211181640625,
      "learning_rate": 7.142381989118286e-05,
      "loss": 0.0002,
      "step": 12690
    },
    {
      "epoch": 5.923453908984831,
      "grad_norm": 0.02001953125,
      "learning_rate": 7.140975859039062e-05,
      "loss": 0.0003,
      "step": 12691
    },
    {
      "epoch": 5.923920653442241,
      "grad_norm": 0.03662109375,
      "learning_rate": 7.139569790517802e-05,
      "loss": 0.0003,
      "step": 12692
    },
    {
      "epoch": 5.92438739789965,
      "grad_norm": 0.0255126953125,
      "learning_rate": 7.138163783584776e-05,
      "loss": 0.002,
      "step": 12693
    },
    {
      "epoch": 5.924854142357059,
      "grad_norm": 0.01495361328125,
      "learning_rate": 7.136757838270258e-05,
      "loss": 0.0002,
      "step": 12694
    },
    {
      "epoch": 5.925320886814469,
      "grad_norm": 0.036865234375,
      "learning_rate": 7.135351954604523e-05,
      "loss": 0.0003,
      "step": 12695
    },
    {
      "epoch": 5.925787631271879,
      "grad_norm": 0.061279296875,
      "learning_rate": 7.133946132617839e-05,
      "loss": 0.0006,
      "step": 12696
    },
    {
      "epoch": 5.926254375729288,
      "grad_norm": 0.030517578125,
      "learning_rate": 7.132540372340472e-05,
      "loss": 0.0003,
      "step": 12697
    },
    {
      "epoch": 5.926721120186698,
      "grad_norm": 0.267578125,
      "learning_rate": 7.131134673802696e-05,
      "loss": 0.0009,
      "step": 12698
    },
    {
      "epoch": 5.927187864644107,
      "grad_norm": 0.349609375,
      "learning_rate": 7.129729037034774e-05,
      "loss": 0.0011,
      "step": 12699
    },
    {
      "epoch": 5.927654609101517,
      "grad_norm": 0.019775390625,
      "learning_rate": 7.128323462066968e-05,
      "loss": 0.0003,
      "step": 12700
    },
    {
      "epoch": 5.928121353558926,
      "grad_norm": 0.03759765625,
      "learning_rate": 7.126917948929544e-05,
      "loss": 0.0004,
      "step": 12701
    },
    {
      "epoch": 5.928588098016336,
      "grad_norm": 0.026123046875,
      "learning_rate": 7.125512497652763e-05,
      "loss": 0.0003,
      "step": 12702
    },
    {
      "epoch": 5.929054842473746,
      "grad_norm": 0.054443359375,
      "learning_rate": 7.124107108266889e-05,
      "loss": 0.0024,
      "step": 12703
    },
    {
      "epoch": 5.9295215869311555,
      "grad_norm": 0.044677734375,
      "learning_rate": 7.122701780802181e-05,
      "loss": 0.0003,
      "step": 12704
    },
    {
      "epoch": 5.929988331388564,
      "grad_norm": 0.05517578125,
      "learning_rate": 7.121296515288895e-05,
      "loss": 0.0004,
      "step": 12705
    },
    {
      "epoch": 5.930455075845974,
      "grad_norm": 0.032470703125,
      "learning_rate": 7.119891311757286e-05,
      "loss": 0.0003,
      "step": 12706
    },
    {
      "epoch": 5.930921820303384,
      "grad_norm": 0.171875,
      "learning_rate": 7.118486170237616e-05,
      "loss": 0.0008,
      "step": 12707
    },
    {
      "epoch": 5.931388564760794,
      "grad_norm": 0.019287109375,
      "learning_rate": 7.117081090760136e-05,
      "loss": 0.0003,
      "step": 12708
    },
    {
      "epoch": 5.931855309218203,
      "grad_norm": 0.171875,
      "learning_rate": 7.115676073355095e-05,
      "loss": 0.0006,
      "step": 12709
    },
    {
      "epoch": 5.9323220536756125,
      "grad_norm": 0.03564453125,
      "learning_rate": 7.114271118052752e-05,
      "loss": 0.0004,
      "step": 12710
    },
    {
      "epoch": 5.932788798133022,
      "grad_norm": 0.0135498046875,
      "learning_rate": 7.112866224883353e-05,
      "loss": 0.0002,
      "step": 12711
    },
    {
      "epoch": 5.933255542590432,
      "grad_norm": 0.10791015625,
      "learning_rate": 7.111461393877145e-05,
      "loss": 0.0029,
      "step": 12712
    },
    {
      "epoch": 5.933722287047841,
      "grad_norm": 0.14453125,
      "learning_rate": 7.110056625064382e-05,
      "loss": 0.0006,
      "step": 12713
    },
    {
      "epoch": 5.934189031505251,
      "grad_norm": 0.033203125,
      "learning_rate": 7.108651918475304e-05,
      "loss": 0.002,
      "step": 12714
    },
    {
      "epoch": 5.934655775962661,
      "grad_norm": 0.024169921875,
      "learning_rate": 7.107247274140161e-05,
      "loss": 0.0003,
      "step": 12715
    },
    {
      "epoch": 5.93512252042007,
      "grad_norm": 0.038818359375,
      "learning_rate": 7.105842692089189e-05,
      "loss": 0.0004,
      "step": 12716
    },
    {
      "epoch": 5.935589264877479,
      "grad_norm": 0.06201171875,
      "learning_rate": 7.104438172352638e-05,
      "loss": 0.0005,
      "step": 12717
    },
    {
      "epoch": 5.936056009334889,
      "grad_norm": 0.09423828125,
      "learning_rate": 7.103033714960746e-05,
      "loss": 0.0029,
      "step": 12718
    },
    {
      "epoch": 5.936522753792299,
      "grad_norm": 0.05322265625,
      "learning_rate": 7.10162931994375e-05,
      "loss": 0.0004,
      "step": 12719
    },
    {
      "epoch": 5.936989498249709,
      "grad_norm": 0.0966796875,
      "learning_rate": 7.100224987331894e-05,
      "loss": 0.0005,
      "step": 12720
    },
    {
      "epoch": 5.937456242707118,
      "grad_norm": 0.050537109375,
      "learning_rate": 7.09882071715541e-05,
      "loss": 0.0004,
      "step": 12721
    },
    {
      "epoch": 5.937922987164527,
      "grad_norm": 0.05615234375,
      "learning_rate": 7.097416509444532e-05,
      "loss": 0.0004,
      "step": 12722
    },
    {
      "epoch": 5.938389731621937,
      "grad_norm": 0.033935546875,
      "learning_rate": 7.096012364229501e-05,
      "loss": 0.0004,
      "step": 12723
    },
    {
      "epoch": 5.938856476079346,
      "grad_norm": 0.03369140625,
      "learning_rate": 7.094608281540545e-05,
      "loss": 0.0004,
      "step": 12724
    },
    {
      "epoch": 5.939323220536756,
      "grad_norm": 0.1533203125,
      "learning_rate": 7.093204261407895e-05,
      "loss": 0.0008,
      "step": 12725
    },
    {
      "epoch": 5.939789964994166,
      "grad_norm": 0.08251953125,
      "learning_rate": 7.091800303861782e-05,
      "loss": 0.0048,
      "step": 12726
    },
    {
      "epoch": 5.9402567094515755,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.090396408932436e-05,
      "loss": 0.0043,
      "step": 12727
    },
    {
      "epoch": 5.940723453908985,
      "grad_norm": 0.2578125,
      "learning_rate": 7.088992576650084e-05,
      "loss": 0.0032,
      "step": 12728
    },
    {
      "epoch": 5.941190198366394,
      "grad_norm": 0.03857421875,
      "learning_rate": 7.087588807044951e-05,
      "loss": 0.0004,
      "step": 12729
    },
    {
      "epoch": 5.941656942823804,
      "grad_norm": 0.259765625,
      "learning_rate": 7.086185100147262e-05,
      "loss": 0.0009,
      "step": 12730
    },
    {
      "epoch": 5.942123687281214,
      "grad_norm": 0.2158203125,
      "learning_rate": 7.084781455987243e-05,
      "loss": 0.0045,
      "step": 12731
    },
    {
      "epoch": 5.942590431738623,
      "grad_norm": 0.076171875,
      "learning_rate": 7.083377874595112e-05,
      "loss": 0.0004,
      "step": 12732
    },
    {
      "epoch": 5.9430571761960325,
      "grad_norm": 0.212890625,
      "learning_rate": 7.081974356001091e-05,
      "loss": 0.0018,
      "step": 12733
    },
    {
      "epoch": 5.943523920653442,
      "grad_norm": 0.022705078125,
      "learning_rate": 7.0805709002354e-05,
      "loss": 0.0003,
      "step": 12734
    },
    {
      "epoch": 5.943990665110852,
      "grad_norm": 0.028564453125,
      "learning_rate": 7.079167507328257e-05,
      "loss": 0.0003,
      "step": 12735
    },
    {
      "epoch": 5.944457409568262,
      "grad_norm": 0.11767578125,
      "learning_rate": 7.077764177309878e-05,
      "loss": 0.0009,
      "step": 12736
    },
    {
      "epoch": 5.944924154025671,
      "grad_norm": 0.050048828125,
      "learning_rate": 7.076360910210479e-05,
      "loss": 0.0003,
      "step": 12737
    },
    {
      "epoch": 5.9453908984830806,
      "grad_norm": 0.1767578125,
      "learning_rate": 7.074957706060275e-05,
      "loss": 0.0007,
      "step": 12738
    },
    {
      "epoch": 5.94585764294049,
      "grad_norm": 0.01165771484375,
      "learning_rate": 7.073554564889473e-05,
      "loss": 0.0002,
      "step": 12739
    },
    {
      "epoch": 5.946324387397899,
      "grad_norm": 0.08203125,
      "learning_rate": 7.072151486728291e-05,
      "loss": 0.0009,
      "step": 12740
    },
    {
      "epoch": 5.946791131855309,
      "grad_norm": 0.039794921875,
      "learning_rate": 7.070748471606936e-05,
      "loss": 0.0025,
      "step": 12741
    },
    {
      "epoch": 5.947257876312719,
      "grad_norm": 0.038818359375,
      "learning_rate": 7.069345519555616e-05,
      "loss": 0.0003,
      "step": 12742
    },
    {
      "epoch": 5.947724620770129,
      "grad_norm": 0.0238037109375,
      "learning_rate": 7.067942630604536e-05,
      "loss": 0.0004,
      "step": 12743
    },
    {
      "epoch": 5.9481913652275376,
      "grad_norm": 0.1669921875,
      "learning_rate": 7.066539804783908e-05,
      "loss": 0.001,
      "step": 12744
    },
    {
      "epoch": 5.948658109684947,
      "grad_norm": 0.01953125,
      "learning_rate": 7.065137042123932e-05,
      "loss": 0.0003,
      "step": 12745
    },
    {
      "epoch": 5.949124854142357,
      "grad_norm": 0.1298828125,
      "learning_rate": 7.06373434265481e-05,
      "loss": 0.0027,
      "step": 12746
    },
    {
      "epoch": 5.949591598599767,
      "grad_norm": 0.1640625,
      "learning_rate": 7.062331706406751e-05,
      "loss": 0.0007,
      "step": 12747
    },
    {
      "epoch": 5.950058343057176,
      "grad_norm": 0.1728515625,
      "learning_rate": 7.060929133409944e-05,
      "loss": 0.0006,
      "step": 12748
    },
    {
      "epoch": 5.950525087514586,
      "grad_norm": 0.173828125,
      "learning_rate": 7.059526623694597e-05,
      "loss": 0.0006,
      "step": 12749
    },
    {
      "epoch": 5.950991831971995,
      "grad_norm": 0.2060546875,
      "learning_rate": 7.058124177290902e-05,
      "loss": 0.001,
      "step": 12750
    },
    {
      "epoch": 5.951458576429405,
      "grad_norm": 0.0673828125,
      "learning_rate": 7.056721794229057e-05,
      "loss": 0.0005,
      "step": 12751
    },
    {
      "epoch": 5.951925320886814,
      "grad_norm": 0.06982421875,
      "learning_rate": 7.05531947453926e-05,
      "loss": 0.0005,
      "step": 12752
    },
    {
      "epoch": 5.952392065344224,
      "grad_norm": 0.05517578125,
      "learning_rate": 7.053917218251699e-05,
      "loss": 0.0004,
      "step": 12753
    },
    {
      "epoch": 5.952858809801634,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.052515025396572e-05,
      "loss": 0.0039,
      "step": 12754
    },
    {
      "epoch": 5.9533255542590435,
      "grad_norm": 0.01422119140625,
      "learning_rate": 7.051112896004063e-05,
      "loss": 0.0002,
      "step": 12755
    },
    {
      "epoch": 5.953792298716452,
      "grad_norm": 0.03173828125,
      "learning_rate": 7.049710830104368e-05,
      "loss": 0.0003,
      "step": 12756
    },
    {
      "epoch": 5.954259043173862,
      "grad_norm": 0.043701171875,
      "learning_rate": 7.048308827727672e-05,
      "loss": 0.0004,
      "step": 12757
    },
    {
      "epoch": 5.954725787631272,
      "grad_norm": 0.035888671875,
      "learning_rate": 7.04690688890416e-05,
      "loss": 0.0003,
      "step": 12758
    },
    {
      "epoch": 5.955192532088682,
      "grad_norm": 0.035888671875,
      "learning_rate": 7.045505013664022e-05,
      "loss": 0.0003,
      "step": 12759
    },
    {
      "epoch": 5.955659276546091,
      "grad_norm": 0.052490234375,
      "learning_rate": 7.044103202037437e-05,
      "loss": 0.0034,
      "step": 12760
    },
    {
      "epoch": 5.9561260210035005,
      "grad_norm": 0.03662109375,
      "learning_rate": 7.04270145405459e-05,
      "loss": 0.0004,
      "step": 12761
    },
    {
      "epoch": 5.95659276546091,
      "grad_norm": 0.03466796875,
      "learning_rate": 7.041299769745662e-05,
      "loss": 0.0004,
      "step": 12762
    },
    {
      "epoch": 5.95705950991832,
      "grad_norm": 0.07568359375,
      "learning_rate": 7.039898149140834e-05,
      "loss": 0.0047,
      "step": 12763
    },
    {
      "epoch": 5.957526254375729,
      "grad_norm": 0.028076171875,
      "learning_rate": 7.038496592270283e-05,
      "loss": 0.0003,
      "step": 12764
    },
    {
      "epoch": 5.957992998833139,
      "grad_norm": 0.0238037109375,
      "learning_rate": 7.037095099164183e-05,
      "loss": 0.0003,
      "step": 12765
    },
    {
      "epoch": 5.958459743290549,
      "grad_norm": 0.047119140625,
      "learning_rate": 7.035693669852716e-05,
      "loss": 0.0004,
      "step": 12766
    },
    {
      "epoch": 5.9589264877479575,
      "grad_norm": 0.040771484375,
      "learning_rate": 7.034292304366053e-05,
      "loss": 0.0004,
      "step": 12767
    },
    {
      "epoch": 5.959393232205367,
      "grad_norm": 0.0167236328125,
      "learning_rate": 7.032891002734365e-05,
      "loss": 0.0003,
      "step": 12768
    },
    {
      "epoch": 5.959859976662777,
      "grad_norm": 0.10546875,
      "learning_rate": 7.031489764987829e-05,
      "loss": 0.0004,
      "step": 12769
    },
    {
      "epoch": 5.960326721120187,
      "grad_norm": 0.027099609375,
      "learning_rate": 7.030088591156611e-05,
      "loss": 0.0004,
      "step": 12770
    },
    {
      "epoch": 5.960793465577597,
      "grad_norm": 0.0191650390625,
      "learning_rate": 7.028687481270879e-05,
      "loss": 0.0003,
      "step": 12771
    },
    {
      "epoch": 5.961260210035006,
      "grad_norm": 0.046142578125,
      "learning_rate": 7.027286435360805e-05,
      "loss": 0.0004,
      "step": 12772
    },
    {
      "epoch": 5.961726954492415,
      "grad_norm": 0.072265625,
      "learning_rate": 7.025885453456551e-05,
      "loss": 0.0005,
      "step": 12773
    },
    {
      "epoch": 5.962193698949825,
      "grad_norm": 0.035888671875,
      "learning_rate": 7.024484535588283e-05,
      "loss": 0.0004,
      "step": 12774
    },
    {
      "epoch": 5.962660443407234,
      "grad_norm": 0.07958984375,
      "learning_rate": 7.023083681786162e-05,
      "loss": 0.0004,
      "step": 12775
    },
    {
      "epoch": 5.963127187864644,
      "grad_norm": 0.62890625,
      "learning_rate": 7.021682892080356e-05,
      "loss": 0.0025,
      "step": 12776
    },
    {
      "epoch": 5.963593932322054,
      "grad_norm": 0.3203125,
      "learning_rate": 7.02028216650102e-05,
      "loss": 0.0015,
      "step": 12777
    },
    {
      "epoch": 5.9640606767794635,
      "grad_norm": 0.146484375,
      "learning_rate": 7.018881505078314e-05,
      "loss": 0.0035,
      "step": 12778
    },
    {
      "epoch": 5.964527421236873,
      "grad_norm": 0.197265625,
      "learning_rate": 7.017480907842399e-05,
      "loss": 0.0007,
      "step": 12779
    },
    {
      "epoch": 5.964994165694282,
      "grad_norm": 0.0216064453125,
      "learning_rate": 7.016080374823429e-05,
      "loss": 0.0003,
      "step": 12780
    },
    {
      "epoch": 5.965460910151692,
      "grad_norm": 0.08984375,
      "learning_rate": 7.014679906051558e-05,
      "loss": 0.001,
      "step": 12781
    },
    {
      "epoch": 5.965927654609102,
      "grad_norm": 0.2216796875,
      "learning_rate": 7.013279501556942e-05,
      "loss": 0.0024,
      "step": 12782
    },
    {
      "epoch": 5.966394399066511,
      "grad_norm": 0.1494140625,
      "learning_rate": 7.011879161369732e-05,
      "loss": 0.0008,
      "step": 12783
    },
    {
      "epoch": 5.9668611435239205,
      "grad_norm": 0.166015625,
      "learning_rate": 7.01047888552008e-05,
      "loss": 0.0027,
      "step": 12784
    },
    {
      "epoch": 5.96732788798133,
      "grad_norm": 0.11083984375,
      "learning_rate": 7.009078674038131e-05,
      "loss": 0.0004,
      "step": 12785
    },
    {
      "epoch": 5.96779463243874,
      "grad_norm": 0.05810546875,
      "learning_rate": 7.007678526954039e-05,
      "loss": 0.0005,
      "step": 12786
    },
    {
      "epoch": 5.968261376896149,
      "grad_norm": 0.1259765625,
      "learning_rate": 7.00627844429795e-05,
      "loss": 0.0004,
      "step": 12787
    },
    {
      "epoch": 5.968728121353559,
      "grad_norm": 0.119140625,
      "learning_rate": 7.004878426100003e-05,
      "loss": 0.0007,
      "step": 12788
    },
    {
      "epoch": 5.969194865810969,
      "grad_norm": 0.1689453125,
      "learning_rate": 7.003478472390351e-05,
      "loss": 0.0007,
      "step": 12789
    },
    {
      "epoch": 5.969661610268378,
      "grad_norm": 0.09326171875,
      "learning_rate": 7.002078583199131e-05,
      "loss": 0.0036,
      "step": 12790
    },
    {
      "epoch": 5.970128354725787,
      "grad_norm": 0.1806640625,
      "learning_rate": 7.000678758556485e-05,
      "loss": 0.0007,
      "step": 12791
    },
    {
      "epoch": 5.970595099183197,
      "grad_norm": 0.06689453125,
      "learning_rate": 6.999278998492553e-05,
      "loss": 0.0005,
      "step": 12792
    },
    {
      "epoch": 5.971061843640607,
      "grad_norm": 0.103515625,
      "learning_rate": 6.997879303037474e-05,
      "loss": 0.0005,
      "step": 12793
    },
    {
      "epoch": 5.971528588098017,
      "grad_norm": 0.11279296875,
      "learning_rate": 6.996479672221386e-05,
      "loss": 0.0007,
      "step": 12794
    },
    {
      "epoch": 5.971995332555426,
      "grad_norm": 0.154296875,
      "learning_rate": 6.995080106074423e-05,
      "loss": 0.0008,
      "step": 12795
    },
    {
      "epoch": 5.972462077012835,
      "grad_norm": 0.044677734375,
      "learning_rate": 6.993680604626718e-05,
      "loss": 0.0004,
      "step": 12796
    },
    {
      "epoch": 5.972928821470245,
      "grad_norm": 0.10693359375,
      "learning_rate": 6.992281167908404e-05,
      "loss": 0.0006,
      "step": 12797
    },
    {
      "epoch": 5.973395565927655,
      "grad_norm": 0.041259765625,
      "learning_rate": 6.990881795949614e-05,
      "loss": 0.0003,
      "step": 12798
    },
    {
      "epoch": 5.973862310385064,
      "grad_norm": 0.095703125,
      "learning_rate": 6.989482488780478e-05,
      "loss": 0.0006,
      "step": 12799
    },
    {
      "epoch": 5.974329054842474,
      "grad_norm": 0.0498046875,
      "learning_rate": 6.988083246431123e-05,
      "loss": 0.0004,
      "step": 12800
    },
    {
      "epoch": 5.9747957992998835,
      "grad_norm": 0.024169921875,
      "learning_rate": 6.98668406893168e-05,
      "loss": 0.0003,
      "step": 12801
    },
    {
      "epoch": 5.975262543757293,
      "grad_norm": 0.046142578125,
      "learning_rate": 6.985284956312271e-05,
      "loss": 0.0003,
      "step": 12802
    },
    {
      "epoch": 5.975729288214702,
      "grad_norm": 0.07763671875,
      "learning_rate": 6.983885908603022e-05,
      "loss": 0.0004,
      "step": 12803
    },
    {
      "epoch": 5.976196032672112,
      "grad_norm": 0.0625,
      "learning_rate": 6.982486925834055e-05,
      "loss": 0.0004,
      "step": 12804
    },
    {
      "epoch": 5.976662777129522,
      "grad_norm": 0.0235595703125,
      "learning_rate": 6.981088008035493e-05,
      "loss": 0.0002,
      "step": 12805
    },
    {
      "epoch": 5.977129521586932,
      "grad_norm": 0.302734375,
      "learning_rate": 6.979689155237456e-05,
      "loss": 0.0033,
      "step": 12806
    },
    {
      "epoch": 5.9775962660443405,
      "grad_norm": 0.035888671875,
      "learning_rate": 6.978290367470061e-05,
      "loss": 0.0004,
      "step": 12807
    },
    {
      "epoch": 5.97806301050175,
      "grad_norm": 0.1494140625,
      "learning_rate": 6.97689164476343e-05,
      "loss": 0.0004,
      "step": 12808
    },
    {
      "epoch": 5.97852975495916,
      "grad_norm": 0.0322265625,
      "learning_rate": 6.975492987147676e-05,
      "loss": 0.0003,
      "step": 12809
    },
    {
      "epoch": 5.978996499416569,
      "grad_norm": 0.10888671875,
      "learning_rate": 6.97409439465291e-05,
      "loss": 0.0006,
      "step": 12810
    },
    {
      "epoch": 5.979463243873979,
      "grad_norm": 0.271484375,
      "learning_rate": 6.972695867309252e-05,
      "loss": 0.0036,
      "step": 12811
    },
    {
      "epoch": 5.979929988331389,
      "grad_norm": 0.037841796875,
      "learning_rate": 6.97129740514681e-05,
      "loss": 0.0004,
      "step": 12812
    },
    {
      "epoch": 5.980396732788798,
      "grad_norm": 0.035400390625,
      "learning_rate": 6.969899008195697e-05,
      "loss": 0.0003,
      "step": 12813
    },
    {
      "epoch": 5.980863477246208,
      "grad_norm": 0.06591796875,
      "learning_rate": 6.968500676486016e-05,
      "loss": 0.0035,
      "step": 12814
    },
    {
      "epoch": 5.981330221703617,
      "grad_norm": 0.376953125,
      "learning_rate": 6.967102410047883e-05,
      "loss": 0.0018,
      "step": 12815
    },
    {
      "epoch": 5.981796966161027,
      "grad_norm": 0.0281982421875,
      "learning_rate": 6.965704208911399e-05,
      "loss": 0.0003,
      "step": 12816
    },
    {
      "epoch": 5.982263710618437,
      "grad_norm": 0.2255859375,
      "learning_rate": 6.964306073106668e-05,
      "loss": 0.0038,
      "step": 12817
    },
    {
      "epoch": 5.9827304550758456,
      "grad_norm": 0.04638671875,
      "learning_rate": 6.962908002663799e-05,
      "loss": 0.0003,
      "step": 12818
    },
    {
      "epoch": 5.983197199533255,
      "grad_norm": 0.0257568359375,
      "learning_rate": 6.96150999761289e-05,
      "loss": 0.0004,
      "step": 12819
    },
    {
      "epoch": 5.983663943990665,
      "grad_norm": 0.0296630859375,
      "learning_rate": 6.960112057984038e-05,
      "loss": 0.0003,
      "step": 12820
    },
    {
      "epoch": 5.984130688448075,
      "grad_norm": 0.08935546875,
      "learning_rate": 6.958714183807348e-05,
      "loss": 0.0006,
      "step": 12821
    },
    {
      "epoch": 5.984597432905485,
      "grad_norm": 0.03955078125,
      "learning_rate": 6.957316375112919e-05,
      "loss": 0.002,
      "step": 12822
    },
    {
      "epoch": 5.985064177362894,
      "grad_norm": 0.0296630859375,
      "learning_rate": 6.955918631930843e-05,
      "loss": 0.0003,
      "step": 12823
    },
    {
      "epoch": 5.985530921820303,
      "grad_norm": 0.11279296875,
      "learning_rate": 6.954520954291214e-05,
      "loss": 0.0005,
      "step": 12824
    },
    {
      "epoch": 5.985997666277713,
      "grad_norm": 0.02734375,
      "learning_rate": 6.953123342224129e-05,
      "loss": 0.0003,
      "step": 12825
    },
    {
      "epoch": 5.986464410735122,
      "grad_norm": 0.173828125,
      "learning_rate": 6.95172579575968e-05,
      "loss": 0.0008,
      "step": 12826
    },
    {
      "epoch": 5.986931155192532,
      "grad_norm": 0.068359375,
      "learning_rate": 6.950328314927955e-05,
      "loss": 0.0005,
      "step": 12827
    },
    {
      "epoch": 5.987397899649942,
      "grad_norm": 0.052001953125,
      "learning_rate": 6.948930899759048e-05,
      "loss": 0.0032,
      "step": 12828
    },
    {
      "epoch": 5.9878646441073515,
      "grad_norm": 0.020751953125,
      "learning_rate": 6.947533550283041e-05,
      "loss": 0.0003,
      "step": 12829
    },
    {
      "epoch": 5.98833138856476,
      "grad_norm": 0.054931640625,
      "learning_rate": 6.946136266530023e-05,
      "loss": 0.0004,
      "step": 12830
    },
    {
      "epoch": 5.98879813302217,
      "grad_norm": 0.0673828125,
      "learning_rate": 6.944739048530082e-05,
      "loss": 0.0026,
      "step": 12831
    },
    {
      "epoch": 5.98926487747958,
      "grad_norm": 0.349609375,
      "learning_rate": 6.9433418963133e-05,
      "loss": 0.0013,
      "step": 12832
    },
    {
      "epoch": 5.98973162193699,
      "grad_norm": 0.054443359375,
      "learning_rate": 6.941944809909756e-05,
      "loss": 0.0005,
      "step": 12833
    },
    {
      "epoch": 5.990198366394399,
      "grad_norm": 0.0142822265625,
      "learning_rate": 6.940547789349533e-05,
      "loss": 0.0002,
      "step": 12834
    },
    {
      "epoch": 5.9906651108518085,
      "grad_norm": 0.07373046875,
      "learning_rate": 6.939150834662711e-05,
      "loss": 0.0006,
      "step": 12835
    },
    {
      "epoch": 5.991131855309218,
      "grad_norm": 0.040283203125,
      "learning_rate": 6.93775394587937e-05,
      "loss": 0.0004,
      "step": 12836
    },
    {
      "epoch": 5.991598599766628,
      "grad_norm": 0.2890625,
      "learning_rate": 6.936357123029579e-05,
      "loss": 0.0008,
      "step": 12837
    },
    {
      "epoch": 5.992065344224037,
      "grad_norm": 0.1904296875,
      "learning_rate": 6.934960366143424e-05,
      "loss": 0.001,
      "step": 12838
    },
    {
      "epoch": 5.992532088681447,
      "grad_norm": 0.2001953125,
      "learning_rate": 6.93356367525097e-05,
      "loss": 0.0008,
      "step": 12839
    },
    {
      "epoch": 5.992998833138857,
      "grad_norm": 0.1376953125,
      "learning_rate": 6.932167050382294e-05,
      "loss": 0.0007,
      "step": 12840
    },
    {
      "epoch": 5.993465577596266,
      "grad_norm": 0.1796875,
      "learning_rate": 6.930770491567463e-05,
      "loss": 0.0046,
      "step": 12841
    },
    {
      "epoch": 5.993932322053675,
      "grad_norm": 0.042236328125,
      "learning_rate": 6.929373998836552e-05,
      "loss": 0.0004,
      "step": 12842
    },
    {
      "epoch": 5.994399066511085,
      "grad_norm": 0.2041015625,
      "learning_rate": 6.927977572219623e-05,
      "loss": 0.0014,
      "step": 12843
    },
    {
      "epoch": 5.994865810968495,
      "grad_norm": 0.0281982421875,
      "learning_rate": 6.926581211746748e-05,
      "loss": 0.0003,
      "step": 12844
    },
    {
      "epoch": 5.995332555425904,
      "grad_norm": 0.05859375,
      "learning_rate": 6.925184917447987e-05,
      "loss": 0.0003,
      "step": 12845
    },
    {
      "epoch": 5.995799299883314,
      "grad_norm": 0.018798828125,
      "learning_rate": 6.923788689353405e-05,
      "loss": 0.0002,
      "step": 12846
    },
    {
      "epoch": 5.996266044340723,
      "grad_norm": 0.08642578125,
      "learning_rate": 6.922392527493069e-05,
      "loss": 0.0005,
      "step": 12847
    },
    {
      "epoch": 5.996732788798133,
      "grad_norm": 0.0179443359375,
      "learning_rate": 6.920996431897036e-05,
      "loss": 0.0003,
      "step": 12848
    },
    {
      "epoch": 5.997199533255543,
      "grad_norm": 0.04541015625,
      "learning_rate": 6.919600402595364e-05,
      "loss": 0.0003,
      "step": 12849
    },
    {
      "epoch": 5.997666277712952,
      "grad_norm": 0.04296875,
      "learning_rate": 6.918204439618114e-05,
      "loss": 0.0014,
      "step": 12850
    },
    {
      "epoch": 5.998133022170362,
      "grad_norm": 0.9296875,
      "learning_rate": 6.916808542995341e-05,
      "loss": 0.0193,
      "step": 12851
    },
    {
      "epoch": 5.9985997666277715,
      "grad_norm": 0.1259765625,
      "learning_rate": 6.915412712757104e-05,
      "loss": 0.0005,
      "step": 12852
    },
    {
      "epoch": 5.99906651108518,
      "grad_norm": 0.158203125,
      "learning_rate": 6.914016948933448e-05,
      "loss": 0.0006,
      "step": 12853
    },
    {
      "epoch": 5.99953325554259,
      "grad_norm": 0.24609375,
      "learning_rate": 6.912621251554435e-05,
      "loss": 0.0018,
      "step": 12854
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.062255859375,
      "learning_rate": 6.911225620650112e-05,
      "loss": 0.0004,
      "step": 12855
    },
    {
      "epoch": 6.00046674445741,
      "grad_norm": 0.0247802734375,
      "learning_rate": 6.909830056250527e-05,
      "loss": 0.0003,
      "step": 12856
    },
    {
      "epoch": 6.000933488914819,
      "grad_norm": 0.384765625,
      "learning_rate": 6.90843455838573e-05,
      "loss": 0.0037,
      "step": 12857
    },
    {
      "epoch": 6.0014002333722285,
      "grad_norm": 0.035400390625,
      "learning_rate": 6.907039127085769e-05,
      "loss": 0.0002,
      "step": 12858
    },
    {
      "epoch": 6.001866977829638,
      "grad_norm": 0.054443359375,
      "learning_rate": 6.905643762380685e-05,
      "loss": 0.0019,
      "step": 12859
    },
    {
      "epoch": 6.002333722287048,
      "grad_norm": 0.11376953125,
      "learning_rate": 6.904248464300525e-05,
      "loss": 0.0025,
      "step": 12860
    },
    {
      "epoch": 6.002800466744457,
      "grad_norm": 0.07275390625,
      "learning_rate": 6.90285323287533e-05,
      "loss": 0.0004,
      "step": 12861
    },
    {
      "epoch": 6.003267211201867,
      "grad_norm": 0.030517578125,
      "learning_rate": 6.901458068135143e-05,
      "loss": 0.0003,
      "step": 12862
    },
    {
      "epoch": 6.003733955659277,
      "grad_norm": 0.1865234375,
      "learning_rate": 6.900062970109998e-05,
      "loss": 0.0008,
      "step": 12863
    },
    {
      "epoch": 6.004200700116686,
      "grad_norm": 0.353515625,
      "learning_rate": 6.89866793882994e-05,
      "loss": 0.0067,
      "step": 12864
    },
    {
      "epoch": 6.004200700116686,
      "eval_loss": 2.306922435760498,
      "eval_runtime": 56.2567,
      "eval_samples_per_second": 32.067,
      "eval_steps_per_second": 4.017,
      "step": 12864
    },
    {
      "epoch": 6.004667444574095,
      "grad_norm": 0.06787109375,
      "learning_rate": 6.897272974325e-05,
      "loss": 0.0004,
      "step": 12865
    },
    {
      "epoch": 6.005134189031505,
      "grad_norm": 0.04052734375,
      "learning_rate": 6.895878076625215e-05,
      "loss": 0.0004,
      "step": 12866
    },
    {
      "epoch": 6.005600933488915,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.894483245760621e-05,
      "loss": 0.0014,
      "step": 12867
    },
    {
      "epoch": 6.006067677946325,
      "grad_norm": 0.056640625,
      "learning_rate": 6.893088481761248e-05,
      "loss": 0.0005,
      "step": 12868
    },
    {
      "epoch": 6.006534422403734,
      "grad_norm": 0.2412109375,
      "learning_rate": 6.891693784657127e-05,
      "loss": 0.0023,
      "step": 12869
    },
    {
      "epoch": 6.007001166861143,
      "grad_norm": 0.03076171875,
      "learning_rate": 6.890299154478288e-05,
      "loss": 0.0003,
      "step": 12870
    },
    {
      "epoch": 6.007467911318553,
      "grad_norm": 0.12060546875,
      "learning_rate": 6.888904591254758e-05,
      "loss": 0.0005,
      "step": 12871
    },
    {
      "epoch": 6.007934655775963,
      "grad_norm": 0.0260009765625,
      "learning_rate": 6.887510095016565e-05,
      "loss": 0.0003,
      "step": 12872
    },
    {
      "epoch": 6.008401400233372,
      "grad_norm": 0.55078125,
      "learning_rate": 6.88611566579373e-05,
      "loss": 0.0045,
      "step": 12873
    },
    {
      "epoch": 6.008868144690782,
      "grad_norm": 0.05029296875,
      "learning_rate": 6.884721303616282e-05,
      "loss": 0.0004,
      "step": 12874
    },
    {
      "epoch": 6.0093348891481915,
      "grad_norm": 0.033935546875,
      "learning_rate": 6.883327008514241e-05,
      "loss": 0.0003,
      "step": 12875
    },
    {
      "epoch": 6.009801633605601,
      "grad_norm": 0.0303955078125,
      "learning_rate": 6.881932780517625e-05,
      "loss": 0.001,
      "step": 12876
    },
    {
      "epoch": 6.01026837806301,
      "grad_norm": 0.0291748046875,
      "learning_rate": 6.880538619656459e-05,
      "loss": 0.0004,
      "step": 12877
    },
    {
      "epoch": 6.01073512252042,
      "grad_norm": 0.31640625,
      "learning_rate": 6.879144525960757e-05,
      "loss": 0.0039,
      "step": 12878
    },
    {
      "epoch": 6.01120186697783,
      "grad_norm": 0.0308837890625,
      "learning_rate": 6.877750499460533e-05,
      "loss": 0.0003,
      "step": 12879
    },
    {
      "epoch": 6.01166861143524,
      "grad_norm": 0.0306396484375,
      "learning_rate": 6.876356540185808e-05,
      "loss": 0.0003,
      "step": 12880
    },
    {
      "epoch": 6.0121353558926485,
      "grad_norm": 0.02001953125,
      "learning_rate": 6.874962648166593e-05,
      "loss": 0.0003,
      "step": 12881
    },
    {
      "epoch": 6.012602100350058,
      "grad_norm": 0.01470947265625,
      "learning_rate": 6.873568823432898e-05,
      "loss": 0.0002,
      "step": 12882
    },
    {
      "epoch": 6.013068844807468,
      "grad_norm": 0.302734375,
      "learning_rate": 6.872175066014734e-05,
      "loss": 0.0011,
      "step": 12883
    },
    {
      "epoch": 6.013535589264878,
      "grad_norm": 0.0208740234375,
      "learning_rate": 6.870781375942114e-05,
      "loss": 0.0003,
      "step": 12884
    },
    {
      "epoch": 6.014002333722287,
      "grad_norm": 0.1884765625,
      "learning_rate": 6.869387753245042e-05,
      "loss": 0.0012,
      "step": 12885
    },
    {
      "epoch": 6.014469078179697,
      "grad_norm": 0.2412109375,
      "learning_rate": 6.867994197953522e-05,
      "loss": 0.0033,
      "step": 12886
    },
    {
      "epoch": 6.014935822637106,
      "grad_norm": 0.01708984375,
      "learning_rate": 6.866600710097566e-05,
      "loss": 0.0003,
      "step": 12887
    },
    {
      "epoch": 6.015402567094516,
      "grad_norm": 0.050537109375,
      "learning_rate": 6.865207289707171e-05,
      "loss": 0.0011,
      "step": 12888
    },
    {
      "epoch": 6.015869311551925,
      "grad_norm": 0.0211181640625,
      "learning_rate": 6.863813936812342e-05,
      "loss": 0.0003,
      "step": 12889
    },
    {
      "epoch": 6.016336056009335,
      "grad_norm": 0.01806640625,
      "learning_rate": 6.862420651443083e-05,
      "loss": 0.0003,
      "step": 12890
    },
    {
      "epoch": 6.016802800466745,
      "grad_norm": 0.0224609375,
      "learning_rate": 6.861027433629382e-05,
      "loss": 0.0002,
      "step": 12891
    },
    {
      "epoch": 6.0172695449241544,
      "grad_norm": 0.029296875,
      "learning_rate": 6.859634283401246e-05,
      "loss": 0.0003,
      "step": 12892
    },
    {
      "epoch": 6.017736289381563,
      "grad_norm": 0.0286865234375,
      "learning_rate": 6.858241200788669e-05,
      "loss": 0.0003,
      "step": 12893
    },
    {
      "epoch": 6.018203033838973,
      "grad_norm": 0.06298828125,
      "learning_rate": 6.856848185821643e-05,
      "loss": 0.0032,
      "step": 12894
    },
    {
      "epoch": 6.018669778296383,
      "grad_norm": 0.470703125,
      "learning_rate": 6.855455238530162e-05,
      "loss": 0.0023,
      "step": 12895
    },
    {
      "epoch": 6.019136522753793,
      "grad_norm": 0.020263671875,
      "learning_rate": 6.85406235894422e-05,
      "loss": 0.0003,
      "step": 12896
    },
    {
      "epoch": 6.019603267211202,
      "grad_norm": 0.0341796875,
      "learning_rate": 6.852669547093807e-05,
      "loss": 0.0004,
      "step": 12897
    },
    {
      "epoch": 6.020070011668611,
      "grad_norm": 0.0771484375,
      "learning_rate": 6.851276803008907e-05,
      "loss": 0.0007,
      "step": 12898
    },
    {
      "epoch": 6.020536756126021,
      "grad_norm": 0.03173828125,
      "learning_rate": 6.849884126719514e-05,
      "loss": 0.0004,
      "step": 12899
    },
    {
      "epoch": 6.02100350058343,
      "grad_norm": 0.1982421875,
      "learning_rate": 6.848491518255611e-05,
      "loss": 0.001,
      "step": 12900
    },
    {
      "epoch": 6.02147024504084,
      "grad_norm": 0.07275390625,
      "learning_rate": 6.847098977647179e-05,
      "loss": 0.0005,
      "step": 12901
    },
    {
      "epoch": 6.02193698949825,
      "grad_norm": 0.21875,
      "learning_rate": 6.845706504924207e-05,
      "loss": 0.0044,
      "step": 12902
    },
    {
      "epoch": 6.0224037339556595,
      "grad_norm": 0.154296875,
      "learning_rate": 6.844314100116675e-05,
      "loss": 0.0109,
      "step": 12903
    },
    {
      "epoch": 6.022870478413068,
      "grad_norm": 0.09619140625,
      "learning_rate": 6.84292176325456e-05,
      "loss": 0.0004,
      "step": 12904
    },
    {
      "epoch": 6.023337222870478,
      "grad_norm": 0.05810546875,
      "learning_rate": 6.841529494367842e-05,
      "loss": 0.0005,
      "step": 12905
    },
    {
      "epoch": 6.023803967327888,
      "grad_norm": 0.0390625,
      "learning_rate": 6.8401372934865e-05,
      "loss": 0.0004,
      "step": 12906
    },
    {
      "epoch": 6.024270711785298,
      "grad_norm": 0.1337890625,
      "learning_rate": 6.838745160640506e-05,
      "loss": 0.0008,
      "step": 12907
    },
    {
      "epoch": 6.024737456242707,
      "grad_norm": 0.041015625,
      "learning_rate": 6.837353095859835e-05,
      "loss": 0.0004,
      "step": 12908
    },
    {
      "epoch": 6.0252042007001165,
      "grad_norm": 0.0654296875,
      "learning_rate": 6.835961099174464e-05,
      "loss": 0.0005,
      "step": 12909
    },
    {
      "epoch": 6.025670945157526,
      "grad_norm": 0.043212890625,
      "learning_rate": 6.83456917061436e-05,
      "loss": 0.0004,
      "step": 12910
    },
    {
      "epoch": 6.026137689614936,
      "grad_norm": 0.06298828125,
      "learning_rate": 6.833177310209495e-05,
      "loss": 0.0004,
      "step": 12911
    },
    {
      "epoch": 6.026604434072345,
      "grad_norm": 0.185546875,
      "learning_rate": 6.831785517989832e-05,
      "loss": 0.0013,
      "step": 12912
    },
    {
      "epoch": 6.027071178529755,
      "grad_norm": 0.0311279296875,
      "learning_rate": 6.830393793985346e-05,
      "loss": 0.0004,
      "step": 12913
    },
    {
      "epoch": 6.027537922987165,
      "grad_norm": 0.259765625,
      "learning_rate": 6.829002138225999e-05,
      "loss": 0.0046,
      "step": 12914
    },
    {
      "epoch": 6.028004667444574,
      "grad_norm": 0.0203857421875,
      "learning_rate": 6.82761055074175e-05,
      "loss": 0.0003,
      "step": 12915
    },
    {
      "epoch": 6.028471411901983,
      "grad_norm": 0.052001953125,
      "learning_rate": 6.826219031562569e-05,
      "loss": 0.0004,
      "step": 12916
    },
    {
      "epoch": 6.028938156359393,
      "grad_norm": 0.107421875,
      "learning_rate": 6.824827580718414e-05,
      "loss": 0.0007,
      "step": 12917
    },
    {
      "epoch": 6.029404900816803,
      "grad_norm": 0.04736328125,
      "learning_rate": 6.823436198239241e-05,
      "loss": 0.0045,
      "step": 12918
    },
    {
      "epoch": 6.029871645274213,
      "grad_norm": 0.099609375,
      "learning_rate": 6.822044884155013e-05,
      "loss": 0.0005,
      "step": 12919
    },
    {
      "epoch": 6.030338389731622,
      "grad_norm": 0.1064453125,
      "learning_rate": 6.820653638495685e-05,
      "loss": 0.0009,
      "step": 12920
    },
    {
      "epoch": 6.030805134189031,
      "grad_norm": 0.11572265625,
      "learning_rate": 6.819262461291212e-05,
      "loss": 0.0009,
      "step": 12921
    },
    {
      "epoch": 6.031271878646441,
      "grad_norm": 0.12158203125,
      "learning_rate": 6.817871352571545e-05,
      "loss": 0.0006,
      "step": 12922
    },
    {
      "epoch": 6.031738623103851,
      "grad_norm": 0.09033203125,
      "learning_rate": 6.816480312366641e-05,
      "loss": 0.0025,
      "step": 12923
    },
    {
      "epoch": 6.03220536756126,
      "grad_norm": 0.1796875,
      "learning_rate": 6.815089340706448e-05,
      "loss": 0.0031,
      "step": 12924
    },
    {
      "epoch": 6.03267211201867,
      "grad_norm": 0.08837890625,
      "learning_rate": 6.813698437620913e-05,
      "loss": 0.0024,
      "step": 12925
    },
    {
      "epoch": 6.0331388564760795,
      "grad_norm": 0.1005859375,
      "learning_rate": 6.812307603139987e-05,
      "loss": 0.0006,
      "step": 12926
    },
    {
      "epoch": 6.033605600933489,
      "grad_norm": 0.109375,
      "learning_rate": 6.810916837293617e-05,
      "loss": 0.0006,
      "step": 12927
    },
    {
      "epoch": 6.034072345390898,
      "grad_norm": 0.035888671875,
      "learning_rate": 6.809526140111742e-05,
      "loss": 0.0003,
      "step": 12928
    },
    {
      "epoch": 6.034539089848308,
      "grad_norm": 0.07958984375,
      "learning_rate": 6.808135511624312e-05,
      "loss": 0.0004,
      "step": 12929
    },
    {
      "epoch": 6.035005834305718,
      "grad_norm": 0.05517578125,
      "learning_rate": 6.806744951861265e-05,
      "loss": 0.0004,
      "step": 12930
    },
    {
      "epoch": 6.035472578763128,
      "grad_norm": 0.026611328125,
      "learning_rate": 6.805354460852544e-05,
      "loss": 0.0003,
      "step": 12931
    },
    {
      "epoch": 6.0359393232205365,
      "grad_norm": 0.33984375,
      "learning_rate": 6.803964038628082e-05,
      "loss": 0.0008,
      "step": 12932
    },
    {
      "epoch": 6.036406067677946,
      "grad_norm": 0.0242919921875,
      "learning_rate": 6.802573685217823e-05,
      "loss": 0.0003,
      "step": 12933
    },
    {
      "epoch": 6.036872812135356,
      "grad_norm": 0.029296875,
      "learning_rate": 6.8011834006517e-05,
      "loss": 0.0004,
      "step": 12934
    },
    {
      "epoch": 6.037339556592766,
      "grad_norm": 0.055419921875,
      "learning_rate": 6.799793184959646e-05,
      "loss": 0.0004,
      "step": 12935
    },
    {
      "epoch": 6.037806301050175,
      "grad_norm": 0.173828125,
      "learning_rate": 6.798403038171597e-05,
      "loss": 0.0031,
      "step": 12936
    },
    {
      "epoch": 6.038273045507585,
      "grad_norm": 0.0458984375,
      "learning_rate": 6.797012960317483e-05,
      "loss": 0.0026,
      "step": 12937
    },
    {
      "epoch": 6.038739789964994,
      "grad_norm": 0.050537109375,
      "learning_rate": 6.795622951427236e-05,
      "loss": 0.0004,
      "step": 12938
    },
    {
      "epoch": 6.039206534422403,
      "grad_norm": 0.1611328125,
      "learning_rate": 6.79423301153078e-05,
      "loss": 0.0016,
      "step": 12939
    },
    {
      "epoch": 6.039673278879813,
      "grad_norm": 0.039306640625,
      "learning_rate": 6.792843140658043e-05,
      "loss": 0.0003,
      "step": 12940
    },
    {
      "epoch": 6.040140023337223,
      "grad_norm": 0.01007080078125,
      "learning_rate": 6.791453338838953e-05,
      "loss": 0.0002,
      "step": 12941
    },
    {
      "epoch": 6.040606767794633,
      "grad_norm": 0.061279296875,
      "learning_rate": 6.790063606103431e-05,
      "loss": 0.0004,
      "step": 12942
    },
    {
      "epoch": 6.041073512252042,
      "grad_norm": 0.05810546875,
      "learning_rate": 6.788673942481403e-05,
      "loss": 0.0004,
      "step": 12943
    },
    {
      "epoch": 6.041540256709451,
      "grad_norm": 0.042236328125,
      "learning_rate": 6.787284348002785e-05,
      "loss": 0.0004,
      "step": 12944
    },
    {
      "epoch": 6.042007001166861,
      "grad_norm": 0.134765625,
      "learning_rate": 6.785894822697503e-05,
      "loss": 0.0018,
      "step": 12945
    },
    {
      "epoch": 6.042473745624271,
      "grad_norm": 0.05615234375,
      "learning_rate": 6.784505366595471e-05,
      "loss": 0.0039,
      "step": 12946
    },
    {
      "epoch": 6.04294049008168,
      "grad_norm": 0.0186767578125,
      "learning_rate": 6.783115979726602e-05,
      "loss": 0.0002,
      "step": 12947
    },
    {
      "epoch": 6.04340723453909,
      "grad_norm": 0.27734375,
      "learning_rate": 6.78172666212082e-05,
      "loss": 0.0021,
      "step": 12948
    },
    {
      "epoch": 6.0438739789964995,
      "grad_norm": 0.0361328125,
      "learning_rate": 6.780337413808031e-05,
      "loss": 0.0019,
      "step": 12949
    },
    {
      "epoch": 6.044340723453909,
      "grad_norm": 0.061767578125,
      "learning_rate": 6.77894823481815e-05,
      "loss": 0.0006,
      "step": 12950
    },
    {
      "epoch": 6.044807467911318,
      "grad_norm": 0.0159912109375,
      "learning_rate": 6.777559125181088e-05,
      "loss": 0.0003,
      "step": 12951
    },
    {
      "epoch": 6.045274212368728,
      "grad_norm": 0.0322265625,
      "learning_rate": 6.776170084926754e-05,
      "loss": 0.0003,
      "step": 12952
    },
    {
      "epoch": 6.045740956826138,
      "grad_norm": 0.09521484375,
      "learning_rate": 6.774781114085056e-05,
      "loss": 0.0039,
      "step": 12953
    },
    {
      "epoch": 6.046207701283548,
      "grad_norm": 0.031005859375,
      "learning_rate": 6.773392212685894e-05,
      "loss": 0.0004,
      "step": 12954
    },
    {
      "epoch": 6.0466744457409565,
      "grad_norm": 0.291015625,
      "learning_rate": 6.772003380759183e-05,
      "loss": 0.001,
      "step": 12955
    },
    {
      "epoch": 6.047141190198366,
      "grad_norm": 0.03173828125,
      "learning_rate": 6.770614618334819e-05,
      "loss": 0.0002,
      "step": 12956
    },
    {
      "epoch": 6.047607934655776,
      "grad_norm": 0.032470703125,
      "learning_rate": 6.769225925442704e-05,
      "loss": 0.0004,
      "step": 12957
    },
    {
      "epoch": 6.048074679113186,
      "grad_norm": 0.056396484375,
      "learning_rate": 6.767837302112741e-05,
      "loss": 0.0004,
      "step": 12958
    },
    {
      "epoch": 6.048541423570595,
      "grad_norm": 0.03173828125,
      "learning_rate": 6.766448748374827e-05,
      "loss": 0.0003,
      "step": 12959
    },
    {
      "epoch": 6.049008168028005,
      "grad_norm": 0.0218505859375,
      "learning_rate": 6.765060264258857e-05,
      "loss": 0.0003,
      "step": 12960
    },
    {
      "epoch": 6.049474912485414,
      "grad_norm": 0.05615234375,
      "learning_rate": 6.763671849794731e-05,
      "loss": 0.0004,
      "step": 12961
    },
    {
      "epoch": 6.049941656942824,
      "grad_norm": 0.212890625,
      "learning_rate": 6.76228350501234e-05,
      "loss": 0.0031,
      "step": 12962
    },
    {
      "epoch": 6.050408401400233,
      "grad_norm": 0.047607421875,
      "learning_rate": 6.760895229941579e-05,
      "loss": 0.0003,
      "step": 12963
    },
    {
      "epoch": 6.050875145857643,
      "grad_norm": 0.0263671875,
      "learning_rate": 6.759507024612335e-05,
      "loss": 0.0002,
      "step": 12964
    },
    {
      "epoch": 6.051341890315053,
      "grad_norm": 0.1669921875,
      "learning_rate": 6.758118889054503e-05,
      "loss": 0.0026,
      "step": 12965
    },
    {
      "epoch": 6.0518086347724624,
      "grad_norm": 0.046875,
      "learning_rate": 6.756730823297967e-05,
      "loss": 0.0003,
      "step": 12966
    },
    {
      "epoch": 6.052275379229871,
      "grad_norm": 0.043701171875,
      "learning_rate": 6.755342827372613e-05,
      "loss": 0.0004,
      "step": 12967
    },
    {
      "epoch": 6.052742123687281,
      "grad_norm": 0.020751953125,
      "learning_rate": 6.753954901308329e-05,
      "loss": 0.0003,
      "step": 12968
    },
    {
      "epoch": 6.053208868144691,
      "grad_norm": 0.02001953125,
      "learning_rate": 6.752567045134998e-05,
      "loss": 0.0003,
      "step": 12969
    },
    {
      "epoch": 6.053675612602101,
      "grad_norm": 0.046630859375,
      "learning_rate": 6.751179258882503e-05,
      "loss": 0.0003,
      "step": 12970
    },
    {
      "epoch": 6.05414235705951,
      "grad_norm": 0.0113525390625,
      "learning_rate": 6.74979154258072e-05,
      "loss": 0.0002,
      "step": 12971
    },
    {
      "epoch": 6.0546091015169194,
      "grad_norm": 0.03515625,
      "learning_rate": 6.748403896259534e-05,
      "loss": 0.0004,
      "step": 12972
    },
    {
      "epoch": 6.055075845974329,
      "grad_norm": 0.039306640625,
      "learning_rate": 6.74701631994882e-05,
      "loss": 0.0003,
      "step": 12973
    },
    {
      "epoch": 6.055542590431739,
      "grad_norm": 0.040771484375,
      "learning_rate": 6.74562881367845e-05,
      "loss": 0.0004,
      "step": 12974
    },
    {
      "epoch": 6.056009334889148,
      "grad_norm": 0.0157470703125,
      "learning_rate": 6.744241377478307e-05,
      "loss": 0.0003,
      "step": 12975
    },
    {
      "epoch": 6.056476079346558,
      "grad_norm": 0.024169921875,
      "learning_rate": 6.742854011378256e-05,
      "loss": 0.0002,
      "step": 12976
    },
    {
      "epoch": 6.0569428238039675,
      "grad_norm": 0.224609375,
      "learning_rate": 6.741466715408174e-05,
      "loss": 0.0014,
      "step": 12977
    },
    {
      "epoch": 6.057409568261377,
      "grad_norm": 0.03271484375,
      "learning_rate": 6.740079489597929e-05,
      "loss": 0.0003,
      "step": 12978
    },
    {
      "epoch": 6.057876312718786,
      "grad_norm": 0.0166015625,
      "learning_rate": 6.73869233397739e-05,
      "loss": 0.0002,
      "step": 12979
    },
    {
      "epoch": 6.058343057176196,
      "grad_norm": 0.033203125,
      "learning_rate": 6.737305248576424e-05,
      "loss": 0.0003,
      "step": 12980
    },
    {
      "epoch": 6.058809801633606,
      "grad_norm": 0.04736328125,
      "learning_rate": 6.735918233424891e-05,
      "loss": 0.0002,
      "step": 12981
    },
    {
      "epoch": 6.059276546091015,
      "grad_norm": 0.06201171875,
      "learning_rate": 6.734531288552665e-05,
      "loss": 0.0026,
      "step": 12982
    },
    {
      "epoch": 6.0597432905484245,
      "grad_norm": 0.04931640625,
      "learning_rate": 6.733144413989603e-05,
      "loss": 0.0003,
      "step": 12983
    },
    {
      "epoch": 6.060210035005834,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.731757609765564e-05,
      "loss": 0.0008,
      "step": 12984
    },
    {
      "epoch": 6.060676779463244,
      "grad_norm": 0.060302734375,
      "learning_rate": 6.730370875910412e-05,
      "loss": 0.0004,
      "step": 12985
    },
    {
      "epoch": 6.061143523920653,
      "grad_norm": 0.026123046875,
      "learning_rate": 6.728984212454006e-05,
      "loss": 0.0002,
      "step": 12986
    },
    {
      "epoch": 6.061610268378063,
      "grad_norm": 0.0478515625,
      "learning_rate": 6.727597619426196e-05,
      "loss": 0.0004,
      "step": 12987
    },
    {
      "epoch": 6.062077012835473,
      "grad_norm": 0.0277099609375,
      "learning_rate": 6.72621109685684e-05,
      "loss": 0.0003,
      "step": 12988
    },
    {
      "epoch": 6.062543757292882,
      "grad_norm": 0.0301513671875,
      "learning_rate": 6.724824644775789e-05,
      "loss": 0.0003,
      "step": 12989
    },
    {
      "epoch": 6.063010501750291,
      "grad_norm": 0.041259765625,
      "learning_rate": 6.723438263212898e-05,
      "loss": 0.0004,
      "step": 12990
    },
    {
      "epoch": 6.063477246207701,
      "grad_norm": 0.0167236328125,
      "learning_rate": 6.72205195219802e-05,
      "loss": 0.0003,
      "step": 12991
    },
    {
      "epoch": 6.063943990665111,
      "grad_norm": 0.068359375,
      "learning_rate": 6.720665711760998e-05,
      "loss": 0.0003,
      "step": 12992
    },
    {
      "epoch": 6.064410735122521,
      "grad_norm": 0.05322265625,
      "learning_rate": 6.71927954193168e-05,
      "loss": 0.0028,
      "step": 12993
    },
    {
      "epoch": 6.06487747957993,
      "grad_norm": 0.047119140625,
      "learning_rate": 6.717893442739916e-05,
      "loss": 0.0004,
      "step": 12994
    },
    {
      "epoch": 6.065344224037339,
      "grad_norm": 0.056640625,
      "learning_rate": 6.716507414215548e-05,
      "loss": 0.0023,
      "step": 12995
    },
    {
      "epoch": 6.065810968494749,
      "grad_norm": 0.031494140625,
      "learning_rate": 6.715121456388415e-05,
      "loss": 0.0004,
      "step": 12996
    },
    {
      "epoch": 6.066277712952159,
      "grad_norm": 0.016357421875,
      "learning_rate": 6.713735569288367e-05,
      "loss": 0.0002,
      "step": 12997
    },
    {
      "epoch": 6.066744457409568,
      "grad_norm": 0.0322265625,
      "learning_rate": 6.712349752945236e-05,
      "loss": 0.0004,
      "step": 12998
    },
    {
      "epoch": 6.067211201866978,
      "grad_norm": 0.04443359375,
      "learning_rate": 6.71096400738886e-05,
      "loss": 0.0019,
      "step": 12999
    },
    {
      "epoch": 6.0676779463243875,
      "grad_norm": 0.036865234375,
      "learning_rate": 6.70957833264908e-05,
      "loss": 0.0003,
      "step": 13000
    },
    {
      "epoch": 6.068144690781797,
      "grad_norm": 0.037109375,
      "learning_rate": 6.708192728755732e-05,
      "loss": 0.0002,
      "step": 13001
    },
    {
      "epoch": 6.068611435239206,
      "grad_norm": 0.02880859375,
      "learning_rate": 6.706807195738644e-05,
      "loss": 0.0026,
      "step": 13002
    },
    {
      "epoch": 6.069078179696616,
      "grad_norm": 0.0169677734375,
      "learning_rate": 6.70542173362765e-05,
      "loss": 0.0002,
      "step": 13003
    },
    {
      "epoch": 6.069544924154026,
      "grad_norm": 0.040283203125,
      "learning_rate": 6.704036342452582e-05,
      "loss": 0.0004,
      "step": 13004
    },
    {
      "epoch": 6.070011668611436,
      "grad_norm": 0.234375,
      "learning_rate": 6.702651022243271e-05,
      "loss": 0.0007,
      "step": 13005
    },
    {
      "epoch": 6.0704784130688445,
      "grad_norm": 0.1201171875,
      "learning_rate": 6.701265773029539e-05,
      "loss": 0.0007,
      "step": 13006
    },
    {
      "epoch": 6.070945157526254,
      "grad_norm": 0.09228515625,
      "learning_rate": 6.699880594841216e-05,
      "loss": 0.0006,
      "step": 13007
    },
    {
      "epoch": 6.071411901983664,
      "grad_norm": 0.24609375,
      "learning_rate": 6.698495487708126e-05,
      "loss": 0.0009,
      "step": 13008
    },
    {
      "epoch": 6.071878646441074,
      "grad_norm": 0.046142578125,
      "learning_rate": 6.697110451660089e-05,
      "loss": 0.0004,
      "step": 13009
    },
    {
      "epoch": 6.072345390898483,
      "grad_norm": 0.02587890625,
      "learning_rate": 6.695725486726931e-05,
      "loss": 0.0003,
      "step": 13010
    },
    {
      "epoch": 6.072812135355893,
      "grad_norm": 0.0322265625,
      "learning_rate": 6.69434059293847e-05,
      "loss": 0.0004,
      "step": 13011
    },
    {
      "epoch": 6.073278879813302,
      "grad_norm": 0.038818359375,
      "learning_rate": 6.692955770324522e-05,
      "loss": 0.0003,
      "step": 13012
    },
    {
      "epoch": 6.073745624270712,
      "grad_norm": 0.1884765625,
      "learning_rate": 6.691571018914904e-05,
      "loss": 0.0008,
      "step": 13013
    },
    {
      "epoch": 6.074212368728121,
      "grad_norm": 0.1474609375,
      "learning_rate": 6.690186338739436e-05,
      "loss": 0.0008,
      "step": 13014
    },
    {
      "epoch": 6.074679113185531,
      "grad_norm": 0.07470703125,
      "learning_rate": 6.688801729827928e-05,
      "loss": 0.0004,
      "step": 13015
    },
    {
      "epoch": 6.075145857642941,
      "grad_norm": 0.054443359375,
      "learning_rate": 6.68741719221019e-05,
      "loss": 0.0018,
      "step": 13016
    },
    {
      "epoch": 6.0756126021003505,
      "grad_norm": 0.059326171875,
      "learning_rate": 6.686032725916038e-05,
      "loss": 0.0003,
      "step": 13017
    },
    {
      "epoch": 6.076079346557759,
      "grad_norm": 0.032470703125,
      "learning_rate": 6.684648330975278e-05,
      "loss": 0.0003,
      "step": 13018
    },
    {
      "epoch": 6.076546091015169,
      "grad_norm": 0.10693359375,
      "learning_rate": 6.683264007417717e-05,
      "loss": 0.0005,
      "step": 13019
    },
    {
      "epoch": 6.077012835472579,
      "grad_norm": 0.0130615234375,
      "learning_rate": 6.681879755273162e-05,
      "loss": 0.0002,
      "step": 13020
    },
    {
      "epoch": 6.077479579929989,
      "grad_norm": 0.296875,
      "learning_rate": 6.680495574571417e-05,
      "loss": 0.0023,
      "step": 13021
    },
    {
      "epoch": 6.077946324387398,
      "grad_norm": 0.0157470703125,
      "learning_rate": 6.679111465342287e-05,
      "loss": 0.0002,
      "step": 13022
    },
    {
      "epoch": 6.0784130688448075,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.677727427615569e-05,
      "loss": 0.0004,
      "step": 13023
    },
    {
      "epoch": 6.078879813302217,
      "grad_norm": 0.0419921875,
      "learning_rate": 6.676343461421068e-05,
      "loss": 0.0003,
      "step": 13024
    },
    {
      "epoch": 6.079346557759626,
      "grad_norm": 0.044189453125,
      "learning_rate": 6.674959566788579e-05,
      "loss": 0.0003,
      "step": 13025
    },
    {
      "epoch": 6.079813302217036,
      "grad_norm": 0.0120849609375,
      "learning_rate": 6.673575743747898e-05,
      "loss": 0.0002,
      "step": 13026
    },
    {
      "epoch": 6.080280046674446,
      "grad_norm": 0.0211181640625,
      "learning_rate": 6.672191992328825e-05,
      "loss": 0.0002,
      "step": 13027
    },
    {
      "epoch": 6.080746791131856,
      "grad_norm": 0.1328125,
      "learning_rate": 6.67080831256115e-05,
      "loss": 0.0008,
      "step": 13028
    },
    {
      "epoch": 6.0812135355892645,
      "grad_norm": 0.0247802734375,
      "learning_rate": 6.669424704474665e-05,
      "loss": 0.0003,
      "step": 13029
    },
    {
      "epoch": 6.081680280046674,
      "grad_norm": 0.05322265625,
      "learning_rate": 6.66804116809916e-05,
      "loss": 0.0003,
      "step": 13030
    },
    {
      "epoch": 6.082147024504084,
      "grad_norm": 0.047119140625,
      "learning_rate": 6.666657703464428e-05,
      "loss": 0.0028,
      "step": 13031
    },
    {
      "epoch": 6.082613768961494,
      "grad_norm": 0.0400390625,
      "learning_rate": 6.665274310600254e-05,
      "loss": 0.0002,
      "step": 13032
    },
    {
      "epoch": 6.083080513418903,
      "grad_norm": 0.0247802734375,
      "learning_rate": 6.663890989536426e-05,
      "loss": 0.0003,
      "step": 13033
    },
    {
      "epoch": 6.083547257876313,
      "grad_norm": 0.197265625,
      "learning_rate": 6.662507740302725e-05,
      "loss": 0.0009,
      "step": 13034
    },
    {
      "epoch": 6.084014002333722,
      "grad_norm": 0.01953125,
      "learning_rate": 6.661124562928933e-05,
      "loss": 0.0002,
      "step": 13035
    },
    {
      "epoch": 6.084480746791132,
      "grad_norm": 0.037109375,
      "learning_rate": 6.659741457444835e-05,
      "loss": 0.0002,
      "step": 13036
    },
    {
      "epoch": 6.084947491248541,
      "grad_norm": 0.032470703125,
      "learning_rate": 6.65835842388021e-05,
      "loss": 0.0026,
      "step": 13037
    },
    {
      "epoch": 6.085414235705951,
      "grad_norm": 0.10888671875,
      "learning_rate": 6.656975462264834e-05,
      "loss": 0.0039,
      "step": 13038
    },
    {
      "epoch": 6.085880980163361,
      "grad_norm": 0.0247802734375,
      "learning_rate": 6.655592572628486e-05,
      "loss": 0.0002,
      "step": 13039
    },
    {
      "epoch": 6.0863477246207704,
      "grad_norm": 0.059326171875,
      "learning_rate": 6.654209755000941e-05,
      "loss": 0.0024,
      "step": 13040
    },
    {
      "epoch": 6.086814469078179,
      "grad_norm": 0.02685546875,
      "learning_rate": 6.652827009411974e-05,
      "loss": 0.0003,
      "step": 13041
    },
    {
      "epoch": 6.087281213535589,
      "grad_norm": 0.08935546875,
      "learning_rate": 6.651444335891353e-05,
      "loss": 0.0022,
      "step": 13042
    },
    {
      "epoch": 6.087747957992999,
      "grad_norm": 0.030029296875,
      "learning_rate": 6.650061734468852e-05,
      "loss": 0.0019,
      "step": 13043
    },
    {
      "epoch": 6.088214702450409,
      "grad_norm": 0.60546875,
      "learning_rate": 6.648679205174238e-05,
      "loss": 0.0016,
      "step": 13044
    },
    {
      "epoch": 6.088681446907818,
      "grad_norm": 0.0203857421875,
      "learning_rate": 6.647296748037278e-05,
      "loss": 0.0002,
      "step": 13045
    },
    {
      "epoch": 6.0891481913652274,
      "grad_norm": 0.01080322265625,
      "learning_rate": 6.64591436308774e-05,
      "loss": 0.0002,
      "step": 13046
    },
    {
      "epoch": 6.089614935822637,
      "grad_norm": 0.048583984375,
      "learning_rate": 6.644532050355388e-05,
      "loss": 0.0004,
      "step": 13047
    },
    {
      "epoch": 6.090081680280047,
      "grad_norm": 0.046142578125,
      "learning_rate": 6.643149809869982e-05,
      "loss": 0.0004,
      "step": 13048
    },
    {
      "epoch": 6.090548424737456,
      "grad_norm": 0.0284423828125,
      "learning_rate": 6.641767641661286e-05,
      "loss": 0.0003,
      "step": 13049
    },
    {
      "epoch": 6.091015169194866,
      "grad_norm": 0.037841796875,
      "learning_rate": 6.64038554575906e-05,
      "loss": 0.0003,
      "step": 13050
    },
    {
      "epoch": 6.0914819136522755,
      "grad_norm": 0.1025390625,
      "learning_rate": 6.639003522193059e-05,
      "loss": 0.0004,
      "step": 13051
    },
    {
      "epoch": 6.091948658109685,
      "grad_norm": 0.134765625,
      "learning_rate": 6.637621570993042e-05,
      "loss": 0.0008,
      "step": 13052
    },
    {
      "epoch": 6.092415402567094,
      "grad_norm": 0.051025390625,
      "learning_rate": 6.636239692188762e-05,
      "loss": 0.0006,
      "step": 13053
    },
    {
      "epoch": 6.092882147024504,
      "grad_norm": 0.044677734375,
      "learning_rate": 6.634857885809975e-05,
      "loss": 0.0004,
      "step": 13054
    },
    {
      "epoch": 6.093348891481914,
      "grad_norm": 0.0087890625,
      "learning_rate": 6.63347615188643e-05,
      "loss": 0.0002,
      "step": 13055
    },
    {
      "epoch": 6.093815635939324,
      "grad_norm": 0.05029296875,
      "learning_rate": 6.632094490447881e-05,
      "loss": 0.0037,
      "step": 13056
    },
    {
      "epoch": 6.0942823803967325,
      "grad_norm": 0.0947265625,
      "learning_rate": 6.630712901524074e-05,
      "loss": 0.0004,
      "step": 13057
    },
    {
      "epoch": 6.094749124854142,
      "grad_norm": 0.018310546875,
      "learning_rate": 6.629331385144754e-05,
      "loss": 0.0002,
      "step": 13058
    },
    {
      "epoch": 6.095215869311552,
      "grad_norm": 0.032470703125,
      "learning_rate": 6.627949941339671e-05,
      "loss": 0.0003,
      "step": 13059
    },
    {
      "epoch": 6.095682613768962,
      "grad_norm": 0.0185546875,
      "learning_rate": 6.626568570138568e-05,
      "loss": 0.0002,
      "step": 13060
    },
    {
      "epoch": 6.096149358226371,
      "grad_norm": 0.0205078125,
      "learning_rate": 6.625187271571184e-05,
      "loss": 0.0002,
      "step": 13061
    },
    {
      "epoch": 6.096616102683781,
      "grad_norm": 0.064453125,
      "learning_rate": 6.623806045667265e-05,
      "loss": 0.002,
      "step": 13062
    },
    {
      "epoch": 6.09708284714119,
      "grad_norm": 0.1904296875,
      "learning_rate": 6.622424892456547e-05,
      "loss": 0.0011,
      "step": 13063
    },
    {
      "epoch": 6.0975495915986,
      "grad_norm": 0.0311279296875,
      "learning_rate": 6.621043811968768e-05,
      "loss": 0.0003,
      "step": 13064
    },
    {
      "epoch": 6.098016336056009,
      "grad_norm": 0.026123046875,
      "learning_rate": 6.619662804233664e-05,
      "loss": 0.0003,
      "step": 13065
    },
    {
      "epoch": 6.098483080513419,
      "grad_norm": 0.0281982421875,
      "learning_rate": 6.618281869280971e-05,
      "loss": 0.0003,
      "step": 13066
    },
    {
      "epoch": 6.098949824970829,
      "grad_norm": 0.057861328125,
      "learning_rate": 6.616901007140422e-05,
      "loss": 0.0009,
      "step": 13067
    },
    {
      "epoch": 6.099416569428238,
      "grad_norm": 0.0869140625,
      "learning_rate": 6.615520217841746e-05,
      "loss": 0.0003,
      "step": 13068
    },
    {
      "epoch": 6.099883313885647,
      "grad_norm": 0.030029296875,
      "learning_rate": 6.614139501414677e-05,
      "loss": 0.0003,
      "step": 13069
    },
    {
      "epoch": 6.100350058343057,
      "grad_norm": 0.03955078125,
      "learning_rate": 6.612758857888942e-05,
      "loss": 0.0003,
      "step": 13070
    },
    {
      "epoch": 6.100816802800467,
      "grad_norm": 0.41796875,
      "learning_rate": 6.611378287294267e-05,
      "loss": 0.0013,
      "step": 13071
    },
    {
      "epoch": 6.101283547257876,
      "grad_norm": 0.119140625,
      "learning_rate": 6.609997789660373e-05,
      "loss": 0.0026,
      "step": 13072
    },
    {
      "epoch": 6.101750291715286,
      "grad_norm": 0.119140625,
      "learning_rate": 6.608617365016993e-05,
      "loss": 0.0006,
      "step": 13073
    },
    {
      "epoch": 6.1022170361726955,
      "grad_norm": 0.04833984375,
      "learning_rate": 6.607237013393842e-05,
      "loss": 0.0004,
      "step": 13074
    },
    {
      "epoch": 6.102683780630105,
      "grad_norm": 0.0174560546875,
      "learning_rate": 6.605856734820642e-05,
      "loss": 0.0003,
      "step": 13075
    },
    {
      "epoch": 6.103150525087514,
      "grad_norm": 0.05859375,
      "learning_rate": 6.604476529327112e-05,
      "loss": 0.0031,
      "step": 13076
    },
    {
      "epoch": 6.103617269544924,
      "grad_norm": 0.0400390625,
      "learning_rate": 6.603096396942971e-05,
      "loss": 0.0004,
      "step": 13077
    },
    {
      "epoch": 6.104084014002334,
      "grad_norm": 0.032470703125,
      "learning_rate": 6.601716337697933e-05,
      "loss": 0.0003,
      "step": 13078
    },
    {
      "epoch": 6.104550758459744,
      "grad_norm": 0.01495361328125,
      "learning_rate": 6.600336351621711e-05,
      "loss": 0.0002,
      "step": 13079
    },
    {
      "epoch": 6.1050175029171525,
      "grad_norm": 0.07080078125,
      "learning_rate": 6.598956438744022e-05,
      "loss": 0.0005,
      "step": 13080
    },
    {
      "epoch": 6.105484247374562,
      "grad_norm": 0.048095703125,
      "learning_rate": 6.597576599094575e-05,
      "loss": 0.0003,
      "step": 13081
    },
    {
      "epoch": 6.105950991831972,
      "grad_norm": 0.043701171875,
      "learning_rate": 6.596196832703078e-05,
      "loss": 0.0005,
      "step": 13082
    },
    {
      "epoch": 6.106417736289382,
      "grad_norm": 0.1484375,
      "learning_rate": 6.59481713959924e-05,
      "loss": 0.0024,
      "step": 13083
    },
    {
      "epoch": 6.106884480746791,
      "grad_norm": 0.021484375,
      "learning_rate": 6.593437519812764e-05,
      "loss": 0.0003,
      "step": 13084
    },
    {
      "epoch": 6.107351225204201,
      "grad_norm": 0.126953125,
      "learning_rate": 6.59205797337336e-05,
      "loss": 0.0005,
      "step": 13085
    },
    {
      "epoch": 6.10781796966161,
      "grad_norm": 0.0311279296875,
      "learning_rate": 6.590678500310729e-05,
      "loss": 0.0003,
      "step": 13086
    },
    {
      "epoch": 6.10828471411902,
      "grad_norm": 0.314453125,
      "learning_rate": 6.589299100654572e-05,
      "loss": 0.0021,
      "step": 13087
    },
    {
      "epoch": 6.108751458576429,
      "grad_norm": 0.019775390625,
      "learning_rate": 6.58791977443459e-05,
      "loss": 0.0003,
      "step": 13088
    },
    {
      "epoch": 6.109218203033839,
      "grad_norm": 0.07763671875,
      "learning_rate": 6.586540521680482e-05,
      "loss": 0.0031,
      "step": 13089
    },
    {
      "epoch": 6.109684947491249,
      "grad_norm": 0.6875,
      "learning_rate": 6.585161342421944e-05,
      "loss": 0.0015,
      "step": 13090
    },
    {
      "epoch": 6.1101516919486585,
      "grad_norm": 0.0615234375,
      "learning_rate": 6.583782236688669e-05,
      "loss": 0.0008,
      "step": 13091
    },
    {
      "epoch": 6.110618436406067,
      "grad_norm": 0.01153564453125,
      "learning_rate": 6.582403204510355e-05,
      "loss": 0.0002,
      "step": 13092
    },
    {
      "epoch": 6.111085180863477,
      "grad_norm": 0.034423828125,
      "learning_rate": 6.58102424591669e-05,
      "loss": 0.0003,
      "step": 13093
    },
    {
      "epoch": 6.111551925320887,
      "grad_norm": 0.0830078125,
      "learning_rate": 6.579645360937366e-05,
      "loss": 0.0006,
      "step": 13094
    },
    {
      "epoch": 6.112018669778297,
      "grad_norm": 0.80078125,
      "learning_rate": 6.578266549602075e-05,
      "loss": 0.0114,
      "step": 13095
    },
    {
      "epoch": 6.112485414235706,
      "grad_norm": 0.041015625,
      "learning_rate": 6.5768878119405e-05,
      "loss": 0.0003,
      "step": 13096
    },
    {
      "epoch": 6.1129521586931155,
      "grad_norm": 0.068359375,
      "learning_rate": 6.575509147982327e-05,
      "loss": 0.0035,
      "step": 13097
    },
    {
      "epoch": 6.113418903150525,
      "grad_norm": 0.78125,
      "learning_rate": 6.574130557757243e-05,
      "loss": 0.0033,
      "step": 13098
    },
    {
      "epoch": 6.113885647607935,
      "grad_norm": 0.035888671875,
      "learning_rate": 6.572752041294931e-05,
      "loss": 0.0004,
      "step": 13099
    },
    {
      "epoch": 6.114352392065344,
      "grad_norm": 0.26953125,
      "learning_rate": 6.571373598625068e-05,
      "loss": 0.0021,
      "step": 13100
    },
    {
      "epoch": 6.114819136522754,
      "grad_norm": 0.091796875,
      "learning_rate": 6.569995229777335e-05,
      "loss": 0.0007,
      "step": 13101
    },
    {
      "epoch": 6.115285880980164,
      "grad_norm": 0.125,
      "learning_rate": 6.568616934781413e-05,
      "loss": 0.0031,
      "step": 13102
    },
    {
      "epoch": 6.115752625437573,
      "grad_norm": 0.1318359375,
      "learning_rate": 6.567238713666974e-05,
      "loss": 0.0009,
      "step": 13103
    },
    {
      "epoch": 6.116219369894982,
      "grad_norm": 0.033203125,
      "learning_rate": 6.565860566463693e-05,
      "loss": 0.0018,
      "step": 13104
    },
    {
      "epoch": 6.116686114352392,
      "grad_norm": 0.2109375,
      "learning_rate": 6.564482493201245e-05,
      "loss": 0.0011,
      "step": 13105
    },
    {
      "epoch": 6.117152858809802,
      "grad_norm": 0.2373046875,
      "learning_rate": 6.563104493909302e-05,
      "loss": 0.0018,
      "step": 13106
    },
    {
      "epoch": 6.117619603267211,
      "grad_norm": 0.154296875,
      "learning_rate": 6.56172656861753e-05,
      "loss": 0.001,
      "step": 13107
    },
    {
      "epoch": 6.118086347724621,
      "grad_norm": 0.126953125,
      "learning_rate": 6.560348717355603e-05,
      "loss": 0.0008,
      "step": 13108
    },
    {
      "epoch": 6.11855309218203,
      "grad_norm": 0.0267333984375,
      "learning_rate": 6.558970940153183e-05,
      "loss": 0.0003,
      "step": 13109
    },
    {
      "epoch": 6.11901983663944,
      "grad_norm": 0.03515625,
      "learning_rate": 6.557593237039936e-05,
      "loss": 0.0004,
      "step": 13110
    },
    {
      "epoch": 6.119486581096849,
      "grad_norm": 0.25390625,
      "learning_rate": 6.556215608045526e-05,
      "loss": 0.0033,
      "step": 13111
    },
    {
      "epoch": 6.119953325554259,
      "grad_norm": 0.291015625,
      "learning_rate": 6.554838053199616e-05,
      "loss": 0.0015,
      "step": 13112
    },
    {
      "epoch": 6.120420070011669,
      "grad_norm": 0.16015625,
      "learning_rate": 6.553460572531866e-05,
      "loss": 0.0043,
      "step": 13113
    },
    {
      "epoch": 6.1208868144690785,
      "grad_norm": 0.0284423828125,
      "learning_rate": 6.552083166071929e-05,
      "loss": 0.0003,
      "step": 13114
    },
    {
      "epoch": 6.121353558926487,
      "grad_norm": 0.051513671875,
      "learning_rate": 6.550705833849471e-05,
      "loss": 0.002,
      "step": 13115
    },
    {
      "epoch": 6.121820303383897,
      "grad_norm": 0.022705078125,
      "learning_rate": 6.549328575894144e-05,
      "loss": 0.0003,
      "step": 13116
    },
    {
      "epoch": 6.122287047841307,
      "grad_norm": 0.24609375,
      "learning_rate": 6.5479513922356e-05,
      "loss": 0.0029,
      "step": 13117
    },
    {
      "epoch": 6.122753792298717,
      "grad_norm": 0.31640625,
      "learning_rate": 6.546574282903492e-05,
      "loss": 0.0012,
      "step": 13118
    },
    {
      "epoch": 6.123220536756126,
      "grad_norm": 0.03125,
      "learning_rate": 6.545197247927474e-05,
      "loss": 0.0004,
      "step": 13119
    },
    {
      "epoch": 6.1236872812135354,
      "grad_norm": 0.0498046875,
      "learning_rate": 6.543820287337192e-05,
      "loss": 0.0004,
      "step": 13120
    },
    {
      "epoch": 6.124154025670945,
      "grad_norm": 0.1123046875,
      "learning_rate": 6.542443401162291e-05,
      "loss": 0.0007,
      "step": 13121
    },
    {
      "epoch": 6.124620770128355,
      "grad_norm": 0.0205078125,
      "learning_rate": 6.54106658943242e-05,
      "loss": 0.0003,
      "step": 13122
    },
    {
      "epoch": 6.125087514585764,
      "grad_norm": 0.0625,
      "learning_rate": 6.539689852177227e-05,
      "loss": 0.0004,
      "step": 13123
    },
    {
      "epoch": 6.125554259043174,
      "grad_norm": 0.02734375,
      "learning_rate": 6.538313189426346e-05,
      "loss": 0.0003,
      "step": 13124
    },
    {
      "epoch": 6.1260210035005835,
      "grad_norm": 0.0238037109375,
      "learning_rate": 6.536936601209424e-05,
      "loss": 0.0003,
      "step": 13125
    },
    {
      "epoch": 6.126487747957993,
      "grad_norm": 0.08154296875,
      "learning_rate": 6.535560087556103e-05,
      "loss": 0.0039,
      "step": 13126
    },
    {
      "epoch": 6.126954492415402,
      "grad_norm": 0.123046875,
      "learning_rate": 6.534183648496013e-05,
      "loss": 0.0006,
      "step": 13127
    },
    {
      "epoch": 6.127421236872812,
      "grad_norm": 0.02001953125,
      "learning_rate": 6.532807284058797e-05,
      "loss": 0.0003,
      "step": 13128
    },
    {
      "epoch": 6.127887981330222,
      "grad_norm": 0.123046875,
      "learning_rate": 6.531430994274088e-05,
      "loss": 0.0007,
      "step": 13129
    },
    {
      "epoch": 6.128354725787632,
      "grad_norm": 0.07666015625,
      "learning_rate": 6.530054779171517e-05,
      "loss": 0.0017,
      "step": 13130
    },
    {
      "epoch": 6.1288214702450405,
      "grad_norm": 0.0791015625,
      "learning_rate": 6.528678638780717e-05,
      "loss": 0.0047,
      "step": 13131
    },
    {
      "epoch": 6.12928821470245,
      "grad_norm": 0.1748046875,
      "learning_rate": 6.527302573131315e-05,
      "loss": 0.0036,
      "step": 13132
    },
    {
      "epoch": 6.12975495915986,
      "grad_norm": 0.05224609375,
      "learning_rate": 6.525926582252941e-05,
      "loss": 0.0004,
      "step": 13133
    },
    {
      "epoch": 6.13022170361727,
      "grad_norm": 0.11328125,
      "learning_rate": 6.524550666175226e-05,
      "loss": 0.0006,
      "step": 13134
    },
    {
      "epoch": 6.130688448074679,
      "grad_norm": 0.228515625,
      "learning_rate": 6.523174824927789e-05,
      "loss": 0.001,
      "step": 13135
    },
    {
      "epoch": 6.131155192532089,
      "grad_norm": 0.1591796875,
      "learning_rate": 6.521799058540255e-05,
      "loss": 0.0033,
      "step": 13136
    },
    {
      "epoch": 6.131621936989498,
      "grad_norm": 0.443359375,
      "learning_rate": 6.520423367042247e-05,
      "loss": 0.0189,
      "step": 13137
    },
    {
      "epoch": 6.132088681446908,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.519047750463386e-05,
      "loss": 0.0006,
      "step": 13138
    },
    {
      "epoch": 6.132555425904317,
      "grad_norm": 0.146484375,
      "learning_rate": 6.517672208833287e-05,
      "loss": 0.0017,
      "step": 13139
    },
    {
      "epoch": 6.133022170361727,
      "grad_norm": 0.482421875,
      "learning_rate": 6.51629674218157e-05,
      "loss": 0.0043,
      "step": 13140
    },
    {
      "epoch": 6.133488914819137,
      "grad_norm": 0.0830078125,
      "learning_rate": 6.51492135053785e-05,
      "loss": 0.002,
      "step": 13141
    },
    {
      "epoch": 6.1339556592765465,
      "grad_norm": 0.045166015625,
      "learning_rate": 6.51354603393174e-05,
      "loss": 0.0004,
      "step": 13142
    },
    {
      "epoch": 6.134422403733955,
      "grad_norm": 0.05908203125,
      "learning_rate": 6.51217079239285e-05,
      "loss": 0.0003,
      "step": 13143
    },
    {
      "epoch": 6.134889148191365,
      "grad_norm": 0.0181884765625,
      "learning_rate": 6.510795625950795e-05,
      "loss": 0.0003,
      "step": 13144
    },
    {
      "epoch": 6.135355892648775,
      "grad_norm": 0.169921875,
      "learning_rate": 6.509420534635182e-05,
      "loss": 0.0025,
      "step": 13145
    },
    {
      "epoch": 6.135822637106184,
      "grad_norm": 0.08447265625,
      "learning_rate": 6.508045518475615e-05,
      "loss": 0.0006,
      "step": 13146
    },
    {
      "epoch": 6.136289381563594,
      "grad_norm": 0.283203125,
      "learning_rate": 6.506670577501705e-05,
      "loss": 0.0024,
      "step": 13147
    },
    {
      "epoch": 6.1367561260210035,
      "grad_norm": 0.1455078125,
      "learning_rate": 6.505295711743052e-05,
      "loss": 0.0006,
      "step": 13148
    },
    {
      "epoch": 6.137222870478413,
      "grad_norm": 0.2158203125,
      "learning_rate": 6.50392092122926e-05,
      "loss": 0.001,
      "step": 13149
    },
    {
      "epoch": 6.137689614935822,
      "grad_norm": 0.0751953125,
      "learning_rate": 6.502546205989929e-05,
      "loss": 0.0006,
      "step": 13150
    },
    {
      "epoch": 6.138156359393232,
      "grad_norm": 0.03564453125,
      "learning_rate": 6.50117156605466e-05,
      "loss": 0.0004,
      "step": 13151
    },
    {
      "epoch": 6.138623103850642,
      "grad_norm": 0.04736328125,
      "learning_rate": 6.499797001453049e-05,
      "loss": 0.0005,
      "step": 13152
    },
    {
      "epoch": 6.139089848308052,
      "grad_norm": 0.09326171875,
      "learning_rate": 6.49842251221469e-05,
      "loss": 0.0007,
      "step": 13153
    },
    {
      "epoch": 6.1395565927654605,
      "grad_norm": 0.103515625,
      "learning_rate": 6.497048098369182e-05,
      "loss": 0.0008,
      "step": 13154
    },
    {
      "epoch": 6.14002333722287,
      "grad_norm": 0.07763671875,
      "learning_rate": 6.495673759946114e-05,
      "loss": 0.0044,
      "step": 13155
    },
    {
      "epoch": 6.14049008168028,
      "grad_norm": 0.041015625,
      "learning_rate": 6.494299496975077e-05,
      "loss": 0.0005,
      "step": 13156
    },
    {
      "epoch": 6.14095682613769,
      "grad_norm": 0.036376953125,
      "learning_rate": 6.492925309485662e-05,
      "loss": 0.0003,
      "step": 13157
    },
    {
      "epoch": 6.141423570595099,
      "grad_norm": 0.0634765625,
      "learning_rate": 6.491551197507458e-05,
      "loss": 0.0007,
      "step": 13158
    },
    {
      "epoch": 6.141890315052509,
      "grad_norm": 0.0250244140625,
      "learning_rate": 6.49017716107005e-05,
      "loss": 0.0004,
      "step": 13159
    },
    {
      "epoch": 6.142357059509918,
      "grad_norm": 0.0216064453125,
      "learning_rate": 6.488803200203017e-05,
      "loss": 0.0004,
      "step": 13160
    },
    {
      "epoch": 6.142823803967328,
      "grad_norm": 0.06298828125,
      "learning_rate": 6.487429314935951e-05,
      "loss": 0.0023,
      "step": 13161
    },
    {
      "epoch": 6.143290548424737,
      "grad_norm": 0.060302734375,
      "learning_rate": 6.48605550529843e-05,
      "loss": 0.0004,
      "step": 13162
    },
    {
      "epoch": 6.143757292882147,
      "grad_norm": 0.138671875,
      "learning_rate": 6.484681771320029e-05,
      "loss": 0.0007,
      "step": 13163
    },
    {
      "epoch": 6.144224037339557,
      "grad_norm": 0.05078125,
      "learning_rate": 6.483308113030332e-05,
      "loss": 0.003,
      "step": 13164
    },
    {
      "epoch": 6.1446907817969665,
      "grad_norm": 0.30078125,
      "learning_rate": 6.481934530458913e-05,
      "loss": 0.0015,
      "step": 13165
    },
    {
      "epoch": 6.145157526254375,
      "grad_norm": 0.11767578125,
      "learning_rate": 6.480561023635345e-05,
      "loss": 0.0005,
      "step": 13166
    },
    {
      "epoch": 6.145624270711785,
      "grad_norm": 0.16796875,
      "learning_rate": 6.479187592589206e-05,
      "loss": 0.0008,
      "step": 13167
    },
    {
      "epoch": 6.146091015169195,
      "grad_norm": 0.07568359375,
      "learning_rate": 6.477814237350064e-05,
      "loss": 0.0004,
      "step": 13168
    },
    {
      "epoch": 6.146557759626605,
      "grad_norm": 0.11181640625,
      "learning_rate": 6.476440957947489e-05,
      "loss": 0.0025,
      "step": 13169
    },
    {
      "epoch": 6.147024504084014,
      "grad_norm": 0.0849609375,
      "learning_rate": 6.475067754411049e-05,
      "loss": 0.0006,
      "step": 13170
    },
    {
      "epoch": 6.1474912485414235,
      "grad_norm": 0.0260009765625,
      "learning_rate": 6.473694626770311e-05,
      "loss": 0.0003,
      "step": 13171
    },
    {
      "epoch": 6.147957992998833,
      "grad_norm": 0.1201171875,
      "learning_rate": 6.472321575054842e-05,
      "loss": 0.0005,
      "step": 13172
    },
    {
      "epoch": 6.148424737456243,
      "grad_norm": 0.04541015625,
      "learning_rate": 6.4709485992942e-05,
      "loss": 0.0004,
      "step": 13173
    },
    {
      "epoch": 6.148891481913652,
      "grad_norm": 0.0164794921875,
      "learning_rate": 6.469575699517953e-05,
      "loss": 0.0003,
      "step": 13174
    },
    {
      "epoch": 6.149358226371062,
      "grad_norm": 0.10107421875,
      "learning_rate": 6.468202875755659e-05,
      "loss": 0.0005,
      "step": 13175
    },
    {
      "epoch": 6.149824970828472,
      "grad_norm": 0.130859375,
      "learning_rate": 6.466830128036872e-05,
      "loss": 0.0007,
      "step": 13176
    },
    {
      "epoch": 6.150291715285881,
      "grad_norm": 0.033935546875,
      "learning_rate": 6.46545745639116e-05,
      "loss": 0.0003,
      "step": 13177
    },
    {
      "epoch": 6.15075845974329,
      "grad_norm": 0.043701171875,
      "learning_rate": 6.464084860848065e-05,
      "loss": 0.0005,
      "step": 13178
    },
    {
      "epoch": 6.1512252042007,
      "grad_norm": 0.039794921875,
      "learning_rate": 6.462712341437147e-05,
      "loss": 0.0004,
      "step": 13179
    },
    {
      "epoch": 6.15169194865811,
      "grad_norm": 0.03173828125,
      "learning_rate": 6.461339898187957e-05,
      "loss": 0.0003,
      "step": 13180
    },
    {
      "epoch": 6.15215869311552,
      "grad_norm": 0.08154296875,
      "learning_rate": 6.459967531130045e-05,
      "loss": 0.0006,
      "step": 13181
    },
    {
      "epoch": 6.152625437572929,
      "grad_norm": 0.240234375,
      "learning_rate": 6.458595240292959e-05,
      "loss": 0.001,
      "step": 13182
    },
    {
      "epoch": 6.153092182030338,
      "grad_norm": 0.0185546875,
      "learning_rate": 6.457223025706247e-05,
      "loss": 0.0003,
      "step": 13183
    },
    {
      "epoch": 6.153558926487748,
      "grad_norm": 0.05908203125,
      "learning_rate": 6.455850887399455e-05,
      "loss": 0.003,
      "step": 13184
    },
    {
      "epoch": 6.154025670945158,
      "grad_norm": 0.0301513671875,
      "learning_rate": 6.454478825402124e-05,
      "loss": 0.0003,
      "step": 13185
    },
    {
      "epoch": 6.154492415402567,
      "grad_norm": 0.09423828125,
      "learning_rate": 6.453106839743799e-05,
      "loss": 0.0004,
      "step": 13186
    },
    {
      "epoch": 6.154959159859977,
      "grad_norm": 0.062255859375,
      "learning_rate": 6.45173493045402e-05,
      "loss": 0.0003,
      "step": 13187
    },
    {
      "epoch": 6.1554259043173865,
      "grad_norm": 0.0517578125,
      "learning_rate": 6.450363097562322e-05,
      "loss": 0.0004,
      "step": 13188
    },
    {
      "epoch": 6.155892648774795,
      "grad_norm": 0.0234375,
      "learning_rate": 6.448991341098247e-05,
      "loss": 0.0003,
      "step": 13189
    },
    {
      "epoch": 6.156359393232205,
      "grad_norm": 0.09423828125,
      "learning_rate": 6.447619661091328e-05,
      "loss": 0.0006,
      "step": 13190
    },
    {
      "epoch": 6.156826137689615,
      "grad_norm": 0.061767578125,
      "learning_rate": 6.4462480575711e-05,
      "loss": 0.0004,
      "step": 13191
    },
    {
      "epoch": 6.157292882147025,
      "grad_norm": 0.166015625,
      "learning_rate": 6.444876530567092e-05,
      "loss": 0.0007,
      "step": 13192
    },
    {
      "epoch": 6.157759626604434,
      "grad_norm": 0.014892578125,
      "learning_rate": 6.443505080108838e-05,
      "loss": 0.0002,
      "step": 13193
    },
    {
      "epoch": 6.1582263710618435,
      "grad_norm": 0.158203125,
      "learning_rate": 6.442133706225866e-05,
      "loss": 0.0008,
      "step": 13194
    },
    {
      "epoch": 6.158693115519253,
      "grad_norm": 0.09716796875,
      "learning_rate": 6.440762408947703e-05,
      "loss": 0.0006,
      "step": 13195
    },
    {
      "epoch": 6.159159859976663,
      "grad_norm": 0.11767578125,
      "learning_rate": 6.439391188303874e-05,
      "loss": 0.0008,
      "step": 13196
    },
    {
      "epoch": 6.159626604434072,
      "grad_norm": 0.0218505859375,
      "learning_rate": 6.438020044323902e-05,
      "loss": 0.0003,
      "step": 13197
    },
    {
      "epoch": 6.160093348891482,
      "grad_norm": 0.0830078125,
      "learning_rate": 6.436648977037312e-05,
      "loss": 0.0007,
      "step": 13198
    },
    {
      "epoch": 6.1605600933488915,
      "grad_norm": 0.0269775390625,
      "learning_rate": 6.435277986473622e-05,
      "loss": 0.0003,
      "step": 13199
    },
    {
      "epoch": 6.161026837806301,
      "grad_norm": 0.1513671875,
      "learning_rate": 6.433907072662355e-05,
      "loss": 0.0008,
      "step": 13200
    },
    {
      "epoch": 6.16149358226371,
      "grad_norm": 0.0224609375,
      "learning_rate": 6.432536235633024e-05,
      "loss": 0.0003,
      "step": 13201
    },
    {
      "epoch": 6.16196032672112,
      "grad_norm": 0.0194091796875,
      "learning_rate": 6.431165475415144e-05,
      "loss": 0.0003,
      "step": 13202
    },
    {
      "epoch": 6.16242707117853,
      "grad_norm": 0.0947265625,
      "learning_rate": 6.429794792038232e-05,
      "loss": 0.0005,
      "step": 13203
    },
    {
      "epoch": 6.16289381563594,
      "grad_norm": 0.07421875,
      "learning_rate": 6.428424185531801e-05,
      "loss": 0.0059,
      "step": 13204
    },
    {
      "epoch": 6.1633605600933485,
      "grad_norm": 0.1796875,
      "learning_rate": 6.427053655925356e-05,
      "loss": 0.0007,
      "step": 13205
    },
    {
      "epoch": 6.163827304550758,
      "grad_norm": 0.54296875,
      "learning_rate": 6.425683203248413e-05,
      "loss": 0.0015,
      "step": 13206
    },
    {
      "epoch": 6.164294049008168,
      "grad_norm": 0.030029296875,
      "learning_rate": 6.424312827530478e-05,
      "loss": 0.0004,
      "step": 13207
    },
    {
      "epoch": 6.164760793465578,
      "grad_norm": 0.06591796875,
      "learning_rate": 6.422942528801054e-05,
      "loss": 0.0012,
      "step": 13208
    },
    {
      "epoch": 6.165227537922987,
      "grad_norm": 0.283203125,
      "learning_rate": 6.421572307089643e-05,
      "loss": 0.0037,
      "step": 13209
    },
    {
      "epoch": 6.165694282380397,
      "grad_norm": 0.042236328125,
      "learning_rate": 6.420202162425755e-05,
      "loss": 0.0003,
      "step": 13210
    },
    {
      "epoch": 6.166161026837806,
      "grad_norm": 0.048828125,
      "learning_rate": 6.418832094838885e-05,
      "loss": 0.0004,
      "step": 13211
    },
    {
      "epoch": 6.166627771295216,
      "grad_norm": 0.041015625,
      "learning_rate": 6.41746210435853e-05,
      "loss": 0.0003,
      "step": 13212
    },
    {
      "epoch": 6.167094515752625,
      "grad_norm": 0.30859375,
      "learning_rate": 6.416092191014193e-05,
      "loss": 0.0039,
      "step": 13213
    },
    {
      "epoch": 6.167561260210035,
      "grad_norm": 0.10888671875,
      "learning_rate": 6.414722354835369e-05,
      "loss": 0.0026,
      "step": 13214
    },
    {
      "epoch": 6.168028004667445,
      "grad_norm": 0.1982421875,
      "learning_rate": 6.41335259585155e-05,
      "loss": 0.0011,
      "step": 13215
    },
    {
      "epoch": 6.1684947491248545,
      "grad_norm": 0.0166015625,
      "learning_rate": 6.411982914092229e-05,
      "loss": 0.0002,
      "step": 13216
    },
    {
      "epoch": 6.168961493582263,
      "grad_norm": 0.119140625,
      "learning_rate": 6.410613309586897e-05,
      "loss": 0.0007,
      "step": 13217
    },
    {
      "epoch": 6.169428238039673,
      "grad_norm": 0.359375,
      "learning_rate": 6.409243782365042e-05,
      "loss": 0.0008,
      "step": 13218
    },
    {
      "epoch": 6.169894982497083,
      "grad_norm": 0.062255859375,
      "learning_rate": 6.407874332456152e-05,
      "loss": 0.0005,
      "step": 13219
    },
    {
      "epoch": 6.170361726954493,
      "grad_norm": 0.2080078125,
      "learning_rate": 6.406504959889714e-05,
      "loss": 0.0073,
      "step": 13220
    },
    {
      "epoch": 6.170828471411902,
      "grad_norm": 0.08349609375,
      "learning_rate": 6.405135664695212e-05,
      "loss": 0.0004,
      "step": 13221
    },
    {
      "epoch": 6.1712952158693115,
      "grad_norm": 0.37109375,
      "learning_rate": 6.403766446902125e-05,
      "loss": 0.001,
      "step": 13222
    },
    {
      "epoch": 6.171761960326721,
      "grad_norm": 0.0322265625,
      "learning_rate": 6.40239730653994e-05,
      "loss": 0.0003,
      "step": 13223
    },
    {
      "epoch": 6.172228704784131,
      "grad_norm": 0.10400390625,
      "learning_rate": 6.401028243638134e-05,
      "loss": 0.0006,
      "step": 13224
    },
    {
      "epoch": 6.17269544924154,
      "grad_norm": 0.431640625,
      "learning_rate": 6.39965925822618e-05,
      "loss": 0.0036,
      "step": 13225
    },
    {
      "epoch": 6.17316219369895,
      "grad_norm": 0.1435546875,
      "learning_rate": 6.398290350333558e-05,
      "loss": 0.0006,
      "step": 13226
    },
    {
      "epoch": 6.17362893815636,
      "grad_norm": 0.201171875,
      "learning_rate": 6.396921519989738e-05,
      "loss": 0.0011,
      "step": 13227
    },
    {
      "epoch": 6.174095682613769,
      "grad_norm": 0.028076171875,
      "learning_rate": 6.395552767224198e-05,
      "loss": 0.0003,
      "step": 13228
    },
    {
      "epoch": 6.174562427071178,
      "grad_norm": 0.0277099609375,
      "learning_rate": 6.394184092066405e-05,
      "loss": 0.0003,
      "step": 13229
    },
    {
      "epoch": 6.175029171528588,
      "grad_norm": 0.0181884765625,
      "learning_rate": 6.392815494545831e-05,
      "loss": 0.0002,
      "step": 13230
    },
    {
      "epoch": 6.175495915985998,
      "grad_norm": 0.07861328125,
      "learning_rate": 6.391446974691938e-05,
      "loss": 0.0007,
      "step": 13231
    },
    {
      "epoch": 6.175962660443407,
      "grad_norm": 0.1025390625,
      "learning_rate": 6.390078532534198e-05,
      "loss": 0.0006,
      "step": 13232
    },
    {
      "epoch": 6.176429404900817,
      "grad_norm": 0.0654296875,
      "learning_rate": 6.388710168102073e-05,
      "loss": 0.0006,
      "step": 13233
    },
    {
      "epoch": 6.176896149358226,
      "grad_norm": 0.08984375,
      "learning_rate": 6.387341881425022e-05,
      "loss": 0.0007,
      "step": 13234
    },
    {
      "epoch": 6.177362893815636,
      "grad_norm": 0.09375,
      "learning_rate": 6.385973672532511e-05,
      "loss": 0.0006,
      "step": 13235
    },
    {
      "epoch": 6.177829638273045,
      "grad_norm": 0.038818359375,
      "learning_rate": 6.384605541453998e-05,
      "loss": 0.0003,
      "step": 13236
    },
    {
      "epoch": 6.178296382730455,
      "grad_norm": 0.2734375,
      "learning_rate": 6.383237488218935e-05,
      "loss": 0.0013,
      "step": 13237
    },
    {
      "epoch": 6.178763127187865,
      "grad_norm": 0.1982421875,
      "learning_rate": 6.381869512856784e-05,
      "loss": 0.0008,
      "step": 13238
    },
    {
      "epoch": 6.1792298716452745,
      "grad_norm": 0.03662109375,
      "learning_rate": 6.380501615396998e-05,
      "loss": 0.0003,
      "step": 13239
    },
    {
      "epoch": 6.179696616102683,
      "grad_norm": 0.232421875,
      "learning_rate": 6.379133795869028e-05,
      "loss": 0.0014,
      "step": 13240
    },
    {
      "epoch": 6.180163360560093,
      "grad_norm": 0.0732421875,
      "learning_rate": 6.377766054302322e-05,
      "loss": 0.0005,
      "step": 13241
    },
    {
      "epoch": 6.180630105017503,
      "grad_norm": 0.126953125,
      "learning_rate": 6.376398390726334e-05,
      "loss": 0.0005,
      "step": 13242
    },
    {
      "epoch": 6.181096849474913,
      "grad_norm": 0.099609375,
      "learning_rate": 6.37503080517051e-05,
      "loss": 0.0006,
      "step": 13243
    },
    {
      "epoch": 6.181563593932322,
      "grad_norm": 0.083984375,
      "learning_rate": 6.37366329766429e-05,
      "loss": 0.0031,
      "step": 13244
    },
    {
      "epoch": 6.1820303383897315,
      "grad_norm": 0.036865234375,
      "learning_rate": 6.372295868237128e-05,
      "loss": 0.0003,
      "step": 13245
    },
    {
      "epoch": 6.182497082847141,
      "grad_norm": 0.251953125,
      "learning_rate": 6.37092851691846e-05,
      "loss": 0.0011,
      "step": 13246
    },
    {
      "epoch": 6.182963827304551,
      "grad_norm": 0.267578125,
      "learning_rate": 6.369561243737724e-05,
      "loss": 0.0013,
      "step": 13247
    },
    {
      "epoch": 6.18343057176196,
      "grad_norm": 0.08251953125,
      "learning_rate": 6.368194048724366e-05,
      "loss": 0.0004,
      "step": 13248
    },
    {
      "epoch": 6.18389731621937,
      "grad_norm": 0.0771484375,
      "learning_rate": 6.366826931907819e-05,
      "loss": 0.0004,
      "step": 13249
    },
    {
      "epoch": 6.18436406067678,
      "grad_norm": 0.0302734375,
      "learning_rate": 6.365459893317519e-05,
      "loss": 0.0003,
      "step": 13250
    },
    {
      "epoch": 6.184830805134189,
      "grad_norm": 0.0546875,
      "learning_rate": 6.364092932982899e-05,
      "loss": 0.0004,
      "step": 13251
    },
    {
      "epoch": 6.185297549591598,
      "grad_norm": 0.0159912109375,
      "learning_rate": 6.362726050933395e-05,
      "loss": 0.0002,
      "step": 13252
    },
    {
      "epoch": 6.185764294049008,
      "grad_norm": 0.08935546875,
      "learning_rate": 6.361359247198433e-05,
      "loss": 0.0005,
      "step": 13253
    },
    {
      "epoch": 6.186231038506418,
      "grad_norm": 0.09423828125,
      "learning_rate": 6.359992521807441e-05,
      "loss": 0.0005,
      "step": 13254
    },
    {
      "epoch": 6.186697782963828,
      "grad_norm": 0.031982421875,
      "learning_rate": 6.358625874789853e-05,
      "loss": 0.0004,
      "step": 13255
    },
    {
      "epoch": 6.187164527421237,
      "grad_norm": 0.03857421875,
      "learning_rate": 6.357259306175088e-05,
      "loss": 0.0004,
      "step": 13256
    },
    {
      "epoch": 6.187631271878646,
      "grad_norm": 0.028564453125,
      "learning_rate": 6.355892815992572e-05,
      "loss": 0.0003,
      "step": 13257
    },
    {
      "epoch": 6.188098016336056,
      "grad_norm": 0.0299072265625,
      "learning_rate": 6.354526404271726e-05,
      "loss": 0.0004,
      "step": 13258
    },
    {
      "epoch": 6.188564760793466,
      "grad_norm": 0.023193359375,
      "learning_rate": 6.353160071041972e-05,
      "loss": 0.0003,
      "step": 13259
    },
    {
      "epoch": 6.189031505250875,
      "grad_norm": 0.0546875,
      "learning_rate": 6.351793816332729e-05,
      "loss": 0.0022,
      "step": 13260
    },
    {
      "epoch": 6.189498249708285,
      "grad_norm": 0.0206298828125,
      "learning_rate": 6.350427640173411e-05,
      "loss": 0.0002,
      "step": 13261
    },
    {
      "epoch": 6.1899649941656945,
      "grad_norm": 0.07275390625,
      "learning_rate": 6.349061542593438e-05,
      "loss": 0.0004,
      "step": 13262
    },
    {
      "epoch": 6.190431738623104,
      "grad_norm": 0.037841796875,
      "learning_rate": 6.34769552362222e-05,
      "loss": 0.0004,
      "step": 13263
    },
    {
      "epoch": 6.190898483080513,
      "grad_norm": 0.0478515625,
      "learning_rate": 6.346329583289166e-05,
      "loss": 0.0005,
      "step": 13264
    },
    {
      "epoch": 6.191365227537923,
      "grad_norm": 0.0771484375,
      "learning_rate": 6.344963721623694e-05,
      "loss": 0.0005,
      "step": 13265
    },
    {
      "epoch": 6.191831971995333,
      "grad_norm": 0.2041015625,
      "learning_rate": 6.343597938655206e-05,
      "loss": 0.0006,
      "step": 13266
    },
    {
      "epoch": 6.1922987164527425,
      "grad_norm": 0.036376953125,
      "learning_rate": 6.342232234413114e-05,
      "loss": 0.0004,
      "step": 13267
    },
    {
      "epoch": 6.1927654609101515,
      "grad_norm": 0.031982421875,
      "learning_rate": 6.340866608926816e-05,
      "loss": 0.0003,
      "step": 13268
    },
    {
      "epoch": 6.193232205367561,
      "grad_norm": 0.04736328125,
      "learning_rate": 6.339501062225724e-05,
      "loss": 0.0004,
      "step": 13269
    },
    {
      "epoch": 6.193698949824971,
      "grad_norm": 0.03857421875,
      "learning_rate": 6.338135594339234e-05,
      "loss": 0.0023,
      "step": 13270
    },
    {
      "epoch": 6.194165694282381,
      "grad_norm": 0.154296875,
      "learning_rate": 6.336770205296746e-05,
      "loss": 0.0009,
      "step": 13271
    },
    {
      "epoch": 6.19463243873979,
      "grad_norm": 0.25390625,
      "learning_rate": 6.335404895127666e-05,
      "loss": 0.0005,
      "step": 13272
    },
    {
      "epoch": 6.1950991831971995,
      "grad_norm": 0.10205078125,
      "learning_rate": 6.334039663861377e-05,
      "loss": 0.0005,
      "step": 13273
    },
    {
      "epoch": 6.195565927654609,
      "grad_norm": 0.0111083984375,
      "learning_rate": 6.332674511527285e-05,
      "loss": 0.0002,
      "step": 13274
    },
    {
      "epoch": 6.196032672112018,
      "grad_norm": 0.10205078125,
      "learning_rate": 6.33130943815478e-05,
      "loss": 0.0046,
      "step": 13275
    },
    {
      "epoch": 6.196499416569428,
      "grad_norm": 0.07763671875,
      "learning_rate": 6.329944443773253e-05,
      "loss": 0.0033,
      "step": 13276
    },
    {
      "epoch": 6.196966161026838,
      "grad_norm": 0.150390625,
      "learning_rate": 6.328579528412094e-05,
      "loss": 0.0005,
      "step": 13277
    },
    {
      "epoch": 6.197432905484248,
      "grad_norm": 0.30859375,
      "learning_rate": 6.327214692100691e-05,
      "loss": 0.0018,
      "step": 13278
    },
    {
      "epoch": 6.1978996499416565,
      "grad_norm": 0.1162109375,
      "learning_rate": 6.32584993486843e-05,
      "loss": 0.0017,
      "step": 13279
    },
    {
      "epoch": 6.198366394399066,
      "grad_norm": 0.1298828125,
      "learning_rate": 6.324485256744696e-05,
      "loss": 0.0011,
      "step": 13280
    },
    {
      "epoch": 6.198833138856476,
      "grad_norm": 0.055419921875,
      "learning_rate": 6.323120657758874e-05,
      "loss": 0.0003,
      "step": 13281
    },
    {
      "epoch": 6.199299883313886,
      "grad_norm": 0.06689453125,
      "learning_rate": 6.321756137940343e-05,
      "loss": 0.0003,
      "step": 13282
    },
    {
      "epoch": 6.199766627771295,
      "grad_norm": 0.1806640625,
      "learning_rate": 6.320391697318481e-05,
      "loss": 0.0008,
      "step": 13283
    },
    {
      "epoch": 6.200233372228705,
      "grad_norm": 0.01953125,
      "learning_rate": 6.319027335922673e-05,
      "loss": 0.0003,
      "step": 13284
    },
    {
      "epoch": 6.200700116686114,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.317663053782289e-05,
      "loss": 0.0004,
      "step": 13285
    },
    {
      "epoch": 6.201166861143524,
      "grad_norm": 0.0693359375,
      "learning_rate": 6.316298850926703e-05,
      "loss": 0.006,
      "step": 13286
    },
    {
      "epoch": 6.201633605600933,
      "grad_norm": 0.44140625,
      "learning_rate": 6.314934727385293e-05,
      "loss": 0.0051,
      "step": 13287
    },
    {
      "epoch": 6.202100350058343,
      "grad_norm": 0.035888671875,
      "learning_rate": 6.313570683187425e-05,
      "loss": 0.0003,
      "step": 13288
    },
    {
      "epoch": 6.202567094515753,
      "grad_norm": 0.291015625,
      "learning_rate": 6.312206718362472e-05,
      "loss": 0.0015,
      "step": 13289
    },
    {
      "epoch": 6.2030338389731625,
      "grad_norm": 0.03076171875,
      "learning_rate": 6.310842832939797e-05,
      "loss": 0.0004,
      "step": 13290
    },
    {
      "epoch": 6.203500583430571,
      "grad_norm": 0.07568359375,
      "learning_rate": 6.309479026948772e-05,
      "loss": 0.0004,
      "step": 13291
    },
    {
      "epoch": 6.203967327887981,
      "grad_norm": 0.064453125,
      "learning_rate": 6.308115300418758e-05,
      "loss": 0.0022,
      "step": 13292
    },
    {
      "epoch": 6.204434072345391,
      "grad_norm": 0.0361328125,
      "learning_rate": 6.306751653379115e-05,
      "loss": 0.0003,
      "step": 13293
    },
    {
      "epoch": 6.204900816802801,
      "grad_norm": 0.020751953125,
      "learning_rate": 6.30538808585921e-05,
      "loss": 0.0003,
      "step": 13294
    },
    {
      "epoch": 6.20536756126021,
      "grad_norm": 0.1328125,
      "learning_rate": 6.304024597888396e-05,
      "loss": 0.0006,
      "step": 13295
    },
    {
      "epoch": 6.2058343057176195,
      "grad_norm": 0.0264892578125,
      "learning_rate": 6.302661189496034e-05,
      "loss": 0.0003,
      "step": 13296
    },
    {
      "epoch": 6.206301050175029,
      "grad_norm": 0.21875,
      "learning_rate": 6.301297860711479e-05,
      "loss": 0.0034,
      "step": 13297
    },
    {
      "epoch": 6.206767794632439,
      "grad_norm": 0.19140625,
      "learning_rate": 6.299934611564087e-05,
      "loss": 0.0007,
      "step": 13298
    },
    {
      "epoch": 6.207234539089848,
      "grad_norm": 0.1201171875,
      "learning_rate": 6.298571442083207e-05,
      "loss": 0.0005,
      "step": 13299
    },
    {
      "epoch": 6.207701283547258,
      "grad_norm": 0.05029296875,
      "learning_rate": 6.297208352298187e-05,
      "loss": 0.0005,
      "step": 13300
    },
    {
      "epoch": 6.208168028004668,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.295845342238384e-05,
      "loss": 0.0022,
      "step": 13301
    },
    {
      "epoch": 6.208634772462077,
      "grad_norm": 0.11181640625,
      "learning_rate": 6.29448241193314e-05,
      "loss": 0.0004,
      "step": 13302
    },
    {
      "epoch": 6.209101516919486,
      "grad_norm": 0.244140625,
      "learning_rate": 6.293119561411797e-05,
      "loss": 0.0087,
      "step": 13303
    },
    {
      "epoch": 6.209568261376896,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.291756790703706e-05,
      "loss": 0.0014,
      "step": 13304
    },
    {
      "epoch": 6.210035005834306,
      "grad_norm": 0.259765625,
      "learning_rate": 6.290394099838207e-05,
      "loss": 0.004,
      "step": 13305
    },
    {
      "epoch": 6.210501750291716,
      "grad_norm": 0.064453125,
      "learning_rate": 6.289031488844636e-05,
      "loss": 0.0005,
      "step": 13306
    },
    {
      "epoch": 6.210968494749125,
      "grad_norm": 0.08154296875,
      "learning_rate": 6.287668957752336e-05,
      "loss": 0.0004,
      "step": 13307
    },
    {
      "epoch": 6.211435239206534,
      "grad_norm": 0.08642578125,
      "learning_rate": 6.286306506590642e-05,
      "loss": 0.0005,
      "step": 13308
    },
    {
      "epoch": 6.211901983663944,
      "grad_norm": 0.06005859375,
      "learning_rate": 6.284944135388891e-05,
      "loss": 0.0005,
      "step": 13309
    },
    {
      "epoch": 6.212368728121354,
      "grad_norm": 0.05078125,
      "learning_rate": 6.283581844176411e-05,
      "loss": 0.0004,
      "step": 13310
    },
    {
      "epoch": 6.212835472578763,
      "grad_norm": 0.060302734375,
      "learning_rate": 6.28221963298254e-05,
      "loss": 0.0049,
      "step": 13311
    },
    {
      "epoch": 6.213302217036173,
      "grad_norm": 0.18359375,
      "learning_rate": 6.280857501836605e-05,
      "loss": 0.0009,
      "step": 13312
    },
    {
      "epoch": 6.2137689614935825,
      "grad_norm": 0.083984375,
      "learning_rate": 6.279495450767934e-05,
      "loss": 0.0004,
      "step": 13313
    },
    {
      "epoch": 6.214235705950992,
      "grad_norm": 0.064453125,
      "learning_rate": 6.278133479805855e-05,
      "loss": 0.0027,
      "step": 13314
    },
    {
      "epoch": 6.214702450408401,
      "grad_norm": 0.059326171875,
      "learning_rate": 6.27677158897969e-05,
      "loss": 0.0004,
      "step": 13315
    },
    {
      "epoch": 6.215169194865811,
      "grad_norm": 0.0125732421875,
      "learning_rate": 6.275409778318765e-05,
      "loss": 0.0002,
      "step": 13316
    },
    {
      "epoch": 6.215635939323221,
      "grad_norm": 0.1865234375,
      "learning_rate": 6.274048047852396e-05,
      "loss": 0.0007,
      "step": 13317
    },
    {
      "epoch": 6.21610268378063,
      "grad_norm": 0.0220947265625,
      "learning_rate": 6.272686397609913e-05,
      "loss": 0.0002,
      "step": 13318
    },
    {
      "epoch": 6.2165694282380395,
      "grad_norm": 0.189453125,
      "learning_rate": 6.271324827620624e-05,
      "loss": 0.0009,
      "step": 13319
    },
    {
      "epoch": 6.217036172695449,
      "grad_norm": 0.09375,
      "learning_rate": 6.269963337913852e-05,
      "loss": 0.0028,
      "step": 13320
    },
    {
      "epoch": 6.217502917152859,
      "grad_norm": 0.1826171875,
      "learning_rate": 6.268601928518905e-05,
      "loss": 0.0011,
      "step": 13321
    },
    {
      "epoch": 6.217969661610268,
      "grad_norm": 0.016845703125,
      "learning_rate": 6.267240599465096e-05,
      "loss": 0.0003,
      "step": 13322
    },
    {
      "epoch": 6.218436406067678,
      "grad_norm": 0.024658203125,
      "learning_rate": 6.26587935078174e-05,
      "loss": 0.0003,
      "step": 13323
    },
    {
      "epoch": 6.218903150525088,
      "grad_norm": 0.259765625,
      "learning_rate": 6.264518182498147e-05,
      "loss": 0.0016,
      "step": 13324
    },
    {
      "epoch": 6.219369894982497,
      "grad_norm": 0.023193359375,
      "learning_rate": 6.26315709464362e-05,
      "loss": 0.0003,
      "step": 13325
    },
    {
      "epoch": 6.219836639439906,
      "grad_norm": 0.1181640625,
      "learning_rate": 6.261796087247467e-05,
      "loss": 0.0011,
      "step": 13326
    },
    {
      "epoch": 6.220303383897316,
      "grad_norm": 0.0230712890625,
      "learning_rate": 6.260435160338992e-05,
      "loss": 0.0003,
      "step": 13327
    },
    {
      "epoch": 6.220770128354726,
      "grad_norm": 0.047119140625,
      "learning_rate": 6.259074313947497e-05,
      "loss": 0.0003,
      "step": 13328
    },
    {
      "epoch": 6.221236872812136,
      "grad_norm": 0.1689453125,
      "learning_rate": 6.257713548102282e-05,
      "loss": 0.0006,
      "step": 13329
    },
    {
      "epoch": 6.221703617269545,
      "grad_norm": 0.026123046875,
      "learning_rate": 6.256352862832649e-05,
      "loss": 0.0003,
      "step": 13330
    },
    {
      "epoch": 6.222170361726954,
      "grad_norm": 0.044189453125,
      "learning_rate": 6.254992258167891e-05,
      "loss": 0.0021,
      "step": 13331
    },
    {
      "epoch": 6.222637106184364,
      "grad_norm": 0.392578125,
      "learning_rate": 6.253631734137301e-05,
      "loss": 0.0017,
      "step": 13332
    },
    {
      "epoch": 6.223103850641774,
      "grad_norm": 0.0274658203125,
      "learning_rate": 6.252271290770182e-05,
      "loss": 0.0004,
      "step": 13333
    },
    {
      "epoch": 6.223570595099183,
      "grad_norm": 0.017822265625,
      "learning_rate": 6.250910928095819e-05,
      "loss": 0.0003,
      "step": 13334
    },
    {
      "epoch": 6.224037339556593,
      "grad_norm": 0.146484375,
      "learning_rate": 6.249550646143501e-05,
      "loss": 0.0054,
      "step": 13335
    },
    {
      "epoch": 6.2245040840140025,
      "grad_norm": 0.0537109375,
      "learning_rate": 6.248190444942521e-05,
      "loss": 0.0003,
      "step": 13336
    },
    {
      "epoch": 6.224970828471412,
      "grad_norm": 0.0693359375,
      "learning_rate": 6.246830324522164e-05,
      "loss": 0.0004,
      "step": 13337
    },
    {
      "epoch": 6.225437572928821,
      "grad_norm": 0.0135498046875,
      "learning_rate": 6.245470284911715e-05,
      "loss": 0.0002,
      "step": 13338
    },
    {
      "epoch": 6.225904317386231,
      "grad_norm": 0.055908203125,
      "learning_rate": 6.244110326140452e-05,
      "loss": 0.0003,
      "step": 13339
    },
    {
      "epoch": 6.226371061843641,
      "grad_norm": 0.053955078125,
      "learning_rate": 6.242750448237664e-05,
      "loss": 0.0004,
      "step": 13340
    },
    {
      "epoch": 6.2268378063010505,
      "grad_norm": 0.0107421875,
      "learning_rate": 6.241390651232628e-05,
      "loss": 0.0002,
      "step": 13341
    },
    {
      "epoch": 6.2273045507584595,
      "grad_norm": 0.1904296875,
      "learning_rate": 6.24003093515462e-05,
      "loss": 0.0008,
      "step": 13342
    },
    {
      "epoch": 6.227771295215869,
      "grad_norm": 0.189453125,
      "learning_rate": 6.238671300032918e-05,
      "loss": 0.0006,
      "step": 13343
    },
    {
      "epoch": 6.228238039673279,
      "grad_norm": 0.0361328125,
      "learning_rate": 6.237311745896797e-05,
      "loss": 0.0003,
      "step": 13344
    },
    {
      "epoch": 6.228704784130689,
      "grad_norm": 0.1455078125,
      "learning_rate": 6.235952272775528e-05,
      "loss": 0.0005,
      "step": 13345
    },
    {
      "epoch": 6.229171528588098,
      "grad_norm": 0.061767578125,
      "learning_rate": 6.234592880698384e-05,
      "loss": 0.0004,
      "step": 13346
    },
    {
      "epoch": 6.2296382730455075,
      "grad_norm": 0.11962890625,
      "learning_rate": 6.233233569694633e-05,
      "loss": 0.0007,
      "step": 13347
    },
    {
      "epoch": 6.230105017502917,
      "grad_norm": 0.047119140625,
      "learning_rate": 6.231874339793544e-05,
      "loss": 0.003,
      "step": 13348
    },
    {
      "epoch": 6.230571761960327,
      "grad_norm": 0.0113525390625,
      "learning_rate": 6.230515191024376e-05,
      "loss": 0.0002,
      "step": 13349
    },
    {
      "epoch": 6.231038506417736,
      "grad_norm": 0.1494140625,
      "learning_rate": 6.229156123416403e-05,
      "loss": 0.0006,
      "step": 13350
    },
    {
      "epoch": 6.231505250875146,
      "grad_norm": 0.027587890625,
      "learning_rate": 6.227797136998881e-05,
      "loss": 0.0003,
      "step": 13351
    },
    {
      "epoch": 6.231971995332556,
      "grad_norm": 0.050537109375,
      "learning_rate": 6.22643823180107e-05,
      "loss": 0.0003,
      "step": 13352
    },
    {
      "epoch": 6.2324387397899645,
      "grad_norm": 0.05615234375,
      "learning_rate": 6.225079407852235e-05,
      "loss": 0.0004,
      "step": 13353
    },
    {
      "epoch": 6.232905484247374,
      "grad_norm": 0.046142578125,
      "learning_rate": 6.223720665181627e-05,
      "loss": 0.0004,
      "step": 13354
    },
    {
      "epoch": 6.233372228704784,
      "grad_norm": 0.1376953125,
      "learning_rate": 6.222362003818501e-05,
      "loss": 0.0043,
      "step": 13355
    },
    {
      "epoch": 6.233838973162194,
      "grad_norm": 0.033203125,
      "learning_rate": 6.221003423792114e-05,
      "loss": 0.0003,
      "step": 13356
    },
    {
      "epoch": 6.234305717619603,
      "grad_norm": 0.059814453125,
      "learning_rate": 6.219644925131716e-05,
      "loss": 0.0023,
      "step": 13357
    },
    {
      "epoch": 6.234772462077013,
      "grad_norm": 0.072265625,
      "learning_rate": 6.218286507866557e-05,
      "loss": 0.0004,
      "step": 13358
    },
    {
      "epoch": 6.235239206534422,
      "grad_norm": 0.08251953125,
      "learning_rate": 6.216928172025884e-05,
      "loss": 0.0006,
      "step": 13359
    },
    {
      "epoch": 6.235705950991832,
      "grad_norm": 0.0283203125,
      "learning_rate": 6.215569917638948e-05,
      "loss": 0.0002,
      "step": 13360
    },
    {
      "epoch": 6.236172695449241,
      "grad_norm": 0.023681640625,
      "learning_rate": 6.214211744734989e-05,
      "loss": 0.0003,
      "step": 13361
    },
    {
      "epoch": 6.236639439906651,
      "grad_norm": 0.04833984375,
      "learning_rate": 6.212853653343251e-05,
      "loss": 0.0018,
      "step": 13362
    },
    {
      "epoch": 6.237106184364061,
      "grad_norm": 0.326171875,
      "learning_rate": 6.211495643492974e-05,
      "loss": 0.0011,
      "step": 13363
    },
    {
      "epoch": 6.2375729288214705,
      "grad_norm": 0.04150390625,
      "learning_rate": 6.210137715213405e-05,
      "loss": 0.0002,
      "step": 13364
    },
    {
      "epoch": 6.238039673278879,
      "grad_norm": 0.029541015625,
      "learning_rate": 6.208779868533769e-05,
      "loss": 0.0003,
      "step": 13365
    },
    {
      "epoch": 6.238506417736289,
      "grad_norm": 0.0159912109375,
      "learning_rate": 6.207422103483314e-05,
      "loss": 0.0002,
      "step": 13366
    },
    {
      "epoch": 6.238973162193699,
      "grad_norm": 0.1708984375,
      "learning_rate": 6.206064420091266e-05,
      "loss": 0.0009,
      "step": 13367
    },
    {
      "epoch": 6.239439906651109,
      "grad_norm": 0.0751953125,
      "learning_rate": 6.204706818386865e-05,
      "loss": 0.0053,
      "step": 13368
    },
    {
      "epoch": 6.239906651108518,
      "grad_norm": 0.3125,
      "learning_rate": 6.203349298399334e-05,
      "loss": 0.0038,
      "step": 13369
    },
    {
      "epoch": 6.2403733955659275,
      "grad_norm": 0.46484375,
      "learning_rate": 6.201991860157906e-05,
      "loss": 0.0016,
      "step": 13370
    },
    {
      "epoch": 6.240840140023337,
      "grad_norm": 0.048828125,
      "learning_rate": 6.200634503691804e-05,
      "loss": 0.003,
      "step": 13371
    },
    {
      "epoch": 6.241306884480747,
      "grad_norm": 0.125,
      "learning_rate": 6.19927722903026e-05,
      "loss": 0.0006,
      "step": 13372
    },
    {
      "epoch": 6.241773628938156,
      "grad_norm": 0.036376953125,
      "learning_rate": 6.197920036202494e-05,
      "loss": 0.0017,
      "step": 13373
    },
    {
      "epoch": 6.242240373395566,
      "grad_norm": 0.09228515625,
      "learning_rate": 6.196562925237725e-05,
      "loss": 0.0004,
      "step": 13374
    },
    {
      "epoch": 6.242707117852976,
      "grad_norm": 0.177734375,
      "learning_rate": 6.19520589616518e-05,
      "loss": 0.0006,
      "step": 13375
    },
    {
      "epoch": 6.243173862310385,
      "grad_norm": 0.047607421875,
      "learning_rate": 6.193848949014071e-05,
      "loss": 0.0005,
      "step": 13376
    },
    {
      "epoch": 6.243640606767794,
      "grad_norm": 0.162109375,
      "learning_rate": 6.192492083813617e-05,
      "loss": 0.0005,
      "step": 13377
    },
    {
      "epoch": 6.244107351225204,
      "grad_norm": 0.12158203125,
      "learning_rate": 6.191135300593034e-05,
      "loss": 0.002,
      "step": 13378
    },
    {
      "epoch": 6.244574095682614,
      "grad_norm": 0.0225830078125,
      "learning_rate": 6.189778599381536e-05,
      "loss": 0.0003,
      "step": 13379
    },
    {
      "epoch": 6.245040840140024,
      "grad_norm": 0.054931640625,
      "learning_rate": 6.188421980208329e-05,
      "loss": 0.0004,
      "step": 13380
    },
    {
      "epoch": 6.245507584597433,
      "grad_norm": 0.02294921875,
      "learning_rate": 6.187065443102626e-05,
      "loss": 0.0003,
      "step": 13381
    },
    {
      "epoch": 6.245974329054842,
      "grad_norm": 0.00946044921875,
      "learning_rate": 6.185708988093635e-05,
      "loss": 0.0002,
      "step": 13382
    },
    {
      "epoch": 6.246441073512252,
      "grad_norm": 0.02294921875,
      "learning_rate": 6.184352615210563e-05,
      "loss": 0.0003,
      "step": 13383
    },
    {
      "epoch": 6.246907817969662,
      "grad_norm": 0.107421875,
      "learning_rate": 6.18299632448261e-05,
      "loss": 0.0006,
      "step": 13384
    },
    {
      "epoch": 6.247374562427071,
      "grad_norm": 0.10302734375,
      "learning_rate": 6.181640115938985e-05,
      "loss": 0.0005,
      "step": 13385
    },
    {
      "epoch": 6.247841306884481,
      "grad_norm": 0.259765625,
      "learning_rate": 6.180283989608882e-05,
      "loss": 0.0015,
      "step": 13386
    },
    {
      "epoch": 6.2483080513418905,
      "grad_norm": 0.28515625,
      "learning_rate": 6.178927945521505e-05,
      "loss": 0.0012,
      "step": 13387
    },
    {
      "epoch": 6.2487747957993,
      "grad_norm": 0.03271484375,
      "learning_rate": 6.177571983706045e-05,
      "loss": 0.0003,
      "step": 13388
    },
    {
      "epoch": 6.249241540256709,
      "grad_norm": 0.080078125,
      "learning_rate": 6.176216104191705e-05,
      "loss": 0.0004,
      "step": 13389
    },
    {
      "epoch": 6.249708284714119,
      "grad_norm": 0.07080078125,
      "learning_rate": 6.174860307007674e-05,
      "loss": 0.0017,
      "step": 13390
    },
    {
      "epoch": 6.250175029171529,
      "grad_norm": 0.048095703125,
      "learning_rate": 6.173504592183142e-05,
      "loss": 0.0004,
      "step": 13391
    },
    {
      "epoch": 6.250641773628939,
      "grad_norm": 0.10009765625,
      "learning_rate": 6.172148959747305e-05,
      "loss": 0.0005,
      "step": 13392
    },
    {
      "epoch": 6.2511085180863475,
      "grad_norm": 0.076171875,
      "learning_rate": 6.170793409729348e-05,
      "loss": 0.0004,
      "step": 13393
    },
    {
      "epoch": 6.251575262543757,
      "grad_norm": 0.1484375,
      "learning_rate": 6.169437942158453e-05,
      "loss": 0.0015,
      "step": 13394
    },
    {
      "epoch": 6.252042007001167,
      "grad_norm": 0.01226806640625,
      "learning_rate": 6.168082557063813e-05,
      "loss": 0.0002,
      "step": 13395
    },
    {
      "epoch": 6.252508751458576,
      "grad_norm": 0.125,
      "learning_rate": 6.166727254474608e-05,
      "loss": 0.004,
      "step": 13396
    },
    {
      "epoch": 6.252975495915986,
      "grad_norm": 0.038818359375,
      "learning_rate": 6.165372034420017e-05,
      "loss": 0.002,
      "step": 13397
    },
    {
      "epoch": 6.253442240373396,
      "grad_norm": 0.034423828125,
      "learning_rate": 6.164016896929218e-05,
      "loss": 0.0004,
      "step": 13398
    },
    {
      "epoch": 6.253908984830805,
      "grad_norm": 0.02880859375,
      "learning_rate": 6.162661842031396e-05,
      "loss": 0.0003,
      "step": 13399
    },
    {
      "epoch": 6.254375729288215,
      "grad_norm": 0.07763671875,
      "learning_rate": 6.161306869755721e-05,
      "loss": 0.0004,
      "step": 13400
    },
    {
      "epoch": 6.254375729288215,
      "eval_loss": 2.343518018722534,
      "eval_runtime": 65.2176,
      "eval_samples_per_second": 27.661,
      "eval_steps_per_second": 3.465,
      "step": 13400
    },
    {
      "epoch": 6.254842473745624,
      "grad_norm": 0.01953125,
      "learning_rate": 6.159951980131364e-05,
      "loss": 0.0002,
      "step": 13401
    },
    {
      "epoch": 6.255309218203034,
      "grad_norm": 0.032958984375,
      "learning_rate": 6.158597173187506e-05,
      "loss": 0.0003,
      "step": 13402
    },
    {
      "epoch": 6.255775962660444,
      "grad_norm": 0.29296875,
      "learning_rate": 6.157242448953314e-05,
      "loss": 0.003,
      "step": 13403
    },
    {
      "epoch": 6.256242707117853,
      "grad_norm": 0.038330078125,
      "learning_rate": 6.15588780745795e-05,
      "loss": 0.0003,
      "step": 13404
    },
    {
      "epoch": 6.256709451575262,
      "grad_norm": 0.043212890625,
      "learning_rate": 6.154533248730591e-05,
      "loss": 0.0006,
      "step": 13405
    },
    {
      "epoch": 6.257176196032672,
      "grad_norm": 0.1259765625,
      "learning_rate": 6.153178772800396e-05,
      "loss": 0.0005,
      "step": 13406
    },
    {
      "epoch": 6.257642940490082,
      "grad_norm": 0.36328125,
      "learning_rate": 6.151824379696533e-05,
      "loss": 0.0018,
      "step": 13407
    },
    {
      "epoch": 6.258109684947491,
      "grad_norm": 0.0703125,
      "learning_rate": 6.150470069448156e-05,
      "loss": 0.0004,
      "step": 13408
    },
    {
      "epoch": 6.258576429404901,
      "grad_norm": 0.05908203125,
      "learning_rate": 6.149115842084433e-05,
      "loss": 0.0004,
      "step": 13409
    },
    {
      "epoch": 6.2590431738623105,
      "grad_norm": 0.73828125,
      "learning_rate": 6.147761697634518e-05,
      "loss": 0.0022,
      "step": 13410
    },
    {
      "epoch": 6.25950991831972,
      "grad_norm": 0.0751953125,
      "learning_rate": 6.146407636127567e-05,
      "loss": 0.0004,
      "step": 13411
    },
    {
      "epoch": 6.259976662777129,
      "grad_norm": 0.11767578125,
      "learning_rate": 6.145053657592737e-05,
      "loss": 0.0047,
      "step": 13412
    },
    {
      "epoch": 6.260443407234539,
      "grad_norm": 0.08349609375,
      "learning_rate": 6.143699762059177e-05,
      "loss": 0.0005,
      "step": 13413
    },
    {
      "epoch": 6.260910151691949,
      "grad_norm": 0.08251953125,
      "learning_rate": 6.142345949556039e-05,
      "loss": 0.0003,
      "step": 13414
    },
    {
      "epoch": 6.2613768961493586,
      "grad_norm": 0.12060546875,
      "learning_rate": 6.140992220112474e-05,
      "loss": 0.0006,
      "step": 13415
    },
    {
      "epoch": 6.2618436406067675,
      "grad_norm": 0.030517578125,
      "learning_rate": 6.139638573757631e-05,
      "loss": 0.0003,
      "step": 13416
    },
    {
      "epoch": 6.262310385064177,
      "grad_norm": 0.1513671875,
      "learning_rate": 6.13828501052065e-05,
      "loss": 0.0078,
      "step": 13417
    },
    {
      "epoch": 6.262777129521587,
      "grad_norm": 0.11767578125,
      "learning_rate": 6.136931530430678e-05,
      "loss": 0.0006,
      "step": 13418
    },
    {
      "epoch": 6.263243873978997,
      "grad_norm": 0.02392578125,
      "learning_rate": 6.135578133516857e-05,
      "loss": 0.0003,
      "step": 13419
    },
    {
      "epoch": 6.263710618436406,
      "grad_norm": 0.0242919921875,
      "learning_rate": 6.134224819808323e-05,
      "loss": 0.0004,
      "step": 13420
    },
    {
      "epoch": 6.2641773628938155,
      "grad_norm": 0.26171875,
      "learning_rate": 6.132871589334221e-05,
      "loss": 0.0011,
      "step": 13421
    },
    {
      "epoch": 6.264644107351225,
      "grad_norm": 0.349609375,
      "learning_rate": 6.131518442123683e-05,
      "loss": 0.001,
      "step": 13422
    },
    {
      "epoch": 6.265110851808635,
      "grad_norm": 0.1533203125,
      "learning_rate": 6.130165378205845e-05,
      "loss": 0.0005,
      "step": 13423
    },
    {
      "epoch": 6.265577596266044,
      "grad_norm": 0.0277099609375,
      "learning_rate": 6.12881239760984e-05,
      "loss": 0.0003,
      "step": 13424
    },
    {
      "epoch": 6.266044340723454,
      "grad_norm": 0.042236328125,
      "learning_rate": 6.1274595003648e-05,
      "loss": 0.0003,
      "step": 13425
    },
    {
      "epoch": 6.266511085180864,
      "grad_norm": 0.0966796875,
      "learning_rate": 6.126106686499853e-05,
      "loss": 0.0021,
      "step": 13426
    },
    {
      "epoch": 6.266977829638273,
      "grad_norm": 0.1630859375,
      "learning_rate": 6.124753956044127e-05,
      "loss": 0.0006,
      "step": 13427
    },
    {
      "epoch": 6.267444574095682,
      "grad_norm": 0.09521484375,
      "learning_rate": 6.12340130902675e-05,
      "loss": 0.002,
      "step": 13428
    },
    {
      "epoch": 6.267911318553092,
      "grad_norm": 0.140625,
      "learning_rate": 6.122048745476843e-05,
      "loss": 0.003,
      "step": 13429
    },
    {
      "epoch": 6.268378063010502,
      "grad_norm": 0.08935546875,
      "learning_rate": 6.120696265423527e-05,
      "loss": 0.0006,
      "step": 13430
    },
    {
      "epoch": 6.268844807467912,
      "grad_norm": 0.2333984375,
      "learning_rate": 6.119343868895926e-05,
      "loss": 0.0009,
      "step": 13431
    },
    {
      "epoch": 6.269311551925321,
      "grad_norm": 0.1484375,
      "learning_rate": 6.117991555923158e-05,
      "loss": 0.0006,
      "step": 13432
    },
    {
      "epoch": 6.26977829638273,
      "grad_norm": 0.0654296875,
      "learning_rate": 6.116639326534336e-05,
      "loss": 0.0003,
      "step": 13433
    },
    {
      "epoch": 6.27024504084014,
      "grad_norm": 0.04638671875,
      "learning_rate": 6.115287180758583e-05,
      "loss": 0.0004,
      "step": 13434
    },
    {
      "epoch": 6.27071178529755,
      "grad_norm": 0.036376953125,
      "learning_rate": 6.113935118625003e-05,
      "loss": 0.0003,
      "step": 13435
    },
    {
      "epoch": 6.271178529754959,
      "grad_norm": 0.040771484375,
      "learning_rate": 6.112583140162712e-05,
      "loss": 0.0003,
      "step": 13436
    },
    {
      "epoch": 6.271645274212369,
      "grad_norm": 0.0262451171875,
      "learning_rate": 6.111231245400822e-05,
      "loss": 0.0003,
      "step": 13437
    },
    {
      "epoch": 6.2721120186697785,
      "grad_norm": 0.361328125,
      "learning_rate": 6.109879434368435e-05,
      "loss": 0.0008,
      "step": 13438
    },
    {
      "epoch": 6.272578763127187,
      "grad_norm": 0.01385498046875,
      "learning_rate": 6.108527707094662e-05,
      "loss": 0.0003,
      "step": 13439
    },
    {
      "epoch": 6.273045507584597,
      "grad_norm": 0.03662109375,
      "learning_rate": 6.107176063608602e-05,
      "loss": 0.0003,
      "step": 13440
    },
    {
      "epoch": 6.273512252042007,
      "grad_norm": 0.27734375,
      "learning_rate": 6.105824503939364e-05,
      "loss": 0.0007,
      "step": 13441
    },
    {
      "epoch": 6.273978996499417,
      "grad_norm": 0.0291748046875,
      "learning_rate": 6.104473028116045e-05,
      "loss": 0.0003,
      "step": 13442
    },
    {
      "epoch": 6.274445740956826,
      "grad_norm": 0.041015625,
      "learning_rate": 6.103121636167739e-05,
      "loss": 0.0004,
      "step": 13443
    },
    {
      "epoch": 6.2749124854142355,
      "grad_norm": 0.1328125,
      "learning_rate": 6.101770328123553e-05,
      "loss": 0.003,
      "step": 13444
    },
    {
      "epoch": 6.275379229871645,
      "grad_norm": 0.2197265625,
      "learning_rate": 6.1004191040125756e-05,
      "loss": 0.0007,
      "step": 13445
    },
    {
      "epoch": 6.275845974329055,
      "grad_norm": 0.05810546875,
      "learning_rate": 6.0990679638639025e-05,
      "loss": 0.0004,
      "step": 13446
    },
    {
      "epoch": 6.276312718786464,
      "grad_norm": 0.1552734375,
      "learning_rate": 6.0977169077066207e-05,
      "loss": 0.0005,
      "step": 13447
    },
    {
      "epoch": 6.276779463243874,
      "grad_norm": 0.376953125,
      "learning_rate": 6.0963659355698274e-05,
      "loss": 0.0014,
      "step": 13448
    },
    {
      "epoch": 6.277246207701284,
      "grad_norm": 0.279296875,
      "learning_rate": 6.0950150474826065e-05,
      "loss": 0.0025,
      "step": 13449
    },
    {
      "epoch": 6.277712952158693,
      "grad_norm": 0.09375,
      "learning_rate": 6.0936642434740396e-05,
      "loss": 0.0038,
      "step": 13450
    },
    {
      "epoch": 6.278179696616102,
      "grad_norm": 0.236328125,
      "learning_rate": 6.092313523573221e-05,
      "loss": 0.0013,
      "step": 13451
    },
    {
      "epoch": 6.278646441073512,
      "grad_norm": 0.0322265625,
      "learning_rate": 6.0909628878092263e-05,
      "loss": 0.0003,
      "step": 13452
    },
    {
      "epoch": 6.279113185530922,
      "grad_norm": 0.058349609375,
      "learning_rate": 6.089612336211136e-05,
      "loss": 0.0003,
      "step": 13453
    },
    {
      "epoch": 6.279579929988332,
      "grad_norm": 0.08251953125,
      "learning_rate": 6.088261868808033e-05,
      "loss": 0.0006,
      "step": 13454
    },
    {
      "epoch": 6.280046674445741,
      "grad_norm": 0.1865234375,
      "learning_rate": 6.086911485628992e-05,
      "loss": 0.0006,
      "step": 13455
    },
    {
      "epoch": 6.28051341890315,
      "grad_norm": 0.0277099609375,
      "learning_rate": 6.0855611867030884e-05,
      "loss": 0.0002,
      "step": 13456
    },
    {
      "epoch": 6.28098016336056,
      "grad_norm": 0.1044921875,
      "learning_rate": 6.084210972059393e-05,
      "loss": 0.0007,
      "step": 13457
    },
    {
      "epoch": 6.28144690781797,
      "grad_norm": 0.220703125,
      "learning_rate": 6.082860841726983e-05,
      "loss": 0.0012,
      "step": 13458
    },
    {
      "epoch": 6.281913652275379,
      "grad_norm": 0.14453125,
      "learning_rate": 6.0815107957349237e-05,
      "loss": 0.0007,
      "step": 13459
    },
    {
      "epoch": 6.282380396732789,
      "grad_norm": 0.19921875,
      "learning_rate": 6.080160834112284e-05,
      "loss": 0.0013,
      "step": 13460
    },
    {
      "epoch": 6.2828471411901985,
      "grad_norm": 0.1474609375,
      "learning_rate": 6.078810956888133e-05,
      "loss": 0.0006,
      "step": 13461
    },
    {
      "epoch": 6.283313885647608,
      "grad_norm": 0.047119140625,
      "learning_rate": 6.0774611640915316e-05,
      "loss": 0.0004,
      "step": 13462
    },
    {
      "epoch": 6.283780630105017,
      "grad_norm": 0.06494140625,
      "learning_rate": 6.0761114557515455e-05,
      "loss": 0.0003,
      "step": 13463
    },
    {
      "epoch": 6.284247374562427,
      "grad_norm": 0.0654296875,
      "learning_rate": 6.074761831897232e-05,
      "loss": 0.0004,
      "step": 13464
    },
    {
      "epoch": 6.284714119019837,
      "grad_norm": 0.1796875,
      "learning_rate": 6.0734122925576495e-05,
      "loss": 0.0028,
      "step": 13465
    },
    {
      "epoch": 6.285180863477247,
      "grad_norm": 0.404296875,
      "learning_rate": 6.072062837761858e-05,
      "loss": 0.0032,
      "step": 13466
    },
    {
      "epoch": 6.2856476079346555,
      "grad_norm": 0.2373046875,
      "learning_rate": 6.0707134675389134e-05,
      "loss": 0.0009,
      "step": 13467
    },
    {
      "epoch": 6.286114352392065,
      "grad_norm": 0.408203125,
      "learning_rate": 6.069364181917866e-05,
      "loss": 0.0062,
      "step": 13468
    },
    {
      "epoch": 6.286581096849475,
      "grad_norm": 0.0908203125,
      "learning_rate": 6.0680149809277676e-05,
      "loss": 0.0004,
      "step": 13469
    },
    {
      "epoch": 6.287047841306885,
      "grad_norm": 0.390625,
      "learning_rate": 6.0666658645976706e-05,
      "loss": 0.0034,
      "step": 13470
    },
    {
      "epoch": 6.287514585764294,
      "grad_norm": 0.263671875,
      "learning_rate": 6.0653168329566225e-05,
      "loss": 0.0016,
      "step": 13471
    },
    {
      "epoch": 6.287981330221704,
      "grad_norm": 0.373046875,
      "learning_rate": 6.063967886033666e-05,
      "loss": 0.0023,
      "step": 13472
    },
    {
      "epoch": 6.288448074679113,
      "grad_norm": 0.0859375,
      "learning_rate": 6.06261902385785e-05,
      "loss": 0.0006,
      "step": 13473
    },
    {
      "epoch": 6.288914819136522,
      "grad_norm": 0.1982421875,
      "learning_rate": 6.061270246458215e-05,
      "loss": 0.0026,
      "step": 13474
    },
    {
      "epoch": 6.289381563593932,
      "grad_norm": 0.3984375,
      "learning_rate": 6.059921553863799e-05,
      "loss": 0.004,
      "step": 13475
    },
    {
      "epoch": 6.289848308051342,
      "grad_norm": 0.15625,
      "learning_rate": 6.0585729461036466e-05,
      "loss": 0.0007,
      "step": 13476
    },
    {
      "epoch": 6.290315052508752,
      "grad_norm": 0.043212890625,
      "learning_rate": 6.057224423206791e-05,
      "loss": 0.0007,
      "step": 13477
    },
    {
      "epoch": 6.2907817969661615,
      "grad_norm": 0.05615234375,
      "learning_rate": 6.05587598520227e-05,
      "loss": 0.0024,
      "step": 13478
    },
    {
      "epoch": 6.29124854142357,
      "grad_norm": 0.0712890625,
      "learning_rate": 6.054527632119111e-05,
      "loss": 0.0041,
      "step": 13479
    },
    {
      "epoch": 6.29171528588098,
      "grad_norm": 0.055908203125,
      "learning_rate": 6.053179363986352e-05,
      "loss": 0.0004,
      "step": 13480
    },
    {
      "epoch": 6.29218203033839,
      "grad_norm": 0.0791015625,
      "learning_rate": 6.0518311808330206e-05,
      "loss": 0.0005,
      "step": 13481
    },
    {
      "epoch": 6.292648774795799,
      "grad_norm": 0.07177734375,
      "learning_rate": 6.050483082688143e-05,
      "loss": 0.0005,
      "step": 13482
    },
    {
      "epoch": 6.293115519253209,
      "grad_norm": 0.058349609375,
      "learning_rate": 6.049135069580749e-05,
      "loss": 0.0005,
      "step": 13483
    },
    {
      "epoch": 6.2935822637106185,
      "grad_norm": 0.1728515625,
      "learning_rate": 6.0477871415398605e-05,
      "loss": 0.0033,
      "step": 13484
    },
    {
      "epoch": 6.294049008168028,
      "grad_norm": 0.0262451171875,
      "learning_rate": 6.046439298594496e-05,
      "loss": 0.0003,
      "step": 13485
    },
    {
      "epoch": 6.294515752625437,
      "grad_norm": 0.1513671875,
      "learning_rate": 6.045091540773684e-05,
      "loss": 0.0014,
      "step": 13486
    },
    {
      "epoch": 6.294982497082847,
      "grad_norm": 0.2451171875,
      "learning_rate": 6.043743868106437e-05,
      "loss": 0.0007,
      "step": 13487
    },
    {
      "epoch": 6.295449241540257,
      "grad_norm": 0.0255126953125,
      "learning_rate": 6.042396280621776e-05,
      "loss": 0.0004,
      "step": 13488
    },
    {
      "epoch": 6.2959159859976666,
      "grad_norm": 0.1044921875,
      "learning_rate": 6.0410487783487124e-05,
      "loss": 0.0052,
      "step": 13489
    },
    {
      "epoch": 6.2963827304550755,
      "grad_norm": 0.0703125,
      "learning_rate": 6.039701361316261e-05,
      "loss": 0.0006,
      "step": 13490
    },
    {
      "epoch": 6.296849474912485,
      "grad_norm": 0.080078125,
      "learning_rate": 6.038354029553436e-05,
      "loss": 0.0005,
      "step": 13491
    },
    {
      "epoch": 6.297316219369895,
      "grad_norm": 0.1318359375,
      "learning_rate": 6.0370067830892405e-05,
      "loss": 0.0006,
      "step": 13492
    },
    {
      "epoch": 6.297782963827305,
      "grad_norm": 0.07763671875,
      "learning_rate": 6.0356596219526884e-05,
      "loss": 0.0038,
      "step": 13493
    },
    {
      "epoch": 6.298249708284714,
      "grad_norm": 0.1767578125,
      "learning_rate": 6.034312546172782e-05,
      "loss": 0.0069,
      "step": 13494
    },
    {
      "epoch": 6.2987164527421236,
      "grad_norm": 0.07666015625,
      "learning_rate": 6.0329655557785293e-05,
      "loss": 0.0003,
      "step": 13495
    },
    {
      "epoch": 6.299183197199533,
      "grad_norm": 0.328125,
      "learning_rate": 6.0316186507989245e-05,
      "loss": 0.002,
      "step": 13496
    },
    {
      "epoch": 6.299649941656943,
      "grad_norm": 0.1611328125,
      "learning_rate": 6.030271831262977e-05,
      "loss": 0.0007,
      "step": 13497
    },
    {
      "epoch": 6.300116686114352,
      "grad_norm": 0.0301513671875,
      "learning_rate": 6.0289250971996825e-05,
      "loss": 0.0004,
      "step": 13498
    },
    {
      "epoch": 6.300583430571762,
      "grad_norm": 0.03662109375,
      "learning_rate": 6.027578448638032e-05,
      "loss": 0.0004,
      "step": 13499
    },
    {
      "epoch": 6.301050175029172,
      "grad_norm": 0.06005859375,
      "learning_rate": 6.026231885607029e-05,
      "loss": 0.0004,
      "step": 13500
    },
    {
      "epoch": 6.301516919486581,
      "grad_norm": 0.1669921875,
      "learning_rate": 6.024885408135661e-05,
      "loss": 0.0008,
      "step": 13501
    },
    {
      "epoch": 6.30198366394399,
      "grad_norm": 0.0576171875,
      "learning_rate": 6.023539016252919e-05,
      "loss": 0.0005,
      "step": 13502
    },
    {
      "epoch": 6.3024504084014,
      "grad_norm": 0.1630859375,
      "learning_rate": 6.022192709987797e-05,
      "loss": 0.0012,
      "step": 13503
    },
    {
      "epoch": 6.30291715285881,
      "grad_norm": 0.0751953125,
      "learning_rate": 6.0208464893692787e-05,
      "loss": 0.0005,
      "step": 13504
    },
    {
      "epoch": 6.30338389731622,
      "grad_norm": 0.1728515625,
      "learning_rate": 6.0195003544263504e-05,
      "loss": 0.0008,
      "step": 13505
    },
    {
      "epoch": 6.303850641773629,
      "grad_norm": 0.74609375,
      "learning_rate": 6.018154305187993e-05,
      "loss": 0.003,
      "step": 13506
    },
    {
      "epoch": 6.304317386231038,
      "grad_norm": 0.040771484375,
      "learning_rate": 6.016808341683196e-05,
      "loss": 0.0005,
      "step": 13507
    },
    {
      "epoch": 6.304784130688448,
      "grad_norm": 0.36328125,
      "learning_rate": 6.0154624639409316e-05,
      "loss": 0.0013,
      "step": 13508
    },
    {
      "epoch": 6.305250875145858,
      "grad_norm": 0.07958984375,
      "learning_rate": 6.0141166719901806e-05,
      "loss": 0.0005,
      "step": 13509
    },
    {
      "epoch": 6.305717619603267,
      "grad_norm": 0.07177734375,
      "learning_rate": 6.012770965859923e-05,
      "loss": 0.0005,
      "step": 13510
    },
    {
      "epoch": 6.306184364060677,
      "grad_norm": 0.10009765625,
      "learning_rate": 6.0114253455791315e-05,
      "loss": 0.003,
      "step": 13511
    },
    {
      "epoch": 6.3066511085180865,
      "grad_norm": 0.03955078125,
      "learning_rate": 6.0100798111767766e-05,
      "loss": 0.0003,
      "step": 13512
    },
    {
      "epoch": 6.307117852975496,
      "grad_norm": 0.0498046875,
      "learning_rate": 6.0087343626818293e-05,
      "loss": 0.0003,
      "step": 13513
    },
    {
      "epoch": 6.307584597432905,
      "grad_norm": 0.06689453125,
      "learning_rate": 6.007389000123258e-05,
      "loss": 0.0003,
      "step": 13514
    },
    {
      "epoch": 6.308051341890315,
      "grad_norm": 0.1943359375,
      "learning_rate": 6.0060437235300325e-05,
      "loss": 0.0007,
      "step": 13515
    },
    {
      "epoch": 6.308518086347725,
      "grad_norm": 0.06787109375,
      "learning_rate": 6.0046985329311165e-05,
      "loss": 0.0004,
      "step": 13516
    },
    {
      "epoch": 6.308984830805134,
      "grad_norm": 0.0537109375,
      "learning_rate": 6.003353428355476e-05,
      "loss": 0.0003,
      "step": 13517
    },
    {
      "epoch": 6.3094515752625435,
      "grad_norm": 0.31640625,
      "learning_rate": 6.002008409832067e-05,
      "loss": 0.003,
      "step": 13518
    },
    {
      "epoch": 6.309918319719953,
      "grad_norm": 0.328125,
      "learning_rate": 6.000663477389856e-05,
      "loss": 0.0016,
      "step": 13519
    },
    {
      "epoch": 6.310385064177363,
      "grad_norm": 0.033935546875,
      "learning_rate": 5.9993186310577974e-05,
      "loss": 0.0004,
      "step": 13520
    },
    {
      "epoch": 6.310851808634773,
      "grad_norm": 0.1416015625,
      "learning_rate": 5.997973870864846e-05,
      "loss": 0.0004,
      "step": 13521
    },
    {
      "epoch": 6.311318553092182,
      "grad_norm": 0.0556640625,
      "learning_rate": 5.996629196839958e-05,
      "loss": 0.0006,
      "step": 13522
    },
    {
      "epoch": 6.311785297549592,
      "grad_norm": 0.173828125,
      "learning_rate": 5.995284609012086e-05,
      "loss": 0.0038,
      "step": 13523
    },
    {
      "epoch": 6.312252042007001,
      "grad_norm": 0.055908203125,
      "learning_rate": 5.9939401074101776e-05,
      "loss": 0.0041,
      "step": 13524
    },
    {
      "epoch": 6.31271878646441,
      "grad_norm": 0.054931640625,
      "learning_rate": 5.992595692063185e-05,
      "loss": 0.0003,
      "step": 13525
    },
    {
      "epoch": 6.31318553092182,
      "grad_norm": 0.047119140625,
      "learning_rate": 5.991251363000054e-05,
      "loss": 0.0004,
      "step": 13526
    },
    {
      "epoch": 6.31365227537923,
      "grad_norm": 0.0546875,
      "learning_rate": 5.989907120249729e-05,
      "loss": 0.0004,
      "step": 13527
    },
    {
      "epoch": 6.31411901983664,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.988562963841151e-05,
      "loss": 0.0009,
      "step": 13528
    },
    {
      "epoch": 6.314585764294049,
      "grad_norm": 0.0264892578125,
      "learning_rate": 5.9872188938032644e-05,
      "loss": 0.0002,
      "step": 13529
    },
    {
      "epoch": 6.315052508751458,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.985874910165008e-05,
      "loss": 0.0008,
      "step": 13530
    },
    {
      "epoch": 6.315519253208868,
      "grad_norm": 0.031494140625,
      "learning_rate": 5.984531012955317e-05,
      "loss": 0.0005,
      "step": 13531
    },
    {
      "epoch": 6.315985997666278,
      "grad_norm": 0.11865234375,
      "learning_rate": 5.98318720220313e-05,
      "loss": 0.0008,
      "step": 13532
    },
    {
      "epoch": 6.316452742123687,
      "grad_norm": 0.055908203125,
      "learning_rate": 5.981843477937379e-05,
      "loss": 0.0003,
      "step": 13533
    },
    {
      "epoch": 6.316919486581097,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.980499840186995e-05,
      "loss": 0.0005,
      "step": 13534
    },
    {
      "epoch": 6.3173862310385065,
      "grad_norm": 0.03173828125,
      "learning_rate": 5.9791562889809095e-05,
      "loss": 0.0003,
      "step": 13535
    },
    {
      "epoch": 6.317852975495916,
      "grad_norm": 0.392578125,
      "learning_rate": 5.9778128243480515e-05,
      "loss": 0.001,
      "step": 13536
    },
    {
      "epoch": 6.318319719953325,
      "grad_norm": 0.08251953125,
      "learning_rate": 5.9764694463173456e-05,
      "loss": 0.0059,
      "step": 13537
    },
    {
      "epoch": 6.318786464410735,
      "grad_norm": 0.08544921875,
      "learning_rate": 5.975126154917715e-05,
      "loss": 0.0003,
      "step": 13538
    },
    {
      "epoch": 6.319253208868145,
      "grad_norm": 0.314453125,
      "learning_rate": 5.973782950178086e-05,
      "loss": 0.0009,
      "step": 13539
    },
    {
      "epoch": 6.319719953325555,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.9724398321273764e-05,
      "loss": 0.0024,
      "step": 13540
    },
    {
      "epoch": 6.3201866977829635,
      "grad_norm": 0.2294921875,
      "learning_rate": 5.971096800794506e-05,
      "loss": 0.0012,
      "step": 13541
    },
    {
      "epoch": 6.320653442240373,
      "grad_norm": 0.06689453125,
      "learning_rate": 5.969753856208392e-05,
      "loss": 0.0032,
      "step": 13542
    },
    {
      "epoch": 6.321120186697783,
      "grad_norm": 0.0289306640625,
      "learning_rate": 5.9684109983979484e-05,
      "loss": 0.0006,
      "step": 13543
    },
    {
      "epoch": 6.321586931155193,
      "grad_norm": 0.06396484375,
      "learning_rate": 5.9670682273920896e-05,
      "loss": 0.0042,
      "step": 13544
    },
    {
      "epoch": 6.322053675612602,
      "grad_norm": 0.2197265625,
      "learning_rate": 5.9657255432197266e-05,
      "loss": 0.001,
      "step": 13545
    },
    {
      "epoch": 6.322520420070012,
      "grad_norm": 0.04833984375,
      "learning_rate": 5.964382945909769e-05,
      "loss": 0.0004,
      "step": 13546
    },
    {
      "epoch": 6.322987164527421,
      "grad_norm": 0.041015625,
      "learning_rate": 5.963040435491125e-05,
      "loss": 0.0002,
      "step": 13547
    },
    {
      "epoch": 6.323453908984831,
      "grad_norm": 0.0859375,
      "learning_rate": 5.961698011992696e-05,
      "loss": 0.0005,
      "step": 13548
    },
    {
      "epoch": 6.32392065344224,
      "grad_norm": 0.10888671875,
      "learning_rate": 5.960355675443392e-05,
      "loss": 0.001,
      "step": 13549
    },
    {
      "epoch": 6.32438739789965,
      "grad_norm": 0.130859375,
      "learning_rate": 5.959013425872113e-05,
      "loss": 0.0027,
      "step": 13550
    },
    {
      "epoch": 6.32485414235706,
      "grad_norm": 0.2431640625,
      "learning_rate": 5.9576712633077556e-05,
      "loss": 0.001,
      "step": 13551
    },
    {
      "epoch": 6.3253208868144695,
      "grad_norm": 0.07080078125,
      "learning_rate": 5.956329187779224e-05,
      "loss": 0.0004,
      "step": 13552
    },
    {
      "epoch": 6.325787631271878,
      "grad_norm": 0.0849609375,
      "learning_rate": 5.954987199315412e-05,
      "loss": 0.0004,
      "step": 13553
    },
    {
      "epoch": 6.326254375729288,
      "grad_norm": 0.283203125,
      "learning_rate": 5.953645297945212e-05,
      "loss": 0.0045,
      "step": 13554
    },
    {
      "epoch": 6.326721120186698,
      "grad_norm": 0.039306640625,
      "learning_rate": 5.952303483697517e-05,
      "loss": 0.0003,
      "step": 13555
    },
    {
      "epoch": 6.327187864644108,
      "grad_norm": 0.1162109375,
      "learning_rate": 5.95096175660122e-05,
      "loss": 0.0005,
      "step": 13556
    },
    {
      "epoch": 6.327654609101517,
      "grad_norm": 0.05517578125,
      "learning_rate": 5.949620116685209e-05,
      "loss": 0.0006,
      "step": 13557
    },
    {
      "epoch": 6.3281213535589265,
      "grad_norm": 0.10595703125,
      "learning_rate": 5.94827856397837e-05,
      "loss": 0.0004,
      "step": 13558
    },
    {
      "epoch": 6.328588098016336,
      "grad_norm": 0.06787109375,
      "learning_rate": 5.946937098509595e-05,
      "loss": 0.0004,
      "step": 13559
    },
    {
      "epoch": 6.329054842473745,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.945595720307755e-05,
      "loss": 0.0008,
      "step": 13560
    },
    {
      "epoch": 6.329521586931155,
      "grad_norm": 0.103515625,
      "learning_rate": 5.944254429401739e-05,
      "loss": 0.0037,
      "step": 13561
    },
    {
      "epoch": 6.329988331388565,
      "grad_norm": 0.353515625,
      "learning_rate": 5.9429132258204256e-05,
      "loss": 0.0057,
      "step": 13562
    },
    {
      "epoch": 6.3304550758459746,
      "grad_norm": 0.076171875,
      "learning_rate": 5.941572109592689e-05,
      "loss": 0.0005,
      "step": 13563
    },
    {
      "epoch": 6.330921820303384,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.94023108074741e-05,
      "loss": 0.0005,
      "step": 13564
    },
    {
      "epoch": 6.331388564760793,
      "grad_norm": 0.037841796875,
      "learning_rate": 5.938890139313461e-05,
      "loss": 0.0004,
      "step": 13565
    },
    {
      "epoch": 6.331855309218203,
      "grad_norm": 0.08447265625,
      "learning_rate": 5.9375492853197124e-05,
      "loss": 0.0006,
      "step": 13566
    },
    {
      "epoch": 6.332322053675613,
      "grad_norm": 0.0693359375,
      "learning_rate": 5.936208518795035e-05,
      "loss": 0.0004,
      "step": 13567
    },
    {
      "epoch": 6.332788798133022,
      "grad_norm": 0.053466796875,
      "learning_rate": 5.934867839768297e-05,
      "loss": 0.0004,
      "step": 13568
    },
    {
      "epoch": 6.3332555425904316,
      "grad_norm": 0.052001953125,
      "learning_rate": 5.933527248268365e-05,
      "loss": 0.0003,
      "step": 13569
    },
    {
      "epoch": 6.333722287047841,
      "grad_norm": 0.027587890625,
      "learning_rate": 5.932186744324102e-05,
      "loss": 0.0003,
      "step": 13570
    },
    {
      "epoch": 6.334189031505251,
      "grad_norm": 0.07421875,
      "learning_rate": 5.930846327964372e-05,
      "loss": 0.0005,
      "step": 13571
    },
    {
      "epoch": 6.33465577596266,
      "grad_norm": 0.0203857421875,
      "learning_rate": 5.9295059992180366e-05,
      "loss": 0.0003,
      "step": 13572
    },
    {
      "epoch": 6.33512252042007,
      "grad_norm": 0.08349609375,
      "learning_rate": 5.9281657581139514e-05,
      "loss": 0.0004,
      "step": 13573
    },
    {
      "epoch": 6.33558926487748,
      "grad_norm": 0.1494140625,
      "learning_rate": 5.926825604680977e-05,
      "loss": 0.0009,
      "step": 13574
    },
    {
      "epoch": 6.336056009334889,
      "grad_norm": 0.25390625,
      "learning_rate": 5.925485538947967e-05,
      "loss": 0.001,
      "step": 13575
    },
    {
      "epoch": 6.336522753792298,
      "grad_norm": 0.032958984375,
      "learning_rate": 5.9241455609437734e-05,
      "loss": 0.0003,
      "step": 13576
    },
    {
      "epoch": 6.336989498249708,
      "grad_norm": 0.0234375,
      "learning_rate": 5.922805670697247e-05,
      "loss": 0.0002,
      "step": 13577
    },
    {
      "epoch": 6.337456242707118,
      "grad_norm": 0.05224609375,
      "learning_rate": 5.921465868237239e-05,
      "loss": 0.0032,
      "step": 13578
    },
    {
      "epoch": 6.337922987164528,
      "grad_norm": 0.1259765625,
      "learning_rate": 5.920126153592598e-05,
      "loss": 0.0021,
      "step": 13579
    },
    {
      "epoch": 6.338389731621937,
      "grad_norm": 0.0211181640625,
      "learning_rate": 5.918786526792164e-05,
      "loss": 0.0003,
      "step": 13580
    },
    {
      "epoch": 6.338856476079346,
      "grad_norm": 0.1435546875,
      "learning_rate": 5.917446987864785e-05,
      "loss": 0.0005,
      "step": 13581
    },
    {
      "epoch": 6.339323220536756,
      "grad_norm": 0.1796875,
      "learning_rate": 5.916107536839305e-05,
      "loss": 0.0011,
      "step": 13582
    },
    {
      "epoch": 6.339789964994166,
      "grad_norm": 0.0135498046875,
      "learning_rate": 5.914768173744557e-05,
      "loss": 0.0002,
      "step": 13583
    },
    {
      "epoch": 6.340256709451575,
      "grad_norm": 0.055419921875,
      "learning_rate": 5.913428898609386e-05,
      "loss": 0.0004,
      "step": 13584
    },
    {
      "epoch": 6.340723453908985,
      "grad_norm": 0.028564453125,
      "learning_rate": 5.912089711462624e-05,
      "loss": 0.0003,
      "step": 13585
    },
    {
      "epoch": 6.3411901983663945,
      "grad_norm": 0.216796875,
      "learning_rate": 5.9107506123331066e-05,
      "loss": 0.0005,
      "step": 13586
    },
    {
      "epoch": 6.341656942823804,
      "grad_norm": 0.036865234375,
      "learning_rate": 5.9094116012496634e-05,
      "loss": 0.0024,
      "step": 13587
    },
    {
      "epoch": 6.342123687281213,
      "grad_norm": 0.0263671875,
      "learning_rate": 5.908072678241129e-05,
      "loss": 0.0003,
      "step": 13588
    },
    {
      "epoch": 6.342590431738623,
      "grad_norm": 0.0277099609375,
      "learning_rate": 5.906733843336329e-05,
      "loss": 0.0003,
      "step": 13589
    },
    {
      "epoch": 6.343057176196033,
      "grad_norm": 0.052490234375,
      "learning_rate": 5.90539509656409e-05,
      "loss": 0.0004,
      "step": 13590
    },
    {
      "epoch": 6.343523920653443,
      "grad_norm": 0.126953125,
      "learning_rate": 5.9040564379532404e-05,
      "loss": 0.0023,
      "step": 13591
    },
    {
      "epoch": 6.3439906651108515,
      "grad_norm": 0.021484375,
      "learning_rate": 5.902717867532599e-05,
      "loss": 0.0003,
      "step": 13592
    },
    {
      "epoch": 6.344457409568261,
      "grad_norm": 0.027099609375,
      "learning_rate": 5.9013793853309876e-05,
      "loss": 0.0003,
      "step": 13593
    },
    {
      "epoch": 6.344924154025671,
      "grad_norm": 0.059814453125,
      "learning_rate": 5.900040991377225e-05,
      "loss": 0.0005,
      "step": 13594
    },
    {
      "epoch": 6.345390898483081,
      "grad_norm": 0.056396484375,
      "learning_rate": 5.8987026857001306e-05,
      "loss": 0.0004,
      "step": 13595
    },
    {
      "epoch": 6.34585764294049,
      "grad_norm": 0.0311279296875,
      "learning_rate": 5.897364468328517e-05,
      "loss": 0.0003,
      "step": 13596
    },
    {
      "epoch": 6.3463243873979,
      "grad_norm": 0.046142578125,
      "learning_rate": 5.8960263392911964e-05,
      "loss": 0.0004,
      "step": 13597
    },
    {
      "epoch": 6.346791131855309,
      "grad_norm": 0.0869140625,
      "learning_rate": 5.8946882986169846e-05,
      "loss": 0.0005,
      "step": 13598
    },
    {
      "epoch": 6.347257876312719,
      "grad_norm": 0.055908203125,
      "learning_rate": 5.8933503463346886e-05,
      "loss": 0.0037,
      "step": 13599
    },
    {
      "epoch": 6.347724620770128,
      "grad_norm": 0.06494140625,
      "learning_rate": 5.8920124824731136e-05,
      "loss": 0.0031,
      "step": 13600
    },
    {
      "epoch": 6.348191365227538,
      "grad_norm": 0.015869140625,
      "learning_rate": 5.8906747070610704e-05,
      "loss": 0.0002,
      "step": 13601
    },
    {
      "epoch": 6.348658109684948,
      "grad_norm": 0.0810546875,
      "learning_rate": 5.889337020127359e-05,
      "loss": 0.0003,
      "step": 13602
    },
    {
      "epoch": 6.349124854142357,
      "grad_norm": 0.039306640625,
      "learning_rate": 5.887999421700782e-05,
      "loss": 0.0003,
      "step": 13603
    },
    {
      "epoch": 6.349591598599766,
      "grad_norm": 0.1650390625,
      "learning_rate": 5.8866619118101406e-05,
      "loss": 0.0037,
      "step": 13604
    },
    {
      "epoch": 6.350058343057176,
      "grad_norm": 0.234375,
      "learning_rate": 5.885324490484232e-05,
      "loss": 0.0035,
      "step": 13605
    },
    {
      "epoch": 6.350525087514586,
      "grad_norm": 0.3125,
      "learning_rate": 5.8839871577518535e-05,
      "loss": 0.0018,
      "step": 13606
    },
    {
      "epoch": 6.350991831971996,
      "grad_norm": 0.10693359375,
      "learning_rate": 5.8826499136418e-05,
      "loss": 0.0036,
      "step": 13607
    },
    {
      "epoch": 6.351458576429405,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.88131275818286e-05,
      "loss": 0.001,
      "step": 13608
    },
    {
      "epoch": 6.3519253208868145,
      "grad_norm": 0.04443359375,
      "learning_rate": 5.8799756914038226e-05,
      "loss": 0.0003,
      "step": 13609
    },
    {
      "epoch": 6.352392065344224,
      "grad_norm": 0.443359375,
      "learning_rate": 5.878638713333484e-05,
      "loss": 0.0133,
      "step": 13610
    },
    {
      "epoch": 6.352858809801633,
      "grad_norm": 0.1640625,
      "learning_rate": 5.877301824000625e-05,
      "loss": 0.0005,
      "step": 13611
    },
    {
      "epoch": 6.353325554259043,
      "grad_norm": 0.125,
      "learning_rate": 5.8759650234340314e-05,
      "loss": 0.0005,
      "step": 13612
    },
    {
      "epoch": 6.353792298716453,
      "grad_norm": 0.208984375,
      "learning_rate": 5.874628311662488e-05,
      "loss": 0.0037,
      "step": 13613
    },
    {
      "epoch": 6.354259043173863,
      "grad_norm": 0.01312255859375,
      "learning_rate": 5.8732916887147735e-05,
      "loss": 0.0002,
      "step": 13614
    },
    {
      "epoch": 6.3547257876312715,
      "grad_norm": 0.04296875,
      "learning_rate": 5.871955154619666e-05,
      "loss": 0.0004,
      "step": 13615
    },
    {
      "epoch": 6.355192532088681,
      "grad_norm": 0.146484375,
      "learning_rate": 5.870618709405946e-05,
      "loss": 0.0005,
      "step": 13616
    },
    {
      "epoch": 6.355659276546091,
      "grad_norm": 0.17578125,
      "learning_rate": 5.869282353102388e-05,
      "loss": 0.0009,
      "step": 13617
    },
    {
      "epoch": 6.356126021003501,
      "grad_norm": 0.045166015625,
      "learning_rate": 5.8679460857377635e-05,
      "loss": 0.0023,
      "step": 13618
    },
    {
      "epoch": 6.35659276546091,
      "grad_norm": 0.044921875,
      "learning_rate": 5.866609907340841e-05,
      "loss": 0.0035,
      "step": 13619
    },
    {
      "epoch": 6.35705950991832,
      "grad_norm": 0.0517578125,
      "learning_rate": 5.8652738179403976e-05,
      "loss": 0.0004,
      "step": 13620
    },
    {
      "epoch": 6.357526254375729,
      "grad_norm": 0.490234375,
      "learning_rate": 5.8639378175651946e-05,
      "loss": 0.0018,
      "step": 13621
    },
    {
      "epoch": 6.357992998833139,
      "grad_norm": 0.062255859375,
      "learning_rate": 5.8626019062439985e-05,
      "loss": 0.0006,
      "step": 13622
    },
    {
      "epoch": 6.358459743290548,
      "grad_norm": 0.1201171875,
      "learning_rate": 5.8612660840055746e-05,
      "loss": 0.0005,
      "step": 13623
    },
    {
      "epoch": 6.358926487747958,
      "grad_norm": 0.0390625,
      "learning_rate": 5.859930350878684e-05,
      "loss": 0.0004,
      "step": 13624
    },
    {
      "epoch": 6.359393232205368,
      "grad_norm": 0.10888671875,
      "learning_rate": 5.858594706892089e-05,
      "loss": 0.0047,
      "step": 13625
    },
    {
      "epoch": 6.3598599766627775,
      "grad_norm": 0.11328125,
      "learning_rate": 5.8572591520745405e-05,
      "loss": 0.0005,
      "step": 13626
    },
    {
      "epoch": 6.360326721120186,
      "grad_norm": 0.193359375,
      "learning_rate": 5.855923686454801e-05,
      "loss": 0.0006,
      "step": 13627
    },
    {
      "epoch": 6.360793465577596,
      "grad_norm": 0.1357421875,
      "learning_rate": 5.854588310061623e-05,
      "loss": 0.0006,
      "step": 13628
    },
    {
      "epoch": 6.361260210035006,
      "grad_norm": 0.22265625,
      "learning_rate": 5.853253022923756e-05,
      "loss": 0.0006,
      "step": 13629
    },
    {
      "epoch": 6.361726954492416,
      "grad_norm": 0.1337890625,
      "learning_rate": 5.851917825069955e-05,
      "loss": 0.0007,
      "step": 13630
    },
    {
      "epoch": 6.362193698949825,
      "grad_norm": 0.32421875,
      "learning_rate": 5.850582716528964e-05,
      "loss": 0.0041,
      "step": 13631
    },
    {
      "epoch": 6.3626604434072345,
      "grad_norm": 0.06494140625,
      "learning_rate": 5.8492476973295294e-05,
      "loss": 0.0004,
      "step": 13632
    },
    {
      "epoch": 6.363127187864644,
      "grad_norm": 0.043212890625,
      "learning_rate": 5.8479127675004006e-05,
      "loss": 0.0003,
      "step": 13633
    },
    {
      "epoch": 6.363593932322054,
      "grad_norm": 0.376953125,
      "learning_rate": 5.846577927070316e-05,
      "loss": 0.0041,
      "step": 13634
    },
    {
      "epoch": 6.364060676779463,
      "grad_norm": 0.1103515625,
      "learning_rate": 5.845243176068016e-05,
      "loss": 0.0005,
      "step": 13635
    },
    {
      "epoch": 6.364527421236873,
      "grad_norm": 0.0159912109375,
      "learning_rate": 5.8439085145222396e-05,
      "loss": 0.0003,
      "step": 13636
    },
    {
      "epoch": 6.364994165694283,
      "grad_norm": 0.032958984375,
      "learning_rate": 5.8425739424617256e-05,
      "loss": 0.0003,
      "step": 13637
    },
    {
      "epoch": 6.365460910151692,
      "grad_norm": 0.06005859375,
      "learning_rate": 5.841239459915206e-05,
      "loss": 0.0003,
      "step": 13638
    },
    {
      "epoch": 6.365927654609101,
      "grad_norm": 0.119140625,
      "learning_rate": 5.8399050669114156e-05,
      "loss": 0.004,
      "step": 13639
    },
    {
      "epoch": 6.366394399066511,
      "grad_norm": 0.07275390625,
      "learning_rate": 5.8385707634790856e-05,
      "loss": 0.0004,
      "step": 13640
    },
    {
      "epoch": 6.366861143523921,
      "grad_norm": 0.08984375,
      "learning_rate": 5.837236549646943e-05,
      "loss": 0.0005,
      "step": 13641
    },
    {
      "epoch": 6.367327887981331,
      "grad_norm": 0.055908203125,
      "learning_rate": 5.835902425443718e-05,
      "loss": 0.0003,
      "step": 13642
    },
    {
      "epoch": 6.3677946324387396,
      "grad_norm": 0.146484375,
      "learning_rate": 5.834568390898132e-05,
      "loss": 0.0037,
      "step": 13643
    },
    {
      "epoch": 6.368261376896149,
      "grad_norm": 0.1865234375,
      "learning_rate": 5.833234446038909e-05,
      "loss": 0.0007,
      "step": 13644
    },
    {
      "epoch": 6.368728121353559,
      "grad_norm": 0.048828125,
      "learning_rate": 5.8319005908947765e-05,
      "loss": 0.0003,
      "step": 13645
    },
    {
      "epoch": 6.369194865810968,
      "grad_norm": 0.01153564453125,
      "learning_rate": 5.8305668254944434e-05,
      "loss": 0.0002,
      "step": 13646
    },
    {
      "epoch": 6.369661610268378,
      "grad_norm": 0.1298828125,
      "learning_rate": 5.8292331498666344e-05,
      "loss": 0.0006,
      "step": 13647
    },
    {
      "epoch": 6.370128354725788,
      "grad_norm": 0.040771484375,
      "learning_rate": 5.827899564040066e-05,
      "loss": 0.0003,
      "step": 13648
    },
    {
      "epoch": 6.370595099183197,
      "grad_norm": 0.037109375,
      "learning_rate": 5.826566068043445e-05,
      "loss": 0.0003,
      "step": 13649
    },
    {
      "epoch": 6.371061843640607,
      "grad_norm": 0.1396484375,
      "learning_rate": 5.825232661905492e-05,
      "loss": 0.0008,
      "step": 13650
    },
    {
      "epoch": 6.371528588098016,
      "grad_norm": 0.05126953125,
      "learning_rate": 5.823899345654906e-05,
      "loss": 0.0003,
      "step": 13651
    },
    {
      "epoch": 6.371995332555426,
      "grad_norm": 0.0810546875,
      "learning_rate": 5.822566119320403e-05,
      "loss": 0.0039,
      "step": 13652
    },
    {
      "epoch": 6.372462077012836,
      "grad_norm": 0.0203857421875,
      "learning_rate": 5.821232982930689e-05,
      "loss": 0.0002,
      "step": 13653
    },
    {
      "epoch": 6.372928821470245,
      "grad_norm": 0.205078125,
      "learning_rate": 5.8198999365144615e-05,
      "loss": 0.0008,
      "step": 13654
    },
    {
      "epoch": 6.373395565927654,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.8185669801004306e-05,
      "loss": 0.0009,
      "step": 13655
    },
    {
      "epoch": 6.373862310385064,
      "grad_norm": 0.06640625,
      "learning_rate": 5.817234113717287e-05,
      "loss": 0.0004,
      "step": 13656
    },
    {
      "epoch": 6.374329054842474,
      "grad_norm": 0.279296875,
      "learning_rate": 5.8159013373937384e-05,
      "loss": 0.0007,
      "step": 13657
    },
    {
      "epoch": 6.374795799299883,
      "grad_norm": 0.043701171875,
      "learning_rate": 5.814568651158472e-05,
      "loss": 0.0004,
      "step": 13658
    },
    {
      "epoch": 6.375262543757293,
      "grad_norm": 0.033935546875,
      "learning_rate": 5.813236055040188e-05,
      "loss": 0.0003,
      "step": 13659
    },
    {
      "epoch": 6.3757292882147025,
      "grad_norm": 0.2890625,
      "learning_rate": 5.811903549067579e-05,
      "loss": 0.0008,
      "step": 13660
    },
    {
      "epoch": 6.376196032672112,
      "grad_norm": 0.05859375,
      "learning_rate": 5.810571133269329e-05,
      "loss": 0.0004,
      "step": 13661
    },
    {
      "epoch": 6.376662777129521,
      "grad_norm": 0.0201416015625,
      "learning_rate": 5.809238807674137e-05,
      "loss": 0.0003,
      "step": 13662
    },
    {
      "epoch": 6.377129521586931,
      "grad_norm": 0.451171875,
      "learning_rate": 5.807906572310677e-05,
      "loss": 0.0039,
      "step": 13663
    },
    {
      "epoch": 6.377596266044341,
      "grad_norm": 0.08056640625,
      "learning_rate": 5.806574427207642e-05,
      "loss": 0.0025,
      "step": 13664
    },
    {
      "epoch": 6.378063010501751,
      "grad_norm": 0.1279296875,
      "learning_rate": 5.805242372393714e-05,
      "loss": 0.0005,
      "step": 13665
    },
    {
      "epoch": 6.3785297549591595,
      "grad_norm": 0.1474609375,
      "learning_rate": 5.8039104078975694e-05,
      "loss": 0.0006,
      "step": 13666
    },
    {
      "epoch": 6.378996499416569,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.802578533747889e-05,
      "loss": 0.0008,
      "step": 13667
    },
    {
      "epoch": 6.379463243873979,
      "grad_norm": 0.046142578125,
      "learning_rate": 5.801246749973354e-05,
      "loss": 0.0004,
      "step": 13668
    },
    {
      "epoch": 6.379929988331389,
      "grad_norm": 0.032470703125,
      "learning_rate": 5.7999150566026306e-05,
      "loss": 0.0004,
      "step": 13669
    },
    {
      "epoch": 6.380396732788798,
      "grad_norm": 0.0615234375,
      "learning_rate": 5.798583453664397e-05,
      "loss": 0.0004,
      "step": 13670
    },
    {
      "epoch": 6.380863477246208,
      "grad_norm": 0.03955078125,
      "learning_rate": 5.7972519411873246e-05,
      "loss": 0.0004,
      "step": 13671
    },
    {
      "epoch": 6.381330221703617,
      "grad_norm": 0.140625,
      "learning_rate": 5.79592051920008e-05,
      "loss": 0.0005,
      "step": 13672
    },
    {
      "epoch": 6.381796966161027,
      "grad_norm": 0.095703125,
      "learning_rate": 5.79458918773133e-05,
      "loss": 0.0006,
      "step": 13673
    },
    {
      "epoch": 6.382263710618436,
      "grad_norm": 0.06396484375,
      "learning_rate": 5.793257946809745e-05,
      "loss": 0.0005,
      "step": 13674
    },
    {
      "epoch": 6.382730455075846,
      "grad_norm": 0.03515625,
      "learning_rate": 5.79192679646398e-05,
      "loss": 0.0004,
      "step": 13675
    },
    {
      "epoch": 6.383197199533256,
      "grad_norm": 0.04248046875,
      "learning_rate": 5.7905957367226994e-05,
      "loss": 0.0003,
      "step": 13676
    },
    {
      "epoch": 6.3836639439906655,
      "grad_norm": 0.0220947265625,
      "learning_rate": 5.789264767614566e-05,
      "loss": 0.0003,
      "step": 13677
    },
    {
      "epoch": 6.384130688448074,
      "grad_norm": 0.130859375,
      "learning_rate": 5.787933889168232e-05,
      "loss": 0.001,
      "step": 13678
    },
    {
      "epoch": 6.384597432905484,
      "grad_norm": 0.1640625,
      "learning_rate": 5.786603101412358e-05,
      "loss": 0.0006,
      "step": 13679
    },
    {
      "epoch": 6.385064177362894,
      "grad_norm": 0.765625,
      "learning_rate": 5.78527240437559e-05,
      "loss": 0.0079,
      "step": 13680
    },
    {
      "epoch": 6.385530921820304,
      "grad_norm": 0.1767578125,
      "learning_rate": 5.783941798086584e-05,
      "loss": 0.0043,
      "step": 13681
    },
    {
      "epoch": 6.385997666277713,
      "grad_norm": 0.0157470703125,
      "learning_rate": 5.782611282573992e-05,
      "loss": 0.0002,
      "step": 13682
    },
    {
      "epoch": 6.3864644107351225,
      "grad_norm": 0.06396484375,
      "learning_rate": 5.7812808578664556e-05,
      "loss": 0.0004,
      "step": 13683
    },
    {
      "epoch": 6.386931155192532,
      "grad_norm": 0.08203125,
      "learning_rate": 5.779950523992621e-05,
      "loss": 0.0007,
      "step": 13684
    },
    {
      "epoch": 6.387397899649942,
      "grad_norm": 0.11376953125,
      "learning_rate": 5.77862028098114e-05,
      "loss": 0.0029,
      "step": 13685
    },
    {
      "epoch": 6.387864644107351,
      "grad_norm": 0.283203125,
      "learning_rate": 5.777290128860643e-05,
      "loss": 0.0016,
      "step": 13686
    },
    {
      "epoch": 6.388331388564761,
      "grad_norm": 0.061279296875,
      "learning_rate": 5.775960067659775e-05,
      "loss": 0.0004,
      "step": 13687
    },
    {
      "epoch": 6.388798133022171,
      "grad_norm": 0.0242919921875,
      "learning_rate": 5.774630097407177e-05,
      "loss": 0.0003,
      "step": 13688
    },
    {
      "epoch": 6.3892648774795795,
      "grad_norm": 0.10791015625,
      "learning_rate": 5.773300218131478e-05,
      "loss": 0.0028,
      "step": 13689
    },
    {
      "epoch": 6.389731621936989,
      "grad_norm": 0.0247802734375,
      "learning_rate": 5.771970429861313e-05,
      "loss": 0.0003,
      "step": 13690
    },
    {
      "epoch": 6.390198366394399,
      "grad_norm": 0.052001953125,
      "learning_rate": 5.770640732625322e-05,
      "loss": 0.0024,
      "step": 13691
    },
    {
      "epoch": 6.390665110851809,
      "grad_norm": 0.443359375,
      "learning_rate": 5.7693111264521235e-05,
      "loss": 0.0051,
      "step": 13692
    },
    {
      "epoch": 6.391131855309218,
      "grad_norm": 0.04150390625,
      "learning_rate": 5.76798161137035e-05,
      "loss": 0.0021,
      "step": 13693
    },
    {
      "epoch": 6.391598599766628,
      "grad_norm": 0.01495361328125,
      "learning_rate": 5.766652187408633e-05,
      "loss": 0.0002,
      "step": 13694
    },
    {
      "epoch": 6.392065344224037,
      "grad_norm": 0.302734375,
      "learning_rate": 5.7653228545955874e-05,
      "loss": 0.0027,
      "step": 13695
    },
    {
      "epoch": 6.392532088681447,
      "grad_norm": 0.06494140625,
      "learning_rate": 5.763993612959838e-05,
      "loss": 0.0006,
      "step": 13696
    },
    {
      "epoch": 6.392998833138856,
      "grad_norm": 0.04443359375,
      "learning_rate": 5.7626644625300107e-05,
      "loss": 0.0003,
      "step": 13697
    },
    {
      "epoch": 6.393465577596266,
      "grad_norm": 0.2421875,
      "learning_rate": 5.761335403334715e-05,
      "loss": 0.0005,
      "step": 13698
    },
    {
      "epoch": 6.393932322053676,
      "grad_norm": 0.0213623046875,
      "learning_rate": 5.7600064354025754e-05,
      "loss": 0.0002,
      "step": 13699
    },
    {
      "epoch": 6.3943990665110855,
      "grad_norm": 0.31640625,
      "learning_rate": 5.7586775587621966e-05,
      "loss": 0.0009,
      "step": 13700
    },
    {
      "epoch": 6.394865810968494,
      "grad_norm": 0.06591796875,
      "learning_rate": 5.757348773442197e-05,
      "loss": 0.0004,
      "step": 13701
    },
    {
      "epoch": 6.395332555425904,
      "grad_norm": 0.39453125,
      "learning_rate": 5.756020079471188e-05,
      "loss": 0.008,
      "step": 13702
    },
    {
      "epoch": 6.395799299883314,
      "grad_norm": 0.0203857421875,
      "learning_rate": 5.754691476877776e-05,
      "loss": 0.0003,
      "step": 13703
    },
    {
      "epoch": 6.396266044340724,
      "grad_norm": 0.3046875,
      "learning_rate": 5.753362965690564e-05,
      "loss": 0.0015,
      "step": 13704
    },
    {
      "epoch": 6.396732788798133,
      "grad_norm": 0.029541015625,
      "learning_rate": 5.7520345459381585e-05,
      "loss": 0.0003,
      "step": 13705
    },
    {
      "epoch": 6.3971995332555425,
      "grad_norm": 0.10693359375,
      "learning_rate": 5.7507062176491654e-05,
      "loss": 0.0005,
      "step": 13706
    },
    {
      "epoch": 6.397666277712952,
      "grad_norm": 0.1416015625,
      "learning_rate": 5.74937798085218e-05,
      "loss": 0.0006,
      "step": 13707
    },
    {
      "epoch": 6.398133022170362,
      "grad_norm": 0.294921875,
      "learning_rate": 5.7480498355758006e-05,
      "loss": 0.0075,
      "step": 13708
    },
    {
      "epoch": 6.398599766627771,
      "grad_norm": 0.0859375,
      "learning_rate": 5.746721781848631e-05,
      "loss": 0.0026,
      "step": 13709
    },
    {
      "epoch": 6.399066511085181,
      "grad_norm": 0.1865234375,
      "learning_rate": 5.745393819699256e-05,
      "loss": 0.0008,
      "step": 13710
    },
    {
      "epoch": 6.399533255542591,
      "grad_norm": 0.087890625,
      "learning_rate": 5.744065949156278e-05,
      "loss": 0.0004,
      "step": 13711
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.04443359375,
      "learning_rate": 5.742738170248276e-05,
      "loss": 0.0004,
      "step": 13712
    },
    {
      "epoch": 6.400466744457409,
      "grad_norm": 0.2119140625,
      "learning_rate": 5.7414104830038465e-05,
      "loss": 0.0012,
      "step": 13713
    },
    {
      "epoch": 6.400933488914819,
      "grad_norm": 0.0291748046875,
      "learning_rate": 5.7400828874515776e-05,
      "loss": 0.0003,
      "step": 13714
    },
    {
      "epoch": 6.401400233372229,
      "grad_norm": 0.158203125,
      "learning_rate": 5.738755383620047e-05,
      "loss": 0.0005,
      "step": 13715
    },
    {
      "epoch": 6.401866977829639,
      "grad_norm": 0.0966796875,
      "learning_rate": 5.737427971537841e-05,
      "loss": 0.0004,
      "step": 13716
    },
    {
      "epoch": 6.402333722287048,
      "grad_norm": 0.72265625,
      "learning_rate": 5.736100651233544e-05,
      "loss": 0.0019,
      "step": 13717
    },
    {
      "epoch": 6.402800466744457,
      "grad_norm": 0.0458984375,
      "learning_rate": 5.734773422735726e-05,
      "loss": 0.0003,
      "step": 13718
    },
    {
      "epoch": 6.403267211201867,
      "grad_norm": 0.03564453125,
      "learning_rate": 5.73344628607297e-05,
      "loss": 0.0003,
      "step": 13719
    },
    {
      "epoch": 6.403733955659277,
      "grad_norm": 0.11865234375,
      "learning_rate": 5.732119241273852e-05,
      "loss": 0.0039,
      "step": 13720
    },
    {
      "epoch": 6.404200700116686,
      "grad_norm": 0.0732421875,
      "learning_rate": 5.7307922883669386e-05,
      "loss": 0.0005,
      "step": 13721
    },
    {
      "epoch": 6.404667444574096,
      "grad_norm": 0.07373046875,
      "learning_rate": 5.7294654273808035e-05,
      "loss": 0.0003,
      "step": 13722
    },
    {
      "epoch": 6.405134189031505,
      "grad_norm": 0.032958984375,
      "learning_rate": 5.72813865834402e-05,
      "loss": 0.0003,
      "step": 13723
    },
    {
      "epoch": 6.405600933488914,
      "grad_norm": 0.14453125,
      "learning_rate": 5.7268119812851475e-05,
      "loss": 0.0035,
      "step": 13724
    },
    {
      "epoch": 6.406067677946324,
      "grad_norm": 0.1728515625,
      "learning_rate": 5.7254853962327546e-05,
      "loss": 0.0007,
      "step": 13725
    },
    {
      "epoch": 6.406534422403734,
      "grad_norm": 0.059814453125,
      "learning_rate": 5.724158903215406e-05,
      "loss": 0.0003,
      "step": 13726
    },
    {
      "epoch": 6.407001166861144,
      "grad_norm": 0.2470703125,
      "learning_rate": 5.722832502261659e-05,
      "loss": 0.0019,
      "step": 13727
    },
    {
      "epoch": 6.4074679113185535,
      "grad_norm": 0.49609375,
      "learning_rate": 5.721506193400073e-05,
      "loss": 0.0078,
      "step": 13728
    },
    {
      "epoch": 6.407934655775962,
      "grad_norm": 0.11328125,
      "learning_rate": 5.720179976659208e-05,
      "loss": 0.0041,
      "step": 13729
    },
    {
      "epoch": 6.408401400233372,
      "grad_norm": 0.171875,
      "learning_rate": 5.718853852067614e-05,
      "loss": 0.0008,
      "step": 13730
    },
    {
      "epoch": 6.408868144690782,
      "grad_norm": 0.08447265625,
      "learning_rate": 5.717527819653849e-05,
      "loss": 0.0024,
      "step": 13731
    },
    {
      "epoch": 6.409334889148191,
      "grad_norm": 0.048095703125,
      "learning_rate": 5.71620187944646e-05,
      "loss": 0.0024,
      "step": 13732
    },
    {
      "epoch": 6.409801633605601,
      "grad_norm": 0.22265625,
      "learning_rate": 5.714876031473997e-05,
      "loss": 0.0007,
      "step": 13733
    },
    {
      "epoch": 6.4102683780630105,
      "grad_norm": 0.04541015625,
      "learning_rate": 5.713550275765011e-05,
      "loss": 0.0005,
      "step": 13734
    },
    {
      "epoch": 6.41073512252042,
      "grad_norm": 0.0390625,
      "learning_rate": 5.712224612348038e-05,
      "loss": 0.0021,
      "step": 13735
    },
    {
      "epoch": 6.411201866977829,
      "grad_norm": 0.40234375,
      "learning_rate": 5.7108990412516285e-05,
      "loss": 0.0051,
      "step": 13736
    },
    {
      "epoch": 6.411668611435239,
      "grad_norm": 0.05029296875,
      "learning_rate": 5.709573562504324e-05,
      "loss": 0.0022,
      "step": 13737
    },
    {
      "epoch": 6.412135355892649,
      "grad_norm": 0.298828125,
      "learning_rate": 5.708248176134658e-05,
      "loss": 0.0018,
      "step": 13738
    },
    {
      "epoch": 6.412602100350059,
      "grad_norm": 0.236328125,
      "learning_rate": 5.7069228821711706e-05,
      "loss": 0.0045,
      "step": 13739
    },
    {
      "epoch": 6.4130688448074675,
      "grad_norm": 0.12158203125,
      "learning_rate": 5.7055976806424004e-05,
      "loss": 0.0007,
      "step": 13740
    },
    {
      "epoch": 6.413535589264877,
      "grad_norm": 0.1630859375,
      "learning_rate": 5.704272571576873e-05,
      "loss": 0.0005,
      "step": 13741
    },
    {
      "epoch": 6.414002333722287,
      "grad_norm": 0.466796875,
      "learning_rate": 5.7029475550031244e-05,
      "loss": 0.0031,
      "step": 13742
    },
    {
      "epoch": 6.414469078179697,
      "grad_norm": 0.02294921875,
      "learning_rate": 5.701622630949684e-05,
      "loss": 0.0003,
      "step": 13743
    },
    {
      "epoch": 6.414935822637106,
      "grad_norm": 0.17578125,
      "learning_rate": 5.700297799445076e-05,
      "loss": 0.0038,
      "step": 13744
    },
    {
      "epoch": 6.415402567094516,
      "grad_norm": 0.0595703125,
      "learning_rate": 5.698973060517825e-05,
      "loss": 0.0033,
      "step": 13745
    },
    {
      "epoch": 6.415869311551925,
      "grad_norm": 0.10888671875,
      "learning_rate": 5.6976484141964615e-05,
      "loss": 0.0006,
      "step": 13746
    },
    {
      "epoch": 6.416336056009335,
      "grad_norm": 0.054931640625,
      "learning_rate": 5.696323860509496e-05,
      "loss": 0.0004,
      "step": 13747
    },
    {
      "epoch": 6.416802800466744,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.694999399485459e-05,
      "loss": 0.0005,
      "step": 13748
    },
    {
      "epoch": 6.417269544924154,
      "grad_norm": 0.22265625,
      "learning_rate": 5.693675031152855e-05,
      "loss": 0.0008,
      "step": 13749
    },
    {
      "epoch": 6.417736289381564,
      "grad_norm": 0.07958984375,
      "learning_rate": 5.69235075554021e-05,
      "loss": 0.0005,
      "step": 13750
    },
    {
      "epoch": 6.4182030338389735,
      "grad_norm": 0.2138671875,
      "learning_rate": 5.69102657267603e-05,
      "loss": 0.0012,
      "step": 13751
    },
    {
      "epoch": 6.418669778296382,
      "grad_norm": 0.1357421875,
      "learning_rate": 5.6897024825888315e-05,
      "loss": 0.0006,
      "step": 13752
    },
    {
      "epoch": 6.419136522753792,
      "grad_norm": 0.291015625,
      "learning_rate": 5.6883784853071176e-05,
      "loss": 0.0049,
      "step": 13753
    },
    {
      "epoch": 6.419603267211202,
      "grad_norm": 0.091796875,
      "learning_rate": 5.687054580859399e-05,
      "loss": 0.0007,
      "step": 13754
    },
    {
      "epoch": 6.420070011668612,
      "grad_norm": 0.40234375,
      "learning_rate": 5.6857307692741824e-05,
      "loss": 0.0049,
      "step": 13755
    },
    {
      "epoch": 6.420536756126021,
      "grad_norm": 0.2470703125,
      "learning_rate": 5.684407050579967e-05,
      "loss": 0.0013,
      "step": 13756
    },
    {
      "epoch": 6.4210035005834305,
      "grad_norm": 0.5703125,
      "learning_rate": 5.683083424805255e-05,
      "loss": 0.0116,
      "step": 13757
    },
    {
      "epoch": 6.42147024504084,
      "grad_norm": 0.333984375,
      "learning_rate": 5.681759891978553e-05,
      "loss": 0.001,
      "step": 13758
    },
    {
      "epoch": 6.42193698949825,
      "grad_norm": 0.04736328125,
      "learning_rate": 5.680436452128346e-05,
      "loss": 0.0004,
      "step": 13759
    },
    {
      "epoch": 6.422403733955659,
      "grad_norm": 0.404296875,
      "learning_rate": 5.679113105283138e-05,
      "loss": 0.0014,
      "step": 13760
    },
    {
      "epoch": 6.422870478413069,
      "grad_norm": 0.032470703125,
      "learning_rate": 5.677789851471416e-05,
      "loss": 0.0003,
      "step": 13761
    },
    {
      "epoch": 6.423337222870479,
      "grad_norm": 0.2412109375,
      "learning_rate": 5.6764666907216735e-05,
      "loss": 0.0019,
      "step": 13762
    },
    {
      "epoch": 6.423803967327888,
      "grad_norm": 0.322265625,
      "learning_rate": 5.6751436230624044e-05,
      "loss": 0.0018,
      "step": 13763
    },
    {
      "epoch": 6.424270711785297,
      "grad_norm": 0.51171875,
      "learning_rate": 5.673820648522088e-05,
      "loss": 0.006,
      "step": 13764
    },
    {
      "epoch": 6.424737456242707,
      "grad_norm": 0.345703125,
      "learning_rate": 5.6724977671292124e-05,
      "loss": 0.0045,
      "step": 13765
    },
    {
      "epoch": 6.425204200700117,
      "grad_norm": 0.1845703125,
      "learning_rate": 5.671174978912266e-05,
      "loss": 0.0029,
      "step": 13766
    },
    {
      "epoch": 6.425670945157526,
      "grad_norm": 0.25390625,
      "learning_rate": 5.669852283899721e-05,
      "loss": 0.0084,
      "step": 13767
    },
    {
      "epoch": 6.426137689614936,
      "grad_norm": 0.275390625,
      "learning_rate": 5.6685296821200604e-05,
      "loss": 0.0024,
      "step": 13768
    },
    {
      "epoch": 6.426604434072345,
      "grad_norm": 0.03759765625,
      "learning_rate": 5.667207173601766e-05,
      "loss": 0.0004,
      "step": 13769
    },
    {
      "epoch": 6.427071178529755,
      "grad_norm": 0.5625,
      "learning_rate": 5.665884758373304e-05,
      "loss": 0.0089,
      "step": 13770
    },
    {
      "epoch": 6.427537922987165,
      "grad_norm": 0.1845703125,
      "learning_rate": 5.664562436463152e-05,
      "loss": 0.0024,
      "step": 13771
    },
    {
      "epoch": 6.428004667444574,
      "grad_norm": 0.125,
      "learning_rate": 5.663240207899786e-05,
      "loss": 0.0005,
      "step": 13772
    },
    {
      "epoch": 6.428471411901984,
      "grad_norm": 0.322265625,
      "learning_rate": 5.661918072711665e-05,
      "loss": 0.0035,
      "step": 13773
    },
    {
      "epoch": 6.4289381563593935,
      "grad_norm": 0.51953125,
      "learning_rate": 5.660596030927261e-05,
      "loss": 0.0027,
      "step": 13774
    },
    {
      "epoch": 6.429404900816802,
      "grad_norm": 0.23046875,
      "learning_rate": 5.659274082575043e-05,
      "loss": 0.0031,
      "step": 13775
    },
    {
      "epoch": 6.429871645274212,
      "grad_norm": 0.2890625,
      "learning_rate": 5.6579522276834676e-05,
      "loss": 0.0018,
      "step": 13776
    },
    {
      "epoch": 6.430338389731622,
      "grad_norm": 0.291015625,
      "learning_rate": 5.656630466280997e-05,
      "loss": 0.0039,
      "step": 13777
    },
    {
      "epoch": 6.430805134189032,
      "grad_norm": 0.1689453125,
      "learning_rate": 5.655308798396095e-05,
      "loss": 0.0011,
      "step": 13778
    },
    {
      "epoch": 6.431271878646441,
      "grad_norm": 0.330078125,
      "learning_rate": 5.653987224057212e-05,
      "loss": 0.0027,
      "step": 13779
    },
    {
      "epoch": 6.4317386231038505,
      "grad_norm": 0.400390625,
      "learning_rate": 5.652665743292811e-05,
      "loss": 0.004,
      "step": 13780
    },
    {
      "epoch": 6.43220536756126,
      "grad_norm": 0.208984375,
      "learning_rate": 5.651344356131334e-05,
      "loss": 0.0016,
      "step": 13781
    },
    {
      "epoch": 6.43267211201867,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.65002306260124e-05,
      "loss": 0.0006,
      "step": 13782
    },
    {
      "epoch": 6.433138856476079,
      "grad_norm": 0.2255859375,
      "learning_rate": 5.648701862730979e-05,
      "loss": 0.0027,
      "step": 13783
    },
    {
      "epoch": 6.433605600933489,
      "grad_norm": 0.28515625,
      "learning_rate": 5.64738075654899e-05,
      "loss": 0.0039,
      "step": 13784
    },
    {
      "epoch": 6.434072345390899,
      "grad_norm": 0.16796875,
      "learning_rate": 5.6460597440837225e-05,
      "loss": 0.0045,
      "step": 13785
    },
    {
      "epoch": 6.434539089848308,
      "grad_norm": 0.13671875,
      "learning_rate": 5.644738825363625e-05,
      "loss": 0.0005,
      "step": 13786
    },
    {
      "epoch": 6.435005834305717,
      "grad_norm": 0.09814453125,
      "learning_rate": 5.6434180004171266e-05,
      "loss": 0.0009,
      "step": 13787
    },
    {
      "epoch": 6.435472578763127,
      "grad_norm": 0.49609375,
      "learning_rate": 5.642097269272674e-05,
      "loss": 0.0019,
      "step": 13788
    },
    {
      "epoch": 6.435939323220537,
      "grad_norm": 0.30078125,
      "learning_rate": 5.640776631958706e-05,
      "loss": 0.0047,
      "step": 13789
    },
    {
      "epoch": 6.436406067677947,
      "grad_norm": 0.025146484375,
      "learning_rate": 5.63945608850365e-05,
      "loss": 0.0003,
      "step": 13790
    },
    {
      "epoch": 6.436872812135356,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.638135638935942e-05,
      "loss": 0.0006,
      "step": 13791
    },
    {
      "epoch": 6.437339556592765,
      "grad_norm": 0.09716796875,
      "learning_rate": 5.636815283284018e-05,
      "loss": 0.0011,
      "step": 13792
    },
    {
      "epoch": 6.437806301050175,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.635495021576297e-05,
      "loss": 0.0036,
      "step": 13793
    },
    {
      "epoch": 6.438273045507585,
      "grad_norm": 0.25,
      "learning_rate": 5.634174853841212e-05,
      "loss": 0.0041,
      "step": 13794
    },
    {
      "epoch": 6.438739789964994,
      "grad_norm": 0.20703125,
      "learning_rate": 5.6328547801071886e-05,
      "loss": 0.0025,
      "step": 13795
    },
    {
      "epoch": 6.439206534422404,
      "grad_norm": 0.33984375,
      "learning_rate": 5.6315348004026446e-05,
      "loss": 0.0064,
      "step": 13796
    },
    {
      "epoch": 6.4396732788798134,
      "grad_norm": 0.341796875,
      "learning_rate": 5.630214914756007e-05,
      "loss": 0.0128,
      "step": 13797
    },
    {
      "epoch": 6.440140023337223,
      "grad_norm": 0.263671875,
      "learning_rate": 5.62889512319569e-05,
      "loss": 0.0054,
      "step": 13798
    },
    {
      "epoch": 6.440606767794632,
      "grad_norm": 0.1552734375,
      "learning_rate": 5.627575425750107e-05,
      "loss": 0.0037,
      "step": 13799
    },
    {
      "epoch": 6.441073512252042,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.6262558224476766e-05,
      "loss": 0.0008,
      "step": 13800
    },
    {
      "epoch": 6.441540256709452,
      "grad_norm": 0.1494140625,
      "learning_rate": 5.624936313316815e-05,
      "loss": 0.001,
      "step": 13801
    },
    {
      "epoch": 6.4420070011668615,
      "grad_norm": 0.380859375,
      "learning_rate": 5.623616898385926e-05,
      "loss": 0.006,
      "step": 13802
    },
    {
      "epoch": 6.44247374562427,
      "grad_norm": 0.03271484375,
      "learning_rate": 5.6222975776834196e-05,
      "loss": 0.0004,
      "step": 13803
    },
    {
      "epoch": 6.44294049008168,
      "grad_norm": 0.09130859375,
      "learning_rate": 5.620978351237708e-05,
      "loss": 0.0034,
      "step": 13804
    },
    {
      "epoch": 6.44340723453909,
      "grad_norm": 0.271484375,
      "learning_rate": 5.6196592190771865e-05,
      "loss": 0.0047,
      "step": 13805
    },
    {
      "epoch": 6.4438739789965,
      "grad_norm": 0.162109375,
      "learning_rate": 5.618340181230263e-05,
      "loss": 0.0026,
      "step": 13806
    },
    {
      "epoch": 6.444340723453909,
      "grad_norm": 0.265625,
      "learning_rate": 5.61702123772534e-05,
      "loss": 0.007,
      "step": 13807
    },
    {
      "epoch": 6.4448074679113185,
      "grad_norm": 0.177734375,
      "learning_rate": 5.615702388590809e-05,
      "loss": 0.0012,
      "step": 13808
    },
    {
      "epoch": 6.445274212368728,
      "grad_norm": 0.3046875,
      "learning_rate": 5.6143836338550735e-05,
      "loss": 0.0066,
      "step": 13809
    },
    {
      "epoch": 6.445740956826137,
      "grad_norm": 0.0712890625,
      "learning_rate": 5.6130649735465225e-05,
      "loss": 0.0008,
      "step": 13810
    },
    {
      "epoch": 6.446207701283547,
      "grad_norm": 0.11279296875,
      "learning_rate": 5.6117464076935475e-05,
      "loss": 0.0008,
      "step": 13811
    },
    {
      "epoch": 6.446674445740957,
      "grad_norm": 0.265625,
      "learning_rate": 5.610427936324546e-05,
      "loss": 0.0014,
      "step": 13812
    },
    {
      "epoch": 6.447141190198367,
      "grad_norm": 0.1533203125,
      "learning_rate": 5.609109559467898e-05,
      "loss": 0.0013,
      "step": 13813
    },
    {
      "epoch": 6.447607934655776,
      "grad_norm": 0.09423828125,
      "learning_rate": 5.6077912771519926e-05,
      "loss": 0.0007,
      "step": 13814
    },
    {
      "epoch": 6.448074679113185,
      "grad_norm": 0.10986328125,
      "learning_rate": 5.6064730894052174e-05,
      "loss": 0.0006,
      "step": 13815
    },
    {
      "epoch": 6.448541423570595,
      "grad_norm": 0.1552734375,
      "learning_rate": 5.605154996255948e-05,
      "loss": 0.0011,
      "step": 13816
    },
    {
      "epoch": 6.449008168028005,
      "grad_norm": 0.2431640625,
      "learning_rate": 5.603836997732568e-05,
      "loss": 0.0013,
      "step": 13817
    },
    {
      "epoch": 6.449474912485414,
      "grad_norm": 0.314453125,
      "learning_rate": 5.6025190938634586e-05,
      "loss": 0.0014,
      "step": 13818
    },
    {
      "epoch": 6.449941656942824,
      "grad_norm": 0.08740234375,
      "learning_rate": 5.601201284676989e-05,
      "loss": 0.0028,
      "step": 13819
    },
    {
      "epoch": 6.450408401400233,
      "grad_norm": 0.53125,
      "learning_rate": 5.599883570201535e-05,
      "loss": 0.0038,
      "step": 13820
    },
    {
      "epoch": 6.450875145857643,
      "grad_norm": 0.263671875,
      "learning_rate": 5.598565950465474e-05,
      "loss": 0.0015,
      "step": 13821
    },
    {
      "epoch": 6.451341890315052,
      "grad_norm": 0.1982421875,
      "learning_rate": 5.597248425497168e-05,
      "loss": 0.0008,
      "step": 13822
    },
    {
      "epoch": 6.451808634772462,
      "grad_norm": 0.0751953125,
      "learning_rate": 5.5959309953249885e-05,
      "loss": 0.0005,
      "step": 13823
    },
    {
      "epoch": 6.452275379229872,
      "grad_norm": 0.216796875,
      "learning_rate": 5.594613659977305e-05,
      "loss": 0.0023,
      "step": 13824
    },
    {
      "epoch": 6.4527421236872815,
      "grad_norm": 0.11865234375,
      "learning_rate": 5.593296419482473e-05,
      "loss": 0.0008,
      "step": 13825
    },
    {
      "epoch": 6.45320886814469,
      "grad_norm": 0.14453125,
      "learning_rate": 5.5919792738688595e-05,
      "loss": 0.0008,
      "step": 13826
    },
    {
      "epoch": 6.4536756126021,
      "grad_norm": 0.2080078125,
      "learning_rate": 5.5906622231648256e-05,
      "loss": 0.0008,
      "step": 13827
    },
    {
      "epoch": 6.45414235705951,
      "grad_norm": 0.2275390625,
      "learning_rate": 5.5893452673987225e-05,
      "loss": 0.0012,
      "step": 13828
    },
    {
      "epoch": 6.45460910151692,
      "grad_norm": 0.11083984375,
      "learning_rate": 5.588028406598913e-05,
      "loss": 0.0027,
      "step": 13829
    },
    {
      "epoch": 6.455075845974329,
      "grad_norm": 0.39453125,
      "learning_rate": 5.586711640793745e-05,
      "loss": 0.0055,
      "step": 13830
    },
    {
      "epoch": 6.4555425904317385,
      "grad_norm": 0.55078125,
      "learning_rate": 5.5853949700115704e-05,
      "loss": 0.002,
      "step": 13831
    },
    {
      "epoch": 6.456009334889148,
      "grad_norm": 0.107421875,
      "learning_rate": 5.584078394280744e-05,
      "loss": 0.0039,
      "step": 13832
    },
    {
      "epoch": 6.456476079346558,
      "grad_norm": 0.2001953125,
      "learning_rate": 5.5827619136296064e-05,
      "loss": 0.0043,
      "step": 13833
    },
    {
      "epoch": 6.456942823803967,
      "grad_norm": 0.55859375,
      "learning_rate": 5.5814455280865056e-05,
      "loss": 0.0026,
      "step": 13834
    },
    {
      "epoch": 6.457409568261377,
      "grad_norm": 0.126953125,
      "learning_rate": 5.580129237679788e-05,
      "loss": 0.0031,
      "step": 13835
    },
    {
      "epoch": 6.457876312718787,
      "grad_norm": 0.06787109375,
      "learning_rate": 5.57881304243779e-05,
      "loss": 0.0005,
      "step": 13836
    },
    {
      "epoch": 6.458343057176196,
      "grad_norm": 0.0859375,
      "learning_rate": 5.57749694238885e-05,
      "loss": 0.0043,
      "step": 13837
    },
    {
      "epoch": 6.458809801633605,
      "grad_norm": 0.035888671875,
      "learning_rate": 5.5761809375613125e-05,
      "loss": 0.0004,
      "step": 13838
    },
    {
      "epoch": 6.459276546091015,
      "grad_norm": 0.1796875,
      "learning_rate": 5.574865027983505e-05,
      "loss": 0.0014,
      "step": 13839
    },
    {
      "epoch": 6.459743290548425,
      "grad_norm": 0.267578125,
      "learning_rate": 5.573549213683762e-05,
      "loss": 0.0011,
      "step": 13840
    },
    {
      "epoch": 6.460210035005835,
      "grad_norm": 0.12890625,
      "learning_rate": 5.5722334946904196e-05,
      "loss": 0.0039,
      "step": 13841
    },
    {
      "epoch": 6.460676779463244,
      "grad_norm": 0.302734375,
      "learning_rate": 5.5709178710318e-05,
      "loss": 0.0016,
      "step": 13842
    },
    {
      "epoch": 6.461143523920653,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.569602342736231e-05,
      "loss": 0.0013,
      "step": 13843
    },
    {
      "epoch": 6.461610268378063,
      "grad_norm": 0.1953125,
      "learning_rate": 5.568286909832043e-05,
      "loss": 0.0024,
      "step": 13844
    },
    {
      "epoch": 6.462077012835473,
      "grad_norm": 0.1025390625,
      "learning_rate": 5.566971572347553e-05,
      "loss": 0.0006,
      "step": 13845
    },
    {
      "epoch": 6.462543757292882,
      "grad_norm": 0.051025390625,
      "learning_rate": 5.5656563303110866e-05,
      "loss": 0.0005,
      "step": 13846
    },
    {
      "epoch": 6.463010501750292,
      "grad_norm": 0.32421875,
      "learning_rate": 5.564341183750959e-05,
      "loss": 0.0019,
      "step": 13847
    },
    {
      "epoch": 6.4634772462077015,
      "grad_norm": 0.1474609375,
      "learning_rate": 5.5630261326954834e-05,
      "loss": 0.0009,
      "step": 13848
    },
    {
      "epoch": 6.463943990665111,
      "grad_norm": 0.353515625,
      "learning_rate": 5.561711177172979e-05,
      "loss": 0.0056,
      "step": 13849
    },
    {
      "epoch": 6.46441073512252,
      "grad_norm": 0.037353515625,
      "learning_rate": 5.56039631721176e-05,
      "loss": 0.0006,
      "step": 13850
    },
    {
      "epoch": 6.46487747957993,
      "grad_norm": 0.173828125,
      "learning_rate": 5.559081552840131e-05,
      "loss": 0.0013,
      "step": 13851
    },
    {
      "epoch": 6.46534422403734,
      "grad_norm": 0.08251953125,
      "learning_rate": 5.557766884086403e-05,
      "loss": 0.0009,
      "step": 13852
    },
    {
      "epoch": 6.465810968494749,
      "grad_norm": 0.240234375,
      "learning_rate": 5.556452310978887e-05,
      "loss": 0.0012,
      "step": 13853
    },
    {
      "epoch": 6.4662777129521585,
      "grad_norm": 0.04150390625,
      "learning_rate": 5.5551378335458784e-05,
      "loss": 0.0006,
      "step": 13854
    },
    {
      "epoch": 6.466744457409568,
      "grad_norm": 0.255859375,
      "learning_rate": 5.553823451815685e-05,
      "loss": 0.0012,
      "step": 13855
    },
    {
      "epoch": 6.467211201866978,
      "grad_norm": 0.28125,
      "learning_rate": 5.5525091658166084e-05,
      "loss": 0.0015,
      "step": 13856
    },
    {
      "epoch": 6.467677946324388,
      "grad_norm": 0.0830078125,
      "learning_rate": 5.551194975576941e-05,
      "loss": 0.0008,
      "step": 13857
    },
    {
      "epoch": 6.468144690781797,
      "grad_norm": 0.1103515625,
      "learning_rate": 5.549880881124985e-05,
      "loss": 0.0032,
      "step": 13858
    },
    {
      "epoch": 6.468611435239207,
      "grad_norm": 0.03662109375,
      "learning_rate": 5.548566882489028e-05,
      "loss": 0.0004,
      "step": 13859
    },
    {
      "epoch": 6.469078179696616,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.547252979697364e-05,
      "loss": 0.0025,
      "step": 13860
    },
    {
      "epoch": 6.469544924154025,
      "grad_norm": 0.443359375,
      "learning_rate": 5.5459391727782894e-05,
      "loss": 0.0097,
      "step": 13861
    },
    {
      "epoch": 6.470011668611435,
      "grad_norm": 0.11962890625,
      "learning_rate": 5.5446254617600804e-05,
      "loss": 0.0078,
      "step": 13862
    },
    {
      "epoch": 6.470478413068845,
      "grad_norm": 0.228515625,
      "learning_rate": 5.54331184667103e-05,
      "loss": 0.0016,
      "step": 13863
    },
    {
      "epoch": 6.470945157526255,
      "grad_norm": 0.0732421875,
      "learning_rate": 5.541998327539424e-05,
      "loss": 0.0006,
      "step": 13864
    },
    {
      "epoch": 6.471411901983664,
      "grad_norm": 0.1611328125,
      "learning_rate": 5.540684904393536e-05,
      "loss": 0.0008,
      "step": 13865
    },
    {
      "epoch": 6.471878646441073,
      "grad_norm": 0.06591796875,
      "learning_rate": 5.5393715772616494e-05,
      "loss": 0.0007,
      "step": 13866
    },
    {
      "epoch": 6.472345390898483,
      "grad_norm": 0.033447265625,
      "learning_rate": 5.538058346172046e-05,
      "loss": 0.0003,
      "step": 13867
    },
    {
      "epoch": 6.472812135355893,
      "grad_norm": 0.220703125,
      "learning_rate": 5.536745211152993e-05,
      "loss": 0.001,
      "step": 13868
    },
    {
      "epoch": 6.473278879813302,
      "grad_norm": 0.353515625,
      "learning_rate": 5.5354321722327685e-05,
      "loss": 0.0011,
      "step": 13869
    },
    {
      "epoch": 6.473745624270712,
      "grad_norm": 0.099609375,
      "learning_rate": 5.534119229439645e-05,
      "loss": 0.0033,
      "step": 13870
    },
    {
      "epoch": 6.4742123687281214,
      "grad_norm": 0.255859375,
      "learning_rate": 5.5328063828018875e-05,
      "loss": 0.0012,
      "step": 13871
    },
    {
      "epoch": 6.474679113185531,
      "grad_norm": 0.0703125,
      "learning_rate": 5.531493632347764e-05,
      "loss": 0.0007,
      "step": 13872
    },
    {
      "epoch": 6.47514585764294,
      "grad_norm": 0.0859375,
      "learning_rate": 5.5301809781055434e-05,
      "loss": 0.0051,
      "step": 13873
    },
    {
      "epoch": 6.47561260210035,
      "grad_norm": 0.1640625,
      "learning_rate": 5.5288684201034835e-05,
      "loss": 0.0012,
      "step": 13874
    },
    {
      "epoch": 6.47607934655776,
      "grad_norm": 0.16015625,
      "learning_rate": 5.527555958369848e-05,
      "loss": 0.001,
      "step": 13875
    },
    {
      "epoch": 6.4765460910151695,
      "grad_norm": 0.07373046875,
      "learning_rate": 5.526243592932897e-05,
      "loss": 0.0037,
      "step": 13876
    },
    {
      "epoch": 6.4770128354725784,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.524931323820881e-05,
      "loss": 0.0006,
      "step": 13877
    },
    {
      "epoch": 6.477479579929988,
      "grad_norm": 0.1494140625,
      "learning_rate": 5.523619151062065e-05,
      "loss": 0.0007,
      "step": 13878
    },
    {
      "epoch": 6.477946324387398,
      "grad_norm": 0.06689453125,
      "learning_rate": 5.5223070746846894e-05,
      "loss": 0.0041,
      "step": 13879
    },
    {
      "epoch": 6.478413068844808,
      "grad_norm": 0.049560546875,
      "learning_rate": 5.520995094717012e-05,
      "loss": 0.0005,
      "step": 13880
    },
    {
      "epoch": 6.478879813302217,
      "grad_norm": 0.3125,
      "learning_rate": 5.519683211187282e-05,
      "loss": 0.0017,
      "step": 13881
    },
    {
      "epoch": 6.4793465577596265,
      "grad_norm": 0.072265625,
      "learning_rate": 5.5183714241237406e-05,
      "loss": 0.0005,
      "step": 13882
    },
    {
      "epoch": 6.479813302217036,
      "grad_norm": 0.173828125,
      "learning_rate": 5.517059733554635e-05,
      "loss": 0.004,
      "step": 13883
    },
    {
      "epoch": 6.480280046674446,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.51574813950821e-05,
      "loss": 0.0008,
      "step": 13884
    },
    {
      "epoch": 6.480746791131855,
      "grad_norm": 0.12890625,
      "learning_rate": 5.5144366420127e-05,
      "loss": 0.001,
      "step": 13885
    },
    {
      "epoch": 6.481213535589265,
      "grad_norm": 0.185546875,
      "learning_rate": 5.5131252410963466e-05,
      "loss": 0.003,
      "step": 13886
    },
    {
      "epoch": 6.481680280046675,
      "grad_norm": 0.138671875,
      "learning_rate": 5.5118139367873866e-05,
      "loss": 0.0007,
      "step": 13887
    },
    {
      "epoch": 6.482147024504084,
      "grad_norm": 0.078125,
      "learning_rate": 5.510502729114051e-05,
      "loss": 0.0006,
      "step": 13888
    },
    {
      "epoch": 6.482613768961493,
      "grad_norm": 0.2392578125,
      "learning_rate": 5.509191618104572e-05,
      "loss": 0.0012,
      "step": 13889
    },
    {
      "epoch": 6.483080513418903,
      "grad_norm": 0.12060546875,
      "learning_rate": 5.5078806037871834e-05,
      "loss": 0.0008,
      "step": 13890
    },
    {
      "epoch": 6.483547257876313,
      "grad_norm": 0.07763671875,
      "learning_rate": 5.506569686190106e-05,
      "loss": 0.0006,
      "step": 13891
    },
    {
      "epoch": 6.484014002333723,
      "grad_norm": 0.220703125,
      "learning_rate": 5.505258865341568e-05,
      "loss": 0.001,
      "step": 13892
    },
    {
      "epoch": 6.484480746791132,
      "grad_norm": 0.42578125,
      "learning_rate": 5.503948141269799e-05,
      "loss": 0.0045,
      "step": 13893
    },
    {
      "epoch": 6.484947491248541,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.502637514003014e-05,
      "loss": 0.0035,
      "step": 13894
    },
    {
      "epoch": 6.485414235705951,
      "grad_norm": 0.2236328125,
      "learning_rate": 5.501326983569429e-05,
      "loss": 0.0009,
      "step": 13895
    },
    {
      "epoch": 6.48588098016336,
      "grad_norm": 0.1435546875,
      "learning_rate": 5.50001654999727e-05,
      "loss": 0.0008,
      "step": 13896
    },
    {
      "epoch": 6.48634772462077,
      "grad_norm": 0.058837890625,
      "learning_rate": 5.498706213314744e-05,
      "loss": 0.0004,
      "step": 13897
    },
    {
      "epoch": 6.48681446907818,
      "grad_norm": 0.09814453125,
      "learning_rate": 5.497395973550066e-05,
      "loss": 0.0007,
      "step": 13898
    },
    {
      "epoch": 6.4872812135355895,
      "grad_norm": 0.0810546875,
      "learning_rate": 5.496085830731455e-05,
      "loss": 0.0006,
      "step": 13899
    },
    {
      "epoch": 6.487747957992999,
      "grad_norm": 0.03466796875,
      "learning_rate": 5.494775784887106e-05,
      "loss": 0.0004,
      "step": 13900
    },
    {
      "epoch": 6.488214702450408,
      "grad_norm": 0.056640625,
      "learning_rate": 5.493465836045235e-05,
      "loss": 0.0005,
      "step": 13901
    },
    {
      "epoch": 6.488681446907818,
      "grad_norm": 0.022705078125,
      "learning_rate": 5.492155984234049e-05,
      "loss": 0.0003,
      "step": 13902
    },
    {
      "epoch": 6.489148191365228,
      "grad_norm": 0.26953125,
      "learning_rate": 5.490846229481741e-05,
      "loss": 0.0017,
      "step": 13903
    },
    {
      "epoch": 6.489614935822637,
      "grad_norm": 0.07763671875,
      "learning_rate": 5.489536571816516e-05,
      "loss": 0.0005,
      "step": 13904
    },
    {
      "epoch": 6.4900816802800465,
      "grad_norm": 0.09619140625,
      "learning_rate": 5.4882270112665776e-05,
      "loss": 0.0007,
      "step": 13905
    },
    {
      "epoch": 6.490548424737456,
      "grad_norm": 0.2265625,
      "learning_rate": 5.486917547860115e-05,
      "loss": 0.001,
      "step": 13906
    },
    {
      "epoch": 6.491015169194866,
      "grad_norm": 0.05029296875,
      "learning_rate": 5.485608181625323e-05,
      "loss": 0.0004,
      "step": 13907
    },
    {
      "epoch": 6.491481913652275,
      "grad_norm": 0.08837890625,
      "learning_rate": 5.484298912590402e-05,
      "loss": 0.0005,
      "step": 13908
    },
    {
      "epoch": 6.491948658109685,
      "grad_norm": 0.06787109375,
      "learning_rate": 5.4829897407835305e-05,
      "loss": 0.0005,
      "step": 13909
    },
    {
      "epoch": 6.492415402567095,
      "grad_norm": 0.236328125,
      "learning_rate": 5.481680666232905e-05,
      "loss": 0.001,
      "step": 13910
    },
    {
      "epoch": 6.492882147024504,
      "grad_norm": 0.22265625,
      "learning_rate": 5.4803716889667045e-05,
      "loss": 0.0009,
      "step": 13911
    },
    {
      "epoch": 6.493348891481913,
      "grad_norm": 0.138671875,
      "learning_rate": 5.4790628090131157e-05,
      "loss": 0.0009,
      "step": 13912
    },
    {
      "epoch": 6.493815635939323,
      "grad_norm": 0.046875,
      "learning_rate": 5.477754026400326e-05,
      "loss": 0.0027,
      "step": 13913
    },
    {
      "epoch": 6.494282380396733,
      "grad_norm": 0.3203125,
      "learning_rate": 5.476445341156504e-05,
      "loss": 0.0018,
      "step": 13914
    },
    {
      "epoch": 6.494749124854143,
      "grad_norm": 0.177734375,
      "learning_rate": 5.475136753309833e-05,
      "loss": 0.0041,
      "step": 13915
    },
    {
      "epoch": 6.495215869311552,
      "grad_norm": 0.412109375,
      "learning_rate": 5.473828262888492e-05,
      "loss": 0.0019,
      "step": 13916
    },
    {
      "epoch": 6.495682613768961,
      "grad_norm": 0.0277099609375,
      "learning_rate": 5.472519869920647e-05,
      "loss": 0.0004,
      "step": 13917
    },
    {
      "epoch": 6.496149358226371,
      "grad_norm": 0.2255859375,
      "learning_rate": 5.4712115744344726e-05,
      "loss": 0.002,
      "step": 13918
    },
    {
      "epoch": 6.496616102683781,
      "grad_norm": 0.04345703125,
      "learning_rate": 5.469903376458142e-05,
      "loss": 0.0004,
      "step": 13919
    },
    {
      "epoch": 6.49708284714119,
      "grad_norm": 0.275390625,
      "learning_rate": 5.468595276019813e-05,
      "loss": 0.0049,
      "step": 13920
    },
    {
      "epoch": 6.4975495915986,
      "grad_norm": 0.048828125,
      "learning_rate": 5.4672872731476565e-05,
      "loss": 0.0004,
      "step": 13921
    },
    {
      "epoch": 6.4980163360560095,
      "grad_norm": 0.05517578125,
      "learning_rate": 5.465979367869838e-05,
      "loss": 0.0005,
      "step": 13922
    },
    {
      "epoch": 6.498483080513419,
      "grad_norm": 0.030517578125,
      "learning_rate": 5.464671560214509e-05,
      "loss": 0.0003,
      "step": 13923
    },
    {
      "epoch": 6.498949824970828,
      "grad_norm": 0.09033203125,
      "learning_rate": 5.463363850209835e-05,
      "loss": 0.0005,
      "step": 13924
    },
    {
      "epoch": 6.499416569428238,
      "grad_norm": 0.03173828125,
      "learning_rate": 5.4620562378839734e-05,
      "loss": 0.0004,
      "step": 13925
    },
    {
      "epoch": 6.499883313885648,
      "grad_norm": 0.263671875,
      "learning_rate": 5.4607487232650724e-05,
      "loss": 0.0013,
      "step": 13926
    },
    {
      "epoch": 6.500350058343058,
      "grad_norm": 0.80078125,
      "learning_rate": 5.4594413063812924e-05,
      "loss": 0.0066,
      "step": 13927
    },
    {
      "epoch": 6.5008168028004665,
      "grad_norm": 0.06005859375,
      "learning_rate": 5.458133987260774e-05,
      "loss": 0.0005,
      "step": 13928
    },
    {
      "epoch": 6.501283547257876,
      "grad_norm": 0.111328125,
      "learning_rate": 5.4568267659316704e-05,
      "loss": 0.0006,
      "step": 13929
    },
    {
      "epoch": 6.501750291715286,
      "grad_norm": 0.341796875,
      "learning_rate": 5.455519642422131e-05,
      "loss": 0.0015,
      "step": 13930
    },
    {
      "epoch": 6.502217036172695,
      "grad_norm": 0.076171875,
      "learning_rate": 5.454212616760292e-05,
      "loss": 0.0004,
      "step": 13931
    },
    {
      "epoch": 6.502683780630105,
      "grad_norm": 0.0299072265625,
      "learning_rate": 5.452905688974299e-05,
      "loss": 0.0004,
      "step": 13932
    },
    {
      "epoch": 6.503150525087515,
      "grad_norm": 0.162109375,
      "learning_rate": 5.451598859092297e-05,
      "loss": 0.0038,
      "step": 13933
    },
    {
      "epoch": 6.503617269544924,
      "grad_norm": 0.171875,
      "learning_rate": 5.450292127142413e-05,
      "loss": 0.001,
      "step": 13934
    },
    {
      "epoch": 6.504084014002334,
      "grad_norm": 0.04345703125,
      "learning_rate": 5.4489854931527874e-05,
      "loss": 0.0004,
      "step": 13935
    },
    {
      "epoch": 6.504550758459743,
      "grad_norm": 0.07421875,
      "learning_rate": 5.4476789571515576e-05,
      "loss": 0.0005,
      "step": 13936
    },
    {
      "epoch": 6.504550758459743,
      "eval_loss": 2.296372175216675,
      "eval_runtime": 84.5427,
      "eval_samples_per_second": 21.338,
      "eval_steps_per_second": 2.673,
      "step": 13936
    },
    {
      "epoch": 6.505017502917153,
      "grad_norm": 0.21484375,
      "learning_rate": 5.446372519166847e-05,
      "loss": 0.0012,
      "step": 13937
    },
    {
      "epoch": 6.505484247374563,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.445066179226787e-05,
      "loss": 0.0008,
      "step": 13938
    },
    {
      "epoch": 6.505950991831972,
      "grad_norm": 0.08544921875,
      "learning_rate": 5.4437599373595116e-05,
      "loss": 0.0006,
      "step": 13939
    },
    {
      "epoch": 6.506417736289381,
      "grad_norm": 0.09375,
      "learning_rate": 5.442453793593136e-05,
      "loss": 0.0007,
      "step": 13940
    },
    {
      "epoch": 6.506884480746791,
      "grad_norm": 0.055419921875,
      "learning_rate": 5.4411477479557904e-05,
      "loss": 0.0004,
      "step": 13941
    },
    {
      "epoch": 6.507351225204201,
      "grad_norm": 0.2099609375,
      "learning_rate": 5.4398418004755914e-05,
      "loss": 0.0011,
      "step": 13942
    },
    {
      "epoch": 6.507817969661611,
      "grad_norm": 0.318359375,
      "learning_rate": 5.4385359511806544e-05,
      "loss": 0.0017,
      "step": 13943
    },
    {
      "epoch": 6.50828471411902,
      "grad_norm": 0.15625,
      "learning_rate": 5.4372302000990995e-05,
      "loss": 0.0026,
      "step": 13944
    },
    {
      "epoch": 6.5087514585764294,
      "grad_norm": 0.3125,
      "learning_rate": 5.435924547259046e-05,
      "loss": 0.0013,
      "step": 13945
    },
    {
      "epoch": 6.509218203033839,
      "grad_norm": 0.21875,
      "learning_rate": 5.4346189926885957e-05,
      "loss": 0.0014,
      "step": 13946
    },
    {
      "epoch": 6.509684947491248,
      "grad_norm": 0.2734375,
      "learning_rate": 5.433313536415864e-05,
      "loss": 0.0013,
      "step": 13947
    },
    {
      "epoch": 6.510151691948658,
      "grad_norm": 0.087890625,
      "learning_rate": 5.4320081784689625e-05,
      "loss": 0.0004,
      "step": 13948
    },
    {
      "epoch": 6.510618436406068,
      "grad_norm": 0.12890625,
      "learning_rate": 5.4307029188759896e-05,
      "loss": 0.0009,
      "step": 13949
    },
    {
      "epoch": 6.5110851808634775,
      "grad_norm": 0.080078125,
      "learning_rate": 5.4293977576650526e-05,
      "loss": 0.0006,
      "step": 13950
    },
    {
      "epoch": 6.5115519253208864,
      "grad_norm": 0.205078125,
      "learning_rate": 5.428092694864256e-05,
      "loss": 0.0031,
      "step": 13951
    },
    {
      "epoch": 6.512018669778296,
      "grad_norm": 0.1044921875,
      "learning_rate": 5.426787730501694e-05,
      "loss": 0.0006,
      "step": 13952
    },
    {
      "epoch": 6.512485414235706,
      "grad_norm": 0.17578125,
      "learning_rate": 5.4254828646054646e-05,
      "loss": 0.0044,
      "step": 13953
    },
    {
      "epoch": 6.512952158693116,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.424178097203669e-05,
      "loss": 0.001,
      "step": 13954
    },
    {
      "epoch": 6.513418903150525,
      "grad_norm": 0.0830078125,
      "learning_rate": 5.422873428324393e-05,
      "loss": 0.0006,
      "step": 13955
    },
    {
      "epoch": 6.5138856476079345,
      "grad_norm": 0.08740234375,
      "learning_rate": 5.421568857995729e-05,
      "loss": 0.002,
      "step": 13956
    },
    {
      "epoch": 6.514352392065344,
      "grad_norm": 0.045654296875,
      "learning_rate": 5.420264386245772e-05,
      "loss": 0.0003,
      "step": 13957
    },
    {
      "epoch": 6.514819136522754,
      "grad_norm": 0.020263671875,
      "learning_rate": 5.418960013102601e-05,
      "loss": 0.0002,
      "step": 13958
    },
    {
      "epoch": 6.515285880980163,
      "grad_norm": 0.220703125,
      "learning_rate": 5.417655738594307e-05,
      "loss": 0.003,
      "step": 13959
    },
    {
      "epoch": 6.515752625437573,
      "grad_norm": 0.026611328125,
      "learning_rate": 5.416351562748965e-05,
      "loss": 0.0002,
      "step": 13960
    },
    {
      "epoch": 6.516219369894983,
      "grad_norm": 0.041748046875,
      "learning_rate": 5.41504748559466e-05,
      "loss": 0.0004,
      "step": 13961
    },
    {
      "epoch": 6.516686114352392,
      "grad_norm": 0.208984375,
      "learning_rate": 5.413743507159473e-05,
      "loss": 0.0026,
      "step": 13962
    },
    {
      "epoch": 6.517152858809801,
      "grad_norm": 0.11767578125,
      "learning_rate": 5.412439627471473e-05,
      "loss": 0.0054,
      "step": 13963
    },
    {
      "epoch": 6.517619603267211,
      "grad_norm": 0.201171875,
      "learning_rate": 5.4111358465587384e-05,
      "loss": 0.0009,
      "step": 13964
    },
    {
      "epoch": 6.518086347724621,
      "grad_norm": 0.1337890625,
      "learning_rate": 5.4098321644493445e-05,
      "loss": 0.0007,
      "step": 13965
    },
    {
      "epoch": 6.518553092182031,
      "grad_norm": 0.24609375,
      "learning_rate": 5.408528581171352e-05,
      "loss": 0.001,
      "step": 13966
    },
    {
      "epoch": 6.51901983663944,
      "grad_norm": 0.2255859375,
      "learning_rate": 5.4072250967528335e-05,
      "loss": 0.0018,
      "step": 13967
    },
    {
      "epoch": 6.519486581096849,
      "grad_norm": 0.046630859375,
      "learning_rate": 5.4059217112218596e-05,
      "loss": 0.0004,
      "step": 13968
    },
    {
      "epoch": 6.519953325554259,
      "grad_norm": 0.4296875,
      "learning_rate": 5.404618424606483e-05,
      "loss": 0.0043,
      "step": 13969
    },
    {
      "epoch": 6.520420070011669,
      "grad_norm": 0.12451171875,
      "learning_rate": 5.403315236934772e-05,
      "loss": 0.0007,
      "step": 13970
    },
    {
      "epoch": 6.520886814469078,
      "grad_norm": 0.0380859375,
      "learning_rate": 5.4020121482347864e-05,
      "loss": 0.0004,
      "step": 13971
    },
    {
      "epoch": 6.521353558926488,
      "grad_norm": 0.283203125,
      "learning_rate": 5.400709158534578e-05,
      "loss": 0.0012,
      "step": 13972
    },
    {
      "epoch": 6.5218203033838975,
      "grad_norm": 0.049560546875,
      "learning_rate": 5.3994062678622036e-05,
      "loss": 0.0004,
      "step": 13973
    },
    {
      "epoch": 6.522287047841306,
      "grad_norm": 0.048828125,
      "learning_rate": 5.398103476245722e-05,
      "loss": 0.0004,
      "step": 13974
    },
    {
      "epoch": 6.522753792298716,
      "grad_norm": 0.11767578125,
      "learning_rate": 5.396800783713173e-05,
      "loss": 0.0009,
      "step": 13975
    },
    {
      "epoch": 6.523220536756126,
      "grad_norm": 0.07763671875,
      "learning_rate": 5.3954981902926147e-05,
      "loss": 0.0004,
      "step": 13976
    },
    {
      "epoch": 6.523687281213536,
      "grad_norm": 0.068359375,
      "learning_rate": 5.394195696012086e-05,
      "loss": 0.0006,
      "step": 13977
    },
    {
      "epoch": 6.524154025670946,
      "grad_norm": 0.056396484375,
      "learning_rate": 5.392893300899633e-05,
      "loss": 0.0028,
      "step": 13978
    },
    {
      "epoch": 6.5246207701283545,
      "grad_norm": 0.1396484375,
      "learning_rate": 5.391591004983304e-05,
      "loss": 0.0008,
      "step": 13979
    },
    {
      "epoch": 6.525087514585764,
      "grad_norm": 0.12890625,
      "learning_rate": 5.390288808291129e-05,
      "loss": 0.0008,
      "step": 13980
    },
    {
      "epoch": 6.525554259043174,
      "grad_norm": 0.37890625,
      "learning_rate": 5.388986710851152e-05,
      "loss": 0.0034,
      "step": 13981
    },
    {
      "epoch": 6.526021003500583,
      "grad_norm": 0.1630859375,
      "learning_rate": 5.3876847126914085e-05,
      "loss": 0.0009,
      "step": 13982
    },
    {
      "epoch": 6.526487747957993,
      "grad_norm": 0.234375,
      "learning_rate": 5.386382813839929e-05,
      "loss": 0.0018,
      "step": 13983
    },
    {
      "epoch": 6.526954492415403,
      "grad_norm": 0.0322265625,
      "learning_rate": 5.3850810143247444e-05,
      "loss": 0.0003,
      "step": 13984
    },
    {
      "epoch": 6.527421236872812,
      "grad_norm": 0.06201171875,
      "learning_rate": 5.383779314173891e-05,
      "loss": 0.0035,
      "step": 13985
    },
    {
      "epoch": 6.527887981330222,
      "grad_norm": 0.056640625,
      "learning_rate": 5.382477713415388e-05,
      "loss": 0.0015,
      "step": 13986
    },
    {
      "epoch": 6.528354725787631,
      "grad_norm": 0.431640625,
      "learning_rate": 5.3811762120772615e-05,
      "loss": 0.0027,
      "step": 13987
    },
    {
      "epoch": 6.528821470245041,
      "grad_norm": 0.054931640625,
      "learning_rate": 5.37987481018754e-05,
      "loss": 0.0004,
      "step": 13988
    },
    {
      "epoch": 6.529288214702451,
      "grad_norm": 0.034912109375,
      "learning_rate": 5.37857350777424e-05,
      "loss": 0.0004,
      "step": 13989
    },
    {
      "epoch": 6.52975495915986,
      "grad_norm": 0.310546875,
      "learning_rate": 5.377272304865376e-05,
      "loss": 0.0007,
      "step": 13990
    },
    {
      "epoch": 6.530221703617269,
      "grad_norm": 0.2265625,
      "learning_rate": 5.375971201488972e-05,
      "loss": 0.0014,
      "step": 13991
    },
    {
      "epoch": 6.530688448074679,
      "grad_norm": 0.1328125,
      "learning_rate": 5.374670197673035e-05,
      "loss": 0.0028,
      "step": 13992
    },
    {
      "epoch": 6.531155192532089,
      "grad_norm": 0.107421875,
      "learning_rate": 5.37336929344558e-05,
      "loss": 0.0007,
      "step": 13993
    },
    {
      "epoch": 6.531621936989498,
      "grad_norm": 0.040771484375,
      "learning_rate": 5.3720684888346225e-05,
      "loss": 0.0004,
      "step": 13994
    },
    {
      "epoch": 6.532088681446908,
      "grad_norm": 0.32421875,
      "learning_rate": 5.37076778386816e-05,
      "loss": 0.0012,
      "step": 13995
    },
    {
      "epoch": 6.5325554259043175,
      "grad_norm": 0.2734375,
      "learning_rate": 5.3694671785742036e-05,
      "loss": 0.0013,
      "step": 13996
    },
    {
      "epoch": 6.533022170361727,
      "grad_norm": 0.2109375,
      "learning_rate": 5.368166672980761e-05,
      "loss": 0.0014,
      "step": 13997
    },
    {
      "epoch": 6.533488914819136,
      "grad_norm": 0.068359375,
      "learning_rate": 5.3668662671158244e-05,
      "loss": 0.0005,
      "step": 13998
    },
    {
      "epoch": 6.533955659276546,
      "grad_norm": 0.042724609375,
      "learning_rate": 5.3655659610073986e-05,
      "loss": 0.0024,
      "step": 13999
    },
    {
      "epoch": 6.534422403733956,
      "grad_norm": 0.11865234375,
      "learning_rate": 5.364265754683483e-05,
      "loss": 0.0005,
      "step": 14000
    },
    {
      "epoch": 6.534889148191366,
      "grad_norm": 0.1650390625,
      "learning_rate": 5.3629656481720646e-05,
      "loss": 0.001,
      "step": 14001
    },
    {
      "epoch": 6.5353558926487745,
      "grad_norm": 0.1630859375,
      "learning_rate": 5.361665641501141e-05,
      "loss": 0.005,
      "step": 14002
    },
    {
      "epoch": 6.535822637106184,
      "grad_norm": 0.0791015625,
      "learning_rate": 5.3603657346987066e-05,
      "loss": 0.0005,
      "step": 14003
    },
    {
      "epoch": 6.536289381563594,
      "grad_norm": 0.0791015625,
      "learning_rate": 5.3590659277927424e-05,
      "loss": 0.0006,
      "step": 14004
    },
    {
      "epoch": 6.536756126021004,
      "grad_norm": 0.039794921875,
      "learning_rate": 5.357766220811238e-05,
      "loss": 0.0004,
      "step": 14005
    },
    {
      "epoch": 6.537222870478413,
      "grad_norm": 0.0242919921875,
      "learning_rate": 5.356466613782181e-05,
      "loss": 0.0002,
      "step": 14006
    },
    {
      "epoch": 6.537689614935823,
      "grad_norm": 0.265625,
      "learning_rate": 5.355167106733545e-05,
      "loss": 0.0016,
      "step": 14007
    },
    {
      "epoch": 6.538156359393232,
      "grad_norm": 0.054443359375,
      "learning_rate": 5.35386769969332e-05,
      "loss": 0.0004,
      "step": 14008
    },
    {
      "epoch": 6.538623103850642,
      "grad_norm": 0.34375,
      "learning_rate": 5.3525683926894746e-05,
      "loss": 0.002,
      "step": 14009
    },
    {
      "epoch": 6.539089848308051,
      "grad_norm": 0.283203125,
      "learning_rate": 5.351269185749988e-05,
      "loss": 0.0009,
      "step": 14010
    },
    {
      "epoch": 6.539556592765461,
      "grad_norm": 0.10400390625,
      "learning_rate": 5.349970078902835e-05,
      "loss": 0.0007,
      "step": 14011
    },
    {
      "epoch": 6.540023337222871,
      "grad_norm": 0.349609375,
      "learning_rate": 5.3486710721759835e-05,
      "loss": 0.0015,
      "step": 14012
    },
    {
      "epoch": 6.5404900816802805,
      "grad_norm": 0.15625,
      "learning_rate": 5.347372165597406e-05,
      "loss": 0.0018,
      "step": 14013
    },
    {
      "epoch": 6.540956826137689,
      "grad_norm": 0.2890625,
      "learning_rate": 5.3460733591950695e-05,
      "loss": 0.0009,
      "step": 14014
    },
    {
      "epoch": 6.541423570595099,
      "grad_norm": 0.08837890625,
      "learning_rate": 5.344774652996936e-05,
      "loss": 0.0006,
      "step": 14015
    },
    {
      "epoch": 6.541890315052509,
      "grad_norm": 0.033447265625,
      "learning_rate": 5.3434760470309666e-05,
      "loss": 0.0004,
      "step": 14016
    },
    {
      "epoch": 6.542357059509918,
      "grad_norm": 0.4375,
      "learning_rate": 5.34217754132513e-05,
      "loss": 0.0018,
      "step": 14017
    },
    {
      "epoch": 6.542823803967328,
      "grad_norm": 0.02880859375,
      "learning_rate": 5.340879135907374e-05,
      "loss": 0.0004,
      "step": 14018
    },
    {
      "epoch": 6.5432905484247375,
      "grad_norm": 0.036376953125,
      "learning_rate": 5.3395808308056596e-05,
      "loss": 0.0003,
      "step": 14019
    },
    {
      "epoch": 6.543757292882147,
      "grad_norm": 0.1650390625,
      "learning_rate": 5.3382826260479454e-05,
      "loss": 0.001,
      "step": 14020
    },
    {
      "epoch": 6.544224037339557,
      "grad_norm": 0.06298828125,
      "learning_rate": 5.336984521662174e-05,
      "loss": 0.0004,
      "step": 14021
    },
    {
      "epoch": 6.544690781796966,
      "grad_norm": 0.1591796875,
      "learning_rate": 5.335686517676299e-05,
      "loss": 0.0041,
      "step": 14022
    },
    {
      "epoch": 6.545157526254376,
      "grad_norm": 0.271484375,
      "learning_rate": 5.334388614118273e-05,
      "loss": 0.0039,
      "step": 14023
    },
    {
      "epoch": 6.5456242707117855,
      "grad_norm": 0.0556640625,
      "learning_rate": 5.333090811016034e-05,
      "loss": 0.0005,
      "step": 14024
    },
    {
      "epoch": 6.5460910151691944,
      "grad_norm": 0.11279296875,
      "learning_rate": 5.3317931083975305e-05,
      "loss": 0.0007,
      "step": 14025
    },
    {
      "epoch": 6.546557759626604,
      "grad_norm": 0.0634765625,
      "learning_rate": 5.330495506290697e-05,
      "loss": 0.0005,
      "step": 14026
    },
    {
      "epoch": 6.547024504084014,
      "grad_norm": 0.12353515625,
      "learning_rate": 5.3291980047234754e-05,
      "loss": 0.0016,
      "step": 14027
    },
    {
      "epoch": 6.547491248541424,
      "grad_norm": 0.1552734375,
      "learning_rate": 5.3279006037238075e-05,
      "loss": 0.0008,
      "step": 14028
    },
    {
      "epoch": 6.547957992998834,
      "grad_norm": 0.19140625,
      "learning_rate": 5.326603303319621e-05,
      "loss": 0.0034,
      "step": 14029
    },
    {
      "epoch": 6.5484247374562425,
      "grad_norm": 0.10546875,
      "learning_rate": 5.3253061035388495e-05,
      "loss": 0.0028,
      "step": 14030
    },
    {
      "epoch": 6.548891481913652,
      "grad_norm": 0.22265625,
      "learning_rate": 5.324009004409427e-05,
      "loss": 0.001,
      "step": 14031
    },
    {
      "epoch": 6.549358226371062,
      "grad_norm": 0.1396484375,
      "learning_rate": 5.322712005959276e-05,
      "loss": 0.0006,
      "step": 14032
    },
    {
      "epoch": 6.549824970828471,
      "grad_norm": 0.24609375,
      "learning_rate": 5.3214151082163255e-05,
      "loss": 0.001,
      "step": 14033
    },
    {
      "epoch": 6.550291715285881,
      "grad_norm": 0.036865234375,
      "learning_rate": 5.320118311208502e-05,
      "loss": 0.0025,
      "step": 14034
    },
    {
      "epoch": 6.550758459743291,
      "grad_norm": 0.048095703125,
      "learning_rate": 5.3188216149637204e-05,
      "loss": 0.0004,
      "step": 14035
    },
    {
      "epoch": 6.5512252042007,
      "grad_norm": 0.169921875,
      "learning_rate": 5.3175250195099025e-05,
      "loss": 0.0007,
      "step": 14036
    },
    {
      "epoch": 6.551691948658109,
      "grad_norm": 0.018310546875,
      "learning_rate": 5.316228524874974e-05,
      "loss": 0.0003,
      "step": 14037
    },
    {
      "epoch": 6.552158693115519,
      "grad_norm": 0.033447265625,
      "learning_rate": 5.314932131086835e-05,
      "loss": 0.0004,
      "step": 14038
    },
    {
      "epoch": 6.552625437572929,
      "grad_norm": 0.306640625,
      "learning_rate": 5.313635838173405e-05,
      "loss": 0.0038,
      "step": 14039
    },
    {
      "epoch": 6.553092182030339,
      "grad_norm": 0.08203125,
      "learning_rate": 5.3123396461625986e-05,
      "loss": 0.001,
      "step": 14040
    },
    {
      "epoch": 6.553558926487748,
      "grad_norm": 0.091796875,
      "learning_rate": 5.3110435550823176e-05,
      "loss": 0.0033,
      "step": 14041
    },
    {
      "epoch": 6.554025670945157,
      "grad_norm": 0.0260009765625,
      "learning_rate": 5.30974756496047e-05,
      "loss": 0.0003,
      "step": 14042
    },
    {
      "epoch": 6.554492415402567,
      "grad_norm": 0.134765625,
      "learning_rate": 5.308451675824966e-05,
      "loss": 0.0026,
      "step": 14043
    },
    {
      "epoch": 6.554959159859977,
      "grad_norm": 0.259765625,
      "learning_rate": 5.307155887703699e-05,
      "loss": 0.0036,
      "step": 14044
    },
    {
      "epoch": 6.555425904317386,
      "grad_norm": 0.251953125,
      "learning_rate": 5.305860200624573e-05,
      "loss": 0.0008,
      "step": 14045
    },
    {
      "epoch": 6.555892648774796,
      "grad_norm": 0.166015625,
      "learning_rate": 5.304564614615488e-05,
      "loss": 0.0008,
      "step": 14046
    },
    {
      "epoch": 6.5563593932322055,
      "grad_norm": 0.171875,
      "learning_rate": 5.3032691297043335e-05,
      "loss": 0.004,
      "step": 14047
    },
    {
      "epoch": 6.556826137689615,
      "grad_norm": 0.01513671875,
      "learning_rate": 5.301973745919004e-05,
      "loss": 0.0002,
      "step": 14048
    },
    {
      "epoch": 6.557292882147024,
      "grad_norm": 0.357421875,
      "learning_rate": 5.300678463287399e-05,
      "loss": 0.002,
      "step": 14049
    },
    {
      "epoch": 6.557759626604434,
      "grad_norm": 0.0849609375,
      "learning_rate": 5.2993832818373935e-05,
      "loss": 0.0005,
      "step": 14050
    },
    {
      "epoch": 6.558226371061844,
      "grad_norm": 0.0625,
      "learning_rate": 5.298088201596884e-05,
      "loss": 0.0006,
      "step": 14051
    },
    {
      "epoch": 6.558693115519253,
      "grad_norm": 0.185546875,
      "learning_rate": 5.296793222593755e-05,
      "loss": 0.001,
      "step": 14052
    },
    {
      "epoch": 6.5591598599766625,
      "grad_norm": 0.14453125,
      "learning_rate": 5.2954983448558824e-05,
      "loss": 0.0008,
      "step": 14053
    },
    {
      "epoch": 6.559626604434072,
      "grad_norm": 0.06494140625,
      "learning_rate": 5.294203568411151e-05,
      "loss": 0.0032,
      "step": 14054
    },
    {
      "epoch": 6.560093348891482,
      "grad_norm": 0.2001953125,
      "learning_rate": 5.2929088932874405e-05,
      "loss": 0.0011,
      "step": 14055
    },
    {
      "epoch": 6.560560093348892,
      "grad_norm": 0.0673828125,
      "learning_rate": 5.2916143195126214e-05,
      "loss": 0.0006,
      "step": 14056
    },
    {
      "epoch": 6.561026837806301,
      "grad_norm": 0.28125,
      "learning_rate": 5.290319847114573e-05,
      "loss": 0.0019,
      "step": 14057
    },
    {
      "epoch": 6.561493582263711,
      "grad_norm": 0.1337890625,
      "learning_rate": 5.289025476121161e-05,
      "loss": 0.0009,
      "step": 14058
    },
    {
      "epoch": 6.56196032672112,
      "grad_norm": 0.08740234375,
      "learning_rate": 5.287731206560257e-05,
      "loss": 0.003,
      "step": 14059
    },
    {
      "epoch": 6.562427071178529,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.286437038459733e-05,
      "loss": 0.0006,
      "step": 14060
    },
    {
      "epoch": 6.562893815635939,
      "grad_norm": 0.13671875,
      "learning_rate": 5.285142971847445e-05,
      "loss": 0.0008,
      "step": 14061
    },
    {
      "epoch": 6.563360560093349,
      "grad_norm": 0.408203125,
      "learning_rate": 5.283849006751258e-05,
      "loss": 0.0017,
      "step": 14062
    },
    {
      "epoch": 6.563827304550759,
      "grad_norm": 0.08447265625,
      "learning_rate": 5.282555143199042e-05,
      "loss": 0.0005,
      "step": 14063
    },
    {
      "epoch": 6.5642940490081685,
      "grad_norm": 0.0810546875,
      "learning_rate": 5.2812613812186404e-05,
      "loss": 0.0006,
      "step": 14064
    },
    {
      "epoch": 6.564760793465577,
      "grad_norm": 0.0311279296875,
      "learning_rate": 5.27996772083792e-05,
      "loss": 0.0004,
      "step": 14065
    },
    {
      "epoch": 6.565227537922987,
      "grad_norm": 0.0634765625,
      "learning_rate": 5.278674162084733e-05,
      "loss": 0.0005,
      "step": 14066
    },
    {
      "epoch": 6.565694282380397,
      "grad_norm": 0.1708984375,
      "learning_rate": 5.2773807049869273e-05,
      "loss": 0.0029,
      "step": 14067
    },
    {
      "epoch": 6.566161026837806,
      "grad_norm": 0.06982421875,
      "learning_rate": 5.2760873495723536e-05,
      "loss": 0.0053,
      "step": 14068
    },
    {
      "epoch": 6.566627771295216,
      "grad_norm": 0.28125,
      "learning_rate": 5.274794095868866e-05,
      "loss": 0.002,
      "step": 14069
    },
    {
      "epoch": 6.5670945157526255,
      "grad_norm": 0.10400390625,
      "learning_rate": 5.2735009439042995e-05,
      "loss": 0.0007,
      "step": 14070
    },
    {
      "epoch": 6.567561260210035,
      "grad_norm": 0.0751953125,
      "learning_rate": 5.272207893706501e-05,
      "loss": 0.0006,
      "step": 14071
    },
    {
      "epoch": 6.568028004667444,
      "grad_norm": 0.2314453125,
      "learning_rate": 5.270914945303317e-05,
      "loss": 0.0008,
      "step": 14072
    },
    {
      "epoch": 6.568494749124854,
      "grad_norm": 0.8125,
      "learning_rate": 5.269622098722578e-05,
      "loss": 0.0086,
      "step": 14073
    },
    {
      "epoch": 6.568961493582264,
      "grad_norm": 0.140625,
      "learning_rate": 5.268329353992122e-05,
      "loss": 0.0049,
      "step": 14074
    },
    {
      "epoch": 6.569428238039674,
      "grad_norm": 0.1435546875,
      "learning_rate": 5.2670367111397903e-05,
      "loss": 0.0027,
      "step": 14075
    },
    {
      "epoch": 6.5698949824970825,
      "grad_norm": 0.058349609375,
      "learning_rate": 5.265744170193404e-05,
      "loss": 0.0004,
      "step": 14076
    },
    {
      "epoch": 6.570361726954492,
      "grad_norm": 0.09130859375,
      "learning_rate": 5.264451731180804e-05,
      "loss": 0.0006,
      "step": 14077
    },
    {
      "epoch": 6.570828471411902,
      "grad_norm": 0.037841796875,
      "learning_rate": 5.263159394129807e-05,
      "loss": 0.0004,
      "step": 14078
    },
    {
      "epoch": 6.571295215869312,
      "grad_norm": 0.078125,
      "learning_rate": 5.2618671590682457e-05,
      "loss": 0.0004,
      "step": 14079
    },
    {
      "epoch": 6.571761960326721,
      "grad_norm": 0.35546875,
      "learning_rate": 5.260575026023944e-05,
      "loss": 0.0012,
      "step": 14080
    },
    {
      "epoch": 6.572228704784131,
      "grad_norm": 0.11474609375,
      "learning_rate": 5.2592829950247166e-05,
      "loss": 0.0006,
      "step": 14081
    },
    {
      "epoch": 6.57269544924154,
      "grad_norm": 0.0830078125,
      "learning_rate": 5.257991066098387e-05,
      "loss": 0.0006,
      "step": 14082
    },
    {
      "epoch": 6.57316219369895,
      "grad_norm": 0.0595703125,
      "learning_rate": 5.2566992392727746e-05,
      "loss": 0.0003,
      "step": 14083
    },
    {
      "epoch": 6.573628938156359,
      "grad_norm": 0.375,
      "learning_rate": 5.255407514575685e-05,
      "loss": 0.0035,
      "step": 14084
    },
    {
      "epoch": 6.574095682613769,
      "grad_norm": 0.14453125,
      "learning_rate": 5.254115892034942e-05,
      "loss": 0.0027,
      "step": 14085
    },
    {
      "epoch": 6.574562427071179,
      "grad_norm": 0.0230712890625,
      "learning_rate": 5.2528243716783445e-05,
      "loss": 0.0003,
      "step": 14086
    },
    {
      "epoch": 6.5750291715285885,
      "grad_norm": 0.2216796875,
      "learning_rate": 5.2515329535337096e-05,
      "loss": 0.0034,
      "step": 14087
    },
    {
      "epoch": 6.575495915985997,
      "grad_norm": 0.166015625,
      "learning_rate": 5.250241637628835e-05,
      "loss": 0.001,
      "step": 14088
    },
    {
      "epoch": 6.575962660443407,
      "grad_norm": 0.0712890625,
      "learning_rate": 5.2489504239915323e-05,
      "loss": 0.0006,
      "step": 14089
    },
    {
      "epoch": 6.576429404900817,
      "grad_norm": 0.181640625,
      "learning_rate": 5.2476593126495933e-05,
      "loss": 0.0009,
      "step": 14090
    },
    {
      "epoch": 6.576896149358227,
      "grad_norm": 0.041748046875,
      "learning_rate": 5.246368303630822e-05,
      "loss": 0.0003,
      "step": 14091
    },
    {
      "epoch": 6.577362893815636,
      "grad_norm": 0.04541015625,
      "learning_rate": 5.245077396963021e-05,
      "loss": 0.0004,
      "step": 14092
    },
    {
      "epoch": 6.5778296382730455,
      "grad_norm": 0.10693359375,
      "learning_rate": 5.2437865926739736e-05,
      "loss": 0.0006,
      "step": 14093
    },
    {
      "epoch": 6.578296382730455,
      "grad_norm": 0.048095703125,
      "learning_rate": 5.242495890791478e-05,
      "loss": 0.0004,
      "step": 14094
    },
    {
      "epoch": 6.578763127187864,
      "grad_norm": 0.07666015625,
      "learning_rate": 5.24120529134333e-05,
      "loss": 0.0057,
      "step": 14095
    },
    {
      "epoch": 6.579229871645274,
      "grad_norm": 0.052490234375,
      "learning_rate": 5.239914794357306e-05,
      "loss": 0.0003,
      "step": 14096
    },
    {
      "epoch": 6.579696616102684,
      "grad_norm": 0.146484375,
      "learning_rate": 5.238624399861198e-05,
      "loss": 0.0006,
      "step": 14097
    },
    {
      "epoch": 6.5801633605600935,
      "grad_norm": 0.251953125,
      "learning_rate": 5.237334107882795e-05,
      "loss": 0.0015,
      "step": 14098
    },
    {
      "epoch": 6.580630105017503,
      "grad_norm": 0.150390625,
      "learning_rate": 5.236043918449869e-05,
      "loss": 0.0005,
      "step": 14099
    },
    {
      "epoch": 6.581096849474912,
      "grad_norm": 0.107421875,
      "learning_rate": 5.2347538315902026e-05,
      "loss": 0.0006,
      "step": 14100
    },
    {
      "epoch": 6.581563593932322,
      "grad_norm": 0.21484375,
      "learning_rate": 5.2334638473315766e-05,
      "loss": 0.002,
      "step": 14101
    },
    {
      "epoch": 6.582030338389732,
      "grad_norm": 0.080078125,
      "learning_rate": 5.232173965701759e-05,
      "loss": 0.0036,
      "step": 14102
    },
    {
      "epoch": 6.582497082847141,
      "grad_norm": 0.251953125,
      "learning_rate": 5.230884186728526e-05,
      "loss": 0.0009,
      "step": 14103
    },
    {
      "epoch": 6.5829638273045505,
      "grad_norm": 0.12890625,
      "learning_rate": 5.2295945104396524e-05,
      "loss": 0.0007,
      "step": 14104
    },
    {
      "epoch": 6.58343057176196,
      "grad_norm": 0.11572265625,
      "learning_rate": 5.2283049368628964e-05,
      "loss": 0.0006,
      "step": 14105
    },
    {
      "epoch": 6.58389731621937,
      "grad_norm": 0.2734375,
      "learning_rate": 5.227015466026035e-05,
      "loss": 0.0011,
      "step": 14106
    },
    {
      "epoch": 6.58436406067678,
      "grad_norm": 0.18359375,
      "learning_rate": 5.225726097956822e-05,
      "loss": 0.0009,
      "step": 14107
    },
    {
      "epoch": 6.584830805134189,
      "grad_norm": 0.376953125,
      "learning_rate": 5.224436832683022e-05,
      "loss": 0.0019,
      "step": 14108
    },
    {
      "epoch": 6.585297549591599,
      "grad_norm": 0.09619140625,
      "learning_rate": 5.223147670232401e-05,
      "loss": 0.0005,
      "step": 14109
    },
    {
      "epoch": 6.585764294049008,
      "grad_norm": 0.12060546875,
      "learning_rate": 5.221858610632706e-05,
      "loss": 0.0006,
      "step": 14110
    },
    {
      "epoch": 6.586231038506417,
      "grad_norm": 0.07568359375,
      "learning_rate": 5.220569653911698e-05,
      "loss": 0.0005,
      "step": 14111
    },
    {
      "epoch": 6.586697782963827,
      "grad_norm": 0.099609375,
      "learning_rate": 5.2192808000971304e-05,
      "loss": 0.0004,
      "step": 14112
    },
    {
      "epoch": 6.587164527421237,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.2179920492167464e-05,
      "loss": 0.0007,
      "step": 14113
    },
    {
      "epoch": 6.587631271878647,
      "grad_norm": 0.19140625,
      "learning_rate": 5.216703401298302e-05,
      "loss": 0.0011,
      "step": 14114
    },
    {
      "epoch": 6.588098016336056,
      "grad_norm": 0.248046875,
      "learning_rate": 5.2154148563695426e-05,
      "loss": 0.0014,
      "step": 14115
    },
    {
      "epoch": 6.588564760793465,
      "grad_norm": 0.1806640625,
      "learning_rate": 5.214126414458207e-05,
      "loss": 0.0009,
      "step": 14116
    },
    {
      "epoch": 6.589031505250875,
      "grad_norm": 0.0186767578125,
      "learning_rate": 5.2128380755920404e-05,
      "loss": 0.0003,
      "step": 14117
    },
    {
      "epoch": 6.589498249708285,
      "grad_norm": 0.138671875,
      "learning_rate": 5.2115498397987836e-05,
      "loss": 0.0006,
      "step": 14118
    },
    {
      "epoch": 6.589964994165694,
      "grad_norm": 0.1474609375,
      "learning_rate": 5.21026170710617e-05,
      "loss": 0.0007,
      "step": 14119
    },
    {
      "epoch": 6.590431738623104,
      "grad_norm": 0.205078125,
      "learning_rate": 5.208973677541934e-05,
      "loss": 0.001,
      "step": 14120
    },
    {
      "epoch": 6.5908984830805135,
      "grad_norm": 0.109375,
      "learning_rate": 5.207685751133815e-05,
      "loss": 0.0004,
      "step": 14121
    },
    {
      "epoch": 6.591365227537923,
      "grad_norm": 0.08203125,
      "learning_rate": 5.2063979279095355e-05,
      "loss": 0.0006,
      "step": 14122
    },
    {
      "epoch": 6.591831971995332,
      "grad_norm": 0.283203125,
      "learning_rate": 5.205110207896827e-05,
      "loss": 0.002,
      "step": 14123
    },
    {
      "epoch": 6.592298716452742,
      "grad_norm": 0.08740234375,
      "learning_rate": 5.20382259112342e-05,
      "loss": 0.0005,
      "step": 14124
    },
    {
      "epoch": 6.592765460910152,
      "grad_norm": 0.1962890625,
      "learning_rate": 5.20253507761703e-05,
      "loss": 0.001,
      "step": 14125
    },
    {
      "epoch": 6.593232205367562,
      "grad_norm": 0.037353515625,
      "learning_rate": 5.2012476674053856e-05,
      "loss": 0.0004,
      "step": 14126
    },
    {
      "epoch": 6.5936989498249705,
      "grad_norm": 0.255859375,
      "learning_rate": 5.199960360516201e-05,
      "loss": 0.0011,
      "step": 14127
    },
    {
      "epoch": 6.59416569428238,
      "grad_norm": 0.06787109375,
      "learning_rate": 5.198673156977194e-05,
      "loss": 0.0004,
      "step": 14128
    },
    {
      "epoch": 6.59463243873979,
      "grad_norm": 0.02880859375,
      "learning_rate": 5.1973860568160845e-05,
      "loss": 0.0019,
      "step": 14129
    },
    {
      "epoch": 6.5950991831972,
      "grad_norm": 0.1875,
      "learning_rate": 5.19609906006058e-05,
      "loss": 0.001,
      "step": 14130
    },
    {
      "epoch": 6.595565927654609,
      "grad_norm": 0.515625,
      "learning_rate": 5.194812166738392e-05,
      "loss": 0.0022,
      "step": 14131
    },
    {
      "epoch": 6.596032672112019,
      "grad_norm": 0.0478515625,
      "learning_rate": 5.193525376877233e-05,
      "loss": 0.0004,
      "step": 14132
    },
    {
      "epoch": 6.596499416569428,
      "grad_norm": 0.064453125,
      "learning_rate": 5.192238690504805e-05,
      "loss": 0.0023,
      "step": 14133
    },
    {
      "epoch": 6.596966161026838,
      "grad_norm": 0.03173828125,
      "learning_rate": 5.19095210764881e-05,
      "loss": 0.0003,
      "step": 14134
    },
    {
      "epoch": 6.597432905484247,
      "grad_norm": 0.1572265625,
      "learning_rate": 5.1896656283369506e-05,
      "loss": 0.0004,
      "step": 14135
    },
    {
      "epoch": 6.597899649941657,
      "grad_norm": 0.103515625,
      "learning_rate": 5.188379252596931e-05,
      "loss": 0.0041,
      "step": 14136
    },
    {
      "epoch": 6.598366394399067,
      "grad_norm": 0.0712890625,
      "learning_rate": 5.187092980456441e-05,
      "loss": 0.0005,
      "step": 14137
    },
    {
      "epoch": 6.598833138856476,
      "grad_norm": 0.0703125,
      "learning_rate": 5.185806811943184e-05,
      "loss": 0.0042,
      "step": 14138
    },
    {
      "epoch": 6.599299883313885,
      "grad_norm": 0.0966796875,
      "learning_rate": 5.184520747084842e-05,
      "loss": 0.0004,
      "step": 14139
    },
    {
      "epoch": 6.599766627771295,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.1832347859091136e-05,
      "loss": 0.007,
      "step": 14140
    },
    {
      "epoch": 6.600233372228705,
      "grad_norm": 0.255859375,
      "learning_rate": 5.181948928443687e-05,
      "loss": 0.0013,
      "step": 14141
    },
    {
      "epoch": 6.600700116686115,
      "grad_norm": 0.193359375,
      "learning_rate": 5.180663174716243e-05,
      "loss": 0.0033,
      "step": 14142
    },
    {
      "epoch": 6.601166861143524,
      "grad_norm": 0.255859375,
      "learning_rate": 5.1793775247544664e-05,
      "loss": 0.0034,
      "step": 14143
    },
    {
      "epoch": 6.6016336056009335,
      "grad_norm": 0.30078125,
      "learning_rate": 5.178091978586045e-05,
      "loss": 0.0017,
      "step": 14144
    },
    {
      "epoch": 6.602100350058343,
      "grad_norm": 0.057373046875,
      "learning_rate": 5.17680653623865e-05,
      "loss": 0.0021,
      "step": 14145
    },
    {
      "epoch": 6.602567094515752,
      "grad_norm": 0.06103515625,
      "learning_rate": 5.17552119773996e-05,
      "loss": 0.0004,
      "step": 14146
    },
    {
      "epoch": 6.603033838973162,
      "grad_norm": 0.011474609375,
      "learning_rate": 5.1742359631176564e-05,
      "loss": 0.0002,
      "step": 14147
    },
    {
      "epoch": 6.603500583430572,
      "grad_norm": 0.185546875,
      "learning_rate": 5.172950832399403e-05,
      "loss": 0.0006,
      "step": 14148
    },
    {
      "epoch": 6.603967327887982,
      "grad_norm": 0.2578125,
      "learning_rate": 5.171665805612873e-05,
      "loss": 0.0014,
      "step": 14149
    },
    {
      "epoch": 6.604434072345391,
      "grad_norm": 0.09375,
      "learning_rate": 5.17038088278574e-05,
      "loss": 0.0048,
      "step": 14150
    },
    {
      "epoch": 6.6049008168028,
      "grad_norm": 0.08056640625,
      "learning_rate": 5.169096063945662e-05,
      "loss": 0.0053,
      "step": 14151
    },
    {
      "epoch": 6.60536756126021,
      "grad_norm": 0.21484375,
      "learning_rate": 5.167811349120304e-05,
      "loss": 0.0021,
      "step": 14152
    },
    {
      "epoch": 6.60583430571762,
      "grad_norm": 0.07958984375,
      "learning_rate": 5.166526738337334e-05,
      "loss": 0.003,
      "step": 14153
    },
    {
      "epoch": 6.606301050175029,
      "grad_norm": 0.1103515625,
      "learning_rate": 5.165242231624402e-05,
      "loss": 0.0007,
      "step": 14154
    },
    {
      "epoch": 6.606767794632439,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.163957829009174e-05,
      "loss": 0.0006,
      "step": 14155
    },
    {
      "epoch": 6.607234539089848,
      "grad_norm": 0.050537109375,
      "learning_rate": 5.1626735305192944e-05,
      "loss": 0.0004,
      "step": 14156
    },
    {
      "epoch": 6.607701283547258,
      "grad_norm": 0.123046875,
      "learning_rate": 5.161389336182421e-05,
      "loss": 0.0006,
      "step": 14157
    },
    {
      "epoch": 6.608168028004667,
      "grad_norm": 0.1884765625,
      "learning_rate": 5.1601052460262076e-05,
      "loss": 0.0016,
      "step": 14158
    },
    {
      "epoch": 6.608634772462077,
      "grad_norm": 0.036865234375,
      "learning_rate": 5.158821260078295e-05,
      "loss": 0.0004,
      "step": 14159
    },
    {
      "epoch": 6.609101516919487,
      "grad_norm": 0.047119140625,
      "learning_rate": 5.15753737836633e-05,
      "loss": 0.0003,
      "step": 14160
    },
    {
      "epoch": 6.6095682613768965,
      "grad_norm": 0.07275390625,
      "learning_rate": 5.156253600917964e-05,
      "loss": 0.0006,
      "step": 14161
    },
    {
      "epoch": 6.610035005834305,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.154969927760829e-05,
      "loss": 0.0005,
      "step": 14162
    },
    {
      "epoch": 6.610501750291715,
      "grad_norm": 0.045166015625,
      "learning_rate": 5.1536863589225646e-05,
      "loss": 0.0028,
      "step": 14163
    },
    {
      "epoch": 6.610968494749125,
      "grad_norm": 0.0235595703125,
      "learning_rate": 5.1524028944308156e-05,
      "loss": 0.0003,
      "step": 14164
    },
    {
      "epoch": 6.611435239206535,
      "grad_norm": 0.03076171875,
      "learning_rate": 5.1511195343132066e-05,
      "loss": 0.0004,
      "step": 14165
    },
    {
      "epoch": 6.611901983663944,
      "grad_norm": 0.041015625,
      "learning_rate": 5.149836278597373e-05,
      "loss": 0.0004,
      "step": 14166
    },
    {
      "epoch": 6.6123687281213535,
      "grad_norm": 0.10205078125,
      "learning_rate": 5.148553127310951e-05,
      "loss": 0.0007,
      "step": 14167
    },
    {
      "epoch": 6.612835472578763,
      "grad_norm": 0.115234375,
      "learning_rate": 5.147270080481559e-05,
      "loss": 0.0009,
      "step": 14168
    },
    {
      "epoch": 6.613302217036173,
      "grad_norm": 0.026123046875,
      "learning_rate": 5.145987138136825e-05,
      "loss": 0.0002,
      "step": 14169
    },
    {
      "epoch": 6.613768961493582,
      "grad_norm": 0.0279541015625,
      "learning_rate": 5.1447043003043784e-05,
      "loss": 0.0003,
      "step": 14170
    },
    {
      "epoch": 6.614235705950992,
      "grad_norm": 0.103515625,
      "learning_rate": 5.1434215670118316e-05,
      "loss": 0.0006,
      "step": 14171
    },
    {
      "epoch": 6.6147024504084015,
      "grad_norm": 0.0703125,
      "learning_rate": 5.142138938286806e-05,
      "loss": 0.0004,
      "step": 14172
    },
    {
      "epoch": 6.615169194865811,
      "grad_norm": 0.0732421875,
      "learning_rate": 5.140856414156923e-05,
      "loss": 0.0003,
      "step": 14173
    },
    {
      "epoch": 6.61563593932322,
      "grad_norm": 0.1787109375,
      "learning_rate": 5.139573994649789e-05,
      "loss": 0.0008,
      "step": 14174
    },
    {
      "epoch": 6.61610268378063,
      "grad_norm": 0.2255859375,
      "learning_rate": 5.138291679793023e-05,
      "loss": 0.0019,
      "step": 14175
    },
    {
      "epoch": 6.61656942823804,
      "grad_norm": 0.490234375,
      "learning_rate": 5.137009469614229e-05,
      "loss": 0.0027,
      "step": 14176
    },
    {
      "epoch": 6.61703617269545,
      "grad_norm": 0.232421875,
      "learning_rate": 5.1357273641410166e-05,
      "loss": 0.0013,
      "step": 14177
    },
    {
      "epoch": 6.6175029171528585,
      "grad_norm": 0.049560546875,
      "learning_rate": 5.1344453634009935e-05,
      "loss": 0.0019,
      "step": 14178
    },
    {
      "epoch": 6.617969661610268,
      "grad_norm": 0.019287109375,
      "learning_rate": 5.133163467421758e-05,
      "loss": 0.0003,
      "step": 14179
    },
    {
      "epoch": 6.618436406067678,
      "grad_norm": 0.251953125,
      "learning_rate": 5.131881676230914e-05,
      "loss": 0.0038,
      "step": 14180
    },
    {
      "epoch": 6.618903150525087,
      "grad_norm": 0.06005859375,
      "learning_rate": 5.1305999898560574e-05,
      "loss": 0.0005,
      "step": 14181
    },
    {
      "epoch": 6.619369894982497,
      "grad_norm": 0.0791015625,
      "learning_rate": 5.1293184083247884e-05,
      "loss": 0.0005,
      "step": 14182
    },
    {
      "epoch": 6.619836639439907,
      "grad_norm": 0.2060546875,
      "learning_rate": 5.128036931664695e-05,
      "loss": 0.0046,
      "step": 14183
    },
    {
      "epoch": 6.620303383897316,
      "grad_norm": 0.03857421875,
      "learning_rate": 5.1267555599033715e-05,
      "loss": 0.0004,
      "step": 14184
    },
    {
      "epoch": 6.620770128354726,
      "grad_norm": 0.059814453125,
      "learning_rate": 5.1254742930684116e-05,
      "loss": 0.0026,
      "step": 14185
    },
    {
      "epoch": 6.621236872812135,
      "grad_norm": 0.06689453125,
      "learning_rate": 5.1241931311873946e-05,
      "loss": 0.0004,
      "step": 14186
    },
    {
      "epoch": 6.621703617269545,
      "grad_norm": 0.06640625,
      "learning_rate": 5.1229120742879135e-05,
      "loss": 0.0004,
      "step": 14187
    },
    {
      "epoch": 6.622170361726955,
      "grad_norm": 0.038818359375,
      "learning_rate": 5.121631122397543e-05,
      "loss": 0.0003,
      "step": 14188
    },
    {
      "epoch": 6.622637106184364,
      "grad_norm": 0.12109375,
      "learning_rate": 5.120350275543866e-05,
      "loss": 0.0006,
      "step": 14189
    },
    {
      "epoch": 6.623103850641773,
      "grad_norm": 0.01385498046875,
      "learning_rate": 5.1190695337544666e-05,
      "loss": 0.0003,
      "step": 14190
    },
    {
      "epoch": 6.623570595099183,
      "grad_norm": 0.2412109375,
      "learning_rate": 5.117788897056912e-05,
      "loss": 0.0012,
      "step": 14191
    },
    {
      "epoch": 6.624037339556593,
      "grad_norm": 0.0272216796875,
      "learning_rate": 5.1165083654787784e-05,
      "loss": 0.0003,
      "step": 14192
    },
    {
      "epoch": 6.624504084014003,
      "grad_norm": 0.037109375,
      "learning_rate": 5.115227939047642e-05,
      "loss": 0.0003,
      "step": 14193
    },
    {
      "epoch": 6.624970828471412,
      "grad_norm": 0.1826171875,
      "learning_rate": 5.113947617791065e-05,
      "loss": 0.0008,
      "step": 14194
    },
    {
      "epoch": 6.6254375729288215,
      "grad_norm": 0.111328125,
      "learning_rate": 5.112667401736616e-05,
      "loss": 0.0005,
      "step": 14195
    },
    {
      "epoch": 6.625904317386231,
      "grad_norm": 0.046142578125,
      "learning_rate": 5.111387290911864e-05,
      "loss": 0.0003,
      "step": 14196
    },
    {
      "epoch": 6.62637106184364,
      "grad_norm": 0.373046875,
      "learning_rate": 5.110107285344365e-05,
      "loss": 0.003,
      "step": 14197
    },
    {
      "epoch": 6.62683780630105,
      "grad_norm": 0.068359375,
      "learning_rate": 5.108827385061681e-05,
      "loss": 0.0003,
      "step": 14198
    },
    {
      "epoch": 6.62730455075846,
      "grad_norm": 0.0498046875,
      "learning_rate": 5.107547590091373e-05,
      "loss": 0.0004,
      "step": 14199
    },
    {
      "epoch": 6.62777129521587,
      "grad_norm": 0.09716796875,
      "learning_rate": 5.1062679004609917e-05,
      "loss": 0.0022,
      "step": 14200
    },
    {
      "epoch": 6.6282380396732785,
      "grad_norm": 0.0732421875,
      "learning_rate": 5.104988316198092e-05,
      "loss": 0.0046,
      "step": 14201
    },
    {
      "epoch": 6.628704784130688,
      "grad_norm": 0.09326171875,
      "learning_rate": 5.103708837330227e-05,
      "loss": 0.0006,
      "step": 14202
    },
    {
      "epoch": 6.629171528588098,
      "grad_norm": 0.1650390625,
      "learning_rate": 5.10242946388494e-05,
      "loss": 0.0033,
      "step": 14203
    },
    {
      "epoch": 6.629638273045508,
      "grad_norm": 0.54296875,
      "learning_rate": 5.101150195889781e-05,
      "loss": 0.0009,
      "step": 14204
    },
    {
      "epoch": 6.630105017502917,
      "grad_norm": 0.033447265625,
      "learning_rate": 5.099871033372299e-05,
      "loss": 0.0003,
      "step": 14205
    },
    {
      "epoch": 6.630571761960327,
      "grad_norm": 0.1220703125,
      "learning_rate": 5.0985919763600244e-05,
      "loss": 0.0005,
      "step": 14206
    },
    {
      "epoch": 6.631038506417736,
      "grad_norm": 0.20703125,
      "learning_rate": 5.0973130248805076e-05,
      "loss": 0.0093,
      "step": 14207
    },
    {
      "epoch": 6.631505250875146,
      "grad_norm": 0.3125,
      "learning_rate": 5.0960341789612766e-05,
      "loss": 0.0019,
      "step": 14208
    },
    {
      "epoch": 6.631971995332555,
      "grad_norm": 0.294921875,
      "learning_rate": 5.09475543862987e-05,
      "loss": 0.001,
      "step": 14209
    },
    {
      "epoch": 6.632438739789965,
      "grad_norm": 0.1689453125,
      "learning_rate": 5.093476803913827e-05,
      "loss": 0.0006,
      "step": 14210
    },
    {
      "epoch": 6.632905484247375,
      "grad_norm": 0.1494140625,
      "learning_rate": 5.092198274840667e-05,
      "loss": 0.0051,
      "step": 14211
    },
    {
      "epoch": 6.6333722287047845,
      "grad_norm": 0.244140625,
      "learning_rate": 5.0909198514379255e-05,
      "loss": 0.0007,
      "step": 14212
    },
    {
      "epoch": 6.633838973162193,
      "grad_norm": 0.076171875,
      "learning_rate": 5.089641533733128e-05,
      "loss": 0.0005,
      "step": 14213
    },
    {
      "epoch": 6.634305717619603,
      "grad_norm": 0.058837890625,
      "learning_rate": 5.088363321753793e-05,
      "loss": 0.0004,
      "step": 14214
    },
    {
      "epoch": 6.634772462077013,
      "grad_norm": 0.03662109375,
      "learning_rate": 5.087085215527445e-05,
      "loss": 0.0003,
      "step": 14215
    },
    {
      "epoch": 6.635239206534423,
      "grad_norm": 0.041259765625,
      "learning_rate": 5.085807215081608e-05,
      "loss": 0.002,
      "step": 14216
    },
    {
      "epoch": 6.635705950991832,
      "grad_norm": 0.1318359375,
      "learning_rate": 5.084529320443791e-05,
      "loss": 0.0009,
      "step": 14217
    },
    {
      "epoch": 6.6361726954492415,
      "grad_norm": 0.0849609375,
      "learning_rate": 5.08325153164151e-05,
      "loss": 0.0004,
      "step": 14218
    },
    {
      "epoch": 6.636639439906651,
      "grad_norm": 0.09033203125,
      "learning_rate": 5.0819738487022824e-05,
      "loss": 0.0005,
      "step": 14219
    },
    {
      "epoch": 6.637106184364061,
      "grad_norm": 0.056640625,
      "learning_rate": 5.080696271653611e-05,
      "loss": 0.0004,
      "step": 14220
    },
    {
      "epoch": 6.63757292882147,
      "grad_norm": 0.029052734375,
      "learning_rate": 5.079418800523008e-05,
      "loss": 0.0004,
      "step": 14221
    },
    {
      "epoch": 6.63803967327888,
      "grad_norm": 0.45703125,
      "learning_rate": 5.078141435337981e-05,
      "loss": 0.0014,
      "step": 14222
    },
    {
      "epoch": 6.63850641773629,
      "grad_norm": 0.07177734375,
      "learning_rate": 5.076864176126026e-05,
      "loss": 0.0005,
      "step": 14223
    },
    {
      "epoch": 6.6389731621936985,
      "grad_norm": 0.5,
      "learning_rate": 5.0755870229146506e-05,
      "loss": 0.0093,
      "step": 14224
    },
    {
      "epoch": 6.639439906651108,
      "grad_norm": 0.047607421875,
      "learning_rate": 5.074309975731346e-05,
      "loss": 0.0005,
      "step": 14225
    },
    {
      "epoch": 6.639906651108518,
      "grad_norm": 0.08154296875,
      "learning_rate": 5.073033034603615e-05,
      "loss": 0.0004,
      "step": 14226
    },
    {
      "epoch": 6.640373395565928,
      "grad_norm": 0.154296875,
      "learning_rate": 5.0717561995589526e-05,
      "loss": 0.0031,
      "step": 14227
    },
    {
      "epoch": 6.640840140023338,
      "grad_norm": 0.1455078125,
      "learning_rate": 5.070479470624845e-05,
      "loss": 0.0006,
      "step": 14228
    },
    {
      "epoch": 6.641306884480747,
      "grad_norm": 0.19921875,
      "learning_rate": 5.0692028478287825e-05,
      "loss": 0.0017,
      "step": 14229
    },
    {
      "epoch": 6.641773628938156,
      "grad_norm": 0.166015625,
      "learning_rate": 5.0679263311982516e-05,
      "loss": 0.0007,
      "step": 14230
    },
    {
      "epoch": 6.642240373395566,
      "grad_norm": 0.384765625,
      "learning_rate": 5.066649920760744e-05,
      "loss": 0.0035,
      "step": 14231
    },
    {
      "epoch": 6.642707117852975,
      "grad_norm": 0.058349609375,
      "learning_rate": 5.0653736165437337e-05,
      "loss": 0.0004,
      "step": 14232
    },
    {
      "epoch": 6.643173862310385,
      "grad_norm": 0.2001953125,
      "learning_rate": 5.0640974185747035e-05,
      "loss": 0.0008,
      "step": 14233
    },
    {
      "epoch": 6.643640606767795,
      "grad_norm": 0.060546875,
      "learning_rate": 5.0628213268811374e-05,
      "loss": 0.0005,
      "step": 14234
    },
    {
      "epoch": 6.6441073512252045,
      "grad_norm": 0.13671875,
      "learning_rate": 5.061545341490501e-05,
      "loss": 0.0008,
      "step": 14235
    },
    {
      "epoch": 6.644574095682614,
      "grad_norm": 0.0179443359375,
      "learning_rate": 5.0602694624302774e-05,
      "loss": 0.0002,
      "step": 14236
    },
    {
      "epoch": 6.645040840140023,
      "grad_norm": 0.1865234375,
      "learning_rate": 5.058993689727929e-05,
      "loss": 0.0007,
      "step": 14237
    },
    {
      "epoch": 6.645507584597433,
      "grad_norm": 0.0703125,
      "learning_rate": 5.057718023410928e-05,
      "loss": 0.0003,
      "step": 14238
    },
    {
      "epoch": 6.645974329054843,
      "grad_norm": 0.049560546875,
      "learning_rate": 5.056442463506745e-05,
      "loss": 0.0004,
      "step": 14239
    },
    {
      "epoch": 6.646441073512252,
      "grad_norm": 0.0257568359375,
      "learning_rate": 5.0551670100428384e-05,
      "loss": 0.0003,
      "step": 14240
    },
    {
      "epoch": 6.6469078179696615,
      "grad_norm": 0.061279296875,
      "learning_rate": 5.0538916630466716e-05,
      "loss": 0.0044,
      "step": 14241
    },
    {
      "epoch": 6.647374562427071,
      "grad_norm": 0.376953125,
      "learning_rate": 5.052616422545708e-05,
      "loss": 0.0026,
      "step": 14242
    },
    {
      "epoch": 6.647841306884481,
      "grad_norm": 0.11376953125,
      "learning_rate": 5.0513412885674e-05,
      "loss": 0.0035,
      "step": 14243
    },
    {
      "epoch": 6.64830805134189,
      "grad_norm": 0.30859375,
      "learning_rate": 5.050066261139203e-05,
      "loss": 0.0017,
      "step": 14244
    },
    {
      "epoch": 6.6487747957993,
      "grad_norm": 0.0703125,
      "learning_rate": 5.048791340288577e-05,
      "loss": 0.0003,
      "step": 14245
    },
    {
      "epoch": 6.6492415402567095,
      "grad_norm": 0.04296875,
      "learning_rate": 5.047516526042962e-05,
      "loss": 0.0004,
      "step": 14246
    },
    {
      "epoch": 6.649708284714119,
      "grad_norm": 0.018798828125,
      "learning_rate": 5.046241818429811e-05,
      "loss": 0.0003,
      "step": 14247
    },
    {
      "epoch": 6.650175029171528,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.044967217476575e-05,
      "loss": 0.0005,
      "step": 14248
    },
    {
      "epoch": 6.650641773628938,
      "grad_norm": 0.220703125,
      "learning_rate": 5.043692723210687e-05,
      "loss": 0.0033,
      "step": 14249
    },
    {
      "epoch": 6.651108518086348,
      "grad_norm": 0.022216796875,
      "learning_rate": 5.0424183356595956e-05,
      "loss": 0.0003,
      "step": 14250
    },
    {
      "epoch": 6.651575262543758,
      "grad_norm": 0.058837890625,
      "learning_rate": 5.0411440548507406e-05,
      "loss": 0.0004,
      "step": 14251
    },
    {
      "epoch": 6.6520420070011665,
      "grad_norm": 0.0810546875,
      "learning_rate": 5.039869880811552e-05,
      "loss": 0.0005,
      "step": 14252
    },
    {
      "epoch": 6.652508751458576,
      "grad_norm": 0.1064453125,
      "learning_rate": 5.038595813569469e-05,
      "loss": 0.003,
      "step": 14253
    },
    {
      "epoch": 6.652975495915986,
      "grad_norm": 0.1796875,
      "learning_rate": 5.037321853151926e-05,
      "loss": 0.001,
      "step": 14254
    },
    {
      "epoch": 6.653442240373396,
      "grad_norm": 0.146484375,
      "learning_rate": 5.036047999586346e-05,
      "loss": 0.008,
      "step": 14255
    },
    {
      "epoch": 6.653908984830805,
      "grad_norm": 0.048828125,
      "learning_rate": 5.0347742529001655e-05,
      "loss": 0.0004,
      "step": 14256
    },
    {
      "epoch": 6.654375729288215,
      "grad_norm": 0.034423828125,
      "learning_rate": 5.033500613120799e-05,
      "loss": 0.0003,
      "step": 14257
    },
    {
      "epoch": 6.654842473745624,
      "grad_norm": 0.042236328125,
      "learning_rate": 5.0322270802756735e-05,
      "loss": 0.0003,
      "step": 14258
    },
    {
      "epoch": 6.655309218203034,
      "grad_norm": 0.478515625,
      "learning_rate": 5.030953654392216e-05,
      "loss": 0.0039,
      "step": 14259
    },
    {
      "epoch": 6.655775962660443,
      "grad_norm": 0.1318359375,
      "learning_rate": 5.029680335497837e-05,
      "loss": 0.0006,
      "step": 14260
    },
    {
      "epoch": 6.656242707117853,
      "grad_norm": 0.140625,
      "learning_rate": 5.028407123619954e-05,
      "loss": 0.0016,
      "step": 14261
    },
    {
      "epoch": 6.656709451575263,
      "grad_norm": 0.259765625,
      "learning_rate": 5.0271340187859836e-05,
      "loss": 0.0059,
      "step": 14262
    },
    {
      "epoch": 6.6571761960326725,
      "grad_norm": 0.10400390625,
      "learning_rate": 5.025861021023335e-05,
      "loss": 0.0005,
      "step": 14263
    },
    {
      "epoch": 6.657642940490081,
      "grad_norm": 0.11181640625,
      "learning_rate": 5.024588130359415e-05,
      "loss": 0.0004,
      "step": 14264
    },
    {
      "epoch": 6.658109684947491,
      "grad_norm": 0.03369140625,
      "learning_rate": 5.023315346821638e-05,
      "loss": 0.0003,
      "step": 14265
    },
    {
      "epoch": 6.658576429404901,
      "grad_norm": 0.06298828125,
      "learning_rate": 5.0220426704373994e-05,
      "loss": 0.0005,
      "step": 14266
    },
    {
      "epoch": 6.65904317386231,
      "grad_norm": 0.345703125,
      "learning_rate": 5.020770101234103e-05,
      "loss": 0.0011,
      "step": 14267
    },
    {
      "epoch": 6.65950991831972,
      "grad_norm": 0.337890625,
      "learning_rate": 5.019497639239157e-05,
      "loss": 0.005,
      "step": 14268
    },
    {
      "epoch": 6.6599766627771295,
      "grad_norm": 0.0927734375,
      "learning_rate": 5.0182252844799495e-05,
      "loss": 0.0023,
      "step": 14269
    },
    {
      "epoch": 6.660443407234539,
      "grad_norm": 0.045166015625,
      "learning_rate": 5.0169530369838766e-05,
      "loss": 0.0005,
      "step": 14270
    },
    {
      "epoch": 6.660910151691949,
      "grad_norm": 0.11181640625,
      "learning_rate": 5.0156808967783385e-05,
      "loss": 0.0006,
      "step": 14271
    },
    {
      "epoch": 6.661376896149358,
      "grad_norm": 0.07373046875,
      "learning_rate": 5.014408863890718e-05,
      "loss": 0.002,
      "step": 14272
    },
    {
      "epoch": 6.661843640606768,
      "grad_norm": 0.1064453125,
      "learning_rate": 5.013136938348409e-05,
      "loss": 0.003,
      "step": 14273
    },
    {
      "epoch": 6.662310385064178,
      "grad_norm": 0.412109375,
      "learning_rate": 5.011865120178791e-05,
      "loss": 0.0048,
      "step": 14274
    },
    {
      "epoch": 6.6627771295215865,
      "grad_norm": 0.03173828125,
      "learning_rate": 5.010593409409252e-05,
      "loss": 0.0004,
      "step": 14275
    },
    {
      "epoch": 6.663243873978996,
      "grad_norm": 0.0947265625,
      "learning_rate": 5.009321806067178e-05,
      "loss": 0.0058,
      "step": 14276
    },
    {
      "epoch": 6.663710618436406,
      "grad_norm": 0.12353515625,
      "learning_rate": 5.008050310179939e-05,
      "loss": 0.0006,
      "step": 14277
    },
    {
      "epoch": 6.664177362893816,
      "grad_norm": 0.0673828125,
      "learning_rate": 5.006778921774915e-05,
      "loss": 0.0005,
      "step": 14278
    },
    {
      "epoch": 6.664644107351226,
      "grad_norm": 0.0400390625,
      "learning_rate": 5.0055076408794786e-05,
      "loss": 0.0004,
      "step": 14279
    },
    {
      "epoch": 6.665110851808635,
      "grad_norm": 0.045166015625,
      "learning_rate": 5.004236467521009e-05,
      "loss": 0.0003,
      "step": 14280
    },
    {
      "epoch": 6.665577596266044,
      "grad_norm": 0.275390625,
      "learning_rate": 5.002965401726868e-05,
      "loss": 0.0013,
      "step": 14281
    },
    {
      "epoch": 6.666044340723454,
      "grad_norm": 0.046142578125,
      "learning_rate": 5.0016944435244254e-05,
      "loss": 0.0004,
      "step": 14282
    },
    {
      "epoch": 6.666511085180863,
      "grad_norm": 0.07373046875,
      "learning_rate": 5.0004235929410514e-05,
      "loss": 0.0004,
      "step": 14283
    },
    {
      "epoch": 6.666977829638273,
      "grad_norm": 0.064453125,
      "learning_rate": 4.999152850004101e-05,
      "loss": 0.0035,
      "step": 14284
    },
    {
      "epoch": 6.667444574095683,
      "grad_norm": 0.21875,
      "learning_rate": 4.997882214740941e-05,
      "loss": 0.0032,
      "step": 14285
    },
    {
      "epoch": 6.6679113185530925,
      "grad_norm": 0.040771484375,
      "learning_rate": 4.996611687178924e-05,
      "loss": 0.0004,
      "step": 14286
    },
    {
      "epoch": 6.668378063010501,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.995341267345408e-05,
      "loss": 0.001,
      "step": 14287
    },
    {
      "epoch": 6.668844807467911,
      "grad_norm": 0.038818359375,
      "learning_rate": 4.994070955267751e-05,
      "loss": 0.0003,
      "step": 14288
    },
    {
      "epoch": 6.669311551925321,
      "grad_norm": 0.05419921875,
      "learning_rate": 4.9928007509732975e-05,
      "loss": 0.0004,
      "step": 14289
    },
    {
      "epoch": 6.669778296382731,
      "grad_norm": 0.0888671875,
      "learning_rate": 4.9915306544893994e-05,
      "loss": 0.0005,
      "step": 14290
    },
    {
      "epoch": 6.67024504084014,
      "grad_norm": 0.11669921875,
      "learning_rate": 4.990260665843406e-05,
      "loss": 0.0028,
      "step": 14291
    },
    {
      "epoch": 6.6707117852975495,
      "grad_norm": 0.0260009765625,
      "learning_rate": 4.988990785062656e-05,
      "loss": 0.0003,
      "step": 14292
    },
    {
      "epoch": 6.671178529754959,
      "grad_norm": 0.158203125,
      "learning_rate": 4.9877210121744945e-05,
      "loss": 0.0005,
      "step": 14293
    },
    {
      "epoch": 6.671645274212369,
      "grad_norm": 0.0546875,
      "learning_rate": 4.9864513472062634e-05,
      "loss": 0.0005,
      "step": 14294
    },
    {
      "epoch": 6.672112018669778,
      "grad_norm": 0.018310546875,
      "learning_rate": 4.985181790185295e-05,
      "loss": 0.0003,
      "step": 14295
    },
    {
      "epoch": 6.672578763127188,
      "grad_norm": 0.0947265625,
      "learning_rate": 4.983912341138926e-05,
      "loss": 0.0004,
      "step": 14296
    },
    {
      "epoch": 6.673045507584598,
      "grad_norm": 0.055908203125,
      "learning_rate": 4.982643000094493e-05,
      "loss": 0.0004,
      "step": 14297
    },
    {
      "epoch": 6.673512252042007,
      "grad_norm": 0.04443359375,
      "learning_rate": 4.98137376707932e-05,
      "loss": 0.0003,
      "step": 14298
    },
    {
      "epoch": 6.673978996499416,
      "grad_norm": 0.08056640625,
      "learning_rate": 4.980104642120737e-05,
      "loss": 0.0005,
      "step": 14299
    },
    {
      "epoch": 6.674445740956826,
      "grad_norm": 0.040771484375,
      "learning_rate": 4.9788356252460764e-05,
      "loss": 0.0004,
      "step": 14300
    },
    {
      "epoch": 6.674912485414236,
      "grad_norm": 0.060546875,
      "learning_rate": 4.9775667164826514e-05,
      "loss": 0.0004,
      "step": 14301
    },
    {
      "epoch": 6.675379229871645,
      "grad_norm": 0.07958984375,
      "learning_rate": 4.976297915857787e-05,
      "loss": 0.0045,
      "step": 14302
    },
    {
      "epoch": 6.675845974329055,
      "grad_norm": 0.02685546875,
      "learning_rate": 4.975029223398806e-05,
      "loss": 0.0004,
      "step": 14303
    },
    {
      "epoch": 6.676312718786464,
      "grad_norm": 0.1875,
      "learning_rate": 4.973760639133018e-05,
      "loss": 0.0009,
      "step": 14304
    },
    {
      "epoch": 6.676779463243874,
      "grad_norm": 0.05810546875,
      "learning_rate": 4.972492163087743e-05,
      "loss": 0.0004,
      "step": 14305
    },
    {
      "epoch": 6.677246207701284,
      "grad_norm": 0.0264892578125,
      "learning_rate": 4.971223795290287e-05,
      "loss": 0.0004,
      "step": 14306
    },
    {
      "epoch": 6.677712952158693,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.9699555357679616e-05,
      "loss": 0.001,
      "step": 14307
    },
    {
      "epoch": 6.678179696616103,
      "grad_norm": 0.0169677734375,
      "learning_rate": 4.968687384548078e-05,
      "loss": 0.0003,
      "step": 14308
    },
    {
      "epoch": 6.6786464410735125,
      "grad_norm": 0.052001953125,
      "learning_rate": 4.967419341657935e-05,
      "loss": 0.001,
      "step": 14309
    },
    {
      "epoch": 6.679113185530921,
      "grad_norm": 0.38671875,
      "learning_rate": 4.966151407124836e-05,
      "loss": 0.0032,
      "step": 14310
    },
    {
      "epoch": 6.679579929988331,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.9648835809760876e-05,
      "loss": 0.0029,
      "step": 14311
    },
    {
      "epoch": 6.680046674445741,
      "grad_norm": 0.064453125,
      "learning_rate": 4.963615863238976e-05,
      "loss": 0.0004,
      "step": 14312
    },
    {
      "epoch": 6.680513418903151,
      "grad_norm": 0.0244140625,
      "learning_rate": 4.962348253940805e-05,
      "loss": 0.0003,
      "step": 14313
    },
    {
      "epoch": 6.6809801633605606,
      "grad_norm": 0.029296875,
      "learning_rate": 4.961080753108869e-05,
      "loss": 0.0004,
      "step": 14314
    },
    {
      "epoch": 6.6814469078179695,
      "grad_norm": 0.275390625,
      "learning_rate": 4.95981336077045e-05,
      "loss": 0.0011,
      "step": 14315
    },
    {
      "epoch": 6.681913652275379,
      "grad_norm": 0.013916015625,
      "learning_rate": 4.958546076952842e-05,
      "loss": 0.0002,
      "step": 14316
    },
    {
      "epoch": 6.682380396732789,
      "grad_norm": 0.087890625,
      "learning_rate": 4.9572789016833354e-05,
      "loss": 0.0004,
      "step": 14317
    },
    {
      "epoch": 6.682847141190198,
      "grad_norm": 0.72265625,
      "learning_rate": 4.956011834989205e-05,
      "loss": 0.0018,
      "step": 14318
    },
    {
      "epoch": 6.683313885647608,
      "grad_norm": 0.09326171875,
      "learning_rate": 4.9547448768977354e-05,
      "loss": 0.0075,
      "step": 14319
    },
    {
      "epoch": 6.6837806301050176,
      "grad_norm": 0.06005859375,
      "learning_rate": 4.95347802743621e-05,
      "loss": 0.0004,
      "step": 14320
    },
    {
      "epoch": 6.684247374562427,
      "grad_norm": 0.055419921875,
      "learning_rate": 4.952211286631899e-05,
      "loss": 0.0005,
      "step": 14321
    },
    {
      "epoch": 6.684714119019836,
      "grad_norm": 0.0247802734375,
      "learning_rate": 4.950944654512083e-05,
      "loss": 0.0003,
      "step": 14322
    },
    {
      "epoch": 6.685180863477246,
      "grad_norm": 0.146484375,
      "learning_rate": 4.9496781311040275e-05,
      "loss": 0.0006,
      "step": 14323
    },
    {
      "epoch": 6.685647607934656,
      "grad_norm": 0.5625,
      "learning_rate": 4.948411716435008e-05,
      "loss": 0.0027,
      "step": 14324
    },
    {
      "epoch": 6.686114352392066,
      "grad_norm": 0.169921875,
      "learning_rate": 4.947145410532286e-05,
      "loss": 0.0024,
      "step": 14325
    },
    {
      "epoch": 6.6865810968494745,
      "grad_norm": 0.0732421875,
      "learning_rate": 4.945879213423132e-05,
      "loss": 0.0032,
      "step": 14326
    },
    {
      "epoch": 6.687047841306884,
      "grad_norm": 0.27734375,
      "learning_rate": 4.944613125134805e-05,
      "loss": 0.0013,
      "step": 14327
    },
    {
      "epoch": 6.687514585764294,
      "grad_norm": 0.1181640625,
      "learning_rate": 4.9433471456945646e-05,
      "loss": 0.0006,
      "step": 14328
    },
    {
      "epoch": 6.687981330221704,
      "grad_norm": 0.0361328125,
      "learning_rate": 4.942081275129674e-05,
      "loss": 0.0003,
      "step": 14329
    },
    {
      "epoch": 6.688448074679113,
      "grad_norm": 0.1181640625,
      "learning_rate": 4.940815513467383e-05,
      "loss": 0.0005,
      "step": 14330
    },
    {
      "epoch": 6.688914819136523,
      "grad_norm": 0.017333984375,
      "learning_rate": 4.9395498607349463e-05,
      "loss": 0.0003,
      "step": 14331
    },
    {
      "epoch": 6.689381563593932,
      "grad_norm": 0.020751953125,
      "learning_rate": 4.9382843169596205e-05,
      "loss": 0.0003,
      "step": 14332
    },
    {
      "epoch": 6.689848308051342,
      "grad_norm": 0.09619140625,
      "learning_rate": 4.9370188821686446e-05,
      "loss": 0.0005,
      "step": 14333
    },
    {
      "epoch": 6.690315052508751,
      "grad_norm": 0.1640625,
      "learning_rate": 4.935753556389273e-05,
      "loss": 0.0007,
      "step": 14334
    },
    {
      "epoch": 6.690781796966161,
      "grad_norm": 0.041015625,
      "learning_rate": 4.934488339648744e-05,
      "loss": 0.0004,
      "step": 14335
    },
    {
      "epoch": 6.691248541423571,
      "grad_norm": 0.208984375,
      "learning_rate": 4.933223231974301e-05,
      "loss": 0.003,
      "step": 14336
    },
    {
      "epoch": 6.6917152858809805,
      "grad_norm": 0.16015625,
      "learning_rate": 4.931958233393187e-05,
      "loss": 0.0007,
      "step": 14337
    },
    {
      "epoch": 6.692182030338389,
      "grad_norm": 0.07666015625,
      "learning_rate": 4.930693343932631e-05,
      "loss": 0.0005,
      "step": 14338
    },
    {
      "epoch": 6.692648774795799,
      "grad_norm": 0.053955078125,
      "learning_rate": 4.929428563619872e-05,
      "loss": 0.0028,
      "step": 14339
    },
    {
      "epoch": 6.693115519253209,
      "grad_norm": 0.1220703125,
      "learning_rate": 4.928163892482146e-05,
      "loss": 0.0006,
      "step": 14340
    },
    {
      "epoch": 6.693582263710619,
      "grad_norm": 0.01275634765625,
      "learning_rate": 4.926899330546675e-05,
      "loss": 0.0002,
      "step": 14341
    },
    {
      "epoch": 6.694049008168028,
      "grad_norm": 0.21484375,
      "learning_rate": 4.925634877840689e-05,
      "loss": 0.0009,
      "step": 14342
    },
    {
      "epoch": 6.6945157526254375,
      "grad_norm": 0.146484375,
      "learning_rate": 4.924370534391419e-05,
      "loss": 0.0033,
      "step": 14343
    },
    {
      "epoch": 6.694982497082847,
      "grad_norm": 0.06494140625,
      "learning_rate": 4.9231063002260794e-05,
      "loss": 0.0006,
      "step": 14344
    },
    {
      "epoch": 6.695449241540256,
      "grad_norm": 0.216796875,
      "learning_rate": 4.921842175371893e-05,
      "loss": 0.0013,
      "step": 14345
    },
    {
      "epoch": 6.695915985997666,
      "grad_norm": 0.1328125,
      "learning_rate": 4.9205781598560844e-05,
      "loss": 0.0031,
      "step": 14346
    },
    {
      "epoch": 6.696382730455076,
      "grad_norm": 0.08984375,
      "learning_rate": 4.919314253705859e-05,
      "loss": 0.0005,
      "step": 14347
    },
    {
      "epoch": 6.696849474912486,
      "grad_norm": 0.265625,
      "learning_rate": 4.9180504569484346e-05,
      "loss": 0.001,
      "step": 14348
    },
    {
      "epoch": 6.697316219369895,
      "grad_norm": 0.08935546875,
      "learning_rate": 4.916786769611026e-05,
      "loss": 0.0004,
      "step": 14349
    },
    {
      "epoch": 6.697782963827304,
      "grad_norm": 0.07763671875,
      "learning_rate": 4.9155231917208344e-05,
      "loss": 0.0005,
      "step": 14350
    },
    {
      "epoch": 6.698249708284714,
      "grad_norm": 0.212890625,
      "learning_rate": 4.9142597233050705e-05,
      "loss": 0.0007,
      "step": 14351
    },
    {
      "epoch": 6.698716452742124,
      "grad_norm": 0.2119140625,
      "learning_rate": 4.91299636439094e-05,
      "loss": 0.001,
      "step": 14352
    },
    {
      "epoch": 6.699183197199533,
      "grad_norm": 0.0703125,
      "learning_rate": 4.91173311500564e-05,
      "loss": 0.0032,
      "step": 14353
    },
    {
      "epoch": 6.699649941656943,
      "grad_norm": 0.083984375,
      "learning_rate": 4.910469975176374e-05,
      "loss": 0.002,
      "step": 14354
    },
    {
      "epoch": 6.700116686114352,
      "grad_norm": 0.0208740234375,
      "learning_rate": 4.9092069449303334e-05,
      "loss": 0.0003,
      "step": 14355
    },
    {
      "epoch": 6.700583430571762,
      "grad_norm": 0.08984375,
      "learning_rate": 4.907944024294715e-05,
      "loss": 0.0005,
      "step": 14356
    },
    {
      "epoch": 6.701050175029172,
      "grad_norm": 0.060546875,
      "learning_rate": 4.906681213296715e-05,
      "loss": 0.0038,
      "step": 14357
    },
    {
      "epoch": 6.701516919486581,
      "grad_norm": 0.34765625,
      "learning_rate": 4.9054185119635165e-05,
      "loss": 0.003,
      "step": 14358
    },
    {
      "epoch": 6.701983663943991,
      "grad_norm": 0.21484375,
      "learning_rate": 4.9041559203223095e-05,
      "loss": 0.001,
      "step": 14359
    },
    {
      "epoch": 6.7024504084014005,
      "grad_norm": 0.158203125,
      "learning_rate": 4.9028934384002834e-05,
      "loss": 0.0005,
      "step": 14360
    },
    {
      "epoch": 6.702917152858809,
      "grad_norm": 0.078125,
      "learning_rate": 4.901631066224613e-05,
      "loss": 0.0005,
      "step": 14361
    },
    {
      "epoch": 6.703383897316219,
      "grad_norm": 0.158203125,
      "learning_rate": 4.900368803822483e-05,
      "loss": 0.0006,
      "step": 14362
    },
    {
      "epoch": 6.703850641773629,
      "grad_norm": 0.34765625,
      "learning_rate": 4.8991066512210736e-05,
      "loss": 0.0098,
      "step": 14363
    },
    {
      "epoch": 6.704317386231039,
      "grad_norm": 0.0732421875,
      "learning_rate": 4.897844608447555e-05,
      "loss": 0.0009,
      "step": 14364
    },
    {
      "epoch": 6.704784130688448,
      "grad_norm": 0.3203125,
      "learning_rate": 4.896582675529102e-05,
      "loss": 0.0027,
      "step": 14365
    },
    {
      "epoch": 6.7052508751458575,
      "grad_norm": 0.27734375,
      "learning_rate": 4.895320852492891e-05,
      "loss": 0.0053,
      "step": 14366
    },
    {
      "epoch": 6.705717619603267,
      "grad_norm": 0.0556640625,
      "learning_rate": 4.8940591393660804e-05,
      "loss": 0.0004,
      "step": 14367
    },
    {
      "epoch": 6.706184364060677,
      "grad_norm": 0.06787109375,
      "learning_rate": 4.8927975361758436e-05,
      "loss": 0.0022,
      "step": 14368
    },
    {
      "epoch": 6.706651108518086,
      "grad_norm": 0.2373046875,
      "learning_rate": 4.891536042949345e-05,
      "loss": 0.0011,
      "step": 14369
    },
    {
      "epoch": 6.707117852975496,
      "grad_norm": 0.06396484375,
      "learning_rate": 4.890274659713739e-05,
      "loss": 0.0004,
      "step": 14370
    },
    {
      "epoch": 6.707584597432906,
      "grad_norm": 0.053955078125,
      "learning_rate": 4.889013386496194e-05,
      "loss": 0.0004,
      "step": 14371
    },
    {
      "epoch": 6.708051341890315,
      "grad_norm": 0.08154296875,
      "learning_rate": 4.8877522233238614e-05,
      "loss": 0.0006,
      "step": 14372
    },
    {
      "epoch": 6.708518086347724,
      "grad_norm": 0.12451171875,
      "learning_rate": 4.8864911702238916e-05,
      "loss": 0.0006,
      "step": 14373
    },
    {
      "epoch": 6.708984830805134,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.885230227223441e-05,
      "loss": 0.0007,
      "step": 14374
    },
    {
      "epoch": 6.709451575262544,
      "grad_norm": 0.115234375,
      "learning_rate": 4.883969394349661e-05,
      "loss": 0.0029,
      "step": 14375
    },
    {
      "epoch": 6.709918319719954,
      "grad_norm": 0.046142578125,
      "learning_rate": 4.882708671629693e-05,
      "loss": 0.0004,
      "step": 14376
    },
    {
      "epoch": 6.710385064177363,
      "grad_norm": 0.10888671875,
      "learning_rate": 4.881448059090685e-05,
      "loss": 0.0008,
      "step": 14377
    },
    {
      "epoch": 6.710851808634772,
      "grad_norm": 0.035888671875,
      "learning_rate": 4.880187556759783e-05,
      "loss": 0.0027,
      "step": 14378
    },
    {
      "epoch": 6.711318553092182,
      "grad_norm": 0.267578125,
      "learning_rate": 4.87892716466412e-05,
      "loss": 0.0096,
      "step": 14379
    },
    {
      "epoch": 6.711785297549592,
      "grad_norm": 0.0615234375,
      "learning_rate": 4.877666882830837e-05,
      "loss": 0.0004,
      "step": 14380
    },
    {
      "epoch": 6.712252042007001,
      "grad_norm": 0.0230712890625,
      "learning_rate": 4.876406711287074e-05,
      "loss": 0.0004,
      "step": 14381
    },
    {
      "epoch": 6.712718786464411,
      "grad_norm": 0.052490234375,
      "learning_rate": 4.875146650059954e-05,
      "loss": 0.0035,
      "step": 14382
    },
    {
      "epoch": 6.7131855309218205,
      "grad_norm": 0.09619140625,
      "learning_rate": 4.873886699176613e-05,
      "loss": 0.0005,
      "step": 14383
    },
    {
      "epoch": 6.71365227537923,
      "grad_norm": 0.390625,
      "learning_rate": 4.872626858664183e-05,
      "loss": 0.0016,
      "step": 14384
    },
    {
      "epoch": 6.714119019836639,
      "grad_norm": 0.08056640625,
      "learning_rate": 4.8713671285497816e-05,
      "loss": 0.0005,
      "step": 14385
    },
    {
      "epoch": 6.714585764294049,
      "grad_norm": 0.02978515625,
      "learning_rate": 4.8701075088605405e-05,
      "loss": 0.0003,
      "step": 14386
    },
    {
      "epoch": 6.715052508751459,
      "grad_norm": 0.0196533203125,
      "learning_rate": 4.8688479996235737e-05,
      "loss": 0.0003,
      "step": 14387
    },
    {
      "epoch": 6.715519253208868,
      "grad_norm": 0.06005859375,
      "learning_rate": 4.867588600866001e-05,
      "loss": 0.0005,
      "step": 14388
    },
    {
      "epoch": 6.7159859976662775,
      "grad_norm": 0.01361083984375,
      "learning_rate": 4.866329312614944e-05,
      "loss": 0.0002,
      "step": 14389
    },
    {
      "epoch": 6.716452742123687,
      "grad_norm": 0.267578125,
      "learning_rate": 4.8650701348975095e-05,
      "loss": 0.0032,
      "step": 14390
    },
    {
      "epoch": 6.716919486581097,
      "grad_norm": 0.036376953125,
      "learning_rate": 4.863811067740812e-05,
      "loss": 0.0003,
      "step": 14391
    },
    {
      "epoch": 6.717386231038507,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.862552111171965e-05,
      "loss": 0.0005,
      "step": 14392
    },
    {
      "epoch": 6.717852975495916,
      "grad_norm": 0.1083984375,
      "learning_rate": 4.861293265218068e-05,
      "loss": 0.0004,
      "step": 14393
    },
    {
      "epoch": 6.7183197199533256,
      "grad_norm": 0.033935546875,
      "learning_rate": 4.860034529906227e-05,
      "loss": 0.0003,
      "step": 14394
    },
    {
      "epoch": 6.718786464410735,
      "grad_norm": 0.1259765625,
      "learning_rate": 4.8587759052635505e-05,
      "loss": 0.0005,
      "step": 14395
    },
    {
      "epoch": 6.719253208868144,
      "grad_norm": 0.043212890625,
      "learning_rate": 4.857517391317128e-05,
      "loss": 0.0004,
      "step": 14396
    },
    {
      "epoch": 6.719719953325554,
      "grad_norm": 0.037353515625,
      "learning_rate": 4.8562589880940636e-05,
      "loss": 0.0003,
      "step": 14397
    },
    {
      "epoch": 6.720186697782964,
      "grad_norm": 0.134765625,
      "learning_rate": 4.855000695621452e-05,
      "loss": 0.0018,
      "step": 14398
    },
    {
      "epoch": 6.720653442240374,
      "grad_norm": 0.025390625,
      "learning_rate": 4.853742513926381e-05,
      "loss": 0.0003,
      "step": 14399
    },
    {
      "epoch": 6.721120186697783,
      "grad_norm": 0.333984375,
      "learning_rate": 4.8524844430359436e-05,
      "loss": 0.0042,
      "step": 14400
    },
    {
      "epoch": 6.721586931155192,
      "grad_norm": 0.1728515625,
      "learning_rate": 4.8512264829772314e-05,
      "loss": 0.0007,
      "step": 14401
    },
    {
      "epoch": 6.722053675612602,
      "grad_norm": 0.095703125,
      "learning_rate": 4.849968633777322e-05,
      "loss": 0.0006,
      "step": 14402
    },
    {
      "epoch": 6.722520420070012,
      "grad_norm": 0.0240478515625,
      "learning_rate": 4.8487108954633054e-05,
      "loss": 0.0004,
      "step": 14403
    },
    {
      "epoch": 6.722987164527421,
      "grad_norm": 0.0185546875,
      "learning_rate": 4.8474532680622556e-05,
      "loss": 0.0002,
      "step": 14404
    },
    {
      "epoch": 6.723453908984831,
      "grad_norm": 0.051513671875,
      "learning_rate": 4.8461957516012526e-05,
      "loss": 0.0022,
      "step": 14405
    },
    {
      "epoch": 6.72392065344224,
      "grad_norm": 0.072265625,
      "learning_rate": 4.844938346107377e-05,
      "loss": 0.0012,
      "step": 14406
    },
    {
      "epoch": 6.72438739789965,
      "grad_norm": 0.0869140625,
      "learning_rate": 4.843681051607696e-05,
      "loss": 0.0004,
      "step": 14407
    },
    {
      "epoch": 6.724854142357059,
      "grad_norm": 0.11376953125,
      "learning_rate": 4.8424238681292825e-05,
      "loss": 0.001,
      "step": 14408
    },
    {
      "epoch": 6.725320886814469,
      "grad_norm": 0.431640625,
      "learning_rate": 4.8411667956992094e-05,
      "loss": 0.0038,
      "step": 14409
    },
    {
      "epoch": 6.725787631271879,
      "grad_norm": 0.0927734375,
      "learning_rate": 4.839909834344535e-05,
      "loss": 0.0005,
      "step": 14410
    },
    {
      "epoch": 6.7262543757292885,
      "grad_norm": 0.05615234375,
      "learning_rate": 4.8386529840923266e-05,
      "loss": 0.0005,
      "step": 14411
    },
    {
      "epoch": 6.726721120186697,
      "grad_norm": 0.041259765625,
      "learning_rate": 4.83739624496965e-05,
      "loss": 0.0024,
      "step": 14412
    },
    {
      "epoch": 6.727187864644107,
      "grad_norm": 0.03369140625,
      "learning_rate": 4.8361396170035566e-05,
      "loss": 0.0003,
      "step": 14413
    },
    {
      "epoch": 6.727654609101517,
      "grad_norm": 0.0135498046875,
      "learning_rate": 4.834883100221107e-05,
      "loss": 0.0002,
      "step": 14414
    },
    {
      "epoch": 6.728121353558927,
      "grad_norm": 0.283203125,
      "learning_rate": 4.833626694649359e-05,
      "loss": 0.0033,
      "step": 14415
    },
    {
      "epoch": 6.728588098016336,
      "grad_norm": 0.037109375,
      "learning_rate": 4.8323704003153555e-05,
      "loss": 0.0004,
      "step": 14416
    },
    {
      "epoch": 6.7290548424737455,
      "grad_norm": 0.052978515625,
      "learning_rate": 4.831114217246152e-05,
      "loss": 0.0004,
      "step": 14417
    },
    {
      "epoch": 6.729521586931155,
      "grad_norm": 0.0281982421875,
      "learning_rate": 4.8298581454687984e-05,
      "loss": 0.0003,
      "step": 14418
    },
    {
      "epoch": 6.729988331388565,
      "grad_norm": 0.11669921875,
      "learning_rate": 4.828602185010335e-05,
      "loss": 0.0006,
      "step": 14419
    },
    {
      "epoch": 6.730455075845974,
      "grad_norm": 0.142578125,
      "learning_rate": 4.8273463358978e-05,
      "loss": 0.0005,
      "step": 14420
    },
    {
      "epoch": 6.730921820303384,
      "grad_norm": 0.03564453125,
      "learning_rate": 4.8260905981582414e-05,
      "loss": 0.0003,
      "step": 14421
    },
    {
      "epoch": 6.731388564760794,
      "grad_norm": 0.392578125,
      "learning_rate": 4.8248349718186894e-05,
      "loss": 0.003,
      "step": 14422
    },
    {
      "epoch": 6.731855309218203,
      "grad_norm": 0.1865234375,
      "learning_rate": 4.823579456906182e-05,
      "loss": 0.0007,
      "step": 14423
    },
    {
      "epoch": 6.732322053675612,
      "grad_norm": 0.054443359375,
      "learning_rate": 4.8223240534477565e-05,
      "loss": 0.0003,
      "step": 14424
    },
    {
      "epoch": 6.732788798133022,
      "grad_norm": 0.026123046875,
      "learning_rate": 4.821068761470434e-05,
      "loss": 0.0003,
      "step": 14425
    },
    {
      "epoch": 6.733255542590432,
      "grad_norm": 0.042236328125,
      "learning_rate": 4.819813581001248e-05,
      "loss": 0.0003,
      "step": 14426
    },
    {
      "epoch": 6.733722287047842,
      "grad_norm": 0.052490234375,
      "learning_rate": 4.818558512067227e-05,
      "loss": 0.0027,
      "step": 14427
    },
    {
      "epoch": 6.734189031505251,
      "grad_norm": 0.203125,
      "learning_rate": 4.817303554695386e-05,
      "loss": 0.0007,
      "step": 14428
    },
    {
      "epoch": 6.73465577596266,
      "grad_norm": 0.1103515625,
      "learning_rate": 4.81604870891275e-05,
      "loss": 0.0024,
      "step": 14429
    },
    {
      "epoch": 6.73512252042007,
      "grad_norm": 0.060791015625,
      "learning_rate": 4.814793974746341e-05,
      "loss": 0.0003,
      "step": 14430
    },
    {
      "epoch": 6.735589264877479,
      "grad_norm": 0.3203125,
      "learning_rate": 4.813539352223167e-05,
      "loss": 0.0013,
      "step": 14431
    },
    {
      "epoch": 6.736056009334889,
      "grad_norm": 0.23046875,
      "learning_rate": 4.812284841370245e-05,
      "loss": 0.001,
      "step": 14432
    },
    {
      "epoch": 6.736522753792299,
      "grad_norm": 0.0380859375,
      "learning_rate": 4.81103044221459e-05,
      "loss": 0.0022,
      "step": 14433
    },
    {
      "epoch": 6.7369894982497085,
      "grad_norm": 0.057861328125,
      "learning_rate": 4.809776154783203e-05,
      "loss": 0.0004,
      "step": 14434
    },
    {
      "epoch": 6.737456242707118,
      "grad_norm": 0.051025390625,
      "learning_rate": 4.8085219791030975e-05,
      "loss": 0.0023,
      "step": 14435
    },
    {
      "epoch": 6.737922987164527,
      "grad_norm": 0.020263671875,
      "learning_rate": 4.8072679152012714e-05,
      "loss": 0.0002,
      "step": 14436
    },
    {
      "epoch": 6.738389731621937,
      "grad_norm": 0.06982421875,
      "learning_rate": 4.8060139631047275e-05,
      "loss": 0.0005,
      "step": 14437
    },
    {
      "epoch": 6.738856476079347,
      "grad_norm": 0.421875,
      "learning_rate": 4.804760122840469e-05,
      "loss": 0.0048,
      "step": 14438
    },
    {
      "epoch": 6.739323220536756,
      "grad_norm": 0.0302734375,
      "learning_rate": 4.803506394435486e-05,
      "loss": 0.0003,
      "step": 14439
    },
    {
      "epoch": 6.7397899649941655,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.802252777916777e-05,
      "loss": 0.0031,
      "step": 14440
    },
    {
      "epoch": 6.740256709451575,
      "grad_norm": 0.03125,
      "learning_rate": 4.800999273311335e-05,
      "loss": 0.0003,
      "step": 14441
    },
    {
      "epoch": 6.740723453908985,
      "grad_norm": 0.049072265625,
      "learning_rate": 4.7997458806461446e-05,
      "loss": 0.0004,
      "step": 14442
    },
    {
      "epoch": 6.741190198366395,
      "grad_norm": 0.19921875,
      "learning_rate": 4.798492599948195e-05,
      "loss": 0.0009,
      "step": 14443
    },
    {
      "epoch": 6.741656942823804,
      "grad_norm": 0.2158203125,
      "learning_rate": 4.797239431244474e-05,
      "loss": 0.0011,
      "step": 14444
    },
    {
      "epoch": 6.742123687281214,
      "grad_norm": 0.0888671875,
      "learning_rate": 4.795986374561956e-05,
      "loss": 0.0008,
      "step": 14445
    },
    {
      "epoch": 6.742590431738623,
      "grad_norm": 0.0279541015625,
      "learning_rate": 4.794733429927627e-05,
      "loss": 0.0003,
      "step": 14446
    },
    {
      "epoch": 6.743057176196032,
      "grad_norm": 0.0634765625,
      "learning_rate": 4.793480597368465e-05,
      "loss": 0.0003,
      "step": 14447
    },
    {
      "epoch": 6.743523920653442,
      "grad_norm": 0.0260009765625,
      "learning_rate": 4.792227876911439e-05,
      "loss": 0.0003,
      "step": 14448
    },
    {
      "epoch": 6.743990665110852,
      "grad_norm": 0.1279296875,
      "learning_rate": 4.7909752685835254e-05,
      "loss": 0.0007,
      "step": 14449
    },
    {
      "epoch": 6.744457409568262,
      "grad_norm": 0.07470703125,
      "learning_rate": 4.789722772411698e-05,
      "loss": 0.0004,
      "step": 14450
    },
    {
      "epoch": 6.744924154025671,
      "grad_norm": 0.046875,
      "learning_rate": 4.788470388422915e-05,
      "loss": 0.0028,
      "step": 14451
    },
    {
      "epoch": 6.74539089848308,
      "grad_norm": 0.330078125,
      "learning_rate": 4.787218116644152e-05,
      "loss": 0.0019,
      "step": 14452
    },
    {
      "epoch": 6.74585764294049,
      "grad_norm": 0.26953125,
      "learning_rate": 4.7859659571023615e-05,
      "loss": 0.0019,
      "step": 14453
    },
    {
      "epoch": 6.7463243873979,
      "grad_norm": 0.5,
      "learning_rate": 4.784713909824511e-05,
      "loss": 0.0015,
      "step": 14454
    },
    {
      "epoch": 6.746791131855309,
      "grad_norm": 0.134765625,
      "learning_rate": 4.783461974837559e-05,
      "loss": 0.0029,
      "step": 14455
    },
    {
      "epoch": 6.747257876312719,
      "grad_norm": 0.2470703125,
      "learning_rate": 4.782210152168455e-05,
      "loss": 0.0011,
      "step": 14456
    },
    {
      "epoch": 6.7477246207701285,
      "grad_norm": 0.036865234375,
      "learning_rate": 4.780958441844155e-05,
      "loss": 0.0004,
      "step": 14457
    },
    {
      "epoch": 6.748191365227538,
      "grad_norm": 0.05908203125,
      "learning_rate": 4.7797068438916145e-05,
      "loss": 0.0019,
      "step": 14458
    },
    {
      "epoch": 6.748658109684947,
      "grad_norm": 0.018310546875,
      "learning_rate": 4.778455358337774e-05,
      "loss": 0.0003,
      "step": 14459
    },
    {
      "epoch": 6.749124854142357,
      "grad_norm": 0.04052734375,
      "learning_rate": 4.777203985209583e-05,
      "loss": 0.0004,
      "step": 14460
    },
    {
      "epoch": 6.749591598599767,
      "grad_norm": 0.265625,
      "learning_rate": 4.7759527245339884e-05,
      "loss": 0.0019,
      "step": 14461
    },
    {
      "epoch": 6.750058343057177,
      "grad_norm": 0.15234375,
      "learning_rate": 4.7747015763379245e-05,
      "loss": 0.0123,
      "step": 14462
    },
    {
      "epoch": 6.7505250875145855,
      "grad_norm": 0.04248046875,
      "learning_rate": 4.7734505406483334e-05,
      "loss": 0.0004,
      "step": 14463
    },
    {
      "epoch": 6.750991831971995,
      "grad_norm": 0.0301513671875,
      "learning_rate": 4.7721996174921544e-05,
      "loss": 0.0003,
      "step": 14464
    },
    {
      "epoch": 6.751458576429405,
      "grad_norm": 0.34375,
      "learning_rate": 4.770948806896315e-05,
      "loss": 0.0025,
      "step": 14465
    },
    {
      "epoch": 6.751925320886815,
      "grad_norm": 0.2177734375,
      "learning_rate": 4.769698108887749e-05,
      "loss": 0.0009,
      "step": 14466
    },
    {
      "epoch": 6.752392065344224,
      "grad_norm": 0.26953125,
      "learning_rate": 4.768447523493393e-05,
      "loss": 0.0035,
      "step": 14467
    },
    {
      "epoch": 6.7528588098016336,
      "grad_norm": 0.05078125,
      "learning_rate": 4.76719705074016e-05,
      "loss": 0.0024,
      "step": 14468
    },
    {
      "epoch": 6.753325554259043,
      "grad_norm": 0.05126953125,
      "learning_rate": 4.7659466906549785e-05,
      "loss": 0.0004,
      "step": 14469
    },
    {
      "epoch": 6.753792298716453,
      "grad_norm": 0.16015625,
      "learning_rate": 4.764696443264778e-05,
      "loss": 0.0007,
      "step": 14470
    },
    {
      "epoch": 6.754259043173862,
      "grad_norm": 0.126953125,
      "learning_rate": 4.763446308596467e-05,
      "loss": 0.0006,
      "step": 14471
    },
    {
      "epoch": 6.754725787631272,
      "grad_norm": 0.0771484375,
      "learning_rate": 4.7621962866769664e-05,
      "loss": 0.0004,
      "step": 14472
    },
    {
      "epoch": 6.754725787631272,
      "eval_loss": 2.3073253631591797,
      "eval_runtime": 85.1391,
      "eval_samples_per_second": 21.189,
      "eval_steps_per_second": 2.654,
      "step": 14472
    },
    {
      "epoch": 6.755192532088682,
      "grad_norm": 0.134765625,
      "learning_rate": 4.760946377533195e-05,
      "loss": 0.0005,
      "step": 14473
    },
    {
      "epoch": 6.7556592765460906,
      "grad_norm": 0.0693359375,
      "learning_rate": 4.759696581192058e-05,
      "loss": 0.0004,
      "step": 14474
    },
    {
      "epoch": 6.7561260210035,
      "grad_norm": 0.046142578125,
      "learning_rate": 4.7584468976804674e-05,
      "loss": 0.0004,
      "step": 14475
    },
    {
      "epoch": 6.75659276546091,
      "grad_norm": 0.0712890625,
      "learning_rate": 4.757197327025335e-05,
      "loss": 0.0006,
      "step": 14476
    },
    {
      "epoch": 6.75705950991832,
      "grad_norm": 0.0157470703125,
      "learning_rate": 4.7559478692535565e-05,
      "loss": 0.0002,
      "step": 14477
    },
    {
      "epoch": 6.75752625437573,
      "grad_norm": 0.033935546875,
      "learning_rate": 4.7546985243920386e-05,
      "loss": 0.0004,
      "step": 14478
    },
    {
      "epoch": 6.757992998833139,
      "grad_norm": 0.11181640625,
      "learning_rate": 4.7534492924676854e-05,
      "loss": 0.0033,
      "step": 14479
    },
    {
      "epoch": 6.758459743290548,
      "grad_norm": 0.306640625,
      "learning_rate": 4.752200173507387e-05,
      "loss": 0.0012,
      "step": 14480
    },
    {
      "epoch": 6.758926487747958,
      "grad_norm": 0.029296875,
      "learning_rate": 4.7509511675380406e-05,
      "loss": 0.0023,
      "step": 14481
    },
    {
      "epoch": 6.759393232205367,
      "grad_norm": 0.2197265625,
      "learning_rate": 4.7497022745865436e-05,
      "loss": 0.0021,
      "step": 14482
    },
    {
      "epoch": 6.759859976662777,
      "grad_norm": 0.072265625,
      "learning_rate": 4.748453494679778e-05,
      "loss": 0.0003,
      "step": 14483
    },
    {
      "epoch": 6.760326721120187,
      "grad_norm": 0.3203125,
      "learning_rate": 4.747204827844639e-05,
      "loss": 0.002,
      "step": 14484
    },
    {
      "epoch": 6.7607934655775965,
      "grad_norm": 0.056396484375,
      "learning_rate": 4.745956274108005e-05,
      "loss": 0.0003,
      "step": 14485
    },
    {
      "epoch": 6.761260210035006,
      "grad_norm": 0.0537109375,
      "learning_rate": 4.744707833496761e-05,
      "loss": 0.0003,
      "step": 14486
    },
    {
      "epoch": 6.761726954492415,
      "grad_norm": 0.08447265625,
      "learning_rate": 4.7434595060377926e-05,
      "loss": 0.0026,
      "step": 14487
    },
    {
      "epoch": 6.762193698949825,
      "grad_norm": 0.07763671875,
      "learning_rate": 4.7422112917579686e-05,
      "loss": 0.0006,
      "step": 14488
    },
    {
      "epoch": 6.762660443407235,
      "grad_norm": 0.0223388671875,
      "learning_rate": 4.74096319068417e-05,
      "loss": 0.0002,
      "step": 14489
    },
    {
      "epoch": 6.763127187864644,
      "grad_norm": 0.275390625,
      "learning_rate": 4.7397152028432724e-05,
      "loss": 0.0045,
      "step": 14490
    },
    {
      "epoch": 6.7635939323220535,
      "grad_norm": 0.0291748046875,
      "learning_rate": 4.738467328262139e-05,
      "loss": 0.0024,
      "step": 14491
    },
    {
      "epoch": 6.764060676779463,
      "grad_norm": 0.059814453125,
      "learning_rate": 4.7372195669676425e-05,
      "loss": 0.0005,
      "step": 14492
    },
    {
      "epoch": 6.764527421236873,
      "grad_norm": 0.0634765625,
      "learning_rate": 4.735971918986651e-05,
      "loss": 0.0005,
      "step": 14493
    },
    {
      "epoch": 6.764994165694282,
      "grad_norm": 0.0186767578125,
      "learning_rate": 4.734724384346021e-05,
      "loss": 0.0003,
      "step": 14494
    },
    {
      "epoch": 6.765460910151692,
      "grad_norm": 0.056396484375,
      "learning_rate": 4.733476963072617e-05,
      "loss": 0.0019,
      "step": 14495
    },
    {
      "epoch": 6.765927654609102,
      "grad_norm": 0.1376953125,
      "learning_rate": 4.732229655193301e-05,
      "loss": 0.0008,
      "step": 14496
    },
    {
      "epoch": 6.766394399066511,
      "grad_norm": 0.056640625,
      "learning_rate": 4.7309824607349204e-05,
      "loss": 0.0032,
      "step": 14497
    },
    {
      "epoch": 6.76686114352392,
      "grad_norm": 0.03857421875,
      "learning_rate": 4.729735379724335e-05,
      "loss": 0.0003,
      "step": 14498
    },
    {
      "epoch": 6.76732788798133,
      "grad_norm": 0.03564453125,
      "learning_rate": 4.728488412188398e-05,
      "loss": 0.0003,
      "step": 14499
    },
    {
      "epoch": 6.76779463243874,
      "grad_norm": 0.197265625,
      "learning_rate": 4.7272415581539495e-05,
      "loss": 0.0039,
      "step": 14500
    },
    {
      "epoch": 6.76826137689615,
      "grad_norm": 0.0169677734375,
      "learning_rate": 4.725994817647845e-05,
      "loss": 0.0003,
      "step": 14501
    },
    {
      "epoch": 6.768728121353559,
      "grad_norm": 0.130859375,
      "learning_rate": 4.72474819069692e-05,
      "loss": 0.0038,
      "step": 14502
    },
    {
      "epoch": 6.769194865810968,
      "grad_norm": 0.028076171875,
      "learning_rate": 4.723501677328019e-05,
      "loss": 0.0003,
      "step": 14503
    },
    {
      "epoch": 6.769661610268378,
      "grad_norm": 0.146484375,
      "learning_rate": 4.7222552775679854e-05,
      "loss": 0.0007,
      "step": 14504
    },
    {
      "epoch": 6.770128354725788,
      "grad_norm": 0.0498046875,
      "learning_rate": 4.7210089914436475e-05,
      "loss": 0.0004,
      "step": 14505
    },
    {
      "epoch": 6.770595099183197,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.719762818981843e-05,
      "loss": 0.0049,
      "step": 14506
    },
    {
      "epoch": 6.771061843640607,
      "grad_norm": 0.01007080078125,
      "learning_rate": 4.7185167602094085e-05,
      "loss": 0.0002,
      "step": 14507
    },
    {
      "epoch": 6.7715285880980165,
      "grad_norm": 0.095703125,
      "learning_rate": 4.717270815153163e-05,
      "loss": 0.0025,
      "step": 14508
    },
    {
      "epoch": 6.771995332555425,
      "grad_norm": 0.0751953125,
      "learning_rate": 4.716024983839938e-05,
      "loss": 0.0006,
      "step": 14509
    },
    {
      "epoch": 6.772462077012835,
      "grad_norm": 0.08740234375,
      "learning_rate": 4.7147792662965614e-05,
      "loss": 0.0004,
      "step": 14510
    },
    {
      "epoch": 6.772928821470245,
      "grad_norm": 0.0306396484375,
      "learning_rate": 4.713533662549848e-05,
      "loss": 0.0004,
      "step": 14511
    },
    {
      "epoch": 6.773395565927655,
      "grad_norm": 0.02587890625,
      "learning_rate": 4.712288172626621e-05,
      "loss": 0.0003,
      "step": 14512
    },
    {
      "epoch": 6.773862310385065,
      "grad_norm": 0.017578125,
      "learning_rate": 4.7110427965537e-05,
      "loss": 0.0003,
      "step": 14513
    },
    {
      "epoch": 6.7743290548424735,
      "grad_norm": 0.173828125,
      "learning_rate": 4.709797534357891e-05,
      "loss": 0.0005,
      "step": 14514
    },
    {
      "epoch": 6.774795799299883,
      "grad_norm": 0.1142578125,
      "learning_rate": 4.7085523860660154e-05,
      "loss": 0.0026,
      "step": 14515
    },
    {
      "epoch": 6.775262543757293,
      "grad_norm": 0.04736328125,
      "learning_rate": 4.707307351704877e-05,
      "loss": 0.0005,
      "step": 14516
    },
    {
      "epoch": 6.775729288214702,
      "grad_norm": 0.0206298828125,
      "learning_rate": 4.706062431301281e-05,
      "loss": 0.0003,
      "step": 14517
    },
    {
      "epoch": 6.776196032672112,
      "grad_norm": 0.12890625,
      "learning_rate": 4.704817624882034e-05,
      "loss": 0.0005,
      "step": 14518
    },
    {
      "epoch": 6.776662777129522,
      "grad_norm": 0.046630859375,
      "learning_rate": 4.703572932473942e-05,
      "loss": 0.0003,
      "step": 14519
    },
    {
      "epoch": 6.777129521586931,
      "grad_norm": 0.02099609375,
      "learning_rate": 4.702328354103798e-05,
      "loss": 0.0002,
      "step": 14520
    },
    {
      "epoch": 6.777596266044341,
      "grad_norm": 0.0299072265625,
      "learning_rate": 4.701083889798402e-05,
      "loss": 0.0003,
      "step": 14521
    },
    {
      "epoch": 6.77806301050175,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.699839539584553e-05,
      "loss": 0.0045,
      "step": 14522
    },
    {
      "epoch": 6.77852975495916,
      "grad_norm": 0.302734375,
      "learning_rate": 4.698595303489036e-05,
      "loss": 0.0019,
      "step": 14523
    },
    {
      "epoch": 6.77899649941657,
      "grad_norm": 0.00885009765625,
      "learning_rate": 4.697351181538644e-05,
      "loss": 0.0002,
      "step": 14524
    },
    {
      "epoch": 6.779463243873979,
      "grad_norm": 0.0233154296875,
      "learning_rate": 4.696107173760168e-05,
      "loss": 0.0003,
      "step": 14525
    },
    {
      "epoch": 6.779929988331388,
      "grad_norm": 0.01446533203125,
      "learning_rate": 4.694863280180386e-05,
      "loss": 0.0003,
      "step": 14526
    },
    {
      "epoch": 6.780396732788798,
      "grad_norm": 0.034912109375,
      "learning_rate": 4.693619500826083e-05,
      "loss": 0.0003,
      "step": 14527
    },
    {
      "epoch": 6.780863477246208,
      "grad_norm": 0.0546875,
      "learning_rate": 4.6923758357240446e-05,
      "loss": 0.0004,
      "step": 14528
    },
    {
      "epoch": 6.781330221703618,
      "grad_norm": 0.04150390625,
      "learning_rate": 4.691132284901039e-05,
      "loss": 0.0003,
      "step": 14529
    },
    {
      "epoch": 6.781796966161027,
      "grad_norm": 0.06298828125,
      "learning_rate": 4.6898888483838446e-05,
      "loss": 0.0004,
      "step": 14530
    },
    {
      "epoch": 6.7822637106184365,
      "grad_norm": 0.07421875,
      "learning_rate": 4.688645526199239e-05,
      "loss": 0.0033,
      "step": 14531
    },
    {
      "epoch": 6.782730455075846,
      "grad_norm": 0.05029296875,
      "learning_rate": 4.687402318373986e-05,
      "loss": 0.0003,
      "step": 14532
    },
    {
      "epoch": 6.783197199533255,
      "grad_norm": 0.061767578125,
      "learning_rate": 4.686159224934857e-05,
      "loss": 0.0025,
      "step": 14533
    },
    {
      "epoch": 6.783663943990665,
      "grad_norm": 0.02587890625,
      "learning_rate": 4.684916245908614e-05,
      "loss": 0.0002,
      "step": 14534
    },
    {
      "epoch": 6.784130688448075,
      "grad_norm": 0.024169921875,
      "learning_rate": 4.68367338132202e-05,
      "loss": 0.0004,
      "step": 14535
    },
    {
      "epoch": 6.784597432905485,
      "grad_norm": 0.08837890625,
      "learning_rate": 4.682430631201842e-05,
      "loss": 0.0052,
      "step": 14536
    },
    {
      "epoch": 6.7850641773628935,
      "grad_norm": 0.025390625,
      "learning_rate": 4.681187995574827e-05,
      "loss": 0.0003,
      "step": 14537
    },
    {
      "epoch": 6.785530921820303,
      "grad_norm": 0.0654296875,
      "learning_rate": 4.6799454744677376e-05,
      "loss": 0.0004,
      "step": 14538
    },
    {
      "epoch": 6.785997666277713,
      "grad_norm": 0.076171875,
      "learning_rate": 4.678703067907327e-05,
      "loss": 0.0023,
      "step": 14539
    },
    {
      "epoch": 6.786464410735123,
      "grad_norm": 0.1845703125,
      "learning_rate": 4.6774607759203407e-05,
      "loss": 0.0005,
      "step": 14540
    },
    {
      "epoch": 6.786931155192532,
      "grad_norm": 0.022216796875,
      "learning_rate": 4.6762185985335284e-05,
      "loss": 0.0002,
      "step": 14541
    },
    {
      "epoch": 6.787397899649942,
      "grad_norm": 0.046875,
      "learning_rate": 4.6749765357736405e-05,
      "loss": 0.0004,
      "step": 14542
    },
    {
      "epoch": 6.787864644107351,
      "grad_norm": 0.142578125,
      "learning_rate": 4.6737345876674134e-05,
      "loss": 0.0007,
      "step": 14543
    },
    {
      "epoch": 6.788331388564761,
      "grad_norm": 0.033203125,
      "learning_rate": 4.672492754241589e-05,
      "loss": 0.0004,
      "step": 14544
    },
    {
      "epoch": 6.78879813302217,
      "grad_norm": 0.0888671875,
      "learning_rate": 4.67125103552291e-05,
      "loss": 0.0025,
      "step": 14545
    },
    {
      "epoch": 6.78926487747958,
      "grad_norm": 0.040771484375,
      "learning_rate": 4.670009431538106e-05,
      "loss": 0.0004,
      "step": 14546
    },
    {
      "epoch": 6.78973162193699,
      "grad_norm": 0.0302734375,
      "learning_rate": 4.6687679423139117e-05,
      "loss": 0.0003,
      "step": 14547
    },
    {
      "epoch": 6.790198366394399,
      "grad_norm": 0.0166015625,
      "learning_rate": 4.667526567877063e-05,
      "loss": 0.0002,
      "step": 14548
    },
    {
      "epoch": 6.790665110851808,
      "grad_norm": 0.345703125,
      "learning_rate": 4.66628530825428e-05,
      "loss": 0.0008,
      "step": 14549
    },
    {
      "epoch": 6.791131855309218,
      "grad_norm": 0.0216064453125,
      "learning_rate": 4.665044163472292e-05,
      "loss": 0.0003,
      "step": 14550
    },
    {
      "epoch": 6.791598599766628,
      "grad_norm": 0.057861328125,
      "learning_rate": 4.663803133557826e-05,
      "loss": 0.0003,
      "step": 14551
    },
    {
      "epoch": 6.792065344224037,
      "grad_norm": 0.0859375,
      "learning_rate": 4.662562218537596e-05,
      "loss": 0.0004,
      "step": 14552
    },
    {
      "epoch": 6.792532088681447,
      "grad_norm": 0.037353515625,
      "learning_rate": 4.661321418438327e-05,
      "loss": 0.0003,
      "step": 14553
    },
    {
      "epoch": 6.792998833138856,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.660080733286727e-05,
      "loss": 0.0005,
      "step": 14554
    },
    {
      "epoch": 6.793465577596266,
      "grad_norm": 0.03564453125,
      "learning_rate": 4.658840163109514e-05,
      "loss": 0.0003,
      "step": 14555
    },
    {
      "epoch": 6.793932322053676,
      "grad_norm": 0.0625,
      "learning_rate": 4.657599707933401e-05,
      "loss": 0.0005,
      "step": 14556
    },
    {
      "epoch": 6.794399066511085,
      "grad_norm": 0.018310546875,
      "learning_rate": 4.656359367785091e-05,
      "loss": 0.0003,
      "step": 14557
    },
    {
      "epoch": 6.794865810968495,
      "grad_norm": 0.11474609375,
      "learning_rate": 4.655119142691293e-05,
      "loss": 0.0005,
      "step": 14558
    },
    {
      "epoch": 6.7953325554259045,
      "grad_norm": 0.025146484375,
      "learning_rate": 4.653879032678714e-05,
      "loss": 0.0003,
      "step": 14559
    },
    {
      "epoch": 6.795799299883313,
      "grad_norm": 0.0247802734375,
      "learning_rate": 4.652639037774047e-05,
      "loss": 0.0003,
      "step": 14560
    },
    {
      "epoch": 6.796266044340723,
      "grad_norm": 0.08544921875,
      "learning_rate": 4.651399158003994e-05,
      "loss": 0.0003,
      "step": 14561
    },
    {
      "epoch": 6.796732788798133,
      "grad_norm": 0.01513671875,
      "learning_rate": 4.650159393395256e-05,
      "loss": 0.0002,
      "step": 14562
    },
    {
      "epoch": 6.797199533255543,
      "grad_norm": 0.2333984375,
      "learning_rate": 4.648919743974521e-05,
      "loss": 0.0009,
      "step": 14563
    },
    {
      "epoch": 6.797666277712953,
      "grad_norm": 0.04296875,
      "learning_rate": 4.647680209768478e-05,
      "loss": 0.0003,
      "step": 14564
    },
    {
      "epoch": 6.7981330221703615,
      "grad_norm": 0.060302734375,
      "learning_rate": 4.646440790803822e-05,
      "loss": 0.0005,
      "step": 14565
    },
    {
      "epoch": 6.798599766627771,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.6452014871072325e-05,
      "loss": 0.0003,
      "step": 14566
    },
    {
      "epoch": 6.799066511085181,
      "grad_norm": 0.06787109375,
      "learning_rate": 4.643962298705395e-05,
      "loss": 0.0033,
      "step": 14567
    },
    {
      "epoch": 6.79953325554259,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.642723225624997e-05,
      "loss": 0.0009,
      "step": 14568
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.033447265625,
      "learning_rate": 4.6414842678927085e-05,
      "loss": 0.0003,
      "step": 14569
    },
    {
      "epoch": 6.80046674445741,
      "grad_norm": 0.0341796875,
      "learning_rate": 4.640245425535208e-05,
      "loss": 0.0004,
      "step": 14570
    },
    {
      "epoch": 6.800933488914819,
      "grad_norm": 0.0693359375,
      "learning_rate": 4.639006698579174e-05,
      "loss": 0.0004,
      "step": 14571
    },
    {
      "epoch": 6.801400233372228,
      "grad_norm": 0.031005859375,
      "learning_rate": 4.637768087051271e-05,
      "loss": 0.0003,
      "step": 14572
    },
    {
      "epoch": 6.801866977829638,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.636529590978169e-05,
      "loss": 0.0005,
      "step": 14573
    },
    {
      "epoch": 6.802333722287048,
      "grad_norm": 0.03759765625,
      "learning_rate": 4.635291210386541e-05,
      "loss": 0.0004,
      "step": 14574
    },
    {
      "epoch": 6.802800466744458,
      "grad_norm": 0.014404296875,
      "learning_rate": 4.6340529453030416e-05,
      "loss": 0.0002,
      "step": 14575
    },
    {
      "epoch": 6.803267211201867,
      "grad_norm": 0.0184326171875,
      "learning_rate": 4.6328147957543354e-05,
      "loss": 0.0002,
      "step": 14576
    },
    {
      "epoch": 6.803733955659276,
      "grad_norm": 0.016357421875,
      "learning_rate": 4.631576761767085e-05,
      "loss": 0.0002,
      "step": 14577
    },
    {
      "epoch": 6.804200700116686,
      "grad_norm": 0.0152587890625,
      "learning_rate": 4.630338843367941e-05,
      "loss": 0.0002,
      "step": 14578
    },
    {
      "epoch": 6.804667444574096,
      "grad_norm": 0.0859375,
      "learning_rate": 4.629101040583558e-05,
      "loss": 0.0004,
      "step": 14579
    },
    {
      "epoch": 6.805134189031505,
      "grad_norm": 0.0732421875,
      "learning_rate": 4.6278633534405925e-05,
      "loss": 0.0004,
      "step": 14580
    },
    {
      "epoch": 6.805600933488915,
      "grad_norm": 0.32421875,
      "learning_rate": 4.6266257819656866e-05,
      "loss": 0.0042,
      "step": 14581
    },
    {
      "epoch": 6.8060676779463245,
      "grad_norm": 0.041015625,
      "learning_rate": 4.6253883261854924e-05,
      "loss": 0.0003,
      "step": 14582
    },
    {
      "epoch": 6.806534422403734,
      "grad_norm": 0.16015625,
      "learning_rate": 4.6241509861266474e-05,
      "loss": 0.0008,
      "step": 14583
    },
    {
      "epoch": 6.807001166861143,
      "grad_norm": 0.2333984375,
      "learning_rate": 4.622913761815797e-05,
      "loss": 0.0009,
      "step": 14584
    },
    {
      "epoch": 6.807467911318553,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.621676653279582e-05,
      "loss": 0.0005,
      "step": 14585
    },
    {
      "epoch": 6.807934655775963,
      "grad_norm": 0.08984375,
      "learning_rate": 4.6204396605446333e-05,
      "loss": 0.0004,
      "step": 14586
    },
    {
      "epoch": 6.808401400233373,
      "grad_norm": 0.0184326171875,
      "learning_rate": 4.6192027836375865e-05,
      "loss": 0.0003,
      "step": 14587
    },
    {
      "epoch": 6.8088681446907815,
      "grad_norm": 0.36328125,
      "learning_rate": 4.6179660225850776e-05,
      "loss": 0.002,
      "step": 14588
    },
    {
      "epoch": 6.809334889148191,
      "grad_norm": 0.048095703125,
      "learning_rate": 4.616729377413729e-05,
      "loss": 0.0003,
      "step": 14589
    },
    {
      "epoch": 6.809801633605601,
      "grad_norm": 0.01312255859375,
      "learning_rate": 4.6154928481501684e-05,
      "loss": 0.0002,
      "step": 14590
    },
    {
      "epoch": 6.810268378063011,
      "grad_norm": 0.017822265625,
      "learning_rate": 4.6142564348210246e-05,
      "loss": 0.0002,
      "step": 14591
    },
    {
      "epoch": 6.81073512252042,
      "grad_norm": 0.037109375,
      "learning_rate": 4.6130201374529126e-05,
      "loss": 0.0003,
      "step": 14592
    },
    {
      "epoch": 6.81120186697783,
      "grad_norm": 0.01373291015625,
      "learning_rate": 4.611783956072454e-05,
      "loss": 0.0002,
      "step": 14593
    },
    {
      "epoch": 6.811668611435239,
      "grad_norm": 0.02880859375,
      "learning_rate": 4.610547890706268e-05,
      "loss": 0.0003,
      "step": 14594
    },
    {
      "epoch": 6.812135355892648,
      "grad_norm": 0.02392578125,
      "learning_rate": 4.609311941380964e-05,
      "loss": 0.0003,
      "step": 14595
    },
    {
      "epoch": 6.812602100350058,
      "grad_norm": 0.03955078125,
      "learning_rate": 4.6080761081231525e-05,
      "loss": 0.0024,
      "step": 14596
    },
    {
      "epoch": 6.813068844807468,
      "grad_norm": 0.09716796875,
      "learning_rate": 4.606840390959449e-05,
      "loss": 0.0004,
      "step": 14597
    },
    {
      "epoch": 6.813535589264878,
      "grad_norm": 0.0556640625,
      "learning_rate": 4.6056047899164535e-05,
      "loss": 0.0021,
      "step": 14598
    },
    {
      "epoch": 6.8140023337222875,
      "grad_norm": 0.060791015625,
      "learning_rate": 4.6043693050207704e-05,
      "loss": 0.0004,
      "step": 14599
    },
    {
      "epoch": 6.814469078179696,
      "grad_norm": 0.0157470703125,
      "learning_rate": 4.6031339362990075e-05,
      "loss": 0.0001,
      "step": 14600
    },
    {
      "epoch": 6.814935822637106,
      "grad_norm": 0.014892578125,
      "learning_rate": 4.601898683777754e-05,
      "loss": 0.0002,
      "step": 14601
    },
    {
      "epoch": 6.815402567094516,
      "grad_norm": 0.11083984375,
      "learning_rate": 4.600663547483617e-05,
      "loss": 0.0005,
      "step": 14602
    },
    {
      "epoch": 6.815869311551925,
      "grad_norm": 0.1787109375,
      "learning_rate": 4.599428527443179e-05,
      "loss": 0.002,
      "step": 14603
    },
    {
      "epoch": 6.816336056009335,
      "grad_norm": 0.020263671875,
      "learning_rate": 4.598193623683037e-05,
      "loss": 0.0003,
      "step": 14604
    },
    {
      "epoch": 6.8168028004667445,
      "grad_norm": 0.197265625,
      "learning_rate": 4.596958836229783e-05,
      "loss": 0.0005,
      "step": 14605
    },
    {
      "epoch": 6.817269544924154,
      "grad_norm": 0.037353515625,
      "learning_rate": 4.595724165109998e-05,
      "loss": 0.0003,
      "step": 14606
    },
    {
      "epoch": 6.817736289381564,
      "grad_norm": 0.05126953125,
      "learning_rate": 4.5944896103502666e-05,
      "loss": 0.0004,
      "step": 14607
    },
    {
      "epoch": 6.818203033838973,
      "grad_norm": 0.07470703125,
      "learning_rate": 4.593255171977176e-05,
      "loss": 0.0005,
      "step": 14608
    },
    {
      "epoch": 6.818669778296383,
      "grad_norm": 0.158203125,
      "learning_rate": 4.592020850017296e-05,
      "loss": 0.0006,
      "step": 14609
    },
    {
      "epoch": 6.819136522753793,
      "grad_norm": 0.169921875,
      "learning_rate": 4.5907866444972104e-05,
      "loss": 0.0005,
      "step": 14610
    },
    {
      "epoch": 6.8196032672112015,
      "grad_norm": 0.049072265625,
      "learning_rate": 4.589552555443487e-05,
      "loss": 0.0004,
      "step": 14611
    },
    {
      "epoch": 6.820070011668611,
      "grad_norm": 0.07421875,
      "learning_rate": 4.588318582882704e-05,
      "loss": 0.0038,
      "step": 14612
    },
    {
      "epoch": 6.820536756126021,
      "grad_norm": 0.034423828125,
      "learning_rate": 4.587084726841422e-05,
      "loss": 0.0018,
      "step": 14613
    },
    {
      "epoch": 6.821003500583431,
      "grad_norm": 0.0255126953125,
      "learning_rate": 4.5858509873462164e-05,
      "loss": 0.0002,
      "step": 14614
    },
    {
      "epoch": 6.82147024504084,
      "grad_norm": 0.08447265625,
      "learning_rate": 4.5846173644236415e-05,
      "loss": 0.0004,
      "step": 14615
    },
    {
      "epoch": 6.82193698949825,
      "grad_norm": 0.453125,
      "learning_rate": 4.583383858100263e-05,
      "loss": 0.0021,
      "step": 14616
    },
    {
      "epoch": 6.822403733955659,
      "grad_norm": 0.1748046875,
      "learning_rate": 4.5821504684026436e-05,
      "loss": 0.001,
      "step": 14617
    },
    {
      "epoch": 6.822870478413069,
      "grad_norm": 0.01080322265625,
      "learning_rate": 4.580917195357333e-05,
      "loss": 0.0002,
      "step": 14618
    },
    {
      "epoch": 6.823337222870478,
      "grad_norm": 0.05615234375,
      "learning_rate": 4.579684038990887e-05,
      "loss": 0.0004,
      "step": 14619
    },
    {
      "epoch": 6.823803967327888,
      "grad_norm": 0.05908203125,
      "learning_rate": 4.578450999329862e-05,
      "loss": 0.0004,
      "step": 14620
    },
    {
      "epoch": 6.824270711785298,
      "grad_norm": 0.06201171875,
      "learning_rate": 4.577218076400798e-05,
      "loss": 0.0003,
      "step": 14621
    },
    {
      "epoch": 6.8247374562427074,
      "grad_norm": 0.0810546875,
      "learning_rate": 4.575985270230245e-05,
      "loss": 0.0045,
      "step": 14622
    },
    {
      "epoch": 6.825204200700116,
      "grad_norm": 0.0888671875,
      "learning_rate": 4.5747525808447524e-05,
      "loss": 0.0003,
      "step": 14623
    },
    {
      "epoch": 6.825670945157526,
      "grad_norm": 0.0203857421875,
      "learning_rate": 4.573520008270851e-05,
      "loss": 0.0002,
      "step": 14624
    },
    {
      "epoch": 6.826137689614936,
      "grad_norm": 0.09423828125,
      "learning_rate": 4.572287552535085e-05,
      "loss": 0.0005,
      "step": 14625
    },
    {
      "epoch": 6.826604434072346,
      "grad_norm": 0.484375,
      "learning_rate": 4.571055213663994e-05,
      "loss": 0.0019,
      "step": 14626
    },
    {
      "epoch": 6.827071178529755,
      "grad_norm": 0.68359375,
      "learning_rate": 4.569822991684104e-05,
      "loss": 0.0021,
      "step": 14627
    },
    {
      "epoch": 6.827537922987164,
      "grad_norm": 0.150390625,
      "learning_rate": 4.568590886621951e-05,
      "loss": 0.0005,
      "step": 14628
    },
    {
      "epoch": 6.828004667444574,
      "grad_norm": 0.07275390625,
      "learning_rate": 4.567358898504065e-05,
      "loss": 0.0046,
      "step": 14629
    },
    {
      "epoch": 6.828471411901984,
      "grad_norm": 0.2578125,
      "learning_rate": 4.5661270273569654e-05,
      "loss": 0.0026,
      "step": 14630
    },
    {
      "epoch": 6.828938156359393,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.564895273207184e-05,
      "loss": 0.0007,
      "step": 14631
    },
    {
      "epoch": 6.829404900816803,
      "grad_norm": 0.146484375,
      "learning_rate": 4.563663636081235e-05,
      "loss": 0.0008,
      "step": 14632
    },
    {
      "epoch": 6.8298716452742125,
      "grad_norm": 0.09375,
      "learning_rate": 4.562432116005638e-05,
      "loss": 0.0005,
      "step": 14633
    },
    {
      "epoch": 6.830338389731622,
      "grad_norm": 0.0264892578125,
      "learning_rate": 4.5612007130069155e-05,
      "loss": 0.0002,
      "step": 14634
    },
    {
      "epoch": 6.830805134189031,
      "grad_norm": 0.045166015625,
      "learning_rate": 4.559969427111571e-05,
      "loss": 0.0003,
      "step": 14635
    },
    {
      "epoch": 6.831271878646441,
      "grad_norm": 0.08544921875,
      "learning_rate": 4.5587382583461225e-05,
      "loss": 0.0006,
      "step": 14636
    },
    {
      "epoch": 6.831738623103851,
      "grad_norm": 0.04345703125,
      "learning_rate": 4.557507206737079e-05,
      "loss": 0.0021,
      "step": 14637
    },
    {
      "epoch": 6.83220536756126,
      "grad_norm": 0.3046875,
      "learning_rate": 4.556276272310941e-05,
      "loss": 0.0013,
      "step": 14638
    },
    {
      "epoch": 6.8326721120186695,
      "grad_norm": 0.0849609375,
      "learning_rate": 4.5550454550942136e-05,
      "loss": 0.0005,
      "step": 14639
    },
    {
      "epoch": 6.833138856476079,
      "grad_norm": 0.040771484375,
      "learning_rate": 4.5538147551134034e-05,
      "loss": 0.0003,
      "step": 14640
    },
    {
      "epoch": 6.833605600933489,
      "grad_norm": 0.0189208984375,
      "learning_rate": 4.552584172395e-05,
      "loss": 0.0002,
      "step": 14641
    },
    {
      "epoch": 6.834072345390899,
      "grad_norm": 0.0196533203125,
      "learning_rate": 4.5513537069655035e-05,
      "loss": 0.0003,
      "step": 14642
    },
    {
      "epoch": 6.834539089848308,
      "grad_norm": 0.08984375,
      "learning_rate": 4.5501233588514115e-05,
      "loss": 0.0006,
      "step": 14643
    },
    {
      "epoch": 6.835005834305718,
      "grad_norm": 0.0341796875,
      "learning_rate": 4.5488931280792055e-05,
      "loss": 0.0003,
      "step": 14644
    },
    {
      "epoch": 6.835472578763127,
      "grad_norm": 0.11767578125,
      "learning_rate": 4.5476630146753784e-05,
      "loss": 0.0011,
      "step": 14645
    },
    {
      "epoch": 6.835939323220536,
      "grad_norm": 0.04296875,
      "learning_rate": 4.546433018666421e-05,
      "loss": 0.0005,
      "step": 14646
    },
    {
      "epoch": 6.836406067677946,
      "grad_norm": 0.25,
      "learning_rate": 4.5452031400788075e-05,
      "loss": 0.0018,
      "step": 14647
    },
    {
      "epoch": 6.836872812135356,
      "grad_norm": 0.059326171875,
      "learning_rate": 4.5439733789390216e-05,
      "loss": 0.0004,
      "step": 14648
    },
    {
      "epoch": 6.837339556592766,
      "grad_norm": 0.1435546875,
      "learning_rate": 4.5427437352735466e-05,
      "loss": 0.0007,
      "step": 14649
    },
    {
      "epoch": 6.8378063010501755,
      "grad_norm": 0.322265625,
      "learning_rate": 4.5415142091088514e-05,
      "loss": 0.0038,
      "step": 14650
    },
    {
      "epoch": 6.838273045507584,
      "grad_norm": 0.032470703125,
      "learning_rate": 4.540284800471414e-05,
      "loss": 0.0002,
      "step": 14651
    },
    {
      "epoch": 6.838739789964994,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.5390555093876996e-05,
      "loss": 0.0051,
      "step": 14652
    },
    {
      "epoch": 6.839206534422404,
      "grad_norm": 0.2041015625,
      "learning_rate": 4.5378263358841786e-05,
      "loss": 0.001,
      "step": 14653
    },
    {
      "epoch": 6.839673278879813,
      "grad_norm": 0.012451171875,
      "learning_rate": 4.5365972799873205e-05,
      "loss": 0.0002,
      "step": 14654
    },
    {
      "epoch": 6.840140023337223,
      "grad_norm": 0.045166015625,
      "learning_rate": 4.535368341723581e-05,
      "loss": 0.0003,
      "step": 14655
    },
    {
      "epoch": 6.8406067677946325,
      "grad_norm": 0.052734375,
      "learning_rate": 4.534139521119425e-05,
      "loss": 0.0025,
      "step": 14656
    },
    {
      "epoch": 6.841073512252042,
      "grad_norm": 0.07568359375,
      "learning_rate": 4.5329108182013124e-05,
      "loss": 0.0036,
      "step": 14657
    },
    {
      "epoch": 6.841540256709451,
      "grad_norm": 0.06591796875,
      "learning_rate": 4.531682232995697e-05,
      "loss": 0.0031,
      "step": 14658
    },
    {
      "epoch": 6.842007001166861,
      "grad_norm": 0.0458984375,
      "learning_rate": 4.530453765529025e-05,
      "loss": 0.0028,
      "step": 14659
    },
    {
      "epoch": 6.842473745624271,
      "grad_norm": 0.037841796875,
      "learning_rate": 4.5292254158277526e-05,
      "loss": 0.0003,
      "step": 14660
    },
    {
      "epoch": 6.842940490081681,
      "grad_norm": 0.03173828125,
      "learning_rate": 4.527997183918331e-05,
      "loss": 0.0003,
      "step": 14661
    },
    {
      "epoch": 6.8434072345390895,
      "grad_norm": 0.05126953125,
      "learning_rate": 4.526769069827197e-05,
      "loss": 0.002,
      "step": 14662
    },
    {
      "epoch": 6.843873978996499,
      "grad_norm": 0.11279296875,
      "learning_rate": 4.5255410735808015e-05,
      "loss": 0.0004,
      "step": 14663
    },
    {
      "epoch": 6.844340723453909,
      "grad_norm": 0.052734375,
      "learning_rate": 4.5243131952055775e-05,
      "loss": 0.0003,
      "step": 14664
    },
    {
      "epoch": 6.844807467911319,
      "grad_norm": 0.0888671875,
      "learning_rate": 4.523085434727965e-05,
      "loss": 0.0034,
      "step": 14665
    },
    {
      "epoch": 6.845274212368728,
      "grad_norm": 0.03515625,
      "learning_rate": 4.521857792174404e-05,
      "loss": 0.0003,
      "step": 14666
    },
    {
      "epoch": 6.845740956826138,
      "grad_norm": 0.031494140625,
      "learning_rate": 4.5206302675713195e-05,
      "loss": 0.0004,
      "step": 14667
    },
    {
      "epoch": 6.846207701283547,
      "grad_norm": 0.224609375,
      "learning_rate": 4.5194028609451445e-05,
      "loss": 0.001,
      "step": 14668
    },
    {
      "epoch": 6.846674445740957,
      "grad_norm": 0.08154296875,
      "learning_rate": 4.5181755723223095e-05,
      "loss": 0.0004,
      "step": 14669
    },
    {
      "epoch": 6.847141190198366,
      "grad_norm": 0.057373046875,
      "learning_rate": 4.516948401729234e-05,
      "loss": 0.0004,
      "step": 14670
    },
    {
      "epoch": 6.847607934655776,
      "grad_norm": 0.04833984375,
      "learning_rate": 4.5157213491923435e-05,
      "loss": 0.0026,
      "step": 14671
    },
    {
      "epoch": 6.848074679113186,
      "grad_norm": 0.0164794921875,
      "learning_rate": 4.51449441473806e-05,
      "loss": 0.0003,
      "step": 14672
    },
    {
      "epoch": 6.8485414235705955,
      "grad_norm": 0.06591796875,
      "learning_rate": 4.513267598392794e-05,
      "loss": 0.0032,
      "step": 14673
    },
    {
      "epoch": 6.849008168028004,
      "grad_norm": 0.2265625,
      "learning_rate": 4.5120409001829646e-05,
      "loss": 0.0008,
      "step": 14674
    },
    {
      "epoch": 6.849474912485414,
      "grad_norm": 0.0220947265625,
      "learning_rate": 4.5108143201349876e-05,
      "loss": 0.0003,
      "step": 14675
    },
    {
      "epoch": 6.849941656942824,
      "grad_norm": 0.0296630859375,
      "learning_rate": 4.5095878582752646e-05,
      "loss": 0.0002,
      "step": 14676
    },
    {
      "epoch": 6.850408401400234,
      "grad_norm": 0.07861328125,
      "learning_rate": 4.508361514630207e-05,
      "loss": 0.0005,
      "step": 14677
    },
    {
      "epoch": 6.850875145857643,
      "grad_norm": 0.033203125,
      "learning_rate": 4.507135289226223e-05,
      "loss": 0.0003,
      "step": 14678
    },
    {
      "epoch": 6.8513418903150525,
      "grad_norm": 0.02587890625,
      "learning_rate": 4.505909182089707e-05,
      "loss": 0.0003,
      "step": 14679
    },
    {
      "epoch": 6.851808634772462,
      "grad_norm": 0.61328125,
      "learning_rate": 4.504683193247065e-05,
      "loss": 0.0025,
      "step": 14680
    },
    {
      "epoch": 6.852275379229871,
      "grad_norm": 0.08984375,
      "learning_rate": 4.503457322724688e-05,
      "loss": 0.0038,
      "step": 14681
    },
    {
      "epoch": 6.852742123687281,
      "grad_norm": 0.123046875,
      "learning_rate": 4.502231570548973e-05,
      "loss": 0.0006,
      "step": 14682
    },
    {
      "epoch": 6.853208868144691,
      "grad_norm": 0.11181640625,
      "learning_rate": 4.501005936746315e-05,
      "loss": 0.0007,
      "step": 14683
    },
    {
      "epoch": 6.853675612602101,
      "grad_norm": 0.087890625,
      "learning_rate": 4.499780421343098e-05,
      "loss": 0.0041,
      "step": 14684
    },
    {
      "epoch": 6.85414235705951,
      "grad_norm": 0.0146484375,
      "learning_rate": 4.498555024365711e-05,
      "loss": 0.0002,
      "step": 14685
    },
    {
      "epoch": 6.854609101516919,
      "grad_norm": 0.02685546875,
      "learning_rate": 4.4973297458405426e-05,
      "loss": 0.0003,
      "step": 14686
    },
    {
      "epoch": 6.855075845974329,
      "grad_norm": 0.041015625,
      "learning_rate": 4.496104585793966e-05,
      "loss": 0.0019,
      "step": 14687
    },
    {
      "epoch": 6.855542590431739,
      "grad_norm": 0.1806640625,
      "learning_rate": 4.494879544252363e-05,
      "loss": 0.0006,
      "step": 14688
    },
    {
      "epoch": 6.856009334889148,
      "grad_norm": 0.11865234375,
      "learning_rate": 4.493654621242116e-05,
      "loss": 0.0005,
      "step": 14689
    },
    {
      "epoch": 6.856476079346558,
      "grad_norm": 0.0201416015625,
      "learning_rate": 4.4924298167895903e-05,
      "loss": 0.0002,
      "step": 14690
    },
    {
      "epoch": 6.856942823803967,
      "grad_norm": 0.040283203125,
      "learning_rate": 4.491205130921162e-05,
      "loss": 0.0003,
      "step": 14691
    },
    {
      "epoch": 6.857409568261377,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.4899805636632044e-05,
      "loss": 0.0005,
      "step": 14692
    },
    {
      "epoch": 6.857876312718787,
      "grad_norm": 0.134765625,
      "learning_rate": 4.488756115042073e-05,
      "loss": 0.0005,
      "step": 14693
    },
    {
      "epoch": 6.858343057176196,
      "grad_norm": 0.1123046875,
      "learning_rate": 4.4875317850841384e-05,
      "loss": 0.0003,
      "step": 14694
    },
    {
      "epoch": 6.858809801633606,
      "grad_norm": 0.031982421875,
      "learning_rate": 4.4863075738157635e-05,
      "loss": 0.0021,
      "step": 14695
    },
    {
      "epoch": 6.8592765460910154,
      "grad_norm": 0.0142822265625,
      "learning_rate": 4.4850834812633e-05,
      "loss": 0.0002,
      "step": 14696
    },
    {
      "epoch": 6.859743290548424,
      "grad_norm": 0.021484375,
      "learning_rate": 4.483859507453109e-05,
      "loss": 0.0003,
      "step": 14697
    },
    {
      "epoch": 6.860210035005834,
      "grad_norm": 0.0576171875,
      "learning_rate": 4.482635652411547e-05,
      "loss": 0.0003,
      "step": 14698
    },
    {
      "epoch": 6.860676779463244,
      "grad_norm": 0.0111083984375,
      "learning_rate": 4.481411916164957e-05,
      "loss": 0.0002,
      "step": 14699
    },
    {
      "epoch": 6.861143523920654,
      "grad_norm": 0.048095703125,
      "learning_rate": 4.4801882987396945e-05,
      "loss": 0.0003,
      "step": 14700
    },
    {
      "epoch": 6.861610268378063,
      "grad_norm": 0.040283203125,
      "learning_rate": 4.4789648001620984e-05,
      "loss": 0.0004,
      "step": 14701
    },
    {
      "epoch": 6.8620770128354724,
      "grad_norm": 0.0155029296875,
      "learning_rate": 4.477741420458517e-05,
      "loss": 0.0002,
      "step": 14702
    },
    {
      "epoch": 6.862543757292882,
      "grad_norm": 0.026611328125,
      "learning_rate": 4.476518159655293e-05,
      "loss": 0.0002,
      "step": 14703
    },
    {
      "epoch": 6.863010501750292,
      "grad_norm": 0.0791015625,
      "learning_rate": 4.475295017778758e-05,
      "loss": 0.0005,
      "step": 14704
    },
    {
      "epoch": 6.863477246207701,
      "grad_norm": 0.07568359375,
      "learning_rate": 4.474071994855251e-05,
      "loss": 0.0025,
      "step": 14705
    },
    {
      "epoch": 6.863943990665111,
      "grad_norm": 0.1484375,
      "learning_rate": 4.472849090911109e-05,
      "loss": 0.0007,
      "step": 14706
    },
    {
      "epoch": 6.8644107351225205,
      "grad_norm": 0.07373046875,
      "learning_rate": 4.471626305972657e-05,
      "loss": 0.0025,
      "step": 14707
    },
    {
      "epoch": 6.86487747957993,
      "grad_norm": 0.035400390625,
      "learning_rate": 4.470403640066223e-05,
      "loss": 0.0003,
      "step": 14708
    },
    {
      "epoch": 6.865344224037339,
      "grad_norm": 0.041259765625,
      "learning_rate": 4.469181093218132e-05,
      "loss": 0.0002,
      "step": 14709
    },
    {
      "epoch": 6.865810968494749,
      "grad_norm": 0.0245361328125,
      "learning_rate": 4.4679586654547136e-05,
      "loss": 0.0003,
      "step": 14710
    },
    {
      "epoch": 6.866277712952159,
      "grad_norm": 0.019287109375,
      "learning_rate": 4.466736356802278e-05,
      "loss": 0.0003,
      "step": 14711
    },
    {
      "epoch": 6.866744457409569,
      "grad_norm": 0.0859375,
      "learning_rate": 4.4655141672871526e-05,
      "loss": 0.0003,
      "step": 14712
    },
    {
      "epoch": 6.8672112018669775,
      "grad_norm": 0.06298828125,
      "learning_rate": 4.464292096935644e-05,
      "loss": 0.002,
      "step": 14713
    },
    {
      "epoch": 6.867677946324387,
      "grad_norm": 0.0223388671875,
      "learning_rate": 4.4630701457740686e-05,
      "loss": 0.0003,
      "step": 14714
    },
    {
      "epoch": 6.868144690781797,
      "grad_norm": 0.0286865234375,
      "learning_rate": 4.461848313828739e-05,
      "loss": 0.0003,
      "step": 14715
    },
    {
      "epoch": 6.868611435239207,
      "grad_norm": 0.02294921875,
      "learning_rate": 4.4606266011259576e-05,
      "loss": 0.0003,
      "step": 14716
    },
    {
      "epoch": 6.869078179696616,
      "grad_norm": 0.06689453125,
      "learning_rate": 4.459405007692031e-05,
      "loss": 0.0033,
      "step": 14717
    },
    {
      "epoch": 6.869544924154026,
      "grad_norm": 0.0546875,
      "learning_rate": 4.4581835335532664e-05,
      "loss": 0.0027,
      "step": 14718
    },
    {
      "epoch": 6.870011668611435,
      "grad_norm": 0.0888671875,
      "learning_rate": 4.456962178735954e-05,
      "loss": 0.0005,
      "step": 14719
    },
    {
      "epoch": 6.870478413068845,
      "grad_norm": 0.0196533203125,
      "learning_rate": 4.455740943266397e-05,
      "loss": 0.0002,
      "step": 14720
    },
    {
      "epoch": 6.870945157526254,
      "grad_norm": 0.056640625,
      "learning_rate": 4.454519827170892e-05,
      "loss": 0.0003,
      "step": 14721
    },
    {
      "epoch": 6.871411901983664,
      "grad_norm": 0.10107421875,
      "learning_rate": 4.453298830475724e-05,
      "loss": 0.0007,
      "step": 14722
    },
    {
      "epoch": 6.871878646441074,
      "grad_norm": 0.021240234375,
      "learning_rate": 4.452077953207187e-05,
      "loss": 0.0002,
      "step": 14723
    },
    {
      "epoch": 6.872345390898483,
      "grad_norm": 0.038818359375,
      "learning_rate": 4.45085719539157e-05,
      "loss": 0.0003,
      "step": 14724
    },
    {
      "epoch": 6.872812135355892,
      "grad_norm": 0.0771484375,
      "learning_rate": 4.449636557055152e-05,
      "loss": 0.0005,
      "step": 14725
    },
    {
      "epoch": 6.873278879813302,
      "grad_norm": 0.0361328125,
      "learning_rate": 4.448416038224217e-05,
      "loss": 0.0003,
      "step": 14726
    },
    {
      "epoch": 6.873745624270712,
      "grad_norm": 0.0810546875,
      "learning_rate": 4.447195638925047e-05,
      "loss": 0.0009,
      "step": 14727
    },
    {
      "epoch": 6.874212368728122,
      "grad_norm": 0.01336669921875,
      "learning_rate": 4.445975359183913e-05,
      "loss": 0.0002,
      "step": 14728
    },
    {
      "epoch": 6.874679113185531,
      "grad_norm": 0.263671875,
      "learning_rate": 4.444755199027092e-05,
      "loss": 0.0013,
      "step": 14729
    },
    {
      "epoch": 6.8751458576429405,
      "grad_norm": 0.01385498046875,
      "learning_rate": 4.443535158480858e-05,
      "loss": 0.0002,
      "step": 14730
    },
    {
      "epoch": 6.87561260210035,
      "grad_norm": 0.01806640625,
      "learning_rate": 4.442315237571474e-05,
      "loss": 0.0002,
      "step": 14731
    },
    {
      "epoch": 6.876079346557759,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.441095436325212e-05,
      "loss": 0.0004,
      "step": 14732
    },
    {
      "epoch": 6.876546091015169,
      "grad_norm": 0.048828125,
      "learning_rate": 4.439875754768331e-05,
      "loss": 0.0022,
      "step": 14733
    },
    {
      "epoch": 6.877012835472579,
      "grad_norm": 0.052734375,
      "learning_rate": 4.438656192927093e-05,
      "loss": 0.0004,
      "step": 14734
    },
    {
      "epoch": 6.877479579929989,
      "grad_norm": 0.6015625,
      "learning_rate": 4.4374367508277614e-05,
      "loss": 0.0011,
      "step": 14735
    },
    {
      "epoch": 6.877946324387398,
      "grad_norm": 0.052978515625,
      "learning_rate": 4.4362174284965854e-05,
      "loss": 0.0027,
      "step": 14736
    },
    {
      "epoch": 6.878413068844807,
      "grad_norm": 0.03076171875,
      "learning_rate": 4.43499822595982e-05,
      "loss": 0.0002,
      "step": 14737
    },
    {
      "epoch": 6.878879813302217,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.433779143243722e-05,
      "loss": 0.0008,
      "step": 14738
    },
    {
      "epoch": 6.879346557759627,
      "grad_norm": 0.024169921875,
      "learning_rate": 4.43256018037453e-05,
      "loss": 0.0003,
      "step": 14739
    },
    {
      "epoch": 6.879813302217036,
      "grad_norm": 0.033447265625,
      "learning_rate": 4.431341337378495e-05,
      "loss": 0.0004,
      "step": 14740
    },
    {
      "epoch": 6.880280046674446,
      "grad_norm": 0.0257568359375,
      "learning_rate": 4.430122614281864e-05,
      "loss": 0.0003,
      "step": 14741
    },
    {
      "epoch": 6.880746791131855,
      "grad_norm": 0.039794921875,
      "learning_rate": 4.4289040111108685e-05,
      "loss": 0.0003,
      "step": 14742
    },
    {
      "epoch": 6.881213535589265,
      "grad_norm": 0.0150146484375,
      "learning_rate": 4.427685527891751e-05,
      "loss": 0.0003,
      "step": 14743
    },
    {
      "epoch": 6.881680280046674,
      "grad_norm": 0.02880859375,
      "learning_rate": 4.42646716465075e-05,
      "loss": 0.0003,
      "step": 14744
    },
    {
      "epoch": 6.882147024504084,
      "grad_norm": 0.201171875,
      "learning_rate": 4.425248921414093e-05,
      "loss": 0.0008,
      "step": 14745
    },
    {
      "epoch": 6.882613768961494,
      "grad_norm": 0.053955078125,
      "learning_rate": 4.4240307982080096e-05,
      "loss": 0.0004,
      "step": 14746
    },
    {
      "epoch": 6.8830805134189035,
      "grad_norm": 0.02099609375,
      "learning_rate": 4.422812795058735e-05,
      "loss": 0.0003,
      "step": 14747
    },
    {
      "epoch": 6.883547257876312,
      "grad_norm": 0.0308837890625,
      "learning_rate": 4.421594911992485e-05,
      "loss": 0.0003,
      "step": 14748
    },
    {
      "epoch": 6.884014002333722,
      "grad_norm": 0.038330078125,
      "learning_rate": 4.420377149035489e-05,
      "loss": 0.0022,
      "step": 14749
    },
    {
      "epoch": 6.884480746791132,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.41915950621396e-05,
      "loss": 0.0006,
      "step": 14750
    },
    {
      "epoch": 6.884947491248542,
      "grad_norm": 0.11279296875,
      "learning_rate": 4.417941983554119e-05,
      "loss": 0.0005,
      "step": 14751
    },
    {
      "epoch": 6.885414235705951,
      "grad_norm": 0.1474609375,
      "learning_rate": 4.4167245810821844e-05,
      "loss": 0.0007,
      "step": 14752
    },
    {
      "epoch": 6.8858809801633605,
      "grad_norm": 0.057373046875,
      "learning_rate": 4.415507298824361e-05,
      "loss": 0.0027,
      "step": 14753
    },
    {
      "epoch": 6.88634772462077,
      "grad_norm": 0.06640625,
      "learning_rate": 4.414290136806864e-05,
      "loss": 0.0004,
      "step": 14754
    },
    {
      "epoch": 6.88681446907818,
      "grad_norm": 0.0296630859375,
      "learning_rate": 4.4130730950558964e-05,
      "loss": 0.0003,
      "step": 14755
    },
    {
      "epoch": 6.887281213535589,
      "grad_norm": 0.0194091796875,
      "learning_rate": 4.411856173597666e-05,
      "loss": 0.0002,
      "step": 14756
    },
    {
      "epoch": 6.887747957992999,
      "grad_norm": 0.052978515625,
      "learning_rate": 4.4106393724583685e-05,
      "loss": 0.0023,
      "step": 14757
    },
    {
      "epoch": 6.888214702450409,
      "grad_norm": 0.03857421875,
      "learning_rate": 4.409422691664207e-05,
      "loss": 0.0003,
      "step": 14758
    },
    {
      "epoch": 6.8886814469078175,
      "grad_norm": 0.07177734375,
      "learning_rate": 4.4082061312413814e-05,
      "loss": 0.0003,
      "step": 14759
    },
    {
      "epoch": 6.889148191365227,
      "grad_norm": 0.09423828125,
      "learning_rate": 4.406989691216079e-05,
      "loss": 0.0005,
      "step": 14760
    },
    {
      "epoch": 6.889614935822637,
      "grad_norm": 0.018310546875,
      "learning_rate": 4.4057733716144965e-05,
      "loss": 0.0002,
      "step": 14761
    },
    {
      "epoch": 6.890081680280047,
      "grad_norm": 0.036376953125,
      "learning_rate": 4.404557172462817e-05,
      "loss": 0.0003,
      "step": 14762
    },
    {
      "epoch": 6.890548424737457,
      "grad_norm": 0.2333984375,
      "learning_rate": 4.403341093787231e-05,
      "loss": 0.004,
      "step": 14763
    },
    {
      "epoch": 6.891015169194866,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.4021251356139235e-05,
      "loss": 0.0009,
      "step": 14764
    },
    {
      "epoch": 6.891481913652275,
      "grad_norm": 0.062255859375,
      "learning_rate": 4.400909297969069e-05,
      "loss": 0.0004,
      "step": 14765
    },
    {
      "epoch": 6.891948658109685,
      "grad_norm": 0.0252685546875,
      "learning_rate": 4.3996935808788496e-05,
      "loss": 0.0002,
      "step": 14766
    },
    {
      "epoch": 6.892415402567094,
      "grad_norm": 0.03369140625,
      "learning_rate": 4.3984779843694454e-05,
      "loss": 0.0002,
      "step": 14767
    },
    {
      "epoch": 6.892882147024504,
      "grad_norm": 0.0167236328125,
      "learning_rate": 4.3972625084670215e-05,
      "loss": 0.0002,
      "step": 14768
    },
    {
      "epoch": 6.893348891481914,
      "grad_norm": 0.017578125,
      "learning_rate": 4.396047153197752e-05,
      "loss": 0.0002,
      "step": 14769
    },
    {
      "epoch": 6.8938156359393234,
      "grad_norm": 0.06591796875,
      "learning_rate": 4.3948319185878086e-05,
      "loss": 0.0003,
      "step": 14770
    },
    {
      "epoch": 6.894282380396733,
      "grad_norm": 0.130859375,
      "learning_rate": 4.393616804663349e-05,
      "loss": 0.0004,
      "step": 14771
    },
    {
      "epoch": 6.894749124854142,
      "grad_norm": 0.044189453125,
      "learning_rate": 4.3924018114505394e-05,
      "loss": 0.0003,
      "step": 14772
    },
    {
      "epoch": 6.895215869311552,
      "grad_norm": 0.0185546875,
      "learning_rate": 4.391186938975546e-05,
      "loss": 0.0002,
      "step": 14773
    },
    {
      "epoch": 6.895682613768962,
      "grad_norm": 0.0189208984375,
      "learning_rate": 4.3899721872645164e-05,
      "loss": 0.0002,
      "step": 14774
    },
    {
      "epoch": 6.896149358226371,
      "grad_norm": 0.02685546875,
      "learning_rate": 4.388757556343609e-05,
      "loss": 0.0003,
      "step": 14775
    },
    {
      "epoch": 6.8966161026837804,
      "grad_norm": 0.064453125,
      "learning_rate": 4.387543046238982e-05,
      "loss": 0.0003,
      "step": 14776
    },
    {
      "epoch": 6.89708284714119,
      "grad_norm": 0.0255126953125,
      "learning_rate": 4.3863286569767756e-05,
      "loss": 0.0003,
      "step": 14777
    },
    {
      "epoch": 6.8975495915986,
      "grad_norm": 0.0654296875,
      "learning_rate": 4.3851143885831436e-05,
      "loss": 0.0004,
      "step": 14778
    },
    {
      "epoch": 6.89801633605601,
      "grad_norm": 0.15234375,
      "learning_rate": 4.383900241084231e-05,
      "loss": 0.0005,
      "step": 14779
    },
    {
      "epoch": 6.898483080513419,
      "grad_norm": 0.028564453125,
      "learning_rate": 4.382686214506173e-05,
      "loss": 0.0003,
      "step": 14780
    },
    {
      "epoch": 6.8989498249708285,
      "grad_norm": 0.03662109375,
      "learning_rate": 4.3814723088751186e-05,
      "loss": 0.0003,
      "step": 14781
    },
    {
      "epoch": 6.899416569428238,
      "grad_norm": 0.038818359375,
      "learning_rate": 4.380258524217194e-05,
      "loss": 0.0003,
      "step": 14782
    },
    {
      "epoch": 6.899883313885647,
      "grad_norm": 0.0302734375,
      "learning_rate": 4.37904486055854e-05,
      "loss": 0.0003,
      "step": 14783
    },
    {
      "epoch": 6.900350058343057,
      "grad_norm": 0.044677734375,
      "learning_rate": 4.37783131792529e-05,
      "loss": 0.0004,
      "step": 14784
    },
    {
      "epoch": 6.900816802800467,
      "grad_norm": 0.306640625,
      "learning_rate": 4.376617896343566e-05,
      "loss": 0.0008,
      "step": 14785
    },
    {
      "epoch": 6.901283547257877,
      "grad_norm": 0.0830078125,
      "learning_rate": 4.3754045958394976e-05,
      "loss": 0.0004,
      "step": 14786
    },
    {
      "epoch": 6.9017502917152855,
      "grad_norm": 0.0361328125,
      "learning_rate": 4.374191416439213e-05,
      "loss": 0.0003,
      "step": 14787
    },
    {
      "epoch": 6.902217036172695,
      "grad_norm": 0.04833984375,
      "learning_rate": 4.3729783581688245e-05,
      "loss": 0.0003,
      "step": 14788
    },
    {
      "epoch": 6.902683780630105,
      "grad_norm": 0.03271484375,
      "learning_rate": 4.371765421054456e-05,
      "loss": 0.0003,
      "step": 14789
    },
    {
      "epoch": 6.903150525087515,
      "grad_norm": 0.2099609375,
      "learning_rate": 4.3705526051222266e-05,
      "loss": 0.0007,
      "step": 14790
    },
    {
      "epoch": 6.903617269544924,
      "grad_norm": 0.035888671875,
      "learning_rate": 4.369339910398241e-05,
      "loss": 0.0002,
      "step": 14791
    },
    {
      "epoch": 6.904084014002334,
      "grad_norm": 0.11083984375,
      "learning_rate": 4.3681273369086154e-05,
      "loss": 0.0005,
      "step": 14792
    },
    {
      "epoch": 6.904550758459743,
      "grad_norm": 0.11083984375,
      "learning_rate": 4.3669148846794595e-05,
      "loss": 0.0006,
      "step": 14793
    },
    {
      "epoch": 6.905017502917153,
      "grad_norm": 0.265625,
      "learning_rate": 4.3657025537368734e-05,
      "loss": 0.0005,
      "step": 14794
    },
    {
      "epoch": 6.905484247374562,
      "grad_norm": 0.0625,
      "learning_rate": 4.364490344106962e-05,
      "loss": 0.0003,
      "step": 14795
    },
    {
      "epoch": 6.905950991831972,
      "grad_norm": 0.04638671875,
      "learning_rate": 4.36327825581583e-05,
      "loss": 0.0022,
      "step": 14796
    },
    {
      "epoch": 6.906417736289382,
      "grad_norm": 0.02783203125,
      "learning_rate": 4.362066288889568e-05,
      "loss": 0.0002,
      "step": 14797
    },
    {
      "epoch": 6.9068844807467915,
      "grad_norm": 0.0296630859375,
      "learning_rate": 4.360854443354278e-05,
      "loss": 0.0019,
      "step": 14798
    },
    {
      "epoch": 6.9073512252042,
      "grad_norm": 0.29296875,
      "learning_rate": 4.3596427192360446e-05,
      "loss": 0.0012,
      "step": 14799
    },
    {
      "epoch": 6.90781796966161,
      "grad_norm": 0.123046875,
      "learning_rate": 4.358431116560962e-05,
      "loss": 0.0024,
      "step": 14800
    },
    {
      "epoch": 6.90828471411902,
      "grad_norm": 0.050537109375,
      "learning_rate": 4.35721963535512e-05,
      "loss": 0.0003,
      "step": 14801
    },
    {
      "epoch": 6.908751458576429,
      "grad_norm": 0.049560546875,
      "learning_rate": 4.3560082756446e-05,
      "loss": 0.0003,
      "step": 14802
    },
    {
      "epoch": 6.909218203033839,
      "grad_norm": 0.11669921875,
      "learning_rate": 4.35479703745548e-05,
      "loss": 0.0005,
      "step": 14803
    },
    {
      "epoch": 6.9096849474912485,
      "grad_norm": 0.01556396484375,
      "learning_rate": 4.353585920813843e-05,
      "loss": 0.0002,
      "step": 14804
    },
    {
      "epoch": 6.910151691948658,
      "grad_norm": 0.02978515625,
      "learning_rate": 4.35237492574577e-05,
      "loss": 0.0002,
      "step": 14805
    },
    {
      "epoch": 6.910618436406068,
      "grad_norm": 0.115234375,
      "learning_rate": 4.351164052277328e-05,
      "loss": 0.0005,
      "step": 14806
    },
    {
      "epoch": 6.911085180863477,
      "grad_norm": 0.059814453125,
      "learning_rate": 4.3499533004345905e-05,
      "loss": 0.0003,
      "step": 14807
    },
    {
      "epoch": 6.911551925320887,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.3487426702436316e-05,
      "loss": 0.0008,
      "step": 14808
    },
    {
      "epoch": 6.912018669778297,
      "grad_norm": 0.03125,
      "learning_rate": 4.34753216173051e-05,
      "loss": 0.0002,
      "step": 14809
    },
    {
      "epoch": 6.9124854142357055,
      "grad_norm": 0.1455078125,
      "learning_rate": 4.346321774921295e-05,
      "loss": 0.0005,
      "step": 14810
    },
    {
      "epoch": 6.912952158693115,
      "grad_norm": 0.02587890625,
      "learning_rate": 4.345111509842041e-05,
      "loss": 0.0002,
      "step": 14811
    },
    {
      "epoch": 6.913418903150525,
      "grad_norm": 0.208984375,
      "learning_rate": 4.3439013665188124e-05,
      "loss": 0.0008,
      "step": 14812
    },
    {
      "epoch": 6.913885647607935,
      "grad_norm": 0.08935546875,
      "learning_rate": 4.3426913449776655e-05,
      "loss": 0.0005,
      "step": 14813
    },
    {
      "epoch": 6.914352392065345,
      "grad_norm": 0.0223388671875,
      "learning_rate": 4.341481445244646e-05,
      "loss": 0.0002,
      "step": 14814
    },
    {
      "epoch": 6.914819136522754,
      "grad_norm": 0.015625,
      "learning_rate": 4.340271667345811e-05,
      "loss": 0.0002,
      "step": 14815
    },
    {
      "epoch": 6.915285880980163,
      "grad_norm": 0.010498046875,
      "learning_rate": 4.33906201130721e-05,
      "loss": 0.0002,
      "step": 14816
    },
    {
      "epoch": 6.915752625437573,
      "grad_norm": 0.0478515625,
      "learning_rate": 4.3378524771548804e-05,
      "loss": 0.0004,
      "step": 14817
    },
    {
      "epoch": 6.916219369894982,
      "grad_norm": 0.04833984375,
      "learning_rate": 4.336643064914869e-05,
      "loss": 0.0003,
      "step": 14818
    },
    {
      "epoch": 6.916686114352392,
      "grad_norm": 0.01708984375,
      "learning_rate": 4.3354337746132215e-05,
      "loss": 0.0002,
      "step": 14819
    },
    {
      "epoch": 6.917152858809802,
      "grad_norm": 0.021484375,
      "learning_rate": 4.334224606275964e-05,
      "loss": 0.0003,
      "step": 14820
    },
    {
      "epoch": 6.9176196032672115,
      "grad_norm": 0.119140625,
      "learning_rate": 4.333015559929138e-05,
      "loss": 0.0032,
      "step": 14821
    },
    {
      "epoch": 6.91808634772462,
      "grad_norm": 0.09033203125,
      "learning_rate": 4.331806635598778e-05,
      "loss": 0.0034,
      "step": 14822
    },
    {
      "epoch": 6.91855309218203,
      "grad_norm": 0.0247802734375,
      "learning_rate": 4.3305978333109064e-05,
      "loss": 0.0002,
      "step": 14823
    },
    {
      "epoch": 6.91901983663944,
      "grad_norm": 0.05078125,
      "learning_rate": 4.329389153091554e-05,
      "loss": 0.0022,
      "step": 14824
    },
    {
      "epoch": 6.91948658109685,
      "grad_norm": 0.1513671875,
      "learning_rate": 4.3281805949667486e-05,
      "loss": 0.0037,
      "step": 14825
    },
    {
      "epoch": 6.919953325554259,
      "grad_norm": 0.031005859375,
      "learning_rate": 4.326972158962504e-05,
      "loss": 0.0003,
      "step": 14826
    },
    {
      "epoch": 6.9204200700116685,
      "grad_norm": 0.1123046875,
      "learning_rate": 4.3257638451048434e-05,
      "loss": 0.0012,
      "step": 14827
    },
    {
      "epoch": 6.920886814469078,
      "grad_norm": 0.035400390625,
      "learning_rate": 4.324555653419786e-05,
      "loss": 0.0002,
      "step": 14828
    },
    {
      "epoch": 6.921353558926488,
      "grad_norm": 0.2421875,
      "learning_rate": 4.3233475839333394e-05,
      "loss": 0.0012,
      "step": 14829
    },
    {
      "epoch": 6.921820303383897,
      "grad_norm": 0.0264892578125,
      "learning_rate": 4.322139636671522e-05,
      "loss": 0.0003,
      "step": 14830
    },
    {
      "epoch": 6.922287047841307,
      "grad_norm": 0.045654296875,
      "learning_rate": 4.3209318116603324e-05,
      "loss": 0.0003,
      "step": 14831
    },
    {
      "epoch": 6.922753792298717,
      "grad_norm": 0.37890625,
      "learning_rate": 4.319724108925783e-05,
      "loss": 0.0011,
      "step": 14832
    },
    {
      "epoch": 6.923220536756126,
      "grad_norm": 0.0289306640625,
      "learning_rate": 4.318516528493879e-05,
      "loss": 0.0003,
      "step": 14833
    },
    {
      "epoch": 6.923687281213535,
      "grad_norm": 0.265625,
      "learning_rate": 4.317309070390615e-05,
      "loss": 0.0009,
      "step": 14834
    },
    {
      "epoch": 6.924154025670945,
      "grad_norm": 0.058837890625,
      "learning_rate": 4.3161017346419916e-05,
      "loss": 0.0017,
      "step": 14835
    },
    {
      "epoch": 6.924620770128355,
      "grad_norm": 0.01953125,
      "learning_rate": 4.3148945212740076e-05,
      "loss": 0.0003,
      "step": 14836
    },
    {
      "epoch": 6.925087514585765,
      "grad_norm": 0.05419921875,
      "learning_rate": 4.313687430312649e-05,
      "loss": 0.0003,
      "step": 14837
    },
    {
      "epoch": 6.925554259043174,
      "grad_norm": 0.1416015625,
      "learning_rate": 4.312480461783909e-05,
      "loss": 0.0029,
      "step": 14838
    },
    {
      "epoch": 6.926021003500583,
      "grad_norm": 0.035400390625,
      "learning_rate": 4.311273615713778e-05,
      "loss": 0.0003,
      "step": 14839
    },
    {
      "epoch": 6.926487747957993,
      "grad_norm": 0.2255859375,
      "learning_rate": 4.310066892128235e-05,
      "loss": 0.0013,
      "step": 14840
    },
    {
      "epoch": 6.926954492415403,
      "grad_norm": 0.02294921875,
      "learning_rate": 4.308860291053264e-05,
      "loss": 0.0003,
      "step": 14841
    },
    {
      "epoch": 6.927421236872812,
      "grad_norm": 0.07861328125,
      "learning_rate": 4.3076538125148505e-05,
      "loss": 0.0004,
      "step": 14842
    },
    {
      "epoch": 6.927887981330222,
      "grad_norm": 0.026611328125,
      "learning_rate": 4.306447456538961e-05,
      "loss": 0.0002,
      "step": 14843
    },
    {
      "epoch": 6.9283547257876315,
      "grad_norm": 0.0478515625,
      "learning_rate": 4.305241223151576e-05,
      "loss": 0.0004,
      "step": 14844
    },
    {
      "epoch": 6.92882147024504,
      "grad_norm": 0.03466796875,
      "learning_rate": 4.3040351123786684e-05,
      "loss": 0.0003,
      "step": 14845
    },
    {
      "epoch": 6.92928821470245,
      "grad_norm": 0.02978515625,
      "learning_rate": 4.3028291242462014e-05,
      "loss": 0.0003,
      "step": 14846
    },
    {
      "epoch": 6.92975495915986,
      "grad_norm": 0.10693359375,
      "learning_rate": 4.301623258780145e-05,
      "loss": 0.0003,
      "step": 14847
    },
    {
      "epoch": 6.93022170361727,
      "grad_norm": 0.02685546875,
      "learning_rate": 4.3004175160064665e-05,
      "loss": 0.0003,
      "step": 14848
    },
    {
      "epoch": 6.9306884480746795,
      "grad_norm": 0.01251220703125,
      "learning_rate": 4.299211895951122e-05,
      "loss": 0.0001,
      "step": 14849
    },
    {
      "epoch": 6.9311551925320884,
      "grad_norm": 0.011474609375,
      "learning_rate": 4.298006398640066e-05,
      "loss": 0.0002,
      "step": 14850
    },
    {
      "epoch": 6.931621936989498,
      "grad_norm": 0.015625,
      "learning_rate": 4.296801024099263e-05,
      "loss": 0.0002,
      "step": 14851
    },
    {
      "epoch": 6.932088681446908,
      "grad_norm": 0.07568359375,
      "learning_rate": 4.295595772354658e-05,
      "loss": 0.0003,
      "step": 14852
    },
    {
      "epoch": 6.932555425904317,
      "grad_norm": 0.111328125,
      "learning_rate": 4.2943906434322056e-05,
      "loss": 0.004,
      "step": 14853
    },
    {
      "epoch": 6.933022170361727,
      "grad_norm": 0.0341796875,
      "learning_rate": 4.293185637357856e-05,
      "loss": 0.0002,
      "step": 14854
    },
    {
      "epoch": 6.9334889148191365,
      "grad_norm": 0.046875,
      "learning_rate": 4.2919807541575486e-05,
      "loss": 0.0003,
      "step": 14855
    },
    {
      "epoch": 6.933955659276546,
      "grad_norm": 0.01239013671875,
      "learning_rate": 4.290775993857229e-05,
      "loss": 0.0002,
      "step": 14856
    },
    {
      "epoch": 6.934422403733956,
      "grad_norm": 0.01104736328125,
      "learning_rate": 4.28957135648284e-05,
      "loss": 0.0002,
      "step": 14857
    },
    {
      "epoch": 6.934889148191365,
      "grad_norm": 0.04345703125,
      "learning_rate": 4.288366842060312e-05,
      "loss": 0.0003,
      "step": 14858
    },
    {
      "epoch": 6.935355892648775,
      "grad_norm": 0.111328125,
      "learning_rate": 4.287162450615582e-05,
      "loss": 0.0003,
      "step": 14859
    },
    {
      "epoch": 6.935822637106185,
      "grad_norm": 0.0220947265625,
      "learning_rate": 4.2859581821745876e-05,
      "loss": 0.0003,
      "step": 14860
    },
    {
      "epoch": 6.9362893815635935,
      "grad_norm": 0.0164794921875,
      "learning_rate": 4.284754036763249e-05,
      "loss": 0.0002,
      "step": 14861
    },
    {
      "epoch": 6.936756126021003,
      "grad_norm": 0.0250244140625,
      "learning_rate": 4.283550014407501e-05,
      "loss": 0.0002,
      "step": 14862
    },
    {
      "epoch": 6.937222870478413,
      "grad_norm": 0.031494140625,
      "learning_rate": 4.282346115133261e-05,
      "loss": 0.0002,
      "step": 14863
    },
    {
      "epoch": 6.937689614935823,
      "grad_norm": 0.0142822265625,
      "learning_rate": 4.2811423389664515e-05,
      "loss": 0.0002,
      "step": 14864
    },
    {
      "epoch": 6.938156359393232,
      "grad_norm": 0.01312255859375,
      "learning_rate": 4.279938685932998e-05,
      "loss": 0.0002,
      "step": 14865
    },
    {
      "epoch": 6.938623103850642,
      "grad_norm": 0.006866455078125,
      "learning_rate": 4.2787351560588064e-05,
      "loss": 0.0002,
      "step": 14866
    },
    {
      "epoch": 6.939089848308051,
      "grad_norm": 0.0693359375,
      "learning_rate": 4.2775317493697954e-05,
      "loss": 0.0004,
      "step": 14867
    },
    {
      "epoch": 6.939556592765461,
      "grad_norm": 0.109375,
      "learning_rate": 4.276328465891878e-05,
      "loss": 0.0004,
      "step": 14868
    },
    {
      "epoch": 6.94002333722287,
      "grad_norm": 0.0230712890625,
      "learning_rate": 4.2751253056509556e-05,
      "loss": 0.0003,
      "step": 14869
    },
    {
      "epoch": 6.94049008168028,
      "grad_norm": 0.1259765625,
      "learning_rate": 4.273922268672937e-05,
      "loss": 0.0004,
      "step": 14870
    },
    {
      "epoch": 6.94095682613769,
      "grad_norm": 0.1396484375,
      "learning_rate": 4.2727193549837296e-05,
      "loss": 0.0004,
      "step": 14871
    },
    {
      "epoch": 6.9414235705950995,
      "grad_norm": 0.296875,
      "learning_rate": 4.271516564609225e-05,
      "loss": 0.0028,
      "step": 14872
    },
    {
      "epoch": 6.941890315052508,
      "grad_norm": 0.07080078125,
      "learning_rate": 4.270313897575324e-05,
      "loss": 0.0004,
      "step": 14873
    },
    {
      "epoch": 6.942357059509918,
      "grad_norm": 0.03662109375,
      "learning_rate": 4.269111353907926e-05,
      "loss": 0.0003,
      "step": 14874
    },
    {
      "epoch": 6.942823803967328,
      "grad_norm": 0.0262451171875,
      "learning_rate": 4.267908933632917e-05,
      "loss": 0.0003,
      "step": 14875
    },
    {
      "epoch": 6.943290548424738,
      "grad_norm": 0.17578125,
      "learning_rate": 4.266706636776187e-05,
      "loss": 0.0005,
      "step": 14876
    },
    {
      "epoch": 6.943757292882147,
      "grad_norm": 0.10302734375,
      "learning_rate": 4.2655044633636275e-05,
      "loss": 0.0003,
      "step": 14877
    },
    {
      "epoch": 6.9442240373395565,
      "grad_norm": 0.0380859375,
      "learning_rate": 4.2643024134211175e-05,
      "loss": 0.002,
      "step": 14878
    },
    {
      "epoch": 6.944690781796966,
      "grad_norm": 0.173828125,
      "learning_rate": 4.263100486974543e-05,
      "loss": 0.001,
      "step": 14879
    },
    {
      "epoch": 6.945157526254376,
      "grad_norm": 0.0301513671875,
      "learning_rate": 4.261898684049776e-05,
      "loss": 0.0003,
      "step": 14880
    },
    {
      "epoch": 6.945624270711785,
      "grad_norm": 0.06005859375,
      "learning_rate": 4.260697004672697e-05,
      "loss": 0.0032,
      "step": 14881
    },
    {
      "epoch": 6.946091015169195,
      "grad_norm": 0.045654296875,
      "learning_rate": 4.259495448869184e-05,
      "loss": 0.0003,
      "step": 14882
    },
    {
      "epoch": 6.946557759626605,
      "grad_norm": 0.023193359375,
      "learning_rate": 4.2582940166650986e-05,
      "loss": 0.0003,
      "step": 14883
    },
    {
      "epoch": 6.947024504084014,
      "grad_norm": 0.294921875,
      "learning_rate": 4.257092708086314e-05,
      "loss": 0.0019,
      "step": 14884
    },
    {
      "epoch": 6.947491248541423,
      "grad_norm": 0.150390625,
      "learning_rate": 4.255891523158699e-05,
      "loss": 0.0006,
      "step": 14885
    },
    {
      "epoch": 6.947957992998833,
      "grad_norm": 0.035888671875,
      "learning_rate": 4.254690461908108e-05,
      "loss": 0.0002,
      "step": 14886
    },
    {
      "epoch": 6.948424737456243,
      "grad_norm": 0.031982421875,
      "learning_rate": 4.253489524360408e-05,
      "loss": 0.0025,
      "step": 14887
    },
    {
      "epoch": 6.948891481913652,
      "grad_norm": 0.0947265625,
      "learning_rate": 4.252288710541457e-05,
      "loss": 0.0004,
      "step": 14888
    },
    {
      "epoch": 6.949358226371062,
      "grad_norm": 0.0546875,
      "learning_rate": 4.251088020477103e-05,
      "loss": 0.0003,
      "step": 14889
    },
    {
      "epoch": 6.949824970828471,
      "grad_norm": 0.267578125,
      "learning_rate": 4.2498874541932035e-05,
      "loss": 0.001,
      "step": 14890
    },
    {
      "epoch": 6.950291715285881,
      "grad_norm": 0.03271484375,
      "learning_rate": 4.2486870117156094e-05,
      "loss": 0.0003,
      "step": 14891
    },
    {
      "epoch": 6.950758459743291,
      "grad_norm": 0.1435546875,
      "learning_rate": 4.247486693070162e-05,
      "loss": 0.0006,
      "step": 14892
    },
    {
      "epoch": 6.9512252042007,
      "grad_norm": 0.01318359375,
      "learning_rate": 4.246286498282709e-05,
      "loss": 0.0002,
      "step": 14893
    },
    {
      "epoch": 6.95169194865811,
      "grad_norm": 0.1123046875,
      "learning_rate": 4.2450864273790944e-05,
      "loss": 0.0007,
      "step": 14894
    },
    {
      "epoch": 6.9521586931155195,
      "grad_norm": 0.060546875,
      "learning_rate": 4.2438864803851505e-05,
      "loss": 0.0004,
      "step": 14895
    },
    {
      "epoch": 6.952625437572928,
      "grad_norm": 0.015869140625,
      "learning_rate": 4.242686657326719e-05,
      "loss": 0.0003,
      "step": 14896
    },
    {
      "epoch": 6.953092182030338,
      "grad_norm": 0.0128173828125,
      "learning_rate": 4.2414869582296365e-05,
      "loss": 0.0002,
      "step": 14897
    },
    {
      "epoch": 6.953558926487748,
      "grad_norm": 0.05126953125,
      "learning_rate": 4.240287383119723e-05,
      "loss": 0.0031,
      "step": 14898
    },
    {
      "epoch": 6.954025670945158,
      "grad_norm": 0.0172119140625,
      "learning_rate": 4.239087932022813e-05,
      "loss": 0.0002,
      "step": 14899
    },
    {
      "epoch": 6.954492415402568,
      "grad_norm": 0.283203125,
      "learning_rate": 4.237888604964735e-05,
      "loss": 0.0036,
      "step": 14900
    },
    {
      "epoch": 6.9549591598599765,
      "grad_norm": 0.1611328125,
      "learning_rate": 4.2366894019713045e-05,
      "loss": 0.0004,
      "step": 14901
    },
    {
      "epoch": 6.955425904317386,
      "grad_norm": 0.06689453125,
      "learning_rate": 4.2354903230683465e-05,
      "loss": 0.0006,
      "step": 14902
    },
    {
      "epoch": 6.955892648774796,
      "grad_norm": 0.05712890625,
      "learning_rate": 4.23429136828168e-05,
      "loss": 0.0004,
      "step": 14903
    },
    {
      "epoch": 6.956359393232205,
      "grad_norm": 0.138671875,
      "learning_rate": 4.233092537637114e-05,
      "loss": 0.0041,
      "step": 14904
    },
    {
      "epoch": 6.956826137689615,
      "grad_norm": 0.02783203125,
      "learning_rate": 4.231893831160465e-05,
      "loss": 0.0002,
      "step": 14905
    },
    {
      "epoch": 6.957292882147025,
      "grad_norm": 0.0859375,
      "learning_rate": 4.230695248877545e-05,
      "loss": 0.0019,
      "step": 14906
    },
    {
      "epoch": 6.957759626604434,
      "grad_norm": 0.04736328125,
      "learning_rate": 4.229496790814153e-05,
      "loss": 0.0005,
      "step": 14907
    },
    {
      "epoch": 6.958226371061843,
      "grad_norm": 0.01025390625,
      "learning_rate": 4.2282984569960994e-05,
      "loss": 0.0002,
      "step": 14908
    },
    {
      "epoch": 6.958693115519253,
      "grad_norm": 0.0245361328125,
      "learning_rate": 4.227100247449186e-05,
      "loss": 0.0002,
      "step": 14909
    },
    {
      "epoch": 6.959159859976663,
      "grad_norm": 0.0244140625,
      "learning_rate": 4.225902162199207e-05,
      "loss": 0.0003,
      "step": 14910
    },
    {
      "epoch": 6.959626604434073,
      "grad_norm": 0.0751953125,
      "learning_rate": 4.224704201271963e-05,
      "loss": 0.0023,
      "step": 14911
    },
    {
      "epoch": 6.960093348891482,
      "grad_norm": 0.220703125,
      "learning_rate": 4.223506364693243e-05,
      "loss": 0.0027,
      "step": 14912
    },
    {
      "epoch": 6.960560093348891,
      "grad_norm": 0.058837890625,
      "learning_rate": 4.222308652488839e-05,
      "loss": 0.0003,
      "step": 14913
    },
    {
      "epoch": 6.961026837806301,
      "grad_norm": 0.0133056640625,
      "learning_rate": 4.2211110646845445e-05,
      "loss": 0.0002,
      "step": 14914
    },
    {
      "epoch": 6.961493582263711,
      "grad_norm": 0.0537109375,
      "learning_rate": 4.219913601306137e-05,
      "loss": 0.004,
      "step": 14915
    },
    {
      "epoch": 6.96196032672112,
      "grad_norm": 0.017822265625,
      "learning_rate": 4.218716262379403e-05,
      "loss": 0.0003,
      "step": 14916
    },
    {
      "epoch": 6.96242707117853,
      "grad_norm": 0.09375,
      "learning_rate": 4.217519047930125e-05,
      "loss": 0.0005,
      "step": 14917
    },
    {
      "epoch": 6.9628938156359395,
      "grad_norm": 0.042236328125,
      "learning_rate": 4.216321957984075e-05,
      "loss": 0.0002,
      "step": 14918
    },
    {
      "epoch": 6.963360560093349,
      "grad_norm": 0.0361328125,
      "learning_rate": 4.21512499256703e-05,
      "loss": 0.0002,
      "step": 14919
    },
    {
      "epoch": 6.963827304550758,
      "grad_norm": 0.07275390625,
      "learning_rate": 4.213928151704766e-05,
      "loss": 0.0004,
      "step": 14920
    },
    {
      "epoch": 6.964294049008168,
      "grad_norm": 0.013671875,
      "learning_rate": 4.212731435423044e-05,
      "loss": 0.0002,
      "step": 14921
    },
    {
      "epoch": 6.964760793465578,
      "grad_norm": 0.0546875,
      "learning_rate": 4.211534843747637e-05,
      "loss": 0.0003,
      "step": 14922
    },
    {
      "epoch": 6.9652275379229875,
      "grad_norm": 0.0177001953125,
      "learning_rate": 4.21033837670431e-05,
      "loss": 0.0002,
      "step": 14923
    },
    {
      "epoch": 6.9656942823803965,
      "grad_norm": 0.013427734375,
      "learning_rate": 4.209142034318818e-05,
      "loss": 0.0003,
      "step": 14924
    },
    {
      "epoch": 6.966161026837806,
      "grad_norm": 0.01904296875,
      "learning_rate": 4.207945816616923e-05,
      "loss": 0.0002,
      "step": 14925
    },
    {
      "epoch": 6.966627771295216,
      "grad_norm": 0.0184326171875,
      "learning_rate": 4.206749723624385e-05,
      "loss": 0.0002,
      "step": 14926
    },
    {
      "epoch": 6.967094515752626,
      "grad_norm": 0.04150390625,
      "learning_rate": 4.205553755366949e-05,
      "loss": 0.0018,
      "step": 14927
    },
    {
      "epoch": 6.967561260210035,
      "grad_norm": 0.019775390625,
      "learning_rate": 4.204357911870373e-05,
      "loss": 0.0002,
      "step": 14928
    },
    {
      "epoch": 6.9680280046674445,
      "grad_norm": 0.03076171875,
      "learning_rate": 4.203162193160398e-05,
      "loss": 0.0002,
      "step": 14929
    },
    {
      "epoch": 6.968494749124854,
      "grad_norm": 0.01226806640625,
      "learning_rate": 4.201966599262773e-05,
      "loss": 0.0001,
      "step": 14930
    },
    {
      "epoch": 6.968961493582263,
      "grad_norm": 0.037841796875,
      "learning_rate": 4.200771130203244e-05,
      "loss": 0.0003,
      "step": 14931
    },
    {
      "epoch": 6.969428238039673,
      "grad_norm": 0.05859375,
      "learning_rate": 4.1995757860075436e-05,
      "loss": 0.0035,
      "step": 14932
    },
    {
      "epoch": 6.969894982497083,
      "grad_norm": 0.2060546875,
      "learning_rate": 4.198380566701411e-05,
      "loss": 0.0008,
      "step": 14933
    },
    {
      "epoch": 6.970361726954493,
      "grad_norm": 0.056640625,
      "learning_rate": 4.1971854723105865e-05,
      "loss": 0.0005,
      "step": 14934
    },
    {
      "epoch": 6.970828471411902,
      "grad_norm": 0.0167236328125,
      "learning_rate": 4.1959905028607926e-05,
      "loss": 0.0002,
      "step": 14935
    },
    {
      "epoch": 6.971295215869311,
      "grad_norm": 0.46484375,
      "learning_rate": 4.194795658377764e-05,
      "loss": 0.002,
      "step": 14936
    },
    {
      "epoch": 6.971761960326721,
      "grad_norm": 0.07666015625,
      "learning_rate": 4.193600938887229e-05,
      "loss": 0.0043,
      "step": 14937
    },
    {
      "epoch": 6.972228704784131,
      "grad_norm": 0.01092529296875,
      "learning_rate": 4.1924063444149044e-05,
      "loss": 0.0002,
      "step": 14938
    },
    {
      "epoch": 6.97269544924154,
      "grad_norm": 0.04296875,
      "learning_rate": 4.191211874986515e-05,
      "loss": 0.0004,
      "step": 14939
    },
    {
      "epoch": 6.97316219369895,
      "grad_norm": 0.095703125,
      "learning_rate": 4.190017530627781e-05,
      "loss": 0.0004,
      "step": 14940
    },
    {
      "epoch": 6.973628938156359,
      "grad_norm": 0.38671875,
      "learning_rate": 4.188823311364414e-05,
      "loss": 0.0044,
      "step": 14941
    },
    {
      "epoch": 6.974095682613769,
      "grad_norm": 0.146484375,
      "learning_rate": 4.187629217222129e-05,
      "loss": 0.0004,
      "step": 14942
    },
    {
      "epoch": 6.974562427071179,
      "grad_norm": 0.05078125,
      "learning_rate": 4.186435248226638e-05,
      "loss": 0.0029,
      "step": 14943
    },
    {
      "epoch": 6.975029171528588,
      "grad_norm": 0.08203125,
      "learning_rate": 4.1852414044036434e-05,
      "loss": 0.0024,
      "step": 14944
    },
    {
      "epoch": 6.975495915985998,
      "grad_norm": 0.00946044921875,
      "learning_rate": 4.1840476857788566e-05,
      "loss": 0.0002,
      "step": 14945
    },
    {
      "epoch": 6.9759626604434075,
      "grad_norm": 0.0341796875,
      "learning_rate": 4.182854092377977e-05,
      "loss": 0.0021,
      "step": 14946
    },
    {
      "epoch": 6.976429404900816,
      "grad_norm": 0.0595703125,
      "learning_rate": 4.181660624226697e-05,
      "loss": 0.0028,
      "step": 14947
    },
    {
      "epoch": 6.976896149358226,
      "grad_norm": 0.0213623046875,
      "learning_rate": 4.180467281350722e-05,
      "loss": 0.0002,
      "step": 14948
    },
    {
      "epoch": 6.977362893815636,
      "grad_norm": 0.0257568359375,
      "learning_rate": 4.179274063775745e-05,
      "loss": 0.0002,
      "step": 14949
    },
    {
      "epoch": 6.977829638273046,
      "grad_norm": 0.0194091796875,
      "learning_rate": 4.1780809715274536e-05,
      "loss": 0.0002,
      "step": 14950
    },
    {
      "epoch": 6.978296382730455,
      "grad_norm": 0.025390625,
      "learning_rate": 4.176888004631537e-05,
      "loss": 0.0002,
      "step": 14951
    },
    {
      "epoch": 6.9787631271878645,
      "grad_norm": 0.0269775390625,
      "learning_rate": 4.1756951631136866e-05,
      "loss": 0.0017,
      "step": 14952
    },
    {
      "epoch": 6.979229871645274,
      "grad_norm": 0.0235595703125,
      "learning_rate": 4.174502446999579e-05,
      "loss": 0.0002,
      "step": 14953
    },
    {
      "epoch": 6.979696616102684,
      "grad_norm": 0.091796875,
      "learning_rate": 4.173309856314895e-05,
      "loss": 0.0004,
      "step": 14954
    },
    {
      "epoch": 6.980163360560093,
      "grad_norm": 0.0296630859375,
      "learning_rate": 4.1721173910853196e-05,
      "loss": 0.0002,
      "step": 14955
    },
    {
      "epoch": 6.980630105017503,
      "grad_norm": 0.0186767578125,
      "learning_rate": 4.1709250513365196e-05,
      "loss": 0.0002,
      "step": 14956
    },
    {
      "epoch": 6.981096849474913,
      "grad_norm": 0.041748046875,
      "learning_rate": 4.16973283709417e-05,
      "loss": 0.0019,
      "step": 14957
    },
    {
      "epoch": 6.981563593932322,
      "grad_norm": 0.0247802734375,
      "learning_rate": 4.1685407483839455e-05,
      "loss": 0.0002,
      "step": 14958
    },
    {
      "epoch": 6.982030338389731,
      "grad_norm": 0.06396484375,
      "learning_rate": 4.167348785231504e-05,
      "loss": 0.0003,
      "step": 14959
    },
    {
      "epoch": 6.982497082847141,
      "grad_norm": 0.046142578125,
      "learning_rate": 4.16615694766252e-05,
      "loss": 0.0002,
      "step": 14960
    },
    {
      "epoch": 6.982963827304551,
      "grad_norm": 0.036376953125,
      "learning_rate": 4.1649652357026445e-05,
      "loss": 0.0002,
      "step": 14961
    },
    {
      "epoch": 6.983430571761961,
      "grad_norm": 0.043701171875,
      "learning_rate": 4.1637736493775425e-05,
      "loss": 0.0003,
      "step": 14962
    },
    {
      "epoch": 6.98389731621937,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.1625821887128724e-05,
      "loss": 0.0005,
      "step": 14963
    },
    {
      "epoch": 6.984364060676779,
      "grad_norm": 0.059814453125,
      "learning_rate": 4.161390853734281e-05,
      "loss": 0.0003,
      "step": 14964
    },
    {
      "epoch": 6.984830805134189,
      "grad_norm": 0.023681640625,
      "learning_rate": 4.160199644467422e-05,
      "loss": 0.0002,
      "step": 14965
    },
    {
      "epoch": 6.985297549591599,
      "grad_norm": 0.037841796875,
      "learning_rate": 4.159008560937948e-05,
      "loss": 0.0004,
      "step": 14966
    },
    {
      "epoch": 6.985764294049008,
      "grad_norm": 0.0228271484375,
      "learning_rate": 4.1578176031714976e-05,
      "loss": 0.0002,
      "step": 14967
    },
    {
      "epoch": 6.986231038506418,
      "grad_norm": 0.03173828125,
      "learning_rate": 4.1566267711937154e-05,
      "loss": 0.0002,
      "step": 14968
    },
    {
      "epoch": 6.9866977829638275,
      "grad_norm": 0.0830078125,
      "learning_rate": 4.155436065030246e-05,
      "loss": 0.0003,
      "step": 14969
    },
    {
      "epoch": 6.987164527421237,
      "grad_norm": 0.053466796875,
      "learning_rate": 4.154245484706719e-05,
      "loss": 0.0003,
      "step": 14970
    },
    {
      "epoch": 6.987631271878646,
      "grad_norm": 0.1884765625,
      "learning_rate": 4.1530550302487725e-05,
      "loss": 0.0005,
      "step": 14971
    },
    {
      "epoch": 6.988098016336056,
      "grad_norm": 0.08984375,
      "learning_rate": 4.151864701682042e-05,
      "loss": 0.0004,
      "step": 14972
    },
    {
      "epoch": 6.988564760793466,
      "grad_norm": 0.1220703125,
      "learning_rate": 4.15067449903215e-05,
      "loss": 0.0004,
      "step": 14973
    },
    {
      "epoch": 6.989031505250875,
      "grad_norm": 0.01953125,
      "learning_rate": 4.1494844223247265e-05,
      "loss": 0.0002,
      "step": 14974
    },
    {
      "epoch": 6.9894982497082845,
      "grad_norm": 0.0272216796875,
      "learning_rate": 4.148294471585397e-05,
      "loss": 0.0003,
      "step": 14975
    },
    {
      "epoch": 6.989964994165694,
      "grad_norm": 0.08203125,
      "learning_rate": 4.147104646839777e-05,
      "loss": 0.0032,
      "step": 14976
    },
    {
      "epoch": 6.990431738623104,
      "grad_norm": 0.0771484375,
      "learning_rate": 4.145914948113492e-05,
      "loss": 0.0004,
      "step": 14977
    },
    {
      "epoch": 6.990898483080514,
      "grad_norm": 0.053955078125,
      "learning_rate": 4.144725375432149e-05,
      "loss": 0.002,
      "step": 14978
    },
    {
      "epoch": 6.991365227537923,
      "grad_norm": 0.00994873046875,
      "learning_rate": 4.143535928821366e-05,
      "loss": 0.0002,
      "step": 14979
    },
    {
      "epoch": 6.991831971995333,
      "grad_norm": 0.0081787109375,
      "learning_rate": 4.142346608306755e-05,
      "loss": 0.0002,
      "step": 14980
    },
    {
      "epoch": 6.992298716452742,
      "grad_norm": 0.05615234375,
      "learning_rate": 4.141157413913917e-05,
      "loss": 0.0003,
      "step": 14981
    },
    {
      "epoch": 6.992765460910151,
      "grad_norm": 0.0186767578125,
      "learning_rate": 4.139968345668461e-05,
      "loss": 0.0002,
      "step": 14982
    },
    {
      "epoch": 6.993232205367561,
      "grad_norm": 0.1025390625,
      "learning_rate": 4.1387794035959924e-05,
      "loss": 0.0021,
      "step": 14983
    },
    {
      "epoch": 6.993698949824971,
      "grad_norm": 0.060546875,
      "learning_rate": 4.1375905877221024e-05,
      "loss": 0.0003,
      "step": 14984
    },
    {
      "epoch": 6.994165694282381,
      "grad_norm": 0.05322265625,
      "learning_rate": 4.1364018980723906e-05,
      "loss": 0.0003,
      "step": 14985
    },
    {
      "epoch": 6.9946324387397905,
      "grad_norm": 0.023681640625,
      "learning_rate": 4.1352133346724565e-05,
      "loss": 0.0002,
      "step": 14986
    },
    {
      "epoch": 6.995099183197199,
      "grad_norm": 0.0306396484375,
      "learning_rate": 4.1340248975478825e-05,
      "loss": 0.0003,
      "step": 14987
    },
    {
      "epoch": 6.995565927654609,
      "grad_norm": 0.0191650390625,
      "learning_rate": 4.1328365867242604e-05,
      "loss": 0.0002,
      "step": 14988
    },
    {
      "epoch": 6.996032672112019,
      "grad_norm": 0.0198974609375,
      "learning_rate": 4.131648402227181e-05,
      "loss": 0.0002,
      "step": 14989
    },
    {
      "epoch": 6.996499416569428,
      "grad_norm": 0.033203125,
      "learning_rate": 4.1304603440822176e-05,
      "loss": 0.0003,
      "step": 14990
    },
    {
      "epoch": 6.996966161026838,
      "grad_norm": 0.0400390625,
      "learning_rate": 4.129272412314956e-05,
      "loss": 0.003,
      "step": 14991
    },
    {
      "epoch": 6.9974329054842475,
      "grad_norm": 0.01458740234375,
      "learning_rate": 4.128084606950976e-05,
      "loss": 0.0002,
      "step": 14992
    },
    {
      "epoch": 6.997899649941657,
      "grad_norm": 0.01953125,
      "learning_rate": 4.12689692801585e-05,
      "loss": 0.0003,
      "step": 14993
    },
    {
      "epoch": 6.998366394399066,
      "grad_norm": 0.047119140625,
      "learning_rate": 4.125709375535146e-05,
      "loss": 0.0015,
      "step": 14994
    },
    {
      "epoch": 6.998833138856476,
      "grad_norm": 0.0311279296875,
      "learning_rate": 4.12452194953444e-05,
      "loss": 0.0003,
      "step": 14995
    },
    {
      "epoch": 6.999299883313886,
      "grad_norm": 0.045166015625,
      "learning_rate": 4.123334650039292e-05,
      "loss": 0.0003,
      "step": 14996
    },
    {
      "epoch": 6.9997666277712955,
      "grad_norm": 0.359375,
      "learning_rate": 4.12214747707527e-05,
      "loss": 0.001,
      "step": 14997
    },
    {
      "epoch": 7.0002333722287045,
      "grad_norm": 0.0849609375,
      "learning_rate": 4.1209604306679375e-05,
      "loss": 0.0046,
      "step": 14998
    },
    {
      "epoch": 7.000700116686114,
      "grad_norm": 0.042724609375,
      "learning_rate": 4.119773510842846e-05,
      "loss": 0.0017,
      "step": 14999
    },
    {
      "epoch": 7.001166861143524,
      "grad_norm": 0.03271484375,
      "learning_rate": 4.118586717625555e-05,
      "loss": 0.0003,
      "step": 15000
    },
    {
      "epoch": 7.001633605600934,
      "grad_norm": 0.130859375,
      "learning_rate": 4.117400051041621e-05,
      "loss": 0.0005,
      "step": 15001
    },
    {
      "epoch": 7.002100350058343,
      "grad_norm": 0.01348876953125,
      "learning_rate": 4.116213511116587e-05,
      "loss": 0.0002,
      "step": 15002
    },
    {
      "epoch": 7.0025670945157525,
      "grad_norm": 0.007537841796875,
      "learning_rate": 4.1150270978760044e-05,
      "loss": 0.0002,
      "step": 15003
    },
    {
      "epoch": 7.003033838973162,
      "grad_norm": 0.023681640625,
      "learning_rate": 4.1138408113454216e-05,
      "loss": 0.0003,
      "step": 15004
    },
    {
      "epoch": 7.003500583430572,
      "grad_norm": 0.0126953125,
      "learning_rate": 4.1126546515503736e-05,
      "loss": 0.0002,
      "step": 15005
    },
    {
      "epoch": 7.003967327887981,
      "grad_norm": 0.0164794921875,
      "learning_rate": 4.111468618516403e-05,
      "loss": 0.0002,
      "step": 15006
    },
    {
      "epoch": 7.004434072345391,
      "grad_norm": 0.051513671875,
      "learning_rate": 4.11028271226905e-05,
      "loss": 0.0003,
      "step": 15007
    },
    {
      "epoch": 7.004900816802801,
      "grad_norm": 0.00927734375,
      "learning_rate": 4.1090969328338416e-05,
      "loss": 0.0002,
      "step": 15008
    },
    {
      "epoch": 7.004900816802801,
      "eval_loss": 2.36678409576416,
      "eval_runtime": 84.4731,
      "eval_samples_per_second": 21.356,
      "eval_steps_per_second": 2.675,
      "step": 15008
    },
    {
      "epoch": 7.00536756126021,
      "grad_norm": 0.022705078125,
      "learning_rate": 4.107911280236315e-05,
      "loss": 0.0002,
      "step": 15009
    },
    {
      "epoch": 7.005834305717619,
      "grad_norm": 0.08837890625,
      "learning_rate": 4.106725754501993e-05,
      "loss": 0.0003,
      "step": 15010
    },
    {
      "epoch": 7.006301050175029,
      "grad_norm": 0.12158203125,
      "learning_rate": 4.105540355656404e-05,
      "loss": 0.0004,
      "step": 15011
    },
    {
      "epoch": 7.006767794632439,
      "grad_norm": 0.01025390625,
      "learning_rate": 4.104355083725076e-05,
      "loss": 0.0002,
      "step": 15012
    },
    {
      "epoch": 7.007234539089849,
      "grad_norm": 0.018310546875,
      "learning_rate": 4.103169938733519e-05,
      "loss": 0.0003,
      "step": 15013
    },
    {
      "epoch": 7.007701283547258,
      "grad_norm": 0.00689697265625,
      "learning_rate": 4.1019849207072556e-05,
      "loss": 0.0002,
      "step": 15014
    },
    {
      "epoch": 7.008168028004667,
      "grad_norm": 0.0152587890625,
      "learning_rate": 4.1008000296718065e-05,
      "loss": 0.0003,
      "step": 15015
    },
    {
      "epoch": 7.008634772462077,
      "grad_norm": 0.016845703125,
      "learning_rate": 4.0996152656526734e-05,
      "loss": 0.0003,
      "step": 15016
    },
    {
      "epoch": 7.009101516919487,
      "grad_norm": 0.026611328125,
      "learning_rate": 4.098430628675369e-05,
      "loss": 0.0003,
      "step": 15017
    },
    {
      "epoch": 7.009568261376896,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.0972461187654066e-05,
      "loss": 0.001,
      "step": 15018
    },
    {
      "epoch": 7.010035005834306,
      "grad_norm": 0.036376953125,
      "learning_rate": 4.09606173594828e-05,
      "loss": 0.0003,
      "step": 15019
    },
    {
      "epoch": 7.0105017502917155,
      "grad_norm": 0.036376953125,
      "learning_rate": 4.094877480249495e-05,
      "loss": 0.0003,
      "step": 15020
    },
    {
      "epoch": 7.010968494749124,
      "grad_norm": 0.0146484375,
      "learning_rate": 4.093693351694553e-05,
      "loss": 0.0003,
      "step": 15021
    },
    {
      "epoch": 7.011435239206534,
      "grad_norm": 0.020751953125,
      "learning_rate": 4.092509350308943e-05,
      "loss": 0.0002,
      "step": 15022
    },
    {
      "epoch": 7.011901983663944,
      "grad_norm": 0.030517578125,
      "learning_rate": 4.0913254761181605e-05,
      "loss": 0.0003,
      "step": 15023
    },
    {
      "epoch": 7.012368728121354,
      "grad_norm": 0.019287109375,
      "learning_rate": 4.090141729147701e-05,
      "loss": 0.0002,
      "step": 15024
    },
    {
      "epoch": 7.012835472578763,
      "grad_norm": 0.0106201171875,
      "learning_rate": 4.088958109423042e-05,
      "loss": 0.0002,
      "step": 15025
    },
    {
      "epoch": 7.0133022170361725,
      "grad_norm": 0.00970458984375,
      "learning_rate": 4.087774616969675e-05,
      "loss": 0.0002,
      "step": 15026
    },
    {
      "epoch": 7.013768961493582,
      "grad_norm": 0.01519775390625,
      "learning_rate": 4.086591251813083e-05,
      "loss": 0.0002,
      "step": 15027
    },
    {
      "epoch": 7.014235705950992,
      "grad_norm": 0.0264892578125,
      "learning_rate": 4.085408013978739e-05,
      "loss": 0.0016,
      "step": 15028
    },
    {
      "epoch": 7.014702450408401,
      "grad_norm": 0.023193359375,
      "learning_rate": 4.0842249034921265e-05,
      "loss": 0.0002,
      "step": 15029
    },
    {
      "epoch": 7.015169194865811,
      "grad_norm": 0.0341796875,
      "learning_rate": 4.083041920378712e-05,
      "loss": 0.0002,
      "step": 15030
    },
    {
      "epoch": 7.015635939323221,
      "grad_norm": 0.01397705078125,
      "learning_rate": 4.0818590646639685e-05,
      "loss": 0.0002,
      "step": 15031
    },
    {
      "epoch": 7.01610268378063,
      "grad_norm": 0.043701171875,
      "learning_rate": 4.080676336373371e-05,
      "loss": 0.0002,
      "step": 15032
    },
    {
      "epoch": 7.016569428238039,
      "grad_norm": 0.0184326171875,
      "learning_rate": 4.079493735532375e-05,
      "loss": 0.0003,
      "step": 15033
    },
    {
      "epoch": 7.017036172695449,
      "grad_norm": 0.0184326171875,
      "learning_rate": 4.078311262166449e-05,
      "loss": 0.0002,
      "step": 15034
    },
    {
      "epoch": 7.017502917152859,
      "grad_norm": 0.0286865234375,
      "learning_rate": 4.077128916301054e-05,
      "loss": 0.0002,
      "step": 15035
    },
    {
      "epoch": 7.017969661610269,
      "grad_norm": 0.041015625,
      "learning_rate": 4.0759466979616414e-05,
      "loss": 0.0003,
      "step": 15036
    },
    {
      "epoch": 7.018436406067678,
      "grad_norm": 0.029052734375,
      "learning_rate": 4.07476460717367e-05,
      "loss": 0.0002,
      "step": 15037
    },
    {
      "epoch": 7.018903150525087,
      "grad_norm": 0.0517578125,
      "learning_rate": 4.073582643962594e-05,
      "loss": 0.0003,
      "step": 15038
    },
    {
      "epoch": 7.019369894982497,
      "grad_norm": 0.04931640625,
      "learning_rate": 4.072400808353856e-05,
      "loss": 0.0028,
      "step": 15039
    },
    {
      "epoch": 7.019836639439907,
      "grad_norm": 0.126953125,
      "learning_rate": 4.071219100372905e-05,
      "loss": 0.0004,
      "step": 15040
    },
    {
      "epoch": 7.020303383897316,
      "grad_norm": 0.0177001953125,
      "learning_rate": 4.070037520045191e-05,
      "loss": 0.0002,
      "step": 15041
    },
    {
      "epoch": 7.020770128354726,
      "grad_norm": 0.0257568359375,
      "learning_rate": 4.068856067396143e-05,
      "loss": 0.0002,
      "step": 15042
    },
    {
      "epoch": 7.0212368728121355,
      "grad_norm": 0.04541015625,
      "learning_rate": 4.067674742451204e-05,
      "loss": 0.0044,
      "step": 15043
    },
    {
      "epoch": 7.021703617269545,
      "grad_norm": 0.032958984375,
      "learning_rate": 4.0664935452358135e-05,
      "loss": 0.0006,
      "step": 15044
    },
    {
      "epoch": 7.022170361726954,
      "grad_norm": 0.0145263671875,
      "learning_rate": 4.065312475775397e-05,
      "loss": 0.0002,
      "step": 15045
    },
    {
      "epoch": 7.022637106184364,
      "grad_norm": 0.043212890625,
      "learning_rate": 4.064131534095387e-05,
      "loss": 0.0003,
      "step": 15046
    },
    {
      "epoch": 7.023103850641774,
      "grad_norm": 0.010986328125,
      "learning_rate": 4.062950720221215e-05,
      "loss": 0.0001,
      "step": 15047
    },
    {
      "epoch": 7.023570595099184,
      "grad_norm": 0.00927734375,
      "learning_rate": 4.0617700341782974e-05,
      "loss": 0.0002,
      "step": 15048
    },
    {
      "epoch": 7.0240373395565925,
      "grad_norm": 0.03515625,
      "learning_rate": 4.060589475992059e-05,
      "loss": 0.0002,
      "step": 15049
    },
    {
      "epoch": 7.024504084014002,
      "grad_norm": 0.025146484375,
      "learning_rate": 4.059409045687924e-05,
      "loss": 0.0002,
      "step": 15050
    },
    {
      "epoch": 7.024970828471412,
      "grad_norm": 0.045654296875,
      "learning_rate": 4.0582287432912993e-05,
      "loss": 0.0022,
      "step": 15051
    },
    {
      "epoch": 7.025437572928822,
      "grad_norm": 0.031494140625,
      "learning_rate": 4.0570485688276015e-05,
      "loss": 0.0002,
      "step": 15052
    },
    {
      "epoch": 7.025904317386231,
      "grad_norm": 0.044921875,
      "learning_rate": 4.0558685223222456e-05,
      "loss": 0.0003,
      "step": 15053
    },
    {
      "epoch": 7.026371061843641,
      "grad_norm": 0.03515625,
      "learning_rate": 4.054688603800634e-05,
      "loss": 0.0019,
      "step": 15054
    },
    {
      "epoch": 7.02683780630105,
      "grad_norm": 0.03466796875,
      "learning_rate": 4.053508813288171e-05,
      "loss": 0.0002,
      "step": 15055
    },
    {
      "epoch": 7.02730455075846,
      "grad_norm": 0.01129150390625,
      "learning_rate": 4.052329150810265e-05,
      "loss": 0.0002,
      "step": 15056
    },
    {
      "epoch": 7.027771295215869,
      "grad_norm": 0.1982421875,
      "learning_rate": 4.051149616392308e-05,
      "loss": 0.0034,
      "step": 15057
    },
    {
      "epoch": 7.028238039673279,
      "grad_norm": 0.197265625,
      "learning_rate": 4.0499702100597035e-05,
      "loss": 0.0016,
      "step": 15058
    },
    {
      "epoch": 7.028704784130689,
      "grad_norm": 0.048583984375,
      "learning_rate": 4.0487909318378394e-05,
      "loss": 0.0027,
      "step": 15059
    },
    {
      "epoch": 7.029171528588098,
      "grad_norm": 0.00860595703125,
      "learning_rate": 4.047611781752109e-05,
      "loss": 0.0001,
      "step": 15060
    },
    {
      "epoch": 7.029638273045507,
      "grad_norm": 0.0181884765625,
      "learning_rate": 4.046432759827904e-05,
      "loss": 0.0002,
      "step": 15061
    },
    {
      "epoch": 7.030105017502917,
      "grad_norm": 0.0272216796875,
      "learning_rate": 4.0452538660906046e-05,
      "loss": 0.0003,
      "step": 15062
    },
    {
      "epoch": 7.030571761960327,
      "grad_norm": 0.0303955078125,
      "learning_rate": 4.0440751005655964e-05,
      "loss": 0.0003,
      "step": 15063
    },
    {
      "epoch": 7.031038506417736,
      "grad_norm": 0.0546875,
      "learning_rate": 4.042896463278263e-05,
      "loss": 0.0003,
      "step": 15064
    },
    {
      "epoch": 7.031505250875146,
      "grad_norm": 0.0174560546875,
      "learning_rate": 4.041717954253974e-05,
      "loss": 0.0002,
      "step": 15065
    },
    {
      "epoch": 7.0319719953325555,
      "grad_norm": 0.01904296875,
      "learning_rate": 4.0405395735181085e-05,
      "loss": 0.0002,
      "step": 15066
    },
    {
      "epoch": 7.032438739789965,
      "grad_norm": 0.0235595703125,
      "learning_rate": 4.039361321096042e-05,
      "loss": 0.0002,
      "step": 15067
    },
    {
      "epoch": 7.032905484247374,
      "grad_norm": 0.01275634765625,
      "learning_rate": 4.038183197013137e-05,
      "loss": 0.0002,
      "step": 15068
    },
    {
      "epoch": 7.033372228704784,
      "grad_norm": 0.013427734375,
      "learning_rate": 4.0370052012947614e-05,
      "loss": 0.0002,
      "step": 15069
    },
    {
      "epoch": 7.033838973162194,
      "grad_norm": 0.412109375,
      "learning_rate": 4.035827333966283e-05,
      "loss": 0.0009,
      "step": 15070
    },
    {
      "epoch": 7.0343057176196035,
      "grad_norm": 0.01318359375,
      "learning_rate": 4.034649595053056e-05,
      "loss": 0.0002,
      "step": 15071
    },
    {
      "epoch": 7.0347724620770125,
      "grad_norm": 0.025146484375,
      "learning_rate": 4.033471984580443e-05,
      "loss": 0.0003,
      "step": 15072
    },
    {
      "epoch": 7.035239206534422,
      "grad_norm": 0.0128173828125,
      "learning_rate": 4.0322945025738014e-05,
      "loss": 0.0002,
      "step": 15073
    },
    {
      "epoch": 7.035705950991832,
      "grad_norm": 0.01458740234375,
      "learning_rate": 4.031117149058477e-05,
      "loss": 0.0002,
      "step": 15074
    },
    {
      "epoch": 7.036172695449242,
      "grad_norm": 0.0302734375,
      "learning_rate": 4.0299399240598224e-05,
      "loss": 0.0003,
      "step": 15075
    },
    {
      "epoch": 7.036639439906651,
      "grad_norm": 0.00799560546875,
      "learning_rate": 4.028762827603188e-05,
      "loss": 0.0001,
      "step": 15076
    },
    {
      "epoch": 7.0371061843640605,
      "grad_norm": 0.0458984375,
      "learning_rate": 4.027585859713913e-05,
      "loss": 0.0005,
      "step": 15077
    },
    {
      "epoch": 7.03757292882147,
      "grad_norm": 0.0830078125,
      "learning_rate": 4.026409020417344e-05,
      "loss": 0.0004,
      "step": 15078
    },
    {
      "epoch": 7.03803967327888,
      "grad_norm": 0.035400390625,
      "learning_rate": 4.0252323097388125e-05,
      "loss": 0.0002,
      "step": 15079
    },
    {
      "epoch": 7.038506417736289,
      "grad_norm": 0.04833984375,
      "learning_rate": 4.02405572770366e-05,
      "loss": 0.0003,
      "step": 15080
    },
    {
      "epoch": 7.038973162193699,
      "grad_norm": 0.006988525390625,
      "learning_rate": 4.02287927433722e-05,
      "loss": 0.0001,
      "step": 15081
    },
    {
      "epoch": 7.039439906651109,
      "grad_norm": 0.00927734375,
      "learning_rate": 4.0217029496648184e-05,
      "loss": 0.0002,
      "step": 15082
    },
    {
      "epoch": 7.039906651108518,
      "grad_norm": 0.0250244140625,
      "learning_rate": 4.020526753711784e-05,
      "loss": 0.0002,
      "step": 15083
    },
    {
      "epoch": 7.040373395565927,
      "grad_norm": 0.048095703125,
      "learning_rate": 4.019350686503448e-05,
      "loss": 0.0007,
      "step": 15084
    },
    {
      "epoch": 7.040840140023337,
      "grad_norm": 0.10107421875,
      "learning_rate": 4.0181747480651235e-05,
      "loss": 0.0004,
      "step": 15085
    },
    {
      "epoch": 7.041306884480747,
      "grad_norm": 0.016845703125,
      "learning_rate": 4.016998938422133e-05,
      "loss": 0.0002,
      "step": 15086
    },
    {
      "epoch": 7.041773628938157,
      "grad_norm": 0.038818359375,
      "learning_rate": 4.015823257599798e-05,
      "loss": 0.0022,
      "step": 15087
    },
    {
      "epoch": 7.042240373395566,
      "grad_norm": 0.2294921875,
      "learning_rate": 4.014647705623426e-05,
      "loss": 0.0006,
      "step": 15088
    },
    {
      "epoch": 7.042707117852975,
      "grad_norm": 0.0101318359375,
      "learning_rate": 4.013472282518329e-05,
      "loss": 0.0001,
      "step": 15089
    },
    {
      "epoch": 7.043173862310385,
      "grad_norm": 0.0260009765625,
      "learning_rate": 4.0122969883098184e-05,
      "loss": 0.0005,
      "step": 15090
    },
    {
      "epoch": 7.043640606767795,
      "grad_norm": 0.0125732421875,
      "learning_rate": 4.011121823023193e-05,
      "loss": 0.0003,
      "step": 15091
    },
    {
      "epoch": 7.044107351225204,
      "grad_norm": 0.1162109375,
      "learning_rate": 4.0099467866837595e-05,
      "loss": 0.0004,
      "step": 15092
    },
    {
      "epoch": 7.044574095682614,
      "grad_norm": 0.016357421875,
      "learning_rate": 4.008771879316822e-05,
      "loss": 0.0002,
      "step": 15093
    },
    {
      "epoch": 7.0450408401400235,
      "grad_norm": 0.095703125,
      "learning_rate": 4.007597100947669e-05,
      "loss": 0.0003,
      "step": 15094
    },
    {
      "epoch": 7.045507584597433,
      "grad_norm": 0.0185546875,
      "learning_rate": 4.006422451601599e-05,
      "loss": 0.0002,
      "step": 15095
    },
    {
      "epoch": 7.045974329054842,
      "grad_norm": 0.07275390625,
      "learning_rate": 4.0052479313039083e-05,
      "loss": 0.0003,
      "step": 15096
    },
    {
      "epoch": 7.046441073512252,
      "grad_norm": 0.01409912109375,
      "learning_rate": 4.004073540079876e-05,
      "loss": 0.0002,
      "step": 15097
    },
    {
      "epoch": 7.046907817969662,
      "grad_norm": 0.00927734375,
      "learning_rate": 4.002899277954794e-05,
      "loss": 0.0002,
      "step": 15098
    },
    {
      "epoch": 7.047374562427072,
      "grad_norm": 0.0263671875,
      "learning_rate": 4.0017251449539474e-05,
      "loss": 0.0001,
      "step": 15099
    },
    {
      "epoch": 7.0478413068844805,
      "grad_norm": 0.11181640625,
      "learning_rate": 4.000551141102611e-05,
      "loss": 0.0004,
      "step": 15100
    },
    {
      "epoch": 7.04830805134189,
      "grad_norm": 0.0517578125,
      "learning_rate": 3.9993772664260645e-05,
      "loss": 0.0032,
      "step": 15101
    },
    {
      "epoch": 7.0487747957993,
      "grad_norm": 0.0262451171875,
      "learning_rate": 3.998203520949586e-05,
      "loss": 0.0003,
      "step": 15102
    },
    {
      "epoch": 7.049241540256709,
      "grad_norm": 0.03173828125,
      "learning_rate": 3.997029904698442e-05,
      "loss": 0.0018,
      "step": 15103
    },
    {
      "epoch": 7.049708284714119,
      "grad_norm": 0.01806640625,
      "learning_rate": 3.995856417697904e-05,
      "loss": 0.0002,
      "step": 15104
    },
    {
      "epoch": 7.050175029171529,
      "grad_norm": 0.023681640625,
      "learning_rate": 3.994683059973243e-05,
      "loss": 0.0002,
      "step": 15105
    },
    {
      "epoch": 7.050641773628938,
      "grad_norm": 0.029052734375,
      "learning_rate": 3.993509831549715e-05,
      "loss": 0.0019,
      "step": 15106
    },
    {
      "epoch": 7.051108518086347,
      "grad_norm": 0.0101318359375,
      "learning_rate": 3.9923367324525896e-05,
      "loss": 0.0002,
      "step": 15107
    },
    {
      "epoch": 7.051575262543757,
      "grad_norm": 0.04345703125,
      "learning_rate": 3.991163762707115e-05,
      "loss": 0.0003,
      "step": 15108
    },
    {
      "epoch": 7.052042007001167,
      "grad_norm": 0.107421875,
      "learning_rate": 3.989990922338551e-05,
      "loss": 0.0003,
      "step": 15109
    },
    {
      "epoch": 7.052508751458577,
      "grad_norm": 0.0184326171875,
      "learning_rate": 3.988818211372155e-05,
      "loss": 0.0002,
      "step": 15110
    },
    {
      "epoch": 7.052975495915986,
      "grad_norm": 0.0147705078125,
      "learning_rate": 3.98764562983317e-05,
      "loss": 0.0002,
      "step": 15111
    },
    {
      "epoch": 7.053442240373395,
      "grad_norm": 0.01416015625,
      "learning_rate": 3.986473177746843e-05,
      "loss": 0.0002,
      "step": 15112
    },
    {
      "epoch": 7.053908984830805,
      "grad_norm": 0.052978515625,
      "learning_rate": 3.9853008551384264e-05,
      "loss": 0.0002,
      "step": 15113
    },
    {
      "epoch": 7.054375729288215,
      "grad_norm": 0.010986328125,
      "learning_rate": 3.984128662033151e-05,
      "loss": 0.0002,
      "step": 15114
    },
    {
      "epoch": 7.054842473745624,
      "grad_norm": 0.0084228515625,
      "learning_rate": 3.982956598456261e-05,
      "loss": 0.0002,
      "step": 15115
    },
    {
      "epoch": 7.055309218203034,
      "grad_norm": 0.138671875,
      "learning_rate": 3.9817846644329936e-05,
      "loss": 0.0004,
      "step": 15116
    },
    {
      "epoch": 7.0557759626604435,
      "grad_norm": 0.05810546875,
      "learning_rate": 3.980612859988577e-05,
      "loss": 0.0009,
      "step": 15117
    },
    {
      "epoch": 7.056242707117853,
      "grad_norm": 0.038818359375,
      "learning_rate": 3.979441185148245e-05,
      "loss": 0.0003,
      "step": 15118
    },
    {
      "epoch": 7.056709451575262,
      "grad_norm": 0.0157470703125,
      "learning_rate": 3.9782696399372264e-05,
      "loss": 0.0002,
      "step": 15119
    },
    {
      "epoch": 7.057176196032672,
      "grad_norm": 0.031982421875,
      "learning_rate": 3.977098224380741e-05,
      "loss": 0.0002,
      "step": 15120
    },
    {
      "epoch": 7.057642940490082,
      "grad_norm": 0.037109375,
      "learning_rate": 3.975926938504013e-05,
      "loss": 0.0018,
      "step": 15121
    },
    {
      "epoch": 7.058109684947492,
      "grad_norm": 0.02001953125,
      "learning_rate": 3.974755782332266e-05,
      "loss": 0.0002,
      "step": 15122
    },
    {
      "epoch": 7.0585764294049005,
      "grad_norm": 0.0283203125,
      "learning_rate": 3.973584755890708e-05,
      "loss": 0.0002,
      "step": 15123
    },
    {
      "epoch": 7.05904317386231,
      "grad_norm": 0.031494140625,
      "learning_rate": 3.972413859204557e-05,
      "loss": 0.0002,
      "step": 15124
    },
    {
      "epoch": 7.05950991831972,
      "grad_norm": 0.036376953125,
      "learning_rate": 3.9712430922990266e-05,
      "loss": 0.0019,
      "step": 15125
    },
    {
      "epoch": 7.05997666277713,
      "grad_norm": 0.03955078125,
      "learning_rate": 3.9700724551993174e-05,
      "loss": 0.0027,
      "step": 15126
    },
    {
      "epoch": 7.060443407234539,
      "grad_norm": 0.01251220703125,
      "learning_rate": 3.9689019479306425e-05,
      "loss": 0.0002,
      "step": 15127
    },
    {
      "epoch": 7.060910151691949,
      "grad_norm": 0.0166015625,
      "learning_rate": 3.9677315705181974e-05,
      "loss": 0.0002,
      "step": 15128
    },
    {
      "epoch": 7.061376896149358,
      "grad_norm": 0.03564453125,
      "learning_rate": 3.966561322987184e-05,
      "loss": 0.0002,
      "step": 15129
    },
    {
      "epoch": 7.061843640606768,
      "grad_norm": 0.061767578125,
      "learning_rate": 3.965391205362803e-05,
      "loss": 0.0003,
      "step": 15130
    },
    {
      "epoch": 7.062310385064177,
      "grad_norm": 0.03466796875,
      "learning_rate": 3.964221217670241e-05,
      "loss": 0.0018,
      "step": 15131
    },
    {
      "epoch": 7.062777129521587,
      "grad_norm": 0.01324462890625,
      "learning_rate": 3.963051359934693e-05,
      "loss": 0.0002,
      "step": 15132
    },
    {
      "epoch": 7.063243873978997,
      "grad_norm": 0.0279541015625,
      "learning_rate": 3.961881632181351e-05,
      "loss": 0.0002,
      "step": 15133
    },
    {
      "epoch": 7.0637106184364065,
      "grad_norm": 0.01519775390625,
      "learning_rate": 3.960712034435393e-05,
      "loss": 0.0002,
      "step": 15134
    },
    {
      "epoch": 7.064177362893815,
      "grad_norm": 0.009521484375,
      "learning_rate": 3.959542566722006e-05,
      "loss": 0.0002,
      "step": 15135
    },
    {
      "epoch": 7.064644107351225,
      "grad_norm": 0.06298828125,
      "learning_rate": 3.958373229066373e-05,
      "loss": 0.002,
      "step": 15136
    },
    {
      "epoch": 7.065110851808635,
      "grad_norm": 0.07958984375,
      "learning_rate": 3.957204021493669e-05,
      "loss": 0.0004,
      "step": 15137
    },
    {
      "epoch": 7.065577596266045,
      "grad_norm": 0.03515625,
      "learning_rate": 3.956034944029062e-05,
      "loss": 0.0003,
      "step": 15138
    },
    {
      "epoch": 7.066044340723454,
      "grad_norm": 0.037109375,
      "learning_rate": 3.9548659966977316e-05,
      "loss": 0.0003,
      "step": 15139
    },
    {
      "epoch": 7.0665110851808635,
      "grad_norm": 0.05322265625,
      "learning_rate": 3.9536971795248414e-05,
      "loss": 0.0003,
      "step": 15140
    },
    {
      "epoch": 7.066977829638273,
      "grad_norm": 0.08935546875,
      "learning_rate": 3.9525284925355586e-05,
      "loss": 0.002,
      "step": 15141
    },
    {
      "epoch": 7.067444574095683,
      "grad_norm": 0.01348876953125,
      "learning_rate": 3.9513599357550515e-05,
      "loss": 0.0002,
      "step": 15142
    },
    {
      "epoch": 7.067911318553092,
      "grad_norm": 0.038330078125,
      "learning_rate": 3.9501915092084716e-05,
      "loss": 0.0002,
      "step": 15143
    },
    {
      "epoch": 7.068378063010502,
      "grad_norm": 0.013671875,
      "learning_rate": 3.949023212920981e-05,
      "loss": 0.0002,
      "step": 15144
    },
    {
      "epoch": 7.0688448074679116,
      "grad_norm": 0.00958251953125,
      "learning_rate": 3.947855046917738e-05,
      "loss": 0.0002,
      "step": 15145
    },
    {
      "epoch": 7.0693115519253205,
      "grad_norm": 0.0262451171875,
      "learning_rate": 3.946687011223888e-05,
      "loss": 0.0002,
      "step": 15146
    },
    {
      "epoch": 7.06977829638273,
      "grad_norm": 0.06201171875,
      "learning_rate": 3.9455191058645813e-05,
      "loss": 0.0004,
      "step": 15147
    },
    {
      "epoch": 7.07024504084014,
      "grad_norm": 0.03466796875,
      "learning_rate": 3.9443513308649704e-05,
      "loss": 0.0015,
      "step": 15148
    },
    {
      "epoch": 7.07071178529755,
      "grad_norm": 0.07568359375,
      "learning_rate": 3.9431836862501906e-05,
      "loss": 0.0003,
      "step": 15149
    },
    {
      "epoch": 7.071178529754959,
      "grad_norm": 0.025390625,
      "learning_rate": 3.942016172045386e-05,
      "loss": 0.0002,
      "step": 15150
    },
    {
      "epoch": 7.0716452742123685,
      "grad_norm": 0.015869140625,
      "learning_rate": 3.940848788275699e-05,
      "loss": 0.0002,
      "step": 15151
    },
    {
      "epoch": 7.072112018669778,
      "grad_norm": 0.018798828125,
      "learning_rate": 3.939681534966255e-05,
      "loss": 0.0002,
      "step": 15152
    },
    {
      "epoch": 7.072578763127188,
      "grad_norm": 0.07958984375,
      "learning_rate": 3.938514412142193e-05,
      "loss": 0.0047,
      "step": 15153
    },
    {
      "epoch": 7.073045507584597,
      "grad_norm": 0.0167236328125,
      "learning_rate": 3.937347419828644e-05,
      "loss": 0.0002,
      "step": 15154
    },
    {
      "epoch": 7.073512252042007,
      "grad_norm": 0.0390625,
      "learning_rate": 3.9361805580507295e-05,
      "loss": 0.0003,
      "step": 15155
    },
    {
      "epoch": 7.073978996499417,
      "grad_norm": 0.022705078125,
      "learning_rate": 3.935013826833578e-05,
      "loss": 0.0002,
      "step": 15156
    },
    {
      "epoch": 7.074445740956826,
      "grad_norm": 0.0169677734375,
      "learning_rate": 3.933847226202304e-05,
      "loss": 0.0002,
      "step": 15157
    },
    {
      "epoch": 7.074912485414235,
      "grad_norm": 0.04150390625,
      "learning_rate": 3.93268075618203e-05,
      "loss": 0.0003,
      "step": 15158
    },
    {
      "epoch": 7.075379229871645,
      "grad_norm": 0.0206298828125,
      "learning_rate": 3.931514416797875e-05,
      "loss": 0.0002,
      "step": 15159
    },
    {
      "epoch": 7.075845974329055,
      "grad_norm": 0.0291748046875,
      "learning_rate": 3.9303482080749445e-05,
      "loss": 0.0004,
      "step": 15160
    },
    {
      "epoch": 7.076312718786465,
      "grad_norm": 0.0284423828125,
      "learning_rate": 3.9291821300383504e-05,
      "loss": 0.0003,
      "step": 15161
    },
    {
      "epoch": 7.076779463243874,
      "grad_norm": 0.01080322265625,
      "learning_rate": 3.9280161827132054e-05,
      "loss": 0.0001,
      "step": 15162
    },
    {
      "epoch": 7.077246207701283,
      "grad_norm": 0.054443359375,
      "learning_rate": 3.926850366124605e-05,
      "loss": 0.0009,
      "step": 15163
    },
    {
      "epoch": 7.077712952158693,
      "grad_norm": 0.048095703125,
      "learning_rate": 3.925684680297654e-05,
      "loss": 0.0034,
      "step": 15164
    },
    {
      "epoch": 7.078179696616103,
      "grad_norm": 0.09375,
      "learning_rate": 3.9245191252574556e-05,
      "loss": 0.002,
      "step": 15165
    },
    {
      "epoch": 7.078646441073512,
      "grad_norm": 0.039794921875,
      "learning_rate": 3.923353701029098e-05,
      "loss": 0.0002,
      "step": 15166
    },
    {
      "epoch": 7.079113185530922,
      "grad_norm": 0.020263671875,
      "learning_rate": 3.922188407637676e-05,
      "loss": 0.0002,
      "step": 15167
    },
    {
      "epoch": 7.0795799299883315,
      "grad_norm": 0.017822265625,
      "learning_rate": 3.921023245108285e-05,
      "loss": 0.0002,
      "step": 15168
    },
    {
      "epoch": 7.080046674445741,
      "grad_norm": 0.00823974609375,
      "learning_rate": 3.9198582134660056e-05,
      "loss": 0.0002,
      "step": 15169
    },
    {
      "epoch": 7.08051341890315,
      "grad_norm": 0.05029296875,
      "learning_rate": 3.918693312735925e-05,
      "loss": 0.0027,
      "step": 15170
    },
    {
      "epoch": 7.08098016336056,
      "grad_norm": 0.01904296875,
      "learning_rate": 3.9175285429431274e-05,
      "loss": 0.0002,
      "step": 15171
    },
    {
      "epoch": 7.08144690781797,
      "grad_norm": 0.006500244140625,
      "learning_rate": 3.916363904112686e-05,
      "loss": 0.0001,
      "step": 15172
    },
    {
      "epoch": 7.08191365227538,
      "grad_norm": 0.0595703125,
      "learning_rate": 3.915199396269679e-05,
      "loss": 0.0026,
      "step": 15173
    },
    {
      "epoch": 7.0823803967327885,
      "grad_norm": 0.01434326171875,
      "learning_rate": 3.914035019439184e-05,
      "loss": 0.0002,
      "step": 15174
    },
    {
      "epoch": 7.082847141190198,
      "grad_norm": 0.042724609375,
      "learning_rate": 3.912870773646263e-05,
      "loss": 0.0036,
      "step": 15175
    },
    {
      "epoch": 7.083313885647608,
      "grad_norm": 0.015380859375,
      "learning_rate": 3.911706658915993e-05,
      "loss": 0.0002,
      "step": 15176
    },
    {
      "epoch": 7.083780630105018,
      "grad_norm": 0.031982421875,
      "learning_rate": 3.91054267527343e-05,
      "loss": 0.0003,
      "step": 15177
    },
    {
      "epoch": 7.084247374562427,
      "grad_norm": 0.050537109375,
      "learning_rate": 3.90937882274364e-05,
      "loss": 0.0002,
      "step": 15178
    },
    {
      "epoch": 7.084714119019837,
      "grad_norm": 0.01397705078125,
      "learning_rate": 3.9082151013516846e-05,
      "loss": 0.0002,
      "step": 15179
    },
    {
      "epoch": 7.085180863477246,
      "grad_norm": 0.00927734375,
      "learning_rate": 3.9070515111226125e-05,
      "loss": 0.0002,
      "step": 15180
    },
    {
      "epoch": 7.085647607934656,
      "grad_norm": 0.00848388671875,
      "learning_rate": 3.9058880520814835e-05,
      "loss": 0.0002,
      "step": 15181
    },
    {
      "epoch": 7.086114352392065,
      "grad_norm": 0.04833984375,
      "learning_rate": 3.904724724253348e-05,
      "loss": 0.003,
      "step": 15182
    },
    {
      "epoch": 7.086581096849475,
      "grad_norm": 0.0230712890625,
      "learning_rate": 3.90356152766325e-05,
      "loss": 0.0002,
      "step": 15183
    },
    {
      "epoch": 7.087047841306885,
      "grad_norm": 0.0123291015625,
      "learning_rate": 3.9023984623362385e-05,
      "loss": 0.0002,
      "step": 15184
    },
    {
      "epoch": 7.0875145857642945,
      "grad_norm": 0.01214599609375,
      "learning_rate": 3.901235528297351e-05,
      "loss": 0.0002,
      "step": 15185
    },
    {
      "epoch": 7.087981330221703,
      "grad_norm": 0.0081787109375,
      "learning_rate": 3.900072725571632e-05,
      "loss": 0.0002,
      "step": 15186
    },
    {
      "epoch": 7.088448074679113,
      "grad_norm": 0.09130859375,
      "learning_rate": 3.898910054184113e-05,
      "loss": 0.0004,
      "step": 15187
    },
    {
      "epoch": 7.088914819136523,
      "grad_norm": 0.0234375,
      "learning_rate": 3.897747514159832e-05,
      "loss": 0.0002,
      "step": 15188
    },
    {
      "epoch": 7.089381563593932,
      "grad_norm": 0.10302734375,
      "learning_rate": 3.896585105523815e-05,
      "loss": 0.0004,
      "step": 15189
    },
    {
      "epoch": 7.089848308051342,
      "grad_norm": 0.068359375,
      "learning_rate": 3.8954228283010916e-05,
      "loss": 0.0026,
      "step": 15190
    },
    {
      "epoch": 7.0903150525087515,
      "grad_norm": 0.0115966796875,
      "learning_rate": 3.894260682516692e-05,
      "loss": 0.0002,
      "step": 15191
    },
    {
      "epoch": 7.090781796966161,
      "grad_norm": 0.0203857421875,
      "learning_rate": 3.893098668195631e-05,
      "loss": 0.0002,
      "step": 15192
    },
    {
      "epoch": 7.09124854142357,
      "grad_norm": 0.09619140625,
      "learning_rate": 3.891936785362931e-05,
      "loss": 0.0062,
      "step": 15193
    },
    {
      "epoch": 7.09171528588098,
      "grad_norm": 0.038330078125,
      "learning_rate": 3.8907750340436125e-05,
      "loss": 0.0016,
      "step": 15194
    },
    {
      "epoch": 7.09218203033839,
      "grad_norm": 0.0269775390625,
      "learning_rate": 3.8896134142626824e-05,
      "loss": 0.0002,
      "step": 15195
    },
    {
      "epoch": 7.0926487747958,
      "grad_norm": 0.006378173828125,
      "learning_rate": 3.888451926045155e-05,
      "loss": 0.0001,
      "step": 15196
    },
    {
      "epoch": 7.0931155192532085,
      "grad_norm": 0.017822265625,
      "learning_rate": 3.887290569416042e-05,
      "loss": 0.0002,
      "step": 15197
    },
    {
      "epoch": 7.093582263710618,
      "grad_norm": 0.0439453125,
      "learning_rate": 3.886129344400342e-05,
      "loss": 0.0026,
      "step": 15198
    },
    {
      "epoch": 7.094049008168028,
      "grad_norm": 0.021484375,
      "learning_rate": 3.884968251023061e-05,
      "loss": 0.0002,
      "step": 15199
    },
    {
      "epoch": 7.094515752625438,
      "grad_norm": 0.0087890625,
      "learning_rate": 3.883807289309202e-05,
      "loss": 0.0001,
      "step": 15200
    },
    {
      "epoch": 7.094982497082847,
      "grad_norm": 0.02734375,
      "learning_rate": 3.882646459283753e-05,
      "loss": 0.0002,
      "step": 15201
    },
    {
      "epoch": 7.095449241540257,
      "grad_norm": 0.076171875,
      "learning_rate": 3.8814857609717134e-05,
      "loss": 0.0005,
      "step": 15202
    },
    {
      "epoch": 7.095915985997666,
      "grad_norm": 0.04345703125,
      "learning_rate": 3.880325194398079e-05,
      "loss": 0.0003,
      "step": 15203
    },
    {
      "epoch": 7.096382730455076,
      "grad_norm": 0.0546875,
      "learning_rate": 3.879164759587828e-05,
      "loss": 0.0002,
      "step": 15204
    },
    {
      "epoch": 7.096849474912485,
      "grad_norm": 0.0263671875,
      "learning_rate": 3.878004456565951e-05,
      "loss": 0.0002,
      "step": 15205
    },
    {
      "epoch": 7.097316219369895,
      "grad_norm": 0.019775390625,
      "learning_rate": 3.876844285357435e-05,
      "loss": 0.0002,
      "step": 15206
    },
    {
      "epoch": 7.097782963827305,
      "grad_norm": 0.01251220703125,
      "learning_rate": 3.875684245987251e-05,
      "loss": 0.0002,
      "step": 15207
    },
    {
      "epoch": 7.0982497082847145,
      "grad_norm": 0.043212890625,
      "learning_rate": 3.874524338480384e-05,
      "loss": 0.0004,
      "step": 15208
    },
    {
      "epoch": 7.098716452742123,
      "grad_norm": 0.01251220703125,
      "learning_rate": 3.873364562861802e-05,
      "loss": 0.0002,
      "step": 15209
    },
    {
      "epoch": 7.099183197199533,
      "grad_norm": 0.018310546875,
      "learning_rate": 3.872204919156477e-05,
      "loss": 0.0002,
      "step": 15210
    },
    {
      "epoch": 7.099649941656943,
      "grad_norm": 0.00872802734375,
      "learning_rate": 3.871045407389383e-05,
      "loss": 0.0002,
      "step": 15211
    },
    {
      "epoch": 7.100116686114353,
      "grad_norm": 0.048095703125,
      "learning_rate": 3.869886027585478e-05,
      "loss": 0.0037,
      "step": 15212
    },
    {
      "epoch": 7.100583430571762,
      "grad_norm": 0.1640625,
      "learning_rate": 3.868726779769728e-05,
      "loss": 0.0007,
      "step": 15213
    },
    {
      "epoch": 7.1010501750291715,
      "grad_norm": 0.0186767578125,
      "learning_rate": 3.8675676639670975e-05,
      "loss": 0.0002,
      "step": 15214
    },
    {
      "epoch": 7.101516919486581,
      "grad_norm": 0.1357421875,
      "learning_rate": 3.866408680202535e-05,
      "loss": 0.0004,
      "step": 15215
    },
    {
      "epoch": 7.101983663943991,
      "grad_norm": 0.0181884765625,
      "learning_rate": 3.8652498285009996e-05,
      "loss": 0.0002,
      "step": 15216
    },
    {
      "epoch": 7.1024504084014,
      "grad_norm": 0.01556396484375,
      "learning_rate": 3.864091108887445e-05,
      "loss": 0.0002,
      "step": 15217
    },
    {
      "epoch": 7.10291715285881,
      "grad_norm": 0.0196533203125,
      "learning_rate": 3.8629325213868136e-05,
      "loss": 0.0002,
      "step": 15218
    },
    {
      "epoch": 7.1033838973162196,
      "grad_norm": 0.05810546875,
      "learning_rate": 3.861774066024054e-05,
      "loss": 0.0016,
      "step": 15219
    },
    {
      "epoch": 7.103850641773629,
      "grad_norm": 0.013427734375,
      "learning_rate": 3.860615742824113e-05,
      "loss": 0.0002,
      "step": 15220
    },
    {
      "epoch": 7.104317386231038,
      "grad_norm": 0.005950927734375,
      "learning_rate": 3.859457551811922e-05,
      "loss": 0.0001,
      "step": 15221
    },
    {
      "epoch": 7.104784130688448,
      "grad_norm": 0.01373291015625,
      "learning_rate": 3.858299493012425e-05,
      "loss": 0.0002,
      "step": 15222
    },
    {
      "epoch": 7.105250875145858,
      "grad_norm": 0.0301513671875,
      "learning_rate": 3.857141566450556e-05,
      "loss": 0.0016,
      "step": 15223
    },
    {
      "epoch": 7.105717619603268,
      "grad_norm": 0.039794921875,
      "learning_rate": 3.8559837721512427e-05,
      "loss": 0.0003,
      "step": 15224
    },
    {
      "epoch": 7.1061843640606766,
      "grad_norm": 0.08642578125,
      "learning_rate": 3.854826110139418e-05,
      "loss": 0.0003,
      "step": 15225
    },
    {
      "epoch": 7.106651108518086,
      "grad_norm": 0.283203125,
      "learning_rate": 3.853668580440003e-05,
      "loss": 0.0007,
      "step": 15226
    },
    {
      "epoch": 7.107117852975496,
      "grad_norm": 0.012451171875,
      "learning_rate": 3.852511183077922e-05,
      "loss": 0.0002,
      "step": 15227
    },
    {
      "epoch": 7.107584597432905,
      "grad_norm": 0.009521484375,
      "learning_rate": 3.8513539180781e-05,
      "loss": 0.0002,
      "step": 15228
    },
    {
      "epoch": 7.108051341890315,
      "grad_norm": 0.039306640625,
      "learning_rate": 3.850196785465446e-05,
      "loss": 0.0003,
      "step": 15229
    },
    {
      "epoch": 7.108518086347725,
      "grad_norm": 0.060546875,
      "learning_rate": 3.849039785264879e-05,
      "loss": 0.0003,
      "step": 15230
    },
    {
      "epoch": 7.108984830805134,
      "grad_norm": 0.052734375,
      "learning_rate": 3.847882917501313e-05,
      "loss": 0.0003,
      "step": 15231
    },
    {
      "epoch": 7.109451575262543,
      "grad_norm": 0.02392578125,
      "learning_rate": 3.846726182199655e-05,
      "loss": 0.0002,
      "step": 15232
    },
    {
      "epoch": 7.109918319719953,
      "grad_norm": 0.03955078125,
      "learning_rate": 3.8455695793848045e-05,
      "loss": 0.0023,
      "step": 15233
    },
    {
      "epoch": 7.110385064177363,
      "grad_norm": 0.0152587890625,
      "learning_rate": 3.84441310908167e-05,
      "loss": 0.0002,
      "step": 15234
    },
    {
      "epoch": 7.110851808634773,
      "grad_norm": 0.033203125,
      "learning_rate": 3.843256771315155e-05,
      "loss": 0.0003,
      "step": 15235
    },
    {
      "epoch": 7.111318553092182,
      "grad_norm": 0.025634765625,
      "learning_rate": 3.842100566110148e-05,
      "loss": 0.0002,
      "step": 15236
    },
    {
      "epoch": 7.111785297549591,
      "grad_norm": 0.039306640625,
      "learning_rate": 3.840944493491553e-05,
      "loss": 0.0027,
      "step": 15237
    },
    {
      "epoch": 7.112252042007001,
      "grad_norm": 0.0269775390625,
      "learning_rate": 3.839788553484253e-05,
      "loss": 0.0002,
      "step": 15238
    },
    {
      "epoch": 7.112718786464411,
      "grad_norm": 0.0164794921875,
      "learning_rate": 3.838632746113139e-05,
      "loss": 0.0002,
      "step": 15239
    },
    {
      "epoch": 7.11318553092182,
      "grad_norm": 0.01611328125,
      "learning_rate": 3.837477071403103e-05,
      "loss": 0.0002,
      "step": 15240
    },
    {
      "epoch": 7.11365227537923,
      "grad_norm": 0.10888671875,
      "learning_rate": 3.8363215293790186e-05,
      "loss": 0.0003,
      "step": 15241
    },
    {
      "epoch": 7.1141190198366395,
      "grad_norm": 0.00860595703125,
      "learning_rate": 3.835166120065771e-05,
      "loss": 0.0002,
      "step": 15242
    },
    {
      "epoch": 7.114585764294049,
      "grad_norm": 0.0106201171875,
      "learning_rate": 3.8340108434882407e-05,
      "loss": 0.0002,
      "step": 15243
    },
    {
      "epoch": 7.115052508751458,
      "grad_norm": 0.02392578125,
      "learning_rate": 3.8328556996712936e-05,
      "loss": 0.0002,
      "step": 15244
    },
    {
      "epoch": 7.115519253208868,
      "grad_norm": 0.0181884765625,
      "learning_rate": 3.831700688639807e-05,
      "loss": 0.0002,
      "step": 15245
    },
    {
      "epoch": 7.115985997666278,
      "grad_norm": 0.09326171875,
      "learning_rate": 3.830545810418651e-05,
      "loss": 0.0003,
      "step": 15246
    },
    {
      "epoch": 7.116452742123688,
      "grad_norm": 0.0302734375,
      "learning_rate": 3.8293910650326855e-05,
      "loss": 0.0003,
      "step": 15247
    },
    {
      "epoch": 7.1169194865810965,
      "grad_norm": 0.04931640625,
      "learning_rate": 3.828236452506777e-05,
      "loss": 0.0031,
      "step": 15248
    },
    {
      "epoch": 7.117386231038506,
      "grad_norm": 0.05078125,
      "learning_rate": 3.827081972865789e-05,
      "loss": 0.0029,
      "step": 15249
    },
    {
      "epoch": 7.117852975495916,
      "grad_norm": 0.011962890625,
      "learning_rate": 3.825927626134571e-05,
      "loss": 0.0002,
      "step": 15250
    },
    {
      "epoch": 7.118319719953326,
      "grad_norm": 0.0167236328125,
      "learning_rate": 3.824773412337982e-05,
      "loss": 0.0003,
      "step": 15251
    },
    {
      "epoch": 7.118786464410735,
      "grad_norm": 0.01116943359375,
      "learning_rate": 3.8236193315008774e-05,
      "loss": 0.0002,
      "step": 15252
    },
    {
      "epoch": 7.119253208868145,
      "grad_norm": 0.01104736328125,
      "learning_rate": 3.822465383648096e-05,
      "loss": 0.0002,
      "step": 15253
    },
    {
      "epoch": 7.119719953325554,
      "grad_norm": 0.02978515625,
      "learning_rate": 3.82131156880449e-05,
      "loss": 0.0002,
      "step": 15254
    },
    {
      "epoch": 7.120186697782964,
      "grad_norm": 0.0162353515625,
      "learning_rate": 3.820157886994905e-05,
      "loss": 0.0002,
      "step": 15255
    },
    {
      "epoch": 7.120653442240373,
      "grad_norm": 0.035888671875,
      "learning_rate": 3.8190043382441735e-05,
      "loss": 0.0007,
      "step": 15256
    },
    {
      "epoch": 7.121120186697783,
      "grad_norm": 0.01116943359375,
      "learning_rate": 3.81785092257714e-05,
      "loss": 0.0002,
      "step": 15257
    },
    {
      "epoch": 7.121586931155193,
      "grad_norm": 0.0185546875,
      "learning_rate": 3.8166976400186317e-05,
      "loss": 0.0002,
      "step": 15258
    },
    {
      "epoch": 7.1220536756126025,
      "grad_norm": 0.0576171875,
      "learning_rate": 3.815544490593484e-05,
      "loss": 0.003,
      "step": 15259
    },
    {
      "epoch": 7.122520420070011,
      "grad_norm": 0.01220703125,
      "learning_rate": 3.814391474326528e-05,
      "loss": 0.0002,
      "step": 15260
    },
    {
      "epoch": 7.122987164527421,
      "grad_norm": 0.0277099609375,
      "learning_rate": 3.8132385912425836e-05,
      "loss": 0.0002,
      "step": 15261
    },
    {
      "epoch": 7.123453908984831,
      "grad_norm": 0.01123046875,
      "learning_rate": 3.812085841366477e-05,
      "loss": 0.0001,
      "step": 15262
    },
    {
      "epoch": 7.123920653442241,
      "grad_norm": 0.087890625,
      "learning_rate": 3.810933224723031e-05,
      "loss": 0.0004,
      "step": 15263
    },
    {
      "epoch": 7.12438739789965,
      "grad_norm": 0.04296875,
      "learning_rate": 3.809780741337057e-05,
      "loss": 0.0029,
      "step": 15264
    },
    {
      "epoch": 7.1248541423570595,
      "grad_norm": 0.01904296875,
      "learning_rate": 3.808628391233371e-05,
      "loss": 0.0002,
      "step": 15265
    },
    {
      "epoch": 7.125320886814469,
      "grad_norm": 0.02392578125,
      "learning_rate": 3.807476174436788e-05,
      "loss": 0.0002,
      "step": 15266
    },
    {
      "epoch": 7.125787631271878,
      "grad_norm": 0.0498046875,
      "learning_rate": 3.806324090972111e-05,
      "loss": 0.0023,
      "step": 15267
    },
    {
      "epoch": 7.126254375729288,
      "grad_norm": 0.078125,
      "learning_rate": 3.805172140864148e-05,
      "loss": 0.0003,
      "step": 15268
    },
    {
      "epoch": 7.126721120186698,
      "grad_norm": 0.022216796875,
      "learning_rate": 3.804020324137706e-05,
      "loss": 0.0002,
      "step": 15269
    },
    {
      "epoch": 7.127187864644108,
      "grad_norm": 0.028564453125,
      "learning_rate": 3.802868640817577e-05,
      "loss": 0.0002,
      "step": 15270
    },
    {
      "epoch": 7.1276546091015165,
      "grad_norm": 0.018310546875,
      "learning_rate": 3.8017170909285624e-05,
      "loss": 0.0002,
      "step": 15271
    },
    {
      "epoch": 7.128121353558926,
      "grad_norm": 0.03271484375,
      "learning_rate": 3.80056567449546e-05,
      "loss": 0.0002,
      "step": 15272
    },
    {
      "epoch": 7.128588098016336,
      "grad_norm": 0.016357421875,
      "learning_rate": 3.7994143915430526e-05,
      "loss": 0.0003,
      "step": 15273
    },
    {
      "epoch": 7.129054842473746,
      "grad_norm": 0.0166015625,
      "learning_rate": 3.798263242096136e-05,
      "loss": 0.0002,
      "step": 15274
    },
    {
      "epoch": 7.129521586931155,
      "grad_norm": 0.08349609375,
      "learning_rate": 3.79711222617949e-05,
      "loss": 0.0003,
      "step": 15275
    },
    {
      "epoch": 7.129988331388565,
      "grad_norm": 0.009765625,
      "learning_rate": 3.7959613438179e-05,
      "loss": 0.0002,
      "step": 15276
    },
    {
      "epoch": 7.130455075845974,
      "grad_norm": 0.125,
      "learning_rate": 3.794810595036149e-05,
      "loss": 0.0015,
      "step": 15277
    },
    {
      "epoch": 7.130921820303384,
      "grad_norm": 0.0286865234375,
      "learning_rate": 3.793659979859008e-05,
      "loss": 0.0002,
      "step": 15278
    },
    {
      "epoch": 7.131388564760793,
      "grad_norm": 0.005035400390625,
      "learning_rate": 3.792509498311256e-05,
      "loss": 0.0001,
      "step": 15279
    },
    {
      "epoch": 7.131855309218203,
      "grad_norm": 0.0186767578125,
      "learning_rate": 3.791359150417659e-05,
      "loss": 0.0002,
      "step": 15280
    },
    {
      "epoch": 7.132322053675613,
      "grad_norm": 0.018798828125,
      "learning_rate": 3.790208936202991e-05,
      "loss": 0.0002,
      "step": 15281
    },
    {
      "epoch": 7.1327887981330225,
      "grad_norm": 0.08740234375,
      "learning_rate": 3.7890588556920124e-05,
      "loss": 0.0003,
      "step": 15282
    },
    {
      "epoch": 7.133255542590431,
      "grad_norm": 0.017578125,
      "learning_rate": 3.787908908909487e-05,
      "loss": 0.0001,
      "step": 15283
    },
    {
      "epoch": 7.133722287047841,
      "grad_norm": 0.0120849609375,
      "learning_rate": 3.786759095880178e-05,
      "loss": 0.0001,
      "step": 15284
    },
    {
      "epoch": 7.134189031505251,
      "grad_norm": 0.00927734375,
      "learning_rate": 3.785609416628837e-05,
      "loss": 0.0001,
      "step": 15285
    },
    {
      "epoch": 7.134655775962661,
      "grad_norm": 0.0206298828125,
      "learning_rate": 3.784459871180223e-05,
      "loss": 0.0002,
      "step": 15286
    },
    {
      "epoch": 7.13512252042007,
      "grad_norm": 0.0093994140625,
      "learning_rate": 3.783310459559081e-05,
      "loss": 0.0002,
      "step": 15287
    },
    {
      "epoch": 7.1355892648774795,
      "grad_norm": 0.0218505859375,
      "learning_rate": 3.782161181790163e-05,
      "loss": 0.0002,
      "step": 15288
    },
    {
      "epoch": 7.136056009334889,
      "grad_norm": 0.02001953125,
      "learning_rate": 3.781012037898216e-05,
      "loss": 0.0002,
      "step": 15289
    },
    {
      "epoch": 7.136522753792299,
      "grad_norm": 0.01068115234375,
      "learning_rate": 3.779863027907976e-05,
      "loss": 0.0002,
      "step": 15290
    },
    {
      "epoch": 7.136989498249708,
      "grad_norm": 0.01312255859375,
      "learning_rate": 3.778714151844188e-05,
      "loss": 0.0002,
      "step": 15291
    },
    {
      "epoch": 7.137456242707118,
      "grad_norm": 0.0537109375,
      "learning_rate": 3.777565409731589e-05,
      "loss": 0.0016,
      "step": 15292
    },
    {
      "epoch": 7.1379229871645276,
      "grad_norm": 0.00946044921875,
      "learning_rate": 3.7764168015949074e-05,
      "loss": 0.0002,
      "step": 15293
    },
    {
      "epoch": 7.138389731621937,
      "grad_norm": 0.018310546875,
      "learning_rate": 3.7752683274588785e-05,
      "loss": 0.0002,
      "step": 15294
    },
    {
      "epoch": 7.138856476079346,
      "grad_norm": 0.0157470703125,
      "learning_rate": 3.774119987348231e-05,
      "loss": 0.0002,
      "step": 15295
    },
    {
      "epoch": 7.139323220536756,
      "grad_norm": 0.07666015625,
      "learning_rate": 3.772971781287685e-05,
      "loss": 0.0003,
      "step": 15296
    },
    {
      "epoch": 7.139789964994166,
      "grad_norm": 0.0120849609375,
      "learning_rate": 3.7718237093019656e-05,
      "loss": 0.0002,
      "step": 15297
    },
    {
      "epoch": 7.140256709451576,
      "grad_norm": 0.01513671875,
      "learning_rate": 3.770675771415796e-05,
      "loss": 0.0002,
      "step": 15298
    },
    {
      "epoch": 7.1407234539089846,
      "grad_norm": 0.043701171875,
      "learning_rate": 3.769527967653884e-05,
      "loss": 0.0003,
      "step": 15299
    },
    {
      "epoch": 7.141190198366394,
      "grad_norm": 0.038818359375,
      "learning_rate": 3.768380298040949e-05,
      "loss": 0.0019,
      "step": 15300
    },
    {
      "epoch": 7.141656942823804,
      "grad_norm": 0.0242919921875,
      "learning_rate": 3.767232762601702e-05,
      "loss": 0.0016,
      "step": 15301
    },
    {
      "epoch": 7.142123687281214,
      "grad_norm": 0.05517578125,
      "learning_rate": 3.766085361360847e-05,
      "loss": 0.002,
      "step": 15302
    },
    {
      "epoch": 7.142590431738623,
      "grad_norm": 0.027587890625,
      "learning_rate": 3.76493809434309e-05,
      "loss": 0.0002,
      "step": 15303
    },
    {
      "epoch": 7.143057176196033,
      "grad_norm": 0.038818359375,
      "learning_rate": 3.763790961573138e-05,
      "loss": 0.0002,
      "step": 15304
    },
    {
      "epoch": 7.143523920653442,
      "grad_norm": 0.0703125,
      "learning_rate": 3.7626439630756816e-05,
      "loss": 0.0027,
      "step": 15305
    },
    {
      "epoch": 7.143990665110852,
      "grad_norm": 0.021728515625,
      "learning_rate": 3.761497098875425e-05,
      "loss": 0.0002,
      "step": 15306
    },
    {
      "epoch": 7.144457409568261,
      "grad_norm": 0.0164794921875,
      "learning_rate": 3.760350368997054e-05,
      "loss": 0.0002,
      "step": 15307
    },
    {
      "epoch": 7.144924154025671,
      "grad_norm": 0.045654296875,
      "learning_rate": 3.759203773465262e-05,
      "loss": 0.002,
      "step": 15308
    },
    {
      "epoch": 7.145390898483081,
      "grad_norm": 0.07958984375,
      "learning_rate": 3.7580573123047416e-05,
      "loss": 0.002,
      "step": 15309
    },
    {
      "epoch": 7.14585764294049,
      "grad_norm": 0.01318359375,
      "learning_rate": 3.7569109855401684e-05,
      "loss": 0.0002,
      "step": 15310
    },
    {
      "epoch": 7.146324387397899,
      "grad_norm": 0.01513671875,
      "learning_rate": 3.7557647931962294e-05,
      "loss": 0.0002,
      "step": 15311
    },
    {
      "epoch": 7.146791131855309,
      "grad_norm": 0.03515625,
      "learning_rate": 3.754618735297606e-05,
      "loss": 0.0002,
      "step": 15312
    },
    {
      "epoch": 7.147257876312719,
      "grad_norm": 0.0068359375,
      "learning_rate": 3.753472811868965e-05,
      "loss": 0.0001,
      "step": 15313
    },
    {
      "epoch": 7.147724620770128,
      "grad_norm": 0.01104736328125,
      "learning_rate": 3.752327022934987e-05,
      "loss": 0.0002,
      "step": 15314
    },
    {
      "epoch": 7.148191365227538,
      "grad_norm": 0.062255859375,
      "learning_rate": 3.751181368520343e-05,
      "loss": 0.0036,
      "step": 15315
    },
    {
      "epoch": 7.1486581096849475,
      "grad_norm": 0.0284423828125,
      "learning_rate": 3.750035848649695e-05,
      "loss": 0.0002,
      "step": 15316
    },
    {
      "epoch": 7.149124854142357,
      "grad_norm": 0.021728515625,
      "learning_rate": 3.748890463347709e-05,
      "loss": 0.0002,
      "step": 15317
    },
    {
      "epoch": 7.149591598599766,
      "grad_norm": 0.0306396484375,
      "learning_rate": 3.74774521263905e-05,
      "loss": 0.0025,
      "step": 15318
    },
    {
      "epoch": 7.150058343057176,
      "grad_norm": 0.0086669921875,
      "learning_rate": 3.7466000965483705e-05,
      "loss": 0.0001,
      "step": 15319
    },
    {
      "epoch": 7.150525087514586,
      "grad_norm": 0.01708984375,
      "learning_rate": 3.745455115100329e-05,
      "loss": 0.0002,
      "step": 15320
    },
    {
      "epoch": 7.150991831971996,
      "grad_norm": 0.0301513671875,
      "learning_rate": 3.7443102683195816e-05,
      "loss": 0.0002,
      "step": 15321
    },
    {
      "epoch": 7.1514585764294045,
      "grad_norm": 0.017333984375,
      "learning_rate": 3.743165556230772e-05,
      "loss": 0.0002,
      "step": 15322
    },
    {
      "epoch": 7.151925320886814,
      "grad_norm": 0.0166015625,
      "learning_rate": 3.742020978858553e-05,
      "loss": 0.0002,
      "step": 15323
    },
    {
      "epoch": 7.152392065344224,
      "grad_norm": 0.00921630859375,
      "learning_rate": 3.740876536227562e-05,
      "loss": 0.0001,
      "step": 15324
    },
    {
      "epoch": 7.152858809801634,
      "grad_norm": 0.01141357421875,
      "learning_rate": 3.7397322283624444e-05,
      "loss": 0.0001,
      "step": 15325
    },
    {
      "epoch": 7.153325554259043,
      "grad_norm": 0.0152587890625,
      "learning_rate": 3.738588055287842e-05,
      "loss": 0.0002,
      "step": 15326
    },
    {
      "epoch": 7.153792298716453,
      "grad_norm": 0.01141357421875,
      "learning_rate": 3.737444017028384e-05,
      "loss": 0.0002,
      "step": 15327
    },
    {
      "epoch": 7.154259043173862,
      "grad_norm": 0.00787353515625,
      "learning_rate": 3.736300113608703e-05,
      "loss": 0.0002,
      "step": 15328
    },
    {
      "epoch": 7.154725787631272,
      "grad_norm": 0.017822265625,
      "learning_rate": 3.73515634505343e-05,
      "loss": 0.0002,
      "step": 15329
    },
    {
      "epoch": 7.155192532088681,
      "grad_norm": 0.024169921875,
      "learning_rate": 3.734012711387195e-05,
      "loss": 0.0003,
      "step": 15330
    },
    {
      "epoch": 7.155659276546091,
      "grad_norm": 0.0142822265625,
      "learning_rate": 3.7328692126346144e-05,
      "loss": 0.0002,
      "step": 15331
    },
    {
      "epoch": 7.156126021003501,
      "grad_norm": 0.00970458984375,
      "learning_rate": 3.7317258488203134e-05,
      "loss": 0.0001,
      "step": 15332
    },
    {
      "epoch": 7.1565927654609105,
      "grad_norm": 0.031005859375,
      "learning_rate": 3.7305826199689144e-05,
      "loss": 0.0002,
      "step": 15333
    },
    {
      "epoch": 7.157059509918319,
      "grad_norm": 0.0086669921875,
      "learning_rate": 3.729439526105023e-05,
      "loss": 0.0002,
      "step": 15334
    },
    {
      "epoch": 7.157526254375729,
      "grad_norm": 0.0089111328125,
      "learning_rate": 3.7282965672532586e-05,
      "loss": 0.0002,
      "step": 15335
    },
    {
      "epoch": 7.157992998833139,
      "grad_norm": 0.06201171875,
      "learning_rate": 3.7271537434382265e-05,
      "loss": 0.0027,
      "step": 15336
    },
    {
      "epoch": 7.158459743290549,
      "grad_norm": 0.039794921875,
      "learning_rate": 3.726011054684533e-05,
      "loss": 0.0026,
      "step": 15337
    },
    {
      "epoch": 7.158926487747958,
      "grad_norm": 0.00811767578125,
      "learning_rate": 3.724868501016785e-05,
      "loss": 0.0002,
      "step": 15338
    },
    {
      "epoch": 7.1593932322053675,
      "grad_norm": 0.0089111328125,
      "learning_rate": 3.723726082459579e-05,
      "loss": 0.0002,
      "step": 15339
    },
    {
      "epoch": 7.159859976662777,
      "grad_norm": 0.06298828125,
      "learning_rate": 3.722583799037512e-05,
      "loss": 0.0003,
      "step": 15340
    },
    {
      "epoch": 7.160326721120187,
      "grad_norm": 0.0234375,
      "learning_rate": 3.721441650775186e-05,
      "loss": 0.0002,
      "step": 15341
    },
    {
      "epoch": 7.160793465577596,
      "grad_norm": 0.037841796875,
      "learning_rate": 3.7202996376971835e-05,
      "loss": 0.0002,
      "step": 15342
    },
    {
      "epoch": 7.161260210035006,
      "grad_norm": 0.029052734375,
      "learning_rate": 3.719157759828097e-05,
      "loss": 0.0002,
      "step": 15343
    },
    {
      "epoch": 7.161726954492416,
      "grad_norm": 0.0146484375,
      "learning_rate": 3.718016017192516e-05,
      "loss": 0.0002,
      "step": 15344
    },
    {
      "epoch": 7.162193698949825,
      "grad_norm": 0.035400390625,
      "learning_rate": 3.716874409815017e-05,
      "loss": 0.0003,
      "step": 15345
    },
    {
      "epoch": 7.162660443407234,
      "grad_norm": 0.041015625,
      "learning_rate": 3.715732937720182e-05,
      "loss": 0.0022,
      "step": 15346
    },
    {
      "epoch": 7.163127187864644,
      "grad_norm": 0.0517578125,
      "learning_rate": 3.7145916009325945e-05,
      "loss": 0.0022,
      "step": 15347
    },
    {
      "epoch": 7.163593932322054,
      "grad_norm": 0.059814453125,
      "learning_rate": 3.713450399476819e-05,
      "loss": 0.0048,
      "step": 15348
    },
    {
      "epoch": 7.164060676779464,
      "grad_norm": 0.0078125,
      "learning_rate": 3.712309333377433e-05,
      "loss": 0.0002,
      "step": 15349
    },
    {
      "epoch": 7.164527421236873,
      "grad_norm": 0.00592041015625,
      "learning_rate": 3.7111684026590054e-05,
      "loss": 0.0001,
      "step": 15350
    },
    {
      "epoch": 7.164994165694282,
      "grad_norm": 0.0458984375,
      "learning_rate": 3.710027607346096e-05,
      "loss": 0.0003,
      "step": 15351
    },
    {
      "epoch": 7.165460910151692,
      "grad_norm": 0.015869140625,
      "learning_rate": 3.7088869474632715e-05,
      "loss": 0.0002,
      "step": 15352
    },
    {
      "epoch": 7.165927654609101,
      "grad_norm": 0.0208740234375,
      "learning_rate": 3.7077464230350956e-05,
      "loss": 0.0002,
      "step": 15353
    },
    {
      "epoch": 7.166394399066511,
      "grad_norm": 0.023681640625,
      "learning_rate": 3.7066060340861153e-05,
      "loss": 0.0002,
      "step": 15354
    },
    {
      "epoch": 7.166861143523921,
      "grad_norm": 0.041748046875,
      "learning_rate": 3.705465780640894e-05,
      "loss": 0.0002,
      "step": 15355
    },
    {
      "epoch": 7.1673278879813305,
      "grad_norm": 0.0380859375,
      "learning_rate": 3.7043256627239754e-05,
      "loss": 0.0029,
      "step": 15356
    },
    {
      "epoch": 7.167794632438739,
      "grad_norm": 0.0172119140625,
      "learning_rate": 3.703185680359911e-05,
      "loss": 0.0002,
      "step": 15357
    },
    {
      "epoch": 7.168261376896149,
      "grad_norm": 0.04248046875,
      "learning_rate": 3.702045833573248e-05,
      "loss": 0.0002,
      "step": 15358
    },
    {
      "epoch": 7.168728121353559,
      "grad_norm": 0.01300048828125,
      "learning_rate": 3.700906122388521e-05,
      "loss": 0.0002,
      "step": 15359
    },
    {
      "epoch": 7.169194865810969,
      "grad_norm": 0.00909423828125,
      "learning_rate": 3.6997665468302755e-05,
      "loss": 0.0001,
      "step": 15360
    },
    {
      "epoch": 7.169661610268378,
      "grad_norm": 0.02099609375,
      "learning_rate": 3.6986271069230506e-05,
      "loss": 0.0002,
      "step": 15361
    },
    {
      "epoch": 7.1701283547257875,
      "grad_norm": 0.025146484375,
      "learning_rate": 3.697487802691372e-05,
      "loss": 0.0002,
      "step": 15362
    },
    {
      "epoch": 7.170595099183197,
      "grad_norm": 0.031494140625,
      "learning_rate": 3.6963486341597716e-05,
      "loss": 0.0002,
      "step": 15363
    },
    {
      "epoch": 7.171061843640607,
      "grad_norm": 0.038818359375,
      "learning_rate": 3.6952096013527845e-05,
      "loss": 0.0028,
      "step": 15364
    },
    {
      "epoch": 7.171528588098016,
      "grad_norm": 0.023681640625,
      "learning_rate": 3.694070704294926e-05,
      "loss": 0.0002,
      "step": 15365
    },
    {
      "epoch": 7.171995332555426,
      "grad_norm": 0.029296875,
      "learning_rate": 3.692931943010722e-05,
      "loss": 0.0002,
      "step": 15366
    },
    {
      "epoch": 7.172462077012836,
      "grad_norm": 0.038818359375,
      "learning_rate": 3.6917933175246954e-05,
      "loss": 0.0003,
      "step": 15367
    },
    {
      "epoch": 7.172928821470245,
      "grad_norm": 0.0947265625,
      "learning_rate": 3.6906548278613525e-05,
      "loss": 0.0004,
      "step": 15368
    },
    {
      "epoch": 7.173395565927654,
      "grad_norm": 0.024658203125,
      "learning_rate": 3.689516474045214e-05,
      "loss": 0.0002,
      "step": 15369
    },
    {
      "epoch": 7.173862310385064,
      "grad_norm": 0.0169677734375,
      "learning_rate": 3.688378256100788e-05,
      "loss": 0.0002,
      "step": 15370
    },
    {
      "epoch": 7.174329054842474,
      "grad_norm": 0.06982421875,
      "learning_rate": 3.68724017405258e-05,
      "loss": 0.0002,
      "step": 15371
    },
    {
      "epoch": 7.174795799299884,
      "grad_norm": 0.018798828125,
      "learning_rate": 3.686102227925093e-05,
      "loss": 0.0002,
      "step": 15372
    },
    {
      "epoch": 7.1752625437572926,
      "grad_norm": 0.07470703125,
      "learning_rate": 3.684964417742836e-05,
      "loss": 0.0003,
      "step": 15373
    },
    {
      "epoch": 7.175729288214702,
      "grad_norm": 0.052490234375,
      "learning_rate": 3.683826743530296e-05,
      "loss": 0.0019,
      "step": 15374
    },
    {
      "epoch": 7.176196032672112,
      "grad_norm": 0.03173828125,
      "learning_rate": 3.682689205311979e-05,
      "loss": 0.0021,
      "step": 15375
    },
    {
      "epoch": 7.176662777129522,
      "grad_norm": 0.0458984375,
      "learning_rate": 3.681551803112373e-05,
      "loss": 0.0026,
      "step": 15376
    },
    {
      "epoch": 7.177129521586931,
      "grad_norm": 0.02490234375,
      "learning_rate": 3.6804145369559626e-05,
      "loss": 0.0002,
      "step": 15377
    },
    {
      "epoch": 7.177596266044341,
      "grad_norm": 0.01611328125,
      "learning_rate": 3.67927740686724e-05,
      "loss": 0.0002,
      "step": 15378
    },
    {
      "epoch": 7.17806301050175,
      "grad_norm": 0.0191650390625,
      "learning_rate": 3.67814041287069e-05,
      "loss": 0.0002,
      "step": 15379
    },
    {
      "epoch": 7.17852975495916,
      "grad_norm": 0.00848388671875,
      "learning_rate": 3.6770035549907875e-05,
      "loss": 0.0001,
      "step": 15380
    },
    {
      "epoch": 7.178996499416569,
      "grad_norm": 0.01416015625,
      "learning_rate": 3.6758668332520154e-05,
      "loss": 0.0002,
      "step": 15381
    },
    {
      "epoch": 7.179463243873979,
      "grad_norm": 0.0191650390625,
      "learning_rate": 3.67473024767885e-05,
      "loss": 0.0002,
      "step": 15382
    },
    {
      "epoch": 7.179929988331389,
      "grad_norm": 0.03125,
      "learning_rate": 3.673593798295757e-05,
      "loss": 0.0002,
      "step": 15383
    },
    {
      "epoch": 7.1803967327887985,
      "grad_norm": 0.018310546875,
      "learning_rate": 3.672457485127209e-05,
      "loss": 0.0002,
      "step": 15384
    },
    {
      "epoch": 7.180863477246207,
      "grad_norm": 0.11962890625,
      "learning_rate": 3.671321308197674e-05,
      "loss": 0.0006,
      "step": 15385
    },
    {
      "epoch": 7.181330221703617,
      "grad_norm": 0.054931640625,
      "learning_rate": 3.67018526753161e-05,
      "loss": 0.0034,
      "step": 15386
    },
    {
      "epoch": 7.181796966161027,
      "grad_norm": 0.058349609375,
      "learning_rate": 3.669049363153485e-05,
      "loss": 0.0033,
      "step": 15387
    },
    {
      "epoch": 7.182263710618437,
      "grad_norm": 0.259765625,
      "learning_rate": 3.6679135950877474e-05,
      "loss": 0.001,
      "step": 15388
    },
    {
      "epoch": 7.182730455075846,
      "grad_norm": 0.016357421875,
      "learning_rate": 3.666777963358855e-05,
      "loss": 0.0002,
      "step": 15389
    },
    {
      "epoch": 7.1831971995332555,
      "grad_norm": 0.0211181640625,
      "learning_rate": 3.665642467991264e-05,
      "loss": 0.0002,
      "step": 15390
    },
    {
      "epoch": 7.183663943990665,
      "grad_norm": 0.007537841796875,
      "learning_rate": 3.664507109009415e-05,
      "loss": 0.0002,
      "step": 15391
    },
    {
      "epoch": 7.184130688448075,
      "grad_norm": 0.0380859375,
      "learning_rate": 3.663371886437758e-05,
      "loss": 0.0017,
      "step": 15392
    },
    {
      "epoch": 7.184597432905484,
      "grad_norm": 0.0233154296875,
      "learning_rate": 3.6622368003007376e-05,
      "loss": 0.0002,
      "step": 15393
    },
    {
      "epoch": 7.185064177362894,
      "grad_norm": 0.0196533203125,
      "learning_rate": 3.661101850622788e-05,
      "loss": 0.0001,
      "step": 15394
    },
    {
      "epoch": 7.185530921820304,
      "grad_norm": 0.0123291015625,
      "learning_rate": 3.659967037428348e-05,
      "loss": 0.0002,
      "step": 15395
    },
    {
      "epoch": 7.1859976662777125,
      "grad_norm": 0.140625,
      "learning_rate": 3.658832360741855e-05,
      "loss": 0.0004,
      "step": 15396
    },
    {
      "epoch": 7.186464410735122,
      "grad_norm": 0.034423828125,
      "learning_rate": 3.657697820587735e-05,
      "loss": 0.0002,
      "step": 15397
    },
    {
      "epoch": 7.186931155192532,
      "grad_norm": 0.02734375,
      "learning_rate": 3.6565634169904164e-05,
      "loss": 0.0003,
      "step": 15398
    },
    {
      "epoch": 7.187397899649942,
      "grad_norm": 0.0157470703125,
      "learning_rate": 3.65542914997433e-05,
      "loss": 0.0002,
      "step": 15399
    },
    {
      "epoch": 7.187864644107351,
      "grad_norm": 0.034912109375,
      "learning_rate": 3.654295019563889e-05,
      "loss": 0.0003,
      "step": 15400
    },
    {
      "epoch": 7.188331388564761,
      "grad_norm": 0.017333984375,
      "learning_rate": 3.653161025783518e-05,
      "loss": 0.0002,
      "step": 15401
    },
    {
      "epoch": 7.18879813302217,
      "grad_norm": 0.050537109375,
      "learning_rate": 3.6520271686576356e-05,
      "loss": 0.0031,
      "step": 15402
    },
    {
      "epoch": 7.18926487747958,
      "grad_norm": 0.05322265625,
      "learning_rate": 3.650893448210647e-05,
      "loss": 0.0049,
      "step": 15403
    },
    {
      "epoch": 7.189731621936989,
      "grad_norm": 0.006439208984375,
      "learning_rate": 3.6497598644669704e-05,
      "loss": 0.0002,
      "step": 15404
    },
    {
      "epoch": 7.190198366394399,
      "grad_norm": 0.006866455078125,
      "learning_rate": 3.6486264174510076e-05,
      "loss": 0.0002,
      "step": 15405
    },
    {
      "epoch": 7.190665110851809,
      "grad_norm": 0.0177001953125,
      "learning_rate": 3.647493107187163e-05,
      "loss": 0.0002,
      "step": 15406
    },
    {
      "epoch": 7.1911318553092185,
      "grad_norm": 0.01385498046875,
      "learning_rate": 3.646359933699845e-05,
      "loss": 0.0002,
      "step": 15407
    },
    {
      "epoch": 7.191598599766627,
      "grad_norm": 0.0196533203125,
      "learning_rate": 3.645226897013443e-05,
      "loss": 0.0002,
      "step": 15408
    },
    {
      "epoch": 7.192065344224037,
      "grad_norm": 0.0279541015625,
      "learning_rate": 3.644093997152357e-05,
      "loss": 0.0003,
      "step": 15409
    },
    {
      "epoch": 7.192532088681447,
      "grad_norm": 0.0303955078125,
      "learning_rate": 3.642961234140982e-05,
      "loss": 0.0002,
      "step": 15410
    },
    {
      "epoch": 7.192998833138857,
      "grad_norm": 0.0177001953125,
      "learning_rate": 3.641828608003702e-05,
      "loss": 0.0002,
      "step": 15411
    },
    {
      "epoch": 7.193465577596266,
      "grad_norm": 0.0308837890625,
      "learning_rate": 3.640696118764907e-05,
      "loss": 0.0002,
      "step": 15412
    },
    {
      "epoch": 7.1939323220536755,
      "grad_norm": 0.012939453125,
      "learning_rate": 3.6395637664489834e-05,
      "loss": 0.0002,
      "step": 15413
    },
    {
      "epoch": 7.194399066511085,
      "grad_norm": 0.005523681640625,
      "learning_rate": 3.638431551080306e-05,
      "loss": 0.0001,
      "step": 15414
    },
    {
      "epoch": 7.194865810968495,
      "grad_norm": 0.01708984375,
      "learning_rate": 3.6372994726832544e-05,
      "loss": 0.0002,
      "step": 15415
    },
    {
      "epoch": 7.195332555425904,
      "grad_norm": 0.06591796875,
      "learning_rate": 3.636167531282209e-05,
      "loss": 0.0004,
      "step": 15416
    },
    {
      "epoch": 7.195799299883314,
      "grad_norm": 0.01226806640625,
      "learning_rate": 3.635035726901533e-05,
      "loss": 0.0002,
      "step": 15417
    },
    {
      "epoch": 7.196266044340724,
      "grad_norm": 0.0079345703125,
      "learning_rate": 3.6339040595656006e-05,
      "loss": 0.0002,
      "step": 15418
    },
    {
      "epoch": 7.196732788798133,
      "grad_norm": 0.0240478515625,
      "learning_rate": 3.6327725292987805e-05,
      "loss": 0.0002,
      "step": 15419
    },
    {
      "epoch": 7.197199533255542,
      "grad_norm": 0.04150390625,
      "learning_rate": 3.631641136125429e-05,
      "loss": 0.0024,
      "step": 15420
    },
    {
      "epoch": 7.197666277712952,
      "grad_norm": 0.009033203125,
      "learning_rate": 3.63050988006991e-05,
      "loss": 0.0002,
      "step": 15421
    },
    {
      "epoch": 7.198133022170362,
      "grad_norm": 0.05810546875,
      "learning_rate": 3.629378761156583e-05,
      "loss": 0.0028,
      "step": 15422
    },
    {
      "epoch": 7.198599766627772,
      "grad_norm": 0.05029296875,
      "learning_rate": 3.6282477794098003e-05,
      "loss": 0.0023,
      "step": 15423
    },
    {
      "epoch": 7.199066511085181,
      "grad_norm": 0.01080322265625,
      "learning_rate": 3.627116934853909e-05,
      "loss": 0.0002,
      "step": 15424
    },
    {
      "epoch": 7.19953325554259,
      "grad_norm": 0.0380859375,
      "learning_rate": 3.6259862275132636e-05,
      "loss": 0.0018,
      "step": 15425
    },
    {
      "epoch": 7.2,
      "grad_norm": 0.1669921875,
      "learning_rate": 3.624855657412204e-05,
      "loss": 0.0005,
      "step": 15426
    },
    {
      "epoch": 7.20046674445741,
      "grad_norm": 0.031005859375,
      "learning_rate": 3.6237252245750755e-05,
      "loss": 0.0002,
      "step": 15427
    },
    {
      "epoch": 7.200933488914819,
      "grad_norm": 0.029052734375,
      "learning_rate": 3.622594929026221e-05,
      "loss": 0.0002,
      "step": 15428
    },
    {
      "epoch": 7.201400233372229,
      "grad_norm": 0.017578125,
      "learning_rate": 3.6214647707899694e-05,
      "loss": 0.0002,
      "step": 15429
    },
    {
      "epoch": 7.2018669778296385,
      "grad_norm": 0.01312255859375,
      "learning_rate": 3.6203347498906595e-05,
      "loss": 0.0002,
      "step": 15430
    },
    {
      "epoch": 7.202333722287048,
      "grad_norm": 0.01263427734375,
      "learning_rate": 3.6192048663526234e-05,
      "loss": 0.0002,
      "step": 15431
    },
    {
      "epoch": 7.202800466744457,
      "grad_norm": 0.0303955078125,
      "learning_rate": 3.618075120200183e-05,
      "loss": 0.0002,
      "step": 15432
    },
    {
      "epoch": 7.203267211201867,
      "grad_norm": 0.00830078125,
      "learning_rate": 3.616945511457665e-05,
      "loss": 0.0002,
      "step": 15433
    },
    {
      "epoch": 7.203733955659277,
      "grad_norm": 0.0289306640625,
      "learning_rate": 3.615816040149398e-05,
      "loss": 0.0003,
      "step": 15434
    },
    {
      "epoch": 7.204200700116687,
      "grad_norm": 0.0267333984375,
      "learning_rate": 3.6146867062996904e-05,
      "loss": 0.0003,
      "step": 15435
    },
    {
      "epoch": 7.2046674445740955,
      "grad_norm": 0.01165771484375,
      "learning_rate": 3.613557509932866e-05,
      "loss": 0.0002,
      "step": 15436
    },
    {
      "epoch": 7.205134189031505,
      "grad_norm": 0.10791015625,
      "learning_rate": 3.612428451073231e-05,
      "loss": 0.0004,
      "step": 15437
    },
    {
      "epoch": 7.205600933488915,
      "grad_norm": 0.009521484375,
      "learning_rate": 3.611299529745098e-05,
      "loss": 0.0001,
      "step": 15438
    },
    {
      "epoch": 7.206067677946324,
      "grad_norm": 0.010498046875,
      "learning_rate": 3.610170745972778e-05,
      "loss": 0.0002,
      "step": 15439
    },
    {
      "epoch": 7.206534422403734,
      "grad_norm": 0.021484375,
      "learning_rate": 3.609042099780568e-05,
      "loss": 0.0002,
      "step": 15440
    },
    {
      "epoch": 7.207001166861144,
      "grad_norm": 0.0086669921875,
      "learning_rate": 3.607913591192774e-05,
      "loss": 0.0001,
      "step": 15441
    },
    {
      "epoch": 7.207467911318553,
      "grad_norm": 0.005859375,
      "learning_rate": 3.606785220233695e-05,
      "loss": 0.0001,
      "step": 15442
    },
    {
      "epoch": 7.207934655775962,
      "grad_norm": 0.1787109375,
      "learning_rate": 3.60565698692762e-05,
      "loss": 0.0024,
      "step": 15443
    },
    {
      "epoch": 7.208401400233372,
      "grad_norm": 0.031494140625,
      "learning_rate": 3.604528891298844e-05,
      "loss": 0.0017,
      "step": 15444
    },
    {
      "epoch": 7.208868144690782,
      "grad_norm": 0.072265625,
      "learning_rate": 3.6034009333716614e-05,
      "loss": 0.0003,
      "step": 15445
    },
    {
      "epoch": 7.209334889148192,
      "grad_norm": 0.0106201171875,
      "learning_rate": 3.60227311317035e-05,
      "loss": 0.0002,
      "step": 15446
    },
    {
      "epoch": 7.209801633605601,
      "grad_norm": 0.0155029296875,
      "learning_rate": 3.6011454307191975e-05,
      "loss": 0.0001,
      "step": 15447
    },
    {
      "epoch": 7.21026837806301,
      "grad_norm": 0.08544921875,
      "learning_rate": 3.600017886042486e-05,
      "loss": 0.0023,
      "step": 15448
    },
    {
      "epoch": 7.21073512252042,
      "grad_norm": 0.0625,
      "learning_rate": 3.5988904791644875e-05,
      "loss": 0.0004,
      "step": 15449
    },
    {
      "epoch": 7.21120186697783,
      "grad_norm": 0.007476806640625,
      "learning_rate": 3.597763210109479e-05,
      "loss": 0.0001,
      "step": 15450
    },
    {
      "epoch": 7.211668611435239,
      "grad_norm": 0.2294921875,
      "learning_rate": 3.596636078901735e-05,
      "loss": 0.0034,
      "step": 15451
    },
    {
      "epoch": 7.212135355892649,
      "grad_norm": 0.013916015625,
      "learning_rate": 3.5955090855655184e-05,
      "loss": 0.0002,
      "step": 15452
    },
    {
      "epoch": 7.212602100350058,
      "grad_norm": 0.015380859375,
      "learning_rate": 3.5943822301250994e-05,
      "loss": 0.0002,
      "step": 15453
    },
    {
      "epoch": 7.213068844807468,
      "grad_norm": 0.01458740234375,
      "learning_rate": 3.593255512604735e-05,
      "loss": 0.0002,
      "step": 15454
    },
    {
      "epoch": 7.213535589264877,
      "grad_norm": 0.0673828125,
      "learning_rate": 3.592128933028689e-05,
      "loss": 0.0003,
      "step": 15455
    },
    {
      "epoch": 7.214002333722287,
      "grad_norm": 0.030029296875,
      "learning_rate": 3.591002491421219e-05,
      "loss": 0.0002,
      "step": 15456
    },
    {
      "epoch": 7.214469078179697,
      "grad_norm": 0.048828125,
      "learning_rate": 3.5898761878065737e-05,
      "loss": 0.0003,
      "step": 15457
    },
    {
      "epoch": 7.2149358226371065,
      "grad_norm": 0.0274658203125,
      "learning_rate": 3.5887500222090045e-05,
      "loss": 0.0015,
      "step": 15458
    },
    {
      "epoch": 7.215402567094515,
      "grad_norm": 0.0189208984375,
      "learning_rate": 3.5876239946527654e-05,
      "loss": 0.0002,
      "step": 15459
    },
    {
      "epoch": 7.215869311551925,
      "grad_norm": 0.04345703125,
      "learning_rate": 3.586498105162093e-05,
      "loss": 0.0029,
      "step": 15460
    },
    {
      "epoch": 7.216336056009335,
      "grad_norm": 0.07421875,
      "learning_rate": 3.5853723537612315e-05,
      "loss": 0.0004,
      "step": 15461
    },
    {
      "epoch": 7.216802800466745,
      "grad_norm": 0.060546875,
      "learning_rate": 3.584246740474425e-05,
      "loss": 0.0018,
      "step": 15462
    },
    {
      "epoch": 7.217269544924154,
      "grad_norm": 0.0279541015625,
      "learning_rate": 3.5831212653259005e-05,
      "loss": 0.0002,
      "step": 15463
    },
    {
      "epoch": 7.2177362893815635,
      "grad_norm": 0.03466796875,
      "learning_rate": 3.581995928339895e-05,
      "loss": 0.0031,
      "step": 15464
    },
    {
      "epoch": 7.218203033838973,
      "grad_norm": 0.0167236328125,
      "learning_rate": 3.580870729540641e-05,
      "loss": 0.0002,
      "step": 15465
    },
    {
      "epoch": 7.218669778296383,
      "grad_norm": 0.0184326171875,
      "learning_rate": 3.579745668952359e-05,
      "loss": 0.0002,
      "step": 15466
    },
    {
      "epoch": 7.219136522753792,
      "grad_norm": 0.047607421875,
      "learning_rate": 3.578620746599277e-05,
      "loss": 0.0035,
      "step": 15467
    },
    {
      "epoch": 7.219603267211202,
      "grad_norm": 0.00927734375,
      "learning_rate": 3.577495962505617e-05,
      "loss": 0.0002,
      "step": 15468
    },
    {
      "epoch": 7.220070011668612,
      "grad_norm": 0.01495361328125,
      "learning_rate": 3.576371316695593e-05,
      "loss": 0.0002,
      "step": 15469
    },
    {
      "epoch": 7.220536756126021,
      "grad_norm": 0.01220703125,
      "learning_rate": 3.575246809193421e-05,
      "loss": 0.0002,
      "step": 15470
    },
    {
      "epoch": 7.22100350058343,
      "grad_norm": 0.00567626953125,
      "learning_rate": 3.57412244002332e-05,
      "loss": 0.0001,
      "step": 15471
    },
    {
      "epoch": 7.22147024504084,
      "grad_norm": 0.04638671875,
      "learning_rate": 3.572998209209486e-05,
      "loss": 0.0002,
      "step": 15472
    },
    {
      "epoch": 7.22193698949825,
      "grad_norm": 0.0164794921875,
      "learning_rate": 3.571874116776132e-05,
      "loss": 0.0001,
      "step": 15473
    },
    {
      "epoch": 7.222403733955659,
      "grad_norm": 0.01080322265625,
      "learning_rate": 3.570750162747463e-05,
      "loss": 0.0001,
      "step": 15474
    },
    {
      "epoch": 7.222870478413069,
      "grad_norm": 0.047607421875,
      "learning_rate": 3.5696263471476724e-05,
      "loss": 0.0023,
      "step": 15475
    },
    {
      "epoch": 7.223337222870478,
      "grad_norm": 0.01531982421875,
      "learning_rate": 3.5685026700009616e-05,
      "loss": 0.0002,
      "step": 15476
    },
    {
      "epoch": 7.223803967327888,
      "grad_norm": 0.005523681640625,
      "learning_rate": 3.5673791313315284e-05,
      "loss": 0.0001,
      "step": 15477
    },
    {
      "epoch": 7.224270711785298,
      "grad_norm": 0.021240234375,
      "learning_rate": 3.566255731163556e-05,
      "loss": 0.0003,
      "step": 15478
    },
    {
      "epoch": 7.224737456242707,
      "grad_norm": 0.10107421875,
      "learning_rate": 3.565132469521235e-05,
      "loss": 0.0004,
      "step": 15479
    },
    {
      "epoch": 7.225204200700117,
      "grad_norm": 0.01153564453125,
      "learning_rate": 3.564009346428756e-05,
      "loss": 0.0002,
      "step": 15480
    },
    {
      "epoch": 7.2256709451575265,
      "grad_norm": 0.056640625,
      "learning_rate": 3.562886361910294e-05,
      "loss": 0.0003,
      "step": 15481
    },
    {
      "epoch": 7.226137689614935,
      "grad_norm": 0.014404296875,
      "learning_rate": 3.561763515990029e-05,
      "loss": 0.0002,
      "step": 15482
    },
    {
      "epoch": 7.226604434072345,
      "grad_norm": 0.064453125,
      "learning_rate": 3.560640808692143e-05,
      "loss": 0.0005,
      "step": 15483
    },
    {
      "epoch": 7.227071178529755,
      "grad_norm": 0.013427734375,
      "learning_rate": 3.559518240040801e-05,
      "loss": 0.0002,
      "step": 15484
    },
    {
      "epoch": 7.227537922987165,
      "grad_norm": 0.0142822265625,
      "learning_rate": 3.5583958100601814e-05,
      "loss": 0.0002,
      "step": 15485
    },
    {
      "epoch": 7.228004667444574,
      "grad_norm": 0.0185546875,
      "learning_rate": 3.5572735187744434e-05,
      "loss": 0.0002,
      "step": 15486
    },
    {
      "epoch": 7.2284714119019835,
      "grad_norm": 0.01055908203125,
      "learning_rate": 3.556151366207755e-05,
      "loss": 0.0001,
      "step": 15487
    },
    {
      "epoch": 7.228938156359393,
      "grad_norm": 0.03515625,
      "learning_rate": 3.55502935238428e-05,
      "loss": 0.0027,
      "step": 15488
    },
    {
      "epoch": 7.229404900816803,
      "grad_norm": 0.09521484375,
      "learning_rate": 3.553907477328171e-05,
      "loss": 0.0004,
      "step": 15489
    },
    {
      "epoch": 7.229871645274212,
      "grad_norm": 0.0625,
      "learning_rate": 3.552785741063586e-05,
      "loss": 0.0039,
      "step": 15490
    },
    {
      "epoch": 7.230338389731622,
      "grad_norm": 0.0162353515625,
      "learning_rate": 3.5516641436146816e-05,
      "loss": 0.0002,
      "step": 15491
    },
    {
      "epoch": 7.230805134189032,
      "grad_norm": 0.00933837890625,
      "learning_rate": 3.550542685005598e-05,
      "loss": 0.0002,
      "step": 15492
    },
    {
      "epoch": 7.231271878646441,
      "grad_norm": 0.29296875,
      "learning_rate": 3.549421365260487e-05,
      "loss": 0.001,
      "step": 15493
    },
    {
      "epoch": 7.23173862310385,
      "grad_norm": 0.0301513671875,
      "learning_rate": 3.5483001844034933e-05,
      "loss": 0.0003,
      "step": 15494
    },
    {
      "epoch": 7.23220536756126,
      "grad_norm": 0.0115966796875,
      "learning_rate": 3.547179142458752e-05,
      "loss": 0.0002,
      "step": 15495
    },
    {
      "epoch": 7.23267211201867,
      "grad_norm": 0.1005859375,
      "learning_rate": 3.546058239450404e-05,
      "loss": 0.0012,
      "step": 15496
    },
    {
      "epoch": 7.23313885647608,
      "grad_norm": 0.00762939453125,
      "learning_rate": 3.544937475402585e-05,
      "loss": 0.0001,
      "step": 15497
    },
    {
      "epoch": 7.233605600933489,
      "grad_norm": 0.0147705078125,
      "learning_rate": 3.543816850339423e-05,
      "loss": 0.0002,
      "step": 15498
    },
    {
      "epoch": 7.234072345390898,
      "grad_norm": 0.006744384765625,
      "learning_rate": 3.5426963642850455e-05,
      "loss": 0.0001,
      "step": 15499
    },
    {
      "epoch": 7.234539089848308,
      "grad_norm": 0.00970458984375,
      "learning_rate": 3.5415760172635836e-05,
      "loss": 0.0002,
      "step": 15500
    },
    {
      "epoch": 7.235005834305718,
      "grad_norm": 0.04638671875,
      "learning_rate": 3.540455809299153e-05,
      "loss": 0.0003,
      "step": 15501
    },
    {
      "epoch": 7.235472578763127,
      "grad_norm": 0.00848388671875,
      "learning_rate": 3.539335740415876e-05,
      "loss": 0.0001,
      "step": 15502
    },
    {
      "epoch": 7.235939323220537,
      "grad_norm": 0.015869140625,
      "learning_rate": 3.5382158106378705e-05,
      "loss": 0.0002,
      "step": 15503
    },
    {
      "epoch": 7.2364060676779465,
      "grad_norm": 0.0218505859375,
      "learning_rate": 3.5370960199892465e-05,
      "loss": 0.0002,
      "step": 15504
    },
    {
      "epoch": 7.236872812135356,
      "grad_norm": 0.01434326171875,
      "learning_rate": 3.5359763684941184e-05,
      "loss": 0.0002,
      "step": 15505
    },
    {
      "epoch": 7.237339556592765,
      "grad_norm": 0.259765625,
      "learning_rate": 3.534856856176587e-05,
      "loss": 0.0059,
      "step": 15506
    },
    {
      "epoch": 7.237806301050175,
      "grad_norm": 0.06640625,
      "learning_rate": 3.533737483060762e-05,
      "loss": 0.0031,
      "step": 15507
    },
    {
      "epoch": 7.238273045507585,
      "grad_norm": 0.021728515625,
      "learning_rate": 3.532618249170746e-05,
      "loss": 0.0003,
      "step": 15508
    },
    {
      "epoch": 7.238739789964995,
      "grad_norm": 0.01123046875,
      "learning_rate": 3.531499154530632e-05,
      "loss": 0.0002,
      "step": 15509
    },
    {
      "epoch": 7.2392065344224035,
      "grad_norm": 0.01116943359375,
      "learning_rate": 3.530380199164518e-05,
      "loss": 0.0001,
      "step": 15510
    },
    {
      "epoch": 7.239673278879813,
      "grad_norm": 0.00732421875,
      "learning_rate": 3.5292613830964994e-05,
      "loss": 0.0001,
      "step": 15511
    },
    {
      "epoch": 7.240140023337223,
      "grad_norm": 0.031494140625,
      "learning_rate": 3.528142706350659e-05,
      "loss": 0.0002,
      "step": 15512
    },
    {
      "epoch": 7.240606767794633,
      "grad_norm": 0.01214599609375,
      "learning_rate": 3.527024168951087e-05,
      "loss": 0.0002,
      "step": 15513
    },
    {
      "epoch": 7.241073512252042,
      "grad_norm": 0.011962890625,
      "learning_rate": 3.525905770921871e-05,
      "loss": 0.0002,
      "step": 15514
    },
    {
      "epoch": 7.241540256709452,
      "grad_norm": 0.01513671875,
      "learning_rate": 3.524787512287082e-05,
      "loss": 0.0002,
      "step": 15515
    },
    {
      "epoch": 7.242007001166861,
      "grad_norm": 0.01007080078125,
      "learning_rate": 3.5236693930708034e-05,
      "loss": 0.0001,
      "step": 15516
    },
    {
      "epoch": 7.24247374562427,
      "grad_norm": 0.0390625,
      "learning_rate": 3.522551413297112e-05,
      "loss": 0.0002,
      "step": 15517
    },
    {
      "epoch": 7.24294049008168,
      "grad_norm": 0.01116943359375,
      "learning_rate": 3.521433572990074e-05,
      "loss": 0.0002,
      "step": 15518
    },
    {
      "epoch": 7.24340723453909,
      "grad_norm": 0.051513671875,
      "learning_rate": 3.520315872173757e-05,
      "loss": 0.0003,
      "step": 15519
    },
    {
      "epoch": 7.2438739789965,
      "grad_norm": 0.00811767578125,
      "learning_rate": 3.5191983108722315e-05,
      "loss": 0.0002,
      "step": 15520
    },
    {
      "epoch": 7.244340723453909,
      "grad_norm": 0.01080322265625,
      "learning_rate": 3.518080889109554e-05,
      "loss": 0.0002,
      "step": 15521
    },
    {
      "epoch": 7.244807467911318,
      "grad_norm": 0.01324462890625,
      "learning_rate": 3.5169636069097866e-05,
      "loss": 0.0001,
      "step": 15522
    },
    {
      "epoch": 7.245274212368728,
      "grad_norm": 0.04052734375,
      "learning_rate": 3.515846464296989e-05,
      "loss": 0.0003,
      "step": 15523
    },
    {
      "epoch": 7.245740956826138,
      "grad_norm": 0.134765625,
      "learning_rate": 3.514729461295208e-05,
      "loss": 0.0005,
      "step": 15524
    },
    {
      "epoch": 7.246207701283547,
      "grad_norm": 0.007293701171875,
      "learning_rate": 3.513612597928497e-05,
      "loss": 0.0002,
      "step": 15525
    },
    {
      "epoch": 7.246674445740957,
      "grad_norm": 0.048583984375,
      "learning_rate": 3.512495874220908e-05,
      "loss": 0.0003,
      "step": 15526
    },
    {
      "epoch": 7.2471411901983664,
      "grad_norm": 0.0673828125,
      "learning_rate": 3.511379290196476e-05,
      "loss": 0.0003,
      "step": 15527
    },
    {
      "epoch": 7.247607934655776,
      "grad_norm": 0.023681640625,
      "learning_rate": 3.510262845879248e-05,
      "loss": 0.0002,
      "step": 15528
    },
    {
      "epoch": 7.248074679113185,
      "grad_norm": 0.009521484375,
      "learning_rate": 3.5091465412932646e-05,
      "loss": 0.0002,
      "step": 15529
    },
    {
      "epoch": 7.248541423570595,
      "grad_norm": 0.011474609375,
      "learning_rate": 3.5080303764625544e-05,
      "loss": 0.0001,
      "step": 15530
    },
    {
      "epoch": 7.249008168028005,
      "grad_norm": 0.038330078125,
      "learning_rate": 3.506914351411154e-05,
      "loss": 0.0017,
      "step": 15531
    },
    {
      "epoch": 7.2494749124854145,
      "grad_norm": 0.015380859375,
      "learning_rate": 3.505798466163094e-05,
      "loss": 0.0002,
      "step": 15532
    },
    {
      "epoch": 7.249941656942823,
      "grad_norm": 0.0291748046875,
      "learning_rate": 3.504682720742395e-05,
      "loss": 0.0003,
      "step": 15533
    },
    {
      "epoch": 7.250408401400233,
      "grad_norm": 0.048828125,
      "learning_rate": 3.503567115173088e-05,
      "loss": 0.0028,
      "step": 15534
    },
    {
      "epoch": 7.250875145857643,
      "grad_norm": 0.021728515625,
      "learning_rate": 3.502451649479185e-05,
      "loss": 0.0002,
      "step": 15535
    },
    {
      "epoch": 7.251341890315053,
      "grad_norm": 0.041748046875,
      "learning_rate": 3.501336323684707e-05,
      "loss": 0.0003,
      "step": 15536
    },
    {
      "epoch": 7.251808634772462,
      "grad_norm": 0.01708984375,
      "learning_rate": 3.5002211378136716e-05,
      "loss": 0.0002,
      "step": 15537
    },
    {
      "epoch": 7.2522753792298715,
      "grad_norm": 0.01214599609375,
      "learning_rate": 3.499106091890084e-05,
      "loss": 0.0001,
      "step": 15538
    },
    {
      "epoch": 7.252742123687281,
      "grad_norm": 0.044921875,
      "learning_rate": 3.4979911859379554e-05,
      "loss": 0.0002,
      "step": 15539
    },
    {
      "epoch": 7.253208868144691,
      "grad_norm": 0.00848388671875,
      "learning_rate": 3.496876419981293e-05,
      "loss": 0.0002,
      "step": 15540
    },
    {
      "epoch": 7.2536756126021,
      "grad_norm": 0.01708984375,
      "learning_rate": 3.495761794044095e-05,
      "loss": 0.0002,
      "step": 15541
    },
    {
      "epoch": 7.25414235705951,
      "grad_norm": 0.06640625,
      "learning_rate": 3.4946473081503605e-05,
      "loss": 0.0003,
      "step": 15542
    },
    {
      "epoch": 7.25460910151692,
      "grad_norm": 0.012939453125,
      "learning_rate": 3.493532962324091e-05,
      "loss": 0.0002,
      "step": 15543
    },
    {
      "epoch": 7.255075845974329,
      "grad_norm": 0.00860595703125,
      "learning_rate": 3.492418756589272e-05,
      "loss": 0.0002,
      "step": 15544
    },
    {
      "epoch": 7.255075845974329,
      "eval_loss": 2.3933210372924805,
      "eval_runtime": 84.4828,
      "eval_samples_per_second": 21.353,
      "eval_steps_per_second": 2.675,
      "step": 15544
    },
    {
      "epoch": 7.255542590431738,
      "grad_norm": 0.0115966796875,
      "learning_rate": 3.491304690969899e-05,
      "loss": 0.0002,
      "step": 15545
    },
    {
      "epoch": 7.256009334889148,
      "grad_norm": 0.0157470703125,
      "learning_rate": 3.490190765489959e-05,
      "loss": 0.0002,
      "step": 15546
    },
    {
      "epoch": 7.256476079346558,
      "grad_norm": 0.010498046875,
      "learning_rate": 3.4890769801734327e-05,
      "loss": 0.0002,
      "step": 15547
    },
    {
      "epoch": 7.256942823803968,
      "grad_norm": 0.0184326171875,
      "learning_rate": 3.487963335044302e-05,
      "loss": 0.0002,
      "step": 15548
    },
    {
      "epoch": 7.257409568261377,
      "grad_norm": 0.1435546875,
      "learning_rate": 3.48684983012655e-05,
      "loss": 0.0004,
      "step": 15549
    },
    {
      "epoch": 7.257876312718786,
      "grad_norm": 0.01806640625,
      "learning_rate": 3.485736465444144e-05,
      "loss": 0.0002,
      "step": 15550
    },
    {
      "epoch": 7.258343057176196,
      "grad_norm": 0.00836181640625,
      "learning_rate": 3.484623241021059e-05,
      "loss": 0.0002,
      "step": 15551
    },
    {
      "epoch": 7.258809801633606,
      "grad_norm": 0.044677734375,
      "learning_rate": 3.483510156881269e-05,
      "loss": 0.0003,
      "step": 15552
    },
    {
      "epoch": 7.259276546091015,
      "grad_norm": 0.0126953125,
      "learning_rate": 3.482397213048732e-05,
      "loss": 0.0001,
      "step": 15553
    },
    {
      "epoch": 7.259743290548425,
      "grad_norm": 0.0478515625,
      "learning_rate": 3.481284409547417e-05,
      "loss": 0.0003,
      "step": 15554
    },
    {
      "epoch": 7.2602100350058345,
      "grad_norm": 0.041015625,
      "learning_rate": 3.480171746401278e-05,
      "loss": 0.0002,
      "step": 15555
    },
    {
      "epoch": 7.260676779463244,
      "grad_norm": 0.00799560546875,
      "learning_rate": 3.4790592236342755e-05,
      "loss": 0.0002,
      "step": 15556
    },
    {
      "epoch": 7.261143523920653,
      "grad_norm": 0.0673828125,
      "learning_rate": 3.4779468412703665e-05,
      "loss": 0.0002,
      "step": 15557
    },
    {
      "epoch": 7.261610268378063,
      "grad_norm": 0.0185546875,
      "learning_rate": 3.476834599333495e-05,
      "loss": 0.0001,
      "step": 15558
    },
    {
      "epoch": 7.262077012835473,
      "grad_norm": 0.0576171875,
      "learning_rate": 3.475722497847612e-05,
      "loss": 0.0022,
      "step": 15559
    },
    {
      "epoch": 7.262543757292882,
      "grad_norm": 0.0228271484375,
      "learning_rate": 3.4746105368366656e-05,
      "loss": 0.0002,
      "step": 15560
    },
    {
      "epoch": 7.2630105017502915,
      "grad_norm": 0.0067138671875,
      "learning_rate": 3.473498716324591e-05,
      "loss": 0.0002,
      "step": 15561
    },
    {
      "epoch": 7.263477246207701,
      "grad_norm": 0.006805419921875,
      "learning_rate": 3.472387036335331e-05,
      "loss": 0.0001,
      "step": 15562
    },
    {
      "epoch": 7.263943990665111,
      "grad_norm": 0.01092529296875,
      "learning_rate": 3.471275496892824e-05,
      "loss": 0.0002,
      "step": 15563
    },
    {
      "epoch": 7.264410735122521,
      "grad_norm": 0.1513671875,
      "learning_rate": 3.470164098020995e-05,
      "loss": 0.0004,
      "step": 15564
    },
    {
      "epoch": 7.26487747957993,
      "grad_norm": 0.017578125,
      "learning_rate": 3.4690528397437785e-05,
      "loss": 0.0002,
      "step": 15565
    },
    {
      "epoch": 7.26534422403734,
      "grad_norm": 0.0189208984375,
      "learning_rate": 3.467941722085107e-05,
      "loss": 0.0002,
      "step": 15566
    },
    {
      "epoch": 7.265810968494749,
      "grad_norm": 0.0074462890625,
      "learning_rate": 3.466830745068892e-05,
      "loss": 0.0001,
      "step": 15567
    },
    {
      "epoch": 7.266277712952158,
      "grad_norm": 0.049560546875,
      "learning_rate": 3.4657199087190585e-05,
      "loss": 0.0002,
      "step": 15568
    },
    {
      "epoch": 7.266744457409568,
      "grad_norm": 0.0654296875,
      "learning_rate": 3.464609213059529e-05,
      "loss": 0.0002,
      "step": 15569
    },
    {
      "epoch": 7.267211201866978,
      "grad_norm": 0.01190185546875,
      "learning_rate": 3.463498658114211e-05,
      "loss": 0.0002,
      "step": 15570
    },
    {
      "epoch": 7.267677946324388,
      "grad_norm": 0.0135498046875,
      "learning_rate": 3.462388243907019e-05,
      "loss": 0.0002,
      "step": 15571
    },
    {
      "epoch": 7.268144690781797,
      "grad_norm": 0.013671875,
      "learning_rate": 3.461277970461866e-05,
      "loss": 0.0002,
      "step": 15572
    },
    {
      "epoch": 7.268611435239206,
      "grad_norm": 0.01263427734375,
      "learning_rate": 3.460167837802648e-05,
      "loss": 0.0002,
      "step": 15573
    },
    {
      "epoch": 7.269078179696616,
      "grad_norm": 0.01141357421875,
      "learning_rate": 3.459057845953274e-05,
      "loss": 0.0001,
      "step": 15574
    },
    {
      "epoch": 7.269544924154026,
      "grad_norm": 0.0174560546875,
      "learning_rate": 3.457947994937645e-05,
      "loss": 0.0002,
      "step": 15575
    },
    {
      "epoch": 7.270011668611435,
      "grad_norm": 0.037353515625,
      "learning_rate": 3.456838284779651e-05,
      "loss": 0.0002,
      "step": 15576
    },
    {
      "epoch": 7.270478413068845,
      "grad_norm": 0.049560546875,
      "learning_rate": 3.4557287155031884e-05,
      "loss": 0.0026,
      "step": 15577
    },
    {
      "epoch": 7.2709451575262545,
      "grad_norm": 0.008056640625,
      "learning_rate": 3.4546192871321505e-05,
      "loss": 0.0002,
      "step": 15578
    },
    {
      "epoch": 7.271411901983664,
      "grad_norm": 0.038818359375,
      "learning_rate": 3.4535099996904183e-05,
      "loss": 0.0002,
      "step": 15579
    },
    {
      "epoch": 7.271878646441073,
      "grad_norm": 0.197265625,
      "learning_rate": 3.452400853201879e-05,
      "loss": 0.0005,
      "step": 15580
    },
    {
      "epoch": 7.272345390898483,
      "grad_norm": 0.046875,
      "learning_rate": 3.4512918476904176e-05,
      "loss": 0.0004,
      "step": 15581
    },
    {
      "epoch": 7.272812135355893,
      "grad_norm": 0.0517578125,
      "learning_rate": 3.450182983179906e-05,
      "loss": 0.0003,
      "step": 15582
    },
    {
      "epoch": 7.273278879813303,
      "grad_norm": 0.08984375,
      "learning_rate": 3.449074259694224e-05,
      "loss": 0.0003,
      "step": 15583
    },
    {
      "epoch": 7.2737456242707115,
      "grad_norm": 0.01031494140625,
      "learning_rate": 3.447965677257239e-05,
      "loss": 0.0002,
      "step": 15584
    },
    {
      "epoch": 7.274212368728121,
      "grad_norm": 0.083984375,
      "learning_rate": 3.4468572358928233e-05,
      "loss": 0.0003,
      "step": 15585
    },
    {
      "epoch": 7.274679113185531,
      "grad_norm": 0.0157470703125,
      "learning_rate": 3.445748935624844e-05,
      "loss": 0.0002,
      "step": 15586
    },
    {
      "epoch": 7.275145857642941,
      "grad_norm": 0.0142822265625,
      "learning_rate": 3.4446407764771596e-05,
      "loss": 0.0002,
      "step": 15587
    },
    {
      "epoch": 7.27561260210035,
      "grad_norm": 0.0380859375,
      "learning_rate": 3.443532758473632e-05,
      "loss": 0.0022,
      "step": 15588
    },
    {
      "epoch": 7.27607934655776,
      "grad_norm": 0.00982666015625,
      "learning_rate": 3.442424881638123e-05,
      "loss": 0.0002,
      "step": 15589
    },
    {
      "epoch": 7.276546091015169,
      "grad_norm": 0.0595703125,
      "learning_rate": 3.4413171459944784e-05,
      "loss": 0.0024,
      "step": 15590
    },
    {
      "epoch": 7.277012835472579,
      "grad_norm": 0.021728515625,
      "learning_rate": 3.440209551566552e-05,
      "loss": 0.0002,
      "step": 15591
    },
    {
      "epoch": 7.277479579929988,
      "grad_norm": 0.01239013671875,
      "learning_rate": 3.439102098378196e-05,
      "loss": 0.0002,
      "step": 15592
    },
    {
      "epoch": 7.277946324387398,
      "grad_norm": 0.00921630859375,
      "learning_rate": 3.437994786453248e-05,
      "loss": 0.0002,
      "step": 15593
    },
    {
      "epoch": 7.278413068844808,
      "grad_norm": 0.009765625,
      "learning_rate": 3.436887615815553e-05,
      "loss": 0.0002,
      "step": 15594
    },
    {
      "epoch": 7.2788798133022174,
      "grad_norm": 0.00885009765625,
      "learning_rate": 3.4357805864889525e-05,
      "loss": 0.0001,
      "step": 15595
    },
    {
      "epoch": 7.279346557759626,
      "grad_norm": 0.0185546875,
      "learning_rate": 3.4346736984972764e-05,
      "loss": 0.0002,
      "step": 15596
    },
    {
      "epoch": 7.279813302217036,
      "grad_norm": 0.0113525390625,
      "learning_rate": 3.4335669518643585e-05,
      "loss": 0.0002,
      "step": 15597
    },
    {
      "epoch": 7.280280046674446,
      "grad_norm": 0.0126953125,
      "learning_rate": 3.4324603466140345e-05,
      "loss": 0.0002,
      "step": 15598
    },
    {
      "epoch": 7.280746791131856,
      "grad_norm": 0.06884765625,
      "learning_rate": 3.4313538827701216e-05,
      "loss": 0.0044,
      "step": 15599
    },
    {
      "epoch": 7.281213535589265,
      "grad_norm": 0.0250244140625,
      "learning_rate": 3.430247560356449e-05,
      "loss": 0.0003,
      "step": 15600
    },
    {
      "epoch": 7.2816802800466744,
      "grad_norm": 0.05322265625,
      "learning_rate": 3.429141379396838e-05,
      "loss": 0.003,
      "step": 15601
    },
    {
      "epoch": 7.282147024504084,
      "grad_norm": 0.01904296875,
      "learning_rate": 3.4280353399151e-05,
      "loss": 0.0001,
      "step": 15602
    },
    {
      "epoch": 7.282613768961493,
      "grad_norm": 0.04736328125,
      "learning_rate": 3.4269294419350573e-05,
      "loss": 0.0006,
      "step": 15603
    },
    {
      "epoch": 7.283080513418903,
      "grad_norm": 0.0274658203125,
      "learning_rate": 3.425823685480513e-05,
      "loss": 0.0002,
      "step": 15604
    },
    {
      "epoch": 7.283547257876313,
      "grad_norm": 0.01116943359375,
      "learning_rate": 3.4247180705752776e-05,
      "loss": 0.0002,
      "step": 15605
    },
    {
      "epoch": 7.2840140023337225,
      "grad_norm": 0.010986328125,
      "learning_rate": 3.423612597243162e-05,
      "loss": 0.0002,
      "step": 15606
    },
    {
      "epoch": 7.2844807467911314,
      "grad_norm": 0.01226806640625,
      "learning_rate": 3.42250726550796e-05,
      "loss": 0.0002,
      "step": 15607
    },
    {
      "epoch": 7.284947491248541,
      "grad_norm": 0.1376953125,
      "learning_rate": 3.421402075393476e-05,
      "loss": 0.0005,
      "step": 15608
    },
    {
      "epoch": 7.285414235705951,
      "grad_norm": 0.1416015625,
      "learning_rate": 3.420297026923507e-05,
      "loss": 0.0005,
      "step": 15609
    },
    {
      "epoch": 7.285880980163361,
      "grad_norm": 0.0108642578125,
      "learning_rate": 3.4191921201218394e-05,
      "loss": 0.0002,
      "step": 15610
    },
    {
      "epoch": 7.28634772462077,
      "grad_norm": 0.029541015625,
      "learning_rate": 3.418087355012266e-05,
      "loss": 0.0002,
      "step": 15611
    },
    {
      "epoch": 7.2868144690781795,
      "grad_norm": 0.029296875,
      "learning_rate": 3.41698273161858e-05,
      "loss": 0.0002,
      "step": 15612
    },
    {
      "epoch": 7.287281213535589,
      "grad_norm": 0.036865234375,
      "learning_rate": 3.4158782499645545e-05,
      "loss": 0.0003,
      "step": 15613
    },
    {
      "epoch": 7.287747957992999,
      "grad_norm": 0.019775390625,
      "learning_rate": 3.414773910073981e-05,
      "loss": 0.0002,
      "step": 15614
    },
    {
      "epoch": 7.288214702450408,
      "grad_norm": 0.0267333984375,
      "learning_rate": 3.4136697119706294e-05,
      "loss": 0.0002,
      "step": 15615
    },
    {
      "epoch": 7.288681446907818,
      "grad_norm": 0.00946044921875,
      "learning_rate": 3.4125656556782746e-05,
      "loss": 0.0002,
      "step": 15616
    },
    {
      "epoch": 7.289148191365228,
      "grad_norm": 0.01055908203125,
      "learning_rate": 3.4114617412206903e-05,
      "loss": 0.0002,
      "step": 15617
    },
    {
      "epoch": 7.289614935822637,
      "grad_norm": 0.05859375,
      "learning_rate": 3.410357968621648e-05,
      "loss": 0.0002,
      "step": 15618
    },
    {
      "epoch": 7.290081680280046,
      "grad_norm": 0.02099609375,
      "learning_rate": 3.409254337904906e-05,
      "loss": 0.0002,
      "step": 15619
    },
    {
      "epoch": 7.290548424737456,
      "grad_norm": 0.032958984375,
      "learning_rate": 3.4081508490942306e-05,
      "loss": 0.002,
      "step": 15620
    },
    {
      "epoch": 7.291015169194866,
      "grad_norm": 0.022216796875,
      "learning_rate": 3.4070475022133855e-05,
      "loss": 0.0004,
      "step": 15621
    },
    {
      "epoch": 7.291481913652276,
      "grad_norm": 0.0177001953125,
      "learning_rate": 3.405944297286119e-05,
      "loss": 0.0002,
      "step": 15622
    },
    {
      "epoch": 7.291948658109685,
      "grad_norm": 0.032470703125,
      "learning_rate": 3.404841234336187e-05,
      "loss": 0.0015,
      "step": 15623
    },
    {
      "epoch": 7.292415402567094,
      "grad_norm": 0.006988525390625,
      "learning_rate": 3.403738313387345e-05,
      "loss": 0.0001,
      "step": 15624
    },
    {
      "epoch": 7.292882147024504,
      "grad_norm": 0.0291748046875,
      "learning_rate": 3.402635534463332e-05,
      "loss": 0.0003,
      "step": 15625
    },
    {
      "epoch": 7.293348891481914,
      "grad_norm": 0.05517578125,
      "learning_rate": 3.401532897587897e-05,
      "loss": 0.0021,
      "step": 15626
    },
    {
      "epoch": 7.293815635939323,
      "grad_norm": 0.0185546875,
      "learning_rate": 3.4004304027847814e-05,
      "loss": 0.0002,
      "step": 15627
    },
    {
      "epoch": 7.294282380396733,
      "grad_norm": 0.0272216796875,
      "learning_rate": 3.399328050077719e-05,
      "loss": 0.0002,
      "step": 15628
    },
    {
      "epoch": 7.2947491248541425,
      "grad_norm": 0.05859375,
      "learning_rate": 3.398225839490447e-05,
      "loss": 0.0028,
      "step": 15629
    },
    {
      "epoch": 7.295215869311552,
      "grad_norm": 0.01055908203125,
      "learning_rate": 3.3971237710467017e-05,
      "loss": 0.0002,
      "step": 15630
    },
    {
      "epoch": 7.295682613768961,
      "grad_norm": 0.048095703125,
      "learning_rate": 3.396021844770204e-05,
      "loss": 0.0003,
      "step": 15631
    },
    {
      "epoch": 7.296149358226371,
      "grad_norm": 0.060791015625,
      "learning_rate": 3.3949200606846865e-05,
      "loss": 0.0044,
      "step": 15632
    },
    {
      "epoch": 7.296616102683781,
      "grad_norm": 0.027587890625,
      "learning_rate": 3.3938184188138644e-05,
      "loss": 0.0002,
      "step": 15633
    },
    {
      "epoch": 7.297082847141191,
      "grad_norm": 0.022216796875,
      "learning_rate": 3.392716919181462e-05,
      "loss": 0.0002,
      "step": 15634
    },
    {
      "epoch": 7.2975495915985995,
      "grad_norm": 0.0111083984375,
      "learning_rate": 3.391615561811199e-05,
      "loss": 0.0002,
      "step": 15635
    },
    {
      "epoch": 7.298016336056009,
      "grad_norm": 0.07958984375,
      "learning_rate": 3.390514346726782e-05,
      "loss": 0.0033,
      "step": 15636
    },
    {
      "epoch": 7.298483080513419,
      "grad_norm": 0.057861328125,
      "learning_rate": 3.389413273951924e-05,
      "loss": 0.0039,
      "step": 15637
    },
    {
      "epoch": 7.298949824970828,
      "grad_norm": 0.01275634765625,
      "learning_rate": 3.388312343510336e-05,
      "loss": 0.0002,
      "step": 15638
    },
    {
      "epoch": 7.299416569428238,
      "grad_norm": 0.0458984375,
      "learning_rate": 3.387211555425715e-05,
      "loss": 0.0021,
      "step": 15639
    },
    {
      "epoch": 7.299883313885648,
      "grad_norm": 0.017822265625,
      "learning_rate": 3.386110909721767e-05,
      "loss": 0.0002,
      "step": 15640
    },
    {
      "epoch": 7.300350058343057,
      "grad_norm": 0.033203125,
      "learning_rate": 3.3850104064221924e-05,
      "loss": 0.0003,
      "step": 15641
    },
    {
      "epoch": 7.300816802800467,
      "grad_norm": 0.01202392578125,
      "learning_rate": 3.38391004555068e-05,
      "loss": 0.0002,
      "step": 15642
    },
    {
      "epoch": 7.301283547257876,
      "grad_norm": 0.039794921875,
      "learning_rate": 3.382809827130925e-05,
      "loss": 0.002,
      "step": 15643
    },
    {
      "epoch": 7.301750291715286,
      "grad_norm": 0.0184326171875,
      "learning_rate": 3.38170975118662e-05,
      "loss": 0.0002,
      "step": 15644
    },
    {
      "epoch": 7.302217036172696,
      "grad_norm": 0.0274658203125,
      "learning_rate": 3.3806098177414435e-05,
      "loss": 0.0017,
      "step": 15645
    },
    {
      "epoch": 7.302683780630105,
      "grad_norm": 0.01318359375,
      "learning_rate": 3.3795100268190815e-05,
      "loss": 0.0002,
      "step": 15646
    },
    {
      "epoch": 7.303150525087514,
      "grad_norm": 0.01287841796875,
      "learning_rate": 3.378410378443219e-05,
      "loss": 0.0002,
      "step": 15647
    },
    {
      "epoch": 7.303617269544924,
      "grad_norm": 0.01123046875,
      "learning_rate": 3.377310872637524e-05,
      "loss": 0.0002,
      "step": 15648
    },
    {
      "epoch": 7.304084014002334,
      "grad_norm": 0.021728515625,
      "learning_rate": 3.376211509425673e-05,
      "loss": 0.0002,
      "step": 15649
    },
    {
      "epoch": 7.304550758459743,
      "grad_norm": 0.04150390625,
      "learning_rate": 3.375112288831341e-05,
      "loss": 0.0003,
      "step": 15650
    },
    {
      "epoch": 7.305017502917153,
      "grad_norm": 0.19921875,
      "learning_rate": 3.374013210878189e-05,
      "loss": 0.0006,
      "step": 15651
    },
    {
      "epoch": 7.3054842473745625,
      "grad_norm": 0.058349609375,
      "learning_rate": 3.372914275589887e-05,
      "loss": 0.0028,
      "step": 15652
    },
    {
      "epoch": 7.305950991831972,
      "grad_norm": 0.0284423828125,
      "learning_rate": 3.3718154829900906e-05,
      "loss": 0.0002,
      "step": 15653
    },
    {
      "epoch": 7.306417736289381,
      "grad_norm": 0.0111083984375,
      "learning_rate": 3.370716833102461e-05,
      "loss": 0.0002,
      "step": 15654
    },
    {
      "epoch": 7.306884480746791,
      "grad_norm": 0.0111083984375,
      "learning_rate": 3.3696183259506566e-05,
      "loss": 0.0002,
      "step": 15655
    },
    {
      "epoch": 7.307351225204201,
      "grad_norm": 0.0093994140625,
      "learning_rate": 3.368519961558322e-05,
      "loss": 0.0002,
      "step": 15656
    },
    {
      "epoch": 7.307817969661611,
      "grad_norm": 0.0230712890625,
      "learning_rate": 3.367421739949111e-05,
      "loss": 0.0002,
      "step": 15657
    },
    {
      "epoch": 7.3082847141190195,
      "grad_norm": 0.031494140625,
      "learning_rate": 3.3663236611466734e-05,
      "loss": 0.0002,
      "step": 15658
    },
    {
      "epoch": 7.308751458576429,
      "grad_norm": 0.01025390625,
      "learning_rate": 3.3652257251746424e-05,
      "loss": 0.0002,
      "step": 15659
    },
    {
      "epoch": 7.309218203033839,
      "grad_norm": 0.0093994140625,
      "learning_rate": 3.364127932056664e-05,
      "loss": 0.0002,
      "step": 15660
    },
    {
      "epoch": 7.309684947491249,
      "grad_norm": 0.01043701171875,
      "learning_rate": 3.3630302818163775e-05,
      "loss": 0.0002,
      "step": 15661
    },
    {
      "epoch": 7.310151691948658,
      "grad_norm": 0.01336669921875,
      "learning_rate": 3.361932774477412e-05,
      "loss": 0.0002,
      "step": 15662
    },
    {
      "epoch": 7.310618436406068,
      "grad_norm": 0.0244140625,
      "learning_rate": 3.3608354100633954e-05,
      "loss": 0.0002,
      "step": 15663
    },
    {
      "epoch": 7.311085180863477,
      "grad_norm": 0.01416015625,
      "learning_rate": 3.359738188597963e-05,
      "loss": 0.0002,
      "step": 15664
    },
    {
      "epoch": 7.311551925320887,
      "grad_norm": 0.0218505859375,
      "learning_rate": 3.358641110104731e-05,
      "loss": 0.002,
      "step": 15665
    },
    {
      "epoch": 7.312018669778296,
      "grad_norm": 0.0286865234375,
      "learning_rate": 3.357544174607324e-05,
      "loss": 0.0002,
      "step": 15666
    },
    {
      "epoch": 7.312485414235706,
      "grad_norm": 0.00921630859375,
      "learning_rate": 3.356447382129365e-05,
      "loss": 0.0002,
      "step": 15667
    },
    {
      "epoch": 7.312952158693116,
      "grad_norm": 0.00946044921875,
      "learning_rate": 3.355350732694462e-05,
      "loss": 0.0002,
      "step": 15668
    },
    {
      "epoch": 7.3134189031505255,
      "grad_norm": 0.134765625,
      "learning_rate": 3.3542542263262286e-05,
      "loss": 0.0003,
      "step": 15669
    },
    {
      "epoch": 7.313885647607934,
      "grad_norm": 0.009033203125,
      "learning_rate": 3.353157863048281e-05,
      "loss": 0.0002,
      "step": 15670
    },
    {
      "epoch": 7.314352392065344,
      "grad_norm": 0.05712890625,
      "learning_rate": 3.3520616428842145e-05,
      "loss": 0.0003,
      "step": 15671
    },
    {
      "epoch": 7.314819136522754,
      "grad_norm": 0.030029296875,
      "learning_rate": 3.350965565857637e-05,
      "loss": 0.0003,
      "step": 15672
    },
    {
      "epoch": 7.315285880980164,
      "grad_norm": 0.06884765625,
      "learning_rate": 3.349869631992152e-05,
      "loss": 0.0003,
      "step": 15673
    },
    {
      "epoch": 7.315752625437573,
      "grad_norm": 0.051513671875,
      "learning_rate": 3.348773841311348e-05,
      "loss": 0.0003,
      "step": 15674
    },
    {
      "epoch": 7.3162193698949824,
      "grad_norm": 0.0262451171875,
      "learning_rate": 3.3476781938388235e-05,
      "loss": 0.0002,
      "step": 15675
    },
    {
      "epoch": 7.316686114352392,
      "grad_norm": 0.0113525390625,
      "learning_rate": 3.346582689598171e-05,
      "loss": 0.0002,
      "step": 15676
    },
    {
      "epoch": 7.317152858809802,
      "grad_norm": 0.0164794921875,
      "learning_rate": 3.345487328612973e-05,
      "loss": 0.0002,
      "step": 15677
    },
    {
      "epoch": 7.317619603267211,
      "grad_norm": 0.0289306640625,
      "learning_rate": 3.3443921109068154e-05,
      "loss": 0.0002,
      "step": 15678
    },
    {
      "epoch": 7.318086347724621,
      "grad_norm": 0.01214599609375,
      "learning_rate": 3.3432970365032834e-05,
      "loss": 0.0002,
      "step": 15679
    },
    {
      "epoch": 7.3185530921820305,
      "grad_norm": 0.006439208984375,
      "learning_rate": 3.342202105425949e-05,
      "loss": 0.0001,
      "step": 15680
    },
    {
      "epoch": 7.3190198366394394,
      "grad_norm": 0.02392578125,
      "learning_rate": 3.34110731769839e-05,
      "loss": 0.0002,
      "step": 15681
    },
    {
      "epoch": 7.319486581096849,
      "grad_norm": 0.04052734375,
      "learning_rate": 3.340012673344183e-05,
      "loss": 0.0027,
      "step": 15682
    },
    {
      "epoch": 7.319953325554259,
      "grad_norm": 0.0159912109375,
      "learning_rate": 3.338918172386888e-05,
      "loss": 0.0002,
      "step": 15683
    },
    {
      "epoch": 7.320420070011669,
      "grad_norm": 0.03466796875,
      "learning_rate": 3.337823814850078e-05,
      "loss": 0.0021,
      "step": 15684
    },
    {
      "epoch": 7.320886814469079,
      "grad_norm": 0.00787353515625,
      "learning_rate": 3.3367296007573115e-05,
      "loss": 0.0001,
      "step": 15685
    },
    {
      "epoch": 7.3213535589264875,
      "grad_norm": 0.0179443359375,
      "learning_rate": 3.335635530132148e-05,
      "loss": 0.0002,
      "step": 15686
    },
    {
      "epoch": 7.321820303383897,
      "grad_norm": 0.048828125,
      "learning_rate": 3.33454160299815e-05,
      "loss": 0.0022,
      "step": 15687
    },
    {
      "epoch": 7.322287047841307,
      "grad_norm": 0.09912109375,
      "learning_rate": 3.333447819378864e-05,
      "loss": 0.0069,
      "step": 15688
    },
    {
      "epoch": 7.322753792298716,
      "grad_norm": 0.004547119140625,
      "learning_rate": 3.332354179297843e-05,
      "loss": 0.0001,
      "step": 15689
    },
    {
      "epoch": 7.323220536756126,
      "grad_norm": 0.00970458984375,
      "learning_rate": 3.331260682778637e-05,
      "loss": 0.0002,
      "step": 15690
    },
    {
      "epoch": 7.323687281213536,
      "grad_norm": 0.33984375,
      "learning_rate": 3.3301673298447844e-05,
      "loss": 0.0008,
      "step": 15691
    },
    {
      "epoch": 7.324154025670945,
      "grad_norm": 0.0223388671875,
      "learning_rate": 3.3290741205198294e-05,
      "loss": 0.0002,
      "step": 15692
    },
    {
      "epoch": 7.324620770128354,
      "grad_norm": 0.0150146484375,
      "learning_rate": 3.327981054827314e-05,
      "loss": 0.0002,
      "step": 15693
    },
    {
      "epoch": 7.325087514585764,
      "grad_norm": 0.010986328125,
      "learning_rate": 3.3268881327907654e-05,
      "loss": 0.0002,
      "step": 15694
    },
    {
      "epoch": 7.325554259043174,
      "grad_norm": 0.007080078125,
      "learning_rate": 3.3257953544337196e-05,
      "loss": 0.0001,
      "step": 15695
    },
    {
      "epoch": 7.326021003500584,
      "grad_norm": 0.011962890625,
      "learning_rate": 3.324702719779709e-05,
      "loss": 0.0002,
      "step": 15696
    },
    {
      "epoch": 7.326487747957993,
      "grad_norm": 0.01068115234375,
      "learning_rate": 3.3236102288522506e-05,
      "loss": 0.0001,
      "step": 15697
    },
    {
      "epoch": 7.326954492415402,
      "grad_norm": 0.016845703125,
      "learning_rate": 3.3225178816748715e-05,
      "loss": 0.0002,
      "step": 15698
    },
    {
      "epoch": 7.327421236872812,
      "grad_norm": 0.01190185546875,
      "learning_rate": 3.321425678271095e-05,
      "loss": 0.0002,
      "step": 15699
    },
    {
      "epoch": 7.327887981330222,
      "grad_norm": 0.010009765625,
      "learning_rate": 3.3203336186644306e-05,
      "loss": 0.0002,
      "step": 15700
    },
    {
      "epoch": 7.328354725787631,
      "grad_norm": 0.006195068359375,
      "learning_rate": 3.319241702878397e-05,
      "loss": 0.0001,
      "step": 15701
    },
    {
      "epoch": 7.328821470245041,
      "grad_norm": 0.0693359375,
      "learning_rate": 3.318149930936499e-05,
      "loss": 0.0058,
      "step": 15702
    },
    {
      "epoch": 7.3292882147024505,
      "grad_norm": 0.1328125,
      "learning_rate": 3.317058302862248e-05,
      "loss": 0.0003,
      "step": 15703
    },
    {
      "epoch": 7.32975495915986,
      "grad_norm": 0.0198974609375,
      "learning_rate": 3.315966818679148e-05,
      "loss": 0.0002,
      "step": 15704
    },
    {
      "epoch": 7.330221703617269,
      "grad_norm": 0.055908203125,
      "learning_rate": 3.3148754784106976e-05,
      "loss": 0.0003,
      "step": 15705
    },
    {
      "epoch": 7.330688448074679,
      "grad_norm": 0.01495361328125,
      "learning_rate": 3.313784282080393e-05,
      "loss": 0.0002,
      "step": 15706
    },
    {
      "epoch": 7.331155192532089,
      "grad_norm": 0.00927734375,
      "learning_rate": 3.3126932297117356e-05,
      "loss": 0.0001,
      "step": 15707
    },
    {
      "epoch": 7.331621936989499,
      "grad_norm": 0.059326171875,
      "learning_rate": 3.3116023213282085e-05,
      "loss": 0.0015,
      "step": 15708
    },
    {
      "epoch": 7.3320886814469075,
      "grad_norm": 0.0211181640625,
      "learning_rate": 3.310511556953305e-05,
      "loss": 0.0002,
      "step": 15709
    },
    {
      "epoch": 7.332555425904317,
      "grad_norm": 0.054443359375,
      "learning_rate": 3.309420936610513e-05,
      "loss": 0.0002,
      "step": 15710
    },
    {
      "epoch": 7.333022170361727,
      "grad_norm": 0.01397705078125,
      "learning_rate": 3.308330460323312e-05,
      "loss": 0.0001,
      "step": 15711
    },
    {
      "epoch": 7.333488914819137,
      "grad_norm": 0.2412109375,
      "learning_rate": 3.307240128115178e-05,
      "loss": 0.0005,
      "step": 15712
    },
    {
      "epoch": 7.333955659276546,
      "grad_norm": 0.050048828125,
      "learning_rate": 3.306149940009592e-05,
      "loss": 0.0033,
      "step": 15713
    },
    {
      "epoch": 7.334422403733956,
      "grad_norm": 0.0166015625,
      "learning_rate": 3.3050598960300205e-05,
      "loss": 0.0002,
      "step": 15714
    },
    {
      "epoch": 7.334889148191365,
      "grad_norm": 0.0164794921875,
      "learning_rate": 3.303969996199939e-05,
      "loss": 0.0002,
      "step": 15715
    },
    {
      "epoch": 7.335355892648775,
      "grad_norm": 0.0247802734375,
      "learning_rate": 3.302880240542816e-05,
      "loss": 0.0002,
      "step": 15716
    },
    {
      "epoch": 7.335822637106184,
      "grad_norm": 0.040771484375,
      "learning_rate": 3.301790629082109e-05,
      "loss": 0.0023,
      "step": 15717
    },
    {
      "epoch": 7.336289381563594,
      "grad_norm": 0.01129150390625,
      "learning_rate": 3.3007011618412806e-05,
      "loss": 0.0002,
      "step": 15718
    },
    {
      "epoch": 7.336756126021004,
      "grad_norm": 0.01611328125,
      "learning_rate": 3.2996118388437936e-05,
      "loss": 0.0002,
      "step": 15719
    },
    {
      "epoch": 7.3372228704784135,
      "grad_norm": 0.01055908203125,
      "learning_rate": 3.2985226601130936e-05,
      "loss": 0.0001,
      "step": 15720
    },
    {
      "epoch": 7.337689614935822,
      "grad_norm": 0.005523681640625,
      "learning_rate": 3.297433625672637e-05,
      "loss": 0.0001,
      "step": 15721
    },
    {
      "epoch": 7.338156359393232,
      "grad_norm": 0.014892578125,
      "learning_rate": 3.296344735545873e-05,
      "loss": 0.0002,
      "step": 15722
    },
    {
      "epoch": 7.338623103850642,
      "grad_norm": 0.011474609375,
      "learning_rate": 3.295255989756243e-05,
      "loss": 0.0002,
      "step": 15723
    },
    {
      "epoch": 7.339089848308051,
      "grad_norm": 0.037353515625,
      "learning_rate": 3.294167388327189e-05,
      "loss": 0.0002,
      "step": 15724
    },
    {
      "epoch": 7.339556592765461,
      "grad_norm": 0.0240478515625,
      "learning_rate": 3.293078931282156e-05,
      "loss": 0.0002,
      "step": 15725
    },
    {
      "epoch": 7.3400233372228705,
      "grad_norm": 0.04150390625,
      "learning_rate": 3.2919906186445714e-05,
      "loss": 0.0017,
      "step": 15726
    },
    {
      "epoch": 7.34049008168028,
      "grad_norm": 0.0167236328125,
      "learning_rate": 3.2909024504378716e-05,
      "loss": 0.0002,
      "step": 15727
    },
    {
      "epoch": 7.34095682613769,
      "grad_norm": 0.00994873046875,
      "learning_rate": 3.2898144266854894e-05,
      "loss": 0.0002,
      "step": 15728
    },
    {
      "epoch": 7.341423570595099,
      "grad_norm": 0.01434326171875,
      "learning_rate": 3.2887265474108434e-05,
      "loss": 0.0002,
      "step": 15729
    },
    {
      "epoch": 7.341890315052509,
      "grad_norm": 0.08984375,
      "learning_rate": 3.287638812637361e-05,
      "loss": 0.0003,
      "step": 15730
    },
    {
      "epoch": 7.342357059509919,
      "grad_norm": 0.00897216796875,
      "learning_rate": 3.286551222388468e-05,
      "loss": 0.0002,
      "step": 15731
    },
    {
      "epoch": 7.3428238039673275,
      "grad_norm": 0.01116943359375,
      "learning_rate": 3.285463776687571e-05,
      "loss": 0.0002,
      "step": 15732
    },
    {
      "epoch": 7.343290548424737,
      "grad_norm": 0.045654296875,
      "learning_rate": 3.284376475558092e-05,
      "loss": 0.0016,
      "step": 15733
    },
    {
      "epoch": 7.343757292882147,
      "grad_norm": 0.0203857421875,
      "learning_rate": 3.2832893190234346e-05,
      "loss": 0.0002,
      "step": 15734
    },
    {
      "epoch": 7.344224037339557,
      "grad_norm": 0.0673828125,
      "learning_rate": 3.282202307107012e-05,
      "loss": 0.0028,
      "step": 15735
    },
    {
      "epoch": 7.344690781796966,
      "grad_norm": 0.0252685546875,
      "learning_rate": 3.2811154398322295e-05,
      "loss": 0.0002,
      "step": 15736
    },
    {
      "epoch": 7.345157526254376,
      "grad_norm": 0.134765625,
      "learning_rate": 3.2800287172224834e-05,
      "loss": 0.0024,
      "step": 15737
    },
    {
      "epoch": 7.345624270711785,
      "grad_norm": 0.053955078125,
      "learning_rate": 3.2789421393011746e-05,
      "loss": 0.0003,
      "step": 15738
    },
    {
      "epoch": 7.346091015169195,
      "grad_norm": 0.0198974609375,
      "learning_rate": 3.2778557060917025e-05,
      "loss": 0.0002,
      "step": 15739
    },
    {
      "epoch": 7.346557759626604,
      "grad_norm": 0.10205078125,
      "learning_rate": 3.276769417617452e-05,
      "loss": 0.0004,
      "step": 15740
    },
    {
      "epoch": 7.347024504084014,
      "grad_norm": 0.0252685546875,
      "learning_rate": 3.275683273901815e-05,
      "loss": 0.0002,
      "step": 15741
    },
    {
      "epoch": 7.347491248541424,
      "grad_norm": 0.00799560546875,
      "learning_rate": 3.274597274968182e-05,
      "loss": 0.0001,
      "step": 15742
    },
    {
      "epoch": 7.3479579929988335,
      "grad_norm": 0.025634765625,
      "learning_rate": 3.273511420839927e-05,
      "loss": 0.0021,
      "step": 15743
    },
    {
      "epoch": 7.348424737456242,
      "grad_norm": 0.0380859375,
      "learning_rate": 3.272425711540435e-05,
      "loss": 0.003,
      "step": 15744
    },
    {
      "epoch": 7.348891481913652,
      "grad_norm": 0.00762939453125,
      "learning_rate": 3.271340147093086e-05,
      "loss": 0.0001,
      "step": 15745
    },
    {
      "epoch": 7.349358226371062,
      "grad_norm": 0.06298828125,
      "learning_rate": 3.270254727521246e-05,
      "loss": 0.0003,
      "step": 15746
    },
    {
      "epoch": 7.349824970828472,
      "grad_norm": 0.01495361328125,
      "learning_rate": 3.269169452848288e-05,
      "loss": 0.0002,
      "step": 15747
    },
    {
      "epoch": 7.350291715285881,
      "grad_norm": 0.07177734375,
      "learning_rate": 3.268084323097582e-05,
      "loss": 0.0003,
      "step": 15748
    },
    {
      "epoch": 7.3507584597432905,
      "grad_norm": 0.0341796875,
      "learning_rate": 3.266999338292487e-05,
      "loss": 0.0018,
      "step": 15749
    },
    {
      "epoch": 7.3512252042007,
      "grad_norm": 0.051025390625,
      "learning_rate": 3.2659144984563695e-05,
      "loss": 0.0027,
      "step": 15750
    },
    {
      "epoch": 7.35169194865811,
      "grad_norm": 0.03759765625,
      "learning_rate": 3.2648298036125825e-05,
      "loss": 0.0002,
      "step": 15751
    },
    {
      "epoch": 7.352158693115519,
      "grad_norm": 0.0081787109375,
      "learning_rate": 3.2637452537844816e-05,
      "loss": 0.0002,
      "step": 15752
    },
    {
      "epoch": 7.352625437572929,
      "grad_norm": 0.2392578125,
      "learning_rate": 3.262660848995423e-05,
      "loss": 0.0007,
      "step": 15753
    },
    {
      "epoch": 7.3530921820303385,
      "grad_norm": 0.1083984375,
      "learning_rate": 3.261576589268748e-05,
      "loss": 0.0059,
      "step": 15754
    },
    {
      "epoch": 7.353558926487748,
      "grad_norm": 0.048828125,
      "learning_rate": 3.2604924746278055e-05,
      "loss": 0.003,
      "step": 15755
    },
    {
      "epoch": 7.354025670945157,
      "grad_norm": 0.0081787109375,
      "learning_rate": 3.259408505095941e-05,
      "loss": 0.0001,
      "step": 15756
    },
    {
      "epoch": 7.354492415402567,
      "grad_norm": 0.0130615234375,
      "learning_rate": 3.2583246806964896e-05,
      "loss": 0.0002,
      "step": 15757
    },
    {
      "epoch": 7.354959159859977,
      "grad_norm": 0.0322265625,
      "learning_rate": 3.2572410014527856e-05,
      "loss": 0.0002,
      "step": 15758
    },
    {
      "epoch": 7.355425904317387,
      "grad_norm": 0.02197265625,
      "learning_rate": 3.256157467388163e-05,
      "loss": 0.0002,
      "step": 15759
    },
    {
      "epoch": 7.3558926487747955,
      "grad_norm": 0.037353515625,
      "learning_rate": 3.255074078525957e-05,
      "loss": 0.0035,
      "step": 15760
    },
    {
      "epoch": 7.356359393232205,
      "grad_norm": 0.0322265625,
      "learning_rate": 3.2539908348894855e-05,
      "loss": 0.0002,
      "step": 15761
    },
    {
      "epoch": 7.356826137689615,
      "grad_norm": 0.006866455078125,
      "learning_rate": 3.25290773650208e-05,
      "loss": 0.0001,
      "step": 15762
    },
    {
      "epoch": 7.357292882147025,
      "grad_norm": 0.005645751953125,
      "learning_rate": 3.251824783387053e-05,
      "loss": 0.0001,
      "step": 15763
    },
    {
      "epoch": 7.357759626604434,
      "grad_norm": 0.0101318359375,
      "learning_rate": 3.250741975567726e-05,
      "loss": 0.0002,
      "step": 15764
    },
    {
      "epoch": 7.358226371061844,
      "grad_norm": 0.0634765625,
      "learning_rate": 3.249659313067417e-05,
      "loss": 0.0025,
      "step": 15765
    },
    {
      "epoch": 7.358693115519253,
      "grad_norm": 0.007598876953125,
      "learning_rate": 3.2485767959094284e-05,
      "loss": 0.0001,
      "step": 15766
    },
    {
      "epoch": 7.359159859976662,
      "grad_norm": 0.01470947265625,
      "learning_rate": 3.2474944241170714e-05,
      "loss": 0.0002,
      "step": 15767
    },
    {
      "epoch": 7.359626604434072,
      "grad_norm": 0.0091552734375,
      "learning_rate": 3.246412197713656e-05,
      "loss": 0.0002,
      "step": 15768
    },
    {
      "epoch": 7.360093348891482,
      "grad_norm": 0.007659912109375,
      "learning_rate": 3.245330116722476e-05,
      "loss": 0.0002,
      "step": 15769
    },
    {
      "epoch": 7.360560093348892,
      "grad_norm": 0.00927734375,
      "learning_rate": 3.244248181166833e-05,
      "loss": 0.0002,
      "step": 15770
    },
    {
      "epoch": 7.3610268378063015,
      "grad_norm": 0.0128173828125,
      "learning_rate": 3.243166391070024e-05,
      "loss": 0.0002,
      "step": 15771
    },
    {
      "epoch": 7.36149358226371,
      "grad_norm": 0.009765625,
      "learning_rate": 3.242084746455337e-05,
      "loss": 0.0001,
      "step": 15772
    },
    {
      "epoch": 7.36196032672112,
      "grad_norm": 0.05078125,
      "learning_rate": 3.241003247346064e-05,
      "loss": 0.0003,
      "step": 15773
    },
    {
      "epoch": 7.36242707117853,
      "grad_norm": 0.006805419921875,
      "learning_rate": 3.239921893765494e-05,
      "loss": 0.0002,
      "step": 15774
    },
    {
      "epoch": 7.362893815635939,
      "grad_norm": 0.09228515625,
      "learning_rate": 3.238840685736901e-05,
      "loss": 0.0003,
      "step": 15775
    },
    {
      "epoch": 7.363360560093349,
      "grad_norm": 0.007049560546875,
      "learning_rate": 3.237759623283572e-05,
      "loss": 0.0001,
      "step": 15776
    },
    {
      "epoch": 7.3638273045507585,
      "grad_norm": 0.1513671875,
      "learning_rate": 3.236678706428783e-05,
      "loss": 0.0029,
      "step": 15777
    },
    {
      "epoch": 7.364294049008168,
      "grad_norm": 0.01318359375,
      "learning_rate": 3.235597935195802e-05,
      "loss": 0.0002,
      "step": 15778
    },
    {
      "epoch": 7.364760793465577,
      "grad_norm": 0.048828125,
      "learning_rate": 3.234517309607903e-05,
      "loss": 0.0026,
      "step": 15779
    },
    {
      "epoch": 7.365227537922987,
      "grad_norm": 0.04296875,
      "learning_rate": 3.233436829688356e-05,
      "loss": 0.0017,
      "step": 15780
    },
    {
      "epoch": 7.365694282380397,
      "grad_norm": 0.08251953125,
      "learning_rate": 3.232356495460419e-05,
      "loss": 0.0025,
      "step": 15781
    },
    {
      "epoch": 7.366161026837807,
      "grad_norm": 0.04296875,
      "learning_rate": 3.231276306947359e-05,
      "loss": 0.0018,
      "step": 15782
    },
    {
      "epoch": 7.3666277712952155,
      "grad_norm": 0.05712890625,
      "learning_rate": 3.230196264172426e-05,
      "loss": 0.0021,
      "step": 15783
    },
    {
      "epoch": 7.367094515752625,
      "grad_norm": 0.00927734375,
      "learning_rate": 3.229116367158879e-05,
      "loss": 0.0001,
      "step": 15784
    },
    {
      "epoch": 7.367561260210035,
      "grad_norm": 0.0113525390625,
      "learning_rate": 3.2280366159299715e-05,
      "loss": 0.0002,
      "step": 15785
    },
    {
      "epoch": 7.368028004667445,
      "grad_norm": 0.01123046875,
      "learning_rate": 3.2269570105089465e-05,
      "loss": 0.0002,
      "step": 15786
    },
    {
      "epoch": 7.368494749124854,
      "grad_norm": 0.01055908203125,
      "learning_rate": 3.225877550919052e-05,
      "loss": 0.0002,
      "step": 15787
    },
    {
      "epoch": 7.368961493582264,
      "grad_norm": 0.07080078125,
      "learning_rate": 3.2247982371835325e-05,
      "loss": 0.0043,
      "step": 15788
    },
    {
      "epoch": 7.369428238039673,
      "grad_norm": 0.01336669921875,
      "learning_rate": 3.2237190693256226e-05,
      "loss": 0.0002,
      "step": 15789
    },
    {
      "epoch": 7.369894982497083,
      "grad_norm": 0.0242919921875,
      "learning_rate": 3.222640047368558e-05,
      "loss": 0.0002,
      "step": 15790
    },
    {
      "epoch": 7.370361726954492,
      "grad_norm": 0.037109375,
      "learning_rate": 3.2215611713355767e-05,
      "loss": 0.0002,
      "step": 15791
    },
    {
      "epoch": 7.370828471411902,
      "grad_norm": 0.0301513671875,
      "learning_rate": 3.2204824412499014e-05,
      "loss": 0.0003,
      "step": 15792
    },
    {
      "epoch": 7.371295215869312,
      "grad_norm": 0.008544921875,
      "learning_rate": 3.2194038571347604e-05,
      "loss": 0.0001,
      "step": 15793
    },
    {
      "epoch": 7.3717619603267215,
      "grad_norm": 0.0196533203125,
      "learning_rate": 3.2183254190133806e-05,
      "loss": 0.0002,
      "step": 15794
    },
    {
      "epoch": 7.37222870478413,
      "grad_norm": 0.01458740234375,
      "learning_rate": 3.217247126908977e-05,
      "loss": 0.0002,
      "step": 15795
    },
    {
      "epoch": 7.37269544924154,
      "grad_norm": 0.050537109375,
      "learning_rate": 3.216168980844767e-05,
      "loss": 0.002,
      "step": 15796
    },
    {
      "epoch": 7.37316219369895,
      "grad_norm": 0.0059814453125,
      "learning_rate": 3.215090980843969e-05,
      "loss": 0.0001,
      "step": 15797
    },
    {
      "epoch": 7.37362893815636,
      "grad_norm": 0.0400390625,
      "learning_rate": 3.214013126929787e-05,
      "loss": 0.002,
      "step": 15798
    },
    {
      "epoch": 7.374095682613769,
      "grad_norm": 0.007110595703125,
      "learning_rate": 3.2129354191254346e-05,
      "loss": 0.0001,
      "step": 15799
    },
    {
      "epoch": 7.3745624270711785,
      "grad_norm": 0.00946044921875,
      "learning_rate": 3.21185785745411e-05,
      "loss": 0.0001,
      "step": 15800
    },
    {
      "epoch": 7.375029171528588,
      "grad_norm": 0.1689453125,
      "learning_rate": 3.210780441939017e-05,
      "loss": 0.0023,
      "step": 15801
    },
    {
      "epoch": 7.375495915985998,
      "grad_norm": 0.0299072265625,
      "learning_rate": 3.2097031726033585e-05,
      "loss": 0.0002,
      "step": 15802
    },
    {
      "epoch": 7.375962660443407,
      "grad_norm": 0.078125,
      "learning_rate": 3.2086260494703205e-05,
      "loss": 0.0026,
      "step": 15803
    },
    {
      "epoch": 7.376429404900817,
      "grad_norm": 0.01446533203125,
      "learning_rate": 3.207549072563099e-05,
      "loss": 0.0002,
      "step": 15804
    },
    {
      "epoch": 7.376896149358227,
      "grad_norm": 0.01416015625,
      "learning_rate": 3.2064722419048854e-05,
      "loss": 0.0002,
      "step": 15805
    },
    {
      "epoch": 7.377362893815636,
      "grad_norm": 0.0196533203125,
      "learning_rate": 3.2053955575188624e-05,
      "loss": 0.0002,
      "step": 15806
    },
    {
      "epoch": 7.377829638273045,
      "grad_norm": 0.052978515625,
      "learning_rate": 3.204319019428208e-05,
      "loss": 0.0021,
      "step": 15807
    },
    {
      "epoch": 7.378296382730455,
      "grad_norm": 0.006927490234375,
      "learning_rate": 3.203242627656105e-05,
      "loss": 0.0001,
      "step": 15808
    },
    {
      "epoch": 7.378763127187865,
      "grad_norm": 0.007293701171875,
      "learning_rate": 3.202166382225732e-05,
      "loss": 0.0002,
      "step": 15809
    },
    {
      "epoch": 7.379229871645274,
      "grad_norm": 0.0302734375,
      "learning_rate": 3.2010902831602576e-05,
      "loss": 0.0021,
      "step": 15810
    },
    {
      "epoch": 7.379696616102684,
      "grad_norm": 0.0203857421875,
      "learning_rate": 3.2000143304828546e-05,
      "loss": 0.0002,
      "step": 15811
    },
    {
      "epoch": 7.380163360560093,
      "grad_norm": 0.006622314453125,
      "learning_rate": 3.198938524216685e-05,
      "loss": 0.0001,
      "step": 15812
    },
    {
      "epoch": 7.380630105017503,
      "grad_norm": 0.01153564453125,
      "learning_rate": 3.197862864384915e-05,
      "loss": 0.0002,
      "step": 15813
    },
    {
      "epoch": 7.381096849474912,
      "grad_norm": 0.004913330078125,
      "learning_rate": 3.196787351010708e-05,
      "loss": 0.0001,
      "step": 15814
    },
    {
      "epoch": 7.381563593932322,
      "grad_norm": 0.0299072265625,
      "learning_rate": 3.195711984117215e-05,
      "loss": 0.0022,
      "step": 15815
    },
    {
      "epoch": 7.382030338389732,
      "grad_norm": 0.01263427734375,
      "learning_rate": 3.194636763727591e-05,
      "loss": 0.0002,
      "step": 15816
    },
    {
      "epoch": 7.3824970828471415,
      "grad_norm": 0.0064697265625,
      "learning_rate": 3.1935616898649926e-05,
      "loss": 0.0001,
      "step": 15817
    },
    {
      "epoch": 7.38296382730455,
      "grad_norm": 0.05224609375,
      "learning_rate": 3.192486762552559e-05,
      "loss": 0.0023,
      "step": 15818
    },
    {
      "epoch": 7.38343057176196,
      "grad_norm": 0.02685546875,
      "learning_rate": 3.191411981813438e-05,
      "loss": 0.0002,
      "step": 15819
    },
    {
      "epoch": 7.38389731621937,
      "grad_norm": 0.032470703125,
      "learning_rate": 3.1903373476707754e-05,
      "loss": 0.0019,
      "step": 15820
    },
    {
      "epoch": 7.38436406067678,
      "grad_norm": 0.00848388671875,
      "learning_rate": 3.1892628601477015e-05,
      "loss": 0.0002,
      "step": 15821
    },
    {
      "epoch": 7.384830805134189,
      "grad_norm": 0.0184326171875,
      "learning_rate": 3.188188519267356e-05,
      "loss": 0.0003,
      "step": 15822
    },
    {
      "epoch": 7.3852975495915985,
      "grad_norm": 0.01031494140625,
      "learning_rate": 3.1871143250528715e-05,
      "loss": 0.0002,
      "step": 15823
    },
    {
      "epoch": 7.385764294049008,
      "grad_norm": 0.0084228515625,
      "learning_rate": 3.1860402775273715e-05,
      "loss": 0.0001,
      "step": 15824
    },
    {
      "epoch": 7.386231038506418,
      "grad_norm": 0.007659912109375,
      "learning_rate": 3.184966376713985e-05,
      "loss": 0.0002,
      "step": 15825
    },
    {
      "epoch": 7.386697782963827,
      "grad_norm": 0.0130615234375,
      "learning_rate": 3.1838926226358356e-05,
      "loss": 0.0002,
      "step": 15826
    },
    {
      "epoch": 7.387164527421237,
      "grad_norm": 0.00921630859375,
      "learning_rate": 3.182819015316039e-05,
      "loss": 0.0002,
      "step": 15827
    },
    {
      "epoch": 7.3876312718786465,
      "grad_norm": 0.06005859375,
      "learning_rate": 3.181745554777712e-05,
      "loss": 0.0051,
      "step": 15828
    },
    {
      "epoch": 7.388098016336056,
      "grad_norm": 0.01434326171875,
      "learning_rate": 3.180672241043972e-05,
      "loss": 0.0002,
      "step": 15829
    },
    {
      "epoch": 7.388564760793465,
      "grad_norm": 0.01153564453125,
      "learning_rate": 3.1795990741379225e-05,
      "loss": 0.0002,
      "step": 15830
    },
    {
      "epoch": 7.389031505250875,
      "grad_norm": 0.047119140625,
      "learning_rate": 3.1785260540826735e-05,
      "loss": 0.0003,
      "step": 15831
    },
    {
      "epoch": 7.389498249708285,
      "grad_norm": 0.0279541015625,
      "learning_rate": 3.1774531809013254e-05,
      "loss": 0.0002,
      "step": 15832
    },
    {
      "epoch": 7.389964994165695,
      "grad_norm": 0.06591796875,
      "learning_rate": 3.1763804546169785e-05,
      "loss": 0.0003,
      "step": 15833
    },
    {
      "epoch": 7.3904317386231035,
      "grad_norm": 0.017578125,
      "learning_rate": 3.1753078752527366e-05,
      "loss": 0.0002,
      "step": 15834
    },
    {
      "epoch": 7.390898483080513,
      "grad_norm": 0.023681640625,
      "learning_rate": 3.174235442831684e-05,
      "loss": 0.0002,
      "step": 15835
    },
    {
      "epoch": 7.391365227537923,
      "grad_norm": 0.01141357421875,
      "learning_rate": 3.173163157376915e-05,
      "loss": 0.0002,
      "step": 15836
    },
    {
      "epoch": 7.391831971995333,
      "grad_norm": 0.06640625,
      "learning_rate": 3.1720910189115214e-05,
      "loss": 0.0017,
      "step": 15837
    },
    {
      "epoch": 7.392298716452742,
      "grad_norm": 0.029052734375,
      "learning_rate": 3.171019027458581e-05,
      "loss": 0.0018,
      "step": 15838
    },
    {
      "epoch": 7.392765460910152,
      "grad_norm": 0.0101318359375,
      "learning_rate": 3.169947183041177e-05,
      "loss": 0.0002,
      "step": 15839
    },
    {
      "epoch": 7.393232205367561,
      "grad_norm": 0.01190185546875,
      "learning_rate": 3.168875485682393e-05,
      "loss": 0.0002,
      "step": 15840
    },
    {
      "epoch": 7.393698949824971,
      "grad_norm": 0.0458984375,
      "learning_rate": 3.1678039354052946e-05,
      "loss": 0.0016,
      "step": 15841
    },
    {
      "epoch": 7.39416569428238,
      "grad_norm": 0.0120849609375,
      "learning_rate": 3.1667325322329575e-05,
      "loss": 0.0002,
      "step": 15842
    },
    {
      "epoch": 7.39463243873979,
      "grad_norm": 0.01190185546875,
      "learning_rate": 3.165661276188455e-05,
      "loss": 0.0002,
      "step": 15843
    },
    {
      "epoch": 7.3950991831972,
      "grad_norm": 0.0107421875,
      "learning_rate": 3.164590167294844e-05,
      "loss": 0.0002,
      "step": 15844
    },
    {
      "epoch": 7.3955659276546095,
      "grad_norm": 0.00885009765625,
      "learning_rate": 3.1635192055751915e-05,
      "loss": 0.0001,
      "step": 15845
    },
    {
      "epoch": 7.396032672112018,
      "grad_norm": 0.0184326171875,
      "learning_rate": 3.162448391052559e-05,
      "loss": 0.0002,
      "step": 15846
    },
    {
      "epoch": 7.396499416569428,
      "grad_norm": 0.00994873046875,
      "learning_rate": 3.161377723749995e-05,
      "loss": 0.0001,
      "step": 15847
    },
    {
      "epoch": 7.396966161026838,
      "grad_norm": 0.08984375,
      "learning_rate": 3.160307203690557e-05,
      "loss": 0.0003,
      "step": 15848
    },
    {
      "epoch": 7.397432905484248,
      "grad_norm": 0.025390625,
      "learning_rate": 3.1592368308972965e-05,
      "loss": 0.0016,
      "step": 15849
    },
    {
      "epoch": 7.397899649941657,
      "grad_norm": 0.009765625,
      "learning_rate": 3.1581666053932536e-05,
      "loss": 0.0001,
      "step": 15850
    },
    {
      "epoch": 7.3983663943990665,
      "grad_norm": 0.05126953125,
      "learning_rate": 3.157096527201478e-05,
      "loss": 0.0026,
      "step": 15851
    },
    {
      "epoch": 7.398833138856476,
      "grad_norm": 0.0185546875,
      "learning_rate": 3.156026596345003e-05,
      "loss": 0.0002,
      "step": 15852
    },
    {
      "epoch": 7.399299883313885,
      "grad_norm": 0.04638671875,
      "learning_rate": 3.1549568128468734e-05,
      "loss": 0.002,
      "step": 15853
    },
    {
      "epoch": 7.399766627771295,
      "grad_norm": 0.04248046875,
      "learning_rate": 3.153887176730114e-05,
      "loss": 0.0032,
      "step": 15854
    },
    {
      "epoch": 7.400233372228705,
      "grad_norm": 0.06396484375,
      "learning_rate": 3.1528176880177626e-05,
      "loss": 0.0003,
      "step": 15855
    },
    {
      "epoch": 7.400700116686115,
      "grad_norm": 0.03955078125,
      "learning_rate": 3.151748346732841e-05,
      "loss": 0.0011,
      "step": 15856
    },
    {
      "epoch": 7.4011668611435235,
      "grad_norm": 0.0274658203125,
      "learning_rate": 3.150679152898376e-05,
      "loss": 0.0002,
      "step": 15857
    },
    {
      "epoch": 7.401633605600933,
      "grad_norm": 0.01141357421875,
      "learning_rate": 3.149610106537391e-05,
      "loss": 0.0001,
      "step": 15858
    },
    {
      "epoch": 7.402100350058343,
      "grad_norm": 0.0078125,
      "learning_rate": 3.148541207672897e-05,
      "loss": 0.0001,
      "step": 15859
    },
    {
      "epoch": 7.402567094515753,
      "grad_norm": 0.0306396484375,
      "learning_rate": 3.147472456327914e-05,
      "loss": 0.0022,
      "step": 15860
    },
    {
      "epoch": 7.403033838973162,
      "grad_norm": 0.00885009765625,
      "learning_rate": 3.146403852525455e-05,
      "loss": 0.0001,
      "step": 15861
    },
    {
      "epoch": 7.403500583430572,
      "grad_norm": 0.042236328125,
      "learning_rate": 3.145335396288521e-05,
      "loss": 0.0017,
      "step": 15862
    },
    {
      "epoch": 7.403967327887981,
      "grad_norm": 0.016845703125,
      "learning_rate": 3.144267087640126e-05,
      "loss": 0.0002,
      "step": 15863
    },
    {
      "epoch": 7.404434072345391,
      "grad_norm": 0.021484375,
      "learning_rate": 3.143198926603262e-05,
      "loss": 0.0002,
      "step": 15864
    },
    {
      "epoch": 7.4049008168028,
      "grad_norm": 0.01336669921875,
      "learning_rate": 3.142130913200935e-05,
      "loss": 0.0001,
      "step": 15865
    },
    {
      "epoch": 7.40536756126021,
      "grad_norm": 0.0284423828125,
      "learning_rate": 3.1410630474561406e-05,
      "loss": 0.0002,
      "step": 15866
    },
    {
      "epoch": 7.40583430571762,
      "grad_norm": 0.0084228515625,
      "learning_rate": 3.139995329391866e-05,
      "loss": 0.0002,
      "step": 15867
    },
    {
      "epoch": 7.4063010501750295,
      "grad_norm": 0.06103515625,
      "learning_rate": 3.138927759031103e-05,
      "loss": 0.0003,
      "step": 15868
    },
    {
      "epoch": 7.406767794632438,
      "grad_norm": 0.028564453125,
      "learning_rate": 3.1378603363968404e-05,
      "loss": 0.0008,
      "step": 15869
    },
    {
      "epoch": 7.407234539089848,
      "grad_norm": 0.04296875,
      "learning_rate": 3.1367930615120555e-05,
      "loss": 0.0003,
      "step": 15870
    },
    {
      "epoch": 7.407701283547258,
      "grad_norm": 0.006988525390625,
      "learning_rate": 3.135725934399731e-05,
      "loss": 0.0001,
      "step": 15871
    },
    {
      "epoch": 7.408168028004668,
      "grad_norm": 0.08984375,
      "learning_rate": 3.134658955082848e-05,
      "loss": 0.006,
      "step": 15872
    },
    {
      "epoch": 7.408634772462077,
      "grad_norm": 0.01409912109375,
      "learning_rate": 3.1335921235843703e-05,
      "loss": 0.0001,
      "step": 15873
    },
    {
      "epoch": 7.4091015169194865,
      "grad_norm": 0.703125,
      "learning_rate": 3.132525439927272e-05,
      "loss": 0.0016,
      "step": 15874
    },
    {
      "epoch": 7.409568261376896,
      "grad_norm": 0.0400390625,
      "learning_rate": 3.131458904134525e-05,
      "loss": 0.0024,
      "step": 15875
    },
    {
      "epoch": 7.410035005834306,
      "grad_norm": 0.042236328125,
      "learning_rate": 3.1303925162290834e-05,
      "loss": 0.0006,
      "step": 15876
    },
    {
      "epoch": 7.410501750291715,
      "grad_norm": 0.00732421875,
      "learning_rate": 3.129326276233915e-05,
      "loss": 0.0001,
      "step": 15877
    },
    {
      "epoch": 7.410968494749125,
      "grad_norm": 0.0341796875,
      "learning_rate": 3.1282601841719774e-05,
      "loss": 0.002,
      "step": 15878
    },
    {
      "epoch": 7.411435239206535,
      "grad_norm": 0.0128173828125,
      "learning_rate": 3.127194240066219e-05,
      "loss": 0.0002,
      "step": 15879
    },
    {
      "epoch": 7.411901983663944,
      "grad_norm": 0.01287841796875,
      "learning_rate": 3.126128443939598e-05,
      "loss": 0.0002,
      "step": 15880
    },
    {
      "epoch": 7.412368728121353,
      "grad_norm": 0.0091552734375,
      "learning_rate": 3.125062795815055e-05,
      "loss": 0.0001,
      "step": 15881
    },
    {
      "epoch": 7.412835472578763,
      "grad_norm": 0.05224609375,
      "learning_rate": 3.123997295715537e-05,
      "loss": 0.0002,
      "step": 15882
    },
    {
      "epoch": 7.413302217036173,
      "grad_norm": 0.039794921875,
      "learning_rate": 3.12293194366399e-05,
      "loss": 0.0027,
      "step": 15883
    },
    {
      "epoch": 7.413768961493583,
      "grad_norm": 0.00762939453125,
      "learning_rate": 3.1218667396833455e-05,
      "loss": 0.0002,
      "step": 15884
    },
    {
      "epoch": 7.414235705950992,
      "grad_norm": 0.006683349609375,
      "learning_rate": 3.120801683796541e-05,
      "loss": 0.0001,
      "step": 15885
    },
    {
      "epoch": 7.414702450408401,
      "grad_norm": 0.00946044921875,
      "learning_rate": 3.1197367760265125e-05,
      "loss": 0.0001,
      "step": 15886
    },
    {
      "epoch": 7.415169194865811,
      "grad_norm": 0.0230712890625,
      "learning_rate": 3.118672016396181e-05,
      "loss": 0.0002,
      "step": 15887
    },
    {
      "epoch": 7.41563593932322,
      "grad_norm": 0.0341796875,
      "learning_rate": 3.117607404928478e-05,
      "loss": 0.0002,
      "step": 15888
    },
    {
      "epoch": 7.41610268378063,
      "grad_norm": 0.11767578125,
      "learning_rate": 3.116542941646324e-05,
      "loss": 0.0059,
      "step": 15889
    },
    {
      "epoch": 7.41656942823804,
      "grad_norm": 0.025146484375,
      "learning_rate": 3.115478626572637e-05,
      "loss": 0.0002,
      "step": 15890
    },
    {
      "epoch": 7.4170361726954495,
      "grad_norm": 0.0137939453125,
      "learning_rate": 3.114414459730333e-05,
      "loss": 0.0002,
      "step": 15891
    },
    {
      "epoch": 7.417502917152859,
      "grad_norm": 0.00872802734375,
      "learning_rate": 3.1133504411423276e-05,
      "loss": 0.0002,
      "step": 15892
    },
    {
      "epoch": 7.417969661610268,
      "grad_norm": 0.0194091796875,
      "learning_rate": 3.112286570831526e-05,
      "loss": 0.0002,
      "step": 15893
    },
    {
      "epoch": 7.418436406067678,
      "grad_norm": 0.08056640625,
      "learning_rate": 3.1112228488208364e-05,
      "loss": 0.0003,
      "step": 15894
    },
    {
      "epoch": 7.418903150525088,
      "grad_norm": 0.0107421875,
      "learning_rate": 3.110159275133165e-05,
      "loss": 0.0002,
      "step": 15895
    },
    {
      "epoch": 7.419369894982497,
      "grad_norm": 0.0240478515625,
      "learning_rate": 3.109095849791406e-05,
      "loss": 0.0002,
      "step": 15896
    },
    {
      "epoch": 7.4198366394399065,
      "grad_norm": 0.0167236328125,
      "learning_rate": 3.108032572818458e-05,
      "loss": 0.0002,
      "step": 15897
    },
    {
      "epoch": 7.420303383897316,
      "grad_norm": 0.03369140625,
      "learning_rate": 3.10696944423722e-05,
      "loss": 0.0002,
      "step": 15898
    },
    {
      "epoch": 7.420770128354726,
      "grad_norm": 0.018310546875,
      "learning_rate": 3.105906464070574e-05,
      "loss": 0.0002,
      "step": 15899
    },
    {
      "epoch": 7.421236872812135,
      "grad_norm": 0.03125,
      "learning_rate": 3.1048436323414145e-05,
      "loss": 0.0028,
      "step": 15900
    },
    {
      "epoch": 7.421703617269545,
      "grad_norm": 0.0128173828125,
      "learning_rate": 3.103780949072621e-05,
      "loss": 0.0002,
      "step": 15901
    },
    {
      "epoch": 7.4221703617269545,
      "grad_norm": 0.0196533203125,
      "learning_rate": 3.1027184142870724e-05,
      "loss": 0.0002,
      "step": 15902
    },
    {
      "epoch": 7.422637106184364,
      "grad_norm": 0.0068359375,
      "learning_rate": 3.101656028007649e-05,
      "loss": 0.0001,
      "step": 15903
    },
    {
      "epoch": 7.423103850641773,
      "grad_norm": 0.02294921875,
      "learning_rate": 3.100593790257228e-05,
      "loss": 0.0002,
      "step": 15904
    },
    {
      "epoch": 7.423570595099183,
      "grad_norm": 0.01336669921875,
      "learning_rate": 3.099531701058676e-05,
      "loss": 0.0002,
      "step": 15905
    },
    {
      "epoch": 7.424037339556593,
      "grad_norm": 0.0103759765625,
      "learning_rate": 3.098469760434861e-05,
      "loss": 0.0002,
      "step": 15906
    },
    {
      "epoch": 7.424504084014003,
      "grad_norm": 0.026611328125,
      "learning_rate": 3.097407968408653e-05,
      "loss": 0.0002,
      "step": 15907
    },
    {
      "epoch": 7.4249708284714115,
      "grad_norm": 0.033203125,
      "learning_rate": 3.096346325002908e-05,
      "loss": 0.0002,
      "step": 15908
    },
    {
      "epoch": 7.425437572928821,
      "grad_norm": 0.04248046875,
      "learning_rate": 3.095284830240485e-05,
      "loss": 0.0025,
      "step": 15909
    },
    {
      "epoch": 7.425904317386231,
      "grad_norm": 0.01458740234375,
      "learning_rate": 3.094223484144244e-05,
      "loss": 0.0002,
      "step": 15910
    },
    {
      "epoch": 7.426371061843641,
      "grad_norm": 0.0206298828125,
      "learning_rate": 3.0931622867370304e-05,
      "loss": 0.0002,
      "step": 15911
    },
    {
      "epoch": 7.42683780630105,
      "grad_norm": 0.00872802734375,
      "learning_rate": 3.0921012380416995e-05,
      "loss": 0.0002,
      "step": 15912
    },
    {
      "epoch": 7.42730455075846,
      "grad_norm": 0.046875,
      "learning_rate": 3.0910403380810905e-05,
      "loss": 0.0025,
      "step": 15913
    },
    {
      "epoch": 7.427771295215869,
      "grad_norm": 0.029052734375,
      "learning_rate": 3.0899795868780476e-05,
      "loss": 0.0002,
      "step": 15914
    },
    {
      "epoch": 7.428238039673279,
      "grad_norm": 0.01202392578125,
      "learning_rate": 3.088918984455415e-05,
      "loss": 0.0002,
      "step": 15915
    },
    {
      "epoch": 7.428704784130688,
      "grad_norm": 0.00958251953125,
      "learning_rate": 3.087858530836021e-05,
      "loss": 0.0001,
      "step": 15916
    },
    {
      "epoch": 7.429171528588098,
      "grad_norm": 0.0172119140625,
      "learning_rate": 3.086798226042702e-05,
      "loss": 0.0002,
      "step": 15917
    },
    {
      "epoch": 7.429638273045508,
      "grad_norm": 0.030517578125,
      "learning_rate": 3.085738070098291e-05,
      "loss": 0.0002,
      "step": 15918
    },
    {
      "epoch": 7.4301050175029175,
      "grad_norm": 0.01318359375,
      "learning_rate": 3.084678063025607e-05,
      "loss": 0.0002,
      "step": 15919
    },
    {
      "epoch": 7.430571761960326,
      "grad_norm": 0.01171875,
      "learning_rate": 3.083618204847478e-05,
      "loss": 0.0002,
      "step": 15920
    },
    {
      "epoch": 7.431038506417736,
      "grad_norm": 0.02783203125,
      "learning_rate": 3.082558495586725e-05,
      "loss": 0.0003,
      "step": 15921
    },
    {
      "epoch": 7.431505250875146,
      "grad_norm": 0.00836181640625,
      "learning_rate": 3.0814989352661604e-05,
      "loss": 0.0001,
      "step": 15922
    },
    {
      "epoch": 7.431971995332556,
      "grad_norm": 0.01446533203125,
      "learning_rate": 3.080439523908599e-05,
      "loss": 0.0002,
      "step": 15923
    },
    {
      "epoch": 7.432438739789965,
      "grad_norm": 0.01043701171875,
      "learning_rate": 3.0793802615368564e-05,
      "loss": 0.0002,
      "step": 15924
    },
    {
      "epoch": 7.4329054842473745,
      "grad_norm": 0.0184326171875,
      "learning_rate": 3.0783211481737314e-05,
      "loss": 0.0001,
      "step": 15925
    },
    {
      "epoch": 7.433372228704784,
      "grad_norm": 0.01007080078125,
      "learning_rate": 3.0772621838420325e-05,
      "loss": 0.0002,
      "step": 15926
    },
    {
      "epoch": 7.433838973162194,
      "grad_norm": 0.09765625,
      "learning_rate": 3.076203368564562e-05,
      "loss": 0.0026,
      "step": 15927
    },
    {
      "epoch": 7.434305717619603,
      "grad_norm": 0.01055908203125,
      "learning_rate": 3.0751447023641124e-05,
      "loss": 0.0002,
      "step": 15928
    },
    {
      "epoch": 7.434772462077013,
      "grad_norm": 0.01007080078125,
      "learning_rate": 3.0740861852634826e-05,
      "loss": 0.0002,
      "step": 15929
    },
    {
      "epoch": 7.435239206534423,
      "grad_norm": 0.0155029296875,
      "learning_rate": 3.0730278172854596e-05,
      "loss": 0.0002,
      "step": 15930
    },
    {
      "epoch": 7.4357059509918315,
      "grad_norm": 0.0174560546875,
      "learning_rate": 3.071969598452834e-05,
      "loss": 0.0002,
      "step": 15931
    },
    {
      "epoch": 7.436172695449241,
      "grad_norm": 0.0234375,
      "learning_rate": 3.0709115287883915e-05,
      "loss": 0.002,
      "step": 15932
    },
    {
      "epoch": 7.436639439906651,
      "grad_norm": 0.06298828125,
      "learning_rate": 3.06985360831491e-05,
      "loss": 0.0003,
      "step": 15933
    },
    {
      "epoch": 7.437106184364061,
      "grad_norm": 0.0634765625,
      "learning_rate": 3.0687958370551686e-05,
      "loss": 0.0069,
      "step": 15934
    },
    {
      "epoch": 7.437572928821471,
      "grad_norm": 0.017333984375,
      "learning_rate": 3.067738215031946e-05,
      "loss": 0.0002,
      "step": 15935
    },
    {
      "epoch": 7.43803967327888,
      "grad_norm": 0.0126953125,
      "learning_rate": 3.0666807422680074e-05,
      "loss": 0.0002,
      "step": 15936
    },
    {
      "epoch": 7.438506417736289,
      "grad_norm": 0.032958984375,
      "learning_rate": 3.065623418786127e-05,
      "loss": 0.002,
      "step": 15937
    },
    {
      "epoch": 7.438973162193699,
      "grad_norm": 0.185546875,
      "learning_rate": 3.0645662446090694e-05,
      "loss": 0.0007,
      "step": 15938
    },
    {
      "epoch": 7.439439906651108,
      "grad_norm": 0.01092529296875,
      "learning_rate": 3.063509219759594e-05,
      "loss": 0.0002,
      "step": 15939
    },
    {
      "epoch": 7.439906651108518,
      "grad_norm": 0.05078125,
      "learning_rate": 3.06245234426046e-05,
      "loss": 0.0002,
      "step": 15940
    },
    {
      "epoch": 7.440373395565928,
      "grad_norm": 0.00958251953125,
      "learning_rate": 3.061395618134428e-05,
      "loss": 0.0002,
      "step": 15941
    },
    {
      "epoch": 7.4408401400233375,
      "grad_norm": 0.0140380859375,
      "learning_rate": 3.060339041404244e-05,
      "loss": 0.0002,
      "step": 15942
    },
    {
      "epoch": 7.441306884480746,
      "grad_norm": 0.0159912109375,
      "learning_rate": 3.05928261409266e-05,
      "loss": 0.0002,
      "step": 15943
    },
    {
      "epoch": 7.441773628938156,
      "grad_norm": 0.036376953125,
      "learning_rate": 3.0582263362224254e-05,
      "loss": 0.0003,
      "step": 15944
    },
    {
      "epoch": 7.442240373395566,
      "grad_norm": 0.0172119140625,
      "learning_rate": 3.057170207816277e-05,
      "loss": 0.0002,
      "step": 15945
    },
    {
      "epoch": 7.442707117852976,
      "grad_norm": 0.011962890625,
      "learning_rate": 3.0561142288969555e-05,
      "loss": 0.0002,
      "step": 15946
    },
    {
      "epoch": 7.443173862310385,
      "grad_norm": 0.04833984375,
      "learning_rate": 3.055058399487204e-05,
      "loss": 0.0026,
      "step": 15947
    },
    {
      "epoch": 7.4436406067677945,
      "grad_norm": 0.01422119140625,
      "learning_rate": 3.054002719609747e-05,
      "loss": 0.0001,
      "step": 15948
    },
    {
      "epoch": 7.444107351225204,
      "grad_norm": 0.03271484375,
      "learning_rate": 3.0529471892873205e-05,
      "loss": 0.0002,
      "step": 15949
    },
    {
      "epoch": 7.444574095682614,
      "grad_norm": 0.0186767578125,
      "learning_rate": 3.0518918085426504e-05,
      "loss": 0.0002,
      "step": 15950
    },
    {
      "epoch": 7.445040840140023,
      "grad_norm": 0.01214599609375,
      "learning_rate": 3.050836577398455e-05,
      "loss": 0.0002,
      "step": 15951
    },
    {
      "epoch": 7.445507584597433,
      "grad_norm": 0.0107421875,
      "learning_rate": 3.0497814958774584e-05,
      "loss": 0.0002,
      "step": 15952
    },
    {
      "epoch": 7.445974329054843,
      "grad_norm": 0.00933837890625,
      "learning_rate": 3.04872656400238e-05,
      "loss": 0.0002,
      "step": 15953
    },
    {
      "epoch": 7.446441073512252,
      "grad_norm": 0.006103515625,
      "learning_rate": 3.047671781795929e-05,
      "loss": 0.0002,
      "step": 15954
    },
    {
      "epoch": 7.446907817969661,
      "grad_norm": 0.015869140625,
      "learning_rate": 3.046617149280817e-05,
      "loss": 0.0002,
      "step": 15955
    },
    {
      "epoch": 7.447374562427071,
      "grad_norm": 0.0130615234375,
      "learning_rate": 3.0455626664797578e-05,
      "loss": 0.0001,
      "step": 15956
    },
    {
      "epoch": 7.447841306884481,
      "grad_norm": 0.0162353515625,
      "learning_rate": 3.0445083334154455e-05,
      "loss": 0.0002,
      "step": 15957
    },
    {
      "epoch": 7.448308051341891,
      "grad_norm": 0.0096435546875,
      "learning_rate": 3.0434541501105874e-05,
      "loss": 0.0001,
      "step": 15958
    },
    {
      "epoch": 7.4487747957993,
      "grad_norm": 0.0380859375,
      "learning_rate": 3.0424001165878824e-05,
      "loss": 0.0015,
      "step": 15959
    },
    {
      "epoch": 7.449241540256709,
      "grad_norm": 0.01318359375,
      "learning_rate": 3.0413462328700194e-05,
      "loss": 0.0002,
      "step": 15960
    },
    {
      "epoch": 7.449708284714119,
      "grad_norm": 0.0164794921875,
      "learning_rate": 3.0402924989796954e-05,
      "loss": 0.0002,
      "step": 15961
    },
    {
      "epoch": 7.450175029171529,
      "grad_norm": 0.0169677734375,
      "learning_rate": 3.0392389149395928e-05,
      "loss": 0.0002,
      "step": 15962
    },
    {
      "epoch": 7.450641773628938,
      "grad_norm": 0.0830078125,
      "learning_rate": 3.0381854807723997e-05,
      "loss": 0.0048,
      "step": 15963
    },
    {
      "epoch": 7.451108518086348,
      "grad_norm": 0.01434326171875,
      "learning_rate": 3.0371321965008014e-05,
      "loss": 0.0002,
      "step": 15964
    },
    {
      "epoch": 7.4515752625437575,
      "grad_norm": 0.05126953125,
      "learning_rate": 3.036079062147468e-05,
      "loss": 0.0033,
      "step": 15965
    },
    {
      "epoch": 7.452042007001167,
      "grad_norm": 0.005828857421875,
      "learning_rate": 3.0350260777350793e-05,
      "loss": 0.0001,
      "step": 15966
    },
    {
      "epoch": 7.452508751458576,
      "grad_norm": 0.0244140625,
      "learning_rate": 3.03397324328631e-05,
      "loss": 0.0002,
      "step": 15967
    },
    {
      "epoch": 7.452975495915986,
      "grad_norm": 0.011474609375,
      "learning_rate": 3.032920558823823e-05,
      "loss": 0.0002,
      "step": 15968
    },
    {
      "epoch": 7.453442240373396,
      "grad_norm": 0.0106201171875,
      "learning_rate": 3.0318680243702868e-05,
      "loss": 0.0002,
      "step": 15969
    },
    {
      "epoch": 7.4539089848308056,
      "grad_norm": 0.009521484375,
      "learning_rate": 3.0308156399483657e-05,
      "loss": 0.0002,
      "step": 15970
    },
    {
      "epoch": 7.4543757292882145,
      "grad_norm": 0.005767822265625,
      "learning_rate": 3.0297634055807134e-05,
      "loss": 0.0001,
      "step": 15971
    },
    {
      "epoch": 7.454842473745624,
      "grad_norm": 0.038818359375,
      "learning_rate": 3.0287113212899887e-05,
      "loss": 0.0002,
      "step": 15972
    },
    {
      "epoch": 7.455309218203034,
      "grad_norm": 0.025390625,
      "learning_rate": 3.0276593870988467e-05,
      "loss": 0.0025,
      "step": 15973
    },
    {
      "epoch": 7.455775962660443,
      "grad_norm": 0.02001953125,
      "learning_rate": 3.0266076030299316e-05,
      "loss": 0.0002,
      "step": 15974
    },
    {
      "epoch": 7.456242707117853,
      "grad_norm": 0.043212890625,
      "learning_rate": 3.025555969105891e-05,
      "loss": 0.0002,
      "step": 15975
    },
    {
      "epoch": 7.4567094515752625,
      "grad_norm": 0.1337890625,
      "learning_rate": 3.024504485349373e-05,
      "loss": 0.0003,
      "step": 15976
    },
    {
      "epoch": 7.457176196032672,
      "grad_norm": 0.006988525390625,
      "learning_rate": 3.023453151783009e-05,
      "loss": 0.0002,
      "step": 15977
    },
    {
      "epoch": 7.457642940490082,
      "grad_norm": 0.037353515625,
      "learning_rate": 3.0224019684294424e-05,
      "loss": 0.0002,
      "step": 15978
    },
    {
      "epoch": 7.458109684947491,
      "grad_norm": 0.0198974609375,
      "learning_rate": 3.0213509353113002e-05,
      "loss": 0.0002,
      "step": 15979
    },
    {
      "epoch": 7.458576429404901,
      "grad_norm": 0.0140380859375,
      "learning_rate": 3.020300052451216e-05,
      "loss": 0.0002,
      "step": 15980
    },
    {
      "epoch": 7.459043173862311,
      "grad_norm": 0.050537109375,
      "learning_rate": 3.0192493198718175e-05,
      "loss": 0.0002,
      "step": 15981
    },
    {
      "epoch": 7.4595099183197195,
      "grad_norm": 0.0194091796875,
      "learning_rate": 3.0181987375957245e-05,
      "loss": 0.0003,
      "step": 15982
    },
    {
      "epoch": 7.459976662777129,
      "grad_norm": 0.037353515625,
      "learning_rate": 3.0171483056455584e-05,
      "loss": 0.0003,
      "step": 15983
    },
    {
      "epoch": 7.460443407234539,
      "grad_norm": 0.08740234375,
      "learning_rate": 3.0160980240439406e-05,
      "loss": 0.0003,
      "step": 15984
    },
    {
      "epoch": 7.460910151691949,
      "grad_norm": 0.040771484375,
      "learning_rate": 3.015047892813477e-05,
      "loss": 0.0021,
      "step": 15985
    },
    {
      "epoch": 7.461376896149358,
      "grad_norm": 0.013671875,
      "learning_rate": 3.0139979119767835e-05,
      "loss": 0.0001,
      "step": 15986
    },
    {
      "epoch": 7.461843640606768,
      "grad_norm": 0.006103515625,
      "learning_rate": 3.0129480815564683e-05,
      "loss": 0.0001,
      "step": 15987
    },
    {
      "epoch": 7.462310385064177,
      "grad_norm": 0.01007080078125,
      "learning_rate": 3.0118984015751306e-05,
      "loss": 0.0001,
      "step": 15988
    },
    {
      "epoch": 7.462777129521587,
      "grad_norm": 0.01385498046875,
      "learning_rate": 3.010848872055373e-05,
      "loss": 0.0002,
      "step": 15989
    },
    {
      "epoch": 7.463243873978996,
      "grad_norm": 0.01177978515625,
      "learning_rate": 3.009799493019797e-05,
      "loss": 0.0002,
      "step": 15990
    },
    {
      "epoch": 7.463710618436406,
      "grad_norm": 0.01336669921875,
      "learning_rate": 3.0087502644909915e-05,
      "loss": 0.0002,
      "step": 15991
    },
    {
      "epoch": 7.464177362893816,
      "grad_norm": 0.051513671875,
      "learning_rate": 3.0077011864915483e-05,
      "loss": 0.0015,
      "step": 15992
    },
    {
      "epoch": 7.4646441073512255,
      "grad_norm": 0.043212890625,
      "learning_rate": 3.0066522590440604e-05,
      "loss": 0.0033,
      "step": 15993
    },
    {
      "epoch": 7.465110851808634,
      "grad_norm": 0.01239013671875,
      "learning_rate": 3.0056034821711065e-05,
      "loss": 0.0001,
      "step": 15994
    },
    {
      "epoch": 7.465577596266044,
      "grad_norm": 0.033935546875,
      "learning_rate": 3.0045548558952685e-05,
      "loss": 0.002,
      "step": 15995
    },
    {
      "epoch": 7.466044340723454,
      "grad_norm": 0.0057373046875,
      "learning_rate": 3.0035063802391338e-05,
      "loss": 0.0001,
      "step": 15996
    },
    {
      "epoch": 7.466511085180864,
      "grad_norm": 0.01611328125,
      "learning_rate": 3.0024580552252623e-05,
      "loss": 0.0002,
      "step": 15997
    },
    {
      "epoch": 7.466977829638273,
      "grad_norm": 0.04736328125,
      "learning_rate": 3.0014098808762324e-05,
      "loss": 0.0016,
      "step": 15998
    },
    {
      "epoch": 7.4674445740956825,
      "grad_norm": 0.009521484375,
      "learning_rate": 3.0003618572146163e-05,
      "loss": 0.0001,
      "step": 15999
    },
    {
      "epoch": 7.467911318553092,
      "grad_norm": 0.019287109375,
      "learning_rate": 2.9993139842629737e-05,
      "loss": 0.0002,
      "step": 16000
    },
    {
      "epoch": 7.468378063010502,
      "grad_norm": 0.051025390625,
      "learning_rate": 2.9982662620438674e-05,
      "loss": 0.002,
      "step": 16001
    },
    {
      "epoch": 7.468844807467911,
      "grad_norm": 0.04736328125,
      "learning_rate": 2.9972186905798595e-05,
      "loss": 0.0021,
      "step": 16002
    },
    {
      "epoch": 7.469311551925321,
      "grad_norm": 0.0172119140625,
      "learning_rate": 2.9961712698935007e-05,
      "loss": 0.0001,
      "step": 16003
    },
    {
      "epoch": 7.469778296382731,
      "grad_norm": 0.0419921875,
      "learning_rate": 2.9951240000073443e-05,
      "loss": 0.0002,
      "step": 16004
    },
    {
      "epoch": 7.47024504084014,
      "grad_norm": 0.06005859375,
      "learning_rate": 2.9940768809439445e-05,
      "loss": 0.0021,
      "step": 16005
    },
    {
      "epoch": 7.470711785297549,
      "grad_norm": 0.0458984375,
      "learning_rate": 2.99302991272584e-05,
      "loss": 0.0003,
      "step": 16006
    },
    {
      "epoch": 7.471178529754959,
      "grad_norm": 0.021484375,
      "learning_rate": 2.9919830953755744e-05,
      "loss": 0.0002,
      "step": 16007
    },
    {
      "epoch": 7.471645274212369,
      "grad_norm": 0.0272216796875,
      "learning_rate": 2.9909364289156916e-05,
      "loss": 0.0002,
      "step": 16008
    },
    {
      "epoch": 7.472112018669779,
      "grad_norm": 0.0206298828125,
      "learning_rate": 2.989889913368722e-05,
      "loss": 0.0002,
      "step": 16009
    },
    {
      "epoch": 7.472578763127188,
      "grad_norm": 0.046875,
      "learning_rate": 2.988843548757203e-05,
      "loss": 0.0002,
      "step": 16010
    },
    {
      "epoch": 7.473045507584597,
      "grad_norm": 0.01397705078125,
      "learning_rate": 2.987797335103658e-05,
      "loss": 0.0002,
      "step": 16011
    },
    {
      "epoch": 7.473512252042007,
      "grad_norm": 0.023681640625,
      "learning_rate": 2.986751272430617e-05,
      "loss": 0.0002,
      "step": 16012
    },
    {
      "epoch": 7.473978996499417,
      "grad_norm": 0.0157470703125,
      "learning_rate": 2.9857053607606055e-05,
      "loss": 0.0002,
      "step": 16013
    },
    {
      "epoch": 7.474445740956826,
      "grad_norm": 0.018798828125,
      "learning_rate": 2.9846596001161366e-05,
      "loss": 0.0002,
      "step": 16014
    },
    {
      "epoch": 7.474912485414236,
      "grad_norm": 0.0098876953125,
      "learning_rate": 2.98361399051973e-05,
      "loss": 0.0001,
      "step": 16015
    },
    {
      "epoch": 7.4753792298716455,
      "grad_norm": 0.09375,
      "learning_rate": 2.9825685319939035e-05,
      "loss": 0.0002,
      "step": 16016
    },
    {
      "epoch": 7.475845974329054,
      "grad_norm": 0.022705078125,
      "learning_rate": 2.981523224561158e-05,
      "loss": 0.0002,
      "step": 16017
    },
    {
      "epoch": 7.476312718786464,
      "grad_norm": 0.07861328125,
      "learning_rate": 2.9804780682440048e-05,
      "loss": 0.0003,
      "step": 16018
    },
    {
      "epoch": 7.476779463243874,
      "grad_norm": 0.029052734375,
      "learning_rate": 2.9794330630649503e-05,
      "loss": 0.0002,
      "step": 16019
    },
    {
      "epoch": 7.477246207701284,
      "grad_norm": 0.287109375,
      "learning_rate": 2.9783882090464886e-05,
      "loss": 0.0009,
      "step": 16020
    },
    {
      "epoch": 7.477712952158694,
      "grad_norm": 0.06689453125,
      "learning_rate": 2.97734350621112e-05,
      "loss": 0.0013,
      "step": 16021
    },
    {
      "epoch": 7.4781796966161025,
      "grad_norm": 0.02001953125,
      "learning_rate": 2.976298954581339e-05,
      "loss": 0.0002,
      "step": 16022
    },
    {
      "epoch": 7.478646441073512,
      "grad_norm": 0.0267333984375,
      "learning_rate": 2.9752545541796327e-05,
      "loss": 0.0001,
      "step": 16023
    },
    {
      "epoch": 7.479113185530922,
      "grad_norm": 0.05810546875,
      "learning_rate": 2.9742103050284886e-05,
      "loss": 0.0024,
      "step": 16024
    },
    {
      "epoch": 7.479579929988331,
      "grad_norm": 0.0157470703125,
      "learning_rate": 2.973166207150396e-05,
      "loss": 0.0001,
      "step": 16025
    },
    {
      "epoch": 7.480046674445741,
      "grad_norm": 0.0078125,
      "learning_rate": 2.9721222605678278e-05,
      "loss": 0.0001,
      "step": 16026
    },
    {
      "epoch": 7.480513418903151,
      "grad_norm": 0.0625,
      "learning_rate": 2.9710784653032654e-05,
      "loss": 0.0002,
      "step": 16027
    },
    {
      "epoch": 7.48098016336056,
      "grad_norm": 0.0205078125,
      "learning_rate": 2.970034821379185e-05,
      "loss": 0.0002,
      "step": 16028
    },
    {
      "epoch": 7.481446907817969,
      "grad_norm": 0.07373046875,
      "learning_rate": 2.9689913288180515e-05,
      "loss": 0.0003,
      "step": 16029
    },
    {
      "epoch": 7.481913652275379,
      "grad_norm": 0.01397705078125,
      "learning_rate": 2.96794798764234e-05,
      "loss": 0.0002,
      "step": 16030
    },
    {
      "epoch": 7.482380396732789,
      "grad_norm": 0.0146484375,
      "learning_rate": 2.9669047978745058e-05,
      "loss": 0.0002,
      "step": 16031
    },
    {
      "epoch": 7.482847141190199,
      "grad_norm": 0.01202392578125,
      "learning_rate": 2.9658617595370162e-05,
      "loss": 0.0001,
      "step": 16032
    },
    {
      "epoch": 7.483313885647608,
      "grad_norm": 0.00958251953125,
      "learning_rate": 2.9648188726523295e-05,
      "loss": 0.0002,
      "step": 16033
    },
    {
      "epoch": 7.483780630105017,
      "grad_norm": 0.009521484375,
      "learning_rate": 2.963776137242895e-05,
      "loss": 0.0001,
      "step": 16034
    },
    {
      "epoch": 7.484247374562427,
      "grad_norm": 0.008544921875,
      "learning_rate": 2.9627335533311672e-05,
      "loss": 0.0002,
      "step": 16035
    },
    {
      "epoch": 7.484714119019837,
      "grad_norm": 0.0091552734375,
      "learning_rate": 2.961691120939598e-05,
      "loss": 0.0002,
      "step": 16036
    },
    {
      "epoch": 7.485180863477246,
      "grad_norm": 0.01361083984375,
      "learning_rate": 2.960648840090624e-05,
      "loss": 0.0002,
      "step": 16037
    },
    {
      "epoch": 7.485647607934656,
      "grad_norm": 0.0634765625,
      "learning_rate": 2.9596067108066905e-05,
      "loss": 0.0003,
      "step": 16038
    },
    {
      "epoch": 7.4861143523920655,
      "grad_norm": 0.007110595703125,
      "learning_rate": 2.9585647331102406e-05,
      "loss": 0.0001,
      "step": 16039
    },
    {
      "epoch": 7.486581096849475,
      "grad_norm": 0.25390625,
      "learning_rate": 2.9575229070237e-05,
      "loss": 0.0011,
      "step": 16040
    },
    {
      "epoch": 7.487047841306884,
      "grad_norm": 0.1650390625,
      "learning_rate": 2.9564812325695056e-05,
      "loss": 0.0003,
      "step": 16041
    },
    {
      "epoch": 7.487514585764294,
      "grad_norm": 0.01214599609375,
      "learning_rate": 2.955439709770088e-05,
      "loss": 0.0002,
      "step": 16042
    },
    {
      "epoch": 7.487981330221704,
      "grad_norm": 0.0179443359375,
      "learning_rate": 2.9543983386478668e-05,
      "loss": 0.0002,
      "step": 16043
    },
    {
      "epoch": 7.4884480746791136,
      "grad_norm": 0.0277099609375,
      "learning_rate": 2.9533571192252696e-05,
      "loss": 0.0013,
      "step": 16044
    },
    {
      "epoch": 7.4889148191365225,
      "grad_norm": 0.048828125,
      "learning_rate": 2.9523160515247116e-05,
      "loss": 0.0003,
      "step": 16045
    },
    {
      "epoch": 7.489381563593932,
      "grad_norm": 0.00848388671875,
      "learning_rate": 2.9512751355686064e-05,
      "loss": 0.0002,
      "step": 16046
    },
    {
      "epoch": 7.489848308051342,
      "grad_norm": 0.03515625,
      "learning_rate": 2.9502343713793667e-05,
      "loss": 0.002,
      "step": 16047
    },
    {
      "epoch": 7.490315052508752,
      "grad_norm": 0.0556640625,
      "learning_rate": 2.9491937589794072e-05,
      "loss": 0.004,
      "step": 16048
    },
    {
      "epoch": 7.490781796966161,
      "grad_norm": 0.01544189453125,
      "learning_rate": 2.9481532983911252e-05,
      "loss": 0.0001,
      "step": 16049
    },
    {
      "epoch": 7.4912485414235706,
      "grad_norm": 0.048828125,
      "learning_rate": 2.9471129896369275e-05,
      "loss": 0.0002,
      "step": 16050
    },
    {
      "epoch": 7.49171528588098,
      "grad_norm": 0.0228271484375,
      "learning_rate": 2.946072832739215e-05,
      "loss": 0.0002,
      "step": 16051
    },
    {
      "epoch": 7.49218203033839,
      "grad_norm": 0.0101318359375,
      "learning_rate": 2.945032827720379e-05,
      "loss": 0.0002,
      "step": 16052
    },
    {
      "epoch": 7.492648774795799,
      "grad_norm": 0.007354736328125,
      "learning_rate": 2.943992974602813e-05,
      "loss": 0.0001,
      "step": 16053
    },
    {
      "epoch": 7.493115519253209,
      "grad_norm": 0.0098876953125,
      "learning_rate": 2.9429532734089094e-05,
      "loss": 0.0002,
      "step": 16054
    },
    {
      "epoch": 7.493582263710619,
      "grad_norm": 0.01251220703125,
      "learning_rate": 2.9419137241610496e-05,
      "loss": 0.0002,
      "step": 16055
    },
    {
      "epoch": 7.494049008168028,
      "grad_norm": 0.08447265625,
      "learning_rate": 2.9408743268816175e-05,
      "loss": 0.002,
      "step": 16056
    },
    {
      "epoch": 7.494515752625437,
      "grad_norm": 0.00592041015625,
      "learning_rate": 2.9398350815929975e-05,
      "loss": 0.0001,
      "step": 16057
    },
    {
      "epoch": 7.494982497082847,
      "grad_norm": 0.037109375,
      "learning_rate": 2.9387959883175574e-05,
      "loss": 0.0003,
      "step": 16058
    },
    {
      "epoch": 7.495449241540257,
      "grad_norm": 0.02197265625,
      "learning_rate": 2.937757047077678e-05,
      "loss": 0.0002,
      "step": 16059
    },
    {
      "epoch": 7.495915985997666,
      "grad_norm": 0.019775390625,
      "learning_rate": 2.936718257895722e-05,
      "loss": 0.0002,
      "step": 16060
    },
    {
      "epoch": 7.496382730455076,
      "grad_norm": 0.016357421875,
      "learning_rate": 2.9356796207940597e-05,
      "loss": 0.0001,
      "step": 16061
    },
    {
      "epoch": 7.496849474912485,
      "grad_norm": 0.050537109375,
      "learning_rate": 2.9346411357950544e-05,
      "loss": 0.0021,
      "step": 16062
    },
    {
      "epoch": 7.497316219369895,
      "grad_norm": 0.051513671875,
      "learning_rate": 2.9336028029210627e-05,
      "loss": 0.0002,
      "step": 16063
    },
    {
      "epoch": 7.497782963827304,
      "grad_norm": 0.078125,
      "learning_rate": 2.9325646221944426e-05,
      "loss": 0.0033,
      "step": 16064
    },
    {
      "epoch": 7.498249708284714,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.9315265936375504e-05,
      "loss": 0.0001,
      "step": 16065
    },
    {
      "epoch": 7.498716452742124,
      "grad_norm": 0.0162353515625,
      "learning_rate": 2.9304887172727313e-05,
      "loss": 0.0002,
      "step": 16066
    },
    {
      "epoch": 7.4991831971995335,
      "grad_norm": 0.04052734375,
      "learning_rate": 2.9294509931223323e-05,
      "loss": 0.0017,
      "step": 16067
    },
    {
      "epoch": 7.499649941656942,
      "grad_norm": 0.0093994140625,
      "learning_rate": 2.928413421208702e-05,
      "loss": 0.0002,
      "step": 16068
    },
    {
      "epoch": 7.500116686114352,
      "grad_norm": 0.01104736328125,
      "learning_rate": 2.927376001554174e-05,
      "loss": 0.0001,
      "step": 16069
    },
    {
      "epoch": 7.500583430571762,
      "grad_norm": 0.0098876953125,
      "learning_rate": 2.926338734181088e-05,
      "loss": 0.0002,
      "step": 16070
    },
    {
      "epoch": 7.501050175029172,
      "grad_norm": 0.2578125,
      "learning_rate": 2.92530161911178e-05,
      "loss": 0.0008,
      "step": 16071
    },
    {
      "epoch": 7.501516919486581,
      "grad_norm": 0.046142578125,
      "learning_rate": 2.924264656368575e-05,
      "loss": 0.0002,
      "step": 16072
    },
    {
      "epoch": 7.5019836639439905,
      "grad_norm": 0.032470703125,
      "learning_rate": 2.9232278459738026e-05,
      "loss": 0.0012,
      "step": 16073
    },
    {
      "epoch": 7.5024504084014,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.92219118794979e-05,
      "loss": 0.0001,
      "step": 16074
    },
    {
      "epoch": 7.50291715285881,
      "grad_norm": 0.12158203125,
      "learning_rate": 2.9211546823188494e-05,
      "loss": 0.0027,
      "step": 16075
    },
    {
      "epoch": 7.503383897316219,
      "grad_norm": 0.00885009765625,
      "learning_rate": 2.920118329103304e-05,
      "loss": 0.0002,
      "step": 16076
    },
    {
      "epoch": 7.503850641773629,
      "grad_norm": 0.00970458984375,
      "learning_rate": 2.9190821283254688e-05,
      "loss": 0.0001,
      "step": 16077
    },
    {
      "epoch": 7.504317386231039,
      "grad_norm": 0.018310546875,
      "learning_rate": 2.9180460800076493e-05,
      "loss": 0.0002,
      "step": 16078
    },
    {
      "epoch": 7.504784130688448,
      "grad_norm": 0.0159912109375,
      "learning_rate": 2.9170101841721576e-05,
      "loss": 0.0002,
      "step": 16079
    },
    {
      "epoch": 7.505250875145857,
      "grad_norm": 0.00579833984375,
      "learning_rate": 2.9159744408412926e-05,
      "loss": 0.0001,
      "step": 16080
    },
    {
      "epoch": 7.505250875145857,
      "eval_loss": 2.4192376136779785,
      "eval_runtime": 83.0787,
      "eval_samples_per_second": 21.714,
      "eval_steps_per_second": 2.72,
      "step": 16080
    },
    {
      "epoch": 7.505717619603267,
      "grad_norm": 0.050537109375,
      "learning_rate": 2.9149388500373587e-05,
      "loss": 0.0019,
      "step": 16081
    },
    {
      "epoch": 7.506184364060677,
      "grad_norm": 0.0111083984375,
      "learning_rate": 2.9139034117826535e-05,
      "loss": 0.0002,
      "step": 16082
    },
    {
      "epoch": 7.506651108518087,
      "grad_norm": 0.043701171875,
      "learning_rate": 2.9128681260994682e-05,
      "loss": 0.0034,
      "step": 16083
    },
    {
      "epoch": 7.507117852975496,
      "grad_norm": 0.00848388671875,
      "learning_rate": 2.911832993010095e-05,
      "loss": 0.0002,
      "step": 16084
    },
    {
      "epoch": 7.507584597432905,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.910798012536826e-05,
      "loss": 0.0002,
      "step": 16085
    },
    {
      "epoch": 7.508051341890315,
      "grad_norm": 0.01904296875,
      "learning_rate": 2.909763184701937e-05,
      "loss": 0.0002,
      "step": 16086
    },
    {
      "epoch": 7.508518086347725,
      "grad_norm": 0.005889892578125,
      "learning_rate": 2.908728509527714e-05,
      "loss": 0.0002,
      "step": 16087
    },
    {
      "epoch": 7.508984830805134,
      "grad_norm": 0.035888671875,
      "learning_rate": 2.907693987036437e-05,
      "loss": 0.0024,
      "step": 16088
    },
    {
      "epoch": 7.509451575262544,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.9066596172503745e-05,
      "loss": 0.0001,
      "step": 16089
    },
    {
      "epoch": 7.5099183197199535,
      "grad_norm": 0.01531982421875,
      "learning_rate": 2.9056254001918004e-05,
      "loss": 0.0002,
      "step": 16090
    },
    {
      "epoch": 7.510385064177363,
      "grad_norm": 0.00848388671875,
      "learning_rate": 2.9045913358829867e-05,
      "loss": 0.0002,
      "step": 16091
    },
    {
      "epoch": 7.510851808634772,
      "grad_norm": 0.01239013671875,
      "learning_rate": 2.9035574243461927e-05,
      "loss": 0.0001,
      "step": 16092
    },
    {
      "epoch": 7.511318553092182,
      "grad_norm": 0.01385498046875,
      "learning_rate": 2.9025236656036792e-05,
      "loss": 0.0002,
      "step": 16093
    },
    {
      "epoch": 7.511785297549592,
      "grad_norm": 0.0103759765625,
      "learning_rate": 2.9014900596777082e-05,
      "loss": 0.0001,
      "step": 16094
    },
    {
      "epoch": 7.512252042007001,
      "grad_norm": 0.01171875,
      "learning_rate": 2.9004566065905293e-05,
      "loss": 0.0002,
      "step": 16095
    },
    {
      "epoch": 7.5127187864644105,
      "grad_norm": 0.00701904296875,
      "learning_rate": 2.8994233063643973e-05,
      "loss": 0.0001,
      "step": 16096
    },
    {
      "epoch": 7.51318553092182,
      "grad_norm": 0.0233154296875,
      "learning_rate": 2.8983901590215622e-05,
      "loss": 0.0002,
      "step": 16097
    },
    {
      "epoch": 7.51365227537923,
      "grad_norm": 0.0142822265625,
      "learning_rate": 2.8973571645842633e-05,
      "loss": 0.0002,
      "step": 16098
    },
    {
      "epoch": 7.51411901983664,
      "grad_norm": 0.0223388671875,
      "learning_rate": 2.896324323074745e-05,
      "loss": 0.0002,
      "step": 16099
    },
    {
      "epoch": 7.514585764294049,
      "grad_norm": 0.09716796875,
      "learning_rate": 2.895291634515249e-05,
      "loss": 0.0022,
      "step": 16100
    },
    {
      "epoch": 7.515052508751459,
      "grad_norm": 0.01708984375,
      "learning_rate": 2.8942590989280037e-05,
      "loss": 0.0002,
      "step": 16101
    },
    {
      "epoch": 7.515519253208868,
      "grad_norm": 0.0218505859375,
      "learning_rate": 2.8932267163352446e-05,
      "loss": 0.0002,
      "step": 16102
    },
    {
      "epoch": 7.515985997666277,
      "grad_norm": 0.00848388671875,
      "learning_rate": 2.892194486759201e-05,
      "loss": 0.0002,
      "step": 16103
    },
    {
      "epoch": 7.516452742123687,
      "grad_norm": 0.01104736328125,
      "learning_rate": 2.8911624102220948e-05,
      "loss": 0.0002,
      "step": 16104
    },
    {
      "epoch": 7.516919486581097,
      "grad_norm": 0.0084228515625,
      "learning_rate": 2.8901304867461486e-05,
      "loss": 0.0002,
      "step": 16105
    },
    {
      "epoch": 7.517386231038507,
      "grad_norm": 0.00811767578125,
      "learning_rate": 2.8890987163535844e-05,
      "loss": 0.0001,
      "step": 16106
    },
    {
      "epoch": 7.5178529754959165,
      "grad_norm": 0.0125732421875,
      "learning_rate": 2.8880670990666124e-05,
      "loss": 0.0002,
      "step": 16107
    },
    {
      "epoch": 7.518319719953325,
      "grad_norm": 0.0233154296875,
      "learning_rate": 2.887035634907449e-05,
      "loss": 0.0002,
      "step": 16108
    },
    {
      "epoch": 7.518786464410735,
      "grad_norm": 0.0260009765625,
      "learning_rate": 2.886004323898297e-05,
      "loss": 0.0002,
      "step": 16109
    },
    {
      "epoch": 7.519253208868145,
      "grad_norm": 0.02392578125,
      "learning_rate": 2.8849731660613654e-05,
      "loss": 0.0002,
      "step": 16110
    },
    {
      "epoch": 7.519719953325554,
      "grad_norm": 0.01495361328125,
      "learning_rate": 2.883942161418859e-05,
      "loss": 0.0002,
      "step": 16111
    },
    {
      "epoch": 7.520186697782964,
      "grad_norm": 0.00665283203125,
      "learning_rate": 2.8829113099929705e-05,
      "loss": 0.0002,
      "step": 16112
    },
    {
      "epoch": 7.5206534422403735,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.8818806118058973e-05,
      "loss": 0.0001,
      "step": 16113
    },
    {
      "epoch": 7.521120186697783,
      "grad_norm": 0.026611328125,
      "learning_rate": 2.8808500668798354e-05,
      "loss": 0.0002,
      "step": 16114
    },
    {
      "epoch": 7.521586931155192,
      "grad_norm": 0.050537109375,
      "learning_rate": 2.8798196752369677e-05,
      "loss": 0.0009,
      "step": 16115
    },
    {
      "epoch": 7.522053675612602,
      "grad_norm": 0.05810546875,
      "learning_rate": 2.8787894368994816e-05,
      "loss": 0.0018,
      "step": 16116
    },
    {
      "epoch": 7.522520420070012,
      "grad_norm": 0.00823974609375,
      "learning_rate": 2.877759351889564e-05,
      "loss": 0.0001,
      "step": 16117
    },
    {
      "epoch": 7.5229871645274216,
      "grad_norm": 0.1669921875,
      "learning_rate": 2.876729420229386e-05,
      "loss": 0.0006,
      "step": 16118
    },
    {
      "epoch": 7.5234539089848305,
      "grad_norm": 0.032470703125,
      "learning_rate": 2.8756996419411276e-05,
      "loss": 0.0028,
      "step": 16119
    },
    {
      "epoch": 7.52392065344224,
      "grad_norm": 0.10888671875,
      "learning_rate": 2.8746700170469633e-05,
      "loss": 0.0017,
      "step": 16120
    },
    {
      "epoch": 7.52438739789965,
      "grad_norm": 0.01373291015625,
      "learning_rate": 2.873640545569056e-05,
      "loss": 0.0002,
      "step": 16121
    },
    {
      "epoch": 7.52485414235706,
      "grad_norm": 0.01287841796875,
      "learning_rate": 2.8726112275295746e-05,
      "loss": 0.0002,
      "step": 16122
    },
    {
      "epoch": 7.525320886814469,
      "grad_norm": 0.05419921875,
      "learning_rate": 2.871582062950685e-05,
      "loss": 0.0003,
      "step": 16123
    },
    {
      "epoch": 7.5257876312718786,
      "grad_norm": 0.0576171875,
      "learning_rate": 2.87055305185454e-05,
      "loss": 0.0002,
      "step": 16124
    },
    {
      "epoch": 7.526254375729288,
      "grad_norm": 0.007568359375,
      "learning_rate": 2.8695241942632967e-05,
      "loss": 0.0001,
      "step": 16125
    },
    {
      "epoch": 7.526721120186698,
      "grad_norm": 0.00830078125,
      "learning_rate": 2.8684954901991123e-05,
      "loss": 0.0001,
      "step": 16126
    },
    {
      "epoch": 7.527187864644107,
      "grad_norm": 0.01708984375,
      "learning_rate": 2.8674669396841302e-05,
      "loss": 0.0002,
      "step": 16127
    },
    {
      "epoch": 7.527654609101517,
      "grad_norm": 0.07275390625,
      "learning_rate": 2.8664385427405017e-05,
      "loss": 0.0002,
      "step": 16128
    },
    {
      "epoch": 7.528121353558927,
      "grad_norm": 0.047119140625,
      "learning_rate": 2.865410299390362e-05,
      "loss": 0.0022,
      "step": 16129
    },
    {
      "epoch": 7.528588098016336,
      "grad_norm": 0.006591796875,
      "learning_rate": 2.864382209655856e-05,
      "loss": 0.0002,
      "step": 16130
    },
    {
      "epoch": 7.529054842473745,
      "grad_norm": 0.0419921875,
      "learning_rate": 2.8633542735591202e-05,
      "loss": 0.002,
      "step": 16131
    },
    {
      "epoch": 7.529521586931155,
      "grad_norm": 0.06494140625,
      "learning_rate": 2.8623264911222814e-05,
      "loss": 0.0023,
      "step": 16132
    },
    {
      "epoch": 7.529988331388565,
      "grad_norm": 0.01165771484375,
      "learning_rate": 2.8612988623674742e-05,
      "loss": 0.0002,
      "step": 16133
    },
    {
      "epoch": 7.530455075845975,
      "grad_norm": 0.019775390625,
      "learning_rate": 2.8602713873168253e-05,
      "loss": 0.0002,
      "step": 16134
    },
    {
      "epoch": 7.530921820303384,
      "grad_norm": 0.01190185546875,
      "learning_rate": 2.8592440659924525e-05,
      "loss": 0.0002,
      "step": 16135
    },
    {
      "epoch": 7.531388564760793,
      "grad_norm": 0.025146484375,
      "learning_rate": 2.858216898416477e-05,
      "loss": 0.0002,
      "step": 16136
    },
    {
      "epoch": 7.531855309218203,
      "grad_norm": 0.03173828125,
      "learning_rate": 2.857189884611019e-05,
      "loss": 0.0023,
      "step": 16137
    },
    {
      "epoch": 7.532322053675612,
      "grad_norm": 0.0186767578125,
      "learning_rate": 2.856163024598185e-05,
      "loss": 0.0002,
      "step": 16138
    },
    {
      "epoch": 7.532788798133022,
      "grad_norm": 0.0089111328125,
      "learning_rate": 2.8551363184000867e-05,
      "loss": 0.0002,
      "step": 16139
    },
    {
      "epoch": 7.533255542590432,
      "grad_norm": 0.007049560546875,
      "learning_rate": 2.8541097660388372e-05,
      "loss": 0.0002,
      "step": 16140
    },
    {
      "epoch": 7.5337222870478415,
      "grad_norm": 0.0140380859375,
      "learning_rate": 2.8530833675365276e-05,
      "loss": 0.0002,
      "step": 16141
    },
    {
      "epoch": 7.534189031505251,
      "grad_norm": 0.01129150390625,
      "learning_rate": 2.8520571229152614e-05,
      "loss": 0.0002,
      "step": 16142
    },
    {
      "epoch": 7.53465577596266,
      "grad_norm": 0.041015625,
      "learning_rate": 2.85103103219714e-05,
      "loss": 0.0027,
      "step": 16143
    },
    {
      "epoch": 7.53512252042007,
      "grad_norm": 0.007537841796875,
      "learning_rate": 2.85000509540425e-05,
      "loss": 0.0001,
      "step": 16144
    },
    {
      "epoch": 7.53558926487748,
      "grad_norm": 0.037109375,
      "learning_rate": 2.8489793125586828e-05,
      "loss": 0.0002,
      "step": 16145
    },
    {
      "epoch": 7.536056009334889,
      "grad_norm": 0.017578125,
      "learning_rate": 2.847953683682528e-05,
      "loss": 0.0001,
      "step": 16146
    },
    {
      "epoch": 7.5365227537922985,
      "grad_norm": 0.00872802734375,
      "learning_rate": 2.846928208797862e-05,
      "loss": 0.0001,
      "step": 16147
    },
    {
      "epoch": 7.536989498249708,
      "grad_norm": 0.0220947265625,
      "learning_rate": 2.845902887926769e-05,
      "loss": 0.0002,
      "step": 16148
    },
    {
      "epoch": 7.537456242707118,
      "grad_norm": 0.0230712890625,
      "learning_rate": 2.844877721091328e-05,
      "loss": 0.0004,
      "step": 16149
    },
    {
      "epoch": 7.537922987164528,
      "grad_norm": 0.00830078125,
      "learning_rate": 2.8438527083136046e-05,
      "loss": 0.0002,
      "step": 16150
    },
    {
      "epoch": 7.538389731621937,
      "grad_norm": 0.037841796875,
      "learning_rate": 2.8428278496156725e-05,
      "loss": 0.0021,
      "step": 16151
    },
    {
      "epoch": 7.538856476079347,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.8418031450196002e-05,
      "loss": 0.0002,
      "step": 16152
    },
    {
      "epoch": 7.539323220536756,
      "grad_norm": 0.008056640625,
      "learning_rate": 2.8407785945474464e-05,
      "loss": 0.0002,
      "step": 16153
    },
    {
      "epoch": 7.539789964994165,
      "grad_norm": 0.0966796875,
      "learning_rate": 2.8397541982212717e-05,
      "loss": 0.0049,
      "step": 16154
    },
    {
      "epoch": 7.540256709451575,
      "grad_norm": 0.01324462890625,
      "learning_rate": 2.8387299560631374e-05,
      "loss": 0.0002,
      "step": 16155
    },
    {
      "epoch": 7.540723453908985,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.8377058680950896e-05,
      "loss": 0.0001,
      "step": 16156
    },
    {
      "epoch": 7.541190198366395,
      "grad_norm": 0.08349609375,
      "learning_rate": 2.8366819343391805e-05,
      "loss": 0.0003,
      "step": 16157
    },
    {
      "epoch": 7.541656942823804,
      "grad_norm": 0.05224609375,
      "learning_rate": 2.835658154817461e-05,
      "loss": 0.0029,
      "step": 16158
    },
    {
      "epoch": 7.542123687281213,
      "grad_norm": 0.036865234375,
      "learning_rate": 2.834634529551967e-05,
      "loss": 0.0002,
      "step": 16159
    },
    {
      "epoch": 7.542590431738623,
      "grad_norm": 0.01953125,
      "learning_rate": 2.8336110585647445e-05,
      "loss": 0.0002,
      "step": 16160
    },
    {
      "epoch": 7.543057176196033,
      "grad_norm": 0.036865234375,
      "learning_rate": 2.8325877418778246e-05,
      "loss": 0.0021,
      "step": 16161
    },
    {
      "epoch": 7.543523920653442,
      "grad_norm": 0.0196533203125,
      "learning_rate": 2.831564579513243e-05,
      "loss": 0.0002,
      "step": 16162
    },
    {
      "epoch": 7.543990665110852,
      "grad_norm": 0.01348876953125,
      "learning_rate": 2.830541571493032e-05,
      "loss": 0.0001,
      "step": 16163
    },
    {
      "epoch": 7.5444574095682615,
      "grad_norm": 0.028564453125,
      "learning_rate": 2.8295187178392136e-05,
      "loss": 0.0006,
      "step": 16164
    },
    {
      "epoch": 7.544924154025671,
      "grad_norm": 0.007049560546875,
      "learning_rate": 2.8284960185738118e-05,
      "loss": 0.0002,
      "step": 16165
    },
    {
      "epoch": 7.54539089848308,
      "grad_norm": 0.025634765625,
      "learning_rate": 2.8274734737188514e-05,
      "loss": 0.0002,
      "step": 16166
    },
    {
      "epoch": 7.54585764294049,
      "grad_norm": 0.01806640625,
      "learning_rate": 2.826451083296342e-05,
      "loss": 0.0002,
      "step": 16167
    },
    {
      "epoch": 7.5463243873979,
      "grad_norm": 0.08349609375,
      "learning_rate": 2.8254288473283e-05,
      "loss": 0.0035,
      "step": 16168
    },
    {
      "epoch": 7.54679113185531,
      "grad_norm": 0.0201416015625,
      "learning_rate": 2.824406765836738e-05,
      "loss": 0.0002,
      "step": 16169
    },
    {
      "epoch": 7.5472578763127185,
      "grad_norm": 0.01202392578125,
      "learning_rate": 2.823384838843657e-05,
      "loss": 0.0002,
      "step": 16170
    },
    {
      "epoch": 7.547724620770128,
      "grad_norm": 0.05126953125,
      "learning_rate": 2.822363066371062e-05,
      "loss": 0.0003,
      "step": 16171
    },
    {
      "epoch": 7.548191365227538,
      "grad_norm": 0.01324462890625,
      "learning_rate": 2.8213414484409582e-05,
      "loss": 0.0002,
      "step": 16172
    },
    {
      "epoch": 7.548658109684947,
      "grad_norm": 0.00872802734375,
      "learning_rate": 2.8203199850753348e-05,
      "loss": 0.0002,
      "step": 16173
    },
    {
      "epoch": 7.549124854142357,
      "grad_norm": 0.017333984375,
      "learning_rate": 2.8192986762961882e-05,
      "loss": 0.0002,
      "step": 16174
    },
    {
      "epoch": 7.549591598599767,
      "grad_norm": 0.00897216796875,
      "learning_rate": 2.8182775221255108e-05,
      "loss": 0.0002,
      "step": 16175
    },
    {
      "epoch": 7.550058343057176,
      "grad_norm": 0.0052490234375,
      "learning_rate": 2.817256522585283e-05,
      "loss": 0.0001,
      "step": 16176
    },
    {
      "epoch": 7.550525087514586,
      "grad_norm": 0.10546875,
      "learning_rate": 2.8162356776974953e-05,
      "loss": 0.0056,
      "step": 16177
    },
    {
      "epoch": 7.550991831971995,
      "grad_norm": 0.07421875,
      "learning_rate": 2.8152149874841217e-05,
      "loss": 0.0002,
      "step": 16178
    },
    {
      "epoch": 7.551458576429405,
      "grad_norm": 0.00970458984375,
      "learning_rate": 2.8141944519671402e-05,
      "loss": 0.0001,
      "step": 16179
    },
    {
      "epoch": 7.551925320886815,
      "grad_norm": 0.041259765625,
      "learning_rate": 2.8131740711685285e-05,
      "loss": 0.0007,
      "step": 16180
    },
    {
      "epoch": 7.552392065344224,
      "grad_norm": 0.05810546875,
      "learning_rate": 2.8121538451102493e-05,
      "loss": 0.0036,
      "step": 16181
    },
    {
      "epoch": 7.552858809801633,
      "grad_norm": 0.031982421875,
      "learning_rate": 2.811133773814273e-05,
      "loss": 0.0001,
      "step": 16182
    },
    {
      "epoch": 7.553325554259043,
      "grad_norm": 0.010986328125,
      "learning_rate": 2.8101138573025654e-05,
      "loss": 0.0002,
      "step": 16183
    },
    {
      "epoch": 7.553792298716453,
      "grad_norm": 0.01416015625,
      "learning_rate": 2.8090940955970812e-05,
      "loss": 0.0002,
      "step": 16184
    },
    {
      "epoch": 7.554259043173863,
      "grad_norm": 0.038330078125,
      "learning_rate": 2.8080744887197797e-05,
      "loss": 0.0023,
      "step": 16185
    },
    {
      "epoch": 7.554725787631272,
      "grad_norm": 0.0252685546875,
      "learning_rate": 2.8070550366926174e-05,
      "loss": 0.0002,
      "step": 16186
    },
    {
      "epoch": 7.5551925320886815,
      "grad_norm": 0.0380859375,
      "learning_rate": 2.8060357395375402e-05,
      "loss": 0.0018,
      "step": 16187
    },
    {
      "epoch": 7.555659276546091,
      "grad_norm": 0.06787109375,
      "learning_rate": 2.805016597276493e-05,
      "loss": 0.0004,
      "step": 16188
    },
    {
      "epoch": 7.5561260210035,
      "grad_norm": 0.00921630859375,
      "learning_rate": 2.8039976099314237e-05,
      "loss": 0.0001,
      "step": 16189
    },
    {
      "epoch": 7.55659276546091,
      "grad_norm": 0.06298828125,
      "learning_rate": 2.8029787775242678e-05,
      "loss": 0.0041,
      "step": 16190
    },
    {
      "epoch": 7.55705950991832,
      "grad_norm": 0.007659912109375,
      "learning_rate": 2.8019601000769636e-05,
      "loss": 0.0001,
      "step": 16191
    },
    {
      "epoch": 7.55752625437573,
      "grad_norm": 0.01416015625,
      "learning_rate": 2.8009415776114478e-05,
      "loss": 0.0002,
      "step": 16192
    },
    {
      "epoch": 7.557992998833139,
      "grad_norm": 0.17578125,
      "learning_rate": 2.7999232101496443e-05,
      "loss": 0.0007,
      "step": 16193
    },
    {
      "epoch": 7.558459743290548,
      "grad_norm": 0.01531982421875,
      "learning_rate": 2.798904997713482e-05,
      "loss": 0.0001,
      "step": 16194
    },
    {
      "epoch": 7.558926487747958,
      "grad_norm": 0.012451171875,
      "learning_rate": 2.7978869403248885e-05,
      "loss": 0.0002,
      "step": 16195
    },
    {
      "epoch": 7.559393232205368,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.7968690380057773e-05,
      "loss": 0.0002,
      "step": 16196
    },
    {
      "epoch": 7.559859976662777,
      "grad_norm": 0.060546875,
      "learning_rate": 2.7958512907780676e-05,
      "loss": 0.0024,
      "step": 16197
    },
    {
      "epoch": 7.5603267211201866,
      "grad_norm": 0.013916015625,
      "learning_rate": 2.7948336986636757e-05,
      "loss": 0.0002,
      "step": 16198
    },
    {
      "epoch": 7.560793465577596,
      "grad_norm": 0.01123046875,
      "learning_rate": 2.7938162616845055e-05,
      "loss": 0.0001,
      "step": 16199
    },
    {
      "epoch": 7.561260210035006,
      "grad_norm": 0.1611328125,
      "learning_rate": 2.7927989798624666e-05,
      "loss": 0.0018,
      "step": 16200
    },
    {
      "epoch": 7.561726954492415,
      "grad_norm": 0.01129150390625,
      "learning_rate": 2.791781853219465e-05,
      "loss": 0.0002,
      "step": 16201
    },
    {
      "epoch": 7.562193698949825,
      "grad_norm": 0.0198974609375,
      "learning_rate": 2.7907648817773947e-05,
      "loss": 0.0002,
      "step": 16202
    },
    {
      "epoch": 7.562660443407235,
      "grad_norm": 0.51171875,
      "learning_rate": 2.7897480655581553e-05,
      "loss": 0.0039,
      "step": 16203
    },
    {
      "epoch": 7.563127187864644,
      "grad_norm": 0.00604248046875,
      "learning_rate": 2.7887314045836444e-05,
      "loss": 0.0001,
      "step": 16204
    },
    {
      "epoch": 7.563593932322053,
      "grad_norm": 0.00823974609375,
      "learning_rate": 2.787714898875743e-05,
      "loss": 0.0001,
      "step": 16205
    },
    {
      "epoch": 7.564060676779463,
      "grad_norm": 0.0732421875,
      "learning_rate": 2.7866985484563414e-05,
      "loss": 0.0004,
      "step": 16206
    },
    {
      "epoch": 7.564527421236873,
      "grad_norm": 0.020263671875,
      "learning_rate": 2.7856823533473277e-05,
      "loss": 0.0002,
      "step": 16207
    },
    {
      "epoch": 7.564994165694283,
      "grad_norm": 0.0274658203125,
      "learning_rate": 2.784666313570573e-05,
      "loss": 0.0002,
      "step": 16208
    },
    {
      "epoch": 7.565460910151692,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.7836504291479627e-05,
      "loss": 0.0133,
      "step": 16209
    },
    {
      "epoch": 7.565927654609101,
      "grad_norm": 0.0123291015625,
      "learning_rate": 2.7826347001013607e-05,
      "loss": 0.0002,
      "step": 16210
    },
    {
      "epoch": 7.566394399066511,
      "grad_norm": 0.00811767578125,
      "learning_rate": 2.7816191264526426e-05,
      "loss": 0.0002,
      "step": 16211
    },
    {
      "epoch": 7.566861143523921,
      "grad_norm": 0.01318359375,
      "learning_rate": 2.7806037082236768e-05,
      "loss": 0.0001,
      "step": 16212
    },
    {
      "epoch": 7.56732788798133,
      "grad_norm": 0.00738525390625,
      "learning_rate": 2.7795884454363197e-05,
      "loss": 0.0002,
      "step": 16213
    },
    {
      "epoch": 7.56779463243874,
      "grad_norm": 0.2255859375,
      "learning_rate": 2.7785733381124346e-05,
      "loss": 0.0008,
      "step": 16214
    },
    {
      "epoch": 7.5682613768961495,
      "grad_norm": 0.010009765625,
      "learning_rate": 2.7775583862738807e-05,
      "loss": 0.0002,
      "step": 16215
    },
    {
      "epoch": 7.568728121353558,
      "grad_norm": 0.00860595703125,
      "learning_rate": 2.776543589942505e-05,
      "loss": 0.0002,
      "step": 16216
    },
    {
      "epoch": 7.569194865810968,
      "grad_norm": 0.0084228515625,
      "learning_rate": 2.7755289491401602e-05,
      "loss": 0.0002,
      "step": 16217
    },
    {
      "epoch": 7.569661610268378,
      "grad_norm": 0.0147705078125,
      "learning_rate": 2.7745144638886967e-05,
      "loss": 0.0002,
      "step": 16218
    },
    {
      "epoch": 7.570128354725788,
      "grad_norm": 0.07080078125,
      "learning_rate": 2.7735001342099508e-05,
      "loss": 0.008,
      "step": 16219
    },
    {
      "epoch": 7.570595099183198,
      "grad_norm": 0.2119140625,
      "learning_rate": 2.7724859601257634e-05,
      "loss": 0.0036,
      "step": 16220
    },
    {
      "epoch": 7.5710618436406065,
      "grad_norm": 0.0281982421875,
      "learning_rate": 2.771471941657977e-05,
      "loss": 0.0002,
      "step": 16221
    },
    {
      "epoch": 7.571528588098016,
      "grad_norm": 0.01080322265625,
      "learning_rate": 2.7704580788284163e-05,
      "loss": 0.0001,
      "step": 16222
    },
    {
      "epoch": 7.571995332555426,
      "grad_norm": 0.02734375,
      "learning_rate": 2.7694443716589136e-05,
      "loss": 0.0001,
      "step": 16223
    },
    {
      "epoch": 7.572462077012835,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.7684308201713006e-05,
      "loss": 0.0004,
      "step": 16224
    },
    {
      "epoch": 7.572928821470245,
      "grad_norm": 0.019287109375,
      "learning_rate": 2.7674174243873917e-05,
      "loss": 0.0002,
      "step": 16225
    },
    {
      "epoch": 7.573395565927655,
      "grad_norm": 0.02392578125,
      "learning_rate": 2.7664041843290133e-05,
      "loss": 0.0022,
      "step": 16226
    },
    {
      "epoch": 7.573862310385064,
      "grad_norm": 0.021484375,
      "learning_rate": 2.7653911000179754e-05,
      "loss": 0.0002,
      "step": 16227
    },
    {
      "epoch": 7.574329054842474,
      "grad_norm": 0.0615234375,
      "learning_rate": 2.7643781714760942e-05,
      "loss": 0.0032,
      "step": 16228
    },
    {
      "epoch": 7.574795799299883,
      "grad_norm": 0.050048828125,
      "learning_rate": 2.763365398725183e-05,
      "loss": 0.0003,
      "step": 16229
    },
    {
      "epoch": 7.575262543757293,
      "grad_norm": 0.0194091796875,
      "learning_rate": 2.7623527817870397e-05,
      "loss": 0.0002,
      "step": 16230
    },
    {
      "epoch": 7.575729288214703,
      "grad_norm": 0.01019287109375,
      "learning_rate": 2.761340320683472e-05,
      "loss": 0.0002,
      "step": 16231
    },
    {
      "epoch": 7.576196032672112,
      "grad_norm": 0.0810546875,
      "learning_rate": 2.760328015436281e-05,
      "loss": 0.0037,
      "step": 16232
    },
    {
      "epoch": 7.576662777129521,
      "grad_norm": 0.02978515625,
      "learning_rate": 2.759315866067258e-05,
      "loss": 0.0025,
      "step": 16233
    },
    {
      "epoch": 7.577129521586931,
      "grad_norm": 0.013671875,
      "learning_rate": 2.758303872598198e-05,
      "loss": 0.0002,
      "step": 16234
    },
    {
      "epoch": 7.577596266044341,
      "grad_norm": 0.01806640625,
      "learning_rate": 2.7572920350508935e-05,
      "loss": 0.0002,
      "step": 16235
    },
    {
      "epoch": 7.57806301050175,
      "grad_norm": 0.130859375,
      "learning_rate": 2.7562803534471272e-05,
      "loss": 0.0003,
      "step": 16236
    },
    {
      "epoch": 7.57852975495916,
      "grad_norm": 0.0228271484375,
      "learning_rate": 2.755268827808679e-05,
      "loss": 0.0003,
      "step": 16237
    },
    {
      "epoch": 7.5789964994165695,
      "grad_norm": 0.00970458984375,
      "learning_rate": 2.7542574581573345e-05,
      "loss": 0.0001,
      "step": 16238
    },
    {
      "epoch": 7.579463243873979,
      "grad_norm": 0.01361083984375,
      "learning_rate": 2.753246244514862e-05,
      "loss": 0.0002,
      "step": 16239
    },
    {
      "epoch": 7.579929988331388,
      "grad_norm": 0.0093994140625,
      "learning_rate": 2.7522351869030395e-05,
      "loss": 0.0002,
      "step": 16240
    },
    {
      "epoch": 7.580396732788798,
      "grad_norm": 0.162109375,
      "learning_rate": 2.7512242853436376e-05,
      "loss": 0.0004,
      "step": 16241
    },
    {
      "epoch": 7.580863477246208,
      "grad_norm": 0.0120849609375,
      "learning_rate": 2.7502135398584162e-05,
      "loss": 0.0002,
      "step": 16242
    },
    {
      "epoch": 7.581330221703618,
      "grad_norm": 0.00927734375,
      "learning_rate": 2.7492029504691408e-05,
      "loss": 0.0001,
      "step": 16243
    },
    {
      "epoch": 7.5817969661610265,
      "grad_norm": 0.01806640625,
      "learning_rate": 2.7481925171975742e-05,
      "loss": 0.0002,
      "step": 16244
    },
    {
      "epoch": 7.582263710618436,
      "grad_norm": 0.08984375,
      "learning_rate": 2.7471822400654655e-05,
      "loss": 0.0003,
      "step": 16245
    },
    {
      "epoch": 7.582730455075846,
      "grad_norm": 0.01116943359375,
      "learning_rate": 2.746172119094571e-05,
      "loss": 0.0002,
      "step": 16246
    },
    {
      "epoch": 7.583197199533256,
      "grad_norm": 0.0223388671875,
      "learning_rate": 2.745162154306642e-05,
      "loss": 0.0002,
      "step": 16247
    },
    {
      "epoch": 7.583663943990665,
      "grad_norm": 0.01251220703125,
      "learning_rate": 2.7441523457234176e-05,
      "loss": 0.0001,
      "step": 16248
    },
    {
      "epoch": 7.584130688448075,
      "grad_norm": 0.01263427734375,
      "learning_rate": 2.7431426933666438e-05,
      "loss": 0.0002,
      "step": 16249
    },
    {
      "epoch": 7.584597432905484,
      "grad_norm": 0.08349609375,
      "learning_rate": 2.742133197258062e-05,
      "loss": 0.0004,
      "step": 16250
    },
    {
      "epoch": 7.585064177362894,
      "grad_norm": 0.0162353515625,
      "learning_rate": 2.7411238574194033e-05,
      "loss": 0.0002,
      "step": 16251
    },
    {
      "epoch": 7.585530921820303,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.7401146738724017e-05,
      "loss": 0.0002,
      "step": 16252
    },
    {
      "epoch": 7.585997666277713,
      "grad_norm": 0.0654296875,
      "learning_rate": 2.7391056466387887e-05,
      "loss": 0.0019,
      "step": 16253
    },
    {
      "epoch": 7.586464410735123,
      "grad_norm": 0.00823974609375,
      "learning_rate": 2.7380967757402842e-05,
      "loss": 0.0002,
      "step": 16254
    },
    {
      "epoch": 7.5869311551925325,
      "grad_norm": 0.01202392578125,
      "learning_rate": 2.7370880611986127e-05,
      "loss": 0.0002,
      "step": 16255
    },
    {
      "epoch": 7.587397899649941,
      "grad_norm": 0.012939453125,
      "learning_rate": 2.736079503035498e-05,
      "loss": 0.0002,
      "step": 16256
    },
    {
      "epoch": 7.587864644107351,
      "grad_norm": 0.01422119140625,
      "learning_rate": 2.7350711012726472e-05,
      "loss": 0.0002,
      "step": 16257
    },
    {
      "epoch": 7.588331388564761,
      "grad_norm": 0.05078125,
      "learning_rate": 2.7340628559317794e-05,
      "loss": 0.0025,
      "step": 16258
    },
    {
      "epoch": 7.58879813302217,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.7330547670345964e-05,
      "loss": 0.0002,
      "step": 16259
    },
    {
      "epoch": 7.58926487747958,
      "grad_norm": 0.05126953125,
      "learning_rate": 2.7320468346028073e-05,
      "loss": 0.003,
      "step": 16260
    },
    {
      "epoch": 7.5897316219369895,
      "grad_norm": 0.00634765625,
      "learning_rate": 2.731039058658117e-05,
      "loss": 0.0001,
      "step": 16261
    },
    {
      "epoch": 7.590198366394399,
      "grad_norm": 0.0123291015625,
      "learning_rate": 2.7300314392222183e-05,
      "loss": 0.0002,
      "step": 16262
    },
    {
      "epoch": 7.590665110851809,
      "grad_norm": 0.044921875,
      "learning_rate": 2.7290239763168078e-05,
      "loss": 0.0002,
      "step": 16263
    },
    {
      "epoch": 7.591131855309218,
      "grad_norm": 0.033447265625,
      "learning_rate": 2.728016669963582e-05,
      "loss": 0.0002,
      "step": 16264
    },
    {
      "epoch": 7.591598599766628,
      "grad_norm": 0.0068359375,
      "learning_rate": 2.7270095201842237e-05,
      "loss": 0.0001,
      "step": 16265
    },
    {
      "epoch": 7.592065344224038,
      "grad_norm": 0.005828857421875,
      "learning_rate": 2.7260025270004185e-05,
      "loss": 0.0001,
      "step": 16266
    },
    {
      "epoch": 7.5925320886814465,
      "grad_norm": 0.01287841796875,
      "learning_rate": 2.7249956904338536e-05,
      "loss": 0.0002,
      "step": 16267
    },
    {
      "epoch": 7.592998833138856,
      "grad_norm": 0.01043701171875,
      "learning_rate": 2.7239890105061995e-05,
      "loss": 0.0002,
      "step": 16268
    },
    {
      "epoch": 7.593465577596266,
      "grad_norm": 0.01239013671875,
      "learning_rate": 2.7229824872391352e-05,
      "loss": 0.0002,
      "step": 16269
    },
    {
      "epoch": 7.593932322053676,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.721976120654335e-05,
      "loss": 0.0002,
      "step": 16270
    },
    {
      "epoch": 7.594399066511086,
      "grad_norm": 0.04052734375,
      "learning_rate": 2.72096991077346e-05,
      "loss": 0.003,
      "step": 16271
    },
    {
      "epoch": 7.594865810968495,
      "grad_norm": 0.007659912109375,
      "learning_rate": 2.7199638576181796e-05,
      "loss": 0.0001,
      "step": 16272
    },
    {
      "epoch": 7.595332555425904,
      "grad_norm": 0.00994873046875,
      "learning_rate": 2.718957961210158e-05,
      "loss": 0.0002,
      "step": 16273
    },
    {
      "epoch": 7.595799299883314,
      "grad_norm": 0.0189208984375,
      "learning_rate": 2.7179522215710472e-05,
      "loss": 0.0002,
      "step": 16274
    },
    {
      "epoch": 7.596266044340723,
      "grad_norm": 0.0255126953125,
      "learning_rate": 2.7169466387225074e-05,
      "loss": 0.0002,
      "step": 16275
    },
    {
      "epoch": 7.596732788798133,
      "grad_norm": 0.02197265625,
      "learning_rate": 2.7159412126861838e-05,
      "loss": 0.0002,
      "step": 16276
    },
    {
      "epoch": 7.597199533255543,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.7149359434837285e-05,
      "loss": 0.0001,
      "step": 16277
    },
    {
      "epoch": 7.597666277712952,
      "grad_norm": 0.01226806640625,
      "learning_rate": 2.7139308311367873e-05,
      "loss": 0.0002,
      "step": 16278
    },
    {
      "epoch": 7.598133022170361,
      "grad_norm": 0.01031494140625,
      "learning_rate": 2.7129258756669975e-05,
      "loss": 0.0001,
      "step": 16279
    },
    {
      "epoch": 7.598599766627771,
      "grad_norm": 0.017578125,
      "learning_rate": 2.711921077095998e-05,
      "loss": 0.0002,
      "step": 16280
    },
    {
      "epoch": 7.599066511085181,
      "grad_norm": 0.04541015625,
      "learning_rate": 2.7109164354454265e-05,
      "loss": 0.0004,
      "step": 16281
    },
    {
      "epoch": 7.599533255542591,
      "grad_norm": 0.0174560546875,
      "learning_rate": 2.7099119507369088e-05,
      "loss": 0.0002,
      "step": 16282
    },
    {
      "epoch": 7.6,
      "grad_norm": 0.00982666015625,
      "learning_rate": 2.7089076229920784e-05,
      "loss": 0.0001,
      "step": 16283
    },
    {
      "epoch": 7.600466744457409,
      "grad_norm": 0.05908203125,
      "learning_rate": 2.7079034522325543e-05,
      "loss": 0.0025,
      "step": 16284
    },
    {
      "epoch": 7.600933488914819,
      "grad_norm": 0.03271484375,
      "learning_rate": 2.706899438479962e-05,
      "loss": 0.0009,
      "step": 16285
    },
    {
      "epoch": 7.601400233372229,
      "grad_norm": 0.0179443359375,
      "learning_rate": 2.7058955817559128e-05,
      "loss": 0.0002,
      "step": 16286
    },
    {
      "epoch": 7.601866977829638,
      "grad_norm": 0.01171875,
      "learning_rate": 2.704891882082029e-05,
      "loss": 0.0002,
      "step": 16287
    },
    {
      "epoch": 7.602333722287048,
      "grad_norm": 0.00775146484375,
      "learning_rate": 2.703888339479913e-05,
      "loss": 0.0001,
      "step": 16288
    },
    {
      "epoch": 7.6028004667444575,
      "grad_norm": 0.01251220703125,
      "learning_rate": 2.7028849539711764e-05,
      "loss": 0.0002,
      "step": 16289
    },
    {
      "epoch": 7.603267211201867,
      "grad_norm": 0.0076904296875,
      "learning_rate": 2.7018817255774266e-05,
      "loss": 0.0002,
      "step": 16290
    },
    {
      "epoch": 7.603733955659276,
      "grad_norm": 0.01300048828125,
      "learning_rate": 2.7008786543202568e-05,
      "loss": 0.0002,
      "step": 16291
    },
    {
      "epoch": 7.604200700116686,
      "grad_norm": 0.029296875,
      "learning_rate": 2.6998757402212693e-05,
      "loss": 0.0022,
      "step": 16292
    },
    {
      "epoch": 7.604667444574096,
      "grad_norm": 0.0084228515625,
      "learning_rate": 2.6988729833020587e-05,
      "loss": 0.0001,
      "step": 16293
    },
    {
      "epoch": 7.605134189031506,
      "grad_norm": 0.012939453125,
      "learning_rate": 2.6978703835842112e-05,
      "loss": 0.0002,
      "step": 16294
    },
    {
      "epoch": 7.6056009334889145,
      "grad_norm": 0.033935546875,
      "learning_rate": 2.696867941089316e-05,
      "loss": 0.0021,
      "step": 16295
    },
    {
      "epoch": 7.606067677946324,
      "grad_norm": 0.056640625,
      "learning_rate": 2.6958656558389594e-05,
      "loss": 0.0003,
      "step": 16296
    },
    {
      "epoch": 7.606534422403734,
      "grad_norm": 0.197265625,
      "learning_rate": 2.6948635278547175e-05,
      "loss": 0.0011,
      "step": 16297
    },
    {
      "epoch": 7.607001166861144,
      "grad_norm": 0.020751953125,
      "learning_rate": 2.6938615571581683e-05,
      "loss": 0.0002,
      "step": 16298
    },
    {
      "epoch": 7.607467911318553,
      "grad_norm": 0.103515625,
      "learning_rate": 2.692859743770888e-05,
      "loss": 0.0003,
      "step": 16299
    },
    {
      "epoch": 7.607934655775963,
      "grad_norm": 0.12255859375,
      "learning_rate": 2.691858087714444e-05,
      "loss": 0.0003,
      "step": 16300
    },
    {
      "epoch": 7.608401400233372,
      "grad_norm": 0.01080322265625,
      "learning_rate": 2.690856589010402e-05,
      "loss": 0.0001,
      "step": 16301
    },
    {
      "epoch": 7.608868144690781,
      "grad_norm": 0.028076171875,
      "learning_rate": 2.689855247680331e-05,
      "loss": 0.0002,
      "step": 16302
    },
    {
      "epoch": 7.609334889148191,
      "grad_norm": 0.00677490234375,
      "learning_rate": 2.6888540637457848e-05,
      "loss": 0.0001,
      "step": 16303
    },
    {
      "epoch": 7.609801633605601,
      "grad_norm": 0.01397705078125,
      "learning_rate": 2.6878530372283217e-05,
      "loss": 0.0002,
      "step": 16304
    },
    {
      "epoch": 7.610268378063011,
      "grad_norm": 0.0228271484375,
      "learning_rate": 2.6868521681494997e-05,
      "loss": 0.0003,
      "step": 16305
    },
    {
      "epoch": 7.6107351225204205,
      "grad_norm": 0.07958984375,
      "learning_rate": 2.6858514565308602e-05,
      "loss": 0.0023,
      "step": 16306
    },
    {
      "epoch": 7.611201866977829,
      "grad_norm": 0.01202392578125,
      "learning_rate": 2.684850902393957e-05,
      "loss": 0.0001,
      "step": 16307
    },
    {
      "epoch": 7.611668611435239,
      "grad_norm": 0.330078125,
      "learning_rate": 2.6838505057603293e-05,
      "loss": 0.0019,
      "step": 16308
    },
    {
      "epoch": 7.612135355892649,
      "grad_norm": 0.0303955078125,
      "learning_rate": 2.6828502666515154e-05,
      "loss": 0.0019,
      "step": 16309
    },
    {
      "epoch": 7.612602100350058,
      "grad_norm": 0.2236328125,
      "learning_rate": 2.6818501850890576e-05,
      "loss": 0.006,
      "step": 16310
    },
    {
      "epoch": 7.613068844807468,
      "grad_norm": 0.01055908203125,
      "learning_rate": 2.680850261094482e-05,
      "loss": 0.0001,
      "step": 16311
    },
    {
      "epoch": 7.6135355892648775,
      "grad_norm": 0.02587890625,
      "learning_rate": 2.679850494689321e-05,
      "loss": 0.0002,
      "step": 16312
    },
    {
      "epoch": 7.614002333722287,
      "grad_norm": 0.0084228515625,
      "learning_rate": 2.6788508858951034e-05,
      "loss": 0.0002,
      "step": 16313
    },
    {
      "epoch": 7.614469078179697,
      "grad_norm": 0.064453125,
      "learning_rate": 2.6778514347333462e-05,
      "loss": 0.0042,
      "step": 16314
    },
    {
      "epoch": 7.614935822637106,
      "grad_norm": 0.0233154296875,
      "learning_rate": 2.676852141225572e-05,
      "loss": 0.0001,
      "step": 16315
    },
    {
      "epoch": 7.615402567094516,
      "grad_norm": 0.047119140625,
      "learning_rate": 2.6758530053932997e-05,
      "loss": 0.0019,
      "step": 16316
    },
    {
      "epoch": 7.615869311551926,
      "grad_norm": 0.04248046875,
      "learning_rate": 2.6748540272580348e-05,
      "loss": 0.002,
      "step": 16317
    },
    {
      "epoch": 7.6163360560093345,
      "grad_norm": 0.0098876953125,
      "learning_rate": 2.6738552068412903e-05,
      "loss": 0.0002,
      "step": 16318
    },
    {
      "epoch": 7.616802800466744,
      "grad_norm": 0.059814453125,
      "learning_rate": 2.6728565441645748e-05,
      "loss": 0.0055,
      "step": 16319
    },
    {
      "epoch": 7.617269544924154,
      "grad_norm": 0.006195068359375,
      "learning_rate": 2.6718580392493842e-05,
      "loss": 0.0001,
      "step": 16320
    },
    {
      "epoch": 7.617736289381564,
      "grad_norm": 0.039794921875,
      "learning_rate": 2.6708596921172213e-05,
      "loss": 0.0019,
      "step": 16321
    },
    {
      "epoch": 7.618203033838973,
      "grad_norm": 0.016357421875,
      "learning_rate": 2.6698615027895834e-05,
      "loss": 0.0002,
      "step": 16322
    },
    {
      "epoch": 7.618669778296383,
      "grad_norm": 0.01190185546875,
      "learning_rate": 2.6688634712879578e-05,
      "loss": 0.0002,
      "step": 16323
    },
    {
      "epoch": 7.619136522753792,
      "grad_norm": 0.00640869140625,
      "learning_rate": 2.667865597633835e-05,
      "loss": 0.0001,
      "step": 16324
    },
    {
      "epoch": 7.619603267211202,
      "grad_norm": 0.037109375,
      "learning_rate": 2.6668678818487047e-05,
      "loss": 0.0002,
      "step": 16325
    },
    {
      "epoch": 7.620070011668611,
      "grad_norm": 0.00787353515625,
      "learning_rate": 2.6658703239540417e-05,
      "loss": 0.0001,
      "step": 16326
    },
    {
      "epoch": 7.620536756126021,
      "grad_norm": 0.030517578125,
      "learning_rate": 2.664872923971331e-05,
      "loss": 0.0002,
      "step": 16327
    },
    {
      "epoch": 7.621003500583431,
      "grad_norm": 0.00799560546875,
      "learning_rate": 2.6638756819220423e-05,
      "loss": 0.0001,
      "step": 16328
    },
    {
      "epoch": 7.6214702450408405,
      "grad_norm": 0.00653076171875,
      "learning_rate": 2.6628785978276495e-05,
      "loss": 0.0001,
      "step": 16329
    },
    {
      "epoch": 7.621936989498249,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.661881671709624e-05,
      "loss": 0.0001,
      "step": 16330
    },
    {
      "epoch": 7.622403733955659,
      "grad_norm": 0.01361083984375,
      "learning_rate": 2.6608849035894277e-05,
      "loss": 0.0002,
      "step": 16331
    },
    {
      "epoch": 7.622870478413069,
      "grad_norm": 0.01416015625,
      "learning_rate": 2.6598882934885193e-05,
      "loss": 0.0002,
      "step": 16332
    },
    {
      "epoch": 7.623337222870479,
      "grad_norm": 0.02783203125,
      "learning_rate": 2.6588918414283602e-05,
      "loss": 0.0002,
      "step": 16333
    },
    {
      "epoch": 7.623803967327888,
      "grad_norm": 0.01251220703125,
      "learning_rate": 2.657895547430408e-05,
      "loss": 0.0002,
      "step": 16334
    },
    {
      "epoch": 7.6242707117852975,
      "grad_norm": 0.059814453125,
      "learning_rate": 2.6568994115161084e-05,
      "loss": 0.0003,
      "step": 16335
    },
    {
      "epoch": 7.624737456242707,
      "grad_norm": 0.036376953125,
      "learning_rate": 2.655903433706911e-05,
      "loss": 0.0002,
      "step": 16336
    },
    {
      "epoch": 7.625204200700117,
      "grad_norm": 0.0419921875,
      "learning_rate": 2.6549076140242636e-05,
      "loss": 0.002,
      "step": 16337
    },
    {
      "epoch": 7.625670945157526,
      "grad_norm": 0.0164794921875,
      "learning_rate": 2.6539119524896028e-05,
      "loss": 0.0002,
      "step": 16338
    },
    {
      "epoch": 7.626137689614936,
      "grad_norm": 0.01708984375,
      "learning_rate": 2.652916449124371e-05,
      "loss": 0.0002,
      "step": 16339
    },
    {
      "epoch": 7.626604434072346,
      "grad_norm": 0.2216796875,
      "learning_rate": 2.651921103949997e-05,
      "loss": 0.0004,
      "step": 16340
    },
    {
      "epoch": 7.627071178529755,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.650925916987913e-05,
      "loss": 0.0001,
      "step": 16341
    },
    {
      "epoch": 7.627537922987164,
      "grad_norm": 0.0103759765625,
      "learning_rate": 2.649930888259553e-05,
      "loss": 0.0002,
      "step": 16342
    },
    {
      "epoch": 7.628004667444574,
      "grad_norm": 0.01318359375,
      "learning_rate": 2.648936017786332e-05,
      "loss": 0.0001,
      "step": 16343
    },
    {
      "epoch": 7.628471411901984,
      "grad_norm": 0.008544921875,
      "learning_rate": 2.647941305589674e-05,
      "loss": 0.0002,
      "step": 16344
    },
    {
      "epoch": 7.628938156359393,
      "grad_norm": 0.03466796875,
      "learning_rate": 2.6469467516910006e-05,
      "loss": 0.003,
      "step": 16345
    },
    {
      "epoch": 7.629404900816803,
      "grad_norm": 0.022705078125,
      "learning_rate": 2.6459523561117195e-05,
      "loss": 0.0002,
      "step": 16346
    },
    {
      "epoch": 7.629871645274212,
      "grad_norm": 0.00811767578125,
      "learning_rate": 2.6449581188732432e-05,
      "loss": 0.0001,
      "step": 16347
    },
    {
      "epoch": 7.630338389731622,
      "grad_norm": 0.0169677734375,
      "learning_rate": 2.643964039996982e-05,
      "loss": 0.0002,
      "step": 16348
    },
    {
      "epoch": 7.630805134189032,
      "grad_norm": 0.040771484375,
      "learning_rate": 2.6429701195043344e-05,
      "loss": 0.0014,
      "step": 16349
    },
    {
      "epoch": 7.631271878646441,
      "grad_norm": 0.0064697265625,
      "learning_rate": 2.6419763574167022e-05,
      "loss": 0.0001,
      "step": 16350
    },
    {
      "epoch": 7.631738623103851,
      "grad_norm": 0.0118408203125,
      "learning_rate": 2.6409827537554853e-05,
      "loss": 0.0002,
      "step": 16351
    },
    {
      "epoch": 7.6322053675612604,
      "grad_norm": 0.037841796875,
      "learning_rate": 2.6399893085420734e-05,
      "loss": 0.0031,
      "step": 16352
    },
    {
      "epoch": 7.632672112018669,
      "grad_norm": 0.064453125,
      "learning_rate": 2.638996021797856e-05,
      "loss": 0.0042,
      "step": 16353
    },
    {
      "epoch": 7.633138856476079,
      "grad_norm": 0.01385498046875,
      "learning_rate": 2.6380028935442257e-05,
      "loss": 0.0002,
      "step": 16354
    },
    {
      "epoch": 7.633605600933489,
      "grad_norm": 0.08447265625,
      "learning_rate": 2.637009923802558e-05,
      "loss": 0.0069,
      "step": 16355
    },
    {
      "epoch": 7.634072345390899,
      "grad_norm": 0.01300048828125,
      "learning_rate": 2.636017112594239e-05,
      "loss": 0.0002,
      "step": 16356
    },
    {
      "epoch": 7.6345390898483085,
      "grad_norm": 0.03271484375,
      "learning_rate": 2.6350244599406383e-05,
      "loss": 0.0008,
      "step": 16357
    },
    {
      "epoch": 7.635005834305717,
      "grad_norm": 0.0283203125,
      "learning_rate": 2.634031965863134e-05,
      "loss": 0.0002,
      "step": 16358
    },
    {
      "epoch": 7.635472578763127,
      "grad_norm": 0.0106201171875,
      "learning_rate": 2.6330396303830962e-05,
      "loss": 0.0002,
      "step": 16359
    },
    {
      "epoch": 7.635939323220537,
      "grad_norm": 0.0257568359375,
      "learning_rate": 2.6320474535218854e-05,
      "loss": 0.0002,
      "step": 16360
    },
    {
      "epoch": 7.636406067677946,
      "grad_norm": 0.03955078125,
      "learning_rate": 2.6310554353008688e-05,
      "loss": 0.0002,
      "step": 16361
    },
    {
      "epoch": 7.636872812135356,
      "grad_norm": 0.06982421875,
      "learning_rate": 2.6300635757414072e-05,
      "loss": 0.0003,
      "step": 16362
    },
    {
      "epoch": 7.6373395565927655,
      "grad_norm": 0.01556396484375,
      "learning_rate": 2.6290718748648513e-05,
      "loss": 0.0002,
      "step": 16363
    },
    {
      "epoch": 7.637806301050175,
      "grad_norm": 0.01092529296875,
      "learning_rate": 2.6280803326925562e-05,
      "loss": 0.0001,
      "step": 16364
    },
    {
      "epoch": 7.638273045507584,
      "grad_norm": 0.0810546875,
      "learning_rate": 2.6270889492458728e-05,
      "loss": 0.0002,
      "step": 16365
    },
    {
      "epoch": 7.638739789964994,
      "grad_norm": 0.057861328125,
      "learning_rate": 2.6260977245461426e-05,
      "loss": 0.0026,
      "step": 16366
    },
    {
      "epoch": 7.639206534422404,
      "grad_norm": 0.01068115234375,
      "learning_rate": 2.6251066586147098e-05,
      "loss": 0.0001,
      "step": 16367
    },
    {
      "epoch": 7.639673278879814,
      "grad_norm": 0.22265625,
      "learning_rate": 2.6241157514729163e-05,
      "loss": 0.0017,
      "step": 16368
    },
    {
      "epoch": 7.6401400233372225,
      "grad_norm": 0.0306396484375,
      "learning_rate": 2.6231250031420918e-05,
      "loss": 0.0002,
      "step": 16369
    },
    {
      "epoch": 7.640606767794632,
      "grad_norm": 0.00994873046875,
      "learning_rate": 2.62213441364357e-05,
      "loss": 0.0001,
      "step": 16370
    },
    {
      "epoch": 7.641073512252042,
      "grad_norm": 0.031494140625,
      "learning_rate": 2.6211439829986827e-05,
      "loss": 0.0002,
      "step": 16371
    },
    {
      "epoch": 7.641540256709452,
      "grad_norm": 0.0341796875,
      "learning_rate": 2.62015371122875e-05,
      "loss": 0.0011,
      "step": 16372
    },
    {
      "epoch": 7.642007001166861,
      "grad_norm": 0.009033203125,
      "learning_rate": 2.6191635983550966e-05,
      "loss": 0.0001,
      "step": 16373
    },
    {
      "epoch": 7.642473745624271,
      "grad_norm": 0.04150390625,
      "learning_rate": 2.6181736443990424e-05,
      "loss": 0.0026,
      "step": 16374
    },
    {
      "epoch": 7.64294049008168,
      "grad_norm": 0.01214599609375,
      "learning_rate": 2.6171838493818968e-05,
      "loss": 0.0002,
      "step": 16375
    },
    {
      "epoch": 7.64340723453909,
      "grad_norm": 0.08154296875,
      "learning_rate": 2.6161942133249774e-05,
      "loss": 0.0003,
      "step": 16376
    },
    {
      "epoch": 7.643873978996499,
      "grad_norm": 0.0537109375,
      "learning_rate": 2.615204736249587e-05,
      "loss": 0.0035,
      "step": 16377
    },
    {
      "epoch": 7.644340723453909,
      "grad_norm": 0.1337890625,
      "learning_rate": 2.6142154181770305e-05,
      "loss": 0.0004,
      "step": 16378
    },
    {
      "epoch": 7.644807467911319,
      "grad_norm": 0.03271484375,
      "learning_rate": 2.6132262591286148e-05,
      "loss": 0.0002,
      "step": 16379
    },
    {
      "epoch": 7.6452742123687285,
      "grad_norm": 0.01080322265625,
      "learning_rate": 2.612237259125634e-05,
      "loss": 0.0002,
      "step": 16380
    },
    {
      "epoch": 7.645740956826137,
      "grad_norm": 0.01190185546875,
      "learning_rate": 2.611248418189378e-05,
      "loss": 0.0002,
      "step": 16381
    },
    {
      "epoch": 7.646207701283547,
      "grad_norm": 0.010009765625,
      "learning_rate": 2.6102597363411417e-05,
      "loss": 0.0002,
      "step": 16382
    },
    {
      "epoch": 7.646674445740957,
      "grad_norm": 0.0140380859375,
      "learning_rate": 2.6092712136022146e-05,
      "loss": 0.0002,
      "step": 16383
    },
    {
      "epoch": 7.647141190198367,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.608282849993877e-05,
      "loss": 0.0001,
      "step": 16384
    },
    {
      "epoch": 7.647607934655776,
      "grad_norm": 0.02294921875,
      "learning_rate": 2.6072946455374092e-05,
      "loss": 0.0002,
      "step": 16385
    },
    {
      "epoch": 7.6480746791131855,
      "grad_norm": 0.0191650390625,
      "learning_rate": 2.6063066002540948e-05,
      "loss": 0.0002,
      "step": 16386
    },
    {
      "epoch": 7.648541423570595,
      "grad_norm": 0.05615234375,
      "learning_rate": 2.605318714165198e-05,
      "loss": 0.0036,
      "step": 16387
    },
    {
      "epoch": 7.649008168028004,
      "grad_norm": 0.0167236328125,
      "learning_rate": 2.6043309872919985e-05,
      "loss": 0.0002,
      "step": 16388
    },
    {
      "epoch": 7.649474912485414,
      "grad_norm": 0.008544921875,
      "learning_rate": 2.603343419655754e-05,
      "loss": 0.0001,
      "step": 16389
    },
    {
      "epoch": 7.649941656942824,
      "grad_norm": 0.0084228515625,
      "learning_rate": 2.6023560112777334e-05,
      "loss": 0.0002,
      "step": 16390
    },
    {
      "epoch": 7.650408401400234,
      "grad_norm": 0.0086669921875,
      "learning_rate": 2.601368762179198e-05,
      "loss": 0.0002,
      "step": 16391
    },
    {
      "epoch": 7.650875145857643,
      "grad_norm": 0.039306640625,
      "learning_rate": 2.6003816723814e-05,
      "loss": 0.0002,
      "step": 16392
    },
    {
      "epoch": 7.651341890315052,
      "grad_norm": 0.0093994140625,
      "learning_rate": 2.599394741905594e-05,
      "loss": 0.0002,
      "step": 16393
    },
    {
      "epoch": 7.651808634772462,
      "grad_norm": 0.0079345703125,
      "learning_rate": 2.5984079707730336e-05,
      "loss": 0.0002,
      "step": 16394
    },
    {
      "epoch": 7.652275379229872,
      "grad_norm": 0.11376953125,
      "learning_rate": 2.597421359004958e-05,
      "loss": 0.0003,
      "step": 16395
    },
    {
      "epoch": 7.652742123687281,
      "grad_norm": 0.01251220703125,
      "learning_rate": 2.5964349066226147e-05,
      "loss": 0.0001,
      "step": 16396
    },
    {
      "epoch": 7.653208868144691,
      "grad_norm": 0.0135498046875,
      "learning_rate": 2.5954486136472446e-05,
      "loss": 0.0001,
      "step": 16397
    },
    {
      "epoch": 7.6536756126021,
      "grad_norm": 0.00726318359375,
      "learning_rate": 2.5944624801000794e-05,
      "loss": 0.0001,
      "step": 16398
    },
    {
      "epoch": 7.65414235705951,
      "grad_norm": 0.01953125,
      "learning_rate": 2.5934765060023525e-05,
      "loss": 0.0002,
      "step": 16399
    },
    {
      "epoch": 7.65460910151692,
      "grad_norm": 0.006103515625,
      "learning_rate": 2.5924906913752976e-05,
      "loss": 0.0001,
      "step": 16400
    },
    {
      "epoch": 7.655075845974329,
      "grad_norm": 0.08642578125,
      "learning_rate": 2.5915050362401337e-05,
      "loss": 0.0003,
      "step": 16401
    },
    {
      "epoch": 7.655542590431739,
      "grad_norm": 0.0272216796875,
      "learning_rate": 2.590519540618086e-05,
      "loss": 0.0002,
      "step": 16402
    },
    {
      "epoch": 7.6560093348891485,
      "grad_norm": 0.080078125,
      "learning_rate": 2.589534204530377e-05,
      "loss": 0.0028,
      "step": 16403
    },
    {
      "epoch": 7.656476079346557,
      "grad_norm": 0.01287841796875,
      "learning_rate": 2.588549027998216e-05,
      "loss": 0.0002,
      "step": 16404
    },
    {
      "epoch": 7.656942823803967,
      "grad_norm": 0.0048828125,
      "learning_rate": 2.58756401104282e-05,
      "loss": 0.0001,
      "step": 16405
    },
    {
      "epoch": 7.657409568261377,
      "grad_norm": 0.0089111328125,
      "learning_rate": 2.586579153685392e-05,
      "loss": 0.0002,
      "step": 16406
    },
    {
      "epoch": 7.657876312718787,
      "grad_norm": 0.016357421875,
      "learning_rate": 2.5855944559471402e-05,
      "loss": 0.0002,
      "step": 16407
    },
    {
      "epoch": 7.658343057176196,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.584609917849269e-05,
      "loss": 0.0002,
      "step": 16408
    },
    {
      "epoch": 7.6588098016336055,
      "grad_norm": 0.0084228515625,
      "learning_rate": 2.5836255394129715e-05,
      "loss": 0.0001,
      "step": 16409
    },
    {
      "epoch": 7.659276546091015,
      "grad_norm": 0.01055908203125,
      "learning_rate": 2.5826413206594446e-05,
      "loss": 0.0002,
      "step": 16410
    },
    {
      "epoch": 7.659743290548425,
      "grad_norm": 0.01251220703125,
      "learning_rate": 2.5816572616098823e-05,
      "loss": 0.0002,
      "step": 16411
    },
    {
      "epoch": 7.660210035005834,
      "grad_norm": 0.021728515625,
      "learning_rate": 2.580673362285467e-05,
      "loss": 0.0003,
      "step": 16412
    },
    {
      "epoch": 7.660676779463244,
      "grad_norm": 0.0096435546875,
      "learning_rate": 2.5796896227073862e-05,
      "loss": 0.0002,
      "step": 16413
    },
    {
      "epoch": 7.661143523920654,
      "grad_norm": 0.032470703125,
      "learning_rate": 2.5787060428968235e-05,
      "loss": 0.0026,
      "step": 16414
    },
    {
      "epoch": 7.661610268378063,
      "grad_norm": 0.0150146484375,
      "learning_rate": 2.5777226228749517e-05,
      "loss": 0.0001,
      "step": 16415
    },
    {
      "epoch": 7.662077012835472,
      "grad_norm": 0.006103515625,
      "learning_rate": 2.5767393626629466e-05,
      "loss": 0.0001,
      "step": 16416
    },
    {
      "epoch": 7.662543757292882,
      "grad_norm": 0.01409912109375,
      "learning_rate": 2.575756262281982e-05,
      "loss": 0.0002,
      "step": 16417
    },
    {
      "epoch": 7.663010501750292,
      "grad_norm": 0.00543212890625,
      "learning_rate": 2.5747733217532187e-05,
      "loss": 0.0001,
      "step": 16418
    },
    {
      "epoch": 7.663477246207702,
      "grad_norm": 0.06689453125,
      "learning_rate": 2.5737905410978258e-05,
      "loss": 0.0002,
      "step": 16419
    },
    {
      "epoch": 7.663943990665111,
      "grad_norm": 0.02978515625,
      "learning_rate": 2.5728079203369637e-05,
      "loss": 0.0025,
      "step": 16420
    },
    {
      "epoch": 7.66441073512252,
      "grad_norm": 0.0113525390625,
      "learning_rate": 2.5718254594917858e-05,
      "loss": 0.0002,
      "step": 16421
    },
    {
      "epoch": 7.66487747957993,
      "grad_norm": 0.017822265625,
      "learning_rate": 2.570843158583447e-05,
      "loss": 0.0002,
      "step": 16422
    },
    {
      "epoch": 7.665344224037339,
      "grad_norm": 0.0091552734375,
      "learning_rate": 2.5698610176331005e-05,
      "loss": 0.0001,
      "step": 16423
    },
    {
      "epoch": 7.665810968494749,
      "grad_norm": 0.038818359375,
      "learning_rate": 2.5688790366618875e-05,
      "loss": 0.0003,
      "step": 16424
    },
    {
      "epoch": 7.666277712952159,
      "grad_norm": 0.02587890625,
      "learning_rate": 2.5678972156909563e-05,
      "loss": 0.0002,
      "step": 16425
    },
    {
      "epoch": 7.6667444574095684,
      "grad_norm": 0.00640869140625,
      "learning_rate": 2.5669155547414448e-05,
      "loss": 0.0001,
      "step": 16426
    },
    {
      "epoch": 7.667211201866978,
      "grad_norm": 0.08349609375,
      "learning_rate": 2.5659340538344844e-05,
      "loss": 0.0003,
      "step": 16427
    },
    {
      "epoch": 7.667677946324387,
      "grad_norm": 0.03173828125,
      "learning_rate": 2.5649527129912133e-05,
      "loss": 0.0013,
      "step": 16428
    },
    {
      "epoch": 7.668144690781797,
      "grad_norm": 0.007415771484375,
      "learning_rate": 2.563971532232763e-05,
      "loss": 0.0002,
      "step": 16429
    },
    {
      "epoch": 7.668611435239207,
      "grad_norm": 0.0308837890625,
      "learning_rate": 2.5629905115802523e-05,
      "loss": 0.0003,
      "step": 16430
    },
    {
      "epoch": 7.669078179696616,
      "grad_norm": 0.0167236328125,
      "learning_rate": 2.5620096510548063e-05,
      "loss": 0.0002,
      "step": 16431
    },
    {
      "epoch": 7.6695449241540254,
      "grad_norm": 0.08642578125,
      "learning_rate": 2.5610289506775497e-05,
      "loss": 0.0052,
      "step": 16432
    },
    {
      "epoch": 7.670011668611435,
      "grad_norm": 0.033203125,
      "learning_rate": 2.5600484104695898e-05,
      "loss": 0.0002,
      "step": 16433
    },
    {
      "epoch": 7.670478413068845,
      "grad_norm": 0.0260009765625,
      "learning_rate": 2.559068030452042e-05,
      "loss": 0.0002,
      "step": 16434
    },
    {
      "epoch": 7.670945157526255,
      "grad_norm": 0.01318359375,
      "learning_rate": 2.5580878106460194e-05,
      "loss": 0.0002,
      "step": 16435
    },
    {
      "epoch": 7.671411901983664,
      "grad_norm": 0.009765625,
      "learning_rate": 2.5571077510726193e-05,
      "loss": 0.0001,
      "step": 16436
    },
    {
      "epoch": 7.6718786464410735,
      "grad_norm": 0.00872802734375,
      "learning_rate": 2.556127851752951e-05,
      "loss": 0.0002,
      "step": 16437
    },
    {
      "epoch": 7.672345390898483,
      "grad_norm": 0.023681640625,
      "learning_rate": 2.5551481127081057e-05,
      "loss": 0.0002,
      "step": 16438
    },
    {
      "epoch": 7.672812135355892,
      "grad_norm": 0.052490234375,
      "learning_rate": 2.5541685339591825e-05,
      "loss": 0.0004,
      "step": 16439
    },
    {
      "epoch": 7.673278879813302,
      "grad_norm": 0.0218505859375,
      "learning_rate": 2.5531891155272748e-05,
      "loss": 0.0001,
      "step": 16440
    },
    {
      "epoch": 7.673745624270712,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.5522098574334642e-05,
      "loss": 0.0051,
      "step": 16441
    },
    {
      "epoch": 7.674212368728122,
      "grad_norm": 0.0849609375,
      "learning_rate": 2.5512307596988383e-05,
      "loss": 0.0022,
      "step": 16442
    },
    {
      "epoch": 7.674679113185531,
      "grad_norm": 0.0115966796875,
      "learning_rate": 2.5502518223444826e-05,
      "loss": 0.0002,
      "step": 16443
    },
    {
      "epoch": 7.67514585764294,
      "grad_norm": 0.021484375,
      "learning_rate": 2.5492730453914683e-05,
      "loss": 0.0002,
      "step": 16444
    },
    {
      "epoch": 7.67561260210035,
      "grad_norm": 0.0086669921875,
      "learning_rate": 2.5482944288608713e-05,
      "loss": 0.0002,
      "step": 16445
    },
    {
      "epoch": 7.67607934655776,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.547315972773766e-05,
      "loss": 0.0005,
      "step": 16446
    },
    {
      "epoch": 7.676546091015169,
      "grad_norm": 0.0517578125,
      "learning_rate": 2.546337677151214e-05,
      "loss": 0.0002,
      "step": 16447
    },
    {
      "epoch": 7.677012835472579,
      "grad_norm": 0.0634765625,
      "learning_rate": 2.545359542014282e-05,
      "loss": 0.0031,
      "step": 16448
    },
    {
      "epoch": 7.677479579929988,
      "grad_norm": 0.0361328125,
      "learning_rate": 2.5443815673840334e-05,
      "loss": 0.0023,
      "step": 16449
    },
    {
      "epoch": 7.677946324387398,
      "grad_norm": 0.010009765625,
      "learning_rate": 2.543403753281518e-05,
      "loss": 0.0001,
      "step": 16450
    },
    {
      "epoch": 7.678413068844807,
      "grad_norm": 0.0224609375,
      "learning_rate": 2.5424260997277927e-05,
      "loss": 0.0002,
      "step": 16451
    },
    {
      "epoch": 7.678879813302217,
      "grad_norm": 0.0634765625,
      "learning_rate": 2.5414486067439114e-05,
      "loss": 0.0033,
      "step": 16452
    },
    {
      "epoch": 7.679346557759627,
      "grad_norm": 0.01708984375,
      "learning_rate": 2.540471274350914e-05,
      "loss": 0.0002,
      "step": 16453
    },
    {
      "epoch": 7.6798133022170365,
      "grad_norm": 0.00994873046875,
      "learning_rate": 2.5394941025698494e-05,
      "loss": 0.0001,
      "step": 16454
    },
    {
      "epoch": 7.680280046674445,
      "grad_norm": 0.142578125,
      "learning_rate": 2.5385170914217517e-05,
      "loss": 0.0004,
      "step": 16455
    },
    {
      "epoch": 7.680746791131855,
      "grad_norm": 0.036865234375,
      "learning_rate": 2.5375402409276593e-05,
      "loss": 0.0002,
      "step": 16456
    },
    {
      "epoch": 7.681213535589265,
      "grad_norm": 0.031494140625,
      "learning_rate": 2.5365635511086093e-05,
      "loss": 0.0002,
      "step": 16457
    },
    {
      "epoch": 7.681680280046675,
      "grad_norm": 0.00927734375,
      "learning_rate": 2.5355870219856224e-05,
      "loss": 0.0002,
      "step": 16458
    },
    {
      "epoch": 7.682147024504084,
      "grad_norm": 0.01324462890625,
      "learning_rate": 2.53461065357973e-05,
      "loss": 0.0002,
      "step": 16459
    },
    {
      "epoch": 7.6826137689614935,
      "grad_norm": 0.0185546875,
      "learning_rate": 2.5336344459119556e-05,
      "loss": 0.0001,
      "step": 16460
    },
    {
      "epoch": 7.683080513418903,
      "grad_norm": 0.01220703125,
      "learning_rate": 2.5326583990033125e-05,
      "loss": 0.0002,
      "step": 16461
    },
    {
      "epoch": 7.683547257876313,
      "grad_norm": 0.0576171875,
      "learning_rate": 2.531682512874819e-05,
      "loss": 0.0024,
      "step": 16462
    },
    {
      "epoch": 7.684014002333722,
      "grad_norm": 0.01104736328125,
      "learning_rate": 2.530706787547491e-05,
      "loss": 0.0002,
      "step": 16463
    },
    {
      "epoch": 7.684480746791132,
      "grad_norm": 0.038330078125,
      "learning_rate": 2.5297312230423298e-05,
      "loss": 0.0002,
      "step": 16464
    },
    {
      "epoch": 7.684947491248542,
      "grad_norm": 0.01544189453125,
      "learning_rate": 2.528755819380344e-05,
      "loss": 0.0002,
      "step": 16465
    },
    {
      "epoch": 7.6854142357059505,
      "grad_norm": 0.009765625,
      "learning_rate": 2.527780576582538e-05,
      "loss": 0.0001,
      "step": 16466
    },
    {
      "epoch": 7.68588098016336,
      "grad_norm": 0.017822265625,
      "learning_rate": 2.5268054946699025e-05,
      "loss": 0.0002,
      "step": 16467
    },
    {
      "epoch": 7.68634772462077,
      "grad_norm": 0.1884765625,
      "learning_rate": 2.525830573663438e-05,
      "loss": 0.0003,
      "step": 16468
    },
    {
      "epoch": 7.68681446907818,
      "grad_norm": 0.010986328125,
      "learning_rate": 2.524855813584136e-05,
      "loss": 0.0002,
      "step": 16469
    },
    {
      "epoch": 7.68728121353559,
      "grad_norm": 0.01483154296875,
      "learning_rate": 2.52388121445298e-05,
      "loss": 0.0002,
      "step": 16470
    },
    {
      "epoch": 7.687747957992999,
      "grad_norm": 0.00885009765625,
      "learning_rate": 2.5229067762909553e-05,
      "loss": 0.0001,
      "step": 16471
    },
    {
      "epoch": 7.688214702450408,
      "grad_norm": 0.0184326171875,
      "learning_rate": 2.5219324991190474e-05,
      "loss": 0.0002,
      "step": 16472
    },
    {
      "epoch": 7.688681446907818,
      "grad_norm": 0.00982666015625,
      "learning_rate": 2.5209583829582273e-05,
      "loss": 0.0001,
      "step": 16473
    },
    {
      "epoch": 7.689148191365227,
      "grad_norm": 0.0390625,
      "learning_rate": 2.5199844278294727e-05,
      "loss": 0.0035,
      "step": 16474
    },
    {
      "epoch": 7.689614935822637,
      "grad_norm": 0.036865234375,
      "learning_rate": 2.5190106337537545e-05,
      "loss": 0.0024,
      "step": 16475
    },
    {
      "epoch": 7.690081680280047,
      "grad_norm": 0.0478515625,
      "learning_rate": 2.5180370007520336e-05,
      "loss": 0.0043,
      "step": 16476
    },
    {
      "epoch": 7.6905484247374565,
      "grad_norm": 0.008056640625,
      "learning_rate": 2.5170635288452772e-05,
      "loss": 0.0001,
      "step": 16477
    },
    {
      "epoch": 7.691015169194866,
      "grad_norm": 0.049560546875,
      "learning_rate": 2.5160902180544487e-05,
      "loss": 0.004,
      "step": 16478
    },
    {
      "epoch": 7.691481913652275,
      "grad_norm": 0.028564453125,
      "learning_rate": 2.515117068400499e-05,
      "loss": 0.0003,
      "step": 16479
    },
    {
      "epoch": 7.691948658109685,
      "grad_norm": 0.030029296875,
      "learning_rate": 2.5141440799043824e-05,
      "loss": 0.0016,
      "step": 16480
    },
    {
      "epoch": 7.692415402567095,
      "grad_norm": 0.019775390625,
      "learning_rate": 2.5131712525870534e-05,
      "loss": 0.0002,
      "step": 16481
    },
    {
      "epoch": 7.692882147024504,
      "grad_norm": 0.0242919921875,
      "learning_rate": 2.512198586469451e-05,
      "loss": 0.0002,
      "step": 16482
    },
    {
      "epoch": 7.6933488914819135,
      "grad_norm": 0.07421875,
      "learning_rate": 2.5112260815725208e-05,
      "loss": 0.0019,
      "step": 16483
    },
    {
      "epoch": 7.693815635939323,
      "grad_norm": 0.047119140625,
      "learning_rate": 2.5102537379172043e-05,
      "loss": 0.0025,
      "step": 16484
    },
    {
      "epoch": 7.694282380396733,
      "grad_norm": 0.01123046875,
      "learning_rate": 2.509281555524433e-05,
      "loss": 0.0001,
      "step": 16485
    },
    {
      "epoch": 7.694749124854142,
      "grad_norm": 0.01171875,
      "learning_rate": 2.508309534415143e-05,
      "loss": 0.0001,
      "step": 16486
    },
    {
      "epoch": 7.695215869311552,
      "grad_norm": 0.01226806640625,
      "learning_rate": 2.5073376746102582e-05,
      "loss": 0.0001,
      "step": 16487
    },
    {
      "epoch": 7.695682613768962,
      "grad_norm": 0.00799560546875,
      "learning_rate": 2.5063659761307067e-05,
      "loss": 0.0001,
      "step": 16488
    },
    {
      "epoch": 7.696149358226371,
      "grad_norm": 0.1201171875,
      "learning_rate": 2.5053944389974126e-05,
      "loss": 0.0025,
      "step": 16489
    },
    {
      "epoch": 7.69661610268378,
      "grad_norm": 0.01544189453125,
      "learning_rate": 2.5044230632312892e-05,
      "loss": 0.0001,
      "step": 16490
    },
    {
      "epoch": 7.69708284714119,
      "grad_norm": 0.00848388671875,
      "learning_rate": 2.5034518488532532e-05,
      "loss": 0.0001,
      "step": 16491
    },
    {
      "epoch": 7.6975495915986,
      "grad_norm": 0.035888671875,
      "learning_rate": 2.50248079588422e-05,
      "loss": 0.0004,
      "step": 16492
    },
    {
      "epoch": 7.69801633605601,
      "grad_norm": 0.03955078125,
      "learning_rate": 2.5015099043450907e-05,
      "loss": 0.0003,
      "step": 16493
    },
    {
      "epoch": 7.698483080513419,
      "grad_norm": 0.025390625,
      "learning_rate": 2.5005391742567728e-05,
      "loss": 0.0026,
      "step": 16494
    },
    {
      "epoch": 7.698949824970828,
      "grad_norm": 0.007080078125,
      "learning_rate": 2.49956860564017e-05,
      "loss": 0.0002,
      "step": 16495
    },
    {
      "epoch": 7.699416569428238,
      "grad_norm": 0.021240234375,
      "learning_rate": 2.4985981985161743e-05,
      "loss": 0.0002,
      "step": 16496
    },
    {
      "epoch": 7.699883313885648,
      "grad_norm": 0.010498046875,
      "learning_rate": 2.4976279529056822e-05,
      "loss": 0.0001,
      "step": 16497
    },
    {
      "epoch": 7.700350058343057,
      "grad_norm": 0.05810546875,
      "learning_rate": 2.4966578688295872e-05,
      "loss": 0.0034,
      "step": 16498
    },
    {
      "epoch": 7.700816802800467,
      "grad_norm": 0.0096435546875,
      "learning_rate": 2.4956879463087712e-05,
      "loss": 0.0002,
      "step": 16499
    },
    {
      "epoch": 7.7012835472578764,
      "grad_norm": 0.023193359375,
      "learning_rate": 2.494718185364119e-05,
      "loss": 0.0002,
      "step": 16500
    },
    {
      "epoch": 7.701750291715286,
      "grad_norm": 0.0262451171875,
      "learning_rate": 2.4937485860165145e-05,
      "loss": 0.0002,
      "step": 16501
    },
    {
      "epoch": 7.702217036172695,
      "grad_norm": 0.0062255859375,
      "learning_rate": 2.492779148286829e-05,
      "loss": 0.0001,
      "step": 16502
    },
    {
      "epoch": 7.702683780630105,
      "grad_norm": 0.01153564453125,
      "learning_rate": 2.491809872195937e-05,
      "loss": 0.0002,
      "step": 16503
    },
    {
      "epoch": 7.703150525087515,
      "grad_norm": 0.0869140625,
      "learning_rate": 2.4908407577647118e-05,
      "loss": 0.0101,
      "step": 16504
    },
    {
      "epoch": 7.7036172695449245,
      "grad_norm": 0.00567626953125,
      "learning_rate": 2.4898718050140148e-05,
      "loss": 0.0001,
      "step": 16505
    },
    {
      "epoch": 7.7040840140023334,
      "grad_norm": 0.0067138671875,
      "learning_rate": 2.488903013964713e-05,
      "loss": 0.0002,
      "step": 16506
    },
    {
      "epoch": 7.704550758459743,
      "grad_norm": 0.00787353515625,
      "learning_rate": 2.48793438463766e-05,
      "loss": 0.0001,
      "step": 16507
    },
    {
      "epoch": 7.705017502917153,
      "grad_norm": 0.0262451171875,
      "learning_rate": 2.486965917053714e-05,
      "loss": 0.0002,
      "step": 16508
    },
    {
      "epoch": 7.705484247374562,
      "grad_norm": 0.01263427734375,
      "learning_rate": 2.4859976112337315e-05,
      "loss": 0.0001,
      "step": 16509
    },
    {
      "epoch": 7.705950991831972,
      "grad_norm": 0.0101318359375,
      "learning_rate": 2.485029467198553e-05,
      "loss": 0.0002,
      "step": 16510
    },
    {
      "epoch": 7.7064177362893815,
      "grad_norm": 0.0089111328125,
      "learning_rate": 2.4840614849690292e-05,
      "loss": 0.0002,
      "step": 16511
    },
    {
      "epoch": 7.706884480746791,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.483093664566004e-05,
      "loss": 0.0001,
      "step": 16512
    },
    {
      "epoch": 7.707351225204201,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.482126006010308e-05,
      "loss": 0.0002,
      "step": 16513
    },
    {
      "epoch": 7.70781796966161,
      "grad_norm": 0.009765625,
      "learning_rate": 2.4811585093227818e-05,
      "loss": 0.0002,
      "step": 16514
    },
    {
      "epoch": 7.70828471411902,
      "grad_norm": 0.11181640625,
      "learning_rate": 2.480191174524258e-05,
      "loss": 0.0004,
      "step": 16515
    },
    {
      "epoch": 7.70875145857643,
      "grad_norm": 0.01544189453125,
      "learning_rate": 2.479224001635557e-05,
      "loss": 0.0002,
      "step": 16516
    },
    {
      "epoch": 7.7092182030338385,
      "grad_norm": 0.00897216796875,
      "learning_rate": 2.478256990677509e-05,
      "loss": 0.0001,
      "step": 16517
    },
    {
      "epoch": 7.709684947491248,
      "grad_norm": 0.6640625,
      "learning_rate": 2.4772901416709372e-05,
      "loss": 0.0068,
      "step": 16518
    },
    {
      "epoch": 7.710151691948658,
      "grad_norm": 0.048828125,
      "learning_rate": 2.476323454636651e-05,
      "loss": 0.0016,
      "step": 16519
    },
    {
      "epoch": 7.710618436406068,
      "grad_norm": 0.09130859375,
      "learning_rate": 2.4753569295954682e-05,
      "loss": 0.0003,
      "step": 16520
    },
    {
      "epoch": 7.711085180863478,
      "grad_norm": 0.04248046875,
      "learning_rate": 2.4743905665682034e-05,
      "loss": 0.0043,
      "step": 16521
    },
    {
      "epoch": 7.711551925320887,
      "grad_norm": 0.0262451171875,
      "learning_rate": 2.4734243655756596e-05,
      "loss": 0.0002,
      "step": 16522
    },
    {
      "epoch": 7.712018669778296,
      "grad_norm": 0.00836181640625,
      "learning_rate": 2.472458326638636e-05,
      "loss": 0.0001,
      "step": 16523
    },
    {
      "epoch": 7.712485414235706,
      "grad_norm": 0.013671875,
      "learning_rate": 2.47149244977794e-05,
      "loss": 0.0002,
      "step": 16524
    },
    {
      "epoch": 7.712952158693115,
      "grad_norm": 0.01495361328125,
      "learning_rate": 2.4705267350143625e-05,
      "loss": 0.0002,
      "step": 16525
    },
    {
      "epoch": 7.713418903150525,
      "grad_norm": 0.0159912109375,
      "learning_rate": 2.469561182368697e-05,
      "loss": 0.0002,
      "step": 16526
    },
    {
      "epoch": 7.713885647607935,
      "grad_norm": 0.01153564453125,
      "learning_rate": 2.4685957918617385e-05,
      "loss": 0.0002,
      "step": 16527
    },
    {
      "epoch": 7.7143523920653445,
      "grad_norm": 0.01019287109375,
      "learning_rate": 2.467630563514266e-05,
      "loss": 0.0001,
      "step": 16528
    },
    {
      "epoch": 7.714819136522753,
      "grad_norm": 0.0068359375,
      "learning_rate": 2.466665497347064e-05,
      "loss": 0.0001,
      "step": 16529
    },
    {
      "epoch": 7.715285880980163,
      "grad_norm": 0.009033203125,
      "learning_rate": 2.4657005933809153e-05,
      "loss": 0.0002,
      "step": 16530
    },
    {
      "epoch": 7.715752625437573,
      "grad_norm": 0.037841796875,
      "learning_rate": 2.4647358516365903e-05,
      "loss": 0.0002,
      "step": 16531
    },
    {
      "epoch": 7.716219369894983,
      "grad_norm": 0.08447265625,
      "learning_rate": 2.4637712721348615e-05,
      "loss": 0.0004,
      "step": 16532
    },
    {
      "epoch": 7.716686114352392,
      "grad_norm": 0.040283203125,
      "learning_rate": 2.4628068548965033e-05,
      "loss": 0.0002,
      "step": 16533
    },
    {
      "epoch": 7.7171528588098015,
      "grad_norm": 0.028076171875,
      "learning_rate": 2.4618425999422722e-05,
      "loss": 0.0023,
      "step": 16534
    },
    {
      "epoch": 7.717619603267211,
      "grad_norm": 0.005767822265625,
      "learning_rate": 2.4608785072929364e-05,
      "loss": 0.0001,
      "step": 16535
    },
    {
      "epoch": 7.718086347724621,
      "grad_norm": 0.00634765625,
      "learning_rate": 2.459914576969249e-05,
      "loss": 0.0001,
      "step": 16536
    },
    {
      "epoch": 7.71855309218203,
      "grad_norm": 0.009521484375,
      "learning_rate": 2.4589508089919657e-05,
      "loss": 0.0001,
      "step": 16537
    },
    {
      "epoch": 7.71901983663944,
      "grad_norm": 0.006378173828125,
      "learning_rate": 2.457987203381842e-05,
      "loss": 0.0001,
      "step": 16538
    },
    {
      "epoch": 7.71948658109685,
      "grad_norm": 0.01708984375,
      "learning_rate": 2.457023760159619e-05,
      "loss": 0.0002,
      "step": 16539
    },
    {
      "epoch": 7.719953325554259,
      "grad_norm": 0.01953125,
      "learning_rate": 2.456060479346043e-05,
      "loss": 0.0002,
      "step": 16540
    },
    {
      "epoch": 7.720420070011668,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.4550973609618576e-05,
      "loss": 0.0002,
      "step": 16541
    },
    {
      "epoch": 7.720886814469078,
      "grad_norm": 0.048583984375,
      "learning_rate": 2.454134405027796e-05,
      "loss": 0.0003,
      "step": 16542
    },
    {
      "epoch": 7.721353558926488,
      "grad_norm": 0.01123046875,
      "learning_rate": 2.4531716115645908e-05,
      "loss": 0.0002,
      "step": 16543
    },
    {
      "epoch": 7.721820303383898,
      "grad_norm": 0.005950927734375,
      "learning_rate": 2.4522089805929783e-05,
      "loss": 0.0001,
      "step": 16544
    },
    {
      "epoch": 7.722287047841307,
      "grad_norm": 0.1083984375,
      "learning_rate": 2.451246512133678e-05,
      "loss": 0.0024,
      "step": 16545
    },
    {
      "epoch": 7.722753792298716,
      "grad_norm": 0.0106201171875,
      "learning_rate": 2.450284206207416e-05,
      "loss": 0.0002,
      "step": 16546
    },
    {
      "epoch": 7.723220536756126,
      "grad_norm": 0.028076171875,
      "learning_rate": 2.4493220628349135e-05,
      "loss": 0.0001,
      "step": 16547
    },
    {
      "epoch": 7.723687281213536,
      "grad_norm": 0.044189453125,
      "learning_rate": 2.4483600820368825e-05,
      "loss": 0.0018,
      "step": 16548
    },
    {
      "epoch": 7.724154025670945,
      "grad_norm": 0.0257568359375,
      "learning_rate": 2.4473982638340377e-05,
      "loss": 0.0002,
      "step": 16549
    },
    {
      "epoch": 7.724620770128355,
      "grad_norm": 0.0150146484375,
      "learning_rate": 2.44643660824709e-05,
      "loss": 0.0001,
      "step": 16550
    },
    {
      "epoch": 7.7250875145857645,
      "grad_norm": 0.04638671875,
      "learning_rate": 2.4454751152967402e-05,
      "loss": 0.0002,
      "step": 16551
    },
    {
      "epoch": 7.725554259043173,
      "grad_norm": 0.458984375,
      "learning_rate": 2.4445137850036938e-05,
      "loss": 0.0019,
      "step": 16552
    },
    {
      "epoch": 7.726021003500583,
      "grad_norm": 0.04443359375,
      "learning_rate": 2.4435526173886514e-05,
      "loss": 0.0003,
      "step": 16553
    },
    {
      "epoch": 7.726487747957993,
      "grad_norm": 0.007080078125,
      "learning_rate": 2.4425916124723026e-05,
      "loss": 0.0001,
      "step": 16554
    },
    {
      "epoch": 7.726954492415403,
      "grad_norm": 0.0081787109375,
      "learning_rate": 2.441630770275344e-05,
      "loss": 0.0001,
      "step": 16555
    },
    {
      "epoch": 7.727421236872813,
      "grad_norm": 0.056884765625,
      "learning_rate": 2.4406700908184577e-05,
      "loss": 0.002,
      "step": 16556
    },
    {
      "epoch": 7.7278879813302215,
      "grad_norm": 0.08984375,
      "learning_rate": 2.439709574122332e-05,
      "loss": 0.0029,
      "step": 16557
    },
    {
      "epoch": 7.728354725787631,
      "grad_norm": 0.056396484375,
      "learning_rate": 2.43874922020765e-05,
      "loss": 0.0003,
      "step": 16558
    },
    {
      "epoch": 7.728821470245041,
      "grad_norm": 0.01141357421875,
      "learning_rate": 2.4377890290950832e-05,
      "loss": 0.0002,
      "step": 16559
    },
    {
      "epoch": 7.72928821470245,
      "grad_norm": 0.0380859375,
      "learning_rate": 2.4368290008053106e-05,
      "loss": 0.0003,
      "step": 16560
    },
    {
      "epoch": 7.72975495915986,
      "grad_norm": 0.014404296875,
      "learning_rate": 2.4358691353590034e-05,
      "loss": 0.0002,
      "step": 16561
    },
    {
      "epoch": 7.73022170361727,
      "grad_norm": 0.0390625,
      "learning_rate": 2.434909432776823e-05,
      "loss": 0.0018,
      "step": 16562
    },
    {
      "epoch": 7.730688448074679,
      "grad_norm": 0.01312255859375,
      "learning_rate": 2.433949893079436e-05,
      "loss": 0.0002,
      "step": 16563
    },
    {
      "epoch": 7.731155192532089,
      "grad_norm": 0.06591796875,
      "learning_rate": 2.4329905162875057e-05,
      "loss": 0.0003,
      "step": 16564
    },
    {
      "epoch": 7.731621936989498,
      "grad_norm": 0.0076904296875,
      "learning_rate": 2.432031302421682e-05,
      "loss": 0.0001,
      "step": 16565
    },
    {
      "epoch": 7.732088681446908,
      "grad_norm": 0.046142578125,
      "learning_rate": 2.4310722515026218e-05,
      "loss": 0.0003,
      "step": 16566
    },
    {
      "epoch": 7.732555425904318,
      "grad_norm": 0.01031494140625,
      "learning_rate": 2.4301133635509766e-05,
      "loss": 0.0001,
      "step": 16567
    },
    {
      "epoch": 7.733022170361727,
      "grad_norm": 0.1962890625,
      "learning_rate": 2.4291546385873864e-05,
      "loss": 0.0044,
      "step": 16568
    },
    {
      "epoch": 7.733488914819136,
      "grad_norm": 0.04345703125,
      "learning_rate": 2.4281960766324974e-05,
      "loss": 0.0008,
      "step": 16569
    },
    {
      "epoch": 7.733955659276546,
      "grad_norm": 0.0172119140625,
      "learning_rate": 2.427237677706954e-05,
      "loss": 0.0002,
      "step": 16570
    },
    {
      "epoch": 7.734422403733956,
      "grad_norm": 0.00799560546875,
      "learning_rate": 2.4262794418313785e-05,
      "loss": 0.0001,
      "step": 16571
    },
    {
      "epoch": 7.734889148191365,
      "grad_norm": 0.013916015625,
      "learning_rate": 2.425321369026411e-05,
      "loss": 0.0002,
      "step": 16572
    },
    {
      "epoch": 7.735355892648775,
      "grad_norm": 0.006011962890625,
      "learning_rate": 2.4243634593126807e-05,
      "loss": 0.0001,
      "step": 16573
    },
    {
      "epoch": 7.7358226371061845,
      "grad_norm": 0.005706787109375,
      "learning_rate": 2.423405712710808e-05,
      "loss": 0.0001,
      "step": 16574
    },
    {
      "epoch": 7.736289381563594,
      "grad_norm": 0.0101318359375,
      "learning_rate": 2.4224481292414157e-05,
      "loss": 0.0002,
      "step": 16575
    },
    {
      "epoch": 7.736756126021003,
      "grad_norm": 0.01806640625,
      "learning_rate": 2.4214907089251272e-05,
      "loss": 0.0002,
      "step": 16576
    },
    {
      "epoch": 7.737222870478413,
      "grad_norm": 0.00860595703125,
      "learning_rate": 2.4205334517825484e-05,
      "loss": 0.0002,
      "step": 16577
    },
    {
      "epoch": 7.737689614935823,
      "grad_norm": 0.00738525390625,
      "learning_rate": 2.4195763578342934e-05,
      "loss": 0.0002,
      "step": 16578
    },
    {
      "epoch": 7.7381563593932325,
      "grad_norm": 0.018798828125,
      "learning_rate": 2.4186194271009742e-05,
      "loss": 0.0002,
      "step": 16579
    },
    {
      "epoch": 7.7386231038506414,
      "grad_norm": 0.0118408203125,
      "learning_rate": 2.417662659603187e-05,
      "loss": 0.0001,
      "step": 16580
    },
    {
      "epoch": 7.739089848308051,
      "grad_norm": 0.0091552734375,
      "learning_rate": 2.416706055361535e-05,
      "loss": 0.0001,
      "step": 16581
    },
    {
      "epoch": 7.739556592765461,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.4157496143966185e-05,
      "loss": 0.0009,
      "step": 16582
    },
    {
      "epoch": 7.740023337222871,
      "grad_norm": 0.015380859375,
      "learning_rate": 2.4147933367290254e-05,
      "loss": 0.0002,
      "step": 16583
    },
    {
      "epoch": 7.74049008168028,
      "grad_norm": 0.0380859375,
      "learning_rate": 2.4138372223793504e-05,
      "loss": 0.0003,
      "step": 16584
    },
    {
      "epoch": 7.7409568261376895,
      "grad_norm": 0.044189453125,
      "learning_rate": 2.4128812713681736e-05,
      "loss": 0.0016,
      "step": 16585
    },
    {
      "epoch": 7.741423570595099,
      "grad_norm": 0.060791015625,
      "learning_rate": 2.411925483716081e-05,
      "loss": 0.0024,
      "step": 16586
    },
    {
      "epoch": 7.741890315052509,
      "grad_norm": 0.0062255859375,
      "learning_rate": 2.4109698594436557e-05,
      "loss": 0.0001,
      "step": 16587
    },
    {
      "epoch": 7.742357059509918,
      "grad_norm": 0.01129150390625,
      "learning_rate": 2.4100143985714664e-05,
      "loss": 0.0001,
      "step": 16588
    },
    {
      "epoch": 7.742823803967328,
      "grad_norm": 0.02392578125,
      "learning_rate": 2.409059101120088e-05,
      "loss": 0.0002,
      "step": 16589
    },
    {
      "epoch": 7.743290548424738,
      "grad_norm": 0.05712890625,
      "learning_rate": 2.4081039671100913e-05,
      "loss": 0.0002,
      "step": 16590
    },
    {
      "epoch": 7.743757292882147,
      "grad_norm": 0.05029296875,
      "learning_rate": 2.4071489965620387e-05,
      "loss": 0.0002,
      "step": 16591
    },
    {
      "epoch": 7.744224037339556,
      "grad_norm": 0.036376953125,
      "learning_rate": 2.406194189496491e-05,
      "loss": 0.0027,
      "step": 16592
    },
    {
      "epoch": 7.744690781796966,
      "grad_norm": 0.0174560546875,
      "learning_rate": 2.4052395459340116e-05,
      "loss": 0.0001,
      "step": 16593
    },
    {
      "epoch": 7.745157526254376,
      "grad_norm": 0.02880859375,
      "learning_rate": 2.4042850658951477e-05,
      "loss": 0.0002,
      "step": 16594
    },
    {
      "epoch": 7.745624270711785,
      "grad_norm": 0.016357421875,
      "learning_rate": 2.403330749400454e-05,
      "loss": 0.0001,
      "step": 16595
    },
    {
      "epoch": 7.746091015169195,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.4023765964704814e-05,
      "loss": 0.0001,
      "step": 16596
    },
    {
      "epoch": 7.746557759626604,
      "grad_norm": 0.0208740234375,
      "learning_rate": 2.4014226071257672e-05,
      "loss": 0.0002,
      "step": 16597
    },
    {
      "epoch": 7.747024504084014,
      "grad_norm": 0.053955078125,
      "learning_rate": 2.4004687813868553e-05,
      "loss": 0.003,
      "step": 16598
    },
    {
      "epoch": 7.747491248541424,
      "grad_norm": 0.016357421875,
      "learning_rate": 2.3995151192742848e-05,
      "loss": 0.0002,
      "step": 16599
    },
    {
      "epoch": 7.747957992998833,
      "grad_norm": 0.1142578125,
      "learning_rate": 2.3985616208085848e-05,
      "loss": 0.0028,
      "step": 16600
    },
    {
      "epoch": 7.748424737456243,
      "grad_norm": 0.0101318359375,
      "learning_rate": 2.3976082860102855e-05,
      "loss": 0.0002,
      "step": 16601
    },
    {
      "epoch": 7.7488914819136525,
      "grad_norm": 0.083984375,
      "learning_rate": 2.3966551148999195e-05,
      "loss": 0.0035,
      "step": 16602
    },
    {
      "epoch": 7.749358226371061,
      "grad_norm": 0.0111083984375,
      "learning_rate": 2.3957021074980002e-05,
      "loss": 0.0001,
      "step": 16603
    },
    {
      "epoch": 7.749824970828471,
      "grad_norm": 0.00946044921875,
      "learning_rate": 2.3947492638250556e-05,
      "loss": 0.0001,
      "step": 16604
    },
    {
      "epoch": 7.750291715285881,
      "grad_norm": 0.006927490234375,
      "learning_rate": 2.393796583901594e-05,
      "loss": 0.0001,
      "step": 16605
    },
    {
      "epoch": 7.750758459743291,
      "grad_norm": 0.01043701171875,
      "learning_rate": 2.3928440677481324e-05,
      "loss": 0.0002,
      "step": 16606
    },
    {
      "epoch": 7.751225204200701,
      "grad_norm": 0.006072998046875,
      "learning_rate": 2.3918917153851794e-05,
      "loss": 0.0001,
      "step": 16607
    },
    {
      "epoch": 7.7516919486581095,
      "grad_norm": 0.03466796875,
      "learning_rate": 2.390939526833238e-05,
      "loss": 0.0002,
      "step": 16608
    },
    {
      "epoch": 7.752158693115519,
      "grad_norm": 0.0169677734375,
      "learning_rate": 2.3899875021128105e-05,
      "loss": 0.0002,
      "step": 16609
    },
    {
      "epoch": 7.752625437572929,
      "grad_norm": 0.029541015625,
      "learning_rate": 2.389035641244397e-05,
      "loss": 0.0002,
      "step": 16610
    },
    {
      "epoch": 7.753092182030338,
      "grad_norm": 0.012451171875,
      "learning_rate": 2.3880839442484892e-05,
      "loss": 0.0002,
      "step": 16611
    },
    {
      "epoch": 7.753558926487748,
      "grad_norm": 0.00787353515625,
      "learning_rate": 2.3871324111455796e-05,
      "loss": 0.0002,
      "step": 16612
    },
    {
      "epoch": 7.754025670945158,
      "grad_norm": 0.00994873046875,
      "learning_rate": 2.3861810419561582e-05,
      "loss": 0.0002,
      "step": 16613
    },
    {
      "epoch": 7.754492415402567,
      "grad_norm": 0.01495361328125,
      "learning_rate": 2.3852298367007043e-05,
      "loss": 0.0002,
      "step": 16614
    },
    {
      "epoch": 7.754959159859976,
      "grad_norm": 0.0206298828125,
      "learning_rate": 2.3842787953997002e-05,
      "loss": 0.0002,
      "step": 16615
    },
    {
      "epoch": 7.755425904317386,
      "grad_norm": 0.04638671875,
      "learning_rate": 2.383327918073627e-05,
      "loss": 0.0002,
      "step": 16616
    },
    {
      "epoch": 7.755425904317386,
      "eval_loss": 2.4245951175689697,
      "eval_runtime": 84.8559,
      "eval_samples_per_second": 21.26,
      "eval_steps_per_second": 2.663,
      "step": 16616
    },
    {
      "epoch": 7.755892648774796,
      "grad_norm": 0.0247802734375,
      "learning_rate": 2.3823772047429505e-05,
      "loss": 0.0002,
      "step": 16617
    },
    {
      "epoch": 7.756359393232206,
      "grad_norm": 0.03271484375,
      "learning_rate": 2.3814266554281485e-05,
      "loss": 0.0002,
      "step": 16618
    },
    {
      "epoch": 7.756826137689615,
      "grad_norm": 0.00616455078125,
      "learning_rate": 2.380476270149683e-05,
      "loss": 0.0001,
      "step": 16619
    },
    {
      "epoch": 7.757292882147024,
      "grad_norm": 0.010498046875,
      "learning_rate": 2.3795260489280145e-05,
      "loss": 0.0001,
      "step": 16620
    },
    {
      "epoch": 7.757759626604434,
      "grad_norm": 0.05615234375,
      "learning_rate": 2.378575991783606e-05,
      "loss": 0.0021,
      "step": 16621
    },
    {
      "epoch": 7.758226371061844,
      "grad_norm": 0.0274658203125,
      "learning_rate": 2.3776260987369148e-05,
      "loss": 0.0002,
      "step": 16622
    },
    {
      "epoch": 7.758693115519253,
      "grad_norm": 0.0118408203125,
      "learning_rate": 2.3766763698083884e-05,
      "loss": 0.0002,
      "step": 16623
    },
    {
      "epoch": 7.759159859976663,
      "grad_norm": 0.018798828125,
      "learning_rate": 2.3757268050184777e-05,
      "loss": 0.0003,
      "step": 16624
    },
    {
      "epoch": 7.7596266044340725,
      "grad_norm": 0.01385498046875,
      "learning_rate": 2.3747774043876316e-05,
      "loss": 0.0002,
      "step": 16625
    },
    {
      "epoch": 7.760093348891482,
      "grad_norm": 0.00665283203125,
      "learning_rate": 2.373828167936284e-05,
      "loss": 0.0001,
      "step": 16626
    },
    {
      "epoch": 7.760560093348891,
      "grad_norm": 0.0126953125,
      "learning_rate": 2.372879095684879e-05,
      "loss": 0.0001,
      "step": 16627
    },
    {
      "epoch": 7.761026837806301,
      "grad_norm": 0.0281982421875,
      "learning_rate": 2.3719301876538514e-05,
      "loss": 0.0002,
      "step": 16628
    },
    {
      "epoch": 7.761493582263711,
      "grad_norm": 0.07080078125,
      "learning_rate": 2.370981443863628e-05,
      "loss": 0.0004,
      "step": 16629
    },
    {
      "epoch": 7.761960326721121,
      "grad_norm": 0.00933837890625,
      "learning_rate": 2.3700328643346383e-05,
      "loss": 0.0001,
      "step": 16630
    },
    {
      "epoch": 7.7624270711785295,
      "grad_norm": 0.1416015625,
      "learning_rate": 2.3690844490873086e-05,
      "loss": 0.0005,
      "step": 16631
    },
    {
      "epoch": 7.762893815635939,
      "grad_norm": 0.0054931640625,
      "learning_rate": 2.368136198142056e-05,
      "loss": 0.0001,
      "step": 16632
    },
    {
      "epoch": 7.763360560093349,
      "grad_norm": 0.123046875,
      "learning_rate": 2.3671881115193005e-05,
      "loss": 0.0003,
      "step": 16633
    },
    {
      "epoch": 7.763827304550759,
      "grad_norm": 0.05078125,
      "learning_rate": 2.3662401892394525e-05,
      "loss": 0.0038,
      "step": 16634
    },
    {
      "epoch": 7.764294049008168,
      "grad_norm": 0.0146484375,
      "learning_rate": 2.365292431322922e-05,
      "loss": 0.0002,
      "step": 16635
    },
    {
      "epoch": 7.764760793465578,
      "grad_norm": 0.02587890625,
      "learning_rate": 2.364344837790119e-05,
      "loss": 0.0002,
      "step": 16636
    },
    {
      "epoch": 7.765227537922987,
      "grad_norm": 0.0118408203125,
      "learning_rate": 2.3633974086614408e-05,
      "loss": 0.0002,
      "step": 16637
    },
    {
      "epoch": 7.765694282380396,
      "grad_norm": 0.013671875,
      "learning_rate": 2.3624501439572888e-05,
      "loss": 0.0001,
      "step": 16638
    },
    {
      "epoch": 7.766161026837806,
      "grad_norm": 0.05615234375,
      "learning_rate": 2.3615030436980624e-05,
      "loss": 0.0018,
      "step": 16639
    },
    {
      "epoch": 7.766627771295216,
      "grad_norm": 0.0177001953125,
      "learning_rate": 2.3605561079041482e-05,
      "loss": 0.0002,
      "step": 16640
    },
    {
      "epoch": 7.767094515752626,
      "grad_norm": 0.01202392578125,
      "learning_rate": 2.3596093365959372e-05,
      "loss": 0.0002,
      "step": 16641
    },
    {
      "epoch": 7.7675612602100355,
      "grad_norm": 0.0157470703125,
      "learning_rate": 2.358662729793817e-05,
      "loss": 0.0002,
      "step": 16642
    },
    {
      "epoch": 7.768028004667444,
      "grad_norm": 0.017333984375,
      "learning_rate": 2.3577162875181625e-05,
      "loss": 0.0002,
      "step": 16643
    },
    {
      "epoch": 7.768494749124854,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.3567700097893564e-05,
      "loss": 0.0002,
      "step": 16644
    },
    {
      "epoch": 7.768961493582264,
      "grad_norm": 0.00830078125,
      "learning_rate": 2.3558238966277756e-05,
      "loss": 0.0002,
      "step": 16645
    },
    {
      "epoch": 7.769428238039673,
      "grad_norm": 0.0162353515625,
      "learning_rate": 2.354877948053784e-05,
      "loss": 0.0002,
      "step": 16646
    },
    {
      "epoch": 7.769894982497083,
      "grad_norm": 0.03076171875,
      "learning_rate": 2.3539321640877533e-05,
      "loss": 0.0002,
      "step": 16647
    },
    {
      "epoch": 7.7703617269544925,
      "grad_norm": 0.01007080078125,
      "learning_rate": 2.3529865447500487e-05,
      "loss": 0.0001,
      "step": 16648
    },
    {
      "epoch": 7.770828471411902,
      "grad_norm": 0.0115966796875,
      "learning_rate": 2.3520410900610268e-05,
      "loss": 0.0001,
      "step": 16649
    },
    {
      "epoch": 7.771295215869312,
      "grad_norm": 0.01397705078125,
      "learning_rate": 2.3510958000410443e-05,
      "loss": 0.0002,
      "step": 16650
    },
    {
      "epoch": 7.771761960326721,
      "grad_norm": 0.016845703125,
      "learning_rate": 2.3501506747104596e-05,
      "loss": 0.0001,
      "step": 16651
    },
    {
      "epoch": 7.772228704784131,
      "grad_norm": 0.02294921875,
      "learning_rate": 2.3492057140896152e-05,
      "loss": 0.0002,
      "step": 16652
    },
    {
      "epoch": 7.7726954492415405,
      "grad_norm": 0.19921875,
      "learning_rate": 2.3482609181988636e-05,
      "loss": 0.0003,
      "step": 16653
    },
    {
      "epoch": 7.7731621936989495,
      "grad_norm": 0.02587890625,
      "learning_rate": 2.3473162870585407e-05,
      "loss": 0.0002,
      "step": 16654
    },
    {
      "epoch": 7.773628938156359,
      "grad_norm": 0.03515625,
      "learning_rate": 2.3463718206889894e-05,
      "loss": 0.0024,
      "step": 16655
    },
    {
      "epoch": 7.774095682613769,
      "grad_norm": 0.046142578125,
      "learning_rate": 2.3454275191105478e-05,
      "loss": 0.0002,
      "step": 16656
    },
    {
      "epoch": 7.774562427071179,
      "grad_norm": 0.01177978515625,
      "learning_rate": 2.344483382343542e-05,
      "loss": 0.0002,
      "step": 16657
    },
    {
      "epoch": 7.775029171528588,
      "grad_norm": 0.0228271484375,
      "learning_rate": 2.343539410408302e-05,
      "loss": 0.0002,
      "step": 16658
    },
    {
      "epoch": 7.7754959159859975,
      "grad_norm": 0.00994873046875,
      "learning_rate": 2.342595603325156e-05,
      "loss": 0.0002,
      "step": 16659
    },
    {
      "epoch": 7.775962660443407,
      "grad_norm": 0.01422119140625,
      "learning_rate": 2.3416519611144216e-05,
      "loss": 0.0002,
      "step": 16660
    },
    {
      "epoch": 7.776429404900817,
      "grad_norm": 0.007354736328125,
      "learning_rate": 2.3407084837964166e-05,
      "loss": 0.0001,
      "step": 16661
    },
    {
      "epoch": 7.776896149358226,
      "grad_norm": 0.007537841796875,
      "learning_rate": 2.339765171391458e-05,
      "loss": 0.0001,
      "step": 16662
    },
    {
      "epoch": 7.777362893815636,
      "grad_norm": 0.00897216796875,
      "learning_rate": 2.3388220239198532e-05,
      "loss": 0.0002,
      "step": 16663
    },
    {
      "epoch": 7.777829638273046,
      "grad_norm": 0.06396484375,
      "learning_rate": 2.3378790414019093e-05,
      "loss": 0.0033,
      "step": 16664
    },
    {
      "epoch": 7.778296382730455,
      "grad_norm": 0.01007080078125,
      "learning_rate": 2.3369362238579372e-05,
      "loss": 0.0001,
      "step": 16665
    },
    {
      "epoch": 7.778763127187864,
      "grad_norm": 0.00701904296875,
      "learning_rate": 2.3359935713082248e-05,
      "loss": 0.0001,
      "step": 16666
    },
    {
      "epoch": 7.779229871645274,
      "grad_norm": 0.00958251953125,
      "learning_rate": 2.3350510837730733e-05,
      "loss": 0.0002,
      "step": 16667
    },
    {
      "epoch": 7.779696616102684,
      "grad_norm": 0.00982666015625,
      "learning_rate": 2.334108761272781e-05,
      "loss": 0.0002,
      "step": 16668
    },
    {
      "epoch": 7.780163360560094,
      "grad_norm": 0.034423828125,
      "learning_rate": 2.333166603827628e-05,
      "loss": 0.0002,
      "step": 16669
    },
    {
      "epoch": 7.780630105017503,
      "grad_norm": 0.011962890625,
      "learning_rate": 2.332224611457905e-05,
      "loss": 0.0001,
      "step": 16670
    },
    {
      "epoch": 7.781096849474912,
      "grad_norm": 0.1298828125,
      "learning_rate": 2.331282784183897e-05,
      "loss": 0.0037,
      "step": 16671
    },
    {
      "epoch": 7.781563593932322,
      "grad_norm": 0.0283203125,
      "learning_rate": 2.3303411220258752e-05,
      "loss": 0.0002,
      "step": 16672
    },
    {
      "epoch": 7.782030338389731,
      "grad_norm": 0.01263427734375,
      "learning_rate": 2.3293996250041195e-05,
      "loss": 0.0002,
      "step": 16673
    },
    {
      "epoch": 7.782497082847141,
      "grad_norm": 0.00946044921875,
      "learning_rate": 2.328458293138904e-05,
      "loss": 0.0001,
      "step": 16674
    },
    {
      "epoch": 7.782963827304551,
      "grad_norm": 0.02001953125,
      "learning_rate": 2.327517126450489e-05,
      "loss": 0.0002,
      "step": 16675
    },
    {
      "epoch": 7.7834305717619605,
      "grad_norm": 0.0098876953125,
      "learning_rate": 2.3265761249591445e-05,
      "loss": 0.0001,
      "step": 16676
    },
    {
      "epoch": 7.78389731621937,
      "grad_norm": 0.056640625,
      "learning_rate": 2.3256352886851317e-05,
      "loss": 0.0002,
      "step": 16677
    },
    {
      "epoch": 7.784364060676779,
      "grad_norm": 0.04345703125,
      "learning_rate": 2.3246946176487038e-05,
      "loss": 0.0022,
      "step": 16678
    },
    {
      "epoch": 7.784830805134189,
      "grad_norm": 0.032470703125,
      "learning_rate": 2.3237541118701155e-05,
      "loss": 0.0003,
      "step": 16679
    },
    {
      "epoch": 7.785297549591599,
      "grad_norm": 0.0279541015625,
      "learning_rate": 2.322813771369622e-05,
      "loss": 0.0002,
      "step": 16680
    },
    {
      "epoch": 7.785764294049008,
      "grad_norm": 0.0194091796875,
      "learning_rate": 2.3218735961674633e-05,
      "loss": 0.0002,
      "step": 16681
    },
    {
      "epoch": 7.7862310385064175,
      "grad_norm": 0.0068359375,
      "learning_rate": 2.3209335862838844e-05,
      "loss": 0.0001,
      "step": 16682
    },
    {
      "epoch": 7.786697782963827,
      "grad_norm": 0.00518798828125,
      "learning_rate": 2.3199937417391283e-05,
      "loss": 0.0001,
      "step": 16683
    },
    {
      "epoch": 7.787164527421237,
      "grad_norm": 0.02880859375,
      "learning_rate": 2.3190540625534252e-05,
      "loss": 0.0001,
      "step": 16684
    },
    {
      "epoch": 7.787631271878647,
      "grad_norm": 0.0086669921875,
      "learning_rate": 2.3181145487470135e-05,
      "loss": 0.0001,
      "step": 16685
    },
    {
      "epoch": 7.788098016336056,
      "grad_norm": 0.026611328125,
      "learning_rate": 2.317175200340116e-05,
      "loss": 0.0002,
      "step": 16686
    },
    {
      "epoch": 7.788564760793466,
      "grad_norm": 0.01171875,
      "learning_rate": 2.3162360173529608e-05,
      "loss": 0.0002,
      "step": 16687
    },
    {
      "epoch": 7.789031505250875,
      "grad_norm": 0.0269775390625,
      "learning_rate": 2.3152969998057718e-05,
      "loss": 0.0002,
      "step": 16688
    },
    {
      "epoch": 7.789498249708284,
      "grad_norm": 0.043212890625,
      "learning_rate": 2.3143581477187627e-05,
      "loss": 0.0003,
      "step": 16689
    },
    {
      "epoch": 7.789964994165694,
      "grad_norm": 0.07275390625,
      "learning_rate": 2.3134194611121484e-05,
      "loss": 0.0002,
      "step": 16690
    },
    {
      "epoch": 7.790431738623104,
      "grad_norm": 0.00872802734375,
      "learning_rate": 2.3124809400061454e-05,
      "loss": 0.0001,
      "step": 16691
    },
    {
      "epoch": 7.790898483080514,
      "grad_norm": 0.0123291015625,
      "learning_rate": 2.311542584420955e-05,
      "loss": 0.0002,
      "step": 16692
    },
    {
      "epoch": 7.791365227537923,
      "grad_norm": 0.0439453125,
      "learning_rate": 2.310604394376782e-05,
      "loss": 0.0003,
      "step": 16693
    },
    {
      "epoch": 7.791831971995332,
      "grad_norm": 0.0625,
      "learning_rate": 2.3096663698938314e-05,
      "loss": 0.0017,
      "step": 16694
    },
    {
      "epoch": 7.792298716452742,
      "grad_norm": 0.01007080078125,
      "learning_rate": 2.308728510992294e-05,
      "loss": 0.0002,
      "step": 16695
    },
    {
      "epoch": 7.792765460910152,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.3077908176923648e-05,
      "loss": 0.0003,
      "step": 16696
    },
    {
      "epoch": 7.793232205367561,
      "grad_norm": 0.059814453125,
      "learning_rate": 2.306853290014237e-05,
      "loss": 0.0003,
      "step": 16697
    },
    {
      "epoch": 7.793698949824971,
      "grad_norm": 0.0201416015625,
      "learning_rate": 2.3059159279780907e-05,
      "loss": 0.0002,
      "step": 16698
    },
    {
      "epoch": 7.7941656942823805,
      "grad_norm": 0.012451171875,
      "learning_rate": 2.3049787316041103e-05,
      "loss": 0.0002,
      "step": 16699
    },
    {
      "epoch": 7.79463243873979,
      "grad_norm": 0.053955078125,
      "learning_rate": 2.304041700912479e-05,
      "loss": 0.0023,
      "step": 16700
    },
    {
      "epoch": 7.795099183197199,
      "grad_norm": 0.010986328125,
      "learning_rate": 2.3031048359233664e-05,
      "loss": 0.0002,
      "step": 16701
    },
    {
      "epoch": 7.795565927654609,
      "grad_norm": 0.0693359375,
      "learning_rate": 2.302168136656949e-05,
      "loss": 0.0024,
      "step": 16702
    },
    {
      "epoch": 7.796032672112019,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.3012316031333904e-05,
      "loss": 0.0001,
      "step": 16703
    },
    {
      "epoch": 7.796499416569429,
      "grad_norm": 0.0126953125,
      "learning_rate": 2.3002952353728567e-05,
      "loss": 0.0002,
      "step": 16704
    },
    {
      "epoch": 7.7969661610268375,
      "grad_norm": 0.02392578125,
      "learning_rate": 2.299359033395512e-05,
      "loss": 0.0002,
      "step": 16705
    },
    {
      "epoch": 7.797432905484247,
      "grad_norm": 0.01019287109375,
      "learning_rate": 2.2984229972215088e-05,
      "loss": 0.0002,
      "step": 16706
    },
    {
      "epoch": 7.797899649941657,
      "grad_norm": 0.020751953125,
      "learning_rate": 2.2974871268710042e-05,
      "loss": 0.0002,
      "step": 16707
    },
    {
      "epoch": 7.798366394399067,
      "grad_norm": 0.0419921875,
      "learning_rate": 2.2965514223641493e-05,
      "loss": 0.0002,
      "step": 16708
    },
    {
      "epoch": 7.798833138856476,
      "grad_norm": 0.07275390625,
      "learning_rate": 2.2956158837210872e-05,
      "loss": 0.0007,
      "step": 16709
    },
    {
      "epoch": 7.799299883313886,
      "grad_norm": 0.039306640625,
      "learning_rate": 2.294680510961964e-05,
      "loss": 0.0002,
      "step": 16710
    },
    {
      "epoch": 7.799766627771295,
      "grad_norm": 0.0179443359375,
      "learning_rate": 2.2937453041069202e-05,
      "loss": 0.0002,
      "step": 16711
    },
    {
      "epoch": 7.800233372228705,
      "grad_norm": 0.00555419921875,
      "learning_rate": 2.2928102631760883e-05,
      "loss": 0.0001,
      "step": 16712
    },
    {
      "epoch": 7.800700116686114,
      "grad_norm": 0.006591796875,
      "learning_rate": 2.291875388189605e-05,
      "loss": 0.0001,
      "step": 16713
    },
    {
      "epoch": 7.801166861143524,
      "grad_norm": 0.0118408203125,
      "learning_rate": 2.290940679167597e-05,
      "loss": 0.0001,
      "step": 16714
    },
    {
      "epoch": 7.801633605600934,
      "grad_norm": 0.033447265625,
      "learning_rate": 2.290006136130187e-05,
      "loss": 0.0002,
      "step": 16715
    },
    {
      "epoch": 7.802100350058343,
      "grad_norm": 0.01007080078125,
      "learning_rate": 2.2890717590974996e-05,
      "loss": 0.0001,
      "step": 16716
    },
    {
      "epoch": 7.802567094515752,
      "grad_norm": 0.042724609375,
      "learning_rate": 2.288137548089655e-05,
      "loss": 0.0026,
      "step": 16717
    },
    {
      "epoch": 7.803033838973162,
      "grad_norm": 0.00958251953125,
      "learning_rate": 2.2872035031267625e-05,
      "loss": 0.0002,
      "step": 16718
    },
    {
      "epoch": 7.803500583430572,
      "grad_norm": 0.06640625,
      "learning_rate": 2.286269624228936e-05,
      "loss": 0.0003,
      "step": 16719
    },
    {
      "epoch": 7.803967327887982,
      "grad_norm": 0.021728515625,
      "learning_rate": 2.2853359114162865e-05,
      "loss": 0.0002,
      "step": 16720
    },
    {
      "epoch": 7.804434072345391,
      "grad_norm": 0.00921630859375,
      "learning_rate": 2.284402364708911e-05,
      "loss": 0.0001,
      "step": 16721
    },
    {
      "epoch": 7.8049008168028005,
      "grad_norm": 0.0269775390625,
      "learning_rate": 2.283468984126913e-05,
      "loss": 0.0002,
      "step": 16722
    },
    {
      "epoch": 7.80536756126021,
      "grad_norm": 0.017333984375,
      "learning_rate": 2.2825357696903925e-05,
      "loss": 0.0002,
      "step": 16723
    },
    {
      "epoch": 7.805834305717619,
      "grad_norm": 0.013427734375,
      "learning_rate": 2.2816027214194368e-05,
      "loss": 0.0002,
      "step": 16724
    },
    {
      "epoch": 7.806301050175029,
      "grad_norm": 0.044921875,
      "learning_rate": 2.2806698393341387e-05,
      "loss": 0.0002,
      "step": 16725
    },
    {
      "epoch": 7.806767794632439,
      "grad_norm": 0.0086669921875,
      "learning_rate": 2.2797371234545863e-05,
      "loss": 0.0002,
      "step": 16726
    },
    {
      "epoch": 7.8072345390898485,
      "grad_norm": 0.054931640625,
      "learning_rate": 2.2788045738008557e-05,
      "loss": 0.004,
      "step": 16727
    },
    {
      "epoch": 7.807701283547258,
      "grad_norm": 0.0107421875,
      "learning_rate": 2.27787219039303e-05,
      "loss": 0.0002,
      "step": 16728
    },
    {
      "epoch": 7.808168028004667,
      "grad_norm": 0.008056640625,
      "learning_rate": 2.2769399732511863e-05,
      "loss": 0.0001,
      "step": 16729
    },
    {
      "epoch": 7.808634772462077,
      "grad_norm": 0.0250244140625,
      "learning_rate": 2.2760079223953924e-05,
      "loss": 0.0002,
      "step": 16730
    },
    {
      "epoch": 7.809101516919487,
      "grad_norm": 0.005859375,
      "learning_rate": 2.275076037845717e-05,
      "loss": 0.0001,
      "step": 16731
    },
    {
      "epoch": 7.809568261376896,
      "grad_norm": 0.006683349609375,
      "learning_rate": 2.274144319622228e-05,
      "loss": 0.0001,
      "step": 16732
    },
    {
      "epoch": 7.8100350058343055,
      "grad_norm": 0.0390625,
      "learning_rate": 2.273212767744981e-05,
      "loss": 0.0002,
      "step": 16733
    },
    {
      "epoch": 7.810501750291715,
      "grad_norm": 0.015625,
      "learning_rate": 2.2722813822340393e-05,
      "loss": 0.0002,
      "step": 16734
    },
    {
      "epoch": 7.810968494749125,
      "grad_norm": 0.01348876953125,
      "learning_rate": 2.2713501631094503e-05,
      "loss": 0.0002,
      "step": 16735
    },
    {
      "epoch": 7.811435239206534,
      "grad_norm": 0.04150390625,
      "learning_rate": 2.2704191103912674e-05,
      "loss": 0.0037,
      "step": 16736
    },
    {
      "epoch": 7.811901983663944,
      "grad_norm": 0.010986328125,
      "learning_rate": 2.2694882240995408e-05,
      "loss": 0.0002,
      "step": 16737
    },
    {
      "epoch": 7.812368728121354,
      "grad_norm": 0.01324462890625,
      "learning_rate": 2.268557504254306e-05,
      "loss": 0.0002,
      "step": 16738
    },
    {
      "epoch": 7.812835472578763,
      "grad_norm": 0.1025390625,
      "learning_rate": 2.2676269508756066e-05,
      "loss": 0.0023,
      "step": 16739
    },
    {
      "epoch": 7.813302217036172,
      "grad_norm": 0.010986328125,
      "learning_rate": 2.2666965639834813e-05,
      "loss": 0.0001,
      "step": 16740
    },
    {
      "epoch": 7.813768961493582,
      "grad_norm": 0.007537841796875,
      "learning_rate": 2.2657663435979558e-05,
      "loss": 0.0001,
      "step": 16741
    },
    {
      "epoch": 7.814235705950992,
      "grad_norm": 0.0284423828125,
      "learning_rate": 2.264836289739063e-05,
      "loss": 0.0002,
      "step": 16742
    },
    {
      "epoch": 7.814702450408402,
      "grad_norm": 0.03759765625,
      "learning_rate": 2.2639064024268297e-05,
      "loss": 0.0019,
      "step": 16743
    },
    {
      "epoch": 7.815169194865811,
      "grad_norm": 0.0164794921875,
      "learning_rate": 2.2629766816812724e-05,
      "loss": 0.0002,
      "step": 16744
    },
    {
      "epoch": 7.81563593932322,
      "grad_norm": 0.041748046875,
      "learning_rate": 2.262047127522411e-05,
      "loss": 0.0002,
      "step": 16745
    },
    {
      "epoch": 7.81610268378063,
      "grad_norm": 0.0126953125,
      "learning_rate": 2.2611177399702632e-05,
      "loss": 0.0002,
      "step": 16746
    },
    {
      "epoch": 7.81656942823804,
      "grad_norm": 0.012939453125,
      "learning_rate": 2.2601885190448345e-05,
      "loss": 0.0002,
      "step": 16747
    },
    {
      "epoch": 7.817036172695449,
      "grad_norm": 0.0198974609375,
      "learning_rate": 2.2592594647661335e-05,
      "loss": 0.0002,
      "step": 16748
    },
    {
      "epoch": 7.817502917152859,
      "grad_norm": 0.03564453125,
      "learning_rate": 2.2583305771541686e-05,
      "loss": 0.0034,
      "step": 16749
    },
    {
      "epoch": 7.8179696616102685,
      "grad_norm": 0.00933837890625,
      "learning_rate": 2.2574018562289324e-05,
      "loss": 0.0001,
      "step": 16750
    },
    {
      "epoch": 7.818436406067678,
      "grad_norm": 0.041015625,
      "learning_rate": 2.2564733020104277e-05,
      "loss": 0.0002,
      "step": 16751
    },
    {
      "epoch": 7.818903150525087,
      "grad_norm": 0.01544189453125,
      "learning_rate": 2.2555449145186426e-05,
      "loss": 0.0002,
      "step": 16752
    },
    {
      "epoch": 7.819369894982497,
      "grad_norm": 0.039306640625,
      "learning_rate": 2.254616693773567e-05,
      "loss": 0.0031,
      "step": 16753
    },
    {
      "epoch": 7.819836639439907,
      "grad_norm": 0.05712890625,
      "learning_rate": 2.2536886397951907e-05,
      "loss": 0.0002,
      "step": 16754
    },
    {
      "epoch": 7.820303383897317,
      "grad_norm": 0.048828125,
      "learning_rate": 2.25276075260349e-05,
      "loss": 0.0002,
      "step": 16755
    },
    {
      "epoch": 7.8207701283547255,
      "grad_norm": 0.00927734375,
      "learning_rate": 2.2518330322184455e-05,
      "loss": 0.0001,
      "step": 16756
    },
    {
      "epoch": 7.821236872812135,
      "grad_norm": 0.125,
      "learning_rate": 2.250905478660036e-05,
      "loss": 0.0004,
      "step": 16757
    },
    {
      "epoch": 7.821703617269545,
      "grad_norm": 0.007354736328125,
      "learning_rate": 2.249978091948226e-05,
      "loss": 0.0002,
      "step": 16758
    },
    {
      "epoch": 7.822170361726954,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.2490508721029867e-05,
      "loss": 0.0002,
      "step": 16759
    },
    {
      "epoch": 7.822637106184364,
      "grad_norm": 0.034423828125,
      "learning_rate": 2.2481238191442854e-05,
      "loss": 0.0019,
      "step": 16760
    },
    {
      "epoch": 7.823103850641774,
      "grad_norm": 0.0081787109375,
      "learning_rate": 2.2471969330920783e-05,
      "loss": 0.0002,
      "step": 16761
    },
    {
      "epoch": 7.823570595099183,
      "grad_norm": 0.008544921875,
      "learning_rate": 2.2462702139663207e-05,
      "loss": 0.0002,
      "step": 16762
    },
    {
      "epoch": 7.824037339556593,
      "grad_norm": 0.007049560546875,
      "learning_rate": 2.2453436617869704e-05,
      "loss": 0.0001,
      "step": 16763
    },
    {
      "epoch": 7.824504084014002,
      "grad_norm": 0.0126953125,
      "learning_rate": 2.2444172765739734e-05,
      "loss": 0.0001,
      "step": 16764
    },
    {
      "epoch": 7.824970828471412,
      "grad_norm": 0.041259765625,
      "learning_rate": 2.243491058347277e-05,
      "loss": 0.0002,
      "step": 16765
    },
    {
      "epoch": 7.825437572928822,
      "grad_norm": 0.049560546875,
      "learning_rate": 2.2425650071268255e-05,
      "loss": 0.0021,
      "step": 16766
    },
    {
      "epoch": 7.825904317386231,
      "grad_norm": 0.007354736328125,
      "learning_rate": 2.241639122932555e-05,
      "loss": 0.0002,
      "step": 16767
    },
    {
      "epoch": 7.82637106184364,
      "grad_norm": 0.007080078125,
      "learning_rate": 2.2407134057844026e-05,
      "loss": 0.0001,
      "step": 16768
    },
    {
      "epoch": 7.82683780630105,
      "grad_norm": 0.015380859375,
      "learning_rate": 2.2397878557023023e-05,
      "loss": 0.0001,
      "step": 16769
    },
    {
      "epoch": 7.82730455075846,
      "grad_norm": 0.0255126953125,
      "learning_rate": 2.238862472706176e-05,
      "loss": 0.002,
      "step": 16770
    },
    {
      "epoch": 7.82777129521587,
      "grad_norm": 0.0137939453125,
      "learning_rate": 2.2379372568159517e-05,
      "loss": 0.0001,
      "step": 16771
    },
    {
      "epoch": 7.828238039673279,
      "grad_norm": 0.013427734375,
      "learning_rate": 2.237012208051554e-05,
      "loss": 0.0002,
      "step": 16772
    },
    {
      "epoch": 7.8287047841306885,
      "grad_norm": 0.0184326171875,
      "learning_rate": 2.236087326432893e-05,
      "loss": 0.0002,
      "step": 16773
    },
    {
      "epoch": 7.829171528588098,
      "grad_norm": 0.02734375,
      "learning_rate": 2.2351626119798874e-05,
      "loss": 0.0002,
      "step": 16774
    },
    {
      "epoch": 7.829638273045507,
      "grad_norm": 0.00823974609375,
      "learning_rate": 2.234238064712447e-05,
      "loss": 0.0001,
      "step": 16775
    },
    {
      "epoch": 7.830105017502917,
      "grad_norm": 0.008544921875,
      "learning_rate": 2.2333136846504754e-05,
      "loss": 0.0002,
      "step": 16776
    },
    {
      "epoch": 7.830571761960327,
      "grad_norm": 0.01190185546875,
      "learning_rate": 2.2323894718138783e-05,
      "loss": 0.0002,
      "step": 16777
    },
    {
      "epoch": 7.831038506417737,
      "grad_norm": 0.025146484375,
      "learning_rate": 2.2314654262225563e-05,
      "loss": 0.0002,
      "step": 16778
    },
    {
      "epoch": 7.8315052508751455,
      "grad_norm": 0.0791015625,
      "learning_rate": 2.2305415478964e-05,
      "loss": 0.005,
      "step": 16779
    },
    {
      "epoch": 7.831971995332555,
      "grad_norm": 0.005615234375,
      "learning_rate": 2.2296178368553046e-05,
      "loss": 0.0001,
      "step": 16780
    },
    {
      "epoch": 7.832438739789965,
      "grad_norm": 0.4140625,
      "learning_rate": 2.2286942931191613e-05,
      "loss": 0.0011,
      "step": 16781
    },
    {
      "epoch": 7.832905484247375,
      "grad_norm": 0.05126953125,
      "learning_rate": 2.22777091670785e-05,
      "loss": 0.0002,
      "step": 16782
    },
    {
      "epoch": 7.833372228704784,
      "grad_norm": 0.03955078125,
      "learning_rate": 2.2268477076412562e-05,
      "loss": 0.0019,
      "step": 16783
    },
    {
      "epoch": 7.833838973162194,
      "grad_norm": 0.07568359375,
      "learning_rate": 2.225924665939254e-05,
      "loss": 0.0027,
      "step": 16784
    },
    {
      "epoch": 7.834305717619603,
      "grad_norm": 0.023193359375,
      "learning_rate": 2.225001791621718e-05,
      "loss": 0.0002,
      "step": 16785
    },
    {
      "epoch": 7.834772462077013,
      "grad_norm": 0.036865234375,
      "learning_rate": 2.2240790847085234e-05,
      "loss": 0.0027,
      "step": 16786
    },
    {
      "epoch": 7.835239206534422,
      "grad_norm": 0.07275390625,
      "learning_rate": 2.2231565452195312e-05,
      "loss": 0.0041,
      "step": 16787
    },
    {
      "epoch": 7.835705950991832,
      "grad_norm": 0.0220947265625,
      "learning_rate": 2.2222341731746065e-05,
      "loss": 0.0002,
      "step": 16788
    },
    {
      "epoch": 7.836172695449242,
      "grad_norm": 0.06591796875,
      "learning_rate": 2.2213119685936135e-05,
      "loss": 0.0015,
      "step": 16789
    },
    {
      "epoch": 7.8366394399066515,
      "grad_norm": 0.00946044921875,
      "learning_rate": 2.2203899314964004e-05,
      "loss": 0.0001,
      "step": 16790
    },
    {
      "epoch": 7.83710618436406,
      "grad_norm": 0.01806640625,
      "learning_rate": 2.219468061902824e-05,
      "loss": 0.0002,
      "step": 16791
    },
    {
      "epoch": 7.83757292882147,
      "grad_norm": 0.005706787109375,
      "learning_rate": 2.2185463598327362e-05,
      "loss": 0.0001,
      "step": 16792
    },
    {
      "epoch": 7.83803967327888,
      "grad_norm": 0.0076904296875,
      "learning_rate": 2.217624825305975e-05,
      "loss": 0.0001,
      "step": 16793
    },
    {
      "epoch": 7.83850641773629,
      "grad_norm": 0.0595703125,
      "learning_rate": 2.2167034583423873e-05,
      "loss": 0.003,
      "step": 16794
    },
    {
      "epoch": 7.838973162193699,
      "grad_norm": 0.03662109375,
      "learning_rate": 2.2157822589618127e-05,
      "loss": 0.0037,
      "step": 16795
    },
    {
      "epoch": 7.8394399066511085,
      "grad_norm": 0.02392578125,
      "learning_rate": 2.2148612271840784e-05,
      "loss": 0.0002,
      "step": 16796
    },
    {
      "epoch": 7.839906651108518,
      "grad_norm": 0.0115966796875,
      "learning_rate": 2.213940363029021e-05,
      "loss": 0.0002,
      "step": 16797
    },
    {
      "epoch": 7.840373395565928,
      "grad_norm": 0.007293701171875,
      "learning_rate": 2.2130196665164693e-05,
      "loss": 0.0001,
      "step": 16798
    },
    {
      "epoch": 7.840840140023337,
      "grad_norm": 0.00823974609375,
      "learning_rate": 2.2120991376662405e-05,
      "loss": 0.0001,
      "step": 16799
    },
    {
      "epoch": 7.841306884480747,
      "grad_norm": 0.0556640625,
      "learning_rate": 2.2111787764981586e-05,
      "loss": 0.0002,
      "step": 16800
    },
    {
      "epoch": 7.8417736289381565,
      "grad_norm": 0.00872802734375,
      "learning_rate": 2.210258583032042e-05,
      "loss": 0.0001,
      "step": 16801
    },
    {
      "epoch": 7.8422403733955655,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.2093385572876978e-05,
      "loss": 0.0002,
      "step": 16802
    },
    {
      "epoch": 7.842707117852975,
      "grad_norm": 0.007049560546875,
      "learning_rate": 2.208418699284942e-05,
      "loss": 0.0001,
      "step": 16803
    },
    {
      "epoch": 7.843173862310385,
      "grad_norm": 0.0216064453125,
      "learning_rate": 2.2074990090435733e-05,
      "loss": 0.0002,
      "step": 16804
    },
    {
      "epoch": 7.843640606767795,
      "grad_norm": 0.0211181640625,
      "learning_rate": 2.2065794865833967e-05,
      "loss": 0.0003,
      "step": 16805
    },
    {
      "epoch": 7.844107351225205,
      "grad_norm": 0.01287841796875,
      "learning_rate": 2.205660131924213e-05,
      "loss": 0.0002,
      "step": 16806
    },
    {
      "epoch": 7.8445740956826135,
      "grad_norm": 0.01153564453125,
      "learning_rate": 2.2047409450858124e-05,
      "loss": 0.0002,
      "step": 16807
    },
    {
      "epoch": 7.845040840140023,
      "grad_norm": 0.01177978515625,
      "learning_rate": 2.203821926087989e-05,
      "loss": 0.0001,
      "step": 16808
    },
    {
      "epoch": 7.845507584597433,
      "grad_norm": 0.0078125,
      "learning_rate": 2.2029030749505307e-05,
      "loss": 0.0002,
      "step": 16809
    },
    {
      "epoch": 7.845974329054842,
      "grad_norm": 0.017822265625,
      "learning_rate": 2.2019843916932213e-05,
      "loss": 0.0002,
      "step": 16810
    },
    {
      "epoch": 7.846441073512252,
      "grad_norm": 0.0050048828125,
      "learning_rate": 2.201065876335837e-05,
      "loss": 0.0001,
      "step": 16811
    },
    {
      "epoch": 7.846907817969662,
      "grad_norm": 0.01239013671875,
      "learning_rate": 2.2001475288981578e-05,
      "loss": 0.0002,
      "step": 16812
    },
    {
      "epoch": 7.847374562427071,
      "grad_norm": 0.0439453125,
      "learning_rate": 2.1992293493999583e-05,
      "loss": 0.0026,
      "step": 16813
    },
    {
      "epoch": 7.847841306884481,
      "grad_norm": 0.0791015625,
      "learning_rate": 2.198311337861004e-05,
      "loss": 0.0045,
      "step": 16814
    },
    {
      "epoch": 7.84830805134189,
      "grad_norm": 0.05224609375,
      "learning_rate": 2.1973934943010653e-05,
      "loss": 0.0003,
      "step": 16815
    },
    {
      "epoch": 7.8487747957993,
      "grad_norm": 0.036865234375,
      "learning_rate": 2.1964758187398994e-05,
      "loss": 0.0015,
      "step": 16816
    },
    {
      "epoch": 7.84924154025671,
      "grad_norm": 0.01220703125,
      "learning_rate": 2.1955583111972667e-05,
      "loss": 0.0002,
      "step": 16817
    },
    {
      "epoch": 7.849708284714119,
      "grad_norm": 0.0118408203125,
      "learning_rate": 2.194640971692926e-05,
      "loss": 0.0001,
      "step": 16818
    },
    {
      "epoch": 7.850175029171528,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.193723800246622e-05,
      "loss": 0.0002,
      "step": 16819
    },
    {
      "epoch": 7.850641773628938,
      "grad_norm": 0.10009765625,
      "learning_rate": 2.1928067968781063e-05,
      "loss": 0.0003,
      "step": 16820
    },
    {
      "epoch": 7.851108518086348,
      "grad_norm": 0.0101318359375,
      "learning_rate": 2.1918899616071252e-05,
      "loss": 0.0001,
      "step": 16821
    },
    {
      "epoch": 7.851575262543757,
      "grad_norm": 0.01226806640625,
      "learning_rate": 2.1909732944534133e-05,
      "loss": 0.0002,
      "step": 16822
    },
    {
      "epoch": 7.852042007001167,
      "grad_norm": 0.0478515625,
      "learning_rate": 2.1900567954367114e-05,
      "loss": 0.0002,
      "step": 16823
    },
    {
      "epoch": 7.8525087514585765,
      "grad_norm": 0.007659912109375,
      "learning_rate": 2.1891404645767544e-05,
      "loss": 0.0002,
      "step": 16824
    },
    {
      "epoch": 7.852975495915986,
      "grad_norm": 0.0277099609375,
      "learning_rate": 2.1882243018932656e-05,
      "loss": 0.0002,
      "step": 16825
    },
    {
      "epoch": 7.853442240373395,
      "grad_norm": 0.01080322265625,
      "learning_rate": 2.1873083074059762e-05,
      "loss": 0.0001,
      "step": 16826
    },
    {
      "epoch": 7.853908984830805,
      "grad_norm": 0.06494140625,
      "learning_rate": 2.1863924811346102e-05,
      "loss": 0.0024,
      "step": 16827
    },
    {
      "epoch": 7.854375729288215,
      "grad_norm": 0.0126953125,
      "learning_rate": 2.1854768230988797e-05,
      "loss": 0.0001,
      "step": 16828
    },
    {
      "epoch": 7.854842473745625,
      "grad_norm": 0.013427734375,
      "learning_rate": 2.1845613333185043e-05,
      "loss": 0.0002,
      "step": 16829
    },
    {
      "epoch": 7.8553092182030335,
      "grad_norm": 0.0191650390625,
      "learning_rate": 2.1836460118131972e-05,
      "loss": 0.0002,
      "step": 16830
    },
    {
      "epoch": 7.855775962660443,
      "grad_norm": 0.0184326171875,
      "learning_rate": 2.1827308586026608e-05,
      "loss": 0.0002,
      "step": 16831
    },
    {
      "epoch": 7.856242707117853,
      "grad_norm": 0.0272216796875,
      "learning_rate": 2.1818158737066064e-05,
      "loss": 0.0002,
      "step": 16832
    },
    {
      "epoch": 7.856709451575263,
      "grad_norm": 0.033447265625,
      "learning_rate": 2.1809010571447262e-05,
      "loss": 0.0003,
      "step": 16833
    },
    {
      "epoch": 7.857176196032672,
      "grad_norm": 0.0400390625,
      "learning_rate": 2.1799864089367227e-05,
      "loss": 0.0026,
      "step": 16834
    },
    {
      "epoch": 7.857642940490082,
      "grad_norm": 0.0595703125,
      "learning_rate": 2.179071929102291e-05,
      "loss": 0.0002,
      "step": 16835
    },
    {
      "epoch": 7.858109684947491,
      "grad_norm": 0.01373291015625,
      "learning_rate": 2.1781576176611153e-05,
      "loss": 0.0002,
      "step": 16836
    },
    {
      "epoch": 7.858576429404901,
      "grad_norm": 0.022705078125,
      "learning_rate": 2.1772434746328853e-05,
      "loss": 0.0002,
      "step": 16837
    },
    {
      "epoch": 7.85904317386231,
      "grad_norm": 0.025390625,
      "learning_rate": 2.1763295000372842e-05,
      "loss": 0.0002,
      "step": 16838
    },
    {
      "epoch": 7.85950991831972,
      "grad_norm": 0.055908203125,
      "learning_rate": 2.1754156938939873e-05,
      "loss": 0.0026,
      "step": 16839
    },
    {
      "epoch": 7.85997666277713,
      "grad_norm": 0.0087890625,
      "learning_rate": 2.1745020562226715e-05,
      "loss": 0.0002,
      "step": 16840
    },
    {
      "epoch": 7.8604434072345395,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.1735885870430127e-05,
      "loss": 0.0002,
      "step": 16841
    },
    {
      "epoch": 7.860910151691948,
      "grad_norm": 0.0277099609375,
      "learning_rate": 2.1726752863746725e-05,
      "loss": 0.0002,
      "step": 16842
    },
    {
      "epoch": 7.861376896149358,
      "grad_norm": 0.0264892578125,
      "learning_rate": 2.1717621542373168e-05,
      "loss": 0.0002,
      "step": 16843
    },
    {
      "epoch": 7.861843640606768,
      "grad_norm": 0.03466796875,
      "learning_rate": 2.170849190650611e-05,
      "loss": 0.0002,
      "step": 16844
    },
    {
      "epoch": 7.862310385064177,
      "grad_norm": 0.006988525390625,
      "learning_rate": 2.1699363956342058e-05,
      "loss": 0.0001,
      "step": 16845
    },
    {
      "epoch": 7.862777129521587,
      "grad_norm": 0.0164794921875,
      "learning_rate": 2.1690237692077576e-05,
      "loss": 0.0001,
      "step": 16846
    },
    {
      "epoch": 7.8632438739789965,
      "grad_norm": 0.0174560546875,
      "learning_rate": 2.168111311390919e-05,
      "loss": 0.0002,
      "step": 16847
    },
    {
      "epoch": 7.863710618436406,
      "grad_norm": 0.042724609375,
      "learning_rate": 2.167199022203331e-05,
      "loss": 0.0022,
      "step": 16848
    },
    {
      "epoch": 7.864177362893816,
      "grad_norm": 0.01007080078125,
      "learning_rate": 2.166286901664638e-05,
      "loss": 0.0002,
      "step": 16849
    },
    {
      "epoch": 7.864644107351225,
      "grad_norm": 0.00982666015625,
      "learning_rate": 2.165374949794483e-05,
      "loss": 0.0001,
      "step": 16850
    },
    {
      "epoch": 7.865110851808635,
      "grad_norm": 0.0517578125,
      "learning_rate": 2.164463166612496e-05,
      "loss": 0.0027,
      "step": 16851
    },
    {
      "epoch": 7.865577596266045,
      "grad_norm": 0.0072021484375,
      "learning_rate": 2.1635515521383122e-05,
      "loss": 0.0001,
      "step": 16852
    },
    {
      "epoch": 7.8660443407234535,
      "grad_norm": 0.010009765625,
      "learning_rate": 2.1626401063915557e-05,
      "loss": 0.0001,
      "step": 16853
    },
    {
      "epoch": 7.866511085180863,
      "grad_norm": 0.009521484375,
      "learning_rate": 2.1617288293918536e-05,
      "loss": 0.0002,
      "step": 16854
    },
    {
      "epoch": 7.866977829638273,
      "grad_norm": 0.0069580078125,
      "learning_rate": 2.16081772115883e-05,
      "loss": 0.0002,
      "step": 16855
    },
    {
      "epoch": 7.867444574095683,
      "grad_norm": 0.048095703125,
      "learning_rate": 2.1599067817120945e-05,
      "loss": 0.0003,
      "step": 16856
    },
    {
      "epoch": 7.867911318553093,
      "grad_norm": 0.0262451171875,
      "learning_rate": 2.158996011071268e-05,
      "loss": 0.0002,
      "step": 16857
    },
    {
      "epoch": 7.868378063010502,
      "grad_norm": 0.0186767578125,
      "learning_rate": 2.1580854092559545e-05,
      "loss": 0.0002,
      "step": 16858
    },
    {
      "epoch": 7.868844807467911,
      "grad_norm": 0.014404296875,
      "learning_rate": 2.1571749762857652e-05,
      "loss": 0.0001,
      "step": 16859
    },
    {
      "epoch": 7.869311551925321,
      "grad_norm": 0.00970458984375,
      "learning_rate": 2.156264712180298e-05,
      "loss": 0.0001,
      "step": 16860
    },
    {
      "epoch": 7.86977829638273,
      "grad_norm": 0.01513671875,
      "learning_rate": 2.1553546169591543e-05,
      "loss": 0.0002,
      "step": 16861
    },
    {
      "epoch": 7.87024504084014,
      "grad_norm": 0.042236328125,
      "learning_rate": 2.1544446906419324e-05,
      "loss": 0.0016,
      "step": 16862
    },
    {
      "epoch": 7.87071178529755,
      "grad_norm": 0.023681640625,
      "learning_rate": 2.153534933248218e-05,
      "loss": 0.0003,
      "step": 16863
    },
    {
      "epoch": 7.8711785297549595,
      "grad_norm": 0.017822265625,
      "learning_rate": 2.1526253447976053e-05,
      "loss": 0.0002,
      "step": 16864
    },
    {
      "epoch": 7.871645274212368,
      "grad_norm": 0.01446533203125,
      "learning_rate": 2.151715925309673e-05,
      "loss": 0.0001,
      "step": 16865
    },
    {
      "epoch": 7.872112018669778,
      "grad_norm": 0.01251220703125,
      "learning_rate": 2.1508066748040046e-05,
      "loss": 0.0002,
      "step": 16866
    },
    {
      "epoch": 7.872578763127188,
      "grad_norm": 0.0064697265625,
      "learning_rate": 2.1498975933001808e-05,
      "loss": 0.0001,
      "step": 16867
    },
    {
      "epoch": 7.873045507584598,
      "grad_norm": 0.01611328125,
      "learning_rate": 2.1489886808177683e-05,
      "loss": 0.0002,
      "step": 16868
    },
    {
      "epoch": 7.873512252042007,
      "grad_norm": 0.06103515625,
      "learning_rate": 2.1480799373763395e-05,
      "loss": 0.0027,
      "step": 16869
    },
    {
      "epoch": 7.8739789964994165,
      "grad_norm": 0.00933837890625,
      "learning_rate": 2.1471713629954648e-05,
      "loss": 0.0001,
      "step": 16870
    },
    {
      "epoch": 7.874445740956826,
      "grad_norm": 0.0123291015625,
      "learning_rate": 2.146262957694701e-05,
      "loss": 0.0002,
      "step": 16871
    },
    {
      "epoch": 7.874912485414236,
      "grad_norm": 0.0322265625,
      "learning_rate": 2.1453547214936097e-05,
      "loss": 0.0003,
      "step": 16872
    },
    {
      "epoch": 7.875379229871645,
      "grad_norm": 0.0439453125,
      "learning_rate": 2.144446654411749e-05,
      "loss": 0.0026,
      "step": 16873
    },
    {
      "epoch": 7.875845974329055,
      "grad_norm": 0.01153564453125,
      "learning_rate": 2.143538756468665e-05,
      "loss": 0.0002,
      "step": 16874
    },
    {
      "epoch": 7.8763127187864646,
      "grad_norm": 0.010498046875,
      "learning_rate": 2.1426310276839077e-05,
      "loss": 0.0001,
      "step": 16875
    },
    {
      "epoch": 7.876779463243874,
      "grad_norm": 0.1025390625,
      "learning_rate": 2.1417234680770258e-05,
      "loss": 0.0004,
      "step": 16876
    },
    {
      "epoch": 7.877246207701283,
      "grad_norm": 0.0098876953125,
      "learning_rate": 2.1408160776675534e-05,
      "loss": 0.0001,
      "step": 16877
    },
    {
      "epoch": 7.877712952158693,
      "grad_norm": 0.0140380859375,
      "learning_rate": 2.1399088564750314e-05,
      "loss": 0.0002,
      "step": 16878
    },
    {
      "epoch": 7.878179696616103,
      "grad_norm": 0.017578125,
      "learning_rate": 2.1390018045189942e-05,
      "loss": 0.0002,
      "step": 16879
    },
    {
      "epoch": 7.878646441073512,
      "grad_norm": 0.005706787109375,
      "learning_rate": 2.1380949218189685e-05,
      "loss": 0.0001,
      "step": 16880
    },
    {
      "epoch": 7.8791131855309215,
      "grad_norm": 0.04931640625,
      "learning_rate": 2.137188208394484e-05,
      "loss": 0.0026,
      "step": 16881
    },
    {
      "epoch": 7.879579929988331,
      "grad_norm": 0.021240234375,
      "learning_rate": 2.1362816642650584e-05,
      "loss": 0.0002,
      "step": 16882
    },
    {
      "epoch": 7.880046674445741,
      "grad_norm": 0.015869140625,
      "learning_rate": 2.1353752894502145e-05,
      "loss": 0.0002,
      "step": 16883
    },
    {
      "epoch": 7.880513418903151,
      "grad_norm": 0.017822265625,
      "learning_rate": 2.1344690839694683e-05,
      "loss": 0.0002,
      "step": 16884
    },
    {
      "epoch": 7.88098016336056,
      "grad_norm": 0.046142578125,
      "learning_rate": 2.1335630478423275e-05,
      "loss": 0.0017,
      "step": 16885
    },
    {
      "epoch": 7.88144690781797,
      "grad_norm": 0.02294921875,
      "learning_rate": 2.1326571810883022e-05,
      "loss": 0.0002,
      "step": 16886
    },
    {
      "epoch": 7.881913652275379,
      "grad_norm": 0.0120849609375,
      "learning_rate": 2.1317514837268994e-05,
      "loss": 0.0001,
      "step": 16887
    },
    {
      "epoch": 7.882380396732788,
      "grad_norm": 0.00848388671875,
      "learning_rate": 2.130845955777615e-05,
      "loss": 0.0001,
      "step": 16888
    },
    {
      "epoch": 7.882847141190198,
      "grad_norm": 0.0072021484375,
      "learning_rate": 2.1299405972599475e-05,
      "loss": 0.0002,
      "step": 16889
    },
    {
      "epoch": 7.883313885647608,
      "grad_norm": 0.173828125,
      "learning_rate": 2.129035408193394e-05,
      "loss": 0.0004,
      "step": 16890
    },
    {
      "epoch": 7.883780630105018,
      "grad_norm": 0.010986328125,
      "learning_rate": 2.128130388597438e-05,
      "loss": 0.0002,
      "step": 16891
    },
    {
      "epoch": 7.8842473745624275,
      "grad_norm": 0.00811767578125,
      "learning_rate": 2.1272255384915685e-05,
      "loss": 0.0001,
      "step": 16892
    },
    {
      "epoch": 7.884714119019836,
      "grad_norm": 0.0634765625,
      "learning_rate": 2.126320857895271e-05,
      "loss": 0.0021,
      "step": 16893
    },
    {
      "epoch": 7.885180863477246,
      "grad_norm": 0.01226806640625,
      "learning_rate": 2.1254163468280196e-05,
      "loss": 0.0002,
      "step": 16894
    },
    {
      "epoch": 7.885647607934656,
      "grad_norm": 0.033447265625,
      "learning_rate": 2.1245120053092905e-05,
      "loss": 0.0002,
      "step": 16895
    },
    {
      "epoch": 7.886114352392065,
      "grad_norm": 0.0181884765625,
      "learning_rate": 2.1236078333585584e-05,
      "loss": 0.0002,
      "step": 16896
    },
    {
      "epoch": 7.886581096849475,
      "grad_norm": 0.00958251953125,
      "learning_rate": 2.122703830995286e-05,
      "loss": 0.0002,
      "step": 16897
    },
    {
      "epoch": 7.8870478413068845,
      "grad_norm": 0.0115966796875,
      "learning_rate": 2.1217999982389415e-05,
      "loss": 0.0001,
      "step": 16898
    },
    {
      "epoch": 7.887514585764294,
      "grad_norm": 0.01611328125,
      "learning_rate": 2.120896335108985e-05,
      "loss": 0.0002,
      "step": 16899
    },
    {
      "epoch": 7.887981330221704,
      "grad_norm": 0.01165771484375,
      "learning_rate": 2.1199928416248704e-05,
      "loss": 0.0002,
      "step": 16900
    },
    {
      "epoch": 7.888448074679113,
      "grad_norm": 0.006103515625,
      "learning_rate": 2.1190895178060556e-05,
      "loss": 0.0001,
      "step": 16901
    },
    {
      "epoch": 7.888914819136523,
      "grad_norm": 0.00701904296875,
      "learning_rate": 2.1181863636719847e-05,
      "loss": 0.0001,
      "step": 16902
    },
    {
      "epoch": 7.889381563593933,
      "grad_norm": 0.03515625,
      "learning_rate": 2.1172833792421066e-05,
      "loss": 0.002,
      "step": 16903
    },
    {
      "epoch": 7.8898483080513415,
      "grad_norm": 0.05078125,
      "learning_rate": 2.1163805645358657e-05,
      "loss": 0.0048,
      "step": 16904
    },
    {
      "epoch": 7.890315052508751,
      "grad_norm": 0.0107421875,
      "learning_rate": 2.1154779195726992e-05,
      "loss": 0.0002,
      "step": 16905
    },
    {
      "epoch": 7.890781796966161,
      "grad_norm": 0.020751953125,
      "learning_rate": 2.114575444372038e-05,
      "loss": 0.0002,
      "step": 16906
    },
    {
      "epoch": 7.891248541423571,
      "grad_norm": 0.05224609375,
      "learning_rate": 2.113673138953316e-05,
      "loss": 0.0002,
      "step": 16907
    },
    {
      "epoch": 7.89171528588098,
      "grad_norm": 0.30859375,
      "learning_rate": 2.112771003335965e-05,
      "loss": 0.0007,
      "step": 16908
    },
    {
      "epoch": 7.89218203033839,
      "grad_norm": 0.01177978515625,
      "learning_rate": 2.1118690375394012e-05,
      "loss": 0.0001,
      "step": 16909
    },
    {
      "epoch": 7.892648774795799,
      "grad_norm": 0.00921630859375,
      "learning_rate": 2.1109672415830506e-05,
      "loss": 0.0001,
      "step": 16910
    },
    {
      "epoch": 7.893115519253209,
      "grad_norm": 0.01519775390625,
      "learning_rate": 2.1100656154863297e-05,
      "loss": 0.0002,
      "step": 16911
    },
    {
      "epoch": 7.893582263710618,
      "grad_norm": 0.0281982421875,
      "learning_rate": 2.1091641592686485e-05,
      "loss": 0.0002,
      "step": 16912
    },
    {
      "epoch": 7.894049008168028,
      "grad_norm": 0.00732421875,
      "learning_rate": 2.1082628729494204e-05,
      "loss": 0.0001,
      "step": 16913
    },
    {
      "epoch": 7.894515752625438,
      "grad_norm": 0.038818359375,
      "learning_rate": 2.1073617565480453e-05,
      "loss": 0.0002,
      "step": 16914
    },
    {
      "epoch": 7.8949824970828475,
      "grad_norm": 0.041748046875,
      "learning_rate": 2.106460810083929e-05,
      "loss": 0.0003,
      "step": 16915
    },
    {
      "epoch": 7.895449241540256,
      "grad_norm": 0.03515625,
      "learning_rate": 2.1055600335764723e-05,
      "loss": 0.0027,
      "step": 16916
    },
    {
      "epoch": 7.895915985997666,
      "grad_norm": 0.009033203125,
      "learning_rate": 2.1046594270450647e-05,
      "loss": 0.0002,
      "step": 16917
    },
    {
      "epoch": 7.896382730455076,
      "grad_norm": 0.0274658203125,
      "learning_rate": 2.1037589905090994e-05,
      "loss": 0.002,
      "step": 16918
    },
    {
      "epoch": 7.896849474912486,
      "grad_norm": 0.031982421875,
      "learning_rate": 2.102858723987967e-05,
      "loss": 0.0002,
      "step": 16919
    },
    {
      "epoch": 7.897316219369895,
      "grad_norm": 0.09423828125,
      "learning_rate": 2.101958627501045e-05,
      "loss": 0.0036,
      "step": 16920
    },
    {
      "epoch": 7.8977829638273045,
      "grad_norm": 0.0308837890625,
      "learning_rate": 2.1010587010677175e-05,
      "loss": 0.0022,
      "step": 16921
    },
    {
      "epoch": 7.898249708284714,
      "grad_norm": 0.0101318359375,
      "learning_rate": 2.100158944707362e-05,
      "loss": 0.0002,
      "step": 16922
    },
    {
      "epoch": 7.898716452742123,
      "grad_norm": 0.016357421875,
      "learning_rate": 2.0992593584393472e-05,
      "loss": 0.0002,
      "step": 16923
    },
    {
      "epoch": 7.899183197199533,
      "grad_norm": 0.00634765625,
      "learning_rate": 2.0983599422830445e-05,
      "loss": 0.0001,
      "step": 16924
    },
    {
      "epoch": 7.899649941656943,
      "grad_norm": 0.043701171875,
      "learning_rate": 2.0974606962578227e-05,
      "loss": 0.0002,
      "step": 16925
    },
    {
      "epoch": 7.900116686114353,
      "grad_norm": 0.01513671875,
      "learning_rate": 2.0965616203830363e-05,
      "loss": 0.0001,
      "step": 16926
    },
    {
      "epoch": 7.900583430571762,
      "grad_norm": 0.0242919921875,
      "learning_rate": 2.095662714678047e-05,
      "loss": 0.0003,
      "step": 16927
    },
    {
      "epoch": 7.901050175029171,
      "grad_norm": 0.048828125,
      "learning_rate": 2.094763979162213e-05,
      "loss": 0.0025,
      "step": 16928
    },
    {
      "epoch": 7.901516919486581,
      "grad_norm": 0.011962890625,
      "learning_rate": 2.0938654138548797e-05,
      "loss": 0.0002,
      "step": 16929
    },
    {
      "epoch": 7.901983663943991,
      "grad_norm": 0.0093994140625,
      "learning_rate": 2.092967018775397e-05,
      "loss": 0.0001,
      "step": 16930
    },
    {
      "epoch": 7.9024504084014,
      "grad_norm": 0.018310546875,
      "learning_rate": 2.092068793943106e-05,
      "loss": 0.0002,
      "step": 16931
    },
    {
      "epoch": 7.90291715285881,
      "grad_norm": 0.01202392578125,
      "learning_rate": 2.091170739377347e-05,
      "loss": 0.0002,
      "step": 16932
    },
    {
      "epoch": 7.903383897316219,
      "grad_norm": 0.10302734375,
      "learning_rate": 2.0902728550974592e-05,
      "loss": 0.0004,
      "step": 16933
    },
    {
      "epoch": 7.903850641773629,
      "grad_norm": 0.0235595703125,
      "learning_rate": 2.0893751411227712e-05,
      "loss": 0.0002,
      "step": 16934
    },
    {
      "epoch": 7.904317386231039,
      "grad_norm": 0.1337890625,
      "learning_rate": 2.0884775974726134e-05,
      "loss": 0.0003,
      "step": 16935
    },
    {
      "epoch": 7.904784130688448,
      "grad_norm": 0.052978515625,
      "learning_rate": 2.0875802241663135e-05,
      "loss": 0.0029,
      "step": 16936
    },
    {
      "epoch": 7.905250875145858,
      "grad_norm": 0.00927734375,
      "learning_rate": 2.086683021223187e-05,
      "loss": 0.0001,
      "step": 16937
    },
    {
      "epoch": 7.9057176196032675,
      "grad_norm": 0.006134033203125,
      "learning_rate": 2.085785988662555e-05,
      "loss": 0.0001,
      "step": 16938
    },
    {
      "epoch": 7.906184364060676,
      "grad_norm": 0.011474609375,
      "learning_rate": 2.0848891265037353e-05,
      "loss": 0.0002,
      "step": 16939
    },
    {
      "epoch": 7.906651108518086,
      "grad_norm": 0.01025390625,
      "learning_rate": 2.0839924347660312e-05,
      "loss": 0.0001,
      "step": 16940
    },
    {
      "epoch": 7.907117852975496,
      "grad_norm": 0.007781982421875,
      "learning_rate": 2.0830959134687534e-05,
      "loss": 0.0001,
      "step": 16941
    },
    {
      "epoch": 7.907584597432906,
      "grad_norm": 0.012451171875,
      "learning_rate": 2.0821995626312073e-05,
      "loss": 0.0001,
      "step": 16942
    },
    {
      "epoch": 7.908051341890315,
      "grad_norm": 0.00811767578125,
      "learning_rate": 2.0813033822726858e-05,
      "loss": 0.0002,
      "step": 16943
    },
    {
      "epoch": 7.9085180863477245,
      "grad_norm": 0.00994873046875,
      "learning_rate": 2.0804073724124895e-05,
      "loss": 0.0001,
      "step": 16944
    },
    {
      "epoch": 7.908984830805134,
      "grad_norm": 0.01397705078125,
      "learning_rate": 2.0795115330699123e-05,
      "loss": 0.0001,
      "step": 16945
    },
    {
      "epoch": 7.909451575262544,
      "grad_norm": 0.028076171875,
      "learning_rate": 2.078615864264236e-05,
      "loss": 0.002,
      "step": 16946
    },
    {
      "epoch": 7.909918319719953,
      "grad_norm": 0.015380859375,
      "learning_rate": 2.0777203660147504e-05,
      "loss": 0.0002,
      "step": 16947
    },
    {
      "epoch": 7.910385064177363,
      "grad_norm": 0.00811767578125,
      "learning_rate": 2.0768250383407362e-05,
      "loss": 0.0002,
      "step": 16948
    },
    {
      "epoch": 7.9108518086347726,
      "grad_norm": 0.041259765625,
      "learning_rate": 2.075929881261468e-05,
      "loss": 0.0002,
      "step": 16949
    },
    {
      "epoch": 7.911318553092182,
      "grad_norm": 0.0106201171875,
      "learning_rate": 2.075034894796225e-05,
      "loss": 0.0002,
      "step": 16950
    },
    {
      "epoch": 7.911785297549591,
      "grad_norm": 0.01708984375,
      "learning_rate": 2.0741400789642695e-05,
      "loss": 0.0001,
      "step": 16951
    },
    {
      "epoch": 7.912252042007001,
      "grad_norm": 0.0208740234375,
      "learning_rate": 2.073245433784876e-05,
      "loss": 0.0002,
      "step": 16952
    },
    {
      "epoch": 7.912718786464411,
      "grad_norm": 0.012939453125,
      "learning_rate": 2.0723509592772993e-05,
      "loss": 0.0002,
      "step": 16953
    },
    {
      "epoch": 7.913185530921821,
      "grad_norm": 0.0194091796875,
      "learning_rate": 2.071456655460806e-05,
      "loss": 0.0002,
      "step": 16954
    },
    {
      "epoch": 7.9136522753792296,
      "grad_norm": 0.011962890625,
      "learning_rate": 2.0705625223546444e-05,
      "loss": 0.0001,
      "step": 16955
    },
    {
      "epoch": 7.914119019836639,
      "grad_norm": 0.01708984375,
      "learning_rate": 2.0696685599780696e-05,
      "loss": 0.0002,
      "step": 16956
    },
    {
      "epoch": 7.914585764294049,
      "grad_norm": 0.04443359375,
      "learning_rate": 2.0687747683503324e-05,
      "loss": 0.0014,
      "step": 16957
    },
    {
      "epoch": 7.915052508751459,
      "grad_norm": 0.04248046875,
      "learning_rate": 2.067881147490671e-05,
      "loss": 0.0002,
      "step": 16958
    },
    {
      "epoch": 7.915519253208868,
      "grad_norm": 0.00885009765625,
      "learning_rate": 2.06698769741833e-05,
      "loss": 0.0002,
      "step": 16959
    },
    {
      "epoch": 7.915985997666278,
      "grad_norm": 0.0091552734375,
      "learning_rate": 2.0660944181525487e-05,
      "loss": 0.0002,
      "step": 16960
    },
    {
      "epoch": 7.916452742123687,
      "grad_norm": 0.048095703125,
      "learning_rate": 2.0652013097125545e-05,
      "loss": 0.002,
      "step": 16961
    },
    {
      "epoch": 7.916919486581097,
      "grad_norm": 0.00750732421875,
      "learning_rate": 2.0643083721175827e-05,
      "loss": 0.0001,
      "step": 16962
    },
    {
      "epoch": 7.917386231038506,
      "grad_norm": 0.00799560546875,
      "learning_rate": 2.0634156053868537e-05,
      "loss": 0.0001,
      "step": 16963
    },
    {
      "epoch": 7.917852975495916,
      "grad_norm": 0.03564453125,
      "learning_rate": 2.062523009539592e-05,
      "loss": 0.0002,
      "step": 16964
    },
    {
      "epoch": 7.918319719953326,
      "grad_norm": 0.0101318359375,
      "learning_rate": 2.061630584595019e-05,
      "loss": 0.0002,
      "step": 16965
    },
    {
      "epoch": 7.918786464410735,
      "grad_norm": 0.0096435546875,
      "learning_rate": 2.060738330572346e-05,
      "loss": 0.0001,
      "step": 16966
    },
    {
      "epoch": 7.919253208868144,
      "grad_norm": 0.01153564453125,
      "learning_rate": 2.0598462474907842e-05,
      "loss": 0.0002,
      "step": 16967
    },
    {
      "epoch": 7.919719953325554,
      "grad_norm": 0.0103759765625,
      "learning_rate": 2.058954335369546e-05,
      "loss": 0.0002,
      "step": 16968
    },
    {
      "epoch": 7.920186697782964,
      "grad_norm": 0.0126953125,
      "learning_rate": 2.0580625942278288e-05,
      "loss": 0.0002,
      "step": 16969
    },
    {
      "epoch": 7.920653442240374,
      "grad_norm": 0.1123046875,
      "learning_rate": 2.0571710240848352e-05,
      "loss": 0.0003,
      "step": 16970
    },
    {
      "epoch": 7.921120186697783,
      "grad_norm": 0.0230712890625,
      "learning_rate": 2.056279624959766e-05,
      "loss": 0.0002,
      "step": 16971
    },
    {
      "epoch": 7.9215869311551925,
      "grad_norm": 0.008056640625,
      "learning_rate": 2.0553883968718056e-05,
      "loss": 0.0001,
      "step": 16972
    },
    {
      "epoch": 7.922053675612602,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.054497339840149e-05,
      "loss": 0.0002,
      "step": 16973
    },
    {
      "epoch": 7.922520420070011,
      "grad_norm": 0.1904296875,
      "learning_rate": 2.053606453883984e-05,
      "loss": 0.0007,
      "step": 16974
    },
    {
      "epoch": 7.922987164527421,
      "grad_norm": 0.015625,
      "learning_rate": 2.0527157390224848e-05,
      "loss": 0.0002,
      "step": 16975
    },
    {
      "epoch": 7.923453908984831,
      "grad_norm": 0.0096435546875,
      "learning_rate": 2.0518251952748337e-05,
      "loss": 0.0002,
      "step": 16976
    },
    {
      "epoch": 7.923920653442241,
      "grad_norm": 0.01129150390625,
      "learning_rate": 2.0509348226602064e-05,
      "loss": 0.0002,
      "step": 16977
    },
    {
      "epoch": 7.92438739789965,
      "grad_norm": 0.031005859375,
      "learning_rate": 2.050044621197771e-05,
      "loss": 0.0002,
      "step": 16978
    },
    {
      "epoch": 7.924854142357059,
      "grad_norm": 0.0191650390625,
      "learning_rate": 2.0491545909066945e-05,
      "loss": 0.0002,
      "step": 16979
    },
    {
      "epoch": 7.925320886814469,
      "grad_norm": 0.021728515625,
      "learning_rate": 2.0482647318061442e-05,
      "loss": 0.0002,
      "step": 16980
    },
    {
      "epoch": 7.925787631271879,
      "grad_norm": 0.052734375,
      "learning_rate": 2.047375043915274e-05,
      "loss": 0.002,
      "step": 16981
    },
    {
      "epoch": 7.926254375729288,
      "grad_norm": 0.0103759765625,
      "learning_rate": 2.0464855272532456e-05,
      "loss": 0.0001,
      "step": 16982
    },
    {
      "epoch": 7.926721120186698,
      "grad_norm": 0.01226806640625,
      "learning_rate": 2.0455961818392055e-05,
      "loss": 0.0002,
      "step": 16983
    },
    {
      "epoch": 7.927187864644107,
      "grad_norm": 0.033935546875,
      "learning_rate": 2.0447070076923057e-05,
      "loss": 0.0024,
      "step": 16984
    },
    {
      "epoch": 7.927654609101517,
      "grad_norm": 0.005279541015625,
      "learning_rate": 2.0438180048316923e-05,
      "loss": 0.0001,
      "step": 16985
    },
    {
      "epoch": 7.928121353558926,
      "grad_norm": 0.04931640625,
      "learning_rate": 2.042929173276503e-05,
      "loss": 0.0022,
      "step": 16986
    },
    {
      "epoch": 7.928588098016336,
      "grad_norm": 0.02783203125,
      "learning_rate": 2.0420405130458777e-05,
      "loss": 0.0002,
      "step": 16987
    },
    {
      "epoch": 7.929054842473746,
      "grad_norm": 0.08642578125,
      "learning_rate": 2.041152024158951e-05,
      "loss": 0.0024,
      "step": 16988
    },
    {
      "epoch": 7.9295215869311555,
      "grad_norm": 0.01080322265625,
      "learning_rate": 2.0402637066348495e-05,
      "loss": 0.0002,
      "step": 16989
    },
    {
      "epoch": 7.929988331388564,
      "grad_norm": 0.01177978515625,
      "learning_rate": 2.0393755604927022e-05,
      "loss": 0.0002,
      "step": 16990
    },
    {
      "epoch": 7.930455075845974,
      "grad_norm": 0.01190185546875,
      "learning_rate": 2.038487585751634e-05,
      "loss": 0.0002,
      "step": 16991
    },
    {
      "epoch": 7.930921820303384,
      "grad_norm": 0.0264892578125,
      "learning_rate": 2.037599782430758e-05,
      "loss": 0.0002,
      "step": 16992
    },
    {
      "epoch": 7.931388564760794,
      "grad_norm": 0.091796875,
      "learning_rate": 2.0367121505491937e-05,
      "loss": 0.0033,
      "step": 16993
    },
    {
      "epoch": 7.931855309218203,
      "grad_norm": 0.0167236328125,
      "learning_rate": 2.035824690126056e-05,
      "loss": 0.0002,
      "step": 16994
    },
    {
      "epoch": 7.9323220536756125,
      "grad_norm": 0.0084228515625,
      "learning_rate": 2.0349374011804445e-05,
      "loss": 0.0001,
      "step": 16995
    },
    {
      "epoch": 7.932788798133022,
      "grad_norm": 0.00872802734375,
      "learning_rate": 2.0340502837314703e-05,
      "loss": 0.0001,
      "step": 16996
    },
    {
      "epoch": 7.933255542590432,
      "grad_norm": 0.00787353515625,
      "learning_rate": 2.0331633377982328e-05,
      "loss": 0.0002,
      "step": 16997
    },
    {
      "epoch": 7.933722287047841,
      "grad_norm": 0.01409912109375,
      "learning_rate": 2.0322765633998265e-05,
      "loss": 0.0002,
      "step": 16998
    },
    {
      "epoch": 7.934189031505251,
      "grad_norm": 0.01275634765625,
      "learning_rate": 2.0313899605553476e-05,
      "loss": 0.0001,
      "step": 16999
    },
    {
      "epoch": 7.934655775962661,
      "grad_norm": 0.04248046875,
      "learning_rate": 2.0305035292838848e-05,
      "loss": 0.0025,
      "step": 17000
    },
    {
      "epoch": 7.93512252042007,
      "grad_norm": 0.021240234375,
      "learning_rate": 2.0296172696045212e-05,
      "loss": 0.0002,
      "step": 17001
    },
    {
      "epoch": 7.935589264877479,
      "grad_norm": 0.041015625,
      "learning_rate": 2.0287311815363397e-05,
      "loss": 0.0002,
      "step": 17002
    },
    {
      "epoch": 7.936056009334889,
      "grad_norm": 0.01239013671875,
      "learning_rate": 2.0278452650984235e-05,
      "loss": 0.0002,
      "step": 17003
    },
    {
      "epoch": 7.936522753792299,
      "grad_norm": 0.01300048828125,
      "learning_rate": 2.0269595203098422e-05,
      "loss": 0.0002,
      "step": 17004
    },
    {
      "epoch": 7.936989498249709,
      "grad_norm": 0.007293701171875,
      "learning_rate": 2.0260739471896672e-05,
      "loss": 0.0002,
      "step": 17005
    },
    {
      "epoch": 7.937456242707118,
      "grad_norm": 0.0263671875,
      "learning_rate": 2.025188545756971e-05,
      "loss": 0.0002,
      "step": 17006
    },
    {
      "epoch": 7.937922987164527,
      "grad_norm": 0.037841796875,
      "learning_rate": 2.024303316030811e-05,
      "loss": 0.0006,
      "step": 17007
    },
    {
      "epoch": 7.938389731621937,
      "grad_norm": 0.007476806640625,
      "learning_rate": 2.0234182580302497e-05,
      "loss": 0.0002,
      "step": 17008
    },
    {
      "epoch": 7.938856476079346,
      "grad_norm": 0.007110595703125,
      "learning_rate": 2.0225333717743456e-05,
      "loss": 0.0001,
      "step": 17009
    },
    {
      "epoch": 7.939323220536756,
      "grad_norm": 0.00927734375,
      "learning_rate": 2.021648657282147e-05,
      "loss": 0.0002,
      "step": 17010
    },
    {
      "epoch": 7.939789964994166,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.0207641145727074e-05,
      "loss": 0.0002,
      "step": 17011
    },
    {
      "epoch": 7.9402567094515755,
      "grad_norm": 0.009521484375,
      "learning_rate": 2.0198797436650674e-05,
      "loss": 0.0002,
      "step": 17012
    },
    {
      "epoch": 7.940723453908985,
      "grad_norm": 0.007049560546875,
      "learning_rate": 2.0189955445782695e-05,
      "loss": 0.0001,
      "step": 17013
    },
    {
      "epoch": 7.941190198366394,
      "grad_norm": 0.036865234375,
      "learning_rate": 2.018111517331356e-05,
      "loss": 0.0013,
      "step": 17014
    },
    {
      "epoch": 7.941656942823804,
      "grad_norm": 0.00787353515625,
      "learning_rate": 2.017227661943355e-05,
      "loss": 0.0001,
      "step": 17015
    },
    {
      "epoch": 7.942123687281214,
      "grad_norm": 0.01220703125,
      "learning_rate": 2.0163439784332994e-05,
      "loss": 0.0002,
      "step": 17016
    },
    {
      "epoch": 7.942590431738623,
      "grad_norm": 0.01202392578125,
      "learning_rate": 2.015460466820218e-05,
      "loss": 0.0001,
      "step": 17017
    },
    {
      "epoch": 7.9430571761960325,
      "grad_norm": 0.0238037109375,
      "learning_rate": 2.0145771271231297e-05,
      "loss": 0.0002,
      "step": 17018
    },
    {
      "epoch": 7.943523920653442,
      "grad_norm": 0.0076904296875,
      "learning_rate": 2.013693959361056e-05,
      "loss": 0.0001,
      "step": 17019
    },
    {
      "epoch": 7.943990665110852,
      "grad_norm": 0.0123291015625,
      "learning_rate": 2.0128109635530144e-05,
      "loss": 0.0001,
      "step": 17020
    },
    {
      "epoch": 7.944457409568262,
      "grad_norm": 0.0152587890625,
      "learning_rate": 2.0119281397180122e-05,
      "loss": 0.0001,
      "step": 17021
    },
    {
      "epoch": 7.944924154025671,
      "grad_norm": 0.06103515625,
      "learning_rate": 2.0110454878750605e-05,
      "loss": 0.0034,
      "step": 17022
    },
    {
      "epoch": 7.9453908984830806,
      "grad_norm": 0.0208740234375,
      "learning_rate": 2.0101630080431656e-05,
      "loss": 0.0002,
      "step": 17023
    },
    {
      "epoch": 7.94585764294049,
      "grad_norm": 0.010498046875,
      "learning_rate": 2.009280700241324e-05,
      "loss": 0.0001,
      "step": 17024
    },
    {
      "epoch": 7.946324387397899,
      "grad_norm": 0.0108642578125,
      "learning_rate": 2.0083985644885338e-05,
      "loss": 0.0002,
      "step": 17025
    },
    {
      "epoch": 7.946791131855309,
      "grad_norm": 0.005126953125,
      "learning_rate": 2.0075166008037927e-05,
      "loss": 0.0001,
      "step": 17026
    },
    {
      "epoch": 7.947257876312719,
      "grad_norm": 0.011962890625,
      "learning_rate": 2.006634809206084e-05,
      "loss": 0.0002,
      "step": 17027
    },
    {
      "epoch": 7.947724620770129,
      "grad_norm": 0.00970458984375,
      "learning_rate": 2.0057531897143966e-05,
      "loss": 0.0002,
      "step": 17028
    },
    {
      "epoch": 7.9481913652275376,
      "grad_norm": 0.01544189453125,
      "learning_rate": 2.0048717423477158e-05,
      "loss": 0.0002,
      "step": 17029
    },
    {
      "epoch": 7.948658109684947,
      "grad_norm": 0.1767578125,
      "learning_rate": 2.0039904671250143e-05,
      "loss": 0.0004,
      "step": 17030
    },
    {
      "epoch": 7.949124854142357,
      "grad_norm": 0.010009765625,
      "learning_rate": 2.0031093640652722e-05,
      "loss": 0.0002,
      "step": 17031
    },
    {
      "epoch": 7.949591598599767,
      "grad_norm": 0.046875,
      "learning_rate": 2.0022284331874552e-05,
      "loss": 0.0002,
      "step": 17032
    },
    {
      "epoch": 7.950058343057176,
      "grad_norm": 0.05517578125,
      "learning_rate": 2.001347674510534e-05,
      "loss": 0.004,
      "step": 17033
    },
    {
      "epoch": 7.950525087514586,
      "grad_norm": 0.01513671875,
      "learning_rate": 2.000467088053476e-05,
      "loss": 0.0001,
      "step": 17034
    },
    {
      "epoch": 7.950991831971995,
      "grad_norm": 0.004669189453125,
      "learning_rate": 1.999586673835233e-05,
      "loss": 0.0001,
      "step": 17035
    },
    {
      "epoch": 7.951458576429405,
      "grad_norm": 0.00653076171875,
      "learning_rate": 1.998706431874766e-05,
      "loss": 0.0001,
      "step": 17036
    },
    {
      "epoch": 7.951925320886814,
      "grad_norm": 0.0196533203125,
      "learning_rate": 1.99782636219103e-05,
      "loss": 0.0002,
      "step": 17037
    },
    {
      "epoch": 7.952392065344224,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.996946464802968e-05,
      "loss": 0.0001,
      "step": 17038
    },
    {
      "epoch": 7.952858809801634,
      "grad_norm": 0.0294189453125,
      "learning_rate": 1.9960667397295285e-05,
      "loss": 0.0018,
      "step": 17039
    },
    {
      "epoch": 7.9533255542590435,
      "grad_norm": 0.1103515625,
      "learning_rate": 1.9951871869896544e-05,
      "loss": 0.0004,
      "step": 17040
    },
    {
      "epoch": 7.953792298716452,
      "grad_norm": 0.010498046875,
      "learning_rate": 1.99430780660228e-05,
      "loss": 0.0002,
      "step": 17041
    },
    {
      "epoch": 7.954259043173862,
      "grad_norm": 0.0419921875,
      "learning_rate": 1.993428598586341e-05,
      "loss": 0.0002,
      "step": 17042
    },
    {
      "epoch": 7.954725787631272,
      "grad_norm": 0.048095703125,
      "learning_rate": 1.9925495629607695e-05,
      "loss": 0.0002,
      "step": 17043
    },
    {
      "epoch": 7.955192532088682,
      "grad_norm": 0.012451171875,
      "learning_rate": 1.991670699744489e-05,
      "loss": 0.0002,
      "step": 17044
    },
    {
      "epoch": 7.955659276546091,
      "grad_norm": 0.006927490234375,
      "learning_rate": 1.9907920089564225e-05,
      "loss": 0.0001,
      "step": 17045
    },
    {
      "epoch": 7.9561260210035005,
      "grad_norm": 0.004364013671875,
      "learning_rate": 1.9899134906154936e-05,
      "loss": 0.0001,
      "step": 17046
    },
    {
      "epoch": 7.95659276546091,
      "grad_norm": 0.01153564453125,
      "learning_rate": 1.989035144740612e-05,
      "loss": 0.0002,
      "step": 17047
    },
    {
      "epoch": 7.95705950991832,
      "grad_norm": 0.020263671875,
      "learning_rate": 1.9881569713506943e-05,
      "loss": 0.0002,
      "step": 17048
    },
    {
      "epoch": 7.957526254375729,
      "grad_norm": 0.01422119140625,
      "learning_rate": 1.9872789704646465e-05,
      "loss": 0.0001,
      "step": 17049
    },
    {
      "epoch": 7.957992998833139,
      "grad_norm": 0.006072998046875,
      "learning_rate": 1.9864011421013696e-05,
      "loss": 0.0001,
      "step": 17050
    },
    {
      "epoch": 7.958459743290549,
      "grad_norm": 0.04248046875,
      "learning_rate": 1.985523486279768e-05,
      "loss": 0.0023,
      "step": 17051
    },
    {
      "epoch": 7.9589264877479575,
      "grad_norm": 0.0517578125,
      "learning_rate": 1.9846460030187408e-05,
      "loss": 0.0016,
      "step": 17052
    },
    {
      "epoch": 7.959393232205367,
      "grad_norm": 0.0098876953125,
      "learning_rate": 1.9837686923371758e-05,
      "loss": 0.0001,
      "step": 17053
    },
    {
      "epoch": 7.959859976662777,
      "grad_norm": 0.00909423828125,
      "learning_rate": 1.982891554253964e-05,
      "loss": 0.0002,
      "step": 17054
    },
    {
      "epoch": 7.960326721120187,
      "grad_norm": 0.07421875,
      "learning_rate": 1.982014588787996e-05,
      "loss": 0.003,
      "step": 17055
    },
    {
      "epoch": 7.960793465577597,
      "grad_norm": 0.072265625,
      "learning_rate": 1.9811377959581466e-05,
      "loss": 0.0041,
      "step": 17056
    },
    {
      "epoch": 7.961260210035006,
      "grad_norm": 0.005889892578125,
      "learning_rate": 1.9802611757832977e-05,
      "loss": 0.0001,
      "step": 17057
    },
    {
      "epoch": 7.961726954492415,
      "grad_norm": 0.00811767578125,
      "learning_rate": 1.9793847282823263e-05,
      "loss": 0.0001,
      "step": 17058
    },
    {
      "epoch": 7.962193698949825,
      "grad_norm": 0.02099609375,
      "learning_rate": 1.9785084534740984e-05,
      "loss": 0.0002,
      "step": 17059
    },
    {
      "epoch": 7.962660443407234,
      "grad_norm": 0.0113525390625,
      "learning_rate": 1.9776323513774853e-05,
      "loss": 0.0002,
      "step": 17060
    },
    {
      "epoch": 7.963127187864644,
      "grad_norm": 0.0673828125,
      "learning_rate": 1.9767564220113466e-05,
      "loss": 0.0003,
      "step": 17061
    },
    {
      "epoch": 7.963593932322054,
      "grad_norm": 0.00994873046875,
      "learning_rate": 1.975880665394544e-05,
      "loss": 0.0002,
      "step": 17062
    },
    {
      "epoch": 7.9640606767794635,
      "grad_norm": 0.053955078125,
      "learning_rate": 1.9750050815459363e-05,
      "loss": 0.0031,
      "step": 17063
    },
    {
      "epoch": 7.964527421236873,
      "grad_norm": 0.01080322265625,
      "learning_rate": 1.9741296704843704e-05,
      "loss": 0.0002,
      "step": 17064
    },
    {
      "epoch": 7.964994165694282,
      "grad_norm": 0.016845703125,
      "learning_rate": 1.973254432228698e-05,
      "loss": 0.0002,
      "step": 17065
    },
    {
      "epoch": 7.965460910151692,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.9723793667977653e-05,
      "loss": 0.0001,
      "step": 17066
    },
    {
      "epoch": 7.965927654609102,
      "grad_norm": 0.01458740234375,
      "learning_rate": 1.9715044742104095e-05,
      "loss": 0.0002,
      "step": 17067
    },
    {
      "epoch": 7.966394399066511,
      "grad_norm": 0.003753662109375,
      "learning_rate": 1.9706297544854702e-05,
      "loss": 0.0001,
      "step": 17068
    },
    {
      "epoch": 7.9668611435239205,
      "grad_norm": 0.01556396484375,
      "learning_rate": 1.9697552076417837e-05,
      "loss": 0.0002,
      "step": 17069
    },
    {
      "epoch": 7.96732788798133,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.9688808336981747e-05,
      "loss": 0.0001,
      "step": 17070
    },
    {
      "epoch": 7.96779463243874,
      "grad_norm": 0.018798828125,
      "learning_rate": 1.9680066326734715e-05,
      "loss": 0.0002,
      "step": 17071
    },
    {
      "epoch": 7.968261376896149,
      "grad_norm": 0.03271484375,
      "learning_rate": 1.9671326045865e-05,
      "loss": 0.0017,
      "step": 17072
    },
    {
      "epoch": 7.968728121353559,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.9662587494560737e-05,
      "loss": 0.0001,
      "step": 17073
    },
    {
      "epoch": 7.969194865810969,
      "grad_norm": 0.0147705078125,
      "learning_rate": 1.9653850673010098e-05,
      "loss": 0.0002,
      "step": 17074
    },
    {
      "epoch": 7.969661610268378,
      "grad_norm": 0.03662109375,
      "learning_rate": 1.964511558140123e-05,
      "loss": 0.0003,
      "step": 17075
    },
    {
      "epoch": 7.970128354725787,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.9636382219922155e-05,
      "loss": 0.0001,
      "step": 17076
    },
    {
      "epoch": 7.970595099183197,
      "grad_norm": 0.080078125,
      "learning_rate": 1.9627650588760928e-05,
      "loss": 0.0002,
      "step": 17077
    },
    {
      "epoch": 7.971061843640607,
      "grad_norm": 0.0537109375,
      "learning_rate": 1.961892068810559e-05,
      "loss": 0.0028,
      "step": 17078
    },
    {
      "epoch": 7.971528588098017,
      "grad_norm": 0.0123291015625,
      "learning_rate": 1.9610192518144042e-05,
      "loss": 0.0002,
      "step": 17079
    },
    {
      "epoch": 7.971995332555426,
      "grad_norm": 0.0230712890625,
      "learning_rate": 1.960146607906428e-05,
      "loss": 0.0001,
      "step": 17080
    },
    {
      "epoch": 7.972462077012835,
      "grad_norm": 0.2412109375,
      "learning_rate": 1.9592741371054124e-05,
      "loss": 0.0005,
      "step": 17081
    },
    {
      "epoch": 7.972928821470245,
      "grad_norm": 0.046142578125,
      "learning_rate": 1.9584018394301464e-05,
      "loss": 0.0003,
      "step": 17082
    },
    {
      "epoch": 7.973395565927655,
      "grad_norm": 0.009765625,
      "learning_rate": 1.957529714899412e-05,
      "loss": 0.0002,
      "step": 17083
    },
    {
      "epoch": 7.973862310385064,
      "grad_norm": 0.0125732421875,
      "learning_rate": 1.956657763531985e-05,
      "loss": 0.0002,
      "step": 17084
    },
    {
      "epoch": 7.974329054842474,
      "grad_norm": 0.00982666015625,
      "learning_rate": 1.955785985346641e-05,
      "loss": 0.0001,
      "step": 17085
    },
    {
      "epoch": 7.9747957992998835,
      "grad_norm": 0.00933837890625,
      "learning_rate": 1.9549143803621516e-05,
      "loss": 0.0002,
      "step": 17086
    },
    {
      "epoch": 7.975262543757293,
      "grad_norm": 0.007110595703125,
      "learning_rate": 1.95404294859728e-05,
      "loss": 0.0001,
      "step": 17087
    },
    {
      "epoch": 7.975729288214702,
      "grad_norm": 0.036865234375,
      "learning_rate": 1.9531716900707896e-05,
      "loss": 0.0022,
      "step": 17088
    },
    {
      "epoch": 7.976196032672112,
      "grad_norm": 0.01953125,
      "learning_rate": 1.9523006048014448e-05,
      "loss": 0.0002,
      "step": 17089
    },
    {
      "epoch": 7.976662777129522,
      "grad_norm": 0.0045166015625,
      "learning_rate": 1.9514296928079932e-05,
      "loss": 0.0001,
      "step": 17090
    },
    {
      "epoch": 7.977129521586932,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.9505589541091916e-05,
      "loss": 0.0002,
      "step": 17091
    },
    {
      "epoch": 7.9775962660443405,
      "grad_norm": 0.007659912109375,
      "learning_rate": 1.949688388723788e-05,
      "loss": 0.0001,
      "step": 17092
    },
    {
      "epoch": 7.97806301050175,
      "grad_norm": 0.0048828125,
      "learning_rate": 1.9488179966705245e-05,
      "loss": 0.0001,
      "step": 17093
    },
    {
      "epoch": 7.97852975495916,
      "grad_norm": 0.0703125,
      "learning_rate": 1.947947777968141e-05,
      "loss": 0.0022,
      "step": 17094
    },
    {
      "epoch": 7.978996499416569,
      "grad_norm": 0.010498046875,
      "learning_rate": 1.9470777326353816e-05,
      "loss": 0.0002,
      "step": 17095
    },
    {
      "epoch": 7.979463243873979,
      "grad_norm": 0.06494140625,
      "learning_rate": 1.9462078606909686e-05,
      "loss": 0.0014,
      "step": 17096
    },
    {
      "epoch": 7.979929988331389,
      "grad_norm": 0.060302734375,
      "learning_rate": 1.9453381621536348e-05,
      "loss": 0.0024,
      "step": 17097
    },
    {
      "epoch": 7.980396732788798,
      "grad_norm": 0.004302978515625,
      "learning_rate": 1.9444686370421107e-05,
      "loss": 0.0001,
      "step": 17098
    },
    {
      "epoch": 7.980863477246208,
      "grad_norm": 0.017822265625,
      "learning_rate": 1.9435992853751128e-05,
      "loss": 0.0002,
      "step": 17099
    },
    {
      "epoch": 7.981330221703617,
      "grad_norm": 0.0196533203125,
      "learning_rate": 1.94273010717136e-05,
      "loss": 0.0002,
      "step": 17100
    },
    {
      "epoch": 7.981796966161027,
      "grad_norm": 0.04248046875,
      "learning_rate": 1.9418611024495702e-05,
      "loss": 0.0002,
      "step": 17101
    },
    {
      "epoch": 7.982263710618437,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.940992271228449e-05,
      "loss": 0.0001,
      "step": 17102
    },
    {
      "epoch": 7.9827304550758456,
      "grad_norm": 0.0390625,
      "learning_rate": 1.9401236135267064e-05,
      "loss": 0.0019,
      "step": 17103
    },
    {
      "epoch": 7.983197199533255,
      "grad_norm": 0.015625,
      "learning_rate": 1.9392551293630467e-05,
      "loss": 0.0002,
      "step": 17104
    },
    {
      "epoch": 7.983663943990665,
      "grad_norm": 0.06396484375,
      "learning_rate": 1.938386818756166e-05,
      "loss": 0.0045,
      "step": 17105
    },
    {
      "epoch": 7.984130688448075,
      "grad_norm": 0.00927734375,
      "learning_rate": 1.9375186817247615e-05,
      "loss": 0.0002,
      "step": 17106
    },
    {
      "epoch": 7.984597432905485,
      "grad_norm": 0.0286865234375,
      "learning_rate": 1.936650718287527e-05,
      "loss": 0.0002,
      "step": 17107
    },
    {
      "epoch": 7.985064177362894,
      "grad_norm": 0.0546875,
      "learning_rate": 1.935782928463147e-05,
      "loss": 0.0024,
      "step": 17108
    },
    {
      "epoch": 7.985530921820303,
      "grad_norm": 0.1298828125,
      "learning_rate": 1.93491531227031e-05,
      "loss": 0.0003,
      "step": 17109
    },
    {
      "epoch": 7.985997666277713,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.934047869727693e-05,
      "loss": 0.0002,
      "step": 17110
    },
    {
      "epoch": 7.986464410735122,
      "grad_norm": 0.01397705078125,
      "learning_rate": 1.9331806008539742e-05,
      "loss": 0.0002,
      "step": 17111
    },
    {
      "epoch": 7.986931155192532,
      "grad_norm": 0.05029296875,
      "learning_rate": 1.93231350566783e-05,
      "loss": 0.0017,
      "step": 17112
    },
    {
      "epoch": 7.987397899649942,
      "grad_norm": 0.0067138671875,
      "learning_rate": 1.9314465841879247e-05,
      "loss": 0.0001,
      "step": 17113
    },
    {
      "epoch": 7.9878646441073515,
      "grad_norm": 0.007476806640625,
      "learning_rate": 1.9305798364329262e-05,
      "loss": 0.0001,
      "step": 17114
    },
    {
      "epoch": 7.98833138856476,
      "grad_norm": 0.0059814453125,
      "learning_rate": 1.9297132624215007e-05,
      "loss": 0.0001,
      "step": 17115
    },
    {
      "epoch": 7.98879813302217,
      "grad_norm": 0.01007080078125,
      "learning_rate": 1.928846862172299e-05,
      "loss": 0.0002,
      "step": 17116
    },
    {
      "epoch": 7.98926487747958,
      "grad_norm": 0.048095703125,
      "learning_rate": 1.9279806357039798e-05,
      "loss": 0.0006,
      "step": 17117
    },
    {
      "epoch": 7.98973162193699,
      "grad_norm": 0.0272216796875,
      "learning_rate": 1.9271145830351967e-05,
      "loss": 0.0002,
      "step": 17118
    },
    {
      "epoch": 7.990198366394399,
      "grad_norm": 0.099609375,
      "learning_rate": 1.9262487041845913e-05,
      "loss": 0.007,
      "step": 17119
    },
    {
      "epoch": 7.9906651108518085,
      "grad_norm": 0.029052734375,
      "learning_rate": 1.9253829991708083e-05,
      "loss": 0.0019,
      "step": 17120
    },
    {
      "epoch": 7.991131855309218,
      "grad_norm": 0.037109375,
      "learning_rate": 1.924517468012491e-05,
      "loss": 0.0027,
      "step": 17121
    },
    {
      "epoch": 7.991598599766628,
      "grad_norm": 0.058837890625,
      "learning_rate": 1.92365211072827e-05,
      "loss": 0.0024,
      "step": 17122
    },
    {
      "epoch": 7.992065344224037,
      "grad_norm": 0.0201416015625,
      "learning_rate": 1.9227869273367803e-05,
      "loss": 0.0002,
      "step": 17123
    },
    {
      "epoch": 7.992532088681447,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.9219219178566518e-05,
      "loss": 0.0002,
      "step": 17124
    },
    {
      "epoch": 7.992998833138857,
      "grad_norm": 0.007110595703125,
      "learning_rate": 1.9210570823065056e-05,
      "loss": 0.0002,
      "step": 17125
    },
    {
      "epoch": 7.993465577596266,
      "grad_norm": 0.050537109375,
      "learning_rate": 1.9201924207049637e-05,
      "loss": 0.0034,
      "step": 17126
    },
    {
      "epoch": 7.993932322053675,
      "grad_norm": 0.427734375,
      "learning_rate": 1.919327933070646e-05,
      "loss": 0.0017,
      "step": 17127
    },
    {
      "epoch": 7.994399066511085,
      "grad_norm": 0.279296875,
      "learning_rate": 1.9184636194221607e-05,
      "loss": 0.0012,
      "step": 17128
    },
    {
      "epoch": 7.994865810968495,
      "grad_norm": 0.010009765625,
      "learning_rate": 1.917599479778124e-05,
      "loss": 0.0001,
      "step": 17129
    },
    {
      "epoch": 7.995332555425904,
      "grad_norm": 0.01171875,
      "learning_rate": 1.916735514157134e-05,
      "loss": 0.0002,
      "step": 17130
    },
    {
      "epoch": 7.995799299883314,
      "grad_norm": 0.0240478515625,
      "learning_rate": 1.9158717225777977e-05,
      "loss": 0.0002,
      "step": 17131
    },
    {
      "epoch": 7.996266044340723,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.9150081050587155e-05,
      "loss": 0.0001,
      "step": 17132
    },
    {
      "epoch": 7.996732788798133,
      "grad_norm": 0.00738525390625,
      "learning_rate": 1.9141446616184765e-05,
      "loss": 0.0001,
      "step": 17133
    },
    {
      "epoch": 7.997199533255543,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.9132813922756744e-05,
      "loss": 0.0001,
      "step": 17134
    },
    {
      "epoch": 7.997666277712952,
      "grad_norm": 0.006439208984375,
      "learning_rate": 1.9124182970488992e-05,
      "loss": 0.0001,
      "step": 17135
    },
    {
      "epoch": 7.998133022170362,
      "grad_norm": 0.05078125,
      "learning_rate": 1.9115553759567283e-05,
      "loss": 0.0015,
      "step": 17136
    },
    {
      "epoch": 7.9985997666277715,
      "grad_norm": 0.2265625,
      "learning_rate": 1.910692629017745e-05,
      "loss": 0.0008,
      "step": 17137
    },
    {
      "epoch": 7.99906651108518,
      "grad_norm": 0.01953125,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 0.0002,
      "step": 17138
    },
    {
      "epoch": 7.99953325554259,
      "grad_norm": 0.011962890625,
      "learning_rate": 1.9089676576736405e-05,
      "loss": 0.0002,
      "step": 17139
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.03662109375,
      "learning_rate": 1.908105433305658e-05,
      "loss": 0.0029,
      "step": 17140
    },
    {
      "epoch": 8.000466744457409,
      "grad_norm": 0.2734375,
      "learning_rate": 1.907243383165147e-05,
      "loss": 0.0005,
      "step": 17141
    },
    {
      "epoch": 8.00093348891482,
      "grad_norm": 0.004730224609375,
      "learning_rate": 1.9063815072706614e-05,
      "loss": 0.0001,
      "step": 17142
    },
    {
      "epoch": 8.001400233372228,
      "grad_norm": 0.038818359375,
      "learning_rate": 1.9055198056407652e-05,
      "loss": 0.0002,
      "step": 17143
    },
    {
      "epoch": 8.001866977829637,
      "grad_norm": 0.006011962890625,
      "learning_rate": 1.904658278294009e-05,
      "loss": 0.0001,
      "step": 17144
    },
    {
      "epoch": 8.002333722287048,
      "grad_norm": 0.049072265625,
      "learning_rate": 1.9037969252489386e-05,
      "loss": 0.0024,
      "step": 17145
    },
    {
      "epoch": 8.002800466744457,
      "grad_norm": 0.005523681640625,
      "learning_rate": 1.902935746524104e-05,
      "loss": 0.0001,
      "step": 17146
    },
    {
      "epoch": 8.003267211201868,
      "grad_norm": 0.0458984375,
      "learning_rate": 1.902074742138049e-05,
      "loss": 0.0023,
      "step": 17147
    },
    {
      "epoch": 8.003733955659277,
      "grad_norm": 0.007080078125,
      "learning_rate": 1.9012139121093064e-05,
      "loss": 0.0001,
      "step": 17148
    },
    {
      "epoch": 8.004200700116685,
      "grad_norm": 0.00579833984375,
      "learning_rate": 1.9003532564564142e-05,
      "loss": 0.0001,
      "step": 17149
    },
    {
      "epoch": 8.004667444574096,
      "grad_norm": 0.005279541015625,
      "learning_rate": 1.8994927751979063e-05,
      "loss": 0.0001,
      "step": 17150
    },
    {
      "epoch": 8.005134189031505,
      "grad_norm": 0.008056640625,
      "learning_rate": 1.898632468352304e-05,
      "loss": 0.0001,
      "step": 17151
    },
    {
      "epoch": 8.005600933488914,
      "grad_norm": 0.004119873046875,
      "learning_rate": 1.8977723359381327e-05,
      "loss": 0.0001,
      "step": 17152
    },
    {
      "epoch": 8.005600933488914,
      "eval_loss": 2.4350552558898926,
      "eval_runtime": 84.5635,
      "eval_samples_per_second": 21.333,
      "eval_steps_per_second": 2.673,
      "step": 17152
    },
    {
      "epoch": 8.006067677946325,
      "grad_norm": 0.0390625,
      "learning_rate": 1.896912377973915e-05,
      "loss": 0.002,
      "step": 17153
    },
    {
      "epoch": 8.006534422403734,
      "grad_norm": 0.01068115234375,
      "learning_rate": 1.8960525944781628e-05,
      "loss": 0.0002,
      "step": 17154
    },
    {
      "epoch": 8.007001166861144,
      "grad_norm": 0.01092529296875,
      "learning_rate": 1.8951929854693882e-05,
      "loss": 0.0002,
      "step": 17155
    },
    {
      "epoch": 8.007467911318553,
      "grad_norm": 0.025146484375,
      "learning_rate": 1.8943335509661054e-05,
      "loss": 0.0002,
      "step": 17156
    },
    {
      "epoch": 8.007934655775962,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.893474290986811e-05,
      "loss": 0.0028,
      "step": 17157
    },
    {
      "epoch": 8.008401400233373,
      "grad_norm": 0.00848388671875,
      "learning_rate": 1.892615205550009e-05,
      "loss": 0.0001,
      "step": 17158
    },
    {
      "epoch": 8.008868144690782,
      "grad_norm": 0.006317138671875,
      "learning_rate": 1.8917562946742006e-05,
      "loss": 0.0001,
      "step": 17159
    },
    {
      "epoch": 8.00933488914819,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.8908975583778722e-05,
      "loss": 0.0002,
      "step": 17160
    },
    {
      "epoch": 8.009801633605601,
      "grad_norm": 0.0272216796875,
      "learning_rate": 1.8900389966795196e-05,
      "loss": 0.0002,
      "step": 17161
    },
    {
      "epoch": 8.01026837806301,
      "grad_norm": 0.01190185546875,
      "learning_rate": 1.8891806095976227e-05,
      "loss": 0.0002,
      "step": 17162
    },
    {
      "epoch": 8.01073512252042,
      "grad_norm": 0.038818359375,
      "learning_rate": 1.8883223971506657e-05,
      "loss": 0.0013,
      "step": 17163
    },
    {
      "epoch": 8.01120186697783,
      "grad_norm": 0.01275634765625,
      "learning_rate": 1.8874643593571316e-05,
      "loss": 0.0002,
      "step": 17164
    },
    {
      "epoch": 8.011668611435239,
      "grad_norm": 0.02099609375,
      "learning_rate": 1.8866064962354877e-05,
      "loss": 0.0001,
      "step": 17165
    },
    {
      "epoch": 8.01213535589265,
      "grad_norm": 0.01446533203125,
      "learning_rate": 1.8857488078042073e-05,
      "loss": 0.0002,
      "step": 17166
    },
    {
      "epoch": 8.012602100350058,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.884891294081761e-05,
      "loss": 0.0002,
      "step": 17167
    },
    {
      "epoch": 8.013068844807467,
      "grad_norm": 0.0380859375,
      "learning_rate": 1.884033955086607e-05,
      "loss": 0.0017,
      "step": 17168
    },
    {
      "epoch": 8.013535589264878,
      "grad_norm": 0.0115966796875,
      "learning_rate": 1.8831767908372056e-05,
      "loss": 0.0002,
      "step": 17169
    },
    {
      "epoch": 8.014002333722287,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.882319801352017e-05,
      "loss": 0.0002,
      "step": 17170
    },
    {
      "epoch": 8.014469078179697,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.881462986649487e-05,
      "loss": 0.0002,
      "step": 17171
    },
    {
      "epoch": 8.014935822637106,
      "grad_norm": 0.043212890625,
      "learning_rate": 1.880606346748066e-05,
      "loss": 0.0022,
      "step": 17172
    },
    {
      "epoch": 8.015402567094515,
      "grad_norm": 0.09619140625,
      "learning_rate": 1.879749881666203e-05,
      "loss": 0.0046,
      "step": 17173
    },
    {
      "epoch": 8.015869311551926,
      "grad_norm": 0.00555419921875,
      "learning_rate": 1.8788935914223305e-05,
      "loss": 0.0001,
      "step": 17174
    },
    {
      "epoch": 8.016336056009335,
      "grad_norm": 0.005401611328125,
      "learning_rate": 1.8780374760348906e-05,
      "loss": 0.0002,
      "step": 17175
    },
    {
      "epoch": 8.016802800466744,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.8771815355223178e-05,
      "loss": 0.0002,
      "step": 17176
    },
    {
      "epoch": 8.017269544924154,
      "grad_norm": 0.005401611328125,
      "learning_rate": 1.8763257699030358e-05,
      "loss": 0.0001,
      "step": 17177
    },
    {
      "epoch": 8.017736289381563,
      "grad_norm": 0.007293701171875,
      "learning_rate": 1.8754701791954767e-05,
      "loss": 0.0001,
      "step": 17178
    },
    {
      "epoch": 8.018203033838974,
      "grad_norm": 0.0125732421875,
      "learning_rate": 1.8746147634180556e-05,
      "loss": 0.0002,
      "step": 17179
    },
    {
      "epoch": 8.018669778296383,
      "grad_norm": 0.04052734375,
      "learning_rate": 1.8737595225891945e-05,
      "loss": 0.0018,
      "step": 17180
    },
    {
      "epoch": 8.019136522753792,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.872904456727309e-05,
      "loss": 0.0002,
      "step": 17181
    },
    {
      "epoch": 8.019603267211203,
      "grad_norm": 0.058349609375,
      "learning_rate": 1.8720495658508053e-05,
      "loss": 0.005,
      "step": 17182
    },
    {
      "epoch": 8.020070011668611,
      "grad_norm": 0.00848388671875,
      "learning_rate": 1.871194849978093e-05,
      "loss": 0.0001,
      "step": 17183
    },
    {
      "epoch": 8.02053675612602,
      "grad_norm": 0.7109375,
      "learning_rate": 1.8703403091275772e-05,
      "loss": 0.006,
      "step": 17184
    },
    {
      "epoch": 8.021003500583431,
      "grad_norm": 0.01348876953125,
      "learning_rate": 1.8694859433176516e-05,
      "loss": 0.0002,
      "step": 17185
    },
    {
      "epoch": 8.02147024504084,
      "grad_norm": 0.0595703125,
      "learning_rate": 1.868631752566715e-05,
      "loss": 0.0002,
      "step": 17186
    },
    {
      "epoch": 8.021936989498249,
      "grad_norm": 0.014404296875,
      "learning_rate": 1.8677777368931603e-05,
      "loss": 0.0002,
      "step": 17187
    },
    {
      "epoch": 8.02240373395566,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.8669238963153722e-05,
      "loss": 0.0001,
      "step": 17188
    },
    {
      "epoch": 8.022870478413068,
      "grad_norm": 0.01953125,
      "learning_rate": 1.8660702308517363e-05,
      "loss": 0.0005,
      "step": 17189
    },
    {
      "epoch": 8.02333722287048,
      "grad_norm": 0.01263427734375,
      "learning_rate": 1.8652167405206356e-05,
      "loss": 0.0002,
      "step": 17190
    },
    {
      "epoch": 8.023803967327888,
      "grad_norm": 0.0091552734375,
      "learning_rate": 1.864363425340444e-05,
      "loss": 0.0002,
      "step": 17191
    },
    {
      "epoch": 8.024270711785297,
      "grad_norm": 0.01153564453125,
      "learning_rate": 1.863510285329533e-05,
      "loss": 0.0001,
      "step": 17192
    },
    {
      "epoch": 8.024737456242708,
      "grad_norm": 0.054443359375,
      "learning_rate": 1.8626573205062738e-05,
      "loss": 0.0029,
      "step": 17193
    },
    {
      "epoch": 8.025204200700117,
      "grad_norm": 0.0250244140625,
      "learning_rate": 1.8618045308890297e-05,
      "loss": 0.0002,
      "step": 17194
    },
    {
      "epoch": 8.025670945157525,
      "grad_norm": 0.05029296875,
      "learning_rate": 1.8609519164961632e-05,
      "loss": 0.002,
      "step": 17195
    },
    {
      "epoch": 8.026137689614936,
      "grad_norm": 0.006072998046875,
      "learning_rate": 1.8600994773460344e-05,
      "loss": 0.0001,
      "step": 17196
    },
    {
      "epoch": 8.026604434072345,
      "grad_norm": 0.0322265625,
      "learning_rate": 1.859247213456994e-05,
      "loss": 0.0002,
      "step": 17197
    },
    {
      "epoch": 8.027071178529756,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.858395124847393e-05,
      "loss": 0.0002,
      "step": 17198
    },
    {
      "epoch": 8.027537922987165,
      "grad_norm": 0.515625,
      "learning_rate": 1.8575432115355797e-05,
      "loss": 0.0022,
      "step": 17199
    },
    {
      "epoch": 8.028004667444574,
      "grad_norm": 0.064453125,
      "learning_rate": 1.8566914735398933e-05,
      "loss": 0.0002,
      "step": 17200
    },
    {
      "epoch": 8.028471411901984,
      "grad_norm": 0.034423828125,
      "learning_rate": 1.855839910878675e-05,
      "loss": 0.0002,
      "step": 17201
    },
    {
      "epoch": 8.028938156359393,
      "grad_norm": 0.0159912109375,
      "learning_rate": 1.8549885235702624e-05,
      "loss": 0.0002,
      "step": 17202
    },
    {
      "epoch": 8.029404900816802,
      "grad_norm": 0.00677490234375,
      "learning_rate": 1.854137311632981e-05,
      "loss": 0.0001,
      "step": 17203
    },
    {
      "epoch": 8.029871645274213,
      "grad_norm": 0.01190185546875,
      "learning_rate": 1.853286275085161e-05,
      "loss": 0.0001,
      "step": 17204
    },
    {
      "epoch": 8.030338389731622,
      "grad_norm": 0.01043701171875,
      "learning_rate": 1.8524354139451282e-05,
      "loss": 0.0001,
      "step": 17205
    },
    {
      "epoch": 8.030805134189032,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.8515847282311994e-05,
      "loss": 0.0001,
      "step": 17206
    },
    {
      "epoch": 8.031271878646441,
      "grad_norm": 0.0235595703125,
      "learning_rate": 1.8507342179616928e-05,
      "loss": 0.0002,
      "step": 17207
    },
    {
      "epoch": 8.03173862310385,
      "grad_norm": 0.00799560546875,
      "learning_rate": 1.8498838831549216e-05,
      "loss": 0.0001,
      "step": 17208
    },
    {
      "epoch": 8.03220536756126,
      "grad_norm": 0.0198974609375,
      "learning_rate": 1.849033723829191e-05,
      "loss": 0.0002,
      "step": 17209
    },
    {
      "epoch": 8.03267211201867,
      "grad_norm": 0.00823974609375,
      "learning_rate": 1.8481837400028102e-05,
      "loss": 0.0002,
      "step": 17210
    },
    {
      "epoch": 8.033138856476079,
      "grad_norm": 0.01239013671875,
      "learning_rate": 1.8473339316940753e-05,
      "loss": 0.0002,
      "step": 17211
    },
    {
      "epoch": 8.03360560093349,
      "grad_norm": 0.03173828125,
      "learning_rate": 1.846484298921287e-05,
      "loss": 0.0002,
      "step": 17212
    },
    {
      "epoch": 8.034072345390898,
      "grad_norm": 0.0096435546875,
      "learning_rate": 1.8456348417027415e-05,
      "loss": 0.0002,
      "step": 17213
    },
    {
      "epoch": 8.034539089848309,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.844785560056721e-05,
      "loss": 0.0001,
      "step": 17214
    },
    {
      "epoch": 8.035005834305718,
      "grad_norm": 0.06689453125,
      "learning_rate": 1.843936454001517e-05,
      "loss": 0.0022,
      "step": 17215
    },
    {
      "epoch": 8.035472578763127,
      "grad_norm": 0.0162353515625,
      "learning_rate": 1.8430875235554125e-05,
      "loss": 0.0002,
      "step": 17216
    },
    {
      "epoch": 8.035939323220537,
      "grad_norm": 0.010009765625,
      "learning_rate": 1.8422387687366816e-05,
      "loss": 0.0001,
      "step": 17217
    },
    {
      "epoch": 8.036406067677946,
      "grad_norm": 0.00579833984375,
      "learning_rate": 1.841390189563601e-05,
      "loss": 0.0001,
      "step": 17218
    },
    {
      "epoch": 8.036872812135355,
      "grad_norm": 0.01397705078125,
      "learning_rate": 1.840541786054445e-05,
      "loss": 0.0001,
      "step": 17219
    },
    {
      "epoch": 8.037339556592766,
      "grad_norm": 0.0189208984375,
      "learning_rate": 1.8396935582274733e-05,
      "loss": 0.0002,
      "step": 17220
    },
    {
      "epoch": 8.037806301050175,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.8388455061009546e-05,
      "loss": 0.0002,
      "step": 17221
    },
    {
      "epoch": 8.038273045507585,
      "grad_norm": 0.01202392578125,
      "learning_rate": 1.837997629693149e-05,
      "loss": 0.0002,
      "step": 17222
    },
    {
      "epoch": 8.038739789964994,
      "grad_norm": 0.03857421875,
      "learning_rate": 1.8371499290223082e-05,
      "loss": 0.0026,
      "step": 17223
    },
    {
      "epoch": 8.039206534422403,
      "grad_norm": 0.005279541015625,
      "learning_rate": 1.8363024041066867e-05,
      "loss": 0.0001,
      "step": 17224
    },
    {
      "epoch": 8.039673278879814,
      "grad_norm": 0.020263671875,
      "learning_rate": 1.8354550549645343e-05,
      "loss": 0.0002,
      "step": 17225
    },
    {
      "epoch": 8.040140023337223,
      "grad_norm": 0.1103515625,
      "learning_rate": 1.83460788161409e-05,
      "loss": 0.0004,
      "step": 17226
    },
    {
      "epoch": 8.040606767794632,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.833760884073601e-05,
      "loss": 0.0001,
      "step": 17227
    },
    {
      "epoch": 8.041073512252042,
      "grad_norm": 0.00653076171875,
      "learning_rate": 1.8329140623612984e-05,
      "loss": 0.0002,
      "step": 17228
    },
    {
      "epoch": 8.041540256709451,
      "grad_norm": 0.00848388671875,
      "learning_rate": 1.8320674164954175e-05,
      "loss": 0.0001,
      "step": 17229
    },
    {
      "epoch": 8.04200700116686,
      "grad_norm": 0.007049560546875,
      "learning_rate": 1.83122094649419e-05,
      "loss": 0.0001,
      "step": 17230
    },
    {
      "epoch": 8.042473745624271,
      "grad_norm": 0.0174560546875,
      "learning_rate": 1.830374652375836e-05,
      "loss": 0.0002,
      "step": 17231
    },
    {
      "epoch": 8.04294049008168,
      "grad_norm": 0.046142578125,
      "learning_rate": 1.82952853415858e-05,
      "loss": 0.0018,
      "step": 17232
    },
    {
      "epoch": 8.04340723453909,
      "grad_norm": 0.016845703125,
      "learning_rate": 1.8286825918606433e-05,
      "loss": 0.0001,
      "step": 17233
    },
    {
      "epoch": 8.0438739789965,
      "grad_norm": 0.040283203125,
      "learning_rate": 1.8278368255002332e-05,
      "loss": 0.0018,
      "step": 17234
    },
    {
      "epoch": 8.044340723453908,
      "grad_norm": 0.01458740234375,
      "learning_rate": 1.8269912350955642e-05,
      "loss": 0.0001,
      "step": 17235
    },
    {
      "epoch": 8.044807467911319,
      "grad_norm": 0.0267333984375,
      "learning_rate": 1.826145820664844e-05,
      "loss": 0.0002,
      "step": 17236
    },
    {
      "epoch": 8.045274212368728,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.8253005822262715e-05,
      "loss": 0.0001,
      "step": 17237
    },
    {
      "epoch": 8.045740956826137,
      "grad_norm": 0.0137939453125,
      "learning_rate": 1.8244555197980472e-05,
      "loss": 0.0001,
      "step": 17238
    },
    {
      "epoch": 8.046207701283548,
      "grad_norm": 0.047607421875,
      "learning_rate": 1.8236106333983715e-05,
      "loss": 0.0017,
      "step": 17239
    },
    {
      "epoch": 8.046674445740956,
      "grad_norm": 0.007354736328125,
      "learning_rate": 1.8227659230454252e-05,
      "loss": 0.0001,
      "step": 17240
    },
    {
      "epoch": 8.047141190198367,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.8219213887574017e-05,
      "loss": 0.0001,
      "step": 17241
    },
    {
      "epoch": 8.047607934655776,
      "grad_norm": 0.010986328125,
      "learning_rate": 1.8210770305524872e-05,
      "loss": 0.0002,
      "step": 17242
    },
    {
      "epoch": 8.048074679113185,
      "grad_norm": 0.048828125,
      "learning_rate": 1.820232848448856e-05,
      "loss": 0.002,
      "step": 17243
    },
    {
      "epoch": 8.048541423570596,
      "grad_norm": 0.01165771484375,
      "learning_rate": 1.8193888424646876e-05,
      "loss": 0.0001,
      "step": 17244
    },
    {
      "epoch": 8.049008168028005,
      "grad_norm": 0.0135498046875,
      "learning_rate": 1.8185450126181557e-05,
      "loss": 0.0001,
      "step": 17245
    },
    {
      "epoch": 8.049474912485413,
      "grad_norm": 0.006988525390625,
      "learning_rate": 1.817701358927425e-05,
      "loss": 0.0001,
      "step": 17246
    },
    {
      "epoch": 8.049941656942824,
      "grad_norm": 0.01068115234375,
      "learning_rate": 1.816857881410663e-05,
      "loss": 0.0001,
      "step": 17247
    },
    {
      "epoch": 8.050408401400233,
      "grad_norm": 0.01116943359375,
      "learning_rate": 1.8160145800860316e-05,
      "loss": 0.0002,
      "step": 17248
    },
    {
      "epoch": 8.050875145857644,
      "grad_norm": 0.01806640625,
      "learning_rate": 1.8151714549716857e-05,
      "loss": 0.0002,
      "step": 17249
    },
    {
      "epoch": 8.051341890315053,
      "grad_norm": 0.0101318359375,
      "learning_rate": 1.8143285060857784e-05,
      "loss": 0.0002,
      "step": 17250
    },
    {
      "epoch": 8.051808634772462,
      "grad_norm": 0.01055908203125,
      "learning_rate": 1.8134857334464627e-05,
      "loss": 0.0001,
      "step": 17251
    },
    {
      "epoch": 8.052275379229872,
      "grad_norm": 0.007598876953125,
      "learning_rate": 1.812643137071881e-05,
      "loss": 0.0002,
      "step": 17252
    },
    {
      "epoch": 8.052742123687281,
      "grad_norm": 0.005950927734375,
      "learning_rate": 1.8118007169801755e-05,
      "loss": 0.0001,
      "step": 17253
    },
    {
      "epoch": 8.05320886814469,
      "grad_norm": 0.014892578125,
      "learning_rate": 1.810958473189489e-05,
      "loss": 0.0002,
      "step": 17254
    },
    {
      "epoch": 8.0536756126021,
      "grad_norm": 0.03857421875,
      "learning_rate": 1.8101164057179488e-05,
      "loss": 0.002,
      "step": 17255
    },
    {
      "epoch": 8.05414235705951,
      "grad_norm": 0.01141357421875,
      "learning_rate": 1.80927451458369e-05,
      "loss": 0.0001,
      "step": 17256
    },
    {
      "epoch": 8.05460910151692,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.808432799804841e-05,
      "loss": 0.0002,
      "step": 17257
    },
    {
      "epoch": 8.05507584597433,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.8075912613995203e-05,
      "loss": 0.0001,
      "step": 17258
    },
    {
      "epoch": 8.055542590431738,
      "grad_norm": 0.006866455078125,
      "learning_rate": 1.8067498993858513e-05,
      "loss": 0.0001,
      "step": 17259
    },
    {
      "epoch": 8.056009334889149,
      "grad_norm": 0.023681640625,
      "learning_rate": 1.8059087137819453e-05,
      "loss": 0.0022,
      "step": 17260
    },
    {
      "epoch": 8.056476079346558,
      "grad_norm": 0.0155029296875,
      "learning_rate": 1.805067704605916e-05,
      "loss": 0.0002,
      "step": 17261
    },
    {
      "epoch": 8.056942823803967,
      "grad_norm": 0.04345703125,
      "learning_rate": 1.804226871875875e-05,
      "loss": 0.0032,
      "step": 17262
    },
    {
      "epoch": 8.057409568261377,
      "grad_norm": 0.02001953125,
      "learning_rate": 1.803386215609919e-05,
      "loss": 0.0002,
      "step": 17263
    },
    {
      "epoch": 8.057876312718786,
      "grad_norm": 0.0034332275390625,
      "learning_rate": 1.802545735826152e-05,
      "loss": 0.0001,
      "step": 17264
    },
    {
      "epoch": 8.058343057176195,
      "grad_norm": 0.00689697265625,
      "learning_rate": 1.801705432542674e-05,
      "loss": 0.0002,
      "step": 17265
    },
    {
      "epoch": 8.058809801633606,
      "grad_norm": 0.09423828125,
      "learning_rate": 1.8008653057775715e-05,
      "loss": 0.0021,
      "step": 17266
    },
    {
      "epoch": 8.059276546091015,
      "grad_norm": 0.007080078125,
      "learning_rate": 1.8000253555489356e-05,
      "loss": 0.0001,
      "step": 17267
    },
    {
      "epoch": 8.059743290548425,
      "grad_norm": 0.0419921875,
      "learning_rate": 1.799185581874855e-05,
      "loss": 0.0022,
      "step": 17268
    },
    {
      "epoch": 8.060210035005834,
      "grad_norm": 0.035400390625,
      "learning_rate": 1.7983459847734053e-05,
      "loss": 0.0002,
      "step": 17269
    },
    {
      "epoch": 8.060676779463243,
      "grad_norm": 0.0263671875,
      "learning_rate": 1.7975065642626652e-05,
      "loss": 0.0002,
      "step": 17270
    },
    {
      "epoch": 8.061143523920654,
      "grad_norm": 0.0654296875,
      "learning_rate": 1.7966673203607132e-05,
      "loss": 0.0003,
      "step": 17271
    },
    {
      "epoch": 8.061610268378063,
      "grad_norm": 0.00579833984375,
      "learning_rate": 1.795828253085613e-05,
      "loss": 0.0001,
      "step": 17272
    },
    {
      "epoch": 8.062077012835472,
      "grad_norm": 0.005859375,
      "learning_rate": 1.7949893624554326e-05,
      "loss": 0.0001,
      "step": 17273
    },
    {
      "epoch": 8.062543757292882,
      "grad_norm": 0.0260009765625,
      "learning_rate": 1.7941506484882375e-05,
      "loss": 0.0002,
      "step": 17274
    },
    {
      "epoch": 8.063010501750291,
      "grad_norm": 0.00848388671875,
      "learning_rate": 1.7933121112020812e-05,
      "loss": 0.0002,
      "step": 17275
    },
    {
      "epoch": 8.063477246207702,
      "grad_norm": 0.0230712890625,
      "learning_rate": 1.7924737506150236e-05,
      "loss": 0.0002,
      "step": 17276
    },
    {
      "epoch": 8.063943990665111,
      "grad_norm": 0.00677490234375,
      "learning_rate": 1.79163556674511e-05,
      "loss": 0.0001,
      "step": 17277
    },
    {
      "epoch": 8.06441073512252,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.790797559610389e-05,
      "loss": 0.0002,
      "step": 17278
    },
    {
      "epoch": 8.06487747957993,
      "grad_norm": 0.005767822265625,
      "learning_rate": 1.7899597292289083e-05,
      "loss": 0.0001,
      "step": 17279
    },
    {
      "epoch": 8.06534422403734,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.789122075618701e-05,
      "loss": 0.0001,
      "step": 17280
    },
    {
      "epoch": 8.065810968494748,
      "grad_norm": 0.0230712890625,
      "learning_rate": 1.7882845987978048e-05,
      "loss": 0.0001,
      "step": 17281
    },
    {
      "epoch": 8.066277712952159,
      "grad_norm": 0.007171630859375,
      "learning_rate": 1.7874472987842562e-05,
      "loss": 0.0001,
      "step": 17282
    },
    {
      "epoch": 8.066744457409568,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.7866101755960752e-05,
      "loss": 0.0002,
      "step": 17283
    },
    {
      "epoch": 8.067211201866979,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.785773229251291e-05,
      "loss": 0.0001,
      "step": 17284
    },
    {
      "epoch": 8.067677946324388,
      "grad_norm": 0.0284423828125,
      "learning_rate": 1.7849364597679252e-05,
      "loss": 0.0001,
      "step": 17285
    },
    {
      "epoch": 8.068144690781796,
      "grad_norm": 0.028076171875,
      "learning_rate": 1.7840998671639897e-05,
      "loss": 0.0002,
      "step": 17286
    },
    {
      "epoch": 8.068611435239207,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.7832634514575018e-05,
      "loss": 0.0002,
      "step": 17287
    },
    {
      "epoch": 8.069078179696616,
      "grad_norm": 0.02880859375,
      "learning_rate": 1.78242721266647e-05,
      "loss": 0.0002,
      "step": 17288
    },
    {
      "epoch": 8.069544924154025,
      "grad_norm": 0.017578125,
      "learning_rate": 1.7815911508088932e-05,
      "loss": 0.0002,
      "step": 17289
    },
    {
      "epoch": 8.070011668611436,
      "grad_norm": 0.012451171875,
      "learning_rate": 1.7807552659027792e-05,
      "loss": 0.0001,
      "step": 17290
    },
    {
      "epoch": 8.070478413068845,
      "grad_norm": 0.014892578125,
      "learning_rate": 1.7799195579661255e-05,
      "loss": 0.0002,
      "step": 17291
    },
    {
      "epoch": 8.070945157526255,
      "grad_norm": 0.00701904296875,
      "learning_rate": 1.7790840270169217e-05,
      "loss": 0.0001,
      "step": 17292
    },
    {
      "epoch": 8.071411901983664,
      "grad_norm": 0.021484375,
      "learning_rate": 1.7782486730731596e-05,
      "loss": 0.0001,
      "step": 17293
    },
    {
      "epoch": 8.071878646441073,
      "grad_norm": 0.034423828125,
      "learning_rate": 1.7774134961528288e-05,
      "loss": 0.0019,
      "step": 17294
    },
    {
      "epoch": 8.072345390898484,
      "grad_norm": 0.04736328125,
      "learning_rate": 1.7765784962739062e-05,
      "loss": 0.0021,
      "step": 17295
    },
    {
      "epoch": 8.072812135355893,
      "grad_norm": 0.0810546875,
      "learning_rate": 1.7757436734543718e-05,
      "loss": 0.0036,
      "step": 17296
    },
    {
      "epoch": 8.073278879813302,
      "grad_norm": 0.490234375,
      "learning_rate": 1.7749090277122037e-05,
      "loss": 0.0057,
      "step": 17297
    },
    {
      "epoch": 8.073745624270712,
      "grad_norm": 0.18359375,
      "learning_rate": 1.7740745590653683e-05,
      "loss": 0.0005,
      "step": 17298
    },
    {
      "epoch": 8.074212368728121,
      "grad_norm": 0.00665283203125,
      "learning_rate": 1.773240267531835e-05,
      "loss": 0.0001,
      "step": 17299
    },
    {
      "epoch": 8.074679113185532,
      "grad_norm": 0.04638671875,
      "learning_rate": 1.772406153129568e-05,
      "loss": 0.0002,
      "step": 17300
    },
    {
      "epoch": 8.07514585764294,
      "grad_norm": 0.0242919921875,
      "learning_rate": 1.7715722158765224e-05,
      "loss": 0.0002,
      "step": 17301
    },
    {
      "epoch": 8.07561260210035,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.770738455790657e-05,
      "loss": 0.0002,
      "step": 17302
    },
    {
      "epoch": 8.07607934655776,
      "grad_norm": 0.056396484375,
      "learning_rate": 1.769904872889926e-05,
      "loss": 0.0034,
      "step": 17303
    },
    {
      "epoch": 8.07654609101517,
      "grad_norm": 0.0191650390625,
      "learning_rate": 1.7690714671922714e-05,
      "loss": 0.0002,
      "step": 17304
    },
    {
      "epoch": 8.077012835472578,
      "grad_norm": 0.01055908203125,
      "learning_rate": 1.7682382387156405e-05,
      "loss": 0.0001,
      "step": 17305
    },
    {
      "epoch": 8.077479579929989,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.7674051874779775e-05,
      "loss": 0.0001,
      "step": 17306
    },
    {
      "epoch": 8.077946324387398,
      "grad_norm": 0.01806640625,
      "learning_rate": 1.766572313497211e-05,
      "loss": 0.0002,
      "step": 17307
    },
    {
      "epoch": 8.078413068844807,
      "grad_norm": 0.0145263671875,
      "learning_rate": 1.765739616791281e-05,
      "loss": 0.0002,
      "step": 17308
    },
    {
      "epoch": 8.078879813302217,
      "grad_norm": 0.051513671875,
      "learning_rate": 1.7649070973781102e-05,
      "loss": 0.0026,
      "step": 17309
    },
    {
      "epoch": 8.079346557759626,
      "grad_norm": 0.0081787109375,
      "learning_rate": 1.7640747552756263e-05,
      "loss": 0.0001,
      "step": 17310
    },
    {
      "epoch": 8.079813302217037,
      "grad_norm": 0.034912109375,
      "learning_rate": 1.7632425905017535e-05,
      "loss": 0.0026,
      "step": 17311
    },
    {
      "epoch": 8.080280046674446,
      "grad_norm": 0.01214599609375,
      "learning_rate": 1.7624106030744046e-05,
      "loss": 0.0001,
      "step": 17312
    },
    {
      "epoch": 8.080746791131855,
      "grad_norm": 0.015380859375,
      "learning_rate": 1.7615787930114946e-05,
      "loss": 0.0002,
      "step": 17313
    },
    {
      "epoch": 8.081213535589265,
      "grad_norm": 0.0115966796875,
      "learning_rate": 1.7607471603309365e-05,
      "loss": 0.0001,
      "step": 17314
    },
    {
      "epoch": 8.081680280046674,
      "grad_norm": 0.007476806640625,
      "learning_rate": 1.759915705050632e-05,
      "loss": 0.0001,
      "step": 17315
    },
    {
      "epoch": 8.082147024504083,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.759084427188484e-05,
      "loss": 0.0002,
      "step": 17316
    },
    {
      "epoch": 8.082613768961494,
      "grad_norm": 0.0081787109375,
      "learning_rate": 1.7582533267623946e-05,
      "loss": 0.0002,
      "step": 17317
    },
    {
      "epoch": 8.083080513418903,
      "grad_norm": 0.01080322265625,
      "learning_rate": 1.757422403790253e-05,
      "loss": 0.0002,
      "step": 17318
    },
    {
      "epoch": 8.083547257876313,
      "grad_norm": 0.00946044921875,
      "learning_rate": 1.7565916582899524e-05,
      "loss": 0.0002,
      "step": 17319
    },
    {
      "epoch": 8.084014002333722,
      "grad_norm": 0.0108642578125,
      "learning_rate": 1.755761090279382e-05,
      "loss": 0.0002,
      "step": 17320
    },
    {
      "epoch": 8.084480746791131,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.7549306997764193e-05,
      "loss": 0.0002,
      "step": 17321
    },
    {
      "epoch": 8.084947491248542,
      "grad_norm": 0.01904296875,
      "learning_rate": 1.7541004867989475e-05,
      "loss": 0.0002,
      "step": 17322
    },
    {
      "epoch": 8.08541423570595,
      "grad_norm": 0.01507568359375,
      "learning_rate": 1.7532704513648436e-05,
      "loss": 0.0002,
      "step": 17323
    },
    {
      "epoch": 8.08588098016336,
      "grad_norm": 0.0498046875,
      "learning_rate": 1.7524405934919742e-05,
      "loss": 0.002,
      "step": 17324
    },
    {
      "epoch": 8.08634772462077,
      "grad_norm": 0.01708984375,
      "learning_rate": 1.7516109131982093e-05,
      "loss": 0.0002,
      "step": 17325
    },
    {
      "epoch": 8.08681446907818,
      "grad_norm": 0.00823974609375,
      "learning_rate": 1.7507814105014152e-05,
      "loss": 0.0002,
      "step": 17326
    },
    {
      "epoch": 8.08728121353559,
      "grad_norm": 0.01129150390625,
      "learning_rate": 1.7499520854194484e-05,
      "loss": 0.0002,
      "step": 17327
    },
    {
      "epoch": 8.087747957992999,
      "grad_norm": 0.0120849609375,
      "learning_rate": 1.749122937970169e-05,
      "loss": 0.0002,
      "step": 17328
    },
    {
      "epoch": 8.088214702450408,
      "grad_norm": 0.0123291015625,
      "learning_rate": 1.748293968171425e-05,
      "loss": 0.0002,
      "step": 17329
    },
    {
      "epoch": 8.088681446907819,
      "grad_norm": 0.005859375,
      "learning_rate": 1.7474651760410676e-05,
      "loss": 0.0001,
      "step": 17330
    },
    {
      "epoch": 8.089148191365227,
      "grad_norm": 0.046630859375,
      "learning_rate": 1.7466365615969437e-05,
      "loss": 0.0002,
      "step": 17331
    },
    {
      "epoch": 8.089614935822636,
      "grad_norm": 0.005950927734375,
      "learning_rate": 1.7458081248568892e-05,
      "loss": 0.0001,
      "step": 17332
    },
    {
      "epoch": 8.090081680280047,
      "grad_norm": 0.01068115234375,
      "learning_rate": 1.744979865838743e-05,
      "loss": 0.0002,
      "step": 17333
    },
    {
      "epoch": 8.090548424737456,
      "grad_norm": 0.05859375,
      "learning_rate": 1.744151784560343e-05,
      "loss": 0.0019,
      "step": 17334
    },
    {
      "epoch": 8.091015169194867,
      "grad_norm": 0.048583984375,
      "learning_rate": 1.743323881039515e-05,
      "loss": 0.0017,
      "step": 17335
    },
    {
      "epoch": 8.091481913652276,
      "grad_norm": 0.0673828125,
      "learning_rate": 1.742496155294082e-05,
      "loss": 0.0042,
      "step": 17336
    },
    {
      "epoch": 8.091948658109684,
      "grad_norm": 0.00897216796875,
      "learning_rate": 1.7416686073418685e-05,
      "loss": 0.0002,
      "step": 17337
    },
    {
      "epoch": 8.092415402567095,
      "grad_norm": 0.007080078125,
      "learning_rate": 1.7408412372006954e-05,
      "loss": 0.0002,
      "step": 17338
    },
    {
      "epoch": 8.092882147024504,
      "grad_norm": 0.0186767578125,
      "learning_rate": 1.7400140448883717e-05,
      "loss": 0.0001,
      "step": 17339
    },
    {
      "epoch": 8.093348891481913,
      "grad_norm": 0.0169677734375,
      "learning_rate": 1.7391870304227133e-05,
      "loss": 0.0002,
      "step": 17340
    },
    {
      "epoch": 8.093815635939324,
      "grad_norm": 0.035400390625,
      "learning_rate": 1.73836019382152e-05,
      "loss": 0.0002,
      "step": 17341
    },
    {
      "epoch": 8.094282380396733,
      "grad_norm": 0.0185546875,
      "learning_rate": 1.737533535102599e-05,
      "loss": 0.0002,
      "step": 17342
    },
    {
      "epoch": 8.094749124854143,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.7367070542837503e-05,
      "loss": 0.0001,
      "step": 17343
    },
    {
      "epoch": 8.095215869311552,
      "grad_norm": 0.0517578125,
      "learning_rate": 1.7358807513827648e-05,
      "loss": 0.0002,
      "step": 17344
    },
    {
      "epoch": 8.095682613768961,
      "grad_norm": 0.01171875,
      "learning_rate": 1.7350546264174362e-05,
      "loss": 0.0001,
      "step": 17345
    },
    {
      "epoch": 8.096149358226372,
      "grad_norm": 0.010498046875,
      "learning_rate": 1.734228679405554e-05,
      "loss": 0.0001,
      "step": 17346
    },
    {
      "epoch": 8.09661610268378,
      "grad_norm": 0.015380859375,
      "learning_rate": 1.733402910364896e-05,
      "loss": 0.0002,
      "step": 17347
    },
    {
      "epoch": 8.09708284714119,
      "grad_norm": 0.0118408203125,
      "learning_rate": 1.7325773193132466e-05,
      "loss": 0.0001,
      "step": 17348
    },
    {
      "epoch": 8.0975495915986,
      "grad_norm": 0.02734375,
      "learning_rate": 1.731751906268382e-05,
      "loss": 0.0002,
      "step": 17349
    },
    {
      "epoch": 8.09801633605601,
      "grad_norm": 0.040283203125,
      "learning_rate": 1.730926671248071e-05,
      "loss": 0.0023,
      "step": 17350
    },
    {
      "epoch": 8.098483080513418,
      "grad_norm": 0.0142822265625,
      "learning_rate": 1.7301016142700823e-05,
      "loss": 0.0001,
      "step": 17351
    },
    {
      "epoch": 8.098949824970829,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.7292767353521843e-05,
      "loss": 0.0001,
      "step": 17352
    },
    {
      "epoch": 8.099416569428238,
      "grad_norm": 0.00518798828125,
      "learning_rate": 1.7284520345121323e-05,
      "loss": 0.0001,
      "step": 17353
    },
    {
      "epoch": 8.099883313885648,
      "grad_norm": 0.005157470703125,
      "learning_rate": 1.7276275117676855e-05,
      "loss": 0.0001,
      "step": 17354
    },
    {
      "epoch": 8.100350058343057,
      "grad_norm": 0.009765625,
      "learning_rate": 1.726803167136599e-05,
      "loss": 0.0001,
      "step": 17355
    },
    {
      "epoch": 8.100816802800466,
      "grad_norm": 0.130859375,
      "learning_rate": 1.7259790006366184e-05,
      "loss": 0.0004,
      "step": 17356
    },
    {
      "epoch": 8.101283547257877,
      "grad_norm": 0.006683349609375,
      "learning_rate": 1.7251550122854908e-05,
      "loss": 0.0001,
      "step": 17357
    },
    {
      "epoch": 8.101750291715286,
      "grad_norm": 0.0137939453125,
      "learning_rate": 1.7243312021009562e-05,
      "loss": 0.0002,
      "step": 17358
    },
    {
      "epoch": 8.102217036172695,
      "grad_norm": 0.0179443359375,
      "learning_rate": 1.723507570100751e-05,
      "loss": 0.0002,
      "step": 17359
    },
    {
      "epoch": 8.102683780630105,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.7226841163026142e-05,
      "loss": 0.0002,
      "step": 17360
    },
    {
      "epoch": 8.103150525087514,
      "grad_norm": 0.00897216796875,
      "learning_rate": 1.7218608407242697e-05,
      "loss": 0.0002,
      "step": 17361
    },
    {
      "epoch": 8.103617269544925,
      "grad_norm": 0.007415771484375,
      "learning_rate": 1.7210377433834457e-05,
      "loss": 0.0001,
      "step": 17362
    },
    {
      "epoch": 8.104084014002334,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.7202148242978676e-05,
      "loss": 0.0001,
      "step": 17363
    },
    {
      "epoch": 8.104550758459743,
      "grad_norm": 0.048583984375,
      "learning_rate": 1.7193920834852484e-05,
      "loss": 0.0002,
      "step": 17364
    },
    {
      "epoch": 8.105017502917153,
      "grad_norm": 0.01177978515625,
      "learning_rate": 1.7185695209633045e-05,
      "loss": 0.0002,
      "step": 17365
    },
    {
      "epoch": 8.105484247374562,
      "grad_norm": 0.005950927734375,
      "learning_rate": 1.71774713674975e-05,
      "loss": 0.0001,
      "step": 17366
    },
    {
      "epoch": 8.105950991831971,
      "grad_norm": 0.01031494140625,
      "learning_rate": 1.716924930862287e-05,
      "loss": 0.0002,
      "step": 17367
    },
    {
      "epoch": 8.106417736289382,
      "grad_norm": 0.04150390625,
      "learning_rate": 1.7161029033186205e-05,
      "loss": 0.0022,
      "step": 17368
    },
    {
      "epoch": 8.10688448074679,
      "grad_norm": 0.006988525390625,
      "learning_rate": 1.7152810541364517e-05,
      "loss": 0.0001,
      "step": 17369
    },
    {
      "epoch": 8.107351225204201,
      "grad_norm": 0.00653076171875,
      "learning_rate": 1.714459383333471e-05,
      "loss": 0.0001,
      "step": 17370
    },
    {
      "epoch": 8.10781796966161,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.7136378909273732e-05,
      "loss": 0.0002,
      "step": 17371
    },
    {
      "epoch": 8.10828471411902,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.7128165769358485e-05,
      "loss": 0.0002,
      "step": 17372
    },
    {
      "epoch": 8.10875145857643,
      "grad_norm": 0.004608154296875,
      "learning_rate": 1.711995441376575e-05,
      "loss": 0.0001,
      "step": 17373
    },
    {
      "epoch": 8.109218203033839,
      "grad_norm": 0.01031494140625,
      "learning_rate": 1.7111744842672338e-05,
      "loss": 0.0002,
      "step": 17374
    },
    {
      "epoch": 8.109684947491248,
      "grad_norm": 0.037841796875,
      "learning_rate": 1.710353705625507e-05,
      "loss": 0.0002,
      "step": 17375
    },
    {
      "epoch": 8.110151691948658,
      "grad_norm": 0.0111083984375,
      "learning_rate": 1.709533105469059e-05,
      "loss": 0.0002,
      "step": 17376
    },
    {
      "epoch": 8.110618436406067,
      "grad_norm": 0.0296630859375,
      "learning_rate": 1.708712683815564e-05,
      "loss": 0.0018,
      "step": 17377
    },
    {
      "epoch": 8.111085180863478,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.7078924406826823e-05,
      "loss": 0.0002,
      "step": 17378
    },
    {
      "epoch": 8.111551925320887,
      "grad_norm": 0.006927490234375,
      "learning_rate": 1.7070723760880767e-05,
      "loss": 0.0002,
      "step": 17379
    },
    {
      "epoch": 8.112018669778296,
      "grad_norm": 0.0146484375,
      "learning_rate": 1.706252490049406e-05,
      "loss": 0.0002,
      "step": 17380
    },
    {
      "epoch": 8.112485414235707,
      "grad_norm": 0.0279541015625,
      "learning_rate": 1.7054327825843196e-05,
      "loss": 0.0002,
      "step": 17381
    },
    {
      "epoch": 8.112952158693115,
      "grad_norm": 0.01092529296875,
      "learning_rate": 1.7046132537104697e-05,
      "loss": 0.0002,
      "step": 17382
    },
    {
      "epoch": 8.113418903150524,
      "grad_norm": 0.01531982421875,
      "learning_rate": 1.7037939034454984e-05,
      "loss": 0.0002,
      "step": 17383
    },
    {
      "epoch": 8.113885647607935,
      "grad_norm": 0.01171875,
      "learning_rate": 1.702974731807052e-05,
      "loss": 0.0002,
      "step": 17384
    },
    {
      "epoch": 8.114352392065344,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.702155738812763e-05,
      "loss": 0.0001,
      "step": 17385
    },
    {
      "epoch": 8.114819136522755,
      "grad_norm": 0.02783203125,
      "learning_rate": 1.701336924480267e-05,
      "loss": 0.0023,
      "step": 17386
    },
    {
      "epoch": 8.115285880980164,
      "grad_norm": 0.0216064453125,
      "learning_rate": 1.7005182888271965e-05,
      "loss": 0.0002,
      "step": 17387
    },
    {
      "epoch": 8.115752625437572,
      "grad_norm": 0.006439208984375,
      "learning_rate": 1.6996998318711743e-05,
      "loss": 0.0001,
      "step": 17388
    },
    {
      "epoch": 8.116219369894983,
      "grad_norm": 0.0634765625,
      "learning_rate": 1.6988815536298253e-05,
      "loss": 0.0049,
      "step": 17389
    },
    {
      "epoch": 8.116686114352392,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.698063454120765e-05,
      "loss": 0.0002,
      "step": 17390
    },
    {
      "epoch": 8.117152858809801,
      "grad_norm": 0.007110595703125,
      "learning_rate": 1.69724553336161e-05,
      "loss": 0.0001,
      "step": 17391
    },
    {
      "epoch": 8.117619603267212,
      "grad_norm": 0.021240234375,
      "learning_rate": 1.6964277913699732e-05,
      "loss": 0.0002,
      "step": 17392
    },
    {
      "epoch": 8.11808634772462,
      "grad_norm": 0.03564453125,
      "learning_rate": 1.6956102281634556e-05,
      "loss": 0.0002,
      "step": 17393
    },
    {
      "epoch": 8.11855309218203,
      "grad_norm": 0.009765625,
      "learning_rate": 1.694792843759665e-05,
      "loss": 0.0001,
      "step": 17394
    },
    {
      "epoch": 8.11901983663944,
      "grad_norm": 0.01611328125,
      "learning_rate": 1.6939756381762008e-05,
      "loss": 0.0002,
      "step": 17395
    },
    {
      "epoch": 8.119486581096849,
      "grad_norm": 0.006988525390625,
      "learning_rate": 1.693158611430654e-05,
      "loss": 0.0002,
      "step": 17396
    },
    {
      "epoch": 8.11995332555426,
      "grad_norm": 0.00628662109375,
      "learning_rate": 1.69234176354062e-05,
      "loss": 0.0001,
      "step": 17397
    },
    {
      "epoch": 8.120420070011669,
      "grad_norm": 0.011962890625,
      "learning_rate": 1.691525094523687e-05,
      "loss": 0.0001,
      "step": 17398
    },
    {
      "epoch": 8.120886814469078,
      "grad_norm": 0.00537109375,
      "learning_rate": 1.690708604397435e-05,
      "loss": 0.0001,
      "step": 17399
    },
    {
      "epoch": 8.121353558926488,
      "grad_norm": 0.0203857421875,
      "learning_rate": 1.689892293179446e-05,
      "loss": 0.0002,
      "step": 17400
    },
    {
      "epoch": 8.121820303383897,
      "grad_norm": 0.00927734375,
      "learning_rate": 1.6890761608872986e-05,
      "loss": 0.0002,
      "step": 17401
    },
    {
      "epoch": 8.122287047841306,
      "grad_norm": 0.021484375,
      "learning_rate": 1.68826020753856e-05,
      "loss": 0.0002,
      "step": 17402
    },
    {
      "epoch": 8.122753792298717,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.687444433150801e-05,
      "loss": 0.0001,
      "step": 17403
    },
    {
      "epoch": 8.123220536756126,
      "grad_norm": 0.006622314453125,
      "learning_rate": 1.6866288377415894e-05,
      "loss": 0.0001,
      "step": 17404
    },
    {
      "epoch": 8.123687281213536,
      "grad_norm": 0.01116943359375,
      "learning_rate": 1.6858134213284804e-05,
      "loss": 0.0002,
      "step": 17405
    },
    {
      "epoch": 8.124154025670945,
      "grad_norm": 0.0478515625,
      "learning_rate": 1.6849981839290352e-05,
      "loss": 0.0014,
      "step": 17406
    },
    {
      "epoch": 8.124620770128354,
      "grad_norm": 0.12255859375,
      "learning_rate": 1.684183125560802e-05,
      "loss": 0.0003,
      "step": 17407
    },
    {
      "epoch": 8.125087514585765,
      "grad_norm": 0.10400390625,
      "learning_rate": 1.6833682462413326e-05,
      "loss": 0.0005,
      "step": 17408
    },
    {
      "epoch": 8.125554259043174,
      "grad_norm": 0.0142822265625,
      "learning_rate": 1.682553545988175e-05,
      "loss": 0.0002,
      "step": 17409
    },
    {
      "epoch": 8.126021003500583,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.6817390248188658e-05,
      "loss": 0.0001,
      "step": 17410
    },
    {
      "epoch": 8.126487747957993,
      "grad_norm": 0.01708984375,
      "learning_rate": 1.6809246827509438e-05,
      "loss": 0.0002,
      "step": 17411
    },
    {
      "epoch": 8.126954492415402,
      "grad_norm": 0.032958984375,
      "learning_rate": 1.6801105198019472e-05,
      "loss": 0.002,
      "step": 17412
    },
    {
      "epoch": 8.127421236872813,
      "grad_norm": 0.0181884765625,
      "learning_rate": 1.6792965359893997e-05,
      "loss": 0.0002,
      "step": 17413
    },
    {
      "epoch": 8.127887981330222,
      "grad_norm": 0.0126953125,
      "learning_rate": 1.6784827313308295e-05,
      "loss": 0.0001,
      "step": 17414
    },
    {
      "epoch": 8.12835472578763,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.677669105843761e-05,
      "loss": 0.0002,
      "step": 17415
    },
    {
      "epoch": 8.128821470245041,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.6768556595457074e-05,
      "loss": 0.0001,
      "step": 17416
    },
    {
      "epoch": 8.12928821470245,
      "grad_norm": 0.01214599609375,
      "learning_rate": 1.676042392454187e-05,
      "loss": 0.0002,
      "step": 17417
    },
    {
      "epoch": 8.12975495915986,
      "grad_norm": 0.01025390625,
      "learning_rate": 1.675229304586712e-05,
      "loss": 0.0002,
      "step": 17418
    },
    {
      "epoch": 8.13022170361727,
      "grad_norm": 0.019287109375,
      "learning_rate": 1.6744163959607838e-05,
      "loss": 0.0002,
      "step": 17419
    },
    {
      "epoch": 8.130688448074679,
      "grad_norm": 0.007659912109375,
      "learning_rate": 1.673603666593908e-05,
      "loss": 0.0001,
      "step": 17420
    },
    {
      "epoch": 8.13115519253209,
      "grad_norm": 0.0172119140625,
      "learning_rate": 1.672791116503587e-05,
      "loss": 0.0002,
      "step": 17421
    },
    {
      "epoch": 8.131621936989498,
      "grad_norm": 0.0537109375,
      "learning_rate": 1.671978745707309e-05,
      "loss": 0.0003,
      "step": 17422
    },
    {
      "epoch": 8.132088681446907,
      "grad_norm": 0.0279541015625,
      "learning_rate": 1.6711665542225684e-05,
      "loss": 0.0002,
      "step": 17423
    },
    {
      "epoch": 8.132555425904318,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.6703545420668566e-05,
      "loss": 0.0001,
      "step": 17424
    },
    {
      "epoch": 8.133022170361727,
      "grad_norm": 0.01318359375,
      "learning_rate": 1.66954270925765e-05,
      "loss": 0.0002,
      "step": 17425
    },
    {
      "epoch": 8.133488914819136,
      "grad_norm": 0.006378173828125,
      "learning_rate": 1.6687310558124337e-05,
      "loss": 0.0001,
      "step": 17426
    },
    {
      "epoch": 8.133955659276547,
      "grad_norm": 0.09716796875,
      "learning_rate": 1.6679195817486792e-05,
      "loss": 0.0003,
      "step": 17427
    },
    {
      "epoch": 8.134422403733955,
      "grad_norm": 0.06298828125,
      "learning_rate": 1.6671082870838616e-05,
      "loss": 0.0021,
      "step": 17428
    },
    {
      "epoch": 8.134889148191366,
      "grad_norm": 0.007293701171875,
      "learning_rate": 1.6662971718354504e-05,
      "loss": 0.0001,
      "step": 17429
    },
    {
      "epoch": 8.135355892648775,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.6654862360209067e-05,
      "loss": 0.0002,
      "step": 17430
    },
    {
      "epoch": 8.135822637106184,
      "grad_norm": 0.0137939453125,
      "learning_rate": 1.66467547965769e-05,
      "loss": 0.0002,
      "step": 17431
    },
    {
      "epoch": 8.136289381563595,
      "grad_norm": 0.04443359375,
      "learning_rate": 1.663864902763258e-05,
      "loss": 0.0003,
      "step": 17432
    },
    {
      "epoch": 8.136756126021004,
      "grad_norm": 0.009033203125,
      "learning_rate": 1.6630545053550662e-05,
      "loss": 0.0002,
      "step": 17433
    },
    {
      "epoch": 8.137222870478412,
      "grad_norm": 0.006500244140625,
      "learning_rate": 1.6622442874505584e-05,
      "loss": 0.0001,
      "step": 17434
    },
    {
      "epoch": 8.137689614935823,
      "grad_norm": 0.00714111328125,
      "learning_rate": 1.661434249067182e-05,
      "loss": 0.0002,
      "step": 17435
    },
    {
      "epoch": 8.138156359393232,
      "grad_norm": 0.0177001953125,
      "learning_rate": 1.6606243902223806e-05,
      "loss": 0.0002,
      "step": 17436
    },
    {
      "epoch": 8.138623103850641,
      "grad_norm": 0.0272216796875,
      "learning_rate": 1.659814710933586e-05,
      "loss": 0.002,
      "step": 17437
    },
    {
      "epoch": 8.139089848308052,
      "grad_norm": 0.01226806640625,
      "learning_rate": 1.6590052112182376e-05,
      "loss": 0.0002,
      "step": 17438
    },
    {
      "epoch": 8.13955659276546,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.6581958910937578e-05,
      "loss": 0.0001,
      "step": 17439
    },
    {
      "epoch": 8.140023337222871,
      "grad_norm": 0.006011962890625,
      "learning_rate": 1.6573867505775764e-05,
      "loss": 0.0001,
      "step": 17440
    },
    {
      "epoch": 8.14049008168028,
      "grad_norm": 0.049560546875,
      "learning_rate": 1.656577789687117e-05,
      "loss": 0.0002,
      "step": 17441
    },
    {
      "epoch": 8.140956826137689,
      "grad_norm": 0.00482177734375,
      "learning_rate": 1.6557690084397924e-05,
      "loss": 0.0001,
      "step": 17442
    },
    {
      "epoch": 8.1414235705951,
      "grad_norm": 0.01416015625,
      "learning_rate": 1.65496040685302e-05,
      "loss": 0.0002,
      "step": 17443
    },
    {
      "epoch": 8.141890315052509,
      "grad_norm": 0.04931640625,
      "learning_rate": 1.654151984944211e-05,
      "loss": 0.0002,
      "step": 17444
    },
    {
      "epoch": 8.142357059509918,
      "grad_norm": 0.0123291015625,
      "learning_rate": 1.6533437427307662e-05,
      "loss": 0.0002,
      "step": 17445
    },
    {
      "epoch": 8.142823803967328,
      "grad_norm": 0.0185546875,
      "learning_rate": 1.652535680230093e-05,
      "loss": 0.0001,
      "step": 17446
    },
    {
      "epoch": 8.143290548424737,
      "grad_norm": 0.05908203125,
      "learning_rate": 1.65172779745959e-05,
      "loss": 0.0029,
      "step": 17447
    },
    {
      "epoch": 8.143757292882148,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.6509200944366477e-05,
      "loss": 0.0002,
      "step": 17448
    },
    {
      "epoch": 8.144224037339557,
      "grad_norm": 0.11181640625,
      "learning_rate": 1.6501125711786592e-05,
      "loss": 0.0029,
      "step": 17449
    },
    {
      "epoch": 8.144690781796966,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.649305227703013e-05,
      "loss": 0.0002,
      "step": 17450
    },
    {
      "epoch": 8.145157526254376,
      "grad_norm": 0.0191650390625,
      "learning_rate": 1.6484980640270898e-05,
      "loss": 0.0001,
      "step": 17451
    },
    {
      "epoch": 8.145624270711785,
      "grad_norm": 0.006011962890625,
      "learning_rate": 1.6476910801682677e-05,
      "loss": 0.0001,
      "step": 17452
    },
    {
      "epoch": 8.146091015169194,
      "grad_norm": 0.0245361328125,
      "learning_rate": 1.6468842761439274e-05,
      "loss": 0.0002,
      "step": 17453
    },
    {
      "epoch": 8.146557759626605,
      "grad_norm": 0.011962890625,
      "learning_rate": 1.646077651971433e-05,
      "loss": 0.0002,
      "step": 17454
    },
    {
      "epoch": 8.147024504084014,
      "grad_norm": 0.0206298828125,
      "learning_rate": 1.6452712076681566e-05,
      "loss": 0.0002,
      "step": 17455
    },
    {
      "epoch": 8.147491248541424,
      "grad_norm": 0.01129150390625,
      "learning_rate": 1.6444649432514626e-05,
      "loss": 0.0002,
      "step": 17456
    },
    {
      "epoch": 8.147957992998833,
      "grad_norm": 0.006805419921875,
      "learning_rate": 1.643658858738707e-05,
      "loss": 0.0001,
      "step": 17457
    },
    {
      "epoch": 8.148424737456242,
      "grad_norm": 0.01348876953125,
      "learning_rate": 1.6428529541472503e-05,
      "loss": 0.0002,
      "step": 17458
    },
    {
      "epoch": 8.148891481913653,
      "grad_norm": 0.01165771484375,
      "learning_rate": 1.642047229494439e-05,
      "loss": 0.0001,
      "step": 17459
    },
    {
      "epoch": 8.149358226371062,
      "grad_norm": 0.01239013671875,
      "learning_rate": 1.6412416847976243e-05,
      "loss": 0.0002,
      "step": 17460
    },
    {
      "epoch": 8.14982497082847,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.6404363200741523e-05,
      "loss": 0.0001,
      "step": 17461
    },
    {
      "epoch": 8.150291715285881,
      "grad_norm": 0.006622314453125,
      "learning_rate": 1.6396311353413595e-05,
      "loss": 0.0001,
      "step": 17462
    },
    {
      "epoch": 8.15075845974329,
      "grad_norm": 0.052978515625,
      "learning_rate": 1.6388261306165843e-05,
      "loss": 0.0002,
      "step": 17463
    },
    {
      "epoch": 8.151225204200701,
      "grad_norm": 0.06396484375,
      "learning_rate": 1.6380213059171623e-05,
      "loss": 0.003,
      "step": 17464
    },
    {
      "epoch": 8.15169194865811,
      "grad_norm": 0.02001953125,
      "learning_rate": 1.6372166612604168e-05,
      "loss": 0.0002,
      "step": 17465
    },
    {
      "epoch": 8.152158693115519,
      "grad_norm": 0.044677734375,
      "learning_rate": 1.6364121966636746e-05,
      "loss": 0.0017,
      "step": 17466
    },
    {
      "epoch": 8.15262543757293,
      "grad_norm": 0.0400390625,
      "learning_rate": 1.6356079121442603e-05,
      "loss": 0.0015,
      "step": 17467
    },
    {
      "epoch": 8.153092182030338,
      "grad_norm": 0.045166015625,
      "learning_rate": 1.634803807719485e-05,
      "loss": 0.0003,
      "step": 17468
    },
    {
      "epoch": 8.153558926487747,
      "grad_norm": 0.01806640625,
      "learning_rate": 1.633999883406665e-05,
      "loss": 0.0002,
      "step": 17469
    },
    {
      "epoch": 8.154025670945158,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.633196139223113e-05,
      "loss": 0.0001,
      "step": 17470
    },
    {
      "epoch": 8.154492415402567,
      "grad_norm": 0.01171875,
      "learning_rate": 1.632392575186129e-05,
      "loss": 0.0002,
      "step": 17471
    },
    {
      "epoch": 8.154959159859978,
      "grad_norm": 0.03955078125,
      "learning_rate": 1.6315891913130167e-05,
      "loss": 0.0014,
      "step": 17472
    },
    {
      "epoch": 8.155425904317386,
      "grad_norm": 0.00909423828125,
      "learning_rate": 1.630785987621075e-05,
      "loss": 0.0002,
      "step": 17473
    },
    {
      "epoch": 8.155892648774795,
      "grad_norm": 0.0108642578125,
      "learning_rate": 1.6299829641275965e-05,
      "loss": 0.0001,
      "step": 17474
    },
    {
      "epoch": 8.156359393232206,
      "grad_norm": 0.00469970703125,
      "learning_rate": 1.6291801208498723e-05,
      "loss": 0.0001,
      "step": 17475
    },
    {
      "epoch": 8.156826137689615,
      "grad_norm": 0.04345703125,
      "learning_rate": 1.628377457805186e-05,
      "loss": 0.0019,
      "step": 17476
    },
    {
      "epoch": 8.157292882147024,
      "grad_norm": 0.032470703125,
      "learning_rate": 1.6275749750108215e-05,
      "loss": 0.0002,
      "step": 17477
    },
    {
      "epoch": 8.157759626604435,
      "grad_norm": 0.052978515625,
      "learning_rate": 1.6267726724840592e-05,
      "loss": 0.0002,
      "step": 17478
    },
    {
      "epoch": 8.158226371061843,
      "grad_norm": 0.10888671875,
      "learning_rate": 1.625970550242172e-05,
      "loss": 0.0003,
      "step": 17479
    },
    {
      "epoch": 8.158693115519252,
      "grad_norm": 0.00946044921875,
      "learning_rate": 1.6251686083024276e-05,
      "loss": 0.0001,
      "step": 17480
    },
    {
      "epoch": 8.159159859976663,
      "grad_norm": 0.0096435546875,
      "learning_rate": 1.624366846682095e-05,
      "loss": 0.0002,
      "step": 17481
    },
    {
      "epoch": 8.159626604434072,
      "grad_norm": 0.050537109375,
      "learning_rate": 1.623565265398439e-05,
      "loss": 0.0038,
      "step": 17482
    },
    {
      "epoch": 8.160093348891483,
      "grad_norm": 0.00543212890625,
      "learning_rate": 1.6227638644687148e-05,
      "loss": 0.0001,
      "step": 17483
    },
    {
      "epoch": 8.160560093348892,
      "grad_norm": 0.020751953125,
      "learning_rate": 1.6219626439101786e-05,
      "loss": 0.0006,
      "step": 17484
    },
    {
      "epoch": 8.1610268378063,
      "grad_norm": 0.00738525390625,
      "learning_rate": 1.6211616037400858e-05,
      "loss": 0.0001,
      "step": 17485
    },
    {
      "epoch": 8.161493582263711,
      "grad_norm": 0.0185546875,
      "learning_rate": 1.6203607439756773e-05,
      "loss": 0.0001,
      "step": 17486
    },
    {
      "epoch": 8.16196032672112,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.6195600646342014e-05,
      "loss": 0.0002,
      "step": 17487
    },
    {
      "epoch": 8.162427071178529,
      "grad_norm": 0.0284423828125,
      "learning_rate": 1.6187595657328935e-05,
      "loss": 0.0019,
      "step": 17488
    },
    {
      "epoch": 8.16289381563594,
      "grad_norm": 0.0115966796875,
      "learning_rate": 1.6179592472889914e-05,
      "loss": 0.0001,
      "step": 17489
    },
    {
      "epoch": 8.163360560093349,
      "grad_norm": 0.006317138671875,
      "learning_rate": 1.6171591093197292e-05,
      "loss": 0.0001,
      "step": 17490
    },
    {
      "epoch": 8.16382730455076,
      "grad_norm": 0.01129150390625,
      "learning_rate": 1.6163591518423294e-05,
      "loss": 0.0001,
      "step": 17491
    },
    {
      "epoch": 8.164294049008168,
      "grad_norm": 0.005950927734375,
      "learning_rate": 1.6155593748740194e-05,
      "loss": 0.0001,
      "step": 17492
    },
    {
      "epoch": 8.164760793465577,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.61475977843202e-05,
      "loss": 0.0001,
      "step": 17493
    },
    {
      "epoch": 8.165227537922988,
      "grad_norm": 0.006500244140625,
      "learning_rate": 1.613960362533545e-05,
      "loss": 0.0001,
      "step": 17494
    },
    {
      "epoch": 8.165694282380397,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.6131611271958068e-05,
      "loss": 0.0002,
      "step": 17495
    },
    {
      "epoch": 8.166161026837806,
      "grad_norm": 0.00701904296875,
      "learning_rate": 1.612362072436019e-05,
      "loss": 0.0002,
      "step": 17496
    },
    {
      "epoch": 8.166627771295216,
      "grad_norm": 0.0120849609375,
      "learning_rate": 1.6115631982713785e-05,
      "loss": 0.0002,
      "step": 17497
    },
    {
      "epoch": 8.167094515752625,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.610764504719089e-05,
      "loss": 0.0001,
      "step": 17498
    },
    {
      "epoch": 8.167561260210036,
      "grad_norm": 0.06884765625,
      "learning_rate": 1.6099659917963505e-05,
      "loss": 0.0049,
      "step": 17499
    },
    {
      "epoch": 8.168028004667445,
      "grad_norm": 0.0263671875,
      "learning_rate": 1.6091676595203508e-05,
      "loss": 0.0022,
      "step": 17500
    },
    {
      "epoch": 8.168494749124854,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.6083695079082817e-05,
      "loss": 0.0001,
      "step": 17501
    },
    {
      "epoch": 8.168961493582264,
      "grad_norm": 0.00701904296875,
      "learning_rate": 1.6075715369773304e-05,
      "loss": 0.0001,
      "step": 17502
    },
    {
      "epoch": 8.169428238039673,
      "grad_norm": 0.00927734375,
      "learning_rate": 1.606773746744672e-05,
      "loss": 0.0001,
      "step": 17503
    },
    {
      "epoch": 8.169894982497082,
      "grad_norm": 0.022705078125,
      "learning_rate": 1.605976137227487e-05,
      "loss": 0.0001,
      "step": 17504
    },
    {
      "epoch": 8.170361726954493,
      "grad_norm": 0.00787353515625,
      "learning_rate": 1.6051787084429526e-05,
      "loss": 0.0002,
      "step": 17505
    },
    {
      "epoch": 8.170828471411902,
      "grad_norm": 0.00518798828125,
      "learning_rate": 1.6043814604082318e-05,
      "loss": 0.0001,
      "step": 17506
    },
    {
      "epoch": 8.171295215869312,
      "grad_norm": 0.0174560546875,
      "learning_rate": 1.6035843931404958e-05,
      "loss": 0.0002,
      "step": 17507
    },
    {
      "epoch": 8.171761960326721,
      "grad_norm": 0.00653076171875,
      "learning_rate": 1.6027875066569022e-05,
      "loss": 0.0001,
      "step": 17508
    },
    {
      "epoch": 8.17222870478413,
      "grad_norm": 0.00933837890625,
      "learning_rate": 1.6019908009746097e-05,
      "loss": 0.0002,
      "step": 17509
    },
    {
      "epoch": 8.172695449241541,
      "grad_norm": 0.005218505859375,
      "learning_rate": 1.6011942761107757e-05,
      "loss": 0.0001,
      "step": 17510
    },
    {
      "epoch": 8.17316219369895,
      "grad_norm": 0.004913330078125,
      "learning_rate": 1.6003979320825445e-05,
      "loss": 0.0001,
      "step": 17511
    },
    {
      "epoch": 8.173628938156359,
      "grad_norm": 0.0098876953125,
      "learning_rate": 1.599601768907065e-05,
      "loss": 0.0001,
      "step": 17512
    },
    {
      "epoch": 8.17409568261377,
      "grad_norm": 0.014404296875,
      "learning_rate": 1.598805786601484e-05,
      "loss": 0.0001,
      "step": 17513
    },
    {
      "epoch": 8.174562427071178,
      "grad_norm": 0.005126953125,
      "learning_rate": 1.5980099851829322e-05,
      "loss": 0.0001,
      "step": 17514
    },
    {
      "epoch": 8.175029171528589,
      "grad_norm": 0.0123291015625,
      "learning_rate": 1.5972143646685478e-05,
      "loss": 0.0002,
      "step": 17515
    },
    {
      "epoch": 8.175495915985998,
      "grad_norm": 0.0076904296875,
      "learning_rate": 1.5964189250754623e-05,
      "loss": 0.0002,
      "step": 17516
    },
    {
      "epoch": 8.175962660443407,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.5956236664208e-05,
      "loss": 0.0002,
      "step": 17517
    },
    {
      "epoch": 8.176429404900817,
      "grad_norm": 0.10205078125,
      "learning_rate": 1.594828588721685e-05,
      "loss": 0.0002,
      "step": 17518
    },
    {
      "epoch": 8.176896149358226,
      "grad_norm": 0.01055908203125,
      "learning_rate": 1.5940336919952393e-05,
      "loss": 0.0001,
      "step": 17519
    },
    {
      "epoch": 8.177362893815635,
      "grad_norm": 0.007171630859375,
      "learning_rate": 1.593238976258572e-05,
      "loss": 0.0001,
      "step": 17520
    },
    {
      "epoch": 8.177829638273046,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.5924444415287965e-05,
      "loss": 0.0001,
      "step": 17521
    },
    {
      "epoch": 8.178296382730455,
      "grad_norm": 0.00543212890625,
      "learning_rate": 1.5916500878230224e-05,
      "loss": 0.0001,
      "step": 17522
    },
    {
      "epoch": 8.178763127187864,
      "grad_norm": 0.0152587890625,
      "learning_rate": 1.5908559151583502e-05,
      "loss": 0.0002,
      "step": 17523
    },
    {
      "epoch": 8.179229871645274,
      "grad_norm": 0.02294921875,
      "learning_rate": 1.590061923551881e-05,
      "loss": 0.0001,
      "step": 17524
    },
    {
      "epoch": 8.179696616102683,
      "grad_norm": 0.01202392578125,
      "learning_rate": 1.5892681130207086e-05,
      "loss": 0.0001,
      "step": 17525
    },
    {
      "epoch": 8.180163360560094,
      "grad_norm": 0.005706787109375,
      "learning_rate": 1.5884744835819276e-05,
      "loss": 0.0001,
      "step": 17526
    },
    {
      "epoch": 8.180630105017503,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.5876810352526206e-05,
      "loss": 0.0001,
      "step": 17527
    },
    {
      "epoch": 8.181096849474912,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.5868877680498772e-05,
      "loss": 0.0002,
      "step": 17528
    },
    {
      "epoch": 8.181563593932323,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.586094681990773e-05,
      "loss": 0.0001,
      "step": 17529
    },
    {
      "epoch": 8.182030338389731,
      "grad_norm": 0.015625,
      "learning_rate": 1.585301777092385e-05,
      "loss": 0.0002,
      "step": 17530
    },
    {
      "epoch": 8.18249708284714,
      "grad_norm": 0.0162353515625,
      "learning_rate": 1.5845090533717887e-05,
      "loss": 0.0002,
      "step": 17531
    },
    {
      "epoch": 8.182963827304551,
      "grad_norm": 0.01031494140625,
      "learning_rate": 1.583716510846047e-05,
      "loss": 0.0002,
      "step": 17532
    },
    {
      "epoch": 8.18343057176196,
      "grad_norm": 0.0322265625,
      "learning_rate": 1.582924149532228e-05,
      "loss": 0.0006,
      "step": 17533
    },
    {
      "epoch": 8.18389731621937,
      "grad_norm": 0.00897216796875,
      "learning_rate": 1.5821319694473924e-05,
      "loss": 0.0002,
      "step": 17534
    },
    {
      "epoch": 8.18436406067678,
      "grad_norm": 0.0059814453125,
      "learning_rate": 1.581339970608593e-05,
      "loss": 0.0001,
      "step": 17535
    },
    {
      "epoch": 8.184830805134188,
      "grad_norm": 0.0076904296875,
      "learning_rate": 1.5805481530328857e-05,
      "loss": 0.0001,
      "step": 17536
    },
    {
      "epoch": 8.1852975495916,
      "grad_norm": 0.07275390625,
      "learning_rate": 1.5797565167373175e-05,
      "loss": 0.0032,
      "step": 17537
    },
    {
      "epoch": 8.185764294049008,
      "grad_norm": 0.0751953125,
      "learning_rate": 1.578965061738934e-05,
      "loss": 0.0003,
      "step": 17538
    },
    {
      "epoch": 8.186231038506417,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.5781737880547777e-05,
      "loss": 0.0012,
      "step": 17539
    },
    {
      "epoch": 8.186697782963828,
      "grad_norm": 0.033447265625,
      "learning_rate": 1.5773826957018823e-05,
      "loss": 0.0002,
      "step": 17540
    },
    {
      "epoch": 8.187164527421237,
      "grad_norm": 0.006378173828125,
      "learning_rate": 1.5765917846972812e-05,
      "loss": 0.0001,
      "step": 17541
    },
    {
      "epoch": 8.187631271878647,
      "grad_norm": 0.0673828125,
      "learning_rate": 1.5758010550580084e-05,
      "loss": 0.0021,
      "step": 17542
    },
    {
      "epoch": 8.188098016336056,
      "grad_norm": 0.01300048828125,
      "learning_rate": 1.5750105068010824e-05,
      "loss": 0.0001,
      "step": 17543
    },
    {
      "epoch": 8.188564760793465,
      "grad_norm": 0.0223388671875,
      "learning_rate": 1.5742201399435285e-05,
      "loss": 0.0019,
      "step": 17544
    },
    {
      "epoch": 8.189031505250876,
      "grad_norm": 0.01483154296875,
      "learning_rate": 1.573429954502367e-05,
      "loss": 0.0002,
      "step": 17545
    },
    {
      "epoch": 8.189498249708285,
      "grad_norm": 0.01104736328125,
      "learning_rate": 1.5726399504946033e-05,
      "loss": 0.0001,
      "step": 17546
    },
    {
      "epoch": 8.189964994165694,
      "grad_norm": 0.0167236328125,
      "learning_rate": 1.571850127937253e-05,
      "loss": 0.0002,
      "step": 17547
    },
    {
      "epoch": 8.190431738623104,
      "grad_norm": 0.006622314453125,
      "learning_rate": 1.5710604868473234e-05,
      "loss": 0.0001,
      "step": 17548
    },
    {
      "epoch": 8.190898483080513,
      "grad_norm": 0.007080078125,
      "learning_rate": 1.5702710272418107e-05,
      "loss": 0.0001,
      "step": 17549
    },
    {
      "epoch": 8.191365227537924,
      "grad_norm": 0.1240234375,
      "learning_rate": 1.5694817491377156e-05,
      "loss": 0.0025,
      "step": 17550
    },
    {
      "epoch": 8.191831971995333,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.568692652552035e-05,
      "loss": 0.0002,
      "step": 17551
    },
    {
      "epoch": 8.192298716452742,
      "grad_norm": 0.01007080078125,
      "learning_rate": 1.5679037375017547e-05,
      "loss": 0.0002,
      "step": 17552
    },
    {
      "epoch": 8.192765460910152,
      "grad_norm": 0.00750732421875,
      "learning_rate": 1.5671150040038607e-05,
      "loss": 0.0001,
      "step": 17553
    },
    {
      "epoch": 8.193232205367561,
      "grad_norm": 0.0517578125,
      "learning_rate": 1.5663264520753408e-05,
      "loss": 0.002,
      "step": 17554
    },
    {
      "epoch": 8.19369894982497,
      "grad_norm": 0.0174560546875,
      "learning_rate": 1.5655380817331667e-05,
      "loss": 0.0002,
      "step": 17555
    },
    {
      "epoch": 8.19416569428238,
      "grad_norm": 0.0223388671875,
      "learning_rate": 1.5647498929943182e-05,
      "loss": 0.0002,
      "step": 17556
    },
    {
      "epoch": 8.19463243873979,
      "grad_norm": 0.0147705078125,
      "learning_rate": 1.5639618858757622e-05,
      "loss": 0.0002,
      "step": 17557
    },
    {
      "epoch": 8.1950991831972,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.5631740603944657e-05,
      "loss": 0.0001,
      "step": 17558
    },
    {
      "epoch": 8.19556592765461,
      "grad_norm": 0.00933837890625,
      "learning_rate": 1.5623864165673952e-05,
      "loss": 0.0001,
      "step": 17559
    },
    {
      "epoch": 8.196032672112018,
      "grad_norm": 0.00738525390625,
      "learning_rate": 1.561598954411505e-05,
      "loss": 0.0002,
      "step": 17560
    },
    {
      "epoch": 8.196499416569429,
      "grad_norm": 0.0150146484375,
      "learning_rate": 1.5608116739437507e-05,
      "loss": 0.0002,
      "step": 17561
    },
    {
      "epoch": 8.196966161026838,
      "grad_norm": 0.0054931640625,
      "learning_rate": 1.5600245751810882e-05,
      "loss": 0.0001,
      "step": 17562
    },
    {
      "epoch": 8.197432905484247,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.5592376581404578e-05,
      "loss": 0.0001,
      "step": 17563
    },
    {
      "epoch": 8.197899649941657,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.5584509228388056e-05,
      "loss": 0.0002,
      "step": 17564
    },
    {
      "epoch": 8.198366394399066,
      "grad_norm": 0.00616455078125,
      "learning_rate": 1.5576643692930736e-05,
      "loss": 0.0002,
      "step": 17565
    },
    {
      "epoch": 8.198833138856475,
      "grad_norm": 0.025390625,
      "learning_rate": 1.556877997520193e-05,
      "loss": 0.0002,
      "step": 17566
    },
    {
      "epoch": 8.199299883313886,
      "grad_norm": 0.0361328125,
      "learning_rate": 1.5560918075370955e-05,
      "loss": 0.002,
      "step": 17567
    },
    {
      "epoch": 8.199766627771295,
      "grad_norm": 0.028076171875,
      "learning_rate": 1.5553057993607134e-05,
      "loss": 0.0021,
      "step": 17568
    },
    {
      "epoch": 8.200233372228706,
      "grad_norm": 0.0126953125,
      "learning_rate": 1.554519973007965e-05,
      "loss": 0.0001,
      "step": 17569
    },
    {
      "epoch": 8.200700116686114,
      "grad_norm": 0.01324462890625,
      "learning_rate": 1.553734328495772e-05,
      "loss": 0.0001,
      "step": 17570
    },
    {
      "epoch": 8.201166861143523,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.552948865841052e-05,
      "loss": 0.0001,
      "step": 17571
    },
    {
      "epoch": 8.201633605600934,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.5521635850607118e-05,
      "loss": 0.0001,
      "step": 17572
    },
    {
      "epoch": 8.202100350058343,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.5513784861716663e-05,
      "loss": 0.0002,
      "step": 17573
    },
    {
      "epoch": 8.202567094515752,
      "grad_norm": 0.05224609375,
      "learning_rate": 1.550593569190816e-05,
      "loss": 0.0022,
      "step": 17574
    },
    {
      "epoch": 8.203033838973163,
      "grad_norm": 0.05517578125,
      "learning_rate": 1.5498088341350582e-05,
      "loss": 0.0029,
      "step": 17575
    },
    {
      "epoch": 8.203500583430571,
      "grad_norm": 0.007080078125,
      "learning_rate": 1.5490242810212917e-05,
      "loss": 0.0002,
      "step": 17576
    },
    {
      "epoch": 8.203967327887982,
      "grad_norm": 0.03173828125,
      "learning_rate": 1.5482399098664112e-05,
      "loss": 0.0007,
      "step": 17577
    },
    {
      "epoch": 8.204434072345391,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.547455720687301e-05,
      "loss": 0.0001,
      "step": 17578
    },
    {
      "epoch": 8.2049008168028,
      "grad_norm": 0.01025390625,
      "learning_rate": 1.5466717135008468e-05,
      "loss": 0.0001,
      "step": 17579
    },
    {
      "epoch": 8.20536756126021,
      "grad_norm": 0.006561279296875,
      "learning_rate": 1.5458878883239325e-05,
      "loss": 0.0001,
      "step": 17580
    },
    {
      "epoch": 8.20583430571762,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.54510424517343e-05,
      "loss": 0.0001,
      "step": 17581
    },
    {
      "epoch": 8.206301050175028,
      "grad_norm": 0.0625,
      "learning_rate": 1.5443207840662132e-05,
      "loss": 0.003,
      "step": 17582
    },
    {
      "epoch": 8.206767794632439,
      "grad_norm": 0.05029296875,
      "learning_rate": 1.5435375050191547e-05,
      "loss": 0.0018,
      "step": 17583
    },
    {
      "epoch": 8.207234539089848,
      "grad_norm": 0.01263427734375,
      "learning_rate": 1.542754408049113e-05,
      "loss": 0.0002,
      "step": 17584
    },
    {
      "epoch": 8.207701283547259,
      "grad_norm": 0.01422119140625,
      "learning_rate": 1.5419714931729556e-05,
      "loss": 0.0002,
      "step": 17585
    },
    {
      "epoch": 8.208168028004668,
      "grad_norm": 0.021484375,
      "learning_rate": 1.5411887604075338e-05,
      "loss": 0.0002,
      "step": 17586
    },
    {
      "epoch": 8.208634772462077,
      "grad_norm": 0.054443359375,
      "learning_rate": 1.5404062097697024e-05,
      "loss": 0.0011,
      "step": 17587
    },
    {
      "epoch": 8.209101516919487,
      "grad_norm": 0.042236328125,
      "learning_rate": 1.5396238412763154e-05,
      "loss": 0.0026,
      "step": 17588
    },
    {
      "epoch": 8.209568261376896,
      "grad_norm": 0.00830078125,
      "learning_rate": 1.5388416549442098e-05,
      "loss": 0.0001,
      "step": 17589
    },
    {
      "epoch": 8.210035005834305,
      "grad_norm": 0.01458740234375,
      "learning_rate": 1.5380596507902323e-05,
      "loss": 0.0001,
      "step": 17590
    },
    {
      "epoch": 8.210501750291716,
      "grad_norm": 0.031982421875,
      "learning_rate": 1.5372778288312217e-05,
      "loss": 0.0016,
      "step": 17591
    },
    {
      "epoch": 8.210968494749125,
      "grad_norm": 0.0096435546875,
      "learning_rate": 1.5364961890840058e-05,
      "loss": 0.0002,
      "step": 17592
    },
    {
      "epoch": 8.211435239206535,
      "grad_norm": 0.052490234375,
      "learning_rate": 1.5357147315654184e-05,
      "loss": 0.0028,
      "step": 17593
    },
    {
      "epoch": 8.211901983663944,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.5349334562922855e-05,
      "loss": 0.0001,
      "step": 17594
    },
    {
      "epoch": 8.212368728121353,
      "grad_norm": 0.008056640625,
      "learning_rate": 1.5341523632814258e-05,
      "loss": 0.0001,
      "step": 17595
    },
    {
      "epoch": 8.212835472578764,
      "grad_norm": 0.03515625,
      "learning_rate": 1.5333714525496578e-05,
      "loss": 0.0017,
      "step": 17596
    },
    {
      "epoch": 8.213302217036173,
      "grad_norm": 0.01416015625,
      "learning_rate": 1.5325907241137992e-05,
      "loss": 0.0002,
      "step": 17597
    },
    {
      "epoch": 8.213768961493582,
      "grad_norm": 0.0576171875,
      "learning_rate": 1.5318101779906556e-05,
      "loss": 0.0016,
      "step": 17598
    },
    {
      "epoch": 8.214235705950992,
      "grad_norm": 0.05908203125,
      "learning_rate": 1.5310298141970338e-05,
      "loss": 0.0037,
      "step": 17599
    },
    {
      "epoch": 8.214702450408401,
      "grad_norm": 0.005889892578125,
      "learning_rate": 1.5302496327497385e-05,
      "loss": 0.0001,
      "step": 17600
    },
    {
      "epoch": 8.21516919486581,
      "grad_norm": 0.0164794921875,
      "learning_rate": 1.5294696336655634e-05,
      "loss": 0.0001,
      "step": 17601
    },
    {
      "epoch": 8.21563593932322,
      "grad_norm": 0.04248046875,
      "learning_rate": 1.528689816961306e-05,
      "loss": 0.0021,
      "step": 17602
    },
    {
      "epoch": 8.21610268378063,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.527910182653758e-05,
      "loss": 0.0001,
      "step": 17603
    },
    {
      "epoch": 8.21656942823804,
      "grad_norm": 0.011962890625,
      "learning_rate": 1.5271307307597016e-05,
      "loss": 0.0001,
      "step": 17604
    },
    {
      "epoch": 8.21703617269545,
      "grad_norm": 0.0113525390625,
      "learning_rate": 1.526351461295924e-05,
      "loss": 0.0002,
      "step": 17605
    },
    {
      "epoch": 8.217502917152858,
      "grad_norm": 0.01177978515625,
      "learning_rate": 1.5255723742791984e-05,
      "loss": 0.0002,
      "step": 17606
    },
    {
      "epoch": 8.217969661610269,
      "grad_norm": 0.007568359375,
      "learning_rate": 1.5247934697263022e-05,
      "loss": 0.0001,
      "step": 17607
    },
    {
      "epoch": 8.218436406067678,
      "grad_norm": 0.0091552734375,
      "learning_rate": 1.5240147476540079e-05,
      "loss": 0.0001,
      "step": 17608
    },
    {
      "epoch": 8.218903150525087,
      "grad_norm": 0.05126953125,
      "learning_rate": 1.5232362080790785e-05,
      "loss": 0.0018,
      "step": 17609
    },
    {
      "epoch": 8.219369894982497,
      "grad_norm": 0.0220947265625,
      "learning_rate": 1.5224578510182786e-05,
      "loss": 0.0002,
      "step": 17610
    },
    {
      "epoch": 8.219836639439906,
      "grad_norm": 0.03466796875,
      "learning_rate": 1.521679676488369e-05,
      "loss": 0.0026,
      "step": 17611
    },
    {
      "epoch": 8.220303383897317,
      "grad_norm": 0.00567626953125,
      "learning_rate": 1.5209016845061009e-05,
      "loss": 0.0001,
      "step": 17612
    },
    {
      "epoch": 8.220770128354726,
      "grad_norm": 0.028564453125,
      "learning_rate": 1.5201238750882273e-05,
      "loss": 0.0002,
      "step": 17613
    },
    {
      "epoch": 8.221236872812135,
      "grad_norm": 0.007659912109375,
      "learning_rate": 1.5193462482514975e-05,
      "loss": 0.0001,
      "step": 17614
    },
    {
      "epoch": 8.221703617269545,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.51856880401265e-05,
      "loss": 0.0001,
      "step": 17615
    },
    {
      "epoch": 8.222170361726954,
      "grad_norm": 0.02294921875,
      "learning_rate": 1.517791542388427e-05,
      "loss": 0.0002,
      "step": 17616
    },
    {
      "epoch": 8.222637106184363,
      "grad_norm": 0.00982666015625,
      "learning_rate": 1.5170144633955652e-05,
      "loss": 0.0002,
      "step": 17617
    },
    {
      "epoch": 8.223103850641774,
      "grad_norm": 0.048095703125,
      "learning_rate": 1.516237567050791e-05,
      "loss": 0.0027,
      "step": 17618
    },
    {
      "epoch": 8.223570595099183,
      "grad_norm": 0.06689453125,
      "learning_rate": 1.515460853370836e-05,
      "loss": 0.0031,
      "step": 17619
    },
    {
      "epoch": 8.224037339556594,
      "grad_norm": 0.00701904296875,
      "learning_rate": 1.5146843223724249e-05,
      "loss": 0.0002,
      "step": 17620
    },
    {
      "epoch": 8.224504084014002,
      "grad_norm": 0.048828125,
      "learning_rate": 1.513907974072274e-05,
      "loss": 0.002,
      "step": 17621
    },
    {
      "epoch": 8.224970828471411,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.5131318084870983e-05,
      "loss": 0.0002,
      "step": 17622
    },
    {
      "epoch": 8.225437572928822,
      "grad_norm": 0.011962890625,
      "learning_rate": 1.5123558256336135e-05,
      "loss": 0.0002,
      "step": 17623
    },
    {
      "epoch": 8.225904317386231,
      "grad_norm": 0.041259765625,
      "learning_rate": 1.5115800255285217e-05,
      "loss": 0.0028,
      "step": 17624
    },
    {
      "epoch": 8.22637106184364,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.5108044081885308e-05,
      "loss": 0.0002,
      "step": 17625
    },
    {
      "epoch": 8.22683780630105,
      "grad_norm": 0.007720947265625,
      "learning_rate": 1.510028973630343e-05,
      "loss": 0.0001,
      "step": 17626
    },
    {
      "epoch": 8.22730455075846,
      "grad_norm": 0.00811767578125,
      "learning_rate": 1.5092537218706471e-05,
      "loss": 0.0001,
      "step": 17627
    },
    {
      "epoch": 8.22777129521587,
      "grad_norm": 0.044677734375,
      "learning_rate": 1.5084786529261408e-05,
      "loss": 0.0027,
      "step": 17628
    },
    {
      "epoch": 8.228238039673279,
      "grad_norm": 0.0281982421875,
      "learning_rate": 1.507703766813512e-05,
      "loss": 0.002,
      "step": 17629
    },
    {
      "epoch": 8.228704784130688,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.5069290635494415e-05,
      "loss": 0.0001,
      "step": 17630
    },
    {
      "epoch": 8.229171528588099,
      "grad_norm": 0.0908203125,
      "learning_rate": 1.5061545431506118e-05,
      "loss": 0.0041,
      "step": 17631
    },
    {
      "epoch": 8.229638273045508,
      "grad_norm": 0.005340576171875,
      "learning_rate": 1.5053802056337007e-05,
      "loss": 0.0001,
      "step": 17632
    },
    {
      "epoch": 8.230105017502916,
      "grad_norm": 0.023681640625,
      "learning_rate": 1.5046060510153758e-05,
      "loss": 0.0001,
      "step": 17633
    },
    {
      "epoch": 8.230571761960327,
      "grad_norm": 0.0302734375,
      "learning_rate": 1.5038320793123096e-05,
      "loss": 0.0002,
      "step": 17634
    },
    {
      "epoch": 8.231038506417736,
      "grad_norm": 0.007720947265625,
      "learning_rate": 1.5030582905411671e-05,
      "loss": 0.0001,
      "step": 17635
    },
    {
      "epoch": 8.231505250875147,
      "grad_norm": 0.01275634765625,
      "learning_rate": 1.5022846847186045e-05,
      "loss": 0.0002,
      "step": 17636
    },
    {
      "epoch": 8.231971995332556,
      "grad_norm": 0.01373291015625,
      "learning_rate": 1.5015112618612836e-05,
      "loss": 0.0001,
      "step": 17637
    },
    {
      "epoch": 8.232438739789965,
      "grad_norm": 0.00823974609375,
      "learning_rate": 1.5007380219858525e-05,
      "loss": 0.0001,
      "step": 17638
    },
    {
      "epoch": 8.232905484247375,
      "grad_norm": 0.0167236328125,
      "learning_rate": 1.499964965108962e-05,
      "loss": 0.0001,
      "step": 17639
    },
    {
      "epoch": 8.233372228704784,
      "grad_norm": 0.046630859375,
      "learning_rate": 1.499192091247259e-05,
      "loss": 0.0025,
      "step": 17640
    },
    {
      "epoch": 8.233838973162193,
      "grad_norm": 0.006683349609375,
      "learning_rate": 1.498419400417379e-05,
      "loss": 0.0001,
      "step": 17641
    },
    {
      "epoch": 8.234305717619604,
      "grad_norm": 0.0064697265625,
      "learning_rate": 1.4976468926359633e-05,
      "loss": 0.0001,
      "step": 17642
    },
    {
      "epoch": 8.234772462077013,
      "grad_norm": 0.007293701171875,
      "learning_rate": 1.496874567919645e-05,
      "loss": 0.0001,
      "step": 17643
    },
    {
      "epoch": 8.235239206534422,
      "grad_norm": 0.0283203125,
      "learning_rate": 1.4961024262850487e-05,
      "loss": 0.0002,
      "step": 17644
    },
    {
      "epoch": 8.235705950991832,
      "grad_norm": 0.006866455078125,
      "learning_rate": 1.4953304677488033e-05,
      "loss": 0.0001,
      "step": 17645
    },
    {
      "epoch": 8.236172695449241,
      "grad_norm": 0.013427734375,
      "learning_rate": 1.4945586923275313e-05,
      "loss": 0.0001,
      "step": 17646
    },
    {
      "epoch": 8.236639439906652,
      "grad_norm": 0.07666015625,
      "learning_rate": 1.4937871000378445e-05,
      "loss": 0.0065,
      "step": 17647
    },
    {
      "epoch": 8.23710618436406,
      "grad_norm": 0.006500244140625,
      "learning_rate": 1.49301569089636e-05,
      "loss": 0.0001,
      "step": 17648
    },
    {
      "epoch": 8.23757292882147,
      "grad_norm": 0.0096435546875,
      "learning_rate": 1.4922444649196888e-05,
      "loss": 0.0001,
      "step": 17649
    },
    {
      "epoch": 8.23803967327888,
      "grad_norm": 0.09033203125,
      "learning_rate": 1.4914734221244309e-05,
      "loss": 0.0028,
      "step": 17650
    },
    {
      "epoch": 8.23850641773629,
      "grad_norm": 0.005340576171875,
      "learning_rate": 1.4907025625271908e-05,
      "loss": 0.0001,
      "step": 17651
    },
    {
      "epoch": 8.238973162193698,
      "grad_norm": 0.08447265625,
      "learning_rate": 1.4899318861445688e-05,
      "loss": 0.003,
      "step": 17652
    },
    {
      "epoch": 8.239439906651109,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.4891613929931525e-05,
      "loss": 0.0001,
      "step": 17653
    },
    {
      "epoch": 8.239906651108518,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.4883910830895364e-05,
      "loss": 0.0001,
      "step": 17654
    },
    {
      "epoch": 8.240373395565928,
      "grad_norm": 0.01055908203125,
      "learning_rate": 1.4876209564503019e-05,
      "loss": 0.0002,
      "step": 17655
    },
    {
      "epoch": 8.240840140023337,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.4868510130920322e-05,
      "loss": 0.0001,
      "step": 17656
    },
    {
      "epoch": 8.241306884480746,
      "grad_norm": 0.0654296875,
      "learning_rate": 1.4860812530313096e-05,
      "loss": 0.002,
      "step": 17657
    },
    {
      "epoch": 8.241773628938157,
      "grad_norm": 0.0230712890625,
      "learning_rate": 1.4853116762846997e-05,
      "loss": 0.0002,
      "step": 17658
    },
    {
      "epoch": 8.242240373395566,
      "grad_norm": 0.3828125,
      "learning_rate": 1.4845422828687783e-05,
      "loss": 0.001,
      "step": 17659
    },
    {
      "epoch": 8.242707117852975,
      "grad_norm": 0.0081787109375,
      "learning_rate": 1.4837730728001109e-05,
      "loss": 0.0001,
      "step": 17660
    },
    {
      "epoch": 8.243173862310385,
      "grad_norm": 0.01153564453125,
      "learning_rate": 1.4830040460952565e-05,
      "loss": 0.0002,
      "step": 17661
    },
    {
      "epoch": 8.243640606767794,
      "grad_norm": 0.006927490234375,
      "learning_rate": 1.4822352027707743e-05,
      "loss": 0.0001,
      "step": 17662
    },
    {
      "epoch": 8.244107351225205,
      "grad_norm": 0.00421142578125,
      "learning_rate": 1.4814665428432217e-05,
      "loss": 0.0001,
      "step": 17663
    },
    {
      "epoch": 8.244574095682614,
      "grad_norm": 0.0137939453125,
      "learning_rate": 1.4806980663291426e-05,
      "loss": 0.0002,
      "step": 17664
    },
    {
      "epoch": 8.245040840140023,
      "grad_norm": 0.047607421875,
      "learning_rate": 1.4799297732450878e-05,
      "loss": 0.0013,
      "step": 17665
    },
    {
      "epoch": 8.245507584597433,
      "grad_norm": 0.01019287109375,
      "learning_rate": 1.479161663607599e-05,
      "loss": 0.0002,
      "step": 17666
    },
    {
      "epoch": 8.245974329054842,
      "grad_norm": 0.0091552734375,
      "learning_rate": 1.4783937374332124e-05,
      "loss": 0.0001,
      "step": 17667
    },
    {
      "epoch": 8.246441073512251,
      "grad_norm": 0.00762939453125,
      "learning_rate": 1.4776259947384629e-05,
      "loss": 0.0001,
      "step": 17668
    },
    {
      "epoch": 8.246907817969662,
      "grad_norm": 0.00799560546875,
      "learning_rate": 1.4768584355398862e-05,
      "loss": 0.0002,
      "step": 17669
    },
    {
      "epoch": 8.247374562427071,
      "grad_norm": 0.01287841796875,
      "learning_rate": 1.4760910598540001e-05,
      "loss": 0.0002,
      "step": 17670
    },
    {
      "epoch": 8.247841306884482,
      "grad_norm": 0.039306640625,
      "learning_rate": 1.4753238676973302e-05,
      "loss": 0.0002,
      "step": 17671
    },
    {
      "epoch": 8.24830805134189,
      "grad_norm": 0.06787109375,
      "learning_rate": 1.474556859086398e-05,
      "loss": 0.0048,
      "step": 17672
    },
    {
      "epoch": 8.2487747957993,
      "grad_norm": 0.01092529296875,
      "learning_rate": 1.473790034037713e-05,
      "loss": 0.0002,
      "step": 17673
    },
    {
      "epoch": 8.24924154025671,
      "grad_norm": 0.0150146484375,
      "learning_rate": 1.4730233925677895e-05,
      "loss": 0.0001,
      "step": 17674
    },
    {
      "epoch": 8.249708284714119,
      "grad_norm": 0.034423828125,
      "learning_rate": 1.4722569346931358e-05,
      "loss": 0.0015,
      "step": 17675
    },
    {
      "epoch": 8.250175029171528,
      "grad_norm": 0.0517578125,
      "learning_rate": 1.4714906604302492e-05,
      "loss": 0.0003,
      "step": 17676
    },
    {
      "epoch": 8.250641773628939,
      "grad_norm": 0.006866455078125,
      "learning_rate": 1.4707245697956307e-05,
      "loss": 0.0001,
      "step": 17677
    },
    {
      "epoch": 8.251108518086347,
      "grad_norm": 0.046142578125,
      "learning_rate": 1.4699586628057782e-05,
      "loss": 0.0023,
      "step": 17678
    },
    {
      "epoch": 8.251575262543756,
      "grad_norm": 0.0069580078125,
      "learning_rate": 1.4691929394771775e-05,
      "loss": 0.0001,
      "step": 17679
    },
    {
      "epoch": 8.252042007001167,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.4684273998263188e-05,
      "loss": 0.0002,
      "step": 17680
    },
    {
      "epoch": 8.252508751458576,
      "grad_norm": 0.07177734375,
      "learning_rate": 1.4676620438696853e-05,
      "loss": 0.0022,
      "step": 17681
    },
    {
      "epoch": 8.252975495915987,
      "grad_norm": 0.013916015625,
      "learning_rate": 1.466896871623753e-05,
      "loss": 0.0002,
      "step": 17682
    },
    {
      "epoch": 8.253442240373396,
      "grad_norm": 0.041748046875,
      "learning_rate": 1.4661318831049987e-05,
      "loss": 0.0031,
      "step": 17683
    },
    {
      "epoch": 8.253908984830804,
      "grad_norm": 0.00616455078125,
      "learning_rate": 1.4653670783298956e-05,
      "loss": 0.0001,
      "step": 17684
    },
    {
      "epoch": 8.254375729288215,
      "grad_norm": 0.0218505859375,
      "learning_rate": 1.4646024573149064e-05,
      "loss": 0.0002,
      "step": 17685
    },
    {
      "epoch": 8.254842473745624,
      "grad_norm": 0.01031494140625,
      "learning_rate": 1.463838020076499e-05,
      "loss": 0.0001,
      "step": 17686
    },
    {
      "epoch": 8.255309218203033,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.4630737666311278e-05,
      "loss": 0.0001,
      "step": 17687
    },
    {
      "epoch": 8.255775962660444,
      "grad_norm": 0.005279541015625,
      "learning_rate": 1.46230969699525e-05,
      "loss": 0.0001,
      "step": 17688
    },
    {
      "epoch": 8.255775962660444,
      "eval_loss": 2.4444942474365234,
      "eval_runtime": 84.4859,
      "eval_samples_per_second": 21.353,
      "eval_steps_per_second": 2.675,
      "step": 17688
    },
    {
      "epoch": 8.256242707117853,
      "grad_norm": 0.005767822265625,
      "learning_rate": 1.4615458111853197e-05,
      "loss": 0.0001,
      "step": 17689
    },
    {
      "epoch": 8.256709451575263,
      "grad_norm": 0.00830078125,
      "learning_rate": 1.4607821092177786e-05,
      "loss": 0.0001,
      "step": 17690
    },
    {
      "epoch": 8.257176196032672,
      "grad_norm": 0.05126953125,
      "learning_rate": 1.4600185911090736e-05,
      "loss": 0.0021,
      "step": 17691
    },
    {
      "epoch": 8.257642940490081,
      "grad_norm": 0.005950927734375,
      "learning_rate": 1.4592552568756456e-05,
      "loss": 0.0001,
      "step": 17692
    },
    {
      "epoch": 8.258109684947492,
      "grad_norm": 0.006072998046875,
      "learning_rate": 1.4584921065339263e-05,
      "loss": 0.0001,
      "step": 17693
    },
    {
      "epoch": 8.2585764294049,
      "grad_norm": 0.0245361328125,
      "learning_rate": 1.457729140100349e-05,
      "loss": 0.0019,
      "step": 17694
    },
    {
      "epoch": 8.25904317386231,
      "grad_norm": 0.0118408203125,
      "learning_rate": 1.456966357591344e-05,
      "loss": 0.0002,
      "step": 17695
    },
    {
      "epoch": 8.25950991831972,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.456203759023328e-05,
      "loss": 0.0001,
      "step": 17696
    },
    {
      "epoch": 8.25997666277713,
      "grad_norm": 0.00830078125,
      "learning_rate": 1.4554413444127257e-05,
      "loss": 0.0001,
      "step": 17697
    },
    {
      "epoch": 8.26044340723454,
      "grad_norm": 0.01019287109375,
      "learning_rate": 1.4546791137759541e-05,
      "loss": 0.0002,
      "step": 17698
    },
    {
      "epoch": 8.260910151691949,
      "grad_norm": 0.109375,
      "learning_rate": 1.453917067129419e-05,
      "loss": 0.0034,
      "step": 17699
    },
    {
      "epoch": 8.261376896149358,
      "grad_norm": 0.0057373046875,
      "learning_rate": 1.4531552044895325e-05,
      "loss": 0.0001,
      "step": 17700
    },
    {
      "epoch": 8.261843640606768,
      "grad_norm": 0.0115966796875,
      "learning_rate": 1.4523935258726995e-05,
      "loss": 0.0002,
      "step": 17701
    },
    {
      "epoch": 8.262310385064177,
      "grad_norm": 0.00848388671875,
      "learning_rate": 1.4516320312953146e-05,
      "loss": 0.0002,
      "step": 17702
    },
    {
      "epoch": 8.262777129521586,
      "grad_norm": 0.013427734375,
      "learning_rate": 1.4508707207737793e-05,
      "loss": 0.0001,
      "step": 17703
    },
    {
      "epoch": 8.263243873978997,
      "grad_norm": 0.007110595703125,
      "learning_rate": 1.4501095943244803e-05,
      "loss": 0.0001,
      "step": 17704
    },
    {
      "epoch": 8.263710618436406,
      "grad_norm": 0.00823974609375,
      "learning_rate": 1.4493486519638089e-05,
      "loss": 0.0001,
      "step": 17705
    },
    {
      "epoch": 8.264177362893816,
      "grad_norm": 0.00738525390625,
      "learning_rate": 1.4485878937081499e-05,
      "loss": 0.0002,
      "step": 17706
    },
    {
      "epoch": 8.264644107351225,
      "grad_norm": 0.0120849609375,
      "learning_rate": 1.447827319573879e-05,
      "loss": 0.0002,
      "step": 17707
    },
    {
      "epoch": 8.265110851808634,
      "grad_norm": 0.032470703125,
      "learning_rate": 1.4470669295773753e-05,
      "loss": 0.0001,
      "step": 17708
    },
    {
      "epoch": 8.265577596266045,
      "grad_norm": 0.014404296875,
      "learning_rate": 1.4463067237350114e-05,
      "loss": 0.0002,
      "step": 17709
    },
    {
      "epoch": 8.266044340723454,
      "grad_norm": 0.035400390625,
      "learning_rate": 1.4455467020631531e-05,
      "loss": 0.0001,
      "step": 17710
    },
    {
      "epoch": 8.266511085180863,
      "grad_norm": 0.005645751953125,
      "learning_rate": 1.4447868645781648e-05,
      "loss": 0.0001,
      "step": 17711
    },
    {
      "epoch": 8.266977829638273,
      "grad_norm": 0.032958984375,
      "learning_rate": 1.4440272112964103e-05,
      "loss": 0.0016,
      "step": 17712
    },
    {
      "epoch": 8.267444574095682,
      "grad_norm": 0.01104736328125,
      "learning_rate": 1.44326774223424e-05,
      "loss": 0.0001,
      "step": 17713
    },
    {
      "epoch": 8.267911318553093,
      "grad_norm": 0.006134033203125,
      "learning_rate": 1.4425084574080105e-05,
      "loss": 0.0001,
      "step": 17714
    },
    {
      "epoch": 8.268378063010502,
      "grad_norm": 0.004547119140625,
      "learning_rate": 1.4417493568340701e-05,
      "loss": 0.0001,
      "step": 17715
    },
    {
      "epoch": 8.26884480746791,
      "grad_norm": 0.0033416748046875,
      "learning_rate": 1.440990440528759e-05,
      "loss": 0.0001,
      "step": 17716
    },
    {
      "epoch": 8.269311551925322,
      "grad_norm": 0.006072998046875,
      "learning_rate": 1.4402317085084227e-05,
      "loss": 0.0001,
      "step": 17717
    },
    {
      "epoch": 8.26977829638273,
      "grad_norm": 0.055908203125,
      "learning_rate": 1.4394731607893941e-05,
      "loss": 0.0002,
      "step": 17718
    },
    {
      "epoch": 8.27024504084014,
      "grad_norm": 0.00531005859375,
      "learning_rate": 1.4387147973880055e-05,
      "loss": 0.0001,
      "step": 17719
    },
    {
      "epoch": 8.27071178529755,
      "grad_norm": 0.023681640625,
      "learning_rate": 1.437956618320585e-05,
      "loss": 0.0002,
      "step": 17720
    },
    {
      "epoch": 8.271178529754959,
      "grad_norm": 0.01611328125,
      "learning_rate": 1.4371986236034618e-05,
      "loss": 0.0002,
      "step": 17721
    },
    {
      "epoch": 8.271645274212368,
      "grad_norm": 0.0186767578125,
      "learning_rate": 1.4364408132529495e-05,
      "loss": 0.0002,
      "step": 17722
    },
    {
      "epoch": 8.272112018669779,
      "grad_norm": 0.01336669921875,
      "learning_rate": 1.4356831872853682e-05,
      "loss": 0.0001,
      "step": 17723
    },
    {
      "epoch": 8.272578763127187,
      "grad_norm": 0.053466796875,
      "learning_rate": 1.4349257457170328e-05,
      "loss": 0.0033,
      "step": 17724
    },
    {
      "epoch": 8.273045507584598,
      "grad_norm": 0.01080322265625,
      "learning_rate": 1.4341684885642459e-05,
      "loss": 0.0001,
      "step": 17725
    },
    {
      "epoch": 8.273512252042007,
      "grad_norm": 0.014404296875,
      "learning_rate": 1.4334114158433165e-05,
      "loss": 0.0001,
      "step": 17726
    },
    {
      "epoch": 8.273978996499416,
      "grad_norm": 0.01300048828125,
      "learning_rate": 1.432654527570546e-05,
      "loss": 0.0002,
      "step": 17727
    },
    {
      "epoch": 8.274445740956827,
      "grad_norm": 0.0140380859375,
      "learning_rate": 1.4318978237622272e-05,
      "loss": 0.0002,
      "step": 17728
    },
    {
      "epoch": 8.274912485414236,
      "grad_norm": 0.00799560546875,
      "learning_rate": 1.4311413044346544e-05,
      "loss": 0.0001,
      "step": 17729
    },
    {
      "epoch": 8.275379229871644,
      "grad_norm": 0.05615234375,
      "learning_rate": 1.4303849696041182e-05,
      "loss": 0.0032,
      "step": 17730
    },
    {
      "epoch": 8.275845974329055,
      "grad_norm": 0.0059814453125,
      "learning_rate": 1.4296288192869012e-05,
      "loss": 0.0001,
      "step": 17731
    },
    {
      "epoch": 8.276312718786464,
      "grad_norm": 0.0125732421875,
      "learning_rate": 1.4288728534992834e-05,
      "loss": 0.0001,
      "step": 17732
    },
    {
      "epoch": 8.276779463243875,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.4281170722575455e-05,
      "loss": 0.0001,
      "step": 17733
    },
    {
      "epoch": 8.277246207701284,
      "grad_norm": 0.0174560546875,
      "learning_rate": 1.4273614755779552e-05,
      "loss": 0.0001,
      "step": 17734
    },
    {
      "epoch": 8.277712952158693,
      "grad_norm": 0.0234375,
      "learning_rate": 1.4266060634767864e-05,
      "loss": 0.0002,
      "step": 17735
    },
    {
      "epoch": 8.278179696616103,
      "grad_norm": 0.013427734375,
      "learning_rate": 1.4258508359702993e-05,
      "loss": 0.0002,
      "step": 17736
    },
    {
      "epoch": 8.278646441073512,
      "grad_norm": 0.04736328125,
      "learning_rate": 1.4250957930747567e-05,
      "loss": 0.0013,
      "step": 17737
    },
    {
      "epoch": 8.279113185530921,
      "grad_norm": 0.08349609375,
      "learning_rate": 1.4243409348064174e-05,
      "loss": 0.0041,
      "step": 17738
    },
    {
      "epoch": 8.279579929988332,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.423586261181531e-05,
      "loss": 0.0001,
      "step": 17739
    },
    {
      "epoch": 8.28004667444574,
      "grad_norm": 0.031005859375,
      "learning_rate": 1.4228317722163476e-05,
      "loss": 0.0002,
      "step": 17740
    },
    {
      "epoch": 8.280513418903151,
      "grad_norm": 0.00537109375,
      "learning_rate": 1.4220774679271153e-05,
      "loss": 0.0001,
      "step": 17741
    },
    {
      "epoch": 8.28098016336056,
      "grad_norm": 0.00390625,
      "learning_rate": 1.4213233483300703e-05,
      "loss": 0.0001,
      "step": 17742
    },
    {
      "epoch": 8.281446907817969,
      "grad_norm": 0.014404296875,
      "learning_rate": 1.4205694134414527e-05,
      "loss": 0.0002,
      "step": 17743
    },
    {
      "epoch": 8.28191365227538,
      "grad_norm": 0.01068115234375,
      "learning_rate": 1.419815663277495e-05,
      "loss": 0.0001,
      "step": 17744
    },
    {
      "epoch": 8.282380396732789,
      "grad_norm": 0.01116943359375,
      "learning_rate": 1.4190620978544256e-05,
      "loss": 0.0001,
      "step": 17745
    },
    {
      "epoch": 8.282847141190198,
      "grad_norm": 0.0380859375,
      "learning_rate": 1.418308717188469e-05,
      "loss": 0.002,
      "step": 17746
    },
    {
      "epoch": 8.283313885647608,
      "grad_norm": 0.0546875,
      "learning_rate": 1.4175555212958491e-05,
      "loss": 0.0041,
      "step": 17747
    },
    {
      "epoch": 8.283780630105017,
      "grad_norm": 0.0224609375,
      "learning_rate": 1.4168025101927795e-05,
      "loss": 0.0001,
      "step": 17748
    },
    {
      "epoch": 8.284247374562428,
      "grad_norm": 0.004486083984375,
      "learning_rate": 1.4160496838954752e-05,
      "loss": 0.0001,
      "step": 17749
    },
    {
      "epoch": 8.284714119019837,
      "grad_norm": 0.055419921875,
      "learning_rate": 1.4152970424201472e-05,
      "loss": 0.0033,
      "step": 17750
    },
    {
      "epoch": 8.285180863477246,
      "grad_norm": 0.0072021484375,
      "learning_rate": 1.4145445857829964e-05,
      "loss": 0.0002,
      "step": 17751
    },
    {
      "epoch": 8.285647607934656,
      "grad_norm": 0.01165771484375,
      "learning_rate": 1.4137923140002285e-05,
      "loss": 0.0001,
      "step": 17752
    },
    {
      "epoch": 8.286114352392065,
      "grad_norm": 0.008056640625,
      "learning_rate": 1.4130402270880372e-05,
      "loss": 0.0002,
      "step": 17753
    },
    {
      "epoch": 8.286581096849474,
      "grad_norm": 0.010986328125,
      "learning_rate": 1.4122883250626161e-05,
      "loss": 0.0001,
      "step": 17754
    },
    {
      "epoch": 8.287047841306885,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.4115366079401593e-05,
      "loss": 0.0001,
      "step": 17755
    },
    {
      "epoch": 8.287514585764294,
      "grad_norm": 0.03271484375,
      "learning_rate": 1.4107850757368446e-05,
      "loss": 0.0024,
      "step": 17756
    },
    {
      "epoch": 8.287981330221704,
      "grad_norm": 0.046875,
      "learning_rate": 1.410033728468858e-05,
      "loss": 0.0002,
      "step": 17757
    },
    {
      "epoch": 8.288448074679113,
      "grad_norm": 0.0303955078125,
      "learning_rate": 1.4092825661523779e-05,
      "loss": 0.0003,
      "step": 17758
    },
    {
      "epoch": 8.288914819136522,
      "grad_norm": 0.0103759765625,
      "learning_rate": 1.4085315888035744e-05,
      "loss": 0.0001,
      "step": 17759
    },
    {
      "epoch": 8.289381563593933,
      "grad_norm": 0.00628662109375,
      "learning_rate": 1.4077807964386169e-05,
      "loss": 0.0001,
      "step": 17760
    },
    {
      "epoch": 8.289848308051342,
      "grad_norm": 0.01336669921875,
      "learning_rate": 1.4070301890736748e-05,
      "loss": 0.0002,
      "step": 17761
    },
    {
      "epoch": 8.29031505250875,
      "grad_norm": 0.060791015625,
      "learning_rate": 1.4062797667249051e-05,
      "loss": 0.0018,
      "step": 17762
    },
    {
      "epoch": 8.290781796966161,
      "grad_norm": 0.00640869140625,
      "learning_rate": 1.4055295294084658e-05,
      "loss": 0.0001,
      "step": 17763
    },
    {
      "epoch": 8.29124854142357,
      "grad_norm": 0.0078125,
      "learning_rate": 1.4047794771405143e-05,
      "loss": 0.0002,
      "step": 17764
    },
    {
      "epoch": 8.29171528588098,
      "grad_norm": 0.0068359375,
      "learning_rate": 1.4040296099371975e-05,
      "loss": 0.0001,
      "step": 17765
    },
    {
      "epoch": 8.29218203033839,
      "grad_norm": 0.00909423828125,
      "learning_rate": 1.403279927814658e-05,
      "loss": 0.0001,
      "step": 17766
    },
    {
      "epoch": 8.292648774795799,
      "grad_norm": 0.005157470703125,
      "learning_rate": 1.402530430789043e-05,
      "loss": 0.0001,
      "step": 17767
    },
    {
      "epoch": 8.29311551925321,
      "grad_norm": 0.01275634765625,
      "learning_rate": 1.401781118876484e-05,
      "loss": 0.0001,
      "step": 17768
    },
    {
      "epoch": 8.293582263710618,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.401031992093117e-05,
      "loss": 0.0002,
      "step": 17769
    },
    {
      "epoch": 8.294049008168027,
      "grad_norm": 0.0245361328125,
      "learning_rate": 1.4002830504550746e-05,
      "loss": 0.0002,
      "step": 17770
    },
    {
      "epoch": 8.294515752625438,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.3995342939784772e-05,
      "loss": 0.0002,
      "step": 17771
    },
    {
      "epoch": 8.294982497082847,
      "grad_norm": 0.0103759765625,
      "learning_rate": 1.3987857226794487e-05,
      "loss": 0.0002,
      "step": 17772
    },
    {
      "epoch": 8.295449241540256,
      "grad_norm": 0.005035400390625,
      "learning_rate": 1.3980373365741095e-05,
      "loss": 0.0001,
      "step": 17773
    },
    {
      "epoch": 8.295915985997667,
      "grad_norm": 0.035888671875,
      "learning_rate": 1.3972891356785678e-05,
      "loss": 0.0002,
      "step": 17774
    },
    {
      "epoch": 8.296382730455075,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.3965411200089362e-05,
      "loss": 0.0001,
      "step": 17775
    },
    {
      "epoch": 8.296849474912486,
      "grad_norm": 0.010009765625,
      "learning_rate": 1.3957932895813219e-05,
      "loss": 0.0001,
      "step": 17776
    },
    {
      "epoch": 8.297316219369895,
      "grad_norm": 0.0101318359375,
      "learning_rate": 1.395045644411822e-05,
      "loss": 0.0001,
      "step": 17777
    },
    {
      "epoch": 8.297782963827304,
      "grad_norm": 0.0118408203125,
      "learning_rate": 1.3942981845165382e-05,
      "loss": 0.0001,
      "step": 17778
    },
    {
      "epoch": 8.298249708284715,
      "grad_norm": 0.005889892578125,
      "learning_rate": 1.393550909911564e-05,
      "loss": 0.0001,
      "step": 17779
    },
    {
      "epoch": 8.298716452742124,
      "grad_norm": 0.00830078125,
      "learning_rate": 1.3928038206129845e-05,
      "loss": 0.0001,
      "step": 17780
    },
    {
      "epoch": 8.299183197199532,
      "grad_norm": 0.11376953125,
      "learning_rate": 1.3920569166368903e-05,
      "loss": 0.0002,
      "step": 17781
    },
    {
      "epoch": 8.299649941656943,
      "grad_norm": 0.0101318359375,
      "learning_rate": 1.3913101979993625e-05,
      "loss": 0.0002,
      "step": 17782
    },
    {
      "epoch": 8.300116686114352,
      "grad_norm": 0.0400390625,
      "learning_rate": 1.3905636647164766e-05,
      "loss": 0.0022,
      "step": 17783
    },
    {
      "epoch": 8.300583430571763,
      "grad_norm": 0.007110595703125,
      "learning_rate": 1.3898173168043083e-05,
      "loss": 0.0001,
      "step": 17784
    },
    {
      "epoch": 8.301050175029172,
      "grad_norm": 0.006317138671875,
      "learning_rate": 1.3890711542789247e-05,
      "loss": 0.0001,
      "step": 17785
    },
    {
      "epoch": 8.30151691948658,
      "grad_norm": 0.02197265625,
      "learning_rate": 1.3883251771563932e-05,
      "loss": 0.0002,
      "step": 17786
    },
    {
      "epoch": 8.301983663943991,
      "grad_norm": 0.01025390625,
      "learning_rate": 1.3875793854527774e-05,
      "loss": 0.0001,
      "step": 17787
    },
    {
      "epoch": 8.3024504084014,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.3868337791841302e-05,
      "loss": 0.0002,
      "step": 17788
    },
    {
      "epoch": 8.302917152858809,
      "grad_norm": 0.005767822265625,
      "learning_rate": 1.3860883583665096e-05,
      "loss": 0.0001,
      "step": 17789
    },
    {
      "epoch": 8.30338389731622,
      "grad_norm": 0.008056640625,
      "learning_rate": 1.3853431230159653e-05,
      "loss": 0.0002,
      "step": 17790
    },
    {
      "epoch": 8.303850641773629,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.3845980731485398e-05,
      "loss": 0.0001,
      "step": 17791
    },
    {
      "epoch": 8.30431738623104,
      "grad_norm": 0.017822265625,
      "learning_rate": 1.3838532087802768e-05,
      "loss": 0.0002,
      "step": 17792
    },
    {
      "epoch": 8.304784130688448,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.3831085299272162e-05,
      "loss": 0.0002,
      "step": 17793
    },
    {
      "epoch": 8.305250875145857,
      "grad_norm": 0.0033416748046875,
      "learning_rate": 1.3823640366053869e-05,
      "loss": 0.0001,
      "step": 17794
    },
    {
      "epoch": 8.305717619603268,
      "grad_norm": 0.019287109375,
      "learning_rate": 1.381619728830822e-05,
      "loss": 0.0001,
      "step": 17795
    },
    {
      "epoch": 8.306184364060677,
      "grad_norm": 0.0111083984375,
      "learning_rate": 1.3808756066195483e-05,
      "loss": 0.0002,
      "step": 17796
    },
    {
      "epoch": 8.306651108518086,
      "grad_norm": 0.01214599609375,
      "learning_rate": 1.3801316699875833e-05,
      "loss": 0.0002,
      "step": 17797
    },
    {
      "epoch": 8.307117852975496,
      "grad_norm": 0.039306640625,
      "learning_rate": 1.3793879189509484e-05,
      "loss": 0.0032,
      "step": 17798
    },
    {
      "epoch": 8.307584597432905,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.3786443535256577e-05,
      "loss": 0.0001,
      "step": 17799
    },
    {
      "epoch": 8.308051341890316,
      "grad_norm": 0.032470703125,
      "learning_rate": 1.377900973727717e-05,
      "loss": 0.0002,
      "step": 17800
    },
    {
      "epoch": 8.308518086347725,
      "grad_norm": 0.0206298828125,
      "learning_rate": 1.3771577795731361e-05,
      "loss": 0.0002,
      "step": 17801
    },
    {
      "epoch": 8.308984830805134,
      "grad_norm": 0.083984375,
      "learning_rate": 1.3764147710779162e-05,
      "loss": 0.0002,
      "step": 17802
    },
    {
      "epoch": 8.309451575262544,
      "grad_norm": 0.0068359375,
      "learning_rate": 1.3756719482580539e-05,
      "loss": 0.0001,
      "step": 17803
    },
    {
      "epoch": 8.309918319719953,
      "grad_norm": 0.004364013671875,
      "learning_rate": 1.3749293111295437e-05,
      "loss": 0.0001,
      "step": 17804
    },
    {
      "epoch": 8.310385064177362,
      "grad_norm": 0.012451171875,
      "learning_rate": 1.374186859708374e-05,
      "loss": 0.0002,
      "step": 17805
    },
    {
      "epoch": 8.310851808634773,
      "grad_norm": 0.0081787109375,
      "learning_rate": 1.3734445940105312e-05,
      "loss": 0.0001,
      "step": 17806
    },
    {
      "epoch": 8.311318553092182,
      "grad_norm": 0.03076171875,
      "learning_rate": 1.372702514052001e-05,
      "loss": 0.0022,
      "step": 17807
    },
    {
      "epoch": 8.31178529754959,
      "grad_norm": 0.0439453125,
      "learning_rate": 1.3719606198487544e-05,
      "loss": 0.003,
      "step": 17808
    },
    {
      "epoch": 8.312252042007001,
      "grad_norm": 0.06201171875,
      "learning_rate": 1.3712189114167684e-05,
      "loss": 0.0025,
      "step": 17809
    },
    {
      "epoch": 8.31271878646441,
      "grad_norm": 0.024658203125,
      "learning_rate": 1.3704773887720169e-05,
      "loss": 0.0002,
      "step": 17810
    },
    {
      "epoch": 8.313185530921821,
      "grad_norm": 0.0233154296875,
      "learning_rate": 1.3697360519304581e-05,
      "loss": 0.0002,
      "step": 17811
    },
    {
      "epoch": 8.31365227537923,
      "grad_norm": 0.02001953125,
      "learning_rate": 1.3689949009080605e-05,
      "loss": 0.0002,
      "step": 17812
    },
    {
      "epoch": 8.314119019836639,
      "grad_norm": 0.005615234375,
      "learning_rate": 1.3682539357207758e-05,
      "loss": 0.0001,
      "step": 17813
    },
    {
      "epoch": 8.31458576429405,
      "grad_norm": 0.008544921875,
      "learning_rate": 1.3675131563845644e-05,
      "loss": 0.0001,
      "step": 17814
    },
    {
      "epoch": 8.315052508751458,
      "grad_norm": 0.01171875,
      "learning_rate": 1.3667725629153694e-05,
      "loss": 0.0002,
      "step": 17815
    },
    {
      "epoch": 8.315519253208867,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.3660321553291423e-05,
      "loss": 0.0002,
      "step": 17816
    },
    {
      "epoch": 8.315985997666278,
      "grad_norm": 0.01129150390625,
      "learning_rate": 1.3652919336418201e-05,
      "loss": 0.0001,
      "step": 17817
    },
    {
      "epoch": 8.316452742123687,
      "grad_norm": 0.0167236328125,
      "learning_rate": 1.3645518978693439e-05,
      "loss": 0.0002,
      "step": 17818
    },
    {
      "epoch": 8.316919486581098,
      "grad_norm": 0.0390625,
      "learning_rate": 1.3638120480276484e-05,
      "loss": 0.0028,
      "step": 17819
    },
    {
      "epoch": 8.317386231038506,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.3630723841326587e-05,
      "loss": 0.0001,
      "step": 17820
    },
    {
      "epoch": 8.317852975495915,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.3623329062003032e-05,
      "loss": 0.0001,
      "step": 17821
    },
    {
      "epoch": 8.318319719953326,
      "grad_norm": 0.007659912109375,
      "learning_rate": 1.3615936142465068e-05,
      "loss": 0.0001,
      "step": 17822
    },
    {
      "epoch": 8.318786464410735,
      "grad_norm": 0.032470703125,
      "learning_rate": 1.3608545082871827e-05,
      "loss": 0.0003,
      "step": 17823
    },
    {
      "epoch": 8.319253208868144,
      "grad_norm": 0.01007080078125,
      "learning_rate": 1.3601155883382454e-05,
      "loss": 0.0002,
      "step": 17824
    },
    {
      "epoch": 8.319719953325555,
      "grad_norm": 0.009033203125,
      "learning_rate": 1.3593768544156082e-05,
      "loss": 0.0001,
      "step": 17825
    },
    {
      "epoch": 8.320186697782963,
      "grad_norm": 0.012451171875,
      "learning_rate": 1.3586383065351715e-05,
      "loss": 0.0002,
      "step": 17826
    },
    {
      "epoch": 8.320653442240374,
      "grad_norm": 0.0137939453125,
      "learning_rate": 1.3578999447128404e-05,
      "loss": 0.0003,
      "step": 17827
    },
    {
      "epoch": 8.321120186697783,
      "grad_norm": 0.03564453125,
      "learning_rate": 1.3571617689645144e-05,
      "loss": 0.0015,
      "step": 17828
    },
    {
      "epoch": 8.321586931155192,
      "grad_norm": 0.01214599609375,
      "learning_rate": 1.3564237793060819e-05,
      "loss": 0.0002,
      "step": 17829
    },
    {
      "epoch": 8.322053675612603,
      "grad_norm": 0.02001953125,
      "learning_rate": 1.3556859757534357e-05,
      "loss": 0.0001,
      "step": 17830
    },
    {
      "epoch": 8.322520420070012,
      "grad_norm": 0.007598876953125,
      "learning_rate": 1.3549483583224642e-05,
      "loss": 0.0001,
      "step": 17831
    },
    {
      "epoch": 8.32298716452742,
      "grad_norm": 0.0167236328125,
      "learning_rate": 1.3542109270290448e-05,
      "loss": 0.0001,
      "step": 17832
    },
    {
      "epoch": 8.323453908984831,
      "grad_norm": 0.00738525390625,
      "learning_rate": 1.353473681889058e-05,
      "loss": 0.0001,
      "step": 17833
    },
    {
      "epoch": 8.32392065344224,
      "grad_norm": 0.016357421875,
      "learning_rate": 1.3527366229183747e-05,
      "loss": 0.0002,
      "step": 17834
    },
    {
      "epoch": 8.32438739789965,
      "grad_norm": 0.0142822265625,
      "learning_rate": 1.3519997501328652e-05,
      "loss": 0.0001,
      "step": 17835
    },
    {
      "epoch": 8.32485414235706,
      "grad_norm": 0.00762939453125,
      "learning_rate": 1.3512630635483992e-05,
      "loss": 0.0001,
      "step": 17836
    },
    {
      "epoch": 8.325320886814469,
      "grad_norm": 0.039306640625,
      "learning_rate": 1.3505265631808338e-05,
      "loss": 0.0015,
      "step": 17837
    },
    {
      "epoch": 8.32578763127188,
      "grad_norm": 0.01080322265625,
      "learning_rate": 1.3497902490460268e-05,
      "loss": 0.0001,
      "step": 17838
    },
    {
      "epoch": 8.326254375729288,
      "grad_norm": 0.0255126953125,
      "learning_rate": 1.3490541211598362e-05,
      "loss": 0.0002,
      "step": 17839
    },
    {
      "epoch": 8.326721120186697,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.3483181795381073e-05,
      "loss": 0.0002,
      "step": 17840
    },
    {
      "epoch": 8.327187864644108,
      "grad_norm": 0.30859375,
      "learning_rate": 1.3475824241966862e-05,
      "loss": 0.0005,
      "step": 17841
    },
    {
      "epoch": 8.327654609101517,
      "grad_norm": 0.00732421875,
      "learning_rate": 1.346846855151418e-05,
      "loss": 0.0002,
      "step": 17842
    },
    {
      "epoch": 8.328121353558927,
      "grad_norm": 0.1201171875,
      "learning_rate": 1.3461114724181356e-05,
      "loss": 0.0003,
      "step": 17843
    },
    {
      "epoch": 8.328588098016336,
      "grad_norm": 0.00518798828125,
      "learning_rate": 1.3453762760126753e-05,
      "loss": 0.0001,
      "step": 17844
    },
    {
      "epoch": 8.329054842473745,
      "grad_norm": 0.00860595703125,
      "learning_rate": 1.3446412659508679e-05,
      "loss": 0.0001,
      "step": 17845
    },
    {
      "epoch": 8.329521586931156,
      "grad_norm": 0.034912109375,
      "learning_rate": 1.343906442248536e-05,
      "loss": 0.0016,
      "step": 17846
    },
    {
      "epoch": 8.329988331388565,
      "grad_norm": 0.0064697265625,
      "learning_rate": 1.3431718049215014e-05,
      "loss": 0.0001,
      "step": 17847
    },
    {
      "epoch": 8.330455075845974,
      "grad_norm": 0.0034332275390625,
      "learning_rate": 1.342437353985585e-05,
      "loss": 0.0001,
      "step": 17848
    },
    {
      "epoch": 8.330921820303384,
      "grad_norm": 0.0164794921875,
      "learning_rate": 1.3417030894565962e-05,
      "loss": 0.0002,
      "step": 17849
    },
    {
      "epoch": 8.331388564760793,
      "grad_norm": 0.01007080078125,
      "learning_rate": 1.3409690113503471e-05,
      "loss": 0.0001,
      "step": 17850
    },
    {
      "epoch": 8.331855309218202,
      "grad_norm": 0.0281982421875,
      "learning_rate": 1.3402351196826445e-05,
      "loss": 0.0019,
      "step": 17851
    },
    {
      "epoch": 8.332322053675613,
      "grad_norm": 0.09228515625,
      "learning_rate": 1.3395014144692853e-05,
      "loss": 0.0051,
      "step": 17852
    },
    {
      "epoch": 8.332788798133022,
      "grad_norm": 0.046142578125,
      "learning_rate": 1.3387678957260718e-05,
      "loss": 0.0016,
      "step": 17853
    },
    {
      "epoch": 8.333255542590432,
      "grad_norm": 0.017578125,
      "learning_rate": 1.338034563468793e-05,
      "loss": 0.0016,
      "step": 17854
    },
    {
      "epoch": 8.333722287047841,
      "grad_norm": 0.0042724609375,
      "learning_rate": 1.3373014177132415e-05,
      "loss": 0.0001,
      "step": 17855
    },
    {
      "epoch": 8.33418903150525,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.3365684584752035e-05,
      "loss": 0.0001,
      "step": 17856
    },
    {
      "epoch": 8.334655775962661,
      "grad_norm": 0.0057373046875,
      "learning_rate": 1.3358356857704568e-05,
      "loss": 0.0001,
      "step": 17857
    },
    {
      "epoch": 8.33512252042007,
      "grad_norm": 0.007537841796875,
      "learning_rate": 1.3351030996147807e-05,
      "loss": 0.0002,
      "step": 17858
    },
    {
      "epoch": 8.335589264877479,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.3343707000239502e-05,
      "loss": 0.0002,
      "step": 17859
    },
    {
      "epoch": 8.33605600933489,
      "grad_norm": 0.015869140625,
      "learning_rate": 1.3336384870137342e-05,
      "loss": 0.0002,
      "step": 17860
    },
    {
      "epoch": 8.336522753792298,
      "grad_norm": 0.01361083984375,
      "learning_rate": 1.3329064605998954e-05,
      "loss": 0.0001,
      "step": 17861
    },
    {
      "epoch": 8.336989498249709,
      "grad_norm": 0.0052490234375,
      "learning_rate": 1.3321746207981956e-05,
      "loss": 0.0001,
      "step": 17862
    },
    {
      "epoch": 8.337456242707118,
      "grad_norm": 0.005218505859375,
      "learning_rate": 1.3314429676243956e-05,
      "loss": 0.0001,
      "step": 17863
    },
    {
      "epoch": 8.337922987164527,
      "grad_norm": 0.038818359375,
      "learning_rate": 1.330711501094245e-05,
      "loss": 0.0002,
      "step": 17864
    },
    {
      "epoch": 8.338389731621938,
      "grad_norm": 0.038818359375,
      "learning_rate": 1.3299802212234957e-05,
      "loss": 0.002,
      "step": 17865
    },
    {
      "epoch": 8.338856476079346,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.3292491280278907e-05,
      "loss": 0.0001,
      "step": 17866
    },
    {
      "epoch": 8.339323220536755,
      "grad_norm": 0.0159912109375,
      "learning_rate": 1.3285182215231718e-05,
      "loss": 0.0002,
      "step": 17867
    },
    {
      "epoch": 8.339789964994166,
      "grad_norm": 0.00732421875,
      "learning_rate": 1.3277875017250795e-05,
      "loss": 0.0001,
      "step": 17868
    },
    {
      "epoch": 8.340256709451575,
      "grad_norm": 0.07470703125,
      "learning_rate": 1.3270569686493417e-05,
      "loss": 0.0022,
      "step": 17869
    },
    {
      "epoch": 8.340723453908986,
      "grad_norm": 0.00933837890625,
      "learning_rate": 1.3263266223116899e-05,
      "loss": 0.0002,
      "step": 17870
    },
    {
      "epoch": 8.341190198366395,
      "grad_norm": 0.011474609375,
      "learning_rate": 1.3255964627278516e-05,
      "loss": 0.0001,
      "step": 17871
    },
    {
      "epoch": 8.341656942823803,
      "grad_norm": 0.0142822265625,
      "learning_rate": 1.3248664899135444e-05,
      "loss": 0.0002,
      "step": 17872
    },
    {
      "epoch": 8.342123687281214,
      "grad_norm": 0.006500244140625,
      "learning_rate": 1.3241367038844865e-05,
      "loss": 0.0001,
      "step": 17873
    },
    {
      "epoch": 8.342590431738623,
      "grad_norm": 0.043212890625,
      "learning_rate": 1.3234071046563934e-05,
      "loss": 0.002,
      "step": 17874
    },
    {
      "epoch": 8.343057176196032,
      "grad_norm": 0.00592041015625,
      "learning_rate": 1.3226776922449703e-05,
      "loss": 0.0001,
      "step": 17875
    },
    {
      "epoch": 8.343523920653443,
      "grad_norm": 0.0079345703125,
      "learning_rate": 1.3219484666659232e-05,
      "loss": 0.0001,
      "step": 17876
    },
    {
      "epoch": 8.343990665110852,
      "grad_norm": 0.0177001953125,
      "learning_rate": 1.3212194279349576e-05,
      "loss": 0.0001,
      "step": 17877
    },
    {
      "epoch": 8.344457409568262,
      "grad_norm": 0.005645751953125,
      "learning_rate": 1.3204905760677643e-05,
      "loss": 0.0001,
      "step": 17878
    },
    {
      "epoch": 8.344924154025671,
      "grad_norm": 0.00732421875,
      "learning_rate": 1.3197619110800386e-05,
      "loss": 0.0001,
      "step": 17879
    },
    {
      "epoch": 8.34539089848308,
      "grad_norm": 0.004913330078125,
      "learning_rate": 1.319033432987472e-05,
      "loss": 0.0001,
      "step": 17880
    },
    {
      "epoch": 8.34585764294049,
      "grad_norm": 0.05126953125,
      "learning_rate": 1.3183051418057457e-05,
      "loss": 0.0014,
      "step": 17881
    },
    {
      "epoch": 8.3463243873979,
      "grad_norm": 0.033935546875,
      "learning_rate": 1.3175770375505448e-05,
      "loss": 0.0018,
      "step": 17882
    },
    {
      "epoch": 8.346791131855309,
      "grad_norm": 0.02197265625,
      "learning_rate": 1.316849120237541e-05,
      "loss": 0.0002,
      "step": 17883
    },
    {
      "epoch": 8.34725787631272,
      "grad_norm": 0.01361083984375,
      "learning_rate": 1.3161213898824088e-05,
      "loss": 0.0001,
      "step": 17884
    },
    {
      "epoch": 8.347724620770128,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.3153938465008208e-05,
      "loss": 0.0002,
      "step": 17885
    },
    {
      "epoch": 8.348191365227539,
      "grad_norm": 0.0140380859375,
      "learning_rate": 1.3146664901084372e-05,
      "loss": 0.0001,
      "step": 17886
    },
    {
      "epoch": 8.348658109684948,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.3139393207209205e-05,
      "loss": 0.0002,
      "step": 17887
    },
    {
      "epoch": 8.349124854142357,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.3132123383539297e-05,
      "loss": 0.0002,
      "step": 17888
    },
    {
      "epoch": 8.349591598599767,
      "grad_norm": 0.046630859375,
      "learning_rate": 1.3124855430231131e-05,
      "loss": 0.0023,
      "step": 17889
    },
    {
      "epoch": 8.350058343057176,
      "grad_norm": 0.00946044921875,
      "learning_rate": 1.3117589347441217e-05,
      "loss": 0.0001,
      "step": 17890
    },
    {
      "epoch": 8.350525087514585,
      "grad_norm": 0.01214599609375,
      "learning_rate": 1.311032513532603e-05,
      "loss": 0.0001,
      "step": 17891
    },
    {
      "epoch": 8.350991831971996,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.310306279404192e-05,
      "loss": 0.0001,
      "step": 17892
    },
    {
      "epoch": 8.351458576429405,
      "grad_norm": 0.040283203125,
      "learning_rate": 1.3095802323745277e-05,
      "loss": 0.0011,
      "step": 17893
    },
    {
      "epoch": 8.351925320886814,
      "grad_norm": 0.0439453125,
      "learning_rate": 1.3088543724592462e-05,
      "loss": 0.002,
      "step": 17894
    },
    {
      "epoch": 8.352392065344224,
      "grad_norm": 0.05029296875,
      "learning_rate": 1.3081286996739705e-05,
      "loss": 0.0031,
      "step": 17895
    },
    {
      "epoch": 8.352858809801633,
      "grad_norm": 0.033203125,
      "learning_rate": 1.3074032140343284e-05,
      "loss": 0.0002,
      "step": 17896
    },
    {
      "epoch": 8.353325554259044,
      "grad_norm": 0.003753662109375,
      "learning_rate": 1.3066779155559416e-05,
      "loss": 0.0001,
      "step": 17897
    },
    {
      "epoch": 8.353792298716453,
      "grad_norm": 0.006011962890625,
      "learning_rate": 1.3059528042544223e-05,
      "loss": 0.0001,
      "step": 17898
    },
    {
      "epoch": 8.354259043173862,
      "grad_norm": 0.006072998046875,
      "learning_rate": 1.3052278801453854e-05,
      "loss": 0.0001,
      "step": 17899
    },
    {
      "epoch": 8.354725787631272,
      "grad_norm": 0.02099609375,
      "learning_rate": 1.304503143244441e-05,
      "loss": 0.0001,
      "step": 17900
    },
    {
      "epoch": 8.355192532088681,
      "grad_norm": 0.010498046875,
      "learning_rate": 1.3037785935671897e-05,
      "loss": 0.0001,
      "step": 17901
    },
    {
      "epoch": 8.35565927654609,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.303054231129236e-05,
      "loss": 0.0001,
      "step": 17902
    },
    {
      "epoch": 8.3561260210035,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.3023300559461726e-05,
      "loss": 0.0001,
      "step": 17903
    },
    {
      "epoch": 8.35659276546091,
      "grad_norm": 0.0140380859375,
      "learning_rate": 1.301606068033593e-05,
      "loss": 0.0001,
      "step": 17904
    },
    {
      "epoch": 8.35705950991832,
      "grad_norm": 0.011474609375,
      "learning_rate": 1.3008822674070874e-05,
      "loss": 0.0002,
      "step": 17905
    },
    {
      "epoch": 8.35752625437573,
      "grad_norm": 0.00634765625,
      "learning_rate": 1.3001586540822374e-05,
      "loss": 0.0002,
      "step": 17906
    },
    {
      "epoch": 8.357992998833138,
      "grad_norm": 0.055908203125,
      "learning_rate": 1.2994352280746235e-05,
      "loss": 0.0015,
      "step": 17907
    },
    {
      "epoch": 8.358459743290549,
      "grad_norm": 0.005645751953125,
      "learning_rate": 1.2987119893998246e-05,
      "loss": 0.0001,
      "step": 17908
    },
    {
      "epoch": 8.358926487747958,
      "grad_norm": 0.0103759765625,
      "learning_rate": 1.297988938073411e-05,
      "loss": 0.0001,
      "step": 17909
    },
    {
      "epoch": 8.359393232205367,
      "grad_norm": 0.006866455078125,
      "learning_rate": 1.2972660741109488e-05,
      "loss": 0.0001,
      "step": 17910
    },
    {
      "epoch": 8.359859976662777,
      "grad_norm": 0.00537109375,
      "learning_rate": 1.2965433975280039e-05,
      "loss": 0.0001,
      "step": 17911
    },
    {
      "epoch": 8.360326721120186,
      "grad_norm": 0.00701904296875,
      "learning_rate": 1.2958209083401385e-05,
      "loss": 0.0001,
      "step": 17912
    },
    {
      "epoch": 8.360793465577597,
      "grad_norm": 0.076171875,
      "learning_rate": 1.2950986065629035e-05,
      "loss": 0.0044,
      "step": 17913
    },
    {
      "epoch": 8.361260210035006,
      "grad_norm": 0.009765625,
      "learning_rate": 1.2943764922118562e-05,
      "loss": 0.0002,
      "step": 17914
    },
    {
      "epoch": 8.361726954492415,
      "grad_norm": 0.09228515625,
      "learning_rate": 1.2936545653025401e-05,
      "loss": 0.0026,
      "step": 17915
    },
    {
      "epoch": 8.362193698949826,
      "grad_norm": 0.01116943359375,
      "learning_rate": 1.2929328258505013e-05,
      "loss": 0.0002,
      "step": 17916
    },
    {
      "epoch": 8.362660443407234,
      "grad_norm": 0.01611328125,
      "learning_rate": 1.2922112738712811e-05,
      "loss": 0.0002,
      "step": 17917
    },
    {
      "epoch": 8.363127187864643,
      "grad_norm": 0.0140380859375,
      "learning_rate": 1.2914899093804111e-05,
      "loss": 0.0002,
      "step": 17918
    },
    {
      "epoch": 8.363593932322054,
      "grad_norm": 0.07861328125,
      "learning_rate": 1.2907687323934259e-05,
      "loss": 0.0029,
      "step": 17919
    },
    {
      "epoch": 8.364060676779463,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.2900477429258562e-05,
      "loss": 0.0001,
      "step": 17920
    },
    {
      "epoch": 8.364527421236874,
      "grad_norm": 0.006103515625,
      "learning_rate": 1.2893269409932196e-05,
      "loss": 0.0001,
      "step": 17921
    },
    {
      "epoch": 8.364994165694283,
      "grad_norm": 0.0615234375,
      "learning_rate": 1.2886063266110382e-05,
      "loss": 0.0002,
      "step": 17922
    },
    {
      "epoch": 8.365460910151691,
      "grad_norm": 0.010009765625,
      "learning_rate": 1.2878858997948306e-05,
      "loss": 0.0002,
      "step": 17923
    },
    {
      "epoch": 8.365927654609102,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.2871656605601034e-05,
      "loss": 0.0002,
      "step": 17924
    },
    {
      "epoch": 8.366394399066511,
      "grad_norm": 0.007171630859375,
      "learning_rate": 1.2864456089223664e-05,
      "loss": 0.0001,
      "step": 17925
    },
    {
      "epoch": 8.36686114352392,
      "grad_norm": 0.007080078125,
      "learning_rate": 1.285725744897126e-05,
      "loss": 0.0001,
      "step": 17926
    },
    {
      "epoch": 8.36732788798133,
      "grad_norm": 0.046630859375,
      "learning_rate": 1.2850060684998766e-05,
      "loss": 0.0026,
      "step": 17927
    },
    {
      "epoch": 8.36779463243874,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.2842865797461157e-05,
      "loss": 0.0001,
      "step": 17928
    },
    {
      "epoch": 8.36826137689615,
      "grad_norm": 0.0072021484375,
      "learning_rate": 1.2835672786513375e-05,
      "loss": 0.0001,
      "step": 17929
    },
    {
      "epoch": 8.36872812135356,
      "grad_norm": 0.005462646484375,
      "learning_rate": 1.2828481652310253e-05,
      "loss": 0.0001,
      "step": 17930
    },
    {
      "epoch": 8.369194865810968,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.2821292395006646e-05,
      "loss": 0.0002,
      "step": 17931
    },
    {
      "epoch": 8.369661610268379,
      "grad_norm": 0.012939453125,
      "learning_rate": 1.2814105014757327e-05,
      "loss": 0.0002,
      "step": 17932
    },
    {
      "epoch": 8.370128354725788,
      "grad_norm": 0.009033203125,
      "learning_rate": 1.2806919511717064e-05,
      "loss": 0.0002,
      "step": 17933
    },
    {
      "epoch": 8.370595099183197,
      "grad_norm": 0.0048828125,
      "learning_rate": 1.2799735886040588e-05,
      "loss": 0.0001,
      "step": 17934
    },
    {
      "epoch": 8.371061843640607,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.2792554137882518e-05,
      "loss": 0.0002,
      "step": 17935
    },
    {
      "epoch": 8.371528588098016,
      "grad_norm": 0.0125732421875,
      "learning_rate": 1.2785374267397532e-05,
      "loss": 0.0002,
      "step": 17936
    },
    {
      "epoch": 8.371995332555425,
      "grad_norm": 0.004852294921875,
      "learning_rate": 1.2778196274740207e-05,
      "loss": 0.0001,
      "step": 17937
    },
    {
      "epoch": 8.372462077012836,
      "grad_norm": 0.00799560546875,
      "learning_rate": 1.2771020160065083e-05,
      "loss": 0.0001,
      "step": 17938
    },
    {
      "epoch": 8.372928821470245,
      "grad_norm": 0.01171875,
      "learning_rate": 1.2763845923526674e-05,
      "loss": 0.0002,
      "step": 17939
    },
    {
      "epoch": 8.373395565927655,
      "grad_norm": 0.0130615234375,
      "learning_rate": 1.2756673565279464e-05,
      "loss": 0.0002,
      "step": 17940
    },
    {
      "epoch": 8.373862310385064,
      "grad_norm": 0.006744384765625,
      "learning_rate": 1.2749503085477864e-05,
      "loss": 0.0001,
      "step": 17941
    },
    {
      "epoch": 8.374329054842473,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.2742334484276264e-05,
      "loss": 0.0001,
      "step": 17942
    },
    {
      "epoch": 8.374795799299884,
      "grad_norm": 0.017578125,
      "learning_rate": 1.2735167761829037e-05,
      "loss": 0.0002,
      "step": 17943
    },
    {
      "epoch": 8.375262543757293,
      "grad_norm": 0.01318359375,
      "learning_rate": 1.2728002918290439e-05,
      "loss": 0.0001,
      "step": 17944
    },
    {
      "epoch": 8.375729288214702,
      "grad_norm": 0.007781982421875,
      "learning_rate": 1.2720839953814778e-05,
      "loss": 0.0001,
      "step": 17945
    },
    {
      "epoch": 8.376196032672112,
      "grad_norm": 0.0162353515625,
      "learning_rate": 1.271367886855629e-05,
      "loss": 0.0001,
      "step": 17946
    },
    {
      "epoch": 8.376662777129521,
      "grad_norm": 0.040283203125,
      "learning_rate": 1.2706519662669124e-05,
      "loss": 0.0022,
      "step": 17947
    },
    {
      "epoch": 8.377129521586932,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.269936233630744e-05,
      "loss": 0.0001,
      "step": 17948
    },
    {
      "epoch": 8.37759626604434,
      "grad_norm": 0.04638671875,
      "learning_rate": 1.2692206889625369e-05,
      "loss": 0.0013,
      "step": 17949
    },
    {
      "epoch": 8.37806301050175,
      "grad_norm": 0.007568359375,
      "learning_rate": 1.2685053322776929e-05,
      "loss": 0.0001,
      "step": 17950
    },
    {
      "epoch": 8.37852975495916,
      "grad_norm": 0.01287841796875,
      "learning_rate": 1.267790163591619e-05,
      "loss": 0.0002,
      "step": 17951
    },
    {
      "epoch": 8.37899649941657,
      "grad_norm": 0.00787353515625,
      "learning_rate": 1.2670751829197103e-05,
      "loss": 0.0001,
      "step": 17952
    },
    {
      "epoch": 8.379463243873978,
      "grad_norm": 0.017578125,
      "learning_rate": 1.2663603902773612e-05,
      "loss": 0.0002,
      "step": 17953
    },
    {
      "epoch": 8.379929988331389,
      "grad_norm": 0.041748046875,
      "learning_rate": 1.2656457856799653e-05,
      "loss": 0.0015,
      "step": 17954
    },
    {
      "epoch": 8.380396732788798,
      "grad_norm": 0.01007080078125,
      "learning_rate": 1.2649313691429042e-05,
      "loss": 0.0001,
      "step": 17955
    },
    {
      "epoch": 8.380863477246209,
      "grad_norm": 0.006256103515625,
      "learning_rate": 1.264217140681565e-05,
      "loss": 0.0001,
      "step": 17956
    },
    {
      "epoch": 8.381330221703617,
      "grad_norm": 0.0123291015625,
      "learning_rate": 1.2635031003113219e-05,
      "loss": 0.0002,
      "step": 17957
    },
    {
      "epoch": 8.381796966161026,
      "grad_norm": 0.024658203125,
      "learning_rate": 1.2627892480475512e-05,
      "loss": 0.0002,
      "step": 17958
    },
    {
      "epoch": 8.382263710618437,
      "grad_norm": 0.00830078125,
      "learning_rate": 1.2620755839056209e-05,
      "loss": 0.0001,
      "step": 17959
    },
    {
      "epoch": 8.382730455075846,
      "grad_norm": 0.006927490234375,
      "learning_rate": 1.2613621079008975e-05,
      "loss": 0.0001,
      "step": 17960
    },
    {
      "epoch": 8.383197199533255,
      "grad_norm": 0.0103759765625,
      "learning_rate": 1.2606488200487465e-05,
      "loss": 0.0002,
      "step": 17961
    },
    {
      "epoch": 8.383663943990666,
      "grad_norm": 0.01251220703125,
      "learning_rate": 1.2599357203645202e-05,
      "loss": 0.0002,
      "step": 17962
    },
    {
      "epoch": 8.384130688448074,
      "grad_norm": 0.023681640625,
      "learning_rate": 1.2592228088635772e-05,
      "loss": 0.002,
      "step": 17963
    },
    {
      "epoch": 8.384597432905485,
      "grad_norm": 0.024658203125,
      "learning_rate": 1.2585100855612631e-05,
      "loss": 0.0002,
      "step": 17964
    },
    {
      "epoch": 8.385064177362894,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.2577975504729256e-05,
      "loss": 0.0018,
      "step": 17965
    },
    {
      "epoch": 8.385530921820303,
      "grad_norm": 0.0098876953125,
      "learning_rate": 1.257085203613908e-05,
      "loss": 0.0001,
      "step": 17966
    },
    {
      "epoch": 8.385997666277714,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.2563730449995459e-05,
      "loss": 0.0002,
      "step": 17967
    },
    {
      "epoch": 8.386464410735122,
      "grad_norm": 0.0439453125,
      "learning_rate": 1.2556610746451725e-05,
      "loss": 0.0003,
      "step": 17968
    },
    {
      "epoch": 8.386931155192531,
      "grad_norm": 0.02880859375,
      "learning_rate": 1.2549492925661199e-05,
      "loss": 0.0015,
      "step": 17969
    },
    {
      "epoch": 8.387397899649942,
      "grad_norm": 0.02685546875,
      "learning_rate": 1.2542376987777105e-05,
      "loss": 0.0023,
      "step": 17970
    },
    {
      "epoch": 8.387864644107351,
      "grad_norm": 0.01165771484375,
      "learning_rate": 1.2535262932952662e-05,
      "loss": 0.0001,
      "step": 17971
    },
    {
      "epoch": 8.388331388564762,
      "grad_norm": 0.00543212890625,
      "learning_rate": 1.2528150761341073e-05,
      "loss": 0.0001,
      "step": 17972
    },
    {
      "epoch": 8.38879813302217,
      "grad_norm": 0.04736328125,
      "learning_rate": 1.2521040473095435e-05,
      "loss": 0.0002,
      "step": 17973
    },
    {
      "epoch": 8.38926487747958,
      "grad_norm": 0.00909423828125,
      "learning_rate": 1.251393206836885e-05,
      "loss": 0.0001,
      "step": 17974
    },
    {
      "epoch": 8.38973162193699,
      "grad_norm": 0.02783203125,
      "learning_rate": 1.2506825547314405e-05,
      "loss": 0.002,
      "step": 17975
    },
    {
      "epoch": 8.390198366394399,
      "grad_norm": 0.01239013671875,
      "learning_rate": 1.2499720910085055e-05,
      "loss": 0.0001,
      "step": 17976
    },
    {
      "epoch": 8.390665110851808,
      "grad_norm": 0.0064697265625,
      "learning_rate": 1.249261815683379e-05,
      "loss": 0.0001,
      "step": 17977
    },
    {
      "epoch": 8.391131855309219,
      "grad_norm": 0.004058837890625,
      "learning_rate": 1.2485517287713578e-05,
      "loss": 0.0001,
      "step": 17978
    },
    {
      "epoch": 8.391598599766628,
      "grad_norm": 0.037841796875,
      "learning_rate": 1.2478418302877259e-05,
      "loss": 0.0002,
      "step": 17979
    },
    {
      "epoch": 8.392065344224036,
      "grad_norm": 0.00946044921875,
      "learning_rate": 1.2471321202477693e-05,
      "loss": 0.0001,
      "step": 17980
    },
    {
      "epoch": 8.392532088681447,
      "grad_norm": 0.007293701171875,
      "learning_rate": 1.2464225986667721e-05,
      "loss": 0.0001,
      "step": 17981
    },
    {
      "epoch": 8.392998833138856,
      "grad_norm": 0.41796875,
      "learning_rate": 1.2457132655600056e-05,
      "loss": 0.0006,
      "step": 17982
    },
    {
      "epoch": 8.393465577596267,
      "grad_norm": 0.045654296875,
      "learning_rate": 1.2450041209427488e-05,
      "loss": 0.0018,
      "step": 17983
    },
    {
      "epoch": 8.393932322053676,
      "grad_norm": 0.01019287109375,
      "learning_rate": 1.2442951648302648e-05,
      "loss": 0.0002,
      "step": 17984
    },
    {
      "epoch": 8.394399066511085,
      "grad_norm": 0.00823974609375,
      "learning_rate": 1.2435863972378204e-05,
      "loss": 0.0001,
      "step": 17985
    },
    {
      "epoch": 8.394865810968495,
      "grad_norm": 0.01531982421875,
      "learning_rate": 1.2428778181806788e-05,
      "loss": 0.0002,
      "step": 17986
    },
    {
      "epoch": 8.395332555425904,
      "grad_norm": 0.007781982421875,
      "learning_rate": 1.2421694276740914e-05,
      "loss": 0.0002,
      "step": 17987
    },
    {
      "epoch": 8.395799299883313,
      "grad_norm": 0.005401611328125,
      "learning_rate": 1.2414612257333125e-05,
      "loss": 0.0001,
      "step": 17988
    },
    {
      "epoch": 8.396266044340724,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.2407532123735944e-05,
      "loss": 0.0001,
      "step": 17989
    },
    {
      "epoch": 8.396732788798133,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.2400453876101758e-05,
      "loss": 0.0001,
      "step": 17990
    },
    {
      "epoch": 8.397199533255543,
      "grad_norm": 0.00726318359375,
      "learning_rate": 1.2393377514582993e-05,
      "loss": 0.0001,
      "step": 17991
    },
    {
      "epoch": 8.397666277712952,
      "grad_norm": 0.039306640625,
      "learning_rate": 1.2386303039332026e-05,
      "loss": 0.0003,
      "step": 17992
    },
    {
      "epoch": 8.398133022170361,
      "grad_norm": 0.03662109375,
      "learning_rate": 1.2379230450501145e-05,
      "loss": 0.0018,
      "step": 17993
    },
    {
      "epoch": 8.398599766627772,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.237215974824265e-05,
      "loss": 0.0002,
      "step": 17994
    },
    {
      "epoch": 8.39906651108518,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.2365090932708811e-05,
      "loss": 0.0001,
      "step": 17995
    },
    {
      "epoch": 8.39953325554259,
      "grad_norm": 0.00482177734375,
      "learning_rate": 1.2358024004051772e-05,
      "loss": 0.0001,
      "step": 17996
    },
    {
      "epoch": 8.4,
      "grad_norm": 0.01275634765625,
      "learning_rate": 1.235095896242371e-05,
      "loss": 0.0001,
      "step": 17997
    },
    {
      "epoch": 8.40046674445741,
      "grad_norm": 0.01287841796875,
      "learning_rate": 1.2343895807976768e-05,
      "loss": 0.0002,
      "step": 17998
    },
    {
      "epoch": 8.40093348891482,
      "grad_norm": 0.0301513671875,
      "learning_rate": 1.2336834540862997e-05,
      "loss": 0.0002,
      "step": 17999
    },
    {
      "epoch": 8.401400233372229,
      "grad_norm": 0.08203125,
      "learning_rate": 1.2329775161234458e-05,
      "loss": 0.006,
      "step": 18000
    },
    {
      "epoch": 8.401866977829638,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.2322717669243111e-05,
      "loss": 0.0002,
      "step": 18001
    },
    {
      "epoch": 8.402333722287048,
      "grad_norm": 0.02294921875,
      "learning_rate": 1.2315662065040923e-05,
      "loss": 0.0002,
      "step": 18002
    },
    {
      "epoch": 8.402800466744457,
      "grad_norm": 0.0235595703125,
      "learning_rate": 1.230860834877985e-05,
      "loss": 0.0002,
      "step": 18003
    },
    {
      "epoch": 8.403267211201866,
      "grad_norm": 0.006378173828125,
      "learning_rate": 1.2301556520611734e-05,
      "loss": 0.0001,
      "step": 18004
    },
    {
      "epoch": 8.403733955659277,
      "grad_norm": 0.011474609375,
      "learning_rate": 1.229450658068838e-05,
      "loss": 0.0001,
      "step": 18005
    },
    {
      "epoch": 8.404200700116686,
      "grad_norm": 0.00860595703125,
      "learning_rate": 1.2287458529161622e-05,
      "loss": 0.0002,
      "step": 18006
    },
    {
      "epoch": 8.404667444574097,
      "grad_norm": 0.0311279296875,
      "learning_rate": 1.2280412366183214e-05,
      "loss": 0.002,
      "step": 18007
    },
    {
      "epoch": 8.405134189031505,
      "grad_norm": 0.044677734375,
      "learning_rate": 1.2273368091904836e-05,
      "loss": 0.002,
      "step": 18008
    },
    {
      "epoch": 8.405600933488914,
      "grad_norm": 0.004913330078125,
      "learning_rate": 1.2266325706478177e-05,
      "loss": 0.0001,
      "step": 18009
    },
    {
      "epoch": 8.406067677946325,
      "grad_norm": 0.03515625,
      "learning_rate": 1.2259285210054883e-05,
      "loss": 0.0002,
      "step": 18010
    },
    {
      "epoch": 8.406534422403734,
      "grad_norm": 0.00628662109375,
      "learning_rate": 1.225224660278651e-05,
      "loss": 0.0001,
      "step": 18011
    },
    {
      "epoch": 8.407001166861143,
      "grad_norm": 0.01031494140625,
      "learning_rate": 1.224520988482465e-05,
      "loss": 0.0001,
      "step": 18012
    },
    {
      "epoch": 8.407467911318554,
      "grad_norm": 0.0050048828125,
      "learning_rate": 1.2238175056320766e-05,
      "loss": 0.0001,
      "step": 18013
    },
    {
      "epoch": 8.407934655775962,
      "grad_norm": 0.10791015625,
      "learning_rate": 1.2231142117426341e-05,
      "loss": 0.0003,
      "step": 18014
    },
    {
      "epoch": 8.408401400233373,
      "grad_norm": 0.006927490234375,
      "learning_rate": 1.2224111068292831e-05,
      "loss": 0.0001,
      "step": 18015
    },
    {
      "epoch": 8.408868144690782,
      "grad_norm": 0.00762939453125,
      "learning_rate": 1.2217081909071582e-05,
      "loss": 0.0001,
      "step": 18016
    },
    {
      "epoch": 8.409334889148191,
      "grad_norm": 0.0084228515625,
      "learning_rate": 1.221005463991396e-05,
      "loss": 0.0002,
      "step": 18017
    },
    {
      "epoch": 8.409801633605602,
      "grad_norm": 0.005401611328125,
      "learning_rate": 1.2203029260971277e-05,
      "loss": 0.0001,
      "step": 18018
    },
    {
      "epoch": 8.41026837806301,
      "grad_norm": 0.005859375,
      "learning_rate": 1.2196005772394769e-05,
      "loss": 0.0001,
      "step": 18019
    },
    {
      "epoch": 8.41073512252042,
      "grad_norm": 0.024169921875,
      "learning_rate": 1.2188984174335672e-05,
      "loss": 0.0002,
      "step": 18020
    },
    {
      "epoch": 8.41120186697783,
      "grad_norm": 0.005523681640625,
      "learning_rate": 1.2181964466945206e-05,
      "loss": 0.0001,
      "step": 18021
    },
    {
      "epoch": 8.411668611435239,
      "grad_norm": 0.007232666015625,
      "learning_rate": 1.2174946650374452e-05,
      "loss": 0.0001,
      "step": 18022
    },
    {
      "epoch": 8.412135355892648,
      "grad_norm": 0.037353515625,
      "learning_rate": 1.2167930724774546e-05,
      "loss": 0.0002,
      "step": 18023
    },
    {
      "epoch": 8.412602100350059,
      "grad_norm": 0.007049560546875,
      "learning_rate": 1.2160916690296554e-05,
      "loss": 0.0001,
      "step": 18024
    },
    {
      "epoch": 8.413068844807468,
      "grad_norm": 0.0791015625,
      "learning_rate": 1.2153904547091465e-05,
      "loss": 0.0051,
      "step": 18025
    },
    {
      "epoch": 8.413535589264878,
      "grad_norm": 0.0281982421875,
      "learning_rate": 1.2146894295310284e-05,
      "loss": 0.001,
      "step": 18026
    },
    {
      "epoch": 8.414002333722287,
      "grad_norm": 0.01092529296875,
      "learning_rate": 1.2139885935103968e-05,
      "loss": 0.0002,
      "step": 18027
    },
    {
      "epoch": 8.414469078179696,
      "grad_norm": 0.050537109375,
      "learning_rate": 1.2132879466623359e-05,
      "loss": 0.0015,
      "step": 18028
    },
    {
      "epoch": 8.414935822637107,
      "grad_norm": 0.018798828125,
      "learning_rate": 1.212587489001935e-05,
      "loss": 0.0002,
      "step": 18029
    },
    {
      "epoch": 8.415402567094516,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.2118872205442778e-05,
      "loss": 0.0001,
      "step": 18030
    },
    {
      "epoch": 8.415869311551925,
      "grad_norm": 0.005645751953125,
      "learning_rate": 1.2111871413044374e-05,
      "loss": 0.0001,
      "step": 18031
    },
    {
      "epoch": 8.416336056009335,
      "grad_norm": 0.004852294921875,
      "learning_rate": 1.2104872512974908e-05,
      "loss": 0.0001,
      "step": 18032
    },
    {
      "epoch": 8.416802800466744,
      "grad_norm": 0.01141357421875,
      "learning_rate": 1.2097875505385025e-05,
      "loss": 0.0002,
      "step": 18033
    },
    {
      "epoch": 8.417269544924155,
      "grad_norm": 0.005401611328125,
      "learning_rate": 1.2090880390425429e-05,
      "loss": 0.0001,
      "step": 18034
    },
    {
      "epoch": 8.417736289381564,
      "grad_norm": 0.04833984375,
      "learning_rate": 1.2083887168246733e-05,
      "loss": 0.0018,
      "step": 18035
    },
    {
      "epoch": 8.418203033838973,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.2076895838999469e-05,
      "loss": 0.0002,
      "step": 18036
    },
    {
      "epoch": 8.418669778296383,
      "grad_norm": 0.06591796875,
      "learning_rate": 1.2069906402834186e-05,
      "loss": 0.0048,
      "step": 18037
    },
    {
      "epoch": 8.419136522753792,
      "grad_norm": 0.01092529296875,
      "learning_rate": 1.2062918859901396e-05,
      "loss": 0.0001,
      "step": 18038
    },
    {
      "epoch": 8.419603267211201,
      "grad_norm": 0.043701171875,
      "learning_rate": 1.2055933210351523e-05,
      "loss": 0.002,
      "step": 18039
    },
    {
      "epoch": 8.420070011668612,
      "grad_norm": 0.048828125,
      "learning_rate": 1.2048949454334968e-05,
      "loss": 0.0002,
      "step": 18040
    },
    {
      "epoch": 8.42053675612602,
      "grad_norm": 0.0419921875,
      "learning_rate": 1.2041967592002146e-05,
      "loss": 0.0031,
      "step": 18041
    },
    {
      "epoch": 8.421003500583431,
      "grad_norm": 0.058837890625,
      "learning_rate": 1.2034987623503324e-05,
      "loss": 0.003,
      "step": 18042
    },
    {
      "epoch": 8.42147024504084,
      "grad_norm": 0.004302978515625,
      "learning_rate": 1.2028009548988828e-05,
      "loss": 0.0001,
      "step": 18043
    },
    {
      "epoch": 8.42193698949825,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.2021033368608902e-05,
      "loss": 0.0002,
      "step": 18044
    },
    {
      "epoch": 8.42240373395566,
      "grad_norm": 0.0172119140625,
      "learning_rate": 1.2014059082513728e-05,
      "loss": 0.0002,
      "step": 18045
    },
    {
      "epoch": 8.422870478413069,
      "grad_norm": 0.0072021484375,
      "learning_rate": 1.2007086690853486e-05,
      "loss": 0.0001,
      "step": 18046
    },
    {
      "epoch": 8.423337222870478,
      "grad_norm": 0.007049560546875,
      "learning_rate": 1.2000116193778322e-05,
      "loss": 0.0002,
      "step": 18047
    },
    {
      "epoch": 8.423803967327888,
      "grad_norm": 0.00823974609375,
      "learning_rate": 1.1993147591438259e-05,
      "loss": 0.0001,
      "step": 18048
    },
    {
      "epoch": 8.424270711785297,
      "grad_norm": 0.036376953125,
      "learning_rate": 1.1986180883983412e-05,
      "loss": 0.001,
      "step": 18049
    },
    {
      "epoch": 8.424737456242708,
      "grad_norm": 0.011474609375,
      "learning_rate": 1.1979216071563704e-05,
      "loss": 0.0002,
      "step": 18050
    },
    {
      "epoch": 8.425204200700117,
      "grad_norm": 0.0087890625,
      "learning_rate": 1.1972253154329171e-05,
      "loss": 0.0001,
      "step": 18051
    },
    {
      "epoch": 8.425670945157526,
      "grad_norm": 0.034912109375,
      "learning_rate": 1.1965292132429661e-05,
      "loss": 0.0026,
      "step": 18052
    },
    {
      "epoch": 8.426137689614936,
      "grad_norm": 0.00616455078125,
      "learning_rate": 1.195833300601512e-05,
      "loss": 0.0001,
      "step": 18053
    },
    {
      "epoch": 8.426604434072345,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.195137577523533e-05,
      "loss": 0.0002,
      "step": 18054
    },
    {
      "epoch": 8.427071178529754,
      "grad_norm": 0.0712890625,
      "learning_rate": 1.1944420440240111e-05,
      "loss": 0.0048,
      "step": 18055
    },
    {
      "epoch": 8.427537922987165,
      "grad_norm": 0.00860595703125,
      "learning_rate": 1.1937467001179236e-05,
      "loss": 0.0002,
      "step": 18056
    },
    {
      "epoch": 8.428004667444574,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.1930515458202384e-05,
      "loss": 0.0001,
      "step": 18057
    },
    {
      "epoch": 8.428471411901985,
      "grad_norm": 0.0198974609375,
      "learning_rate": 1.1923565811459259e-05,
      "loss": 0.0002,
      "step": 18058
    },
    {
      "epoch": 8.428938156359393,
      "grad_norm": 0.03759765625,
      "learning_rate": 1.1916618061099493e-05,
      "loss": 0.0002,
      "step": 18059
    },
    {
      "epoch": 8.429404900816802,
      "grad_norm": 0.00811767578125,
      "learning_rate": 1.1909672207272649e-05,
      "loss": 0.0001,
      "step": 18060
    },
    {
      "epoch": 8.429871645274213,
      "grad_norm": 0.01123046875,
      "learning_rate": 1.1902728250128325e-05,
      "loss": 0.0002,
      "step": 18061
    },
    {
      "epoch": 8.430338389731622,
      "grad_norm": 0.0169677734375,
      "learning_rate": 1.1895786189815995e-05,
      "loss": 0.0002,
      "step": 18062
    },
    {
      "epoch": 8.43080513418903,
      "grad_norm": 0.060791015625,
      "learning_rate": 1.1888846026485134e-05,
      "loss": 0.0011,
      "step": 18063
    },
    {
      "epoch": 8.431271878646442,
      "grad_norm": 0.00787353515625,
      "learning_rate": 1.1881907760285194e-05,
      "loss": 0.0001,
      "step": 18064
    },
    {
      "epoch": 8.43173862310385,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.187497139136553e-05,
      "loss": 0.0001,
      "step": 18065
    },
    {
      "epoch": 8.43220536756126,
      "grad_norm": 0.01611328125,
      "learning_rate": 1.1868036919875513e-05,
      "loss": 0.0001,
      "step": 18066
    },
    {
      "epoch": 8.43267211201867,
      "grad_norm": 0.032958984375,
      "learning_rate": 1.186110434596447e-05,
      "loss": 0.0002,
      "step": 18067
    },
    {
      "epoch": 8.433138856476079,
      "grad_norm": 0.01544189453125,
      "learning_rate": 1.185417366978162e-05,
      "loss": 0.0002,
      "step": 18068
    },
    {
      "epoch": 8.43360560093349,
      "grad_norm": 0.00579833984375,
      "learning_rate": 1.1847244891476206e-05,
      "loss": 0.0001,
      "step": 18069
    },
    {
      "epoch": 8.434072345390899,
      "grad_norm": 0.0101318359375,
      "learning_rate": 1.1840318011197438e-05,
      "loss": 0.0002,
      "step": 18070
    },
    {
      "epoch": 8.434539089848307,
      "grad_norm": 0.0186767578125,
      "learning_rate": 1.183339302909442e-05,
      "loss": 0.0002,
      "step": 18071
    },
    {
      "epoch": 8.435005834305718,
      "grad_norm": 0.0751953125,
      "learning_rate": 1.1826469945316265e-05,
      "loss": 0.0018,
      "step": 18072
    },
    {
      "epoch": 8.435472578763127,
      "grad_norm": 0.01019287109375,
      "learning_rate": 1.1819548760012068e-05,
      "loss": 0.0001,
      "step": 18073
    },
    {
      "epoch": 8.435939323220536,
      "grad_norm": 0.0030364990234375,
      "learning_rate": 1.1812629473330805e-05,
      "loss": 0.0001,
      "step": 18074
    },
    {
      "epoch": 8.436406067677947,
      "grad_norm": 0.0042724609375,
      "learning_rate": 1.1805712085421473e-05,
      "loss": 0.0001,
      "step": 18075
    },
    {
      "epoch": 8.436872812135356,
      "grad_norm": 0.004791259765625,
      "learning_rate": 1.1798796596433037e-05,
      "loss": 0.0001,
      "step": 18076
    },
    {
      "epoch": 8.437339556592766,
      "grad_norm": 0.0274658203125,
      "learning_rate": 1.1791883006514349e-05,
      "loss": 0.0001,
      "step": 18077
    },
    {
      "epoch": 8.437806301050175,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.1784971315814287e-05,
      "loss": 0.0001,
      "step": 18078
    },
    {
      "epoch": 8.438273045507584,
      "grad_norm": 0.007720947265625,
      "learning_rate": 1.1778061524481688e-05,
      "loss": 0.0001,
      "step": 18079
    },
    {
      "epoch": 8.438739789964995,
      "grad_norm": 0.006195068359375,
      "learning_rate": 1.1771153632665288e-05,
      "loss": 0.0001,
      "step": 18080
    },
    {
      "epoch": 8.439206534422404,
      "grad_norm": 0.00946044921875,
      "learning_rate": 1.1764247640513858e-05,
      "loss": 0.0002,
      "step": 18081
    },
    {
      "epoch": 8.439673278879813,
      "grad_norm": 0.02978515625,
      "learning_rate": 1.175734354817607e-05,
      "loss": 0.0019,
      "step": 18082
    },
    {
      "epoch": 8.440140023337223,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.1750441355800568e-05,
      "loss": 0.0002,
      "step": 18083
    },
    {
      "epoch": 8.440606767794632,
      "grad_norm": 0.07080078125,
      "learning_rate": 1.1743541063536001e-05,
      "loss": 0.0002,
      "step": 18084
    },
    {
      "epoch": 8.441073512252043,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.1736642671530906e-05,
      "loss": 0.0001,
      "step": 18085
    },
    {
      "epoch": 8.441540256709452,
      "grad_norm": 0.007415771484375,
      "learning_rate": 1.172974617993381e-05,
      "loss": 0.0001,
      "step": 18086
    },
    {
      "epoch": 8.44200700116686,
      "grad_norm": 0.007537841796875,
      "learning_rate": 1.1722851588893246e-05,
      "loss": 0.0002,
      "step": 18087
    },
    {
      "epoch": 8.442473745624271,
      "grad_norm": 0.0166015625,
      "learning_rate": 1.171595889855761e-05,
      "loss": 0.0003,
      "step": 18088
    },
    {
      "epoch": 8.44294049008168,
      "grad_norm": 0.0252685546875,
      "learning_rate": 1.1709068109075338e-05,
      "loss": 0.0021,
      "step": 18089
    },
    {
      "epoch": 8.44340723453909,
      "grad_norm": 0.054931640625,
      "learning_rate": 1.17021792205948e-05,
      "loss": 0.003,
      "step": 18090
    },
    {
      "epoch": 8.4438739789965,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.16952922332643e-05,
      "loss": 0.0001,
      "step": 18091
    },
    {
      "epoch": 8.444340723453909,
      "grad_norm": 0.005523681640625,
      "learning_rate": 1.1688407147232128e-05,
      "loss": 0.0001,
      "step": 18092
    },
    {
      "epoch": 8.444807467911318,
      "grad_norm": 0.03857421875,
      "learning_rate": 1.1681523962646557e-05,
      "loss": 0.0026,
      "step": 18093
    },
    {
      "epoch": 8.445274212368728,
      "grad_norm": 0.06396484375,
      "learning_rate": 1.1674642679655745e-05,
      "loss": 0.0021,
      "step": 18094
    },
    {
      "epoch": 8.445740956826137,
      "grad_norm": 0.054931640625,
      "learning_rate": 1.1667763298407875e-05,
      "loss": 0.0002,
      "step": 18095
    },
    {
      "epoch": 8.446207701283548,
      "grad_norm": 0.00909423828125,
      "learning_rate": 1.1660885819051081e-05,
      "loss": 0.0001,
      "step": 18096
    },
    {
      "epoch": 8.446674445740957,
      "grad_norm": 0.0751953125,
      "learning_rate": 1.1654010241733414e-05,
      "loss": 0.0019,
      "step": 18097
    },
    {
      "epoch": 8.447141190198366,
      "grad_norm": 0.00494384765625,
      "learning_rate": 1.1647136566602935e-05,
      "loss": 0.0001,
      "step": 18098
    },
    {
      "epoch": 8.447607934655776,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.1640264793807676e-05,
      "loss": 0.0002,
      "step": 18099
    },
    {
      "epoch": 8.448074679113185,
      "grad_norm": 0.09814453125,
      "learning_rate": 1.16333949234955e-05,
      "loss": 0.003,
      "step": 18100
    },
    {
      "epoch": 8.448541423570596,
      "grad_norm": 0.01336669921875,
      "learning_rate": 1.1626526955814376e-05,
      "loss": 0.0001,
      "step": 18101
    },
    {
      "epoch": 8.449008168028005,
      "grad_norm": 0.0341796875,
      "learning_rate": 1.1619660890912209e-05,
      "loss": 0.0017,
      "step": 18102
    },
    {
      "epoch": 8.449474912485414,
      "grad_norm": 0.014404296875,
      "learning_rate": 1.161279672893678e-05,
      "loss": 0.0002,
      "step": 18103
    },
    {
      "epoch": 8.449941656942825,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.1605934470035906e-05,
      "loss": 0.0001,
      "step": 18104
    },
    {
      "epoch": 8.450408401400233,
      "grad_norm": 0.035400390625,
      "learning_rate": 1.1599074114357355e-05,
      "loss": 0.0023,
      "step": 18105
    },
    {
      "epoch": 8.450875145857642,
      "grad_norm": 0.043701171875,
      "learning_rate": 1.15922156620488e-05,
      "loss": 0.0042,
      "step": 18106
    },
    {
      "epoch": 8.451341890315053,
      "grad_norm": 0.006988525390625,
      "learning_rate": 1.1585359113257931e-05,
      "loss": 0.0001,
      "step": 18107
    },
    {
      "epoch": 8.451808634772462,
      "grad_norm": 0.0361328125,
      "learning_rate": 1.1578504468132401e-05,
      "loss": 0.0002,
      "step": 18108
    },
    {
      "epoch": 8.45227537922987,
      "grad_norm": 0.01055908203125,
      "learning_rate": 1.1571651726819755e-05,
      "loss": 0.0001,
      "step": 18109
    },
    {
      "epoch": 8.452742123687282,
      "grad_norm": 0.01007080078125,
      "learning_rate": 1.1564800889467553e-05,
      "loss": 0.0002,
      "step": 18110
    },
    {
      "epoch": 8.45320886814469,
      "grad_norm": 0.004425048828125,
      "learning_rate": 1.1557951956223334e-05,
      "loss": 0.0001,
      "step": 18111
    },
    {
      "epoch": 8.453675612602101,
      "grad_norm": 0.00787353515625,
      "learning_rate": 1.1551104927234524e-05,
      "loss": 0.0001,
      "step": 18112
    },
    {
      "epoch": 8.45414235705951,
      "grad_norm": 0.01312255859375,
      "learning_rate": 1.1544259802648572e-05,
      "loss": 0.0002,
      "step": 18113
    },
    {
      "epoch": 8.454609101516919,
      "grad_norm": 0.00927734375,
      "learning_rate": 1.1537416582612825e-05,
      "loss": 0.0002,
      "step": 18114
    },
    {
      "epoch": 8.45507584597433,
      "grad_norm": 0.007720947265625,
      "learning_rate": 1.1530575267274658e-05,
      "loss": 0.0001,
      "step": 18115
    },
    {
      "epoch": 8.455542590431739,
      "grad_norm": 0.02587890625,
      "learning_rate": 1.1523735856781392e-05,
      "loss": 0.0002,
      "step": 18116
    },
    {
      "epoch": 8.456009334889147,
      "grad_norm": 0.0205078125,
      "learning_rate": 1.1516898351280226e-05,
      "loss": 0.0002,
      "step": 18117
    },
    {
      "epoch": 8.456476079346558,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.1510062750918416e-05,
      "loss": 0.0001,
      "step": 18118
    },
    {
      "epoch": 8.456942823803967,
      "grad_norm": 0.0247802734375,
      "learning_rate": 1.1503229055843157e-05,
      "loss": 0.0019,
      "step": 18119
    },
    {
      "epoch": 8.457409568261378,
      "grad_norm": 0.008056640625,
      "learning_rate": 1.1496397266201541e-05,
      "loss": 0.0002,
      "step": 18120
    },
    {
      "epoch": 8.457876312718787,
      "grad_norm": 0.0257568359375,
      "learning_rate": 1.1489567382140699e-05,
      "loss": 0.0004,
      "step": 18121
    },
    {
      "epoch": 8.458343057176196,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.1482739403807685e-05,
      "loss": 0.0001,
      "step": 18122
    },
    {
      "epoch": 8.458809801633606,
      "grad_norm": 0.0096435546875,
      "learning_rate": 1.1475913331349486e-05,
      "loss": 0.0001,
      "step": 18123
    },
    {
      "epoch": 8.459276546091015,
      "grad_norm": 0.030517578125,
      "learning_rate": 1.1469089164913093e-05,
      "loss": 0.0025,
      "step": 18124
    },
    {
      "epoch": 8.459743290548424,
      "grad_norm": 0.013671875,
      "learning_rate": 1.1462266904645457e-05,
      "loss": 0.0002,
      "step": 18125
    },
    {
      "epoch": 8.460210035005835,
      "grad_norm": 0.01544189453125,
      "learning_rate": 1.1455446550693427e-05,
      "loss": 0.0002,
      "step": 18126
    },
    {
      "epoch": 8.460676779463244,
      "grad_norm": 0.052001953125,
      "learning_rate": 1.1448628103203884e-05,
      "loss": 0.0018,
      "step": 18127
    },
    {
      "epoch": 8.461143523920654,
      "grad_norm": 0.014892578125,
      "learning_rate": 1.1441811562323635e-05,
      "loss": 0.0001,
      "step": 18128
    },
    {
      "epoch": 8.461610268378063,
      "grad_norm": 0.04833984375,
      "learning_rate": 1.1434996928199426e-05,
      "loss": 0.0002,
      "step": 18129
    },
    {
      "epoch": 8.462077012835472,
      "grad_norm": 0.00836181640625,
      "learning_rate": 1.1428184200978021e-05,
      "loss": 0.0001,
      "step": 18130
    },
    {
      "epoch": 8.462543757292883,
      "grad_norm": 0.0458984375,
      "learning_rate": 1.1421373380806055e-05,
      "loss": 0.0007,
      "step": 18131
    },
    {
      "epoch": 8.463010501750292,
      "grad_norm": 0.02099609375,
      "learning_rate": 1.1414564467830202e-05,
      "loss": 0.0002,
      "step": 18132
    },
    {
      "epoch": 8.4634772462077,
      "grad_norm": 0.07861328125,
      "learning_rate": 1.1407757462197077e-05,
      "loss": 0.0021,
      "step": 18133
    },
    {
      "epoch": 8.463943990665111,
      "grad_norm": 0.00958251953125,
      "learning_rate": 1.140095236405322e-05,
      "loss": 0.0001,
      "step": 18134
    },
    {
      "epoch": 8.46441073512252,
      "grad_norm": 0.01129150390625,
      "learning_rate": 1.1394149173545154e-05,
      "loss": 0.0001,
      "step": 18135
    },
    {
      "epoch": 8.464877479579929,
      "grad_norm": 0.02978515625,
      "learning_rate": 1.138734789081939e-05,
      "loss": 0.002,
      "step": 18136
    },
    {
      "epoch": 8.46534422403734,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.1380548516022326e-05,
      "loss": 0.0001,
      "step": 18137
    },
    {
      "epoch": 8.465810968494749,
      "grad_norm": 0.01080322265625,
      "learning_rate": 1.1373751049300374e-05,
      "loss": 0.0001,
      "step": 18138
    },
    {
      "epoch": 8.46627771295216,
      "grad_norm": 0.036376953125,
      "learning_rate": 1.1366955490799912e-05,
      "loss": 0.0018,
      "step": 18139
    },
    {
      "epoch": 8.466744457409568,
      "grad_norm": 0.00897216796875,
      "learning_rate": 1.1360161840667217e-05,
      "loss": 0.0002,
      "step": 18140
    },
    {
      "epoch": 8.467211201866977,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.1353370099048589e-05,
      "loss": 0.0001,
      "step": 18141
    },
    {
      "epoch": 8.467677946324388,
      "grad_norm": 0.056640625,
      "learning_rate": 1.1346580266090279e-05,
      "loss": 0.0023,
      "step": 18142
    },
    {
      "epoch": 8.468144690781797,
      "grad_norm": 0.0115966796875,
      "learning_rate": 1.1339792341938438e-05,
      "loss": 0.0001,
      "step": 18143
    },
    {
      "epoch": 8.468611435239206,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.1333006326739237e-05,
      "loss": 0.0001,
      "step": 18144
    },
    {
      "epoch": 8.469078179696616,
      "grad_norm": 0.00439453125,
      "learning_rate": 1.1326222220638805e-05,
      "loss": 0.0001,
      "step": 18145
    },
    {
      "epoch": 8.469544924154025,
      "grad_norm": 0.01953125,
      "learning_rate": 1.1319440023783179e-05,
      "loss": 0.0002,
      "step": 18146
    },
    {
      "epoch": 8.470011668611436,
      "grad_norm": 0.00848388671875,
      "learning_rate": 1.131265973631842e-05,
      "loss": 0.0001,
      "step": 18147
    },
    {
      "epoch": 8.470478413068845,
      "grad_norm": 0.016845703125,
      "learning_rate": 1.1305881358390491e-05,
      "loss": 0.0002,
      "step": 18148
    },
    {
      "epoch": 8.470945157526254,
      "grad_norm": 0.01287841796875,
      "learning_rate": 1.129910489014534e-05,
      "loss": 0.0002,
      "step": 18149
    },
    {
      "epoch": 8.471411901983664,
      "grad_norm": 0.004852294921875,
      "learning_rate": 1.1292330331728862e-05,
      "loss": 0.0001,
      "step": 18150
    },
    {
      "epoch": 8.471878646441073,
      "grad_norm": 0.007415771484375,
      "learning_rate": 1.1285557683286962e-05,
      "loss": 0.0001,
      "step": 18151
    },
    {
      "epoch": 8.472345390898482,
      "grad_norm": 0.0174560546875,
      "learning_rate": 1.1278786944965402e-05,
      "loss": 0.0002,
      "step": 18152
    },
    {
      "epoch": 8.472812135355893,
      "grad_norm": 0.01190185546875,
      "learning_rate": 1.127201811691001e-05,
      "loss": 0.0001,
      "step": 18153
    },
    {
      "epoch": 8.473278879813302,
      "grad_norm": 0.0260009765625,
      "learning_rate": 1.1265251199266535e-05,
      "loss": 0.0005,
      "step": 18154
    },
    {
      "epoch": 8.473745624270713,
      "grad_norm": 0.0498046875,
      "learning_rate": 1.1258486192180628e-05,
      "loss": 0.0035,
      "step": 18155
    },
    {
      "epoch": 8.474212368728121,
      "grad_norm": 0.006103515625,
      "learning_rate": 1.1251723095797972e-05,
      "loss": 0.0001,
      "step": 18156
    },
    {
      "epoch": 8.47467911318553,
      "grad_norm": 0.01708984375,
      "learning_rate": 1.1244961910264218e-05,
      "loss": 0.0002,
      "step": 18157
    },
    {
      "epoch": 8.475145857642941,
      "grad_norm": 0.0291748046875,
      "learning_rate": 1.123820263572487e-05,
      "loss": 0.0002,
      "step": 18158
    },
    {
      "epoch": 8.47561260210035,
      "grad_norm": 0.006744384765625,
      "learning_rate": 1.1231445272325514e-05,
      "loss": 0.0002,
      "step": 18159
    },
    {
      "epoch": 8.476079346557759,
      "grad_norm": 0.04541015625,
      "learning_rate": 1.1224689820211653e-05,
      "loss": 0.0022,
      "step": 18160
    },
    {
      "epoch": 8.47654609101517,
      "grad_norm": 0.005218505859375,
      "learning_rate": 1.1217936279528696e-05,
      "loss": 0.0001,
      "step": 18161
    },
    {
      "epoch": 8.477012835472578,
      "grad_norm": 0.019287109375,
      "learning_rate": 1.1211184650422102e-05,
      "loss": 0.0002,
      "step": 18162
    },
    {
      "epoch": 8.47747957992999,
      "grad_norm": 0.034423828125,
      "learning_rate": 1.1204434933037188e-05,
      "loss": 0.0004,
      "step": 18163
    },
    {
      "epoch": 8.477946324387398,
      "grad_norm": 0.00555419921875,
      "learning_rate": 1.1197687127519307e-05,
      "loss": 0.0001,
      "step": 18164
    },
    {
      "epoch": 8.478413068844807,
      "grad_norm": 0.0123291015625,
      "learning_rate": 1.1190941234013774e-05,
      "loss": 0.0002,
      "step": 18165
    },
    {
      "epoch": 8.478879813302218,
      "grad_norm": 0.00897216796875,
      "learning_rate": 1.1184197252665795e-05,
      "loss": 0.0002,
      "step": 18166
    },
    {
      "epoch": 8.479346557759627,
      "grad_norm": 0.006256103515625,
      "learning_rate": 1.117745518362059e-05,
      "loss": 0.0001,
      "step": 18167
    },
    {
      "epoch": 8.479813302217035,
      "grad_norm": 0.00750732421875,
      "learning_rate": 1.1170715027023348e-05,
      "loss": 0.0001,
      "step": 18168
    },
    {
      "epoch": 8.480280046674446,
      "grad_norm": 0.00775146484375,
      "learning_rate": 1.1163976783019137e-05,
      "loss": 0.0001,
      "step": 18169
    },
    {
      "epoch": 8.480746791131855,
      "grad_norm": 0.051025390625,
      "learning_rate": 1.1157240451753081e-05,
      "loss": 0.002,
      "step": 18170
    },
    {
      "epoch": 8.481213535589266,
      "grad_norm": 0.00927734375,
      "learning_rate": 1.1150506033370234e-05,
      "loss": 0.0001,
      "step": 18171
    },
    {
      "epoch": 8.481680280046675,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.1143773528015544e-05,
      "loss": 0.0001,
      "step": 18172
    },
    {
      "epoch": 8.482147024504084,
      "grad_norm": 0.00970458984375,
      "learning_rate": 1.1137042935834008e-05,
      "loss": 0.0001,
      "step": 18173
    },
    {
      "epoch": 8.482613768961494,
      "grad_norm": 0.0093994140625,
      "learning_rate": 1.1130314256970542e-05,
      "loss": 0.0001,
      "step": 18174
    },
    {
      "epoch": 8.483080513418903,
      "grad_norm": 0.00604248046875,
      "learning_rate": 1.1123587491569998e-05,
      "loss": 0.0001,
      "step": 18175
    },
    {
      "epoch": 8.483547257876312,
      "grad_norm": 0.0245361328125,
      "learning_rate": 1.1116862639777225e-05,
      "loss": 0.0002,
      "step": 18176
    },
    {
      "epoch": 8.484014002333723,
      "grad_norm": 0.06884765625,
      "learning_rate": 1.1110139701737033e-05,
      "loss": 0.0018,
      "step": 18177
    },
    {
      "epoch": 8.484480746791132,
      "grad_norm": 0.006683349609375,
      "learning_rate": 1.1103418677594145e-05,
      "loss": 0.0001,
      "step": 18178
    },
    {
      "epoch": 8.48494749124854,
      "grad_norm": 0.057861328125,
      "learning_rate": 1.1096699567493307e-05,
      "loss": 0.0018,
      "step": 18179
    },
    {
      "epoch": 8.485414235705951,
      "grad_norm": 0.0146484375,
      "learning_rate": 1.1089982371579134e-05,
      "loss": 0.0002,
      "step": 18180
    },
    {
      "epoch": 8.48588098016336,
      "grad_norm": 0.0712890625,
      "learning_rate": 1.1083267089996296e-05,
      "loss": 0.0043,
      "step": 18181
    },
    {
      "epoch": 8.48634772462077,
      "grad_norm": 0.007354736328125,
      "learning_rate": 1.1076553722889393e-05,
      "loss": 0.0001,
      "step": 18182
    },
    {
      "epoch": 8.48681446907818,
      "grad_norm": 0.00531005859375,
      "learning_rate": 1.106984227040293e-05,
      "loss": 0.0001,
      "step": 18183
    },
    {
      "epoch": 8.487281213535589,
      "grad_norm": 0.0108642578125,
      "learning_rate": 1.1063132732681436e-05,
      "loss": 0.0001,
      "step": 18184
    },
    {
      "epoch": 8.487747957993,
      "grad_norm": 0.00994873046875,
      "learning_rate": 1.1056425109869384e-05,
      "loss": 0.0002,
      "step": 18185
    },
    {
      "epoch": 8.488214702450408,
      "grad_norm": 0.04833984375,
      "learning_rate": 1.1049719402111169e-05,
      "loss": 0.0002,
      "step": 18186
    },
    {
      "epoch": 8.488681446907817,
      "grad_norm": 0.006256103515625,
      "learning_rate": 1.1043015609551189e-05,
      "loss": 0.0001,
      "step": 18187
    },
    {
      "epoch": 8.489148191365228,
      "grad_norm": 0.11572265625,
      "learning_rate": 1.1036313732333802e-05,
      "loss": 0.0003,
      "step": 18188
    },
    {
      "epoch": 8.489614935822637,
      "grad_norm": 0.05126953125,
      "learning_rate": 1.1029613770603264e-05,
      "loss": 0.0031,
      "step": 18189
    },
    {
      "epoch": 8.490081680280047,
      "grad_norm": 0.01043701171875,
      "learning_rate": 1.1022915724503857e-05,
      "loss": 0.0001,
      "step": 18190
    },
    {
      "epoch": 8.490548424737456,
      "grad_norm": 0.00677490234375,
      "learning_rate": 1.1016219594179811e-05,
      "loss": 0.0001,
      "step": 18191
    },
    {
      "epoch": 8.491015169194865,
      "grad_norm": 0.00811767578125,
      "learning_rate": 1.1009525379775276e-05,
      "loss": 0.0001,
      "step": 18192
    },
    {
      "epoch": 8.491481913652276,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.1002833081434394e-05,
      "loss": 0.0001,
      "step": 18193
    },
    {
      "epoch": 8.491948658109685,
      "grad_norm": 0.0081787109375,
      "learning_rate": 1.0996142699301271e-05,
      "loss": 0.0001,
      "step": 18194
    },
    {
      "epoch": 8.492415402567094,
      "grad_norm": 0.00714111328125,
      "learning_rate": 1.0989454233519958e-05,
      "loss": 0.0001,
      "step": 18195
    },
    {
      "epoch": 8.492882147024504,
      "grad_norm": 0.010498046875,
      "learning_rate": 1.098276768423443e-05,
      "loss": 0.0001,
      "step": 18196
    },
    {
      "epoch": 8.493348891481913,
      "grad_norm": 0.00994873046875,
      "learning_rate": 1.0976083051588704e-05,
      "loss": 0.0001,
      "step": 18197
    },
    {
      "epoch": 8.493815635939324,
      "grad_norm": 0.006561279296875,
      "learning_rate": 1.0969400335726654e-05,
      "loss": 0.0001,
      "step": 18198
    },
    {
      "epoch": 8.494282380396733,
      "grad_norm": 0.01025390625,
      "learning_rate": 1.0962719536792198e-05,
      "loss": 0.0001,
      "step": 18199
    },
    {
      "epoch": 8.494749124854142,
      "grad_norm": 0.00408935546875,
      "learning_rate": 1.0956040654929211e-05,
      "loss": 0.0001,
      "step": 18200
    },
    {
      "epoch": 8.495215869311552,
      "grad_norm": 0.007476806640625,
      "learning_rate": 1.0949363690281433e-05,
      "loss": 0.0001,
      "step": 18201
    },
    {
      "epoch": 8.495682613768961,
      "grad_norm": 0.0186767578125,
      "learning_rate": 1.094268864299266e-05,
      "loss": 0.0002,
      "step": 18202
    },
    {
      "epoch": 8.49614935822637,
      "grad_norm": 0.0048828125,
      "learning_rate": 1.0936015513206632e-05,
      "loss": 0.0001,
      "step": 18203
    },
    {
      "epoch": 8.496616102683781,
      "grad_norm": 0.0966796875,
      "learning_rate": 1.0929344301066991e-05,
      "loss": 0.0058,
      "step": 18204
    },
    {
      "epoch": 8.49708284714119,
      "grad_norm": 0.01202392578125,
      "learning_rate": 1.0922675006717386e-05,
      "loss": 0.0001,
      "step": 18205
    },
    {
      "epoch": 8.4975495915986,
      "grad_norm": 0.012451171875,
      "learning_rate": 1.091600763030145e-05,
      "loss": 0.0001,
      "step": 18206
    },
    {
      "epoch": 8.49801633605601,
      "grad_norm": 0.0189208984375,
      "learning_rate": 1.0909342171962688e-05,
      "loss": 0.0002,
      "step": 18207
    },
    {
      "epoch": 8.498483080513418,
      "grad_norm": 0.0625,
      "learning_rate": 1.0902678631844632e-05,
      "loss": 0.0027,
      "step": 18208
    },
    {
      "epoch": 8.498949824970829,
      "grad_norm": 0.006988525390625,
      "learning_rate": 1.0896017010090787e-05,
      "loss": 0.0001,
      "step": 18209
    },
    {
      "epoch": 8.499416569428238,
      "grad_norm": 0.00567626953125,
      "learning_rate": 1.0889357306844528e-05,
      "loss": 0.0001,
      "step": 18210
    },
    {
      "epoch": 8.499883313885647,
      "grad_norm": 0.01116943359375,
      "learning_rate": 1.0882699522249306e-05,
      "loss": 0.0001,
      "step": 18211
    },
    {
      "epoch": 8.500350058343058,
      "grad_norm": 0.0048828125,
      "learning_rate": 1.087604365644842e-05,
      "loss": 0.0001,
      "step": 18212
    },
    {
      "epoch": 8.500816802800466,
      "grad_norm": 0.00750732421875,
      "learning_rate": 1.0869389709585187e-05,
      "loss": 0.0001,
      "step": 18213
    },
    {
      "epoch": 8.501283547257877,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.0862737681802914e-05,
      "loss": 0.0001,
      "step": 18214
    },
    {
      "epoch": 8.501750291715286,
      "grad_norm": 0.00994873046875,
      "learning_rate": 1.0856087573244789e-05,
      "loss": 0.0001,
      "step": 18215
    },
    {
      "epoch": 8.502217036172695,
      "grad_norm": 0.005828857421875,
      "learning_rate": 1.0849439384053995e-05,
      "loss": 0.0001,
      "step": 18216
    },
    {
      "epoch": 8.502683780630106,
      "grad_norm": 0.01025390625,
      "learning_rate": 1.0842793114373707e-05,
      "loss": 0.0001,
      "step": 18217
    },
    {
      "epoch": 8.503150525087515,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.0836148764346976e-05,
      "loss": 0.0002,
      "step": 18218
    },
    {
      "epoch": 8.503617269544923,
      "grad_norm": 0.003997802734375,
      "learning_rate": 1.0829506334116901e-05,
      "loss": 0.0001,
      "step": 18219
    },
    {
      "epoch": 8.504084014002334,
      "grad_norm": 0.006805419921875,
      "learning_rate": 1.0822865823826511e-05,
      "loss": 0.0001,
      "step": 18220
    },
    {
      "epoch": 8.504550758459743,
      "grad_norm": 0.046142578125,
      "learning_rate": 1.0816227233618736e-05,
      "loss": 0.0027,
      "step": 18221
    },
    {
      "epoch": 8.505017502917152,
      "grad_norm": 0.047119140625,
      "learning_rate": 1.0809590563636552e-05,
      "loss": 0.0015,
      "step": 18222
    },
    {
      "epoch": 8.505484247374563,
      "grad_norm": 0.006591796875,
      "learning_rate": 1.0802955814022852e-05,
      "loss": 0.0001,
      "step": 18223
    },
    {
      "epoch": 8.505950991831972,
      "grad_norm": 0.0196533203125,
      "learning_rate": 1.0796322984920459e-05,
      "loss": 0.0002,
      "step": 18224
    },
    {
      "epoch": 8.505950991831972,
      "eval_loss": 2.4529030323028564,
      "eval_runtime": 83.7213,
      "eval_samples_per_second": 21.548,
      "eval_steps_per_second": 2.699,
      "step": 18224
    },
    {
      "epoch": 8.506417736289382,
      "grad_norm": 0.01171875,
      "learning_rate": 1.0789692076472202e-05,
      "loss": 0.0002,
      "step": 18225
    },
    {
      "epoch": 8.506884480746791,
      "grad_norm": 0.00921630859375,
      "learning_rate": 1.0783063088820889e-05,
      "loss": 0.0001,
      "step": 18226
    },
    {
      "epoch": 8.5073512252042,
      "grad_norm": 0.0113525390625,
      "learning_rate": 1.0776436022109182e-05,
      "loss": 0.0001,
      "step": 18227
    },
    {
      "epoch": 8.50781796966161,
      "grad_norm": 0.00689697265625,
      "learning_rate": 1.0769810876479836e-05,
      "loss": 0.0001,
      "step": 18228
    },
    {
      "epoch": 8.50828471411902,
      "grad_norm": 0.091796875,
      "learning_rate": 1.0763187652075434e-05,
      "loss": 0.0027,
      "step": 18229
    },
    {
      "epoch": 8.50875145857643,
      "grad_norm": 0.052001953125,
      "learning_rate": 1.075656634903861e-05,
      "loss": 0.0002,
      "step": 18230
    },
    {
      "epoch": 8.50921820303384,
      "grad_norm": 0.0037689208984375,
      "learning_rate": 1.0749946967511959e-05,
      "loss": 0.0001,
      "step": 18231
    },
    {
      "epoch": 8.509684947491248,
      "grad_norm": 0.01190185546875,
      "learning_rate": 1.0743329507637945e-05,
      "loss": 0.0002,
      "step": 18232
    },
    {
      "epoch": 8.510151691948659,
      "grad_norm": 0.053955078125,
      "learning_rate": 1.0736713969559086e-05,
      "loss": 0.0029,
      "step": 18233
    },
    {
      "epoch": 8.510618436406068,
      "grad_norm": 0.00787353515625,
      "learning_rate": 1.0730100353417839e-05,
      "loss": 0.0001,
      "step": 18234
    },
    {
      "epoch": 8.511085180863477,
      "grad_norm": 0.031982421875,
      "learning_rate": 1.072348865935654e-05,
      "loss": 0.0001,
      "step": 18235
    },
    {
      "epoch": 8.511551925320887,
      "grad_norm": 0.0172119140625,
      "learning_rate": 1.07168788875176e-05,
      "loss": 0.0002,
      "step": 18236
    },
    {
      "epoch": 8.512018669778296,
      "grad_norm": 0.0166015625,
      "learning_rate": 1.071027103804334e-05,
      "loss": 0.0002,
      "step": 18237
    },
    {
      "epoch": 8.512485414235705,
      "grad_norm": 0.00909423828125,
      "learning_rate": 1.0703665111075977e-05,
      "loss": 0.0001,
      "step": 18238
    },
    {
      "epoch": 8.512952158693116,
      "grad_norm": 0.005218505859375,
      "learning_rate": 1.06970611067578e-05,
      "loss": 0.0001,
      "step": 18239
    },
    {
      "epoch": 8.513418903150525,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.0690459025230992e-05,
      "loss": 0.0001,
      "step": 18240
    },
    {
      "epoch": 8.513885647607935,
      "grad_norm": 0.04296875,
      "learning_rate": 1.0683858866637663e-05,
      "loss": 0.0016,
      "step": 18241
    },
    {
      "epoch": 8.514352392065344,
      "grad_norm": 0.010498046875,
      "learning_rate": 1.0677260631119979e-05,
      "loss": 0.0002,
      "step": 18242
    },
    {
      "epoch": 8.514819136522753,
      "grad_norm": 0.060791015625,
      "learning_rate": 1.0670664318819978e-05,
      "loss": 0.0002,
      "step": 18243
    },
    {
      "epoch": 8.515285880980164,
      "grad_norm": 0.0634765625,
      "learning_rate": 1.0664069929879671e-05,
      "loss": 0.0015,
      "step": 18244
    },
    {
      "epoch": 8.515752625437573,
      "grad_norm": 0.006561279296875,
      "learning_rate": 1.0657477464441056e-05,
      "loss": 0.0001,
      "step": 18245
    },
    {
      "epoch": 8.516219369894982,
      "grad_norm": 0.054931640625,
      "learning_rate": 1.0650886922646098e-05,
      "loss": 0.0002,
      "step": 18246
    },
    {
      "epoch": 8.516686114352392,
      "grad_norm": 0.060791015625,
      "learning_rate": 1.0644298304636657e-05,
      "loss": 0.0018,
      "step": 18247
    },
    {
      "epoch": 8.517152858809801,
      "grad_norm": 0.0078125,
      "learning_rate": 1.0637711610554613e-05,
      "loss": 0.0002,
      "step": 18248
    },
    {
      "epoch": 8.517619603267212,
      "grad_norm": 0.0125732421875,
      "learning_rate": 1.0631126840541817e-05,
      "loss": 0.0001,
      "step": 18249
    },
    {
      "epoch": 8.518086347724621,
      "grad_norm": 0.09375,
      "learning_rate": 1.0624543994739988e-05,
      "loss": 0.0047,
      "step": 18250
    },
    {
      "epoch": 8.51855309218203,
      "grad_norm": 0.007659912109375,
      "learning_rate": 1.0617963073290892e-05,
      "loss": 0.0001,
      "step": 18251
    },
    {
      "epoch": 8.51901983663944,
      "grad_norm": 0.007171630859375,
      "learning_rate": 1.061138407633624e-05,
      "loss": 0.0001,
      "step": 18252
    },
    {
      "epoch": 8.51948658109685,
      "grad_norm": 0.00714111328125,
      "learning_rate": 1.0604807004017658e-05,
      "loss": 0.0001,
      "step": 18253
    },
    {
      "epoch": 8.519953325554258,
      "grad_norm": 0.051513671875,
      "learning_rate": 1.059823185647676e-05,
      "loss": 0.0028,
      "step": 18254
    },
    {
      "epoch": 8.520420070011669,
      "grad_norm": 0.01409912109375,
      "learning_rate": 1.059165863385514e-05,
      "loss": 0.0002,
      "step": 18255
    },
    {
      "epoch": 8.520886814469078,
      "grad_norm": 0.01226806640625,
      "learning_rate": 1.0585087336294287e-05,
      "loss": 0.0002,
      "step": 18256
    },
    {
      "epoch": 8.521353558926489,
      "grad_norm": 0.02001953125,
      "learning_rate": 1.0578517963935719e-05,
      "loss": 0.0002,
      "step": 18257
    },
    {
      "epoch": 8.521820303383898,
      "grad_norm": 0.009765625,
      "learning_rate": 1.0571950516920892e-05,
      "loss": 0.0001,
      "step": 18258
    },
    {
      "epoch": 8.522287047841306,
      "grad_norm": 0.007659912109375,
      "learning_rate": 1.056538499539117e-05,
      "loss": 0.0001,
      "step": 18259
    },
    {
      "epoch": 8.522753792298717,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.0558821399487962e-05,
      "loss": 0.0001,
      "step": 18260
    },
    {
      "epoch": 8.523220536756126,
      "grad_norm": 0.01141357421875,
      "learning_rate": 1.0552259729352542e-05,
      "loss": 0.0002,
      "step": 18261
    },
    {
      "epoch": 8.523687281213535,
      "grad_norm": 0.0311279296875,
      "learning_rate": 1.0545699985126223e-05,
      "loss": 0.0003,
      "step": 18262
    },
    {
      "epoch": 8.524154025670946,
      "grad_norm": 0.06591796875,
      "learning_rate": 1.0539142166950245e-05,
      "loss": 0.0025,
      "step": 18263
    },
    {
      "epoch": 8.524620770128355,
      "grad_norm": 0.058837890625,
      "learning_rate": 1.0532586274965771e-05,
      "loss": 0.0029,
      "step": 18264
    },
    {
      "epoch": 8.525087514585763,
      "grad_norm": 0.00762939453125,
      "learning_rate": 1.0526032309313994e-05,
      "loss": 0.0001,
      "step": 18265
    },
    {
      "epoch": 8.525554259043174,
      "grad_norm": 0.06689453125,
      "learning_rate": 1.0519480270136029e-05,
      "loss": 0.0002,
      "step": 18266
    },
    {
      "epoch": 8.526021003500583,
      "grad_norm": 0.0206298828125,
      "learning_rate": 1.0512930157572908e-05,
      "loss": 0.0002,
      "step": 18267
    },
    {
      "epoch": 8.526487747957994,
      "grad_norm": 0.018310546875,
      "learning_rate": 1.0506381971765699e-05,
      "loss": 0.0002,
      "step": 18268
    },
    {
      "epoch": 8.526954492415403,
      "grad_norm": 0.052734375,
      "learning_rate": 1.0499835712855388e-05,
      "loss": 0.0023,
      "step": 18269
    },
    {
      "epoch": 8.527421236872812,
      "grad_norm": 0.009521484375,
      "learning_rate": 1.0493291380982906e-05,
      "loss": 0.0002,
      "step": 18270
    },
    {
      "epoch": 8.527887981330222,
      "grad_norm": 0.00616455078125,
      "learning_rate": 1.0486748976289162e-05,
      "loss": 0.0001,
      "step": 18271
    },
    {
      "epoch": 8.528354725787631,
      "grad_norm": 0.00531005859375,
      "learning_rate": 1.0480208498915056e-05,
      "loss": 0.0001,
      "step": 18272
    },
    {
      "epoch": 8.528821470245042,
      "grad_norm": 0.00592041015625,
      "learning_rate": 1.0473669949001363e-05,
      "loss": 0.0001,
      "step": 18273
    },
    {
      "epoch": 8.52928821470245,
      "grad_norm": 0.0194091796875,
      "learning_rate": 1.0467133326688894e-05,
      "loss": 0.0001,
      "step": 18274
    },
    {
      "epoch": 8.52975495915986,
      "grad_norm": 0.037841796875,
      "learning_rate": 1.046059863211839e-05,
      "loss": 0.001,
      "step": 18275
    },
    {
      "epoch": 8.53022170361727,
      "grad_norm": 0.0242919921875,
      "learning_rate": 1.045406586543053e-05,
      "loss": 0.0002,
      "step": 18276
    },
    {
      "epoch": 8.53068844807468,
      "grad_norm": 0.023681640625,
      "learning_rate": 1.0447535026765987e-05,
      "loss": 0.0002,
      "step": 18277
    },
    {
      "epoch": 8.531155192532088,
      "grad_norm": 0.0062255859375,
      "learning_rate": 1.0441006116265394e-05,
      "loss": 0.0001,
      "step": 18278
    },
    {
      "epoch": 8.531621936989499,
      "grad_norm": 0.037109375,
      "learning_rate": 1.0434479134069285e-05,
      "loss": 0.0023,
      "step": 18279
    },
    {
      "epoch": 8.532088681446908,
      "grad_norm": 0.046630859375,
      "learning_rate": 1.0427954080318236e-05,
      "loss": 0.0024,
      "step": 18280
    },
    {
      "epoch": 8.532555425904317,
      "grad_norm": 0.01220703125,
      "learning_rate": 1.04214309551527e-05,
      "loss": 0.0002,
      "step": 18281
    },
    {
      "epoch": 8.533022170361727,
      "grad_norm": 0.00946044921875,
      "learning_rate": 1.0414909758713142e-05,
      "loss": 0.0001,
      "step": 18282
    },
    {
      "epoch": 8.533488914819136,
      "grad_norm": 0.013916015625,
      "learning_rate": 1.0408390491139985e-05,
      "loss": 0.0001,
      "step": 18283
    },
    {
      "epoch": 8.533955659276547,
      "grad_norm": 0.0155029296875,
      "learning_rate": 1.0401873152573572e-05,
      "loss": 0.0002,
      "step": 18284
    },
    {
      "epoch": 8.534422403733956,
      "grad_norm": 0.01544189453125,
      "learning_rate": 1.0395357743154244e-05,
      "loss": 0.0001,
      "step": 18285
    },
    {
      "epoch": 8.534889148191365,
      "grad_norm": 0.013916015625,
      "learning_rate": 1.0388844263022302e-05,
      "loss": 0.0002,
      "step": 18286
    },
    {
      "epoch": 8.535355892648775,
      "grad_norm": 0.00897216796875,
      "learning_rate": 1.0382332712317944e-05,
      "loss": 0.0001,
      "step": 18287
    },
    {
      "epoch": 8.535822637106184,
      "grad_norm": 0.0089111328125,
      "learning_rate": 1.0375823091181392e-05,
      "loss": 0.0001,
      "step": 18288
    },
    {
      "epoch": 8.536289381563593,
      "grad_norm": 0.004913330078125,
      "learning_rate": 1.0369315399752832e-05,
      "loss": 0.0001,
      "step": 18289
    },
    {
      "epoch": 8.536756126021004,
      "grad_norm": 0.00628662109375,
      "learning_rate": 1.0362809638172366e-05,
      "loss": 0.0002,
      "step": 18290
    },
    {
      "epoch": 8.537222870478413,
      "grad_norm": 0.029052734375,
      "learning_rate": 1.0356305806580036e-05,
      "loss": 0.0002,
      "step": 18291
    },
    {
      "epoch": 8.537689614935823,
      "grad_norm": 0.007720947265625,
      "learning_rate": 1.0349803905115918e-05,
      "loss": 0.0001,
      "step": 18292
    },
    {
      "epoch": 8.538156359393232,
      "grad_norm": 0.02880859375,
      "learning_rate": 1.0343303933919979e-05,
      "loss": 0.0003,
      "step": 18293
    },
    {
      "epoch": 8.538623103850641,
      "grad_norm": 0.015625,
      "learning_rate": 1.0336805893132173e-05,
      "loss": 0.0002,
      "step": 18294
    },
    {
      "epoch": 8.539089848308052,
      "grad_norm": 0.0157470703125,
      "learning_rate": 1.0330309782892444e-05,
      "loss": 0.0001,
      "step": 18295
    },
    {
      "epoch": 8.53955659276546,
      "grad_norm": 0.01043701171875,
      "learning_rate": 1.0323815603340625e-05,
      "loss": 0.0001,
      "step": 18296
    },
    {
      "epoch": 8.54002333722287,
      "grad_norm": 0.0113525390625,
      "learning_rate": 1.0317323354616537e-05,
      "loss": 0.0001,
      "step": 18297
    },
    {
      "epoch": 8.54049008168028,
      "grad_norm": 0.00732421875,
      "learning_rate": 1.0310833036860012e-05,
      "loss": 0.0001,
      "step": 18298
    },
    {
      "epoch": 8.54095682613769,
      "grad_norm": 0.011962890625,
      "learning_rate": 1.030434465021074e-05,
      "loss": 0.0001,
      "step": 18299
    },
    {
      "epoch": 8.5414235705951,
      "grad_norm": 0.0106201171875,
      "learning_rate": 1.0297858194808452e-05,
      "loss": 0.0001,
      "step": 18300
    },
    {
      "epoch": 8.541890315052509,
      "grad_norm": 0.010498046875,
      "learning_rate": 1.0291373670792815e-05,
      "loss": 0.0001,
      "step": 18301
    },
    {
      "epoch": 8.542357059509918,
      "grad_norm": 0.014404296875,
      "learning_rate": 1.0284891078303426e-05,
      "loss": 0.0001,
      "step": 18302
    },
    {
      "epoch": 8.542823803967329,
      "grad_norm": 0.0311279296875,
      "learning_rate": 1.0278410417479867e-05,
      "loss": 0.0002,
      "step": 18303
    },
    {
      "epoch": 8.543290548424737,
      "grad_norm": 0.0186767578125,
      "learning_rate": 1.02719316884617e-05,
      "loss": 0.0002,
      "step": 18304
    },
    {
      "epoch": 8.543757292882146,
      "grad_norm": 0.0191650390625,
      "learning_rate": 1.0265454891388383e-05,
      "loss": 0.0002,
      "step": 18305
    },
    {
      "epoch": 8.544224037339557,
      "grad_norm": 0.0264892578125,
      "learning_rate": 1.0258980026399379e-05,
      "loss": 0.0002,
      "step": 18306
    },
    {
      "epoch": 8.544690781796966,
      "grad_norm": 0.0286865234375,
      "learning_rate": 1.0252507093634123e-05,
      "loss": 0.002,
      "step": 18307
    },
    {
      "epoch": 8.545157526254375,
      "grad_norm": 0.01055908203125,
      "learning_rate": 1.0246036093231947e-05,
      "loss": 0.0001,
      "step": 18308
    },
    {
      "epoch": 8.545624270711786,
      "grad_norm": 0.01434326171875,
      "learning_rate": 1.023956702533222e-05,
      "loss": 0.0002,
      "step": 18309
    },
    {
      "epoch": 8.546091015169194,
      "grad_norm": 0.00811767578125,
      "learning_rate": 1.0233099890074182e-05,
      "loss": 0.0001,
      "step": 18310
    },
    {
      "epoch": 8.546557759626605,
      "grad_norm": 0.0045166015625,
      "learning_rate": 1.0226634687597114e-05,
      "loss": 0.0001,
      "step": 18311
    },
    {
      "epoch": 8.547024504084014,
      "grad_norm": 0.06982421875,
      "learning_rate": 1.0220171418040214e-05,
      "loss": 0.0023,
      "step": 18312
    },
    {
      "epoch": 8.547491248541423,
      "grad_norm": 0.00811767578125,
      "learning_rate": 1.0213710081542616e-05,
      "loss": 0.0001,
      "step": 18313
    },
    {
      "epoch": 8.547957992998834,
      "grad_norm": 0.00579833984375,
      "learning_rate": 1.0207250678243463e-05,
      "loss": 0.0001,
      "step": 18314
    },
    {
      "epoch": 8.548424737456243,
      "grad_norm": 0.0081787109375,
      "learning_rate": 1.0200793208281844e-05,
      "loss": 0.0002,
      "step": 18315
    },
    {
      "epoch": 8.548891481913651,
      "grad_norm": 0.126953125,
      "learning_rate": 1.0194337671796772e-05,
      "loss": 0.0017,
      "step": 18316
    },
    {
      "epoch": 8.549358226371062,
      "grad_norm": 0.0086669921875,
      "learning_rate": 1.0187884068927244e-05,
      "loss": 0.0001,
      "step": 18317
    },
    {
      "epoch": 8.549824970828471,
      "grad_norm": 0.01336669921875,
      "learning_rate": 1.0181432399812241e-05,
      "loss": 0.0002,
      "step": 18318
    },
    {
      "epoch": 8.550291715285882,
      "grad_norm": 0.01263427734375,
      "learning_rate": 1.0174982664590638e-05,
      "loss": 0.0002,
      "step": 18319
    },
    {
      "epoch": 8.55075845974329,
      "grad_norm": 0.030029296875,
      "learning_rate": 1.0168534863401314e-05,
      "loss": 0.0025,
      "step": 18320
    },
    {
      "epoch": 8.5512252042007,
      "grad_norm": 0.0634765625,
      "learning_rate": 1.0162088996383124e-05,
      "loss": 0.0032,
      "step": 18321
    },
    {
      "epoch": 8.55169194865811,
      "grad_norm": 0.005340576171875,
      "learning_rate": 1.0155645063674824e-05,
      "loss": 0.0001,
      "step": 18322
    },
    {
      "epoch": 8.55215869311552,
      "grad_norm": 0.04150390625,
      "learning_rate": 1.0149203065415158e-05,
      "loss": 0.0002,
      "step": 18323
    },
    {
      "epoch": 8.552625437572928,
      "grad_norm": 0.08984375,
      "learning_rate": 1.0142763001742873e-05,
      "loss": 0.0051,
      "step": 18324
    },
    {
      "epoch": 8.553092182030339,
      "grad_norm": 0.0888671875,
      "learning_rate": 1.0136324872796576e-05,
      "loss": 0.0044,
      "step": 18325
    },
    {
      "epoch": 8.553558926487748,
      "grad_norm": 0.00799560546875,
      "learning_rate": 1.0129888678714905e-05,
      "loss": 0.0002,
      "step": 18326
    },
    {
      "epoch": 8.554025670945158,
      "grad_norm": 0.0167236328125,
      "learning_rate": 1.0123454419636458e-05,
      "loss": 0.0002,
      "step": 18327
    },
    {
      "epoch": 8.554492415402567,
      "grad_norm": 0.0157470703125,
      "learning_rate": 1.0117022095699746e-05,
      "loss": 0.0001,
      "step": 18328
    },
    {
      "epoch": 8.554959159859976,
      "grad_norm": 0.006866455078125,
      "learning_rate": 1.0110591707043293e-05,
      "loss": 0.0002,
      "step": 18329
    },
    {
      "epoch": 8.555425904317387,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.0104163253805521e-05,
      "loss": 0.0001,
      "step": 18330
    },
    {
      "epoch": 8.555892648774796,
      "grad_norm": 0.00848388671875,
      "learning_rate": 1.0097736736124852e-05,
      "loss": 0.0001,
      "step": 18331
    },
    {
      "epoch": 8.556359393232205,
      "grad_norm": 0.0125732421875,
      "learning_rate": 1.0091312154139677e-05,
      "loss": 0.0001,
      "step": 18332
    },
    {
      "epoch": 8.556826137689615,
      "grad_norm": 0.006988525390625,
      "learning_rate": 1.0084889507988282e-05,
      "loss": 0.0001,
      "step": 18333
    },
    {
      "epoch": 8.557292882147024,
      "grad_norm": 0.0380859375,
      "learning_rate": 1.0078468797808982e-05,
      "loss": 0.0017,
      "step": 18334
    },
    {
      "epoch": 8.557759626604435,
      "grad_norm": 0.0162353515625,
      "learning_rate": 1.0072050023740043e-05,
      "loss": 0.0001,
      "step": 18335
    },
    {
      "epoch": 8.558226371061844,
      "grad_norm": 0.00634765625,
      "learning_rate": 1.0065633185919621e-05,
      "loss": 0.0001,
      "step": 18336
    },
    {
      "epoch": 8.558693115519253,
      "grad_norm": 0.00616455078125,
      "learning_rate": 1.0059218284485895e-05,
      "loss": 0.0001,
      "step": 18337
    },
    {
      "epoch": 8.559159859976663,
      "grad_norm": 0.013916015625,
      "learning_rate": 1.0052805319577018e-05,
      "loss": 0.0001,
      "step": 18338
    },
    {
      "epoch": 8.559626604434072,
      "grad_norm": 0.00872802734375,
      "learning_rate": 1.0046394291331029e-05,
      "loss": 0.0001,
      "step": 18339
    },
    {
      "epoch": 8.560093348891481,
      "grad_norm": 0.005126953125,
      "learning_rate": 1.003998519988597e-05,
      "loss": 0.0001,
      "step": 18340
    },
    {
      "epoch": 8.560560093348892,
      "grad_norm": 0.00885009765625,
      "learning_rate": 1.0033578045379854e-05,
      "loss": 0.0001,
      "step": 18341
    },
    {
      "epoch": 8.5610268378063,
      "grad_norm": 0.005584716796875,
      "learning_rate": 1.0027172827950604e-05,
      "loss": 0.0001,
      "step": 18342
    },
    {
      "epoch": 8.561493582263711,
      "grad_norm": 0.00653076171875,
      "learning_rate": 1.0020769547736153e-05,
      "loss": 0.0001,
      "step": 18343
    },
    {
      "epoch": 8.56196032672112,
      "grad_norm": 0.060302734375,
      "learning_rate": 1.001436820487438e-05,
      "loss": 0.0013,
      "step": 18344
    },
    {
      "epoch": 8.56242707117853,
      "grad_norm": 0.0107421875,
      "learning_rate": 1.0007968799503086e-05,
      "loss": 0.0001,
      "step": 18345
    },
    {
      "epoch": 8.56289381563594,
      "grad_norm": 0.00787353515625,
      "learning_rate": 1.0001571331760073e-05,
      "loss": 0.0001,
      "step": 18346
    },
    {
      "epoch": 8.563360560093349,
      "grad_norm": 0.0107421875,
      "learning_rate": 9.995175801783108e-06,
      "loss": 0.0001,
      "step": 18347
    },
    {
      "epoch": 8.563827304550758,
      "grad_norm": 0.007110595703125,
      "learning_rate": 9.988782209709846e-06,
      "loss": 0.0002,
      "step": 18348
    },
    {
      "epoch": 8.564294049008168,
      "grad_norm": 0.050537109375,
      "learning_rate": 9.982390555677968e-06,
      "loss": 0.0012,
      "step": 18349
    },
    {
      "epoch": 8.564760793465577,
      "grad_norm": 0.09716796875,
      "learning_rate": 9.976000839825118e-06,
      "loss": 0.0038,
      "step": 18350
    },
    {
      "epoch": 8.565227537922986,
      "grad_norm": 0.07958984375,
      "learning_rate": 9.96961306228883e-06,
      "loss": 0.0023,
      "step": 18351
    },
    {
      "epoch": 8.565694282380397,
      "grad_norm": 0.025146484375,
      "learning_rate": 9.963227223206672e-06,
      "loss": 0.0001,
      "step": 18352
    },
    {
      "epoch": 8.566161026837806,
      "grad_norm": 0.00616455078125,
      "learning_rate": 9.956843322716148e-06,
      "loss": 0.0001,
      "step": 18353
    },
    {
      "epoch": 8.566627771295217,
      "grad_norm": 0.03857421875,
      "learning_rate": 9.950461360954667e-06,
      "loss": 0.0022,
      "step": 18354
    },
    {
      "epoch": 8.567094515752625,
      "grad_norm": 0.01055908203125,
      "learning_rate": 9.944081338059663e-06,
      "loss": 0.0002,
      "step": 18355
    },
    {
      "epoch": 8.567561260210034,
      "grad_norm": 0.01519775390625,
      "learning_rate": 9.93770325416854e-06,
      "loss": 0.0002,
      "step": 18356
    },
    {
      "epoch": 8.568028004667445,
      "grad_norm": 0.007598876953125,
      "learning_rate": 9.931327109418554e-06,
      "loss": 0.0001,
      "step": 18357
    },
    {
      "epoch": 8.568494749124854,
      "grad_norm": 0.007415771484375,
      "learning_rate": 9.92495290394706e-06,
      "loss": 0.0001,
      "step": 18358
    },
    {
      "epoch": 8.568961493582263,
      "grad_norm": 0.0286865234375,
      "learning_rate": 9.91858063789125e-06,
      "loss": 0.0002,
      "step": 18359
    },
    {
      "epoch": 8.569428238039674,
      "grad_norm": 0.005218505859375,
      "learning_rate": 9.912210311388336e-06,
      "loss": 0.0001,
      "step": 18360
    },
    {
      "epoch": 8.569894982497082,
      "grad_norm": 0.0888671875,
      "learning_rate": 9.90584192457552e-06,
      "loss": 0.005,
      "step": 18361
    },
    {
      "epoch": 8.570361726954493,
      "grad_norm": 0.00726318359375,
      "learning_rate": 9.899475477589858e-06,
      "loss": 0.0001,
      "step": 18362
    },
    {
      "epoch": 8.570828471411902,
      "grad_norm": 0.04248046875,
      "learning_rate": 9.893110970568465e-06,
      "loss": 0.0014,
      "step": 18363
    },
    {
      "epoch": 8.571295215869311,
      "grad_norm": 0.0208740234375,
      "learning_rate": 9.886748403648383e-06,
      "loss": 0.0004,
      "step": 18364
    },
    {
      "epoch": 8.571761960326722,
      "grad_norm": 0.00537109375,
      "learning_rate": 9.880387776966582e-06,
      "loss": 0.0001,
      "step": 18365
    },
    {
      "epoch": 8.57222870478413,
      "grad_norm": 0.05810546875,
      "learning_rate": 9.874029090660008e-06,
      "loss": 0.0031,
      "step": 18366
    },
    {
      "epoch": 8.57269544924154,
      "grad_norm": 0.0159912109375,
      "learning_rate": 9.867672344865608e-06,
      "loss": 0.0001,
      "step": 18367
    },
    {
      "epoch": 8.57316219369895,
      "grad_norm": 0.0517578125,
      "learning_rate": 9.861317539720205e-06,
      "loss": 0.0034,
      "step": 18368
    },
    {
      "epoch": 8.573628938156359,
      "grad_norm": 0.00714111328125,
      "learning_rate": 9.854964675360646e-06,
      "loss": 0.0001,
      "step": 18369
    },
    {
      "epoch": 8.57409568261377,
      "grad_norm": 0.007080078125,
      "learning_rate": 9.84861375192374e-06,
      "loss": 0.0001,
      "step": 18370
    },
    {
      "epoch": 8.574562427071179,
      "grad_norm": 0.05859375,
      "learning_rate": 9.842264769546184e-06,
      "loss": 0.0027,
      "step": 18371
    },
    {
      "epoch": 8.575029171528588,
      "grad_norm": 0.007171630859375,
      "learning_rate": 9.835917728364697e-06,
      "loss": 0.0001,
      "step": 18372
    },
    {
      "epoch": 8.575495915985998,
      "grad_norm": 0.005889892578125,
      "learning_rate": 9.82957262851596e-06,
      "loss": 0.0002,
      "step": 18373
    },
    {
      "epoch": 8.575962660443407,
      "grad_norm": 0.01483154296875,
      "learning_rate": 9.82322947013654e-06,
      "loss": 0.0001,
      "step": 18374
    },
    {
      "epoch": 8.576429404900816,
      "grad_norm": 0.042724609375,
      "learning_rate": 9.816888253363055e-06,
      "loss": 0.0016,
      "step": 18375
    },
    {
      "epoch": 8.576896149358227,
      "grad_norm": 0.058837890625,
      "learning_rate": 9.810548978332046e-06,
      "loss": 0.0029,
      "step": 18376
    },
    {
      "epoch": 8.577362893815636,
      "grad_norm": 0.004913330078125,
      "learning_rate": 9.804211645179961e-06,
      "loss": 0.0001,
      "step": 18377
    },
    {
      "epoch": 8.577829638273045,
      "grad_norm": 0.01263427734375,
      "learning_rate": 9.797876254043293e-06,
      "loss": 0.0001,
      "step": 18378
    },
    {
      "epoch": 8.578296382730455,
      "grad_norm": 0.04931640625,
      "learning_rate": 9.791542805058419e-06,
      "loss": 0.0028,
      "step": 18379
    },
    {
      "epoch": 8.578763127187864,
      "grad_norm": 0.044921875,
      "learning_rate": 9.785211298361707e-06,
      "loss": 0.0002,
      "step": 18380
    },
    {
      "epoch": 8.579229871645275,
      "grad_norm": 0.0189208984375,
      "learning_rate": 9.778881734089507e-06,
      "loss": 0.0002,
      "step": 18381
    },
    {
      "epoch": 8.579696616102684,
      "grad_norm": 0.005584716796875,
      "learning_rate": 9.772554112378075e-06,
      "loss": 0.0001,
      "step": 18382
    },
    {
      "epoch": 8.580163360560093,
      "grad_norm": 0.01129150390625,
      "learning_rate": 9.766228433363655e-06,
      "loss": 0.0001,
      "step": 18383
    },
    {
      "epoch": 8.580630105017503,
      "grad_norm": 0.0177001953125,
      "learning_rate": 9.759904697182475e-06,
      "loss": 0.0002,
      "step": 18384
    },
    {
      "epoch": 8.581096849474912,
      "grad_norm": 0.01416015625,
      "learning_rate": 9.753582903970648e-06,
      "loss": 0.0001,
      "step": 18385
    },
    {
      "epoch": 8.581563593932323,
      "grad_norm": 0.1044921875,
      "learning_rate": 9.74726305386433e-06,
      "loss": 0.0042,
      "step": 18386
    },
    {
      "epoch": 8.582030338389732,
      "grad_norm": 0.0927734375,
      "learning_rate": 9.740945146999558e-06,
      "loss": 0.0046,
      "step": 18387
    },
    {
      "epoch": 8.58249708284714,
      "grad_norm": 0.0068359375,
      "learning_rate": 9.734629183512411e-06,
      "loss": 0.0001,
      "step": 18388
    },
    {
      "epoch": 8.582963827304551,
      "grad_norm": 0.01458740234375,
      "learning_rate": 9.728315163538815e-06,
      "loss": 0.0001,
      "step": 18389
    },
    {
      "epoch": 8.58343057176196,
      "grad_norm": 0.01611328125,
      "learning_rate": 9.722003087214782e-06,
      "loss": 0.0002,
      "step": 18390
    },
    {
      "epoch": 8.58389731621937,
      "grad_norm": 0.0198974609375,
      "learning_rate": 9.715692954676158e-06,
      "loss": 0.0001,
      "step": 18391
    },
    {
      "epoch": 8.58436406067678,
      "grad_norm": 0.04736328125,
      "learning_rate": 9.709384766058837e-06,
      "loss": 0.0006,
      "step": 18392
    },
    {
      "epoch": 8.584830805134189,
      "grad_norm": 0.0123291015625,
      "learning_rate": 9.703078521498665e-06,
      "loss": 0.0001,
      "step": 18393
    },
    {
      "epoch": 8.585297549591598,
      "grad_norm": 0.009765625,
      "learning_rate": 9.696774221131389e-06,
      "loss": 0.0001,
      "step": 18394
    },
    {
      "epoch": 8.585764294049008,
      "grad_norm": 0.0057373046875,
      "learning_rate": 9.690471865092742e-06,
      "loss": 0.0001,
      "step": 18395
    },
    {
      "epoch": 8.586231038506417,
      "grad_norm": 0.017822265625,
      "learning_rate": 9.684171453518464e-06,
      "loss": 0.0002,
      "step": 18396
    },
    {
      "epoch": 8.586697782963828,
      "grad_norm": 0.01531982421875,
      "learning_rate": 9.677872986544157e-06,
      "loss": 0.0002,
      "step": 18397
    },
    {
      "epoch": 8.587164527421237,
      "grad_norm": 0.1640625,
      "learning_rate": 9.671576464305466e-06,
      "loss": 0.0004,
      "step": 18398
    },
    {
      "epoch": 8.587631271878646,
      "grad_norm": 0.0059814453125,
      "learning_rate": 9.66528188693796e-06,
      "loss": 0.0001,
      "step": 18399
    },
    {
      "epoch": 8.588098016336057,
      "grad_norm": 0.004852294921875,
      "learning_rate": 9.658989254577155e-06,
      "loss": 0.0001,
      "step": 18400
    },
    {
      "epoch": 8.588564760793465,
      "grad_norm": 0.01513671875,
      "learning_rate": 9.652698567358543e-06,
      "loss": 0.0001,
      "step": 18401
    },
    {
      "epoch": 8.589031505250874,
      "grad_norm": 0.015380859375,
      "learning_rate": 9.6464098254176e-06,
      "loss": 0.0002,
      "step": 18402
    },
    {
      "epoch": 8.589498249708285,
      "grad_norm": 0.0093994140625,
      "learning_rate": 9.640123028889669e-06,
      "loss": 0.0001,
      "step": 18403
    },
    {
      "epoch": 8.589964994165694,
      "grad_norm": 0.00909423828125,
      "learning_rate": 9.63383817791016e-06,
      "loss": 0.0001,
      "step": 18404
    },
    {
      "epoch": 8.590431738623105,
      "grad_norm": 0.00592041015625,
      "learning_rate": 9.627555272614386e-06,
      "loss": 0.0001,
      "step": 18405
    },
    {
      "epoch": 8.590898483080514,
      "grad_norm": 0.060546875,
      "learning_rate": 9.621274313137607e-06,
      "loss": 0.0026,
      "step": 18406
    },
    {
      "epoch": 8.591365227537922,
      "grad_norm": 0.0634765625,
      "learning_rate": 9.614995299615081e-06,
      "loss": 0.0037,
      "step": 18407
    },
    {
      "epoch": 8.591831971995333,
      "grad_norm": 0.01104736328125,
      "learning_rate": 9.608718232181967e-06,
      "loss": 0.0002,
      "step": 18408
    },
    {
      "epoch": 8.592298716452742,
      "grad_norm": 0.054931640625,
      "learning_rate": 9.602443110973446e-06,
      "loss": 0.0011,
      "step": 18409
    },
    {
      "epoch": 8.592765460910151,
      "grad_norm": 0.01409912109375,
      "learning_rate": 9.596169936124643e-06,
      "loss": 0.0002,
      "step": 18410
    },
    {
      "epoch": 8.593232205367562,
      "grad_norm": 0.010498046875,
      "learning_rate": 9.589898707770594e-06,
      "loss": 0.0002,
      "step": 18411
    },
    {
      "epoch": 8.59369894982497,
      "grad_norm": 0.0289306640625,
      "learning_rate": 9.583629426046326e-06,
      "loss": 0.0004,
      "step": 18412
    },
    {
      "epoch": 8.594165694282381,
      "grad_norm": 0.006927490234375,
      "learning_rate": 9.57736209108686e-06,
      "loss": 0.0002,
      "step": 18413
    },
    {
      "epoch": 8.59463243873979,
      "grad_norm": 0.064453125,
      "learning_rate": 9.571096703027093e-06,
      "loss": 0.0026,
      "step": 18414
    },
    {
      "epoch": 8.595099183197199,
      "grad_norm": 0.0054931640625,
      "learning_rate": 9.564833262001948e-06,
      "loss": 0.0001,
      "step": 18415
    },
    {
      "epoch": 8.59556592765461,
      "grad_norm": 0.01080322265625,
      "learning_rate": 9.558571768146296e-06,
      "loss": 0.0002,
      "step": 18416
    },
    {
      "epoch": 8.596032672112019,
      "grad_norm": 0.01129150390625,
      "learning_rate": 9.552312221594928e-06,
      "loss": 0.0001,
      "step": 18417
    },
    {
      "epoch": 8.596499416569428,
      "grad_norm": 0.0286865234375,
      "learning_rate": 9.546054622482625e-06,
      "loss": 0.0002,
      "step": 18418
    },
    {
      "epoch": 8.596966161026838,
      "grad_norm": 0.03173828125,
      "learning_rate": 9.539798970944148e-06,
      "loss": 0.0014,
      "step": 18419
    },
    {
      "epoch": 8.597432905484247,
      "grad_norm": 0.0076904296875,
      "learning_rate": 9.533545267114152e-06,
      "loss": 0.0001,
      "step": 18420
    },
    {
      "epoch": 8.597899649941656,
      "grad_norm": 0.005157470703125,
      "learning_rate": 9.527293511127289e-06,
      "loss": 0.0001,
      "step": 18421
    },
    {
      "epoch": 8.598366394399067,
      "grad_norm": 0.0242919921875,
      "learning_rate": 9.521043703118204e-06,
      "loss": 0.0003,
      "step": 18422
    },
    {
      "epoch": 8.598833138856476,
      "grad_norm": 0.040771484375,
      "learning_rate": 9.514795843221425e-06,
      "loss": 0.0003,
      "step": 18423
    },
    {
      "epoch": 8.599299883313886,
      "grad_norm": 0.03955078125,
      "learning_rate": 9.508549931571475e-06,
      "loss": 0.0026,
      "step": 18424
    },
    {
      "epoch": 8.599766627771295,
      "grad_norm": 0.08837890625,
      "learning_rate": 9.502305968302872e-06,
      "loss": 0.0049,
      "step": 18425
    },
    {
      "epoch": 8.600233372228704,
      "grad_norm": 0.0096435546875,
      "learning_rate": 9.496063953550004e-06,
      "loss": 0.0001,
      "step": 18426
    },
    {
      "epoch": 8.600700116686115,
      "grad_norm": 0.0135498046875,
      "learning_rate": 9.489823887447325e-06,
      "loss": 0.0002,
      "step": 18427
    },
    {
      "epoch": 8.601166861143524,
      "grad_norm": 0.06689453125,
      "learning_rate": 9.483585770129133e-06,
      "loss": 0.0018,
      "step": 18428
    },
    {
      "epoch": 8.601633605600934,
      "grad_norm": 0.0157470703125,
      "learning_rate": 9.477349601729779e-06,
      "loss": 0.0001,
      "step": 18429
    },
    {
      "epoch": 8.602100350058343,
      "grad_norm": 0.01361083984375,
      "learning_rate": 9.471115382383533e-06,
      "loss": 0.0001,
      "step": 18430
    },
    {
      "epoch": 8.602567094515752,
      "grad_norm": 0.004913330078125,
      "learning_rate": 9.464883112224598e-06,
      "loss": 0.0001,
      "step": 18431
    },
    {
      "epoch": 8.603033838973163,
      "grad_norm": 0.013916015625,
      "learning_rate": 9.45865279138719e-06,
      "loss": 0.0001,
      "step": 18432
    },
    {
      "epoch": 8.603500583430572,
      "grad_norm": 0.1025390625,
      "learning_rate": 9.452424420005456e-06,
      "loss": 0.003,
      "step": 18433
    },
    {
      "epoch": 8.60396732788798,
      "grad_norm": 0.00555419921875,
      "learning_rate": 9.446197998213491e-06,
      "loss": 0.0001,
      "step": 18434
    },
    {
      "epoch": 8.604434072345391,
      "grad_norm": 0.01470947265625,
      "learning_rate": 9.439973526145329e-06,
      "loss": 0.0001,
      "step": 18435
    },
    {
      "epoch": 8.6049008168028,
      "grad_norm": 0.006439208984375,
      "learning_rate": 9.43375100393502e-06,
      "loss": 0.0002,
      "step": 18436
    },
    {
      "epoch": 8.60536756126021,
      "grad_norm": 0.013427734375,
      "learning_rate": 9.427530431716558e-06,
      "loss": 0.0001,
      "step": 18437
    },
    {
      "epoch": 8.60583430571762,
      "grad_norm": 0.0078125,
      "learning_rate": 9.421311809623835e-06,
      "loss": 0.0001,
      "step": 18438
    },
    {
      "epoch": 8.606301050175029,
      "grad_norm": 0.017578125,
      "learning_rate": 9.415095137790785e-06,
      "loss": 0.0002,
      "step": 18439
    },
    {
      "epoch": 8.60676779463244,
      "grad_norm": 0.00689697265625,
      "learning_rate": 9.40888041635123e-06,
      "loss": 0.0001,
      "step": 18440
    },
    {
      "epoch": 8.607234539089848,
      "grad_norm": 0.00909423828125,
      "learning_rate": 9.40266764543899e-06,
      "loss": 0.0001,
      "step": 18441
    },
    {
      "epoch": 8.607701283547257,
      "grad_norm": 0.0615234375,
      "learning_rate": 9.39645682518785e-06,
      "loss": 0.0031,
      "step": 18442
    },
    {
      "epoch": 8.608168028004668,
      "grad_norm": 0.0242919921875,
      "learning_rate": 9.390247955731512e-06,
      "loss": 0.0002,
      "step": 18443
    },
    {
      "epoch": 8.608634772462077,
      "grad_norm": 0.01123046875,
      "learning_rate": 9.384041037203661e-06,
      "loss": 0.0001,
      "step": 18444
    },
    {
      "epoch": 8.609101516919486,
      "grad_norm": 0.007354736328125,
      "learning_rate": 9.377836069737966e-06,
      "loss": 0.0001,
      "step": 18445
    },
    {
      "epoch": 8.609568261376896,
      "grad_norm": 0.0162353515625,
      "learning_rate": 9.371633053467988e-06,
      "loss": 0.0002,
      "step": 18446
    },
    {
      "epoch": 8.610035005834305,
      "grad_norm": 0.0272216796875,
      "learning_rate": 9.365431988527307e-06,
      "loss": 0.0002,
      "step": 18447
    },
    {
      "epoch": 8.610501750291716,
      "grad_norm": 0.00811767578125,
      "learning_rate": 9.35923287504945e-06,
      "loss": 0.0002,
      "step": 18448
    },
    {
      "epoch": 8.610968494749125,
      "grad_norm": 0.006591796875,
      "learning_rate": 9.353035713167858e-06,
      "loss": 0.0001,
      "step": 18449
    },
    {
      "epoch": 8.611435239206534,
      "grad_norm": 0.009033203125,
      "learning_rate": 9.346840503015985e-06,
      "loss": 0.0002,
      "step": 18450
    },
    {
      "epoch": 8.611901983663945,
      "grad_norm": 0.0301513671875,
      "learning_rate": 9.340647244727241e-06,
      "loss": 0.0002,
      "step": 18451
    },
    {
      "epoch": 8.612368728121353,
      "grad_norm": 0.032958984375,
      "learning_rate": 9.334455938434916e-06,
      "loss": 0.0002,
      "step": 18452
    },
    {
      "epoch": 8.612835472578762,
      "grad_norm": 0.01116943359375,
      "learning_rate": 9.32826658427236e-06,
      "loss": 0.0002,
      "step": 18453
    },
    {
      "epoch": 8.613302217036173,
      "grad_norm": 0.06591796875,
      "learning_rate": 9.322079182372845e-06,
      "loss": 0.006,
      "step": 18454
    },
    {
      "epoch": 8.613768961493582,
      "grad_norm": 0.007781982421875,
      "learning_rate": 9.315893732869563e-06,
      "loss": 0.0001,
      "step": 18455
    },
    {
      "epoch": 8.614235705950993,
      "grad_norm": 0.0123291015625,
      "learning_rate": 9.309710235895697e-06,
      "loss": 0.0001,
      "step": 18456
    },
    {
      "epoch": 8.614702450408402,
      "grad_norm": 0.0296630859375,
      "learning_rate": 9.303528691584407e-06,
      "loss": 0.0002,
      "step": 18457
    },
    {
      "epoch": 8.61516919486581,
      "grad_norm": 0.005706787109375,
      "learning_rate": 9.297349100068764e-06,
      "loss": 0.0001,
      "step": 18458
    },
    {
      "epoch": 8.615635939323221,
      "grad_norm": 0.0079345703125,
      "learning_rate": 9.29117146148184e-06,
      "loss": 0.0001,
      "step": 18459
    },
    {
      "epoch": 8.61610268378063,
      "grad_norm": 0.0130615234375,
      "learning_rate": 9.284995775956628e-06,
      "loss": 0.0002,
      "step": 18460
    },
    {
      "epoch": 8.616569428238039,
      "grad_norm": 0.00689697265625,
      "learning_rate": 9.2788220436261e-06,
      "loss": 0.0001,
      "step": 18461
    },
    {
      "epoch": 8.61703617269545,
      "grad_norm": 0.0042724609375,
      "learning_rate": 9.272650264623206e-06,
      "loss": 0.0001,
      "step": 18462
    },
    {
      "epoch": 8.617502917152859,
      "grad_norm": 0.007354736328125,
      "learning_rate": 9.266480439080804e-06,
      "loss": 0.0002,
      "step": 18463
    },
    {
      "epoch": 8.617969661610267,
      "grad_norm": 0.00726318359375,
      "learning_rate": 9.260312567131746e-06,
      "loss": 0.0001,
      "step": 18464
    },
    {
      "epoch": 8.618436406067678,
      "grad_norm": 0.068359375,
      "learning_rate": 9.254146648908857e-06,
      "loss": 0.0042,
      "step": 18465
    },
    {
      "epoch": 8.618903150525087,
      "grad_norm": 0.006011962890625,
      "learning_rate": 9.247982684544854e-06,
      "loss": 0.0001,
      "step": 18466
    },
    {
      "epoch": 8.619369894982498,
      "grad_norm": 0.007049560546875,
      "learning_rate": 9.241820674172486e-06,
      "loss": 0.0002,
      "step": 18467
    },
    {
      "epoch": 8.619836639439907,
      "grad_norm": 0.018798828125,
      "learning_rate": 9.235660617924424e-06,
      "loss": 0.0002,
      "step": 18468
    },
    {
      "epoch": 8.620303383897316,
      "grad_norm": 0.00811767578125,
      "learning_rate": 9.229502515933275e-06,
      "loss": 0.0001,
      "step": 18469
    },
    {
      "epoch": 8.620770128354726,
      "grad_norm": 0.0133056640625,
      "learning_rate": 9.223346368331643e-06,
      "loss": 0.0001,
      "step": 18470
    },
    {
      "epoch": 8.621236872812135,
      "grad_norm": 0.012451171875,
      "learning_rate": 9.21719217525211e-06,
      "loss": 0.0001,
      "step": 18471
    },
    {
      "epoch": 8.621703617269546,
      "grad_norm": 0.01165771484375,
      "learning_rate": 9.211039936827137e-06,
      "loss": 0.0002,
      "step": 18472
    },
    {
      "epoch": 8.622170361726955,
      "grad_norm": 0.00897216796875,
      "learning_rate": 9.204889653189197e-06,
      "loss": 0.0001,
      "step": 18473
    },
    {
      "epoch": 8.622637106184364,
      "grad_norm": 0.02294921875,
      "learning_rate": 9.19874132447075e-06,
      "loss": 0.0005,
      "step": 18474
    },
    {
      "epoch": 8.623103850641774,
      "grad_norm": 0.0166015625,
      "learning_rate": 9.192594950804124e-06,
      "loss": 0.0002,
      "step": 18475
    },
    {
      "epoch": 8.623570595099183,
      "grad_norm": 0.05908203125,
      "learning_rate": 9.1864505323217e-06,
      "loss": 0.0018,
      "step": 18476
    },
    {
      "epoch": 8.624037339556592,
      "grad_norm": 0.0126953125,
      "learning_rate": 9.18030806915574e-06,
      "loss": 0.0002,
      "step": 18477
    },
    {
      "epoch": 8.624504084014003,
      "grad_norm": 0.0167236328125,
      "learning_rate": 9.174167561438519e-06,
      "loss": 0.0002,
      "step": 18478
    },
    {
      "epoch": 8.624970828471412,
      "grad_norm": 0.01507568359375,
      "learning_rate": 9.168029009302248e-06,
      "loss": 0.0002,
      "step": 18479
    },
    {
      "epoch": 8.62543757292882,
      "grad_norm": 0.006072998046875,
      "learning_rate": 9.16189241287908e-06,
      "loss": 0.0001,
      "step": 18480
    },
    {
      "epoch": 8.625904317386231,
      "grad_norm": 0.007293701171875,
      "learning_rate": 9.155757772301177e-06,
      "loss": 0.0001,
      "step": 18481
    },
    {
      "epoch": 8.62637106184364,
      "grad_norm": 0.034423828125,
      "learning_rate": 9.149625087700587e-06,
      "loss": 0.0002,
      "step": 18482
    },
    {
      "epoch": 8.626837806301051,
      "grad_norm": 0.005767822265625,
      "learning_rate": 9.143494359209382e-06,
      "loss": 0.0001,
      "step": 18483
    },
    {
      "epoch": 8.62730455075846,
      "grad_norm": 0.014404296875,
      "learning_rate": 9.137365586959523e-06,
      "loss": 0.0002,
      "step": 18484
    },
    {
      "epoch": 8.627771295215869,
      "grad_norm": 0.005950927734375,
      "learning_rate": 9.131238771083007e-06,
      "loss": 0.0001,
      "step": 18485
    },
    {
      "epoch": 8.62823803967328,
      "grad_norm": 0.00628662109375,
      "learning_rate": 9.125113911711759e-06,
      "loss": 0.0001,
      "step": 18486
    },
    {
      "epoch": 8.628704784130688,
      "grad_norm": 0.01123046875,
      "learning_rate": 9.118991008977618e-06,
      "loss": 0.0001,
      "step": 18487
    },
    {
      "epoch": 8.629171528588097,
      "grad_norm": 0.008544921875,
      "learning_rate": 9.112870063012446e-06,
      "loss": 0.0001,
      "step": 18488
    },
    {
      "epoch": 8.629638273045508,
      "grad_norm": 0.017333984375,
      "learning_rate": 9.106751073948016e-06,
      "loss": 0.0001,
      "step": 18489
    },
    {
      "epoch": 8.630105017502917,
      "grad_norm": 0.0576171875,
      "learning_rate": 9.100634041916078e-06,
      "loss": 0.0003,
      "step": 18490
    },
    {
      "epoch": 8.630571761960327,
      "grad_norm": 0.02197265625,
      "learning_rate": 9.09451896704836e-06,
      "loss": 0.0001,
      "step": 18491
    },
    {
      "epoch": 8.631038506417736,
      "grad_norm": 0.012939453125,
      "learning_rate": 9.08840584947649e-06,
      "loss": 0.0002,
      "step": 18492
    },
    {
      "epoch": 8.631505250875145,
      "grad_norm": 0.0263671875,
      "learning_rate": 9.082294689332116e-06,
      "loss": 0.0002,
      "step": 18493
    },
    {
      "epoch": 8.631971995332556,
      "grad_norm": 0.020263671875,
      "learning_rate": 9.076185486746824e-06,
      "loss": 0.0002,
      "step": 18494
    },
    {
      "epoch": 8.632438739789965,
      "grad_norm": 0.0086669921875,
      "learning_rate": 9.070078241852131e-06,
      "loss": 0.0002,
      "step": 18495
    },
    {
      "epoch": 8.632905484247374,
      "grad_norm": 0.035400390625,
      "learning_rate": 9.063972954779531e-06,
      "loss": 0.0017,
      "step": 18496
    },
    {
      "epoch": 8.633372228704784,
      "grad_norm": 0.007080078125,
      "learning_rate": 9.057869625660509e-06,
      "loss": 0.0001,
      "step": 18497
    },
    {
      "epoch": 8.633838973162193,
      "grad_norm": 0.01190185546875,
      "learning_rate": 9.051768254626447e-06,
      "loss": 0.0001,
      "step": 18498
    },
    {
      "epoch": 8.634305717619604,
      "grad_norm": 0.005828857421875,
      "learning_rate": 9.045668841808719e-06,
      "loss": 0.0001,
      "step": 18499
    },
    {
      "epoch": 8.634772462077013,
      "grad_norm": 0.0108642578125,
      "learning_rate": 9.039571387338663e-06,
      "loss": 0.0001,
      "step": 18500
    },
    {
      "epoch": 8.635239206534422,
      "grad_norm": 0.10693359375,
      "learning_rate": 9.033475891347554e-06,
      "loss": 0.0006,
      "step": 18501
    },
    {
      "epoch": 8.635705950991833,
      "grad_norm": 0.07080078125,
      "learning_rate": 9.027382353966629e-06,
      "loss": 0.0017,
      "step": 18502
    },
    {
      "epoch": 8.636172695449241,
      "grad_norm": 0.0098876953125,
      "learning_rate": 9.02129077532713e-06,
      "loss": 0.0001,
      "step": 18503
    },
    {
      "epoch": 8.63663943990665,
      "grad_norm": 0.11083984375,
      "learning_rate": 9.01520115556015e-06,
      "loss": 0.0003,
      "step": 18504
    },
    {
      "epoch": 8.637106184364061,
      "grad_norm": 0.006500244140625,
      "learning_rate": 9.00911349479684e-06,
      "loss": 0.0002,
      "step": 18505
    },
    {
      "epoch": 8.63757292882147,
      "grad_norm": 0.06396484375,
      "learning_rate": 9.003027793168295e-06,
      "loss": 0.0049,
      "step": 18506
    },
    {
      "epoch": 8.638039673278879,
      "grad_norm": 0.01055908203125,
      "learning_rate": 8.996944050805511e-06,
      "loss": 0.0002,
      "step": 18507
    },
    {
      "epoch": 8.63850641773629,
      "grad_norm": 0.005706787109375,
      "learning_rate": 8.990862267839495e-06,
      "loss": 0.0001,
      "step": 18508
    },
    {
      "epoch": 8.638973162193698,
      "grad_norm": 0.00653076171875,
      "learning_rate": 8.984782444401185e-06,
      "loss": 0.0001,
      "step": 18509
    },
    {
      "epoch": 8.63943990665111,
      "grad_norm": 0.0184326171875,
      "learning_rate": 8.978704580621489e-06,
      "loss": 0.0002,
      "step": 18510
    },
    {
      "epoch": 8.639906651108518,
      "grad_norm": 0.043701171875,
      "learning_rate": 8.97262867663129e-06,
      "loss": 0.0023,
      "step": 18511
    },
    {
      "epoch": 8.640373395565927,
      "grad_norm": 0.0230712890625,
      "learning_rate": 8.966554732561383e-06,
      "loss": 0.0002,
      "step": 18512
    },
    {
      "epoch": 8.640840140023338,
      "grad_norm": 0.0208740234375,
      "learning_rate": 8.960482748542542e-06,
      "loss": 0.0002,
      "step": 18513
    },
    {
      "epoch": 8.641306884480747,
      "grad_norm": 0.04541015625,
      "learning_rate": 8.954412724705541e-06,
      "loss": 0.0016,
      "step": 18514
    },
    {
      "epoch": 8.641773628938157,
      "grad_norm": 0.02099609375,
      "learning_rate": 8.948344661181041e-06,
      "loss": 0.0002,
      "step": 18515
    },
    {
      "epoch": 8.642240373395566,
      "grad_norm": 0.00653076171875,
      "learning_rate": 8.942278558099693e-06,
      "loss": 0.0001,
      "step": 18516
    },
    {
      "epoch": 8.642707117852975,
      "grad_norm": 0.06640625,
      "learning_rate": 8.936214415592137e-06,
      "loss": 0.0035,
      "step": 18517
    },
    {
      "epoch": 8.643173862310386,
      "grad_norm": 0.0045166015625,
      "learning_rate": 8.930152233788901e-06,
      "loss": 0.0001,
      "step": 18518
    },
    {
      "epoch": 8.643640606767795,
      "grad_norm": 0.00445556640625,
      "learning_rate": 8.92409201282054e-06,
      "loss": 0.0001,
      "step": 18519
    },
    {
      "epoch": 8.644107351225204,
      "grad_norm": 0.006988525390625,
      "learning_rate": 8.918033752817546e-06,
      "loss": 0.0001,
      "step": 18520
    },
    {
      "epoch": 8.644574095682614,
      "grad_norm": 0.005859375,
      "learning_rate": 8.911977453910314e-06,
      "loss": 0.0002,
      "step": 18521
    },
    {
      "epoch": 8.645040840140023,
      "grad_norm": 0.00799560546875,
      "learning_rate": 8.905923116229287e-06,
      "loss": 0.0001,
      "step": 18522
    },
    {
      "epoch": 8.645507584597432,
      "grad_norm": 0.031494140625,
      "learning_rate": 8.899870739904815e-06,
      "loss": 0.0002,
      "step": 18523
    },
    {
      "epoch": 8.645974329054843,
      "grad_norm": 0.007293701171875,
      "learning_rate": 8.893820325067182e-06,
      "loss": 0.0002,
      "step": 18524
    },
    {
      "epoch": 8.646441073512252,
      "grad_norm": 0.03271484375,
      "learning_rate": 8.887771871846707e-06,
      "loss": 0.0002,
      "step": 18525
    },
    {
      "epoch": 8.646907817969662,
      "grad_norm": 0.08349609375,
      "learning_rate": 8.881725380373574e-06,
      "loss": 0.0028,
      "step": 18526
    },
    {
      "epoch": 8.647374562427071,
      "grad_norm": 0.068359375,
      "learning_rate": 8.87568085077799e-06,
      "loss": 0.0002,
      "step": 18527
    },
    {
      "epoch": 8.64784130688448,
      "grad_norm": 0.045654296875,
      "learning_rate": 8.869638283190119e-06,
      "loss": 0.002,
      "step": 18528
    },
    {
      "epoch": 8.64830805134189,
      "grad_norm": 0.025634765625,
      "learning_rate": 8.863597677740043e-06,
      "loss": 0.0003,
      "step": 18529
    },
    {
      "epoch": 8.6487747957993,
      "grad_norm": 0.01043701171875,
      "learning_rate": 8.857559034557817e-06,
      "loss": 0.0001,
      "step": 18530
    },
    {
      "epoch": 8.649241540256709,
      "grad_norm": 0.0263671875,
      "learning_rate": 8.851522353773456e-06,
      "loss": 0.0002,
      "step": 18531
    },
    {
      "epoch": 8.64970828471412,
      "grad_norm": 0.0308837890625,
      "learning_rate": 8.84548763551698e-06,
      "loss": 0.0018,
      "step": 18532
    },
    {
      "epoch": 8.650175029171528,
      "grad_norm": 0.01556396484375,
      "learning_rate": 8.839454879918274e-06,
      "loss": 0.0001,
      "step": 18533
    },
    {
      "epoch": 8.650641773628939,
      "grad_norm": 0.01483154296875,
      "learning_rate": 8.833424087107245e-06,
      "loss": 0.0001,
      "step": 18534
    },
    {
      "epoch": 8.651108518086348,
      "grad_norm": 0.04931640625,
      "learning_rate": 8.827395257213766e-06,
      "loss": 0.0023,
      "step": 18535
    },
    {
      "epoch": 8.651575262543757,
      "grad_norm": 0.00762939453125,
      "learning_rate": 8.821368390367601e-06,
      "loss": 0.0001,
      "step": 18536
    },
    {
      "epoch": 8.652042007001167,
      "grad_norm": 0.140625,
      "learning_rate": 8.815343486698568e-06,
      "loss": 0.0003,
      "step": 18537
    },
    {
      "epoch": 8.652508751458576,
      "grad_norm": 0.0125732421875,
      "learning_rate": 8.80932054633633e-06,
      "loss": 0.0001,
      "step": 18538
    },
    {
      "epoch": 8.652975495915985,
      "grad_norm": 0.00823974609375,
      "learning_rate": 8.803299569410606e-06,
      "loss": 0.0002,
      "step": 18539
    },
    {
      "epoch": 8.653442240373396,
      "grad_norm": 0.029296875,
      "learning_rate": 8.797280556051046e-06,
      "loss": 0.0022,
      "step": 18540
    },
    {
      "epoch": 8.653908984830805,
      "grad_norm": 0.0111083984375,
      "learning_rate": 8.791263506387205e-06,
      "loss": 0.0001,
      "step": 18541
    },
    {
      "epoch": 8.654375729288216,
      "grad_norm": 0.08740234375,
      "learning_rate": 8.785248420548665e-06,
      "loss": 0.0031,
      "step": 18542
    },
    {
      "epoch": 8.654842473745624,
      "grad_norm": 0.00555419921875,
      "learning_rate": 8.779235298664945e-06,
      "loss": 0.0001,
      "step": 18543
    },
    {
      "epoch": 8.655309218203033,
      "grad_norm": 0.05126953125,
      "learning_rate": 8.773224140865477e-06,
      "loss": 0.0003,
      "step": 18544
    },
    {
      "epoch": 8.655775962660444,
      "grad_norm": 0.037841796875,
      "learning_rate": 8.767214947279723e-06,
      "loss": 0.0027,
      "step": 18545
    },
    {
      "epoch": 8.656242707117853,
      "grad_norm": 0.0595703125,
      "learning_rate": 8.761207718037067e-06,
      "loss": 0.0002,
      "step": 18546
    },
    {
      "epoch": 8.656709451575262,
      "grad_norm": 0.01123046875,
      "learning_rate": 8.75520245326682e-06,
      "loss": 0.0001,
      "step": 18547
    },
    {
      "epoch": 8.657176196032673,
      "grad_norm": 0.00958251953125,
      "learning_rate": 8.749199153098298e-06,
      "loss": 0.0001,
      "step": 18548
    },
    {
      "epoch": 8.657642940490081,
      "grad_norm": 0.0185546875,
      "learning_rate": 8.743197817660787e-06,
      "loss": 0.0003,
      "step": 18549
    },
    {
      "epoch": 8.65810968494749,
      "grad_norm": 0.00946044921875,
      "learning_rate": 8.73719844708346e-06,
      "loss": 0.0001,
      "step": 18550
    },
    {
      "epoch": 8.658576429404901,
      "grad_norm": 0.01116943359375,
      "learning_rate": 8.731201041495507e-06,
      "loss": 0.0001,
      "step": 18551
    },
    {
      "epoch": 8.65904317386231,
      "grad_norm": 0.0093994140625,
      "learning_rate": 8.725205601026088e-06,
      "loss": 0.0002,
      "step": 18552
    },
    {
      "epoch": 8.65950991831972,
      "grad_norm": 0.00457763671875,
      "learning_rate": 8.719212125804233e-06,
      "loss": 0.0001,
      "step": 18553
    },
    {
      "epoch": 8.65997666277713,
      "grad_norm": 0.0181884765625,
      "learning_rate": 8.713220615959016e-06,
      "loss": 0.0002,
      "step": 18554
    },
    {
      "epoch": 8.660443407234538,
      "grad_norm": 0.00836181640625,
      "learning_rate": 8.70723107161947e-06,
      "loss": 0.0001,
      "step": 18555
    },
    {
      "epoch": 8.660910151691949,
      "grad_norm": 0.0216064453125,
      "learning_rate": 8.701243492914513e-06,
      "loss": 0.0002,
      "step": 18556
    },
    {
      "epoch": 8.661376896149358,
      "grad_norm": 0.00689697265625,
      "learning_rate": 8.695257879973095e-06,
      "loss": 0.0001,
      "step": 18557
    },
    {
      "epoch": 8.661843640606769,
      "grad_norm": 0.00592041015625,
      "learning_rate": 8.689274232924061e-06,
      "loss": 0.0001,
      "step": 18558
    },
    {
      "epoch": 8.662310385064178,
      "grad_norm": 0.00537109375,
      "learning_rate": 8.683292551896272e-06,
      "loss": 0.0001,
      "step": 18559
    },
    {
      "epoch": 8.662777129521587,
      "grad_norm": 0.0177001953125,
      "learning_rate": 8.677312837018514e-06,
      "loss": 0.0002,
      "step": 18560
    },
    {
      "epoch": 8.663243873978997,
      "grad_norm": 0.0169677734375,
      "learning_rate": 8.67133508841953e-06,
      "loss": 0.0005,
      "step": 18561
    },
    {
      "epoch": 8.663710618436406,
      "grad_norm": 0.007232666015625,
      "learning_rate": 8.665359306228028e-06,
      "loss": 0.0001,
      "step": 18562
    },
    {
      "epoch": 8.664177362893815,
      "grad_norm": 0.01251220703125,
      "learning_rate": 8.659385490572702e-06,
      "loss": 0.0001,
      "step": 18563
    },
    {
      "epoch": 8.664644107351226,
      "grad_norm": 0.1240234375,
      "learning_rate": 8.65341364158212e-06,
      "loss": 0.004,
      "step": 18564
    },
    {
      "epoch": 8.665110851808635,
      "grad_norm": 0.01531982421875,
      "learning_rate": 8.647443759384899e-06,
      "loss": 0.0001,
      "step": 18565
    },
    {
      "epoch": 8.665577596266044,
      "grad_norm": 0.00897216796875,
      "learning_rate": 8.641475844109593e-06,
      "loss": 0.0001,
      "step": 18566
    },
    {
      "epoch": 8.666044340723454,
      "grad_norm": 0.005767822265625,
      "learning_rate": 8.635509895884653e-06,
      "loss": 0.0001,
      "step": 18567
    },
    {
      "epoch": 8.666511085180863,
      "grad_norm": 0.01153564453125,
      "learning_rate": 8.629545914838555e-06,
      "loss": 0.0001,
      "step": 18568
    },
    {
      "epoch": 8.666977829638274,
      "grad_norm": 0.01312255859375,
      "learning_rate": 8.623583901099741e-06,
      "loss": 0.0001,
      "step": 18569
    },
    {
      "epoch": 8.667444574095683,
      "grad_norm": 0.006439208984375,
      "learning_rate": 8.617623854796519e-06,
      "loss": 0.0001,
      "step": 18570
    },
    {
      "epoch": 8.667911318553092,
      "grad_norm": 0.0076904296875,
      "learning_rate": 8.611665776057243e-06,
      "loss": 0.0001,
      "step": 18571
    },
    {
      "epoch": 8.668378063010502,
      "grad_norm": 0.01263427734375,
      "learning_rate": 8.605709665010232e-06,
      "loss": 0.0002,
      "step": 18572
    },
    {
      "epoch": 8.668844807467911,
      "grad_norm": 0.04931640625,
      "learning_rate": 8.59975552178367e-06,
      "loss": 0.0025,
      "step": 18573
    },
    {
      "epoch": 8.66931155192532,
      "grad_norm": 0.0113525390625,
      "learning_rate": 8.59380334650579e-06,
      "loss": 0.0001,
      "step": 18574
    },
    {
      "epoch": 8.66977829638273,
      "grad_norm": 0.01519775390625,
      "learning_rate": 8.587853139304735e-06,
      "loss": 0.0001,
      "step": 18575
    },
    {
      "epoch": 8.67024504084014,
      "grad_norm": 0.00970458984375,
      "learning_rate": 8.581904900308624e-06,
      "loss": 0.0001,
      "step": 18576
    },
    {
      "epoch": 8.67071178529755,
      "grad_norm": 0.01239013671875,
      "learning_rate": 8.575958629645554e-06,
      "loss": 0.0001,
      "step": 18577
    },
    {
      "epoch": 8.67117852975496,
      "grad_norm": 0.033203125,
      "learning_rate": 8.570014327443521e-06,
      "loss": 0.0002,
      "step": 18578
    },
    {
      "epoch": 8.671645274212368,
      "grad_norm": 0.03857421875,
      "learning_rate": 8.564071993830503e-06,
      "loss": 0.0025,
      "step": 18579
    },
    {
      "epoch": 8.672112018669779,
      "grad_norm": 0.007049560546875,
      "learning_rate": 8.558131628934474e-06,
      "loss": 0.0001,
      "step": 18580
    },
    {
      "epoch": 8.672578763127188,
      "grad_norm": 0.005462646484375,
      "learning_rate": 8.552193232883344e-06,
      "loss": 0.0001,
      "step": 18581
    },
    {
      "epoch": 8.673045507584597,
      "grad_norm": 0.016845703125,
      "learning_rate": 8.546256805804931e-06,
      "loss": 0.0002,
      "step": 18582
    },
    {
      "epoch": 8.673512252042007,
      "grad_norm": 0.048583984375,
      "learning_rate": 8.54032234782709e-06,
      "loss": 0.0022,
      "step": 18583
    },
    {
      "epoch": 8.673978996499416,
      "grad_norm": 0.0174560546875,
      "learning_rate": 8.534389859077596e-06,
      "loss": 0.0001,
      "step": 18584
    },
    {
      "epoch": 8.674445740956827,
      "grad_norm": 0.005401611328125,
      "learning_rate": 8.528459339684158e-06,
      "loss": 0.0001,
      "step": 18585
    },
    {
      "epoch": 8.674912485414236,
      "grad_norm": 0.0155029296875,
      "learning_rate": 8.522530789774474e-06,
      "loss": 0.0001,
      "step": 18586
    },
    {
      "epoch": 8.675379229871645,
      "grad_norm": 0.015625,
      "learning_rate": 8.51660420947622e-06,
      "loss": 0.0002,
      "step": 18587
    },
    {
      "epoch": 8.675845974329055,
      "grad_norm": 0.00982666015625,
      "learning_rate": 8.51067959891696e-06,
      "loss": 0.0001,
      "step": 18588
    },
    {
      "epoch": 8.676312718786464,
      "grad_norm": 0.00421142578125,
      "learning_rate": 8.504756958224292e-06,
      "loss": 0.0001,
      "step": 18589
    },
    {
      "epoch": 8.676779463243873,
      "grad_norm": 0.0050048828125,
      "learning_rate": 8.498836287525713e-06,
      "loss": 0.0001,
      "step": 18590
    },
    {
      "epoch": 8.677246207701284,
      "grad_norm": 0.00665283203125,
      "learning_rate": 8.492917586948712e-06,
      "loss": 0.0002,
      "step": 18591
    },
    {
      "epoch": 8.677712952158693,
      "grad_norm": 0.060546875,
      "learning_rate": 8.48700085662073e-06,
      "loss": 0.0031,
      "step": 18592
    },
    {
      "epoch": 8.678179696616102,
      "grad_norm": 0.0069580078125,
      "learning_rate": 8.481086096669144e-06,
      "loss": 0.0001,
      "step": 18593
    },
    {
      "epoch": 8.678646441073512,
      "grad_norm": 0.010009765625,
      "learning_rate": 8.475173307221317e-06,
      "loss": 0.0001,
      "step": 18594
    },
    {
      "epoch": 8.679113185530921,
      "grad_norm": 0.0087890625,
      "learning_rate": 8.469262488404572e-06,
      "loss": 0.0001,
      "step": 18595
    },
    {
      "epoch": 8.679579929988332,
      "grad_norm": 0.011962890625,
      "learning_rate": 8.46335364034615e-06,
      "loss": 0.0001,
      "step": 18596
    },
    {
      "epoch": 8.680046674445741,
      "grad_norm": 0.01031494140625,
      "learning_rate": 8.457446763173283e-06,
      "loss": 0.0001,
      "step": 18597
    },
    {
      "epoch": 8.68051341890315,
      "grad_norm": 0.006195068359375,
      "learning_rate": 8.451541857013168e-06,
      "loss": 0.0001,
      "step": 18598
    },
    {
      "epoch": 8.68098016336056,
      "grad_norm": 0.019775390625,
      "learning_rate": 8.445638921992915e-06,
      "loss": 0.0002,
      "step": 18599
    },
    {
      "epoch": 8.68144690781797,
      "grad_norm": 0.005950927734375,
      "learning_rate": 8.439737958239635e-06,
      "loss": 0.0001,
      "step": 18600
    },
    {
      "epoch": 8.68191365227538,
      "grad_norm": 0.010986328125,
      "learning_rate": 8.433838965880414e-06,
      "loss": 0.0001,
      "step": 18601
    },
    {
      "epoch": 8.682380396732789,
      "grad_norm": 0.006561279296875,
      "learning_rate": 8.427941945042206e-06,
      "loss": 0.0001,
      "step": 18602
    },
    {
      "epoch": 8.682847141190198,
      "grad_norm": 0.01458740234375,
      "learning_rate": 8.422046895852009e-06,
      "loss": 0.0002,
      "step": 18603
    },
    {
      "epoch": 8.683313885647609,
      "grad_norm": 0.034912109375,
      "learning_rate": 8.416153818436778e-06,
      "loss": 0.0002,
      "step": 18604
    },
    {
      "epoch": 8.683780630105018,
      "grad_norm": 0.049560546875,
      "learning_rate": 8.410262712923345e-06,
      "loss": 0.0002,
      "step": 18605
    },
    {
      "epoch": 8.684247374562426,
      "grad_norm": 0.005859375,
      "learning_rate": 8.404373579438585e-06,
      "loss": 0.0001,
      "step": 18606
    },
    {
      "epoch": 8.684714119019837,
      "grad_norm": 0.035888671875,
      "learning_rate": 8.398486418109286e-06,
      "loss": 0.002,
      "step": 18607
    },
    {
      "epoch": 8.685180863477246,
      "grad_norm": 0.040771484375,
      "learning_rate": 8.392601229062203e-06,
      "loss": 0.003,
      "step": 18608
    },
    {
      "epoch": 8.685647607934655,
      "grad_norm": 0.0087890625,
      "learning_rate": 8.386718012424066e-06,
      "loss": 0.0001,
      "step": 18609
    },
    {
      "epoch": 8.686114352392066,
      "grad_norm": 0.0279541015625,
      "learning_rate": 8.380836768321531e-06,
      "loss": 0.0005,
      "step": 18610
    },
    {
      "epoch": 8.686581096849475,
      "grad_norm": 0.004669189453125,
      "learning_rate": 8.37495749688123e-06,
      "loss": 0.0001,
      "step": 18611
    },
    {
      "epoch": 8.687047841306885,
      "grad_norm": 0.007781982421875,
      "learning_rate": 8.369080198229762e-06,
      "loss": 0.0001,
      "step": 18612
    },
    {
      "epoch": 8.687514585764294,
      "grad_norm": 0.005340576171875,
      "learning_rate": 8.363204872493657e-06,
      "loss": 0.0001,
      "step": 18613
    },
    {
      "epoch": 8.687981330221703,
      "grad_norm": 0.0086669921875,
      "learning_rate": 8.357331519799415e-06,
      "loss": 0.0001,
      "step": 18614
    },
    {
      "epoch": 8.688448074679114,
      "grad_norm": 0.0308837890625,
      "learning_rate": 8.351460140273537e-06,
      "loss": 0.0003,
      "step": 18615
    },
    {
      "epoch": 8.688914819136523,
      "grad_norm": 0.01458740234375,
      "learning_rate": 8.345590734042375e-06,
      "loss": 0.0001,
      "step": 18616
    },
    {
      "epoch": 8.689381563593932,
      "grad_norm": 0.01092529296875,
      "learning_rate": 8.33972330123235e-06,
      "loss": 0.0001,
      "step": 18617
    },
    {
      "epoch": 8.689848308051342,
      "grad_norm": 0.01177978515625,
      "learning_rate": 8.333857841969783e-06,
      "loss": 0.0001,
      "step": 18618
    },
    {
      "epoch": 8.690315052508751,
      "grad_norm": 0.052978515625,
      "learning_rate": 8.327994356380953e-06,
      "loss": 0.002,
      "step": 18619
    },
    {
      "epoch": 8.690781796966162,
      "grad_norm": 0.005950927734375,
      "learning_rate": 8.322132844592124e-06,
      "loss": 0.0001,
      "step": 18620
    },
    {
      "epoch": 8.69124854142357,
      "grad_norm": 0.0135498046875,
      "learning_rate": 8.316273306729493e-06,
      "loss": 0.0001,
      "step": 18621
    },
    {
      "epoch": 8.69171528588098,
      "grad_norm": 0.021484375,
      "learning_rate": 8.31041574291922e-06,
      "loss": 0.0002,
      "step": 18622
    },
    {
      "epoch": 8.69218203033839,
      "grad_norm": 0.00897216796875,
      "learning_rate": 8.304560153287421e-06,
      "loss": 0.0002,
      "step": 18623
    },
    {
      "epoch": 8.6926487747958,
      "grad_norm": 0.007720947265625,
      "learning_rate": 8.298706537960198e-06,
      "loss": 0.0001,
      "step": 18624
    },
    {
      "epoch": 8.693115519253208,
      "grad_norm": 0.212890625,
      "learning_rate": 8.292854897063574e-06,
      "loss": 0.0004,
      "step": 18625
    },
    {
      "epoch": 8.693582263710619,
      "grad_norm": 0.01043701171875,
      "learning_rate": 8.28700523072351e-06,
      "loss": 0.0002,
      "step": 18626
    },
    {
      "epoch": 8.694049008168028,
      "grad_norm": 0.0186767578125,
      "learning_rate": 8.281157539065998e-06,
      "loss": 0.0002,
      "step": 18627
    },
    {
      "epoch": 8.694515752625438,
      "grad_norm": 0.201171875,
      "learning_rate": 8.275311822216913e-06,
      "loss": 0.0003,
      "step": 18628
    },
    {
      "epoch": 8.694982497082847,
      "grad_norm": 0.010009765625,
      "learning_rate": 8.269468080302134e-06,
      "loss": 0.0001,
      "step": 18629
    },
    {
      "epoch": 8.695449241540256,
      "grad_norm": 0.08642578125,
      "learning_rate": 8.263626313447503e-06,
      "loss": 0.0049,
      "step": 18630
    },
    {
      "epoch": 8.695915985997667,
      "grad_norm": 0.01507568359375,
      "learning_rate": 8.257786521778766e-06,
      "loss": 0.0002,
      "step": 18631
    },
    {
      "epoch": 8.696382730455076,
      "grad_norm": 0.0064697265625,
      "learning_rate": 8.251948705421664e-06,
      "loss": 0.0001,
      "step": 18632
    },
    {
      "epoch": 8.696849474912485,
      "grad_norm": 0.0177001953125,
      "learning_rate": 8.246112864501931e-06,
      "loss": 0.0002,
      "step": 18633
    },
    {
      "epoch": 8.697316219369895,
      "grad_norm": 0.01251220703125,
      "learning_rate": 8.240278999145168e-06,
      "loss": 0.0001,
      "step": 18634
    },
    {
      "epoch": 8.697782963827304,
      "grad_norm": 0.0341796875,
      "learning_rate": 8.234447109477007e-06,
      "loss": 0.0021,
      "step": 18635
    },
    {
      "epoch": 8.698249708284713,
      "grad_norm": 0.019287109375,
      "learning_rate": 8.228617195623034e-06,
      "loss": 0.0002,
      "step": 18636
    },
    {
      "epoch": 8.698716452742124,
      "grad_norm": 0.01007080078125,
      "learning_rate": 8.222789257708741e-06,
      "loss": 0.0001,
      "step": 18637
    },
    {
      "epoch": 8.699183197199533,
      "grad_norm": 0.007293701171875,
      "learning_rate": 8.216963295859637e-06,
      "loss": 0.0001,
      "step": 18638
    },
    {
      "epoch": 8.699649941656943,
      "grad_norm": 0.00921630859375,
      "learning_rate": 8.211139310201133e-06,
      "loss": 0.0001,
      "step": 18639
    },
    {
      "epoch": 8.700116686114352,
      "grad_norm": 0.0081787109375,
      "learning_rate": 8.20531730085864e-06,
      "loss": 0.0001,
      "step": 18640
    },
    {
      "epoch": 8.700583430571761,
      "grad_norm": 0.01373291015625,
      "learning_rate": 8.199497267957523e-06,
      "loss": 0.0002,
      "step": 18641
    },
    {
      "epoch": 8.701050175029172,
      "grad_norm": 0.010009765625,
      "learning_rate": 8.193679211623084e-06,
      "loss": 0.0002,
      "step": 18642
    },
    {
      "epoch": 8.701516919486581,
      "grad_norm": 0.009033203125,
      "learning_rate": 8.187863131980578e-06,
      "loss": 0.0001,
      "step": 18643
    },
    {
      "epoch": 8.701983663943992,
      "grad_norm": 0.031494140625,
      "learning_rate": 8.18204902915527e-06,
      "loss": 0.0002,
      "step": 18644
    },
    {
      "epoch": 8.7024504084014,
      "grad_norm": 0.01025390625,
      "learning_rate": 8.176236903272304e-06,
      "loss": 0.0001,
      "step": 18645
    },
    {
      "epoch": 8.70291715285881,
      "grad_norm": 0.01068115234375,
      "learning_rate": 8.170426754456829e-06,
      "loss": 0.0001,
      "step": 18646
    },
    {
      "epoch": 8.70338389731622,
      "grad_norm": 0.0091552734375,
      "learning_rate": 8.164618582833983e-06,
      "loss": 0.0001,
      "step": 18647
    },
    {
      "epoch": 8.703850641773629,
      "grad_norm": 0.041748046875,
      "learning_rate": 8.15881238852877e-06,
      "loss": 0.0015,
      "step": 18648
    },
    {
      "epoch": 8.704317386231038,
      "grad_norm": 0.00653076171875,
      "learning_rate": 8.153008171666221e-06,
      "loss": 0.0001,
      "step": 18649
    },
    {
      "epoch": 8.704784130688449,
      "grad_norm": 0.00537109375,
      "learning_rate": 8.147205932371326e-06,
      "loss": 0.0001,
      "step": 18650
    },
    {
      "epoch": 8.705250875145857,
      "grad_norm": 0.0244140625,
      "learning_rate": 8.141405670768987e-06,
      "loss": 0.0002,
      "step": 18651
    },
    {
      "epoch": 8.705717619603266,
      "grad_norm": 0.00665283203125,
      "learning_rate": 8.135607386984112e-06,
      "loss": 0.0001,
      "step": 18652
    },
    {
      "epoch": 8.706184364060677,
      "grad_norm": 0.0269775390625,
      "learning_rate": 8.129811081141548e-06,
      "loss": 0.0002,
      "step": 18653
    },
    {
      "epoch": 8.706651108518086,
      "grad_norm": 0.07666015625,
      "learning_rate": 8.12401675336606e-06,
      "loss": 0.0059,
      "step": 18654
    },
    {
      "epoch": 8.707117852975497,
      "grad_norm": 0.0166015625,
      "learning_rate": 8.118224403782448e-06,
      "loss": 0.0002,
      "step": 18655
    },
    {
      "epoch": 8.707584597432906,
      "grad_norm": 0.00836181640625,
      "learning_rate": 8.112434032515393e-06,
      "loss": 0.0001,
      "step": 18656
    },
    {
      "epoch": 8.708051341890314,
      "grad_norm": 0.034912109375,
      "learning_rate": 8.106645639689581e-06,
      "loss": 0.0031,
      "step": 18657
    },
    {
      "epoch": 8.708518086347725,
      "grad_norm": 0.005615234375,
      "learning_rate": 8.100859225429669e-06,
      "loss": 0.0001,
      "step": 18658
    },
    {
      "epoch": 8.708984830805134,
      "grad_norm": 0.01336669921875,
      "learning_rate": 8.095074789860202e-06,
      "loss": 0.0001,
      "step": 18659
    },
    {
      "epoch": 8.709451575262543,
      "grad_norm": 0.03515625,
      "learning_rate": 8.089292333105748e-06,
      "loss": 0.0002,
      "step": 18660
    },
    {
      "epoch": 8.709918319719954,
      "grad_norm": 0.04736328125,
      "learning_rate": 8.083511855290815e-06,
      "loss": 0.0013,
      "step": 18661
    },
    {
      "epoch": 8.710385064177363,
      "grad_norm": 0.0118408203125,
      "learning_rate": 8.077733356539851e-06,
      "loss": 0.0001,
      "step": 18662
    },
    {
      "epoch": 8.710851808634773,
      "grad_norm": 0.015869140625,
      "learning_rate": 8.071956836977268e-06,
      "loss": 0.0002,
      "step": 18663
    },
    {
      "epoch": 8.711318553092182,
      "grad_norm": 0.0169677734375,
      "learning_rate": 8.066182296727464e-06,
      "loss": 0.0002,
      "step": 18664
    },
    {
      "epoch": 8.711785297549591,
      "grad_norm": 0.07421875,
      "learning_rate": 8.060409735914753e-06,
      "loss": 0.0028,
      "step": 18665
    },
    {
      "epoch": 8.712252042007002,
      "grad_norm": 0.007720947265625,
      "learning_rate": 8.054639154663424e-06,
      "loss": 0.0001,
      "step": 18666
    },
    {
      "epoch": 8.71271878646441,
      "grad_norm": 0.007080078125,
      "learning_rate": 8.048870553097742e-06,
      "loss": 0.0001,
      "step": 18667
    },
    {
      "epoch": 8.71318553092182,
      "grad_norm": 0.00677490234375,
      "learning_rate": 8.043103931341878e-06,
      "loss": 0.0001,
      "step": 18668
    },
    {
      "epoch": 8.71365227537923,
      "grad_norm": 0.052490234375,
      "learning_rate": 8.037339289520029e-06,
      "loss": 0.0028,
      "step": 18669
    },
    {
      "epoch": 8.71411901983664,
      "grad_norm": 0.0673828125,
      "learning_rate": 8.031576627756299e-06,
      "loss": 0.0025,
      "step": 18670
    },
    {
      "epoch": 8.71458576429405,
      "grad_norm": 0.053955078125,
      "learning_rate": 8.025815946174764e-06,
      "loss": 0.0033,
      "step": 18671
    },
    {
      "epoch": 8.715052508751459,
      "grad_norm": 0.01043701171875,
      "learning_rate": 8.020057244899448e-06,
      "loss": 0.0001,
      "step": 18672
    },
    {
      "epoch": 8.715519253208868,
      "grad_norm": 0.00836181640625,
      "learning_rate": 8.014300524054386e-06,
      "loss": 0.0001,
      "step": 18673
    },
    {
      "epoch": 8.715985997666278,
      "grad_norm": 0.01263427734375,
      "learning_rate": 8.008545783763467e-06,
      "loss": 0.0002,
      "step": 18674
    },
    {
      "epoch": 8.716452742123687,
      "grad_norm": 0.0068359375,
      "learning_rate": 8.002793024150624e-06,
      "loss": 0.0001,
      "step": 18675
    },
    {
      "epoch": 8.716919486581096,
      "grad_norm": 0.00927734375,
      "learning_rate": 7.997042245339737e-06,
      "loss": 0.0002,
      "step": 18676
    },
    {
      "epoch": 8.717386231038507,
      "grad_norm": 0.005157470703125,
      "learning_rate": 7.991293447454584e-06,
      "loss": 0.0001,
      "step": 18677
    },
    {
      "epoch": 8.717852975495916,
      "grad_norm": 0.0130615234375,
      "learning_rate": 7.985546630618968e-06,
      "loss": 0.0001,
      "step": 18678
    },
    {
      "epoch": 8.718319719953325,
      "grad_norm": 0.0272216796875,
      "learning_rate": 7.979801794956654e-06,
      "loss": 0.0001,
      "step": 18679
    },
    {
      "epoch": 8.718786464410735,
      "grad_norm": 0.05859375,
      "learning_rate": 7.974058940591277e-06,
      "loss": 0.0002,
      "step": 18680
    },
    {
      "epoch": 8.719253208868144,
      "grad_norm": 0.0299072265625,
      "learning_rate": 7.968318067646518e-06,
      "loss": 0.0002,
      "step": 18681
    },
    {
      "epoch": 8.719719953325555,
      "grad_norm": 0.00830078125,
      "learning_rate": 7.96257917624601e-06,
      "loss": 0.0001,
      "step": 18682
    },
    {
      "epoch": 8.720186697782964,
      "grad_norm": 0.006011962890625,
      "learning_rate": 7.956842266513265e-06,
      "loss": 0.0001,
      "step": 18683
    },
    {
      "epoch": 8.720653442240373,
      "grad_norm": 0.0047607421875,
      "learning_rate": 7.951107338571829e-06,
      "loss": 0.0001,
      "step": 18684
    },
    {
      "epoch": 8.721120186697783,
      "grad_norm": 0.007598876953125,
      "learning_rate": 7.94537439254519e-06,
      "loss": 0.0001,
      "step": 18685
    },
    {
      "epoch": 8.721586931155192,
      "grad_norm": 0.0072021484375,
      "learning_rate": 7.939643428556776e-06,
      "loss": 0.0001,
      "step": 18686
    },
    {
      "epoch": 8.722053675612603,
      "grad_norm": 0.0048828125,
      "learning_rate": 7.933914446729984e-06,
      "loss": 0.0001,
      "step": 18687
    },
    {
      "epoch": 8.722520420070012,
      "grad_norm": 0.007781982421875,
      "learning_rate": 7.928187447188152e-06,
      "loss": 0.0002,
      "step": 18688
    },
    {
      "epoch": 8.72298716452742,
      "grad_norm": 0.01312255859375,
      "learning_rate": 7.9224624300546e-06,
      "loss": 0.0001,
      "step": 18689
    },
    {
      "epoch": 8.723453908984832,
      "grad_norm": 0.005157470703125,
      "learning_rate": 7.91673939545261e-06,
      "loss": 0.0001,
      "step": 18690
    },
    {
      "epoch": 8.72392065344224,
      "grad_norm": 0.00836181640625,
      "learning_rate": 7.911018343505372e-06,
      "loss": 0.0001,
      "step": 18691
    },
    {
      "epoch": 8.72438739789965,
      "grad_norm": 0.0693359375,
      "learning_rate": 7.905299274336087e-06,
      "loss": 0.0002,
      "step": 18692
    },
    {
      "epoch": 8.72485414235706,
      "grad_norm": 0.0238037109375,
      "learning_rate": 7.899582188067912e-06,
      "loss": 0.0003,
      "step": 18693
    },
    {
      "epoch": 8.725320886814469,
      "grad_norm": 0.012939453125,
      "learning_rate": 7.893867084823892e-06,
      "loss": 0.0002,
      "step": 18694
    },
    {
      "epoch": 8.725787631271878,
      "grad_norm": 0.00677490234375,
      "learning_rate": 7.888153964727108e-06,
      "loss": 0.0001,
      "step": 18695
    },
    {
      "epoch": 8.726254375729289,
      "grad_norm": 0.00848388671875,
      "learning_rate": 7.882442827900593e-06,
      "loss": 0.0001,
      "step": 18696
    },
    {
      "epoch": 8.726721120186697,
      "grad_norm": 0.005645751953125,
      "learning_rate": 7.876733674467263e-06,
      "loss": 0.0002,
      "step": 18697
    },
    {
      "epoch": 8.727187864644108,
      "grad_norm": 0.0059814453125,
      "learning_rate": 7.871026504550072e-06,
      "loss": 0.0001,
      "step": 18698
    },
    {
      "epoch": 8.727654609101517,
      "grad_norm": 0.005645751953125,
      "learning_rate": 7.865321318271923e-06,
      "loss": 0.0001,
      "step": 18699
    },
    {
      "epoch": 8.728121353558926,
      "grad_norm": 0.006927490234375,
      "learning_rate": 7.859618115755607e-06,
      "loss": 0.0001,
      "step": 18700
    },
    {
      "epoch": 8.728588098016337,
      "grad_norm": 0.0087890625,
      "learning_rate": 7.853916897123948e-06,
      "loss": 0.0001,
      "step": 18701
    },
    {
      "epoch": 8.729054842473746,
      "grad_norm": 0.004608154296875,
      "learning_rate": 7.848217662499713e-06,
      "loss": 0.0001,
      "step": 18702
    },
    {
      "epoch": 8.729521586931154,
      "grad_norm": 0.0096435546875,
      "learning_rate": 7.842520412005583e-06,
      "loss": 0.0002,
      "step": 18703
    },
    {
      "epoch": 8.729988331388565,
      "grad_norm": 0.0269775390625,
      "learning_rate": 7.836825145764247e-06,
      "loss": 0.0002,
      "step": 18704
    },
    {
      "epoch": 8.730455075845974,
      "grad_norm": 0.068359375,
      "learning_rate": 7.831131863898312e-06,
      "loss": 0.0038,
      "step": 18705
    },
    {
      "epoch": 8.730921820303385,
      "grad_norm": 0.03173828125,
      "learning_rate": 7.825440566530373e-06,
      "loss": 0.0002,
      "step": 18706
    },
    {
      "epoch": 8.731388564760794,
      "grad_norm": 0.0169677734375,
      "learning_rate": 7.81975125378298e-06,
      "loss": 0.0002,
      "step": 18707
    },
    {
      "epoch": 8.731855309218203,
      "grad_norm": 0.01312255859375,
      "learning_rate": 7.814063925778603e-06,
      "loss": 0.0002,
      "step": 18708
    },
    {
      "epoch": 8.732322053675613,
      "grad_norm": 0.007415771484375,
      "learning_rate": 7.808378582639708e-06,
      "loss": 0.0002,
      "step": 18709
    },
    {
      "epoch": 8.732788798133022,
      "grad_norm": 0.0166015625,
      "learning_rate": 7.802695224488732e-06,
      "loss": 0.0001,
      "step": 18710
    },
    {
      "epoch": 8.733255542590431,
      "grad_norm": 0.00555419921875,
      "learning_rate": 7.797013851447999e-06,
      "loss": 0.0001,
      "step": 18711
    },
    {
      "epoch": 8.733722287047842,
      "grad_norm": 0.0048828125,
      "learning_rate": 7.791334463639866e-06,
      "loss": 0.0001,
      "step": 18712
    },
    {
      "epoch": 8.73418903150525,
      "grad_norm": 0.05224609375,
      "learning_rate": 7.785657061186624e-06,
      "loss": 0.0039,
      "step": 18713
    },
    {
      "epoch": 8.734655775962661,
      "grad_norm": 0.03271484375,
      "learning_rate": 7.779981644210488e-06,
      "loss": 0.0005,
      "step": 18714
    },
    {
      "epoch": 8.73512252042007,
      "grad_norm": 0.0093994140625,
      "learning_rate": 7.774308212833658e-06,
      "loss": 0.0002,
      "step": 18715
    },
    {
      "epoch": 8.735589264877479,
      "grad_norm": 0.025390625,
      "learning_rate": 7.768636767178317e-06,
      "loss": 0.0002,
      "step": 18716
    },
    {
      "epoch": 8.73605600933489,
      "grad_norm": 0.005828857421875,
      "learning_rate": 7.762967307366542e-06,
      "loss": 0.0001,
      "step": 18717
    },
    {
      "epoch": 8.736522753792299,
      "grad_norm": 0.041259765625,
      "learning_rate": 7.757299833520425e-06,
      "loss": 0.0019,
      "step": 18718
    },
    {
      "epoch": 8.736989498249708,
      "grad_norm": 0.006927490234375,
      "learning_rate": 7.751634345762005e-06,
      "loss": 0.0001,
      "step": 18719
    },
    {
      "epoch": 8.737456242707118,
      "grad_norm": 0.034912109375,
      "learning_rate": 7.745970844213235e-06,
      "loss": 0.0017,
      "step": 18720
    },
    {
      "epoch": 8.737922987164527,
      "grad_norm": 0.0078125,
      "learning_rate": 7.740309328996064e-06,
      "loss": 0.0002,
      "step": 18721
    },
    {
      "epoch": 8.738389731621936,
      "grad_norm": 0.013427734375,
      "learning_rate": 7.734649800232418e-06,
      "loss": 0.0002,
      "step": 18722
    },
    {
      "epoch": 8.738856476079347,
      "grad_norm": 0.041259765625,
      "learning_rate": 7.72899225804411e-06,
      "loss": 0.002,
      "step": 18723
    },
    {
      "epoch": 8.739323220536756,
      "grad_norm": 0.05712890625,
      "learning_rate": 7.723336702552975e-06,
      "loss": 0.0024,
      "step": 18724
    },
    {
      "epoch": 8.739789964994166,
      "grad_norm": 0.03466796875,
      "learning_rate": 7.717683133880804e-06,
      "loss": 0.0019,
      "step": 18725
    },
    {
      "epoch": 8.740256709451575,
      "grad_norm": 0.00982666015625,
      "learning_rate": 7.712031552149279e-06,
      "loss": 0.0001,
      "step": 18726
    },
    {
      "epoch": 8.740723453908984,
      "grad_norm": 0.0203857421875,
      "learning_rate": 7.706381957480114e-06,
      "loss": 0.0003,
      "step": 18727
    },
    {
      "epoch": 8.741190198366395,
      "grad_norm": 0.0810546875,
      "learning_rate": 7.700734349994965e-06,
      "loss": 0.0021,
      "step": 18728
    },
    {
      "epoch": 8.741656942823804,
      "grad_norm": 0.0107421875,
      "learning_rate": 7.695088729815403e-06,
      "loss": 0.0002,
      "step": 18729
    },
    {
      "epoch": 8.742123687281214,
      "grad_norm": 0.037353515625,
      "learning_rate": 7.689445097062985e-06,
      "loss": 0.0022,
      "step": 18730
    },
    {
      "epoch": 8.742590431738623,
      "grad_norm": 0.01177978515625,
      "learning_rate": 7.68380345185925e-06,
      "loss": 0.0002,
      "step": 18731
    },
    {
      "epoch": 8.743057176196032,
      "grad_norm": 0.00628662109375,
      "learning_rate": 7.678163794325644e-06,
      "loss": 0.0001,
      "step": 18732
    },
    {
      "epoch": 8.743523920653443,
      "grad_norm": 0.00885009765625,
      "learning_rate": 7.6725261245836e-06,
      "loss": 0.0001,
      "step": 18733
    },
    {
      "epoch": 8.743990665110852,
      "grad_norm": 0.005950927734375,
      "learning_rate": 7.666890442754527e-06,
      "loss": 0.0001,
      "step": 18734
    },
    {
      "epoch": 8.74445740956826,
      "grad_norm": 0.006744384765625,
      "learning_rate": 7.661256748959722e-06,
      "loss": 0.0001,
      "step": 18735
    },
    {
      "epoch": 8.744924154025671,
      "grad_norm": 0.0203857421875,
      "learning_rate": 7.655625043320536e-06,
      "loss": 0.0002,
      "step": 18736
    },
    {
      "epoch": 8.74539089848308,
      "grad_norm": 0.01104736328125,
      "learning_rate": 7.649995325958181e-06,
      "loss": 0.0002,
      "step": 18737
    },
    {
      "epoch": 8.74585764294049,
      "grad_norm": 0.003814697265625,
      "learning_rate": 7.644367596993896e-06,
      "loss": 0.0001,
      "step": 18738
    },
    {
      "epoch": 8.7463243873979,
      "grad_norm": 0.006927490234375,
      "learning_rate": 7.638741856548858e-06,
      "loss": 0.0002,
      "step": 18739
    },
    {
      "epoch": 8.746791131855309,
      "grad_norm": 0.005767822265625,
      "learning_rate": 7.633118104744175e-06,
      "loss": 0.0001,
      "step": 18740
    },
    {
      "epoch": 8.74725787631272,
      "grad_norm": 0.016357421875,
      "learning_rate": 7.627496341700935e-06,
      "loss": 0.0001,
      "step": 18741
    },
    {
      "epoch": 8.747724620770128,
      "grad_norm": 0.010986328125,
      "learning_rate": 7.62187656754021e-06,
      "loss": 0.0002,
      "step": 18742
    },
    {
      "epoch": 8.748191365227537,
      "grad_norm": 0.0078125,
      "learning_rate": 7.616258782382957e-06,
      "loss": 0.0001,
      "step": 18743
    },
    {
      "epoch": 8.748658109684948,
      "grad_norm": 0.051025390625,
      "learning_rate": 7.610642986350159e-06,
      "loss": 0.0002,
      "step": 18744
    },
    {
      "epoch": 8.749124854142357,
      "grad_norm": 0.014404296875,
      "learning_rate": 7.605029179562728e-06,
      "loss": 0.0002,
      "step": 18745
    },
    {
      "epoch": 8.749591598599766,
      "grad_norm": 0.0146484375,
      "learning_rate": 7.599417362141536e-06,
      "loss": 0.0001,
      "step": 18746
    },
    {
      "epoch": 8.750058343057177,
      "grad_norm": 0.005889892578125,
      "learning_rate": 7.593807534207398e-06,
      "loss": 0.0001,
      "step": 18747
    },
    {
      "epoch": 8.750525087514585,
      "grad_norm": 0.01043701171875,
      "learning_rate": 7.588199695881127e-06,
      "loss": 0.0001,
      "step": 18748
    },
    {
      "epoch": 8.750991831971996,
      "grad_norm": 0.01806640625,
      "learning_rate": 7.582593847283426e-06,
      "loss": 0.0001,
      "step": 18749
    },
    {
      "epoch": 8.751458576429405,
      "grad_norm": 0.0341796875,
      "learning_rate": 7.576989988535033e-06,
      "loss": 0.0015,
      "step": 18750
    },
    {
      "epoch": 8.751925320886814,
      "grad_norm": 0.01312255859375,
      "learning_rate": 7.571388119756595e-06,
      "loss": 0.0002,
      "step": 18751
    },
    {
      "epoch": 8.752392065344225,
      "grad_norm": 0.04541015625,
      "learning_rate": 7.565788241068705e-06,
      "loss": 0.0002,
      "step": 18752
    },
    {
      "epoch": 8.752858809801634,
      "grad_norm": 0.01068115234375,
      "learning_rate": 7.560190352591956e-06,
      "loss": 0.0001,
      "step": 18753
    },
    {
      "epoch": 8.753325554259042,
      "grad_norm": 0.005828857421875,
      "learning_rate": 7.554594454446895e-06,
      "loss": 0.0001,
      "step": 18754
    },
    {
      "epoch": 8.753792298716453,
      "grad_norm": 0.04345703125,
      "learning_rate": 7.549000546753959e-06,
      "loss": 0.003,
      "step": 18755
    },
    {
      "epoch": 8.754259043173862,
      "grad_norm": 0.00762939453125,
      "learning_rate": 7.543408629633641e-06,
      "loss": 0.0001,
      "step": 18756
    },
    {
      "epoch": 8.754725787631273,
      "grad_norm": 0.00921630859375,
      "learning_rate": 7.537818703206301e-06,
      "loss": 0.0001,
      "step": 18757
    },
    {
      "epoch": 8.755192532088682,
      "grad_norm": 0.046630859375,
      "learning_rate": 7.532230767592307e-06,
      "loss": 0.0018,
      "step": 18758
    },
    {
      "epoch": 8.75565927654609,
      "grad_norm": 0.01129150390625,
      "learning_rate": 7.526644822911999e-06,
      "loss": 0.0001,
      "step": 18759
    },
    {
      "epoch": 8.756126021003501,
      "grad_norm": 0.01092529296875,
      "learning_rate": 7.521060869285612e-06,
      "loss": 0.0002,
      "step": 18760
    },
    {
      "epoch": 8.756126021003501,
      "eval_loss": 2.457089900970459,
      "eval_runtime": 86.3037,
      "eval_samples_per_second": 20.903,
      "eval_steps_per_second": 2.619,
      "step": 18760
    },
    {
      "epoch": 8.75659276546091,
      "grad_norm": 0.01708984375,
      "learning_rate": 7.5154789068333955e-06,
      "loss": 0.0002,
      "step": 18761
    },
    {
      "epoch": 8.757059509918319,
      "grad_norm": 0.076171875,
      "learning_rate": 7.509898935675541e-06,
      "loss": 0.0015,
      "step": 18762
    },
    {
      "epoch": 8.75752625437573,
      "grad_norm": 0.00799560546875,
      "learning_rate": 7.504320955932165e-06,
      "loss": 0.0001,
      "step": 18763
    },
    {
      "epoch": 8.757992998833139,
      "grad_norm": 0.006317138671875,
      "learning_rate": 7.4987449677233924e-06,
      "loss": 0.0001,
      "step": 18764
    },
    {
      "epoch": 8.758459743290548,
      "grad_norm": 0.06201171875,
      "learning_rate": 7.493170971169283e-06,
      "loss": 0.0002,
      "step": 18765
    },
    {
      "epoch": 8.758926487747958,
      "grad_norm": 0.05029296875,
      "learning_rate": 7.487598966389831e-06,
      "loss": 0.0021,
      "step": 18766
    },
    {
      "epoch": 8.759393232205367,
      "grad_norm": 0.1005859375,
      "learning_rate": 7.482028953505016e-06,
      "loss": 0.0019,
      "step": 18767
    },
    {
      "epoch": 8.759859976662778,
      "grad_norm": 0.01446533203125,
      "learning_rate": 7.476460932634799e-06,
      "loss": 0.0001,
      "step": 18768
    },
    {
      "epoch": 8.760326721120187,
      "grad_norm": 0.0091552734375,
      "learning_rate": 7.470894903898995e-06,
      "loss": 0.0002,
      "step": 18769
    },
    {
      "epoch": 8.760793465577596,
      "grad_norm": 0.0126953125,
      "learning_rate": 7.465330867417497e-06,
      "loss": 0.0002,
      "step": 18770
    },
    {
      "epoch": 8.761260210035006,
      "grad_norm": 0.03857421875,
      "learning_rate": 7.459768823310098e-06,
      "loss": 0.0003,
      "step": 18771
    },
    {
      "epoch": 8.761726954492415,
      "grad_norm": 0.0079345703125,
      "learning_rate": 7.454208771696547e-06,
      "loss": 0.0001,
      "step": 18772
    },
    {
      "epoch": 8.762193698949824,
      "grad_norm": 0.080078125,
      "learning_rate": 7.448650712696559e-06,
      "loss": 0.0058,
      "step": 18773
    },
    {
      "epoch": 8.762660443407235,
      "grad_norm": 0.043701171875,
      "learning_rate": 7.443094646429827e-06,
      "loss": 0.0027,
      "step": 18774
    },
    {
      "epoch": 8.763127187864644,
      "grad_norm": 0.0069580078125,
      "learning_rate": 7.437540573015945e-06,
      "loss": 0.0001,
      "step": 18775
    },
    {
      "epoch": 8.763593932322054,
      "grad_norm": 0.044189453125,
      "learning_rate": 7.431988492574504e-06,
      "loss": 0.0036,
      "step": 18776
    },
    {
      "epoch": 8.764060676779463,
      "grad_norm": 0.010498046875,
      "learning_rate": 7.426438405225078e-06,
      "loss": 0.0002,
      "step": 18777
    },
    {
      "epoch": 8.764527421236872,
      "grad_norm": 0.0072021484375,
      "learning_rate": 7.4208903110871365e-06,
      "loss": 0.0002,
      "step": 18778
    },
    {
      "epoch": 8.764994165694283,
      "grad_norm": 0.0146484375,
      "learning_rate": 7.415344210280139e-06,
      "loss": 0.0002,
      "step": 18779
    },
    {
      "epoch": 8.765460910151692,
      "grad_norm": 0.01165771484375,
      "learning_rate": 7.4098001029235255e-06,
      "loss": 0.0002,
      "step": 18780
    },
    {
      "epoch": 8.7659276546091,
      "grad_norm": 0.00909423828125,
      "learning_rate": 7.40425798913662e-06,
      "loss": 0.0001,
      "step": 18781
    },
    {
      "epoch": 8.766394399066511,
      "grad_norm": 0.04345703125,
      "learning_rate": 7.398717869038785e-06,
      "loss": 0.0027,
      "step": 18782
    },
    {
      "epoch": 8.76686114352392,
      "grad_norm": 0.006072998046875,
      "learning_rate": 7.393179742749312e-06,
      "loss": 0.0001,
      "step": 18783
    },
    {
      "epoch": 8.767327887981331,
      "grad_norm": 0.08154296875,
      "learning_rate": 7.387643610387418e-06,
      "loss": 0.005,
      "step": 18784
    },
    {
      "epoch": 8.76779463243874,
      "grad_norm": 0.00811767578125,
      "learning_rate": 7.3821094720723185e-06,
      "loss": 0.0001,
      "step": 18785
    },
    {
      "epoch": 8.768261376896149,
      "grad_norm": 0.0498046875,
      "learning_rate": 7.376577327923151e-06,
      "loss": 0.002,
      "step": 18786
    },
    {
      "epoch": 8.76872812135356,
      "grad_norm": 0.01904296875,
      "learning_rate": 7.3710471780590425e-06,
      "loss": 0.0002,
      "step": 18787
    },
    {
      "epoch": 8.769194865810968,
      "grad_norm": 0.0146484375,
      "learning_rate": 7.3655190225990766e-06,
      "loss": 0.0001,
      "step": 18788
    },
    {
      "epoch": 8.769661610268377,
      "grad_norm": 0.007659912109375,
      "learning_rate": 7.359992861662246e-06,
      "loss": 0.0001,
      "step": 18789
    },
    {
      "epoch": 8.770128354725788,
      "grad_norm": 0.05029296875,
      "learning_rate": 7.354468695367567e-06,
      "loss": 0.0033,
      "step": 18790
    },
    {
      "epoch": 8.770595099183197,
      "grad_norm": 0.0113525390625,
      "learning_rate": 7.348946523833977e-06,
      "loss": 0.0002,
      "step": 18791
    },
    {
      "epoch": 8.771061843640608,
      "grad_norm": 0.005157470703125,
      "learning_rate": 7.343426347180349e-06,
      "loss": 0.0001,
      "step": 18792
    },
    {
      "epoch": 8.771528588098016,
      "grad_norm": 0.01190185546875,
      "learning_rate": 7.337908165525553e-06,
      "loss": 0.0001,
      "step": 18793
    },
    {
      "epoch": 8.771995332555425,
      "grad_norm": 0.013916015625,
      "learning_rate": 7.332391978988429e-06,
      "loss": 0.0002,
      "step": 18794
    },
    {
      "epoch": 8.772462077012836,
      "grad_norm": 0.007476806640625,
      "learning_rate": 7.326877787687703e-06,
      "loss": 0.0001,
      "step": 18795
    },
    {
      "epoch": 8.772928821470245,
      "grad_norm": 0.023193359375,
      "learning_rate": 7.321365591742113e-06,
      "loss": 0.0002,
      "step": 18796
    },
    {
      "epoch": 8.773395565927654,
      "grad_norm": 0.039306640625,
      "learning_rate": 7.3158553912703765e-06,
      "loss": 0.0025,
      "step": 18797
    },
    {
      "epoch": 8.773862310385065,
      "grad_norm": 0.009033203125,
      "learning_rate": 7.310347186391087e-06,
      "loss": 0.0001,
      "step": 18798
    },
    {
      "epoch": 8.774329054842473,
      "grad_norm": 0.055419921875,
      "learning_rate": 7.304840977222871e-06,
      "loss": 0.0013,
      "step": 18799
    },
    {
      "epoch": 8.774795799299884,
      "grad_norm": 0.006439208984375,
      "learning_rate": 7.29933676388429e-06,
      "loss": 0.0002,
      "step": 18800
    },
    {
      "epoch": 8.775262543757293,
      "grad_norm": 0.006378173828125,
      "learning_rate": 7.293834546493816e-06,
      "loss": 0.0001,
      "step": 18801
    },
    {
      "epoch": 8.775729288214702,
      "grad_norm": 0.0155029296875,
      "learning_rate": 7.288334325169955e-06,
      "loss": 0.0002,
      "step": 18802
    },
    {
      "epoch": 8.776196032672113,
      "grad_norm": 0.0164794921875,
      "learning_rate": 7.282836100031132e-06,
      "loss": 0.0002,
      "step": 18803
    },
    {
      "epoch": 8.776662777129522,
      "grad_norm": 0.01904296875,
      "learning_rate": 7.2773398711956986e-06,
      "loss": 0.0002,
      "step": 18804
    },
    {
      "epoch": 8.77712952158693,
      "grad_norm": 0.013916015625,
      "learning_rate": 7.271845638782038e-06,
      "loss": 0.0001,
      "step": 18805
    },
    {
      "epoch": 8.777596266044341,
      "grad_norm": 0.01019287109375,
      "learning_rate": 7.26635340290841e-06,
      "loss": 0.0002,
      "step": 18806
    },
    {
      "epoch": 8.77806301050175,
      "grad_norm": 0.033447265625,
      "learning_rate": 7.260863163693077e-06,
      "loss": 0.0002,
      "step": 18807
    },
    {
      "epoch": 8.778529754959159,
      "grad_norm": 0.0283203125,
      "learning_rate": 7.255374921254277e-06,
      "loss": 0.0007,
      "step": 18808
    },
    {
      "epoch": 8.77899649941657,
      "grad_norm": 0.0052490234375,
      "learning_rate": 7.249888675710137e-06,
      "loss": 0.0001,
      "step": 18809
    },
    {
      "epoch": 8.779463243873979,
      "grad_norm": 0.00848388671875,
      "learning_rate": 7.244404427178797e-06,
      "loss": 0.0001,
      "step": 18810
    },
    {
      "epoch": 8.77992998833139,
      "grad_norm": 0.01202392578125,
      "learning_rate": 7.238922175778373e-06,
      "loss": 0.0001,
      "step": 18811
    },
    {
      "epoch": 8.780396732788798,
      "grad_norm": 0.0103759765625,
      "learning_rate": 7.233441921626849e-06,
      "loss": 0.0002,
      "step": 18812
    },
    {
      "epoch": 8.780863477246207,
      "grad_norm": 0.0064697265625,
      "learning_rate": 7.227963664842252e-06,
      "loss": 0.0001,
      "step": 18813
    },
    {
      "epoch": 8.781330221703618,
      "grad_norm": 0.007049560546875,
      "learning_rate": 7.222487405542544e-06,
      "loss": 0.0001,
      "step": 18814
    },
    {
      "epoch": 8.781796966161027,
      "grad_norm": 0.013671875,
      "learning_rate": 7.217013143845597e-06,
      "loss": 0.0002,
      "step": 18815
    },
    {
      "epoch": 8.782263710618436,
      "grad_norm": 0.0194091796875,
      "learning_rate": 7.211540879869328e-06,
      "loss": 0.0002,
      "step": 18816
    },
    {
      "epoch": 8.782730455075846,
      "grad_norm": 0.00567626953125,
      "learning_rate": 7.20607061373153e-06,
      "loss": 0.0001,
      "step": 18817
    },
    {
      "epoch": 8.783197199533255,
      "grad_norm": 0.00579833984375,
      "learning_rate": 7.200602345549967e-06,
      "loss": 0.0001,
      "step": 18818
    },
    {
      "epoch": 8.783663943990666,
      "grad_norm": 0.0205078125,
      "learning_rate": 7.19513607544241e-06,
      "loss": 0.0002,
      "step": 18819
    },
    {
      "epoch": 8.784130688448075,
      "grad_norm": 0.01165771484375,
      "learning_rate": 7.189671803526554e-06,
      "loss": 0.0001,
      "step": 18820
    },
    {
      "epoch": 8.784597432905484,
      "grad_norm": 0.02001953125,
      "learning_rate": 7.184209529920017e-06,
      "loss": 0.0001,
      "step": 18821
    },
    {
      "epoch": 8.785064177362894,
      "grad_norm": 0.03662109375,
      "learning_rate": 7.178749254740435e-06,
      "loss": 0.0011,
      "step": 18822
    },
    {
      "epoch": 8.785530921820303,
      "grad_norm": 0.00848388671875,
      "learning_rate": 7.173290978105385e-06,
      "loss": 0.0001,
      "step": 18823
    },
    {
      "epoch": 8.785997666277712,
      "grad_norm": 0.0079345703125,
      "learning_rate": 7.1678347001323585e-06,
      "loss": 0.0002,
      "step": 18824
    },
    {
      "epoch": 8.786464410735123,
      "grad_norm": 0.042724609375,
      "learning_rate": 7.162380420938841e-06,
      "loss": 0.0016,
      "step": 18825
    },
    {
      "epoch": 8.786931155192532,
      "grad_norm": 0.00689697265625,
      "learning_rate": 7.156928140642305e-06,
      "loss": 0.0001,
      "step": 18826
    },
    {
      "epoch": 8.787397899649942,
      "grad_norm": 0.00927734375,
      "learning_rate": 7.151477859360089e-06,
      "loss": 0.0001,
      "step": 18827
    },
    {
      "epoch": 8.787864644107351,
      "grad_norm": 0.006805419921875,
      "learning_rate": 7.146029577209579e-06,
      "loss": 0.0001,
      "step": 18828
    },
    {
      "epoch": 8.78833138856476,
      "grad_norm": 0.011962890625,
      "learning_rate": 7.140583294308101e-06,
      "loss": 0.0001,
      "step": 18829
    },
    {
      "epoch": 8.788798133022171,
      "grad_norm": 0.0196533203125,
      "learning_rate": 7.135139010772862e-06,
      "loss": 0.0003,
      "step": 18830
    },
    {
      "epoch": 8.78926487747958,
      "grad_norm": 0.052001953125,
      "learning_rate": 7.129696726721125e-06,
      "loss": 0.0022,
      "step": 18831
    },
    {
      "epoch": 8.789731621936989,
      "grad_norm": 0.00823974609375,
      "learning_rate": 7.124256442270072e-06,
      "loss": 0.0001,
      "step": 18832
    },
    {
      "epoch": 8.7901983663944,
      "grad_norm": 0.0069580078125,
      "learning_rate": 7.1188181575368106e-06,
      "loss": 0.0001,
      "step": 18833
    },
    {
      "epoch": 8.790665110851808,
      "grad_norm": 0.052001953125,
      "learning_rate": 7.1133818726384585e-06,
      "loss": 0.0028,
      "step": 18834
    },
    {
      "epoch": 8.791131855309219,
      "grad_norm": 0.004547119140625,
      "learning_rate": 7.107947587692043e-06,
      "loss": 0.0001,
      "step": 18835
    },
    {
      "epoch": 8.791598599766628,
      "grad_norm": 0.005889892578125,
      "learning_rate": 7.102515302814583e-06,
      "loss": 0.0001,
      "step": 18836
    },
    {
      "epoch": 8.792065344224037,
      "grad_norm": 0.005584716796875,
      "learning_rate": 7.097085018123051e-06,
      "loss": 0.0001,
      "step": 18837
    },
    {
      "epoch": 8.792532088681448,
      "grad_norm": 0.01123046875,
      "learning_rate": 7.091656733734342e-06,
      "loss": 0.0002,
      "step": 18838
    },
    {
      "epoch": 8.792998833138856,
      "grad_norm": 0.0084228515625,
      "learning_rate": 7.086230449765352e-06,
      "loss": 0.0001,
      "step": 18839
    },
    {
      "epoch": 8.793465577596265,
      "grad_norm": 0.009521484375,
      "learning_rate": 7.080806166332921e-06,
      "loss": 0.0001,
      "step": 18840
    },
    {
      "epoch": 8.793932322053676,
      "grad_norm": 0.01153564453125,
      "learning_rate": 7.07538388355381e-06,
      "loss": 0.0001,
      "step": 18841
    },
    {
      "epoch": 8.794399066511085,
      "grad_norm": 0.01153564453125,
      "learning_rate": 7.0699636015447934e-06,
      "loss": 0.0001,
      "step": 18842
    },
    {
      "epoch": 8.794865810968496,
      "grad_norm": 0.005950927734375,
      "learning_rate": 7.064545320422578e-06,
      "loss": 0.0001,
      "step": 18843
    },
    {
      "epoch": 8.795332555425905,
      "grad_norm": 0.007568359375,
      "learning_rate": 7.059129040303802e-06,
      "loss": 0.0002,
      "step": 18844
    },
    {
      "epoch": 8.795799299883313,
      "grad_norm": 0.00537109375,
      "learning_rate": 7.053714761305097e-06,
      "loss": 0.0001,
      "step": 18845
    },
    {
      "epoch": 8.796266044340724,
      "grad_norm": 0.0091552734375,
      "learning_rate": 7.048302483543056e-06,
      "loss": 0.0001,
      "step": 18846
    },
    {
      "epoch": 8.796732788798133,
      "grad_norm": 0.0103759765625,
      "learning_rate": 7.042892207134178e-06,
      "loss": 0.0001,
      "step": 18847
    },
    {
      "epoch": 8.797199533255542,
      "grad_norm": 0.005767822265625,
      "learning_rate": 7.037483932194966e-06,
      "loss": 0.0001,
      "step": 18848
    },
    {
      "epoch": 8.797666277712953,
      "grad_norm": 0.00897216796875,
      "learning_rate": 7.032077658841885e-06,
      "loss": 0.0001,
      "step": 18849
    },
    {
      "epoch": 8.798133022170362,
      "grad_norm": 0.0240478515625,
      "learning_rate": 7.026673387191307e-06,
      "loss": 0.0002,
      "step": 18850
    },
    {
      "epoch": 8.79859976662777,
      "grad_norm": 0.00537109375,
      "learning_rate": 7.021271117359607e-06,
      "loss": 0.0001,
      "step": 18851
    },
    {
      "epoch": 8.799066511085181,
      "grad_norm": 0.01239013671875,
      "learning_rate": 7.015870849463113e-06,
      "loss": 0.0002,
      "step": 18852
    },
    {
      "epoch": 8.79953325554259,
      "grad_norm": 0.043701171875,
      "learning_rate": 7.010472583618078e-06,
      "loss": 0.0021,
      "step": 18853
    },
    {
      "epoch": 8.8,
      "grad_norm": 0.00787353515625,
      "learning_rate": 7.0050763199407625e-06,
      "loss": 0.0002,
      "step": 18854
    },
    {
      "epoch": 8.80046674445741,
      "grad_norm": 0.0142822265625,
      "learning_rate": 6.999682058547308e-06,
      "loss": 0.0001,
      "step": 18855
    },
    {
      "epoch": 8.800933488914819,
      "grad_norm": 0.00897216796875,
      "learning_rate": 6.9942897995538884e-06,
      "loss": 0.0001,
      "step": 18856
    },
    {
      "epoch": 8.80140023337223,
      "grad_norm": 0.0240478515625,
      "learning_rate": 6.988899543076611e-06,
      "loss": 0.0002,
      "step": 18857
    },
    {
      "epoch": 8.801866977829638,
      "grad_norm": 0.00836181640625,
      "learning_rate": 6.983511289231514e-06,
      "loss": 0.0002,
      "step": 18858
    },
    {
      "epoch": 8.802333722287047,
      "grad_norm": 0.00665283203125,
      "learning_rate": 6.9781250381346195e-06,
      "loss": 0.0001,
      "step": 18859
    },
    {
      "epoch": 8.802800466744458,
      "grad_norm": 0.04931640625,
      "learning_rate": 6.972740789901921e-06,
      "loss": 0.0002,
      "step": 18860
    },
    {
      "epoch": 8.803267211201867,
      "grad_norm": 0.046875,
      "learning_rate": 6.967358544649316e-06,
      "loss": 0.0034,
      "step": 18861
    },
    {
      "epoch": 8.803733955659277,
      "grad_norm": 0.00726318359375,
      "learning_rate": 6.9619783024926884e-06,
      "loss": 0.0001,
      "step": 18862
    },
    {
      "epoch": 8.804200700116686,
      "grad_norm": 0.018798828125,
      "learning_rate": 6.956600063547925e-06,
      "loss": 0.0001,
      "step": 18863
    },
    {
      "epoch": 8.804667444574095,
      "grad_norm": 0.007537841796875,
      "learning_rate": 6.9512238279307975e-06,
      "loss": 0.0001,
      "step": 18864
    },
    {
      "epoch": 8.805134189031506,
      "grad_norm": 0.0079345703125,
      "learning_rate": 6.945849595757037e-06,
      "loss": 0.0001,
      "step": 18865
    },
    {
      "epoch": 8.805600933488915,
      "grad_norm": 0.0120849609375,
      "learning_rate": 6.940477367142395e-06,
      "loss": 0.0001,
      "step": 18866
    },
    {
      "epoch": 8.806067677946324,
      "grad_norm": 0.0198974609375,
      "learning_rate": 6.935107142202513e-06,
      "loss": 0.0001,
      "step": 18867
    },
    {
      "epoch": 8.806534422403734,
      "grad_norm": 0.00592041015625,
      "learning_rate": 6.929738921053042e-06,
      "loss": 0.0002,
      "step": 18868
    },
    {
      "epoch": 8.807001166861143,
      "grad_norm": 0.014892578125,
      "learning_rate": 6.924372703809556e-06,
      "loss": 0.0002,
      "step": 18869
    },
    {
      "epoch": 8.807467911318554,
      "grad_norm": 0.009765625,
      "learning_rate": 6.919008490587586e-06,
      "loss": 0.0002,
      "step": 18870
    },
    {
      "epoch": 8.807934655775963,
      "grad_norm": 0.01953125,
      "learning_rate": 6.913646281502639e-06,
      "loss": 0.0001,
      "step": 18871
    },
    {
      "epoch": 8.808401400233372,
      "grad_norm": 0.00640869140625,
      "learning_rate": 6.908286076670179e-06,
      "loss": 0.0001,
      "step": 18872
    },
    {
      "epoch": 8.808868144690782,
      "grad_norm": 0.007110595703125,
      "learning_rate": 6.902927876205601e-06,
      "loss": 0.0001,
      "step": 18873
    },
    {
      "epoch": 8.809334889148191,
      "grad_norm": 0.00494384765625,
      "learning_rate": 6.8975716802242685e-06,
      "loss": 0.0001,
      "step": 18874
    },
    {
      "epoch": 8.8098016336056,
      "grad_norm": 0.00506591796875,
      "learning_rate": 6.892217488841535e-06,
      "loss": 0.0001,
      "step": 18875
    },
    {
      "epoch": 8.81026837806301,
      "grad_norm": 0.00897216796875,
      "learning_rate": 6.886865302172651e-06,
      "loss": 0.0001,
      "step": 18876
    },
    {
      "epoch": 8.81073512252042,
      "grad_norm": 0.01092529296875,
      "learning_rate": 6.88151512033286e-06,
      "loss": 0.0001,
      "step": 18877
    },
    {
      "epoch": 8.811201866977829,
      "grad_norm": 0.012451171875,
      "learning_rate": 6.8761669434373896e-06,
      "loss": 0.0001,
      "step": 18878
    },
    {
      "epoch": 8.81166861143524,
      "grad_norm": 0.0308837890625,
      "learning_rate": 6.87082077160135e-06,
      "loss": 0.0018,
      "step": 18879
    },
    {
      "epoch": 8.812135355892648,
      "grad_norm": 0.06689453125,
      "learning_rate": 6.865476604939869e-06,
      "loss": 0.0016,
      "step": 18880
    },
    {
      "epoch": 8.812602100350059,
      "grad_norm": 0.0191650390625,
      "learning_rate": 6.860134443568022e-06,
      "loss": 0.0002,
      "step": 18881
    },
    {
      "epoch": 8.813068844807468,
      "grad_norm": 0.00762939453125,
      "learning_rate": 6.854794287600807e-06,
      "loss": 0.0001,
      "step": 18882
    },
    {
      "epoch": 8.813535589264877,
      "grad_norm": 0.041259765625,
      "learning_rate": 6.8494561371532425e-06,
      "loss": 0.003,
      "step": 18883
    },
    {
      "epoch": 8.814002333722287,
      "grad_norm": 0.058837890625,
      "learning_rate": 6.844119992340214e-06,
      "loss": 0.0002,
      "step": 18884
    },
    {
      "epoch": 8.814469078179696,
      "grad_norm": 0.0113525390625,
      "learning_rate": 6.838785853276642e-06,
      "loss": 0.0001,
      "step": 18885
    },
    {
      "epoch": 8.814935822637107,
      "grad_norm": 0.019775390625,
      "learning_rate": 6.833453720077387e-06,
      "loss": 0.0002,
      "step": 18886
    },
    {
      "epoch": 8.815402567094516,
      "grad_norm": 0.00457763671875,
      "learning_rate": 6.828123592857239e-06,
      "loss": 0.0001,
      "step": 18887
    },
    {
      "epoch": 8.815869311551925,
      "grad_norm": 0.006134033203125,
      "learning_rate": 6.822795471730958e-06,
      "loss": 0.0001,
      "step": 18888
    },
    {
      "epoch": 8.816336056009336,
      "grad_norm": 0.01153564453125,
      "learning_rate": 6.817469356813289e-06,
      "loss": 0.0001,
      "step": 18889
    },
    {
      "epoch": 8.816802800466744,
      "grad_norm": 0.01324462890625,
      "learning_rate": 6.812145248218871e-06,
      "loss": 0.0002,
      "step": 18890
    },
    {
      "epoch": 8.817269544924153,
      "grad_norm": 0.0140380859375,
      "learning_rate": 6.806823146062368e-06,
      "loss": 0.0001,
      "step": 18891
    },
    {
      "epoch": 8.817736289381564,
      "grad_norm": 0.008544921875,
      "learning_rate": 6.801503050458369e-06,
      "loss": 0.0001,
      "step": 18892
    },
    {
      "epoch": 8.818203033838973,
      "grad_norm": 0.039794921875,
      "learning_rate": 6.7961849615214125e-06,
      "loss": 0.0004,
      "step": 18893
    },
    {
      "epoch": 8.818669778296382,
      "grad_norm": 0.01031494140625,
      "learning_rate": 6.790868879365997e-06,
      "loss": 0.0001,
      "step": 18894
    },
    {
      "epoch": 8.819136522753793,
      "grad_norm": 0.0067138671875,
      "learning_rate": 6.785554804106608e-06,
      "loss": 0.0001,
      "step": 18895
    },
    {
      "epoch": 8.819603267211201,
      "grad_norm": 0.004913330078125,
      "learning_rate": 6.780242735857634e-06,
      "loss": 0.0001,
      "step": 18896
    },
    {
      "epoch": 8.820070011668612,
      "grad_norm": 0.0556640625,
      "learning_rate": 6.774932674733458e-06,
      "loss": 0.0002,
      "step": 18897
    },
    {
      "epoch": 8.820536756126021,
      "grad_norm": 0.037109375,
      "learning_rate": 6.769624620848436e-06,
      "loss": 0.0006,
      "step": 18898
    },
    {
      "epoch": 8.82100350058343,
      "grad_norm": 0.0123291015625,
      "learning_rate": 6.7643185743168196e-06,
      "loss": 0.0001,
      "step": 18899
    },
    {
      "epoch": 8.82147024504084,
      "grad_norm": 0.0133056640625,
      "learning_rate": 6.759014535252872e-06,
      "loss": 0.0002,
      "step": 18900
    },
    {
      "epoch": 8.82193698949825,
      "grad_norm": 0.00885009765625,
      "learning_rate": 6.753712503770804e-06,
      "loss": 0.0001,
      "step": 18901
    },
    {
      "epoch": 8.822403733955658,
      "grad_norm": 0.00909423828125,
      "learning_rate": 6.748412479984756e-06,
      "loss": 0.0001,
      "step": 18902
    },
    {
      "epoch": 8.82287047841307,
      "grad_norm": 0.051025390625,
      "learning_rate": 6.743114464008871e-06,
      "loss": 0.0015,
      "step": 18903
    },
    {
      "epoch": 8.823337222870478,
      "grad_norm": 0.01531982421875,
      "learning_rate": 6.737818455957179e-06,
      "loss": 0.0002,
      "step": 18904
    },
    {
      "epoch": 8.823803967327889,
      "grad_norm": 0.0167236328125,
      "learning_rate": 6.732524455943734e-06,
      "loss": 0.0001,
      "step": 18905
    },
    {
      "epoch": 8.824270711785298,
      "grad_norm": 0.005767822265625,
      "learning_rate": 6.727232464082534e-06,
      "loss": 0.0001,
      "step": 18906
    },
    {
      "epoch": 8.824737456242707,
      "grad_norm": 0.00469970703125,
      "learning_rate": 6.721942480487487e-06,
      "loss": 0.0001,
      "step": 18907
    },
    {
      "epoch": 8.825204200700117,
      "grad_norm": 0.006988525390625,
      "learning_rate": 6.7166545052725236e-06,
      "loss": 0.0001,
      "step": 18908
    },
    {
      "epoch": 8.825670945157526,
      "grad_norm": 0.004241943359375,
      "learning_rate": 6.711368538551488e-06,
      "loss": 0.0001,
      "step": 18909
    },
    {
      "epoch": 8.826137689614935,
      "grad_norm": 0.0162353515625,
      "learning_rate": 6.7060845804381875e-06,
      "loss": 0.0001,
      "step": 18910
    },
    {
      "epoch": 8.826604434072346,
      "grad_norm": 0.010498046875,
      "learning_rate": 6.700802631046399e-06,
      "loss": 0.0001,
      "step": 18911
    },
    {
      "epoch": 8.827071178529755,
      "grad_norm": 0.0118408203125,
      "learning_rate": 6.695522690489864e-06,
      "loss": 0.0002,
      "step": 18912
    },
    {
      "epoch": 8.827537922987165,
      "grad_norm": 0.041259765625,
      "learning_rate": 6.690244758882247e-06,
      "loss": 0.0028,
      "step": 18913
    },
    {
      "epoch": 8.828004667444574,
      "grad_norm": 0.007720947265625,
      "learning_rate": 6.684968836337169e-06,
      "loss": 0.0001,
      "step": 18914
    },
    {
      "epoch": 8.828471411901983,
      "grad_norm": 0.0439453125,
      "learning_rate": 6.679694922968271e-06,
      "loss": 0.0018,
      "step": 18915
    },
    {
      "epoch": 8.828938156359394,
      "grad_norm": 0.01239013671875,
      "learning_rate": 6.674423018889054e-06,
      "loss": 0.0002,
      "step": 18916
    },
    {
      "epoch": 8.829404900816803,
      "grad_norm": 0.0712890625,
      "learning_rate": 6.669153124213068e-06,
      "loss": 0.0034,
      "step": 18917
    },
    {
      "epoch": 8.829871645274212,
      "grad_norm": 0.0732421875,
      "learning_rate": 6.66388523905378e-06,
      "loss": 0.0018,
      "step": 18918
    },
    {
      "epoch": 8.830338389731622,
      "grad_norm": 0.007171630859375,
      "learning_rate": 6.658619363524576e-06,
      "loss": 0.0002,
      "step": 18919
    },
    {
      "epoch": 8.830805134189031,
      "grad_norm": 0.00732421875,
      "learning_rate": 6.653355497738867e-06,
      "loss": 0.0001,
      "step": 18920
    },
    {
      "epoch": 8.83127187864644,
      "grad_norm": 0.017822265625,
      "learning_rate": 6.648093641810005e-06,
      "loss": 0.0002,
      "step": 18921
    },
    {
      "epoch": 8.83173862310385,
      "grad_norm": 0.0128173828125,
      "learning_rate": 6.6428337958512444e-06,
      "loss": 0.0002,
      "step": 18922
    },
    {
      "epoch": 8.83220536756126,
      "grad_norm": 0.01019287109375,
      "learning_rate": 6.63757595997585e-06,
      "loss": 0.0001,
      "step": 18923
    },
    {
      "epoch": 8.83267211201867,
      "grad_norm": 0.00531005859375,
      "learning_rate": 6.632320134297054e-06,
      "loss": 0.0001,
      "step": 18924
    },
    {
      "epoch": 8.83313885647608,
      "grad_norm": 0.00927734375,
      "learning_rate": 6.627066318927977e-06,
      "loss": 0.0001,
      "step": 18925
    },
    {
      "epoch": 8.833605600933488,
      "grad_norm": 0.00787353515625,
      "learning_rate": 6.6218145139817615e-06,
      "loss": 0.0001,
      "step": 18926
    },
    {
      "epoch": 8.834072345390899,
      "grad_norm": 0.0252685546875,
      "learning_rate": 6.616564719571505e-06,
      "loss": 0.0003,
      "step": 18927
    },
    {
      "epoch": 8.834539089848308,
      "grad_norm": 0.044921875,
      "learning_rate": 6.611316935810208e-06,
      "loss": 0.0002,
      "step": 18928
    },
    {
      "epoch": 8.835005834305719,
      "grad_norm": 0.010986328125,
      "learning_rate": 6.6060711628108676e-06,
      "loss": 0.0002,
      "step": 18929
    },
    {
      "epoch": 8.835472578763127,
      "grad_norm": 0.053955078125,
      "learning_rate": 6.600827400686449e-06,
      "loss": 0.0017,
      "step": 18930
    },
    {
      "epoch": 8.835939323220536,
      "grad_norm": 0.006805419921875,
      "learning_rate": 6.59558564954984e-06,
      "loss": 0.0001,
      "step": 18931
    },
    {
      "epoch": 8.836406067677947,
      "grad_norm": 0.00628662109375,
      "learning_rate": 6.5903459095138955e-06,
      "loss": 0.0001,
      "step": 18932
    },
    {
      "epoch": 8.836872812135356,
      "grad_norm": 0.01019287109375,
      "learning_rate": 6.585108180691468e-06,
      "loss": 0.0001,
      "step": 18933
    },
    {
      "epoch": 8.837339556592765,
      "grad_norm": 0.005401611328125,
      "learning_rate": 6.57987246319528e-06,
      "loss": 0.0001,
      "step": 18934
    },
    {
      "epoch": 8.837806301050176,
      "grad_norm": 0.0107421875,
      "learning_rate": 6.5746387571381185e-06,
      "loss": 0.0002,
      "step": 18935
    },
    {
      "epoch": 8.838273045507584,
      "grad_norm": 0.01104736328125,
      "learning_rate": 6.569407062632616e-06,
      "loss": 0.0001,
      "step": 18936
    },
    {
      "epoch": 8.838739789964993,
      "grad_norm": 0.0108642578125,
      "learning_rate": 6.564177379791447e-06,
      "loss": 0.0002,
      "step": 18937
    },
    {
      "epoch": 8.839206534422404,
      "grad_norm": 0.007293701171875,
      "learning_rate": 6.558949708727225e-06,
      "loss": 0.0001,
      "step": 18938
    },
    {
      "epoch": 8.839673278879813,
      "grad_norm": 0.0098876953125,
      "learning_rate": 6.553724049552468e-06,
      "loss": 0.0002,
      "step": 18939
    },
    {
      "epoch": 8.840140023337224,
      "grad_norm": 0.006866455078125,
      "learning_rate": 6.54850040237972e-06,
      "loss": 0.0001,
      "step": 18940
    },
    {
      "epoch": 8.840606767794633,
      "grad_norm": 0.0576171875,
      "learning_rate": 6.5432787673214594e-06,
      "loss": 0.0025,
      "step": 18941
    },
    {
      "epoch": 8.841073512252041,
      "grad_norm": 0.00341796875,
      "learning_rate": 6.538059144490083e-06,
      "loss": 0.0001,
      "step": 18942
    },
    {
      "epoch": 8.841540256709452,
      "grad_norm": 0.006103515625,
      "learning_rate": 6.5328415339979795e-06,
      "loss": 0.0001,
      "step": 18943
    },
    {
      "epoch": 8.842007001166861,
      "grad_norm": 0.0220947265625,
      "learning_rate": 6.527625935957526e-06,
      "loss": 0.0001,
      "step": 18944
    },
    {
      "epoch": 8.84247374562427,
      "grad_norm": 0.0908203125,
      "learning_rate": 6.522412350480989e-06,
      "loss": 0.0002,
      "step": 18945
    },
    {
      "epoch": 8.84294049008168,
      "grad_norm": 0.0086669921875,
      "learning_rate": 6.517200777680621e-06,
      "loss": 0.0001,
      "step": 18946
    },
    {
      "epoch": 8.84340723453909,
      "grad_norm": 0.00604248046875,
      "learning_rate": 6.511991217668656e-06,
      "loss": 0.0001,
      "step": 18947
    },
    {
      "epoch": 8.8438739789965,
      "grad_norm": 0.00897216796875,
      "learning_rate": 6.506783670557226e-06,
      "loss": 0.0001,
      "step": 18948
    },
    {
      "epoch": 8.844340723453909,
      "grad_norm": 0.049560546875,
      "learning_rate": 6.501578136458486e-06,
      "loss": 0.0018,
      "step": 18949
    },
    {
      "epoch": 8.844807467911318,
      "grad_norm": 0.0198974609375,
      "learning_rate": 6.496374615484513e-06,
      "loss": 0.0002,
      "step": 18950
    },
    {
      "epoch": 8.845274212368729,
      "grad_norm": 0.01129150390625,
      "learning_rate": 6.4911731077473284e-06,
      "loss": 0.0001,
      "step": 18951
    },
    {
      "epoch": 8.845740956826138,
      "grad_norm": 0.0128173828125,
      "learning_rate": 6.485973613358953e-06,
      "loss": 0.0001,
      "step": 18952
    },
    {
      "epoch": 8.846207701283546,
      "grad_norm": 0.01275634765625,
      "learning_rate": 6.48077613243131e-06,
      "loss": 0.0001,
      "step": 18953
    },
    {
      "epoch": 8.846674445740957,
      "grad_norm": 0.0050048828125,
      "learning_rate": 6.475580665076308e-06,
      "loss": 0.0001,
      "step": 18954
    },
    {
      "epoch": 8.847141190198366,
      "grad_norm": 0.0302734375,
      "learning_rate": 6.4703872114058484e-06,
      "loss": 0.0002,
      "step": 18955
    },
    {
      "epoch": 8.847607934655777,
      "grad_norm": 0.006378173828125,
      "learning_rate": 6.465195771531696e-06,
      "loss": 0.0001,
      "step": 18956
    },
    {
      "epoch": 8.848074679113186,
      "grad_norm": 0.0238037109375,
      "learning_rate": 6.460006345565672e-06,
      "loss": 0.0002,
      "step": 18957
    },
    {
      "epoch": 8.848541423570595,
      "grad_norm": 0.004180908203125,
      "learning_rate": 6.4548189336195e-06,
      "loss": 0.0001,
      "step": 18958
    },
    {
      "epoch": 8.849008168028005,
      "grad_norm": 0.068359375,
      "learning_rate": 6.449633535804866e-06,
      "loss": 0.0043,
      "step": 18959
    },
    {
      "epoch": 8.849474912485414,
      "grad_norm": 0.006622314453125,
      "learning_rate": 6.444450152233416e-06,
      "loss": 0.0001,
      "step": 18960
    },
    {
      "epoch": 8.849941656942823,
      "grad_norm": 0.115234375,
      "learning_rate": 6.4392687830167385e-06,
      "loss": 0.0004,
      "step": 18961
    },
    {
      "epoch": 8.850408401400234,
      "grad_norm": 0.0093994140625,
      "learning_rate": 6.434089428266443e-06,
      "loss": 0.0002,
      "step": 18962
    },
    {
      "epoch": 8.850875145857643,
      "grad_norm": 0.0361328125,
      "learning_rate": 6.428912088093985e-06,
      "loss": 0.0014,
      "step": 18963
    },
    {
      "epoch": 8.851341890315052,
      "grad_norm": 0.0072021484375,
      "learning_rate": 6.423736762610899e-06,
      "loss": 0.0001,
      "step": 18964
    },
    {
      "epoch": 8.851808634772462,
      "grad_norm": 0.004913330078125,
      "learning_rate": 6.418563451928561e-06,
      "loss": 0.0001,
      "step": 18965
    },
    {
      "epoch": 8.852275379229871,
      "grad_norm": 0.044677734375,
      "learning_rate": 6.413392156158382e-06,
      "loss": 0.0005,
      "step": 18966
    },
    {
      "epoch": 8.852742123687282,
      "grad_norm": 0.06396484375,
      "learning_rate": 6.408222875411718e-06,
      "loss": 0.0025,
      "step": 18967
    },
    {
      "epoch": 8.85320886814469,
      "grad_norm": 0.006805419921875,
      "learning_rate": 6.403055609799846e-06,
      "loss": 0.0001,
      "step": 18968
    },
    {
      "epoch": 8.8536756126021,
      "grad_norm": 0.006195068359375,
      "learning_rate": 6.397890359434034e-06,
      "loss": 0.0001,
      "step": 18969
    },
    {
      "epoch": 8.85414235705951,
      "grad_norm": 0.0076904296875,
      "learning_rate": 6.3927271244255124e-06,
      "loss": 0.0001,
      "step": 18970
    },
    {
      "epoch": 8.85460910151692,
      "grad_norm": 0.060302734375,
      "learning_rate": 6.3875659048854286e-06,
      "loss": 0.0019,
      "step": 18971
    },
    {
      "epoch": 8.85507584597433,
      "grad_norm": 0.004669189453125,
      "learning_rate": 6.3824067009249035e-06,
      "loss": 0.0001,
      "step": 18972
    },
    {
      "epoch": 8.855542590431739,
      "grad_norm": 0.01495361328125,
      "learning_rate": 6.377249512655048e-06,
      "loss": 0.0001,
      "step": 18973
    },
    {
      "epoch": 8.856009334889148,
      "grad_norm": 0.01287841796875,
      "learning_rate": 6.372094340186874e-06,
      "loss": 0.0001,
      "step": 18974
    },
    {
      "epoch": 8.856476079346558,
      "grad_norm": 0.004547119140625,
      "learning_rate": 6.366941183631392e-06,
      "loss": 0.0001,
      "step": 18975
    },
    {
      "epoch": 8.856942823803967,
      "grad_norm": 0.0228271484375,
      "learning_rate": 6.361790043099569e-06,
      "loss": 0.0001,
      "step": 18976
    },
    {
      "epoch": 8.857409568261376,
      "grad_norm": 0.01513671875,
      "learning_rate": 6.356640918702283e-06,
      "loss": 0.0002,
      "step": 18977
    },
    {
      "epoch": 8.857876312718787,
      "grad_norm": 0.0186767578125,
      "learning_rate": 6.351493810550413e-06,
      "loss": 0.0002,
      "step": 18978
    },
    {
      "epoch": 8.858343057176196,
      "grad_norm": 0.011962890625,
      "learning_rate": 6.3463487187548135e-06,
      "loss": 0.0001,
      "step": 18979
    },
    {
      "epoch": 8.858809801633605,
      "grad_norm": 0.007049560546875,
      "learning_rate": 6.3412056434262064e-06,
      "loss": 0.0001,
      "step": 18980
    },
    {
      "epoch": 8.859276546091015,
      "grad_norm": 0.005401611328125,
      "learning_rate": 6.336064584675372e-06,
      "loss": 0.0001,
      "step": 18981
    },
    {
      "epoch": 8.859743290548424,
      "grad_norm": 0.00714111328125,
      "learning_rate": 6.330925542612997e-06,
      "loss": 0.0001,
      "step": 18982
    },
    {
      "epoch": 8.860210035005835,
      "grad_norm": 0.048583984375,
      "learning_rate": 6.325788517349706e-06,
      "loss": 0.0008,
      "step": 18983
    },
    {
      "epoch": 8.860676779463244,
      "grad_norm": 0.0296630859375,
      "learning_rate": 6.320653508996144e-06,
      "loss": 0.0002,
      "step": 18984
    },
    {
      "epoch": 8.861143523920653,
      "grad_norm": 0.007415771484375,
      "learning_rate": 6.315520517662832e-06,
      "loss": 0.0001,
      "step": 18985
    },
    {
      "epoch": 8.861610268378064,
      "grad_norm": 0.0177001953125,
      "learning_rate": 6.3103895434603044e-06,
      "loss": 0.0002,
      "step": 18986
    },
    {
      "epoch": 8.862077012835472,
      "grad_norm": 0.043212890625,
      "learning_rate": 6.305260586499062e-06,
      "loss": 0.0022,
      "step": 18987
    },
    {
      "epoch": 8.862543757292881,
      "grad_norm": 0.004913330078125,
      "learning_rate": 6.300133646889494e-06,
      "loss": 0.0001,
      "step": 18988
    },
    {
      "epoch": 8.863010501750292,
      "grad_norm": 0.01141357421875,
      "learning_rate": 6.295008724742013e-06,
      "loss": 0.0001,
      "step": 18989
    },
    {
      "epoch": 8.863477246207701,
      "grad_norm": 0.00799560546875,
      "learning_rate": 6.289885820166974e-06,
      "loss": 0.0001,
      "step": 18990
    },
    {
      "epoch": 8.863943990665112,
      "grad_norm": 0.0172119140625,
      "learning_rate": 6.2847649332746565e-06,
      "loss": 0.0002,
      "step": 18991
    },
    {
      "epoch": 8.86441073512252,
      "grad_norm": 0.053955078125,
      "learning_rate": 6.279646064175327e-06,
      "loss": 0.0022,
      "step": 18992
    },
    {
      "epoch": 8.86487747957993,
      "grad_norm": 0.033935546875,
      "learning_rate": 6.274529212979208e-06,
      "loss": 0.0009,
      "step": 18993
    },
    {
      "epoch": 8.86534422403734,
      "grad_norm": 0.1162109375,
      "learning_rate": 6.269414379796468e-06,
      "loss": 0.0034,
      "step": 18994
    },
    {
      "epoch": 8.865810968494749,
      "grad_norm": 0.005401611328125,
      "learning_rate": 6.264301564737218e-06,
      "loss": 0.0001,
      "step": 18995
    },
    {
      "epoch": 8.866277712952158,
      "grad_norm": 0.01483154296875,
      "learning_rate": 6.259190767911571e-06,
      "loss": 0.0002,
      "step": 18996
    },
    {
      "epoch": 8.866744457409569,
      "grad_norm": 0.033935546875,
      "learning_rate": 6.25408198942955e-06,
      "loss": 0.0006,
      "step": 18997
    },
    {
      "epoch": 8.867211201866978,
      "grad_norm": 0.006011962890625,
      "learning_rate": 6.248975229401155e-06,
      "loss": 0.0001,
      "step": 18998
    },
    {
      "epoch": 8.867677946324388,
      "grad_norm": 0.00494384765625,
      "learning_rate": 6.243870487936354e-06,
      "loss": 0.0001,
      "step": 18999
    },
    {
      "epoch": 8.868144690781797,
      "grad_norm": 0.01116943359375,
      "learning_rate": 6.238767765145026e-06,
      "loss": 0.0001,
      "step": 19000
    },
    {
      "epoch": 8.868611435239206,
      "grad_norm": 0.0257568359375,
      "learning_rate": 6.233667061137072e-06,
      "loss": 0.0002,
      "step": 19001
    },
    {
      "epoch": 8.869078179696617,
      "grad_norm": 0.00518798828125,
      "learning_rate": 6.228568376022293e-06,
      "loss": 0.0001,
      "step": 19002
    },
    {
      "epoch": 8.869544924154026,
      "grad_norm": 0.00701904296875,
      "learning_rate": 6.223471709910478e-06,
      "loss": 0.0001,
      "step": 19003
    },
    {
      "epoch": 8.870011668611435,
      "grad_norm": 0.040283203125,
      "learning_rate": 6.218377062911373e-06,
      "loss": 0.0002,
      "step": 19004
    },
    {
      "epoch": 8.870478413068845,
      "grad_norm": 0.0133056640625,
      "learning_rate": 6.213284435134659e-06,
      "loss": 0.0001,
      "step": 19005
    },
    {
      "epoch": 8.870945157526254,
      "grad_norm": 0.01318359375,
      "learning_rate": 6.208193826689979e-06,
      "loss": 0.0002,
      "step": 19006
    },
    {
      "epoch": 8.871411901983663,
      "grad_norm": 0.0245361328125,
      "learning_rate": 6.203105237686968e-06,
      "loss": 0.0001,
      "step": 19007
    },
    {
      "epoch": 8.871878646441074,
      "grad_norm": 0.006927490234375,
      "learning_rate": 6.198018668235173e-06,
      "loss": 0.0001,
      "step": 19008
    },
    {
      "epoch": 8.872345390898483,
      "grad_norm": 0.03271484375,
      "learning_rate": 6.192934118444093e-06,
      "loss": 0.0021,
      "step": 19009
    },
    {
      "epoch": 8.872812135355893,
      "grad_norm": 0.0205078125,
      "learning_rate": 6.1878515884232215e-06,
      "loss": 0.0002,
      "step": 19010
    },
    {
      "epoch": 8.873278879813302,
      "grad_norm": 0.0120849609375,
      "learning_rate": 6.182771078282013e-06,
      "loss": 0.0001,
      "step": 19011
    },
    {
      "epoch": 8.873745624270711,
      "grad_norm": 0.0201416015625,
      "learning_rate": 6.177692588129813e-06,
      "loss": 0.0002,
      "step": 19012
    },
    {
      "epoch": 8.874212368728122,
      "grad_norm": 0.06640625,
      "learning_rate": 6.172616118076002e-06,
      "loss": 0.0003,
      "step": 19013
    },
    {
      "epoch": 8.87467911318553,
      "grad_norm": 0.006378173828125,
      "learning_rate": 6.167541668229859e-06,
      "loss": 0.0001,
      "step": 19014
    },
    {
      "epoch": 8.875145857642941,
      "grad_norm": 0.006927490234375,
      "learning_rate": 6.162469238700652e-06,
      "loss": 0.0001,
      "step": 19015
    },
    {
      "epoch": 8.87561260210035,
      "grad_norm": 0.00531005859375,
      "learning_rate": 6.157398829597616e-06,
      "loss": 0.0001,
      "step": 19016
    },
    {
      "epoch": 8.87607934655776,
      "grad_norm": 0.006927490234375,
      "learning_rate": 6.152330441029885e-06,
      "loss": 0.0001,
      "step": 19017
    },
    {
      "epoch": 8.87654609101517,
      "grad_norm": 0.005126953125,
      "learning_rate": 6.147264073106607e-06,
      "loss": 0.0001,
      "step": 19018
    },
    {
      "epoch": 8.877012835472579,
      "grad_norm": 0.0400390625,
      "learning_rate": 6.14219972593687e-06,
      "loss": 0.0016,
      "step": 19019
    },
    {
      "epoch": 8.877479579929988,
      "grad_norm": 0.0093994140625,
      "learning_rate": 6.137137399629711e-06,
      "loss": 0.0001,
      "step": 19020
    },
    {
      "epoch": 8.877946324387398,
      "grad_norm": 0.01055908203125,
      "learning_rate": 6.13207709429412e-06,
      "loss": 0.0002,
      "step": 19021
    },
    {
      "epoch": 8.878413068844807,
      "grad_norm": 0.005279541015625,
      "learning_rate": 6.1270188100390645e-06,
      "loss": 0.0001,
      "step": 19022
    },
    {
      "epoch": 8.878879813302216,
      "grad_norm": 0.01080322265625,
      "learning_rate": 6.121962546973437e-06,
      "loss": 0.0002,
      "step": 19023
    },
    {
      "epoch": 8.879346557759627,
      "grad_norm": 0.006378173828125,
      "learning_rate": 6.116908305206126e-06,
      "loss": 0.0001,
      "step": 19024
    },
    {
      "epoch": 8.879813302217036,
      "grad_norm": 0.008056640625,
      "learning_rate": 6.1118560848459464e-06,
      "loss": 0.0001,
      "step": 19025
    },
    {
      "epoch": 8.880280046674446,
      "grad_norm": 0.044677734375,
      "learning_rate": 6.106805886001665e-06,
      "loss": 0.0026,
      "step": 19026
    },
    {
      "epoch": 8.880746791131855,
      "grad_norm": 0.049560546875,
      "learning_rate": 6.10175770878203e-06,
      "loss": 0.0002,
      "step": 19027
    },
    {
      "epoch": 8.881213535589264,
      "grad_norm": 0.00433349609375,
      "learning_rate": 6.096711553295753e-06,
      "loss": 0.0001,
      "step": 19028
    },
    {
      "epoch": 8.881680280046675,
      "grad_norm": 0.01409912109375,
      "learning_rate": 6.091667419651437e-06,
      "loss": 0.0002,
      "step": 19029
    },
    {
      "epoch": 8.882147024504084,
      "grad_norm": 0.00799560546875,
      "learning_rate": 6.086625307957727e-06,
      "loss": 0.0001,
      "step": 19030
    },
    {
      "epoch": 8.882613768961493,
      "grad_norm": 0.006378173828125,
      "learning_rate": 6.081585218323183e-06,
      "loss": 0.0001,
      "step": 19031
    },
    {
      "epoch": 8.883080513418903,
      "grad_norm": 0.0162353515625,
      "learning_rate": 6.076547150856293e-06,
      "loss": 0.0002,
      "step": 19032
    },
    {
      "epoch": 8.883547257876312,
      "grad_norm": 0.0301513671875,
      "learning_rate": 6.071511105665572e-06,
      "loss": 0.0023,
      "step": 19033
    },
    {
      "epoch": 8.884014002333723,
      "grad_norm": 0.00726318359375,
      "learning_rate": 6.066477082859412e-06,
      "loss": 0.0002,
      "step": 19034
    },
    {
      "epoch": 8.884480746791132,
      "grad_norm": 0.00482177734375,
      "learning_rate": 6.061445082546235e-06,
      "loss": 0.0001,
      "step": 19035
    },
    {
      "epoch": 8.88494749124854,
      "grad_norm": 0.0059814453125,
      "learning_rate": 6.056415104834379e-06,
      "loss": 0.0001,
      "step": 19036
    },
    {
      "epoch": 8.885414235705952,
      "grad_norm": 0.00921630859375,
      "learning_rate": 6.051387149832122e-06,
      "loss": 0.0001,
      "step": 19037
    },
    {
      "epoch": 8.88588098016336,
      "grad_norm": 0.00970458984375,
      "learning_rate": 6.046361217647733e-06,
      "loss": 0.0001,
      "step": 19038
    },
    {
      "epoch": 8.88634772462077,
      "grad_norm": 0.008544921875,
      "learning_rate": 6.041337308389461e-06,
      "loss": 0.0001,
      "step": 19039
    },
    {
      "epoch": 8.88681446907818,
      "grad_norm": 0.033203125,
      "learning_rate": 6.0363154221654174e-06,
      "loss": 0.0002,
      "step": 19040
    },
    {
      "epoch": 8.887281213535589,
      "grad_norm": 0.0069580078125,
      "learning_rate": 6.03129555908376e-06,
      "loss": 0.0001,
      "step": 19041
    },
    {
      "epoch": 8.887747957993,
      "grad_norm": 0.0185546875,
      "learning_rate": 6.026277719252582e-06,
      "loss": 0.0001,
      "step": 19042
    },
    {
      "epoch": 8.888214702450409,
      "grad_norm": 0.0067138671875,
      "learning_rate": 6.021261902779896e-06,
      "loss": 0.0002,
      "step": 19043
    },
    {
      "epoch": 8.888681446907817,
      "grad_norm": 0.00921630859375,
      "learning_rate": 6.016248109773715e-06,
      "loss": 0.0002,
      "step": 19044
    },
    {
      "epoch": 8.889148191365228,
      "grad_norm": 0.007232666015625,
      "learning_rate": 6.011236340342008e-06,
      "loss": 0.0001,
      "step": 19045
    },
    {
      "epoch": 8.889614935822637,
      "grad_norm": 0.0654296875,
      "learning_rate": 6.006226594592645e-06,
      "loss": 0.0012,
      "step": 19046
    },
    {
      "epoch": 8.890081680280046,
      "grad_norm": 0.0078125,
      "learning_rate": 6.001218872633507e-06,
      "loss": 0.0001,
      "step": 19047
    },
    {
      "epoch": 8.890548424737457,
      "grad_norm": 0.00836181640625,
      "learning_rate": 5.99621317457244e-06,
      "loss": 0.0001,
      "step": 19048
    },
    {
      "epoch": 8.891015169194866,
      "grad_norm": 0.034912109375,
      "learning_rate": 5.991209500517192e-06,
      "loss": 0.0018,
      "step": 19049
    },
    {
      "epoch": 8.891481913652274,
      "grad_norm": 0.045654296875,
      "learning_rate": 5.986207850575521e-06,
      "loss": 0.0048,
      "step": 19050
    },
    {
      "epoch": 8.891948658109685,
      "grad_norm": 0.006744384765625,
      "learning_rate": 5.981208224855084e-06,
      "loss": 0.0001,
      "step": 19051
    },
    {
      "epoch": 8.892415402567094,
      "grad_norm": 0.00543212890625,
      "learning_rate": 5.976210623463552e-06,
      "loss": 0.0001,
      "step": 19052
    },
    {
      "epoch": 8.892882147024505,
      "grad_norm": 0.00469970703125,
      "learning_rate": 5.9712150465085385e-06,
      "loss": 0.0001,
      "step": 19053
    },
    {
      "epoch": 8.893348891481914,
      "grad_norm": 0.0286865234375,
      "learning_rate": 5.96622149409759e-06,
      "loss": 0.0002,
      "step": 19054
    },
    {
      "epoch": 8.893815635939323,
      "grad_norm": 0.0120849609375,
      "learning_rate": 5.961229966338222e-06,
      "loss": 0.0002,
      "step": 19055
    },
    {
      "epoch": 8.894282380396733,
      "grad_norm": 0.0079345703125,
      "learning_rate": 5.956240463337903e-06,
      "loss": 0.0002,
      "step": 19056
    },
    {
      "epoch": 8.894749124854142,
      "grad_norm": 0.004730224609375,
      "learning_rate": 5.951252985204092e-06,
      "loss": 0.0001,
      "step": 19057
    },
    {
      "epoch": 8.895215869311553,
      "grad_norm": 0.07373046875,
      "learning_rate": 5.946267532044136e-06,
      "loss": 0.002,
      "step": 19058
    },
    {
      "epoch": 8.895682613768962,
      "grad_norm": 0.0390625,
      "learning_rate": 5.941284103965384e-06,
      "loss": 0.0029,
      "step": 19059
    },
    {
      "epoch": 8.89614935822637,
      "grad_norm": 0.01324462890625,
      "learning_rate": 5.93630270107517e-06,
      "loss": 0.0002,
      "step": 19060
    },
    {
      "epoch": 8.896616102683781,
      "grad_norm": 0.0576171875,
      "learning_rate": 5.9313233234807084e-06,
      "loss": 0.0016,
      "step": 19061
    },
    {
      "epoch": 8.89708284714119,
      "grad_norm": 0.01043701171875,
      "learning_rate": 5.9263459712892375e-06,
      "loss": 0.0001,
      "step": 19062
    },
    {
      "epoch": 8.8975495915986,
      "grad_norm": 0.0086669921875,
      "learning_rate": 5.921370644607904e-06,
      "loss": 0.0001,
      "step": 19063
    },
    {
      "epoch": 8.89801633605601,
      "grad_norm": 0.005828857421875,
      "learning_rate": 5.916397343543833e-06,
      "loss": 0.0001,
      "step": 19064
    },
    {
      "epoch": 8.898483080513419,
      "grad_norm": 0.01385498046875,
      "learning_rate": 5.91142606820414e-06,
      "loss": 0.0002,
      "step": 19065
    },
    {
      "epoch": 8.898949824970828,
      "grad_norm": 0.01226806640625,
      "learning_rate": 5.9064568186958265e-06,
      "loss": 0.0001,
      "step": 19066
    },
    {
      "epoch": 8.899416569428238,
      "grad_norm": 0.009521484375,
      "learning_rate": 5.901489595125887e-06,
      "loss": 0.0001,
      "step": 19067
    },
    {
      "epoch": 8.899883313885647,
      "grad_norm": 0.01080322265625,
      "learning_rate": 5.8965243976013e-06,
      "loss": 0.0002,
      "step": 19068
    },
    {
      "epoch": 8.900350058343058,
      "grad_norm": 0.060302734375,
      "learning_rate": 5.891561226228925e-06,
      "loss": 0.0003,
      "step": 19069
    },
    {
      "epoch": 8.900816802800467,
      "grad_norm": 0.01416015625,
      "learning_rate": 5.886600081115668e-06,
      "loss": 0.0001,
      "step": 19070
    },
    {
      "epoch": 8.901283547257876,
      "grad_norm": 0.01116943359375,
      "learning_rate": 5.88164096236834e-06,
      "loss": 0.0002,
      "step": 19071
    },
    {
      "epoch": 8.901750291715286,
      "grad_norm": 0.01263427734375,
      "learning_rate": 5.876683870093691e-06,
      "loss": 0.0002,
      "step": 19072
    },
    {
      "epoch": 8.902217036172695,
      "grad_norm": 0.0791015625,
      "learning_rate": 5.871728804398469e-06,
      "loss": 0.004,
      "step": 19073
    },
    {
      "epoch": 8.902683780630104,
      "grad_norm": 0.032958984375,
      "learning_rate": 5.866775765389387e-06,
      "loss": 0.0002,
      "step": 19074
    },
    {
      "epoch": 8.903150525087515,
      "grad_norm": 0.006256103515625,
      "learning_rate": 5.861824753173039e-06,
      "loss": 0.0001,
      "step": 19075
    },
    {
      "epoch": 8.903617269544924,
      "grad_norm": 0.00872802734375,
      "learning_rate": 5.856875767856063e-06,
      "loss": 0.0002,
      "step": 19076
    },
    {
      "epoch": 8.904084014002335,
      "grad_norm": 0.01385498046875,
      "learning_rate": 5.851928809545015e-06,
      "loss": 0.0002,
      "step": 19077
    },
    {
      "epoch": 8.904550758459743,
      "grad_norm": 0.005767822265625,
      "learning_rate": 5.846983878346379e-06,
      "loss": 0.0001,
      "step": 19078
    },
    {
      "epoch": 8.905017502917152,
      "grad_norm": 0.00970458984375,
      "learning_rate": 5.8420409743666584e-06,
      "loss": 0.0001,
      "step": 19079
    },
    {
      "epoch": 8.905484247374563,
      "grad_norm": 0.059326171875,
      "learning_rate": 5.837100097712267e-06,
      "loss": 0.0005,
      "step": 19080
    },
    {
      "epoch": 8.905950991831972,
      "grad_norm": 0.040771484375,
      "learning_rate": 5.8321612484895764e-06,
      "loss": 0.0019,
      "step": 19081
    },
    {
      "epoch": 8.90641773628938,
      "grad_norm": 0.0206298828125,
      "learning_rate": 5.827224426804956e-06,
      "loss": 0.0002,
      "step": 19082
    },
    {
      "epoch": 8.906884480746792,
      "grad_norm": 0.01300048828125,
      "learning_rate": 5.8222896327646546e-06,
      "loss": 0.0002,
      "step": 19083
    },
    {
      "epoch": 8.9073512252042,
      "grad_norm": 0.0791015625,
      "learning_rate": 5.817356866474955e-06,
      "loss": 0.0041,
      "step": 19084
    },
    {
      "epoch": 8.907817969661611,
      "grad_norm": 0.0087890625,
      "learning_rate": 5.81242612804207e-06,
      "loss": 0.0002,
      "step": 19085
    },
    {
      "epoch": 8.90828471411902,
      "grad_norm": 0.0361328125,
      "learning_rate": 5.807497417572139e-06,
      "loss": 0.0002,
      "step": 19086
    },
    {
      "epoch": 8.908751458576429,
      "grad_norm": 0.048828125,
      "learning_rate": 5.8025707351713086e-06,
      "loss": 0.0031,
      "step": 19087
    },
    {
      "epoch": 8.90921820303384,
      "grad_norm": 0.0263671875,
      "learning_rate": 5.79764608094564e-06,
      "loss": 0.0002,
      "step": 19088
    },
    {
      "epoch": 8.909684947491249,
      "grad_norm": 0.0126953125,
      "learning_rate": 5.79272345500117e-06,
      "loss": 0.0001,
      "step": 19089
    },
    {
      "epoch": 8.910151691948657,
      "grad_norm": 0.068359375,
      "learning_rate": 5.78780285744388e-06,
      "loss": 0.0051,
      "step": 19090
    },
    {
      "epoch": 8.910618436406068,
      "grad_norm": 0.004180908203125,
      "learning_rate": 5.782884288379753e-06,
      "loss": 0.0001,
      "step": 19091
    },
    {
      "epoch": 8.911085180863477,
      "grad_norm": 0.01446533203125,
      "learning_rate": 5.777967747914637e-06,
      "loss": 0.0002,
      "step": 19092
    },
    {
      "epoch": 8.911551925320886,
      "grad_norm": 0.01312255859375,
      "learning_rate": 5.773053236154413e-06,
      "loss": 0.0001,
      "step": 19093
    },
    {
      "epoch": 8.912018669778297,
      "grad_norm": 0.010498046875,
      "learning_rate": 5.76814075320492e-06,
      "loss": 0.0001,
      "step": 19094
    },
    {
      "epoch": 8.912485414235706,
      "grad_norm": 0.00604248046875,
      "learning_rate": 5.763230299171895e-06,
      "loss": 0.0001,
      "step": 19095
    },
    {
      "epoch": 8.912952158693116,
      "grad_norm": 0.0255126953125,
      "learning_rate": 5.758321874161088e-06,
      "loss": 0.0002,
      "step": 19096
    },
    {
      "epoch": 8.913418903150525,
      "grad_norm": 0.01611328125,
      "learning_rate": 5.753415478278168e-06,
      "loss": 0.0001,
      "step": 19097
    },
    {
      "epoch": 8.913885647607934,
      "grad_norm": 0.0054931640625,
      "learning_rate": 5.748511111628785e-06,
      "loss": 0.0002,
      "step": 19098
    },
    {
      "epoch": 8.914352392065345,
      "grad_norm": 0.00762939453125,
      "learning_rate": 5.743608774318521e-06,
      "loss": 0.0001,
      "step": 19099
    },
    {
      "epoch": 8.914819136522754,
      "grad_norm": 0.0103759765625,
      "learning_rate": 5.7387084664529574e-06,
      "loss": 0.0001,
      "step": 19100
    },
    {
      "epoch": 8.915285880980164,
      "grad_norm": 0.0123291015625,
      "learning_rate": 5.733810188137567e-06,
      "loss": 0.0001,
      "step": 19101
    },
    {
      "epoch": 8.915752625437573,
      "grad_norm": 0.006805419921875,
      "learning_rate": 5.728913939477854e-06,
      "loss": 0.0001,
      "step": 19102
    },
    {
      "epoch": 8.916219369894982,
      "grad_norm": 0.007232666015625,
      "learning_rate": 5.724019720579222e-06,
      "loss": 0.0002,
      "step": 19103
    },
    {
      "epoch": 8.916686114352393,
      "grad_norm": 0.01068115234375,
      "learning_rate": 5.719127531547019e-06,
      "loss": 0.0002,
      "step": 19104
    },
    {
      "epoch": 8.917152858809802,
      "grad_norm": 0.00958251953125,
      "learning_rate": 5.7142373724866195e-06,
      "loss": 0.0002,
      "step": 19105
    },
    {
      "epoch": 8.91761960326721,
      "grad_norm": 0.01068115234375,
      "learning_rate": 5.709349243503303e-06,
      "loss": 0.0002,
      "step": 19106
    },
    {
      "epoch": 8.918086347724621,
      "grad_norm": 0.01513671875,
      "learning_rate": 5.70446314470231e-06,
      "loss": 0.0001,
      "step": 19107
    },
    {
      "epoch": 8.91855309218203,
      "grad_norm": 0.00592041015625,
      "learning_rate": 5.699579076188844e-06,
      "loss": 0.0001,
      "step": 19108
    },
    {
      "epoch": 8.919019836639439,
      "grad_norm": 0.008056640625,
      "learning_rate": 5.6946970380680866e-06,
      "loss": 0.0001,
      "step": 19109
    },
    {
      "epoch": 8.91948658109685,
      "grad_norm": 0.02294921875,
      "learning_rate": 5.689817030445121e-06,
      "loss": 0.0002,
      "step": 19110
    },
    {
      "epoch": 8.919953325554259,
      "grad_norm": 0.0225830078125,
      "learning_rate": 5.6849390534250315e-06,
      "loss": 0.0002,
      "step": 19111
    },
    {
      "epoch": 8.92042007001167,
      "grad_norm": 0.01141357421875,
      "learning_rate": 5.680063107112865e-06,
      "loss": 0.0002,
      "step": 19112
    },
    {
      "epoch": 8.920886814469078,
      "grad_norm": 0.0096435546875,
      "learning_rate": 5.675189191613573e-06,
      "loss": 0.0001,
      "step": 19113
    },
    {
      "epoch": 8.921353558926487,
      "grad_norm": 0.007781982421875,
      "learning_rate": 5.6703173070321245e-06,
      "loss": 0.0001,
      "step": 19114
    },
    {
      "epoch": 8.921820303383898,
      "grad_norm": 0.0191650390625,
      "learning_rate": 5.665447453473394e-06,
      "loss": 0.0002,
      "step": 19115
    },
    {
      "epoch": 8.922287047841307,
      "grad_norm": 0.06591796875,
      "learning_rate": 5.660579631042251e-06,
      "loss": 0.0044,
      "step": 19116
    },
    {
      "epoch": 8.922753792298716,
      "grad_norm": 0.0054931640625,
      "learning_rate": 5.6557138398435015e-06,
      "loss": 0.0001,
      "step": 19117
    },
    {
      "epoch": 8.923220536756126,
      "grad_norm": 0.00543212890625,
      "learning_rate": 5.650850079981906e-06,
      "loss": 0.0001,
      "step": 19118
    },
    {
      "epoch": 8.923687281213535,
      "grad_norm": 0.014892578125,
      "learning_rate": 5.645988351562192e-06,
      "loss": 0.0002,
      "step": 19119
    },
    {
      "epoch": 8.924154025670946,
      "grad_norm": 0.0093994140625,
      "learning_rate": 5.641128654689043e-06,
      "loss": 0.0001,
      "step": 19120
    },
    {
      "epoch": 8.924620770128355,
      "grad_norm": 0.00872802734375,
      "learning_rate": 5.636270989467074e-06,
      "loss": 0.0001,
      "step": 19121
    },
    {
      "epoch": 8.925087514585764,
      "grad_norm": 0.0078125,
      "learning_rate": 5.631415356000891e-06,
      "loss": 0.0002,
      "step": 19122
    },
    {
      "epoch": 8.925554259043174,
      "grad_norm": 0.0216064453125,
      "learning_rate": 5.626561754395054e-06,
      "loss": 0.0001,
      "step": 19123
    },
    {
      "epoch": 8.926021003500583,
      "grad_norm": 0.00531005859375,
      "learning_rate": 5.621710184754037e-06,
      "loss": 0.0001,
      "step": 19124
    },
    {
      "epoch": 8.926487747957992,
      "grad_norm": 0.006256103515625,
      "learning_rate": 5.616860647182309e-06,
      "loss": 0.0001,
      "step": 19125
    },
    {
      "epoch": 8.926954492415403,
      "grad_norm": 0.0111083984375,
      "learning_rate": 5.612013141784312e-06,
      "loss": 0.0002,
      "step": 19126
    },
    {
      "epoch": 8.927421236872812,
      "grad_norm": 0.007049560546875,
      "learning_rate": 5.607167668664382e-06,
      "loss": 0.0001,
      "step": 19127
    },
    {
      "epoch": 8.927887981330223,
      "grad_norm": 0.015869140625,
      "learning_rate": 5.60232422792687e-06,
      "loss": 0.0002,
      "step": 19128
    },
    {
      "epoch": 8.928354725787631,
      "grad_norm": 0.0361328125,
      "learning_rate": 5.597482819676058e-06,
      "loss": 0.0017,
      "step": 19129
    },
    {
      "epoch": 8.92882147024504,
      "grad_norm": 0.00421142578125,
      "learning_rate": 5.592643444016166e-06,
      "loss": 0.0001,
      "step": 19130
    },
    {
      "epoch": 8.929288214702451,
      "grad_norm": 0.01708984375,
      "learning_rate": 5.587806101051429e-06,
      "loss": 0.0001,
      "step": 19131
    },
    {
      "epoch": 8.92975495915986,
      "grad_norm": 0.006134033203125,
      "learning_rate": 5.582970790885955e-06,
      "loss": 0.0001,
      "step": 19132
    },
    {
      "epoch": 8.930221703617269,
      "grad_norm": 0.005615234375,
      "learning_rate": 5.578137513623882e-06,
      "loss": 0.0001,
      "step": 19133
    },
    {
      "epoch": 8.93068844807468,
      "grad_norm": 0.00701904296875,
      "learning_rate": 5.573306269369294e-06,
      "loss": 0.0002,
      "step": 19134
    },
    {
      "epoch": 8.931155192532088,
      "grad_norm": 0.006378173828125,
      "learning_rate": 5.568477058226163e-06,
      "loss": 0.0001,
      "step": 19135
    },
    {
      "epoch": 8.931621936989497,
      "grad_norm": 0.01239013671875,
      "learning_rate": 5.563649880298505e-06,
      "loss": 0.0002,
      "step": 19136
    },
    {
      "epoch": 8.932088681446908,
      "grad_norm": 0.038330078125,
      "learning_rate": 5.558824735690249e-06,
      "loss": 0.0019,
      "step": 19137
    },
    {
      "epoch": 8.932555425904317,
      "grad_norm": 0.0050048828125,
      "learning_rate": 5.5540016245052785e-06,
      "loss": 0.0001,
      "step": 19138
    },
    {
      "epoch": 8.933022170361728,
      "grad_norm": 0.01531982421875,
      "learning_rate": 5.5491805468474324e-06,
      "loss": 0.0001,
      "step": 19139
    },
    {
      "epoch": 8.933488914819137,
      "grad_norm": 0.04345703125,
      "learning_rate": 5.544361502820539e-06,
      "loss": 0.0019,
      "step": 19140
    },
    {
      "epoch": 8.933955659276545,
      "grad_norm": 0.013427734375,
      "learning_rate": 5.539544492528337e-06,
      "loss": 0.0002,
      "step": 19141
    },
    {
      "epoch": 8.934422403733956,
      "grad_norm": 0.01300048828125,
      "learning_rate": 5.534729516074533e-06,
      "loss": 0.0001,
      "step": 19142
    },
    {
      "epoch": 8.934889148191365,
      "grad_norm": 0.00830078125,
      "learning_rate": 5.529916573562843e-06,
      "loss": 0.0001,
      "step": 19143
    },
    {
      "epoch": 8.935355892648776,
      "grad_norm": 0.007720947265625,
      "learning_rate": 5.525105665096842e-06,
      "loss": 0.0001,
      "step": 19144
    },
    {
      "epoch": 8.935822637106185,
      "grad_norm": 0.016357421875,
      "learning_rate": 5.520296790780144e-06,
      "loss": 0.0001,
      "step": 19145
    },
    {
      "epoch": 8.936289381563594,
      "grad_norm": 0.04296875,
      "learning_rate": 5.51548995071629e-06,
      "loss": 0.002,
      "step": 19146
    },
    {
      "epoch": 8.936756126021004,
      "grad_norm": 0.00762939453125,
      "learning_rate": 5.5106851450087535e-06,
      "loss": 0.0001,
      "step": 19147
    },
    {
      "epoch": 8.937222870478413,
      "grad_norm": 0.0164794921875,
      "learning_rate": 5.505882373761018e-06,
      "loss": 0.0001,
      "step": 19148
    },
    {
      "epoch": 8.937689614935822,
      "grad_norm": 0.0137939453125,
      "learning_rate": 5.501081637076478e-06,
      "loss": 0.0001,
      "step": 19149
    },
    {
      "epoch": 8.938156359393233,
      "grad_norm": 0.06884765625,
      "learning_rate": 5.496282935058505e-06,
      "loss": 0.0022,
      "step": 19150
    },
    {
      "epoch": 8.938623103850642,
      "grad_norm": 0.0169677734375,
      "learning_rate": 5.491486267810397e-06,
      "loss": 0.0001,
      "step": 19151
    },
    {
      "epoch": 8.93908984830805,
      "grad_norm": 0.00494384765625,
      "learning_rate": 5.48669163543547e-06,
      "loss": 0.0001,
      "step": 19152
    },
    {
      "epoch": 8.939556592765461,
      "grad_norm": 0.041259765625,
      "learning_rate": 5.48189903803692e-06,
      "loss": 0.0031,
      "step": 19153
    },
    {
      "epoch": 8.94002333722287,
      "grad_norm": 0.005706787109375,
      "learning_rate": 5.47710847571794e-06,
      "loss": 0.0001,
      "step": 19154
    },
    {
      "epoch": 8.94049008168028,
      "grad_norm": 0.00469970703125,
      "learning_rate": 5.472319948581717e-06,
      "loss": 0.0001,
      "step": 19155
    },
    {
      "epoch": 8.94095682613769,
      "grad_norm": 0.07958984375,
      "learning_rate": 5.4675334567313104e-06,
      "loss": 0.0021,
      "step": 19156
    },
    {
      "epoch": 8.941423570595099,
      "grad_norm": 0.0084228515625,
      "learning_rate": 5.462749000269795e-06,
      "loss": 0.0002,
      "step": 19157
    },
    {
      "epoch": 8.94189031505251,
      "grad_norm": 0.00775146484375,
      "learning_rate": 5.4579665793001886e-06,
      "loss": 0.0001,
      "step": 19158
    },
    {
      "epoch": 8.942357059509918,
      "grad_norm": 0.007354736328125,
      "learning_rate": 5.453186193925453e-06,
      "loss": 0.0001,
      "step": 19159
    },
    {
      "epoch": 8.942823803967327,
      "grad_norm": 0.060302734375,
      "learning_rate": 5.4484078442485265e-06,
      "loss": 0.0016,
      "step": 19160
    },
    {
      "epoch": 8.943290548424738,
      "grad_norm": 0.0556640625,
      "learning_rate": 5.443631530372284e-06,
      "loss": 0.0027,
      "step": 19161
    },
    {
      "epoch": 8.943757292882147,
      "grad_norm": 0.01025390625,
      "learning_rate": 5.438857252399565e-06,
      "loss": 0.0001,
      "step": 19162
    },
    {
      "epoch": 8.944224037339557,
      "grad_norm": 0.004486083984375,
      "learning_rate": 5.434085010433177e-06,
      "loss": 0.0001,
      "step": 19163
    },
    {
      "epoch": 8.944690781796966,
      "grad_norm": 0.01177978515625,
      "learning_rate": 5.429314804575847e-06,
      "loss": 0.0001,
      "step": 19164
    },
    {
      "epoch": 8.945157526254375,
      "grad_norm": 0.0186767578125,
      "learning_rate": 5.4245466349302945e-06,
      "loss": 0.0002,
      "step": 19165
    },
    {
      "epoch": 8.945624270711786,
      "grad_norm": 0.0106201171875,
      "learning_rate": 5.4197805015992035e-06,
      "loss": 0.0001,
      "step": 19166
    },
    {
      "epoch": 8.946091015169195,
      "grad_norm": 0.007354736328125,
      "learning_rate": 5.415016404685158e-06,
      "loss": 0.0001,
      "step": 19167
    },
    {
      "epoch": 8.946557759626604,
      "grad_norm": 0.055419921875,
      "learning_rate": 5.410254344290766e-06,
      "loss": 0.0002,
      "step": 19168
    },
    {
      "epoch": 8.947024504084014,
      "grad_norm": 0.00750732421875,
      "learning_rate": 5.405494320518545e-06,
      "loss": 0.0001,
      "step": 19169
    },
    {
      "epoch": 8.947491248541423,
      "grad_norm": 0.010009765625,
      "learning_rate": 5.400736333470979e-06,
      "loss": 0.0002,
      "step": 19170
    },
    {
      "epoch": 8.947957992998834,
      "grad_norm": 0.022705078125,
      "learning_rate": 5.39598038325052e-06,
      "loss": 0.0002,
      "step": 19171
    },
    {
      "epoch": 8.948424737456243,
      "grad_norm": 0.00531005859375,
      "learning_rate": 5.391226469959587e-06,
      "loss": 0.0001,
      "step": 19172
    },
    {
      "epoch": 8.948891481913652,
      "grad_norm": 0.0096435546875,
      "learning_rate": 5.386474593700497e-06,
      "loss": 0.0001,
      "step": 19173
    },
    {
      "epoch": 8.949358226371062,
      "grad_norm": 0.017578125,
      "learning_rate": 5.38172475457559e-06,
      "loss": 0.0002,
      "step": 19174
    },
    {
      "epoch": 8.949824970828471,
      "grad_norm": 0.01025390625,
      "learning_rate": 5.376976952687141e-06,
      "loss": 0.0002,
      "step": 19175
    },
    {
      "epoch": 8.95029171528588,
      "grad_norm": 0.01409912109375,
      "learning_rate": 5.372231188137345e-06,
      "loss": 0.0001,
      "step": 19176
    },
    {
      "epoch": 8.950758459743291,
      "grad_norm": 0.0093994140625,
      "learning_rate": 5.367487461028409e-06,
      "loss": 0.0002,
      "step": 19177
    },
    {
      "epoch": 8.9512252042007,
      "grad_norm": 0.00634765625,
      "learning_rate": 5.362745771462474e-06,
      "loss": 0.0001,
      "step": 19178
    },
    {
      "epoch": 8.951691948658109,
      "grad_norm": 0.006256103515625,
      "learning_rate": 5.358006119541614e-06,
      "loss": 0.0002,
      "step": 19179
    },
    {
      "epoch": 8.95215869311552,
      "grad_norm": 0.005615234375,
      "learning_rate": 5.353268505367903e-06,
      "loss": 0.0001,
      "step": 19180
    },
    {
      "epoch": 8.952625437572928,
      "grad_norm": 0.005950927734375,
      "learning_rate": 5.348532929043326e-06,
      "loss": 0.0001,
      "step": 19181
    },
    {
      "epoch": 8.953092182030339,
      "grad_norm": 0.035888671875,
      "learning_rate": 5.343799390669846e-06,
      "loss": 0.0032,
      "step": 19182
    },
    {
      "epoch": 8.953558926487748,
      "grad_norm": 0.009765625,
      "learning_rate": 5.339067890349403e-06,
      "loss": 0.0002,
      "step": 19183
    },
    {
      "epoch": 8.954025670945157,
      "grad_norm": 0.0107421875,
      "learning_rate": 5.334338428183838e-06,
      "loss": 0.0002,
      "step": 19184
    },
    {
      "epoch": 8.954492415402568,
      "grad_norm": 0.00933837890625,
      "learning_rate": 5.329611004275004e-06,
      "loss": 0.0002,
      "step": 19185
    },
    {
      "epoch": 8.954959159859976,
      "grad_norm": 0.0654296875,
      "learning_rate": 5.324885618724696e-06,
      "loss": 0.0018,
      "step": 19186
    },
    {
      "epoch": 8.955425904317387,
      "grad_norm": 0.0111083984375,
      "learning_rate": 5.320162271634632e-06,
      "loss": 0.0002,
      "step": 19187
    },
    {
      "epoch": 8.955892648774796,
      "grad_norm": 0.00823974609375,
      "learning_rate": 5.315440963106522e-06,
      "loss": 0.0001,
      "step": 19188
    },
    {
      "epoch": 8.956359393232205,
      "grad_norm": 0.006561279296875,
      "learning_rate": 5.310721693242038e-06,
      "loss": 0.0001,
      "step": 19189
    },
    {
      "epoch": 8.956826137689616,
      "grad_norm": 0.01239013671875,
      "learning_rate": 5.3060044621427665e-06,
      "loss": 0.0002,
      "step": 19190
    },
    {
      "epoch": 8.957292882147025,
      "grad_norm": 0.09228515625,
      "learning_rate": 5.30128926991027e-06,
      "loss": 0.0005,
      "step": 19191
    },
    {
      "epoch": 8.957759626604433,
      "grad_norm": 0.005767822265625,
      "learning_rate": 5.296576116646113e-06,
      "loss": 0.0001,
      "step": 19192
    },
    {
      "epoch": 8.958226371061844,
      "grad_norm": 0.005767822265625,
      "learning_rate": 5.2918650024517235e-06,
      "loss": 0.0001,
      "step": 19193
    },
    {
      "epoch": 8.958693115519253,
      "grad_norm": 0.059326171875,
      "learning_rate": 5.287155927428566e-06,
      "loss": 0.005,
      "step": 19194
    },
    {
      "epoch": 8.959159859976662,
      "grad_norm": 0.00537109375,
      "learning_rate": 5.282448891678038e-06,
      "loss": 0.0001,
      "step": 19195
    },
    {
      "epoch": 8.959626604434073,
      "grad_norm": 0.0157470703125,
      "learning_rate": 5.277743895301468e-06,
      "loss": 0.0001,
      "step": 19196
    },
    {
      "epoch": 8.960093348891482,
      "grad_norm": 0.017578125,
      "learning_rate": 5.273040938400165e-06,
      "loss": 0.0002,
      "step": 19197
    },
    {
      "epoch": 8.960560093348892,
      "grad_norm": 0.004119873046875,
      "learning_rate": 5.268340021075424e-06,
      "loss": 0.0001,
      "step": 19198
    },
    {
      "epoch": 8.961026837806301,
      "grad_norm": 0.00909423828125,
      "learning_rate": 5.263641143428399e-06,
      "loss": 0.0001,
      "step": 19199
    },
    {
      "epoch": 8.96149358226371,
      "grad_norm": 0.006744384765625,
      "learning_rate": 5.2589443055602876e-06,
      "loss": 0.0001,
      "step": 19200
    },
    {
      "epoch": 8.96196032672112,
      "grad_norm": 0.027099609375,
      "learning_rate": 5.25424950757224e-06,
      "loss": 0.0002,
      "step": 19201
    },
    {
      "epoch": 8.96242707117853,
      "grad_norm": 0.0069580078125,
      "learning_rate": 5.24955674956531e-06,
      "loss": 0.0001,
      "step": 19202
    },
    {
      "epoch": 8.962893815635939,
      "grad_norm": 0.0194091796875,
      "learning_rate": 5.24486603164055e-06,
      "loss": 0.0001,
      "step": 19203
    },
    {
      "epoch": 8.96336056009335,
      "grad_norm": 0.013916015625,
      "learning_rate": 5.240177353898968e-06,
      "loss": 0.0002,
      "step": 19204
    },
    {
      "epoch": 8.963827304550758,
      "grad_norm": 0.00628662109375,
      "learning_rate": 5.235490716441494e-06,
      "loss": 0.0001,
      "step": 19205
    },
    {
      "epoch": 8.964294049008169,
      "grad_norm": 0.01123046875,
      "learning_rate": 5.230806119369036e-06,
      "loss": 0.0001,
      "step": 19206
    },
    {
      "epoch": 8.964760793465578,
      "grad_norm": 0.00482177734375,
      "learning_rate": 5.226123562782503e-06,
      "loss": 0.0001,
      "step": 19207
    },
    {
      "epoch": 8.965227537922987,
      "grad_norm": 0.00567626953125,
      "learning_rate": 5.2214430467826595e-06,
      "loss": 0.0001,
      "step": 19208
    },
    {
      "epoch": 8.965694282380397,
      "grad_norm": 0.0135498046875,
      "learning_rate": 5.216764571470301e-06,
      "loss": 0.0001,
      "step": 19209
    },
    {
      "epoch": 8.966161026837806,
      "grad_norm": 0.0269775390625,
      "learning_rate": 5.2120881369461915e-06,
      "loss": 0.0002,
      "step": 19210
    },
    {
      "epoch": 8.966627771295215,
      "grad_norm": 0.00860595703125,
      "learning_rate": 5.207413743310974e-06,
      "loss": 0.0002,
      "step": 19211
    },
    {
      "epoch": 8.967094515752626,
      "grad_norm": 0.007171630859375,
      "learning_rate": 5.202741390665322e-06,
      "loss": 0.0001,
      "step": 19212
    },
    {
      "epoch": 8.967561260210035,
      "grad_norm": 0.003692626953125,
      "learning_rate": 5.198071079109823e-06,
      "loss": 0.0001,
      "step": 19213
    },
    {
      "epoch": 8.968028004667445,
      "grad_norm": 0.00775146484375,
      "learning_rate": 5.193402808745029e-06,
      "loss": 0.0001,
      "step": 19214
    },
    {
      "epoch": 8.968494749124854,
      "grad_norm": 0.006134033203125,
      "learning_rate": 5.188736579671483e-06,
      "loss": 0.0001,
      "step": 19215
    },
    {
      "epoch": 8.968961493582263,
      "grad_norm": 0.004913330078125,
      "learning_rate": 5.1840723919896135e-06,
      "loss": 0.0001,
      "step": 19216
    },
    {
      "epoch": 8.969428238039674,
      "grad_norm": 0.0142822265625,
      "learning_rate": 5.179410245799865e-06,
      "loss": 0.0002,
      "step": 19217
    },
    {
      "epoch": 8.969894982497083,
      "grad_norm": 0.0086669921875,
      "learning_rate": 5.1747501412026444e-06,
      "loss": 0.0001,
      "step": 19218
    },
    {
      "epoch": 8.970361726954492,
      "grad_norm": 0.006927490234375,
      "learning_rate": 5.170092078298239e-06,
      "loss": 0.0002,
      "step": 19219
    },
    {
      "epoch": 8.970828471411902,
      "grad_norm": 0.080078125,
      "learning_rate": 5.165436057186968e-06,
      "loss": 0.0041,
      "step": 19220
    },
    {
      "epoch": 8.971295215869311,
      "grad_norm": 0.030029296875,
      "learning_rate": 5.160782077969095e-06,
      "loss": 0.0023,
      "step": 19221
    },
    {
      "epoch": 8.97176196032672,
      "grad_norm": 0.0081787109375,
      "learning_rate": 5.156130140744797e-06,
      "loss": 0.0001,
      "step": 19222
    },
    {
      "epoch": 8.972228704784131,
      "grad_norm": 0.0037078857421875,
      "learning_rate": 5.151480245614249e-06,
      "loss": 0.0001,
      "step": 19223
    },
    {
      "epoch": 8.97269544924154,
      "grad_norm": 0.01019287109375,
      "learning_rate": 5.146832392677581e-06,
      "loss": 0.0001,
      "step": 19224
    },
    {
      "epoch": 8.97316219369895,
      "grad_norm": 0.0164794921875,
      "learning_rate": 5.142186582034836e-06,
      "loss": 0.0002,
      "step": 19225
    },
    {
      "epoch": 8.97362893815636,
      "grad_norm": 0.034912109375,
      "learning_rate": 5.137542813786056e-06,
      "loss": 0.0002,
      "step": 19226
    },
    {
      "epoch": 8.974095682613768,
      "grad_norm": 0.01129150390625,
      "learning_rate": 5.13290108803125e-06,
      "loss": 0.0001,
      "step": 19227
    },
    {
      "epoch": 8.974562427071179,
      "grad_norm": 0.005767822265625,
      "learning_rate": 5.128261404870327e-06,
      "loss": 0.0001,
      "step": 19228
    },
    {
      "epoch": 8.975029171528588,
      "grad_norm": 0.0419921875,
      "learning_rate": 5.123623764403207e-06,
      "loss": 0.0023,
      "step": 19229
    },
    {
      "epoch": 8.975495915985999,
      "grad_norm": 0.047119140625,
      "learning_rate": 5.1189881667297325e-06,
      "loss": 0.0026,
      "step": 19230
    },
    {
      "epoch": 8.975962660443408,
      "grad_norm": 0.007080078125,
      "learning_rate": 5.114354611949712e-06,
      "loss": 0.0001,
      "step": 19231
    },
    {
      "epoch": 8.976429404900816,
      "grad_norm": 0.0205078125,
      "learning_rate": 5.109723100162922e-06,
      "loss": 0.0003,
      "step": 19232
    },
    {
      "epoch": 8.976896149358227,
      "grad_norm": 0.01123046875,
      "learning_rate": 5.10509363146906e-06,
      "loss": 0.0002,
      "step": 19233
    },
    {
      "epoch": 8.977362893815636,
      "grad_norm": 0.005157470703125,
      "learning_rate": 5.100466205967836e-06,
      "loss": 0.0001,
      "step": 19234
    },
    {
      "epoch": 8.977829638273045,
      "grad_norm": 0.00921630859375,
      "learning_rate": 5.095840823758869e-06,
      "loss": 0.0001,
      "step": 19235
    },
    {
      "epoch": 8.978296382730456,
      "grad_norm": 0.006866455078125,
      "learning_rate": 5.0912174849417236e-06,
      "loss": 0.0001,
      "step": 19236
    },
    {
      "epoch": 8.978763127187865,
      "grad_norm": 0.00482177734375,
      "learning_rate": 5.086596189615989e-06,
      "loss": 0.0001,
      "step": 19237
    },
    {
      "epoch": 8.979229871645273,
      "grad_norm": 0.0076904296875,
      "learning_rate": 5.08197693788115e-06,
      "loss": 0.0001,
      "step": 19238
    },
    {
      "epoch": 8.979696616102684,
      "grad_norm": 0.10302734375,
      "learning_rate": 5.07735972983664e-06,
      "loss": 0.0003,
      "step": 19239
    },
    {
      "epoch": 8.980163360560093,
      "grad_norm": 0.06494140625,
      "learning_rate": 5.07274456558191e-06,
      "loss": 0.0019,
      "step": 19240
    },
    {
      "epoch": 8.980630105017504,
      "grad_norm": 0.01214599609375,
      "learning_rate": 5.068131445216317e-06,
      "loss": 0.0002,
      "step": 19241
    },
    {
      "epoch": 8.981096849474913,
      "grad_norm": 0.044677734375,
      "learning_rate": 5.06352036883917e-06,
      "loss": 0.0002,
      "step": 19242
    },
    {
      "epoch": 8.981563593932322,
      "grad_norm": 0.0167236328125,
      "learning_rate": 5.0589113365497655e-06,
      "loss": 0.0002,
      "step": 19243
    },
    {
      "epoch": 8.982030338389732,
      "grad_norm": 0.038330078125,
      "learning_rate": 5.054304348447347e-06,
      "loss": 0.0019,
      "step": 19244
    },
    {
      "epoch": 8.982497082847141,
      "grad_norm": 0.006011962890625,
      "learning_rate": 5.049699404631092e-06,
      "loss": 0.0001,
      "step": 19245
    },
    {
      "epoch": 8.98296382730455,
      "grad_norm": 0.039306640625,
      "learning_rate": 5.045096505200165e-06,
      "loss": 0.0001,
      "step": 19246
    },
    {
      "epoch": 8.98343057176196,
      "grad_norm": 0.00994873046875,
      "learning_rate": 5.040495650253674e-06,
      "loss": 0.0001,
      "step": 19247
    },
    {
      "epoch": 8.98389731621937,
      "grad_norm": 0.01141357421875,
      "learning_rate": 5.035896839890652e-06,
      "loss": 0.0001,
      "step": 19248
    },
    {
      "epoch": 8.98436406067678,
      "grad_norm": 0.007110595703125,
      "learning_rate": 5.031300074210132e-06,
      "loss": 0.0001,
      "step": 19249
    },
    {
      "epoch": 8.98483080513419,
      "grad_norm": 0.008544921875,
      "learning_rate": 5.026705353311112e-06,
      "loss": 0.0001,
      "step": 19250
    },
    {
      "epoch": 8.985297549591598,
      "grad_norm": 0.07763671875,
      "learning_rate": 5.022112677292479e-06,
      "loss": 0.004,
      "step": 19251
    },
    {
      "epoch": 8.985764294049009,
      "grad_norm": 0.01080322265625,
      "learning_rate": 5.017522046253142e-06,
      "loss": 0.0001,
      "step": 19252
    },
    {
      "epoch": 8.986231038506418,
      "grad_norm": 0.06982421875,
      "learning_rate": 5.012933460291958e-06,
      "loss": 0.0019,
      "step": 19253
    },
    {
      "epoch": 8.986697782963827,
      "grad_norm": 0.0078125,
      "learning_rate": 5.008346919507689e-06,
      "loss": 0.0001,
      "step": 19254
    },
    {
      "epoch": 8.987164527421237,
      "grad_norm": 0.0089111328125,
      "learning_rate": 5.003762423999114e-06,
      "loss": 0.0001,
      "step": 19255
    },
    {
      "epoch": 8.987631271878646,
      "grad_norm": 0.0113525390625,
      "learning_rate": 4.999179973864942e-06,
      "loss": 0.0002,
      "step": 19256
    },
    {
      "epoch": 8.988098016336057,
      "grad_norm": 0.006561279296875,
      "learning_rate": 4.994599569203817e-06,
      "loss": 0.0001,
      "step": 19257
    },
    {
      "epoch": 8.988564760793466,
      "grad_norm": 0.003448486328125,
      "learning_rate": 4.990021210114371e-06,
      "loss": 0.0001,
      "step": 19258
    },
    {
      "epoch": 8.989031505250875,
      "grad_norm": 0.004730224609375,
      "learning_rate": 4.985444896695201e-06,
      "loss": 0.0001,
      "step": 19259
    },
    {
      "epoch": 8.989498249708285,
      "grad_norm": 0.00537109375,
      "learning_rate": 4.980870629044809e-06,
      "loss": 0.0001,
      "step": 19260
    },
    {
      "epoch": 8.989964994165694,
      "grad_norm": 0.00836181640625,
      "learning_rate": 4.9762984072617035e-06,
      "loss": 0.0002,
      "step": 19261
    },
    {
      "epoch": 8.990431738623103,
      "grad_norm": 0.015380859375,
      "learning_rate": 4.971728231444317e-06,
      "loss": 0.0002,
      "step": 19262
    },
    {
      "epoch": 8.990898483080514,
      "grad_norm": 0.052734375,
      "learning_rate": 4.96716010169106e-06,
      "loss": 0.0022,
      "step": 19263
    },
    {
      "epoch": 8.991365227537923,
      "grad_norm": 0.07568359375,
      "learning_rate": 4.962594018100297e-06,
      "loss": 0.0002,
      "step": 19264
    },
    {
      "epoch": 8.991831971995332,
      "grad_norm": 0.035400390625,
      "learning_rate": 4.958029980770318e-06,
      "loss": 0.001,
      "step": 19265
    },
    {
      "epoch": 8.992298716452742,
      "grad_norm": 0.0050048828125,
      "learning_rate": 4.953467989799409e-06,
      "loss": 0.0001,
      "step": 19266
    },
    {
      "epoch": 8.992765460910151,
      "grad_norm": 0.046142578125,
      "learning_rate": 4.948908045285794e-06,
      "loss": 0.0002,
      "step": 19267
    },
    {
      "epoch": 8.993232205367562,
      "grad_norm": 0.0230712890625,
      "learning_rate": 4.944350147327648e-06,
      "loss": 0.0001,
      "step": 19268
    },
    {
      "epoch": 8.99369894982497,
      "grad_norm": 0.0751953125,
      "learning_rate": 4.939794296023104e-06,
      "loss": 0.0021,
      "step": 19269
    },
    {
      "epoch": 8.99416569428238,
      "grad_norm": 0.00897216796875,
      "learning_rate": 4.935240491470272e-06,
      "loss": 0.0002,
      "step": 19270
    },
    {
      "epoch": 8.99463243873979,
      "grad_norm": 0.01025390625,
      "learning_rate": 4.9306887337671744e-06,
      "loss": 0.0001,
      "step": 19271
    },
    {
      "epoch": 8.9950991831972,
      "grad_norm": 0.005462646484375,
      "learning_rate": 4.926139023011822e-06,
      "loss": 0.0001,
      "step": 19272
    },
    {
      "epoch": 8.995565927654608,
      "grad_norm": 0.0419921875,
      "learning_rate": 4.921591359302202e-06,
      "loss": 0.0024,
      "step": 19273
    },
    {
      "epoch": 8.996032672112019,
      "grad_norm": 0.0081787109375,
      "learning_rate": 4.9170457427362036e-06,
      "loss": 0.0001,
      "step": 19274
    },
    {
      "epoch": 8.996499416569428,
      "grad_norm": 0.00872802734375,
      "learning_rate": 4.912502173411704e-06,
      "loss": 0.0001,
      "step": 19275
    },
    {
      "epoch": 8.996966161026839,
      "grad_norm": 0.00762939453125,
      "learning_rate": 4.907960651426546e-06,
      "loss": 0.0001,
      "step": 19276
    },
    {
      "epoch": 8.997432905484247,
      "grad_norm": 0.0079345703125,
      "learning_rate": 4.903421176878487e-06,
      "loss": 0.0001,
      "step": 19277
    },
    {
      "epoch": 8.997899649941656,
      "grad_norm": 0.006256103515625,
      "learning_rate": 4.898883749865291e-06,
      "loss": 0.0001,
      "step": 19278
    }
  ],
  "logging_steps": 1,
  "max_steps": 21420,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 2142,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.045162537989492e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}