{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999605631581023,
  "eval_steps": 768,
  "global_step": 6339,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00015774736759080333,
      "grad_norm": 10.605186462402344,
      "learning_rate": 7.8125e-09,
      "loss": 1.7188,
      "step": 1
    },
    {
      "epoch": 0.00031549473518160666,
      "grad_norm": 10.251529693603516,
      "learning_rate": 1.5625e-08,
      "loss": 1.7404,
      "step": 2
    },
    {
      "epoch": 0.00047324210277240996,
      "grad_norm": 8.667778015136719,
      "learning_rate": 2.34375e-08,
      "loss": 1.6564,
      "step": 3
    },
    {
      "epoch": 0.0006309894703632133,
      "grad_norm": 9.559566497802734,
      "learning_rate": 3.125e-08,
      "loss": 1.692,
      "step": 4
    },
    {
      "epoch": 0.0007887368379540167,
      "grad_norm": 9.882808685302734,
      "learning_rate": 3.9062499999999997e-08,
      "loss": 1.7072,
      "step": 5
    },
    {
      "epoch": 0.0009464842055448199,
      "grad_norm": 9.359736442565918,
      "learning_rate": 4.6875e-08,
      "loss": 1.699,
      "step": 6
    },
    {
      "epoch": 0.0011042315731356234,
      "grad_norm": 9.632488250732422,
      "learning_rate": 5.46875e-08,
      "loss": 1.713,
      "step": 7
    },
    {
      "epoch": 0.0012619789407264266,
      "grad_norm": 10.159477233886719,
      "learning_rate": 6.25e-08,
      "loss": 1.7004,
      "step": 8
    },
    {
      "epoch": 0.0014197263083172299,
      "grad_norm": 8.86357593536377,
      "learning_rate": 7.03125e-08,
      "loss": 1.6718,
      "step": 9
    },
    {
      "epoch": 0.0015774736759080333,
      "grad_norm": 9.487836837768555,
      "learning_rate": 7.812499999999999e-08,
      "loss": 1.6739,
      "step": 10
    },
    {
      "epoch": 0.0017352210434988366,
      "grad_norm": 11.373818397521973,
      "learning_rate": 8.59375e-08,
      "loss": 1.7022,
      "step": 11
    },
    {
      "epoch": 0.0018929684110896398,
      "grad_norm": 8.63364315032959,
      "learning_rate": 9.375e-08,
      "loss": 1.6508,
      "step": 12
    },
    {
      "epoch": 0.0020507157786804433,
      "grad_norm": 9.603653907775879,
      "learning_rate": 1.015625e-07,
      "loss": 1.7101,
      "step": 13
    },
    {
      "epoch": 0.0022084631462712468,
      "grad_norm": 9.238273620605469,
      "learning_rate": 1.09375e-07,
      "loss": 1.6795,
      "step": 14
    },
    {
      "epoch": 0.00236621051386205,
      "grad_norm": 9.683836936950684,
      "learning_rate": 1.1718749999999999e-07,
      "loss": 1.6917,
      "step": 15
    },
    {
      "epoch": 0.0025239578814528533,
      "grad_norm": 10.214293479919434,
      "learning_rate": 1.25e-07,
      "loss": 1.6798,
      "step": 16
    },
    {
      "epoch": 0.0026817052490436567,
      "grad_norm": 9.098894119262695,
      "learning_rate": 1.328125e-07,
      "loss": 1.6695,
      "step": 17
    },
    {
      "epoch": 0.0028394526166344598,
      "grad_norm": 9.310189247131348,
      "learning_rate": 1.40625e-07,
      "loss": 1.6801,
      "step": 18
    },
    {
      "epoch": 0.0029971999842252632,
      "grad_norm": 9.319585800170898,
      "learning_rate": 1.4843749999999998e-07,
      "loss": 1.6781,
      "step": 19
    },
    {
      "epoch": 0.0031549473518160667,
      "grad_norm": 9.21566104888916,
      "learning_rate": 1.5624999999999999e-07,
      "loss": 1.6803,
      "step": 20
    },
    {
      "epoch": 0.0033126947194068697,
      "grad_norm": 8.562273025512695,
      "learning_rate": 1.640625e-07,
      "loss": 1.6577,
      "step": 21
    },
    {
      "epoch": 0.003470442086997673,
      "grad_norm": 15.55296802520752,
      "learning_rate": 1.71875e-07,
      "loss": 1.6569,
      "step": 22
    },
    {
      "epoch": 0.0036281894545884767,
      "grad_norm": 9.303814888000488,
      "learning_rate": 1.796875e-07,
      "loss": 1.6736,
      "step": 23
    },
    {
      "epoch": 0.0037859368221792797,
      "grad_norm": 10.956175804138184,
      "learning_rate": 1.875e-07,
      "loss": 1.7147,
      "step": 24
    },
    {
      "epoch": 0.003943684189770084,
      "grad_norm": 8.357747077941895,
      "learning_rate": 1.9531249999999998e-07,
      "loss": 1.6438,
      "step": 25
    },
    {
      "epoch": 0.004101431557360887,
      "grad_norm": 9.625614166259766,
      "learning_rate": 2.03125e-07,
      "loss": 1.6928,
      "step": 26
    },
    {
      "epoch": 0.00425917892495169,
      "grad_norm": 9.122293472290039,
      "learning_rate": 2.109375e-07,
      "loss": 1.6726,
      "step": 27
    },
    {
      "epoch": 0.0044169262925424935,
      "grad_norm": 9.233643531799316,
      "learning_rate": 2.1875e-07,
      "loss": 1.6593,
      "step": 28
    },
    {
      "epoch": 0.004574673660133297,
      "grad_norm": 8.721983909606934,
      "learning_rate": 2.265625e-07,
      "loss": 1.6597,
      "step": 29
    },
    {
      "epoch": 0.0047324210277241,
      "grad_norm": 8.40044116973877,
      "learning_rate": 2.3437499999999998e-07,
      "loss": 1.635,
      "step": 30
    },
    {
      "epoch": 0.0048901683953149035,
      "grad_norm": 9.595640182495117,
      "learning_rate": 2.421875e-07,
      "loss": 1.6837,
      "step": 31
    },
    {
      "epoch": 0.0050479157629057065,
      "grad_norm": 8.767451286315918,
      "learning_rate": 2.5e-07,
      "loss": 1.6323,
      "step": 32
    },
    {
      "epoch": 0.0052056631304965096,
      "grad_norm": 8.943890571594238,
      "learning_rate": 2.5781249999999997e-07,
      "loss": 1.6379,
      "step": 33
    },
    {
      "epoch": 0.0053634104980873135,
      "grad_norm": 9.733226776123047,
      "learning_rate": 2.65625e-07,
      "loss": 1.65,
      "step": 34
    },
    {
      "epoch": 0.0055211578656781165,
      "grad_norm": 8.848211288452148,
      "learning_rate": 2.734375e-07,
      "loss": 1.636,
      "step": 35
    },
    {
      "epoch": 0.0056789052332689195,
      "grad_norm": 8.562799453735352,
      "learning_rate": 2.8125e-07,
      "loss": 1.6429,
      "step": 36
    },
    {
      "epoch": 0.005836652600859723,
      "grad_norm": 8.89713191986084,
      "learning_rate": 2.890625e-07,
      "loss": 1.6226,
      "step": 37
    },
    {
      "epoch": 0.0059943999684505265,
      "grad_norm": 9.156657218933105,
      "learning_rate": 2.9687499999999996e-07,
      "loss": 1.6312,
      "step": 38
    },
    {
      "epoch": 0.0061521473360413295,
      "grad_norm": 9.411709785461426,
      "learning_rate": 3.046875e-07,
      "loss": 1.6134,
      "step": 39
    },
    {
      "epoch": 0.006309894703632133,
      "grad_norm": 8.37478256225586,
      "learning_rate": 3.1249999999999997e-07,
      "loss": 1.6248,
      "step": 40
    },
    {
      "epoch": 0.006467642071222936,
      "grad_norm": 9.141951560974121,
      "learning_rate": 3.203125e-07,
      "loss": 1.6487,
      "step": 41
    },
    {
      "epoch": 0.0066253894388137394,
      "grad_norm": 9.102831840515137,
      "learning_rate": 3.28125e-07,
      "loss": 1.6006,
      "step": 42
    },
    {
      "epoch": 0.006783136806404543,
      "grad_norm": 8.59057903289795,
      "learning_rate": 3.3593749999999996e-07,
      "loss": 1.6389,
      "step": 43
    },
    {
      "epoch": 0.006940884173995346,
      "grad_norm": 7.083738803863525,
      "learning_rate": 3.4375e-07,
      "loss": 1.5743,
      "step": 44
    },
    {
      "epoch": 0.007098631541586149,
      "grad_norm": 7.837166786193848,
      "learning_rate": 3.5156249999999997e-07,
      "loss": 1.5881,
      "step": 45
    },
    {
      "epoch": 0.007256378909176953,
      "grad_norm": 9.42550277709961,
      "learning_rate": 3.59375e-07,
      "loss": 1.6255,
      "step": 46
    },
    {
      "epoch": 0.007414126276767756,
      "grad_norm": 9.34886646270752,
      "learning_rate": 3.671875e-07,
      "loss": 1.6094,
      "step": 47
    },
    {
      "epoch": 0.007571873644358559,
      "grad_norm": 8.45085334777832,
      "learning_rate": 3.75e-07,
      "loss": 1.5775,
      "step": 48
    },
    {
      "epoch": 0.007729621011949363,
      "grad_norm": 9.391676902770996,
      "learning_rate": 3.828125e-07,
      "loss": 1.6273,
      "step": 49
    },
    {
      "epoch": 0.007887368379540167,
      "grad_norm": 8.811574935913086,
      "learning_rate": 3.9062499999999997e-07,
      "loss": 1.5805,
      "step": 50
    },
    {
      "epoch": 0.00804511574713097,
      "grad_norm": 8.590651512145996,
      "learning_rate": 3.984375e-07,
      "loss": 1.6052,
      "step": 51
    },
    {
      "epoch": 0.008202863114721773,
      "grad_norm": 9.803258895874023,
      "learning_rate": 4.0625e-07,
      "loss": 1.5911,
      "step": 52
    },
    {
      "epoch": 0.008360610482312576,
      "grad_norm": 10.158894538879395,
      "learning_rate": 4.140625e-07,
      "loss": 1.6011,
      "step": 53
    },
    {
      "epoch": 0.00851835784990338,
      "grad_norm": 8.869569778442383,
      "learning_rate": 4.21875e-07,
      "loss": 1.5658,
      "step": 54
    },
    {
      "epoch": 0.008676105217494182,
      "grad_norm": 9.200840950012207,
      "learning_rate": 4.2968749999999996e-07,
      "loss": 1.5631,
      "step": 55
    },
    {
      "epoch": 0.008833852585084987,
      "grad_norm": 9.003901481628418,
      "learning_rate": 4.375e-07,
      "loss": 1.5568,
      "step": 56
    },
    {
      "epoch": 0.00899159995267579,
      "grad_norm": 10.028145790100098,
      "learning_rate": 4.4531249999999997e-07,
      "loss": 1.5362,
      "step": 57
    },
    {
      "epoch": 0.009149347320266593,
      "grad_norm": 8.84991455078125,
      "learning_rate": 4.53125e-07,
      "loss": 1.5227,
      "step": 58
    },
    {
      "epoch": 0.009307094687857396,
      "grad_norm": 9.381906509399414,
      "learning_rate": 4.609375e-07,
      "loss": 1.5435,
      "step": 59
    },
    {
      "epoch": 0.0094648420554482,
      "grad_norm": 9.109251976013184,
      "learning_rate": 4.6874999999999996e-07,
      "loss": 1.5378,
      "step": 60
    },
    {
      "epoch": 0.009622589423039002,
      "grad_norm": 8.686064720153809,
      "learning_rate": 4.765625e-07,
      "loss": 1.5229,
      "step": 61
    },
    {
      "epoch": 0.009780336790629807,
      "grad_norm": 8.528047561645508,
      "learning_rate": 4.84375e-07,
      "loss": 1.5224,
      "step": 62
    },
    {
      "epoch": 0.00993808415822061,
      "grad_norm": 9.493647575378418,
      "learning_rate": 4.921875e-07,
      "loss": 1.5203,
      "step": 63
    },
    {
      "epoch": 0.010095831525811413,
      "grad_norm": 8.9859037399292,
      "learning_rate": 5e-07,
      "loss": 1.5174,
      "step": 64
    },
    {
      "epoch": 0.010253578893402216,
      "grad_norm": 9.881206512451172,
      "learning_rate": 5.078125e-07,
      "loss": 1.4765,
      "step": 65
    },
    {
      "epoch": 0.010411326260993019,
      "grad_norm": 9.306422233581543,
      "learning_rate": 5.156249999999999e-07,
      "loss": 1.4665,
      "step": 66
    },
    {
      "epoch": 0.010569073628583822,
      "grad_norm": 9.779091835021973,
      "learning_rate": 5.234375e-07,
      "loss": 1.486,
      "step": 67
    },
    {
      "epoch": 0.010726820996174627,
      "grad_norm": 10.422354698181152,
      "learning_rate": 5.3125e-07,
      "loss": 1.4697,
      "step": 68
    },
    {
      "epoch": 0.01088456836376543,
      "grad_norm": 9.95809555053711,
      "learning_rate": 5.390625e-07,
      "loss": 1.453,
      "step": 69
    },
    {
      "epoch": 0.011042315731356233,
      "grad_norm": 11.298678398132324,
      "learning_rate": 5.46875e-07,
      "loss": 1.4186,
      "step": 70
    },
    {
      "epoch": 0.011200063098947036,
      "grad_norm": 10.737797737121582,
      "learning_rate": 5.546874999999999e-07,
      "loss": 1.43,
      "step": 71
    },
    {
      "epoch": 0.011357810466537839,
      "grad_norm": 10.393301010131836,
      "learning_rate": 5.625e-07,
      "loss": 1.4079,
      "step": 72
    },
    {
      "epoch": 0.011515557834128644,
      "grad_norm": 10.883868217468262,
      "learning_rate": 5.703125e-07,
      "loss": 1.4192,
      "step": 73
    },
    {
      "epoch": 0.011673305201719447,
      "grad_norm": 16.47789192199707,
      "learning_rate": 5.78125e-07,
      "loss": 1.3894,
      "step": 74
    },
    {
      "epoch": 0.01183105256931025,
      "grad_norm": 11.472434997558594,
      "learning_rate": 5.859375e-07,
      "loss": 1.3653,
      "step": 75
    },
    {
      "epoch": 0.011988799936901053,
      "grad_norm": 12.478020668029785,
      "learning_rate": 5.937499999999999e-07,
      "loss": 1.3557,
      "step": 76
    },
    {
      "epoch": 0.012146547304491856,
      "grad_norm": 14.180948257446289,
      "learning_rate": 6.015625e-07,
      "loss": 1.3193,
      "step": 77
    },
    {
      "epoch": 0.012304294672082659,
      "grad_norm": 12.813949584960938,
      "learning_rate": 6.09375e-07,
      "loss": 1.3267,
      "step": 78
    },
    {
      "epoch": 0.012462042039673464,
      "grad_norm": 11.938271522521973,
      "learning_rate": 6.171875e-07,
      "loss": 1.2896,
      "step": 79
    },
    {
      "epoch": 0.012619789407264267,
      "grad_norm": 12.180197715759277,
      "learning_rate": 6.249999999999999e-07,
      "loss": 1.2662,
      "step": 80
    },
    {
      "epoch": 0.01277753677485507,
      "grad_norm": 10.459107398986816,
      "learning_rate": 6.328124999999999e-07,
      "loss": 1.2404,
      "step": 81
    },
    {
      "epoch": 0.012935284142445873,
      "grad_norm": 12.862483024597168,
      "learning_rate": 6.40625e-07,
      "loss": 1.1687,
      "step": 82
    },
    {
      "epoch": 0.013093031510036676,
      "grad_norm": 8.410504341125488,
      "learning_rate": 6.484375e-07,
      "loss": 1.2366,
      "step": 83
    },
    {
      "epoch": 0.013250778877627479,
      "grad_norm": 708.03125,
      "learning_rate": 6.5625e-07,
      "loss": 1.2195,
      "step": 84
    },
    {
      "epoch": 0.013408526245218284,
      "grad_norm": 8.076370239257812,
      "learning_rate": 6.640624999999999e-07,
      "loss": 1.1526,
      "step": 85
    },
    {
      "epoch": 0.013566273612809087,
      "grad_norm": 7.006606101989746,
      "learning_rate": 6.718749999999999e-07,
      "loss": 1.2265,
      "step": 86
    },
    {
      "epoch": 0.01372402098039989,
      "grad_norm": 7.100209712982178,
      "learning_rate": 6.796875e-07,
      "loss": 1.1768,
      "step": 87
    },
    {
      "epoch": 0.013881768347990693,
      "grad_norm": 8.758187294006348,
      "learning_rate": 6.875e-07,
      "loss": 1.1765,
      "step": 88
    },
    {
      "epoch": 0.014039515715581496,
      "grad_norm": 8.036221504211426,
      "learning_rate": 6.953125e-07,
      "loss": 1.1236,
      "step": 89
    },
    {
      "epoch": 0.014197263083172299,
      "grad_norm": 6.954113006591797,
      "learning_rate": 7.031249999999999e-07,
      "loss": 1.1204,
      "step": 90
    },
    {
      "epoch": 0.014355010450763104,
      "grad_norm": 7.305809497833252,
      "learning_rate": 7.109374999999999e-07,
      "loss": 1.1557,
      "step": 91
    },
    {
      "epoch": 0.014512757818353907,
      "grad_norm": 5.923920631408691,
      "learning_rate": 7.1875e-07,
      "loss": 1.0172,
      "step": 92
    },
    {
      "epoch": 0.01467050518594471,
      "grad_norm": 8.816319465637207,
      "learning_rate": 7.265625e-07,
      "loss": 1.1,
      "step": 93
    },
    {
      "epoch": 0.014828252553535513,
      "grad_norm": 8.399360656738281,
      "learning_rate": 7.34375e-07,
      "loss": 1.1316,
      "step": 94
    },
    {
      "epoch": 0.014985999921126316,
      "grad_norm": 8.26098346710205,
      "learning_rate": 7.421874999999999e-07,
      "loss": 1.059,
      "step": 95
    },
    {
      "epoch": 0.015143747288717119,
      "grad_norm": 7.330342769622803,
      "learning_rate": 7.5e-07,
      "loss": 1.0831,
      "step": 96
    },
    {
      "epoch": 0.015301494656307923,
      "grad_norm": 7.985196113586426,
      "learning_rate": 7.578125e-07,
      "loss": 1.0961,
      "step": 97
    },
    {
      "epoch": 0.015459242023898727,
      "grad_norm": 17.839303970336914,
      "learning_rate": 7.65625e-07,
      "loss": 1.0891,
      "step": 98
    },
    {
      "epoch": 0.01561698939148953,
      "grad_norm": 6.162847995758057,
      "learning_rate": 7.734375e-07,
      "loss": 1.0715,
      "step": 99
    },
    {
      "epoch": 0.015774736759080334,
      "grad_norm": 120.85552978515625,
      "learning_rate": 7.812499999999999e-07,
      "loss": 1.0538,
      "step": 100
    },
    {
      "epoch": 0.015932484126671136,
      "grad_norm": 6.262155055999756,
      "learning_rate": 7.890625e-07,
      "loss": 0.9359,
      "step": 101
    },
    {
      "epoch": 0.01609023149426194,
      "grad_norm": 8.068511962890625,
      "learning_rate": 7.96875e-07,
      "loss": 1.0444,
      "step": 102
    },
    {
      "epoch": 0.01624797886185274,
      "grad_norm": 5.990355968475342,
      "learning_rate": 8.046875e-07,
      "loss": 0.9598,
      "step": 103
    },
    {
      "epoch": 0.016405726229443546,
      "grad_norm": 7.656558990478516,
      "learning_rate": 8.125e-07,
      "loss": 1.0295,
      "step": 104
    },
    {
      "epoch": 0.01656347359703435,
      "grad_norm": 8.8897123336792,
      "learning_rate": 8.203124999999999e-07,
      "loss": 0.9684,
      "step": 105
    },
    {
      "epoch": 0.016721220964625153,
      "grad_norm": 6.440278053283691,
      "learning_rate": 8.28125e-07,
      "loss": 0.9391,
      "step": 106
    },
    {
      "epoch": 0.016878968332215957,
      "grad_norm": 6.96622896194458,
      "learning_rate": 8.359375e-07,
      "loss": 0.9678,
      "step": 107
    },
    {
      "epoch": 0.01703671569980676,
      "grad_norm": 6.747005939483643,
      "learning_rate": 8.4375e-07,
      "loss": 0.9858,
      "step": 108
    },
    {
      "epoch": 0.017194463067397563,
      "grad_norm": 7.434170722961426,
      "learning_rate": 8.515624999999999e-07,
      "loss": 0.9164,
      "step": 109
    },
    {
      "epoch": 0.017352210434988365,
      "grad_norm": 5.481407165527344,
      "learning_rate": 8.593749999999999e-07,
      "loss": 0.9099,
      "step": 110
    },
    {
      "epoch": 0.01750995780257917,
      "grad_norm": 7.978529930114746,
      "learning_rate": 8.671875e-07,
      "loss": 0.8959,
      "step": 111
    },
    {
      "epoch": 0.017667705170169974,
      "grad_norm": 15.46996021270752,
      "learning_rate": 8.75e-07,
      "loss": 0.9717,
      "step": 112
    },
    {
      "epoch": 0.017825452537760775,
      "grad_norm": 6.016064643859863,
      "learning_rate": 8.828125e-07,
      "loss": 0.9127,
      "step": 113
    },
    {
      "epoch": 0.01798319990535158,
      "grad_norm": 7.6805009841918945,
      "learning_rate": 8.906249999999999e-07,
      "loss": 0.8833,
      "step": 114
    },
    {
      "epoch": 0.01814094727294238,
      "grad_norm": 5.962435722351074,
      "learning_rate": 8.984374999999999e-07,
      "loss": 0.9002,
      "step": 115
    },
    {
      "epoch": 0.018298694640533186,
      "grad_norm": 34.05613327026367,
      "learning_rate": 9.0625e-07,
      "loss": 0.8052,
      "step": 116
    },
    {
      "epoch": 0.01845644200812399,
      "grad_norm": 7.545765399932861,
      "learning_rate": 9.140625e-07,
      "loss": 0.8975,
      "step": 117
    },
    {
      "epoch": 0.018614189375714792,
      "grad_norm": 6.937775611877441,
      "learning_rate": 9.21875e-07,
      "loss": 0.8179,
      "step": 118
    },
    {
      "epoch": 0.018771936743305597,
      "grad_norm": 7.694420337677002,
      "learning_rate": 9.296874999999999e-07,
      "loss": 0.8432,
      "step": 119
    },
    {
      "epoch": 0.0189296841108964,
      "grad_norm": 6.258421897888184,
      "learning_rate": 9.374999999999999e-07,
      "loss": 0.8252,
      "step": 120
    },
    {
      "epoch": 0.019087431478487203,
      "grad_norm": 6.795842170715332,
      "learning_rate": 9.453125e-07,
      "loss": 0.7777,
      "step": 121
    },
    {
      "epoch": 0.019245178846078004,
      "grad_norm": 8.250446319580078,
      "learning_rate": 9.53125e-07,
      "loss": 0.923,
      "step": 122
    },
    {
      "epoch": 0.01940292621366881,
      "grad_norm": 5.323596477508545,
      "learning_rate": 9.609374999999999e-07,
      "loss": 0.7119,
      "step": 123
    },
    {
      "epoch": 0.019560673581259614,
      "grad_norm": 6.934390544891357,
      "learning_rate": 9.6875e-07,
      "loss": 0.8482,
      "step": 124
    },
    {
      "epoch": 0.019718420948850415,
      "grad_norm": 6.222138404846191,
      "learning_rate": 9.765625e-07,
      "loss": 0.8095,
      "step": 125
    },
    {
      "epoch": 0.01987616831644122,
      "grad_norm": 8.063146591186523,
      "learning_rate": 9.84375e-07,
      "loss": 0.7474,
      "step": 126
    },
    {
      "epoch": 0.02003391568403202,
      "grad_norm": 8.791608810424805,
      "learning_rate": 9.921875e-07,
      "loss": 0.7099,
      "step": 127
    },
    {
      "epoch": 0.020191663051622826,
      "grad_norm": 9.693697929382324,
      "learning_rate": 1e-06,
      "loss": 0.6944,
      "step": 128
    },
    {
      "epoch": 0.02034941041921363,
      "grad_norm": 9.193809509277344,
      "learning_rate": 9.998389953308647e-07,
      "loss": 0.7153,
      "step": 129
    },
    {
      "epoch": 0.020507157786804432,
      "grad_norm": 5.666736125946045,
      "learning_rate": 9.996779906617292e-07,
      "loss": 0.6735,
      "step": 130
    },
    {
      "epoch": 0.020664905154395237,
      "grad_norm": 8.746193885803223,
      "learning_rate": 9.995169859925937e-07,
      "loss": 0.7298,
      "step": 131
    },
    {
      "epoch": 0.020822652521986038,
      "grad_norm": 30.897165298461914,
      "learning_rate": 9.993559813234584e-07,
      "loss": 0.6898,
      "step": 132
    },
    {
      "epoch": 0.020980399889576843,
      "grad_norm": 5.508756160736084,
      "learning_rate": 9.991949766543229e-07,
      "loss": 0.6829,
      "step": 133
    },
    {
      "epoch": 0.021138147257167644,
      "grad_norm": 9.121854782104492,
      "learning_rate": 9.990339719851876e-07,
      "loss": 0.6911,
      "step": 134
    },
    {
      "epoch": 0.02129589462475845,
      "grad_norm": 4.720840930938721,
      "learning_rate": 9.98872967316052e-07,
      "loss": 0.5923,
      "step": 135
    },
    {
      "epoch": 0.021453641992349254,
      "grad_norm": 5.468916416168213,
      "learning_rate": 9.987119626469168e-07,
      "loss": 0.6364,
      "step": 136
    },
    {
      "epoch": 0.021611389359940055,
      "grad_norm": 7.017789840698242,
      "learning_rate": 9.985509579777813e-07,
      "loss": 0.6458,
      "step": 137
    },
    {
      "epoch": 0.02176913672753086,
      "grad_norm": 7.074862003326416,
      "learning_rate": 9.98389953308646e-07,
      "loss": 0.5727,
      "step": 138
    },
    {
      "epoch": 0.02192688409512166,
      "grad_norm": 5.27205753326416,
      "learning_rate": 9.982289486395105e-07,
      "loss": 0.5757,
      "step": 139
    },
    {
      "epoch": 0.022084631462712466,
      "grad_norm": 5.0011677742004395,
      "learning_rate": 9.98067943970375e-07,
      "loss": 0.5238,
      "step": 140
    },
    {
      "epoch": 0.02224237883030327,
      "grad_norm": 6.091749668121338,
      "learning_rate": 9.979069393012397e-07,
      "loss": 0.5863,
      "step": 141
    },
    {
      "epoch": 0.022400126197894072,
      "grad_norm": 10.343565940856934,
      "learning_rate": 9.977459346321042e-07,
      "loss": 0.5043,
      "step": 142
    },
    {
      "epoch": 0.022557873565484877,
      "grad_norm": 9.453807830810547,
      "learning_rate": 9.975849299629687e-07,
      "loss": 0.5515,
      "step": 143
    },
    {
      "epoch": 0.022715620933075678,
      "grad_norm": 17.887365341186523,
      "learning_rate": 9.974239252938334e-07,
      "loss": 0.5898,
      "step": 144
    },
    {
      "epoch": 0.022873368300666483,
      "grad_norm": 4.972782611846924,
      "learning_rate": 9.972629206246982e-07,
      "loss": 0.5454,
      "step": 145
    },
    {
      "epoch": 0.023031115668257288,
      "grad_norm": 4.8516526222229,
      "learning_rate": 9.971019159555627e-07,
      "loss": 0.539,
      "step": 146
    },
    {
      "epoch": 0.02318886303584809,
      "grad_norm": 11.199058532714844,
      "learning_rate": 9.969409112864272e-07,
      "loss": 0.4849,
      "step": 147
    },
    {
      "epoch": 0.023346610403438894,
      "grad_norm": 4.304190158843994,
      "learning_rate": 9.967799066172919e-07,
      "loss": 0.5409,
      "step": 148
    },
    {
      "epoch": 0.023504357771029695,
      "grad_norm": 6.043247222900391,
      "learning_rate": 9.966189019481564e-07,
      "loss": 0.4697,
      "step": 149
    },
    {
      "epoch": 0.0236621051386205,
      "grad_norm": 4.9490966796875,
      "learning_rate": 9.96457897279021e-07,
      "loss": 0.5018,
      "step": 150
    },
    {
      "epoch": 0.0238198525062113,
      "grad_norm": 8.612019538879395,
      "learning_rate": 9.962968926098856e-07,
      "loss": 0.537,
      "step": 151
    },
    {
      "epoch": 0.023977599873802106,
      "grad_norm": 5.846463203430176,
      "learning_rate": 9.961358879407503e-07,
      "loss": 0.5215,
      "step": 152
    },
    {
      "epoch": 0.02413534724139291,
      "grad_norm": 8.96881103515625,
      "learning_rate": 9.959748832716148e-07,
      "loss": 0.5182,
      "step": 153
    },
    {
      "epoch": 0.024293094608983712,
      "grad_norm": 5.526561260223389,
      "learning_rate": 9.958138786024795e-07,
      "loss": 0.4682,
      "step": 154
    },
    {
      "epoch": 0.024450841976574517,
      "grad_norm": 5.215470314025879,
      "learning_rate": 9.95652873933344e-07,
      "loss": 0.5077,
      "step": 155
    },
    {
      "epoch": 0.024608589344165318,
      "grad_norm": 7.315175533294678,
      "learning_rate": 9.954918692642085e-07,
      "loss": 0.4901,
      "step": 156
    },
    {
      "epoch": 0.024766336711756123,
      "grad_norm": 5.9891037940979,
      "learning_rate": 9.953308645950732e-07,
      "loss": 0.4722,
      "step": 157
    },
    {
      "epoch": 0.024924084079346927,
      "grad_norm": 10.207083702087402,
      "learning_rate": 9.95169859925938e-07,
      "loss": 0.4682,
      "step": 158
    },
    {
      "epoch": 0.02508183144693773,
      "grad_norm": 4.114821434020996,
      "learning_rate": 9.950088552568024e-07,
      "loss": 0.4392,
      "step": 159
    },
    {
      "epoch": 0.025239578814528534,
      "grad_norm": 4.719483375549316,
      "learning_rate": 9.94847850587667e-07,
      "loss": 0.4616,
      "step": 160
    },
    {
      "epoch": 0.025397326182119335,
      "grad_norm": 5.122009754180908,
      "learning_rate": 9.946868459185316e-07,
      "loss": 0.4797,
      "step": 161
    },
    {
      "epoch": 0.02555507354971014,
      "grad_norm": 6.682754039764404,
      "learning_rate": 9.945258412493961e-07,
      "loss": 0.3869,
      "step": 162
    },
    {
      "epoch": 0.02571282091730094,
      "grad_norm": 5.727705001831055,
      "learning_rate": 9.943648365802609e-07,
      "loss": 0.4822,
      "step": 163
    },
    {
      "epoch": 0.025870568284891746,
      "grad_norm": 5.69813346862793,
      "learning_rate": 9.942038319111254e-07,
      "loss": 0.4607,
      "step": 164
    },
    {
      "epoch": 0.02602831565248255,
      "grad_norm": 14.972919464111328,
      "learning_rate": 9.940428272419899e-07,
      "loss": 0.3776,
      "step": 165
    },
    {
      "epoch": 0.02618606302007335,
      "grad_norm": 4.726632118225098,
      "learning_rate": 9.938818225728546e-07,
      "loss": 0.401,
      "step": 166
    },
    {
      "epoch": 0.026343810387664156,
      "grad_norm": 4.312942028045654,
      "learning_rate": 9.937208179037193e-07,
      "loss": 0.3945,
      "step": 167
    },
    {
      "epoch": 0.026501557755254958,
      "grad_norm": 10.081215858459473,
      "learning_rate": 9.935598132345838e-07,
      "loss": 0.4185,
      "step": 168
    },
    {
      "epoch": 0.026659305122845763,
      "grad_norm": 6.823802947998047,
      "learning_rate": 9.933988085654483e-07,
      "loss": 0.3749,
      "step": 169
    },
    {
      "epoch": 0.026817052490436567,
      "grad_norm": 5.444103717803955,
      "learning_rate": 9.93237803896313e-07,
      "loss": 0.3603,
      "step": 170
    },
    {
      "epoch": 0.02697479985802737,
      "grad_norm": 7.1029887199401855,
      "learning_rate": 9.930767992271775e-07,
      "loss": 0.3454,
      "step": 171
    },
    {
      "epoch": 0.027132547225618173,
      "grad_norm": 19.086397171020508,
      "learning_rate": 9.92915794558042e-07,
      "loss": 0.3951,
      "step": 172
    },
    {
      "epoch": 0.027290294593208975,
      "grad_norm": 6.493707180023193,
      "learning_rate": 9.927547898889067e-07,
      "loss": 0.3904,
      "step": 173
    },
    {
      "epoch": 0.02744804196079978,
      "grad_norm": 15.91336441040039,
      "learning_rate": 9.925937852197714e-07,
      "loss": 0.4409,
      "step": 174
    },
    {
      "epoch": 0.027605789328390584,
      "grad_norm": 3.995539426803589,
      "learning_rate": 9.92432780550636e-07,
      "loss": 0.3163,
      "step": 175
    },
    {
      "epoch": 0.027763536695981385,
      "grad_norm": 13.101820945739746,
      "learning_rate": 9.922717758815006e-07,
      "loss": 0.3018,
      "step": 176
    },
    {
      "epoch": 0.02792128406357219,
      "grad_norm": 5.566201686859131,
      "learning_rate": 9.921107712123651e-07,
      "loss": 0.3522,
      "step": 177
    },
    {
      "epoch": 0.02807903143116299,
      "grad_norm": 4.240898609161377,
      "learning_rate": 9.919497665432296e-07,
      "loss": 0.3409,
      "step": 178
    },
    {
      "epoch": 0.028236778798753796,
      "grad_norm": 5.057911396026611,
      "learning_rate": 9.917887618740943e-07,
      "loss": 0.2774,
      "step": 179
    },
    {
      "epoch": 0.028394526166344598,
      "grad_norm": 10.959158897399902,
      "learning_rate": 9.916277572049588e-07,
      "loss": 0.292,
      "step": 180
    },
    {
      "epoch": 0.028552273533935402,
      "grad_norm": 10.364619255065918,
      "learning_rate": 9.914667525358236e-07,
      "loss": 0.3145,
      "step": 181
    },
    {
      "epoch": 0.028710020901526207,
      "grad_norm": 6.3560004234313965,
      "learning_rate": 9.91305747866688e-07,
      "loss": 0.3154,
      "step": 182
    },
    {
      "epoch": 0.02886776826911701,
      "grad_norm": 17.252134323120117,
      "learning_rate": 9.911447431975528e-07,
      "loss": 0.3492,
      "step": 183
    },
    {
      "epoch": 0.029025515636707813,
      "grad_norm": 5.359389305114746,
      "learning_rate": 9.909837385284173e-07,
      "loss": 0.2964,
      "step": 184
    },
    {
      "epoch": 0.029183263004298615,
      "grad_norm": 9.04397201538086,
      "learning_rate": 9.908227338592818e-07,
      "loss": 0.2977,
      "step": 185
    },
    {
      "epoch": 0.02934101037188942,
      "grad_norm": 13.794440269470215,
      "learning_rate": 9.906617291901465e-07,
      "loss": 0.2825,
      "step": 186
    },
    {
      "epoch": 0.029498757739480224,
      "grad_norm": 5.914454460144043,
      "learning_rate": 9.90500724521011e-07,
      "loss": 0.3258,
      "step": 187
    },
    {
      "epoch": 0.029656505107071025,
      "grad_norm": 4.176006317138672,
      "learning_rate": 9.903397198518757e-07,
      "loss": 0.2532,
      "step": 188
    },
    {
      "epoch": 0.02981425247466183,
      "grad_norm": 4.511537075042725,
      "learning_rate": 9.901787151827402e-07,
      "loss": 0.3124,
      "step": 189
    },
    {
      "epoch": 0.02997199984225263,
      "grad_norm": 12.462079048156738,
      "learning_rate": 9.90017710513605e-07,
      "loss": 0.252,
      "step": 190
    },
    {
      "epoch": 0.030129747209843436,
      "grad_norm": 8.845741271972656,
      "learning_rate": 9.898567058444694e-07,
      "loss": 0.3113,
      "step": 191
    },
    {
      "epoch": 0.030287494577434237,
      "grad_norm": 7.979472637176514,
      "learning_rate": 9.896957011753341e-07,
      "loss": 0.25,
      "step": 192
    },
    {
      "epoch": 0.030445241945025042,
      "grad_norm": 5.637496471405029,
      "learning_rate": 9.895346965061986e-07,
      "loss": 0.2759,
      "step": 193
    },
    {
      "epoch": 0.030602989312615847,
      "grad_norm": 6.822127819061279,
      "learning_rate": 9.893736918370631e-07,
      "loss": 0.2119,
      "step": 194
    },
    {
      "epoch": 0.03076073668020665,
      "grad_norm": 7.106624126434326,
      "learning_rate": 9.892126871679278e-07,
      "loss": 0.3029,
      "step": 195
    },
    {
      "epoch": 0.030918484047797453,
      "grad_norm": 8.90538501739502,
      "learning_rate": 9.890516824987925e-07,
      "loss": 0.3089,
      "step": 196
    },
    {
      "epoch": 0.031076231415388254,
      "grad_norm": 7.088816165924072,
      "learning_rate": 9.88890677829657e-07,
      "loss": 0.3151,
      "step": 197
    },
    {
      "epoch": 0.03123397878297906,
      "grad_norm": 3.6720592975616455,
      "learning_rate": 9.887296731605215e-07,
      "loss": 0.2142,
      "step": 198
    },
    {
      "epoch": 0.031391726150569864,
      "grad_norm": 4.644659042358398,
      "learning_rate": 9.885686684913863e-07,
      "loss": 0.1983,
      "step": 199
    },
    {
      "epoch": 0.03154947351816067,
      "grad_norm": 6.448646545410156,
      "learning_rate": 9.884076638222508e-07,
      "loss": 0.2224,
      "step": 200
    },
    {
      "epoch": 0.031707220885751466,
      "grad_norm": 6.825787544250488,
      "learning_rate": 9.882466591531155e-07,
      "loss": 0.2749,
      "step": 201
    },
    {
      "epoch": 0.03186496825334227,
      "grad_norm": 5.165946960449219,
      "learning_rate": 9.8808565448398e-07,
      "loss": 0.2671,
      "step": 202
    },
    {
      "epoch": 0.032022715620933076,
      "grad_norm": 5.763522148132324,
      "learning_rate": 9.879246498148445e-07,
      "loss": 0.2456,
      "step": 203
    },
    {
      "epoch": 0.03218046298852388,
      "grad_norm": 5.943336486816406,
      "learning_rate": 9.877636451457092e-07,
      "loss": 0.255,
      "step": 204
    },
    {
      "epoch": 0.032338210356114686,
      "grad_norm": 13.404995918273926,
      "learning_rate": 9.87602640476574e-07,
      "loss": 0.2258,
      "step": 205
    },
    {
      "epoch": 0.03249595772370548,
      "grad_norm": 6.111392498016357,
      "learning_rate": 9.874416358074384e-07,
      "loss": 0.1813,
      "step": 206
    },
    {
      "epoch": 0.03265370509129629,
      "grad_norm": 6.81652307510376,
      "learning_rate": 9.87280631138303e-07,
      "loss": 0.3109,
      "step": 207
    },
    {
      "epoch": 0.03281145245888709,
      "grad_norm": 6.360020637512207,
      "learning_rate": 9.871196264691676e-07,
      "loss": 0.23,
      "step": 208
    },
    {
      "epoch": 0.0329691998264779,
      "grad_norm": 9.0701904296875,
      "learning_rate": 9.869586218000321e-07,
      "loss": 0.2352,
      "step": 209
    },
    {
      "epoch": 0.0331269471940687,
      "grad_norm": 5.986306667327881,
      "learning_rate": 9.867976171308966e-07,
      "loss": 0.2041,
      "step": 210
    },
    {
      "epoch": 0.0332846945616595,
      "grad_norm": 8.649664878845215,
      "learning_rate": 9.866366124617613e-07,
      "loss": 0.2144,
      "step": 211
    },
    {
      "epoch": 0.033442441929250305,
      "grad_norm": 8.318222045898438,
      "learning_rate": 9.86475607792626e-07,
      "loss": 0.2599,
      "step": 212
    },
    {
      "epoch": 0.03360018929684111,
      "grad_norm": 4.279948711395264,
      "learning_rate": 9.863146031234905e-07,
      "loss": 0.2252,
      "step": 213
    },
    {
      "epoch": 0.033757936664431915,
      "grad_norm": 9.401074409484863,
      "learning_rate": 9.86153598454355e-07,
      "loss": 0.2168,
      "step": 214
    },
    {
      "epoch": 0.03391568403202271,
      "grad_norm": 8.848161697387695,
      "learning_rate": 9.859925937852198e-07,
      "loss": 0.2156,
      "step": 215
    },
    {
      "epoch": 0.03407343139961352,
      "grad_norm": 6.798628807067871,
      "learning_rate": 9.858315891160843e-07,
      "loss": 0.2374,
      "step": 216
    },
    {
      "epoch": 0.03423117876720432,
      "grad_norm": 8.470605850219727,
      "learning_rate": 9.85670584446949e-07,
      "loss": 0.1854,
      "step": 217
    },
    {
      "epoch": 0.03438892613479513,
      "grad_norm": 5.13956880569458,
      "learning_rate": 9.855095797778135e-07,
      "loss": 0.1908,
      "step": 218
    },
    {
      "epoch": 0.03454667350238593,
      "grad_norm": 6.615530967712402,
      "learning_rate": 9.853485751086782e-07,
      "loss": 0.1661,
      "step": 219
    },
    {
      "epoch": 0.03470442086997673,
      "grad_norm": 9.583069801330566,
      "learning_rate": 9.851875704395427e-07,
      "loss": 0.2093,
      "step": 220
    },
    {
      "epoch": 0.034862168237567534,
      "grad_norm": 6.675043106079102,
      "learning_rate": 9.850265657704074e-07,
      "loss": 0.1603,
      "step": 221
    },
    {
      "epoch": 0.03501991560515834,
      "grad_norm": 7.040927886962891,
      "learning_rate": 9.848655611012719e-07,
      "loss": 0.2393,
      "step": 222
    },
    {
      "epoch": 0.035177662972749144,
      "grad_norm": 5.56865930557251,
      "learning_rate": 9.847045564321364e-07,
      "loss": 0.2089,
      "step": 223
    },
    {
      "epoch": 0.03533541034033995,
      "grad_norm": 8.290128707885742,
      "learning_rate": 9.84543551763001e-07,
      "loss": 0.1896,
      "step": 224
    },
    {
      "epoch": 0.035493157707930746,
      "grad_norm": 6.122547626495361,
      "learning_rate": 9.843825470938656e-07,
      "loss": 0.2031,
      "step": 225
    },
    {
      "epoch": 0.03565090507552155,
      "grad_norm": 6.4090094566345215,
      "learning_rate": 9.842215424247303e-07,
      "loss": 0.1658,
      "step": 226
    },
    {
      "epoch": 0.035808652443112356,
      "grad_norm": 7.354506969451904,
      "learning_rate": 9.840605377555948e-07,
      "loss": 0.1616,
      "step": 227
    },
    {
      "epoch": 0.03596639981070316,
      "grad_norm": 4.053621768951416,
      "learning_rate": 9.838995330864595e-07,
      "loss": 0.1381,
      "step": 228
    },
    {
      "epoch": 0.036124147178293965,
      "grad_norm": 8.635991096496582,
      "learning_rate": 9.83738528417324e-07,
      "loss": 0.148,
      "step": 229
    },
    {
      "epoch": 0.03628189454588476,
      "grad_norm": 5.523995399475098,
      "learning_rate": 9.835775237481887e-07,
      "loss": 0.1908,
      "step": 230
    },
    {
      "epoch": 0.03643964191347557,
      "grad_norm": 10.925482749938965,
      "learning_rate": 9.834165190790532e-07,
      "loss": 0.1971,
      "step": 231
    },
    {
      "epoch": 0.03659738928106637,
      "grad_norm": 7.450096607208252,
      "learning_rate": 9.832555144099177e-07,
      "loss": 0.1653,
      "step": 232
    },
    {
      "epoch": 0.03675513664865718,
      "grad_norm": 7.926517486572266,
      "learning_rate": 9.830945097407825e-07,
      "loss": 0.1813,
      "step": 233
    },
    {
      "epoch": 0.03691288401624798,
      "grad_norm": 16.86855125427246,
      "learning_rate": 9.829335050716472e-07,
      "loss": 0.153,
      "step": 234
    },
    {
      "epoch": 0.03707063138383878,
      "grad_norm": 4.815715312957764,
      "learning_rate": 9.827725004025117e-07,
      "loss": 0.2121,
      "step": 235
    },
    {
      "epoch": 0.037228378751429585,
      "grad_norm": 5.170910358428955,
      "learning_rate": 9.826114957333762e-07,
      "loss": 0.1429,
      "step": 236
    },
    {
      "epoch": 0.03738612611902039,
      "grad_norm": 5.570741653442383,
      "learning_rate": 9.824504910642409e-07,
      "loss": 0.1527,
      "step": 237
    },
    {
      "epoch": 0.037543873486611194,
      "grad_norm": 8.256614685058594,
      "learning_rate": 9.822894863951054e-07,
      "loss": 0.1797,
      "step": 238
    },
    {
      "epoch": 0.03770162085420199,
      "grad_norm": 4.869175910949707,
      "learning_rate": 9.821284817259699e-07,
      "loss": 0.1974,
      "step": 239
    },
    {
      "epoch": 0.0378593682217928,
      "grad_norm": 4.715397834777832,
      "learning_rate": 9.819674770568346e-07,
      "loss": 0.1887,
      "step": 240
    },
    {
      "epoch": 0.0380171155893836,
      "grad_norm": 7.1969404220581055,
      "learning_rate": 9.818064723876993e-07,
      "loss": 0.1525,
      "step": 241
    },
    {
      "epoch": 0.038174862956974406,
      "grad_norm": 5.895638942718506,
      "learning_rate": 9.816454677185638e-07,
      "loss": 0.1723,
      "step": 242
    },
    {
      "epoch": 0.03833261032456521,
      "grad_norm": 5.916856288909912,
      "learning_rate": 9.814844630494283e-07,
      "loss": 0.1762,
      "step": 243
    },
    {
      "epoch": 0.03849035769215601,
      "grad_norm": 7.5934367179870605,
      "learning_rate": 9.81323458380293e-07,
      "loss": 0.1887,
      "step": 244
    },
    {
      "epoch": 0.038648105059746814,
      "grad_norm": 6.210063934326172,
      "learning_rate": 9.811624537111575e-07,
      "loss": 0.1499,
      "step": 245
    },
    {
      "epoch": 0.03880585242733762,
      "grad_norm": 4.598802089691162,
      "learning_rate": 9.810014490420222e-07,
      "loss": 0.1259,
      "step": 246
    },
    {
      "epoch": 0.03896359979492842,
      "grad_norm": 12.598335266113281,
      "learning_rate": 9.808404443728867e-07,
      "loss": 0.2052,
      "step": 247
    },
    {
      "epoch": 0.03912134716251923,
      "grad_norm": 7.708802700042725,
      "learning_rate": 9.806794397037512e-07,
      "loss": 0.2005,
      "step": 248
    },
    {
      "epoch": 0.039279094530110026,
      "grad_norm": 6.15869665145874,
      "learning_rate": 9.80518435034616e-07,
      "loss": 0.1366,
      "step": 249
    },
    {
      "epoch": 0.03943684189770083,
      "grad_norm": 5.237630367279053,
      "learning_rate": 9.803574303654807e-07,
      "loss": 0.1429,
      "step": 250
    },
    {
      "epoch": 0.039594589265291635,
      "grad_norm": 4.9005842208862305,
      "learning_rate": 9.801964256963452e-07,
      "loss": 0.1147,
      "step": 251
    },
    {
      "epoch": 0.03975233663288244,
      "grad_norm": 129.9768524169922,
      "learning_rate": 9.800354210272097e-07,
      "loss": 0.1916,
      "step": 252
    },
    {
      "epoch": 0.039910084000473245,
      "grad_norm": 6.41878604888916,
      "learning_rate": 9.798744163580744e-07,
      "loss": 0.1334,
      "step": 253
    },
    {
      "epoch": 0.04006783136806404,
      "grad_norm": 6.262964725494385,
      "learning_rate": 9.797134116889389e-07,
      "loss": 0.1779,
      "step": 254
    },
    {
      "epoch": 0.04022557873565485,
      "grad_norm": 6.25661563873291,
      "learning_rate": 9.795524070198036e-07,
      "loss": 0.1479,
      "step": 255
    },
    {
      "epoch": 0.04038332610324565,
      "grad_norm": 5.191387176513672,
      "learning_rate": 9.79391402350668e-07,
      "loss": 0.133,
      "step": 256
    },
    {
      "epoch": 0.04054107347083646,
      "grad_norm": 7.769615650177002,
      "learning_rate": 9.792303976815328e-07,
      "loss": 0.1854,
      "step": 257
    },
    {
      "epoch": 0.04069882083842726,
      "grad_norm": 7.134732723236084,
      "learning_rate": 9.790693930123973e-07,
      "loss": 0.1926,
      "step": 258
    },
    {
      "epoch": 0.04085656820601806,
      "grad_norm": 5.413559436798096,
      "learning_rate": 9.78908388343262e-07,
      "loss": 0.1899,
      "step": 259
    },
    {
      "epoch": 0.041014315573608864,
      "grad_norm": 4.809391975402832,
      "learning_rate": 9.787473836741265e-07,
      "loss": 0.1293,
      "step": 260
    },
    {
      "epoch": 0.04117206294119967,
      "grad_norm": 5.961005687713623,
      "learning_rate": 9.78586379004991e-07,
      "loss": 0.1242,
      "step": 261
    },
    {
      "epoch": 0.041329810308790474,
      "grad_norm": 6.710011959075928,
      "learning_rate": 9.784253743358557e-07,
      "loss": 0.2335,
      "step": 262
    },
    {
      "epoch": 0.04148755767638128,
      "grad_norm": 8.22215747833252,
      "learning_rate": 9.782643696667202e-07,
      "loss": 0.2066,
      "step": 263
    },
    {
      "epoch": 0.041645305043972076,
      "grad_norm": 7.861912250518799,
      "learning_rate": 9.78103364997585e-07,
      "loss": 0.148,
      "step": 264
    },
    {
      "epoch": 0.04180305241156288,
      "grad_norm": 7.59419584274292,
      "learning_rate": 9.779423603284494e-07,
      "loss": 0.1698,
      "step": 265
    },
    {
      "epoch": 0.041960799779153686,
      "grad_norm": 6.258653163909912,
      "learning_rate": 9.777813556593141e-07,
      "loss": 0.2071,
      "step": 266
    },
    {
      "epoch": 0.04211854714674449,
      "grad_norm": 9.125118255615234,
      "learning_rate": 9.776203509901786e-07,
      "loss": 0.1292,
      "step": 267
    },
    {
      "epoch": 0.04227629451433529,
      "grad_norm": 6.597840309143066,
      "learning_rate": 9.774593463210434e-07,
      "loss": 0.1656,
      "step": 268
    },
    {
      "epoch": 0.04243404188192609,
      "grad_norm": 5.397624492645264,
      "learning_rate": 9.772983416519079e-07,
      "loss": 0.1225,
      "step": 269
    },
    {
      "epoch": 0.0425917892495169,
      "grad_norm": 5.821804523468018,
      "learning_rate": 9.771373369827724e-07,
      "loss": 0.1419,
      "step": 270
    },
    {
      "epoch": 0.0427495366171077,
      "grad_norm": 8.308530807495117,
      "learning_rate": 9.76976332313637e-07,
      "loss": 0.1593,
      "step": 271
    },
    {
      "epoch": 0.04290728398469851,
      "grad_norm": 6.071272373199463,
      "learning_rate": 9.768153276445018e-07,
      "loss": 0.1264,
      "step": 272
    },
    {
      "epoch": 0.043065031352289306,
      "grad_norm": 5.041172027587891,
      "learning_rate": 9.766543229753663e-07,
      "loss": 0.1934,
      "step": 273
    },
    {
      "epoch": 0.04322277871988011,
      "grad_norm": 6.917977809906006,
      "learning_rate": 9.764933183062308e-07,
      "loss": 0.0926,
      "step": 274
    },
    {
      "epoch": 0.043380526087470915,
      "grad_norm": 8.757829666137695,
      "learning_rate": 9.763323136370955e-07,
      "loss": 0.1772,
      "step": 275
    },
    {
      "epoch": 0.04353827345506172,
      "grad_norm": 4.275970458984375,
      "learning_rate": 9.7617130896796e-07,
      "loss": 0.1311,
      "step": 276
    },
    {
      "epoch": 0.043696020822652525,
      "grad_norm": 3.909147262573242,
      "learning_rate": 9.760103042988245e-07,
      "loss": 0.1522,
      "step": 277
    },
    {
      "epoch": 0.04385376819024332,
      "grad_norm": 9.127202987670898,
      "learning_rate": 9.758492996296892e-07,
      "loss": 0.1836,
      "step": 278
    },
    {
      "epoch": 0.04401151555783413,
      "grad_norm": 59.553279876708984,
      "learning_rate": 9.75688294960554e-07,
      "loss": 0.1497,
      "step": 279
    },
    {
      "epoch": 0.04416926292542493,
      "grad_norm": 8.238790512084961,
      "learning_rate": 9.755272902914184e-07,
      "loss": 0.1657,
      "step": 280
    },
    {
      "epoch": 0.04432701029301574,
      "grad_norm": 6.173616409301758,
      "learning_rate": 9.75366285622283e-07,
      "loss": 0.1609,
      "step": 281
    },
    {
      "epoch": 0.04448475766060654,
      "grad_norm": 4.654531955718994,
      "learning_rate": 9.752052809531476e-07,
      "loss": 0.115,
      "step": 282
    },
    {
      "epoch": 0.04464250502819734,
      "grad_norm": 3.280255079269409,
      "learning_rate": 9.750442762840121e-07,
      "loss": 0.0819,
      "step": 283
    },
    {
      "epoch": 0.044800252395788144,
      "grad_norm": 14.191688537597656,
      "learning_rate": 9.748832716148768e-07,
      "loss": 0.1864,
      "step": 284
    },
    {
      "epoch": 0.04495799976337895,
      "grad_norm": 8.970497131347656,
      "learning_rate": 9.747222669457413e-07,
      "loss": 0.1204,
      "step": 285
    },
    {
      "epoch": 0.045115747130969754,
      "grad_norm": 3.3083322048187256,
      "learning_rate": 9.74561262276606e-07,
      "loss": 0.1487,
      "step": 286
    },
    {
      "epoch": 0.04527349449856056,
      "grad_norm": 10.429438591003418,
      "learning_rate": 9.744002576074706e-07,
      "loss": 0.1775,
      "step": 287
    },
    {
      "epoch": 0.045431241866151356,
      "grad_norm": 5.250263690948486,
      "learning_rate": 9.742392529383353e-07,
      "loss": 0.165,
      "step": 288
    },
    {
      "epoch": 0.04558898923374216,
      "grad_norm": 20.89002227783203,
      "learning_rate": 9.740782482691998e-07,
      "loss": 0.2032,
      "step": 289
    },
    {
      "epoch": 0.045746736601332966,
      "grad_norm": 5.709809303283691,
      "learning_rate": 9.739172436000643e-07,
      "loss": 0.145,
      "step": 290
    },
    {
      "epoch": 0.04590448396892377,
      "grad_norm": 13.901089668273926,
      "learning_rate": 9.73756238930929e-07,
      "loss": 0.0993,
      "step": 291
    },
    {
      "epoch": 0.046062231336514575,
      "grad_norm": 9.392309188842773,
      "learning_rate": 9.735952342617935e-07,
      "loss": 0.1707,
      "step": 292
    },
    {
      "epoch": 0.04621997870410537,
      "grad_norm": 11.4054594039917,
      "learning_rate": 9.734342295926582e-07,
      "loss": 0.1482,
      "step": 293
    },
    {
      "epoch": 0.04637772607169618,
      "grad_norm": 4.401249885559082,
      "learning_rate": 9.732732249235227e-07,
      "loss": 0.1087,
      "step": 294
    },
    {
      "epoch": 0.04653547343928698,
      "grad_norm": 3.186814069747925,
      "learning_rate": 9.731122202543874e-07,
      "loss": 0.1321,
      "step": 295
    },
    {
      "epoch": 0.04669322080687779,
      "grad_norm": 3.693132162094116,
      "learning_rate": 9.72951215585252e-07,
      "loss": 0.1089,
      "step": 296
    },
    {
      "epoch": 0.046850968174468585,
      "grad_norm": 8.566479682922363,
      "learning_rate": 9.727902109161166e-07,
      "loss": 0.1497,
      "step": 297
    },
    {
      "epoch": 0.04700871554205939,
      "grad_norm": 4.2073211669921875,
      "learning_rate": 9.726292062469811e-07,
      "loss": 0.1217,
      "step": 298
    },
    {
      "epoch": 0.047166462909650195,
      "grad_norm": 6.404031276702881,
      "learning_rate": 9.724682015778456e-07,
      "loss": 0.1168,
      "step": 299
    },
    {
      "epoch": 0.047324210277241,
      "grad_norm": 9.464739799499512,
      "learning_rate": 9.723071969087103e-07,
      "loss": 0.1161,
      "step": 300
    },
    {
      "epoch": 0.047481957644831804,
      "grad_norm": 5.61821985244751,
      "learning_rate": 9.72146192239575e-07,
      "loss": 0.1246,
      "step": 301
    },
    {
      "epoch": 0.0476397050124226,
      "grad_norm": 7.7798991203308105,
      "learning_rate": 9.719851875704395e-07,
      "loss": 0.1115,
      "step": 302
    },
    {
      "epoch": 0.04779745238001341,
      "grad_norm": 4.6815290451049805,
      "learning_rate": 9.71824182901304e-07,
      "loss": 0.1037,
      "step": 303
    },
    {
      "epoch": 0.04795519974760421,
      "grad_norm": 4.269176006317139,
      "learning_rate": 9.716631782321688e-07,
      "loss": 0.1384,
      "step": 304
    },
    {
      "epoch": 0.048112947115195016,
      "grad_norm": 7.13702917098999,
      "learning_rate": 9.715021735630333e-07,
      "loss": 0.1228,
      "step": 305
    },
    {
      "epoch": 0.04827069448278582,
      "grad_norm": 4.8622727394104,
      "learning_rate": 9.713411688938978e-07,
      "loss": 0.1779,
      "step": 306
    },
    {
      "epoch": 0.04842844185037662,
      "grad_norm": 4.0259013175964355,
      "learning_rate": 9.711801642247625e-07,
      "loss": 0.1285,
      "step": 307
    },
    {
      "epoch": 0.048586189217967424,
      "grad_norm": 11.432190895080566,
      "learning_rate": 9.71019159555627e-07,
      "loss": 0.1798,
      "step": 308
    },
    {
      "epoch": 0.04874393658555823,
      "grad_norm": 10.696998596191406,
      "learning_rate": 9.708581548864917e-07,
      "loss": 0.1703,
      "step": 309
    },
    {
      "epoch": 0.04890168395314903,
      "grad_norm": 7.325720310211182,
      "learning_rate": 9.706971502173562e-07,
      "loss": 0.1329,
      "step": 310
    },
    {
      "epoch": 0.04905943132073984,
      "grad_norm": 4.719336032867432,
      "learning_rate": 9.70536145548221e-07,
      "loss": 0.0911,
      "step": 311
    },
    {
      "epoch": 0.049217178688330636,
      "grad_norm": 3.4688565731048584,
      "learning_rate": 9.703751408790854e-07,
      "loss": 0.0969,
      "step": 312
    },
    {
      "epoch": 0.04937492605592144,
      "grad_norm": 5.5488433837890625,
      "learning_rate": 9.702141362099501e-07,
      "loss": 0.1138,
      "step": 313
    },
    {
      "epoch": 0.049532673423512245,
      "grad_norm": 10.009937286376953,
      "learning_rate": 9.700531315408146e-07,
      "loss": 0.1413,
      "step": 314
    },
    {
      "epoch": 0.04969042079110305,
      "grad_norm": 6.785848617553711,
      "learning_rate": 9.698921268716791e-07,
      "loss": 0.1184,
      "step": 315
    },
    {
      "epoch": 0.049848168158693855,
      "grad_norm": 15.140206336975098,
      "learning_rate": 9.697311222025438e-07,
      "loss": 0.1195,
      "step": 316
    },
    {
      "epoch": 0.05000591552628465,
      "grad_norm": 7.2620930671691895,
      "learning_rate": 9.695701175334085e-07,
      "loss": 0.1006,
      "step": 317
    },
    {
      "epoch": 0.05016366289387546,
      "grad_norm": 6.647378444671631,
      "learning_rate": 9.69409112864273e-07,
      "loss": 0.0997,
      "step": 318
    },
    {
      "epoch": 0.05032141026146626,
      "grad_norm": 3.3054749965667725,
      "learning_rate": 9.692481081951375e-07,
      "loss": 0.0707,
      "step": 319
    },
    {
      "epoch": 0.05047915762905707,
      "grad_norm": 8.351949691772461,
      "learning_rate": 9.690871035260022e-07,
      "loss": 0.1555,
      "step": 320
    },
    {
      "epoch": 0.05063690499664787,
      "grad_norm": 6.779848575592041,
      "learning_rate": 9.689260988568667e-07,
      "loss": 0.0918,
      "step": 321
    },
    {
      "epoch": 0.05079465236423867,
      "grad_norm": 4.860285758972168,
      "learning_rate": 9.687650941877315e-07,
      "loss": 0.1497,
      "step": 322
    },
    {
      "epoch": 0.050952399731829474,
      "grad_norm": 9.101405143737793,
      "learning_rate": 9.68604089518596e-07,
      "loss": 0.0948,
      "step": 323
    },
    {
      "epoch": 0.05111014709942028,
      "grad_norm": 7.681159019470215,
      "learning_rate": 9.684430848494607e-07,
      "loss": 0.1159,
      "step": 324
    },
    {
      "epoch": 0.051267894467011084,
      "grad_norm": 11.528342247009277,
      "learning_rate": 9.682820801803252e-07,
      "loss": 0.186,
      "step": 325
    },
    {
      "epoch": 0.05142564183460188,
      "grad_norm": 9.96857738494873,
      "learning_rate": 9.681210755111899e-07,
      "loss": 0.1376,
      "step": 326
    },
    {
      "epoch": 0.051583389202192687,
      "grad_norm": 4.946990966796875,
      "learning_rate": 9.679600708420544e-07,
      "loss": 0.1328,
      "step": 327
    },
    {
      "epoch": 0.05174113656978349,
      "grad_norm": 7.803967475891113,
      "learning_rate": 9.677990661729189e-07,
      "loss": 0.1327,
      "step": 328
    },
    {
      "epoch": 0.051898883937374296,
      "grad_norm": 7.988615036010742,
      "learning_rate": 9.676380615037836e-07,
      "loss": 0.1666,
      "step": 329
    },
    {
      "epoch": 0.0520566313049651,
      "grad_norm": 5.772839546203613,
      "learning_rate": 9.67477056834648e-07,
      "loss": 0.0978,
      "step": 330
    },
    {
      "epoch": 0.0522143786725559,
      "grad_norm": 10.387200355529785,
      "learning_rate": 9.673160521655128e-07,
      "loss": 0.1152,
      "step": 331
    },
    {
      "epoch": 0.0523721260401467,
      "grad_norm": 5.836713790893555,
      "learning_rate": 9.671550474963773e-07,
      "loss": 0.0987,
      "step": 332
    },
    {
      "epoch": 0.05252987340773751,
      "grad_norm": 9.30050277709961,
      "learning_rate": 9.66994042827242e-07,
      "loss": 0.1603,
      "step": 333
    },
    {
      "epoch": 0.05268762077532831,
      "grad_norm": 7.9058451652526855,
      "learning_rate": 9.668330381581065e-07,
      "loss": 0.1947,
      "step": 334
    },
    {
      "epoch": 0.05284536814291912,
      "grad_norm": 7.316274642944336,
      "learning_rate": 9.66672033488971e-07,
      "loss": 0.1462,
      "step": 335
    },
    {
      "epoch": 0.053003115510509916,
      "grad_norm": 4.279013156890869,
      "learning_rate": 9.665110288198357e-07,
      "loss": 0.1366,
      "step": 336
    },
    {
      "epoch": 0.05316086287810072,
      "grad_norm": 10.732933044433594,
      "learning_rate": 9.663500241507002e-07,
      "loss": 0.1455,
      "step": 337
    },
    {
      "epoch": 0.053318610245691525,
      "grad_norm": 3.3139238357543945,
      "learning_rate": 9.66189019481565e-07,
      "loss": 0.0694,
      "step": 338
    },
    {
      "epoch": 0.05347635761328233,
      "grad_norm": 6.18437385559082,
      "learning_rate": 9.660280148124297e-07,
      "loss": 0.082,
      "step": 339
    },
    {
      "epoch": 0.053634104980873135,
      "grad_norm": 12.412300109863281,
      "learning_rate": 9.658670101432942e-07,
      "loss": 0.1503,
      "step": 340
    },
    {
      "epoch": 0.05379185234846393,
      "grad_norm": 6.401512145996094,
      "learning_rate": 9.657060054741587e-07,
      "loss": 0.0825,
      "step": 341
    },
    {
      "epoch": 0.05394959971605474,
      "grad_norm": 7.424494743347168,
      "learning_rate": 9.655450008050234e-07,
      "loss": 0.0997,
      "step": 342
    },
    {
      "epoch": 0.05410734708364554,
      "grad_norm": 9.988298416137695,
      "learning_rate": 9.653839961358879e-07,
      "loss": 0.2279,
      "step": 343
    },
    {
      "epoch": 0.05426509445123635,
      "grad_norm": 10.520110130310059,
      "learning_rate": 9.652229914667524e-07,
      "loss": 0.1491,
      "step": 344
    },
    {
      "epoch": 0.05442284181882715,
      "grad_norm": 5.183448314666748,
      "learning_rate": 9.65061986797617e-07,
      "loss": 0.2024,
      "step": 345
    },
    {
      "epoch": 0.05458058918641795,
      "grad_norm": 4.082580089569092,
      "learning_rate": 9.649009821284818e-07,
      "loss": 0.1181,
      "step": 346
    },
    {
      "epoch": 0.054738336554008754,
      "grad_norm": 10.461071968078613,
      "learning_rate": 9.647399774593463e-07,
      "loss": 0.1077,
      "step": 347
    },
    {
      "epoch": 0.05489608392159956,
      "grad_norm": 4.259407997131348,
      "learning_rate": 9.645789727902108e-07,
      "loss": 0.1165,
      "step": 348
    },
    {
      "epoch": 0.055053831289190364,
      "grad_norm": 7.56532621383667,
      "learning_rate": 9.644179681210755e-07,
      "loss": 0.1025,
      "step": 349
    },
    {
      "epoch": 0.05521157865678117,
      "grad_norm": 4.267522811889648,
      "learning_rate": 9.6425696345194e-07,
      "loss": 0.1156,
      "step": 350
    },
    {
      "epoch": 0.055369326024371966,
      "grad_norm": 4.609867095947266,
      "learning_rate": 9.640959587828047e-07,
      "loss": 0.1236,
      "step": 351
    },
    {
      "epoch": 0.05552707339196277,
      "grad_norm": 12.157086372375488,
      "learning_rate": 9.639349541136692e-07,
      "loss": 0.1221,
      "step": 352
    },
    {
      "epoch": 0.055684820759553576,
      "grad_norm": 9.895761489868164,
      "learning_rate": 9.637739494445337e-07,
      "loss": 0.1346,
      "step": 353
    },
    {
      "epoch": 0.05584256812714438,
      "grad_norm": 10.083800315856934,
      "learning_rate": 9.636129447753984e-07,
      "loss": 0.1094,
      "step": 354
    },
    {
      "epoch": 0.05600031549473518,
      "grad_norm": 5.506272315979004,
      "learning_rate": 9.634519401062632e-07,
      "loss": 0.0883,
      "step": 355
    },
    {
      "epoch": 0.05615806286232598,
      "grad_norm": 7.488781452178955,
      "learning_rate": 9.632909354371277e-07,
      "loss": 0.1136,
      "step": 356
    },
    {
      "epoch": 0.05631581022991679,
      "grad_norm": 5.5719146728515625,
      "learning_rate": 9.631299307679922e-07,
      "loss": 0.1914,
      "step": 357
    },
    {
      "epoch": 0.05647355759750759,
      "grad_norm": 11.344675064086914,
      "learning_rate": 9.629689260988569e-07,
      "loss": 0.1,
      "step": 358
    },
    {
      "epoch": 0.0566313049650984,
      "grad_norm": 12.2915620803833,
      "learning_rate": 9.628079214297214e-07,
      "loss": 0.1473,
      "step": 359
    },
    {
      "epoch": 0.056789052332689195,
      "grad_norm": 7.278500080108643,
      "learning_rate": 9.626469167605859e-07,
      "loss": 0.1206,
      "step": 360
    },
    {
      "epoch": 0.05694679970028,
      "grad_norm": 10.03773307800293,
      "learning_rate": 9.624859120914506e-07,
      "loss": 0.1689,
      "step": 361
    },
    {
      "epoch": 0.057104547067870805,
      "grad_norm": 7.41524600982666,
      "learning_rate": 9.623249074223153e-07,
      "loss": 0.0993,
      "step": 362
    },
    {
      "epoch": 0.05726229443546161,
      "grad_norm": 8.416111946105957,
      "learning_rate": 9.621639027531798e-07,
      "loss": 0.1379,
      "step": 363
    },
    {
      "epoch": 0.057420041803052414,
      "grad_norm": 5.612109184265137,
      "learning_rate": 9.620028980840445e-07,
      "loss": 0.0697,
      "step": 364
    },
    {
      "epoch": 0.05757778917064321,
      "grad_norm": 5.46721887588501,
      "learning_rate": 9.61841893414909e-07,
      "loss": 0.0799,
      "step": 365
    },
    {
      "epoch": 0.05773553653823402,
      "grad_norm": 8.063959121704102,
      "learning_rate": 9.616808887457735e-07,
      "loss": 0.073,
      "step": 366
    },
    {
      "epoch": 0.05789328390582482,
      "grad_norm": 7.9931559562683105,
      "learning_rate": 9.615198840766382e-07,
      "loss": 0.1438,
      "step": 367
    },
    {
      "epoch": 0.058051031273415626,
      "grad_norm": 5.741889953613281,
      "learning_rate": 9.613588794075027e-07,
      "loss": 0.0832,
      "step": 368
    },
    {
      "epoch": 0.05820877864100643,
      "grad_norm": 4.304195880889893,
      "learning_rate": 9.611978747383674e-07,
      "loss": 0.1258,
      "step": 369
    },
    {
      "epoch": 0.05836652600859723,
      "grad_norm": 9.83571720123291,
      "learning_rate": 9.61036870069232e-07,
      "loss": 0.1328,
      "step": 370
    },
    {
      "epoch": 0.058524273376188034,
      "grad_norm": 4.696883201599121,
      "learning_rate": 9.608758654000966e-07,
      "loss": 0.1282,
      "step": 371
    },
    {
      "epoch": 0.05868202074377884,
      "grad_norm": 5.337697505950928,
      "learning_rate": 9.607148607309611e-07,
      "loss": 0.1203,
      "step": 372
    },
    {
      "epoch": 0.05883976811136964,
      "grad_norm": 6.526247501373291,
      "learning_rate": 9.605538560618256e-07,
      "loss": 0.1294,
      "step": 373
    },
    {
      "epoch": 0.05899751547896045,
      "grad_norm": 4.350911617279053,
      "learning_rate": 9.603928513926904e-07,
      "loss": 0.1189,
      "step": 374
    },
    {
      "epoch": 0.059155262846551246,
      "grad_norm": 6.42826509475708,
      "learning_rate": 9.602318467235549e-07,
      "loss": 0.0815,
      "step": 375
    },
    {
      "epoch": 0.05931301021414205,
      "grad_norm": 7.661411285400391,
      "learning_rate": 9.600708420544196e-07,
      "loss": 0.0944,
      "step": 376
    },
    {
      "epoch": 0.059470757581732855,
      "grad_norm": 5.3927764892578125,
      "learning_rate": 9.59909837385284e-07,
      "loss": 0.0781,
      "step": 377
    },
    {
      "epoch": 0.05962850494932366,
      "grad_norm": 5.518739223480225,
      "learning_rate": 9.597488327161488e-07,
      "loss": 0.0927,
      "step": 378
    },
    {
      "epoch": 0.059786252316914465,
      "grad_norm": 5.976505756378174,
      "learning_rate": 9.595878280470133e-07,
      "loss": 0.1466,
      "step": 379
    },
    {
      "epoch": 0.05994399968450526,
      "grad_norm": 5.26506233215332,
      "learning_rate": 9.59426823377878e-07,
      "loss": 0.1124,
      "step": 380
    },
    {
      "epoch": 0.06010174705209607,
      "grad_norm": 4.402028560638428,
      "learning_rate": 9.592658187087425e-07,
      "loss": 0.0617,
      "step": 381
    },
    {
      "epoch": 0.06025949441968687,
      "grad_norm": 5.284600257873535,
      "learning_rate": 9.59104814039607e-07,
      "loss": 0.0855,
      "step": 382
    },
    {
      "epoch": 0.06041724178727768,
      "grad_norm": 3.978062391281128,
      "learning_rate": 9.589438093704717e-07,
      "loss": 0.1116,
      "step": 383
    },
    {
      "epoch": 0.060574989154868475,
      "grad_norm": 4.164371490478516,
      "learning_rate": 9.587828047013364e-07,
      "loss": 0.1204,
      "step": 384
    },
    {
      "epoch": 0.06073273652245928,
      "grad_norm": 7.169449806213379,
      "learning_rate": 9.58621800032201e-07,
      "loss": 0.1236,
      "step": 385
    },
    {
      "epoch": 0.060890483890050084,
      "grad_norm": 6.995155334472656,
      "learning_rate": 9.584607953630654e-07,
      "loss": 0.1152,
      "step": 386
    },
    {
      "epoch": 0.06104823125764089,
      "grad_norm": 8.239535331726074,
      "learning_rate": 9.582997906939301e-07,
      "loss": 0.0693,
      "step": 387
    },
    {
      "epoch": 0.061205978625231694,
      "grad_norm": 4.845700263977051,
      "learning_rate": 9.581387860247946e-07,
      "loss": 0.1102,
      "step": 388
    },
    {
      "epoch": 0.06136372599282249,
      "grad_norm": 5.36713981628418,
      "learning_rate": 9.579777813556593e-07,
      "loss": 0.0923,
      "step": 389
    },
    {
      "epoch": 0.0615214733604133,
      "grad_norm": 7.444408416748047,
      "learning_rate": 9.578167766865238e-07,
      "loss": 0.1109,
      "step": 390
    },
    {
      "epoch": 0.0616792207280041,
      "grad_norm": 5.161166191101074,
      "learning_rate": 9.576557720173886e-07,
      "loss": 0.101,
      "step": 391
    },
    {
      "epoch": 0.061836968095594906,
      "grad_norm": 6.004199028015137,
      "learning_rate": 9.57494767348253e-07,
      "loss": 0.1083,
      "step": 392
    },
    {
      "epoch": 0.06199471546318571,
      "grad_norm": 9.09014892578125,
      "learning_rate": 9.573337626791178e-07,
      "loss": 0.1364,
      "step": 393
    },
    {
      "epoch": 0.06215246283077651,
      "grad_norm": 4.672868251800537,
      "learning_rate": 9.571727580099823e-07,
      "loss": 0.1331,
      "step": 394
    },
    {
      "epoch": 0.06231021019836731,
      "grad_norm": 8.290534019470215,
      "learning_rate": 9.570117533408468e-07,
      "loss": 0.1124,
      "step": 395
    },
    {
      "epoch": 0.06246795756595812,
      "grad_norm": 4.446094989776611,
      "learning_rate": 9.568507486717115e-07,
      "loss": 0.1108,
      "step": 396
    },
    {
      "epoch": 0.06262570493354892,
      "grad_norm": 11.480560302734375,
      "learning_rate": 9.56689744002576e-07,
      "loss": 0.1627,
      "step": 397
    },
    {
      "epoch": 0.06278345230113973,
      "grad_norm": 5.206017971038818,
      "learning_rate": 9.565287393334405e-07,
      "loss": 0.1112,
      "step": 398
    },
    {
      "epoch": 0.06294119966873053,
      "grad_norm": 7.868137836456299,
      "learning_rate": 9.563677346643052e-07,
      "loss": 0.1475,
      "step": 399
    },
    {
      "epoch": 0.06309894703632134,
      "grad_norm": 11.411048889160156,
      "learning_rate": 9.5620672999517e-07,
      "loss": 0.1333,
      "step": 400
    },
    {
      "epoch": 0.06325669440391213,
      "grad_norm": 6.512931823730469,
      "learning_rate": 9.560457253260344e-07,
      "loss": 0.1181,
      "step": 401
    },
    {
      "epoch": 0.06341444177150293,
      "grad_norm": 4.9762864112854,
      "learning_rate": 9.55884720656899e-07,
      "loss": 0.123,
      "step": 402
    },
    {
      "epoch": 0.06357218913909374,
      "grad_norm": 4.826909065246582,
      "learning_rate": 9.557237159877636e-07,
      "loss": 0.1399,
      "step": 403
    },
    {
      "epoch": 0.06372993650668454,
      "grad_norm": 23.437101364135742,
      "learning_rate": 9.555627113186281e-07,
      "loss": 0.1496,
      "step": 404
    },
    {
      "epoch": 0.06388768387427535,
      "grad_norm": 8.380960464477539,
      "learning_rate": 9.554017066494928e-07,
      "loss": 0.1632,
      "step": 405
    },
    {
      "epoch": 0.06404543124186615,
      "grad_norm": 6.791145324707031,
      "learning_rate": 9.552407019803573e-07,
      "loss": 0.1224,
      "step": 406
    },
    {
      "epoch": 0.06420317860945696,
      "grad_norm": 8.910951614379883,
      "learning_rate": 9.55079697311222e-07,
      "loss": 0.1235,
      "step": 407
    },
    {
      "epoch": 0.06436092597704776,
      "grad_norm": 6.297551155090332,
      "learning_rate": 9.549186926420865e-07,
      "loss": 0.0877,
      "step": 408
    },
    {
      "epoch": 0.06451867334463857,
      "grad_norm": 5.6142497062683105,
      "learning_rate": 9.547576879729513e-07,
      "loss": 0.0899,
      "step": 409
    },
    {
      "epoch": 0.06467642071222937,
      "grad_norm": 5.524415016174316,
      "learning_rate": 9.545966833038158e-07,
      "loss": 0.1246,
      "step": 410
    },
    {
      "epoch": 0.06483416807982016,
      "grad_norm": 6.010134220123291,
      "learning_rate": 9.544356786346803e-07,
      "loss": 0.0955,
      "step": 411
    },
    {
      "epoch": 0.06499191544741097,
      "grad_norm": 6.693892002105713,
      "learning_rate": 9.54274673965545e-07,
      "loss": 0.128,
      "step": 412
    },
    {
      "epoch": 0.06514966281500177,
      "grad_norm": 9.232756614685059,
      "learning_rate": 9.541136692964095e-07,
      "loss": 0.0881,
      "step": 413
    },
    {
      "epoch": 0.06530741018259258,
      "grad_norm": 6.193661689758301,
      "learning_rate": 9.539526646272742e-07,
      "loss": 0.1144,
      "step": 414
    },
    {
      "epoch": 0.06546515755018338,
      "grad_norm": 4.4394683837890625,
      "learning_rate": 9.537916599581387e-07,
      "loss": 0.0497,
      "step": 415
    },
    {
      "epoch": 0.06562290491777419,
      "grad_norm": 4.97087287902832,
      "learning_rate": 9.536306552890033e-07,
      "loss": 0.0984,
      "step": 416
    },
    {
      "epoch": 0.06578065228536499,
      "grad_norm": 3.874796152114868,
      "learning_rate": 9.534696506198679e-07,
      "loss": 0.1162,
      "step": 417
    },
    {
      "epoch": 0.0659383996529558,
      "grad_norm": 7.771538734436035,
      "learning_rate": 9.533086459507326e-07,
      "loss": 0.1724,
      "step": 418
    },
    {
      "epoch": 0.0660961470205466,
      "grad_norm": 5.318310260772705,
      "learning_rate": 9.531476412815971e-07,
      "loss": 0.1136,
      "step": 419
    },
    {
      "epoch": 0.0662538943881374,
      "grad_norm": 7.3136372566223145,
      "learning_rate": 9.529866366124617e-07,
      "loss": 0.104,
      "step": 420
    },
    {
      "epoch": 0.0664116417557282,
      "grad_norm": 10.768095970153809,
      "learning_rate": 9.528256319433263e-07,
      "loss": 0.125,
      "step": 421
    },
    {
      "epoch": 0.066569389123319,
      "grad_norm": 2.916339159011841,
      "learning_rate": 9.526646272741909e-07,
      "loss": 0.0639,
      "step": 422
    },
    {
      "epoch": 0.0667271364909098,
      "grad_norm": 9.273578643798828,
      "learning_rate": 9.525036226050554e-07,
      "loss": 0.11,
      "step": 423
    },
    {
      "epoch": 0.06688488385850061,
      "grad_norm": 4.831707000732422,
      "learning_rate": 9.5234261793592e-07,
      "loss": 0.1128,
      "step": 424
    },
    {
      "epoch": 0.06704263122609141,
      "grad_norm": 10.874245643615723,
      "learning_rate": 9.521816132667847e-07,
      "loss": 0.1701,
      "step": 425
    },
    {
      "epoch": 0.06720037859368222,
      "grad_norm": 5.706363201141357,
      "learning_rate": 9.520206085976494e-07,
      "loss": 0.1301,
      "step": 426
    },
    {
      "epoch": 0.06735812596127302,
      "grad_norm": 4.783963680267334,
      "learning_rate": 9.518596039285139e-07,
      "loss": 0.0934,
      "step": 427
    },
    {
      "epoch": 0.06751587332886383,
      "grad_norm": 8.126564979553223,
      "learning_rate": 9.516985992593785e-07,
      "loss": 0.1474,
      "step": 428
    },
    {
      "epoch": 0.06767362069645463,
      "grad_norm": 5.196165084838867,
      "learning_rate": 9.515375945902431e-07,
      "loss": 0.0903,
      "step": 429
    },
    {
      "epoch": 0.06783136806404542,
      "grad_norm": 15.124307632446289,
      "learning_rate": 9.513765899211077e-07,
      "loss": 0.1804,
      "step": 430
    },
    {
      "epoch": 0.06798911543163623,
      "grad_norm": 5.490636825561523,
      "learning_rate": 9.512155852519723e-07,
      "loss": 0.0729,
      "step": 431
    },
    {
      "epoch": 0.06814686279922703,
      "grad_norm": 6.503512859344482,
      "learning_rate": 9.510545805828368e-07,
      "loss": 0.1389,
      "step": 432
    },
    {
      "epoch": 0.06830461016681784,
      "grad_norm": 10.483194351196289,
      "learning_rate": 9.508935759137015e-07,
      "loss": 0.0925,
      "step": 433
    },
    {
      "epoch": 0.06846235753440864,
      "grad_norm": 6.307246208190918,
      "learning_rate": 9.507325712445661e-07,
      "loss": 0.0968,
      "step": 434
    },
    {
      "epoch": 0.06862010490199945,
      "grad_norm": 6.193094253540039,
      "learning_rate": 9.505715665754307e-07,
      "loss": 0.1141,
      "step": 435
    },
    {
      "epoch": 0.06877785226959025,
      "grad_norm": 68.03524017333984,
      "learning_rate": 9.504105619062952e-07,
      "loss": 0.1235,
      "step": 436
    },
    {
      "epoch": 0.06893559963718106,
      "grad_norm": 5.203610420227051,
      "learning_rate": 9.502495572371598e-07,
      "loss": 0.0933,
      "step": 437
    },
    {
      "epoch": 0.06909334700477186,
      "grad_norm": 7.363839626312256,
      "learning_rate": 9.500885525680244e-07,
      "loss": 0.1038,
      "step": 438
    },
    {
      "epoch": 0.06925109437236267,
      "grad_norm": 5.326133728027344,
      "learning_rate": 9.49927547898889e-07,
      "loss": 0.1036,
      "step": 439
    },
    {
      "epoch": 0.06940884173995346,
      "grad_norm": 7.377047061920166,
      "learning_rate": 9.497665432297536e-07,
      "loss": 0.0791,
      "step": 440
    },
    {
      "epoch": 0.06956658910754426,
      "grad_norm": 6.677900791168213,
      "learning_rate": 9.496055385606182e-07,
      "loss": 0.1382,
      "step": 441
    },
    {
      "epoch": 0.06972433647513507,
      "grad_norm": 4.663441181182861,
      "learning_rate": 9.494445338914828e-07,
      "loss": 0.0971,
      "step": 442
    },
    {
      "epoch": 0.06988208384272587,
      "grad_norm": 8.59867000579834,
      "learning_rate": 9.492835292223474e-07,
      "loss": 0.1171,
      "step": 443
    },
    {
      "epoch": 0.07003983121031668,
      "grad_norm": 7.579411029815674,
      "learning_rate": 9.49122524553212e-07,
      "loss": 0.1495,
      "step": 444
    },
    {
      "epoch": 0.07019757857790748,
      "grad_norm": 7.643759727478027,
      "learning_rate": 9.489615198840766e-07,
      "loss": 0.0885,
      "step": 445
    },
    {
      "epoch": 0.07035532594549829,
      "grad_norm": 9.882652282714844,
      "learning_rate": 9.488005152149412e-07,
      "loss": 0.0885,
      "step": 446
    },
    {
      "epoch": 0.07051307331308909,
      "grad_norm": 5.170772075653076,
      "learning_rate": 9.486395105458058e-07,
      "loss": 0.1332,
      "step": 447
    },
    {
      "epoch": 0.0706708206806799,
      "grad_norm": 6.967252254486084,
      "learning_rate": 9.484785058766704e-07,
      "loss": 0.0816,
      "step": 448
    },
    {
      "epoch": 0.0708285680482707,
      "grad_norm": 7.689639568328857,
      "learning_rate": 9.48317501207535e-07,
      "loss": 0.097,
      "step": 449
    },
    {
      "epoch": 0.07098631541586149,
      "grad_norm": 4.52255916595459,
      "learning_rate": 9.481564965383996e-07,
      "loss": 0.0501,
      "step": 450
    },
    {
      "epoch": 0.0711440627834523,
      "grad_norm": 6.25842809677124,
      "learning_rate": 9.479954918692642e-07,
      "loss": 0.1615,
      "step": 451
    },
    {
      "epoch": 0.0713018101510431,
      "grad_norm": 7.390439987182617,
      "learning_rate": 9.478344872001287e-07,
      "loss": 0.0939,
      "step": 452
    },
    {
      "epoch": 0.0714595575186339,
      "grad_norm": 6.982767581939697,
      "learning_rate": 9.476734825309933e-07,
      "loss": 0.057,
      "step": 453
    },
    {
      "epoch": 0.07161730488622471,
      "grad_norm": 5.735555171966553,
      "learning_rate": 9.475124778618579e-07,
      "loss": 0.0916,
      "step": 454
    },
    {
      "epoch": 0.07177505225381552,
      "grad_norm": 6.87061071395874,
      "learning_rate": 9.473514731927226e-07,
      "loss": 0.1062,
      "step": 455
    },
    {
      "epoch": 0.07193279962140632,
      "grad_norm": 4.926663875579834,
      "learning_rate": 9.471904685235872e-07,
      "loss": 0.0669,
      "step": 456
    },
    {
      "epoch": 0.07209054698899713,
      "grad_norm": 6.2761311531066895,
      "learning_rate": 9.470294638544517e-07,
      "loss": 0.1445,
      "step": 457
    },
    {
      "epoch": 0.07224829435658793,
      "grad_norm": 6.018259525299072,
      "learning_rate": 9.468684591853163e-07,
      "loss": 0.1123,
      "step": 458
    },
    {
      "epoch": 0.07240604172417872,
      "grad_norm": 6.722275257110596,
      "learning_rate": 9.467074545161809e-07,
      "loss": 0.1306,
      "step": 459
    },
    {
      "epoch": 0.07256378909176953,
      "grad_norm": 5.744132995605469,
      "learning_rate": 9.465464498470455e-07,
      "loss": 0.0828,
      "step": 460
    },
    {
      "epoch": 0.07272153645936033,
      "grad_norm": 3.327827215194702,
      "learning_rate": 9.4638544517791e-07,
      "loss": 0.0999,
      "step": 461
    },
    {
      "epoch": 0.07287928382695114,
      "grad_norm": 6.590383052825928,
      "learning_rate": 9.462244405087747e-07,
      "loss": 0.0878,
      "step": 462
    },
    {
      "epoch": 0.07303703119454194,
      "grad_norm": 4.286330223083496,
      "learning_rate": 9.460634358396394e-07,
      "loss": 0.0908,
      "step": 463
    },
    {
      "epoch": 0.07319477856213275,
      "grad_norm": 6.676567077636719,
      "learning_rate": 9.45902431170504e-07,
      "loss": 0.1406,
      "step": 464
    },
    {
      "epoch": 0.07335252592972355,
      "grad_norm": 52.79438018798828,
      "learning_rate": 9.457414265013685e-07,
      "loss": 0.1012,
      "step": 465
    },
    {
      "epoch": 0.07351027329731435,
      "grad_norm": 4.823317527770996,
      "learning_rate": 9.455804218322331e-07,
      "loss": 0.1164,
      "step": 466
    },
    {
      "epoch": 0.07366802066490516,
      "grad_norm": 4.268604755401611,
      "learning_rate": 9.454194171630977e-07,
      "loss": 0.0871,
      "step": 467
    },
    {
      "epoch": 0.07382576803249596,
      "grad_norm": 7.7923760414123535,
      "learning_rate": 9.452584124939623e-07,
      "loss": 0.0609,
      "step": 468
    },
    {
      "epoch": 0.07398351540008676,
      "grad_norm": 7.410219669342041,
      "learning_rate": 9.450974078248268e-07,
      "loss": 0.0679,
      "step": 469
    },
    {
      "epoch": 0.07414126276767756,
      "grad_norm": 4.796772480010986,
      "learning_rate": 9.449364031556915e-07,
      "loss": 0.0514,
      "step": 470
    },
    {
      "epoch": 0.07429901013526836,
      "grad_norm": 11.331911087036133,
      "learning_rate": 9.447753984865561e-07,
      "loss": 0.1235,
      "step": 471
    },
    {
      "epoch": 0.07445675750285917,
      "grad_norm": 6.432934284210205,
      "learning_rate": 9.446143938174207e-07,
      "loss": 0.1039,
      "step": 472
    },
    {
      "epoch": 0.07461450487044997,
      "grad_norm": 6.318420886993408,
      "learning_rate": 9.444533891482852e-07,
      "loss": 0.1477,
      "step": 473
    },
    {
      "epoch": 0.07477225223804078,
      "grad_norm": 12.588046073913574,
      "learning_rate": 9.442923844791498e-07,
      "loss": 0.0832,
      "step": 474
    },
    {
      "epoch": 0.07492999960563158,
      "grad_norm": 6.165432453155518,
      "learning_rate": 9.441313798100144e-07,
      "loss": 0.1009,
      "step": 475
    },
    {
      "epoch": 0.07508774697322239,
      "grad_norm": 5.55441951751709,
      "learning_rate": 9.43970375140879e-07,
      "loss": 0.0968,
      "step": 476
    },
    {
      "epoch": 0.0752454943408132,
      "grad_norm": 11.043139457702637,
      "learning_rate": 9.438093704717436e-07,
      "loss": 0.0829,
      "step": 477
    },
    {
      "epoch": 0.07540324170840398,
      "grad_norm": 4.340719699859619,
      "learning_rate": 9.436483658026082e-07,
      "loss": 0.1333,
      "step": 478
    },
    {
      "epoch": 0.07556098907599479,
      "grad_norm": 4.043093204498291,
      "learning_rate": 9.434873611334729e-07,
      "loss": 0.1048,
      "step": 479
    },
    {
      "epoch": 0.0757187364435856,
      "grad_norm": 7.1659417152404785,
      "learning_rate": 9.433263564643375e-07,
      "loss": 0.0923,
      "step": 480
    },
    {
      "epoch": 0.0758764838111764,
      "grad_norm": 12.38833999633789,
      "learning_rate": 9.431653517952021e-07,
      "loss": 0.1171,
      "step": 481
    },
    {
      "epoch": 0.0760342311787672,
      "grad_norm": 8.507014274597168,
      "learning_rate": 9.430043471260666e-07,
      "loss": 0.077,
      "step": 482
    },
    {
      "epoch": 0.07619197854635801,
      "grad_norm": 3.959944725036621,
      "learning_rate": 9.428433424569312e-07,
      "loss": 0.0501,
      "step": 483
    },
    {
      "epoch": 0.07634972591394881,
      "grad_norm": 4.138267993927002,
      "learning_rate": 9.426823377877958e-07,
      "loss": 0.1075,
      "step": 484
    },
    {
      "epoch": 0.07650747328153962,
      "grad_norm": 6.563967704772949,
      "learning_rate": 9.425213331186605e-07,
      "loss": 0.1285,
      "step": 485
    },
    {
      "epoch": 0.07666522064913042,
      "grad_norm": 5.721851825714111,
      "learning_rate": 9.42360328449525e-07,
      "loss": 0.0935,
      "step": 486
    },
    {
      "epoch": 0.07682296801672123,
      "grad_norm": 4.658355712890625,
      "learning_rate": 9.421993237803896e-07,
      "loss": 0.0714,
      "step": 487
    },
    {
      "epoch": 0.07698071538431202,
      "grad_norm": 7.072540760040283,
      "learning_rate": 9.420383191112542e-07,
      "loss": 0.1746,
      "step": 488
    },
    {
      "epoch": 0.07713846275190282,
      "grad_norm": 8.619388580322266,
      "learning_rate": 9.418773144421188e-07,
      "loss": 0.0629,
      "step": 489
    },
    {
      "epoch": 0.07729621011949363,
      "grad_norm": 4.010007858276367,
      "learning_rate": 9.417163097729833e-07,
      "loss": 0.0591,
      "step": 490
    },
    {
      "epoch": 0.07745395748708443,
      "grad_norm": 8.81683349609375,
      "learning_rate": 9.415553051038479e-07,
      "loss": 0.1324,
      "step": 491
    },
    {
      "epoch": 0.07761170485467524,
      "grad_norm": 5.686316013336182,
      "learning_rate": 9.413943004347125e-07,
      "loss": 0.0603,
      "step": 492
    },
    {
      "epoch": 0.07776945222226604,
      "grad_norm": 14.068903923034668,
      "learning_rate": 9.412332957655772e-07,
      "loss": 0.12,
      "step": 493
    },
    {
      "epoch": 0.07792719958985685,
      "grad_norm": 9.79300594329834,
      "learning_rate": 9.410722910964417e-07,
      "loss": 0.0626,
      "step": 494
    },
    {
      "epoch": 0.07808494695744765,
      "grad_norm": 3.8315680027008057,
      "learning_rate": 9.409112864273063e-07,
      "loss": 0.0477,
      "step": 495
    },
    {
      "epoch": 0.07824269432503846,
      "grad_norm": 6.023341178894043,
      "learning_rate": 9.40750281758171e-07,
      "loss": 0.1123,
      "step": 496
    },
    {
      "epoch": 0.07840044169262926,
      "grad_norm": 7.395686626434326,
      "learning_rate": 9.405892770890356e-07,
      "loss": 0.1365,
      "step": 497
    },
    {
      "epoch": 0.07855818906022005,
      "grad_norm": 3.773207426071167,
      "learning_rate": 9.404282724199001e-07,
      "loss": 0.0772,
      "step": 498
    },
    {
      "epoch": 0.07871593642781086,
      "grad_norm": 5.029270648956299,
      "learning_rate": 9.402672677507647e-07,
      "loss": 0.0928,
      "step": 499
    },
    {
      "epoch": 0.07887368379540166,
      "grad_norm": 8.06304931640625,
      "learning_rate": 9.401062630816294e-07,
      "loss": 0.0926,
      "step": 500
    },
    {
      "epoch": 0.07903143116299247,
      "grad_norm": 8.494391441345215,
      "learning_rate": 9.39945258412494e-07,
      "loss": 0.0712,
      "step": 501
    },
    {
      "epoch": 0.07918917853058327,
      "grad_norm": 19.12388038635254,
      "learning_rate": 9.397842537433586e-07,
      "loss": 0.1247,
      "step": 502
    },
    {
      "epoch": 0.07934692589817408,
      "grad_norm": 4.007694721221924,
      "learning_rate": 9.396232490742231e-07,
      "loss": 0.0871,
      "step": 503
    },
    {
      "epoch": 0.07950467326576488,
      "grad_norm": 4.083090305328369,
      "learning_rate": 9.394622444050877e-07,
      "loss": 0.0929,
      "step": 504
    },
    {
      "epoch": 0.07966242063335569,
      "grad_norm": 5.226316452026367,
      "learning_rate": 9.393012397359523e-07,
      "loss": 0.0758,
      "step": 505
    },
    {
      "epoch": 0.07982016800094649,
      "grad_norm": 5.83637809753418,
      "learning_rate": 9.391402350668169e-07,
      "loss": 0.1174,
      "step": 506
    },
    {
      "epoch": 0.07997791536853728,
      "grad_norm": 10.742006301879883,
      "learning_rate": 9.389792303976814e-07,
      "loss": 0.0948,
      "step": 507
    },
    {
      "epoch": 0.08013566273612809,
      "grad_norm": 11.444242477416992,
      "learning_rate": 9.388182257285461e-07,
      "loss": 0.1807,
      "step": 508
    },
    {
      "epoch": 0.08029341010371889,
      "grad_norm": 7.078170299530029,
      "learning_rate": 9.386572210594107e-07,
      "loss": 0.0794,
      "step": 509
    },
    {
      "epoch": 0.0804511574713097,
      "grad_norm": 4.803638935089111,
      "learning_rate": 9.384962163902753e-07,
      "loss": 0.1344,
      "step": 510
    },
    {
      "epoch": 0.0806089048389005,
      "grad_norm": 6.092920303344727,
      "learning_rate": 9.383352117211398e-07,
      "loss": 0.0774,
      "step": 511
    },
    {
      "epoch": 0.0807666522064913,
      "grad_norm": 17.177648544311523,
      "learning_rate": 9.381742070520044e-07,
      "loss": 0.1563,
      "step": 512
    },
    {
      "epoch": 0.08092439957408211,
      "grad_norm": 11.093384742736816,
      "learning_rate": 9.38013202382869e-07,
      "loss": 0.0795,
      "step": 513
    },
    {
      "epoch": 0.08108214694167291,
      "grad_norm": 9.071053504943848,
      "learning_rate": 9.378521977137337e-07,
      "loss": 0.1089,
      "step": 514
    },
    {
      "epoch": 0.08123989430926372,
      "grad_norm": 6.980352401733398,
      "learning_rate": 9.376911930445983e-07,
      "loss": 0.1469,
      "step": 515
    },
    {
      "epoch": 0.08139764167685452,
      "grad_norm": 6.9618988037109375,
      "learning_rate": 9.375301883754629e-07,
      "loss": 0.1253,
      "step": 516
    },
    {
      "epoch": 0.08155538904444531,
      "grad_norm": 6.925991535186768,
      "learning_rate": 9.373691837063275e-07,
      "loss": 0.1222,
      "step": 517
    },
    {
      "epoch": 0.08171313641203612,
      "grad_norm": 4.175725936889648,
      "learning_rate": 9.372081790371921e-07,
      "loss": 0.055,
      "step": 518
    },
    {
      "epoch": 0.08187088377962692,
      "grad_norm": 9.227218627929688,
      "learning_rate": 9.370471743680566e-07,
      "loss": 0.127,
      "step": 519
    },
    {
      "epoch": 0.08202863114721773,
      "grad_norm": 4.0665459632873535,
      "learning_rate": 9.368861696989212e-07,
      "loss": 0.0779,
      "step": 520
    },
    {
      "epoch": 0.08218637851480853,
      "grad_norm": 11.169240951538086,
      "learning_rate": 9.367251650297858e-07,
      "loss": 0.1303,
      "step": 521
    },
    {
      "epoch": 0.08234412588239934,
      "grad_norm": 9.30782413482666,
      "learning_rate": 9.365641603606504e-07,
      "loss": 0.0757,
      "step": 522
    },
    {
      "epoch": 0.08250187324999014,
      "grad_norm": 15.76601791381836,
      "learning_rate": 9.36403155691515e-07,
      "loss": 0.0687,
      "step": 523
    },
    {
      "epoch": 0.08265962061758095,
      "grad_norm": 4.672974109649658,
      "learning_rate": 9.362421510223796e-07,
      "loss": 0.1032,
      "step": 524
    },
    {
      "epoch": 0.08281736798517175,
      "grad_norm": 5.200424671173096,
      "learning_rate": 9.360811463532442e-07,
      "loss": 0.0743,
      "step": 525
    },
    {
      "epoch": 0.08297511535276256,
      "grad_norm": 7.3553900718688965,
      "learning_rate": 9.359201416841088e-07,
      "loss": 0.0567,
      "step": 526
    },
    {
      "epoch": 0.08313286272035335,
      "grad_norm": 4.422732353210449,
      "learning_rate": 9.357591370149734e-07,
      "loss": 0.0665,
      "step": 527
    },
    {
      "epoch": 0.08329061008794415,
      "grad_norm": 7.514669418334961,
      "learning_rate": 9.355981323458379e-07,
      "loss": 0.0975,
      "step": 528
    },
    {
      "epoch": 0.08344835745553496,
      "grad_norm": 8.483025550842285,
      "learning_rate": 9.354371276767025e-07,
      "loss": 0.0869,
      "step": 529
    },
    {
      "epoch": 0.08360610482312576,
      "grad_norm": 13.895788192749023,
      "learning_rate": 9.352761230075672e-07,
      "loss": 0.1096,
      "step": 530
    },
    {
      "epoch": 0.08376385219071657,
      "grad_norm": 5.938143730163574,
      "learning_rate": 9.351151183384319e-07,
      "loss": 0.0868,
      "step": 531
    },
    {
      "epoch": 0.08392159955830737,
      "grad_norm": 6.248524188995361,
      "learning_rate": 9.349541136692964e-07,
      "loss": 0.1188,
      "step": 532
    },
    {
      "epoch": 0.08407934692589818,
      "grad_norm": 4.3282856941223145,
      "learning_rate": 9.34793109000161e-07,
      "loss": 0.0617,
      "step": 533
    },
    {
      "epoch": 0.08423709429348898,
      "grad_norm": 12.2025785446167,
      "learning_rate": 9.346321043310256e-07,
      "loss": 0.1276,
      "step": 534
    },
    {
      "epoch": 0.08439484166107979,
      "grad_norm": 7.238867282867432,
      "learning_rate": 9.344710996618902e-07,
      "loss": 0.1071,
      "step": 535
    },
    {
      "epoch": 0.08455258902867058,
      "grad_norm": 3.5196988582611084,
      "learning_rate": 9.343100949927547e-07,
      "loss": 0.0471,
      "step": 536
    },
    {
      "epoch": 0.08471033639626138,
      "grad_norm": 4.482356071472168,
      "learning_rate": 9.341490903236193e-07,
      "loss": 0.0739,
      "step": 537
    },
    {
      "epoch": 0.08486808376385219,
      "grad_norm": 3.579318046569824,
      "learning_rate": 9.33988085654484e-07,
      "loss": 0.0514,
      "step": 538
    },
    {
      "epoch": 0.08502583113144299,
      "grad_norm": 5.4972639083862305,
      "learning_rate": 9.338270809853486e-07,
      "loss": 0.0577,
      "step": 539
    },
    {
      "epoch": 0.0851835784990338,
      "grad_norm": 4.826931476593018,
      "learning_rate": 9.336660763162131e-07,
      "loss": 0.0622,
      "step": 540
    },
    {
      "epoch": 0.0853413258666246,
      "grad_norm": 4.429808139801025,
      "learning_rate": 9.335050716470777e-07,
      "loss": 0.067,
      "step": 541
    },
    {
      "epoch": 0.0854990732342154,
      "grad_norm": 5.156866073608398,
      "learning_rate": 9.333440669779423e-07,
      "loss": 0.0924,
      "step": 542
    },
    {
      "epoch": 0.08565682060180621,
      "grad_norm": 3.470766067504883,
      "learning_rate": 9.331830623088069e-07,
      "loss": 0.082,
      "step": 543
    },
    {
      "epoch": 0.08581456796939702,
      "grad_norm": 8.749771118164062,
      "learning_rate": 9.330220576396714e-07,
      "loss": 0.0935,
      "step": 544
    },
    {
      "epoch": 0.08597231533698782,
      "grad_norm": 8.987846374511719,
      "learning_rate": 9.328610529705361e-07,
      "loss": 0.1161,
      "step": 545
    },
    {
      "epoch": 0.08613006270457861,
      "grad_norm": 7.917525291442871,
      "learning_rate": 9.327000483014007e-07,
      "loss": 0.068,
      "step": 546
    },
    {
      "epoch": 0.08628781007216942,
      "grad_norm": 7.498785018920898,
      "learning_rate": 9.325390436322653e-07,
      "loss": 0.1017,
      "step": 547
    },
    {
      "epoch": 0.08644555743976022,
      "grad_norm": 6.345965385437012,
      "learning_rate": 9.3237803896313e-07,
      "loss": 0.1245,
      "step": 548
    },
    {
      "epoch": 0.08660330480735103,
      "grad_norm": 5.038855075836182,
      "learning_rate": 9.322170342939944e-07,
      "loss": 0.0764,
      "step": 549
    },
    {
      "epoch": 0.08676105217494183,
      "grad_norm": 7.756494045257568,
      "learning_rate": 9.320560296248591e-07,
      "loss": 0.0708,
      "step": 550
    },
    {
      "epoch": 0.08691879954253263,
      "grad_norm": 9.325003623962402,
      "learning_rate": 9.318950249557237e-07,
      "loss": 0.0832,
      "step": 551
    },
    {
      "epoch": 0.08707654691012344,
      "grad_norm": 8.301737785339355,
      "learning_rate": 9.317340202865883e-07,
      "loss": 0.0767,
      "step": 552
    },
    {
      "epoch": 0.08723429427771424,
      "grad_norm": 8.105409622192383,
      "learning_rate": 9.315730156174529e-07,
      "loss": 0.1151,
      "step": 553
    },
    {
      "epoch": 0.08739204164530505,
      "grad_norm": 7.55850887298584,
      "learning_rate": 9.314120109483175e-07,
      "loss": 0.0657,
      "step": 554
    },
    {
      "epoch": 0.08754978901289585,
      "grad_norm": 7.7853851318359375,
      "learning_rate": 9.312510062791821e-07,
      "loss": 0.1194,
      "step": 555
    },
    {
      "epoch": 0.08770753638048664,
      "grad_norm": 5.793752193450928,
      "learning_rate": 9.310900016100467e-07,
      "loss": 0.0739,
      "step": 556
    },
    {
      "epoch": 0.08786528374807745,
      "grad_norm": 7.338489532470703,
      "learning_rate": 9.309289969409112e-07,
      "loss": 0.118,
      "step": 557
    },
    {
      "epoch": 0.08802303111566825,
      "grad_norm": 5.680853843688965,
      "learning_rate": 9.307679922717758e-07,
      "loss": 0.0967,
      "step": 558
    },
    {
      "epoch": 0.08818077848325906,
      "grad_norm": 4.661684989929199,
      "learning_rate": 9.306069876026404e-07,
      "loss": 0.0893,
      "step": 559
    },
    {
      "epoch": 0.08833852585084986,
      "grad_norm": 8.23542594909668,
      "learning_rate": 9.304459829335051e-07,
      "loss": 0.1228,
      "step": 560
    },
    {
      "epoch": 0.08849627321844067,
      "grad_norm": 4.448554515838623,
      "learning_rate": 9.302849782643696e-07,
      "loss": 0.1152,
      "step": 561
    },
    {
      "epoch": 0.08865402058603147,
      "grad_norm": 3.9955506324768066,
      "learning_rate": 9.301239735952342e-07,
      "loss": 0.0488,
      "step": 562
    },
    {
      "epoch": 0.08881176795362228,
      "grad_norm": 4.207185745239258,
      "learning_rate": 9.299629689260988e-07,
      "loss": 0.0612,
      "step": 563
    },
    {
      "epoch": 0.08896951532121308,
      "grad_norm": 3.6789309978485107,
      "learning_rate": 9.298019642569634e-07,
      "loss": 0.0847,
      "step": 564
    },
    {
      "epoch": 0.08912726268880387,
      "grad_norm": 3.0475735664367676,
      "learning_rate": 9.296409595878279e-07,
      "loss": 0.0582,
      "step": 565
    },
    {
      "epoch": 0.08928501005639468,
      "grad_norm": 5.452734470367432,
      "learning_rate": 9.294799549186925e-07,
      "loss": 0.0614,
      "step": 566
    },
    {
      "epoch": 0.08944275742398548,
      "grad_norm": 8.011670112609863,
      "learning_rate": 9.293189502495572e-07,
      "loss": 0.0633,
      "step": 567
    },
    {
      "epoch": 0.08960050479157629,
      "grad_norm": 5.0398993492126465,
      "learning_rate": 9.291579455804219e-07,
      "loss": 0.0647,
      "step": 568
    },
    {
      "epoch": 0.08975825215916709,
      "grad_norm": 3.8250749111175537,
      "learning_rate": 9.289969409112864e-07,
      "loss": 0.0717,
      "step": 569
    },
    {
      "epoch": 0.0899159995267579,
      "grad_norm": 3.7562711238861084,
      "learning_rate": 9.28835936242151e-07,
      "loss": 0.0505,
      "step": 570
    },
    {
      "epoch": 0.0900737468943487,
      "grad_norm": 8.792325973510742,
      "learning_rate": 9.286749315730156e-07,
      "loss": 0.1248,
      "step": 571
    },
    {
      "epoch": 0.09023149426193951,
      "grad_norm": 6.8114118576049805,
      "learning_rate": 9.285139269038802e-07,
      "loss": 0.0419,
      "step": 572
    },
    {
      "epoch": 0.09038924162953031,
      "grad_norm": 5.860121726989746,
      "learning_rate": 9.283529222347448e-07,
      "loss": 0.0777,
      "step": 573
    },
    {
      "epoch": 0.09054698899712112,
      "grad_norm": 4.751367092132568,
      "learning_rate": 9.281919175656093e-07,
      "loss": 0.0588,
      "step": 574
    },
    {
      "epoch": 0.09070473636471191,
      "grad_norm": 5.606316566467285,
      "learning_rate": 9.28030912896474e-07,
      "loss": 0.0737,
      "step": 575
    },
    {
      "epoch": 0.09086248373230271,
      "grad_norm": 5.022028923034668,
      "learning_rate": 9.278699082273386e-07,
      "loss": 0.078,
      "step": 576
    },
    {
      "epoch": 0.09102023109989352,
      "grad_norm": 7.044178485870361,
      "learning_rate": 9.277089035582032e-07,
      "loss": 0.1334,
      "step": 577
    },
    {
      "epoch": 0.09117797846748432,
      "grad_norm": 3.5324037075042725,
      "learning_rate": 9.275478988890677e-07,
      "loss": 0.0626,
      "step": 578
    },
    {
      "epoch": 0.09133572583507513,
      "grad_norm": 5.3293962478637695,
      "learning_rate": 9.273868942199323e-07,
      "loss": 0.0676,
      "step": 579
    },
    {
      "epoch": 0.09149347320266593,
      "grad_norm": 5.908418655395508,
      "learning_rate": 9.272258895507969e-07,
      "loss": 0.095,
      "step": 580
    },
    {
      "epoch": 0.09165122057025674,
      "grad_norm": 5.927475452423096,
      "learning_rate": 9.270648848816615e-07,
      "loss": 0.1069,
      "step": 581
    },
    {
      "epoch": 0.09180896793784754,
      "grad_norm": 9.431535720825195,
      "learning_rate": 9.26903880212526e-07,
      "loss": 0.0812,
      "step": 582
    },
    {
      "epoch": 0.09196671530543835,
      "grad_norm": 4.260199546813965,
      "learning_rate": 9.267428755433907e-07,
      "loss": 0.0458,
      "step": 583
    },
    {
      "epoch": 0.09212446267302915,
      "grad_norm": 6.684591770172119,
      "learning_rate": 9.265818708742554e-07,
      "loss": 0.1157,
      "step": 584
    },
    {
      "epoch": 0.09228221004061994,
      "grad_norm": 9.132654190063477,
      "learning_rate": 9.2642086620512e-07,
      "loss": 0.1202,
      "step": 585
    },
    {
      "epoch": 0.09243995740821075,
      "grad_norm": 5.602100372314453,
      "learning_rate": 9.262598615359845e-07,
      "loss": 0.0618,
      "step": 586
    },
    {
      "epoch": 0.09259770477580155,
      "grad_norm": 5.033472537994385,
      "learning_rate": 9.260988568668491e-07,
      "loss": 0.0484,
      "step": 587
    },
    {
      "epoch": 0.09275545214339236,
      "grad_norm": 7.528144359588623,
      "learning_rate": 9.259378521977137e-07,
      "loss": 0.0992,
      "step": 588
    },
    {
      "epoch": 0.09291319951098316,
      "grad_norm": 10.023838996887207,
      "learning_rate": 9.257768475285783e-07,
      "loss": 0.0755,
      "step": 589
    },
    {
      "epoch": 0.09307094687857397,
      "grad_norm": 5.312619209289551,
      "learning_rate": 9.256158428594428e-07,
      "loss": 0.0948,
      "step": 590
    },
    {
      "epoch": 0.09322869424616477,
      "grad_norm": 3.866603136062622,
      "learning_rate": 9.254548381903075e-07,
      "loss": 0.0632,
      "step": 591
    },
    {
      "epoch": 0.09338644161375557,
      "grad_norm": 5.804541110992432,
      "learning_rate": 9.252938335211721e-07,
      "loss": 0.1006,
      "step": 592
    },
    {
      "epoch": 0.09354418898134638,
      "grad_norm": 9.800780296325684,
      "learning_rate": 9.251328288520367e-07,
      "loss": 0.0794,
      "step": 593
    },
    {
      "epoch": 0.09370193634893717,
      "grad_norm": 6.569456577301025,
      "learning_rate": 9.249718241829013e-07,
      "loss": 0.0752,
      "step": 594
    },
    {
      "epoch": 0.09385968371652798,
      "grad_norm": 4.535599708557129,
      "learning_rate": 9.248108195137658e-07,
      "loss": 0.0926,
      "step": 595
    },
    {
      "epoch": 0.09401743108411878,
      "grad_norm": 5.393555164337158,
      "learning_rate": 9.246498148446304e-07,
      "loss": 0.0546,
      "step": 596
    },
    {
      "epoch": 0.09417517845170958,
      "grad_norm": 9.806357383728027,
      "learning_rate": 9.24488810175495e-07,
      "loss": 0.0646,
      "step": 597
    },
    {
      "epoch": 0.09433292581930039,
      "grad_norm": 4.138769626617432,
      "learning_rate": 9.243278055063597e-07,
      "loss": 0.0725,
      "step": 598
    },
    {
      "epoch": 0.0944906731868912,
      "grad_norm": 4.142092227935791,
      "learning_rate": 9.241668008372242e-07,
      "loss": 0.1253,
      "step": 599
    },
    {
      "epoch": 0.094648420554482,
      "grad_norm": 6.697494029998779,
      "learning_rate": 9.240057961680888e-07,
      "loss": 0.119,
      "step": 600
    },
    {
      "epoch": 0.0948061679220728,
      "grad_norm": 4.450563430786133,
      "learning_rate": 9.238447914989534e-07,
      "loss": 0.1033,
      "step": 601
    },
    {
      "epoch": 0.09496391528966361,
      "grad_norm": 8.81712532043457,
      "learning_rate": 9.236837868298181e-07,
      "loss": 0.0871,
      "step": 602
    },
    {
      "epoch": 0.09512166265725441,
      "grad_norm": 2.71614933013916,
      "learning_rate": 9.235227821606826e-07,
      "loss": 0.0315,
      "step": 603
    },
    {
      "epoch": 0.0952794100248452,
      "grad_norm": 5.211971282958984,
      "learning_rate": 9.233617774915472e-07,
      "loss": 0.1361,
      "step": 604
    },
    {
      "epoch": 0.09543715739243601,
      "grad_norm": 4.793868541717529,
      "learning_rate": 9.232007728224119e-07,
      "loss": 0.0529,
      "step": 605
    },
    {
      "epoch": 0.09559490476002681,
      "grad_norm": 5.744448661804199,
      "learning_rate": 9.230397681532765e-07,
      "loss": 0.0878,
      "step": 606
    },
    {
      "epoch": 0.09575265212761762,
      "grad_norm": 7.048700332641602,
      "learning_rate": 9.22878763484141e-07,
      "loss": 0.0548,
      "step": 607
    },
    {
      "epoch": 0.09591039949520842,
      "grad_norm": 3.6247646808624268,
      "learning_rate": 9.227177588150056e-07,
      "loss": 0.0841,
      "step": 608
    },
    {
      "epoch": 0.09606814686279923,
      "grad_norm": 15.860011100769043,
      "learning_rate": 9.225567541458702e-07,
      "loss": 0.1282,
      "step": 609
    },
    {
      "epoch": 0.09622589423039003,
      "grad_norm": 4.523679733276367,
      "learning_rate": 9.223957494767348e-07,
      "loss": 0.0486,
      "step": 610
    },
    {
      "epoch": 0.09638364159798084,
      "grad_norm": 5.463552951812744,
      "learning_rate": 9.222347448075993e-07,
      "loss": 0.1049,
      "step": 611
    },
    {
      "epoch": 0.09654138896557164,
      "grad_norm": 6.040287017822266,
      "learning_rate": 9.220737401384639e-07,
      "loss": 0.1149,
      "step": 612
    },
    {
      "epoch": 0.09669913633316245,
      "grad_norm": 5.69699764251709,
      "learning_rate": 9.219127354693286e-07,
      "loss": 0.0711,
      "step": 613
    },
    {
      "epoch": 0.09685688370075324,
      "grad_norm": 8.530750274658203,
      "learning_rate": 9.217517308001932e-07,
      "loss": 0.056,
      "step": 614
    },
    {
      "epoch": 0.09701463106834404,
      "grad_norm": 5.582493305206299,
      "learning_rate": 9.215907261310577e-07,
      "loss": 0.0793,
      "step": 615
    },
    {
      "epoch": 0.09717237843593485,
      "grad_norm": 13.381341934204102,
      "learning_rate": 9.214297214619223e-07,
      "loss": 0.1297,
      "step": 616
    },
    {
      "epoch": 0.09733012580352565,
      "grad_norm": 8.435693740844727,
      "learning_rate": 9.212687167927869e-07,
      "loss": 0.1178,
      "step": 617
    },
    {
      "epoch": 0.09748787317111646,
      "grad_norm": 7.7765703201293945,
      "learning_rate": 9.211077121236515e-07,
      "loss": 0.1133,
      "step": 618
    },
    {
      "epoch": 0.09764562053870726,
      "grad_norm": 6.829525947570801,
      "learning_rate": 9.209467074545161e-07,
      "loss": 0.124,
      "step": 619
    },
    {
      "epoch": 0.09780336790629807,
      "grad_norm": 7.434627532958984,
      "learning_rate": 9.207857027853806e-07,
      "loss": 0.0649,
      "step": 620
    },
    {
      "epoch": 0.09796111527388887,
      "grad_norm": 8.623050689697266,
      "learning_rate": 9.206246981162454e-07,
      "loss": 0.1165,
      "step": 621
    },
    {
      "epoch": 0.09811886264147968,
      "grad_norm": 6.121469974517822,
      "learning_rate": 9.2046369344711e-07,
      "loss": 0.0831,
      "step": 622
    },
    {
      "epoch": 0.09827661000907047,
      "grad_norm": 6.457248210906982,
      "learning_rate": 9.203026887779746e-07,
      "loss": 0.0733,
      "step": 623
    },
    {
      "epoch": 0.09843435737666127,
      "grad_norm": 6.442403316497803,
      "learning_rate": 9.201416841088391e-07,
      "loss": 0.0985,
      "step": 624
    },
    {
      "epoch": 0.09859210474425208,
      "grad_norm": 6.161303997039795,
      "learning_rate": 9.199806794397037e-07,
      "loss": 0.082,
      "step": 625
    },
    {
      "epoch": 0.09874985211184288,
      "grad_norm": 7.090782642364502,
      "learning_rate": 9.198196747705683e-07,
      "loss": 0.0338,
      "step": 626
    },
    {
      "epoch": 0.09890759947943369,
      "grad_norm": 6.823625087738037,
      "learning_rate": 9.196586701014329e-07,
      "loss": 0.0977,
      "step": 627
    },
    {
      "epoch": 0.09906534684702449,
      "grad_norm": 6.264141082763672,
      "learning_rate": 9.194976654322975e-07,
      "loss": 0.0556,
      "step": 628
    },
    {
      "epoch": 0.0992230942146153,
      "grad_norm": 4.8218841552734375,
      "learning_rate": 9.193366607631621e-07,
      "loss": 0.0698,
      "step": 629
    },
    {
      "epoch": 0.0993808415822061,
      "grad_norm": 5.638354778289795,
      "learning_rate": 9.191756560940267e-07,
      "loss": 0.0851,
      "step": 630
    },
    {
      "epoch": 0.0995385889497969,
      "grad_norm": 4.278707981109619,
      "learning_rate": 9.190146514248913e-07,
      "loss": 0.0892,
      "step": 631
    },
    {
      "epoch": 0.09969633631738771,
      "grad_norm": 5.753275394439697,
      "learning_rate": 9.188536467557558e-07,
      "loss": 0.083,
      "step": 632
    },
    {
      "epoch": 0.0998540836849785,
      "grad_norm": 5.237423419952393,
      "learning_rate": 9.186926420866204e-07,
      "loss": 0.0786,
      "step": 633
    },
    {
      "epoch": 0.1000118310525693,
      "grad_norm": 12.854137420654297,
      "learning_rate": 9.18531637417485e-07,
      "loss": 0.125,
      "step": 634
    },
    {
      "epoch": 0.10016957842016011,
      "grad_norm": 8.807177543640137,
      "learning_rate": 9.183706327483497e-07,
      "loss": 0.1198,
      "step": 635
    },
    {
      "epoch": 0.10032732578775092,
      "grad_norm": 7.128739833831787,
      "learning_rate": 9.182096280792142e-07,
      "loss": 0.0891,
      "step": 636
    },
    {
      "epoch": 0.10048507315534172,
      "grad_norm": 5.885706424713135,
      "learning_rate": 9.180486234100789e-07,
      "loss": 0.0565,
      "step": 637
    },
    {
      "epoch": 0.10064282052293252,
      "grad_norm": 11.299050331115723,
      "learning_rate": 9.178876187409435e-07,
      "loss": 0.1356,
      "step": 638
    },
    {
      "epoch": 0.10080056789052333,
      "grad_norm": 6.594298839569092,
      "learning_rate": 9.177266140718081e-07,
      "loss": 0.0788,
      "step": 639
    },
    {
      "epoch": 0.10095831525811413,
      "grad_norm": 5.020556926727295,
      "learning_rate": 9.175656094026727e-07,
      "loss": 0.0933,
      "step": 640
    },
    {
      "epoch": 0.10111606262570494,
      "grad_norm": 7.084664344787598,
      "learning_rate": 9.174046047335372e-07,
      "loss": 0.099,
      "step": 641
    },
    {
      "epoch": 0.10127380999329574,
      "grad_norm": 6.887861728668213,
      "learning_rate": 9.172436000644018e-07,
      "loss": 0.0652,
      "step": 642
    },
    {
      "epoch": 0.10143155736088653,
      "grad_norm": 9.95553207397461,
      "learning_rate": 9.170825953952665e-07,
      "loss": 0.105,
      "step": 643
    },
    {
      "epoch": 0.10158930472847734,
      "grad_norm": 52.05778121948242,
      "learning_rate": 9.169215907261311e-07,
      "loss": 0.0901,
      "step": 644
    },
    {
      "epoch": 0.10174705209606814,
      "grad_norm": 7.226211071014404,
      "learning_rate": 9.167605860569956e-07,
      "loss": 0.0862,
      "step": 645
    },
    {
      "epoch": 0.10190479946365895,
      "grad_norm": 4.614170551300049,
      "learning_rate": 9.165995813878602e-07,
      "loss": 0.0907,
      "step": 646
    },
    {
      "epoch": 0.10206254683124975,
      "grad_norm": 3.3297107219696045,
      "learning_rate": 9.164385767187248e-07,
      "loss": 0.0389,
      "step": 647
    },
    {
      "epoch": 0.10222029419884056,
      "grad_norm": 8.3242826461792,
      "learning_rate": 9.162775720495894e-07,
      "loss": 0.0745,
      "step": 648
    },
    {
      "epoch": 0.10237804156643136,
      "grad_norm": 4.23464822769165,
      "learning_rate": 9.161165673804539e-07,
      "loss": 0.0714,
      "step": 649
    },
    {
      "epoch": 0.10253578893402217,
      "grad_norm": 5.633603096008301,
      "learning_rate": 9.159555627113185e-07,
      "loss": 0.0813,
      "step": 650
    },
    {
      "epoch": 0.10269353630161297,
      "grad_norm": 10.278568267822266,
      "learning_rate": 9.157945580421832e-07,
      "loss": 0.1132,
      "step": 651
    },
    {
      "epoch": 0.10285128366920376,
      "grad_norm": 22.042381286621094,
      "learning_rate": 9.156335533730478e-07,
      "loss": 0.0874,
      "step": 652
    },
    {
      "epoch": 0.10300903103679457,
      "grad_norm": 9.171586990356445,
      "learning_rate": 9.154725487039123e-07,
      "loss": 0.1028,
      "step": 653
    },
    {
      "epoch": 0.10316677840438537,
      "grad_norm": 8.835233688354492,
      "learning_rate": 9.153115440347769e-07,
      "loss": 0.0658,
      "step": 654
    },
    {
      "epoch": 0.10332452577197618,
      "grad_norm": 4.944420337677002,
      "learning_rate": 9.151505393656416e-07,
      "loss": 0.083,
      "step": 655
    },
    {
      "epoch": 0.10348227313956698,
      "grad_norm": 7.005538463592529,
      "learning_rate": 9.149895346965062e-07,
      "loss": 0.1406,
      "step": 656
    },
    {
      "epoch": 0.10364002050715779,
      "grad_norm": 11.50197982788086,
      "learning_rate": 9.148285300273707e-07,
      "loss": 0.1425,
      "step": 657
    },
    {
      "epoch": 0.10379776787474859,
      "grad_norm": 3.9614639282226562,
      "learning_rate": 9.146675253582354e-07,
      "loss": 0.0601,
      "step": 658
    },
    {
      "epoch": 0.1039555152423394,
      "grad_norm": 3.4335761070251465,
      "learning_rate": 9.145065206891e-07,
      "loss": 0.0668,
      "step": 659
    },
    {
      "epoch": 0.1041132626099302,
      "grad_norm": 6.671053886413574,
      "learning_rate": 9.143455160199646e-07,
      "loss": 0.0929,
      "step": 660
    },
    {
      "epoch": 0.104271009977521,
      "grad_norm": 3.743391990661621,
      "learning_rate": 9.141845113508291e-07,
      "loss": 0.0758,
      "step": 661
    },
    {
      "epoch": 0.1044287573451118,
      "grad_norm": 6.328144073486328,
      "learning_rate": 9.140235066816937e-07,
      "loss": 0.0597,
      "step": 662
    },
    {
      "epoch": 0.1045865047127026,
      "grad_norm": 5.457065105438232,
      "learning_rate": 9.138625020125583e-07,
      "loss": 0.0971,
      "step": 663
    },
    {
      "epoch": 0.1047442520802934,
      "grad_norm": 5.440207481384277,
      "learning_rate": 9.137014973434229e-07,
      "loss": 0.1172,
      "step": 664
    },
    {
      "epoch": 0.10490199944788421,
      "grad_norm": 3.1148476600646973,
      "learning_rate": 9.135404926742876e-07,
      "loss": 0.0433,
      "step": 665
    },
    {
      "epoch": 0.10505974681547502,
      "grad_norm": 3.905869483947754,
      "learning_rate": 9.133794880051521e-07,
      "loss": 0.0362,
      "step": 666
    },
    {
      "epoch": 0.10521749418306582,
      "grad_norm": 6.365758895874023,
      "learning_rate": 9.132184833360167e-07,
      "loss": 0.0725,
      "step": 667
    },
    {
      "epoch": 0.10537524155065663,
      "grad_norm": 7.652137756347656,
      "learning_rate": 9.130574786668813e-07,
      "loss": 0.0773,
      "step": 668
    },
    {
      "epoch": 0.10553298891824743,
      "grad_norm": 4.283510208129883,
      "learning_rate": 9.128964739977459e-07,
      "loss": 0.0729,
      "step": 669
    },
    {
      "epoch": 0.10569073628583824,
      "grad_norm": 6.367171764373779,
      "learning_rate": 9.127354693286104e-07,
      "loss": 0.1088,
      "step": 670
    },
    {
      "epoch": 0.10584848365342904,
      "grad_norm": 6.150599479675293,
      "learning_rate": 9.12574464659475e-07,
      "loss": 0.062,
      "step": 671
    },
    {
      "epoch": 0.10600623102101983,
      "grad_norm": 5.077731132507324,
      "learning_rate": 9.124134599903396e-07,
      "loss": 0.0469,
      "step": 672
    },
    {
      "epoch": 0.10616397838861064,
      "grad_norm": 4.419490337371826,
      "learning_rate": 9.122524553212044e-07,
      "loss": 0.0919,
      "step": 673
    },
    {
      "epoch": 0.10632172575620144,
      "grad_norm": 5.757754802703857,
      "learning_rate": 9.120914506520689e-07,
      "loss": 0.0529,
      "step": 674
    },
    {
      "epoch": 0.10647947312379225,
      "grad_norm": 5.260188579559326,
      "learning_rate": 9.119304459829335e-07,
      "loss": 0.0714,
      "step": 675
    },
    {
      "epoch": 0.10663722049138305,
      "grad_norm": 1.9488763809204102,
      "learning_rate": 9.117694413137981e-07,
      "loss": 0.0251,
      "step": 676
    },
    {
      "epoch": 0.10679496785897385,
      "grad_norm": 8.528887748718262,
      "learning_rate": 9.116084366446627e-07,
      "loss": 0.1339,
      "step": 677
    },
    {
      "epoch": 0.10695271522656466,
      "grad_norm": 8.196385383605957,
      "learning_rate": 9.114474319755272e-07,
      "loss": 0.1058,
      "step": 678
    },
    {
      "epoch": 0.10711046259415546,
      "grad_norm": 7.687489032745361,
      "learning_rate": 9.112864273063918e-07,
      "loss": 0.0996,
      "step": 679
    },
    {
      "epoch": 0.10726820996174627,
      "grad_norm": 3.2332191467285156,
      "learning_rate": 9.111254226372565e-07,
      "loss": 0.0705,
      "step": 680
    },
    {
      "epoch": 0.10742595732933706,
      "grad_norm": 6.897693634033203,
      "learning_rate": 9.109644179681211e-07,
      "loss": 0.1039,
      "step": 681
    },
    {
      "epoch": 0.10758370469692786,
      "grad_norm": 6.623891830444336,
      "learning_rate": 9.108034132989856e-07,
      "loss": 0.1054,
      "step": 682
    },
    {
      "epoch": 0.10774145206451867,
      "grad_norm": 5.592535018920898,
      "learning_rate": 9.106424086298502e-07,
      "loss": 0.0701,
      "step": 683
    },
    {
      "epoch": 0.10789919943210947,
      "grad_norm": 11.811220169067383,
      "learning_rate": 9.104814039607148e-07,
      "loss": 0.1015,
      "step": 684
    },
    {
      "epoch": 0.10805694679970028,
      "grad_norm": 4.838003158569336,
      "learning_rate": 9.103203992915794e-07,
      "loss": 0.0752,
      "step": 685
    },
    {
      "epoch": 0.10821469416729108,
      "grad_norm": 5.717372894287109,
      "learning_rate": 9.101593946224439e-07,
      "loss": 0.1081,
      "step": 686
    },
    {
      "epoch": 0.10837244153488189,
      "grad_norm": 6.643096923828125,
      "learning_rate": 9.099983899533085e-07,
      "loss": 0.0901,
      "step": 687
    },
    {
      "epoch": 0.1085301889024727,
      "grad_norm": 7.170506477355957,
      "learning_rate": 9.098373852841732e-07,
      "loss": 0.0657,
      "step": 688
    },
    {
      "epoch": 0.1086879362700635,
      "grad_norm": 5.268221378326416,
      "learning_rate": 9.096763806150379e-07,
      "loss": 0.0793,
      "step": 689
    },
    {
      "epoch": 0.1088456836376543,
      "grad_norm": 5.993264198303223,
      "learning_rate": 9.095153759459025e-07,
      "loss": 0.0979,
      "step": 690
    },
    {
      "epoch": 0.1090034310052451,
      "grad_norm": 6.888185501098633,
      "learning_rate": 9.09354371276767e-07,
      "loss": 0.1079,
      "step": 691
    },
    {
      "epoch": 0.1091611783728359,
      "grad_norm": 3.7119643688201904,
      "learning_rate": 9.091933666076316e-07,
      "loss": 0.0458,
      "step": 692
    },
    {
      "epoch": 0.1093189257404267,
      "grad_norm": 8.08841609954834,
      "learning_rate": 9.090323619384962e-07,
      "loss": 0.0827,
      "step": 693
    },
    {
      "epoch": 0.10947667310801751,
      "grad_norm": 5.303411960601807,
      "learning_rate": 9.088713572693608e-07,
      "loss": 0.0643,
      "step": 694
    },
    {
      "epoch": 0.10963442047560831,
      "grad_norm": 7.4574971199035645,
      "learning_rate": 9.087103526002253e-07,
      "loss": 0.1011,
      "step": 695
    },
    {
      "epoch": 0.10979216784319912,
      "grad_norm": 7.693200588226318,
      "learning_rate": 9.0854934793109e-07,
      "loss": 0.0873,
      "step": 696
    },
    {
      "epoch": 0.10994991521078992,
      "grad_norm": 5.869884014129639,
      "learning_rate": 9.083883432619546e-07,
      "loss": 0.0639,
      "step": 697
    },
    {
      "epoch": 0.11010766257838073,
      "grad_norm": 5.635510444641113,
      "learning_rate": 9.082273385928192e-07,
      "loss": 0.0461,
      "step": 698
    },
    {
      "epoch": 0.11026540994597153,
      "grad_norm": 5.684610366821289,
      "learning_rate": 9.080663339236837e-07,
      "loss": 0.0598,
      "step": 699
    },
    {
      "epoch": 0.11042315731356234,
      "grad_norm": 8.58759880065918,
      "learning_rate": 9.079053292545483e-07,
      "loss": 0.0769,
      "step": 700
    },
    {
      "epoch": 0.11058090468115313,
      "grad_norm": 6.110040664672852,
      "learning_rate": 9.077443245854129e-07,
      "loss": 0.076,
      "step": 701
    },
    {
      "epoch": 0.11073865204874393,
      "grad_norm": 5.440070152282715,
      "learning_rate": 9.075833199162775e-07,
      "loss": 0.0723,
      "step": 702
    },
    {
      "epoch": 0.11089639941633474,
      "grad_norm": 7.805794715881348,
      "learning_rate": 9.074223152471421e-07,
      "loss": 0.0898,
      "step": 703
    },
    {
      "epoch": 0.11105414678392554,
      "grad_norm": 7.114494323730469,
      "learning_rate": 9.072613105780067e-07,
      "loss": 0.0765,
      "step": 704
    },
    {
      "epoch": 0.11121189415151635,
      "grad_norm": 6.303781986236572,
      "learning_rate": 9.071003059088713e-07,
      "loss": 0.1273,
      "step": 705
    },
    {
      "epoch": 0.11136964151910715,
      "grad_norm": 3.669435977935791,
      "learning_rate": 9.069393012397359e-07,
      "loss": 0.0703,
      "step": 706
    },
    {
      "epoch": 0.11152738888669796,
      "grad_norm": 4.772210121154785,
      "learning_rate": 9.067782965706004e-07,
      "loss": 0.0752,
      "step": 707
    },
    {
      "epoch": 0.11168513625428876,
      "grad_norm": 3.1551268100738525,
      "learning_rate": 9.06617291901465e-07,
      "loss": 0.0615,
      "step": 708
    },
    {
      "epoch": 0.11184288362187957,
      "grad_norm": 3.9344019889831543,
      "learning_rate": 9.064562872323297e-07,
      "loss": 0.0664,
      "step": 709
    },
    {
      "epoch": 0.11200063098947036,
      "grad_norm": 3.228260040283203,
      "learning_rate": 9.062952825631944e-07,
      "loss": 0.0814,
      "step": 710
    },
    {
      "epoch": 0.11215837835706116,
      "grad_norm": 5.874837875366211,
      "learning_rate": 9.06134277894059e-07,
      "loss": 0.0872,
      "step": 711
    },
    {
      "epoch": 0.11231612572465197,
      "grad_norm": 9.931432723999023,
      "learning_rate": 9.059732732249235e-07,
      "loss": 0.0745,
      "step": 712
    },
    {
      "epoch": 0.11247387309224277,
      "grad_norm": 10.438408851623535,
      "learning_rate": 9.058122685557881e-07,
      "loss": 0.1201,
      "step": 713
    },
    {
      "epoch": 0.11263162045983358,
      "grad_norm": 5.636549949645996,
      "learning_rate": 9.056512638866527e-07,
      "loss": 0.066,
      "step": 714
    },
    {
      "epoch": 0.11278936782742438,
      "grad_norm": 5.628836154937744,
      "learning_rate": 9.054902592175173e-07,
      "loss": 0.0778,
      "step": 715
    },
    {
      "epoch": 0.11294711519501519,
      "grad_norm": 3.7619338035583496,
      "learning_rate": 9.053292545483818e-07,
      "loss": 0.0459,
      "step": 716
    },
    {
      "epoch": 0.11310486256260599,
      "grad_norm": 6.656398773193359,
      "learning_rate": 9.051682498792464e-07,
      "loss": 0.1077,
      "step": 717
    },
    {
      "epoch": 0.1132626099301968,
      "grad_norm": 7.641002655029297,
      "learning_rate": 9.050072452101111e-07,
      "loss": 0.1254,
      "step": 718
    },
    {
      "epoch": 0.1134203572977876,
      "grad_norm": 6.015817642211914,
      "learning_rate": 9.048462405409757e-07,
      "loss": 0.0674,
      "step": 719
    },
    {
      "epoch": 0.11357810466537839,
      "grad_norm": 7.3636698722839355,
      "learning_rate": 9.046852358718402e-07,
      "loss": 0.0895,
      "step": 720
    },
    {
      "epoch": 0.1137358520329692,
      "grad_norm": 10.179230690002441,
      "learning_rate": 9.045242312027048e-07,
      "loss": 0.1238,
      "step": 721
    },
    {
      "epoch": 0.11389359940056,
      "grad_norm": 3.5914306640625,
      "learning_rate": 9.043632265335694e-07,
      "loss": 0.0765,
      "step": 722
    },
    {
      "epoch": 0.1140513467681508,
      "grad_norm": 4.742732048034668,
      "learning_rate": 9.04202221864434e-07,
      "loss": 0.0714,
      "step": 723
    },
    {
      "epoch": 0.11420909413574161,
      "grad_norm": 5.424478530883789,
      "learning_rate": 9.040412171952985e-07,
      "loss": 0.0517,
      "step": 724
    },
    {
      "epoch": 0.11436684150333241,
      "grad_norm": 11.305291175842285,
      "learning_rate": 9.038802125261631e-07,
      "loss": 0.1395,
      "step": 725
    },
    {
      "epoch": 0.11452458887092322,
      "grad_norm": 3.5070836544036865,
      "learning_rate": 9.037192078570279e-07,
      "loss": 0.0719,
      "step": 726
    },
    {
      "epoch": 0.11468233623851402,
      "grad_norm": 10.200228691101074,
      "learning_rate": 9.035582031878925e-07,
      "loss": 0.1091,
      "step": 727
    },
    {
      "epoch": 0.11484008360610483,
      "grad_norm": 3.7730553150177,
      "learning_rate": 9.03397198518757e-07,
      "loss": 0.0639,
      "step": 728
    },
    {
      "epoch": 0.11499783097369563,
      "grad_norm": 5.975857257843018,
      "learning_rate": 9.032361938496216e-07,
      "loss": 0.0646,
      "step": 729
    },
    {
      "epoch": 0.11515557834128642,
      "grad_norm": 4.199368476867676,
      "learning_rate": 9.030751891804862e-07,
      "loss": 0.1113,
      "step": 730
    },
    {
      "epoch": 0.11531332570887723,
      "grad_norm": 7.727891445159912,
      "learning_rate": 9.029141845113508e-07,
      "loss": 0.1398,
      "step": 731
    },
    {
      "epoch": 0.11547107307646803,
      "grad_norm": 4.718742847442627,
      "learning_rate": 9.027531798422153e-07,
      "loss": 0.0419,
      "step": 732
    },
    {
      "epoch": 0.11562882044405884,
      "grad_norm": 4.954360008239746,
      "learning_rate": 9.0259217517308e-07,
      "loss": 0.1133,
      "step": 733
    },
    {
      "epoch": 0.11578656781164964,
      "grad_norm": 2.603778839111328,
      "learning_rate": 9.024311705039446e-07,
      "loss": 0.0611,
      "step": 734
    },
    {
      "epoch": 0.11594431517924045,
      "grad_norm": 6.089944362640381,
      "learning_rate": 9.022701658348092e-07,
      "loss": 0.0555,
      "step": 735
    },
    {
      "epoch": 0.11610206254683125,
      "grad_norm": 3.6806843280792236,
      "learning_rate": 9.021091611656738e-07,
      "loss": 0.0568,
      "step": 736
    },
    {
      "epoch": 0.11625980991442206,
      "grad_norm": 7.865287780761719,
      "learning_rate": 9.019481564965383e-07,
      "loss": 0.0858,
      "step": 737
    },
    {
      "epoch": 0.11641755728201286,
      "grad_norm": 4.575242042541504,
      "learning_rate": 9.017871518274029e-07,
      "loss": 0.0885,
      "step": 738
    },
    {
      "epoch": 0.11657530464960365,
      "grad_norm": 7.171676158905029,
      "learning_rate": 9.016261471582675e-07,
      "loss": 0.0523,
      "step": 739
    },
    {
      "epoch": 0.11673305201719446,
      "grad_norm": 4.008410930633545,
      "learning_rate": 9.014651424891322e-07,
      "loss": 0.0532,
      "step": 740
    },
    {
      "epoch": 0.11689079938478526,
      "grad_norm": 6.986484050750732,
      "learning_rate": 9.013041378199967e-07,
      "loss": 0.0436,
      "step": 741
    },
    {
      "epoch": 0.11704854675237607,
      "grad_norm": 5.582924842834473,
      "learning_rate": 9.011431331508613e-07,
      "loss": 0.0587,
      "step": 742
    },
    {
      "epoch": 0.11720629411996687,
      "grad_norm": 3.2100069522857666,
      "learning_rate": 9.00982128481726e-07,
      "loss": 0.064,
      "step": 743
    },
    {
      "epoch": 0.11736404148755768,
      "grad_norm": 4.56215238571167,
      "learning_rate": 9.008211238125906e-07,
      "loss": 0.0859,
      "step": 744
    },
    {
      "epoch": 0.11752178885514848,
      "grad_norm": 6.167077541351318,
      "learning_rate": 9.006601191434551e-07,
      "loss": 0.0664,
      "step": 745
    },
    {
      "epoch": 0.11767953622273929,
      "grad_norm": 7.485244274139404,
      "learning_rate": 9.004991144743197e-07,
      "loss": 0.1289,
      "step": 746
    },
    {
      "epoch": 0.11783728359033009,
      "grad_norm": 10.55856990814209,
      "learning_rate": 9.003381098051843e-07,
      "loss": 0.08,
      "step": 747
    },
    {
      "epoch": 0.1179950309579209,
      "grad_norm": 15.230066299438477,
      "learning_rate": 9.00177105136049e-07,
      "loss": 0.0937,
      "step": 748
    },
    {
      "epoch": 0.11815277832551169,
      "grad_norm": 5.164748191833496,
      "learning_rate": 9.000161004669135e-07,
      "loss": 0.0805,
      "step": 749
    },
    {
      "epoch": 0.11831052569310249,
      "grad_norm": 7.331662178039551,
      "learning_rate": 8.998550957977781e-07,
      "loss": 0.056,
      "step": 750
    },
    {
      "epoch": 0.1184682730606933,
      "grad_norm": 7.451263904571533,
      "learning_rate": 8.996940911286427e-07,
      "loss": 0.0618,
      "step": 751
    },
    {
      "epoch": 0.1186260204282841,
      "grad_norm": 3.639286994934082,
      "learning_rate": 8.995330864595073e-07,
      "loss": 0.0487,
      "step": 752
    },
    {
      "epoch": 0.1187837677958749,
      "grad_norm": 5.050006866455078,
      "learning_rate": 8.993720817903718e-07,
      "loss": 0.0723,
      "step": 753
    },
    {
      "epoch": 0.11894151516346571,
      "grad_norm": 5.0852837562561035,
      "learning_rate": 8.992110771212364e-07,
      "loss": 0.1024,
      "step": 754
    },
    {
      "epoch": 0.11909926253105652,
      "grad_norm": 5.076583385467529,
      "learning_rate": 8.99050072452101e-07,
      "loss": 0.1035,
      "step": 755
    },
    {
      "epoch": 0.11925700989864732,
      "grad_norm": 8.556181907653809,
      "learning_rate": 8.988890677829657e-07,
      "loss": 0.0744,
      "step": 756
    },
    {
      "epoch": 0.11941475726623813,
      "grad_norm": 4.467113018035889,
      "learning_rate": 8.987280631138303e-07,
      "loss": 0.0343,
      "step": 757
    },
    {
      "epoch": 0.11957250463382893,
      "grad_norm": 5.956662178039551,
      "learning_rate": 8.985670584446948e-07,
      "loss": 0.0757,
      "step": 758
    },
    {
      "epoch": 0.11973025200141972,
      "grad_norm": 9.729576110839844,
      "learning_rate": 8.984060537755594e-07,
      "loss": 0.0774,
      "step": 759
    },
    {
      "epoch": 0.11988799936901053,
      "grad_norm": 5.245589256286621,
      "learning_rate": 8.98245049106424e-07,
      "loss": 0.082,
      "step": 760
    },
    {
      "epoch": 0.12004574673660133,
      "grad_norm": 9.502345085144043,
      "learning_rate": 8.980840444372887e-07,
      "loss": 0.1337,
      "step": 761
    },
    {
      "epoch": 0.12020349410419214,
      "grad_norm": 6.284233570098877,
      "learning_rate": 8.979230397681532e-07,
      "loss": 0.0802,
      "step": 762
    },
    {
      "epoch": 0.12036124147178294,
      "grad_norm": 4.558141231536865,
      "learning_rate": 8.977620350990179e-07,
      "loss": 0.0639,
      "step": 763
    },
    {
      "epoch": 0.12051898883937374,
      "grad_norm": 3.1438100337982178,
      "learning_rate": 8.976010304298825e-07,
      "loss": 0.0714,
      "step": 764
    },
    {
      "epoch": 0.12067673620696455,
      "grad_norm": 5.920631408691406,
      "learning_rate": 8.974400257607471e-07,
      "loss": 0.1316,
      "step": 765
    },
    {
      "epoch": 0.12083448357455535,
      "grad_norm": 4.774268627166748,
      "learning_rate": 8.972790210916116e-07,
      "loss": 0.0631,
      "step": 766
    },
    {
      "epoch": 0.12099223094214616,
      "grad_norm": 4.985866069793701,
      "learning_rate": 8.971180164224762e-07,
      "loss": 0.1117,
      "step": 767
    },
    {
      "epoch": 0.12114997830973695,
      "grad_norm": 4.642230987548828,
      "learning_rate": 8.969570117533408e-07,
      "loss": 0.0664,
      "step": 768
    },
    {
      "epoch": 0.12114997830973695,
      "eval_accuracy": 0.980177959626334,
      "eval_f1": 0.980177959626334,
      "eval_loss": 0.068110391497612,
      "eval_runtime": 4727.7868,
      "eval_samples_per_second": 42.907,
      "eval_steps_per_second": 2.682,
      "step": 768
    },
    {
      "epoch": 0.12130772567732775,
      "grad_norm": 6.041478633880615,
      "learning_rate": 8.967960070842054e-07,
      "loss": 0.0674,
      "step": 769
    },
    {
      "epoch": 0.12146547304491856,
      "grad_norm": 5.090635776519775,
      "learning_rate": 8.966350024150699e-07,
      "loss": 0.043,
      "step": 770
    },
    {
      "epoch": 0.12162322041250936,
      "grad_norm": 8.246053695678711,
      "learning_rate": 8.964739977459346e-07,
      "loss": 0.0305,
      "step": 771
    },
    {
      "epoch": 0.12178096778010017,
      "grad_norm": 8.57614517211914,
      "learning_rate": 8.963129930767992e-07,
      "loss": 0.1028,
      "step": 772
    },
    {
      "epoch": 0.12193871514769097,
      "grad_norm": 7.169241428375244,
      "learning_rate": 8.961519884076638e-07,
      "loss": 0.0653,
      "step": 773
    },
    {
      "epoch": 0.12209646251528178,
      "grad_norm": 5.16396951675415,
      "learning_rate": 8.959909837385283e-07,
      "loss": 0.0891,
      "step": 774
    },
    {
      "epoch": 0.12225420988287258,
      "grad_norm": 6.805628299713135,
      "learning_rate": 8.958299790693929e-07,
      "loss": 0.0815,
      "step": 775
    },
    {
      "epoch": 0.12241195725046339,
      "grad_norm": 7.052548885345459,
      "learning_rate": 8.956689744002575e-07,
      "loss": 0.0992,
      "step": 776
    },
    {
      "epoch": 0.12256970461805419,
      "grad_norm": 3.9773361682891846,
      "learning_rate": 8.955079697311221e-07,
      "loss": 0.0624,
      "step": 777
    },
    {
      "epoch": 0.12272745198564498,
      "grad_norm": 6.479729175567627,
      "learning_rate": 8.953469650619868e-07,
      "loss": 0.0691,
      "step": 778
    },
    {
      "epoch": 0.12288519935323579,
      "grad_norm": 7.385898113250732,
      "learning_rate": 8.951859603928514e-07,
      "loss": 0.1458,
      "step": 779
    },
    {
      "epoch": 0.1230429467208266,
      "grad_norm": 6.460968017578125,
      "learning_rate": 8.95024955723716e-07,
      "loss": 0.0868,
      "step": 780
    },
    {
      "epoch": 0.1232006940884174,
      "grad_norm": 7.318996906280518,
      "learning_rate": 8.948639510545806e-07,
      "loss": 0.088,
      "step": 781
    },
    {
      "epoch": 0.1233584414560082,
      "grad_norm": 7.164853572845459,
      "learning_rate": 8.947029463854452e-07,
      "loss": 0.0804,
      "step": 782
    },
    {
      "epoch": 0.12351618882359901,
      "grad_norm": 11.578691482543945,
      "learning_rate": 8.945419417163097e-07,
      "loss": 0.075,
      "step": 783
    },
    {
      "epoch": 0.12367393619118981,
      "grad_norm": 2.8673319816589355,
      "learning_rate": 8.943809370471743e-07,
      "loss": 0.0567,
      "step": 784
    },
    {
      "epoch": 0.12383168355878062,
      "grad_norm": 7.566540241241455,
      "learning_rate": 8.942199323780389e-07,
      "loss": 0.0914,
      "step": 785
    },
    {
      "epoch": 0.12398943092637142,
      "grad_norm": 5.447342395782471,
      "learning_rate": 8.940589277089036e-07,
      "loss": 0.0883,
      "step": 786
    },
    {
      "epoch": 0.12414717829396221,
      "grad_norm": 3.222623348236084,
      "learning_rate": 8.938979230397681e-07,
      "loss": 0.0385,
      "step": 787
    },
    {
      "epoch": 0.12430492566155302,
      "grad_norm": 12.599985122680664,
      "learning_rate": 8.937369183706327e-07,
      "loss": 0.1081,
      "step": 788
    },
    {
      "epoch": 0.12446267302914382,
      "grad_norm": 4.441560745239258,
      "learning_rate": 8.935759137014973e-07,
      "loss": 0.0628,
      "step": 789
    },
    {
      "epoch": 0.12462042039673463,
      "grad_norm": 5.921494483947754,
      "learning_rate": 8.934149090323619e-07,
      "loss": 0.1036,
      "step": 790
    },
    {
      "epoch": 0.12477816776432543,
      "grad_norm": 8.730775833129883,
      "learning_rate": 8.932539043632264e-07,
      "loss": 0.0895,
      "step": 791
    },
    {
      "epoch": 0.12493591513191624,
      "grad_norm": 4.8975911140441895,
      "learning_rate": 8.93092899694091e-07,
      "loss": 0.0645,
      "step": 792
    },
    {
      "epoch": 0.12509366249950704,
      "grad_norm": 4.241909503936768,
      "learning_rate": 8.929318950249557e-07,
      "loss": 0.0622,
      "step": 793
    },
    {
      "epoch": 0.12525140986709785,
      "grad_norm": 6.301782131195068,
      "learning_rate": 8.927708903558203e-07,
      "loss": 0.0722,
      "step": 794
    },
    {
      "epoch": 0.12540915723468865,
      "grad_norm": 2.9589102268218994,
      "learning_rate": 8.926098856866848e-07,
      "loss": 0.0362,
      "step": 795
    },
    {
      "epoch": 0.12556690460227946,
      "grad_norm": 6.449861526489258,
      "learning_rate": 8.924488810175495e-07,
      "loss": 0.0808,
      "step": 796
    },
    {
      "epoch": 0.12572465196987026,
      "grad_norm": 4.228991508483887,
      "learning_rate": 8.922878763484141e-07,
      "loss": 0.0605,
      "step": 797
    },
    {
      "epoch": 0.12588239933746107,
      "grad_norm": 9.525351524353027,
      "learning_rate": 8.921268716792787e-07,
      "loss": 0.1215,
      "step": 798
    },
    {
      "epoch": 0.12604014670505187,
      "grad_norm": 5.587832450866699,
      "learning_rate": 8.919658670101432e-07,
      "loss": 0.0933,
      "step": 799
    },
    {
      "epoch": 0.12619789407264267,
      "grad_norm": 3.7256529331207275,
      "learning_rate": 8.918048623410078e-07,
      "loss": 0.0882,
      "step": 800
    },
    {
      "epoch": 0.12635564144023348,
      "grad_norm": 4.9349236488342285,
      "learning_rate": 8.916438576718725e-07,
      "loss": 0.0573,
      "step": 801
    },
    {
      "epoch": 0.12651338880782426,
      "grad_norm": 7.595797061920166,
      "learning_rate": 8.914828530027371e-07,
      "loss": 0.0727,
      "step": 802
    },
    {
      "epoch": 0.12667113617541506,
      "grad_norm": 8.809452056884766,
      "learning_rate": 8.913218483336017e-07,
      "loss": 0.098,
      "step": 803
    },
    {
      "epoch": 0.12682888354300587,
      "grad_norm": 4.205000877380371,
      "learning_rate": 8.911608436644662e-07,
      "loss": 0.0784,
      "step": 804
    },
    {
      "epoch": 0.12698663091059667,
      "grad_norm": 3.946079969406128,
      "learning_rate": 8.909998389953308e-07,
      "loss": 0.0487,
      "step": 805
    },
    {
      "epoch": 0.12714437827818748,
      "grad_norm": 7.7502570152282715,
      "learning_rate": 8.908388343261954e-07,
      "loss": 0.0747,
      "step": 806
    },
    {
      "epoch": 0.12730212564577828,
      "grad_norm": 4.870746612548828,
      "learning_rate": 8.9067782965706e-07,
      "loss": 0.0653,
      "step": 807
    },
    {
      "epoch": 0.12745987301336908,
      "grad_norm": 5.179272174835205,
      "learning_rate": 8.905168249879246e-07,
      "loss": 0.0597,
      "step": 808
    },
    {
      "epoch": 0.1276176203809599,
      "grad_norm": 5.307404518127441,
      "learning_rate": 8.903558203187892e-07,
      "loss": 0.0847,
      "step": 809
    },
    {
      "epoch": 0.1277753677485507,
      "grad_norm": 6.88541841506958,
      "learning_rate": 8.901948156496538e-07,
      "loss": 0.0781,
      "step": 810
    },
    {
      "epoch": 0.1279331151161415,
      "grad_norm": 10.111740112304688,
      "learning_rate": 8.900338109805184e-07,
      "loss": 0.0569,
      "step": 811
    },
    {
      "epoch": 0.1280908624837323,
      "grad_norm": 6.572051048278809,
      "learning_rate": 8.898728063113829e-07,
      "loss": 0.0949,
      "step": 812
    },
    {
      "epoch": 0.1282486098513231,
      "grad_norm": 7.902374744415283,
      "learning_rate": 8.897118016422476e-07,
      "loss": 0.0693,
      "step": 813
    },
    {
      "epoch": 0.1284063572189139,
      "grad_norm": 7.52032995223999,
      "learning_rate": 8.895507969731122e-07,
      "loss": 0.046,
      "step": 814
    },
    {
      "epoch": 0.12856410458650472,
      "grad_norm": 4.0166239738464355,
      "learning_rate": 8.893897923039768e-07,
      "loss": 0.0889,
      "step": 815
    },
    {
      "epoch": 0.12872185195409552,
      "grad_norm": 7.301949977874756,
      "learning_rate": 8.892287876348414e-07,
      "loss": 0.0758,
      "step": 816
    },
    {
      "epoch": 0.12887959932168633,
      "grad_norm": 3.707253932952881,
      "learning_rate": 8.89067782965706e-07,
      "loss": 0.0677,
      "step": 817
    },
    {
      "epoch": 0.12903734668927713,
      "grad_norm": 4.05390739440918,
      "learning_rate": 8.889067782965706e-07,
      "loss": 0.0543,
      "step": 818
    },
    {
      "epoch": 0.12919509405686794,
      "grad_norm": 3.006601095199585,
      "learning_rate": 8.887457736274352e-07,
      "loss": 0.031,
      "step": 819
    },
    {
      "epoch": 0.12935284142445874,
      "grad_norm": 6.196488857269287,
      "learning_rate": 8.885847689582997e-07,
      "loss": 0.129,
      "step": 820
    },
    {
      "epoch": 0.12951058879204952,
      "grad_norm": 4.589459419250488,
      "learning_rate": 8.884237642891643e-07,
      "loss": 0.0635,
      "step": 821
    },
    {
      "epoch": 0.12966833615964032,
      "grad_norm": 7.604683876037598,
      "learning_rate": 8.882627596200289e-07,
      "loss": 0.0734,
      "step": 822
    },
    {
      "epoch": 0.12982608352723113,
      "grad_norm": 6.847631454467773,
      "learning_rate": 8.881017549508936e-07,
      "loss": 0.0493,
      "step": 823
    },
    {
      "epoch": 0.12998383089482193,
      "grad_norm": 5.853129863739014,
      "learning_rate": 8.879407502817581e-07,
      "loss": 0.0716,
      "step": 824
    },
    {
      "epoch": 0.13014157826241274,
      "grad_norm": 8.411120414733887,
      "learning_rate": 8.877797456126227e-07,
      "loss": 0.111,
      "step": 825
    },
    {
      "epoch": 0.13029932563000354,
      "grad_norm": 10.987421989440918,
      "learning_rate": 8.876187409434873e-07,
      "loss": 0.118,
      "step": 826
    },
    {
      "epoch": 0.13045707299759435,
      "grad_norm": 4.021767616271973,
      "learning_rate": 8.874577362743519e-07,
      "loss": 0.066,
      "step": 827
    },
    {
      "epoch": 0.13061482036518515,
      "grad_norm": 9.605419158935547,
      "learning_rate": 8.872967316052165e-07,
      "loss": 0.0765,
      "step": 828
    },
    {
      "epoch": 0.13077256773277596,
      "grad_norm": 7.201488018035889,
      "learning_rate": 8.87135726936081e-07,
      "loss": 0.0735,
      "step": 829
    },
    {
      "epoch": 0.13093031510036676,
      "grad_norm": 5.675152778625488,
      "learning_rate": 8.869747222669456e-07,
      "loss": 0.0825,
      "step": 830
    },
    {
      "epoch": 0.13108806246795757,
      "grad_norm": 6.44282865524292,
      "learning_rate": 8.868137175978104e-07,
      "loss": 0.0875,
      "step": 831
    },
    {
      "epoch": 0.13124580983554837,
      "grad_norm": 5.081498146057129,
      "learning_rate": 8.86652712928675e-07,
      "loss": 0.0719,
      "step": 832
    },
    {
      "epoch": 0.13140355720313918,
      "grad_norm": 7.329372882843018,
      "learning_rate": 8.864917082595395e-07,
      "loss": 0.0707,
      "step": 833
    },
    {
      "epoch": 0.13156130457072998,
      "grad_norm": 5.028716564178467,
      "learning_rate": 8.863307035904041e-07,
      "loss": 0.0587,
      "step": 834
    },
    {
      "epoch": 0.13171905193832079,
      "grad_norm": 4.705847263336182,
      "learning_rate": 8.861696989212687e-07,
      "loss": 0.0481,
      "step": 835
    },
    {
      "epoch": 0.1318767993059116,
      "grad_norm": 7.79337215423584,
      "learning_rate": 8.860086942521333e-07,
      "loss": 0.1407,
      "step": 836
    },
    {
      "epoch": 0.1320345466735024,
      "grad_norm": 4.879453182220459,
      "learning_rate": 8.858476895829978e-07,
      "loss": 0.0852,
      "step": 837
    },
    {
      "epoch": 0.1321922940410932,
      "grad_norm": 4.048136234283447,
      "learning_rate": 8.856866849138625e-07,
      "loss": 0.0419,
      "step": 838
    },
    {
      "epoch": 0.132350041408684,
      "grad_norm": 5.209859371185303,
      "learning_rate": 8.855256802447271e-07,
      "loss": 0.0496,
      "step": 839
    },
    {
      "epoch": 0.1325077887762748,
      "grad_norm": 6.314807891845703,
      "learning_rate": 8.853646755755917e-07,
      "loss": 0.1217,
      "step": 840
    },
    {
      "epoch": 0.1326655361438656,
      "grad_norm": 5.608424663543701,
      "learning_rate": 8.852036709064562e-07,
      "loss": 0.0795,
      "step": 841
    },
    {
      "epoch": 0.1328232835114564,
      "grad_norm": 8.808143615722656,
      "learning_rate": 8.850426662373208e-07,
      "loss": 0.0573,
      "step": 842
    },
    {
      "epoch": 0.1329810308790472,
      "grad_norm": 6.290168762207031,
      "learning_rate": 8.848816615681854e-07,
      "loss": 0.0691,
      "step": 843
    },
    {
      "epoch": 0.133138778246638,
      "grad_norm": 7.928153038024902,
      "learning_rate": 8.8472065689905e-07,
      "loss": 0.0669,
      "step": 844
    },
    {
      "epoch": 0.1332965256142288,
      "grad_norm": 10.569119453430176,
      "learning_rate": 8.845596522299145e-07,
      "loss": 0.1071,
      "step": 845
    },
    {
      "epoch": 0.1334542729818196,
      "grad_norm": 8.861801147460938,
      "learning_rate": 8.843986475607792e-07,
      "loss": 0.071,
      "step": 846
    },
    {
      "epoch": 0.13361202034941042,
      "grad_norm": 8.680996894836426,
      "learning_rate": 8.842376428916438e-07,
      "loss": 0.0885,
      "step": 847
    },
    {
      "epoch": 0.13376976771700122,
      "grad_norm": 4.941028118133545,
      "learning_rate": 8.840766382225085e-07,
      "loss": 0.0677,
      "step": 848
    },
    {
      "epoch": 0.13392751508459202,
      "grad_norm": 4.489545822143555,
      "learning_rate": 8.83915633553373e-07,
      "loss": 0.04,
      "step": 849
    },
    {
      "epoch": 0.13408526245218283,
      "grad_norm": 10.692946434020996,
      "learning_rate": 8.837546288842376e-07,
      "loss": 0.0641,
      "step": 850
    },
    {
      "epoch": 0.13424300981977363,
      "grad_norm": 7.841832160949707,
      "learning_rate": 8.835936242151022e-07,
      "loss": 0.0606,
      "step": 851
    },
    {
      "epoch": 0.13440075718736444,
      "grad_norm": 7.537064552307129,
      "learning_rate": 8.834326195459668e-07,
      "loss": 0.0785,
      "step": 852
    },
    {
      "epoch": 0.13455850455495524,
      "grad_norm": 8.915136337280273,
      "learning_rate": 8.832716148768315e-07,
      "loss": 0.0517,
      "step": 853
    },
    {
      "epoch": 0.13471625192254605,
      "grad_norm": 7.323801517486572,
      "learning_rate": 8.83110610207696e-07,
      "loss": 0.0766,
      "step": 854
    },
    {
      "epoch": 0.13487399929013685,
      "grad_norm": 4.316955089569092,
      "learning_rate": 8.829496055385606e-07,
      "loss": 0.0428,
      "step": 855
    },
    {
      "epoch": 0.13503174665772766,
      "grad_norm": 8.194013595581055,
      "learning_rate": 8.827886008694252e-07,
      "loss": 0.0681,
      "step": 856
    },
    {
      "epoch": 0.13518949402531846,
      "grad_norm": 6.799248218536377,
      "learning_rate": 8.826275962002898e-07,
      "loss": 0.0697,
      "step": 857
    },
    {
      "epoch": 0.13534724139290927,
      "grad_norm": 11.271206855773926,
      "learning_rate": 8.824665915311543e-07,
      "loss": 0.0967,
      "step": 858
    },
    {
      "epoch": 0.13550498876050007,
      "grad_norm": 4.689577102661133,
      "learning_rate": 8.823055868620189e-07,
      "loss": 0.0548,
      "step": 859
    },
    {
      "epoch": 0.13566273612809085,
      "grad_norm": 5.299933910369873,
      "learning_rate": 8.821445821928835e-07,
      "loss": 0.0993,
      "step": 860
    },
    {
      "epoch": 0.13582048349568165,
      "grad_norm": 6.0095601081848145,
      "learning_rate": 8.819835775237482e-07,
      "loss": 0.0766,
      "step": 861
    },
    {
      "epoch": 0.13597823086327246,
      "grad_norm": 9.537050247192383,
      "learning_rate": 8.818225728546127e-07,
      "loss": 0.0891,
      "step": 862
    },
    {
      "epoch": 0.13613597823086326,
      "grad_norm": 5.553176403045654,
      "learning_rate": 8.816615681854773e-07,
      "loss": 0.0674,
      "step": 863
    },
    {
      "epoch": 0.13629372559845407,
      "grad_norm": 4.781669616699219,
      "learning_rate": 8.815005635163419e-07,
      "loss": 0.03,
      "step": 864
    },
    {
      "epoch": 0.13645147296604487,
      "grad_norm": 5.221872329711914,
      "learning_rate": 8.813395588472065e-07,
      "loss": 0.0742,
      "step": 865
    },
    {
      "epoch": 0.13660922033363568,
      "grad_norm": 4.754615783691406,
      "learning_rate": 8.81178554178071e-07,
      "loss": 0.0593,
      "step": 866
    },
    {
      "epoch": 0.13676696770122648,
      "grad_norm": 3.2132647037506104,
      "learning_rate": 8.810175495089357e-07,
      "loss": 0.0935,
      "step": 867
    },
    {
      "epoch": 0.1369247150688173,
      "grad_norm": 3.5838680267333984,
      "learning_rate": 8.808565448398004e-07,
      "loss": 0.0559,
      "step": 868
    },
    {
      "epoch": 0.1370824624364081,
      "grad_norm": 2.9936325550079346,
      "learning_rate": 8.80695540170665e-07,
      "loss": 0.0246,
      "step": 869
    },
    {
      "epoch": 0.1372402098039989,
      "grad_norm": 10.4288330078125,
      "learning_rate": 8.805345355015295e-07,
      "loss": 0.1194,
      "step": 870
    },
    {
      "epoch": 0.1373979571715897,
      "grad_norm": 5.2311110496521,
      "learning_rate": 8.803735308323941e-07,
      "loss": 0.0953,
      "step": 871
    },
    {
      "epoch": 0.1375557045391805,
      "grad_norm": 1.4418354034423828,
      "learning_rate": 8.802125261632587e-07,
      "loss": 0.0196,
      "step": 872
    },
    {
      "epoch": 0.1377134519067713,
      "grad_norm": 5.885601997375488,
      "learning_rate": 8.800515214941233e-07,
      "loss": 0.1315,
      "step": 873
    },
    {
      "epoch": 0.13787119927436212,
      "grad_norm": 4.334197998046875,
      "learning_rate": 8.798905168249879e-07,
      "loss": 0.0832,
      "step": 874
    },
    {
      "epoch": 0.13802894664195292,
      "grad_norm": 4.670304775238037,
      "learning_rate": 8.797295121558524e-07,
      "loss": 0.0383,
      "step": 875
    },
    {
      "epoch": 0.13818669400954373,
      "grad_norm": 7.2617669105529785,
      "learning_rate": 8.795685074867171e-07,
      "loss": 0.1078,
      "step": 876
    },
    {
      "epoch": 0.13834444137713453,
      "grad_norm": 5.498252868652344,
      "learning_rate": 8.794075028175817e-07,
      "loss": 0.0972,
      "step": 877
    },
    {
      "epoch": 0.13850218874472534,
      "grad_norm": 6.531783103942871,
      "learning_rate": 8.792464981484463e-07,
      "loss": 0.0993,
      "step": 878
    },
    {
      "epoch": 0.1386599361123161,
      "grad_norm": 4.48145055770874,
      "learning_rate": 8.790854934793108e-07,
      "loss": 0.0363,
      "step": 879
    },
    {
      "epoch": 0.13881768347990692,
      "grad_norm": 3.881742000579834,
      "learning_rate": 8.789244888101754e-07,
      "loss": 0.0352,
      "step": 880
    },
    {
      "epoch": 0.13897543084749772,
      "grad_norm": 2.8471145629882812,
      "learning_rate": 8.7876348414104e-07,
      "loss": 0.0559,
      "step": 881
    },
    {
      "epoch": 0.13913317821508853,
      "grad_norm": 4.34542179107666,
      "learning_rate": 8.786024794719046e-07,
      "loss": 0.037,
      "step": 882
    },
    {
      "epoch": 0.13929092558267933,
      "grad_norm": 7.625082015991211,
      "learning_rate": 8.784414748027693e-07,
      "loss": 0.061,
      "step": 883
    },
    {
      "epoch": 0.13944867295027014,
      "grad_norm": 5.236334323883057,
      "learning_rate": 8.782804701336339e-07,
      "loss": 0.0552,
      "step": 884
    },
    {
      "epoch": 0.13960642031786094,
      "grad_norm": 5.2504963874816895,
      "learning_rate": 8.781194654644985e-07,
      "loss": 0.0959,
      "step": 885
    },
    {
      "epoch": 0.13976416768545175,
      "grad_norm": 4.942289352416992,
      "learning_rate": 8.779584607953631e-07,
      "loss": 0.0733,
      "step": 886
    },
    {
      "epoch": 0.13992191505304255,
      "grad_norm": 5.178107738494873,
      "learning_rate": 8.777974561262276e-07,
      "loss": 0.0996,
      "step": 887
    },
    {
      "epoch": 0.14007966242063336,
      "grad_norm": 5.820649147033691,
      "learning_rate": 8.776364514570922e-07,
      "loss": 0.0443,
      "step": 888
    },
    {
      "epoch": 0.14023740978822416,
      "grad_norm": 3.677860736846924,
      "learning_rate": 8.774754467879568e-07,
      "loss": 0.055,
      "step": 889
    },
    {
      "epoch": 0.14039515715581496,
      "grad_norm": 3.097083568572998,
      "learning_rate": 8.773144421188214e-07,
      "loss": 0.0293,
      "step": 890
    },
    {
      "epoch": 0.14055290452340577,
      "grad_norm": 5.426209926605225,
      "learning_rate": 8.77153437449686e-07,
      "loss": 0.0501,
      "step": 891
    },
    {
      "epoch": 0.14071065189099657,
      "grad_norm": 8.975652694702148,
      "learning_rate": 8.769924327805506e-07,
      "loss": 0.0475,
      "step": 892
    },
    {
      "epoch": 0.14086839925858738,
      "grad_norm": 3.8866279125213623,
      "learning_rate": 8.768314281114152e-07,
      "loss": 0.0776,
      "step": 893
    },
    {
      "epoch": 0.14102614662617818,
      "grad_norm": 5.4157280921936035,
      "learning_rate": 8.766704234422798e-07,
      "loss": 0.0702,
      "step": 894
    },
    {
      "epoch": 0.141183893993769,
      "grad_norm": 7.1885600090026855,
      "learning_rate": 8.765094187731443e-07,
      "loss": 0.0748,
      "step": 895
    },
    {
      "epoch": 0.1413416413613598,
      "grad_norm": 10.391587257385254,
      "learning_rate": 8.763484141040089e-07,
      "loss": 0.1019,
      "step": 896
    },
    {
      "epoch": 0.1414993887289506,
      "grad_norm": 8.128127098083496,
      "learning_rate": 8.761874094348735e-07,
      "loss": 0.1077,
      "step": 897
    },
    {
      "epoch": 0.1416571360965414,
      "grad_norm": 7.078885078430176,
      "learning_rate": 8.760264047657382e-07,
      "loss": 0.0775,
      "step": 898
    },
    {
      "epoch": 0.14181488346413218,
      "grad_norm": 5.855307579040527,
      "learning_rate": 8.758654000966028e-07,
      "loss": 0.0534,
      "step": 899
    },
    {
      "epoch": 0.14197263083172298,
      "grad_norm": 8.8378324508667,
      "learning_rate": 8.757043954274673e-07,
      "loss": 0.1473,
      "step": 900
    },
    {
      "epoch": 0.1421303781993138,
      "grad_norm": 6.6874518394470215,
      "learning_rate": 8.75543390758332e-07,
      "loss": 0.0767,
      "step": 901
    },
    {
      "epoch": 0.1422881255669046,
      "grad_norm": 4.136755466461182,
      "learning_rate": 8.753823860891966e-07,
      "loss": 0.0864,
      "step": 902
    },
    {
      "epoch": 0.1424458729344954,
      "grad_norm": 10.590389251708984,
      "learning_rate": 8.752213814200612e-07,
      "loss": 0.0944,
      "step": 903
    },
    {
      "epoch": 0.1426036203020862,
      "grad_norm": 9.044832229614258,
      "learning_rate": 8.750603767509257e-07,
      "loss": 0.1435,
      "step": 904
    },
    {
      "epoch": 0.142761367669677,
      "grad_norm": 6.6368021965026855,
      "learning_rate": 8.748993720817903e-07,
      "loss": 0.0932,
      "step": 905
    },
    {
      "epoch": 0.1429191150372678,
      "grad_norm": 6.0270562171936035,
      "learning_rate": 8.74738367412655e-07,
      "loss": 0.0668,
      "step": 906
    },
    {
      "epoch": 0.14307686240485862,
      "grad_norm": 5.822785377502441,
      "learning_rate": 8.745773627435196e-07,
      "loss": 0.0734,
      "step": 907
    },
    {
      "epoch": 0.14323460977244942,
      "grad_norm": 5.617862701416016,
      "learning_rate": 8.744163580743841e-07,
      "loss": 0.0667,
      "step": 908
    },
    {
      "epoch": 0.14339235714004023,
      "grad_norm": 7.765786647796631,
      "learning_rate": 8.742553534052487e-07,
      "loss": 0.0538,
      "step": 909
    },
    {
      "epoch": 0.14355010450763103,
      "grad_norm": 5.429433822631836,
      "learning_rate": 8.740943487361133e-07,
      "loss": 0.0769,
      "step": 910
    },
    {
      "epoch": 0.14370785187522184,
      "grad_norm": 4.444888591766357,
      "learning_rate": 8.739333440669779e-07,
      "loss": 0.053,
      "step": 911
    },
    {
      "epoch": 0.14386559924281264,
      "grad_norm": 8.31460189819336,
      "learning_rate": 8.737723393978424e-07,
      "loss": 0.096,
      "step": 912
    },
    {
      "epoch": 0.14402334661040345,
      "grad_norm": 8.856794357299805,
      "learning_rate": 8.736113347287071e-07,
      "loss": 0.084,
      "step": 913
    },
    {
      "epoch": 0.14418109397799425,
      "grad_norm": 5.17337703704834,
      "learning_rate": 8.734503300595717e-07,
      "loss": 0.0826,
      "step": 914
    },
    {
      "epoch": 0.14433884134558506,
      "grad_norm": 5.695919990539551,
      "learning_rate": 8.732893253904363e-07,
      "loss": 0.0729,
      "step": 915
    },
    {
      "epoch": 0.14449658871317586,
      "grad_norm": 2.785297393798828,
      "learning_rate": 8.731283207213008e-07,
      "loss": 0.0503,
      "step": 916
    },
    {
      "epoch": 0.14465433608076667,
      "grad_norm": 4.552045822143555,
      "learning_rate": 8.729673160521654e-07,
      "loss": 0.0618,
      "step": 917
    },
    {
      "epoch": 0.14481208344835744,
      "grad_norm": 11.460582733154297,
      "learning_rate": 8.7280631138303e-07,
      "loss": 0.1406,
      "step": 918
    },
    {
      "epoch": 0.14496983081594825,
      "grad_norm": 7.194328308105469,
      "learning_rate": 8.726453067138947e-07,
      "loss": 0.1099,
      "step": 919
    },
    {
      "epoch": 0.14512757818353905,
      "grad_norm": 5.801459312438965,
      "learning_rate": 8.724843020447593e-07,
      "loss": 0.0688,
      "step": 920
    },
    {
      "epoch": 0.14528532555112986,
      "grad_norm": 8.886085510253906,
      "learning_rate": 8.723232973756239e-07,
      "loss": 0.1025,
      "step": 921
    },
    {
      "epoch": 0.14544307291872066,
      "grad_norm": 6.849817276000977,
      "learning_rate": 8.721622927064885e-07,
      "loss": 0.1059,
      "step": 922
    },
    {
      "epoch": 0.14560082028631147,
      "grad_norm": 3.839834213256836,
      "learning_rate": 8.720012880373531e-07,
      "loss": 0.0568,
      "step": 923
    },
    {
      "epoch": 0.14575856765390227,
      "grad_norm": 11.36760139465332,
      "learning_rate": 8.718402833682177e-07,
      "loss": 0.1034,
      "step": 924
    },
    {
      "epoch": 0.14591631502149308,
      "grad_norm": 6.963069915771484,
      "learning_rate": 8.716792786990822e-07,
      "loss": 0.0885,
      "step": 925
    },
    {
      "epoch": 0.14607406238908388,
      "grad_norm": 13.822622299194336,
      "learning_rate": 8.715182740299468e-07,
      "loss": 0.0581,
      "step": 926
    },
    {
      "epoch": 0.14623180975667469,
      "grad_norm": 3.9957492351531982,
      "learning_rate": 8.713572693608114e-07,
      "loss": 0.0641,
      "step": 927
    },
    {
      "epoch": 0.1463895571242655,
      "grad_norm": 5.5053229331970215,
      "learning_rate": 8.711962646916761e-07,
      "loss": 0.0436,
      "step": 928
    },
    {
      "epoch": 0.1465473044918563,
      "grad_norm": 5.216220378875732,
      "learning_rate": 8.710352600225406e-07,
      "loss": 0.0793,
      "step": 929
    },
    {
      "epoch": 0.1467050518594471,
      "grad_norm": 5.825531005859375,
      "learning_rate": 8.708742553534052e-07,
      "loss": 0.1178,
      "step": 930
    },
    {
      "epoch": 0.1468627992270379,
      "grad_norm": 6.612665176391602,
      "learning_rate": 8.707132506842698e-07,
      "loss": 0.0723,
      "step": 931
    },
    {
      "epoch": 0.1470205465946287,
      "grad_norm": 25.09065055847168,
      "learning_rate": 8.705522460151344e-07,
      "loss": 0.0635,
      "step": 932
    },
    {
      "epoch": 0.14717829396221951,
      "grad_norm": 5.503942966461182,
      "learning_rate": 8.703912413459989e-07,
      "loss": 0.0533,
      "step": 933
    },
    {
      "epoch": 0.14733604132981032,
      "grad_norm": 5.840237140655518,
      "learning_rate": 8.702302366768635e-07,
      "loss": 0.0705,
      "step": 934
    },
    {
      "epoch": 0.14749378869740112,
      "grad_norm": 4.470847129821777,
      "learning_rate": 8.700692320077281e-07,
      "loss": 0.0629,
      "step": 935
    },
    {
      "epoch": 0.14765153606499193,
      "grad_norm": 13.115463256835938,
      "learning_rate": 8.699082273385929e-07,
      "loss": 0.1077,
      "step": 936
    },
    {
      "epoch": 0.1478092834325827,
      "grad_norm": 7.014602184295654,
      "learning_rate": 8.697472226694574e-07,
      "loss": 0.0464,
      "step": 937
    },
    {
      "epoch": 0.1479670308001735,
      "grad_norm": 10.690193176269531,
      "learning_rate": 8.69586218000322e-07,
      "loss": 0.1457,
      "step": 938
    },
    {
      "epoch": 0.14812477816776431,
      "grad_norm": 7.957321643829346,
      "learning_rate": 8.694252133311866e-07,
      "loss": 0.0614,
      "step": 939
    },
    {
      "epoch": 0.14828252553535512,
      "grad_norm": 5.306655406951904,
      "learning_rate": 8.692642086620512e-07,
      "loss": 0.0963,
      "step": 940
    },
    {
      "epoch": 0.14844027290294592,
      "grad_norm": 6.521378040313721,
      "learning_rate": 8.691032039929157e-07,
      "loss": 0.0836,
      "step": 941
    },
    {
      "epoch": 0.14859802027053673,
      "grad_norm": 6.258962631225586,
      "learning_rate": 8.689421993237803e-07,
      "loss": 0.0904,
      "step": 942
    },
    {
      "epoch": 0.14875576763812753,
      "grad_norm": 5.522530555725098,
      "learning_rate": 8.68781194654645e-07,
      "loss": 0.0627,
      "step": 943
    },
    {
      "epoch": 0.14891351500571834,
      "grad_norm": 3.896024465560913,
      "learning_rate": 8.686201899855096e-07,
      "loss": 0.0576,
      "step": 944
    },
    {
      "epoch": 0.14907126237330914,
      "grad_norm": 8.584480285644531,
      "learning_rate": 8.684591853163742e-07,
      "loss": 0.0483,
      "step": 945
    },
    {
      "epoch": 0.14922900974089995,
      "grad_norm": 4.397729873657227,
      "learning_rate": 8.682981806472387e-07,
      "loss": 0.053,
      "step": 946
    },
    {
      "epoch": 0.14938675710849075,
      "grad_norm": 4.222675323486328,
      "learning_rate": 8.681371759781033e-07,
      "loss": 0.0363,
      "step": 947
    },
    {
      "epoch": 0.14954450447608156,
      "grad_norm": 6.897716522216797,
      "learning_rate": 8.679761713089679e-07,
      "loss": 0.0565,
      "step": 948
    },
    {
      "epoch": 0.14970225184367236,
      "grad_norm": 3.7198383808135986,
      "learning_rate": 8.678151666398325e-07,
      "loss": 0.0737,
      "step": 949
    },
    {
      "epoch": 0.14985999921126317,
      "grad_norm": 5.227583885192871,
      "learning_rate": 8.67654161970697e-07,
      "loss": 0.0567,
      "step": 950
    },
    {
      "epoch": 0.15001774657885397,
      "grad_norm": 4.30453634262085,
      "learning_rate": 8.674931573015617e-07,
      "loss": 0.0723,
      "step": 951
    },
    {
      "epoch": 0.15017549394644478,
      "grad_norm": 7.840729713439941,
      "learning_rate": 8.673321526324263e-07,
      "loss": 0.0971,
      "step": 952
    },
    {
      "epoch": 0.15033324131403558,
      "grad_norm": 5.92392110824585,
      "learning_rate": 8.67171147963291e-07,
      "loss": 0.0314,
      "step": 953
    },
    {
      "epoch": 0.1504909886816264,
      "grad_norm": 4.194683074951172,
      "learning_rate": 8.670101432941555e-07,
      "loss": 0.0544,
      "step": 954
    },
    {
      "epoch": 0.1506487360492172,
      "grad_norm": 2.73091721534729,
      "learning_rate": 8.668491386250201e-07,
      "loss": 0.0282,
      "step": 955
    },
    {
      "epoch": 0.15080648341680797,
      "grad_norm": 4.007771015167236,
      "learning_rate": 8.666881339558847e-07,
      "loss": 0.0445,
      "step": 956
    },
    {
      "epoch": 0.15096423078439877,
      "grad_norm": 4.4001054763793945,
      "learning_rate": 8.665271292867493e-07,
      "loss": 0.0619,
      "step": 957
    },
    {
      "epoch": 0.15112197815198958,
      "grad_norm": 5.6491522789001465,
      "learning_rate": 8.663661246176138e-07,
      "loss": 0.0527,
      "step": 958
    },
    {
      "epoch": 0.15127972551958038,
      "grad_norm": 4.243597984313965,
      "learning_rate": 8.662051199484785e-07,
      "loss": 0.0589,
      "step": 959
    },
    {
      "epoch": 0.1514374728871712,
      "grad_norm": 3.485116720199585,
      "learning_rate": 8.660441152793431e-07,
      "loss": 0.0274,
      "step": 960
    },
    {
      "epoch": 0.151595220254762,
      "grad_norm": 4.097177505493164,
      "learning_rate": 8.658831106102077e-07,
      "loss": 0.0724,
      "step": 961
    },
    {
      "epoch": 0.1517529676223528,
      "grad_norm": 3.5488181114196777,
      "learning_rate": 8.657221059410722e-07,
      "loss": 0.0375,
      "step": 962
    },
    {
      "epoch": 0.1519107149899436,
      "grad_norm": 3.5617055892944336,
      "learning_rate": 8.655611012719368e-07,
      "loss": 0.0635,
      "step": 963
    },
    {
      "epoch": 0.1520684623575344,
      "grad_norm": 5.917942047119141,
      "learning_rate": 8.654000966028014e-07,
      "loss": 0.0732,
      "step": 964
    },
    {
      "epoch": 0.1522262097251252,
      "grad_norm": 7.556006908416748,
      "learning_rate": 8.65239091933666e-07,
      "loss": 0.0382,
      "step": 965
    },
    {
      "epoch": 0.15238395709271602,
      "grad_norm": 10.01443862915039,
      "learning_rate": 8.650780872645307e-07,
      "loss": 0.0893,
      "step": 966
    },
    {
      "epoch": 0.15254170446030682,
      "grad_norm": 8.719974517822266,
      "learning_rate": 8.649170825953952e-07,
      "loss": 0.0729,
      "step": 967
    },
    {
      "epoch": 0.15269945182789763,
      "grad_norm": 9.003377914428711,
      "learning_rate": 8.647560779262598e-07,
      "loss": 0.1022,
      "step": 968
    },
    {
      "epoch": 0.15285719919548843,
      "grad_norm": 6.390311241149902,
      "learning_rate": 8.645950732571244e-07,
      "loss": 0.1335,
      "step": 969
    },
    {
      "epoch": 0.15301494656307923,
      "grad_norm": 14.419649124145508,
      "learning_rate": 8.64434068587989e-07,
      "loss": 0.1022,
      "step": 970
    },
    {
      "epoch": 0.15317269393067004,
      "grad_norm": 5.124694347381592,
      "learning_rate": 8.642730639188535e-07,
      "loss": 0.1289,
      "step": 971
    },
    {
      "epoch": 0.15333044129826084,
      "grad_norm": 8.405393600463867,
      "learning_rate": 8.641120592497182e-07,
      "loss": 0.0545,
      "step": 972
    },
    {
      "epoch": 0.15348818866585165,
      "grad_norm": 7.67399263381958,
      "learning_rate": 8.639510545805829e-07,
      "loss": 0.0632,
      "step": 973
    },
    {
      "epoch": 0.15364593603344245,
      "grad_norm": 6.533464431762695,
      "learning_rate": 8.637900499114475e-07,
      "loss": 0.0677,
      "step": 974
    },
    {
      "epoch": 0.15380368340103326,
      "grad_norm": 9.284419059753418,
      "learning_rate": 8.63629045242312e-07,
      "loss": 0.087,
      "step": 975
    },
    {
      "epoch": 0.15396143076862404,
      "grad_norm": 4.728756904602051,
      "learning_rate": 8.634680405731766e-07,
      "loss": 0.0651,
      "step": 976
    },
    {
      "epoch": 0.15411917813621484,
      "grad_norm": 8.1788969039917,
      "learning_rate": 8.633070359040412e-07,
      "loss": 0.0738,
      "step": 977
    },
    {
      "epoch": 0.15427692550380565,
      "grad_norm": 13.664631843566895,
      "learning_rate": 8.631460312349058e-07,
      "loss": 0.0805,
      "step": 978
    },
    {
      "epoch": 0.15443467287139645,
      "grad_norm": 10.1797456741333,
      "learning_rate": 8.629850265657703e-07,
      "loss": 0.0571,
      "step": 979
    },
    {
      "epoch": 0.15459242023898725,
      "grad_norm": 4.150187015533447,
      "learning_rate": 8.628240218966349e-07,
      "loss": 0.0303,
      "step": 980
    },
    {
      "epoch": 0.15475016760657806,
      "grad_norm": 3.7078182697296143,
      "learning_rate": 8.626630172274996e-07,
      "loss": 0.0406,
      "step": 981
    },
    {
      "epoch": 0.15490791497416886,
      "grad_norm": 7.318813323974609,
      "learning_rate": 8.625020125583642e-07,
      "loss": 0.0502,
      "step": 982
    },
    {
      "epoch": 0.15506566234175967,
      "grad_norm": 6.322595596313477,
      "learning_rate": 8.623410078892287e-07,
      "loss": 0.0582,
      "step": 983
    },
    {
      "epoch": 0.15522340970935047,
      "grad_norm": 5.230018615722656,
      "learning_rate": 8.621800032200933e-07,
      "loss": 0.0487,
      "step": 984
    },
    {
      "epoch": 0.15538115707694128,
      "grad_norm": 5.482806205749512,
      "learning_rate": 8.620189985509579e-07,
      "loss": 0.0419,
      "step": 985
    },
    {
      "epoch": 0.15553890444453208,
      "grad_norm": 6.209947109222412,
      "learning_rate": 8.618579938818225e-07,
      "loss": 0.0504,
      "step": 986
    },
    {
      "epoch": 0.1556966518121229,
      "grad_norm": 8.243134498596191,
      "learning_rate": 8.61696989212687e-07,
      "loss": 0.0585,
      "step": 987
    },
    {
      "epoch": 0.1558543991797137,
      "grad_norm": 4.481822490692139,
      "learning_rate": 8.615359845435516e-07,
      "loss": 0.0529,
      "step": 988
    },
    {
      "epoch": 0.1560121465473045,
      "grad_norm": 7.807392597198486,
      "learning_rate": 8.613749798744164e-07,
      "loss": 0.0348,
      "step": 989
    },
    {
      "epoch": 0.1561698939148953,
      "grad_norm": 9.053619384765625,
      "learning_rate": 8.61213975205281e-07,
      "loss": 0.0759,
      "step": 990
    },
    {
      "epoch": 0.1563276412824861,
      "grad_norm": 9.76534366607666,
      "learning_rate": 8.610529705361456e-07,
      "loss": 0.0552,
      "step": 991
    },
    {
      "epoch": 0.1564853886500769,
      "grad_norm": 9.809041976928711,
      "learning_rate": 8.608919658670101e-07,
      "loss": 0.0747,
      "step": 992
    },
    {
      "epoch": 0.15664313601766772,
      "grad_norm": 6.268856048583984,
      "learning_rate": 8.607309611978747e-07,
      "loss": 0.0861,
      "step": 993
    },
    {
      "epoch": 0.15680088338525852,
      "grad_norm": 142.48008728027344,
      "learning_rate": 8.605699565287393e-07,
      "loss": 0.1093,
      "step": 994
    },
    {
      "epoch": 0.1569586307528493,
      "grad_norm": 4.752850532531738,
      "learning_rate": 8.604089518596039e-07,
      "loss": 0.1125,
      "step": 995
    },
    {
      "epoch": 0.1571163781204401,
      "grad_norm": 6.792891025543213,
      "learning_rate": 8.602479471904685e-07,
      "loss": 0.0506,
      "step": 996
    },
    {
      "epoch": 0.1572741254880309,
      "grad_norm": 5.308335304260254,
      "learning_rate": 8.600869425213331e-07,
      "loss": 0.037,
      "step": 997
    },
    {
      "epoch": 0.1574318728556217,
      "grad_norm": 4.207362651824951,
      "learning_rate": 8.599259378521977e-07,
      "loss": 0.052,
      "step": 998
    },
    {
      "epoch": 0.15758962022321252,
      "grad_norm": 6.1989922523498535,
      "learning_rate": 8.597649331830623e-07,
      "loss": 0.0505,
      "step": 999
    },
    {
      "epoch": 0.15774736759080332,
      "grad_norm": 6.644219398498535,
      "learning_rate": 8.596039285139268e-07,
      "loss": 0.0708,
      "step": 1000
    },
    {
      "epoch": 0.15790511495839413,
      "grad_norm": 8.120538711547852,
      "learning_rate": 8.594429238447914e-07,
      "loss": 0.0471,
      "step": 1001
    },
    {
      "epoch": 0.15806286232598493,
      "grad_norm": 3.678455352783203,
      "learning_rate": 8.59281919175656e-07,
      "loss": 0.0294,
      "step": 1002
    },
    {
      "epoch": 0.15822060969357574,
      "grad_norm": 6.847627639770508,
      "learning_rate": 8.591209145065207e-07,
      "loss": 0.1008,
      "step": 1003
    },
    {
      "epoch": 0.15837835706116654,
      "grad_norm": 3.4793381690979004,
      "learning_rate": 8.589599098373852e-07,
      "loss": 0.0329,
      "step": 1004
    },
    {
      "epoch": 0.15853610442875735,
      "grad_norm": 7.70238733291626,
      "learning_rate": 8.587989051682498e-07,
      "loss": 0.0634,
      "step": 1005
    },
    {
      "epoch": 0.15869385179634815,
      "grad_norm": 4.5630106925964355,
      "learning_rate": 8.586379004991145e-07,
      "loss": 0.0774,
      "step": 1006
    },
    {
      "epoch": 0.15885159916393896,
      "grad_norm": 8.479558944702148,
      "learning_rate": 8.584768958299791e-07,
      "loss": 0.0769,
      "step": 1007
    },
    {
      "epoch": 0.15900934653152976,
      "grad_norm": 8.264494895935059,
      "learning_rate": 8.583158911608436e-07,
      "loss": 0.0663,
      "step": 1008
    },
    {
      "epoch": 0.15916709389912057,
      "grad_norm": 4.046088218688965,
      "learning_rate": 8.581548864917082e-07,
      "loss": 0.036,
      "step": 1009
    },
    {
      "epoch": 0.15932484126671137,
      "grad_norm": 7.706320762634277,
      "learning_rate": 8.579938818225728e-07,
      "loss": 0.0693,
      "step": 1010
    },
    {
      "epoch": 0.15948258863430217,
      "grad_norm": 3.29152250289917,
      "learning_rate": 8.578328771534375e-07,
      "loss": 0.0391,
      "step": 1011
    },
    {
      "epoch": 0.15964033600189298,
      "grad_norm": 3.251323699951172,
      "learning_rate": 8.576718724843021e-07,
      "loss": 0.0414,
      "step": 1012
    },
    {
      "epoch": 0.15979808336948378,
      "grad_norm": 5.912606716156006,
      "learning_rate": 8.575108678151666e-07,
      "loss": 0.067,
      "step": 1013
    },
    {
      "epoch": 0.15995583073707456,
      "grad_norm": 5.435115814208984,
      "learning_rate": 8.573498631460312e-07,
      "loss": 0.0865,
      "step": 1014
    },
    {
      "epoch": 0.16011357810466537,
      "grad_norm": 4.942360877990723,
      "learning_rate": 8.571888584768958e-07,
      "loss": 0.1016,
      "step": 1015
    },
    {
      "epoch": 0.16027132547225617,
      "grad_norm": 4.604283332824707,
      "learning_rate": 8.570278538077604e-07,
      "loss": 0.0371,
      "step": 1016
    },
    {
      "epoch": 0.16042907283984698,
      "grad_norm": 6.982027530670166,
      "learning_rate": 8.568668491386249e-07,
      "loss": 0.0997,
      "step": 1017
    },
    {
      "epoch": 0.16058682020743778,
      "grad_norm": 6.0777668952941895,
      "learning_rate": 8.567058444694896e-07,
      "loss": 0.0893,
      "step": 1018
    },
    {
      "epoch": 0.16074456757502859,
      "grad_norm": 3.114046573638916,
      "learning_rate": 8.565448398003542e-07,
      "loss": 0.0447,
      "step": 1019
    },
    {
      "epoch": 0.1609023149426194,
      "grad_norm": 5.120523452758789,
      "learning_rate": 8.563838351312188e-07,
      "loss": 0.0506,
      "step": 1020
    },
    {
      "epoch": 0.1610600623102102,
      "grad_norm": 6.959108352661133,
      "learning_rate": 8.562228304620833e-07,
      "loss": 0.0949,
      "step": 1021
    },
    {
      "epoch": 0.161217809677801,
      "grad_norm": 6.767091751098633,
      "learning_rate": 8.560618257929479e-07,
      "loss": 0.1029,
      "step": 1022
    },
    {
      "epoch": 0.1613755570453918,
      "grad_norm": 6.899235725402832,
      "learning_rate": 8.559008211238125e-07,
      "loss": 0.0677,
      "step": 1023
    },
    {
      "epoch": 0.1615333044129826,
      "grad_norm": 7.200080871582031,
      "learning_rate": 8.557398164546772e-07,
      "loss": 0.0737,
      "step": 1024
    },
    {
      "epoch": 0.1616910517805734,
      "grad_norm": 4.691421985626221,
      "learning_rate": 8.555788117855417e-07,
      "loss": 0.0601,
      "step": 1025
    },
    {
      "epoch": 0.16184879914816422,
      "grad_norm": 6.024470806121826,
      "learning_rate": 8.554178071164064e-07,
      "loss": 0.0558,
      "step": 1026
    },
    {
      "epoch": 0.16200654651575502,
      "grad_norm": 4.782193183898926,
      "learning_rate": 8.55256802447271e-07,
      "loss": 0.0446,
      "step": 1027
    },
    {
      "epoch": 0.16216429388334583,
      "grad_norm": 4.7500176429748535,
      "learning_rate": 8.550957977781356e-07,
      "loss": 0.0442,
      "step": 1028
    },
    {
      "epoch": 0.16232204125093663,
      "grad_norm": 6.006361961364746,
      "learning_rate": 8.549347931090001e-07,
      "loss": 0.0875,
      "step": 1029
    },
    {
      "epoch": 0.16247978861852744,
      "grad_norm": 5.781008720397949,
      "learning_rate": 8.547737884398647e-07,
      "loss": 0.0889,
      "step": 1030
    },
    {
      "epoch": 0.16263753598611824,
      "grad_norm": 6.284361362457275,
      "learning_rate": 8.546127837707293e-07,
      "loss": 0.063,
      "step": 1031
    },
    {
      "epoch": 0.16279528335370905,
      "grad_norm": 15.59167766571045,
      "learning_rate": 8.544517791015939e-07,
      "loss": 0.0708,
      "step": 1032
    },
    {
      "epoch": 0.16295303072129985,
      "grad_norm": 5.2419304847717285,
      "learning_rate": 8.542907744324584e-07,
      "loss": 0.0629,
      "step": 1033
    },
    {
      "epoch": 0.16311077808889063,
      "grad_norm": 5.976012229919434,
      "learning_rate": 8.541297697633231e-07,
      "loss": 0.0489,
      "step": 1034
    },
    {
      "epoch": 0.16326852545648143,
      "grad_norm": 6.138513088226318,
      "learning_rate": 8.539687650941877e-07,
      "loss": 0.0612,
      "step": 1035
    },
    {
      "epoch": 0.16342627282407224,
      "grad_norm": 14.731651306152344,
      "learning_rate": 8.538077604250523e-07,
      "loss": 0.06,
      "step": 1036
    },
    {
      "epoch": 0.16358402019166304,
      "grad_norm": 7.656831741333008,
      "learning_rate": 8.536467557559169e-07,
      "loss": 0.1064,
      "step": 1037
    },
    {
      "epoch": 0.16374176755925385,
      "grad_norm": 5.557590484619141,
      "learning_rate": 8.534857510867814e-07,
      "loss": 0.1172,
      "step": 1038
    },
    {
      "epoch": 0.16389951492684465,
      "grad_norm": 5.93520450592041,
      "learning_rate": 8.53324746417646e-07,
      "loss": 0.1034,
      "step": 1039
    },
    {
      "epoch": 0.16405726229443546,
      "grad_norm": 9.557320594787598,
      "learning_rate": 8.531637417485106e-07,
      "loss": 0.0541,
      "step": 1040
    },
    {
      "epoch": 0.16421500966202626,
      "grad_norm": 5.482620716094971,
      "learning_rate": 8.530027370793754e-07,
      "loss": 0.0638,
      "step": 1041
    },
    {
      "epoch": 0.16437275702961707,
      "grad_norm": 7.604795455932617,
      "learning_rate": 8.528417324102399e-07,
      "loss": 0.0622,
      "step": 1042
    },
    {
      "epoch": 0.16453050439720787,
      "grad_norm": 5.067941188812256,
      "learning_rate": 8.526807277411045e-07,
      "loss": 0.0624,
      "step": 1043
    },
    {
      "epoch": 0.16468825176479868,
      "grad_norm": 7.364793300628662,
      "learning_rate": 8.525197230719691e-07,
      "loss": 0.0741,
      "step": 1044
    },
    {
      "epoch": 0.16484599913238948,
      "grad_norm": 5.77717399597168,
      "learning_rate": 8.523587184028337e-07,
      "loss": 0.0471,
      "step": 1045
    },
    {
      "epoch": 0.16500374649998029,
      "grad_norm": 7.496662139892578,
      "learning_rate": 8.521977137336982e-07,
      "loss": 0.0667,
      "step": 1046
    },
    {
      "epoch": 0.1651614938675711,
      "grad_norm": 8.605033874511719,
      "learning_rate": 8.520367090645628e-07,
      "loss": 0.0818,
      "step": 1047
    },
    {
      "epoch": 0.1653192412351619,
      "grad_norm": 6.1481781005859375,
      "learning_rate": 8.518757043954275e-07,
      "loss": 0.0884,
      "step": 1048
    },
    {
      "epoch": 0.1654769886027527,
      "grad_norm": 4.9542059898376465,
      "learning_rate": 8.517146997262921e-07,
      "loss": 0.0778,
      "step": 1049
    },
    {
      "epoch": 0.1656347359703435,
      "grad_norm": 7.27318000793457,
      "learning_rate": 8.515536950571566e-07,
      "loss": 0.0653,
      "step": 1050
    },
    {
      "epoch": 0.1657924833379343,
      "grad_norm": 5.166891098022461,
      "learning_rate": 8.513926903880212e-07,
      "loss": 0.0554,
      "step": 1051
    },
    {
      "epoch": 0.16595023070552511,
      "grad_norm": 9.332530975341797,
      "learning_rate": 8.512316857188858e-07,
      "loss": 0.0753,
      "step": 1052
    },
    {
      "epoch": 0.1661079780731159,
      "grad_norm": 4.665710926055908,
      "learning_rate": 8.510706810497504e-07,
      "loss": 0.0662,
      "step": 1053
    },
    {
      "epoch": 0.1662657254407067,
      "grad_norm": 5.497588634490967,
      "learning_rate": 8.509096763806149e-07,
      "loss": 0.0852,
      "step": 1054
    },
    {
      "epoch": 0.1664234728082975,
      "grad_norm": 5.279402732849121,
      "learning_rate": 8.507486717114795e-07,
      "loss": 0.0511,
      "step": 1055
    },
    {
      "epoch": 0.1665812201758883,
      "grad_norm": 4.8032097816467285,
      "learning_rate": 8.505876670423442e-07,
      "loss": 0.0743,
      "step": 1056
    },
    {
      "epoch": 0.1667389675434791,
      "grad_norm": 7.598001480102539,
      "learning_rate": 8.504266623732088e-07,
      "loss": 0.0997,
      "step": 1057
    },
    {
      "epoch": 0.16689671491106992,
      "grad_norm": 6.238626956939697,
      "learning_rate": 8.502656577040733e-07,
      "loss": 0.0529,
      "step": 1058
    },
    {
      "epoch": 0.16705446227866072,
      "grad_norm": 9.464849472045898,
      "learning_rate": 8.50104653034938e-07,
      "loss": 0.0858,
      "step": 1059
    },
    {
      "epoch": 0.16721220964625153,
      "grad_norm": 4.487782955169678,
      "learning_rate": 8.499436483658026e-07,
      "loss": 0.0772,
      "step": 1060
    },
    {
      "epoch": 0.16736995701384233,
      "grad_norm": 6.810683250427246,
      "learning_rate": 8.497826436966672e-07,
      "loss": 0.1174,
      "step": 1061
    },
    {
      "epoch": 0.16752770438143313,
      "grad_norm": 4.956742286682129,
      "learning_rate": 8.496216390275318e-07,
      "loss": 0.09,
      "step": 1062
    },
    {
      "epoch": 0.16768545174902394,
      "grad_norm": 6.974895477294922,
      "learning_rate": 8.494606343583963e-07,
      "loss": 0.0801,
      "step": 1063
    },
    {
      "epoch": 0.16784319911661474,
      "grad_norm": 4.244524002075195,
      "learning_rate": 8.49299629689261e-07,
      "loss": 0.0587,
      "step": 1064
    },
    {
      "epoch": 0.16800094648420555,
      "grad_norm": 3.368680715560913,
      "learning_rate": 8.491386250201256e-07,
      "loss": 0.0819,
      "step": 1065
    },
    {
      "epoch": 0.16815869385179635,
      "grad_norm": 6.945837020874023,
      "learning_rate": 8.489776203509902e-07,
      "loss": 0.115,
      "step": 1066
    },
    {
      "epoch": 0.16831644121938716,
      "grad_norm": 10.10990047454834,
      "learning_rate": 8.488166156818547e-07,
      "loss": 0.104,
      "step": 1067
    },
    {
      "epoch": 0.16847418858697796,
      "grad_norm": 4.6603007316589355,
      "learning_rate": 8.486556110127193e-07,
      "loss": 0.0301,
      "step": 1068
    },
    {
      "epoch": 0.16863193595456877,
      "grad_norm": 4.7445969581604,
      "learning_rate": 8.484946063435839e-07,
      "loss": 0.0885,
      "step": 1069
    },
    {
      "epoch": 0.16878968332215957,
      "grad_norm": 7.254500389099121,
      "learning_rate": 8.483336016744485e-07,
      "loss": 0.0929,
      "step": 1070
    },
    {
      "epoch": 0.16894743068975038,
      "grad_norm": 3.6324713230133057,
      "learning_rate": 8.481725970053131e-07,
      "loss": 0.0365,
      "step": 1071
    },
    {
      "epoch": 0.16910517805734115,
      "grad_norm": 8.888936042785645,
      "learning_rate": 8.480115923361777e-07,
      "loss": 0.0819,
      "step": 1072
    },
    {
      "epoch": 0.16926292542493196,
      "grad_norm": 11.244965553283691,
      "learning_rate": 8.478505876670423e-07,
      "loss": 0.0637,
      "step": 1073
    },
    {
      "epoch": 0.16942067279252276,
      "grad_norm": 7.391339302062988,
      "learning_rate": 8.476895829979069e-07,
      "loss": 0.08,
      "step": 1074
    },
    {
      "epoch": 0.16957842016011357,
      "grad_norm": 12.288595199584961,
      "learning_rate": 8.475285783287714e-07,
      "loss": 0.0703,
      "step": 1075
    },
    {
      "epoch": 0.16973616752770437,
      "grad_norm": 5.894357681274414,
      "learning_rate": 8.47367573659636e-07,
      "loss": 0.0843,
      "step": 1076
    },
    {
      "epoch": 0.16989391489529518,
      "grad_norm": 8.744696617126465,
      "learning_rate": 8.472065689905007e-07,
      "loss": 0.0985,
      "step": 1077
    },
    {
      "epoch": 0.17005166226288598,
      "grad_norm": 6.621856212615967,
      "learning_rate": 8.470455643213654e-07,
      "loss": 0.0672,
      "step": 1078
    },
    {
      "epoch": 0.1702094096304768,
      "grad_norm": 6.276869297027588,
      "learning_rate": 8.468845596522299e-07,
      "loss": 0.0482,
      "step": 1079
    },
    {
      "epoch": 0.1703671569980676,
      "grad_norm": 5.934649467468262,
      "learning_rate": 8.467235549830945e-07,
      "loss": 0.0662,
      "step": 1080
    },
    {
      "epoch": 0.1705249043656584,
      "grad_norm": 4.509552478790283,
      "learning_rate": 8.465625503139591e-07,
      "loss": 0.0608,
      "step": 1081
    },
    {
      "epoch": 0.1706826517332492,
      "grad_norm": 5.281670570373535,
      "learning_rate": 8.464015456448237e-07,
      "loss": 0.0638,
      "step": 1082
    },
    {
      "epoch": 0.17084039910084,
      "grad_norm": 4.402935981750488,
      "learning_rate": 8.462405409756883e-07,
      "loss": 0.0386,
      "step": 1083
    },
    {
      "epoch": 0.1709981464684308,
      "grad_norm": 4.183838367462158,
      "learning_rate": 8.460795363065528e-07,
      "loss": 0.0955,
      "step": 1084
    },
    {
      "epoch": 0.17115589383602162,
      "grad_norm": 5.913980960845947,
      "learning_rate": 8.459185316374174e-07,
      "loss": 0.0793,
      "step": 1085
    },
    {
      "epoch": 0.17131364120361242,
      "grad_norm": 8.913019180297852,
      "learning_rate": 8.457575269682821e-07,
      "loss": 0.0787,
      "step": 1086
    },
    {
      "epoch": 0.17147138857120323,
      "grad_norm": 5.881781578063965,
      "learning_rate": 8.455965222991467e-07,
      "loss": 0.0781,
      "step": 1087
    },
    {
      "epoch": 0.17162913593879403,
      "grad_norm": 7.500895023345947,
      "learning_rate": 8.454355176300112e-07,
      "loss": 0.0884,
      "step": 1088
    },
    {
      "epoch": 0.17178688330638484,
      "grad_norm": 8.74000358581543,
      "learning_rate": 8.452745129608758e-07,
      "loss": 0.0661,
      "step": 1089
    },
    {
      "epoch": 0.17194463067397564,
      "grad_norm": 7.282954692840576,
      "learning_rate": 8.451135082917404e-07,
      "loss": 0.1357,
      "step": 1090
    },
    {
      "epoch": 0.17210237804156645,
      "grad_norm": 3.7355973720550537,
      "learning_rate": 8.44952503622605e-07,
      "loss": 0.0493,
      "step": 1091
    },
    {
      "epoch": 0.17226012540915722,
      "grad_norm": 3.9269330501556396,
      "learning_rate": 8.447914989534695e-07,
      "loss": 0.0325,
      "step": 1092
    },
    {
      "epoch": 0.17241787277674803,
      "grad_norm": 6.021385192871094,
      "learning_rate": 8.446304942843341e-07,
      "loss": 0.1451,
      "step": 1093
    },
    {
      "epoch": 0.17257562014433883,
      "grad_norm": 3.8144242763519287,
      "learning_rate": 8.444694896151989e-07,
      "loss": 0.0452,
      "step": 1094
    },
    {
      "epoch": 0.17273336751192964,
      "grad_norm": 6.527584075927734,
      "learning_rate": 8.443084849460635e-07,
      "loss": 0.0695,
      "step": 1095
    },
    {
      "epoch": 0.17289111487952044,
      "grad_norm": 3.24444317817688,
      "learning_rate": 8.44147480276928e-07,
      "loss": 0.0555,
      "step": 1096
    },
    {
      "epoch": 0.17304886224711125,
      "grad_norm": 6.450403213500977,
      "learning_rate": 8.439864756077926e-07,
      "loss": 0.0499,
      "step": 1097
    },
    {
      "epoch": 0.17320660961470205,
      "grad_norm": 6.687957286834717,
      "learning_rate": 8.438254709386572e-07,
      "loss": 0.0643,
      "step": 1098
    },
    {
      "epoch": 0.17336435698229286,
      "grad_norm": 7.201638698577881,
      "learning_rate": 8.436644662695218e-07,
      "loss": 0.0889,
      "step": 1099
    },
    {
      "epoch": 0.17352210434988366,
      "grad_norm": 3.677889347076416,
      "learning_rate": 8.435034616003863e-07,
      "loss": 0.0551,
      "step": 1100
    },
    {
      "epoch": 0.17367985171747446,
      "grad_norm": 6.6136627197265625,
      "learning_rate": 8.43342456931251e-07,
      "loss": 0.0787,
      "step": 1101
    },
    {
      "epoch": 0.17383759908506527,
      "grad_norm": 2.215071678161621,
      "learning_rate": 8.431814522621156e-07,
      "loss": 0.0369,
      "step": 1102
    },
    {
      "epoch": 0.17399534645265607,
      "grad_norm": 5.720557689666748,
      "learning_rate": 8.430204475929802e-07,
      "loss": 0.0522,
      "step": 1103
    },
    {
      "epoch": 0.17415309382024688,
      "grad_norm": 3.361826181411743,
      "learning_rate": 8.428594429238447e-07,
      "loss": 0.032,
      "step": 1104
    },
    {
      "epoch": 0.17431084118783768,
      "grad_norm": 4.106998920440674,
      "learning_rate": 8.426984382547093e-07,
      "loss": 0.0742,
      "step": 1105
    },
    {
      "epoch": 0.1744685885554285,
      "grad_norm": 5.182785987854004,
      "learning_rate": 8.425374335855739e-07,
      "loss": 0.0779,
      "step": 1106
    },
    {
      "epoch": 0.1746263359230193,
      "grad_norm": 5.918130397796631,
      "learning_rate": 8.423764289164385e-07,
      "loss": 0.0382,
      "step": 1107
    },
    {
      "epoch": 0.1747840832906101,
      "grad_norm": 8.346705436706543,
      "learning_rate": 8.422154242473032e-07,
      "loss": 0.0692,
      "step": 1108
    },
    {
      "epoch": 0.1749418306582009,
      "grad_norm": 5.706119060516357,
      "learning_rate": 8.420544195781677e-07,
      "loss": 0.0713,
      "step": 1109
    },
    {
      "epoch": 0.1750995780257917,
      "grad_norm": 5.7744927406311035,
      "learning_rate": 8.418934149090323e-07,
      "loss": 0.0614,
      "step": 1110
    },
    {
      "epoch": 0.17525732539338248,
      "grad_norm": 5.5424394607543945,
      "learning_rate": 8.41732410239897e-07,
      "loss": 0.0786,
      "step": 1111
    },
    {
      "epoch": 0.1754150727609733,
      "grad_norm": 8.933950424194336,
      "learning_rate": 8.415714055707616e-07,
      "loss": 0.0916,
      "step": 1112
    },
    {
      "epoch": 0.1755728201285641,
      "grad_norm": 4.873121738433838,
      "learning_rate": 8.414104009016261e-07,
      "loss": 0.0791,
      "step": 1113
    },
    {
      "epoch": 0.1757305674961549,
      "grad_norm": 11.223404884338379,
      "learning_rate": 8.412493962324907e-07,
      "loss": 0.0894,
      "step": 1114
    },
    {
      "epoch": 0.1758883148637457,
      "grad_norm": 8.899439811706543,
      "learning_rate": 8.410883915633553e-07,
      "loss": 0.0627,
      "step": 1115
    },
    {
      "epoch": 0.1760460622313365,
      "grad_norm": 8.759340286254883,
      "learning_rate": 8.4092738689422e-07,
      "loss": 0.0678,
      "step": 1116
    },
    {
      "epoch": 0.1762038095989273,
      "grad_norm": 8.697700500488281,
      "learning_rate": 8.407663822250845e-07,
      "loss": 0.0842,
      "step": 1117
    },
    {
      "epoch": 0.17636155696651812,
      "grad_norm": 5.81970739364624,
      "learning_rate": 8.406053775559491e-07,
      "loss": 0.0806,
      "step": 1118
    },
    {
      "epoch": 0.17651930433410892,
      "grad_norm": 3.0905957221984863,
      "learning_rate": 8.404443728868137e-07,
      "loss": 0.0616,
      "step": 1119
    },
    {
      "epoch": 0.17667705170169973,
      "grad_norm": 5.192538261413574,
      "learning_rate": 8.402833682176783e-07,
      "loss": 0.068,
      "step": 1120
    },
    {
      "epoch": 0.17683479906929053,
      "grad_norm": 5.564635276794434,
      "learning_rate": 8.401223635485428e-07,
      "loss": 0.0597,
      "step": 1121
    },
    {
      "epoch": 0.17699254643688134,
      "grad_norm": 7.305998802185059,
      "learning_rate": 8.399613588794074e-07,
      "loss": 0.092,
      "step": 1122
    },
    {
      "epoch": 0.17715029380447214,
      "grad_norm": 8.529744148254395,
      "learning_rate": 8.39800354210272e-07,
      "loss": 0.0492,
      "step": 1123
    },
    {
      "epoch": 0.17730804117206295,
      "grad_norm": 6.663406848907471,
      "learning_rate": 8.396393495411367e-07,
      "loss": 0.058,
      "step": 1124
    },
    {
      "epoch": 0.17746578853965375,
      "grad_norm": 5.353763580322266,
      "learning_rate": 8.394783448720012e-07,
      "loss": 0.0416,
      "step": 1125
    },
    {
      "epoch": 0.17762353590724456,
      "grad_norm": 5.439206123352051,
      "learning_rate": 8.393173402028658e-07,
      "loss": 0.034,
      "step": 1126
    },
    {
      "epoch": 0.17778128327483536,
      "grad_norm": 5.433595180511475,
      "learning_rate": 8.391563355337304e-07,
      "loss": 0.0859,
      "step": 1127
    },
    {
      "epoch": 0.17793903064242617,
      "grad_norm": 5.620081901550293,
      "learning_rate": 8.38995330864595e-07,
      "loss": 0.0746,
      "step": 1128
    },
    {
      "epoch": 0.17809677801001697,
      "grad_norm": 5.438671112060547,
      "learning_rate": 8.388343261954597e-07,
      "loss": 0.051,
      "step": 1129
    },
    {
      "epoch": 0.17825452537760775,
      "grad_norm": 5.7582221031188965,
      "learning_rate": 8.386733215263242e-07,
      "loss": 0.0663,
      "step": 1130
    },
    {
      "epoch": 0.17841227274519855,
      "grad_norm": 3.84271240234375,
      "learning_rate": 8.385123168571889e-07,
      "loss": 0.0603,
      "step": 1131
    },
    {
      "epoch": 0.17857002011278936,
      "grad_norm": 10.873653411865234,
      "learning_rate": 8.383513121880535e-07,
      "loss": 0.0839,
      "step": 1132
    },
    {
      "epoch": 0.17872776748038016,
      "grad_norm": 6.634921073913574,
      "learning_rate": 8.381903075189181e-07,
      "loss": 0.0687,
      "step": 1133
    },
    {
      "epoch": 0.17888551484797097,
      "grad_norm": 9.545279502868652,
      "learning_rate": 8.380293028497826e-07,
      "loss": 0.0471,
      "step": 1134
    },
    {
      "epoch": 0.17904326221556177,
      "grad_norm": 5.004254341125488,
      "learning_rate": 8.378682981806472e-07,
      "loss": 0.0417,
      "step": 1135
    },
    {
      "epoch": 0.17920100958315258,
      "grad_norm": 6.581553936004639,
      "learning_rate": 8.377072935115118e-07,
      "loss": 0.0982,
      "step": 1136
    },
    {
      "epoch": 0.17935875695074338,
      "grad_norm": 6.660752296447754,
      "learning_rate": 8.375462888423764e-07,
      "loss": 0.0807,
      "step": 1137
    },
    {
      "epoch": 0.17951650431833419,
      "grad_norm": 3.5822994709014893,
      "learning_rate": 8.373852841732409e-07,
      "loss": 0.0605,
      "step": 1138
    },
    {
      "epoch": 0.179674251685925,
      "grad_norm": 7.432138442993164,
      "learning_rate": 8.372242795041056e-07,
      "loss": 0.0533,
      "step": 1139
    },
    {
      "epoch": 0.1798319990535158,
      "grad_norm": 5.05093240737915,
      "learning_rate": 8.370632748349702e-07,
      "loss": 0.0563,
      "step": 1140
    },
    {
      "epoch": 0.1799897464211066,
      "grad_norm": 3.2812182903289795,
      "learning_rate": 8.369022701658348e-07,
      "loss": 0.0371,
      "step": 1141
    },
    {
      "epoch": 0.1801474937886974,
      "grad_norm": 5.70815372467041,
      "learning_rate": 8.367412654966993e-07,
      "loss": 0.0748,
      "step": 1142
    },
    {
      "epoch": 0.1803052411562882,
      "grad_norm": 5.593191146850586,
      "learning_rate": 8.365802608275639e-07,
      "loss": 0.0634,
      "step": 1143
    },
    {
      "epoch": 0.18046298852387901,
      "grad_norm": 5.0286478996276855,
      "learning_rate": 8.364192561584285e-07,
      "loss": 0.0528,
      "step": 1144
    },
    {
      "epoch": 0.18062073589146982,
      "grad_norm": 4.4398698806762695,
      "learning_rate": 8.362582514892931e-07,
      "loss": 0.0998,
      "step": 1145
    },
    {
      "epoch": 0.18077848325906062,
      "grad_norm": 4.408929824829102,
      "learning_rate": 8.360972468201577e-07,
      "loss": 0.0658,
      "step": 1146
    },
    {
      "epoch": 0.18093623062665143,
      "grad_norm": 3.4798245429992676,
      "learning_rate": 8.359362421510224e-07,
      "loss": 0.057,
      "step": 1147
    },
    {
      "epoch": 0.18109397799424223,
      "grad_norm": 2.3720192909240723,
      "learning_rate": 8.35775237481887e-07,
      "loss": 0.023,
      "step": 1148
    },
    {
      "epoch": 0.18125172536183304,
      "grad_norm": 1.9957060813903809,
      "learning_rate": 8.356142328127516e-07,
      "loss": 0.022,
      "step": 1149
    },
    {
      "epoch": 0.18140947272942382,
      "grad_norm": 6.715579509735107,
      "learning_rate": 8.354532281436161e-07,
      "loss": 0.139,
      "step": 1150
    },
    {
      "epoch": 0.18156722009701462,
      "grad_norm": 6.875566005706787,
      "learning_rate": 8.352922234744807e-07,
      "loss": 0.0692,
      "step": 1151
    },
    {
      "epoch": 0.18172496746460542,
      "grad_norm": 3.6216447353363037,
      "learning_rate": 8.351312188053453e-07,
      "loss": 0.0336,
      "step": 1152
    },
    {
      "epoch": 0.18188271483219623,
      "grad_norm": 4.116306781768799,
      "learning_rate": 8.349702141362099e-07,
      "loss": 0.0441,
      "step": 1153
    },
    {
      "epoch": 0.18204046219978703,
      "grad_norm": 6.943159580230713,
      "learning_rate": 8.348092094670746e-07,
      "loss": 0.0913,
      "step": 1154
    },
    {
      "epoch": 0.18219820956737784,
      "grad_norm": 3.5021755695343018,
      "learning_rate": 8.346482047979391e-07,
      "loss": 0.0607,
      "step": 1155
    },
    {
      "epoch": 0.18235595693496864,
      "grad_norm": 5.7585649490356445,
      "learning_rate": 8.344872001288037e-07,
      "loss": 0.0786,
      "step": 1156
    },
    {
      "epoch": 0.18251370430255945,
      "grad_norm": 5.901808738708496,
      "learning_rate": 8.343261954596683e-07,
      "loss": 0.0541,
      "step": 1157
    },
    {
      "epoch": 0.18267145167015025,
      "grad_norm": 8.584582328796387,
      "learning_rate": 8.341651907905329e-07,
      "loss": 0.0558,
      "step": 1158
    },
    {
      "epoch": 0.18282919903774106,
      "grad_norm": 6.354942798614502,
      "learning_rate": 8.340041861213974e-07,
      "loss": 0.0514,
      "step": 1159
    },
    {
      "epoch": 0.18298694640533186,
      "grad_norm": 5.005200386047363,
      "learning_rate": 8.33843181452262e-07,
      "loss": 0.081,
      "step": 1160
    },
    {
      "epoch": 0.18314469377292267,
      "grad_norm": 4.33692741394043,
      "learning_rate": 8.336821767831267e-07,
      "loss": 0.0676,
      "step": 1161
    },
    {
      "epoch": 0.18330244114051347,
      "grad_norm": 5.106905460357666,
      "learning_rate": 8.335211721139913e-07,
      "loss": 0.0686,
      "step": 1162
    },
    {
      "epoch": 0.18346018850810428,
      "grad_norm": 6.16445255279541,
      "learning_rate": 8.333601674448558e-07,
      "loss": 0.0909,
      "step": 1163
    },
    {
      "epoch": 0.18361793587569508,
      "grad_norm": 3.997206211090088,
      "learning_rate": 8.331991627757204e-07,
      "loss": 0.0472,
      "step": 1164
    },
    {
      "epoch": 0.1837756832432859,
      "grad_norm": 4.126903533935547,
      "learning_rate": 8.330381581065851e-07,
      "loss": 0.0588,
      "step": 1165
    },
    {
      "epoch": 0.1839334306108767,
      "grad_norm": 10.28675365447998,
      "learning_rate": 8.328771534374497e-07,
      "loss": 0.0675,
      "step": 1166
    },
    {
      "epoch": 0.1840911779784675,
      "grad_norm": 7.186713695526123,
      "learning_rate": 8.327161487683142e-07,
      "loss": 0.0862,
      "step": 1167
    },
    {
      "epoch": 0.1842489253460583,
      "grad_norm": 5.181392192840576,
      "learning_rate": 8.325551440991788e-07,
      "loss": 0.0534,
      "step": 1168
    },
    {
      "epoch": 0.18440667271364908,
      "grad_norm": 4.184267044067383,
      "learning_rate": 8.323941394300435e-07,
      "loss": 0.0542,
      "step": 1169
    },
    {
      "epoch": 0.18456442008123988,
      "grad_norm": 8.53089714050293,
      "learning_rate": 8.322331347609081e-07,
      "loss": 0.067,
      "step": 1170
    },
    {
      "epoch": 0.1847221674488307,
      "grad_norm": 6.048624515533447,
      "learning_rate": 8.320721300917726e-07,
      "loss": 0.1088,
      "step": 1171
    },
    {
      "epoch": 0.1848799148164215,
      "grad_norm": 6.11414098739624,
      "learning_rate": 8.319111254226372e-07,
      "loss": 0.0945,
      "step": 1172
    },
    {
      "epoch": 0.1850376621840123,
      "grad_norm": 8.064154624938965,
      "learning_rate": 8.317501207535018e-07,
      "loss": 0.0806,
      "step": 1173
    },
    {
      "epoch": 0.1851954095516031,
      "grad_norm": 4.403115272521973,
      "learning_rate": 8.315891160843664e-07,
      "loss": 0.0323,
      "step": 1174
    },
    {
      "epoch": 0.1853531569191939,
      "grad_norm": 3.644397497177124,
      "learning_rate": 8.31428111415231e-07,
      "loss": 0.0336,
      "step": 1175
    },
    {
      "epoch": 0.1855109042867847,
      "grad_norm": 2.348759412765503,
      "learning_rate": 8.312671067460956e-07,
      "loss": 0.0217,
      "step": 1176
    },
    {
      "epoch": 0.18566865165437552,
      "grad_norm": 8.147252082824707,
      "learning_rate": 8.311061020769602e-07,
      "loss": 0.0559,
      "step": 1177
    },
    {
      "epoch": 0.18582639902196632,
      "grad_norm": 3.9271461963653564,
      "learning_rate": 8.309450974078248e-07,
      "loss": 0.0753,
      "step": 1178
    },
    {
      "epoch": 0.18598414638955713,
      "grad_norm": 3.8691606521606445,
      "learning_rate": 8.307840927386894e-07,
      "loss": 0.063,
      "step": 1179
    },
    {
      "epoch": 0.18614189375714793,
      "grad_norm": 7.3613152503967285,
      "learning_rate": 8.306230880695539e-07,
      "loss": 0.0477,
      "step": 1180
    },
    {
      "epoch": 0.18629964112473874,
      "grad_norm": 6.239074230194092,
      "learning_rate": 8.304620834004185e-07,
      "loss": 0.0313,
      "step": 1181
    },
    {
      "epoch": 0.18645738849232954,
      "grad_norm": 6.47470235824585,
      "learning_rate": 8.303010787312832e-07,
      "loss": 0.0405,
      "step": 1182
    },
    {
      "epoch": 0.18661513585992034,
      "grad_norm": 14.254313468933105,
      "learning_rate": 8.301400740621478e-07,
      "loss": 0.1435,
      "step": 1183
    },
    {
      "epoch": 0.18677288322751115,
      "grad_norm": 4.88731050491333,
      "learning_rate": 8.299790693930124e-07,
      "loss": 0.0442,
      "step": 1184
    },
    {
      "epoch": 0.18693063059510195,
      "grad_norm": 4.566811561584473,
      "learning_rate": 8.29818064723877e-07,
      "loss": 0.0496,
      "step": 1185
    },
    {
      "epoch": 0.18708837796269276,
      "grad_norm": 5.288136005401611,
      "learning_rate": 8.296570600547416e-07,
      "loss": 0.0606,
      "step": 1186
    },
    {
      "epoch": 0.18724612533028356,
      "grad_norm": 14.072057723999023,
      "learning_rate": 8.294960553856062e-07,
      "loss": 0.0592,
      "step": 1187
    },
    {
      "epoch": 0.18740387269787434,
      "grad_norm": 3.6332509517669678,
      "learning_rate": 8.293350507164707e-07,
      "loss": 0.091,
      "step": 1188
    },
    {
      "epoch": 0.18756162006546515,
      "grad_norm": 7.204614162445068,
      "learning_rate": 8.291740460473353e-07,
      "loss": 0.0825,
      "step": 1189
    },
    {
      "epoch": 0.18771936743305595,
      "grad_norm": 5.5783867835998535,
      "learning_rate": 8.290130413781999e-07,
      "loss": 0.0624,
      "step": 1190
    },
    {
      "epoch": 0.18787711480064676,
      "grad_norm": 8.766402244567871,
      "learning_rate": 8.288520367090646e-07,
      "loss": 0.0828,
      "step": 1191
    },
    {
      "epoch": 0.18803486216823756,
      "grad_norm": 4.802553176879883,
      "learning_rate": 8.286910320399291e-07,
      "loss": 0.074,
      "step": 1192
    },
    {
      "epoch": 0.18819260953582836,
      "grad_norm": 11.380311965942383,
      "learning_rate": 8.285300273707937e-07,
      "loss": 0.0936,
      "step": 1193
    },
    {
      "epoch": 0.18835035690341917,
      "grad_norm": 9.392463684082031,
      "learning_rate": 8.283690227016583e-07,
      "loss": 0.094,
      "step": 1194
    },
    {
      "epoch": 0.18850810427100997,
      "grad_norm": 7.408383369445801,
      "learning_rate": 8.282080180325229e-07,
      "loss": 0.0684,
      "step": 1195
    },
    {
      "epoch": 0.18866585163860078,
      "grad_norm": 6.016057014465332,
      "learning_rate": 8.280470133633874e-07,
      "loss": 0.0684,
      "step": 1196
    },
    {
      "epoch": 0.18882359900619158,
      "grad_norm": 4.20705509185791,
      "learning_rate": 8.27886008694252e-07,
      "loss": 0.0592,
      "step": 1197
    },
    {
      "epoch": 0.1889813463737824,
      "grad_norm": 6.25236701965332,
      "learning_rate": 8.277250040251166e-07,
      "loss": 0.109,
      "step": 1198
    },
    {
      "epoch": 0.1891390937413732,
      "grad_norm": 8.855053901672363,
      "learning_rate": 8.275639993559814e-07,
      "loss": 0.0567,
      "step": 1199
    },
    {
      "epoch": 0.189296841108964,
      "grad_norm": 7.839231014251709,
      "learning_rate": 8.27402994686846e-07,
      "loss": 0.08,
      "step": 1200
    },
    {
      "epoch": 0.1894545884765548,
      "grad_norm": 5.923698902130127,
      "learning_rate": 8.272419900177105e-07,
      "loss": 0.0526,
      "step": 1201
    },
    {
      "epoch": 0.1896123358441456,
      "grad_norm": 3.367675542831421,
      "learning_rate": 8.270809853485751e-07,
      "loss": 0.032,
      "step": 1202
    },
    {
      "epoch": 0.1897700832117364,
      "grad_norm": 7.34796667098999,
      "learning_rate": 8.269199806794397e-07,
      "loss": 0.0781,
      "step": 1203
    },
    {
      "epoch": 0.18992783057932722,
      "grad_norm": 4.488637447357178,
      "learning_rate": 8.267589760103043e-07,
      "loss": 0.0655,
      "step": 1204
    },
    {
      "epoch": 0.19008557794691802,
      "grad_norm": 7.351264476776123,
      "learning_rate": 8.265979713411688e-07,
      "loss": 0.0438,
      "step": 1205
    },
    {
      "epoch": 0.19024332531450883,
      "grad_norm": 6.292731761932373,
      "learning_rate": 8.264369666720335e-07,
      "loss": 0.053,
      "step": 1206
    },
    {
      "epoch": 0.19040107268209963,
      "grad_norm": 9.767741203308105,
      "learning_rate": 8.262759620028981e-07,
      "loss": 0.066,
      "step": 1207
    },
    {
      "epoch": 0.1905588200496904,
      "grad_norm": 6.028949737548828,
      "learning_rate": 8.261149573337627e-07,
      "loss": 0.053,
      "step": 1208
    },
    {
      "epoch": 0.1907165674172812,
      "grad_norm": 9.830450057983398,
      "learning_rate": 8.259539526646272e-07,
      "loss": 0.0771,
      "step": 1209
    },
    {
      "epoch": 0.19087431478487202,
      "grad_norm": 6.367222309112549,
      "learning_rate": 8.257929479954918e-07,
      "loss": 0.0933,
      "step": 1210
    },
    {
      "epoch": 0.19103206215246282,
      "grad_norm": 3.649705410003662,
      "learning_rate": 8.256319433263564e-07,
      "loss": 0.0728,
      "step": 1211
    },
    {
      "epoch": 0.19118980952005363,
      "grad_norm": 6.903989315032959,
      "learning_rate": 8.25470938657221e-07,
      "loss": 0.1195,
      "step": 1212
    },
    {
      "epoch": 0.19134755688764443,
      "grad_norm": 9.6062593460083,
      "learning_rate": 8.253099339880855e-07,
      "loss": 0.1184,
      "step": 1213
    },
    {
      "epoch": 0.19150530425523524,
      "grad_norm": 4.583438873291016,
      "learning_rate": 8.251489293189502e-07,
      "loss": 0.0317,
      "step": 1214
    },
    {
      "epoch": 0.19166305162282604,
      "grad_norm": 3.6098432540893555,
      "learning_rate": 8.249879246498148e-07,
      "loss": 0.0639,
      "step": 1215
    },
    {
      "epoch": 0.19182079899041685,
      "grad_norm": 5.552302360534668,
      "learning_rate": 8.248269199806794e-07,
      "loss": 0.0482,
      "step": 1216
    },
    {
      "epoch": 0.19197854635800765,
      "grad_norm": 4.402776718139648,
      "learning_rate": 8.24665915311544e-07,
      "loss": 0.0933,
      "step": 1217
    },
    {
      "epoch": 0.19213629372559846,
      "grad_norm": 6.575020790100098,
      "learning_rate": 8.245049106424086e-07,
      "loss": 0.0679,
      "step": 1218
    },
    {
      "epoch": 0.19229404109318926,
      "grad_norm": 6.469968795776367,
      "learning_rate": 8.243439059732732e-07,
      "loss": 0.0729,
      "step": 1219
    },
    {
      "epoch": 0.19245178846078007,
      "grad_norm": 5.68861722946167,
      "learning_rate": 8.241829013041378e-07,
      "loss": 0.1167,
      "step": 1220
    },
    {
      "epoch": 0.19260953582837087,
      "grad_norm": 8.055570602416992,
      "learning_rate": 8.240218966350024e-07,
      "loss": 0.0498,
      "step": 1221
    },
    {
      "epoch": 0.19276728319596168,
      "grad_norm": 4.277229309082031,
      "learning_rate": 8.23860891965867e-07,
      "loss": 0.03,
      "step": 1222
    },
    {
      "epoch": 0.19292503056355248,
      "grad_norm": 5.417783737182617,
      "learning_rate": 8.236998872967316e-07,
      "loss": 0.0699,
      "step": 1223
    },
    {
      "epoch": 0.19308277793114328,
      "grad_norm": 5.803797245025635,
      "learning_rate": 8.235388826275962e-07,
      "loss": 0.0586,
      "step": 1224
    },
    {
      "epoch": 0.1932405252987341,
      "grad_norm": 3.9364662170410156,
      "learning_rate": 8.233778779584608e-07,
      "loss": 0.0489,
      "step": 1225
    },
    {
      "epoch": 0.1933982726663249,
      "grad_norm": 7.794634819030762,
      "learning_rate": 8.232168732893253e-07,
      "loss": 0.1499,
      "step": 1226
    },
    {
      "epoch": 0.19355602003391567,
      "grad_norm": 3.498791456222534,
      "learning_rate": 8.230558686201899e-07,
      "loss": 0.0737,
      "step": 1227
    },
    {
      "epoch": 0.19371376740150648,
      "grad_norm": 6.459693908691406,
      "learning_rate": 8.228948639510545e-07,
      "loss": 0.0896,
      "step": 1228
    },
    {
      "epoch": 0.19387151476909728,
      "grad_norm": 4.633272171020508,
      "learning_rate": 8.227338592819192e-07,
      "loss": 0.0434,
      "step": 1229
    },
    {
      "epoch": 0.19402926213668809,
      "grad_norm": 6.239524841308594,
      "learning_rate": 8.225728546127837e-07,
      "loss": 0.0456,
      "step": 1230
    },
    {
      "epoch": 0.1941870095042789,
      "grad_norm": 3.814872980117798,
      "learning_rate": 8.224118499436483e-07,
      "loss": 0.06,
      "step": 1231
    },
    {
      "epoch": 0.1943447568718697,
      "grad_norm": 4.028870582580566,
      "learning_rate": 8.222508452745129e-07,
      "loss": 0.0408,
      "step": 1232
    },
    {
      "epoch": 0.1945025042394605,
      "grad_norm": 12.71030330657959,
      "learning_rate": 8.220898406053775e-07,
      "loss": 0.0828,
      "step": 1233
    },
    {
      "epoch": 0.1946602516070513,
      "grad_norm": 2.8151774406433105,
      "learning_rate": 8.21928835936242e-07,
      "loss": 0.0308,
      "step": 1234
    },
    {
      "epoch": 0.1948179989746421,
      "grad_norm": 8.508167266845703,
      "learning_rate": 8.217678312671067e-07,
      "loss": 0.0902,
      "step": 1235
    },
    {
      "epoch": 0.19497574634223291,
      "grad_norm": 7.739496231079102,
      "learning_rate": 8.216068265979714e-07,
      "loss": 0.0605,
      "step": 1236
    },
    {
      "epoch": 0.19513349370982372,
      "grad_norm": 5.401721477508545,
      "learning_rate": 8.21445821928836e-07,
      "loss": 0.0559,
      "step": 1237
    },
    {
      "epoch": 0.19529124107741452,
      "grad_norm": 3.6133265495300293,
      "learning_rate": 8.212848172597005e-07,
      "loss": 0.0669,
      "step": 1238
    },
    {
      "epoch": 0.19544898844500533,
      "grad_norm": 6.507582664489746,
      "learning_rate": 8.211238125905651e-07,
      "loss": 0.0815,
      "step": 1239
    },
    {
      "epoch": 0.19560673581259613,
      "grad_norm": 3.1642470359802246,
      "learning_rate": 8.209628079214297e-07,
      "loss": 0.031,
      "step": 1240
    },
    {
      "epoch": 0.19576448318018694,
      "grad_norm": 4.73131799697876,
      "learning_rate": 8.208018032522943e-07,
      "loss": 0.0462,
      "step": 1241
    },
    {
      "epoch": 0.19592223054777774,
      "grad_norm": 8.675853729248047,
      "learning_rate": 8.206407985831588e-07,
      "loss": 0.1086,
      "step": 1242
    },
    {
      "epoch": 0.19607997791536855,
      "grad_norm": 3.636523962020874,
      "learning_rate": 8.204797939140234e-07,
      "loss": 0.0459,
      "step": 1243
    },
    {
      "epoch": 0.19623772528295935,
      "grad_norm": 8.941878318786621,
      "learning_rate": 8.203187892448881e-07,
      "loss": 0.0617,
      "step": 1244
    },
    {
      "epoch": 0.19639547265055016,
      "grad_norm": 1.8642269372940063,
      "learning_rate": 8.201577845757527e-07,
      "loss": 0.0185,
      "step": 1245
    },
    {
      "epoch": 0.19655322001814093,
      "grad_norm": 11.972249984741211,
      "learning_rate": 8.199967799066173e-07,
      "loss": 0.0908,
      "step": 1246
    },
    {
      "epoch": 0.19671096738573174,
      "grad_norm": 3.977522850036621,
      "learning_rate": 8.198357752374818e-07,
      "loss": 0.0561,
      "step": 1247
    },
    {
      "epoch": 0.19686871475332254,
      "grad_norm": 7.002127647399902,
      "learning_rate": 8.196747705683464e-07,
      "loss": 0.1224,
      "step": 1248
    },
    {
      "epoch": 0.19702646212091335,
      "grad_norm": 3.3856682777404785,
      "learning_rate": 8.19513765899211e-07,
      "loss": 0.0444,
      "step": 1249
    },
    {
      "epoch": 0.19718420948850415,
      "grad_norm": 7.356138706207275,
      "learning_rate": 8.193527612300756e-07,
      "loss": 0.0639,
      "step": 1250
    },
    {
      "epoch": 0.19734195685609496,
      "grad_norm": 5.135031700134277,
      "learning_rate": 8.191917565609402e-07,
      "loss": 0.0695,
      "step": 1251
    },
    {
      "epoch": 0.19749970422368576,
      "grad_norm": 5.035935878753662,
      "learning_rate": 8.190307518918049e-07,
      "loss": 0.0493,
      "step": 1252
    },
    {
      "epoch": 0.19765745159127657,
      "grad_norm": 3.736492872238159,
      "learning_rate": 8.188697472226695e-07,
      "loss": 0.0306,
      "step": 1253
    },
    {
      "epoch": 0.19781519895886737,
      "grad_norm": 8.930371284484863,
      "learning_rate": 8.187087425535341e-07,
      "loss": 0.087,
      "step": 1254
    },
    {
      "epoch": 0.19797294632645818,
      "grad_norm": 7.436129570007324,
      "learning_rate": 8.185477378843986e-07,
      "loss": 0.0928,
      "step": 1255
    },
    {
      "epoch": 0.19813069369404898,
      "grad_norm": 9.20053768157959,
      "learning_rate": 8.183867332152632e-07,
      "loss": 0.1073,
      "step": 1256
    },
    {
      "epoch": 0.1982884410616398,
      "grad_norm": 9.247213363647461,
      "learning_rate": 8.182257285461278e-07,
      "loss": 0.0914,
      "step": 1257
    },
    {
      "epoch": 0.1984461884292306,
      "grad_norm": 5.168614387512207,
      "learning_rate": 8.180647238769924e-07,
      "loss": 0.0525,
      "step": 1258
    },
    {
      "epoch": 0.1986039357968214,
      "grad_norm": 4.512944221496582,
      "learning_rate": 8.17903719207857e-07,
      "loss": 0.0874,
      "step": 1259
    },
    {
      "epoch": 0.1987616831644122,
      "grad_norm": 4.579786777496338,
      "learning_rate": 8.177427145387216e-07,
      "loss": 0.0633,
      "step": 1260
    },
    {
      "epoch": 0.198919430532003,
      "grad_norm": 7.017485618591309,
      "learning_rate": 8.175817098695862e-07,
      "loss": 0.072,
      "step": 1261
    },
    {
      "epoch": 0.1990771778995938,
      "grad_norm": 6.463603496551514,
      "learning_rate": 8.174207052004508e-07,
      "loss": 0.0697,
      "step": 1262
    },
    {
      "epoch": 0.19923492526718461,
      "grad_norm": 5.57963752746582,
      "learning_rate": 8.172597005313153e-07,
      "loss": 0.1092,
      "step": 1263
    },
    {
      "epoch": 0.19939267263477542,
      "grad_norm": 7.083054065704346,
      "learning_rate": 8.170986958621799e-07,
      "loss": 0.0666,
      "step": 1264
    },
    {
      "epoch": 0.1995504200023662,
      "grad_norm": 8.915925979614258,
      "learning_rate": 8.169376911930445e-07,
      "loss": 0.1097,
      "step": 1265
    },
    {
      "epoch": 0.199708167369957,
      "grad_norm": 6.554561614990234,
      "learning_rate": 8.167766865239092e-07,
      "loss": 0.0687,
      "step": 1266
    },
    {
      "epoch": 0.1998659147375478,
      "grad_norm": 7.839459419250488,
      "learning_rate": 8.166156818547737e-07,
      "loss": 0.0548,
      "step": 1267
    },
    {
      "epoch": 0.2000236621051386,
      "grad_norm": 6.417100429534912,
      "learning_rate": 8.164546771856383e-07,
      "loss": 0.0362,
      "step": 1268
    },
    {
      "epoch": 0.20018140947272942,
      "grad_norm": 6.159739017486572,
      "learning_rate": 8.16293672516503e-07,
      "loss": 0.0846,
      "step": 1269
    },
    {
      "epoch": 0.20033915684032022,
      "grad_norm": 3.370793342590332,
      "learning_rate": 8.161326678473676e-07,
      "loss": 0.0212,
      "step": 1270
    },
    {
      "epoch": 0.20049690420791103,
      "grad_norm": 7.1980204582214355,
      "learning_rate": 8.159716631782322e-07,
      "loss": 0.0891,
      "step": 1271
    },
    {
      "epoch": 0.20065465157550183,
      "grad_norm": 6.110872268676758,
      "learning_rate": 8.158106585090967e-07,
      "loss": 0.051,
      "step": 1272
    },
    {
      "epoch": 0.20081239894309263,
      "grad_norm": 3.0168726444244385,
      "learning_rate": 8.156496538399613e-07,
      "loss": 0.0416,
      "step": 1273
    },
    {
      "epoch": 0.20097014631068344,
      "grad_norm": 6.868875980377197,
      "learning_rate": 8.15488649170826e-07,
      "loss": 0.0664,
      "step": 1274
    },
    {
      "epoch": 0.20112789367827424,
      "grad_norm": 4.881394863128662,
      "learning_rate": 8.153276445016906e-07,
      "loss": 0.066,
      "step": 1275
    },
    {
      "epoch": 0.20128564104586505,
      "grad_norm": 8.150283813476562,
      "learning_rate": 8.151666398325551e-07,
      "loss": 0.1026,
      "step": 1276
    },
    {
      "epoch": 0.20144338841345585,
      "grad_norm": 8.353609085083008,
      "learning_rate": 8.150056351634197e-07,
      "loss": 0.0588,
      "step": 1277
    },
    {
      "epoch": 0.20160113578104666,
      "grad_norm": 8.24417781829834,
      "learning_rate": 8.148446304942843e-07,
      "loss": 0.0824,
      "step": 1278
    },
    {
      "epoch": 0.20175888314863746,
      "grad_norm": 3.7639694213867188,
      "learning_rate": 8.146836258251489e-07,
      "loss": 0.0743,
      "step": 1279
    },
    {
      "epoch": 0.20191663051622827,
      "grad_norm": 4.55227518081665,
      "learning_rate": 8.145226211560134e-07,
      "loss": 0.0685,
      "step": 1280
    },
    {
      "epoch": 0.20207437788381907,
      "grad_norm": 5.401048183441162,
      "learning_rate": 8.143616164868781e-07,
      "loss": 0.0287,
      "step": 1281
    },
    {
      "epoch": 0.20223212525140988,
      "grad_norm": 5.142983913421631,
      "learning_rate": 8.142006118177427e-07,
      "loss": 0.0818,
      "step": 1282
    },
    {
      "epoch": 0.20238987261900068,
      "grad_norm": 5.776869773864746,
      "learning_rate": 8.140396071486073e-07,
      "loss": 0.0632,
      "step": 1283
    },
    {
      "epoch": 0.2025476199865915,
      "grad_norm": 4.408137321472168,
      "learning_rate": 8.138786024794718e-07,
      "loss": 0.0448,
      "step": 1284
    },
    {
      "epoch": 0.20270536735418226,
      "grad_norm": 7.559304237365723,
      "learning_rate": 8.137175978103364e-07,
      "loss": 0.0956,
      "step": 1285
    },
    {
      "epoch": 0.20286311472177307,
      "grad_norm": 5.189079284667969,
      "learning_rate": 8.13556593141201e-07,
      "loss": 0.0409,
      "step": 1286
    },
    {
      "epoch": 0.20302086208936387,
      "grad_norm": 4.444090843200684,
      "learning_rate": 8.133955884720657e-07,
      "loss": 0.0585,
      "step": 1287
    },
    {
      "epoch": 0.20317860945695468,
      "grad_norm": 2.579726457595825,
      "learning_rate": 8.132345838029302e-07,
      "loss": 0.0307,
      "step": 1288
    },
    {
      "epoch": 0.20333635682454548,
      "grad_norm": 9.362491607666016,
      "learning_rate": 8.130735791337949e-07,
      "loss": 0.0689,
      "step": 1289
    },
    {
      "epoch": 0.2034941041921363,
      "grad_norm": 6.831413745880127,
      "learning_rate": 8.129125744646595e-07,
      "loss": 0.0439,
      "step": 1290
    },
    {
      "epoch": 0.2036518515597271,
      "grad_norm": 2.6339004039764404,
      "learning_rate": 8.127515697955241e-07,
      "loss": 0.028,
      "step": 1291
    },
    {
      "epoch": 0.2038095989273179,
      "grad_norm": 7.570190906524658,
      "learning_rate": 8.125905651263887e-07,
      "loss": 0.0692,
      "step": 1292
    },
    {
      "epoch": 0.2039673462949087,
      "grad_norm": 4.936234474182129,
      "learning_rate": 8.124295604572532e-07,
      "loss": 0.0602,
      "step": 1293
    },
    {
      "epoch": 0.2041250936624995,
      "grad_norm": 4.387109756469727,
      "learning_rate": 8.122685557881178e-07,
      "loss": 0.0481,
      "step": 1294
    },
    {
      "epoch": 0.2042828410300903,
      "grad_norm": 5.723708152770996,
      "learning_rate": 8.121075511189824e-07,
      "loss": 0.0404,
      "step": 1295
    },
    {
      "epoch": 0.20444058839768112,
      "grad_norm": 3.7193493843078613,
      "learning_rate": 8.119465464498471e-07,
      "loss": 0.0392,
      "step": 1296
    },
    {
      "epoch": 0.20459833576527192,
      "grad_norm": 8.105212211608887,
      "learning_rate": 8.117855417807116e-07,
      "loss": 0.0664,
      "step": 1297
    },
    {
      "epoch": 0.20475608313286273,
      "grad_norm": 5.5150628089904785,
      "learning_rate": 8.116245371115762e-07,
      "loss": 0.0491,
      "step": 1298
    },
    {
      "epoch": 0.20491383050045353,
      "grad_norm": 7.704408645629883,
      "learning_rate": 8.114635324424408e-07,
      "loss": 0.0674,
      "step": 1299
    },
    {
      "epoch": 0.20507157786804434,
      "grad_norm": 4.652005672454834,
      "learning_rate": 8.113025277733054e-07,
      "loss": 0.0475,
      "step": 1300
    },
    {
      "epoch": 0.20522932523563514,
      "grad_norm": 6.252089023590088,
      "learning_rate": 8.111415231041699e-07,
      "loss": 0.051,
      "step": 1301
    },
    {
      "epoch": 0.20538707260322595,
      "grad_norm": 8.732755661010742,
      "learning_rate": 8.109805184350345e-07,
      "loss": 0.0956,
      "step": 1302
    },
    {
      "epoch": 0.20554481997081675,
      "grad_norm": 5.888700485229492,
      "learning_rate": 8.108195137658991e-07,
      "loss": 0.1075,
      "step": 1303
    },
    {
      "epoch": 0.20570256733840753,
      "grad_norm": 5.2593674659729,
      "learning_rate": 8.106585090967639e-07,
      "loss": 0.0555,
      "step": 1304
    },
    {
      "epoch": 0.20586031470599833,
      "grad_norm": 7.008612155914307,
      "learning_rate": 8.104975044276284e-07,
      "loss": 0.0594,
      "step": 1305
    },
    {
      "epoch": 0.20601806207358914,
      "grad_norm": 9.058858871459961,
      "learning_rate": 8.10336499758493e-07,
      "loss": 0.0775,
      "step": 1306
    },
    {
      "epoch": 0.20617580944117994,
      "grad_norm": 5.718960762023926,
      "learning_rate": 8.101754950893576e-07,
      "loss": 0.0513,
      "step": 1307
    },
    {
      "epoch": 0.20633355680877075,
      "grad_norm": 4.485774040222168,
      "learning_rate": 8.100144904202222e-07,
      "loss": 0.0474,
      "step": 1308
    },
    {
      "epoch": 0.20649130417636155,
      "grad_norm": 10.537572860717773,
      "learning_rate": 8.098534857510867e-07,
      "loss": 0.0431,
      "step": 1309
    },
    {
      "epoch": 0.20664905154395236,
      "grad_norm": 5.365055561065674,
      "learning_rate": 8.096924810819513e-07,
      "loss": 0.0398,
      "step": 1310
    },
    {
      "epoch": 0.20680679891154316,
      "grad_norm": 8.628813743591309,
      "learning_rate": 8.09531476412816e-07,
      "loss": 0.0722,
      "step": 1311
    },
    {
      "epoch": 0.20696454627913397,
      "grad_norm": 6.088448524475098,
      "learning_rate": 8.093704717436806e-07,
      "loss": 0.1014,
      "step": 1312
    },
    {
      "epoch": 0.20712229364672477,
      "grad_norm": 3.6947081089019775,
      "learning_rate": 8.092094670745451e-07,
      "loss": 0.0686,
      "step": 1313
    },
    {
      "epoch": 0.20728004101431557,
      "grad_norm": 3.5171139240264893,
      "learning_rate": 8.090484624054097e-07,
      "loss": 0.038,
      "step": 1314
    },
    {
      "epoch": 0.20743778838190638,
      "grad_norm": 6.769224166870117,
      "learning_rate": 8.088874577362743e-07,
      "loss": 0.0867,
      "step": 1315
    },
    {
      "epoch": 0.20759553574949718,
      "grad_norm": 5.466546058654785,
      "learning_rate": 8.087264530671389e-07,
      "loss": 0.0652,
      "step": 1316
    },
    {
      "epoch": 0.207753283117088,
      "grad_norm": 3.244213581085205,
      "learning_rate": 8.085654483980035e-07,
      "loss": 0.0783,
      "step": 1317
    },
    {
      "epoch": 0.2079110304846788,
      "grad_norm": 3.2699074745178223,
      "learning_rate": 8.08404443728868e-07,
      "loss": 0.0478,
      "step": 1318
    },
    {
      "epoch": 0.2080687778522696,
      "grad_norm": 7.128431797027588,
      "learning_rate": 8.082434390597327e-07,
      "loss": 0.046,
      "step": 1319
    },
    {
      "epoch": 0.2082265252198604,
      "grad_norm": 7.5372467041015625,
      "learning_rate": 8.080824343905973e-07,
      "loss": 0.0688,
      "step": 1320
    },
    {
      "epoch": 0.2083842725874512,
      "grad_norm": 2.642981767654419,
      "learning_rate": 8.07921429721462e-07,
      "loss": 0.0256,
      "step": 1321
    },
    {
      "epoch": 0.208542019955042,
      "grad_norm": 7.668417930603027,
      "learning_rate": 8.077604250523264e-07,
      "loss": 0.0748,
      "step": 1322
    },
    {
      "epoch": 0.2086997673226328,
      "grad_norm": 8.645655632019043,
      "learning_rate": 8.075994203831911e-07,
      "loss": 0.0862,
      "step": 1323
    },
    {
      "epoch": 0.2088575146902236,
      "grad_norm": 2.2922632694244385,
      "learning_rate": 8.074384157140557e-07,
      "loss": 0.0268,
      "step": 1324
    },
    {
      "epoch": 0.2090152620578144,
      "grad_norm": 5.980663776397705,
      "learning_rate": 8.072774110449203e-07,
      "loss": 0.0716,
      "step": 1325
    },
    {
      "epoch": 0.2091730094254052,
      "grad_norm": 4.555629730224609,
      "learning_rate": 8.071164063757848e-07,
      "loss": 0.0333,
      "step": 1326
    },
    {
      "epoch": 0.209330756792996,
      "grad_norm": 4.295013427734375,
      "learning_rate": 8.069554017066495e-07,
      "loss": 0.025,
      "step": 1327
    },
    {
      "epoch": 0.2094885041605868,
      "grad_norm": 7.122177600860596,
      "learning_rate": 8.067943970375141e-07,
      "loss": 0.047,
      "step": 1328
    },
    {
      "epoch": 0.20964625152817762,
      "grad_norm": 5.486662864685059,
      "learning_rate": 8.066333923683787e-07,
      "loss": 0.0724,
      "step": 1329
    },
    {
      "epoch": 0.20980399889576842,
      "grad_norm": 3.3127543926239014,
      "learning_rate": 8.064723876992432e-07,
      "loss": 0.0235,
      "step": 1330
    },
    {
      "epoch": 0.20996174626335923,
      "grad_norm": 4.185500621795654,
      "learning_rate": 8.063113830301078e-07,
      "loss": 0.0151,
      "step": 1331
    },
    {
      "epoch": 0.21011949363095003,
      "grad_norm": 6.932736873626709,
      "learning_rate": 8.061503783609724e-07,
      "loss": 0.1283,
      "step": 1332
    },
    {
      "epoch": 0.21027724099854084,
      "grad_norm": 3.9714646339416504,
      "learning_rate": 8.05989373691837e-07,
      "loss": 0.0159,
      "step": 1333
    },
    {
      "epoch": 0.21043498836613164,
      "grad_norm": 6.8552350997924805,
      "learning_rate": 8.058283690227016e-07,
      "loss": 0.0685,
      "step": 1334
    },
    {
      "epoch": 0.21059273573372245,
      "grad_norm": 5.01610803604126,
      "learning_rate": 8.056673643535662e-07,
      "loss": 0.0534,
      "step": 1335
    },
    {
      "epoch": 0.21075048310131325,
      "grad_norm": 5.589885711669922,
      "learning_rate": 8.055063596844308e-07,
      "loss": 0.0417,
      "step": 1336
    },
    {
      "epoch": 0.21090823046890406,
      "grad_norm": 5.473964214324951,
      "learning_rate": 8.053453550152954e-07,
      "loss": 0.0433,
      "step": 1337
    },
    {
      "epoch": 0.21106597783649486,
      "grad_norm": 2.978450298309326,
      "learning_rate": 8.0518435034616e-07,
      "loss": 0.054,
      "step": 1338
    },
    {
      "epoch": 0.21122372520408567,
      "grad_norm": 8.028962135314941,
      "learning_rate": 8.050233456770245e-07,
      "loss": 0.0431,
      "step": 1339
    },
    {
      "epoch": 0.21138147257167647,
      "grad_norm": 4.471803188323975,
      "learning_rate": 8.048623410078891e-07,
      "loss": 0.0774,
      "step": 1340
    },
    {
      "epoch": 0.21153921993926728,
      "grad_norm": 6.427778720855713,
      "learning_rate": 8.047013363387539e-07,
      "loss": 0.0652,
      "step": 1341
    },
    {
      "epoch": 0.21169696730685808,
      "grad_norm": 4.534457206726074,
      "learning_rate": 8.045403316696185e-07,
      "loss": 0.061,
      "step": 1342
    },
    {
      "epoch": 0.21185471467444886,
      "grad_norm": 5.0695576667785645,
      "learning_rate": 8.04379327000483e-07,
      "loss": 0.0607,
      "step": 1343
    },
    {
      "epoch": 0.21201246204203966,
      "grad_norm": 7.301562786102295,
      "learning_rate": 8.042183223313476e-07,
      "loss": 0.0793,
      "step": 1344
    },
    {
      "epoch": 0.21217020940963047,
      "grad_norm": 5.415078639984131,
      "learning_rate": 8.040573176622122e-07,
      "loss": 0.0633,
      "step": 1345
    },
    {
      "epoch": 0.21232795677722127,
      "grad_norm": 7.417275428771973,
      "learning_rate": 8.038963129930768e-07,
      "loss": 0.0794,
      "step": 1346
    },
    {
      "epoch": 0.21248570414481208,
      "grad_norm": 5.172476291656494,
      "learning_rate": 8.037353083239413e-07,
      "loss": 0.0586,
      "step": 1347
    },
    {
      "epoch": 0.21264345151240288,
      "grad_norm": 4.431525230407715,
      "learning_rate": 8.035743036548059e-07,
      "loss": 0.0413,
      "step": 1348
    },
    {
      "epoch": 0.21280119887999369,
      "grad_norm": 5.791142463684082,
      "learning_rate": 8.034132989856706e-07,
      "loss": 0.0636,
      "step": 1349
    },
    {
      "epoch": 0.2129589462475845,
      "grad_norm": 4.702826023101807,
      "learning_rate": 8.032522943165352e-07,
      "loss": 0.0536,
      "step": 1350
    },
    {
      "epoch": 0.2131166936151753,
      "grad_norm": 8.566353797912598,
      "learning_rate": 8.030912896473997e-07,
      "loss": 0.0381,
      "step": 1351
    },
    {
      "epoch": 0.2132744409827661,
      "grad_norm": 6.175302028656006,
      "learning_rate": 8.029302849782643e-07,
      "loss": 0.0686,
      "step": 1352
    },
    {
      "epoch": 0.2134321883503569,
      "grad_norm": 4.509758472442627,
      "learning_rate": 8.027692803091289e-07,
      "loss": 0.0375,
      "step": 1353
    },
    {
      "epoch": 0.2135899357179477,
      "grad_norm": 9.859237670898438,
      "learning_rate": 8.026082756399935e-07,
      "loss": 0.0583,
      "step": 1354
    },
    {
      "epoch": 0.21374768308553851,
      "grad_norm": 3.6694836616516113,
      "learning_rate": 8.02447270970858e-07,
      "loss": 0.0303,
      "step": 1355
    },
    {
      "epoch": 0.21390543045312932,
      "grad_norm": 4.9124579429626465,
      "learning_rate": 8.022862663017226e-07,
      "loss": 0.0792,
      "step": 1356
    },
    {
      "epoch": 0.21406317782072012,
      "grad_norm": 4.5436110496521,
      "learning_rate": 8.021252616325874e-07,
      "loss": 0.075,
      "step": 1357
    },
    {
      "epoch": 0.21422092518831093,
      "grad_norm": 7.841489791870117,
      "learning_rate": 8.01964256963452e-07,
      "loss": 0.0614,
      "step": 1358
    },
    {
      "epoch": 0.21437867255590173,
      "grad_norm": 10.84503173828125,
      "learning_rate": 8.018032522943165e-07,
      "loss": 0.1507,
      "step": 1359
    },
    {
      "epoch": 0.21453641992349254,
      "grad_norm": 5.0677103996276855,
      "learning_rate": 8.016422476251811e-07,
      "loss": 0.058,
      "step": 1360
    },
    {
      "epoch": 0.21469416729108334,
      "grad_norm": 5.633657932281494,
      "learning_rate": 8.014812429560457e-07,
      "loss": 0.0513,
      "step": 1361
    },
    {
      "epoch": 0.21485191465867412,
      "grad_norm": 6.73211669921875,
      "learning_rate": 8.013202382869103e-07,
      "loss": 0.0705,
      "step": 1362
    },
    {
      "epoch": 0.21500966202626492,
      "grad_norm": 10.13681697845459,
      "learning_rate": 8.011592336177749e-07,
      "loss": 0.0725,
      "step": 1363
    },
    {
      "epoch": 0.21516740939385573,
      "grad_norm": 8.70879077911377,
      "learning_rate": 8.009982289486395e-07,
      "loss": 0.0515,
      "step": 1364
    },
    {
      "epoch": 0.21532515676144653,
      "grad_norm": 4.02731466293335,
      "learning_rate": 8.008372242795041e-07,
      "loss": 0.0526,
      "step": 1365
    },
    {
      "epoch": 0.21548290412903734,
      "grad_norm": 6.5245466232299805,
      "learning_rate": 8.006762196103687e-07,
      "loss": 0.087,
      "step": 1366
    },
    {
      "epoch": 0.21564065149662814,
      "grad_norm": 5.074929237365723,
      "learning_rate": 8.005152149412333e-07,
      "loss": 0.0595,
      "step": 1367
    },
    {
      "epoch": 0.21579839886421895,
      "grad_norm": 4.632253170013428,
      "learning_rate": 8.003542102720978e-07,
      "loss": 0.0488,
      "step": 1368
    },
    {
      "epoch": 0.21595614623180975,
      "grad_norm": 4.121384620666504,
      "learning_rate": 8.001932056029624e-07,
      "loss": 0.0288,
      "step": 1369
    },
    {
      "epoch": 0.21611389359940056,
      "grad_norm": 6.601824760437012,
      "learning_rate": 8.00032200933827e-07,
      "loss": 0.0931,
      "step": 1370
    },
    {
      "epoch": 0.21627164096699136,
      "grad_norm": 6.737264633178711,
      "learning_rate": 7.998711962646917e-07,
      "loss": 0.04,
      "step": 1371
    },
    {
      "epoch": 0.21642938833458217,
      "grad_norm": 7.603921413421631,
      "learning_rate": 7.997101915955562e-07,
      "loss": 0.05,
      "step": 1372
    },
    {
      "epoch": 0.21658713570217297,
      "grad_norm": 4.314795017242432,
      "learning_rate": 7.995491869264208e-07,
      "loss": 0.0385,
      "step": 1373
    },
    {
      "epoch": 0.21674488306976378,
      "grad_norm": 7.3531174659729,
      "learning_rate": 7.993881822572854e-07,
      "loss": 0.0681,
      "step": 1374
    },
    {
      "epoch": 0.21690263043735458,
      "grad_norm": 14.709685325622559,
      "learning_rate": 7.9922717758815e-07,
      "loss": 0.0719,
      "step": 1375
    },
    {
      "epoch": 0.2170603778049454,
      "grad_norm": 6.657068252563477,
      "learning_rate": 7.990661729190146e-07,
      "loss": 0.0705,
      "step": 1376
    },
    {
      "epoch": 0.2172181251725362,
      "grad_norm": 5.743774890899658,
      "learning_rate": 7.989051682498792e-07,
      "loss": 0.0613,
      "step": 1377
    },
    {
      "epoch": 0.217375872540127,
      "grad_norm": 4.418606281280518,
      "learning_rate": 7.987441635807438e-07,
      "loss": 0.0876,
      "step": 1378
    },
    {
      "epoch": 0.2175336199077178,
      "grad_norm": 7.777172088623047,
      "learning_rate": 7.985831589116085e-07,
      "loss": 0.0558,
      "step": 1379
    },
    {
      "epoch": 0.2176913672753086,
      "grad_norm": 1.5176472663879395,
      "learning_rate": 7.98422154242473e-07,
      "loss": 0.0112,
      "step": 1380
    },
    {
      "epoch": 0.21784911464289938,
      "grad_norm": 5.456096172332764,
      "learning_rate": 7.982611495733376e-07,
      "loss": 0.043,
      "step": 1381
    },
    {
      "epoch": 0.2180068620104902,
      "grad_norm": 11.711322784423828,
      "learning_rate": 7.981001449042022e-07,
      "loss": 0.1085,
      "step": 1382
    },
    {
      "epoch": 0.218164609378081,
      "grad_norm": 3.6353113651275635,
      "learning_rate": 7.979391402350668e-07,
      "loss": 0.0245,
      "step": 1383
    },
    {
      "epoch": 0.2183223567456718,
      "grad_norm": 8.09837818145752,
      "learning_rate": 7.977781355659313e-07,
      "loss": 0.0565,
      "step": 1384
    },
    {
      "epoch": 0.2184801041132626,
      "grad_norm": 6.092850208282471,
      "learning_rate": 7.976171308967959e-07,
      "loss": 0.0585,
      "step": 1385
    },
    {
      "epoch": 0.2186378514808534,
      "grad_norm": 7.001495361328125,
      "learning_rate": 7.974561262276606e-07,
      "loss": 0.0659,
      "step": 1386
    },
    {
      "epoch": 0.2187955988484442,
      "grad_norm": 5.3207197189331055,
      "learning_rate": 7.972951215585252e-07,
      "loss": 0.0883,
      "step": 1387
    },
    {
      "epoch": 0.21895334621603502,
      "grad_norm": 5.347443103790283,
      "learning_rate": 7.971341168893898e-07,
      "loss": 0.0562,
      "step": 1388
    },
    {
      "epoch": 0.21911109358362582,
      "grad_norm": 6.29995059967041,
      "learning_rate": 7.969731122202543e-07,
      "loss": 0.0423,
      "step": 1389
    },
    {
      "epoch": 0.21926884095121663,
      "grad_norm": 2.837951183319092,
      "learning_rate": 7.968121075511189e-07,
      "loss": 0.0388,
      "step": 1390
    },
    {
      "epoch": 0.21942658831880743,
      "grad_norm": 4.2165985107421875,
      "learning_rate": 7.966511028819835e-07,
      "loss": 0.0278,
      "step": 1391
    },
    {
      "epoch": 0.21958433568639824,
      "grad_norm": 4.890751361846924,
      "learning_rate": 7.964900982128481e-07,
      "loss": 0.0794,
      "step": 1392
    },
    {
      "epoch": 0.21974208305398904,
      "grad_norm": 5.265214920043945,
      "learning_rate": 7.963290935437126e-07,
      "loss": 0.0353,
      "step": 1393
    },
    {
      "epoch": 0.21989983042157984,
      "grad_norm": 3.7690224647521973,
      "learning_rate": 7.961680888745774e-07,
      "loss": 0.0438,
      "step": 1394
    },
    {
      "epoch": 0.22005757778917065,
      "grad_norm": 5.014218330383301,
      "learning_rate": 7.96007084205442e-07,
      "loss": 0.0512,
      "step": 1395
    },
    {
      "epoch": 0.22021532515676145,
      "grad_norm": 9.617183685302734,
      "learning_rate": 7.958460795363066e-07,
      "loss": 0.0949,
      "step": 1396
    },
    {
      "epoch": 0.22037307252435226,
      "grad_norm": 2.744088649749756,
      "learning_rate": 7.956850748671711e-07,
      "loss": 0.0335,
      "step": 1397
    },
    {
      "epoch": 0.22053081989194306,
      "grad_norm": 3.987654447555542,
      "learning_rate": 7.955240701980357e-07,
      "loss": 0.0356,
      "step": 1398
    },
    {
      "epoch": 0.22068856725953387,
      "grad_norm": 8.036004066467285,
      "learning_rate": 7.953630655289003e-07,
      "loss": 0.0422,
      "step": 1399
    },
    {
      "epoch": 0.22084631462712467,
      "grad_norm": 6.1750030517578125,
      "learning_rate": 7.952020608597649e-07,
      "loss": 0.0475,
      "step": 1400
    },
    {
      "epoch": 0.22100406199471545,
      "grad_norm": 5.037681579589844,
      "learning_rate": 7.950410561906294e-07,
      "loss": 0.0372,
      "step": 1401
    },
    {
      "epoch": 0.22116180936230626,
      "grad_norm": 3.706131935119629,
      "learning_rate": 7.948800515214941e-07,
      "loss": 0.0447,
      "step": 1402
    },
    {
      "epoch": 0.22131955672989706,
      "grad_norm": 3.950047731399536,
      "learning_rate": 7.947190468523587e-07,
      "loss": 0.0664,
      "step": 1403
    },
    {
      "epoch": 0.22147730409748786,
      "grad_norm": 3.1588940620422363,
      "learning_rate": 7.945580421832233e-07,
      "loss": 0.0358,
      "step": 1404
    },
    {
      "epoch": 0.22163505146507867,
      "grad_norm": 3.942070722579956,
      "learning_rate": 7.943970375140878e-07,
      "loss": 0.0551,
      "step": 1405
    },
    {
      "epoch": 0.22179279883266947,
      "grad_norm": 4.074398994445801,
      "learning_rate": 7.942360328449524e-07,
      "loss": 0.0352,
      "step": 1406
    },
    {
      "epoch": 0.22195054620026028,
      "grad_norm": 4.072987079620361,
      "learning_rate": 7.94075028175817e-07,
      "loss": 0.0698,
      "step": 1407
    },
    {
      "epoch": 0.22210829356785108,
      "grad_norm": 6.349443435668945,
      "learning_rate": 7.939140235066816e-07,
      "loss": 0.0742,
      "step": 1408
    },
    {
      "epoch": 0.2222660409354419,
      "grad_norm": 5.44382381439209,
      "learning_rate": 7.937530188375463e-07,
      "loss": 0.0634,
      "step": 1409
    },
    {
      "epoch": 0.2224237883030327,
      "grad_norm": 5.967154502868652,
      "learning_rate": 7.935920141684109e-07,
      "loss": 0.0765,
      "step": 1410
    },
    {
      "epoch": 0.2225815356706235,
      "grad_norm": 7.5302839279174805,
      "learning_rate": 7.934310094992755e-07,
      "loss": 0.0446,
      "step": 1411
    },
    {
      "epoch": 0.2227392830382143,
      "grad_norm": 3.2310304641723633,
      "learning_rate": 7.932700048301401e-07,
      "loss": 0.0167,
      "step": 1412
    },
    {
      "epoch": 0.2228970304058051,
      "grad_norm": 7.038295745849609,
      "learning_rate": 7.931090001610047e-07,
      "loss": 0.0896,
      "step": 1413
    },
    {
      "epoch": 0.2230547777733959,
      "grad_norm": 6.531866550445557,
      "learning_rate": 7.929479954918692e-07,
      "loss": 0.0629,
      "step": 1414
    },
    {
      "epoch": 0.22321252514098672,
      "grad_norm": 8.073969841003418,
      "learning_rate": 7.927869908227338e-07,
      "loss": 0.0976,
      "step": 1415
    },
    {
      "epoch": 0.22337027250857752,
      "grad_norm": 4.7604851722717285,
      "learning_rate": 7.926259861535985e-07,
      "loss": 0.0232,
      "step": 1416
    },
    {
      "epoch": 0.22352801987616833,
      "grad_norm": 6.101916790008545,
      "learning_rate": 7.924649814844631e-07,
      "loss": 0.0616,
      "step": 1417
    },
    {
      "epoch": 0.22368576724375913,
      "grad_norm": 6.440509796142578,
      "learning_rate": 7.923039768153276e-07,
      "loss": 0.0754,
      "step": 1418
    },
    {
      "epoch": 0.22384351461134994,
      "grad_norm": 8.148072242736816,
      "learning_rate": 7.921429721461922e-07,
      "loss": 0.0685,
      "step": 1419
    },
    {
      "epoch": 0.2240012619789407,
      "grad_norm": 10.16176986694336,
      "learning_rate": 7.919819674770568e-07,
      "loss": 0.0519,
      "step": 1420
    },
    {
      "epoch": 0.22415900934653152,
      "grad_norm": 6.655317306518555,
      "learning_rate": 7.918209628079214e-07,
      "loss": 0.0949,
      "step": 1421
    },
    {
      "epoch": 0.22431675671412232,
      "grad_norm": 7.478221416473389,
      "learning_rate": 7.916599581387859e-07,
      "loss": 0.0734,
      "step": 1422
    },
    {
      "epoch": 0.22447450408171313,
      "grad_norm": 5.7581281661987305,
      "learning_rate": 7.914989534696505e-07,
      "loss": 0.0766,
      "step": 1423
    },
    {
      "epoch": 0.22463225144930393,
      "grad_norm": 10.845542907714844,
      "learning_rate": 7.913379488005152e-07,
      "loss": 0.0732,
      "step": 1424
    },
    {
      "epoch": 0.22478999881689474,
      "grad_norm": 8.282747268676758,
      "learning_rate": 7.911769441313798e-07,
      "loss": 0.0962,
      "step": 1425
    },
    {
      "epoch": 0.22494774618448554,
      "grad_norm": 5.4240193367004395,
      "learning_rate": 7.910159394622443e-07,
      "loss": 0.0231,
      "step": 1426
    },
    {
      "epoch": 0.22510549355207635,
      "grad_norm": 7.236989498138428,
      "learning_rate": 7.908549347931089e-07,
      "loss": 0.0724,
      "step": 1427
    },
    {
      "epoch": 0.22526324091966715,
      "grad_norm": 11.251788139343262,
      "learning_rate": 7.906939301239736e-07,
      "loss": 0.0582,
      "step": 1428
    },
    {
      "epoch": 0.22542098828725796,
      "grad_norm": 5.494878768920898,
      "learning_rate": 7.905329254548382e-07,
      "loss": 0.0589,
      "step": 1429
    },
    {
      "epoch": 0.22557873565484876,
      "grad_norm": 8.22620677947998,
      "learning_rate": 7.903719207857027e-07,
      "loss": 0.0915,
      "step": 1430
    },
    {
      "epoch": 0.22573648302243957,
      "grad_norm": 9.50594711303711,
      "learning_rate": 7.902109161165673e-07,
      "loss": 0.0721,
      "step": 1431
    },
    {
      "epoch": 0.22589423039003037,
      "grad_norm": 7.40491247177124,
      "learning_rate": 7.90049911447432e-07,
      "loss": 0.0521,
      "step": 1432
    },
    {
      "epoch": 0.22605197775762118,
      "grad_norm": 13.662528991699219,
      "learning_rate": 7.898889067782966e-07,
      "loss": 0.117,
      "step": 1433
    },
    {
      "epoch": 0.22620972512521198,
      "grad_norm": 7.314889430999756,
      "learning_rate": 7.897279021091612e-07,
      "loss": 0.0799,
      "step": 1434
    },
    {
      "epoch": 0.22636747249280278,
      "grad_norm": 6.094870567321777,
      "learning_rate": 7.895668974400257e-07,
      "loss": 0.0715,
      "step": 1435
    },
    {
      "epoch": 0.2265252198603936,
      "grad_norm": 2.415949583053589,
      "learning_rate": 7.894058927708903e-07,
      "loss": 0.0255,
      "step": 1436
    },
    {
      "epoch": 0.2266829672279844,
      "grad_norm": 3.3921942710876465,
      "learning_rate": 7.892448881017549e-07,
      "loss": 0.0576,
      "step": 1437
    },
    {
      "epoch": 0.2268407145955752,
      "grad_norm": 6.06171178817749,
      "learning_rate": 7.890838834326195e-07,
      "loss": 0.0725,
      "step": 1438
    },
    {
      "epoch": 0.22699846196316598,
      "grad_norm": 6.461246490478516,
      "learning_rate": 7.889228787634841e-07,
      "loss": 0.0768,
      "step": 1439
    },
    {
      "epoch": 0.22715620933075678,
      "grad_norm": 5.678943157196045,
      "learning_rate": 7.887618740943487e-07,
      "loss": 0.0812,
      "step": 1440
    },
    {
      "epoch": 0.22731395669834759,
      "grad_norm": 4.532436370849609,
      "learning_rate": 7.886008694252133e-07,
      "loss": 0.0417,
      "step": 1441
    },
    {
      "epoch": 0.2274717040659384,
      "grad_norm": 7.049404144287109,
      "learning_rate": 7.884398647560779e-07,
      "loss": 0.1101,
      "step": 1442
    },
    {
      "epoch": 0.2276294514335292,
      "grad_norm": 5.901423931121826,
      "learning_rate": 7.882788600869424e-07,
      "loss": 0.1286,
      "step": 1443
    },
    {
      "epoch": 0.22778719880112,
      "grad_norm": 7.790620803833008,
      "learning_rate": 7.88117855417807e-07,
      "loss": 0.0405,
      "step": 1444
    },
    {
      "epoch": 0.2279449461687108,
      "grad_norm": 5.334627628326416,
      "learning_rate": 7.879568507486716e-07,
      "loss": 0.0479,
      "step": 1445
    },
    {
      "epoch": 0.2281026935363016,
      "grad_norm": 4.426734924316406,
      "learning_rate": 7.877958460795364e-07,
      "loss": 0.0293,
      "step": 1446
    },
    {
      "epoch": 0.22826044090389241,
      "grad_norm": 4.468956470489502,
      "learning_rate": 7.876348414104009e-07,
      "loss": 0.0251,
      "step": 1447
    },
    {
      "epoch": 0.22841818827148322,
      "grad_norm": 5.049088954925537,
      "learning_rate": 7.874738367412655e-07,
      "loss": 0.0462,
      "step": 1448
    },
    {
      "epoch": 0.22857593563907402,
      "grad_norm": 8.321356773376465,
      "learning_rate": 7.873128320721301e-07,
      "loss": 0.0737,
      "step": 1449
    },
    {
      "epoch": 0.22873368300666483,
      "grad_norm": 7.106385231018066,
      "learning_rate": 7.871518274029947e-07,
      "loss": 0.0603,
      "step": 1450
    },
    {
      "epoch": 0.22889143037425563,
      "grad_norm": 5.824122428894043,
      "learning_rate": 7.869908227338592e-07,
      "loss": 0.0539,
      "step": 1451
    },
    {
      "epoch": 0.22904917774184644,
      "grad_norm": 6.70082950592041,
      "learning_rate": 7.868298180647238e-07,
      "loss": 0.1212,
      "step": 1452
    },
    {
      "epoch": 0.22920692510943724,
      "grad_norm": 4.612785339355469,
      "learning_rate": 7.866688133955884e-07,
      "loss": 0.0456,
      "step": 1453
    },
    {
      "epoch": 0.22936467247702805,
      "grad_norm": 9.111032485961914,
      "learning_rate": 7.865078087264531e-07,
      "loss": 0.0929,
      "step": 1454
    },
    {
      "epoch": 0.22952241984461885,
      "grad_norm": 5.721334457397461,
      "learning_rate": 7.863468040573177e-07,
      "loss": 0.1015,
      "step": 1455
    },
    {
      "epoch": 0.22968016721220966,
      "grad_norm": 4.232248783111572,
      "learning_rate": 7.861857993881822e-07,
      "loss": 0.0533,
      "step": 1456
    },
    {
      "epoch": 0.22983791457980046,
      "grad_norm": 8.858681678771973,
      "learning_rate": 7.860247947190468e-07,
      "loss": 0.1098,
      "step": 1457
    },
    {
      "epoch": 0.22999566194739127,
      "grad_norm": 4.750298023223877,
      "learning_rate": 7.858637900499114e-07,
      "loss": 0.0663,
      "step": 1458
    },
    {
      "epoch": 0.23015340931498204,
      "grad_norm": 5.576642036437988,
      "learning_rate": 7.85702785380776e-07,
      "loss": 0.1292,
      "step": 1459
    },
    {
      "epoch": 0.23031115668257285,
      "grad_norm": 4.532076835632324,
      "learning_rate": 7.855417807116405e-07,
      "loss": 0.0765,
      "step": 1460
    },
    {
      "epoch": 0.23046890405016365,
      "grad_norm": 4.859536647796631,
      "learning_rate": 7.853807760425051e-07,
      "loss": 0.0325,
      "step": 1461
    },
    {
      "epoch": 0.23062665141775446,
      "grad_norm": 3.2009525299072266,
      "learning_rate": 7.852197713733698e-07,
      "loss": 0.0568,
      "step": 1462
    },
    {
      "epoch": 0.23078439878534526,
      "grad_norm": 8.067152976989746,
      "learning_rate": 7.850587667042345e-07,
      "loss": 0.0931,
      "step": 1463
    },
    {
      "epoch": 0.23094214615293607,
      "grad_norm": 6.133688449859619,
      "learning_rate": 7.84897762035099e-07,
      "loss": 0.0596,
      "step": 1464
    },
    {
      "epoch": 0.23109989352052687,
      "grad_norm": 5.80610466003418,
      "learning_rate": 7.847367573659636e-07,
      "loss": 0.0347,
      "step": 1465
    },
    {
      "epoch": 0.23125764088811768,
      "grad_norm": 8.714088439941406,
      "learning_rate": 7.845757526968282e-07,
      "loss": 0.1071,
      "step": 1466
    },
    {
      "epoch": 0.23141538825570848,
      "grad_norm": 6.408808708190918,
      "learning_rate": 7.844147480276928e-07,
      "loss": 0.0635,
      "step": 1467
    },
    {
      "epoch": 0.2315731356232993,
      "grad_norm": 5.011511325836182,
      "learning_rate": 7.842537433585573e-07,
      "loss": 0.0445,
      "step": 1468
    },
    {
      "epoch": 0.2317308829908901,
      "grad_norm": 7.067837715148926,
      "learning_rate": 7.84092738689422e-07,
      "loss": 0.0737,
      "step": 1469
    },
    {
      "epoch": 0.2318886303584809,
      "grad_norm": 4.492575645446777,
      "learning_rate": 7.839317340202866e-07,
      "loss": 0.0341,
      "step": 1470
    },
    {
      "epoch": 0.2320463777260717,
      "grad_norm": 2.238722085952759,
      "learning_rate": 7.837707293511512e-07,
      "loss": 0.0211,
      "step": 1471
    },
    {
      "epoch": 0.2322041250936625,
      "grad_norm": 1.7708102464675903,
      "learning_rate": 7.836097246820157e-07,
      "loss": 0.0257,
      "step": 1472
    },
    {
      "epoch": 0.2323618724612533,
      "grad_norm": 7.906129360198975,
      "learning_rate": 7.834487200128803e-07,
      "loss": 0.0535,
      "step": 1473
    },
    {
      "epoch": 0.23251961982884412,
      "grad_norm": 5.424860954284668,
      "learning_rate": 7.832877153437449e-07,
      "loss": 0.0592,
      "step": 1474
    },
    {
      "epoch": 0.23267736719643492,
      "grad_norm": 6.904181003570557,
      "learning_rate": 7.831267106746095e-07,
      "loss": 0.0919,
      "step": 1475
    },
    {
      "epoch": 0.23283511456402572,
      "grad_norm": 2.7707386016845703,
      "learning_rate": 7.82965706005474e-07,
      "loss": 0.0337,
      "step": 1476
    },
    {
      "epoch": 0.23299286193161653,
      "grad_norm": 5.375368595123291,
      "learning_rate": 7.828047013363387e-07,
      "loss": 0.0896,
      "step": 1477
    },
    {
      "epoch": 0.2331506092992073,
      "grad_norm": 6.6298017501831055,
      "learning_rate": 7.826436966672033e-07,
      "loss": 0.076,
      "step": 1478
    },
    {
      "epoch": 0.2333083566667981,
      "grad_norm": 3.2984378337860107,
      "learning_rate": 7.824826919980679e-07,
      "loss": 0.0453,
      "step": 1479
    },
    {
      "epoch": 0.23346610403438892,
      "grad_norm": 3.9714133739471436,
      "learning_rate": 7.823216873289326e-07,
      "loss": 0.0458,
      "step": 1480
    },
    {
      "epoch": 0.23362385140197972,
      "grad_norm": 4.95203161239624,
      "learning_rate": 7.82160682659797e-07,
      "loss": 0.0577,
      "step": 1481
    },
    {
      "epoch": 0.23378159876957053,
      "grad_norm": 6.471922874450684,
      "learning_rate": 7.819996779906617e-07,
      "loss": 0.0984,
      "step": 1482
    },
    {
      "epoch": 0.23393934613716133,
      "grad_norm": 10.565709114074707,
      "learning_rate": 7.818386733215263e-07,
      "loss": 0.0924,
      "step": 1483
    },
    {
      "epoch": 0.23409709350475214,
      "grad_norm": 5.166487216949463,
      "learning_rate": 7.81677668652391e-07,
      "loss": 0.0671,
      "step": 1484
    },
    {
      "epoch": 0.23425484087234294,
      "grad_norm": 4.182351112365723,
      "learning_rate": 7.815166639832555e-07,
      "loss": 0.0479,
      "step": 1485
    },
    {
      "epoch": 0.23441258823993374,
      "grad_norm": 6.305501937866211,
      "learning_rate": 7.813556593141201e-07,
      "loss": 0.044,
      "step": 1486
    },
    {
      "epoch": 0.23457033560752455,
      "grad_norm": 11.590909004211426,
      "learning_rate": 7.811946546449847e-07,
      "loss": 0.0694,
      "step": 1487
    },
    {
      "epoch": 0.23472808297511535,
      "grad_norm": 5.491660118103027,
      "learning_rate": 7.810336499758493e-07,
      "loss": 0.0617,
      "step": 1488
    },
    {
      "epoch": 0.23488583034270616,
      "grad_norm": 7.87093448638916,
      "learning_rate": 7.808726453067138e-07,
      "loss": 0.0719,
      "step": 1489
    },
    {
      "epoch": 0.23504357771029696,
      "grad_norm": 4.106072902679443,
      "learning_rate": 7.807116406375784e-07,
      "loss": 0.0308,
      "step": 1490
    },
    {
      "epoch": 0.23520132507788777,
      "grad_norm": 5.372685432434082,
      "learning_rate": 7.80550635968443e-07,
      "loss": 0.0881,
      "step": 1491
    },
    {
      "epoch": 0.23535907244547857,
      "grad_norm": 4.103046894073486,
      "learning_rate": 7.803896312993077e-07,
      "loss": 0.0203,
      "step": 1492
    },
    {
      "epoch": 0.23551681981306938,
      "grad_norm": 5.021068096160889,
      "learning_rate": 7.802286266301722e-07,
      "loss": 0.0348,
      "step": 1493
    },
    {
      "epoch": 0.23567456718066018,
      "grad_norm": 5.834413528442383,
      "learning_rate": 7.800676219610368e-07,
      "loss": 0.0302,
      "step": 1494
    },
    {
      "epoch": 0.235832314548251,
      "grad_norm": 7.424976348876953,
      "learning_rate": 7.799066172919014e-07,
      "loss": 0.0567,
      "step": 1495
    },
    {
      "epoch": 0.2359900619158418,
      "grad_norm": 2.5958831310272217,
      "learning_rate": 7.79745612622766e-07,
      "loss": 0.0213,
      "step": 1496
    },
    {
      "epoch": 0.23614780928343257,
      "grad_norm": 5.730758190155029,
      "learning_rate": 7.795846079536305e-07,
      "loss": 0.0216,
      "step": 1497
    },
    {
      "epoch": 0.23630555665102337,
      "grad_norm": 5.889350891113281,
      "learning_rate": 7.794236032844951e-07,
      "loss": 0.0231,
      "step": 1498
    },
    {
      "epoch": 0.23646330401861418,
      "grad_norm": 3.6521618366241455,
      "learning_rate": 7.792625986153599e-07,
      "loss": 0.0447,
      "step": 1499
    },
    {
      "epoch": 0.23662105138620498,
      "grad_norm": 5.1871161460876465,
      "learning_rate": 7.791015939462245e-07,
      "loss": 0.0608,
      "step": 1500
    },
    {
      "epoch": 0.2367787987537958,
      "grad_norm": 7.352590084075928,
      "learning_rate": 7.789405892770891e-07,
      "loss": 0.0774,
      "step": 1501
    },
    {
      "epoch": 0.2369365461213866,
      "grad_norm": 8.463130950927734,
      "learning_rate": 7.787795846079536e-07,
      "loss": 0.0804,
      "step": 1502
    },
    {
      "epoch": 0.2370942934889774,
      "grad_norm": 4.437898635864258,
      "learning_rate": 7.786185799388182e-07,
      "loss": 0.047,
      "step": 1503
    },
    {
      "epoch": 0.2372520408565682,
      "grad_norm": 4.352744102478027,
      "learning_rate": 7.784575752696828e-07,
      "loss": 0.0437,
      "step": 1504
    },
    {
      "epoch": 0.237409788224159,
      "grad_norm": 4.607551097869873,
      "learning_rate": 7.782965706005474e-07,
      "loss": 0.0632,
      "step": 1505
    },
    {
      "epoch": 0.2375675355917498,
      "grad_norm": 4.5728325843811035,
      "learning_rate": 7.781355659314119e-07,
      "loss": 0.048,
      "step": 1506
    },
    {
      "epoch": 0.23772528295934062,
      "grad_norm": 4.126950740814209,
      "learning_rate": 7.779745612622766e-07,
      "loss": 0.0274,
      "step": 1507
    },
    {
      "epoch": 0.23788303032693142,
      "grad_norm": 7.351589679718018,
      "learning_rate": 7.778135565931412e-07,
      "loss": 0.0453,
      "step": 1508
    },
    {
      "epoch": 0.23804077769452223,
      "grad_norm": 4.392864227294922,
      "learning_rate": 7.776525519240058e-07,
      "loss": 0.098,
      "step": 1509
    },
    {
      "epoch": 0.23819852506211303,
      "grad_norm": 3.9069530963897705,
      "learning_rate": 7.774915472548703e-07,
      "loss": 0.0409,
      "step": 1510
    },
    {
      "epoch": 0.23835627242970384,
      "grad_norm": 6.062881946563721,
      "learning_rate": 7.773305425857349e-07,
      "loss": 0.0518,
      "step": 1511
    },
    {
      "epoch": 0.23851401979729464,
      "grad_norm": 6.978428840637207,
      "learning_rate": 7.771695379165995e-07,
      "loss": 0.1017,
      "step": 1512
    },
    {
      "epoch": 0.23867176716488545,
      "grad_norm": 5.4202656745910645,
      "learning_rate": 7.770085332474641e-07,
      "loss": 0.043,
      "step": 1513
    },
    {
      "epoch": 0.23882951453247625,
      "grad_norm": 2.6532363891601562,
      "learning_rate": 7.768475285783287e-07,
      "loss": 0.0538,
      "step": 1514
    },
    {
      "epoch": 0.23898726190006706,
      "grad_norm": 4.850531101226807,
      "learning_rate": 7.766865239091933e-07,
      "loss": 0.0668,
      "step": 1515
    },
    {
      "epoch": 0.23914500926765786,
      "grad_norm": 5.29673957824707,
      "learning_rate": 7.76525519240058e-07,
      "loss": 0.0392,
      "step": 1516
    },
    {
      "epoch": 0.23930275663524864,
      "grad_norm": 2.9764435291290283,
      "learning_rate": 7.763645145709226e-07,
      "loss": 0.0531,
      "step": 1517
    },
    {
      "epoch": 0.23946050400283944,
      "grad_norm": 13.339690208435059,
      "learning_rate": 7.762035099017871e-07,
      "loss": 0.0657,
      "step": 1518
    },
    {
      "epoch": 0.23961825137043025,
      "grad_norm": 2.644252300262451,
      "learning_rate": 7.760425052326517e-07,
      "loss": 0.0242,
      "step": 1519
    },
    {
      "epoch": 0.23977599873802105,
      "grad_norm": 4.701986312866211,
      "learning_rate": 7.758815005635163e-07,
      "loss": 0.0519,
      "step": 1520
    },
    {
      "epoch": 0.23993374610561186,
      "grad_norm": 3.8381757736206055,
      "learning_rate": 7.757204958943809e-07,
      "loss": 0.0239,
      "step": 1521
    },
    {
      "epoch": 0.24009149347320266,
      "grad_norm": 4.815766334533691,
      "learning_rate": 7.755594912252455e-07,
      "loss": 0.0388,
      "step": 1522
    },
    {
      "epoch": 0.24024924084079347,
      "grad_norm": 9.112265586853027,
      "learning_rate": 7.753984865561101e-07,
      "loss": 0.0802,
      "step": 1523
    },
    {
      "epoch": 0.24040698820838427,
      "grad_norm": 3.9090425968170166,
      "learning_rate": 7.752374818869747e-07,
      "loss": 0.0289,
      "step": 1524
    },
    {
      "epoch": 0.24056473557597507,
      "grad_norm": 5.512587070465088,
      "learning_rate": 7.750764772178393e-07,
      "loss": 0.0351,
      "step": 1525
    },
    {
      "epoch": 0.24072248294356588,
      "grad_norm": 10.256319999694824,
      "learning_rate": 7.749154725487039e-07,
      "loss": 0.0861,
      "step": 1526
    },
    {
      "epoch": 0.24088023031115668,
      "grad_norm": 6.231867790222168,
      "learning_rate": 7.747544678795684e-07,
      "loss": 0.0685,
      "step": 1527
    },
    {
      "epoch": 0.2410379776787475,
      "grad_norm": 6.662831783294678,
      "learning_rate": 7.74593463210433e-07,
      "loss": 0.0743,
      "step": 1528
    },
    {
      "epoch": 0.2411957250463383,
      "grad_norm": 4.1656646728515625,
      "learning_rate": 7.744324585412977e-07,
      "loss": 0.0482,
      "step": 1529
    },
    {
      "epoch": 0.2413534724139291,
      "grad_norm": 4.898168087005615,
      "learning_rate": 7.742714538721623e-07,
      "loss": 0.0809,
      "step": 1530
    },
    {
      "epoch": 0.2415112197815199,
      "grad_norm": 4.342949390411377,
      "learning_rate": 7.741104492030268e-07,
      "loss": 0.0342,
      "step": 1531
    },
    {
      "epoch": 0.2416689671491107,
      "grad_norm": 8.505386352539062,
      "learning_rate": 7.739494445338914e-07,
      "loss": 0.1033,
      "step": 1532
    },
    {
      "epoch": 0.2418267145167015,
      "grad_norm": 3.9101195335388184,
      "learning_rate": 7.73788439864756e-07,
      "loss": 0.0373,
      "step": 1533
    },
    {
      "epoch": 0.24198446188429232,
      "grad_norm": 4.149836540222168,
      "learning_rate": 7.736274351956207e-07,
      "loss": 0.0493,
      "step": 1534
    },
    {
      "epoch": 0.24214220925188312,
      "grad_norm": 3.7258353233337402,
      "learning_rate": 7.734664305264852e-07,
      "loss": 0.0283,
      "step": 1535
    },
    {
      "epoch": 0.2422999566194739,
      "grad_norm": 15.704261779785156,
      "learning_rate": 7.733054258573498e-07,
      "loss": 0.0666,
      "step": 1536
    },
    {
      "epoch": 0.2422999566194739,
      "eval_accuracy": 0.9838110965960908,
      "eval_f1": 0.9838110965960908,
      "eval_loss": 0.05488148331642151,
      "eval_runtime": 4723.7889,
      "eval_samples_per_second": 42.943,
      "eval_steps_per_second": 2.684,
      "step": 1536
    },
    {
      "epoch": 0.2424577039870647,
      "grad_norm": 5.802548408508301,
      "learning_rate": 7.731444211882145e-07,
      "loss": 0.0763,
      "step": 1537
    },
    {
      "epoch": 0.2426154513546555,
      "grad_norm": 8.643044471740723,
      "learning_rate": 7.729834165190791e-07,
      "loss": 0.0463,
      "step": 1538
    },
    {
      "epoch": 0.24277319872224631,
      "grad_norm": 8.600137710571289,
      "learning_rate": 7.728224118499436e-07,
      "loss": 0.0757,
      "step": 1539
    },
    {
      "epoch": 0.24293094608983712,
      "grad_norm": 6.230480670928955,
      "learning_rate": 7.726614071808082e-07,
      "loss": 0.0523,
      "step": 1540
    },
    {
      "epoch": 0.24308869345742792,
      "grad_norm": 2.5582406520843506,
      "learning_rate": 7.725004025116728e-07,
      "loss": 0.0326,
      "step": 1541
    },
    {
      "epoch": 0.24324644082501873,
      "grad_norm": 10.910799980163574,
      "learning_rate": 7.723393978425374e-07,
      "loss": 0.0665,
      "step": 1542
    },
    {
      "epoch": 0.24340418819260953,
      "grad_norm": 6.541328430175781,
      "learning_rate": 7.721783931734019e-07,
      "loss": 0.09,
      "step": 1543
    },
    {
      "epoch": 0.24356193556020034,
      "grad_norm": 4.641464710235596,
      "learning_rate": 7.720173885042666e-07,
      "loss": 0.0548,
      "step": 1544
    },
    {
      "epoch": 0.24371968292779114,
      "grad_norm": 4.99911642074585,
      "learning_rate": 7.718563838351312e-07,
      "loss": 0.0283,
      "step": 1545
    },
    {
      "epoch": 0.24387743029538195,
      "grad_norm": 6.955615520477295,
      "learning_rate": 7.716953791659958e-07,
      "loss": 0.1133,
      "step": 1546
    },
    {
      "epoch": 0.24403517766297275,
      "grad_norm": 5.527281284332275,
      "learning_rate": 7.715343744968604e-07,
      "loss": 0.0799,
      "step": 1547
    },
    {
      "epoch": 0.24419292503056356,
      "grad_norm": 5.46390962600708,
      "learning_rate": 7.713733698277249e-07,
      "loss": 0.0429,
      "step": 1548
    },
    {
      "epoch": 0.24435067239815436,
      "grad_norm": 3.071164608001709,
      "learning_rate": 7.712123651585895e-07,
      "loss": 0.0365,
      "step": 1549
    },
    {
      "epoch": 0.24450841976574517,
      "grad_norm": 5.630229949951172,
      "learning_rate": 7.710513604894541e-07,
      "loss": 0.0669,
      "step": 1550
    },
    {
      "epoch": 0.24466616713333597,
      "grad_norm": 6.220067501068115,
      "learning_rate": 7.708903558203188e-07,
      "loss": 0.0709,
      "step": 1551
    },
    {
      "epoch": 0.24482391450092678,
      "grad_norm": 7.007501125335693,
      "learning_rate": 7.707293511511834e-07,
      "loss": 0.0735,
      "step": 1552
    },
    {
      "epoch": 0.24498166186851758,
      "grad_norm": 6.205754280090332,
      "learning_rate": 7.70568346482048e-07,
      "loss": 0.1477,
      "step": 1553
    },
    {
      "epoch": 0.24513940923610839,
      "grad_norm": 5.855749130249023,
      "learning_rate": 7.704073418129126e-07,
      "loss": 0.0297,
      "step": 1554
    },
    {
      "epoch": 0.24529715660369916,
      "grad_norm": 4.720376014709473,
      "learning_rate": 7.702463371437772e-07,
      "loss": 0.0562,
      "step": 1555
    },
    {
      "epoch": 0.24545490397128997,
      "grad_norm": 4.498953342437744,
      "learning_rate": 7.700853324746417e-07,
      "loss": 0.09,
      "step": 1556
    },
    {
      "epoch": 0.24561265133888077,
      "grad_norm": 3.5153050422668457,
      "learning_rate": 7.699243278055063e-07,
      "loss": 0.0588,
      "step": 1557
    },
    {
      "epoch": 0.24577039870647158,
      "grad_norm": 5.03090238571167,
      "learning_rate": 7.697633231363709e-07,
      "loss": 0.0915,
      "step": 1558
    },
    {
      "epoch": 0.24592814607406238,
      "grad_norm": 5.303913116455078,
      "learning_rate": 7.696023184672356e-07,
      "loss": 0.0934,
      "step": 1559
    },
    {
      "epoch": 0.2460858934416532,
      "grad_norm": 6.614214897155762,
      "learning_rate": 7.694413137981001e-07,
      "loss": 0.0796,
      "step": 1560
    },
    {
      "epoch": 0.246243640809244,
      "grad_norm": 5.710455417633057,
      "learning_rate": 7.692803091289647e-07,
      "loss": 0.0503,
      "step": 1561
    },
    {
      "epoch": 0.2464013881768348,
      "grad_norm": 3.4222452640533447,
      "learning_rate": 7.691193044598293e-07,
      "loss": 0.024,
      "step": 1562
    },
    {
      "epoch": 0.2465591355444256,
      "grad_norm": 3.0048317909240723,
      "learning_rate": 7.689582997906939e-07,
      "loss": 0.0356,
      "step": 1563
    },
    {
      "epoch": 0.2467168829120164,
      "grad_norm": 5.317868232727051,
      "learning_rate": 7.687972951215584e-07,
      "loss": 0.0575,
      "step": 1564
    },
    {
      "epoch": 0.2468746302796072,
      "grad_norm": 6.72121524810791,
      "learning_rate": 7.68636290452423e-07,
      "loss": 0.0526,
      "step": 1565
    },
    {
      "epoch": 0.24703237764719801,
      "grad_norm": 2.8865790367126465,
      "learning_rate": 7.684752857832876e-07,
      "loss": 0.0194,
      "step": 1566
    },
    {
      "epoch": 0.24719012501478882,
      "grad_norm": 6.781136989593506,
      "learning_rate": 7.683142811141523e-07,
      "loss": 0.0424,
      "step": 1567
    },
    {
      "epoch": 0.24734787238237962,
      "grad_norm": 5.859102725982666,
      "learning_rate": 7.681532764450168e-07,
      "loss": 0.0393,
      "step": 1568
    },
    {
      "epoch": 0.24750561974997043,
      "grad_norm": 4.861766338348389,
      "learning_rate": 7.679922717758815e-07,
      "loss": 0.0692,
      "step": 1569
    },
    {
      "epoch": 0.24766336711756123,
      "grad_norm": 5.112473011016846,
      "learning_rate": 7.678312671067461e-07,
      "loss": 0.046,
      "step": 1570
    },
    {
      "epoch": 0.24782111448515204,
      "grad_norm": 4.80613899230957,
      "learning_rate": 7.676702624376107e-07,
      "loss": 0.0207,
      "step": 1571
    },
    {
      "epoch": 0.24797886185274284,
      "grad_norm": 4.593803882598877,
      "learning_rate": 7.675092577684753e-07,
      "loss": 0.025,
      "step": 1572
    },
    {
      "epoch": 0.24813660922033365,
      "grad_norm": 4.530654430389404,
      "learning_rate": 7.673482530993398e-07,
      "loss": 0.0583,
      "step": 1573
    },
    {
      "epoch": 0.24829435658792443,
      "grad_norm": 5.102366924285889,
      "learning_rate": 7.671872484302045e-07,
      "loss": 0.0383,
      "step": 1574
    },
    {
      "epoch": 0.24845210395551523,
      "grad_norm": 7.638766288757324,
      "learning_rate": 7.670262437610691e-07,
      "loss": 0.0549,
      "step": 1575
    },
    {
      "epoch": 0.24860985132310603,
      "grad_norm": 9.579071998596191,
      "learning_rate": 7.668652390919337e-07,
      "loss": 0.0531,
      "step": 1576
    },
    {
      "epoch": 0.24876759869069684,
      "grad_norm": 8.592358589172363,
      "learning_rate": 7.667042344227982e-07,
      "loss": 0.082,
      "step": 1577
    },
    {
      "epoch": 0.24892534605828764,
      "grad_norm": 5.915648937225342,
      "learning_rate": 7.665432297536628e-07,
      "loss": 0.0534,
      "step": 1578
    },
    {
      "epoch": 0.24908309342587845,
      "grad_norm": 7.780115127563477,
      "learning_rate": 7.663822250845274e-07,
      "loss": 0.0693,
      "step": 1579
    },
    {
      "epoch": 0.24924084079346925,
      "grad_norm": 2.9398467540740967,
      "learning_rate": 7.66221220415392e-07,
      "loss": 0.0384,
      "step": 1580
    },
    {
      "epoch": 0.24939858816106006,
      "grad_norm": 6.325009346008301,
      "learning_rate": 7.660602157462565e-07,
      "loss": 0.034,
      "step": 1581
    },
    {
      "epoch": 0.24955633552865086,
      "grad_norm": 4.324703216552734,
      "learning_rate": 7.658992110771212e-07,
      "loss": 0.0769,
      "step": 1582
    },
    {
      "epoch": 0.24971408289624167,
      "grad_norm": 6.651468276977539,
      "learning_rate": 7.657382064079858e-07,
      "loss": 0.0459,
      "step": 1583
    },
    {
      "epoch": 0.24987183026383247,
      "grad_norm": 4.541459083557129,
      "learning_rate": 7.655772017388504e-07,
      "loss": 0.0362,
      "step": 1584
    },
    {
      "epoch": 0.25002957763142325,
      "grad_norm": 6.376233100891113,
      "learning_rate": 7.654161970697149e-07,
      "loss": 0.0735,
      "step": 1585
    },
    {
      "epoch": 0.2501873249990141,
      "grad_norm": 2.1708052158355713,
      "learning_rate": 7.652551924005796e-07,
      "loss": 0.0235,
      "step": 1586
    },
    {
      "epoch": 0.25034507236660486,
      "grad_norm": 2.9557206630706787,
      "learning_rate": 7.650941877314442e-07,
      "loss": 0.027,
      "step": 1587
    },
    {
      "epoch": 0.2505028197341957,
      "grad_norm": 5.037871360778809,
      "learning_rate": 7.649331830623088e-07,
      "loss": 0.0485,
      "step": 1588
    },
    {
      "epoch": 0.25066056710178647,
      "grad_norm": 4.366393566131592,
      "learning_rate": 7.647721783931734e-07,
      "loss": 0.0365,
      "step": 1589
    },
    {
      "epoch": 0.2508183144693773,
      "grad_norm": 2.030135154724121,
      "learning_rate": 7.64611173724038e-07,
      "loss": 0.0278,
      "step": 1590
    },
    {
      "epoch": 0.2509760618369681,
      "grad_norm": 8.954212188720703,
      "learning_rate": 7.644501690549026e-07,
      "loss": 0.1446,
      "step": 1591
    },
    {
      "epoch": 0.2511338092045589,
      "grad_norm": 3.5425732135772705,
      "learning_rate": 7.642891643857672e-07,
      "loss": 0.0451,
      "step": 1592
    },
    {
      "epoch": 0.2512915565721497,
      "grad_norm": 3.917052745819092,
      "learning_rate": 7.641281597166317e-07,
      "loss": 0.0396,
      "step": 1593
    },
    {
      "epoch": 0.2514493039397405,
      "grad_norm": 5.786766529083252,
      "learning_rate": 7.639671550474963e-07,
      "loss": 0.0824,
      "step": 1594
    },
    {
      "epoch": 0.2516070513073313,
      "grad_norm": 4.057224750518799,
      "learning_rate": 7.638061503783609e-07,
      "loss": 0.0378,
      "step": 1595
    },
    {
      "epoch": 0.25176479867492213,
      "grad_norm": 5.296603679656982,
      "learning_rate": 7.636451457092255e-07,
      "loss": 0.0246,
      "step": 1596
    },
    {
      "epoch": 0.2519225460425129,
      "grad_norm": 6.155651569366455,
      "learning_rate": 7.634841410400902e-07,
      "loss": 0.0818,
      "step": 1597
    },
    {
      "epoch": 0.25208029341010374,
      "grad_norm": 4.915137767791748,
      "learning_rate": 7.633231363709547e-07,
      "loss": 0.0715,
      "step": 1598
    },
    {
      "epoch": 0.2522380407776945,
      "grad_norm": 5.413578510284424,
      "learning_rate": 7.631621317018193e-07,
      "loss": 0.0527,
      "step": 1599
    },
    {
      "epoch": 0.25239578814528535,
      "grad_norm": 4.126835346221924,
      "learning_rate": 7.630011270326839e-07,
      "loss": 0.0705,
      "step": 1600
    },
    {
      "epoch": 0.2525535355128761,
      "grad_norm": 6.050610542297363,
      "learning_rate": 7.628401223635485e-07,
      "loss": 0.081,
      "step": 1601
    },
    {
      "epoch": 0.25271128288046696,
      "grad_norm": 4.388615608215332,
      "learning_rate": 7.62679117694413e-07,
      "loss": 0.0301,
      "step": 1602
    },
    {
      "epoch": 0.25286903024805774,
      "grad_norm": 11.63635540008545,
      "learning_rate": 7.625181130252776e-07,
      "loss": 0.0639,
      "step": 1603
    },
    {
      "epoch": 0.2530267776156485,
      "grad_norm": 3.628242254257202,
      "learning_rate": 7.623571083561424e-07,
      "loss": 0.0335,
      "step": 1604
    },
    {
      "epoch": 0.25318452498323935,
      "grad_norm": 10.505038261413574,
      "learning_rate": 7.62196103687007e-07,
      "loss": 0.0713,
      "step": 1605
    },
    {
      "epoch": 0.2533422723508301,
      "grad_norm": 12.409416198730469,
      "learning_rate": 7.620350990178715e-07,
      "loss": 0.1076,
      "step": 1606
    },
    {
      "epoch": 0.25350001971842095,
      "grad_norm": 4.488946914672852,
      "learning_rate": 7.618740943487361e-07,
      "loss": 0.0381,
      "step": 1607
    },
    {
      "epoch": 0.25365776708601173,
      "grad_norm": 10.696093559265137,
      "learning_rate": 7.617130896796007e-07,
      "loss": 0.0654,
      "step": 1608
    },
    {
      "epoch": 0.25381551445360256,
      "grad_norm": 5.161859035491943,
      "learning_rate": 7.615520850104653e-07,
      "loss": 0.0353,
      "step": 1609
    },
    {
      "epoch": 0.25397326182119334,
      "grad_norm": 5.989293098449707,
      "learning_rate": 7.613910803413298e-07,
      "loss": 0.059,
      "step": 1610
    },
    {
      "epoch": 0.2541310091887842,
      "grad_norm": 5.287972927093506,
      "learning_rate": 7.612300756721944e-07,
      "loss": 0.0252,
      "step": 1611
    },
    {
      "epoch": 0.25428875655637495,
      "grad_norm": 3.6004478931427,
      "learning_rate": 7.610690710030591e-07,
      "loss": 0.0318,
      "step": 1612
    },
    {
      "epoch": 0.2544465039239658,
      "grad_norm": 6.214033603668213,
      "learning_rate": 7.609080663339237e-07,
      "loss": 0.0651,
      "step": 1613
    },
    {
      "epoch": 0.25460425129155656,
      "grad_norm": 5.563488006591797,
      "learning_rate": 7.607470616647882e-07,
      "loss": 0.0827,
      "step": 1614
    },
    {
      "epoch": 0.2547619986591474,
      "grad_norm": 7.151546001434326,
      "learning_rate": 7.605860569956528e-07,
      "loss": 0.0592,
      "step": 1615
    },
    {
      "epoch": 0.25491974602673817,
      "grad_norm": 6.595117092132568,
      "learning_rate": 7.604250523265174e-07,
      "loss": 0.0646,
      "step": 1616
    },
    {
      "epoch": 0.255077493394329,
      "grad_norm": 5.908832550048828,
      "learning_rate": 7.60264047657382e-07,
      "loss": 0.0643,
      "step": 1617
    },
    {
      "epoch": 0.2552352407619198,
      "grad_norm": 6.553600788116455,
      "learning_rate": 7.601030429882466e-07,
      "loss": 0.039,
      "step": 1618
    },
    {
      "epoch": 0.2553929881295106,
      "grad_norm": 2.677720069885254,
      "learning_rate": 7.599420383191112e-07,
      "loss": 0.0396,
      "step": 1619
    },
    {
      "epoch": 0.2555507354971014,
      "grad_norm": 6.571342468261719,
      "learning_rate": 7.597810336499758e-07,
      "loss": 0.1106,
      "step": 1620
    },
    {
      "epoch": 0.2557084828646922,
      "grad_norm": 11.148405075073242,
      "learning_rate": 7.596200289808405e-07,
      "loss": 0.0835,
      "step": 1621
    },
    {
      "epoch": 0.255866230232283,
      "grad_norm": 4.706027984619141,
      "learning_rate": 7.594590243117051e-07,
      "loss": 0.036,
      "step": 1622
    },
    {
      "epoch": 0.2560239775998738,
      "grad_norm": 4.21353816986084,
      "learning_rate": 7.592980196425696e-07,
      "loss": 0.0497,
      "step": 1623
    },
    {
      "epoch": 0.2561817249674646,
      "grad_norm": 8.491364479064941,
      "learning_rate": 7.591370149734342e-07,
      "loss": 0.0477,
      "step": 1624
    },
    {
      "epoch": 0.2563394723350554,
      "grad_norm": 4.532919406890869,
      "learning_rate": 7.589760103042988e-07,
      "loss": 0.0794,
      "step": 1625
    },
    {
      "epoch": 0.2564972197026462,
      "grad_norm": 4.545327663421631,
      "learning_rate": 7.588150056351634e-07,
      "loss": 0.0403,
      "step": 1626
    },
    {
      "epoch": 0.256654967070237,
      "grad_norm": 3.378826379776001,
      "learning_rate": 7.58654000966028e-07,
      "loss": 0.0306,
      "step": 1627
    },
    {
      "epoch": 0.2568127144378278,
      "grad_norm": 3.9472615718841553,
      "learning_rate": 7.584929962968926e-07,
      "loss": 0.0343,
      "step": 1628
    },
    {
      "epoch": 0.2569704618054186,
      "grad_norm": 5.402244567871094,
      "learning_rate": 7.583319916277572e-07,
      "loss": 0.0513,
      "step": 1629
    },
    {
      "epoch": 0.25712820917300944,
      "grad_norm": 5.572901725769043,
      "learning_rate": 7.581709869586218e-07,
      "loss": 0.0621,
      "step": 1630
    },
    {
      "epoch": 0.2572859565406002,
      "grad_norm": 4.3812713623046875,
      "learning_rate": 7.580099822894863e-07,
      "loss": 0.032,
      "step": 1631
    },
    {
      "epoch": 0.25744370390819105,
      "grad_norm": 7.747415065765381,
      "learning_rate": 7.578489776203509e-07,
      "loss": 0.0566,
      "step": 1632
    },
    {
      "epoch": 0.2576014512757818,
      "grad_norm": 1.9615048170089722,
      "learning_rate": 7.576879729512155e-07,
      "loss": 0.0216,
      "step": 1633
    },
    {
      "epoch": 0.25775919864337266,
      "grad_norm": 4.1552653312683105,
      "learning_rate": 7.575269682820802e-07,
      "loss": 0.0268,
      "step": 1634
    },
    {
      "epoch": 0.25791694601096343,
      "grad_norm": 5.072568893432617,
      "learning_rate": 7.573659636129447e-07,
      "loss": 0.0688,
      "step": 1635
    },
    {
      "epoch": 0.25807469337855427,
      "grad_norm": 2.2559962272644043,
      "learning_rate": 7.572049589438093e-07,
      "loss": 0.0331,
      "step": 1636
    },
    {
      "epoch": 0.25823244074614504,
      "grad_norm": 4.044931888580322,
      "learning_rate": 7.570439542746739e-07,
      "loss": 0.0391,
      "step": 1637
    },
    {
      "epoch": 0.2583901881137359,
      "grad_norm": 5.895998001098633,
      "learning_rate": 7.568829496055385e-07,
      "loss": 0.0722,
      "step": 1638
    },
    {
      "epoch": 0.25854793548132665,
      "grad_norm": 6.41295051574707,
      "learning_rate": 7.56721944936403e-07,
      "loss": 0.0673,
      "step": 1639
    },
    {
      "epoch": 0.2587056828489175,
      "grad_norm": 3.9313929080963135,
      "learning_rate": 7.565609402672677e-07,
      "loss": 0.0268,
      "step": 1640
    },
    {
      "epoch": 0.25886343021650826,
      "grad_norm": 6.33918571472168,
      "learning_rate": 7.563999355981323e-07,
      "loss": 0.0792,
      "step": 1641
    },
    {
      "epoch": 0.25902117758409904,
      "grad_norm": 5.593765735626221,
      "learning_rate": 7.56238930928997e-07,
      "loss": 0.0371,
      "step": 1642
    },
    {
      "epoch": 0.25917892495168987,
      "grad_norm": 4.316140174865723,
      "learning_rate": 7.560779262598616e-07,
      "loss": 0.0639,
      "step": 1643
    },
    {
      "epoch": 0.25933667231928065,
      "grad_norm": 5.701627731323242,
      "learning_rate": 7.559169215907261e-07,
      "loss": 0.0407,
      "step": 1644
    },
    {
      "epoch": 0.2594944196868715,
      "grad_norm": 4.160895347595215,
      "learning_rate": 7.557559169215907e-07,
      "loss": 0.0504,
      "step": 1645
    },
    {
      "epoch": 0.25965216705446226,
      "grad_norm": 3.8167333602905273,
      "learning_rate": 7.555949122524553e-07,
      "loss": 0.0584,
      "step": 1646
    },
    {
      "epoch": 0.2598099144220531,
      "grad_norm": 6.050884246826172,
      "learning_rate": 7.554339075833199e-07,
      "loss": 0.0618,
      "step": 1647
    },
    {
      "epoch": 0.25996766178964387,
      "grad_norm": 7.4451422691345215,
      "learning_rate": 7.552729029141844e-07,
      "loss": 0.0527,
      "step": 1648
    },
    {
      "epoch": 0.2601254091572347,
      "grad_norm": 2.439089059829712,
      "learning_rate": 7.551118982450491e-07,
      "loss": 0.0596,
      "step": 1649
    },
    {
      "epoch": 0.2602831565248255,
      "grad_norm": 2.6400222778320312,
      "learning_rate": 7.549508935759137e-07,
      "loss": 0.0165,
      "step": 1650
    },
    {
      "epoch": 0.2604409038924163,
      "grad_norm": 4.077780723571777,
      "learning_rate": 7.547898889067783e-07,
      "loss": 0.0783,
      "step": 1651
    },
    {
      "epoch": 0.2605986512600071,
      "grad_norm": 5.45765495300293,
      "learning_rate": 7.546288842376428e-07,
      "loss": 0.0693,
      "step": 1652
    },
    {
      "epoch": 0.2607563986275979,
      "grad_norm": 5.8802289962768555,
      "learning_rate": 7.544678795685074e-07,
      "loss": 0.0947,
      "step": 1653
    },
    {
      "epoch": 0.2609141459951887,
      "grad_norm": 11.156391143798828,
      "learning_rate": 7.54306874899372e-07,
      "loss": 0.0526,
      "step": 1654
    },
    {
      "epoch": 0.26107189336277953,
      "grad_norm": 4.166376113891602,
      "learning_rate": 7.541458702302366e-07,
      "loss": 0.0389,
      "step": 1655
    },
    {
      "epoch": 0.2612296407303703,
      "grad_norm": 7.223203659057617,
      "learning_rate": 7.539848655611011e-07,
      "loss": 0.0954,
      "step": 1656
    },
    {
      "epoch": 0.26138738809796114,
      "grad_norm": 8.58761215209961,
      "learning_rate": 7.538238608919659e-07,
      "loss": 0.1029,
      "step": 1657
    },
    {
      "epoch": 0.2615451354655519,
      "grad_norm": 3.298081159591675,
      "learning_rate": 7.536628562228305e-07,
      "loss": 0.0364,
      "step": 1658
    },
    {
      "epoch": 0.26170288283314275,
      "grad_norm": 5.4777350425720215,
      "learning_rate": 7.535018515536951e-07,
      "loss": 0.0827,
      "step": 1659
    },
    {
      "epoch": 0.2618606302007335,
      "grad_norm": 5.848228931427002,
      "learning_rate": 7.533408468845596e-07,
      "loss": 0.0533,
      "step": 1660
    },
    {
      "epoch": 0.2620183775683243,
      "grad_norm": 7.866878986358643,
      "learning_rate": 7.531798422154242e-07,
      "loss": 0.0796,
      "step": 1661
    },
    {
      "epoch": 0.26217612493591513,
      "grad_norm": 2.7035603523254395,
      "learning_rate": 7.530188375462888e-07,
      "loss": 0.0213,
      "step": 1662
    },
    {
      "epoch": 0.2623338723035059,
      "grad_norm": 5.373941898345947,
      "learning_rate": 7.528578328771534e-07,
      "loss": 0.0618,
      "step": 1663
    },
    {
      "epoch": 0.26249161967109674,
      "grad_norm": 8.629209518432617,
      "learning_rate": 7.526968282080181e-07,
      "loss": 0.0901,
      "step": 1664
    },
    {
      "epoch": 0.2626493670386875,
      "grad_norm": 5.4822001457214355,
      "learning_rate": 7.525358235388826e-07,
      "loss": 0.0914,
      "step": 1665
    },
    {
      "epoch": 0.26280711440627835,
      "grad_norm": 4.280784606933594,
      "learning_rate": 7.523748188697472e-07,
      "loss": 0.0918,
      "step": 1666
    },
    {
      "epoch": 0.26296486177386913,
      "grad_norm": 3.0485968589782715,
      "learning_rate": 7.522138142006118e-07,
      "loss": 0.0525,
      "step": 1667
    },
    {
      "epoch": 0.26312260914145996,
      "grad_norm": 7.273502826690674,
      "learning_rate": 7.520528095314764e-07,
      "loss": 0.0763,
      "step": 1668
    },
    {
      "epoch": 0.26328035650905074,
      "grad_norm": 6.036385536193848,
      "learning_rate": 7.518918048623409e-07,
      "loss": 0.0641,
      "step": 1669
    },
    {
      "epoch": 0.26343810387664157,
      "grad_norm": 5.04290771484375,
      "learning_rate": 7.517308001932055e-07,
      "loss": 0.1048,
      "step": 1670
    },
    {
      "epoch": 0.26359585124423235,
      "grad_norm": 5.335455417633057,
      "learning_rate": 7.515697955240701e-07,
      "loss": 0.0769,
      "step": 1671
    },
    {
      "epoch": 0.2637535986118232,
      "grad_norm": 5.425839424133301,
      "learning_rate": 7.514087908549348e-07,
      "loss": 0.0333,
      "step": 1672
    },
    {
      "epoch": 0.26391134597941396,
      "grad_norm": 5.6087260246276855,
      "learning_rate": 7.512477861857993e-07,
      "loss": 0.061,
      "step": 1673
    },
    {
      "epoch": 0.2640690933470048,
      "grad_norm": 8.458930969238281,
      "learning_rate": 7.51086781516664e-07,
      "loss": 0.042,
      "step": 1674
    },
    {
      "epoch": 0.26422684071459557,
      "grad_norm": 9.07277774810791,
      "learning_rate": 7.509257768475286e-07,
      "loss": 0.0753,
      "step": 1675
    },
    {
      "epoch": 0.2643845880821864,
      "grad_norm": 5.768187522888184,
      "learning_rate": 7.507647721783932e-07,
      "loss": 0.026,
      "step": 1676
    },
    {
      "epoch": 0.2645423354497772,
      "grad_norm": 4.110996723175049,
      "learning_rate": 7.506037675092577e-07,
      "loss": 0.0463,
      "step": 1677
    },
    {
      "epoch": 0.264700082817368,
      "grad_norm": 2.877336025238037,
      "learning_rate": 7.504427628401223e-07,
      "loss": 0.0265,
      "step": 1678
    },
    {
      "epoch": 0.2648578301849588,
      "grad_norm": 7.911360740661621,
      "learning_rate": 7.50281758170987e-07,
      "loss": 0.0371,
      "step": 1679
    },
    {
      "epoch": 0.2650155775525496,
      "grad_norm": 6.589733600616455,
      "learning_rate": 7.501207535018516e-07,
      "loss": 0.0654,
      "step": 1680
    },
    {
      "epoch": 0.2651733249201404,
      "grad_norm": 5.051817893981934,
      "learning_rate": 7.499597488327161e-07,
      "loss": 0.0715,
      "step": 1681
    },
    {
      "epoch": 0.2653310722877312,
      "grad_norm": 5.483920574188232,
      "learning_rate": 7.497987441635807e-07,
      "loss": 0.0436,
      "step": 1682
    },
    {
      "epoch": 0.265488819655322,
      "grad_norm": 4.896912097930908,
      "learning_rate": 7.496377394944453e-07,
      "loss": 0.0575,
      "step": 1683
    },
    {
      "epoch": 0.2656465670229128,
      "grad_norm": 7.662925720214844,
      "learning_rate": 7.494767348253099e-07,
      "loss": 0.0479,
      "step": 1684
    },
    {
      "epoch": 0.2658043143905036,
      "grad_norm": 4.1565260887146,
      "learning_rate": 7.493157301561744e-07,
      "loss": 0.0353,
      "step": 1685
    },
    {
      "epoch": 0.2659620617580944,
      "grad_norm": 5.179498195648193,
      "learning_rate": 7.49154725487039e-07,
      "loss": 0.034,
      "step": 1686
    },
    {
      "epoch": 0.2661198091256852,
      "grad_norm": 5.550738334655762,
      "learning_rate": 7.489937208179037e-07,
      "loss": 0.0402,
      "step": 1687
    },
    {
      "epoch": 0.266277556493276,
      "grad_norm": 4.709192752838135,
      "learning_rate": 7.488327161487683e-07,
      "loss": 0.0757,
      "step": 1688
    },
    {
      "epoch": 0.26643530386086683,
      "grad_norm": 5.049219131469727,
      "learning_rate": 7.486717114796329e-07,
      "loss": 0.0789,
      "step": 1689
    },
    {
      "epoch": 0.2665930512284576,
      "grad_norm": 3.1105828285217285,
      "learning_rate": 7.485107068104974e-07,
      "loss": 0.0318,
      "step": 1690
    },
    {
      "epoch": 0.26675079859604844,
      "grad_norm": 4.199319362640381,
      "learning_rate": 7.48349702141362e-07,
      "loss": 0.0842,
      "step": 1691
    },
    {
      "epoch": 0.2669085459636392,
      "grad_norm": 4.327901840209961,
      "learning_rate": 7.481886974722267e-07,
      "loss": 0.045,
      "step": 1692
    },
    {
      "epoch": 0.26706629333123005,
      "grad_norm": 11.174415588378906,
      "learning_rate": 7.480276928030913e-07,
      "loss": 0.1128,
      "step": 1693
    },
    {
      "epoch": 0.26722404069882083,
      "grad_norm": 5.058723449707031,
      "learning_rate": 7.478666881339558e-07,
      "loss": 0.071,
      "step": 1694
    },
    {
      "epoch": 0.26738178806641166,
      "grad_norm": 4.530946731567383,
      "learning_rate": 7.477056834648205e-07,
      "loss": 0.0458,
      "step": 1695
    },
    {
      "epoch": 0.26753953543400244,
      "grad_norm": 8.258978843688965,
      "learning_rate": 7.475446787956851e-07,
      "loss": 0.0286,
      "step": 1696
    },
    {
      "epoch": 0.2676972828015933,
      "grad_norm": 2.9627537727355957,
      "learning_rate": 7.473836741265497e-07,
      "loss": 0.0275,
      "step": 1697
    },
    {
      "epoch": 0.26785503016918405,
      "grad_norm": 6.273197174072266,
      "learning_rate": 7.472226694574142e-07,
      "loss": 0.0506,
      "step": 1698
    },
    {
      "epoch": 0.2680127775367749,
      "grad_norm": 5.220770835876465,
      "learning_rate": 7.470616647882788e-07,
      "loss": 0.0442,
      "step": 1699
    },
    {
      "epoch": 0.26817052490436566,
      "grad_norm": 8.129053115844727,
      "learning_rate": 7.469006601191434e-07,
      "loss": 0.0536,
      "step": 1700
    },
    {
      "epoch": 0.26832827227195644,
      "grad_norm": 5.066961765289307,
      "learning_rate": 7.46739655450008e-07,
      "loss": 0.0518,
      "step": 1701
    },
    {
      "epoch": 0.26848601963954727,
      "grad_norm": 5.97332763671875,
      "learning_rate": 7.465786507808726e-07,
      "loss": 0.112,
      "step": 1702
    },
    {
      "epoch": 0.26864376700713805,
      "grad_norm": 4.293923854827881,
      "learning_rate": 7.464176461117372e-07,
      "loss": 0.0402,
      "step": 1703
    },
    {
      "epoch": 0.2688015143747289,
      "grad_norm": 4.722505569458008,
      "learning_rate": 7.462566414426018e-07,
      "loss": 0.0439,
      "step": 1704
    },
    {
      "epoch": 0.26895926174231966,
      "grad_norm": 4.975640296936035,
      "learning_rate": 7.460956367734664e-07,
      "loss": 0.0282,
      "step": 1705
    },
    {
      "epoch": 0.2691170091099105,
      "grad_norm": 19.83115005493164,
      "learning_rate": 7.459346321043309e-07,
      "loss": 0.0639,
      "step": 1706
    },
    {
      "epoch": 0.26927475647750126,
      "grad_norm": 7.357720851898193,
      "learning_rate": 7.457736274351955e-07,
      "loss": 0.0362,
      "step": 1707
    },
    {
      "epoch": 0.2694325038450921,
      "grad_norm": 10.837241172790527,
      "learning_rate": 7.456126227660601e-07,
      "loss": 0.0884,
      "step": 1708
    },
    {
      "epoch": 0.2695902512126829,
      "grad_norm": 3.5052623748779297,
      "learning_rate": 7.454516180969249e-07,
      "loss": 0.0322,
      "step": 1709
    },
    {
      "epoch": 0.2697479985802737,
      "grad_norm": 10.239457130432129,
      "learning_rate": 7.452906134277895e-07,
      "loss": 0.0449,
      "step": 1710
    },
    {
      "epoch": 0.2699057459478645,
      "grad_norm": 4.1607985496521,
      "learning_rate": 7.45129608758654e-07,
      "loss": 0.0814,
      "step": 1711
    },
    {
      "epoch": 0.2700634933154553,
      "grad_norm": 4.808945178985596,
      "learning_rate": 7.449686040895186e-07,
      "loss": 0.0529,
      "step": 1712
    },
    {
      "epoch": 0.2702212406830461,
      "grad_norm": 3.835298776626587,
      "learning_rate": 7.448075994203832e-07,
      "loss": 0.0599,
      "step": 1713
    },
    {
      "epoch": 0.2703789880506369,
      "grad_norm": 8.88935375213623,
      "learning_rate": 7.446465947512478e-07,
      "loss": 0.0413,
      "step": 1714
    },
    {
      "epoch": 0.2705367354182277,
      "grad_norm": 16.979846954345703,
      "learning_rate": 7.444855900821123e-07,
      "loss": 0.073,
      "step": 1715
    },
    {
      "epoch": 0.27069448278581854,
      "grad_norm": 4.93610954284668,
      "learning_rate": 7.443245854129769e-07,
      "loss": 0.1082,
      "step": 1716
    },
    {
      "epoch": 0.2708522301534093,
      "grad_norm": 5.720614910125732,
      "learning_rate": 7.441635807438416e-07,
      "loss": 0.063,
      "step": 1717
    },
    {
      "epoch": 0.27100997752100014,
      "grad_norm": 6.2367329597473145,
      "learning_rate": 7.440025760747062e-07,
      "loss": 0.0386,
      "step": 1718
    },
    {
      "epoch": 0.2711677248885909,
      "grad_norm": 6.605106830596924,
      "learning_rate": 7.438415714055707e-07,
      "loss": 0.0822,
      "step": 1719
    },
    {
      "epoch": 0.2713254722561817,
      "grad_norm": 11.338903427124023,
      "learning_rate": 7.436805667364353e-07,
      "loss": 0.0729,
      "step": 1720
    },
    {
      "epoch": 0.27148321962377253,
      "grad_norm": 8.502854347229004,
      "learning_rate": 7.435195620672999e-07,
      "loss": 0.0316,
      "step": 1721
    },
    {
      "epoch": 0.2716409669913633,
      "grad_norm": 5.40214729309082,
      "learning_rate": 7.433585573981645e-07,
      "loss": 0.0399,
      "step": 1722
    },
    {
      "epoch": 0.27179871435895414,
      "grad_norm": 5.6454548835754395,
      "learning_rate": 7.43197552729029e-07,
      "loss": 0.0589,
      "step": 1723
    },
    {
      "epoch": 0.2719564617265449,
      "grad_norm": 3.156519651412964,
      "learning_rate": 7.430365480598937e-07,
      "loss": 0.0531,
      "step": 1724
    },
    {
      "epoch": 0.27211420909413575,
      "grad_norm": 4.796392917633057,
      "learning_rate": 7.428755433907583e-07,
      "loss": 0.0282,
      "step": 1725
    },
    {
      "epoch": 0.2722719564617265,
      "grad_norm": 4.910034656524658,
      "learning_rate": 7.42714538721623e-07,
      "loss": 0.0437,
      "step": 1726
    },
    {
      "epoch": 0.27242970382931736,
      "grad_norm": 7.996124267578125,
      "learning_rate": 7.425535340524875e-07,
      "loss": 0.082,
      "step": 1727
    },
    {
      "epoch": 0.27258745119690814,
      "grad_norm": 9.234084129333496,
      "learning_rate": 7.423925293833521e-07,
      "loss": 0.0594,
      "step": 1728
    },
    {
      "epoch": 0.27274519856449897,
      "grad_norm": 7.054522514343262,
      "learning_rate": 7.422315247142167e-07,
      "loss": 0.0441,
      "step": 1729
    },
    {
      "epoch": 0.27290294593208975,
      "grad_norm": 5.047627925872803,
      "learning_rate": 7.420705200450813e-07,
      "loss": 0.0526,
      "step": 1730
    },
    {
      "epoch": 0.2730606932996806,
      "grad_norm": 5.253188610076904,
      "learning_rate": 7.419095153759458e-07,
      "loss": 0.0338,
      "step": 1731
    },
    {
      "epoch": 0.27321844066727136,
      "grad_norm": 8.9965238571167,
      "learning_rate": 7.417485107068105e-07,
      "loss": 0.0817,
      "step": 1732
    },
    {
      "epoch": 0.2733761880348622,
      "grad_norm": 3.8212831020355225,
      "learning_rate": 7.415875060376751e-07,
      "loss": 0.0258,
      "step": 1733
    },
    {
      "epoch": 0.27353393540245297,
      "grad_norm": 1.7131558656692505,
      "learning_rate": 7.414265013685397e-07,
      "loss": 0.0187,
      "step": 1734
    },
    {
      "epoch": 0.2736916827700438,
      "grad_norm": 2.771113634109497,
      "learning_rate": 7.412654966994043e-07,
      "loss": 0.032,
      "step": 1735
    },
    {
      "epoch": 0.2738494301376346,
      "grad_norm": 3.408644199371338,
      "learning_rate": 7.411044920302688e-07,
      "loss": 0.0673,
      "step": 1736
    },
    {
      "epoch": 0.2740071775052254,
      "grad_norm": 8.400445938110352,
      "learning_rate": 7.409434873611334e-07,
      "loss": 0.0787,
      "step": 1737
    },
    {
      "epoch": 0.2741649248728162,
      "grad_norm": 6.512462615966797,
      "learning_rate": 7.40782482691998e-07,
      "loss": 0.0315,
      "step": 1738
    },
    {
      "epoch": 0.27432267224040696,
      "grad_norm": 6.375429153442383,
      "learning_rate": 7.406214780228627e-07,
      "loss": 0.1009,
      "step": 1739
    },
    {
      "epoch": 0.2744804196079978,
      "grad_norm": 3.734159231185913,
      "learning_rate": 7.404604733537272e-07,
      "loss": 0.0553,
      "step": 1740
    },
    {
      "epoch": 0.27463816697558857,
      "grad_norm": 2.839085340499878,
      "learning_rate": 7.402994686845918e-07,
      "loss": 0.0152,
      "step": 1741
    },
    {
      "epoch": 0.2747959143431794,
      "grad_norm": 7.769985675811768,
      "learning_rate": 7.401384640154564e-07,
      "loss": 0.0598,
      "step": 1742
    },
    {
      "epoch": 0.2749536617107702,
      "grad_norm": 9.466279029846191,
      "learning_rate": 7.39977459346321e-07,
      "loss": 0.0468,
      "step": 1743
    },
    {
      "epoch": 0.275111409078361,
      "grad_norm": 4.23649263381958,
      "learning_rate": 7.398164546771855e-07,
      "loss": 0.0529,
      "step": 1744
    },
    {
      "epoch": 0.2752691564459518,
      "grad_norm": 2.059992790222168,
      "learning_rate": 7.396554500080502e-07,
      "loss": 0.0523,
      "step": 1745
    },
    {
      "epoch": 0.2754269038135426,
      "grad_norm": 8.96181869506836,
      "learning_rate": 7.394944453389148e-07,
      "loss": 0.0925,
      "step": 1746
    },
    {
      "epoch": 0.2755846511811334,
      "grad_norm": 4.507099628448486,
      "learning_rate": 7.393334406697795e-07,
      "loss": 0.0548,
      "step": 1747
    },
    {
      "epoch": 0.27574239854872423,
      "grad_norm": 5.502170562744141,
      "learning_rate": 7.39172436000644e-07,
      "loss": 0.0999,
      "step": 1748
    },
    {
      "epoch": 0.275900145916315,
      "grad_norm": 3.1486527919769287,
      "learning_rate": 7.390114313315086e-07,
      "loss": 0.0366,
      "step": 1749
    },
    {
      "epoch": 0.27605789328390584,
      "grad_norm": 7.729282379150391,
      "learning_rate": 7.388504266623732e-07,
      "loss": 0.0846,
      "step": 1750
    },
    {
      "epoch": 0.2762156406514966,
      "grad_norm": 5.723428249359131,
      "learning_rate": 7.386894219932378e-07,
      "loss": 0.0655,
      "step": 1751
    },
    {
      "epoch": 0.27637338801908745,
      "grad_norm": 9.416695594787598,
      "learning_rate": 7.385284173241023e-07,
      "loss": 0.0954,
      "step": 1752
    },
    {
      "epoch": 0.27653113538667823,
      "grad_norm": 6.169697284698486,
      "learning_rate": 7.383674126549669e-07,
      "loss": 0.0746,
      "step": 1753
    },
    {
      "epoch": 0.27668888275426906,
      "grad_norm": 4.9604973793029785,
      "learning_rate": 7.382064079858316e-07,
      "loss": 0.0756,
      "step": 1754
    },
    {
      "epoch": 0.27684663012185984,
      "grad_norm": 6.345669269561768,
      "learning_rate": 7.380454033166962e-07,
      "loss": 0.0553,
      "step": 1755
    },
    {
      "epoch": 0.27700437748945067,
      "grad_norm": 6.011808395385742,
      "learning_rate": 7.378843986475607e-07,
      "loss": 0.0634,
      "step": 1756
    },
    {
      "epoch": 0.27716212485704145,
      "grad_norm": 9.210107803344727,
      "learning_rate": 7.377233939784253e-07,
      "loss": 0.0733,
      "step": 1757
    },
    {
      "epoch": 0.2773198722246322,
      "grad_norm": 5.413785934448242,
      "learning_rate": 7.375623893092899e-07,
      "loss": 0.031,
      "step": 1758
    },
    {
      "epoch": 0.27747761959222306,
      "grad_norm": 5.12843132019043,
      "learning_rate": 7.374013846401545e-07,
      "loss": 0.082,
      "step": 1759
    },
    {
      "epoch": 0.27763536695981383,
      "grad_norm": 3.3309543132781982,
      "learning_rate": 7.372403799710191e-07,
      "loss": 0.0643,
      "step": 1760
    },
    {
      "epoch": 0.27779311432740467,
      "grad_norm": 6.6534953117370605,
      "learning_rate": 7.370793753018836e-07,
      "loss": 0.085,
      "step": 1761
    },
    {
      "epoch": 0.27795086169499544,
      "grad_norm": 7.947381496429443,
      "learning_rate": 7.369183706327484e-07,
      "loss": 0.086,
      "step": 1762
    },
    {
      "epoch": 0.2781086090625863,
      "grad_norm": 4.031906604766846,
      "learning_rate": 7.36757365963613e-07,
      "loss": 0.0528,
      "step": 1763
    },
    {
      "epoch": 0.27826635643017705,
      "grad_norm": 4.700561046600342,
      "learning_rate": 7.365963612944776e-07,
      "loss": 0.0422,
      "step": 1764
    },
    {
      "epoch": 0.2784241037977679,
      "grad_norm": 4.3197021484375,
      "learning_rate": 7.364353566253421e-07,
      "loss": 0.0754,
      "step": 1765
    },
    {
      "epoch": 0.27858185116535866,
      "grad_norm": 4.787306308746338,
      "learning_rate": 7.362743519562067e-07,
      "loss": 0.0489,
      "step": 1766
    },
    {
      "epoch": 0.2787395985329495,
      "grad_norm": 11.220168113708496,
      "learning_rate": 7.361133472870713e-07,
      "loss": 0.1168,
      "step": 1767
    },
    {
      "epoch": 0.27889734590054027,
      "grad_norm": 9.612006187438965,
      "learning_rate": 7.359523426179359e-07,
      "loss": 0.0759,
      "step": 1768
    },
    {
      "epoch": 0.2790550932681311,
      "grad_norm": 3.2283174991607666,
      "learning_rate": 7.357913379488004e-07,
      "loss": 0.0286,
      "step": 1769
    },
    {
      "epoch": 0.2792128406357219,
      "grad_norm": 1.8329448699951172,
      "learning_rate": 7.356303332796651e-07,
      "loss": 0.0155,
      "step": 1770
    },
    {
      "epoch": 0.2793705880033127,
      "grad_norm": 6.679264545440674,
      "learning_rate": 7.354693286105297e-07,
      "loss": 0.03,
      "step": 1771
    },
    {
      "epoch": 0.2795283353709035,
      "grad_norm": 6.550128936767578,
      "learning_rate": 7.353083239413943e-07,
      "loss": 0.0833,
      "step": 1772
    },
    {
      "epoch": 0.2796860827384943,
      "grad_norm": 6.05768346786499,
      "learning_rate": 7.351473192722588e-07,
      "loss": 0.0443,
      "step": 1773
    },
    {
      "epoch": 0.2798438301060851,
      "grad_norm": 6.712532043457031,
      "learning_rate": 7.349863146031234e-07,
      "loss": 0.0761,
      "step": 1774
    },
    {
      "epoch": 0.28000157747367593,
      "grad_norm": 3.4038772583007812,
      "learning_rate": 7.34825309933988e-07,
      "loss": 0.018,
      "step": 1775
    },
    {
      "epoch": 0.2801593248412667,
      "grad_norm": 5.155402183532715,
      "learning_rate": 7.346643052648526e-07,
      "loss": 0.041,
      "step": 1776
    },
    {
      "epoch": 0.2803170722088575,
      "grad_norm": 5.3723530769348145,
      "learning_rate": 7.345033005957172e-07,
      "loss": 0.0416,
      "step": 1777
    },
    {
      "epoch": 0.2804748195764483,
      "grad_norm": 6.428321361541748,
      "learning_rate": 7.343422959265818e-07,
      "loss": 0.0798,
      "step": 1778
    },
    {
      "epoch": 0.2806325669440391,
      "grad_norm": 4.262994766235352,
      "learning_rate": 7.341812912574465e-07,
      "loss": 0.0375,
      "step": 1779
    },
    {
      "epoch": 0.28079031431162993,
      "grad_norm": 5.236276626586914,
      "learning_rate": 7.340202865883111e-07,
      "loss": 0.0359,
      "step": 1780
    },
    {
      "epoch": 0.2809480616792207,
      "grad_norm": 4.757218837738037,
      "learning_rate": 7.338592819191757e-07,
      "loss": 0.0331,
      "step": 1781
    },
    {
      "epoch": 0.28110580904681154,
      "grad_norm": 4.1465678215026855,
      "learning_rate": 7.336982772500402e-07,
      "loss": 0.0346,
      "step": 1782
    },
    {
      "epoch": 0.2812635564144023,
      "grad_norm": 3.4765822887420654,
      "learning_rate": 7.335372725809048e-07,
      "loss": 0.0375,
      "step": 1783
    },
    {
      "epoch": 0.28142130378199315,
      "grad_norm": 6.680180549621582,
      "learning_rate": 7.333762679117695e-07,
      "loss": 0.0249,
      "step": 1784
    },
    {
      "epoch": 0.2815790511495839,
      "grad_norm": 6.115419864654541,
      "learning_rate": 7.332152632426341e-07,
      "loss": 0.0497,
      "step": 1785
    },
    {
      "epoch": 0.28173679851717476,
      "grad_norm": 11.377446174621582,
      "learning_rate": 7.330542585734986e-07,
      "loss": 0.0664,
      "step": 1786
    },
    {
      "epoch": 0.28189454588476553,
      "grad_norm": 4.166162490844727,
      "learning_rate": 7.328932539043632e-07,
      "loss": 0.0746,
      "step": 1787
    },
    {
      "epoch": 0.28205229325235637,
      "grad_norm": 3.433807849884033,
      "learning_rate": 7.327322492352278e-07,
      "loss": 0.0693,
      "step": 1788
    },
    {
      "epoch": 0.28221004061994714,
      "grad_norm": 4.4710693359375,
      "learning_rate": 7.325712445660924e-07,
      "loss": 0.0485,
      "step": 1789
    },
    {
      "epoch": 0.282367787987538,
      "grad_norm": 4.173577785491943,
      "learning_rate": 7.324102398969569e-07,
      "loss": 0.0441,
      "step": 1790
    },
    {
      "epoch": 0.28252553535512875,
      "grad_norm": 7.301423072814941,
      "learning_rate": 7.322492352278215e-07,
      "loss": 0.0499,
      "step": 1791
    },
    {
      "epoch": 0.2826832827227196,
      "grad_norm": 8.932049751281738,
      "learning_rate": 7.320882305586862e-07,
      "loss": 0.0746,
      "step": 1792
    },
    {
      "epoch": 0.28284103009031036,
      "grad_norm": 6.043249130249023,
      "learning_rate": 7.319272258895508e-07,
      "loss": 0.0605,
      "step": 1793
    },
    {
      "epoch": 0.2829987774579012,
      "grad_norm": 4.231344223022461,
      "learning_rate": 7.317662212204153e-07,
      "loss": 0.0488,
      "step": 1794
    },
    {
      "epoch": 0.283156524825492,
      "grad_norm": 8.559317588806152,
      "learning_rate": 7.316052165512799e-07,
      "loss": 0.1268,
      "step": 1795
    },
    {
      "epoch": 0.2833142721930828,
      "grad_norm": 11.188226699829102,
      "learning_rate": 7.314442118821445e-07,
      "loss": 0.0812,
      "step": 1796
    },
    {
      "epoch": 0.2834720195606736,
      "grad_norm": 5.2370500564575195,
      "learning_rate": 7.312832072130092e-07,
      "loss": 0.0689,
      "step": 1797
    },
    {
      "epoch": 0.28362976692826436,
      "grad_norm": 2.636385679244995,
      "learning_rate": 7.311222025438737e-07,
      "loss": 0.0198,
      "step": 1798
    },
    {
      "epoch": 0.2837875142958552,
      "grad_norm": 7.2173895835876465,
      "learning_rate": 7.309611978747383e-07,
      "loss": 0.0523,
      "step": 1799
    },
    {
      "epoch": 0.28394526166344597,
      "grad_norm": 4.4801411628723145,
      "learning_rate": 7.30800193205603e-07,
      "loss": 0.0791,
      "step": 1800
    },
    {
      "epoch": 0.2841030090310368,
      "grad_norm": 2.042889356613159,
      "learning_rate": 7.306391885364676e-07,
      "loss": 0.0115,
      "step": 1801
    },
    {
      "epoch": 0.2842607563986276,
      "grad_norm": 6.690401077270508,
      "learning_rate": 7.304781838673321e-07,
      "loss": 0.0838,
      "step": 1802
    },
    {
      "epoch": 0.2844185037662184,
      "grad_norm": 3.1295642852783203,
      "learning_rate": 7.303171791981967e-07,
      "loss": 0.0304,
      "step": 1803
    },
    {
      "epoch": 0.2845762511338092,
      "grad_norm": 6.915700912475586,
      "learning_rate": 7.301561745290613e-07,
      "loss": 0.0409,
      "step": 1804
    },
    {
      "epoch": 0.2847339985014,
      "grad_norm": 4.392620086669922,
      "learning_rate": 7.299951698599259e-07,
      "loss": 0.114,
      "step": 1805
    },
    {
      "epoch": 0.2848917458689908,
      "grad_norm": 7.388136863708496,
      "learning_rate": 7.298341651907905e-07,
      "loss": 0.0718,
      "step": 1806
    },
    {
      "epoch": 0.28504949323658163,
      "grad_norm": 10.599838256835938,
      "learning_rate": 7.296731605216551e-07,
      "loss": 0.0849,
      "step": 1807
    },
    {
      "epoch": 0.2852072406041724,
      "grad_norm": 4.139700412750244,
      "learning_rate": 7.295121558525197e-07,
      "loss": 0.0318,
      "step": 1808
    },
    {
      "epoch": 0.28536498797176324,
      "grad_norm": 5.016416072845459,
      "learning_rate": 7.293511511833843e-07,
      "loss": 0.0412,
      "step": 1809
    },
    {
      "epoch": 0.285522735339354,
      "grad_norm": 11.005081176757812,
      "learning_rate": 7.291901465142489e-07,
      "loss": 0.0772,
      "step": 1810
    },
    {
      "epoch": 0.28568048270694485,
      "grad_norm": 3.681155204772949,
      "learning_rate": 7.290291418451134e-07,
      "loss": 0.0408,
      "step": 1811
    },
    {
      "epoch": 0.2858382300745356,
      "grad_norm": 6.3219146728515625,
      "learning_rate": 7.28868137175978e-07,
      "loss": 0.0363,
      "step": 1812
    },
    {
      "epoch": 0.28599597744212646,
      "grad_norm": 6.1140522956848145,
      "learning_rate": 7.287071325068426e-07,
      "loss": 0.0821,
      "step": 1813
    },
    {
      "epoch": 0.28615372480971724,
      "grad_norm": 8.338149070739746,
      "learning_rate": 7.285461278377074e-07,
      "loss": 0.0644,
      "step": 1814
    },
    {
      "epoch": 0.28631147217730807,
      "grad_norm": 8.887350082397461,
      "learning_rate": 7.283851231685719e-07,
      "loss": 0.0998,
      "step": 1815
    },
    {
      "epoch": 0.28646921954489885,
      "grad_norm": 3.9195213317871094,
      "learning_rate": 7.282241184994365e-07,
      "loss": 0.0464,
      "step": 1816
    },
    {
      "epoch": 0.2866269669124896,
      "grad_norm": 6.6678996086120605,
      "learning_rate": 7.280631138303011e-07,
      "loss": 0.0583,
      "step": 1817
    },
    {
      "epoch": 0.28678471428008045,
      "grad_norm": 6.425255298614502,
      "learning_rate": 7.279021091611657e-07,
      "loss": 0.0823,
      "step": 1818
    },
    {
      "epoch": 0.28694246164767123,
      "grad_norm": 3.6952288150787354,
      "learning_rate": 7.277411044920302e-07,
      "loss": 0.0389,
      "step": 1819
    },
    {
      "epoch": 0.28710020901526206,
      "grad_norm": 10.184599876403809,
      "learning_rate": 7.275800998228948e-07,
      "loss": 0.0359,
      "step": 1820
    },
    {
      "epoch": 0.28725795638285284,
      "grad_norm": 4.51015567779541,
      "learning_rate": 7.274190951537594e-07,
      "loss": 0.0753,
      "step": 1821
    },
    {
      "epoch": 0.2874157037504437,
      "grad_norm": 20.935108184814453,
      "learning_rate": 7.272580904846241e-07,
      "loss": 0.0988,
      "step": 1822
    },
    {
      "epoch": 0.28757345111803445,
      "grad_norm": 4.455567359924316,
      "learning_rate": 7.270970858154886e-07,
      "loss": 0.0348,
      "step": 1823
    },
    {
      "epoch": 0.2877311984856253,
      "grad_norm": 3.6792962551116943,
      "learning_rate": 7.269360811463532e-07,
      "loss": 0.0244,
      "step": 1824
    },
    {
      "epoch": 0.28788894585321606,
      "grad_norm": 4.358467102050781,
      "learning_rate": 7.267750764772178e-07,
      "loss": 0.0365,
      "step": 1825
    },
    {
      "epoch": 0.2880466932208069,
      "grad_norm": 10.892244338989258,
      "learning_rate": 7.266140718080824e-07,
      "loss": 0.0535,
      "step": 1826
    },
    {
      "epoch": 0.28820444058839767,
      "grad_norm": 3.952486991882324,
      "learning_rate": 7.26453067138947e-07,
      "loss": 0.0583,
      "step": 1827
    },
    {
      "epoch": 0.2883621879559885,
      "grad_norm": 2.9721004962921143,
      "learning_rate": 7.262920624698115e-07,
      "loss": 0.0282,
      "step": 1828
    },
    {
      "epoch": 0.2885199353235793,
      "grad_norm": 8.202765464782715,
      "learning_rate": 7.261310578006761e-07,
      "loss": 0.0941,
      "step": 1829
    },
    {
      "epoch": 0.2886776826911701,
      "grad_norm": 6.014126777648926,
      "learning_rate": 7.259700531315408e-07,
      "loss": 0.0527,
      "step": 1830
    },
    {
      "epoch": 0.2888354300587609,
      "grad_norm": 4.782886505126953,
      "learning_rate": 7.258090484624055e-07,
      "loss": 0.0494,
      "step": 1831
    },
    {
      "epoch": 0.2889931774263517,
      "grad_norm": 2.926882028579712,
      "learning_rate": 7.2564804379327e-07,
      "loss": 0.0217,
      "step": 1832
    },
    {
      "epoch": 0.2891509247939425,
      "grad_norm": 5.702027797698975,
      "learning_rate": 7.254870391241346e-07,
      "loss": 0.0771,
      "step": 1833
    },
    {
      "epoch": 0.28930867216153333,
      "grad_norm": 7.120554447174072,
      "learning_rate": 7.253260344549992e-07,
      "loss": 0.0353,
      "step": 1834
    },
    {
      "epoch": 0.2894664195291241,
      "grad_norm": 23.75204849243164,
      "learning_rate": 7.251650297858638e-07,
      "loss": 0.0424,
      "step": 1835
    },
    {
      "epoch": 0.2896241668967149,
      "grad_norm": 6.956763744354248,
      "learning_rate": 7.250040251167283e-07,
      "loss": 0.0921,
      "step": 1836
    },
    {
      "epoch": 0.2897819142643057,
      "grad_norm": 3.723970651626587,
      "learning_rate": 7.24843020447593e-07,
      "loss": 0.0291,
      "step": 1837
    },
    {
      "epoch": 0.2899396616318965,
      "grad_norm": 5.069478511810303,
      "learning_rate": 7.246820157784576e-07,
      "loss": 0.0632,
      "step": 1838
    },
    {
      "epoch": 0.2900974089994873,
      "grad_norm": 5.687950134277344,
      "learning_rate": 7.245210111093222e-07,
      "loss": 0.0449,
      "step": 1839
    },
    {
      "epoch": 0.2902551563670781,
      "grad_norm": 4.0997114181518555,
      "learning_rate": 7.243600064401867e-07,
      "loss": 0.0797,
      "step": 1840
    },
    {
      "epoch": 0.29041290373466894,
      "grad_norm": 5.048920154571533,
      "learning_rate": 7.241990017710513e-07,
      "loss": 0.0234,
      "step": 1841
    },
    {
      "epoch": 0.2905706511022597,
      "grad_norm": 7.273456573486328,
      "learning_rate": 7.240379971019159e-07,
      "loss": 0.0473,
      "step": 1842
    },
    {
      "epoch": 0.29072839846985055,
      "grad_norm": 6.181704998016357,
      "learning_rate": 7.238769924327805e-07,
      "loss": 0.0879,
      "step": 1843
    },
    {
      "epoch": 0.2908861458374413,
      "grad_norm": 5.6075310707092285,
      "learning_rate": 7.23715987763645e-07,
      "loss": 0.0599,
      "step": 1844
    },
    {
      "epoch": 0.29104389320503216,
      "grad_norm": 6.415543079376221,
      "learning_rate": 7.235549830945097e-07,
      "loss": 0.065,
      "step": 1845
    },
    {
      "epoch": 0.29120164057262293,
      "grad_norm": 4.180621147155762,
      "learning_rate": 7.233939784253743e-07,
      "loss": 0.0609,
      "step": 1846
    },
    {
      "epoch": 0.29135938794021377,
      "grad_norm": 3.543720006942749,
      "learning_rate": 7.232329737562389e-07,
      "loss": 0.0442,
      "step": 1847
    },
    {
      "epoch": 0.29151713530780454,
      "grad_norm": 8.813889503479004,
      "learning_rate": 7.230719690871034e-07,
      "loss": 0.1239,
      "step": 1848
    },
    {
      "epoch": 0.2916748826753954,
      "grad_norm": 7.704339981079102,
      "learning_rate": 7.22910964417968e-07,
      "loss": 0.0891,
      "step": 1849
    },
    {
      "epoch": 0.29183263004298615,
      "grad_norm": 7.240747928619385,
      "learning_rate": 7.227499597488327e-07,
      "loss": 0.0787,
      "step": 1850
    },
    {
      "epoch": 0.291990377410577,
      "grad_norm": 4.727330684661865,
      "learning_rate": 7.225889550796973e-07,
      "loss": 0.0357,
      "step": 1851
    },
    {
      "epoch": 0.29214812477816776,
      "grad_norm": 4.392090797424316,
      "learning_rate": 7.22427950410562e-07,
      "loss": 0.0307,
      "step": 1852
    },
    {
      "epoch": 0.2923058721457586,
      "grad_norm": 4.477465629577637,
      "learning_rate": 7.222669457414265e-07,
      "loss": 0.073,
      "step": 1853
    },
    {
      "epoch": 0.29246361951334937,
      "grad_norm": 9.257769584655762,
      "learning_rate": 7.221059410722911e-07,
      "loss": 0.0808,
      "step": 1854
    },
    {
      "epoch": 0.29262136688094015,
      "grad_norm": 4.829099655151367,
      "learning_rate": 7.219449364031557e-07,
      "loss": 0.0491,
      "step": 1855
    },
    {
      "epoch": 0.292779114248531,
      "grad_norm": 6.342634677886963,
      "learning_rate": 7.217839317340203e-07,
      "loss": 0.0632,
      "step": 1856
    },
    {
      "epoch": 0.29293686161612176,
      "grad_norm": 3.0647029876708984,
      "learning_rate": 7.216229270648848e-07,
      "loss": 0.0425,
      "step": 1857
    },
    {
      "epoch": 0.2930946089837126,
      "grad_norm": 4.071573257446289,
      "learning_rate": 7.214619223957494e-07,
      "loss": 0.0321,
      "step": 1858
    },
    {
      "epoch": 0.29325235635130337,
      "grad_norm": 5.210506439208984,
      "learning_rate": 7.21300917726614e-07,
      "loss": 0.0836,
      "step": 1859
    },
    {
      "epoch": 0.2934101037188942,
      "grad_norm": 9.28438663482666,
      "learning_rate": 7.211399130574787e-07,
      "loss": 0.0645,
      "step": 1860
    },
    {
      "epoch": 0.293567851086485,
      "grad_norm": 6.821744918823242,
      "learning_rate": 7.209789083883432e-07,
      "loss": 0.0871,
      "step": 1861
    },
    {
      "epoch": 0.2937255984540758,
      "grad_norm": 3.003976821899414,
      "learning_rate": 7.208179037192078e-07,
      "loss": 0.0582,
      "step": 1862
    },
    {
      "epoch": 0.2938833458216666,
      "grad_norm": 6.32629919052124,
      "learning_rate": 7.206568990500724e-07,
      "loss": 0.042,
      "step": 1863
    },
    {
      "epoch": 0.2940410931892574,
      "grad_norm": 8.073723793029785,
      "learning_rate": 7.20495894380937e-07,
      "loss": 0.0911,
      "step": 1864
    },
    {
      "epoch": 0.2941988405568482,
      "grad_norm": 7.266777038574219,
      "learning_rate": 7.203348897118015e-07,
      "loss": 0.1142,
      "step": 1865
    },
    {
      "epoch": 0.29435658792443903,
      "grad_norm": 6.750946044921875,
      "learning_rate": 7.201738850426661e-07,
      "loss": 0.0666,
      "step": 1866
    },
    {
      "epoch": 0.2945143352920298,
      "grad_norm": 3.6444435119628906,
      "learning_rate": 7.200128803735309e-07,
      "loss": 0.0288,
      "step": 1867
    },
    {
      "epoch": 0.29467208265962064,
      "grad_norm": 3.0052032470703125,
      "learning_rate": 7.198518757043955e-07,
      "loss": 0.0189,
      "step": 1868
    },
    {
      "epoch": 0.2948298300272114,
      "grad_norm": 5.2682719230651855,
      "learning_rate": 7.1969087103526e-07,
      "loss": 0.0834,
      "step": 1869
    },
    {
      "epoch": 0.29498757739480225,
      "grad_norm": 5.890918254852295,
      "learning_rate": 7.195298663661246e-07,
      "loss": 0.0704,
      "step": 1870
    },
    {
      "epoch": 0.295145324762393,
      "grad_norm": 3.2959954738616943,
      "learning_rate": 7.193688616969892e-07,
      "loss": 0.0186,
      "step": 1871
    },
    {
      "epoch": 0.29530307212998386,
      "grad_norm": 3.670823097229004,
      "learning_rate": 7.192078570278538e-07,
      "loss": 0.0651,
      "step": 1872
    },
    {
      "epoch": 0.29546081949757463,
      "grad_norm": 5.776731491088867,
      "learning_rate": 7.190468523587184e-07,
      "loss": 0.0644,
      "step": 1873
    },
    {
      "epoch": 0.2956185668651654,
      "grad_norm": 2.765413761138916,
      "learning_rate": 7.188858476895829e-07,
      "loss": 0.0554,
      "step": 1874
    },
    {
      "epoch": 0.29577631423275624,
      "grad_norm": 7.823866844177246,
      "learning_rate": 7.187248430204476e-07,
      "loss": 0.0641,
      "step": 1875
    },
    {
      "epoch": 0.295934061600347,
      "grad_norm": 3.9487831592559814,
      "learning_rate": 7.185638383513122e-07,
      "loss": 0.0637,
      "step": 1876
    },
    {
      "epoch": 0.29609180896793785,
      "grad_norm": 3.203082323074341,
      "learning_rate": 7.184028336821768e-07,
      "loss": 0.0401,
      "step": 1877
    },
    {
      "epoch": 0.29624955633552863,
      "grad_norm": 3.9091076850891113,
      "learning_rate": 7.182418290130413e-07,
      "loss": 0.0448,
      "step": 1878
    },
    {
      "epoch": 0.29640730370311946,
      "grad_norm": 4.002121925354004,
      "learning_rate": 7.180808243439059e-07,
      "loss": 0.0873,
      "step": 1879
    },
    {
      "epoch": 0.29656505107071024,
      "grad_norm": 6.394353866577148,
      "learning_rate": 7.179198196747705e-07,
      "loss": 0.0338,
      "step": 1880
    },
    {
      "epoch": 0.29672279843830107,
      "grad_norm": 6.340281963348389,
      "learning_rate": 7.177588150056351e-07,
      "loss": 0.0635,
      "step": 1881
    },
    {
      "epoch": 0.29688054580589185,
      "grad_norm": 4.5851640701293945,
      "learning_rate": 7.175978103364997e-07,
      "loss": 0.0468,
      "step": 1882
    },
    {
      "epoch": 0.2970382931734827,
      "grad_norm": 5.020273685455322,
      "learning_rate": 7.174368056673643e-07,
      "loss": 0.0341,
      "step": 1883
    },
    {
      "epoch": 0.29719604054107346,
      "grad_norm": 11.44133186340332,
      "learning_rate": 7.17275800998229e-07,
      "loss": 0.0875,
      "step": 1884
    },
    {
      "epoch": 0.2973537879086643,
      "grad_norm": 4.677864074707031,
      "learning_rate": 7.171147963290936e-07,
      "loss": 0.0216,
      "step": 1885
    },
    {
      "epoch": 0.29751153527625507,
      "grad_norm": 10.180984497070312,
      "learning_rate": 7.169537916599581e-07,
      "loss": 0.076,
      "step": 1886
    },
    {
      "epoch": 0.2976692826438459,
      "grad_norm": 3.750933885574341,
      "learning_rate": 7.167927869908227e-07,
      "loss": 0.0374,
      "step": 1887
    },
    {
      "epoch": 0.2978270300114367,
      "grad_norm": 6.10151481628418,
      "learning_rate": 7.166317823216873e-07,
      "loss": 0.0442,
      "step": 1888
    },
    {
      "epoch": 0.2979847773790275,
      "grad_norm": 8.401910781860352,
      "learning_rate": 7.164707776525519e-07,
      "loss": 0.0532,
      "step": 1889
    },
    {
      "epoch": 0.2981425247466183,
      "grad_norm": 4.196383476257324,
      "learning_rate": 7.163097729834165e-07,
      "loss": 0.0568,
      "step": 1890
    },
    {
      "epoch": 0.2983002721142091,
      "grad_norm": 3.7810068130493164,
      "learning_rate": 7.161487683142811e-07,
      "loss": 0.0284,
      "step": 1891
    },
    {
      "epoch": 0.2984580194817999,
      "grad_norm": 8.036553382873535,
      "learning_rate": 7.159877636451457e-07,
      "loss": 0.0707,
      "step": 1892
    },
    {
      "epoch": 0.2986157668493907,
      "grad_norm": 5.693993091583252,
      "learning_rate": 7.158267589760103e-07,
      "loss": 0.0834,
      "step": 1893
    },
    {
      "epoch": 0.2987735142169815,
      "grad_norm": 5.394157409667969,
      "learning_rate": 7.156657543068748e-07,
      "loss": 0.0629,
      "step": 1894
    },
    {
      "epoch": 0.2989312615845723,
      "grad_norm": 7.30886173248291,
      "learning_rate": 7.155047496377394e-07,
      "loss": 0.1106,
      "step": 1895
    },
    {
      "epoch": 0.2990890089521631,
      "grad_norm": 6.162013053894043,
      "learning_rate": 7.15343744968604e-07,
      "loss": 0.057,
      "step": 1896
    },
    {
      "epoch": 0.2992467563197539,
      "grad_norm": 6.323513507843018,
      "learning_rate": 7.151827402994687e-07,
      "loss": 0.0469,
      "step": 1897
    },
    {
      "epoch": 0.2994045036873447,
      "grad_norm": 4.471866607666016,
      "learning_rate": 7.150217356303333e-07,
      "loss": 0.0706,
      "step": 1898
    },
    {
      "epoch": 0.2995622510549355,
      "grad_norm": 7.489285469055176,
      "learning_rate": 7.148607309611978e-07,
      "loss": 0.1049,
      "step": 1899
    },
    {
      "epoch": 0.29971999842252633,
      "grad_norm": 4.253059387207031,
      "learning_rate": 7.146997262920624e-07,
      "loss": 0.0361,
      "step": 1900
    },
    {
      "epoch": 0.2998777457901171,
      "grad_norm": 8.71242904663086,
      "learning_rate": 7.14538721622927e-07,
      "loss": 0.1012,
      "step": 1901
    },
    {
      "epoch": 0.30003549315770794,
      "grad_norm": 6.077664852142334,
      "learning_rate": 7.143777169537917e-07,
      "loss": 0.0932,
      "step": 1902
    },
    {
      "epoch": 0.3001932405252987,
      "grad_norm": 3.9309680461883545,
      "learning_rate": 7.142167122846562e-07,
      "loss": 0.0244,
      "step": 1903
    },
    {
      "epoch": 0.30035098789288955,
      "grad_norm": 6.765064716339111,
      "learning_rate": 7.140557076155208e-07,
      "loss": 0.0232,
      "step": 1904
    },
    {
      "epoch": 0.30050873526048033,
      "grad_norm": 6.2488884925842285,
      "learning_rate": 7.138947029463855e-07,
      "loss": 0.0723,
      "step": 1905
    },
    {
      "epoch": 0.30066648262807116,
      "grad_norm": 5.702681064605713,
      "learning_rate": 7.137336982772501e-07,
      "loss": 0.0544,
      "step": 1906
    },
    {
      "epoch": 0.30082422999566194,
      "grad_norm": 5.73229455947876,
      "learning_rate": 7.135726936081146e-07,
      "loss": 0.082,
      "step": 1907
    },
    {
      "epoch": 0.3009819773632528,
      "grad_norm": 7.134407043457031,
      "learning_rate": 7.134116889389792e-07,
      "loss": 0.0366,
      "step": 1908
    },
    {
      "epoch": 0.30113972473084355,
      "grad_norm": 3.8968775272369385,
      "learning_rate": 7.132506842698438e-07,
      "loss": 0.0505,
      "step": 1909
    },
    {
      "epoch": 0.3012974720984344,
      "grad_norm": 7.5660929679870605,
      "learning_rate": 7.130896796007084e-07,
      "loss": 0.0701,
      "step": 1910
    },
    {
      "epoch": 0.30145521946602516,
      "grad_norm": 11.128520011901855,
      "learning_rate": 7.129286749315729e-07,
      "loss": 0.0387,
      "step": 1911
    },
    {
      "epoch": 0.30161296683361594,
      "grad_norm": 4.411167144775391,
      "learning_rate": 7.127676702624376e-07,
      "loss": 0.0559,
      "step": 1912
    },
    {
      "epoch": 0.30177071420120677,
      "grad_norm": 5.5612263679504395,
      "learning_rate": 7.126066655933022e-07,
      "loss": 0.0235,
      "step": 1913
    },
    {
      "epoch": 0.30192846156879755,
      "grad_norm": 3.3991105556488037,
      "learning_rate": 7.124456609241668e-07,
      "loss": 0.0544,
      "step": 1914
    },
    {
      "epoch": 0.3020862089363884,
      "grad_norm": 3.648372173309326,
      "learning_rate": 7.122846562550313e-07,
      "loss": 0.0262,
      "step": 1915
    },
    {
      "epoch": 0.30224395630397916,
      "grad_norm": 2.522200345993042,
      "learning_rate": 7.121236515858959e-07,
      "loss": 0.0177,
      "step": 1916
    },
    {
      "epoch": 0.30240170367157,
      "grad_norm": 3.3338704109191895,
      "learning_rate": 7.119626469167605e-07,
      "loss": 0.0337,
      "step": 1917
    },
    {
      "epoch": 0.30255945103916076,
      "grad_norm": 4.939285755157471,
      "learning_rate": 7.118016422476251e-07,
      "loss": 0.0648,
      "step": 1918
    },
    {
      "epoch": 0.3027171984067516,
      "grad_norm": 3.7481281757354736,
      "learning_rate": 7.116406375784896e-07,
      "loss": 0.058,
      "step": 1919
    },
    {
      "epoch": 0.3028749457743424,
      "grad_norm": 2.6262574195861816,
      "learning_rate": 7.114796329093544e-07,
      "loss": 0.0416,
      "step": 1920
    },
    {
      "epoch": 0.3030326931419332,
      "grad_norm": 7.282456398010254,
      "learning_rate": 7.11318628240219e-07,
      "loss": 0.0788,
      "step": 1921
    },
    {
      "epoch": 0.303190440509524,
      "grad_norm": 4.289928913116455,
      "learning_rate": 7.111576235710836e-07,
      "loss": 0.0326,
      "step": 1922
    },
    {
      "epoch": 0.3033481878771148,
      "grad_norm": 6.543341159820557,
      "learning_rate": 7.109966189019482e-07,
      "loss": 0.0668,
      "step": 1923
    },
    {
      "epoch": 0.3035059352447056,
      "grad_norm": 10.349213600158691,
      "learning_rate": 7.108356142328127e-07,
      "loss": 0.0905,
      "step": 1924
    },
    {
      "epoch": 0.3036636826122964,
      "grad_norm": 7.131984233856201,
      "learning_rate": 7.106746095636773e-07,
      "loss": 0.0634,
      "step": 1925
    },
    {
      "epoch": 0.3038214299798872,
      "grad_norm": 4.043402671813965,
      "learning_rate": 7.105136048945419e-07,
      "loss": 0.0415,
      "step": 1926
    },
    {
      "epoch": 0.30397917734747804,
      "grad_norm": 2.7604012489318848,
      "learning_rate": 7.103526002254066e-07,
      "loss": 0.0153,
      "step": 1927
    },
    {
      "epoch": 0.3041369247150688,
      "grad_norm": 7.04880952835083,
      "learning_rate": 7.101915955562711e-07,
      "loss": 0.1251,
      "step": 1928
    },
    {
      "epoch": 0.30429467208265965,
      "grad_norm": 4.090600967407227,
      "learning_rate": 7.100305908871357e-07,
      "loss": 0.0533,
      "step": 1929
    },
    {
      "epoch": 0.3044524194502504,
      "grad_norm": 4.4699177742004395,
      "learning_rate": 7.098695862180003e-07,
      "loss": 0.0567,
      "step": 1930
    },
    {
      "epoch": 0.30461016681784125,
      "grad_norm": 53.74309539794922,
      "learning_rate": 7.097085815488649e-07,
      "loss": 0.1118,
      "step": 1931
    },
    {
      "epoch": 0.30476791418543203,
      "grad_norm": 5.222226142883301,
      "learning_rate": 7.095475768797294e-07,
      "loss": 0.0486,
      "step": 1932
    },
    {
      "epoch": 0.3049256615530228,
      "grad_norm": 4.8929266929626465,
      "learning_rate": 7.09386572210594e-07,
      "loss": 0.0416,
      "step": 1933
    },
    {
      "epoch": 0.30508340892061364,
      "grad_norm": 4.212865829467773,
      "learning_rate": 7.092255675414586e-07,
      "loss": 0.0697,
      "step": 1934
    },
    {
      "epoch": 0.3052411562882044,
      "grad_norm": 4.160580635070801,
      "learning_rate": 7.090645628723233e-07,
      "loss": 0.0234,
      "step": 1935
    },
    {
      "epoch": 0.30539890365579525,
      "grad_norm": 4.000143527984619,
      "learning_rate": 7.089035582031878e-07,
      "loss": 0.0592,
      "step": 1936
    },
    {
      "epoch": 0.30555665102338603,
      "grad_norm": 9.259430885314941,
      "learning_rate": 7.087425535340524e-07,
      "loss": 0.0591,
      "step": 1937
    },
    {
      "epoch": 0.30571439839097686,
      "grad_norm": 8.322980880737305,
      "learning_rate": 7.085815488649171e-07,
      "loss": 0.0901,
      "step": 1938
    },
    {
      "epoch": 0.30587214575856764,
      "grad_norm": 6.6517133712768555,
      "learning_rate": 7.084205441957817e-07,
      "loss": 0.0672,
      "step": 1939
    },
    {
      "epoch": 0.30602989312615847,
      "grad_norm": 8.325651168823242,
      "learning_rate": 7.082595395266462e-07,
      "loss": 0.0746,
      "step": 1940
    },
    {
      "epoch": 0.30618764049374925,
      "grad_norm": 5.0230393409729,
      "learning_rate": 7.080985348575108e-07,
      "loss": 0.0477,
      "step": 1941
    },
    {
      "epoch": 0.3063453878613401,
      "grad_norm": 2.52583909034729,
      "learning_rate": 7.079375301883755e-07,
      "loss": 0.0285,
      "step": 1942
    },
    {
      "epoch": 0.30650313522893086,
      "grad_norm": 5.606345176696777,
      "learning_rate": 7.077765255192401e-07,
      "loss": 0.072,
      "step": 1943
    },
    {
      "epoch": 0.3066608825965217,
      "grad_norm": 7.923591613769531,
      "learning_rate": 7.076155208501047e-07,
      "loss": 0.0639,
      "step": 1944
    },
    {
      "epoch": 0.30681862996411247,
      "grad_norm": 4.813804626464844,
      "learning_rate": 7.074545161809692e-07,
      "loss": 0.0549,
      "step": 1945
    },
    {
      "epoch": 0.3069763773317033,
      "grad_norm": 7.287686347961426,
      "learning_rate": 7.072935115118338e-07,
      "loss": 0.0409,
      "step": 1946
    },
    {
      "epoch": 0.3071341246992941,
      "grad_norm": 4.390560150146484,
      "learning_rate": 7.071325068426984e-07,
      "loss": 0.0591,
      "step": 1947
    },
    {
      "epoch": 0.3072918720668849,
      "grad_norm": 3.507387161254883,
      "learning_rate": 7.06971502173563e-07,
      "loss": 0.0565,
      "step": 1948
    },
    {
      "epoch": 0.3074496194344757,
      "grad_norm": 8.796087265014648,
      "learning_rate": 7.068104975044275e-07,
      "loss": 0.0663,
      "step": 1949
    },
    {
      "epoch": 0.3076073668020665,
      "grad_norm": 5.937930583953857,
      "learning_rate": 7.066494928352922e-07,
      "loss": 0.0525,
      "step": 1950
    },
    {
      "epoch": 0.3077651141696573,
      "grad_norm": 5.049323558807373,
      "learning_rate": 7.064884881661568e-07,
      "loss": 0.0718,
      "step": 1951
    },
    {
      "epoch": 0.30792286153724807,
      "grad_norm": 7.667891502380371,
      "learning_rate": 7.063274834970214e-07,
      "loss": 0.0425,
      "step": 1952
    },
    {
      "epoch": 0.3080806089048389,
      "grad_norm": 6.416477680206299,
      "learning_rate": 7.061664788278859e-07,
      "loss": 0.0658,
      "step": 1953
    },
    {
      "epoch": 0.3082383562724297,
      "grad_norm": 11.372632026672363,
      "learning_rate": 7.060054741587505e-07,
      "loss": 0.0367,
      "step": 1954
    },
    {
      "epoch": 0.3083961036400205,
      "grad_norm": 3.07737135887146,
      "learning_rate": 7.058444694896152e-07,
      "loss": 0.0255,
      "step": 1955
    },
    {
      "epoch": 0.3085538510076113,
      "grad_norm": 6.110159397125244,
      "learning_rate": 7.056834648204798e-07,
      "loss": 0.1018,
      "step": 1956
    },
    {
      "epoch": 0.3087115983752021,
      "grad_norm": 12.801488876342773,
      "learning_rate": 7.055224601513444e-07,
      "loss": 0.076,
      "step": 1957
    },
    {
      "epoch": 0.3088693457427929,
      "grad_norm": 5.292741298675537,
      "learning_rate": 7.05361455482209e-07,
      "loss": 0.0743,
      "step": 1958
    },
    {
      "epoch": 0.30902709311038373,
      "grad_norm": 4.4499430656433105,
      "learning_rate": 7.052004508130736e-07,
      "loss": 0.0944,
      "step": 1959
    },
    {
      "epoch": 0.3091848404779745,
      "grad_norm": 6.13045072555542,
      "learning_rate": 7.050394461439382e-07,
      "loss": 0.0544,
      "step": 1960
    },
    {
      "epoch": 0.30934258784556534,
      "grad_norm": 8.339452743530273,
      "learning_rate": 7.048784414748027e-07,
      "loss": 0.0413,
      "step": 1961
    },
    {
      "epoch": 0.3095003352131561,
      "grad_norm": 7.852583408355713,
      "learning_rate": 7.047174368056673e-07,
      "loss": 0.0628,
      "step": 1962
    },
    {
      "epoch": 0.30965808258074695,
      "grad_norm": 6.660775184631348,
      "learning_rate": 7.045564321365319e-07,
      "loss": 0.0393,
      "step": 1963
    },
    {
      "epoch": 0.30981582994833773,
      "grad_norm": 4.972191333770752,
      "learning_rate": 7.043954274673965e-07,
      "loss": 0.0336,
      "step": 1964
    },
    {
      "epoch": 0.30997357731592856,
      "grad_norm": 2.4434168338775635,
      "learning_rate": 7.042344227982611e-07,
      "loss": 0.027,
      "step": 1965
    },
    {
      "epoch": 0.31013132468351934,
      "grad_norm": 4.470157146453857,
      "learning_rate": 7.040734181291257e-07,
      "loss": 0.0742,
      "step": 1966
    },
    {
      "epoch": 0.31028907205111017,
      "grad_norm": 7.267810344696045,
      "learning_rate": 7.039124134599903e-07,
      "loss": 0.1028,
      "step": 1967
    },
    {
      "epoch": 0.31044681941870095,
      "grad_norm": 6.394923686981201,
      "learning_rate": 7.037514087908549e-07,
      "loss": 0.0615,
      "step": 1968
    },
    {
      "epoch": 0.3106045667862918,
      "grad_norm": 6.019500732421875,
      "learning_rate": 7.035904041217195e-07,
      "loss": 0.0602,
      "step": 1969
    },
    {
      "epoch": 0.31076231415388256,
      "grad_norm": 4.828829765319824,
      "learning_rate": 7.03429399452584e-07,
      "loss": 0.0455,
      "step": 1970
    },
    {
      "epoch": 0.31092006152147333,
      "grad_norm": 9.272650718688965,
      "learning_rate": 7.032683947834486e-07,
      "loss": 0.0343,
      "step": 1971
    },
    {
      "epoch": 0.31107780888906417,
      "grad_norm": 5.205511093139648,
      "learning_rate": 7.031073901143134e-07,
      "loss": 0.0532,
      "step": 1972
    },
    {
      "epoch": 0.31123555625665494,
      "grad_norm": 5.682951927185059,
      "learning_rate": 7.02946385445178e-07,
      "loss": 0.045,
      "step": 1973
    },
    {
      "epoch": 0.3113933036242458,
      "grad_norm": 4.142481327056885,
      "learning_rate": 7.027853807760425e-07,
      "loss": 0.0319,
      "step": 1974
    },
    {
      "epoch": 0.31155105099183655,
      "grad_norm": 6.376063823699951,
      "learning_rate": 7.026243761069071e-07,
      "loss": 0.0631,
      "step": 1975
    },
    {
      "epoch": 0.3117087983594274,
      "grad_norm": 7.768744468688965,
      "learning_rate": 7.024633714377717e-07,
      "loss": 0.0538,
      "step": 1976
    },
    {
      "epoch": 0.31186654572701816,
      "grad_norm": 7.69910192489624,
      "learning_rate": 7.023023667686363e-07,
      "loss": 0.0719,
      "step": 1977
    },
    {
      "epoch": 0.312024293094609,
      "grad_norm": 3.8771684169769287,
      "learning_rate": 7.021413620995008e-07,
      "loss": 0.0442,
      "step": 1978
    },
    {
      "epoch": 0.3121820404621998,
      "grad_norm": 11.783376693725586,
      "learning_rate": 7.019803574303654e-07,
      "loss": 0.0493,
      "step": 1979
    },
    {
      "epoch": 0.3123397878297906,
      "grad_norm": 5.049818992614746,
      "learning_rate": 7.018193527612301e-07,
      "loss": 0.0325,
      "step": 1980
    },
    {
      "epoch": 0.3124975351973814,
      "grad_norm": 5.231262683868408,
      "learning_rate": 7.016583480920947e-07,
      "loss": 0.0374,
      "step": 1981
    },
    {
      "epoch": 0.3126552825649722,
      "grad_norm": 2.590451955795288,
      "learning_rate": 7.014973434229592e-07,
      "loss": 0.0494,
      "step": 1982
    },
    {
      "epoch": 0.312813029932563,
      "grad_norm": 4.199472427368164,
      "learning_rate": 7.013363387538238e-07,
      "loss": 0.0495,
      "step": 1983
    },
    {
      "epoch": 0.3129707773001538,
      "grad_norm": 3.365995168685913,
      "learning_rate": 7.011753340846884e-07,
      "loss": 0.0169,
      "step": 1984
    },
    {
      "epoch": 0.3131285246677446,
      "grad_norm": 7.31893253326416,
      "learning_rate": 7.01014329415553e-07,
      "loss": 0.1077,
      "step": 1985
    },
    {
      "epoch": 0.31328627203533543,
      "grad_norm": 4.630749225616455,
      "learning_rate": 7.008533247464175e-07,
      "loss": 0.0563,
      "step": 1986
    },
    {
      "epoch": 0.3134440194029262,
      "grad_norm": 4.555026054382324,
      "learning_rate": 7.006923200772822e-07,
      "loss": 0.0628,
      "step": 1987
    },
    {
      "epoch": 0.31360176677051704,
      "grad_norm": 3.6789534091949463,
      "learning_rate": 7.005313154081468e-07,
      "loss": 0.0419,
      "step": 1988
    },
    {
      "epoch": 0.3137595141381078,
      "grad_norm": 5.142852306365967,
      "learning_rate": 7.003703107390114e-07,
      "loss": 0.0742,
      "step": 1989
    },
    {
      "epoch": 0.3139172615056986,
      "grad_norm": 5.34221076965332,
      "learning_rate": 7.002093060698761e-07,
      "loss": 0.029,
      "step": 1990
    },
    {
      "epoch": 0.31407500887328943,
      "grad_norm": 6.0126261711120605,
      "learning_rate": 7.000483014007406e-07,
      "loss": 0.0674,
      "step": 1991
    },
    {
      "epoch": 0.3142327562408802,
      "grad_norm": 6.048847198486328,
      "learning_rate": 6.998872967316052e-07,
      "loss": 0.0903,
      "step": 1992
    },
    {
      "epoch": 0.31439050360847104,
      "grad_norm": 5.697565078735352,
      "learning_rate": 6.997262920624698e-07,
      "loss": 0.0496,
      "step": 1993
    },
    {
      "epoch": 0.3145482509760618,
      "grad_norm": 2.6553454399108887,
      "learning_rate": 6.995652873933344e-07,
      "loss": 0.0292,
      "step": 1994
    },
    {
      "epoch": 0.31470599834365265,
      "grad_norm": 3.822661876678467,
      "learning_rate": 6.99404282724199e-07,
      "loss": 0.0382,
      "step": 1995
    },
    {
      "epoch": 0.3148637457112434,
      "grad_norm": 5.864170074462891,
      "learning_rate": 6.992432780550636e-07,
      "loss": 0.0975,
      "step": 1996
    },
    {
      "epoch": 0.31502149307883426,
      "grad_norm": 2.8939883708953857,
      "learning_rate": 6.990822733859282e-07,
      "loss": 0.0157,
      "step": 1997
    },
    {
      "epoch": 0.31517924044642504,
      "grad_norm": 6.828817367553711,
      "learning_rate": 6.989212687167928e-07,
      "loss": 0.0615,
      "step": 1998
    },
    {
      "epoch": 0.31533698781401587,
      "grad_norm": 6.745560646057129,
      "learning_rate": 6.987602640476573e-07,
      "loss": 0.0683,
      "step": 1999
    },
    {
      "epoch": 0.31549473518160664,
      "grad_norm": 7.2444539070129395,
      "learning_rate": 6.985992593785219e-07,
      "loss": 0.0238,
      "step": 2000
    },
    {
      "epoch": 0.3156524825491975,
      "grad_norm": 2.476250648498535,
      "learning_rate": 6.984382547093865e-07,
      "loss": 0.0303,
      "step": 2001
    },
    {
      "epoch": 0.31581022991678825,
      "grad_norm": 3.634782075881958,
      "learning_rate": 6.982772500402512e-07,
      "loss": 0.0332,
      "step": 2002
    },
    {
      "epoch": 0.3159679772843791,
      "grad_norm": 6.093087673187256,
      "learning_rate": 6.981162453711157e-07,
      "loss": 0.0594,
      "step": 2003
    },
    {
      "epoch": 0.31612572465196986,
      "grad_norm": 4.593890190124512,
      "learning_rate": 6.979552407019803e-07,
      "loss": 0.0619,
      "step": 2004
    },
    {
      "epoch": 0.3162834720195607,
      "grad_norm": 6.26422643661499,
      "learning_rate": 6.977942360328449e-07,
      "loss": 0.0453,
      "step": 2005
    },
    {
      "epoch": 0.3164412193871515,
      "grad_norm": 4.618278503417969,
      "learning_rate": 6.976332313637095e-07,
      "loss": 0.0746,
      "step": 2006
    },
    {
      "epoch": 0.3165989667547423,
      "grad_norm": 3.2007484436035156,
      "learning_rate": 6.97472226694574e-07,
      "loss": 0.0333,
      "step": 2007
    },
    {
      "epoch": 0.3167567141223331,
      "grad_norm": 4.1666154861450195,
      "learning_rate": 6.973112220254387e-07,
      "loss": 0.0394,
      "step": 2008
    },
    {
      "epoch": 0.31691446148992386,
      "grad_norm": 2.576082468032837,
      "learning_rate": 6.971502173563033e-07,
      "loss": 0.0605,
      "step": 2009
    },
    {
      "epoch": 0.3170722088575147,
      "grad_norm": 10.811822891235352,
      "learning_rate": 6.96989212687168e-07,
      "loss": 0.0873,
      "step": 2010
    },
    {
      "epoch": 0.31722995622510547,
      "grad_norm": 4.870930194854736,
      "learning_rate": 6.968282080180325e-07,
      "loss": 0.0439,
      "step": 2011
    },
    {
      "epoch": 0.3173877035926963,
      "grad_norm": 5.0976643562316895,
      "learning_rate": 6.966672033488971e-07,
      "loss": 0.0692,
      "step": 2012
    },
    {
      "epoch": 0.3175454509602871,
      "grad_norm": 6.9528656005859375,
      "learning_rate": 6.965061986797617e-07,
      "loss": 0.0664,
      "step": 2013
    },
    {
      "epoch": 0.3177031983278779,
      "grad_norm": 7.9897589683532715,
      "learning_rate": 6.963451940106263e-07,
      "loss": 0.0913,
      "step": 2014
    },
    {
      "epoch": 0.3178609456954687,
      "grad_norm": 6.256965637207031,
      "learning_rate": 6.961841893414909e-07,
      "loss": 0.0549,
      "step": 2015
    },
    {
      "epoch": 0.3180186930630595,
      "grad_norm": 7.676211833953857,
      "learning_rate": 6.960231846723554e-07,
      "loss": 0.0549,
      "step": 2016
    },
    {
      "epoch": 0.3181764404306503,
      "grad_norm": 4.518301486968994,
      "learning_rate": 6.958621800032201e-07,
      "loss": 0.0432,
      "step": 2017
    },
    {
      "epoch": 0.31833418779824113,
      "grad_norm": 10.347399711608887,
      "learning_rate": 6.957011753340847e-07,
      "loss": 0.0594,
      "step": 2018
    },
    {
      "epoch": 0.3184919351658319,
      "grad_norm": 5.0956926345825195,
      "learning_rate": 6.955401706649493e-07,
      "loss": 0.0662,
      "step": 2019
    },
    {
      "epoch": 0.31864968253342274,
      "grad_norm": 5.088799476623535,
      "learning_rate": 6.953791659958138e-07,
      "loss": 0.0416,
      "step": 2020
    },
    {
      "epoch": 0.3188074299010135,
      "grad_norm": 3.090458869934082,
      "learning_rate": 6.952181613266784e-07,
      "loss": 0.0323,
      "step": 2021
    },
    {
      "epoch": 0.31896517726860435,
      "grad_norm": 5.809577465057373,
      "learning_rate": 6.95057156657543e-07,
      "loss": 0.0568,
      "step": 2022
    },
    {
      "epoch": 0.3191229246361951,
      "grad_norm": 4.143522262573242,
      "learning_rate": 6.948961519884076e-07,
      "loss": 0.0466,
      "step": 2023
    },
    {
      "epoch": 0.31928067200378596,
      "grad_norm": 6.0525126457214355,
      "learning_rate": 6.947351473192721e-07,
      "loss": 0.056,
      "step": 2024
    },
    {
      "epoch": 0.31943841937137674,
      "grad_norm": 4.867067337036133,
      "learning_rate": 6.945741426501369e-07,
      "loss": 0.0598,
      "step": 2025
    },
    {
      "epoch": 0.31959616673896757,
      "grad_norm": 4.798086643218994,
      "learning_rate": 6.944131379810015e-07,
      "loss": 0.0743,
      "step": 2026
    },
    {
      "epoch": 0.31975391410655835,
      "grad_norm": 4.135916233062744,
      "learning_rate": 6.942521333118661e-07,
      "loss": 0.057,
      "step": 2027
    },
    {
      "epoch": 0.3199116614741491,
      "grad_norm": 4.452865123748779,
      "learning_rate": 6.940911286427306e-07,
      "loss": 0.0386,
      "step": 2028
    },
    {
      "epoch": 0.32006940884173996,
      "grad_norm": 7.033433437347412,
      "learning_rate": 6.939301239735952e-07,
      "loss": 0.0528,
      "step": 2029
    },
    {
      "epoch": 0.32022715620933073,
      "grad_norm": 14.667048454284668,
      "learning_rate": 6.937691193044598e-07,
      "loss": 0.0659,
      "step": 2030
    },
    {
      "epoch": 0.32038490357692156,
      "grad_norm": 3.494577407836914,
      "learning_rate": 6.936081146353244e-07,
      "loss": 0.0451,
      "step": 2031
    },
    {
      "epoch": 0.32054265094451234,
      "grad_norm": 3.9050979614257812,
      "learning_rate": 6.934471099661889e-07,
      "loss": 0.0464,
      "step": 2032
    },
    {
      "epoch": 0.3207003983121032,
      "grad_norm": 8.84255599975586,
      "learning_rate": 6.932861052970536e-07,
      "loss": 0.101,
      "step": 2033
    },
    {
      "epoch": 0.32085814567969395,
      "grad_norm": 6.650998115539551,
      "learning_rate": 6.931251006279182e-07,
      "loss": 0.0759,
      "step": 2034
    },
    {
      "epoch": 0.3210158930472848,
      "grad_norm": 5.78363037109375,
      "learning_rate": 6.929640959587828e-07,
      "loss": 0.0775,
      "step": 2035
    },
    {
      "epoch": 0.32117364041487556,
      "grad_norm": 5.970606803894043,
      "learning_rate": 6.928030912896474e-07,
      "loss": 0.0682,
      "step": 2036
    },
    {
      "epoch": 0.3213313877824664,
      "grad_norm": 7.525320529937744,
      "learning_rate": 6.926420866205119e-07,
      "loss": 0.0684,
      "step": 2037
    },
    {
      "epoch": 0.32148913515005717,
      "grad_norm": 7.529545783996582,
      "learning_rate": 6.924810819513765e-07,
      "loss": 0.0678,
      "step": 2038
    },
    {
      "epoch": 0.321646882517648,
      "grad_norm": 6.4445366859436035,
      "learning_rate": 6.923200772822411e-07,
      "loss": 0.0823,
      "step": 2039
    },
    {
      "epoch": 0.3218046298852388,
      "grad_norm": 3.999107599258423,
      "learning_rate": 6.921590726131058e-07,
      "loss": 0.0734,
      "step": 2040
    },
    {
      "epoch": 0.3219623772528296,
      "grad_norm": 3.815798282623291,
      "learning_rate": 6.919980679439703e-07,
      "loss": 0.0503,
      "step": 2041
    },
    {
      "epoch": 0.3221201246204204,
      "grad_norm": 4.9361958503723145,
      "learning_rate": 6.91837063274835e-07,
      "loss": 0.0446,
      "step": 2042
    },
    {
      "epoch": 0.3222778719880112,
      "grad_norm": 7.048590660095215,
      "learning_rate": 6.916760586056996e-07,
      "loss": 0.0762,
      "step": 2043
    },
    {
      "epoch": 0.322435619355602,
      "grad_norm": 10.876458168029785,
      "learning_rate": 6.915150539365642e-07,
      "loss": 0.0634,
      "step": 2044
    },
    {
      "epoch": 0.32259336672319283,
      "grad_norm": 9.228300094604492,
      "learning_rate": 6.913540492674287e-07,
      "loss": 0.0479,
      "step": 2045
    },
    {
      "epoch": 0.3227511140907836,
      "grad_norm": 5.568813323974609,
      "learning_rate": 6.911930445982933e-07,
      "loss": 0.0547,
      "step": 2046
    },
    {
      "epoch": 0.32290886145837444,
      "grad_norm": 7.572629451751709,
      "learning_rate": 6.91032039929158e-07,
      "loss": 0.0675,
      "step": 2047
    },
    {
      "epoch": 0.3230666088259652,
      "grad_norm": 5.944652080535889,
      "learning_rate": 6.908710352600226e-07,
      "loss": 0.0353,
      "step": 2048
    },
    {
      "epoch": 0.323224356193556,
      "grad_norm": 8.71141242980957,
      "learning_rate": 6.907100305908871e-07,
      "loss": 0.0571,
      "step": 2049
    },
    {
      "epoch": 0.3233821035611468,
      "grad_norm": 7.83239221572876,
      "learning_rate": 6.905490259217517e-07,
      "loss": 0.0259,
      "step": 2050
    },
    {
      "epoch": 0.3235398509287376,
      "grad_norm": 5.153806209564209,
      "learning_rate": 6.903880212526163e-07,
      "loss": 0.064,
      "step": 2051
    },
    {
      "epoch": 0.32369759829632844,
      "grad_norm": 4.7819623947143555,
      "learning_rate": 6.902270165834809e-07,
      "loss": 0.0314,
      "step": 2052
    },
    {
      "epoch": 0.3238553456639192,
      "grad_norm": 6.805207252502441,
      "learning_rate": 6.900660119143454e-07,
      "loss": 0.0347,
      "step": 2053
    },
    {
      "epoch": 0.32401309303151005,
      "grad_norm": 3.347442626953125,
      "learning_rate": 6.8990500724521e-07,
      "loss": 0.0356,
      "step": 2054
    },
    {
      "epoch": 0.3241708403991008,
      "grad_norm": 5.032720565795898,
      "learning_rate": 6.897440025760747e-07,
      "loss": 0.0661,
      "step": 2055
    },
    {
      "epoch": 0.32432858776669166,
      "grad_norm": 5.20153284072876,
      "learning_rate": 6.895829979069393e-07,
      "loss": 0.0597,
      "step": 2056
    },
    {
      "epoch": 0.32448633513428243,
      "grad_norm": 6.703059196472168,
      "learning_rate": 6.894219932378038e-07,
      "loss": 0.0479,
      "step": 2057
    },
    {
      "epoch": 0.32464408250187327,
      "grad_norm": 7.035309791564941,
      "learning_rate": 6.892609885686684e-07,
      "loss": 0.0238,
      "step": 2058
    },
    {
      "epoch": 0.32480182986946404,
      "grad_norm": 2.6613142490386963,
      "learning_rate": 6.89099983899533e-07,
      "loss": 0.0788,
      "step": 2059
    },
    {
      "epoch": 0.3249595772370549,
      "grad_norm": 4.252102851867676,
      "learning_rate": 6.889389792303976e-07,
      "loss": 0.0417,
      "step": 2060
    },
    {
      "epoch": 0.32511732460464565,
      "grad_norm": 2.8893206119537354,
      "learning_rate": 6.887779745612623e-07,
      "loss": 0.015,
      "step": 2061
    },
    {
      "epoch": 0.3252750719722365,
      "grad_norm": 7.74230432510376,
      "learning_rate": 6.886169698921268e-07,
      "loss": 0.0819,
      "step": 2062
    },
    {
      "epoch": 0.32543281933982726,
      "grad_norm": 3.7796642780303955,
      "learning_rate": 6.884559652229915e-07,
      "loss": 0.0581,
      "step": 2063
    },
    {
      "epoch": 0.3255905667074181,
      "grad_norm": 7.510829448699951,
      "learning_rate": 6.882949605538561e-07,
      "loss": 0.0723,
      "step": 2064
    },
    {
      "epoch": 0.32574831407500887,
      "grad_norm": 7.586187362670898,
      "learning_rate": 6.881339558847207e-07,
      "loss": 0.0408,
      "step": 2065
    },
    {
      "epoch": 0.3259060614425997,
      "grad_norm": 5.421665191650391,
      "learning_rate": 6.879729512155852e-07,
      "loss": 0.0322,
      "step": 2066
    },
    {
      "epoch": 0.3260638088101905,
      "grad_norm": 6.571201324462891,
      "learning_rate": 6.878119465464498e-07,
      "loss": 0.0686,
      "step": 2067
    },
    {
      "epoch": 0.32622155617778126,
      "grad_norm": 5.433219909667969,
      "learning_rate": 6.876509418773144e-07,
      "loss": 0.0555,
      "step": 2068
    },
    {
      "epoch": 0.3263793035453721,
      "grad_norm": 8.206216812133789,
      "learning_rate": 6.87489937208179e-07,
      "loss": 0.079,
      "step": 2069
    },
    {
      "epoch": 0.32653705091296287,
      "grad_norm": 5.314496040344238,
      "learning_rate": 6.873289325390436e-07,
      "loss": 0.0483,
      "step": 2070
    },
    {
      "epoch": 0.3266947982805537,
      "grad_norm": 3.5508322715759277,
      "learning_rate": 6.871679278699082e-07,
      "loss": 0.0287,
      "step": 2071
    },
    {
      "epoch": 0.3268525456481445,
      "grad_norm": 3.366732597351074,
      "learning_rate": 6.870069232007728e-07,
      "loss": 0.0152,
      "step": 2072
    },
    {
      "epoch": 0.3270102930157353,
      "grad_norm": 5.403148651123047,
      "learning_rate": 6.868459185316374e-07,
      "loss": 0.0548,
      "step": 2073
    },
    {
      "epoch": 0.3271680403833261,
      "grad_norm": 15.252777099609375,
      "learning_rate": 6.866849138625019e-07,
      "loss": 0.1246,
      "step": 2074
    },
    {
      "epoch": 0.3273257877509169,
      "grad_norm": 3.936343193054199,
      "learning_rate": 6.865239091933665e-07,
      "loss": 0.0413,
      "step": 2075
    },
    {
      "epoch": 0.3274835351185077,
      "grad_norm": 3.2121474742889404,
      "learning_rate": 6.863629045242311e-07,
      "loss": 0.0259,
      "step": 2076
    },
    {
      "epoch": 0.32764128248609853,
      "grad_norm": 3.9176294803619385,
      "learning_rate": 6.862018998550959e-07,
      "loss": 0.0343,
      "step": 2077
    },
    {
      "epoch": 0.3277990298536893,
      "grad_norm": 6.870373249053955,
      "learning_rate": 6.860408951859604e-07,
      "loss": 0.0602,
      "step": 2078
    },
    {
      "epoch": 0.32795677722128014,
      "grad_norm": 4.5205817222595215,
      "learning_rate": 6.85879890516825e-07,
      "loss": 0.0696,
      "step": 2079
    },
    {
      "epoch": 0.3281145245888709,
      "grad_norm": 4.577205657958984,
      "learning_rate": 6.857188858476896e-07,
      "loss": 0.0316,
      "step": 2080
    },
    {
      "epoch": 0.32827227195646175,
      "grad_norm": 10.280963897705078,
      "learning_rate": 6.855578811785542e-07,
      "loss": 0.0953,
      "step": 2081
    },
    {
      "epoch": 0.3284300193240525,
      "grad_norm": 6.28256368637085,
      "learning_rate": 6.853968765094187e-07,
      "loss": 0.0338,
      "step": 2082
    },
    {
      "epoch": 0.32858776669164336,
      "grad_norm": 9.391764640808105,
      "learning_rate": 6.852358718402833e-07,
      "loss": 0.0908,
      "step": 2083
    },
    {
      "epoch": 0.32874551405923413,
      "grad_norm": 5.13310432434082,
      "learning_rate": 6.850748671711479e-07,
      "loss": 0.0657,
      "step": 2084
    },
    {
      "epoch": 0.32890326142682497,
      "grad_norm": 2.295919179916382,
      "learning_rate": 6.849138625020126e-07,
      "loss": 0.0181,
      "step": 2085
    },
    {
      "epoch": 0.32906100879441574,
      "grad_norm": 9.700862884521484,
      "learning_rate": 6.847528578328772e-07,
      "loss": 0.042,
      "step": 2086
    },
    {
      "epoch": 0.3292187561620065,
      "grad_norm": 15.523784637451172,
      "learning_rate": 6.845918531637417e-07,
      "loss": 0.111,
      "step": 2087
    },
    {
      "epoch": 0.32937650352959735,
      "grad_norm": 9.510817527770996,
      "learning_rate": 6.844308484946063e-07,
      "loss": 0.0721,
      "step": 2088
    },
    {
      "epoch": 0.32953425089718813,
      "grad_norm": 3.8640923500061035,
      "learning_rate": 6.842698438254709e-07,
      "loss": 0.0596,
      "step": 2089
    },
    {
      "epoch": 0.32969199826477896,
      "grad_norm": 10.860950469970703,
      "learning_rate": 6.841088391563355e-07,
      "loss": 0.0594,
      "step": 2090
    },
    {
      "epoch": 0.32984974563236974,
      "grad_norm": 11.403396606445312,
      "learning_rate": 6.839478344872e-07,
      "loss": 0.0427,
      "step": 2091
    },
    {
      "epoch": 0.33000749299996057,
      "grad_norm": 3.601780652999878,
      "learning_rate": 6.837868298180647e-07,
      "loss": 0.0308,
      "step": 2092
    },
    {
      "epoch": 0.33016524036755135,
      "grad_norm": 1.7029801607131958,
      "learning_rate": 6.836258251489293e-07,
      "loss": 0.0165,
      "step": 2093
    },
    {
      "epoch": 0.3303229877351422,
      "grad_norm": 4.657985687255859,
      "learning_rate": 6.834648204797939e-07,
      "loss": 0.0248,
      "step": 2094
    },
    {
      "epoch": 0.33048073510273296,
      "grad_norm": 8.094704627990723,
      "learning_rate": 6.833038158106584e-07,
      "loss": 0.0859,
      "step": 2095
    },
    {
      "epoch": 0.3306384824703238,
      "grad_norm": 4.248706340789795,
      "learning_rate": 6.831428111415231e-07,
      "loss": 0.0423,
      "step": 2096
    },
    {
      "epoch": 0.33079622983791457,
      "grad_norm": 4.41000509262085,
      "learning_rate": 6.829818064723877e-07,
      "loss": 0.0379,
      "step": 2097
    },
    {
      "epoch": 0.3309539772055054,
      "grad_norm": 7.829663276672363,
      "learning_rate": 6.828208018032523e-07,
      "loss": 0.0746,
      "step": 2098
    },
    {
      "epoch": 0.3311117245730962,
      "grad_norm": 4.953271865844727,
      "learning_rate": 6.826597971341168e-07,
      "loss": 0.0817,
      "step": 2099
    },
    {
      "epoch": 0.331269471940687,
      "grad_norm": 5.108656883239746,
      "learning_rate": 6.824987924649815e-07,
      "loss": 0.0614,
      "step": 2100
    },
    {
      "epoch": 0.3314272193082778,
      "grad_norm": 5.445789337158203,
      "learning_rate": 6.823377877958461e-07,
      "loss": 0.0454,
      "step": 2101
    },
    {
      "epoch": 0.3315849666758686,
      "grad_norm": 5.153439521789551,
      "learning_rate": 6.821767831267107e-07,
      "loss": 0.0689,
      "step": 2102
    },
    {
      "epoch": 0.3317427140434594,
      "grad_norm": 4.924667835235596,
      "learning_rate": 6.820157784575752e-07,
      "loss": 0.0417,
      "step": 2103
    },
    {
      "epoch": 0.33190046141105023,
      "grad_norm": 7.9560418128967285,
      "learning_rate": 6.818547737884398e-07,
      "loss": 0.0388,
      "step": 2104
    },
    {
      "epoch": 0.332058208778641,
      "grad_norm": 5.098888397216797,
      "learning_rate": 6.816937691193044e-07,
      "loss": 0.0628,
      "step": 2105
    },
    {
      "epoch": 0.3322159561462318,
      "grad_norm": 5.160165309906006,
      "learning_rate": 6.81532764450169e-07,
      "loss": 0.035,
      "step": 2106
    },
    {
      "epoch": 0.3323737035138226,
      "grad_norm": 7.469124794006348,
      "learning_rate": 6.813717597810337e-07,
      "loss": 0.0266,
      "step": 2107
    },
    {
      "epoch": 0.3325314508814134,
      "grad_norm": 8.128046035766602,
      "learning_rate": 6.812107551118982e-07,
      "loss": 0.0644,
      "step": 2108
    },
    {
      "epoch": 0.3326891982490042,
      "grad_norm": 5.858038425445557,
      "learning_rate": 6.810497504427628e-07,
      "loss": 0.0426,
      "step": 2109
    },
    {
      "epoch": 0.332846945616595,
      "grad_norm": 7.542439937591553,
      "learning_rate": 6.808887457736274e-07,
      "loss": 0.0654,
      "step": 2110
    },
    {
      "epoch": 0.33300469298418583,
      "grad_norm": 5.132564544677734,
      "learning_rate": 6.80727741104492e-07,
      "loss": 0.0339,
      "step": 2111
    },
    {
      "epoch": 0.3331624403517766,
      "grad_norm": 5.81063175201416,
      "learning_rate": 6.805667364353565e-07,
      "loss": 0.0678,
      "step": 2112
    },
    {
      "epoch": 0.33332018771936744,
      "grad_norm": 2.5806543827056885,
      "learning_rate": 6.804057317662211e-07,
      "loss": 0.03,
      "step": 2113
    },
    {
      "epoch": 0.3334779350869582,
      "grad_norm": 2.1244475841522217,
      "learning_rate": 6.802447270970858e-07,
      "loss": 0.0173,
      "step": 2114
    },
    {
      "epoch": 0.33363568245454905,
      "grad_norm": 4.667263507843018,
      "learning_rate": 6.800837224279505e-07,
      "loss": 0.0466,
      "step": 2115
    },
    {
      "epoch": 0.33379342982213983,
      "grad_norm": 7.054257392883301,
      "learning_rate": 6.79922717758815e-07,
      "loss": 0.0268,
      "step": 2116
    },
    {
      "epoch": 0.33395117718973066,
      "grad_norm": 4.43825101852417,
      "learning_rate": 6.797617130896796e-07,
      "loss": 0.0492,
      "step": 2117
    },
    {
      "epoch": 0.33410892455732144,
      "grad_norm": 5.486238479614258,
      "learning_rate": 6.796007084205442e-07,
      "loss": 0.0377,
      "step": 2118
    },
    {
      "epoch": 0.3342666719249123,
      "grad_norm": 4.388202667236328,
      "learning_rate": 6.794397037514088e-07,
      "loss": 0.0197,
      "step": 2119
    },
    {
      "epoch": 0.33442441929250305,
      "grad_norm": 6.367566108703613,
      "learning_rate": 6.792786990822733e-07,
      "loss": 0.0588,
      "step": 2120
    },
    {
      "epoch": 0.3345821666600939,
      "grad_norm": 9.597057342529297,
      "learning_rate": 6.791176944131379e-07,
      "loss": 0.0593,
      "step": 2121
    },
    {
      "epoch": 0.33473991402768466,
      "grad_norm": 3.9881293773651123,
      "learning_rate": 6.789566897440026e-07,
      "loss": 0.0538,
      "step": 2122
    },
    {
      "epoch": 0.3348976613952755,
      "grad_norm": 8.376191139221191,
      "learning_rate": 6.787956850748672e-07,
      "loss": 0.0617,
      "step": 2123
    },
    {
      "epoch": 0.33505540876286627,
      "grad_norm": 1.8899438381195068,
      "learning_rate": 6.786346804057317e-07,
      "loss": 0.0139,
      "step": 2124
    },
    {
      "epoch": 0.33521315613045705,
      "grad_norm": 8.477391242980957,
      "learning_rate": 6.784736757365963e-07,
      "loss": 0.0518,
      "step": 2125
    },
    {
      "epoch": 0.3353709034980479,
      "grad_norm": 9.249786376953125,
      "learning_rate": 6.783126710674609e-07,
      "loss": 0.0346,
      "step": 2126
    },
    {
      "epoch": 0.33552865086563866,
      "grad_norm": 8.91741943359375,
      "learning_rate": 6.781516663983255e-07,
      "loss": 0.0738,
      "step": 2127
    },
    {
      "epoch": 0.3356863982332295,
      "grad_norm": 6.2260942459106445,
      "learning_rate": 6.7799066172919e-07,
      "loss": 0.062,
      "step": 2128
    },
    {
      "epoch": 0.33584414560082027,
      "grad_norm": 6.918528079986572,
      "learning_rate": 6.778296570600546e-07,
      "loss": 0.0304,
      "step": 2129
    },
    {
      "epoch": 0.3360018929684111,
      "grad_norm": 7.220362186431885,
      "learning_rate": 6.776686523909194e-07,
      "loss": 0.0494,
      "step": 2130
    },
    {
      "epoch": 0.3361596403360019,
      "grad_norm": 3.4409945011138916,
      "learning_rate": 6.77507647721784e-07,
      "loss": 0.0577,
      "step": 2131
    },
    {
      "epoch": 0.3363173877035927,
      "grad_norm": 5.775416851043701,
      "learning_rate": 6.773466430526486e-07,
      "loss": 0.0523,
      "step": 2132
    },
    {
      "epoch": 0.3364751350711835,
      "grad_norm": 5.859851360321045,
      "learning_rate": 6.771856383835131e-07,
      "loss": 0.0648,
      "step": 2133
    },
    {
      "epoch": 0.3366328824387743,
      "grad_norm": 4.478068828582764,
      "learning_rate": 6.770246337143777e-07,
      "loss": 0.0491,
      "step": 2134
    },
    {
      "epoch": 0.3367906298063651,
      "grad_norm": 8.653120994567871,
      "learning_rate": 6.768636290452423e-07,
      "loss": 0.081,
      "step": 2135
    },
    {
      "epoch": 0.3369483771739559,
      "grad_norm": 6.8522868156433105,
      "learning_rate": 6.767026243761069e-07,
      "loss": 0.0527,
      "step": 2136
    },
    {
      "epoch": 0.3371061245415467,
      "grad_norm": 6.237538814544678,
      "learning_rate": 6.765416197069714e-07,
      "loss": 0.0711,
      "step": 2137
    },
    {
      "epoch": 0.33726387190913754,
      "grad_norm": 4.676800727844238,
      "learning_rate": 6.763806150378361e-07,
      "loss": 0.0688,
      "step": 2138
    },
    {
      "epoch": 0.3374216192767283,
      "grad_norm": 6.894059181213379,
      "learning_rate": 6.762196103687007e-07,
      "loss": 0.0857,
      "step": 2139
    },
    {
      "epoch": 0.33757936664431915,
      "grad_norm": 5.634133815765381,
      "learning_rate": 6.760586056995653e-07,
      "loss": 0.0701,
      "step": 2140
    },
    {
      "epoch": 0.3377371140119099,
      "grad_norm": 8.917547225952148,
      "learning_rate": 6.758976010304298e-07,
      "loss": 0.0747,
      "step": 2141
    },
    {
      "epoch": 0.33789486137950075,
      "grad_norm": 10.881660461425781,
      "learning_rate": 6.757365963612944e-07,
      "loss": 0.0441,
      "step": 2142
    },
    {
      "epoch": 0.33805260874709153,
      "grad_norm": 7.538335800170898,
      "learning_rate": 6.75575591692159e-07,
      "loss": 0.0394,
      "step": 2143
    },
    {
      "epoch": 0.3382103561146823,
      "grad_norm": 8.540990829467773,
      "learning_rate": 6.754145870230236e-07,
      "loss": 0.0725,
      "step": 2144
    },
    {
      "epoch": 0.33836810348227314,
      "grad_norm": 7.7875189781188965,
      "learning_rate": 6.752535823538882e-07,
      "loss": 0.0582,
      "step": 2145
    },
    {
      "epoch": 0.3385258508498639,
      "grad_norm": 4.958803176879883,
      "learning_rate": 6.750925776847528e-07,
      "loss": 0.0378,
      "step": 2146
    },
    {
      "epoch": 0.33868359821745475,
      "grad_norm": 5.001570224761963,
      "learning_rate": 6.749315730156174e-07,
      "loss": 0.0312,
      "step": 2147
    },
    {
      "epoch": 0.33884134558504553,
      "grad_norm": 12.070775985717773,
      "learning_rate": 6.74770568346482e-07,
      "loss": 0.0415,
      "step": 2148
    },
    {
      "epoch": 0.33899909295263636,
      "grad_norm": 9.16423225402832,
      "learning_rate": 6.746095636773466e-07,
      "loss": 0.0561,
      "step": 2149
    },
    {
      "epoch": 0.33915684032022714,
      "grad_norm": 4.144524574279785,
      "learning_rate": 6.744485590082112e-07,
      "loss": 0.0262,
      "step": 2150
    },
    {
      "epoch": 0.33931458768781797,
      "grad_norm": 12.374677658081055,
      "learning_rate": 6.742875543390758e-07,
      "loss": 0.0611,
      "step": 2151
    },
    {
      "epoch": 0.33947233505540875,
      "grad_norm": 9.37118148803711,
      "learning_rate": 6.741265496699405e-07,
      "loss": 0.0578,
      "step": 2152
    },
    {
      "epoch": 0.3396300824229996,
      "grad_norm": 9.701456069946289,
      "learning_rate": 6.739655450008051e-07,
      "loss": 0.0411,
      "step": 2153
    },
    {
      "epoch": 0.33978782979059036,
      "grad_norm": 8.668558120727539,
      "learning_rate": 6.738045403316696e-07,
      "loss": 0.0331,
      "step": 2154
    },
    {
      "epoch": 0.3399455771581812,
      "grad_norm": 4.742426872253418,
      "learning_rate": 6.736435356625342e-07,
      "loss": 0.0515,
      "step": 2155
    },
    {
      "epoch": 0.34010332452577197,
      "grad_norm": 4.369631767272949,
      "learning_rate": 6.734825309933988e-07,
      "loss": 0.0561,
      "step": 2156
    },
    {
      "epoch": 0.3402610718933628,
      "grad_norm": 7.481386661529541,
      "learning_rate": 6.733215263242634e-07,
      "loss": 0.1153,
      "step": 2157
    },
    {
      "epoch": 0.3404188192609536,
      "grad_norm": 5.711971282958984,
      "learning_rate": 6.731605216551279e-07,
      "loss": 0.0301,
      "step": 2158
    },
    {
      "epoch": 0.3405765666285444,
      "grad_norm": 4.880026340484619,
      "learning_rate": 6.729995169859925e-07,
      "loss": 0.0467,
      "step": 2159
    },
    {
      "epoch": 0.3407343139961352,
      "grad_norm": 3.742311954498291,
      "learning_rate": 6.728385123168572e-07,
      "loss": 0.0437,
      "step": 2160
    },
    {
      "epoch": 0.340892061363726,
      "grad_norm": 3.4775466918945312,
      "learning_rate": 6.726775076477218e-07,
      "loss": 0.0258,
      "step": 2161
    },
    {
      "epoch": 0.3410498087313168,
      "grad_norm": 8.224218368530273,
      "learning_rate": 6.725165029785863e-07,
      "loss": 0.0583,
      "step": 2162
    },
    {
      "epoch": 0.34120755609890757,
      "grad_norm": 5.834990501403809,
      "learning_rate": 6.723554983094509e-07,
      "loss": 0.0643,
      "step": 2163
    },
    {
      "epoch": 0.3413653034664984,
      "grad_norm": 3.2723138332366943,
      "learning_rate": 6.721944936403155e-07,
      "loss": 0.035,
      "step": 2164
    },
    {
      "epoch": 0.3415230508340892,
      "grad_norm": 4.137942790985107,
      "learning_rate": 6.720334889711801e-07,
      "loss": 0.0326,
      "step": 2165
    },
    {
      "epoch": 0.34168079820168,
      "grad_norm": 9.54454517364502,
      "learning_rate": 6.718724843020446e-07,
      "loss": 0.085,
      "step": 2166
    },
    {
      "epoch": 0.3418385455692708,
      "grad_norm": 5.521014213562012,
      "learning_rate": 6.717114796329093e-07,
      "loss": 0.0306,
      "step": 2167
    },
    {
      "epoch": 0.3419962929368616,
      "grad_norm": 5.938337802886963,
      "learning_rate": 6.71550474963774e-07,
      "loss": 0.0266,
      "step": 2168
    },
    {
      "epoch": 0.3421540403044524,
      "grad_norm": 2.8671207427978516,
      "learning_rate": 6.713894702946386e-07,
      "loss": 0.0155,
      "step": 2169
    },
    {
      "epoch": 0.34231178767204323,
      "grad_norm": 2.2926113605499268,
      "learning_rate": 6.712284656255031e-07,
      "loss": 0.0418,
      "step": 2170
    },
    {
      "epoch": 0.342469535039634,
      "grad_norm": 4.5548553466796875,
      "learning_rate": 6.710674609563677e-07,
      "loss": 0.0669,
      "step": 2171
    },
    {
      "epoch": 0.34262728240722484,
      "grad_norm": 5.34604024887085,
      "learning_rate": 6.709064562872323e-07,
      "loss": 0.0505,
      "step": 2172
    },
    {
      "epoch": 0.3427850297748156,
      "grad_norm": 5.072606563568115,
      "learning_rate": 6.707454516180969e-07,
      "loss": 0.0515,
      "step": 2173
    },
    {
      "epoch": 0.34294277714240645,
      "grad_norm": 13.048532485961914,
      "learning_rate": 6.705844469489614e-07,
      "loss": 0.0748,
      "step": 2174
    },
    {
      "epoch": 0.34310052450999723,
      "grad_norm": 6.24669075012207,
      "learning_rate": 6.704234422798261e-07,
      "loss": 0.0358,
      "step": 2175
    },
    {
      "epoch": 0.34325827187758806,
      "grad_norm": 4.102876663208008,
      "learning_rate": 6.702624376106907e-07,
      "loss": 0.0476,
      "step": 2176
    },
    {
      "epoch": 0.34341601924517884,
      "grad_norm": 3.69059681892395,
      "learning_rate": 6.701014329415553e-07,
      "loss": 0.0222,
      "step": 2177
    },
    {
      "epoch": 0.34357376661276967,
      "grad_norm": 3.0619795322418213,
      "learning_rate": 6.699404282724199e-07,
      "loss": 0.0407,
      "step": 2178
    },
    {
      "epoch": 0.34373151398036045,
      "grad_norm": 8.129076957702637,
      "learning_rate": 6.697794236032844e-07,
      "loss": 0.0876,
      "step": 2179
    },
    {
      "epoch": 0.3438892613479513,
      "grad_norm": 3.9379684925079346,
      "learning_rate": 6.69618418934149e-07,
      "loss": 0.024,
      "step": 2180
    },
    {
      "epoch": 0.34404700871554206,
      "grad_norm": 6.143450736999512,
      "learning_rate": 6.694574142650136e-07,
      "loss": 0.0589,
      "step": 2181
    },
    {
      "epoch": 0.3442047560831329,
      "grad_norm": 4.693875312805176,
      "learning_rate": 6.692964095958783e-07,
      "loss": 0.0715,
      "step": 2182
    },
    {
      "epoch": 0.34436250345072367,
      "grad_norm": 5.9264750480651855,
      "learning_rate": 6.691354049267428e-07,
      "loss": 0.0389,
      "step": 2183
    },
    {
      "epoch": 0.34452025081831444,
      "grad_norm": 5.538174629211426,
      "learning_rate": 6.689744002576075e-07,
      "loss": 0.0425,
      "step": 2184
    },
    {
      "epoch": 0.3446779981859053,
      "grad_norm": 8.8797607421875,
      "learning_rate": 6.688133955884721e-07,
      "loss": 0.0537,
      "step": 2185
    },
    {
      "epoch": 0.34483574555349605,
      "grad_norm": 5.80300760269165,
      "learning_rate": 6.686523909193367e-07,
      "loss": 0.0605,
      "step": 2186
    },
    {
      "epoch": 0.3449934929210869,
      "grad_norm": 3.030482053756714,
      "learning_rate": 6.684913862502012e-07,
      "loss": 0.0254,
      "step": 2187
    },
    {
      "epoch": 0.34515124028867766,
      "grad_norm": 3.2663259506225586,
      "learning_rate": 6.683303815810658e-07,
      "loss": 0.0215,
      "step": 2188
    },
    {
      "epoch": 0.3453089876562685,
      "grad_norm": 3.619297504425049,
      "learning_rate": 6.681693769119304e-07,
      "loss": 0.0278,
      "step": 2189
    },
    {
      "epoch": 0.3454667350238593,
      "grad_norm": 1.7165285348892212,
      "learning_rate": 6.680083722427951e-07,
      "loss": 0.0126,
      "step": 2190
    },
    {
      "epoch": 0.3456244823914501,
      "grad_norm": 5.1468892097473145,
      "learning_rate": 6.678473675736596e-07,
      "loss": 0.0386,
      "step": 2191
    },
    {
      "epoch": 0.3457822297590409,
      "grad_norm": 5.578167915344238,
      "learning_rate": 6.676863629045242e-07,
      "loss": 0.0546,
      "step": 2192
    },
    {
      "epoch": 0.3459399771266317,
      "grad_norm": 8.097320556640625,
      "learning_rate": 6.675253582353888e-07,
      "loss": 0.0392,
      "step": 2193
    },
    {
      "epoch": 0.3460977244942225,
      "grad_norm": 6.397778034210205,
      "learning_rate": 6.673643535662534e-07,
      "loss": 0.0714,
      "step": 2194
    },
    {
      "epoch": 0.3462554718618133,
      "grad_norm": 7.528390407562256,
      "learning_rate": 6.672033488971179e-07,
      "loss": 0.0277,
      "step": 2195
    },
    {
      "epoch": 0.3464132192294041,
      "grad_norm": 7.353246212005615,
      "learning_rate": 6.670423442279825e-07,
      "loss": 0.0518,
      "step": 2196
    },
    {
      "epoch": 0.34657096659699493,
      "grad_norm": 5.379136562347412,
      "learning_rate": 6.668813395588471e-07,
      "loss": 0.0307,
      "step": 2197
    },
    {
      "epoch": 0.3467287139645857,
      "grad_norm": 7.021356582641602,
      "learning_rate": 6.667203348897118e-07,
      "loss": 0.0911,
      "step": 2198
    },
    {
      "epoch": 0.34688646133217654,
      "grad_norm": 8.127105712890625,
      "learning_rate": 6.665593302205764e-07,
      "loss": 0.086,
      "step": 2199
    },
    {
      "epoch": 0.3470442086997673,
      "grad_norm": 7.807665824890137,
      "learning_rate": 6.663983255514409e-07,
      "loss": 0.072,
      "step": 2200
    },
    {
      "epoch": 0.34720195606735815,
      "grad_norm": 5.4244184494018555,
      "learning_rate": 6.662373208823056e-07,
      "loss": 0.0761,
      "step": 2201
    },
    {
      "epoch": 0.34735970343494893,
      "grad_norm": 2.1045219898223877,
      "learning_rate": 6.660763162131702e-07,
      "loss": 0.0111,
      "step": 2202
    },
    {
      "epoch": 0.3475174508025397,
      "grad_norm": 6.108265399932861,
      "learning_rate": 6.659153115440348e-07,
      "loss": 0.0861,
      "step": 2203
    },
    {
      "epoch": 0.34767519817013054,
      "grad_norm": 3.343996524810791,
      "learning_rate": 6.657543068748993e-07,
      "loss": 0.0218,
      "step": 2204
    },
    {
      "epoch": 0.3478329455377213,
      "grad_norm": 4.500611305236816,
      "learning_rate": 6.65593302205764e-07,
      "loss": 0.0495,
      "step": 2205
    },
    {
      "epoch": 0.34799069290531215,
      "grad_norm": 7.43673038482666,
      "learning_rate": 6.654322975366286e-07,
      "loss": 0.0949,
      "step": 2206
    },
    {
      "epoch": 0.3481484402729029,
      "grad_norm": 10.482646942138672,
      "learning_rate": 6.652712928674932e-07,
      "loss": 0.1303,
      "step": 2207
    },
    {
      "epoch": 0.34830618764049376,
      "grad_norm": 11.60330867767334,
      "learning_rate": 6.651102881983577e-07,
      "loss": 0.1006,
      "step": 2208
    },
    {
      "epoch": 0.34846393500808454,
      "grad_norm": 5.609463691711426,
      "learning_rate": 6.649492835292223e-07,
      "loss": 0.0344,
      "step": 2209
    },
    {
      "epoch": 0.34862168237567537,
      "grad_norm": 6.145679473876953,
      "learning_rate": 6.647882788600869e-07,
      "loss": 0.0276,
      "step": 2210
    },
    {
      "epoch": 0.34877942974326615,
      "grad_norm": 6.1248369216918945,
      "learning_rate": 6.646272741909515e-07,
      "loss": 0.0551,
      "step": 2211
    },
    {
      "epoch": 0.348937177110857,
      "grad_norm": 5.781682968139648,
      "learning_rate": 6.64466269521816e-07,
      "loss": 0.0471,
      "step": 2212
    },
    {
      "epoch": 0.34909492447844775,
      "grad_norm": 6.478988170623779,
      "learning_rate": 6.643052648526807e-07,
      "loss": 0.0834,
      "step": 2213
    },
    {
      "epoch": 0.3492526718460386,
      "grad_norm": 4.307476043701172,
      "learning_rate": 6.641442601835453e-07,
      "loss": 0.054,
      "step": 2214
    },
    {
      "epoch": 0.34941041921362936,
      "grad_norm": 4.316431522369385,
      "learning_rate": 6.639832555144099e-07,
      "loss": 0.0367,
      "step": 2215
    },
    {
      "epoch": 0.3495681665812202,
      "grad_norm": 9.591572761535645,
      "learning_rate": 6.638222508452744e-07,
      "loss": 0.0952,
      "step": 2216
    },
    {
      "epoch": 0.349725913948811,
      "grad_norm": 6.154400825500488,
      "learning_rate": 6.63661246176139e-07,
      "loss": 0.0371,
      "step": 2217
    },
    {
      "epoch": 0.3498836613164018,
      "grad_norm": 4.875076770782471,
      "learning_rate": 6.635002415070036e-07,
      "loss": 0.066,
      "step": 2218
    },
    {
      "epoch": 0.3500414086839926,
      "grad_norm": 3.7643606662750244,
      "learning_rate": 6.633392368378683e-07,
      "loss": 0.0476,
      "step": 2219
    },
    {
      "epoch": 0.3501991560515834,
      "grad_norm": 4.582226753234863,
      "learning_rate": 6.631782321687329e-07,
      "loss": 0.0952,
      "step": 2220
    },
    {
      "epoch": 0.3503569034191742,
      "grad_norm": 4.8520989418029785,
      "learning_rate": 6.630172274995975e-07,
      "loss": 0.0425,
      "step": 2221
    },
    {
      "epoch": 0.35051465078676497,
      "grad_norm": 7.374439239501953,
      "learning_rate": 6.628562228304621e-07,
      "loss": 0.0993,
      "step": 2222
    },
    {
      "epoch": 0.3506723981543558,
      "grad_norm": 3.0643153190612793,
      "learning_rate": 6.626952181613267e-07,
      "loss": 0.0241,
      "step": 2223
    },
    {
      "epoch": 0.3508301455219466,
      "grad_norm": 5.558948040008545,
      "learning_rate": 6.625342134921913e-07,
      "loss": 0.0403,
      "step": 2224
    },
    {
      "epoch": 0.3509878928895374,
      "grad_norm": 7.733405590057373,
      "learning_rate": 6.623732088230558e-07,
      "loss": 0.0742,
      "step": 2225
    },
    {
      "epoch": 0.3511456402571282,
      "grad_norm": 5.296177864074707,
      "learning_rate": 6.622122041539204e-07,
      "loss": 0.0502,
      "step": 2226
    },
    {
      "epoch": 0.351303387624719,
      "grad_norm": 3.479292869567871,
      "learning_rate": 6.62051199484785e-07,
      "loss": 0.0352,
      "step": 2227
    },
    {
      "epoch": 0.3514611349923098,
      "grad_norm": 9.535543441772461,
      "learning_rate": 6.618901948156497e-07,
      "loss": 0.0658,
      "step": 2228
    },
    {
      "epoch": 0.35161888235990063,
      "grad_norm": 3.5360796451568604,
      "learning_rate": 6.617291901465142e-07,
      "loss": 0.0382,
      "step": 2229
    },
    {
      "epoch": 0.3517766297274914,
      "grad_norm": 3.81123948097229,
      "learning_rate": 6.615681854773788e-07,
      "loss": 0.0648,
      "step": 2230
    },
    {
      "epoch": 0.35193437709508224,
      "grad_norm": 4.938370227813721,
      "learning_rate": 6.614071808082434e-07,
      "loss": 0.046,
      "step": 2231
    },
    {
      "epoch": 0.352092124462673,
      "grad_norm": 7.3840179443359375,
      "learning_rate": 6.61246176139108e-07,
      "loss": 0.1033,
      "step": 2232
    },
    {
      "epoch": 0.35224987183026385,
      "grad_norm": 2.876157522201538,
      "learning_rate": 6.610851714699725e-07,
      "loss": 0.0433,
      "step": 2233
    },
    {
      "epoch": 0.3524076191978546,
      "grad_norm": 7.421253204345703,
      "learning_rate": 6.609241668008371e-07,
      "loss": 0.0561,
      "step": 2234
    },
    {
      "epoch": 0.35256536656544546,
      "grad_norm": 5.2663044929504395,
      "learning_rate": 6.607631621317018e-07,
      "loss": 0.053,
      "step": 2235
    },
    {
      "epoch": 0.35272311393303624,
      "grad_norm": 7.223364353179932,
      "learning_rate": 6.606021574625665e-07,
      "loss": 0.1221,
      "step": 2236
    },
    {
      "epoch": 0.35288086130062707,
      "grad_norm": 6.649539470672607,
      "learning_rate": 6.60441152793431e-07,
      "loss": 0.0576,
      "step": 2237
    },
    {
      "epoch": 0.35303860866821785,
      "grad_norm": 6.069492340087891,
      "learning_rate": 6.602801481242956e-07,
      "loss": 0.055,
      "step": 2238
    },
    {
      "epoch": 0.3531963560358087,
      "grad_norm": 6.612998962402344,
      "learning_rate": 6.601191434551602e-07,
      "loss": 0.0765,
      "step": 2239
    },
    {
      "epoch": 0.35335410340339946,
      "grad_norm": 4.131660461425781,
      "learning_rate": 6.599581387860248e-07,
      "loss": 0.0332,
      "step": 2240
    },
    {
      "epoch": 0.35351185077099023,
      "grad_norm": 2.3796064853668213,
      "learning_rate": 6.597971341168893e-07,
      "loss": 0.022,
      "step": 2241
    },
    {
      "epoch": 0.35366959813858106,
      "grad_norm": 7.201269149780273,
      "learning_rate": 6.596361294477539e-07,
      "loss": 0.0531,
      "step": 2242
    },
    {
      "epoch": 0.35382734550617184,
      "grad_norm": 6.009885311126709,
      "learning_rate": 6.594751247786186e-07,
      "loss": 0.0577,
      "step": 2243
    },
    {
      "epoch": 0.3539850928737627,
      "grad_norm": 7.4139299392700195,
      "learning_rate": 6.593141201094832e-07,
      "loss": 0.0682,
      "step": 2244
    },
    {
      "epoch": 0.35414284024135345,
      "grad_norm": 4.9820756912231445,
      "learning_rate": 6.591531154403478e-07,
      "loss": 0.0236,
      "step": 2245
    },
    {
      "epoch": 0.3543005876089443,
      "grad_norm": 5.673033237457275,
      "learning_rate": 6.589921107712123e-07,
      "loss": 0.0483,
      "step": 2246
    },
    {
      "epoch": 0.35445833497653506,
      "grad_norm": 3.904254198074341,
      "learning_rate": 6.588311061020769e-07,
      "loss": 0.0353,
      "step": 2247
    },
    {
      "epoch": 0.3546160823441259,
      "grad_norm": 4.530683994293213,
      "learning_rate": 6.586701014329415e-07,
      "loss": 0.0606,
      "step": 2248
    },
    {
      "epoch": 0.35477382971171667,
      "grad_norm": 4.462540149688721,
      "learning_rate": 6.585090967638061e-07,
      "loss": 0.0283,
      "step": 2249
    },
    {
      "epoch": 0.3549315770793075,
      "grad_norm": 4.321871757507324,
      "learning_rate": 6.583480920946707e-07,
      "loss": 0.0524,
      "step": 2250
    },
    {
      "epoch": 0.3550893244468983,
      "grad_norm": 2.95967435836792,
      "learning_rate": 6.581870874255353e-07,
      "loss": 0.0286,
      "step": 2251
    },
    {
      "epoch": 0.3552470718144891,
      "grad_norm": 6.752710342407227,
      "learning_rate": 6.580260827563999e-07,
      "loss": 0.0682,
      "step": 2252
    },
    {
      "epoch": 0.3554048191820799,
      "grad_norm": 7.123131275177002,
      "learning_rate": 6.578650780872646e-07,
      "loss": 0.0915,
      "step": 2253
    },
    {
      "epoch": 0.3555625665496707,
      "grad_norm": 7.508081912994385,
      "learning_rate": 6.57704073418129e-07,
      "loss": 0.0787,
      "step": 2254
    },
    {
      "epoch": 0.3557203139172615,
      "grad_norm": 12.460843086242676,
      "learning_rate": 6.575430687489937e-07,
      "loss": 0.0714,
      "step": 2255
    },
    {
      "epoch": 0.35587806128485233,
      "grad_norm": 7.170982360839844,
      "learning_rate": 6.573820640798583e-07,
      "loss": 0.0487,
      "step": 2256
    },
    {
      "epoch": 0.3560358086524431,
      "grad_norm": 5.3124165534973145,
      "learning_rate": 6.572210594107229e-07,
      "loss": 0.0449,
      "step": 2257
    },
    {
      "epoch": 0.35619355602003394,
      "grad_norm": 10.241024017333984,
      "learning_rate": 6.570600547415875e-07,
      "loss": 0.0925,
      "step": 2258
    },
    {
      "epoch": 0.3563513033876247,
      "grad_norm": 7.437023639678955,
      "learning_rate": 6.568990500724521e-07,
      "loss": 0.0456,
      "step": 2259
    },
    {
      "epoch": 0.3565090507552155,
      "grad_norm": 4.243247032165527,
      "learning_rate": 6.567380454033167e-07,
      "loss": 0.0599,
      "step": 2260
    },
    {
      "epoch": 0.35666679812280633,
      "grad_norm": 8.64215087890625,
      "learning_rate": 6.565770407341813e-07,
      "loss": 0.0651,
      "step": 2261
    },
    {
      "epoch": 0.3568245454903971,
      "grad_norm": 4.423739433288574,
      "learning_rate": 6.564160360650458e-07,
      "loss": 0.044,
      "step": 2262
    },
    {
      "epoch": 0.35698229285798794,
      "grad_norm": 4.106833457946777,
      "learning_rate": 6.562550313959104e-07,
      "loss": 0.0575,
      "step": 2263
    },
    {
      "epoch": 0.3571400402255787,
      "grad_norm": 7.103672504425049,
      "learning_rate": 6.56094026726775e-07,
      "loss": 0.0622,
      "step": 2264
    },
    {
      "epoch": 0.35729778759316955,
      "grad_norm": 2.7301411628723145,
      "learning_rate": 6.559330220576397e-07,
      "loss": 0.0235,
      "step": 2265
    },
    {
      "epoch": 0.3574555349607603,
      "grad_norm": 6.06307315826416,
      "learning_rate": 6.557720173885042e-07,
      "loss": 0.0643,
      "step": 2266
    },
    {
      "epoch": 0.35761328232835116,
      "grad_norm": 3.6312015056610107,
      "learning_rate": 6.556110127193688e-07,
      "loss": 0.0479,
      "step": 2267
    },
    {
      "epoch": 0.35777102969594193,
      "grad_norm": 5.553654670715332,
      "learning_rate": 6.554500080502334e-07,
      "loss": 0.0557,
      "step": 2268
    },
    {
      "epoch": 0.35792877706353277,
      "grad_norm": 8.613818168640137,
      "learning_rate": 6.55289003381098e-07,
      "loss": 0.0714,
      "step": 2269
    },
    {
      "epoch": 0.35808652443112354,
      "grad_norm": 4.059633255004883,
      "learning_rate": 6.551279987119626e-07,
      "loss": 0.0385,
      "step": 2270
    },
    {
      "epoch": 0.3582442717987144,
      "grad_norm": 4.890695095062256,
      "learning_rate": 6.549669940428271e-07,
      "loss": 0.0311,
      "step": 2271
    },
    {
      "epoch": 0.35840201916630515,
      "grad_norm": 5.212426662445068,
      "learning_rate": 6.548059893736918e-07,
      "loss": 0.0394,
      "step": 2272
    },
    {
      "epoch": 0.358559766533896,
      "grad_norm": 3.5437400341033936,
      "learning_rate": 6.546449847045565e-07,
      "loss": 0.0366,
      "step": 2273
    },
    {
      "epoch": 0.35871751390148676,
      "grad_norm": 24.547109603881836,
      "learning_rate": 6.544839800354211e-07,
      "loss": 0.12,
      "step": 2274
    },
    {
      "epoch": 0.3588752612690776,
      "grad_norm": 7.82403039932251,
      "learning_rate": 6.543229753662856e-07,
      "loss": 0.073,
      "step": 2275
    },
    {
      "epoch": 0.35903300863666837,
      "grad_norm": 3.1485228538513184,
      "learning_rate": 6.541619706971502e-07,
      "loss": 0.0837,
      "step": 2276
    },
    {
      "epoch": 0.3591907560042592,
      "grad_norm": 11.192998886108398,
      "learning_rate": 6.540009660280148e-07,
      "loss": 0.053,
      "step": 2277
    },
    {
      "epoch": 0.35934850337185,
      "grad_norm": 5.769438743591309,
      "learning_rate": 6.538399613588794e-07,
      "loss": 0.0845,
      "step": 2278
    },
    {
      "epoch": 0.35950625073944076,
      "grad_norm": 6.332764148712158,
      "learning_rate": 6.536789566897439e-07,
      "loss": 0.0645,
      "step": 2279
    },
    {
      "epoch": 0.3596639981070316,
      "grad_norm": 1.8677188158035278,
      "learning_rate": 6.535179520206086e-07,
      "loss": 0.0168,
      "step": 2280
    },
    {
      "epoch": 0.35982174547462237,
      "grad_norm": 2.9039602279663086,
      "learning_rate": 6.533569473514732e-07,
      "loss": 0.0204,
      "step": 2281
    },
    {
      "epoch": 0.3599794928422132,
      "grad_norm": 6.307854175567627,
      "learning_rate": 6.531959426823378e-07,
      "loss": 0.0508,
      "step": 2282
    },
    {
      "epoch": 0.360137240209804,
      "grad_norm": 4.860239028930664,
      "learning_rate": 6.530349380132023e-07,
      "loss": 0.06,
      "step": 2283
    },
    {
      "epoch": 0.3602949875773948,
      "grad_norm": 6.497222423553467,
      "learning_rate": 6.528739333440669e-07,
      "loss": 0.0523,
      "step": 2284
    },
    {
      "epoch": 0.3604527349449856,
      "grad_norm": 6.6986002922058105,
      "learning_rate": 6.527129286749315e-07,
      "loss": 0.1066,
      "step": 2285
    },
    {
      "epoch": 0.3606104823125764,
      "grad_norm": 4.625892162322998,
      "learning_rate": 6.525519240057961e-07,
      "loss": 0.0288,
      "step": 2286
    },
    {
      "epoch": 0.3607682296801672,
      "grad_norm": 6.019299507141113,
      "learning_rate": 6.523909193366606e-07,
      "loss": 0.0382,
      "step": 2287
    },
    {
      "epoch": 0.36092597704775803,
      "grad_norm": 5.150125503540039,
      "learning_rate": 6.522299146675253e-07,
      "loss": 0.0304,
      "step": 2288
    },
    {
      "epoch": 0.3610837244153488,
      "grad_norm": 9.732718467712402,
      "learning_rate": 6.5206890999839e-07,
      "loss": 0.0852,
      "step": 2289
    },
    {
      "epoch": 0.36124147178293964,
      "grad_norm": 5.909388542175293,
      "learning_rate": 6.519079053292546e-07,
      "loss": 0.0282,
      "step": 2290
    },
    {
      "epoch": 0.3613992191505304,
      "grad_norm": 5.037200450897217,
      "learning_rate": 6.517469006601191e-07,
      "loss": 0.0325,
      "step": 2291
    },
    {
      "epoch": 0.36155696651812125,
      "grad_norm": 9.236132621765137,
      "learning_rate": 6.515858959909837e-07,
      "loss": 0.0724,
      "step": 2292
    },
    {
      "epoch": 0.361714713885712,
      "grad_norm": 4.459693908691406,
      "learning_rate": 6.514248913218483e-07,
      "loss": 0.0343,
      "step": 2293
    },
    {
      "epoch": 0.36187246125330286,
      "grad_norm": 5.420177936553955,
      "learning_rate": 6.512638866527129e-07,
      "loss": 0.0282,
      "step": 2294
    },
    {
      "epoch": 0.36203020862089363,
      "grad_norm": 3.4490017890930176,
      "learning_rate": 6.511028819835776e-07,
      "loss": 0.0183,
      "step": 2295
    },
    {
      "epoch": 0.36218795598848447,
      "grad_norm": 3.956369400024414,
      "learning_rate": 6.509418773144421e-07,
      "loss": 0.0497,
      "step": 2296
    },
    {
      "epoch": 0.36234570335607524,
      "grad_norm": 6.420275688171387,
      "learning_rate": 6.507808726453067e-07,
      "loss": 0.0508,
      "step": 2297
    },
    {
      "epoch": 0.3625034507236661,
      "grad_norm": 7.002887725830078,
      "learning_rate": 6.506198679761713e-07,
      "loss": 0.039,
      "step": 2298
    },
    {
      "epoch": 0.36266119809125685,
      "grad_norm": 6.247574806213379,
      "learning_rate": 6.504588633070359e-07,
      "loss": 0.025,
      "step": 2299
    },
    {
      "epoch": 0.36281894545884763,
      "grad_norm": 3.8853237628936768,
      "learning_rate": 6.502978586379004e-07,
      "loss": 0.0901,
      "step": 2300
    },
    {
      "epoch": 0.36297669282643846,
      "grad_norm": 5.598296642303467,
      "learning_rate": 6.50136853968765e-07,
      "loss": 0.0545,
      "step": 2301
    },
    {
      "epoch": 0.36313444019402924,
      "grad_norm": 3.694289445877075,
      "learning_rate": 6.499758492996296e-07,
      "loss": 0.0464,
      "step": 2302
    },
    {
      "epoch": 0.3632921875616201,
      "grad_norm": 6.625827312469482,
      "learning_rate": 6.498148446304943e-07,
      "loss": 0.0307,
      "step": 2303
    },
    {
      "epoch": 0.36344993492921085,
      "grad_norm": 3.9422924518585205,
      "learning_rate": 6.496538399613588e-07,
      "loss": 0.0507,
      "step": 2304
    },
    {
      "epoch": 0.36344993492921085,
      "eval_accuracy": 0.9849744891671391,
      "eval_f1": 0.9849744891671391,
      "eval_loss": 0.05127612501382828,
      "eval_runtime": 4719.973,
      "eval_samples_per_second": 42.978,
      "eval_steps_per_second": 2.686,
      "step": 2304
    },
    {
      "epoch": 0.3636076822968017,
      "grad_norm": 6.840505599975586,
      "learning_rate": 6.494928352922234e-07,
      "loss": 0.1002,
      "step": 2305
    },
    {
      "epoch": 0.36376542966439246,
      "grad_norm": 3.1464266777038574,
      "learning_rate": 6.49331830623088e-07,
      "loss": 0.0337,
      "step": 2306
    },
    {
      "epoch": 0.3639231770319833,
      "grad_norm": 5.0393595695495605,
      "learning_rate": 6.491708259539527e-07,
      "loss": 0.0617,
      "step": 2307
    },
    {
      "epoch": 0.36408092439957407,
      "grad_norm": 6.115160942077637,
      "learning_rate": 6.490098212848172e-07,
      "loss": 0.0848,
      "step": 2308
    },
    {
      "epoch": 0.3642386717671649,
      "grad_norm": 4.03957462310791,
      "learning_rate": 6.488488166156818e-07,
      "loss": 0.0258,
      "step": 2309
    },
    {
      "epoch": 0.3643964191347557,
      "grad_norm": 4.076841831207275,
      "learning_rate": 6.486878119465465e-07,
      "loss": 0.0319,
      "step": 2310
    },
    {
      "epoch": 0.3645541665023465,
      "grad_norm": 2.939751386642456,
      "learning_rate": 6.485268072774111e-07,
      "loss": 0.0257,
      "step": 2311
    },
    {
      "epoch": 0.3647119138699373,
      "grad_norm": 8.846038818359375,
      "learning_rate": 6.483658026082756e-07,
      "loss": 0.0496,
      "step": 2312
    },
    {
      "epoch": 0.3648696612375281,
      "grad_norm": 4.721747398376465,
      "learning_rate": 6.482047979391402e-07,
      "loss": 0.0591,
      "step": 2313
    },
    {
      "epoch": 0.3650274086051189,
      "grad_norm": 6.604150295257568,
      "learning_rate": 6.480437932700048e-07,
      "loss": 0.1203,
      "step": 2314
    },
    {
      "epoch": 0.36518515597270973,
      "grad_norm": 5.340870380401611,
      "learning_rate": 6.478827886008694e-07,
      "loss": 0.0305,
      "step": 2315
    },
    {
      "epoch": 0.3653429033403005,
      "grad_norm": 5.70967435836792,
      "learning_rate": 6.47721783931734e-07,
      "loss": 0.0727,
      "step": 2316
    },
    {
      "epoch": 0.36550065070789134,
      "grad_norm": 7.697112083435059,
      "learning_rate": 6.475607792625985e-07,
      "loss": 0.0524,
      "step": 2317
    },
    {
      "epoch": 0.3656583980754821,
      "grad_norm": 3.6987898349761963,
      "learning_rate": 6.473997745934632e-07,
      "loss": 0.0369,
      "step": 2318
    },
    {
      "epoch": 0.3658161454430729,
      "grad_norm": 5.392115592956543,
      "learning_rate": 6.472387699243278e-07,
      "loss": 0.0531,
      "step": 2319
    },
    {
      "epoch": 0.3659738928106637,
      "grad_norm": 6.497808456420898,
      "learning_rate": 6.470777652551924e-07,
      "loss": 0.0477,
      "step": 2320
    },
    {
      "epoch": 0.3661316401782545,
      "grad_norm": 4.177874565124512,
      "learning_rate": 6.469167605860569e-07,
      "loss": 0.0568,
      "step": 2321
    },
    {
      "epoch": 0.36628938754584534,
      "grad_norm": 8.207234382629395,
      "learning_rate": 6.467557559169215e-07,
      "loss": 0.0627,
      "step": 2322
    },
    {
      "epoch": 0.3664471349134361,
      "grad_norm": 6.176510810852051,
      "learning_rate": 6.465947512477861e-07,
      "loss": 0.0439,
      "step": 2323
    },
    {
      "epoch": 0.36660488228102694,
      "grad_norm": 3.681514024734497,
      "learning_rate": 6.464337465786508e-07,
      "loss": 0.0454,
      "step": 2324
    },
    {
      "epoch": 0.3667626296486177,
      "grad_norm": 6.141844272613525,
      "learning_rate": 6.462727419095154e-07,
      "loss": 0.0576,
      "step": 2325
    },
    {
      "epoch": 0.36692037701620855,
      "grad_norm": 3.4974746704101562,
      "learning_rate": 6.4611173724038e-07,
      "loss": 0.0504,
      "step": 2326
    },
    {
      "epoch": 0.36707812438379933,
      "grad_norm": 2.7725894451141357,
      "learning_rate": 6.459507325712446e-07,
      "loss": 0.0173,
      "step": 2327
    },
    {
      "epoch": 0.36723587175139016,
      "grad_norm": 9.339814186096191,
      "learning_rate": 6.457897279021092e-07,
      "loss": 0.1016,
      "step": 2328
    },
    {
      "epoch": 0.36739361911898094,
      "grad_norm": 5.565303802490234,
      "learning_rate": 6.456287232329737e-07,
      "loss": 0.0458,
      "step": 2329
    },
    {
      "epoch": 0.3675513664865718,
      "grad_norm": 4.297898292541504,
      "learning_rate": 6.454677185638383e-07,
      "loss": 0.0355,
      "step": 2330
    },
    {
      "epoch": 0.36770911385416255,
      "grad_norm": 4.813329219818115,
      "learning_rate": 6.453067138947029e-07,
      "loss": 0.0243,
      "step": 2331
    },
    {
      "epoch": 0.3678668612217534,
      "grad_norm": 5.359735488891602,
      "learning_rate": 6.451457092255675e-07,
      "loss": 0.0619,
      "step": 2332
    },
    {
      "epoch": 0.36802460858934416,
      "grad_norm": 4.5004563331604,
      "learning_rate": 6.449847045564321e-07,
      "loss": 0.0238,
      "step": 2333
    },
    {
      "epoch": 0.368182355956935,
      "grad_norm": 3.844249725341797,
      "learning_rate": 6.448236998872967e-07,
      "loss": 0.0561,
      "step": 2334
    },
    {
      "epoch": 0.36834010332452577,
      "grad_norm": 4.186479091644287,
      "learning_rate": 6.446626952181613e-07,
      "loss": 0.0609,
      "step": 2335
    },
    {
      "epoch": 0.3684978506921166,
      "grad_norm": 4.213637351989746,
      "learning_rate": 6.445016905490259e-07,
      "loss": 0.0343,
      "step": 2336
    },
    {
      "epoch": 0.3686555980597074,
      "grad_norm": 6.416787624359131,
      "learning_rate": 6.443406858798904e-07,
      "loss": 0.0787,
      "step": 2337
    },
    {
      "epoch": 0.36881334542729816,
      "grad_norm": 7.132354259490967,
      "learning_rate": 6.44179681210755e-07,
      "loss": 0.0677,
      "step": 2338
    },
    {
      "epoch": 0.368971092794889,
      "grad_norm": 9.036219596862793,
      "learning_rate": 6.440186765416196e-07,
      "loss": 0.0536,
      "step": 2339
    },
    {
      "epoch": 0.36912884016247977,
      "grad_norm": 3.9812729358673096,
      "learning_rate": 6.438576718724843e-07,
      "loss": 0.0452,
      "step": 2340
    },
    {
      "epoch": 0.3692865875300706,
      "grad_norm": 7.19404935836792,
      "learning_rate": 6.43696667203349e-07,
      "loss": 0.0442,
      "step": 2341
    },
    {
      "epoch": 0.3694443348976614,
      "grad_norm": 6.11970329284668,
      "learning_rate": 6.435356625342135e-07,
      "loss": 0.06,
      "step": 2342
    },
    {
      "epoch": 0.3696020822652522,
      "grad_norm": 3.8445143699645996,
      "learning_rate": 6.433746578650781e-07,
      "loss": 0.0497,
      "step": 2343
    },
    {
      "epoch": 0.369759829632843,
      "grad_norm": 5.607524394989014,
      "learning_rate": 6.432136531959427e-07,
      "loss": 0.0453,
      "step": 2344
    },
    {
      "epoch": 0.3699175770004338,
      "grad_norm": 6.61794376373291,
      "learning_rate": 6.430526485268073e-07,
      "loss": 0.0903,
      "step": 2345
    },
    {
      "epoch": 0.3700753243680246,
      "grad_norm": 6.52325963973999,
      "learning_rate": 6.428916438576718e-07,
      "loss": 0.0769,
      "step": 2346
    },
    {
      "epoch": 0.3702330717356154,
      "grad_norm": 4.8962082862854,
      "learning_rate": 6.427306391885364e-07,
      "loss": 0.0378,
      "step": 2347
    },
    {
      "epoch": 0.3703908191032062,
      "grad_norm": 4.14684534072876,
      "learning_rate": 6.425696345194011e-07,
      "loss": 0.0498,
      "step": 2348
    },
    {
      "epoch": 0.37054856647079704,
      "grad_norm": 6.0636820793151855,
      "learning_rate": 6.424086298502657e-07,
      "loss": 0.0654,
      "step": 2349
    },
    {
      "epoch": 0.3707063138383878,
      "grad_norm": 5.343741416931152,
      "learning_rate": 6.422476251811302e-07,
      "loss": 0.0331,
      "step": 2350
    },
    {
      "epoch": 0.37086406120597865,
      "grad_norm": 2.8783695697784424,
      "learning_rate": 6.420866205119948e-07,
      "loss": 0.0281,
      "step": 2351
    },
    {
      "epoch": 0.3710218085735694,
      "grad_norm": 2.5364882946014404,
      "learning_rate": 6.419256158428594e-07,
      "loss": 0.0251,
      "step": 2352
    },
    {
      "epoch": 0.37117955594116026,
      "grad_norm": 4.4716033935546875,
      "learning_rate": 6.41764611173724e-07,
      "loss": 0.0267,
      "step": 2353
    },
    {
      "epoch": 0.37133730330875103,
      "grad_norm": 7.375990390777588,
      "learning_rate": 6.416036065045885e-07,
      "loss": 0.0686,
      "step": 2354
    },
    {
      "epoch": 0.37149505067634186,
      "grad_norm": 2.959165096282959,
      "learning_rate": 6.414426018354532e-07,
      "loss": 0.0246,
      "step": 2355
    },
    {
      "epoch": 0.37165279804393264,
      "grad_norm": 5.672486305236816,
      "learning_rate": 6.412815971663178e-07,
      "loss": 0.0935,
      "step": 2356
    },
    {
      "epoch": 0.3718105454115234,
      "grad_norm": 3.0972232818603516,
      "learning_rate": 6.411205924971824e-07,
      "loss": 0.0622,
      "step": 2357
    },
    {
      "epoch": 0.37196829277911425,
      "grad_norm": 5.868617534637451,
      "learning_rate": 6.409595878280469e-07,
      "loss": 0.0666,
      "step": 2358
    },
    {
      "epoch": 0.37212604014670503,
      "grad_norm": 1.829486608505249,
      "learning_rate": 6.407985831589115e-07,
      "loss": 0.0112,
      "step": 2359
    },
    {
      "epoch": 0.37228378751429586,
      "grad_norm": 3.74314022064209,
      "learning_rate": 6.406375784897762e-07,
      "loss": 0.044,
      "step": 2360
    },
    {
      "epoch": 0.37244153488188664,
      "grad_norm": 4.790905475616455,
      "learning_rate": 6.404765738206408e-07,
      "loss": 0.0469,
      "step": 2361
    },
    {
      "epoch": 0.37259928224947747,
      "grad_norm": 4.5560760498046875,
      "learning_rate": 6.403155691515054e-07,
      "loss": 0.0396,
      "step": 2362
    },
    {
      "epoch": 0.37275702961706825,
      "grad_norm": 3.961839437484741,
      "learning_rate": 6.4015456448237e-07,
      "loss": 0.0653,
      "step": 2363
    },
    {
      "epoch": 0.3729147769846591,
      "grad_norm": 6.641368389129639,
      "learning_rate": 6.399935598132346e-07,
      "loss": 0.0646,
      "step": 2364
    },
    {
      "epoch": 0.37307252435224986,
      "grad_norm": 6.974590301513672,
      "learning_rate": 6.398325551440992e-07,
      "loss": 0.0984,
      "step": 2365
    },
    {
      "epoch": 0.3732302717198407,
      "grad_norm": 7.7075395584106445,
      "learning_rate": 6.396715504749638e-07,
      "loss": 0.031,
      "step": 2366
    },
    {
      "epoch": 0.37338801908743147,
      "grad_norm": 6.126717567443848,
      "learning_rate": 6.395105458058283e-07,
      "loss": 0.0444,
      "step": 2367
    },
    {
      "epoch": 0.3735457664550223,
      "grad_norm": 11.811305046081543,
      "learning_rate": 6.393495411366929e-07,
      "loss": 0.0598,
      "step": 2368
    },
    {
      "epoch": 0.3737035138226131,
      "grad_norm": 4.188122272491455,
      "learning_rate": 6.391885364675575e-07,
      "loss": 0.0436,
      "step": 2369
    },
    {
      "epoch": 0.3738612611902039,
      "grad_norm": 4.03995943069458,
      "learning_rate": 6.390275317984222e-07,
      "loss": 0.0401,
      "step": 2370
    },
    {
      "epoch": 0.3740190085577947,
      "grad_norm": 8.820528030395508,
      "learning_rate": 6.388665271292867e-07,
      "loss": 0.063,
      "step": 2371
    },
    {
      "epoch": 0.3741767559253855,
      "grad_norm": 10.953715324401855,
      "learning_rate": 6.387055224601513e-07,
      "loss": 0.0597,
      "step": 2372
    },
    {
      "epoch": 0.3743345032929763,
      "grad_norm": 4.182105541229248,
      "learning_rate": 6.385445177910159e-07,
      "loss": 0.0396,
      "step": 2373
    },
    {
      "epoch": 0.3744922506605671,
      "grad_norm": 5.633617877960205,
      "learning_rate": 6.383835131218805e-07,
      "loss": 0.0612,
      "step": 2374
    },
    {
      "epoch": 0.3746499980281579,
      "grad_norm": 6.203413963317871,
      "learning_rate": 6.38222508452745e-07,
      "loss": 0.0568,
      "step": 2375
    },
    {
      "epoch": 0.3748077453957487,
      "grad_norm": 4.6891584396362305,
      "learning_rate": 6.380615037836096e-07,
      "loss": 0.0448,
      "step": 2376
    },
    {
      "epoch": 0.3749654927633395,
      "grad_norm": 3.4268014430999756,
      "learning_rate": 6.379004991144743e-07,
      "loss": 0.0294,
      "step": 2377
    },
    {
      "epoch": 0.3751232401309303,
      "grad_norm": 6.151249408721924,
      "learning_rate": 6.37739494445339e-07,
      "loss": 0.0439,
      "step": 2378
    },
    {
      "epoch": 0.3752809874985211,
      "grad_norm": 5.077561855316162,
      "learning_rate": 6.375784897762035e-07,
      "loss": 0.0379,
      "step": 2379
    },
    {
      "epoch": 0.3754387348661119,
      "grad_norm": 7.715423583984375,
      "learning_rate": 6.374174851070681e-07,
      "loss": 0.032,
      "step": 2380
    },
    {
      "epoch": 0.37559648223370273,
      "grad_norm": 5.999269962310791,
      "learning_rate": 6.372564804379327e-07,
      "loss": 0.052,
      "step": 2381
    },
    {
      "epoch": 0.3757542296012935,
      "grad_norm": 6.657078742980957,
      "learning_rate": 6.370954757687973e-07,
      "loss": 0.072,
      "step": 2382
    },
    {
      "epoch": 0.37591197696888434,
      "grad_norm": 6.159919261932373,
      "learning_rate": 6.369344710996618e-07,
      "loss": 0.0386,
      "step": 2383
    },
    {
      "epoch": 0.3760697243364751,
      "grad_norm": 8.939737319946289,
      "learning_rate": 6.367734664305264e-07,
      "loss": 0.0474,
      "step": 2384
    },
    {
      "epoch": 0.37622747170406595,
      "grad_norm": 3.6807610988616943,
      "learning_rate": 6.366124617613911e-07,
      "loss": 0.026,
      "step": 2385
    },
    {
      "epoch": 0.37638521907165673,
      "grad_norm": 5.3889665603637695,
      "learning_rate": 6.364514570922557e-07,
      "loss": 0.0222,
      "step": 2386
    },
    {
      "epoch": 0.37654296643924756,
      "grad_norm": 5.128452777862549,
      "learning_rate": 6.362904524231203e-07,
      "loss": 0.0553,
      "step": 2387
    },
    {
      "epoch": 0.37670071380683834,
      "grad_norm": 4.838190078735352,
      "learning_rate": 6.361294477539848e-07,
      "loss": 0.047,
      "step": 2388
    },
    {
      "epoch": 0.37685846117442917,
      "grad_norm": 3.7896814346313477,
      "learning_rate": 6.359684430848494e-07,
      "loss": 0.052,
      "step": 2389
    },
    {
      "epoch": 0.37701620854201995,
      "grad_norm": 1.4873887300491333,
      "learning_rate": 6.35807438415714e-07,
      "loss": 0.0108,
      "step": 2390
    },
    {
      "epoch": 0.3771739559096108,
      "grad_norm": 5.165665626525879,
      "learning_rate": 6.356464337465786e-07,
      "loss": 0.0651,
      "step": 2391
    },
    {
      "epoch": 0.37733170327720156,
      "grad_norm": 6.901916980743408,
      "learning_rate": 6.354854290774431e-07,
      "loss": 0.0468,
      "step": 2392
    },
    {
      "epoch": 0.3774894506447924,
      "grad_norm": 3.8376920223236084,
      "learning_rate": 6.353244244083078e-07,
      "loss": 0.046,
      "step": 2393
    },
    {
      "epoch": 0.37764719801238317,
      "grad_norm": 3.073216199874878,
      "learning_rate": 6.351634197391725e-07,
      "loss": 0.056,
      "step": 2394
    },
    {
      "epoch": 0.37780494537997394,
      "grad_norm": 7.888088226318359,
      "learning_rate": 6.350024150700371e-07,
      "loss": 0.0889,
      "step": 2395
    },
    {
      "epoch": 0.3779626927475648,
      "grad_norm": 7.4682393074035645,
      "learning_rate": 6.348414104009016e-07,
      "loss": 0.0683,
      "step": 2396
    },
    {
      "epoch": 0.37812044011515555,
      "grad_norm": 5.220827102661133,
      "learning_rate": 6.346804057317662e-07,
      "loss": 0.0423,
      "step": 2397
    },
    {
      "epoch": 0.3782781874827464,
      "grad_norm": 4.824035167694092,
      "learning_rate": 6.345194010626308e-07,
      "loss": 0.0361,
      "step": 2398
    },
    {
      "epoch": 0.37843593485033716,
      "grad_norm": 3.3719117641448975,
      "learning_rate": 6.343583963934954e-07,
      "loss": 0.0251,
      "step": 2399
    },
    {
      "epoch": 0.378593682217928,
      "grad_norm": 3.443359375,
      "learning_rate": 6.341973917243599e-07,
      "loss": 0.0138,
      "step": 2400
    },
    {
      "epoch": 0.3787514295855188,
      "grad_norm": 4.2096781730651855,
      "learning_rate": 6.340363870552246e-07,
      "loss": 0.04,
      "step": 2401
    },
    {
      "epoch": 0.3789091769531096,
      "grad_norm": 5.901605606079102,
      "learning_rate": 6.338753823860892e-07,
      "loss": 0.0486,
      "step": 2402
    },
    {
      "epoch": 0.3790669243207004,
      "grad_norm": 8.777657508850098,
      "learning_rate": 6.337143777169538e-07,
      "loss": 0.047,
      "step": 2403
    },
    {
      "epoch": 0.3792246716882912,
      "grad_norm": 6.585122585296631,
      "learning_rate": 6.335533730478183e-07,
      "loss": 0.0529,
      "step": 2404
    },
    {
      "epoch": 0.379382419055882,
      "grad_norm": 7.888032913208008,
      "learning_rate": 6.333923683786829e-07,
      "loss": 0.0799,
      "step": 2405
    },
    {
      "epoch": 0.3795401664234728,
      "grad_norm": 5.162048816680908,
      "learning_rate": 6.332313637095475e-07,
      "loss": 0.0865,
      "step": 2406
    },
    {
      "epoch": 0.3796979137910636,
      "grad_norm": 4.420879364013672,
      "learning_rate": 6.330703590404121e-07,
      "loss": 0.0229,
      "step": 2407
    },
    {
      "epoch": 0.37985566115865443,
      "grad_norm": 5.623416423797607,
      "learning_rate": 6.329093543712768e-07,
      "loss": 0.0429,
      "step": 2408
    },
    {
      "epoch": 0.3800134085262452,
      "grad_norm": 5.633999824523926,
      "learning_rate": 6.327483497021413e-07,
      "loss": 0.0733,
      "step": 2409
    },
    {
      "epoch": 0.38017115589383604,
      "grad_norm": 2.145211935043335,
      "learning_rate": 6.325873450330059e-07,
      "loss": 0.0268,
      "step": 2410
    },
    {
      "epoch": 0.3803289032614268,
      "grad_norm": 7.416122913360596,
      "learning_rate": 6.324263403638705e-07,
      "loss": 0.0431,
      "step": 2411
    },
    {
      "epoch": 0.38048665062901765,
      "grad_norm": 5.949075698852539,
      "learning_rate": 6.322653356947352e-07,
      "loss": 0.0566,
      "step": 2412
    },
    {
      "epoch": 0.38064439799660843,
      "grad_norm": 4.465049743652344,
      "learning_rate": 6.321043310255997e-07,
      "loss": 0.0542,
      "step": 2413
    },
    {
      "epoch": 0.38080214536419926,
      "grad_norm": 5.267343521118164,
      "learning_rate": 6.319433263564643e-07,
      "loss": 0.0359,
      "step": 2414
    },
    {
      "epoch": 0.38095989273179004,
      "grad_norm": 14.05527114868164,
      "learning_rate": 6.31782321687329e-07,
      "loss": 0.0905,
      "step": 2415
    },
    {
      "epoch": 0.3811176400993808,
      "grad_norm": 7.6798224449157715,
      "learning_rate": 6.316213170181936e-07,
      "loss": 0.0884,
      "step": 2416
    },
    {
      "epoch": 0.38127538746697165,
      "grad_norm": 9.165621757507324,
      "learning_rate": 6.314603123490581e-07,
      "loss": 0.0417,
      "step": 2417
    },
    {
      "epoch": 0.3814331348345624,
      "grad_norm": 11.236653327941895,
      "learning_rate": 6.312993076799227e-07,
      "loss": 0.0908,
      "step": 2418
    },
    {
      "epoch": 0.38159088220215326,
      "grad_norm": 4.317572593688965,
      "learning_rate": 6.311383030107873e-07,
      "loss": 0.0681,
      "step": 2419
    },
    {
      "epoch": 0.38174862956974404,
      "grad_norm": 6.037778377532959,
      "learning_rate": 6.309772983416519e-07,
      "loss": 0.0733,
      "step": 2420
    },
    {
      "epoch": 0.38190637693733487,
      "grad_norm": 4.881504535675049,
      "learning_rate": 6.308162936725164e-07,
      "loss": 0.0721,
      "step": 2421
    },
    {
      "epoch": 0.38206412430492565,
      "grad_norm": 4.847290992736816,
      "learning_rate": 6.30655289003381e-07,
      "loss": 0.0366,
      "step": 2422
    },
    {
      "epoch": 0.3822218716725165,
      "grad_norm": 9.645156860351562,
      "learning_rate": 6.304942843342457e-07,
      "loss": 0.0441,
      "step": 2423
    },
    {
      "epoch": 0.38237961904010725,
      "grad_norm": 4.416476249694824,
      "learning_rate": 6.303332796651103e-07,
      "loss": 0.0503,
      "step": 2424
    },
    {
      "epoch": 0.3825373664076981,
      "grad_norm": 7.714351177215576,
      "learning_rate": 6.301722749959748e-07,
      "loss": 0.0625,
      "step": 2425
    },
    {
      "epoch": 0.38269511377528886,
      "grad_norm": 7.266255855560303,
      "learning_rate": 6.300112703268394e-07,
      "loss": 0.0309,
      "step": 2426
    },
    {
      "epoch": 0.3828528611428797,
      "grad_norm": 2.3878281116485596,
      "learning_rate": 6.29850265657704e-07,
      "loss": 0.0249,
      "step": 2427
    },
    {
      "epoch": 0.3830106085104705,
      "grad_norm": 6.210450649261475,
      "learning_rate": 6.296892609885686e-07,
      "loss": 0.0318,
      "step": 2428
    },
    {
      "epoch": 0.3831683558780613,
      "grad_norm": 3.793093204498291,
      "learning_rate": 6.295282563194331e-07,
      "loss": 0.0483,
      "step": 2429
    },
    {
      "epoch": 0.3833261032456521,
      "grad_norm": 5.334296703338623,
      "learning_rate": 6.293672516502979e-07,
      "loss": 0.0808,
      "step": 2430
    },
    {
      "epoch": 0.3834838506132429,
      "grad_norm": 6.60718297958374,
      "learning_rate": 6.292062469811625e-07,
      "loss": 0.0304,
      "step": 2431
    },
    {
      "epoch": 0.3836415979808337,
      "grad_norm": 3.425261974334717,
      "learning_rate": 6.290452423120271e-07,
      "loss": 0.0645,
      "step": 2432
    },
    {
      "epoch": 0.3837993453484245,
      "grad_norm": 3.1075892448425293,
      "learning_rate": 6.288842376428917e-07,
      "loss": 0.0284,
      "step": 2433
    },
    {
      "epoch": 0.3839570927160153,
      "grad_norm": 5.557161331176758,
      "learning_rate": 6.287232329737562e-07,
      "loss": 0.0459,
      "step": 2434
    },
    {
      "epoch": 0.3841148400836061,
      "grad_norm": 5.449305534362793,
      "learning_rate": 6.285622283046208e-07,
      "loss": 0.0582,
      "step": 2435
    },
    {
      "epoch": 0.3842725874511969,
      "grad_norm": 2.887645721435547,
      "learning_rate": 6.284012236354854e-07,
      "loss": 0.0209,
      "step": 2436
    },
    {
      "epoch": 0.3844303348187877,
      "grad_norm": 4.608367919921875,
      "learning_rate": 6.2824021896635e-07,
      "loss": 0.017,
      "step": 2437
    },
    {
      "epoch": 0.3845880821863785,
      "grad_norm": 10.7024564743042,
      "learning_rate": 6.280792142972146e-07,
      "loss": 0.0878,
      "step": 2438
    },
    {
      "epoch": 0.3847458295539693,
      "grad_norm": 4.104433059692383,
      "learning_rate": 6.279182096280792e-07,
      "loss": 0.0483,
      "step": 2439
    },
    {
      "epoch": 0.38490357692156013,
      "grad_norm": 7.577397346496582,
      "learning_rate": 6.277572049589438e-07,
      "loss": 0.0798,
      "step": 2440
    },
    {
      "epoch": 0.3850613242891509,
      "grad_norm": 4.732962608337402,
      "learning_rate": 6.275962002898084e-07,
      "loss": 0.0842,
      "step": 2441
    },
    {
      "epoch": 0.38521907165674174,
      "grad_norm": 5.1730852127075195,
      "learning_rate": 6.274351956206729e-07,
      "loss": 0.0453,
      "step": 2442
    },
    {
      "epoch": 0.3853768190243325,
      "grad_norm": 3.8691246509552,
      "learning_rate": 6.272741909515375e-07,
      "loss": 0.0496,
      "step": 2443
    },
    {
      "epoch": 0.38553456639192335,
      "grad_norm": 2.767871618270874,
      "learning_rate": 6.271131862824021e-07,
      "loss": 0.0443,
      "step": 2444
    },
    {
      "epoch": 0.3856923137595141,
      "grad_norm": 5.965371131896973,
      "learning_rate": 6.269521816132668e-07,
      "loss": 0.0637,
      "step": 2445
    },
    {
      "epoch": 0.38585006112710496,
      "grad_norm": 13.240874290466309,
      "learning_rate": 6.267911769441313e-07,
      "loss": 0.0599,
      "step": 2446
    },
    {
      "epoch": 0.38600780849469574,
      "grad_norm": 3.1452207565307617,
      "learning_rate": 6.26630172274996e-07,
      "loss": 0.0299,
      "step": 2447
    },
    {
      "epoch": 0.38616555586228657,
      "grad_norm": 8.212433815002441,
      "learning_rate": 6.264691676058606e-07,
      "loss": 0.0393,
      "step": 2448
    },
    {
      "epoch": 0.38632330322987735,
      "grad_norm": 6.709705829620361,
      "learning_rate": 6.263081629367252e-07,
      "loss": 0.0846,
      "step": 2449
    },
    {
      "epoch": 0.3864810505974682,
      "grad_norm": 4.421270847320557,
      "learning_rate": 6.261471582675897e-07,
      "loss": 0.0293,
      "step": 2450
    },
    {
      "epoch": 0.38663879796505896,
      "grad_norm": 4.260890483856201,
      "learning_rate": 6.259861535984543e-07,
      "loss": 0.0533,
      "step": 2451
    },
    {
      "epoch": 0.3867965453326498,
      "grad_norm": 7.362264156341553,
      "learning_rate": 6.258251489293189e-07,
      "loss": 0.0538,
      "step": 2452
    },
    {
      "epoch": 0.38695429270024057,
      "grad_norm": 6.69329309463501,
      "learning_rate": 6.256641442601836e-07,
      "loss": 0.0681,
      "step": 2453
    },
    {
      "epoch": 0.38711204006783134,
      "grad_norm": 3.216433048248291,
      "learning_rate": 6.255031395910481e-07,
      "loss": 0.0334,
      "step": 2454
    },
    {
      "epoch": 0.3872697874354222,
      "grad_norm": 4.169561386108398,
      "learning_rate": 6.253421349219127e-07,
      "loss": 0.0167,
      "step": 2455
    },
    {
      "epoch": 0.38742753480301295,
      "grad_norm": 10.017484664916992,
      "learning_rate": 6.251811302527773e-07,
      "loss": 0.09,
      "step": 2456
    },
    {
      "epoch": 0.3875852821706038,
      "grad_norm": 8.480804443359375,
      "learning_rate": 6.250201255836419e-07,
      "loss": 0.0646,
      "step": 2457
    },
    {
      "epoch": 0.38774302953819456,
      "grad_norm": 6.587338447570801,
      "learning_rate": 6.248591209145065e-07,
      "loss": 0.0596,
      "step": 2458
    },
    {
      "epoch": 0.3879007769057854,
      "grad_norm": 4.2713751792907715,
      "learning_rate": 6.24698116245371e-07,
      "loss": 0.054,
      "step": 2459
    },
    {
      "epoch": 0.38805852427337617,
      "grad_norm": 3.821355104446411,
      "learning_rate": 6.245371115762357e-07,
      "loss": 0.0197,
      "step": 2460
    },
    {
      "epoch": 0.388216271640967,
      "grad_norm": 4.111475467681885,
      "learning_rate": 6.243761069071003e-07,
      "loss": 0.0562,
      "step": 2461
    },
    {
      "epoch": 0.3883740190085578,
      "grad_norm": 3.1913421154022217,
      "learning_rate": 6.242151022379649e-07,
      "loss": 0.021,
      "step": 2462
    },
    {
      "epoch": 0.3885317663761486,
      "grad_norm": 4.067933559417725,
      "learning_rate": 6.240540975688294e-07,
      "loss": 0.0314,
      "step": 2463
    },
    {
      "epoch": 0.3886895137437394,
      "grad_norm": 6.808985233306885,
      "learning_rate": 6.23893092899694e-07,
      "loss": 0.112,
      "step": 2464
    },
    {
      "epoch": 0.3888472611113302,
      "grad_norm": 8.162635803222656,
      "learning_rate": 6.237320882305587e-07,
      "loss": 0.06,
      "step": 2465
    },
    {
      "epoch": 0.389005008478921,
      "grad_norm": 6.6350483894348145,
      "learning_rate": 6.235710835614233e-07,
      "loss": 0.035,
      "step": 2466
    },
    {
      "epoch": 0.38916275584651183,
      "grad_norm": 6.872002601623535,
      "learning_rate": 6.234100788922878e-07,
      "loss": 0.051,
      "step": 2467
    },
    {
      "epoch": 0.3893205032141026,
      "grad_norm": 7.840211391448975,
      "learning_rate": 6.232490742231525e-07,
      "loss": 0.0767,
      "step": 2468
    },
    {
      "epoch": 0.38947825058169344,
      "grad_norm": 3.9296979904174805,
      "learning_rate": 6.230880695540171e-07,
      "loss": 0.0416,
      "step": 2469
    },
    {
      "epoch": 0.3896359979492842,
      "grad_norm": 3.726431369781494,
      "learning_rate": 6.229270648848817e-07,
      "loss": 0.0314,
      "step": 2470
    },
    {
      "epoch": 0.38979374531687505,
      "grad_norm": 4.923587799072266,
      "learning_rate": 6.227660602157462e-07,
      "loss": 0.0323,
      "step": 2471
    },
    {
      "epoch": 0.38995149268446583,
      "grad_norm": 6.875946044921875,
      "learning_rate": 6.226050555466108e-07,
      "loss": 0.1052,
      "step": 2472
    },
    {
      "epoch": 0.3901092400520566,
      "grad_norm": 10.48278522491455,
      "learning_rate": 6.224440508774754e-07,
      "loss": 0.1129,
      "step": 2473
    },
    {
      "epoch": 0.39026698741964744,
      "grad_norm": 7.798339366912842,
      "learning_rate": 6.2228304620834e-07,
      "loss": 0.0719,
      "step": 2474
    },
    {
      "epoch": 0.3904247347872382,
      "grad_norm": 4.402515888214111,
      "learning_rate": 6.221220415392045e-07,
      "loss": 0.0881,
      "step": 2475
    },
    {
      "epoch": 0.39058248215482905,
      "grad_norm": 6.922409534454346,
      "learning_rate": 6.219610368700692e-07,
      "loss": 0.0552,
      "step": 2476
    },
    {
      "epoch": 0.3907402295224198,
      "grad_norm": 4.888563632965088,
      "learning_rate": 6.218000322009338e-07,
      "loss": 0.0481,
      "step": 2477
    },
    {
      "epoch": 0.39089797689001066,
      "grad_norm": 4.72125768661499,
      "learning_rate": 6.216390275317984e-07,
      "loss": 0.0999,
      "step": 2478
    },
    {
      "epoch": 0.39105572425760143,
      "grad_norm": 4.6083526611328125,
      "learning_rate": 6.21478022862663e-07,
      "loss": 0.0648,
      "step": 2479
    },
    {
      "epoch": 0.39121347162519227,
      "grad_norm": 2.519092559814453,
      "learning_rate": 6.213170181935275e-07,
      "loss": 0.0162,
      "step": 2480
    },
    {
      "epoch": 0.39137121899278304,
      "grad_norm": 2.008596658706665,
      "learning_rate": 6.211560135243921e-07,
      "loss": 0.0214,
      "step": 2481
    },
    {
      "epoch": 0.3915289663603739,
      "grad_norm": 3.965940237045288,
      "learning_rate": 6.209950088552567e-07,
      "loss": 0.0445,
      "step": 2482
    },
    {
      "epoch": 0.39168671372796465,
      "grad_norm": 4.958406448364258,
      "learning_rate": 6.208340041861215e-07,
      "loss": 0.0409,
      "step": 2483
    },
    {
      "epoch": 0.3918444610955555,
      "grad_norm": 7.141300678253174,
      "learning_rate": 6.20672999516986e-07,
      "loss": 0.0733,
      "step": 2484
    },
    {
      "epoch": 0.39200220846314626,
      "grad_norm": 3.656527280807495,
      "learning_rate": 6.205119948478506e-07,
      "loss": 0.0803,
      "step": 2485
    },
    {
      "epoch": 0.3921599558307371,
      "grad_norm": 8.250334739685059,
      "learning_rate": 6.203509901787152e-07,
      "loss": 0.0888,
      "step": 2486
    },
    {
      "epoch": 0.39231770319832787,
      "grad_norm": 4.159690856933594,
      "learning_rate": 6.201899855095798e-07,
      "loss": 0.0697,
      "step": 2487
    },
    {
      "epoch": 0.3924754505659187,
      "grad_norm": 4.09848690032959,
      "learning_rate": 6.200289808404443e-07,
      "loss": 0.0268,
      "step": 2488
    },
    {
      "epoch": 0.3926331979335095,
      "grad_norm": 4.453529357910156,
      "learning_rate": 6.198679761713089e-07,
      "loss": 0.0382,
      "step": 2489
    },
    {
      "epoch": 0.3927909453011003,
      "grad_norm": 7.759806156158447,
      "learning_rate": 6.197069715021736e-07,
      "loss": 0.0614,
      "step": 2490
    },
    {
      "epoch": 0.3929486926686911,
      "grad_norm": 4.390666961669922,
      "learning_rate": 6.195459668330382e-07,
      "loss": 0.064,
      "step": 2491
    },
    {
      "epoch": 0.39310644003628187,
      "grad_norm": 3.7275447845458984,
      "learning_rate": 6.193849621639027e-07,
      "loss": 0.0275,
      "step": 2492
    },
    {
      "epoch": 0.3932641874038727,
      "grad_norm": 3.584869384765625,
      "learning_rate": 6.192239574947673e-07,
      "loss": 0.0412,
      "step": 2493
    },
    {
      "epoch": 0.3934219347714635,
      "grad_norm": 2.2914278507232666,
      "learning_rate": 6.190629528256319e-07,
      "loss": 0.0227,
      "step": 2494
    },
    {
      "epoch": 0.3935796821390543,
      "grad_norm": 4.413097858428955,
      "learning_rate": 6.189019481564965e-07,
      "loss": 0.0276,
      "step": 2495
    },
    {
      "epoch": 0.3937374295066451,
      "grad_norm": 6.845357894897461,
      "learning_rate": 6.18740943487361e-07,
      "loss": 0.0717,
      "step": 2496
    },
    {
      "epoch": 0.3938951768742359,
      "grad_norm": 6.780287742614746,
      "learning_rate": 6.185799388182256e-07,
      "loss": 0.0354,
      "step": 2497
    },
    {
      "epoch": 0.3940529242418267,
      "grad_norm": 4.552005290985107,
      "learning_rate": 6.184189341490903e-07,
      "loss": 0.0259,
      "step": 2498
    },
    {
      "epoch": 0.39421067160941753,
      "grad_norm": 3.2987165451049805,
      "learning_rate": 6.18257929479955e-07,
      "loss": 0.0228,
      "step": 2499
    },
    {
      "epoch": 0.3943684189770083,
      "grad_norm": 8.419435501098633,
      "learning_rate": 6.180969248108195e-07,
      "loss": 0.0468,
      "step": 2500
    },
    {
      "epoch": 0.39452616634459914,
      "grad_norm": 3.106199026107788,
      "learning_rate": 6.179359201416841e-07,
      "loss": 0.0242,
      "step": 2501
    },
    {
      "epoch": 0.3946839137121899,
      "grad_norm": 6.093924045562744,
      "learning_rate": 6.177749154725487e-07,
      "loss": 0.0662,
      "step": 2502
    },
    {
      "epoch": 0.39484166107978075,
      "grad_norm": 7.137367248535156,
      "learning_rate": 6.176139108034133e-07,
      "loss": 0.0656,
      "step": 2503
    },
    {
      "epoch": 0.3949994084473715,
      "grad_norm": 2.205962657928467,
      "learning_rate": 6.174529061342779e-07,
      "loss": 0.0266,
      "step": 2504
    },
    {
      "epoch": 0.39515715581496236,
      "grad_norm": 5.25399923324585,
      "learning_rate": 6.172919014651424e-07,
      "loss": 0.0439,
      "step": 2505
    },
    {
      "epoch": 0.39531490318255313,
      "grad_norm": 3.3727121353149414,
      "learning_rate": 6.171308967960071e-07,
      "loss": 0.0462,
      "step": 2506
    },
    {
      "epoch": 0.39547265055014397,
      "grad_norm": 7.983486175537109,
      "learning_rate": 6.169698921268717e-07,
      "loss": 0.0919,
      "step": 2507
    },
    {
      "epoch": 0.39563039791773474,
      "grad_norm": 4.4629998207092285,
      "learning_rate": 6.168088874577363e-07,
      "loss": 0.0512,
      "step": 2508
    },
    {
      "epoch": 0.3957881452853256,
      "grad_norm": 6.60710334777832,
      "learning_rate": 6.166478827886008e-07,
      "loss": 0.06,
      "step": 2509
    },
    {
      "epoch": 0.39594589265291635,
      "grad_norm": 6.631765842437744,
      "learning_rate": 6.164868781194654e-07,
      "loss": 0.0604,
      "step": 2510
    },
    {
      "epoch": 0.39610364002050713,
      "grad_norm": 3.6337196826934814,
      "learning_rate": 6.1632587345033e-07,
      "loss": 0.0427,
      "step": 2511
    },
    {
      "epoch": 0.39626138738809796,
      "grad_norm": 3.338477849960327,
      "learning_rate": 6.161648687811946e-07,
      "loss": 0.0232,
      "step": 2512
    },
    {
      "epoch": 0.39641913475568874,
      "grad_norm": 2.910904884338379,
      "learning_rate": 6.160038641120592e-07,
      "loss": 0.0176,
      "step": 2513
    },
    {
      "epoch": 0.3965768821232796,
      "grad_norm": 3.373647689819336,
      "learning_rate": 6.158428594429238e-07,
      "loss": 0.0555,
      "step": 2514
    },
    {
      "epoch": 0.39673462949087035,
      "grad_norm": 7.375838756561279,
      "learning_rate": 6.156818547737884e-07,
      "loss": 0.0661,
      "step": 2515
    },
    {
      "epoch": 0.3968923768584612,
      "grad_norm": 7.225650310516357,
      "learning_rate": 6.15520850104653e-07,
      "loss": 0.0495,
      "step": 2516
    },
    {
      "epoch": 0.39705012422605196,
      "grad_norm": 3.6008362770080566,
      "learning_rate": 6.153598454355175e-07,
      "loss": 0.0244,
      "step": 2517
    },
    {
      "epoch": 0.3972078715936428,
      "grad_norm": 5.5542097091674805,
      "learning_rate": 6.151988407663822e-07,
      "loss": 0.0395,
      "step": 2518
    },
    {
      "epoch": 0.39736561896123357,
      "grad_norm": 7.228381156921387,
      "learning_rate": 6.150378360972468e-07,
      "loss": 0.0545,
      "step": 2519
    },
    {
      "epoch": 0.3975233663288244,
      "grad_norm": 3.6476194858551025,
      "learning_rate": 6.148768314281115e-07,
      "loss": 0.0329,
      "step": 2520
    },
    {
      "epoch": 0.3976811136964152,
      "grad_norm": 6.134904384613037,
      "learning_rate": 6.14715826758976e-07,
      "loss": 0.0756,
      "step": 2521
    },
    {
      "epoch": 0.397838861064006,
      "grad_norm": 5.202237129211426,
      "learning_rate": 6.145548220898406e-07,
      "loss": 0.0278,
      "step": 2522
    },
    {
      "epoch": 0.3979966084315968,
      "grad_norm": 2.656900644302368,
      "learning_rate": 6.143938174207052e-07,
      "loss": 0.0261,
      "step": 2523
    },
    {
      "epoch": 0.3981543557991876,
      "grad_norm": 7.788497447967529,
      "learning_rate": 6.142328127515698e-07,
      "loss": 0.1033,
      "step": 2524
    },
    {
      "epoch": 0.3983121031667784,
      "grad_norm": 5.942392349243164,
      "learning_rate": 6.140718080824344e-07,
      "loss": 0.0538,
      "step": 2525
    },
    {
      "epoch": 0.39846985053436923,
      "grad_norm": 6.155275344848633,
      "learning_rate": 6.139108034132989e-07,
      "loss": 0.025,
      "step": 2526
    },
    {
      "epoch": 0.39862759790196,
      "grad_norm": 6.200692653656006,
      "learning_rate": 6.137497987441635e-07,
      "loss": 0.031,
      "step": 2527
    },
    {
      "epoch": 0.39878534526955084,
      "grad_norm": 5.335763931274414,
      "learning_rate": 6.135887940750282e-07,
      "loss": 0.0535,
      "step": 2528
    },
    {
      "epoch": 0.3989430926371416,
      "grad_norm": 3.2778477668762207,
      "learning_rate": 6.134277894058928e-07,
      "loss": 0.0251,
      "step": 2529
    },
    {
      "epoch": 0.3991008400047324,
      "grad_norm": 5.646211624145508,
      "learning_rate": 6.132667847367573e-07,
      "loss": 0.0627,
      "step": 2530
    },
    {
      "epoch": 0.3992585873723232,
      "grad_norm": 5.945457935333252,
      "learning_rate": 6.131057800676219e-07,
      "loss": 0.0752,
      "step": 2531
    },
    {
      "epoch": 0.399416334739914,
      "grad_norm": 6.385833740234375,
      "learning_rate": 6.129447753984865e-07,
      "loss": 0.0846,
      "step": 2532
    },
    {
      "epoch": 0.39957408210750484,
      "grad_norm": 3.2346737384796143,
      "learning_rate": 6.127837707293511e-07,
      "loss": 0.0308,
      "step": 2533
    },
    {
      "epoch": 0.3997318294750956,
      "grad_norm": 6.200742721557617,
      "learning_rate": 6.126227660602156e-07,
      "loss": 0.0366,
      "step": 2534
    },
    {
      "epoch": 0.39988957684268644,
      "grad_norm": 4.204428195953369,
      "learning_rate": 6.124617613910802e-07,
      "loss": 0.0508,
      "step": 2535
    },
    {
      "epoch": 0.4000473242102772,
      "grad_norm": 3.469731330871582,
      "learning_rate": 6.12300756721945e-07,
      "loss": 0.0241,
      "step": 2536
    },
    {
      "epoch": 0.40020507157786805,
      "grad_norm": 4.995206356048584,
      "learning_rate": 6.121397520528096e-07,
      "loss": 0.0272,
      "step": 2537
    },
    {
      "epoch": 0.40036281894545883,
      "grad_norm": 4.293266296386719,
      "learning_rate": 6.119787473836741e-07,
      "loss": 0.0607,
      "step": 2538
    },
    {
      "epoch": 0.40052056631304966,
      "grad_norm": 3.424766778945923,
      "learning_rate": 6.118177427145387e-07,
      "loss": 0.0165,
      "step": 2539
    },
    {
      "epoch": 0.40067831368064044,
      "grad_norm": 10.498326301574707,
      "learning_rate": 6.116567380454033e-07,
      "loss": 0.0671,
      "step": 2540
    },
    {
      "epoch": 0.4008360610482313,
      "grad_norm": 6.873740196228027,
      "learning_rate": 6.114957333762679e-07,
      "loss": 0.0668,
      "step": 2541
    },
    {
      "epoch": 0.40099380841582205,
      "grad_norm": 10.00786018371582,
      "learning_rate": 6.113347287071324e-07,
      "loss": 0.051,
      "step": 2542
    },
    {
      "epoch": 0.4011515557834129,
      "grad_norm": 3.9746105670928955,
      "learning_rate": 6.111737240379971e-07,
      "loss": 0.033,
      "step": 2543
    },
    {
      "epoch": 0.40130930315100366,
      "grad_norm": 4.75880241394043,
      "learning_rate": 6.110127193688617e-07,
      "loss": 0.1254,
      "step": 2544
    },
    {
      "epoch": 0.4014670505185945,
      "grad_norm": 6.771790981292725,
      "learning_rate": 6.108517146997263e-07,
      "loss": 0.1063,
      "step": 2545
    },
    {
      "epoch": 0.40162479788618527,
      "grad_norm": 7.923023223876953,
      "learning_rate": 6.106907100305908e-07,
      "loss": 0.1006,
      "step": 2546
    },
    {
      "epoch": 0.4017825452537761,
      "grad_norm": 8.773269653320312,
      "learning_rate": 6.105297053614554e-07,
      "loss": 0.0558,
      "step": 2547
    },
    {
      "epoch": 0.4019402926213669,
      "grad_norm": 4.2924394607543945,
      "learning_rate": 6.1036870069232e-07,
      "loss": 0.0447,
      "step": 2548
    },
    {
      "epoch": 0.4020980399889577,
      "grad_norm": 5.82992696762085,
      "learning_rate": 6.102076960231846e-07,
      "loss": 0.052,
      "step": 2549
    },
    {
      "epoch": 0.4022557873565485,
      "grad_norm": 6.2320146560668945,
      "learning_rate": 6.100466913540493e-07,
      "loss": 0.0841,
      "step": 2550
    },
    {
      "epoch": 0.40241353472413927,
      "grad_norm": 2.127149820327759,
      "learning_rate": 6.098856866849138e-07,
      "loss": 0.0128,
      "step": 2551
    },
    {
      "epoch": 0.4025712820917301,
      "grad_norm": 6.342316150665283,
      "learning_rate": 6.097246820157785e-07,
      "loss": 0.0536,
      "step": 2552
    },
    {
      "epoch": 0.4027290294593209,
      "grad_norm": 3.6407859325408936,
      "learning_rate": 6.095636773466431e-07,
      "loss": 0.0214,
      "step": 2553
    },
    {
      "epoch": 0.4028867768269117,
      "grad_norm": 7.800268650054932,
      "learning_rate": 6.094026726775077e-07,
      "loss": 0.0891,
      "step": 2554
    },
    {
      "epoch": 0.4030445241945025,
      "grad_norm": 5.903280735015869,
      "learning_rate": 6.092416680083722e-07,
      "loss": 0.0631,
      "step": 2555
    },
    {
      "epoch": 0.4032022715620933,
      "grad_norm": 6.41977071762085,
      "learning_rate": 6.090806633392368e-07,
      "loss": 0.0454,
      "step": 2556
    },
    {
      "epoch": 0.4033600189296841,
      "grad_norm": 4.016145706176758,
      "learning_rate": 6.089196586701014e-07,
      "loss": 0.0651,
      "step": 2557
    },
    {
      "epoch": 0.4035177662972749,
      "grad_norm": 3.340258836746216,
      "learning_rate": 6.087586540009661e-07,
      "loss": 0.0342,
      "step": 2558
    },
    {
      "epoch": 0.4036755136648657,
      "grad_norm": 3.5221705436706543,
      "learning_rate": 6.085976493318306e-07,
      "loss": 0.0483,
      "step": 2559
    },
    {
      "epoch": 0.40383326103245654,
      "grad_norm": 5.156766414642334,
      "learning_rate": 6.084366446626952e-07,
      "loss": 0.0859,
      "step": 2560
    },
    {
      "epoch": 0.4039910084000473,
      "grad_norm": 7.001066207885742,
      "learning_rate": 6.082756399935598e-07,
      "loss": 0.0256,
      "step": 2561
    },
    {
      "epoch": 0.40414875576763815,
      "grad_norm": 6.314644813537598,
      "learning_rate": 6.081146353244244e-07,
      "loss": 0.0437,
      "step": 2562
    },
    {
      "epoch": 0.4043065031352289,
      "grad_norm": 4.919968128204346,
      "learning_rate": 6.079536306552889e-07,
      "loss": 0.0605,
      "step": 2563
    },
    {
      "epoch": 0.40446425050281976,
      "grad_norm": 2.208249092102051,
      "learning_rate": 6.077926259861535e-07,
      "loss": 0.0173,
      "step": 2564
    },
    {
      "epoch": 0.40462199787041053,
      "grad_norm": 5.167953014373779,
      "learning_rate": 6.076316213170181e-07,
      "loss": 0.0134,
      "step": 2565
    },
    {
      "epoch": 0.40477974523800136,
      "grad_norm": 4.2204508781433105,
      "learning_rate": 6.074706166478828e-07,
      "loss": 0.0451,
      "step": 2566
    },
    {
      "epoch": 0.40493749260559214,
      "grad_norm": 6.463189125061035,
      "learning_rate": 6.073096119787473e-07,
      "loss": 0.0691,
      "step": 2567
    },
    {
      "epoch": 0.405095239973183,
      "grad_norm": 5.235622406005859,
      "learning_rate": 6.071486073096119e-07,
      "loss": 0.0357,
      "step": 2568
    },
    {
      "epoch": 0.40525298734077375,
      "grad_norm": 5.960876941680908,
      "learning_rate": 6.069876026404765e-07,
      "loss": 0.0904,
      "step": 2569
    },
    {
      "epoch": 0.40541073470836453,
      "grad_norm": 6.063017845153809,
      "learning_rate": 6.068265979713412e-07,
      "loss": 0.0544,
      "step": 2570
    },
    {
      "epoch": 0.40556848207595536,
      "grad_norm": 7.733518600463867,
      "learning_rate": 6.066655933022058e-07,
      "loss": 0.0542,
      "step": 2571
    },
    {
      "epoch": 0.40572622944354614,
      "grad_norm": 4.373366355895996,
      "learning_rate": 6.065045886330703e-07,
      "loss": 0.0462,
      "step": 2572
    },
    {
      "epoch": 0.40588397681113697,
      "grad_norm": 5.196827411651611,
      "learning_rate": 6.06343583963935e-07,
      "loss": 0.0704,
      "step": 2573
    },
    {
      "epoch": 0.40604172417872775,
      "grad_norm": 6.645310401916504,
      "learning_rate": 6.061825792947996e-07,
      "loss": 0.0962,
      "step": 2574
    },
    {
      "epoch": 0.4061994715463186,
      "grad_norm": 5.274722576141357,
      "learning_rate": 6.060215746256642e-07,
      "loss": 0.0731,
      "step": 2575
    },
    {
      "epoch": 0.40635721891390936,
      "grad_norm": 6.908938884735107,
      "learning_rate": 6.058605699565287e-07,
      "loss": 0.1043,
      "step": 2576
    },
    {
      "epoch": 0.4065149662815002,
      "grad_norm": 5.653018951416016,
      "learning_rate": 6.056995652873933e-07,
      "loss": 0.0202,
      "step": 2577
    },
    {
      "epoch": 0.40667271364909097,
      "grad_norm": 6.999205589294434,
      "learning_rate": 6.055385606182579e-07,
      "loss": 0.0547,
      "step": 2578
    },
    {
      "epoch": 0.4068304610166818,
      "grad_norm": 4.579869270324707,
      "learning_rate": 6.053775559491225e-07,
      "loss": 0.0541,
      "step": 2579
    },
    {
      "epoch": 0.4069882083842726,
      "grad_norm": 9.910844802856445,
      "learning_rate": 6.05216551279987e-07,
      "loss": 0.0433,
      "step": 2580
    },
    {
      "epoch": 0.4071459557518634,
      "grad_norm": 4.5902838706970215,
      "learning_rate": 6.050555466108517e-07,
      "loss": 0.0377,
      "step": 2581
    },
    {
      "epoch": 0.4073037031194542,
      "grad_norm": 4.332019329071045,
      "learning_rate": 6.048945419417163e-07,
      "loss": 0.0365,
      "step": 2582
    },
    {
      "epoch": 0.407461450487045,
      "grad_norm": 3.858030319213867,
      "learning_rate": 6.047335372725809e-07,
      "loss": 0.048,
      "step": 2583
    },
    {
      "epoch": 0.4076191978546358,
      "grad_norm": 1.9246933460235596,
      "learning_rate": 6.045725326034454e-07,
      "loss": 0.0315,
      "step": 2584
    },
    {
      "epoch": 0.40777694522222663,
      "grad_norm": 4.131516456604004,
      "learning_rate": 6.0441152793431e-07,
      "loss": 0.0313,
      "step": 2585
    },
    {
      "epoch": 0.4079346925898174,
      "grad_norm": 15.895787239074707,
      "learning_rate": 6.042505232651746e-07,
      "loss": 0.0437,
      "step": 2586
    },
    {
      "epoch": 0.40809243995740824,
      "grad_norm": 3.734407424926758,
      "learning_rate": 6.040895185960392e-07,
      "loss": 0.0283,
      "step": 2587
    },
    {
      "epoch": 0.408250187324999,
      "grad_norm": 5.200179100036621,
      "learning_rate": 6.039285139269039e-07,
      "loss": 0.0238,
      "step": 2588
    },
    {
      "epoch": 0.4084079346925898,
      "grad_norm": 3.5924243927001953,
      "learning_rate": 6.037675092577685e-07,
      "loss": 0.0358,
      "step": 2589
    },
    {
      "epoch": 0.4085656820601806,
      "grad_norm": 3.209386110305786,
      "learning_rate": 6.036065045886331e-07,
      "loss": 0.0237,
      "step": 2590
    },
    {
      "epoch": 0.4087234294277714,
      "grad_norm": 3.4143080711364746,
      "learning_rate": 6.034454999194977e-07,
      "loss": 0.0161,
      "step": 2591
    },
    {
      "epoch": 0.40888117679536223,
      "grad_norm": 6.51207160949707,
      "learning_rate": 6.032844952503622e-07,
      "loss": 0.0492,
      "step": 2592
    },
    {
      "epoch": 0.409038924162953,
      "grad_norm": 9.512336730957031,
      "learning_rate": 6.031234905812268e-07,
      "loss": 0.0506,
      "step": 2593
    },
    {
      "epoch": 0.40919667153054384,
      "grad_norm": 4.175022602081299,
      "learning_rate": 6.029624859120914e-07,
      "loss": 0.0726,
      "step": 2594
    },
    {
      "epoch": 0.4093544188981346,
      "grad_norm": 3.06833553314209,
      "learning_rate": 6.02801481242956e-07,
      "loss": 0.0467,
      "step": 2595
    },
    {
      "epoch": 0.40951216626572545,
      "grad_norm": 4.335803508758545,
      "learning_rate": 6.026404765738207e-07,
      "loss": 0.053,
      "step": 2596
    },
    {
      "epoch": 0.40966991363331623,
      "grad_norm": 7.851492404937744,
      "learning_rate": 6.024794719046852e-07,
      "loss": 0.0855,
      "step": 2597
    },
    {
      "epoch": 0.40982766100090706,
      "grad_norm": 6.55589485168457,
      "learning_rate": 6.023184672355498e-07,
      "loss": 0.033,
      "step": 2598
    },
    {
      "epoch": 0.40998540836849784,
      "grad_norm": 2.5253841876983643,
      "learning_rate": 6.021574625664144e-07,
      "loss": 0.0139,
      "step": 2599
    },
    {
      "epoch": 0.41014315573608867,
      "grad_norm": 3.553863763809204,
      "learning_rate": 6.01996457897279e-07,
      "loss": 0.0537,
      "step": 2600
    },
    {
      "epoch": 0.41030090310367945,
      "grad_norm": 4.84967041015625,
      "learning_rate": 6.018354532281435e-07,
      "loss": 0.0793,
      "step": 2601
    },
    {
      "epoch": 0.4104586504712703,
      "grad_norm": 7.575808048248291,
      "learning_rate": 6.016744485590081e-07,
      "loss": 0.0572,
      "step": 2602
    },
    {
      "epoch": 0.41061639783886106,
      "grad_norm": 8.631465911865234,
      "learning_rate": 6.015134438898728e-07,
      "loss": 0.0718,
      "step": 2603
    },
    {
      "epoch": 0.4107741452064519,
      "grad_norm": 4.6790056228637695,
      "learning_rate": 6.013524392207374e-07,
      "loss": 0.0554,
      "step": 2604
    },
    {
      "epoch": 0.41093189257404267,
      "grad_norm": 4.117552280426025,
      "learning_rate": 6.01191434551602e-07,
      "loss": 0.0465,
      "step": 2605
    },
    {
      "epoch": 0.4110896399416335,
      "grad_norm": 6.920306205749512,
      "learning_rate": 6.010304298824666e-07,
      "loss": 0.0572,
      "step": 2606
    },
    {
      "epoch": 0.4112473873092243,
      "grad_norm": 3.1878037452697754,
      "learning_rate": 6.008694252133312e-07,
      "loss": 0.0251,
      "step": 2607
    },
    {
      "epoch": 0.41140513467681505,
      "grad_norm": 6.447464466094971,
      "learning_rate": 6.007084205441958e-07,
      "loss": 0.0467,
      "step": 2608
    },
    {
      "epoch": 0.4115628820444059,
      "grad_norm": 7.5503363609313965,
      "learning_rate": 6.005474158750603e-07,
      "loss": 0.0475,
      "step": 2609
    },
    {
      "epoch": 0.41172062941199666,
      "grad_norm": 4.07452917098999,
      "learning_rate": 6.003864112059249e-07,
      "loss": 0.0227,
      "step": 2610
    },
    {
      "epoch": 0.4118783767795875,
      "grad_norm": 5.6531524658203125,
      "learning_rate": 6.002254065367896e-07,
      "loss": 0.045,
      "step": 2611
    },
    {
      "epoch": 0.4120361241471783,
      "grad_norm": 4.197863578796387,
      "learning_rate": 6.000644018676542e-07,
      "loss": 0.0225,
      "step": 2612
    },
    {
      "epoch": 0.4121938715147691,
      "grad_norm": 5.841701030731201,
      "learning_rate": 5.999033971985187e-07,
      "loss": 0.0415,
      "step": 2613
    },
    {
      "epoch": 0.4123516188823599,
      "grad_norm": 3.9903364181518555,
      "learning_rate": 5.997423925293833e-07,
      "loss": 0.0393,
      "step": 2614
    },
    {
      "epoch": 0.4125093662499507,
      "grad_norm": 3.155518054962158,
      "learning_rate": 5.995813878602479e-07,
      "loss": 0.0145,
      "step": 2615
    },
    {
      "epoch": 0.4126671136175415,
      "grad_norm": 4.071169853210449,
      "learning_rate": 5.994203831911125e-07,
      "loss": 0.0456,
      "step": 2616
    },
    {
      "epoch": 0.4128248609851323,
      "grad_norm": 3.968475818634033,
      "learning_rate": 5.99259378521977e-07,
      "loss": 0.0245,
      "step": 2617
    },
    {
      "epoch": 0.4129826083527231,
      "grad_norm": 6.552299976348877,
      "learning_rate": 5.990983738528417e-07,
      "loss": 0.0944,
      "step": 2618
    },
    {
      "epoch": 0.41314035572031393,
      "grad_norm": 8.035197257995605,
      "learning_rate": 5.989373691837063e-07,
      "loss": 0.1221,
      "step": 2619
    },
    {
      "epoch": 0.4132981030879047,
      "grad_norm": 8.004145622253418,
      "learning_rate": 5.987763645145709e-07,
      "loss": 0.0546,
      "step": 2620
    },
    {
      "epoch": 0.41345585045549554,
      "grad_norm": 2.6789891719818115,
      "learning_rate": 5.986153598454355e-07,
      "loss": 0.0318,
      "step": 2621
    },
    {
      "epoch": 0.4136135978230863,
      "grad_norm": 5.315359115600586,
      "learning_rate": 5.984543551763e-07,
      "loss": 0.0578,
      "step": 2622
    },
    {
      "epoch": 0.41377134519067715,
      "grad_norm": 4.650211811065674,
      "learning_rate": 5.982933505071647e-07,
      "loss": 0.067,
      "step": 2623
    },
    {
      "epoch": 0.41392909255826793,
      "grad_norm": 4.35648250579834,
      "learning_rate": 5.981323458380293e-07,
      "loss": 0.0275,
      "step": 2624
    },
    {
      "epoch": 0.41408683992585876,
      "grad_norm": 6.478522300720215,
      "learning_rate": 5.979713411688939e-07,
      "loss": 0.0403,
      "step": 2625
    },
    {
      "epoch": 0.41424458729344954,
      "grad_norm": 5.901090621948242,
      "learning_rate": 5.978103364997585e-07,
      "loss": 0.0356,
      "step": 2626
    },
    {
      "epoch": 0.4144023346610403,
      "grad_norm": 12.616568565368652,
      "learning_rate": 5.976493318306231e-07,
      "loss": 0.0691,
      "step": 2627
    },
    {
      "epoch": 0.41456008202863115,
      "grad_norm": 5.331027984619141,
      "learning_rate": 5.974883271614877e-07,
      "loss": 0.0284,
      "step": 2628
    },
    {
      "epoch": 0.4147178293962219,
      "grad_norm": 5.250926494598389,
      "learning_rate": 5.973273224923523e-07,
      "loss": 0.0418,
      "step": 2629
    },
    {
      "epoch": 0.41487557676381276,
      "grad_norm": 4.558555603027344,
      "learning_rate": 5.971663178232168e-07,
      "loss": 0.0379,
      "step": 2630
    },
    {
      "epoch": 0.41503332413140354,
      "grad_norm": 5.668402671813965,
      "learning_rate": 5.970053131540814e-07,
      "loss": 0.0452,
      "step": 2631
    },
    {
      "epoch": 0.41519107149899437,
      "grad_norm": 5.460912227630615,
      "learning_rate": 5.96844308484946e-07,
      "loss": 0.0203,
      "step": 2632
    },
    {
      "epoch": 0.41534881886658515,
      "grad_norm": 5.2398271560668945,
      "learning_rate": 5.966833038158107e-07,
      "loss": 0.0538,
      "step": 2633
    },
    {
      "epoch": 0.415506566234176,
      "grad_norm": 11.468836784362793,
      "learning_rate": 5.965222991466752e-07,
      "loss": 0.0935,
      "step": 2634
    },
    {
      "epoch": 0.41566431360176676,
      "grad_norm": 7.559725761413574,
      "learning_rate": 5.963612944775398e-07,
      "loss": 0.0673,
      "step": 2635
    },
    {
      "epoch": 0.4158220609693576,
      "grad_norm": 5.004055976867676,
      "learning_rate": 5.962002898084044e-07,
      "loss": 0.0191,
      "step": 2636
    },
    {
      "epoch": 0.41597980833694836,
      "grad_norm": 4.78782320022583,
      "learning_rate": 5.96039285139269e-07,
      "loss": 0.0261,
      "step": 2637
    },
    {
      "epoch": 0.4161375557045392,
      "grad_norm": 8.144420623779297,
      "learning_rate": 5.958782804701335e-07,
      "loss": 0.0584,
      "step": 2638
    },
    {
      "epoch": 0.41629530307213,
      "grad_norm": 3.625196695327759,
      "learning_rate": 5.957172758009981e-07,
      "loss": 0.0481,
      "step": 2639
    },
    {
      "epoch": 0.4164530504397208,
      "grad_norm": 4.784508228302002,
      "learning_rate": 5.955562711318627e-07,
      "loss": 0.0469,
      "step": 2640
    },
    {
      "epoch": 0.4166107978073116,
      "grad_norm": 5.008841514587402,
      "learning_rate": 5.953952664627275e-07,
      "loss": 0.0434,
      "step": 2641
    },
    {
      "epoch": 0.4167685451749024,
      "grad_norm": 4.575802803039551,
      "learning_rate": 5.952342617935921e-07,
      "loss": 0.0395,
      "step": 2642
    },
    {
      "epoch": 0.4169262925424932,
      "grad_norm": 4.045619964599609,
      "learning_rate": 5.950732571244566e-07,
      "loss": 0.0249,
      "step": 2643
    },
    {
      "epoch": 0.417084039910084,
      "grad_norm": 4.226224422454834,
      "learning_rate": 5.949122524553212e-07,
      "loss": 0.0365,
      "step": 2644
    },
    {
      "epoch": 0.4172417872776748,
      "grad_norm": 9.37044906616211,
      "learning_rate": 5.947512477861858e-07,
      "loss": 0.0638,
      "step": 2645
    },
    {
      "epoch": 0.4173995346452656,
      "grad_norm": 6.7245001792907715,
      "learning_rate": 5.945902431170504e-07,
      "loss": 0.0299,
      "step": 2646
    },
    {
      "epoch": 0.4175572820128564,
      "grad_norm": 9.082123756408691,
      "learning_rate": 5.944292384479149e-07,
      "loss": 0.0668,
      "step": 2647
    },
    {
      "epoch": 0.4177150293804472,
      "grad_norm": 7.415581226348877,
      "learning_rate": 5.942682337787796e-07,
      "loss": 0.1205,
      "step": 2648
    },
    {
      "epoch": 0.417872776748038,
      "grad_norm": 2.4082369804382324,
      "learning_rate": 5.941072291096442e-07,
      "loss": 0.0129,
      "step": 2649
    },
    {
      "epoch": 0.4180305241156288,
      "grad_norm": 8.189739227294922,
      "learning_rate": 5.939462244405088e-07,
      "loss": 0.1089,
      "step": 2650
    },
    {
      "epoch": 0.41818827148321963,
      "grad_norm": 4.564749240875244,
      "learning_rate": 5.937852197713733e-07,
      "loss": 0.061,
      "step": 2651
    },
    {
      "epoch": 0.4183460188508104,
      "grad_norm": 2.6857526302337646,
      "learning_rate": 5.936242151022379e-07,
      "loss": 0.0141,
      "step": 2652
    },
    {
      "epoch": 0.41850376621840124,
      "grad_norm": 5.747800827026367,
      "learning_rate": 5.934632104331025e-07,
      "loss": 0.0419,
      "step": 2653
    },
    {
      "epoch": 0.418661513585992,
      "grad_norm": 3.823982000350952,
      "learning_rate": 5.933022057639671e-07,
      "loss": 0.0235,
      "step": 2654
    },
    {
      "epoch": 0.41881926095358285,
      "grad_norm": 6.821854591369629,
      "learning_rate": 5.931412010948316e-07,
      "loss": 0.057,
      "step": 2655
    },
    {
      "epoch": 0.4189770083211736,
      "grad_norm": 9.907840728759766,
      "learning_rate": 5.929801964256963e-07,
      "loss": 0.0684,
      "step": 2656
    },
    {
      "epoch": 0.41913475568876446,
      "grad_norm": 6.711484909057617,
      "learning_rate": 5.92819191756561e-07,
      "loss": 0.0456,
      "step": 2657
    },
    {
      "epoch": 0.41929250305635524,
      "grad_norm": 7.317268371582031,
      "learning_rate": 5.926581870874256e-07,
      "loss": 0.0809,
      "step": 2658
    },
    {
      "epoch": 0.41945025042394607,
      "grad_norm": 6.277600288391113,
      "learning_rate": 5.924971824182901e-07,
      "loss": 0.057,
      "step": 2659
    },
    {
      "epoch": 0.41960799779153685,
      "grad_norm": 5.619283676147461,
      "learning_rate": 5.923361777491547e-07,
      "loss": 0.0484,
      "step": 2660
    },
    {
      "epoch": 0.4197657451591277,
      "grad_norm": 6.379857540130615,
      "learning_rate": 5.921751730800193e-07,
      "loss": 0.0595,
      "step": 2661
    },
    {
      "epoch": 0.41992349252671846,
      "grad_norm": 8.349502563476562,
      "learning_rate": 5.920141684108839e-07,
      "loss": 0.1024,
      "step": 2662
    },
    {
      "epoch": 0.4200812398943093,
      "grad_norm": 6.646510601043701,
      "learning_rate": 5.918531637417485e-07,
      "loss": 0.0624,
      "step": 2663
    },
    {
      "epoch": 0.42023898726190007,
      "grad_norm": 10.572782516479492,
      "learning_rate": 5.916921590726131e-07,
      "loss": 0.1357,
      "step": 2664
    },
    {
      "epoch": 0.4203967346294909,
      "grad_norm": 6.063112735748291,
      "learning_rate": 5.915311544034777e-07,
      "loss": 0.0458,
      "step": 2665
    },
    {
      "epoch": 0.4205544819970817,
      "grad_norm": 3.0055274963378906,
      "learning_rate": 5.913701497343423e-07,
      "loss": 0.0322,
      "step": 2666
    },
    {
      "epoch": 0.42071222936467245,
      "grad_norm": 4.2171711921691895,
      "learning_rate": 5.912091450652069e-07,
      "loss": 0.0556,
      "step": 2667
    },
    {
      "epoch": 0.4208699767322633,
      "grad_norm": 8.459450721740723,
      "learning_rate": 5.910481403960714e-07,
      "loss": 0.0401,
      "step": 2668
    },
    {
      "epoch": 0.42102772409985406,
      "grad_norm": 2.972092866897583,
      "learning_rate": 5.90887135726936e-07,
      "loss": 0.0295,
      "step": 2669
    },
    {
      "epoch": 0.4211854714674449,
      "grad_norm": 7.867188930511475,
      "learning_rate": 5.907261310578006e-07,
      "loss": 0.07,
      "step": 2670
    },
    {
      "epoch": 0.42134321883503567,
      "grad_norm": 5.135982513427734,
      "learning_rate": 5.905651263886653e-07,
      "loss": 0.0521,
      "step": 2671
    },
    {
      "epoch": 0.4215009662026265,
      "grad_norm": 4.097264766693115,
      "learning_rate": 5.904041217195298e-07,
      "loss": 0.034,
      "step": 2672
    },
    {
      "epoch": 0.4216587135702173,
      "grad_norm": 3.2581069469451904,
      "learning_rate": 5.902431170503944e-07,
      "loss": 0.031,
      "step": 2673
    },
    {
      "epoch": 0.4218164609378081,
      "grad_norm": 4.152832508087158,
      "learning_rate": 5.90082112381259e-07,
      "loss": 0.0388,
      "step": 2674
    },
    {
      "epoch": 0.4219742083053989,
      "grad_norm": 7.619700908660889,
      "learning_rate": 5.899211077121237e-07,
      "loss": 0.0503,
      "step": 2675
    },
    {
      "epoch": 0.4221319556729897,
      "grad_norm": 3.360625743865967,
      "learning_rate": 5.897601030429882e-07,
      "loss": 0.0218,
      "step": 2676
    },
    {
      "epoch": 0.4222897030405805,
      "grad_norm": 4.949860572814941,
      "learning_rate": 5.895990983738528e-07,
      "loss": 0.0276,
      "step": 2677
    },
    {
      "epoch": 0.42244745040817133,
      "grad_norm": 5.772740840911865,
      "learning_rate": 5.894380937047175e-07,
      "loss": 0.0352,
      "step": 2678
    },
    {
      "epoch": 0.4226051977757621,
      "grad_norm": 4.35075569152832,
      "learning_rate": 5.892770890355821e-07,
      "loss": 0.0484,
      "step": 2679
    },
    {
      "epoch": 0.42276294514335294,
      "grad_norm": 2.8517985343933105,
      "learning_rate": 5.891160843664466e-07,
      "loss": 0.0601,
      "step": 2680
    },
    {
      "epoch": 0.4229206925109437,
      "grad_norm": 2.550802230834961,
      "learning_rate": 5.889550796973112e-07,
      "loss": 0.0134,
      "step": 2681
    },
    {
      "epoch": 0.42307843987853455,
      "grad_norm": 3.6672005653381348,
      "learning_rate": 5.887940750281758e-07,
      "loss": 0.0518,
      "step": 2682
    },
    {
      "epoch": 0.42323618724612533,
      "grad_norm": 6.912148475646973,
      "learning_rate": 5.886330703590404e-07,
      "loss": 0.0822,
      "step": 2683
    },
    {
      "epoch": 0.42339393461371616,
      "grad_norm": 5.84259033203125,
      "learning_rate": 5.884720656899049e-07,
      "loss": 0.0667,
      "step": 2684
    },
    {
      "epoch": 0.42355168198130694,
      "grad_norm": 4.002005100250244,
      "learning_rate": 5.883110610207695e-07,
      "loss": 0.0494,
      "step": 2685
    },
    {
      "epoch": 0.4237094293488977,
      "grad_norm": 5.479091167449951,
      "learning_rate": 5.881500563516342e-07,
      "loss": 0.0369,
      "step": 2686
    },
    {
      "epoch": 0.42386717671648855,
      "grad_norm": 4.5175652503967285,
      "learning_rate": 5.879890516824988e-07,
      "loss": 0.0381,
      "step": 2687
    },
    {
      "epoch": 0.4240249240840793,
      "grad_norm": 3.294260025024414,
      "learning_rate": 5.878280470133634e-07,
      "loss": 0.0269,
      "step": 2688
    },
    {
      "epoch": 0.42418267145167016,
      "grad_norm": 3.613359212875366,
      "learning_rate": 5.876670423442279e-07,
      "loss": 0.0457,
      "step": 2689
    },
    {
      "epoch": 0.42434041881926093,
      "grad_norm": 5.717574596405029,
      "learning_rate": 5.875060376750925e-07,
      "loss": 0.0528,
      "step": 2690
    },
    {
      "epoch": 0.42449816618685177,
      "grad_norm": 3.467639207839966,
      "learning_rate": 5.873450330059571e-07,
      "loss": 0.0303,
      "step": 2691
    },
    {
      "epoch": 0.42465591355444254,
      "grad_norm": 35.10895538330078,
      "learning_rate": 5.871840283368217e-07,
      "loss": 0.0447,
      "step": 2692
    },
    {
      "epoch": 0.4248136609220334,
      "grad_norm": 7.386102676391602,
      "learning_rate": 5.870230236676864e-07,
      "loss": 0.0437,
      "step": 2693
    },
    {
      "epoch": 0.42497140828962415,
      "grad_norm": 4.7817816734313965,
      "learning_rate": 5.86862018998551e-07,
      "loss": 0.0501,
      "step": 2694
    },
    {
      "epoch": 0.425129155657215,
      "grad_norm": 6.132143974304199,
      "learning_rate": 5.867010143294156e-07,
      "loss": 0.0751,
      "step": 2695
    },
    {
      "epoch": 0.42528690302480576,
      "grad_norm": 4.20089054107666,
      "learning_rate": 5.865400096602802e-07,
      "loss": 0.0579,
      "step": 2696
    },
    {
      "epoch": 0.4254446503923966,
      "grad_norm": 1.796299695968628,
      "learning_rate": 5.863790049911447e-07,
      "loss": 0.0098,
      "step": 2697
    },
    {
      "epoch": 0.42560239775998737,
      "grad_norm": 3.0523252487182617,
      "learning_rate": 5.862180003220093e-07,
      "loss": 0.024,
      "step": 2698
    },
    {
      "epoch": 0.4257601451275782,
      "grad_norm": 4.074103355407715,
      "learning_rate": 5.860569956528739e-07,
      "loss": 0.0611,
      "step": 2699
    },
    {
      "epoch": 0.425917892495169,
      "grad_norm": 7.769861221313477,
      "learning_rate": 5.858959909837385e-07,
      "loss": 0.1272,
      "step": 2700
    },
    {
      "epoch": 0.4260756398627598,
      "grad_norm": 6.271566867828369,
      "learning_rate": 5.857349863146031e-07,
      "loss": 0.0559,
      "step": 2701
    },
    {
      "epoch": 0.4262333872303506,
      "grad_norm": 6.888360977172852,
      "learning_rate": 5.855739816454677e-07,
      "loss": 0.0598,
      "step": 2702
    },
    {
      "epoch": 0.4263911345979414,
      "grad_norm": 8.045308113098145,
      "learning_rate": 5.854129769763323e-07,
      "loss": 0.0598,
      "step": 2703
    },
    {
      "epoch": 0.4265488819655322,
      "grad_norm": 5.838111877441406,
      "learning_rate": 5.852519723071969e-07,
      "loss": 0.051,
      "step": 2704
    },
    {
      "epoch": 0.426706629333123,
      "grad_norm": 7.2040557861328125,
      "learning_rate": 5.850909676380614e-07,
      "loss": 0.0525,
      "step": 2705
    },
    {
      "epoch": 0.4268643767007138,
      "grad_norm": 3.799769878387451,
      "learning_rate": 5.84929962968926e-07,
      "loss": 0.0349,
      "step": 2706
    },
    {
      "epoch": 0.4270221240683046,
      "grad_norm": 3.744792938232422,
      "learning_rate": 5.847689582997906e-07,
      "loss": 0.0325,
      "step": 2707
    },
    {
      "epoch": 0.4271798714358954,
      "grad_norm": 7.798043727874756,
      "learning_rate": 5.846079536306553e-07,
      "loss": 0.0347,
      "step": 2708
    },
    {
      "epoch": 0.4273376188034862,
      "grad_norm": 4.0187764167785645,
      "learning_rate": 5.844469489615198e-07,
      "loss": 0.0298,
      "step": 2709
    },
    {
      "epoch": 0.42749536617107703,
      "grad_norm": 8.834007263183594,
      "learning_rate": 5.842859442923844e-07,
      "loss": 0.034,
      "step": 2710
    },
    {
      "epoch": 0.4276531135386678,
      "grad_norm": 7.793360710144043,
      "learning_rate": 5.841249396232491e-07,
      "loss": 0.0674,
      "step": 2711
    },
    {
      "epoch": 0.42781086090625864,
      "grad_norm": 1.7781180143356323,
      "learning_rate": 5.839639349541137e-07,
      "loss": 0.0082,
      "step": 2712
    },
    {
      "epoch": 0.4279686082738494,
      "grad_norm": 6.2372941970825195,
      "learning_rate": 5.838029302849783e-07,
      "loss": 0.0319,
      "step": 2713
    },
    {
      "epoch": 0.42812635564144025,
      "grad_norm": 24.141183853149414,
      "learning_rate": 5.836419256158428e-07,
      "loss": 0.0079,
      "step": 2714
    },
    {
      "epoch": 0.428284103009031,
      "grad_norm": 13.051074028015137,
      "learning_rate": 5.834809209467074e-07,
      "loss": 0.0286,
      "step": 2715
    },
    {
      "epoch": 0.42844185037662186,
      "grad_norm": 11.537595748901367,
      "learning_rate": 5.833199162775721e-07,
      "loss": 0.0876,
      "step": 2716
    },
    {
      "epoch": 0.42859959774421263,
      "grad_norm": 3.638516664505005,
      "learning_rate": 5.831589116084367e-07,
      "loss": 0.0311,
      "step": 2717
    },
    {
      "epoch": 0.42875734511180347,
      "grad_norm": 8.744967460632324,
      "learning_rate": 5.829979069393012e-07,
      "loss": 0.0528,
      "step": 2718
    },
    {
      "epoch": 0.42891509247939424,
      "grad_norm": 12.257424354553223,
      "learning_rate": 5.828369022701658e-07,
      "loss": 0.0731,
      "step": 2719
    },
    {
      "epoch": 0.4290728398469851,
      "grad_norm": 6.0300116539001465,
      "learning_rate": 5.826758976010304e-07,
      "loss": 0.0724,
      "step": 2720
    },
    {
      "epoch": 0.42923058721457585,
      "grad_norm": 8.850271224975586,
      "learning_rate": 5.82514892931895e-07,
      "loss": 0.0757,
      "step": 2721
    },
    {
      "epoch": 0.4293883345821667,
      "grad_norm": 8.085939407348633,
      "learning_rate": 5.823538882627595e-07,
      "loss": 0.0785,
      "step": 2722
    },
    {
      "epoch": 0.42954608194975746,
      "grad_norm": 2.521941900253296,
      "learning_rate": 5.821928835936242e-07,
      "loss": 0.0375,
      "step": 2723
    },
    {
      "epoch": 0.42970382931734824,
      "grad_norm": 4.878946781158447,
      "learning_rate": 5.820318789244888e-07,
      "loss": 0.0276,
      "step": 2724
    },
    {
      "epoch": 0.4298615766849391,
      "grad_norm": 5.004683494567871,
      "learning_rate": 5.818708742553534e-07,
      "loss": 0.0587,
      "step": 2725
    },
    {
      "epoch": 0.43001932405252985,
      "grad_norm": 3.161196231842041,
      "learning_rate": 5.817098695862179e-07,
      "loss": 0.0352,
      "step": 2726
    },
    {
      "epoch": 0.4301770714201207,
      "grad_norm": 4.550516128540039,
      "learning_rate": 5.815488649170825e-07,
      "loss": 0.0362,
      "step": 2727
    },
    {
      "epoch": 0.43033481878771146,
      "grad_norm": 3.5981874465942383,
      "learning_rate": 5.813878602479471e-07,
      "loss": 0.0324,
      "step": 2728
    },
    {
      "epoch": 0.4304925661553023,
      "grad_norm": 7.343104839324951,
      "learning_rate": 5.812268555788118e-07,
      "loss": 0.0488,
      "step": 2729
    },
    {
      "epoch": 0.43065031352289307,
      "grad_norm": 4.300084114074707,
      "learning_rate": 5.810658509096763e-07,
      "loss": 0.064,
      "step": 2730
    },
    {
      "epoch": 0.4308080608904839,
      "grad_norm": 6.5787153244018555,
      "learning_rate": 5.80904846240541e-07,
      "loss": 0.0865,
      "step": 2731
    },
    {
      "epoch": 0.4309658082580747,
      "grad_norm": 6.296169281005859,
      "learning_rate": 5.807438415714056e-07,
      "loss": 0.0759,
      "step": 2732
    },
    {
      "epoch": 0.4311235556256655,
      "grad_norm": 4.846346378326416,
      "learning_rate": 5.805828369022702e-07,
      "loss": 0.0779,
      "step": 2733
    },
    {
      "epoch": 0.4312813029932563,
      "grad_norm": 3.2218270301818848,
      "learning_rate": 5.804218322331348e-07,
      "loss": 0.0375,
      "step": 2734
    },
    {
      "epoch": 0.4314390503608471,
      "grad_norm": 5.525953769683838,
      "learning_rate": 5.802608275639993e-07,
      "loss": 0.0571,
      "step": 2735
    },
    {
      "epoch": 0.4315967977284379,
      "grad_norm": 6.556750774383545,
      "learning_rate": 5.800998228948639e-07,
      "loss": 0.0584,
      "step": 2736
    },
    {
      "epoch": 0.43175454509602873,
      "grad_norm": 4.103269577026367,
      "learning_rate": 5.799388182257285e-07,
      "loss": 0.0402,
      "step": 2737
    },
    {
      "epoch": 0.4319122924636195,
      "grad_norm": 6.050023078918457,
      "learning_rate": 5.797778135565932e-07,
      "loss": 0.0632,
      "step": 2738
    },
    {
      "epoch": 0.43207003983121034,
      "grad_norm": 3.4731900691986084,
      "learning_rate": 5.796168088874577e-07,
      "loss": 0.0248,
      "step": 2739
    },
    {
      "epoch": 0.4322277871988011,
      "grad_norm": 10.451546669006348,
      "learning_rate": 5.794558042183223e-07,
      "loss": 0.0933,
      "step": 2740
    },
    {
      "epoch": 0.43238553456639195,
      "grad_norm": 3.39365291595459,
      "learning_rate": 5.792947995491869e-07,
      "loss": 0.0462,
      "step": 2741
    },
    {
      "epoch": 0.4325432819339827,
      "grad_norm": 3.792912483215332,
      "learning_rate": 5.791337948800515e-07,
      "loss": 0.0165,
      "step": 2742
    },
    {
      "epoch": 0.4327010293015735,
      "grad_norm": 2.6316964626312256,
      "learning_rate": 5.78972790210916e-07,
      "loss": 0.0131,
      "step": 2743
    },
    {
      "epoch": 0.43285877666916434,
      "grad_norm": 6.959858417510986,
      "learning_rate": 5.788117855417806e-07,
      "loss": 0.0567,
      "step": 2744
    },
    {
      "epoch": 0.4330165240367551,
      "grad_norm": 10.351690292358398,
      "learning_rate": 5.786507808726452e-07,
      "loss": 0.0672,
      "step": 2745
    },
    {
      "epoch": 0.43317427140434595,
      "grad_norm": 5.9646830558776855,
      "learning_rate": 5.7848977620351e-07,
      "loss": 0.0437,
      "step": 2746
    },
    {
      "epoch": 0.4333320187719367,
      "grad_norm": 5.861454010009766,
      "learning_rate": 5.783287715343745e-07,
      "loss": 0.0524,
      "step": 2747
    },
    {
      "epoch": 0.43348976613952755,
      "grad_norm": 8.58069896697998,
      "learning_rate": 5.781677668652391e-07,
      "loss": 0.0347,
      "step": 2748
    },
    {
      "epoch": 0.43364751350711833,
      "grad_norm": 5.208137035369873,
      "learning_rate": 5.780067621961037e-07,
      "loss": 0.0622,
      "step": 2749
    },
    {
      "epoch": 0.43380526087470916,
      "grad_norm": 3.031008243560791,
      "learning_rate": 5.778457575269683e-07,
      "loss": 0.0423,
      "step": 2750
    },
    {
      "epoch": 0.43396300824229994,
      "grad_norm": 6.715963840484619,
      "learning_rate": 5.776847528578328e-07,
      "loss": 0.0508,
      "step": 2751
    },
    {
      "epoch": 0.4341207556098908,
      "grad_norm": 3.5705153942108154,
      "learning_rate": 5.775237481886974e-07,
      "loss": 0.0146,
      "step": 2752
    },
    {
      "epoch": 0.43427850297748155,
      "grad_norm": 4.775838375091553,
      "learning_rate": 5.773627435195621e-07,
      "loss": 0.1016,
      "step": 2753
    },
    {
      "epoch": 0.4344362503450724,
      "grad_norm": 6.519713401794434,
      "learning_rate": 5.772017388504267e-07,
      "loss": 0.0768,
      "step": 2754
    },
    {
      "epoch": 0.43459399771266316,
      "grad_norm": 12.248930931091309,
      "learning_rate": 5.770407341812912e-07,
      "loss": 0.0351,
      "step": 2755
    },
    {
      "epoch": 0.434751745080254,
      "grad_norm": 2.580695152282715,
      "learning_rate": 5.768797295121558e-07,
      "loss": 0.0195,
      "step": 2756
    },
    {
      "epoch": 0.43490949244784477,
      "grad_norm": 9.897636413574219,
      "learning_rate": 5.767187248430204e-07,
      "loss": 0.0939,
      "step": 2757
    },
    {
      "epoch": 0.4350672398154356,
      "grad_norm": 26.89352798461914,
      "learning_rate": 5.76557720173885e-07,
      "loss": 0.0545,
      "step": 2758
    },
    {
      "epoch": 0.4352249871830264,
      "grad_norm": 4.561646461486816,
      "learning_rate": 5.763967155047496e-07,
      "loss": 0.0573,
      "step": 2759
    },
    {
      "epoch": 0.4353827345506172,
      "grad_norm": 2.932121753692627,
      "learning_rate": 5.762357108356141e-07,
      "loss": 0.03,
      "step": 2760
    },
    {
      "epoch": 0.435540481918208,
      "grad_norm": 7.722933769226074,
      "learning_rate": 5.760747061664788e-07,
      "loss": 0.0513,
      "step": 2761
    },
    {
      "epoch": 0.43569822928579877,
      "grad_norm": 4.556098937988281,
      "learning_rate": 5.759137014973434e-07,
      "loss": 0.0715,
      "step": 2762
    },
    {
      "epoch": 0.4358559766533896,
      "grad_norm": 5.859991073608398,
      "learning_rate": 5.757526968282081e-07,
      "loss": 0.071,
      "step": 2763
    },
    {
      "epoch": 0.4360137240209804,
      "grad_norm": 7.509644508361816,
      "learning_rate": 5.755916921590726e-07,
      "loss": 0.084,
      "step": 2764
    },
    {
      "epoch": 0.4361714713885712,
      "grad_norm": 7.722994804382324,
      "learning_rate": 5.754306874899372e-07,
      "loss": 0.0973,
      "step": 2765
    },
    {
      "epoch": 0.436329218756162,
      "grad_norm": 6.072455883026123,
      "learning_rate": 5.752696828208018e-07,
      "loss": 0.025,
      "step": 2766
    },
    {
      "epoch": 0.4364869661237528,
      "grad_norm": 6.797508239746094,
      "learning_rate": 5.751086781516664e-07,
      "loss": 0.0842,
      "step": 2767
    },
    {
      "epoch": 0.4366447134913436,
      "grad_norm": 10.333138465881348,
      "learning_rate": 5.74947673482531e-07,
      "loss": 0.1186,
      "step": 2768
    },
    {
      "epoch": 0.4368024608589344,
      "grad_norm": 6.689957141876221,
      "learning_rate": 5.747866688133956e-07,
      "loss": 0.0428,
      "step": 2769
    },
    {
      "epoch": 0.4369602082265252,
      "grad_norm": 5.772156715393066,
      "learning_rate": 5.746256641442602e-07,
      "loss": 0.0744,
      "step": 2770
    },
    {
      "epoch": 0.43711795559411604,
      "grad_norm": 7.585948944091797,
      "learning_rate": 5.744646594751248e-07,
      "loss": 0.0448,
      "step": 2771
    },
    {
      "epoch": 0.4372757029617068,
      "grad_norm": 4.13621187210083,
      "learning_rate": 5.743036548059893e-07,
      "loss": 0.0339,
      "step": 2772
    },
    {
      "epoch": 0.43743345032929765,
      "grad_norm": 7.188488483428955,
      "learning_rate": 5.741426501368539e-07,
      "loss": 0.0341,
      "step": 2773
    },
    {
      "epoch": 0.4375911976968884,
      "grad_norm": 2.731147289276123,
      "learning_rate": 5.739816454677185e-07,
      "loss": 0.0304,
      "step": 2774
    },
    {
      "epoch": 0.43774894506447926,
      "grad_norm": 4.252033710479736,
      "learning_rate": 5.738206407985831e-07,
      "loss": 0.0355,
      "step": 2775
    },
    {
      "epoch": 0.43790669243207003,
      "grad_norm": 11.173271179199219,
      "learning_rate": 5.736596361294477e-07,
      "loss": 0.0743,
      "step": 2776
    },
    {
      "epoch": 0.43806443979966087,
      "grad_norm": 5.05621337890625,
      "learning_rate": 5.734986314603123e-07,
      "loss": 0.0661,
      "step": 2777
    },
    {
      "epoch": 0.43822218716725164,
      "grad_norm": 7.741921901702881,
      "learning_rate": 5.733376267911769e-07,
      "loss": 0.0532,
      "step": 2778
    },
    {
      "epoch": 0.4383799345348425,
      "grad_norm": 3.764737367630005,
      "learning_rate": 5.731766221220415e-07,
      "loss": 0.0505,
      "step": 2779
    },
    {
      "epoch": 0.43853768190243325,
      "grad_norm": 4.446591854095459,
      "learning_rate": 5.730156174529061e-07,
      "loss": 0.0296,
      "step": 2780
    },
    {
      "epoch": 0.43869542927002403,
      "grad_norm": 7.5737833976745605,
      "learning_rate": 5.728546127837706e-07,
      "loss": 0.0662,
      "step": 2781
    },
    {
      "epoch": 0.43885317663761486,
      "grad_norm": 6.038362503051758,
      "learning_rate": 5.726936081146353e-07,
      "loss": 0.0429,
      "step": 2782
    },
    {
      "epoch": 0.43901092400520564,
      "grad_norm": 10.07390308380127,
      "learning_rate": 5.725326034455e-07,
      "loss": 0.0397,
      "step": 2783
    },
    {
      "epoch": 0.43916867137279647,
      "grad_norm": 3.347057819366455,
      "learning_rate": 5.723715987763646e-07,
      "loss": 0.0326,
      "step": 2784
    },
    {
      "epoch": 0.43932641874038725,
      "grad_norm": 3.8456451892852783,
      "learning_rate": 5.722105941072291e-07,
      "loss": 0.0722,
      "step": 2785
    },
    {
      "epoch": 0.4394841661079781,
      "grad_norm": 5.601995944976807,
      "learning_rate": 5.720495894380937e-07,
      "loss": 0.0328,
      "step": 2786
    },
    {
      "epoch": 0.43964191347556886,
      "grad_norm": 7.750946521759033,
      "learning_rate": 5.718885847689583e-07,
      "loss": 0.073,
      "step": 2787
    },
    {
      "epoch": 0.4397996608431597,
      "grad_norm": 3.2933003902435303,
      "learning_rate": 5.717275800998229e-07,
      "loss": 0.0139,
      "step": 2788
    },
    {
      "epoch": 0.43995740821075047,
      "grad_norm": 5.042623519897461,
      "learning_rate": 5.715665754306874e-07,
      "loss": 0.0895,
      "step": 2789
    },
    {
      "epoch": 0.4401151555783413,
      "grad_norm": 2.9454407691955566,
      "learning_rate": 5.71405570761552e-07,
      "loss": 0.0171,
      "step": 2790
    },
    {
      "epoch": 0.4402729029459321,
      "grad_norm": 5.160466194152832,
      "learning_rate": 5.712445660924167e-07,
      "loss": 0.0472,
      "step": 2791
    },
    {
      "epoch": 0.4404306503135229,
      "grad_norm": 3.523851156234741,
      "learning_rate": 5.710835614232813e-07,
      "loss": 0.0342,
      "step": 2792
    },
    {
      "epoch": 0.4405883976811137,
      "grad_norm": 4.827638626098633,
      "learning_rate": 5.709225567541458e-07,
      "loss": 0.0385,
      "step": 2793
    },
    {
      "epoch": 0.4407461450487045,
      "grad_norm": 5.04912805557251,
      "learning_rate": 5.707615520850104e-07,
      "loss": 0.0793,
      "step": 2794
    },
    {
      "epoch": 0.4409038924162953,
      "grad_norm": 6.973541259765625,
      "learning_rate": 5.70600547415875e-07,
      "loss": 0.0738,
      "step": 2795
    },
    {
      "epoch": 0.44106163978388613,
      "grad_norm": 5.731598377227783,
      "learning_rate": 5.704395427467396e-07,
      "loss": 0.0474,
      "step": 2796
    },
    {
      "epoch": 0.4412193871514769,
      "grad_norm": 4.190987586975098,
      "learning_rate": 5.702785380776041e-07,
      "loss": 0.0313,
      "step": 2797
    },
    {
      "epoch": 0.44137713451906774,
      "grad_norm": 3.4082014560699463,
      "learning_rate": 5.701175334084689e-07,
      "loss": 0.0202,
      "step": 2798
    },
    {
      "epoch": 0.4415348818866585,
      "grad_norm": 5.786175727844238,
      "learning_rate": 5.699565287393335e-07,
      "loss": 0.0601,
      "step": 2799
    },
    {
      "epoch": 0.44169262925424935,
      "grad_norm": 4.239172458648682,
      "learning_rate": 5.697955240701981e-07,
      "loss": 0.0689,
      "step": 2800
    },
    {
      "epoch": 0.4418503766218401,
      "grad_norm": 3.3087871074676514,
      "learning_rate": 5.696345194010626e-07,
      "loss": 0.0495,
      "step": 2801
    },
    {
      "epoch": 0.4420081239894309,
      "grad_norm": 6.40525484085083,
      "learning_rate": 5.694735147319272e-07,
      "loss": 0.0477,
      "step": 2802
    },
    {
      "epoch": 0.44216587135702173,
      "grad_norm": 5.296397686004639,
      "learning_rate": 5.693125100627918e-07,
      "loss": 0.0669,
      "step": 2803
    },
    {
      "epoch": 0.4423236187246125,
      "grad_norm": 8.928070068359375,
      "learning_rate": 5.691515053936564e-07,
      "loss": 0.0561,
      "step": 2804
    },
    {
      "epoch": 0.44248136609220334,
      "grad_norm": 3.2695870399475098,
      "learning_rate": 5.68990500724521e-07,
      "loss": 0.0434,
      "step": 2805
    },
    {
      "epoch": 0.4426391134597941,
      "grad_norm": 4.327723979949951,
      "learning_rate": 5.688294960553856e-07,
      "loss": 0.0267,
      "step": 2806
    },
    {
      "epoch": 0.44279686082738495,
      "grad_norm": 4.668676853179932,
      "learning_rate": 5.686684913862502e-07,
      "loss": 0.0436,
      "step": 2807
    },
    {
      "epoch": 0.44295460819497573,
      "grad_norm": 4.603848457336426,
      "learning_rate": 5.685074867171148e-07,
      "loss": 0.0389,
      "step": 2808
    },
    {
      "epoch": 0.44311235556256656,
      "grad_norm": 5.100075721740723,
      "learning_rate": 5.683464820479794e-07,
      "loss": 0.0364,
      "step": 2809
    },
    {
      "epoch": 0.44327010293015734,
      "grad_norm": 2.836162567138672,
      "learning_rate": 5.681854773788439e-07,
      "loss": 0.0326,
      "step": 2810
    },
    {
      "epoch": 0.44342785029774817,
      "grad_norm": 2.471791982650757,
      "learning_rate": 5.680244727097085e-07,
      "loss": 0.0176,
      "step": 2811
    },
    {
      "epoch": 0.44358559766533895,
      "grad_norm": 2.9325854778289795,
      "learning_rate": 5.678634680405731e-07,
      "loss": 0.0227,
      "step": 2812
    },
    {
      "epoch": 0.4437433450329298,
      "grad_norm": 2.0947155952453613,
      "learning_rate": 5.677024633714378e-07,
      "loss": 0.0172,
      "step": 2813
    },
    {
      "epoch": 0.44390109240052056,
      "grad_norm": 4.812472343444824,
      "learning_rate": 5.675414587023023e-07,
      "loss": 0.0524,
      "step": 2814
    },
    {
      "epoch": 0.4440588397681114,
      "grad_norm": 3.985248565673828,
      "learning_rate": 5.673804540331669e-07,
      "loss": 0.058,
      "step": 2815
    },
    {
      "epoch": 0.44421658713570217,
      "grad_norm": 3.3267018795013428,
      "learning_rate": 5.672194493640316e-07,
      "loss": 0.0511,
      "step": 2816
    },
    {
      "epoch": 0.444374334503293,
      "grad_norm": 13.52624225616455,
      "learning_rate": 5.670584446948962e-07,
      "loss": 0.0704,
      "step": 2817
    },
    {
      "epoch": 0.4445320818708838,
      "grad_norm": 3.9268667697906494,
      "learning_rate": 5.668974400257607e-07,
      "loss": 0.0291,
      "step": 2818
    },
    {
      "epoch": 0.4446898292384746,
      "grad_norm": 6.879584789276123,
      "learning_rate": 5.667364353566253e-07,
      "loss": 0.0462,
      "step": 2819
    },
    {
      "epoch": 0.4448475766060654,
      "grad_norm": 14.13100528717041,
      "learning_rate": 5.665754306874899e-07,
      "loss": 0.1236,
      "step": 2820
    },
    {
      "epoch": 0.44500532397365616,
      "grad_norm": 4.642329692840576,
      "learning_rate": 5.664144260183546e-07,
      "loss": 0.0347,
      "step": 2821
    },
    {
      "epoch": 0.445163071341247,
      "grad_norm": 9.75646686553955,
      "learning_rate": 5.662534213492191e-07,
      "loss": 0.0472,
      "step": 2822
    },
    {
      "epoch": 0.4453208187088378,
      "grad_norm": 2.149906635284424,
      "learning_rate": 5.660924166800837e-07,
      "loss": 0.0131,
      "step": 2823
    },
    {
      "epoch": 0.4454785660764286,
      "grad_norm": 3.265413761138916,
      "learning_rate": 5.659314120109483e-07,
      "loss": 0.034,
      "step": 2824
    },
    {
      "epoch": 0.4456363134440194,
      "grad_norm": 8.077981948852539,
      "learning_rate": 5.657704073418129e-07,
      "loss": 0.0458,
      "step": 2825
    },
    {
      "epoch": 0.4457940608116102,
      "grad_norm": 2.121990919113159,
      "learning_rate": 5.656094026726774e-07,
      "loss": 0.0165,
      "step": 2826
    },
    {
      "epoch": 0.445951808179201,
      "grad_norm": 2.8277318477630615,
      "learning_rate": 5.65448398003542e-07,
      "loss": 0.0289,
      "step": 2827
    },
    {
      "epoch": 0.4461095555467918,
      "grad_norm": 5.254321098327637,
      "learning_rate": 5.652873933344067e-07,
      "loss": 0.0265,
      "step": 2828
    },
    {
      "epoch": 0.4462673029143826,
      "grad_norm": 6.698361396789551,
      "learning_rate": 5.651263886652713e-07,
      "loss": 0.0487,
      "step": 2829
    },
    {
      "epoch": 0.44642505028197343,
      "grad_norm": 4.19152307510376,
      "learning_rate": 5.649653839961359e-07,
      "loss": 0.0237,
      "step": 2830
    },
    {
      "epoch": 0.4465827976495642,
      "grad_norm": 5.102362155914307,
      "learning_rate": 5.648043793270004e-07,
      "loss": 0.0231,
      "step": 2831
    },
    {
      "epoch": 0.44674054501715504,
      "grad_norm": 5.332268238067627,
      "learning_rate": 5.64643374657865e-07,
      "loss": 0.0409,
      "step": 2832
    },
    {
      "epoch": 0.4468982923847458,
      "grad_norm": 4.980330467224121,
      "learning_rate": 5.644823699887296e-07,
      "loss": 0.0329,
      "step": 2833
    },
    {
      "epoch": 0.44705603975233665,
      "grad_norm": 7.554039478302002,
      "learning_rate": 5.643213653195943e-07,
      "loss": 0.0562,
      "step": 2834
    },
    {
      "epoch": 0.44721378711992743,
      "grad_norm": 3.8478784561157227,
      "learning_rate": 5.641603606504588e-07,
      "loss": 0.0334,
      "step": 2835
    },
    {
      "epoch": 0.44737153448751826,
      "grad_norm": 5.280252933502197,
      "learning_rate": 5.639993559813235e-07,
      "loss": 0.0533,
      "step": 2836
    },
    {
      "epoch": 0.44752928185510904,
      "grad_norm": 7.56386661529541,
      "learning_rate": 5.638383513121881e-07,
      "loss": 0.0215,
      "step": 2837
    },
    {
      "epoch": 0.4476870292226999,
      "grad_norm": 5.807894229888916,
      "learning_rate": 5.636773466430527e-07,
      "loss": 0.0957,
      "step": 2838
    },
    {
      "epoch": 0.44784477659029065,
      "grad_norm": 5.296294689178467,
      "learning_rate": 5.635163419739172e-07,
      "loss": 0.0531,
      "step": 2839
    },
    {
      "epoch": 0.4480025239578814,
      "grad_norm": 4.388183116912842,
      "learning_rate": 5.633553373047818e-07,
      "loss": 0.0389,
      "step": 2840
    },
    {
      "epoch": 0.44816027132547226,
      "grad_norm": 4.946773052215576,
      "learning_rate": 5.631943326356464e-07,
      "loss": 0.0606,
      "step": 2841
    },
    {
      "epoch": 0.44831801869306304,
      "grad_norm": 4.141297340393066,
      "learning_rate": 5.63033327966511e-07,
      "loss": 0.0792,
      "step": 2842
    },
    {
      "epoch": 0.44847576606065387,
      "grad_norm": 4.016687870025635,
      "learning_rate": 5.628723232973755e-07,
      "loss": 0.0392,
      "step": 2843
    },
    {
      "epoch": 0.44863351342824465,
      "grad_norm": 6.216261863708496,
      "learning_rate": 5.627113186282402e-07,
      "loss": 0.0651,
      "step": 2844
    },
    {
      "epoch": 0.4487912607958355,
      "grad_norm": 2.2858870029449463,
      "learning_rate": 5.625503139591048e-07,
      "loss": 0.0208,
      "step": 2845
    },
    {
      "epoch": 0.44894900816342626,
      "grad_norm": 10.226709365844727,
      "learning_rate": 5.623893092899694e-07,
      "loss": 0.0896,
      "step": 2846
    },
    {
      "epoch": 0.4491067555310171,
      "grad_norm": 2.742591142654419,
      "learning_rate": 5.622283046208339e-07,
      "loss": 0.0343,
      "step": 2847
    },
    {
      "epoch": 0.44926450289860786,
      "grad_norm": 6.542262554168701,
      "learning_rate": 5.620672999516985e-07,
      "loss": 0.0589,
      "step": 2848
    },
    {
      "epoch": 0.4494222502661987,
      "grad_norm": 5.806894779205322,
      "learning_rate": 5.619062952825631e-07,
      "loss": 0.0275,
      "step": 2849
    },
    {
      "epoch": 0.4495799976337895,
      "grad_norm": 4.3196234703063965,
      "learning_rate": 5.617452906134277e-07,
      "loss": 0.075,
      "step": 2850
    },
    {
      "epoch": 0.4497377450013803,
      "grad_norm": 8.437948226928711,
      "learning_rate": 5.615842859442925e-07,
      "loss": 0.0619,
      "step": 2851
    },
    {
      "epoch": 0.4498954923689711,
      "grad_norm": 5.742941379547119,
      "learning_rate": 5.61423281275157e-07,
      "loss": 0.0528,
      "step": 2852
    },
    {
      "epoch": 0.4500532397365619,
      "grad_norm": 4.836208343505859,
      "learning_rate": 5.612622766060216e-07,
      "loss": 0.0452,
      "step": 2853
    },
    {
      "epoch": 0.4502109871041527,
      "grad_norm": 8.764130592346191,
      "learning_rate": 5.611012719368862e-07,
      "loss": 0.0533,
      "step": 2854
    },
    {
      "epoch": 0.4503687344717435,
      "grad_norm": 7.907119274139404,
      "learning_rate": 5.609402672677508e-07,
      "loss": 0.0899,
      "step": 2855
    },
    {
      "epoch": 0.4505264818393343,
      "grad_norm": 3.403799533843994,
      "learning_rate": 5.607792625986153e-07,
      "loss": 0.0305,
      "step": 2856
    },
    {
      "epoch": 0.45068422920692514,
      "grad_norm": 5.970966815948486,
      "learning_rate": 5.606182579294799e-07,
      "loss": 0.0641,
      "step": 2857
    },
    {
      "epoch": 0.4508419765745159,
      "grad_norm": 8.332094192504883,
      "learning_rate": 5.604572532603446e-07,
      "loss": 0.0651,
      "step": 2858
    },
    {
      "epoch": 0.4509997239421067,
      "grad_norm": 3.923067092895508,
      "learning_rate": 5.602962485912092e-07,
      "loss": 0.0323,
      "step": 2859
    },
    {
      "epoch": 0.4511574713096975,
      "grad_norm": 6.301210880279541,
      "learning_rate": 5.601352439220737e-07,
      "loss": 0.049,
      "step": 2860
    },
    {
      "epoch": 0.4513152186772883,
      "grad_norm": 2.444934129714966,
      "learning_rate": 5.599742392529383e-07,
      "loss": 0.0126,
      "step": 2861
    },
    {
      "epoch": 0.45147296604487913,
      "grad_norm": 5.900997638702393,
      "learning_rate": 5.598132345838029e-07,
      "loss": 0.1177,
      "step": 2862
    },
    {
      "epoch": 0.4516307134124699,
      "grad_norm": 8.015467643737793,
      "learning_rate": 5.596522299146675e-07,
      "loss": 0.0777,
      "step": 2863
    },
    {
      "epoch": 0.45178846078006074,
      "grad_norm": 2.3847241401672363,
      "learning_rate": 5.59491225245532e-07,
      "loss": 0.0204,
      "step": 2864
    },
    {
      "epoch": 0.4519462081476515,
      "grad_norm": 7.526757717132568,
      "learning_rate": 5.593302205763966e-07,
      "loss": 0.0971,
      "step": 2865
    },
    {
      "epoch": 0.45210395551524235,
      "grad_norm": 5.946081638336182,
      "learning_rate": 5.591692159072613e-07,
      "loss": 0.0577,
      "step": 2866
    },
    {
      "epoch": 0.4522617028828331,
      "grad_norm": 6.8733439445495605,
      "learning_rate": 5.590082112381259e-07,
      "loss": 0.1003,
      "step": 2867
    },
    {
      "epoch": 0.45241945025042396,
      "grad_norm": 4.146086692810059,
      "learning_rate": 5.588472065689904e-07,
      "loss": 0.0449,
      "step": 2868
    },
    {
      "epoch": 0.45257719761801474,
      "grad_norm": 3.731523036956787,
      "learning_rate": 5.58686201899855e-07,
      "loss": 0.0298,
      "step": 2869
    },
    {
      "epoch": 0.45273494498560557,
      "grad_norm": 5.817784309387207,
      "learning_rate": 5.585251972307197e-07,
      "loss": 0.0903,
      "step": 2870
    },
    {
      "epoch": 0.45289269235319635,
      "grad_norm": 8.980182647705078,
      "learning_rate": 5.583641925615843e-07,
      "loss": 0.0368,
      "step": 2871
    },
    {
      "epoch": 0.4530504397207872,
      "grad_norm": 5.589970111846924,
      "learning_rate": 5.582031878924488e-07,
      "loss": 0.0631,
      "step": 2872
    },
    {
      "epoch": 0.45320818708837796,
      "grad_norm": 3.7583048343658447,
      "learning_rate": 5.580421832233134e-07,
      "loss": 0.0333,
      "step": 2873
    },
    {
      "epoch": 0.4533659344559688,
      "grad_norm": 4.043829917907715,
      "learning_rate": 5.578811785541781e-07,
      "loss": 0.0355,
      "step": 2874
    },
    {
      "epoch": 0.45352368182355957,
      "grad_norm": 3.1130857467651367,
      "learning_rate": 5.577201738850427e-07,
      "loss": 0.0241,
      "step": 2875
    },
    {
      "epoch": 0.4536814291911504,
      "grad_norm": 4.681988716125488,
      "learning_rate": 5.575591692159073e-07,
      "loss": 0.0582,
      "step": 2876
    },
    {
      "epoch": 0.4538391765587412,
      "grad_norm": 2.468484401702881,
      "learning_rate": 5.573981645467718e-07,
      "loss": 0.0468,
      "step": 2877
    },
    {
      "epoch": 0.45399692392633195,
      "grad_norm": 8.368632316589355,
      "learning_rate": 5.572371598776364e-07,
      "loss": 0.0586,
      "step": 2878
    },
    {
      "epoch": 0.4541546712939228,
      "grad_norm": 5.181319713592529,
      "learning_rate": 5.57076155208501e-07,
      "loss": 0.0387,
      "step": 2879
    },
    {
      "epoch": 0.45431241866151356,
      "grad_norm": 5.580580711364746,
      "learning_rate": 5.569151505393656e-07,
      "loss": 0.0583,
      "step": 2880
    },
    {
      "epoch": 0.4544701660291044,
      "grad_norm": 3.7161035537719727,
      "learning_rate": 5.567541458702302e-07,
      "loss": 0.016,
      "step": 2881
    },
    {
      "epoch": 0.45462791339669517,
      "grad_norm": 6.218023777008057,
      "learning_rate": 5.565931412010948e-07,
      "loss": 0.0725,
      "step": 2882
    },
    {
      "epoch": 0.454785660764286,
      "grad_norm": 6.219332695007324,
      "learning_rate": 5.564321365319594e-07,
      "loss": 0.0468,
      "step": 2883
    },
    {
      "epoch": 0.4549434081318768,
      "grad_norm": 4.914670467376709,
      "learning_rate": 5.56271131862824e-07,
      "loss": 0.0206,
      "step": 2884
    },
    {
      "epoch": 0.4551011554994676,
      "grad_norm": 4.154045581817627,
      "learning_rate": 5.561101271936885e-07,
      "loss": 0.0351,
      "step": 2885
    },
    {
      "epoch": 0.4552589028670584,
      "grad_norm": 2.9888126850128174,
      "learning_rate": 5.559491225245531e-07,
      "loss": 0.0342,
      "step": 2886
    },
    {
      "epoch": 0.4554166502346492,
      "grad_norm": 5.366695880889893,
      "learning_rate": 5.557881178554178e-07,
      "loss": 0.0649,
      "step": 2887
    },
    {
      "epoch": 0.45557439760224,
      "grad_norm": 7.164974689483643,
      "learning_rate": 5.556271131862825e-07,
      "loss": 0.0316,
      "step": 2888
    },
    {
      "epoch": 0.45573214496983083,
      "grad_norm": 3.472078323364258,
      "learning_rate": 5.55466108517147e-07,
      "loss": 0.0455,
      "step": 2889
    },
    {
      "epoch": 0.4558898923374216,
      "grad_norm": 5.037097930908203,
      "learning_rate": 5.553051038480116e-07,
      "loss": 0.0471,
      "step": 2890
    },
    {
      "epoch": 0.45604763970501244,
      "grad_norm": 3.8302767276763916,
      "learning_rate": 5.551440991788762e-07,
      "loss": 0.0178,
      "step": 2891
    },
    {
      "epoch": 0.4562053870726032,
      "grad_norm": 5.1232218742370605,
      "learning_rate": 5.549830945097408e-07,
      "loss": 0.0493,
      "step": 2892
    },
    {
      "epoch": 0.45636313444019405,
      "grad_norm": 7.466702461242676,
      "learning_rate": 5.548220898406053e-07,
      "loss": 0.0394,
      "step": 2893
    },
    {
      "epoch": 0.45652088180778483,
      "grad_norm": 5.414917945861816,
      "learning_rate": 5.546610851714699e-07,
      "loss": 0.0427,
      "step": 2894
    },
    {
      "epoch": 0.45667862917537566,
      "grad_norm": 4.600940227508545,
      "learning_rate": 5.545000805023345e-07,
      "loss": 0.0606,
      "step": 2895
    },
    {
      "epoch": 0.45683637654296644,
      "grad_norm": 7.288424491882324,
      "learning_rate": 5.543390758331992e-07,
      "loss": 0.1097,
      "step": 2896
    },
    {
      "epoch": 0.4569941239105572,
      "grad_norm": 3.1181650161743164,
      "learning_rate": 5.541780711640638e-07,
      "loss": 0.0246,
      "step": 2897
    },
    {
      "epoch": 0.45715187127814805,
      "grad_norm": 8.209742546081543,
      "learning_rate": 5.540170664949283e-07,
      "loss": 0.0713,
      "step": 2898
    },
    {
      "epoch": 0.4573096186457388,
      "grad_norm": 7.193093299865723,
      "learning_rate": 5.538560618257929e-07,
      "loss": 0.0545,
      "step": 2899
    },
    {
      "epoch": 0.45746736601332966,
      "grad_norm": 3.752289295196533,
      "learning_rate": 5.536950571566575e-07,
      "loss": 0.0217,
      "step": 2900
    },
    {
      "epoch": 0.45762511338092043,
      "grad_norm": 4.784655570983887,
      "learning_rate": 5.535340524875221e-07,
      "loss": 0.0609,
      "step": 2901
    },
    {
      "epoch": 0.45778286074851127,
      "grad_norm": 5.997508525848389,
      "learning_rate": 5.533730478183866e-07,
      "loss": 0.0772,
      "step": 2902
    },
    {
      "epoch": 0.45794060811610204,
      "grad_norm": 4.745800018310547,
      "learning_rate": 5.532120431492512e-07,
      "loss": 0.0283,
      "step": 2903
    },
    {
      "epoch": 0.4580983554836929,
      "grad_norm": 3.1995737552642822,
      "learning_rate": 5.53051038480116e-07,
      "loss": 0.0358,
      "step": 2904
    },
    {
      "epoch": 0.45825610285128365,
      "grad_norm": 4.824692249298096,
      "learning_rate": 5.528900338109806e-07,
      "loss": 0.0268,
      "step": 2905
    },
    {
      "epoch": 0.4584138502188745,
      "grad_norm": 3.300462007522583,
      "learning_rate": 5.527290291418451e-07,
      "loss": 0.029,
      "step": 2906
    },
    {
      "epoch": 0.45857159758646526,
      "grad_norm": 4.107739448547363,
      "learning_rate": 5.525680244727097e-07,
      "loss": 0.0452,
      "step": 2907
    },
    {
      "epoch": 0.4587293449540561,
      "grad_norm": 4.978026390075684,
      "learning_rate": 5.524070198035743e-07,
      "loss": 0.0862,
      "step": 2908
    },
    {
      "epoch": 0.45888709232164687,
      "grad_norm": 5.25403356552124,
      "learning_rate": 5.522460151344389e-07,
      "loss": 0.0919,
      "step": 2909
    },
    {
      "epoch": 0.4590448396892377,
      "grad_norm": 3.5655996799468994,
      "learning_rate": 5.520850104653034e-07,
      "loss": 0.0295,
      "step": 2910
    },
    {
      "epoch": 0.4592025870568285,
      "grad_norm": 3.7482802867889404,
      "learning_rate": 5.519240057961681e-07,
      "loss": 0.0835,
      "step": 2911
    },
    {
      "epoch": 0.4593603344244193,
      "grad_norm": 9.63411808013916,
      "learning_rate": 5.517630011270327e-07,
      "loss": 0.0399,
      "step": 2912
    },
    {
      "epoch": 0.4595180817920101,
      "grad_norm": 2.809464931488037,
      "learning_rate": 5.516019964578973e-07,
      "loss": 0.0347,
      "step": 2913
    },
    {
      "epoch": 0.4596758291596009,
      "grad_norm": 4.120750904083252,
      "learning_rate": 5.514409917887618e-07,
      "loss": 0.0421,
      "step": 2914
    },
    {
      "epoch": 0.4598335765271917,
      "grad_norm": 6.6847615242004395,
      "learning_rate": 5.512799871196264e-07,
      "loss": 0.0666,
      "step": 2915
    },
    {
      "epoch": 0.45999132389478253,
      "grad_norm": 40.42054748535156,
      "learning_rate": 5.51118982450491e-07,
      "loss": 0.0489,
      "step": 2916
    },
    {
      "epoch": 0.4601490712623733,
      "grad_norm": 8.054227828979492,
      "learning_rate": 5.509579777813556e-07,
      "loss": 0.0498,
      "step": 2917
    },
    {
      "epoch": 0.4603068186299641,
      "grad_norm": 5.607865333557129,
      "learning_rate": 5.507969731122201e-07,
      "loss": 0.0602,
      "step": 2918
    },
    {
      "epoch": 0.4604645659975549,
      "grad_norm": 5.769576072692871,
      "learning_rate": 5.506359684430848e-07,
      "loss": 0.0322,
      "step": 2919
    },
    {
      "epoch": 0.4606223133651457,
      "grad_norm": 6.5281901359558105,
      "learning_rate": 5.504749637739494e-07,
      "loss": 0.0606,
      "step": 2920
    },
    {
      "epoch": 0.46078006073273653,
      "grad_norm": 4.875824451446533,
      "learning_rate": 5.50313959104814e-07,
      "loss": 0.0408,
      "step": 2921
    },
    {
      "epoch": 0.4609378081003273,
      "grad_norm": 11.520488739013672,
      "learning_rate": 5.501529544356787e-07,
      "loss": 0.061,
      "step": 2922
    },
    {
      "epoch": 0.46109555546791814,
      "grad_norm": 8.879790306091309,
      "learning_rate": 5.499919497665432e-07,
      "loss": 0.0576,
      "step": 2923
    },
    {
      "epoch": 0.4612533028355089,
      "grad_norm": 4.364360809326172,
      "learning_rate": 5.498309450974078e-07,
      "loss": 0.0545,
      "step": 2924
    },
    {
      "epoch": 0.46141105020309975,
      "grad_norm": 4.412820339202881,
      "learning_rate": 5.496699404282724e-07,
      "loss": 0.0885,
      "step": 2925
    },
    {
      "epoch": 0.4615687975706905,
      "grad_norm": 3.2861177921295166,
      "learning_rate": 5.495089357591371e-07,
      "loss": 0.0113,
      "step": 2926
    },
    {
      "epoch": 0.46172654493828136,
      "grad_norm": 7.449189186096191,
      "learning_rate": 5.493479310900016e-07,
      "loss": 0.0354,
      "step": 2927
    },
    {
      "epoch": 0.46188429230587214,
      "grad_norm": 10.700511932373047,
      "learning_rate": 5.491869264208662e-07,
      "loss": 0.0757,
      "step": 2928
    },
    {
      "epoch": 0.46204203967346297,
      "grad_norm": 7.325472831726074,
      "learning_rate": 5.490259217517308e-07,
      "loss": 0.0699,
      "step": 2929
    },
    {
      "epoch": 0.46219978704105374,
      "grad_norm": 2.8952784538269043,
      "learning_rate": 5.488649170825954e-07,
      "loss": 0.0302,
      "step": 2930
    },
    {
      "epoch": 0.4623575344086446,
      "grad_norm": 3.699751615524292,
      "learning_rate": 5.487039124134599e-07,
      "loss": 0.0282,
      "step": 2931
    },
    {
      "epoch": 0.46251528177623535,
      "grad_norm": 5.765395641326904,
      "learning_rate": 5.485429077443245e-07,
      "loss": 0.0356,
      "step": 2932
    },
    {
      "epoch": 0.4626730291438262,
      "grad_norm": 8.634234428405762,
      "learning_rate": 5.483819030751891e-07,
      "loss": 0.0492,
      "step": 2933
    },
    {
      "epoch": 0.46283077651141696,
      "grad_norm": 3.448993444442749,
      "learning_rate": 5.482208984060538e-07,
      "loss": 0.0394,
      "step": 2934
    },
    {
      "epoch": 0.4629885238790078,
      "grad_norm": 5.2671613693237305,
      "learning_rate": 5.480598937369183e-07,
      "loss": 0.0296,
      "step": 2935
    },
    {
      "epoch": 0.4631462712465986,
      "grad_norm": 4.191190242767334,
      "learning_rate": 5.478988890677829e-07,
      "loss": 0.0207,
      "step": 2936
    },
    {
      "epoch": 0.46330401861418935,
      "grad_norm": 3.1752512454986572,
      "learning_rate": 5.477378843986475e-07,
      "loss": 0.0424,
      "step": 2937
    },
    {
      "epoch": 0.4634617659817802,
      "grad_norm": 4.166040420532227,
      "learning_rate": 5.475768797295121e-07,
      "loss": 0.0471,
      "step": 2938
    },
    {
      "epoch": 0.46361951334937096,
      "grad_norm": 4.05285120010376,
      "learning_rate": 5.474158750603766e-07,
      "loss": 0.0403,
      "step": 2939
    },
    {
      "epoch": 0.4637772607169618,
      "grad_norm": 6.6873459815979,
      "learning_rate": 5.472548703912413e-07,
      "loss": 0.0365,
      "step": 2940
    },
    {
      "epoch": 0.46393500808455257,
      "grad_norm": 9.0433988571167,
      "learning_rate": 5.47093865722106e-07,
      "loss": 0.0514,
      "step": 2941
    },
    {
      "epoch": 0.4640927554521434,
      "grad_norm": 11.118558883666992,
      "learning_rate": 5.469328610529706e-07,
      "loss": 0.059,
      "step": 2942
    },
    {
      "epoch": 0.4642505028197342,
      "grad_norm": 3.9807562828063965,
      "learning_rate": 5.467718563838352e-07,
      "loss": 0.0517,
      "step": 2943
    },
    {
      "epoch": 0.464408250187325,
      "grad_norm": 5.029097080230713,
      "learning_rate": 5.466108517146997e-07,
      "loss": 0.0534,
      "step": 2944
    },
    {
      "epoch": 0.4645659975549158,
      "grad_norm": 4.308042526245117,
      "learning_rate": 5.464498470455643e-07,
      "loss": 0.0603,
      "step": 2945
    },
    {
      "epoch": 0.4647237449225066,
      "grad_norm": 4.594994068145752,
      "learning_rate": 5.462888423764289e-07,
      "loss": 0.0369,
      "step": 2946
    },
    {
      "epoch": 0.4648814922900974,
      "grad_norm": 5.747838020324707,
      "learning_rate": 5.461278377072935e-07,
      "loss": 0.0466,
      "step": 2947
    },
    {
      "epoch": 0.46503923965768823,
      "grad_norm": 7.009786128997803,
      "learning_rate": 5.45966833038158e-07,
      "loss": 0.0876,
      "step": 2948
    },
    {
      "epoch": 0.465196987025279,
      "grad_norm": 7.380682468414307,
      "learning_rate": 5.458058283690227e-07,
      "loss": 0.0836,
      "step": 2949
    },
    {
      "epoch": 0.46535473439286984,
      "grad_norm": 5.737110614776611,
      "learning_rate": 5.456448236998873e-07,
      "loss": 0.0674,
      "step": 2950
    },
    {
      "epoch": 0.4655124817604606,
      "grad_norm": 5.259807586669922,
      "learning_rate": 5.454838190307519e-07,
      "loss": 0.0371,
      "step": 2951
    },
    {
      "epoch": 0.46567022912805145,
      "grad_norm": 6.416214466094971,
      "learning_rate": 5.453228143616164e-07,
      "loss": 0.0218,
      "step": 2952
    },
    {
      "epoch": 0.4658279764956422,
      "grad_norm": 6.674959659576416,
      "learning_rate": 5.45161809692481e-07,
      "loss": 0.0538,
      "step": 2953
    },
    {
      "epoch": 0.46598572386323306,
      "grad_norm": 4.255826473236084,
      "learning_rate": 5.450008050233456e-07,
      "loss": 0.0544,
      "step": 2954
    },
    {
      "epoch": 0.46614347123082384,
      "grad_norm": 6.422432899475098,
      "learning_rate": 5.448398003542102e-07,
      "loss": 0.0384,
      "step": 2955
    },
    {
      "epoch": 0.4663012185984146,
      "grad_norm": 5.41008996963501,
      "learning_rate": 5.446787956850748e-07,
      "loss": 0.0451,
      "step": 2956
    },
    {
      "epoch": 0.46645896596600545,
      "grad_norm": 7.205507278442383,
      "learning_rate": 5.445177910159395e-07,
      "loss": 0.0911,
      "step": 2957
    },
    {
      "epoch": 0.4666167133335962,
      "grad_norm": 6.411682605743408,
      "learning_rate": 5.443567863468041e-07,
      "loss": 0.0752,
      "step": 2958
    },
    {
      "epoch": 0.46677446070118706,
      "grad_norm": 2.7243564128875732,
      "learning_rate": 5.441957816776687e-07,
      "loss": 0.0226,
      "step": 2959
    },
    {
      "epoch": 0.46693220806877783,
      "grad_norm": 6.036257266998291,
      "learning_rate": 5.440347770085332e-07,
      "loss": 0.0283,
      "step": 2960
    },
    {
      "epoch": 0.46708995543636866,
      "grad_norm": 6.654210567474365,
      "learning_rate": 5.438737723393978e-07,
      "loss": 0.0515,
      "step": 2961
    },
    {
      "epoch": 0.46724770280395944,
      "grad_norm": 2.447610855102539,
      "learning_rate": 5.437127676702624e-07,
      "loss": 0.0234,
      "step": 2962
    },
    {
      "epoch": 0.4674054501715503,
      "grad_norm": 4.649781703948975,
      "learning_rate": 5.43551763001127e-07,
      "loss": 0.0442,
      "step": 2963
    },
    {
      "epoch": 0.46756319753914105,
      "grad_norm": 5.978750228881836,
      "learning_rate": 5.433907583319916e-07,
      "loss": 0.0762,
      "step": 2964
    },
    {
      "epoch": 0.4677209449067319,
      "grad_norm": 4.587388038635254,
      "learning_rate": 5.432297536628562e-07,
      "loss": 0.0646,
      "step": 2965
    },
    {
      "epoch": 0.46787869227432266,
      "grad_norm": 5.262384414672852,
      "learning_rate": 5.430687489937208e-07,
      "loss": 0.1034,
      "step": 2966
    },
    {
      "epoch": 0.4680364396419135,
      "grad_norm": 5.238274097442627,
      "learning_rate": 5.429077443245854e-07,
      "loss": 0.0515,
      "step": 2967
    },
    {
      "epoch": 0.46819418700950427,
      "grad_norm": 6.681858062744141,
      "learning_rate": 5.4274673965545e-07,
      "loss": 0.0602,
      "step": 2968
    },
    {
      "epoch": 0.4683519343770951,
      "grad_norm": 6.393606662750244,
      "learning_rate": 5.425857349863145e-07,
      "loss": 0.0282,
      "step": 2969
    },
    {
      "epoch": 0.4685096817446859,
      "grad_norm": 7.102097988128662,
      "learning_rate": 5.424247303171791e-07,
      "loss": 0.0544,
      "step": 2970
    },
    {
      "epoch": 0.4686674291122767,
      "grad_norm": 4.880553722381592,
      "learning_rate": 5.422637256480438e-07,
      "loss": 0.0459,
      "step": 2971
    },
    {
      "epoch": 0.4688251764798675,
      "grad_norm": 3.7544023990631104,
      "learning_rate": 5.421027209789084e-07,
      "loss": 0.0378,
      "step": 2972
    },
    {
      "epoch": 0.4689829238474583,
      "grad_norm": 6.994016170501709,
      "learning_rate": 5.419417163097729e-07,
      "loss": 0.0429,
      "step": 2973
    },
    {
      "epoch": 0.4691406712150491,
      "grad_norm": 5.7882080078125,
      "learning_rate": 5.417807116406376e-07,
      "loss": 0.0392,
      "step": 2974
    },
    {
      "epoch": 0.4692984185826399,
      "grad_norm": 5.978911876678467,
      "learning_rate": 5.416197069715022e-07,
      "loss": 0.0514,
      "step": 2975
    },
    {
      "epoch": 0.4694561659502307,
      "grad_norm": 4.241064071655273,
      "learning_rate": 5.414587023023668e-07,
      "loss": 0.0513,
      "step": 2976
    },
    {
      "epoch": 0.4696139133178215,
      "grad_norm": 3.8420863151550293,
      "learning_rate": 5.412976976332313e-07,
      "loss": 0.0757,
      "step": 2977
    },
    {
      "epoch": 0.4697716606854123,
      "grad_norm": 8.296415328979492,
      "learning_rate": 5.411366929640959e-07,
      "loss": 0.0469,
      "step": 2978
    },
    {
      "epoch": 0.4699294080530031,
      "grad_norm": 3.1178243160247803,
      "learning_rate": 5.409756882949606e-07,
      "loss": 0.0351,
      "step": 2979
    },
    {
      "epoch": 0.4700871554205939,
      "grad_norm": 3.7371997833251953,
      "learning_rate": 5.408146836258252e-07,
      "loss": 0.0506,
      "step": 2980
    },
    {
      "epoch": 0.4702449027881847,
      "grad_norm": 6.004145622253418,
      "learning_rate": 5.406536789566897e-07,
      "loss": 0.0451,
      "step": 2981
    },
    {
      "epoch": 0.47040265015577554,
      "grad_norm": 4.889219760894775,
      "learning_rate": 5.404926742875543e-07,
      "loss": 0.0768,
      "step": 2982
    },
    {
      "epoch": 0.4705603975233663,
      "grad_norm": 4.599617958068848,
      "learning_rate": 5.403316696184189e-07,
      "loss": 0.0357,
      "step": 2983
    },
    {
      "epoch": 0.47071814489095715,
      "grad_norm": 2.7061350345611572,
      "learning_rate": 5.401706649492835e-07,
      "loss": 0.0362,
      "step": 2984
    },
    {
      "epoch": 0.4708758922585479,
      "grad_norm": 4.982329368591309,
      "learning_rate": 5.40009660280148e-07,
      "loss": 0.0432,
      "step": 2985
    },
    {
      "epoch": 0.47103363962613876,
      "grad_norm": 6.297840595245361,
      "learning_rate": 5.398486556110127e-07,
      "loss": 0.0526,
      "step": 2986
    },
    {
      "epoch": 0.47119138699372953,
      "grad_norm": 6.057762622833252,
      "learning_rate": 5.396876509418773e-07,
      "loss": 0.0927,
      "step": 2987
    },
    {
      "epoch": 0.47134913436132037,
      "grad_norm": 2.123640775680542,
      "learning_rate": 5.395266462727419e-07,
      "loss": 0.0295,
      "step": 2988
    },
    {
      "epoch": 0.47150688172891114,
      "grad_norm": 5.117542743682861,
      "learning_rate": 5.393656416036064e-07,
      "loss": 0.0397,
      "step": 2989
    },
    {
      "epoch": 0.471664629096502,
      "grad_norm": 10.117819786071777,
      "learning_rate": 5.39204636934471e-07,
      "loss": 0.072,
      "step": 2990
    },
    {
      "epoch": 0.47182237646409275,
      "grad_norm": 6.594187259674072,
      "learning_rate": 5.390436322653356e-07,
      "loss": 0.0768,
      "step": 2991
    },
    {
      "epoch": 0.4719801238316836,
      "grad_norm": 4.6469902992248535,
      "learning_rate": 5.388826275962003e-07,
      "loss": 0.0366,
      "step": 2992
    },
    {
      "epoch": 0.47213787119927436,
      "grad_norm": 5.465667247772217,
      "learning_rate": 5.387216229270649e-07,
      "loss": 0.0386,
      "step": 2993
    },
    {
      "epoch": 0.47229561856686514,
      "grad_norm": 9.858426094055176,
      "learning_rate": 5.385606182579295e-07,
      "loss": 0.0833,
      "step": 2994
    },
    {
      "epoch": 0.47245336593445597,
      "grad_norm": 9.21403694152832,
      "learning_rate": 5.383996135887941e-07,
      "loss": 0.081,
      "step": 2995
    },
    {
      "epoch": 0.47261111330204675,
      "grad_norm": 6.342504978179932,
      "learning_rate": 5.382386089196587e-07,
      "loss": 0.0432,
      "step": 2996
    },
    {
      "epoch": 0.4727688606696376,
      "grad_norm": 3.481053352355957,
      "learning_rate": 5.380776042505233e-07,
      "loss": 0.0196,
      "step": 2997
    },
    {
      "epoch": 0.47292660803722836,
      "grad_norm": 7.552100658416748,
      "learning_rate": 5.379165995813878e-07,
      "loss": 0.058,
      "step": 2998
    },
    {
      "epoch": 0.4730843554048192,
      "grad_norm": 5.199467658996582,
      "learning_rate": 5.377555949122524e-07,
      "loss": 0.0364,
      "step": 2999
    },
    {
      "epoch": 0.47324210277240997,
      "grad_norm": 5.026432991027832,
      "learning_rate": 5.37594590243117e-07,
      "loss": 0.0343,
      "step": 3000
    },
    {
      "epoch": 0.4733998501400008,
      "grad_norm": 4.435891151428223,
      "learning_rate": 5.374335855739817e-07,
      "loss": 0.0389,
      "step": 3001
    },
    {
      "epoch": 0.4735575975075916,
      "grad_norm": 3.941455364227295,
      "learning_rate": 5.372725809048462e-07,
      "loss": 0.0467,
      "step": 3002
    },
    {
      "epoch": 0.4737153448751824,
      "grad_norm": 3.924062728881836,
      "learning_rate": 5.371115762357108e-07,
      "loss": 0.0469,
      "step": 3003
    },
    {
      "epoch": 0.4738730922427732,
      "grad_norm": 4.199190139770508,
      "learning_rate": 5.369505715665754e-07,
      "loss": 0.0437,
      "step": 3004
    },
    {
      "epoch": 0.474030839610364,
      "grad_norm": 6.40954065322876,
      "learning_rate": 5.3678956689744e-07,
      "loss": 0.0552,
      "step": 3005
    },
    {
      "epoch": 0.4741885869779548,
      "grad_norm": 4.79932975769043,
      "learning_rate": 5.366285622283045e-07,
      "loss": 0.0377,
      "step": 3006
    },
    {
      "epoch": 0.47434633434554563,
      "grad_norm": 6.0069684982299805,
      "learning_rate": 5.364675575591691e-07,
      "loss": 0.0373,
      "step": 3007
    },
    {
      "epoch": 0.4745040817131364,
      "grad_norm": 3.9204750061035156,
      "learning_rate": 5.363065528900337e-07,
      "loss": 0.02,
      "step": 3008
    },
    {
      "epoch": 0.47466182908072724,
      "grad_norm": 6.374783515930176,
      "learning_rate": 5.361455482208985e-07,
      "loss": 0.0774,
      "step": 3009
    },
    {
      "epoch": 0.474819576448318,
      "grad_norm": 2.727414608001709,
      "learning_rate": 5.35984543551763e-07,
      "loss": 0.0176,
      "step": 3010
    },
    {
      "epoch": 0.47497732381590885,
      "grad_norm": 11.19305419921875,
      "learning_rate": 5.358235388826276e-07,
      "loss": 0.0782,
      "step": 3011
    },
    {
      "epoch": 0.4751350711834996,
      "grad_norm": 3.4844961166381836,
      "learning_rate": 5.356625342134922e-07,
      "loss": 0.0546,
      "step": 3012
    },
    {
      "epoch": 0.4752928185510904,
      "grad_norm": 10.611740112304688,
      "learning_rate": 5.355015295443568e-07,
      "loss": 0.0442,
      "step": 3013
    },
    {
      "epoch": 0.47545056591868123,
      "grad_norm": 4.291416168212891,
      "learning_rate": 5.353405248752214e-07,
      "loss": 0.0392,
      "step": 3014
    },
    {
      "epoch": 0.475608313286272,
      "grad_norm": 6.187273025512695,
      "learning_rate": 5.351795202060859e-07,
      "loss": 0.0579,
      "step": 3015
    },
    {
      "epoch": 0.47576606065386284,
      "grad_norm": 7.718811511993408,
      "learning_rate": 5.350185155369506e-07,
      "loss": 0.0708,
      "step": 3016
    },
    {
      "epoch": 0.4759238080214536,
      "grad_norm": 4.694831848144531,
      "learning_rate": 5.348575108678152e-07,
      "loss": 0.0477,
      "step": 3017
    },
    {
      "epoch": 0.47608155538904445,
      "grad_norm": 3.8272786140441895,
      "learning_rate": 5.346965061986798e-07,
      "loss": 0.0349,
      "step": 3018
    },
    {
      "epoch": 0.47623930275663523,
      "grad_norm": 3.613823175430298,
      "learning_rate": 5.345355015295443e-07,
      "loss": 0.0891,
      "step": 3019
    },
    {
      "epoch": 0.47639705012422606,
      "grad_norm": 3.9235403537750244,
      "learning_rate": 5.343744968604089e-07,
      "loss": 0.0155,
      "step": 3020
    },
    {
      "epoch": 0.47655479749181684,
      "grad_norm": 9.554059028625488,
      "learning_rate": 5.342134921912735e-07,
      "loss": 0.0535,
      "step": 3021
    },
    {
      "epoch": 0.47671254485940767,
      "grad_norm": 3.8005659580230713,
      "learning_rate": 5.340524875221381e-07,
      "loss": 0.0461,
      "step": 3022
    },
    {
      "epoch": 0.47687029222699845,
      "grad_norm": 5.433043479919434,
      "learning_rate": 5.338914828530026e-07,
      "loss": 0.0447,
      "step": 3023
    },
    {
      "epoch": 0.4770280395945893,
      "grad_norm": 4.371764659881592,
      "learning_rate": 5.337304781838673e-07,
      "loss": 0.0495,
      "step": 3024
    },
    {
      "epoch": 0.47718578696218006,
      "grad_norm": 7.512784004211426,
      "learning_rate": 5.335694735147319e-07,
      "loss": 0.0635,
      "step": 3025
    },
    {
      "epoch": 0.4773435343297709,
      "grad_norm": 3.749889850616455,
      "learning_rate": 5.334084688455965e-07,
      "loss": 0.0284,
      "step": 3026
    },
    {
      "epoch": 0.47750128169736167,
      "grad_norm": 3.9005444049835205,
      "learning_rate": 5.33247464176461e-07,
      "loss": 0.0524,
      "step": 3027
    },
    {
      "epoch": 0.4776590290649525,
      "grad_norm": 2.9798567295074463,
      "learning_rate": 5.330864595073257e-07,
      "loss": 0.0374,
      "step": 3028
    },
    {
      "epoch": 0.4778167764325433,
      "grad_norm": 3.3086020946502686,
      "learning_rate": 5.329254548381903e-07,
      "loss": 0.0379,
      "step": 3029
    },
    {
      "epoch": 0.4779745238001341,
      "grad_norm": 6.154900550842285,
      "learning_rate": 5.327644501690549e-07,
      "loss": 0.115,
      "step": 3030
    },
    {
      "epoch": 0.4781322711677249,
      "grad_norm": 6.707804203033447,
      "learning_rate": 5.326034454999195e-07,
      "loss": 0.0655,
      "step": 3031
    },
    {
      "epoch": 0.4782900185353157,
      "grad_norm": 11.475181579589844,
      "learning_rate": 5.324424408307841e-07,
      "loss": 0.0138,
      "step": 3032
    },
    {
      "epoch": 0.4784477659029065,
      "grad_norm": 5.882479190826416,
      "learning_rate": 5.322814361616487e-07,
      "loss": 0.0397,
      "step": 3033
    },
    {
      "epoch": 0.4786055132704973,
      "grad_norm": 4.037417888641357,
      "learning_rate": 5.321204314925133e-07,
      "loss": 0.0269,
      "step": 3034
    },
    {
      "epoch": 0.4787632606380881,
      "grad_norm": 4.259555339813232,
      "learning_rate": 5.319594268233778e-07,
      "loss": 0.0343,
      "step": 3035
    },
    {
      "epoch": 0.4789210080056789,
      "grad_norm": 4.306237697601318,
      "learning_rate": 5.317984221542424e-07,
      "loss": 0.0457,
      "step": 3036
    },
    {
      "epoch": 0.4790787553732697,
      "grad_norm": 6.921058654785156,
      "learning_rate": 5.31637417485107e-07,
      "loss": 0.0475,
      "step": 3037
    },
    {
      "epoch": 0.4792365027408605,
      "grad_norm": 6.094809055328369,
      "learning_rate": 5.314764128159716e-07,
      "loss": 0.062,
      "step": 3038
    },
    {
      "epoch": 0.4793942501084513,
      "grad_norm": 6.288085460662842,
      "learning_rate": 5.313154081468363e-07,
      "loss": 0.0811,
      "step": 3039
    },
    {
      "epoch": 0.4795519974760421,
      "grad_norm": 6.041825294494629,
      "learning_rate": 5.311544034777008e-07,
      "loss": 0.0332,
      "step": 3040
    },
    {
      "epoch": 0.47970974484363293,
      "grad_norm": 4.585716247558594,
      "learning_rate": 5.309933988085654e-07,
      "loss": 0.0374,
      "step": 3041
    },
    {
      "epoch": 0.4798674922112237,
      "grad_norm": 5.681454181671143,
      "learning_rate": 5.3083239413943e-07,
      "loss": 0.0506,
      "step": 3042
    },
    {
      "epoch": 0.48002523957881454,
      "grad_norm": 6.8209452629089355,
      "learning_rate": 5.306713894702946e-07,
      "loss": 0.082,
      "step": 3043
    },
    {
      "epoch": 0.4801829869464053,
      "grad_norm": 3.82092022895813,
      "learning_rate": 5.305103848011591e-07,
      "loss": 0.0435,
      "step": 3044
    },
    {
      "epoch": 0.48034073431399615,
      "grad_norm": 5.553289890289307,
      "learning_rate": 5.303493801320238e-07,
      "loss": 0.0445,
      "step": 3045
    },
    {
      "epoch": 0.48049848168158693,
      "grad_norm": 4.531376361846924,
      "learning_rate": 5.301883754628885e-07,
      "loss": 0.0408,
      "step": 3046
    },
    {
      "epoch": 0.48065622904917776,
      "grad_norm": 3.5035979747772217,
      "learning_rate": 5.300273707937531e-07,
      "loss": 0.0273,
      "step": 3047
    },
    {
      "epoch": 0.48081397641676854,
      "grad_norm": 4.718772888183594,
      "learning_rate": 5.298663661246176e-07,
      "loss": 0.0424,
      "step": 3048
    },
    {
      "epoch": 0.4809717237843594,
      "grad_norm": 3.934015989303589,
      "learning_rate": 5.297053614554822e-07,
      "loss": 0.0445,
      "step": 3049
    },
    {
      "epoch": 0.48112947115195015,
      "grad_norm": 5.095694541931152,
      "learning_rate": 5.295443567863468e-07,
      "loss": 0.0314,
      "step": 3050
    },
    {
      "epoch": 0.481287218519541,
      "grad_norm": 6.989995956420898,
      "learning_rate": 5.293833521172114e-07,
      "loss": 0.0525,
      "step": 3051
    },
    {
      "epoch": 0.48144496588713176,
      "grad_norm": 7.8963470458984375,
      "learning_rate": 5.292223474480759e-07,
      "loss": 0.0641,
      "step": 3052
    },
    {
      "epoch": 0.48160271325472254,
      "grad_norm": 4.29169225692749,
      "learning_rate": 5.290613427789405e-07,
      "loss": 0.0245,
      "step": 3053
    },
    {
      "epoch": 0.48176046062231337,
      "grad_norm": 6.149662017822266,
      "learning_rate": 5.289003381098052e-07,
      "loss": 0.0321,
      "step": 3054
    },
    {
      "epoch": 0.48191820798990415,
      "grad_norm": 3.3679161071777344,
      "learning_rate": 5.287393334406698e-07,
      "loss": 0.0216,
      "step": 3055
    },
    {
      "epoch": 0.482075955357495,
      "grad_norm": 8.198663711547852,
      "learning_rate": 5.285783287715343e-07,
      "loss": 0.0607,
      "step": 3056
    },
    {
      "epoch": 0.48223370272508576,
      "grad_norm": 5.388924598693848,
      "learning_rate": 5.284173241023989e-07,
      "loss": 0.0236,
      "step": 3057
    },
    {
      "epoch": 0.4823914500926766,
      "grad_norm": 5.866364479064941,
      "learning_rate": 5.282563194332635e-07,
      "loss": 0.068,
      "step": 3058
    },
    {
      "epoch": 0.48254919746026737,
      "grad_norm": 7.354922294616699,
      "learning_rate": 5.280953147641281e-07,
      "loss": 0.06,
      "step": 3059
    },
    {
      "epoch": 0.4827069448278582,
      "grad_norm": 6.349786758422852,
      "learning_rate": 5.279343100949927e-07,
      "loss": 0.0549,
      "step": 3060
    },
    {
      "epoch": 0.482864692195449,
      "grad_norm": 2.514930248260498,
      "learning_rate": 5.277733054258573e-07,
      "loss": 0.0511,
      "step": 3061
    },
    {
      "epoch": 0.4830224395630398,
      "grad_norm": 8.368856430053711,
      "learning_rate": 5.27612300756722e-07,
      "loss": 0.076,
      "step": 3062
    },
    {
      "epoch": 0.4831801869306306,
      "grad_norm": 3.9374332427978516,
      "learning_rate": 5.274512960875866e-07,
      "loss": 0.0585,
      "step": 3063
    },
    {
      "epoch": 0.4833379342982214,
      "grad_norm": 7.416379451751709,
      "learning_rate": 5.272902914184512e-07,
      "loss": 0.0622,
      "step": 3064
    },
    {
      "epoch": 0.4834956816658122,
      "grad_norm": 6.479883670806885,
      "learning_rate": 5.271292867493157e-07,
      "loss": 0.0311,
      "step": 3065
    },
    {
      "epoch": 0.483653429033403,
      "grad_norm": 3.6350669860839844,
      "learning_rate": 5.269682820801803e-07,
      "loss": 0.0399,
      "step": 3066
    },
    {
      "epoch": 0.4838111764009938,
      "grad_norm": 4.032829284667969,
      "learning_rate": 5.268072774110449e-07,
      "loss": 0.0329,
      "step": 3067
    },
    {
      "epoch": 0.48396892376858464,
      "grad_norm": 3.473764657974243,
      "learning_rate": 5.266462727419095e-07,
      "loss": 0.0383,
      "step": 3068
    },
    {
      "epoch": 0.4841266711361754,
      "grad_norm": 5.921380043029785,
      "learning_rate": 5.264852680727741e-07,
      "loss": 0.0277,
      "step": 3069
    },
    {
      "epoch": 0.48428441850376625,
      "grad_norm": 3.3913817405700684,
      "learning_rate": 5.263242634036387e-07,
      "loss": 0.0313,
      "step": 3070
    },
    {
      "epoch": 0.484442165871357,
      "grad_norm": 3.7090003490448,
      "learning_rate": 5.261632587345033e-07,
      "loss": 0.0352,
      "step": 3071
    },
    {
      "epoch": 0.4845999132389478,
      "grad_norm": 7.192994594573975,
      "learning_rate": 5.260022540653679e-07,
      "loss": 0.0455,
      "step": 3072
    },
    {
      "epoch": 0.4845999132389478,
      "eval_accuracy": 0.9856942150797368,
      "eval_f1": 0.9856942150797368,
      "eval_loss": 0.046789322048425674,
      "eval_runtime": 4718.3444,
      "eval_samples_per_second": 42.993,
      "eval_steps_per_second": 2.687,
      "step": 3072
    },
    {
      "epoch": 0.48475766060653863,
      "grad_norm": 6.613557815551758,
      "learning_rate": 5.258412493962324e-07,
      "loss": 0.0582,
      "step": 3073
    },
    {
      "epoch": 0.4849154079741294,
      "grad_norm": 4.4686055183410645,
      "learning_rate": 5.25680244727097e-07,
      "loss": 0.0596,
      "step": 3074
    },
    {
      "epoch": 0.48507315534172024,
      "grad_norm": 6.73822021484375,
      "learning_rate": 5.255192400579616e-07,
      "loss": 0.0509,
      "step": 3075
    },
    {
      "epoch": 0.485230902709311,
      "grad_norm": 6.796287536621094,
      "learning_rate": 5.253582353888263e-07,
      "loss": 0.0428,
      "step": 3076
    },
    {
      "epoch": 0.48538865007690185,
      "grad_norm": 6.841584205627441,
      "learning_rate": 5.251972307196908e-07,
      "loss": 0.0494,
      "step": 3077
    },
    {
      "epoch": 0.48554639744449263,
      "grad_norm": 6.251879692077637,
      "learning_rate": 5.250362260505554e-07,
      "loss": 0.0518,
      "step": 3078
    },
    {
      "epoch": 0.48570414481208346,
      "grad_norm": 7.815091133117676,
      "learning_rate": 5.2487522138142e-07,
      "loss": 0.1038,
      "step": 3079
    },
    {
      "epoch": 0.48586189217967424,
      "grad_norm": 5.727319717407227,
      "learning_rate": 5.247142167122847e-07,
      "loss": 0.0509,
      "step": 3080
    },
    {
      "epoch": 0.48601963954726507,
      "grad_norm": 3.4534387588500977,
      "learning_rate": 5.245532120431492e-07,
      "loss": 0.0771,
      "step": 3081
    },
    {
      "epoch": 0.48617738691485585,
      "grad_norm": 4.194572925567627,
      "learning_rate": 5.243922073740138e-07,
      "loss": 0.0443,
      "step": 3082
    },
    {
      "epoch": 0.4863351342824467,
      "grad_norm": 10.300954818725586,
      "learning_rate": 5.242312027048784e-07,
      "loss": 0.0689,
      "step": 3083
    },
    {
      "epoch": 0.48649288165003746,
      "grad_norm": 5.091367244720459,
      "learning_rate": 5.240701980357431e-07,
      "loss": 0.0846,
      "step": 3084
    },
    {
      "epoch": 0.4866506290176283,
      "grad_norm": 2.955735445022583,
      "learning_rate": 5.239091933666077e-07,
      "loss": 0.0502,
      "step": 3085
    },
    {
      "epoch": 0.48680837638521907,
      "grad_norm": 4.298361778259277,
      "learning_rate": 5.237481886974722e-07,
      "loss": 0.0258,
      "step": 3086
    },
    {
      "epoch": 0.4869661237528099,
      "grad_norm": 3.9187076091766357,
      "learning_rate": 5.235871840283368e-07,
      "loss": 0.0644,
      "step": 3087
    },
    {
      "epoch": 0.4871238711204007,
      "grad_norm": 5.889786243438721,
      "learning_rate": 5.234261793592014e-07,
      "loss": 0.0795,
      "step": 3088
    },
    {
      "epoch": 0.4872816184879915,
      "grad_norm": 12.091011047363281,
      "learning_rate": 5.23265174690066e-07,
      "loss": 0.0682,
      "step": 3089
    },
    {
      "epoch": 0.4874393658555823,
      "grad_norm": 5.780660629272461,
      "learning_rate": 5.231041700209305e-07,
      "loss": 0.0319,
      "step": 3090
    },
    {
      "epoch": 0.48759711322317306,
      "grad_norm": 1.9263288974761963,
      "learning_rate": 5.229431653517952e-07,
      "loss": 0.0138,
      "step": 3091
    },
    {
      "epoch": 0.4877548605907639,
      "grad_norm": 6.742379188537598,
      "learning_rate": 5.227821606826598e-07,
      "loss": 0.0789,
      "step": 3092
    },
    {
      "epoch": 0.48791260795835467,
      "grad_norm": 3.9017274379730225,
      "learning_rate": 5.226211560135244e-07,
      "loss": 0.0577,
      "step": 3093
    },
    {
      "epoch": 0.4880703553259455,
      "grad_norm": 6.11548376083374,
      "learning_rate": 5.224601513443889e-07,
      "loss": 0.0603,
      "step": 3094
    },
    {
      "epoch": 0.4882281026935363,
      "grad_norm": 3.4503278732299805,
      "learning_rate": 5.222991466752535e-07,
      "loss": 0.0297,
      "step": 3095
    },
    {
      "epoch": 0.4883858500611271,
      "grad_norm": 4.641238212585449,
      "learning_rate": 5.221381420061181e-07,
      "loss": 0.0481,
      "step": 3096
    },
    {
      "epoch": 0.4885435974287179,
      "grad_norm": 5.323633193969727,
      "learning_rate": 5.219771373369828e-07,
      "loss": 0.022,
      "step": 3097
    },
    {
      "epoch": 0.4887013447963087,
      "grad_norm": 5.186425685882568,
      "learning_rate": 5.218161326678473e-07,
      "loss": 0.0522,
      "step": 3098
    },
    {
      "epoch": 0.4888590921638995,
      "grad_norm": 4.983971118927002,
      "learning_rate": 5.21655127998712e-07,
      "loss": 0.0486,
      "step": 3099
    },
    {
      "epoch": 0.48901683953149033,
      "grad_norm": 4.3277082443237305,
      "learning_rate": 5.214941233295766e-07,
      "loss": 0.0249,
      "step": 3100
    },
    {
      "epoch": 0.4891745868990811,
      "grad_norm": 4.370128154754639,
      "learning_rate": 5.213331186604412e-07,
      "loss": 0.0343,
      "step": 3101
    },
    {
      "epoch": 0.48933233426667194,
      "grad_norm": 4.3571553230285645,
      "learning_rate": 5.211721139913057e-07,
      "loss": 0.0693,
      "step": 3102
    },
    {
      "epoch": 0.4894900816342627,
      "grad_norm": 2.895589590072632,
      "learning_rate": 5.210111093221703e-07,
      "loss": 0.0307,
      "step": 3103
    },
    {
      "epoch": 0.48964782900185355,
      "grad_norm": 18.217578887939453,
      "learning_rate": 5.208501046530349e-07,
      "loss": 0.0699,
      "step": 3104
    },
    {
      "epoch": 0.48980557636944433,
      "grad_norm": 3.3057503700256348,
      "learning_rate": 5.206890999838995e-07,
      "loss": 0.0223,
      "step": 3105
    },
    {
      "epoch": 0.48996332373703516,
      "grad_norm": 5.851504802703857,
      "learning_rate": 5.205280953147642e-07,
      "loss": 0.0717,
      "step": 3106
    },
    {
      "epoch": 0.49012107110462594,
      "grad_norm": 4.29199743270874,
      "learning_rate": 5.203670906456287e-07,
      "loss": 0.0359,
      "step": 3107
    },
    {
      "epoch": 0.49027881847221677,
      "grad_norm": 12.454666137695312,
      "learning_rate": 5.202060859764933e-07,
      "loss": 0.0706,
      "step": 3108
    },
    {
      "epoch": 0.49043656583980755,
      "grad_norm": 5.318867206573486,
      "learning_rate": 5.200450813073579e-07,
      "loss": 0.0397,
      "step": 3109
    },
    {
      "epoch": 0.4905943132073983,
      "grad_norm": 3.4172487258911133,
      "learning_rate": 5.198840766382225e-07,
      "loss": 0.0545,
      "step": 3110
    },
    {
      "epoch": 0.49075206057498916,
      "grad_norm": 3.336772918701172,
      "learning_rate": 5.19723071969087e-07,
      "loss": 0.0617,
      "step": 3111
    },
    {
      "epoch": 0.49090980794257993,
      "grad_norm": 4.109711647033691,
      "learning_rate": 5.195620672999516e-07,
      "loss": 0.017,
      "step": 3112
    },
    {
      "epoch": 0.49106755531017077,
      "grad_norm": 5.944070816040039,
      "learning_rate": 5.194010626308162e-07,
      "loss": 0.0421,
      "step": 3113
    },
    {
      "epoch": 0.49122530267776154,
      "grad_norm": 9.251203536987305,
      "learning_rate": 5.19240057961681e-07,
      "loss": 0.087,
      "step": 3114
    },
    {
      "epoch": 0.4913830500453524,
      "grad_norm": 2.984105110168457,
      "learning_rate": 5.190790532925455e-07,
      "loss": 0.0252,
      "step": 3115
    },
    {
      "epoch": 0.49154079741294315,
      "grad_norm": 8.3684663772583,
      "learning_rate": 5.189180486234101e-07,
      "loss": 0.0767,
      "step": 3116
    },
    {
      "epoch": 0.491698544780534,
      "grad_norm": 4.011730194091797,
      "learning_rate": 5.187570439542747e-07,
      "loss": 0.0203,
      "step": 3117
    },
    {
      "epoch": 0.49185629214812476,
      "grad_norm": 5.830489158630371,
      "learning_rate": 5.185960392851393e-07,
      "loss": 0.0597,
      "step": 3118
    },
    {
      "epoch": 0.4920140395157156,
      "grad_norm": 3.307762861251831,
      "learning_rate": 5.184350346160038e-07,
      "loss": 0.034,
      "step": 3119
    },
    {
      "epoch": 0.4921717868833064,
      "grad_norm": 4.353969573974609,
      "learning_rate": 5.182740299468684e-07,
      "loss": 0.0443,
      "step": 3120
    },
    {
      "epoch": 0.4923295342508972,
      "grad_norm": 11.109829902648926,
      "learning_rate": 5.181130252777331e-07,
      "loss": 0.1123,
      "step": 3121
    },
    {
      "epoch": 0.492487281618488,
      "grad_norm": 4.706302642822266,
      "learning_rate": 5.179520206085977e-07,
      "loss": 0.0456,
      "step": 3122
    },
    {
      "epoch": 0.4926450289860788,
      "grad_norm": 4.336117744445801,
      "learning_rate": 5.177910159394622e-07,
      "loss": 0.0527,
      "step": 3123
    },
    {
      "epoch": 0.4928027763536696,
      "grad_norm": 6.7515974044799805,
      "learning_rate": 5.176300112703268e-07,
      "loss": 0.0409,
      "step": 3124
    },
    {
      "epoch": 0.4929605237212604,
      "grad_norm": 8.648667335510254,
      "learning_rate": 5.174690066011914e-07,
      "loss": 0.0723,
      "step": 3125
    },
    {
      "epoch": 0.4931182710888512,
      "grad_norm": 6.021974563598633,
      "learning_rate": 5.17308001932056e-07,
      "loss": 0.0595,
      "step": 3126
    },
    {
      "epoch": 0.49327601845644203,
      "grad_norm": 5.117554664611816,
      "learning_rate": 5.171469972629205e-07,
      "loss": 0.0302,
      "step": 3127
    },
    {
      "epoch": 0.4934337658240328,
      "grad_norm": 4.950992584228516,
      "learning_rate": 5.169859925937851e-07,
      "loss": 0.0225,
      "step": 3128
    },
    {
      "epoch": 0.4935915131916236,
      "grad_norm": 4.997193336486816,
      "learning_rate": 5.168249879246498e-07,
      "loss": 0.055,
      "step": 3129
    },
    {
      "epoch": 0.4937492605592144,
      "grad_norm": 4.263675689697266,
      "learning_rate": 5.166639832555144e-07,
      "loss": 0.0546,
      "step": 3130
    },
    {
      "epoch": 0.4939070079268052,
      "grad_norm": 7.349658489227295,
      "learning_rate": 5.16502978586379e-07,
      "loss": 0.0453,
      "step": 3131
    },
    {
      "epoch": 0.49406475529439603,
      "grad_norm": 7.558027267456055,
      "learning_rate": 5.163419739172435e-07,
      "loss": 0.0497,
      "step": 3132
    },
    {
      "epoch": 0.4942225026619868,
      "grad_norm": 9.713709831237793,
      "learning_rate": 5.161809692481082e-07,
      "loss": 0.0714,
      "step": 3133
    },
    {
      "epoch": 0.49438025002957764,
      "grad_norm": 2.164876699447632,
      "learning_rate": 5.160199645789728e-07,
      "loss": 0.0163,
      "step": 3134
    },
    {
      "epoch": 0.4945379973971684,
      "grad_norm": 5.119956970214844,
      "learning_rate": 5.158589599098374e-07,
      "loss": 0.0382,
      "step": 3135
    },
    {
      "epoch": 0.49469574476475925,
      "grad_norm": 6.049105167388916,
      "learning_rate": 5.15697955240702e-07,
      "loss": 0.0897,
      "step": 3136
    },
    {
      "epoch": 0.49485349213235,
      "grad_norm": 6.990386962890625,
      "learning_rate": 5.155369505715666e-07,
      "loss": 0.1112,
      "step": 3137
    },
    {
      "epoch": 0.49501123949994086,
      "grad_norm": 6.017998695373535,
      "learning_rate": 5.153759459024312e-07,
      "loss": 0.0496,
      "step": 3138
    },
    {
      "epoch": 0.49516898686753164,
      "grad_norm": 1.9739807844161987,
      "learning_rate": 5.152149412332958e-07,
      "loss": 0.0136,
      "step": 3139
    },
    {
      "epoch": 0.49532673423512247,
      "grad_norm": 3.636338949203491,
      "learning_rate": 5.150539365641603e-07,
      "loss": 0.0613,
      "step": 3140
    },
    {
      "epoch": 0.49548448160271324,
      "grad_norm": 4.632494926452637,
      "learning_rate": 5.148929318950249e-07,
      "loss": 0.0413,
      "step": 3141
    },
    {
      "epoch": 0.4956422289703041,
      "grad_norm": 6.799563884735107,
      "learning_rate": 5.147319272258895e-07,
      "loss": 0.0237,
      "step": 3142
    },
    {
      "epoch": 0.49579997633789485,
      "grad_norm": 4.903563022613525,
      "learning_rate": 5.145709225567541e-07,
      "loss": 0.0705,
      "step": 3143
    },
    {
      "epoch": 0.4959577237054857,
      "grad_norm": 5.018761157989502,
      "learning_rate": 5.144099178876187e-07,
      "loss": 0.0361,
      "step": 3144
    },
    {
      "epoch": 0.49611547107307646,
      "grad_norm": 7.929637432098389,
      "learning_rate": 5.142489132184833e-07,
      "loss": 0.0772,
      "step": 3145
    },
    {
      "epoch": 0.4962732184406673,
      "grad_norm": 4.951676845550537,
      "learning_rate": 5.140879085493479e-07,
      "loss": 0.0297,
      "step": 3146
    },
    {
      "epoch": 0.4964309658082581,
      "grad_norm": 3.4608020782470703,
      "learning_rate": 5.139269038802125e-07,
      "loss": 0.0394,
      "step": 3147
    },
    {
      "epoch": 0.49658871317584885,
      "grad_norm": 9.678434371948242,
      "learning_rate": 5.13765899211077e-07,
      "loss": 0.0365,
      "step": 3148
    },
    {
      "epoch": 0.4967464605434397,
      "grad_norm": 5.920650005340576,
      "learning_rate": 5.136048945419416e-07,
      "loss": 0.0343,
      "step": 3149
    },
    {
      "epoch": 0.49690420791103046,
      "grad_norm": 7.1352362632751465,
      "learning_rate": 5.134438898728063e-07,
      "loss": 0.0492,
      "step": 3150
    },
    {
      "epoch": 0.4970619552786213,
      "grad_norm": 2.299804449081421,
      "learning_rate": 5.13282885203671e-07,
      "loss": 0.0239,
      "step": 3151
    },
    {
      "epoch": 0.49721970264621207,
      "grad_norm": 3.965886354446411,
      "learning_rate": 5.131218805345355e-07,
      "loss": 0.0375,
      "step": 3152
    },
    {
      "epoch": 0.4973774500138029,
      "grad_norm": 5.144937038421631,
      "learning_rate": 5.129608758654001e-07,
      "loss": 0.0323,
      "step": 3153
    },
    {
      "epoch": 0.4975351973813937,
      "grad_norm": 3.9022388458251953,
      "learning_rate": 5.127998711962647e-07,
      "loss": 0.0294,
      "step": 3154
    },
    {
      "epoch": 0.4976929447489845,
      "grad_norm": 8.25753116607666,
      "learning_rate": 5.126388665271293e-07,
      "loss": 0.026,
      "step": 3155
    },
    {
      "epoch": 0.4978506921165753,
      "grad_norm": 5.1440606117248535,
      "learning_rate": 5.124778618579939e-07,
      "loss": 0.0534,
      "step": 3156
    },
    {
      "epoch": 0.4980084394841661,
      "grad_norm": 2.1168277263641357,
      "learning_rate": 5.123168571888584e-07,
      "loss": 0.0136,
      "step": 3157
    },
    {
      "epoch": 0.4981661868517569,
      "grad_norm": 4.708176612854004,
      "learning_rate": 5.12155852519723e-07,
      "loss": 0.0433,
      "step": 3158
    },
    {
      "epoch": 0.49832393421934773,
      "grad_norm": 4.0774407386779785,
      "learning_rate": 5.119948478505877e-07,
      "loss": 0.0318,
      "step": 3159
    },
    {
      "epoch": 0.4984816815869385,
      "grad_norm": 5.956881523132324,
      "learning_rate": 5.118338431814523e-07,
      "loss": 0.0275,
      "step": 3160
    },
    {
      "epoch": 0.49863942895452934,
      "grad_norm": 7.400354385375977,
      "learning_rate": 5.116728385123168e-07,
      "loss": 0.0594,
      "step": 3161
    },
    {
      "epoch": 0.4987971763221201,
      "grad_norm": 6.836795806884766,
      "learning_rate": 5.115118338431814e-07,
      "loss": 0.0577,
      "step": 3162
    },
    {
      "epoch": 0.49895492368971095,
      "grad_norm": 4.228177547454834,
      "learning_rate": 5.11350829174046e-07,
      "loss": 0.0313,
      "step": 3163
    },
    {
      "epoch": 0.4991126710573017,
      "grad_norm": 4.468157768249512,
      "learning_rate": 5.111898245049106e-07,
      "loss": 0.0346,
      "step": 3164
    },
    {
      "epoch": 0.49927041842489256,
      "grad_norm": 5.014430999755859,
      "learning_rate": 5.110288198357751e-07,
      "loss": 0.0458,
      "step": 3165
    },
    {
      "epoch": 0.49942816579248334,
      "grad_norm": 4.235401630401611,
      "learning_rate": 5.108678151666398e-07,
      "loss": 0.0332,
      "step": 3166
    },
    {
      "epoch": 0.49958591316007417,
      "grad_norm": 4.435441970825195,
      "learning_rate": 5.107068104975045e-07,
      "loss": 0.0547,
      "step": 3167
    },
    {
      "epoch": 0.49974366052766495,
      "grad_norm": 6.001528739929199,
      "learning_rate": 5.105458058283691e-07,
      "loss": 0.0698,
      "step": 3168
    },
    {
      "epoch": 0.4999014078952557,
      "grad_norm": 3.8304319381713867,
      "learning_rate": 5.103848011592336e-07,
      "loss": 0.0239,
      "step": 3169
    },
    {
      "epoch": 0.5000591552628465,
      "grad_norm": 8.665897369384766,
      "learning_rate": 5.102237964900982e-07,
      "loss": 0.0695,
      "step": 3170
    },
    {
      "epoch": 0.5002169026304374,
      "grad_norm": 4.512013912200928,
      "learning_rate": 5.100627918209628e-07,
      "loss": 0.0348,
      "step": 3171
    },
    {
      "epoch": 0.5003746499980282,
      "grad_norm": 3.1103832721710205,
      "learning_rate": 5.099017871518274e-07,
      "loss": 0.0267,
      "step": 3172
    },
    {
      "epoch": 0.5005323973656189,
      "grad_norm": 10.89181137084961,
      "learning_rate": 5.097407824826919e-07,
      "loss": 0.0783,
      "step": 3173
    },
    {
      "epoch": 0.5006901447332097,
      "grad_norm": 6.386903762817383,
      "learning_rate": 5.095797778135566e-07,
      "loss": 0.0395,
      "step": 3174
    },
    {
      "epoch": 0.5008478921008006,
      "grad_norm": 9.237397193908691,
      "learning_rate": 5.094187731444212e-07,
      "loss": 0.039,
      "step": 3175
    },
    {
      "epoch": 0.5010056394683914,
      "grad_norm": 5.752801895141602,
      "learning_rate": 5.092577684752858e-07,
      "loss": 0.0715,
      "step": 3176
    },
    {
      "epoch": 0.5011633868359822,
      "grad_norm": 7.930622577667236,
      "learning_rate": 5.090967638061504e-07,
      "loss": 0.0375,
      "step": 3177
    },
    {
      "epoch": 0.5013211342035729,
      "grad_norm": 4.135472297668457,
      "learning_rate": 5.089357591370149e-07,
      "loss": 0.0542,
      "step": 3178
    },
    {
      "epoch": 0.5014788815711638,
      "grad_norm": 1.8181886672973633,
      "learning_rate": 5.087747544678795e-07,
      "loss": 0.0121,
      "step": 3179
    },
    {
      "epoch": 0.5016366289387546,
      "grad_norm": 9.15134334564209,
      "learning_rate": 5.086137497987441e-07,
      "loss": 0.0468,
      "step": 3180
    },
    {
      "epoch": 0.5017943763063454,
      "grad_norm": 8.41822624206543,
      "learning_rate": 5.084527451296088e-07,
      "loss": 0.0768,
      "step": 3181
    },
    {
      "epoch": 0.5019521236739362,
      "grad_norm": 8.632811546325684,
      "learning_rate": 5.082917404604733e-07,
      "loss": 0.0682,
      "step": 3182
    },
    {
      "epoch": 0.502109871041527,
      "grad_norm": 6.422868251800537,
      "learning_rate": 5.081307357913379e-07,
      "loss": 0.0613,
      "step": 3183
    },
    {
      "epoch": 0.5022676184091178,
      "grad_norm": 5.798893451690674,
      "learning_rate": 5.079697311222025e-07,
      "loss": 0.1024,
      "step": 3184
    },
    {
      "epoch": 0.5024253657767086,
      "grad_norm": 4.80502986907959,
      "learning_rate": 5.078087264530672e-07,
      "loss": 0.0473,
      "step": 3185
    },
    {
      "epoch": 0.5025831131442994,
      "grad_norm": 4.152266979217529,
      "learning_rate": 5.076477217839317e-07,
      "loss": 0.0536,
      "step": 3186
    },
    {
      "epoch": 0.5027408605118902,
      "grad_norm": 9.28447151184082,
      "learning_rate": 5.074867171147963e-07,
      "loss": 0.0701,
      "step": 3187
    },
    {
      "epoch": 0.502898607879481,
      "grad_norm": 4.702619552612305,
      "learning_rate": 5.073257124456609e-07,
      "loss": 0.0235,
      "step": 3188
    },
    {
      "epoch": 0.5030563552470718,
      "grad_norm": 13.083337783813477,
      "learning_rate": 5.071647077765256e-07,
      "loss": 0.0684,
      "step": 3189
    },
    {
      "epoch": 0.5032141026146626,
      "grad_norm": 6.463047981262207,
      "learning_rate": 5.070037031073901e-07,
      "loss": 0.0337,
      "step": 3190
    },
    {
      "epoch": 0.5033718499822534,
      "grad_norm": 4.493224143981934,
      "learning_rate": 5.068426984382547e-07,
      "loss": 0.0475,
      "step": 3191
    },
    {
      "epoch": 0.5035295973498443,
      "grad_norm": 7.519652366638184,
      "learning_rate": 5.066816937691193e-07,
      "loss": 0.0244,
      "step": 3192
    },
    {
      "epoch": 0.503687344717435,
      "grad_norm": 8.25917911529541,
      "learning_rate": 5.065206890999839e-07,
      "loss": 0.0964,
      "step": 3193
    },
    {
      "epoch": 0.5038450920850258,
      "grad_norm": 1.5638859272003174,
      "learning_rate": 5.063596844308484e-07,
      "loss": 0.013,
      "step": 3194
    },
    {
      "epoch": 0.5040028394526166,
      "grad_norm": 4.545310974121094,
      "learning_rate": 5.06198679761713e-07,
      "loss": 0.0201,
      "step": 3195
    },
    {
      "epoch": 0.5041605868202075,
      "grad_norm": 18.580646514892578,
      "learning_rate": 5.060376750925777e-07,
      "loss": 0.1505,
      "step": 3196
    },
    {
      "epoch": 0.5043183341877983,
      "grad_norm": 10.399683952331543,
      "learning_rate": 5.058766704234423e-07,
      "loss": 0.0689,
      "step": 3197
    },
    {
      "epoch": 0.504476081555389,
      "grad_norm": 5.154493808746338,
      "learning_rate": 5.057156657543068e-07,
      "loss": 0.039,
      "step": 3198
    },
    {
      "epoch": 0.5046338289229798,
      "grad_norm": 3.546096086502075,
      "learning_rate": 5.055546610851714e-07,
      "loss": 0.0701,
      "step": 3199
    },
    {
      "epoch": 0.5047915762905707,
      "grad_norm": 3.3994908332824707,
      "learning_rate": 5.05393656416036e-07,
      "loss": 0.02,
      "step": 3200
    },
    {
      "epoch": 0.5049493236581615,
      "grad_norm": 4.627392292022705,
      "learning_rate": 5.052326517469006e-07,
      "loss": 0.0547,
      "step": 3201
    },
    {
      "epoch": 0.5051070710257523,
      "grad_norm": 4.102443218231201,
      "learning_rate": 5.050716470777652e-07,
      "loss": 0.0996,
      "step": 3202
    },
    {
      "epoch": 0.505264818393343,
      "grad_norm": 3.814687490463257,
      "learning_rate": 5.049106424086297e-07,
      "loss": 0.0521,
      "step": 3203
    },
    {
      "epoch": 0.5054225657609339,
      "grad_norm": 3.8533315658569336,
      "learning_rate": 5.047496377394945e-07,
      "loss": 0.0597,
      "step": 3204
    },
    {
      "epoch": 0.5055803131285247,
      "grad_norm": 2.4979841709136963,
      "learning_rate": 5.045886330703591e-07,
      "loss": 0.0137,
      "step": 3205
    },
    {
      "epoch": 0.5057380604961155,
      "grad_norm": 6.9564595222473145,
      "learning_rate": 5.044276284012237e-07,
      "loss": 0.0576,
      "step": 3206
    },
    {
      "epoch": 0.5058958078637062,
      "grad_norm": 4.046496868133545,
      "learning_rate": 5.042666237320882e-07,
      "loss": 0.0534,
      "step": 3207
    },
    {
      "epoch": 0.506053555231297,
      "grad_norm": 5.28488826751709,
      "learning_rate": 5.041056190629528e-07,
      "loss": 0.0442,
      "step": 3208
    },
    {
      "epoch": 0.5062113025988879,
      "grad_norm": 2.557673931121826,
      "learning_rate": 5.039446143938174e-07,
      "loss": 0.0223,
      "step": 3209
    },
    {
      "epoch": 0.5063690499664787,
      "grad_norm": 4.47113561630249,
      "learning_rate": 5.03783609724682e-07,
      "loss": 0.0511,
      "step": 3210
    },
    {
      "epoch": 0.5065267973340695,
      "grad_norm": 3.6947097778320312,
      "learning_rate": 5.036226050555465e-07,
      "loss": 0.0437,
      "step": 3211
    },
    {
      "epoch": 0.5066845447016602,
      "grad_norm": 6.7887282371521,
      "learning_rate": 5.034616003864112e-07,
      "loss": 0.0456,
      "step": 3212
    },
    {
      "epoch": 0.5068422920692511,
      "grad_norm": 6.798984050750732,
      "learning_rate": 5.033005957172758e-07,
      "loss": 0.0816,
      "step": 3213
    },
    {
      "epoch": 0.5070000394368419,
      "grad_norm": 3.5052080154418945,
      "learning_rate": 5.031395910481404e-07,
      "loss": 0.0491,
      "step": 3214
    },
    {
      "epoch": 0.5071577868044327,
      "grad_norm": 3.1653616428375244,
      "learning_rate": 5.029785863790049e-07,
      "loss": 0.061,
      "step": 3215
    },
    {
      "epoch": 0.5073155341720235,
      "grad_norm": 6.844089508056641,
      "learning_rate": 5.028175817098695e-07,
      "loss": 0.0316,
      "step": 3216
    },
    {
      "epoch": 0.5074732815396144,
      "grad_norm": 4.814005374908447,
      "learning_rate": 5.026565770407341e-07,
      "loss": 0.0205,
      "step": 3217
    },
    {
      "epoch": 0.5076310289072051,
      "grad_norm": 7.20102071762085,
      "learning_rate": 5.024955723715987e-07,
      "loss": 0.0267,
      "step": 3218
    },
    {
      "epoch": 0.5077887762747959,
      "grad_norm": 2.9849743843078613,
      "learning_rate": 5.023345677024633e-07,
      "loss": 0.0361,
      "step": 3219
    },
    {
      "epoch": 0.5079465236423867,
      "grad_norm": 5.194559574127197,
      "learning_rate": 5.02173563033328e-07,
      "loss": 0.0817,
      "step": 3220
    },
    {
      "epoch": 0.5081042710099776,
      "grad_norm": 2.9066145420074463,
      "learning_rate": 5.020125583641926e-07,
      "loss": 0.0301,
      "step": 3221
    },
    {
      "epoch": 0.5082620183775683,
      "grad_norm": 1.0378518104553223,
      "learning_rate": 5.018515536950572e-07,
      "loss": 0.0082,
      "step": 3222
    },
    {
      "epoch": 0.5084197657451591,
      "grad_norm": 7.024827480316162,
      "learning_rate": 5.016905490259218e-07,
      "loss": 0.053,
      "step": 3223
    },
    {
      "epoch": 0.5085775131127499,
      "grad_norm": 3.3456127643585205,
      "learning_rate": 5.015295443567863e-07,
      "loss": 0.0311,
      "step": 3224
    },
    {
      "epoch": 0.5087352604803408,
      "grad_norm": 4.731700420379639,
      "learning_rate": 5.013685396876509e-07,
      "loss": 0.0175,
      "step": 3225
    },
    {
      "epoch": 0.5088930078479316,
      "grad_norm": 4.278652667999268,
      "learning_rate": 5.012075350185156e-07,
      "loss": 0.0186,
      "step": 3226
    },
    {
      "epoch": 0.5090507552155223,
      "grad_norm": 1.8399741649627686,
      "learning_rate": 5.010465303493802e-07,
      "loss": 0.0135,
      "step": 3227
    },
    {
      "epoch": 0.5092085025831131,
      "grad_norm": 3.6055192947387695,
      "learning_rate": 5.008855256802447e-07,
      "loss": 0.0453,
      "step": 3228
    },
    {
      "epoch": 0.5093662499507039,
      "grad_norm": 5.718883991241455,
      "learning_rate": 5.007245210111093e-07,
      "loss": 0.0458,
      "step": 3229
    },
    {
      "epoch": 0.5095239973182948,
      "grad_norm": 2.1946067810058594,
      "learning_rate": 5.005635163419739e-07,
      "loss": 0.0119,
      "step": 3230
    },
    {
      "epoch": 0.5096817446858856,
      "grad_norm": 5.639341831207275,
      "learning_rate": 5.004025116728385e-07,
      "loss": 0.0614,
      "step": 3231
    },
    {
      "epoch": 0.5098394920534763,
      "grad_norm": 8.0573148727417,
      "learning_rate": 5.00241507003703e-07,
      "loss": 0.0545,
      "step": 3232
    },
    {
      "epoch": 0.5099972394210671,
      "grad_norm": 6.907105445861816,
      "learning_rate": 5.000805023345676e-07,
      "loss": 0.0691,
      "step": 3233
    },
    {
      "epoch": 0.510154986788658,
      "grad_norm": 5.254229545593262,
      "learning_rate": 4.999194976654323e-07,
      "loss": 0.0162,
      "step": 3234
    },
    {
      "epoch": 0.5103127341562488,
      "grad_norm": 6.818365097045898,
      "learning_rate": 4.997584929962968e-07,
      "loss": 0.0641,
      "step": 3235
    },
    {
      "epoch": 0.5104704815238396,
      "grad_norm": 6.1585235595703125,
      "learning_rate": 4.995974883271614e-07,
      "loss": 0.0598,
      "step": 3236
    },
    {
      "epoch": 0.5106282288914303,
      "grad_norm": 12.957852363586426,
      "learning_rate": 4.99436483658026e-07,
      "loss": 0.0614,
      "step": 3237
    },
    {
      "epoch": 0.5107859762590212,
      "grad_norm": 5.405330657958984,
      "learning_rate": 4.992754789888907e-07,
      "loss": 0.029,
      "step": 3238
    },
    {
      "epoch": 0.510943723626612,
      "grad_norm": 3.506260871887207,
      "learning_rate": 4.991144743197553e-07,
      "loss": 0.058,
      "step": 3239
    },
    {
      "epoch": 0.5111014709942028,
      "grad_norm": 4.510875225067139,
      "learning_rate": 4.989534696506199e-07,
      "loss": 0.0475,
      "step": 3240
    },
    {
      "epoch": 0.5112592183617936,
      "grad_norm": 3.320171356201172,
      "learning_rate": 4.987924649814844e-07,
      "loss": 0.0306,
      "step": 3241
    },
    {
      "epoch": 0.5114169657293844,
      "grad_norm": 5.9775471687316895,
      "learning_rate": 4.986314603123491e-07,
      "loss": 0.0306,
      "step": 3242
    },
    {
      "epoch": 0.5115747130969752,
      "grad_norm": 6.424386501312256,
      "learning_rate": 4.984704556432136e-07,
      "loss": 0.0405,
      "step": 3243
    },
    {
      "epoch": 0.511732460464566,
      "grad_norm": 4.2295074462890625,
      "learning_rate": 4.983094509740782e-07,
      "loss": 0.0497,
      "step": 3244
    },
    {
      "epoch": 0.5118902078321568,
      "grad_norm": 6.756084442138672,
      "learning_rate": 4.981484463049428e-07,
      "loss": 0.0642,
      "step": 3245
    },
    {
      "epoch": 0.5120479551997476,
      "grad_norm": 4.199583053588867,
      "learning_rate": 4.979874416358074e-07,
      "loss": 0.044,
      "step": 3246
    },
    {
      "epoch": 0.5122057025673384,
      "grad_norm": 3.0998520851135254,
      "learning_rate": 4.97826436966672e-07,
      "loss": 0.0295,
      "step": 3247
    },
    {
      "epoch": 0.5123634499349292,
      "grad_norm": 8.630373001098633,
      "learning_rate": 4.976654322975366e-07,
      "loss": 0.097,
      "step": 3248
    },
    {
      "epoch": 0.51252119730252,
      "grad_norm": 3.19980525970459,
      "learning_rate": 4.975044276284012e-07,
      "loss": 0.0329,
      "step": 3249
    },
    {
      "epoch": 0.5126789446701108,
      "grad_norm": 3.5462422370910645,
      "learning_rate": 4.973434229592658e-07,
      "loss": 0.072,
      "step": 3250
    },
    {
      "epoch": 0.5128366920377017,
      "grad_norm": 4.932615756988525,
      "learning_rate": 4.971824182901304e-07,
      "loss": 0.0243,
      "step": 3251
    },
    {
      "epoch": 0.5129944394052924,
      "grad_norm": 4.566836357116699,
      "learning_rate": 4.970214136209949e-07,
      "loss": 0.0441,
      "step": 3252
    },
    {
      "epoch": 0.5131521867728832,
      "grad_norm": 3.056807279586792,
      "learning_rate": 4.968604089518596e-07,
      "loss": 0.0196,
      "step": 3253
    },
    {
      "epoch": 0.513309934140474,
      "grad_norm": 6.167160987854004,
      "learning_rate": 4.966994042827241e-07,
      "loss": 0.0564,
      "step": 3254
    },
    {
      "epoch": 0.5134676815080649,
      "grad_norm": 6.918657302856445,
      "learning_rate": 4.965383996135887e-07,
      "loss": 0.0687,
      "step": 3255
    },
    {
      "epoch": 0.5136254288756557,
      "grad_norm": 6.711324214935303,
      "learning_rate": 4.963773949444534e-07,
      "loss": 0.0826,
      "step": 3256
    },
    {
      "epoch": 0.5137831762432464,
      "grad_norm": 5.007664203643799,
      "learning_rate": 4.96216390275318e-07,
      "loss": 0.026,
      "step": 3257
    },
    {
      "epoch": 0.5139409236108372,
      "grad_norm": 7.329284191131592,
      "learning_rate": 4.960553856061826e-07,
      "loss": 0.0754,
      "step": 3258
    },
    {
      "epoch": 0.5140986709784281,
      "grad_norm": 5.177643775939941,
      "learning_rate": 4.958943809370472e-07,
      "loss": 0.0217,
      "step": 3259
    },
    {
      "epoch": 0.5142564183460189,
      "grad_norm": 4.8610076904296875,
      "learning_rate": 4.957333762679118e-07,
      "loss": 0.0401,
      "step": 3260
    },
    {
      "epoch": 0.5144141657136097,
      "grad_norm": 7.865502834320068,
      "learning_rate": 4.955723715987764e-07,
      "loss": 0.0476,
      "step": 3261
    },
    {
      "epoch": 0.5145719130812004,
      "grad_norm": 2.3840301036834717,
      "learning_rate": 4.954113669296409e-07,
      "loss": 0.0136,
      "step": 3262
    },
    {
      "epoch": 0.5147296604487913,
      "grad_norm": 5.249300479888916,
      "learning_rate": 4.952503622605055e-07,
      "loss": 0.0661,
      "step": 3263
    },
    {
      "epoch": 0.5148874078163821,
      "grad_norm": 3.7689268589019775,
      "learning_rate": 4.950893575913701e-07,
      "loss": 0.0366,
      "step": 3264
    },
    {
      "epoch": 0.5150451551839729,
      "grad_norm": 6.984394550323486,
      "learning_rate": 4.949283529222347e-07,
      "loss": 0.0762,
      "step": 3265
    },
    {
      "epoch": 0.5152029025515636,
      "grad_norm": 3.6469602584838867,
      "learning_rate": 4.947673482530993e-07,
      "loss": 0.0213,
      "step": 3266
    },
    {
      "epoch": 0.5153606499191544,
      "grad_norm": 5.703409194946289,
      "learning_rate": 4.946063435839639e-07,
      "loss": 0.0646,
      "step": 3267
    },
    {
      "epoch": 0.5155183972867453,
      "grad_norm": 6.856706619262695,
      "learning_rate": 4.944453389148285e-07,
      "loss": 0.0539,
      "step": 3268
    },
    {
      "epoch": 0.5156761446543361,
      "grad_norm": 3.3388614654541016,
      "learning_rate": 4.942843342456931e-07,
      "loss": 0.0436,
      "step": 3269
    },
    {
      "epoch": 0.5158338920219269,
      "grad_norm": 4.2724528312683105,
      "learning_rate": 4.941233295765577e-07,
      "loss": 0.0298,
      "step": 3270
    },
    {
      "epoch": 0.5159916393895176,
      "grad_norm": 8.368443489074707,
      "learning_rate": 4.939623249074222e-07,
      "loss": 0.0533,
      "step": 3271
    },
    {
      "epoch": 0.5161493867571085,
      "grad_norm": 2.5894930362701416,
      "learning_rate": 4.93801320238287e-07,
      "loss": 0.0225,
      "step": 3272
    },
    {
      "epoch": 0.5163071341246993,
      "grad_norm": 9.17452621459961,
      "learning_rate": 4.936403155691515e-07,
      "loss": 0.0401,
      "step": 3273
    },
    {
      "epoch": 0.5164648814922901,
      "grad_norm": 4.909448623657227,
      "learning_rate": 4.934793109000161e-07,
      "loss": 0.0384,
      "step": 3274
    },
    {
      "epoch": 0.5166226288598809,
      "grad_norm": 5.070178508758545,
      "learning_rate": 4.933183062308807e-07,
      "loss": 0.0648,
      "step": 3275
    },
    {
      "epoch": 0.5167803762274717,
      "grad_norm": 5.238038539886475,
      "learning_rate": 4.931573015617453e-07,
      "loss": 0.0428,
      "step": 3276
    },
    {
      "epoch": 0.5169381235950625,
      "grad_norm": 6.299678802490234,
      "learning_rate": 4.929962968926099e-07,
      "loss": 0.0649,
      "step": 3277
    },
    {
      "epoch": 0.5170958709626533,
      "grad_norm": 6.64601469039917,
      "learning_rate": 4.928352922234745e-07,
      "loss": 0.0456,
      "step": 3278
    },
    {
      "epoch": 0.5172536183302441,
      "grad_norm": 8.029877662658691,
      "learning_rate": 4.926742875543391e-07,
      "loss": 0.0942,
      "step": 3279
    },
    {
      "epoch": 0.517411365697835,
      "grad_norm": 3.4317123889923096,
      "learning_rate": 4.925132828852037e-07,
      "loss": 0.0171,
      "step": 3280
    },
    {
      "epoch": 0.5175691130654257,
      "grad_norm": 8.241918563842773,
      "learning_rate": 4.923522782160682e-07,
      "loss": 0.043,
      "step": 3281
    },
    {
      "epoch": 0.5177268604330165,
      "grad_norm": 4.322054386138916,
      "learning_rate": 4.921912735469328e-07,
      "loss": 0.0377,
      "step": 3282
    },
    {
      "epoch": 0.5178846078006073,
      "grad_norm": 3.4600963592529297,
      "learning_rate": 4.920302688777974e-07,
      "loss": 0.0574,
      "step": 3283
    },
    {
      "epoch": 0.5180423551681981,
      "grad_norm": 5.678745746612549,
      "learning_rate": 4.91869264208662e-07,
      "loss": 0.0507,
      "step": 3284
    },
    {
      "epoch": 0.518200102535789,
      "grad_norm": 5.465057373046875,
      "learning_rate": 4.917082595395266e-07,
      "loss": 0.04,
      "step": 3285
    },
    {
      "epoch": 0.5183578499033797,
      "grad_norm": 4.774681568145752,
      "learning_rate": 4.915472548703912e-07,
      "loss": 0.0463,
      "step": 3286
    },
    {
      "epoch": 0.5185155972709705,
      "grad_norm": 7.449644088745117,
      "learning_rate": 4.913862502012558e-07,
      "loss": 0.0871,
      "step": 3287
    },
    {
      "epoch": 0.5186733446385613,
      "grad_norm": 5.768908500671387,
      "learning_rate": 4.912252455321204e-07,
      "loss": 0.056,
      "step": 3288
    },
    {
      "epoch": 0.5188310920061522,
      "grad_norm": 4.329742908477783,
      "learning_rate": 4.910642408629849e-07,
      "loss": 0.0313,
      "step": 3289
    },
    {
      "epoch": 0.518988839373743,
      "grad_norm": 3.8610966205596924,
      "learning_rate": 4.909032361938497e-07,
      "loss": 0.0284,
      "step": 3290
    },
    {
      "epoch": 0.5191465867413337,
      "grad_norm": 5.501697540283203,
      "learning_rate": 4.907422315247142e-07,
      "loss": 0.0617,
      "step": 3291
    },
    {
      "epoch": 0.5193043341089245,
      "grad_norm": 2.665956497192383,
      "learning_rate": 4.905812268555788e-07,
      "loss": 0.0299,
      "step": 3292
    },
    {
      "epoch": 0.5194620814765154,
      "grad_norm": 6.519895076751709,
      "learning_rate": 4.904202221864434e-07,
      "loss": 0.082,
      "step": 3293
    },
    {
      "epoch": 0.5196198288441062,
      "grad_norm": 9.04831314086914,
      "learning_rate": 4.90259217517308e-07,
      "loss": 0.0611,
      "step": 3294
    },
    {
      "epoch": 0.519777576211697,
      "grad_norm": 5.466322898864746,
      "learning_rate": 4.900982128481726e-07,
      "loss": 0.0283,
      "step": 3295
    },
    {
      "epoch": 0.5199353235792877,
      "grad_norm": 3.772711992263794,
      "learning_rate": 4.899372081790372e-07,
      "loss": 0.0278,
      "step": 3296
    },
    {
      "epoch": 0.5200930709468786,
      "grad_norm": 4.77584981918335,
      "learning_rate": 4.897762035099018e-07,
      "loss": 0.0745,
      "step": 3297
    },
    {
      "epoch": 0.5202508183144694,
      "grad_norm": 4.437483787536621,
      "learning_rate": 4.896151988407664e-07,
      "loss": 0.0628,
      "step": 3298
    },
    {
      "epoch": 0.5204085656820602,
      "grad_norm": 7.5456318855285645,
      "learning_rate": 4.89454194171631e-07,
      "loss": 0.0256,
      "step": 3299
    },
    {
      "epoch": 0.520566313049651,
      "grad_norm": 2.4569687843322754,
      "learning_rate": 4.892931895024955e-07,
      "loss": 0.0174,
      "step": 3300
    },
    {
      "epoch": 0.5207240604172418,
      "grad_norm": 4.919608116149902,
      "learning_rate": 4.891321848333601e-07,
      "loss": 0.0366,
      "step": 3301
    },
    {
      "epoch": 0.5208818077848326,
      "grad_norm": 6.518052101135254,
      "learning_rate": 4.889711801642247e-07,
      "loss": 0.0594,
      "step": 3302
    },
    {
      "epoch": 0.5210395551524234,
      "grad_norm": 2.119096517562866,
      "learning_rate": 4.888101754950893e-07,
      "loss": 0.0145,
      "step": 3303
    },
    {
      "epoch": 0.5211973025200142,
      "grad_norm": 5.68674898147583,
      "learning_rate": 4.886491708259539e-07,
      "loss": 0.0382,
      "step": 3304
    },
    {
      "epoch": 0.521355049887605,
      "grad_norm": 4.043545722961426,
      "learning_rate": 4.884881661568185e-07,
      "loss": 0.0218,
      "step": 3305
    },
    {
      "epoch": 0.5215127972551958,
      "grad_norm": 7.708515644073486,
      "learning_rate": 4.883271614876831e-07,
      "loss": 0.087,
      "step": 3306
    },
    {
      "epoch": 0.5216705446227866,
      "grad_norm": 6.87803316116333,
      "learning_rate": 4.881661568185477e-07,
      "loss": 0.1043,
      "step": 3307
    },
    {
      "epoch": 0.5218282919903774,
      "grad_norm": 6.581982612609863,
      "learning_rate": 4.880051521494122e-07,
      "loss": 0.0574,
      "step": 3308
    },
    {
      "epoch": 0.5219860393579682,
      "grad_norm": 5.4126458168029785,
      "learning_rate": 4.87844147480277e-07,
      "loss": 0.0517,
      "step": 3309
    },
    {
      "epoch": 0.5221437867255591,
      "grad_norm": 5.3862385749816895,
      "learning_rate": 4.876831428111415e-07,
      "loss": 0.0655,
      "step": 3310
    },
    {
      "epoch": 0.5223015340931498,
      "grad_norm": 8.10891056060791,
      "learning_rate": 4.875221381420061e-07,
      "loss": 0.0664,
      "step": 3311
    },
    {
      "epoch": 0.5224592814607406,
      "grad_norm": 8.003843307495117,
      "learning_rate": 4.873611334728707e-07,
      "loss": 0.0771,
      "step": 3312
    },
    {
      "epoch": 0.5226170288283314,
      "grad_norm": 4.2681989669799805,
      "learning_rate": 4.872001288037353e-07,
      "loss": 0.0245,
      "step": 3313
    },
    {
      "epoch": 0.5227747761959223,
      "grad_norm": 3.433547019958496,
      "learning_rate": 4.870391241345999e-07,
      "loss": 0.0145,
      "step": 3314
    },
    {
      "epoch": 0.522932523563513,
      "grad_norm": 2.166034698486328,
      "learning_rate": 4.868781194654645e-07,
      "loss": 0.0134,
      "step": 3315
    },
    {
      "epoch": 0.5230902709311038,
      "grad_norm": 2.5870463848114014,
      "learning_rate": 4.867171147963291e-07,
      "loss": 0.012,
      "step": 3316
    },
    {
      "epoch": 0.5232480182986946,
      "grad_norm": 2.9055380821228027,
      "learning_rate": 4.865561101271937e-07,
      "loss": 0.0302,
      "step": 3317
    },
    {
      "epoch": 0.5234057656662855,
      "grad_norm": 4.043204307556152,
      "learning_rate": 4.863951054580583e-07,
      "loss": 0.0554,
      "step": 3318
    },
    {
      "epoch": 0.5235635130338763,
      "grad_norm": 5.544101715087891,
      "learning_rate": 4.862341007889228e-07,
      "loss": 0.0413,
      "step": 3319
    },
    {
      "epoch": 0.523721260401467,
      "grad_norm": 3.7651774883270264,
      "learning_rate": 4.860730961197875e-07,
      "loss": 0.0378,
      "step": 3320
    },
    {
      "epoch": 0.5238790077690578,
      "grad_norm": 3.9472198486328125,
      "learning_rate": 4.85912091450652e-07,
      "loss": 0.0482,
      "step": 3321
    },
    {
      "epoch": 0.5240367551366486,
      "grad_norm": 9.249954223632812,
      "learning_rate": 4.857510867815166e-07,
      "loss": 0.0849,
      "step": 3322
    },
    {
      "epoch": 0.5241945025042395,
      "grad_norm": 5.982876300811768,
      "learning_rate": 4.855900821123812e-07,
      "loss": 0.0421,
      "step": 3323
    },
    {
      "epoch": 0.5243522498718303,
      "grad_norm": 53.606571197509766,
      "learning_rate": 4.854290774432458e-07,
      "loss": 0.0513,
      "step": 3324
    },
    {
      "epoch": 0.524509997239421,
      "grad_norm": 1.2392445802688599,
      "learning_rate": 4.852680727741104e-07,
      "loss": 0.01,
      "step": 3325
    },
    {
      "epoch": 0.5246677446070118,
      "grad_norm": 6.681481838226318,
      "learning_rate": 4.851070681049751e-07,
      "loss": 0.0309,
      "step": 3326
    },
    {
      "epoch": 0.5248254919746027,
      "grad_norm": 8.890432357788086,
      "learning_rate": 4.849460634358396e-07,
      "loss": 0.0671,
      "step": 3327
    },
    {
      "epoch": 0.5249832393421935,
      "grad_norm": 4.494774341583252,
      "learning_rate": 4.847850587667043e-07,
      "loss": 0.0346,
      "step": 3328
    },
    {
      "epoch": 0.5251409867097843,
      "grad_norm": 3.080538272857666,
      "learning_rate": 4.846240540975688e-07,
      "loss": 0.0134,
      "step": 3329
    },
    {
      "epoch": 0.525298734077375,
      "grad_norm": 5.779559135437012,
      "learning_rate": 4.844630494284334e-07,
      "loss": 0.0403,
      "step": 3330
    },
    {
      "epoch": 0.5254564814449659,
      "grad_norm": 15.656715393066406,
      "learning_rate": 4.84302044759298e-07,
      "loss": 0.1181,
      "step": 3331
    },
    {
      "epoch": 0.5256142288125567,
      "grad_norm": 5.662373065948486,
      "learning_rate": 4.841410400901626e-07,
      "loss": 0.0813,
      "step": 3332
    },
    {
      "epoch": 0.5257719761801475,
      "grad_norm": 3.515956163406372,
      "learning_rate": 4.839800354210272e-07,
      "loss": 0.0248,
      "step": 3333
    },
    {
      "epoch": 0.5259297235477383,
      "grad_norm": 4.553384304046631,
      "learning_rate": 4.838190307518918e-07,
      "loss": 0.017,
      "step": 3334
    },
    {
      "epoch": 0.5260874709153291,
      "grad_norm": 1.009729266166687,
      "learning_rate": 4.836580260827564e-07,
      "loss": 0.0068,
      "step": 3335
    },
    {
      "epoch": 0.5262452182829199,
      "grad_norm": 3.6028075218200684,
      "learning_rate": 4.83497021413621e-07,
      "loss": 0.0338,
      "step": 3336
    },
    {
      "epoch": 0.5264029656505107,
      "grad_norm": 4.565405368804932,
      "learning_rate": 4.833360167444855e-07,
      "loss": 0.0383,
      "step": 3337
    },
    {
      "epoch": 0.5265607130181015,
      "grad_norm": 6.493062973022461,
      "learning_rate": 4.831750120753501e-07,
      "loss": 0.0615,
      "step": 3338
    },
    {
      "epoch": 0.5267184603856924,
      "grad_norm": 4.326895236968994,
      "learning_rate": 4.830140074062148e-07,
      "loss": 0.0196,
      "step": 3339
    },
    {
      "epoch": 0.5268762077532831,
      "grad_norm": 3.374591588973999,
      "learning_rate": 4.828530027370793e-07,
      "loss": 0.0305,
      "step": 3340
    },
    {
      "epoch": 0.5270339551208739,
      "grad_norm": 5.829579830169678,
      "learning_rate": 4.826919980679439e-07,
      "loss": 0.0316,
      "step": 3341
    },
    {
      "epoch": 0.5271917024884647,
      "grad_norm": 5.114638328552246,
      "learning_rate": 4.825309933988085e-07,
      "loss": 0.0455,
      "step": 3342
    },
    {
      "epoch": 0.5273494498560555,
      "grad_norm": 4.803951740264893,
      "learning_rate": 4.823699887296732e-07,
      "loss": 0.033,
      "step": 3343
    },
    {
      "epoch": 0.5275071972236464,
      "grad_norm": 5.90579080581665,
      "learning_rate": 4.822089840605378e-07,
      "loss": 0.0466,
      "step": 3344
    },
    {
      "epoch": 0.5276649445912371,
      "grad_norm": 3.841439962387085,
      "learning_rate": 4.820479793914024e-07,
      "loss": 0.0698,
      "step": 3345
    },
    {
      "epoch": 0.5278226919588279,
      "grad_norm": 4.033648490905762,
      "learning_rate": 4.818869747222669e-07,
      "loss": 0.0406,
      "step": 3346
    },
    {
      "epoch": 0.5279804393264187,
      "grad_norm": 5.283125400543213,
      "learning_rate": 4.817259700531316e-07,
      "loss": 0.0563,
      "step": 3347
    },
    {
      "epoch": 0.5281381866940096,
      "grad_norm": 2.5686020851135254,
      "learning_rate": 4.815649653839961e-07,
      "loss": 0.0134,
      "step": 3348
    },
    {
      "epoch": 0.5282959340616004,
      "grad_norm": 4.076895236968994,
      "learning_rate": 4.814039607148607e-07,
      "loss": 0.072,
      "step": 3349
    },
    {
      "epoch": 0.5284536814291911,
      "grad_norm": 5.695146560668945,
      "learning_rate": 4.812429560457253e-07,
      "loss": 0.0476,
      "step": 3350
    },
    {
      "epoch": 0.5286114287967819,
      "grad_norm": 6.99587869644165,
      "learning_rate": 4.810819513765899e-07,
      "loss": 0.0783,
      "step": 3351
    },
    {
      "epoch": 0.5287691761643728,
      "grad_norm": 3.8918187618255615,
      "learning_rate": 4.809209467074545e-07,
      "loss": 0.0522,
      "step": 3352
    },
    {
      "epoch": 0.5289269235319636,
      "grad_norm": 5.451104164123535,
      "learning_rate": 4.807599420383191e-07,
      "loss": 0.0443,
      "step": 3353
    },
    {
      "epoch": 0.5290846708995544,
      "grad_norm": 5.873027801513672,
      "learning_rate": 4.805989373691837e-07,
      "loss": 0.0751,
      "step": 3354
    },
    {
      "epoch": 0.5292424182671451,
      "grad_norm": 4.282705783843994,
      "learning_rate": 4.804379327000483e-07,
      "loss": 0.0341,
      "step": 3355
    },
    {
      "epoch": 0.529400165634736,
      "grad_norm": 6.071550369262695,
      "learning_rate": 4.802769280309128e-07,
      "loss": 0.0269,
      "step": 3356
    },
    {
      "epoch": 0.5295579130023268,
      "grad_norm": 8.23840618133545,
      "learning_rate": 4.801159233617774e-07,
      "loss": 0.076,
      "step": 3357
    },
    {
      "epoch": 0.5297156603699176,
      "grad_norm": 4.4299445152282715,
      "learning_rate": 4.79954918692642e-07,
      "loss": 0.0623,
      "step": 3358
    },
    {
      "epoch": 0.5298734077375084,
      "grad_norm": 3.180269956588745,
      "learning_rate": 4.797939140235066e-07,
      "loss": 0.0258,
      "step": 3359
    },
    {
      "epoch": 0.5300311551050992,
      "grad_norm": 8.468398094177246,
      "learning_rate": 4.796329093543712e-07,
      "loss": 0.038,
      "step": 3360
    },
    {
      "epoch": 0.53018890247269,
      "grad_norm": 5.929890155792236,
      "learning_rate": 4.794719046852359e-07,
      "loss": 0.0591,
      "step": 3361
    },
    {
      "epoch": 0.5303466498402808,
      "grad_norm": 3.943211078643799,
      "learning_rate": 4.793109000161005e-07,
      "loss": 0.0222,
      "step": 3362
    },
    {
      "epoch": 0.5305043972078716,
      "grad_norm": 2.4576566219329834,
      "learning_rate": 4.791498953469651e-07,
      "loss": 0.0434,
      "step": 3363
    },
    {
      "epoch": 0.5306621445754623,
      "grad_norm": 5.097107887268066,
      "learning_rate": 4.789888906778297e-07,
      "loss": 0.058,
      "step": 3364
    },
    {
      "epoch": 0.5308198919430532,
      "grad_norm": 2.5385961532592773,
      "learning_rate": 4.788278860086943e-07,
      "loss": 0.0179,
      "step": 3365
    },
    {
      "epoch": 0.530977639310644,
      "grad_norm": 4.564829349517822,
      "learning_rate": 4.786668813395589e-07,
      "loss": 0.0831,
      "step": 3366
    },
    {
      "epoch": 0.5311353866782348,
      "grad_norm": 5.847906112670898,
      "learning_rate": 4.785058766704234e-07,
      "loss": 0.0614,
      "step": 3367
    },
    {
      "epoch": 0.5312931340458256,
      "grad_norm": 4.3448052406311035,
      "learning_rate": 4.78344872001288e-07,
      "loss": 0.0612,
      "step": 3368
    },
    {
      "epoch": 0.5314508814134165,
      "grad_norm": 3.5759341716766357,
      "learning_rate": 4.781838673321526e-07,
      "loss": 0.0267,
      "step": 3369
    },
    {
      "epoch": 0.5316086287810072,
      "grad_norm": 6.660635471343994,
      "learning_rate": 4.780228626630172e-07,
      "loss": 0.0415,
      "step": 3370
    },
    {
      "epoch": 0.531766376148598,
      "grad_norm": 8.639548301696777,
      "learning_rate": 4.778618579938818e-07,
      "loss": 0.0597,
      "step": 3371
    },
    {
      "epoch": 0.5319241235161888,
      "grad_norm": 4.539547443389893,
      "learning_rate": 4.777008533247464e-07,
      "loss": 0.0328,
      "step": 3372
    },
    {
      "epoch": 0.5320818708837797,
      "grad_norm": 51.51942443847656,
      "learning_rate": 4.77539848655611e-07,
      "loss": 0.078,
      "step": 3373
    },
    {
      "epoch": 0.5322396182513704,
      "grad_norm": 4.7608642578125,
      "learning_rate": 4.773788439864756e-07,
      "loss": 0.0426,
      "step": 3374
    },
    {
      "epoch": 0.5323973656189612,
      "grad_norm": 7.365528583526611,
      "learning_rate": 4.772178393173401e-07,
      "loss": 0.0379,
      "step": 3375
    },
    {
      "epoch": 0.532555112986552,
      "grad_norm": 4.892368793487549,
      "learning_rate": 4.770568346482047e-07,
      "loss": 0.0493,
      "step": 3376
    },
    {
      "epoch": 0.5327128603541429,
      "grad_norm": 6.183288097381592,
      "learning_rate": 4.768958299790693e-07,
      "loss": 0.0611,
      "step": 3377
    },
    {
      "epoch": 0.5328706077217337,
      "grad_norm": 6.039890766143799,
      "learning_rate": 4.7673482530993395e-07,
      "loss": 0.1078,
      "step": 3378
    },
    {
      "epoch": 0.5330283550893244,
      "grad_norm": 7.143927574157715,
      "learning_rate": 4.7657382064079856e-07,
      "loss": 0.0925,
      "step": 3379
    },
    {
      "epoch": 0.5331861024569152,
      "grad_norm": 5.521488666534424,
      "learning_rate": 4.7641281597166316e-07,
      "loss": 0.0179,
      "step": 3380
    },
    {
      "epoch": 0.533343849824506,
      "grad_norm": 7.703474044799805,
      "learning_rate": 4.762518113025277e-07,
      "loss": 0.0355,
      "step": 3381
    },
    {
      "epoch": 0.5335015971920969,
      "grad_norm": 3.419008255004883,
      "learning_rate": 4.760908066333924e-07,
      "loss": 0.0234,
      "step": 3382
    },
    {
      "epoch": 0.5336593445596877,
      "grad_norm": 6.611175537109375,
      "learning_rate": 4.7592980196425693e-07,
      "loss": 0.0407,
      "step": 3383
    },
    {
      "epoch": 0.5338170919272784,
      "grad_norm": 6.516146183013916,
      "learning_rate": 4.7576879729512153e-07,
      "loss": 0.0252,
      "step": 3384
    },
    {
      "epoch": 0.5339748392948692,
      "grad_norm": 4.867481708526611,
      "learning_rate": 4.7560779262598614e-07,
      "loss": 0.0462,
      "step": 3385
    },
    {
      "epoch": 0.5341325866624601,
      "grad_norm": 7.547916412353516,
      "learning_rate": 4.7544678795685075e-07,
      "loss": 0.0423,
      "step": 3386
    },
    {
      "epoch": 0.5342903340300509,
      "grad_norm": 4.116353988647461,
      "learning_rate": 4.7528578328771535e-07,
      "loss": 0.0303,
      "step": 3387
    },
    {
      "epoch": 0.5344480813976417,
      "grad_norm": 5.257046699523926,
      "learning_rate": 4.751247786185799e-07,
      "loss": 0.0622,
      "step": 3388
    },
    {
      "epoch": 0.5346058287652324,
      "grad_norm": 2.8776915073394775,
      "learning_rate": 4.749637739494445e-07,
      "loss": 0.0421,
      "step": 3389
    },
    {
      "epoch": 0.5347635761328233,
      "grad_norm": 8.064032554626465,
      "learning_rate": 4.748027692803091e-07,
      "loss": 0.0531,
      "step": 3390
    },
    {
      "epoch": 0.5349213235004141,
      "grad_norm": 4.772770404815674,
      "learning_rate": 4.746417646111737e-07,
      "loss": 0.0409,
      "step": 3391
    },
    {
      "epoch": 0.5350790708680049,
      "grad_norm": 7.8889360427856445,
      "learning_rate": 4.744807599420383e-07,
      "loss": 0.0888,
      "step": 3392
    },
    {
      "epoch": 0.5352368182355957,
      "grad_norm": 4.489109516143799,
      "learning_rate": 4.743197552729029e-07,
      "loss": 0.0489,
      "step": 3393
    },
    {
      "epoch": 0.5353945656031865,
      "grad_norm": 4.8228864669799805,
      "learning_rate": 4.741587506037675e-07,
      "loss": 0.037,
      "step": 3394
    },
    {
      "epoch": 0.5355523129707773,
      "grad_norm": 5.193973064422607,
      "learning_rate": 4.739977459346321e-07,
      "loss": 0.0576,
      "step": 3395
    },
    {
      "epoch": 0.5357100603383681,
      "grad_norm": 4.263784408569336,
      "learning_rate": 4.7383674126549665e-07,
      "loss": 0.0334,
      "step": 3396
    },
    {
      "epoch": 0.5358678077059589,
      "grad_norm": 2.4240798950195312,
      "learning_rate": 4.736757365963613e-07,
      "loss": 0.0224,
      "step": 3397
    },
    {
      "epoch": 0.5360255550735498,
      "grad_norm": 5.9223408699035645,
      "learning_rate": 4.7351473192722586e-07,
      "loss": 0.0561,
      "step": 3398
    },
    {
      "epoch": 0.5361833024411405,
      "grad_norm": 2.8138952255249023,
      "learning_rate": 4.7335372725809047e-07,
      "loss": 0.0387,
      "step": 3399
    },
    {
      "epoch": 0.5363410498087313,
      "grad_norm": 8.36697769165039,
      "learning_rate": 4.73192722588955e-07,
      "loss": 0.0304,
      "step": 3400
    },
    {
      "epoch": 0.5364987971763221,
      "grad_norm": 3.208568572998047,
      "learning_rate": 4.730317179198197e-07,
      "loss": 0.0454,
      "step": 3401
    },
    {
      "epoch": 0.5366565445439129,
      "grad_norm": 6.757406711578369,
      "learning_rate": 4.7287071325068424e-07,
      "loss": 0.0463,
      "step": 3402
    },
    {
      "epoch": 0.5368142919115038,
      "grad_norm": 7.119345664978027,
      "learning_rate": 4.7270970858154884e-07,
      "loss": 0.0323,
      "step": 3403
    },
    {
      "epoch": 0.5369720392790945,
      "grad_norm": 6.767031192779541,
      "learning_rate": 4.725487039124134e-07,
      "loss": 0.0429,
      "step": 3404
    },
    {
      "epoch": 0.5371297866466853,
      "grad_norm": 3.0922765731811523,
      "learning_rate": 4.7238769924327805e-07,
      "loss": 0.0472,
      "step": 3405
    },
    {
      "epoch": 0.5372875340142761,
      "grad_norm": 4.668428421020508,
      "learning_rate": 4.722266945741426e-07,
      "loss": 0.0537,
      "step": 3406
    },
    {
      "epoch": 0.537445281381867,
      "grad_norm": 2.73225998878479,
      "learning_rate": 4.720656899050072e-07,
      "loss": 0.0197,
      "step": 3407
    },
    {
      "epoch": 0.5376030287494578,
      "grad_norm": 4.740433692932129,
      "learning_rate": 4.719046852358718e-07,
      "loss": 0.0222,
      "step": 3408
    },
    {
      "epoch": 0.5377607761170485,
      "grad_norm": 3.2959320545196533,
      "learning_rate": 4.7174368056673643e-07,
      "loss": 0.0237,
      "step": 3409
    },
    {
      "epoch": 0.5379185234846393,
      "grad_norm": 7.326497554779053,
      "learning_rate": 4.7158267589760103e-07,
      "loss": 0.0294,
      "step": 3410
    },
    {
      "epoch": 0.5380762708522302,
      "grad_norm": 6.463442325592041,
      "learning_rate": 4.714216712284656e-07,
      "loss": 0.0528,
      "step": 3411
    },
    {
      "epoch": 0.538234018219821,
      "grad_norm": 3.855602979660034,
      "learning_rate": 4.7126066655933025e-07,
      "loss": 0.0431,
      "step": 3412
    },
    {
      "epoch": 0.5383917655874118,
      "grad_norm": 3.002051830291748,
      "learning_rate": 4.710996618901948e-07,
      "loss": 0.0461,
      "step": 3413
    },
    {
      "epoch": 0.5385495129550025,
      "grad_norm": 4.882457256317139,
      "learning_rate": 4.709386572210594e-07,
      "loss": 0.0435,
      "step": 3414
    },
    {
      "epoch": 0.5387072603225934,
      "grad_norm": 5.218534469604492,
      "learning_rate": 4.7077765255192396e-07,
      "loss": 0.0467,
      "step": 3415
    },
    {
      "epoch": 0.5388650076901842,
      "grad_norm": 2.1747610569000244,
      "learning_rate": 4.706166478827886e-07,
      "loss": 0.0134,
      "step": 3416
    },
    {
      "epoch": 0.539022755057775,
      "grad_norm": 7.50874662399292,
      "learning_rate": 4.7045564321365317e-07,
      "loss": 0.0691,
      "step": 3417
    },
    {
      "epoch": 0.5391805024253657,
      "grad_norm": 4.116052150726318,
      "learning_rate": 4.702946385445178e-07,
      "loss": 0.047,
      "step": 3418
    },
    {
      "epoch": 0.5393382497929565,
      "grad_norm": 9.27847671508789,
      "learning_rate": 4.7013363387538233e-07,
      "loss": 0.0587,
      "step": 3419
    },
    {
      "epoch": 0.5394959971605474,
      "grad_norm": 3.903451919555664,
      "learning_rate": 4.69972629206247e-07,
      "loss": 0.044,
      "step": 3420
    },
    {
      "epoch": 0.5396537445281382,
      "grad_norm": 3.332679033279419,
      "learning_rate": 4.6981162453711154e-07,
      "loss": 0.0169,
      "step": 3421
    },
    {
      "epoch": 0.539811491895729,
      "grad_norm": 6.710604190826416,
      "learning_rate": 4.6965061986797615e-07,
      "loss": 0.0864,
      "step": 3422
    },
    {
      "epoch": 0.5399692392633197,
      "grad_norm": 5.891706466674805,
      "learning_rate": 4.694896151988407e-07,
      "loss": 0.0597,
      "step": 3423
    },
    {
      "epoch": 0.5401269866309106,
      "grad_norm": 6.030353546142578,
      "learning_rate": 4.6932861052970536e-07,
      "loss": 0.0466,
      "step": 3424
    },
    {
      "epoch": 0.5402847339985014,
      "grad_norm": 5.598007678985596,
      "learning_rate": 4.691676058605699e-07,
      "loss": 0.0471,
      "step": 3425
    },
    {
      "epoch": 0.5404424813660922,
      "grad_norm": 8.186391830444336,
      "learning_rate": 4.690066011914345e-07,
      "loss": 0.0636,
      "step": 3426
    },
    {
      "epoch": 0.540600228733683,
      "grad_norm": 8.03498649597168,
      "learning_rate": 4.6884559652229913e-07,
      "loss": 0.0664,
      "step": 3427
    },
    {
      "epoch": 0.5407579761012739,
      "grad_norm": 3.3238210678100586,
      "learning_rate": 4.6868459185316374e-07,
      "loss": 0.0298,
      "step": 3428
    },
    {
      "epoch": 0.5409157234688646,
      "grad_norm": 3.5968828201293945,
      "learning_rate": 4.685235871840283e-07,
      "loss": 0.0577,
      "step": 3429
    },
    {
      "epoch": 0.5410734708364554,
      "grad_norm": 5.372389316558838,
      "learning_rate": 4.683625825148929e-07,
      "loss": 0.0455,
      "step": 3430
    },
    {
      "epoch": 0.5412312182040462,
      "grad_norm": 1.568935751914978,
      "learning_rate": 4.682015778457575e-07,
      "loss": 0.014,
      "step": 3431
    },
    {
      "epoch": 0.5413889655716371,
      "grad_norm": 3.2657830715179443,
      "learning_rate": 4.680405731766221e-07,
      "loss": 0.0208,
      "step": 3432
    },
    {
      "epoch": 0.5415467129392278,
      "grad_norm": 8.174827575683594,
      "learning_rate": 4.678795685074867e-07,
      "loss": 0.034,
      "step": 3433
    },
    {
      "epoch": 0.5417044603068186,
      "grad_norm": 2.3991222381591797,
      "learning_rate": 4.6771856383835127e-07,
      "loss": 0.0119,
      "step": 3434
    },
    {
      "epoch": 0.5418622076744094,
      "grad_norm": 6.57170295715332,
      "learning_rate": 4.675575591692159e-07,
      "loss": 0.0433,
      "step": 3435
    },
    {
      "epoch": 0.5420199550420003,
      "grad_norm": 2.901010036468506,
      "learning_rate": 4.673965545000805e-07,
      "loss": 0.039,
      "step": 3436
    },
    {
      "epoch": 0.5421777024095911,
      "grad_norm": 5.223124027252197,
      "learning_rate": 4.672355498309451e-07,
      "loss": 0.0515,
      "step": 3437
    },
    {
      "epoch": 0.5423354497771818,
      "grad_norm": 3.257049560546875,
      "learning_rate": 4.6707454516180964e-07,
      "loss": 0.0343,
      "step": 3438
    },
    {
      "epoch": 0.5424931971447726,
      "grad_norm": 11.468480110168457,
      "learning_rate": 4.669135404926743e-07,
      "loss": 0.0522,
      "step": 3439
    },
    {
      "epoch": 0.5426509445123634,
      "grad_norm": 8.88228988647461,
      "learning_rate": 4.6675253582353885e-07,
      "loss": 0.0772,
      "step": 3440
    },
    {
      "epoch": 0.5428086918799543,
      "grad_norm": 3.969876289367676,
      "learning_rate": 4.6659153115440346e-07,
      "loss": 0.0363,
      "step": 3441
    },
    {
      "epoch": 0.5429664392475451,
      "grad_norm": 8.859699249267578,
      "learning_rate": 4.6643052648526806e-07,
      "loss": 0.0277,
      "step": 3442
    },
    {
      "epoch": 0.5431241866151358,
      "grad_norm": 6.532208442687988,
      "learning_rate": 4.6626952181613267e-07,
      "loss": 0.023,
      "step": 3443
    },
    {
      "epoch": 0.5432819339827266,
      "grad_norm": 145.56295776367188,
      "learning_rate": 4.661085171469972e-07,
      "loss": 0.0473,
      "step": 3444
    },
    {
      "epoch": 0.5434396813503175,
      "grad_norm": 5.529518127441406,
      "learning_rate": 4.6594751247786183e-07,
      "loss": 0.0286,
      "step": 3445
    },
    {
      "epoch": 0.5435974287179083,
      "grad_norm": 5.401988983154297,
      "learning_rate": 4.6578650780872644e-07,
      "loss": 0.0503,
      "step": 3446
    },
    {
      "epoch": 0.5437551760854991,
      "grad_norm": 6.497735023498535,
      "learning_rate": 4.6562550313959104e-07,
      "loss": 0.0424,
      "step": 3447
    },
    {
      "epoch": 0.5439129234530898,
      "grad_norm": 5.0852203369140625,
      "learning_rate": 4.654644984704556e-07,
      "loss": 0.0297,
      "step": 3448
    },
    {
      "epoch": 0.5440706708206807,
      "grad_norm": 3.3863515853881836,
      "learning_rate": 4.653034938013202e-07,
      "loss": 0.0153,
      "step": 3449
    },
    {
      "epoch": 0.5442284181882715,
      "grad_norm": 2.5436203479766846,
      "learning_rate": 4.651424891321848e-07,
      "loss": 0.0303,
      "step": 3450
    },
    {
      "epoch": 0.5443861655558623,
      "grad_norm": 7.061236381530762,
      "learning_rate": 4.649814844630494e-07,
      "loss": 0.0787,
      "step": 3451
    },
    {
      "epoch": 0.544543912923453,
      "grad_norm": 2.437614679336548,
      "learning_rate": 4.6482047979391397e-07,
      "loss": 0.0102,
      "step": 3452
    },
    {
      "epoch": 0.5447016602910439,
      "grad_norm": 1.9190566539764404,
      "learning_rate": 4.646594751247786e-07,
      "loss": 0.0167,
      "step": 3453
    },
    {
      "epoch": 0.5448594076586347,
      "grad_norm": 4.904125213623047,
      "learning_rate": 4.644984704556432e-07,
      "loss": 0.0565,
      "step": 3454
    },
    {
      "epoch": 0.5450171550262255,
      "grad_norm": 5.0562005043029785,
      "learning_rate": 4.643374657865078e-07,
      "loss": 0.0126,
      "step": 3455
    },
    {
      "epoch": 0.5451749023938163,
      "grad_norm": 2.606792449951172,
      "learning_rate": 4.641764611173724e-07,
      "loss": 0.0179,
      "step": 3456
    },
    {
      "epoch": 0.545332649761407,
      "grad_norm": 6.193460464477539,
      "learning_rate": 4.64015456448237e-07,
      "loss": 0.0728,
      "step": 3457
    },
    {
      "epoch": 0.5454903971289979,
      "grad_norm": 7.9752116203308105,
      "learning_rate": 4.638544517791016e-07,
      "loss": 0.059,
      "step": 3458
    },
    {
      "epoch": 0.5456481444965887,
      "grad_norm": 5.858904838562012,
      "learning_rate": 4.6369344710996616e-07,
      "loss": 0.082,
      "step": 3459
    },
    {
      "epoch": 0.5458058918641795,
      "grad_norm": 4.883467197418213,
      "learning_rate": 4.6353244244083077e-07,
      "loss": 0.0474,
      "step": 3460
    },
    {
      "epoch": 0.5459636392317703,
      "grad_norm": 1.261867880821228,
      "learning_rate": 4.6337143777169537e-07,
      "loss": 0.008,
      "step": 3461
    },
    {
      "epoch": 0.5461213865993612,
      "grad_norm": 11.196954727172852,
      "learning_rate": 4.6321043310256e-07,
      "loss": 0.0654,
      "step": 3462
    },
    {
      "epoch": 0.5462791339669519,
      "grad_norm": 7.130752086639404,
      "learning_rate": 4.6304942843342453e-07,
      "loss": 0.067,
      "step": 3463
    },
    {
      "epoch": 0.5464368813345427,
      "grad_norm": 9.090606689453125,
      "learning_rate": 4.6288842376428914e-07,
      "loss": 0.0608,
      "step": 3464
    },
    {
      "epoch": 0.5465946287021335,
      "grad_norm": 4.5959343910217285,
      "learning_rate": 4.6272741909515375e-07,
      "loss": 0.0368,
      "step": 3465
    },
    {
      "epoch": 0.5467523760697244,
      "grad_norm": 3.92128849029541,
      "learning_rate": 4.6256641442601835e-07,
      "loss": 0.0206,
      "step": 3466
    },
    {
      "epoch": 0.5469101234373152,
      "grad_norm": 4.205663204193115,
      "learning_rate": 4.624054097568829e-07,
      "loss": 0.0664,
      "step": 3467
    },
    {
      "epoch": 0.5470678708049059,
      "grad_norm": 11.866006851196289,
      "learning_rate": 4.622444050877475e-07,
      "loss": 0.0505,
      "step": 3468
    },
    {
      "epoch": 0.5472256181724967,
      "grad_norm": 5.364625930786133,
      "learning_rate": 4.620834004186121e-07,
      "loss": 0.0548,
      "step": 3469
    },
    {
      "epoch": 0.5473833655400876,
      "grad_norm": 4.128734588623047,
      "learning_rate": 4.619223957494767e-07,
      "loss": 0.0664,
      "step": 3470
    },
    {
      "epoch": 0.5475411129076784,
      "grad_norm": 6.497686862945557,
      "learning_rate": 4.617613910803413e-07,
      "loss": 0.0823,
      "step": 3471
    },
    {
      "epoch": 0.5476988602752692,
      "grad_norm": 4.692132949829102,
      "learning_rate": 4.6160038641120594e-07,
      "loss": 0.0204,
      "step": 3472
    },
    {
      "epoch": 0.5478566076428599,
      "grad_norm": 3.7560946941375732,
      "learning_rate": 4.614393817420705e-07,
      "loss": 0.0337,
      "step": 3473
    },
    {
      "epoch": 0.5480143550104508,
      "grad_norm": 5.997998237609863,
      "learning_rate": 4.612783770729351e-07,
      "loss": 0.0326,
      "step": 3474
    },
    {
      "epoch": 0.5481721023780416,
      "grad_norm": 4.13024377822876,
      "learning_rate": 4.6111737240379965e-07,
      "loss": 0.0536,
      "step": 3475
    },
    {
      "epoch": 0.5483298497456324,
      "grad_norm": 4.070066928863525,
      "learning_rate": 4.609563677346643e-07,
      "loss": 0.0432,
      "step": 3476
    },
    {
      "epoch": 0.5484875971132231,
      "grad_norm": 2.8549342155456543,
      "learning_rate": 4.6079536306552886e-07,
      "loss": 0.0234,
      "step": 3477
    },
    {
      "epoch": 0.5486453444808139,
      "grad_norm": 5.868303298950195,
      "learning_rate": 4.6063435839639347e-07,
      "loss": 0.0337,
      "step": 3478
    },
    {
      "epoch": 0.5488030918484048,
      "grad_norm": 3.397012710571289,
      "learning_rate": 4.604733537272581e-07,
      "loss": 0.011,
      "step": 3479
    },
    {
      "epoch": 0.5489608392159956,
      "grad_norm": 8.067663192749023,
      "learning_rate": 4.603123490581227e-07,
      "loss": 0.0464,
      "step": 3480
    },
    {
      "epoch": 0.5491185865835864,
      "grad_norm": 3.4272048473358154,
      "learning_rate": 4.601513443889873e-07,
      "loss": 0.0421,
      "step": 3481
    },
    {
      "epoch": 0.5492763339511771,
      "grad_norm": 5.533441543579102,
      "learning_rate": 4.5999033971985184e-07,
      "loss": 0.0545,
      "step": 3482
    },
    {
      "epoch": 0.549434081318768,
      "grad_norm": 4.946138381958008,
      "learning_rate": 4.5982933505071645e-07,
      "loss": 0.0834,
      "step": 3483
    },
    {
      "epoch": 0.5495918286863588,
      "grad_norm": 4.812375545501709,
      "learning_rate": 4.5966833038158105e-07,
      "loss": 0.0384,
      "step": 3484
    },
    {
      "epoch": 0.5497495760539496,
      "grad_norm": 3.619523286819458,
      "learning_rate": 4.5950732571244566e-07,
      "loss": 0.0135,
      "step": 3485
    },
    {
      "epoch": 0.5499073234215404,
      "grad_norm": 6.208311080932617,
      "learning_rate": 4.593463210433102e-07,
      "loss": 0.0358,
      "step": 3486
    },
    {
      "epoch": 0.5500650707891312,
      "grad_norm": 4.9354248046875,
      "learning_rate": 4.5918531637417487e-07,
      "loss": 0.0558,
      "step": 3487
    },
    {
      "epoch": 0.550222818156722,
      "grad_norm": 4.226499557495117,
      "learning_rate": 4.590243117050394e-07,
      "loss": 0.0209,
      "step": 3488
    },
    {
      "epoch": 0.5503805655243128,
      "grad_norm": 7.435657024383545,
      "learning_rate": 4.5886330703590403e-07,
      "loss": 0.0549,
      "step": 3489
    },
    {
      "epoch": 0.5505383128919036,
      "grad_norm": 5.155275344848633,
      "learning_rate": 4.587023023667686e-07,
      "loss": 0.0373,
      "step": 3490
    },
    {
      "epoch": 0.5506960602594945,
      "grad_norm": 5.168789863586426,
      "learning_rate": 4.5854129769763324e-07,
      "loss": 0.0215,
      "step": 3491
    },
    {
      "epoch": 0.5508538076270852,
      "grad_norm": 6.964191436767578,
      "learning_rate": 4.583802930284978e-07,
      "loss": 0.0551,
      "step": 3492
    },
    {
      "epoch": 0.551011554994676,
      "grad_norm": 4.6709885597229,
      "learning_rate": 4.582192883593624e-07,
      "loss": 0.0284,
      "step": 3493
    },
    {
      "epoch": 0.5511693023622668,
      "grad_norm": 6.141894340515137,
      "learning_rate": 4.5805828369022696e-07,
      "loss": 0.0963,
      "step": 3494
    },
    {
      "epoch": 0.5513270497298577,
      "grad_norm": 1.8826123476028442,
      "learning_rate": 4.578972790210916e-07,
      "loss": 0.014,
      "step": 3495
    },
    {
      "epoch": 0.5514847970974485,
      "grad_norm": 5.1856584548950195,
      "learning_rate": 4.5773627435195617e-07,
      "loss": 0.0543,
      "step": 3496
    },
    {
      "epoch": 0.5516425444650392,
      "grad_norm": 7.011806964874268,
      "learning_rate": 4.575752696828208e-07,
      "loss": 0.0453,
      "step": 3497
    },
    {
      "epoch": 0.55180029183263,
      "grad_norm": 4.731129169464111,
      "learning_rate": 4.5741426501368533e-07,
      "loss": 0.042,
      "step": 3498
    },
    {
      "epoch": 0.5519580392002208,
      "grad_norm": 4.678847312927246,
      "learning_rate": 4.5725326034455e-07,
      "loss": 0.0519,
      "step": 3499
    },
    {
      "epoch": 0.5521157865678117,
      "grad_norm": 3.7979085445404053,
      "learning_rate": 4.5709225567541454e-07,
      "loss": 0.0578,
      "step": 3500
    },
    {
      "epoch": 0.5522735339354025,
      "grad_norm": 7.19497013092041,
      "learning_rate": 4.5693125100627915e-07,
      "loss": 0.0694,
      "step": 3501
    },
    {
      "epoch": 0.5524312813029932,
      "grad_norm": 5.810506343841553,
      "learning_rate": 4.567702463371438e-07,
      "loss": 0.0364,
      "step": 3502
    },
    {
      "epoch": 0.552589028670584,
      "grad_norm": 4.69302225112915,
      "learning_rate": 4.5660924166800836e-07,
      "loss": 0.1141,
      "step": 3503
    },
    {
      "epoch": 0.5527467760381749,
      "grad_norm": 3.8634958267211914,
      "learning_rate": 4.5644823699887297e-07,
      "loss": 0.0404,
      "step": 3504
    },
    {
      "epoch": 0.5529045234057657,
      "grad_norm": 7.281254768371582,
      "learning_rate": 4.562872323297375e-07,
      "loss": 0.1027,
      "step": 3505
    },
    {
      "epoch": 0.5530622707733565,
      "grad_norm": 5.159707069396973,
      "learning_rate": 4.561262276606022e-07,
      "loss": 0.0189,
      "step": 3506
    },
    {
      "epoch": 0.5532200181409472,
      "grad_norm": 5.031998157501221,
      "learning_rate": 4.5596522299146673e-07,
      "loss": 0.0293,
      "step": 3507
    },
    {
      "epoch": 0.5533777655085381,
      "grad_norm": 4.738980293273926,
      "learning_rate": 4.5580421832233134e-07,
      "loss": 0.0493,
      "step": 3508
    },
    {
      "epoch": 0.5535355128761289,
      "grad_norm": 6.045560359954834,
      "learning_rate": 4.556432136531959e-07,
      "loss": 0.0848,
      "step": 3509
    },
    {
      "epoch": 0.5536932602437197,
      "grad_norm": 2.5067272186279297,
      "learning_rate": 4.5548220898406055e-07,
      "loss": 0.0101,
      "step": 3510
    },
    {
      "epoch": 0.5538510076113105,
      "grad_norm": 3.580920696258545,
      "learning_rate": 4.553212043149251e-07,
      "loss": 0.0405,
      "step": 3511
    },
    {
      "epoch": 0.5540087549789013,
      "grad_norm": 4.838515758514404,
      "learning_rate": 4.551601996457897e-07,
      "loss": 0.0187,
      "step": 3512
    },
    {
      "epoch": 0.5541665023464921,
      "grad_norm": 5.049856185913086,
      "learning_rate": 4.5499919497665427e-07,
      "loss": 0.0584,
      "step": 3513
    },
    {
      "epoch": 0.5543242497140829,
      "grad_norm": 1.609534502029419,
      "learning_rate": 4.548381903075189e-07,
      "loss": 0.0104,
      "step": 3514
    },
    {
      "epoch": 0.5544819970816737,
      "grad_norm": 4.600698471069336,
      "learning_rate": 4.546771856383835e-07,
      "loss": 0.0333,
      "step": 3515
    },
    {
      "epoch": 0.5546397444492644,
      "grad_norm": 7.993060111999512,
      "learning_rate": 4.545161809692481e-07,
      "loss": 0.0558,
      "step": 3516
    },
    {
      "epoch": 0.5547974918168553,
      "grad_norm": 3.101815938949585,
      "learning_rate": 4.5435517630011264e-07,
      "loss": 0.0228,
      "step": 3517
    },
    {
      "epoch": 0.5549552391844461,
      "grad_norm": 5.370943546295166,
      "learning_rate": 4.541941716309773e-07,
      "loss": 0.1041,
      "step": 3518
    },
    {
      "epoch": 0.5551129865520369,
      "grad_norm": 7.163643836975098,
      "learning_rate": 4.5403316696184185e-07,
      "loss": 0.0535,
      "step": 3519
    },
    {
      "epoch": 0.5552707339196277,
      "grad_norm": 3.8371834754943848,
      "learning_rate": 4.5387216229270646e-07,
      "loss": 0.0255,
      "step": 3520
    },
    {
      "epoch": 0.5554284812872186,
      "grad_norm": 8.119474411010742,
      "learning_rate": 4.5371115762357106e-07,
      "loss": 0.0674,
      "step": 3521
    },
    {
      "epoch": 0.5555862286548093,
      "grad_norm": 4.2197980880737305,
      "learning_rate": 4.5355015295443567e-07,
      "loss": 0.0612,
      "step": 3522
    },
    {
      "epoch": 0.5557439760224001,
      "grad_norm": 6.642730712890625,
      "learning_rate": 4.533891482853002e-07,
      "loss": 0.0458,
      "step": 3523
    },
    {
      "epoch": 0.5559017233899909,
      "grad_norm": 7.063992023468018,
      "learning_rate": 4.5322814361616483e-07,
      "loss": 0.0785,
      "step": 3524
    },
    {
      "epoch": 0.5560594707575818,
      "grad_norm": 4.294404029846191,
      "learning_rate": 4.530671389470295e-07,
      "loss": 0.0348,
      "step": 3525
    },
    {
      "epoch": 0.5562172181251726,
      "grad_norm": 4.23181676864624,
      "learning_rate": 4.5290613427789404e-07,
      "loss": 0.0357,
      "step": 3526
    },
    {
      "epoch": 0.5563749654927633,
      "grad_norm": 10.125638008117676,
      "learning_rate": 4.5274512960875865e-07,
      "loss": 0.0384,
      "step": 3527
    },
    {
      "epoch": 0.5565327128603541,
      "grad_norm": 5.969682693481445,
      "learning_rate": 4.525841249396232e-07,
      "loss": 0.0476,
      "step": 3528
    },
    {
      "epoch": 0.556690460227945,
      "grad_norm": 4.976404666900635,
      "learning_rate": 4.5242312027048786e-07,
      "loss": 0.0816,
      "step": 3529
    },
    {
      "epoch": 0.5568482075955358,
      "grad_norm": 5.278900146484375,
      "learning_rate": 4.522621156013524e-07,
      "loss": 0.0294,
      "step": 3530
    },
    {
      "epoch": 0.5570059549631265,
      "grad_norm": 0.712295413017273,
      "learning_rate": 4.52101110932217e-07,
      "loss": 0.0058,
      "step": 3531
    },
    {
      "epoch": 0.5571637023307173,
      "grad_norm": 4.528964519500732,
      "learning_rate": 4.519401062630816e-07,
      "loss": 0.0404,
      "step": 3532
    },
    {
      "epoch": 0.5573214496983082,
      "grad_norm": 5.830653190612793,
      "learning_rate": 4.5177910159394623e-07,
      "loss": 0.0474,
      "step": 3533
    },
    {
      "epoch": 0.557479197065899,
      "grad_norm": 5.37799072265625,
      "learning_rate": 4.516180969248108e-07,
      "loss": 0.0791,
      "step": 3534
    },
    {
      "epoch": 0.5576369444334898,
      "grad_norm": 4.497136116027832,
      "learning_rate": 4.514570922556754e-07,
      "loss": 0.0459,
      "step": 3535
    },
    {
      "epoch": 0.5577946918010805,
      "grad_norm": 12.904130935668945,
      "learning_rate": 4.5129608758654e-07,
      "loss": 0.0399,
      "step": 3536
    },
    {
      "epoch": 0.5579524391686713,
      "grad_norm": 7.827426433563232,
      "learning_rate": 4.511350829174046e-07,
      "loss": 0.0686,
      "step": 3537
    },
    {
      "epoch": 0.5581101865362622,
      "grad_norm": 5.223310947418213,
      "learning_rate": 4.5097407824826916e-07,
      "loss": 0.0697,
      "step": 3538
    },
    {
      "epoch": 0.558267933903853,
      "grad_norm": 5.207493782043457,
      "learning_rate": 4.5081307357913377e-07,
      "loss": 0.0308,
      "step": 3539
    },
    {
      "epoch": 0.5584256812714438,
      "grad_norm": 9.767026901245117,
      "learning_rate": 4.5065206890999837e-07,
      "loss": 0.0277,
      "step": 3540
    },
    {
      "epoch": 0.5585834286390345,
      "grad_norm": 3.7552871704101562,
      "learning_rate": 4.50491064240863e-07,
      "loss": 0.0664,
      "step": 3541
    },
    {
      "epoch": 0.5587411760066254,
      "grad_norm": 3.1752941608428955,
      "learning_rate": 4.5033005957172753e-07,
      "loss": 0.0214,
      "step": 3542
    },
    {
      "epoch": 0.5588989233742162,
      "grad_norm": 3.1909866333007812,
      "learning_rate": 4.5016905490259214e-07,
      "loss": 0.0194,
      "step": 3543
    },
    {
      "epoch": 0.559056670741807,
      "grad_norm": 7.0103325843811035,
      "learning_rate": 4.5000805023345674e-07,
      "loss": 0.1283,
      "step": 3544
    },
    {
      "epoch": 0.5592144181093978,
      "grad_norm": 8.944089889526367,
      "learning_rate": 4.4984704556432135e-07,
      "loss": 0.0369,
      "step": 3545
    },
    {
      "epoch": 0.5593721654769886,
      "grad_norm": 2.797886848449707,
      "learning_rate": 4.496860408951859e-07,
      "loss": 0.0237,
      "step": 3546
    },
    {
      "epoch": 0.5595299128445794,
      "grad_norm": 4.817679405212402,
      "learning_rate": 4.495250362260505e-07,
      "loss": 0.0402,
      "step": 3547
    },
    {
      "epoch": 0.5596876602121702,
      "grad_norm": 4.421696662902832,
      "learning_rate": 4.4936403155691517e-07,
      "loss": 0.0787,
      "step": 3548
    },
    {
      "epoch": 0.559845407579761,
      "grad_norm": 4.182826519012451,
      "learning_rate": 4.492030268877797e-07,
      "loss": 0.0542,
      "step": 3549
    },
    {
      "epoch": 0.5600031549473519,
      "grad_norm": 4.986604690551758,
      "learning_rate": 4.4904202221864433e-07,
      "loss": 0.06,
      "step": 3550
    },
    {
      "epoch": 0.5601609023149426,
      "grad_norm": 6.201351165771484,
      "learning_rate": 4.4888101754950894e-07,
      "loss": 0.0344,
      "step": 3551
    },
    {
      "epoch": 0.5603186496825334,
      "grad_norm": 9.955351829528809,
      "learning_rate": 4.4872001288037354e-07,
      "loss": 0.0868,
      "step": 3552
    },
    {
      "epoch": 0.5604763970501242,
      "grad_norm": 6.308249473571777,
      "learning_rate": 4.485590082112381e-07,
      "loss": 0.0451,
      "step": 3553
    },
    {
      "epoch": 0.560634144417715,
      "grad_norm": 5.036297798156738,
      "learning_rate": 4.483980035421027e-07,
      "loss": 0.034,
      "step": 3554
    },
    {
      "epoch": 0.5607918917853059,
      "grad_norm": 6.0406293869018555,
      "learning_rate": 4.482369988729673e-07,
      "loss": 0.0182,
      "step": 3555
    },
    {
      "epoch": 0.5609496391528966,
      "grad_norm": 6.802767753601074,
      "learning_rate": 4.480759942038319e-07,
      "loss": 0.0706,
      "step": 3556
    },
    {
      "epoch": 0.5611073865204874,
      "grad_norm": 2.260598659515381,
      "learning_rate": 4.4791498953469647e-07,
      "loss": 0.0112,
      "step": 3557
    },
    {
      "epoch": 0.5612651338880782,
      "grad_norm": 5.509818077087402,
      "learning_rate": 4.477539848655611e-07,
      "loss": 0.0924,
      "step": 3558
    },
    {
      "epoch": 0.5614228812556691,
      "grad_norm": 2.3354532718658447,
      "learning_rate": 4.475929801964257e-07,
      "loss": 0.033,
      "step": 3559
    },
    {
      "epoch": 0.5615806286232599,
      "grad_norm": 1.7844414710998535,
      "learning_rate": 4.474319755272903e-07,
      "loss": 0.0152,
      "step": 3560
    },
    {
      "epoch": 0.5617383759908506,
      "grad_norm": 6.738806247711182,
      "learning_rate": 4.4727097085815484e-07,
      "loss": 0.0756,
      "step": 3561
    },
    {
      "epoch": 0.5618961233584414,
      "grad_norm": 4.664470195770264,
      "learning_rate": 4.4710996618901945e-07,
      "loss": 0.0299,
      "step": 3562
    },
    {
      "epoch": 0.5620538707260323,
      "grad_norm": 7.961429119110107,
      "learning_rate": 4.4694896151988405e-07,
      "loss": 0.0919,
      "step": 3563
    },
    {
      "epoch": 0.5622116180936231,
      "grad_norm": 8.812355995178223,
      "learning_rate": 4.4678795685074866e-07,
      "loss": 0.0479,
      "step": 3564
    },
    {
      "epoch": 0.5623693654612139,
      "grad_norm": 2.8166677951812744,
      "learning_rate": 4.466269521816132e-07,
      "loss": 0.0226,
      "step": 3565
    },
    {
      "epoch": 0.5625271128288046,
      "grad_norm": 4.626858234405518,
      "learning_rate": 4.4646594751247787e-07,
      "loss": 0.0438,
      "step": 3566
    },
    {
      "epoch": 0.5626848601963955,
      "grad_norm": 6.821376800537109,
      "learning_rate": 4.463049428433424e-07,
      "loss": 0.0307,
      "step": 3567
    },
    {
      "epoch": 0.5628426075639863,
      "grad_norm": 1.3527711629867554,
      "learning_rate": 4.4614393817420703e-07,
      "loss": 0.0126,
      "step": 3568
    },
    {
      "epoch": 0.5630003549315771,
      "grad_norm": 6.533590793609619,
      "learning_rate": 4.459829335050716e-07,
      "loss": 0.0544,
      "step": 3569
    },
    {
      "epoch": 0.5631581022991679,
      "grad_norm": 4.931443214416504,
      "learning_rate": 4.4582192883593624e-07,
      "loss": 0.0374,
      "step": 3570
    },
    {
      "epoch": 0.5633158496667587,
      "grad_norm": 4.620832920074463,
      "learning_rate": 4.4566092416680085e-07,
      "loss": 0.0592,
      "step": 3571
    },
    {
      "epoch": 0.5634735970343495,
      "grad_norm": 4.254405498504639,
      "learning_rate": 4.454999194976654e-07,
      "loss": 0.0428,
      "step": 3572
    },
    {
      "epoch": 0.5636313444019403,
      "grad_norm": 3.803039312362671,
      "learning_rate": 4.4533891482853e-07,
      "loss": 0.0283,
      "step": 3573
    },
    {
      "epoch": 0.5637890917695311,
      "grad_norm": 1.895159125328064,
      "learning_rate": 4.451779101593946e-07,
      "loss": 0.0143,
      "step": 3574
    },
    {
      "epoch": 0.5639468391371218,
      "grad_norm": 8.910772323608398,
      "learning_rate": 4.450169054902592e-07,
      "loss": 0.0906,
      "step": 3575
    },
    {
      "epoch": 0.5641045865047127,
      "grad_norm": 7.452020168304443,
      "learning_rate": 4.448559008211238e-07,
      "loss": 0.0389,
      "step": 3576
    },
    {
      "epoch": 0.5642623338723035,
      "grad_norm": 6.074785232543945,
      "learning_rate": 4.446948961519884e-07,
      "loss": 0.0544,
      "step": 3577
    },
    {
      "epoch": 0.5644200812398943,
      "grad_norm": 2.867938756942749,
      "learning_rate": 4.44533891482853e-07,
      "loss": 0.0141,
      "step": 3578
    },
    {
      "epoch": 0.5645778286074851,
      "grad_norm": 5.436214447021484,
      "learning_rate": 4.443728868137176e-07,
      "loss": 0.0427,
      "step": 3579
    },
    {
      "epoch": 0.564735575975076,
      "grad_norm": 5.736272811889648,
      "learning_rate": 4.4421188214458215e-07,
      "loss": 0.0629,
      "step": 3580
    },
    {
      "epoch": 0.5648933233426667,
      "grad_norm": 7.325777530670166,
      "learning_rate": 4.440508774754468e-07,
      "loss": 0.0482,
      "step": 3581
    },
    {
      "epoch": 0.5650510707102575,
      "grad_norm": 3.6818907260894775,
      "learning_rate": 4.4388987280631136e-07,
      "loss": 0.0268,
      "step": 3582
    },
    {
      "epoch": 0.5652088180778483,
      "grad_norm": 5.712748050689697,
      "learning_rate": 4.4372886813717597e-07,
      "loss": 0.0583,
      "step": 3583
    },
    {
      "epoch": 0.5653665654454392,
      "grad_norm": 2.976290464401245,
      "learning_rate": 4.435678634680405e-07,
      "loss": 0.0155,
      "step": 3584
    },
    {
      "epoch": 0.56552431281303,
      "grad_norm": 4.396505832672119,
      "learning_rate": 4.434068587989052e-07,
      "loss": 0.0215,
      "step": 3585
    },
    {
      "epoch": 0.5656820601806207,
      "grad_norm": 3.87809157371521,
      "learning_rate": 4.4324585412976973e-07,
      "loss": 0.0238,
      "step": 3586
    },
    {
      "epoch": 0.5658398075482115,
      "grad_norm": 5.459232807159424,
      "learning_rate": 4.4308484946063434e-07,
      "loss": 0.0462,
      "step": 3587
    },
    {
      "epoch": 0.5659975549158024,
      "grad_norm": 7.381917476654053,
      "learning_rate": 4.429238447914989e-07,
      "loss": 0.0517,
      "step": 3588
    },
    {
      "epoch": 0.5661553022833932,
      "grad_norm": 6.301811218261719,
      "learning_rate": 4.4276284012236355e-07,
      "loss": 0.0395,
      "step": 3589
    },
    {
      "epoch": 0.566313049650984,
      "grad_norm": 6.701537609100342,
      "learning_rate": 4.426018354532281e-07,
      "loss": 0.064,
      "step": 3590
    },
    {
      "epoch": 0.5664707970185747,
      "grad_norm": 6.956577301025391,
      "learning_rate": 4.424408307840927e-07,
      "loss": 0.027,
      "step": 3591
    },
    {
      "epoch": 0.5666285443861656,
      "grad_norm": 6.830028057098389,
      "learning_rate": 4.4227982611495726e-07,
      "loss": 0.0788,
      "step": 3592
    },
    {
      "epoch": 0.5667862917537564,
      "grad_norm": 5.745498180389404,
      "learning_rate": 4.421188214458219e-07,
      "loss": 0.0823,
      "step": 3593
    },
    {
      "epoch": 0.5669440391213472,
      "grad_norm": 8.539166450500488,
      "learning_rate": 4.419578167766865e-07,
      "loss": 0.021,
      "step": 3594
    },
    {
      "epoch": 0.5671017864889379,
      "grad_norm": 8.19500732421875,
      "learning_rate": 4.417968121075511e-07,
      "loss": 0.0428,
      "step": 3595
    },
    {
      "epoch": 0.5672595338565287,
      "grad_norm": 5.17337703704834,
      "learning_rate": 4.4163580743841574e-07,
      "loss": 0.0982,
      "step": 3596
    },
    {
      "epoch": 0.5674172812241196,
      "grad_norm": 5.357944965362549,
      "learning_rate": 4.414748027692803e-07,
      "loss": 0.0188,
      "step": 3597
    },
    {
      "epoch": 0.5675750285917104,
      "grad_norm": 5.059239864349365,
      "learning_rate": 4.413137981001449e-07,
      "loss": 0.0285,
      "step": 3598
    },
    {
      "epoch": 0.5677327759593012,
      "grad_norm": 3.6183338165283203,
      "learning_rate": 4.4115279343100946e-07,
      "loss": 0.0534,
      "step": 3599
    },
    {
      "epoch": 0.5678905233268919,
      "grad_norm": 3.2750072479248047,
      "learning_rate": 4.409917887618741e-07,
      "loss": 0.0229,
      "step": 3600
    },
    {
      "epoch": 0.5680482706944828,
      "grad_norm": 6.083227634429932,
      "learning_rate": 4.4083078409273867e-07,
      "loss": 0.0634,
      "step": 3601
    },
    {
      "epoch": 0.5682060180620736,
      "grad_norm": 6.175106525421143,
      "learning_rate": 4.406697794236033e-07,
      "loss": 0.0628,
      "step": 3602
    },
    {
      "epoch": 0.5683637654296644,
      "grad_norm": 3.048811197280884,
      "learning_rate": 4.4050877475446783e-07,
      "loss": 0.0191,
      "step": 3603
    },
    {
      "epoch": 0.5685215127972552,
      "grad_norm": 3.7438251972198486,
      "learning_rate": 4.403477700853325e-07,
      "loss": 0.034,
      "step": 3604
    },
    {
      "epoch": 0.568679260164846,
      "grad_norm": 2.688021659851074,
      "learning_rate": 4.4018676541619704e-07,
      "loss": 0.0154,
      "step": 3605
    },
    {
      "epoch": 0.5688370075324368,
      "grad_norm": 4.568440914154053,
      "learning_rate": 4.4002576074706165e-07,
      "loss": 0.0536,
      "step": 3606
    },
    {
      "epoch": 0.5689947549000276,
      "grad_norm": 4.3514323234558105,
      "learning_rate": 4.398647560779262e-07,
      "loss": 0.0249,
      "step": 3607
    },
    {
      "epoch": 0.5691525022676184,
      "grad_norm": 9.23238468170166,
      "learning_rate": 4.3970375140879086e-07,
      "loss": 0.072,
      "step": 3608
    },
    {
      "epoch": 0.5693102496352093,
      "grad_norm": 25.24394989013672,
      "learning_rate": 4.395427467396554e-07,
      "loss": 0.0431,
      "step": 3609
    },
    {
      "epoch": 0.5694679970028,
      "grad_norm": 5.3971052169799805,
      "learning_rate": 4.3938174207052e-07,
      "loss": 0.0692,
      "step": 3610
    },
    {
      "epoch": 0.5696257443703908,
      "grad_norm": 1.751876950263977,
      "learning_rate": 4.392207374013846e-07,
      "loss": 0.0128,
      "step": 3611
    },
    {
      "epoch": 0.5697834917379816,
      "grad_norm": 4.0099005699157715,
      "learning_rate": 4.3905973273224923e-07,
      "loss": 0.0545,
      "step": 3612
    },
    {
      "epoch": 0.5699412391055724,
      "grad_norm": 5.732677936553955,
      "learning_rate": 4.388987280631138e-07,
      "loss": 0.0563,
      "step": 3613
    },
    {
      "epoch": 0.5700989864731633,
      "grad_norm": 6.623098373413086,
      "learning_rate": 4.387377233939784e-07,
      "loss": 0.0269,
      "step": 3614
    },
    {
      "epoch": 0.570256733840754,
      "grad_norm": 7.234887599945068,
      "learning_rate": 4.38576718724843e-07,
      "loss": 0.0952,
      "step": 3615
    },
    {
      "epoch": 0.5704144812083448,
      "grad_norm": 5.600711345672607,
      "learning_rate": 4.384157140557076e-07,
      "loss": 0.0298,
      "step": 3616
    },
    {
      "epoch": 0.5705722285759356,
      "grad_norm": 5.513890743255615,
      "learning_rate": 4.3825470938657216e-07,
      "loss": 0.0275,
      "step": 3617
    },
    {
      "epoch": 0.5707299759435265,
      "grad_norm": 4.5403337478637695,
      "learning_rate": 4.3809370471743676e-07,
      "loss": 0.0458,
      "step": 3618
    },
    {
      "epoch": 0.5708877233111173,
      "grad_norm": 2.449573516845703,
      "learning_rate": 4.379327000483014e-07,
      "loss": 0.0124,
      "step": 3619
    },
    {
      "epoch": 0.571045470678708,
      "grad_norm": 6.583084583282471,
      "learning_rate": 4.37771695379166e-07,
      "loss": 0.0841,
      "step": 3620
    },
    {
      "epoch": 0.5712032180462988,
      "grad_norm": 3.6035895347595215,
      "learning_rate": 4.376106907100306e-07,
      "loss": 0.0375,
      "step": 3621
    },
    {
      "epoch": 0.5713609654138897,
      "grad_norm": 9.386648178100586,
      "learning_rate": 4.3744968604089514e-07,
      "loss": 0.0745,
      "step": 3622
    },
    {
      "epoch": 0.5715187127814805,
      "grad_norm": 5.065177917480469,
      "learning_rate": 4.372886813717598e-07,
      "loss": 0.0708,
      "step": 3623
    },
    {
      "epoch": 0.5716764601490713,
      "grad_norm": 5.430810928344727,
      "learning_rate": 4.3712767670262435e-07,
      "loss": 0.0348,
      "step": 3624
    },
    {
      "epoch": 0.571834207516662,
      "grad_norm": 2.6487369537353516,
      "learning_rate": 4.3696667203348896e-07,
      "loss": 0.0116,
      "step": 3625
    },
    {
      "epoch": 0.5719919548842529,
      "grad_norm": 6.235917568206787,
      "learning_rate": 4.3680566736435356e-07,
      "loss": 0.0348,
      "step": 3626
    },
    {
      "epoch": 0.5721497022518437,
      "grad_norm": 6.639988899230957,
      "learning_rate": 4.3664466269521817e-07,
      "loss": 0.0418,
      "step": 3627
    },
    {
      "epoch": 0.5723074496194345,
      "grad_norm": 24.25504493713379,
      "learning_rate": 4.364836580260827e-07,
      "loss": 0.0543,
      "step": 3628
    },
    {
      "epoch": 0.5724651969870252,
      "grad_norm": 5.3725175857543945,
      "learning_rate": 4.3632265335694733e-07,
      "loss": 0.0462,
      "step": 3629
    },
    {
      "epoch": 0.5726229443546161,
      "grad_norm": 2.526221752166748,
      "learning_rate": 4.3616164868781193e-07,
      "loss": 0.0144,
      "step": 3630
    },
    {
      "epoch": 0.5727806917222069,
      "grad_norm": 9.240471839904785,
      "learning_rate": 4.3600064401867654e-07,
      "loss": 0.0432,
      "step": 3631
    },
    {
      "epoch": 0.5729384390897977,
      "grad_norm": 7.248495101928711,
      "learning_rate": 4.358396393495411e-07,
      "loss": 0.0247,
      "step": 3632
    },
    {
      "epoch": 0.5730961864573885,
      "grad_norm": 12.687851905822754,
      "learning_rate": 4.356786346804057e-07,
      "loss": 0.0641,
      "step": 3633
    },
    {
      "epoch": 0.5732539338249792,
      "grad_norm": 7.592212677001953,
      "learning_rate": 4.355176300112703e-07,
      "loss": 0.0675,
      "step": 3634
    },
    {
      "epoch": 0.5734116811925701,
      "grad_norm": 6.328648090362549,
      "learning_rate": 4.353566253421349e-07,
      "loss": 0.0605,
      "step": 3635
    },
    {
      "epoch": 0.5735694285601609,
      "grad_norm": 2.8092939853668213,
      "learning_rate": 4.3519562067299947e-07,
      "loss": 0.0199,
      "step": 3636
    },
    {
      "epoch": 0.5737271759277517,
      "grad_norm": 2.8142776489257812,
      "learning_rate": 4.3503461600386407e-07,
      "loss": 0.0202,
      "step": 3637
    },
    {
      "epoch": 0.5738849232953425,
      "grad_norm": 3.231452465057373,
      "learning_rate": 4.348736113347287e-07,
      "loss": 0.0193,
      "step": 3638
    },
    {
      "epoch": 0.5740426706629334,
      "grad_norm": 3.44266676902771,
      "learning_rate": 4.347126066655933e-07,
      "loss": 0.0451,
      "step": 3639
    },
    {
      "epoch": 0.5742004180305241,
      "grad_norm": 7.478565692901611,
      "learning_rate": 4.3455160199645784e-07,
      "loss": 0.0898,
      "step": 3640
    },
    {
      "epoch": 0.5743581653981149,
      "grad_norm": 6.21815824508667,
      "learning_rate": 4.343905973273225e-07,
      "loss": 0.0224,
      "step": 3641
    },
    {
      "epoch": 0.5745159127657057,
      "grad_norm": 4.524927616119385,
      "learning_rate": 4.342295926581871e-07,
      "loss": 0.0594,
      "step": 3642
    },
    {
      "epoch": 0.5746736601332966,
      "grad_norm": 4.425346851348877,
      "learning_rate": 4.3406858798905166e-07,
      "loss": 0.0448,
      "step": 3643
    },
    {
      "epoch": 0.5748314075008873,
      "grad_norm": 2.897800922393799,
      "learning_rate": 4.3390758331991626e-07,
      "loss": 0.026,
      "step": 3644
    },
    {
      "epoch": 0.5749891548684781,
      "grad_norm": 5.019111156463623,
      "learning_rate": 4.3374657865078087e-07,
      "loss": 0.059,
      "step": 3645
    },
    {
      "epoch": 0.5751469022360689,
      "grad_norm": 4.009286880493164,
      "learning_rate": 4.335855739816455e-07,
      "loss": 0.0288,
      "step": 3646
    },
    {
      "epoch": 0.5753046496036598,
      "grad_norm": 7.221583843231201,
      "learning_rate": 4.3342456931251003e-07,
      "loss": 0.0882,
      "step": 3647
    },
    {
      "epoch": 0.5754623969712506,
      "grad_norm": 5.97765588760376,
      "learning_rate": 4.3326356464337464e-07,
      "loss": 0.0824,
      "step": 3648
    },
    {
      "epoch": 0.5756201443388413,
      "grad_norm": 5.377807140350342,
      "learning_rate": 4.3310255997423924e-07,
      "loss": 0.0706,
      "step": 3649
    },
    {
      "epoch": 0.5757778917064321,
      "grad_norm": 2.622694492340088,
      "learning_rate": 4.3294155530510385e-07,
      "loss": 0.0229,
      "step": 3650
    },
    {
      "epoch": 0.5759356390740229,
      "grad_norm": 3.528857707977295,
      "learning_rate": 4.327805506359684e-07,
      "loss": 0.0358,
      "step": 3651
    },
    {
      "epoch": 0.5760933864416138,
      "grad_norm": 2.9627106189727783,
      "learning_rate": 4.32619545966833e-07,
      "loss": 0.0154,
      "step": 3652
    },
    {
      "epoch": 0.5762511338092046,
      "grad_norm": 5.691407203674316,
      "learning_rate": 4.324585412976976e-07,
      "loss": 0.0374,
      "step": 3653
    },
    {
      "epoch": 0.5764088811767953,
      "grad_norm": 7.110293388366699,
      "learning_rate": 4.322975366285622e-07,
      "loss": 0.0608,
      "step": 3654
    },
    {
      "epoch": 0.5765666285443861,
      "grad_norm": 4.146986961364746,
      "learning_rate": 4.321365319594268e-07,
      "loss": 0.0254,
      "step": 3655
    },
    {
      "epoch": 0.576724375911977,
      "grad_norm": 5.152608394622803,
      "learning_rate": 4.3197552729029143e-07,
      "loss": 0.0481,
      "step": 3656
    },
    {
      "epoch": 0.5768821232795678,
      "grad_norm": 7.030542373657227,
      "learning_rate": 4.31814522621156e-07,
      "loss": 0.0655,
      "step": 3657
    },
    {
      "epoch": 0.5770398706471586,
      "grad_norm": 6.886842250823975,
      "learning_rate": 4.316535179520206e-07,
      "loss": 0.0739,
      "step": 3658
    },
    {
      "epoch": 0.5771976180147493,
      "grad_norm": 3.774353504180908,
      "learning_rate": 4.3149251328288515e-07,
      "loss": 0.016,
      "step": 3659
    },
    {
      "epoch": 0.5773553653823402,
      "grad_norm": 5.763378143310547,
      "learning_rate": 4.313315086137498e-07,
      "loss": 0.0478,
      "step": 3660
    },
    {
      "epoch": 0.577513112749931,
      "grad_norm": 4.991652011871338,
      "learning_rate": 4.3117050394461436e-07,
      "loss": 0.0489,
      "step": 3661
    },
    {
      "epoch": 0.5776708601175218,
      "grad_norm": 5.139795303344727,
      "learning_rate": 4.3100949927547897e-07,
      "loss": 0.0507,
      "step": 3662
    },
    {
      "epoch": 0.5778286074851126,
      "grad_norm": 5.135206699371338,
      "learning_rate": 4.308484946063435e-07,
      "loss": 0.0642,
      "step": 3663
    },
    {
      "epoch": 0.5779863548527034,
      "grad_norm": 3.217456340789795,
      "learning_rate": 4.306874899372082e-07,
      "loss": 0.03,
      "step": 3664
    },
    {
      "epoch": 0.5781441022202942,
      "grad_norm": 6.17140007019043,
      "learning_rate": 4.305264852680728e-07,
      "loss": 0.106,
      "step": 3665
    },
    {
      "epoch": 0.578301849587885,
      "grad_norm": 9.629501342773438,
      "learning_rate": 4.3036548059893734e-07,
      "loss": 0.0751,
      "step": 3666
    },
    {
      "epoch": 0.5784595969554758,
      "grad_norm": 4.909124851226807,
      "learning_rate": 4.3020447592980194e-07,
      "loss": 0.0262,
      "step": 3667
    },
    {
      "epoch": 0.5786173443230667,
      "grad_norm": 3.7654974460601807,
      "learning_rate": 4.3004347126066655e-07,
      "loss": 0.0168,
      "step": 3668
    },
    {
      "epoch": 0.5787750916906574,
      "grad_norm": 2.6891441345214844,
      "learning_rate": 4.2988246659153116e-07,
      "loss": 0.0144,
      "step": 3669
    },
    {
      "epoch": 0.5789328390582482,
      "grad_norm": 3.555137872695923,
      "learning_rate": 4.297214619223957e-07,
      "loss": 0.0339,
      "step": 3670
    },
    {
      "epoch": 0.579090586425839,
      "grad_norm": 5.062921047210693,
      "learning_rate": 4.2956045725326037e-07,
      "loss": 0.044,
      "step": 3671
    },
    {
      "epoch": 0.5792483337934298,
      "grad_norm": 6.692593097686768,
      "learning_rate": 4.293994525841249e-07,
      "loss": 0.0726,
      "step": 3672
    },
    {
      "epoch": 0.5794060811610207,
      "grad_norm": 7.257974147796631,
      "learning_rate": 4.2923844791498953e-07,
      "loss": 0.0226,
      "step": 3673
    },
    {
      "epoch": 0.5795638285286114,
      "grad_norm": 8.177369117736816,
      "learning_rate": 4.290774432458541e-07,
      "loss": 0.0357,
      "step": 3674
    },
    {
      "epoch": 0.5797215758962022,
      "grad_norm": 6.1049065589904785,
      "learning_rate": 4.2891643857671874e-07,
      "loss": 0.0286,
      "step": 3675
    },
    {
      "epoch": 0.579879323263793,
      "grad_norm": 1.6039459705352783,
      "learning_rate": 4.287554339075833e-07,
      "loss": 0.0111,
      "step": 3676
    },
    {
      "epoch": 0.5800370706313839,
      "grad_norm": 23.160669326782227,
      "learning_rate": 4.285944292384479e-07,
      "loss": 0.0616,
      "step": 3677
    },
    {
      "epoch": 0.5801948179989747,
      "grad_norm": 6.465285778045654,
      "learning_rate": 4.2843342456931245e-07,
      "loss": 0.0437,
      "step": 3678
    },
    {
      "epoch": 0.5803525653665654,
      "grad_norm": 3.6227431297302246,
      "learning_rate": 4.282724199001771e-07,
      "loss": 0.0228,
      "step": 3679
    },
    {
      "epoch": 0.5805103127341562,
      "grad_norm": 4.434861183166504,
      "learning_rate": 4.2811141523104167e-07,
      "loss": 0.0439,
      "step": 3680
    },
    {
      "epoch": 0.5806680601017471,
      "grad_norm": 4.3915228843688965,
      "learning_rate": 4.279504105619063e-07,
      "loss": 0.0152,
      "step": 3681
    },
    {
      "epoch": 0.5808258074693379,
      "grad_norm": 10.597392082214355,
      "learning_rate": 4.2778940589277083e-07,
      "loss": 0.0628,
      "step": 3682
    },
    {
      "epoch": 0.5809835548369287,
      "grad_norm": 11.902427673339844,
      "learning_rate": 4.276284012236355e-07,
      "loss": 0.064,
      "step": 3683
    },
    {
      "epoch": 0.5811413022045194,
      "grad_norm": 6.83543586730957,
      "learning_rate": 4.2746739655450004e-07,
      "loss": 0.0525,
      "step": 3684
    },
    {
      "epoch": 0.5812990495721103,
      "grad_norm": 3.0453226566314697,
      "learning_rate": 4.2730639188536465e-07,
      "loss": 0.0467,
      "step": 3685
    },
    {
      "epoch": 0.5814567969397011,
      "grad_norm": 5.81014347076416,
      "learning_rate": 4.271453872162292e-07,
      "loss": 0.0578,
      "step": 3686
    },
    {
      "epoch": 0.5816145443072919,
      "grad_norm": 2.190941333770752,
      "learning_rate": 4.2698438254709386e-07,
      "loss": 0.0091,
      "step": 3687
    },
    {
      "epoch": 0.5817722916748826,
      "grad_norm": 3.178262710571289,
      "learning_rate": 4.2682337787795847e-07,
      "loss": 0.0694,
      "step": 3688
    },
    {
      "epoch": 0.5819300390424734,
      "grad_norm": 4.956207752227783,
      "learning_rate": 4.26662373208823e-07,
      "loss": 0.033,
      "step": 3689
    },
    {
      "epoch": 0.5820877864100643,
      "grad_norm": 7.0127105712890625,
      "learning_rate": 4.265013685396877e-07,
      "loss": 0.0465,
      "step": 3690
    },
    {
      "epoch": 0.5822455337776551,
      "grad_norm": 0.9440305829048157,
      "learning_rate": 4.2634036387055223e-07,
      "loss": 0.005,
      "step": 3691
    },
    {
      "epoch": 0.5824032811452459,
      "grad_norm": 5.224591255187988,
      "learning_rate": 4.2617935920141684e-07,
      "loss": 0.0556,
      "step": 3692
    },
    {
      "epoch": 0.5825610285128366,
      "grad_norm": 6.958468914031982,
      "learning_rate": 4.260183545322814e-07,
      "loss": 0.0655,
      "step": 3693
    },
    {
      "epoch": 0.5827187758804275,
      "grad_norm": 7.18191385269165,
      "learning_rate": 4.2585734986314605e-07,
      "loss": 0.0813,
      "step": 3694
    },
    {
      "epoch": 0.5828765232480183,
      "grad_norm": 6.075129985809326,
      "learning_rate": 4.256963451940106e-07,
      "loss": 0.0164,
      "step": 3695
    },
    {
      "epoch": 0.5830342706156091,
      "grad_norm": 4.780632495880127,
      "learning_rate": 4.255353405248752e-07,
      "loss": 0.0145,
      "step": 3696
    },
    {
      "epoch": 0.5831920179831999,
      "grad_norm": 5.390576362609863,
      "learning_rate": 4.2537433585573976e-07,
      "loss": 0.0299,
      "step": 3697
    },
    {
      "epoch": 0.5833497653507907,
      "grad_norm": 6.3759846687316895,
      "learning_rate": 4.252133311866044e-07,
      "loss": 0.0443,
      "step": 3698
    },
    {
      "epoch": 0.5835075127183815,
      "grad_norm": 4.434875011444092,
      "learning_rate": 4.25052326517469e-07,
      "loss": 0.0518,
      "step": 3699
    },
    {
      "epoch": 0.5836652600859723,
      "grad_norm": 5.077605724334717,
      "learning_rate": 4.248913218483336e-07,
      "loss": 0.0371,
      "step": 3700
    },
    {
      "epoch": 0.5838230074535631,
      "grad_norm": 3.5893185138702393,
      "learning_rate": 4.2473031717919814e-07,
      "loss": 0.0552,
      "step": 3701
    },
    {
      "epoch": 0.583980754821154,
      "grad_norm": 3.928100347518921,
      "learning_rate": 4.245693125100628e-07,
      "loss": 0.0393,
      "step": 3702
    },
    {
      "epoch": 0.5841385021887447,
      "grad_norm": 4.908411026000977,
      "learning_rate": 4.2440830784092735e-07,
      "loss": 0.0357,
      "step": 3703
    },
    {
      "epoch": 0.5842962495563355,
      "grad_norm": 4.357424736022949,
      "learning_rate": 4.2424730317179195e-07,
      "loss": 0.0249,
      "step": 3704
    },
    {
      "epoch": 0.5844539969239263,
      "grad_norm": 5.003686904907227,
      "learning_rate": 4.2408629850265656e-07,
      "loss": 0.0497,
      "step": 3705
    },
    {
      "epoch": 0.5846117442915172,
      "grad_norm": 4.742331027984619,
      "learning_rate": 4.2392529383352117e-07,
      "loss": 0.027,
      "step": 3706
    },
    {
      "epoch": 0.584769491659108,
      "grad_norm": 8.138925552368164,
      "learning_rate": 4.237642891643857e-07,
      "loss": 0.0615,
      "step": 3707
    },
    {
      "epoch": 0.5849272390266987,
      "grad_norm": 11.250143051147461,
      "learning_rate": 4.2360328449525033e-07,
      "loss": 0.0441,
      "step": 3708
    },
    {
      "epoch": 0.5850849863942895,
      "grad_norm": 3.3631234169006348,
      "learning_rate": 4.2344227982611493e-07,
      "loss": 0.0212,
      "step": 3709
    },
    {
      "epoch": 0.5852427337618803,
      "grad_norm": 5.691203594207764,
      "learning_rate": 4.2328127515697954e-07,
      "loss": 0.0664,
      "step": 3710
    },
    {
      "epoch": 0.5854004811294712,
      "grad_norm": 8.211243629455566,
      "learning_rate": 4.2312027048784415e-07,
      "loss": 0.0247,
      "step": 3711
    },
    {
      "epoch": 0.585558228497062,
      "grad_norm": 4.621741771697998,
      "learning_rate": 4.229592658187087e-07,
      "loss": 0.0175,
      "step": 3712
    },
    {
      "epoch": 0.5857159758646527,
      "grad_norm": 7.450106143951416,
      "learning_rate": 4.2279826114957336e-07,
      "loss": 0.0811,
      "step": 3713
    },
    {
      "epoch": 0.5858737232322435,
      "grad_norm": 3.560211181640625,
      "learning_rate": 4.226372564804379e-07,
      "loss": 0.0275,
      "step": 3714
    },
    {
      "epoch": 0.5860314705998344,
      "grad_norm": 5.3267388343811035,
      "learning_rate": 4.224762518113025e-07,
      "loss": 0.0469,
      "step": 3715
    },
    {
      "epoch": 0.5861892179674252,
      "grad_norm": 8.223593711853027,
      "learning_rate": 4.2231524714216707e-07,
      "loss": 0.0383,
      "step": 3716
    },
    {
      "epoch": 0.586346965335016,
      "grad_norm": 7.442598819732666,
      "learning_rate": 4.2215424247303173e-07,
      "loss": 0.0441,
      "step": 3717
    },
    {
      "epoch": 0.5865047127026067,
      "grad_norm": 6.4629807472229,
      "learning_rate": 4.219932378038963e-07,
      "loss": 0.0241,
      "step": 3718
    },
    {
      "epoch": 0.5866624600701976,
      "grad_norm": 5.52560567855835,
      "learning_rate": 4.218322331347609e-07,
      "loss": 0.0468,
      "step": 3719
    },
    {
      "epoch": 0.5868202074377884,
      "grad_norm": 3.707911729812622,
      "learning_rate": 4.216712284656255e-07,
      "loss": 0.0286,
      "step": 3720
    },
    {
      "epoch": 0.5869779548053792,
      "grad_norm": 6.9518585205078125,
      "learning_rate": 4.215102237964901e-07,
      "loss": 0.0383,
      "step": 3721
    },
    {
      "epoch": 0.58713570217297,
      "grad_norm": 5.35220193862915,
      "learning_rate": 4.2134921912735466e-07,
      "loss": 0.077,
      "step": 3722
    },
    {
      "epoch": 0.5872934495405608,
      "grad_norm": 5.377768516540527,
      "learning_rate": 4.2118821445821926e-07,
      "loss": 0.0225,
      "step": 3723
    },
    {
      "epoch": 0.5874511969081516,
      "grad_norm": 7.410951614379883,
      "learning_rate": 4.2102720978908387e-07,
      "loss": 0.0295,
      "step": 3724
    },
    {
      "epoch": 0.5876089442757424,
      "grad_norm": 3.232762336730957,
      "learning_rate": 4.208662051199485e-07,
      "loss": 0.0387,
      "step": 3725
    },
    {
      "epoch": 0.5877666916433332,
      "grad_norm": 4.976396560668945,
      "learning_rate": 4.2070520045081303e-07,
      "loss": 0.0253,
      "step": 3726
    },
    {
      "epoch": 0.5879244390109241,
      "grad_norm": 5.500836372375488,
      "learning_rate": 4.2054419578167763e-07,
      "loss": 0.0648,
      "step": 3727
    },
    {
      "epoch": 0.5880821863785148,
      "grad_norm": 3.2531464099884033,
      "learning_rate": 4.2038319111254224e-07,
      "loss": 0.0253,
      "step": 3728
    },
    {
      "epoch": 0.5882399337461056,
      "grad_norm": 3.775691032409668,
      "learning_rate": 4.2022218644340685e-07,
      "loss": 0.0228,
      "step": 3729
    },
    {
      "epoch": 0.5883976811136964,
      "grad_norm": 6.692509174346924,
      "learning_rate": 4.200611817742714e-07,
      "loss": 0.0512,
      "step": 3730
    },
    {
      "epoch": 0.5885554284812872,
      "grad_norm": 8.168987274169922,
      "learning_rate": 4.19900177105136e-07,
      "loss": 0.0431,
      "step": 3731
    },
    {
      "epoch": 0.5887131758488781,
      "grad_norm": 9.085684776306152,
      "learning_rate": 4.197391724360006e-07,
      "loss": 0.0628,
      "step": 3732
    },
    {
      "epoch": 0.5888709232164688,
      "grad_norm": 3.0501174926757812,
      "learning_rate": 4.195781677668652e-07,
      "loss": 0.0272,
      "step": 3733
    },
    {
      "epoch": 0.5890286705840596,
      "grad_norm": 5.695446014404297,
      "learning_rate": 4.194171630977298e-07,
      "loss": 0.0235,
      "step": 3734
    },
    {
      "epoch": 0.5891864179516504,
      "grad_norm": 7.079178810119629,
      "learning_rate": 4.1925615842859443e-07,
      "loss": 0.0655,
      "step": 3735
    },
    {
      "epoch": 0.5893441653192413,
      "grad_norm": 4.774590969085693,
      "learning_rate": 4.1909515375945904e-07,
      "loss": 0.0675,
      "step": 3736
    },
    {
      "epoch": 0.589501912686832,
      "grad_norm": 4.56651496887207,
      "learning_rate": 4.189341490903236e-07,
      "loss": 0.0344,
      "step": 3737
    },
    {
      "epoch": 0.5896596600544228,
      "grad_norm": 6.133149147033691,
      "learning_rate": 4.187731444211882e-07,
      "loss": 0.0771,
      "step": 3738
    },
    {
      "epoch": 0.5898174074220136,
      "grad_norm": 5.865187168121338,
      "learning_rate": 4.186121397520528e-07,
      "loss": 0.0435,
      "step": 3739
    },
    {
      "epoch": 0.5899751547896045,
      "grad_norm": 4.250848770141602,
      "learning_rate": 4.184511350829174e-07,
      "loss": 0.0368,
      "step": 3740
    },
    {
      "epoch": 0.5901329021571953,
      "grad_norm": 7.351658821105957,
      "learning_rate": 4.1829013041378196e-07,
      "loss": 0.0586,
      "step": 3741
    },
    {
      "epoch": 0.590290649524786,
      "grad_norm": 7.026547908782959,
      "learning_rate": 4.1812912574464657e-07,
      "loss": 0.0723,
      "step": 3742
    },
    {
      "epoch": 0.5904483968923768,
      "grad_norm": 4.544672012329102,
      "learning_rate": 4.179681210755112e-07,
      "loss": 0.0287,
      "step": 3743
    },
    {
      "epoch": 0.5906061442599677,
      "grad_norm": 4.992133617401123,
      "learning_rate": 4.178071164063758e-07,
      "loss": 0.0434,
      "step": 3744
    },
    {
      "epoch": 0.5907638916275585,
      "grad_norm": 6.560860633850098,
      "learning_rate": 4.1764611173724034e-07,
      "loss": 0.0939,
      "step": 3745
    },
    {
      "epoch": 0.5909216389951493,
      "grad_norm": 3.0830299854278564,
      "learning_rate": 4.1748510706810494e-07,
      "loss": 0.0418,
      "step": 3746
    },
    {
      "epoch": 0.59107938636274,
      "grad_norm": 4.333657264709473,
      "learning_rate": 4.1732410239896955e-07,
      "loss": 0.0564,
      "step": 3747
    },
    {
      "epoch": 0.5912371337303308,
      "grad_norm": 11.215478897094727,
      "learning_rate": 4.1716309772983416e-07,
      "loss": 0.0601,
      "step": 3748
    },
    {
      "epoch": 0.5913948810979217,
      "grad_norm": 8.15578556060791,
      "learning_rate": 4.170020930606987e-07,
      "loss": 0.0632,
      "step": 3749
    },
    {
      "epoch": 0.5915526284655125,
      "grad_norm": 6.492042064666748,
      "learning_rate": 4.1684108839156337e-07,
      "loss": 0.0423,
      "step": 3750
    },
    {
      "epoch": 0.5917103758331033,
      "grad_norm": 6.464126110076904,
      "learning_rate": 4.166800837224279e-07,
      "loss": 0.0563,
      "step": 3751
    },
    {
      "epoch": 0.591868123200694,
      "grad_norm": 4.373297214508057,
      "learning_rate": 4.1651907905329253e-07,
      "loss": 0.0265,
      "step": 3752
    },
    {
      "epoch": 0.5920258705682849,
      "grad_norm": 6.5860772132873535,
      "learning_rate": 4.163580743841571e-07,
      "loss": 0.0612,
      "step": 3753
    },
    {
      "epoch": 0.5921836179358757,
      "grad_norm": 10.905440330505371,
      "learning_rate": 4.1619706971502174e-07,
      "loss": 0.0791,
      "step": 3754
    },
    {
      "epoch": 0.5923413653034665,
      "grad_norm": 5.605947017669678,
      "learning_rate": 4.160360650458863e-07,
      "loss": 0.0905,
      "step": 3755
    },
    {
      "epoch": 0.5924991126710573,
      "grad_norm": 6.811803817749023,
      "learning_rate": 4.158750603767509e-07,
      "loss": 0.0751,
      "step": 3756
    },
    {
      "epoch": 0.5926568600386481,
      "grad_norm": 5.964085102081299,
      "learning_rate": 4.157140557076155e-07,
      "loss": 0.0321,
      "step": 3757
    },
    {
      "epoch": 0.5928146074062389,
      "grad_norm": 5.20408296585083,
      "learning_rate": 4.155530510384801e-07,
      "loss": 0.0687,
      "step": 3758
    },
    {
      "epoch": 0.5929723547738297,
      "grad_norm": 6.977663993835449,
      "learning_rate": 4.153920463693447e-07,
      "loss": 0.1494,
      "step": 3759
    },
    {
      "epoch": 0.5931301021414205,
      "grad_norm": 5.758049488067627,
      "learning_rate": 4.1523104170020927e-07,
      "loss": 0.0372,
      "step": 3760
    },
    {
      "epoch": 0.5932878495090114,
      "grad_norm": 6.385850429534912,
      "learning_rate": 4.150700370310739e-07,
      "loss": 0.0405,
      "step": 3761
    },
    {
      "epoch": 0.5934455968766021,
      "grad_norm": 5.329529285430908,
      "learning_rate": 4.149090323619385e-07,
      "loss": 0.0484,
      "step": 3762
    },
    {
      "epoch": 0.5936033442441929,
      "grad_norm": 6.6791887283325195,
      "learning_rate": 4.147480276928031e-07,
      "loss": 0.0831,
      "step": 3763
    },
    {
      "epoch": 0.5937610916117837,
      "grad_norm": 2.605978012084961,
      "learning_rate": 4.1458702302366765e-07,
      "loss": 0.0175,
      "step": 3764
    },
    {
      "epoch": 0.5939188389793746,
      "grad_norm": 5.62919807434082,
      "learning_rate": 4.144260183545323e-07,
      "loss": 0.0628,
      "step": 3765
    },
    {
      "epoch": 0.5940765863469654,
      "grad_norm": 3.591205596923828,
      "learning_rate": 4.1426501368539686e-07,
      "loss": 0.0138,
      "step": 3766
    },
    {
      "epoch": 0.5942343337145561,
      "grad_norm": 7.075249195098877,
      "learning_rate": 4.1410400901626146e-07,
      "loss": 0.0881,
      "step": 3767
    },
    {
      "epoch": 0.5943920810821469,
      "grad_norm": 7.193486213684082,
      "learning_rate": 4.13943004347126e-07,
      "loss": 0.0773,
      "step": 3768
    },
    {
      "epoch": 0.5945498284497377,
      "grad_norm": 5.35740852355957,
      "learning_rate": 4.137819996779907e-07,
      "loss": 0.0772,
      "step": 3769
    },
    {
      "epoch": 0.5947075758173286,
      "grad_norm": 6.029299259185791,
      "learning_rate": 4.1362099500885523e-07,
      "loss": 0.0478,
      "step": 3770
    },
    {
      "epoch": 0.5948653231849194,
      "grad_norm": 3.6259782314300537,
      "learning_rate": 4.1345999033971984e-07,
      "loss": 0.0134,
      "step": 3771
    },
    {
      "epoch": 0.5950230705525101,
      "grad_norm": 4.932665824890137,
      "learning_rate": 4.132989856705844e-07,
      "loss": 0.0329,
      "step": 3772
    },
    {
      "epoch": 0.5951808179201009,
      "grad_norm": 4.984086990356445,
      "learning_rate": 4.1313798100144905e-07,
      "loss": 0.0355,
      "step": 3773
    },
    {
      "epoch": 0.5953385652876918,
      "grad_norm": 3.5823488235473633,
      "learning_rate": 4.129769763323136e-07,
      "loss": 0.0333,
      "step": 3774
    },
    {
      "epoch": 0.5954963126552826,
      "grad_norm": 4.083855152130127,
      "learning_rate": 4.128159716631782e-07,
      "loss": 0.0549,
      "step": 3775
    },
    {
      "epoch": 0.5956540600228734,
      "grad_norm": 3.4935083389282227,
      "learning_rate": 4.1265496699404276e-07,
      "loss": 0.0655,
      "step": 3776
    },
    {
      "epoch": 0.5958118073904641,
      "grad_norm": 5.366272926330566,
      "learning_rate": 4.124939623249074e-07,
      "loss": 0.0264,
      "step": 3777
    },
    {
      "epoch": 0.595969554758055,
      "grad_norm": 2.7977328300476074,
      "learning_rate": 4.12332957655772e-07,
      "loss": 0.0135,
      "step": 3778
    },
    {
      "epoch": 0.5961273021256458,
      "grad_norm": 5.729579925537109,
      "learning_rate": 4.121719529866366e-07,
      "loss": 0.0309,
      "step": 3779
    },
    {
      "epoch": 0.5962850494932366,
      "grad_norm": 5.854665279388428,
      "learning_rate": 4.120109483175012e-07,
      "loss": 0.0598,
      "step": 3780
    },
    {
      "epoch": 0.5964427968608274,
      "grad_norm": 2.5261032581329346,
      "learning_rate": 4.118499436483658e-07,
      "loss": 0.038,
      "step": 3781
    },
    {
      "epoch": 0.5966005442284182,
      "grad_norm": 3.297360420227051,
      "learning_rate": 4.116889389792304e-07,
      "loss": 0.0344,
      "step": 3782
    },
    {
      "epoch": 0.596758291596009,
      "grad_norm": 3.670506715774536,
      "learning_rate": 4.1152793431009495e-07,
      "loss": 0.0349,
      "step": 3783
    },
    {
      "epoch": 0.5969160389635998,
      "grad_norm": 7.786942005157471,
      "learning_rate": 4.113669296409596e-07,
      "loss": 0.0665,
      "step": 3784
    },
    {
      "epoch": 0.5970737863311906,
      "grad_norm": 1.6498265266418457,
      "learning_rate": 4.1120592497182417e-07,
      "loss": 0.0122,
      "step": 3785
    },
    {
      "epoch": 0.5972315336987813,
      "grad_norm": 4.289611339569092,
      "learning_rate": 4.1104492030268877e-07,
      "loss": 0.0219,
      "step": 3786
    },
    {
      "epoch": 0.5973892810663722,
      "grad_norm": 8.479968070983887,
      "learning_rate": 4.108839156335533e-07,
      "loss": 0.0511,
      "step": 3787
    },
    {
      "epoch": 0.597547028433963,
      "grad_norm": 9.507231712341309,
      "learning_rate": 4.10722910964418e-07,
      "loss": 0.0369,
      "step": 3788
    },
    {
      "epoch": 0.5977047758015538,
      "grad_norm": 5.04258394241333,
      "learning_rate": 4.1056190629528254e-07,
      "loss": 0.042,
      "step": 3789
    },
    {
      "epoch": 0.5978625231691446,
      "grad_norm": 6.5770182609558105,
      "learning_rate": 4.1040090162614714e-07,
      "loss": 0.066,
      "step": 3790
    },
    {
      "epoch": 0.5980202705367355,
      "grad_norm": 7.160120964050293,
      "learning_rate": 4.102398969570117e-07,
      "loss": 0.027,
      "step": 3791
    },
    {
      "epoch": 0.5981780179043262,
      "grad_norm": 2.9768872261047363,
      "learning_rate": 4.1007889228787636e-07,
      "loss": 0.0131,
      "step": 3792
    },
    {
      "epoch": 0.598335765271917,
      "grad_norm": 5.2741923332214355,
      "learning_rate": 4.099178876187409e-07,
      "loss": 0.0514,
      "step": 3793
    },
    {
      "epoch": 0.5984935126395078,
      "grad_norm": 7.782166957855225,
      "learning_rate": 4.097568829496055e-07,
      "loss": 0.0507,
      "step": 3794
    },
    {
      "epoch": 0.5986512600070987,
      "grad_norm": 6.158569812774658,
      "learning_rate": 4.095958782804701e-07,
      "loss": 0.0807,
      "step": 3795
    },
    {
      "epoch": 0.5988090073746895,
      "grad_norm": 4.537302494049072,
      "learning_rate": 4.0943487361133473e-07,
      "loss": 0.0341,
      "step": 3796
    },
    {
      "epoch": 0.5989667547422802,
      "grad_norm": 4.648359775543213,
      "learning_rate": 4.092738689421993e-07,
      "loss": 0.0287,
      "step": 3797
    },
    {
      "epoch": 0.599124502109871,
      "grad_norm": 6.254510879516602,
      "learning_rate": 4.091128642730639e-07,
      "loss": 0.0519,
      "step": 3798
    },
    {
      "epoch": 0.5992822494774619,
      "grad_norm": 5.403506755828857,
      "learning_rate": 4.089518596039285e-07,
      "loss": 0.0362,
      "step": 3799
    },
    {
      "epoch": 0.5994399968450527,
      "grad_norm": 7.727915287017822,
      "learning_rate": 4.087908549347931e-07,
      "loss": 0.0411,
      "step": 3800
    },
    {
      "epoch": 0.5995977442126434,
      "grad_norm": 6.238685607910156,
      "learning_rate": 4.0862985026565766e-07,
      "loss": 0.064,
      "step": 3801
    },
    {
      "epoch": 0.5997554915802342,
      "grad_norm": 6.48141622543335,
      "learning_rate": 4.0846884559652226e-07,
      "loss": 0.0529,
      "step": 3802
    },
    {
      "epoch": 0.5999132389478251,
      "grad_norm": 7.0403313636779785,
      "learning_rate": 4.0830784092738687e-07,
      "loss": 0.0498,
      "step": 3803
    },
    {
      "epoch": 0.6000709863154159,
      "grad_norm": 4.721019268035889,
      "learning_rate": 4.081468362582515e-07,
      "loss": 0.0254,
      "step": 3804
    },
    {
      "epoch": 0.6002287336830067,
      "grad_norm": 5.8263726234436035,
      "learning_rate": 4.079858315891161e-07,
      "loss": 0.0395,
      "step": 3805
    },
    {
      "epoch": 0.6003864810505974,
      "grad_norm": 3.3491733074188232,
      "learning_rate": 4.0782482691998063e-07,
      "loss": 0.0234,
      "step": 3806
    },
    {
      "epoch": 0.6005442284181882,
      "grad_norm": 4.7357587814331055,
      "learning_rate": 4.076638222508453e-07,
      "loss": 0.0301,
      "step": 3807
    },
    {
      "epoch": 0.6007019757857791,
      "grad_norm": 4.4766411781311035,
      "learning_rate": 4.0750281758170985e-07,
      "loss": 0.0508,
      "step": 3808
    },
    {
      "epoch": 0.6008597231533699,
      "grad_norm": 4.65160608291626,
      "learning_rate": 4.0734181291257445e-07,
      "loss": 0.0491,
      "step": 3809
    },
    {
      "epoch": 0.6010174705209607,
      "grad_norm": 4.908505439758301,
      "learning_rate": 4.0718080824343906e-07,
      "loss": 0.0352,
      "step": 3810
    },
    {
      "epoch": 0.6011752178885514,
      "grad_norm": 5.558905124664307,
      "learning_rate": 4.0701980357430367e-07,
      "loss": 0.0613,
      "step": 3811
    },
    {
      "epoch": 0.6013329652561423,
      "grad_norm": 5.562036037445068,
      "learning_rate": 4.068587989051682e-07,
      "loss": 0.0634,
      "step": 3812
    },
    {
      "epoch": 0.6014907126237331,
      "grad_norm": 5.081418991088867,
      "learning_rate": 4.066977942360328e-07,
      "loss": 0.0427,
      "step": 3813
    },
    {
      "epoch": 0.6016484599913239,
      "grad_norm": 5.784884452819824,
      "learning_rate": 4.0653678956689743e-07,
      "loss": 0.0286,
      "step": 3814
    },
    {
      "epoch": 0.6018062073589147,
      "grad_norm": 5.610983848571777,
      "learning_rate": 4.0637578489776204e-07,
      "loss": 0.082,
      "step": 3815
    },
    {
      "epoch": 0.6019639547265055,
      "grad_norm": 3.5575551986694336,
      "learning_rate": 4.062147802286266e-07,
      "loss": 0.0235,
      "step": 3816
    },
    {
      "epoch": 0.6021217020940963,
      "grad_norm": 6.529111385345459,
      "learning_rate": 4.060537755594912e-07,
      "loss": 0.0784,
      "step": 3817
    },
    {
      "epoch": 0.6022794494616871,
      "grad_norm": 2.607593536376953,
      "learning_rate": 4.058927708903558e-07,
      "loss": 0.0117,
      "step": 3818
    },
    {
      "epoch": 0.6024371968292779,
      "grad_norm": 6.788741111755371,
      "learning_rate": 4.057317662212204e-07,
      "loss": 0.0758,
      "step": 3819
    },
    {
      "epoch": 0.6025949441968688,
      "grad_norm": 4.868200778961182,
      "learning_rate": 4.0557076155208496e-07,
      "loss": 0.0335,
      "step": 3820
    },
    {
      "epoch": 0.6027526915644595,
      "grad_norm": 3.741621494293213,
      "learning_rate": 4.0540975688294957e-07,
      "loss": 0.0558,
      "step": 3821
    },
    {
      "epoch": 0.6029104389320503,
      "grad_norm": 4.564605236053467,
      "learning_rate": 4.052487522138142e-07,
      "loss": 0.0287,
      "step": 3822
    },
    {
      "epoch": 0.6030681862996411,
      "grad_norm": 4.796411991119385,
      "learning_rate": 4.050877475446788e-07,
      "loss": 0.0293,
      "step": 3823
    },
    {
      "epoch": 0.6032259336672319,
      "grad_norm": 4.420485019683838,
      "learning_rate": 4.0492674287554334e-07,
      "loss": 0.0641,
      "step": 3824
    },
    {
      "epoch": 0.6033836810348228,
      "grad_norm": 3.2823305130004883,
      "learning_rate": 4.04765738206408e-07,
      "loss": 0.0296,
      "step": 3825
    },
    {
      "epoch": 0.6035414284024135,
      "grad_norm": 3.3454949855804443,
      "learning_rate": 4.0460473353727255e-07,
      "loss": 0.0272,
      "step": 3826
    },
    {
      "epoch": 0.6036991757700043,
      "grad_norm": 6.38124418258667,
      "learning_rate": 4.0444372886813715e-07,
      "loss": 0.0661,
      "step": 3827
    },
    {
      "epoch": 0.6038569231375951,
      "grad_norm": 7.184398651123047,
      "learning_rate": 4.0428272419900176e-07,
      "loss": 0.0561,
      "step": 3828
    },
    {
      "epoch": 0.604014670505186,
      "grad_norm": 5.573564529418945,
      "learning_rate": 4.0412171952986637e-07,
      "loss": 0.0812,
      "step": 3829
    },
    {
      "epoch": 0.6041724178727768,
      "grad_norm": 6.937236785888672,
      "learning_rate": 4.03960714860731e-07,
      "loss": 0.0272,
      "step": 3830
    },
    {
      "epoch": 0.6043301652403675,
      "grad_norm": 8.94973373413086,
      "learning_rate": 4.0379971019159553e-07,
      "loss": 0.1069,
      "step": 3831
    },
    {
      "epoch": 0.6044879126079583,
      "grad_norm": 5.745685577392578,
      "learning_rate": 4.0363870552246013e-07,
      "loss": 0.0458,
      "step": 3832
    },
    {
      "epoch": 0.6046456599755492,
      "grad_norm": 1.876373291015625,
      "learning_rate": 4.0347770085332474e-07,
      "loss": 0.0111,
      "step": 3833
    },
    {
      "epoch": 0.60480340734314,
      "grad_norm": 3.7778372764587402,
      "learning_rate": 4.0331669618418935e-07,
      "loss": 0.046,
      "step": 3834
    },
    {
      "epoch": 0.6049611547107308,
      "grad_norm": 5.344654560089111,
      "learning_rate": 4.031556915150539e-07,
      "loss": 0.0454,
      "step": 3835
    },
    {
      "epoch": 0.6051189020783215,
      "grad_norm": 4.426279544830322,
      "learning_rate": 4.029946868459185e-07,
      "loss": 0.0722,
      "step": 3836
    },
    {
      "epoch": 0.6052766494459124,
      "grad_norm": 2.89872407913208,
      "learning_rate": 4.028336821767831e-07,
      "loss": 0.0113,
      "step": 3837
    },
    {
      "epoch": 0.6054343968135032,
      "grad_norm": 3.577136993408203,
      "learning_rate": 4.026726775076477e-07,
      "loss": 0.0428,
      "step": 3838
    },
    {
      "epoch": 0.605592144181094,
      "grad_norm": 6.849196910858154,
      "learning_rate": 4.0251167283851227e-07,
      "loss": 0.0206,
      "step": 3839
    },
    {
      "epoch": 0.6057498915486847,
      "grad_norm": 6.689548492431641,
      "learning_rate": 4.0235066816937693e-07,
      "loss": 0.0241,
      "step": 3840
    },
    {
      "epoch": 0.6057498915486847,
      "eval_accuracy": 0.9873752187523107,
      "eval_f1": 0.9873752187523107,
      "eval_loss": 0.04061572626233101,
      "eval_runtime": 4708.8828,
      "eval_samples_per_second": 43.079,
      "eval_steps_per_second": 2.693,
      "step": 3840
    },
    {
      "epoch": 0.6059076389162756,
      "grad_norm": 5.607239723205566,
      "learning_rate": 4.021896635002415e-07,
      "loss": 0.0529,
      "step": 3841
    },
    {
      "epoch": 0.6060653862838664,
      "grad_norm": 5.150136947631836,
      "learning_rate": 4.020286588311061e-07,
      "loss": 0.0286,
      "step": 3842
    },
    {
      "epoch": 0.6062231336514572,
      "grad_norm": 2.0050036907196045,
      "learning_rate": 4.0186765416197064e-07,
      "loss": 0.0084,
      "step": 3843
    },
    {
      "epoch": 0.606380881019048,
      "grad_norm": 4.918299674987793,
      "learning_rate": 4.017066494928353e-07,
      "loss": 0.0292,
      "step": 3844
    },
    {
      "epoch": 0.6065386283866387,
      "grad_norm": 4.557953357696533,
      "learning_rate": 4.0154564482369986e-07,
      "loss": 0.0518,
      "step": 3845
    },
    {
      "epoch": 0.6066963757542296,
      "grad_norm": 4.684080123901367,
      "learning_rate": 4.0138464015456446e-07,
      "loss": 0.046,
      "step": 3846
    },
    {
      "epoch": 0.6068541231218204,
      "grad_norm": 4.371344089508057,
      "learning_rate": 4.01223635485429e-07,
      "loss": 0.032,
      "step": 3847
    },
    {
      "epoch": 0.6070118704894112,
      "grad_norm": 4.31361722946167,
      "learning_rate": 4.010626308162937e-07,
      "loss": 0.0412,
      "step": 3848
    },
    {
      "epoch": 0.607169617857002,
      "grad_norm": 2.3368191719055176,
      "learning_rate": 4.0090162614715823e-07,
      "loss": 0.0292,
      "step": 3849
    },
    {
      "epoch": 0.6073273652245929,
      "grad_norm": 4.166183948516846,
      "learning_rate": 4.0074062147802284e-07,
      "loss": 0.03,
      "step": 3850
    },
    {
      "epoch": 0.6074851125921836,
      "grad_norm": 106.28617095947266,
      "learning_rate": 4.0057961680888744e-07,
      "loss": 0.0394,
      "step": 3851
    },
    {
      "epoch": 0.6076428599597744,
      "grad_norm": 7.088550090789795,
      "learning_rate": 4.0041861213975205e-07,
      "loss": 0.074,
      "step": 3852
    },
    {
      "epoch": 0.6078006073273652,
      "grad_norm": 6.301978588104248,
      "learning_rate": 4.0025760747061665e-07,
      "loss": 0.0727,
      "step": 3853
    },
    {
      "epoch": 0.6079583546949561,
      "grad_norm": 4.252588272094727,
      "learning_rate": 4.000966028014812e-07,
      "loss": 0.0327,
      "step": 3854
    },
    {
      "epoch": 0.6081161020625468,
      "grad_norm": 5.474897861480713,
      "learning_rate": 3.9993559813234587e-07,
      "loss": 0.0953,
      "step": 3855
    },
    {
      "epoch": 0.6082738494301376,
      "grad_norm": 3.9940476417541504,
      "learning_rate": 3.997745934632104e-07,
      "loss": 0.039,
      "step": 3856
    },
    {
      "epoch": 0.6084315967977284,
      "grad_norm": 5.6145100593566895,
      "learning_rate": 3.99613588794075e-07,
      "loss": 0.0515,
      "step": 3857
    },
    {
      "epoch": 0.6085893441653193,
      "grad_norm": 5.668785572052002,
      "learning_rate": 3.994525841249396e-07,
      "loss": 0.084,
      "step": 3858
    },
    {
      "epoch": 0.6087470915329101,
      "grad_norm": 3.3662006855010986,
      "learning_rate": 3.9929157945580424e-07,
      "loss": 0.0436,
      "step": 3859
    },
    {
      "epoch": 0.6089048389005008,
      "grad_norm": 4.437554359436035,
      "learning_rate": 3.991305747866688e-07,
      "loss": 0.024,
      "step": 3860
    },
    {
      "epoch": 0.6090625862680916,
      "grad_norm": 2.6105315685272217,
      "learning_rate": 3.989695701175334e-07,
      "loss": 0.014,
      "step": 3861
    },
    {
      "epoch": 0.6092203336356825,
      "grad_norm": 5.951709270477295,
      "learning_rate": 3.9880856544839795e-07,
      "loss": 0.0506,
      "step": 3862
    },
    {
      "epoch": 0.6093780810032733,
      "grad_norm": 4.6668243408203125,
      "learning_rate": 3.986475607792626e-07,
      "loss": 0.0567,
      "step": 3863
    },
    {
      "epoch": 0.6095358283708641,
      "grad_norm": 6.604398727416992,
      "learning_rate": 3.9848655611012716e-07,
      "loss": 0.0624,
      "step": 3864
    },
    {
      "epoch": 0.6096935757384548,
      "grad_norm": 7.9031291007995605,
      "learning_rate": 3.9832555144099177e-07,
      "loss": 0.0755,
      "step": 3865
    },
    {
      "epoch": 0.6098513231060456,
      "grad_norm": 6.474466323852539,
      "learning_rate": 3.981645467718563e-07,
      "loss": 0.041,
      "step": 3866
    },
    {
      "epoch": 0.6100090704736365,
      "grad_norm": 2.133225917816162,
      "learning_rate": 3.98003542102721e-07,
      "loss": 0.0162,
      "step": 3867
    },
    {
      "epoch": 0.6101668178412273,
      "grad_norm": 4.724400520324707,
      "learning_rate": 3.9784253743358554e-07,
      "loss": 0.0895,
      "step": 3868
    },
    {
      "epoch": 0.6103245652088181,
      "grad_norm": 3.4139459133148193,
      "learning_rate": 3.9768153276445014e-07,
      "loss": 0.0322,
      "step": 3869
    },
    {
      "epoch": 0.6104823125764088,
      "grad_norm": 5.139689922332764,
      "learning_rate": 3.975205280953147e-07,
      "loss": 0.0203,
      "step": 3870
    },
    {
      "epoch": 0.6106400599439997,
      "grad_norm": 2.9104020595550537,
      "learning_rate": 3.9735952342617936e-07,
      "loss": 0.0128,
      "step": 3871
    },
    {
      "epoch": 0.6107978073115905,
      "grad_norm": 3.117351770401001,
      "learning_rate": 3.971985187570439e-07,
      "loss": 0.0219,
      "step": 3872
    },
    {
      "epoch": 0.6109555546791813,
      "grad_norm": 4.9865593910217285,
      "learning_rate": 3.970375140879085e-07,
      "loss": 0.0384,
      "step": 3873
    },
    {
      "epoch": 0.6111133020467721,
      "grad_norm": 5.56674861907959,
      "learning_rate": 3.968765094187732e-07,
      "loss": 0.0512,
      "step": 3874
    },
    {
      "epoch": 0.6112710494143629,
      "grad_norm": 6.8554840087890625,
      "learning_rate": 3.9671550474963773e-07,
      "loss": 0.0607,
      "step": 3875
    },
    {
      "epoch": 0.6114287967819537,
      "grad_norm": 4.936234474182129,
      "learning_rate": 3.9655450008050233e-07,
      "loss": 0.0505,
      "step": 3876
    },
    {
      "epoch": 0.6115865441495445,
      "grad_norm": 4.334475517272949,
      "learning_rate": 3.963934954113669e-07,
      "loss": 0.0195,
      "step": 3877
    },
    {
      "epoch": 0.6117442915171353,
      "grad_norm": 3.8036322593688965,
      "learning_rate": 3.9623249074223155e-07,
      "loss": 0.0201,
      "step": 3878
    },
    {
      "epoch": 0.6119020388847262,
      "grad_norm": 4.9143805503845215,
      "learning_rate": 3.960714860730961e-07,
      "loss": 0.023,
      "step": 3879
    },
    {
      "epoch": 0.6120597862523169,
      "grad_norm": 4.306915760040283,
      "learning_rate": 3.959104814039607e-07,
      "loss": 0.0203,
      "step": 3880
    },
    {
      "epoch": 0.6122175336199077,
      "grad_norm": 4.71891450881958,
      "learning_rate": 3.9574947673482526e-07,
      "loss": 0.0478,
      "step": 3881
    },
    {
      "epoch": 0.6123752809874985,
      "grad_norm": 5.184467315673828,
      "learning_rate": 3.955884720656899e-07,
      "loss": 0.0596,
      "step": 3882
    },
    {
      "epoch": 0.6125330283550893,
      "grad_norm": 5.844971656799316,
      "learning_rate": 3.9542746739655447e-07,
      "loss": 0.0636,
      "step": 3883
    },
    {
      "epoch": 0.6126907757226802,
      "grad_norm": 3.3564276695251465,
      "learning_rate": 3.952664627274191e-07,
      "loss": 0.0571,
      "step": 3884
    },
    {
      "epoch": 0.6128485230902709,
      "grad_norm": 4.913923740386963,
      "learning_rate": 3.9510545805828363e-07,
      "loss": 0.0473,
      "step": 3885
    },
    {
      "epoch": 0.6130062704578617,
      "grad_norm": 4.008642196655273,
      "learning_rate": 3.949444533891483e-07,
      "loss": 0.0325,
      "step": 3886
    },
    {
      "epoch": 0.6131640178254525,
      "grad_norm": 3.20485782623291,
      "learning_rate": 3.9478344872001285e-07,
      "loss": 0.0301,
      "step": 3887
    },
    {
      "epoch": 0.6133217651930434,
      "grad_norm": 3.5149877071380615,
      "learning_rate": 3.9462244405087745e-07,
      "loss": 0.0348,
      "step": 3888
    },
    {
      "epoch": 0.6134795125606342,
      "grad_norm": 6.214327335357666,
      "learning_rate": 3.9446143938174206e-07,
      "loss": 0.0578,
      "step": 3889
    },
    {
      "epoch": 0.6136372599282249,
      "grad_norm": 3.9520163536071777,
      "learning_rate": 3.9430043471260666e-07,
      "loss": 0.0373,
      "step": 3890
    },
    {
      "epoch": 0.6137950072958157,
      "grad_norm": 3.7997848987579346,
      "learning_rate": 3.941394300434712e-07,
      "loss": 0.0307,
      "step": 3891
    },
    {
      "epoch": 0.6139527546634066,
      "grad_norm": 4.809603214263916,
      "learning_rate": 3.939784253743358e-07,
      "loss": 0.0336,
      "step": 3892
    },
    {
      "epoch": 0.6141105020309974,
      "grad_norm": 3.4347429275512695,
      "learning_rate": 3.9381742070520043e-07,
      "loss": 0.0324,
      "step": 3893
    },
    {
      "epoch": 0.6142682493985882,
      "grad_norm": 6.23806095123291,
      "learning_rate": 3.9365641603606504e-07,
      "loss": 0.0605,
      "step": 3894
    },
    {
      "epoch": 0.6144259967661789,
      "grad_norm": 8.719413757324219,
      "learning_rate": 3.934954113669296e-07,
      "loss": 0.0445,
      "step": 3895
    },
    {
      "epoch": 0.6145837441337698,
      "grad_norm": 4.386685371398926,
      "learning_rate": 3.933344066977942e-07,
      "loss": 0.0327,
      "step": 3896
    },
    {
      "epoch": 0.6147414915013606,
      "grad_norm": 4.640712738037109,
      "learning_rate": 3.9317340202865886e-07,
      "loss": 0.053,
      "step": 3897
    },
    {
      "epoch": 0.6148992388689514,
      "grad_norm": 4.2591471672058105,
      "learning_rate": 3.930123973595234e-07,
      "loss": 0.0306,
      "step": 3898
    },
    {
      "epoch": 0.6150569862365421,
      "grad_norm": 4.054599285125732,
      "learning_rate": 3.92851392690388e-07,
      "loss": 0.0303,
      "step": 3899
    },
    {
      "epoch": 0.615214733604133,
      "grad_norm": 9.246919631958008,
      "learning_rate": 3.9269038802125257e-07,
      "loss": 0.0719,
      "step": 3900
    },
    {
      "epoch": 0.6153724809717238,
      "grad_norm": 2.701117992401123,
      "learning_rate": 3.9252938335211723e-07,
      "loss": 0.0411,
      "step": 3901
    },
    {
      "epoch": 0.6155302283393146,
      "grad_norm": 5.678535461425781,
      "learning_rate": 3.923683786829818e-07,
      "loss": 0.0645,
      "step": 3902
    },
    {
      "epoch": 0.6156879757069054,
      "grad_norm": 2.281376838684082,
      "learning_rate": 3.922073740138464e-07,
      "loss": 0.0174,
      "step": 3903
    },
    {
      "epoch": 0.6158457230744961,
      "grad_norm": 5.359318256378174,
      "learning_rate": 3.92046369344711e-07,
      "loss": 0.0297,
      "step": 3904
    },
    {
      "epoch": 0.616003470442087,
      "grad_norm": 3.9732272624969482,
      "learning_rate": 3.918853646755756e-07,
      "loss": 0.0116,
      "step": 3905
    },
    {
      "epoch": 0.6161612178096778,
      "grad_norm": 4.528304576873779,
      "learning_rate": 3.9172436000644015e-07,
      "loss": 0.0219,
      "step": 3906
    },
    {
      "epoch": 0.6163189651772686,
      "grad_norm": 3.9752628803253174,
      "learning_rate": 3.9156335533730476e-07,
      "loss": 0.0238,
      "step": 3907
    },
    {
      "epoch": 0.6164767125448594,
      "grad_norm": 3.861325979232788,
      "learning_rate": 3.9140235066816937e-07,
      "loss": 0.031,
      "step": 3908
    },
    {
      "epoch": 0.6166344599124503,
      "grad_norm": 3.2465317249298096,
      "learning_rate": 3.9124134599903397e-07,
      "loss": 0.0262,
      "step": 3909
    },
    {
      "epoch": 0.616792207280041,
      "grad_norm": 4.051419258117676,
      "learning_rate": 3.910803413298985e-07,
      "loss": 0.0194,
      "step": 3910
    },
    {
      "epoch": 0.6169499546476318,
      "grad_norm": 6.009376525878906,
      "learning_rate": 3.9091933666076313e-07,
      "loss": 0.0467,
      "step": 3911
    },
    {
      "epoch": 0.6171077020152226,
      "grad_norm": 4.511587619781494,
      "learning_rate": 3.9075833199162774e-07,
      "loss": 0.0319,
      "step": 3912
    },
    {
      "epoch": 0.6172654493828135,
      "grad_norm": 2.984797954559326,
      "learning_rate": 3.9059732732249234e-07,
      "loss": 0.0412,
      "step": 3913
    },
    {
      "epoch": 0.6174231967504042,
      "grad_norm": 5.0735883712768555,
      "learning_rate": 3.904363226533569e-07,
      "loss": 0.0567,
      "step": 3914
    },
    {
      "epoch": 0.617580944117995,
      "grad_norm": 5.829403877258301,
      "learning_rate": 3.902753179842215e-07,
      "loss": 0.0313,
      "step": 3915
    },
    {
      "epoch": 0.6177386914855858,
      "grad_norm": 3.133254051208496,
      "learning_rate": 3.901143133150861e-07,
      "loss": 0.0169,
      "step": 3916
    },
    {
      "epoch": 0.6178964388531767,
      "grad_norm": 8.683976173400879,
      "learning_rate": 3.899533086459507e-07,
      "loss": 0.0718,
      "step": 3917
    },
    {
      "epoch": 0.6180541862207675,
      "grad_norm": 4.144251346588135,
      "learning_rate": 3.8979230397681527e-07,
      "loss": 0.0549,
      "step": 3918
    },
    {
      "epoch": 0.6182119335883582,
      "grad_norm": 4.386746406555176,
      "learning_rate": 3.8963129930767993e-07,
      "loss": 0.0664,
      "step": 3919
    },
    {
      "epoch": 0.618369680955949,
      "grad_norm": 5.533723831176758,
      "learning_rate": 3.8947029463854454e-07,
      "loss": 0.0232,
      "step": 3920
    },
    {
      "epoch": 0.6185274283235398,
      "grad_norm": 2.7240710258483887,
      "learning_rate": 3.893092899694091e-07,
      "loss": 0.021,
      "step": 3921
    },
    {
      "epoch": 0.6186851756911307,
      "grad_norm": 5.716734409332275,
      "learning_rate": 3.891482853002737e-07,
      "loss": 0.0409,
      "step": 3922
    },
    {
      "epoch": 0.6188429230587215,
      "grad_norm": 11.965015411376953,
      "learning_rate": 3.889872806311383e-07,
      "loss": 0.0609,
      "step": 3923
    },
    {
      "epoch": 0.6190006704263122,
      "grad_norm": 7.14946985244751,
      "learning_rate": 3.888262759620029e-07,
      "loss": 0.1178,
      "step": 3924
    },
    {
      "epoch": 0.619158417793903,
      "grad_norm": 4.608770370483398,
      "learning_rate": 3.8866527129286746e-07,
      "loss": 0.0373,
      "step": 3925
    },
    {
      "epoch": 0.6193161651614939,
      "grad_norm": 3.3860247135162354,
      "learning_rate": 3.8850426662373207e-07,
      "loss": 0.0155,
      "step": 3926
    },
    {
      "epoch": 0.6194739125290847,
      "grad_norm": 3.5459256172180176,
      "learning_rate": 3.883432619545967e-07,
      "loss": 0.0292,
      "step": 3927
    },
    {
      "epoch": 0.6196316598966755,
      "grad_norm": 6.250884532928467,
      "learning_rate": 3.881822572854613e-07,
      "loss": 0.037,
      "step": 3928
    },
    {
      "epoch": 0.6197894072642662,
      "grad_norm": 8.23530101776123,
      "learning_rate": 3.8802125261632583e-07,
      "loss": 0.0614,
      "step": 3929
    },
    {
      "epoch": 0.6199471546318571,
      "grad_norm": 2.993102788925171,
      "learning_rate": 3.8786024794719044e-07,
      "loss": 0.0096,
      "step": 3930
    },
    {
      "epoch": 0.6201049019994479,
      "grad_norm": 5.495903015136719,
      "learning_rate": 3.8769924327805505e-07,
      "loss": 0.0476,
      "step": 3931
    },
    {
      "epoch": 0.6202626493670387,
      "grad_norm": 3.0880420207977295,
      "learning_rate": 3.8753823860891965e-07,
      "loss": 0.0136,
      "step": 3932
    },
    {
      "epoch": 0.6204203967346295,
      "grad_norm": 6.1710052490234375,
      "learning_rate": 3.873772339397842e-07,
      "loss": 0.0548,
      "step": 3933
    },
    {
      "epoch": 0.6205781441022203,
      "grad_norm": 8.93266773223877,
      "learning_rate": 3.8721622927064887e-07,
      "loss": 0.0567,
      "step": 3934
    },
    {
      "epoch": 0.6207358914698111,
      "grad_norm": 8.327640533447266,
      "learning_rate": 3.870552246015134e-07,
      "loss": 0.0502,
      "step": 3935
    },
    {
      "epoch": 0.6208936388374019,
      "grad_norm": 4.471561908721924,
      "learning_rate": 3.86894219932378e-07,
      "loss": 0.058,
      "step": 3936
    },
    {
      "epoch": 0.6210513862049927,
      "grad_norm": 10.008733749389648,
      "learning_rate": 3.867332152632426e-07,
      "loss": 0.0678,
      "step": 3937
    },
    {
      "epoch": 0.6212091335725836,
      "grad_norm": 11.19927978515625,
      "learning_rate": 3.8657221059410724e-07,
      "loss": 0.068,
      "step": 3938
    },
    {
      "epoch": 0.6213668809401743,
      "grad_norm": 5.79336404800415,
      "learning_rate": 3.864112059249718e-07,
      "loss": 0.0683,
      "step": 3939
    },
    {
      "epoch": 0.6215246283077651,
      "grad_norm": 4.393022537231445,
      "learning_rate": 3.862502012558364e-07,
      "loss": 0.05,
      "step": 3940
    },
    {
      "epoch": 0.6216823756753559,
      "grad_norm": 5.813998699188232,
      "learning_rate": 3.8608919658670095e-07,
      "loss": 0.0492,
      "step": 3941
    },
    {
      "epoch": 0.6218401230429467,
      "grad_norm": 4.290627479553223,
      "learning_rate": 3.859281919175656e-07,
      "loss": 0.0462,
      "step": 3942
    },
    {
      "epoch": 0.6219978704105376,
      "grad_norm": 6.6334547996521,
      "learning_rate": 3.857671872484302e-07,
      "loss": 0.0395,
      "step": 3943
    },
    {
      "epoch": 0.6221556177781283,
      "grad_norm": 7.830145835876465,
      "learning_rate": 3.8560618257929477e-07,
      "loss": 0.0584,
      "step": 3944
    },
    {
      "epoch": 0.6223133651457191,
      "grad_norm": 5.8135857582092285,
      "learning_rate": 3.854451779101594e-07,
      "loss": 0.0472,
      "step": 3945
    },
    {
      "epoch": 0.6224711125133099,
      "grad_norm": 14.138686180114746,
      "learning_rate": 3.85284173241024e-07,
      "loss": 0.0655,
      "step": 3946
    },
    {
      "epoch": 0.6226288598809008,
      "grad_norm": 5.751094818115234,
      "learning_rate": 3.851231685718886e-07,
      "loss": 0.0585,
      "step": 3947
    },
    {
      "epoch": 0.6227866072484916,
      "grad_norm": 5.578413009643555,
      "learning_rate": 3.8496216390275314e-07,
      "loss": 0.0653,
      "step": 3948
    },
    {
      "epoch": 0.6229443546160823,
      "grad_norm": 3.894172191619873,
      "learning_rate": 3.848011592336178e-07,
      "loss": 0.0321,
      "step": 3949
    },
    {
      "epoch": 0.6231021019836731,
      "grad_norm": 5.169582366943359,
      "learning_rate": 3.8464015456448235e-07,
      "loss": 0.0373,
      "step": 3950
    },
    {
      "epoch": 0.623259849351264,
      "grad_norm": 3.197829246520996,
      "learning_rate": 3.8447914989534696e-07,
      "loss": 0.0319,
      "step": 3951
    },
    {
      "epoch": 0.6234175967188548,
      "grad_norm": 5.065234184265137,
      "learning_rate": 3.843181452262115e-07,
      "loss": 0.0582,
      "step": 3952
    },
    {
      "epoch": 0.6235753440864455,
      "grad_norm": 3.9794607162475586,
      "learning_rate": 3.841571405570762e-07,
      "loss": 0.0352,
      "step": 3953
    },
    {
      "epoch": 0.6237330914540363,
      "grad_norm": 5.315174579620361,
      "learning_rate": 3.8399613588794073e-07,
      "loss": 0.0537,
      "step": 3954
    },
    {
      "epoch": 0.6238908388216272,
      "grad_norm": 6.888188362121582,
      "learning_rate": 3.8383513121880533e-07,
      "loss": 0.0542,
      "step": 3955
    },
    {
      "epoch": 0.624048586189218,
      "grad_norm": 7.044703960418701,
      "learning_rate": 3.836741265496699e-07,
      "loss": 0.0565,
      "step": 3956
    },
    {
      "epoch": 0.6242063335568088,
      "grad_norm": 6.392137050628662,
      "learning_rate": 3.8351312188053455e-07,
      "loss": 0.0599,
      "step": 3957
    },
    {
      "epoch": 0.6243640809243995,
      "grad_norm": 3.4717397689819336,
      "learning_rate": 3.833521172113991e-07,
      "loss": 0.0163,
      "step": 3958
    },
    {
      "epoch": 0.6245218282919904,
      "grad_norm": 6.944422245025635,
      "learning_rate": 3.831911125422637e-07,
      "loss": 0.061,
      "step": 3959
    },
    {
      "epoch": 0.6246795756595812,
      "grad_norm": 4.242948532104492,
      "learning_rate": 3.8303010787312826e-07,
      "loss": 0.0408,
      "step": 3960
    },
    {
      "epoch": 0.624837323027172,
      "grad_norm": 2.8211543560028076,
      "learning_rate": 3.828691032039929e-07,
      "loss": 0.0193,
      "step": 3961
    },
    {
      "epoch": 0.6249950703947628,
      "grad_norm": 3.514871835708618,
      "learning_rate": 3.8270809853485747e-07,
      "loss": 0.0356,
      "step": 3962
    },
    {
      "epoch": 0.6251528177623535,
      "grad_norm": 4.88889741897583,
      "learning_rate": 3.825470938657221e-07,
      "loss": 0.0494,
      "step": 3963
    },
    {
      "epoch": 0.6253105651299444,
      "grad_norm": 10.978367805480957,
      "learning_rate": 3.823860891965867e-07,
      "loss": 0.0581,
      "step": 3964
    },
    {
      "epoch": 0.6254683124975352,
      "grad_norm": 5.499526500701904,
      "learning_rate": 3.822250845274513e-07,
      "loss": 0.0469,
      "step": 3965
    },
    {
      "epoch": 0.625626059865126,
      "grad_norm": 4.112061500549316,
      "learning_rate": 3.8206407985831584e-07,
      "loss": 0.0152,
      "step": 3966
    },
    {
      "epoch": 0.6257838072327168,
      "grad_norm": 4.566140174865723,
      "learning_rate": 3.8190307518918045e-07,
      "loss": 0.0517,
      "step": 3967
    },
    {
      "epoch": 0.6259415546003076,
      "grad_norm": 5.12914514541626,
      "learning_rate": 3.817420705200451e-07,
      "loss": 0.0567,
      "step": 3968
    },
    {
      "epoch": 0.6260993019678984,
      "grad_norm": 4.391212463378906,
      "learning_rate": 3.8158106585090966e-07,
      "loss": 0.0767,
      "step": 3969
    },
    {
      "epoch": 0.6262570493354892,
      "grad_norm": 4.2883124351501465,
      "learning_rate": 3.8142006118177427e-07,
      "loss": 0.0355,
      "step": 3970
    },
    {
      "epoch": 0.62641479670308,
      "grad_norm": 6.136410236358643,
      "learning_rate": 3.812590565126388e-07,
      "loss": 0.0739,
      "step": 3971
    },
    {
      "epoch": 0.6265725440706709,
      "grad_norm": 3.8228116035461426,
      "learning_rate": 3.810980518435035e-07,
      "loss": 0.0314,
      "step": 3972
    },
    {
      "epoch": 0.6267302914382616,
      "grad_norm": 5.828356742858887,
      "learning_rate": 3.8093704717436804e-07,
      "loss": 0.0812,
      "step": 3973
    },
    {
      "epoch": 0.6268880388058524,
      "grad_norm": 5.330719947814941,
      "learning_rate": 3.8077604250523264e-07,
      "loss": 0.0753,
      "step": 3974
    },
    {
      "epoch": 0.6270457861734432,
      "grad_norm": 6.100672245025635,
      "learning_rate": 3.806150378360972e-07,
      "loss": 0.0713,
      "step": 3975
    },
    {
      "epoch": 0.6272035335410341,
      "grad_norm": 4.633836269378662,
      "learning_rate": 3.8045403316696185e-07,
      "loss": 0.0396,
      "step": 3976
    },
    {
      "epoch": 0.6273612809086249,
      "grad_norm": 7.444503307342529,
      "learning_rate": 3.802930284978264e-07,
      "loss": 0.1086,
      "step": 3977
    },
    {
      "epoch": 0.6275190282762156,
      "grad_norm": 5.09481954574585,
      "learning_rate": 3.80132023828691e-07,
      "loss": 0.033,
      "step": 3978
    },
    {
      "epoch": 0.6276767756438064,
      "grad_norm": 5.039127349853516,
      "learning_rate": 3.799710191595556e-07,
      "loss": 0.0468,
      "step": 3979
    },
    {
      "epoch": 0.6278345230113972,
      "grad_norm": 8.150583267211914,
      "learning_rate": 3.7981001449042023e-07,
      "loss": 0.0346,
      "step": 3980
    },
    {
      "epoch": 0.6279922703789881,
      "grad_norm": 8.27899169921875,
      "learning_rate": 3.796490098212848e-07,
      "loss": 0.0524,
      "step": 3981
    },
    {
      "epoch": 0.6281500177465789,
      "grad_norm": 5.138139247894287,
      "learning_rate": 3.794880051521494e-07,
      "loss": 0.0493,
      "step": 3982
    },
    {
      "epoch": 0.6283077651141696,
      "grad_norm": 8.71139907836914,
      "learning_rate": 3.79327000483014e-07,
      "loss": 0.0797,
      "step": 3983
    },
    {
      "epoch": 0.6284655124817604,
      "grad_norm": 3.703125,
      "learning_rate": 3.791659958138786e-07,
      "loss": 0.0775,
      "step": 3984
    },
    {
      "epoch": 0.6286232598493513,
      "grad_norm": 3.0508437156677246,
      "learning_rate": 3.7900499114474315e-07,
      "loss": 0.0183,
      "step": 3985
    },
    {
      "epoch": 0.6287810072169421,
      "grad_norm": 4.136172294616699,
      "learning_rate": 3.7884398647560776e-07,
      "loss": 0.0507,
      "step": 3986
    },
    {
      "epoch": 0.6289387545845329,
      "grad_norm": 6.061408042907715,
      "learning_rate": 3.7868298180647237e-07,
      "loss": 0.0328,
      "step": 3987
    },
    {
      "epoch": 0.6290965019521236,
      "grad_norm": 4.876362323760986,
      "learning_rate": 3.7852197713733697e-07,
      "loss": 0.0309,
      "step": 3988
    },
    {
      "epoch": 0.6292542493197145,
      "grad_norm": 5.385231018066406,
      "learning_rate": 3.783609724682015e-07,
      "loss": 0.048,
      "step": 3989
    },
    {
      "epoch": 0.6294119966873053,
      "grad_norm": 5.413110733032227,
      "learning_rate": 3.7819996779906613e-07,
      "loss": 0.0398,
      "step": 3990
    },
    {
      "epoch": 0.6295697440548961,
      "grad_norm": 5.107495307922363,
      "learning_rate": 3.780389631299308e-07,
      "loss": 0.0457,
      "step": 3991
    },
    {
      "epoch": 0.6297274914224869,
      "grad_norm": 3.7703804969787598,
      "learning_rate": 3.7787795846079534e-07,
      "loss": 0.0275,
      "step": 3992
    },
    {
      "epoch": 0.6298852387900777,
      "grad_norm": 5.415693283081055,
      "learning_rate": 3.7771695379165995e-07,
      "loss": 0.0603,
      "step": 3993
    },
    {
      "epoch": 0.6300429861576685,
      "grad_norm": 6.143565654754639,
      "learning_rate": 3.7755594912252456e-07,
      "loss": 0.019,
      "step": 3994
    },
    {
      "epoch": 0.6302007335252593,
      "grad_norm": 4.309868812561035,
      "learning_rate": 3.7739494445338916e-07,
      "loss": 0.0542,
      "step": 3995
    },
    {
      "epoch": 0.6303584808928501,
      "grad_norm": 6.542187690734863,
      "learning_rate": 3.772339397842537e-07,
      "loss": 0.0651,
      "step": 3996
    },
    {
      "epoch": 0.630516228260441,
      "grad_norm": 4.054347991943359,
      "learning_rate": 3.770729351151183e-07,
      "loss": 0.0683,
      "step": 3997
    },
    {
      "epoch": 0.6306739756280317,
      "grad_norm": 4.607563495635986,
      "learning_rate": 3.7691193044598293e-07,
      "loss": 0.0368,
      "step": 3998
    },
    {
      "epoch": 0.6308317229956225,
      "grad_norm": 4.678775787353516,
      "learning_rate": 3.7675092577684753e-07,
      "loss": 0.0334,
      "step": 3999
    },
    {
      "epoch": 0.6309894703632133,
      "grad_norm": 6.153867244720459,
      "learning_rate": 3.765899211077121e-07,
      "loss": 0.0921,
      "step": 4000
    },
    {
      "epoch": 0.6311472177308041,
      "grad_norm": 2.6871447563171387,
      "learning_rate": 3.764289164385767e-07,
      "loss": 0.0273,
      "step": 4001
    },
    {
      "epoch": 0.631304965098395,
      "grad_norm": 3.4612224102020264,
      "learning_rate": 3.762679117694413e-07,
      "loss": 0.0384,
      "step": 4002
    },
    {
      "epoch": 0.6314627124659857,
      "grad_norm": 7.3668365478515625,
      "learning_rate": 3.761069071003059e-07,
      "loss": 0.0666,
      "step": 4003
    },
    {
      "epoch": 0.6316204598335765,
      "grad_norm": 4.346674919128418,
      "learning_rate": 3.7594590243117046e-07,
      "loss": 0.0554,
      "step": 4004
    },
    {
      "epoch": 0.6317782072011673,
      "grad_norm": 8.219521522521973,
      "learning_rate": 3.7578489776203507e-07,
      "loss": 0.0513,
      "step": 4005
    },
    {
      "epoch": 0.6319359545687582,
      "grad_norm": 5.268881797790527,
      "learning_rate": 3.7562389309289967e-07,
      "loss": 0.0428,
      "step": 4006
    },
    {
      "epoch": 0.632093701936349,
      "grad_norm": 6.4654107093811035,
      "learning_rate": 3.754628884237643e-07,
      "loss": 0.0729,
      "step": 4007
    },
    {
      "epoch": 0.6322514493039397,
      "grad_norm": 5.147934436798096,
      "learning_rate": 3.7530188375462883e-07,
      "loss": 0.0577,
      "step": 4008
    },
    {
      "epoch": 0.6324091966715305,
      "grad_norm": 6.916091442108154,
      "learning_rate": 3.751408790854935e-07,
      "loss": 0.0428,
      "step": 4009
    },
    {
      "epoch": 0.6325669440391214,
      "grad_norm": 4.3061628341674805,
      "learning_rate": 3.7497987441635805e-07,
      "loss": 0.0426,
      "step": 4010
    },
    {
      "epoch": 0.6327246914067122,
      "grad_norm": 4.644337177276611,
      "learning_rate": 3.7481886974722265e-07,
      "loss": 0.0304,
      "step": 4011
    },
    {
      "epoch": 0.632882438774303,
      "grad_norm": 2.602245807647705,
      "learning_rate": 3.746578650780872e-07,
      "loss": 0.0155,
      "step": 4012
    },
    {
      "epoch": 0.6330401861418937,
      "grad_norm": 4.101451873779297,
      "learning_rate": 3.7449686040895186e-07,
      "loss": 0.0265,
      "step": 4013
    },
    {
      "epoch": 0.6331979335094846,
      "grad_norm": 3.6709542274475098,
      "learning_rate": 3.7433585573981647e-07,
      "loss": 0.0258,
      "step": 4014
    },
    {
      "epoch": 0.6333556808770754,
      "grad_norm": 2.2103958129882812,
      "learning_rate": 3.74174851070681e-07,
      "loss": 0.0298,
      "step": 4015
    },
    {
      "epoch": 0.6335134282446662,
      "grad_norm": 4.5828752517700195,
      "learning_rate": 3.7401384640154563e-07,
      "loss": 0.0494,
      "step": 4016
    },
    {
      "epoch": 0.6336711756122569,
      "grad_norm": 3.546704053878784,
      "learning_rate": 3.7385284173241024e-07,
      "loss": 0.0414,
      "step": 4017
    },
    {
      "epoch": 0.6338289229798477,
      "grad_norm": 3.1935534477233887,
      "learning_rate": 3.7369183706327484e-07,
      "loss": 0.0275,
      "step": 4018
    },
    {
      "epoch": 0.6339866703474386,
      "grad_norm": 3.213179111480713,
      "learning_rate": 3.735308323941394e-07,
      "loss": 0.0375,
      "step": 4019
    },
    {
      "epoch": 0.6341444177150294,
      "grad_norm": 3.9565305709838867,
      "learning_rate": 3.73369827725004e-07,
      "loss": 0.0496,
      "step": 4020
    },
    {
      "epoch": 0.6343021650826202,
      "grad_norm": 4.46652889251709,
      "learning_rate": 3.732088230558686e-07,
      "loss": 0.0328,
      "step": 4021
    },
    {
      "epoch": 0.6344599124502109,
      "grad_norm": 4.671411991119385,
      "learning_rate": 3.730478183867332e-07,
      "loss": 0.0509,
      "step": 4022
    },
    {
      "epoch": 0.6346176598178018,
      "grad_norm": 5.742166042327881,
      "learning_rate": 3.7288681371759777e-07,
      "loss": 0.0404,
      "step": 4023
    },
    {
      "epoch": 0.6347754071853926,
      "grad_norm": 8.105299949645996,
      "learning_rate": 3.7272580904846243e-07,
      "loss": 0.0561,
      "step": 4024
    },
    {
      "epoch": 0.6349331545529834,
      "grad_norm": 5.17732048034668,
      "learning_rate": 3.72564804379327e-07,
      "loss": 0.0545,
      "step": 4025
    },
    {
      "epoch": 0.6350909019205742,
      "grad_norm": 4.43886137008667,
      "learning_rate": 3.724037997101916e-07,
      "loss": 0.0506,
      "step": 4026
    },
    {
      "epoch": 0.635248649288165,
      "grad_norm": 7.071035861968994,
      "learning_rate": 3.7224279504105614e-07,
      "loss": 0.0611,
      "step": 4027
    },
    {
      "epoch": 0.6354063966557558,
      "grad_norm": 3.3725736141204834,
      "learning_rate": 3.720817903719208e-07,
      "loss": 0.0499,
      "step": 4028
    },
    {
      "epoch": 0.6355641440233466,
      "grad_norm": 3.5755679607391357,
      "learning_rate": 3.7192078570278535e-07,
      "loss": 0.0449,
      "step": 4029
    },
    {
      "epoch": 0.6357218913909374,
      "grad_norm": 6.882693767547607,
      "learning_rate": 3.7175978103364996e-07,
      "loss": 0.0309,
      "step": 4030
    },
    {
      "epoch": 0.6358796387585283,
      "grad_norm": 1.3837300539016724,
      "learning_rate": 3.715987763645145e-07,
      "loss": 0.0114,
      "step": 4031
    },
    {
      "epoch": 0.636037386126119,
      "grad_norm": 5.463102340698242,
      "learning_rate": 3.7143777169537917e-07,
      "loss": 0.0325,
      "step": 4032
    },
    {
      "epoch": 0.6361951334937098,
      "grad_norm": 5.296006202697754,
      "learning_rate": 3.712767670262437e-07,
      "loss": 0.0346,
      "step": 4033
    },
    {
      "epoch": 0.6363528808613006,
      "grad_norm": 5.083899974822998,
      "learning_rate": 3.7111576235710833e-07,
      "loss": 0.0698,
      "step": 4034
    },
    {
      "epoch": 0.6365106282288915,
      "grad_norm": 3.9612481594085693,
      "learning_rate": 3.709547576879729e-07,
      "loss": 0.0361,
      "step": 4035
    },
    {
      "epoch": 0.6366683755964823,
      "grad_norm": 4.612145900726318,
      "learning_rate": 3.7079375301883755e-07,
      "loss": 0.026,
      "step": 4036
    },
    {
      "epoch": 0.636826122964073,
      "grad_norm": 2.1052379608154297,
      "learning_rate": 3.7063274834970215e-07,
      "loss": 0.0223,
      "step": 4037
    },
    {
      "epoch": 0.6369838703316638,
      "grad_norm": 5.490352630615234,
      "learning_rate": 3.704717436805667e-07,
      "loss": 0.0305,
      "step": 4038
    },
    {
      "epoch": 0.6371416176992546,
      "grad_norm": 3.52217173576355,
      "learning_rate": 3.7031073901143136e-07,
      "loss": 0.0358,
      "step": 4039
    },
    {
      "epoch": 0.6372993650668455,
      "grad_norm": 5.325096130371094,
      "learning_rate": 3.701497343422959e-07,
      "loss": 0.0169,
      "step": 4040
    },
    {
      "epoch": 0.6374571124344363,
      "grad_norm": 3.8855066299438477,
      "learning_rate": 3.699887296731605e-07,
      "loss": 0.048,
      "step": 4041
    },
    {
      "epoch": 0.637614859802027,
      "grad_norm": 7.730265140533447,
      "learning_rate": 3.698277250040251e-07,
      "loss": 0.0633,
      "step": 4042
    },
    {
      "epoch": 0.6377726071696178,
      "grad_norm": 6.336199760437012,
      "learning_rate": 3.6966672033488974e-07,
      "loss": 0.0216,
      "step": 4043
    },
    {
      "epoch": 0.6379303545372087,
      "grad_norm": 3.8117284774780273,
      "learning_rate": 3.695057156657543e-07,
      "loss": 0.0483,
      "step": 4044
    },
    {
      "epoch": 0.6380881019047995,
      "grad_norm": 2.870159864425659,
      "learning_rate": 3.693447109966189e-07,
      "loss": 0.0116,
      "step": 4045
    },
    {
      "epoch": 0.6382458492723903,
      "grad_norm": 3.458928108215332,
      "learning_rate": 3.6918370632748345e-07,
      "loss": 0.0194,
      "step": 4046
    },
    {
      "epoch": 0.638403596639981,
      "grad_norm": 3.858642101287842,
      "learning_rate": 3.690227016583481e-07,
      "loss": 0.0431,
      "step": 4047
    },
    {
      "epoch": 0.6385613440075719,
      "grad_norm": 5.640664100646973,
      "learning_rate": 3.6886169698921266e-07,
      "loss": 0.0745,
      "step": 4048
    },
    {
      "epoch": 0.6387190913751627,
      "grad_norm": 7.989383697509766,
      "learning_rate": 3.6870069232007727e-07,
      "loss": 0.0548,
      "step": 4049
    },
    {
      "epoch": 0.6388768387427535,
      "grad_norm": 6.7757673263549805,
      "learning_rate": 3.685396876509418e-07,
      "loss": 0.0579,
      "step": 4050
    },
    {
      "epoch": 0.6390345861103442,
      "grad_norm": 8.522056579589844,
      "learning_rate": 3.683786829818065e-07,
      "loss": 0.0603,
      "step": 4051
    },
    {
      "epoch": 0.6391923334779351,
      "grad_norm": 4.27354097366333,
      "learning_rate": 3.6821767831267103e-07,
      "loss": 0.0373,
      "step": 4052
    },
    {
      "epoch": 0.6393500808455259,
      "grad_norm": 3.444039821624756,
      "learning_rate": 3.6805667364353564e-07,
      "loss": 0.0394,
      "step": 4053
    },
    {
      "epoch": 0.6395078282131167,
      "grad_norm": 8.019424438476562,
      "learning_rate": 3.678956689744002e-07,
      "loss": 0.0641,
      "step": 4054
    },
    {
      "epoch": 0.6396655755807075,
      "grad_norm": 7.16523551940918,
      "learning_rate": 3.6773466430526485e-07,
      "loss": 0.0622,
      "step": 4055
    },
    {
      "epoch": 0.6398233229482982,
      "grad_norm": 4.8991875648498535,
      "learning_rate": 3.675736596361294e-07,
      "loss": 0.0196,
      "step": 4056
    },
    {
      "epoch": 0.6399810703158891,
      "grad_norm": 2.6943492889404297,
      "learning_rate": 3.67412654966994e-07,
      "loss": 0.0322,
      "step": 4057
    },
    {
      "epoch": 0.6401388176834799,
      "grad_norm": 7.7071709632873535,
      "learning_rate": 3.672516502978586e-07,
      "loss": 0.043,
      "step": 4058
    },
    {
      "epoch": 0.6402965650510707,
      "grad_norm": 7.444522380828857,
      "learning_rate": 3.670906456287232e-07,
      "loss": 0.0534,
      "step": 4059
    },
    {
      "epoch": 0.6404543124186615,
      "grad_norm": 2.598670244216919,
      "learning_rate": 3.6692964095958783e-07,
      "loss": 0.0112,
      "step": 4060
    },
    {
      "epoch": 0.6406120597862524,
      "grad_norm": 3.433713436126709,
      "learning_rate": 3.667686362904524e-07,
      "loss": 0.018,
      "step": 4061
    },
    {
      "epoch": 0.6407698071538431,
      "grad_norm": 5.276607513427734,
      "learning_rate": 3.6660763162131704e-07,
      "loss": 0.0593,
      "step": 4062
    },
    {
      "epoch": 0.6409275545214339,
      "grad_norm": 6.290041923522949,
      "learning_rate": 3.664466269521816e-07,
      "loss": 0.0257,
      "step": 4063
    },
    {
      "epoch": 0.6410853018890247,
      "grad_norm": 7.995701313018799,
      "learning_rate": 3.662856222830462e-07,
      "loss": 0.0123,
      "step": 4064
    },
    {
      "epoch": 0.6412430492566156,
      "grad_norm": 7.102702617645264,
      "learning_rate": 3.6612461761391076e-07,
      "loss": 0.0304,
      "step": 4065
    },
    {
      "epoch": 0.6414007966242063,
      "grad_norm": 4.317173480987549,
      "learning_rate": 3.659636129447754e-07,
      "loss": 0.0496,
      "step": 4066
    },
    {
      "epoch": 0.6415585439917971,
      "grad_norm": 4.227144718170166,
      "learning_rate": 3.6580260827563997e-07,
      "loss": 0.0602,
      "step": 4067
    },
    {
      "epoch": 0.6417162913593879,
      "grad_norm": 3.030519962310791,
      "learning_rate": 3.656416036065046e-07,
      "loss": 0.0215,
      "step": 4068
    },
    {
      "epoch": 0.6418740387269788,
      "grad_norm": 4.835078716278076,
      "learning_rate": 3.6548059893736913e-07,
      "loss": 0.0244,
      "step": 4069
    },
    {
      "epoch": 0.6420317860945696,
      "grad_norm": 4.8524956703186035,
      "learning_rate": 3.653195942682338e-07,
      "loss": 0.0243,
      "step": 4070
    },
    {
      "epoch": 0.6421895334621603,
      "grad_norm": 4.568508148193359,
      "learning_rate": 3.6515858959909834e-07,
      "loss": 0.0424,
      "step": 4071
    },
    {
      "epoch": 0.6423472808297511,
      "grad_norm": 5.0349812507629395,
      "learning_rate": 3.6499758492996295e-07,
      "loss": 0.0319,
      "step": 4072
    },
    {
      "epoch": 0.642505028197342,
      "grad_norm": 7.431118488311768,
      "learning_rate": 3.6483658026082756e-07,
      "loss": 0.0182,
      "step": 4073
    },
    {
      "epoch": 0.6426627755649328,
      "grad_norm": 10.326835632324219,
      "learning_rate": 3.6467557559169216e-07,
      "loss": 0.0409,
      "step": 4074
    },
    {
      "epoch": 0.6428205229325236,
      "grad_norm": 2.9283664226531982,
      "learning_rate": 3.645145709225567e-07,
      "loss": 0.03,
      "step": 4075
    },
    {
      "epoch": 0.6429782703001143,
      "grad_norm": 3.3292901515960693,
      "learning_rate": 3.643535662534213e-07,
      "loss": 0.0343,
      "step": 4076
    },
    {
      "epoch": 0.6431360176677051,
      "grad_norm": 2.3271267414093018,
      "learning_rate": 3.6419256158428593e-07,
      "loss": 0.0124,
      "step": 4077
    },
    {
      "epoch": 0.643293765035296,
      "grad_norm": 8.10965347290039,
      "learning_rate": 3.6403155691515053e-07,
      "loss": 0.0259,
      "step": 4078
    },
    {
      "epoch": 0.6434515124028868,
      "grad_norm": 8.534149169921875,
      "learning_rate": 3.638705522460151e-07,
      "loss": 0.0792,
      "step": 4079
    },
    {
      "epoch": 0.6436092597704776,
      "grad_norm": 4.74871301651001,
      "learning_rate": 3.637095475768797e-07,
      "loss": 0.0523,
      "step": 4080
    },
    {
      "epoch": 0.6437670071380683,
      "grad_norm": 3.940587043762207,
      "learning_rate": 3.635485429077443e-07,
      "loss": 0.0595,
      "step": 4081
    },
    {
      "epoch": 0.6439247545056592,
      "grad_norm": 4.647641181945801,
      "learning_rate": 3.633875382386089e-07,
      "loss": 0.0433,
      "step": 4082
    },
    {
      "epoch": 0.64408250187325,
      "grad_norm": 2.905236005783081,
      "learning_rate": 3.632265335694735e-07,
      "loss": 0.0263,
      "step": 4083
    },
    {
      "epoch": 0.6442402492408408,
      "grad_norm": 4.514106273651123,
      "learning_rate": 3.6306552890033807e-07,
      "loss": 0.0355,
      "step": 4084
    },
    {
      "epoch": 0.6443979966084316,
      "grad_norm": 5.877621173858643,
      "learning_rate": 3.629045242312027e-07,
      "loss": 0.0353,
      "step": 4085
    },
    {
      "epoch": 0.6445557439760224,
      "grad_norm": 5.227679252624512,
      "learning_rate": 3.627435195620673e-07,
      "loss": 0.0419,
      "step": 4086
    },
    {
      "epoch": 0.6447134913436132,
      "grad_norm": 7.069372653961182,
      "learning_rate": 3.625825148929319e-07,
      "loss": 0.074,
      "step": 4087
    },
    {
      "epoch": 0.644871238711204,
      "grad_norm": 3.09944486618042,
      "learning_rate": 3.624215102237965e-07,
      "loss": 0.0096,
      "step": 4088
    },
    {
      "epoch": 0.6450289860787948,
      "grad_norm": 6.890988349914551,
      "learning_rate": 3.622605055546611e-07,
      "loss": 0.0276,
      "step": 4089
    },
    {
      "epoch": 0.6451867334463857,
      "grad_norm": 12.69340705871582,
      "learning_rate": 3.6209950088552565e-07,
      "loss": 0.0508,
      "step": 4090
    },
    {
      "epoch": 0.6453444808139764,
      "grad_norm": 6.514010429382324,
      "learning_rate": 3.6193849621639026e-07,
      "loss": 0.0544,
      "step": 4091
    },
    {
      "epoch": 0.6455022281815672,
      "grad_norm": 4.55770206451416,
      "learning_rate": 3.6177749154725486e-07,
      "loss": 0.0171,
      "step": 4092
    },
    {
      "epoch": 0.645659975549158,
      "grad_norm": 5.213406085968018,
      "learning_rate": 3.6161648687811947e-07,
      "loss": 0.0405,
      "step": 4093
    },
    {
      "epoch": 0.6458177229167489,
      "grad_norm": 5.1065993309021,
      "learning_rate": 3.61455482208984e-07,
      "loss": 0.0696,
      "step": 4094
    },
    {
      "epoch": 0.6459754702843397,
      "grad_norm": 5.2595014572143555,
      "learning_rate": 3.6129447753984863e-07,
      "loss": 0.0381,
      "step": 4095
    },
    {
      "epoch": 0.6461332176519304,
      "grad_norm": 5.897598743438721,
      "learning_rate": 3.6113347287071324e-07,
      "loss": 0.0506,
      "step": 4096
    },
    {
      "epoch": 0.6462909650195212,
      "grad_norm": 5.073671340942383,
      "learning_rate": 3.6097246820157784e-07,
      "loss": 0.0504,
      "step": 4097
    },
    {
      "epoch": 0.646448712387112,
      "grad_norm": 7.696348667144775,
      "learning_rate": 3.608114635324424e-07,
      "loss": 0.0859,
      "step": 4098
    },
    {
      "epoch": 0.6466064597547029,
      "grad_norm": 7.650763511657715,
      "learning_rate": 3.60650458863307e-07,
      "loss": 0.0776,
      "step": 4099
    },
    {
      "epoch": 0.6467642071222937,
      "grad_norm": 9.237130165100098,
      "learning_rate": 3.604894541941716e-07,
      "loss": 0.0533,
      "step": 4100
    },
    {
      "epoch": 0.6469219544898844,
      "grad_norm": 5.437690734863281,
      "learning_rate": 3.603284495250362e-07,
      "loss": 0.0593,
      "step": 4101
    },
    {
      "epoch": 0.6470797018574752,
      "grad_norm": 5.095547199249268,
      "learning_rate": 3.6016744485590077e-07,
      "loss": 0.0623,
      "step": 4102
    },
    {
      "epoch": 0.6472374492250661,
      "grad_norm": 0.9722776412963867,
      "learning_rate": 3.6000644018676543e-07,
      "loss": 0.0071,
      "step": 4103
    },
    {
      "epoch": 0.6473951965926569,
      "grad_norm": 3.607344150543213,
      "learning_rate": 3.5984543551763e-07,
      "loss": 0.0378,
      "step": 4104
    },
    {
      "epoch": 0.6475529439602477,
      "grad_norm": 3.9352469444274902,
      "learning_rate": 3.596844308484946e-07,
      "loss": 0.0436,
      "step": 4105
    },
    {
      "epoch": 0.6477106913278384,
      "grad_norm": 5.423374652862549,
      "learning_rate": 3.595234261793592e-07,
      "loss": 0.0451,
      "step": 4106
    },
    {
      "epoch": 0.6478684386954293,
      "grad_norm": 2.667495012283325,
      "learning_rate": 3.593624215102238e-07,
      "loss": 0.0253,
      "step": 4107
    },
    {
      "epoch": 0.6480261860630201,
      "grad_norm": 7.697785377502441,
      "learning_rate": 3.592014168410884e-07,
      "loss": 0.0471,
      "step": 4108
    },
    {
      "epoch": 0.6481839334306109,
      "grad_norm": 5.53313684463501,
      "learning_rate": 3.5904041217195296e-07,
      "loss": 0.0425,
      "step": 4109
    },
    {
      "epoch": 0.6483416807982016,
      "grad_norm": 2.814230442047119,
      "learning_rate": 3.5887940750281757e-07,
      "loss": 0.0162,
      "step": 4110
    },
    {
      "epoch": 0.6484994281657925,
      "grad_norm": 3.392780065536499,
      "learning_rate": 3.5871840283368217e-07,
      "loss": 0.05,
      "step": 4111
    },
    {
      "epoch": 0.6486571755333833,
      "grad_norm": 11.159510612487793,
      "learning_rate": 3.585573981645468e-07,
      "loss": 0.0645,
      "step": 4112
    },
    {
      "epoch": 0.6488149229009741,
      "grad_norm": 5.5175042152404785,
      "learning_rate": 3.5839639349541133e-07,
      "loss": 0.0227,
      "step": 4113
    },
    {
      "epoch": 0.6489726702685649,
      "grad_norm": 5.363088607788086,
      "learning_rate": 3.5823538882627594e-07,
      "loss": 0.034,
      "step": 4114
    },
    {
      "epoch": 0.6491304176361556,
      "grad_norm": 4.83908224105835,
      "learning_rate": 3.5807438415714054e-07,
      "loss": 0.0319,
      "step": 4115
    },
    {
      "epoch": 0.6492881650037465,
      "grad_norm": 4.075836181640625,
      "learning_rate": 3.5791337948800515e-07,
      "loss": 0.0155,
      "step": 4116
    },
    {
      "epoch": 0.6494459123713373,
      "grad_norm": 3.4654319286346436,
      "learning_rate": 3.577523748188697e-07,
      "loss": 0.0382,
      "step": 4117
    },
    {
      "epoch": 0.6496036597389281,
      "grad_norm": 5.039188385009766,
      "learning_rate": 3.5759137014973436e-07,
      "loss": 0.0611,
      "step": 4118
    },
    {
      "epoch": 0.6497614071065189,
      "grad_norm": 5.757129192352295,
      "learning_rate": 3.574303654805989e-07,
      "loss": 0.038,
      "step": 4119
    },
    {
      "epoch": 0.6499191544741098,
      "grad_norm": 6.571725368499756,
      "learning_rate": 3.572693608114635e-07,
      "loss": 0.0511,
      "step": 4120
    },
    {
      "epoch": 0.6500769018417005,
      "grad_norm": 3.8214361667633057,
      "learning_rate": 3.571083561423281e-07,
      "loss": 0.0881,
      "step": 4121
    },
    {
      "epoch": 0.6502346492092913,
      "grad_norm": 12.41899299621582,
      "learning_rate": 3.5694735147319274e-07,
      "loss": 0.0888,
      "step": 4122
    },
    {
      "epoch": 0.6503923965768821,
      "grad_norm": 4.831263542175293,
      "learning_rate": 3.567863468040573e-07,
      "loss": 0.0286,
      "step": 4123
    },
    {
      "epoch": 0.650550143944473,
      "grad_norm": 8.912391662597656,
      "learning_rate": 3.566253421349219e-07,
      "loss": 0.1056,
      "step": 4124
    },
    {
      "epoch": 0.6507078913120637,
      "grad_norm": 4.068164825439453,
      "learning_rate": 3.5646433746578645e-07,
      "loss": 0.0564,
      "step": 4125
    },
    {
      "epoch": 0.6508656386796545,
      "grad_norm": 3.028892993927002,
      "learning_rate": 3.563033327966511e-07,
      "loss": 0.0219,
      "step": 4126
    },
    {
      "epoch": 0.6510233860472453,
      "grad_norm": 4.021695137023926,
      "learning_rate": 3.5614232812751566e-07,
      "loss": 0.0556,
      "step": 4127
    },
    {
      "epoch": 0.6511811334148362,
      "grad_norm": 5.660754680633545,
      "learning_rate": 3.5598132345838027e-07,
      "loss": 0.0738,
      "step": 4128
    },
    {
      "epoch": 0.651338880782427,
      "grad_norm": 3.951024055480957,
      "learning_rate": 3.558203187892448e-07,
      "loss": 0.0117,
      "step": 4129
    },
    {
      "epoch": 0.6514966281500177,
      "grad_norm": 2.7424662113189697,
      "learning_rate": 3.556593141201095e-07,
      "loss": 0.0325,
      "step": 4130
    },
    {
      "epoch": 0.6516543755176085,
      "grad_norm": 4.798673629760742,
      "learning_rate": 3.554983094509741e-07,
      "loss": 0.0704,
      "step": 4131
    },
    {
      "epoch": 0.6518121228851994,
      "grad_norm": 10.368926048278809,
      "learning_rate": 3.5533730478183864e-07,
      "loss": 0.0329,
      "step": 4132
    },
    {
      "epoch": 0.6519698702527902,
      "grad_norm": 8.735679626464844,
      "learning_rate": 3.551763001127033e-07,
      "loss": 0.0419,
      "step": 4133
    },
    {
      "epoch": 0.652127617620381,
      "grad_norm": 7.130101680755615,
      "learning_rate": 3.5501529544356785e-07,
      "loss": 0.0541,
      "step": 4134
    },
    {
      "epoch": 0.6522853649879717,
      "grad_norm": 4.491020679473877,
      "learning_rate": 3.5485429077443246e-07,
      "loss": 0.04,
      "step": 4135
    },
    {
      "epoch": 0.6524431123555625,
      "grad_norm": 7.604978084564209,
      "learning_rate": 3.54693286105297e-07,
      "loss": 0.0552,
      "step": 4136
    },
    {
      "epoch": 0.6526008597231534,
      "grad_norm": 3.4317190647125244,
      "learning_rate": 3.5453228143616167e-07,
      "loss": 0.0243,
      "step": 4137
    },
    {
      "epoch": 0.6527586070907442,
      "grad_norm": 16.422517776489258,
      "learning_rate": 3.543712767670262e-07,
      "loss": 0.1136,
      "step": 4138
    },
    {
      "epoch": 0.652916354458335,
      "grad_norm": 2.868276834487915,
      "learning_rate": 3.5421027209789083e-07,
      "loss": 0.0241,
      "step": 4139
    },
    {
      "epoch": 0.6530741018259257,
      "grad_norm": 5.0669989585876465,
      "learning_rate": 3.540492674287554e-07,
      "loss": 0.0287,
      "step": 4140
    },
    {
      "epoch": 0.6532318491935166,
      "grad_norm": 5.366792678833008,
      "learning_rate": 3.5388826275962004e-07,
      "loss": 0.0286,
      "step": 4141
    },
    {
      "epoch": 0.6533895965611074,
      "grad_norm": 6.7800493240356445,
      "learning_rate": 3.537272580904846e-07,
      "loss": 0.0189,
      "step": 4142
    },
    {
      "epoch": 0.6535473439286982,
      "grad_norm": 4.256573677062988,
      "learning_rate": 3.535662534213492e-07,
      "loss": 0.0498,
      "step": 4143
    },
    {
      "epoch": 0.653705091296289,
      "grad_norm": 6.63882303237915,
      "learning_rate": 3.5340524875221376e-07,
      "loss": 0.0751,
      "step": 4144
    },
    {
      "epoch": 0.6538628386638798,
      "grad_norm": 7.1804890632629395,
      "learning_rate": 3.532442440830784e-07,
      "loss": 0.0514,
      "step": 4145
    },
    {
      "epoch": 0.6540205860314706,
      "grad_norm": 5.477293491363525,
      "learning_rate": 3.5308323941394297e-07,
      "loss": 0.033,
      "step": 4146
    },
    {
      "epoch": 0.6541783333990614,
      "grad_norm": 5.306337356567383,
      "learning_rate": 3.529222347448076e-07,
      "loss": 0.0401,
      "step": 4147
    },
    {
      "epoch": 0.6543360807666522,
      "grad_norm": 5.703723430633545,
      "learning_rate": 3.527612300756722e-07,
      "loss": 0.051,
      "step": 4148
    },
    {
      "epoch": 0.6544938281342431,
      "grad_norm": 6.859851360321045,
      "learning_rate": 3.526002254065368e-07,
      "loss": 0.0838,
      "step": 4149
    },
    {
      "epoch": 0.6546515755018338,
      "grad_norm": 4.944122314453125,
      "learning_rate": 3.5243922073740134e-07,
      "loss": 0.0609,
      "step": 4150
    },
    {
      "epoch": 0.6548093228694246,
      "grad_norm": 2.2758069038391113,
      "learning_rate": 3.5227821606826595e-07,
      "loss": 0.0182,
      "step": 4151
    },
    {
      "epoch": 0.6549670702370154,
      "grad_norm": 8.949913024902344,
      "learning_rate": 3.5211721139913055e-07,
      "loss": 0.0527,
      "step": 4152
    },
    {
      "epoch": 0.6551248176046062,
      "grad_norm": 4.411393165588379,
      "learning_rate": 3.5195620672999516e-07,
      "loss": 0.0257,
      "step": 4153
    },
    {
      "epoch": 0.6552825649721971,
      "grad_norm": 4.506752014160156,
      "learning_rate": 3.5179520206085977e-07,
      "loss": 0.0277,
      "step": 4154
    },
    {
      "epoch": 0.6554403123397878,
      "grad_norm": 4.197580814361572,
      "learning_rate": 3.516341973917243e-07,
      "loss": 0.0301,
      "step": 4155
    },
    {
      "epoch": 0.6555980597073786,
      "grad_norm": 4.70909309387207,
      "learning_rate": 3.51473192722589e-07,
      "loss": 0.0328,
      "step": 4156
    },
    {
      "epoch": 0.6557558070749694,
      "grad_norm": 4.624414443969727,
      "learning_rate": 3.5131218805345353e-07,
      "loss": 0.0461,
      "step": 4157
    },
    {
      "epoch": 0.6559135544425603,
      "grad_norm": 4.482852935791016,
      "learning_rate": 3.5115118338431814e-07,
      "loss": 0.0185,
      "step": 4158
    },
    {
      "epoch": 0.656071301810151,
      "grad_norm": 7.920719623565674,
      "learning_rate": 3.509901787151827e-07,
      "loss": 0.06,
      "step": 4159
    },
    {
      "epoch": 0.6562290491777418,
      "grad_norm": 3.98606538772583,
      "learning_rate": 3.5082917404604735e-07,
      "loss": 0.0386,
      "step": 4160
    },
    {
      "epoch": 0.6563867965453326,
      "grad_norm": 3.9634904861450195,
      "learning_rate": 3.506681693769119e-07,
      "loss": 0.0345,
      "step": 4161
    },
    {
      "epoch": 0.6565445439129235,
      "grad_norm": 5.764398097991943,
      "learning_rate": 3.505071647077765e-07,
      "loss": 0.0681,
      "step": 4162
    },
    {
      "epoch": 0.6567022912805143,
      "grad_norm": 4.86509370803833,
      "learning_rate": 3.503461600386411e-07,
      "loss": 0.0353,
      "step": 4163
    },
    {
      "epoch": 0.656860038648105,
      "grad_norm": 5.676183700561523,
      "learning_rate": 3.501851553695057e-07,
      "loss": 0.0266,
      "step": 4164
    },
    {
      "epoch": 0.6570177860156958,
      "grad_norm": 6.578132629394531,
      "learning_rate": 3.500241507003703e-07,
      "loss": 0.064,
      "step": 4165
    },
    {
      "epoch": 0.6571755333832867,
      "grad_norm": 3.3060572147369385,
      "learning_rate": 3.498631460312349e-07,
      "loss": 0.032,
      "step": 4166
    },
    {
      "epoch": 0.6573332807508775,
      "grad_norm": 4.167490005493164,
      "learning_rate": 3.497021413620995e-07,
      "loss": 0.0383,
      "step": 4167
    },
    {
      "epoch": 0.6574910281184683,
      "grad_norm": 3.2244651317596436,
      "learning_rate": 3.495411366929641e-07,
      "loss": 0.0161,
      "step": 4168
    },
    {
      "epoch": 0.657648775486059,
      "grad_norm": 6.713063716888428,
      "learning_rate": 3.4938013202382865e-07,
      "loss": 0.0506,
      "step": 4169
    },
    {
      "epoch": 0.6578065228536499,
      "grad_norm": 8.036482810974121,
      "learning_rate": 3.4921912735469326e-07,
      "loss": 0.0809,
      "step": 4170
    },
    {
      "epoch": 0.6579642702212407,
      "grad_norm": 9.210870742797852,
      "learning_rate": 3.4905812268555786e-07,
      "loss": 0.0779,
      "step": 4171
    },
    {
      "epoch": 0.6581220175888315,
      "grad_norm": 6.666898727416992,
      "learning_rate": 3.4889711801642247e-07,
      "loss": 0.0443,
      "step": 4172
    },
    {
      "epoch": 0.6582797649564223,
      "grad_norm": 3.237283229827881,
      "learning_rate": 3.48736113347287e-07,
      "loss": 0.0131,
      "step": 4173
    },
    {
      "epoch": 0.658437512324013,
      "grad_norm": 6.588843822479248,
      "learning_rate": 3.4857510867815163e-07,
      "loss": 0.0341,
      "step": 4174
    },
    {
      "epoch": 0.6585952596916039,
      "grad_norm": 9.462082862854004,
      "learning_rate": 3.4841410400901623e-07,
      "loss": 0.0871,
      "step": 4175
    },
    {
      "epoch": 0.6587530070591947,
      "grad_norm": 2.9450438022613525,
      "learning_rate": 3.4825309933988084e-07,
      "loss": 0.0546,
      "step": 4176
    },
    {
      "epoch": 0.6589107544267855,
      "grad_norm": 4.6205058097839355,
      "learning_rate": 3.4809209467074545e-07,
      "loss": 0.0616,
      "step": 4177
    },
    {
      "epoch": 0.6590685017943763,
      "grad_norm": 5.313398838043213,
      "learning_rate": 3.4793109000161005e-07,
      "loss": 0.0259,
      "step": 4178
    },
    {
      "epoch": 0.6592262491619671,
      "grad_norm": 12.58399772644043,
      "learning_rate": 3.4777008533247466e-07,
      "loss": 0.0592,
      "step": 4179
    },
    {
      "epoch": 0.6593839965295579,
      "grad_norm": 4.588766098022461,
      "learning_rate": 3.476090806633392e-07,
      "loss": 0.0348,
      "step": 4180
    },
    {
      "epoch": 0.6595417438971487,
      "grad_norm": 4.431708812713623,
      "learning_rate": 3.474480759942038e-07,
      "loss": 0.0355,
      "step": 4181
    },
    {
      "epoch": 0.6596994912647395,
      "grad_norm": 3.6338794231414795,
      "learning_rate": 3.472870713250684e-07,
      "loss": 0.0579,
      "step": 4182
    },
    {
      "epoch": 0.6598572386323304,
      "grad_norm": 5.266646385192871,
      "learning_rate": 3.4712606665593303e-07,
      "loss": 0.054,
      "step": 4183
    },
    {
      "epoch": 0.6600149859999211,
      "grad_norm": 1.8733805418014526,
      "learning_rate": 3.469650619867976e-07,
      "loss": 0.0286,
      "step": 4184
    },
    {
      "epoch": 0.6601727333675119,
      "grad_norm": 2.6201236248016357,
      "learning_rate": 3.468040573176622e-07,
      "loss": 0.0237,
      "step": 4185
    },
    {
      "epoch": 0.6603304807351027,
      "grad_norm": 4.8239827156066895,
      "learning_rate": 3.466430526485268e-07,
      "loss": 0.0303,
      "step": 4186
    },
    {
      "epoch": 0.6604882281026936,
      "grad_norm": 4.803821563720703,
      "learning_rate": 3.464820479793914e-07,
      "loss": 0.033,
      "step": 4187
    },
    {
      "epoch": 0.6606459754702844,
      "grad_norm": 6.170940399169922,
      "learning_rate": 3.4632104331025596e-07,
      "loss": 0.0473,
      "step": 4188
    },
    {
      "epoch": 0.6608037228378751,
      "grad_norm": 1.6020370721817017,
      "learning_rate": 3.4616003864112056e-07,
      "loss": 0.0126,
      "step": 4189
    },
    {
      "epoch": 0.6609614702054659,
      "grad_norm": 5.833771705627441,
      "learning_rate": 3.4599903397198517e-07,
      "loss": 0.0836,
      "step": 4190
    },
    {
      "epoch": 0.6611192175730567,
      "grad_norm": 6.305184841156006,
      "learning_rate": 3.458380293028498e-07,
      "loss": 0.0286,
      "step": 4191
    },
    {
      "epoch": 0.6612769649406476,
      "grad_norm": 12.768031120300293,
      "learning_rate": 3.4567702463371433e-07,
      "loss": 0.0645,
      "step": 4192
    },
    {
      "epoch": 0.6614347123082384,
      "grad_norm": 5.400166988372803,
      "learning_rate": 3.45516019964579e-07,
      "loss": 0.0257,
      "step": 4193
    },
    {
      "epoch": 0.6615924596758291,
      "grad_norm": 2.253767728805542,
      "learning_rate": 3.4535501529544354e-07,
      "loss": 0.0267,
      "step": 4194
    },
    {
      "epoch": 0.6617502070434199,
      "grad_norm": 4.741762638092041,
      "learning_rate": 3.4519401062630815e-07,
      "loss": 0.0288,
      "step": 4195
    },
    {
      "epoch": 0.6619079544110108,
      "grad_norm": 5.254599094390869,
      "learning_rate": 3.450330059571727e-07,
      "loss": 0.0667,
      "step": 4196
    },
    {
      "epoch": 0.6620657017786016,
      "grad_norm": 2.3707597255706787,
      "learning_rate": 3.4487200128803736e-07,
      "loss": 0.051,
      "step": 4197
    },
    {
      "epoch": 0.6622234491461924,
      "grad_norm": 3.265528440475464,
      "learning_rate": 3.447109966189019e-07,
      "loss": 0.0385,
      "step": 4198
    },
    {
      "epoch": 0.6623811965137831,
      "grad_norm": 3.6523993015289307,
      "learning_rate": 3.445499919497665e-07,
      "loss": 0.0626,
      "step": 4199
    },
    {
      "epoch": 0.662538943881374,
      "grad_norm": 4.667764186859131,
      "learning_rate": 3.4438898728063113e-07,
      "loss": 0.0559,
      "step": 4200
    },
    {
      "epoch": 0.6626966912489648,
      "grad_norm": 8.858838081359863,
      "learning_rate": 3.4422798261149573e-07,
      "loss": 0.0963,
      "step": 4201
    },
    {
      "epoch": 0.6628544386165556,
      "grad_norm": 7.327144622802734,
      "learning_rate": 3.4406697794236034e-07,
      "loss": 0.0662,
      "step": 4202
    },
    {
      "epoch": 0.6630121859841464,
      "grad_norm": 4.257870197296143,
      "learning_rate": 3.439059732732249e-07,
      "loss": 0.0517,
      "step": 4203
    },
    {
      "epoch": 0.6631699333517372,
      "grad_norm": 6.316591739654541,
      "learning_rate": 3.437449686040895e-07,
      "loss": 0.0466,
      "step": 4204
    },
    {
      "epoch": 0.663327680719328,
      "grad_norm": 4.2004594802856445,
      "learning_rate": 3.435839639349541e-07,
      "loss": 0.0359,
      "step": 4205
    },
    {
      "epoch": 0.6634854280869188,
      "grad_norm": 6.217618942260742,
      "learning_rate": 3.434229592658187e-07,
      "loss": 0.0308,
      "step": 4206
    },
    {
      "epoch": 0.6636431754545096,
      "grad_norm": 3.3948912620544434,
      "learning_rate": 3.4326195459668327e-07,
      "loss": 0.0388,
      "step": 4207
    },
    {
      "epoch": 0.6638009228221005,
      "grad_norm": 5.694115161895752,
      "learning_rate": 3.431009499275479e-07,
      "loss": 0.0309,
      "step": 4208
    },
    {
      "epoch": 0.6639586701896912,
      "grad_norm": 3.7904810905456543,
      "learning_rate": 3.429399452584125e-07,
      "loss": 0.0198,
      "step": 4209
    },
    {
      "epoch": 0.664116417557282,
      "grad_norm": 6.0549092292785645,
      "learning_rate": 3.427789405892771e-07,
      "loss": 0.0374,
      "step": 4210
    },
    {
      "epoch": 0.6642741649248728,
      "grad_norm": 3.5876405239105225,
      "learning_rate": 3.4261793592014164e-07,
      "loss": 0.028,
      "step": 4211
    },
    {
      "epoch": 0.6644319122924636,
      "grad_norm": 3.949674606323242,
      "learning_rate": 3.424569312510063e-07,
      "loss": 0.0501,
      "step": 4212
    },
    {
      "epoch": 0.6645896596600545,
      "grad_norm": 4.2892069816589355,
      "learning_rate": 3.4229592658187085e-07,
      "loss": 0.0243,
      "step": 4213
    },
    {
      "epoch": 0.6647474070276452,
      "grad_norm": 6.020479679107666,
      "learning_rate": 3.4213492191273546e-07,
      "loss": 0.1264,
      "step": 4214
    },
    {
      "epoch": 0.664905154395236,
      "grad_norm": 3.7382099628448486,
      "learning_rate": 3.419739172436e-07,
      "loss": 0.0297,
      "step": 4215
    },
    {
      "epoch": 0.6650629017628268,
      "grad_norm": 3.4760491847991943,
      "learning_rate": 3.4181291257446467e-07,
      "loss": 0.02,
      "step": 4216
    },
    {
      "epoch": 0.6652206491304177,
      "grad_norm": 5.966982364654541,
      "learning_rate": 3.416519079053292e-07,
      "loss": 0.0269,
      "step": 4217
    },
    {
      "epoch": 0.6653783964980085,
      "grad_norm": 4.919923782348633,
      "learning_rate": 3.4149090323619383e-07,
      "loss": 0.0698,
      "step": 4218
    },
    {
      "epoch": 0.6655361438655992,
      "grad_norm": 10.699564933776855,
      "learning_rate": 3.413298985670584e-07,
      "loss": 0.06,
      "step": 4219
    },
    {
      "epoch": 0.66569389123319,
      "grad_norm": 1.5768576860427856,
      "learning_rate": 3.4116889389792304e-07,
      "loss": 0.0097,
      "step": 4220
    },
    {
      "epoch": 0.6658516386007809,
      "grad_norm": 4.370302200317383,
      "learning_rate": 3.410078892287876e-07,
      "loss": 0.0334,
      "step": 4221
    },
    {
      "epoch": 0.6660093859683717,
      "grad_norm": 9.208818435668945,
      "learning_rate": 3.408468845596522e-07,
      "loss": 0.0628,
      "step": 4222
    },
    {
      "epoch": 0.6661671333359624,
      "grad_norm": 4.493042469024658,
      "learning_rate": 3.4068587989051686e-07,
      "loss": 0.0419,
      "step": 4223
    },
    {
      "epoch": 0.6663248807035532,
      "grad_norm": 2.3730244636535645,
      "learning_rate": 3.405248752213814e-07,
      "loss": 0.0316,
      "step": 4224
    },
    {
      "epoch": 0.6664826280711441,
      "grad_norm": 5.553833961486816,
      "learning_rate": 3.40363870552246e-07,
      "loss": 0.0477,
      "step": 4225
    },
    {
      "epoch": 0.6666403754387349,
      "grad_norm": 6.563319206237793,
      "learning_rate": 3.402028658831106e-07,
      "loss": 0.0242,
      "step": 4226
    },
    {
      "epoch": 0.6667981228063257,
      "grad_norm": 3.1789028644561768,
      "learning_rate": 3.4004186121397523e-07,
      "loss": 0.0207,
      "step": 4227
    },
    {
      "epoch": 0.6669558701739164,
      "grad_norm": 6.585756778717041,
      "learning_rate": 3.398808565448398e-07,
      "loss": 0.0366,
      "step": 4228
    },
    {
      "epoch": 0.6671136175415073,
      "grad_norm": 4.605175971984863,
      "learning_rate": 3.397198518757044e-07,
      "loss": 0.0557,
      "step": 4229
    },
    {
      "epoch": 0.6672713649090981,
      "grad_norm": 1.7443026304244995,
      "learning_rate": 3.3955884720656895e-07,
      "loss": 0.0055,
      "step": 4230
    },
    {
      "epoch": 0.6674291122766889,
      "grad_norm": 10.741480827331543,
      "learning_rate": 3.393978425374336e-07,
      "loss": 0.0625,
      "step": 4231
    },
    {
      "epoch": 0.6675868596442797,
      "grad_norm": 5.386422634124756,
      "learning_rate": 3.3923683786829816e-07,
      "loss": 0.0417,
      "step": 4232
    },
    {
      "epoch": 0.6677446070118704,
      "grad_norm": 3.263134002685547,
      "learning_rate": 3.3907583319916277e-07,
      "loss": 0.0708,
      "step": 4233
    },
    {
      "epoch": 0.6679023543794613,
      "grad_norm": 16.562847137451172,
      "learning_rate": 3.389148285300273e-07,
      "loss": 0.0416,
      "step": 4234
    },
    {
      "epoch": 0.6680601017470521,
      "grad_norm": 4.665866851806641,
      "learning_rate": 3.38753823860892e-07,
      "loss": 0.0315,
      "step": 4235
    },
    {
      "epoch": 0.6682178491146429,
      "grad_norm": 6.751397609710693,
      "learning_rate": 3.3859281919175653e-07,
      "loss": 0.055,
      "step": 4236
    },
    {
      "epoch": 0.6683755964822337,
      "grad_norm": 6.178576946258545,
      "learning_rate": 3.3843181452262114e-07,
      "loss": 0.0387,
      "step": 4237
    },
    {
      "epoch": 0.6685333438498245,
      "grad_norm": 1.5225474834442139,
      "learning_rate": 3.382708098534857e-07,
      "loss": 0.0115,
      "step": 4238
    },
    {
      "epoch": 0.6686910912174153,
      "grad_norm": 5.776130199432373,
      "learning_rate": 3.3810980518435035e-07,
      "loss": 0.038,
      "step": 4239
    },
    {
      "epoch": 0.6688488385850061,
      "grad_norm": 4.210666656494141,
      "learning_rate": 3.379488005152149e-07,
      "loss": 0.0325,
      "step": 4240
    },
    {
      "epoch": 0.6690065859525969,
      "grad_norm": 4.568084239959717,
      "learning_rate": 3.377877958460795e-07,
      "loss": 0.0552,
      "step": 4241
    },
    {
      "epoch": 0.6691643333201878,
      "grad_norm": 3.740183115005493,
      "learning_rate": 3.376267911769441e-07,
      "loss": 0.0464,
      "step": 4242
    },
    {
      "epoch": 0.6693220806877785,
      "grad_norm": 3.6826813220977783,
      "learning_rate": 3.374657865078087e-07,
      "loss": 0.0426,
      "step": 4243
    },
    {
      "epoch": 0.6694798280553693,
      "grad_norm": 7.331096649169922,
      "learning_rate": 3.373047818386733e-07,
      "loss": 0.0952,
      "step": 4244
    },
    {
      "epoch": 0.6696375754229601,
      "grad_norm": 10.238526344299316,
      "learning_rate": 3.371437771695379e-07,
      "loss": 0.0551,
      "step": 4245
    },
    {
      "epoch": 0.669795322790551,
      "grad_norm": 7.426093578338623,
      "learning_rate": 3.3698277250040254e-07,
      "loss": 0.0571,
      "step": 4246
    },
    {
      "epoch": 0.6699530701581418,
      "grad_norm": 4.035054683685303,
      "learning_rate": 3.368217678312671e-07,
      "loss": 0.0272,
      "step": 4247
    },
    {
      "epoch": 0.6701108175257325,
      "grad_norm": 5.987761497497559,
      "learning_rate": 3.366607631621317e-07,
      "loss": 0.0643,
      "step": 4248
    },
    {
      "epoch": 0.6702685648933233,
      "grad_norm": 4.998387813568115,
      "learning_rate": 3.3649975849299625e-07,
      "loss": 0.031,
      "step": 4249
    },
    {
      "epoch": 0.6704263122609141,
      "grad_norm": 2.5684804916381836,
      "learning_rate": 3.363387538238609e-07,
      "loss": 0.0174,
      "step": 4250
    },
    {
      "epoch": 0.670584059628505,
      "grad_norm": 4.180668830871582,
      "learning_rate": 3.3617774915472547e-07,
      "loss": 0.0449,
      "step": 4251
    },
    {
      "epoch": 0.6707418069960958,
      "grad_norm": 8.460077285766602,
      "learning_rate": 3.360167444855901e-07,
      "loss": 0.0965,
      "step": 4252
    },
    {
      "epoch": 0.6708995543636865,
      "grad_norm": 4.493109703063965,
      "learning_rate": 3.3585573981645463e-07,
      "loss": 0.0589,
      "step": 4253
    },
    {
      "epoch": 0.6710573017312773,
      "grad_norm": 6.921128749847412,
      "learning_rate": 3.356947351473193e-07,
      "loss": 0.0487,
      "step": 4254
    },
    {
      "epoch": 0.6712150490988682,
      "grad_norm": 7.861953258514404,
      "learning_rate": 3.3553373047818384e-07,
      "loss": 0.1062,
      "step": 4255
    },
    {
      "epoch": 0.671372796466459,
      "grad_norm": 3.8344569206237793,
      "learning_rate": 3.3537272580904845e-07,
      "loss": 0.0313,
      "step": 4256
    },
    {
      "epoch": 0.6715305438340498,
      "grad_norm": 4.822232723236084,
      "learning_rate": 3.3521172113991305e-07,
      "loss": 0.0378,
      "step": 4257
    },
    {
      "epoch": 0.6716882912016405,
      "grad_norm": 12.531158447265625,
      "learning_rate": 3.3505071647077766e-07,
      "loss": 0.1177,
      "step": 4258
    },
    {
      "epoch": 0.6718460385692314,
      "grad_norm": 6.85503625869751,
      "learning_rate": 3.348897118016422e-07,
      "loss": 0.0474,
      "step": 4259
    },
    {
      "epoch": 0.6720037859368222,
      "grad_norm": 5.491405487060547,
      "learning_rate": 3.347287071325068e-07,
      "loss": 0.0428,
      "step": 4260
    },
    {
      "epoch": 0.672161533304413,
      "grad_norm": 1.6274034976959229,
      "learning_rate": 3.345677024633714e-07,
      "loss": 0.0176,
      "step": 4261
    },
    {
      "epoch": 0.6723192806720037,
      "grad_norm": 11.389778137207031,
      "learning_rate": 3.3440669779423603e-07,
      "loss": 0.0727,
      "step": 4262
    },
    {
      "epoch": 0.6724770280395946,
      "grad_norm": 4.566437244415283,
      "learning_rate": 3.342456931251006e-07,
      "loss": 0.0446,
      "step": 4263
    },
    {
      "epoch": 0.6726347754071854,
      "grad_norm": 6.433427333831787,
      "learning_rate": 3.340846884559652e-07,
      "loss": 0.0446,
      "step": 4264
    },
    {
      "epoch": 0.6727925227747762,
      "grad_norm": 6.395633697509766,
      "learning_rate": 3.339236837868298e-07,
      "loss": 0.0712,
      "step": 4265
    },
    {
      "epoch": 0.672950270142367,
      "grad_norm": 6.634777545928955,
      "learning_rate": 3.337626791176944e-07,
      "loss": 0.0385,
      "step": 4266
    },
    {
      "epoch": 0.6731080175099579,
      "grad_norm": 3.537296772003174,
      "learning_rate": 3.3360167444855896e-07,
      "loss": 0.0313,
      "step": 4267
    },
    {
      "epoch": 0.6732657648775486,
      "grad_norm": 4.52726936340332,
      "learning_rate": 3.3344066977942356e-07,
      "loss": 0.0617,
      "step": 4268
    },
    {
      "epoch": 0.6734235122451394,
      "grad_norm": 7.097104072570801,
      "learning_rate": 3.332796651102882e-07,
      "loss": 0.0645,
      "step": 4269
    },
    {
      "epoch": 0.6735812596127302,
      "grad_norm": 3.4680495262145996,
      "learning_rate": 3.331186604411528e-07,
      "loss": 0.0266,
      "step": 4270
    },
    {
      "epoch": 0.673739006980321,
      "grad_norm": 3.1925270557403564,
      "learning_rate": 3.329576557720174e-07,
      "loss": 0.0421,
      "step": 4271
    },
    {
      "epoch": 0.6738967543479119,
      "grad_norm": 151.88665771484375,
      "learning_rate": 3.32796651102882e-07,
      "loss": 0.0802,
      "step": 4272
    },
    {
      "epoch": 0.6740545017155026,
      "grad_norm": 4.57705020904541,
      "learning_rate": 3.326356464337466e-07,
      "loss": 0.0307,
      "step": 4273
    },
    {
      "epoch": 0.6742122490830934,
      "grad_norm": 5.487161636352539,
      "learning_rate": 3.3247464176461115e-07,
      "loss": 0.0508,
      "step": 4274
    },
    {
      "epoch": 0.6743699964506842,
      "grad_norm": 5.27452278137207,
      "learning_rate": 3.3231363709547575e-07,
      "loss": 0.0557,
      "step": 4275
    },
    {
      "epoch": 0.6745277438182751,
      "grad_norm": 6.933518409729004,
      "learning_rate": 3.3215263242634036e-07,
      "loss": 0.0624,
      "step": 4276
    },
    {
      "epoch": 0.6746854911858658,
      "grad_norm": 6.866750717163086,
      "learning_rate": 3.3199162775720497e-07,
      "loss": 0.0453,
      "step": 4277
    },
    {
      "epoch": 0.6748432385534566,
      "grad_norm": 4.619298458099365,
      "learning_rate": 3.318306230880695e-07,
      "loss": 0.0476,
      "step": 4278
    },
    {
      "epoch": 0.6750009859210474,
      "grad_norm": 5.475695610046387,
      "learning_rate": 3.3166961841893413e-07,
      "loss": 0.0608,
      "step": 4279
    },
    {
      "epoch": 0.6751587332886383,
      "grad_norm": 5.759555816650391,
      "learning_rate": 3.3150861374979873e-07,
      "loss": 0.044,
      "step": 4280
    },
    {
      "epoch": 0.6753164806562291,
      "grad_norm": 6.68592643737793,
      "learning_rate": 3.3134760908066334e-07,
      "loss": 0.0689,
      "step": 4281
    },
    {
      "epoch": 0.6754742280238198,
      "grad_norm": 3.2656190395355225,
      "learning_rate": 3.311866044115279e-07,
      "loss": 0.0328,
      "step": 4282
    },
    {
      "epoch": 0.6756319753914106,
      "grad_norm": 4.156400680541992,
      "learning_rate": 3.310255997423925e-07,
      "loss": 0.0413,
      "step": 4283
    },
    {
      "epoch": 0.6757897227590015,
      "grad_norm": 5.9173583984375,
      "learning_rate": 3.308645950732571e-07,
      "loss": 0.0737,
      "step": 4284
    },
    {
      "epoch": 0.6759474701265923,
      "grad_norm": 2.6478850841522217,
      "learning_rate": 3.307035904041217e-07,
      "loss": 0.0155,
      "step": 4285
    },
    {
      "epoch": 0.6761052174941831,
      "grad_norm": 5.950133800506592,
      "learning_rate": 3.3054258573498626e-07,
      "loss": 0.0385,
      "step": 4286
    },
    {
      "epoch": 0.6762629648617738,
      "grad_norm": 5.7482099533081055,
      "learning_rate": 3.303815810658509e-07,
      "loss": 0.0617,
      "step": 4287
    },
    {
      "epoch": 0.6764207122293646,
      "grad_norm": 1.9521363973617554,
      "learning_rate": 3.302205763967155e-07,
      "loss": 0.0122,
      "step": 4288
    },
    {
      "epoch": 0.6765784595969555,
      "grad_norm": 2.926424264907837,
      "learning_rate": 3.300595717275801e-07,
      "loss": 0.0215,
      "step": 4289
    },
    {
      "epoch": 0.6767362069645463,
      "grad_norm": 5.0904860496521,
      "learning_rate": 3.2989856705844464e-07,
      "loss": 0.0627,
      "step": 4290
    },
    {
      "epoch": 0.6768939543321371,
      "grad_norm": 3.662292003631592,
      "learning_rate": 3.297375623893093e-07,
      "loss": 0.049,
      "step": 4291
    },
    {
      "epoch": 0.6770517016997278,
      "grad_norm": 4.862106800079346,
      "learning_rate": 3.295765577201739e-07,
      "loss": 0.038,
      "step": 4292
    },
    {
      "epoch": 0.6772094490673187,
      "grad_norm": 7.885111331939697,
      "learning_rate": 3.2941555305103846e-07,
      "loss": 0.0509,
      "step": 4293
    },
    {
      "epoch": 0.6773671964349095,
      "grad_norm": 4.848024368286133,
      "learning_rate": 3.2925454838190306e-07,
      "loss": 0.0497,
      "step": 4294
    },
    {
      "epoch": 0.6775249438025003,
      "grad_norm": 7.610829830169678,
      "learning_rate": 3.2909354371276767e-07,
      "loss": 0.06,
      "step": 4295
    },
    {
      "epoch": 0.6776826911700911,
      "grad_norm": 5.329433917999268,
      "learning_rate": 3.289325390436323e-07,
      "loss": 0.0599,
      "step": 4296
    },
    {
      "epoch": 0.6778404385376819,
      "grad_norm": 4.244685173034668,
      "learning_rate": 3.2877153437449683e-07,
      "loss": 0.021,
      "step": 4297
    },
    {
      "epoch": 0.6779981859052727,
      "grad_norm": 3.7655065059661865,
      "learning_rate": 3.2861052970536143e-07,
      "loss": 0.0361,
      "step": 4298
    },
    {
      "epoch": 0.6781559332728635,
      "grad_norm": 6.28010892868042,
      "learning_rate": 3.2844952503622604e-07,
      "loss": 0.0568,
      "step": 4299
    },
    {
      "epoch": 0.6783136806404543,
      "grad_norm": 7.3299055099487305,
      "learning_rate": 3.2828852036709065e-07,
      "loss": 0.0582,
      "step": 4300
    },
    {
      "epoch": 0.6784714280080452,
      "grad_norm": 7.427062034606934,
      "learning_rate": 3.281275156979552e-07,
      "loss": 0.0659,
      "step": 4301
    },
    {
      "epoch": 0.6786291753756359,
      "grad_norm": 3.0374765396118164,
      "learning_rate": 3.2796651102881986e-07,
      "loss": 0.0342,
      "step": 4302
    },
    {
      "epoch": 0.6787869227432267,
      "grad_norm": 10.06717586517334,
      "learning_rate": 3.278055063596844e-07,
      "loss": 0.0529,
      "step": 4303
    },
    {
      "epoch": 0.6789446701108175,
      "grad_norm": 7.636187553405762,
      "learning_rate": 3.27644501690549e-07,
      "loss": 0.0607,
      "step": 4304
    },
    {
      "epoch": 0.6791024174784084,
      "grad_norm": 2.9388339519500732,
      "learning_rate": 3.2748349702141357e-07,
      "loss": 0.0135,
      "step": 4305
    },
    {
      "epoch": 0.6792601648459992,
      "grad_norm": 3.9617867469787598,
      "learning_rate": 3.2732249235227823e-07,
      "loss": 0.0261,
      "step": 4306
    },
    {
      "epoch": 0.6794179122135899,
      "grad_norm": 3.3003311157226562,
      "learning_rate": 3.271614876831428e-07,
      "loss": 0.0218,
      "step": 4307
    },
    {
      "epoch": 0.6795756595811807,
      "grad_norm": 3.664931535720825,
      "learning_rate": 3.270004830140074e-07,
      "loss": 0.0604,
      "step": 4308
    },
    {
      "epoch": 0.6797334069487715,
      "grad_norm": 4.4985857009887695,
      "learning_rate": 3.2683947834487195e-07,
      "loss": 0.0699,
      "step": 4309
    },
    {
      "epoch": 0.6798911543163624,
      "grad_norm": 4.862767219543457,
      "learning_rate": 3.266784736757366e-07,
      "loss": 0.0434,
      "step": 4310
    },
    {
      "epoch": 0.6800489016839532,
      "grad_norm": 4.629154205322266,
      "learning_rate": 3.2651746900660116e-07,
      "loss": 0.031,
      "step": 4311
    },
    {
      "epoch": 0.6802066490515439,
      "grad_norm": 5.003152370452881,
      "learning_rate": 3.2635646433746576e-07,
      "loss": 0.0605,
      "step": 4312
    },
    {
      "epoch": 0.6803643964191347,
      "grad_norm": 8.366888046264648,
      "learning_rate": 3.261954596683303e-07,
      "loss": 0.0345,
      "step": 4313
    },
    {
      "epoch": 0.6805221437867256,
      "grad_norm": 5.949161529541016,
      "learning_rate": 3.26034454999195e-07,
      "loss": 0.0417,
      "step": 4314
    },
    {
      "epoch": 0.6806798911543164,
      "grad_norm": 2.953150749206543,
      "learning_rate": 3.2587345033005953e-07,
      "loss": 0.0154,
      "step": 4315
    },
    {
      "epoch": 0.6808376385219072,
      "grad_norm": 4.614923477172852,
      "learning_rate": 3.2571244566092414e-07,
      "loss": 0.0489,
      "step": 4316
    },
    {
      "epoch": 0.6809953858894979,
      "grad_norm": 7.076268672943115,
      "learning_rate": 3.255514409917888e-07,
      "loss": 0.0555,
      "step": 4317
    },
    {
      "epoch": 0.6811531332570888,
      "grad_norm": 5.182340145111084,
      "learning_rate": 3.2539043632265335e-07,
      "loss": 0.0308,
      "step": 4318
    },
    {
      "epoch": 0.6813108806246796,
      "grad_norm": 5.0712666511535645,
      "learning_rate": 3.2522943165351796e-07,
      "loss": 0.0365,
      "step": 4319
    },
    {
      "epoch": 0.6814686279922704,
      "grad_norm": 5.439001083374023,
      "learning_rate": 3.250684269843825e-07,
      "loss": 0.0423,
      "step": 4320
    },
    {
      "epoch": 0.6816263753598611,
      "grad_norm": 7.498855113983154,
      "learning_rate": 3.2490742231524717e-07,
      "loss": 0.0813,
      "step": 4321
    },
    {
      "epoch": 0.681784122727452,
      "grad_norm": 9.36682415008545,
      "learning_rate": 3.247464176461117e-07,
      "loss": 0.0418,
      "step": 4322
    },
    {
      "epoch": 0.6819418700950428,
      "grad_norm": 6.6287522315979,
      "learning_rate": 3.2458541297697633e-07,
      "loss": 0.1025,
      "step": 4323
    },
    {
      "epoch": 0.6820996174626336,
      "grad_norm": 3.7365763187408447,
      "learning_rate": 3.244244083078409e-07,
      "loss": 0.0277,
      "step": 4324
    },
    {
      "epoch": 0.6822573648302244,
      "grad_norm": 4.683732509613037,
      "learning_rate": 3.2426340363870554e-07,
      "loss": 0.0367,
      "step": 4325
    },
    {
      "epoch": 0.6824151121978151,
      "grad_norm": 6.683684349060059,
      "learning_rate": 3.241023989695701e-07,
      "loss": 0.0389,
      "step": 4326
    },
    {
      "epoch": 0.682572859565406,
      "grad_norm": 5.876407623291016,
      "learning_rate": 3.239413943004347e-07,
      "loss": 0.0854,
      "step": 4327
    },
    {
      "epoch": 0.6827306069329968,
      "grad_norm": 3.6485280990600586,
      "learning_rate": 3.2378038963129925e-07,
      "loss": 0.0292,
      "step": 4328
    },
    {
      "epoch": 0.6828883543005876,
      "grad_norm": 5.209390640258789,
      "learning_rate": 3.236193849621639e-07,
      "loss": 0.0215,
      "step": 4329
    },
    {
      "epoch": 0.6830461016681784,
      "grad_norm": 4.231175422668457,
      "learning_rate": 3.2345838029302847e-07,
      "loss": 0.0724,
      "step": 4330
    },
    {
      "epoch": 0.6832038490357693,
      "grad_norm": 3.57532000541687,
      "learning_rate": 3.2329737562389307e-07,
      "loss": 0.0354,
      "step": 4331
    },
    {
      "epoch": 0.68336159640336,
      "grad_norm": 5.371700763702393,
      "learning_rate": 3.231363709547577e-07,
      "loss": 0.0392,
      "step": 4332
    },
    {
      "epoch": 0.6835193437709508,
      "grad_norm": 5.248849391937256,
      "learning_rate": 3.229753662856223e-07,
      "loss": 0.0308,
      "step": 4333
    },
    {
      "epoch": 0.6836770911385416,
      "grad_norm": 5.769176959991455,
      "learning_rate": 3.2281436161648684e-07,
      "loss": 0.0463,
      "step": 4334
    },
    {
      "epoch": 0.6838348385061325,
      "grad_norm": 2.052978992462158,
      "learning_rate": 3.2265335694735144e-07,
      "loss": 0.0169,
      "step": 4335
    },
    {
      "epoch": 0.6839925858737232,
      "grad_norm": 4.024712085723877,
      "learning_rate": 3.2249235227821605e-07,
      "loss": 0.0284,
      "step": 4336
    },
    {
      "epoch": 0.684150333241314,
      "grad_norm": 4.762681484222412,
      "learning_rate": 3.2233134760908066e-07,
      "loss": 0.0418,
      "step": 4337
    },
    {
      "epoch": 0.6843080806089048,
      "grad_norm": 3.959506034851074,
      "learning_rate": 3.221703429399452e-07,
      "loss": 0.0366,
      "step": 4338
    },
    {
      "epoch": 0.6844658279764957,
      "grad_norm": 3.946669816970825,
      "learning_rate": 3.220093382708098e-07,
      "loss": 0.0447,
      "step": 4339
    },
    {
      "epoch": 0.6846235753440865,
      "grad_norm": 5.558835029602051,
      "learning_rate": 3.218483336016745e-07,
      "loss": 0.0557,
      "step": 4340
    },
    {
      "epoch": 0.6847813227116772,
      "grad_norm": 2.9637818336486816,
      "learning_rate": 3.2168732893253903e-07,
      "loss": 0.0119,
      "step": 4341
    },
    {
      "epoch": 0.684939070079268,
      "grad_norm": 5.928901195526123,
      "learning_rate": 3.2152632426340364e-07,
      "loss": 0.0738,
      "step": 4342
    },
    {
      "epoch": 0.6850968174468589,
      "grad_norm": 4.637601375579834,
      "learning_rate": 3.213653195942682e-07,
      "loss": 0.0391,
      "step": 4343
    },
    {
      "epoch": 0.6852545648144497,
      "grad_norm": 5.620077610015869,
      "learning_rate": 3.2120431492513285e-07,
      "loss": 0.0341,
      "step": 4344
    },
    {
      "epoch": 0.6854123121820405,
      "grad_norm": 5.471549987792969,
      "learning_rate": 3.210433102559974e-07,
      "loss": 0.0798,
      "step": 4345
    },
    {
      "epoch": 0.6855700595496312,
      "grad_norm": 17.964046478271484,
      "learning_rate": 3.20882305586862e-07,
      "loss": 0.0637,
      "step": 4346
    },
    {
      "epoch": 0.685727806917222,
      "grad_norm": 3.0591983795166016,
      "learning_rate": 3.207213009177266e-07,
      "loss": 0.0194,
      "step": 4347
    },
    {
      "epoch": 0.6858855542848129,
      "grad_norm": 5.701319217681885,
      "learning_rate": 3.205602962485912e-07,
      "loss": 0.0551,
      "step": 4348
    },
    {
      "epoch": 0.6860433016524037,
      "grad_norm": 4.1603851318359375,
      "learning_rate": 3.203992915794558e-07,
      "loss": 0.0474,
      "step": 4349
    },
    {
      "epoch": 0.6862010490199945,
      "grad_norm": 9.443543434143066,
      "learning_rate": 3.202382869103204e-07,
      "loss": 0.0295,
      "step": 4350
    },
    {
      "epoch": 0.6863587963875852,
      "grad_norm": 5.248782634735107,
      "learning_rate": 3.20077282241185e-07,
      "loss": 0.0428,
      "step": 4351
    },
    {
      "epoch": 0.6865165437551761,
      "grad_norm": 4.237235069274902,
      "learning_rate": 3.199162775720496e-07,
      "loss": 0.0349,
      "step": 4352
    },
    {
      "epoch": 0.6866742911227669,
      "grad_norm": 3.3554368019104004,
      "learning_rate": 3.1975527290291415e-07,
      "loss": 0.019,
      "step": 4353
    },
    {
      "epoch": 0.6868320384903577,
      "grad_norm": 6.664609909057617,
      "learning_rate": 3.1959426823377875e-07,
      "loss": 0.0383,
      "step": 4354
    },
    {
      "epoch": 0.6869897858579485,
      "grad_norm": 4.52626371383667,
      "learning_rate": 3.1943326356464336e-07,
      "loss": 0.0221,
      "step": 4355
    },
    {
      "epoch": 0.6871475332255393,
      "grad_norm": 5.13330078125,
      "learning_rate": 3.1927225889550797e-07,
      "loss": 0.0715,
      "step": 4356
    },
    {
      "epoch": 0.6873052805931301,
      "grad_norm": 6.823228359222412,
      "learning_rate": 3.191112542263725e-07,
      "loss": 0.0701,
      "step": 4357
    },
    {
      "epoch": 0.6874630279607209,
      "grad_norm": 5.131317138671875,
      "learning_rate": 3.189502495572371e-07,
      "loss": 0.0364,
      "step": 4358
    },
    {
      "epoch": 0.6876207753283117,
      "grad_norm": 1.8192733526229858,
      "learning_rate": 3.1878924488810173e-07,
      "loss": 0.0108,
      "step": 4359
    },
    {
      "epoch": 0.6877785226959026,
      "grad_norm": 3.5216405391693115,
      "learning_rate": 3.1862824021896634e-07,
      "loss": 0.0194,
      "step": 4360
    },
    {
      "epoch": 0.6879362700634933,
      "grad_norm": 3.8323073387145996,
      "learning_rate": 3.184672355498309e-07,
      "loss": 0.0571,
      "step": 4361
    },
    {
      "epoch": 0.6880940174310841,
      "grad_norm": 9.057934761047363,
      "learning_rate": 3.1830623088069555e-07,
      "loss": 0.0722,
      "step": 4362
    },
    {
      "epoch": 0.6882517647986749,
      "grad_norm": 5.102464199066162,
      "learning_rate": 3.1814522621156016e-07,
      "loss": 0.0408,
      "step": 4363
    },
    {
      "epoch": 0.6884095121662658,
      "grad_norm": 9.215035438537598,
      "learning_rate": 3.179842215424247e-07,
      "loss": 0.0647,
      "step": 4364
    },
    {
      "epoch": 0.6885672595338566,
      "grad_norm": 4.922355651855469,
      "learning_rate": 3.178232168732893e-07,
      "loss": 0.0174,
      "step": 4365
    },
    {
      "epoch": 0.6887250069014473,
      "grad_norm": 3.5283076763153076,
      "learning_rate": 3.176622122041539e-07,
      "loss": 0.0623,
      "step": 4366
    },
    {
      "epoch": 0.6888827542690381,
      "grad_norm": 4.64548397064209,
      "learning_rate": 3.1750120753501853e-07,
      "loss": 0.0238,
      "step": 4367
    },
    {
      "epoch": 0.6890405016366289,
      "grad_norm": 4.4998250007629395,
      "learning_rate": 3.173402028658831e-07,
      "loss": 0.0281,
      "step": 4368
    },
    {
      "epoch": 0.6891982490042198,
      "grad_norm": 5.0620927810668945,
      "learning_rate": 3.171791981967477e-07,
      "loss": 0.0402,
      "step": 4369
    },
    {
      "epoch": 0.6893559963718106,
      "grad_norm": 5.032283306121826,
      "learning_rate": 3.170181935276123e-07,
      "loss": 0.0911,
      "step": 4370
    },
    {
      "epoch": 0.6895137437394013,
      "grad_norm": 4.686389923095703,
      "learning_rate": 3.168571888584769e-07,
      "loss": 0.029,
      "step": 4371
    },
    {
      "epoch": 0.6896714911069921,
      "grad_norm": 3.5803253650665283,
      "learning_rate": 3.1669618418934146e-07,
      "loss": 0.0494,
      "step": 4372
    },
    {
      "epoch": 0.689829238474583,
      "grad_norm": 4.782825469970703,
      "learning_rate": 3.1653517952020606e-07,
      "loss": 0.0845,
      "step": 4373
    },
    {
      "epoch": 0.6899869858421738,
      "grad_norm": 4.182551383972168,
      "learning_rate": 3.1637417485107067e-07,
      "loss": 0.0609,
      "step": 4374
    },
    {
      "epoch": 0.6901447332097645,
      "grad_norm": 3.656662702560425,
      "learning_rate": 3.162131701819353e-07,
      "loss": 0.0251,
      "step": 4375
    },
    {
      "epoch": 0.6903024805773553,
      "grad_norm": 3.6005845069885254,
      "learning_rate": 3.1605216551279983e-07,
      "loss": 0.0217,
      "step": 4376
    },
    {
      "epoch": 0.6904602279449462,
      "grad_norm": 5.404850006103516,
      "learning_rate": 3.158911608436645e-07,
      "loss": 0.0388,
      "step": 4377
    },
    {
      "epoch": 0.690617975312537,
      "grad_norm": 5.392271995544434,
      "learning_rate": 3.1573015617452904e-07,
      "loss": 0.0436,
      "step": 4378
    },
    {
      "epoch": 0.6907757226801278,
      "grad_norm": 4.791945934295654,
      "learning_rate": 3.1556915150539365e-07,
      "loss": 0.0294,
      "step": 4379
    },
    {
      "epoch": 0.6909334700477185,
      "grad_norm": 2.6300947666168213,
      "learning_rate": 3.154081468362582e-07,
      "loss": 0.0428,
      "step": 4380
    },
    {
      "epoch": 0.6910912174153094,
      "grad_norm": 6.325275421142578,
      "learning_rate": 3.1524714216712286e-07,
      "loss": 0.0422,
      "step": 4381
    },
    {
      "epoch": 0.6912489647829002,
      "grad_norm": 5.496462821960449,
      "learning_rate": 3.150861374979874e-07,
      "loss": 0.0606,
      "step": 4382
    },
    {
      "epoch": 0.691406712150491,
      "grad_norm": 3.19685435295105,
      "learning_rate": 3.14925132828852e-07,
      "loss": 0.0368,
      "step": 4383
    },
    {
      "epoch": 0.6915644595180818,
      "grad_norm": 4.607407093048096,
      "learning_rate": 3.1476412815971657e-07,
      "loss": 0.0168,
      "step": 4384
    },
    {
      "epoch": 0.6917222068856725,
      "grad_norm": 6.012766361236572,
      "learning_rate": 3.1460312349058123e-07,
      "loss": 0.0433,
      "step": 4385
    },
    {
      "epoch": 0.6918799542532634,
      "grad_norm": 5.831282138824463,
      "learning_rate": 3.1444211882144584e-07,
      "loss": 0.0495,
      "step": 4386
    },
    {
      "epoch": 0.6920377016208542,
      "grad_norm": 5.386518478393555,
      "learning_rate": 3.142811141523104e-07,
      "loss": 0.0516,
      "step": 4387
    },
    {
      "epoch": 0.692195448988445,
      "grad_norm": 5.470829010009766,
      "learning_rate": 3.14120109483175e-07,
      "loss": 0.0507,
      "step": 4388
    },
    {
      "epoch": 0.6923531963560358,
      "grad_norm": 5.152629852294922,
      "learning_rate": 3.139591048140396e-07,
      "loss": 0.0675,
      "step": 4389
    },
    {
      "epoch": 0.6925109437236266,
      "grad_norm": 3.937241792678833,
      "learning_rate": 3.137981001449042e-07,
      "loss": 0.0344,
      "step": 4390
    },
    {
      "epoch": 0.6926686910912174,
      "grad_norm": 5.884132385253906,
      "learning_rate": 3.1363709547576876e-07,
      "loss": 0.041,
      "step": 4391
    },
    {
      "epoch": 0.6928264384588082,
      "grad_norm": 5.485620498657227,
      "learning_rate": 3.134760908066334e-07,
      "loss": 0.0671,
      "step": 4392
    },
    {
      "epoch": 0.692984185826399,
      "grad_norm": 2.248553514480591,
      "learning_rate": 3.13315086137498e-07,
      "loss": 0.0109,
      "step": 4393
    },
    {
      "epoch": 0.6931419331939899,
      "grad_norm": 5.195814609527588,
      "learning_rate": 3.131540814683626e-07,
      "loss": 0.0627,
      "step": 4394
    },
    {
      "epoch": 0.6932996805615806,
      "grad_norm": 10.239777565002441,
      "learning_rate": 3.1299307679922714e-07,
      "loss": 0.1075,
      "step": 4395
    },
    {
      "epoch": 0.6934574279291714,
      "grad_norm": 4.752469062805176,
      "learning_rate": 3.128320721300918e-07,
      "loss": 0.0669,
      "step": 4396
    },
    {
      "epoch": 0.6936151752967622,
      "grad_norm": 5.5641889572143555,
      "learning_rate": 3.1267106746095635e-07,
      "loss": 0.0323,
      "step": 4397
    },
    {
      "epoch": 0.6937729226643531,
      "grad_norm": 3.373412847518921,
      "learning_rate": 3.1251006279182095e-07,
      "loss": 0.0484,
      "step": 4398
    },
    {
      "epoch": 0.6939306700319439,
      "grad_norm": 5.519466400146484,
      "learning_rate": 3.123490581226855e-07,
      "loss": 0.0485,
      "step": 4399
    },
    {
      "epoch": 0.6940884173995346,
      "grad_norm": 8.3934965133667,
      "learning_rate": 3.1218805345355017e-07,
      "loss": 0.0333,
      "step": 4400
    },
    {
      "epoch": 0.6942461647671254,
      "grad_norm": 3.610666275024414,
      "learning_rate": 3.120270487844147e-07,
      "loss": 0.0364,
      "step": 4401
    },
    {
      "epoch": 0.6944039121347163,
      "grad_norm": 3.121730089187622,
      "learning_rate": 3.1186604411527933e-07,
      "loss": 0.0299,
      "step": 4402
    },
    {
      "epoch": 0.6945616595023071,
      "grad_norm": 4.997406005859375,
      "learning_rate": 3.117050394461439e-07,
      "loss": 0.0348,
      "step": 4403
    },
    {
      "epoch": 0.6947194068698979,
      "grad_norm": 3.996657133102417,
      "learning_rate": 3.1154403477700854e-07,
      "loss": 0.0232,
      "step": 4404
    },
    {
      "epoch": 0.6948771542374886,
      "grad_norm": 2.573070526123047,
      "learning_rate": 3.113830301078731e-07,
      "loss": 0.0403,
      "step": 4405
    },
    {
      "epoch": 0.6950349016050794,
      "grad_norm": 6.505340576171875,
      "learning_rate": 3.112220254387377e-07,
      "loss": 0.0318,
      "step": 4406
    },
    {
      "epoch": 0.6951926489726703,
      "grad_norm": 3.81855845451355,
      "learning_rate": 3.1106102076960225e-07,
      "loss": 0.05,
      "step": 4407
    },
    {
      "epoch": 0.6953503963402611,
      "grad_norm": 5.036444187164307,
      "learning_rate": 3.109000161004669e-07,
      "loss": 0.0554,
      "step": 4408
    },
    {
      "epoch": 0.6955081437078519,
      "grad_norm": 4.578446388244629,
      "learning_rate": 3.107390114313315e-07,
      "loss": 0.0224,
      "step": 4409
    },
    {
      "epoch": 0.6956658910754426,
      "grad_norm": 4.750903606414795,
      "learning_rate": 3.1057800676219607e-07,
      "loss": 0.0295,
      "step": 4410
    },
    {
      "epoch": 0.6958236384430335,
      "grad_norm": 4.846446514129639,
      "learning_rate": 3.1041700209306073e-07,
      "loss": 0.0276,
      "step": 4411
    },
    {
      "epoch": 0.6959813858106243,
      "grad_norm": 35.4178581237793,
      "learning_rate": 3.102559974239253e-07,
      "loss": 0.0763,
      "step": 4412
    },
    {
      "epoch": 0.6961391331782151,
      "grad_norm": 5.210877418518066,
      "learning_rate": 3.100949927547899e-07,
      "loss": 0.0483,
      "step": 4413
    },
    {
      "epoch": 0.6962968805458059,
      "grad_norm": 9.870617866516113,
      "learning_rate": 3.0993398808565444e-07,
      "loss": 0.0415,
      "step": 4414
    },
    {
      "epoch": 0.6964546279133967,
      "grad_norm": 5.791896343231201,
      "learning_rate": 3.097729834165191e-07,
      "loss": 0.0401,
      "step": 4415
    },
    {
      "epoch": 0.6966123752809875,
      "grad_norm": 3.8255116939544678,
      "learning_rate": 3.0961197874738366e-07,
      "loss": 0.0138,
      "step": 4416
    },
    {
      "epoch": 0.6967701226485783,
      "grad_norm": 5.592365264892578,
      "learning_rate": 3.0945097407824826e-07,
      "loss": 0.041,
      "step": 4417
    },
    {
      "epoch": 0.6969278700161691,
      "grad_norm": 2.6533119678497314,
      "learning_rate": 3.092899694091128e-07,
      "loss": 0.0253,
      "step": 4418
    },
    {
      "epoch": 0.69708561738376,
      "grad_norm": 7.8833112716674805,
      "learning_rate": 3.091289647399775e-07,
      "loss": 0.061,
      "step": 4419
    },
    {
      "epoch": 0.6972433647513507,
      "grad_norm": 5.523496627807617,
      "learning_rate": 3.0896796007084203e-07,
      "loss": 0.0687,
      "step": 4420
    },
    {
      "epoch": 0.6974011121189415,
      "grad_norm": 6.058192253112793,
      "learning_rate": 3.0880695540170664e-07,
      "loss": 0.03,
      "step": 4421
    },
    {
      "epoch": 0.6975588594865323,
      "grad_norm": 5.791531562805176,
      "learning_rate": 3.086459507325712e-07,
      "loss": 0.0248,
      "step": 4422
    },
    {
      "epoch": 0.6977166068541231,
      "grad_norm": 9.704782485961914,
      "learning_rate": 3.0848494606343585e-07,
      "loss": 0.0689,
      "step": 4423
    },
    {
      "epoch": 0.697874354221714,
      "grad_norm": 8.3525972366333,
      "learning_rate": 3.083239413943004e-07,
      "loss": 0.0311,
      "step": 4424
    },
    {
      "epoch": 0.6980321015893047,
      "grad_norm": 4.394040107727051,
      "learning_rate": 3.08162936725165e-07,
      "loss": 0.0606,
      "step": 4425
    },
    {
      "epoch": 0.6981898489568955,
      "grad_norm": 3.5000298023223877,
      "learning_rate": 3.080019320560296e-07,
      "loss": 0.0668,
      "step": 4426
    },
    {
      "epoch": 0.6983475963244863,
      "grad_norm": 5.3555498123168945,
      "learning_rate": 3.078409273868942e-07,
      "loss": 0.0319,
      "step": 4427
    },
    {
      "epoch": 0.6985053436920772,
      "grad_norm": 4.745055198669434,
      "learning_rate": 3.0767992271775877e-07,
      "loss": 0.0202,
      "step": 4428
    },
    {
      "epoch": 0.698663091059668,
      "grad_norm": 5.928339004516602,
      "learning_rate": 3.075189180486234e-07,
      "loss": 0.0817,
      "step": 4429
    },
    {
      "epoch": 0.6988208384272587,
      "grad_norm": 4.530652046203613,
      "learning_rate": 3.07357913379488e-07,
      "loss": 0.029,
      "step": 4430
    },
    {
      "epoch": 0.6989785857948495,
      "grad_norm": 6.590553283691406,
      "learning_rate": 3.071969087103526e-07,
      "loss": 0.0702,
      "step": 4431
    },
    {
      "epoch": 0.6991363331624404,
      "grad_norm": 6.570598125457764,
      "learning_rate": 3.070359040412172e-07,
      "loss": 0.0358,
      "step": 4432
    },
    {
      "epoch": 0.6992940805300312,
      "grad_norm": 4.274868965148926,
      "learning_rate": 3.0687489937208175e-07,
      "loss": 0.0381,
      "step": 4433
    },
    {
      "epoch": 0.699451827897622,
      "grad_norm": 4.528099536895752,
      "learning_rate": 3.067138947029464e-07,
      "loss": 0.0535,
      "step": 4434
    },
    {
      "epoch": 0.6996095752652127,
      "grad_norm": 1.9978628158569336,
      "learning_rate": 3.0655289003381096e-07,
      "loss": 0.01,
      "step": 4435
    },
    {
      "epoch": 0.6997673226328036,
      "grad_norm": 5.906517505645752,
      "learning_rate": 3.0639188536467557e-07,
      "loss": 0.059,
      "step": 4436
    },
    {
      "epoch": 0.6999250700003944,
      "grad_norm": 3.6237752437591553,
      "learning_rate": 3.062308806955401e-07,
      "loss": 0.0171,
      "step": 4437
    },
    {
      "epoch": 0.7000828173679852,
      "grad_norm": 5.734447002410889,
      "learning_rate": 3.060698760264048e-07,
      "loss": 0.0327,
      "step": 4438
    },
    {
      "epoch": 0.7002405647355759,
      "grad_norm": 9.282461166381836,
      "learning_rate": 3.0590887135726934e-07,
      "loss": 0.0311,
      "step": 4439
    },
    {
      "epoch": 0.7003983121031668,
      "grad_norm": 5.420535087585449,
      "learning_rate": 3.0574786668813394e-07,
      "loss": 0.0411,
      "step": 4440
    },
    {
      "epoch": 0.7005560594707576,
      "grad_norm": 2.7260584831237793,
      "learning_rate": 3.0558686201899855e-07,
      "loss": 0.0234,
      "step": 4441
    },
    {
      "epoch": 0.7007138068383484,
      "grad_norm": 6.108238697052002,
      "learning_rate": 3.0542585734986316e-07,
      "loss": 0.0253,
      "step": 4442
    },
    {
      "epoch": 0.7008715542059392,
      "grad_norm": 8.544012069702148,
      "learning_rate": 3.052648526807277e-07,
      "loss": 0.0689,
      "step": 4443
    },
    {
      "epoch": 0.7010293015735299,
      "grad_norm": 5.6943359375,
      "learning_rate": 3.051038480115923e-07,
      "loss": 0.0669,
      "step": 4444
    },
    {
      "epoch": 0.7011870489411208,
      "grad_norm": 4.468897819519043,
      "learning_rate": 3.049428433424569e-07,
      "loss": 0.0433,
      "step": 4445
    },
    {
      "epoch": 0.7013447963087116,
      "grad_norm": 3.731135129928589,
      "learning_rate": 3.0478183867332153e-07,
      "loss": 0.0478,
      "step": 4446
    },
    {
      "epoch": 0.7015025436763024,
      "grad_norm": 8.033485412597656,
      "learning_rate": 3.046208340041861e-07,
      "loss": 0.0291,
      "step": 4447
    },
    {
      "epoch": 0.7016602910438932,
      "grad_norm": 2.8984217643737793,
      "learning_rate": 3.044598293350507e-07,
      "loss": 0.0151,
      "step": 4448
    },
    {
      "epoch": 0.701818038411484,
      "grad_norm": 3.4804484844207764,
      "learning_rate": 3.042988246659153e-07,
      "loss": 0.0242,
      "step": 4449
    },
    {
      "epoch": 0.7019757857790748,
      "grad_norm": 4.877875328063965,
      "learning_rate": 3.041378199967799e-07,
      "loss": 0.0262,
      "step": 4450
    },
    {
      "epoch": 0.7021335331466656,
      "grad_norm": 4.9602532386779785,
      "learning_rate": 3.0397681532764445e-07,
      "loss": 0.0668,
      "step": 4451
    },
    {
      "epoch": 0.7022912805142564,
      "grad_norm": 4.77786111831665,
      "learning_rate": 3.0381581065850906e-07,
      "loss": 0.0644,
      "step": 4452
    },
    {
      "epoch": 0.7024490278818473,
      "grad_norm": 4.0027546882629395,
      "learning_rate": 3.0365480598937367e-07,
      "loss": 0.0445,
      "step": 4453
    },
    {
      "epoch": 0.702606775249438,
      "grad_norm": 3.758492946624756,
      "learning_rate": 3.0349380132023827e-07,
      "loss": 0.0331,
      "step": 4454
    },
    {
      "epoch": 0.7027645226170288,
      "grad_norm": 2.2905874252319336,
      "learning_rate": 3.033327966511029e-07,
      "loss": 0.0215,
      "step": 4455
    },
    {
      "epoch": 0.7029222699846196,
      "grad_norm": 3.43929386138916,
      "learning_rate": 3.031717919819675e-07,
      "loss": 0.025,
      "step": 4456
    },
    {
      "epoch": 0.7030800173522105,
      "grad_norm": 2.7031850814819336,
      "learning_rate": 3.030107873128321e-07,
      "loss": 0.0158,
      "step": 4457
    },
    {
      "epoch": 0.7032377647198013,
      "grad_norm": 4.073042869567871,
      "learning_rate": 3.0284978264369665e-07,
      "loss": 0.0372,
      "step": 4458
    },
    {
      "epoch": 0.703395512087392,
      "grad_norm": 5.968804836273193,
      "learning_rate": 3.0268877797456125e-07,
      "loss": 0.041,
      "step": 4459
    },
    {
      "epoch": 0.7035532594549828,
      "grad_norm": 4.01667594909668,
      "learning_rate": 3.0252777330542586e-07,
      "loss": 0.0682,
      "step": 4460
    },
    {
      "epoch": 0.7037110068225737,
      "grad_norm": 3.069646120071411,
      "learning_rate": 3.0236676863629046e-07,
      "loss": 0.024,
      "step": 4461
    },
    {
      "epoch": 0.7038687541901645,
      "grad_norm": 2.835047960281372,
      "learning_rate": 3.02205763967155e-07,
      "loss": 0.0191,
      "step": 4462
    },
    {
      "epoch": 0.7040265015577553,
      "grad_norm": 3.3007359504699707,
      "learning_rate": 3.020447592980196e-07,
      "loss": 0.0283,
      "step": 4463
    },
    {
      "epoch": 0.704184248925346,
      "grad_norm": 4.909847259521484,
      "learning_rate": 3.0188375462888423e-07,
      "loss": 0.0513,
      "step": 4464
    },
    {
      "epoch": 0.7043419962929368,
      "grad_norm": 4.200301170349121,
      "learning_rate": 3.0172274995974884e-07,
      "loss": 0.0357,
      "step": 4465
    },
    {
      "epoch": 0.7044997436605277,
      "grad_norm": 6.148838043212891,
      "learning_rate": 3.015617452906134e-07,
      "loss": 0.0334,
      "step": 4466
    },
    {
      "epoch": 0.7046574910281185,
      "grad_norm": 5.063936710357666,
      "learning_rate": 3.01400740621478e-07,
      "loss": 0.0635,
      "step": 4467
    },
    {
      "epoch": 0.7048152383957093,
      "grad_norm": 5.4391865730285645,
      "learning_rate": 3.012397359523426e-07,
      "loss": 0.0799,
      "step": 4468
    },
    {
      "epoch": 0.7049729857633,
      "grad_norm": 3.486837863922119,
      "learning_rate": 3.010787312832072e-07,
      "loss": 0.042,
      "step": 4469
    },
    {
      "epoch": 0.7051307331308909,
      "grad_norm": 5.314830780029297,
      "learning_rate": 3.0091772661407176e-07,
      "loss": 0.0422,
      "step": 4470
    },
    {
      "epoch": 0.7052884804984817,
      "grad_norm": 5.49745512008667,
      "learning_rate": 3.007567219449364e-07,
      "loss": 0.0641,
      "step": 4471
    },
    {
      "epoch": 0.7054462278660725,
      "grad_norm": 3.8079562187194824,
      "learning_rate": 3.00595717275801e-07,
      "loss": 0.0256,
      "step": 4472
    },
    {
      "epoch": 0.7056039752336632,
      "grad_norm": 8.788825988769531,
      "learning_rate": 3.004347126066656e-07,
      "loss": 0.0535,
      "step": 4473
    },
    {
      "epoch": 0.7057617226012541,
      "grad_norm": 5.572489261627197,
      "learning_rate": 3.0027370793753013e-07,
      "loss": 0.0549,
      "step": 4474
    },
    {
      "epoch": 0.7059194699688449,
      "grad_norm": 6.502420425415039,
      "learning_rate": 3.001127032683948e-07,
      "loss": 0.046,
      "step": 4475
    },
    {
      "epoch": 0.7060772173364357,
      "grad_norm": 4.948294162750244,
      "learning_rate": 2.9995169859925935e-07,
      "loss": 0.0298,
      "step": 4476
    },
    {
      "epoch": 0.7062349647040265,
      "grad_norm": 5.6751861572265625,
      "learning_rate": 2.9979069393012395e-07,
      "loss": 0.0739,
      "step": 4477
    },
    {
      "epoch": 0.7063927120716174,
      "grad_norm": 4.962527275085449,
      "learning_rate": 2.996296892609885e-07,
      "loss": 0.0486,
      "step": 4478
    },
    {
      "epoch": 0.7065504594392081,
      "grad_norm": 5.656927585601807,
      "learning_rate": 2.9946868459185317e-07,
      "loss": 0.0707,
      "step": 4479
    },
    {
      "epoch": 0.7067082068067989,
      "grad_norm": 13.633844375610352,
      "learning_rate": 2.9930767992271777e-07,
      "loss": 0.0937,
      "step": 4480
    },
    {
      "epoch": 0.7068659541743897,
      "grad_norm": 7.616853713989258,
      "learning_rate": 2.991466752535823e-07,
      "loss": 0.046,
      "step": 4481
    },
    {
      "epoch": 0.7070237015419805,
      "grad_norm": 1.5346670150756836,
      "learning_rate": 2.9898567058444693e-07,
      "loss": 0.0096,
      "step": 4482
    },
    {
      "epoch": 0.7071814489095714,
      "grad_norm": 8.293448448181152,
      "learning_rate": 2.9882466591531154e-07,
      "loss": 0.0729,
      "step": 4483
    },
    {
      "epoch": 0.7073391962771621,
      "grad_norm": 6.246132850646973,
      "learning_rate": 2.9866366124617614e-07,
      "loss": 0.0421,
      "step": 4484
    },
    {
      "epoch": 0.7074969436447529,
      "grad_norm": 5.36680269241333,
      "learning_rate": 2.985026565770407e-07,
      "loss": 0.1121,
      "step": 4485
    },
    {
      "epoch": 0.7076546910123437,
      "grad_norm": 4.480698108673096,
      "learning_rate": 2.9834165190790536e-07,
      "loss": 0.0389,
      "step": 4486
    },
    {
      "epoch": 0.7078124383799346,
      "grad_norm": 6.817065238952637,
      "learning_rate": 2.981806472387699e-07,
      "loss": 0.0429,
      "step": 4487
    },
    {
      "epoch": 0.7079701857475253,
      "grad_norm": 4.782590389251709,
      "learning_rate": 2.980196425696345e-07,
      "loss": 0.0253,
      "step": 4488
    },
    {
      "epoch": 0.7081279331151161,
      "grad_norm": 3.480205774307251,
      "learning_rate": 2.9785863790049907e-07,
      "loss": 0.0375,
      "step": 4489
    },
    {
      "epoch": 0.7082856804827069,
      "grad_norm": 2.1709165573120117,
      "learning_rate": 2.9769763323136373e-07,
      "loss": 0.0093,
      "step": 4490
    },
    {
      "epoch": 0.7084434278502978,
      "grad_norm": 3.2164363861083984,
      "learning_rate": 2.975366285622283e-07,
      "loss": 0.0374,
      "step": 4491
    },
    {
      "epoch": 0.7086011752178886,
      "grad_norm": 4.7639241218566895,
      "learning_rate": 2.973756238930929e-07,
      "loss": 0.0433,
      "step": 4492
    },
    {
      "epoch": 0.7087589225854793,
      "grad_norm": 8.232230186462402,
      "learning_rate": 2.9721461922395744e-07,
      "loss": 0.0485,
      "step": 4493
    },
    {
      "epoch": 0.7089166699530701,
      "grad_norm": 6.158698558807373,
      "learning_rate": 2.970536145548221e-07,
      "loss": 0.0252,
      "step": 4494
    },
    {
      "epoch": 0.709074417320661,
      "grad_norm": 6.094226360321045,
      "learning_rate": 2.9689260988568666e-07,
      "loss": 0.0234,
      "step": 4495
    },
    {
      "epoch": 0.7092321646882518,
      "grad_norm": 3.0969483852386475,
      "learning_rate": 2.9673160521655126e-07,
      "loss": 0.0301,
      "step": 4496
    },
    {
      "epoch": 0.7093899120558426,
      "grad_norm": 5.906505584716797,
      "learning_rate": 2.965706005474158e-07,
      "loss": 0.0379,
      "step": 4497
    },
    {
      "epoch": 0.7095476594234333,
      "grad_norm": 3.609701633453369,
      "learning_rate": 2.964095958782805e-07,
      "loss": 0.064,
      "step": 4498
    },
    {
      "epoch": 0.7097054067910242,
      "grad_norm": 4.088521480560303,
      "learning_rate": 2.9624859120914503e-07,
      "loss": 0.0166,
      "step": 4499
    },
    {
      "epoch": 0.709863154158615,
      "grad_norm": 4.423680305480957,
      "learning_rate": 2.9608758654000963e-07,
      "loss": 0.044,
      "step": 4500
    },
    {
      "epoch": 0.7100209015262058,
      "grad_norm": 4.046323776245117,
      "learning_rate": 2.9592658187087424e-07,
      "loss": 0.0407,
      "step": 4501
    },
    {
      "epoch": 0.7101786488937966,
      "grad_norm": 4.431769847869873,
      "learning_rate": 2.9576557720173885e-07,
      "loss": 0.0502,
      "step": 4502
    },
    {
      "epoch": 0.7103363962613873,
      "grad_norm": 7.08558464050293,
      "learning_rate": 2.9560457253260345e-07,
      "loss": 0.0451,
      "step": 4503
    },
    {
      "epoch": 0.7104941436289782,
      "grad_norm": 8.421886444091797,
      "learning_rate": 2.95443567863468e-07,
      "loss": 0.0634,
      "step": 4504
    },
    {
      "epoch": 0.710651890996569,
      "grad_norm": 6.394400119781494,
      "learning_rate": 2.9528256319433267e-07,
      "loss": 0.0411,
      "step": 4505
    },
    {
      "epoch": 0.7108096383641598,
      "grad_norm": 3.922478437423706,
      "learning_rate": 2.951215585251972e-07,
      "loss": 0.0279,
      "step": 4506
    },
    {
      "epoch": 0.7109673857317506,
      "grad_norm": 5.17605447769165,
      "learning_rate": 2.949605538560618e-07,
      "loss": 0.0348,
      "step": 4507
    },
    {
      "epoch": 0.7111251330993414,
      "grad_norm": 2.986701726913452,
      "learning_rate": 2.947995491869264e-07,
      "loss": 0.0251,
      "step": 4508
    },
    {
      "epoch": 0.7112828804669322,
      "grad_norm": 8.349322319030762,
      "learning_rate": 2.9463854451779104e-07,
      "loss": 0.0598,
      "step": 4509
    },
    {
      "epoch": 0.711440627834523,
      "grad_norm": 6.769835948944092,
      "learning_rate": 2.944775398486556e-07,
      "loss": 0.0446,
      "step": 4510
    },
    {
      "epoch": 0.7115983752021138,
      "grad_norm": 2.4589574337005615,
      "learning_rate": 2.943165351795202e-07,
      "loss": 0.0303,
      "step": 4511
    },
    {
      "epoch": 0.7117561225697047,
      "grad_norm": 4.1198506355285645,
      "learning_rate": 2.9415553051038475e-07,
      "loss": 0.021,
      "step": 4512
    },
    {
      "epoch": 0.7119138699372954,
      "grad_norm": 6.685397148132324,
      "learning_rate": 2.939945258412494e-07,
      "loss": 0.0374,
      "step": 4513
    },
    {
      "epoch": 0.7120716173048862,
      "grad_norm": 6.449472904205322,
      "learning_rate": 2.9383352117211396e-07,
      "loss": 0.0613,
      "step": 4514
    },
    {
      "epoch": 0.712229364672477,
      "grad_norm": 4.5486369132995605,
      "learning_rate": 2.9367251650297857e-07,
      "loss": 0.0372,
      "step": 4515
    },
    {
      "epoch": 0.7123871120400679,
      "grad_norm": 3.8935794830322266,
      "learning_rate": 2.935115118338432e-07,
      "loss": 0.0285,
      "step": 4516
    },
    {
      "epoch": 0.7125448594076587,
      "grad_norm": 4.628503322601318,
      "learning_rate": 2.933505071647078e-07,
      "loss": 0.0472,
      "step": 4517
    },
    {
      "epoch": 0.7127026067752494,
      "grad_norm": 2.7762725353240967,
      "learning_rate": 2.9318950249557234e-07,
      "loss": 0.0279,
      "step": 4518
    },
    {
      "epoch": 0.7128603541428402,
      "grad_norm": 10.613434791564941,
      "learning_rate": 2.9302849782643694e-07,
      "loss": 0.0497,
      "step": 4519
    },
    {
      "epoch": 0.713018101510431,
      "grad_norm": 4.48843240737915,
      "learning_rate": 2.9286749315730155e-07,
      "loss": 0.0392,
      "step": 4520
    },
    {
      "epoch": 0.7131758488780219,
      "grad_norm": 3.927433967590332,
      "learning_rate": 2.9270648848816615e-07,
      "loss": 0.038,
      "step": 4521
    },
    {
      "epoch": 0.7133335962456127,
      "grad_norm": 9.800247192382812,
      "learning_rate": 2.925454838190307e-07,
      "loss": 0.0511,
      "step": 4522
    },
    {
      "epoch": 0.7134913436132034,
      "grad_norm": 7.325588703155518,
      "learning_rate": 2.923844791498953e-07,
      "loss": 0.0644,
      "step": 4523
    },
    {
      "epoch": 0.7136490909807942,
      "grad_norm": 4.694719314575195,
      "learning_rate": 2.922234744807599e-07,
      "loss": 0.0313,
      "step": 4524
    },
    {
      "epoch": 0.7138068383483851,
      "grad_norm": 3.6894118785858154,
      "learning_rate": 2.9206246981162453e-07,
      "loss": 0.0345,
      "step": 4525
    },
    {
      "epoch": 0.7139645857159759,
      "grad_norm": 4.581935882568359,
      "learning_rate": 2.9190146514248913e-07,
      "loss": 0.0372,
      "step": 4526
    },
    {
      "epoch": 0.7141223330835667,
      "grad_norm": 7.473629951477051,
      "learning_rate": 2.917404604733537e-07,
      "loss": 0.0286,
      "step": 4527
    },
    {
      "epoch": 0.7142800804511574,
      "grad_norm": 5.640164375305176,
      "learning_rate": 2.9157945580421835e-07,
      "loss": 0.0412,
      "step": 4528
    },
    {
      "epoch": 0.7144378278187483,
      "grad_norm": 6.2254838943481445,
      "learning_rate": 2.914184511350829e-07,
      "loss": 0.0532,
      "step": 4529
    },
    {
      "epoch": 0.7145955751863391,
      "grad_norm": 4.138742446899414,
      "learning_rate": 2.912574464659475e-07,
      "loss": 0.0364,
      "step": 4530
    },
    {
      "epoch": 0.7147533225539299,
      "grad_norm": 4.648903846740723,
      "learning_rate": 2.910964417968121e-07,
      "loss": 0.0359,
      "step": 4531
    },
    {
      "epoch": 0.7149110699215206,
      "grad_norm": 4.27373743057251,
      "learning_rate": 2.909354371276767e-07,
      "loss": 0.0402,
      "step": 4532
    },
    {
      "epoch": 0.7150688172891115,
      "grad_norm": 4.147173881530762,
      "learning_rate": 2.9077443245854127e-07,
      "loss": 0.035,
      "step": 4533
    },
    {
      "epoch": 0.7152265646567023,
      "grad_norm": 3.1072118282318115,
      "learning_rate": 2.906134277894059e-07,
      "loss": 0.0137,
      "step": 4534
    },
    {
      "epoch": 0.7153843120242931,
      "grad_norm": 11.774784088134766,
      "learning_rate": 2.904524231202705e-07,
      "loss": 0.1105,
      "step": 4535
    },
    {
      "epoch": 0.7155420593918839,
      "grad_norm": 4.649820804595947,
      "learning_rate": 2.902914184511351e-07,
      "loss": 0.0727,
      "step": 4536
    },
    {
      "epoch": 0.7156998067594748,
      "grad_norm": 7.751923561096191,
      "learning_rate": 2.9013041378199964e-07,
      "loss": 0.0485,
      "step": 4537
    },
    {
      "epoch": 0.7158575541270655,
      "grad_norm": 5.118620872497559,
      "learning_rate": 2.8996940911286425e-07,
      "loss": 0.0403,
      "step": 4538
    },
    {
      "epoch": 0.7160153014946563,
      "grad_norm": 7.4496846199035645,
      "learning_rate": 2.8980840444372886e-07,
      "loss": 0.0397,
      "step": 4539
    },
    {
      "epoch": 0.7161730488622471,
      "grad_norm": 7.6184587478637695,
      "learning_rate": 2.8964739977459346e-07,
      "loss": 0.0584,
      "step": 4540
    },
    {
      "epoch": 0.7163307962298379,
      "grad_norm": 3.5897741317749023,
      "learning_rate": 2.89486395105458e-07,
      "loss": 0.049,
      "step": 4541
    },
    {
      "epoch": 0.7164885435974288,
      "grad_norm": 5.4391913414001465,
      "learning_rate": 2.893253904363226e-07,
      "loss": 0.0216,
      "step": 4542
    },
    {
      "epoch": 0.7166462909650195,
      "grad_norm": 2.8665518760681152,
      "learning_rate": 2.8916438576718723e-07,
      "loss": 0.0303,
      "step": 4543
    },
    {
      "epoch": 0.7168040383326103,
      "grad_norm": 2.100839138031006,
      "learning_rate": 2.8900338109805184e-07,
      "loss": 0.0122,
      "step": 4544
    },
    {
      "epoch": 0.7169617857002011,
      "grad_norm": 7.321889400482178,
      "learning_rate": 2.888423764289164e-07,
      "loss": 0.0942,
      "step": 4545
    },
    {
      "epoch": 0.717119533067792,
      "grad_norm": 4.987734317779541,
      "learning_rate": 2.8868137175978105e-07,
      "loss": 0.0475,
      "step": 4546
    },
    {
      "epoch": 0.7172772804353827,
      "grad_norm": 6.623134613037109,
      "learning_rate": 2.885203670906456e-07,
      "loss": 0.0415,
      "step": 4547
    },
    {
      "epoch": 0.7174350278029735,
      "grad_norm": 5.465873718261719,
      "learning_rate": 2.883593624215102e-07,
      "loss": 0.085,
      "step": 4548
    },
    {
      "epoch": 0.7175927751705643,
      "grad_norm": 7.853731155395508,
      "learning_rate": 2.881983577523748e-07,
      "loss": 0.0748,
      "step": 4549
    },
    {
      "epoch": 0.7177505225381552,
      "grad_norm": 6.531794548034668,
      "learning_rate": 2.880373530832394e-07,
      "loss": 0.0394,
      "step": 4550
    },
    {
      "epoch": 0.717908269905746,
      "grad_norm": 6.5806732177734375,
      "learning_rate": 2.8787634841410403e-07,
      "loss": 0.0241,
      "step": 4551
    },
    {
      "epoch": 0.7180660172733367,
      "grad_norm": 6.493992328643799,
      "learning_rate": 2.877153437449686e-07,
      "loss": 0.0675,
      "step": 4552
    },
    {
      "epoch": 0.7182237646409275,
      "grad_norm": 1.3564178943634033,
      "learning_rate": 2.875543390758332e-07,
      "loss": 0.0079,
      "step": 4553
    },
    {
      "epoch": 0.7183815120085184,
      "grad_norm": 6.057300567626953,
      "learning_rate": 2.873933344066978e-07,
      "loss": 0.0408,
      "step": 4554
    },
    {
      "epoch": 0.7185392593761092,
      "grad_norm": 4.63604736328125,
      "learning_rate": 2.872323297375624e-07,
      "loss": 0.0221,
      "step": 4555
    },
    {
      "epoch": 0.7186970067437,
      "grad_norm": 8.910985946655273,
      "learning_rate": 2.8707132506842695e-07,
      "loss": 0.0456,
      "step": 4556
    },
    {
      "epoch": 0.7188547541112907,
      "grad_norm": 6.352619647979736,
      "learning_rate": 2.8691032039929156e-07,
      "loss": 0.056,
      "step": 4557
    },
    {
      "epoch": 0.7190125014788815,
      "grad_norm": 4.340364456176758,
      "learning_rate": 2.8674931573015616e-07,
      "loss": 0.04,
      "step": 4558
    },
    {
      "epoch": 0.7191702488464724,
      "grad_norm": 2.3495678901672363,
      "learning_rate": 2.8658831106102077e-07,
      "loss": 0.0139,
      "step": 4559
    },
    {
      "epoch": 0.7193279962140632,
      "grad_norm": 5.017070770263672,
      "learning_rate": 2.864273063918853e-07,
      "loss": 0.075,
      "step": 4560
    },
    {
      "epoch": 0.719485743581654,
      "grad_norm": 3.2221689224243164,
      "learning_rate": 2.8626630172275e-07,
      "loss": 0.043,
      "step": 4561
    },
    {
      "epoch": 0.7196434909492447,
      "grad_norm": 7.304450988769531,
      "learning_rate": 2.8610529705361454e-07,
      "loss": 0.0876,
      "step": 4562
    },
    {
      "epoch": 0.7198012383168356,
      "grad_norm": 3.5030410289764404,
      "learning_rate": 2.8594429238447914e-07,
      "loss": 0.0383,
      "step": 4563
    },
    {
      "epoch": 0.7199589856844264,
      "grad_norm": 3.965207815170288,
      "learning_rate": 2.857832877153437e-07,
      "loss": 0.032,
      "step": 4564
    },
    {
      "epoch": 0.7201167330520172,
      "grad_norm": 4.125488758087158,
      "learning_rate": 2.8562228304620836e-07,
      "loss": 0.0227,
      "step": 4565
    },
    {
      "epoch": 0.720274480419608,
      "grad_norm": 4.313589096069336,
      "learning_rate": 2.854612783770729e-07,
      "loss": 0.0622,
      "step": 4566
    },
    {
      "epoch": 0.7204322277871988,
      "grad_norm": 3.845998764038086,
      "learning_rate": 2.853002737079375e-07,
      "loss": 0.0717,
      "step": 4567
    },
    {
      "epoch": 0.7205899751547896,
      "grad_norm": 8.11777400970459,
      "learning_rate": 2.8513926903880207e-07,
      "loss": 0.0392,
      "step": 4568
    },
    {
      "epoch": 0.7207477225223804,
      "grad_norm": 3.3960371017456055,
      "learning_rate": 2.8497826436966673e-07,
      "loss": 0.0266,
      "step": 4569
    },
    {
      "epoch": 0.7209054698899712,
      "grad_norm": 10.87304973602295,
      "learning_rate": 2.848172597005313e-07,
      "loss": 0.0588,
      "step": 4570
    },
    {
      "epoch": 0.7210632172575621,
      "grad_norm": 5.718992710113525,
      "learning_rate": 2.846562550313959e-07,
      "loss": 0.0481,
      "step": 4571
    },
    {
      "epoch": 0.7212209646251528,
      "grad_norm": 6.245044708251953,
      "learning_rate": 2.844952503622605e-07,
      "loss": 0.0281,
      "step": 4572
    },
    {
      "epoch": 0.7213787119927436,
      "grad_norm": 2.650266647338867,
      "learning_rate": 2.843342456931251e-07,
      "loss": 0.0347,
      "step": 4573
    },
    {
      "epoch": 0.7215364593603344,
      "grad_norm": 4.317817211151123,
      "learning_rate": 2.841732410239897e-07,
      "loss": 0.0436,
      "step": 4574
    },
    {
      "epoch": 0.7216942067279253,
      "grad_norm": 7.154784679412842,
      "learning_rate": 2.8401223635485426e-07,
      "loss": 0.0695,
      "step": 4575
    },
    {
      "epoch": 0.7218519540955161,
      "grad_norm": 6.304006576538086,
      "learning_rate": 2.838512316857189e-07,
      "loss": 0.0525,
      "step": 4576
    },
    {
      "epoch": 0.7220097014631068,
      "grad_norm": 3.2809009552001953,
      "learning_rate": 2.8369022701658347e-07,
      "loss": 0.0288,
      "step": 4577
    },
    {
      "epoch": 0.7221674488306976,
      "grad_norm": 6.112624645233154,
      "learning_rate": 2.835292223474481e-07,
      "loss": 0.0714,
      "step": 4578
    },
    {
      "epoch": 0.7223251961982884,
      "grad_norm": 11.613422393798828,
      "learning_rate": 2.8336821767831263e-07,
      "loss": 0.0675,
      "step": 4579
    },
    {
      "epoch": 0.7224829435658793,
      "grad_norm": 9.109901428222656,
      "learning_rate": 2.832072130091773e-07,
      "loss": 0.0615,
      "step": 4580
    },
    {
      "epoch": 0.72264069093347,
      "grad_norm": 6.590017318725586,
      "learning_rate": 2.8304620834004185e-07,
      "loss": 0.032,
      "step": 4581
    },
    {
      "epoch": 0.7227984383010608,
      "grad_norm": 4.4322404861450195,
      "learning_rate": 2.8288520367090645e-07,
      "loss": 0.073,
      "step": 4582
    },
    {
      "epoch": 0.7229561856686516,
      "grad_norm": 7.764130115509033,
      "learning_rate": 2.82724199001771e-07,
      "loss": 0.0585,
      "step": 4583
    },
    {
      "epoch": 0.7231139330362425,
      "grad_norm": 5.1639790534973145,
      "learning_rate": 2.8256319433263566e-07,
      "loss": 0.0843,
      "step": 4584
    },
    {
      "epoch": 0.7232716804038333,
      "grad_norm": 4.606882095336914,
      "learning_rate": 2.824021896635002e-07,
      "loss": 0.0319,
      "step": 4585
    },
    {
      "epoch": 0.723429427771424,
      "grad_norm": 6.314964294433594,
      "learning_rate": 2.822411849943648e-07,
      "loss": 0.0854,
      "step": 4586
    },
    {
      "epoch": 0.7235871751390148,
      "grad_norm": 4.091706275939941,
      "learning_rate": 2.820801803252294e-07,
      "loss": 0.043,
      "step": 4587
    },
    {
      "epoch": 0.7237449225066057,
      "grad_norm": 3.6330947875976562,
      "learning_rate": 2.8191917565609404e-07,
      "loss": 0.0287,
      "step": 4588
    },
    {
      "epoch": 0.7239026698741965,
      "grad_norm": 7.67761754989624,
      "learning_rate": 2.817581709869586e-07,
      "loss": 0.0151,
      "step": 4589
    },
    {
      "epoch": 0.7240604172417873,
      "grad_norm": 5.9519548416137695,
      "learning_rate": 2.815971663178232e-07,
      "loss": 0.0776,
      "step": 4590
    },
    {
      "epoch": 0.724218164609378,
      "grad_norm": 4.592087268829346,
      "learning_rate": 2.8143616164868775e-07,
      "loss": 0.0279,
      "step": 4591
    },
    {
      "epoch": 0.7243759119769689,
      "grad_norm": 4.141416072845459,
      "learning_rate": 2.812751569795524e-07,
      "loss": 0.0213,
      "step": 4592
    },
    {
      "epoch": 0.7245336593445597,
      "grad_norm": 5.703773498535156,
      "learning_rate": 2.8111415231041696e-07,
      "loss": 0.046,
      "step": 4593
    },
    {
      "epoch": 0.7246914067121505,
      "grad_norm": 4.934125900268555,
      "learning_rate": 2.8095314764128157e-07,
      "loss": 0.0524,
      "step": 4594
    },
    {
      "epoch": 0.7248491540797413,
      "grad_norm": 2.652515411376953,
      "learning_rate": 2.8079214297214623e-07,
      "loss": 0.0161,
      "step": 4595
    },
    {
      "epoch": 0.7250069014473322,
      "grad_norm": 4.344799518585205,
      "learning_rate": 2.806311383030108e-07,
      "loss": 0.0194,
      "step": 4596
    },
    {
      "epoch": 0.7251646488149229,
      "grad_norm": 5.285390377044678,
      "learning_rate": 2.804701336338754e-07,
      "loss": 0.0399,
      "step": 4597
    },
    {
      "epoch": 0.7253223961825137,
      "grad_norm": 4.879339694976807,
      "learning_rate": 2.8030912896473994e-07,
      "loss": 0.037,
      "step": 4598
    },
    {
      "epoch": 0.7254801435501045,
      "grad_norm": 9.88895034790039,
      "learning_rate": 2.801481242956046e-07,
      "loss": 0.044,
      "step": 4599
    },
    {
      "epoch": 0.7256378909176953,
      "grad_norm": 1.890272617340088,
      "learning_rate": 2.7998711962646915e-07,
      "loss": 0.0226,
      "step": 4600
    },
    {
      "epoch": 0.7257956382852861,
      "grad_norm": 4.78530740737915,
      "learning_rate": 2.7982611495733376e-07,
      "loss": 0.0408,
      "step": 4601
    },
    {
      "epoch": 0.7259533856528769,
      "grad_norm": 2.758941173553467,
      "learning_rate": 2.796651102881983e-07,
      "loss": 0.0129,
      "step": 4602
    },
    {
      "epoch": 0.7261111330204677,
      "grad_norm": 5.772910118103027,
      "learning_rate": 2.7950410561906297e-07,
      "loss": 0.0544,
      "step": 4603
    },
    {
      "epoch": 0.7262688803880585,
      "grad_norm": 4.390839099884033,
      "learning_rate": 2.793431009499275e-07,
      "loss": 0.055,
      "step": 4604
    },
    {
      "epoch": 0.7264266277556494,
      "grad_norm": 3.2692339420318604,
      "learning_rate": 2.7918209628079213e-07,
      "loss": 0.0265,
      "step": 4605
    },
    {
      "epoch": 0.7265843751232401,
      "grad_norm": 3.9459526538848877,
      "learning_rate": 2.790210916116567e-07,
      "loss": 0.0135,
      "step": 4606
    },
    {
      "epoch": 0.7267421224908309,
      "grad_norm": 3.0343422889709473,
      "learning_rate": 2.7886008694252134e-07,
      "loss": 0.0212,
      "step": 4607
    },
    {
      "epoch": 0.7268998698584217,
      "grad_norm": 4.5064520835876465,
      "learning_rate": 2.786990822733859e-07,
      "loss": 0.0312,
      "step": 4608
    },
    {
      "epoch": 0.7268998698584217,
      "eval_accuracy": 0.9879914224446033,
      "eval_f1": 0.9879914224446033,
      "eval_loss": 0.039093952625989914,
      "eval_runtime": 4699.0523,
      "eval_samples_per_second": 43.169,
      "eval_steps_per_second": 2.698,
      "step": 4608
    },
    {
      "epoch": 0.7270576172260126,
      "grad_norm": 3.6038103103637695,
      "learning_rate": 2.785380776042505e-07,
      "loss": 0.0276,
      "step": 4609
    },
    {
      "epoch": 0.7272153645936034,
      "grad_norm": 1.399121880531311,
      "learning_rate": 2.783770729351151e-07,
      "loss": 0.0091,
      "step": 4610
    },
    {
      "epoch": 0.7273731119611941,
      "grad_norm": 3.0181891918182373,
      "learning_rate": 2.782160682659797e-07,
      "loss": 0.0277,
      "step": 4611
    },
    {
      "epoch": 0.7275308593287849,
      "grad_norm": 4.86851692199707,
      "learning_rate": 2.7805506359684427e-07,
      "loss": 0.0495,
      "step": 4612
    },
    {
      "epoch": 0.7276886066963758,
      "grad_norm": 6.538103103637695,
      "learning_rate": 2.778940589277089e-07,
      "loss": 0.0689,
      "step": 4613
    },
    {
      "epoch": 0.7278463540639666,
      "grad_norm": 4.0547895431518555,
      "learning_rate": 2.777330542585735e-07,
      "loss": 0.0293,
      "step": 4614
    },
    {
      "epoch": 0.7280041014315574,
      "grad_norm": 3.82523250579834,
      "learning_rate": 2.775720495894381e-07,
      "loss": 0.0363,
      "step": 4615
    },
    {
      "epoch": 0.7281618487991481,
      "grad_norm": 3.744596242904663,
      "learning_rate": 2.7741104492030264e-07,
      "loss": 0.0347,
      "step": 4616
    },
    {
      "epoch": 0.7283195961667389,
      "grad_norm": 7.278544902801514,
      "learning_rate": 2.7725004025116725e-07,
      "loss": 0.1236,
      "step": 4617
    },
    {
      "epoch": 0.7284773435343298,
      "grad_norm": 8.369022369384766,
      "learning_rate": 2.770890355820319e-07,
      "loss": 0.0403,
      "step": 4618
    },
    {
      "epoch": 0.7286350909019206,
      "grad_norm": 6.324163913726807,
      "learning_rate": 2.7692803091289646e-07,
      "loss": 0.0488,
      "step": 4619
    },
    {
      "epoch": 0.7287928382695114,
      "grad_norm": 3.8759469985961914,
      "learning_rate": 2.7676702624376107e-07,
      "loss": 0.0257,
      "step": 4620
    },
    {
      "epoch": 0.7289505856371021,
      "grad_norm": 4.767875671386719,
      "learning_rate": 2.766060215746256e-07,
      "loss": 0.0265,
      "step": 4621
    },
    {
      "epoch": 0.729108333004693,
      "grad_norm": 5.644853115081787,
      "learning_rate": 2.764450169054903e-07,
      "loss": 0.0562,
      "step": 4622
    },
    {
      "epoch": 0.7292660803722838,
      "grad_norm": 6.3105597496032715,
      "learning_rate": 2.7628401223635483e-07,
      "loss": 0.072,
      "step": 4623
    },
    {
      "epoch": 0.7294238277398746,
      "grad_norm": 6.4134297370910645,
      "learning_rate": 2.7612300756721944e-07,
      "loss": 0.0305,
      "step": 4624
    },
    {
      "epoch": 0.7295815751074654,
      "grad_norm": 3.7911131381988525,
      "learning_rate": 2.7596200289808405e-07,
      "loss": 0.0552,
      "step": 4625
    },
    {
      "epoch": 0.7297393224750562,
      "grad_norm": 2.7404658794403076,
      "learning_rate": 2.7580099822894865e-07,
      "loss": 0.0138,
      "step": 4626
    },
    {
      "epoch": 0.729897069842647,
      "grad_norm": 4.408675670623779,
      "learning_rate": 2.756399935598132e-07,
      "loss": 0.0398,
      "step": 4627
    },
    {
      "epoch": 0.7300548172102378,
      "grad_norm": 2.576265811920166,
      "learning_rate": 2.754789888906778e-07,
      "loss": 0.0088,
      "step": 4628
    },
    {
      "epoch": 0.7302125645778286,
      "grad_norm": 4.607842922210693,
      "learning_rate": 2.753179842215424e-07,
      "loss": 0.039,
      "step": 4629
    },
    {
      "epoch": 0.7303703119454195,
      "grad_norm": 6.559534549713135,
      "learning_rate": 2.75156979552407e-07,
      "loss": 0.1064,
      "step": 4630
    },
    {
      "epoch": 0.7305280593130102,
      "grad_norm": 5.801240921020508,
      "learning_rate": 2.749959748832716e-07,
      "loss": 0.0354,
      "step": 4631
    },
    {
      "epoch": 0.730685806680601,
      "grad_norm": 5.777365684509277,
      "learning_rate": 2.748349702141362e-07,
      "loss": 0.046,
      "step": 4632
    },
    {
      "epoch": 0.7308435540481918,
      "grad_norm": 7.259184837341309,
      "learning_rate": 2.746739655450008e-07,
      "loss": 0.0827,
      "step": 4633
    },
    {
      "epoch": 0.7310013014157827,
      "grad_norm": 7.7218918800354,
      "learning_rate": 2.745129608758654e-07,
      "loss": 0.0804,
      "step": 4634
    },
    {
      "epoch": 0.7311590487833735,
      "grad_norm": 7.663471221923828,
      "learning_rate": 2.7435195620672995e-07,
      "loss": 0.0958,
      "step": 4635
    },
    {
      "epoch": 0.7313167961509642,
      "grad_norm": 4.536482810974121,
      "learning_rate": 2.7419095153759456e-07,
      "loss": 0.0327,
      "step": 4636
    },
    {
      "epoch": 0.731474543518555,
      "grad_norm": 5.311334609985352,
      "learning_rate": 2.7402994686845916e-07,
      "loss": 0.0725,
      "step": 4637
    },
    {
      "epoch": 0.7316322908861458,
      "grad_norm": 7.605411052703857,
      "learning_rate": 2.7386894219932377e-07,
      "loss": 0.0739,
      "step": 4638
    },
    {
      "epoch": 0.7317900382537367,
      "grad_norm": 6.093417167663574,
      "learning_rate": 2.737079375301883e-07,
      "loss": 0.0232,
      "step": 4639
    },
    {
      "epoch": 0.7319477856213275,
      "grad_norm": 5.984975337982178,
      "learning_rate": 2.73546932861053e-07,
      "loss": 0.0578,
      "step": 4640
    },
    {
      "epoch": 0.7321055329889182,
      "grad_norm": 5.4426751136779785,
      "learning_rate": 2.733859281919176e-07,
      "loss": 0.0651,
      "step": 4641
    },
    {
      "epoch": 0.732263280356509,
      "grad_norm": 6.997135639190674,
      "learning_rate": 2.7322492352278214e-07,
      "loss": 0.0658,
      "step": 4642
    },
    {
      "epoch": 0.7324210277240999,
      "grad_norm": 4.691622734069824,
      "learning_rate": 2.7306391885364675e-07,
      "loss": 0.0178,
      "step": 4643
    },
    {
      "epoch": 0.7325787750916907,
      "grad_norm": 5.306899547576904,
      "learning_rate": 2.7290291418451136e-07,
      "loss": 0.0451,
      "step": 4644
    },
    {
      "epoch": 0.7327365224592814,
      "grad_norm": 6.233757495880127,
      "learning_rate": 2.7274190951537596e-07,
      "loss": 0.0713,
      "step": 4645
    },
    {
      "epoch": 0.7328942698268722,
      "grad_norm": 6.5986456871032715,
      "learning_rate": 2.725809048462405e-07,
      "loss": 0.0324,
      "step": 4646
    },
    {
      "epoch": 0.7330520171944631,
      "grad_norm": 5.2762837409973145,
      "learning_rate": 2.724199001771051e-07,
      "loss": 0.0331,
      "step": 4647
    },
    {
      "epoch": 0.7332097645620539,
      "grad_norm": 10.506940841674805,
      "learning_rate": 2.7225889550796973e-07,
      "loss": 0.0518,
      "step": 4648
    },
    {
      "epoch": 0.7333675119296447,
      "grad_norm": 2.849714756011963,
      "learning_rate": 2.7209789083883433e-07,
      "loss": 0.0188,
      "step": 4649
    },
    {
      "epoch": 0.7335252592972354,
      "grad_norm": 7.017124652862549,
      "learning_rate": 2.719368861696989e-07,
      "loss": 0.0746,
      "step": 4650
    },
    {
      "epoch": 0.7336830066648263,
      "grad_norm": 6.85901403427124,
      "learning_rate": 2.717758815005635e-07,
      "loss": 0.0417,
      "step": 4651
    },
    {
      "epoch": 0.7338407540324171,
      "grad_norm": 6.239452838897705,
      "learning_rate": 2.716148768314281e-07,
      "loss": 0.0657,
      "step": 4652
    },
    {
      "epoch": 0.7339985014000079,
      "grad_norm": 8.635309219360352,
      "learning_rate": 2.714538721622927e-07,
      "loss": 0.0552,
      "step": 4653
    },
    {
      "epoch": 0.7341562487675987,
      "grad_norm": 3.114839553833008,
      "learning_rate": 2.7129286749315726e-07,
      "loss": 0.0377,
      "step": 4654
    },
    {
      "epoch": 0.7343139961351894,
      "grad_norm": 5.728065013885498,
      "learning_rate": 2.711318628240219e-07,
      "loss": 0.0805,
      "step": 4655
    },
    {
      "epoch": 0.7344717435027803,
      "grad_norm": 3.9795284271240234,
      "learning_rate": 2.7097085815488647e-07,
      "loss": 0.0591,
      "step": 4656
    },
    {
      "epoch": 0.7346294908703711,
      "grad_norm": 4.951369762420654,
      "learning_rate": 2.708098534857511e-07,
      "loss": 0.0399,
      "step": 4657
    },
    {
      "epoch": 0.7347872382379619,
      "grad_norm": 6.942314624786377,
      "learning_rate": 2.7064884881661563e-07,
      "loss": 0.0601,
      "step": 4658
    },
    {
      "epoch": 0.7349449856055527,
      "grad_norm": 2.2104973793029785,
      "learning_rate": 2.704878441474803e-07,
      "loss": 0.0175,
      "step": 4659
    },
    {
      "epoch": 0.7351027329731435,
      "grad_norm": 2.437842607498169,
      "learning_rate": 2.7032683947834484e-07,
      "loss": 0.0091,
      "step": 4660
    },
    {
      "epoch": 0.7352604803407343,
      "grad_norm": 4.740623950958252,
      "learning_rate": 2.7016583480920945e-07,
      "loss": 0.0325,
      "step": 4661
    },
    {
      "epoch": 0.7354182277083251,
      "grad_norm": 6.783057689666748,
      "learning_rate": 2.70004830140074e-07,
      "loss": 0.0595,
      "step": 4662
    },
    {
      "epoch": 0.7355759750759159,
      "grad_norm": 3.6913976669311523,
      "learning_rate": 2.6984382547093866e-07,
      "loss": 0.0233,
      "step": 4663
    },
    {
      "epoch": 0.7357337224435068,
      "grad_norm": 4.711301326751709,
      "learning_rate": 2.696828208018032e-07,
      "loss": 0.0345,
      "step": 4664
    },
    {
      "epoch": 0.7358914698110975,
      "grad_norm": 6.14579439163208,
      "learning_rate": 2.695218161326678e-07,
      "loss": 0.0343,
      "step": 4665
    },
    {
      "epoch": 0.7360492171786883,
      "grad_norm": 7.727104187011719,
      "learning_rate": 2.6936081146353243e-07,
      "loss": 0.1267,
      "step": 4666
    },
    {
      "epoch": 0.7362069645462791,
      "grad_norm": 3.1093430519104004,
      "learning_rate": 2.6919980679439704e-07,
      "loss": 0.0222,
      "step": 4667
    },
    {
      "epoch": 0.73636471191387,
      "grad_norm": 2.481738805770874,
      "learning_rate": 2.6903880212526164e-07,
      "loss": 0.0277,
      "step": 4668
    },
    {
      "epoch": 0.7365224592814608,
      "grad_norm": 7.11629056930542,
      "learning_rate": 2.688777974561262e-07,
      "loss": 0.0476,
      "step": 4669
    },
    {
      "epoch": 0.7366802066490515,
      "grad_norm": 9.372925758361816,
      "learning_rate": 2.6871679278699085e-07,
      "loss": 0.0442,
      "step": 4670
    },
    {
      "epoch": 0.7368379540166423,
      "grad_norm": 6.089367389678955,
      "learning_rate": 2.685557881178554e-07,
      "loss": 0.0311,
      "step": 4671
    },
    {
      "epoch": 0.7369957013842332,
      "grad_norm": 3.7933833599090576,
      "learning_rate": 2.6839478344872e-07,
      "loss": 0.0246,
      "step": 4672
    },
    {
      "epoch": 0.737153448751824,
      "grad_norm": 5.042723655700684,
      "learning_rate": 2.6823377877958457e-07,
      "loss": 0.0699,
      "step": 4673
    },
    {
      "epoch": 0.7373111961194148,
      "grad_norm": 3.5024068355560303,
      "learning_rate": 2.6807277411044923e-07,
      "loss": 0.0311,
      "step": 4674
    },
    {
      "epoch": 0.7374689434870055,
      "grad_norm": 4.534403324127197,
      "learning_rate": 2.679117694413138e-07,
      "loss": 0.0586,
      "step": 4675
    },
    {
      "epoch": 0.7376266908545963,
      "grad_norm": 4.200473308563232,
      "learning_rate": 2.677507647721784e-07,
      "loss": 0.0246,
      "step": 4676
    },
    {
      "epoch": 0.7377844382221872,
      "grad_norm": 4.237485885620117,
      "learning_rate": 2.6758976010304294e-07,
      "loss": 0.0132,
      "step": 4677
    },
    {
      "epoch": 0.737942185589778,
      "grad_norm": 3.843857526779175,
      "learning_rate": 2.674287554339076e-07,
      "loss": 0.0278,
      "step": 4678
    },
    {
      "epoch": 0.7380999329573688,
      "grad_norm": 5.551311492919922,
      "learning_rate": 2.6726775076477215e-07,
      "loss": 0.0603,
      "step": 4679
    },
    {
      "epoch": 0.7382576803249595,
      "grad_norm": 7.5500030517578125,
      "learning_rate": 2.6710674609563676e-07,
      "loss": 0.0653,
      "step": 4680
    },
    {
      "epoch": 0.7384154276925504,
      "grad_norm": 2.9558167457580566,
      "learning_rate": 2.669457414265013e-07,
      "loss": 0.0334,
      "step": 4681
    },
    {
      "epoch": 0.7385731750601412,
      "grad_norm": 4.379537105560303,
      "learning_rate": 2.6678473675736597e-07,
      "loss": 0.035,
      "step": 4682
    },
    {
      "epoch": 0.738730922427732,
      "grad_norm": 7.399539947509766,
      "learning_rate": 2.666237320882305e-07,
      "loss": 0.0568,
      "step": 4683
    },
    {
      "epoch": 0.7388886697953228,
      "grad_norm": 6.342048645019531,
      "learning_rate": 2.6646272741909513e-07,
      "loss": 0.0672,
      "step": 4684
    },
    {
      "epoch": 0.7390464171629136,
      "grad_norm": 5.140990257263184,
      "learning_rate": 2.6630172274995974e-07,
      "loss": 0.0545,
      "step": 4685
    },
    {
      "epoch": 0.7392041645305044,
      "grad_norm": 4.986282825469971,
      "learning_rate": 2.6614071808082434e-07,
      "loss": 0.0561,
      "step": 4686
    },
    {
      "epoch": 0.7393619118980952,
      "grad_norm": 5.449554443359375,
      "learning_rate": 2.659797134116889e-07,
      "loss": 0.0331,
      "step": 4687
    },
    {
      "epoch": 0.739519659265686,
      "grad_norm": 1.2541812658309937,
      "learning_rate": 2.658187087425535e-07,
      "loss": 0.0108,
      "step": 4688
    },
    {
      "epoch": 0.7396774066332769,
      "grad_norm": 39.77605056762695,
      "learning_rate": 2.6565770407341816e-07,
      "loss": 0.061,
      "step": 4689
    },
    {
      "epoch": 0.7398351540008676,
      "grad_norm": 3.652667284011841,
      "learning_rate": 2.654966994042827e-07,
      "loss": 0.0678,
      "step": 4690
    },
    {
      "epoch": 0.7399929013684584,
      "grad_norm": 3.6541528701782227,
      "learning_rate": 2.653356947351473e-07,
      "loss": 0.049,
      "step": 4691
    },
    {
      "epoch": 0.7401506487360492,
      "grad_norm": 4.591233730316162,
      "learning_rate": 2.651746900660119e-07,
      "loss": 0.0388,
      "step": 4692
    },
    {
      "epoch": 0.74030839610364,
      "grad_norm": 5.591604232788086,
      "learning_rate": 2.6501368539687654e-07,
      "loss": 0.0264,
      "step": 4693
    },
    {
      "epoch": 0.7404661434712309,
      "grad_norm": 5.747438907623291,
      "learning_rate": 2.648526807277411e-07,
      "loss": 0.0393,
      "step": 4694
    },
    {
      "epoch": 0.7406238908388216,
      "grad_norm": 7.057928562164307,
      "learning_rate": 2.646916760586057e-07,
      "loss": 0.0511,
      "step": 4695
    },
    {
      "epoch": 0.7407816382064124,
      "grad_norm": 8.497698783874512,
      "learning_rate": 2.6453067138947025e-07,
      "loss": 0.0786,
      "step": 4696
    },
    {
      "epoch": 0.7409393855740032,
      "grad_norm": 6.685943126678467,
      "learning_rate": 2.643696667203349e-07,
      "loss": 0.0275,
      "step": 4697
    },
    {
      "epoch": 0.7410971329415941,
      "grad_norm": 7.371028423309326,
      "learning_rate": 2.6420866205119946e-07,
      "loss": 0.0547,
      "step": 4698
    },
    {
      "epoch": 0.7412548803091848,
      "grad_norm": 7.228631496429443,
      "learning_rate": 2.6404765738206407e-07,
      "loss": 0.0289,
      "step": 4699
    },
    {
      "epoch": 0.7414126276767756,
      "grad_norm": 5.806507587432861,
      "learning_rate": 2.6388665271292867e-07,
      "loss": 0.0653,
      "step": 4700
    },
    {
      "epoch": 0.7415703750443664,
      "grad_norm": 5.1528496742248535,
      "learning_rate": 2.637256480437933e-07,
      "loss": 0.0441,
      "step": 4701
    },
    {
      "epoch": 0.7417281224119573,
      "grad_norm": 3.8632359504699707,
      "learning_rate": 2.6356464337465783e-07,
      "loss": 0.0503,
      "step": 4702
    },
    {
      "epoch": 0.7418858697795481,
      "grad_norm": 2.901906728744507,
      "learning_rate": 2.6340363870552244e-07,
      "loss": 0.0336,
      "step": 4703
    },
    {
      "epoch": 0.7420436171471388,
      "grad_norm": 2.611889362335205,
      "learning_rate": 2.6324263403638705e-07,
      "loss": 0.0109,
      "step": 4704
    },
    {
      "epoch": 0.7422013645147296,
      "grad_norm": 4.360488414764404,
      "learning_rate": 2.6308162936725165e-07,
      "loss": 0.0228,
      "step": 4705
    },
    {
      "epoch": 0.7423591118823205,
      "grad_norm": 4.6343302726745605,
      "learning_rate": 2.629206246981162e-07,
      "loss": 0.0449,
      "step": 4706
    },
    {
      "epoch": 0.7425168592499113,
      "grad_norm": 3.0819201469421387,
      "learning_rate": 2.627596200289808e-07,
      "loss": 0.015,
      "step": 4707
    },
    {
      "epoch": 0.7426746066175021,
      "grad_norm": 10.442688941955566,
      "learning_rate": 2.625986153598454e-07,
      "loss": 0.045,
      "step": 4708
    },
    {
      "epoch": 0.7428323539850928,
      "grad_norm": 4.94475793838501,
      "learning_rate": 2.6243761069071e-07,
      "loss": 0.0193,
      "step": 4709
    },
    {
      "epoch": 0.7429901013526837,
      "grad_norm": 6.157139301300049,
      "learning_rate": 2.622766060215746e-07,
      "loss": 0.056,
      "step": 4710
    },
    {
      "epoch": 0.7431478487202745,
      "grad_norm": 4.2757744789123535,
      "learning_rate": 2.621156013524392e-07,
      "loss": 0.0355,
      "step": 4711
    },
    {
      "epoch": 0.7433055960878653,
      "grad_norm": 3.8923239707946777,
      "learning_rate": 2.6195459668330384e-07,
      "loss": 0.0271,
      "step": 4712
    },
    {
      "epoch": 0.7434633434554561,
      "grad_norm": 10.463128089904785,
      "learning_rate": 2.617935920141684e-07,
      "loss": 0.0635,
      "step": 4713
    },
    {
      "epoch": 0.7436210908230468,
      "grad_norm": 2.57322359085083,
      "learning_rate": 2.61632587345033e-07,
      "loss": 0.0156,
      "step": 4714
    },
    {
      "epoch": 0.7437788381906377,
      "grad_norm": 7.563078880310059,
      "learning_rate": 2.614715826758976e-07,
      "loss": 0.0282,
      "step": 4715
    },
    {
      "epoch": 0.7439365855582285,
      "grad_norm": 6.335556507110596,
      "learning_rate": 2.613105780067622e-07,
      "loss": 0.0361,
      "step": 4716
    },
    {
      "epoch": 0.7440943329258193,
      "grad_norm": 2.6851494312286377,
      "learning_rate": 2.6114957333762677e-07,
      "loss": 0.0136,
      "step": 4717
    },
    {
      "epoch": 0.7442520802934101,
      "grad_norm": 4.868751049041748,
      "learning_rate": 2.609885686684914e-07,
      "loss": 0.0407,
      "step": 4718
    },
    {
      "epoch": 0.744409827661001,
      "grad_norm": 7.605571746826172,
      "learning_rate": 2.60827563999356e-07,
      "loss": 0.0511,
      "step": 4719
    },
    {
      "epoch": 0.7445675750285917,
      "grad_norm": 4.598052501678467,
      "learning_rate": 2.606665593302206e-07,
      "loss": 0.0302,
      "step": 4720
    },
    {
      "epoch": 0.7447253223961825,
      "grad_norm": 5.403987884521484,
      "learning_rate": 2.6050555466108514e-07,
      "loss": 0.047,
      "step": 4721
    },
    {
      "epoch": 0.7448830697637733,
      "grad_norm": 9.109344482421875,
      "learning_rate": 2.6034454999194975e-07,
      "loss": 0.0505,
      "step": 4722
    },
    {
      "epoch": 0.7450408171313642,
      "grad_norm": 4.874695301055908,
      "learning_rate": 2.6018354532281435e-07,
      "loss": 0.0435,
      "step": 4723
    },
    {
      "epoch": 0.7451985644989549,
      "grad_norm": 3.578183174133301,
      "learning_rate": 2.6002254065367896e-07,
      "loss": 0.0343,
      "step": 4724
    },
    {
      "epoch": 0.7453563118665457,
      "grad_norm": 4.437819957733154,
      "learning_rate": 2.598615359845435e-07,
      "loss": 0.0496,
      "step": 4725
    },
    {
      "epoch": 0.7455140592341365,
      "grad_norm": 2.6773672103881836,
      "learning_rate": 2.597005313154081e-07,
      "loss": 0.0352,
      "step": 4726
    },
    {
      "epoch": 0.7456718066017274,
      "grad_norm": 3.715035915374756,
      "learning_rate": 2.595395266462727e-07,
      "loss": 0.0368,
      "step": 4727
    },
    {
      "epoch": 0.7458295539693182,
      "grad_norm": 5.758171081542969,
      "learning_rate": 2.5937852197713733e-07,
      "loss": 0.0559,
      "step": 4728
    },
    {
      "epoch": 0.7459873013369089,
      "grad_norm": 6.195015907287598,
      "learning_rate": 2.592175173080019e-07,
      "loss": 0.0625,
      "step": 4729
    },
    {
      "epoch": 0.7461450487044997,
      "grad_norm": 3.536844491958618,
      "learning_rate": 2.5905651263886655e-07,
      "loss": 0.033,
      "step": 4730
    },
    {
      "epoch": 0.7463027960720906,
      "grad_norm": 3.621981143951416,
      "learning_rate": 2.588955079697311e-07,
      "loss": 0.0297,
      "step": 4731
    },
    {
      "epoch": 0.7464605434396814,
      "grad_norm": 7.923707485198975,
      "learning_rate": 2.587345033005957e-07,
      "loss": 0.0379,
      "step": 4732
    },
    {
      "epoch": 0.7466182908072722,
      "grad_norm": 6.867339611053467,
      "learning_rate": 2.5857349863146026e-07,
      "loss": 0.0469,
      "step": 4733
    },
    {
      "epoch": 0.7467760381748629,
      "grad_norm": 10.134578704833984,
      "learning_rate": 2.584124939623249e-07,
      "loss": 0.0595,
      "step": 4734
    },
    {
      "epoch": 0.7469337855424537,
      "grad_norm": 2.841083288192749,
      "learning_rate": 2.582514892931895e-07,
      "loss": 0.0295,
      "step": 4735
    },
    {
      "epoch": 0.7470915329100446,
      "grad_norm": 4.944960117340088,
      "learning_rate": 2.580904846240541e-07,
      "loss": 0.0282,
      "step": 4736
    },
    {
      "epoch": 0.7472492802776354,
      "grad_norm": 2.846627950668335,
      "learning_rate": 2.579294799549187e-07,
      "loss": 0.0371,
      "step": 4737
    },
    {
      "epoch": 0.7474070276452262,
      "grad_norm": 7.171413421630859,
      "learning_rate": 2.577684752857833e-07,
      "loss": 0.0766,
      "step": 4738
    },
    {
      "epoch": 0.7475647750128169,
      "grad_norm": 5.6841864585876465,
      "learning_rate": 2.576074706166479e-07,
      "loss": 0.0812,
      "step": 4739
    },
    {
      "epoch": 0.7477225223804078,
      "grad_norm": 10.416234970092773,
      "learning_rate": 2.5744646594751245e-07,
      "loss": 0.0474,
      "step": 4740
    },
    {
      "epoch": 0.7478802697479986,
      "grad_norm": 3.202110528945923,
      "learning_rate": 2.5728546127837706e-07,
      "loss": 0.0367,
      "step": 4741
    },
    {
      "epoch": 0.7480380171155894,
      "grad_norm": 6.204504013061523,
      "learning_rate": 2.5712445660924166e-07,
      "loss": 0.0462,
      "step": 4742
    },
    {
      "epoch": 0.7481957644831801,
      "grad_norm": 9.100051879882812,
      "learning_rate": 2.5696345194010627e-07,
      "loss": 0.0655,
      "step": 4743
    },
    {
      "epoch": 0.748353511850771,
      "grad_norm": 5.076653480529785,
      "learning_rate": 2.568024472709708e-07,
      "loss": 0.0378,
      "step": 4744
    },
    {
      "epoch": 0.7485112592183618,
      "grad_norm": 5.283294200897217,
      "learning_rate": 2.566414426018355e-07,
      "loss": 0.0436,
      "step": 4745
    },
    {
      "epoch": 0.7486690065859526,
      "grad_norm": 3.0343127250671387,
      "learning_rate": 2.5648043793270003e-07,
      "loss": 0.0343,
      "step": 4746
    },
    {
      "epoch": 0.7488267539535434,
      "grad_norm": 6.469725131988525,
      "learning_rate": 2.5631943326356464e-07,
      "loss": 0.0501,
      "step": 4747
    },
    {
      "epoch": 0.7489845013211343,
      "grad_norm": 4.294254302978516,
      "learning_rate": 2.561584285944292e-07,
      "loss": 0.0278,
      "step": 4748
    },
    {
      "epoch": 0.749142248688725,
      "grad_norm": 5.547506332397461,
      "learning_rate": 2.5599742392529385e-07,
      "loss": 0.0406,
      "step": 4749
    },
    {
      "epoch": 0.7492999960563158,
      "grad_norm": 3.277902126312256,
      "learning_rate": 2.558364192561584e-07,
      "loss": 0.0488,
      "step": 4750
    },
    {
      "epoch": 0.7494577434239066,
      "grad_norm": 6.372631549835205,
      "learning_rate": 2.55675414587023e-07,
      "loss": 0.055,
      "step": 4751
    },
    {
      "epoch": 0.7496154907914974,
      "grad_norm": 6.90498161315918,
      "learning_rate": 2.5551440991788757e-07,
      "loss": 0.0632,
      "step": 4752
    },
    {
      "epoch": 0.7497732381590883,
      "grad_norm": 3.1088144779205322,
      "learning_rate": 2.553534052487522e-07,
      "loss": 0.0329,
      "step": 4753
    },
    {
      "epoch": 0.749930985526679,
      "grad_norm": 3.4850716590881348,
      "learning_rate": 2.551924005796168e-07,
      "loss": 0.0423,
      "step": 4754
    },
    {
      "epoch": 0.7500887328942698,
      "grad_norm": 3.101902484893799,
      "learning_rate": 2.550313959104814e-07,
      "loss": 0.0278,
      "step": 4755
    },
    {
      "epoch": 0.7502464802618606,
      "grad_norm": 3.260136842727661,
      "learning_rate": 2.5487039124134594e-07,
      "loss": 0.0163,
      "step": 4756
    },
    {
      "epoch": 0.7504042276294515,
      "grad_norm": 3.9728457927703857,
      "learning_rate": 2.547093865722106e-07,
      "loss": 0.0537,
      "step": 4757
    },
    {
      "epoch": 0.7505619749970422,
      "grad_norm": 6.179561614990234,
      "learning_rate": 2.545483819030752e-07,
      "loss": 0.0656,
      "step": 4758
    },
    {
      "epoch": 0.750719722364633,
      "grad_norm": 3.508845567703247,
      "learning_rate": 2.5438737723393976e-07,
      "loss": 0.0186,
      "step": 4759
    },
    {
      "epoch": 0.7508774697322238,
      "grad_norm": 10.673635482788086,
      "learning_rate": 2.542263725648044e-07,
      "loss": 0.0387,
      "step": 4760
    },
    {
      "epoch": 0.7510352170998147,
      "grad_norm": 4.301787853240967,
      "learning_rate": 2.5406536789566897e-07,
      "loss": 0.0845,
      "step": 4761
    },
    {
      "epoch": 0.7511929644674055,
      "grad_norm": 4.449602127075195,
      "learning_rate": 2.539043632265336e-07,
      "loss": 0.0535,
      "step": 4762
    },
    {
      "epoch": 0.7513507118349962,
      "grad_norm": 3.7055530548095703,
      "learning_rate": 2.5374335855739813e-07,
      "loss": 0.0383,
      "step": 4763
    },
    {
      "epoch": 0.751508459202587,
      "grad_norm": 3.549971103668213,
      "learning_rate": 2.535823538882628e-07,
      "loss": 0.07,
      "step": 4764
    },
    {
      "epoch": 0.7516662065701779,
      "grad_norm": 3.0408430099487305,
      "learning_rate": 2.5342134921912734e-07,
      "loss": 0.0534,
      "step": 4765
    },
    {
      "epoch": 0.7518239539377687,
      "grad_norm": 6.597678184509277,
      "learning_rate": 2.5326034454999195e-07,
      "loss": 0.036,
      "step": 4766
    },
    {
      "epoch": 0.7519817013053595,
      "grad_norm": 3.500997543334961,
      "learning_rate": 2.530993398808565e-07,
      "loss": 0.0546,
      "step": 4767
    },
    {
      "epoch": 0.7521394486729502,
      "grad_norm": 7.335076808929443,
      "learning_rate": 2.5293833521172116e-07,
      "loss": 0.032,
      "step": 4768
    },
    {
      "epoch": 0.7522971960405411,
      "grad_norm": 6.85729455947876,
      "learning_rate": 2.527773305425857e-07,
      "loss": 0.0292,
      "step": 4769
    },
    {
      "epoch": 0.7524549434081319,
      "grad_norm": 6.948343753814697,
      "learning_rate": 2.526163258734503e-07,
      "loss": 0.039,
      "step": 4770
    },
    {
      "epoch": 0.7526126907757227,
      "grad_norm": 2.67183518409729,
      "learning_rate": 2.524553212043149e-07,
      "loss": 0.037,
      "step": 4771
    },
    {
      "epoch": 0.7527704381433135,
      "grad_norm": 3.0378963947296143,
      "learning_rate": 2.5229431653517953e-07,
      "loss": 0.0303,
      "step": 4772
    },
    {
      "epoch": 0.7529281855109042,
      "grad_norm": 6.4038591384887695,
      "learning_rate": 2.521333118660441e-07,
      "loss": 0.0595,
      "step": 4773
    },
    {
      "epoch": 0.7530859328784951,
      "grad_norm": 7.847699165344238,
      "learning_rate": 2.519723071969087e-07,
      "loss": 0.0433,
      "step": 4774
    },
    {
      "epoch": 0.7532436802460859,
      "grad_norm": 7.712286472320557,
      "learning_rate": 2.5181130252777325e-07,
      "loss": 0.0308,
      "step": 4775
    },
    {
      "epoch": 0.7534014276136767,
      "grad_norm": 3.324777603149414,
      "learning_rate": 2.516502978586379e-07,
      "loss": 0.0352,
      "step": 4776
    },
    {
      "epoch": 0.7535591749812675,
      "grad_norm": 5.5077314376831055,
      "learning_rate": 2.5148929318950246e-07,
      "loss": 0.041,
      "step": 4777
    },
    {
      "epoch": 0.7537169223488583,
      "grad_norm": 3.103212594985962,
      "learning_rate": 2.5132828852036707e-07,
      "loss": 0.0167,
      "step": 4778
    },
    {
      "epoch": 0.7538746697164491,
      "grad_norm": 3.603749990463257,
      "learning_rate": 2.5116728385123167e-07,
      "loss": 0.0508,
      "step": 4779
    },
    {
      "epoch": 0.7540324170840399,
      "grad_norm": 3.0347187519073486,
      "learning_rate": 2.510062791820963e-07,
      "loss": 0.0269,
      "step": 4780
    },
    {
      "epoch": 0.7541901644516307,
      "grad_norm": 3.86840558052063,
      "learning_rate": 2.508452745129609e-07,
      "loss": 0.0381,
      "step": 4781
    },
    {
      "epoch": 0.7543479118192216,
      "grad_norm": 9.875972747802734,
      "learning_rate": 2.5068426984382544e-07,
      "loss": 0.0881,
      "step": 4782
    },
    {
      "epoch": 0.7545056591868123,
      "grad_norm": 7.191247940063477,
      "learning_rate": 2.505232651746901e-07,
      "loss": 0.0269,
      "step": 4783
    },
    {
      "epoch": 0.7546634065544031,
      "grad_norm": 7.005562782287598,
      "learning_rate": 2.5036226050555465e-07,
      "loss": 0.0612,
      "step": 4784
    },
    {
      "epoch": 0.7548211539219939,
      "grad_norm": 3.008082628250122,
      "learning_rate": 2.5020125583641926e-07,
      "loss": 0.0405,
      "step": 4785
    },
    {
      "epoch": 0.7549789012895848,
      "grad_norm": 3.388357162475586,
      "learning_rate": 2.500402511672838e-07,
      "loss": 0.0251,
      "step": 4786
    },
    {
      "epoch": 0.7551366486571756,
      "grad_norm": 3.2952568531036377,
      "learning_rate": 2.498792464981484e-07,
      "loss": 0.0421,
      "step": 4787
    },
    {
      "epoch": 0.7552943960247663,
      "grad_norm": 8.339786529541016,
      "learning_rate": 2.49718241829013e-07,
      "loss": 0.0306,
      "step": 4788
    },
    {
      "epoch": 0.7554521433923571,
      "grad_norm": 4.5375494956970215,
      "learning_rate": 2.4955723715987763e-07,
      "loss": 0.0301,
      "step": 4789
    },
    {
      "epoch": 0.7556098907599479,
      "grad_norm": 3.0443336963653564,
      "learning_rate": 2.493962324907422e-07,
      "loss": 0.02,
      "step": 4790
    },
    {
      "epoch": 0.7557676381275388,
      "grad_norm": 4.50595760345459,
      "learning_rate": 2.492352278216068e-07,
      "loss": 0.0594,
      "step": 4791
    },
    {
      "epoch": 0.7559253854951296,
      "grad_norm": 4.146176338195801,
      "learning_rate": 2.490742231524714e-07,
      "loss": 0.0403,
      "step": 4792
    },
    {
      "epoch": 0.7560831328627203,
      "grad_norm": 5.834567546844482,
      "learning_rate": 2.48913218483336e-07,
      "loss": 0.0349,
      "step": 4793
    },
    {
      "epoch": 0.7562408802303111,
      "grad_norm": 6.704777717590332,
      "learning_rate": 2.487522138142006e-07,
      "loss": 0.0567,
      "step": 4794
    },
    {
      "epoch": 0.756398627597902,
      "grad_norm": 2.64961576461792,
      "learning_rate": 2.485912091450652e-07,
      "loss": 0.0282,
      "step": 4795
    },
    {
      "epoch": 0.7565563749654928,
      "grad_norm": 5.034839153289795,
      "learning_rate": 2.484302044759298e-07,
      "loss": 0.069,
      "step": 4796
    },
    {
      "epoch": 0.7567141223330835,
      "grad_norm": 3.204643487930298,
      "learning_rate": 2.482691998067944e-07,
      "loss": 0.0705,
      "step": 4797
    },
    {
      "epoch": 0.7568718697006743,
      "grad_norm": 3.6070191860198975,
      "learning_rate": 2.48108195137659e-07,
      "loss": 0.0594,
      "step": 4798
    },
    {
      "epoch": 0.7570296170682652,
      "grad_norm": 3.545212745666504,
      "learning_rate": 2.479471904685236e-07,
      "loss": 0.0552,
      "step": 4799
    },
    {
      "epoch": 0.757187364435856,
      "grad_norm": 4.373169898986816,
      "learning_rate": 2.477861857993882e-07,
      "loss": 0.0253,
      "step": 4800
    },
    {
      "epoch": 0.7573451118034468,
      "grad_norm": 0.9404663443565369,
      "learning_rate": 2.4762518113025275e-07,
      "loss": 0.0071,
      "step": 4801
    },
    {
      "epoch": 0.7575028591710375,
      "grad_norm": 7.398576259613037,
      "learning_rate": 2.4746417646111735e-07,
      "loss": 0.0461,
      "step": 4802
    },
    {
      "epoch": 0.7576606065386284,
      "grad_norm": 4.485061168670654,
      "learning_rate": 2.4730317179198196e-07,
      "loss": 0.0346,
      "step": 4803
    },
    {
      "epoch": 0.7578183539062192,
      "grad_norm": 6.237451553344727,
      "learning_rate": 2.4714216712284657e-07,
      "loss": 0.0425,
      "step": 4804
    },
    {
      "epoch": 0.75797610127381,
      "grad_norm": 3.685185670852661,
      "learning_rate": 2.469811624537111e-07,
      "loss": 0.0268,
      "step": 4805
    },
    {
      "epoch": 0.7581338486414008,
      "grad_norm": 7.0791335105896,
      "learning_rate": 2.468201577845757e-07,
      "loss": 0.0225,
      "step": 4806
    },
    {
      "epoch": 0.7582915960089917,
      "grad_norm": 8.132930755615234,
      "learning_rate": 2.4665915311544033e-07,
      "loss": 0.0414,
      "step": 4807
    },
    {
      "epoch": 0.7584493433765824,
      "grad_norm": 4.797114372253418,
      "learning_rate": 2.4649814844630494e-07,
      "loss": 0.0268,
      "step": 4808
    },
    {
      "epoch": 0.7586070907441732,
      "grad_norm": 3.8750391006469727,
      "learning_rate": 2.4633714377716954e-07,
      "loss": 0.0629,
      "step": 4809
    },
    {
      "epoch": 0.758764838111764,
      "grad_norm": 3.5614655017852783,
      "learning_rate": 2.461761391080341e-07,
      "loss": 0.0593,
      "step": 4810
    },
    {
      "epoch": 0.7589225854793548,
      "grad_norm": 4.513403415679932,
      "learning_rate": 2.460151344388987e-07,
      "loss": 0.0251,
      "step": 4811
    },
    {
      "epoch": 0.7590803328469456,
      "grad_norm": 2.8573057651519775,
      "learning_rate": 2.458541297697633e-07,
      "loss": 0.0245,
      "step": 4812
    },
    {
      "epoch": 0.7592380802145364,
      "grad_norm": 4.854127407073975,
      "learning_rate": 2.456931251006279e-07,
      "loss": 0.0691,
      "step": 4813
    },
    {
      "epoch": 0.7593958275821272,
      "grad_norm": 6.474213123321533,
      "learning_rate": 2.4553212043149247e-07,
      "loss": 0.0518,
      "step": 4814
    },
    {
      "epoch": 0.759553574949718,
      "grad_norm": 3.787691354751587,
      "learning_rate": 2.453711157623571e-07,
      "loss": 0.0236,
      "step": 4815
    },
    {
      "epoch": 0.7597113223173089,
      "grad_norm": 4.183278560638428,
      "learning_rate": 2.452101110932217e-07,
      "loss": 0.0768,
      "step": 4816
    },
    {
      "epoch": 0.7598690696848996,
      "grad_norm": 5.145864486694336,
      "learning_rate": 2.450491064240863e-07,
      "loss": 0.0346,
      "step": 4817
    },
    {
      "epoch": 0.7600268170524904,
      "grad_norm": 4.315591335296631,
      "learning_rate": 2.448881017549509e-07,
      "loss": 0.0258,
      "step": 4818
    },
    {
      "epoch": 0.7601845644200812,
      "grad_norm": 9.347548484802246,
      "learning_rate": 2.447270970858155e-07,
      "loss": 0.1037,
      "step": 4819
    },
    {
      "epoch": 0.7603423117876721,
      "grad_norm": 10.202075004577637,
      "learning_rate": 2.4456609241668005e-07,
      "loss": 0.0554,
      "step": 4820
    },
    {
      "epoch": 0.7605000591552629,
      "grad_norm": 8.322285652160645,
      "learning_rate": 2.4440508774754466e-07,
      "loss": 0.0497,
      "step": 4821
    },
    {
      "epoch": 0.7606578065228536,
      "grad_norm": 3.859376907348633,
      "learning_rate": 2.4424408307840927e-07,
      "loss": 0.0547,
      "step": 4822
    },
    {
      "epoch": 0.7608155538904444,
      "grad_norm": 4.298503875732422,
      "learning_rate": 2.440830784092739e-07,
      "loss": 0.0352,
      "step": 4823
    },
    {
      "epoch": 0.7609733012580353,
      "grad_norm": 6.906977653503418,
      "learning_rate": 2.439220737401385e-07,
      "loss": 0.0628,
      "step": 4824
    },
    {
      "epoch": 0.7611310486256261,
      "grad_norm": 4.583846092224121,
      "learning_rate": 2.4376106907100303e-07,
      "loss": 0.0759,
      "step": 4825
    },
    {
      "epoch": 0.7612887959932169,
      "grad_norm": 3.9792728424072266,
      "learning_rate": 2.4360006440186764e-07,
      "loss": 0.0299,
      "step": 4826
    },
    {
      "epoch": 0.7614465433608076,
      "grad_norm": 5.429749011993408,
      "learning_rate": 2.4343905973273225e-07,
      "loss": 0.0402,
      "step": 4827
    },
    {
      "epoch": 0.7616042907283985,
      "grad_norm": 2.6782314777374268,
      "learning_rate": 2.4327805506359685e-07,
      "loss": 0.017,
      "step": 4828
    },
    {
      "epoch": 0.7617620380959893,
      "grad_norm": 7.9765424728393555,
      "learning_rate": 2.431170503944614e-07,
      "loss": 0.0417,
      "step": 4829
    },
    {
      "epoch": 0.7619197854635801,
      "grad_norm": 4.836733341217041,
      "learning_rate": 2.42956045725326e-07,
      "loss": 0.0232,
      "step": 4830
    },
    {
      "epoch": 0.7620775328311709,
      "grad_norm": 1.0174866914749146,
      "learning_rate": 2.427950410561906e-07,
      "loss": 0.0072,
      "step": 4831
    },
    {
      "epoch": 0.7622352801987616,
      "grad_norm": 4.296733856201172,
      "learning_rate": 2.426340363870552e-07,
      "loss": 0.0384,
      "step": 4832
    },
    {
      "epoch": 0.7623930275663525,
      "grad_norm": 3.270857572555542,
      "learning_rate": 2.424730317179198e-07,
      "loss": 0.0171,
      "step": 4833
    },
    {
      "epoch": 0.7625507749339433,
      "grad_norm": 2.496772050857544,
      "learning_rate": 2.423120270487844e-07,
      "loss": 0.0202,
      "step": 4834
    },
    {
      "epoch": 0.7627085223015341,
      "grad_norm": 4.638273239135742,
      "learning_rate": 2.42151022379649e-07,
      "loss": 0.0372,
      "step": 4835
    },
    {
      "epoch": 0.7628662696691249,
      "grad_norm": 4.851541519165039,
      "learning_rate": 2.419900177105136e-07,
      "loss": 0.0433,
      "step": 4836
    },
    {
      "epoch": 0.7630240170367157,
      "grad_norm": 2.6814563274383545,
      "learning_rate": 2.418290130413782e-07,
      "loss": 0.0153,
      "step": 4837
    },
    {
      "epoch": 0.7631817644043065,
      "grad_norm": 8.24600601196289,
      "learning_rate": 2.4166800837224276e-07,
      "loss": 0.1217,
      "step": 4838
    },
    {
      "epoch": 0.7633395117718973,
      "grad_norm": 4.169066905975342,
      "learning_rate": 2.415070037031074e-07,
      "loss": 0.0368,
      "step": 4839
    },
    {
      "epoch": 0.7634972591394881,
      "grad_norm": 4.953155994415283,
      "learning_rate": 2.4134599903397197e-07,
      "loss": 0.0358,
      "step": 4840
    },
    {
      "epoch": 0.763655006507079,
      "grad_norm": 6.2820611000061035,
      "learning_rate": 2.411849943648366e-07,
      "loss": 0.0455,
      "step": 4841
    },
    {
      "epoch": 0.7638127538746697,
      "grad_norm": 5.614377021789551,
      "learning_rate": 2.410239896957012e-07,
      "loss": 0.057,
      "step": 4842
    },
    {
      "epoch": 0.7639705012422605,
      "grad_norm": 3.532853841781616,
      "learning_rate": 2.408629850265658e-07,
      "loss": 0.0217,
      "step": 4843
    },
    {
      "epoch": 0.7641282486098513,
      "grad_norm": 7.9854416847229,
      "learning_rate": 2.4070198035743034e-07,
      "loss": 0.0536,
      "step": 4844
    },
    {
      "epoch": 0.7642859959774422,
      "grad_norm": 9.11436939239502,
      "learning_rate": 2.4054097568829495e-07,
      "loss": 0.0833,
      "step": 4845
    },
    {
      "epoch": 0.764443743345033,
      "grad_norm": 3.794739007949829,
      "learning_rate": 2.4037997101915955e-07,
      "loss": 0.0385,
      "step": 4846
    },
    {
      "epoch": 0.7646014907126237,
      "grad_norm": 4.817052364349365,
      "learning_rate": 2.4021896635002416e-07,
      "loss": 0.0394,
      "step": 4847
    },
    {
      "epoch": 0.7647592380802145,
      "grad_norm": 46.40493392944336,
      "learning_rate": 2.400579616808887e-07,
      "loss": 0.0332,
      "step": 4848
    },
    {
      "epoch": 0.7649169854478053,
      "grad_norm": 10.331274032592773,
      "learning_rate": 2.398969570117533e-07,
      "loss": 0.0661,
      "step": 4849
    },
    {
      "epoch": 0.7650747328153962,
      "grad_norm": 4.1327128410339355,
      "learning_rate": 2.3973595234261793e-07,
      "loss": 0.0244,
      "step": 4850
    },
    {
      "epoch": 0.765232480182987,
      "grad_norm": 8.767778396606445,
      "learning_rate": 2.3957494767348253e-07,
      "loss": 0.1145,
      "step": 4851
    },
    {
      "epoch": 0.7653902275505777,
      "grad_norm": 5.427033424377441,
      "learning_rate": 2.3941394300434714e-07,
      "loss": 0.0392,
      "step": 4852
    },
    {
      "epoch": 0.7655479749181685,
      "grad_norm": 5.3446221351623535,
      "learning_rate": 2.392529383352117e-07,
      "loss": 0.0689,
      "step": 4853
    },
    {
      "epoch": 0.7657057222857594,
      "grad_norm": 5.317961692810059,
      "learning_rate": 2.390919336660763e-07,
      "loss": 0.0306,
      "step": 4854
    },
    {
      "epoch": 0.7658634696533502,
      "grad_norm": 9.808844566345215,
      "learning_rate": 2.389309289969409e-07,
      "loss": 0.0767,
      "step": 4855
    },
    {
      "epoch": 0.766021217020941,
      "grad_norm": 8.848456382751465,
      "learning_rate": 2.387699243278055e-07,
      "loss": 0.09,
      "step": 4856
    },
    {
      "epoch": 0.7661789643885317,
      "grad_norm": 6.596203327178955,
      "learning_rate": 2.3860891965867006e-07,
      "loss": 0.0475,
      "step": 4857
    },
    {
      "epoch": 0.7663367117561226,
      "grad_norm": 5.3128981590271,
      "learning_rate": 2.3844791498953467e-07,
      "loss": 0.0464,
      "step": 4858
    },
    {
      "epoch": 0.7664944591237134,
      "grad_norm": 4.245983600616455,
      "learning_rate": 2.3828691032039928e-07,
      "loss": 0.05,
      "step": 4859
    },
    {
      "epoch": 0.7666522064913042,
      "grad_norm": 6.538334369659424,
      "learning_rate": 2.3812590565126386e-07,
      "loss": 0.0277,
      "step": 4860
    },
    {
      "epoch": 0.7668099538588949,
      "grad_norm": 8.421479225158691,
      "learning_rate": 2.3796490098212846e-07,
      "loss": 0.06,
      "step": 4861
    },
    {
      "epoch": 0.7669677012264858,
      "grad_norm": 2.751904249191284,
      "learning_rate": 2.3780389631299307e-07,
      "loss": 0.0292,
      "step": 4862
    },
    {
      "epoch": 0.7671254485940766,
      "grad_norm": 4.167398452758789,
      "learning_rate": 2.3764289164385768e-07,
      "loss": 0.0197,
      "step": 4863
    },
    {
      "epoch": 0.7672831959616674,
      "grad_norm": 7.360714435577393,
      "learning_rate": 2.3748188697472226e-07,
      "loss": 0.0757,
      "step": 4864
    },
    {
      "epoch": 0.7674409433292582,
      "grad_norm": 10.587705612182617,
      "learning_rate": 2.3732088230558686e-07,
      "loss": 0.1102,
      "step": 4865
    },
    {
      "epoch": 0.767598690696849,
      "grad_norm": 5.881148815155029,
      "learning_rate": 2.3715987763645144e-07,
      "loss": 0.0518,
      "step": 4866
    },
    {
      "epoch": 0.7677564380644398,
      "grad_norm": 4.342334747314453,
      "learning_rate": 2.3699887296731605e-07,
      "loss": 0.0632,
      "step": 4867
    },
    {
      "epoch": 0.7679141854320306,
      "grad_norm": 7.428899765014648,
      "learning_rate": 2.3683786829818065e-07,
      "loss": 0.0457,
      "step": 4868
    },
    {
      "epoch": 0.7680719327996214,
      "grad_norm": 5.535422325134277,
      "learning_rate": 2.3667686362904523e-07,
      "loss": 0.055,
      "step": 4869
    },
    {
      "epoch": 0.7682296801672122,
      "grad_norm": 2.0176048278808594,
      "learning_rate": 2.3651585895990984e-07,
      "loss": 0.0137,
      "step": 4870
    },
    {
      "epoch": 0.768387427534803,
      "grad_norm": 2.525188446044922,
      "learning_rate": 2.3635485429077442e-07,
      "loss": 0.0233,
      "step": 4871
    },
    {
      "epoch": 0.7685451749023938,
      "grad_norm": 5.715892314910889,
      "learning_rate": 2.3619384962163903e-07,
      "loss": 0.0611,
      "step": 4872
    },
    {
      "epoch": 0.7687029222699846,
      "grad_norm": 4.827230930328369,
      "learning_rate": 2.360328449525036e-07,
      "loss": 0.0436,
      "step": 4873
    },
    {
      "epoch": 0.7688606696375754,
      "grad_norm": 4.874945163726807,
      "learning_rate": 2.3587184028336821e-07,
      "loss": 0.037,
      "step": 4874
    },
    {
      "epoch": 0.7690184170051663,
      "grad_norm": 4.637301445007324,
      "learning_rate": 2.357108356142328e-07,
      "loss": 0.0269,
      "step": 4875
    },
    {
      "epoch": 0.769176164372757,
      "grad_norm": 2.1692140102386475,
      "learning_rate": 2.355498309450974e-07,
      "loss": 0.0154,
      "step": 4876
    },
    {
      "epoch": 0.7693339117403478,
      "grad_norm": 4.651496887207031,
      "learning_rate": 2.3538882627596198e-07,
      "loss": 0.0496,
      "step": 4877
    },
    {
      "epoch": 0.7694916591079386,
      "grad_norm": 4.78239631652832,
      "learning_rate": 2.3522782160682659e-07,
      "loss": 0.0236,
      "step": 4878
    },
    {
      "epoch": 0.7696494064755295,
      "grad_norm": 4.077250957489014,
      "learning_rate": 2.3506681693769117e-07,
      "loss": 0.0462,
      "step": 4879
    },
    {
      "epoch": 0.7698071538431203,
      "grad_norm": 11.404817581176758,
      "learning_rate": 2.3490581226855577e-07,
      "loss": 0.1007,
      "step": 4880
    },
    {
      "epoch": 0.769964901210711,
      "grad_norm": 7.535063743591309,
      "learning_rate": 2.3474480759942035e-07,
      "loss": 0.0468,
      "step": 4881
    },
    {
      "epoch": 0.7701226485783018,
      "grad_norm": 5.084873676300049,
      "learning_rate": 2.3458380293028496e-07,
      "loss": 0.0323,
      "step": 4882
    },
    {
      "epoch": 0.7702803959458927,
      "grad_norm": 8.215431213378906,
      "learning_rate": 2.3442279826114956e-07,
      "loss": 0.0517,
      "step": 4883
    },
    {
      "epoch": 0.7704381433134835,
      "grad_norm": 1.797682285308838,
      "learning_rate": 2.3426179359201414e-07,
      "loss": 0.0081,
      "step": 4884
    },
    {
      "epoch": 0.7705958906810743,
      "grad_norm": 2.626028060913086,
      "learning_rate": 2.3410078892287875e-07,
      "loss": 0.016,
      "step": 4885
    },
    {
      "epoch": 0.770753638048665,
      "grad_norm": 3.058671236038208,
      "learning_rate": 2.3393978425374336e-07,
      "loss": 0.0163,
      "step": 4886
    },
    {
      "epoch": 0.7709113854162558,
      "grad_norm": 5.610447883605957,
      "learning_rate": 2.3377877958460796e-07,
      "loss": 0.0266,
      "step": 4887
    },
    {
      "epoch": 0.7710691327838467,
      "grad_norm": 5.915511131286621,
      "learning_rate": 2.3361777491547254e-07,
      "loss": 0.0222,
      "step": 4888
    },
    {
      "epoch": 0.7712268801514375,
      "grad_norm": 5.027588844299316,
      "learning_rate": 2.3345677024633715e-07,
      "loss": 0.018,
      "step": 4889
    },
    {
      "epoch": 0.7713846275190283,
      "grad_norm": 5.529494762420654,
      "learning_rate": 2.3329576557720173e-07,
      "loss": 0.0898,
      "step": 4890
    },
    {
      "epoch": 0.771542374886619,
      "grad_norm": 7.25527286529541,
      "learning_rate": 2.3313476090806634e-07,
      "loss": 0.0528,
      "step": 4891
    },
    {
      "epoch": 0.7717001222542099,
      "grad_norm": 4.3180389404296875,
      "learning_rate": 2.3297375623893092e-07,
      "loss": 0.0305,
      "step": 4892
    },
    {
      "epoch": 0.7718578696218007,
      "grad_norm": 2.8147454261779785,
      "learning_rate": 2.3281275156979552e-07,
      "loss": 0.0238,
      "step": 4893
    },
    {
      "epoch": 0.7720156169893915,
      "grad_norm": 3.026052236557007,
      "learning_rate": 2.326517469006601e-07,
      "loss": 0.0238,
      "step": 4894
    },
    {
      "epoch": 0.7721733643569823,
      "grad_norm": 6.311300754547119,
      "learning_rate": 2.324907422315247e-07,
      "loss": 0.0361,
      "step": 4895
    },
    {
      "epoch": 0.7723311117245731,
      "grad_norm": 6.82096529006958,
      "learning_rate": 2.323297375623893e-07,
      "loss": 0.0481,
      "step": 4896
    },
    {
      "epoch": 0.7724888590921639,
      "grad_norm": 10.022536277770996,
      "learning_rate": 2.321687328932539e-07,
      "loss": 0.085,
      "step": 4897
    },
    {
      "epoch": 0.7726466064597547,
      "grad_norm": 6.319077491760254,
      "learning_rate": 2.320077282241185e-07,
      "loss": 0.0896,
      "step": 4898
    },
    {
      "epoch": 0.7728043538273455,
      "grad_norm": 4.465986728668213,
      "learning_rate": 2.3184672355498308e-07,
      "loss": 0.017,
      "step": 4899
    },
    {
      "epoch": 0.7729621011949364,
      "grad_norm": 3.6244585514068604,
      "learning_rate": 2.3168571888584769e-07,
      "loss": 0.0214,
      "step": 4900
    },
    {
      "epoch": 0.7731198485625271,
      "grad_norm": 4.184035301208496,
      "learning_rate": 2.3152471421671227e-07,
      "loss": 0.038,
      "step": 4901
    },
    {
      "epoch": 0.7732775959301179,
      "grad_norm": 5.894932270050049,
      "learning_rate": 2.3136370954757687e-07,
      "loss": 0.0434,
      "step": 4902
    },
    {
      "epoch": 0.7734353432977087,
      "grad_norm": 4.606961727142334,
      "learning_rate": 2.3120270487844145e-07,
      "loss": 0.0793,
      "step": 4903
    },
    {
      "epoch": 0.7735930906652996,
      "grad_norm": 7.030303955078125,
      "learning_rate": 2.3104170020930606e-07,
      "loss": 0.0397,
      "step": 4904
    },
    {
      "epoch": 0.7737508380328904,
      "grad_norm": 4.425064563751221,
      "learning_rate": 2.3088069554017064e-07,
      "loss": 0.0951,
      "step": 4905
    },
    {
      "epoch": 0.7739085854004811,
      "grad_norm": 4.096029281616211,
      "learning_rate": 2.3071969087103524e-07,
      "loss": 0.0241,
      "step": 4906
    },
    {
      "epoch": 0.7740663327680719,
      "grad_norm": 8.110779762268066,
      "learning_rate": 2.3055868620189982e-07,
      "loss": 0.0728,
      "step": 4907
    },
    {
      "epoch": 0.7742240801356627,
      "grad_norm": 4.087162494659424,
      "learning_rate": 2.3039768153276443e-07,
      "loss": 0.0117,
      "step": 4908
    },
    {
      "epoch": 0.7743818275032536,
      "grad_norm": 4.035387992858887,
      "learning_rate": 2.3023667686362904e-07,
      "loss": 0.0131,
      "step": 4909
    },
    {
      "epoch": 0.7745395748708443,
      "grad_norm": 6.301119804382324,
      "learning_rate": 2.3007567219449364e-07,
      "loss": 0.0493,
      "step": 4910
    },
    {
      "epoch": 0.7746973222384351,
      "grad_norm": 2.393937349319458,
      "learning_rate": 2.2991466752535822e-07,
      "loss": 0.02,
      "step": 4911
    },
    {
      "epoch": 0.7748550696060259,
      "grad_norm": 5.6482930183410645,
      "learning_rate": 2.2975366285622283e-07,
      "loss": 0.0235,
      "step": 4912
    },
    {
      "epoch": 0.7750128169736168,
      "grad_norm": 6.207182884216309,
      "learning_rate": 2.2959265818708744e-07,
      "loss": 0.0537,
      "step": 4913
    },
    {
      "epoch": 0.7751705643412076,
      "grad_norm": 3.2115135192871094,
      "learning_rate": 2.2943165351795202e-07,
      "loss": 0.0363,
      "step": 4914
    },
    {
      "epoch": 0.7753283117087983,
      "grad_norm": 2.6930439472198486,
      "learning_rate": 2.2927064884881662e-07,
      "loss": 0.033,
      "step": 4915
    },
    {
      "epoch": 0.7754860590763891,
      "grad_norm": 5.9401984214782715,
      "learning_rate": 2.291096441796812e-07,
      "loss": 0.0288,
      "step": 4916
    },
    {
      "epoch": 0.77564380644398,
      "grad_norm": 3.733520984649658,
      "learning_rate": 2.289486395105458e-07,
      "loss": 0.0402,
      "step": 4917
    },
    {
      "epoch": 0.7758015538115708,
      "grad_norm": 3.3514657020568848,
      "learning_rate": 2.287876348414104e-07,
      "loss": 0.0177,
      "step": 4918
    },
    {
      "epoch": 0.7759593011791616,
      "grad_norm": 7.077025890350342,
      "learning_rate": 2.28626630172275e-07,
      "loss": 0.0973,
      "step": 4919
    },
    {
      "epoch": 0.7761170485467523,
      "grad_norm": 5.048889636993408,
      "learning_rate": 2.2846562550313957e-07,
      "loss": 0.0482,
      "step": 4920
    },
    {
      "epoch": 0.7762747959143432,
      "grad_norm": 5.32190465927124,
      "learning_rate": 2.2830462083400418e-07,
      "loss": 0.0361,
      "step": 4921
    },
    {
      "epoch": 0.776432543281934,
      "grad_norm": 3.7085556983947754,
      "learning_rate": 2.2814361616486876e-07,
      "loss": 0.0413,
      "step": 4922
    },
    {
      "epoch": 0.7765902906495248,
      "grad_norm": 5.97633695602417,
      "learning_rate": 2.2798261149573337e-07,
      "loss": 0.0395,
      "step": 4923
    },
    {
      "epoch": 0.7767480380171156,
      "grad_norm": 6.187735080718994,
      "learning_rate": 2.2782160682659795e-07,
      "loss": 0.0464,
      "step": 4924
    },
    {
      "epoch": 0.7769057853847063,
      "grad_norm": 6.129651069641113,
      "learning_rate": 2.2766060215746255e-07,
      "loss": 0.0365,
      "step": 4925
    },
    {
      "epoch": 0.7770635327522972,
      "grad_norm": 3.3210835456848145,
      "learning_rate": 2.2749959748832713e-07,
      "loss": 0.0744,
      "step": 4926
    },
    {
      "epoch": 0.777221280119888,
      "grad_norm": 4.758389472961426,
      "learning_rate": 2.2733859281919174e-07,
      "loss": 0.037,
      "step": 4927
    },
    {
      "epoch": 0.7773790274874788,
      "grad_norm": 2.596815824508667,
      "learning_rate": 2.2717758815005632e-07,
      "loss": 0.0134,
      "step": 4928
    },
    {
      "epoch": 0.7775367748550696,
      "grad_norm": 7.266978740692139,
      "learning_rate": 2.2701658348092093e-07,
      "loss": 0.0338,
      "step": 4929
    },
    {
      "epoch": 0.7776945222226604,
      "grad_norm": 5.165182113647461,
      "learning_rate": 2.2685557881178553e-07,
      "loss": 0.0537,
      "step": 4930
    },
    {
      "epoch": 0.7778522695902512,
      "grad_norm": 8.269397735595703,
      "learning_rate": 2.266945741426501e-07,
      "loss": 0.0432,
      "step": 4931
    },
    {
      "epoch": 0.778010016957842,
      "grad_norm": 5.935682773590088,
      "learning_rate": 2.2653356947351474e-07,
      "loss": 0.0484,
      "step": 4932
    },
    {
      "epoch": 0.7781677643254328,
      "grad_norm": 2.917829751968384,
      "learning_rate": 2.2637256480437932e-07,
      "loss": 0.0229,
      "step": 4933
    },
    {
      "epoch": 0.7783255116930237,
      "grad_norm": 3.5857861042022705,
      "learning_rate": 2.2621156013524393e-07,
      "loss": 0.0288,
      "step": 4934
    },
    {
      "epoch": 0.7784832590606144,
      "grad_norm": 7.50002384185791,
      "learning_rate": 2.260505554661085e-07,
      "loss": 0.0308,
      "step": 4935
    },
    {
      "epoch": 0.7786410064282052,
      "grad_norm": 2.9288904666900635,
      "learning_rate": 2.2588955079697312e-07,
      "loss": 0.0164,
      "step": 4936
    },
    {
      "epoch": 0.778798753795796,
      "grad_norm": 3.809661388397217,
      "learning_rate": 2.257285461278377e-07,
      "loss": 0.024,
      "step": 4937
    },
    {
      "epoch": 0.7789565011633869,
      "grad_norm": 5.5322442054748535,
      "learning_rate": 2.255675414587023e-07,
      "loss": 0.0491,
      "step": 4938
    },
    {
      "epoch": 0.7791142485309777,
      "grad_norm": 2.1326024532318115,
      "learning_rate": 2.2540653678956688e-07,
      "loss": 0.0106,
      "step": 4939
    },
    {
      "epoch": 0.7792719958985684,
      "grad_norm": 7.786957263946533,
      "learning_rate": 2.252455321204315e-07,
      "loss": 0.0653,
      "step": 4940
    },
    {
      "epoch": 0.7794297432661592,
      "grad_norm": 4.649642467498779,
      "learning_rate": 2.2508452745129607e-07,
      "loss": 0.0257,
      "step": 4941
    },
    {
      "epoch": 0.7795874906337501,
      "grad_norm": 3.2936179637908936,
      "learning_rate": 2.2492352278216068e-07,
      "loss": 0.0304,
      "step": 4942
    },
    {
      "epoch": 0.7797452380013409,
      "grad_norm": 3.508375883102417,
      "learning_rate": 2.2476251811302526e-07,
      "loss": 0.0197,
      "step": 4943
    },
    {
      "epoch": 0.7799029853689317,
      "grad_norm": 4.908533573150635,
      "learning_rate": 2.2460151344388986e-07,
      "loss": 0.0472,
      "step": 4944
    },
    {
      "epoch": 0.7800607327365224,
      "grad_norm": 3.5363972187042236,
      "learning_rate": 2.2444050877475447e-07,
      "loss": 0.0446,
      "step": 4945
    },
    {
      "epoch": 0.7802184801041132,
      "grad_norm": 8.4215726852417,
      "learning_rate": 2.2427950410561905e-07,
      "loss": 0.0679,
      "step": 4946
    },
    {
      "epoch": 0.7803762274717041,
      "grad_norm": 4.550530910491943,
      "learning_rate": 2.2411849943648365e-07,
      "loss": 0.0251,
      "step": 4947
    },
    {
      "epoch": 0.7805339748392949,
      "grad_norm": 4.5460662841796875,
      "learning_rate": 2.2395749476734823e-07,
      "loss": 0.0345,
      "step": 4948
    },
    {
      "epoch": 0.7806917222068857,
      "grad_norm": 6.487603187561035,
      "learning_rate": 2.2379649009821284e-07,
      "loss": 0.0756,
      "step": 4949
    },
    {
      "epoch": 0.7808494695744764,
      "grad_norm": 7.149035453796387,
      "learning_rate": 2.2363548542907742e-07,
      "loss": 0.0377,
      "step": 4950
    },
    {
      "epoch": 0.7810072169420673,
      "grad_norm": 5.912919521331787,
      "learning_rate": 2.2347448075994203e-07,
      "loss": 0.0654,
      "step": 4951
    },
    {
      "epoch": 0.7811649643096581,
      "grad_norm": 7.855524063110352,
      "learning_rate": 2.233134760908066e-07,
      "loss": 0.0607,
      "step": 4952
    },
    {
      "epoch": 0.7813227116772489,
      "grad_norm": 7.1649065017700195,
      "learning_rate": 2.231524714216712e-07,
      "loss": 0.0493,
      "step": 4953
    },
    {
      "epoch": 0.7814804590448396,
      "grad_norm": 2.527353525161743,
      "learning_rate": 2.229914667525358e-07,
      "loss": 0.0193,
      "step": 4954
    },
    {
      "epoch": 0.7816382064124305,
      "grad_norm": 9.294767379760742,
      "learning_rate": 2.2283046208340042e-07,
      "loss": 0.0814,
      "step": 4955
    },
    {
      "epoch": 0.7817959537800213,
      "grad_norm": 7.593995094299316,
      "learning_rate": 2.22669457414265e-07,
      "loss": 0.049,
      "step": 4956
    },
    {
      "epoch": 0.7819537011476121,
      "grad_norm": 9.311671257019043,
      "learning_rate": 2.225084527451296e-07,
      "loss": 0.0796,
      "step": 4957
    },
    {
      "epoch": 0.7821114485152029,
      "grad_norm": 5.528035640716553,
      "learning_rate": 2.223474480759942e-07,
      "loss": 0.0503,
      "step": 4958
    },
    {
      "epoch": 0.7822691958827938,
      "grad_norm": 3.1407878398895264,
      "learning_rate": 2.221864434068588e-07,
      "loss": 0.0189,
      "step": 4959
    },
    {
      "epoch": 0.7824269432503845,
      "grad_norm": 2.911999464035034,
      "learning_rate": 2.220254387377234e-07,
      "loss": 0.0364,
      "step": 4960
    },
    {
      "epoch": 0.7825846906179753,
      "grad_norm": 6.310225963592529,
      "learning_rate": 2.2186443406858798e-07,
      "loss": 0.0259,
      "step": 4961
    },
    {
      "epoch": 0.7827424379855661,
      "grad_norm": 5.104500770568848,
      "learning_rate": 2.217034293994526e-07,
      "loss": 0.0329,
      "step": 4962
    },
    {
      "epoch": 0.782900185353157,
      "grad_norm": 4.815276145935059,
      "learning_rate": 2.2154242473031717e-07,
      "loss": 0.0276,
      "step": 4963
    },
    {
      "epoch": 0.7830579327207478,
      "grad_norm": 4.592746734619141,
      "learning_rate": 2.2138142006118178e-07,
      "loss": 0.0459,
      "step": 4964
    },
    {
      "epoch": 0.7832156800883385,
      "grad_norm": 4.836174488067627,
      "learning_rate": 2.2122041539204636e-07,
      "loss": 0.0271,
      "step": 4965
    },
    {
      "epoch": 0.7833734274559293,
      "grad_norm": 4.292660713195801,
      "learning_rate": 2.2105941072291096e-07,
      "loss": 0.0321,
      "step": 4966
    },
    {
      "epoch": 0.7835311748235201,
      "grad_norm": 7.005777359008789,
      "learning_rate": 2.2089840605377554e-07,
      "loss": 0.0299,
      "step": 4967
    },
    {
      "epoch": 0.783688922191111,
      "grad_norm": 4.866457462310791,
      "learning_rate": 2.2073740138464015e-07,
      "loss": 0.034,
      "step": 4968
    },
    {
      "epoch": 0.7838466695587017,
      "grad_norm": 5.929502010345459,
      "learning_rate": 2.2057639671550473e-07,
      "loss": 0.0552,
      "step": 4969
    },
    {
      "epoch": 0.7840044169262925,
      "grad_norm": 7.180016040802002,
      "learning_rate": 2.2041539204636933e-07,
      "loss": 0.0523,
      "step": 4970
    },
    {
      "epoch": 0.7841621642938833,
      "grad_norm": 5.354877948760986,
      "learning_rate": 2.2025438737723391e-07,
      "loss": 0.06,
      "step": 4971
    },
    {
      "epoch": 0.7843199116614742,
      "grad_norm": 5.529969692230225,
      "learning_rate": 2.2009338270809852e-07,
      "loss": 0.0269,
      "step": 4972
    },
    {
      "epoch": 0.784477659029065,
      "grad_norm": 2.358067512512207,
      "learning_rate": 2.199323780389631e-07,
      "loss": 0.04,
      "step": 4973
    },
    {
      "epoch": 0.7846354063966557,
      "grad_norm": 4.004404544830322,
      "learning_rate": 2.197713733698277e-07,
      "loss": 0.0311,
      "step": 4974
    },
    {
      "epoch": 0.7847931537642465,
      "grad_norm": 3.5707602500915527,
      "learning_rate": 2.196103687006923e-07,
      "loss": 0.0191,
      "step": 4975
    },
    {
      "epoch": 0.7849509011318374,
      "grad_norm": 22.91334342956543,
      "learning_rate": 2.194493640315569e-07,
      "loss": 0.0181,
      "step": 4976
    },
    {
      "epoch": 0.7851086484994282,
      "grad_norm": 4.836089134216309,
      "learning_rate": 2.192883593624215e-07,
      "loss": 0.0604,
      "step": 4977
    },
    {
      "epoch": 0.785266395867019,
      "grad_norm": 5.083307266235352,
      "learning_rate": 2.1912735469328608e-07,
      "loss": 0.0234,
      "step": 4978
    },
    {
      "epoch": 0.7854241432346097,
      "grad_norm": 3.707251787185669,
      "learning_rate": 2.189663500241507e-07,
      "loss": 0.0142,
      "step": 4979
    },
    {
      "epoch": 0.7855818906022006,
      "grad_norm": 6.500045299530029,
      "learning_rate": 2.188053453550153e-07,
      "loss": 0.0267,
      "step": 4980
    },
    {
      "epoch": 0.7857396379697914,
      "grad_norm": 9.503015518188477,
      "learning_rate": 2.186443406858799e-07,
      "loss": 0.0521,
      "step": 4981
    },
    {
      "epoch": 0.7858973853373822,
      "grad_norm": 4.674173355102539,
      "learning_rate": 2.1848333601674448e-07,
      "loss": 0.0601,
      "step": 4982
    },
    {
      "epoch": 0.786055132704973,
      "grad_norm": 3.1530070304870605,
      "learning_rate": 2.1832233134760908e-07,
      "loss": 0.0256,
      "step": 4983
    },
    {
      "epoch": 0.7862128800725637,
      "grad_norm": 5.259850978851318,
      "learning_rate": 2.1816132667847366e-07,
      "loss": 0.0371,
      "step": 4984
    },
    {
      "epoch": 0.7863706274401546,
      "grad_norm": 4.983250617980957,
      "learning_rate": 2.1800032200933827e-07,
      "loss": 0.0252,
      "step": 4985
    },
    {
      "epoch": 0.7865283748077454,
      "grad_norm": 4.136760234832764,
      "learning_rate": 2.1783931734020285e-07,
      "loss": 0.046,
      "step": 4986
    },
    {
      "epoch": 0.7866861221753362,
      "grad_norm": 3.8495824337005615,
      "learning_rate": 2.1767831267106746e-07,
      "loss": 0.0566,
      "step": 4987
    },
    {
      "epoch": 0.786843869542927,
      "grad_norm": 5.45047664642334,
      "learning_rate": 2.1751730800193204e-07,
      "loss": 0.0453,
      "step": 4988
    },
    {
      "epoch": 0.7870016169105178,
      "grad_norm": 7.654728412628174,
      "learning_rate": 2.1735630333279664e-07,
      "loss": 0.0835,
      "step": 4989
    },
    {
      "epoch": 0.7871593642781086,
      "grad_norm": 6.550886631011963,
      "learning_rate": 2.1719529866366125e-07,
      "loss": 0.0563,
      "step": 4990
    },
    {
      "epoch": 0.7873171116456994,
      "grad_norm": 7.195198059082031,
      "learning_rate": 2.1703429399452583e-07,
      "loss": 0.0516,
      "step": 4991
    },
    {
      "epoch": 0.7874748590132902,
      "grad_norm": 5.970688819885254,
      "learning_rate": 2.1687328932539044e-07,
      "loss": 0.0673,
      "step": 4992
    },
    {
      "epoch": 0.7876326063808811,
      "grad_norm": 5.075960159301758,
      "learning_rate": 2.1671228465625501e-07,
      "loss": 0.0229,
      "step": 4993
    },
    {
      "epoch": 0.7877903537484718,
      "grad_norm": 6.305851936340332,
      "learning_rate": 2.1655127998711962e-07,
      "loss": 0.0966,
      "step": 4994
    },
    {
      "epoch": 0.7879481011160626,
      "grad_norm": 4.926904201507568,
      "learning_rate": 2.163902753179842e-07,
      "loss": 0.0955,
      "step": 4995
    },
    {
      "epoch": 0.7881058484836534,
      "grad_norm": 5.612282752990723,
      "learning_rate": 2.162292706488488e-07,
      "loss": 0.0505,
      "step": 4996
    },
    {
      "epoch": 0.7882635958512443,
      "grad_norm": 4.992518901824951,
      "learning_rate": 2.160682659797134e-07,
      "loss": 0.0394,
      "step": 4997
    },
    {
      "epoch": 0.7884213432188351,
      "grad_norm": 2.6907243728637695,
      "learning_rate": 2.15907261310578e-07,
      "loss": 0.022,
      "step": 4998
    },
    {
      "epoch": 0.7885790905864258,
      "grad_norm": 3.63834547996521,
      "learning_rate": 2.1574625664144257e-07,
      "loss": 0.0419,
      "step": 4999
    },
    {
      "epoch": 0.7887368379540166,
      "grad_norm": 5.431453704833984,
      "learning_rate": 2.1558525197230718e-07,
      "loss": 0.0387,
      "step": 5000
    },
    {
      "epoch": 0.7888945853216075,
      "grad_norm": 3.6970741748809814,
      "learning_rate": 2.1542424730317176e-07,
      "loss": 0.0143,
      "step": 5001
    },
    {
      "epoch": 0.7890523326891983,
      "grad_norm": 4.054006576538086,
      "learning_rate": 2.152632426340364e-07,
      "loss": 0.0461,
      "step": 5002
    },
    {
      "epoch": 0.789210080056789,
      "grad_norm": 5.707550048828125,
      "learning_rate": 2.1510223796490097e-07,
      "loss": 0.0571,
      "step": 5003
    },
    {
      "epoch": 0.7893678274243798,
      "grad_norm": 2.3995659351348877,
      "learning_rate": 2.1494123329576558e-07,
      "loss": 0.0114,
      "step": 5004
    },
    {
      "epoch": 0.7895255747919706,
      "grad_norm": 6.082056999206543,
      "learning_rate": 2.1478022862663018e-07,
      "loss": 0.0759,
      "step": 5005
    },
    {
      "epoch": 0.7896833221595615,
      "grad_norm": 8.491081237792969,
      "learning_rate": 2.1461922395749476e-07,
      "loss": 0.0741,
      "step": 5006
    },
    {
      "epoch": 0.7898410695271523,
      "grad_norm": 3.2995965480804443,
      "learning_rate": 2.1445821928835937e-07,
      "loss": 0.0358,
      "step": 5007
    },
    {
      "epoch": 0.789998816894743,
      "grad_norm": 3.908900737762451,
      "learning_rate": 2.1429721461922395e-07,
      "loss": 0.0341,
      "step": 5008
    },
    {
      "epoch": 0.7901565642623338,
      "grad_norm": 3.2598395347595215,
      "learning_rate": 2.1413620995008856e-07,
      "loss": 0.0123,
      "step": 5009
    },
    {
      "epoch": 0.7903143116299247,
      "grad_norm": 5.370056629180908,
      "learning_rate": 2.1397520528095314e-07,
      "loss": 0.0787,
      "step": 5010
    },
    {
      "epoch": 0.7904720589975155,
      "grad_norm": 4.599198818206787,
      "learning_rate": 2.1381420061181774e-07,
      "loss": 0.0424,
      "step": 5011
    },
    {
      "epoch": 0.7906298063651063,
      "grad_norm": 7.477070331573486,
      "learning_rate": 2.1365319594268232e-07,
      "loss": 0.0829,
      "step": 5012
    },
    {
      "epoch": 0.790787553732697,
      "grad_norm": 3.146921396255493,
      "learning_rate": 2.1349219127354693e-07,
      "loss": 0.0374,
      "step": 5013
    },
    {
      "epoch": 0.7909453011002879,
      "grad_norm": 5.969034194946289,
      "learning_rate": 2.133311866044115e-07,
      "loss": 0.0502,
      "step": 5014
    },
    {
      "epoch": 0.7911030484678787,
      "grad_norm": 6.728024005889893,
      "learning_rate": 2.1317018193527612e-07,
      "loss": 0.0762,
      "step": 5015
    },
    {
      "epoch": 0.7912607958354695,
      "grad_norm": 3.958031177520752,
      "learning_rate": 2.130091772661407e-07,
      "loss": 0.0425,
      "step": 5016
    },
    {
      "epoch": 0.7914185432030603,
      "grad_norm": 7.5270586013793945,
      "learning_rate": 2.128481725970053e-07,
      "loss": 0.0662,
      "step": 5017
    },
    {
      "epoch": 0.7915762905706512,
      "grad_norm": 4.066029071807861,
      "learning_rate": 2.1268716792786988e-07,
      "loss": 0.0555,
      "step": 5018
    },
    {
      "epoch": 0.7917340379382419,
      "grad_norm": 6.936734199523926,
      "learning_rate": 2.125261632587345e-07,
      "loss": 0.0457,
      "step": 5019
    },
    {
      "epoch": 0.7918917853058327,
      "grad_norm": 4.055303573608398,
      "learning_rate": 2.1236515858959907e-07,
      "loss": 0.0227,
      "step": 5020
    },
    {
      "epoch": 0.7920495326734235,
      "grad_norm": 6.9554619789123535,
      "learning_rate": 2.1220415392046367e-07,
      "loss": 0.0445,
      "step": 5021
    },
    {
      "epoch": 0.7922072800410143,
      "grad_norm": 5.465765476226807,
      "learning_rate": 2.1204314925132828e-07,
      "loss": 0.0621,
      "step": 5022
    },
    {
      "epoch": 0.7923650274086051,
      "grad_norm": 5.453721046447754,
      "learning_rate": 2.1188214458219286e-07,
      "loss": 0.0395,
      "step": 5023
    },
    {
      "epoch": 0.7925227747761959,
      "grad_norm": 8.362287521362305,
      "learning_rate": 2.1172113991305747e-07,
      "loss": 0.0564,
      "step": 5024
    },
    {
      "epoch": 0.7926805221437867,
      "grad_norm": 6.847805500030518,
      "learning_rate": 2.1156013524392207e-07,
      "loss": 0.0989,
      "step": 5025
    },
    {
      "epoch": 0.7928382695113775,
      "grad_norm": 6.537115573883057,
      "learning_rate": 2.1139913057478668e-07,
      "loss": 0.0343,
      "step": 5026
    },
    {
      "epoch": 0.7929960168789684,
      "grad_norm": 5.798552989959717,
      "learning_rate": 2.1123812590565126e-07,
      "loss": 0.0401,
      "step": 5027
    },
    {
      "epoch": 0.7931537642465591,
      "grad_norm": 6.021036624908447,
      "learning_rate": 2.1107712123651587e-07,
      "loss": 0.0705,
      "step": 5028
    },
    {
      "epoch": 0.7933115116141499,
      "grad_norm": 11.112351417541504,
      "learning_rate": 2.1091611656738045e-07,
      "loss": 0.0628,
      "step": 5029
    },
    {
      "epoch": 0.7934692589817407,
      "grad_norm": 4.815724849700928,
      "learning_rate": 2.1075511189824505e-07,
      "loss": 0.0242,
      "step": 5030
    },
    {
      "epoch": 0.7936270063493316,
      "grad_norm": 8.125972747802734,
      "learning_rate": 2.1059410722910963e-07,
      "loss": 0.0861,
      "step": 5031
    },
    {
      "epoch": 0.7937847537169224,
      "grad_norm": 6.67203950881958,
      "learning_rate": 2.1043310255997424e-07,
      "loss": 0.0603,
      "step": 5032
    },
    {
      "epoch": 0.7939425010845131,
      "grad_norm": 5.425326347351074,
      "learning_rate": 2.1027209789083882e-07,
      "loss": 0.0695,
      "step": 5033
    },
    {
      "epoch": 0.7941002484521039,
      "grad_norm": 3.3202567100524902,
      "learning_rate": 2.1011109322170342e-07,
      "loss": 0.0209,
      "step": 5034
    },
    {
      "epoch": 0.7942579958196948,
      "grad_norm": 4.705724239349365,
      "learning_rate": 2.09950088552568e-07,
      "loss": 0.0535,
      "step": 5035
    },
    {
      "epoch": 0.7944157431872856,
      "grad_norm": 8.349897384643555,
      "learning_rate": 2.097890838834326e-07,
      "loss": 0.0793,
      "step": 5036
    },
    {
      "epoch": 0.7945734905548764,
      "grad_norm": 4.8175153732299805,
      "learning_rate": 2.0962807921429722e-07,
      "loss": 0.0428,
      "step": 5037
    },
    {
      "epoch": 0.7947312379224671,
      "grad_norm": 5.381453037261963,
      "learning_rate": 2.094670745451618e-07,
      "loss": 0.06,
      "step": 5038
    },
    {
      "epoch": 0.794888985290058,
      "grad_norm": 7.311165809631348,
      "learning_rate": 2.093060698760264e-07,
      "loss": 0.0732,
      "step": 5039
    },
    {
      "epoch": 0.7950467326576488,
      "grad_norm": 4.570580005645752,
      "learning_rate": 2.0914506520689098e-07,
      "loss": 0.0466,
      "step": 5040
    },
    {
      "epoch": 0.7952044800252396,
      "grad_norm": 6.096408367156982,
      "learning_rate": 2.089840605377556e-07,
      "loss": 0.087,
      "step": 5041
    },
    {
      "epoch": 0.7953622273928304,
      "grad_norm": 3.1089320182800293,
      "learning_rate": 2.0882305586862017e-07,
      "loss": 0.0161,
      "step": 5042
    },
    {
      "epoch": 0.7955199747604211,
      "grad_norm": 4.253561496734619,
      "learning_rate": 2.0866205119948477e-07,
      "loss": 0.0497,
      "step": 5043
    },
    {
      "epoch": 0.795677722128012,
      "grad_norm": 5.132200241088867,
      "learning_rate": 2.0850104653034935e-07,
      "loss": 0.0631,
      "step": 5044
    },
    {
      "epoch": 0.7958354694956028,
      "grad_norm": 1.9497289657592773,
      "learning_rate": 2.0834004186121396e-07,
      "loss": 0.0427,
      "step": 5045
    },
    {
      "epoch": 0.7959932168631936,
      "grad_norm": 3.9391345977783203,
      "learning_rate": 2.0817903719207854e-07,
      "loss": 0.0766,
      "step": 5046
    },
    {
      "epoch": 0.7961509642307844,
      "grad_norm": 9.721490859985352,
      "learning_rate": 2.0801803252294315e-07,
      "loss": 0.0479,
      "step": 5047
    },
    {
      "epoch": 0.7963087115983752,
      "grad_norm": 4.6253767013549805,
      "learning_rate": 2.0785702785380775e-07,
      "loss": 0.0686,
      "step": 5048
    },
    {
      "epoch": 0.796466458965966,
      "grad_norm": 5.105081081390381,
      "learning_rate": 2.0769602318467236e-07,
      "loss": 0.0437,
      "step": 5049
    },
    {
      "epoch": 0.7966242063335568,
      "grad_norm": 4.698444366455078,
      "learning_rate": 2.0753501851553694e-07,
      "loss": 0.0604,
      "step": 5050
    },
    {
      "epoch": 0.7967819537011476,
      "grad_norm": 3.0366082191467285,
      "learning_rate": 2.0737401384640155e-07,
      "loss": 0.0203,
      "step": 5051
    },
    {
      "epoch": 0.7969397010687385,
      "grad_norm": 1.8577876091003418,
      "learning_rate": 2.0721300917726615e-07,
      "loss": 0.0143,
      "step": 5052
    },
    {
      "epoch": 0.7970974484363292,
      "grad_norm": 3.2034897804260254,
      "learning_rate": 2.0705200450813073e-07,
      "loss": 0.0335,
      "step": 5053
    },
    {
      "epoch": 0.79725519580392,
      "grad_norm": 7.292186737060547,
      "learning_rate": 2.0689099983899534e-07,
      "loss": 0.054,
      "step": 5054
    },
    {
      "epoch": 0.7974129431715108,
      "grad_norm": 5.905879497528076,
      "learning_rate": 2.0672999516985992e-07,
      "loss": 0.042,
      "step": 5055
    },
    {
      "epoch": 0.7975706905391017,
      "grad_norm": 4.840546131134033,
      "learning_rate": 2.0656899050072452e-07,
      "loss": 0.0408,
      "step": 5056
    },
    {
      "epoch": 0.7977284379066925,
      "grad_norm": 6.667255401611328,
      "learning_rate": 2.064079858315891e-07,
      "loss": 0.035,
      "step": 5057
    },
    {
      "epoch": 0.7978861852742832,
      "grad_norm": 5.428292751312256,
      "learning_rate": 2.062469811624537e-07,
      "loss": 0.0349,
      "step": 5058
    },
    {
      "epoch": 0.798043932641874,
      "grad_norm": 3.1888558864593506,
      "learning_rate": 2.060859764933183e-07,
      "loss": 0.0363,
      "step": 5059
    },
    {
      "epoch": 0.7982016800094648,
      "grad_norm": 3.992692470550537,
      "learning_rate": 2.059249718241829e-07,
      "loss": 0.0483,
      "step": 5060
    },
    {
      "epoch": 0.7983594273770557,
      "grad_norm": 4.776655197143555,
      "learning_rate": 2.0576396715504748e-07,
      "loss": 0.072,
      "step": 5061
    },
    {
      "epoch": 0.7985171747446465,
      "grad_norm": 8.188023567199707,
      "learning_rate": 2.0560296248591208e-07,
      "loss": 0.0304,
      "step": 5062
    },
    {
      "epoch": 0.7986749221122372,
      "grad_norm": 2.867313861846924,
      "learning_rate": 2.0544195781677666e-07,
      "loss": 0.008,
      "step": 5063
    },
    {
      "epoch": 0.798832669479828,
      "grad_norm": 9.445868492126465,
      "learning_rate": 2.0528095314764127e-07,
      "loss": 0.0736,
      "step": 5064
    },
    {
      "epoch": 0.7989904168474189,
      "grad_norm": 5.495667934417725,
      "learning_rate": 2.0511994847850585e-07,
      "loss": 0.0257,
      "step": 5065
    },
    {
      "epoch": 0.7991481642150097,
      "grad_norm": 6.540268898010254,
      "learning_rate": 2.0495894380937046e-07,
      "loss": 0.0598,
      "step": 5066
    },
    {
      "epoch": 0.7993059115826004,
      "grad_norm": 4.974075794219971,
      "learning_rate": 2.0479793914023506e-07,
      "loss": 0.056,
      "step": 5067
    },
    {
      "epoch": 0.7994636589501912,
      "grad_norm": 0.9020988941192627,
      "learning_rate": 2.0463693447109964e-07,
      "loss": 0.0085,
      "step": 5068
    },
    {
      "epoch": 0.7996214063177821,
      "grad_norm": 7.461383819580078,
      "learning_rate": 2.0447592980196425e-07,
      "loss": 0.0706,
      "step": 5069
    },
    {
      "epoch": 0.7997791536853729,
      "grad_norm": 3.6474037170410156,
      "learning_rate": 2.0431492513282883e-07,
      "loss": 0.0239,
      "step": 5070
    },
    {
      "epoch": 0.7999369010529637,
      "grad_norm": 6.408823013305664,
      "learning_rate": 2.0415392046369343e-07,
      "loss": 0.0448,
      "step": 5071
    },
    {
      "epoch": 0.8000946484205544,
      "grad_norm": 3.4695098400115967,
      "learning_rate": 2.0399291579455804e-07,
      "loss": 0.0334,
      "step": 5072
    },
    {
      "epoch": 0.8002523957881453,
      "grad_norm": 5.335095405578613,
      "learning_rate": 2.0383191112542265e-07,
      "loss": 0.0565,
      "step": 5073
    },
    {
      "epoch": 0.8004101431557361,
      "grad_norm": 4.53596305847168,
      "learning_rate": 2.0367090645628723e-07,
      "loss": 0.0516,
      "step": 5074
    },
    {
      "epoch": 0.8005678905233269,
      "grad_norm": 6.096803188323975,
      "learning_rate": 2.0350990178715183e-07,
      "loss": 0.0783,
      "step": 5075
    },
    {
      "epoch": 0.8007256378909177,
      "grad_norm": 2.34279465675354,
      "learning_rate": 2.033488971180164e-07,
      "loss": 0.0354,
      "step": 5076
    },
    {
      "epoch": 0.8008833852585086,
      "grad_norm": 3.428112506866455,
      "learning_rate": 2.0318789244888102e-07,
      "loss": 0.0316,
      "step": 5077
    },
    {
      "epoch": 0.8010411326260993,
      "grad_norm": 4.4523115158081055,
      "learning_rate": 2.030268877797456e-07,
      "loss": 0.0419,
      "step": 5078
    },
    {
      "epoch": 0.8011988799936901,
      "grad_norm": 8.772183418273926,
      "learning_rate": 2.028658831106102e-07,
      "loss": 0.0461,
      "step": 5079
    },
    {
      "epoch": 0.8013566273612809,
      "grad_norm": 3.7308502197265625,
      "learning_rate": 2.0270487844147478e-07,
      "loss": 0.0295,
      "step": 5080
    },
    {
      "epoch": 0.8015143747288717,
      "grad_norm": 5.455733299255371,
      "learning_rate": 2.025438737723394e-07,
      "loss": 0.0342,
      "step": 5081
    },
    {
      "epoch": 0.8016721220964625,
      "grad_norm": 3.7311489582061768,
      "learning_rate": 2.02382869103204e-07,
      "loss": 0.0563,
      "step": 5082
    },
    {
      "epoch": 0.8018298694640533,
      "grad_norm": 3.39882755279541,
      "learning_rate": 2.0222186443406858e-07,
      "loss": 0.0284,
      "step": 5083
    },
    {
      "epoch": 0.8019876168316441,
      "grad_norm": 3.38927960395813,
      "learning_rate": 2.0206085976493318e-07,
      "loss": 0.0239,
      "step": 5084
    },
    {
      "epoch": 0.8021453641992349,
      "grad_norm": 4.0973005294799805,
      "learning_rate": 2.0189985509579776e-07,
      "loss": 0.0294,
      "step": 5085
    },
    {
      "epoch": 0.8023031115668258,
      "grad_norm": 3.903991937637329,
      "learning_rate": 2.0173885042666237e-07,
      "loss": 0.0475,
      "step": 5086
    },
    {
      "epoch": 0.8024608589344165,
      "grad_norm": 4.17649507522583,
      "learning_rate": 2.0157784575752695e-07,
      "loss": 0.0355,
      "step": 5087
    },
    {
      "epoch": 0.8026186063020073,
      "grad_norm": 7.29484748840332,
      "learning_rate": 2.0141684108839156e-07,
      "loss": 0.0483,
      "step": 5088
    },
    {
      "epoch": 0.8027763536695981,
      "grad_norm": 4.985876083374023,
      "learning_rate": 2.0125583641925614e-07,
      "loss": 0.0327,
      "step": 5089
    },
    {
      "epoch": 0.802934101037189,
      "grad_norm": 10.11892318725586,
      "learning_rate": 2.0109483175012074e-07,
      "loss": 0.0445,
      "step": 5090
    },
    {
      "epoch": 0.8030918484047798,
      "grad_norm": 3.6641905307769775,
      "learning_rate": 2.0093382708098532e-07,
      "loss": 0.0733,
      "step": 5091
    },
    {
      "epoch": 0.8032495957723705,
      "grad_norm": 5.7901930809021,
      "learning_rate": 2.0077282241184993e-07,
      "loss": 0.0476,
      "step": 5092
    },
    {
      "epoch": 0.8034073431399613,
      "grad_norm": 6.130947113037109,
      "learning_rate": 2.006118177427145e-07,
      "loss": 0.0747,
      "step": 5093
    },
    {
      "epoch": 0.8035650905075522,
      "grad_norm": 4.188258647918701,
      "learning_rate": 2.0045081307357911e-07,
      "loss": 0.0255,
      "step": 5094
    },
    {
      "epoch": 0.803722837875143,
      "grad_norm": 5.715501308441162,
      "learning_rate": 2.0028980840444372e-07,
      "loss": 0.0808,
      "step": 5095
    },
    {
      "epoch": 0.8038805852427338,
      "grad_norm": 2.4276111125946045,
      "learning_rate": 2.0012880373530833e-07,
      "loss": 0.0302,
      "step": 5096
    },
    {
      "epoch": 0.8040383326103245,
      "grad_norm": 7.273693084716797,
      "learning_rate": 1.9996779906617293e-07,
      "loss": 0.0404,
      "step": 5097
    },
    {
      "epoch": 0.8041960799779154,
      "grad_norm": 3.8964407444000244,
      "learning_rate": 1.998067943970375e-07,
      "loss": 0.0211,
      "step": 5098
    },
    {
      "epoch": 0.8043538273455062,
      "grad_norm": 3.150508165359497,
      "learning_rate": 1.9964578972790212e-07,
      "loss": 0.0354,
      "step": 5099
    },
    {
      "epoch": 0.804511574713097,
      "grad_norm": 3.0548505783081055,
      "learning_rate": 1.994847850587667e-07,
      "loss": 0.0541,
      "step": 5100
    },
    {
      "epoch": 0.8046693220806878,
      "grad_norm": 3.6085751056671143,
      "learning_rate": 1.993237803896313e-07,
      "loss": 0.0259,
      "step": 5101
    },
    {
      "epoch": 0.8048270694482785,
      "grad_norm": 3.171452283859253,
      "learning_rate": 1.9916277572049589e-07,
      "loss": 0.0194,
      "step": 5102
    },
    {
      "epoch": 0.8049848168158694,
      "grad_norm": 6.587282180786133,
      "learning_rate": 1.990017710513605e-07,
      "loss": 0.0498,
      "step": 5103
    },
    {
      "epoch": 0.8051425641834602,
      "grad_norm": 3.8433663845062256,
      "learning_rate": 1.9884076638222507e-07,
      "loss": 0.0298,
      "step": 5104
    },
    {
      "epoch": 0.805300311551051,
      "grad_norm": 6.593711853027344,
      "learning_rate": 1.9867976171308968e-07,
      "loss": 0.0736,
      "step": 5105
    },
    {
      "epoch": 0.8054580589186418,
      "grad_norm": 7.378602981567383,
      "learning_rate": 1.9851875704395426e-07,
      "loss": 0.03,
      "step": 5106
    },
    {
      "epoch": 0.8056158062862326,
      "grad_norm": 7.948828220367432,
      "learning_rate": 1.9835775237481886e-07,
      "loss": 0.0381,
      "step": 5107
    },
    {
      "epoch": 0.8057735536538234,
      "grad_norm": 4.1958465576171875,
      "learning_rate": 1.9819674770568344e-07,
      "loss": 0.0262,
      "step": 5108
    },
    {
      "epoch": 0.8059313010214142,
      "grad_norm": 4.811060428619385,
      "learning_rate": 1.9803574303654805e-07,
      "loss": 0.0777,
      "step": 5109
    },
    {
      "epoch": 0.806089048389005,
      "grad_norm": 3.1506600379943848,
      "learning_rate": 1.9787473836741263e-07,
      "loss": 0.0367,
      "step": 5110
    },
    {
      "epoch": 0.8062467957565959,
      "grad_norm": 3.457339286804199,
      "learning_rate": 1.9771373369827724e-07,
      "loss": 0.0443,
      "step": 5111
    },
    {
      "epoch": 0.8064045431241866,
      "grad_norm": 5.397216320037842,
      "learning_rate": 1.9755272902914182e-07,
      "loss": 0.0317,
      "step": 5112
    },
    {
      "epoch": 0.8065622904917774,
      "grad_norm": 3.1989307403564453,
      "learning_rate": 1.9739172436000642e-07,
      "loss": 0.0334,
      "step": 5113
    },
    {
      "epoch": 0.8067200378593682,
      "grad_norm": 3.706082582473755,
      "learning_rate": 1.9723071969087103e-07,
      "loss": 0.0301,
      "step": 5114
    },
    {
      "epoch": 0.8068777852269591,
      "grad_norm": 6.416573524475098,
      "learning_rate": 1.970697150217356e-07,
      "loss": 0.0246,
      "step": 5115
    },
    {
      "epoch": 0.8070355325945499,
      "grad_norm": 3.8669443130493164,
      "learning_rate": 1.9690871035260022e-07,
      "loss": 0.044,
      "step": 5116
    },
    {
      "epoch": 0.8071932799621406,
      "grad_norm": 4.944881439208984,
      "learning_rate": 1.967477056834648e-07,
      "loss": 0.089,
      "step": 5117
    },
    {
      "epoch": 0.8073510273297314,
      "grad_norm": 2.441859006881714,
      "learning_rate": 1.9658670101432943e-07,
      "loss": 0.0148,
      "step": 5118
    },
    {
      "epoch": 0.8075087746973222,
      "grad_norm": 6.643697261810303,
      "learning_rate": 1.96425696345194e-07,
      "loss": 0.0612,
      "step": 5119
    },
    {
      "epoch": 0.8076665220649131,
      "grad_norm": 3.9879868030548096,
      "learning_rate": 1.9626469167605861e-07,
      "loss": 0.0216,
      "step": 5120
    },
    {
      "epoch": 0.8078242694325038,
      "grad_norm": 5.297037601470947,
      "learning_rate": 1.961036870069232e-07,
      "loss": 0.0826,
      "step": 5121
    },
    {
      "epoch": 0.8079820168000946,
      "grad_norm": 4.819685935974121,
      "learning_rate": 1.959426823377878e-07,
      "loss": 0.0512,
      "step": 5122
    },
    {
      "epoch": 0.8081397641676854,
      "grad_norm": 5.230579853057861,
      "learning_rate": 1.9578167766865238e-07,
      "loss": 0.0202,
      "step": 5123
    },
    {
      "epoch": 0.8082975115352763,
      "grad_norm": 5.653816223144531,
      "learning_rate": 1.9562067299951699e-07,
      "loss": 0.0703,
      "step": 5124
    },
    {
      "epoch": 0.8084552589028671,
      "grad_norm": 4.329024791717529,
      "learning_rate": 1.9545966833038157e-07,
      "loss": 0.0398,
      "step": 5125
    },
    {
      "epoch": 0.8086130062704578,
      "grad_norm": 4.0327229499816895,
      "learning_rate": 1.9529866366124617e-07,
      "loss": 0.0449,
      "step": 5126
    },
    {
      "epoch": 0.8087707536380486,
      "grad_norm": 4.008519172668457,
      "learning_rate": 1.9513765899211075e-07,
      "loss": 0.0302,
      "step": 5127
    },
    {
      "epoch": 0.8089285010056395,
      "grad_norm": 5.778116703033447,
      "learning_rate": 1.9497665432297536e-07,
      "loss": 0.0319,
      "step": 5128
    },
    {
      "epoch": 0.8090862483732303,
      "grad_norm": 6.866086006164551,
      "learning_rate": 1.9481564965383996e-07,
      "loss": 0.0707,
      "step": 5129
    },
    {
      "epoch": 0.8092439957408211,
      "grad_norm": 5.5323686599731445,
      "learning_rate": 1.9465464498470454e-07,
      "loss": 0.0583,
      "step": 5130
    },
    {
      "epoch": 0.8094017431084118,
      "grad_norm": 5.858497619628906,
      "learning_rate": 1.9449364031556915e-07,
      "loss": 0.0484,
      "step": 5131
    },
    {
      "epoch": 0.8095594904760027,
      "grad_norm": 6.386364936828613,
      "learning_rate": 1.9433263564643373e-07,
      "loss": 0.0744,
      "step": 5132
    },
    {
      "epoch": 0.8097172378435935,
      "grad_norm": 2.88401460647583,
      "learning_rate": 1.9417163097729834e-07,
      "loss": 0.0156,
      "step": 5133
    },
    {
      "epoch": 0.8098749852111843,
      "grad_norm": 7.814047336578369,
      "learning_rate": 1.9401062630816292e-07,
      "loss": 0.0846,
      "step": 5134
    },
    {
      "epoch": 0.8100327325787751,
      "grad_norm": 5.588277339935303,
      "learning_rate": 1.9384962163902752e-07,
      "loss": 0.0675,
      "step": 5135
    },
    {
      "epoch": 0.810190479946366,
      "grad_norm": 6.536828994750977,
      "learning_rate": 1.936886169698921e-07,
      "loss": 0.0659,
      "step": 5136
    },
    {
      "epoch": 0.8103482273139567,
      "grad_norm": 6.740087985992432,
      "learning_rate": 1.935276123007567e-07,
      "loss": 0.0569,
      "step": 5137
    },
    {
      "epoch": 0.8105059746815475,
      "grad_norm": 5.166973114013672,
      "learning_rate": 1.933666076316213e-07,
      "loss": 0.025,
      "step": 5138
    },
    {
      "epoch": 0.8106637220491383,
      "grad_norm": 5.403120994567871,
      "learning_rate": 1.932056029624859e-07,
      "loss": 0.0587,
      "step": 5139
    },
    {
      "epoch": 0.8108214694167291,
      "grad_norm": 3.0564119815826416,
      "learning_rate": 1.9304459829335048e-07,
      "loss": 0.0176,
      "step": 5140
    },
    {
      "epoch": 0.81097921678432,
      "grad_norm": 7.829894065856934,
      "learning_rate": 1.928835936242151e-07,
      "loss": 0.0628,
      "step": 5141
    },
    {
      "epoch": 0.8111369641519107,
      "grad_norm": 5.746363639831543,
      "learning_rate": 1.927225889550797e-07,
      "loss": 0.0854,
      "step": 5142
    },
    {
      "epoch": 0.8112947115195015,
      "grad_norm": 4.095083713531494,
      "learning_rate": 1.925615842859443e-07,
      "loss": 0.054,
      "step": 5143
    },
    {
      "epoch": 0.8114524588870923,
      "grad_norm": 7.036856651306152,
      "learning_rate": 1.924005796168089e-07,
      "loss": 0.0686,
      "step": 5144
    },
    {
      "epoch": 0.8116102062546832,
      "grad_norm": 4.661142826080322,
      "learning_rate": 1.9223957494767348e-07,
      "loss": 0.0418,
      "step": 5145
    },
    {
      "epoch": 0.8117679536222739,
      "grad_norm": 10.97108268737793,
      "learning_rate": 1.920785702785381e-07,
      "loss": 0.0449,
      "step": 5146
    },
    {
      "epoch": 0.8119257009898647,
      "grad_norm": 6.629733085632324,
      "learning_rate": 1.9191756560940267e-07,
      "loss": 0.0372,
      "step": 5147
    },
    {
      "epoch": 0.8120834483574555,
      "grad_norm": 5.715473651885986,
      "learning_rate": 1.9175656094026727e-07,
      "loss": 0.0555,
      "step": 5148
    },
    {
      "epoch": 0.8122411957250464,
      "grad_norm": 4.085668087005615,
      "learning_rate": 1.9159555627113185e-07,
      "loss": 0.0411,
      "step": 5149
    },
    {
      "epoch": 0.8123989430926372,
      "grad_norm": 3.9186437129974365,
      "learning_rate": 1.9143455160199646e-07,
      "loss": 0.0254,
      "step": 5150
    },
    {
      "epoch": 0.8125566904602279,
      "grad_norm": 4.551998138427734,
      "learning_rate": 1.9127354693286104e-07,
      "loss": 0.047,
      "step": 5151
    },
    {
      "epoch": 0.8127144378278187,
      "grad_norm": 2.5676891803741455,
      "learning_rate": 1.9111254226372565e-07,
      "loss": 0.0142,
      "step": 5152
    },
    {
      "epoch": 0.8128721851954096,
      "grad_norm": 3.801438808441162,
      "learning_rate": 1.9095153759459023e-07,
      "loss": 0.0271,
      "step": 5153
    },
    {
      "epoch": 0.8130299325630004,
      "grad_norm": 6.762763023376465,
      "learning_rate": 1.9079053292545483e-07,
      "loss": 0.0453,
      "step": 5154
    },
    {
      "epoch": 0.8131876799305912,
      "grad_norm": 4.866330146789551,
      "learning_rate": 1.906295282563194e-07,
      "loss": 0.0637,
      "step": 5155
    },
    {
      "epoch": 0.8133454272981819,
      "grad_norm": 3.441608190536499,
      "learning_rate": 1.9046852358718402e-07,
      "loss": 0.0299,
      "step": 5156
    },
    {
      "epoch": 0.8135031746657727,
      "grad_norm": 5.448775291442871,
      "learning_rate": 1.903075189180486e-07,
      "loss": 0.0421,
      "step": 5157
    },
    {
      "epoch": 0.8136609220333636,
      "grad_norm": 4.715697765350342,
      "learning_rate": 1.901465142489132e-07,
      "loss": 0.0696,
      "step": 5158
    },
    {
      "epoch": 0.8138186694009544,
      "grad_norm": 3.6033568382263184,
      "learning_rate": 1.899855095797778e-07,
      "loss": 0.039,
      "step": 5159
    },
    {
      "epoch": 0.8139764167685452,
      "grad_norm": 6.063899993896484,
      "learning_rate": 1.898245049106424e-07,
      "loss": 0.04,
      "step": 5160
    },
    {
      "epoch": 0.8141341641361359,
      "grad_norm": 7.023772239685059,
      "learning_rate": 1.89663500241507e-07,
      "loss": 0.0609,
      "step": 5161
    },
    {
      "epoch": 0.8142919115037268,
      "grad_norm": 4.2832441329956055,
      "learning_rate": 1.8950249557237158e-07,
      "loss": 0.0452,
      "step": 5162
    },
    {
      "epoch": 0.8144496588713176,
      "grad_norm": 3.4046990871429443,
      "learning_rate": 1.8934149090323618e-07,
      "loss": 0.0176,
      "step": 5163
    },
    {
      "epoch": 0.8146074062389084,
      "grad_norm": 3.943274974822998,
      "learning_rate": 1.8918048623410076e-07,
      "loss": 0.0433,
      "step": 5164
    },
    {
      "epoch": 0.8147651536064991,
      "grad_norm": 5.466538906097412,
      "learning_rate": 1.890194815649654e-07,
      "loss": 0.0649,
      "step": 5165
    },
    {
      "epoch": 0.81492290097409,
      "grad_norm": 4.183228015899658,
      "learning_rate": 1.8885847689582997e-07,
      "loss": 0.0194,
      "step": 5166
    },
    {
      "epoch": 0.8150806483416808,
      "grad_norm": 7.976485729217529,
      "learning_rate": 1.8869747222669458e-07,
      "loss": 0.0629,
      "step": 5167
    },
    {
      "epoch": 0.8152383957092716,
      "grad_norm": 7.346073627471924,
      "learning_rate": 1.8853646755755916e-07,
      "loss": 0.0646,
      "step": 5168
    },
    {
      "epoch": 0.8153961430768624,
      "grad_norm": 3.0371408462524414,
      "learning_rate": 1.8837546288842377e-07,
      "loss": 0.0143,
      "step": 5169
    },
    {
      "epoch": 0.8155538904444533,
      "grad_norm": 3.4316720962524414,
      "learning_rate": 1.8821445821928835e-07,
      "loss": 0.02,
      "step": 5170
    },
    {
      "epoch": 0.815711637812044,
      "grad_norm": 6.701343536376953,
      "learning_rate": 1.8805345355015295e-07,
      "loss": 0.0246,
      "step": 5171
    },
    {
      "epoch": 0.8158693851796348,
      "grad_norm": 5.175650119781494,
      "learning_rate": 1.8789244888101753e-07,
      "loss": 0.0418,
      "step": 5172
    },
    {
      "epoch": 0.8160271325472256,
      "grad_norm": 5.159764766693115,
      "learning_rate": 1.8773144421188214e-07,
      "loss": 0.0232,
      "step": 5173
    },
    {
      "epoch": 0.8161848799148165,
      "grad_norm": 3.392036199569702,
      "learning_rate": 1.8757043954274675e-07,
      "loss": 0.0177,
      "step": 5174
    },
    {
      "epoch": 0.8163426272824073,
      "grad_norm": 3.4437549114227295,
      "learning_rate": 1.8740943487361133e-07,
      "loss": 0.0169,
      "step": 5175
    },
    {
      "epoch": 0.816500374649998,
      "grad_norm": 5.620139122009277,
      "learning_rate": 1.8724843020447593e-07,
      "loss": 0.0358,
      "step": 5176
    },
    {
      "epoch": 0.8166581220175888,
      "grad_norm": 4.827464580535889,
      "learning_rate": 1.870874255353405e-07,
      "loss": 0.0136,
      "step": 5177
    },
    {
      "epoch": 0.8168158693851796,
      "grad_norm": 5.521566390991211,
      "learning_rate": 1.8692642086620512e-07,
      "loss": 0.05,
      "step": 5178
    },
    {
      "epoch": 0.8169736167527705,
      "grad_norm": 7.122973442077637,
      "learning_rate": 1.867654161970697e-07,
      "loss": 0.0473,
      "step": 5179
    },
    {
      "epoch": 0.8171313641203612,
      "grad_norm": 5.359594345092773,
      "learning_rate": 1.866044115279343e-07,
      "loss": 0.0563,
      "step": 5180
    },
    {
      "epoch": 0.817289111487952,
      "grad_norm": 2.6821999549865723,
      "learning_rate": 1.8644340685879888e-07,
      "loss": 0.0166,
      "step": 5181
    },
    {
      "epoch": 0.8174468588555428,
      "grad_norm": 4.8910322189331055,
      "learning_rate": 1.862824021896635e-07,
      "loss": 0.0386,
      "step": 5182
    },
    {
      "epoch": 0.8176046062231337,
      "grad_norm": 3.7409276962280273,
      "learning_rate": 1.8612139752052807e-07,
      "loss": 0.0184,
      "step": 5183
    },
    {
      "epoch": 0.8177623535907245,
      "grad_norm": 6.966063022613525,
      "learning_rate": 1.8596039285139268e-07,
      "loss": 0.0756,
      "step": 5184
    },
    {
      "epoch": 0.8179201009583152,
      "grad_norm": 7.367618560791016,
      "learning_rate": 1.8579938818225726e-07,
      "loss": 0.045,
      "step": 5185
    },
    {
      "epoch": 0.818077848325906,
      "grad_norm": 5.712251663208008,
      "learning_rate": 1.8563838351312186e-07,
      "loss": 0.0427,
      "step": 5186
    },
    {
      "epoch": 0.8182355956934969,
      "grad_norm": 2.799457550048828,
      "learning_rate": 1.8547737884398644e-07,
      "loss": 0.0176,
      "step": 5187
    },
    {
      "epoch": 0.8183933430610877,
      "grad_norm": 5.2669782638549805,
      "learning_rate": 1.8531637417485108e-07,
      "loss": 0.0583,
      "step": 5188
    },
    {
      "epoch": 0.8185510904286785,
      "grad_norm": 1.8714468479156494,
      "learning_rate": 1.8515536950571568e-07,
      "loss": 0.0123,
      "step": 5189
    },
    {
      "epoch": 0.8187088377962692,
      "grad_norm": 3.939018964767456,
      "learning_rate": 1.8499436483658026e-07,
      "loss": 0.0358,
      "step": 5190
    },
    {
      "epoch": 0.8188665851638601,
      "grad_norm": 6.186793327331543,
      "learning_rate": 1.8483336016744487e-07,
      "loss": 0.0285,
      "step": 5191
    },
    {
      "epoch": 0.8190243325314509,
      "grad_norm": 2.6424143314361572,
      "learning_rate": 1.8467235549830945e-07,
      "loss": 0.0173,
      "step": 5192
    },
    {
      "epoch": 0.8191820798990417,
      "grad_norm": 4.678024768829346,
      "learning_rate": 1.8451135082917405e-07,
      "loss": 0.0517,
      "step": 5193
    },
    {
      "epoch": 0.8193398272666325,
      "grad_norm": 4.546450138092041,
      "learning_rate": 1.8435034616003863e-07,
      "loss": 0.0284,
      "step": 5194
    },
    {
      "epoch": 0.8194975746342232,
      "grad_norm": 5.08975887298584,
      "learning_rate": 1.8418934149090324e-07,
      "loss": 0.0222,
      "step": 5195
    },
    {
      "epoch": 0.8196553220018141,
      "grad_norm": 7.609801769256592,
      "learning_rate": 1.8402833682176782e-07,
      "loss": 0.0459,
      "step": 5196
    },
    {
      "epoch": 0.8198130693694049,
      "grad_norm": 4.874221324920654,
      "learning_rate": 1.8386733215263243e-07,
      "loss": 0.051,
      "step": 5197
    },
    {
      "epoch": 0.8199708167369957,
      "grad_norm": 6.313220500946045,
      "learning_rate": 1.83706327483497e-07,
      "loss": 0.0341,
      "step": 5198
    },
    {
      "epoch": 0.8201285641045865,
      "grad_norm": 4.605165481567383,
      "learning_rate": 1.835453228143616e-07,
      "loss": 0.0263,
      "step": 5199
    },
    {
      "epoch": 0.8202863114721773,
      "grad_norm": 2.869143009185791,
      "learning_rate": 1.833843181452262e-07,
      "loss": 0.0208,
      "step": 5200
    },
    {
      "epoch": 0.8204440588397681,
      "grad_norm": 4.082322120666504,
      "learning_rate": 1.832233134760908e-07,
      "loss": 0.0386,
      "step": 5201
    },
    {
      "epoch": 0.8206018062073589,
      "grad_norm": 7.996222496032715,
      "learning_rate": 1.8306230880695538e-07,
      "loss": 0.0393,
      "step": 5202
    },
    {
      "epoch": 0.8207595535749497,
      "grad_norm": 2.8665354251861572,
      "learning_rate": 1.8290130413781999e-07,
      "loss": 0.0491,
      "step": 5203
    },
    {
      "epoch": 0.8209173009425406,
      "grad_norm": 5.036356449127197,
      "learning_rate": 1.8274029946868456e-07,
      "loss": 0.0599,
      "step": 5204
    },
    {
      "epoch": 0.8210750483101313,
      "grad_norm": 4.692917823791504,
      "learning_rate": 1.8257929479954917e-07,
      "loss": 0.0389,
      "step": 5205
    },
    {
      "epoch": 0.8212327956777221,
      "grad_norm": 3.754473924636841,
      "learning_rate": 1.8241829013041378e-07,
      "loss": 0.0729,
      "step": 5206
    },
    {
      "epoch": 0.8213905430453129,
      "grad_norm": 6.935403823852539,
      "learning_rate": 1.8225728546127836e-07,
      "loss": 0.0671,
      "step": 5207
    },
    {
      "epoch": 0.8215482904129038,
      "grad_norm": 10.570488929748535,
      "learning_rate": 1.8209628079214296e-07,
      "loss": 0.0494,
      "step": 5208
    },
    {
      "epoch": 0.8217060377804946,
      "grad_norm": 5.963524341583252,
      "learning_rate": 1.8193527612300754e-07,
      "loss": 0.0522,
      "step": 5209
    },
    {
      "epoch": 0.8218637851480853,
      "grad_norm": 5.608829498291016,
      "learning_rate": 1.8177427145387215e-07,
      "loss": 0.0384,
      "step": 5210
    },
    {
      "epoch": 0.8220215325156761,
      "grad_norm": 6.981807708740234,
      "learning_rate": 1.8161326678473676e-07,
      "loss": 0.0441,
      "step": 5211
    },
    {
      "epoch": 0.822179279883267,
      "grad_norm": 5.255591869354248,
      "learning_rate": 1.8145226211560136e-07,
      "loss": 0.0339,
      "step": 5212
    },
    {
      "epoch": 0.8223370272508578,
      "grad_norm": 4.443271160125732,
      "learning_rate": 1.8129125744646594e-07,
      "loss": 0.0226,
      "step": 5213
    },
    {
      "epoch": 0.8224947746184486,
      "grad_norm": 1.6246569156646729,
      "learning_rate": 1.8113025277733055e-07,
      "loss": 0.0107,
      "step": 5214
    },
    {
      "epoch": 0.8226525219860393,
      "grad_norm": 2.8058571815490723,
      "learning_rate": 1.8096924810819513e-07,
      "loss": 0.0123,
      "step": 5215
    },
    {
      "epoch": 0.8228102693536301,
      "grad_norm": 3.151439666748047,
      "learning_rate": 1.8080824343905973e-07,
      "loss": 0.0323,
      "step": 5216
    },
    {
      "epoch": 0.822968016721221,
      "grad_norm": 2.786860942840576,
      "learning_rate": 1.8064723876992431e-07,
      "loss": 0.0376,
      "step": 5217
    },
    {
      "epoch": 0.8231257640888118,
      "grad_norm": 6.642549991607666,
      "learning_rate": 1.8048623410078892e-07,
      "loss": 0.0361,
      "step": 5218
    },
    {
      "epoch": 0.8232835114564026,
      "grad_norm": 5.647134304046631,
      "learning_rate": 1.803252294316535e-07,
      "loss": 0.0205,
      "step": 5219
    },
    {
      "epoch": 0.8234412588239933,
      "grad_norm": 2.1456258296966553,
      "learning_rate": 1.801642247625181e-07,
      "loss": 0.013,
      "step": 5220
    },
    {
      "epoch": 0.8235990061915842,
      "grad_norm": 3.01601243019104,
      "learning_rate": 1.8000322009338271e-07,
      "loss": 0.0324,
      "step": 5221
    },
    {
      "epoch": 0.823756753559175,
      "grad_norm": 4.266296863555908,
      "learning_rate": 1.798422154242473e-07,
      "loss": 0.0257,
      "step": 5222
    },
    {
      "epoch": 0.8239145009267658,
      "grad_norm": 7.477838039398193,
      "learning_rate": 1.796812107551119e-07,
      "loss": 0.0249,
      "step": 5223
    },
    {
      "epoch": 0.8240722482943565,
      "grad_norm": 4.718059062957764,
      "learning_rate": 1.7952020608597648e-07,
      "loss": 0.0673,
      "step": 5224
    },
    {
      "epoch": 0.8242299956619474,
      "grad_norm": 5.125923156738281,
      "learning_rate": 1.7935920141684109e-07,
      "loss": 0.0544,
      "step": 5225
    },
    {
      "epoch": 0.8243877430295382,
      "grad_norm": 4.098823547363281,
      "learning_rate": 1.7919819674770567e-07,
      "loss": 0.0362,
      "step": 5226
    },
    {
      "epoch": 0.824545490397129,
      "grad_norm": 2.892537832260132,
      "learning_rate": 1.7903719207857027e-07,
      "loss": 0.0105,
      "step": 5227
    },
    {
      "epoch": 0.8247032377647198,
      "grad_norm": 5.221391201019287,
      "learning_rate": 1.7887618740943485e-07,
      "loss": 0.0258,
      "step": 5228
    },
    {
      "epoch": 0.8248609851323107,
      "grad_norm": 6.827422618865967,
      "learning_rate": 1.7871518274029946e-07,
      "loss": 0.0796,
      "step": 5229
    },
    {
      "epoch": 0.8250187324999014,
      "grad_norm": 4.7586164474487305,
      "learning_rate": 1.7855417807116404e-07,
      "loss": 0.0388,
      "step": 5230
    },
    {
      "epoch": 0.8251764798674922,
      "grad_norm": 6.9503583908081055,
      "learning_rate": 1.7839317340202864e-07,
      "loss": 0.048,
      "step": 5231
    },
    {
      "epoch": 0.825334227235083,
      "grad_norm": 4.58003044128418,
      "learning_rate": 1.7823216873289322e-07,
      "loss": 0.0331,
      "step": 5232
    },
    {
      "epoch": 0.8254919746026739,
      "grad_norm": 5.491656303405762,
      "learning_rate": 1.7807116406375783e-07,
      "loss": 0.0615,
      "step": 5233
    },
    {
      "epoch": 0.8256497219702646,
      "grad_norm": 6.24632453918457,
      "learning_rate": 1.779101593946224e-07,
      "loss": 0.0404,
      "step": 5234
    },
    {
      "epoch": 0.8258074693378554,
      "grad_norm": 5.307912349700928,
      "learning_rate": 1.7774915472548704e-07,
      "loss": 0.0657,
      "step": 5235
    },
    {
      "epoch": 0.8259652167054462,
      "grad_norm": 3.9404730796813965,
      "learning_rate": 1.7758815005635165e-07,
      "loss": 0.0289,
      "step": 5236
    },
    {
      "epoch": 0.826122964073037,
      "grad_norm": 6.570104122161865,
      "learning_rate": 1.7742714538721623e-07,
      "loss": 0.0791,
      "step": 5237
    },
    {
      "epoch": 0.8262807114406279,
      "grad_norm": 4.156282901763916,
      "learning_rate": 1.7726614071808084e-07,
      "loss": 0.0244,
      "step": 5238
    },
    {
      "epoch": 0.8264384588082186,
      "grad_norm": 6.332943439483643,
      "learning_rate": 1.7710513604894542e-07,
      "loss": 0.0629,
      "step": 5239
    },
    {
      "epoch": 0.8265962061758094,
      "grad_norm": 6.7248616218566895,
      "learning_rate": 1.7694413137981002e-07,
      "loss": 0.0543,
      "step": 5240
    },
    {
      "epoch": 0.8267539535434002,
      "grad_norm": 4.556899070739746,
      "learning_rate": 1.767831267106746e-07,
      "loss": 0.0377,
      "step": 5241
    },
    {
      "epoch": 0.8269117009109911,
      "grad_norm": 5.604135513305664,
      "learning_rate": 1.766221220415392e-07,
      "loss": 0.0738,
      "step": 5242
    },
    {
      "epoch": 0.8270694482785819,
      "grad_norm": 2.578310012817383,
      "learning_rate": 1.764611173724038e-07,
      "loss": 0.0125,
      "step": 5243
    },
    {
      "epoch": 0.8272271956461726,
      "grad_norm": 4.124502182006836,
      "learning_rate": 1.763001127032684e-07,
      "loss": 0.0156,
      "step": 5244
    },
    {
      "epoch": 0.8273849430137634,
      "grad_norm": 3.3038899898529053,
      "learning_rate": 1.7613910803413297e-07,
      "loss": 0.019,
      "step": 5245
    },
    {
      "epoch": 0.8275426903813543,
      "grad_norm": 6.650944232940674,
      "learning_rate": 1.7597810336499758e-07,
      "loss": 0.0192,
      "step": 5246
    },
    {
      "epoch": 0.8277004377489451,
      "grad_norm": 7.623540878295898,
      "learning_rate": 1.7581709869586216e-07,
      "loss": 0.0686,
      "step": 5247
    },
    {
      "epoch": 0.8278581851165359,
      "grad_norm": 0.5060501098632812,
      "learning_rate": 1.7565609402672677e-07,
      "loss": 0.0036,
      "step": 5248
    },
    {
      "epoch": 0.8280159324841266,
      "grad_norm": 7.243015289306641,
      "learning_rate": 1.7549508935759135e-07,
      "loss": 0.0171,
      "step": 5249
    },
    {
      "epoch": 0.8281736798517175,
      "grad_norm": 5.172285556793213,
      "learning_rate": 1.7533408468845595e-07,
      "loss": 0.0337,
      "step": 5250
    },
    {
      "epoch": 0.8283314272193083,
      "grad_norm": 2.3561525344848633,
      "learning_rate": 1.7517308001932056e-07,
      "loss": 0.0355,
      "step": 5251
    },
    {
      "epoch": 0.8284891745868991,
      "grad_norm": 6.328184127807617,
      "learning_rate": 1.7501207535018514e-07,
      "loss": 0.0404,
      "step": 5252
    },
    {
      "epoch": 0.8286469219544899,
      "grad_norm": 3.560062885284424,
      "learning_rate": 1.7485107068104974e-07,
      "loss": 0.0185,
      "step": 5253
    },
    {
      "epoch": 0.8288046693220806,
      "grad_norm": 6.333250999450684,
      "learning_rate": 1.7469006601191432e-07,
      "loss": 0.0203,
      "step": 5254
    },
    {
      "epoch": 0.8289624166896715,
      "grad_norm": 4.132623672485352,
      "learning_rate": 1.7452906134277893e-07,
      "loss": 0.0472,
      "step": 5255
    },
    {
      "epoch": 0.8291201640572623,
      "grad_norm": 3.312601089477539,
      "learning_rate": 1.743680566736435e-07,
      "loss": 0.0277,
      "step": 5256
    },
    {
      "epoch": 0.8292779114248531,
      "grad_norm": 2.555586814880371,
      "learning_rate": 1.7420705200450812e-07,
      "loss": 0.0375,
      "step": 5257
    },
    {
      "epoch": 0.8294356587924439,
      "grad_norm": 5.3458476066589355,
      "learning_rate": 1.7404604733537272e-07,
      "loss": 0.0582,
      "step": 5258
    },
    {
      "epoch": 0.8295934061600347,
      "grad_norm": 4.48042106628418,
      "learning_rate": 1.7388504266623733e-07,
      "loss": 0.0634,
      "step": 5259
    },
    {
      "epoch": 0.8297511535276255,
      "grad_norm": 3.472592830657959,
      "learning_rate": 1.737240379971019e-07,
      "loss": 0.0282,
      "step": 5260
    },
    {
      "epoch": 0.8299089008952163,
      "grad_norm": 5.665646553039551,
      "learning_rate": 1.7356303332796652e-07,
      "loss": 0.0448,
      "step": 5261
    },
    {
      "epoch": 0.8300666482628071,
      "grad_norm": 3.956422805786133,
      "learning_rate": 1.734020286588311e-07,
      "loss": 0.0191,
      "step": 5262
    },
    {
      "epoch": 0.830224395630398,
      "grad_norm": 3.4300477504730225,
      "learning_rate": 1.732410239896957e-07,
      "loss": 0.0185,
      "step": 5263
    },
    {
      "epoch": 0.8303821429979887,
      "grad_norm": 5.574986934661865,
      "learning_rate": 1.7308001932056028e-07,
      "loss": 0.0243,
      "step": 5264
    },
    {
      "epoch": 0.8305398903655795,
      "grad_norm": 16.364606857299805,
      "learning_rate": 1.729190146514249e-07,
      "loss": 0.034,
      "step": 5265
    },
    {
      "epoch": 0.8306976377331703,
      "grad_norm": 2.714670419692993,
      "learning_rate": 1.727580099822895e-07,
      "loss": 0.0267,
      "step": 5266
    },
    {
      "epoch": 0.8308553851007612,
      "grad_norm": 4.444028377532959,
      "learning_rate": 1.7259700531315407e-07,
      "loss": 0.0221,
      "step": 5267
    },
    {
      "epoch": 0.831013132468352,
      "grad_norm": 7.142070770263672,
      "learning_rate": 1.7243600064401868e-07,
      "loss": 0.031,
      "step": 5268
    },
    {
      "epoch": 0.8311708798359427,
      "grad_norm": 6.350658893585205,
      "learning_rate": 1.7227499597488326e-07,
      "loss": 0.0409,
      "step": 5269
    },
    {
      "epoch": 0.8313286272035335,
      "grad_norm": 4.477631568908691,
      "learning_rate": 1.7211399130574787e-07,
      "loss": 0.0261,
      "step": 5270
    },
    {
      "epoch": 0.8314863745711244,
      "grad_norm": 4.212431907653809,
      "learning_rate": 1.7195298663661245e-07,
      "loss": 0.0339,
      "step": 5271
    },
    {
      "epoch": 0.8316441219387152,
      "grad_norm": 4.496842861175537,
      "learning_rate": 1.7179198196747705e-07,
      "loss": 0.0229,
      "step": 5272
    },
    {
      "epoch": 0.831801869306306,
      "grad_norm": 3.2212107181549072,
      "learning_rate": 1.7163097729834163e-07,
      "loss": 0.0382,
      "step": 5273
    },
    {
      "epoch": 0.8319596166738967,
      "grad_norm": 5.017549514770508,
      "learning_rate": 1.7146997262920624e-07,
      "loss": 0.0323,
      "step": 5274
    },
    {
      "epoch": 0.8321173640414875,
      "grad_norm": 3.7942099571228027,
      "learning_rate": 1.7130896796007082e-07,
      "loss": 0.0189,
      "step": 5275
    },
    {
      "epoch": 0.8322751114090784,
      "grad_norm": 6.848123550415039,
      "learning_rate": 1.7114796329093543e-07,
      "loss": 0.1382,
      "step": 5276
    },
    {
      "epoch": 0.8324328587766692,
      "grad_norm": 3.2568774223327637,
      "learning_rate": 1.709869586218e-07,
      "loss": 0.0191,
      "step": 5277
    },
    {
      "epoch": 0.83259060614426,
      "grad_norm": 3.9052367210388184,
      "learning_rate": 1.708259539526646e-07,
      "loss": 0.0481,
      "step": 5278
    },
    {
      "epoch": 0.8327483535118507,
      "grad_norm": 5.097133636474609,
      "learning_rate": 1.706649492835292e-07,
      "loss": 0.0609,
      "step": 5279
    },
    {
      "epoch": 0.8329061008794416,
      "grad_norm": 4.586767196655273,
      "learning_rate": 1.705039446143938e-07,
      "loss": 0.0763,
      "step": 5280
    },
    {
      "epoch": 0.8330638482470324,
      "grad_norm": 4.707311630249023,
      "learning_rate": 1.7034293994525843e-07,
      "loss": 0.0413,
      "step": 5281
    },
    {
      "epoch": 0.8332215956146232,
      "grad_norm": 5.52780818939209,
      "learning_rate": 1.70181935276123e-07,
      "loss": 0.0304,
      "step": 5282
    },
    {
      "epoch": 0.8333793429822139,
      "grad_norm": 4.097690105438232,
      "learning_rate": 1.7002093060698762e-07,
      "loss": 0.0355,
      "step": 5283
    },
    {
      "epoch": 0.8335370903498048,
      "grad_norm": 4.840817451477051,
      "learning_rate": 1.698599259378522e-07,
      "loss": 0.0417,
      "step": 5284
    },
    {
      "epoch": 0.8336948377173956,
      "grad_norm": 5.8652825355529785,
      "learning_rate": 1.696989212687168e-07,
      "loss": 0.0465,
      "step": 5285
    },
    {
      "epoch": 0.8338525850849864,
      "grad_norm": 4.693187236785889,
      "learning_rate": 1.6953791659958138e-07,
      "loss": 0.0599,
      "step": 5286
    },
    {
      "epoch": 0.8340103324525772,
      "grad_norm": 4.07180643081665,
      "learning_rate": 1.69376911930446e-07,
      "loss": 0.029,
      "step": 5287
    },
    {
      "epoch": 0.834168079820168,
      "grad_norm": 4.234261512756348,
      "learning_rate": 1.6921590726131057e-07,
      "loss": 0.0163,
      "step": 5288
    },
    {
      "epoch": 0.8343258271877588,
      "grad_norm": 3.8681535720825195,
      "learning_rate": 1.6905490259217518e-07,
      "loss": 0.0309,
      "step": 5289
    },
    {
      "epoch": 0.8344835745553496,
      "grad_norm": 6.554484844207764,
      "learning_rate": 1.6889389792303976e-07,
      "loss": 0.0334,
      "step": 5290
    },
    {
      "epoch": 0.8346413219229404,
      "grad_norm": 4.8437347412109375,
      "learning_rate": 1.6873289325390436e-07,
      "loss": 0.0338,
      "step": 5291
    },
    {
      "epoch": 0.8347990692905312,
      "grad_norm": 5.575812816619873,
      "learning_rate": 1.6857188858476894e-07,
      "loss": 0.0506,
      "step": 5292
    },
    {
      "epoch": 0.834956816658122,
      "grad_norm": 4.8221282958984375,
      "learning_rate": 1.6841088391563355e-07,
      "loss": 0.051,
      "step": 5293
    },
    {
      "epoch": 0.8351145640257128,
      "grad_norm": 5.950834274291992,
      "learning_rate": 1.6824987924649813e-07,
      "loss": 0.045,
      "step": 5294
    },
    {
      "epoch": 0.8352723113933036,
      "grad_norm": 7.225146293640137,
      "learning_rate": 1.6808887457736273e-07,
      "loss": 0.0799,
      "step": 5295
    },
    {
      "epoch": 0.8354300587608944,
      "grad_norm": 7.208869934082031,
      "learning_rate": 1.6792786990822731e-07,
      "loss": 0.0418,
      "step": 5296
    },
    {
      "epoch": 0.8355878061284853,
      "grad_norm": 5.060133934020996,
      "learning_rate": 1.6776686523909192e-07,
      "loss": 0.0589,
      "step": 5297
    },
    {
      "epoch": 0.835745553496076,
      "grad_norm": 4.54489278793335,
      "learning_rate": 1.6760586056995653e-07,
      "loss": 0.0413,
      "step": 5298
    },
    {
      "epoch": 0.8359033008636668,
      "grad_norm": 3.2688772678375244,
      "learning_rate": 1.674448559008211e-07,
      "loss": 0.0497,
      "step": 5299
    },
    {
      "epoch": 0.8360610482312576,
      "grad_norm": 5.300332069396973,
      "learning_rate": 1.672838512316857e-07,
      "loss": 0.0306,
      "step": 5300
    },
    {
      "epoch": 0.8362187955988485,
      "grad_norm": 3.340242862701416,
      "learning_rate": 1.671228465625503e-07,
      "loss": 0.0298,
      "step": 5301
    },
    {
      "epoch": 0.8363765429664393,
      "grad_norm": 6.670624256134033,
      "learning_rate": 1.669618418934149e-07,
      "loss": 0.0678,
      "step": 5302
    },
    {
      "epoch": 0.83653429033403,
      "grad_norm": 6.236144542694092,
      "learning_rate": 1.6680083722427948e-07,
      "loss": 0.0304,
      "step": 5303
    },
    {
      "epoch": 0.8366920377016208,
      "grad_norm": 6.884690761566162,
      "learning_rate": 1.666398325551441e-07,
      "loss": 0.063,
      "step": 5304
    },
    {
      "epoch": 0.8368497850692117,
      "grad_norm": 6.910121917724609,
      "learning_rate": 1.664788278860087e-07,
      "loss": 0.0861,
      "step": 5305
    },
    {
      "epoch": 0.8370075324368025,
      "grad_norm": 5.235419273376465,
      "learning_rate": 1.663178232168733e-07,
      "loss": 0.0339,
      "step": 5306
    },
    {
      "epoch": 0.8371652798043933,
      "grad_norm": 3.1473639011383057,
      "learning_rate": 1.6615681854773788e-07,
      "loss": 0.0244,
      "step": 5307
    },
    {
      "epoch": 0.837323027171984,
      "grad_norm": 3.6600468158721924,
      "learning_rate": 1.6599581387860248e-07,
      "loss": 0.0158,
      "step": 5308
    },
    {
      "epoch": 0.8374807745395749,
      "grad_norm": 2.8669915199279785,
      "learning_rate": 1.6583480920946706e-07,
      "loss": 0.0271,
      "step": 5309
    },
    {
      "epoch": 0.8376385219071657,
      "grad_norm": 2.6794514656066895,
      "learning_rate": 1.6567380454033167e-07,
      "loss": 0.0231,
      "step": 5310
    },
    {
      "epoch": 0.8377962692747565,
      "grad_norm": 4.003549098968506,
      "learning_rate": 1.6551279987119625e-07,
      "loss": 0.0417,
      "step": 5311
    },
    {
      "epoch": 0.8379540166423473,
      "grad_norm": 4.020390033721924,
      "learning_rate": 1.6535179520206086e-07,
      "loss": 0.0539,
      "step": 5312
    },
    {
      "epoch": 0.838111764009938,
      "grad_norm": 16.512651443481445,
      "learning_rate": 1.6519079053292546e-07,
      "loss": 0.0534,
      "step": 5313
    },
    {
      "epoch": 0.8382695113775289,
      "grad_norm": 4.025285243988037,
      "learning_rate": 1.6502978586379004e-07,
      "loss": 0.0322,
      "step": 5314
    },
    {
      "epoch": 0.8384272587451197,
      "grad_norm": 5.083260536193848,
      "learning_rate": 1.6486878119465465e-07,
      "loss": 0.0327,
      "step": 5315
    },
    {
      "epoch": 0.8385850061127105,
      "grad_norm": 6.532556056976318,
      "learning_rate": 1.6470777652551923e-07,
      "loss": 0.0248,
      "step": 5316
    },
    {
      "epoch": 0.8387427534803013,
      "grad_norm": 4.935925006866455,
      "learning_rate": 1.6454677185638383e-07,
      "loss": 0.0638,
      "step": 5317
    },
    {
      "epoch": 0.8389005008478921,
      "grad_norm": 3.878962516784668,
      "learning_rate": 1.6438576718724841e-07,
      "loss": 0.0259,
      "step": 5318
    },
    {
      "epoch": 0.8390582482154829,
      "grad_norm": 1.2626736164093018,
      "learning_rate": 1.6422476251811302e-07,
      "loss": 0.0073,
      "step": 5319
    },
    {
      "epoch": 0.8392159955830737,
      "grad_norm": 2.795637369155884,
      "learning_rate": 1.640637578489776e-07,
      "loss": 0.0282,
      "step": 5320
    },
    {
      "epoch": 0.8393737429506645,
      "grad_norm": 11.341633796691895,
      "learning_rate": 1.639027531798422e-07,
      "loss": 0.0723,
      "step": 5321
    },
    {
      "epoch": 0.8395314903182554,
      "grad_norm": 5.826061725616455,
      "learning_rate": 1.6374174851070679e-07,
      "loss": 0.0436,
      "step": 5322
    },
    {
      "epoch": 0.8396892376858461,
      "grad_norm": 4.152256965637207,
      "learning_rate": 1.635807438415714e-07,
      "loss": 0.0494,
      "step": 5323
    },
    {
      "epoch": 0.8398469850534369,
      "grad_norm": 4.842311859130859,
      "learning_rate": 1.6341973917243597e-07,
      "loss": 0.0235,
      "step": 5324
    },
    {
      "epoch": 0.8400047324210277,
      "grad_norm": 5.246630668640137,
      "learning_rate": 1.6325873450330058e-07,
      "loss": 0.0339,
      "step": 5325
    },
    {
      "epoch": 0.8401624797886186,
      "grad_norm": 5.315804481506348,
      "learning_rate": 1.6309772983416516e-07,
      "loss": 0.0516,
      "step": 5326
    },
    {
      "epoch": 0.8403202271562094,
      "grad_norm": 4.90289831161499,
      "learning_rate": 1.6293672516502977e-07,
      "loss": 0.0272,
      "step": 5327
    },
    {
      "epoch": 0.8404779745238001,
      "grad_norm": 9.051569938659668,
      "learning_rate": 1.627757204958944e-07,
      "loss": 0.0263,
      "step": 5328
    },
    {
      "epoch": 0.8406357218913909,
      "grad_norm": 7.007421970367432,
      "learning_rate": 1.6261471582675898e-07,
      "loss": 0.0619,
      "step": 5329
    },
    {
      "epoch": 0.8407934692589818,
      "grad_norm": 4.381661415100098,
      "learning_rate": 1.6245371115762358e-07,
      "loss": 0.054,
      "step": 5330
    },
    {
      "epoch": 0.8409512166265726,
      "grad_norm": 7.271567344665527,
      "learning_rate": 1.6229270648848816e-07,
      "loss": 0.0838,
      "step": 5331
    },
    {
      "epoch": 0.8411089639941634,
      "grad_norm": 3.0706746578216553,
      "learning_rate": 1.6213170181935277e-07,
      "loss": 0.0313,
      "step": 5332
    },
    {
      "epoch": 0.8412667113617541,
      "grad_norm": 3.5723705291748047,
      "learning_rate": 1.6197069715021735e-07,
      "loss": 0.0391,
      "step": 5333
    },
    {
      "epoch": 0.8414244587293449,
      "grad_norm": 3.5169169902801514,
      "learning_rate": 1.6180969248108196e-07,
      "loss": 0.0394,
      "step": 5334
    },
    {
      "epoch": 0.8415822060969358,
      "grad_norm": 9.128049850463867,
      "learning_rate": 1.6164868781194654e-07,
      "loss": 0.058,
      "step": 5335
    },
    {
      "epoch": 0.8417399534645266,
      "grad_norm": 4.708001136779785,
      "learning_rate": 1.6148768314281114e-07,
      "loss": 0.0326,
      "step": 5336
    },
    {
      "epoch": 0.8418977008321173,
      "grad_norm": 6.781467914581299,
      "learning_rate": 1.6132667847367572e-07,
      "loss": 0.0401,
      "step": 5337
    },
    {
      "epoch": 0.8420554481997081,
      "grad_norm": 7.906466007232666,
      "learning_rate": 1.6116567380454033e-07,
      "loss": 0.1097,
      "step": 5338
    },
    {
      "epoch": 0.842213195567299,
      "grad_norm": 6.7985053062438965,
      "learning_rate": 1.610046691354049e-07,
      "loss": 0.0895,
      "step": 5339
    },
    {
      "epoch": 0.8423709429348898,
      "grad_norm": 7.104395389556885,
      "learning_rate": 1.6084366446626951e-07,
      "loss": 0.027,
      "step": 5340
    },
    {
      "epoch": 0.8425286903024806,
      "grad_norm": 7.683440208435059,
      "learning_rate": 1.606826597971341e-07,
      "loss": 0.0694,
      "step": 5341
    },
    {
      "epoch": 0.8426864376700713,
      "grad_norm": 5.992101669311523,
      "learning_rate": 1.605216551279987e-07,
      "loss": 0.0298,
      "step": 5342
    },
    {
      "epoch": 0.8428441850376622,
      "grad_norm": 3.087597131729126,
      "learning_rate": 1.603606504588633e-07,
      "loss": 0.0214,
      "step": 5343
    },
    {
      "epoch": 0.843001932405253,
      "grad_norm": 2.7837986946105957,
      "learning_rate": 1.601996457897279e-07,
      "loss": 0.0121,
      "step": 5344
    },
    {
      "epoch": 0.8431596797728438,
      "grad_norm": 4.8075056076049805,
      "learning_rate": 1.600386411205925e-07,
      "loss": 0.0549,
      "step": 5345
    },
    {
      "epoch": 0.8433174271404346,
      "grad_norm": 5.946511745452881,
      "learning_rate": 1.5987763645145707e-07,
      "loss": 0.0535,
      "step": 5346
    },
    {
      "epoch": 0.8434751745080254,
      "grad_norm": 8.470784187316895,
      "learning_rate": 1.5971663178232168e-07,
      "loss": 0.0281,
      "step": 5347
    },
    {
      "epoch": 0.8436329218756162,
      "grad_norm": 3.7917988300323486,
      "learning_rate": 1.5955562711318626e-07,
      "loss": 0.0509,
      "step": 5348
    },
    {
      "epoch": 0.843790669243207,
      "grad_norm": 6.6693806648254395,
      "learning_rate": 1.5939462244405087e-07,
      "loss": 0.0705,
      "step": 5349
    },
    {
      "epoch": 0.8439484166107978,
      "grad_norm": 1.5791327953338623,
      "learning_rate": 1.5923361777491545e-07,
      "loss": 0.0124,
      "step": 5350
    },
    {
      "epoch": 0.8441061639783886,
      "grad_norm": 3.5993335247039795,
      "learning_rate": 1.5907261310578008e-07,
      "loss": 0.0396,
      "step": 5351
    },
    {
      "epoch": 0.8442639113459794,
      "grad_norm": 3.5184741020202637,
      "learning_rate": 1.5891160843664466e-07,
      "loss": 0.04,
      "step": 5352
    },
    {
      "epoch": 0.8444216587135702,
      "grad_norm": 8.009190559387207,
      "learning_rate": 1.5875060376750926e-07,
      "loss": 0.1181,
      "step": 5353
    },
    {
      "epoch": 0.844579406081161,
      "grad_norm": 5.545644760131836,
      "learning_rate": 1.5858959909837384e-07,
      "loss": 0.0656,
      "step": 5354
    },
    {
      "epoch": 0.8447371534487518,
      "grad_norm": 5.302511692047119,
      "learning_rate": 1.5842859442923845e-07,
      "loss": 0.0347,
      "step": 5355
    },
    {
      "epoch": 0.8448949008163427,
      "grad_norm": 5.997876167297363,
      "learning_rate": 1.5826758976010303e-07,
      "loss": 0.0379,
      "step": 5356
    },
    {
      "epoch": 0.8450526481839334,
      "grad_norm": 5.522672653198242,
      "learning_rate": 1.5810658509096764e-07,
      "loss": 0.0662,
      "step": 5357
    },
    {
      "epoch": 0.8452103955515242,
      "grad_norm": 4.134479522705078,
      "learning_rate": 1.5794558042183224e-07,
      "loss": 0.0398,
      "step": 5358
    },
    {
      "epoch": 0.845368142919115,
      "grad_norm": 3.993116617202759,
      "learning_rate": 1.5778457575269682e-07,
      "loss": 0.0573,
      "step": 5359
    },
    {
      "epoch": 0.8455258902867059,
      "grad_norm": 5.000351428985596,
      "learning_rate": 1.5762357108356143e-07,
      "loss": 0.059,
      "step": 5360
    },
    {
      "epoch": 0.8456836376542967,
      "grad_norm": 10.278120994567871,
      "learning_rate": 1.57462566414426e-07,
      "loss": 0.1084,
      "step": 5361
    },
    {
      "epoch": 0.8458413850218874,
      "grad_norm": 5.670780181884766,
      "learning_rate": 1.5730156174529062e-07,
      "loss": 0.0568,
      "step": 5362
    },
    {
      "epoch": 0.8459991323894782,
      "grad_norm": 3.0353498458862305,
      "learning_rate": 1.571405570761552e-07,
      "loss": 0.0196,
      "step": 5363
    },
    {
      "epoch": 0.8461568797570691,
      "grad_norm": 7.127408981323242,
      "learning_rate": 1.569795524070198e-07,
      "loss": 0.0673,
      "step": 5364
    },
    {
      "epoch": 0.8463146271246599,
      "grad_norm": 3.319708824157715,
      "learning_rate": 1.5681854773788438e-07,
      "loss": 0.0223,
      "step": 5365
    },
    {
      "epoch": 0.8464723744922507,
      "grad_norm": 3.6849279403686523,
      "learning_rate": 1.56657543068749e-07,
      "loss": 0.0148,
      "step": 5366
    },
    {
      "epoch": 0.8466301218598414,
      "grad_norm": 4.387140274047852,
      "learning_rate": 1.5649653839961357e-07,
      "loss": 0.0553,
      "step": 5367
    },
    {
      "epoch": 0.8467878692274323,
      "grad_norm": 6.39334774017334,
      "learning_rate": 1.5633553373047817e-07,
      "loss": 0.0516,
      "step": 5368
    },
    {
      "epoch": 0.8469456165950231,
      "grad_norm": 5.241713047027588,
      "learning_rate": 1.5617452906134275e-07,
      "loss": 0.0204,
      "step": 5369
    },
    {
      "epoch": 0.8471033639626139,
      "grad_norm": 4.3344197273254395,
      "learning_rate": 1.5601352439220736e-07,
      "loss": 0.0257,
      "step": 5370
    },
    {
      "epoch": 0.8472611113302047,
      "grad_norm": 1.5527998208999634,
      "learning_rate": 1.5585251972307194e-07,
      "loss": 0.0101,
      "step": 5371
    },
    {
      "epoch": 0.8474188586977954,
      "grad_norm": 4.683798313140869,
      "learning_rate": 1.5569151505393655e-07,
      "loss": 0.0377,
      "step": 5372
    },
    {
      "epoch": 0.8475766060653863,
      "grad_norm": 6.119536399841309,
      "learning_rate": 1.5553051038480113e-07,
      "loss": 0.0286,
      "step": 5373
    },
    {
      "epoch": 0.8477343534329771,
      "grad_norm": 9.703241348266602,
      "learning_rate": 1.5536950571566576e-07,
      "loss": 0.0563,
      "step": 5374
    },
    {
      "epoch": 0.8478921008005679,
      "grad_norm": 5.909456729888916,
      "learning_rate": 1.5520850104653037e-07,
      "loss": 0.06,
      "step": 5375
    },
    {
      "epoch": 0.8480498481681586,
      "grad_norm": 6.890008449554443,
      "learning_rate": 1.5504749637739495e-07,
      "loss": 0.0846,
      "step": 5376
    },
    {
      "epoch": 0.8480498481681586,
      "eval_accuracy": 0.9880505779990634,
      "eval_f1": 0.9880505779990634,
      "eval_loss": 0.038994282484054565,
      "eval_runtime": 4696.7739,
      "eval_samples_per_second": 43.19,
      "eval_steps_per_second": 2.7,
      "step": 5376
    },
    {
      "epoch": 0.8482075955357495,
      "grad_norm": 2.2304201126098633,
      "learning_rate": 1.5488649170825955e-07,
      "loss": 0.0505,
      "step": 5377
    },
    {
      "epoch": 0.8483653429033403,
      "grad_norm": 8.14090633392334,
      "learning_rate": 1.5472548703912413e-07,
      "loss": 0.052,
      "step": 5378
    },
    {
      "epoch": 0.8485230902709311,
      "grad_norm": 3.4045495986938477,
      "learning_rate": 1.5456448236998874e-07,
      "loss": 0.0606,
      "step": 5379
    },
    {
      "epoch": 0.8486808376385219,
      "grad_norm": 9.605330467224121,
      "learning_rate": 1.5440347770085332e-07,
      "loss": 0.0482,
      "step": 5380
    },
    {
      "epoch": 0.8488385850061128,
      "grad_norm": 3.838212013244629,
      "learning_rate": 1.5424247303171792e-07,
      "loss": 0.0585,
      "step": 5381
    },
    {
      "epoch": 0.8489963323737035,
      "grad_norm": 8.023015022277832,
      "learning_rate": 1.540814683625825e-07,
      "loss": 0.0737,
      "step": 5382
    },
    {
      "epoch": 0.8491540797412943,
      "grad_norm": 4.286295413970947,
      "learning_rate": 1.539204636934471e-07,
      "loss": 0.023,
      "step": 5383
    },
    {
      "epoch": 0.8493118271088851,
      "grad_norm": 3.6643738746643066,
      "learning_rate": 1.537594590243117e-07,
      "loss": 0.0323,
      "step": 5384
    },
    {
      "epoch": 0.849469574476476,
      "grad_norm": 4.660480976104736,
      "learning_rate": 1.535984543551763e-07,
      "loss": 0.0373,
      "step": 5385
    },
    {
      "epoch": 0.8496273218440668,
      "grad_norm": 7.360021114349365,
      "learning_rate": 1.5343744968604088e-07,
      "loss": 0.0304,
      "step": 5386
    },
    {
      "epoch": 0.8497850692116575,
      "grad_norm": 5.344607830047607,
      "learning_rate": 1.5327644501690548e-07,
      "loss": 0.0419,
      "step": 5387
    },
    {
      "epoch": 0.8499428165792483,
      "grad_norm": 4.966273784637451,
      "learning_rate": 1.5311544034777006e-07,
      "loss": 0.0308,
      "step": 5388
    },
    {
      "epoch": 0.8501005639468391,
      "grad_norm": 2.0803072452545166,
      "learning_rate": 1.5295443567863467e-07,
      "loss": 0.0167,
      "step": 5389
    },
    {
      "epoch": 0.85025831131443,
      "grad_norm": 4.4739532470703125,
      "learning_rate": 1.5279343100949927e-07,
      "loss": 0.0309,
      "step": 5390
    },
    {
      "epoch": 0.8504160586820207,
      "grad_norm": 4.332695960998535,
      "learning_rate": 1.5263242634036385e-07,
      "loss": 0.0297,
      "step": 5391
    },
    {
      "epoch": 0.8505738060496115,
      "grad_norm": 6.460211753845215,
      "learning_rate": 1.5247142167122846e-07,
      "loss": 0.0442,
      "step": 5392
    },
    {
      "epoch": 0.8507315534172023,
      "grad_norm": 5.7827067375183105,
      "learning_rate": 1.5231041700209304e-07,
      "loss": 0.0314,
      "step": 5393
    },
    {
      "epoch": 0.8508893007847932,
      "grad_norm": 3.955542802810669,
      "learning_rate": 1.5214941233295765e-07,
      "loss": 0.0606,
      "step": 5394
    },
    {
      "epoch": 0.851047048152384,
      "grad_norm": 8.374228477478027,
      "learning_rate": 1.5198840766382223e-07,
      "loss": 0.0665,
      "step": 5395
    },
    {
      "epoch": 0.8512047955199747,
      "grad_norm": 5.542219638824463,
      "learning_rate": 1.5182740299468683e-07,
      "loss": 0.0426,
      "step": 5396
    },
    {
      "epoch": 0.8513625428875655,
      "grad_norm": 3.1665616035461426,
      "learning_rate": 1.5166639832555144e-07,
      "loss": 0.0483,
      "step": 5397
    },
    {
      "epoch": 0.8515202902551564,
      "grad_norm": 7.315969467163086,
      "learning_rate": 1.5150539365641605e-07,
      "loss": 0.0549,
      "step": 5398
    },
    {
      "epoch": 0.8516780376227472,
      "grad_norm": 1.9205162525177002,
      "learning_rate": 1.5134438898728063e-07,
      "loss": 0.0366,
      "step": 5399
    },
    {
      "epoch": 0.851835784990338,
      "grad_norm": 4.792670249938965,
      "learning_rate": 1.5118338431814523e-07,
      "loss": 0.0555,
      "step": 5400
    },
    {
      "epoch": 0.8519935323579287,
      "grad_norm": 4.623531818389893,
      "learning_rate": 1.510223796490098e-07,
      "loss": 0.0252,
      "step": 5401
    },
    {
      "epoch": 0.8521512797255196,
      "grad_norm": 6.570959091186523,
      "learning_rate": 1.5086137497987442e-07,
      "loss": 0.031,
      "step": 5402
    },
    {
      "epoch": 0.8523090270931104,
      "grad_norm": 5.0001654624938965,
      "learning_rate": 1.50700370310739e-07,
      "loss": 0.0171,
      "step": 5403
    },
    {
      "epoch": 0.8524667744607012,
      "grad_norm": 2.375577449798584,
      "learning_rate": 1.505393656416036e-07,
      "loss": 0.0141,
      "step": 5404
    },
    {
      "epoch": 0.852624521828292,
      "grad_norm": 2.5384368896484375,
      "learning_rate": 1.503783609724682e-07,
      "loss": 0.0117,
      "step": 5405
    },
    {
      "epoch": 0.8527822691958828,
      "grad_norm": 4.021308422088623,
      "learning_rate": 1.502173563033328e-07,
      "loss": 0.0564,
      "step": 5406
    },
    {
      "epoch": 0.8529400165634736,
      "grad_norm": 3.8227062225341797,
      "learning_rate": 1.500563516341974e-07,
      "loss": 0.0469,
      "step": 5407
    },
    {
      "epoch": 0.8530977639310644,
      "grad_norm": 4.229830741882324,
      "learning_rate": 1.4989534696506198e-07,
      "loss": 0.0156,
      "step": 5408
    },
    {
      "epoch": 0.8532555112986552,
      "grad_norm": 5.805696964263916,
      "learning_rate": 1.4973434229592658e-07,
      "loss": 0.0552,
      "step": 5409
    },
    {
      "epoch": 0.853413258666246,
      "grad_norm": 4.083367824554443,
      "learning_rate": 1.4957333762679116e-07,
      "loss": 0.0477,
      "step": 5410
    },
    {
      "epoch": 0.8535710060338368,
      "grad_norm": 2.8954594135284424,
      "learning_rate": 1.4941233295765577e-07,
      "loss": 0.0396,
      "step": 5411
    },
    {
      "epoch": 0.8537287534014276,
      "grad_norm": 4.661309719085693,
      "learning_rate": 1.4925132828852035e-07,
      "loss": 0.0522,
      "step": 5412
    },
    {
      "epoch": 0.8538865007690184,
      "grad_norm": 7.065108299255371,
      "learning_rate": 1.4909032361938496e-07,
      "loss": 0.0316,
      "step": 5413
    },
    {
      "epoch": 0.8540442481366092,
      "grad_norm": 3.9269495010375977,
      "learning_rate": 1.4892931895024954e-07,
      "loss": 0.0467,
      "step": 5414
    },
    {
      "epoch": 0.8542019955042001,
      "grad_norm": 5.224636554718018,
      "learning_rate": 1.4876831428111414e-07,
      "loss": 0.0492,
      "step": 5415
    },
    {
      "epoch": 0.8543597428717908,
      "grad_norm": 3.75856351852417,
      "learning_rate": 1.4860730961197872e-07,
      "loss": 0.0448,
      "step": 5416
    },
    {
      "epoch": 0.8545174902393816,
      "grad_norm": 6.5601606369018555,
      "learning_rate": 1.4844630494284333e-07,
      "loss": 0.0275,
      "step": 5417
    },
    {
      "epoch": 0.8546752376069724,
      "grad_norm": 3.714090347290039,
      "learning_rate": 1.482853002737079e-07,
      "loss": 0.025,
      "step": 5418
    },
    {
      "epoch": 0.8548329849745633,
      "grad_norm": 8.23410701751709,
      "learning_rate": 1.4812429560457251e-07,
      "loss": 0.0547,
      "step": 5419
    },
    {
      "epoch": 0.8549907323421541,
      "grad_norm": 3.264073133468628,
      "learning_rate": 1.4796329093543712e-07,
      "loss": 0.019,
      "step": 5420
    },
    {
      "epoch": 0.8551484797097448,
      "grad_norm": 3.837836980819702,
      "learning_rate": 1.4780228626630173e-07,
      "loss": 0.0235,
      "step": 5421
    },
    {
      "epoch": 0.8553062270773356,
      "grad_norm": 8.193350791931152,
      "learning_rate": 1.4764128159716633e-07,
      "loss": 0.0353,
      "step": 5422
    },
    {
      "epoch": 0.8554639744449265,
      "grad_norm": 3.9345805644989014,
      "learning_rate": 1.474802769280309e-07,
      "loss": 0.0205,
      "step": 5423
    },
    {
      "epoch": 0.8556217218125173,
      "grad_norm": 4.9703898429870605,
      "learning_rate": 1.4731927225889552e-07,
      "loss": 0.0445,
      "step": 5424
    },
    {
      "epoch": 0.855779469180108,
      "grad_norm": 6.9233551025390625,
      "learning_rate": 1.471582675897601e-07,
      "loss": 0.0958,
      "step": 5425
    },
    {
      "epoch": 0.8559372165476988,
      "grad_norm": 4.656965732574463,
      "learning_rate": 1.469972629206247e-07,
      "loss": 0.0548,
      "step": 5426
    },
    {
      "epoch": 0.8560949639152896,
      "grad_norm": 6.408620834350586,
      "learning_rate": 1.4683625825148928e-07,
      "loss": 0.0559,
      "step": 5427
    },
    {
      "epoch": 0.8562527112828805,
      "grad_norm": 4.187609672546387,
      "learning_rate": 1.466752535823539e-07,
      "loss": 0.0439,
      "step": 5428
    },
    {
      "epoch": 0.8564104586504713,
      "grad_norm": 3.907073497772217,
      "learning_rate": 1.4651424891321847e-07,
      "loss": 0.0207,
      "step": 5429
    },
    {
      "epoch": 0.856568206018062,
      "grad_norm": 2.5755913257598877,
      "learning_rate": 1.4635324424408308e-07,
      "loss": 0.0192,
      "step": 5430
    },
    {
      "epoch": 0.8567259533856528,
      "grad_norm": 7.500949859619141,
      "learning_rate": 1.4619223957494766e-07,
      "loss": 0.0721,
      "step": 5431
    },
    {
      "epoch": 0.8568837007532437,
      "grad_norm": 4.847621440887451,
      "learning_rate": 1.4603123490581226e-07,
      "loss": 0.0414,
      "step": 5432
    },
    {
      "epoch": 0.8570414481208345,
      "grad_norm": 11.874192237854004,
      "learning_rate": 1.4587023023667684e-07,
      "loss": 0.064,
      "step": 5433
    },
    {
      "epoch": 0.8571991954884253,
      "grad_norm": 3.3340604305267334,
      "learning_rate": 1.4570922556754145e-07,
      "loss": 0.0541,
      "step": 5434
    },
    {
      "epoch": 0.857356942856016,
      "grad_norm": 3.7429940700531006,
      "learning_rate": 1.4554822089840606e-07,
      "loss": 0.0274,
      "step": 5435
    },
    {
      "epoch": 0.8575146902236069,
      "grad_norm": 5.898879051208496,
      "learning_rate": 1.4538721622927064e-07,
      "loss": 0.0472,
      "step": 5436
    },
    {
      "epoch": 0.8576724375911977,
      "grad_norm": 5.683361530303955,
      "learning_rate": 1.4522621156013524e-07,
      "loss": 0.0394,
      "step": 5437
    },
    {
      "epoch": 0.8578301849587885,
      "grad_norm": 6.243908882141113,
      "learning_rate": 1.4506520689099982e-07,
      "loss": 0.0507,
      "step": 5438
    },
    {
      "epoch": 0.8579879323263793,
      "grad_norm": 8.795913696289062,
      "learning_rate": 1.4490420222186443e-07,
      "loss": 0.0618,
      "step": 5439
    },
    {
      "epoch": 0.8581456796939702,
      "grad_norm": 11.45658016204834,
      "learning_rate": 1.44743197552729e-07,
      "loss": 0.0555,
      "step": 5440
    },
    {
      "epoch": 0.8583034270615609,
      "grad_norm": 5.12626838684082,
      "learning_rate": 1.4458219288359361e-07,
      "loss": 0.026,
      "step": 5441
    },
    {
      "epoch": 0.8584611744291517,
      "grad_norm": 7.450155735015869,
      "learning_rate": 1.444211882144582e-07,
      "loss": 0.0952,
      "step": 5442
    },
    {
      "epoch": 0.8586189217967425,
      "grad_norm": 4.901927471160889,
      "learning_rate": 1.442601835453228e-07,
      "loss": 0.0459,
      "step": 5443
    },
    {
      "epoch": 0.8587766691643334,
      "grad_norm": 5.0577898025512695,
      "learning_rate": 1.440991788761874e-07,
      "loss": 0.0187,
      "step": 5444
    },
    {
      "epoch": 0.8589344165319241,
      "grad_norm": 5.3275580406188965,
      "learning_rate": 1.4393817420705201e-07,
      "loss": 0.0212,
      "step": 5445
    },
    {
      "epoch": 0.8590921638995149,
      "grad_norm": 3.313403367996216,
      "learning_rate": 1.437771695379166e-07,
      "loss": 0.0366,
      "step": 5446
    },
    {
      "epoch": 0.8592499112671057,
      "grad_norm": 9.613713264465332,
      "learning_rate": 1.436161648687812e-07,
      "loss": 0.0365,
      "step": 5447
    },
    {
      "epoch": 0.8594076586346965,
      "grad_norm": 4.729363918304443,
      "learning_rate": 1.4345516019964578e-07,
      "loss": 0.0644,
      "step": 5448
    },
    {
      "epoch": 0.8595654060022874,
      "grad_norm": 7.8069539070129395,
      "learning_rate": 1.4329415553051039e-07,
      "loss": 0.0562,
      "step": 5449
    },
    {
      "epoch": 0.8597231533698781,
      "grad_norm": 3.9969279766082764,
      "learning_rate": 1.43133150861375e-07,
      "loss": 0.0683,
      "step": 5450
    },
    {
      "epoch": 0.8598809007374689,
      "grad_norm": 7.209540367126465,
      "learning_rate": 1.4297214619223957e-07,
      "loss": 0.0496,
      "step": 5451
    },
    {
      "epoch": 0.8600386481050597,
      "grad_norm": 2.844700813293457,
      "learning_rate": 1.4281114152310418e-07,
      "loss": 0.0204,
      "step": 5452
    },
    {
      "epoch": 0.8601963954726506,
      "grad_norm": 6.2221455574035645,
      "learning_rate": 1.4265013685396876e-07,
      "loss": 0.0448,
      "step": 5453
    },
    {
      "epoch": 0.8603541428402414,
      "grad_norm": 5.791203498840332,
      "learning_rate": 1.4248913218483336e-07,
      "loss": 0.0466,
      "step": 5454
    },
    {
      "epoch": 0.8605118902078321,
      "grad_norm": 4.409770488739014,
      "learning_rate": 1.4232812751569794e-07,
      "loss": 0.0212,
      "step": 5455
    },
    {
      "epoch": 0.8606696375754229,
      "grad_norm": 3.2753994464874268,
      "learning_rate": 1.4216712284656255e-07,
      "loss": 0.024,
      "step": 5456
    },
    {
      "epoch": 0.8608273849430138,
      "grad_norm": 6.3212127685546875,
      "learning_rate": 1.4200611817742713e-07,
      "loss": 0.0755,
      "step": 5457
    },
    {
      "epoch": 0.8609851323106046,
      "grad_norm": 3.5120387077331543,
      "learning_rate": 1.4184511350829174e-07,
      "loss": 0.0335,
      "step": 5458
    },
    {
      "epoch": 0.8611428796781954,
      "grad_norm": 3.85371732711792,
      "learning_rate": 1.4168410883915632e-07,
      "loss": 0.0366,
      "step": 5459
    },
    {
      "epoch": 0.8613006270457861,
      "grad_norm": 6.459498405456543,
      "learning_rate": 1.4152310417002092e-07,
      "loss": 0.0612,
      "step": 5460
    },
    {
      "epoch": 0.861458374413377,
      "grad_norm": 5.595515251159668,
      "learning_rate": 1.413620995008855e-07,
      "loss": 0.0859,
      "step": 5461
    },
    {
      "epoch": 0.8616161217809678,
      "grad_norm": 4.2606987953186035,
      "learning_rate": 1.412010948317501e-07,
      "loss": 0.0686,
      "step": 5462
    },
    {
      "epoch": 0.8617738691485586,
      "grad_norm": 5.481800556182861,
      "learning_rate": 1.410400901626147e-07,
      "loss": 0.0427,
      "step": 5463
    },
    {
      "epoch": 0.8619316165161494,
      "grad_norm": 4.921341896057129,
      "learning_rate": 1.408790854934793e-07,
      "loss": 0.0245,
      "step": 5464
    },
    {
      "epoch": 0.8620893638837402,
      "grad_norm": 4.414101600646973,
      "learning_rate": 1.4071808082434387e-07,
      "loss": 0.0324,
      "step": 5465
    },
    {
      "epoch": 0.862247111251331,
      "grad_norm": 6.160189628601074,
      "learning_rate": 1.4055707615520848e-07,
      "loss": 0.0534,
      "step": 5466
    },
    {
      "epoch": 0.8624048586189218,
      "grad_norm": 3.8047962188720703,
      "learning_rate": 1.4039607148607311e-07,
      "loss": 0.0171,
      "step": 5467
    },
    {
      "epoch": 0.8625626059865126,
      "grad_norm": 4.260643482208252,
      "learning_rate": 1.402350668169377e-07,
      "loss": 0.0369,
      "step": 5468
    },
    {
      "epoch": 0.8627203533541034,
      "grad_norm": 3.534762382507324,
      "learning_rate": 1.400740621478023e-07,
      "loss": 0.0357,
      "step": 5469
    },
    {
      "epoch": 0.8628781007216942,
      "grad_norm": 3.0478262901306152,
      "learning_rate": 1.3991305747866688e-07,
      "loss": 0.0175,
      "step": 5470
    },
    {
      "epoch": 0.863035848089285,
      "grad_norm": 6.904594898223877,
      "learning_rate": 1.3975205280953149e-07,
      "loss": 0.077,
      "step": 5471
    },
    {
      "epoch": 0.8631935954568758,
      "grad_norm": 9.13009262084961,
      "learning_rate": 1.3959104814039607e-07,
      "loss": 0.0544,
      "step": 5472
    },
    {
      "epoch": 0.8633513428244666,
      "grad_norm": 6.062447547912598,
      "learning_rate": 1.3943004347126067e-07,
      "loss": 0.0977,
      "step": 5473
    },
    {
      "epoch": 0.8635090901920575,
      "grad_norm": 5.987272262573242,
      "learning_rate": 1.3926903880212525e-07,
      "loss": 0.0302,
      "step": 5474
    },
    {
      "epoch": 0.8636668375596482,
      "grad_norm": 4.56599760055542,
      "learning_rate": 1.3910803413298986e-07,
      "loss": 0.0235,
      "step": 5475
    },
    {
      "epoch": 0.863824584927239,
      "grad_norm": 5.920708179473877,
      "learning_rate": 1.3894702946385444e-07,
      "loss": 0.0559,
      "step": 5476
    },
    {
      "epoch": 0.8639823322948298,
      "grad_norm": 4.275696754455566,
      "learning_rate": 1.3878602479471904e-07,
      "loss": 0.0248,
      "step": 5477
    },
    {
      "epoch": 0.8641400796624207,
      "grad_norm": 6.5294270515441895,
      "learning_rate": 1.3862502012558362e-07,
      "loss": 0.0582,
      "step": 5478
    },
    {
      "epoch": 0.8642978270300115,
      "grad_norm": 5.760613918304443,
      "learning_rate": 1.3846401545644823e-07,
      "loss": 0.0661,
      "step": 5479
    },
    {
      "epoch": 0.8644555743976022,
      "grad_norm": 4.418890953063965,
      "learning_rate": 1.383030107873128e-07,
      "loss": 0.0307,
      "step": 5480
    },
    {
      "epoch": 0.864613321765193,
      "grad_norm": 3.454737424850464,
      "learning_rate": 1.3814200611817742e-07,
      "loss": 0.0341,
      "step": 5481
    },
    {
      "epoch": 0.8647710691327839,
      "grad_norm": 3.304800510406494,
      "learning_rate": 1.3798100144904202e-07,
      "loss": 0.0207,
      "step": 5482
    },
    {
      "epoch": 0.8649288165003747,
      "grad_norm": 5.844666481018066,
      "learning_rate": 1.378199967799066e-07,
      "loss": 0.0622,
      "step": 5483
    },
    {
      "epoch": 0.8650865638679655,
      "grad_norm": 2.9924535751342773,
      "learning_rate": 1.376589921107712e-07,
      "loss": 0.0347,
      "step": 5484
    },
    {
      "epoch": 0.8652443112355562,
      "grad_norm": 5.538552284240723,
      "learning_rate": 1.374979874416358e-07,
      "loss": 0.0497,
      "step": 5485
    },
    {
      "epoch": 0.865402058603147,
      "grad_norm": 5.983892917633057,
      "learning_rate": 1.373369827725004e-07,
      "loss": 0.1164,
      "step": 5486
    },
    {
      "epoch": 0.8655598059707379,
      "grad_norm": 3.1277856826782227,
      "learning_rate": 1.3717597810336498e-07,
      "loss": 0.0589,
      "step": 5487
    },
    {
      "epoch": 0.8657175533383287,
      "grad_norm": 3.137880325317383,
      "learning_rate": 1.3701497343422958e-07,
      "loss": 0.0211,
      "step": 5488
    },
    {
      "epoch": 0.8658753007059194,
      "grad_norm": 7.256457328796387,
      "learning_rate": 1.3685396876509416e-07,
      "loss": 0.0381,
      "step": 5489
    },
    {
      "epoch": 0.8660330480735102,
      "grad_norm": 5.043852806091309,
      "learning_rate": 1.366929640959588e-07,
      "loss": 0.0212,
      "step": 5490
    },
    {
      "epoch": 0.8661907954411011,
      "grad_norm": 11.189506530761719,
      "learning_rate": 1.3653195942682337e-07,
      "loss": 0.0846,
      "step": 5491
    },
    {
      "epoch": 0.8663485428086919,
      "grad_norm": 5.468672752380371,
      "learning_rate": 1.3637095475768798e-07,
      "loss": 0.1226,
      "step": 5492
    },
    {
      "epoch": 0.8665062901762827,
      "grad_norm": 3.6421420574188232,
      "learning_rate": 1.3620995008855256e-07,
      "loss": 0.0461,
      "step": 5493
    },
    {
      "epoch": 0.8666640375438734,
      "grad_norm": 4.256086349487305,
      "learning_rate": 1.3604894541941717e-07,
      "loss": 0.0468,
      "step": 5494
    },
    {
      "epoch": 0.8668217849114643,
      "grad_norm": 3.2741804122924805,
      "learning_rate": 1.3588794075028175e-07,
      "loss": 0.0261,
      "step": 5495
    },
    {
      "epoch": 0.8669795322790551,
      "grad_norm": 6.785216808319092,
      "learning_rate": 1.3572693608114635e-07,
      "loss": 0.0376,
      "step": 5496
    },
    {
      "epoch": 0.8671372796466459,
      "grad_norm": 8.623574256896973,
      "learning_rate": 1.3556593141201096e-07,
      "loss": 0.0441,
      "step": 5497
    },
    {
      "epoch": 0.8672950270142367,
      "grad_norm": 4.828185558319092,
      "learning_rate": 1.3540492674287554e-07,
      "loss": 0.0513,
      "step": 5498
    },
    {
      "epoch": 0.8674527743818276,
      "grad_norm": 6.098627090454102,
      "learning_rate": 1.3524392207374015e-07,
      "loss": 0.0362,
      "step": 5499
    },
    {
      "epoch": 0.8676105217494183,
      "grad_norm": 3.566786289215088,
      "learning_rate": 1.3508291740460473e-07,
      "loss": 0.0375,
      "step": 5500
    },
    {
      "epoch": 0.8677682691170091,
      "grad_norm": 5.9243550300598145,
      "learning_rate": 1.3492191273546933e-07,
      "loss": 0.0663,
      "step": 5501
    },
    {
      "epoch": 0.8679260164845999,
      "grad_norm": 4.2992048263549805,
      "learning_rate": 1.347609080663339e-07,
      "loss": 0.0516,
      "step": 5502
    },
    {
      "epoch": 0.8680837638521908,
      "grad_norm": 3.649446964263916,
      "learning_rate": 1.3459990339719852e-07,
      "loss": 0.0556,
      "step": 5503
    },
    {
      "epoch": 0.8682415112197815,
      "grad_norm": 3.712911367416382,
      "learning_rate": 1.344388987280631e-07,
      "loss": 0.0167,
      "step": 5504
    },
    {
      "epoch": 0.8683992585873723,
      "grad_norm": 2.6621758937835693,
      "learning_rate": 1.342778940589277e-07,
      "loss": 0.019,
      "step": 5505
    },
    {
      "epoch": 0.8685570059549631,
      "grad_norm": 5.428889274597168,
      "learning_rate": 1.3411688938979228e-07,
      "loss": 0.0389,
      "step": 5506
    },
    {
      "epoch": 0.8687147533225539,
      "grad_norm": 2.7606325149536133,
      "learning_rate": 1.339558847206569e-07,
      "loss": 0.0338,
      "step": 5507
    },
    {
      "epoch": 0.8688725006901448,
      "grad_norm": 4.508447170257568,
      "learning_rate": 1.3379488005152147e-07,
      "loss": 0.027,
      "step": 5508
    },
    {
      "epoch": 0.8690302480577355,
      "grad_norm": 3.3050265312194824,
      "learning_rate": 1.3363387538238608e-07,
      "loss": 0.0173,
      "step": 5509
    },
    {
      "epoch": 0.8691879954253263,
      "grad_norm": 4.372236251831055,
      "learning_rate": 1.3347287071325066e-07,
      "loss": 0.0327,
      "step": 5510
    },
    {
      "epoch": 0.8693457427929171,
      "grad_norm": 2.6069273948669434,
      "learning_rate": 1.3331186604411526e-07,
      "loss": 0.0166,
      "step": 5511
    },
    {
      "epoch": 0.869503490160508,
      "grad_norm": 3.1929118633270264,
      "learning_rate": 1.3315086137497987e-07,
      "loss": 0.0508,
      "step": 5512
    },
    {
      "epoch": 0.8696612375280988,
      "grad_norm": 4.904454708099365,
      "learning_rate": 1.3298985670584445e-07,
      "loss": 0.0689,
      "step": 5513
    },
    {
      "epoch": 0.8698189848956895,
      "grad_norm": 3.687384605407715,
      "learning_rate": 1.3282885203670908e-07,
      "loss": 0.0288,
      "step": 5514
    },
    {
      "epoch": 0.8699767322632803,
      "grad_norm": 3.5571999549865723,
      "learning_rate": 1.3266784736757366e-07,
      "loss": 0.0211,
      "step": 5515
    },
    {
      "epoch": 0.8701344796308712,
      "grad_norm": 5.523207664489746,
      "learning_rate": 1.3250684269843827e-07,
      "loss": 0.0323,
      "step": 5516
    },
    {
      "epoch": 0.870292226998462,
      "grad_norm": 2.4000661373138428,
      "learning_rate": 1.3234583802930285e-07,
      "loss": 0.012,
      "step": 5517
    },
    {
      "epoch": 0.8704499743660528,
      "grad_norm": 4.749390602111816,
      "learning_rate": 1.3218483336016745e-07,
      "loss": 0.0679,
      "step": 5518
    },
    {
      "epoch": 0.8706077217336435,
      "grad_norm": 6.371307373046875,
      "learning_rate": 1.3202382869103203e-07,
      "loss": 0.037,
      "step": 5519
    },
    {
      "epoch": 0.8707654691012344,
      "grad_norm": 7.201481819152832,
      "learning_rate": 1.3186282402189664e-07,
      "loss": 0.0438,
      "step": 5520
    },
    {
      "epoch": 0.8709232164688252,
      "grad_norm": 3.805671215057373,
      "learning_rate": 1.3170181935276122e-07,
      "loss": 0.0228,
      "step": 5521
    },
    {
      "epoch": 0.871080963836416,
      "grad_norm": 6.359230041503906,
      "learning_rate": 1.3154081468362583e-07,
      "loss": 0.0374,
      "step": 5522
    },
    {
      "epoch": 0.8712387112040068,
      "grad_norm": 4.945654392242432,
      "learning_rate": 1.313798100144904e-07,
      "loss": 0.0359,
      "step": 5523
    },
    {
      "epoch": 0.8713964585715975,
      "grad_norm": 11.090949058532715,
      "learning_rate": 1.31218805345355e-07,
      "loss": 0.0652,
      "step": 5524
    },
    {
      "epoch": 0.8715542059391884,
      "grad_norm": 4.84529447555542,
      "learning_rate": 1.310578006762196e-07,
      "loss": 0.0421,
      "step": 5525
    },
    {
      "epoch": 0.8717119533067792,
      "grad_norm": 4.299323081970215,
      "learning_rate": 1.308967960070842e-07,
      "loss": 0.0262,
      "step": 5526
    },
    {
      "epoch": 0.87186970067437,
      "grad_norm": 4.873415946960449,
      "learning_rate": 1.307357913379488e-07,
      "loss": 0.0235,
      "step": 5527
    },
    {
      "epoch": 0.8720274480419608,
      "grad_norm": 5.196324348449707,
      "learning_rate": 1.3057478666881338e-07,
      "loss": 0.0603,
      "step": 5528
    },
    {
      "epoch": 0.8721851954095516,
      "grad_norm": 4.5741353034973145,
      "learning_rate": 1.30413781999678e-07,
      "loss": 0.0368,
      "step": 5529
    },
    {
      "epoch": 0.8723429427771424,
      "grad_norm": 4.3620924949646,
      "learning_rate": 1.3025277733054257e-07,
      "loss": 0.0488,
      "step": 5530
    },
    {
      "epoch": 0.8725006901447332,
      "grad_norm": 3.900864362716675,
      "learning_rate": 1.3009177266140718e-07,
      "loss": 0.0229,
      "step": 5531
    },
    {
      "epoch": 0.872658437512324,
      "grad_norm": 3.649635076522827,
      "learning_rate": 1.2993076799227176e-07,
      "loss": 0.0163,
      "step": 5532
    },
    {
      "epoch": 0.8728161848799149,
      "grad_norm": 3.126142740249634,
      "learning_rate": 1.2976976332313636e-07,
      "loss": 0.014,
      "step": 5533
    },
    {
      "epoch": 0.8729739322475056,
      "grad_norm": 5.796621799468994,
      "learning_rate": 1.2960875865400094e-07,
      "loss": 0.0625,
      "step": 5534
    },
    {
      "epoch": 0.8731316796150964,
      "grad_norm": 4.14392614364624,
      "learning_rate": 1.2944775398486555e-07,
      "loss": 0.0959,
      "step": 5535
    },
    {
      "epoch": 0.8732894269826872,
      "grad_norm": 5.45209264755249,
      "learning_rate": 1.2928674931573013e-07,
      "loss": 0.0518,
      "step": 5536
    },
    {
      "epoch": 0.8734471743502781,
      "grad_norm": 5.426156044006348,
      "learning_rate": 1.2912574464659476e-07,
      "loss": 0.0293,
      "step": 5537
    },
    {
      "epoch": 0.8736049217178689,
      "grad_norm": 2.479907274246216,
      "learning_rate": 1.2896473997745934e-07,
      "loss": 0.0229,
      "step": 5538
    },
    {
      "epoch": 0.8737626690854596,
      "grad_norm": 11.546285629272461,
      "learning_rate": 1.2880373530832395e-07,
      "loss": 0.0648,
      "step": 5539
    },
    {
      "epoch": 0.8739204164530504,
      "grad_norm": 3.793890953063965,
      "learning_rate": 1.2864273063918853e-07,
      "loss": 0.0261,
      "step": 5540
    },
    {
      "epoch": 0.8740781638206413,
      "grad_norm": 3.6029372215270996,
      "learning_rate": 1.2848172597005313e-07,
      "loss": 0.0133,
      "step": 5541
    },
    {
      "epoch": 0.8742359111882321,
      "grad_norm": 5.791674613952637,
      "learning_rate": 1.2832072130091774e-07,
      "loss": 0.0385,
      "step": 5542
    },
    {
      "epoch": 0.8743936585558229,
      "grad_norm": 5.276854991912842,
      "learning_rate": 1.2815971663178232e-07,
      "loss": 0.0487,
      "step": 5543
    },
    {
      "epoch": 0.8745514059234136,
      "grad_norm": 2.5421369075775146,
      "learning_rate": 1.2799871196264693e-07,
      "loss": 0.0203,
      "step": 5544
    },
    {
      "epoch": 0.8747091532910044,
      "grad_norm": 7.317483425140381,
      "learning_rate": 1.278377072935115e-07,
      "loss": 0.0333,
      "step": 5545
    },
    {
      "epoch": 0.8748669006585953,
      "grad_norm": 4.536862850189209,
      "learning_rate": 1.276767026243761e-07,
      "loss": 0.0141,
      "step": 5546
    },
    {
      "epoch": 0.8750246480261861,
      "grad_norm": 5.074760437011719,
      "learning_rate": 1.275156979552407e-07,
      "loss": 0.053,
      "step": 5547
    },
    {
      "epoch": 0.8751823953937768,
      "grad_norm": 3.548044204711914,
      "learning_rate": 1.273546932861053e-07,
      "loss": 0.0801,
      "step": 5548
    },
    {
      "epoch": 0.8753401427613676,
      "grad_norm": 1.4778392314910889,
      "learning_rate": 1.2719368861696988e-07,
      "loss": 0.0175,
      "step": 5549
    },
    {
      "epoch": 0.8754978901289585,
      "grad_norm": 3.6699726581573486,
      "learning_rate": 1.2703268394783449e-07,
      "loss": 0.0194,
      "step": 5550
    },
    {
      "epoch": 0.8756556374965493,
      "grad_norm": 7.689450263977051,
      "learning_rate": 1.2687167927869907e-07,
      "loss": 0.031,
      "step": 5551
    },
    {
      "epoch": 0.8758133848641401,
      "grad_norm": 3.877007246017456,
      "learning_rate": 1.2671067460956367e-07,
      "loss": 0.0222,
      "step": 5552
    },
    {
      "epoch": 0.8759711322317308,
      "grad_norm": 4.741959571838379,
      "learning_rate": 1.2654966994042825e-07,
      "loss": 0.0274,
      "step": 5553
    },
    {
      "epoch": 0.8761288795993217,
      "grad_norm": 7.20140266418457,
      "learning_rate": 1.2638866527129286e-07,
      "loss": 0.0272,
      "step": 5554
    },
    {
      "epoch": 0.8762866269669125,
      "grad_norm": 7.812570095062256,
      "learning_rate": 1.2622766060215744e-07,
      "loss": 0.0382,
      "step": 5555
    },
    {
      "epoch": 0.8764443743345033,
      "grad_norm": 4.745429992675781,
      "learning_rate": 1.2606665593302204e-07,
      "loss": 0.0347,
      "step": 5556
    },
    {
      "epoch": 0.8766021217020941,
      "grad_norm": 6.139153957366943,
      "learning_rate": 1.2590565126388662e-07,
      "loss": 0.0167,
      "step": 5557
    },
    {
      "epoch": 0.876759869069685,
      "grad_norm": 5.597256660461426,
      "learning_rate": 1.2574464659475123e-07,
      "loss": 0.0236,
      "step": 5558
    },
    {
      "epoch": 0.8769176164372757,
      "grad_norm": 7.539785385131836,
      "learning_rate": 1.2558364192561584e-07,
      "loss": 0.0815,
      "step": 5559
    },
    {
      "epoch": 0.8770753638048665,
      "grad_norm": 3.764596462249756,
      "learning_rate": 1.2542263725648044e-07,
      "loss": 0.025,
      "step": 5560
    },
    {
      "epoch": 0.8772331111724573,
      "grad_norm": 4.3173089027404785,
      "learning_rate": 1.2526163258734505e-07,
      "loss": 0.0335,
      "step": 5561
    },
    {
      "epoch": 0.8773908585400481,
      "grad_norm": 11.727649688720703,
      "learning_rate": 1.2510062791820963e-07,
      "loss": 0.0641,
      "step": 5562
    },
    {
      "epoch": 0.877548605907639,
      "grad_norm": 5.548543930053711,
      "learning_rate": 1.249396232490742e-07,
      "loss": 0.0704,
      "step": 5563
    },
    {
      "epoch": 0.8777063532752297,
      "grad_norm": 5.145328044891357,
      "learning_rate": 1.2477861857993881e-07,
      "loss": 0.0552,
      "step": 5564
    },
    {
      "epoch": 0.8778641006428205,
      "grad_norm": 5.675923824310303,
      "learning_rate": 1.246176139108034e-07,
      "loss": 0.077,
      "step": 5565
    },
    {
      "epoch": 0.8780218480104113,
      "grad_norm": 4.026497840881348,
      "learning_rate": 1.24456609241668e-07,
      "loss": 0.047,
      "step": 5566
    },
    {
      "epoch": 0.8781795953780022,
      "grad_norm": 3.609802484512329,
      "learning_rate": 1.242956045725326e-07,
      "loss": 0.0405,
      "step": 5567
    },
    {
      "epoch": 0.8783373427455929,
      "grad_norm": 5.737876892089844,
      "learning_rate": 1.241345999033972e-07,
      "loss": 0.0442,
      "step": 5568
    },
    {
      "epoch": 0.8784950901131837,
      "grad_norm": 3.8690743446350098,
      "learning_rate": 1.239735952342618e-07,
      "loss": 0.0143,
      "step": 5569
    },
    {
      "epoch": 0.8786528374807745,
      "grad_norm": 3.8035223484039307,
      "learning_rate": 1.2381259056512637e-07,
      "loss": 0.0281,
      "step": 5570
    },
    {
      "epoch": 0.8788105848483654,
      "grad_norm": 4.356854438781738,
      "learning_rate": 1.2365158589599098e-07,
      "loss": 0.0248,
      "step": 5571
    },
    {
      "epoch": 0.8789683322159562,
      "grad_norm": 4.11252498626709,
      "learning_rate": 1.2349058122685556e-07,
      "loss": 0.0224,
      "step": 5572
    },
    {
      "epoch": 0.8791260795835469,
      "grad_norm": 4.7247843742370605,
      "learning_rate": 1.2332957655772017e-07,
      "loss": 0.0334,
      "step": 5573
    },
    {
      "epoch": 0.8792838269511377,
      "grad_norm": 7.541321754455566,
      "learning_rate": 1.2316857188858477e-07,
      "loss": 0.0474,
      "step": 5574
    },
    {
      "epoch": 0.8794415743187286,
      "grad_norm": 5.661472797393799,
      "learning_rate": 1.2300756721944935e-07,
      "loss": 0.0308,
      "step": 5575
    },
    {
      "epoch": 0.8795993216863194,
      "grad_norm": 2.8416740894317627,
      "learning_rate": 1.2284656255031396e-07,
      "loss": 0.0192,
      "step": 5576
    },
    {
      "epoch": 0.8797570690539102,
      "grad_norm": 10.018465995788574,
      "learning_rate": 1.2268555788117854e-07,
      "loss": 0.0559,
      "step": 5577
    },
    {
      "epoch": 0.8799148164215009,
      "grad_norm": 6.063082695007324,
      "learning_rate": 1.2252455321204314e-07,
      "loss": 0.0564,
      "step": 5578
    },
    {
      "epoch": 0.8800725637890918,
      "grad_norm": 4.855288982391357,
      "learning_rate": 1.2236354854290775e-07,
      "loss": 0.0629,
      "step": 5579
    },
    {
      "epoch": 0.8802303111566826,
      "grad_norm": 1.4225459098815918,
      "learning_rate": 1.2220254387377233e-07,
      "loss": 0.0077,
      "step": 5580
    },
    {
      "epoch": 0.8803880585242734,
      "grad_norm": 4.527211666107178,
      "learning_rate": 1.2204153920463694e-07,
      "loss": 0.0538,
      "step": 5581
    },
    {
      "epoch": 0.8805458058918642,
      "grad_norm": 7.736018180847168,
      "learning_rate": 1.2188053453550152e-07,
      "loss": 0.0576,
      "step": 5582
    },
    {
      "epoch": 0.8807035532594549,
      "grad_norm": 5.84458589553833,
      "learning_rate": 1.2171952986636612e-07,
      "loss": 0.0382,
      "step": 5583
    },
    {
      "epoch": 0.8808613006270458,
      "grad_norm": 2.6455740928649902,
      "learning_rate": 1.215585251972307e-07,
      "loss": 0.0397,
      "step": 5584
    },
    {
      "epoch": 0.8810190479946366,
      "grad_norm": 4.986097812652588,
      "learning_rate": 1.213975205280953e-07,
      "loss": 0.0612,
      "step": 5585
    },
    {
      "epoch": 0.8811767953622274,
      "grad_norm": 6.404510498046875,
      "learning_rate": 1.212365158589599e-07,
      "loss": 0.0222,
      "step": 5586
    },
    {
      "epoch": 0.8813345427298181,
      "grad_norm": 5.908301830291748,
      "learning_rate": 1.210755111898245e-07,
      "loss": 0.0553,
      "step": 5587
    },
    {
      "epoch": 0.881492290097409,
      "grad_norm": 7.566300392150879,
      "learning_rate": 1.209145065206891e-07,
      "loss": 0.0942,
      "step": 5588
    },
    {
      "epoch": 0.8816500374649998,
      "grad_norm": 2.8671958446502686,
      "learning_rate": 1.207535018515537e-07,
      "loss": 0.035,
      "step": 5589
    },
    {
      "epoch": 0.8818077848325906,
      "grad_norm": 3.029541015625,
      "learning_rate": 1.205924971824183e-07,
      "loss": 0.0174,
      "step": 5590
    },
    {
      "epoch": 0.8819655322001814,
      "grad_norm": 6.2882537841796875,
      "learning_rate": 1.204314925132829e-07,
      "loss": 0.0445,
      "step": 5591
    },
    {
      "epoch": 0.8821232795677723,
      "grad_norm": 9.018379211425781,
      "learning_rate": 1.2027048784414747e-07,
      "loss": 0.0387,
      "step": 5592
    },
    {
      "epoch": 0.882281026935363,
      "grad_norm": 3.9891650676727295,
      "learning_rate": 1.2010948317501208e-07,
      "loss": 0.0155,
      "step": 5593
    },
    {
      "epoch": 0.8824387743029538,
      "grad_norm": 3.620929479598999,
      "learning_rate": 1.1994847850587666e-07,
      "loss": 0.031,
      "step": 5594
    },
    {
      "epoch": 0.8825965216705446,
      "grad_norm": 3.1831936836242676,
      "learning_rate": 1.1978747383674127e-07,
      "loss": 0.0383,
      "step": 5595
    },
    {
      "epoch": 0.8827542690381355,
      "grad_norm": 4.132047176361084,
      "learning_rate": 1.1962646916760585e-07,
      "loss": 0.0222,
      "step": 5596
    },
    {
      "epoch": 0.8829120164057263,
      "grad_norm": 7.14155912399292,
      "learning_rate": 1.1946546449847045e-07,
      "loss": 0.0605,
      "step": 5597
    },
    {
      "epoch": 0.883069763773317,
      "grad_norm": 19.78788948059082,
      "learning_rate": 1.1930445982933503e-07,
      "loss": 0.0424,
      "step": 5598
    },
    {
      "epoch": 0.8832275111409078,
      "grad_norm": 2.621591091156006,
      "learning_rate": 1.1914345516019964e-07,
      "loss": 0.0312,
      "step": 5599
    },
    {
      "epoch": 0.8833852585084987,
      "grad_norm": 5.4610066413879395,
      "learning_rate": 1.1898245049106423e-07,
      "loss": 0.0424,
      "step": 5600
    },
    {
      "epoch": 0.8835430058760895,
      "grad_norm": 5.091216087341309,
      "learning_rate": 1.1882144582192884e-07,
      "loss": 0.0666,
      "step": 5601
    },
    {
      "epoch": 0.8837007532436802,
      "grad_norm": 7.490744113922119,
      "learning_rate": 1.1866044115279343e-07,
      "loss": 0.0576,
      "step": 5602
    },
    {
      "epoch": 0.883858500611271,
      "grad_norm": 6.292675495147705,
      "learning_rate": 1.1849943648365802e-07,
      "loss": 0.044,
      "step": 5603
    },
    {
      "epoch": 0.8840162479788618,
      "grad_norm": 4.969595909118652,
      "learning_rate": 1.1833843181452262e-07,
      "loss": 0.0326,
      "step": 5604
    },
    {
      "epoch": 0.8841739953464527,
      "grad_norm": 4.043492317199707,
      "learning_rate": 1.1817742714538721e-07,
      "loss": 0.0191,
      "step": 5605
    },
    {
      "epoch": 0.8843317427140435,
      "grad_norm": 4.971721649169922,
      "learning_rate": 1.180164224762518e-07,
      "loss": 0.0381,
      "step": 5606
    },
    {
      "epoch": 0.8844894900816342,
      "grad_norm": 3.6937193870544434,
      "learning_rate": 1.178554178071164e-07,
      "loss": 0.0471,
      "step": 5607
    },
    {
      "epoch": 0.884647237449225,
      "grad_norm": 7.5754714012146,
      "learning_rate": 1.1769441313798099e-07,
      "loss": 0.0814,
      "step": 5608
    },
    {
      "epoch": 0.8848049848168159,
      "grad_norm": 4.852900505065918,
      "learning_rate": 1.1753340846884558e-07,
      "loss": 0.0518,
      "step": 5609
    },
    {
      "epoch": 0.8849627321844067,
      "grad_norm": 4.317257404327393,
      "learning_rate": 1.1737240379971018e-07,
      "loss": 0.0588,
      "step": 5610
    },
    {
      "epoch": 0.8851204795519975,
      "grad_norm": 5.594083786010742,
      "learning_rate": 1.1721139913057478e-07,
      "loss": 0.0359,
      "step": 5611
    },
    {
      "epoch": 0.8852782269195882,
      "grad_norm": 3.8825714588165283,
      "learning_rate": 1.1705039446143938e-07,
      "loss": 0.0558,
      "step": 5612
    },
    {
      "epoch": 0.8854359742871791,
      "grad_norm": 4.055095195770264,
      "learning_rate": 1.1688938979230398e-07,
      "loss": 0.0386,
      "step": 5613
    },
    {
      "epoch": 0.8855937216547699,
      "grad_norm": 4.435608863830566,
      "learning_rate": 1.1672838512316857e-07,
      "loss": 0.0153,
      "step": 5614
    },
    {
      "epoch": 0.8857514690223607,
      "grad_norm": 3.319115161895752,
      "learning_rate": 1.1656738045403317e-07,
      "loss": 0.0179,
      "step": 5615
    },
    {
      "epoch": 0.8859092163899515,
      "grad_norm": 3.764820098876953,
      "learning_rate": 1.1640637578489776e-07,
      "loss": 0.0494,
      "step": 5616
    },
    {
      "epoch": 0.8860669637575423,
      "grad_norm": 4.6623406410217285,
      "learning_rate": 1.1624537111576235e-07,
      "loss": 0.0426,
      "step": 5617
    },
    {
      "epoch": 0.8862247111251331,
      "grad_norm": 4.292932510375977,
      "learning_rate": 1.1608436644662695e-07,
      "loss": 0.0578,
      "step": 5618
    },
    {
      "epoch": 0.8863824584927239,
      "grad_norm": 4.933620452880859,
      "learning_rate": 1.1592336177749154e-07,
      "loss": 0.0407,
      "step": 5619
    },
    {
      "epoch": 0.8865402058603147,
      "grad_norm": 0.9307185411453247,
      "learning_rate": 1.1576235710835613e-07,
      "loss": 0.0053,
      "step": 5620
    },
    {
      "epoch": 0.8866979532279055,
      "grad_norm": 4.064803600311279,
      "learning_rate": 1.1560135243922073e-07,
      "loss": 0.0326,
      "step": 5621
    },
    {
      "epoch": 0.8868557005954963,
      "grad_norm": 5.850526332855225,
      "learning_rate": 1.1544034777008532e-07,
      "loss": 0.0516,
      "step": 5622
    },
    {
      "epoch": 0.8870134479630871,
      "grad_norm": 3.1925485134124756,
      "learning_rate": 1.1527934310094991e-07,
      "loss": 0.0234,
      "step": 5623
    },
    {
      "epoch": 0.8871711953306779,
      "grad_norm": 7.044651508331299,
      "learning_rate": 1.1511833843181452e-07,
      "loss": 0.0467,
      "step": 5624
    },
    {
      "epoch": 0.8873289426982687,
      "grad_norm": 6.3519978523254395,
      "learning_rate": 1.1495733376267911e-07,
      "loss": 0.0638,
      "step": 5625
    },
    {
      "epoch": 0.8874866900658596,
      "grad_norm": 4.373105525970459,
      "learning_rate": 1.1479632909354372e-07,
      "loss": 0.0268,
      "step": 5626
    },
    {
      "epoch": 0.8876444374334503,
      "grad_norm": 7.837530612945557,
      "learning_rate": 1.1463532442440831e-07,
      "loss": 0.054,
      "step": 5627
    },
    {
      "epoch": 0.8878021848010411,
      "grad_norm": 3.8669676780700684,
      "learning_rate": 1.144743197552729e-07,
      "loss": 0.0259,
      "step": 5628
    },
    {
      "epoch": 0.8879599321686319,
      "grad_norm": 5.265133380889893,
      "learning_rate": 1.143133150861375e-07,
      "loss": 0.0153,
      "step": 5629
    },
    {
      "epoch": 0.8881176795362228,
      "grad_norm": 3.393134593963623,
      "learning_rate": 1.1415231041700209e-07,
      "loss": 0.0203,
      "step": 5630
    },
    {
      "epoch": 0.8882754269038136,
      "grad_norm": 3.8280105590820312,
      "learning_rate": 1.1399130574786668e-07,
      "loss": 0.0243,
      "step": 5631
    },
    {
      "epoch": 0.8884331742714043,
      "grad_norm": 5.004944324493408,
      "learning_rate": 1.1383030107873128e-07,
      "loss": 0.0711,
      "step": 5632
    },
    {
      "epoch": 0.8885909216389951,
      "grad_norm": 3.1511619091033936,
      "learning_rate": 1.1366929640959587e-07,
      "loss": 0.03,
      "step": 5633
    },
    {
      "epoch": 0.888748669006586,
      "grad_norm": 8.524422645568848,
      "learning_rate": 1.1350829174046046e-07,
      "loss": 0.0457,
      "step": 5634
    },
    {
      "epoch": 0.8889064163741768,
      "grad_norm": 5.389379024505615,
      "learning_rate": 1.1334728707132506e-07,
      "loss": 0.0539,
      "step": 5635
    },
    {
      "epoch": 0.8890641637417676,
      "grad_norm": 4.989780902862549,
      "learning_rate": 1.1318628240218966e-07,
      "loss": 0.0295,
      "step": 5636
    },
    {
      "epoch": 0.8892219111093583,
      "grad_norm": 21.108285903930664,
      "learning_rate": 1.1302527773305426e-07,
      "loss": 0.0565,
      "step": 5637
    },
    {
      "epoch": 0.8893796584769492,
      "grad_norm": 4.350667476654053,
      "learning_rate": 1.1286427306391885e-07,
      "loss": 0.0329,
      "step": 5638
    },
    {
      "epoch": 0.88953740584454,
      "grad_norm": 5.962086200714111,
      "learning_rate": 1.1270326839478344e-07,
      "loss": 0.0349,
      "step": 5639
    },
    {
      "epoch": 0.8896951532121308,
      "grad_norm": 8.710920333862305,
      "learning_rate": 1.1254226372564803e-07,
      "loss": 0.033,
      "step": 5640
    },
    {
      "epoch": 0.8898529005797216,
      "grad_norm": 3.4920713901519775,
      "learning_rate": 1.1238125905651263e-07,
      "loss": 0.0182,
      "step": 5641
    },
    {
      "epoch": 0.8900106479473123,
      "grad_norm": 5.161077499389648,
      "learning_rate": 1.1222025438737723e-07,
      "loss": 0.0682,
      "step": 5642
    },
    {
      "epoch": 0.8901683953149032,
      "grad_norm": 3.349120855331421,
      "learning_rate": 1.1205924971824183e-07,
      "loss": 0.0304,
      "step": 5643
    },
    {
      "epoch": 0.890326142682494,
      "grad_norm": 6.49302864074707,
      "learning_rate": 1.1189824504910642e-07,
      "loss": 0.0597,
      "step": 5644
    },
    {
      "epoch": 0.8904838900500848,
      "grad_norm": 4.746656894683838,
      "learning_rate": 1.1173724037997101e-07,
      "loss": 0.0291,
      "step": 5645
    },
    {
      "epoch": 0.8906416374176755,
      "grad_norm": 4.701689720153809,
      "learning_rate": 1.115762357108356e-07,
      "loss": 0.0465,
      "step": 5646
    },
    {
      "epoch": 0.8907993847852664,
      "grad_norm": 4.649779319763184,
      "learning_rate": 1.1141523104170021e-07,
      "loss": 0.0535,
      "step": 5647
    },
    {
      "epoch": 0.8909571321528572,
      "grad_norm": 3.2291858196258545,
      "learning_rate": 1.112542263725648e-07,
      "loss": 0.0285,
      "step": 5648
    },
    {
      "epoch": 0.891114879520448,
      "grad_norm": 4.260729789733887,
      "learning_rate": 1.110932217034294e-07,
      "loss": 0.0249,
      "step": 5649
    },
    {
      "epoch": 0.8912726268880388,
      "grad_norm": 3.859715700149536,
      "learning_rate": 1.1093221703429399e-07,
      "loss": 0.0289,
      "step": 5650
    },
    {
      "epoch": 0.8914303742556297,
      "grad_norm": 5.544247150421143,
      "learning_rate": 1.1077121236515858e-07,
      "loss": 0.0311,
      "step": 5651
    },
    {
      "epoch": 0.8915881216232204,
      "grad_norm": 4.745953559875488,
      "learning_rate": 1.1061020769602318e-07,
      "loss": 0.0807,
      "step": 5652
    },
    {
      "epoch": 0.8917458689908112,
      "grad_norm": 4.021941184997559,
      "learning_rate": 1.1044920302688777e-07,
      "loss": 0.0193,
      "step": 5653
    },
    {
      "epoch": 0.891903616358402,
      "grad_norm": 6.016777515411377,
      "learning_rate": 1.1028819835775236e-07,
      "loss": 0.0643,
      "step": 5654
    },
    {
      "epoch": 0.8920613637259929,
      "grad_norm": 3.4665982723236084,
      "learning_rate": 1.1012719368861696e-07,
      "loss": 0.0571,
      "step": 5655
    },
    {
      "epoch": 0.8922191110935836,
      "grad_norm": 2.344822406768799,
      "learning_rate": 1.0996618901948155e-07,
      "loss": 0.0413,
      "step": 5656
    },
    {
      "epoch": 0.8923768584611744,
      "grad_norm": 4.093752384185791,
      "learning_rate": 1.0980518435034616e-07,
      "loss": 0.0189,
      "step": 5657
    },
    {
      "epoch": 0.8925346058287652,
      "grad_norm": 6.18466854095459,
      "learning_rate": 1.0964417968121075e-07,
      "loss": 0.0839,
      "step": 5658
    },
    {
      "epoch": 0.892692353196356,
      "grad_norm": 2.5295774936676025,
      "learning_rate": 1.0948317501207536e-07,
      "loss": 0.0328,
      "step": 5659
    },
    {
      "epoch": 0.8928501005639469,
      "grad_norm": 6.288724422454834,
      "learning_rate": 1.0932217034293995e-07,
      "loss": 0.0529,
      "step": 5660
    },
    {
      "epoch": 0.8930078479315376,
      "grad_norm": 6.388190746307373,
      "learning_rate": 1.0916116567380454e-07,
      "loss": 0.0637,
      "step": 5661
    },
    {
      "epoch": 0.8931655952991284,
      "grad_norm": 4.875352382659912,
      "learning_rate": 1.0900016100466914e-07,
      "loss": 0.0607,
      "step": 5662
    },
    {
      "epoch": 0.8933233426667192,
      "grad_norm": 5.0760416984558105,
      "learning_rate": 1.0883915633553373e-07,
      "loss": 0.0463,
      "step": 5663
    },
    {
      "epoch": 0.8934810900343101,
      "grad_norm": 6.9283952713012695,
      "learning_rate": 1.0867815166639832e-07,
      "loss": 0.0366,
      "step": 5664
    },
    {
      "epoch": 0.8936388374019009,
      "grad_norm": 5.368336200714111,
      "learning_rate": 1.0851714699726291e-07,
      "loss": 0.0461,
      "step": 5665
    },
    {
      "epoch": 0.8937965847694916,
      "grad_norm": 2.2210261821746826,
      "learning_rate": 1.0835614232812751e-07,
      "loss": 0.0205,
      "step": 5666
    },
    {
      "epoch": 0.8939543321370824,
      "grad_norm": 4.281381607055664,
      "learning_rate": 1.081951376589921e-07,
      "loss": 0.0409,
      "step": 5667
    },
    {
      "epoch": 0.8941120795046733,
      "grad_norm": 4.237894058227539,
      "learning_rate": 1.080341329898567e-07,
      "loss": 0.0498,
      "step": 5668
    },
    {
      "epoch": 0.8942698268722641,
      "grad_norm": 3.750728130340576,
      "learning_rate": 1.0787312832072129e-07,
      "loss": 0.0535,
      "step": 5669
    },
    {
      "epoch": 0.8944275742398549,
      "grad_norm": 7.401185035705566,
      "learning_rate": 1.0771212365158588e-07,
      "loss": 0.0527,
      "step": 5670
    },
    {
      "epoch": 0.8945853216074456,
      "grad_norm": 5.04086446762085,
      "learning_rate": 1.0755111898245049e-07,
      "loss": 0.0443,
      "step": 5671
    },
    {
      "epoch": 0.8947430689750365,
      "grad_norm": 5.274075508117676,
      "learning_rate": 1.0739011431331509e-07,
      "loss": 0.0281,
      "step": 5672
    },
    {
      "epoch": 0.8949008163426273,
      "grad_norm": 6.070065975189209,
      "learning_rate": 1.0722910964417969e-07,
      "loss": 0.0435,
      "step": 5673
    },
    {
      "epoch": 0.8950585637102181,
      "grad_norm": 4.067622184753418,
      "learning_rate": 1.0706810497504428e-07,
      "loss": 0.0234,
      "step": 5674
    },
    {
      "epoch": 0.8952163110778089,
      "grad_norm": 5.839089393615723,
      "learning_rate": 1.0690710030590887e-07,
      "loss": 0.06,
      "step": 5675
    },
    {
      "epoch": 0.8953740584453997,
      "grad_norm": 5.103932857513428,
      "learning_rate": 1.0674609563677346e-07,
      "loss": 0.0285,
      "step": 5676
    },
    {
      "epoch": 0.8955318058129905,
      "grad_norm": 3.5581488609313965,
      "learning_rate": 1.0658509096763806e-07,
      "loss": 0.0353,
      "step": 5677
    },
    {
      "epoch": 0.8956895531805813,
      "grad_norm": 7.937819004058838,
      "learning_rate": 1.0642408629850265e-07,
      "loss": 0.0493,
      "step": 5678
    },
    {
      "epoch": 0.8958473005481721,
      "grad_norm": 7.44099760055542,
      "learning_rate": 1.0626308162936724e-07,
      "loss": 0.0642,
      "step": 5679
    },
    {
      "epoch": 0.8960050479157629,
      "grad_norm": 4.186953544616699,
      "learning_rate": 1.0610207696023184e-07,
      "loss": 0.0364,
      "step": 5680
    },
    {
      "epoch": 0.8961627952833537,
      "grad_norm": 3.0762956142425537,
      "learning_rate": 1.0594107229109643e-07,
      "loss": 0.0212,
      "step": 5681
    },
    {
      "epoch": 0.8963205426509445,
      "grad_norm": 5.375474452972412,
      "learning_rate": 1.0578006762196104e-07,
      "loss": 0.0589,
      "step": 5682
    },
    {
      "epoch": 0.8964782900185353,
      "grad_norm": 7.132761001586914,
      "learning_rate": 1.0561906295282563e-07,
      "loss": 0.0608,
      "step": 5683
    },
    {
      "epoch": 0.8966360373861261,
      "grad_norm": 5.897538661956787,
      "learning_rate": 1.0545805828369022e-07,
      "loss": 0.0642,
      "step": 5684
    },
    {
      "epoch": 0.896793784753717,
      "grad_norm": 5.504117965698242,
      "learning_rate": 1.0529705361455482e-07,
      "loss": 0.0168,
      "step": 5685
    },
    {
      "epoch": 0.8969515321213077,
      "grad_norm": 11.507904052734375,
      "learning_rate": 1.0513604894541941e-07,
      "loss": 0.0618,
      "step": 5686
    },
    {
      "epoch": 0.8971092794888985,
      "grad_norm": 7.014697074890137,
      "learning_rate": 1.04975044276284e-07,
      "loss": 0.0459,
      "step": 5687
    },
    {
      "epoch": 0.8972670268564893,
      "grad_norm": 3.058468818664551,
      "learning_rate": 1.0481403960714861e-07,
      "loss": 0.0183,
      "step": 5688
    },
    {
      "epoch": 0.8974247742240802,
      "grad_norm": 10.896299362182617,
      "learning_rate": 1.046530349380132e-07,
      "loss": 0.0667,
      "step": 5689
    },
    {
      "epoch": 0.897582521591671,
      "grad_norm": 5.112190246582031,
      "learning_rate": 1.044920302688778e-07,
      "loss": 0.0228,
      "step": 5690
    },
    {
      "epoch": 0.8977402689592617,
      "grad_norm": 3.9673049449920654,
      "learning_rate": 1.0433102559974239e-07,
      "loss": 0.0489,
      "step": 5691
    },
    {
      "epoch": 0.8978980163268525,
      "grad_norm": 6.478947639465332,
      "learning_rate": 1.0417002093060698e-07,
      "loss": 0.0684,
      "step": 5692
    },
    {
      "epoch": 0.8980557636944434,
      "grad_norm": 3.892296552658081,
      "learning_rate": 1.0400901626147157e-07,
      "loss": 0.0657,
      "step": 5693
    },
    {
      "epoch": 0.8982135110620342,
      "grad_norm": 4.397469520568848,
      "learning_rate": 1.0384801159233618e-07,
      "loss": 0.0319,
      "step": 5694
    },
    {
      "epoch": 0.898371258429625,
      "grad_norm": 4.551697731018066,
      "learning_rate": 1.0368700692320077e-07,
      "loss": 0.0425,
      "step": 5695
    },
    {
      "epoch": 0.8985290057972157,
      "grad_norm": 4.86274528503418,
      "learning_rate": 1.0352600225406537e-07,
      "loss": 0.0459,
      "step": 5696
    },
    {
      "epoch": 0.8986867531648066,
      "grad_norm": 3.842259168624878,
      "learning_rate": 1.0336499758492996e-07,
      "loss": 0.0317,
      "step": 5697
    },
    {
      "epoch": 0.8988445005323974,
      "grad_norm": 7.030752658843994,
      "learning_rate": 1.0320399291579455e-07,
      "loss": 0.0573,
      "step": 5698
    },
    {
      "epoch": 0.8990022478999882,
      "grad_norm": 5.519429683685303,
      "learning_rate": 1.0304298824665915e-07,
      "loss": 0.1221,
      "step": 5699
    },
    {
      "epoch": 0.899159995267579,
      "grad_norm": 7.13485860824585,
      "learning_rate": 1.0288198357752374e-07,
      "loss": 0.049,
      "step": 5700
    },
    {
      "epoch": 0.8993177426351697,
      "grad_norm": 1.5401393175125122,
      "learning_rate": 1.0272097890838833e-07,
      "loss": 0.0075,
      "step": 5701
    },
    {
      "epoch": 0.8994754900027606,
      "grad_norm": 4.875319480895996,
      "learning_rate": 1.0255997423925292e-07,
      "loss": 0.0361,
      "step": 5702
    },
    {
      "epoch": 0.8996332373703514,
      "grad_norm": 3.3554129600524902,
      "learning_rate": 1.0239896957011753e-07,
      "loss": 0.0293,
      "step": 5703
    },
    {
      "epoch": 0.8997909847379422,
      "grad_norm": 8.17493724822998,
      "learning_rate": 1.0223796490098212e-07,
      "loss": 0.0382,
      "step": 5704
    },
    {
      "epoch": 0.899948732105533,
      "grad_norm": 3.584365129470825,
      "learning_rate": 1.0207696023184672e-07,
      "loss": 0.0409,
      "step": 5705
    },
    {
      "epoch": 0.9001064794731238,
      "grad_norm": 4.148134231567383,
      "learning_rate": 1.0191595556271132e-07,
      "loss": 0.0425,
      "step": 5706
    },
    {
      "epoch": 0.9002642268407146,
      "grad_norm": 9.026497840881348,
      "learning_rate": 1.0175495089357592e-07,
      "loss": 0.0745,
      "step": 5707
    },
    {
      "epoch": 0.9004219742083054,
      "grad_norm": 8.715054512023926,
      "learning_rate": 1.0159394622444051e-07,
      "loss": 0.0645,
      "step": 5708
    },
    {
      "epoch": 0.9005797215758962,
      "grad_norm": 5.331550598144531,
      "learning_rate": 1.014329415553051e-07,
      "loss": 0.0175,
      "step": 5709
    },
    {
      "epoch": 0.900737468943487,
      "grad_norm": 4.127692699432373,
      "learning_rate": 1.012719368861697e-07,
      "loss": 0.0376,
      "step": 5710
    },
    {
      "epoch": 0.9008952163110778,
      "grad_norm": 6.854963302612305,
      "learning_rate": 1.0111093221703429e-07,
      "loss": 0.0876,
      "step": 5711
    },
    {
      "epoch": 0.9010529636786686,
      "grad_norm": 4.6490044593811035,
      "learning_rate": 1.0094992754789888e-07,
      "loss": 0.039,
      "step": 5712
    },
    {
      "epoch": 0.9012107110462594,
      "grad_norm": 6.645543098449707,
      "learning_rate": 1.0078892287876347e-07,
      "loss": 0.0349,
      "step": 5713
    },
    {
      "epoch": 0.9013684584138503,
      "grad_norm": 2.1745362281799316,
      "learning_rate": 1.0062791820962807e-07,
      "loss": 0.0121,
      "step": 5714
    },
    {
      "epoch": 0.901526205781441,
      "grad_norm": 2.395117998123169,
      "learning_rate": 1.0046691354049266e-07,
      "loss": 0.0068,
      "step": 5715
    },
    {
      "epoch": 0.9016839531490318,
      "grad_norm": 4.637762069702148,
      "learning_rate": 1.0030590887135725e-07,
      "loss": 0.0404,
      "step": 5716
    },
    {
      "epoch": 0.9018417005166226,
      "grad_norm": 2.9174087047576904,
      "learning_rate": 1.0014490420222186e-07,
      "loss": 0.0251,
      "step": 5717
    },
    {
      "epoch": 0.9019994478842134,
      "grad_norm": 4.064671039581299,
      "learning_rate": 9.998389953308647e-08,
      "loss": 0.0236,
      "step": 5718
    },
    {
      "epoch": 0.9021571952518043,
      "grad_norm": 8.09548568725586,
      "learning_rate": 9.982289486395106e-08,
      "loss": 0.0497,
      "step": 5719
    },
    {
      "epoch": 0.902314942619395,
      "grad_norm": 5.048455238342285,
      "learning_rate": 9.966189019481565e-08,
      "loss": 0.0369,
      "step": 5720
    },
    {
      "epoch": 0.9024726899869858,
      "grad_norm": 5.491619110107422,
      "learning_rate": 9.950088552568025e-08,
      "loss": 0.0354,
      "step": 5721
    },
    {
      "epoch": 0.9026304373545766,
      "grad_norm": 6.2281317710876465,
      "learning_rate": 9.933988085654484e-08,
      "loss": 0.0335,
      "step": 5722
    },
    {
      "epoch": 0.9027881847221675,
      "grad_norm": 1.4659321308135986,
      "learning_rate": 9.917887618740943e-08,
      "loss": 0.0068,
      "step": 5723
    },
    {
      "epoch": 0.9029459320897583,
      "grad_norm": 4.197894096374512,
      "learning_rate": 9.901787151827403e-08,
      "loss": 0.0271,
      "step": 5724
    },
    {
      "epoch": 0.903103679457349,
      "grad_norm": 3.937116861343384,
      "learning_rate": 9.885686684913862e-08,
      "loss": 0.0565,
      "step": 5725
    },
    {
      "epoch": 0.9032614268249398,
      "grad_norm": 2.79811429977417,
      "learning_rate": 9.869586218000321e-08,
      "loss": 0.0418,
      "step": 5726
    },
    {
      "epoch": 0.9034191741925307,
      "grad_norm": 4.082679271697998,
      "learning_rate": 9.85348575108678e-08,
      "loss": 0.0116,
      "step": 5727
    },
    {
      "epoch": 0.9035769215601215,
      "grad_norm": 5.4862446784973145,
      "learning_rate": 9.83738528417324e-08,
      "loss": 0.0516,
      "step": 5728
    },
    {
      "epoch": 0.9037346689277123,
      "grad_norm": 3.0281641483306885,
      "learning_rate": 9.8212848172597e-08,
      "loss": 0.0332,
      "step": 5729
    },
    {
      "epoch": 0.903892416295303,
      "grad_norm": 6.902345180511475,
      "learning_rate": 9.80518435034616e-08,
      "loss": 0.0528,
      "step": 5730
    },
    {
      "epoch": 0.9040501636628939,
      "grad_norm": 9.338648796081543,
      "learning_rate": 9.789083883432619e-08,
      "loss": 0.0614,
      "step": 5731
    },
    {
      "epoch": 0.9042079110304847,
      "grad_norm": 4.462130546569824,
      "learning_rate": 9.772983416519078e-08,
      "loss": 0.0512,
      "step": 5732
    },
    {
      "epoch": 0.9043656583980755,
      "grad_norm": 4.22948694229126,
      "learning_rate": 9.756882949605538e-08,
      "loss": 0.0714,
      "step": 5733
    },
    {
      "epoch": 0.9045234057656663,
      "grad_norm": 4.789216995239258,
      "learning_rate": 9.740782482691998e-08,
      "loss": 0.0626,
      "step": 5734
    },
    {
      "epoch": 0.9046811531332571,
      "grad_norm": 2.0657951831817627,
      "learning_rate": 9.724682015778458e-08,
      "loss": 0.0188,
      "step": 5735
    },
    {
      "epoch": 0.9048389005008479,
      "grad_norm": 3.498124122619629,
      "learning_rate": 9.708581548864917e-08,
      "loss": 0.0583,
      "step": 5736
    },
    {
      "epoch": 0.9049966478684387,
      "grad_norm": 2.890901565551758,
      "learning_rate": 9.692481081951376e-08,
      "loss": 0.0251,
      "step": 5737
    },
    {
      "epoch": 0.9051543952360295,
      "grad_norm": 5.114566326141357,
      "learning_rate": 9.676380615037835e-08,
      "loss": 0.0712,
      "step": 5738
    },
    {
      "epoch": 0.9053121426036203,
      "grad_norm": 4.458266735076904,
      "learning_rate": 9.660280148124295e-08,
      "loss": 0.0312,
      "step": 5739
    },
    {
      "epoch": 0.9054698899712111,
      "grad_norm": 4.075973987579346,
      "learning_rate": 9.644179681210755e-08,
      "loss": 0.0414,
      "step": 5740
    },
    {
      "epoch": 0.9056276373388019,
      "grad_norm": 4.164230823516846,
      "learning_rate": 9.628079214297215e-08,
      "loss": 0.0381,
      "step": 5741
    },
    {
      "epoch": 0.9057853847063927,
      "grad_norm": 2.8527257442474365,
      "learning_rate": 9.611978747383674e-08,
      "loss": 0.0186,
      "step": 5742
    },
    {
      "epoch": 0.9059431320739835,
      "grad_norm": 2.837355136871338,
      "learning_rate": 9.595878280470133e-08,
      "loss": 0.0278,
      "step": 5743
    },
    {
      "epoch": 0.9061008794415744,
      "grad_norm": 4.097024917602539,
      "learning_rate": 9.579777813556593e-08,
      "loss": 0.0384,
      "step": 5744
    },
    {
      "epoch": 0.9062586268091651,
      "grad_norm": 5.411277770996094,
      "learning_rate": 9.563677346643052e-08,
      "loss": 0.0378,
      "step": 5745
    },
    {
      "epoch": 0.9064163741767559,
      "grad_norm": 11.387405395507812,
      "learning_rate": 9.547576879729511e-08,
      "loss": 0.0499,
      "step": 5746
    },
    {
      "epoch": 0.9065741215443467,
      "grad_norm": 3.678576707839966,
      "learning_rate": 9.53147641281597e-08,
      "loss": 0.0359,
      "step": 5747
    },
    {
      "epoch": 0.9067318689119376,
      "grad_norm": 4.650317668914795,
      "learning_rate": 9.51537594590243e-08,
      "loss": 0.0657,
      "step": 5748
    },
    {
      "epoch": 0.9068896162795284,
      "grad_norm": 2.559847116470337,
      "learning_rate": 9.49927547898889e-08,
      "loss": 0.0326,
      "step": 5749
    },
    {
      "epoch": 0.9070473636471191,
      "grad_norm": 5.558595180511475,
      "learning_rate": 9.48317501207535e-08,
      "loss": 0.0775,
      "step": 5750
    },
    {
      "epoch": 0.9072051110147099,
      "grad_norm": 3.4408504962921143,
      "learning_rate": 9.467074545161809e-08,
      "loss": 0.0156,
      "step": 5751
    },
    {
      "epoch": 0.9073628583823008,
      "grad_norm": 3.8436343669891357,
      "learning_rate": 9.45097407824827e-08,
      "loss": 0.0348,
      "step": 5752
    },
    {
      "epoch": 0.9075206057498916,
      "grad_norm": 4.63906192779541,
      "learning_rate": 9.434873611334729e-08,
      "loss": 0.0233,
      "step": 5753
    },
    {
      "epoch": 0.9076783531174824,
      "grad_norm": 4.913653373718262,
      "learning_rate": 9.418773144421188e-08,
      "loss": 0.0455,
      "step": 5754
    },
    {
      "epoch": 0.9078361004850731,
      "grad_norm": 6.2280707359313965,
      "learning_rate": 9.402672677507648e-08,
      "loss": 0.087,
      "step": 5755
    },
    {
      "epoch": 0.9079938478526639,
      "grad_norm": 5.642157554626465,
      "learning_rate": 9.386572210594107e-08,
      "loss": 0.0851,
      "step": 5756
    },
    {
      "epoch": 0.9081515952202548,
      "grad_norm": 5.007143020629883,
      "learning_rate": 9.370471743680566e-08,
      "loss": 0.0507,
      "step": 5757
    },
    {
      "epoch": 0.9083093425878456,
      "grad_norm": 3.5131096839904785,
      "learning_rate": 9.354371276767026e-08,
      "loss": 0.0378,
      "step": 5758
    },
    {
      "epoch": 0.9084670899554363,
      "grad_norm": 4.084439277648926,
      "learning_rate": 9.338270809853485e-08,
      "loss": 0.0372,
      "step": 5759
    },
    {
      "epoch": 0.9086248373230271,
      "grad_norm": 3.037368059158325,
      "learning_rate": 9.322170342939944e-08,
      "loss": 0.0143,
      "step": 5760
    },
    {
      "epoch": 0.908782584690618,
      "grad_norm": 2.3745875358581543,
      "learning_rate": 9.306069876026404e-08,
      "loss": 0.0217,
      "step": 5761
    },
    {
      "epoch": 0.9089403320582088,
      "grad_norm": 4.520737171173096,
      "learning_rate": 9.289969409112863e-08,
      "loss": 0.0823,
      "step": 5762
    },
    {
      "epoch": 0.9090980794257996,
      "grad_norm": 5.6762542724609375,
      "learning_rate": 9.273868942199322e-08,
      "loss": 0.0649,
      "step": 5763
    },
    {
      "epoch": 0.9092558267933903,
      "grad_norm": 7.527730464935303,
      "learning_rate": 9.257768475285784e-08,
      "loss": 0.0406,
      "step": 5764
    },
    {
      "epoch": 0.9094135741609812,
      "grad_norm": 2.8972980976104736,
      "learning_rate": 9.241668008372243e-08,
      "loss": 0.02,
      "step": 5765
    },
    {
      "epoch": 0.909571321528572,
      "grad_norm": 3.597255229949951,
      "learning_rate": 9.225567541458703e-08,
      "loss": 0.0402,
      "step": 5766
    },
    {
      "epoch": 0.9097290688961628,
      "grad_norm": 5.695924758911133,
      "learning_rate": 9.209467074545162e-08,
      "loss": 0.051,
      "step": 5767
    },
    {
      "epoch": 0.9098868162637536,
      "grad_norm": 6.006032943725586,
      "learning_rate": 9.193366607631621e-08,
      "loss": 0.0357,
      "step": 5768
    },
    {
      "epoch": 0.9100445636313444,
      "grad_norm": 2.889230728149414,
      "learning_rate": 9.17726614071808e-08,
      "loss": 0.0176,
      "step": 5769
    },
    {
      "epoch": 0.9102023109989352,
      "grad_norm": 3.2964913845062256,
      "learning_rate": 9.16116567380454e-08,
      "loss": 0.0093,
      "step": 5770
    },
    {
      "epoch": 0.910360058366526,
      "grad_norm": 3.3851239681243896,
      "learning_rate": 9.145065206890999e-08,
      "loss": 0.0181,
      "step": 5771
    },
    {
      "epoch": 0.9105178057341168,
      "grad_norm": 4.152960300445557,
      "learning_rate": 9.128964739977459e-08,
      "loss": 0.0339,
      "step": 5772
    },
    {
      "epoch": 0.9106755531017077,
      "grad_norm": 6.54844331741333,
      "learning_rate": 9.112864273063918e-08,
      "loss": 0.0459,
      "step": 5773
    },
    {
      "epoch": 0.9108333004692984,
      "grad_norm": 4.084027290344238,
      "learning_rate": 9.096763806150377e-08,
      "loss": 0.0379,
      "step": 5774
    },
    {
      "epoch": 0.9109910478368892,
      "grad_norm": 6.559889316558838,
      "learning_rate": 9.080663339236838e-08,
      "loss": 0.045,
      "step": 5775
    },
    {
      "epoch": 0.91114879520448,
      "grad_norm": 2.890805721282959,
      "learning_rate": 9.064562872323297e-08,
      "loss": 0.022,
      "step": 5776
    },
    {
      "epoch": 0.9113065425720708,
      "grad_norm": 3.939265251159668,
      "learning_rate": 9.048462405409756e-08,
      "loss": 0.0385,
      "step": 5777
    },
    {
      "epoch": 0.9114642899396617,
      "grad_norm": 4.184532642364502,
      "learning_rate": 9.032361938496216e-08,
      "loss": 0.0332,
      "step": 5778
    },
    {
      "epoch": 0.9116220373072524,
      "grad_norm": 3.5728583335876465,
      "learning_rate": 9.016261471582675e-08,
      "loss": 0.0374,
      "step": 5779
    },
    {
      "epoch": 0.9117797846748432,
      "grad_norm": 5.712306499481201,
      "learning_rate": 9.000161004669136e-08,
      "loss": 0.0582,
      "step": 5780
    },
    {
      "epoch": 0.911937532042434,
      "grad_norm": 4.514135360717773,
      "learning_rate": 8.984060537755595e-08,
      "loss": 0.0317,
      "step": 5781
    },
    {
      "epoch": 0.9120952794100249,
      "grad_norm": 4.458296775817871,
      "learning_rate": 8.967960070842054e-08,
      "loss": 0.049,
      "step": 5782
    },
    {
      "epoch": 0.9122530267776157,
      "grad_norm": 2.0265932083129883,
      "learning_rate": 8.951859603928514e-08,
      "loss": 0.0106,
      "step": 5783
    },
    {
      "epoch": 0.9124107741452064,
      "grad_norm": 2.6315975189208984,
      "learning_rate": 8.935759137014973e-08,
      "loss": 0.0152,
      "step": 5784
    },
    {
      "epoch": 0.9125685215127972,
      "grad_norm": 4.49893045425415,
      "learning_rate": 8.919658670101432e-08,
      "loss": 0.0805,
      "step": 5785
    },
    {
      "epoch": 0.9127262688803881,
      "grad_norm": 7.787235736846924,
      "learning_rate": 8.903558203187892e-08,
      "loss": 0.0452,
      "step": 5786
    },
    {
      "epoch": 0.9128840162479789,
      "grad_norm": 5.546772003173828,
      "learning_rate": 8.887457736274352e-08,
      "loss": 0.0824,
      "step": 5787
    },
    {
      "epoch": 0.9130417636155697,
      "grad_norm": 5.021836280822754,
      "learning_rate": 8.871357269360811e-08,
      "loss": 0.0442,
      "step": 5788
    },
    {
      "epoch": 0.9131995109831604,
      "grad_norm": 3.255812883377075,
      "learning_rate": 8.855256802447271e-08,
      "loss": 0.0178,
      "step": 5789
    },
    {
      "epoch": 0.9133572583507513,
      "grad_norm": 5.3505096435546875,
      "learning_rate": 8.83915633553373e-08,
      "loss": 0.0491,
      "step": 5790
    },
    {
      "epoch": 0.9135150057183421,
      "grad_norm": 4.129081726074219,
      "learning_rate": 8.82305586862019e-08,
      "loss": 0.0313,
      "step": 5791
    },
    {
      "epoch": 0.9136727530859329,
      "grad_norm": 5.043010711669922,
      "learning_rate": 8.806955401706649e-08,
      "loss": 0.0213,
      "step": 5792
    },
    {
      "epoch": 0.9138305004535237,
      "grad_norm": 2.7486824989318848,
      "learning_rate": 8.790854934793108e-08,
      "loss": 0.0356,
      "step": 5793
    },
    {
      "epoch": 0.9139882478211144,
      "grad_norm": 2.983541250228882,
      "learning_rate": 8.774754467879567e-08,
      "loss": 0.0216,
      "step": 5794
    },
    {
      "epoch": 0.9141459951887053,
      "grad_norm": 2.1109204292297363,
      "learning_rate": 8.758654000966028e-08,
      "loss": 0.058,
      "step": 5795
    },
    {
      "epoch": 0.9143037425562961,
      "grad_norm": 3.1736154556274414,
      "learning_rate": 8.742553534052487e-08,
      "loss": 0.0378,
      "step": 5796
    },
    {
      "epoch": 0.9144614899238869,
      "grad_norm": 6.466217517852783,
      "learning_rate": 8.726453067138947e-08,
      "loss": 0.0473,
      "step": 5797
    },
    {
      "epoch": 0.9146192372914776,
      "grad_norm": 3.0933151245117188,
      "learning_rate": 8.710352600225406e-08,
      "loss": 0.0267,
      "step": 5798
    },
    {
      "epoch": 0.9147769846590685,
      "grad_norm": 7.6968770027160645,
      "learning_rate": 8.694252133311866e-08,
      "loss": 0.0586,
      "step": 5799
    },
    {
      "epoch": 0.9149347320266593,
      "grad_norm": 4.800207614898682,
      "learning_rate": 8.678151666398326e-08,
      "loss": 0.0374,
      "step": 5800
    },
    {
      "epoch": 0.9150924793942501,
      "grad_norm": 4.6078314781188965,
      "learning_rate": 8.662051199484785e-08,
      "loss": 0.0229,
      "step": 5801
    },
    {
      "epoch": 0.9152502267618409,
      "grad_norm": 2.8267221450805664,
      "learning_rate": 8.645950732571244e-08,
      "loss": 0.0293,
      "step": 5802
    },
    {
      "epoch": 0.9154079741294318,
      "grad_norm": 5.6814141273498535,
      "learning_rate": 8.629850265657704e-08,
      "loss": 0.0464,
      "step": 5803
    },
    {
      "epoch": 0.9155657214970225,
      "grad_norm": 10.271818161010742,
      "learning_rate": 8.613749798744163e-08,
      "loss": 0.0479,
      "step": 5804
    },
    {
      "epoch": 0.9157234688646133,
      "grad_norm": 5.682190418243408,
      "learning_rate": 8.597649331830622e-08,
      "loss": 0.0398,
      "step": 5805
    },
    {
      "epoch": 0.9158812162322041,
      "grad_norm": 5.16942834854126,
      "learning_rate": 8.581548864917082e-08,
      "loss": 0.0333,
      "step": 5806
    },
    {
      "epoch": 0.916038963599795,
      "grad_norm": 3.6980783939361572,
      "learning_rate": 8.565448398003541e-08,
      "loss": 0.0293,
      "step": 5807
    },
    {
      "epoch": 0.9161967109673858,
      "grad_norm": 5.025379657745361,
      "learning_rate": 8.54934793109e-08,
      "loss": 0.0434,
      "step": 5808
    },
    {
      "epoch": 0.9163544583349765,
      "grad_norm": 0.8971937298774719,
      "learning_rate": 8.53324746417646e-08,
      "loss": 0.0076,
      "step": 5809
    },
    {
      "epoch": 0.9165122057025673,
      "grad_norm": 7.095747947692871,
      "learning_rate": 8.517146997262922e-08,
      "loss": 0.0386,
      "step": 5810
    },
    {
      "epoch": 0.9166699530701582,
      "grad_norm": 2.7633774280548096,
      "learning_rate": 8.501046530349381e-08,
      "loss": 0.0246,
      "step": 5811
    },
    {
      "epoch": 0.916827700437749,
      "grad_norm": 6.486352920532227,
      "learning_rate": 8.48494606343584e-08,
      "loss": 0.0559,
      "step": 5812
    },
    {
      "epoch": 0.9169854478053397,
      "grad_norm": 7.435791015625,
      "learning_rate": 8.4688455965223e-08,
      "loss": 0.0635,
      "step": 5813
    },
    {
      "epoch": 0.9171431951729305,
      "grad_norm": 5.165439128875732,
      "learning_rate": 8.452745129608759e-08,
      "loss": 0.0384,
      "step": 5814
    },
    {
      "epoch": 0.9173009425405213,
      "grad_norm": 7.617971420288086,
      "learning_rate": 8.436644662695218e-08,
      "loss": 0.0721,
      "step": 5815
    },
    {
      "epoch": 0.9174586899081122,
      "grad_norm": 5.212454319000244,
      "learning_rate": 8.420544195781677e-08,
      "loss": 0.0636,
      "step": 5816
    },
    {
      "epoch": 0.917616437275703,
      "grad_norm": 6.4869489669799805,
      "learning_rate": 8.404443728868137e-08,
      "loss": 0.0487,
      "step": 5817
    },
    {
      "epoch": 0.9177741846432937,
      "grad_norm": 3.043604850769043,
      "learning_rate": 8.388343261954596e-08,
      "loss": 0.0208,
      "step": 5818
    },
    {
      "epoch": 0.9179319320108845,
      "grad_norm": 5.8964667320251465,
      "learning_rate": 8.372242795041055e-08,
      "loss": 0.0658,
      "step": 5819
    },
    {
      "epoch": 0.9180896793784754,
      "grad_norm": 6.04457950592041,
      "learning_rate": 8.356142328127515e-08,
      "loss": 0.053,
      "step": 5820
    },
    {
      "epoch": 0.9182474267460662,
      "grad_norm": 9.761399269104004,
      "learning_rate": 8.340041861213974e-08,
      "loss": 0.0687,
      "step": 5821
    },
    {
      "epoch": 0.918405174113657,
      "grad_norm": 2.7290096282958984,
      "learning_rate": 8.323941394300435e-08,
      "loss": 0.0194,
      "step": 5822
    },
    {
      "epoch": 0.9185629214812477,
      "grad_norm": 3.8870458602905273,
      "learning_rate": 8.307840927386894e-08,
      "loss": 0.0371,
      "step": 5823
    },
    {
      "epoch": 0.9187206688488386,
      "grad_norm": 4.762032985687256,
      "learning_rate": 8.291740460473353e-08,
      "loss": 0.0522,
      "step": 5824
    },
    {
      "epoch": 0.9188784162164294,
      "grad_norm": 4.974912643432617,
      "learning_rate": 8.275639993559812e-08,
      "loss": 0.0464,
      "step": 5825
    },
    {
      "epoch": 0.9190361635840202,
      "grad_norm": 4.439513206481934,
      "learning_rate": 8.259539526646273e-08,
      "loss": 0.0447,
      "step": 5826
    },
    {
      "epoch": 0.919193910951611,
      "grad_norm": 5.005949974060059,
      "learning_rate": 8.243439059732732e-08,
      "loss": 0.04,
      "step": 5827
    },
    {
      "epoch": 0.9193516583192018,
      "grad_norm": 6.744287014007568,
      "learning_rate": 8.227338592819192e-08,
      "loss": 0.0569,
      "step": 5828
    },
    {
      "epoch": 0.9195094056867926,
      "grad_norm": 7.064107894897461,
      "learning_rate": 8.211238125905651e-08,
      "loss": 0.0651,
      "step": 5829
    },
    {
      "epoch": 0.9196671530543834,
      "grad_norm": 6.484222888946533,
      "learning_rate": 8.19513765899211e-08,
      "loss": 0.0483,
      "step": 5830
    },
    {
      "epoch": 0.9198249004219742,
      "grad_norm": 5.973041534423828,
      "learning_rate": 8.17903719207857e-08,
      "loss": 0.0357,
      "step": 5831
    },
    {
      "epoch": 0.9199826477895651,
      "grad_norm": 11.105122566223145,
      "learning_rate": 8.162936725165029e-08,
      "loss": 0.0715,
      "step": 5832
    },
    {
      "epoch": 0.9201403951571558,
      "grad_norm": 6.121046543121338,
      "learning_rate": 8.146836258251488e-08,
      "loss": 0.0478,
      "step": 5833
    },
    {
      "epoch": 0.9202981425247466,
      "grad_norm": 4.047852039337158,
      "learning_rate": 8.130735791337949e-08,
      "loss": 0.0526,
      "step": 5834
    },
    {
      "epoch": 0.9204558898923374,
      "grad_norm": 7.392566204071045,
      "learning_rate": 8.114635324424408e-08,
      "loss": 0.0226,
      "step": 5835
    },
    {
      "epoch": 0.9206136372599282,
      "grad_norm": 5.444931983947754,
      "learning_rate": 8.098534857510868e-08,
      "loss": 0.034,
      "step": 5836
    },
    {
      "epoch": 0.9207713846275191,
      "grad_norm": 6.684887886047363,
      "learning_rate": 8.082434390597327e-08,
      "loss": 0.0243,
      "step": 5837
    },
    {
      "epoch": 0.9209291319951098,
      "grad_norm": 6.164416790008545,
      "learning_rate": 8.066333923683786e-08,
      "loss": 0.0401,
      "step": 5838
    },
    {
      "epoch": 0.9210868793627006,
      "grad_norm": 4.6397175788879395,
      "learning_rate": 8.050233456770245e-08,
      "loss": 0.0423,
      "step": 5839
    },
    {
      "epoch": 0.9212446267302914,
      "grad_norm": 5.3267340660095215,
      "learning_rate": 8.034132989856705e-08,
      "loss": 0.0586,
      "step": 5840
    },
    {
      "epoch": 0.9214023740978823,
      "grad_norm": 6.337176322937012,
      "learning_rate": 8.018032522943165e-08,
      "loss": 0.0669,
      "step": 5841
    },
    {
      "epoch": 0.9215601214654731,
      "grad_norm": 6.352117538452148,
      "learning_rate": 8.001932056029625e-08,
      "loss": 0.0827,
      "step": 5842
    },
    {
      "epoch": 0.9217178688330638,
      "grad_norm": 4.213643550872803,
      "learning_rate": 7.985831589116084e-08,
      "loss": 0.0245,
      "step": 5843
    },
    {
      "epoch": 0.9218756162006546,
      "grad_norm": 3.4744551181793213,
      "learning_rate": 7.969731122202543e-08,
      "loss": 0.0189,
      "step": 5844
    },
    {
      "epoch": 0.9220333635682455,
      "grad_norm": 4.864378929138184,
      "learning_rate": 7.953630655289004e-08,
      "loss": 0.0386,
      "step": 5845
    },
    {
      "epoch": 0.9221911109358363,
      "grad_norm": 4.443397045135498,
      "learning_rate": 7.937530188375463e-08,
      "loss": 0.0565,
      "step": 5846
    },
    {
      "epoch": 0.922348858303427,
      "grad_norm": 3.475379228591919,
      "learning_rate": 7.921429721461923e-08,
      "loss": 0.0184,
      "step": 5847
    },
    {
      "epoch": 0.9225066056710178,
      "grad_norm": 6.629957675933838,
      "learning_rate": 7.905329254548382e-08,
      "loss": 0.0682,
      "step": 5848
    },
    {
      "epoch": 0.9226643530386087,
      "grad_norm": 9.265862464904785,
      "learning_rate": 7.889228787634841e-08,
      "loss": 0.0616,
      "step": 5849
    },
    {
      "epoch": 0.9228221004061995,
      "grad_norm": 7.067864418029785,
      "learning_rate": 7.8731283207213e-08,
      "loss": 0.0488,
      "step": 5850
    },
    {
      "epoch": 0.9229798477737903,
      "grad_norm": 8.622649192810059,
      "learning_rate": 7.85702785380776e-08,
      "loss": 0.0287,
      "step": 5851
    },
    {
      "epoch": 0.923137595141381,
      "grad_norm": 3.200631618499756,
      "learning_rate": 7.840927386894219e-08,
      "loss": 0.0204,
      "step": 5852
    },
    {
      "epoch": 0.9232953425089718,
      "grad_norm": 6.694149017333984,
      "learning_rate": 7.824826919980678e-08,
      "loss": 0.0626,
      "step": 5853
    },
    {
      "epoch": 0.9234530898765627,
      "grad_norm": 3.36803936958313,
      "learning_rate": 7.808726453067138e-08,
      "loss": 0.0211,
      "step": 5854
    },
    {
      "epoch": 0.9236108372441535,
      "grad_norm": 5.772575855255127,
      "learning_rate": 7.792625986153597e-08,
      "loss": 0.0626,
      "step": 5855
    },
    {
      "epoch": 0.9237685846117443,
      "grad_norm": 3.5682120323181152,
      "learning_rate": 7.776525519240056e-08,
      "loss": 0.0298,
      "step": 5856
    },
    {
      "epoch": 0.923926331979335,
      "grad_norm": 9.057229042053223,
      "learning_rate": 7.760425052326518e-08,
      "loss": 0.0443,
      "step": 5857
    },
    {
      "epoch": 0.9240840793469259,
      "grad_norm": 8.135175704956055,
      "learning_rate": 7.744324585412978e-08,
      "loss": 0.0441,
      "step": 5858
    },
    {
      "epoch": 0.9242418267145167,
      "grad_norm": 3.027372121810913,
      "learning_rate": 7.728224118499437e-08,
      "loss": 0.0251,
      "step": 5859
    },
    {
      "epoch": 0.9243995740821075,
      "grad_norm": 16.04706382751465,
      "learning_rate": 7.712123651585896e-08,
      "loss": 0.0292,
      "step": 5860
    },
    {
      "epoch": 0.9245573214496983,
      "grad_norm": 2.796297550201416,
      "learning_rate": 7.696023184672356e-08,
      "loss": 0.0483,
      "step": 5861
    },
    {
      "epoch": 0.9247150688172892,
      "grad_norm": 7.089400291442871,
      "learning_rate": 7.679922717758815e-08,
      "loss": 0.0588,
      "step": 5862
    },
    {
      "epoch": 0.9248728161848799,
      "grad_norm": 11.826553344726562,
      "learning_rate": 7.663822250845274e-08,
      "loss": 0.0606,
      "step": 5863
    },
    {
      "epoch": 0.9250305635524707,
      "grad_norm": 4.3761677742004395,
      "learning_rate": 7.647721783931733e-08,
      "loss": 0.0277,
      "step": 5864
    },
    {
      "epoch": 0.9251883109200615,
      "grad_norm": 6.777541637420654,
      "learning_rate": 7.631621317018193e-08,
      "loss": 0.0583,
      "step": 5865
    },
    {
      "epoch": 0.9253460582876524,
      "grad_norm": 2.7931265830993652,
      "learning_rate": 7.615520850104652e-08,
      "loss": 0.0378,
      "step": 5866
    },
    {
      "epoch": 0.9255038056552432,
      "grad_norm": 4.795775413513184,
      "learning_rate": 7.599420383191111e-08,
      "loss": 0.0494,
      "step": 5867
    },
    {
      "epoch": 0.9256615530228339,
      "grad_norm": 4.076138019561768,
      "learning_rate": 7.583319916277572e-08,
      "loss": 0.0381,
      "step": 5868
    },
    {
      "epoch": 0.9258193003904247,
      "grad_norm": 7.07107400894165,
      "learning_rate": 7.567219449364031e-08,
      "loss": 0.0688,
      "step": 5869
    },
    {
      "epoch": 0.9259770477580156,
      "grad_norm": 4.784503936767578,
      "learning_rate": 7.55111898245049e-08,
      "loss": 0.0284,
      "step": 5870
    },
    {
      "epoch": 0.9261347951256064,
      "grad_norm": 2.3995094299316406,
      "learning_rate": 7.53501851553695e-08,
      "loss": 0.033,
      "step": 5871
    },
    {
      "epoch": 0.9262925424931971,
      "grad_norm": 5.072754383087158,
      "learning_rate": 7.51891804862341e-08,
      "loss": 0.0532,
      "step": 5872
    },
    {
      "epoch": 0.9264502898607879,
      "grad_norm": 3.824201822280884,
      "learning_rate": 7.50281758170987e-08,
      "loss": 0.0318,
      "step": 5873
    },
    {
      "epoch": 0.9266080372283787,
      "grad_norm": 5.722204208374023,
      "learning_rate": 7.486717114796329e-08,
      "loss": 0.0457,
      "step": 5874
    },
    {
      "epoch": 0.9267657845959696,
      "grad_norm": 18.639663696289062,
      "learning_rate": 7.470616647882788e-08,
      "loss": 0.0937,
      "step": 5875
    },
    {
      "epoch": 0.9269235319635604,
      "grad_norm": 2.2406606674194336,
      "learning_rate": 7.454516180969248e-08,
      "loss": 0.014,
      "step": 5876
    },
    {
      "epoch": 0.9270812793311511,
      "grad_norm": 4.865085601806641,
      "learning_rate": 7.438415714055707e-08,
      "loss": 0.0375,
      "step": 5877
    },
    {
      "epoch": 0.9272390266987419,
      "grad_norm": 4.385605335235596,
      "learning_rate": 7.422315247142166e-08,
      "loss": 0.0146,
      "step": 5878
    },
    {
      "epoch": 0.9273967740663328,
      "grad_norm": 4.06872034072876,
      "learning_rate": 7.406214780228626e-08,
      "loss": 0.0286,
      "step": 5879
    },
    {
      "epoch": 0.9275545214339236,
      "grad_norm": 10.020902633666992,
      "learning_rate": 7.390114313315086e-08,
      "loss": 0.0696,
      "step": 5880
    },
    {
      "epoch": 0.9277122688015144,
      "grad_norm": 3.6544883251190186,
      "learning_rate": 7.374013846401546e-08,
      "loss": 0.0246,
      "step": 5881
    },
    {
      "epoch": 0.9278700161691051,
      "grad_norm": 6.031393051147461,
      "learning_rate": 7.357913379488005e-08,
      "loss": 0.0259,
      "step": 5882
    },
    {
      "epoch": 0.928027763536696,
      "grad_norm": 3.699489116668701,
      "learning_rate": 7.341812912574464e-08,
      "loss": 0.0367,
      "step": 5883
    },
    {
      "epoch": 0.9281855109042868,
      "grad_norm": 4.587629318237305,
      "learning_rate": 7.325712445660924e-08,
      "loss": 0.0326,
      "step": 5884
    },
    {
      "epoch": 0.9283432582718776,
      "grad_norm": 1.859129548072815,
      "learning_rate": 7.309611978747383e-08,
      "loss": 0.0118,
      "step": 5885
    },
    {
      "epoch": 0.9285010056394684,
      "grad_norm": 4.384711742401123,
      "learning_rate": 7.293511511833842e-08,
      "loss": 0.0323,
      "step": 5886
    },
    {
      "epoch": 0.9286587530070592,
      "grad_norm": 7.893499851226807,
      "learning_rate": 7.277411044920303e-08,
      "loss": 0.1054,
      "step": 5887
    },
    {
      "epoch": 0.92881650037465,
      "grad_norm": 16.19036293029785,
      "learning_rate": 7.261310578006762e-08,
      "loss": 0.15,
      "step": 5888
    },
    {
      "epoch": 0.9289742477422408,
      "grad_norm": 4.184976100921631,
      "learning_rate": 7.245210111093221e-08,
      "loss": 0.0512,
      "step": 5889
    },
    {
      "epoch": 0.9291319951098316,
      "grad_norm": 5.087081432342529,
      "learning_rate": 7.229109644179681e-08,
      "loss": 0.0444,
      "step": 5890
    },
    {
      "epoch": 0.9292897424774224,
      "grad_norm": 6.60899543762207,
      "learning_rate": 7.21300917726614e-08,
      "loss": 0.026,
      "step": 5891
    },
    {
      "epoch": 0.9294474898450132,
      "grad_norm": 2.9725310802459717,
      "learning_rate": 7.196908710352601e-08,
      "loss": 0.0231,
      "step": 5892
    },
    {
      "epoch": 0.929605237212604,
      "grad_norm": 6.0414910316467285,
      "learning_rate": 7.18080824343906e-08,
      "loss": 0.0713,
      "step": 5893
    },
    {
      "epoch": 0.9297629845801948,
      "grad_norm": 4.175647735595703,
      "learning_rate": 7.164707776525519e-08,
      "loss": 0.023,
      "step": 5894
    },
    {
      "epoch": 0.9299207319477856,
      "grad_norm": 6.7042236328125,
      "learning_rate": 7.148607309611979e-08,
      "loss": 0.0254,
      "step": 5895
    },
    {
      "epoch": 0.9300784793153765,
      "grad_norm": 33.338382720947266,
      "learning_rate": 7.132506842698438e-08,
      "loss": 0.0215,
      "step": 5896
    },
    {
      "epoch": 0.9302362266829672,
      "grad_norm": 7.470001220703125,
      "learning_rate": 7.116406375784897e-08,
      "loss": 0.0527,
      "step": 5897
    },
    {
      "epoch": 0.930393974050558,
      "grad_norm": 4.17733907699585,
      "learning_rate": 7.100305908871357e-08,
      "loss": 0.0468,
      "step": 5898
    },
    {
      "epoch": 0.9305517214181488,
      "grad_norm": 5.177865505218506,
      "learning_rate": 7.084205441957816e-08,
      "loss": 0.0356,
      "step": 5899
    },
    {
      "epoch": 0.9307094687857397,
      "grad_norm": 5.825542449951172,
      "learning_rate": 7.068104975044275e-08,
      "loss": 0.0502,
      "step": 5900
    },
    {
      "epoch": 0.9308672161533305,
      "grad_norm": 5.751056671142578,
      "learning_rate": 7.052004508130734e-08,
      "loss": 0.0249,
      "step": 5901
    },
    {
      "epoch": 0.9310249635209212,
      "grad_norm": 4.16907262802124,
      "learning_rate": 7.035904041217194e-08,
      "loss": 0.0824,
      "step": 5902
    },
    {
      "epoch": 0.931182710888512,
      "grad_norm": 4.030389785766602,
      "learning_rate": 7.019803574303656e-08,
      "loss": 0.0541,
      "step": 5903
    },
    {
      "epoch": 0.9313404582561029,
      "grad_norm": 5.3484625816345215,
      "learning_rate": 7.003703107390115e-08,
      "loss": 0.0447,
      "step": 5904
    },
    {
      "epoch": 0.9314982056236937,
      "grad_norm": 5.1839094161987305,
      "learning_rate": 6.987602640476574e-08,
      "loss": 0.0312,
      "step": 5905
    },
    {
      "epoch": 0.9316559529912845,
      "grad_norm": 7.408259868621826,
      "learning_rate": 6.971502173563034e-08,
      "loss": 0.0373,
      "step": 5906
    },
    {
      "epoch": 0.9318137003588752,
      "grad_norm": 3.3852338790893555,
      "learning_rate": 6.955401706649493e-08,
      "loss": 0.0355,
      "step": 5907
    },
    {
      "epoch": 0.9319714477264661,
      "grad_norm": 5.942039489746094,
      "learning_rate": 6.939301239735952e-08,
      "loss": 0.0355,
      "step": 5908
    },
    {
      "epoch": 0.9321291950940569,
      "grad_norm": 5.299680709838867,
      "learning_rate": 6.923200772822412e-08,
      "loss": 0.0726,
      "step": 5909
    },
    {
      "epoch": 0.9322869424616477,
      "grad_norm": 1.9223432540893555,
      "learning_rate": 6.907100305908871e-08,
      "loss": 0.0134,
      "step": 5910
    },
    {
      "epoch": 0.9324446898292384,
      "grad_norm": 4.402111530303955,
      "learning_rate": 6.89099983899533e-08,
      "loss": 0.0334,
      "step": 5911
    },
    {
      "epoch": 0.9326024371968292,
      "grad_norm": 4.189789295196533,
      "learning_rate": 6.87489937208179e-08,
      "loss": 0.0312,
      "step": 5912
    },
    {
      "epoch": 0.9327601845644201,
      "grad_norm": 4.966956615447998,
      "learning_rate": 6.858798905168249e-08,
      "loss": 0.0456,
      "step": 5913
    },
    {
      "epoch": 0.9329179319320109,
      "grad_norm": 4.631741523742676,
      "learning_rate": 6.842698438254708e-08,
      "loss": 0.061,
      "step": 5914
    },
    {
      "epoch": 0.9330756792996017,
      "grad_norm": 6.761546611785889,
      "learning_rate": 6.826597971341169e-08,
      "loss": 0.0348,
      "step": 5915
    },
    {
      "epoch": 0.9332334266671924,
      "grad_norm": 3.471755027770996,
      "learning_rate": 6.810497504427628e-08,
      "loss": 0.0353,
      "step": 5916
    },
    {
      "epoch": 0.9333911740347833,
      "grad_norm": 5.140878200531006,
      "learning_rate": 6.794397037514087e-08,
      "loss": 0.0413,
      "step": 5917
    },
    {
      "epoch": 0.9335489214023741,
      "grad_norm": 5.290947914123535,
      "learning_rate": 6.778296570600548e-08,
      "loss": 0.0488,
      "step": 5918
    },
    {
      "epoch": 0.9337066687699649,
      "grad_norm": 7.529231548309326,
      "learning_rate": 6.762196103687007e-08,
      "loss": 0.0771,
      "step": 5919
    },
    {
      "epoch": 0.9338644161375557,
      "grad_norm": 4.304989337921143,
      "learning_rate": 6.746095636773467e-08,
      "loss": 0.0184,
      "step": 5920
    },
    {
      "epoch": 0.9340221635051466,
      "grad_norm": 4.198561191558838,
      "learning_rate": 6.729995169859926e-08,
      "loss": 0.028,
      "step": 5921
    },
    {
      "epoch": 0.9341799108727373,
      "grad_norm": 3.3721861839294434,
      "learning_rate": 6.713894702946385e-08,
      "loss": 0.0357,
      "step": 5922
    },
    {
      "epoch": 0.9343376582403281,
      "grad_norm": 4.256961345672607,
      "learning_rate": 6.697794236032845e-08,
      "loss": 0.0516,
      "step": 5923
    },
    {
      "epoch": 0.9344954056079189,
      "grad_norm": 5.280449390411377,
      "learning_rate": 6.681693769119304e-08,
      "loss": 0.0345,
      "step": 5924
    },
    {
      "epoch": 0.9346531529755098,
      "grad_norm": 2.9795844554901123,
      "learning_rate": 6.665593302205763e-08,
      "loss": 0.0453,
      "step": 5925
    },
    {
      "epoch": 0.9348109003431005,
      "grad_norm": 9.320393562316895,
      "learning_rate": 6.649492835292222e-08,
      "loss": 0.0659,
      "step": 5926
    },
    {
      "epoch": 0.9349686477106913,
      "grad_norm": 6.018735885620117,
      "learning_rate": 6.633392368378683e-08,
      "loss": 0.0435,
      "step": 5927
    },
    {
      "epoch": 0.9351263950782821,
      "grad_norm": 3.991745710372925,
      "learning_rate": 6.617291901465142e-08,
      "loss": 0.0246,
      "step": 5928
    },
    {
      "epoch": 0.9352841424458729,
      "grad_norm": 3.104764461517334,
      "learning_rate": 6.601191434551602e-08,
      "loss": 0.0304,
      "step": 5929
    },
    {
      "epoch": 0.9354418898134638,
      "grad_norm": 3.9032888412475586,
      "learning_rate": 6.585090967638061e-08,
      "loss": 0.0579,
      "step": 5930
    },
    {
      "epoch": 0.9355996371810545,
      "grad_norm": 5.164892196655273,
      "learning_rate": 6.56899050072452e-08,
      "loss": 0.0503,
      "step": 5931
    },
    {
      "epoch": 0.9357573845486453,
      "grad_norm": 8.195326805114746,
      "learning_rate": 6.55289003381098e-08,
      "loss": 0.0421,
      "step": 5932
    },
    {
      "epoch": 0.9359151319162361,
      "grad_norm": 7.684947967529297,
      "learning_rate": 6.53678956689744e-08,
      "loss": 0.0261,
      "step": 5933
    },
    {
      "epoch": 0.936072879283827,
      "grad_norm": 4.523174285888672,
      "learning_rate": 6.5206890999839e-08,
      "loss": 0.0563,
      "step": 5934
    },
    {
      "epoch": 0.9362306266514178,
      "grad_norm": 11.877717018127441,
      "learning_rate": 6.504588633070359e-08,
      "loss": 0.0922,
      "step": 5935
    },
    {
      "epoch": 0.9363883740190085,
      "grad_norm": 3.477673053741455,
      "learning_rate": 6.488488166156818e-08,
      "loss": 0.0202,
      "step": 5936
    },
    {
      "epoch": 0.9365461213865993,
      "grad_norm": 3.0422277450561523,
      "learning_rate": 6.472387699243277e-08,
      "loss": 0.0588,
      "step": 5937
    },
    {
      "epoch": 0.9367038687541902,
      "grad_norm": 4.104816913604736,
      "learning_rate": 6.456287232329738e-08,
      "loss": 0.0496,
      "step": 5938
    },
    {
      "epoch": 0.936861616121781,
      "grad_norm": 4.961500644683838,
      "learning_rate": 6.440186765416197e-08,
      "loss": 0.0594,
      "step": 5939
    },
    {
      "epoch": 0.9370193634893718,
      "grad_norm": 2.4108762741088867,
      "learning_rate": 6.424086298502657e-08,
      "loss": 0.0311,
      "step": 5940
    },
    {
      "epoch": 0.9371771108569625,
      "grad_norm": 5.95065975189209,
      "learning_rate": 6.407985831589116e-08,
      "loss": 0.0914,
      "step": 5941
    },
    {
      "epoch": 0.9373348582245534,
      "grad_norm": 8.552593231201172,
      "learning_rate": 6.391885364675575e-08,
      "loss": 0.0549,
      "step": 5942
    },
    {
      "epoch": 0.9374926055921442,
      "grad_norm": 5.621298313140869,
      "learning_rate": 6.375784897762035e-08,
      "loss": 0.0223,
      "step": 5943
    },
    {
      "epoch": 0.937650352959735,
      "grad_norm": 3.9253087043762207,
      "learning_rate": 6.359684430848494e-08,
      "loss": 0.0693,
      "step": 5944
    },
    {
      "epoch": 0.9378081003273258,
      "grad_norm": 6.854350566864014,
      "learning_rate": 6.343583963934953e-08,
      "loss": 0.0254,
      "step": 5945
    },
    {
      "epoch": 0.9379658476949166,
      "grad_norm": 4.1480278968811035,
      "learning_rate": 6.327483497021413e-08,
      "loss": 0.0681,
      "step": 5946
    },
    {
      "epoch": 0.9381235950625074,
      "grad_norm": 13.152607917785645,
      "learning_rate": 6.311383030107872e-08,
      "loss": 0.043,
      "step": 5947
    },
    {
      "epoch": 0.9382813424300982,
      "grad_norm": 13.167405128479004,
      "learning_rate": 6.295282563194331e-08,
      "loss": 0.0761,
      "step": 5948
    },
    {
      "epoch": 0.938439089797689,
      "grad_norm": 6.634943962097168,
      "learning_rate": 6.279182096280792e-08,
      "loss": 0.0764,
      "step": 5949
    },
    {
      "epoch": 0.9385968371652798,
      "grad_norm": 4.601792335510254,
      "learning_rate": 6.263081629367252e-08,
      "loss": 0.0447,
      "step": 5950
    },
    {
      "epoch": 0.9387545845328706,
      "grad_norm": 3.340606927871704,
      "learning_rate": 6.24698116245371e-08,
      "loss": 0.0354,
      "step": 5951
    },
    {
      "epoch": 0.9389123319004614,
      "grad_norm": 4.565184116363525,
      "learning_rate": 6.23088069554017e-08,
      "loss": 0.0227,
      "step": 5952
    },
    {
      "epoch": 0.9390700792680522,
      "grad_norm": 4.615509510040283,
      "learning_rate": 6.21478022862663e-08,
      "loss": 0.0612,
      "step": 5953
    },
    {
      "epoch": 0.939227826635643,
      "grad_norm": 2.9482510089874268,
      "learning_rate": 6.19867976171309e-08,
      "loss": 0.0296,
      "step": 5954
    },
    {
      "epoch": 0.9393855740032339,
      "grad_norm": 5.325509548187256,
      "learning_rate": 6.182579294799549e-08,
      "loss": 0.0212,
      "step": 5955
    },
    {
      "epoch": 0.9395433213708246,
      "grad_norm": 3.9348642826080322,
      "learning_rate": 6.166478827886008e-08,
      "loss": 0.024,
      "step": 5956
    },
    {
      "epoch": 0.9397010687384154,
      "grad_norm": 6.6087822914123535,
      "learning_rate": 6.150378360972468e-08,
      "loss": 0.0791,
      "step": 5957
    },
    {
      "epoch": 0.9398588161060062,
      "grad_norm": 6.441550254821777,
      "learning_rate": 6.134277894058927e-08,
      "loss": 0.0762,
      "step": 5958
    },
    {
      "epoch": 0.9400165634735971,
      "grad_norm": 4.344478607177734,
      "learning_rate": 6.118177427145388e-08,
      "loss": 0.0811,
      "step": 5959
    },
    {
      "epoch": 0.9401743108411879,
      "grad_norm": 3.4004554748535156,
      "learning_rate": 6.102076960231847e-08,
      "loss": 0.0282,
      "step": 5960
    },
    {
      "epoch": 0.9403320582087786,
      "grad_norm": 7.548405170440674,
      "learning_rate": 6.085976493318306e-08,
      "loss": 0.0444,
      "step": 5961
    },
    {
      "epoch": 0.9404898055763694,
      "grad_norm": 4.60588264465332,
      "learning_rate": 6.069876026404765e-08,
      "loss": 0.0214,
      "step": 5962
    },
    {
      "epoch": 0.9406475529439603,
      "grad_norm": 6.029146194458008,
      "learning_rate": 6.053775559491225e-08,
      "loss": 0.1035,
      "step": 5963
    },
    {
      "epoch": 0.9408053003115511,
      "grad_norm": 11.042107582092285,
      "learning_rate": 6.037675092577685e-08,
      "loss": 0.0226,
      "step": 5964
    },
    {
      "epoch": 0.9409630476791419,
      "grad_norm": 3.1142709255218506,
      "learning_rate": 6.021574625664145e-08,
      "loss": 0.0591,
      "step": 5965
    },
    {
      "epoch": 0.9411207950467326,
      "grad_norm": 2.8489291667938232,
      "learning_rate": 6.005474158750604e-08,
      "loss": 0.0252,
      "step": 5966
    },
    {
      "epoch": 0.9412785424143235,
      "grad_norm": 3.7460086345672607,
      "learning_rate": 5.989373691837063e-08,
      "loss": 0.0498,
      "step": 5967
    },
    {
      "epoch": 0.9414362897819143,
      "grad_norm": 3.1322598457336426,
      "learning_rate": 5.973273224923523e-08,
      "loss": 0.0292,
      "step": 5968
    },
    {
      "epoch": 0.9415940371495051,
      "grad_norm": 3.797261953353882,
      "learning_rate": 5.957172758009982e-08,
      "loss": 0.0148,
      "step": 5969
    },
    {
      "epoch": 0.9417517845170958,
      "grad_norm": 5.342018127441406,
      "learning_rate": 5.941072291096442e-08,
      "loss": 0.022,
      "step": 5970
    },
    {
      "epoch": 0.9419095318846866,
      "grad_norm": 6.661462306976318,
      "learning_rate": 5.924971824182901e-08,
      "loss": 0.0404,
      "step": 5971
    },
    {
      "epoch": 0.9420672792522775,
      "grad_norm": 2.1923716068267822,
      "learning_rate": 5.9088713572693605e-08,
      "loss": 0.0223,
      "step": 5972
    },
    {
      "epoch": 0.9422250266198683,
      "grad_norm": 5.065052509307861,
      "learning_rate": 5.89277089035582e-08,
      "loss": 0.0451,
      "step": 5973
    },
    {
      "epoch": 0.9423827739874591,
      "grad_norm": 6.447623252868652,
      "learning_rate": 5.876670423442279e-08,
      "loss": 0.064,
      "step": 5974
    },
    {
      "epoch": 0.9425405213550498,
      "grad_norm": 2.33776593208313,
      "learning_rate": 5.860569956528739e-08,
      "loss": 0.0204,
      "step": 5975
    },
    {
      "epoch": 0.9426982687226407,
      "grad_norm": 5.81252384185791,
      "learning_rate": 5.844469489615199e-08,
      "loss": 0.1056,
      "step": 5976
    },
    {
      "epoch": 0.9428560160902315,
      "grad_norm": 3.33357310295105,
      "learning_rate": 5.8283690227016584e-08,
      "loss": 0.0193,
      "step": 5977
    },
    {
      "epoch": 0.9430137634578223,
      "grad_norm": 4.074812412261963,
      "learning_rate": 5.812268555788118e-08,
      "loss": 0.0771,
      "step": 5978
    },
    {
      "epoch": 0.9431715108254131,
      "grad_norm": 2.8009839057922363,
      "learning_rate": 5.796168088874577e-08,
      "loss": 0.0192,
      "step": 5979
    },
    {
      "epoch": 0.943329258193004,
      "grad_norm": 4.452914237976074,
      "learning_rate": 5.780067621961036e-08,
      "loss": 0.0808,
      "step": 5980
    },
    {
      "epoch": 0.9434870055605947,
      "grad_norm": 3.7786972522735596,
      "learning_rate": 5.7639671550474956e-08,
      "loss": 0.0401,
      "step": 5981
    },
    {
      "epoch": 0.9436447529281855,
      "grad_norm": 3.053673505783081,
      "learning_rate": 5.7478666881339556e-08,
      "loss": 0.017,
      "step": 5982
    },
    {
      "epoch": 0.9438025002957763,
      "grad_norm": 6.5977630615234375,
      "learning_rate": 5.7317662212204156e-08,
      "loss": 0.0398,
      "step": 5983
    },
    {
      "epoch": 0.9439602476633672,
      "grad_norm": 3.19356369972229,
      "learning_rate": 5.715665754306875e-08,
      "loss": 0.0184,
      "step": 5984
    },
    {
      "epoch": 0.944117995030958,
      "grad_norm": 8.475666046142578,
      "learning_rate": 5.699565287393334e-08,
      "loss": 0.0418,
      "step": 5985
    },
    {
      "epoch": 0.9442757423985487,
      "grad_norm": 5.271622657775879,
      "learning_rate": 5.6834648204797935e-08,
      "loss": 0.0307,
      "step": 5986
    },
    {
      "epoch": 0.9444334897661395,
      "grad_norm": 7.270249843597412,
      "learning_rate": 5.667364353566253e-08,
      "loss": 0.0475,
      "step": 5987
    },
    {
      "epoch": 0.9445912371337303,
      "grad_norm": 4.548971652984619,
      "learning_rate": 5.651263886652713e-08,
      "loss": 0.0427,
      "step": 5988
    },
    {
      "epoch": 0.9447489845013212,
      "grad_norm": 4.646120071411133,
      "learning_rate": 5.635163419739172e-08,
      "loss": 0.0509,
      "step": 5989
    },
    {
      "epoch": 0.9449067318689119,
      "grad_norm": 5.53933572769165,
      "learning_rate": 5.6190629528256314e-08,
      "loss": 0.0686,
      "step": 5990
    },
    {
      "epoch": 0.9450644792365027,
      "grad_norm": 6.857639312744141,
      "learning_rate": 5.6029624859120913e-08,
      "loss": 0.0714,
      "step": 5991
    },
    {
      "epoch": 0.9452222266040935,
      "grad_norm": 3.0770936012268066,
      "learning_rate": 5.5868620189985507e-08,
      "loss": 0.0294,
      "step": 5992
    },
    {
      "epoch": 0.9453799739716844,
      "grad_norm": 5.394449710845947,
      "learning_rate": 5.5707615520850106e-08,
      "loss": 0.0787,
      "step": 5993
    },
    {
      "epoch": 0.9455377213392752,
      "grad_norm": 6.045137882232666,
      "learning_rate": 5.55466108517147e-08,
      "loss": 0.0477,
      "step": 5994
    },
    {
      "epoch": 0.9456954687068659,
      "grad_norm": 7.0374603271484375,
      "learning_rate": 5.538560618257929e-08,
      "loss": 0.0668,
      "step": 5995
    },
    {
      "epoch": 0.9458532160744567,
      "grad_norm": 3.169250249862671,
      "learning_rate": 5.5224601513443885e-08,
      "loss": 0.0134,
      "step": 5996
    },
    {
      "epoch": 0.9460109634420476,
      "grad_norm": 3.9192862510681152,
      "learning_rate": 5.506359684430848e-08,
      "loss": 0.0398,
      "step": 5997
    },
    {
      "epoch": 0.9461687108096384,
      "grad_norm": 10.280298233032227,
      "learning_rate": 5.490259217517308e-08,
      "loss": 0.0552,
      "step": 5998
    },
    {
      "epoch": 0.9463264581772292,
      "grad_norm": 2.0093114376068115,
      "learning_rate": 5.474158750603768e-08,
      "loss": 0.011,
      "step": 5999
    },
    {
      "epoch": 0.9464842055448199,
      "grad_norm": 3.823469877243042,
      "learning_rate": 5.458058283690227e-08,
      "loss": 0.0554,
      "step": 6000
    },
    {
      "epoch": 0.9466419529124108,
      "grad_norm": 6.715463638305664,
      "learning_rate": 5.4419578167766864e-08,
      "loss": 0.0386,
      "step": 6001
    },
    {
      "epoch": 0.9467997002800016,
      "grad_norm": 2.5882365703582764,
      "learning_rate": 5.425857349863146e-08,
      "loss": 0.0243,
      "step": 6002
    },
    {
      "epoch": 0.9469574476475924,
      "grad_norm": 3.5130350589752197,
      "learning_rate": 5.409756882949605e-08,
      "loss": 0.027,
      "step": 6003
    },
    {
      "epoch": 0.9471151950151832,
      "grad_norm": 3.3332033157348633,
      "learning_rate": 5.3936564160360643e-08,
      "loss": 0.034,
      "step": 6004
    },
    {
      "epoch": 0.947272942382774,
      "grad_norm": 2.0243804454803467,
      "learning_rate": 5.377555949122524e-08,
      "loss": 0.0151,
      "step": 6005
    },
    {
      "epoch": 0.9474306897503648,
      "grad_norm": 4.765673637390137,
      "learning_rate": 5.361455482208984e-08,
      "loss": 0.0329,
      "step": 6006
    },
    {
      "epoch": 0.9475884371179556,
      "grad_norm": 9.25135612487793,
      "learning_rate": 5.3453550152954436e-08,
      "loss": 0.0903,
      "step": 6007
    },
    {
      "epoch": 0.9477461844855464,
      "grad_norm": 6.916815757751465,
      "learning_rate": 5.329254548381903e-08,
      "loss": 0.0375,
      "step": 6008
    },
    {
      "epoch": 0.9479039318531371,
      "grad_norm": 4.9020094871521,
      "learning_rate": 5.313154081468362e-08,
      "loss": 0.0186,
      "step": 6009
    },
    {
      "epoch": 0.948061679220728,
      "grad_norm": 5.087789058685303,
      "learning_rate": 5.2970536145548215e-08,
      "loss": 0.0336,
      "step": 6010
    },
    {
      "epoch": 0.9482194265883188,
      "grad_norm": 3.896568536758423,
      "learning_rate": 5.2809531476412815e-08,
      "loss": 0.0462,
      "step": 6011
    },
    {
      "epoch": 0.9483771739559096,
      "grad_norm": 2.7155940532684326,
      "learning_rate": 5.264852680727741e-08,
      "loss": 0.0225,
      "step": 6012
    },
    {
      "epoch": 0.9485349213235004,
      "grad_norm": 5.200143337249756,
      "learning_rate": 5.2487522138142e-08,
      "loss": 0.024,
      "step": 6013
    },
    {
      "epoch": 0.9486926686910913,
      "grad_norm": 6.716253757476807,
      "learning_rate": 5.23265174690066e-08,
      "loss": 0.0562,
      "step": 6014
    },
    {
      "epoch": 0.948850416058682,
      "grad_norm": 3.017744302749634,
      "learning_rate": 5.2165512799871194e-08,
      "loss": 0.0344,
      "step": 6015
    },
    {
      "epoch": 0.9490081634262728,
      "grad_norm": 5.271416664123535,
      "learning_rate": 5.200450813073579e-08,
      "loss": 0.0305,
      "step": 6016
    },
    {
      "epoch": 0.9491659107938636,
      "grad_norm": 4.114175319671631,
      "learning_rate": 5.1843503461600386e-08,
      "loss": 0.0245,
      "step": 6017
    },
    {
      "epoch": 0.9493236581614545,
      "grad_norm": 2.249285936355591,
      "learning_rate": 5.168249879246498e-08,
      "loss": 0.0172,
      "step": 6018
    },
    {
      "epoch": 0.9494814055290453,
      "grad_norm": 6.96615743637085,
      "learning_rate": 5.152149412332957e-08,
      "loss": 0.0578,
      "step": 6019
    },
    {
      "epoch": 0.949639152896636,
      "grad_norm": 4.631272792816162,
      "learning_rate": 5.1360489454194166e-08,
      "loss": 0.0302,
      "step": 6020
    },
    {
      "epoch": 0.9497969002642268,
      "grad_norm": 4.814763069152832,
      "learning_rate": 5.1199484785058765e-08,
      "loss": 0.0646,
      "step": 6021
    },
    {
      "epoch": 0.9499546476318177,
      "grad_norm": 3.347012996673584,
      "learning_rate": 5.103848011592336e-08,
      "loss": 0.022,
      "step": 6022
    },
    {
      "epoch": 0.9501123949994085,
      "grad_norm": 3.0205132961273193,
      "learning_rate": 5.087747544678796e-08,
      "loss": 0.0391,
      "step": 6023
    },
    {
      "epoch": 0.9502701423669992,
      "grad_norm": 5.37874698638916,
      "learning_rate": 5.071647077765255e-08,
      "loss": 0.0513,
      "step": 6024
    },
    {
      "epoch": 0.95042788973459,
      "grad_norm": 4.942624568939209,
      "learning_rate": 5.0555466108517144e-08,
      "loss": 0.0548,
      "step": 6025
    },
    {
      "epoch": 0.9505856371021808,
      "grad_norm": 4.22371768951416,
      "learning_rate": 5.039446143938174e-08,
      "loss": 0.0291,
      "step": 6026
    },
    {
      "epoch": 0.9507433844697717,
      "grad_norm": 4.233780860900879,
      "learning_rate": 5.023345677024633e-08,
      "loss": 0.0218,
      "step": 6027
    },
    {
      "epoch": 0.9509011318373625,
      "grad_norm": 7.730014801025391,
      "learning_rate": 5.007245210111093e-08,
      "loss": 0.0375,
      "step": 6028
    },
    {
      "epoch": 0.9510588792049532,
      "grad_norm": 3.876887559890747,
      "learning_rate": 4.991144743197553e-08,
      "loss": 0.0279,
      "step": 6029
    },
    {
      "epoch": 0.951216626572544,
      "grad_norm": 4.554478168487549,
      "learning_rate": 4.975044276284012e-08,
      "loss": 0.0443,
      "step": 6030
    },
    {
      "epoch": 0.9513743739401349,
      "grad_norm": 5.302878379821777,
      "learning_rate": 4.9589438093704716e-08,
      "loss": 0.0309,
      "step": 6031
    },
    {
      "epoch": 0.9515321213077257,
      "grad_norm": 26.972261428833008,
      "learning_rate": 4.942843342456931e-08,
      "loss": 0.0218,
      "step": 6032
    },
    {
      "epoch": 0.9516898686753165,
      "grad_norm": 2.710054636001587,
      "learning_rate": 4.92674287554339e-08,
      "loss": 0.0134,
      "step": 6033
    },
    {
      "epoch": 0.9518476160429072,
      "grad_norm": 2.822371482849121,
      "learning_rate": 4.91064240862985e-08,
      "loss": 0.0187,
      "step": 6034
    },
    {
      "epoch": 0.9520053634104981,
      "grad_norm": 3.546046018600464,
      "learning_rate": 4.8945419417163095e-08,
      "loss": 0.0231,
      "step": 6035
    },
    {
      "epoch": 0.9521631107780889,
      "grad_norm": 2.5862648487091064,
      "learning_rate": 4.878441474802769e-08,
      "loss": 0.021,
      "step": 6036
    },
    {
      "epoch": 0.9523208581456797,
      "grad_norm": 4.88242244720459,
      "learning_rate": 4.862341007889229e-08,
      "loss": 0.0493,
      "step": 6037
    },
    {
      "epoch": 0.9524786055132705,
      "grad_norm": 4.476553440093994,
      "learning_rate": 4.846240540975688e-08,
      "loss": 0.035,
      "step": 6038
    },
    {
      "epoch": 0.9526363528808613,
      "grad_norm": 3.3754684925079346,
      "learning_rate": 4.8301400740621474e-08,
      "loss": 0.0362,
      "step": 6039
    },
    {
      "epoch": 0.9527941002484521,
      "grad_norm": 2.319370746612549,
      "learning_rate": 4.8140396071486074e-08,
      "loss": 0.0339,
      "step": 6040
    },
    {
      "epoch": 0.9529518476160429,
      "grad_norm": 4.899821758270264,
      "learning_rate": 4.797939140235067e-08,
      "loss": 0.06,
      "step": 6041
    },
    {
      "epoch": 0.9531095949836337,
      "grad_norm": 3.4484035968780518,
      "learning_rate": 4.781838673321526e-08,
      "loss": 0.031,
      "step": 6042
    },
    {
      "epoch": 0.9532673423512246,
      "grad_norm": 11.35166072845459,
      "learning_rate": 4.765738206407985e-08,
      "loss": 0.0473,
      "step": 6043
    },
    {
      "epoch": 0.9534250897188153,
      "grad_norm": 8.086166381835938,
      "learning_rate": 4.749637739494445e-08,
      "loss": 0.0767,
      "step": 6044
    },
    {
      "epoch": 0.9535828370864061,
      "grad_norm": 3.089240550994873,
      "learning_rate": 4.7335372725809046e-08,
      "loss": 0.0167,
      "step": 6045
    },
    {
      "epoch": 0.9537405844539969,
      "grad_norm": 10.3706693649292,
      "learning_rate": 4.7174368056673645e-08,
      "loss": 0.0625,
      "step": 6046
    },
    {
      "epoch": 0.9538983318215877,
      "grad_norm": 5.132999420166016,
      "learning_rate": 4.701336338753824e-08,
      "loss": 0.0456,
      "step": 6047
    },
    {
      "epoch": 0.9540560791891786,
      "grad_norm": 4.445789813995361,
      "learning_rate": 4.685235871840283e-08,
      "loss": 0.0553,
      "step": 6048
    },
    {
      "epoch": 0.9542138265567693,
      "grad_norm": 5.825806617736816,
      "learning_rate": 4.6691354049267425e-08,
      "loss": 0.0214,
      "step": 6049
    },
    {
      "epoch": 0.9543715739243601,
      "grad_norm": 5.145627975463867,
      "learning_rate": 4.653034938013202e-08,
      "loss": 0.0293,
      "step": 6050
    },
    {
      "epoch": 0.9545293212919509,
      "grad_norm": 4.06556510925293,
      "learning_rate": 4.636934471099661e-08,
      "loss": 0.0218,
      "step": 6051
    },
    {
      "epoch": 0.9546870686595418,
      "grad_norm": 3.483142614364624,
      "learning_rate": 4.620834004186122e-08,
      "loss": 0.0395,
      "step": 6052
    },
    {
      "epoch": 0.9548448160271326,
      "grad_norm": 5.402103900909424,
      "learning_rate": 4.604733537272581e-08,
      "loss": 0.0573,
      "step": 6053
    },
    {
      "epoch": 0.9550025633947233,
      "grad_norm": 8.97451114654541,
      "learning_rate": 4.58863307035904e-08,
      "loss": 0.0466,
      "step": 6054
    },
    {
      "epoch": 0.9551603107623141,
      "grad_norm": 3.5415401458740234,
      "learning_rate": 4.5725326034454996e-08,
      "loss": 0.0236,
      "step": 6055
    },
    {
      "epoch": 0.955318058129905,
      "grad_norm": 5.875942230224609,
      "learning_rate": 4.556432136531959e-08,
      "loss": 0.0526,
      "step": 6056
    },
    {
      "epoch": 0.9554758054974958,
      "grad_norm": 10.333345413208008,
      "learning_rate": 4.540331669618419e-08,
      "loss": 0.0621,
      "step": 6057
    },
    {
      "epoch": 0.9556335528650866,
      "grad_norm": 4.999233722686768,
      "learning_rate": 4.524231202704878e-08,
      "loss": 0.0257,
      "step": 6058
    },
    {
      "epoch": 0.9557913002326773,
      "grad_norm": 16.694101333618164,
      "learning_rate": 4.5081307357913375e-08,
      "loss": 0.0475,
      "step": 6059
    },
    {
      "epoch": 0.9559490476002682,
      "grad_norm": 4.000310897827148,
      "learning_rate": 4.4920302688777975e-08,
      "loss": 0.0371,
      "step": 6060
    },
    {
      "epoch": 0.956106794967859,
      "grad_norm": 4.216620922088623,
      "learning_rate": 4.475929801964257e-08,
      "loss": 0.0647,
      "step": 6061
    },
    {
      "epoch": 0.9562645423354498,
      "grad_norm": 3.991231918334961,
      "learning_rate": 4.459829335050716e-08,
      "loss": 0.0342,
      "step": 6062
    },
    {
      "epoch": 0.9564222897030406,
      "grad_norm": 6.847813129425049,
      "learning_rate": 4.443728868137176e-08,
      "loss": 0.0357,
      "step": 6063
    },
    {
      "epoch": 0.9565800370706314,
      "grad_norm": 3.854752779006958,
      "learning_rate": 4.4276284012236354e-08,
      "loss": 0.0705,
      "step": 6064
    },
    {
      "epoch": 0.9567377844382222,
      "grad_norm": 4.1908745765686035,
      "learning_rate": 4.411527934310095e-08,
      "loss": 0.0351,
      "step": 6065
    },
    {
      "epoch": 0.956895531805813,
      "grad_norm": 6.255070686340332,
      "learning_rate": 4.395427467396554e-08,
      "loss": 0.0405,
      "step": 6066
    },
    {
      "epoch": 0.9570532791734038,
      "grad_norm": 4.168485164642334,
      "learning_rate": 4.379327000483014e-08,
      "loss": 0.0946,
      "step": 6067
    },
    {
      "epoch": 0.9572110265409945,
      "grad_norm": 6.225264549255371,
      "learning_rate": 4.363226533569473e-08,
      "loss": 0.059,
      "step": 6068
    },
    {
      "epoch": 0.9573687739085854,
      "grad_norm": 4.073039531707764,
      "learning_rate": 4.347126066655933e-08,
      "loss": 0.0436,
      "step": 6069
    },
    {
      "epoch": 0.9575265212761762,
      "grad_norm": 3.783607244491577,
      "learning_rate": 4.3310255997423926e-08,
      "loss": 0.0524,
      "step": 6070
    },
    {
      "epoch": 0.957684268643767,
      "grad_norm": 2.8810980319976807,
      "learning_rate": 4.314925132828852e-08,
      "loss": 0.0207,
      "step": 6071
    },
    {
      "epoch": 0.9578420160113578,
      "grad_norm": 3.771087408065796,
      "learning_rate": 4.298824665915311e-08,
      "loss": 0.0233,
      "step": 6072
    },
    {
      "epoch": 0.9579997633789487,
      "grad_norm": 3.440264940261841,
      "learning_rate": 4.2827241990017705e-08,
      "loss": 0.0235,
      "step": 6073
    },
    {
      "epoch": 0.9581575107465394,
      "grad_norm": 6.002028942108154,
      "learning_rate": 4.26662373208823e-08,
      "loss": 0.0176,
      "step": 6074
    },
    {
      "epoch": 0.9583152581141302,
      "grad_norm": 3.9315836429595947,
      "learning_rate": 4.2505232651746904e-08,
      "loss": 0.0219,
      "step": 6075
    },
    {
      "epoch": 0.958473005481721,
      "grad_norm": 5.8993144035339355,
      "learning_rate": 4.23442279826115e-08,
      "loss": 0.0366,
      "step": 6076
    },
    {
      "epoch": 0.9586307528493119,
      "grad_norm": 4.322342395782471,
      "learning_rate": 4.218322331347609e-08,
      "loss": 0.0327,
      "step": 6077
    },
    {
      "epoch": 0.9587885002169027,
      "grad_norm": 3.4553074836730957,
      "learning_rate": 4.2022218644340683e-08,
      "loss": 0.0229,
      "step": 6078
    },
    {
      "epoch": 0.9589462475844934,
      "grad_norm": 5.900453567504883,
      "learning_rate": 4.1861213975205277e-08,
      "loss": 0.0379,
      "step": 6079
    },
    {
      "epoch": 0.9591039949520842,
      "grad_norm": 4.2595906257629395,
      "learning_rate": 4.170020930606987e-08,
      "loss": 0.0671,
      "step": 6080
    },
    {
      "epoch": 0.9592617423196751,
      "grad_norm": 4.484315872192383,
      "learning_rate": 4.153920463693447e-08,
      "loss": 0.0657,
      "step": 6081
    },
    {
      "epoch": 0.9594194896872659,
      "grad_norm": 5.0750274658203125,
      "learning_rate": 4.137819996779906e-08,
      "loss": 0.0484,
      "step": 6082
    },
    {
      "epoch": 0.9595772370548566,
      "grad_norm": 3.9562439918518066,
      "learning_rate": 4.121719529866366e-08,
      "loss": 0.0203,
      "step": 6083
    },
    {
      "epoch": 0.9597349844224474,
      "grad_norm": 2.25351619720459,
      "learning_rate": 4.1056190629528255e-08,
      "loss": 0.0264,
      "step": 6084
    },
    {
      "epoch": 0.9598927317900382,
      "grad_norm": 4.723305702209473,
      "learning_rate": 4.089518596039285e-08,
      "loss": 0.0391,
      "step": 6085
    },
    {
      "epoch": 0.9600504791576291,
      "grad_norm": 4.14780855178833,
      "learning_rate": 4.073418129125744e-08,
      "loss": 0.0317,
      "step": 6086
    },
    {
      "epoch": 0.9602082265252199,
      "grad_norm": 6.291996955871582,
      "learning_rate": 4.057317662212204e-08,
      "loss": 0.0471,
      "step": 6087
    },
    {
      "epoch": 0.9603659738928106,
      "grad_norm": 6.449141979217529,
      "learning_rate": 4.0412171952986634e-08,
      "loss": 0.0746,
      "step": 6088
    },
    {
      "epoch": 0.9605237212604014,
      "grad_norm": 7.540985584259033,
      "learning_rate": 4.025116728385123e-08,
      "loss": 0.0899,
      "step": 6089
    },
    {
      "epoch": 0.9606814686279923,
      "grad_norm": 4.7788920402526855,
      "learning_rate": 4.009016261471583e-08,
      "loss": 0.0269,
      "step": 6090
    },
    {
      "epoch": 0.9608392159955831,
      "grad_norm": 3.3653805255889893,
      "learning_rate": 3.992915794558042e-08,
      "loss": 0.0301,
      "step": 6091
    },
    {
      "epoch": 0.9609969633631739,
      "grad_norm": 6.402235507965088,
      "learning_rate": 3.976815327644502e-08,
      "loss": 0.049,
      "step": 6092
    },
    {
      "epoch": 0.9611547107307646,
      "grad_norm": 5.099925518035889,
      "learning_rate": 3.960714860730961e-08,
      "loss": 0.0632,
      "step": 6093
    },
    {
      "epoch": 0.9613124580983555,
      "grad_norm": 3.989698648452759,
      "learning_rate": 3.9446143938174206e-08,
      "loss": 0.0402,
      "step": 6094
    },
    {
      "epoch": 0.9614702054659463,
      "grad_norm": 5.849874019622803,
      "learning_rate": 3.92851392690388e-08,
      "loss": 0.0483,
      "step": 6095
    },
    {
      "epoch": 0.9616279528335371,
      "grad_norm": 6.019822120666504,
      "learning_rate": 3.912413459990339e-08,
      "loss": 0.0846,
      "step": 6096
    },
    {
      "epoch": 0.9617857002011279,
      "grad_norm": 7.444940090179443,
      "learning_rate": 3.8963129930767985e-08,
      "loss": 0.0414,
      "step": 6097
    },
    {
      "epoch": 0.9619434475687187,
      "grad_norm": 4.172461986541748,
      "learning_rate": 3.880212526163259e-08,
      "loss": 0.0756,
      "step": 6098
    },
    {
      "epoch": 0.9621011949363095,
      "grad_norm": 4.624844074249268,
      "learning_rate": 3.8641120592497184e-08,
      "loss": 0.0413,
      "step": 6099
    },
    {
      "epoch": 0.9622589423039003,
      "grad_norm": 7.130695819854736,
      "learning_rate": 3.848011592336178e-08,
      "loss": 0.0825,
      "step": 6100
    },
    {
      "epoch": 0.9624166896714911,
      "grad_norm": 3.7383406162261963,
      "learning_rate": 3.831911125422637e-08,
      "loss": 0.0435,
      "step": 6101
    },
    {
      "epoch": 0.962574437039082,
      "grad_norm": 3.9024624824523926,
      "learning_rate": 3.8158106585090964e-08,
      "loss": 0.0307,
      "step": 6102
    },
    {
      "epoch": 0.9627321844066727,
      "grad_norm": 4.302260398864746,
      "learning_rate": 3.799710191595556e-08,
      "loss": 0.0398,
      "step": 6103
    },
    {
      "epoch": 0.9628899317742635,
      "grad_norm": 5.5469441413879395,
      "learning_rate": 3.7836097246820156e-08,
      "loss": 0.0922,
      "step": 6104
    },
    {
      "epoch": 0.9630476791418543,
      "grad_norm": 3.2649874687194824,
      "learning_rate": 3.767509257768475e-08,
      "loss": 0.0238,
      "step": 6105
    },
    {
      "epoch": 0.9632054265094451,
      "grad_norm": 5.246920585632324,
      "learning_rate": 3.751408790854935e-08,
      "loss": 0.0472,
      "step": 6106
    },
    {
      "epoch": 0.963363173877036,
      "grad_norm": 4.312435150146484,
      "learning_rate": 3.735308323941394e-08,
      "loss": 0.0311,
      "step": 6107
    },
    {
      "epoch": 0.9635209212446267,
      "grad_norm": 7.261901378631592,
      "learning_rate": 3.7192078570278535e-08,
      "loss": 0.0479,
      "step": 6108
    },
    {
      "epoch": 0.9636786686122175,
      "grad_norm": 6.1279988288879395,
      "learning_rate": 3.703107390114313e-08,
      "loss": 0.0531,
      "step": 6109
    },
    {
      "epoch": 0.9638364159798083,
      "grad_norm": 5.212375164031982,
      "learning_rate": 3.687006923200773e-08,
      "loss": 0.0665,
      "step": 6110
    },
    {
      "epoch": 0.9639941633473992,
      "grad_norm": 3.3017594814300537,
      "learning_rate": 3.670906456287232e-08,
      "loss": 0.0344,
      "step": 6111
    },
    {
      "epoch": 0.96415191071499,
      "grad_norm": 6.285422325134277,
      "learning_rate": 3.6548059893736914e-08,
      "loss": 0.0567,
      "step": 6112
    },
    {
      "epoch": 0.9643096580825807,
      "grad_norm": 6.064441680908203,
      "learning_rate": 3.6387055224601514e-08,
      "loss": 0.0473,
      "step": 6113
    },
    {
      "epoch": 0.9644674054501715,
      "grad_norm": 4.454376220703125,
      "learning_rate": 3.622605055546611e-08,
      "loss": 0.0173,
      "step": 6114
    },
    {
      "epoch": 0.9646251528177624,
      "grad_norm": 5.166300296783447,
      "learning_rate": 3.60650458863307e-08,
      "loss": 0.0396,
      "step": 6115
    },
    {
      "epoch": 0.9647829001853532,
      "grad_norm": 19.746580123901367,
      "learning_rate": 3.59040412171953e-08,
      "loss": 0.1012,
      "step": 6116
    },
    {
      "epoch": 0.964940647552944,
      "grad_norm": 3.5285303592681885,
      "learning_rate": 3.574303654805989e-08,
      "loss": 0.0267,
      "step": 6117
    },
    {
      "epoch": 0.9650983949205347,
      "grad_norm": 4.301242828369141,
      "learning_rate": 3.5582031878924486e-08,
      "loss": 0.0195,
      "step": 6118
    },
    {
      "epoch": 0.9652561422881256,
      "grad_norm": 6.223407745361328,
      "learning_rate": 3.542102720978908e-08,
      "loss": 0.0286,
      "step": 6119
    },
    {
      "epoch": 0.9654138896557164,
      "grad_norm": 6.488699436187744,
      "learning_rate": 3.526002254065367e-08,
      "loss": 0.0337,
      "step": 6120
    },
    {
      "epoch": 0.9655716370233072,
      "grad_norm": 1.9815067052841187,
      "learning_rate": 3.509901787151828e-08,
      "loss": 0.0139,
      "step": 6121
    },
    {
      "epoch": 0.965729384390898,
      "grad_norm": 6.164742946624756,
      "learning_rate": 3.493801320238287e-08,
      "loss": 0.0587,
      "step": 6122
    },
    {
      "epoch": 0.9658871317584887,
      "grad_norm": 1.9928150177001953,
      "learning_rate": 3.4777008533247465e-08,
      "loss": 0.0176,
      "step": 6123
    },
    {
      "epoch": 0.9660448791260796,
      "grad_norm": 9.271769523620605,
      "learning_rate": 3.461600386411206e-08,
      "loss": 0.0497,
      "step": 6124
    },
    {
      "epoch": 0.9662026264936704,
      "grad_norm": 5.085504055023193,
      "learning_rate": 3.445499919497665e-08,
      "loss": 0.0592,
      "step": 6125
    },
    {
      "epoch": 0.9663603738612612,
      "grad_norm": 3.6184051036834717,
      "learning_rate": 3.4293994525841244e-08,
      "loss": 0.0758,
      "step": 6126
    },
    {
      "epoch": 0.966518121228852,
      "grad_norm": 6.860569000244141,
      "learning_rate": 3.4132989856705844e-08,
      "loss": 0.0336,
      "step": 6127
    },
    {
      "epoch": 0.9666758685964428,
      "grad_norm": 3.6731934547424316,
      "learning_rate": 3.3971985187570437e-08,
      "loss": 0.0332,
      "step": 6128
    },
    {
      "epoch": 0.9668336159640336,
      "grad_norm": 7.188792705535889,
      "learning_rate": 3.3810980518435036e-08,
      "loss": 0.0437,
      "step": 6129
    },
    {
      "epoch": 0.9669913633316244,
      "grad_norm": 4.120716571807861,
      "learning_rate": 3.364997584929963e-08,
      "loss": 0.0531,
      "step": 6130
    },
    {
      "epoch": 0.9671491106992152,
      "grad_norm": 2.426354169845581,
      "learning_rate": 3.348897118016422e-08,
      "loss": 0.0112,
      "step": 6131
    },
    {
      "epoch": 0.967306858066806,
      "grad_norm": 6.214199066162109,
      "learning_rate": 3.3327966511028816e-08,
      "loss": 0.0596,
      "step": 6132
    },
    {
      "epoch": 0.9674646054343968,
      "grad_norm": 3.4293365478515625,
      "learning_rate": 3.3166961841893415e-08,
      "loss": 0.0193,
      "step": 6133
    },
    {
      "epoch": 0.9676223528019876,
      "grad_norm": 2.90401291847229,
      "learning_rate": 3.300595717275801e-08,
      "loss": 0.013,
      "step": 6134
    },
    {
      "epoch": 0.9677801001695784,
      "grad_norm": 9.00876522064209,
      "learning_rate": 3.28449525036226e-08,
      "loss": 0.0345,
      "step": 6135
    },
    {
      "epoch": 0.9679378475371693,
      "grad_norm": 2.3712689876556396,
      "learning_rate": 3.26839478344872e-08,
      "loss": 0.0312,
      "step": 6136
    },
    {
      "epoch": 0.96809559490476,
      "grad_norm": 7.52162504196167,
      "learning_rate": 3.2522943165351794e-08,
      "loss": 0.1005,
      "step": 6137
    },
    {
      "epoch": 0.9682533422723508,
      "grad_norm": 5.890244007110596,
      "learning_rate": 3.236193849621639e-08,
      "loss": 0.0537,
      "step": 6138
    },
    {
      "epoch": 0.9684110896399416,
      "grad_norm": 5.929839134216309,
      "learning_rate": 3.220093382708099e-08,
      "loss": 0.0804,
      "step": 6139
    },
    {
      "epoch": 0.9685688370075325,
      "grad_norm": 2.614980936050415,
      "learning_rate": 3.203992915794558e-08,
      "loss": 0.0131,
      "step": 6140
    },
    {
      "epoch": 0.9687265843751233,
      "grad_norm": 5.683056354522705,
      "learning_rate": 3.187892448881017e-08,
      "loss": 0.0567,
      "step": 6141
    },
    {
      "epoch": 0.968884331742714,
      "grad_norm": 2.9339380264282227,
      "learning_rate": 3.1717919819674766e-08,
      "loss": 0.0213,
      "step": 6142
    },
    {
      "epoch": 0.9690420791103048,
      "grad_norm": 5.3598222732543945,
      "learning_rate": 3.155691515053936e-08,
      "loss": 0.0229,
      "step": 6143
    },
    {
      "epoch": 0.9691998264778956,
      "grad_norm": 4.69875431060791,
      "learning_rate": 3.139591048140396e-08,
      "loss": 0.0512,
      "step": 6144
    },
    {
      "epoch": 0.9691998264778956,
      "eval_accuracy": 0.9882576224396736,
      "eval_f1": 0.9882576224396736,
      "eval_loss": 0.038487568497657776,
      "eval_runtime": 4708.4491,
      "eval_samples_per_second": 43.083,
      "eval_steps_per_second": 2.693,
      "step": 6144
    },
    {
      "epoch": 0.9693575738454865,
      "grad_norm": 7.932188987731934,
      "learning_rate": 3.123490581226855e-08,
      "loss": 0.0437,
      "step": 6145
    },
    {
      "epoch": 0.9695153212130773,
      "grad_norm": 4.773967742919922,
      "learning_rate": 3.107390114313315e-08,
      "loss": 0.0307,
      "step": 6146
    },
    {
      "epoch": 0.969673068580668,
      "grad_norm": 4.265565395355225,
      "learning_rate": 3.0912896473997745e-08,
      "loss": 0.0284,
      "step": 6147
    },
    {
      "epoch": 0.9698308159482588,
      "grad_norm": 7.66091775894165,
      "learning_rate": 3.075189180486234e-08,
      "loss": 0.0528,
      "step": 6148
    },
    {
      "epoch": 0.9699885633158497,
      "grad_norm": 2.410403251647949,
      "learning_rate": 3.059088713572694e-08,
      "loss": 0.0158,
      "step": 6149
    },
    {
      "epoch": 0.9701463106834405,
      "grad_norm": 5.615338325500488,
      "learning_rate": 3.042988246659153e-08,
      "loss": 0.0306,
      "step": 6150
    },
    {
      "epoch": 0.9703040580510313,
      "grad_norm": 3.4606375694274902,
      "learning_rate": 3.0268877797456124e-08,
      "loss": 0.0633,
      "step": 6151
    },
    {
      "epoch": 0.970461805418622,
      "grad_norm": 5.322042942047119,
      "learning_rate": 3.0107873128320724e-08,
      "loss": 0.0615,
      "step": 6152
    },
    {
      "epoch": 0.9706195527862129,
      "grad_norm": 15.94222640991211,
      "learning_rate": 2.9946868459185317e-08,
      "loss": 0.0809,
      "step": 6153
    },
    {
      "epoch": 0.9707773001538037,
      "grad_norm": 1.9718109369277954,
      "learning_rate": 2.978586379004991e-08,
      "loss": 0.0109,
      "step": 6154
    },
    {
      "epoch": 0.9709350475213945,
      "grad_norm": 6.080210208892822,
      "learning_rate": 2.9624859120914506e-08,
      "loss": 0.0287,
      "step": 6155
    },
    {
      "epoch": 0.9710927948889853,
      "grad_norm": 4.3839569091796875,
      "learning_rate": 2.94638544517791e-08,
      "loss": 0.0642,
      "step": 6156
    },
    {
      "epoch": 0.9712505422565761,
      "grad_norm": 6.569242000579834,
      "learning_rate": 2.9302849782643696e-08,
      "loss": 0.0574,
      "step": 6157
    },
    {
      "epoch": 0.9714082896241669,
      "grad_norm": 4.099246978759766,
      "learning_rate": 2.9141845113508292e-08,
      "loss": 0.0293,
      "step": 6158
    },
    {
      "epoch": 0.9715660369917577,
      "grad_norm": 4.485950946807861,
      "learning_rate": 2.8980840444372885e-08,
      "loss": 0.0284,
      "step": 6159
    },
    {
      "epoch": 0.9717237843593485,
      "grad_norm": 2.693380355834961,
      "learning_rate": 2.8819835775237478e-08,
      "loss": 0.0228,
      "step": 6160
    },
    {
      "epoch": 0.9718815317269393,
      "grad_norm": 7.15558385848999,
      "learning_rate": 2.8658831106102078e-08,
      "loss": 0.0478,
      "step": 6161
    },
    {
      "epoch": 0.9720392790945301,
      "grad_norm": 2.9982993602752686,
      "learning_rate": 2.849782643696667e-08,
      "loss": 0.0467,
      "step": 6162
    },
    {
      "epoch": 0.9721970264621209,
      "grad_norm": 9.508600234985352,
      "learning_rate": 2.8336821767831264e-08,
      "loss": 0.0535,
      "step": 6163
    },
    {
      "epoch": 0.9723547738297117,
      "grad_norm": 1.484305739402771,
      "learning_rate": 2.817581709869586e-08,
      "loss": 0.0108,
      "step": 6164
    },
    {
      "epoch": 0.9725125211973025,
      "grad_norm": 7.577390670776367,
      "learning_rate": 2.8014812429560457e-08,
      "loss": 0.0834,
      "step": 6165
    },
    {
      "epoch": 0.9726702685648934,
      "grad_norm": 3.675633192062378,
      "learning_rate": 2.7853807760425053e-08,
      "loss": 0.0142,
      "step": 6166
    },
    {
      "epoch": 0.9728280159324841,
      "grad_norm": 4.311420440673828,
      "learning_rate": 2.7692803091289646e-08,
      "loss": 0.0394,
      "step": 6167
    },
    {
      "epoch": 0.9729857633000749,
      "grad_norm": 3.139410972595215,
      "learning_rate": 2.753179842215424e-08,
      "loss": 0.0108,
      "step": 6168
    },
    {
      "epoch": 0.9731435106676657,
      "grad_norm": 3.7535455226898193,
      "learning_rate": 2.737079375301884e-08,
      "loss": 0.0435,
      "step": 6169
    },
    {
      "epoch": 0.9733012580352566,
      "grad_norm": 4.458008766174316,
      "learning_rate": 2.7209789083883432e-08,
      "loss": 0.0391,
      "step": 6170
    },
    {
      "epoch": 0.9734590054028474,
      "grad_norm": 5.988290786743164,
      "learning_rate": 2.7048784414748025e-08,
      "loss": 0.0514,
      "step": 6171
    },
    {
      "epoch": 0.9736167527704381,
      "grad_norm": 6.685472011566162,
      "learning_rate": 2.688777974561262e-08,
      "loss": 0.0549,
      "step": 6172
    },
    {
      "epoch": 0.9737745001380289,
      "grad_norm": 4.843076705932617,
      "learning_rate": 2.6726775076477218e-08,
      "loss": 0.0521,
      "step": 6173
    },
    {
      "epoch": 0.9739322475056198,
      "grad_norm": 4.8866753578186035,
      "learning_rate": 2.656577040734181e-08,
      "loss": 0.0232,
      "step": 6174
    },
    {
      "epoch": 0.9740899948732106,
      "grad_norm": 7.195474624633789,
      "learning_rate": 2.6404765738206407e-08,
      "loss": 0.0383,
      "step": 6175
    },
    {
      "epoch": 0.9742477422408014,
      "grad_norm": 5.306886672973633,
      "learning_rate": 2.6243761069071e-08,
      "loss": 0.0575,
      "step": 6176
    },
    {
      "epoch": 0.9744054896083921,
      "grad_norm": 5.701973915100098,
      "learning_rate": 2.6082756399935597e-08,
      "loss": 0.0962,
      "step": 6177
    },
    {
      "epoch": 0.974563236975983,
      "grad_norm": 5.497990131378174,
      "learning_rate": 2.5921751730800193e-08,
      "loss": 0.0427,
      "step": 6178
    },
    {
      "epoch": 0.9747209843435738,
      "grad_norm": 3.343956708908081,
      "learning_rate": 2.5760747061664786e-08,
      "loss": 0.0153,
      "step": 6179
    },
    {
      "epoch": 0.9748787317111646,
      "grad_norm": 4.167031288146973,
      "learning_rate": 2.5599742392529383e-08,
      "loss": 0.0753,
      "step": 6180
    },
    {
      "epoch": 0.9750364790787553,
      "grad_norm": 2.9136710166931152,
      "learning_rate": 2.543873772339398e-08,
      "loss": 0.0248,
      "step": 6181
    },
    {
      "epoch": 0.9751942264463461,
      "grad_norm": 7.567819118499756,
      "learning_rate": 2.5277733054258572e-08,
      "loss": 0.0817,
      "step": 6182
    },
    {
      "epoch": 0.975351973813937,
      "grad_norm": 3.8892838954925537,
      "learning_rate": 2.5116728385123165e-08,
      "loss": 0.0153,
      "step": 6183
    },
    {
      "epoch": 0.9755097211815278,
      "grad_norm": 4.018087387084961,
      "learning_rate": 2.4955723715987765e-08,
      "loss": 0.0217,
      "step": 6184
    },
    {
      "epoch": 0.9756674685491186,
      "grad_norm": 5.014780521392822,
      "learning_rate": 2.4794719046852358e-08,
      "loss": 0.0355,
      "step": 6185
    },
    {
      "epoch": 0.9758252159167093,
      "grad_norm": 5.33723258972168,
      "learning_rate": 2.463371437771695e-08,
      "loss": 0.049,
      "step": 6186
    },
    {
      "epoch": 0.9759829632843002,
      "grad_norm": 2.67022967338562,
      "learning_rate": 2.4472709708581547e-08,
      "loss": 0.0454,
      "step": 6187
    },
    {
      "epoch": 0.976140710651891,
      "grad_norm": 1.7317997217178345,
      "learning_rate": 2.4311705039446144e-08,
      "loss": 0.0114,
      "step": 6188
    },
    {
      "epoch": 0.9762984580194818,
      "grad_norm": 5.719967365264893,
      "learning_rate": 2.4150700370310737e-08,
      "loss": 0.0529,
      "step": 6189
    },
    {
      "epoch": 0.9764562053870726,
      "grad_norm": 5.98928689956665,
      "learning_rate": 2.3989695701175333e-08,
      "loss": 0.0404,
      "step": 6190
    },
    {
      "epoch": 0.9766139527546635,
      "grad_norm": 3.8387529850006104,
      "learning_rate": 2.3828691032039926e-08,
      "loss": 0.0513,
      "step": 6191
    },
    {
      "epoch": 0.9767717001222542,
      "grad_norm": 2.3278558254241943,
      "learning_rate": 2.3667686362904523e-08,
      "loss": 0.0174,
      "step": 6192
    },
    {
      "epoch": 0.976929447489845,
      "grad_norm": 5.587123394012451,
      "learning_rate": 2.350668169376912e-08,
      "loss": 0.0136,
      "step": 6193
    },
    {
      "epoch": 0.9770871948574358,
      "grad_norm": 2.9909167289733887,
      "learning_rate": 2.3345677024633712e-08,
      "loss": 0.0427,
      "step": 6194
    },
    {
      "epoch": 0.9772449422250267,
      "grad_norm": 5.721266746520996,
      "learning_rate": 2.3184672355498305e-08,
      "loss": 0.062,
      "step": 6195
    },
    {
      "epoch": 0.9774026895926174,
      "grad_norm": 7.145239353179932,
      "learning_rate": 2.3023667686362905e-08,
      "loss": 0.044,
      "step": 6196
    },
    {
      "epoch": 0.9775604369602082,
      "grad_norm": 5.093533515930176,
      "learning_rate": 2.2862663017227498e-08,
      "loss": 0.0513,
      "step": 6197
    },
    {
      "epoch": 0.977718184327799,
      "grad_norm": 3.0863187313079834,
      "learning_rate": 2.2701658348092095e-08,
      "loss": 0.0454,
      "step": 6198
    },
    {
      "epoch": 0.9778759316953899,
      "grad_norm": 6.698001384735107,
      "learning_rate": 2.2540653678956688e-08,
      "loss": 0.0363,
      "step": 6199
    },
    {
      "epoch": 0.9780336790629807,
      "grad_norm": 3.7999961376190186,
      "learning_rate": 2.2379649009821284e-08,
      "loss": 0.0198,
      "step": 6200
    },
    {
      "epoch": 0.9781914264305714,
      "grad_norm": 6.384427070617676,
      "learning_rate": 2.221864434068588e-08,
      "loss": 0.0518,
      "step": 6201
    },
    {
      "epoch": 0.9783491737981622,
      "grad_norm": 3.1141104698181152,
      "learning_rate": 2.2057639671550473e-08,
      "loss": 0.012,
      "step": 6202
    },
    {
      "epoch": 0.978506921165753,
      "grad_norm": 4.341085910797119,
      "learning_rate": 2.189663500241507e-08,
      "loss": 0.0257,
      "step": 6203
    },
    {
      "epoch": 0.9786646685333439,
      "grad_norm": 6.542384624481201,
      "learning_rate": 2.1735630333279666e-08,
      "loss": 0.0359,
      "step": 6204
    },
    {
      "epoch": 0.9788224159009347,
      "grad_norm": 3.0185017585754395,
      "learning_rate": 2.157462566414426e-08,
      "loss": 0.0473,
      "step": 6205
    },
    {
      "epoch": 0.9789801632685254,
      "grad_norm": 6.168667793273926,
      "learning_rate": 2.1413620995008852e-08,
      "loss": 0.0841,
      "step": 6206
    },
    {
      "epoch": 0.9791379106361162,
      "grad_norm": 2.684790849685669,
      "learning_rate": 2.1252616325873452e-08,
      "loss": 0.0145,
      "step": 6207
    },
    {
      "epoch": 0.9792956580037071,
      "grad_norm": 6.765401840209961,
      "learning_rate": 2.1091611656738045e-08,
      "loss": 0.0161,
      "step": 6208
    },
    {
      "epoch": 0.9794534053712979,
      "grad_norm": 4.448896884918213,
      "learning_rate": 2.0930606987602638e-08,
      "loss": 0.0366,
      "step": 6209
    },
    {
      "epoch": 0.9796111527388887,
      "grad_norm": 4.182690620422363,
      "learning_rate": 2.0769602318467235e-08,
      "loss": 0.0153,
      "step": 6210
    },
    {
      "epoch": 0.9797689001064794,
      "grad_norm": 5.692211627960205,
      "learning_rate": 2.060859764933183e-08,
      "loss": 0.0534,
      "step": 6211
    },
    {
      "epoch": 0.9799266474740703,
      "grad_norm": 5.768784523010254,
      "learning_rate": 2.0447592980196424e-08,
      "loss": 0.0538,
      "step": 6212
    },
    {
      "epoch": 0.9800843948416611,
      "grad_norm": 4.9965596199035645,
      "learning_rate": 2.028658831106102e-08,
      "loss": 0.0812,
      "step": 6213
    },
    {
      "epoch": 0.9802421422092519,
      "grad_norm": 5.822160243988037,
      "learning_rate": 2.0125583641925614e-08,
      "loss": 0.0514,
      "step": 6214
    },
    {
      "epoch": 0.9803998895768427,
      "grad_norm": 9.266775131225586,
      "learning_rate": 1.996457897279021e-08,
      "loss": 0.0384,
      "step": 6215
    },
    {
      "epoch": 0.9805576369444335,
      "grad_norm": 5.855165004730225,
      "learning_rate": 1.9803574303654806e-08,
      "loss": 0.0849,
      "step": 6216
    },
    {
      "epoch": 0.9807153843120243,
      "grad_norm": 4.29231071472168,
      "learning_rate": 1.96425696345194e-08,
      "loss": 0.019,
      "step": 6217
    },
    {
      "epoch": 0.9808731316796151,
      "grad_norm": 5.040156841278076,
      "learning_rate": 1.9481564965383993e-08,
      "loss": 0.0281,
      "step": 6218
    },
    {
      "epoch": 0.9810308790472059,
      "grad_norm": 3.730966806411743,
      "learning_rate": 1.9320560296248592e-08,
      "loss": 0.0219,
      "step": 6219
    },
    {
      "epoch": 0.9811886264147966,
      "grad_norm": 3.881930112838745,
      "learning_rate": 1.9159555627113185e-08,
      "loss": 0.0335,
      "step": 6220
    },
    {
      "epoch": 0.9813463737823875,
      "grad_norm": 6.038278102874756,
      "learning_rate": 1.899855095797778e-08,
      "loss": 0.098,
      "step": 6221
    },
    {
      "epoch": 0.9815041211499783,
      "grad_norm": 3.6147210597991943,
      "learning_rate": 1.8837546288842375e-08,
      "loss": 0.0243,
      "step": 6222
    },
    {
      "epoch": 0.9816618685175691,
      "grad_norm": 5.513983726501465,
      "learning_rate": 1.867654161970697e-08,
      "loss": 0.076,
      "step": 6223
    },
    {
      "epoch": 0.9818196158851599,
      "grad_norm": 7.426619529724121,
      "learning_rate": 1.8515536950571564e-08,
      "loss": 0.0333,
      "step": 6224
    },
    {
      "epoch": 0.9819773632527508,
      "grad_norm": 3.414789915084839,
      "learning_rate": 1.835453228143616e-08,
      "loss": 0.0503,
      "step": 6225
    },
    {
      "epoch": 0.9821351106203415,
      "grad_norm": 9.266975402832031,
      "learning_rate": 1.8193527612300757e-08,
      "loss": 0.043,
      "step": 6226
    },
    {
      "epoch": 0.9822928579879323,
      "grad_norm": 3.6152501106262207,
      "learning_rate": 1.803252294316535e-08,
      "loss": 0.0377,
      "step": 6227
    },
    {
      "epoch": 0.9824506053555231,
      "grad_norm": 2.5475409030914307,
      "learning_rate": 1.7871518274029946e-08,
      "loss": 0.0451,
      "step": 6228
    },
    {
      "epoch": 0.982608352723114,
      "grad_norm": 2.931168556213379,
      "learning_rate": 1.771051360489454e-08,
      "loss": 0.0191,
      "step": 6229
    },
    {
      "epoch": 0.9827661000907048,
      "grad_norm": 1.9964200258255005,
      "learning_rate": 1.754950893575914e-08,
      "loss": 0.0101,
      "step": 6230
    },
    {
      "epoch": 0.9829238474582955,
      "grad_norm": 5.205655097961426,
      "learning_rate": 1.7388504266623732e-08,
      "loss": 0.0309,
      "step": 6231
    },
    {
      "epoch": 0.9830815948258863,
      "grad_norm": 3.76123046875,
      "learning_rate": 1.7227499597488325e-08,
      "loss": 0.0293,
      "step": 6232
    },
    {
      "epoch": 0.9832393421934772,
      "grad_norm": 4.543227672576904,
      "learning_rate": 1.7066494928352922e-08,
      "loss": 0.0499,
      "step": 6233
    },
    {
      "epoch": 0.983397089561068,
      "grad_norm": 6.56539249420166,
      "learning_rate": 1.6905490259217518e-08,
      "loss": 0.0418,
      "step": 6234
    },
    {
      "epoch": 0.9835548369286587,
      "grad_norm": 3.8159878253936768,
      "learning_rate": 1.674448559008211e-08,
      "loss": 0.0395,
      "step": 6235
    },
    {
      "epoch": 0.9837125842962495,
      "grad_norm": 4.285485744476318,
      "learning_rate": 1.6583480920946708e-08,
      "loss": 0.0381,
      "step": 6236
    },
    {
      "epoch": 0.9838703316638404,
      "grad_norm": 4.0999627113342285,
      "learning_rate": 1.64224762518113e-08,
      "loss": 0.04,
      "step": 6237
    },
    {
      "epoch": 0.9840280790314312,
      "grad_norm": 5.697142601013184,
      "learning_rate": 1.6261471582675897e-08,
      "loss": 0.038,
      "step": 6238
    },
    {
      "epoch": 0.984185826399022,
      "grad_norm": 3.3728394508361816,
      "learning_rate": 1.6100466913540494e-08,
      "loss": 0.0185,
      "step": 6239
    },
    {
      "epoch": 0.9843435737666127,
      "grad_norm": 4.262262344360352,
      "learning_rate": 1.5939462244405087e-08,
      "loss": 0.0318,
      "step": 6240
    },
    {
      "epoch": 0.9845013211342035,
      "grad_norm": 5.060286998748779,
      "learning_rate": 1.577845757526968e-08,
      "loss": 0.0268,
      "step": 6241
    },
    {
      "epoch": 0.9846590685017944,
      "grad_norm": 8.937932968139648,
      "learning_rate": 1.5617452906134276e-08,
      "loss": 0.0384,
      "step": 6242
    },
    {
      "epoch": 0.9848168158693852,
      "grad_norm": 5.5496110916137695,
      "learning_rate": 1.5456448236998872e-08,
      "loss": 0.0177,
      "step": 6243
    },
    {
      "epoch": 0.984974563236976,
      "grad_norm": 2.7683727741241455,
      "learning_rate": 1.529544356786347e-08,
      "loss": 0.0172,
      "step": 6244
    },
    {
      "epoch": 0.9851323106045667,
      "grad_norm": 4.80905294418335,
      "learning_rate": 1.5134438898728062e-08,
      "loss": 0.0532,
      "step": 6245
    },
    {
      "epoch": 0.9852900579721576,
      "grad_norm": 6.039252281188965,
      "learning_rate": 1.4973434229592658e-08,
      "loss": 0.0267,
      "step": 6246
    },
    {
      "epoch": 0.9854478053397484,
      "grad_norm": 5.504446983337402,
      "learning_rate": 1.4812429560457253e-08,
      "loss": 0.0386,
      "step": 6247
    },
    {
      "epoch": 0.9856055527073392,
      "grad_norm": 3.770838975906372,
      "learning_rate": 1.4651424891321848e-08,
      "loss": 0.0688,
      "step": 6248
    },
    {
      "epoch": 0.98576330007493,
      "grad_norm": 2.9576942920684814,
      "learning_rate": 1.4490420222186443e-08,
      "loss": 0.0292,
      "step": 6249
    },
    {
      "epoch": 0.9859210474425208,
      "grad_norm": 2.8827409744262695,
      "learning_rate": 1.4329415553051039e-08,
      "loss": 0.0583,
      "step": 6250
    },
    {
      "epoch": 0.9860787948101116,
      "grad_norm": 5.529241561889648,
      "learning_rate": 1.4168410883915632e-08,
      "loss": 0.0552,
      "step": 6251
    },
    {
      "epoch": 0.9862365421777024,
      "grad_norm": 7.1768293380737305,
      "learning_rate": 1.4007406214780228e-08,
      "loss": 0.0735,
      "step": 6252
    },
    {
      "epoch": 0.9863942895452932,
      "grad_norm": 5.332305431365967,
      "learning_rate": 1.3846401545644823e-08,
      "loss": 0.0702,
      "step": 6253
    },
    {
      "epoch": 0.9865520369128841,
      "grad_norm": 4.009863376617432,
      "learning_rate": 1.368539687650942e-08,
      "loss": 0.0185,
      "step": 6254
    },
    {
      "epoch": 0.9867097842804748,
      "grad_norm": 9.313681602478027,
      "learning_rate": 1.3524392207374013e-08,
      "loss": 0.0651,
      "step": 6255
    },
    {
      "epoch": 0.9868675316480656,
      "grad_norm": 4.814515113830566,
      "learning_rate": 1.3363387538238609e-08,
      "loss": 0.0566,
      "step": 6256
    },
    {
      "epoch": 0.9870252790156564,
      "grad_norm": 1.7556432485580444,
      "learning_rate": 1.3202382869103204e-08,
      "loss": 0.0116,
      "step": 6257
    },
    {
      "epoch": 0.9871830263832472,
      "grad_norm": 3.093322515487671,
      "learning_rate": 1.3041378199967798e-08,
      "loss": 0.0229,
      "step": 6258
    },
    {
      "epoch": 0.9873407737508381,
      "grad_norm": 5.676116466522217,
      "learning_rate": 1.2880373530832393e-08,
      "loss": 0.0502,
      "step": 6259
    },
    {
      "epoch": 0.9874985211184288,
      "grad_norm": 4.5010857582092285,
      "learning_rate": 1.271936886169699e-08,
      "loss": 0.0453,
      "step": 6260
    },
    {
      "epoch": 0.9876562684860196,
      "grad_norm": 5.022231101989746,
      "learning_rate": 1.2558364192561583e-08,
      "loss": 0.0413,
      "step": 6261
    },
    {
      "epoch": 0.9878140158536104,
      "grad_norm": 2.8914976119995117,
      "learning_rate": 1.2397359523426179e-08,
      "loss": 0.0098,
      "step": 6262
    },
    {
      "epoch": 0.9879717632212013,
      "grad_norm": 4.519913196563721,
      "learning_rate": 1.2236354854290774e-08,
      "loss": 0.0255,
      "step": 6263
    },
    {
      "epoch": 0.9881295105887921,
      "grad_norm": 3.478266477584839,
      "learning_rate": 1.2075350185155368e-08,
      "loss": 0.0163,
      "step": 6264
    },
    {
      "epoch": 0.9882872579563828,
      "grad_norm": 5.429773807525635,
      "learning_rate": 1.1914345516019963e-08,
      "loss": 0.0466,
      "step": 6265
    },
    {
      "epoch": 0.9884450053239736,
      "grad_norm": 5.458247184753418,
      "learning_rate": 1.175334084688456e-08,
      "loss": 0.0484,
      "step": 6266
    },
    {
      "epoch": 0.9886027526915645,
      "grad_norm": 4.081238269805908,
      "learning_rate": 1.1592336177749153e-08,
      "loss": 0.0208,
      "step": 6267
    },
    {
      "epoch": 0.9887605000591553,
      "grad_norm": 6.506401062011719,
      "learning_rate": 1.1431331508613749e-08,
      "loss": 0.0308,
      "step": 6268
    },
    {
      "epoch": 0.988918247426746,
      "grad_norm": 5.5140838623046875,
      "learning_rate": 1.1270326839478344e-08,
      "loss": 0.0566,
      "step": 6269
    },
    {
      "epoch": 0.9890759947943368,
      "grad_norm": 6.035703659057617,
      "learning_rate": 1.110932217034294e-08,
      "loss": 0.0482,
      "step": 6270
    },
    {
      "epoch": 0.9892337421619277,
      "grad_norm": 5.993376731872559,
      "learning_rate": 1.0948317501207535e-08,
      "loss": 0.0602,
      "step": 6271
    },
    {
      "epoch": 0.9893914895295185,
      "grad_norm": 4.419226169586182,
      "learning_rate": 1.078731283207213e-08,
      "loss": 0.0416,
      "step": 6272
    },
    {
      "epoch": 0.9895492368971093,
      "grad_norm": 4.298280239105225,
      "learning_rate": 1.0626308162936726e-08,
      "loss": 0.0289,
      "step": 6273
    },
    {
      "epoch": 0.9897069842647,
      "grad_norm": 3.5037503242492676,
      "learning_rate": 1.0465303493801319e-08,
      "loss": 0.0404,
      "step": 6274
    },
    {
      "epoch": 0.9898647316322909,
      "grad_norm": 5.784923553466797,
      "learning_rate": 1.0304298824665916e-08,
      "loss": 0.0409,
      "step": 6275
    },
    {
      "epoch": 0.9900224789998817,
      "grad_norm": 3.270423173904419,
      "learning_rate": 1.014329415553051e-08,
      "loss": 0.0308,
      "step": 6276
    },
    {
      "epoch": 0.9901802263674725,
      "grad_norm": 3.4138495922088623,
      "learning_rate": 9.982289486395105e-09,
      "loss": 0.016,
      "step": 6277
    },
    {
      "epoch": 0.9903379737350633,
      "grad_norm": 7.211464881896973,
      "learning_rate": 9.8212848172597e-09,
      "loss": 0.0435,
      "step": 6278
    },
    {
      "epoch": 0.990495721102654,
      "grad_norm": 5.604741096496582,
      "learning_rate": 9.660280148124296e-09,
      "loss": 0.0377,
      "step": 6279
    },
    {
      "epoch": 0.9906534684702449,
      "grad_norm": 5.227348327636719,
      "learning_rate": 9.49927547898889e-09,
      "loss": 0.0307,
      "step": 6280
    },
    {
      "epoch": 0.9908112158378357,
      "grad_norm": 4.079761505126953,
      "learning_rate": 9.338270809853486e-09,
      "loss": 0.0744,
      "step": 6281
    },
    {
      "epoch": 0.9909689632054265,
      "grad_norm": 7.002961158752441,
      "learning_rate": 9.17726614071808e-09,
      "loss": 0.0492,
      "step": 6282
    },
    {
      "epoch": 0.9911267105730173,
      "grad_norm": 6.92935037612915,
      "learning_rate": 9.016261471582675e-09,
      "loss": 0.0401,
      "step": 6283
    },
    {
      "epoch": 0.9912844579406082,
      "grad_norm": 3.1296815872192383,
      "learning_rate": 8.85525680244727e-09,
      "loss": 0.0186,
      "step": 6284
    },
    {
      "epoch": 0.9914422053081989,
      "grad_norm": 4.725761890411377,
      "learning_rate": 8.694252133311866e-09,
      "loss": 0.038,
      "step": 6285
    },
    {
      "epoch": 0.9915999526757897,
      "grad_norm": 8.60621452331543,
      "learning_rate": 8.533247464176461e-09,
      "loss": 0.0718,
      "step": 6286
    },
    {
      "epoch": 0.9917577000433805,
      "grad_norm": 2.2828660011291504,
      "learning_rate": 8.372242795041056e-09,
      "loss": 0.0181,
      "step": 6287
    },
    {
      "epoch": 0.9919154474109714,
      "grad_norm": 5.0544867515563965,
      "learning_rate": 8.21123812590565e-09,
      "loss": 0.0244,
      "step": 6288
    },
    {
      "epoch": 0.9920731947785622,
      "grad_norm": 8.647851943969727,
      "learning_rate": 8.050233456770247e-09,
      "loss": 0.0428,
      "step": 6289
    },
    {
      "epoch": 0.9922309421461529,
      "grad_norm": 2.197507858276367,
      "learning_rate": 7.88922878763484e-09,
      "loss": 0.0136,
      "step": 6290
    },
    {
      "epoch": 0.9923886895137437,
      "grad_norm": 3.088834524154663,
      "learning_rate": 7.728224118499436e-09,
      "loss": 0.0199,
      "step": 6291
    },
    {
      "epoch": 0.9925464368813346,
      "grad_norm": 5.041046619415283,
      "learning_rate": 7.567219449364031e-09,
      "loss": 0.0793,
      "step": 6292
    },
    {
      "epoch": 0.9927041842489254,
      "grad_norm": 4.296901702880859,
      "learning_rate": 7.4062147802286265e-09,
      "loss": 0.0247,
      "step": 6293
    },
    {
      "epoch": 0.9928619316165161,
      "grad_norm": 5.000230312347412,
      "learning_rate": 7.245210111093221e-09,
      "loss": 0.0306,
      "step": 6294
    },
    {
      "epoch": 0.9930196789841069,
      "grad_norm": 4.815591812133789,
      "learning_rate": 7.084205441957816e-09,
      "loss": 0.0305,
      "step": 6295
    },
    {
      "epoch": 0.9931774263516977,
      "grad_norm": 2.091221809387207,
      "learning_rate": 6.9232007728224115e-09,
      "loss": 0.0372,
      "step": 6296
    },
    {
      "epoch": 0.9933351737192886,
      "grad_norm": 7.173551559448242,
      "learning_rate": 6.762196103687006e-09,
      "loss": 0.0485,
      "step": 6297
    },
    {
      "epoch": 0.9934929210868794,
      "grad_norm": 9.944113731384277,
      "learning_rate": 6.601191434551602e-09,
      "loss": 0.1298,
      "step": 6298
    },
    {
      "epoch": 0.9936506684544701,
      "grad_norm": 2.1898152828216553,
      "learning_rate": 6.4401867654161966e-09,
      "loss": 0.0382,
      "step": 6299
    },
    {
      "epoch": 0.9938084158220609,
      "grad_norm": 7.77823543548584,
      "learning_rate": 6.279182096280791e-09,
      "loss": 0.0489,
      "step": 6300
    },
    {
      "epoch": 0.9939661631896518,
      "grad_norm": 1.7604883909225464,
      "learning_rate": 6.118177427145387e-09,
      "loss": 0.0093,
      "step": 6301
    },
    {
      "epoch": 0.9941239105572426,
      "grad_norm": 6.472969055175781,
      "learning_rate": 5.957172758009982e-09,
      "loss": 0.037,
      "step": 6302
    },
    {
      "epoch": 0.9942816579248334,
      "grad_norm": 4.241221904754639,
      "learning_rate": 5.796168088874576e-09,
      "loss": 0.0185,
      "step": 6303
    },
    {
      "epoch": 0.9944394052924241,
      "grad_norm": 4.061306476593018,
      "learning_rate": 5.635163419739172e-09,
      "loss": 0.0231,
      "step": 6304
    },
    {
      "epoch": 0.994597152660015,
      "grad_norm": 5.12284517288208,
      "learning_rate": 5.4741587506037675e-09,
      "loss": 0.0429,
      "step": 6305
    },
    {
      "epoch": 0.9947549000276058,
      "grad_norm": 5.1497368812561035,
      "learning_rate": 5.313154081468363e-09,
      "loss": 0.0668,
      "step": 6306
    },
    {
      "epoch": 0.9949126473951966,
      "grad_norm": 8.419878005981445,
      "learning_rate": 5.152149412332958e-09,
      "loss": 0.0274,
      "step": 6307
    },
    {
      "epoch": 0.9950703947627874,
      "grad_norm": 5.198375225067139,
      "learning_rate": 4.9911447431975525e-09,
      "loss": 0.0306,
      "step": 6308
    },
    {
      "epoch": 0.9952281421303782,
      "grad_norm": 2.8158767223358154,
      "learning_rate": 4.830140074062148e-09,
      "loss": 0.0337,
      "step": 6309
    },
    {
      "epoch": 0.995385889497969,
      "grad_norm": 3.8012020587921143,
      "learning_rate": 4.669135404926743e-09,
      "loss": 0.0688,
      "step": 6310
    },
    {
      "epoch": 0.9955436368655598,
      "grad_norm": 4.970787525177002,
      "learning_rate": 4.5081307357913375e-09,
      "loss": 0.0504,
      "step": 6311
    },
    {
      "epoch": 0.9957013842331506,
      "grad_norm": 4.551598072052002,
      "learning_rate": 4.347126066655933e-09,
      "loss": 0.0289,
      "step": 6312
    },
    {
      "epoch": 0.9958591316007415,
      "grad_norm": 4.775918006896973,
      "learning_rate": 4.186121397520528e-09,
      "loss": 0.0391,
      "step": 6313
    },
    {
      "epoch": 0.9960168789683322,
      "grad_norm": 18.21245765686035,
      "learning_rate": 4.025116728385123e-09,
      "loss": 0.0501,
      "step": 6314
    },
    {
      "epoch": 0.996174626335923,
      "grad_norm": 5.041269779205322,
      "learning_rate": 3.864112059249718e-09,
      "loss": 0.0385,
      "step": 6315
    },
    {
      "epoch": 0.9963323737035138,
      "grad_norm": 5.841076850891113,
      "learning_rate": 3.7031073901143133e-09,
      "loss": 0.0392,
      "step": 6316
    },
    {
      "epoch": 0.9964901210711046,
      "grad_norm": 5.748498439788818,
      "learning_rate": 3.542102720978908e-09,
      "loss": 0.04,
      "step": 6317
    },
    {
      "epoch": 0.9966478684386955,
      "grad_norm": 4.236159324645996,
      "learning_rate": 3.381098051843503e-09,
      "loss": 0.0744,
      "step": 6318
    },
    {
      "epoch": 0.9968056158062862,
      "grad_norm": 3.2368459701538086,
      "learning_rate": 3.2200933827080983e-09,
      "loss": 0.0172,
      "step": 6319
    },
    {
      "epoch": 0.996963363173877,
      "grad_norm": 6.562069892883301,
      "learning_rate": 3.0590887135726934e-09,
      "loss": 0.0441,
      "step": 6320
    },
    {
      "epoch": 0.9971211105414678,
      "grad_norm": 3.658777952194214,
      "learning_rate": 2.898084044437288e-09,
      "loss": 0.0324,
      "step": 6321
    },
    {
      "epoch": 0.9972788579090587,
      "grad_norm": 7.275805473327637,
      "learning_rate": 2.7370793753018837e-09,
      "loss": 0.0652,
      "step": 6322
    },
    {
      "epoch": 0.9974366052766495,
      "grad_norm": 8.29052734375,
      "learning_rate": 2.576074706166479e-09,
      "loss": 0.0461,
      "step": 6323
    },
    {
      "epoch": 0.9975943526442402,
      "grad_norm": 7.682597637176514,
      "learning_rate": 2.415070037031074e-09,
      "loss": 0.0827,
      "step": 6324
    },
    {
      "epoch": 0.997752100011831,
      "grad_norm": 7.659365653991699,
      "learning_rate": 2.2540653678956688e-09,
      "loss": 0.0365,
      "step": 6325
    },
    {
      "epoch": 0.9979098473794219,
      "grad_norm": 8.81132984161377,
      "learning_rate": 2.093060698760264e-09,
      "loss": 0.0539,
      "step": 6326
    },
    {
      "epoch": 0.9980675947470127,
      "grad_norm": 3.7465226650238037,
      "learning_rate": 1.932056029624859e-09,
      "loss": 0.0224,
      "step": 6327
    },
    {
      "epoch": 0.9982253421146035,
      "grad_norm": 5.081471920013428,
      "learning_rate": 1.771051360489454e-09,
      "loss": 0.04,
      "step": 6328
    },
    {
      "epoch": 0.9983830894821942,
      "grad_norm": 2.309492349624634,
      "learning_rate": 1.6100466913540491e-09,
      "loss": 0.0248,
      "step": 6329
    },
    {
      "epoch": 0.9985408368497851,
      "grad_norm": 13.910364151000977,
      "learning_rate": 1.449042022218644e-09,
      "loss": 0.0439,
      "step": 6330
    },
    {
      "epoch": 0.9986985842173759,
      "grad_norm": 3.937494993209839,
      "learning_rate": 1.2880373530832394e-09,
      "loss": 0.0307,
      "step": 6331
    },
    {
      "epoch": 0.9988563315849667,
      "grad_norm": 4.481929302215576,
      "learning_rate": 1.1270326839478344e-09,
      "loss": 0.0442,
      "step": 6332
    },
    {
      "epoch": 0.9990140789525574,
      "grad_norm": 6.280672550201416,
      "learning_rate": 9.660280148124295e-10,
      "loss": 0.0558,
      "step": 6333
    },
    {
      "epoch": 0.9991718263201483,
      "grad_norm": 4.704923629760742,
      "learning_rate": 8.050233456770246e-10,
      "loss": 0.0198,
      "step": 6334
    },
    {
      "epoch": 0.9993295736877391,
      "grad_norm": 2.0602962970733643,
      "learning_rate": 6.440186765416197e-10,
      "loss": 0.0178,
      "step": 6335
    },
    {
      "epoch": 0.9994873210553299,
      "grad_norm": 3.155426025390625,
      "learning_rate": 4.830140074062148e-10,
      "loss": 0.0311,
      "step": 6336
    },
    {
      "epoch": 0.9996450684229207,
      "grad_norm": 5.537027835845947,
      "learning_rate": 3.2200933827080986e-10,
      "loss": 0.0388,
      "step": 6337
    },
    {
      "epoch": 0.9998028157905114,
      "grad_norm": 3.089198589324951,
      "learning_rate": 1.6100466913540493e-10,
      "loss": 0.0198,
      "step": 6338
    },
    {
      "epoch": 0.9999605631581023,
      "grad_norm": 5.744074821472168,
      "learning_rate": 0.0,
      "loss": 0.0391,
      "step": 6339
    },
    {
      "epoch": 0.9999605631581023,
      "step": 6339,
      "total_flos": 1.5123386349736428e+18,
      "train_loss": 0.08585457195308795,
      "train_runtime": 144531.5462,
      "train_samples_per_second": 11.228,
      "train_steps_per_second": 0.044
    }
  ],
  "logging_steps": 1,
  "max_steps": 6339,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 768,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.5123386349736428e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}