{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 7902,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00012655024044545685,
      "grad_norm": 13.985199928283691,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 1
    },
    {
      "epoch": 0.0002531004808909137,
      "grad_norm": 13.90275764465332,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.768,
      "step": 2
    },
    {
      "epoch": 0.00037965072133637056,
      "grad_norm": 15.523126602172852,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.783,
      "step": 3
    },
    {
      "epoch": 0.0005062009617818274,
      "grad_norm": 10.846606254577637,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.7935,
      "step": 4
    },
    {
      "epoch": 0.0006327512022272843,
      "grad_norm": 5.396270751953125,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.6984,
      "step": 5
    },
    {
      "epoch": 0.0007593014426727411,
      "grad_norm": 4.489923000335693,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.7164,
      "step": 6
    },
    {
      "epoch": 0.000885851683118198,
      "grad_norm": 4.679110050201416,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.7107,
      "step": 7
    },
    {
      "epoch": 0.0010124019235636548,
      "grad_norm": 3.9017746448516846,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.7157,
      "step": 8
    },
    {
      "epoch": 0.0011389521640091116,
      "grad_norm": 4.059990882873535,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.7068,
      "step": 9
    },
    {
      "epoch": 0.0012655024044545685,
      "grad_norm": 3.291783571243286,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.708,
      "step": 10
    },
    {
      "epoch": 0.0013920526449000253,
      "grad_norm": 2.2814252376556396,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.6423,
      "step": 11
    },
    {
      "epoch": 0.0015186028853454822,
      "grad_norm": 3.1553704738616943,
      "learning_rate": 2.2e-06,
      "loss": 0.7116,
      "step": 12
    },
    {
      "epoch": 0.001645153125790939,
      "grad_norm": 2.388195037841797,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.6937,
      "step": 13
    },
    {
      "epoch": 0.001771703366236396,
      "grad_norm": 2.6007742881774902,
      "learning_rate": 2.6e-06,
      "loss": 0.6504,
      "step": 14
    },
    {
      "epoch": 0.0018982536066818527,
      "grad_norm": 2.8062355518341064,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.6291,
      "step": 15
    },
    {
      "epoch": 0.0020248038471273096,
      "grad_norm": 2.6179542541503906,
      "learning_rate": 3e-06,
      "loss": 0.7159,
      "step": 16
    },
    {
      "epoch": 0.0021513540875727666,
      "grad_norm": 2.2260546684265137,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.6576,
      "step": 17
    },
    {
      "epoch": 0.002277904328018223,
      "grad_norm": 2.672590732574463,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.6977,
      "step": 18
    },
    {
      "epoch": 0.00240445456846368,
      "grad_norm": 2.074707269668579,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.6249,
      "step": 19
    },
    {
      "epoch": 0.002531004808909137,
      "grad_norm": 2.7172651290893555,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.6818,
      "step": 20
    },
    {
      "epoch": 0.0026575550493545936,
      "grad_norm": 2.395345449447632,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.6681,
      "step": 21
    },
    {
      "epoch": 0.0027841052898000505,
      "grad_norm": 2.186948299407959,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.6435,
      "step": 22
    },
    {
      "epoch": 0.0029106555302455075,
      "grad_norm": 2.578268051147461,
      "learning_rate": 4.4e-06,
      "loss": 0.6776,
      "step": 23
    },
    {
      "epoch": 0.0030372057706909645,
      "grad_norm": 2.4086413383483887,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.6449,
      "step": 24
    },
    {
      "epoch": 0.003163756011136421,
      "grad_norm": 2.376239538192749,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.6321,
      "step": 25
    },
    {
      "epoch": 0.003290306251581878,
      "grad_norm": 2.3646175861358643,
      "learning_rate": 5e-06,
      "loss": 0.667,
      "step": 26
    },
    {
      "epoch": 0.003416856492027335,
      "grad_norm": 2.2437045574188232,
      "learning_rate": 5.2e-06,
      "loss": 0.6641,
      "step": 27
    },
    {
      "epoch": 0.003543406732472792,
      "grad_norm": 2.119607925415039,
      "learning_rate": 5.400000000000001e-06,
      "loss": 0.6009,
      "step": 28
    },
    {
      "epoch": 0.0036699569729182484,
      "grad_norm": 2.3002877235412598,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.6576,
      "step": 29
    },
    {
      "epoch": 0.0037965072133637054,
      "grad_norm": 2.6401987075805664,
      "learning_rate": 5.8e-06,
      "loss": 0.6859,
      "step": 30
    },
    {
      "epoch": 0.003923057453809162,
      "grad_norm": 2.3456244468688965,
      "learning_rate": 6e-06,
      "loss": 0.6596,
      "step": 31
    },
    {
      "epoch": 0.004049607694254619,
      "grad_norm": 2.084927797317505,
      "learning_rate": 6.200000000000001e-06,
      "loss": 0.6216,
      "step": 32
    },
    {
      "epoch": 0.004176157934700076,
      "grad_norm": 2.192030429840088,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.6664,
      "step": 33
    },
    {
      "epoch": 0.004302708175145533,
      "grad_norm": 2.009577512741089,
      "learning_rate": 6.600000000000001e-06,
      "loss": 0.6352,
      "step": 34
    },
    {
      "epoch": 0.00442925841559099,
      "grad_norm": 2.1298763751983643,
      "learning_rate": 6.800000000000001e-06,
      "loss": 0.6321,
      "step": 35
    },
    {
      "epoch": 0.004555808656036446,
      "grad_norm": 2.0489702224731445,
      "learning_rate": 7e-06,
      "loss": 0.6465,
      "step": 36
    },
    {
      "epoch": 0.004682358896481904,
      "grad_norm": 2.4162447452545166,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.6626,
      "step": 37
    },
    {
      "epoch": 0.00480890913692736,
      "grad_norm": 2.492039203643799,
      "learning_rate": 7.4e-06,
      "loss": 0.6669,
      "step": 38
    },
    {
      "epoch": 0.004935459377372817,
      "grad_norm": 2.5284948348999023,
      "learning_rate": 7.600000000000001e-06,
      "loss": 0.6416,
      "step": 39
    },
    {
      "epoch": 0.005062009617818274,
      "grad_norm": 2.5235719680786133,
      "learning_rate": 7.800000000000002e-06,
      "loss": 0.6728,
      "step": 40
    },
    {
      "epoch": 0.005188559858263731,
      "grad_norm": 2.4154090881347656,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.6956,
      "step": 41
    },
    {
      "epoch": 0.005315110098709187,
      "grad_norm": 2.3730292320251465,
      "learning_rate": 8.2e-06,
      "loss": 0.6166,
      "step": 42
    },
    {
      "epoch": 0.0054416603391546445,
      "grad_norm": 5.592038154602051,
      "learning_rate": 8.400000000000001e-06,
      "loss": 0.6709,
      "step": 43
    },
    {
      "epoch": 0.005568210579600101,
      "grad_norm": 2.4537770748138428,
      "learning_rate": 8.6e-06,
      "loss": 0.6433,
      "step": 44
    },
    {
      "epoch": 0.0056947608200455585,
      "grad_norm": 3.7874631881713867,
      "learning_rate": 8.8e-06,
      "loss": 0.6847,
      "step": 45
    },
    {
      "epoch": 0.005821311060491015,
      "grad_norm": 2.700901746749878,
      "learning_rate": 9e-06,
      "loss": 0.6626,
      "step": 46
    },
    {
      "epoch": 0.0059478613009364715,
      "grad_norm": 2.618072748184204,
      "learning_rate": 9.200000000000002e-06,
      "loss": 0.6895,
      "step": 47
    },
    {
      "epoch": 0.006074411541381929,
      "grad_norm": 2.9922256469726562,
      "learning_rate": 9.4e-06,
      "loss": 0.6448,
      "step": 48
    },
    {
      "epoch": 0.0062009617818273854,
      "grad_norm": 2.824721336364746,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.6438,
      "step": 49
    },
    {
      "epoch": 0.006327512022272842,
      "grad_norm": 2.491943836212158,
      "learning_rate": 9.800000000000001e-06,
      "loss": 0.6394,
      "step": 50
    },
    {
      "epoch": 0.006454062262718299,
      "grad_norm": 2.1831438541412354,
      "learning_rate": 1e-05,
      "loss": 0.6752,
      "step": 51
    },
    {
      "epoch": 0.006580612503163756,
      "grad_norm": 2.1266467571258545,
      "learning_rate": 1.02e-05,
      "loss": 0.6544,
      "step": 52
    },
    {
      "epoch": 0.006707162743609213,
      "grad_norm": 2.2077813148498535,
      "learning_rate": 1.04e-05,
      "loss": 0.6491,
      "step": 53
    },
    {
      "epoch": 0.00683371298405467,
      "grad_norm": 3.0424718856811523,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 0.6487,
      "step": 54
    },
    {
      "epoch": 0.006960263224500126,
      "grad_norm": 3.327303886413574,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 0.6387,
      "step": 55
    },
    {
      "epoch": 0.007086813464945584,
      "grad_norm": 2.359013557434082,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.6622,
      "step": 56
    },
    {
      "epoch": 0.00721336370539104,
      "grad_norm": 2.148637294769287,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.6655,
      "step": 57
    },
    {
      "epoch": 0.007339913945836497,
      "grad_norm": 2.2850229740142822,
      "learning_rate": 1.14e-05,
      "loss": 0.6663,
      "step": 58
    },
    {
      "epoch": 0.007466464186281954,
      "grad_norm": 3.1060843467712402,
      "learning_rate": 1.16e-05,
      "loss": 0.6801,
      "step": 59
    },
    {
      "epoch": 0.007593014426727411,
      "grad_norm": 4.193774700164795,
      "learning_rate": 1.18e-05,
      "loss": 0.6891,
      "step": 60
    },
    {
      "epoch": 0.007719564667172867,
      "grad_norm": 2.3206233978271484,
      "learning_rate": 1.2e-05,
      "loss": 0.6533,
      "step": 61
    },
    {
      "epoch": 0.007846114907618324,
      "grad_norm": 3.7517285346984863,
      "learning_rate": 1.22e-05,
      "loss": 0.7151,
      "step": 62
    },
    {
      "epoch": 0.007972665148063782,
      "grad_norm": 4.658071517944336,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 0.6897,
      "step": 63
    },
    {
      "epoch": 0.008099215388509239,
      "grad_norm": 2.69758939743042,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 0.7273,
      "step": 64
    },
    {
      "epoch": 0.008225765628954695,
      "grad_norm": 6.138875484466553,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.6641,
      "step": 65
    },
    {
      "epoch": 0.008352315869400152,
      "grad_norm": 3.6908342838287354,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.7003,
      "step": 66
    },
    {
      "epoch": 0.008478866109845608,
      "grad_norm": 6.324772834777832,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 0.7493,
      "step": 67
    },
    {
      "epoch": 0.008605416350291066,
      "grad_norm": 4.708766460418701,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 0.6987,
      "step": 68
    },
    {
      "epoch": 0.008731966590736523,
      "grad_norm": 7.057864189147949,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.7002,
      "step": 69
    },
    {
      "epoch": 0.00885851683118198,
      "grad_norm": 4.916528224945068,
      "learning_rate": 1.38e-05,
      "loss": 0.7147,
      "step": 70
    },
    {
      "epoch": 0.008985067071627436,
      "grad_norm": 5.019859313964844,
      "learning_rate": 1.4e-05,
      "loss": 0.7184,
      "step": 71
    },
    {
      "epoch": 0.009111617312072893,
      "grad_norm": 3.636657953262329,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 0.6409,
      "step": 72
    },
    {
      "epoch": 0.009238167552518349,
      "grad_norm": 9.65031909942627,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.6962,
      "step": 73
    },
    {
      "epoch": 0.009364717792963807,
      "grad_norm": 5.363977909088135,
      "learning_rate": 1.46e-05,
      "loss": 0.6974,
      "step": 74
    },
    {
      "epoch": 0.009491268033409264,
      "grad_norm": 6.134232521057129,
      "learning_rate": 1.48e-05,
      "loss": 0.7416,
      "step": 75
    },
    {
      "epoch": 0.00961781827385472,
      "grad_norm": 3.060453176498413,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.7085,
      "step": 76
    },
    {
      "epoch": 0.009744368514300177,
      "grad_norm": 10.826395034790039,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.7256,
      "step": 77
    },
    {
      "epoch": 0.009870918754745633,
      "grad_norm": 6.343544960021973,
      "learning_rate": 1.54e-05,
      "loss": 0.6727,
      "step": 78
    },
    {
      "epoch": 0.009997468995191092,
      "grad_norm": 7.497647762298584,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 0.6696,
      "step": 79
    },
    {
      "epoch": 0.010124019235636548,
      "grad_norm": 6.527174949645996,
      "learning_rate": 1.58e-05,
      "loss": 0.7316,
      "step": 80
    },
    {
      "epoch": 0.010250569476082005,
      "grad_norm": 2.816755533218384,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.6978,
      "step": 81
    },
    {
      "epoch": 0.010377119716527461,
      "grad_norm": 5.435158729553223,
      "learning_rate": 1.62e-05,
      "loss": 0.7275,
      "step": 82
    },
    {
      "epoch": 0.010503669956972918,
      "grad_norm": 4.304297924041748,
      "learning_rate": 1.64e-05,
      "loss": 0.7074,
      "step": 83
    },
    {
      "epoch": 0.010630220197418374,
      "grad_norm": 2.6429708003997803,
      "learning_rate": 1.66e-05,
      "loss": 0.6743,
      "step": 84
    },
    {
      "epoch": 0.010756770437863833,
      "grad_norm": 5.842764377593994,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.6799,
      "step": 85
    },
    {
      "epoch": 0.010883320678309289,
      "grad_norm": 2.9549334049224854,
      "learning_rate": 1.7e-05,
      "loss": 0.6524,
      "step": 86
    },
    {
      "epoch": 0.011009870918754746,
      "grad_norm": 6.567975997924805,
      "learning_rate": 1.72e-05,
      "loss": 0.7405,
      "step": 87
    },
    {
      "epoch": 0.011136421159200202,
      "grad_norm": 3.0751874446868896,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 0.711,
      "step": 88
    },
    {
      "epoch": 0.011262971399645659,
      "grad_norm": 4.5287580490112305,
      "learning_rate": 1.76e-05,
      "loss": 0.702,
      "step": 89
    },
    {
      "epoch": 0.011389521640091117,
      "grad_norm": 3.2978765964508057,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 0.7004,
      "step": 90
    },
    {
      "epoch": 0.011516071880536573,
      "grad_norm": 2.714090585708618,
      "learning_rate": 1.8e-05,
      "loss": 0.6589,
      "step": 91
    },
    {
      "epoch": 0.01164262212098203,
      "grad_norm": 4.804320335388184,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 0.736,
      "step": 92
    },
    {
      "epoch": 0.011769172361427487,
      "grad_norm": 3.215153455734253,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.7135,
      "step": 93
    },
    {
      "epoch": 0.011895722601872943,
      "grad_norm": 3.2461957931518555,
      "learning_rate": 1.86e-05,
      "loss": 0.6701,
      "step": 94
    },
    {
      "epoch": 0.0120222728423184,
      "grad_norm": 3.767245054244995,
      "learning_rate": 1.88e-05,
      "loss": 0.6852,
      "step": 95
    },
    {
      "epoch": 0.012148823082763858,
      "grad_norm": 2.4842722415924072,
      "learning_rate": 1.9e-05,
      "loss": 0.6794,
      "step": 96
    },
    {
      "epoch": 0.012275373323209314,
      "grad_norm": 10.805033683776855,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.6712,
      "step": 97
    },
    {
      "epoch": 0.012401923563654771,
      "grad_norm": 6.633640289306641,
      "learning_rate": 1.94e-05,
      "loss": 0.7051,
      "step": 98
    },
    {
      "epoch": 0.012528473804100227,
      "grad_norm": 2.86954665184021,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 0.7021,
      "step": 99
    },
    {
      "epoch": 0.012655024044545684,
      "grad_norm": 4.390101432800293,
      "learning_rate": 1.98e-05,
      "loss": 0.7163,
      "step": 100
    },
    {
      "epoch": 0.012781574284991142,
      "grad_norm": 3.448770046234131,
      "learning_rate": 2e-05,
      "loss": 0.7371,
      "step": 101
    },
    {
      "epoch": 0.012908124525436599,
      "grad_norm": 2.783313512802124,
      "learning_rate": 1.9999999839414654e-05,
      "loss": 0.6805,
      "step": 102
    },
    {
      "epoch": 0.013034674765882055,
      "grad_norm": 3.9792027473449707,
      "learning_rate": 1.9999999357658617e-05,
      "loss": 0.707,
      "step": 103
    },
    {
      "epoch": 0.013161225006327512,
      "grad_norm": 2.8279881477355957,
      "learning_rate": 1.9999998554731902e-05,
      "loss": 0.753,
      "step": 104
    },
    {
      "epoch": 0.013287775246772968,
      "grad_norm": 2.4803669452667236,
      "learning_rate": 1.9999997430634542e-05,
      "loss": 0.7095,
      "step": 105
    },
    {
      "epoch": 0.013414325487218427,
      "grad_norm": 3.6263887882232666,
      "learning_rate": 1.9999995985366568e-05,
      "loss": 0.7624,
      "step": 106
    },
    {
      "epoch": 0.013540875727663883,
      "grad_norm": 4.736232280731201,
      "learning_rate": 1.9999994218928024e-05,
      "loss": 0.757,
      "step": 107
    },
    {
      "epoch": 0.01366742596810934,
      "grad_norm": 9.867573738098145,
      "learning_rate": 1.9999992131318973e-05,
      "loss": 0.7686,
      "step": 108
    },
    {
      "epoch": 0.013793976208554796,
      "grad_norm": 24.362350463867188,
      "learning_rate": 1.9999989722539482e-05,
      "loss": 0.7586,
      "step": 109
    },
    {
      "epoch": 0.013920526449000253,
      "grad_norm": 22.680953979492188,
      "learning_rate": 1.9999986992589623e-05,
      "loss": 0.8037,
      "step": 110
    },
    {
      "epoch": 0.01404707668944571,
      "grad_norm": 5.34781551361084,
      "learning_rate": 1.9999983941469486e-05,
      "loss": 0.795,
      "step": 111
    },
    {
      "epoch": 0.014173626929891167,
      "grad_norm": 5.674261093139648,
      "learning_rate": 1.999998056917917e-05,
      "loss": 0.7813,
      "step": 112
    },
    {
      "epoch": 0.014300177170336624,
      "grad_norm": 10.219779968261719,
      "learning_rate": 1.9999976875718785e-05,
      "loss": 0.811,
      "step": 113
    },
    {
      "epoch": 0.01442672741078208,
      "grad_norm": 3.841041088104248,
      "learning_rate": 1.9999972861088444e-05,
      "loss": 0.7428,
      "step": 114
    },
    {
      "epoch": 0.014553277651227537,
      "grad_norm": 3.9405486583709717,
      "learning_rate": 1.999996852528828e-05,
      "loss": 0.7496,
      "step": 115
    },
    {
      "epoch": 0.014679827891672994,
      "grad_norm": 4.051082134246826,
      "learning_rate": 1.999996386831843e-05,
      "loss": 0.7509,
      "step": 116
    },
    {
      "epoch": 0.014806378132118452,
      "grad_norm": 2.405705690383911,
      "learning_rate": 1.999995889017905e-05,
      "loss": 0.7201,
      "step": 117
    },
    {
      "epoch": 0.014932928372563908,
      "grad_norm": 3.078101634979248,
      "learning_rate": 1.999995359087029e-05,
      "loss": 0.7167,
      "step": 118
    },
    {
      "epoch": 0.015059478613009365,
      "grad_norm": 3.6934361457824707,
      "learning_rate": 1.9999947970392325e-05,
      "loss": 0.7537,
      "step": 119
    },
    {
      "epoch": 0.015186028853454821,
      "grad_norm": 21.252214431762695,
      "learning_rate": 1.9999942028745337e-05,
      "loss": 0.7486,
      "step": 120
    },
    {
      "epoch": 0.015312579093900278,
      "grad_norm": 222.65286254882812,
      "learning_rate": 1.999993576592952e-05,
      "loss": 1.0504,
      "step": 121
    },
    {
      "epoch": 0.015439129334345734,
      "grad_norm": 6.828272819519043,
      "learning_rate": 1.9999929181945066e-05,
      "loss": 0.7515,
      "step": 122
    },
    {
      "epoch": 0.015565679574791193,
      "grad_norm": 3.362840175628662,
      "learning_rate": 1.9999922276792188e-05,
      "loss": 0.7343,
      "step": 123
    },
    {
      "epoch": 0.015692229815236648,
      "grad_norm": 3.3566877841949463,
      "learning_rate": 1.9999915050471118e-05,
      "loss": 0.7287,
      "step": 124
    },
    {
      "epoch": 0.015818780055682106,
      "grad_norm": 4.026240348815918,
      "learning_rate": 1.9999907502982076e-05,
      "loss": 0.736,
      "step": 125
    },
    {
      "epoch": 0.015945330296127564,
      "grad_norm": 2.2892649173736572,
      "learning_rate": 1.9999899634325314e-05,
      "loss": 0.7411,
      "step": 126
    },
    {
      "epoch": 0.01607188053657302,
      "grad_norm": 6.7854132652282715,
      "learning_rate": 1.9999891444501075e-05,
      "loss": 0.8091,
      "step": 127
    },
    {
      "epoch": 0.016198430777018477,
      "grad_norm": 4.323059558868408,
      "learning_rate": 1.999988293350963e-05,
      "loss": 0.7177,
      "step": 128
    },
    {
      "epoch": 0.016324981017463932,
      "grad_norm": 2.9057679176330566,
      "learning_rate": 1.9999874101351246e-05,
      "loss": 0.7143,
      "step": 129
    },
    {
      "epoch": 0.01645153125790939,
      "grad_norm": 3.0279417037963867,
      "learning_rate": 1.9999864948026217e-05,
      "loss": 0.7111,
      "step": 130
    },
    {
      "epoch": 0.01657808149835485,
      "grad_norm": 2.3679051399230957,
      "learning_rate": 1.9999855473534825e-05,
      "loss": 0.7401,
      "step": 131
    },
    {
      "epoch": 0.016704631738800303,
      "grad_norm": 3.5330636501312256,
      "learning_rate": 1.999984567787738e-05,
      "loss": 0.7278,
      "step": 132
    },
    {
      "epoch": 0.01683118197924576,
      "grad_norm": 2.417140245437622,
      "learning_rate": 1.9999835561054194e-05,
      "loss": 0.7505,
      "step": 133
    },
    {
      "epoch": 0.016957732219691216,
      "grad_norm": 8.322010040283203,
      "learning_rate": 1.9999825123065596e-05,
      "loss": 0.7541,
      "step": 134
    },
    {
      "epoch": 0.017084282460136675,
      "grad_norm": 3.951509475708008,
      "learning_rate": 1.9999814363911917e-05,
      "loss": 0.7898,
      "step": 135
    },
    {
      "epoch": 0.017210832700582133,
      "grad_norm": 6.829751491546631,
      "learning_rate": 1.9999803283593508e-05,
      "loss": 0.7261,
      "step": 136
    },
    {
      "epoch": 0.017337382941027588,
      "grad_norm": 10.25860595703125,
      "learning_rate": 1.999979188211072e-05,
      "loss": 0.7468,
      "step": 137
    },
    {
      "epoch": 0.017463933181473046,
      "grad_norm": 3.416598320007324,
      "learning_rate": 1.9999780159463923e-05,
      "loss": 0.759,
      "step": 138
    },
    {
      "epoch": 0.0175904834219185,
      "grad_norm": 3.0253524780273438,
      "learning_rate": 1.999976811565349e-05,
      "loss": 0.7304,
      "step": 139
    },
    {
      "epoch": 0.01771703366236396,
      "grad_norm": 3.8044586181640625,
      "learning_rate": 1.9999755750679807e-05,
      "loss": 0.7546,
      "step": 140
    },
    {
      "epoch": 0.017843583902809414,
      "grad_norm": 13.11278247833252,
      "learning_rate": 1.9999743064543272e-05,
      "loss": 0.7314,
      "step": 141
    },
    {
      "epoch": 0.017970134143254872,
      "grad_norm": 8.058473587036133,
      "learning_rate": 1.99997300572443e-05,
      "loss": 0.794,
      "step": 142
    },
    {
      "epoch": 0.01809668438370033,
      "grad_norm": 2.691558361053467,
      "learning_rate": 1.9999716728783297e-05,
      "loss": 0.6653,
      "step": 143
    },
    {
      "epoch": 0.018223234624145785,
      "grad_norm": 2.783745527267456,
      "learning_rate": 1.9999703079160697e-05,
      "loss": 0.7564,
      "step": 144
    },
    {
      "epoch": 0.018349784864591243,
      "grad_norm": 2.112461805343628,
      "learning_rate": 1.9999689108376938e-05,
      "loss": 0.6958,
      "step": 145
    },
    {
      "epoch": 0.018476335105036698,
      "grad_norm": 2.8818328380584717,
      "learning_rate": 1.999967481643247e-05,
      "loss": 0.7576,
      "step": 146
    },
    {
      "epoch": 0.018602885345482156,
      "grad_norm": 3.6163551807403564,
      "learning_rate": 1.999966020332775e-05,
      "loss": 0.7395,
      "step": 147
    },
    {
      "epoch": 0.018729435585927615,
      "grad_norm": 2.533327102661133,
      "learning_rate": 1.999964526906325e-05,
      "loss": 0.7295,
      "step": 148
    },
    {
      "epoch": 0.01885598582637307,
      "grad_norm": 2.2075982093811035,
      "learning_rate": 1.9999630013639442e-05,
      "loss": 0.7231,
      "step": 149
    },
    {
      "epoch": 0.018982536066818528,
      "grad_norm": 2.785245418548584,
      "learning_rate": 1.9999614437056826e-05,
      "loss": 0.7302,
      "step": 150
    },
    {
      "epoch": 0.019109086307263982,
      "grad_norm": 7.382096767425537,
      "learning_rate": 1.9999598539315895e-05,
      "loss": 0.8391,
      "step": 151
    },
    {
      "epoch": 0.01923563654770944,
      "grad_norm": 3.258136034011841,
      "learning_rate": 1.9999582320417166e-05,
      "loss": 0.6944,
      "step": 152
    },
    {
      "epoch": 0.0193621867881549,
      "grad_norm": 3.9974422454833984,
      "learning_rate": 1.9999565780361153e-05,
      "loss": 0.755,
      "step": 153
    },
    {
      "epoch": 0.019488737028600354,
      "grad_norm": 2.681788921356201,
      "learning_rate": 1.9999548919148392e-05,
      "loss": 0.7622,
      "step": 154
    },
    {
      "epoch": 0.019615287269045812,
      "grad_norm": 4.3360371589660645,
      "learning_rate": 1.9999531736779424e-05,
      "loss": 0.7399,
      "step": 155
    },
    {
      "epoch": 0.019741837509491267,
      "grad_norm": 3.9696414470672607,
      "learning_rate": 1.99995142332548e-05,
      "loss": 0.774,
      "step": 156
    },
    {
      "epoch": 0.019868387749936725,
      "grad_norm": 3.979137420654297,
      "learning_rate": 1.999949640857508e-05,
      "loss": 0.7782,
      "step": 157
    },
    {
      "epoch": 0.019994937990382183,
      "grad_norm": 7.949512004852295,
      "learning_rate": 1.9999478262740835e-05,
      "loss": 0.7409,
      "step": 158
    },
    {
      "epoch": 0.020121488230827638,
      "grad_norm": 4.83364725112915,
      "learning_rate": 1.999945979575266e-05,
      "loss": 0.7538,
      "step": 159
    },
    {
      "epoch": 0.020248038471273096,
      "grad_norm": 4.368710041046143,
      "learning_rate": 1.9999441007611132e-05,
      "loss": 0.7529,
      "step": 160
    },
    {
      "epoch": 0.02037458871171855,
      "grad_norm": 2.436755657196045,
      "learning_rate": 1.9999421898316863e-05,
      "loss": 0.7309,
      "step": 161
    },
    {
      "epoch": 0.02050113895216401,
      "grad_norm": 5.59641170501709,
      "learning_rate": 1.999940246787047e-05,
      "loss": 0.7609,
      "step": 162
    },
    {
      "epoch": 0.020627689192609464,
      "grad_norm": 4.055682182312012,
      "learning_rate": 1.999938271627257e-05,
      "loss": 0.7781,
      "step": 163
    },
    {
      "epoch": 0.020754239433054922,
      "grad_norm": 5.317326545715332,
      "learning_rate": 1.9999362643523798e-05,
      "loss": 0.7688,
      "step": 164
    },
    {
      "epoch": 0.02088078967350038,
      "grad_norm": 4.865894794464111,
      "learning_rate": 1.9999342249624806e-05,
      "loss": 0.7454,
      "step": 165
    },
    {
      "epoch": 0.021007339913945836,
      "grad_norm": 3.6462562084198,
      "learning_rate": 1.9999321534576236e-05,
      "loss": 0.7775,
      "step": 166
    },
    {
      "epoch": 0.021133890154391294,
      "grad_norm": 4.900111198425293,
      "learning_rate": 1.9999300498378767e-05,
      "loss": 0.7993,
      "step": 167
    },
    {
      "epoch": 0.02126044039483675,
      "grad_norm": 5.05611515045166,
      "learning_rate": 1.9999279141033065e-05,
      "loss": 0.7926,
      "step": 168
    },
    {
      "epoch": 0.021386990635282207,
      "grad_norm": 28.14910125732422,
      "learning_rate": 1.999925746253982e-05,
      "loss": 0.804,
      "step": 169
    },
    {
      "epoch": 0.021513540875727665,
      "grad_norm": 6.826397895812988,
      "learning_rate": 1.999923546289973e-05,
      "loss": 0.7801,
      "step": 170
    },
    {
      "epoch": 0.02164009111617312,
      "grad_norm": 4.764292240142822,
      "learning_rate": 1.9999213142113496e-05,
      "loss": 0.7484,
      "step": 171
    },
    {
      "epoch": 0.021766641356618578,
      "grad_norm": 4.07310676574707,
      "learning_rate": 1.999919050018184e-05,
      "loss": 0.7072,
      "step": 172
    },
    {
      "epoch": 0.021893191597064033,
      "grad_norm": 3.295194625854492,
      "learning_rate": 1.9999167537105492e-05,
      "loss": 0.7815,
      "step": 173
    },
    {
      "epoch": 0.02201974183750949,
      "grad_norm": 4.278547763824463,
      "learning_rate": 1.999914425288518e-05,
      "loss": 0.7899,
      "step": 174
    },
    {
      "epoch": 0.02214629207795495,
      "grad_norm": 3.4261250495910645,
      "learning_rate": 1.9999120647521653e-05,
      "loss": 0.7137,
      "step": 175
    },
    {
      "epoch": 0.022272842318400404,
      "grad_norm": 3.7300071716308594,
      "learning_rate": 1.9999096721015677e-05,
      "loss": 0.7533,
      "step": 176
    },
    {
      "epoch": 0.022399392558845863,
      "grad_norm": 2.1310386657714844,
      "learning_rate": 1.9999072473368018e-05,
      "loss": 0.7336,
      "step": 177
    },
    {
      "epoch": 0.022525942799291317,
      "grad_norm": 3.6934549808502197,
      "learning_rate": 1.999904790457945e-05,
      "loss": 0.6966,
      "step": 178
    },
    {
      "epoch": 0.022652493039736776,
      "grad_norm": 5.027427673339844,
      "learning_rate": 1.9999023014650765e-05,
      "loss": 0.6876,
      "step": 179
    },
    {
      "epoch": 0.022779043280182234,
      "grad_norm": 3.2989590167999268,
      "learning_rate": 1.9998997803582764e-05,
      "loss": 0.7285,
      "step": 180
    },
    {
      "epoch": 0.02290559352062769,
      "grad_norm": 12.07487678527832,
      "learning_rate": 1.9998972271376257e-05,
      "loss": 0.7226,
      "step": 181
    },
    {
      "epoch": 0.023032143761073147,
      "grad_norm": 4.070852756500244,
      "learning_rate": 1.9998946418032062e-05,
      "loss": 0.7294,
      "step": 182
    },
    {
      "epoch": 0.0231586940015186,
      "grad_norm": 2.3300113677978516,
      "learning_rate": 1.9998920243551006e-05,
      "loss": 0.7685,
      "step": 183
    },
    {
      "epoch": 0.02328524424196406,
      "grad_norm": 3.874284029006958,
      "learning_rate": 1.9998893747933935e-05,
      "loss": 0.7704,
      "step": 184
    },
    {
      "epoch": 0.023411794482409518,
      "grad_norm": 2.4091031551361084,
      "learning_rate": 1.9998866931181696e-05,
      "loss": 0.741,
      "step": 185
    },
    {
      "epoch": 0.023538344722854973,
      "grad_norm": 2.836761474609375,
      "learning_rate": 1.9998839793295157e-05,
      "loss": 0.7223,
      "step": 186
    },
    {
      "epoch": 0.02366489496330043,
      "grad_norm": 3.4797942638397217,
      "learning_rate": 1.999881233427518e-05,
      "loss": 0.7444,
      "step": 187
    },
    {
      "epoch": 0.023791445203745886,
      "grad_norm": 2.600398302078247,
      "learning_rate": 1.9998784554122658e-05,
      "loss": 0.7731,
      "step": 188
    },
    {
      "epoch": 0.023917995444191344,
      "grad_norm": 3.92728328704834,
      "learning_rate": 1.9998756452838473e-05,
      "loss": 0.7079,
      "step": 189
    },
    {
      "epoch": 0.0240445456846368,
      "grad_norm": 3.2434329986572266,
      "learning_rate": 1.999872803042353e-05,
      "loss": 0.6872,
      "step": 190
    },
    {
      "epoch": 0.024171095925082257,
      "grad_norm": 3.332825183868408,
      "learning_rate": 1.999869928687875e-05,
      "loss": 0.7472,
      "step": 191
    },
    {
      "epoch": 0.024297646165527716,
      "grad_norm": 2.1622445583343506,
      "learning_rate": 1.9998670222205043e-05,
      "loss": 0.6923,
      "step": 192
    },
    {
      "epoch": 0.02442419640597317,
      "grad_norm": 4.413276195526123,
      "learning_rate": 1.9998640836403353e-05,
      "loss": 0.7034,
      "step": 193
    },
    {
      "epoch": 0.02455074664641863,
      "grad_norm": 3.1784801483154297,
      "learning_rate": 1.999861112947462e-05,
      "loss": 0.779,
      "step": 194
    },
    {
      "epoch": 0.024677296886864084,
      "grad_norm": 4.370777606964111,
      "learning_rate": 1.9998581101419798e-05,
      "loss": 0.7561,
      "step": 195
    },
    {
      "epoch": 0.024803847127309542,
      "grad_norm": 3.040710210800171,
      "learning_rate": 1.999855075223985e-05,
      "loss": 0.7362,
      "step": 196
    },
    {
      "epoch": 0.024930397367755,
      "grad_norm": 3.4362075328826904,
      "learning_rate": 1.9998520081935753e-05,
      "loss": 0.7204,
      "step": 197
    },
    {
      "epoch": 0.025056947608200455,
      "grad_norm": 2.894932746887207,
      "learning_rate": 1.999848909050849e-05,
      "loss": 0.7364,
      "step": 198
    },
    {
      "epoch": 0.025183497848645913,
      "grad_norm": 4.424647808074951,
      "learning_rate": 1.9998457777959058e-05,
      "loss": 0.7402,
      "step": 199
    },
    {
      "epoch": 0.025310048089091368,
      "grad_norm": 3.429141044616699,
      "learning_rate": 1.9998426144288464e-05,
      "loss": 0.7258,
      "step": 200
    },
    {
      "epoch": 0.025436598329536826,
      "grad_norm": 3.7962710857391357,
      "learning_rate": 1.999839418949772e-05,
      "loss": 0.7745,
      "step": 201
    },
    {
      "epoch": 0.025563148569982284,
      "grad_norm": 3.2279765605926514,
      "learning_rate": 1.9998361913587854e-05,
      "loss": 0.7044,
      "step": 202
    },
    {
      "epoch": 0.02568969881042774,
      "grad_norm": 2.6983156204223633,
      "learning_rate": 1.999832931655991e-05,
      "loss": 0.7766,
      "step": 203
    },
    {
      "epoch": 0.025816249050873197,
      "grad_norm": 3.016213893890381,
      "learning_rate": 1.999829639841492e-05,
      "loss": 0.7569,
      "step": 204
    },
    {
      "epoch": 0.025942799291318652,
      "grad_norm": 2.720350742340088,
      "learning_rate": 1.9998263159153954e-05,
      "loss": 0.7922,
      "step": 205
    },
    {
      "epoch": 0.02606934953176411,
      "grad_norm": 2.494453191757202,
      "learning_rate": 1.999822959877807e-05,
      "loss": 0.7425,
      "step": 206
    },
    {
      "epoch": 0.02619589977220957,
      "grad_norm": 2.9743306636810303,
      "learning_rate": 1.9998195717288354e-05,
      "loss": 0.6963,
      "step": 207
    },
    {
      "epoch": 0.026322450012655024,
      "grad_norm": 2.4955286979675293,
      "learning_rate": 1.999816151468589e-05,
      "loss": 0.7019,
      "step": 208
    },
    {
      "epoch": 0.026449000253100482,
      "grad_norm": 3.312103748321533,
      "learning_rate": 1.999812699097178e-05,
      "loss": 0.7086,
      "step": 209
    },
    {
      "epoch": 0.026575550493545937,
      "grad_norm": 2.874286651611328,
      "learning_rate": 1.9998092146147128e-05,
      "loss": 0.735,
      "step": 210
    },
    {
      "epoch": 0.026702100733991395,
      "grad_norm": 3.22044038772583,
      "learning_rate": 1.9998056980213056e-05,
      "loss": 0.7901,
      "step": 211
    },
    {
      "epoch": 0.026828650974436853,
      "grad_norm": 2.4791576862335205,
      "learning_rate": 1.9998021493170688e-05,
      "loss": 0.7228,
      "step": 212
    },
    {
      "epoch": 0.026955201214882308,
      "grad_norm": 3.1535239219665527,
      "learning_rate": 1.9997985685021175e-05,
      "loss": 0.6958,
      "step": 213
    },
    {
      "epoch": 0.027081751455327766,
      "grad_norm": 2.962524890899658,
      "learning_rate": 1.999794955576566e-05,
      "loss": 0.7338,
      "step": 214
    },
    {
      "epoch": 0.02720830169577322,
      "grad_norm": 2.8971126079559326,
      "learning_rate": 1.99979131054053e-05,
      "loss": 0.7276,
      "step": 215
    },
    {
      "epoch": 0.02733485193621868,
      "grad_norm": 2.4257972240448,
      "learning_rate": 1.999787633394127e-05,
      "loss": 0.7648,
      "step": 216
    },
    {
      "epoch": 0.027461402176664134,
      "grad_norm": 2.8326356410980225,
      "learning_rate": 1.9997839241374754e-05,
      "loss": 0.7431,
      "step": 217
    },
    {
      "epoch": 0.027587952417109592,
      "grad_norm": 2.0882227420806885,
      "learning_rate": 1.9997801827706938e-05,
      "loss": 0.7294,
      "step": 218
    },
    {
      "epoch": 0.02771450265755505,
      "grad_norm": 2.8033554553985596,
      "learning_rate": 1.9997764092939025e-05,
      "loss": 0.7437,
      "step": 219
    },
    {
      "epoch": 0.027841052898000505,
      "grad_norm": 2.0529937744140625,
      "learning_rate": 1.9997726037072228e-05,
      "loss": 0.7734,
      "step": 220
    },
    {
      "epoch": 0.027967603138445964,
      "grad_norm": 2.444610118865967,
      "learning_rate": 1.9997687660107766e-05,
      "loss": 0.6743,
      "step": 221
    },
    {
      "epoch": 0.02809415337889142,
      "grad_norm": 2.213343620300293,
      "learning_rate": 1.999764896204688e-05,
      "loss": 0.7623,
      "step": 222
    },
    {
      "epoch": 0.028220703619336877,
      "grad_norm": 2.073878049850464,
      "learning_rate": 1.9997609942890803e-05,
      "loss": 0.6845,
      "step": 223
    },
    {
      "epoch": 0.028347253859782335,
      "grad_norm": 1.9515246152877808,
      "learning_rate": 1.9997570602640792e-05,
      "loss": 0.7144,
      "step": 224
    },
    {
      "epoch": 0.02847380410022779,
      "grad_norm": 1.8842471837997437,
      "learning_rate": 1.9997530941298114e-05,
      "loss": 0.671,
      "step": 225
    },
    {
      "epoch": 0.028600354340673248,
      "grad_norm": 2.285245895385742,
      "learning_rate": 1.9997490958864033e-05,
      "loss": 0.773,
      "step": 226
    },
    {
      "epoch": 0.028726904581118703,
      "grad_norm": 2.2714853286743164,
      "learning_rate": 1.9997450655339846e-05,
      "loss": 0.7746,
      "step": 227
    },
    {
      "epoch": 0.02885345482156416,
      "grad_norm": 1.9219987392425537,
      "learning_rate": 1.999741003072684e-05,
      "loss": 0.7757,
      "step": 228
    },
    {
      "epoch": 0.02898000506200962,
      "grad_norm": 1.900895357131958,
      "learning_rate": 1.999736908502632e-05,
      "loss": 0.6832,
      "step": 229
    },
    {
      "epoch": 0.029106555302455074,
      "grad_norm": 2.2520174980163574,
      "learning_rate": 1.99973278182396e-05,
      "loss": 0.7307,
      "step": 230
    },
    {
      "epoch": 0.029233105542900532,
      "grad_norm": 2.078585624694824,
      "learning_rate": 1.999728623036801e-05,
      "loss": 0.6988,
      "step": 231
    },
    {
      "epoch": 0.029359655783345987,
      "grad_norm": 1.8943878412246704,
      "learning_rate": 1.9997244321412882e-05,
      "loss": 0.7578,
      "step": 232
    },
    {
      "epoch": 0.029486206023791445,
      "grad_norm": 2.0495598316192627,
      "learning_rate": 1.9997202091375564e-05,
      "loss": 0.7262,
      "step": 233
    },
    {
      "epoch": 0.029612756264236904,
      "grad_norm": 2.0213913917541504,
      "learning_rate": 1.999715954025741e-05,
      "loss": 0.7252,
      "step": 234
    },
    {
      "epoch": 0.02973930650468236,
      "grad_norm": 2.074827194213867,
      "learning_rate": 1.9997116668059787e-05,
      "loss": 0.7614,
      "step": 235
    },
    {
      "epoch": 0.029865856745127817,
      "grad_norm": 2.16278076171875,
      "learning_rate": 1.9997073474784073e-05,
      "loss": 0.7825,
      "step": 236
    },
    {
      "epoch": 0.02999240698557327,
      "grad_norm": 1.8708511590957642,
      "learning_rate": 1.9997029960431656e-05,
      "loss": 0.6799,
      "step": 237
    },
    {
      "epoch": 0.03011895722601873,
      "grad_norm": 2.147273302078247,
      "learning_rate": 1.9996986125003933e-05,
      "loss": 0.7362,
      "step": 238
    },
    {
      "epoch": 0.030245507466464185,
      "grad_norm": 1.8549391031265259,
      "learning_rate": 1.9996941968502307e-05,
      "loss": 0.7581,
      "step": 239
    },
    {
      "epoch": 0.030372057706909643,
      "grad_norm": 2.0326991081237793,
      "learning_rate": 1.9996897490928207e-05,
      "loss": 0.7528,
      "step": 240
    },
    {
      "epoch": 0.0304986079473551,
      "grad_norm": 2.019186496734619,
      "learning_rate": 1.999685269228305e-05,
      "loss": 0.7785,
      "step": 241
    },
    {
      "epoch": 0.030625158187800556,
      "grad_norm": 1.9348907470703125,
      "learning_rate": 1.9996807572568283e-05,
      "loss": 0.7944,
      "step": 242
    },
    {
      "epoch": 0.030751708428246014,
      "grad_norm": 2.0608372688293457,
      "learning_rate": 1.999676213178535e-05,
      "loss": 0.7612,
      "step": 243
    },
    {
      "epoch": 0.03087825866869147,
      "grad_norm": 1.990561604499817,
      "learning_rate": 1.9996716369935715e-05,
      "loss": 0.7787,
      "step": 244
    },
    {
      "epoch": 0.031004808909136927,
      "grad_norm": 2.054804801940918,
      "learning_rate": 1.9996670287020844e-05,
      "loss": 0.8107,
      "step": 245
    },
    {
      "epoch": 0.031131359149582385,
      "grad_norm": 1.9413708448410034,
      "learning_rate": 1.999662388304222e-05,
      "loss": 0.7441,
      "step": 246
    },
    {
      "epoch": 0.031257909390027844,
      "grad_norm": 2.116816520690918,
      "learning_rate": 1.999657715800133e-05,
      "loss": 0.7927,
      "step": 247
    },
    {
      "epoch": 0.031384459630473295,
      "grad_norm": 2.0755362510681152,
      "learning_rate": 1.9996530111899676e-05,
      "loss": 0.7747,
      "step": 248
    },
    {
      "epoch": 0.03151100987091875,
      "grad_norm": 2.250007390975952,
      "learning_rate": 1.999648274473877e-05,
      "loss": 0.8186,
      "step": 249
    },
    {
      "epoch": 0.03163756011136421,
      "grad_norm": 1.947838306427002,
      "learning_rate": 1.9996435056520133e-05,
      "loss": 0.771,
      "step": 250
    },
    {
      "epoch": 0.03176411035180967,
      "grad_norm": 2.1759774684906006,
      "learning_rate": 1.999638704724529e-05,
      "loss": 0.7954,
      "step": 251
    },
    {
      "epoch": 0.03189066059225513,
      "grad_norm": 2.377180814743042,
      "learning_rate": 1.99963387169158e-05,
      "loss": 0.7572,
      "step": 252
    },
    {
      "epoch": 0.03201721083270058,
      "grad_norm": 2.0934760570526123,
      "learning_rate": 1.9996290065533194e-05,
      "loss": 0.7585,
      "step": 253
    },
    {
      "epoch": 0.03214376107314604,
      "grad_norm": 1.848320484161377,
      "learning_rate": 1.999624109309905e-05,
      "loss": 0.7658,
      "step": 254
    },
    {
      "epoch": 0.032270311313591496,
      "grad_norm": 1.933855414390564,
      "learning_rate": 1.9996191799614934e-05,
      "loss": 0.7638,
      "step": 255
    },
    {
      "epoch": 0.032396861554036954,
      "grad_norm": 1.8370720148086548,
      "learning_rate": 1.9996142185082433e-05,
      "loss": 0.7646,
      "step": 256
    },
    {
      "epoch": 0.03252341179448241,
      "grad_norm": 2.232774257659912,
      "learning_rate": 1.9996092249503138e-05,
      "loss": 0.7572,
      "step": 257
    },
    {
      "epoch": 0.032649962034927864,
      "grad_norm": 2.464170455932617,
      "learning_rate": 1.9996041992878653e-05,
      "loss": 0.7957,
      "step": 258
    },
    {
      "epoch": 0.03277651227537332,
      "grad_norm": 1.7014336585998535,
      "learning_rate": 1.999599141521059e-05,
      "loss": 0.7305,
      "step": 259
    },
    {
      "epoch": 0.03290306251581878,
      "grad_norm": 1.7367585897445679,
      "learning_rate": 1.9995940516500575e-05,
      "loss": 0.7536,
      "step": 260
    },
    {
      "epoch": 0.03302961275626424,
      "grad_norm": 1.7794029712677002,
      "learning_rate": 1.9995889296750247e-05,
      "loss": 0.6931,
      "step": 261
    },
    {
      "epoch": 0.0331561629967097,
      "grad_norm": 1.8764920234680176,
      "learning_rate": 1.9995837755961246e-05,
      "loss": 0.7603,
      "step": 262
    },
    {
      "epoch": 0.03328271323715515,
      "grad_norm": 2.0262672901153564,
      "learning_rate": 1.9995785894135228e-05,
      "loss": 0.834,
      "step": 263
    },
    {
      "epoch": 0.033409263477600606,
      "grad_norm": 2.5078697204589844,
      "learning_rate": 1.999573371127386e-05,
      "loss": 0.7747,
      "step": 264
    },
    {
      "epoch": 0.033535813718046065,
      "grad_norm": 2.3635966777801514,
      "learning_rate": 1.9995681207378816e-05,
      "loss": 0.7554,
      "step": 265
    },
    {
      "epoch": 0.03366236395849152,
      "grad_norm": 1.6598420143127441,
      "learning_rate": 1.9995628382451783e-05,
      "loss": 0.7558,
      "step": 266
    },
    {
      "epoch": 0.03378891419893698,
      "grad_norm": 2.2354466915130615,
      "learning_rate": 1.999557523649446e-05,
      "loss": 0.7936,
      "step": 267
    },
    {
      "epoch": 0.03391546443938243,
      "grad_norm": 2.0556702613830566,
      "learning_rate": 1.999552176950855e-05,
      "loss": 0.724,
      "step": 268
    },
    {
      "epoch": 0.03404201467982789,
      "grad_norm": 2.067535638809204,
      "learning_rate": 1.9995467981495777e-05,
      "loss": 0.7813,
      "step": 269
    },
    {
      "epoch": 0.03416856492027335,
      "grad_norm": 2.232144832611084,
      "learning_rate": 1.999541387245786e-05,
      "loss": 0.7432,
      "step": 270
    },
    {
      "epoch": 0.03429511516071881,
      "grad_norm": 2.1276612281799316,
      "learning_rate": 1.9995359442396543e-05,
      "loss": 0.8142,
      "step": 271
    },
    {
      "epoch": 0.034421665401164266,
      "grad_norm": 1.889432668685913,
      "learning_rate": 1.999530469131357e-05,
      "loss": 0.7713,
      "step": 272
    },
    {
      "epoch": 0.03454821564160972,
      "grad_norm": 1.9064579010009766,
      "learning_rate": 1.9995249619210702e-05,
      "loss": 0.7733,
      "step": 273
    },
    {
      "epoch": 0.034674765882055175,
      "grad_norm": 2.1462016105651855,
      "learning_rate": 1.9995194226089704e-05,
      "loss": 0.7324,
      "step": 274
    },
    {
      "epoch": 0.03480131612250063,
      "grad_norm": 2.419970989227295,
      "learning_rate": 1.999513851195236e-05,
      "loss": 0.7815,
      "step": 275
    },
    {
      "epoch": 0.03492786636294609,
      "grad_norm": 1.8160194158554077,
      "learning_rate": 1.999508247680046e-05,
      "loss": 0.7527,
      "step": 276
    },
    {
      "epoch": 0.03505441660339154,
      "grad_norm": 3.6053144931793213,
      "learning_rate": 1.9995026120635796e-05,
      "loss": 0.794,
      "step": 277
    },
    {
      "epoch": 0.035180966843837,
      "grad_norm": 3.404911756515503,
      "learning_rate": 1.9994969443460186e-05,
      "loss": 0.7704,
      "step": 278
    },
    {
      "epoch": 0.03530751708428246,
      "grad_norm": 3.8246371746063232,
      "learning_rate": 1.9994912445275446e-05,
      "loss": 0.8092,
      "step": 279
    },
    {
      "epoch": 0.03543406732472792,
      "grad_norm": 3.4314613342285156,
      "learning_rate": 1.999485512608341e-05,
      "loss": 0.8368,
      "step": 280
    },
    {
      "epoch": 0.035560617565173376,
      "grad_norm": 2.902418613433838,
      "learning_rate": 1.9994797485885917e-05,
      "loss": 0.7448,
      "step": 281
    },
    {
      "epoch": 0.03568716780561883,
      "grad_norm": 1.986924648284912,
      "learning_rate": 1.9994739524684815e-05,
      "loss": 0.7501,
      "step": 282
    },
    {
      "epoch": 0.035813718046064286,
      "grad_norm": 2.040584087371826,
      "learning_rate": 1.999468124248197e-05,
      "loss": 0.7835,
      "step": 283
    },
    {
      "epoch": 0.035940268286509744,
      "grad_norm": 2.250995397567749,
      "learning_rate": 1.9994622639279252e-05,
      "loss": 0.7113,
      "step": 284
    },
    {
      "epoch": 0.0360668185269552,
      "grad_norm": 1.9429740905761719,
      "learning_rate": 1.999456371507855e-05,
      "loss": 0.8311,
      "step": 285
    },
    {
      "epoch": 0.03619336876740066,
      "grad_norm": 2.109989881515503,
      "learning_rate": 1.999450446988174e-05,
      "loss": 0.8076,
      "step": 286
    },
    {
      "epoch": 0.03631991900784611,
      "grad_norm": 2.342918872833252,
      "learning_rate": 1.999444490369074e-05,
      "loss": 0.7094,
      "step": 287
    },
    {
      "epoch": 0.03644646924829157,
      "grad_norm": 1.9303691387176514,
      "learning_rate": 1.999438501650746e-05,
      "loss": 0.7104,
      "step": 288
    },
    {
      "epoch": 0.03657301948873703,
      "grad_norm": 2.112776756286621,
      "learning_rate": 1.999432480833382e-05,
      "loss": 0.7684,
      "step": 289
    },
    {
      "epoch": 0.03669956972918249,
      "grad_norm": 1.8414201736450195,
      "learning_rate": 1.9994264279171754e-05,
      "loss": 0.7321,
      "step": 290
    },
    {
      "epoch": 0.036826119969627945,
      "grad_norm": 2.161665678024292,
      "learning_rate": 1.9994203429023208e-05,
      "loss": 0.7754,
      "step": 291
    },
    {
      "epoch": 0.036952670210073396,
      "grad_norm": 1.8337818384170532,
      "learning_rate": 1.9994142257890135e-05,
      "loss": 0.7158,
      "step": 292
    },
    {
      "epoch": 0.037079220450518854,
      "grad_norm": 1.9573699235916138,
      "learning_rate": 1.99940807657745e-05,
      "loss": 0.7392,
      "step": 293
    },
    {
      "epoch": 0.03720577069096431,
      "grad_norm": 2.105797529220581,
      "learning_rate": 1.999401895267828e-05,
      "loss": 0.7438,
      "step": 294
    },
    {
      "epoch": 0.03733232093140977,
      "grad_norm": 1.8444557189941406,
      "learning_rate": 1.9993956818603454e-05,
      "loss": 0.7815,
      "step": 295
    },
    {
      "epoch": 0.03745887117185523,
      "grad_norm": 1.8910044431686401,
      "learning_rate": 1.9993894363552025e-05,
      "loss": 0.7619,
      "step": 296
    },
    {
      "epoch": 0.03758542141230068,
      "grad_norm": 2.292133331298828,
      "learning_rate": 1.9993831587525996e-05,
      "loss": 0.7497,
      "step": 297
    },
    {
      "epoch": 0.03771197165274614,
      "grad_norm": 1.9697457551956177,
      "learning_rate": 1.9993768490527382e-05,
      "loss": 0.7478,
      "step": 298
    },
    {
      "epoch": 0.0378385218931916,
      "grad_norm": 2.1313259601593018,
      "learning_rate": 1.999370507255821e-05,
      "loss": 0.7124,
      "step": 299
    },
    {
      "epoch": 0.037965072133637055,
      "grad_norm": 2.06640362739563,
      "learning_rate": 1.9993641333620518e-05,
      "loss": 0.727,
      "step": 300
    },
    {
      "epoch": 0.038091622374082514,
      "grad_norm": 1.8245208263397217,
      "learning_rate": 1.999357727371635e-05,
      "loss": 0.7469,
      "step": 301
    },
    {
      "epoch": 0.038218172614527965,
      "grad_norm": 2.583322286605835,
      "learning_rate": 1.999351289284777e-05,
      "loss": 0.7938,
      "step": 302
    },
    {
      "epoch": 0.03834472285497342,
      "grad_norm": 1.9097955226898193,
      "learning_rate": 1.9993448191016838e-05,
      "loss": 0.7396,
      "step": 303
    },
    {
      "epoch": 0.03847127309541888,
      "grad_norm": 1.6975648403167725,
      "learning_rate": 1.9993383168225637e-05,
      "loss": 0.7325,
      "step": 304
    },
    {
      "epoch": 0.03859782333586434,
      "grad_norm": 1.9074190855026245,
      "learning_rate": 1.9993317824476254e-05,
      "loss": 0.8349,
      "step": 305
    },
    {
      "epoch": 0.0387243735763098,
      "grad_norm": 1.9295637607574463,
      "learning_rate": 1.999325215977079e-05,
      "loss": 0.7393,
      "step": 306
    },
    {
      "epoch": 0.03885092381675525,
      "grad_norm": 1.7606990337371826,
      "learning_rate": 1.9993186174111348e-05,
      "loss": 0.767,
      "step": 307
    },
    {
      "epoch": 0.03897747405720071,
      "grad_norm": 1.7623753547668457,
      "learning_rate": 1.999311986750005e-05,
      "loss": 0.7442,
      "step": 308
    },
    {
      "epoch": 0.039104024297646166,
      "grad_norm": 1.989001989364624,
      "learning_rate": 1.9993053239939028e-05,
      "loss": 0.8056,
      "step": 309
    },
    {
      "epoch": 0.039230574538091624,
      "grad_norm": 2.2212886810302734,
      "learning_rate": 1.999298629143042e-05,
      "loss": 0.7787,
      "step": 310
    },
    {
      "epoch": 0.03935712477853708,
      "grad_norm": 2.1541669368743896,
      "learning_rate": 1.9992919021976376e-05,
      "loss": 0.7775,
      "step": 311
    },
    {
      "epoch": 0.039483675018982534,
      "grad_norm": 1.7051433324813843,
      "learning_rate": 1.9992851431579057e-05,
      "loss": 0.7756,
      "step": 312
    },
    {
      "epoch": 0.03961022525942799,
      "grad_norm": 1.9671646356582642,
      "learning_rate": 1.9992783520240636e-05,
      "loss": 0.7917,
      "step": 313
    },
    {
      "epoch": 0.03973677549987345,
      "grad_norm": 1.7287921905517578,
      "learning_rate": 1.9992715287963287e-05,
      "loss": 0.7663,
      "step": 314
    },
    {
      "epoch": 0.03986332574031891,
      "grad_norm": 1.9209833145141602,
      "learning_rate": 1.999264673474921e-05,
      "loss": 0.7617,
      "step": 315
    },
    {
      "epoch": 0.03998987598076437,
      "grad_norm": 1.9496666193008423,
      "learning_rate": 1.9992577860600605e-05,
      "loss": 0.7922,
      "step": 316
    },
    {
      "epoch": 0.04011642622120982,
      "grad_norm": 2.0900402069091797,
      "learning_rate": 1.9992508665519676e-05,
      "loss": 0.8089,
      "step": 317
    },
    {
      "epoch": 0.040242976461655276,
      "grad_norm": 2.588820219039917,
      "learning_rate": 1.9992439149508652e-05,
      "loss": 0.8393,
      "step": 318
    },
    {
      "epoch": 0.040369526702100735,
      "grad_norm": 1.9903364181518555,
      "learning_rate": 1.9992369312569768e-05,
      "loss": 0.7707,
      "step": 319
    },
    {
      "epoch": 0.04049607694254619,
      "grad_norm": 1.8376017808914185,
      "learning_rate": 1.9992299154705264e-05,
      "loss": 0.7888,
      "step": 320
    },
    {
      "epoch": 0.04062262718299165,
      "grad_norm": 1.8400070667266846,
      "learning_rate": 1.9992228675917392e-05,
      "loss": 0.7803,
      "step": 321
    },
    {
      "epoch": 0.0407491774234371,
      "grad_norm": 2.0549070835113525,
      "learning_rate": 1.9992157876208418e-05,
      "loss": 0.7558,
      "step": 322
    },
    {
      "epoch": 0.04087572766388256,
      "grad_norm": 1.8893535137176514,
      "learning_rate": 1.9992086755580616e-05,
      "loss": 0.7009,
      "step": 323
    },
    {
      "epoch": 0.04100227790432802,
      "grad_norm": 1.9745639562606812,
      "learning_rate": 1.9992015314036264e-05,
      "loss": 0.7654,
      "step": 324
    },
    {
      "epoch": 0.04112882814477348,
      "grad_norm": 2.042276382446289,
      "learning_rate": 1.9991943551577666e-05,
      "loss": 0.7556,
      "step": 325
    },
    {
      "epoch": 0.04125537838521893,
      "grad_norm": 2.136768341064453,
      "learning_rate": 1.999187146820712e-05,
      "loss": 0.765,
      "step": 326
    },
    {
      "epoch": 0.04138192862566439,
      "grad_norm": 2.7210731506347656,
      "learning_rate": 1.9991799063926943e-05,
      "loss": 0.7783,
      "step": 327
    },
    {
      "epoch": 0.041508478866109845,
      "grad_norm": 1.8116244077682495,
      "learning_rate": 1.9991726338739458e-05,
      "loss": 0.7637,
      "step": 328
    },
    {
      "epoch": 0.0416350291065553,
      "grad_norm": 3.0680904388427734,
      "learning_rate": 1.9991653292647005e-05,
      "loss": 0.7892,
      "step": 329
    },
    {
      "epoch": 0.04176157934700076,
      "grad_norm": 2.0143771171569824,
      "learning_rate": 1.9991579925651933e-05,
      "loss": 0.775,
      "step": 330
    },
    {
      "epoch": 0.04188812958744621,
      "grad_norm": 1.8708196878433228,
      "learning_rate": 1.999150623775659e-05,
      "loss": 0.7464,
      "step": 331
    },
    {
      "epoch": 0.04201467982789167,
      "grad_norm": 2.6441519260406494,
      "learning_rate": 1.9991432228963347e-05,
      "loss": 0.7321,
      "step": 332
    },
    {
      "epoch": 0.04214123006833713,
      "grad_norm": 1.7815757989883423,
      "learning_rate": 1.9991357899274577e-05,
      "loss": 0.7335,
      "step": 333
    },
    {
      "epoch": 0.04226778030878259,
      "grad_norm": 2.107880115509033,
      "learning_rate": 1.9991283248692674e-05,
      "loss": 0.7445,
      "step": 334
    },
    {
      "epoch": 0.042394330549228046,
      "grad_norm": 2.083987236022949,
      "learning_rate": 1.9991208277220034e-05,
      "loss": 0.7626,
      "step": 335
    },
    {
      "epoch": 0.0425208807896735,
      "grad_norm": 2.1020870208740234,
      "learning_rate": 1.999113298485906e-05,
      "loss": 0.7377,
      "step": 336
    },
    {
      "epoch": 0.042647431030118955,
      "grad_norm": 2.028149366378784,
      "learning_rate": 1.9991057371612173e-05,
      "loss": 0.7858,
      "step": 337
    },
    {
      "epoch": 0.042773981270564414,
      "grad_norm": 1.6566681861877441,
      "learning_rate": 1.99909814374818e-05,
      "loss": 0.7335,
      "step": 338
    },
    {
      "epoch": 0.04290053151100987,
      "grad_norm": 2.0452022552490234,
      "learning_rate": 1.9990905182470384e-05,
      "loss": 0.7979,
      "step": 339
    },
    {
      "epoch": 0.04302708175145533,
      "grad_norm": 2.492017984390259,
      "learning_rate": 1.9990828606580373e-05,
      "loss": 0.7773,
      "step": 340
    },
    {
      "epoch": 0.04315363199190078,
      "grad_norm": 2.086411952972412,
      "learning_rate": 1.9990751709814224e-05,
      "loss": 0.7369,
      "step": 341
    },
    {
      "epoch": 0.04328018223234624,
      "grad_norm": 2.125770330429077,
      "learning_rate": 1.9990674492174407e-05,
      "loss": 0.8038,
      "step": 342
    },
    {
      "epoch": 0.0434067324727917,
      "grad_norm": 2.240492582321167,
      "learning_rate": 1.99905969536634e-05,
      "loss": 0.7855,
      "step": 343
    },
    {
      "epoch": 0.043533282713237156,
      "grad_norm": 1.9689252376556396,
      "learning_rate": 1.99905190942837e-05,
      "loss": 0.7683,
      "step": 344
    },
    {
      "epoch": 0.043659832953682615,
      "grad_norm": 1.806531310081482,
      "learning_rate": 1.9990440914037804e-05,
      "loss": 0.7409,
      "step": 345
    },
    {
      "epoch": 0.043786383194128066,
      "grad_norm": 2.3503682613372803,
      "learning_rate": 1.999036241292822e-05,
      "loss": 0.7619,
      "step": 346
    },
    {
      "epoch": 0.043912933434573524,
      "grad_norm": 2.55704665184021,
      "learning_rate": 1.999028359095747e-05,
      "loss": 0.7214,
      "step": 347
    },
    {
      "epoch": 0.04403948367501898,
      "grad_norm": 2.238823652267456,
      "learning_rate": 1.999020444812809e-05,
      "loss": 0.7072,
      "step": 348
    },
    {
      "epoch": 0.04416603391546444,
      "grad_norm": 3.4982070922851562,
      "learning_rate": 1.999012498444262e-05,
      "loss": 0.7691,
      "step": 349
    },
    {
      "epoch": 0.0442925841559099,
      "grad_norm": 2.826392650604248,
      "learning_rate": 1.9990045199903607e-05,
      "loss": 0.8047,
      "step": 350
    },
    {
      "epoch": 0.04441913439635535,
      "grad_norm": 3.0347354412078857,
      "learning_rate": 1.998996509451362e-05,
      "loss": 0.7575,
      "step": 351
    },
    {
      "epoch": 0.04454568463680081,
      "grad_norm": 2.335083246231079,
      "learning_rate": 1.998988466827523e-05,
      "loss": 0.7912,
      "step": 352
    },
    {
      "epoch": 0.04467223487724627,
      "grad_norm": 3.9997000694274902,
      "learning_rate": 1.998980392119102e-05,
      "loss": 0.781,
      "step": 353
    },
    {
      "epoch": 0.044798785117691725,
      "grad_norm": 2.6936283111572266,
      "learning_rate": 1.998972285326358e-05,
      "loss": 0.7538,
      "step": 354
    },
    {
      "epoch": 0.04492533535813718,
      "grad_norm": 3.65978741645813,
      "learning_rate": 1.9989641464495516e-05,
      "loss": 0.7612,
      "step": 355
    },
    {
      "epoch": 0.045051885598582635,
      "grad_norm": 2.7851903438568115,
      "learning_rate": 1.9989559754889448e-05,
      "loss": 0.7579,
      "step": 356
    },
    {
      "epoch": 0.04517843583902809,
      "grad_norm": 3.1100780963897705,
      "learning_rate": 1.998947772444799e-05,
      "loss": 0.7725,
      "step": 357
    },
    {
      "epoch": 0.04530498607947355,
      "grad_norm": 2.899186611175537,
      "learning_rate": 1.9989395373173784e-05,
      "loss": 0.7519,
      "step": 358
    },
    {
      "epoch": 0.04543153631991901,
      "grad_norm": 2.5418968200683594,
      "learning_rate": 1.998931270106947e-05,
      "loss": 0.7418,
      "step": 359
    },
    {
      "epoch": 0.04555808656036447,
      "grad_norm": 2.5076675415039062,
      "learning_rate": 1.9989229708137702e-05,
      "loss": 0.7535,
      "step": 360
    },
    {
      "epoch": 0.04568463680080992,
      "grad_norm": 2.117948055267334,
      "learning_rate": 1.9989146394381156e-05,
      "loss": 0.7095,
      "step": 361
    },
    {
      "epoch": 0.04581118704125538,
      "grad_norm": 2.429708242416382,
      "learning_rate": 1.9989062759802497e-05,
      "loss": 0.7739,
      "step": 362
    },
    {
      "epoch": 0.045937737281700836,
      "grad_norm": 2.624659776687622,
      "learning_rate": 1.9988978804404416e-05,
      "loss": 0.7428,
      "step": 363
    },
    {
      "epoch": 0.046064287522146294,
      "grad_norm": 2.0526256561279297,
      "learning_rate": 1.9988894528189605e-05,
      "loss": 0.76,
      "step": 364
    },
    {
      "epoch": 0.04619083776259175,
      "grad_norm": 2.656740665435791,
      "learning_rate": 1.9988809931160777e-05,
      "loss": 0.7928,
      "step": 365
    },
    {
      "epoch": 0.0463173880030372,
      "grad_norm": 2.0593314170837402,
      "learning_rate": 1.9988725013320646e-05,
      "loss": 0.7596,
      "step": 366
    },
    {
      "epoch": 0.04644393824348266,
      "grad_norm": 2.4573006629943848,
      "learning_rate": 1.9988639774671936e-05,
      "loss": 0.6914,
      "step": 367
    },
    {
      "epoch": 0.04657048848392812,
      "grad_norm": 2.2691891193389893,
      "learning_rate": 1.998855421521739e-05,
      "loss": 0.755,
      "step": 368
    },
    {
      "epoch": 0.04669703872437358,
      "grad_norm": 2.2925331592559814,
      "learning_rate": 1.9988468334959754e-05,
      "loss": 0.7303,
      "step": 369
    },
    {
      "epoch": 0.046823588964819036,
      "grad_norm": 1.7519915103912354,
      "learning_rate": 1.9988382133901785e-05,
      "loss": 0.7369,
      "step": 370
    },
    {
      "epoch": 0.04695013920526449,
      "grad_norm": 2.054250717163086,
      "learning_rate": 1.9988295612046254e-05,
      "loss": 0.7303,
      "step": 371
    },
    {
      "epoch": 0.047076689445709946,
      "grad_norm": 1.5741499662399292,
      "learning_rate": 1.9988208769395935e-05,
      "loss": 0.7041,
      "step": 372
    },
    {
      "epoch": 0.047203239686155404,
      "grad_norm": 2.123243570327759,
      "learning_rate": 1.998812160595362e-05,
      "loss": 0.7245,
      "step": 373
    },
    {
      "epoch": 0.04732978992660086,
      "grad_norm": 3.9763295650482178,
      "learning_rate": 1.9988034121722113e-05,
      "loss": 0.7904,
      "step": 374
    },
    {
      "epoch": 0.04745634016704632,
      "grad_norm": 2.4123129844665527,
      "learning_rate": 1.9987946316704218e-05,
      "loss": 0.7457,
      "step": 375
    },
    {
      "epoch": 0.04758289040749177,
      "grad_norm": 2.2609922885894775,
      "learning_rate": 1.9987858190902755e-05,
      "loss": 0.7519,
      "step": 376
    },
    {
      "epoch": 0.04770944064793723,
      "grad_norm": 8.226103782653809,
      "learning_rate": 1.998776974432056e-05,
      "loss": 0.7838,
      "step": 377
    },
    {
      "epoch": 0.04783599088838269,
      "grad_norm": 2.0057852268218994,
      "learning_rate": 1.9987680976960466e-05,
      "loss": 0.7634,
      "step": 378
    },
    {
      "epoch": 0.04796254112882815,
      "grad_norm": 2.2674946784973145,
      "learning_rate": 1.9987591888825326e-05,
      "loss": 0.7527,
      "step": 379
    },
    {
      "epoch": 0.0480890913692736,
      "grad_norm": 1.840714454650879,
      "learning_rate": 1.9987502479918006e-05,
      "loss": 0.7384,
      "step": 380
    },
    {
      "epoch": 0.04821564160971906,
      "grad_norm": 1.9293302297592163,
      "learning_rate": 1.998741275024137e-05,
      "loss": 0.7586,
      "step": 381
    },
    {
      "epoch": 0.048342191850164515,
      "grad_norm": 2.3589794635772705,
      "learning_rate": 1.998732269979831e-05,
      "loss": 0.7402,
      "step": 382
    },
    {
      "epoch": 0.04846874209060997,
      "grad_norm": 1.617301106452942,
      "learning_rate": 1.9987232328591712e-05,
      "loss": 0.7108,
      "step": 383
    },
    {
      "epoch": 0.04859529233105543,
      "grad_norm": 2.2161426544189453,
      "learning_rate": 1.9987141636624477e-05,
      "loss": 0.7299,
      "step": 384
    },
    {
      "epoch": 0.04872184257150088,
      "grad_norm": 1.8740520477294922,
      "learning_rate": 1.998705062389952e-05,
      "loss": 0.7317,
      "step": 385
    },
    {
      "epoch": 0.04884839281194634,
      "grad_norm": 1.848888874053955,
      "learning_rate": 1.9986959290419762e-05,
      "loss": 0.7389,
      "step": 386
    },
    {
      "epoch": 0.0489749430523918,
      "grad_norm": 2.688734292984009,
      "learning_rate": 1.9986867636188137e-05,
      "loss": 0.7403,
      "step": 387
    },
    {
      "epoch": 0.04910149329283726,
      "grad_norm": 1.7303390502929688,
      "learning_rate": 1.998677566120759e-05,
      "loss": 0.7206,
      "step": 388
    },
    {
      "epoch": 0.049228043533282716,
      "grad_norm": 1.8409491777420044,
      "learning_rate": 1.998668336548108e-05,
      "loss": 0.7664,
      "step": 389
    },
    {
      "epoch": 0.04935459377372817,
      "grad_norm": 1.9803327322006226,
      "learning_rate": 1.998659074901156e-05,
      "loss": 0.7502,
      "step": 390
    },
    {
      "epoch": 0.049481144014173625,
      "grad_norm": 1.7713662385940552,
      "learning_rate": 1.9986497811802015e-05,
      "loss": 0.7804,
      "step": 391
    },
    {
      "epoch": 0.049607694254619084,
      "grad_norm": 1.910206913948059,
      "learning_rate": 1.9986404553855423e-05,
      "loss": 0.7119,
      "step": 392
    },
    {
      "epoch": 0.04973424449506454,
      "grad_norm": 1.7927131652832031,
      "learning_rate": 1.9986310975174778e-05,
      "loss": 0.7087,
      "step": 393
    },
    {
      "epoch": 0.04986079473551,
      "grad_norm": 1.7266032695770264,
      "learning_rate": 1.9986217075763094e-05,
      "loss": 0.7608,
      "step": 394
    },
    {
      "epoch": 0.04998734497595545,
      "grad_norm": 1.6485451459884644,
      "learning_rate": 1.998612285562338e-05,
      "loss": 0.7315,
      "step": 395
    },
    {
      "epoch": 0.05011389521640091,
      "grad_norm": 1.6894564628601074,
      "learning_rate": 1.9986028314758664e-05,
      "loss": 0.7496,
      "step": 396
    },
    {
      "epoch": 0.05024044545684637,
      "grad_norm": 1.649774193763733,
      "learning_rate": 1.9985933453171983e-05,
      "loss": 0.7615,
      "step": 397
    },
    {
      "epoch": 0.050366995697291826,
      "grad_norm": 3.749188184738159,
      "learning_rate": 1.998583827086638e-05,
      "loss": 0.7756,
      "step": 398
    },
    {
      "epoch": 0.050493545937737284,
      "grad_norm": 1.8198847770690918,
      "learning_rate": 1.998574276784492e-05,
      "loss": 0.6881,
      "step": 399
    },
    {
      "epoch": 0.050620096178182736,
      "grad_norm": 2.641719341278076,
      "learning_rate": 1.998564694411066e-05,
      "loss": 0.7145,
      "step": 400
    },
    {
      "epoch": 0.050746646418628194,
      "grad_norm": 2.4125821590423584,
      "learning_rate": 1.9985550799666686e-05,
      "loss": 0.8025,
      "step": 401
    },
    {
      "epoch": 0.05087319665907365,
      "grad_norm": 3.2811150550842285,
      "learning_rate": 1.998545433451608e-05,
      "loss": 0.7869,
      "step": 402
    },
    {
      "epoch": 0.05099974689951911,
      "grad_norm": 1.7750908136367798,
      "learning_rate": 1.9985357548661945e-05,
      "loss": 0.7767,
      "step": 403
    },
    {
      "epoch": 0.05112629713996457,
      "grad_norm": 3.4508118629455566,
      "learning_rate": 1.9985260442107387e-05,
      "loss": 0.7795,
      "step": 404
    },
    {
      "epoch": 0.05125284738041002,
      "grad_norm": 2.2264323234558105,
      "learning_rate": 1.9985163014855523e-05,
      "loss": 0.797,
      "step": 405
    },
    {
      "epoch": 0.05137939762085548,
      "grad_norm": 3.349602460861206,
      "learning_rate": 1.9985065266909484e-05,
      "loss": 0.7193,
      "step": 406
    },
    {
      "epoch": 0.05150594786130094,
      "grad_norm": 2.6841611862182617,
      "learning_rate": 1.9984967198272414e-05,
      "loss": 0.723,
      "step": 407
    },
    {
      "epoch": 0.051632498101746395,
      "grad_norm": 2.505018949508667,
      "learning_rate": 1.9984868808947454e-05,
      "loss": 0.7645,
      "step": 408
    },
    {
      "epoch": 0.05175904834219185,
      "grad_norm": 2.2371463775634766,
      "learning_rate": 1.9984770098937768e-05,
      "loss": 0.774,
      "step": 409
    },
    {
      "epoch": 0.051885598582637305,
      "grad_norm": 2.507427930831909,
      "learning_rate": 1.9984671068246525e-05,
      "loss": 0.7836,
      "step": 410
    },
    {
      "epoch": 0.05201214882308276,
      "grad_norm": 2.0163538455963135,
      "learning_rate": 1.998457171687691e-05,
      "loss": 0.7339,
      "step": 411
    },
    {
      "epoch": 0.05213869906352822,
      "grad_norm": 3.3020756244659424,
      "learning_rate": 1.9984472044832114e-05,
      "loss": 0.7167,
      "step": 412
    },
    {
      "epoch": 0.05226524930397368,
      "grad_norm": 2.411696672439575,
      "learning_rate": 1.998437205211533e-05,
      "loss": 0.7104,
      "step": 413
    },
    {
      "epoch": 0.05239179954441914,
      "grad_norm": 2.6806347370147705,
      "learning_rate": 1.9984271738729775e-05,
      "loss": 0.759,
      "step": 414
    },
    {
      "epoch": 0.05251834978486459,
      "grad_norm": 2.42179274559021,
      "learning_rate": 1.9984171104678674e-05,
      "loss": 0.7745,
      "step": 415
    },
    {
      "epoch": 0.05264490002531005,
      "grad_norm": 2.218935251235962,
      "learning_rate": 1.998407014996525e-05,
      "loss": 0.707,
      "step": 416
    },
    {
      "epoch": 0.052771450265755505,
      "grad_norm": 2.5953571796417236,
      "learning_rate": 1.9983968874592753e-05,
      "loss": 0.7255,
      "step": 417
    },
    {
      "epoch": 0.052898000506200964,
      "grad_norm": 2.2375288009643555,
      "learning_rate": 1.9983867278564433e-05,
      "loss": 0.7231,
      "step": 418
    },
    {
      "epoch": 0.05302455074664642,
      "grad_norm": 2.0732016563415527,
      "learning_rate": 1.9983765361883555e-05,
      "loss": 0.703,
      "step": 419
    },
    {
      "epoch": 0.05315110098709187,
      "grad_norm": 1.8623052835464478,
      "learning_rate": 1.9983663124553388e-05,
      "loss": 0.7722,
      "step": 420
    },
    {
      "epoch": 0.05327765122753733,
      "grad_norm": 2.018340587615967,
      "learning_rate": 1.998356056657722e-05,
      "loss": 0.7397,
      "step": 421
    },
    {
      "epoch": 0.05340420146798279,
      "grad_norm": 2.055734157562256,
      "learning_rate": 1.9983457687958342e-05,
      "loss": 0.7522,
      "step": 422
    },
    {
      "epoch": 0.05353075170842825,
      "grad_norm": 2.1646268367767334,
      "learning_rate": 1.9983354488700063e-05,
      "loss": 0.7418,
      "step": 423
    },
    {
      "epoch": 0.053657301948873706,
      "grad_norm": 1.6597952842712402,
      "learning_rate": 1.998325096880569e-05,
      "loss": 0.7568,
      "step": 424
    },
    {
      "epoch": 0.05378385218931916,
      "grad_norm": 1.5719071626663208,
      "learning_rate": 1.9983147128278553e-05,
      "loss": 0.7238,
      "step": 425
    },
    {
      "epoch": 0.053910402429764616,
      "grad_norm": 1.7099846601486206,
      "learning_rate": 1.998304296712198e-05,
      "loss": 0.7378,
      "step": 426
    },
    {
      "epoch": 0.054036952670210074,
      "grad_norm": 1.8653504848480225,
      "learning_rate": 1.998293848533933e-05,
      "loss": 0.7313,
      "step": 427
    },
    {
      "epoch": 0.05416350291065553,
      "grad_norm": 1.5161350965499878,
      "learning_rate": 1.9982833682933946e-05,
      "loss": 0.6936,
      "step": 428
    },
    {
      "epoch": 0.054290053151100984,
      "grad_norm": 1.5212278366088867,
      "learning_rate": 1.99827285599092e-05,
      "loss": 0.7425,
      "step": 429
    },
    {
      "epoch": 0.05441660339154644,
      "grad_norm": 1.589483380317688,
      "learning_rate": 1.9982623116268466e-05,
      "loss": 0.7385,
      "step": 430
    },
    {
      "epoch": 0.0545431536319919,
      "grad_norm": 1.609422206878662,
      "learning_rate": 1.998251735201513e-05,
      "loss": 0.7519,
      "step": 431
    },
    {
      "epoch": 0.05466970387243736,
      "grad_norm": 1.576114296913147,
      "learning_rate": 1.998241126715259e-05,
      "loss": 0.7277,
      "step": 432
    },
    {
      "epoch": 0.05479625411288282,
      "grad_norm": 1.6274117231369019,
      "learning_rate": 1.9982304861684257e-05,
      "loss": 0.7687,
      "step": 433
    },
    {
      "epoch": 0.05492280435332827,
      "grad_norm": 1.5008888244628906,
      "learning_rate": 1.998219813561354e-05,
      "loss": 0.7212,
      "step": 434
    },
    {
      "epoch": 0.055049354593773726,
      "grad_norm": 1.5968364477157593,
      "learning_rate": 1.9982091088943872e-05,
      "loss": 0.7375,
      "step": 435
    },
    {
      "epoch": 0.055175904834219185,
      "grad_norm": 1.4820812940597534,
      "learning_rate": 1.998198372167869e-05,
      "loss": 0.7378,
      "step": 436
    },
    {
      "epoch": 0.05530245507466464,
      "grad_norm": 1.5850917100906372,
      "learning_rate": 1.9981876033821446e-05,
      "loss": 0.7443,
      "step": 437
    },
    {
      "epoch": 0.0554290053151101,
      "grad_norm": 1.7735756635665894,
      "learning_rate": 1.9981768025375594e-05,
      "loss": 0.7457,
      "step": 438
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 1.5781291723251343,
      "learning_rate": 1.9981659696344603e-05,
      "loss": 0.7573,
      "step": 439
    },
    {
      "epoch": 0.05568210579600101,
      "grad_norm": 1.5923293828964233,
      "learning_rate": 1.9981551046731954e-05,
      "loss": 0.694,
      "step": 440
    },
    {
      "epoch": 0.05580865603644647,
      "grad_norm": 1.7094613313674927,
      "learning_rate": 1.9981442076541136e-05,
      "loss": 0.7574,
      "step": 441
    },
    {
      "epoch": 0.05593520627689193,
      "grad_norm": 1.6411291360855103,
      "learning_rate": 1.998133278577565e-05,
      "loss": 0.7432,
      "step": 442
    },
    {
      "epoch": 0.056061756517337386,
      "grad_norm": 1.9845683574676514,
      "learning_rate": 1.9981223174439003e-05,
      "loss": 0.7479,
      "step": 443
    },
    {
      "epoch": 0.05618830675778284,
      "grad_norm": 1.6264742612838745,
      "learning_rate": 1.9981113242534718e-05,
      "loss": 0.7592,
      "step": 444
    },
    {
      "epoch": 0.056314856998228295,
      "grad_norm": 1.5474179983139038,
      "learning_rate": 1.9981002990066325e-05,
      "loss": 0.7395,
      "step": 445
    },
    {
      "epoch": 0.05644140723867375,
      "grad_norm": 1.9311972856521606,
      "learning_rate": 1.9980892417037365e-05,
      "loss": 0.737,
      "step": 446
    },
    {
      "epoch": 0.05656795747911921,
      "grad_norm": 1.591695785522461,
      "learning_rate": 1.9980781523451387e-05,
      "loss": 0.7464,
      "step": 447
    },
    {
      "epoch": 0.05669450771956467,
      "grad_norm": 1.743865728378296,
      "learning_rate": 1.9980670309311958e-05,
      "loss": 0.7423,
      "step": 448
    },
    {
      "epoch": 0.05682105796001012,
      "grad_norm": 1.764783263206482,
      "learning_rate": 1.9980558774622643e-05,
      "loss": 0.7461,
      "step": 449
    },
    {
      "epoch": 0.05694760820045558,
      "grad_norm": 1.5077862739562988,
      "learning_rate": 1.9980446919387032e-05,
      "loss": 0.7241,
      "step": 450
    },
    {
      "epoch": 0.05707415844090104,
      "grad_norm": 1.8361295461654663,
      "learning_rate": 1.9980334743608712e-05,
      "loss": 0.7843,
      "step": 451
    },
    {
      "epoch": 0.057200708681346496,
      "grad_norm": 1.8968926668167114,
      "learning_rate": 1.9980222247291285e-05,
      "loss": 0.7356,
      "step": 452
    },
    {
      "epoch": 0.057327258921791954,
      "grad_norm": 1.6816670894622803,
      "learning_rate": 1.998010943043837e-05,
      "loss": 0.7373,
      "step": 453
    },
    {
      "epoch": 0.057453809162237406,
      "grad_norm": 1.6320533752441406,
      "learning_rate": 1.9979996293053583e-05,
      "loss": 0.7688,
      "step": 454
    },
    {
      "epoch": 0.057580359402682864,
      "grad_norm": 1.940863847732544,
      "learning_rate": 1.997988283514056e-05,
      "loss": 0.7713,
      "step": 455
    },
    {
      "epoch": 0.05770690964312832,
      "grad_norm": 1.5767347812652588,
      "learning_rate": 1.997976905670295e-05,
      "loss": 0.7347,
      "step": 456
    },
    {
      "epoch": 0.05783345988357378,
      "grad_norm": 2.1415600776672363,
      "learning_rate": 1.9979654957744397e-05,
      "loss": 0.7302,
      "step": 457
    },
    {
      "epoch": 0.05796001012401924,
      "grad_norm": 1.7806148529052734,
      "learning_rate": 1.9979540538268577e-05,
      "loss": 0.6972,
      "step": 458
    },
    {
      "epoch": 0.05808656036446469,
      "grad_norm": 1.5478593111038208,
      "learning_rate": 1.997942579827916e-05,
      "loss": 0.7154,
      "step": 459
    },
    {
      "epoch": 0.05821311060491015,
      "grad_norm": 1.9780018329620361,
      "learning_rate": 1.997931073777983e-05,
      "loss": 0.7482,
      "step": 460
    },
    {
      "epoch": 0.058339660845355606,
      "grad_norm": 1.6623482704162598,
      "learning_rate": 1.997919535677428e-05,
      "loss": 0.7751,
      "step": 461
    },
    {
      "epoch": 0.058466211085801065,
      "grad_norm": 1.5676614046096802,
      "learning_rate": 1.9979079655266222e-05,
      "loss": 0.7862,
      "step": 462
    },
    {
      "epoch": 0.05859276132624652,
      "grad_norm": 1.5995910167694092,
      "learning_rate": 1.9978963633259367e-05,
      "loss": 0.7431,
      "step": 463
    },
    {
      "epoch": 0.058719311566691974,
      "grad_norm": 1.5758317708969116,
      "learning_rate": 1.9978847290757443e-05,
      "loss": 0.7053,
      "step": 464
    },
    {
      "epoch": 0.05884586180713743,
      "grad_norm": 1.714404821395874,
      "learning_rate": 1.9978730627764186e-05,
      "loss": 0.7109,
      "step": 465
    },
    {
      "epoch": 0.05897241204758289,
      "grad_norm": 1.6472957134246826,
      "learning_rate": 1.9978613644283347e-05,
      "loss": 0.8058,
      "step": 466
    },
    {
      "epoch": 0.05909896228802835,
      "grad_norm": 2.0002057552337646,
      "learning_rate": 1.9978496340318676e-05,
      "loss": 0.677,
      "step": 467
    },
    {
      "epoch": 0.05922551252847381,
      "grad_norm": 1.6078999042510986,
      "learning_rate": 1.9978378715873946e-05,
      "loss": 0.7575,
      "step": 468
    },
    {
      "epoch": 0.05935206276891926,
      "grad_norm": 1.8979408740997314,
      "learning_rate": 1.9978260770952935e-05,
      "loss": 0.7245,
      "step": 469
    },
    {
      "epoch": 0.05947861300936472,
      "grad_norm": 1.6448585987091064,
      "learning_rate": 1.9978142505559423e-05,
      "loss": 0.75,
      "step": 470
    },
    {
      "epoch": 0.059605163249810175,
      "grad_norm": 1.6064947843551636,
      "learning_rate": 1.9978023919697217e-05,
      "loss": 0.7934,
      "step": 471
    },
    {
      "epoch": 0.059731713490255633,
      "grad_norm": 2.0620546340942383,
      "learning_rate": 1.9977905013370128e-05,
      "loss": 0.7068,
      "step": 472
    },
    {
      "epoch": 0.05985826373070109,
      "grad_norm": 1.5774785280227661,
      "learning_rate": 1.9977785786581964e-05,
      "loss": 0.7339,
      "step": 473
    },
    {
      "epoch": 0.05998481397114654,
      "grad_norm": 1.8163349628448486,
      "learning_rate": 1.9977666239336563e-05,
      "loss": 0.7672,
      "step": 474
    },
    {
      "epoch": 0.060111364211592,
      "grad_norm": 2.4444737434387207,
      "learning_rate": 1.9977546371637758e-05,
      "loss": 0.7325,
      "step": 475
    },
    {
      "epoch": 0.06023791445203746,
      "grad_norm": 1.5919833183288574,
      "learning_rate": 1.9977426183489407e-05,
      "loss": 0.7622,
      "step": 476
    },
    {
      "epoch": 0.06036446469248292,
      "grad_norm": 2.559626340866089,
      "learning_rate": 1.997730567489536e-05,
      "loss": 0.7754,
      "step": 477
    },
    {
      "epoch": 0.06049101493292837,
      "grad_norm": 1.5898832082748413,
      "learning_rate": 1.9977184845859497e-05,
      "loss": 0.783,
      "step": 478
    },
    {
      "epoch": 0.06061756517337383,
      "grad_norm": 2.305893898010254,
      "learning_rate": 1.9977063696385692e-05,
      "loss": 0.7249,
      "step": 479
    },
    {
      "epoch": 0.060744115413819286,
      "grad_norm": 1.5691099166870117,
      "learning_rate": 1.997694222647784e-05,
      "loss": 0.7733,
      "step": 480
    },
    {
      "epoch": 0.060870665654264744,
      "grad_norm": 2.1325223445892334,
      "learning_rate": 1.9976820436139843e-05,
      "loss": 0.7633,
      "step": 481
    },
    {
      "epoch": 0.0609972158947102,
      "grad_norm": 1.4885525703430176,
      "learning_rate": 1.9976698325375607e-05,
      "loss": 0.7366,
      "step": 482
    },
    {
      "epoch": 0.061123766135155654,
      "grad_norm": 1.9422506093978882,
      "learning_rate": 1.9976575894189056e-05,
      "loss": 0.7693,
      "step": 483
    },
    {
      "epoch": 0.06125031637560111,
      "grad_norm": 2.8598029613494873,
      "learning_rate": 1.9976453142584128e-05,
      "loss": 0.7461,
      "step": 484
    },
    {
      "epoch": 0.06137686661604657,
      "grad_norm": 1.788918137550354,
      "learning_rate": 1.9976330070564755e-05,
      "loss": 0.7384,
      "step": 485
    },
    {
      "epoch": 0.06150341685649203,
      "grad_norm": 1.8950351476669312,
      "learning_rate": 1.99762066781349e-05,
      "loss": 0.7085,
      "step": 486
    },
    {
      "epoch": 0.06162996709693749,
      "grad_norm": 1.6177431344985962,
      "learning_rate": 1.997608296529852e-05,
      "loss": 0.7718,
      "step": 487
    },
    {
      "epoch": 0.06175651733738294,
      "grad_norm": 1.9570757150650024,
      "learning_rate": 1.9975958932059587e-05,
      "loss": 0.7458,
      "step": 488
    },
    {
      "epoch": 0.061883067577828396,
      "grad_norm": 1.810065746307373,
      "learning_rate": 1.9975834578422092e-05,
      "loss": 0.7323,
      "step": 489
    },
    {
      "epoch": 0.062009617818273854,
      "grad_norm": 1.703405737876892,
      "learning_rate": 1.997570990439002e-05,
      "loss": 0.7873,
      "step": 490
    },
    {
      "epoch": 0.06213616805871931,
      "grad_norm": 2.167327404022217,
      "learning_rate": 1.9975584909967382e-05,
      "loss": 0.753,
      "step": 491
    },
    {
      "epoch": 0.06226271829916477,
      "grad_norm": 1.5923563241958618,
      "learning_rate": 1.9975459595158186e-05,
      "loss": 0.723,
      "step": 492
    },
    {
      "epoch": 0.06238926853961022,
      "grad_norm": 1.698549747467041,
      "learning_rate": 1.9975333959966466e-05,
      "loss": 0.7401,
      "step": 493
    },
    {
      "epoch": 0.06251581878005569,
      "grad_norm": 2.045020580291748,
      "learning_rate": 1.9975208004396247e-05,
      "loss": 0.7686,
      "step": 494
    },
    {
      "epoch": 0.06264236902050115,
      "grad_norm": 1.652764916419983,
      "learning_rate": 1.9975081728451583e-05,
      "loss": 0.7253,
      "step": 495
    },
    {
      "epoch": 0.06276891926094659,
      "grad_norm": 1.693906545639038,
      "learning_rate": 1.997495513213652e-05,
      "loss": 0.779,
      "step": 496
    },
    {
      "epoch": 0.06289546950139205,
      "grad_norm": 1.6372843980789185,
      "learning_rate": 1.9974828215455133e-05,
      "loss": 0.7656,
      "step": 497
    },
    {
      "epoch": 0.0630220197418375,
      "grad_norm": 1.632106900215149,
      "learning_rate": 1.9974700978411495e-05,
      "loss": 0.7725,
      "step": 498
    },
    {
      "epoch": 0.06314856998228296,
      "grad_norm": 1.718024492263794,
      "learning_rate": 1.997457342100969e-05,
      "loss": 0.71,
      "step": 499
    },
    {
      "epoch": 0.06327512022272842,
      "grad_norm": 1.56805419921875,
      "learning_rate": 1.997444554325382e-05,
      "loss": 0.7703,
      "step": 500
    },
    {
      "epoch": 0.06340167046317388,
      "grad_norm": 1.5445226430892944,
      "learning_rate": 1.9974317345147985e-05,
      "loss": 0.765,
      "step": 501
    },
    {
      "epoch": 0.06352822070361934,
      "grad_norm": 1.9556280374526978,
      "learning_rate": 1.997418882669631e-05,
      "loss": 0.7632,
      "step": 502
    },
    {
      "epoch": 0.0636547709440648,
      "grad_norm": 1.7386813163757324,
      "learning_rate": 1.9974059987902916e-05,
      "loss": 0.7614,
      "step": 503
    },
    {
      "epoch": 0.06378132118451026,
      "grad_norm": 1.6317558288574219,
      "learning_rate": 1.9973930828771944e-05,
      "loss": 0.7261,
      "step": 504
    },
    {
      "epoch": 0.0639078714249557,
      "grad_norm": 1.8482298851013184,
      "learning_rate": 1.9973801349307542e-05,
      "loss": 0.7429,
      "step": 505
    },
    {
      "epoch": 0.06403442166540116,
      "grad_norm": 3.433553695678711,
      "learning_rate": 1.997367154951387e-05,
      "loss": 0.7439,
      "step": 506
    },
    {
      "epoch": 0.06416097190584662,
      "grad_norm": 1.7826015949249268,
      "learning_rate": 1.9973541429395092e-05,
      "loss": 0.7285,
      "step": 507
    },
    {
      "epoch": 0.06428752214629208,
      "grad_norm": 1.8994264602661133,
      "learning_rate": 1.9973410988955396e-05,
      "loss": 0.7383,
      "step": 508
    },
    {
      "epoch": 0.06441407238673753,
      "grad_norm": 1.5677827596664429,
      "learning_rate": 1.9973280228198963e-05,
      "loss": 0.6947,
      "step": 509
    },
    {
      "epoch": 0.06454062262718299,
      "grad_norm": 2.2271389961242676,
      "learning_rate": 1.9973149147129993e-05,
      "loss": 0.7678,
      "step": 510
    },
    {
      "epoch": 0.06466717286762845,
      "grad_norm": 2.625534772872925,
      "learning_rate": 1.99730177457527e-05,
      "loss": 0.7778,
      "step": 511
    },
    {
      "epoch": 0.06479372310807391,
      "grad_norm": 2.141263484954834,
      "learning_rate": 1.9972886024071302e-05,
      "loss": 0.7272,
      "step": 512
    },
    {
      "epoch": 0.06492027334851937,
      "grad_norm": 2.0913381576538086,
      "learning_rate": 1.997275398209003e-05,
      "loss": 0.7659,
      "step": 513
    },
    {
      "epoch": 0.06504682358896482,
      "grad_norm": 1.6604971885681152,
      "learning_rate": 1.9972621619813127e-05,
      "loss": 0.7786,
      "step": 514
    },
    {
      "epoch": 0.06517337382941027,
      "grad_norm": 2.4928152561187744,
      "learning_rate": 1.997248893724484e-05,
      "loss": 0.697,
      "step": 515
    },
    {
      "epoch": 0.06529992406985573,
      "grad_norm": 1.5817365646362305,
      "learning_rate": 1.9972355934389433e-05,
      "loss": 0.7678,
      "step": 516
    },
    {
      "epoch": 0.06542647431030119,
      "grad_norm": 1.7203079462051392,
      "learning_rate": 1.9972222611251177e-05,
      "loss": 0.7564,
      "step": 517
    },
    {
      "epoch": 0.06555302455074664,
      "grad_norm": 1.5923702716827393,
      "learning_rate": 1.9972088967834352e-05,
      "loss": 0.7663,
      "step": 518
    },
    {
      "epoch": 0.0656795747911921,
      "grad_norm": 1.4810463190078735,
      "learning_rate": 1.9971955004143256e-05,
      "loss": 0.7356,
      "step": 519
    },
    {
      "epoch": 0.06580612503163756,
      "grad_norm": 1.553257942199707,
      "learning_rate": 1.9971820720182186e-05,
      "loss": 0.7516,
      "step": 520
    },
    {
      "epoch": 0.06593267527208302,
      "grad_norm": 1.7238121032714844,
      "learning_rate": 1.997168611595546e-05,
      "loss": 0.7394,
      "step": 521
    },
    {
      "epoch": 0.06605922551252848,
      "grad_norm": 1.6913905143737793,
      "learning_rate": 1.997155119146739e-05,
      "loss": 0.7384,
      "step": 522
    },
    {
      "epoch": 0.06618577575297394,
      "grad_norm": 1.7064911127090454,
      "learning_rate": 1.9971415946722323e-05,
      "loss": 0.7869,
      "step": 523
    },
    {
      "epoch": 0.0663123259934194,
      "grad_norm": 1.5547065734863281,
      "learning_rate": 1.997128038172459e-05,
      "loss": 0.6959,
      "step": 524
    },
    {
      "epoch": 0.06643887623386484,
      "grad_norm": 1.5766282081604004,
      "learning_rate": 1.997114449647856e-05,
      "loss": 0.7463,
      "step": 525
    },
    {
      "epoch": 0.0665654264743103,
      "grad_norm": 1.6752922534942627,
      "learning_rate": 1.997100829098858e-05,
      "loss": 0.7166,
      "step": 526
    },
    {
      "epoch": 0.06669197671475575,
      "grad_norm": 1.4589229822158813,
      "learning_rate": 1.997087176525904e-05,
      "loss": 0.7429,
      "step": 527
    },
    {
      "epoch": 0.06681852695520121,
      "grad_norm": 1.5544849634170532,
      "learning_rate": 1.997073491929432e-05,
      "loss": 0.7274,
      "step": 528
    },
    {
      "epoch": 0.06694507719564667,
      "grad_norm": 1.8031620979309082,
      "learning_rate": 1.9970597753098806e-05,
      "loss": 0.6934,
      "step": 529
    },
    {
      "epoch": 0.06707162743609213,
      "grad_norm": 1.8216593265533447,
      "learning_rate": 1.9970460266676914e-05,
      "loss": 0.7797,
      "step": 530
    },
    {
      "epoch": 0.06719817767653759,
      "grad_norm": 1.7411493062973022,
      "learning_rate": 1.9970322460033058e-05,
      "loss": 0.8206,
      "step": 531
    },
    {
      "epoch": 0.06732472791698305,
      "grad_norm": 2.175513744354248,
      "learning_rate": 1.9970184333171662e-05,
      "loss": 0.7614,
      "step": 532
    },
    {
      "epoch": 0.0674512781574285,
      "grad_norm": 1.634366512298584,
      "learning_rate": 1.9970045886097158e-05,
      "loss": 0.7299,
      "step": 533
    },
    {
      "epoch": 0.06757782839787396,
      "grad_norm": 1.7188187837600708,
      "learning_rate": 1.9969907118814002e-05,
      "loss": 0.7724,
      "step": 534
    },
    {
      "epoch": 0.0677043786383194,
      "grad_norm": 1.8019986152648926,
      "learning_rate": 1.9969768031326642e-05,
      "loss": 0.8164,
      "step": 535
    },
    {
      "epoch": 0.06783092887876487,
      "grad_norm": 1.5235289335250854,
      "learning_rate": 1.9969628623639553e-05,
      "loss": 0.7179,
      "step": 536
    },
    {
      "epoch": 0.06795747911921032,
      "grad_norm": 2.0712671279907227,
      "learning_rate": 1.9969488895757206e-05,
      "loss": 0.7402,
      "step": 537
    },
    {
      "epoch": 0.06808402935965578,
      "grad_norm": 1.807409405708313,
      "learning_rate": 1.996934884768409e-05,
      "loss": 0.7363,
      "step": 538
    },
    {
      "epoch": 0.06821057960010124,
      "grad_norm": 1.7568745613098145,
      "learning_rate": 1.9969208479424704e-05,
      "loss": 0.7975,
      "step": 539
    },
    {
      "epoch": 0.0683371298405467,
      "grad_norm": 1.6273751258850098,
      "learning_rate": 1.996906779098356e-05,
      "loss": 0.7074,
      "step": 540
    },
    {
      "epoch": 0.06846368008099216,
      "grad_norm": 1.8117600679397583,
      "learning_rate": 1.996892678236517e-05,
      "loss": 0.7698,
      "step": 541
    },
    {
      "epoch": 0.06859023032143761,
      "grad_norm": 1.6508816480636597,
      "learning_rate": 1.996878545357406e-05,
      "loss": 0.7895,
      "step": 542
    },
    {
      "epoch": 0.06871678056188307,
      "grad_norm": 1.611555576324463,
      "learning_rate": 1.996864380461478e-05,
      "loss": 0.7427,
      "step": 543
    },
    {
      "epoch": 0.06884333080232853,
      "grad_norm": 1.7622883319854736,
      "learning_rate": 1.9968501835491872e-05,
      "loss": 0.7351,
      "step": 544
    },
    {
      "epoch": 0.06896988104277398,
      "grad_norm": 1.6539791822433472,
      "learning_rate": 1.99683595462099e-05,
      "loss": 0.7091,
      "step": 545
    },
    {
      "epoch": 0.06909643128321943,
      "grad_norm": 1.5874444246292114,
      "learning_rate": 1.9968216936773425e-05,
      "loss": 0.7701,
      "step": 546
    },
    {
      "epoch": 0.06922298152366489,
      "grad_norm": 1.733399510383606,
      "learning_rate": 1.996807400718704e-05,
      "loss": 0.7247,
      "step": 547
    },
    {
      "epoch": 0.06934953176411035,
      "grad_norm": 1.646468997001648,
      "learning_rate": 1.9967930757455325e-05,
      "loss": 0.7482,
      "step": 548
    },
    {
      "epoch": 0.06947608200455581,
      "grad_norm": 1.853247880935669,
      "learning_rate": 1.9967787187582886e-05,
      "loss": 0.7658,
      "step": 549
    },
    {
      "epoch": 0.06960263224500127,
      "grad_norm": 1.9162883758544922,
      "learning_rate": 1.9967643297574332e-05,
      "loss": 0.7244,
      "step": 550
    },
    {
      "epoch": 0.06972918248544673,
      "grad_norm": 2.2719955444335938,
      "learning_rate": 1.996749908743429e-05,
      "loss": 0.7708,
      "step": 551
    },
    {
      "epoch": 0.06985573272589218,
      "grad_norm": 1.7887380123138428,
      "learning_rate": 1.996735455716738e-05,
      "loss": 0.7571,
      "step": 552
    },
    {
      "epoch": 0.06998228296633764,
      "grad_norm": 1.734606146812439,
      "learning_rate": 1.9967209706778256e-05,
      "loss": 0.7204,
      "step": 553
    },
    {
      "epoch": 0.07010883320678309,
      "grad_norm": 1.9300377368927002,
      "learning_rate": 1.996706453627156e-05,
      "loss": 0.7689,
      "step": 554
    },
    {
      "epoch": 0.07023538344722854,
      "grad_norm": 1.7058435678482056,
      "learning_rate": 1.996691904565196e-05,
      "loss": 0.7649,
      "step": 555
    },
    {
      "epoch": 0.070361933687674,
      "grad_norm": 1.7513755559921265,
      "learning_rate": 1.996677323492413e-05,
      "loss": 0.7465,
      "step": 556
    },
    {
      "epoch": 0.07048848392811946,
      "grad_norm": 1.8265395164489746,
      "learning_rate": 1.996662710409275e-05,
      "loss": 0.7842,
      "step": 557
    },
    {
      "epoch": 0.07061503416856492,
      "grad_norm": 2.447066068649292,
      "learning_rate": 1.9966480653162516e-05,
      "loss": 0.7658,
      "step": 558
    },
    {
      "epoch": 0.07074158440901038,
      "grad_norm": 3.4016926288604736,
      "learning_rate": 1.996633388213813e-05,
      "loss": 0.7508,
      "step": 559
    },
    {
      "epoch": 0.07086813464945584,
      "grad_norm": 2.4929800033569336,
      "learning_rate": 1.9966186791024303e-05,
      "loss": 0.7667,
      "step": 560
    },
    {
      "epoch": 0.0709946848899013,
      "grad_norm": 3.7309176921844482,
      "learning_rate": 1.996603937982576e-05,
      "loss": 0.7717,
      "step": 561
    },
    {
      "epoch": 0.07112123513034675,
      "grad_norm": 2.1441762447357178,
      "learning_rate": 1.996589164854724e-05,
      "loss": 0.7792,
      "step": 562
    },
    {
      "epoch": 0.07124778537079221,
      "grad_norm": 1.5813885927200317,
      "learning_rate": 1.9965743597193487e-05,
      "loss": 0.747,
      "step": 563
    },
    {
      "epoch": 0.07137433561123765,
      "grad_norm": 1.6694176197052002,
      "learning_rate": 1.996559522576925e-05,
      "loss": 0.7567,
      "step": 564
    },
    {
      "epoch": 0.07150088585168311,
      "grad_norm": 1.7345050573349,
      "learning_rate": 1.99654465342793e-05,
      "loss": 0.7096,
      "step": 565
    },
    {
      "epoch": 0.07162743609212857,
      "grad_norm": 3.3624837398529053,
      "learning_rate": 1.996529752272841e-05,
      "loss": 0.7789,
      "step": 566
    },
    {
      "epoch": 0.07175398633257403,
      "grad_norm": 7.170213222503662,
      "learning_rate": 1.9965148191121368e-05,
      "loss": 0.7522,
      "step": 567
    },
    {
      "epoch": 0.07188053657301949,
      "grad_norm": 2.0450081825256348,
      "learning_rate": 1.9964998539462966e-05,
      "loss": 0.725,
      "step": 568
    },
    {
      "epoch": 0.07200708681346495,
      "grad_norm": 1.7531481981277466,
      "learning_rate": 1.9964848567758016e-05,
      "loss": 0.7744,
      "step": 569
    },
    {
      "epoch": 0.0721336370539104,
      "grad_norm": 5.371631145477295,
      "learning_rate": 1.996469827601133e-05,
      "loss": 0.7399,
      "step": 570
    },
    {
      "epoch": 0.07226018729435586,
      "grad_norm": 2.915452718734741,
      "learning_rate": 1.9964547664227737e-05,
      "loss": 0.7661,
      "step": 571
    },
    {
      "epoch": 0.07238673753480132,
      "grad_norm": 1.8081164360046387,
      "learning_rate": 1.9964396732412073e-05,
      "loss": 0.7595,
      "step": 572
    },
    {
      "epoch": 0.07251328777524678,
      "grad_norm": 2.1392085552215576,
      "learning_rate": 1.9964245480569185e-05,
      "loss": 0.7167,
      "step": 573
    },
    {
      "epoch": 0.07263983801569222,
      "grad_norm": 1.8647606372833252,
      "learning_rate": 1.9964093908703934e-05,
      "loss": 0.7862,
      "step": 574
    },
    {
      "epoch": 0.07276638825613768,
      "grad_norm": 2.071840286254883,
      "learning_rate": 1.996394201682119e-05,
      "loss": 0.7878,
      "step": 575
    },
    {
      "epoch": 0.07289293849658314,
      "grad_norm": 1.9130667448043823,
      "learning_rate": 1.9963789804925818e-05,
      "loss": 0.7297,
      "step": 576
    },
    {
      "epoch": 0.0730194887370286,
      "grad_norm": 1.6019047498703003,
      "learning_rate": 1.996363727302272e-05,
      "loss": 0.7257,
      "step": 577
    },
    {
      "epoch": 0.07314603897747406,
      "grad_norm": 2.3450000286102295,
      "learning_rate": 1.996348442111679e-05,
      "loss": 0.7547,
      "step": 578
    },
    {
      "epoch": 0.07327258921791951,
      "grad_norm": 2.027743339538574,
      "learning_rate": 1.9963331249212943e-05,
      "loss": 0.7735,
      "step": 579
    },
    {
      "epoch": 0.07339913945836497,
      "grad_norm": 1.4903640747070312,
      "learning_rate": 1.996317775731609e-05,
      "loss": 0.7077,
      "step": 580
    },
    {
      "epoch": 0.07352568969881043,
      "grad_norm": 2.327136516571045,
      "learning_rate": 1.9963023945431167e-05,
      "loss": 0.7715,
      "step": 581
    },
    {
      "epoch": 0.07365223993925589,
      "grad_norm": 1.6279491186141968,
      "learning_rate": 1.9962869813563105e-05,
      "loss": 0.7166,
      "step": 582
    },
    {
      "epoch": 0.07377879017970135,
      "grad_norm": 5.478910446166992,
      "learning_rate": 1.9962715361716864e-05,
      "loss": 0.7186,
      "step": 583
    },
    {
      "epoch": 0.07390534042014679,
      "grad_norm": 2.1897311210632324,
      "learning_rate": 1.9962560589897402e-05,
      "loss": 0.7251,
      "step": 584
    },
    {
      "epoch": 0.07403189066059225,
      "grad_norm": 1.726898193359375,
      "learning_rate": 1.9962405498109686e-05,
      "loss": 0.77,
      "step": 585
    },
    {
      "epoch": 0.07415844090103771,
      "grad_norm": 2.018368721008301,
      "learning_rate": 1.9962250086358702e-05,
      "loss": 0.7613,
      "step": 586
    },
    {
      "epoch": 0.07428499114148317,
      "grad_norm": 1.723206639289856,
      "learning_rate": 1.996209435464944e-05,
      "loss": 0.7153,
      "step": 587
    },
    {
      "epoch": 0.07441154138192863,
      "grad_norm": 1.6836010217666626,
      "learning_rate": 1.9961938302986897e-05,
      "loss": 0.7402,
      "step": 588
    },
    {
      "epoch": 0.07453809162237408,
      "grad_norm": 2.208045482635498,
      "learning_rate": 1.996178193137609e-05,
      "loss": 0.7588,
      "step": 589
    },
    {
      "epoch": 0.07466464186281954,
      "grad_norm": 1.7726982831954956,
      "learning_rate": 1.9961625239822045e-05,
      "loss": 0.7738,
      "step": 590
    },
    {
      "epoch": 0.074791192103265,
      "grad_norm": 2.4224860668182373,
      "learning_rate": 1.9961468228329784e-05,
      "loss": 0.737,
      "step": 591
    },
    {
      "epoch": 0.07491774234371046,
      "grad_norm": 1.7084523439407349,
      "learning_rate": 1.996131089690436e-05,
      "loss": 0.6942,
      "step": 592
    },
    {
      "epoch": 0.07504429258415592,
      "grad_norm": 1.588350772857666,
      "learning_rate": 1.9961153245550815e-05,
      "loss": 0.7699,
      "step": 593
    },
    {
      "epoch": 0.07517084282460136,
      "grad_norm": 1.9261937141418457,
      "learning_rate": 1.9960995274274222e-05,
      "loss": 0.775,
      "step": 594
    },
    {
      "epoch": 0.07529739306504682,
      "grad_norm": 1.837291955947876,
      "learning_rate": 1.996083698307965e-05,
      "loss": 0.7518,
      "step": 595
    },
    {
      "epoch": 0.07542394330549228,
      "grad_norm": 1.7726119756698608,
      "learning_rate": 1.9960678371972184e-05,
      "loss": 0.7272,
      "step": 596
    },
    {
      "epoch": 0.07555049354593774,
      "grad_norm": 1.9974706172943115,
      "learning_rate": 1.996051944095692e-05,
      "loss": 0.7095,
      "step": 597
    },
    {
      "epoch": 0.0756770437863832,
      "grad_norm": 1.5493909120559692,
      "learning_rate": 1.996036019003896e-05,
      "loss": 0.7312,
      "step": 598
    },
    {
      "epoch": 0.07580359402682865,
      "grad_norm": 2.0072567462921143,
      "learning_rate": 1.9960200619223418e-05,
      "loss": 0.7526,
      "step": 599
    },
    {
      "epoch": 0.07593014426727411,
      "grad_norm": 1.6817883253097534,
      "learning_rate": 1.996004072851542e-05,
      "loss": 0.7393,
      "step": 600
    },
    {
      "epoch": 0.07605669450771957,
      "grad_norm": 1.8753128051757812,
      "learning_rate": 1.9959880517920102e-05,
      "loss": 0.7763,
      "step": 601
    },
    {
      "epoch": 0.07618324474816503,
      "grad_norm": 1.8623238801956177,
      "learning_rate": 1.9959719987442607e-05,
      "loss": 0.6941,
      "step": 602
    },
    {
      "epoch": 0.07630979498861047,
      "grad_norm": 1.5703284740447998,
      "learning_rate": 1.9959559137088097e-05,
      "loss": 0.7337,
      "step": 603
    },
    {
      "epoch": 0.07643634522905593,
      "grad_norm": 1.859286904335022,
      "learning_rate": 1.995939796686173e-05,
      "loss": 0.7325,
      "step": 604
    },
    {
      "epoch": 0.07656289546950139,
      "grad_norm": 1.7669713497161865,
      "learning_rate": 1.9959236476768685e-05,
      "loss": 0.7548,
      "step": 605
    },
    {
      "epoch": 0.07668944570994685,
      "grad_norm": 15.930581092834473,
      "learning_rate": 1.9959074666814154e-05,
      "loss": 0.7138,
      "step": 606
    },
    {
      "epoch": 0.0768159959503923,
      "grad_norm": 1.8625502586364746,
      "learning_rate": 1.9958912537003323e-05,
      "loss": 0.6931,
      "step": 607
    },
    {
      "epoch": 0.07694254619083776,
      "grad_norm": 1.5301278829574585,
      "learning_rate": 1.995875008734141e-05,
      "loss": 0.7177,
      "step": 608
    },
    {
      "epoch": 0.07706909643128322,
      "grad_norm": 1.5992401838302612,
      "learning_rate": 1.9958587317833626e-05,
      "loss": 0.7284,
      "step": 609
    },
    {
      "epoch": 0.07719564667172868,
      "grad_norm": 1.5269906520843506,
      "learning_rate": 1.9958424228485204e-05,
      "loss": 0.7461,
      "step": 610
    },
    {
      "epoch": 0.07732219691217414,
      "grad_norm": 3.3848605155944824,
      "learning_rate": 1.9958260819301375e-05,
      "loss": 0.7319,
      "step": 611
    },
    {
      "epoch": 0.0774487471526196,
      "grad_norm": 1.6256392002105713,
      "learning_rate": 1.9958097090287393e-05,
      "loss": 0.6871,
      "step": 612
    },
    {
      "epoch": 0.07757529739306504,
      "grad_norm": 2.0586154460906982,
      "learning_rate": 1.995793304144851e-05,
      "loss": 0.7543,
      "step": 613
    },
    {
      "epoch": 0.0777018476335105,
      "grad_norm": 2.861776113510132,
      "learning_rate": 1.9957768672790006e-05,
      "loss": 0.7478,
      "step": 614
    },
    {
      "epoch": 0.07782839787395596,
      "grad_norm": 2.3442533016204834,
      "learning_rate": 1.9957603984317145e-05,
      "loss": 0.7884,
      "step": 615
    },
    {
      "epoch": 0.07795494811440142,
      "grad_norm": 1.632901668548584,
      "learning_rate": 1.9957438976035232e-05,
      "loss": 0.7279,
      "step": 616
    },
    {
      "epoch": 0.07808149835484687,
      "grad_norm": 1.739268183708191,
      "learning_rate": 1.9957273647949555e-05,
      "loss": 0.709,
      "step": 617
    },
    {
      "epoch": 0.07820804859529233,
      "grad_norm": 2.088954448699951,
      "learning_rate": 1.995710800006543e-05,
      "loss": 0.7607,
      "step": 618
    },
    {
      "epoch": 0.07833459883573779,
      "grad_norm": 1.5475139617919922,
      "learning_rate": 1.9956942032388175e-05,
      "loss": 0.779,
      "step": 619
    },
    {
      "epoch": 0.07846114907618325,
      "grad_norm": 1.8258075714111328,
      "learning_rate": 1.9956775744923115e-05,
      "loss": 0.7538,
      "step": 620
    },
    {
      "epoch": 0.0785876993166287,
      "grad_norm": 1.7472097873687744,
      "learning_rate": 1.99566091376756e-05,
      "loss": 0.7292,
      "step": 621
    },
    {
      "epoch": 0.07871424955707416,
      "grad_norm": 1.554799199104309,
      "learning_rate": 1.995644221065098e-05,
      "loss": 0.7267,
      "step": 622
    },
    {
      "epoch": 0.07884079979751961,
      "grad_norm": 2.825268507003784,
      "learning_rate": 1.9956274963854606e-05,
      "loss": 0.8059,
      "step": 623
    },
    {
      "epoch": 0.07896735003796507,
      "grad_norm": 1.5879746675491333,
      "learning_rate": 1.9956107397291862e-05,
      "loss": 0.757,
      "step": 624
    },
    {
      "epoch": 0.07909390027841053,
      "grad_norm": 2.200544595718384,
      "learning_rate": 1.9955939510968122e-05,
      "loss": 0.7416,
      "step": 625
    },
    {
      "epoch": 0.07922045051885598,
      "grad_norm": 1.7468293905258179,
      "learning_rate": 1.995577130488878e-05,
      "loss": 0.6972,
      "step": 626
    },
    {
      "epoch": 0.07934700075930144,
      "grad_norm": 1.9627777338027954,
      "learning_rate": 1.9955602779059246e-05,
      "loss": 0.6856,
      "step": 627
    },
    {
      "epoch": 0.0794735509997469,
      "grad_norm": 1.9618955850601196,
      "learning_rate": 1.9955433933484917e-05,
      "loss": 0.7328,
      "step": 628
    },
    {
      "epoch": 0.07960010124019236,
      "grad_norm": 1.6855698823928833,
      "learning_rate": 1.9955264768171225e-05,
      "loss": 0.7151,
      "step": 629
    },
    {
      "epoch": 0.07972665148063782,
      "grad_norm": 2.300311803817749,
      "learning_rate": 1.9955095283123603e-05,
      "loss": 0.7243,
      "step": 630
    },
    {
      "epoch": 0.07985320172108328,
      "grad_norm": 1.7692513465881348,
      "learning_rate": 1.9954925478347494e-05,
      "loss": 0.7357,
      "step": 631
    },
    {
      "epoch": 0.07997975196152873,
      "grad_norm": 2.2853214740753174,
      "learning_rate": 1.995475535384835e-05,
      "loss": 0.7795,
      "step": 632
    },
    {
      "epoch": 0.08010630220197418,
      "grad_norm": 1.6379685401916504,
      "learning_rate": 1.995458490963164e-05,
      "loss": 0.7467,
      "step": 633
    },
    {
      "epoch": 0.08023285244241964,
      "grad_norm": 2.326428174972534,
      "learning_rate": 1.995441414570283e-05,
      "loss": 0.688,
      "step": 634
    },
    {
      "epoch": 0.0803594026828651,
      "grad_norm": 1.5409183502197266,
      "learning_rate": 1.9954243062067414e-05,
      "loss": 0.7318,
      "step": 635
    },
    {
      "epoch": 0.08048595292331055,
      "grad_norm": 1.9967628717422485,
      "learning_rate": 1.9954071658730875e-05,
      "loss": 0.7791,
      "step": 636
    },
    {
      "epoch": 0.08061250316375601,
      "grad_norm": 1.6326056718826294,
      "learning_rate": 1.9953899935698726e-05,
      "loss": 0.7135,
      "step": 637
    },
    {
      "epoch": 0.08073905340420147,
      "grad_norm": 1.5500733852386475,
      "learning_rate": 1.995372789297648e-05,
      "loss": 0.7292,
      "step": 638
    },
    {
      "epoch": 0.08086560364464693,
      "grad_norm": 1.4537745714187622,
      "learning_rate": 1.9953555530569666e-05,
      "loss": 0.698,
      "step": 639
    },
    {
      "epoch": 0.08099215388509239,
      "grad_norm": 1.5395584106445312,
      "learning_rate": 1.995338284848382e-05,
      "loss": 0.7421,
      "step": 640
    },
    {
      "epoch": 0.08111870412553784,
      "grad_norm": 1.6508269309997559,
      "learning_rate": 1.9953209846724477e-05,
      "loss": 0.7085,
      "step": 641
    },
    {
      "epoch": 0.0812452543659833,
      "grad_norm": 1.6048578023910522,
      "learning_rate": 1.9953036525297205e-05,
      "loss": 0.7461,
      "step": 642
    },
    {
      "epoch": 0.08137180460642875,
      "grad_norm": 1.8240708112716675,
      "learning_rate": 1.9952862884207566e-05,
      "loss": 0.7323,
      "step": 643
    },
    {
      "epoch": 0.0814983548468742,
      "grad_norm": 1.7071568965911865,
      "learning_rate": 1.995268892346114e-05,
      "loss": 0.7287,
      "step": 644
    },
    {
      "epoch": 0.08162490508731966,
      "grad_norm": 1.5078516006469727,
      "learning_rate": 1.9952514643063508e-05,
      "loss": 0.7138,
      "step": 645
    },
    {
      "epoch": 0.08175145532776512,
      "grad_norm": 1.7329046726226807,
      "learning_rate": 1.9952340043020273e-05,
      "loss": 0.7897,
      "step": 646
    },
    {
      "epoch": 0.08187800556821058,
      "grad_norm": 1.7746914625167847,
      "learning_rate": 1.9952165123337043e-05,
      "loss": 0.7079,
      "step": 647
    },
    {
      "epoch": 0.08200455580865604,
      "grad_norm": 1.4582864046096802,
      "learning_rate": 1.9951989884019428e-05,
      "loss": 0.7321,
      "step": 648
    },
    {
      "epoch": 0.0821311060491015,
      "grad_norm": 1.894044280052185,
      "learning_rate": 1.9951814325073067e-05,
      "loss": 0.7574,
      "step": 649
    },
    {
      "epoch": 0.08225765628954695,
      "grad_norm": 1.4459939002990723,
      "learning_rate": 1.995163844650359e-05,
      "loss": 0.6766,
      "step": 650
    },
    {
      "epoch": 0.08238420652999241,
      "grad_norm": 1.5752893686294556,
      "learning_rate": 1.995146224831665e-05,
      "loss": 0.7349,
      "step": 651
    },
    {
      "epoch": 0.08251075677043786,
      "grad_norm": 1.650840163230896,
      "learning_rate": 1.9951285730517903e-05,
      "loss": 0.7276,
      "step": 652
    },
    {
      "epoch": 0.08263730701088332,
      "grad_norm": 1.660454273223877,
      "learning_rate": 1.9951108893113025e-05,
      "loss": 0.7605,
      "step": 653
    },
    {
      "epoch": 0.08276385725132877,
      "grad_norm": 1.583268404006958,
      "learning_rate": 1.9950931736107688e-05,
      "loss": 0.7444,
      "step": 654
    },
    {
      "epoch": 0.08289040749177423,
      "grad_norm": 1.834311842918396,
      "learning_rate": 1.9950754259507584e-05,
      "loss": 0.7374,
      "step": 655
    },
    {
      "epoch": 0.08301695773221969,
      "grad_norm": 1.3522001504898071,
      "learning_rate": 1.9950576463318416e-05,
      "loss": 0.7156,
      "step": 656
    },
    {
      "epoch": 0.08314350797266515,
      "grad_norm": 2.0906245708465576,
      "learning_rate": 1.995039834754589e-05,
      "loss": 0.7368,
      "step": 657
    },
    {
      "epoch": 0.0832700582131106,
      "grad_norm": 1.6079574823379517,
      "learning_rate": 1.9950219912195727e-05,
      "loss": 0.7405,
      "step": 658
    },
    {
      "epoch": 0.08339660845355606,
      "grad_norm": 1.934524655342102,
      "learning_rate": 1.995004115727366e-05,
      "loss": 0.6673,
      "step": 659
    },
    {
      "epoch": 0.08352315869400152,
      "grad_norm": 1.738564372062683,
      "learning_rate": 1.994986208278543e-05,
      "loss": 0.743,
      "step": 660
    },
    {
      "epoch": 0.08364970893444698,
      "grad_norm": 1.4737071990966797,
      "learning_rate": 1.9949682688736788e-05,
      "loss": 0.6732,
      "step": 661
    },
    {
      "epoch": 0.08377625917489243,
      "grad_norm": 1.7974488735198975,
      "learning_rate": 1.9949502975133495e-05,
      "loss": 0.7001,
      "step": 662
    },
    {
      "epoch": 0.08390280941533788,
      "grad_norm": 1.5915745496749878,
      "learning_rate": 1.994932294198132e-05,
      "loss": 0.7092,
      "step": 663
    },
    {
      "epoch": 0.08402935965578334,
      "grad_norm": 1.4814543724060059,
      "learning_rate": 1.9949142589286053e-05,
      "loss": 0.7365,
      "step": 664
    },
    {
      "epoch": 0.0841559098962288,
      "grad_norm": 1.865308403968811,
      "learning_rate": 1.994896191705348e-05,
      "loss": 0.7111,
      "step": 665
    },
    {
      "epoch": 0.08428246013667426,
      "grad_norm": 1.6176215410232544,
      "learning_rate": 1.9948780925289403e-05,
      "loss": 0.7442,
      "step": 666
    },
    {
      "epoch": 0.08440901037711972,
      "grad_norm": 1.4959415197372437,
      "learning_rate": 1.994859961399964e-05,
      "loss": 0.7421,
      "step": 667
    },
    {
      "epoch": 0.08453556061756518,
      "grad_norm": 1.539615273475647,
      "learning_rate": 1.994841798319001e-05,
      "loss": 0.7051,
      "step": 668
    },
    {
      "epoch": 0.08466211085801063,
      "grad_norm": 1.39666748046875,
      "learning_rate": 1.9948236032866347e-05,
      "loss": 0.7265,
      "step": 669
    },
    {
      "epoch": 0.08478866109845609,
      "grad_norm": 1.4861972332000732,
      "learning_rate": 1.9948053763034497e-05,
      "loss": 0.7447,
      "step": 670
    },
    {
      "epoch": 0.08491521133890155,
      "grad_norm": 1.4765129089355469,
      "learning_rate": 1.994787117370031e-05,
      "loss": 0.7535,
      "step": 671
    },
    {
      "epoch": 0.085041761579347,
      "grad_norm": 1.4568405151367188,
      "learning_rate": 1.9947688264869658e-05,
      "loss": 0.7323,
      "step": 672
    },
    {
      "epoch": 0.08516831181979245,
      "grad_norm": 2.5113260746002197,
      "learning_rate": 1.9947505036548406e-05,
      "loss": 0.7138,
      "step": 673
    },
    {
      "epoch": 0.08529486206023791,
      "grad_norm": 1.6963971853256226,
      "learning_rate": 1.9947321488742448e-05,
      "loss": 0.7994,
      "step": 674
    },
    {
      "epoch": 0.08542141230068337,
      "grad_norm": 1.5398598909378052,
      "learning_rate": 1.9947137621457667e-05,
      "loss": 0.7428,
      "step": 675
    },
    {
      "epoch": 0.08554796254112883,
      "grad_norm": 1.5029847621917725,
      "learning_rate": 1.9946953434699977e-05,
      "loss": 0.7693,
      "step": 676
    },
    {
      "epoch": 0.08567451278157429,
      "grad_norm": 1.5882307291030884,
      "learning_rate": 1.9946768928475297e-05,
      "loss": 0.7488,
      "step": 677
    },
    {
      "epoch": 0.08580106302201974,
      "grad_norm": 1.590442419052124,
      "learning_rate": 1.994658410278954e-05,
      "loss": 0.6972,
      "step": 678
    },
    {
      "epoch": 0.0859276132624652,
      "grad_norm": 1.4750480651855469,
      "learning_rate": 1.9946398957648654e-05,
      "loss": 0.7161,
      "step": 679
    },
    {
      "epoch": 0.08605416350291066,
      "grad_norm": 2.771658182144165,
      "learning_rate": 1.994621349305858e-05,
      "loss": 0.7763,
      "step": 680
    },
    {
      "epoch": 0.08618071374335612,
      "grad_norm": 1.5032761096954346,
      "learning_rate": 1.994602770902528e-05,
      "loss": 0.801,
      "step": 681
    },
    {
      "epoch": 0.08630726398380156,
      "grad_norm": 1.444166660308838,
      "learning_rate": 1.994584160555471e-05,
      "loss": 0.7037,
      "step": 682
    },
    {
      "epoch": 0.08643381422424702,
      "grad_norm": 1.5051970481872559,
      "learning_rate": 1.9945655182652858e-05,
      "loss": 0.7626,
      "step": 683
    },
    {
      "epoch": 0.08656036446469248,
      "grad_norm": 1.4996848106384277,
      "learning_rate": 1.9945468440325703e-05,
      "loss": 0.7316,
      "step": 684
    },
    {
      "epoch": 0.08668691470513794,
      "grad_norm": 1.834889531135559,
      "learning_rate": 1.994528137857925e-05,
      "loss": 0.7119,
      "step": 685
    },
    {
      "epoch": 0.0868134649455834,
      "grad_norm": 1.550125002861023,
      "learning_rate": 1.99450939974195e-05,
      "loss": 0.7314,
      "step": 686
    },
    {
      "epoch": 0.08694001518602885,
      "grad_norm": 1.788459300994873,
      "learning_rate": 1.9944906296852477e-05,
      "loss": 0.7464,
      "step": 687
    },
    {
      "epoch": 0.08706656542647431,
      "grad_norm": 4.154918670654297,
      "learning_rate": 1.9944718276884206e-05,
      "loss": 0.7715,
      "step": 688
    },
    {
      "epoch": 0.08719311566691977,
      "grad_norm": 1.6998522281646729,
      "learning_rate": 1.9944529937520726e-05,
      "loss": 0.804,
      "step": 689
    },
    {
      "epoch": 0.08731966590736523,
      "grad_norm": 2.181408405303955,
      "learning_rate": 1.9944341278768088e-05,
      "loss": 0.7447,
      "step": 690
    },
    {
      "epoch": 0.08744621614781069,
      "grad_norm": 1.5223848819732666,
      "learning_rate": 1.994415230063235e-05,
      "loss": 0.7665,
      "step": 691
    },
    {
      "epoch": 0.08757276638825613,
      "grad_norm": 5.318248271942139,
      "learning_rate": 1.994396300311958e-05,
      "loss": 0.7358,
      "step": 692
    },
    {
      "epoch": 0.08769931662870159,
      "grad_norm": 2.2478368282318115,
      "learning_rate": 1.994377338623586e-05,
      "loss": 0.7716,
      "step": 693
    },
    {
      "epoch": 0.08782586686914705,
      "grad_norm": 2.879570960998535,
      "learning_rate": 1.9943583449987277e-05,
      "loss": 0.7617,
      "step": 694
    },
    {
      "epoch": 0.0879524171095925,
      "grad_norm": 2.532179117202759,
      "learning_rate": 1.9943393194379933e-05,
      "loss": 0.7595,
      "step": 695
    },
    {
      "epoch": 0.08807896735003796,
      "grad_norm": 1.974387526512146,
      "learning_rate": 1.9943202619419935e-05,
      "loss": 0.7134,
      "step": 696
    },
    {
      "epoch": 0.08820551759048342,
      "grad_norm": 2.3433732986450195,
      "learning_rate": 1.9943011725113416e-05,
      "loss": 0.7068,
      "step": 697
    },
    {
      "epoch": 0.08833206783092888,
      "grad_norm": 1.8994289636611938,
      "learning_rate": 1.994282051146649e-05,
      "loss": 0.6816,
      "step": 698
    },
    {
      "epoch": 0.08845861807137434,
      "grad_norm": 2.1026957035064697,
      "learning_rate": 1.994262897848531e-05,
      "loss": 0.791,
      "step": 699
    },
    {
      "epoch": 0.0885851683118198,
      "grad_norm": 1.9434138536453247,
      "learning_rate": 1.9942437126176022e-05,
      "loss": 0.7144,
      "step": 700
    },
    {
      "epoch": 0.08871171855226524,
      "grad_norm": 1.6989952325820923,
      "learning_rate": 1.994224495454479e-05,
      "loss": 0.742,
      "step": 701
    },
    {
      "epoch": 0.0888382687927107,
      "grad_norm": 2.5013272762298584,
      "learning_rate": 1.9942052463597787e-05,
      "loss": 0.7496,
      "step": 702
    },
    {
      "epoch": 0.08896481903315616,
      "grad_norm": 1.5888795852661133,
      "learning_rate": 1.994185965334119e-05,
      "loss": 0.758,
      "step": 703
    },
    {
      "epoch": 0.08909136927360162,
      "grad_norm": 2.4282405376434326,
      "learning_rate": 1.9941666523781196e-05,
      "loss": 0.7066,
      "step": 704
    },
    {
      "epoch": 0.08921791951404708,
      "grad_norm": 1.9977871179580688,
      "learning_rate": 1.994147307492401e-05,
      "loss": 0.7629,
      "step": 705
    },
    {
      "epoch": 0.08934446975449253,
      "grad_norm": 2.509422540664673,
      "learning_rate": 1.994127930677584e-05,
      "loss": 0.6932,
      "step": 706
    },
    {
      "epoch": 0.08947101999493799,
      "grad_norm": 2.2290127277374268,
      "learning_rate": 1.9941085219342912e-05,
      "loss": 0.7225,
      "step": 707
    },
    {
      "epoch": 0.08959757023538345,
      "grad_norm": 2.1876792907714844,
      "learning_rate": 1.9940890812631456e-05,
      "loss": 0.7409,
      "step": 708
    },
    {
      "epoch": 0.08972412047582891,
      "grad_norm": 2.349973201751709,
      "learning_rate": 1.9940696086647723e-05,
      "loss": 0.738,
      "step": 709
    },
    {
      "epoch": 0.08985067071627437,
      "grad_norm": 1.7748953104019165,
      "learning_rate": 1.994050104139796e-05,
      "loss": 0.7235,
      "step": 710
    },
    {
      "epoch": 0.08997722095671981,
      "grad_norm": 1.9518067836761475,
      "learning_rate": 1.9940305676888433e-05,
      "loss": 0.7388,
      "step": 711
    },
    {
      "epoch": 0.09010377119716527,
      "grad_norm": 1.7122337818145752,
      "learning_rate": 1.994010999312542e-05,
      "loss": 0.7484,
      "step": 712
    },
    {
      "epoch": 0.09023032143761073,
      "grad_norm": 1.8457072973251343,
      "learning_rate": 1.9939913990115197e-05,
      "loss": 0.7232,
      "step": 713
    },
    {
      "epoch": 0.09035687167805619,
      "grad_norm": 1.7313016653060913,
      "learning_rate": 1.993971766786407e-05,
      "loss": 0.7382,
      "step": 714
    },
    {
      "epoch": 0.09048342191850164,
      "grad_norm": 1.559822916984558,
      "learning_rate": 1.9939521026378343e-05,
      "loss": 0.7082,
      "step": 715
    },
    {
      "epoch": 0.0906099721589471,
      "grad_norm": 1.8260371685028076,
      "learning_rate": 1.993932406566432e-05,
      "loss": 0.7305,
      "step": 716
    },
    {
      "epoch": 0.09073652239939256,
      "grad_norm": 1.3476608991622925,
      "learning_rate": 1.993912678572834e-05,
      "loss": 0.6985,
      "step": 717
    },
    {
      "epoch": 0.09086307263983802,
      "grad_norm": 1.5482807159423828,
      "learning_rate": 1.9938929186576736e-05,
      "loss": 0.7025,
      "step": 718
    },
    {
      "epoch": 0.09098962288028348,
      "grad_norm": 1.5975584983825684,
      "learning_rate": 1.9938731268215847e-05,
      "loss": 0.6928,
      "step": 719
    },
    {
      "epoch": 0.09111617312072894,
      "grad_norm": 1.337191104888916,
      "learning_rate": 1.9938533030652034e-05,
      "loss": 0.6855,
      "step": 720
    },
    {
      "epoch": 0.09124272336117438,
      "grad_norm": 2.256329298019409,
      "learning_rate": 1.9938334473891666e-05,
      "loss": 0.8133,
      "step": 721
    },
    {
      "epoch": 0.09136927360161984,
      "grad_norm": 1.3841300010681152,
      "learning_rate": 1.9938135597941124e-05,
      "loss": 0.721,
      "step": 722
    },
    {
      "epoch": 0.0914958238420653,
      "grad_norm": 1.4175554513931274,
      "learning_rate": 1.9937936402806785e-05,
      "loss": 0.7352,
      "step": 723
    },
    {
      "epoch": 0.09162237408251075,
      "grad_norm": 1.3515031337738037,
      "learning_rate": 1.9937736888495053e-05,
      "loss": 0.69,
      "step": 724
    },
    {
      "epoch": 0.09174892432295621,
      "grad_norm": 1.451002597808838,
      "learning_rate": 1.993753705501233e-05,
      "loss": 0.7499,
      "step": 725
    },
    {
      "epoch": 0.09187547456340167,
      "grad_norm": 1.6199744939804077,
      "learning_rate": 1.9937336902365044e-05,
      "loss": 0.7865,
      "step": 726
    },
    {
      "epoch": 0.09200202480384713,
      "grad_norm": 1.5397891998291016,
      "learning_rate": 1.9937136430559615e-05,
      "loss": 0.7277,
      "step": 727
    },
    {
      "epoch": 0.09212857504429259,
      "grad_norm": 1.585427165031433,
      "learning_rate": 1.9936935639602484e-05,
      "loss": 0.755,
      "step": 728
    },
    {
      "epoch": 0.09225512528473805,
      "grad_norm": 1.413102626800537,
      "learning_rate": 1.99367345295001e-05,
      "loss": 0.6863,
      "step": 729
    },
    {
      "epoch": 0.0923816755251835,
      "grad_norm": 1.8748239278793335,
      "learning_rate": 1.9936533100258924e-05,
      "loss": 0.7046,
      "step": 730
    },
    {
      "epoch": 0.09250822576562895,
      "grad_norm": 1.5159116983413696,
      "learning_rate": 1.9936331351885424e-05,
      "loss": 0.7225,
      "step": 731
    },
    {
      "epoch": 0.0926347760060744,
      "grad_norm": 1.5068730115890503,
      "learning_rate": 1.9936129284386077e-05,
      "loss": 0.7372,
      "step": 732
    },
    {
      "epoch": 0.09276132624651987,
      "grad_norm": 1.6879994869232178,
      "learning_rate": 1.9935926897767373e-05,
      "loss": 0.7612,
      "step": 733
    },
    {
      "epoch": 0.09288787648696532,
      "grad_norm": 1.5808929204940796,
      "learning_rate": 1.993572419203582e-05,
      "loss": 0.7179,
      "step": 734
    },
    {
      "epoch": 0.09301442672741078,
      "grad_norm": 1.5718052387237549,
      "learning_rate": 1.9935521167197917e-05,
      "loss": 0.739,
      "step": 735
    },
    {
      "epoch": 0.09314097696785624,
      "grad_norm": 1.4241702556610107,
      "learning_rate": 1.9935317823260187e-05,
      "loss": 0.7234,
      "step": 736
    },
    {
      "epoch": 0.0932675272083017,
      "grad_norm": 1.4086230993270874,
      "learning_rate": 1.993511416022917e-05,
      "loss": 0.7281,
      "step": 737
    },
    {
      "epoch": 0.09339407744874716,
      "grad_norm": 2.8655402660369873,
      "learning_rate": 1.9934910178111396e-05,
      "loss": 0.7505,
      "step": 738
    },
    {
      "epoch": 0.09352062768919261,
      "grad_norm": 1.601930022239685,
      "learning_rate": 1.9934705876913425e-05,
      "loss": 0.735,
      "step": 739
    },
    {
      "epoch": 0.09364717792963807,
      "grad_norm": 1.4804420471191406,
      "learning_rate": 1.993450125664181e-05,
      "loss": 0.7417,
      "step": 740
    },
    {
      "epoch": 0.09377372817008352,
      "grad_norm": 1.38815176486969,
      "learning_rate": 1.9934296317303132e-05,
      "loss": 0.6898,
      "step": 741
    },
    {
      "epoch": 0.09390027841052898,
      "grad_norm": 1.518764853477478,
      "learning_rate": 1.9934091058903967e-05,
      "loss": 0.7342,
      "step": 742
    },
    {
      "epoch": 0.09402682865097443,
      "grad_norm": 1.5901776552200317,
      "learning_rate": 1.9933885481450906e-05,
      "loss": 0.7241,
      "step": 743
    },
    {
      "epoch": 0.09415337889141989,
      "grad_norm": 1.4839508533477783,
      "learning_rate": 1.993367958495056e-05,
      "loss": 0.7215,
      "step": 744
    },
    {
      "epoch": 0.09427992913186535,
      "grad_norm": 1.557447075843811,
      "learning_rate": 1.993347336940953e-05,
      "loss": 0.6972,
      "step": 745
    },
    {
      "epoch": 0.09440647937231081,
      "grad_norm": 1.8715488910675049,
      "learning_rate": 1.9933266834834448e-05,
      "loss": 0.7154,
      "step": 746
    },
    {
      "epoch": 0.09453302961275627,
      "grad_norm": 1.5072920322418213,
      "learning_rate": 1.9933059981231943e-05,
      "loss": 0.7603,
      "step": 747
    },
    {
      "epoch": 0.09465957985320173,
      "grad_norm": 1.4526628255844116,
      "learning_rate": 1.9932852808608662e-05,
      "loss": 0.7162,
      "step": 748
    },
    {
      "epoch": 0.09478613009364718,
      "grad_norm": 1.6535840034484863,
      "learning_rate": 1.9932645316971255e-05,
      "loss": 0.7602,
      "step": 749
    },
    {
      "epoch": 0.09491268033409264,
      "grad_norm": 1.355115294456482,
      "learning_rate": 1.9932437506326386e-05,
      "loss": 0.7085,
      "step": 750
    },
    {
      "epoch": 0.09503923057453809,
      "grad_norm": 3.139521598815918,
      "learning_rate": 1.9932229376680733e-05,
      "loss": 0.7201,
      "step": 751
    },
    {
      "epoch": 0.09516578081498354,
      "grad_norm": 2.028580665588379,
      "learning_rate": 1.9932020928040976e-05,
      "loss": 0.7036,
      "step": 752
    },
    {
      "epoch": 0.095292331055429,
      "grad_norm": 2.004106044769287,
      "learning_rate": 1.9931812160413817e-05,
      "loss": 0.7235,
      "step": 753
    },
    {
      "epoch": 0.09541888129587446,
      "grad_norm": 1.385477900505066,
      "learning_rate": 1.993160307380595e-05,
      "loss": 0.7276,
      "step": 754
    },
    {
      "epoch": 0.09554543153631992,
      "grad_norm": 2.6343648433685303,
      "learning_rate": 1.99313936682241e-05,
      "loss": 0.7133,
      "step": 755
    },
    {
      "epoch": 0.09567198177676538,
      "grad_norm": 1.490881085395813,
      "learning_rate": 1.9931183943674992e-05,
      "loss": 0.7289,
      "step": 756
    },
    {
      "epoch": 0.09579853201721084,
      "grad_norm": 1.603025197982788,
      "learning_rate": 1.9930973900165354e-05,
      "loss": 0.78,
      "step": 757
    },
    {
      "epoch": 0.0959250822576563,
      "grad_norm": 1.540832757949829,
      "learning_rate": 1.9930763537701937e-05,
      "loss": 0.7362,
      "step": 758
    },
    {
      "epoch": 0.09605163249810175,
      "grad_norm": 1.4202687740325928,
      "learning_rate": 1.99305528562915e-05,
      "loss": 0.7169,
      "step": 759
    },
    {
      "epoch": 0.0961781827385472,
      "grad_norm": 1.5323998928070068,
      "learning_rate": 1.9930341855940803e-05,
      "loss": 0.7862,
      "step": 760
    },
    {
      "epoch": 0.09630473297899265,
      "grad_norm": 1.5710397958755493,
      "learning_rate": 1.9930130536656626e-05,
      "loss": 0.746,
      "step": 761
    },
    {
      "epoch": 0.09643128321943811,
      "grad_norm": 2.0334360599517822,
      "learning_rate": 1.992991889844576e-05,
      "loss": 0.7505,
      "step": 762
    },
    {
      "epoch": 0.09655783345988357,
      "grad_norm": 1.4783095121383667,
      "learning_rate": 1.9929706941314995e-05,
      "loss": 0.7054,
      "step": 763
    },
    {
      "epoch": 0.09668438370032903,
      "grad_norm": 1.62554132938385,
      "learning_rate": 1.9929494665271142e-05,
      "loss": 0.7028,
      "step": 764
    },
    {
      "epoch": 0.09681093394077449,
      "grad_norm": 1.539354920387268,
      "learning_rate": 1.9929282070321017e-05,
      "loss": 0.762,
      "step": 765
    },
    {
      "epoch": 0.09693748418121995,
      "grad_norm": 1.3176792860031128,
      "learning_rate": 1.9929069156471455e-05,
      "loss": 0.7201,
      "step": 766
    },
    {
      "epoch": 0.0970640344216654,
      "grad_norm": 1.7022274732589722,
      "learning_rate": 1.9928855923729282e-05,
      "loss": 0.759,
      "step": 767
    },
    {
      "epoch": 0.09719058466211086,
      "grad_norm": 1.7778435945510864,
      "learning_rate": 1.9928642372101356e-05,
      "loss": 0.7269,
      "step": 768
    },
    {
      "epoch": 0.09731713490255632,
      "grad_norm": 1.5161689519882202,
      "learning_rate": 1.992842850159453e-05,
      "loss": 0.7318,
      "step": 769
    },
    {
      "epoch": 0.09744368514300177,
      "grad_norm": 1.64460027217865,
      "learning_rate": 1.9928214312215675e-05,
      "loss": 0.729,
      "step": 770
    },
    {
      "epoch": 0.09757023538344722,
      "grad_norm": 1.5539460182189941,
      "learning_rate": 1.9927999803971676e-05,
      "loss": 0.7367,
      "step": 771
    },
    {
      "epoch": 0.09769678562389268,
      "grad_norm": 1.677174687385559,
      "learning_rate": 1.9927784976869416e-05,
      "loss": 0.743,
      "step": 772
    },
    {
      "epoch": 0.09782333586433814,
      "grad_norm": 1.4366185665130615,
      "learning_rate": 1.9927569830915788e-05,
      "loss": 0.6795,
      "step": 773
    },
    {
      "epoch": 0.0979498861047836,
      "grad_norm": 1.7313652038574219,
      "learning_rate": 1.992735436611772e-05,
      "loss": 0.7968,
      "step": 774
    },
    {
      "epoch": 0.09807643634522906,
      "grad_norm": 1.434493899345398,
      "learning_rate": 1.9927138582482116e-05,
      "loss": 0.7269,
      "step": 775
    },
    {
      "epoch": 0.09820298658567451,
      "grad_norm": 1.6430329084396362,
      "learning_rate": 1.992692248001591e-05,
      "loss": 0.7476,
      "step": 776
    },
    {
      "epoch": 0.09832953682611997,
      "grad_norm": 1.657827615737915,
      "learning_rate": 1.9926706058726048e-05,
      "loss": 0.7735,
      "step": 777
    },
    {
      "epoch": 0.09845608706656543,
      "grad_norm": 1.5162452459335327,
      "learning_rate": 1.9926489318619473e-05,
      "loss": 0.7433,
      "step": 778
    },
    {
      "epoch": 0.09858263730701089,
      "grad_norm": 4.805769920349121,
      "learning_rate": 1.9926272259703154e-05,
      "loss": 0.7911,
      "step": 779
    },
    {
      "epoch": 0.09870918754745633,
      "grad_norm": 1.5198724269866943,
      "learning_rate": 1.9926054881984057e-05,
      "loss": 0.7186,
      "step": 780
    },
    {
      "epoch": 0.09883573778790179,
      "grad_norm": 1.8706762790679932,
      "learning_rate": 1.9925837185469166e-05,
      "loss": 0.7376,
      "step": 781
    },
    {
      "epoch": 0.09896228802834725,
      "grad_norm": 1.4808868169784546,
      "learning_rate": 1.992561917016547e-05,
      "loss": 0.7802,
      "step": 782
    },
    {
      "epoch": 0.09908883826879271,
      "grad_norm": 1.5038726329803467,
      "learning_rate": 1.992540083607997e-05,
      "loss": 0.739,
      "step": 783
    },
    {
      "epoch": 0.09921538850923817,
      "grad_norm": 1.5597747564315796,
      "learning_rate": 1.992518218321969e-05,
      "loss": 0.7421,
      "step": 784
    },
    {
      "epoch": 0.09934193874968363,
      "grad_norm": 1.4369347095489502,
      "learning_rate": 1.9924963211591638e-05,
      "loss": 0.7627,
      "step": 785
    },
    {
      "epoch": 0.09946848899012908,
      "grad_norm": 1.7528847455978394,
      "learning_rate": 1.9924743921202852e-05,
      "loss": 0.7411,
      "step": 786
    },
    {
      "epoch": 0.09959503923057454,
      "grad_norm": 1.5660029649734497,
      "learning_rate": 1.992452431206038e-05,
      "loss": 0.7455,
      "step": 787
    },
    {
      "epoch": 0.09972158947102,
      "grad_norm": 1.4422589540481567,
      "learning_rate": 1.9924304384171264e-05,
      "loss": 0.731,
      "step": 788
    },
    {
      "epoch": 0.09984813971146546,
      "grad_norm": 1.4866676330566406,
      "learning_rate": 1.992408413754258e-05,
      "loss": 0.7486,
      "step": 789
    },
    {
      "epoch": 0.0999746899519109,
      "grad_norm": 1.4360122680664062,
      "learning_rate": 1.992386357218139e-05,
      "loss": 0.7647,
      "step": 790
    },
    {
      "epoch": 0.10010124019235636,
      "grad_norm": 1.5635185241699219,
      "learning_rate": 1.9923642688094786e-05,
      "loss": 0.7205,
      "step": 791
    },
    {
      "epoch": 0.10022779043280182,
      "grad_norm": 1.562855839729309,
      "learning_rate": 1.9923421485289862e-05,
      "loss": 0.735,
      "step": 792
    },
    {
      "epoch": 0.10035434067324728,
      "grad_norm": 1.3436577320098877,
      "learning_rate": 1.9923199963773718e-05,
      "loss": 0.7187,
      "step": 793
    },
    {
      "epoch": 0.10048089091369274,
      "grad_norm": 1.4289807081222534,
      "learning_rate": 1.992297812355347e-05,
      "loss": 0.7172,
      "step": 794
    },
    {
      "epoch": 0.1006074411541382,
      "grad_norm": 1.5727053880691528,
      "learning_rate": 1.9922755964636247e-05,
      "loss": 0.7445,
      "step": 795
    },
    {
      "epoch": 0.10073399139458365,
      "grad_norm": 1.450014352798462,
      "learning_rate": 1.9922533487029177e-05,
      "loss": 0.7104,
      "step": 796
    },
    {
      "epoch": 0.10086054163502911,
      "grad_norm": 1.8131129741668701,
      "learning_rate": 1.992231069073941e-05,
      "loss": 0.7609,
      "step": 797
    },
    {
      "epoch": 0.10098709187547457,
      "grad_norm": 1.7955565452575684,
      "learning_rate": 1.99220875757741e-05,
      "loss": 0.7441,
      "step": 798
    },
    {
      "epoch": 0.10111364211592003,
      "grad_norm": 1.4657824039459229,
      "learning_rate": 1.9921864142140414e-05,
      "loss": 0.7459,
      "step": 799
    },
    {
      "epoch": 0.10124019235636547,
      "grad_norm": 1.9435704946517944,
      "learning_rate": 1.992164038984553e-05,
      "loss": 0.7473,
      "step": 800
    },
    {
      "epoch": 0.10136674259681093,
      "grad_norm": 2.0092828273773193,
      "learning_rate": 1.9921416318896628e-05,
      "loss": 0.7524,
      "step": 801
    },
    {
      "epoch": 0.10149329283725639,
      "grad_norm": 1.660022497177124,
      "learning_rate": 1.992119192930091e-05,
      "loss": 0.7895,
      "step": 802
    },
    {
      "epoch": 0.10161984307770185,
      "grad_norm": 1.9563835859298706,
      "learning_rate": 1.9920967221065578e-05,
      "loss": 0.7431,
      "step": 803
    },
    {
      "epoch": 0.1017463933181473,
      "grad_norm": 1.4714224338531494,
      "learning_rate": 1.9920742194197853e-05,
      "loss": 0.7675,
      "step": 804
    },
    {
      "epoch": 0.10187294355859276,
      "grad_norm": 1.5493030548095703,
      "learning_rate": 1.9920516848704964e-05,
      "loss": 0.7461,
      "step": 805
    },
    {
      "epoch": 0.10199949379903822,
      "grad_norm": 1.5526721477508545,
      "learning_rate": 1.9920291184594143e-05,
      "loss": 0.7013,
      "step": 806
    },
    {
      "epoch": 0.10212604403948368,
      "grad_norm": 1.4344581365585327,
      "learning_rate": 1.992006520187264e-05,
      "loss": 0.6673,
      "step": 807
    },
    {
      "epoch": 0.10225259427992914,
      "grad_norm": 1.5164128541946411,
      "learning_rate": 1.9919838900547716e-05,
      "loss": 0.7153,
      "step": 808
    },
    {
      "epoch": 0.10237914452037458,
      "grad_norm": 1.6903530359268188,
      "learning_rate": 1.991961228062663e-05,
      "loss": 0.7471,
      "step": 809
    },
    {
      "epoch": 0.10250569476082004,
      "grad_norm": 1.8673560619354248,
      "learning_rate": 1.9919385342116672e-05,
      "loss": 0.7078,
      "step": 810
    },
    {
      "epoch": 0.1026322450012655,
      "grad_norm": 1.7774335145950317,
      "learning_rate": 1.9919158085025122e-05,
      "loss": 0.7083,
      "step": 811
    },
    {
      "epoch": 0.10275879524171096,
      "grad_norm": 1.6315817832946777,
      "learning_rate": 1.9918930509359287e-05,
      "loss": 0.7488,
      "step": 812
    },
    {
      "epoch": 0.10288534548215642,
      "grad_norm": 1.3469077348709106,
      "learning_rate": 1.9918702615126468e-05,
      "loss": 0.747,
      "step": 813
    },
    {
      "epoch": 0.10301189572260187,
      "grad_norm": 1.4092113971710205,
      "learning_rate": 1.9918474402333987e-05,
      "loss": 0.7215,
      "step": 814
    },
    {
      "epoch": 0.10313844596304733,
      "grad_norm": 1.5088467597961426,
      "learning_rate": 1.9918245870989175e-05,
      "loss": 0.7424,
      "step": 815
    },
    {
      "epoch": 0.10326499620349279,
      "grad_norm": 1.484425663948059,
      "learning_rate": 1.991801702109937e-05,
      "loss": 0.7256,
      "step": 816
    },
    {
      "epoch": 0.10339154644393825,
      "grad_norm": 1.5516234636306763,
      "learning_rate": 1.9917787852671922e-05,
      "loss": 0.6942,
      "step": 817
    },
    {
      "epoch": 0.1035180966843837,
      "grad_norm": 1.4225513935089111,
      "learning_rate": 1.9917558365714192e-05,
      "loss": 0.7357,
      "step": 818
    },
    {
      "epoch": 0.10364464692482915,
      "grad_norm": 1.5368125438690186,
      "learning_rate": 1.9917328560233547e-05,
      "loss": 0.719,
      "step": 819
    },
    {
      "epoch": 0.10377119716527461,
      "grad_norm": 1.536887764930725,
      "learning_rate": 1.9917098436237378e-05,
      "loss": 0.7321,
      "step": 820
    },
    {
      "epoch": 0.10389774740572007,
      "grad_norm": 1.358853816986084,
      "learning_rate": 1.9916867993733066e-05,
      "loss": 0.6968,
      "step": 821
    },
    {
      "epoch": 0.10402429764616553,
      "grad_norm": 1.5096031427383423,
      "learning_rate": 1.9916637232728014e-05,
      "loss": 0.7186,
      "step": 822
    },
    {
      "epoch": 0.10415084788661098,
      "grad_norm": 1.462568759918213,
      "learning_rate": 1.9916406153229633e-05,
      "loss": 0.7188,
      "step": 823
    },
    {
      "epoch": 0.10427739812705644,
      "grad_norm": 1.5053701400756836,
      "learning_rate": 1.991617475524535e-05,
      "loss": 0.7608,
      "step": 824
    },
    {
      "epoch": 0.1044039483675019,
      "grad_norm": 1.3465765714645386,
      "learning_rate": 1.9915943038782594e-05,
      "loss": 0.7096,
      "step": 825
    },
    {
      "epoch": 0.10453049860794736,
      "grad_norm": 1.4358009099960327,
      "learning_rate": 1.9915711003848802e-05,
      "loss": 0.7133,
      "step": 826
    },
    {
      "epoch": 0.10465704884839282,
      "grad_norm": 1.480311632156372,
      "learning_rate": 1.991547865045143e-05,
      "loss": 0.7686,
      "step": 827
    },
    {
      "epoch": 0.10478359908883828,
      "grad_norm": 1.4674137830734253,
      "learning_rate": 1.991524597859794e-05,
      "loss": 0.7548,
      "step": 828
    },
    {
      "epoch": 0.10491014932928372,
      "grad_norm": 1.596968412399292,
      "learning_rate": 1.9915012988295808e-05,
      "loss": 0.7534,
      "step": 829
    },
    {
      "epoch": 0.10503669956972918,
      "grad_norm": 1.343299388885498,
      "learning_rate": 1.9914779679552514e-05,
      "loss": 0.7109,
      "step": 830
    },
    {
      "epoch": 0.10516324981017464,
      "grad_norm": 1.4922771453857422,
      "learning_rate": 1.991454605237555e-05,
      "loss": 0.773,
      "step": 831
    },
    {
      "epoch": 0.1052898000506201,
      "grad_norm": 1.363261103630066,
      "learning_rate": 1.991431210677242e-05,
      "loss": 0.7235,
      "step": 832
    },
    {
      "epoch": 0.10541635029106555,
      "grad_norm": 1.499193787574768,
      "learning_rate": 1.9914077842750643e-05,
      "loss": 0.7285,
      "step": 833
    },
    {
      "epoch": 0.10554290053151101,
      "grad_norm": 1.5785777568817139,
      "learning_rate": 1.9913843260317734e-05,
      "loss": 0.7438,
      "step": 834
    },
    {
      "epoch": 0.10566945077195647,
      "grad_norm": 1.4620304107666016,
      "learning_rate": 1.9913608359481237e-05,
      "loss": 0.7249,
      "step": 835
    },
    {
      "epoch": 0.10579600101240193,
      "grad_norm": 1.4176222085952759,
      "learning_rate": 1.9913373140248687e-05,
      "loss": 0.7302,
      "step": 836
    },
    {
      "epoch": 0.10592255125284739,
      "grad_norm": 1.4623842239379883,
      "learning_rate": 1.991313760262764e-05,
      "loss": 0.7381,
      "step": 837
    },
    {
      "epoch": 0.10604910149329284,
      "grad_norm": 1.475247859954834,
      "learning_rate": 1.9912901746625665e-05,
      "loss": 0.6964,
      "step": 838
    },
    {
      "epoch": 0.10617565173373829,
      "grad_norm": 1.6196223497390747,
      "learning_rate": 1.9912665572250337e-05,
      "loss": 0.7773,
      "step": 839
    },
    {
      "epoch": 0.10630220197418375,
      "grad_norm": 1.4485702514648438,
      "learning_rate": 1.991242907950924e-05,
      "loss": 0.7017,
      "step": 840
    },
    {
      "epoch": 0.1064287522146292,
      "grad_norm": 1.696723461151123,
      "learning_rate": 1.991219226840997e-05,
      "loss": 0.743,
      "step": 841
    },
    {
      "epoch": 0.10655530245507466,
      "grad_norm": 1.4031288623809814,
      "learning_rate": 1.9911955138960128e-05,
      "loss": 0.7404,
      "step": 842
    },
    {
      "epoch": 0.10668185269552012,
      "grad_norm": 1.4992191791534424,
      "learning_rate": 1.9911717691167336e-05,
      "loss": 0.7463,
      "step": 843
    },
    {
      "epoch": 0.10680840293596558,
      "grad_norm": 1.404259204864502,
      "learning_rate": 1.9911479925039217e-05,
      "loss": 0.7195,
      "step": 844
    },
    {
      "epoch": 0.10693495317641104,
      "grad_norm": 1.4543205499649048,
      "learning_rate": 1.9911241840583407e-05,
      "loss": 0.7595,
      "step": 845
    },
    {
      "epoch": 0.1070615034168565,
      "grad_norm": 1.5569204092025757,
      "learning_rate": 1.9911003437807554e-05,
      "loss": 0.7401,
      "step": 846
    },
    {
      "epoch": 0.10718805365730195,
      "grad_norm": 1.9021282196044922,
      "learning_rate": 1.9910764716719314e-05,
      "loss": 0.7292,
      "step": 847
    },
    {
      "epoch": 0.10731460389774741,
      "grad_norm": 1.3589918613433838,
      "learning_rate": 1.991052567732636e-05,
      "loss": 0.7216,
      "step": 848
    },
    {
      "epoch": 0.10744115413819286,
      "grad_norm": 1.485809326171875,
      "learning_rate": 1.9910286319636356e-05,
      "loss": 0.7279,
      "step": 849
    },
    {
      "epoch": 0.10756770437863832,
      "grad_norm": 1.499064564704895,
      "learning_rate": 1.9910046643657003e-05,
      "loss": 0.6934,
      "step": 850
    },
    {
      "epoch": 0.10769425461908377,
      "grad_norm": 1.6317020654678345,
      "learning_rate": 1.9909806649395986e-05,
      "loss": 0.7059,
      "step": 851
    },
    {
      "epoch": 0.10782080485952923,
      "grad_norm": 1.4445220232009888,
      "learning_rate": 1.9909566336861026e-05,
      "loss": 0.7373,
      "step": 852
    },
    {
      "epoch": 0.10794735509997469,
      "grad_norm": 1.4202994108200073,
      "learning_rate": 1.990932570605983e-05,
      "loss": 0.695,
      "step": 853
    },
    {
      "epoch": 0.10807390534042015,
      "grad_norm": 1.5110903978347778,
      "learning_rate": 1.9909084757000133e-05,
      "loss": 0.724,
      "step": 854
    },
    {
      "epoch": 0.1082004555808656,
      "grad_norm": 1.5666476488113403,
      "learning_rate": 1.9908843489689675e-05,
      "loss": 0.7657,
      "step": 855
    },
    {
      "epoch": 0.10832700582131106,
      "grad_norm": 1.443267583847046,
      "learning_rate": 1.99086019041362e-05,
      "loss": 0.7002,
      "step": 856
    },
    {
      "epoch": 0.10845355606175652,
      "grad_norm": 1.3765579462051392,
      "learning_rate": 1.9908360000347468e-05,
      "loss": 0.6709,
      "step": 857
    },
    {
      "epoch": 0.10858010630220197,
      "grad_norm": 1.8026766777038574,
      "learning_rate": 1.9908117778331247e-05,
      "loss": 0.7591,
      "step": 858
    },
    {
      "epoch": 0.10870665654264743,
      "grad_norm": 1.4030425548553467,
      "learning_rate": 1.990787523809532e-05,
      "loss": 0.7552,
      "step": 859
    },
    {
      "epoch": 0.10883320678309288,
      "grad_norm": 1.5715428590774536,
      "learning_rate": 1.9907632379647475e-05,
      "loss": 0.7428,
      "step": 860
    },
    {
      "epoch": 0.10895975702353834,
      "grad_norm": 1.4859100580215454,
      "learning_rate": 1.990738920299551e-05,
      "loss": 0.767,
      "step": 861
    },
    {
      "epoch": 0.1090863072639838,
      "grad_norm": 1.3077685832977295,
      "learning_rate": 1.990714570814724e-05,
      "loss": 0.716,
      "step": 862
    },
    {
      "epoch": 0.10921285750442926,
      "grad_norm": 1.4010231494903564,
      "learning_rate": 1.9906901895110483e-05,
      "loss": 0.7516,
      "step": 863
    },
    {
      "epoch": 0.10933940774487472,
      "grad_norm": 1.6055885553359985,
      "learning_rate": 1.9906657763893065e-05,
      "loss": 0.7454,
      "step": 864
    },
    {
      "epoch": 0.10946595798532018,
      "grad_norm": 1.3307240009307861,
      "learning_rate": 1.9906413314502834e-05,
      "loss": 0.6512,
      "step": 865
    },
    {
      "epoch": 0.10959250822576563,
      "grad_norm": 1.790787696838379,
      "learning_rate": 1.9906168546947634e-05,
      "loss": 0.7714,
      "step": 866
    },
    {
      "epoch": 0.10971905846621109,
      "grad_norm": 1.554275631904602,
      "learning_rate": 1.9905923461235333e-05,
      "loss": 0.7073,
      "step": 867
    },
    {
      "epoch": 0.10984560870665654,
      "grad_norm": 1.4901196956634521,
      "learning_rate": 1.99056780573738e-05,
      "loss": 0.7216,
      "step": 868
    },
    {
      "epoch": 0.109972158947102,
      "grad_norm": 2.202862024307251,
      "learning_rate": 1.9905432335370913e-05,
      "loss": 0.7482,
      "step": 869
    },
    {
      "epoch": 0.11009870918754745,
      "grad_norm": 1.4708445072174072,
      "learning_rate": 1.990518629523457e-05,
      "loss": 0.7397,
      "step": 870
    },
    {
      "epoch": 0.11022525942799291,
      "grad_norm": 1.6943504810333252,
      "learning_rate": 1.9904939936972667e-05,
      "loss": 0.7263,
      "step": 871
    },
    {
      "epoch": 0.11035180966843837,
      "grad_norm": 1.347771406173706,
      "learning_rate": 1.9904693260593122e-05,
      "loss": 0.7303,
      "step": 872
    },
    {
      "epoch": 0.11047835990888383,
      "grad_norm": 1.4452368021011353,
      "learning_rate": 1.990444626610385e-05,
      "loss": 0.7106,
      "step": 873
    },
    {
      "epoch": 0.11060491014932929,
      "grad_norm": 1.9836878776550293,
      "learning_rate": 1.9904198953512798e-05,
      "loss": 0.7569,
      "step": 874
    },
    {
      "epoch": 0.11073146038977474,
      "grad_norm": 1.5442403554916382,
      "learning_rate": 1.990395132282789e-05,
      "loss": 0.7761,
      "step": 875
    },
    {
      "epoch": 0.1108580106302202,
      "grad_norm": 1.3954970836639404,
      "learning_rate": 1.9903703374057094e-05,
      "loss": 0.6983,
      "step": 876
    },
    {
      "epoch": 0.11098456087066566,
      "grad_norm": 1.5577490329742432,
      "learning_rate": 1.9903455107208364e-05,
      "loss": 0.7315,
      "step": 877
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 1.3955304622650146,
      "learning_rate": 1.990320652228968e-05,
      "loss": 0.7433,
      "step": 878
    },
    {
      "epoch": 0.11123766135155656,
      "grad_norm": 1.582599401473999,
      "learning_rate": 1.9902957619309025e-05,
      "loss": 0.7426,
      "step": 879
    },
    {
      "epoch": 0.11136421159200202,
      "grad_norm": 1.6910432577133179,
      "learning_rate": 1.990270839827439e-05,
      "loss": 0.7661,
      "step": 880
    },
    {
      "epoch": 0.11149076183244748,
      "grad_norm": 1.3405661582946777,
      "learning_rate": 1.990245885919378e-05,
      "loss": 0.7125,
      "step": 881
    },
    {
      "epoch": 0.11161731207289294,
      "grad_norm": 1.5667096376419067,
      "learning_rate": 1.990220900207521e-05,
      "loss": 0.6348,
      "step": 882
    },
    {
      "epoch": 0.1117438623133384,
      "grad_norm": 2.0841429233551025,
      "learning_rate": 1.9901958826926705e-05,
      "loss": 0.7149,
      "step": 883
    },
    {
      "epoch": 0.11187041255378385,
      "grad_norm": 1.659557819366455,
      "learning_rate": 1.9901708333756302e-05,
      "loss": 0.7706,
      "step": 884
    },
    {
      "epoch": 0.11199696279422931,
      "grad_norm": 2.029585599899292,
      "learning_rate": 1.9901457522572042e-05,
      "loss": 0.7267,
      "step": 885
    },
    {
      "epoch": 0.11212351303467477,
      "grad_norm": 1.621406078338623,
      "learning_rate": 1.990120639338198e-05,
      "loss": 0.742,
      "step": 886
    },
    {
      "epoch": 0.11225006327512023,
      "grad_norm": 2.4419679641723633,
      "learning_rate": 1.9900954946194192e-05,
      "loss": 0.7354,
      "step": 887
    },
    {
      "epoch": 0.11237661351556567,
      "grad_norm": 1.7079962491989136,
      "learning_rate": 1.9900703181016737e-05,
      "loss": 0.6786,
      "step": 888
    },
    {
      "epoch": 0.11250316375601113,
      "grad_norm": 1.5894176959991455,
      "learning_rate": 1.9900451097857713e-05,
      "loss": 0.7504,
      "step": 889
    },
    {
      "epoch": 0.11262971399645659,
      "grad_norm": 1.5512795448303223,
      "learning_rate": 1.990019869672521e-05,
      "loss": 0.7272,
      "step": 890
    },
    {
      "epoch": 0.11275626423690205,
      "grad_norm": 1.5827910900115967,
      "learning_rate": 1.989994597762734e-05,
      "loss": 0.7327,
      "step": 891
    },
    {
      "epoch": 0.1128828144773475,
      "grad_norm": 1.5007028579711914,
      "learning_rate": 1.989969294057222e-05,
      "loss": 0.7113,
      "step": 892
    },
    {
      "epoch": 0.11300936471779296,
      "grad_norm": 1.4675143957138062,
      "learning_rate": 1.9899439585567966e-05,
      "loss": 0.766,
      "step": 893
    },
    {
      "epoch": 0.11313591495823842,
      "grad_norm": 1.586624264717102,
      "learning_rate": 1.9899185912622726e-05,
      "loss": 0.7381,
      "step": 894
    },
    {
      "epoch": 0.11326246519868388,
      "grad_norm": 1.5194711685180664,
      "learning_rate": 1.9898931921744643e-05,
      "loss": 0.7203,
      "step": 895
    },
    {
      "epoch": 0.11338901543912934,
      "grad_norm": 1.6630206108093262,
      "learning_rate": 1.9898677612941874e-05,
      "loss": 0.7473,
      "step": 896
    },
    {
      "epoch": 0.1135155656795748,
      "grad_norm": 1.460914134979248,
      "learning_rate": 1.989842298622259e-05,
      "loss": 0.7197,
      "step": 897
    },
    {
      "epoch": 0.11364211592002024,
      "grad_norm": 1.5589309930801392,
      "learning_rate": 1.9898168041594963e-05,
      "loss": 0.7086,
      "step": 898
    },
    {
      "epoch": 0.1137686661604657,
      "grad_norm": 1.5489461421966553,
      "learning_rate": 1.989791277906719e-05,
      "loss": 0.7332,
      "step": 899
    },
    {
      "epoch": 0.11389521640091116,
      "grad_norm": 1.326927900314331,
      "learning_rate": 1.989765719864746e-05,
      "loss": 0.6884,
      "step": 900
    },
    {
      "epoch": 0.11402176664135662,
      "grad_norm": 1.4106143712997437,
      "learning_rate": 1.9897401300343985e-05,
      "loss": 0.7212,
      "step": 901
    },
    {
      "epoch": 0.11414831688180208,
      "grad_norm": 1.4296071529388428,
      "learning_rate": 1.9897145084164984e-05,
      "loss": 0.7148,
      "step": 902
    },
    {
      "epoch": 0.11427486712224753,
      "grad_norm": 1.3831626176834106,
      "learning_rate": 1.989688855011869e-05,
      "loss": 0.7583,
      "step": 903
    },
    {
      "epoch": 0.11440141736269299,
      "grad_norm": 1.4502894878387451,
      "learning_rate": 1.9896631698213338e-05,
      "loss": 0.6753,
      "step": 904
    },
    {
      "epoch": 0.11452796760313845,
      "grad_norm": 1.3960176706314087,
      "learning_rate": 1.9896374528457175e-05,
      "loss": 0.6802,
      "step": 905
    },
    {
      "epoch": 0.11465451784358391,
      "grad_norm": 1.7331746816635132,
      "learning_rate": 1.989611704085846e-05,
      "loss": 0.7722,
      "step": 906
    },
    {
      "epoch": 0.11478106808402935,
      "grad_norm": 1.3385294675827026,
      "learning_rate": 1.9895859235425474e-05,
      "loss": 0.7153,
      "step": 907
    },
    {
      "epoch": 0.11490761832447481,
      "grad_norm": 1.4386317729949951,
      "learning_rate": 1.9895601112166484e-05,
      "loss": 0.6986,
      "step": 908
    },
    {
      "epoch": 0.11503416856492027,
      "grad_norm": 1.7691932916641235,
      "learning_rate": 1.9895342671089785e-05,
      "loss": 0.7111,
      "step": 909
    },
    {
      "epoch": 0.11516071880536573,
      "grad_norm": 1.3420500755310059,
      "learning_rate": 1.9895083912203678e-05,
      "loss": 0.7166,
      "step": 910
    },
    {
      "epoch": 0.11528726904581119,
      "grad_norm": 1.8500834703445435,
      "learning_rate": 1.9894824835516474e-05,
      "loss": 0.7111,
      "step": 911
    },
    {
      "epoch": 0.11541381928625664,
      "grad_norm": 4.7799835205078125,
      "learning_rate": 1.989456544103649e-05,
      "loss": 0.7603,
      "step": 912
    },
    {
      "epoch": 0.1155403695267021,
      "grad_norm": 2.5567824840545654,
      "learning_rate": 1.989430572877206e-05,
      "loss": 0.7437,
      "step": 913
    },
    {
      "epoch": 0.11566691976714756,
      "grad_norm": 1.5591068267822266,
      "learning_rate": 1.9894045698731528e-05,
      "loss": 0.7248,
      "step": 914
    },
    {
      "epoch": 0.11579347000759302,
      "grad_norm": 1.8818336725234985,
      "learning_rate": 1.9893785350923238e-05,
      "loss": 0.7083,
      "step": 915
    },
    {
      "epoch": 0.11592002024803848,
      "grad_norm": 1.5813599824905396,
      "learning_rate": 1.989352468535556e-05,
      "loss": 0.743,
      "step": 916
    },
    {
      "epoch": 0.11604657048848392,
      "grad_norm": 1.4396576881408691,
      "learning_rate": 1.9893263702036858e-05,
      "loss": 0.701,
      "step": 917
    },
    {
      "epoch": 0.11617312072892938,
      "grad_norm": 2.2035129070281982,
      "learning_rate": 1.989300240097552e-05,
      "loss": 0.7178,
      "step": 918
    },
    {
      "epoch": 0.11629967096937484,
      "grad_norm": 1.503414511680603,
      "learning_rate": 1.9892740782179937e-05,
      "loss": 0.7208,
      "step": 919
    },
    {
      "epoch": 0.1164262212098203,
      "grad_norm": 2.5508668422698975,
      "learning_rate": 1.9892478845658507e-05,
      "loss": 0.7422,
      "step": 920
    },
    {
      "epoch": 0.11655277145026575,
      "grad_norm": 1.443010687828064,
      "learning_rate": 1.989221659141965e-05,
      "loss": 0.7141,
      "step": 921
    },
    {
      "epoch": 0.11667932169071121,
      "grad_norm": 2.385265588760376,
      "learning_rate": 1.989195401947178e-05,
      "loss": 0.7051,
      "step": 922
    },
    {
      "epoch": 0.11680587193115667,
      "grad_norm": 1.833487868309021,
      "learning_rate": 1.9891691129823337e-05,
      "loss": 0.7302,
      "step": 923
    },
    {
      "epoch": 0.11693242217160213,
      "grad_norm": 2.457515239715576,
      "learning_rate": 1.989142792248276e-05,
      "loss": 0.7285,
      "step": 924
    },
    {
      "epoch": 0.11705897241204759,
      "grad_norm": 2.152526378631592,
      "learning_rate": 1.989116439745851e-05,
      "loss": 0.7288,
      "step": 925
    },
    {
      "epoch": 0.11718552265249305,
      "grad_norm": 2.2381486892700195,
      "learning_rate": 1.989090055475904e-05,
      "loss": 0.7116,
      "step": 926
    },
    {
      "epoch": 0.11731207289293849,
      "grad_norm": 1.7628812789916992,
      "learning_rate": 1.9890636394392836e-05,
      "loss": 0.6799,
      "step": 927
    },
    {
      "epoch": 0.11743862313338395,
      "grad_norm": 1.9741804599761963,
      "learning_rate": 1.989037191636837e-05,
      "loss": 0.7353,
      "step": 928
    },
    {
      "epoch": 0.1175651733738294,
      "grad_norm": 1.7093483209609985,
      "learning_rate": 1.9890107120694142e-05,
      "loss": 0.6879,
      "step": 929
    },
    {
      "epoch": 0.11769172361427487,
      "grad_norm": 1.7342928647994995,
      "learning_rate": 1.9889842007378655e-05,
      "loss": 0.7372,
      "step": 930
    },
    {
      "epoch": 0.11781827385472032,
      "grad_norm": 1.6592991352081299,
      "learning_rate": 1.9889576576430425e-05,
      "loss": 0.7254,
      "step": 931
    },
    {
      "epoch": 0.11794482409516578,
      "grad_norm": 2.2503223419189453,
      "learning_rate": 1.9889310827857978e-05,
      "loss": 0.7265,
      "step": 932
    },
    {
      "epoch": 0.11807137433561124,
      "grad_norm": 1.5699291229248047,
      "learning_rate": 1.9889044761669844e-05,
      "loss": 0.7032,
      "step": 933
    },
    {
      "epoch": 0.1181979245760567,
      "grad_norm": 2.1852786540985107,
      "learning_rate": 1.9888778377874576e-05,
      "loss": 0.6827,
      "step": 934
    },
    {
      "epoch": 0.11832447481650216,
      "grad_norm": 2.206437110900879,
      "learning_rate": 1.9888511676480723e-05,
      "loss": 0.6864,
      "step": 935
    },
    {
      "epoch": 0.11845102505694761,
      "grad_norm": 1.915076494216919,
      "learning_rate": 1.9888244657496854e-05,
      "loss": 0.7662,
      "step": 936
    },
    {
      "epoch": 0.11857757529739306,
      "grad_norm": 2.3391952514648438,
      "learning_rate": 1.988797732093154e-05,
      "loss": 0.7659,
      "step": 937
    },
    {
      "epoch": 0.11870412553783852,
      "grad_norm": 1.4723795652389526,
      "learning_rate": 1.9887709666793373e-05,
      "loss": 0.7495,
      "step": 938
    },
    {
      "epoch": 0.11883067577828398,
      "grad_norm": 2.1229209899902344,
      "learning_rate": 1.988744169509095e-05,
      "loss": 0.7319,
      "step": 939
    },
    {
      "epoch": 0.11895722601872943,
      "grad_norm": 1.2992359399795532,
      "learning_rate": 1.9887173405832873e-05,
      "loss": 0.7278,
      "step": 940
    },
    {
      "epoch": 0.11908377625917489,
      "grad_norm": 1.4991666078567505,
      "learning_rate": 1.9886904799027757e-05,
      "loss": 0.6578,
      "step": 941
    },
    {
      "epoch": 0.11921032649962035,
      "grad_norm": 1.3382859230041504,
      "learning_rate": 1.9886635874684236e-05,
      "loss": 0.7159,
      "step": 942
    },
    {
      "epoch": 0.11933687674006581,
      "grad_norm": 1.704626202583313,
      "learning_rate": 1.988636663281094e-05,
      "loss": 0.7196,
      "step": 943
    },
    {
      "epoch": 0.11946342698051127,
      "grad_norm": 1.445652961730957,
      "learning_rate": 1.988609707341652e-05,
      "loss": 0.7041,
      "step": 944
    },
    {
      "epoch": 0.11958997722095673,
      "grad_norm": 1.3373247385025024,
      "learning_rate": 1.9885827196509634e-05,
      "loss": 0.7002,
      "step": 945
    },
    {
      "epoch": 0.11971652746140218,
      "grad_norm": 1.4224271774291992,
      "learning_rate": 1.9885557002098948e-05,
      "loss": 0.6723,
      "step": 946
    },
    {
      "epoch": 0.11984307770184763,
      "grad_norm": 1.8158930540084839,
      "learning_rate": 1.988528649019314e-05,
      "loss": 0.7311,
      "step": 947
    },
    {
      "epoch": 0.11996962794229309,
      "grad_norm": 1.4352871179580688,
      "learning_rate": 1.98850156608009e-05,
      "loss": 0.6859,
      "step": 948
    },
    {
      "epoch": 0.12009617818273854,
      "grad_norm": 1.6294814348220825,
      "learning_rate": 1.9884744513930922e-05,
      "loss": 0.7549,
      "step": 949
    },
    {
      "epoch": 0.120222728423184,
      "grad_norm": 1.7172322273254395,
      "learning_rate": 1.9884473049591918e-05,
      "loss": 0.7115,
      "step": 950
    },
    {
      "epoch": 0.12034927866362946,
      "grad_norm": 1.456181287765503,
      "learning_rate": 1.9884201267792608e-05,
      "loss": 0.7187,
      "step": 951
    },
    {
      "epoch": 0.12047582890407492,
      "grad_norm": 1.5850032567977905,
      "learning_rate": 1.9883929168541715e-05,
      "loss": 0.7327,
      "step": 952
    },
    {
      "epoch": 0.12060237914452038,
      "grad_norm": 1.4637863636016846,
      "learning_rate": 1.9883656751847983e-05,
      "loss": 0.6675,
      "step": 953
    },
    {
      "epoch": 0.12072892938496584,
      "grad_norm": 1.565523624420166,
      "learning_rate": 1.988338401772016e-05,
      "loss": 0.7783,
      "step": 954
    },
    {
      "epoch": 0.1208554796254113,
      "grad_norm": 1.4565293788909912,
      "learning_rate": 1.9883110966167007e-05,
      "loss": 0.695,
      "step": 955
    },
    {
      "epoch": 0.12098202986585674,
      "grad_norm": 1.6588215827941895,
      "learning_rate": 1.988283759719729e-05,
      "loss": 0.7089,
      "step": 956
    },
    {
      "epoch": 0.1211085801063022,
      "grad_norm": 1.5119205713272095,
      "learning_rate": 1.988256391081979e-05,
      "loss": 0.6855,
      "step": 957
    },
    {
      "epoch": 0.12123513034674765,
      "grad_norm": 1.4645875692367554,
      "learning_rate": 1.9882289907043298e-05,
      "loss": 0.7115,
      "step": 958
    },
    {
      "epoch": 0.12136168058719311,
      "grad_norm": 1.5104066133499146,
      "learning_rate": 1.9882015585876614e-05,
      "loss": 0.6822,
      "step": 959
    },
    {
      "epoch": 0.12148823082763857,
      "grad_norm": 1.4040604829788208,
      "learning_rate": 1.988174094732855e-05,
      "loss": 0.7004,
      "step": 960
    },
    {
      "epoch": 0.12161478106808403,
      "grad_norm": 1.3956875801086426,
      "learning_rate": 1.988146599140792e-05,
      "loss": 0.7241,
      "step": 961
    },
    {
      "epoch": 0.12174133130852949,
      "grad_norm": 1.3745182752609253,
      "learning_rate": 1.9881190718123567e-05,
      "loss": 0.6621,
      "step": 962
    },
    {
      "epoch": 0.12186788154897495,
      "grad_norm": 1.7807087898254395,
      "learning_rate": 1.988091512748432e-05,
      "loss": 0.7953,
      "step": 963
    },
    {
      "epoch": 0.1219944317894204,
      "grad_norm": 1.479649305343628,
      "learning_rate": 1.9880639219499034e-05,
      "loss": 0.743,
      "step": 964
    },
    {
      "epoch": 0.12212098202986586,
      "grad_norm": 1.414460301399231,
      "learning_rate": 1.9880362994176573e-05,
      "loss": 0.7269,
      "step": 965
    },
    {
      "epoch": 0.12224753227031131,
      "grad_norm": 1.4834524393081665,
      "learning_rate": 1.9880086451525805e-05,
      "loss": 0.7032,
      "step": 966
    },
    {
      "epoch": 0.12237408251075677,
      "grad_norm": 1.3599659204483032,
      "learning_rate": 1.9879809591555617e-05,
      "loss": 0.7059,
      "step": 967
    },
    {
      "epoch": 0.12250063275120222,
      "grad_norm": 1.3363984823226929,
      "learning_rate": 1.9879532414274893e-05,
      "loss": 0.6989,
      "step": 968
    },
    {
      "epoch": 0.12262718299164768,
      "grad_norm": 1.4159510135650635,
      "learning_rate": 1.987925491969254e-05,
      "loss": 0.6941,
      "step": 969
    },
    {
      "epoch": 0.12275373323209314,
      "grad_norm": 1.4465556144714355,
      "learning_rate": 1.9878977107817473e-05,
      "loss": 0.6996,
      "step": 970
    },
    {
      "epoch": 0.1228802834725386,
      "grad_norm": 1.3270994424819946,
      "learning_rate": 1.987869897865861e-05,
      "loss": 0.7389,
      "step": 971
    },
    {
      "epoch": 0.12300683371298406,
      "grad_norm": 1.628600835800171,
      "learning_rate": 1.9878420532224886e-05,
      "loss": 0.7368,
      "step": 972
    },
    {
      "epoch": 0.12313338395342951,
      "grad_norm": 1.520208716392517,
      "learning_rate": 1.9878141768525235e-05,
      "loss": 0.7368,
      "step": 973
    },
    {
      "epoch": 0.12325993419387497,
      "grad_norm": 1.6098644733428955,
      "learning_rate": 1.9877862687568625e-05,
      "loss": 0.7311,
      "step": 974
    },
    {
      "epoch": 0.12338648443432043,
      "grad_norm": 1.4917585849761963,
      "learning_rate": 1.9877583289364013e-05,
      "loss": 0.7249,
      "step": 975
    },
    {
      "epoch": 0.12351303467476588,
      "grad_norm": 1.604479432106018,
      "learning_rate": 1.987730357392037e-05,
      "loss": 0.7501,
      "step": 976
    },
    {
      "epoch": 0.12363958491521133,
      "grad_norm": 1.4240484237670898,
      "learning_rate": 1.987702354124668e-05,
      "loss": 0.6912,
      "step": 977
    },
    {
      "epoch": 0.12376613515565679,
      "grad_norm": 1.5388692617416382,
      "learning_rate": 1.987674319135194e-05,
      "loss": 0.7781,
      "step": 978
    },
    {
      "epoch": 0.12389268539610225,
      "grad_norm": 1.3599058389663696,
      "learning_rate": 1.9876462524245152e-05,
      "loss": 0.7033,
      "step": 979
    },
    {
      "epoch": 0.12401923563654771,
      "grad_norm": 3.2703309059143066,
      "learning_rate": 1.9876181539935327e-05,
      "loss": 0.7878,
      "step": 980
    },
    {
      "epoch": 0.12414578587699317,
      "grad_norm": 1.3605518341064453,
      "learning_rate": 1.9875900238431495e-05,
      "loss": 0.6893,
      "step": 981
    },
    {
      "epoch": 0.12427233611743863,
      "grad_norm": 1.4645575284957886,
      "learning_rate": 1.987561861974269e-05,
      "loss": 0.7587,
      "step": 982
    },
    {
      "epoch": 0.12439888635788408,
      "grad_norm": 1.3291693925857544,
      "learning_rate": 1.9875336683877953e-05,
      "loss": 0.7381,
      "step": 983
    },
    {
      "epoch": 0.12452543659832954,
      "grad_norm": 1.411341905593872,
      "learning_rate": 1.987505443084634e-05,
      "loss": 0.7162,
      "step": 984
    },
    {
      "epoch": 0.124651986838775,
      "grad_norm": 1.6751728057861328,
      "learning_rate": 1.9874771860656922e-05,
      "loss": 0.7299,
      "step": 985
    },
    {
      "epoch": 0.12477853707922044,
      "grad_norm": 1.3982819318771362,
      "learning_rate": 1.9874488973318765e-05,
      "loss": 0.6883,
      "step": 986
    },
    {
      "epoch": 0.1249050873196659,
      "grad_norm": 1.5910639762878418,
      "learning_rate": 1.9874205768840958e-05,
      "loss": 0.712,
      "step": 987
    },
    {
      "epoch": 0.12503163756011137,
      "grad_norm": 1.492058515548706,
      "learning_rate": 1.98739222472326e-05,
      "loss": 0.7721,
      "step": 988
    },
    {
      "epoch": 0.12515818780055682,
      "grad_norm": 1.8080620765686035,
      "learning_rate": 1.9873638408502797e-05,
      "loss": 0.75,
      "step": 989
    },
    {
      "epoch": 0.1252847380410023,
      "grad_norm": 1.4867298603057861,
      "learning_rate": 1.987335425266066e-05,
      "loss": 0.7188,
      "step": 990
    },
    {
      "epoch": 0.12541128828144774,
      "grad_norm": 1.4081100225448608,
      "learning_rate": 1.987306977971532e-05,
      "loss": 0.7618,
      "step": 991
    },
    {
      "epoch": 0.12553783852189318,
      "grad_norm": 1.3052310943603516,
      "learning_rate": 1.9872784989675905e-05,
      "loss": 0.6909,
      "step": 992
    },
    {
      "epoch": 0.12566438876233865,
      "grad_norm": 1.4310415983200073,
      "learning_rate": 1.9872499882551574e-05,
      "loss": 0.6902,
      "step": 993
    },
    {
      "epoch": 0.1257909390027841,
      "grad_norm": 1.4493794441223145,
      "learning_rate": 1.9872214458351477e-05,
      "loss": 0.7128,
      "step": 994
    },
    {
      "epoch": 0.12591748924322957,
      "grad_norm": 1.3193589448928833,
      "learning_rate": 1.987192871708478e-05,
      "loss": 0.6814,
      "step": 995
    },
    {
      "epoch": 0.126044039483675,
      "grad_norm": 1.418829321861267,
      "learning_rate": 1.987164265876066e-05,
      "loss": 0.7065,
      "step": 996
    },
    {
      "epoch": 0.12617058972412049,
      "grad_norm": 1.3695859909057617,
      "learning_rate": 1.9871356283388307e-05,
      "loss": 0.7596,
      "step": 997
    },
    {
      "epoch": 0.12629713996456593,
      "grad_norm": 1.5387675762176514,
      "learning_rate": 1.9871069590976922e-05,
      "loss": 0.6643,
      "step": 998
    },
    {
      "epoch": 0.1264236902050114,
      "grad_norm": 1.4276633262634277,
      "learning_rate": 1.9870782581535705e-05,
      "loss": 0.7078,
      "step": 999
    },
    {
      "epoch": 0.12655024044545685,
      "grad_norm": 1.4148565530776978,
      "learning_rate": 1.9870495255073877e-05,
      "loss": 0.7412,
      "step": 1000
    },
    {
      "epoch": 0.1266767906859023,
      "grad_norm": 1.346156120300293,
      "learning_rate": 1.987020761160067e-05,
      "loss": 0.7166,
      "step": 1001
    },
    {
      "epoch": 0.12680334092634776,
      "grad_norm": 1.409359097480774,
      "learning_rate": 1.9869919651125315e-05,
      "loss": 0.7087,
      "step": 1002
    },
    {
      "epoch": 0.1269298911667932,
      "grad_norm": 1.3479506969451904,
      "learning_rate": 1.9869631373657067e-05,
      "loss": 0.6866,
      "step": 1003
    },
    {
      "epoch": 0.12705644140723868,
      "grad_norm": 1.2483811378479004,
      "learning_rate": 1.986934277920518e-05,
      "loss": 0.709,
      "step": 1004
    },
    {
      "epoch": 0.12718299164768412,
      "grad_norm": 1.276661992073059,
      "learning_rate": 1.9869053867778926e-05,
      "loss": 0.6837,
      "step": 1005
    },
    {
      "epoch": 0.1273095418881296,
      "grad_norm": 1.3566193580627441,
      "learning_rate": 1.9868764639387584e-05,
      "loss": 0.7188,
      "step": 1006
    },
    {
      "epoch": 0.12743609212857504,
      "grad_norm": 1.369962215423584,
      "learning_rate": 1.986847509404044e-05,
      "loss": 0.7407,
      "step": 1007
    },
    {
      "epoch": 0.1275626423690205,
      "grad_norm": 1.3317837715148926,
      "learning_rate": 1.9868185231746794e-05,
      "loss": 0.6972,
      "step": 1008
    },
    {
      "epoch": 0.12768919260946596,
      "grad_norm": 1.3249095678329468,
      "learning_rate": 1.9867895052515958e-05,
      "loss": 0.7032,
      "step": 1009
    },
    {
      "epoch": 0.1278157428499114,
      "grad_norm": 1.3725255727767944,
      "learning_rate": 1.986760455635725e-05,
      "loss": 0.666,
      "step": 1010
    },
    {
      "epoch": 0.12794229309035687,
      "grad_norm": 1.7813315391540527,
      "learning_rate": 1.986731374328e-05,
      "loss": 0.743,
      "step": 1011
    },
    {
      "epoch": 0.12806884333080232,
      "grad_norm": 1.4598991870880127,
      "learning_rate": 1.986702261329355e-05,
      "loss": 0.6989,
      "step": 1012
    },
    {
      "epoch": 0.1281953935712478,
      "grad_norm": 1.647645115852356,
      "learning_rate": 1.9866731166407248e-05,
      "loss": 0.7141,
      "step": 1013
    },
    {
      "epoch": 0.12832194381169323,
      "grad_norm": 1.618118405342102,
      "learning_rate": 1.9866439402630454e-05,
      "loss": 0.6892,
      "step": 1014
    },
    {
      "epoch": 0.1284484940521387,
      "grad_norm": 1.4833314418792725,
      "learning_rate": 1.986614732197254e-05,
      "loss": 0.7704,
      "step": 1015
    },
    {
      "epoch": 0.12857504429258415,
      "grad_norm": 1.6827833652496338,
      "learning_rate": 1.9865854924442885e-05,
      "loss": 0.7559,
      "step": 1016
    },
    {
      "epoch": 0.12870159453302962,
      "grad_norm": 1.3908610343933105,
      "learning_rate": 1.9865562210050884e-05,
      "loss": 0.7039,
      "step": 1017
    },
    {
      "epoch": 0.12882814477347507,
      "grad_norm": 1.391330599784851,
      "learning_rate": 1.9865269178805934e-05,
      "loss": 0.7447,
      "step": 1018
    },
    {
      "epoch": 0.12895469501392054,
      "grad_norm": 1.3229349851608276,
      "learning_rate": 1.9864975830717448e-05,
      "loss": 0.6754,
      "step": 1019
    },
    {
      "epoch": 0.12908124525436598,
      "grad_norm": 1.421649694442749,
      "learning_rate": 1.9864682165794845e-05,
      "loss": 0.7417,
      "step": 1020
    },
    {
      "epoch": 0.12920779549481143,
      "grad_norm": 1.8537667989730835,
      "learning_rate": 1.9864388184047564e-05,
      "loss": 0.7284,
      "step": 1021
    },
    {
      "epoch": 0.1293343457352569,
      "grad_norm": 1.5346688032150269,
      "learning_rate": 1.9864093885485037e-05,
      "loss": 0.7265,
      "step": 1022
    },
    {
      "epoch": 0.12946089597570234,
      "grad_norm": 1.5469647645950317,
      "learning_rate": 1.9863799270116725e-05,
      "loss": 0.7298,
      "step": 1023
    },
    {
      "epoch": 0.12958744621614782,
      "grad_norm": 1.7426066398620605,
      "learning_rate": 1.9863504337952083e-05,
      "loss": 0.7451,
      "step": 1024
    },
    {
      "epoch": 0.12971399645659326,
      "grad_norm": 1.3992360830307007,
      "learning_rate": 1.9863209089000585e-05,
      "loss": 0.7432,
      "step": 1025
    },
    {
      "epoch": 0.12984054669703873,
      "grad_norm": 1.5705806016921997,
      "learning_rate": 1.9862913523271715e-05,
      "loss": 0.7653,
      "step": 1026
    },
    {
      "epoch": 0.12996709693748418,
      "grad_norm": 1.357663631439209,
      "learning_rate": 1.9862617640774968e-05,
      "loss": 0.703,
      "step": 1027
    },
    {
      "epoch": 0.13009364717792965,
      "grad_norm": 1.4717273712158203,
      "learning_rate": 1.9862321441519846e-05,
      "loss": 0.732,
      "step": 1028
    },
    {
      "epoch": 0.1302201974183751,
      "grad_norm": 1.3332781791687012,
      "learning_rate": 1.986202492551586e-05,
      "loss": 0.6982,
      "step": 1029
    },
    {
      "epoch": 0.13034674765882054,
      "grad_norm": 1.34749174118042,
      "learning_rate": 1.9861728092772527e-05,
      "loss": 0.7133,
      "step": 1030
    },
    {
      "epoch": 0.130473297899266,
      "grad_norm": 1.372266411781311,
      "learning_rate": 1.9861430943299395e-05,
      "loss": 0.6911,
      "step": 1031
    },
    {
      "epoch": 0.13059984813971146,
      "grad_norm": 1.3901941776275635,
      "learning_rate": 1.9861133477105997e-05,
      "loss": 0.7446,
      "step": 1032
    },
    {
      "epoch": 0.13072639838015693,
      "grad_norm": 1.4268280267715454,
      "learning_rate": 1.986083569420189e-05,
      "loss": 0.7309,
      "step": 1033
    },
    {
      "epoch": 0.13085294862060237,
      "grad_norm": 1.563790202140808,
      "learning_rate": 1.9860537594596633e-05,
      "loss": 0.703,
      "step": 1034
    },
    {
      "epoch": 0.13097949886104784,
      "grad_norm": 1.566311240196228,
      "learning_rate": 1.986023917829981e-05,
      "loss": 0.7137,
      "step": 1035
    },
    {
      "epoch": 0.1311060491014933,
      "grad_norm": 1.4129443168640137,
      "learning_rate": 1.9859940445320994e-05,
      "loss": 0.6699,
      "step": 1036
    },
    {
      "epoch": 0.13123259934193876,
      "grad_norm": 1.419777750968933,
      "learning_rate": 1.9859641395669785e-05,
      "loss": 0.716,
      "step": 1037
    },
    {
      "epoch": 0.1313591495823842,
      "grad_norm": 1.7544318437576294,
      "learning_rate": 1.9859342029355793e-05,
      "loss": 0.7373,
      "step": 1038
    },
    {
      "epoch": 0.13148569982282968,
      "grad_norm": 1.393851399421692,
      "learning_rate": 1.9859042346388627e-05,
      "loss": 0.7098,
      "step": 1039
    },
    {
      "epoch": 0.13161225006327512,
      "grad_norm": 1.743149995803833,
      "learning_rate": 1.985874234677791e-05,
      "loss": 0.7426,
      "step": 1040
    },
    {
      "epoch": 0.13173880030372057,
      "grad_norm": 1.6342633962631226,
      "learning_rate": 1.985844203053328e-05,
      "loss": 0.7602,
      "step": 1041
    },
    {
      "epoch": 0.13186535054416604,
      "grad_norm": 1.4077647924423218,
      "learning_rate": 1.9858141397664382e-05,
      "loss": 0.7245,
      "step": 1042
    },
    {
      "epoch": 0.13199190078461148,
      "grad_norm": 1.5660897493362427,
      "learning_rate": 1.985784044818087e-05,
      "loss": 0.705,
      "step": 1043
    },
    {
      "epoch": 0.13211845102505695,
      "grad_norm": 1.3453203439712524,
      "learning_rate": 1.985753918209241e-05,
      "loss": 0.7415,
      "step": 1044
    },
    {
      "epoch": 0.1322450012655024,
      "grad_norm": 1.4615741968154907,
      "learning_rate": 1.9857237599408684e-05,
      "loss": 0.7041,
      "step": 1045
    },
    {
      "epoch": 0.13237155150594787,
      "grad_norm": 1.2403738498687744,
      "learning_rate": 1.9856935700139367e-05,
      "loss": 0.6444,
      "step": 1046
    },
    {
      "epoch": 0.13249810174639332,
      "grad_norm": 1.3161225318908691,
      "learning_rate": 1.9856633484294163e-05,
      "loss": 0.6918,
      "step": 1047
    },
    {
      "epoch": 0.1326246519868388,
      "grad_norm": 1.5872657299041748,
      "learning_rate": 1.9856330951882775e-05,
      "loss": 0.7353,
      "step": 1048
    },
    {
      "epoch": 0.13275120222728423,
      "grad_norm": 1.2746602296829224,
      "learning_rate": 1.9856028102914924e-05,
      "loss": 0.7071,
      "step": 1049
    },
    {
      "epoch": 0.13287775246772968,
      "grad_norm": 1.8670061826705933,
      "learning_rate": 1.985572493740033e-05,
      "loss": 0.7772,
      "step": 1050
    },
    {
      "epoch": 0.13300430270817515,
      "grad_norm": 1.8144220113754272,
      "learning_rate": 1.985542145534873e-05,
      "loss": 0.7135,
      "step": 1051
    },
    {
      "epoch": 0.1331308529486206,
      "grad_norm": 1.3191373348236084,
      "learning_rate": 1.9855117656769878e-05,
      "loss": 0.7028,
      "step": 1052
    },
    {
      "epoch": 0.13325740318906606,
      "grad_norm": 1.3742867708206177,
      "learning_rate": 1.9854813541673525e-05,
      "loss": 0.7112,
      "step": 1053
    },
    {
      "epoch": 0.1333839534295115,
      "grad_norm": 1.3153440952301025,
      "learning_rate": 1.9854509110069444e-05,
      "loss": 0.695,
      "step": 1054
    },
    {
      "epoch": 0.13351050366995698,
      "grad_norm": 1.2031164169311523,
      "learning_rate": 1.9854204361967402e-05,
      "loss": 0.6969,
      "step": 1055
    },
    {
      "epoch": 0.13363705391040243,
      "grad_norm": 1.3659594058990479,
      "learning_rate": 1.9853899297377198e-05,
      "loss": 0.7451,
      "step": 1056
    },
    {
      "epoch": 0.1337636041508479,
      "grad_norm": 1.4110047817230225,
      "learning_rate": 1.9853593916308624e-05,
      "loss": 0.6879,
      "step": 1057
    },
    {
      "epoch": 0.13389015439129334,
      "grad_norm": 5.904182434082031,
      "learning_rate": 1.9853288218771486e-05,
      "loss": 0.7295,
      "step": 1058
    },
    {
      "epoch": 0.1340167046317388,
      "grad_norm": 1.8530023097991943,
      "learning_rate": 1.985298220477561e-05,
      "loss": 0.7287,
      "step": 1059
    },
    {
      "epoch": 0.13414325487218426,
      "grad_norm": 1.6028395891189575,
      "learning_rate": 1.985267587433082e-05,
      "loss": 0.7205,
      "step": 1060
    },
    {
      "epoch": 0.1342698051126297,
      "grad_norm": 1.431804895401001,
      "learning_rate": 1.985236922744695e-05,
      "loss": 0.7501,
      "step": 1061
    },
    {
      "epoch": 0.13439635535307518,
      "grad_norm": 2.1176390647888184,
      "learning_rate": 1.9852062264133854e-05,
      "loss": 0.7883,
      "step": 1062
    },
    {
      "epoch": 0.13452290559352062,
      "grad_norm": 2.950989246368408,
      "learning_rate": 1.9851754984401386e-05,
      "loss": 0.746,
      "step": 1063
    },
    {
      "epoch": 0.1346494558339661,
      "grad_norm": 1.731371521949768,
      "learning_rate": 1.985144738825942e-05,
      "loss": 0.7112,
      "step": 1064
    },
    {
      "epoch": 0.13477600607441154,
      "grad_norm": 1.6559282541275024,
      "learning_rate": 1.9851139475717834e-05,
      "loss": 0.6818,
      "step": 1065
    },
    {
      "epoch": 0.134902556314857,
      "grad_norm": 1.6687043905258179,
      "learning_rate": 1.9850831246786516e-05,
      "loss": 0.6858,
      "step": 1066
    },
    {
      "epoch": 0.13502910655530245,
      "grad_norm": 2.0243048667907715,
      "learning_rate": 1.985052270147537e-05,
      "loss": 0.7129,
      "step": 1067
    },
    {
      "epoch": 0.13515565679574792,
      "grad_norm": 1.4248886108398438,
      "learning_rate": 1.9850213839794296e-05,
      "loss": 0.7105,
      "step": 1068
    },
    {
      "epoch": 0.13528220703619337,
      "grad_norm": 1.9948289394378662,
      "learning_rate": 1.984990466175322e-05,
      "loss": 0.6915,
      "step": 1069
    },
    {
      "epoch": 0.1354087572766388,
      "grad_norm": 1.5190576314926147,
      "learning_rate": 1.9849595167362072e-05,
      "loss": 0.7094,
      "step": 1070
    },
    {
      "epoch": 0.13553530751708429,
      "grad_norm": 2.085909605026245,
      "learning_rate": 1.984928535663079e-05,
      "loss": 0.6943,
      "step": 1071
    },
    {
      "epoch": 0.13566185775752973,
      "grad_norm": 1.5484418869018555,
      "learning_rate": 1.9848975229569328e-05,
      "loss": 0.7116,
      "step": 1072
    },
    {
      "epoch": 0.1357884079979752,
      "grad_norm": 2.0897481441497803,
      "learning_rate": 1.984866478618764e-05,
      "loss": 0.75,
      "step": 1073
    },
    {
      "epoch": 0.13591495823842065,
      "grad_norm": 1.5771828889846802,
      "learning_rate": 1.98483540264957e-05,
      "loss": 0.6663,
      "step": 1074
    },
    {
      "epoch": 0.13604150847886612,
      "grad_norm": 2.2784533500671387,
      "learning_rate": 1.9848042950503492e-05,
      "loss": 0.7511,
      "step": 1075
    },
    {
      "epoch": 0.13616805871931156,
      "grad_norm": 1.6186012029647827,
      "learning_rate": 1.9847731558221005e-05,
      "loss": 0.7595,
      "step": 1076
    },
    {
      "epoch": 0.13629460895975704,
      "grad_norm": 2.2613446712493896,
      "learning_rate": 1.9847419849658232e-05,
      "loss": 0.7443,
      "step": 1077
    },
    {
      "epoch": 0.13642115920020248,
      "grad_norm": 2.2935702800750732,
      "learning_rate": 1.9847107824825194e-05,
      "loss": 0.6897,
      "step": 1078
    },
    {
      "epoch": 0.13654770944064792,
      "grad_norm": 1.5083117485046387,
      "learning_rate": 1.9846795483731914e-05,
      "loss": 0.6912,
      "step": 1079
    },
    {
      "epoch": 0.1366742596810934,
      "grad_norm": 1.894834280014038,
      "learning_rate": 1.9846482826388414e-05,
      "loss": 0.6972,
      "step": 1080
    },
    {
      "epoch": 0.13680080992153884,
      "grad_norm": 1.3171061277389526,
      "learning_rate": 1.9846169852804743e-05,
      "loss": 0.6844,
      "step": 1081
    },
    {
      "epoch": 0.1369273601619843,
      "grad_norm": 1.8938465118408203,
      "learning_rate": 1.9845856562990947e-05,
      "loss": 0.7393,
      "step": 1082
    },
    {
      "epoch": 0.13705391040242976,
      "grad_norm": 1.3566538095474243,
      "learning_rate": 1.9845542956957095e-05,
      "loss": 0.7077,
      "step": 1083
    },
    {
      "epoch": 0.13718046064287523,
      "grad_norm": 1.4441255331039429,
      "learning_rate": 1.984522903471325e-05,
      "loss": 0.7404,
      "step": 1084
    },
    {
      "epoch": 0.13730701088332067,
      "grad_norm": 1.6459498405456543,
      "learning_rate": 1.9844914796269502e-05,
      "loss": 0.7173,
      "step": 1085
    },
    {
      "epoch": 0.13743356112376615,
      "grad_norm": 1.3637651205062866,
      "learning_rate": 1.984460024163594e-05,
      "loss": 0.7152,
      "step": 1086
    },
    {
      "epoch": 0.1375601113642116,
      "grad_norm": 1.4726169109344482,
      "learning_rate": 1.984428537082267e-05,
      "loss": 0.7017,
      "step": 1087
    },
    {
      "epoch": 0.13768666160465706,
      "grad_norm": 1.3233355283737183,
      "learning_rate": 1.9843970183839804e-05,
      "loss": 0.7421,
      "step": 1088
    },
    {
      "epoch": 0.1378132118451025,
      "grad_norm": 1.6742472648620605,
      "learning_rate": 1.9843654680697457e-05,
      "loss": 0.7336,
      "step": 1089
    },
    {
      "epoch": 0.13793976208554795,
      "grad_norm": 1.6620330810546875,
      "learning_rate": 1.9843338861405772e-05,
      "loss": 0.7046,
      "step": 1090
    },
    {
      "epoch": 0.13806631232599342,
      "grad_norm": 1.6886900663375854,
      "learning_rate": 1.9843022725974886e-05,
      "loss": 0.6724,
      "step": 1091
    },
    {
      "epoch": 0.13819286256643887,
      "grad_norm": 1.3449492454528809,
      "learning_rate": 1.9842706274414957e-05,
      "loss": 0.7118,
      "step": 1092
    },
    {
      "epoch": 0.13831941280688434,
      "grad_norm": 1.8862601518630981,
      "learning_rate": 1.9842389506736148e-05,
      "loss": 0.7404,
      "step": 1093
    },
    {
      "epoch": 0.13844596304732978,
      "grad_norm": 1.3016871213912964,
      "learning_rate": 1.9842072422948627e-05,
      "loss": 0.714,
      "step": 1094
    },
    {
      "epoch": 0.13857251328777526,
      "grad_norm": 1.3852330446243286,
      "learning_rate": 1.984175502306258e-05,
      "loss": 0.735,
      "step": 1095
    },
    {
      "epoch": 0.1386990635282207,
      "grad_norm": 1.3952821493148804,
      "learning_rate": 1.9841437307088208e-05,
      "loss": 0.7121,
      "step": 1096
    },
    {
      "epoch": 0.13882561376866617,
      "grad_norm": 1.3784925937652588,
      "learning_rate": 1.9841119275035705e-05,
      "loss": 0.7673,
      "step": 1097
    },
    {
      "epoch": 0.13895216400911162,
      "grad_norm": 2.304252862930298,
      "learning_rate": 1.984080092691529e-05,
      "loss": 0.7319,
      "step": 1098
    },
    {
      "epoch": 0.13907871424955706,
      "grad_norm": 1.4510414600372314,
      "learning_rate": 1.984048226273719e-05,
      "loss": 0.7693,
      "step": 1099
    },
    {
      "epoch": 0.13920526449000253,
      "grad_norm": 1.7344343662261963,
      "learning_rate": 1.9840163282511636e-05,
      "loss": 0.6861,
      "step": 1100
    },
    {
      "epoch": 0.13933181473044798,
      "grad_norm": 1.4922350645065308,
      "learning_rate": 1.9839843986248874e-05,
      "loss": 0.7644,
      "step": 1101
    },
    {
      "epoch": 0.13945836497089345,
      "grad_norm": 1.8472124338150024,
      "learning_rate": 1.9839524373959155e-05,
      "loss": 0.6899,
      "step": 1102
    },
    {
      "epoch": 0.1395849152113389,
      "grad_norm": 1.6263045072555542,
      "learning_rate": 1.9839204445652754e-05,
      "loss": 0.7084,
      "step": 1103
    },
    {
      "epoch": 0.13971146545178437,
      "grad_norm": 1.947512149810791,
      "learning_rate": 1.9838884201339935e-05,
      "loss": 0.7106,
      "step": 1104
    },
    {
      "epoch": 0.1398380156922298,
      "grad_norm": 1.612810492515564,
      "learning_rate": 1.983856364103099e-05,
      "loss": 0.6802,
      "step": 1105
    },
    {
      "epoch": 0.13996456593267528,
      "grad_norm": 1.8875095844268799,
      "learning_rate": 1.983824276473621e-05,
      "loss": 0.7337,
      "step": 1106
    },
    {
      "epoch": 0.14009111617312073,
      "grad_norm": 2.1237359046936035,
      "learning_rate": 1.9837921572465902e-05,
      "loss": 0.7005,
      "step": 1107
    },
    {
      "epoch": 0.14021766641356617,
      "grad_norm": 1.68097722530365,
      "learning_rate": 1.9837600064230384e-05,
      "loss": 0.6617,
      "step": 1108
    },
    {
      "epoch": 0.14034421665401164,
      "grad_norm": 1.3452650308609009,
      "learning_rate": 1.9837278240039983e-05,
      "loss": 0.7047,
      "step": 1109
    },
    {
      "epoch": 0.1404707668944571,
      "grad_norm": 1.8015680313110352,
      "learning_rate": 1.9836956099905032e-05,
      "loss": 0.6958,
      "step": 1110
    },
    {
      "epoch": 0.14059731713490256,
      "grad_norm": 1.4769086837768555,
      "learning_rate": 1.9836633643835875e-05,
      "loss": 0.6866,
      "step": 1111
    },
    {
      "epoch": 0.140723867375348,
      "grad_norm": 1.8468232154846191,
      "learning_rate": 1.9836310871842873e-05,
      "loss": 0.7344,
      "step": 1112
    },
    {
      "epoch": 0.14085041761579348,
      "grad_norm": 1.396087646484375,
      "learning_rate": 1.983598778393639e-05,
      "loss": 0.6762,
      "step": 1113
    },
    {
      "epoch": 0.14097696785623892,
      "grad_norm": 2.159205436706543,
      "learning_rate": 1.9835664380126806e-05,
      "loss": 0.6928,
      "step": 1114
    },
    {
      "epoch": 0.1411035180966844,
      "grad_norm": 1.7167434692382812,
      "learning_rate": 1.9835340660424503e-05,
      "loss": 0.6444,
      "step": 1115
    },
    {
      "epoch": 0.14123006833712984,
      "grad_norm": 1.9127838611602783,
      "learning_rate": 1.9835016624839884e-05,
      "loss": 0.7196,
      "step": 1116
    },
    {
      "epoch": 0.1413566185775753,
      "grad_norm": 1.7396206855773926,
      "learning_rate": 1.9834692273383346e-05,
      "loss": 0.6933,
      "step": 1117
    },
    {
      "epoch": 0.14148316881802075,
      "grad_norm": 1.4106205701828003,
      "learning_rate": 1.9834367606065315e-05,
      "loss": 0.6669,
      "step": 1118
    },
    {
      "epoch": 0.1416097190584662,
      "grad_norm": 1.681012749671936,
      "learning_rate": 1.9834042622896217e-05,
      "loss": 0.7053,
      "step": 1119
    },
    {
      "epoch": 0.14173626929891167,
      "grad_norm": 1.4903134107589722,
      "learning_rate": 1.9833717323886486e-05,
      "loss": 0.7103,
      "step": 1120
    },
    {
      "epoch": 0.14186281953935712,
      "grad_norm": 1.6424251794815063,
      "learning_rate": 1.9833391709046575e-05,
      "loss": 0.6959,
      "step": 1121
    },
    {
      "epoch": 0.1419893697798026,
      "grad_norm": 1.3932400941848755,
      "learning_rate": 1.9833065778386937e-05,
      "loss": 0.7159,
      "step": 1122
    },
    {
      "epoch": 0.14211592002024803,
      "grad_norm": 1.4303759336471558,
      "learning_rate": 1.983273953191804e-05,
      "loss": 0.663,
      "step": 1123
    },
    {
      "epoch": 0.1422424702606935,
      "grad_norm": 1.3888238668441772,
      "learning_rate": 1.9832412969650365e-05,
      "loss": 0.7339,
      "step": 1124
    },
    {
      "epoch": 0.14236902050113895,
      "grad_norm": 1.3313859701156616,
      "learning_rate": 1.9832086091594397e-05,
      "loss": 0.7199,
      "step": 1125
    },
    {
      "epoch": 0.14249557074158442,
      "grad_norm": 1.4089207649230957,
      "learning_rate": 1.983175889776064e-05,
      "loss": 0.7084,
      "step": 1126
    },
    {
      "epoch": 0.14262212098202987,
      "grad_norm": 1.4639602899551392,
      "learning_rate": 1.9831431388159597e-05,
      "loss": 0.6803,
      "step": 1127
    },
    {
      "epoch": 0.1427486712224753,
      "grad_norm": 1.4514799118041992,
      "learning_rate": 1.9831103562801788e-05,
      "loss": 0.7042,
      "step": 1128
    },
    {
      "epoch": 0.14287522146292078,
      "grad_norm": 1.3652178049087524,
      "learning_rate": 1.9830775421697743e-05,
      "loss": 0.6781,
      "step": 1129
    },
    {
      "epoch": 0.14300177170336623,
      "grad_norm": 1.7537041902542114,
      "learning_rate": 1.9830446964858e-05,
      "loss": 0.7115,
      "step": 1130
    },
    {
      "epoch": 0.1431283219438117,
      "grad_norm": 1.6503397226333618,
      "learning_rate": 1.9830118192293106e-05,
      "loss": 0.663,
      "step": 1131
    },
    {
      "epoch": 0.14325487218425714,
      "grad_norm": 1.544651746749878,
      "learning_rate": 1.9829789104013624e-05,
      "loss": 0.6651,
      "step": 1132
    },
    {
      "epoch": 0.14338142242470261,
      "grad_norm": 1.294142723083496,
      "learning_rate": 1.9829459700030123e-05,
      "loss": 0.7393,
      "step": 1133
    },
    {
      "epoch": 0.14350797266514806,
      "grad_norm": 4.539606094360352,
      "learning_rate": 1.982912998035318e-05,
      "loss": 0.7378,
      "step": 1134
    },
    {
      "epoch": 0.14363452290559353,
      "grad_norm": 1.6243301630020142,
      "learning_rate": 1.9828799944993388e-05,
      "loss": 0.6738,
      "step": 1135
    },
    {
      "epoch": 0.14376107314603898,
      "grad_norm": 1.6365278959274292,
      "learning_rate": 1.9828469593961343e-05,
      "loss": 0.7012,
      "step": 1136
    },
    {
      "epoch": 0.14388762338648445,
      "grad_norm": 1.5093907117843628,
      "learning_rate": 1.9828138927267657e-05,
      "loss": 0.6939,
      "step": 1137
    },
    {
      "epoch": 0.1440141736269299,
      "grad_norm": 1.5736477375030518,
      "learning_rate": 1.982780794492295e-05,
      "loss": 0.6918,
      "step": 1138
    },
    {
      "epoch": 0.14414072386737534,
      "grad_norm": 1.7004963159561157,
      "learning_rate": 1.982747664693785e-05,
      "loss": 0.7267,
      "step": 1139
    },
    {
      "epoch": 0.1442672741078208,
      "grad_norm": 1.6399154663085938,
      "learning_rate": 1.9827145033323e-05,
      "loss": 0.7285,
      "step": 1140
    },
    {
      "epoch": 0.14439382434826625,
      "grad_norm": 1.7531399726867676,
      "learning_rate": 1.982681310408905e-05,
      "loss": 0.7045,
      "step": 1141
    },
    {
      "epoch": 0.14452037458871173,
      "grad_norm": 1.5202357769012451,
      "learning_rate": 1.982648085924666e-05,
      "loss": 0.7367,
      "step": 1142
    },
    {
      "epoch": 0.14464692482915717,
      "grad_norm": 1.5135363340377808,
      "learning_rate": 1.98261482988065e-05,
      "loss": 0.7177,
      "step": 1143
    },
    {
      "epoch": 0.14477347506960264,
      "grad_norm": 1.5126445293426514,
      "learning_rate": 1.9825815422779257e-05,
      "loss": 0.7052,
      "step": 1144
    },
    {
      "epoch": 0.1449000253100481,
      "grad_norm": 1.5774952173233032,
      "learning_rate": 1.982548223117561e-05,
      "loss": 0.6899,
      "step": 1145
    },
    {
      "epoch": 0.14502657555049356,
      "grad_norm": 1.3692030906677246,
      "learning_rate": 1.982514872400627e-05,
      "loss": 0.6649,
      "step": 1146
    },
    {
      "epoch": 0.145153125790939,
      "grad_norm": 1.3965767621994019,
      "learning_rate": 1.9824814901281943e-05,
      "loss": 0.7049,
      "step": 1147
    },
    {
      "epoch": 0.14527967603138445,
      "grad_norm": 1.4991382360458374,
      "learning_rate": 1.9824480763013355e-05,
      "loss": 0.7294,
      "step": 1148
    },
    {
      "epoch": 0.14540622627182992,
      "grad_norm": 1.3945945501327515,
      "learning_rate": 1.9824146309211235e-05,
      "loss": 0.7111,
      "step": 1149
    },
    {
      "epoch": 0.14553277651227536,
      "grad_norm": 1.3451007604599,
      "learning_rate": 1.9823811539886324e-05,
      "loss": 0.7378,
      "step": 1150
    },
    {
      "epoch": 0.14565932675272084,
      "grad_norm": 1.3474198579788208,
      "learning_rate": 1.9823476455049374e-05,
      "loss": 0.7088,
      "step": 1151
    },
    {
      "epoch": 0.14578587699316628,
      "grad_norm": 1.4041929244995117,
      "learning_rate": 1.9823141054711148e-05,
      "loss": 0.6855,
      "step": 1152
    },
    {
      "epoch": 0.14591242723361175,
      "grad_norm": 1.4323797225952148,
      "learning_rate": 1.9822805338882422e-05,
      "loss": 0.6991,
      "step": 1153
    },
    {
      "epoch": 0.1460389774740572,
      "grad_norm": 1.4221374988555908,
      "learning_rate": 1.9822469307573967e-05,
      "loss": 0.7244,
      "step": 1154
    },
    {
      "epoch": 0.14616552771450267,
      "grad_norm": 1.2907145023345947,
      "learning_rate": 1.9822132960796586e-05,
      "loss": 0.7327,
      "step": 1155
    },
    {
      "epoch": 0.1462920779549481,
      "grad_norm": 1.5872886180877686,
      "learning_rate": 1.982179629856108e-05,
      "loss": 0.7288,
      "step": 1156
    },
    {
      "epoch": 0.14641862819539356,
      "grad_norm": 1.4552171230316162,
      "learning_rate": 1.9821459320878253e-05,
      "loss": 0.6975,
      "step": 1157
    },
    {
      "epoch": 0.14654517843583903,
      "grad_norm": 1.3317043781280518,
      "learning_rate": 1.9821122027758936e-05,
      "loss": 0.7409,
      "step": 1158
    },
    {
      "epoch": 0.14667172867628447,
      "grad_norm": 1.3776739835739136,
      "learning_rate": 1.982078441921396e-05,
      "loss": 0.7244,
      "step": 1159
    },
    {
      "epoch": 0.14679827891672995,
      "grad_norm": 1.3748347759246826,
      "learning_rate": 1.982044649525417e-05,
      "loss": 0.7455,
      "step": 1160
    },
    {
      "epoch": 0.1469248291571754,
      "grad_norm": 1.5151103734970093,
      "learning_rate": 1.9820108255890417e-05,
      "loss": 0.7159,
      "step": 1161
    },
    {
      "epoch": 0.14705137939762086,
      "grad_norm": 1.3173834085464478,
      "learning_rate": 1.9819769701133563e-05,
      "loss": 0.7267,
      "step": 1162
    },
    {
      "epoch": 0.1471779296380663,
      "grad_norm": 1.3565303087234497,
      "learning_rate": 1.981943083099448e-05,
      "loss": 0.7537,
      "step": 1163
    },
    {
      "epoch": 0.14730447987851178,
      "grad_norm": 1.3715691566467285,
      "learning_rate": 1.981909164548406e-05,
      "loss": 0.7052,
      "step": 1164
    },
    {
      "epoch": 0.14743103011895722,
      "grad_norm": 1.5895023345947266,
      "learning_rate": 1.9818752144613186e-05,
      "loss": 0.6807,
      "step": 1165
    },
    {
      "epoch": 0.1475575803594027,
      "grad_norm": 1.3169116973876953,
      "learning_rate": 1.9818412328392772e-05,
      "loss": 0.6911,
      "step": 1166
    },
    {
      "epoch": 0.14768413059984814,
      "grad_norm": 1.3031060695648193,
      "learning_rate": 1.9818072196833722e-05,
      "loss": 0.7158,
      "step": 1167
    },
    {
      "epoch": 0.14781068084029358,
      "grad_norm": 1.304284691810608,
      "learning_rate": 1.9817731749946966e-05,
      "loss": 0.6718,
      "step": 1168
    },
    {
      "epoch": 0.14793723108073906,
      "grad_norm": 1.333270788192749,
      "learning_rate": 1.9817390987743433e-05,
      "loss": 0.6952,
      "step": 1169
    },
    {
      "epoch": 0.1480637813211845,
      "grad_norm": 1.3223015069961548,
      "learning_rate": 1.9817049910234078e-05,
      "loss": 0.6514,
      "step": 1170
    },
    {
      "epoch": 0.14819033156162997,
      "grad_norm": 1.4705158472061157,
      "learning_rate": 1.981670851742984e-05,
      "loss": 0.7586,
      "step": 1171
    },
    {
      "epoch": 0.14831688180207542,
      "grad_norm": 1.3985443115234375,
      "learning_rate": 1.98163668093417e-05,
      "loss": 0.7296,
      "step": 1172
    },
    {
      "epoch": 0.1484434320425209,
      "grad_norm": 1.3800677061080933,
      "learning_rate": 1.981602478598062e-05,
      "loss": 0.7233,
      "step": 1173
    },
    {
      "epoch": 0.14856998228296633,
      "grad_norm": 1.3602551221847534,
      "learning_rate": 1.9815682447357593e-05,
      "loss": 0.7001,
      "step": 1174
    },
    {
      "epoch": 0.1486965325234118,
      "grad_norm": 1.396471619606018,
      "learning_rate": 1.9815339793483605e-05,
      "loss": 0.673,
      "step": 1175
    },
    {
      "epoch": 0.14882308276385725,
      "grad_norm": 1.2576171159744263,
      "learning_rate": 1.981499682436967e-05,
      "loss": 0.6807,
      "step": 1176
    },
    {
      "epoch": 0.1489496330043027,
      "grad_norm": 1.3613699674606323,
      "learning_rate": 1.9814653540026797e-05,
      "loss": 0.7016,
      "step": 1177
    },
    {
      "epoch": 0.14907618324474817,
      "grad_norm": 1.3201247453689575,
      "learning_rate": 1.9814309940466013e-05,
      "loss": 0.6803,
      "step": 1178
    },
    {
      "epoch": 0.1492027334851936,
      "grad_norm": 1.5779638290405273,
      "learning_rate": 1.9813966025698357e-05,
      "loss": 0.6994,
      "step": 1179
    },
    {
      "epoch": 0.14932928372563908,
      "grad_norm": 1.4408332109451294,
      "learning_rate": 1.981362179573487e-05,
      "loss": 0.7304,
      "step": 1180
    },
    {
      "epoch": 0.14945583396608453,
      "grad_norm": 1.3950140476226807,
      "learning_rate": 1.9813277250586613e-05,
      "loss": 0.6866,
      "step": 1181
    },
    {
      "epoch": 0.14958238420653,
      "grad_norm": 1.400380253791809,
      "learning_rate": 1.9812932390264646e-05,
      "loss": 0.7231,
      "step": 1182
    },
    {
      "epoch": 0.14970893444697544,
      "grad_norm": 1.3051351308822632,
      "learning_rate": 1.9812587214780046e-05,
      "loss": 0.6869,
      "step": 1183
    },
    {
      "epoch": 0.14983548468742092,
      "grad_norm": 1.3692625761032104,
      "learning_rate": 1.98122417241439e-05,
      "loss": 0.6902,
      "step": 1184
    },
    {
      "epoch": 0.14996203492786636,
      "grad_norm": 1.3521186113357544,
      "learning_rate": 1.9811895918367306e-05,
      "loss": 0.7413,
      "step": 1185
    },
    {
      "epoch": 0.15008858516831183,
      "grad_norm": 1.490319848060608,
      "learning_rate": 1.9811549797461366e-05,
      "loss": 0.6417,
      "step": 1186
    },
    {
      "epoch": 0.15021513540875728,
      "grad_norm": 1.3041462898254395,
      "learning_rate": 1.9811203361437198e-05,
      "loss": 0.7453,
      "step": 1187
    },
    {
      "epoch": 0.15034168564920272,
      "grad_norm": 1.437191367149353,
      "learning_rate": 1.9810856610305934e-05,
      "loss": 0.6874,
      "step": 1188
    },
    {
      "epoch": 0.1504682358896482,
      "grad_norm": 1.460975170135498,
      "learning_rate": 1.9810509544078706e-05,
      "loss": 0.6918,
      "step": 1189
    },
    {
      "epoch": 0.15059478613009364,
      "grad_norm": 1.413447618484497,
      "learning_rate": 1.9810162162766656e-05,
      "loss": 0.7681,
      "step": 1190
    },
    {
      "epoch": 0.1507213363705391,
      "grad_norm": 1.5216823816299438,
      "learning_rate": 1.9809814466380945e-05,
      "loss": 0.662,
      "step": 1191
    },
    {
      "epoch": 0.15084788661098456,
      "grad_norm": 1.3892854452133179,
      "learning_rate": 1.9809466454932744e-05,
      "loss": 0.7272,
      "step": 1192
    },
    {
      "epoch": 0.15097443685143003,
      "grad_norm": 1.780741810798645,
      "learning_rate": 1.980911812843323e-05,
      "loss": 0.7525,
      "step": 1193
    },
    {
      "epoch": 0.15110098709187547,
      "grad_norm": 1.5574004650115967,
      "learning_rate": 1.980876948689358e-05,
      "loss": 0.7027,
      "step": 1194
    },
    {
      "epoch": 0.15122753733232094,
      "grad_norm": 1.8690134286880493,
      "learning_rate": 1.9808420530325004e-05,
      "loss": 0.6886,
      "step": 1195
    },
    {
      "epoch": 0.1513540875727664,
      "grad_norm": 1.2087221145629883,
      "learning_rate": 1.98080712587387e-05,
      "loss": 0.7136,
      "step": 1196
    },
    {
      "epoch": 0.15148063781321183,
      "grad_norm": 1.8461849689483643,
      "learning_rate": 1.980772167214589e-05,
      "loss": 0.7742,
      "step": 1197
    },
    {
      "epoch": 0.1516071880536573,
      "grad_norm": 1.271973967552185,
      "learning_rate": 1.9807371770557802e-05,
      "loss": 0.6975,
      "step": 1198
    },
    {
      "epoch": 0.15173373829410275,
      "grad_norm": 1.9026466608047485,
      "learning_rate": 1.9807021553985673e-05,
      "loss": 0.74,
      "step": 1199
    },
    {
      "epoch": 0.15186028853454822,
      "grad_norm": 1.2797188758850098,
      "learning_rate": 1.980667102244075e-05,
      "loss": 0.737,
      "step": 1200
    },
    {
      "epoch": 0.15198683877499367,
      "grad_norm": 2.34269642829895,
      "learning_rate": 1.980632017593429e-05,
      "loss": 0.6872,
      "step": 1201
    },
    {
      "epoch": 0.15211338901543914,
      "grad_norm": 1.694174885749817,
      "learning_rate": 1.9805969014477567e-05,
      "loss": 0.8005,
      "step": 1202
    },
    {
      "epoch": 0.15223993925588458,
      "grad_norm": 1.628565788269043,
      "learning_rate": 1.9805617538081853e-05,
      "loss": 0.7925,
      "step": 1203
    },
    {
      "epoch": 0.15236648949633005,
      "grad_norm": 1.3031736612319946,
      "learning_rate": 1.980526574675844e-05,
      "loss": 0.6836,
      "step": 1204
    },
    {
      "epoch": 0.1524930397367755,
      "grad_norm": 1.5224523544311523,
      "learning_rate": 1.9804913640518623e-05,
      "loss": 0.6917,
      "step": 1205
    },
    {
      "epoch": 0.15261958997722094,
      "grad_norm": 1.2927451133728027,
      "learning_rate": 1.9804561219373714e-05,
      "loss": 0.71,
      "step": 1206
    },
    {
      "epoch": 0.15274614021766642,
      "grad_norm": 1.2862625122070312,
      "learning_rate": 1.980420848333503e-05,
      "loss": 0.7056,
      "step": 1207
    },
    {
      "epoch": 0.15287269045811186,
      "grad_norm": 1.5093903541564941,
      "learning_rate": 1.98038554324139e-05,
      "loss": 0.7362,
      "step": 1208
    },
    {
      "epoch": 0.15299924069855733,
      "grad_norm": 1.2472777366638184,
      "learning_rate": 1.9803502066621665e-05,
      "loss": 0.7175,
      "step": 1209
    },
    {
      "epoch": 0.15312579093900278,
      "grad_norm": 1.809510588645935,
      "learning_rate": 1.980314838596967e-05,
      "loss": 0.7106,
      "step": 1210
    },
    {
      "epoch": 0.15325234117944825,
      "grad_norm": 1.2891818284988403,
      "learning_rate": 1.980279439046928e-05,
      "loss": 0.6989,
      "step": 1211
    },
    {
      "epoch": 0.1533788914198937,
      "grad_norm": 1.5668342113494873,
      "learning_rate": 1.980244008013186e-05,
      "loss": 0.7286,
      "step": 1212
    },
    {
      "epoch": 0.15350544166033916,
      "grad_norm": 1.3399709463119507,
      "learning_rate": 1.9802085454968786e-05,
      "loss": 0.7307,
      "step": 1213
    },
    {
      "epoch": 0.1536319919007846,
      "grad_norm": 1.4724574089050293,
      "learning_rate": 1.9801730514991457e-05,
      "loss": 0.7019,
      "step": 1214
    },
    {
      "epoch": 0.15375854214123008,
      "grad_norm": 1.2500977516174316,
      "learning_rate": 1.9801375260211267e-05,
      "loss": 0.6877,
      "step": 1215
    },
    {
      "epoch": 0.15388509238167553,
      "grad_norm": 1.6199941635131836,
      "learning_rate": 1.9801019690639624e-05,
      "loss": 0.7283,
      "step": 1216
    },
    {
      "epoch": 0.15401164262212097,
      "grad_norm": 1.3434911966323853,
      "learning_rate": 1.9800663806287952e-05,
      "loss": 0.7285,
      "step": 1217
    },
    {
      "epoch": 0.15413819286256644,
      "grad_norm": 1.2925642728805542,
      "learning_rate": 1.9800307607167677e-05,
      "loss": 0.6523,
      "step": 1218
    },
    {
      "epoch": 0.1542647431030119,
      "grad_norm": 1.5035452842712402,
      "learning_rate": 1.9799951093290243e-05,
      "loss": 0.7102,
      "step": 1219
    },
    {
      "epoch": 0.15439129334345736,
      "grad_norm": 1.3362501859664917,
      "learning_rate": 1.9799594264667096e-05,
      "loss": 0.7283,
      "step": 1220
    },
    {
      "epoch": 0.1545178435839028,
      "grad_norm": 1.6003369092941284,
      "learning_rate": 1.9799237121309697e-05,
      "loss": 0.7304,
      "step": 1221
    },
    {
      "epoch": 0.15464439382434828,
      "grad_norm": 1.7702730894088745,
      "learning_rate": 1.9798879663229522e-05,
      "loss": 0.7569,
      "step": 1222
    },
    {
      "epoch": 0.15477094406479372,
      "grad_norm": 1.3201602697372437,
      "learning_rate": 1.9798521890438047e-05,
      "loss": 0.6885,
      "step": 1223
    },
    {
      "epoch": 0.1548974943052392,
      "grad_norm": 1.3022280931472778,
      "learning_rate": 1.9798163802946765e-05,
      "loss": 0.6707,
      "step": 1224
    },
    {
      "epoch": 0.15502404454568464,
      "grad_norm": 1.2472718954086304,
      "learning_rate": 1.979780540076717e-05,
      "loss": 0.7396,
      "step": 1225
    },
    {
      "epoch": 0.15515059478613008,
      "grad_norm": 1.3867162466049194,
      "learning_rate": 1.979744668391078e-05,
      "loss": 0.6744,
      "step": 1226
    },
    {
      "epoch": 0.15527714502657555,
      "grad_norm": 1.3982288837432861,
      "learning_rate": 1.979708765238911e-05,
      "loss": 0.6867,
      "step": 1227
    },
    {
      "epoch": 0.155403695267021,
      "grad_norm": 1.3572982549667358,
      "learning_rate": 1.97967283062137e-05,
      "loss": 0.7896,
      "step": 1228
    },
    {
      "epoch": 0.15553024550746647,
      "grad_norm": 1.88454270362854,
      "learning_rate": 1.979636864539608e-05,
      "loss": 0.7773,
      "step": 1229
    },
    {
      "epoch": 0.1556567957479119,
      "grad_norm": 1.31227445602417,
      "learning_rate": 1.9796008669947814e-05,
      "loss": 0.7225,
      "step": 1230
    },
    {
      "epoch": 0.15578334598835739,
      "grad_norm": 1.652184009552002,
      "learning_rate": 1.9795648379880452e-05,
      "loss": 0.7409,
      "step": 1231
    },
    {
      "epoch": 0.15590989622880283,
      "grad_norm": 1.3673312664031982,
      "learning_rate": 1.979528777520557e-05,
      "loss": 0.7335,
      "step": 1232
    },
    {
      "epoch": 0.1560364464692483,
      "grad_norm": 1.276179313659668,
      "learning_rate": 1.9794926855934752e-05,
      "loss": 0.7194,
      "step": 1233
    },
    {
      "epoch": 0.15616299670969375,
      "grad_norm": 1.4279032945632935,
      "learning_rate": 1.9794565622079583e-05,
      "loss": 0.7185,
      "step": 1234
    },
    {
      "epoch": 0.15628954695013922,
      "grad_norm": 1.3168892860412598,
      "learning_rate": 1.979420407365167e-05,
      "loss": 0.6925,
      "step": 1235
    },
    {
      "epoch": 0.15641609719058466,
      "grad_norm": 1.4966228008270264,
      "learning_rate": 1.9793842210662625e-05,
      "loss": 0.7421,
      "step": 1236
    },
    {
      "epoch": 0.1565426474310301,
      "grad_norm": 1.5230294466018677,
      "learning_rate": 1.979348003312407e-05,
      "loss": 0.6828,
      "step": 1237
    },
    {
      "epoch": 0.15666919767147558,
      "grad_norm": 1.3253307342529297,
      "learning_rate": 1.9793117541047633e-05,
      "loss": 0.6771,
      "step": 1238
    },
    {
      "epoch": 0.15679574791192102,
      "grad_norm": 1.7673194408416748,
      "learning_rate": 1.979275473444496e-05,
      "loss": 0.6965,
      "step": 1239
    },
    {
      "epoch": 0.1569222981523665,
      "grad_norm": 1.2240883111953735,
      "learning_rate": 1.9792391613327703e-05,
      "loss": 0.6937,
      "step": 1240
    },
    {
      "epoch": 0.15704884839281194,
      "grad_norm": 1.626938819885254,
      "learning_rate": 1.9792028177707522e-05,
      "loss": 0.6881,
      "step": 1241
    },
    {
      "epoch": 0.1571753986332574,
      "grad_norm": 13.830275535583496,
      "learning_rate": 1.979166442759609e-05,
      "loss": 0.7706,
      "step": 1242
    },
    {
      "epoch": 0.15730194887370286,
      "grad_norm": 2.3513882160186768,
      "learning_rate": 1.9791300363005095e-05,
      "loss": 0.6622,
      "step": 1243
    },
    {
      "epoch": 0.15742849911414833,
      "grad_norm": 1.2931212186813354,
      "learning_rate": 1.979093598394622e-05,
      "loss": 0.7086,
      "step": 1244
    },
    {
      "epoch": 0.15755504935459377,
      "grad_norm": 1.7150992155075073,
      "learning_rate": 1.9790571290431176e-05,
      "loss": 0.6994,
      "step": 1245
    },
    {
      "epoch": 0.15768159959503922,
      "grad_norm": 1.3131043910980225,
      "learning_rate": 1.9790206282471677e-05,
      "loss": 0.6978,
      "step": 1246
    },
    {
      "epoch": 0.1578081498354847,
      "grad_norm": 1.3990962505340576,
      "learning_rate": 1.9789840960079436e-05,
      "loss": 0.7173,
      "step": 1247
    },
    {
      "epoch": 0.15793470007593013,
      "grad_norm": 1.3620359897613525,
      "learning_rate": 1.9789475323266193e-05,
      "loss": 0.7162,
      "step": 1248
    },
    {
      "epoch": 0.1580612503163756,
      "grad_norm": 1.3808149099349976,
      "learning_rate": 1.9789109372043692e-05,
      "loss": 0.7483,
      "step": 1249
    },
    {
      "epoch": 0.15818780055682105,
      "grad_norm": 1.8259354829788208,
      "learning_rate": 1.9788743106423683e-05,
      "loss": 0.6861,
      "step": 1250
    },
    {
      "epoch": 0.15831435079726652,
      "grad_norm": 2.716019630432129,
      "learning_rate": 1.9788376526417932e-05,
      "loss": 0.6906,
      "step": 1251
    },
    {
      "epoch": 0.15844090103771197,
      "grad_norm": 1.8369535207748413,
      "learning_rate": 1.9788009632038213e-05,
      "loss": 0.7289,
      "step": 1252
    },
    {
      "epoch": 0.15856745127815744,
      "grad_norm": 1.4988173246383667,
      "learning_rate": 1.9787642423296308e-05,
      "loss": 0.6789,
      "step": 1253
    },
    {
      "epoch": 0.15869400151860288,
      "grad_norm": 1.5819765329360962,
      "learning_rate": 1.9787274900204006e-05,
      "loss": 0.7059,
      "step": 1254
    },
    {
      "epoch": 0.15882055175904833,
      "grad_norm": 1.236707091331482,
      "learning_rate": 1.9786907062773118e-05,
      "loss": 0.7125,
      "step": 1255
    },
    {
      "epoch": 0.1589471019994938,
      "grad_norm": 1.6258342266082764,
      "learning_rate": 1.9786538911015456e-05,
      "loss": 0.6732,
      "step": 1256
    },
    {
      "epoch": 0.15907365223993924,
      "grad_norm": 1.2907849550247192,
      "learning_rate": 1.978617044494284e-05,
      "loss": 0.7174,
      "step": 1257
    },
    {
      "epoch": 0.15920020248038472,
      "grad_norm": 1.4665979146957397,
      "learning_rate": 1.978580166456711e-05,
      "loss": 0.7278,
      "step": 1258
    },
    {
      "epoch": 0.15932675272083016,
      "grad_norm": 1.3249282836914062,
      "learning_rate": 1.978543256990011e-05,
      "loss": 0.6379,
      "step": 1259
    },
    {
      "epoch": 0.15945330296127563,
      "grad_norm": 1.5587635040283203,
      "learning_rate": 1.978506316095369e-05,
      "loss": 0.7018,
      "step": 1260
    },
    {
      "epoch": 0.15957985320172108,
      "grad_norm": 2.16178560256958,
      "learning_rate": 1.9784693437739717e-05,
      "loss": 0.7088,
      "step": 1261
    },
    {
      "epoch": 0.15970640344216655,
      "grad_norm": 1.6479535102844238,
      "learning_rate": 1.9784323400270064e-05,
      "loss": 0.7119,
      "step": 1262
    },
    {
      "epoch": 0.159832953682612,
      "grad_norm": 1.8855435848236084,
      "learning_rate": 1.978395304855661e-05,
      "loss": 0.6865,
      "step": 1263
    },
    {
      "epoch": 0.15995950392305747,
      "grad_norm": 1.9203925132751465,
      "learning_rate": 1.9783582382611263e-05,
      "loss": 0.6958,
      "step": 1264
    },
    {
      "epoch": 0.1600860541635029,
      "grad_norm": 1.4652434587478638,
      "learning_rate": 1.978321140244592e-05,
      "loss": 0.7103,
      "step": 1265
    },
    {
      "epoch": 0.16021260440394836,
      "grad_norm": 1.4174835681915283,
      "learning_rate": 1.9782840108072492e-05,
      "loss": 0.6547,
      "step": 1266
    },
    {
      "epoch": 0.16033915464439383,
      "grad_norm": 1.5422433614730835,
      "learning_rate": 1.978246849950291e-05,
      "loss": 0.6876,
      "step": 1267
    },
    {
      "epoch": 0.16046570488483927,
      "grad_norm": 1.2734493017196655,
      "learning_rate": 1.978209657674911e-05,
      "loss": 0.7207,
      "step": 1268
    },
    {
      "epoch": 0.16059225512528474,
      "grad_norm": 1.7123507261276245,
      "learning_rate": 1.9781724339823028e-05,
      "loss": 0.6763,
      "step": 1269
    },
    {
      "epoch": 0.1607188053657302,
      "grad_norm": 1.3318103551864624,
      "learning_rate": 1.9781351788736626e-05,
      "loss": 0.6901,
      "step": 1270
    },
    {
      "epoch": 0.16084535560617566,
      "grad_norm": 1.6476786136627197,
      "learning_rate": 1.978097892350187e-05,
      "loss": 0.6894,
      "step": 1271
    },
    {
      "epoch": 0.1609719058466211,
      "grad_norm": 1.285781741142273,
      "learning_rate": 1.9780605744130735e-05,
      "loss": 0.7088,
      "step": 1272
    },
    {
      "epoch": 0.16109845608706658,
      "grad_norm": 1.4896700382232666,
      "learning_rate": 1.9780232250635204e-05,
      "loss": 0.6943,
      "step": 1273
    },
    {
      "epoch": 0.16122500632751202,
      "grad_norm": 1.4769500494003296,
      "learning_rate": 1.9779858443027273e-05,
      "loss": 0.7365,
      "step": 1274
    },
    {
      "epoch": 0.16135155656795747,
      "grad_norm": 1.6431047916412354,
      "learning_rate": 1.9779484321318952e-05,
      "loss": 0.7626,
      "step": 1275
    },
    {
      "epoch": 0.16147810680840294,
      "grad_norm": 1.286101222038269,
      "learning_rate": 1.977910988552225e-05,
      "loss": 0.7429,
      "step": 1276
    },
    {
      "epoch": 0.16160465704884838,
      "grad_norm": 1.3987494707107544,
      "learning_rate": 1.9778735135649192e-05,
      "loss": 0.6754,
      "step": 1277
    },
    {
      "epoch": 0.16173120728929385,
      "grad_norm": 1.236694574356079,
      "learning_rate": 1.9778360071711823e-05,
      "loss": 0.7179,
      "step": 1278
    },
    {
      "epoch": 0.1618577575297393,
      "grad_norm": 1.283210039138794,
      "learning_rate": 1.9777984693722185e-05,
      "loss": 0.7135,
      "step": 1279
    },
    {
      "epoch": 0.16198430777018477,
      "grad_norm": 1.381028652191162,
      "learning_rate": 1.977760900169233e-05,
      "loss": 0.7005,
      "step": 1280
    },
    {
      "epoch": 0.16211085801063022,
      "grad_norm": 1.2387903928756714,
      "learning_rate": 1.9777232995634326e-05,
      "loss": 0.7202,
      "step": 1281
    },
    {
      "epoch": 0.1622374082510757,
      "grad_norm": 1.3638108968734741,
      "learning_rate": 1.977685667556025e-05,
      "loss": 0.7359,
      "step": 1282
    },
    {
      "epoch": 0.16236395849152113,
      "grad_norm": 1.3051482439041138,
      "learning_rate": 1.977648004148219e-05,
      "loss": 0.7145,
      "step": 1283
    },
    {
      "epoch": 0.1624905087319666,
      "grad_norm": 1.2551755905151367,
      "learning_rate": 1.9776103093412237e-05,
      "loss": 0.6786,
      "step": 1284
    },
    {
      "epoch": 0.16261705897241205,
      "grad_norm": 1.2534486055374146,
      "learning_rate": 1.9775725831362505e-05,
      "loss": 0.6772,
      "step": 1285
    },
    {
      "epoch": 0.1627436092128575,
      "grad_norm": 1.3088812828063965,
      "learning_rate": 1.9775348255345107e-05,
      "loss": 0.7534,
      "step": 1286
    },
    {
      "epoch": 0.16287015945330297,
      "grad_norm": 1.2970775365829468,
      "learning_rate": 1.9774970365372165e-05,
      "loss": 0.7098,
      "step": 1287
    },
    {
      "epoch": 0.1629967096937484,
      "grad_norm": 1.254194736480713,
      "learning_rate": 1.9774592161455827e-05,
      "loss": 0.7041,
      "step": 1288
    },
    {
      "epoch": 0.16312325993419388,
      "grad_norm": 1.3945815563201904,
      "learning_rate": 1.9774213643608227e-05,
      "loss": 0.6951,
      "step": 1289
    },
    {
      "epoch": 0.16324981017463933,
      "grad_norm": 1.3243417739868164,
      "learning_rate": 1.9773834811841527e-05,
      "loss": 0.6995,
      "step": 1290
    },
    {
      "epoch": 0.1633763604150848,
      "grad_norm": 1.2912318706512451,
      "learning_rate": 1.97734556661679e-05,
      "loss": 0.6984,
      "step": 1291
    },
    {
      "epoch": 0.16350291065553024,
      "grad_norm": 1.27584969997406,
      "learning_rate": 1.9773076206599515e-05,
      "loss": 0.6808,
      "step": 1292
    },
    {
      "epoch": 0.16362946089597571,
      "grad_norm": 1.5523064136505127,
      "learning_rate": 1.977269643314856e-05,
      "loss": 0.6988,
      "step": 1293
    },
    {
      "epoch": 0.16375601113642116,
      "grad_norm": 1.2604564428329468,
      "learning_rate": 1.9772316345827236e-05,
      "loss": 0.6832,
      "step": 1294
    },
    {
      "epoch": 0.1638825613768666,
      "grad_norm": 1.5310419797897339,
      "learning_rate": 1.9771935944647747e-05,
      "loss": 0.6942,
      "step": 1295
    },
    {
      "epoch": 0.16400911161731208,
      "grad_norm": 1.506679892539978,
      "learning_rate": 1.9771555229622313e-05,
      "loss": 0.7585,
      "step": 1296
    },
    {
      "epoch": 0.16413566185775752,
      "grad_norm": 1.3438594341278076,
      "learning_rate": 1.977117420076316e-05,
      "loss": 0.6739,
      "step": 1297
    },
    {
      "epoch": 0.164262212098203,
      "grad_norm": 1.562716007232666,
      "learning_rate": 1.9770792858082525e-05,
      "loss": 0.7231,
      "step": 1298
    },
    {
      "epoch": 0.16438876233864844,
      "grad_norm": 1.2477607727050781,
      "learning_rate": 1.9770411201592655e-05,
      "loss": 0.6961,
      "step": 1299
    },
    {
      "epoch": 0.1645153125790939,
      "grad_norm": 1.3825671672821045,
      "learning_rate": 1.977002923130581e-05,
      "loss": 0.7398,
      "step": 1300
    },
    {
      "epoch": 0.16464186281953935,
      "grad_norm": 1.297499179840088,
      "learning_rate": 1.9769646947234262e-05,
      "loss": 0.7128,
      "step": 1301
    },
    {
      "epoch": 0.16476841305998483,
      "grad_norm": 1.3222262859344482,
      "learning_rate": 1.9769264349390278e-05,
      "loss": 0.7007,
      "step": 1302
    },
    {
      "epoch": 0.16489496330043027,
      "grad_norm": 1.461938738822937,
      "learning_rate": 1.9768881437786154e-05,
      "loss": 0.7282,
      "step": 1303
    },
    {
      "epoch": 0.16502151354087571,
      "grad_norm": 1.2663376331329346,
      "learning_rate": 1.9768498212434183e-05,
      "loss": 0.7391,
      "step": 1304
    },
    {
      "epoch": 0.16514806378132119,
      "grad_norm": 1.3202067613601685,
      "learning_rate": 1.9768114673346676e-05,
      "loss": 0.7242,
      "step": 1305
    },
    {
      "epoch": 0.16527461402176663,
      "grad_norm": 1.2892793416976929,
      "learning_rate": 1.9767730820535953e-05,
      "loss": 0.6778,
      "step": 1306
    },
    {
      "epoch": 0.1654011642622121,
      "grad_norm": 1.2282859086990356,
      "learning_rate": 1.976734665401434e-05,
      "loss": 0.6883,
      "step": 1307
    },
    {
      "epoch": 0.16552771450265755,
      "grad_norm": 1.1947331428527832,
      "learning_rate": 1.9766962173794174e-05,
      "loss": 0.7313,
      "step": 1308
    },
    {
      "epoch": 0.16565426474310302,
      "grad_norm": 1.312265157699585,
      "learning_rate": 1.9766577379887804e-05,
      "loss": 0.7184,
      "step": 1309
    },
    {
      "epoch": 0.16578081498354846,
      "grad_norm": 1.35601007938385,
      "learning_rate": 1.9766192272307593e-05,
      "loss": 0.6855,
      "step": 1310
    },
    {
      "epoch": 0.16590736522399394,
      "grad_norm": 1.3649295568466187,
      "learning_rate": 1.97658068510659e-05,
      "loss": 0.7244,
      "step": 1311
    },
    {
      "epoch": 0.16603391546443938,
      "grad_norm": 1.2974942922592163,
      "learning_rate": 1.9765421116175112e-05,
      "loss": 0.6862,
      "step": 1312
    },
    {
      "epoch": 0.16616046570488485,
      "grad_norm": 1.2533499002456665,
      "learning_rate": 1.9765035067647616e-05,
      "loss": 0.7377,
      "step": 1313
    },
    {
      "epoch": 0.1662870159453303,
      "grad_norm": 1.2447832822799683,
      "learning_rate": 1.9764648705495807e-05,
      "loss": 0.6894,
      "step": 1314
    },
    {
      "epoch": 0.16641356618577574,
      "grad_norm": 1.459912896156311,
      "learning_rate": 1.97642620297321e-05,
      "loss": 0.6876,
      "step": 1315
    },
    {
      "epoch": 0.1665401164262212,
      "grad_norm": 1.2897275686264038,
      "learning_rate": 1.9763875040368906e-05,
      "loss": 0.6724,
      "step": 1316
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 1.2195154428482056,
      "learning_rate": 1.976348773741866e-05,
      "loss": 0.7251,
      "step": 1317
    },
    {
      "epoch": 0.16679321690711213,
      "grad_norm": 1.397018313407898,
      "learning_rate": 1.97631001208938e-05,
      "loss": 0.7574,
      "step": 1318
    },
    {
      "epoch": 0.16691976714755757,
      "grad_norm": 1.2579833269119263,
      "learning_rate": 1.9762712190806778e-05,
      "loss": 0.7012,
      "step": 1319
    },
    {
      "epoch": 0.16704631738800305,
      "grad_norm": 1.4021692276000977,
      "learning_rate": 1.9762323947170045e-05,
      "loss": 0.7125,
      "step": 1320
    },
    {
      "epoch": 0.1671728676284485,
      "grad_norm": 1.1936827898025513,
      "learning_rate": 1.9761935389996077e-05,
      "loss": 0.6801,
      "step": 1321
    },
    {
      "epoch": 0.16729941786889396,
      "grad_norm": 1.282652735710144,
      "learning_rate": 1.976154651929735e-05,
      "loss": 0.6807,
      "step": 1322
    },
    {
      "epoch": 0.1674259681093394,
      "grad_norm": 1.4512253999710083,
      "learning_rate": 1.9761157335086355e-05,
      "loss": 0.7401,
      "step": 1323
    },
    {
      "epoch": 0.16755251834978485,
      "grad_norm": 1.2687034606933594,
      "learning_rate": 1.976076783737559e-05,
      "loss": 0.7252,
      "step": 1324
    },
    {
      "epoch": 0.16767906859023032,
      "grad_norm": 1.2657051086425781,
      "learning_rate": 1.9760378026177568e-05,
      "loss": 0.6872,
      "step": 1325
    },
    {
      "epoch": 0.16780561883067577,
      "grad_norm": 1.215227484703064,
      "learning_rate": 1.9759987901504804e-05,
      "loss": 0.6958,
      "step": 1326
    },
    {
      "epoch": 0.16793216907112124,
      "grad_norm": 1.2847038507461548,
      "learning_rate": 1.975959746336983e-05,
      "loss": 0.694,
      "step": 1327
    },
    {
      "epoch": 0.16805871931156668,
      "grad_norm": 1.163077712059021,
      "learning_rate": 1.975920671178519e-05,
      "loss": 0.6955,
      "step": 1328
    },
    {
      "epoch": 0.16818526955201216,
      "grad_norm": 1.4300591945648193,
      "learning_rate": 1.9758815646763424e-05,
      "loss": 0.6983,
      "step": 1329
    },
    {
      "epoch": 0.1683118197924576,
      "grad_norm": 1.273438811302185,
      "learning_rate": 1.97584242683171e-05,
      "loss": 0.6972,
      "step": 1330
    },
    {
      "epoch": 0.16843837003290307,
      "grad_norm": 1.4212759733200073,
      "learning_rate": 1.9758032576458786e-05,
      "loss": 0.749,
      "step": 1331
    },
    {
      "epoch": 0.16856492027334852,
      "grad_norm": 1.3870741128921509,
      "learning_rate": 1.9757640571201057e-05,
      "loss": 0.6698,
      "step": 1332
    },
    {
      "epoch": 0.168691470513794,
      "grad_norm": 1.2818174362182617,
      "learning_rate": 1.975724825255651e-05,
      "loss": 0.7403,
      "step": 1333
    },
    {
      "epoch": 0.16881802075423943,
      "grad_norm": 1.1581439971923828,
      "learning_rate": 1.9756855620537744e-05,
      "loss": 0.6884,
      "step": 1334
    },
    {
      "epoch": 0.16894457099468488,
      "grad_norm": 1.175638198852539,
      "learning_rate": 1.9756462675157364e-05,
      "loss": 0.6561,
      "step": 1335
    },
    {
      "epoch": 0.16907112123513035,
      "grad_norm": 1.259161114692688,
      "learning_rate": 1.9756069416427998e-05,
      "loss": 0.681,
      "step": 1336
    },
    {
      "epoch": 0.1691976714755758,
      "grad_norm": 1.3485016822814941,
      "learning_rate": 1.975567584436227e-05,
      "loss": 0.6965,
      "step": 1337
    },
    {
      "epoch": 0.16932422171602127,
      "grad_norm": 1.2628765106201172,
      "learning_rate": 1.975528195897282e-05,
      "loss": 0.7278,
      "step": 1338
    },
    {
      "epoch": 0.1694507719564667,
      "grad_norm": 1.2986897230148315,
      "learning_rate": 1.97548877602723e-05,
      "loss": 0.6744,
      "step": 1339
    },
    {
      "epoch": 0.16957732219691218,
      "grad_norm": 1.2808492183685303,
      "learning_rate": 1.975449324827338e-05,
      "loss": 0.7303,
      "step": 1340
    },
    {
      "epoch": 0.16970387243735763,
      "grad_norm": 1.2651822566986084,
      "learning_rate": 1.9754098422988713e-05,
      "loss": 0.7324,
      "step": 1341
    },
    {
      "epoch": 0.1698304226778031,
      "grad_norm": 1.389469861984253,
      "learning_rate": 1.9753703284430992e-05,
      "loss": 0.6925,
      "step": 1342
    },
    {
      "epoch": 0.16995697291824854,
      "grad_norm": 1.265027642250061,
      "learning_rate": 1.9753307832612908e-05,
      "loss": 0.7306,
      "step": 1343
    },
    {
      "epoch": 0.170083523158694,
      "grad_norm": 1.3813601732254028,
      "learning_rate": 1.9752912067547152e-05,
      "loss": 0.728,
      "step": 1344
    },
    {
      "epoch": 0.17021007339913946,
      "grad_norm": 1.2720451354980469,
      "learning_rate": 1.9752515989246444e-05,
      "loss": 0.704,
      "step": 1345
    },
    {
      "epoch": 0.1703366236395849,
      "grad_norm": 1.2904952764511108,
      "learning_rate": 1.97521195977235e-05,
      "loss": 0.7424,
      "step": 1346
    },
    {
      "epoch": 0.17046317388003038,
      "grad_norm": 1.3648642301559448,
      "learning_rate": 1.9751722892991054e-05,
      "loss": 0.7451,
      "step": 1347
    },
    {
      "epoch": 0.17058972412047582,
      "grad_norm": 2.368821859359741,
      "learning_rate": 1.9751325875061847e-05,
      "loss": 0.7483,
      "step": 1348
    },
    {
      "epoch": 0.1707162743609213,
      "grad_norm": 1.3269779682159424,
      "learning_rate": 1.975092854394863e-05,
      "loss": 0.7125,
      "step": 1349
    },
    {
      "epoch": 0.17084282460136674,
      "grad_norm": 1.276856541633606,
      "learning_rate": 1.9750530899664158e-05,
      "loss": 0.6943,
      "step": 1350
    },
    {
      "epoch": 0.1709693748418122,
      "grad_norm": 1.4871817827224731,
      "learning_rate": 1.9750132942221212e-05,
      "loss": 0.7647,
      "step": 1351
    },
    {
      "epoch": 0.17109592508225765,
      "grad_norm": 1.2924468517303467,
      "learning_rate": 1.9749734671632564e-05,
      "loss": 0.7396,
      "step": 1352
    },
    {
      "epoch": 0.1712224753227031,
      "grad_norm": 1.2759222984313965,
      "learning_rate": 1.9749336087911015e-05,
      "loss": 0.6886,
      "step": 1353
    },
    {
      "epoch": 0.17134902556314857,
      "grad_norm": 1.5031390190124512,
      "learning_rate": 1.9748937191069353e-05,
      "loss": 0.7316,
      "step": 1354
    },
    {
      "epoch": 0.17147557580359402,
      "grad_norm": 1.2903990745544434,
      "learning_rate": 1.97485379811204e-05,
      "loss": 0.7005,
      "step": 1355
    },
    {
      "epoch": 0.1716021260440395,
      "grad_norm": 1.3040269613265991,
      "learning_rate": 1.974813845807698e-05,
      "loss": 0.7041,
      "step": 1356
    },
    {
      "epoch": 0.17172867628448493,
      "grad_norm": 1.298585295677185,
      "learning_rate": 1.9747738621951913e-05,
      "loss": 0.7249,
      "step": 1357
    },
    {
      "epoch": 0.1718552265249304,
      "grad_norm": 1.2083685398101807,
      "learning_rate": 1.974733847275805e-05,
      "loss": 0.7127,
      "step": 1358
    },
    {
      "epoch": 0.17198177676537585,
      "grad_norm": 1.2665727138519287,
      "learning_rate": 1.974693801050824e-05,
      "loss": 0.7294,
      "step": 1359
    },
    {
      "epoch": 0.17210832700582132,
      "grad_norm": 1.2917219400405884,
      "learning_rate": 1.9746537235215343e-05,
      "loss": 0.7333,
      "step": 1360
    },
    {
      "epoch": 0.17223487724626677,
      "grad_norm": 1.2375973463058472,
      "learning_rate": 1.9746136146892232e-05,
      "loss": 0.7098,
      "step": 1361
    },
    {
      "epoch": 0.17236142748671224,
      "grad_norm": 1.3495126962661743,
      "learning_rate": 1.974573474555179e-05,
      "loss": 0.6887,
      "step": 1362
    },
    {
      "epoch": 0.17248797772715768,
      "grad_norm": 1.2476613521575928,
      "learning_rate": 1.9745333031206905e-05,
      "loss": 0.7065,
      "step": 1363
    },
    {
      "epoch": 0.17261452796760313,
      "grad_norm": 1.248517394065857,
      "learning_rate": 1.9744931003870483e-05,
      "loss": 0.6781,
      "step": 1364
    },
    {
      "epoch": 0.1727410782080486,
      "grad_norm": 1.2442021369934082,
      "learning_rate": 1.9744528663555433e-05,
      "loss": 0.6879,
      "step": 1365
    },
    {
      "epoch": 0.17286762844849404,
      "grad_norm": 1.28376042842865,
      "learning_rate": 1.974412601027468e-05,
      "loss": 0.727,
      "step": 1366
    },
    {
      "epoch": 0.17299417868893951,
      "grad_norm": 1.4384384155273438,
      "learning_rate": 1.974372304404115e-05,
      "loss": 0.7459,
      "step": 1367
    },
    {
      "epoch": 0.17312072892938496,
      "grad_norm": 1.6707600355148315,
      "learning_rate": 1.9743319764867793e-05,
      "loss": 0.7503,
      "step": 1368
    },
    {
      "epoch": 0.17324727916983043,
      "grad_norm": 1.4593451023101807,
      "learning_rate": 1.974291617276756e-05,
      "loss": 0.753,
      "step": 1369
    },
    {
      "epoch": 0.17337382941027588,
      "grad_norm": 1.6350181102752686,
      "learning_rate": 1.9742512267753405e-05,
      "loss": 0.6891,
      "step": 1370
    },
    {
      "epoch": 0.17350037965072135,
      "grad_norm": 1.3516654968261719,
      "learning_rate": 1.974210804983831e-05,
      "loss": 0.7239,
      "step": 1371
    },
    {
      "epoch": 0.1736269298911668,
      "grad_norm": 1.8106036186218262,
      "learning_rate": 1.9741703519035253e-05,
      "loss": 0.7431,
      "step": 1372
    },
    {
      "epoch": 0.17375348013161224,
      "grad_norm": 1.3177207708358765,
      "learning_rate": 1.9741298675357226e-05,
      "loss": 0.6642,
      "step": 1373
    },
    {
      "epoch": 0.1738800303720577,
      "grad_norm": 1.4837316274642944,
      "learning_rate": 1.974089351881723e-05,
      "loss": 0.7133,
      "step": 1374
    },
    {
      "epoch": 0.17400658061250315,
      "grad_norm": 2.6417932510375977,
      "learning_rate": 1.9740488049428285e-05,
      "loss": 0.755,
      "step": 1375
    },
    {
      "epoch": 0.17413313085294863,
      "grad_norm": 1.3808915615081787,
      "learning_rate": 1.9740082267203406e-05,
      "loss": 0.683,
      "step": 1376
    },
    {
      "epoch": 0.17425968109339407,
      "grad_norm": 1.5025885105133057,
      "learning_rate": 1.9739676172155622e-05,
      "loss": 0.775,
      "step": 1377
    },
    {
      "epoch": 0.17438623133383954,
      "grad_norm": 1.3967547416687012,
      "learning_rate": 1.9739269764297985e-05,
      "loss": 0.7127,
      "step": 1378
    },
    {
      "epoch": 0.174512781574285,
      "grad_norm": 1.646643877029419,
      "learning_rate": 1.9738863043643546e-05,
      "loss": 0.7318,
      "step": 1379
    },
    {
      "epoch": 0.17463933181473046,
      "grad_norm": 1.752781629562378,
      "learning_rate": 1.9738456010205365e-05,
      "loss": 0.6959,
      "step": 1380
    },
    {
      "epoch": 0.1747658820551759,
      "grad_norm": 1.4531840085983276,
      "learning_rate": 1.973804866399651e-05,
      "loss": 0.7306,
      "step": 1381
    },
    {
      "epoch": 0.17489243229562138,
      "grad_norm": 1.9876139163970947,
      "learning_rate": 1.9737641005030076e-05,
      "loss": 0.722,
      "step": 1382
    },
    {
      "epoch": 0.17501898253606682,
      "grad_norm": 1.5604699850082397,
      "learning_rate": 1.9737233033319144e-05,
      "loss": 0.7138,
      "step": 1383
    },
    {
      "epoch": 0.17514553277651226,
      "grad_norm": 1.8556336164474487,
      "learning_rate": 1.973682474887682e-05,
      "loss": 0.7181,
      "step": 1384
    },
    {
      "epoch": 0.17527208301695774,
      "grad_norm": 1.7296404838562012,
      "learning_rate": 1.9736416151716222e-05,
      "loss": 0.7628,
      "step": 1385
    },
    {
      "epoch": 0.17539863325740318,
      "grad_norm": 1.6389529705047607,
      "learning_rate": 1.973600724185047e-05,
      "loss": 0.6727,
      "step": 1386
    },
    {
      "epoch": 0.17552518349784865,
      "grad_norm": 1.2351360321044922,
      "learning_rate": 1.9735598019292692e-05,
      "loss": 0.7287,
      "step": 1387
    },
    {
      "epoch": 0.1756517337382941,
      "grad_norm": 1.445870041847229,
      "learning_rate": 1.973518848405604e-05,
      "loss": 0.7363,
      "step": 1388
    },
    {
      "epoch": 0.17577828397873957,
      "grad_norm": 1.1325939893722534,
      "learning_rate": 1.973477863615366e-05,
      "loss": 0.6092,
      "step": 1389
    },
    {
      "epoch": 0.175904834219185,
      "grad_norm": 1.7345926761627197,
      "learning_rate": 1.9734368475598718e-05,
      "loss": 0.737,
      "step": 1390
    },
    {
      "epoch": 0.17603138445963049,
      "grad_norm": 1.232146978378296,
      "learning_rate": 1.9733958002404387e-05,
      "loss": 0.6646,
      "step": 1391
    },
    {
      "epoch": 0.17615793470007593,
      "grad_norm": 1.332629919052124,
      "learning_rate": 1.973354721658385e-05,
      "loss": 0.6728,
      "step": 1392
    },
    {
      "epoch": 0.17628448494052137,
      "grad_norm": 1.3182220458984375,
      "learning_rate": 1.97331361181503e-05,
      "loss": 0.6702,
      "step": 1393
    },
    {
      "epoch": 0.17641103518096685,
      "grad_norm": 1.2221462726593018,
      "learning_rate": 1.9732724707116942e-05,
      "loss": 0.6856,
      "step": 1394
    },
    {
      "epoch": 0.1765375854214123,
      "grad_norm": 1.4908490180969238,
      "learning_rate": 1.9732312983496986e-05,
      "loss": 0.6942,
      "step": 1395
    },
    {
      "epoch": 0.17666413566185776,
      "grad_norm": 1.331132411956787,
      "learning_rate": 1.9731900947303657e-05,
      "loss": 0.7458,
      "step": 1396
    },
    {
      "epoch": 0.1767906859023032,
      "grad_norm": 1.8232083320617676,
      "learning_rate": 1.973148859855019e-05,
      "loss": 0.7408,
      "step": 1397
    },
    {
      "epoch": 0.17691723614274868,
      "grad_norm": 1.1791061162948608,
      "learning_rate": 1.9731075937249823e-05,
      "loss": 0.6631,
      "step": 1398
    },
    {
      "epoch": 0.17704378638319412,
      "grad_norm": 1.363901138305664,
      "learning_rate": 1.973066296341582e-05,
      "loss": 0.67,
      "step": 1399
    },
    {
      "epoch": 0.1771703366236396,
      "grad_norm": 1.3545746803283691,
      "learning_rate": 1.973024967706143e-05,
      "loss": 0.7182,
      "step": 1400
    },
    {
      "epoch": 0.17729688686408504,
      "grad_norm": 1.3294436931610107,
      "learning_rate": 1.9729836078199942e-05,
      "loss": 0.7375,
      "step": 1401
    },
    {
      "epoch": 0.17742343710453048,
      "grad_norm": 1.2490442991256714,
      "learning_rate": 1.9729422166844628e-05,
      "loss": 0.6788,
      "step": 1402
    },
    {
      "epoch": 0.17754998734497596,
      "grad_norm": 1.2975002527236938,
      "learning_rate": 1.9729007943008786e-05,
      "loss": 0.7526,
      "step": 1403
    },
    {
      "epoch": 0.1776765375854214,
      "grad_norm": 1.5997247695922852,
      "learning_rate": 1.972859340670572e-05,
      "loss": 0.7174,
      "step": 1404
    },
    {
      "epoch": 0.17780308782586687,
      "grad_norm": 1.3995827436447144,
      "learning_rate": 1.9728178557948742e-05,
      "loss": 0.6955,
      "step": 1405
    },
    {
      "epoch": 0.17792963806631232,
      "grad_norm": 1.278583288192749,
      "learning_rate": 1.972776339675118e-05,
      "loss": 0.7241,
      "step": 1406
    },
    {
      "epoch": 0.1780561883067578,
      "grad_norm": 1.3563110828399658,
      "learning_rate": 1.972734792312636e-05,
      "loss": 0.7377,
      "step": 1407
    },
    {
      "epoch": 0.17818273854720323,
      "grad_norm": 1.247473120689392,
      "learning_rate": 1.9726932137087634e-05,
      "loss": 0.7354,
      "step": 1408
    },
    {
      "epoch": 0.1783092887876487,
      "grad_norm": 1.3573999404907227,
      "learning_rate": 1.972651603864835e-05,
      "loss": 0.691,
      "step": 1409
    },
    {
      "epoch": 0.17843583902809415,
      "grad_norm": 1.334663987159729,
      "learning_rate": 1.972609962782188e-05,
      "loss": 0.6992,
      "step": 1410
    },
    {
      "epoch": 0.17856238926853962,
      "grad_norm": 1.3088182210922241,
      "learning_rate": 1.9725682904621583e-05,
      "loss": 0.6817,
      "step": 1411
    },
    {
      "epoch": 0.17868893950898507,
      "grad_norm": 1.3291181325912476,
      "learning_rate": 1.9725265869060856e-05,
      "loss": 0.7272,
      "step": 1412
    },
    {
      "epoch": 0.1788154897494305,
      "grad_norm": 1.294813871383667,
      "learning_rate": 1.9724848521153087e-05,
      "loss": 0.7005,
      "step": 1413
    },
    {
      "epoch": 0.17894203998987598,
      "grad_norm": 1.3090065717697144,
      "learning_rate": 1.9724430860911683e-05,
      "loss": 0.6994,
      "step": 1414
    },
    {
      "epoch": 0.17906859023032143,
      "grad_norm": 1.4385325908660889,
      "learning_rate": 1.9724012888350058e-05,
      "loss": 0.6804,
      "step": 1415
    },
    {
      "epoch": 0.1791951404707669,
      "grad_norm": 1.2816325426101685,
      "learning_rate": 1.9723594603481633e-05,
      "loss": 0.7201,
      "step": 1416
    },
    {
      "epoch": 0.17932169071121234,
      "grad_norm": 1.2426400184631348,
      "learning_rate": 1.9723176006319847e-05,
      "loss": 0.744,
      "step": 1417
    },
    {
      "epoch": 0.17944824095165782,
      "grad_norm": 1.4314335584640503,
      "learning_rate": 1.972275709687814e-05,
      "loss": 0.7001,
      "step": 1418
    },
    {
      "epoch": 0.17957479119210326,
      "grad_norm": 1.3344509601593018,
      "learning_rate": 1.9722337875169963e-05,
      "loss": 0.667,
      "step": 1419
    },
    {
      "epoch": 0.17970134143254873,
      "grad_norm": 1.3746684789657593,
      "learning_rate": 1.972191834120879e-05,
      "loss": 0.7121,
      "step": 1420
    },
    {
      "epoch": 0.17982789167299418,
      "grad_norm": 1.250834584236145,
      "learning_rate": 1.9721498495008086e-05,
      "loss": 0.6972,
      "step": 1421
    },
    {
      "epoch": 0.17995444191343962,
      "grad_norm": 1.215623378753662,
      "learning_rate": 1.972107833658134e-05,
      "loss": 0.712,
      "step": 1422
    },
    {
      "epoch": 0.1800809921538851,
      "grad_norm": 1.6032688617706299,
      "learning_rate": 1.9720657865942044e-05,
      "loss": 0.6823,
      "step": 1423
    },
    {
      "epoch": 0.18020754239433054,
      "grad_norm": 1.235183835029602,
      "learning_rate": 1.9720237083103703e-05,
      "loss": 0.7065,
      "step": 1424
    },
    {
      "epoch": 0.180334092634776,
      "grad_norm": 1.2651077508926392,
      "learning_rate": 1.9719815988079835e-05,
      "loss": 0.6608,
      "step": 1425
    },
    {
      "epoch": 0.18046064287522146,
      "grad_norm": 1.4532278776168823,
      "learning_rate": 1.9719394580883957e-05,
      "loss": 0.7187,
      "step": 1426
    },
    {
      "epoch": 0.18058719311566693,
      "grad_norm": 1.2685941457748413,
      "learning_rate": 1.971897286152961e-05,
      "loss": 0.699,
      "step": 1427
    },
    {
      "epoch": 0.18071374335611237,
      "grad_norm": 1.3798880577087402,
      "learning_rate": 1.9718550830030337e-05,
      "loss": 0.7337,
      "step": 1428
    },
    {
      "epoch": 0.18084029359655784,
      "grad_norm": 1.315252423286438,
      "learning_rate": 1.9718128486399685e-05,
      "loss": 0.6899,
      "step": 1429
    },
    {
      "epoch": 0.1809668438370033,
      "grad_norm": 1.4540470838546753,
      "learning_rate": 1.971770583065123e-05,
      "loss": 0.7165,
      "step": 1430
    },
    {
      "epoch": 0.18109339407744876,
      "grad_norm": 1.5037881135940552,
      "learning_rate": 1.971728286279854e-05,
      "loss": 0.6969,
      "step": 1431
    },
    {
      "epoch": 0.1812199443178942,
      "grad_norm": 1.495856761932373,
      "learning_rate": 1.97168595828552e-05,
      "loss": 0.6684,
      "step": 1432
    },
    {
      "epoch": 0.18134649455833965,
      "grad_norm": 1.4863567352294922,
      "learning_rate": 1.9716435990834807e-05,
      "loss": 0.7158,
      "step": 1433
    },
    {
      "epoch": 0.18147304479878512,
      "grad_norm": 1.2845147848129272,
      "learning_rate": 1.971601208675096e-05,
      "loss": 0.6817,
      "step": 1434
    },
    {
      "epoch": 0.18159959503923057,
      "grad_norm": 1.322800636291504,
      "learning_rate": 1.9715587870617285e-05,
      "loss": 0.6885,
      "step": 1435
    },
    {
      "epoch": 0.18172614527967604,
      "grad_norm": 1.1973204612731934,
      "learning_rate": 1.971516334244739e-05,
      "loss": 0.7202,
      "step": 1436
    },
    {
      "epoch": 0.18185269552012148,
      "grad_norm": 1.5601669549942017,
      "learning_rate": 1.971473850225492e-05,
      "loss": 0.7695,
      "step": 1437
    },
    {
      "epoch": 0.18197924576056695,
      "grad_norm": 1.3710458278656006,
      "learning_rate": 1.971431335005352e-05,
      "loss": 0.7419,
      "step": 1438
    },
    {
      "epoch": 0.1821057960010124,
      "grad_norm": 1.2860478162765503,
      "learning_rate": 1.971388788585684e-05,
      "loss": 0.6979,
      "step": 1439
    },
    {
      "epoch": 0.18223234624145787,
      "grad_norm": 1.877659559249878,
      "learning_rate": 1.9713462109678552e-05,
      "loss": 0.7344,
      "step": 1440
    },
    {
      "epoch": 0.18235889648190332,
      "grad_norm": 1.4165942668914795,
      "learning_rate": 1.9713036021532324e-05,
      "loss": 0.6659,
      "step": 1441
    },
    {
      "epoch": 0.18248544672234876,
      "grad_norm": 1.5124757289886475,
      "learning_rate": 1.971260962143184e-05,
      "loss": 0.7308,
      "step": 1442
    },
    {
      "epoch": 0.18261199696279423,
      "grad_norm": 1.334116816520691,
      "learning_rate": 1.97121829093908e-05,
      "loss": 0.7138,
      "step": 1443
    },
    {
      "epoch": 0.18273854720323968,
      "grad_norm": 1.4634581804275513,
      "learning_rate": 1.9711755885422906e-05,
      "loss": 0.7505,
      "step": 1444
    },
    {
      "epoch": 0.18286509744368515,
      "grad_norm": 1.2061057090759277,
      "learning_rate": 1.971132854954187e-05,
      "loss": 0.7341,
      "step": 1445
    },
    {
      "epoch": 0.1829916476841306,
      "grad_norm": 1.3759366273880005,
      "learning_rate": 1.9710900901761424e-05,
      "loss": 0.7026,
      "step": 1446
    },
    {
      "epoch": 0.18311819792457606,
      "grad_norm": 1.2317211627960205,
      "learning_rate": 1.9710472942095296e-05,
      "loss": 0.6649,
      "step": 1447
    },
    {
      "epoch": 0.1832447481650215,
      "grad_norm": 1.8511899709701538,
      "learning_rate": 1.9710044670557234e-05,
      "loss": 0.7345,
      "step": 1448
    },
    {
      "epoch": 0.18337129840546698,
      "grad_norm": 1.2633020877838135,
      "learning_rate": 1.9709616087160994e-05,
      "loss": 0.7716,
      "step": 1449
    },
    {
      "epoch": 0.18349784864591243,
      "grad_norm": 1.5829133987426758,
      "learning_rate": 1.9709187191920334e-05,
      "loss": 0.6813,
      "step": 1450
    },
    {
      "epoch": 0.18362439888635787,
      "grad_norm": 1.3500844240188599,
      "learning_rate": 1.9708757984849035e-05,
      "loss": 0.7086,
      "step": 1451
    },
    {
      "epoch": 0.18375094912680334,
      "grad_norm": 1.3518285751342773,
      "learning_rate": 1.9708328465960884e-05,
      "loss": 0.6907,
      "step": 1452
    },
    {
      "epoch": 0.1838774993672488,
      "grad_norm": 1.2986679077148438,
      "learning_rate": 1.970789863526967e-05,
      "loss": 0.7067,
      "step": 1453
    },
    {
      "epoch": 0.18400404960769426,
      "grad_norm": 1.1724722385406494,
      "learning_rate": 1.9707468492789203e-05,
      "loss": 0.7261,
      "step": 1454
    },
    {
      "epoch": 0.1841305998481397,
      "grad_norm": 1.2288825511932373,
      "learning_rate": 1.9707038038533294e-05,
      "loss": 0.7293,
      "step": 1455
    },
    {
      "epoch": 0.18425715008858518,
      "grad_norm": 1.2502776384353638,
      "learning_rate": 1.970660727251577e-05,
      "loss": 0.7232,
      "step": 1456
    },
    {
      "epoch": 0.18438370032903062,
      "grad_norm": 1.2854580879211426,
      "learning_rate": 1.970617619475046e-05,
      "loss": 0.7145,
      "step": 1457
    },
    {
      "epoch": 0.1845102505694761,
      "grad_norm": 1.3183971643447876,
      "learning_rate": 1.970574480525122e-05,
      "loss": 0.7006,
      "step": 1458
    },
    {
      "epoch": 0.18463680080992154,
      "grad_norm": 1.2312012910842896,
      "learning_rate": 1.9705313104031897e-05,
      "loss": 0.6997,
      "step": 1459
    },
    {
      "epoch": 0.184763351050367,
      "grad_norm": 1.2111928462982178,
      "learning_rate": 1.970488109110636e-05,
      "loss": 0.6465,
      "step": 1460
    },
    {
      "epoch": 0.18488990129081245,
      "grad_norm": 1.1704070568084717,
      "learning_rate": 1.970444876648848e-05,
      "loss": 0.6545,
      "step": 1461
    },
    {
      "epoch": 0.1850164515312579,
      "grad_norm": 1.4949567317962646,
      "learning_rate": 1.9704016130192146e-05,
      "loss": 0.7097,
      "step": 1462
    },
    {
      "epoch": 0.18514300177170337,
      "grad_norm": 1.2500032186508179,
      "learning_rate": 1.970358318223125e-05,
      "loss": 0.6947,
      "step": 1463
    },
    {
      "epoch": 0.1852695520121488,
      "grad_norm": 1.5865957736968994,
      "learning_rate": 1.9703149922619696e-05,
      "loss": 0.7512,
      "step": 1464
    },
    {
      "epoch": 0.18539610225259429,
      "grad_norm": 1.345358967781067,
      "learning_rate": 1.9702716351371402e-05,
      "loss": 0.7248,
      "step": 1465
    },
    {
      "epoch": 0.18552265249303973,
      "grad_norm": 1.3517688512802124,
      "learning_rate": 1.9702282468500296e-05,
      "loss": 0.6751,
      "step": 1466
    },
    {
      "epoch": 0.1856492027334852,
      "grad_norm": 1.3240675926208496,
      "learning_rate": 1.9701848274020308e-05,
      "loss": 0.7653,
      "step": 1467
    },
    {
      "epoch": 0.18577575297393065,
      "grad_norm": 1.2473992109298706,
      "learning_rate": 1.9701413767945382e-05,
      "loss": 0.6888,
      "step": 1468
    },
    {
      "epoch": 0.18590230321437612,
      "grad_norm": 1.2643026113510132,
      "learning_rate": 1.9700978950289475e-05,
      "loss": 0.731,
      "step": 1469
    },
    {
      "epoch": 0.18602885345482156,
      "grad_norm": 1.3135473728179932,
      "learning_rate": 1.9700543821066556e-05,
      "loss": 0.6964,
      "step": 1470
    },
    {
      "epoch": 0.186155403695267,
      "grad_norm": 1.3790580034255981,
      "learning_rate": 1.9700108380290593e-05,
      "loss": 0.7275,
      "step": 1471
    },
    {
      "epoch": 0.18628195393571248,
      "grad_norm": 1.3130995035171509,
      "learning_rate": 1.9699672627975577e-05,
      "loss": 0.6996,
      "step": 1472
    },
    {
      "epoch": 0.18640850417615792,
      "grad_norm": 1.2255648374557495,
      "learning_rate": 1.96992365641355e-05,
      "loss": 0.6924,
      "step": 1473
    },
    {
      "epoch": 0.1865350544166034,
      "grad_norm": 1.2650445699691772,
      "learning_rate": 1.9698800188784367e-05,
      "loss": 0.7114,
      "step": 1474
    },
    {
      "epoch": 0.18666160465704884,
      "grad_norm": 1.2190639972686768,
      "learning_rate": 1.9698363501936197e-05,
      "loss": 0.6871,
      "step": 1475
    },
    {
      "epoch": 0.1867881548974943,
      "grad_norm": 1.2413949966430664,
      "learning_rate": 1.969792650360501e-05,
      "loss": 0.7057,
      "step": 1476
    },
    {
      "epoch": 0.18691470513793976,
      "grad_norm": 1.248470664024353,
      "learning_rate": 1.969748919380484e-05,
      "loss": 0.6917,
      "step": 1477
    },
    {
      "epoch": 0.18704125537838523,
      "grad_norm": 1.3117204904556274,
      "learning_rate": 1.969705157254974e-05,
      "loss": 0.6983,
      "step": 1478
    },
    {
      "epoch": 0.18716780561883067,
      "grad_norm": 1.2054753303527832,
      "learning_rate": 1.9696613639853756e-05,
      "loss": 0.6963,
      "step": 1479
    },
    {
      "epoch": 0.18729435585927615,
      "grad_norm": 1.224846363067627,
      "learning_rate": 1.969617539573096e-05,
      "loss": 0.6607,
      "step": 1480
    },
    {
      "epoch": 0.1874209060997216,
      "grad_norm": 1.2858130931854248,
      "learning_rate": 1.9695736840195427e-05,
      "loss": 0.7277,
      "step": 1481
    },
    {
      "epoch": 0.18754745634016703,
      "grad_norm": 1.1927601099014282,
      "learning_rate": 1.9695297973261232e-05,
      "loss": 0.6825,
      "step": 1482
    },
    {
      "epoch": 0.1876740065806125,
      "grad_norm": 1.2945138216018677,
      "learning_rate": 1.9694858794942483e-05,
      "loss": 0.7096,
      "step": 1483
    },
    {
      "epoch": 0.18780055682105795,
      "grad_norm": 1.344256043434143,
      "learning_rate": 1.969441930525328e-05,
      "loss": 0.6971,
      "step": 1484
    },
    {
      "epoch": 0.18792710706150342,
      "grad_norm": 1.3018643856048584,
      "learning_rate": 1.969397950420774e-05,
      "loss": 0.717,
      "step": 1485
    },
    {
      "epoch": 0.18805365730194887,
      "grad_norm": 1.2183735370635986,
      "learning_rate": 1.9693539391819983e-05,
      "loss": 0.7192,
      "step": 1486
    },
    {
      "epoch": 0.18818020754239434,
      "grad_norm": 1.2363461256027222,
      "learning_rate": 1.9693098968104148e-05,
      "loss": 0.7127,
      "step": 1487
    },
    {
      "epoch": 0.18830675778283978,
      "grad_norm": 1.290531873703003,
      "learning_rate": 1.969265823307438e-05,
      "loss": 0.7268,
      "step": 1488
    },
    {
      "epoch": 0.18843330802328526,
      "grad_norm": 1.2175843715667725,
      "learning_rate": 1.9692217186744833e-05,
      "loss": 0.6856,
      "step": 1489
    },
    {
      "epoch": 0.1885598582637307,
      "grad_norm": 1.2366515398025513,
      "learning_rate": 1.969177582912967e-05,
      "loss": 0.6985,
      "step": 1490
    },
    {
      "epoch": 0.18868640850417615,
      "grad_norm": 1.2166697978973389,
      "learning_rate": 1.969133416024307e-05,
      "loss": 0.6722,
      "step": 1491
    },
    {
      "epoch": 0.18881295874462162,
      "grad_norm": 1.2616201639175415,
      "learning_rate": 1.969089218009922e-05,
      "loss": 0.7326,
      "step": 1492
    },
    {
      "epoch": 0.18893950898506706,
      "grad_norm": 1.1744694709777832,
      "learning_rate": 1.9690449888712312e-05,
      "loss": 0.6862,
      "step": 1493
    },
    {
      "epoch": 0.18906605922551253,
      "grad_norm": 1.3838828802108765,
      "learning_rate": 1.9690007286096547e-05,
      "loss": 0.7186,
      "step": 1494
    },
    {
      "epoch": 0.18919260946595798,
      "grad_norm": 1.1714913845062256,
      "learning_rate": 1.9689564372266146e-05,
      "loss": 0.6737,
      "step": 1495
    },
    {
      "epoch": 0.18931915970640345,
      "grad_norm": 1.246382236480713,
      "learning_rate": 1.9689121147235334e-05,
      "loss": 0.6787,
      "step": 1496
    },
    {
      "epoch": 0.1894457099468489,
      "grad_norm": 1.3053051233291626,
      "learning_rate": 1.968867761101834e-05,
      "loss": 0.7209,
      "step": 1497
    },
    {
      "epoch": 0.18957226018729437,
      "grad_norm": 1.1892896890640259,
      "learning_rate": 1.9688233763629417e-05,
      "loss": 0.6528,
      "step": 1498
    },
    {
      "epoch": 0.1896988104277398,
      "grad_norm": 1.3166921138763428,
      "learning_rate": 1.968778960508282e-05,
      "loss": 0.7343,
      "step": 1499
    },
    {
      "epoch": 0.18982536066818528,
      "grad_norm": 1.2456257343292236,
      "learning_rate": 1.9687345135392803e-05,
      "loss": 0.6839,
      "step": 1500
    },
    {
      "epoch": 0.18995191090863073,
      "grad_norm": 1.5761576890945435,
      "learning_rate": 1.9686900354573653e-05,
      "loss": 0.7735,
      "step": 1501
    },
    {
      "epoch": 0.19007846114907617,
      "grad_norm": 1.2125911712646484,
      "learning_rate": 1.968645526263965e-05,
      "loss": 0.6844,
      "step": 1502
    },
    {
      "epoch": 0.19020501138952164,
      "grad_norm": 1.3297758102416992,
      "learning_rate": 1.968600985960509e-05,
      "loss": 0.7046,
      "step": 1503
    },
    {
      "epoch": 0.1903315616299671,
      "grad_norm": 1.354677677154541,
      "learning_rate": 1.9685564145484277e-05,
      "loss": 0.7192,
      "step": 1504
    },
    {
      "epoch": 0.19045811187041256,
      "grad_norm": 1.2670120000839233,
      "learning_rate": 1.9685118120291528e-05,
      "loss": 0.6599,
      "step": 1505
    },
    {
      "epoch": 0.190584662110858,
      "grad_norm": 1.3982356786727905,
      "learning_rate": 1.9684671784041166e-05,
      "loss": 0.688,
      "step": 1506
    },
    {
      "epoch": 0.19071121235130348,
      "grad_norm": 1.273277759552002,
      "learning_rate": 1.9684225136747525e-05,
      "loss": 0.6756,
      "step": 1507
    },
    {
      "epoch": 0.19083776259174892,
      "grad_norm": 1.2288867235183716,
      "learning_rate": 1.9683778178424955e-05,
      "loss": 0.7014,
      "step": 1508
    },
    {
      "epoch": 0.1909643128321944,
      "grad_norm": 1.2327594757080078,
      "learning_rate": 1.9683330909087805e-05,
      "loss": 0.6423,
      "step": 1509
    },
    {
      "epoch": 0.19109086307263984,
      "grad_norm": 1.1865266561508179,
      "learning_rate": 1.9682883328750445e-05,
      "loss": 0.7221,
      "step": 1510
    },
    {
      "epoch": 0.19121741331308528,
      "grad_norm": 1.2746052742004395,
      "learning_rate": 1.9682435437427244e-05,
      "loss": 0.6933,
      "step": 1511
    },
    {
      "epoch": 0.19134396355353075,
      "grad_norm": 4.857626914978027,
      "learning_rate": 1.96819872351326e-05,
      "loss": 0.692,
      "step": 1512
    },
    {
      "epoch": 0.1914705137939762,
      "grad_norm": 1.194216251373291,
      "learning_rate": 1.9681538721880887e-05,
      "loss": 0.6514,
      "step": 1513
    },
    {
      "epoch": 0.19159706403442167,
      "grad_norm": 1.2347618341445923,
      "learning_rate": 1.968108989768653e-05,
      "loss": 0.7245,
      "step": 1514
    },
    {
      "epoch": 0.19172361427486712,
      "grad_norm": 1.434093713760376,
      "learning_rate": 1.968064076256393e-05,
      "loss": 0.6845,
      "step": 1515
    },
    {
      "epoch": 0.1918501645153126,
      "grad_norm": 1.4627560377120972,
      "learning_rate": 1.9680191316527522e-05,
      "loss": 0.7287,
      "step": 1516
    },
    {
      "epoch": 0.19197671475575803,
      "grad_norm": 1.3027520179748535,
      "learning_rate": 1.9679741559591733e-05,
      "loss": 0.6821,
      "step": 1517
    },
    {
      "epoch": 0.1921032649962035,
      "grad_norm": 1.6070083379745483,
      "learning_rate": 1.9679291491771014e-05,
      "loss": 0.6969,
      "step": 1518
    },
    {
      "epoch": 0.19222981523664895,
      "grad_norm": 2.15567684173584,
      "learning_rate": 1.9678841113079818e-05,
      "loss": 0.6619,
      "step": 1519
    },
    {
      "epoch": 0.1923563654770944,
      "grad_norm": 1.3539752960205078,
      "learning_rate": 1.96783904235326e-05,
      "loss": 0.6698,
      "step": 1520
    },
    {
      "epoch": 0.19248291571753987,
      "grad_norm": 1.3915995359420776,
      "learning_rate": 1.9677939423143854e-05,
      "loss": 0.6953,
      "step": 1521
    },
    {
      "epoch": 0.1926094659579853,
      "grad_norm": 1.224461555480957,
      "learning_rate": 1.9677488111928055e-05,
      "loss": 0.6819,
      "step": 1522
    },
    {
      "epoch": 0.19273601619843078,
      "grad_norm": 1.2621554136276245,
      "learning_rate": 1.967703648989969e-05,
      "loss": 0.7071,
      "step": 1523
    },
    {
      "epoch": 0.19286256643887623,
      "grad_norm": 1.2402311563491821,
      "learning_rate": 1.9676584557073275e-05,
      "loss": 0.6707,
      "step": 1524
    },
    {
      "epoch": 0.1929891166793217,
      "grad_norm": 1.2271008491516113,
      "learning_rate": 1.967613231346332e-05,
      "loss": 0.6658,
      "step": 1525
    },
    {
      "epoch": 0.19311566691976714,
      "grad_norm": 1.5869712829589844,
      "learning_rate": 1.967567975908435e-05,
      "loss": 0.7363,
      "step": 1526
    },
    {
      "epoch": 0.19324221716021261,
      "grad_norm": 1.2843143939971924,
      "learning_rate": 1.9675226893950904e-05,
      "loss": 0.7001,
      "step": 1527
    },
    {
      "epoch": 0.19336876740065806,
      "grad_norm": 1.260161280632019,
      "learning_rate": 1.967477371807752e-05,
      "loss": 0.6985,
      "step": 1528
    },
    {
      "epoch": 0.19349531764110353,
      "grad_norm": 1.291739583015442,
      "learning_rate": 1.9674320231478753e-05,
      "loss": 0.7359,
      "step": 1529
    },
    {
      "epoch": 0.19362186788154898,
      "grad_norm": 1.18709135055542,
      "learning_rate": 1.9673866434169175e-05,
      "loss": 0.6716,
      "step": 1530
    },
    {
      "epoch": 0.19374841812199442,
      "grad_norm": 1.2636618614196777,
      "learning_rate": 1.9673412326163355e-05,
      "loss": 0.7182,
      "step": 1531
    },
    {
      "epoch": 0.1938749683624399,
      "grad_norm": 1.1825413703918457,
      "learning_rate": 1.967295790747588e-05,
      "loss": 0.6856,
      "step": 1532
    },
    {
      "epoch": 0.19400151860288534,
      "grad_norm": 1.1985816955566406,
      "learning_rate": 1.9672503178121342e-05,
      "loss": 0.7024,
      "step": 1533
    },
    {
      "epoch": 0.1941280688433308,
      "grad_norm": 1.3487489223480225,
      "learning_rate": 1.9672048138114345e-05,
      "loss": 0.7185,
      "step": 1534
    },
    {
      "epoch": 0.19425461908377625,
      "grad_norm": 1.5425299406051636,
      "learning_rate": 1.9671592787469504e-05,
      "loss": 0.6848,
      "step": 1535
    },
    {
      "epoch": 0.19438116932422173,
      "grad_norm": 1.2308995723724365,
      "learning_rate": 1.9671137126201448e-05,
      "loss": 0.6886,
      "step": 1536
    },
    {
      "epoch": 0.19450771956466717,
      "grad_norm": 1.5431321859359741,
      "learning_rate": 1.9670681154324808e-05,
      "loss": 0.7312,
      "step": 1537
    },
    {
      "epoch": 0.19463426980511264,
      "grad_norm": 1.3237091302871704,
      "learning_rate": 1.967022487185423e-05,
      "loss": 0.6946,
      "step": 1538
    },
    {
      "epoch": 0.1947608200455581,
      "grad_norm": 1.3964502811431885,
      "learning_rate": 1.9669768278804366e-05,
      "loss": 0.7765,
      "step": 1539
    },
    {
      "epoch": 0.19488737028600353,
      "grad_norm": 1.2780307531356812,
      "learning_rate": 1.966931137518988e-05,
      "loss": 0.6939,
      "step": 1540
    },
    {
      "epoch": 0.195013920526449,
      "grad_norm": 1.3457484245300293,
      "learning_rate": 1.966885416102545e-05,
      "loss": 0.6817,
      "step": 1541
    },
    {
      "epoch": 0.19514047076689445,
      "grad_norm": 1.285263180732727,
      "learning_rate": 1.966839663632576e-05,
      "loss": 0.7195,
      "step": 1542
    },
    {
      "epoch": 0.19526702100733992,
      "grad_norm": 1.2988955974578857,
      "learning_rate": 1.96679388011055e-05,
      "loss": 0.7195,
      "step": 1543
    },
    {
      "epoch": 0.19539357124778536,
      "grad_norm": 1.225723147392273,
      "learning_rate": 1.966748065537938e-05,
      "loss": 0.7133,
      "step": 1544
    },
    {
      "epoch": 0.19552012148823084,
      "grad_norm": 1.4564158916473389,
      "learning_rate": 1.9667022199162108e-05,
      "loss": 0.7388,
      "step": 1545
    },
    {
      "epoch": 0.19564667172867628,
      "grad_norm": 1.3514500856399536,
      "learning_rate": 1.9666563432468414e-05,
      "loss": 0.723,
      "step": 1546
    },
    {
      "epoch": 0.19577322196912175,
      "grad_norm": 1.3112412691116333,
      "learning_rate": 1.966610435531303e-05,
      "loss": 0.7179,
      "step": 1547
    },
    {
      "epoch": 0.1958997722095672,
      "grad_norm": 1.6777987480163574,
      "learning_rate": 1.96656449677107e-05,
      "loss": 0.6942,
      "step": 1548
    },
    {
      "epoch": 0.19602632245001267,
      "grad_norm": 1.1401196718215942,
      "learning_rate": 1.966518526967618e-05,
      "loss": 0.6545,
      "step": 1549
    },
    {
      "epoch": 0.1961528726904581,
      "grad_norm": 1.27152681350708,
      "learning_rate": 1.9664725261224232e-05,
      "loss": 0.7578,
      "step": 1550
    },
    {
      "epoch": 0.19627942293090356,
      "grad_norm": 1.2120342254638672,
      "learning_rate": 1.9664264942369628e-05,
      "loss": 0.712,
      "step": 1551
    },
    {
      "epoch": 0.19640597317134903,
      "grad_norm": 1.2955023050308228,
      "learning_rate": 1.966380431312716e-05,
      "loss": 0.7262,
      "step": 1552
    },
    {
      "epoch": 0.19653252341179447,
      "grad_norm": 1.2949409484863281,
      "learning_rate": 1.9663343373511612e-05,
      "loss": 0.6926,
      "step": 1553
    },
    {
      "epoch": 0.19665907365223995,
      "grad_norm": 1.2359402179718018,
      "learning_rate": 1.9662882123537795e-05,
      "loss": 0.7265,
      "step": 1554
    },
    {
      "epoch": 0.1967856238926854,
      "grad_norm": 1.219754695892334,
      "learning_rate": 1.966242056322052e-05,
      "loss": 0.6982,
      "step": 1555
    },
    {
      "epoch": 0.19691217413313086,
      "grad_norm": 1.252565860748291,
      "learning_rate": 1.9661958692574613e-05,
      "loss": 0.693,
      "step": 1556
    },
    {
      "epoch": 0.1970387243735763,
      "grad_norm": 1.2708735466003418,
      "learning_rate": 1.9661496511614904e-05,
      "loss": 0.6906,
      "step": 1557
    },
    {
      "epoch": 0.19716527461402178,
      "grad_norm": 1.1530851125717163,
      "learning_rate": 1.966103402035624e-05,
      "loss": 0.6732,
      "step": 1558
    },
    {
      "epoch": 0.19729182485446722,
      "grad_norm": 1.1931413412094116,
      "learning_rate": 1.966057121881348e-05,
      "loss": 0.6695,
      "step": 1559
    },
    {
      "epoch": 0.19741837509491267,
      "grad_norm": 1.3538968563079834,
      "learning_rate": 1.9660108107001478e-05,
      "loss": 0.7264,
      "step": 1560
    },
    {
      "epoch": 0.19754492533535814,
      "grad_norm": 1.3528797626495361,
      "learning_rate": 1.965964468493511e-05,
      "loss": 0.7336,
      "step": 1561
    },
    {
      "epoch": 0.19767147557580358,
      "grad_norm": 1.2320576906204224,
      "learning_rate": 1.965918095262927e-05,
      "loss": 0.7168,
      "step": 1562
    },
    {
      "epoch": 0.19779802581624906,
      "grad_norm": 1.4164888858795166,
      "learning_rate": 1.9658716910098835e-05,
      "loss": 0.7348,
      "step": 1563
    },
    {
      "epoch": 0.1979245760566945,
      "grad_norm": 1.3960223197937012,
      "learning_rate": 1.965825255735872e-05,
      "loss": 0.6857,
      "step": 1564
    },
    {
      "epoch": 0.19805112629713997,
      "grad_norm": 1.3486295938491821,
      "learning_rate": 1.965778789442384e-05,
      "loss": 0.6864,
      "step": 1565
    },
    {
      "epoch": 0.19817767653758542,
      "grad_norm": 1.3493682146072388,
      "learning_rate": 1.965732292130911e-05,
      "loss": 0.7222,
      "step": 1566
    },
    {
      "epoch": 0.1983042267780309,
      "grad_norm": 1.8527848720550537,
      "learning_rate": 1.965685763802947e-05,
      "loss": 0.6759,
      "step": 1567
    },
    {
      "epoch": 0.19843077701847633,
      "grad_norm": 1.414131760597229,
      "learning_rate": 1.9656392044599858e-05,
      "loss": 0.7244,
      "step": 1568
    },
    {
      "epoch": 0.19855732725892178,
      "grad_norm": 1.3655788898468018,
      "learning_rate": 1.9655926141035238e-05,
      "loss": 0.7159,
      "step": 1569
    },
    {
      "epoch": 0.19868387749936725,
      "grad_norm": 1.187276840209961,
      "learning_rate": 1.9655459927350562e-05,
      "loss": 0.6628,
      "step": 1570
    },
    {
      "epoch": 0.1988104277398127,
      "grad_norm": 1.3477691411972046,
      "learning_rate": 1.965499340356081e-05,
      "loss": 0.711,
      "step": 1571
    },
    {
      "epoch": 0.19893697798025817,
      "grad_norm": 1.3619394302368164,
      "learning_rate": 1.965452656968096e-05,
      "loss": 0.7316,
      "step": 1572
    },
    {
      "epoch": 0.1990635282207036,
      "grad_norm": 1.2609972953796387,
      "learning_rate": 1.9654059425726013e-05,
      "loss": 0.7347,
      "step": 1573
    },
    {
      "epoch": 0.19919007846114908,
      "grad_norm": 1.2593607902526855,
      "learning_rate": 1.965359197171097e-05,
      "loss": 0.7278,
      "step": 1574
    },
    {
      "epoch": 0.19931662870159453,
      "grad_norm": 1.2738368511199951,
      "learning_rate": 1.9653124207650834e-05,
      "loss": 0.735,
      "step": 1575
    },
    {
      "epoch": 0.19944317894204,
      "grad_norm": 1.1572697162628174,
      "learning_rate": 1.9652656133560645e-05,
      "loss": 0.6561,
      "step": 1576
    },
    {
      "epoch": 0.19956972918248544,
      "grad_norm": 1.170464038848877,
      "learning_rate": 1.965218774945542e-05,
      "loss": 0.6572,
      "step": 1577
    },
    {
      "epoch": 0.19969627942293092,
      "grad_norm": 1.124376654624939,
      "learning_rate": 1.9651719055350217e-05,
      "loss": 0.6596,
      "step": 1578
    },
    {
      "epoch": 0.19982282966337636,
      "grad_norm": 1.2687394618988037,
      "learning_rate": 1.965125005126008e-05,
      "loss": 0.7159,
      "step": 1579
    },
    {
      "epoch": 0.1999493799038218,
      "grad_norm": 1.334679126739502,
      "learning_rate": 1.9650780737200072e-05,
      "loss": 0.7022,
      "step": 1580
    },
    {
      "epoch": 0.20007593014426728,
      "grad_norm": 1.199448823928833,
      "learning_rate": 1.9650311113185266e-05,
      "loss": 0.6935,
      "step": 1581
    },
    {
      "epoch": 0.20020248038471272,
      "grad_norm": 1.4542646408081055,
      "learning_rate": 1.9649841179230754e-05,
      "loss": 0.7056,
      "step": 1582
    },
    {
      "epoch": 0.2003290306251582,
      "grad_norm": 1.2624573707580566,
      "learning_rate": 1.9649370935351616e-05,
      "loss": 0.7049,
      "step": 1583
    },
    {
      "epoch": 0.20045558086560364,
      "grad_norm": 1.5164581537246704,
      "learning_rate": 1.964890038156296e-05,
      "loss": 0.7266,
      "step": 1584
    },
    {
      "epoch": 0.2005821311060491,
      "grad_norm": 2.1790850162506104,
      "learning_rate": 1.9648429517879904e-05,
      "loss": 0.8057,
      "step": 1585
    },
    {
      "epoch": 0.20070868134649456,
      "grad_norm": 1.2718777656555176,
      "learning_rate": 1.9647958344317562e-05,
      "loss": 0.6656,
      "step": 1586
    },
    {
      "epoch": 0.20083523158694003,
      "grad_norm": 1.4100289344787598,
      "learning_rate": 1.9647486860891077e-05,
      "loss": 0.6978,
      "step": 1587
    },
    {
      "epoch": 0.20096178182738547,
      "grad_norm": 2.5536015033721924,
      "learning_rate": 1.9647015067615582e-05,
      "loss": 0.7058,
      "step": 1588
    },
    {
      "epoch": 0.20108833206783092,
      "grad_norm": 1.613206386566162,
      "learning_rate": 1.964654296450623e-05,
      "loss": 0.7234,
      "step": 1589
    },
    {
      "epoch": 0.2012148823082764,
      "grad_norm": 1.6448583602905273,
      "learning_rate": 1.964607055157819e-05,
      "loss": 0.6887,
      "step": 1590
    },
    {
      "epoch": 0.20134143254872183,
      "grad_norm": 1.5863806009292603,
      "learning_rate": 1.964559782884663e-05,
      "loss": 0.6994,
      "step": 1591
    },
    {
      "epoch": 0.2014679827891673,
      "grad_norm": 1.2958152294158936,
      "learning_rate": 1.9645124796326734e-05,
      "loss": 0.7096,
      "step": 1592
    },
    {
      "epoch": 0.20159453302961275,
      "grad_norm": 1.2837938070297241,
      "learning_rate": 1.9644651454033696e-05,
      "loss": 0.7286,
      "step": 1593
    },
    {
      "epoch": 0.20172108327005822,
      "grad_norm": 1.3233197927474976,
      "learning_rate": 1.9644177801982715e-05,
      "loss": 0.7448,
      "step": 1594
    },
    {
      "epoch": 0.20184763351050367,
      "grad_norm": 1.2355883121490479,
      "learning_rate": 1.9643703840189007e-05,
      "loss": 0.6496,
      "step": 1595
    },
    {
      "epoch": 0.20197418375094914,
      "grad_norm": 1.2781611680984497,
      "learning_rate": 1.964322956866779e-05,
      "loss": 0.6905,
      "step": 1596
    },
    {
      "epoch": 0.20210073399139458,
      "grad_norm": 1.4397051334381104,
      "learning_rate": 1.9642754987434298e-05,
      "loss": 0.71,
      "step": 1597
    },
    {
      "epoch": 0.20222728423184005,
      "grad_norm": 1.5227223634719849,
      "learning_rate": 1.9642280096503773e-05,
      "loss": 0.6824,
      "step": 1598
    },
    {
      "epoch": 0.2023538344722855,
      "grad_norm": 1.7234582901000977,
      "learning_rate": 1.964180489589147e-05,
      "loss": 0.7076,
      "step": 1599
    },
    {
      "epoch": 0.20248038471273094,
      "grad_norm": 1.400059461593628,
      "learning_rate": 1.9641329385612648e-05,
      "loss": 0.6958,
      "step": 1600
    },
    {
      "epoch": 0.20260693495317642,
      "grad_norm": 1.4760463237762451,
      "learning_rate": 1.964085356568258e-05,
      "loss": 0.6864,
      "step": 1601
    },
    {
      "epoch": 0.20273348519362186,
      "grad_norm": 1.3604785203933716,
      "learning_rate": 1.9640377436116546e-05,
      "loss": 0.7564,
      "step": 1602
    },
    {
      "epoch": 0.20286003543406733,
      "grad_norm": 1.4099897146224976,
      "learning_rate": 1.9639900996929842e-05,
      "loss": 0.7341,
      "step": 1603
    },
    {
      "epoch": 0.20298658567451278,
      "grad_norm": 1.2749874591827393,
      "learning_rate": 1.9639424248137766e-05,
      "loss": 0.7664,
      "step": 1604
    },
    {
      "epoch": 0.20311313591495825,
      "grad_norm": 1.247374176979065,
      "learning_rate": 1.963894718975563e-05,
      "loss": 0.6438,
      "step": 1605
    },
    {
      "epoch": 0.2032396861554037,
      "grad_norm": 1.1076833009719849,
      "learning_rate": 1.963846982179876e-05,
      "loss": 0.6609,
      "step": 1606
    },
    {
      "epoch": 0.20336623639584916,
      "grad_norm": 1.3145843744277954,
      "learning_rate": 1.963799214428248e-05,
      "loss": 0.6981,
      "step": 1607
    },
    {
      "epoch": 0.2034927866362946,
      "grad_norm": 1.2145676612854004,
      "learning_rate": 1.963751415722214e-05,
      "loss": 0.7235,
      "step": 1608
    },
    {
      "epoch": 0.20361933687674005,
      "grad_norm": 1.3255066871643066,
      "learning_rate": 1.9637035860633085e-05,
      "loss": 0.7006,
      "step": 1609
    },
    {
      "epoch": 0.20374588711718553,
      "grad_norm": 1.5862802267074585,
      "learning_rate": 1.963655725453068e-05,
      "loss": 0.7147,
      "step": 1610
    },
    {
      "epoch": 0.20387243735763097,
      "grad_norm": 1.194977045059204,
      "learning_rate": 1.9636078338930296e-05,
      "loss": 0.7307,
      "step": 1611
    },
    {
      "epoch": 0.20399898759807644,
      "grad_norm": 1.549065113067627,
      "learning_rate": 1.9635599113847313e-05,
      "loss": 0.7038,
      "step": 1612
    },
    {
      "epoch": 0.2041255378385219,
      "grad_norm": 1.2363486289978027,
      "learning_rate": 1.963511957929712e-05,
      "loss": 0.7113,
      "step": 1613
    },
    {
      "epoch": 0.20425208807896736,
      "grad_norm": 1.5444289445877075,
      "learning_rate": 1.9634639735295126e-05,
      "loss": 0.6763,
      "step": 1614
    },
    {
      "epoch": 0.2043786383194128,
      "grad_norm": 1.2053565979003906,
      "learning_rate": 1.963415958185674e-05,
      "loss": 0.6806,
      "step": 1615
    },
    {
      "epoch": 0.20450518855985828,
      "grad_norm": 6.753561973571777,
      "learning_rate": 1.9633679118997376e-05,
      "loss": 0.6939,
      "step": 1616
    },
    {
      "epoch": 0.20463173880030372,
      "grad_norm": 2.491490602493286,
      "learning_rate": 1.963319834673247e-05,
      "loss": 0.6785,
      "step": 1617
    },
    {
      "epoch": 0.20475828904074916,
      "grad_norm": 1.2844572067260742,
      "learning_rate": 1.9632717265077466e-05,
      "loss": 0.7098,
      "step": 1618
    },
    {
      "epoch": 0.20488483928119464,
      "grad_norm": 2.059366464614868,
      "learning_rate": 1.963223587404781e-05,
      "loss": 0.6514,
      "step": 1619
    },
    {
      "epoch": 0.20501138952164008,
      "grad_norm": 1.5953902006149292,
      "learning_rate": 1.9631754173658964e-05,
      "loss": 0.6974,
      "step": 1620
    },
    {
      "epoch": 0.20513793976208555,
      "grad_norm": 2.0511622428894043,
      "learning_rate": 1.96312721639264e-05,
      "loss": 0.7139,
      "step": 1621
    },
    {
      "epoch": 0.205264490002531,
      "grad_norm": 2.1361067295074463,
      "learning_rate": 1.96307898448656e-05,
      "loss": 0.723,
      "step": 1622
    },
    {
      "epoch": 0.20539104024297647,
      "grad_norm": 1.400399923324585,
      "learning_rate": 1.963030721649205e-05,
      "loss": 0.6817,
      "step": 1623
    },
    {
      "epoch": 0.2055175904834219,
      "grad_norm": 1.9142087697982788,
      "learning_rate": 1.9629824278821256e-05,
      "loss": 0.7082,
      "step": 1624
    },
    {
      "epoch": 0.20564414072386739,
      "grad_norm": 1.1915558576583862,
      "learning_rate": 1.9629341031868724e-05,
      "loss": 0.6796,
      "step": 1625
    },
    {
      "epoch": 0.20577069096431283,
      "grad_norm": 1.610068440437317,
      "learning_rate": 1.9628857475649976e-05,
      "loss": 0.6715,
      "step": 1626
    },
    {
      "epoch": 0.2058972412047583,
      "grad_norm": 1.2053229808807373,
      "learning_rate": 1.9628373610180542e-05,
      "loss": 0.6872,
      "step": 1627
    },
    {
      "epoch": 0.20602379144520375,
      "grad_norm": 1.4604718685150146,
      "learning_rate": 1.9627889435475968e-05,
      "loss": 0.6937,
      "step": 1628
    },
    {
      "epoch": 0.2061503416856492,
      "grad_norm": 1.1615575551986694,
      "learning_rate": 1.9627404951551794e-05,
      "loss": 0.6896,
      "step": 1629
    },
    {
      "epoch": 0.20627689192609466,
      "grad_norm": 2.7280936241149902,
      "learning_rate": 1.962692015842359e-05,
      "loss": 0.7772,
      "step": 1630
    },
    {
      "epoch": 0.2064034421665401,
      "grad_norm": 1.3559974431991577,
      "learning_rate": 1.962643505610692e-05,
      "loss": 0.6844,
      "step": 1631
    },
    {
      "epoch": 0.20652999240698558,
      "grad_norm": 1.4279286861419678,
      "learning_rate": 1.9625949644617367e-05,
      "loss": 0.6748,
      "step": 1632
    },
    {
      "epoch": 0.20665654264743102,
      "grad_norm": 1.2395954132080078,
      "learning_rate": 1.9625463923970514e-05,
      "loss": 0.7156,
      "step": 1633
    },
    {
      "epoch": 0.2067830928878765,
      "grad_norm": 1.6281431913375854,
      "learning_rate": 1.9624977894181972e-05,
      "loss": 0.6773,
      "step": 1634
    },
    {
      "epoch": 0.20690964312832194,
      "grad_norm": 1.294625997543335,
      "learning_rate": 1.9624491555267346e-05,
      "loss": 0.6718,
      "step": 1635
    },
    {
      "epoch": 0.2070361933687674,
      "grad_norm": 1.7690510749816895,
      "learning_rate": 1.9624004907242256e-05,
      "loss": 0.7048,
      "step": 1636
    },
    {
      "epoch": 0.20716274360921286,
      "grad_norm": 1.415603756904602,
      "learning_rate": 1.962351795012233e-05,
      "loss": 0.7292,
      "step": 1637
    },
    {
      "epoch": 0.2072892938496583,
      "grad_norm": 1.5629616975784302,
      "learning_rate": 1.9623030683923208e-05,
      "loss": 0.625,
      "step": 1638
    },
    {
      "epoch": 0.20741584409010377,
      "grad_norm": 1.5758380889892578,
      "learning_rate": 1.9622543108660542e-05,
      "loss": 0.6679,
      "step": 1639
    },
    {
      "epoch": 0.20754239433054922,
      "grad_norm": 1.6248940229415894,
      "learning_rate": 1.9622055224349988e-05,
      "loss": 0.7536,
      "step": 1640
    },
    {
      "epoch": 0.2076689445709947,
      "grad_norm": 1.8008595705032349,
      "learning_rate": 1.9621567031007218e-05,
      "loss": 0.7167,
      "step": 1641
    },
    {
      "epoch": 0.20779549481144013,
      "grad_norm": 1.1835217475891113,
      "learning_rate": 1.962107852864791e-05,
      "loss": 0.6886,
      "step": 1642
    },
    {
      "epoch": 0.2079220450518856,
      "grad_norm": 2.161879539489746,
      "learning_rate": 1.9620589717287757e-05,
      "loss": 0.7487,
      "step": 1643
    },
    {
      "epoch": 0.20804859529233105,
      "grad_norm": 1.6750376224517822,
      "learning_rate": 1.962010059694245e-05,
      "loss": 0.6757,
      "step": 1644
    },
    {
      "epoch": 0.20817514553277652,
      "grad_norm": 1.6980773210525513,
      "learning_rate": 1.961961116762771e-05,
      "loss": 0.7332,
      "step": 1645
    },
    {
      "epoch": 0.20830169577322197,
      "grad_norm": 1.7329051494598389,
      "learning_rate": 1.961912142935924e-05,
      "loss": 0.6808,
      "step": 1646
    },
    {
      "epoch": 0.20842824601366744,
      "grad_norm": 1.2277202606201172,
      "learning_rate": 1.9618631382152785e-05,
      "loss": 0.6775,
      "step": 1647
    },
    {
      "epoch": 0.20855479625411288,
      "grad_norm": 1.530221700668335,
      "learning_rate": 1.9618141026024074e-05,
      "loss": 0.7106,
      "step": 1648
    },
    {
      "epoch": 0.20868134649455833,
      "grad_norm": 1.216953158378601,
      "learning_rate": 1.961765036098886e-05,
      "loss": 0.717,
      "step": 1649
    },
    {
      "epoch": 0.2088078967350038,
      "grad_norm": 1.3490759134292603,
      "learning_rate": 1.96171593870629e-05,
      "loss": 0.7257,
      "step": 1650
    },
    {
      "epoch": 0.20893444697544925,
      "grad_norm": 1.2838283777236938,
      "learning_rate": 1.9616668104261964e-05,
      "loss": 0.6957,
      "step": 1651
    },
    {
      "epoch": 0.20906099721589472,
      "grad_norm": 1.2271021604537964,
      "learning_rate": 1.961617651260183e-05,
      "loss": 0.6645,
      "step": 1652
    },
    {
      "epoch": 0.20918754745634016,
      "grad_norm": 1.2553613185882568,
      "learning_rate": 1.9615684612098285e-05,
      "loss": 0.6793,
      "step": 1653
    },
    {
      "epoch": 0.20931409769678563,
      "grad_norm": 1.3883463144302368,
      "learning_rate": 1.961519240276713e-05,
      "loss": 0.701,
      "step": 1654
    },
    {
      "epoch": 0.20944064793723108,
      "grad_norm": 1.5031473636627197,
      "learning_rate": 1.9614699884624172e-05,
      "loss": 0.7153,
      "step": 1655
    },
    {
      "epoch": 0.20956719817767655,
      "grad_norm": 1.2364914417266846,
      "learning_rate": 1.961420705768523e-05,
      "loss": 0.6797,
      "step": 1656
    },
    {
      "epoch": 0.209693748418122,
      "grad_norm": 1.501976490020752,
      "learning_rate": 1.9613713921966127e-05,
      "loss": 0.7188,
      "step": 1657
    },
    {
      "epoch": 0.20982029865856744,
      "grad_norm": 1.1195874214172363,
      "learning_rate": 1.961322047748271e-05,
      "loss": 0.6598,
      "step": 1658
    },
    {
      "epoch": 0.2099468488990129,
      "grad_norm": 1.445251226425171,
      "learning_rate": 1.9612726724250823e-05,
      "loss": 0.7422,
      "step": 1659
    },
    {
      "epoch": 0.21007339913945836,
      "grad_norm": 1.1241436004638672,
      "learning_rate": 1.9612232662286316e-05,
      "loss": 0.6417,
      "step": 1660
    },
    {
      "epoch": 0.21019994937990383,
      "grad_norm": 1.43875253200531,
      "learning_rate": 1.9611738291605073e-05,
      "loss": 0.7773,
      "step": 1661
    },
    {
      "epoch": 0.21032649962034927,
      "grad_norm": 1.3248752355575562,
      "learning_rate": 1.9611243612222957e-05,
      "loss": 0.6869,
      "step": 1662
    },
    {
      "epoch": 0.21045304986079474,
      "grad_norm": 1.2964569330215454,
      "learning_rate": 1.9610748624155865e-05,
      "loss": 0.6714,
      "step": 1663
    },
    {
      "epoch": 0.2105796001012402,
      "grad_norm": 1.2111291885375977,
      "learning_rate": 1.9610253327419687e-05,
      "loss": 0.6874,
      "step": 1664
    },
    {
      "epoch": 0.21070615034168566,
      "grad_norm": 1.191536784172058,
      "learning_rate": 1.960975772203034e-05,
      "loss": 0.6788,
      "step": 1665
    },
    {
      "epoch": 0.2108327005821311,
      "grad_norm": 1.2847646474838257,
      "learning_rate": 1.960926180800373e-05,
      "loss": 0.697,
      "step": 1666
    },
    {
      "epoch": 0.21095925082257655,
      "grad_norm": 1.3109086751937866,
      "learning_rate": 1.9608765585355795e-05,
      "loss": 0.6958,
      "step": 1667
    },
    {
      "epoch": 0.21108580106302202,
      "grad_norm": 1.2270761728286743,
      "learning_rate": 1.9608269054102466e-05,
      "loss": 0.6811,
      "step": 1668
    },
    {
      "epoch": 0.21121235130346747,
      "grad_norm": 1.5621494054794312,
      "learning_rate": 1.9607772214259693e-05,
      "loss": 0.717,
      "step": 1669
    },
    {
      "epoch": 0.21133890154391294,
      "grad_norm": 1.8028022050857544,
      "learning_rate": 1.960727506584343e-05,
      "loss": 0.6896,
      "step": 1670
    },
    {
      "epoch": 0.21146545178435838,
      "grad_norm": 1.4521104097366333,
      "learning_rate": 1.9606777608869648e-05,
      "loss": 0.7157,
      "step": 1671
    },
    {
      "epoch": 0.21159200202480385,
      "grad_norm": 1.2811082601547241,
      "learning_rate": 1.960627984335432e-05,
      "loss": 0.75,
      "step": 1672
    },
    {
      "epoch": 0.2117185522652493,
      "grad_norm": 1.3587536811828613,
      "learning_rate": 1.960578176931343e-05,
      "loss": 0.6946,
      "step": 1673
    },
    {
      "epoch": 0.21184510250569477,
      "grad_norm": 1.2241768836975098,
      "learning_rate": 1.9605283386762986e-05,
      "loss": 0.6981,
      "step": 1674
    },
    {
      "epoch": 0.21197165274614022,
      "grad_norm": 1.2860398292541504,
      "learning_rate": 1.9604784695718984e-05,
      "loss": 0.6844,
      "step": 1675
    },
    {
      "epoch": 0.2120982029865857,
      "grad_norm": 1.2260700464248657,
      "learning_rate": 1.9604285696197444e-05,
      "loss": 0.6818,
      "step": 1676
    },
    {
      "epoch": 0.21222475322703113,
      "grad_norm": 1.3050811290740967,
      "learning_rate": 1.9603786388214395e-05,
      "loss": 0.7157,
      "step": 1677
    },
    {
      "epoch": 0.21235130346747658,
      "grad_norm": 1.3034312725067139,
      "learning_rate": 1.9603286771785868e-05,
      "loss": 0.6952,
      "step": 1678
    },
    {
      "epoch": 0.21247785370792205,
      "grad_norm": 1.1671481132507324,
      "learning_rate": 1.960278684692791e-05,
      "loss": 0.672,
      "step": 1679
    },
    {
      "epoch": 0.2126044039483675,
      "grad_norm": 1.2904198169708252,
      "learning_rate": 1.9602286613656582e-05,
      "loss": 0.687,
      "step": 1680
    },
    {
      "epoch": 0.21273095418881297,
      "grad_norm": 1.3265031576156616,
      "learning_rate": 1.960178607198795e-05,
      "loss": 0.7269,
      "step": 1681
    },
    {
      "epoch": 0.2128575044292584,
      "grad_norm": 1.2561595439910889,
      "learning_rate": 1.960128522193808e-05,
      "loss": 0.6978,
      "step": 1682
    },
    {
      "epoch": 0.21298405466970388,
      "grad_norm": 1.4157966375350952,
      "learning_rate": 1.960078406352307e-05,
      "loss": 0.7046,
      "step": 1683
    },
    {
      "epoch": 0.21311060491014933,
      "grad_norm": 1.6269859075546265,
      "learning_rate": 1.9600282596759008e-05,
      "loss": 0.7226,
      "step": 1684
    },
    {
      "epoch": 0.2132371551505948,
      "grad_norm": 1.4148619174957275,
      "learning_rate": 1.9599780821662003e-05,
      "loss": 0.697,
      "step": 1685
    },
    {
      "epoch": 0.21336370539104024,
      "grad_norm": 1.2876383066177368,
      "learning_rate": 1.959927873824817e-05,
      "loss": 0.7137,
      "step": 1686
    },
    {
      "epoch": 0.2134902556314857,
      "grad_norm": 1.3770002126693726,
      "learning_rate": 1.9598776346533634e-05,
      "loss": 0.7374,
      "step": 1687
    },
    {
      "epoch": 0.21361680587193116,
      "grad_norm": 1.2667595148086548,
      "learning_rate": 1.959827364653453e-05,
      "loss": 0.7304,
      "step": 1688
    },
    {
      "epoch": 0.2137433561123766,
      "grad_norm": 3.1715126037597656,
      "learning_rate": 1.959777063826701e-05,
      "loss": 0.6464,
      "step": 1689
    },
    {
      "epoch": 0.21386990635282208,
      "grad_norm": 1.2449829578399658,
      "learning_rate": 1.9597267321747216e-05,
      "loss": 0.6646,
      "step": 1690
    },
    {
      "epoch": 0.21399645659326752,
      "grad_norm": 1.5769342184066772,
      "learning_rate": 1.959676369699132e-05,
      "loss": 0.6877,
      "step": 1691
    },
    {
      "epoch": 0.214123006833713,
      "grad_norm": 1.1906845569610596,
      "learning_rate": 1.95962597640155e-05,
      "loss": 0.7043,
      "step": 1692
    },
    {
      "epoch": 0.21424955707415844,
      "grad_norm": 1.5738296508789062,
      "learning_rate": 1.959575552283594e-05,
      "loss": 0.648,
      "step": 1693
    },
    {
      "epoch": 0.2143761073146039,
      "grad_norm": 1.2734756469726562,
      "learning_rate": 1.9595250973468825e-05,
      "loss": 0.6722,
      "step": 1694
    },
    {
      "epoch": 0.21450265755504935,
      "grad_norm": 1.4610594511032104,
      "learning_rate": 1.9594746115930374e-05,
      "loss": 0.7373,
      "step": 1695
    },
    {
      "epoch": 0.21462920779549483,
      "grad_norm": 1.1622661352157593,
      "learning_rate": 1.9594240950236792e-05,
      "loss": 0.6547,
      "step": 1696
    },
    {
      "epoch": 0.21475575803594027,
      "grad_norm": 1.3597232103347778,
      "learning_rate": 1.9593735476404304e-05,
      "loss": 0.7242,
      "step": 1697
    },
    {
      "epoch": 0.21488230827638571,
      "grad_norm": 1.3428728580474854,
      "learning_rate": 1.959322969444915e-05,
      "loss": 0.6725,
      "step": 1698
    },
    {
      "epoch": 0.21500885851683119,
      "grad_norm": 1.2740367650985718,
      "learning_rate": 1.9592723604387572e-05,
      "loss": 0.6541,
      "step": 1699
    },
    {
      "epoch": 0.21513540875727663,
      "grad_norm": 1.227531909942627,
      "learning_rate": 1.959221720623582e-05,
      "loss": 0.687,
      "step": 1700
    },
    {
      "epoch": 0.2152619589977221,
      "grad_norm": 1.2393560409545898,
      "learning_rate": 1.959171050001016e-05,
      "loss": 0.7038,
      "step": 1701
    },
    {
      "epoch": 0.21538850923816755,
      "grad_norm": 1.1840736865997314,
      "learning_rate": 1.9591203485726867e-05,
      "loss": 0.68,
      "step": 1702
    },
    {
      "epoch": 0.21551505947861302,
      "grad_norm": 1.1577540636062622,
      "learning_rate": 1.9590696163402228e-05,
      "loss": 0.678,
      "step": 1703
    },
    {
      "epoch": 0.21564160971905846,
      "grad_norm": 1.2725622653961182,
      "learning_rate": 1.9590188533052533e-05,
      "loss": 0.6804,
      "step": 1704
    },
    {
      "epoch": 0.21576815995950394,
      "grad_norm": 1.3712525367736816,
      "learning_rate": 1.9589680594694085e-05,
      "loss": 0.6606,
      "step": 1705
    },
    {
      "epoch": 0.21589471019994938,
      "grad_norm": 1.4411678314208984,
      "learning_rate": 1.9589172348343196e-05,
      "loss": 0.6597,
      "step": 1706
    },
    {
      "epoch": 0.21602126044039482,
      "grad_norm": 1.38253915309906,
      "learning_rate": 1.9588663794016196e-05,
      "loss": 0.6667,
      "step": 1707
    },
    {
      "epoch": 0.2161478106808403,
      "grad_norm": 1.985404133796692,
      "learning_rate": 1.9588154931729412e-05,
      "loss": 0.7117,
      "step": 1708
    },
    {
      "epoch": 0.21627436092128574,
      "grad_norm": 2.1242270469665527,
      "learning_rate": 1.958764576149919e-05,
      "loss": 0.7054,
      "step": 1709
    },
    {
      "epoch": 0.2164009111617312,
      "grad_norm": 1.2916743755340576,
      "learning_rate": 1.958713628334188e-05,
      "loss": 0.706,
      "step": 1710
    },
    {
      "epoch": 0.21652746140217666,
      "grad_norm": 1.2691528797149658,
      "learning_rate": 1.9586626497273854e-05,
      "loss": 0.7105,
      "step": 1711
    },
    {
      "epoch": 0.21665401164262213,
      "grad_norm": 1.4036459922790527,
      "learning_rate": 1.9586116403311474e-05,
      "loss": 0.7192,
      "step": 1712
    },
    {
      "epoch": 0.21678056188306757,
      "grad_norm": 1.3612534999847412,
      "learning_rate": 1.9585606001471124e-05,
      "loss": 0.7089,
      "step": 1713
    },
    {
      "epoch": 0.21690711212351305,
      "grad_norm": 1.22440505027771,
      "learning_rate": 1.95850952917692e-05,
      "loss": 0.6978,
      "step": 1714
    },
    {
      "epoch": 0.2170336623639585,
      "grad_norm": 1.288277268409729,
      "learning_rate": 1.9584584274222107e-05,
      "loss": 0.6822,
      "step": 1715
    },
    {
      "epoch": 0.21716021260440393,
      "grad_norm": 1.328555941581726,
      "learning_rate": 1.9584072948846254e-05,
      "loss": 0.6865,
      "step": 1716
    },
    {
      "epoch": 0.2172867628448494,
      "grad_norm": 1.2860289812088013,
      "learning_rate": 1.9583561315658063e-05,
      "loss": 0.6628,
      "step": 1717
    },
    {
      "epoch": 0.21741331308529485,
      "grad_norm": 1.1546012163162231,
      "learning_rate": 1.9583049374673965e-05,
      "loss": 0.6923,
      "step": 1718
    },
    {
      "epoch": 0.21753986332574032,
      "grad_norm": 1.2521576881408691,
      "learning_rate": 1.9582537125910406e-05,
      "loss": 0.6678,
      "step": 1719
    },
    {
      "epoch": 0.21766641356618577,
      "grad_norm": 1.2066609859466553,
      "learning_rate": 1.9582024569383833e-05,
      "loss": 0.6835,
      "step": 1720
    },
    {
      "epoch": 0.21779296380663124,
      "grad_norm": 1.2604353427886963,
      "learning_rate": 1.9581511705110712e-05,
      "loss": 0.6994,
      "step": 1721
    },
    {
      "epoch": 0.21791951404707668,
      "grad_norm": 1.2612601518630981,
      "learning_rate": 1.9580998533107516e-05,
      "loss": 0.6971,
      "step": 1722
    },
    {
      "epoch": 0.21804606428752216,
      "grad_norm": 1.3836424350738525,
      "learning_rate": 1.9580485053390722e-05,
      "loss": 0.6664,
      "step": 1723
    },
    {
      "epoch": 0.2181726145279676,
      "grad_norm": 1.5006049871444702,
      "learning_rate": 1.9579971265976825e-05,
      "loss": 0.7003,
      "step": 1724
    },
    {
      "epoch": 0.21829916476841307,
      "grad_norm": 1.367750644683838,
      "learning_rate": 1.957945717088232e-05,
      "loss": 0.68,
      "step": 1725
    },
    {
      "epoch": 0.21842571500885852,
      "grad_norm": 1.2269175052642822,
      "learning_rate": 1.957894276812373e-05,
      "loss": 0.6577,
      "step": 1726
    },
    {
      "epoch": 0.21855226524930396,
      "grad_norm": 1.1693462133407593,
      "learning_rate": 1.957842805771756e-05,
      "loss": 0.6685,
      "step": 1727
    },
    {
      "epoch": 0.21867881548974943,
      "grad_norm": 1.2783880233764648,
      "learning_rate": 1.9577913039680356e-05,
      "loss": 0.7044,
      "step": 1728
    },
    {
      "epoch": 0.21880536573019488,
      "grad_norm": 1.174206256866455,
      "learning_rate": 1.957739771402865e-05,
      "loss": 0.6712,
      "step": 1729
    },
    {
      "epoch": 0.21893191597064035,
      "grad_norm": 1.7605361938476562,
      "learning_rate": 1.9576882080779e-05,
      "loss": 0.7177,
      "step": 1730
    },
    {
      "epoch": 0.2190584662110858,
      "grad_norm": 1.1412365436553955,
      "learning_rate": 1.957636613994796e-05,
      "loss": 0.7125,
      "step": 1731
    },
    {
      "epoch": 0.21918501645153127,
      "grad_norm": 1.2537004947662354,
      "learning_rate": 1.95758498915521e-05,
      "loss": 0.6507,
      "step": 1732
    },
    {
      "epoch": 0.2193115666919767,
      "grad_norm": 1.1295982599258423,
      "learning_rate": 1.9575333335608005e-05,
      "loss": 0.6775,
      "step": 1733
    },
    {
      "epoch": 0.21943811693242218,
      "grad_norm": 1.2160670757293701,
      "learning_rate": 1.9574816472132267e-05,
      "loss": 0.689,
      "step": 1734
    },
    {
      "epoch": 0.21956466717286763,
      "grad_norm": 1.2153502702713013,
      "learning_rate": 1.9574299301141477e-05,
      "loss": 0.7267,
      "step": 1735
    },
    {
      "epoch": 0.21969121741331307,
      "grad_norm": 1.2247169017791748,
      "learning_rate": 1.9573781822652253e-05,
      "loss": 0.6663,
      "step": 1736
    },
    {
      "epoch": 0.21981776765375854,
      "grad_norm": 1.5008653402328491,
      "learning_rate": 1.957326403668121e-05,
      "loss": 0.7317,
      "step": 1737
    },
    {
      "epoch": 0.219944317894204,
      "grad_norm": 1.2424105405807495,
      "learning_rate": 1.9572745943244987e-05,
      "loss": 0.7123,
      "step": 1738
    },
    {
      "epoch": 0.22007086813464946,
      "grad_norm": 1.495927333831787,
      "learning_rate": 1.957222754236021e-05,
      "loss": 0.7205,
      "step": 1739
    },
    {
      "epoch": 0.2201974183750949,
      "grad_norm": 1.3056093454360962,
      "learning_rate": 1.957170883404354e-05,
      "loss": 0.7079,
      "step": 1740
    },
    {
      "epoch": 0.22032396861554038,
      "grad_norm": 1.253096342086792,
      "learning_rate": 1.9571189818311625e-05,
      "loss": 0.6976,
      "step": 1741
    },
    {
      "epoch": 0.22045051885598582,
      "grad_norm": 1.2876646518707275,
      "learning_rate": 1.957067049518115e-05,
      "loss": 0.7006,
      "step": 1742
    },
    {
      "epoch": 0.2205770690964313,
      "grad_norm": 1.1811401844024658,
      "learning_rate": 1.957015086466878e-05,
      "loss": 0.6472,
      "step": 1743
    },
    {
      "epoch": 0.22070361933687674,
      "grad_norm": 1.2132327556610107,
      "learning_rate": 1.956963092679121e-05,
      "loss": 0.7005,
      "step": 1744
    },
    {
      "epoch": 0.2208301695773222,
      "grad_norm": 1.2026102542877197,
      "learning_rate": 1.956911068156514e-05,
      "loss": 0.7195,
      "step": 1745
    },
    {
      "epoch": 0.22095671981776766,
      "grad_norm": 1.207983136177063,
      "learning_rate": 1.9568590129007272e-05,
      "loss": 0.6656,
      "step": 1746
    },
    {
      "epoch": 0.2210832700582131,
      "grad_norm": 1.3305339813232422,
      "learning_rate": 1.956806926913433e-05,
      "loss": 0.6341,
      "step": 1747
    },
    {
      "epoch": 0.22120982029865857,
      "grad_norm": 1.4073107242584229,
      "learning_rate": 1.9567548101963044e-05,
      "loss": 0.6884,
      "step": 1748
    },
    {
      "epoch": 0.22133637053910402,
      "grad_norm": 1.2711278200149536,
      "learning_rate": 1.956702662751015e-05,
      "loss": 0.6853,
      "step": 1749
    },
    {
      "epoch": 0.2214629207795495,
      "grad_norm": 1.2278355360031128,
      "learning_rate": 1.95665048457924e-05,
      "loss": 0.6685,
      "step": 1750
    },
    {
      "epoch": 0.22158947101999493,
      "grad_norm": 1.183292031288147,
      "learning_rate": 1.9565982756826543e-05,
      "loss": 0.7004,
      "step": 1751
    },
    {
      "epoch": 0.2217160212604404,
      "grad_norm": 1.2002910375595093,
      "learning_rate": 1.9565460360629358e-05,
      "loss": 0.7275,
      "step": 1752
    },
    {
      "epoch": 0.22184257150088585,
      "grad_norm": 17.713176727294922,
      "learning_rate": 1.956493765721761e-05,
      "loss": 0.6682,
      "step": 1753
    },
    {
      "epoch": 0.22196912174133132,
      "grad_norm": 1.33478844165802,
      "learning_rate": 1.9564414646608103e-05,
      "loss": 0.7123,
      "step": 1754
    },
    {
      "epoch": 0.22209567198177677,
      "grad_norm": 1.2045024633407593,
      "learning_rate": 1.956389132881762e-05,
      "loss": 0.645,
      "step": 1755
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 1.184607982635498,
      "learning_rate": 1.9563367703862976e-05,
      "loss": 0.6585,
      "step": 1756
    },
    {
      "epoch": 0.22234877246266768,
      "grad_norm": 2.310594320297241,
      "learning_rate": 1.9562843771760988e-05,
      "loss": 0.6898,
      "step": 1757
    },
    {
      "epoch": 0.22247532270311313,
      "grad_norm": 1.376968502998352,
      "learning_rate": 1.9562319532528478e-05,
      "loss": 0.7064,
      "step": 1758
    },
    {
      "epoch": 0.2226018729435586,
      "grad_norm": 1.68683660030365,
      "learning_rate": 1.9561794986182286e-05,
      "loss": 0.683,
      "step": 1759
    },
    {
      "epoch": 0.22272842318400404,
      "grad_norm": 1.6651475429534912,
      "learning_rate": 1.9561270132739265e-05,
      "loss": 0.647,
      "step": 1760
    },
    {
      "epoch": 0.22285497342444952,
      "grad_norm": 1.406321406364441,
      "learning_rate": 1.9560744972216263e-05,
      "loss": 0.7161,
      "step": 1761
    },
    {
      "epoch": 0.22298152366489496,
      "grad_norm": 1.5299592018127441,
      "learning_rate": 1.956021950463015e-05,
      "loss": 0.6901,
      "step": 1762
    },
    {
      "epoch": 0.22310807390534043,
      "grad_norm": 1.3885606527328491,
      "learning_rate": 1.95596937299978e-05,
      "loss": 0.6815,
      "step": 1763
    },
    {
      "epoch": 0.22323462414578588,
      "grad_norm": 1.3310248851776123,
      "learning_rate": 1.9559167648336105e-05,
      "loss": 0.6854,
      "step": 1764
    },
    {
      "epoch": 0.22336117438623132,
      "grad_norm": 1.522231936454773,
      "learning_rate": 1.955864125966196e-05,
      "loss": 0.7268,
      "step": 1765
    },
    {
      "epoch": 0.2234877246266768,
      "grad_norm": 1.1657569408416748,
      "learning_rate": 1.9558114563992263e-05,
      "loss": 0.6849,
      "step": 1766
    },
    {
      "epoch": 0.22361427486712224,
      "grad_norm": 1.568596601486206,
      "learning_rate": 1.955758756134394e-05,
      "loss": 0.6946,
      "step": 1767
    },
    {
      "epoch": 0.2237408251075677,
      "grad_norm": 1.3488593101501465,
      "learning_rate": 1.955706025173391e-05,
      "loss": 0.7249,
      "step": 1768
    },
    {
      "epoch": 0.22386737534801315,
      "grad_norm": 1.7654660940170288,
      "learning_rate": 1.9556532635179113e-05,
      "loss": 0.6999,
      "step": 1769
    },
    {
      "epoch": 0.22399392558845863,
      "grad_norm": 1.2630524635314941,
      "learning_rate": 1.955600471169649e-05,
      "loss": 0.67,
      "step": 1770
    },
    {
      "epoch": 0.22412047582890407,
      "grad_norm": 1.8584041595458984,
      "learning_rate": 1.9555476481303003e-05,
      "loss": 0.7141,
      "step": 1771
    },
    {
      "epoch": 0.22424702606934954,
      "grad_norm": 1.8873634338378906,
      "learning_rate": 1.9554947944015614e-05,
      "loss": 0.7006,
      "step": 1772
    },
    {
      "epoch": 0.224373576309795,
      "grad_norm": 1.4616169929504395,
      "learning_rate": 1.9554419099851295e-05,
      "loss": 0.6878,
      "step": 1773
    },
    {
      "epoch": 0.22450012655024046,
      "grad_norm": 1.8282945156097412,
      "learning_rate": 1.9553889948827035e-05,
      "loss": 0.6719,
      "step": 1774
    },
    {
      "epoch": 0.2246266767906859,
      "grad_norm": 1.1866344213485718,
      "learning_rate": 1.9553360490959824e-05,
      "loss": 0.7723,
      "step": 1775
    },
    {
      "epoch": 0.22475322703113135,
      "grad_norm": 1.643118977546692,
      "learning_rate": 1.955283072626667e-05,
      "loss": 0.7294,
      "step": 1776
    },
    {
      "epoch": 0.22487977727157682,
      "grad_norm": 1.233877420425415,
      "learning_rate": 1.9552300654764592e-05,
      "loss": 0.6749,
      "step": 1777
    },
    {
      "epoch": 0.22500632751202226,
      "grad_norm": 1.6793513298034668,
      "learning_rate": 1.9551770276470608e-05,
      "loss": 0.6841,
      "step": 1778
    },
    {
      "epoch": 0.22513287775246774,
      "grad_norm": 1.3726403713226318,
      "learning_rate": 1.9551239591401748e-05,
      "loss": 0.6728,
      "step": 1779
    },
    {
      "epoch": 0.22525942799291318,
      "grad_norm": 1.3199126720428467,
      "learning_rate": 1.9550708599575068e-05,
      "loss": 0.6621,
      "step": 1780
    },
    {
      "epoch": 0.22538597823335865,
      "grad_norm": 1.3049207925796509,
      "learning_rate": 1.9550177301007617e-05,
      "loss": 0.7196,
      "step": 1781
    },
    {
      "epoch": 0.2255125284738041,
      "grad_norm": 1.4387173652648926,
      "learning_rate": 1.9549645695716454e-05,
      "loss": 0.6875,
      "step": 1782
    },
    {
      "epoch": 0.22563907871424957,
      "grad_norm": 1.2417938709259033,
      "learning_rate": 1.9549113783718655e-05,
      "loss": 0.6734,
      "step": 1783
    },
    {
      "epoch": 0.225765628954695,
      "grad_norm": 1.5864157676696777,
      "learning_rate": 1.9548581565031307e-05,
      "loss": 0.6684,
      "step": 1784
    },
    {
      "epoch": 0.22589217919514046,
      "grad_norm": 1.462911605834961,
      "learning_rate": 1.95480490396715e-05,
      "loss": 0.7017,
      "step": 1785
    },
    {
      "epoch": 0.22601872943558593,
      "grad_norm": 1.6802489757537842,
      "learning_rate": 1.954751620765634e-05,
      "loss": 0.6983,
      "step": 1786
    },
    {
      "epoch": 0.22614527967603137,
      "grad_norm": 1.7295048236846924,
      "learning_rate": 1.9546983069002934e-05,
      "loss": 0.7,
      "step": 1787
    },
    {
      "epoch": 0.22627182991647685,
      "grad_norm": 1.3953408002853394,
      "learning_rate": 1.9546449623728412e-05,
      "loss": 0.6729,
      "step": 1788
    },
    {
      "epoch": 0.2263983801569223,
      "grad_norm": 1.5908665657043457,
      "learning_rate": 1.9545915871849904e-05,
      "loss": 0.7024,
      "step": 1789
    },
    {
      "epoch": 0.22652493039736776,
      "grad_norm": 1.2763930559158325,
      "learning_rate": 1.954538181338455e-05,
      "loss": 0.6583,
      "step": 1790
    },
    {
      "epoch": 0.2266514806378132,
      "grad_norm": 1.3946034908294678,
      "learning_rate": 1.9544847448349506e-05,
      "loss": 0.7122,
      "step": 1791
    },
    {
      "epoch": 0.22677803087825868,
      "grad_norm": 1.1451077461242676,
      "learning_rate": 1.954431277676193e-05,
      "loss": 0.6315,
      "step": 1792
    },
    {
      "epoch": 0.22690458111870412,
      "grad_norm": 1.1141955852508545,
      "learning_rate": 1.9543777798639e-05,
      "loss": 0.6992,
      "step": 1793
    },
    {
      "epoch": 0.2270311313591496,
      "grad_norm": 1.2341190576553345,
      "learning_rate": 1.9543242513997893e-05,
      "loss": 0.742,
      "step": 1794
    },
    {
      "epoch": 0.22715768159959504,
      "grad_norm": 1.1940155029296875,
      "learning_rate": 1.9542706922855804e-05,
      "loss": 0.7076,
      "step": 1795
    },
    {
      "epoch": 0.22728423184004048,
      "grad_norm": 3.60657000541687,
      "learning_rate": 1.954217102522993e-05,
      "loss": 0.7537,
      "step": 1796
    },
    {
      "epoch": 0.22741078208048596,
      "grad_norm": 1.2712105512619019,
      "learning_rate": 1.954163482113749e-05,
      "loss": 0.7533,
      "step": 1797
    },
    {
      "epoch": 0.2275373323209314,
      "grad_norm": 1.3817026615142822,
      "learning_rate": 1.95410983105957e-05,
      "loss": 0.7008,
      "step": 1798
    },
    {
      "epoch": 0.22766388256137687,
      "grad_norm": 14.493037223815918,
      "learning_rate": 1.954056149362179e-05,
      "loss": 0.7048,
      "step": 1799
    },
    {
      "epoch": 0.22779043280182232,
      "grad_norm": 1.682161808013916,
      "learning_rate": 1.9540024370233004e-05,
      "loss": 0.7305,
      "step": 1800
    },
    {
      "epoch": 0.2279169830422678,
      "grad_norm": 1.2907742261886597,
      "learning_rate": 1.9539486940446593e-05,
      "loss": 0.7041,
      "step": 1801
    },
    {
      "epoch": 0.22804353328271323,
      "grad_norm": 1.397018313407898,
      "learning_rate": 1.9538949204279814e-05,
      "loss": 0.6915,
      "step": 1802
    },
    {
      "epoch": 0.2281700835231587,
      "grad_norm": 1.3266624212265015,
      "learning_rate": 1.9538411161749943e-05,
      "loss": 0.7243,
      "step": 1803
    },
    {
      "epoch": 0.22829663376360415,
      "grad_norm": 3.146108865737915,
      "learning_rate": 1.9537872812874258e-05,
      "loss": 0.6817,
      "step": 1804
    },
    {
      "epoch": 0.2284231840040496,
      "grad_norm": 1.3825069665908813,
      "learning_rate": 1.9537334157670047e-05,
      "loss": 0.6935,
      "step": 1805
    },
    {
      "epoch": 0.22854973424449507,
      "grad_norm": 1.1787012815475464,
      "learning_rate": 1.9536795196154608e-05,
      "loss": 0.6803,
      "step": 1806
    },
    {
      "epoch": 0.2286762844849405,
      "grad_norm": 1.4747250080108643,
      "learning_rate": 1.953625592834526e-05,
      "loss": 0.7162,
      "step": 1807
    },
    {
      "epoch": 0.22880283472538598,
      "grad_norm": 1.1680315732955933,
      "learning_rate": 1.9535716354259313e-05,
      "loss": 0.6594,
      "step": 1808
    },
    {
      "epoch": 0.22892938496583143,
      "grad_norm": 1.1918264627456665,
      "learning_rate": 1.9535176473914104e-05,
      "loss": 0.7057,
      "step": 1809
    },
    {
      "epoch": 0.2290559352062769,
      "grad_norm": 1.8603367805480957,
      "learning_rate": 1.953463628732697e-05,
      "loss": 0.7037,
      "step": 1810
    },
    {
      "epoch": 0.22918248544672234,
      "grad_norm": 1.3642618656158447,
      "learning_rate": 1.9534095794515254e-05,
      "loss": 0.6536,
      "step": 1811
    },
    {
      "epoch": 0.22930903568716782,
      "grad_norm": 1.309769868850708,
      "learning_rate": 1.9533554995496325e-05,
      "loss": 0.7037,
      "step": 1812
    },
    {
      "epoch": 0.22943558592761326,
      "grad_norm": 1.291043758392334,
      "learning_rate": 1.9533013890287545e-05,
      "loss": 0.6715,
      "step": 1813
    },
    {
      "epoch": 0.2295621361680587,
      "grad_norm": 1.287350058555603,
      "learning_rate": 1.9532472478906296e-05,
      "loss": 0.7116,
      "step": 1814
    },
    {
      "epoch": 0.22968868640850418,
      "grad_norm": 1.2031333446502686,
      "learning_rate": 1.9531930761369965e-05,
      "loss": 0.6833,
      "step": 1815
    },
    {
      "epoch": 0.22981523664894962,
      "grad_norm": 1.8785243034362793,
      "learning_rate": 1.953138873769595e-05,
      "loss": 0.705,
      "step": 1816
    },
    {
      "epoch": 0.2299417868893951,
      "grad_norm": 1.1785389184951782,
      "learning_rate": 1.9530846407901664e-05,
      "loss": 0.6584,
      "step": 1817
    },
    {
      "epoch": 0.23006833712984054,
      "grad_norm": 1.389325499534607,
      "learning_rate": 1.953030377200452e-05,
      "loss": 0.6827,
      "step": 1818
    },
    {
      "epoch": 0.230194887370286,
      "grad_norm": 1.2255662679672241,
      "learning_rate": 1.9529760830021945e-05,
      "loss": 0.6776,
      "step": 1819
    },
    {
      "epoch": 0.23032143761073146,
      "grad_norm": 1.5149942636489868,
      "learning_rate": 1.952921758197138e-05,
      "loss": 0.7162,
      "step": 1820
    },
    {
      "epoch": 0.23044798785117693,
      "grad_norm": 1.247291922569275,
      "learning_rate": 1.9528674027870268e-05,
      "loss": 0.6559,
      "step": 1821
    },
    {
      "epoch": 0.23057453809162237,
      "grad_norm": 1.1396353244781494,
      "learning_rate": 1.9528130167736073e-05,
      "loss": 0.6546,
      "step": 1822
    },
    {
      "epoch": 0.23070108833206784,
      "grad_norm": 1.156215786933899,
      "learning_rate": 1.952758600158626e-05,
      "loss": 0.6311,
      "step": 1823
    },
    {
      "epoch": 0.2308276385725133,
      "grad_norm": 1.2444770336151123,
      "learning_rate": 1.9527041529438303e-05,
      "loss": 0.6937,
      "step": 1824
    },
    {
      "epoch": 0.23095418881295873,
      "grad_norm": 1.2092866897583008,
      "learning_rate": 1.952649675130969e-05,
      "loss": 0.655,
      "step": 1825
    },
    {
      "epoch": 0.2310807390534042,
      "grad_norm": 1.1920949220657349,
      "learning_rate": 1.9525951667217917e-05,
      "loss": 0.6995,
      "step": 1826
    },
    {
      "epoch": 0.23120728929384965,
      "grad_norm": 1.36337411403656,
      "learning_rate": 1.9525406277180498e-05,
      "loss": 0.6417,
      "step": 1827
    },
    {
      "epoch": 0.23133383953429512,
      "grad_norm": 1.2544351816177368,
      "learning_rate": 1.9524860581214934e-05,
      "loss": 0.6763,
      "step": 1828
    },
    {
      "epoch": 0.23146038977474057,
      "grad_norm": 1.176026701927185,
      "learning_rate": 1.952431457933877e-05,
      "loss": 0.6987,
      "step": 1829
    },
    {
      "epoch": 0.23158694001518604,
      "grad_norm": 1.1660873889923096,
      "learning_rate": 1.9523768271569524e-05,
      "loss": 0.6761,
      "step": 1830
    },
    {
      "epoch": 0.23171349025563148,
      "grad_norm": 1.2126883268356323,
      "learning_rate": 1.9523221657924755e-05,
      "loss": 0.672,
      "step": 1831
    },
    {
      "epoch": 0.23184004049607695,
      "grad_norm": 1.1858030557632446,
      "learning_rate": 1.9522674738422013e-05,
      "loss": 0.6672,
      "step": 1832
    },
    {
      "epoch": 0.2319665907365224,
      "grad_norm": 1.1751772165298462,
      "learning_rate": 1.9522127513078866e-05,
      "loss": 0.7114,
      "step": 1833
    },
    {
      "epoch": 0.23209314097696784,
      "grad_norm": 1.2079445123672485,
      "learning_rate": 1.9521579981912884e-05,
      "loss": 0.7106,
      "step": 1834
    },
    {
      "epoch": 0.23221969121741332,
      "grad_norm": 1.3109147548675537,
      "learning_rate": 1.9521032144941655e-05,
      "loss": 0.7158,
      "step": 1835
    },
    {
      "epoch": 0.23234624145785876,
      "grad_norm": 1.14360511302948,
      "learning_rate": 1.9520484002182777e-05,
      "loss": 0.6739,
      "step": 1836
    },
    {
      "epoch": 0.23247279169830423,
      "grad_norm": 1.1544556617736816,
      "learning_rate": 1.951993555365385e-05,
      "loss": 0.6631,
      "step": 1837
    },
    {
      "epoch": 0.23259934193874968,
      "grad_norm": 1.1638606786727905,
      "learning_rate": 1.9519386799372494e-05,
      "loss": 0.6829,
      "step": 1838
    },
    {
      "epoch": 0.23272589217919515,
      "grad_norm": 1.127072811126709,
      "learning_rate": 1.951883773935633e-05,
      "loss": 0.6333,
      "step": 1839
    },
    {
      "epoch": 0.2328524424196406,
      "grad_norm": 1.1732773780822754,
      "learning_rate": 1.951828837362299e-05,
      "loss": 0.7212,
      "step": 1840
    },
    {
      "epoch": 0.23297899266008607,
      "grad_norm": 1.187272310256958,
      "learning_rate": 1.9517738702190122e-05,
      "loss": 0.7339,
      "step": 1841
    },
    {
      "epoch": 0.2331055429005315,
      "grad_norm": 1.199961543083191,
      "learning_rate": 1.951718872507538e-05,
      "loss": 0.6659,
      "step": 1842
    },
    {
      "epoch": 0.23323209314097698,
      "grad_norm": 1.263596773147583,
      "learning_rate": 1.9516638442296423e-05,
      "loss": 0.7093,
      "step": 1843
    },
    {
      "epoch": 0.23335864338142243,
      "grad_norm": 1.201698899269104,
      "learning_rate": 1.9516087853870928e-05,
      "loss": 0.6798,
      "step": 1844
    },
    {
      "epoch": 0.23348519362186787,
      "grad_norm": 1.2291878461837769,
      "learning_rate": 1.951553695981658e-05,
      "loss": 0.6437,
      "step": 1845
    },
    {
      "epoch": 0.23361174386231334,
      "grad_norm": 1.1325397491455078,
      "learning_rate": 1.9514985760151068e-05,
      "loss": 0.6598,
      "step": 1846
    },
    {
      "epoch": 0.2337382941027588,
      "grad_norm": 1.1792720556259155,
      "learning_rate": 1.9514434254892092e-05,
      "loss": 0.717,
      "step": 1847
    },
    {
      "epoch": 0.23386484434320426,
      "grad_norm": 1.2474173307418823,
      "learning_rate": 1.951388244405738e-05,
      "loss": 0.6465,
      "step": 1848
    },
    {
      "epoch": 0.2339913945836497,
      "grad_norm": 1.2849096059799194,
      "learning_rate": 1.9513330327664633e-05,
      "loss": 0.7407,
      "step": 1849
    },
    {
      "epoch": 0.23411794482409518,
      "grad_norm": 1.2685158252716064,
      "learning_rate": 1.95127779057316e-05,
      "loss": 0.6756,
      "step": 1850
    },
    {
      "epoch": 0.23424449506454062,
      "grad_norm": 1.2960704565048218,
      "learning_rate": 1.9512225178276015e-05,
      "loss": 0.697,
      "step": 1851
    },
    {
      "epoch": 0.2343710453049861,
      "grad_norm": 1.1558823585510254,
      "learning_rate": 1.9511672145315636e-05,
      "loss": 0.6973,
      "step": 1852
    },
    {
      "epoch": 0.23449759554543154,
      "grad_norm": 1.2331045866012573,
      "learning_rate": 1.951111880686822e-05,
      "loss": 0.6748,
      "step": 1853
    },
    {
      "epoch": 0.23462414578587698,
      "grad_norm": 1.267952561378479,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 0.6706,
      "step": 1854
    },
    {
      "epoch": 0.23475069602632245,
      "grad_norm": 1.1614494323730469,
      "learning_rate": 1.9510011213583374e-05,
      "loss": 0.6428,
      "step": 1855
    },
    {
      "epoch": 0.2348772462667679,
      "grad_norm": 1.2557008266448975,
      "learning_rate": 1.950945695878152e-05,
      "loss": 0.6934,
      "step": 1856
    },
    {
      "epoch": 0.23500379650721337,
      "grad_norm": 1.2266710996627808,
      "learning_rate": 1.9508902398563775e-05,
      "loss": 0.6935,
      "step": 1857
    },
    {
      "epoch": 0.2351303467476588,
      "grad_norm": 1.2143179178237915,
      "learning_rate": 1.9508347532947946e-05,
      "loss": 0.6728,
      "step": 1858
    },
    {
      "epoch": 0.23525689698810429,
      "grad_norm": 1.2797080278396606,
      "learning_rate": 1.9507792361951864e-05,
      "loss": 0.6725,
      "step": 1859
    },
    {
      "epoch": 0.23538344722854973,
      "grad_norm": 1.244854211807251,
      "learning_rate": 1.950723688559335e-05,
      "loss": 0.7222,
      "step": 1860
    },
    {
      "epoch": 0.2355099974689952,
      "grad_norm": 1.2376166582107544,
      "learning_rate": 1.950668110389025e-05,
      "loss": 0.6972,
      "step": 1861
    },
    {
      "epoch": 0.23563654770944065,
      "grad_norm": 1.2259522676467896,
      "learning_rate": 1.9506125016860412e-05,
      "loss": 0.6829,
      "step": 1862
    },
    {
      "epoch": 0.2357630979498861,
      "grad_norm": 1.1959218978881836,
      "learning_rate": 1.9505568624521696e-05,
      "loss": 0.6655,
      "step": 1863
    },
    {
      "epoch": 0.23588964819033156,
      "grad_norm": 1.1768079996109009,
      "learning_rate": 1.950501192689197e-05,
      "loss": 0.639,
      "step": 1864
    },
    {
      "epoch": 0.236016198430777,
      "grad_norm": 1.171636939048767,
      "learning_rate": 1.9504454923989116e-05,
      "loss": 0.6475,
      "step": 1865
    },
    {
      "epoch": 0.23614274867122248,
      "grad_norm": 1.3487186431884766,
      "learning_rate": 1.950389761583102e-05,
      "loss": 0.7037,
      "step": 1866
    },
    {
      "epoch": 0.23626929891166792,
      "grad_norm": 4.094715118408203,
      "learning_rate": 1.9503340002435585e-05,
      "loss": 0.6917,
      "step": 1867
    },
    {
      "epoch": 0.2363958491521134,
      "grad_norm": 1.6452888250350952,
      "learning_rate": 1.950278208382072e-05,
      "loss": 0.7124,
      "step": 1868
    },
    {
      "epoch": 0.23652239939255884,
      "grad_norm": 1.3878976106643677,
      "learning_rate": 1.9502223860004343e-05,
      "loss": 0.7351,
      "step": 1869
    },
    {
      "epoch": 0.2366489496330043,
      "grad_norm": 1.4509079456329346,
      "learning_rate": 1.9501665331004376e-05,
      "loss": 0.6702,
      "step": 1870
    },
    {
      "epoch": 0.23677549987344976,
      "grad_norm": 1.270859956741333,
      "learning_rate": 1.9501106496838768e-05,
      "loss": 0.6874,
      "step": 1871
    },
    {
      "epoch": 0.23690205011389523,
      "grad_norm": 1.2845653295516968,
      "learning_rate": 1.9500547357525457e-05,
      "loss": 0.6237,
      "step": 1872
    },
    {
      "epoch": 0.23702860035434067,
      "grad_norm": 1.305277943611145,
      "learning_rate": 1.9499987913082412e-05,
      "loss": 0.7351,
      "step": 1873
    },
    {
      "epoch": 0.23715515059478612,
      "grad_norm": 1.2960280179977417,
      "learning_rate": 1.949942816352759e-05,
      "loss": 0.716,
      "step": 1874
    },
    {
      "epoch": 0.2372817008352316,
      "grad_norm": 1.2455377578735352,
      "learning_rate": 1.9498868108878973e-05,
      "loss": 0.7094,
      "step": 1875
    },
    {
      "epoch": 0.23740825107567703,
      "grad_norm": 1.1325632333755493,
      "learning_rate": 1.949830774915455e-05,
      "loss": 0.7131,
      "step": 1876
    },
    {
      "epoch": 0.2375348013161225,
      "grad_norm": 1.1845518350601196,
      "learning_rate": 1.9497747084372318e-05,
      "loss": 0.6793,
      "step": 1877
    },
    {
      "epoch": 0.23766135155656795,
      "grad_norm": 1.6691175699234009,
      "learning_rate": 1.9497186114550278e-05,
      "loss": 0.6704,
      "step": 1878
    },
    {
      "epoch": 0.23778790179701342,
      "grad_norm": 1.136674165725708,
      "learning_rate": 1.9496624839706456e-05,
      "loss": 0.6714,
      "step": 1879
    },
    {
      "epoch": 0.23791445203745887,
      "grad_norm": 1.2732514142990112,
      "learning_rate": 1.949606325985887e-05,
      "loss": 0.6632,
      "step": 1880
    },
    {
      "epoch": 0.23804100227790434,
      "grad_norm": 1.3061705827713013,
      "learning_rate": 1.949550137502556e-05,
      "loss": 0.7585,
      "step": 1881
    },
    {
      "epoch": 0.23816755251834978,
      "grad_norm": 3.675151824951172,
      "learning_rate": 1.9494939185224576e-05,
      "loss": 0.7269,
      "step": 1882
    },
    {
      "epoch": 0.23829410275879523,
      "grad_norm": 1.208816647529602,
      "learning_rate": 1.949437669047397e-05,
      "loss": 0.6661,
      "step": 1883
    },
    {
      "epoch": 0.2384206529992407,
      "grad_norm": 1.2818204164505005,
      "learning_rate": 1.9493813890791803e-05,
      "loss": 0.7181,
      "step": 1884
    },
    {
      "epoch": 0.23854720323968615,
      "grad_norm": 1.4528851509094238,
      "learning_rate": 1.9493250786196156e-05,
      "loss": 0.7202,
      "step": 1885
    },
    {
      "epoch": 0.23867375348013162,
      "grad_norm": 1.190704345703125,
      "learning_rate": 1.9492687376705115e-05,
      "loss": 0.6673,
      "step": 1886
    },
    {
      "epoch": 0.23880030372057706,
      "grad_norm": 1.2506771087646484,
      "learning_rate": 1.949212366233677e-05,
      "loss": 0.6657,
      "step": 1887
    },
    {
      "epoch": 0.23892685396102253,
      "grad_norm": 1.2396209239959717,
      "learning_rate": 1.9491559643109236e-05,
      "loss": 0.7047,
      "step": 1888
    },
    {
      "epoch": 0.23905340420146798,
      "grad_norm": 1.295107364654541,
      "learning_rate": 1.9490995319040614e-05,
      "loss": 0.7275,
      "step": 1889
    },
    {
      "epoch": 0.23917995444191345,
      "grad_norm": 1.2297542095184326,
      "learning_rate": 1.949043069014904e-05,
      "loss": 0.6949,
      "step": 1890
    },
    {
      "epoch": 0.2393065046823589,
      "grad_norm": 1.2025843858718872,
      "learning_rate": 1.9489865756452642e-05,
      "loss": 0.6763,
      "step": 1891
    },
    {
      "epoch": 0.23943305492280437,
      "grad_norm": 1.2218292951583862,
      "learning_rate": 1.9489300517969565e-05,
      "loss": 0.682,
      "step": 1892
    },
    {
      "epoch": 0.2395596051632498,
      "grad_norm": 1.1721712350845337,
      "learning_rate": 1.9488734974717965e-05,
      "loss": 0.6857,
      "step": 1893
    },
    {
      "epoch": 0.23968615540369526,
      "grad_norm": 1.1570247411727905,
      "learning_rate": 1.9488169126716e-05,
      "loss": 0.7042,
      "step": 1894
    },
    {
      "epoch": 0.23981270564414073,
      "grad_norm": 1.1461262702941895,
      "learning_rate": 1.9487602973981853e-05,
      "loss": 0.6835,
      "step": 1895
    },
    {
      "epoch": 0.23993925588458617,
      "grad_norm": 1.1641522645950317,
      "learning_rate": 1.94870365165337e-05,
      "loss": 0.6522,
      "step": 1896
    },
    {
      "epoch": 0.24006580612503164,
      "grad_norm": 1.6170024871826172,
      "learning_rate": 1.948646975438973e-05,
      "loss": 0.7088,
      "step": 1897
    },
    {
      "epoch": 0.2401923563654771,
      "grad_norm": 1.181496262550354,
      "learning_rate": 1.9485902687568158e-05,
      "loss": 0.6684,
      "step": 1898
    },
    {
      "epoch": 0.24031890660592256,
      "grad_norm": 1.9641883373260498,
      "learning_rate": 1.9485335316087187e-05,
      "loss": 0.6882,
      "step": 1899
    },
    {
      "epoch": 0.240445456846368,
      "grad_norm": 1.1738696098327637,
      "learning_rate": 1.9484767639965044e-05,
      "loss": 0.6862,
      "step": 1900
    },
    {
      "epoch": 0.24057200708681348,
      "grad_norm": 1.2199281454086304,
      "learning_rate": 1.9484199659219952e-05,
      "loss": 0.6628,
      "step": 1901
    },
    {
      "epoch": 0.24069855732725892,
      "grad_norm": 1.1094003915786743,
      "learning_rate": 1.9483631373870167e-05,
      "loss": 0.671,
      "step": 1902
    },
    {
      "epoch": 0.24082510756770437,
      "grad_norm": 1.1532506942749023,
      "learning_rate": 1.9483062783933935e-05,
      "loss": 0.6574,
      "step": 1903
    },
    {
      "epoch": 0.24095165780814984,
      "grad_norm": 1.183184266090393,
      "learning_rate": 1.948249388942951e-05,
      "loss": 0.6627,
      "step": 1904
    },
    {
      "epoch": 0.24107820804859528,
      "grad_norm": 1.141011118888855,
      "learning_rate": 1.9481924690375176e-05,
      "loss": 0.6557,
      "step": 1905
    },
    {
      "epoch": 0.24120475828904075,
      "grad_norm": 1.1403331756591797,
      "learning_rate": 1.9481355186789203e-05,
      "loss": 0.6663,
      "step": 1906
    },
    {
      "epoch": 0.2413313085294862,
      "grad_norm": 1.171991229057312,
      "learning_rate": 1.9480785378689885e-05,
      "loss": 0.7613,
      "step": 1907
    },
    {
      "epoch": 0.24145785876993167,
      "grad_norm": 1.2879053354263306,
      "learning_rate": 1.9480215266095527e-05,
      "loss": 0.675,
      "step": 1908
    },
    {
      "epoch": 0.24158440901037712,
      "grad_norm": 1.1402301788330078,
      "learning_rate": 1.9479644849024436e-05,
      "loss": 0.6569,
      "step": 1909
    },
    {
      "epoch": 0.2417109592508226,
      "grad_norm": 1.1769901514053345,
      "learning_rate": 1.947907412749493e-05,
      "loss": 0.6726,
      "step": 1910
    },
    {
      "epoch": 0.24183750949126803,
      "grad_norm": 1.2106738090515137,
      "learning_rate": 1.947850310152534e-05,
      "loss": 0.6662,
      "step": 1911
    },
    {
      "epoch": 0.24196405973171348,
      "grad_norm": 1.1447632312774658,
      "learning_rate": 1.947793177113401e-05,
      "loss": 0.6845,
      "step": 1912
    },
    {
      "epoch": 0.24209060997215895,
      "grad_norm": 1.2221596240997314,
      "learning_rate": 1.947736013633929e-05,
      "loss": 0.6904,
      "step": 1913
    },
    {
      "epoch": 0.2422171602126044,
      "grad_norm": 1.2264511585235596,
      "learning_rate": 1.9476788197159526e-05,
      "loss": 0.7164,
      "step": 1914
    },
    {
      "epoch": 0.24234371045304987,
      "grad_norm": 2.7189788818359375,
      "learning_rate": 1.94762159536131e-05,
      "loss": 0.7321,
      "step": 1915
    },
    {
      "epoch": 0.2424702606934953,
      "grad_norm": 1.2568784952163696,
      "learning_rate": 1.9475643405718388e-05,
      "loss": 0.6679,
      "step": 1916
    },
    {
      "epoch": 0.24259681093394078,
      "grad_norm": 1.341949224472046,
      "learning_rate": 1.947507055349378e-05,
      "loss": 0.6845,
      "step": 1917
    },
    {
      "epoch": 0.24272336117438623,
      "grad_norm": 1.3587394952774048,
      "learning_rate": 1.947449739695767e-05,
      "loss": 0.6981,
      "step": 1918
    },
    {
      "epoch": 0.2428499114148317,
      "grad_norm": 1.5399433374404907,
      "learning_rate": 1.9473923936128467e-05,
      "loss": 0.665,
      "step": 1919
    },
    {
      "epoch": 0.24297646165527714,
      "grad_norm": 1.198602318763733,
      "learning_rate": 1.9473350171024594e-05,
      "loss": 0.7068,
      "step": 1920
    },
    {
      "epoch": 0.24310301189572261,
      "grad_norm": 1.4703457355499268,
      "learning_rate": 1.947277610166447e-05,
      "loss": 0.6948,
      "step": 1921
    },
    {
      "epoch": 0.24322956213616806,
      "grad_norm": 1.1906366348266602,
      "learning_rate": 1.9472201728066542e-05,
      "loss": 0.6916,
      "step": 1922
    },
    {
      "epoch": 0.2433561123766135,
      "grad_norm": 1.522653579711914,
      "learning_rate": 1.947162705024925e-05,
      "loss": 0.7009,
      "step": 1923
    },
    {
      "epoch": 0.24348266261705898,
      "grad_norm": 1.1504346132278442,
      "learning_rate": 1.947105206823105e-05,
      "loss": 0.6719,
      "step": 1924
    },
    {
      "epoch": 0.24360921285750442,
      "grad_norm": 1.5603631734848022,
      "learning_rate": 1.947047678203042e-05,
      "loss": 0.6805,
      "step": 1925
    },
    {
      "epoch": 0.2437357630979499,
      "grad_norm": 1.4896115064620972,
      "learning_rate": 1.946990119166582e-05,
      "loss": 0.7024,
      "step": 1926
    },
    {
      "epoch": 0.24386231333839534,
      "grad_norm": 1.613944172859192,
      "learning_rate": 1.946932529715575e-05,
      "loss": 0.6952,
      "step": 1927
    },
    {
      "epoch": 0.2439888635788408,
      "grad_norm": 1.260860800743103,
      "learning_rate": 1.9468749098518703e-05,
      "loss": 0.677,
      "step": 1928
    },
    {
      "epoch": 0.24411541381928625,
      "grad_norm": 1.4944961071014404,
      "learning_rate": 1.9468172595773178e-05,
      "loss": 0.6808,
      "step": 1929
    },
    {
      "epoch": 0.24424196405973173,
      "grad_norm": 1.264902949333191,
      "learning_rate": 1.9467595788937698e-05,
      "loss": 0.7093,
      "step": 1930
    },
    {
      "epoch": 0.24436851430017717,
      "grad_norm": 1.2682998180389404,
      "learning_rate": 1.9467018678030783e-05,
      "loss": 0.6218,
      "step": 1931
    },
    {
      "epoch": 0.24449506454062261,
      "grad_norm": 1.1725460290908813,
      "learning_rate": 1.9466441263070974e-05,
      "loss": 0.667,
      "step": 1932
    },
    {
      "epoch": 0.2446216147810681,
      "grad_norm": 1.4257855415344238,
      "learning_rate": 1.946586354407681e-05,
      "loss": 0.7407,
      "step": 1933
    },
    {
      "epoch": 0.24474816502151353,
      "grad_norm": 1.2309156656265259,
      "learning_rate": 1.9465285521066846e-05,
      "loss": 0.7175,
      "step": 1934
    },
    {
      "epoch": 0.244874715261959,
      "grad_norm": 1.6187973022460938,
      "learning_rate": 1.9464707194059655e-05,
      "loss": 0.686,
      "step": 1935
    },
    {
      "epoch": 0.24500126550240445,
      "grad_norm": 1.309920310974121,
      "learning_rate": 1.9464128563073804e-05,
      "loss": 0.7071,
      "step": 1936
    },
    {
      "epoch": 0.24512781574284992,
      "grad_norm": 1.5028973817825317,
      "learning_rate": 1.9463549628127875e-05,
      "loss": 0.6561,
      "step": 1937
    },
    {
      "epoch": 0.24525436598329536,
      "grad_norm": 1.3248621225357056,
      "learning_rate": 1.9462970389240467e-05,
      "loss": 0.7134,
      "step": 1938
    },
    {
      "epoch": 0.24538091622374084,
      "grad_norm": 1.25055992603302,
      "learning_rate": 1.9462390846430176e-05,
      "loss": 0.6946,
      "step": 1939
    },
    {
      "epoch": 0.24550746646418628,
      "grad_norm": 1.1234986782073975,
      "learning_rate": 1.9461810999715625e-05,
      "loss": 0.6579,
      "step": 1940
    },
    {
      "epoch": 0.24563401670463175,
      "grad_norm": 1.3533774614334106,
      "learning_rate": 1.946123084911543e-05,
      "loss": 0.6746,
      "step": 1941
    },
    {
      "epoch": 0.2457605669450772,
      "grad_norm": 1.2318578958511353,
      "learning_rate": 1.9460650394648227e-05,
      "loss": 0.7297,
      "step": 1942
    },
    {
      "epoch": 0.24588711718552264,
      "grad_norm": 1.4808573722839355,
      "learning_rate": 1.946006963633266e-05,
      "loss": 0.6685,
      "step": 1943
    },
    {
      "epoch": 0.2460136674259681,
      "grad_norm": 1.1492873430252075,
      "learning_rate": 1.945948857418737e-05,
      "loss": 0.683,
      "step": 1944
    },
    {
      "epoch": 0.24614021766641356,
      "grad_norm": 1.334990382194519,
      "learning_rate": 1.9458907208231036e-05,
      "loss": 0.7291,
      "step": 1945
    },
    {
      "epoch": 0.24626676790685903,
      "grad_norm": 1.0972230434417725,
      "learning_rate": 1.945832553848232e-05,
      "loss": 0.649,
      "step": 1946
    },
    {
      "epoch": 0.24639331814730447,
      "grad_norm": 1.204565405845642,
      "learning_rate": 1.9457743564959903e-05,
      "loss": 0.7,
      "step": 1947
    },
    {
      "epoch": 0.24651986838774995,
      "grad_norm": 1.1605517864227295,
      "learning_rate": 1.9457161287682476e-05,
      "loss": 0.7292,
      "step": 1948
    },
    {
      "epoch": 0.2466464186281954,
      "grad_norm": 1.200202226638794,
      "learning_rate": 1.945657870666875e-05,
      "loss": 0.6985,
      "step": 1949
    },
    {
      "epoch": 0.24677296886864086,
      "grad_norm": 1.1840875148773193,
      "learning_rate": 1.9455995821937422e-05,
      "loss": 0.662,
      "step": 1950
    },
    {
      "epoch": 0.2468995191090863,
      "grad_norm": 1.2063456773757935,
      "learning_rate": 1.9455412633507217e-05,
      "loss": 0.6686,
      "step": 1951
    },
    {
      "epoch": 0.24702606934953175,
      "grad_norm": 1.290706992149353,
      "learning_rate": 1.9454829141396875e-05,
      "loss": 0.6967,
      "step": 1952
    },
    {
      "epoch": 0.24715261958997722,
      "grad_norm": 1.2153680324554443,
      "learning_rate": 1.945424534562512e-05,
      "loss": 0.671,
      "step": 1953
    },
    {
      "epoch": 0.24727916983042267,
      "grad_norm": 1.3990291357040405,
      "learning_rate": 1.9453661246210712e-05,
      "loss": 0.6743,
      "step": 1954
    },
    {
      "epoch": 0.24740572007086814,
      "grad_norm": 1.2162106037139893,
      "learning_rate": 1.9453076843172408e-05,
      "loss": 0.6605,
      "step": 1955
    },
    {
      "epoch": 0.24753227031131358,
      "grad_norm": 1.5196470022201538,
      "learning_rate": 1.945249213652898e-05,
      "loss": 0.6415,
      "step": 1956
    },
    {
      "epoch": 0.24765882055175906,
      "grad_norm": 1.2611533403396606,
      "learning_rate": 1.9451907126299204e-05,
      "loss": 0.6627,
      "step": 1957
    },
    {
      "epoch": 0.2477853707922045,
      "grad_norm": 1.7409261465072632,
      "learning_rate": 1.945132181250187e-05,
      "loss": 0.7044,
      "step": 1958
    },
    {
      "epoch": 0.24791192103264997,
      "grad_norm": 1.2183095216751099,
      "learning_rate": 1.9450736195155776e-05,
      "loss": 0.6386,
      "step": 1959
    },
    {
      "epoch": 0.24803847127309542,
      "grad_norm": 1.5059089660644531,
      "learning_rate": 1.9450150274279726e-05,
      "loss": 0.6496,
      "step": 1960
    },
    {
      "epoch": 0.24816502151354086,
      "grad_norm": 3.0920071601867676,
      "learning_rate": 1.9449564049892547e-05,
      "loss": 0.7467,
      "step": 1961
    },
    {
      "epoch": 0.24829157175398633,
      "grad_norm": 1.2610329389572144,
      "learning_rate": 1.9448977522013063e-05,
      "loss": 0.6912,
      "step": 1962
    },
    {
      "epoch": 0.24841812199443178,
      "grad_norm": 1.102611780166626,
      "learning_rate": 1.944839069066011e-05,
      "loss": 0.6175,
      "step": 1963
    },
    {
      "epoch": 0.24854467223487725,
      "grad_norm": 1.2263998985290527,
      "learning_rate": 1.9447803555852533e-05,
      "loss": 0.7001,
      "step": 1964
    },
    {
      "epoch": 0.2486712224753227,
      "grad_norm": 1.11858069896698,
      "learning_rate": 1.9447216117609193e-05,
      "loss": 0.6715,
      "step": 1965
    },
    {
      "epoch": 0.24879777271576817,
      "grad_norm": 1.1926791667938232,
      "learning_rate": 1.9446628375948958e-05,
      "loss": 0.7058,
      "step": 1966
    },
    {
      "epoch": 0.2489243229562136,
      "grad_norm": 1.158455491065979,
      "learning_rate": 1.9446040330890703e-05,
      "loss": 0.622,
      "step": 1967
    },
    {
      "epoch": 0.24905087319665908,
      "grad_norm": 1.2271329164505005,
      "learning_rate": 1.9445451982453313e-05,
      "loss": 0.6849,
      "step": 1968
    },
    {
      "epoch": 0.24917742343710453,
      "grad_norm": 1.2266674041748047,
      "learning_rate": 1.9444863330655685e-05,
      "loss": 0.6674,
      "step": 1969
    },
    {
      "epoch": 0.24930397367755,
      "grad_norm": 1.1806366443634033,
      "learning_rate": 1.9444274375516724e-05,
      "loss": 0.6847,
      "step": 1970
    },
    {
      "epoch": 0.24943052391799544,
      "grad_norm": 1.2973955869674683,
      "learning_rate": 1.944368511705535e-05,
      "loss": 0.6696,
      "step": 1971
    },
    {
      "epoch": 0.2495570741584409,
      "grad_norm": 1.1577467918395996,
      "learning_rate": 1.944309555529048e-05,
      "loss": 0.6563,
      "step": 1972
    },
    {
      "epoch": 0.24968362439888636,
      "grad_norm": 1.4302666187286377,
      "learning_rate": 1.9442505690241053e-05,
      "loss": 0.6563,
      "step": 1973
    },
    {
      "epoch": 0.2498101746393318,
      "grad_norm": 1.232669711112976,
      "learning_rate": 1.9441915521926017e-05,
      "loss": 0.718,
      "step": 1974
    },
    {
      "epoch": 0.24993672487977728,
      "grad_norm": 1.2343558073043823,
      "learning_rate": 1.944132505036432e-05,
      "loss": 0.6795,
      "step": 1975
    },
    {
      "epoch": 0.25006327512022275,
      "grad_norm": 1.2770065069198608,
      "learning_rate": 1.944073427557493e-05,
      "loss": 0.7053,
      "step": 1976
    },
    {
      "epoch": 0.2501898253606682,
      "grad_norm": 1.1646666526794434,
      "learning_rate": 1.9440143197576823e-05,
      "loss": 0.6345,
      "step": 1977
    },
    {
      "epoch": 0.25031637560111364,
      "grad_norm": 1.193252682685852,
      "learning_rate": 1.943955181638898e-05,
      "loss": 0.7103,
      "step": 1978
    },
    {
      "epoch": 0.2504429258415591,
      "grad_norm": 1.1943954229354858,
      "learning_rate": 1.9438960132030395e-05,
      "loss": 0.7163,
      "step": 1979
    },
    {
      "epoch": 0.2505694760820046,
      "grad_norm": 1.0859686136245728,
      "learning_rate": 1.9438368144520073e-05,
      "loss": 0.6564,
      "step": 1980
    },
    {
      "epoch": 0.25069602632245,
      "grad_norm": 1.1679168939590454,
      "learning_rate": 1.9437775853877025e-05,
      "loss": 0.7181,
      "step": 1981
    },
    {
      "epoch": 0.25082257656289547,
      "grad_norm": 1.2882598638534546,
      "learning_rate": 1.943718326012027e-05,
      "loss": 0.6618,
      "step": 1982
    },
    {
      "epoch": 0.2509491268033409,
      "grad_norm": 1.2266868352890015,
      "learning_rate": 1.9436590363268845e-05,
      "loss": 0.7008,
      "step": 1983
    },
    {
      "epoch": 0.25107567704378636,
      "grad_norm": 1.1542975902557373,
      "learning_rate": 1.943599716334179e-05,
      "loss": 0.6797,
      "step": 1984
    },
    {
      "epoch": 0.25120222728423186,
      "grad_norm": 1.2518969774246216,
      "learning_rate": 1.943540366035816e-05,
      "loss": 0.6637,
      "step": 1985
    },
    {
      "epoch": 0.2513287775246773,
      "grad_norm": 1.1469261646270752,
      "learning_rate": 1.9434809854337014e-05,
      "loss": 0.6924,
      "step": 1986
    },
    {
      "epoch": 0.25145532776512275,
      "grad_norm": 1.1725984811782837,
      "learning_rate": 1.9434215745297425e-05,
      "loss": 0.6821,
      "step": 1987
    },
    {
      "epoch": 0.2515818780055682,
      "grad_norm": 1.1158244609832764,
      "learning_rate": 1.943362133325847e-05,
      "loss": 0.6529,
      "step": 1988
    },
    {
      "epoch": 0.2517084282460137,
      "grad_norm": 1.173640251159668,
      "learning_rate": 1.9433026618239244e-05,
      "loss": 0.694,
      "step": 1989
    },
    {
      "epoch": 0.25183497848645914,
      "grad_norm": 1.1324524879455566,
      "learning_rate": 1.9432431600258848e-05,
      "loss": 0.6421,
      "step": 1990
    },
    {
      "epoch": 0.2519615287269046,
      "grad_norm": 1.4761695861816406,
      "learning_rate": 1.943183627933639e-05,
      "loss": 0.6602,
      "step": 1991
    },
    {
      "epoch": 0.25208807896735,
      "grad_norm": 1.5011529922485352,
      "learning_rate": 1.943124065549099e-05,
      "loss": 0.8841,
      "step": 1992
    },
    {
      "epoch": 0.25221462920779547,
      "grad_norm": 1.4668446779251099,
      "learning_rate": 1.9430644728741774e-05,
      "loss": 0.727,
      "step": 1993
    },
    {
      "epoch": 0.25234117944824097,
      "grad_norm": 1.1739670038223267,
      "learning_rate": 1.9430048499107893e-05,
      "loss": 0.6536,
      "step": 1994
    },
    {
      "epoch": 0.2524677296886864,
      "grad_norm": 1.2502620220184326,
      "learning_rate": 1.942945196660848e-05,
      "loss": 0.6442,
      "step": 1995
    },
    {
      "epoch": 0.25259427992913186,
      "grad_norm": 1.152512788772583,
      "learning_rate": 1.942885513126271e-05,
      "loss": 0.6744,
      "step": 1996
    },
    {
      "epoch": 0.2527208301695773,
      "grad_norm": 1.4219805002212524,
      "learning_rate": 1.942825799308974e-05,
      "loss": 0.6751,
      "step": 1997
    },
    {
      "epoch": 0.2528473804100228,
      "grad_norm": 1.151161551475525,
      "learning_rate": 1.9427660552108754e-05,
      "loss": 0.6803,
      "step": 1998
    },
    {
      "epoch": 0.25297393065046825,
      "grad_norm": 1.1760233640670776,
      "learning_rate": 1.9427062808338936e-05,
      "loss": 0.7034,
      "step": 1999
    },
    {
      "epoch": 0.2531004808909137,
      "grad_norm": 1.3354183435440063,
      "learning_rate": 1.9426464761799487e-05,
      "loss": 0.6632,
      "step": 2000
    },
    {
      "epoch": 0.25322703113135914,
      "grad_norm": 1.162401556968689,
      "learning_rate": 1.9425866412509616e-05,
      "loss": 0.6485,
      "step": 2001
    },
    {
      "epoch": 0.2533535813718046,
      "grad_norm": 1.3028697967529297,
      "learning_rate": 1.9425267760488537e-05,
      "loss": 0.7598,
      "step": 2002
    },
    {
      "epoch": 0.2534801316122501,
      "grad_norm": 1.2078185081481934,
      "learning_rate": 1.9424668805755476e-05,
      "loss": 0.7073,
      "step": 2003
    },
    {
      "epoch": 0.2536066818526955,
      "grad_norm": 1.1807013750076294,
      "learning_rate": 1.9424069548329676e-05,
      "loss": 0.7178,
      "step": 2004
    },
    {
      "epoch": 0.25373323209314097,
      "grad_norm": 1.1452748775482178,
      "learning_rate": 1.9423469988230375e-05,
      "loss": 0.6431,
      "step": 2005
    },
    {
      "epoch": 0.2538597823335864,
      "grad_norm": 1.1303726434707642,
      "learning_rate": 1.9422870125476833e-05,
      "loss": 0.7394,
      "step": 2006
    },
    {
      "epoch": 0.2539863325740319,
      "grad_norm": 1.1141248941421509,
      "learning_rate": 1.942226996008832e-05,
      "loss": 0.6159,
      "step": 2007
    },
    {
      "epoch": 0.25411288281447736,
      "grad_norm": 1.2776416540145874,
      "learning_rate": 1.9421669492084105e-05,
      "loss": 0.7012,
      "step": 2008
    },
    {
      "epoch": 0.2542394330549228,
      "grad_norm": 1.148169994354248,
      "learning_rate": 1.9421068721483473e-05,
      "loss": 0.6644,
      "step": 2009
    },
    {
      "epoch": 0.25436598329536825,
      "grad_norm": 1.1616098880767822,
      "learning_rate": 1.9420467648305724e-05,
      "loss": 0.6508,
      "step": 2010
    },
    {
      "epoch": 0.2544925335358137,
      "grad_norm": 1.2397229671478271,
      "learning_rate": 1.941986627257016e-05,
      "loss": 0.6878,
      "step": 2011
    },
    {
      "epoch": 0.2546190837762592,
      "grad_norm": 1.3194842338562012,
      "learning_rate": 1.94192645942961e-05,
      "loss": 0.667,
      "step": 2012
    },
    {
      "epoch": 0.25474563401670464,
      "grad_norm": 1.1656527519226074,
      "learning_rate": 1.941866261350286e-05,
      "loss": 0.6605,
      "step": 2013
    },
    {
      "epoch": 0.2548721842571501,
      "grad_norm": 1.1911921501159668,
      "learning_rate": 1.9418060330209777e-05,
      "loss": 0.663,
      "step": 2014
    },
    {
      "epoch": 0.2549987344975955,
      "grad_norm": 1.2629910707473755,
      "learning_rate": 1.9417457744436197e-05,
      "loss": 0.682,
      "step": 2015
    },
    {
      "epoch": 0.255125284738041,
      "grad_norm": 1.1896568536758423,
      "learning_rate": 1.941685485620147e-05,
      "loss": 0.7168,
      "step": 2016
    },
    {
      "epoch": 0.25525183497848647,
      "grad_norm": 1.2289396524429321,
      "learning_rate": 1.9416251665524966e-05,
      "loss": 0.6311,
      "step": 2017
    },
    {
      "epoch": 0.2553783852189319,
      "grad_norm": 1.2118463516235352,
      "learning_rate": 1.9415648172426047e-05,
      "loss": 0.6885,
      "step": 2018
    },
    {
      "epoch": 0.25550493545937736,
      "grad_norm": 1.2185109853744507,
      "learning_rate": 1.94150443769241e-05,
      "loss": 0.701,
      "step": 2019
    },
    {
      "epoch": 0.2556314856998228,
      "grad_norm": 1.3058035373687744,
      "learning_rate": 1.9414440279038526e-05,
      "loss": 0.6742,
      "step": 2020
    },
    {
      "epoch": 0.2557580359402683,
      "grad_norm": 1.2528880834579468,
      "learning_rate": 1.941383587878871e-05,
      "loss": 0.7197,
      "step": 2021
    },
    {
      "epoch": 0.25588458618071375,
      "grad_norm": 1.4886484146118164,
      "learning_rate": 1.941323117619408e-05,
      "loss": 0.677,
      "step": 2022
    },
    {
      "epoch": 0.2560111364211592,
      "grad_norm": 1.1693609952926636,
      "learning_rate": 1.9412626171274042e-05,
      "loss": 0.6381,
      "step": 2023
    },
    {
      "epoch": 0.25613768666160464,
      "grad_norm": 1.353184700012207,
      "learning_rate": 1.941202086404804e-05,
      "loss": 0.7005,
      "step": 2024
    },
    {
      "epoch": 0.25626423690205014,
      "grad_norm": 1.2072772979736328,
      "learning_rate": 1.9411415254535508e-05,
      "loss": 0.7305,
      "step": 2025
    },
    {
      "epoch": 0.2563907871424956,
      "grad_norm": 1.2374389171600342,
      "learning_rate": 1.94108093427559e-05,
      "loss": 0.6653,
      "step": 2026
    },
    {
      "epoch": 0.256517337382941,
      "grad_norm": 2.068427085876465,
      "learning_rate": 1.941020312872867e-05,
      "loss": 0.6988,
      "step": 2027
    },
    {
      "epoch": 0.25664388762338647,
      "grad_norm": 1.3928765058517456,
      "learning_rate": 1.9409596612473295e-05,
      "loss": 0.6411,
      "step": 2028
    },
    {
      "epoch": 0.25677043786383197,
      "grad_norm": 1.3619613647460938,
      "learning_rate": 1.9408989794009247e-05,
      "loss": 0.72,
      "step": 2029
    },
    {
      "epoch": 0.2568969881042774,
      "grad_norm": 1.207789659500122,
      "learning_rate": 1.940838267335602e-05,
      "loss": 0.692,
      "step": 2030
    },
    {
      "epoch": 0.25702353834472286,
      "grad_norm": 1.4588181972503662,
      "learning_rate": 1.9407775250533117e-05,
      "loss": 0.7532,
      "step": 2031
    },
    {
      "epoch": 0.2571500885851683,
      "grad_norm": 1.3076136112213135,
      "learning_rate": 1.940716752556004e-05,
      "loss": 0.6589,
      "step": 2032
    },
    {
      "epoch": 0.25727663882561375,
      "grad_norm": 1.3680468797683716,
      "learning_rate": 1.9406559498456307e-05,
      "loss": 0.6767,
      "step": 2033
    },
    {
      "epoch": 0.25740318906605925,
      "grad_norm": 4.177816390991211,
      "learning_rate": 1.940595116924145e-05,
      "loss": 0.7014,
      "step": 2034
    },
    {
      "epoch": 0.2575297393065047,
      "grad_norm": 1.5397911071777344,
      "learning_rate": 1.940534253793501e-05,
      "loss": 0.6963,
      "step": 2035
    },
    {
      "epoch": 0.25765628954695013,
      "grad_norm": 1.3052754402160645,
      "learning_rate": 1.9404733604556526e-05,
      "loss": 0.7194,
      "step": 2036
    },
    {
      "epoch": 0.2577828397873956,
      "grad_norm": 1.3174420595169067,
      "learning_rate": 1.9404124369125556e-05,
      "loss": 0.6959,
      "step": 2037
    },
    {
      "epoch": 0.2579093900278411,
      "grad_norm": 1.1546287536621094,
      "learning_rate": 1.940351483166167e-05,
      "loss": 0.6929,
      "step": 2038
    },
    {
      "epoch": 0.2580359402682865,
      "grad_norm": 1.305334210395813,
      "learning_rate": 1.9402904992184448e-05,
      "loss": 0.7127,
      "step": 2039
    },
    {
      "epoch": 0.25816249050873197,
      "grad_norm": 1.3078056573867798,
      "learning_rate": 1.940229485071347e-05,
      "loss": 0.6707,
      "step": 2040
    },
    {
      "epoch": 0.2582890407491774,
      "grad_norm": 1.2039052248001099,
      "learning_rate": 1.9401684407268336e-05,
      "loss": 0.6424,
      "step": 2041
    },
    {
      "epoch": 0.25841559098962286,
      "grad_norm": 1.4154140949249268,
      "learning_rate": 1.9401073661868647e-05,
      "loss": 0.7282,
      "step": 2042
    },
    {
      "epoch": 0.25854214123006836,
      "grad_norm": 1.1612393856048584,
      "learning_rate": 1.9400462614534026e-05,
      "loss": 0.6946,
      "step": 2043
    },
    {
      "epoch": 0.2586686914705138,
      "grad_norm": 1.4600396156311035,
      "learning_rate": 1.939985126528409e-05,
      "loss": 0.7173,
      "step": 2044
    },
    {
      "epoch": 0.25879524171095925,
      "grad_norm": 1.2670230865478516,
      "learning_rate": 1.939923961413848e-05,
      "loss": 0.6623,
      "step": 2045
    },
    {
      "epoch": 0.2589217919514047,
      "grad_norm": 1.3010796308517456,
      "learning_rate": 1.9398627661116837e-05,
      "loss": 0.6511,
      "step": 2046
    },
    {
      "epoch": 0.2590483421918502,
      "grad_norm": 1.162835717201233,
      "learning_rate": 1.9398015406238815e-05,
      "loss": 0.6489,
      "step": 2047
    },
    {
      "epoch": 0.25917489243229563,
      "grad_norm": 1.2870159149169922,
      "learning_rate": 1.9397402849524075e-05,
      "loss": 0.6898,
      "step": 2048
    },
    {
      "epoch": 0.2593014426727411,
      "grad_norm": 1.1825729608535767,
      "learning_rate": 1.93967899909923e-05,
      "loss": 0.6743,
      "step": 2049
    },
    {
      "epoch": 0.2594279929131865,
      "grad_norm": 1.347347378730774,
      "learning_rate": 1.9396176830663165e-05,
      "loss": 0.6728,
      "step": 2050
    },
    {
      "epoch": 0.25955454315363197,
      "grad_norm": 1.2052510976791382,
      "learning_rate": 1.9395563368556365e-05,
      "loss": 0.6574,
      "step": 2051
    },
    {
      "epoch": 0.25968109339407747,
      "grad_norm": 1.3357588052749634,
      "learning_rate": 1.9394949604691604e-05,
      "loss": 0.7165,
      "step": 2052
    },
    {
      "epoch": 0.2598076436345229,
      "grad_norm": 1.100272297859192,
      "learning_rate": 1.9394335539088594e-05,
      "loss": 0.7084,
      "step": 2053
    },
    {
      "epoch": 0.25993419387496836,
      "grad_norm": 1.4561413526535034,
      "learning_rate": 1.9393721171767055e-05,
      "loss": 0.6832,
      "step": 2054
    },
    {
      "epoch": 0.2600607441154138,
      "grad_norm": 1.09897780418396,
      "learning_rate": 1.9393106502746718e-05,
      "loss": 0.6773,
      "step": 2055
    },
    {
      "epoch": 0.2601872943558593,
      "grad_norm": 1.4931520223617554,
      "learning_rate": 1.9392491532047327e-05,
      "loss": 0.6409,
      "step": 2056
    },
    {
      "epoch": 0.26031384459630474,
      "grad_norm": 1.247262954711914,
      "learning_rate": 1.9391876259688637e-05,
      "loss": 0.7089,
      "step": 2057
    },
    {
      "epoch": 0.2604403948367502,
      "grad_norm": 1.5674159526824951,
      "learning_rate": 1.93912606856904e-05,
      "loss": 0.6778,
      "step": 2058
    },
    {
      "epoch": 0.26056694507719563,
      "grad_norm": 1.2154968976974487,
      "learning_rate": 1.939064481007239e-05,
      "loss": 0.7487,
      "step": 2059
    },
    {
      "epoch": 0.2606934953176411,
      "grad_norm": 1.423398494720459,
      "learning_rate": 1.9390028632854387e-05,
      "loss": 0.7119,
      "step": 2060
    },
    {
      "epoch": 0.2608200455580866,
      "grad_norm": 1.1241225004196167,
      "learning_rate": 1.9389412154056188e-05,
      "loss": 0.6989,
      "step": 2061
    },
    {
      "epoch": 0.260946595798532,
      "grad_norm": 1.4133213758468628,
      "learning_rate": 1.9388795373697576e-05,
      "loss": 0.6902,
      "step": 2062
    },
    {
      "epoch": 0.26107314603897747,
      "grad_norm": 1.1340422630310059,
      "learning_rate": 1.9388178291798376e-05,
      "loss": 0.6647,
      "step": 2063
    },
    {
      "epoch": 0.2611996962794229,
      "grad_norm": 1.214118242263794,
      "learning_rate": 1.93875609083784e-05,
      "loss": 0.7076,
      "step": 2064
    },
    {
      "epoch": 0.2613262465198684,
      "grad_norm": 1.2092719078063965,
      "learning_rate": 1.938694322345748e-05,
      "loss": 0.7144,
      "step": 2065
    },
    {
      "epoch": 0.26145279676031385,
      "grad_norm": 1.2199033498764038,
      "learning_rate": 1.9386325237055448e-05,
      "loss": 0.686,
      "step": 2066
    },
    {
      "epoch": 0.2615793470007593,
      "grad_norm": 1.1443315744400024,
      "learning_rate": 1.9385706949192157e-05,
      "loss": 0.6884,
      "step": 2067
    },
    {
      "epoch": 0.26170589724120474,
      "grad_norm": 1.172389268875122,
      "learning_rate": 1.9385088359887463e-05,
      "loss": 0.7749,
      "step": 2068
    },
    {
      "epoch": 0.2618324474816502,
      "grad_norm": 1.299575924873352,
      "learning_rate": 1.9384469469161235e-05,
      "loss": 0.7012,
      "step": 2069
    },
    {
      "epoch": 0.2619589977220957,
      "grad_norm": 1.229143500328064,
      "learning_rate": 1.9383850277033346e-05,
      "loss": 0.6847,
      "step": 2070
    },
    {
      "epoch": 0.26208554796254113,
      "grad_norm": 1.2026817798614502,
      "learning_rate": 1.9383230783523687e-05,
      "loss": 0.6613,
      "step": 2071
    },
    {
      "epoch": 0.2622120982029866,
      "grad_norm": 1.204367995262146,
      "learning_rate": 1.938261098865215e-05,
      "loss": 0.6658,
      "step": 2072
    },
    {
      "epoch": 0.262338648443432,
      "grad_norm": 1.1597437858581543,
      "learning_rate": 1.9381990892438648e-05,
      "loss": 0.6719,
      "step": 2073
    },
    {
      "epoch": 0.2624651986838775,
      "grad_norm": 1.085219144821167,
      "learning_rate": 1.9381370494903087e-05,
      "loss": 0.657,
      "step": 2074
    },
    {
      "epoch": 0.26259174892432297,
      "grad_norm": 1.1863634586334229,
      "learning_rate": 1.93807497960654e-05,
      "loss": 0.6297,
      "step": 2075
    },
    {
      "epoch": 0.2627182991647684,
      "grad_norm": 1.0968337059020996,
      "learning_rate": 1.938012879594552e-05,
      "loss": 0.6798,
      "step": 2076
    },
    {
      "epoch": 0.26284484940521385,
      "grad_norm": 1.1581826210021973,
      "learning_rate": 1.937950749456339e-05,
      "loss": 0.6907,
      "step": 2077
    },
    {
      "epoch": 0.26297139964565935,
      "grad_norm": 1.2843053340911865,
      "learning_rate": 1.9378885891938965e-05,
      "loss": 0.7496,
      "step": 2078
    },
    {
      "epoch": 0.2630979498861048,
      "grad_norm": 1.0795516967773438,
      "learning_rate": 1.9378263988092213e-05,
      "loss": 0.6437,
      "step": 2079
    },
    {
      "epoch": 0.26322450012655024,
      "grad_norm": 1.2919865846633911,
      "learning_rate": 1.93776417830431e-05,
      "loss": 0.6883,
      "step": 2080
    },
    {
      "epoch": 0.2633510503669957,
      "grad_norm": 2.2591195106506348,
      "learning_rate": 1.9377019276811614e-05,
      "loss": 0.7085,
      "step": 2081
    },
    {
      "epoch": 0.26347760060744113,
      "grad_norm": 1.6106594800949097,
      "learning_rate": 1.937639646941775e-05,
      "loss": 0.6791,
      "step": 2082
    },
    {
      "epoch": 0.26360415084788663,
      "grad_norm": 1.1482434272766113,
      "learning_rate": 1.937577336088151e-05,
      "loss": 0.6183,
      "step": 2083
    },
    {
      "epoch": 0.2637307010883321,
      "grad_norm": 1.1470928192138672,
      "learning_rate": 1.9375149951222903e-05,
      "loss": 0.651,
      "step": 2084
    },
    {
      "epoch": 0.2638572513287775,
      "grad_norm": 1.284377098083496,
      "learning_rate": 1.9374526240461954e-05,
      "loss": 0.7138,
      "step": 2085
    },
    {
      "epoch": 0.26398380156922296,
      "grad_norm": 1.1764346361160278,
      "learning_rate": 1.937390222861869e-05,
      "loss": 0.693,
      "step": 2086
    },
    {
      "epoch": 0.26411035180966846,
      "grad_norm": 1.349522590637207,
      "learning_rate": 1.937327791571316e-05,
      "loss": 0.7211,
      "step": 2087
    },
    {
      "epoch": 0.2642369020501139,
      "grad_norm": 1.2037619352340698,
      "learning_rate": 1.937265330176541e-05,
      "loss": 0.6843,
      "step": 2088
    },
    {
      "epoch": 0.26436345229055935,
      "grad_norm": 1.1878496408462524,
      "learning_rate": 1.9372028386795505e-05,
      "loss": 0.6777,
      "step": 2089
    },
    {
      "epoch": 0.2644900025310048,
      "grad_norm": 1.1540637016296387,
      "learning_rate": 1.937140317082351e-05,
      "loss": 0.6853,
      "step": 2090
    },
    {
      "epoch": 0.26461655277145024,
      "grad_norm": 1.1316492557525635,
      "learning_rate": 1.9370777653869508e-05,
      "loss": 0.6697,
      "step": 2091
    },
    {
      "epoch": 0.26474310301189574,
      "grad_norm": 1.2886861562728882,
      "learning_rate": 1.937015183595359e-05,
      "loss": 0.7086,
      "step": 2092
    },
    {
      "epoch": 0.2648696532523412,
      "grad_norm": 1.2235668897628784,
      "learning_rate": 1.936952571709585e-05,
      "loss": 0.6515,
      "step": 2093
    },
    {
      "epoch": 0.26499620349278663,
      "grad_norm": 1.2737947702407837,
      "learning_rate": 1.9368899297316402e-05,
      "loss": 0.6595,
      "step": 2094
    },
    {
      "epoch": 0.2651227537332321,
      "grad_norm": 1.3966999053955078,
      "learning_rate": 1.9368272576635364e-05,
      "loss": 0.7437,
      "step": 2095
    },
    {
      "epoch": 0.2652493039736776,
      "grad_norm": 1.1850228309631348,
      "learning_rate": 1.9367645555072868e-05,
      "loss": 0.6916,
      "step": 2096
    },
    {
      "epoch": 0.265375854214123,
      "grad_norm": 1.3056720495224,
      "learning_rate": 1.9367018232649043e-05,
      "loss": 0.701,
      "step": 2097
    },
    {
      "epoch": 0.26550240445456846,
      "grad_norm": 1.2085974216461182,
      "learning_rate": 1.9366390609384048e-05,
      "loss": 0.7143,
      "step": 2098
    },
    {
      "epoch": 0.2656289546950139,
      "grad_norm": 1.1650186777114868,
      "learning_rate": 1.936576268529803e-05,
      "loss": 0.706,
      "step": 2099
    },
    {
      "epoch": 0.26575550493545935,
      "grad_norm": 1.1991664171218872,
      "learning_rate": 1.936513446041116e-05,
      "loss": 0.6679,
      "step": 2100
    },
    {
      "epoch": 0.26588205517590485,
      "grad_norm": 1.1791162490844727,
      "learning_rate": 1.9364505934743617e-05,
      "loss": 0.6681,
      "step": 2101
    },
    {
      "epoch": 0.2660086054163503,
      "grad_norm": 1.1412855386734009,
      "learning_rate": 1.9363877108315585e-05,
      "loss": 0.6795,
      "step": 2102
    },
    {
      "epoch": 0.26613515565679574,
      "grad_norm": 1.2970107793807983,
      "learning_rate": 1.9363247981147265e-05,
      "loss": 0.6871,
      "step": 2103
    },
    {
      "epoch": 0.2662617058972412,
      "grad_norm": 1.2306640148162842,
      "learning_rate": 1.9362618553258854e-05,
      "loss": 0.701,
      "step": 2104
    },
    {
      "epoch": 0.2663882561376867,
      "grad_norm": 1.1683470010757446,
      "learning_rate": 1.936198882467057e-05,
      "loss": 0.725,
      "step": 2105
    },
    {
      "epoch": 0.26651480637813213,
      "grad_norm": 1.1902347803115845,
      "learning_rate": 1.9361358795402646e-05,
      "loss": 0.6426,
      "step": 2106
    },
    {
      "epoch": 0.2666413566185776,
      "grad_norm": 1.1586556434631348,
      "learning_rate": 1.9360728465475303e-05,
      "loss": 0.666,
      "step": 2107
    },
    {
      "epoch": 0.266767906859023,
      "grad_norm": 1.1510807275772095,
      "learning_rate": 1.93600978349088e-05,
      "loss": 0.6331,
      "step": 2108
    },
    {
      "epoch": 0.26689445709946846,
      "grad_norm": 1.1458871364593506,
      "learning_rate": 1.9359466903723382e-05,
      "loss": 0.7192,
      "step": 2109
    },
    {
      "epoch": 0.26702100733991396,
      "grad_norm": 1.1664769649505615,
      "learning_rate": 1.935883567193931e-05,
      "loss": 0.6746,
      "step": 2110
    },
    {
      "epoch": 0.2671475575803594,
      "grad_norm": 1.1976444721221924,
      "learning_rate": 1.9358204139576865e-05,
      "loss": 0.6914,
      "step": 2111
    },
    {
      "epoch": 0.26727410782080485,
      "grad_norm": 1.149725317955017,
      "learning_rate": 1.9357572306656328e-05,
      "loss": 0.6508,
      "step": 2112
    },
    {
      "epoch": 0.2674006580612503,
      "grad_norm": 1.239849328994751,
      "learning_rate": 1.9356940173197994e-05,
      "loss": 0.6505,
      "step": 2113
    },
    {
      "epoch": 0.2675272083016958,
      "grad_norm": 1.1391817331314087,
      "learning_rate": 1.9356307739222155e-05,
      "loss": 0.6707,
      "step": 2114
    },
    {
      "epoch": 0.26765375854214124,
      "grad_norm": 1.1560676097869873,
      "learning_rate": 1.9355675004749134e-05,
      "loss": 0.6814,
      "step": 2115
    },
    {
      "epoch": 0.2677803087825867,
      "grad_norm": 1.134958267211914,
      "learning_rate": 1.9355041969799245e-05,
      "loss": 0.6938,
      "step": 2116
    },
    {
      "epoch": 0.26790685902303213,
      "grad_norm": 1.1510262489318848,
      "learning_rate": 1.9354408634392825e-05,
      "loss": 0.6689,
      "step": 2117
    },
    {
      "epoch": 0.2680334092634776,
      "grad_norm": 1.1023435592651367,
      "learning_rate": 1.935377499855021e-05,
      "loss": 0.6912,
      "step": 2118
    },
    {
      "epoch": 0.2681599595039231,
      "grad_norm": 1.1429834365844727,
      "learning_rate": 1.9353141062291757e-05,
      "loss": 0.6975,
      "step": 2119
    },
    {
      "epoch": 0.2682865097443685,
      "grad_norm": 1.1274863481521606,
      "learning_rate": 1.935250682563782e-05,
      "loss": 0.6659,
      "step": 2120
    },
    {
      "epoch": 0.26841305998481396,
      "grad_norm": 1.104219913482666,
      "learning_rate": 1.935187228860877e-05,
      "loss": 0.6754,
      "step": 2121
    },
    {
      "epoch": 0.2685396102252594,
      "grad_norm": 1.1674686670303345,
      "learning_rate": 1.9351237451224988e-05,
      "loss": 0.6746,
      "step": 2122
    },
    {
      "epoch": 0.2686661604657049,
      "grad_norm": 20.50112533569336,
      "learning_rate": 1.9350602313506863e-05,
      "loss": 0.6671,
      "step": 2123
    },
    {
      "epoch": 0.26879271070615035,
      "grad_norm": 1.1579077243804932,
      "learning_rate": 1.9349966875474792e-05,
      "loss": 0.6264,
      "step": 2124
    },
    {
      "epoch": 0.2689192609465958,
      "grad_norm": 1.2569726705551147,
      "learning_rate": 1.9349331137149186e-05,
      "loss": 0.6844,
      "step": 2125
    },
    {
      "epoch": 0.26904581118704124,
      "grad_norm": 1.1597986221313477,
      "learning_rate": 1.934869509855046e-05,
      "loss": 0.6373,
      "step": 2126
    },
    {
      "epoch": 0.26917236142748674,
      "grad_norm": 1.3148577213287354,
      "learning_rate": 1.9348058759699046e-05,
      "loss": 0.7147,
      "step": 2127
    },
    {
      "epoch": 0.2692989116679322,
      "grad_norm": 1.422682523727417,
      "learning_rate": 1.9347422120615378e-05,
      "loss": 0.6985,
      "step": 2128
    },
    {
      "epoch": 0.26942546190837763,
      "grad_norm": 1.2521461248397827,
      "learning_rate": 1.9346785181319903e-05,
      "loss": 0.6598,
      "step": 2129
    },
    {
      "epoch": 0.2695520121488231,
      "grad_norm": 1.1216530799865723,
      "learning_rate": 1.9346147941833074e-05,
      "loss": 0.7293,
      "step": 2130
    },
    {
      "epoch": 0.2696785623892685,
      "grad_norm": 1.14635169506073,
      "learning_rate": 1.934551040217537e-05,
      "loss": 0.6539,
      "step": 2131
    },
    {
      "epoch": 0.269805112629714,
      "grad_norm": 1.288740873336792,
      "learning_rate": 1.9344872562367253e-05,
      "loss": 0.7295,
      "step": 2132
    },
    {
      "epoch": 0.26993166287015946,
      "grad_norm": 1.8977441787719727,
      "learning_rate": 1.934423442242922e-05,
      "loss": 0.6859,
      "step": 2133
    },
    {
      "epoch": 0.2700582131106049,
      "grad_norm": 1.2748366594314575,
      "learning_rate": 1.9343595982381753e-05,
      "loss": 0.7508,
      "step": 2134
    },
    {
      "epoch": 0.27018476335105035,
      "grad_norm": 1.1268023252487183,
      "learning_rate": 1.9342957242245365e-05,
      "loss": 0.6843,
      "step": 2135
    },
    {
      "epoch": 0.27031131359149585,
      "grad_norm": 1.2781767845153809,
      "learning_rate": 1.9342318202040572e-05,
      "loss": 0.7106,
      "step": 2136
    },
    {
      "epoch": 0.2704378638319413,
      "grad_norm": 1.1970219612121582,
      "learning_rate": 1.9341678861787897e-05,
      "loss": 0.6453,
      "step": 2137
    },
    {
      "epoch": 0.27056441407238674,
      "grad_norm": 1.1933460235595703,
      "learning_rate": 1.934103922150787e-05,
      "loss": 0.7009,
      "step": 2138
    },
    {
      "epoch": 0.2706909643128322,
      "grad_norm": 1.2293802499771118,
      "learning_rate": 1.9340399281221036e-05,
      "loss": 0.7017,
      "step": 2139
    },
    {
      "epoch": 0.2708175145532776,
      "grad_norm": 1.2156130075454712,
      "learning_rate": 1.9339759040947953e-05,
      "loss": 0.695,
      "step": 2140
    },
    {
      "epoch": 0.2709440647937231,
      "grad_norm": 1.4736647605895996,
      "learning_rate": 1.9339118500709176e-05,
      "loss": 0.6677,
      "step": 2141
    },
    {
      "epoch": 0.27107061503416857,
      "grad_norm": 1.2698308229446411,
      "learning_rate": 1.9338477660525283e-05,
      "loss": 0.7227,
      "step": 2142
    },
    {
      "epoch": 0.271197165274614,
      "grad_norm": 1.5246448516845703,
      "learning_rate": 1.9337836520416848e-05,
      "loss": 0.7195,
      "step": 2143
    },
    {
      "epoch": 0.27132371551505946,
      "grad_norm": 1.135532259941101,
      "learning_rate": 1.9337195080404475e-05,
      "loss": 0.6845,
      "step": 2144
    },
    {
      "epoch": 0.27145026575550496,
      "grad_norm": 1.1550017595291138,
      "learning_rate": 1.9336553340508754e-05,
      "loss": 0.6618,
      "step": 2145
    },
    {
      "epoch": 0.2715768159959504,
      "grad_norm": 1.507252812385559,
      "learning_rate": 1.9335911300750302e-05,
      "loss": 0.7082,
      "step": 2146
    },
    {
      "epoch": 0.27170336623639585,
      "grad_norm": 1.538630723953247,
      "learning_rate": 1.9335268961149736e-05,
      "loss": 0.6394,
      "step": 2147
    },
    {
      "epoch": 0.2718299164768413,
      "grad_norm": 2.3565633296966553,
      "learning_rate": 1.9334626321727688e-05,
      "loss": 0.6806,
      "step": 2148
    },
    {
      "epoch": 0.27195646671728674,
      "grad_norm": 1.6933844089508057,
      "learning_rate": 1.9333983382504797e-05,
      "loss": 0.6921,
      "step": 2149
    },
    {
      "epoch": 0.27208301695773224,
      "grad_norm": 1.115331768989563,
      "learning_rate": 1.9333340143501716e-05,
      "loss": 0.68,
      "step": 2150
    },
    {
      "epoch": 0.2722095671981777,
      "grad_norm": 1.3804632425308228,
      "learning_rate": 1.9332696604739096e-05,
      "loss": 0.6776,
      "step": 2151
    },
    {
      "epoch": 0.2723361174386231,
      "grad_norm": 1.0968255996704102,
      "learning_rate": 1.9332052766237615e-05,
      "loss": 0.6949,
      "step": 2152
    },
    {
      "epoch": 0.27246266767906857,
      "grad_norm": 1.3586957454681396,
      "learning_rate": 1.9331408628017943e-05,
      "loss": 0.6791,
      "step": 2153
    },
    {
      "epoch": 0.27258921791951407,
      "grad_norm": 1.2968069314956665,
      "learning_rate": 1.9330764190100773e-05,
      "loss": 0.7315,
      "step": 2154
    },
    {
      "epoch": 0.2727157681599595,
      "grad_norm": 1.6576945781707764,
      "learning_rate": 1.9330119452506798e-05,
      "loss": 0.6664,
      "step": 2155
    },
    {
      "epoch": 0.27284231840040496,
      "grad_norm": 3.9400856494903564,
      "learning_rate": 1.9329474415256727e-05,
      "loss": 0.7197,
      "step": 2156
    },
    {
      "epoch": 0.2729688686408504,
      "grad_norm": 1.4246536493301392,
      "learning_rate": 1.932882907837128e-05,
      "loss": 0.7307,
      "step": 2157
    },
    {
      "epoch": 0.27309541888129585,
      "grad_norm": 1.1763046979904175,
      "learning_rate": 1.9328183441871182e-05,
      "loss": 0.6325,
      "step": 2158
    },
    {
      "epoch": 0.27322196912174135,
      "grad_norm": 1.242796540260315,
      "learning_rate": 1.9327537505777165e-05,
      "loss": 0.6703,
      "step": 2159
    },
    {
      "epoch": 0.2733485193621868,
      "grad_norm": 1.181427240371704,
      "learning_rate": 1.932689127010998e-05,
      "loss": 0.6712,
      "step": 2160
    },
    {
      "epoch": 0.27347506960263224,
      "grad_norm": 1.1430915594100952,
      "learning_rate": 1.9326244734890374e-05,
      "loss": 0.6655,
      "step": 2161
    },
    {
      "epoch": 0.2736016198430777,
      "grad_norm": 1.183580994606018,
      "learning_rate": 1.9325597900139127e-05,
      "loss": 0.6445,
      "step": 2162
    },
    {
      "epoch": 0.2737281700835232,
      "grad_norm": 1.1577163934707642,
      "learning_rate": 1.9324950765876995e-05,
      "loss": 0.7146,
      "step": 2163
    },
    {
      "epoch": 0.2738547203239686,
      "grad_norm": 1.355529546737671,
      "learning_rate": 1.9324303332124775e-05,
      "loss": 0.6987,
      "step": 2164
    },
    {
      "epoch": 0.27398127056441407,
      "grad_norm": 1.1210706233978271,
      "learning_rate": 1.9323655598903254e-05,
      "loss": 0.6957,
      "step": 2165
    },
    {
      "epoch": 0.2741078208048595,
      "grad_norm": 1.309954047203064,
      "learning_rate": 1.9323007566233238e-05,
      "loss": 0.6557,
      "step": 2166
    },
    {
      "epoch": 0.27423437104530496,
      "grad_norm": 1.2238829135894775,
      "learning_rate": 1.932235923413554e-05,
      "loss": 0.678,
      "step": 2167
    },
    {
      "epoch": 0.27436092128575046,
      "grad_norm": 1.417290210723877,
      "learning_rate": 1.9321710602630985e-05,
      "loss": 0.7334,
      "step": 2168
    },
    {
      "epoch": 0.2744874715261959,
      "grad_norm": 1.0932973623275757,
      "learning_rate": 1.93210616717404e-05,
      "loss": 0.6818,
      "step": 2169
    },
    {
      "epoch": 0.27461402176664135,
      "grad_norm": 1.215461254119873,
      "learning_rate": 1.932041244148463e-05,
      "loss": 0.6745,
      "step": 2170
    },
    {
      "epoch": 0.2747405720070868,
      "grad_norm": 1.1842972040176392,
      "learning_rate": 1.9319762911884524e-05,
      "loss": 0.6633,
      "step": 2171
    },
    {
      "epoch": 0.2748671222475323,
      "grad_norm": 1.1260366439819336,
      "learning_rate": 1.9319113082960946e-05,
      "loss": 0.6855,
      "step": 2172
    },
    {
      "epoch": 0.27499367248797774,
      "grad_norm": 1.1815721988677979,
      "learning_rate": 1.9318462954734766e-05,
      "loss": 0.6334,
      "step": 2173
    },
    {
      "epoch": 0.2751202227284232,
      "grad_norm": 1.1849300861358643,
      "learning_rate": 1.9317812527226862e-05,
      "loss": 0.6907,
      "step": 2174
    },
    {
      "epoch": 0.2752467729688686,
      "grad_norm": 1.1980412006378174,
      "learning_rate": 1.9317161800458124e-05,
      "loss": 0.6613,
      "step": 2175
    },
    {
      "epoch": 0.2753733232093141,
      "grad_norm": 1.1317112445831299,
      "learning_rate": 1.9316510774449452e-05,
      "loss": 0.6768,
      "step": 2176
    },
    {
      "epoch": 0.27549987344975957,
      "grad_norm": 1.1860630512237549,
      "learning_rate": 1.931585944922176e-05,
      "loss": 0.6902,
      "step": 2177
    },
    {
      "epoch": 0.275626423690205,
      "grad_norm": 1.1117267608642578,
      "learning_rate": 1.9315207824795957e-05,
      "loss": 0.6545,
      "step": 2178
    },
    {
      "epoch": 0.27575297393065046,
      "grad_norm": 1.2172279357910156,
      "learning_rate": 1.9314555901192976e-05,
      "loss": 0.6817,
      "step": 2179
    },
    {
      "epoch": 0.2758795241710959,
      "grad_norm": 1.2184144258499146,
      "learning_rate": 1.931390367843376e-05,
      "loss": 0.7193,
      "step": 2180
    },
    {
      "epoch": 0.2760060744115414,
      "grad_norm": 1.4010224342346191,
      "learning_rate": 1.931325115653925e-05,
      "loss": 0.6868,
      "step": 2181
    },
    {
      "epoch": 0.27613262465198685,
      "grad_norm": 1.2569243907928467,
      "learning_rate": 1.9312598335530404e-05,
      "loss": 0.7166,
      "step": 2182
    },
    {
      "epoch": 0.2762591748924323,
      "grad_norm": 1.1359059810638428,
      "learning_rate": 1.9311945215428194e-05,
      "loss": 0.6607,
      "step": 2183
    },
    {
      "epoch": 0.27638572513287774,
      "grad_norm": 1.1964271068572998,
      "learning_rate": 1.9311291796253588e-05,
      "loss": 0.6821,
      "step": 2184
    },
    {
      "epoch": 0.27651227537332324,
      "grad_norm": 1.172772765159607,
      "learning_rate": 1.9310638078027576e-05,
      "loss": 0.6988,
      "step": 2185
    },
    {
      "epoch": 0.2766388256137687,
      "grad_norm": 1.1443603038787842,
      "learning_rate": 1.9309984060771154e-05,
      "loss": 0.6889,
      "step": 2186
    },
    {
      "epoch": 0.2767653758542141,
      "grad_norm": 1.192618489265442,
      "learning_rate": 1.9309329744505326e-05,
      "loss": 0.7149,
      "step": 2187
    },
    {
      "epoch": 0.27689192609465957,
      "grad_norm": 1.2422521114349365,
      "learning_rate": 1.9308675129251107e-05,
      "loss": 0.6979,
      "step": 2188
    },
    {
      "epoch": 0.277018476335105,
      "grad_norm": 1.1483198404312134,
      "learning_rate": 1.9308020215029522e-05,
      "loss": 0.6797,
      "step": 2189
    },
    {
      "epoch": 0.2771450265755505,
      "grad_norm": 1.3981598615646362,
      "learning_rate": 1.9307365001861607e-05,
      "loss": 0.6614,
      "step": 2190
    },
    {
      "epoch": 0.27727157681599596,
      "grad_norm": 1.422987937927246,
      "learning_rate": 1.93067094897684e-05,
      "loss": 0.6765,
      "step": 2191
    },
    {
      "epoch": 0.2773981270564414,
      "grad_norm": 1.2425094842910767,
      "learning_rate": 1.9306053678770956e-05,
      "loss": 0.673,
      "step": 2192
    },
    {
      "epoch": 0.27752467729688685,
      "grad_norm": 1.2945060729980469,
      "learning_rate": 1.930539756889034e-05,
      "loss": 0.6807,
      "step": 2193
    },
    {
      "epoch": 0.27765122753733235,
      "grad_norm": 1.2932056188583374,
      "learning_rate": 1.9304741160147625e-05,
      "loss": 0.6834,
      "step": 2194
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 1.1786513328552246,
      "learning_rate": 1.9304084452563893e-05,
      "loss": 0.6646,
      "step": 2195
    },
    {
      "epoch": 0.27790432801822323,
      "grad_norm": 1.2575902938842773,
      "learning_rate": 1.930342744616023e-05,
      "loss": 0.6639,
      "step": 2196
    },
    {
      "epoch": 0.2780308782586687,
      "grad_norm": 1.0688217878341675,
      "learning_rate": 1.930277014095774e-05,
      "loss": 0.6104,
      "step": 2197
    },
    {
      "epoch": 0.2781574284991141,
      "grad_norm": 1.2915323972702026,
      "learning_rate": 1.9302112536977535e-05,
      "loss": 0.673,
      "step": 2198
    },
    {
      "epoch": 0.2782839787395596,
      "grad_norm": 1.2574577331542969,
      "learning_rate": 1.9301454634240736e-05,
      "loss": 0.7172,
      "step": 2199
    },
    {
      "epoch": 0.27841052898000507,
      "grad_norm": 1.1708816289901733,
      "learning_rate": 1.930079643276847e-05,
      "loss": 0.6964,
      "step": 2200
    },
    {
      "epoch": 0.2785370792204505,
      "grad_norm": 1.2664152383804321,
      "learning_rate": 1.9300137932581882e-05,
      "loss": 0.6766,
      "step": 2201
    },
    {
      "epoch": 0.27866362946089596,
      "grad_norm": 1.2139124870300293,
      "learning_rate": 1.9299479133702114e-05,
      "loss": 0.6702,
      "step": 2202
    },
    {
      "epoch": 0.27879017970134146,
      "grad_norm": 1.1167672872543335,
      "learning_rate": 1.929882003615033e-05,
      "loss": 0.6472,
      "step": 2203
    },
    {
      "epoch": 0.2789167299417869,
      "grad_norm": 1.2696774005889893,
      "learning_rate": 1.9298160639947695e-05,
      "loss": 0.6765,
      "step": 2204
    },
    {
      "epoch": 0.27904328018223234,
      "grad_norm": 1.2548308372497559,
      "learning_rate": 1.9297500945115388e-05,
      "loss": 0.7295,
      "step": 2205
    },
    {
      "epoch": 0.2791698304226778,
      "grad_norm": 1.3489569425582886,
      "learning_rate": 1.92968409516746e-05,
      "loss": 0.6742,
      "step": 2206
    },
    {
      "epoch": 0.27929638066312323,
      "grad_norm": 1.3010839223861694,
      "learning_rate": 1.9296180659646517e-05,
      "loss": 0.7009,
      "step": 2207
    },
    {
      "epoch": 0.27942293090356873,
      "grad_norm": 1.2307424545288086,
      "learning_rate": 1.929552006905236e-05,
      "loss": 0.6904,
      "step": 2208
    },
    {
      "epoch": 0.2795494811440142,
      "grad_norm": 1.166365146636963,
      "learning_rate": 1.9294859179913334e-05,
      "loss": 0.6962,
      "step": 2209
    },
    {
      "epoch": 0.2796760313844596,
      "grad_norm": 1.2474901676177979,
      "learning_rate": 1.9294197992250675e-05,
      "loss": 0.6827,
      "step": 2210
    },
    {
      "epoch": 0.27980258162490507,
      "grad_norm": 1.1594730615615845,
      "learning_rate": 1.9293536506085607e-05,
      "loss": 0.6577,
      "step": 2211
    },
    {
      "epoch": 0.27992913186535057,
      "grad_norm": 1.2556869983673096,
      "learning_rate": 1.9292874721439386e-05,
      "loss": 0.6834,
      "step": 2212
    },
    {
      "epoch": 0.280055682105796,
      "grad_norm": 1.1926097869873047,
      "learning_rate": 1.9292212638333256e-05,
      "loss": 0.6452,
      "step": 2213
    },
    {
      "epoch": 0.28018223234624146,
      "grad_norm": 1.283728003501892,
      "learning_rate": 1.929155025678849e-05,
      "loss": 0.6921,
      "step": 2214
    },
    {
      "epoch": 0.2803087825866869,
      "grad_norm": 1.2302265167236328,
      "learning_rate": 1.9290887576826357e-05,
      "loss": 0.7351,
      "step": 2215
    },
    {
      "epoch": 0.28043533282713234,
      "grad_norm": 1.6110005378723145,
      "learning_rate": 1.9290224598468143e-05,
      "loss": 0.6724,
      "step": 2216
    },
    {
      "epoch": 0.28056188306757784,
      "grad_norm": 1.6159050464630127,
      "learning_rate": 1.928956132173514e-05,
      "loss": 0.6746,
      "step": 2217
    },
    {
      "epoch": 0.2806884333080233,
      "grad_norm": 1.480055570602417,
      "learning_rate": 1.9288897746648646e-05,
      "loss": 0.6609,
      "step": 2218
    },
    {
      "epoch": 0.28081498354846873,
      "grad_norm": 1.535912036895752,
      "learning_rate": 1.9288233873229982e-05,
      "loss": 0.6661,
      "step": 2219
    },
    {
      "epoch": 0.2809415337889142,
      "grad_norm": 1.2746376991271973,
      "learning_rate": 1.9287569701500463e-05,
      "loss": 0.6562,
      "step": 2220
    },
    {
      "epoch": 0.2810680840293597,
      "grad_norm": 1.2111423015594482,
      "learning_rate": 1.928690523148142e-05,
      "loss": 0.673,
      "step": 2221
    },
    {
      "epoch": 0.2811946342698051,
      "grad_norm": 1.4137012958526611,
      "learning_rate": 1.92862404631942e-05,
      "loss": 0.6751,
      "step": 2222
    },
    {
      "epoch": 0.28132118451025057,
      "grad_norm": 1.1527185440063477,
      "learning_rate": 1.9285575396660144e-05,
      "loss": 0.6283,
      "step": 2223
    },
    {
      "epoch": 0.281447734750696,
      "grad_norm": 1.3329445123672485,
      "learning_rate": 1.9284910031900622e-05,
      "loss": 0.6575,
      "step": 2224
    },
    {
      "epoch": 0.2815742849911415,
      "grad_norm": 1.2693668603897095,
      "learning_rate": 1.9284244368937e-05,
      "loss": 0.7431,
      "step": 2225
    },
    {
      "epoch": 0.28170083523158695,
      "grad_norm": 1.599587321281433,
      "learning_rate": 1.928357840779065e-05,
      "loss": 0.697,
      "step": 2226
    },
    {
      "epoch": 0.2818273854720324,
      "grad_norm": 1.1031957864761353,
      "learning_rate": 1.928291214848297e-05,
      "loss": 0.6385,
      "step": 2227
    },
    {
      "epoch": 0.28195393571247784,
      "grad_norm": 1.4942517280578613,
      "learning_rate": 1.9282245591035355e-05,
      "loss": 0.677,
      "step": 2228
    },
    {
      "epoch": 0.2820804859529233,
      "grad_norm": 1.3027976751327515,
      "learning_rate": 1.928157873546921e-05,
      "loss": 0.6779,
      "step": 2229
    },
    {
      "epoch": 0.2822070361933688,
      "grad_norm": 1.4064185619354248,
      "learning_rate": 1.928091158180596e-05,
      "loss": 0.6916,
      "step": 2230
    },
    {
      "epoch": 0.28233358643381423,
      "grad_norm": 1.331271767616272,
      "learning_rate": 1.9280244130067024e-05,
      "loss": 0.6605,
      "step": 2231
    },
    {
      "epoch": 0.2824601366742597,
      "grad_norm": 1.2583575248718262,
      "learning_rate": 1.9279576380273844e-05,
      "loss": 0.6632,
      "step": 2232
    },
    {
      "epoch": 0.2825866869147051,
      "grad_norm": 1.2423722743988037,
      "learning_rate": 1.9278908332447863e-05,
      "loss": 0.6317,
      "step": 2233
    },
    {
      "epoch": 0.2827132371551506,
      "grad_norm": 1.151948094367981,
      "learning_rate": 1.9278239986610537e-05,
      "loss": 0.6304,
      "step": 2234
    },
    {
      "epoch": 0.28283978739559607,
      "grad_norm": 1.2797582149505615,
      "learning_rate": 1.9277571342783335e-05,
      "loss": 0.713,
      "step": 2235
    },
    {
      "epoch": 0.2829663376360415,
      "grad_norm": 1.2839981317520142,
      "learning_rate": 1.9276902400987725e-05,
      "loss": 0.6477,
      "step": 2236
    },
    {
      "epoch": 0.28309288787648695,
      "grad_norm": 1.2694467306137085,
      "learning_rate": 1.92762331612452e-05,
      "loss": 0.6841,
      "step": 2237
    },
    {
      "epoch": 0.2832194381169324,
      "grad_norm": 1.3442816734313965,
      "learning_rate": 1.9275563623577244e-05,
      "loss": 0.6576,
      "step": 2238
    },
    {
      "epoch": 0.2833459883573779,
      "grad_norm": 1.2035210132598877,
      "learning_rate": 1.9274893788005374e-05,
      "loss": 0.6738,
      "step": 2239
    },
    {
      "epoch": 0.28347253859782334,
      "grad_norm": 1.356966495513916,
      "learning_rate": 1.927422365455109e-05,
      "loss": 0.6708,
      "step": 2240
    },
    {
      "epoch": 0.2835990888382688,
      "grad_norm": 1.2260533571243286,
      "learning_rate": 1.927355322323592e-05,
      "loss": 0.6398,
      "step": 2241
    },
    {
      "epoch": 0.28372563907871423,
      "grad_norm": 1.2787985801696777,
      "learning_rate": 1.9272882494081396e-05,
      "loss": 0.6868,
      "step": 2242
    },
    {
      "epoch": 0.28385218931915973,
      "grad_norm": 1.143290400505066,
      "learning_rate": 1.927221146710906e-05,
      "loss": 0.7007,
      "step": 2243
    },
    {
      "epoch": 0.2839787395596052,
      "grad_norm": 1.3524729013442993,
      "learning_rate": 1.9271540142340464e-05,
      "loss": 0.6968,
      "step": 2244
    },
    {
      "epoch": 0.2841052898000506,
      "grad_norm": 1.2061817646026611,
      "learning_rate": 1.9270868519797172e-05,
      "loss": 0.7463,
      "step": 2245
    },
    {
      "epoch": 0.28423184004049606,
      "grad_norm": 1.3158925771713257,
      "learning_rate": 1.9270196599500744e-05,
      "loss": 0.7266,
      "step": 2246
    },
    {
      "epoch": 0.2843583902809415,
      "grad_norm": 1.1990960836410522,
      "learning_rate": 1.926952438147277e-05,
      "loss": 0.7127,
      "step": 2247
    },
    {
      "epoch": 0.284484940521387,
      "grad_norm": 2.7317373752593994,
      "learning_rate": 1.9268851865734842e-05,
      "loss": 0.7179,
      "step": 2248
    },
    {
      "epoch": 0.28461149076183245,
      "grad_norm": 2.437161684036255,
      "learning_rate": 1.9268179052308552e-05,
      "loss": 0.6743,
      "step": 2249
    },
    {
      "epoch": 0.2847380410022779,
      "grad_norm": 1.1724286079406738,
      "learning_rate": 1.9267505941215505e-05,
      "loss": 0.6693,
      "step": 2250
    },
    {
      "epoch": 0.28486459124272334,
      "grad_norm": 1.9456478357315063,
      "learning_rate": 1.9266832532477332e-05,
      "loss": 0.7119,
      "step": 2251
    },
    {
      "epoch": 0.28499114148316884,
      "grad_norm": 1.4567795991897583,
      "learning_rate": 1.926615882611565e-05,
      "loss": 0.7047,
      "step": 2252
    },
    {
      "epoch": 0.2851176917236143,
      "grad_norm": 1.5203088521957397,
      "learning_rate": 1.92654848221521e-05,
      "loss": 0.6596,
      "step": 2253
    },
    {
      "epoch": 0.28524424196405973,
      "grad_norm": 1.4917539358139038,
      "learning_rate": 1.9264810520608336e-05,
      "loss": 0.6761,
      "step": 2254
    },
    {
      "epoch": 0.2853707922045052,
      "grad_norm": 1.3721147775650024,
      "learning_rate": 1.9264135921506e-05,
      "loss": 0.7006,
      "step": 2255
    },
    {
      "epoch": 0.2854973424449506,
      "grad_norm": 1.390321969985962,
      "learning_rate": 1.9263461024866773e-05,
      "loss": 0.652,
      "step": 2256
    },
    {
      "epoch": 0.2856238926853961,
      "grad_norm": 1.210954189300537,
      "learning_rate": 1.926278583071232e-05,
      "loss": 0.6621,
      "step": 2257
    },
    {
      "epoch": 0.28575044292584156,
      "grad_norm": 1.2137054204940796,
      "learning_rate": 1.926211033906433e-05,
      "loss": 0.6453,
      "step": 2258
    },
    {
      "epoch": 0.285876993166287,
      "grad_norm": 1.076522707939148,
      "learning_rate": 1.92614345499445e-05,
      "loss": 0.6525,
      "step": 2259
    },
    {
      "epoch": 0.28600354340673245,
      "grad_norm": 1.4629164934158325,
      "learning_rate": 1.9260758463374536e-05,
      "loss": 0.6637,
      "step": 2260
    },
    {
      "epoch": 0.28613009364717795,
      "grad_norm": 1.245474100112915,
      "learning_rate": 1.9260082079376145e-05,
      "loss": 0.716,
      "step": 2261
    },
    {
      "epoch": 0.2862566438876234,
      "grad_norm": 1.3017228841781616,
      "learning_rate": 1.9259405397971052e-05,
      "loss": 0.6793,
      "step": 2262
    },
    {
      "epoch": 0.28638319412806884,
      "grad_norm": 1.2524045705795288,
      "learning_rate": 1.9258728419180995e-05,
      "loss": 0.7089,
      "step": 2263
    },
    {
      "epoch": 0.2865097443685143,
      "grad_norm": 1.6127194166183472,
      "learning_rate": 1.925805114302771e-05,
      "loss": 0.7108,
      "step": 2264
    },
    {
      "epoch": 0.28663629460895973,
      "grad_norm": 1.1124303340911865,
      "learning_rate": 1.925737356953296e-05,
      "loss": 0.6748,
      "step": 2265
    },
    {
      "epoch": 0.28676284484940523,
      "grad_norm": 1.4668159484863281,
      "learning_rate": 1.925669569871849e-05,
      "loss": 0.7168,
      "step": 2266
    },
    {
      "epoch": 0.2868893950898507,
      "grad_norm": 1.3156027793884277,
      "learning_rate": 1.9256017530606087e-05,
      "loss": 0.689,
      "step": 2267
    },
    {
      "epoch": 0.2870159453302961,
      "grad_norm": 1.440192461013794,
      "learning_rate": 1.925533906521752e-05,
      "loss": 0.6783,
      "step": 2268
    },
    {
      "epoch": 0.28714249557074156,
      "grad_norm": 1.3108162879943848,
      "learning_rate": 1.925466030257459e-05,
      "loss": 0.6732,
      "step": 2269
    },
    {
      "epoch": 0.28726904581118706,
      "grad_norm": 1.2536064386367798,
      "learning_rate": 1.9253981242699087e-05,
      "loss": 0.6252,
      "step": 2270
    },
    {
      "epoch": 0.2873955960516325,
      "grad_norm": 1.2843822240829468,
      "learning_rate": 1.9253301885612828e-05,
      "loss": 0.6828,
      "step": 2271
    },
    {
      "epoch": 0.28752214629207795,
      "grad_norm": 3.710359811782837,
      "learning_rate": 1.9252622231337626e-05,
      "loss": 0.6755,
      "step": 2272
    },
    {
      "epoch": 0.2876486965325234,
      "grad_norm": 1.856526494026184,
      "learning_rate": 1.9251942279895318e-05,
      "loss": 0.7017,
      "step": 2273
    },
    {
      "epoch": 0.2877752467729689,
      "grad_norm": 1.1952158212661743,
      "learning_rate": 1.9251262031307733e-05,
      "loss": 0.6537,
      "step": 2274
    },
    {
      "epoch": 0.28790179701341434,
      "grad_norm": 1.65059494972229,
      "learning_rate": 1.925058148559672e-05,
      "loss": 0.6799,
      "step": 2275
    },
    {
      "epoch": 0.2880283472538598,
      "grad_norm": 1.3483970165252686,
      "learning_rate": 1.9249900642784142e-05,
      "loss": 0.672,
      "step": 2276
    },
    {
      "epoch": 0.28815489749430523,
      "grad_norm": 1.4251182079315186,
      "learning_rate": 1.9249219502891857e-05,
      "loss": 0.6541,
      "step": 2277
    },
    {
      "epoch": 0.2882814477347507,
      "grad_norm": 1.6314789056777954,
      "learning_rate": 1.924853806594175e-05,
      "loss": 0.6922,
      "step": 2278
    },
    {
      "epoch": 0.2884079979751962,
      "grad_norm": 1.191184401512146,
      "learning_rate": 1.9247856331955704e-05,
      "loss": 0.6939,
      "step": 2279
    },
    {
      "epoch": 0.2885345482156416,
      "grad_norm": 1.6518797874450684,
      "learning_rate": 1.924717430095561e-05,
      "loss": 0.7189,
      "step": 2280
    },
    {
      "epoch": 0.28866109845608706,
      "grad_norm": 1.1681867837905884,
      "learning_rate": 1.9246491972963377e-05,
      "loss": 0.7219,
      "step": 2281
    },
    {
      "epoch": 0.2887876486965325,
      "grad_norm": 1.4760500192642212,
      "learning_rate": 1.9245809348000917e-05,
      "loss": 0.6917,
      "step": 2282
    },
    {
      "epoch": 0.288914198936978,
      "grad_norm": 1.1698871850967407,
      "learning_rate": 1.9245126426090154e-05,
      "loss": 0.6405,
      "step": 2283
    },
    {
      "epoch": 0.28904074917742345,
      "grad_norm": 1.3783793449401855,
      "learning_rate": 1.9244443207253027e-05,
      "loss": 0.7243,
      "step": 2284
    },
    {
      "epoch": 0.2891672994178689,
      "grad_norm": 1.2991875410079956,
      "learning_rate": 1.9243759691511472e-05,
      "loss": 0.7011,
      "step": 2285
    },
    {
      "epoch": 0.28929384965831434,
      "grad_norm": 1.4108848571777344,
      "learning_rate": 1.9243075878887444e-05,
      "loss": 0.682,
      "step": 2286
    },
    {
      "epoch": 0.2894203998987598,
      "grad_norm": 1.3751459121704102,
      "learning_rate": 1.9242391769402906e-05,
      "loss": 0.6654,
      "step": 2287
    },
    {
      "epoch": 0.2895469501392053,
      "grad_norm": 1.2144001722335815,
      "learning_rate": 1.9241707363079827e-05,
      "loss": 0.6857,
      "step": 2288
    },
    {
      "epoch": 0.28967350037965073,
      "grad_norm": 1.4326225519180298,
      "learning_rate": 1.9241022659940192e-05,
      "loss": 0.675,
      "step": 2289
    },
    {
      "epoch": 0.2898000506200962,
      "grad_norm": 1.1351265907287598,
      "learning_rate": 1.9240337660005993e-05,
      "loss": 0.6234,
      "step": 2290
    },
    {
      "epoch": 0.2899266008605416,
      "grad_norm": 1.3048040866851807,
      "learning_rate": 1.9239652363299224e-05,
      "loss": 0.6936,
      "step": 2291
    },
    {
      "epoch": 0.2900531511009871,
      "grad_norm": 1.1020585298538208,
      "learning_rate": 1.9238966769841897e-05,
      "loss": 0.697,
      "step": 2292
    },
    {
      "epoch": 0.29017970134143256,
      "grad_norm": 1.0910873413085938,
      "learning_rate": 1.9238280879656032e-05,
      "loss": 0.6428,
      "step": 2293
    },
    {
      "epoch": 0.290306251581878,
      "grad_norm": 1.0858328342437744,
      "learning_rate": 1.9237594692763654e-05,
      "loss": 0.6571,
      "step": 2294
    },
    {
      "epoch": 0.29043280182232345,
      "grad_norm": 1.269611120223999,
      "learning_rate": 1.9236908209186808e-05,
      "loss": 0.7032,
      "step": 2295
    },
    {
      "epoch": 0.2905593520627689,
      "grad_norm": 1.187048077583313,
      "learning_rate": 1.923622142894754e-05,
      "loss": 0.6831,
      "step": 2296
    },
    {
      "epoch": 0.2906859023032144,
      "grad_norm": 1.3301125764846802,
      "learning_rate": 1.9235534352067905e-05,
      "loss": 0.6528,
      "step": 2297
    },
    {
      "epoch": 0.29081245254365984,
      "grad_norm": 1.1387379169464111,
      "learning_rate": 1.923484697856997e-05,
      "loss": 0.7081,
      "step": 2298
    },
    {
      "epoch": 0.2909390027841053,
      "grad_norm": 1.2405732870101929,
      "learning_rate": 1.9234159308475814e-05,
      "loss": 0.7081,
      "step": 2299
    },
    {
      "epoch": 0.2910655530245507,
      "grad_norm": 1.2447412014007568,
      "learning_rate": 1.923347134180752e-05,
      "loss": 0.6612,
      "step": 2300
    },
    {
      "epoch": 0.2911921032649962,
      "grad_norm": 1.145851492881775,
      "learning_rate": 1.9232783078587185e-05,
      "loss": 0.6445,
      "step": 2301
    },
    {
      "epoch": 0.29131865350544167,
      "grad_norm": 1.2117632627487183,
      "learning_rate": 1.9232094518836915e-05,
      "loss": 0.7085,
      "step": 2302
    },
    {
      "epoch": 0.2914452037458871,
      "grad_norm": 1.1200906038284302,
      "learning_rate": 1.923140566257882e-05,
      "loss": 0.718,
      "step": 2303
    },
    {
      "epoch": 0.29157175398633256,
      "grad_norm": 1.2394739389419556,
      "learning_rate": 1.9230716509835032e-05,
      "loss": 0.6827,
      "step": 2304
    },
    {
      "epoch": 0.291698304226778,
      "grad_norm": 1.1278444528579712,
      "learning_rate": 1.923002706062768e-05,
      "loss": 0.691,
      "step": 2305
    },
    {
      "epoch": 0.2918248544672235,
      "grad_norm": 1.0927510261535645,
      "learning_rate": 1.9229337314978905e-05,
      "loss": 0.674,
      "step": 2306
    },
    {
      "epoch": 0.29195140470766895,
      "grad_norm": 1.115313172340393,
      "learning_rate": 1.922864727291086e-05,
      "loss": 0.6846,
      "step": 2307
    },
    {
      "epoch": 0.2920779549481144,
      "grad_norm": 1.121808648109436,
      "learning_rate": 1.922795693444571e-05,
      "loss": 0.6576,
      "step": 2308
    },
    {
      "epoch": 0.29220450518855984,
      "grad_norm": 1.0711945295333862,
      "learning_rate": 1.9227266299605628e-05,
      "loss": 0.6669,
      "step": 2309
    },
    {
      "epoch": 0.29233105542900534,
      "grad_norm": 1.1301854848861694,
      "learning_rate": 1.9226575368412787e-05,
      "loss": 0.6832,
      "step": 2310
    },
    {
      "epoch": 0.2924576056694508,
      "grad_norm": 1.1584711074829102,
      "learning_rate": 1.922588414088939e-05,
      "loss": 0.6989,
      "step": 2311
    },
    {
      "epoch": 0.2925841559098962,
      "grad_norm": 1.2252802848815918,
      "learning_rate": 1.9225192617057625e-05,
      "loss": 0.7024,
      "step": 2312
    },
    {
      "epoch": 0.29271070615034167,
      "grad_norm": 1.1455875635147095,
      "learning_rate": 1.9224500796939712e-05,
      "loss": 0.6822,
      "step": 2313
    },
    {
      "epoch": 0.2928372563907871,
      "grad_norm": 1.1023941040039062,
      "learning_rate": 1.922380868055786e-05,
      "loss": 0.6322,
      "step": 2314
    },
    {
      "epoch": 0.2929638066312326,
      "grad_norm": 1.1679918766021729,
      "learning_rate": 1.922311626793431e-05,
      "loss": 0.6533,
      "step": 2315
    },
    {
      "epoch": 0.29309035687167806,
      "grad_norm": 1.1022127866744995,
      "learning_rate": 1.9222423559091287e-05,
      "loss": 0.6618,
      "step": 2316
    },
    {
      "epoch": 0.2932169071121235,
      "grad_norm": 1.0780224800109863,
      "learning_rate": 1.922173055405105e-05,
      "loss": 0.6826,
      "step": 2317
    },
    {
      "epoch": 0.29334345735256895,
      "grad_norm": 1.1358656883239746,
      "learning_rate": 1.9221037252835847e-05,
      "loss": 0.6047,
      "step": 2318
    },
    {
      "epoch": 0.29347000759301445,
      "grad_norm": 1.0887959003448486,
      "learning_rate": 1.9220343655467953e-05,
      "loss": 0.6507,
      "step": 2319
    },
    {
      "epoch": 0.2935965578334599,
      "grad_norm": 1.0857595205307007,
      "learning_rate": 1.921964976196964e-05,
      "loss": 0.6522,
      "step": 2320
    },
    {
      "epoch": 0.29372310807390534,
      "grad_norm": 1.276652216911316,
      "learning_rate": 1.9218955572363192e-05,
      "loss": 0.6609,
      "step": 2321
    },
    {
      "epoch": 0.2938496583143508,
      "grad_norm": 1.1804572343826294,
      "learning_rate": 1.921826108667091e-05,
      "loss": 0.6763,
      "step": 2322
    },
    {
      "epoch": 0.2939762085547963,
      "grad_norm": 1.1562458276748657,
      "learning_rate": 1.921756630491509e-05,
      "loss": 0.6307,
      "step": 2323
    },
    {
      "epoch": 0.2941027587952417,
      "grad_norm": 1.1932611465454102,
      "learning_rate": 1.9216871227118057e-05,
      "loss": 0.6731,
      "step": 2324
    },
    {
      "epoch": 0.29422930903568717,
      "grad_norm": 1.1228519678115845,
      "learning_rate": 1.9216175853302132e-05,
      "loss": 0.6799,
      "step": 2325
    },
    {
      "epoch": 0.2943558592761326,
      "grad_norm": 1.2664060592651367,
      "learning_rate": 1.921548018348964e-05,
      "loss": 0.7049,
      "step": 2326
    },
    {
      "epoch": 0.29448240951657806,
      "grad_norm": 1.262054443359375,
      "learning_rate": 1.9214784217702935e-05,
      "loss": 0.7258,
      "step": 2327
    },
    {
      "epoch": 0.29460895975702356,
      "grad_norm": 1.1400460004806519,
      "learning_rate": 1.9214087955964363e-05,
      "loss": 0.5963,
      "step": 2328
    },
    {
      "epoch": 0.294735509997469,
      "grad_norm": 1.170332670211792,
      "learning_rate": 1.9213391398296288e-05,
      "loss": 0.6915,
      "step": 2329
    },
    {
      "epoch": 0.29486206023791445,
      "grad_norm": 1.2916465997695923,
      "learning_rate": 1.921269454472108e-05,
      "loss": 0.6984,
      "step": 2330
    },
    {
      "epoch": 0.2949886104783599,
      "grad_norm": 1.1446070671081543,
      "learning_rate": 1.921199739526112e-05,
      "loss": 0.6876,
      "step": 2331
    },
    {
      "epoch": 0.2951151607188054,
      "grad_norm": 1.140730619430542,
      "learning_rate": 1.92112999499388e-05,
      "loss": 0.6735,
      "step": 2332
    },
    {
      "epoch": 0.29524171095925084,
      "grad_norm": 1.1533358097076416,
      "learning_rate": 1.9210602208776522e-05,
      "loss": 0.6922,
      "step": 2333
    },
    {
      "epoch": 0.2953682611996963,
      "grad_norm": 2.2694900035858154,
      "learning_rate": 1.9209904171796684e-05,
      "loss": 0.6759,
      "step": 2334
    },
    {
      "epoch": 0.2954948114401417,
      "grad_norm": 1.2650765180587769,
      "learning_rate": 1.920920583902172e-05,
      "loss": 0.7251,
      "step": 2335
    },
    {
      "epoch": 0.29562136168058717,
      "grad_norm": 1.2907081842422485,
      "learning_rate": 1.9208507210474054e-05,
      "loss": 0.6699,
      "step": 2336
    },
    {
      "epoch": 0.29574791192103267,
      "grad_norm": 1.1730892658233643,
      "learning_rate": 1.9207808286176117e-05,
      "loss": 0.6387,
      "step": 2337
    },
    {
      "epoch": 0.2958744621614781,
      "grad_norm": 1.5014188289642334,
      "learning_rate": 1.920710906615036e-05,
      "loss": 0.6915,
      "step": 2338
    },
    {
      "epoch": 0.29600101240192356,
      "grad_norm": 1.2729839086532593,
      "learning_rate": 1.9206409550419242e-05,
      "loss": 0.6872,
      "step": 2339
    },
    {
      "epoch": 0.296127562642369,
      "grad_norm": 2.454040050506592,
      "learning_rate": 1.9205709739005228e-05,
      "loss": 0.6625,
      "step": 2340
    },
    {
      "epoch": 0.2962541128828145,
      "grad_norm": 1.5475025177001953,
      "learning_rate": 1.9205009631930795e-05,
      "loss": 0.6877,
      "step": 2341
    },
    {
      "epoch": 0.29638066312325995,
      "grad_norm": 1.1420613527297974,
      "learning_rate": 1.9204309229218428e-05,
      "loss": 0.6477,
      "step": 2342
    },
    {
      "epoch": 0.2965072133637054,
      "grad_norm": 1.3507921695709229,
      "learning_rate": 1.920360853089062e-05,
      "loss": 0.6801,
      "step": 2343
    },
    {
      "epoch": 0.29663376360415084,
      "grad_norm": 1.2669190168380737,
      "learning_rate": 1.9202907536969877e-05,
      "loss": 0.7269,
      "step": 2344
    },
    {
      "epoch": 0.2967603138445963,
      "grad_norm": 1.2385772466659546,
      "learning_rate": 1.9202206247478714e-05,
      "loss": 0.6257,
      "step": 2345
    },
    {
      "epoch": 0.2968868640850418,
      "grad_norm": 1.0978447198867798,
      "learning_rate": 1.9201504662439653e-05,
      "loss": 0.6577,
      "step": 2346
    },
    {
      "epoch": 0.2970134143254872,
      "grad_norm": 1.1677873134613037,
      "learning_rate": 1.9200802781875225e-05,
      "loss": 0.6479,
      "step": 2347
    },
    {
      "epoch": 0.29713996456593267,
      "grad_norm": 1.1437983512878418,
      "learning_rate": 1.920010060580798e-05,
      "loss": 0.6677,
      "step": 2348
    },
    {
      "epoch": 0.2972665148063781,
      "grad_norm": 1.4852381944656372,
      "learning_rate": 1.9199398134260457e-05,
      "loss": 0.6619,
      "step": 2349
    },
    {
      "epoch": 0.2973930650468236,
      "grad_norm": 1.1245867013931274,
      "learning_rate": 1.9198695367255227e-05,
      "loss": 0.6838,
      "step": 2350
    },
    {
      "epoch": 0.29751961528726906,
      "grad_norm": 2.487992286682129,
      "learning_rate": 1.9197992304814858e-05,
      "loss": 0.7392,
      "step": 2351
    },
    {
      "epoch": 0.2976461655277145,
      "grad_norm": 1.289067029953003,
      "learning_rate": 1.919728894696193e-05,
      "loss": 0.6589,
      "step": 2352
    },
    {
      "epoch": 0.29777271576815995,
      "grad_norm": 1.1078583002090454,
      "learning_rate": 1.9196585293719033e-05,
      "loss": 0.6578,
      "step": 2353
    },
    {
      "epoch": 0.2978992660086054,
      "grad_norm": 1.210647702217102,
      "learning_rate": 1.919588134510877e-05,
      "loss": 0.6979,
      "step": 2354
    },
    {
      "epoch": 0.2980258162490509,
      "grad_norm": 1.8224499225616455,
      "learning_rate": 1.9195177101153744e-05,
      "loss": 0.6912,
      "step": 2355
    },
    {
      "epoch": 0.29815236648949633,
      "grad_norm": 1.2925302982330322,
      "learning_rate": 1.9194472561876572e-05,
      "loss": 0.6636,
      "step": 2356
    },
    {
      "epoch": 0.2982789167299418,
      "grad_norm": 1.123360514640808,
      "learning_rate": 1.9193767727299892e-05,
      "loss": 0.738,
      "step": 2357
    },
    {
      "epoch": 0.2984054669703872,
      "grad_norm": 1.398868203163147,
      "learning_rate": 1.919306259744633e-05,
      "loss": 0.6702,
      "step": 2358
    },
    {
      "epoch": 0.2985320172108327,
      "grad_norm": 1.3486449718475342,
      "learning_rate": 1.919235717233854e-05,
      "loss": 0.7129,
      "step": 2359
    },
    {
      "epoch": 0.29865856745127817,
      "grad_norm": 1.3076082468032837,
      "learning_rate": 1.9191651451999177e-05,
      "loss": 0.6444,
      "step": 2360
    },
    {
      "epoch": 0.2987851176917236,
      "grad_norm": 1.1353919506072998,
      "learning_rate": 1.91909454364509e-05,
      "loss": 0.6824,
      "step": 2361
    },
    {
      "epoch": 0.29891166793216906,
      "grad_norm": 1.1026968955993652,
      "learning_rate": 1.919023912571639e-05,
      "loss": 0.6712,
      "step": 2362
    },
    {
      "epoch": 0.2990382181726145,
      "grad_norm": 1.2026370763778687,
      "learning_rate": 1.9189532519818334e-05,
      "loss": 0.6677,
      "step": 2363
    },
    {
      "epoch": 0.29916476841306,
      "grad_norm": 1.1124355792999268,
      "learning_rate": 1.918882561877942e-05,
      "loss": 0.6748,
      "step": 2364
    },
    {
      "epoch": 0.29929131865350544,
      "grad_norm": 1.06806218624115,
      "learning_rate": 1.918811842262236e-05,
      "loss": 0.6731,
      "step": 2365
    },
    {
      "epoch": 0.2994178688939509,
      "grad_norm": 1.3255932331085205,
      "learning_rate": 1.9187410931369855e-05,
      "loss": 0.6855,
      "step": 2366
    },
    {
      "epoch": 0.29954441913439633,
      "grad_norm": 1.080070972442627,
      "learning_rate": 1.918670314504464e-05,
      "loss": 0.6787,
      "step": 2367
    },
    {
      "epoch": 0.29967096937484183,
      "grad_norm": 1.34927237033844,
      "learning_rate": 1.918599506366944e-05,
      "loss": 0.6479,
      "step": 2368
    },
    {
      "epoch": 0.2997975196152873,
      "grad_norm": 1.0977826118469238,
      "learning_rate": 1.9185286687266995e-05,
      "loss": 0.6317,
      "step": 2369
    },
    {
      "epoch": 0.2999240698557327,
      "grad_norm": 1.2675760984420776,
      "learning_rate": 1.9184578015860057e-05,
      "loss": 0.6191,
      "step": 2370
    },
    {
      "epoch": 0.30005062009617817,
      "grad_norm": 1.1091456413269043,
      "learning_rate": 1.9183869049471396e-05,
      "loss": 0.6654,
      "step": 2371
    },
    {
      "epoch": 0.30017717033662367,
      "grad_norm": 1.3059979677200317,
      "learning_rate": 1.9183159788123765e-05,
      "loss": 0.6964,
      "step": 2372
    },
    {
      "epoch": 0.3003037205770691,
      "grad_norm": 1.2906736135482788,
      "learning_rate": 1.918245023183996e-05,
      "loss": 0.7033,
      "step": 2373
    },
    {
      "epoch": 0.30043027081751456,
      "grad_norm": 1.3428544998168945,
      "learning_rate": 1.9181740380642758e-05,
      "loss": 0.6854,
      "step": 2374
    },
    {
      "epoch": 0.30055682105796,
      "grad_norm": 1.25877845287323,
      "learning_rate": 1.9181030234554963e-05,
      "loss": 0.7328,
      "step": 2375
    },
    {
      "epoch": 0.30068337129840544,
      "grad_norm": 1.247884750366211,
      "learning_rate": 1.9180319793599383e-05,
      "loss": 0.6675,
      "step": 2376
    },
    {
      "epoch": 0.30080992153885094,
      "grad_norm": 1.2537747621536255,
      "learning_rate": 1.917960905779883e-05,
      "loss": 0.7043,
      "step": 2377
    },
    {
      "epoch": 0.3009364717792964,
      "grad_norm": 1.147627830505371,
      "learning_rate": 1.9178898027176135e-05,
      "loss": 0.6821,
      "step": 2378
    },
    {
      "epoch": 0.30106302201974183,
      "grad_norm": 1.0992374420166016,
      "learning_rate": 1.917818670175414e-05,
      "loss": 0.6592,
      "step": 2379
    },
    {
      "epoch": 0.3011895722601873,
      "grad_norm": 1.2735240459442139,
      "learning_rate": 1.917747508155568e-05,
      "loss": 0.7171,
      "step": 2380
    },
    {
      "epoch": 0.3013161225006328,
      "grad_norm": 1.1190608739852905,
      "learning_rate": 1.9176763166603613e-05,
      "loss": 0.7016,
      "step": 2381
    },
    {
      "epoch": 0.3014426727410782,
      "grad_norm": 1.221461534500122,
      "learning_rate": 1.9176050956920806e-05,
      "loss": 0.6697,
      "step": 2382
    },
    {
      "epoch": 0.30156922298152367,
      "grad_norm": 1.15225350856781,
      "learning_rate": 1.9175338452530134e-05,
      "loss": 0.6743,
      "step": 2383
    },
    {
      "epoch": 0.3016957732219691,
      "grad_norm": 1.209125280380249,
      "learning_rate": 1.9174625653454476e-05,
      "loss": 0.7563,
      "step": 2384
    },
    {
      "epoch": 0.30182232346241455,
      "grad_norm": 1.1908372640609741,
      "learning_rate": 1.917391255971673e-05,
      "loss": 0.6728,
      "step": 2385
    },
    {
      "epoch": 0.30194887370286005,
      "grad_norm": 1.214199423789978,
      "learning_rate": 1.9173199171339795e-05,
      "loss": 0.7089,
      "step": 2386
    },
    {
      "epoch": 0.3020754239433055,
      "grad_norm": 1.1526983976364136,
      "learning_rate": 1.917248548834659e-05,
      "loss": 0.6549,
      "step": 2387
    },
    {
      "epoch": 0.30220197418375094,
      "grad_norm": 1.1317170858383179,
      "learning_rate": 1.9171771510760026e-05,
      "loss": 0.6798,
      "step": 2388
    },
    {
      "epoch": 0.3023285244241964,
      "grad_norm": 1.0835803747177124,
      "learning_rate": 1.9171057238603038e-05,
      "loss": 0.6093,
      "step": 2389
    },
    {
      "epoch": 0.3024550746646419,
      "grad_norm": 1.1072524785995483,
      "learning_rate": 1.917034267189857e-05,
      "loss": 0.6937,
      "step": 2390
    },
    {
      "epoch": 0.30258162490508733,
      "grad_norm": 1.1591894626617432,
      "learning_rate": 1.9169627810669563e-05,
      "loss": 0.6303,
      "step": 2391
    },
    {
      "epoch": 0.3027081751455328,
      "grad_norm": 1.169550895690918,
      "learning_rate": 1.916891265493899e-05,
      "loss": 0.7195,
      "step": 2392
    },
    {
      "epoch": 0.3028347253859782,
      "grad_norm": 1.2952815294265747,
      "learning_rate": 1.9168197204729804e-05,
      "loss": 0.696,
      "step": 2393
    },
    {
      "epoch": 0.30296127562642367,
      "grad_norm": 1.210940957069397,
      "learning_rate": 1.9167481460064994e-05,
      "loss": 0.6908,
      "step": 2394
    },
    {
      "epoch": 0.30308782586686916,
      "grad_norm": 1.1449395418167114,
      "learning_rate": 1.9166765420967543e-05,
      "loss": 0.6808,
      "step": 2395
    },
    {
      "epoch": 0.3032143761073146,
      "grad_norm": 1.1225558519363403,
      "learning_rate": 1.916604908746045e-05,
      "loss": 0.6524,
      "step": 2396
    },
    {
      "epoch": 0.30334092634776005,
      "grad_norm": 1.0526472330093384,
      "learning_rate": 1.9165332459566723e-05,
      "loss": 0.6448,
      "step": 2397
    },
    {
      "epoch": 0.3034674765882055,
      "grad_norm": 1.125791311264038,
      "learning_rate": 1.9164615537309377e-05,
      "loss": 0.6455,
      "step": 2398
    },
    {
      "epoch": 0.303594026828651,
      "grad_norm": 1.2099051475524902,
      "learning_rate": 1.9163898320711433e-05,
      "loss": 0.6818,
      "step": 2399
    },
    {
      "epoch": 0.30372057706909644,
      "grad_norm": 1.2460461854934692,
      "learning_rate": 1.916318080979593e-05,
      "loss": 0.6438,
      "step": 2400
    },
    {
      "epoch": 0.3038471273095419,
      "grad_norm": 1.3023167848587036,
      "learning_rate": 1.9162463004585915e-05,
      "loss": 0.6849,
      "step": 2401
    },
    {
      "epoch": 0.30397367754998733,
      "grad_norm": 1.2899408340454102,
      "learning_rate": 1.9161744905104435e-05,
      "loss": 0.698,
      "step": 2402
    },
    {
      "epoch": 0.3041002277904328,
      "grad_norm": 1.248580813407898,
      "learning_rate": 1.916102651137456e-05,
      "loss": 0.6622,
      "step": 2403
    },
    {
      "epoch": 0.3042267780308783,
      "grad_norm": 1.4394384622573853,
      "learning_rate": 1.9160307823419355e-05,
      "loss": 0.744,
      "step": 2404
    },
    {
      "epoch": 0.3043533282713237,
      "grad_norm": 1.2196393013000488,
      "learning_rate": 1.9159588841261908e-05,
      "loss": 0.6967,
      "step": 2405
    },
    {
      "epoch": 0.30447987851176916,
      "grad_norm": 1.3984012603759766,
      "learning_rate": 1.9158869564925313e-05,
      "loss": 0.6458,
      "step": 2406
    },
    {
      "epoch": 0.3046064287522146,
      "grad_norm": 1.4400266408920288,
      "learning_rate": 1.9158149994432665e-05,
      "loss": 0.6957,
      "step": 2407
    },
    {
      "epoch": 0.3047329789926601,
      "grad_norm": 1.4418392181396484,
      "learning_rate": 1.9157430129807077e-05,
      "loss": 0.7204,
      "step": 2408
    },
    {
      "epoch": 0.30485952923310555,
      "grad_norm": 1.5575100183486938,
      "learning_rate": 1.9156709971071666e-05,
      "loss": 0.7182,
      "step": 2409
    },
    {
      "epoch": 0.304986079473551,
      "grad_norm": 1.1602870225906372,
      "learning_rate": 1.9155989518249568e-05,
      "loss": 0.7065,
      "step": 2410
    },
    {
      "epoch": 0.30511262971399644,
      "grad_norm": 1.2147818803787231,
      "learning_rate": 1.9155268771363915e-05,
      "loss": 0.6673,
      "step": 2411
    },
    {
      "epoch": 0.3052391799544419,
      "grad_norm": 2.1966774463653564,
      "learning_rate": 1.915454773043786e-05,
      "loss": 0.6672,
      "step": 2412
    },
    {
      "epoch": 0.3053657301948874,
      "grad_norm": 1.209800362586975,
      "learning_rate": 1.9153826395494558e-05,
      "loss": 0.6948,
      "step": 2413
    },
    {
      "epoch": 0.30549228043533283,
      "grad_norm": 1.1305210590362549,
      "learning_rate": 1.915310476655718e-05,
      "loss": 0.6528,
      "step": 2414
    },
    {
      "epoch": 0.3056188306757783,
      "grad_norm": 1.2479077577590942,
      "learning_rate": 1.9152382843648895e-05,
      "loss": 0.6733,
      "step": 2415
    },
    {
      "epoch": 0.3057453809162237,
      "grad_norm": 1.1095272302627563,
      "learning_rate": 1.9151660626792897e-05,
      "loss": 0.6654,
      "step": 2416
    },
    {
      "epoch": 0.3058719311566692,
      "grad_norm": 1.131777286529541,
      "learning_rate": 1.9150938116012377e-05,
      "loss": 0.6409,
      "step": 2417
    },
    {
      "epoch": 0.30599848139711466,
      "grad_norm": 1.1941022872924805,
      "learning_rate": 1.915021531133054e-05,
      "loss": 0.7177,
      "step": 2418
    },
    {
      "epoch": 0.3061250316375601,
      "grad_norm": 1.1381275653839111,
      "learning_rate": 1.9149492212770605e-05,
      "loss": 0.6558,
      "step": 2419
    },
    {
      "epoch": 0.30625158187800555,
      "grad_norm": 1.1410025358200073,
      "learning_rate": 1.914876882035579e-05,
      "loss": 0.6963,
      "step": 2420
    },
    {
      "epoch": 0.30637813211845105,
      "grad_norm": 1.151899814605713,
      "learning_rate": 1.914804513410933e-05,
      "loss": 0.6405,
      "step": 2421
    },
    {
      "epoch": 0.3065046823588965,
      "grad_norm": 1.1564102172851562,
      "learning_rate": 1.914732115405447e-05,
      "loss": 0.6731,
      "step": 2422
    },
    {
      "epoch": 0.30663123259934194,
      "grad_norm": 1.1620115041732788,
      "learning_rate": 1.914659688021446e-05,
      "loss": 0.6638,
      "step": 2423
    },
    {
      "epoch": 0.3067577828397874,
      "grad_norm": 1.3998125791549683,
      "learning_rate": 1.9145872312612562e-05,
      "loss": 0.6661,
      "step": 2424
    },
    {
      "epoch": 0.30688433308023283,
      "grad_norm": 1.1782217025756836,
      "learning_rate": 1.9145147451272045e-05,
      "loss": 0.6626,
      "step": 2425
    },
    {
      "epoch": 0.30701088332067833,
      "grad_norm": 1.3754316568374634,
      "learning_rate": 1.9144422296216193e-05,
      "loss": 0.6942,
      "step": 2426
    },
    {
      "epoch": 0.3071374335611238,
      "grad_norm": 1.1836981773376465,
      "learning_rate": 1.9143696847468295e-05,
      "loss": 0.6348,
      "step": 2427
    },
    {
      "epoch": 0.3072639838015692,
      "grad_norm": 1.2203402519226074,
      "learning_rate": 1.9142971105051645e-05,
      "loss": 0.6499,
      "step": 2428
    },
    {
      "epoch": 0.30739053404201466,
      "grad_norm": 1.2126156091690063,
      "learning_rate": 1.9142245068989564e-05,
      "loss": 0.6586,
      "step": 2429
    },
    {
      "epoch": 0.30751708428246016,
      "grad_norm": 1.27278733253479,
      "learning_rate": 1.9141518739305354e-05,
      "loss": 0.6843,
      "step": 2430
    },
    {
      "epoch": 0.3076436345229056,
      "grad_norm": 1.1349947452545166,
      "learning_rate": 1.914079211602236e-05,
      "loss": 0.6625,
      "step": 2431
    },
    {
      "epoch": 0.30777018476335105,
      "grad_norm": 1.2768971920013428,
      "learning_rate": 1.91400651991639e-05,
      "loss": 0.6929,
      "step": 2432
    },
    {
      "epoch": 0.3078967350037965,
      "grad_norm": 1.1746339797973633,
      "learning_rate": 1.913933798875334e-05,
      "loss": 0.65,
      "step": 2433
    },
    {
      "epoch": 0.30802328524424194,
      "grad_norm": 1.1998722553253174,
      "learning_rate": 1.913861048481402e-05,
      "loss": 0.6156,
      "step": 2434
    },
    {
      "epoch": 0.30814983548468744,
      "grad_norm": 1.121586561203003,
      "learning_rate": 1.9137882687369313e-05,
      "loss": 0.6442,
      "step": 2435
    },
    {
      "epoch": 0.3082763857251329,
      "grad_norm": 1.1027894020080566,
      "learning_rate": 1.9137154596442595e-05,
      "loss": 0.6792,
      "step": 2436
    },
    {
      "epoch": 0.30840293596557833,
      "grad_norm": 1.2581855058670044,
      "learning_rate": 1.9136426212057243e-05,
      "loss": 0.6887,
      "step": 2437
    },
    {
      "epoch": 0.3085294862060238,
      "grad_norm": 1.2536678314208984,
      "learning_rate": 1.9135697534236655e-05,
      "loss": 0.6752,
      "step": 2438
    },
    {
      "epoch": 0.3086560364464693,
      "grad_norm": 1.1216659545898438,
      "learning_rate": 1.9134968563004238e-05,
      "loss": 0.6385,
      "step": 2439
    },
    {
      "epoch": 0.3087825866869147,
      "grad_norm": 1.173002004623413,
      "learning_rate": 1.9134239298383402e-05,
      "loss": 0.6781,
      "step": 2440
    },
    {
      "epoch": 0.30890913692736016,
      "grad_norm": 1.3690868616104126,
      "learning_rate": 1.913350974039756e-05,
      "loss": 0.6691,
      "step": 2441
    },
    {
      "epoch": 0.3090356871678056,
      "grad_norm": 1.195446491241455,
      "learning_rate": 1.9132779889070156e-05,
      "loss": 0.6912,
      "step": 2442
    },
    {
      "epoch": 0.30916223740825105,
      "grad_norm": 1.285412311553955,
      "learning_rate": 1.9132049744424623e-05,
      "loss": 0.6603,
      "step": 2443
    },
    {
      "epoch": 0.30928878764869655,
      "grad_norm": 1.199991226196289,
      "learning_rate": 1.9131319306484414e-05,
      "loss": 0.6597,
      "step": 2444
    },
    {
      "epoch": 0.309415337889142,
      "grad_norm": 1.1918500661849976,
      "learning_rate": 1.9130588575272987e-05,
      "loss": 0.6346,
      "step": 2445
    },
    {
      "epoch": 0.30954188812958744,
      "grad_norm": 1.2525718212127686,
      "learning_rate": 1.9129857550813817e-05,
      "loss": 0.6833,
      "step": 2446
    },
    {
      "epoch": 0.3096684383700329,
      "grad_norm": 1.1659340858459473,
      "learning_rate": 1.912912623313037e-05,
      "loss": 0.6449,
      "step": 2447
    },
    {
      "epoch": 0.3097949886104784,
      "grad_norm": 1.1340223550796509,
      "learning_rate": 1.9128394622246143e-05,
      "loss": 0.6836,
      "step": 2448
    },
    {
      "epoch": 0.30992153885092383,
      "grad_norm": 1.3298460245132446,
      "learning_rate": 1.9127662718184632e-05,
      "loss": 0.6824,
      "step": 2449
    },
    {
      "epoch": 0.31004808909136927,
      "grad_norm": 1.1289913654327393,
      "learning_rate": 1.9126930520969342e-05,
      "loss": 0.6731,
      "step": 2450
    },
    {
      "epoch": 0.3101746393318147,
      "grad_norm": 1.3816134929656982,
      "learning_rate": 1.912619803062379e-05,
      "loss": 0.7164,
      "step": 2451
    },
    {
      "epoch": 0.31030118957226016,
      "grad_norm": 1.2343436479568481,
      "learning_rate": 1.91254652471715e-05,
      "loss": 0.7017,
      "step": 2452
    },
    {
      "epoch": 0.31042773981270566,
      "grad_norm": 1.5834800004959106,
      "learning_rate": 1.9124732170636015e-05,
      "loss": 0.6976,
      "step": 2453
    },
    {
      "epoch": 0.3105542900531511,
      "grad_norm": 1.2320058345794678,
      "learning_rate": 1.9123998801040864e-05,
      "loss": 0.7138,
      "step": 2454
    },
    {
      "epoch": 0.31068084029359655,
      "grad_norm": 1.396832823753357,
      "learning_rate": 1.9123265138409613e-05,
      "loss": 0.6795,
      "step": 2455
    },
    {
      "epoch": 0.310807390534042,
      "grad_norm": 1.3174705505371094,
      "learning_rate": 1.9122531182765817e-05,
      "loss": 0.728,
      "step": 2456
    },
    {
      "epoch": 0.3109339407744875,
      "grad_norm": 1.1898387670516968,
      "learning_rate": 1.9121796934133057e-05,
      "loss": 0.6434,
      "step": 2457
    },
    {
      "epoch": 0.31106049101493294,
      "grad_norm": 1.3090736865997314,
      "learning_rate": 1.912106239253491e-05,
      "loss": 0.6778,
      "step": 2458
    },
    {
      "epoch": 0.3111870412553784,
      "grad_norm": 1.085234522819519,
      "learning_rate": 1.912032755799497e-05,
      "loss": 0.6748,
      "step": 2459
    },
    {
      "epoch": 0.3113135914958238,
      "grad_norm": 1.1057937145233154,
      "learning_rate": 1.9119592430536832e-05,
      "loss": 0.6663,
      "step": 2460
    },
    {
      "epoch": 0.31144014173626927,
      "grad_norm": 1.0657390356063843,
      "learning_rate": 1.9118857010184107e-05,
      "loss": 0.6512,
      "step": 2461
    },
    {
      "epoch": 0.31156669197671477,
      "grad_norm": 1.1700886487960815,
      "learning_rate": 1.9118121296960424e-05,
      "loss": 0.7014,
      "step": 2462
    },
    {
      "epoch": 0.3116932422171602,
      "grad_norm": 1.7461519241333008,
      "learning_rate": 1.91173852908894e-05,
      "loss": 0.6864,
      "step": 2463
    },
    {
      "epoch": 0.31181979245760566,
      "grad_norm": 1.221987009048462,
      "learning_rate": 1.911664899199468e-05,
      "loss": 0.6433,
      "step": 2464
    },
    {
      "epoch": 0.3119463426980511,
      "grad_norm": 1.2448540925979614,
      "learning_rate": 1.911591240029991e-05,
      "loss": 0.6977,
      "step": 2465
    },
    {
      "epoch": 0.3120728929384966,
      "grad_norm": 1.1318249702453613,
      "learning_rate": 1.911517551582875e-05,
      "loss": 0.6813,
      "step": 2466
    },
    {
      "epoch": 0.31219944317894205,
      "grad_norm": 1.1172972917556763,
      "learning_rate": 1.9114438338604865e-05,
      "loss": 0.6905,
      "step": 2467
    },
    {
      "epoch": 0.3123259934193875,
      "grad_norm": 1.2400152683258057,
      "learning_rate": 1.9113700868651928e-05,
      "loss": 0.7113,
      "step": 2468
    },
    {
      "epoch": 0.31245254365983294,
      "grad_norm": 1.2104047536849976,
      "learning_rate": 1.9112963105993624e-05,
      "loss": 0.6876,
      "step": 2469
    },
    {
      "epoch": 0.31257909390027844,
      "grad_norm": 1.0745267868041992,
      "learning_rate": 1.9112225050653653e-05,
      "loss": 0.642,
      "step": 2470
    },
    {
      "epoch": 0.3127056441407239,
      "grad_norm": 1.1658568382263184,
      "learning_rate": 1.9111486702655714e-05,
      "loss": 0.7064,
      "step": 2471
    },
    {
      "epoch": 0.3128321943811693,
      "grad_norm": 1.1242197751998901,
      "learning_rate": 1.9110748062023528e-05,
      "loss": 0.6948,
      "step": 2472
    },
    {
      "epoch": 0.31295874462161477,
      "grad_norm": 1.4100626707077026,
      "learning_rate": 1.911000912878081e-05,
      "loss": 0.7363,
      "step": 2473
    },
    {
      "epoch": 0.3130852948620602,
      "grad_norm": 1.1010181903839111,
      "learning_rate": 1.9109269902951296e-05,
      "loss": 0.6513,
      "step": 2474
    },
    {
      "epoch": 0.3132118451025057,
      "grad_norm": 1.261474847793579,
      "learning_rate": 1.9108530384558722e-05,
      "loss": 0.6752,
      "step": 2475
    },
    {
      "epoch": 0.31333839534295116,
      "grad_norm": 1.1236003637313843,
      "learning_rate": 1.910779057362685e-05,
      "loss": 0.6495,
      "step": 2476
    },
    {
      "epoch": 0.3134649455833966,
      "grad_norm": 1.1568374633789062,
      "learning_rate": 1.9107050470179433e-05,
      "loss": 0.666,
      "step": 2477
    },
    {
      "epoch": 0.31359149582384205,
      "grad_norm": 1.1833186149597168,
      "learning_rate": 1.9106310074240243e-05,
      "loss": 0.7006,
      "step": 2478
    },
    {
      "epoch": 0.31371804606428755,
      "grad_norm": 1.1587326526641846,
      "learning_rate": 1.910556938583306e-05,
      "loss": 0.633,
      "step": 2479
    },
    {
      "epoch": 0.313844596304733,
      "grad_norm": 1.1253560781478882,
      "learning_rate": 1.910482840498167e-05,
      "loss": 0.6225,
      "step": 2480
    },
    {
      "epoch": 0.31397114654517844,
      "grad_norm": 1.261959433555603,
      "learning_rate": 1.9104087131709873e-05,
      "loss": 0.6583,
      "step": 2481
    },
    {
      "epoch": 0.3140976967856239,
      "grad_norm": 1.1201939582824707,
      "learning_rate": 1.9103345566041477e-05,
      "loss": 0.6717,
      "step": 2482
    },
    {
      "epoch": 0.3142242470260693,
      "grad_norm": 1.1574106216430664,
      "learning_rate": 1.91026037080003e-05,
      "loss": 0.681,
      "step": 2483
    },
    {
      "epoch": 0.3143507972665148,
      "grad_norm": 1.192556381225586,
      "learning_rate": 1.9101861557610165e-05,
      "loss": 0.6836,
      "step": 2484
    },
    {
      "epoch": 0.31447734750696027,
      "grad_norm": 1.3849091529846191,
      "learning_rate": 1.9101119114894908e-05,
      "loss": 0.6689,
      "step": 2485
    },
    {
      "epoch": 0.3146038977474057,
      "grad_norm": 1.1657263040542603,
      "learning_rate": 1.910037637987838e-05,
      "loss": 0.6767,
      "step": 2486
    },
    {
      "epoch": 0.31473044798785116,
      "grad_norm": 1.324154257774353,
      "learning_rate": 1.9099633352584426e-05,
      "loss": 0.621,
      "step": 2487
    },
    {
      "epoch": 0.31485699822829666,
      "grad_norm": 1.324389100074768,
      "learning_rate": 1.9098890033036918e-05,
      "loss": 0.6821,
      "step": 2488
    },
    {
      "epoch": 0.3149835484687421,
      "grad_norm": 1.2039180994033813,
      "learning_rate": 1.9098146421259726e-05,
      "loss": 0.7106,
      "step": 2489
    },
    {
      "epoch": 0.31511009870918755,
      "grad_norm": 1.1268612146377563,
      "learning_rate": 1.9097402517276732e-05,
      "loss": 0.7153,
      "step": 2490
    },
    {
      "epoch": 0.315236648949633,
      "grad_norm": 1.1404730081558228,
      "learning_rate": 1.9096658321111828e-05,
      "loss": 0.669,
      "step": 2491
    },
    {
      "epoch": 0.31536319919007844,
      "grad_norm": 1.085402488708496,
      "learning_rate": 1.9095913832788916e-05,
      "loss": 0.6727,
      "step": 2492
    },
    {
      "epoch": 0.31548974943052394,
      "grad_norm": 1.172845482826233,
      "learning_rate": 1.9095169052331906e-05,
      "loss": 0.6701,
      "step": 2493
    },
    {
      "epoch": 0.3156162996709694,
      "grad_norm": 1.1348145008087158,
      "learning_rate": 1.9094423979764725e-05,
      "loss": 0.7163,
      "step": 2494
    },
    {
      "epoch": 0.3157428499114148,
      "grad_norm": 1.0996739864349365,
      "learning_rate": 1.9093678615111294e-05,
      "loss": 0.6307,
      "step": 2495
    },
    {
      "epoch": 0.31586940015186027,
      "grad_norm": 1.129888892173767,
      "learning_rate": 1.9092932958395553e-05,
      "loss": 0.6731,
      "step": 2496
    },
    {
      "epoch": 0.31599595039230577,
      "grad_norm": 1.1936547756195068,
      "learning_rate": 1.909218700964145e-05,
      "loss": 0.7052,
      "step": 2497
    },
    {
      "epoch": 0.3161225006327512,
      "grad_norm": 1.1524752378463745,
      "learning_rate": 1.9091440768872946e-05,
      "loss": 0.6217,
      "step": 2498
    },
    {
      "epoch": 0.31624905087319666,
      "grad_norm": 1.0908414125442505,
      "learning_rate": 1.9090694236114005e-05,
      "loss": 0.6302,
      "step": 2499
    },
    {
      "epoch": 0.3163756011136421,
      "grad_norm": 1.179583191871643,
      "learning_rate": 1.908994741138861e-05,
      "loss": 0.6544,
      "step": 2500
    },
    {
      "epoch": 0.31650215135408755,
      "grad_norm": 1.1493048667907715,
      "learning_rate": 1.9089200294720737e-05,
      "loss": 0.6752,
      "step": 2501
    },
    {
      "epoch": 0.31662870159453305,
      "grad_norm": 1.1533101797103882,
      "learning_rate": 1.908845288613439e-05,
      "loss": 0.7027,
      "step": 2502
    },
    {
      "epoch": 0.3167552518349785,
      "grad_norm": 1.1428030729293823,
      "learning_rate": 1.9087705185653566e-05,
      "loss": 0.7163,
      "step": 2503
    },
    {
      "epoch": 0.31688180207542394,
      "grad_norm": 1.2021300792694092,
      "learning_rate": 1.9086957193302288e-05,
      "loss": 0.6447,
      "step": 2504
    },
    {
      "epoch": 0.3170083523158694,
      "grad_norm": 1.101332664489746,
      "learning_rate": 1.9086208909104567e-05,
      "loss": 0.6447,
      "step": 2505
    },
    {
      "epoch": 0.3171349025563149,
      "grad_norm": 1.1093209981918335,
      "learning_rate": 1.9085460333084447e-05,
      "loss": 0.7071,
      "step": 2506
    },
    {
      "epoch": 0.3172614527967603,
      "grad_norm": 1.147874355316162,
      "learning_rate": 1.9084711465265962e-05,
      "loss": 0.6902,
      "step": 2507
    },
    {
      "epoch": 0.31738800303720577,
      "grad_norm": 1.1686620712280273,
      "learning_rate": 1.9083962305673172e-05,
      "loss": 0.6853,
      "step": 2508
    },
    {
      "epoch": 0.3175145532776512,
      "grad_norm": 1.3153820037841797,
      "learning_rate": 1.908321285433013e-05,
      "loss": 0.7193,
      "step": 2509
    },
    {
      "epoch": 0.31764110351809666,
      "grad_norm": 1.1049870252609253,
      "learning_rate": 1.908246311126091e-05,
      "loss": 0.6526,
      "step": 2510
    },
    {
      "epoch": 0.31776765375854216,
      "grad_norm": 1.1822243928909302,
      "learning_rate": 1.9081713076489592e-05,
      "loss": 0.6208,
      "step": 2511
    },
    {
      "epoch": 0.3178942039989876,
      "grad_norm": 1.104270100593567,
      "learning_rate": 1.908096275004026e-05,
      "loss": 0.6351,
      "step": 2512
    },
    {
      "epoch": 0.31802075423943305,
      "grad_norm": 1.1572846174240112,
      "learning_rate": 1.9080212131937017e-05,
      "loss": 0.7142,
      "step": 2513
    },
    {
      "epoch": 0.3181473044798785,
      "grad_norm": 1.228797435760498,
      "learning_rate": 1.907946122220397e-05,
      "loss": 0.6711,
      "step": 2514
    },
    {
      "epoch": 0.318273854720324,
      "grad_norm": 1.1193865537643433,
      "learning_rate": 1.9078710020865235e-05,
      "loss": 0.6409,
      "step": 2515
    },
    {
      "epoch": 0.31840040496076943,
      "grad_norm": 1.1347503662109375,
      "learning_rate": 1.9077958527944944e-05,
      "loss": 0.6761,
      "step": 2516
    },
    {
      "epoch": 0.3185269552012149,
      "grad_norm": 1.2460780143737793,
      "learning_rate": 1.9077206743467222e-05,
      "loss": 0.6467,
      "step": 2517
    },
    {
      "epoch": 0.3186535054416603,
      "grad_norm": 1.1753673553466797,
      "learning_rate": 1.9076454667456223e-05,
      "loss": 0.6506,
      "step": 2518
    },
    {
      "epoch": 0.3187800556821058,
      "grad_norm": 1.2035521268844604,
      "learning_rate": 1.9075702299936094e-05,
      "loss": 0.7069,
      "step": 2519
    },
    {
      "epoch": 0.31890660592255127,
      "grad_norm": 1.3195048570632935,
      "learning_rate": 1.907494964093101e-05,
      "loss": 0.6847,
      "step": 2520
    },
    {
      "epoch": 0.3190331561629967,
      "grad_norm": 1.2037506103515625,
      "learning_rate": 1.907419669046513e-05,
      "loss": 0.5967,
      "step": 2521
    },
    {
      "epoch": 0.31915970640344216,
      "grad_norm": 1.2109975814819336,
      "learning_rate": 1.907344344856265e-05,
      "loss": 0.6491,
      "step": 2522
    },
    {
      "epoch": 0.3192862566438876,
      "grad_norm": 1.1287477016448975,
      "learning_rate": 1.907268991524775e-05,
      "loss": 0.6516,
      "step": 2523
    },
    {
      "epoch": 0.3194128068843331,
      "grad_norm": 1.1476210355758667,
      "learning_rate": 1.9071936090544642e-05,
      "loss": 0.6771,
      "step": 2524
    },
    {
      "epoch": 0.31953935712477854,
      "grad_norm": 1.101014494895935,
      "learning_rate": 1.9071181974477527e-05,
      "loss": 0.6746,
      "step": 2525
    },
    {
      "epoch": 0.319665907365224,
      "grad_norm": 1.1901377439498901,
      "learning_rate": 1.907042756707063e-05,
      "loss": 0.6624,
      "step": 2526
    },
    {
      "epoch": 0.31979245760566943,
      "grad_norm": 1.0437265634536743,
      "learning_rate": 1.9069672868348187e-05,
      "loss": 0.637,
      "step": 2527
    },
    {
      "epoch": 0.31991900784611493,
      "grad_norm": 1.0727907419204712,
      "learning_rate": 1.9068917878334426e-05,
      "loss": 0.666,
      "step": 2528
    },
    {
      "epoch": 0.3200455580865604,
      "grad_norm": 1.115738868713379,
      "learning_rate": 1.9068162597053596e-05,
      "loss": 0.6953,
      "step": 2529
    },
    {
      "epoch": 0.3201721083270058,
      "grad_norm": 1.0848329067230225,
      "learning_rate": 1.906740702452996e-05,
      "loss": 0.6961,
      "step": 2530
    },
    {
      "epoch": 0.32029865856745127,
      "grad_norm": 1.1181355714797974,
      "learning_rate": 1.9066651160787783e-05,
      "loss": 0.7022,
      "step": 2531
    },
    {
      "epoch": 0.3204252088078967,
      "grad_norm": 1.0923906564712524,
      "learning_rate": 1.9065895005851337e-05,
      "loss": 0.6431,
      "step": 2532
    },
    {
      "epoch": 0.3205517590483422,
      "grad_norm": 1.1214449405670166,
      "learning_rate": 1.906513855974491e-05,
      "loss": 0.7016,
      "step": 2533
    },
    {
      "epoch": 0.32067830928878766,
      "grad_norm": 1.076848030090332,
      "learning_rate": 1.90643818224928e-05,
      "loss": 0.6518,
      "step": 2534
    },
    {
      "epoch": 0.3208048595292331,
      "grad_norm": 1.082580804824829,
      "learning_rate": 1.9063624794119308e-05,
      "loss": 0.6742,
      "step": 2535
    },
    {
      "epoch": 0.32093140976967854,
      "grad_norm": 1.3041726350784302,
      "learning_rate": 1.906286747464875e-05,
      "loss": 0.6592,
      "step": 2536
    },
    {
      "epoch": 0.32105796001012404,
      "grad_norm": 1.1777820587158203,
      "learning_rate": 1.9062109864105448e-05,
      "loss": 0.6764,
      "step": 2537
    },
    {
      "epoch": 0.3211845102505695,
      "grad_norm": 1.0588692426681519,
      "learning_rate": 1.906135196251373e-05,
      "loss": 0.6671,
      "step": 2538
    },
    {
      "epoch": 0.32131106049101493,
      "grad_norm": 1.1874797344207764,
      "learning_rate": 1.9060593769897942e-05,
      "loss": 0.6587,
      "step": 2539
    },
    {
      "epoch": 0.3214376107314604,
      "grad_norm": 1.0815770626068115,
      "learning_rate": 1.9059835286282434e-05,
      "loss": 0.653,
      "step": 2540
    },
    {
      "epoch": 0.3215641609719058,
      "grad_norm": 1.2093102931976318,
      "learning_rate": 1.9059076511691566e-05,
      "loss": 0.6983,
      "step": 2541
    },
    {
      "epoch": 0.3216907112123513,
      "grad_norm": 1.1219230890274048,
      "learning_rate": 1.905831744614971e-05,
      "loss": 0.6419,
      "step": 2542
    },
    {
      "epoch": 0.32181726145279677,
      "grad_norm": 1.1977219581604004,
      "learning_rate": 1.905755808968124e-05,
      "loss": 0.6831,
      "step": 2543
    },
    {
      "epoch": 0.3219438116932422,
      "grad_norm": 1.1079010963439941,
      "learning_rate": 1.9056798442310544e-05,
      "loss": 0.6561,
      "step": 2544
    },
    {
      "epoch": 0.32207036193368765,
      "grad_norm": 1.3943511247634888,
      "learning_rate": 1.9056038504062025e-05,
      "loss": 0.6805,
      "step": 2545
    },
    {
      "epoch": 0.32219691217413315,
      "grad_norm": 1.1440469026565552,
      "learning_rate": 1.9055278274960088e-05,
      "loss": 0.6911,
      "step": 2546
    },
    {
      "epoch": 0.3223234624145786,
      "grad_norm": 1.313088297843933,
      "learning_rate": 1.9054517755029152e-05,
      "loss": 0.7241,
      "step": 2547
    },
    {
      "epoch": 0.32245001265502404,
      "grad_norm": 1.1264877319335938,
      "learning_rate": 1.9053756944293634e-05,
      "loss": 0.6488,
      "step": 2548
    },
    {
      "epoch": 0.3225765628954695,
      "grad_norm": 1.131474256515503,
      "learning_rate": 1.905299584277798e-05,
      "loss": 0.6467,
      "step": 2549
    },
    {
      "epoch": 0.32270311313591493,
      "grad_norm": 1.4494996070861816,
      "learning_rate": 1.9052234450506626e-05,
      "loss": 0.6741,
      "step": 2550
    },
    {
      "epoch": 0.32282966337636043,
      "grad_norm": 1.1640692949295044,
      "learning_rate": 1.905147276750403e-05,
      "loss": 0.6767,
      "step": 2551
    },
    {
      "epoch": 0.3229562136168059,
      "grad_norm": 1.155745506286621,
      "learning_rate": 1.9050710793794654e-05,
      "loss": 0.6486,
      "step": 2552
    },
    {
      "epoch": 0.3230827638572513,
      "grad_norm": 1.1438724994659424,
      "learning_rate": 1.9049948529402968e-05,
      "loss": 0.6041,
      "step": 2553
    },
    {
      "epoch": 0.32320931409769676,
      "grad_norm": 1.1376025676727295,
      "learning_rate": 1.9049185974353457e-05,
      "loss": 0.6633,
      "step": 2554
    },
    {
      "epoch": 0.32333586433814226,
      "grad_norm": 1.1695468425750732,
      "learning_rate": 1.9048423128670612e-05,
      "loss": 0.6703,
      "step": 2555
    },
    {
      "epoch": 0.3234624145785877,
      "grad_norm": 1.0465813875198364,
      "learning_rate": 1.9047659992378927e-05,
      "loss": 0.6287,
      "step": 2556
    },
    {
      "epoch": 0.32358896481903315,
      "grad_norm": 1.1635348796844482,
      "learning_rate": 1.9046896565502924e-05,
      "loss": 0.7054,
      "step": 2557
    },
    {
      "epoch": 0.3237155150594786,
      "grad_norm": 1.210452675819397,
      "learning_rate": 1.904613284806711e-05,
      "loss": 0.622,
      "step": 2558
    },
    {
      "epoch": 0.32384206529992404,
      "grad_norm": 1.392088532447815,
      "learning_rate": 1.904536884009602e-05,
      "loss": 0.6734,
      "step": 2559
    },
    {
      "epoch": 0.32396861554036954,
      "grad_norm": 1.1447644233703613,
      "learning_rate": 1.904460454161419e-05,
      "loss": 0.703,
      "step": 2560
    },
    {
      "epoch": 0.324095165780815,
      "grad_norm": 1.1171029806137085,
      "learning_rate": 1.9043839952646167e-05,
      "loss": 0.6686,
      "step": 2561
    },
    {
      "epoch": 0.32422171602126043,
      "grad_norm": 1.1134803295135498,
      "learning_rate": 1.9043075073216507e-05,
      "loss": 0.652,
      "step": 2562
    },
    {
      "epoch": 0.3243482662617059,
      "grad_norm": 1.1549540758132935,
      "learning_rate": 1.904230990334978e-05,
      "loss": 0.6497,
      "step": 2563
    },
    {
      "epoch": 0.3244748165021514,
      "grad_norm": 1.1721065044403076,
      "learning_rate": 1.904154444307055e-05,
      "loss": 0.6802,
      "step": 2564
    },
    {
      "epoch": 0.3246013667425968,
      "grad_norm": 1.1454843282699585,
      "learning_rate": 1.9040778692403412e-05,
      "loss": 0.6411,
      "step": 2565
    },
    {
      "epoch": 0.32472791698304226,
      "grad_norm": 1.1188325881958008,
      "learning_rate": 1.9040012651372955e-05,
      "loss": 0.6744,
      "step": 2566
    },
    {
      "epoch": 0.3248544672234877,
      "grad_norm": 1.0781182050704956,
      "learning_rate": 1.9039246320003784e-05,
      "loss": 0.6828,
      "step": 2567
    },
    {
      "epoch": 0.3249810174639332,
      "grad_norm": 1.1191622018814087,
      "learning_rate": 1.903847969832051e-05,
      "loss": 0.6456,
      "step": 2568
    },
    {
      "epoch": 0.32510756770437865,
      "grad_norm": 1.1278207302093506,
      "learning_rate": 1.9037712786347755e-05,
      "loss": 0.6623,
      "step": 2569
    },
    {
      "epoch": 0.3252341179448241,
      "grad_norm": 1.0386601686477661,
      "learning_rate": 1.903694558411015e-05,
      "loss": 0.6637,
      "step": 2570
    },
    {
      "epoch": 0.32536066818526954,
      "grad_norm": 1.1140486001968384,
      "learning_rate": 1.9036178091632335e-05,
      "loss": 0.6421,
      "step": 2571
    },
    {
      "epoch": 0.325487218425715,
      "grad_norm": 1.4298408031463623,
      "learning_rate": 1.903541030893896e-05,
      "loss": 0.7301,
      "step": 2572
    },
    {
      "epoch": 0.3256137686661605,
      "grad_norm": 1.1318457126617432,
      "learning_rate": 1.9034642236054683e-05,
      "loss": 0.6767,
      "step": 2573
    },
    {
      "epoch": 0.32574031890660593,
      "grad_norm": 1.2183383703231812,
      "learning_rate": 1.9033873873004177e-05,
      "loss": 0.6933,
      "step": 2574
    },
    {
      "epoch": 0.3258668691470514,
      "grad_norm": 1.1089524030685425,
      "learning_rate": 1.903310521981211e-05,
      "loss": 0.6535,
      "step": 2575
    },
    {
      "epoch": 0.3259934193874968,
      "grad_norm": 1.054723858833313,
      "learning_rate": 1.9032336276503178e-05,
      "loss": 0.6431,
      "step": 2576
    },
    {
      "epoch": 0.3261199696279423,
      "grad_norm": 1.1599452495574951,
      "learning_rate": 1.9031567043102073e-05,
      "loss": 0.6514,
      "step": 2577
    },
    {
      "epoch": 0.32624651986838776,
      "grad_norm": 1.21922767162323,
      "learning_rate": 1.90307975196335e-05,
      "loss": 0.6976,
      "step": 2578
    },
    {
      "epoch": 0.3263730701088332,
      "grad_norm": 1.05745530128479,
      "learning_rate": 1.9030027706122175e-05,
      "loss": 0.6478,
      "step": 2579
    },
    {
      "epoch": 0.32649962034927865,
      "grad_norm": 1.169604778289795,
      "learning_rate": 1.9029257602592825e-05,
      "loss": 0.7094,
      "step": 2580
    },
    {
      "epoch": 0.3266261705897241,
      "grad_norm": 1.152143955230713,
      "learning_rate": 1.9028487209070178e-05,
      "loss": 0.7169,
      "step": 2581
    },
    {
      "epoch": 0.3267527208301696,
      "grad_norm": 1.0936825275421143,
      "learning_rate": 1.902771652557898e-05,
      "loss": 0.6667,
      "step": 2582
    },
    {
      "epoch": 0.32687927107061504,
      "grad_norm": 1.175484299659729,
      "learning_rate": 1.9026945552143985e-05,
      "loss": 0.6814,
      "step": 2583
    },
    {
      "epoch": 0.3270058213110605,
      "grad_norm": 1.2572977542877197,
      "learning_rate": 1.902617428878995e-05,
      "loss": 0.6936,
      "step": 2584
    },
    {
      "epoch": 0.32713237155150593,
      "grad_norm": 1.3970571756362915,
      "learning_rate": 1.9025402735541647e-05,
      "loss": 0.6615,
      "step": 2585
    },
    {
      "epoch": 0.32725892179195143,
      "grad_norm": 1.2229403257369995,
      "learning_rate": 1.9024630892423857e-05,
      "loss": 0.6393,
      "step": 2586
    },
    {
      "epoch": 0.3273854720323969,
      "grad_norm": 1.144653081893921,
      "learning_rate": 1.902385875946137e-05,
      "loss": 0.6673,
      "step": 2587
    },
    {
      "epoch": 0.3275120222728423,
      "grad_norm": 1.1159390211105347,
      "learning_rate": 1.9023086336678984e-05,
      "loss": 0.6642,
      "step": 2588
    },
    {
      "epoch": 0.32763857251328776,
      "grad_norm": 1.2287980318069458,
      "learning_rate": 1.9022313624101504e-05,
      "loss": 0.679,
      "step": 2589
    },
    {
      "epoch": 0.3277651227537332,
      "grad_norm": 1.6126234531402588,
      "learning_rate": 1.9021540621753754e-05,
      "loss": 0.6651,
      "step": 2590
    },
    {
      "epoch": 0.3278916729941787,
      "grad_norm": 1.1472835540771484,
      "learning_rate": 1.9020767329660553e-05,
      "loss": 0.676,
      "step": 2591
    },
    {
      "epoch": 0.32801822323462415,
      "grad_norm": 1.0393867492675781,
      "learning_rate": 1.9019993747846743e-05,
      "loss": 0.6386,
      "step": 2592
    },
    {
      "epoch": 0.3281447734750696,
      "grad_norm": 1.1323829889297485,
      "learning_rate": 1.9019219876337164e-05,
      "loss": 0.6662,
      "step": 2593
    },
    {
      "epoch": 0.32827132371551504,
      "grad_norm": 1.1378740072250366,
      "learning_rate": 1.9018445715156674e-05,
      "loss": 0.6614,
      "step": 2594
    },
    {
      "epoch": 0.32839787395596054,
      "grad_norm": 1.144293189048767,
      "learning_rate": 1.9017671264330136e-05,
      "loss": 0.6461,
      "step": 2595
    },
    {
      "epoch": 0.328524424196406,
      "grad_norm": 1.1850624084472656,
      "learning_rate": 1.9016896523882424e-05,
      "loss": 0.6863,
      "step": 2596
    },
    {
      "epoch": 0.32865097443685143,
      "grad_norm": 1.263000249862671,
      "learning_rate": 1.901612149383842e-05,
      "loss": 0.7094,
      "step": 2597
    },
    {
      "epoch": 0.3287775246772969,
      "grad_norm": 1.0556870698928833,
      "learning_rate": 1.9015346174223012e-05,
      "loss": 0.6705,
      "step": 2598
    },
    {
      "epoch": 0.3289040749177423,
      "grad_norm": 1.075974464416504,
      "learning_rate": 1.9014570565061104e-05,
      "loss": 0.6749,
      "step": 2599
    },
    {
      "epoch": 0.3290306251581878,
      "grad_norm": 1.121903419494629,
      "learning_rate": 1.901379466637761e-05,
      "loss": 0.6757,
      "step": 2600
    },
    {
      "epoch": 0.32915717539863326,
      "grad_norm": 1.615971326828003,
      "learning_rate": 1.9013018478197443e-05,
      "loss": 0.6658,
      "step": 2601
    },
    {
      "epoch": 0.3292837256390787,
      "grad_norm": 1.2637332677841187,
      "learning_rate": 1.9012242000545536e-05,
      "loss": 0.6916,
      "step": 2602
    },
    {
      "epoch": 0.32941027587952415,
      "grad_norm": 1.1683894395828247,
      "learning_rate": 1.9011465233446824e-05,
      "loss": 0.7086,
      "step": 2603
    },
    {
      "epoch": 0.32953682611996965,
      "grad_norm": 1.1785240173339844,
      "learning_rate": 1.9010688176926258e-05,
      "loss": 0.6392,
      "step": 2604
    },
    {
      "epoch": 0.3296633763604151,
      "grad_norm": 1.108154058456421,
      "learning_rate": 1.9009910831008795e-05,
      "loss": 0.6858,
      "step": 2605
    },
    {
      "epoch": 0.32978992660086054,
      "grad_norm": 1.0966063737869263,
      "learning_rate": 1.9009133195719394e-05,
      "loss": 0.6569,
      "step": 2606
    },
    {
      "epoch": 0.329916476841306,
      "grad_norm": 1.11668062210083,
      "learning_rate": 1.900835527108304e-05,
      "loss": 0.6626,
      "step": 2607
    },
    {
      "epoch": 0.33004302708175143,
      "grad_norm": 1.1222115755081177,
      "learning_rate": 1.9007577057124713e-05,
      "loss": 0.6853,
      "step": 2608
    },
    {
      "epoch": 0.3301695773221969,
      "grad_norm": 1.1083064079284668,
      "learning_rate": 1.9006798553869406e-05,
      "loss": 0.6798,
      "step": 2609
    },
    {
      "epoch": 0.33029612756264237,
      "grad_norm": 1.0915979146957397,
      "learning_rate": 1.9006019761342127e-05,
      "loss": 0.647,
      "step": 2610
    },
    {
      "epoch": 0.3304226778030878,
      "grad_norm": 1.2276513576507568,
      "learning_rate": 1.900524067956788e-05,
      "loss": 0.7288,
      "step": 2611
    },
    {
      "epoch": 0.33054922804353326,
      "grad_norm": 1.0951697826385498,
      "learning_rate": 1.9004461308571694e-05,
      "loss": 0.6528,
      "step": 2612
    },
    {
      "epoch": 0.33067577828397876,
      "grad_norm": 1.281195044517517,
      "learning_rate": 1.90036816483786e-05,
      "loss": 0.6925,
      "step": 2613
    },
    {
      "epoch": 0.3308023285244242,
      "grad_norm": 1.0923312902450562,
      "learning_rate": 1.9002901699013635e-05,
      "loss": 0.6903,
      "step": 2614
    },
    {
      "epoch": 0.33092887876486965,
      "grad_norm": 1.2658277750015259,
      "learning_rate": 1.900212146050185e-05,
      "loss": 0.6734,
      "step": 2615
    },
    {
      "epoch": 0.3310554290053151,
      "grad_norm": 1.145256519317627,
      "learning_rate": 1.9001340932868298e-05,
      "loss": 0.709,
      "step": 2616
    },
    {
      "epoch": 0.3311819792457606,
      "grad_norm": 1.1719635725021362,
      "learning_rate": 1.9000560116138063e-05,
      "loss": 0.6625,
      "step": 2617
    },
    {
      "epoch": 0.33130852948620604,
      "grad_norm": 1.124320149421692,
      "learning_rate": 1.8999779010336204e-05,
      "loss": 0.6523,
      "step": 2618
    },
    {
      "epoch": 0.3314350797266515,
      "grad_norm": 1.1721012592315674,
      "learning_rate": 1.8998997615487825e-05,
      "loss": 0.6879,
      "step": 2619
    },
    {
      "epoch": 0.3315616299670969,
      "grad_norm": 1.0614389181137085,
      "learning_rate": 1.8998215931618008e-05,
      "loss": 0.64,
      "step": 2620
    },
    {
      "epoch": 0.33168818020754237,
      "grad_norm": 1.3070275783538818,
      "learning_rate": 1.899743395875187e-05,
      "loss": 0.677,
      "step": 2621
    },
    {
      "epoch": 0.33181473044798787,
      "grad_norm": 1.0810248851776123,
      "learning_rate": 1.8996651696914514e-05,
      "loss": 0.6418,
      "step": 2622
    },
    {
      "epoch": 0.3319412806884333,
      "grad_norm": 1.1543712615966797,
      "learning_rate": 1.8995869146131073e-05,
      "loss": 0.6598,
      "step": 2623
    },
    {
      "epoch": 0.33206783092887876,
      "grad_norm": 1.1596903800964355,
      "learning_rate": 1.8995086306426676e-05,
      "loss": 0.6808,
      "step": 2624
    },
    {
      "epoch": 0.3321943811693242,
      "grad_norm": 1.0884535312652588,
      "learning_rate": 1.8994303177826467e-05,
      "loss": 0.6987,
      "step": 2625
    },
    {
      "epoch": 0.3323209314097697,
      "grad_norm": 1.2069261074066162,
      "learning_rate": 1.8993519760355597e-05,
      "loss": 0.6888,
      "step": 2626
    },
    {
      "epoch": 0.33244748165021515,
      "grad_norm": 1.1123932600021362,
      "learning_rate": 1.8992736054039225e-05,
      "loss": 0.6914,
      "step": 2627
    },
    {
      "epoch": 0.3325740318906606,
      "grad_norm": 1.06263267993927,
      "learning_rate": 1.8991952058902528e-05,
      "loss": 0.6528,
      "step": 2628
    },
    {
      "epoch": 0.33270058213110604,
      "grad_norm": 1.1120414733886719,
      "learning_rate": 1.899116777497068e-05,
      "loss": 0.6524,
      "step": 2629
    },
    {
      "epoch": 0.3328271323715515,
      "grad_norm": 1.0540144443511963,
      "learning_rate": 1.8990383202268873e-05,
      "loss": 0.6495,
      "step": 2630
    },
    {
      "epoch": 0.332953682611997,
      "grad_norm": 1.0752556324005127,
      "learning_rate": 1.89895983408223e-05,
      "loss": 0.6746,
      "step": 2631
    },
    {
      "epoch": 0.3330802328524424,
      "grad_norm": 1.1828514337539673,
      "learning_rate": 1.8988813190656175e-05,
      "loss": 0.6805,
      "step": 2632
    },
    {
      "epoch": 0.33320678309288787,
      "grad_norm": 1.1061418056488037,
      "learning_rate": 1.898802775179571e-05,
      "loss": 0.7036,
      "step": 2633
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 1.0966936349868774,
      "learning_rate": 1.8987242024266134e-05,
      "loss": 0.6405,
      "step": 2634
    },
    {
      "epoch": 0.3334598835737788,
      "grad_norm": 1.0929055213928223,
      "learning_rate": 1.898645600809268e-05,
      "loss": 0.6714,
      "step": 2635
    },
    {
      "epoch": 0.33358643381422426,
      "grad_norm": 1.0913196802139282,
      "learning_rate": 1.8985669703300593e-05,
      "loss": 0.6684,
      "step": 2636
    },
    {
      "epoch": 0.3337129840546697,
      "grad_norm": 1.1435409784317017,
      "learning_rate": 1.8984883109915127e-05,
      "loss": 0.6715,
      "step": 2637
    },
    {
      "epoch": 0.33383953429511515,
      "grad_norm": 1.0885552167892456,
      "learning_rate": 1.898409622796155e-05,
      "loss": 0.6576,
      "step": 2638
    },
    {
      "epoch": 0.3339660845355606,
      "grad_norm": 1.0886621475219727,
      "learning_rate": 1.8983309057465124e-05,
      "loss": 0.6527,
      "step": 2639
    },
    {
      "epoch": 0.3340926347760061,
      "grad_norm": 1.1152197122573853,
      "learning_rate": 1.8982521598451136e-05,
      "loss": 0.695,
      "step": 2640
    },
    {
      "epoch": 0.33421918501645154,
      "grad_norm": 1.1300621032714844,
      "learning_rate": 1.898173385094488e-05,
      "loss": 0.7009,
      "step": 2641
    },
    {
      "epoch": 0.334345735256897,
      "grad_norm": 1.0654009580612183,
      "learning_rate": 1.8980945814971652e-05,
      "loss": 0.6608,
      "step": 2642
    },
    {
      "epoch": 0.3344722854973424,
      "grad_norm": 1.0521208047866821,
      "learning_rate": 1.8980157490556764e-05,
      "loss": 0.6322,
      "step": 2643
    },
    {
      "epoch": 0.3345988357377879,
      "grad_norm": 1.162584662437439,
      "learning_rate": 1.8979368877725534e-05,
      "loss": 0.644,
      "step": 2644
    },
    {
      "epoch": 0.33472538597823337,
      "grad_norm": 1.372947335243225,
      "learning_rate": 1.897857997650329e-05,
      "loss": 0.7053,
      "step": 2645
    },
    {
      "epoch": 0.3348519362186788,
      "grad_norm": 1.2150038480758667,
      "learning_rate": 1.8977790786915366e-05,
      "loss": 0.6569,
      "step": 2646
    },
    {
      "epoch": 0.33497848645912426,
      "grad_norm": 1.0911668539047241,
      "learning_rate": 1.897700130898711e-05,
      "loss": 0.6369,
      "step": 2647
    },
    {
      "epoch": 0.3351050366995697,
      "grad_norm": 1.1425471305847168,
      "learning_rate": 1.8976211542743882e-05,
      "loss": 0.7116,
      "step": 2648
    },
    {
      "epoch": 0.3352315869400152,
      "grad_norm": 1.1955746412277222,
      "learning_rate": 1.897542148821104e-05,
      "loss": 0.6537,
      "step": 2649
    },
    {
      "epoch": 0.33535813718046065,
      "grad_norm": 1.1624016761779785,
      "learning_rate": 1.8974631145413966e-05,
      "loss": 0.6591,
      "step": 2650
    },
    {
      "epoch": 0.3354846874209061,
      "grad_norm": 1.1661415100097656,
      "learning_rate": 1.8973840514378034e-05,
      "loss": 0.6876,
      "step": 2651
    },
    {
      "epoch": 0.33561123766135154,
      "grad_norm": 1.1356258392333984,
      "learning_rate": 1.8973049595128646e-05,
      "loss": 0.643,
      "step": 2652
    },
    {
      "epoch": 0.33573778790179704,
      "grad_norm": 1.176040768623352,
      "learning_rate": 1.89722583876912e-05,
      "loss": 0.6624,
      "step": 2653
    },
    {
      "epoch": 0.3358643381422425,
      "grad_norm": 1.1429955959320068,
      "learning_rate": 1.8971466892091105e-05,
      "loss": 0.6738,
      "step": 2654
    },
    {
      "epoch": 0.3359908883826879,
      "grad_norm": 1.2114317417144775,
      "learning_rate": 1.897067510835379e-05,
      "loss": 0.6846,
      "step": 2655
    },
    {
      "epoch": 0.33611743862313337,
      "grad_norm": 1.1597586870193481,
      "learning_rate": 1.8969883036504673e-05,
      "loss": 0.6417,
      "step": 2656
    },
    {
      "epoch": 0.3362439888635788,
      "grad_norm": 1.1367509365081787,
      "learning_rate": 1.8969090676569203e-05,
      "loss": 0.6709,
      "step": 2657
    },
    {
      "epoch": 0.3363705391040243,
      "grad_norm": 1.1446309089660645,
      "learning_rate": 1.896829802857282e-05,
      "loss": 0.657,
      "step": 2658
    },
    {
      "epoch": 0.33649708934446976,
      "grad_norm": 1.104920744895935,
      "learning_rate": 1.8967505092540987e-05,
      "loss": 0.668,
      "step": 2659
    },
    {
      "epoch": 0.3366236395849152,
      "grad_norm": 1.1271021366119385,
      "learning_rate": 1.896671186849917e-05,
      "loss": 0.671,
      "step": 2660
    },
    {
      "epoch": 0.33675018982536065,
      "grad_norm": 1.0712887048721313,
      "learning_rate": 1.8965918356472843e-05,
      "loss": 0.6575,
      "step": 2661
    },
    {
      "epoch": 0.33687674006580615,
      "grad_norm": 1.0816929340362549,
      "learning_rate": 1.8965124556487495e-05,
      "loss": 0.636,
      "step": 2662
    },
    {
      "epoch": 0.3370032903062516,
      "grad_norm": 1.2621643543243408,
      "learning_rate": 1.8964330468568617e-05,
      "loss": 0.6949,
      "step": 2663
    },
    {
      "epoch": 0.33712984054669703,
      "grad_norm": 1.096777081489563,
      "learning_rate": 1.8963536092741713e-05,
      "loss": 0.659,
      "step": 2664
    },
    {
      "epoch": 0.3372563907871425,
      "grad_norm": 1.1065999269485474,
      "learning_rate": 1.8962741429032297e-05,
      "loss": 0.705,
      "step": 2665
    },
    {
      "epoch": 0.337382941027588,
      "grad_norm": 1.1941393613815308,
      "learning_rate": 1.8961946477465894e-05,
      "loss": 0.6897,
      "step": 2666
    },
    {
      "epoch": 0.3375094912680334,
      "grad_norm": 1.1510530710220337,
      "learning_rate": 1.896115123806803e-05,
      "loss": 0.6955,
      "step": 2667
    },
    {
      "epoch": 0.33763604150847887,
      "grad_norm": 1.115094542503357,
      "learning_rate": 1.8960355710864248e-05,
      "loss": 0.6663,
      "step": 2668
    },
    {
      "epoch": 0.3377625917489243,
      "grad_norm": 1.1198387145996094,
      "learning_rate": 1.8959559895880102e-05,
      "loss": 0.6905,
      "step": 2669
    },
    {
      "epoch": 0.33788914198936976,
      "grad_norm": 1.1916743516921997,
      "learning_rate": 1.8958763793141144e-05,
      "loss": 0.6898,
      "step": 2670
    },
    {
      "epoch": 0.33801569222981526,
      "grad_norm": 1.160999059677124,
      "learning_rate": 1.895796740267295e-05,
      "loss": 0.6781,
      "step": 2671
    },
    {
      "epoch": 0.3381422424702607,
      "grad_norm": 1.1366519927978516,
      "learning_rate": 1.895717072450109e-05,
      "loss": 0.6492,
      "step": 2672
    },
    {
      "epoch": 0.33826879271070615,
      "grad_norm": 1.1206393241882324,
      "learning_rate": 1.8956373758651158e-05,
      "loss": 0.6202,
      "step": 2673
    },
    {
      "epoch": 0.3383953429511516,
      "grad_norm": 1.1358542442321777,
      "learning_rate": 1.895557650514875e-05,
      "loss": 0.6806,
      "step": 2674
    },
    {
      "epoch": 0.3385218931915971,
      "grad_norm": 1.125508427619934,
      "learning_rate": 1.8954778964019463e-05,
      "loss": 0.632,
      "step": 2675
    },
    {
      "epoch": 0.33864844343204253,
      "grad_norm": 1.2233844995498657,
      "learning_rate": 1.8953981135288915e-05,
      "loss": 0.6455,
      "step": 2676
    },
    {
      "epoch": 0.338774993672488,
      "grad_norm": 1.1739882230758667,
      "learning_rate": 1.8953183018982737e-05,
      "loss": 0.7001,
      "step": 2677
    },
    {
      "epoch": 0.3389015439129334,
      "grad_norm": 1.2931798696517944,
      "learning_rate": 1.895238461512656e-05,
      "loss": 0.7007,
      "step": 2678
    },
    {
      "epoch": 0.33902809415337887,
      "grad_norm": 1.087073564529419,
      "learning_rate": 1.895158592374602e-05,
      "loss": 0.6568,
      "step": 2679
    },
    {
      "epoch": 0.33915464439382437,
      "grad_norm": 1.4621703624725342,
      "learning_rate": 1.895078694486677e-05,
      "loss": 0.6755,
      "step": 2680
    },
    {
      "epoch": 0.3392811946342698,
      "grad_norm": 1.1064414978027344,
      "learning_rate": 1.8949987678514476e-05,
      "loss": 0.6527,
      "step": 2681
    },
    {
      "epoch": 0.33940774487471526,
      "grad_norm": 1.3941171169281006,
      "learning_rate": 1.8949188124714805e-05,
      "loss": 0.6977,
      "step": 2682
    },
    {
      "epoch": 0.3395342951151607,
      "grad_norm": 1.165744662284851,
      "learning_rate": 1.894838828349344e-05,
      "loss": 0.6995,
      "step": 2683
    },
    {
      "epoch": 0.3396608453556062,
      "grad_norm": 1.6009420156478882,
      "learning_rate": 1.8947588154876063e-05,
      "loss": 0.7326,
      "step": 2684
    },
    {
      "epoch": 0.33978739559605164,
      "grad_norm": 1.4050874710083008,
      "learning_rate": 1.8946787738888376e-05,
      "loss": 0.652,
      "step": 2685
    },
    {
      "epoch": 0.3399139458364971,
      "grad_norm": 1.4728556871414185,
      "learning_rate": 1.8945987035556084e-05,
      "loss": 0.6802,
      "step": 2686
    },
    {
      "epoch": 0.34004049607694253,
      "grad_norm": 1.3937196731567383,
      "learning_rate": 1.8945186044904904e-05,
      "loss": 0.6292,
      "step": 2687
    },
    {
      "epoch": 0.340167046317388,
      "grad_norm": 1.402928113937378,
      "learning_rate": 1.8944384766960562e-05,
      "loss": 0.7225,
      "step": 2688
    },
    {
      "epoch": 0.3402935965578335,
      "grad_norm": 1.6199166774749756,
      "learning_rate": 1.894358320174879e-05,
      "loss": 0.6902,
      "step": 2689
    },
    {
      "epoch": 0.3404201467982789,
      "grad_norm": 1.0562843084335327,
      "learning_rate": 1.894278134929534e-05,
      "loss": 0.6714,
      "step": 2690
    },
    {
      "epoch": 0.34054669703872437,
      "grad_norm": 1.4320658445358276,
      "learning_rate": 1.894197920962596e-05,
      "loss": 0.7206,
      "step": 2691
    },
    {
      "epoch": 0.3406732472791698,
      "grad_norm": 1.1347448825836182,
      "learning_rate": 1.8941176782766405e-05,
      "loss": 0.6501,
      "step": 2692
    },
    {
      "epoch": 0.3407997975196153,
      "grad_norm": 1.2868074178695679,
      "learning_rate": 1.894037406874246e-05,
      "loss": 0.6773,
      "step": 2693
    },
    {
      "epoch": 0.34092634776006076,
      "grad_norm": 1.1062426567077637,
      "learning_rate": 1.8939571067579895e-05,
      "loss": 0.6652,
      "step": 2694
    },
    {
      "epoch": 0.3410528980005062,
      "grad_norm": 1.2794101238250732,
      "learning_rate": 1.8938767779304507e-05,
      "loss": 0.6265,
      "step": 2695
    },
    {
      "epoch": 0.34117944824095164,
      "grad_norm": 1.05411958694458,
      "learning_rate": 1.893796420394209e-05,
      "loss": 0.6762,
      "step": 2696
    },
    {
      "epoch": 0.3413059984813971,
      "grad_norm": 1.2128822803497314,
      "learning_rate": 1.893716034151846e-05,
      "loss": 0.7158,
      "step": 2697
    },
    {
      "epoch": 0.3414325487218426,
      "grad_norm": 1.12364661693573,
      "learning_rate": 1.8936356192059425e-05,
      "loss": 0.6728,
      "step": 2698
    },
    {
      "epoch": 0.34155909896228803,
      "grad_norm": 1.2549161911010742,
      "learning_rate": 1.893555175559082e-05,
      "loss": 0.6488,
      "step": 2699
    },
    {
      "epoch": 0.3416856492027335,
      "grad_norm": 1.2018449306488037,
      "learning_rate": 1.8934747032138475e-05,
      "loss": 0.6641,
      "step": 2700
    },
    {
      "epoch": 0.3418121994431789,
      "grad_norm": 1.1499736309051514,
      "learning_rate": 1.8933942021728238e-05,
      "loss": 0.6854,
      "step": 2701
    },
    {
      "epoch": 0.3419387496836244,
      "grad_norm": 1.2001876831054688,
      "learning_rate": 1.8933136724385967e-05,
      "loss": 0.7081,
      "step": 2702
    },
    {
      "epoch": 0.34206529992406987,
      "grad_norm": 1.136047601699829,
      "learning_rate": 1.8932331140137518e-05,
      "loss": 0.6448,
      "step": 2703
    },
    {
      "epoch": 0.3421918501645153,
      "grad_norm": 1.2020750045776367,
      "learning_rate": 1.8931525269008775e-05,
      "loss": 0.6961,
      "step": 2704
    },
    {
      "epoch": 0.34231840040496075,
      "grad_norm": 1.1369059085845947,
      "learning_rate": 1.893071911102561e-05,
      "loss": 0.6601,
      "step": 2705
    },
    {
      "epoch": 0.3424449506454062,
      "grad_norm": 1.1432462930679321,
      "learning_rate": 1.8929912666213916e-05,
      "loss": 0.6619,
      "step": 2706
    },
    {
      "epoch": 0.3425715008858517,
      "grad_norm": 1.2749577760696411,
      "learning_rate": 1.89291059345996e-05,
      "loss": 0.6365,
      "step": 2707
    },
    {
      "epoch": 0.34269805112629714,
      "grad_norm": 1.1860865354537964,
      "learning_rate": 1.8928298916208567e-05,
      "loss": 0.6671,
      "step": 2708
    },
    {
      "epoch": 0.3428246013667426,
      "grad_norm": 1.1258915662765503,
      "learning_rate": 1.8927491611066736e-05,
      "loss": 0.6768,
      "step": 2709
    },
    {
      "epoch": 0.34295115160718803,
      "grad_norm": 1.3076982498168945,
      "learning_rate": 1.8926684019200037e-05,
      "loss": 0.7345,
      "step": 2710
    },
    {
      "epoch": 0.34307770184763353,
      "grad_norm": 1.1037375926971436,
      "learning_rate": 1.8925876140634404e-05,
      "loss": 0.6989,
      "step": 2711
    },
    {
      "epoch": 0.343204252088079,
      "grad_norm": 1.3273427486419678,
      "learning_rate": 1.892506797539579e-05,
      "loss": 0.737,
      "step": 2712
    },
    {
      "epoch": 0.3433308023285244,
      "grad_norm": 1.2671356201171875,
      "learning_rate": 1.8924259523510145e-05,
      "loss": 0.6973,
      "step": 2713
    },
    {
      "epoch": 0.34345735256896986,
      "grad_norm": 1.2170273065567017,
      "learning_rate": 1.8923450785003436e-05,
      "loss": 0.6763,
      "step": 2714
    },
    {
      "epoch": 0.34358390280941536,
      "grad_norm": 1.1017708778381348,
      "learning_rate": 1.892264175990164e-05,
      "loss": 0.6322,
      "step": 2715
    },
    {
      "epoch": 0.3437104530498608,
      "grad_norm": 1.5644844770431519,
      "learning_rate": 1.8921832448230737e-05,
      "loss": 0.6937,
      "step": 2716
    },
    {
      "epoch": 0.34383700329030625,
      "grad_norm": 1.2295260429382324,
      "learning_rate": 1.892102285001672e-05,
      "loss": 0.6421,
      "step": 2717
    },
    {
      "epoch": 0.3439635535307517,
      "grad_norm": 1.9988603591918945,
      "learning_rate": 1.8920212965285592e-05,
      "loss": 0.6892,
      "step": 2718
    },
    {
      "epoch": 0.34409010377119714,
      "grad_norm": 1.4714902639389038,
      "learning_rate": 1.8919402794063365e-05,
      "loss": 0.6606,
      "step": 2719
    },
    {
      "epoch": 0.34421665401164264,
      "grad_norm": 1.1301995515823364,
      "learning_rate": 1.8918592336376055e-05,
      "loss": 0.6343,
      "step": 2720
    },
    {
      "epoch": 0.3443432042520881,
      "grad_norm": 1.2612367868423462,
      "learning_rate": 1.8917781592249697e-05,
      "loss": 0.6519,
      "step": 2721
    },
    {
      "epoch": 0.34446975449253353,
      "grad_norm": 1.0977823734283447,
      "learning_rate": 1.891697056171033e-05,
      "loss": 0.6959,
      "step": 2722
    },
    {
      "epoch": 0.344596304732979,
      "grad_norm": 1.185202956199646,
      "learning_rate": 1.8916159244783995e-05,
      "loss": 0.6812,
      "step": 2723
    },
    {
      "epoch": 0.3447228549734245,
      "grad_norm": 1.0837106704711914,
      "learning_rate": 1.8915347641496754e-05,
      "loss": 0.6942,
      "step": 2724
    },
    {
      "epoch": 0.3448494052138699,
      "grad_norm": 1.1101964712142944,
      "learning_rate": 1.8914535751874673e-05,
      "loss": 0.6552,
      "step": 2725
    },
    {
      "epoch": 0.34497595545431536,
      "grad_norm": 1.0653594732284546,
      "learning_rate": 1.891372357594383e-05,
      "loss": 0.6512,
      "step": 2726
    },
    {
      "epoch": 0.3451025056947608,
      "grad_norm": 1.2294996976852417,
      "learning_rate": 1.8912911113730307e-05,
      "loss": 0.6806,
      "step": 2727
    },
    {
      "epoch": 0.34522905593520625,
      "grad_norm": 1.098734736442566,
      "learning_rate": 1.8912098365260196e-05,
      "loss": 0.6281,
      "step": 2728
    },
    {
      "epoch": 0.34535560617565175,
      "grad_norm": 1.1127207279205322,
      "learning_rate": 1.89112853305596e-05,
      "loss": 0.6655,
      "step": 2729
    },
    {
      "epoch": 0.3454821564160972,
      "grad_norm": 1.253007173538208,
      "learning_rate": 1.891047200965464e-05,
      "loss": 0.676,
      "step": 2730
    },
    {
      "epoch": 0.34560870665654264,
      "grad_norm": 1.1969695091247559,
      "learning_rate": 1.8909658402571427e-05,
      "loss": 0.642,
      "step": 2731
    },
    {
      "epoch": 0.3457352568969881,
      "grad_norm": 1.178655743598938,
      "learning_rate": 1.8908844509336095e-05,
      "loss": 0.6599,
      "step": 2732
    },
    {
      "epoch": 0.3458618071374336,
      "grad_norm": 1.2378122806549072,
      "learning_rate": 1.890803032997478e-05,
      "loss": 0.6425,
      "step": 2733
    },
    {
      "epoch": 0.34598835737787903,
      "grad_norm": 1.1257367134094238,
      "learning_rate": 1.8907215864513642e-05,
      "loss": 0.6512,
      "step": 2734
    },
    {
      "epoch": 0.3461149076183245,
      "grad_norm": 1.3557201623916626,
      "learning_rate": 1.8906401112978835e-05,
      "loss": 0.6826,
      "step": 2735
    },
    {
      "epoch": 0.3462414578587699,
      "grad_norm": 1.0794357061386108,
      "learning_rate": 1.890558607539652e-05,
      "loss": 0.6221,
      "step": 2736
    },
    {
      "epoch": 0.34636800809921536,
      "grad_norm": 1.3222334384918213,
      "learning_rate": 1.8904770751792875e-05,
      "loss": 0.685,
      "step": 2737
    },
    {
      "epoch": 0.34649455833966086,
      "grad_norm": 1.1301466226577759,
      "learning_rate": 1.890395514219409e-05,
      "loss": 0.7039,
      "step": 2738
    },
    {
      "epoch": 0.3466211085801063,
      "grad_norm": 1.4951961040496826,
      "learning_rate": 1.890313924662636e-05,
      "loss": 0.6696,
      "step": 2739
    },
    {
      "epoch": 0.34674765882055175,
      "grad_norm": 1.2405693531036377,
      "learning_rate": 1.8902323065115888e-05,
      "loss": 0.6611,
      "step": 2740
    },
    {
      "epoch": 0.3468742090609972,
      "grad_norm": 1.068050742149353,
      "learning_rate": 1.890150659768889e-05,
      "loss": 0.6367,
      "step": 2741
    },
    {
      "epoch": 0.3470007593014427,
      "grad_norm": 1.1129833459854126,
      "learning_rate": 1.890068984437158e-05,
      "loss": 0.6657,
      "step": 2742
    },
    {
      "epoch": 0.34712730954188814,
      "grad_norm": 1.0847632884979248,
      "learning_rate": 1.88998728051902e-05,
      "loss": 0.6651,
      "step": 2743
    },
    {
      "epoch": 0.3472538597823336,
      "grad_norm": 1.0535658597946167,
      "learning_rate": 1.889905548017098e-05,
      "loss": 0.5968,
      "step": 2744
    },
    {
      "epoch": 0.34738041002277903,
      "grad_norm": 1.085997223854065,
      "learning_rate": 1.889823786934018e-05,
      "loss": 0.6387,
      "step": 2745
    },
    {
      "epoch": 0.3475069602632245,
      "grad_norm": 1.0437999963760376,
      "learning_rate": 1.8897419972724056e-05,
      "loss": 0.6437,
      "step": 2746
    },
    {
      "epoch": 0.34763351050367,
      "grad_norm": 1.1121196746826172,
      "learning_rate": 1.8896601790348876e-05,
      "loss": 0.6984,
      "step": 2747
    },
    {
      "epoch": 0.3477600607441154,
      "grad_norm": 1.2738202810287476,
      "learning_rate": 1.8895783322240918e-05,
      "loss": 0.6875,
      "step": 2748
    },
    {
      "epoch": 0.34788661098456086,
      "grad_norm": 1.1313283443450928,
      "learning_rate": 1.8894964568426466e-05,
      "loss": 0.6581,
      "step": 2749
    },
    {
      "epoch": 0.3480131612250063,
      "grad_norm": 1.1433151960372925,
      "learning_rate": 1.889414552893182e-05,
      "loss": 0.697,
      "step": 2750
    },
    {
      "epoch": 0.3481397114654518,
      "grad_norm": 1.117059350013733,
      "learning_rate": 1.8893326203783285e-05,
      "loss": 0.6972,
      "step": 2751
    },
    {
      "epoch": 0.34826626170589725,
      "grad_norm": 1.0967841148376465,
      "learning_rate": 1.8892506593007172e-05,
      "loss": 0.6447,
      "step": 2752
    },
    {
      "epoch": 0.3483928119463427,
      "grad_norm": 1.1369096040725708,
      "learning_rate": 1.8891686696629805e-05,
      "loss": 0.7157,
      "step": 2753
    },
    {
      "epoch": 0.34851936218678814,
      "grad_norm": 1.1119064092636108,
      "learning_rate": 1.8890866514677523e-05,
      "loss": 0.6929,
      "step": 2754
    },
    {
      "epoch": 0.3486459124272336,
      "grad_norm": 1.1272499561309814,
      "learning_rate": 1.8890046047176658e-05,
      "loss": 0.7042,
      "step": 2755
    },
    {
      "epoch": 0.3487724626676791,
      "grad_norm": 1.110547661781311,
      "learning_rate": 1.8889225294153567e-05,
      "loss": 0.6354,
      "step": 2756
    },
    {
      "epoch": 0.34889901290812453,
      "grad_norm": 1.1098524332046509,
      "learning_rate": 1.8888404255634612e-05,
      "loss": 0.6786,
      "step": 2757
    },
    {
      "epoch": 0.34902556314857,
      "grad_norm": 1.2065773010253906,
      "learning_rate": 1.8887582931646154e-05,
      "loss": 0.6454,
      "step": 2758
    },
    {
      "epoch": 0.3491521133890154,
      "grad_norm": 1.1307262182235718,
      "learning_rate": 1.8886761322214583e-05,
      "loss": 0.6802,
      "step": 2759
    },
    {
      "epoch": 0.3492786636294609,
      "grad_norm": 1.2015904188156128,
      "learning_rate": 1.8885939427366277e-05,
      "loss": 0.6619,
      "step": 2760
    },
    {
      "epoch": 0.34940521386990636,
      "grad_norm": 1.1568855047225952,
      "learning_rate": 1.8885117247127635e-05,
      "loss": 0.6584,
      "step": 2761
    },
    {
      "epoch": 0.3495317641103518,
      "grad_norm": 1.1914006471633911,
      "learning_rate": 1.888429478152507e-05,
      "loss": 0.6619,
      "step": 2762
    },
    {
      "epoch": 0.34965831435079725,
      "grad_norm": 1.1677125692367554,
      "learning_rate": 1.8883472030584987e-05,
      "loss": 0.6883,
      "step": 2763
    },
    {
      "epoch": 0.34978486459124275,
      "grad_norm": 1.087209701538086,
      "learning_rate": 1.8882648994333816e-05,
      "loss": 0.6323,
      "step": 2764
    },
    {
      "epoch": 0.3499114148316882,
      "grad_norm": 1.2281850576400757,
      "learning_rate": 1.888182567279799e-05,
      "loss": 0.7242,
      "step": 2765
    },
    {
      "epoch": 0.35003796507213364,
      "grad_norm": 1.0646789073944092,
      "learning_rate": 1.888100206600395e-05,
      "loss": 0.7023,
      "step": 2766
    },
    {
      "epoch": 0.3501645153125791,
      "grad_norm": 1.150014877319336,
      "learning_rate": 1.8880178173978154e-05,
      "loss": 0.6929,
      "step": 2767
    },
    {
      "epoch": 0.3502910655530245,
      "grad_norm": 1.1031701564788818,
      "learning_rate": 1.8879353996747053e-05,
      "loss": 0.6764,
      "step": 2768
    },
    {
      "epoch": 0.35041761579347,
      "grad_norm": 1.0327949523925781,
      "learning_rate": 1.8878529534337123e-05,
      "loss": 0.6687,
      "step": 2769
    },
    {
      "epoch": 0.35054416603391547,
      "grad_norm": 1.081382393836975,
      "learning_rate": 1.8877704786774845e-05,
      "loss": 0.6551,
      "step": 2770
    },
    {
      "epoch": 0.3506707162743609,
      "grad_norm": 1.1102046966552734,
      "learning_rate": 1.8876879754086702e-05,
      "loss": 0.6613,
      "step": 2771
    },
    {
      "epoch": 0.35079726651480636,
      "grad_norm": 1.2041163444519043,
      "learning_rate": 1.8876054436299195e-05,
      "loss": 0.6933,
      "step": 2772
    },
    {
      "epoch": 0.35092381675525186,
      "grad_norm": 1.1177877187728882,
      "learning_rate": 1.8875228833438833e-05,
      "loss": 0.6758,
      "step": 2773
    },
    {
      "epoch": 0.3510503669956973,
      "grad_norm": 1.1661721467971802,
      "learning_rate": 1.8874402945532127e-05,
      "loss": 0.6878,
      "step": 2774
    },
    {
      "epoch": 0.35117691723614275,
      "grad_norm": 1.1396093368530273,
      "learning_rate": 1.8873576772605605e-05,
      "loss": 0.6439,
      "step": 2775
    },
    {
      "epoch": 0.3513034674765882,
      "grad_norm": 1.1720246076583862,
      "learning_rate": 1.88727503146858e-05,
      "loss": 0.6824,
      "step": 2776
    },
    {
      "epoch": 0.35143001771703364,
      "grad_norm": 1.217507243156433,
      "learning_rate": 1.887192357179926e-05,
      "loss": 0.6681,
      "step": 2777
    },
    {
      "epoch": 0.35155656795747914,
      "grad_norm": 1.1114476919174194,
      "learning_rate": 1.887109654397253e-05,
      "loss": 0.6958,
      "step": 2778
    },
    {
      "epoch": 0.3516831181979246,
      "grad_norm": 1.0438114404678345,
      "learning_rate": 1.8870269231232178e-05,
      "loss": 0.6418,
      "step": 2779
    },
    {
      "epoch": 0.35180966843837,
      "grad_norm": 1.1088560819625854,
      "learning_rate": 1.886944163360477e-05,
      "loss": 0.6377,
      "step": 2780
    },
    {
      "epoch": 0.35193621867881547,
      "grad_norm": 1.1321133375167847,
      "learning_rate": 1.8868613751116888e-05,
      "loss": 0.6551,
      "step": 2781
    },
    {
      "epoch": 0.35206276891926097,
      "grad_norm": 1.1099997758865356,
      "learning_rate": 1.8867785583795122e-05,
      "loss": 0.6541,
      "step": 2782
    },
    {
      "epoch": 0.3521893191597064,
      "grad_norm": 1.2128396034240723,
      "learning_rate": 1.8866957131666067e-05,
      "loss": 0.7133,
      "step": 2783
    },
    {
      "epoch": 0.35231586940015186,
      "grad_norm": 1.1000994443893433,
      "learning_rate": 1.886612839475634e-05,
      "loss": 0.643,
      "step": 2784
    },
    {
      "epoch": 0.3524424196405973,
      "grad_norm": 1.1087950468063354,
      "learning_rate": 1.8865299373092546e-05,
      "loss": 0.6657,
      "step": 2785
    },
    {
      "epoch": 0.35256896988104275,
      "grad_norm": 1.175474762916565,
      "learning_rate": 1.8864470066701317e-05,
      "loss": 0.6618,
      "step": 2786
    },
    {
      "epoch": 0.35269552012148825,
      "grad_norm": 1.1561086177825928,
      "learning_rate": 1.8863640475609287e-05,
      "loss": 0.6881,
      "step": 2787
    },
    {
      "epoch": 0.3528220703619337,
      "grad_norm": 1.0654094219207764,
      "learning_rate": 1.8862810599843094e-05,
      "loss": 0.6777,
      "step": 2788
    },
    {
      "epoch": 0.35294862060237914,
      "grad_norm": 1.347825050354004,
      "learning_rate": 1.8861980439429402e-05,
      "loss": 0.6838,
      "step": 2789
    },
    {
      "epoch": 0.3530751708428246,
      "grad_norm": 1.089776873588562,
      "learning_rate": 1.886114999439487e-05,
      "loss": 0.6652,
      "step": 2790
    },
    {
      "epoch": 0.3532017210832701,
      "grad_norm": 1.0986690521240234,
      "learning_rate": 1.8860319264766163e-05,
      "loss": 0.6418,
      "step": 2791
    },
    {
      "epoch": 0.3533282713237155,
      "grad_norm": 1.179577350616455,
      "learning_rate": 1.8859488250569966e-05,
      "loss": 0.6626,
      "step": 2792
    },
    {
      "epoch": 0.35345482156416097,
      "grad_norm": 1.003167748451233,
      "learning_rate": 1.8858656951832972e-05,
      "loss": 0.6167,
      "step": 2793
    },
    {
      "epoch": 0.3535813718046064,
      "grad_norm": 1.0756618976593018,
      "learning_rate": 1.8857825368581873e-05,
      "loss": 0.6771,
      "step": 2794
    },
    {
      "epoch": 0.35370792204505186,
      "grad_norm": 1.1905122995376587,
      "learning_rate": 1.8856993500843382e-05,
      "loss": 0.6658,
      "step": 2795
    },
    {
      "epoch": 0.35383447228549736,
      "grad_norm": 1.1605842113494873,
      "learning_rate": 1.8856161348644217e-05,
      "loss": 0.6589,
      "step": 2796
    },
    {
      "epoch": 0.3539610225259428,
      "grad_norm": 1.0856819152832031,
      "learning_rate": 1.88553289120111e-05,
      "loss": 0.658,
      "step": 2797
    },
    {
      "epoch": 0.35408757276638825,
      "grad_norm": 1.5580005645751953,
      "learning_rate": 1.885449619097077e-05,
      "loss": 0.7248,
      "step": 2798
    },
    {
      "epoch": 0.3542141230068337,
      "grad_norm": 1.1172577142715454,
      "learning_rate": 1.885366318554997e-05,
      "loss": 0.674,
      "step": 2799
    },
    {
      "epoch": 0.3543406732472792,
      "grad_norm": 1.018964409828186,
      "learning_rate": 1.8852829895775454e-05,
      "loss": 0.6271,
      "step": 2800
    },
    {
      "epoch": 0.35446722348772464,
      "grad_norm": 1.2653615474700928,
      "learning_rate": 1.8851996321673983e-05,
      "loss": 0.7111,
      "step": 2801
    },
    {
      "epoch": 0.3545937737281701,
      "grad_norm": 1.1818190813064575,
      "learning_rate": 1.885116246327233e-05,
      "loss": 0.6524,
      "step": 2802
    },
    {
      "epoch": 0.3547203239686155,
      "grad_norm": 1.1676102876663208,
      "learning_rate": 1.8850328320597276e-05,
      "loss": 0.6463,
      "step": 2803
    },
    {
      "epoch": 0.35484687420906097,
      "grad_norm": 1.1466397047042847,
      "learning_rate": 1.8849493893675617e-05,
      "loss": 0.6754,
      "step": 2804
    },
    {
      "epoch": 0.35497342444950647,
      "grad_norm": 1.1715682744979858,
      "learning_rate": 1.8848659182534142e-05,
      "loss": 0.6734,
      "step": 2805
    },
    {
      "epoch": 0.3550999746899519,
      "grad_norm": 1.1023685932159424,
      "learning_rate": 1.8847824187199667e-05,
      "loss": 0.6849,
      "step": 2806
    },
    {
      "epoch": 0.35522652493039736,
      "grad_norm": 1.0526208877563477,
      "learning_rate": 1.884698890769901e-05,
      "loss": 0.6826,
      "step": 2807
    },
    {
      "epoch": 0.3553530751708428,
      "grad_norm": 1.243398666381836,
      "learning_rate": 1.884615334405899e-05,
      "loss": 0.6661,
      "step": 2808
    },
    {
      "epoch": 0.3554796254112883,
      "grad_norm": 1.0376516580581665,
      "learning_rate": 1.884531749630645e-05,
      "loss": 0.6854,
      "step": 2809
    },
    {
      "epoch": 0.35560617565173375,
      "grad_norm": 1.0896108150482178,
      "learning_rate": 1.884448136446823e-05,
      "loss": 0.6207,
      "step": 2810
    },
    {
      "epoch": 0.3557327258921792,
      "grad_norm": 1.1433814764022827,
      "learning_rate": 1.8843644948571195e-05,
      "loss": 0.7212,
      "step": 2811
    },
    {
      "epoch": 0.35585927613262464,
      "grad_norm": 1.096231460571289,
      "learning_rate": 1.8842808248642192e-05,
      "loss": 0.6484,
      "step": 2812
    },
    {
      "epoch": 0.35598582637307014,
      "grad_norm": 1.0716303586959839,
      "learning_rate": 1.8841971264708107e-05,
      "loss": 0.6632,
      "step": 2813
    },
    {
      "epoch": 0.3561123766135156,
      "grad_norm": 1.1107029914855957,
      "learning_rate": 1.8841133996795815e-05,
      "loss": 0.6636,
      "step": 2814
    },
    {
      "epoch": 0.356238926853961,
      "grad_norm": 1.1322418451309204,
      "learning_rate": 1.8840296444932203e-05,
      "loss": 0.7166,
      "step": 2815
    },
    {
      "epoch": 0.35636547709440647,
      "grad_norm": 1.1781336069107056,
      "learning_rate": 1.8839458609144178e-05,
      "loss": 0.661,
      "step": 2816
    },
    {
      "epoch": 0.3564920273348519,
      "grad_norm": 1.1010043621063232,
      "learning_rate": 1.8838620489458647e-05,
      "loss": 0.6629,
      "step": 2817
    },
    {
      "epoch": 0.3566185775752974,
      "grad_norm": 1.1558082103729248,
      "learning_rate": 1.8837782085902526e-05,
      "loss": 0.6586,
      "step": 2818
    },
    {
      "epoch": 0.35674512781574286,
      "grad_norm": 1.0713602304458618,
      "learning_rate": 1.8836943398502744e-05,
      "loss": 0.6712,
      "step": 2819
    },
    {
      "epoch": 0.3568716780561883,
      "grad_norm": 1.230086326599121,
      "learning_rate": 1.8836104427286236e-05,
      "loss": 0.6617,
      "step": 2820
    },
    {
      "epoch": 0.35699822829663375,
      "grad_norm": 1.048031210899353,
      "learning_rate": 1.8835265172279952e-05,
      "loss": 0.6586,
      "step": 2821
    },
    {
      "epoch": 0.35712477853707925,
      "grad_norm": 1.2796345949172974,
      "learning_rate": 1.8834425633510838e-05,
      "loss": 0.6379,
      "step": 2822
    },
    {
      "epoch": 0.3572513287775247,
      "grad_norm": 1.0225762128829956,
      "learning_rate": 1.883358581100586e-05,
      "loss": 0.6663,
      "step": 2823
    },
    {
      "epoch": 0.35737787901797013,
      "grad_norm": 1.2033003568649292,
      "learning_rate": 1.8832745704791992e-05,
      "loss": 0.6338,
      "step": 2824
    },
    {
      "epoch": 0.3575044292584156,
      "grad_norm": 1.1470179557800293,
      "learning_rate": 1.8831905314896223e-05,
      "loss": 0.6594,
      "step": 2825
    },
    {
      "epoch": 0.357630979498861,
      "grad_norm": 1.1921502351760864,
      "learning_rate": 1.883106464134553e-05,
      "loss": 0.6995,
      "step": 2826
    },
    {
      "epoch": 0.3577575297393065,
      "grad_norm": 1.099787712097168,
      "learning_rate": 1.883022368416692e-05,
      "loss": 0.6836,
      "step": 2827
    },
    {
      "epoch": 0.35788407997975197,
      "grad_norm": 1.21066415309906,
      "learning_rate": 1.8829382443387403e-05,
      "loss": 0.6606,
      "step": 2828
    },
    {
      "epoch": 0.3580106302201974,
      "grad_norm": 1.1296007633209229,
      "learning_rate": 1.8828540919033997e-05,
      "loss": 0.6391,
      "step": 2829
    },
    {
      "epoch": 0.35813718046064286,
      "grad_norm": 1.0939797163009644,
      "learning_rate": 1.8827699111133728e-05,
      "loss": 0.6797,
      "step": 2830
    },
    {
      "epoch": 0.35826373070108836,
      "grad_norm": 1.1967202425003052,
      "learning_rate": 1.8826857019713635e-05,
      "loss": 0.699,
      "step": 2831
    },
    {
      "epoch": 0.3583902809415338,
      "grad_norm": 1.1733694076538086,
      "learning_rate": 1.8826014644800755e-05,
      "loss": 0.6568,
      "step": 2832
    },
    {
      "epoch": 0.35851683118197925,
      "grad_norm": 1.1151119470596313,
      "learning_rate": 1.8825171986422154e-05,
      "loss": 0.691,
      "step": 2833
    },
    {
      "epoch": 0.3586433814224247,
      "grad_norm": 1.2306156158447266,
      "learning_rate": 1.882432904460489e-05,
      "loss": 0.6708,
      "step": 2834
    },
    {
      "epoch": 0.35876993166287013,
      "grad_norm": 1.1065365076065063,
      "learning_rate": 1.8823485819376037e-05,
      "loss": 0.6965,
      "step": 2835
    },
    {
      "epoch": 0.35889648190331563,
      "grad_norm": 1.2552039623260498,
      "learning_rate": 1.8822642310762673e-05,
      "loss": 0.6791,
      "step": 2836
    },
    {
      "epoch": 0.3590230321437611,
      "grad_norm": 1.1441631317138672,
      "learning_rate": 1.8821798518791896e-05,
      "loss": 0.6603,
      "step": 2837
    },
    {
      "epoch": 0.3591495823842065,
      "grad_norm": 1.143978476524353,
      "learning_rate": 1.88209544434908e-05,
      "loss": 0.697,
      "step": 2838
    },
    {
      "epoch": 0.35927613262465197,
      "grad_norm": 1.201582431793213,
      "learning_rate": 1.8820110084886498e-05,
      "loss": 0.6376,
      "step": 2839
    },
    {
      "epoch": 0.35940268286509747,
      "grad_norm": 1.200173258781433,
      "learning_rate": 1.881926544300611e-05,
      "loss": 0.7203,
      "step": 2840
    },
    {
      "epoch": 0.3595292331055429,
      "grad_norm": 1.1302556991577148,
      "learning_rate": 1.8818420517876754e-05,
      "loss": 0.7076,
      "step": 2841
    },
    {
      "epoch": 0.35965578334598836,
      "grad_norm": 1.171454906463623,
      "learning_rate": 1.881757530952558e-05,
      "loss": 0.6813,
      "step": 2842
    },
    {
      "epoch": 0.3597823335864338,
      "grad_norm": 1.164805293083191,
      "learning_rate": 1.8816729817979722e-05,
      "loss": 0.6526,
      "step": 2843
    },
    {
      "epoch": 0.35990888382687924,
      "grad_norm": 1.2544326782226562,
      "learning_rate": 1.8815884043266338e-05,
      "loss": 0.7103,
      "step": 2844
    },
    {
      "epoch": 0.36003543406732474,
      "grad_norm": 1.1308091878890991,
      "learning_rate": 1.8815037985412598e-05,
      "loss": 0.7011,
      "step": 2845
    },
    {
      "epoch": 0.3601619843077702,
      "grad_norm": 1.0905157327651978,
      "learning_rate": 1.8814191644445667e-05,
      "loss": 0.7001,
      "step": 2846
    },
    {
      "epoch": 0.36028853454821563,
      "grad_norm": 1.068256139755249,
      "learning_rate": 1.881334502039273e-05,
      "loss": 0.6328,
      "step": 2847
    },
    {
      "epoch": 0.3604150847886611,
      "grad_norm": 1.1309993267059326,
      "learning_rate": 1.881249811328098e-05,
      "loss": 0.6625,
      "step": 2848
    },
    {
      "epoch": 0.3605416350291066,
      "grad_norm": 1.026354432106018,
      "learning_rate": 1.8811650923137616e-05,
      "loss": 0.6395,
      "step": 2849
    },
    {
      "epoch": 0.360668185269552,
      "grad_norm": 1.1987693309783936,
      "learning_rate": 1.881080344998984e-05,
      "loss": 0.6344,
      "step": 2850
    },
    {
      "epoch": 0.36079473550999747,
      "grad_norm": 1.0821301937103271,
      "learning_rate": 1.8809955693864882e-05,
      "loss": 0.6048,
      "step": 2851
    },
    {
      "epoch": 0.3609212857504429,
      "grad_norm": 1.134945273399353,
      "learning_rate": 1.8809107654789965e-05,
      "loss": 0.6334,
      "step": 2852
    },
    {
      "epoch": 0.36104783599088835,
      "grad_norm": 1.0829273462295532,
      "learning_rate": 1.8808259332792325e-05,
      "loss": 0.6523,
      "step": 2853
    },
    {
      "epoch": 0.36117438623133385,
      "grad_norm": 1.1467435359954834,
      "learning_rate": 1.8807410727899206e-05,
      "loss": 0.6857,
      "step": 2854
    },
    {
      "epoch": 0.3613009364717793,
      "grad_norm": 1.1503900289535522,
      "learning_rate": 1.8806561840137862e-05,
      "loss": 0.6793,
      "step": 2855
    },
    {
      "epoch": 0.36142748671222474,
      "grad_norm": 1.1948004961013794,
      "learning_rate": 1.880571266953556e-05,
      "loss": 0.6949,
      "step": 2856
    },
    {
      "epoch": 0.3615540369526702,
      "grad_norm": 1.1748077869415283,
      "learning_rate": 1.8804863216119574e-05,
      "loss": 0.6295,
      "step": 2857
    },
    {
      "epoch": 0.3616805871931157,
      "grad_norm": 1.0759912729263306,
      "learning_rate": 1.8804013479917186e-05,
      "loss": 0.6579,
      "step": 2858
    },
    {
      "epoch": 0.36180713743356113,
      "grad_norm": 1.1248908042907715,
      "learning_rate": 1.880316346095568e-05,
      "loss": 0.6425,
      "step": 2859
    },
    {
      "epoch": 0.3619336876740066,
      "grad_norm": 1.065864086151123,
      "learning_rate": 1.8802313159262364e-05,
      "loss": 0.6473,
      "step": 2860
    },
    {
      "epoch": 0.362060237914452,
      "grad_norm": 1.2371751070022583,
      "learning_rate": 1.8801462574864543e-05,
      "loss": 0.6446,
      "step": 2861
    },
    {
      "epoch": 0.3621867881548975,
      "grad_norm": 1.0983498096466064,
      "learning_rate": 1.8800611707789537e-05,
      "loss": 0.6748,
      "step": 2862
    },
    {
      "epoch": 0.36231333839534297,
      "grad_norm": 1.27419912815094,
      "learning_rate": 1.8799760558064672e-05,
      "loss": 0.6757,
      "step": 2863
    },
    {
      "epoch": 0.3624398886357884,
      "grad_norm": 1.1012303829193115,
      "learning_rate": 1.8798909125717286e-05,
      "loss": 0.6591,
      "step": 2864
    },
    {
      "epoch": 0.36256643887623385,
      "grad_norm": 1.0924758911132812,
      "learning_rate": 1.879805741077472e-05,
      "loss": 0.6638,
      "step": 2865
    },
    {
      "epoch": 0.3626929891166793,
      "grad_norm": 1.031409502029419,
      "learning_rate": 1.879720541326434e-05,
      "loss": 0.6579,
      "step": 2866
    },
    {
      "epoch": 0.3628195393571248,
      "grad_norm": 1.1194196939468384,
      "learning_rate": 1.8796353133213495e-05,
      "loss": 0.6997,
      "step": 2867
    },
    {
      "epoch": 0.36294608959757024,
      "grad_norm": 1.042445421218872,
      "learning_rate": 1.879550057064957e-05,
      "loss": 0.6898,
      "step": 2868
    },
    {
      "epoch": 0.3630726398380157,
      "grad_norm": 1.0526130199432373,
      "learning_rate": 1.8794647725599935e-05,
      "loss": 0.6315,
      "step": 2869
    },
    {
      "epoch": 0.36319919007846113,
      "grad_norm": 1.0614184141159058,
      "learning_rate": 1.8793794598091993e-05,
      "loss": 0.6602,
      "step": 2870
    },
    {
      "epoch": 0.36332574031890663,
      "grad_norm": 1.14036226272583,
      "learning_rate": 1.8792941188153136e-05,
      "loss": 0.6574,
      "step": 2871
    },
    {
      "epoch": 0.3634522905593521,
      "grad_norm": 1.0242919921875,
      "learning_rate": 1.8792087495810777e-05,
      "loss": 0.6305,
      "step": 2872
    },
    {
      "epoch": 0.3635788407997975,
      "grad_norm": 1.1315464973449707,
      "learning_rate": 1.8791233521092332e-05,
      "loss": 0.6926,
      "step": 2873
    },
    {
      "epoch": 0.36370539104024296,
      "grad_norm": 1.1324976682662964,
      "learning_rate": 1.8790379264025226e-05,
      "loss": 0.6455,
      "step": 2874
    },
    {
      "epoch": 0.3638319412806884,
      "grad_norm": 1.0778757333755493,
      "learning_rate": 1.87895247246369e-05,
      "loss": 0.6474,
      "step": 2875
    },
    {
      "epoch": 0.3639584915211339,
      "grad_norm": 1.0934447050094604,
      "learning_rate": 1.8788669902954798e-05,
      "loss": 0.6889,
      "step": 2876
    },
    {
      "epoch": 0.36408504176157935,
      "grad_norm": 1.0694963932037354,
      "learning_rate": 1.8787814799006374e-05,
      "loss": 0.6677,
      "step": 2877
    },
    {
      "epoch": 0.3642115920020248,
      "grad_norm": 1.174481749534607,
      "learning_rate": 1.878695941281909e-05,
      "loss": 0.6779,
      "step": 2878
    },
    {
      "epoch": 0.36433814224247024,
      "grad_norm": 1.0583173036575317,
      "learning_rate": 1.878610374442042e-05,
      "loss": 0.669,
      "step": 2879
    },
    {
      "epoch": 0.36446469248291574,
      "grad_norm": 1.1533608436584473,
      "learning_rate": 1.8785247793837845e-05,
      "loss": 0.6703,
      "step": 2880
    },
    {
      "epoch": 0.3645912427233612,
      "grad_norm": 1.2395575046539307,
      "learning_rate": 1.8784391561098854e-05,
      "loss": 0.648,
      "step": 2881
    },
    {
      "epoch": 0.36471779296380663,
      "grad_norm": 1.0346198081970215,
      "learning_rate": 1.878353504623095e-05,
      "loss": 0.6549,
      "step": 2882
    },
    {
      "epoch": 0.3648443432042521,
      "grad_norm": 1.237114429473877,
      "learning_rate": 1.878267824926164e-05,
      "loss": 0.7007,
      "step": 2883
    },
    {
      "epoch": 0.3649708934446975,
      "grad_norm": 1.0982191562652588,
      "learning_rate": 1.878182117021844e-05,
      "loss": 0.7148,
      "step": 2884
    },
    {
      "epoch": 0.365097443685143,
      "grad_norm": 1.0487560033798218,
      "learning_rate": 1.878096380912888e-05,
      "loss": 0.6507,
      "step": 2885
    },
    {
      "epoch": 0.36522399392558846,
      "grad_norm": 1.1363095045089722,
      "learning_rate": 1.8780106166020497e-05,
      "loss": 0.6714,
      "step": 2886
    },
    {
      "epoch": 0.3653505441660339,
      "grad_norm": 1.0340757369995117,
      "learning_rate": 1.8779248240920832e-05,
      "loss": 0.6321,
      "step": 2887
    },
    {
      "epoch": 0.36547709440647935,
      "grad_norm": 1.0443366765975952,
      "learning_rate": 1.877839003385744e-05,
      "loss": 0.6399,
      "step": 2888
    },
    {
      "epoch": 0.36560364464692485,
      "grad_norm": 1.0302683115005493,
      "learning_rate": 1.8777531544857886e-05,
      "loss": 0.6905,
      "step": 2889
    },
    {
      "epoch": 0.3657301948873703,
      "grad_norm": 1.0848406553268433,
      "learning_rate": 1.877667277394974e-05,
      "loss": 0.6373,
      "step": 2890
    },
    {
      "epoch": 0.36585674512781574,
      "grad_norm": 1.023097038269043,
      "learning_rate": 1.877581372116058e-05,
      "loss": 0.65,
      "step": 2891
    },
    {
      "epoch": 0.3659832953682612,
      "grad_norm": 1.2021889686584473,
      "learning_rate": 1.877495438651801e-05,
      "loss": 0.6772,
      "step": 2892
    },
    {
      "epoch": 0.36610984560870663,
      "grad_norm": 1.085697889328003,
      "learning_rate": 1.8774094770049613e-05,
      "loss": 0.6795,
      "step": 2893
    },
    {
      "epoch": 0.36623639584915213,
      "grad_norm": 1.137204885482788,
      "learning_rate": 1.8773234871783007e-05,
      "loss": 0.6104,
      "step": 2894
    },
    {
      "epoch": 0.3663629460895976,
      "grad_norm": 1.2504000663757324,
      "learning_rate": 1.8772374691745806e-05,
      "loss": 0.7019,
      "step": 2895
    },
    {
      "epoch": 0.366489496330043,
      "grad_norm": 1.0388952493667603,
      "learning_rate": 1.8771514229965637e-05,
      "loss": 0.6697,
      "step": 2896
    },
    {
      "epoch": 0.36661604657048846,
      "grad_norm": 1.154547929763794,
      "learning_rate": 1.8770653486470135e-05,
      "loss": 0.6728,
      "step": 2897
    },
    {
      "epoch": 0.36674259681093396,
      "grad_norm": 1.069966197013855,
      "learning_rate": 1.8769792461286946e-05,
      "loss": 0.6633,
      "step": 2898
    },
    {
      "epoch": 0.3668691470513794,
      "grad_norm": 1.182846188545227,
      "learning_rate": 1.876893115444372e-05,
      "loss": 0.6946,
      "step": 2899
    },
    {
      "epoch": 0.36699569729182485,
      "grad_norm": 1.060046672821045,
      "learning_rate": 1.8768069565968123e-05,
      "loss": 0.6562,
      "step": 2900
    },
    {
      "epoch": 0.3671222475322703,
      "grad_norm": 1.1423090696334839,
      "learning_rate": 1.8767207695887827e-05,
      "loss": 0.7251,
      "step": 2901
    },
    {
      "epoch": 0.36724879777271574,
      "grad_norm": 1.128447413444519,
      "learning_rate": 1.8766345544230513e-05,
      "loss": 0.7205,
      "step": 2902
    },
    {
      "epoch": 0.36737534801316124,
      "grad_norm": 1.1993446350097656,
      "learning_rate": 1.8765483111023866e-05,
      "loss": 0.6762,
      "step": 2903
    },
    {
      "epoch": 0.3675018982536067,
      "grad_norm": 1.0524110794067383,
      "learning_rate": 1.876462039629559e-05,
      "loss": 0.6059,
      "step": 2904
    },
    {
      "epoch": 0.36762844849405213,
      "grad_norm": 1.0869832038879395,
      "learning_rate": 1.8763757400073394e-05,
      "loss": 0.6448,
      "step": 2905
    },
    {
      "epoch": 0.3677549987344976,
      "grad_norm": 0.989641010761261,
      "learning_rate": 1.8762894122384992e-05,
      "loss": 0.6305,
      "step": 2906
    },
    {
      "epoch": 0.3678815489749431,
      "grad_norm": 1.0234320163726807,
      "learning_rate": 1.8762030563258105e-05,
      "loss": 0.6845,
      "step": 2907
    },
    {
      "epoch": 0.3680080992153885,
      "grad_norm": 1.089969515800476,
      "learning_rate": 1.8761166722720477e-05,
      "loss": 0.6317,
      "step": 2908
    },
    {
      "epoch": 0.36813464945583396,
      "grad_norm": 1.0780372619628906,
      "learning_rate": 1.876030260079985e-05,
      "loss": 0.6454,
      "step": 2909
    },
    {
      "epoch": 0.3682611996962794,
      "grad_norm": 1.0337499380111694,
      "learning_rate": 1.8759438197523973e-05,
      "loss": 0.6344,
      "step": 2910
    },
    {
      "epoch": 0.3683877499367249,
      "grad_norm": 1.1309664249420166,
      "learning_rate": 1.8758573512920608e-05,
      "loss": 0.6776,
      "step": 2911
    },
    {
      "epoch": 0.36851430017717035,
      "grad_norm": 1.0919432640075684,
      "learning_rate": 1.8757708547017533e-05,
      "loss": 0.6304,
      "step": 2912
    },
    {
      "epoch": 0.3686408504176158,
      "grad_norm": 1.2873797416687012,
      "learning_rate": 1.875684329984252e-05,
      "loss": 0.6707,
      "step": 2913
    },
    {
      "epoch": 0.36876740065806124,
      "grad_norm": 1.063444972038269,
      "learning_rate": 1.8755977771423363e-05,
      "loss": 0.6408,
      "step": 2914
    },
    {
      "epoch": 0.3688939508985067,
      "grad_norm": 1.18365478515625,
      "learning_rate": 1.875511196178786e-05,
      "loss": 0.6011,
      "step": 2915
    },
    {
      "epoch": 0.3690205011389522,
      "grad_norm": 1.1556155681610107,
      "learning_rate": 1.8754245870963814e-05,
      "loss": 0.6837,
      "step": 2916
    },
    {
      "epoch": 0.36914705137939763,
      "grad_norm": 2.062662124633789,
      "learning_rate": 1.8753379498979047e-05,
      "loss": 0.6904,
      "step": 2917
    },
    {
      "epoch": 0.3692736016198431,
      "grad_norm": 1.3646788597106934,
      "learning_rate": 1.8752512845861382e-05,
      "loss": 0.6415,
      "step": 2918
    },
    {
      "epoch": 0.3694001518602885,
      "grad_norm": 1.0888125896453857,
      "learning_rate": 1.8751645911638656e-05,
      "loss": 0.7302,
      "step": 2919
    },
    {
      "epoch": 0.369526702100734,
      "grad_norm": 1.3719323873519897,
      "learning_rate": 1.8750778696338705e-05,
      "loss": 0.708,
      "step": 2920
    },
    {
      "epoch": 0.36965325234117946,
      "grad_norm": 1.1359347105026245,
      "learning_rate": 1.8749911199989386e-05,
      "loss": 0.7011,
      "step": 2921
    },
    {
      "epoch": 0.3697798025816249,
      "grad_norm": 1.0498058795928955,
      "learning_rate": 1.874904342261856e-05,
      "loss": 0.6372,
      "step": 2922
    },
    {
      "epoch": 0.36990635282207035,
      "grad_norm": 1.0409060716629028,
      "learning_rate": 1.87481753642541e-05,
      "loss": 0.6476,
      "step": 2923
    },
    {
      "epoch": 0.3700329030625158,
      "grad_norm": 1.4390454292297363,
      "learning_rate": 1.874730702492389e-05,
      "loss": 0.6901,
      "step": 2924
    },
    {
      "epoch": 0.3701594533029613,
      "grad_norm": 1.0890412330627441,
      "learning_rate": 1.8746438404655802e-05,
      "loss": 0.6682,
      "step": 2925
    },
    {
      "epoch": 0.37028600354340674,
      "grad_norm": 1.2107499837875366,
      "learning_rate": 1.874556950347775e-05,
      "loss": 0.6383,
      "step": 2926
    },
    {
      "epoch": 0.3704125537838522,
      "grad_norm": 1.062759280204773,
      "learning_rate": 1.874470032141763e-05,
      "loss": 0.6413,
      "step": 2927
    },
    {
      "epoch": 0.3705391040242976,
      "grad_norm": 1.3091328144073486,
      "learning_rate": 1.8743830858503363e-05,
      "loss": 0.6395,
      "step": 2928
    },
    {
      "epoch": 0.3706656542647431,
      "grad_norm": 1.1976318359375,
      "learning_rate": 1.874296111476287e-05,
      "loss": 0.6983,
      "step": 2929
    },
    {
      "epoch": 0.37079220450518857,
      "grad_norm": 1.16410493850708,
      "learning_rate": 1.874209109022409e-05,
      "loss": 0.6456,
      "step": 2930
    },
    {
      "epoch": 0.370918754745634,
      "grad_norm": 1.0910305976867676,
      "learning_rate": 1.8741220784914964e-05,
      "loss": 0.6266,
      "step": 2931
    },
    {
      "epoch": 0.37104530498607946,
      "grad_norm": 1.1433706283569336,
      "learning_rate": 1.874035019886344e-05,
      "loss": 0.6446,
      "step": 2932
    },
    {
      "epoch": 0.3711718552265249,
      "grad_norm": 1.0679960250854492,
      "learning_rate": 1.873947933209748e-05,
      "loss": 0.6789,
      "step": 2933
    },
    {
      "epoch": 0.3712984054669704,
      "grad_norm": 1.2347091436386108,
      "learning_rate": 1.8738608184645057e-05,
      "loss": 0.6062,
      "step": 2934
    },
    {
      "epoch": 0.37142495570741585,
      "grad_norm": 1.1851966381072998,
      "learning_rate": 1.8737736756534145e-05,
      "loss": 0.6531,
      "step": 2935
    },
    {
      "epoch": 0.3715515059478613,
      "grad_norm": 1.3474055528640747,
      "learning_rate": 1.8736865047792734e-05,
      "loss": 0.6582,
      "step": 2936
    },
    {
      "epoch": 0.37167805618830674,
      "grad_norm": 1.2433713674545288,
      "learning_rate": 1.873599305844882e-05,
      "loss": 0.6791,
      "step": 2937
    },
    {
      "epoch": 0.37180460642875224,
      "grad_norm": 1.2115737199783325,
      "learning_rate": 1.8735120788530412e-05,
      "loss": 0.6483,
      "step": 2938
    },
    {
      "epoch": 0.3719311566691977,
      "grad_norm": 1.1454299688339233,
      "learning_rate": 1.873424823806552e-05,
      "loss": 0.6331,
      "step": 2939
    },
    {
      "epoch": 0.3720577069096431,
      "grad_norm": 1.2019070386886597,
      "learning_rate": 1.8733375407082172e-05,
      "loss": 0.6761,
      "step": 2940
    },
    {
      "epoch": 0.37218425715008857,
      "grad_norm": 1.236263394355774,
      "learning_rate": 1.87325022956084e-05,
      "loss": 0.6957,
      "step": 2941
    },
    {
      "epoch": 0.372310807390534,
      "grad_norm": 1.248347282409668,
      "learning_rate": 1.873162890367224e-05,
      "loss": 0.7106,
      "step": 2942
    },
    {
      "epoch": 0.3724373576309795,
      "grad_norm": 1.244560956954956,
      "learning_rate": 1.873075523130175e-05,
      "loss": 0.6584,
      "step": 2943
    },
    {
      "epoch": 0.37256390787142496,
      "grad_norm": 1.2027522325515747,
      "learning_rate": 1.8729881278524985e-05,
      "loss": 0.691,
      "step": 2944
    },
    {
      "epoch": 0.3726904581118704,
      "grad_norm": 1.126194715499878,
      "learning_rate": 1.8729007045370018e-05,
      "loss": 0.6611,
      "step": 2945
    },
    {
      "epoch": 0.37281700835231585,
      "grad_norm": 1.1446839570999146,
      "learning_rate": 1.8728132531864926e-05,
      "loss": 0.6902,
      "step": 2946
    },
    {
      "epoch": 0.37294355859276135,
      "grad_norm": 1.055340051651001,
      "learning_rate": 1.8727257738037793e-05,
      "loss": 0.6381,
      "step": 2947
    },
    {
      "epoch": 0.3730701088332068,
      "grad_norm": 1.097338318824768,
      "learning_rate": 1.8726382663916714e-05,
      "loss": 0.7259,
      "step": 2948
    },
    {
      "epoch": 0.37319665907365224,
      "grad_norm": 1.0713136196136475,
      "learning_rate": 1.87255073095298e-05,
      "loss": 0.6674,
      "step": 2949
    },
    {
      "epoch": 0.3733232093140977,
      "grad_norm": 1.0438917875289917,
      "learning_rate": 1.872463167490516e-05,
      "loss": 0.6349,
      "step": 2950
    },
    {
      "epoch": 0.3734497595545431,
      "grad_norm": 0.9662273526191711,
      "learning_rate": 1.8723755760070916e-05,
      "loss": 0.6428,
      "step": 2951
    },
    {
      "epoch": 0.3735763097949886,
      "grad_norm": 1.168408989906311,
      "learning_rate": 1.8722879565055204e-05,
      "loss": 0.6755,
      "step": 2952
    },
    {
      "epoch": 0.37370286003543407,
      "grad_norm": 1.1032123565673828,
      "learning_rate": 1.872200308988616e-05,
      "loss": 0.7007,
      "step": 2953
    },
    {
      "epoch": 0.3738294102758795,
      "grad_norm": 1.3144729137420654,
      "learning_rate": 1.8721126334591937e-05,
      "loss": 0.6768,
      "step": 2954
    },
    {
      "epoch": 0.37395596051632496,
      "grad_norm": 1.106518030166626,
      "learning_rate": 1.8720249299200695e-05,
      "loss": 0.704,
      "step": 2955
    },
    {
      "epoch": 0.37408251075677046,
      "grad_norm": 1.3199691772460938,
      "learning_rate": 1.87193719837406e-05,
      "loss": 0.6711,
      "step": 2956
    },
    {
      "epoch": 0.3742090609972159,
      "grad_norm": 1.2058138847351074,
      "learning_rate": 1.8718494388239826e-05,
      "loss": 0.68,
      "step": 2957
    },
    {
      "epoch": 0.37433561123766135,
      "grad_norm": 1.4292750358581543,
      "learning_rate": 1.8717616512726562e-05,
      "loss": 0.6627,
      "step": 2958
    },
    {
      "epoch": 0.3744621614781068,
      "grad_norm": 1.1042755842208862,
      "learning_rate": 1.8716738357229004e-05,
      "loss": 0.6489,
      "step": 2959
    },
    {
      "epoch": 0.3745887117185523,
      "grad_norm": 1.38535737991333,
      "learning_rate": 1.871585992177535e-05,
      "loss": 0.6852,
      "step": 2960
    },
    {
      "epoch": 0.37471526195899774,
      "grad_norm": 1.2102270126342773,
      "learning_rate": 1.8714981206393822e-05,
      "loss": 0.6757,
      "step": 2961
    },
    {
      "epoch": 0.3748418121994432,
      "grad_norm": 1.2433370351791382,
      "learning_rate": 1.871410221111263e-05,
      "loss": 0.6243,
      "step": 2962
    },
    {
      "epoch": 0.3749683624398886,
      "grad_norm": 1.2545957565307617,
      "learning_rate": 1.8713222935960015e-05,
      "loss": 0.7283,
      "step": 2963
    },
    {
      "epoch": 0.37509491268033407,
      "grad_norm": 1.1263327598571777,
      "learning_rate": 1.8712343380964216e-05,
      "loss": 0.6621,
      "step": 2964
    },
    {
      "epoch": 0.37522146292077957,
      "grad_norm": 1.3181078433990479,
      "learning_rate": 1.8711463546153476e-05,
      "loss": 0.6025,
      "step": 2965
    },
    {
      "epoch": 0.375348013161225,
      "grad_norm": 1.122555136680603,
      "learning_rate": 1.8710583431556057e-05,
      "loss": 0.6876,
      "step": 2966
    },
    {
      "epoch": 0.37547456340167046,
      "grad_norm": 1.1468145847320557,
      "learning_rate": 1.8709703037200223e-05,
      "loss": 0.6983,
      "step": 2967
    },
    {
      "epoch": 0.3756011136421159,
      "grad_norm": 1.0048121213912964,
      "learning_rate": 1.870882236311425e-05,
      "loss": 0.6369,
      "step": 2968
    },
    {
      "epoch": 0.3757276638825614,
      "grad_norm": 1.2081165313720703,
      "learning_rate": 1.8707941409326425e-05,
      "loss": 0.702,
      "step": 2969
    },
    {
      "epoch": 0.37585421412300685,
      "grad_norm": 1.0397460460662842,
      "learning_rate": 1.8707060175865036e-05,
      "loss": 0.6579,
      "step": 2970
    },
    {
      "epoch": 0.3759807643634523,
      "grad_norm": 1.1787896156311035,
      "learning_rate": 1.8706178662758396e-05,
      "loss": 0.6304,
      "step": 2971
    },
    {
      "epoch": 0.37610731460389774,
      "grad_norm": 1.051835060119629,
      "learning_rate": 1.870529687003481e-05,
      "loss": 0.6294,
      "step": 2972
    },
    {
      "epoch": 0.3762338648443432,
      "grad_norm": 1.0449624061584473,
      "learning_rate": 1.8704414797722594e-05,
      "loss": 0.6628,
      "step": 2973
    },
    {
      "epoch": 0.3763604150847887,
      "grad_norm": 1.1239902973175049,
      "learning_rate": 1.8703532445850085e-05,
      "loss": 0.667,
      "step": 2974
    },
    {
      "epoch": 0.3764869653252341,
      "grad_norm": 1.0490485429763794,
      "learning_rate": 1.870264981444562e-05,
      "loss": 0.6634,
      "step": 2975
    },
    {
      "epoch": 0.37661351556567957,
      "grad_norm": 1.0758813619613647,
      "learning_rate": 1.8701766903537548e-05,
      "loss": 0.6373,
      "step": 2976
    },
    {
      "epoch": 0.376740065806125,
      "grad_norm": 1.149254322052002,
      "learning_rate": 1.8700883713154222e-05,
      "loss": 0.6847,
      "step": 2977
    },
    {
      "epoch": 0.3768666160465705,
      "grad_norm": 1.0697414875030518,
      "learning_rate": 1.8700000243324004e-05,
      "loss": 0.6733,
      "step": 2978
    },
    {
      "epoch": 0.37699316628701596,
      "grad_norm": 1.0854767560958862,
      "learning_rate": 1.869911649407528e-05,
      "loss": 0.6391,
      "step": 2979
    },
    {
      "epoch": 0.3771197165274614,
      "grad_norm": 1.173665165901184,
      "learning_rate": 1.8698232465436423e-05,
      "loss": 0.6623,
      "step": 2980
    },
    {
      "epoch": 0.37724626676790685,
      "grad_norm": 1.07884681224823,
      "learning_rate": 1.8697348157435828e-05,
      "loss": 0.6725,
      "step": 2981
    },
    {
      "epoch": 0.3773728170083523,
      "grad_norm": 1.0558823347091675,
      "learning_rate": 1.86964635701019e-05,
      "loss": 0.6573,
      "step": 2982
    },
    {
      "epoch": 0.3774993672487978,
      "grad_norm": 1.208953619003296,
      "learning_rate": 1.8695578703463046e-05,
      "loss": 0.6615,
      "step": 2983
    },
    {
      "epoch": 0.37762591748924323,
      "grad_norm": 1.1360911130905151,
      "learning_rate": 1.8694693557547684e-05,
      "loss": 0.672,
      "step": 2984
    },
    {
      "epoch": 0.3777524677296887,
      "grad_norm": 1.1435760259628296,
      "learning_rate": 1.869380813238425e-05,
      "loss": 0.7007,
      "step": 2985
    },
    {
      "epoch": 0.3778790179701341,
      "grad_norm": 1.065979242324829,
      "learning_rate": 1.8692922428001166e-05,
      "loss": 0.6481,
      "step": 2986
    },
    {
      "epoch": 0.3780055682105796,
      "grad_norm": 1.0729353427886963,
      "learning_rate": 1.8692036444426897e-05,
      "loss": 0.7027,
      "step": 2987
    },
    {
      "epoch": 0.37813211845102507,
      "grad_norm": 1.16510808467865,
      "learning_rate": 1.8691150181689887e-05,
      "loss": 0.6563,
      "step": 2988
    },
    {
      "epoch": 0.3782586686914705,
      "grad_norm": 1.1614526510238647,
      "learning_rate": 1.8690263639818597e-05,
      "loss": 0.6995,
      "step": 2989
    },
    {
      "epoch": 0.37838521893191596,
      "grad_norm": 1.0689175128936768,
      "learning_rate": 1.868937681884151e-05,
      "loss": 0.6425,
      "step": 2990
    },
    {
      "epoch": 0.3785117691723614,
      "grad_norm": 1.1318020820617676,
      "learning_rate": 1.8688489718787103e-05,
      "loss": 0.6882,
      "step": 2991
    },
    {
      "epoch": 0.3786383194128069,
      "grad_norm": 1.0448355674743652,
      "learning_rate": 1.868760233968387e-05,
      "loss": 0.6286,
      "step": 2992
    },
    {
      "epoch": 0.37876486965325235,
      "grad_norm": 1.152972936630249,
      "learning_rate": 1.8686714681560304e-05,
      "loss": 0.6603,
      "step": 2993
    },
    {
      "epoch": 0.3788914198936978,
      "grad_norm": 1.160873532295227,
      "learning_rate": 1.868582674444492e-05,
      "loss": 0.7211,
      "step": 2994
    },
    {
      "epoch": 0.37901797013414323,
      "grad_norm": 1.1397182941436768,
      "learning_rate": 1.8684938528366235e-05,
      "loss": 0.6535,
      "step": 2995
    },
    {
      "epoch": 0.37914452037458873,
      "grad_norm": 1.3654909133911133,
      "learning_rate": 1.8684050033352776e-05,
      "loss": 0.6437,
      "step": 2996
    },
    {
      "epoch": 0.3792710706150342,
      "grad_norm": 1.1168630123138428,
      "learning_rate": 1.8683161259433077e-05,
      "loss": 0.6921,
      "step": 2997
    },
    {
      "epoch": 0.3793976208554796,
      "grad_norm": 1.1186308860778809,
      "learning_rate": 1.8682272206635686e-05,
      "loss": 0.6557,
      "step": 2998
    },
    {
      "epoch": 0.37952417109592507,
      "grad_norm": 1.0849076509475708,
      "learning_rate": 1.8681382874989152e-05,
      "loss": 0.69,
      "step": 2999
    },
    {
      "epoch": 0.37965072133637057,
      "grad_norm": 1.0244297981262207,
      "learning_rate": 1.8680493264522043e-05,
      "loss": 0.6661,
      "step": 3000
    },
    {
      "epoch": 0.379777271576816,
      "grad_norm": 0.9954763650894165,
      "learning_rate": 1.8679603375262926e-05,
      "loss": 0.6234,
      "step": 3001
    },
    {
      "epoch": 0.37990382181726146,
      "grad_norm": 1.089093565940857,
      "learning_rate": 1.8678713207240383e-05,
      "loss": 0.6733,
      "step": 3002
    },
    {
      "epoch": 0.3800303720577069,
      "grad_norm": 1.02705717086792,
      "learning_rate": 1.867782276048301e-05,
      "loss": 0.6417,
      "step": 3003
    },
    {
      "epoch": 0.38015692229815234,
      "grad_norm": 1.0365369319915771,
      "learning_rate": 1.8676932035019396e-05,
      "loss": 0.6337,
      "step": 3004
    },
    {
      "epoch": 0.38028347253859784,
      "grad_norm": 1.1228113174438477,
      "learning_rate": 1.867604103087815e-05,
      "loss": 0.684,
      "step": 3005
    },
    {
      "epoch": 0.3804100227790433,
      "grad_norm": 1.104498028755188,
      "learning_rate": 1.8675149748087892e-05,
      "loss": 0.6582,
      "step": 3006
    },
    {
      "epoch": 0.38053657301948873,
      "grad_norm": 1.1619032621383667,
      "learning_rate": 1.867425818667725e-05,
      "loss": 0.6648,
      "step": 3007
    },
    {
      "epoch": 0.3806631232599342,
      "grad_norm": 1.0988856554031372,
      "learning_rate": 1.867336634667485e-05,
      "loss": 0.6857,
      "step": 3008
    },
    {
      "epoch": 0.3807896735003797,
      "grad_norm": 1.0808483362197876,
      "learning_rate": 1.8672474228109344e-05,
      "loss": 0.6691,
      "step": 3009
    },
    {
      "epoch": 0.3809162237408251,
      "grad_norm": 1.1448111534118652,
      "learning_rate": 1.8671581831009378e-05,
      "loss": 0.6811,
      "step": 3010
    },
    {
      "epoch": 0.38104277398127057,
      "grad_norm": 1.3668915033340454,
      "learning_rate": 1.8670689155403618e-05,
      "loss": 0.7094,
      "step": 3011
    },
    {
      "epoch": 0.381169324221716,
      "grad_norm": 0.9847873449325562,
      "learning_rate": 1.8669796201320725e-05,
      "loss": 0.6413,
      "step": 3012
    },
    {
      "epoch": 0.38129587446216145,
      "grad_norm": 1.0893404483795166,
      "learning_rate": 1.8668902968789392e-05,
      "loss": 0.718,
      "step": 3013
    },
    {
      "epoch": 0.38142242470260695,
      "grad_norm": 1.0632299184799194,
      "learning_rate": 1.8668009457838296e-05,
      "loss": 0.6496,
      "step": 3014
    },
    {
      "epoch": 0.3815489749430524,
      "grad_norm": 1.004468560218811,
      "learning_rate": 1.8667115668496136e-05,
      "loss": 0.6348,
      "step": 3015
    },
    {
      "epoch": 0.38167552518349784,
      "grad_norm": 1.0683302879333496,
      "learning_rate": 1.866622160079162e-05,
      "loss": 0.6996,
      "step": 3016
    },
    {
      "epoch": 0.3818020754239433,
      "grad_norm": 1.1366000175476074,
      "learning_rate": 1.8665327254753466e-05,
      "loss": 0.6841,
      "step": 3017
    },
    {
      "epoch": 0.3819286256643888,
      "grad_norm": 1.1922308206558228,
      "learning_rate": 1.866443263041039e-05,
      "loss": 0.6808,
      "step": 3018
    },
    {
      "epoch": 0.38205517590483423,
      "grad_norm": 1.1673305034637451,
      "learning_rate": 1.866353772779113e-05,
      "loss": 0.7253,
      "step": 3019
    },
    {
      "epoch": 0.3821817261452797,
      "grad_norm": 1.1389377117156982,
      "learning_rate": 1.8662642546924425e-05,
      "loss": 0.6758,
      "step": 3020
    },
    {
      "epoch": 0.3823082763857251,
      "grad_norm": 0.9768390655517578,
      "learning_rate": 1.8661747087839027e-05,
      "loss": 0.6336,
      "step": 3021
    },
    {
      "epoch": 0.38243482662617057,
      "grad_norm": 1.1002428531646729,
      "learning_rate": 1.8660851350563696e-05,
      "loss": 0.671,
      "step": 3022
    },
    {
      "epoch": 0.38256137686661607,
      "grad_norm": 1.0138039588928223,
      "learning_rate": 1.86599553351272e-05,
      "loss": 0.6554,
      "step": 3023
    },
    {
      "epoch": 0.3826879271070615,
      "grad_norm": 1.0876187086105347,
      "learning_rate": 1.865905904155832e-05,
      "loss": 0.645,
      "step": 3024
    },
    {
      "epoch": 0.38281447734750695,
      "grad_norm": 1.1032459735870361,
      "learning_rate": 1.8658162469885835e-05,
      "loss": 0.6531,
      "step": 3025
    },
    {
      "epoch": 0.3829410275879524,
      "grad_norm": 1.0563875436782837,
      "learning_rate": 1.8657265620138543e-05,
      "loss": 0.6689,
      "step": 3026
    },
    {
      "epoch": 0.3830675778283979,
      "grad_norm": 1.205802083015442,
      "learning_rate": 1.8656368492345252e-05,
      "loss": 0.6959,
      "step": 3027
    },
    {
      "epoch": 0.38319412806884334,
      "grad_norm": 1.0943418741226196,
      "learning_rate": 1.865547108653477e-05,
      "loss": 0.6625,
      "step": 3028
    },
    {
      "epoch": 0.3833206783092888,
      "grad_norm": 1.2681344747543335,
      "learning_rate": 1.865457340273592e-05,
      "loss": 0.657,
      "step": 3029
    },
    {
      "epoch": 0.38344722854973423,
      "grad_norm": 1.094083547592163,
      "learning_rate": 1.865367544097754e-05,
      "loss": 0.6652,
      "step": 3030
    },
    {
      "epoch": 0.3835737787901797,
      "grad_norm": 1.2158397436141968,
      "learning_rate": 1.8652777201288458e-05,
      "loss": 0.6447,
      "step": 3031
    },
    {
      "epoch": 0.3837003290306252,
      "grad_norm": 1.1166341304779053,
      "learning_rate": 1.8651878683697534e-05,
      "loss": 0.6818,
      "step": 3032
    },
    {
      "epoch": 0.3838268792710706,
      "grad_norm": 1.17692232131958,
      "learning_rate": 1.8650979888233616e-05,
      "loss": 0.6622,
      "step": 3033
    },
    {
      "epoch": 0.38395342951151606,
      "grad_norm": 1.0818978548049927,
      "learning_rate": 1.865008081492558e-05,
      "loss": 0.6521,
      "step": 3034
    },
    {
      "epoch": 0.3840799797519615,
      "grad_norm": 1.347763180732727,
      "learning_rate": 1.864918146380229e-05,
      "loss": 0.6841,
      "step": 3035
    },
    {
      "epoch": 0.384206529992407,
      "grad_norm": 1.1639262437820435,
      "learning_rate": 1.864828183489264e-05,
      "loss": 0.656,
      "step": 3036
    },
    {
      "epoch": 0.38433308023285245,
      "grad_norm": 1.1059741973876953,
      "learning_rate": 1.8647381928225523e-05,
      "loss": 0.6571,
      "step": 3037
    },
    {
      "epoch": 0.3844596304732979,
      "grad_norm": 1.2013012170791626,
      "learning_rate": 1.864648174382984e-05,
      "loss": 0.6843,
      "step": 3038
    },
    {
      "epoch": 0.38458618071374334,
      "grad_norm": 1.1417930126190186,
      "learning_rate": 1.86455812817345e-05,
      "loss": 0.6493,
      "step": 3039
    },
    {
      "epoch": 0.3847127309541888,
      "grad_norm": 1.0958887338638306,
      "learning_rate": 1.8644680541968424e-05,
      "loss": 0.6685,
      "step": 3040
    },
    {
      "epoch": 0.3848392811946343,
      "grad_norm": 1.0734211206436157,
      "learning_rate": 1.8643779524560543e-05,
      "loss": 0.6512,
      "step": 3041
    },
    {
      "epoch": 0.38496583143507973,
      "grad_norm": 1.1432565450668335,
      "learning_rate": 1.8642878229539793e-05,
      "loss": 0.706,
      "step": 3042
    },
    {
      "epoch": 0.3850923816755252,
      "grad_norm": 1.0865916013717651,
      "learning_rate": 1.8641976656935125e-05,
      "loss": 0.6885,
      "step": 3043
    },
    {
      "epoch": 0.3852189319159706,
      "grad_norm": 1.0400816202163696,
      "learning_rate": 1.8641074806775486e-05,
      "loss": 0.6671,
      "step": 3044
    },
    {
      "epoch": 0.3853454821564161,
      "grad_norm": 1.0607047080993652,
      "learning_rate": 1.8640172679089853e-05,
      "loss": 0.6595,
      "step": 3045
    },
    {
      "epoch": 0.38547203239686156,
      "grad_norm": 1.0926406383514404,
      "learning_rate": 1.863927027390719e-05,
      "loss": 0.697,
      "step": 3046
    },
    {
      "epoch": 0.385598582637307,
      "grad_norm": 1.4612618684768677,
      "learning_rate": 1.8638367591256482e-05,
      "loss": 0.7196,
      "step": 3047
    },
    {
      "epoch": 0.38572513287775245,
      "grad_norm": 1.0459989309310913,
      "learning_rate": 1.8637464631166723e-05,
      "loss": 0.6529,
      "step": 3048
    },
    {
      "epoch": 0.38585168311819795,
      "grad_norm": 1.1387603282928467,
      "learning_rate": 1.863656139366691e-05,
      "loss": 0.6567,
      "step": 3049
    },
    {
      "epoch": 0.3859782333586434,
      "grad_norm": 1.1586390733718872,
      "learning_rate": 1.8635657878786055e-05,
      "loss": 0.6742,
      "step": 3050
    },
    {
      "epoch": 0.38610478359908884,
      "grad_norm": 1.101112961769104,
      "learning_rate": 1.8634754086553178e-05,
      "loss": 0.6531,
      "step": 3051
    },
    {
      "epoch": 0.3862313338395343,
      "grad_norm": 1.065043568611145,
      "learning_rate": 1.86338500169973e-05,
      "loss": 0.6489,
      "step": 3052
    },
    {
      "epoch": 0.38635788407997973,
      "grad_norm": 1.0761244297027588,
      "learning_rate": 1.8632945670147462e-05,
      "loss": 0.6443,
      "step": 3053
    },
    {
      "epoch": 0.38648443432042523,
      "grad_norm": 1.081400752067566,
      "learning_rate": 1.8632041046032706e-05,
      "loss": 0.674,
      "step": 3054
    },
    {
      "epoch": 0.3866109845608707,
      "grad_norm": 1.124401569366455,
      "learning_rate": 1.863113614468209e-05,
      "loss": 0.7237,
      "step": 3055
    },
    {
      "epoch": 0.3867375348013161,
      "grad_norm": 1.0188703536987305,
      "learning_rate": 1.8630230966124674e-05,
      "loss": 0.6479,
      "step": 3056
    },
    {
      "epoch": 0.38686408504176156,
      "grad_norm": 1.1203112602233887,
      "learning_rate": 1.8629325510389528e-05,
      "loss": 0.665,
      "step": 3057
    },
    {
      "epoch": 0.38699063528220706,
      "grad_norm": 1.1983388662338257,
      "learning_rate": 1.8628419777505737e-05,
      "loss": 0.66,
      "step": 3058
    },
    {
      "epoch": 0.3871171855226525,
      "grad_norm": 1.0550005435943604,
      "learning_rate": 1.8627513767502388e-05,
      "loss": 0.6821,
      "step": 3059
    },
    {
      "epoch": 0.38724373576309795,
      "grad_norm": 1.1346067190170288,
      "learning_rate": 1.8626607480408578e-05,
      "loss": 0.6219,
      "step": 3060
    },
    {
      "epoch": 0.3873702860035434,
      "grad_norm": 7.169003486633301,
      "learning_rate": 1.8625700916253415e-05,
      "loss": 0.7302,
      "step": 3061
    },
    {
      "epoch": 0.38749683624398884,
      "grad_norm": 1.4371531009674072,
      "learning_rate": 1.8624794075066017e-05,
      "loss": 0.6978,
      "step": 3062
    },
    {
      "epoch": 0.38762338648443434,
      "grad_norm": 1.0971049070358276,
      "learning_rate": 1.8623886956875507e-05,
      "loss": 0.6903,
      "step": 3063
    },
    {
      "epoch": 0.3877499367248798,
      "grad_norm": 1.0785351991653442,
      "learning_rate": 1.8622979561711017e-05,
      "loss": 0.6804,
      "step": 3064
    },
    {
      "epoch": 0.38787648696532523,
      "grad_norm": 1.1050162315368652,
      "learning_rate": 1.86220718896017e-05,
      "loss": 0.637,
      "step": 3065
    },
    {
      "epoch": 0.3880030372057707,
      "grad_norm": 1.0188689231872559,
      "learning_rate": 1.8621163940576697e-05,
      "loss": 0.6633,
      "step": 3066
    },
    {
      "epoch": 0.3881295874462162,
      "grad_norm": 1.026931643486023,
      "learning_rate": 1.862025571466517e-05,
      "loss": 0.6041,
      "step": 3067
    },
    {
      "epoch": 0.3882561376866616,
      "grad_norm": 1.331531286239624,
      "learning_rate": 1.861934721189629e-05,
      "loss": 0.7274,
      "step": 3068
    },
    {
      "epoch": 0.38838268792710706,
      "grad_norm": 1.082899570465088,
      "learning_rate": 1.861843843229924e-05,
      "loss": 0.6647,
      "step": 3069
    },
    {
      "epoch": 0.3885092381675525,
      "grad_norm": 1.1088200807571411,
      "learning_rate": 1.8617529375903205e-05,
      "loss": 0.676,
      "step": 3070
    },
    {
      "epoch": 0.38863578840799795,
      "grad_norm": 1.0375852584838867,
      "learning_rate": 1.8616620042737378e-05,
      "loss": 0.64,
      "step": 3071
    },
    {
      "epoch": 0.38876233864844345,
      "grad_norm": 1.0916281938552856,
      "learning_rate": 1.8615710432830966e-05,
      "loss": 0.6603,
      "step": 3072
    },
    {
      "epoch": 0.3888888888888889,
      "grad_norm": 1.0983717441558838,
      "learning_rate": 1.861480054621318e-05,
      "loss": 0.6773,
      "step": 3073
    },
    {
      "epoch": 0.38901543912933434,
      "grad_norm": 1.0538969039916992,
      "learning_rate": 1.861389038291325e-05,
      "loss": 0.6625,
      "step": 3074
    },
    {
      "epoch": 0.3891419893697798,
      "grad_norm": 1.060303807258606,
      "learning_rate": 1.86129799429604e-05,
      "loss": 0.6876,
      "step": 3075
    },
    {
      "epoch": 0.3892685396102253,
      "grad_norm": 1.1989209651947021,
      "learning_rate": 1.8612069226383876e-05,
      "loss": 0.6544,
      "step": 3076
    },
    {
      "epoch": 0.38939508985067073,
      "grad_norm": 1.1048470735549927,
      "learning_rate": 1.8611158233212925e-05,
      "loss": 0.6605,
      "step": 3077
    },
    {
      "epoch": 0.3895216400911162,
      "grad_norm": 1.0586085319519043,
      "learning_rate": 1.8610246963476805e-05,
      "loss": 0.6457,
      "step": 3078
    },
    {
      "epoch": 0.3896481903315616,
      "grad_norm": 1.1497958898544312,
      "learning_rate": 1.8609335417204788e-05,
      "loss": 0.6509,
      "step": 3079
    },
    {
      "epoch": 0.38977474057200706,
      "grad_norm": 1.1493924856185913,
      "learning_rate": 1.8608423594426143e-05,
      "loss": 0.6448,
      "step": 3080
    },
    {
      "epoch": 0.38990129081245256,
      "grad_norm": 1.1170318126678467,
      "learning_rate": 1.8607511495170163e-05,
      "loss": 0.6218,
      "step": 3081
    },
    {
      "epoch": 0.390027841052898,
      "grad_norm": 1.2892712354660034,
      "learning_rate": 1.8606599119466136e-05,
      "loss": 0.6591,
      "step": 3082
    },
    {
      "epoch": 0.39015439129334345,
      "grad_norm": 1.0991783142089844,
      "learning_rate": 1.8605686467343363e-05,
      "loss": 0.5978,
      "step": 3083
    },
    {
      "epoch": 0.3902809415337889,
      "grad_norm": 1.0486305952072144,
      "learning_rate": 1.8604773538831164e-05,
      "loss": 0.6666,
      "step": 3084
    },
    {
      "epoch": 0.3904074917742344,
      "grad_norm": 1.144901990890503,
      "learning_rate": 1.8603860333958854e-05,
      "loss": 0.7036,
      "step": 3085
    },
    {
      "epoch": 0.39053404201467984,
      "grad_norm": 0.9873310923576355,
      "learning_rate": 1.860294685275576e-05,
      "loss": 0.588,
      "step": 3086
    },
    {
      "epoch": 0.3906605922551253,
      "grad_norm": 1.1622213125228882,
      "learning_rate": 1.860203309525123e-05,
      "loss": 0.6638,
      "step": 3087
    },
    {
      "epoch": 0.3907871424955707,
      "grad_norm": 1.023512601852417,
      "learning_rate": 1.8601119061474598e-05,
      "loss": 0.6615,
      "step": 3088
    },
    {
      "epoch": 0.39091369273601617,
      "grad_norm": 1.1699488162994385,
      "learning_rate": 1.860020475145523e-05,
      "loss": 0.6673,
      "step": 3089
    },
    {
      "epoch": 0.39104024297646167,
      "grad_norm": 1.119884967803955,
      "learning_rate": 1.859929016522249e-05,
      "loss": 0.6824,
      "step": 3090
    },
    {
      "epoch": 0.3911667932169071,
      "grad_norm": 1.1315079927444458,
      "learning_rate": 1.859837530280575e-05,
      "loss": 0.6162,
      "step": 3091
    },
    {
      "epoch": 0.39129334345735256,
      "grad_norm": 1.0675368309020996,
      "learning_rate": 1.8597460164234386e-05,
      "loss": 0.6489,
      "step": 3092
    },
    {
      "epoch": 0.391419893697798,
      "grad_norm": 1.279061198234558,
      "learning_rate": 1.8596544749537803e-05,
      "loss": 0.6849,
      "step": 3093
    },
    {
      "epoch": 0.3915464439382435,
      "grad_norm": 1.1814274787902832,
      "learning_rate": 1.859562905874539e-05,
      "loss": 0.685,
      "step": 3094
    },
    {
      "epoch": 0.39167299417868895,
      "grad_norm": 1.1203621625900269,
      "learning_rate": 1.8594713091886563e-05,
      "loss": 0.6513,
      "step": 3095
    },
    {
      "epoch": 0.3917995444191344,
      "grad_norm": 1.129793405532837,
      "learning_rate": 1.8593796848990736e-05,
      "loss": 0.6301,
      "step": 3096
    },
    {
      "epoch": 0.39192609465957984,
      "grad_norm": 1.1149784326553345,
      "learning_rate": 1.8592880330087342e-05,
      "loss": 0.687,
      "step": 3097
    },
    {
      "epoch": 0.39205264490002534,
      "grad_norm": 1.0853538513183594,
      "learning_rate": 1.859196353520581e-05,
      "loss": 0.636,
      "step": 3098
    },
    {
      "epoch": 0.3921791951404708,
      "grad_norm": 1.1061296463012695,
      "learning_rate": 1.859104646437559e-05,
      "loss": 0.7071,
      "step": 3099
    },
    {
      "epoch": 0.3923057453809162,
      "grad_norm": 1.0851918458938599,
      "learning_rate": 1.859012911762613e-05,
      "loss": 0.6767,
      "step": 3100
    },
    {
      "epoch": 0.39243229562136167,
      "grad_norm": 1.185187578201294,
      "learning_rate": 1.8589211494986896e-05,
      "loss": 0.6746,
      "step": 3101
    },
    {
      "epoch": 0.3925588458618071,
      "grad_norm": 1.0677894353866577,
      "learning_rate": 1.8588293596487365e-05,
      "loss": 0.6355,
      "step": 3102
    },
    {
      "epoch": 0.3926853961022526,
      "grad_norm": 1.1001925468444824,
      "learning_rate": 1.8587375422157002e-05,
      "loss": 0.6927,
      "step": 3103
    },
    {
      "epoch": 0.39281194634269806,
      "grad_norm": 0.9788049459457397,
      "learning_rate": 1.8586456972025312e-05,
      "loss": 0.6164,
      "step": 3104
    },
    {
      "epoch": 0.3929384965831435,
      "grad_norm": 1.1942178010940552,
      "learning_rate": 1.8585538246121785e-05,
      "loss": 0.6852,
      "step": 3105
    },
    {
      "epoch": 0.39306504682358895,
      "grad_norm": 1.095186710357666,
      "learning_rate": 1.8584619244475928e-05,
      "loss": 0.6597,
      "step": 3106
    },
    {
      "epoch": 0.39319159706403445,
      "grad_norm": 1.0218037366867065,
      "learning_rate": 1.858369996711726e-05,
      "loss": 0.605,
      "step": 3107
    },
    {
      "epoch": 0.3933181473044799,
      "grad_norm": 1.0161601305007935,
      "learning_rate": 1.85827804140753e-05,
      "loss": 0.6735,
      "step": 3108
    },
    {
      "epoch": 0.39344469754492534,
      "grad_norm": 0.9792773127555847,
      "learning_rate": 1.8581860585379584e-05,
      "loss": 0.6348,
      "step": 3109
    },
    {
      "epoch": 0.3935712477853708,
      "grad_norm": 1.0077967643737793,
      "learning_rate": 1.858094048105966e-05,
      "loss": 0.6356,
      "step": 3110
    },
    {
      "epoch": 0.3936977980258162,
      "grad_norm": 1.0479403734207153,
      "learning_rate": 1.858002010114507e-05,
      "loss": 0.6636,
      "step": 3111
    },
    {
      "epoch": 0.3938243482662617,
      "grad_norm": 1.0450148582458496,
      "learning_rate": 1.857909944566538e-05,
      "loss": 0.694,
      "step": 3112
    },
    {
      "epoch": 0.39395089850670717,
      "grad_norm": 1.0747712850570679,
      "learning_rate": 1.8578178514650152e-05,
      "loss": 0.659,
      "step": 3113
    },
    {
      "epoch": 0.3940774487471526,
      "grad_norm": 1.0359846353530884,
      "learning_rate": 1.857725730812897e-05,
      "loss": 0.6636,
      "step": 3114
    },
    {
      "epoch": 0.39420399898759806,
      "grad_norm": 1.1241058111190796,
      "learning_rate": 1.8576335826131423e-05,
      "loss": 0.5894,
      "step": 3115
    },
    {
      "epoch": 0.39433054922804356,
      "grad_norm": 1.274362325668335,
      "learning_rate": 1.8575414068687098e-05,
      "loss": 0.6848,
      "step": 3116
    },
    {
      "epoch": 0.394457099468489,
      "grad_norm": 1.1428688764572144,
      "learning_rate": 1.8574492035825605e-05,
      "loss": 0.7023,
      "step": 3117
    },
    {
      "epoch": 0.39458364970893445,
      "grad_norm": 1.1245439052581787,
      "learning_rate": 1.857356972757655e-05,
      "loss": 0.6388,
      "step": 3118
    },
    {
      "epoch": 0.3947101999493799,
      "grad_norm": 1.0174925327301025,
      "learning_rate": 1.8572647143969568e-05,
      "loss": 0.6696,
      "step": 3119
    },
    {
      "epoch": 0.39483675018982534,
      "grad_norm": 1.1020185947418213,
      "learning_rate": 1.8571724285034277e-05,
      "loss": 0.6652,
      "step": 3120
    },
    {
      "epoch": 0.39496330043027084,
      "grad_norm": 0.997175931930542,
      "learning_rate": 1.8570801150800323e-05,
      "loss": 0.6057,
      "step": 3121
    },
    {
      "epoch": 0.3950898506707163,
      "grad_norm": 1.0601389408111572,
      "learning_rate": 1.8569877741297347e-05,
      "loss": 0.6632,
      "step": 3122
    },
    {
      "epoch": 0.3952164009111617,
      "grad_norm": 0.9657734632492065,
      "learning_rate": 1.8568954056555016e-05,
      "loss": 0.6362,
      "step": 3123
    },
    {
      "epoch": 0.39534295115160717,
      "grad_norm": 1.0915333032608032,
      "learning_rate": 1.856803009660299e-05,
      "loss": 0.6958,
      "step": 3124
    },
    {
      "epoch": 0.39546950139205267,
      "grad_norm": 1.0567071437835693,
      "learning_rate": 1.856710586147095e-05,
      "loss": 0.6626,
      "step": 3125
    },
    {
      "epoch": 0.3955960516324981,
      "grad_norm": 1.052115559577942,
      "learning_rate": 1.856618135118857e-05,
      "loss": 0.6344,
      "step": 3126
    },
    {
      "epoch": 0.39572260187294356,
      "grad_norm": 1.0764107704162598,
      "learning_rate": 1.8565256565785553e-05,
      "loss": 0.6885,
      "step": 3127
    },
    {
      "epoch": 0.395849152113389,
      "grad_norm": 1.1369692087173462,
      "learning_rate": 1.856433150529159e-05,
      "loss": 0.6867,
      "step": 3128
    },
    {
      "epoch": 0.39597570235383445,
      "grad_norm": 1.0260438919067383,
      "learning_rate": 1.85634061697364e-05,
      "loss": 0.669,
      "step": 3129
    },
    {
      "epoch": 0.39610225259427995,
      "grad_norm": 1.101898193359375,
      "learning_rate": 1.8562480559149697e-05,
      "loss": 0.6409,
      "step": 3130
    },
    {
      "epoch": 0.3962288028347254,
      "grad_norm": 1.1033155918121338,
      "learning_rate": 1.856155467356121e-05,
      "loss": 0.6285,
      "step": 3131
    },
    {
      "epoch": 0.39635535307517084,
      "grad_norm": 1.166475534439087,
      "learning_rate": 1.8560628513000676e-05,
      "loss": 0.6117,
      "step": 3132
    },
    {
      "epoch": 0.3964819033156163,
      "grad_norm": 1.169384479522705,
      "learning_rate": 1.855970207749784e-05,
      "loss": 0.7069,
      "step": 3133
    },
    {
      "epoch": 0.3966084535560618,
      "grad_norm": 1.101737141609192,
      "learning_rate": 1.855877536708246e-05,
      "loss": 0.6419,
      "step": 3134
    },
    {
      "epoch": 0.3967350037965072,
      "grad_norm": 0.9939728379249573,
      "learning_rate": 1.8557848381784294e-05,
      "loss": 0.6359,
      "step": 3135
    },
    {
      "epoch": 0.39686155403695267,
      "grad_norm": 1.146917462348938,
      "learning_rate": 1.8556921121633117e-05,
      "loss": 0.6906,
      "step": 3136
    },
    {
      "epoch": 0.3969881042773981,
      "grad_norm": 1.1801221370697021,
      "learning_rate": 1.8555993586658713e-05,
      "loss": 0.6912,
      "step": 3137
    },
    {
      "epoch": 0.39711465451784356,
      "grad_norm": 1.1332752704620361,
      "learning_rate": 1.8555065776890863e-05,
      "loss": 0.6801,
      "step": 3138
    },
    {
      "epoch": 0.39724120475828906,
      "grad_norm": 1.1599879264831543,
      "learning_rate": 1.855413769235937e-05,
      "loss": 0.673,
      "step": 3139
    },
    {
      "epoch": 0.3973677549987345,
      "grad_norm": 1.109193205833435,
      "learning_rate": 1.8553209333094048e-05,
      "loss": 0.678,
      "step": 3140
    },
    {
      "epoch": 0.39749430523917995,
      "grad_norm": 1.358920931816101,
      "learning_rate": 1.85522806991247e-05,
      "loss": 0.6773,
      "step": 3141
    },
    {
      "epoch": 0.3976208554796254,
      "grad_norm": 1.07080078125,
      "learning_rate": 1.855135179048116e-05,
      "loss": 0.634,
      "step": 3142
    },
    {
      "epoch": 0.3977474057200709,
      "grad_norm": 1.6908509731292725,
      "learning_rate": 1.8550422607193262e-05,
      "loss": 0.6871,
      "step": 3143
    },
    {
      "epoch": 0.39787395596051633,
      "grad_norm": 1.317725419998169,
      "learning_rate": 1.8549493149290848e-05,
      "loss": 0.6442,
      "step": 3144
    },
    {
      "epoch": 0.3980005062009618,
      "grad_norm": 1.2910133600234985,
      "learning_rate": 1.8548563416803764e-05,
      "loss": 0.6752,
      "step": 3145
    },
    {
      "epoch": 0.3981270564414072,
      "grad_norm": 1.2265598773956299,
      "learning_rate": 1.8547633409761877e-05,
      "loss": 0.68,
      "step": 3146
    },
    {
      "epoch": 0.3982536066818527,
      "grad_norm": 1.1126247644424438,
      "learning_rate": 1.8546703128195053e-05,
      "loss": 0.6399,
      "step": 3147
    },
    {
      "epoch": 0.39838015692229817,
      "grad_norm": 1.106188416481018,
      "learning_rate": 1.854577257213317e-05,
      "loss": 0.6712,
      "step": 3148
    },
    {
      "epoch": 0.3985067071627436,
      "grad_norm": 1.203925609588623,
      "learning_rate": 1.8544841741606112e-05,
      "loss": 0.7069,
      "step": 3149
    },
    {
      "epoch": 0.39863325740318906,
      "grad_norm": 1.2822904586791992,
      "learning_rate": 1.854391063664378e-05,
      "loss": 0.6412,
      "step": 3150
    },
    {
      "epoch": 0.3987598076436345,
      "grad_norm": 1.2014613151550293,
      "learning_rate": 1.854297925727608e-05,
      "loss": 0.6889,
      "step": 3151
    },
    {
      "epoch": 0.39888635788408,
      "grad_norm": 1.3088983297348022,
      "learning_rate": 1.8542047603532915e-05,
      "loss": 0.6451,
      "step": 3152
    },
    {
      "epoch": 0.39901290812452544,
      "grad_norm": 1.0184674263000488,
      "learning_rate": 1.8541115675444217e-05,
      "loss": 0.6406,
      "step": 3153
    },
    {
      "epoch": 0.3991394583649709,
      "grad_norm": 1.1973875761032104,
      "learning_rate": 1.854018347303991e-05,
      "loss": 0.6416,
      "step": 3154
    },
    {
      "epoch": 0.39926600860541633,
      "grad_norm": 1.0532914400100708,
      "learning_rate": 1.8539250996349942e-05,
      "loss": 0.6656,
      "step": 3155
    },
    {
      "epoch": 0.39939255884586183,
      "grad_norm": 1.2532520294189453,
      "learning_rate": 1.853831824540425e-05,
      "loss": 0.6321,
      "step": 3156
    },
    {
      "epoch": 0.3995191090863073,
      "grad_norm": 1.0276128053665161,
      "learning_rate": 1.8537385220232797e-05,
      "loss": 0.6421,
      "step": 3157
    },
    {
      "epoch": 0.3996456593267527,
      "grad_norm": 1.229143500328064,
      "learning_rate": 1.8536451920865554e-05,
      "loss": 0.64,
      "step": 3158
    },
    {
      "epoch": 0.39977220956719817,
      "grad_norm": 1.0530014038085938,
      "learning_rate": 1.8535518347332485e-05,
      "loss": 0.6812,
      "step": 3159
    },
    {
      "epoch": 0.3998987598076436,
      "grad_norm": 1.2565771341323853,
      "learning_rate": 1.8534584499663584e-05,
      "loss": 0.6601,
      "step": 3160
    },
    {
      "epoch": 0.4000253100480891,
      "grad_norm": 1.0618367195129395,
      "learning_rate": 1.853365037788884e-05,
      "loss": 0.6872,
      "step": 3161
    },
    {
      "epoch": 0.40015186028853456,
      "grad_norm": 1.1178395748138428,
      "learning_rate": 1.853271598203825e-05,
      "loss": 0.6729,
      "step": 3162
    },
    {
      "epoch": 0.40027841052898,
      "grad_norm": 1.0578042268753052,
      "learning_rate": 1.853178131214183e-05,
      "loss": 0.6453,
      "step": 3163
    },
    {
      "epoch": 0.40040496076942544,
      "grad_norm": 1.0989981889724731,
      "learning_rate": 1.8530846368229594e-05,
      "loss": 0.6863,
      "step": 3164
    },
    {
      "epoch": 0.40053151100987094,
      "grad_norm": 1.0538841485977173,
      "learning_rate": 1.8529911150331574e-05,
      "loss": 0.6202,
      "step": 3165
    },
    {
      "epoch": 0.4006580612503164,
      "grad_norm": 1.1137702465057373,
      "learning_rate": 1.8528975658477802e-05,
      "loss": 0.6332,
      "step": 3166
    },
    {
      "epoch": 0.40078461149076183,
      "grad_norm": 1.1317658424377441,
      "learning_rate": 1.852803989269833e-05,
      "loss": 0.6803,
      "step": 3167
    },
    {
      "epoch": 0.4009111617312073,
      "grad_norm": 1.2161988019943237,
      "learning_rate": 1.8527103853023204e-05,
      "loss": 0.7082,
      "step": 3168
    },
    {
      "epoch": 0.4010377119716527,
      "grad_norm": 1.0385340452194214,
      "learning_rate": 1.8526167539482493e-05,
      "loss": 0.6733,
      "step": 3169
    },
    {
      "epoch": 0.4011642622120982,
      "grad_norm": 1.1426050662994385,
      "learning_rate": 1.8525230952106265e-05,
      "loss": 0.6811,
      "step": 3170
    },
    {
      "epoch": 0.40129081245254367,
      "grad_norm": 1.18229079246521,
      "learning_rate": 1.85242940909246e-05,
      "loss": 0.693,
      "step": 3171
    },
    {
      "epoch": 0.4014173626929891,
      "grad_norm": 1.1419517993927002,
      "learning_rate": 1.852335695596759e-05,
      "loss": 0.6544,
      "step": 3172
    },
    {
      "epoch": 0.40154391293343455,
      "grad_norm": 1.0733261108398438,
      "learning_rate": 1.8522419547265334e-05,
      "loss": 0.6376,
      "step": 3173
    },
    {
      "epoch": 0.40167046317388005,
      "grad_norm": 1.1835963726043701,
      "learning_rate": 1.852148186484793e-05,
      "loss": 0.6624,
      "step": 3174
    },
    {
      "epoch": 0.4017970134143255,
      "grad_norm": 1.1827261447906494,
      "learning_rate": 1.852054390874551e-05,
      "loss": 0.6902,
      "step": 3175
    },
    {
      "epoch": 0.40192356365477094,
      "grad_norm": 1.1123480796813965,
      "learning_rate": 1.8519605678988185e-05,
      "loss": 0.6889,
      "step": 3176
    },
    {
      "epoch": 0.4020501138952164,
      "grad_norm": 1.1182643175125122,
      "learning_rate": 1.851866717560609e-05,
      "loss": 0.6676,
      "step": 3177
    },
    {
      "epoch": 0.40217666413566183,
      "grad_norm": 1.0202085971832275,
      "learning_rate": 1.8517728398629372e-05,
      "loss": 0.6603,
      "step": 3178
    },
    {
      "epoch": 0.40230321437610733,
      "grad_norm": 1.1024456024169922,
      "learning_rate": 1.8516789348088175e-05,
      "loss": 0.6401,
      "step": 3179
    },
    {
      "epoch": 0.4024297646165528,
      "grad_norm": 1.1205799579620361,
      "learning_rate": 1.8515850024012666e-05,
      "loss": 0.7023,
      "step": 3180
    },
    {
      "epoch": 0.4025563148569982,
      "grad_norm": 1.0896635055541992,
      "learning_rate": 1.8514910426433007e-05,
      "loss": 0.7032,
      "step": 3181
    },
    {
      "epoch": 0.40268286509744367,
      "grad_norm": 1.1419507265090942,
      "learning_rate": 1.851397055537938e-05,
      "loss": 0.6732,
      "step": 3182
    },
    {
      "epoch": 0.40280941533788917,
      "grad_norm": 1.0520559549331665,
      "learning_rate": 1.8513030410881963e-05,
      "loss": 0.6165,
      "step": 3183
    },
    {
      "epoch": 0.4029359655783346,
      "grad_norm": 1.1409149169921875,
      "learning_rate": 1.851208999297096e-05,
      "loss": 0.6169,
      "step": 3184
    },
    {
      "epoch": 0.40306251581878005,
      "grad_norm": 1.0982904434204102,
      "learning_rate": 1.8511149301676575e-05,
      "loss": 0.6686,
      "step": 3185
    },
    {
      "epoch": 0.4031890660592255,
      "grad_norm": 1.0890427827835083,
      "learning_rate": 1.8510208337029012e-05,
      "loss": 0.6486,
      "step": 3186
    },
    {
      "epoch": 0.40331561629967094,
      "grad_norm": 1.069460153579712,
      "learning_rate": 1.8509267099058493e-05,
      "loss": 0.6617,
      "step": 3187
    },
    {
      "epoch": 0.40344216654011644,
      "grad_norm": 1.0773403644561768,
      "learning_rate": 1.8508325587795256e-05,
      "loss": 0.6389,
      "step": 3188
    },
    {
      "epoch": 0.4035687167805619,
      "grad_norm": 1.0890603065490723,
      "learning_rate": 1.8507383803269535e-05,
      "loss": 0.6244,
      "step": 3189
    },
    {
      "epoch": 0.40369526702100733,
      "grad_norm": 1.0224772691726685,
      "learning_rate": 1.850644174551157e-05,
      "loss": 0.655,
      "step": 3190
    },
    {
      "epoch": 0.4038218172614528,
      "grad_norm": 1.2080745697021484,
      "learning_rate": 1.850549941455163e-05,
      "loss": 0.6837,
      "step": 3191
    },
    {
      "epoch": 0.4039483675018983,
      "grad_norm": 1.024741530418396,
      "learning_rate": 1.8504556810419974e-05,
      "loss": 0.6663,
      "step": 3192
    },
    {
      "epoch": 0.4040749177423437,
      "grad_norm": 1.6229256391525269,
      "learning_rate": 1.8503613933146873e-05,
      "loss": 0.6786,
      "step": 3193
    },
    {
      "epoch": 0.40420146798278916,
      "grad_norm": 1.1306045055389404,
      "learning_rate": 1.8502670782762614e-05,
      "loss": 0.6425,
      "step": 3194
    },
    {
      "epoch": 0.4043280182232346,
      "grad_norm": 1.1615949869155884,
      "learning_rate": 1.8501727359297485e-05,
      "loss": 0.673,
      "step": 3195
    },
    {
      "epoch": 0.4044545684636801,
      "grad_norm": 1.053791880607605,
      "learning_rate": 1.8500783662781785e-05,
      "loss": 0.649,
      "step": 3196
    },
    {
      "epoch": 0.40458111870412555,
      "grad_norm": 1.2641717195510864,
      "learning_rate": 1.8499839693245827e-05,
      "loss": 0.6459,
      "step": 3197
    },
    {
      "epoch": 0.404707668944571,
      "grad_norm": 1.1321290731430054,
      "learning_rate": 1.8498895450719926e-05,
      "loss": 0.6784,
      "step": 3198
    },
    {
      "epoch": 0.40483421918501644,
      "grad_norm": 1.1559072732925415,
      "learning_rate": 1.849795093523441e-05,
      "loss": 0.7129,
      "step": 3199
    },
    {
      "epoch": 0.4049607694254619,
      "grad_norm": 1.1517341136932373,
      "learning_rate": 1.849700614681961e-05,
      "loss": 0.6565,
      "step": 3200
    },
    {
      "epoch": 0.4050873196659074,
      "grad_norm": 1.1203734874725342,
      "learning_rate": 1.849606108550588e-05,
      "loss": 0.6655,
      "step": 3201
    },
    {
      "epoch": 0.40521386990635283,
      "grad_norm": 1.1843093633651733,
      "learning_rate": 1.8495115751323557e-05,
      "loss": 0.6603,
      "step": 3202
    },
    {
      "epoch": 0.4053404201467983,
      "grad_norm": 1.3350251913070679,
      "learning_rate": 1.8494170144303014e-05,
      "loss": 0.6726,
      "step": 3203
    },
    {
      "epoch": 0.4054669703872437,
      "grad_norm": 1.2116140127182007,
      "learning_rate": 1.849322426447462e-05,
      "loss": 0.6787,
      "step": 3204
    },
    {
      "epoch": 0.4055935206276892,
      "grad_norm": 1.2176642417907715,
      "learning_rate": 1.8492278111868748e-05,
      "loss": 0.6983,
      "step": 3205
    },
    {
      "epoch": 0.40572007086813466,
      "grad_norm": 1.177191972732544,
      "learning_rate": 1.849133168651579e-05,
      "loss": 0.6686,
      "step": 3206
    },
    {
      "epoch": 0.4058466211085801,
      "grad_norm": 1.0697097778320312,
      "learning_rate": 1.8490384988446143e-05,
      "loss": 0.6099,
      "step": 3207
    },
    {
      "epoch": 0.40597317134902555,
      "grad_norm": 1.1146050691604614,
      "learning_rate": 1.848943801769021e-05,
      "loss": 0.6423,
      "step": 3208
    },
    {
      "epoch": 0.406099721589471,
      "grad_norm": 1.0062044858932495,
      "learning_rate": 1.8488490774278408e-05,
      "loss": 0.6468,
      "step": 3209
    },
    {
      "epoch": 0.4062262718299165,
      "grad_norm": 1.1228471994400024,
      "learning_rate": 1.8487543258241155e-05,
      "loss": 0.6277,
      "step": 3210
    },
    {
      "epoch": 0.40635282207036194,
      "grad_norm": 1.066148281097412,
      "learning_rate": 1.8486595469608883e-05,
      "loss": 0.6675,
      "step": 3211
    },
    {
      "epoch": 0.4064793723108074,
      "grad_norm": 1.0606883764266968,
      "learning_rate": 1.8485647408412036e-05,
      "loss": 0.6373,
      "step": 3212
    },
    {
      "epoch": 0.40660592255125283,
      "grad_norm": 1.0336192846298218,
      "learning_rate": 1.8484699074681062e-05,
      "loss": 0.6633,
      "step": 3213
    },
    {
      "epoch": 0.40673247279169833,
      "grad_norm": 1.12703275680542,
      "learning_rate": 1.848375046844642e-05,
      "loss": 0.6695,
      "step": 3214
    },
    {
      "epoch": 0.4068590230321438,
      "grad_norm": 1.0270061492919922,
      "learning_rate": 1.8482801589738573e-05,
      "loss": 0.6529,
      "step": 3215
    },
    {
      "epoch": 0.4069855732725892,
      "grad_norm": 1.1753886938095093,
      "learning_rate": 1.8481852438587994e-05,
      "loss": 0.6663,
      "step": 3216
    },
    {
      "epoch": 0.40711212351303466,
      "grad_norm": 1.1111700534820557,
      "learning_rate": 1.8480903015025175e-05,
      "loss": 0.6869,
      "step": 3217
    },
    {
      "epoch": 0.4072386737534801,
      "grad_norm": 1.1205871105194092,
      "learning_rate": 1.8479953319080602e-05,
      "loss": 0.6886,
      "step": 3218
    },
    {
      "epoch": 0.4073652239939256,
      "grad_norm": 1.0984103679656982,
      "learning_rate": 1.8479003350784775e-05,
      "loss": 0.6701,
      "step": 3219
    },
    {
      "epoch": 0.40749177423437105,
      "grad_norm": 1.2022993564605713,
      "learning_rate": 1.8478053110168215e-05,
      "loss": 0.6831,
      "step": 3220
    },
    {
      "epoch": 0.4076183244748165,
      "grad_norm": 1.034580945968628,
      "learning_rate": 1.847710259726143e-05,
      "loss": 0.6441,
      "step": 3221
    },
    {
      "epoch": 0.40774487471526194,
      "grad_norm": 0.9717069864273071,
      "learning_rate": 1.8476151812094953e-05,
      "loss": 0.6279,
      "step": 3222
    },
    {
      "epoch": 0.40787142495570744,
      "grad_norm": 1.0540454387664795,
      "learning_rate": 1.8475200754699318e-05,
      "loss": 0.6582,
      "step": 3223
    },
    {
      "epoch": 0.4079979751961529,
      "grad_norm": 1.049422264099121,
      "learning_rate": 1.8474249425105068e-05,
      "loss": 0.6505,
      "step": 3224
    },
    {
      "epoch": 0.40812452543659833,
      "grad_norm": 1.0277273654937744,
      "learning_rate": 1.847329782334276e-05,
      "loss": 0.6471,
      "step": 3225
    },
    {
      "epoch": 0.4082510756770438,
      "grad_norm": 1.0622191429138184,
      "learning_rate": 1.847234594944296e-05,
      "loss": 0.6841,
      "step": 3226
    },
    {
      "epoch": 0.4083776259174892,
      "grad_norm": 1.0225958824157715,
      "learning_rate": 1.8471393803436234e-05,
      "loss": 0.6754,
      "step": 3227
    },
    {
      "epoch": 0.4085041761579347,
      "grad_norm": 0.9996317028999329,
      "learning_rate": 1.8470441385353168e-05,
      "loss": 0.6311,
      "step": 3228
    },
    {
      "epoch": 0.40863072639838016,
      "grad_norm": 1.2035064697265625,
      "learning_rate": 1.846948869522434e-05,
      "loss": 0.6804,
      "step": 3229
    },
    {
      "epoch": 0.4087572766388256,
      "grad_norm": 1.1441806554794312,
      "learning_rate": 1.8468535733080358e-05,
      "loss": 0.626,
      "step": 3230
    },
    {
      "epoch": 0.40888382687927105,
      "grad_norm": 1.037033200263977,
      "learning_rate": 1.846758249895183e-05,
      "loss": 0.6504,
      "step": 3231
    },
    {
      "epoch": 0.40901037711971655,
      "grad_norm": 1.1364187002182007,
      "learning_rate": 1.846662899286936e-05,
      "loss": 0.6434,
      "step": 3232
    },
    {
      "epoch": 0.409136927360162,
      "grad_norm": 1.045065999031067,
      "learning_rate": 1.8465675214863578e-05,
      "loss": 0.6157,
      "step": 3233
    },
    {
      "epoch": 0.40926347760060744,
      "grad_norm": 1.0937246084213257,
      "learning_rate": 1.846472116496512e-05,
      "loss": 0.6848,
      "step": 3234
    },
    {
      "epoch": 0.4093900278410529,
      "grad_norm": 1.0502619743347168,
      "learning_rate": 1.846376684320462e-05,
      "loss": 0.663,
      "step": 3235
    },
    {
      "epoch": 0.40951657808149833,
      "grad_norm": 1.0260025262832642,
      "learning_rate": 1.846281224961273e-05,
      "loss": 0.6776,
      "step": 3236
    },
    {
      "epoch": 0.40964312832194383,
      "grad_norm": 1.0697742700576782,
      "learning_rate": 1.8461857384220117e-05,
      "loss": 0.6709,
      "step": 3237
    },
    {
      "epoch": 0.4097696785623893,
      "grad_norm": 0.9975712299346924,
      "learning_rate": 1.8460902247057437e-05,
      "loss": 0.6335,
      "step": 3238
    },
    {
      "epoch": 0.4098962288028347,
      "grad_norm": 1.0526072978973389,
      "learning_rate": 1.8459946838155373e-05,
      "loss": 0.6745,
      "step": 3239
    },
    {
      "epoch": 0.41002277904328016,
      "grad_norm": 1.0756182670593262,
      "learning_rate": 1.845899115754461e-05,
      "loss": 0.6488,
      "step": 3240
    },
    {
      "epoch": 0.41014932928372566,
      "grad_norm": 1.0354440212249756,
      "learning_rate": 1.8458035205255837e-05,
      "loss": 0.701,
      "step": 3241
    },
    {
      "epoch": 0.4102758795241711,
      "grad_norm": 1.0215880870819092,
      "learning_rate": 1.845707898131976e-05,
      "loss": 0.6806,
      "step": 3242
    },
    {
      "epoch": 0.41040242976461655,
      "grad_norm": 1.008567214012146,
      "learning_rate": 1.845612248576709e-05,
      "loss": 0.6714,
      "step": 3243
    },
    {
      "epoch": 0.410528980005062,
      "grad_norm": 1.0365089178085327,
      "learning_rate": 1.8455165718628546e-05,
      "loss": 0.6831,
      "step": 3244
    },
    {
      "epoch": 0.4106555302455075,
      "grad_norm": 1.043015718460083,
      "learning_rate": 1.8454208679934857e-05,
      "loss": 0.6157,
      "step": 3245
    },
    {
      "epoch": 0.41078208048595294,
      "grad_norm": 0.9571185111999512,
      "learning_rate": 1.8453251369716757e-05,
      "loss": 0.6419,
      "step": 3246
    },
    {
      "epoch": 0.4109086307263984,
      "grad_norm": 0.9946694374084473,
      "learning_rate": 1.8452293788004997e-05,
      "loss": 0.6042,
      "step": 3247
    },
    {
      "epoch": 0.4110351809668438,
      "grad_norm": 1.0811208486557007,
      "learning_rate": 1.8451335934830328e-05,
      "loss": 0.6657,
      "step": 3248
    },
    {
      "epoch": 0.41116173120728927,
      "grad_norm": 1.1867878437042236,
      "learning_rate": 1.8450377810223517e-05,
      "loss": 0.6986,
      "step": 3249
    },
    {
      "epoch": 0.41128828144773477,
      "grad_norm": 1.022910475730896,
      "learning_rate": 1.8449419414215332e-05,
      "loss": 0.6246,
      "step": 3250
    },
    {
      "epoch": 0.4114148316881802,
      "grad_norm": 2.3407044410705566,
      "learning_rate": 1.8448460746836558e-05,
      "loss": 0.6577,
      "step": 3251
    },
    {
      "epoch": 0.41154138192862566,
      "grad_norm": 1.0669562816619873,
      "learning_rate": 1.8447501808117986e-05,
      "loss": 0.634,
      "step": 3252
    },
    {
      "epoch": 0.4116679321690711,
      "grad_norm": 1.006485939025879,
      "learning_rate": 1.8446542598090403e-05,
      "loss": 0.6501,
      "step": 3253
    },
    {
      "epoch": 0.4117944824095166,
      "grad_norm": 1.0299805402755737,
      "learning_rate": 1.8445583116784632e-05,
      "loss": 0.6432,
      "step": 3254
    },
    {
      "epoch": 0.41192103264996205,
      "grad_norm": 0.9969426989555359,
      "learning_rate": 1.8444623364231478e-05,
      "loss": 0.6255,
      "step": 3255
    },
    {
      "epoch": 0.4120475828904075,
      "grad_norm": 1.1473517417907715,
      "learning_rate": 1.8443663340461766e-05,
      "loss": 0.6805,
      "step": 3256
    },
    {
      "epoch": 0.41217413313085294,
      "grad_norm": 1.1403262615203857,
      "learning_rate": 1.844270304550633e-05,
      "loss": 0.6554,
      "step": 3257
    },
    {
      "epoch": 0.4123006833712984,
      "grad_norm": 1.0796451568603516,
      "learning_rate": 1.8441742479396018e-05,
      "loss": 0.6285,
      "step": 3258
    },
    {
      "epoch": 0.4124272336117439,
      "grad_norm": 1.1100473403930664,
      "learning_rate": 1.8440781642161674e-05,
      "loss": 0.7126,
      "step": 3259
    },
    {
      "epoch": 0.4125537838521893,
      "grad_norm": 1.0439268350601196,
      "learning_rate": 1.843982053383416e-05,
      "loss": 0.664,
      "step": 3260
    },
    {
      "epoch": 0.41268033409263477,
      "grad_norm": 1.1259201765060425,
      "learning_rate": 1.843885915444434e-05,
      "loss": 0.6906,
      "step": 3261
    },
    {
      "epoch": 0.4128068843330802,
      "grad_norm": 1.1520041227340698,
      "learning_rate": 1.8437897504023096e-05,
      "loss": 0.6677,
      "step": 3262
    },
    {
      "epoch": 0.4129334345735257,
      "grad_norm": 1.0641061067581177,
      "learning_rate": 1.843693558260131e-05,
      "loss": 0.6904,
      "step": 3263
    },
    {
      "epoch": 0.41305998481397116,
      "grad_norm": 1.0481367111206055,
      "learning_rate": 1.8435973390209878e-05,
      "loss": 0.6727,
      "step": 3264
    },
    {
      "epoch": 0.4131865350544166,
      "grad_norm": 1.0115565061569214,
      "learning_rate": 1.84350109268797e-05,
      "loss": 0.6762,
      "step": 3265
    },
    {
      "epoch": 0.41331308529486205,
      "grad_norm": 1.040140151977539,
      "learning_rate": 1.8434048192641694e-05,
      "loss": 0.6004,
      "step": 3266
    },
    {
      "epoch": 0.4134396355353075,
      "grad_norm": 1.0976085662841797,
      "learning_rate": 1.8433085187526773e-05,
      "loss": 0.642,
      "step": 3267
    },
    {
      "epoch": 0.413566185775753,
      "grad_norm": 1.0021257400512695,
      "learning_rate": 1.843212191156587e-05,
      "loss": 0.6014,
      "step": 3268
    },
    {
      "epoch": 0.41369273601619844,
      "grad_norm": 1.0169793367385864,
      "learning_rate": 1.843115836478992e-05,
      "loss": 0.6439,
      "step": 3269
    },
    {
      "epoch": 0.4138192862566439,
      "grad_norm": 1.0744960308074951,
      "learning_rate": 1.843019454722987e-05,
      "loss": 0.6572,
      "step": 3270
    },
    {
      "epoch": 0.4139458364970893,
      "grad_norm": 1.161246418952942,
      "learning_rate": 1.8429230458916672e-05,
      "loss": 0.6525,
      "step": 3271
    },
    {
      "epoch": 0.4140723867375348,
      "grad_norm": 1.1052401065826416,
      "learning_rate": 1.84282660998813e-05,
      "loss": 0.6987,
      "step": 3272
    },
    {
      "epoch": 0.41419893697798027,
      "grad_norm": 1.0347111225128174,
      "learning_rate": 1.8427301470154716e-05,
      "loss": 0.659,
      "step": 3273
    },
    {
      "epoch": 0.4143254872184257,
      "grad_norm": 1.0754157304763794,
      "learning_rate": 1.8426336569767905e-05,
      "loss": 0.6437,
      "step": 3274
    },
    {
      "epoch": 0.41445203745887116,
      "grad_norm": 1.0930627584457397,
      "learning_rate": 1.8425371398751858e-05,
      "loss": 0.6588,
      "step": 3275
    },
    {
      "epoch": 0.4145785876993166,
      "grad_norm": 1.0384052991867065,
      "learning_rate": 1.842440595713757e-05,
      "loss": 0.6451,
      "step": 3276
    },
    {
      "epoch": 0.4147051379397621,
      "grad_norm": 1.083706021308899,
      "learning_rate": 1.842344024495605e-05,
      "loss": 0.6585,
      "step": 3277
    },
    {
      "epoch": 0.41483168818020755,
      "grad_norm": 1.052836298942566,
      "learning_rate": 1.8422474262238315e-05,
      "loss": 0.6341,
      "step": 3278
    },
    {
      "epoch": 0.414958238420653,
      "grad_norm": 1.0964864492416382,
      "learning_rate": 1.8421508009015387e-05,
      "loss": 0.6611,
      "step": 3279
    },
    {
      "epoch": 0.41508478866109844,
      "grad_norm": 1.0131479501724243,
      "learning_rate": 1.84205414853183e-05,
      "loss": 0.6527,
      "step": 3280
    },
    {
      "epoch": 0.41521133890154394,
      "grad_norm": 1.0769774913787842,
      "learning_rate": 1.84195746911781e-05,
      "loss": 0.7259,
      "step": 3281
    },
    {
      "epoch": 0.4153378891419894,
      "grad_norm": 1.0690926313400269,
      "learning_rate": 1.8418607626625832e-05,
      "loss": 0.6451,
      "step": 3282
    },
    {
      "epoch": 0.4154644393824348,
      "grad_norm": 1.0546820163726807,
      "learning_rate": 1.841764029169256e-05,
      "loss": 0.6656,
      "step": 3283
    },
    {
      "epoch": 0.41559098962288027,
      "grad_norm": 1.0624217987060547,
      "learning_rate": 1.8416672686409347e-05,
      "loss": 0.6385,
      "step": 3284
    },
    {
      "epoch": 0.4157175398633257,
      "grad_norm": 1.0837849378585815,
      "learning_rate": 1.841570481080727e-05,
      "loss": 0.6574,
      "step": 3285
    },
    {
      "epoch": 0.4158440901037712,
      "grad_norm": 1.1490540504455566,
      "learning_rate": 1.841473666491742e-05,
      "loss": 0.6546,
      "step": 3286
    },
    {
      "epoch": 0.41597064034421666,
      "grad_norm": 1.108541488647461,
      "learning_rate": 1.8413768248770885e-05,
      "loss": 0.674,
      "step": 3287
    },
    {
      "epoch": 0.4160971905846621,
      "grad_norm": 1.0413808822631836,
      "learning_rate": 1.8412799562398772e-05,
      "loss": 0.6273,
      "step": 3288
    },
    {
      "epoch": 0.41622374082510755,
      "grad_norm": 1.1278965473175049,
      "learning_rate": 1.841183060583219e-05,
      "loss": 0.6852,
      "step": 3289
    },
    {
      "epoch": 0.41635029106555305,
      "grad_norm": 1.1131694316864014,
      "learning_rate": 1.8410861379102257e-05,
      "loss": 0.6676,
      "step": 3290
    },
    {
      "epoch": 0.4164768413059985,
      "grad_norm": 1.0306941270828247,
      "learning_rate": 1.840989188224011e-05,
      "loss": 0.6543,
      "step": 3291
    },
    {
      "epoch": 0.41660339154644394,
      "grad_norm": 1.0052138566970825,
      "learning_rate": 1.8408922115276874e-05,
      "loss": 0.6554,
      "step": 3292
    },
    {
      "epoch": 0.4167299417868894,
      "grad_norm": 0.9769648313522339,
      "learning_rate": 1.8407952078243706e-05,
      "loss": 0.6286,
      "step": 3293
    },
    {
      "epoch": 0.4168564920273349,
      "grad_norm": 1.0491201877593994,
      "learning_rate": 1.8406981771171755e-05,
      "loss": 0.6296,
      "step": 3294
    },
    {
      "epoch": 0.4169830422677803,
      "grad_norm": 1.0712871551513672,
      "learning_rate": 1.8406011194092185e-05,
      "loss": 0.6598,
      "step": 3295
    },
    {
      "epoch": 0.41710959250822577,
      "grad_norm": 1.0309923887252808,
      "learning_rate": 1.8405040347036168e-05,
      "loss": 0.6578,
      "step": 3296
    },
    {
      "epoch": 0.4172361427486712,
      "grad_norm": 1.089337706565857,
      "learning_rate": 1.8404069230034888e-05,
      "loss": 0.6545,
      "step": 3297
    },
    {
      "epoch": 0.41736269298911666,
      "grad_norm": 1.059401035308838,
      "learning_rate": 1.840309784311953e-05,
      "loss": 0.6928,
      "step": 3298
    },
    {
      "epoch": 0.41748924322956216,
      "grad_norm": 1.0883984565734863,
      "learning_rate": 1.8402126186321298e-05,
      "loss": 0.7101,
      "step": 3299
    },
    {
      "epoch": 0.4176157934700076,
      "grad_norm": 1.0361326932907104,
      "learning_rate": 1.840115425967139e-05,
      "loss": 0.6693,
      "step": 3300
    },
    {
      "epoch": 0.41774234371045305,
      "grad_norm": 1.126055359840393,
      "learning_rate": 1.840018206320103e-05,
      "loss": 0.6474,
      "step": 3301
    },
    {
      "epoch": 0.4178688939508985,
      "grad_norm": 1.0593388080596924,
      "learning_rate": 1.8399209596941438e-05,
      "loss": 0.6849,
      "step": 3302
    },
    {
      "epoch": 0.417995444191344,
      "grad_norm": 1.1356630325317383,
      "learning_rate": 1.839823686092385e-05,
      "loss": 0.6329,
      "step": 3303
    },
    {
      "epoch": 0.41812199443178943,
      "grad_norm": 1.036507248878479,
      "learning_rate": 1.83972638551795e-05,
      "loss": 0.7005,
      "step": 3304
    },
    {
      "epoch": 0.4182485446722349,
      "grad_norm": 1.4863362312316895,
      "learning_rate": 1.8396290579739644e-05,
      "loss": 0.6981,
      "step": 3305
    },
    {
      "epoch": 0.4183750949126803,
      "grad_norm": 1.0505326986312866,
      "learning_rate": 1.839531703463554e-05,
      "loss": 0.642,
      "step": 3306
    },
    {
      "epoch": 0.41850164515312577,
      "grad_norm": 1.3031657934188843,
      "learning_rate": 1.8394343219898452e-05,
      "loss": 0.6535,
      "step": 3307
    },
    {
      "epoch": 0.41862819539357127,
      "grad_norm": 1.1635587215423584,
      "learning_rate": 1.8393369135559665e-05,
      "loss": 0.6665,
      "step": 3308
    },
    {
      "epoch": 0.4187547456340167,
      "grad_norm": 1.104845643043518,
      "learning_rate": 1.8392394781650454e-05,
      "loss": 0.65,
      "step": 3309
    },
    {
      "epoch": 0.41888129587446216,
      "grad_norm": 1.297116756439209,
      "learning_rate": 1.839142015820212e-05,
      "loss": 0.6276,
      "step": 3310
    },
    {
      "epoch": 0.4190078461149076,
      "grad_norm": 1.0765498876571655,
      "learning_rate": 1.8390445265245958e-05,
      "loss": 0.6329,
      "step": 3311
    },
    {
      "epoch": 0.4191343963553531,
      "grad_norm": 1.0768942832946777,
      "learning_rate": 1.838947010281328e-05,
      "loss": 0.6572,
      "step": 3312
    },
    {
      "epoch": 0.41926094659579854,
      "grad_norm": 1.0137029886245728,
      "learning_rate": 1.838849467093541e-05,
      "loss": 0.6823,
      "step": 3313
    },
    {
      "epoch": 0.419387496836244,
      "grad_norm": 1.0864596366882324,
      "learning_rate": 1.8387518969643676e-05,
      "loss": 0.6749,
      "step": 3314
    },
    {
      "epoch": 0.41951404707668943,
      "grad_norm": 0.9962771534919739,
      "learning_rate": 1.8386542998969407e-05,
      "loss": 0.6424,
      "step": 3315
    },
    {
      "epoch": 0.4196405973171349,
      "grad_norm": 1.0335692167282104,
      "learning_rate": 1.8385566758943956e-05,
      "loss": 0.6593,
      "step": 3316
    },
    {
      "epoch": 0.4197671475575804,
      "grad_norm": 0.9958950877189636,
      "learning_rate": 1.8384590249598674e-05,
      "loss": 0.6507,
      "step": 3317
    },
    {
      "epoch": 0.4198936977980258,
      "grad_norm": 1.120845913887024,
      "learning_rate": 1.8383613470964922e-05,
      "loss": 0.624,
      "step": 3318
    },
    {
      "epoch": 0.42002024803847127,
      "grad_norm": 1.1289750337600708,
      "learning_rate": 1.8382636423074073e-05,
      "loss": 0.6635,
      "step": 3319
    },
    {
      "epoch": 0.4201467982789167,
      "grad_norm": 1.274202823638916,
      "learning_rate": 1.838165910595751e-05,
      "loss": 0.6307,
      "step": 3320
    },
    {
      "epoch": 0.4202733485193622,
      "grad_norm": 1.089768409729004,
      "learning_rate": 1.838068151964662e-05,
      "loss": 0.6887,
      "step": 3321
    },
    {
      "epoch": 0.42039989875980766,
      "grad_norm": 1.0452234745025635,
      "learning_rate": 1.8379703664172796e-05,
      "loss": 0.637,
      "step": 3322
    },
    {
      "epoch": 0.4205264490002531,
      "grad_norm": 1.0248467922210693,
      "learning_rate": 1.8378725539567447e-05,
      "loss": 0.6414,
      "step": 3323
    },
    {
      "epoch": 0.42065299924069854,
      "grad_norm": 1.1199495792388916,
      "learning_rate": 1.8377747145861984e-05,
      "loss": 0.6513,
      "step": 3324
    },
    {
      "epoch": 0.420779549481144,
      "grad_norm": 1.0700539350509644,
      "learning_rate": 1.837676848308784e-05,
      "loss": 0.6618,
      "step": 3325
    },
    {
      "epoch": 0.4209060997215895,
      "grad_norm": 1.0178276300430298,
      "learning_rate": 1.8375789551276436e-05,
      "loss": 0.6473,
      "step": 3326
    },
    {
      "epoch": 0.42103264996203493,
      "grad_norm": 1.0892287492752075,
      "learning_rate": 1.8374810350459214e-05,
      "loss": 0.6579,
      "step": 3327
    },
    {
      "epoch": 0.4211592002024804,
      "grad_norm": 1.197449803352356,
      "learning_rate": 1.837383088066763e-05,
      "loss": 0.6881,
      "step": 3328
    },
    {
      "epoch": 0.4212857504429258,
      "grad_norm": 1.0975373983383179,
      "learning_rate": 1.8372851141933138e-05,
      "loss": 0.6729,
      "step": 3329
    },
    {
      "epoch": 0.4214123006833713,
      "grad_norm": 1.0080722570419312,
      "learning_rate": 1.83718711342872e-05,
      "loss": 0.677,
      "step": 3330
    },
    {
      "epoch": 0.42153885092381677,
      "grad_norm": 1.1914184093475342,
      "learning_rate": 1.8370890857761297e-05,
      "loss": 0.6638,
      "step": 3331
    },
    {
      "epoch": 0.4216654011642622,
      "grad_norm": 1.0814416408538818,
      "learning_rate": 1.836991031238691e-05,
      "loss": 0.6668,
      "step": 3332
    },
    {
      "epoch": 0.42179195140470765,
      "grad_norm": 1.1341063976287842,
      "learning_rate": 1.8368929498195532e-05,
      "loss": 0.657,
      "step": 3333
    },
    {
      "epoch": 0.4219185016451531,
      "grad_norm": 1.0252817869186401,
      "learning_rate": 1.836794841521866e-05,
      "loss": 0.6719,
      "step": 3334
    },
    {
      "epoch": 0.4220450518855986,
      "grad_norm": 1.1699254512786865,
      "learning_rate": 1.8366967063487808e-05,
      "loss": 0.6574,
      "step": 3335
    },
    {
      "epoch": 0.42217160212604404,
      "grad_norm": 1.0397437810897827,
      "learning_rate": 1.8365985443034497e-05,
      "loss": 0.6615,
      "step": 3336
    },
    {
      "epoch": 0.4222981523664895,
      "grad_norm": 1.1312376260757446,
      "learning_rate": 1.8365003553890246e-05,
      "loss": 0.6759,
      "step": 3337
    },
    {
      "epoch": 0.42242470260693493,
      "grad_norm": 1.0337687730789185,
      "learning_rate": 1.8364021396086596e-05,
      "loss": 0.6808,
      "step": 3338
    },
    {
      "epoch": 0.42255125284738043,
      "grad_norm": 1.1176934242248535,
      "learning_rate": 1.8363038969655086e-05,
      "loss": 0.6655,
      "step": 3339
    },
    {
      "epoch": 0.4226778030878259,
      "grad_norm": 1.0462372303009033,
      "learning_rate": 1.8362056274627277e-05,
      "loss": 0.6196,
      "step": 3340
    },
    {
      "epoch": 0.4228043533282713,
      "grad_norm": 1.1951428651809692,
      "learning_rate": 1.8361073311034722e-05,
      "loss": 0.6729,
      "step": 3341
    },
    {
      "epoch": 0.42293090356871677,
      "grad_norm": 1.007615327835083,
      "learning_rate": 1.8360090078908993e-05,
      "loss": 0.6353,
      "step": 3342
    },
    {
      "epoch": 0.42305745380916226,
      "grad_norm": 1.0721940994262695,
      "learning_rate": 1.835910657828167e-05,
      "loss": 0.6186,
      "step": 3343
    },
    {
      "epoch": 0.4231840040496077,
      "grad_norm": 1.0695900917053223,
      "learning_rate": 1.8358122809184344e-05,
      "loss": 0.6694,
      "step": 3344
    },
    {
      "epoch": 0.42331055429005315,
      "grad_norm": 1.027365803718567,
      "learning_rate": 1.8357138771648602e-05,
      "loss": 0.6706,
      "step": 3345
    },
    {
      "epoch": 0.4234371045304986,
      "grad_norm": 1.087697148323059,
      "learning_rate": 1.8356154465706053e-05,
      "loss": 0.6458,
      "step": 3346
    },
    {
      "epoch": 0.42356365477094404,
      "grad_norm": 1.048728585243225,
      "learning_rate": 1.835516989138831e-05,
      "loss": 0.6899,
      "step": 3347
    },
    {
      "epoch": 0.42369020501138954,
      "grad_norm": 1.0591965913772583,
      "learning_rate": 1.8354185048726997e-05,
      "loss": 0.6624,
      "step": 3348
    },
    {
      "epoch": 0.423816755251835,
      "grad_norm": 1.0202052593231201,
      "learning_rate": 1.835319993775374e-05,
      "loss": 0.6302,
      "step": 3349
    },
    {
      "epoch": 0.42394330549228043,
      "grad_norm": 1.0285941362380981,
      "learning_rate": 1.835221455850018e-05,
      "loss": 0.6518,
      "step": 3350
    },
    {
      "epoch": 0.4240698557327259,
      "grad_norm": 1.0713657140731812,
      "learning_rate": 1.8351228910997963e-05,
      "loss": 0.6352,
      "step": 3351
    },
    {
      "epoch": 0.4241964059731714,
      "grad_norm": 1.063902735710144,
      "learning_rate": 1.8350242995278747e-05,
      "loss": 0.6968,
      "step": 3352
    },
    {
      "epoch": 0.4243229562136168,
      "grad_norm": 1.140015959739685,
      "learning_rate": 1.83492568113742e-05,
      "loss": 0.6665,
      "step": 3353
    },
    {
      "epoch": 0.42444950645406226,
      "grad_norm": 1.038071870803833,
      "learning_rate": 1.8348270359315987e-05,
      "loss": 0.7256,
      "step": 3354
    },
    {
      "epoch": 0.4245760566945077,
      "grad_norm": 1.1291675567626953,
      "learning_rate": 1.8347283639135795e-05,
      "loss": 0.656,
      "step": 3355
    },
    {
      "epoch": 0.42470260693495315,
      "grad_norm": 1.0217084884643555,
      "learning_rate": 1.8346296650865314e-05,
      "loss": 0.6509,
      "step": 3356
    },
    {
      "epoch": 0.42482915717539865,
      "grad_norm": 1.0433610677719116,
      "learning_rate": 1.8345309394536244e-05,
      "loss": 0.6473,
      "step": 3357
    },
    {
      "epoch": 0.4249557074158441,
      "grad_norm": 1.0704652070999146,
      "learning_rate": 1.834432187018029e-05,
      "loss": 0.6871,
      "step": 3358
    },
    {
      "epoch": 0.42508225765628954,
      "grad_norm": 0.9865548610687256,
      "learning_rate": 1.8343334077829172e-05,
      "loss": 0.6414,
      "step": 3359
    },
    {
      "epoch": 0.425208807896735,
      "grad_norm": 1.096318006515503,
      "learning_rate": 1.8342346017514617e-05,
      "loss": 0.6475,
      "step": 3360
    },
    {
      "epoch": 0.4253353581371805,
      "grad_norm": 1.0335251092910767,
      "learning_rate": 1.8341357689268347e-05,
      "loss": 0.6447,
      "step": 3361
    },
    {
      "epoch": 0.42546190837762593,
      "grad_norm": 1.0797291994094849,
      "learning_rate": 1.8340369093122118e-05,
      "loss": 0.659,
      "step": 3362
    },
    {
      "epoch": 0.4255884586180714,
      "grad_norm": 1.0269932746887207,
      "learning_rate": 1.8339380229107672e-05,
      "loss": 0.65,
      "step": 3363
    },
    {
      "epoch": 0.4257150088585168,
      "grad_norm": 0.9808313846588135,
      "learning_rate": 1.8338391097256773e-05,
      "loss": 0.6174,
      "step": 3364
    },
    {
      "epoch": 0.42584155909896226,
      "grad_norm": 1.2062054872512817,
      "learning_rate": 1.8337401697601185e-05,
      "loss": 0.7083,
      "step": 3365
    },
    {
      "epoch": 0.42596810933940776,
      "grad_norm": 1.024771809577942,
      "learning_rate": 1.8336412030172688e-05,
      "loss": 0.6395,
      "step": 3366
    },
    {
      "epoch": 0.4260946595798532,
      "grad_norm": 1.130638837814331,
      "learning_rate": 1.8335422095003067e-05,
      "loss": 0.6414,
      "step": 3367
    },
    {
      "epoch": 0.42622120982029865,
      "grad_norm": 1.0194361209869385,
      "learning_rate": 1.8334431892124113e-05,
      "loss": 0.6605,
      "step": 3368
    },
    {
      "epoch": 0.4263477600607441,
      "grad_norm": 1.028486967086792,
      "learning_rate": 1.8333441421567632e-05,
      "loss": 0.6311,
      "step": 3369
    },
    {
      "epoch": 0.4264743103011896,
      "grad_norm": 1.023818850517273,
      "learning_rate": 1.8332450683365434e-05,
      "loss": 0.6698,
      "step": 3370
    },
    {
      "epoch": 0.42660086054163504,
      "grad_norm": 1.0422108173370361,
      "learning_rate": 1.8331459677549335e-05,
      "loss": 0.6508,
      "step": 3371
    },
    {
      "epoch": 0.4267274107820805,
      "grad_norm": 1.0317667722702026,
      "learning_rate": 1.833046840415117e-05,
      "loss": 0.6377,
      "step": 3372
    },
    {
      "epoch": 0.42685396102252593,
      "grad_norm": 1.1612333059310913,
      "learning_rate": 1.8329476863202768e-05,
      "loss": 0.6438,
      "step": 3373
    },
    {
      "epoch": 0.4269805112629714,
      "grad_norm": 1.8474048376083374,
      "learning_rate": 1.8328485054735982e-05,
      "loss": 0.7463,
      "step": 3374
    },
    {
      "epoch": 0.4271070615034169,
      "grad_norm": 1.0511761903762817,
      "learning_rate": 1.832749297878266e-05,
      "loss": 0.6179,
      "step": 3375
    },
    {
      "epoch": 0.4272336117438623,
      "grad_norm": 1.0720878839492798,
      "learning_rate": 1.8326500635374665e-05,
      "loss": 0.6905,
      "step": 3376
    },
    {
      "epoch": 0.42736016198430776,
      "grad_norm": 1.1417826414108276,
      "learning_rate": 1.8325508024543875e-05,
      "loss": 0.6496,
      "step": 3377
    },
    {
      "epoch": 0.4274867122247532,
      "grad_norm": 1.2553162574768066,
      "learning_rate": 1.832451514632216e-05,
      "loss": 0.643,
      "step": 3378
    },
    {
      "epoch": 0.4276132624651987,
      "grad_norm": 0.9942009449005127,
      "learning_rate": 1.8323522000741417e-05,
      "loss": 0.6547,
      "step": 3379
    },
    {
      "epoch": 0.42773981270564415,
      "grad_norm": 1.0498970746994019,
      "learning_rate": 1.832252858783354e-05,
      "loss": 0.6583,
      "step": 3380
    },
    {
      "epoch": 0.4278663629460896,
      "grad_norm": 1.0842331647872925,
      "learning_rate": 1.832153490763043e-05,
      "loss": 0.6563,
      "step": 3381
    },
    {
      "epoch": 0.42799291318653504,
      "grad_norm": 1.08688223361969,
      "learning_rate": 1.8320540960164007e-05,
      "loss": 0.645,
      "step": 3382
    },
    {
      "epoch": 0.4281194634269805,
      "grad_norm": 1.0754328966140747,
      "learning_rate": 1.8319546745466188e-05,
      "loss": 0.6552,
      "step": 3383
    },
    {
      "epoch": 0.428246013667426,
      "grad_norm": 1.1599172353744507,
      "learning_rate": 1.831855226356891e-05,
      "loss": 0.6858,
      "step": 3384
    },
    {
      "epoch": 0.42837256390787143,
      "grad_norm": 1.0912997722625732,
      "learning_rate": 1.831755751450411e-05,
      "loss": 0.666,
      "step": 3385
    },
    {
      "epoch": 0.4284991141483169,
      "grad_norm": 1.5459743738174438,
      "learning_rate": 1.831656249830374e-05,
      "loss": 0.718,
      "step": 3386
    },
    {
      "epoch": 0.4286256643887623,
      "grad_norm": 1.0518476963043213,
      "learning_rate": 1.8315567214999748e-05,
      "loss": 0.6138,
      "step": 3387
    },
    {
      "epoch": 0.4287522146292078,
      "grad_norm": 1.1716803312301636,
      "learning_rate": 1.831457166462411e-05,
      "loss": 0.6402,
      "step": 3388
    },
    {
      "epoch": 0.42887876486965326,
      "grad_norm": 0.9888066053390503,
      "learning_rate": 1.8313575847208797e-05,
      "loss": 0.6555,
      "step": 3389
    },
    {
      "epoch": 0.4290053151100987,
      "grad_norm": 1.2509783506393433,
      "learning_rate": 1.8312579762785784e-05,
      "loss": 0.6916,
      "step": 3390
    },
    {
      "epoch": 0.42913186535054415,
      "grad_norm": 1.0792813301086426,
      "learning_rate": 1.8311583411387076e-05,
      "loss": 0.6943,
      "step": 3391
    },
    {
      "epoch": 0.42925841559098965,
      "grad_norm": 1.0602021217346191,
      "learning_rate": 1.8310586793044662e-05,
      "loss": 0.6832,
      "step": 3392
    },
    {
      "epoch": 0.4293849658314351,
      "grad_norm": 1.053147315979004,
      "learning_rate": 1.830958990779055e-05,
      "loss": 0.623,
      "step": 3393
    },
    {
      "epoch": 0.42951151607188054,
      "grad_norm": 1.0312350988388062,
      "learning_rate": 1.830859275565677e-05,
      "loss": 0.6606,
      "step": 3394
    },
    {
      "epoch": 0.429638066312326,
      "grad_norm": 1.1678457260131836,
      "learning_rate": 1.8307595336675336e-05,
      "loss": 0.6515,
      "step": 3395
    },
    {
      "epoch": 0.42976461655277143,
      "grad_norm": 1.0186891555786133,
      "learning_rate": 1.830659765087828e-05,
      "loss": 0.6044,
      "step": 3396
    },
    {
      "epoch": 0.42989116679321693,
      "grad_norm": 1.0275965929031372,
      "learning_rate": 1.8305599698297652e-05,
      "loss": 0.6688,
      "step": 3397
    },
    {
      "epoch": 0.43001771703366237,
      "grad_norm": 1.044425129890442,
      "learning_rate": 1.8304601478965504e-05,
      "loss": 0.6485,
      "step": 3398
    },
    {
      "epoch": 0.4301442672741078,
      "grad_norm": 1.0266170501708984,
      "learning_rate": 1.830360299291389e-05,
      "loss": 0.6206,
      "step": 3399
    },
    {
      "epoch": 0.43027081751455326,
      "grad_norm": 1.0461266040802002,
      "learning_rate": 1.830260424017488e-05,
      "loss": 0.6368,
      "step": 3400
    },
    {
      "epoch": 0.43039736775499876,
      "grad_norm": 1.0026519298553467,
      "learning_rate": 1.8301605220780555e-05,
      "loss": 0.6419,
      "step": 3401
    },
    {
      "epoch": 0.4305239179954442,
      "grad_norm": 0.9455111622810364,
      "learning_rate": 1.8300605934762995e-05,
      "loss": 0.6218,
      "step": 3402
    },
    {
      "epoch": 0.43065046823588965,
      "grad_norm": 1.074019193649292,
      "learning_rate": 1.82996063821543e-05,
      "loss": 0.6552,
      "step": 3403
    },
    {
      "epoch": 0.4307770184763351,
      "grad_norm": 1.0141286849975586,
      "learning_rate": 1.829860656298657e-05,
      "loss": 0.643,
      "step": 3404
    },
    {
      "epoch": 0.43090356871678054,
      "grad_norm": 1.022855520248413,
      "learning_rate": 1.829760647729191e-05,
      "loss": 0.6402,
      "step": 3405
    },
    {
      "epoch": 0.43103011895722604,
      "grad_norm": 1.0064784288406372,
      "learning_rate": 1.8296606125102453e-05,
      "loss": 0.6402,
      "step": 3406
    },
    {
      "epoch": 0.4311566691976715,
      "grad_norm": 1.1125068664550781,
      "learning_rate": 1.8295605506450315e-05,
      "loss": 0.6798,
      "step": 3407
    },
    {
      "epoch": 0.4312832194381169,
      "grad_norm": 1.1181949377059937,
      "learning_rate": 1.8294604621367643e-05,
      "loss": 0.6271,
      "step": 3408
    },
    {
      "epoch": 0.43140976967856237,
      "grad_norm": 1.0811539888381958,
      "learning_rate": 1.8293603469886573e-05,
      "loss": 0.6689,
      "step": 3409
    },
    {
      "epoch": 0.43153631991900787,
      "grad_norm": 1.007168173789978,
      "learning_rate": 1.8292602052039265e-05,
      "loss": 0.6532,
      "step": 3410
    },
    {
      "epoch": 0.4316628701594533,
      "grad_norm": 1.0271369218826294,
      "learning_rate": 1.8291600367857884e-05,
      "loss": 0.6562,
      "step": 3411
    },
    {
      "epoch": 0.43178942039989876,
      "grad_norm": 1.0417978763580322,
      "learning_rate": 1.8290598417374592e-05,
      "loss": 0.6855,
      "step": 3412
    },
    {
      "epoch": 0.4319159706403442,
      "grad_norm": 1.1341822147369385,
      "learning_rate": 1.8289596200621574e-05,
      "loss": 0.6725,
      "step": 3413
    },
    {
      "epoch": 0.43204252088078965,
      "grad_norm": 1.0831360816955566,
      "learning_rate": 1.8288593717631023e-05,
      "loss": 0.6876,
      "step": 3414
    },
    {
      "epoch": 0.43216907112123515,
      "grad_norm": 1.1396985054016113,
      "learning_rate": 1.8287590968435127e-05,
      "loss": 0.6755,
      "step": 3415
    },
    {
      "epoch": 0.4322956213616806,
      "grad_norm": 1.0663349628448486,
      "learning_rate": 1.82865879530661e-05,
      "loss": 0.6622,
      "step": 3416
    },
    {
      "epoch": 0.43242217160212604,
      "grad_norm": 1.086037516593933,
      "learning_rate": 1.8285584671556146e-05,
      "loss": 0.6323,
      "step": 3417
    },
    {
      "epoch": 0.4325487218425715,
      "grad_norm": 1.0802186727523804,
      "learning_rate": 1.8284581123937497e-05,
      "loss": 0.6564,
      "step": 3418
    },
    {
      "epoch": 0.432675272083017,
      "grad_norm": 1.0888352394104004,
      "learning_rate": 1.8283577310242377e-05,
      "loss": 0.6562,
      "step": 3419
    },
    {
      "epoch": 0.4328018223234624,
      "grad_norm": 1.0256433486938477,
      "learning_rate": 1.828257323050303e-05,
      "loss": 0.6073,
      "step": 3420
    },
    {
      "epoch": 0.43292837256390787,
      "grad_norm": 0.9727943539619446,
      "learning_rate": 1.8281568884751704e-05,
      "loss": 0.6388,
      "step": 3421
    },
    {
      "epoch": 0.4330549228043533,
      "grad_norm": 1.1003700494766235,
      "learning_rate": 1.8280564273020654e-05,
      "loss": 0.6438,
      "step": 3422
    },
    {
      "epoch": 0.43318147304479876,
      "grad_norm": 1.10560941696167,
      "learning_rate": 1.8279559395342143e-05,
      "loss": 0.6559,
      "step": 3423
    },
    {
      "epoch": 0.43330802328524426,
      "grad_norm": 1.033234715461731,
      "learning_rate": 1.8278554251748447e-05,
      "loss": 0.6624,
      "step": 3424
    },
    {
      "epoch": 0.4334345735256897,
      "grad_norm": 1.0684884786605835,
      "learning_rate": 1.827754884227185e-05,
      "loss": 0.6422,
      "step": 3425
    },
    {
      "epoch": 0.43356112376613515,
      "grad_norm": 1.284649133682251,
      "learning_rate": 1.8276543166944642e-05,
      "loss": 0.6573,
      "step": 3426
    },
    {
      "epoch": 0.4336876740065806,
      "grad_norm": 1.0501270294189453,
      "learning_rate": 1.8275537225799122e-05,
      "loss": 0.6846,
      "step": 3427
    },
    {
      "epoch": 0.4338142242470261,
      "grad_norm": 1.1874053478240967,
      "learning_rate": 1.8274531018867595e-05,
      "loss": 0.6711,
      "step": 3428
    },
    {
      "epoch": 0.43394077448747154,
      "grad_norm": 1.0871448516845703,
      "learning_rate": 1.8273524546182382e-05,
      "loss": 0.6095,
      "step": 3429
    },
    {
      "epoch": 0.434067324727917,
      "grad_norm": 1.111251950263977,
      "learning_rate": 1.8272517807775804e-05,
      "loss": 0.631,
      "step": 3430
    },
    {
      "epoch": 0.4341938749683624,
      "grad_norm": 1.085345983505249,
      "learning_rate": 1.82715108036802e-05,
      "loss": 0.6512,
      "step": 3431
    },
    {
      "epoch": 0.43432042520880787,
      "grad_norm": 1.1527870893478394,
      "learning_rate": 1.8270503533927905e-05,
      "loss": 0.648,
      "step": 3432
    },
    {
      "epoch": 0.43444697544925337,
      "grad_norm": 0.9924759864807129,
      "learning_rate": 1.8269495998551273e-05,
      "loss": 0.6809,
      "step": 3433
    },
    {
      "epoch": 0.4345735256896988,
      "grad_norm": 1.0286539793014526,
      "learning_rate": 1.8268488197582663e-05,
      "loss": 0.6463,
      "step": 3434
    },
    {
      "epoch": 0.43470007593014426,
      "grad_norm": 1.0460678339004517,
      "learning_rate": 1.826748013105444e-05,
      "loss": 0.6023,
      "step": 3435
    },
    {
      "epoch": 0.4348266261705897,
      "grad_norm": 1.123197078704834,
      "learning_rate": 1.826647179899899e-05,
      "loss": 0.6332,
      "step": 3436
    },
    {
      "epoch": 0.4349531764110352,
      "grad_norm": 1.2395833730697632,
      "learning_rate": 1.8265463201448686e-05,
      "loss": 0.69,
      "step": 3437
    },
    {
      "epoch": 0.43507972665148065,
      "grad_norm": 0.9960422515869141,
      "learning_rate": 1.8264454338435922e-05,
      "loss": 0.6417,
      "step": 3438
    },
    {
      "epoch": 0.4352062768919261,
      "grad_norm": 1.0988210439682007,
      "learning_rate": 1.826344520999311e-05,
      "loss": 0.6428,
      "step": 3439
    },
    {
      "epoch": 0.43533282713237154,
      "grad_norm": 1.00052809715271,
      "learning_rate": 1.826243581615265e-05,
      "loss": 0.6478,
      "step": 3440
    },
    {
      "epoch": 0.43545937737281704,
      "grad_norm": 1.0128724575042725,
      "learning_rate": 1.8261426156946967e-05,
      "loss": 0.5967,
      "step": 3441
    },
    {
      "epoch": 0.4355859276132625,
      "grad_norm": 1.1452668905258179,
      "learning_rate": 1.826041623240848e-05,
      "loss": 0.624,
      "step": 3442
    },
    {
      "epoch": 0.4357124778537079,
      "grad_norm": 0.9999600052833557,
      "learning_rate": 1.8259406042569633e-05,
      "loss": 0.6754,
      "step": 3443
    },
    {
      "epoch": 0.43583902809415337,
      "grad_norm": 1.0484496355056763,
      "learning_rate": 1.825839558746287e-05,
      "loss": 0.6851,
      "step": 3444
    },
    {
      "epoch": 0.4359655783345988,
      "grad_norm": 0.9838557839393616,
      "learning_rate": 1.825738486712064e-05,
      "loss": 0.6369,
      "step": 3445
    },
    {
      "epoch": 0.4360921285750443,
      "grad_norm": 1.0113316774368286,
      "learning_rate": 1.8256373881575408e-05,
      "loss": 0.6327,
      "step": 3446
    },
    {
      "epoch": 0.43621867881548976,
      "grad_norm": 0.9934133291244507,
      "learning_rate": 1.825536263085964e-05,
      "loss": 0.6357,
      "step": 3447
    },
    {
      "epoch": 0.4363452290559352,
      "grad_norm": 1.0482301712036133,
      "learning_rate": 1.8254351115005816e-05,
      "loss": 0.6412,
      "step": 3448
    },
    {
      "epoch": 0.43647177929638065,
      "grad_norm": 1.0968645811080933,
      "learning_rate": 1.8253339334046423e-05,
      "loss": 0.6181,
      "step": 3449
    },
    {
      "epoch": 0.43659832953682615,
      "grad_norm": 1.0756559371948242,
      "learning_rate": 1.825232728801396e-05,
      "loss": 0.6733,
      "step": 3450
    },
    {
      "epoch": 0.4367248797772716,
      "grad_norm": 1.061323642730713,
      "learning_rate": 1.825131497694092e-05,
      "loss": 0.6764,
      "step": 3451
    },
    {
      "epoch": 0.43685143001771704,
      "grad_norm": 1.1218111515045166,
      "learning_rate": 1.8250302400859832e-05,
      "loss": 0.6748,
      "step": 3452
    },
    {
      "epoch": 0.4369779802581625,
      "grad_norm": 0.9927155375480652,
      "learning_rate": 1.8249289559803202e-05,
      "loss": 0.6356,
      "step": 3453
    },
    {
      "epoch": 0.4371045304986079,
      "grad_norm": 1.0765554904937744,
      "learning_rate": 1.8248276453803567e-05,
      "loss": 0.658,
      "step": 3454
    },
    {
      "epoch": 0.4372310807390534,
      "grad_norm": 1.139829158782959,
      "learning_rate": 1.8247263082893466e-05,
      "loss": 0.7113,
      "step": 3455
    },
    {
      "epoch": 0.43735763097949887,
      "grad_norm": 1.0052869319915771,
      "learning_rate": 1.8246249447105442e-05,
      "loss": 0.633,
      "step": 3456
    },
    {
      "epoch": 0.4374841812199443,
      "grad_norm": 1.018242597579956,
      "learning_rate": 1.8245235546472053e-05,
      "loss": 0.6392,
      "step": 3457
    },
    {
      "epoch": 0.43761073146038976,
      "grad_norm": 1.8514626026153564,
      "learning_rate": 1.8244221381025857e-05,
      "loss": 0.69,
      "step": 3458
    },
    {
      "epoch": 0.43773728170083526,
      "grad_norm": 1.0649828910827637,
      "learning_rate": 1.824320695079943e-05,
      "loss": 0.6485,
      "step": 3459
    },
    {
      "epoch": 0.4378638319412807,
      "grad_norm": 1.0710225105285645,
      "learning_rate": 1.8242192255825354e-05,
      "loss": 0.6648,
      "step": 3460
    },
    {
      "epoch": 0.43799038218172615,
      "grad_norm": 1.0285509824752808,
      "learning_rate": 1.8241177296136215e-05,
      "loss": 0.6482,
      "step": 3461
    },
    {
      "epoch": 0.4381169324221716,
      "grad_norm": 1.0604625940322876,
      "learning_rate": 1.824016207176461e-05,
      "loss": 0.6758,
      "step": 3462
    },
    {
      "epoch": 0.43824348266261703,
      "grad_norm": 1.083574652671814,
      "learning_rate": 1.823914658274315e-05,
      "loss": 0.6369,
      "step": 3463
    },
    {
      "epoch": 0.43837003290306253,
      "grad_norm": 1.0591503381729126,
      "learning_rate": 1.8238130829104444e-05,
      "loss": 0.6778,
      "step": 3464
    },
    {
      "epoch": 0.438496583143508,
      "grad_norm": 1.0678844451904297,
      "learning_rate": 1.8237114810881117e-05,
      "loss": 0.6783,
      "step": 3465
    },
    {
      "epoch": 0.4386231333839534,
      "grad_norm": 1.1193602085113525,
      "learning_rate": 1.8236098528105802e-05,
      "loss": 0.582,
      "step": 3466
    },
    {
      "epoch": 0.43874968362439887,
      "grad_norm": 1.149772047996521,
      "learning_rate": 1.8235081980811138e-05,
      "loss": 0.7005,
      "step": 3467
    },
    {
      "epoch": 0.43887623386484437,
      "grad_norm": 1.1162923574447632,
      "learning_rate": 1.8234065169029775e-05,
      "loss": 0.6504,
      "step": 3468
    },
    {
      "epoch": 0.4390027841052898,
      "grad_norm": 1.0058400630950928,
      "learning_rate": 1.8233048092794365e-05,
      "loss": 0.6518,
      "step": 3469
    },
    {
      "epoch": 0.43912933434573526,
      "grad_norm": 1.6459484100341797,
      "learning_rate": 1.823203075213758e-05,
      "loss": 0.7097,
      "step": 3470
    },
    {
      "epoch": 0.4392558845861807,
      "grad_norm": 1.1038315296173096,
      "learning_rate": 1.823101314709209e-05,
      "loss": 0.6714,
      "step": 3471
    },
    {
      "epoch": 0.43938243482662614,
      "grad_norm": 0.9566551446914673,
      "learning_rate": 1.8229995277690576e-05,
      "loss": 0.6023,
      "step": 3472
    },
    {
      "epoch": 0.43950898506707164,
      "grad_norm": 1.0233265161514282,
      "learning_rate": 1.8228977143965733e-05,
      "loss": 0.6357,
      "step": 3473
    },
    {
      "epoch": 0.4396355353075171,
      "grad_norm": 1.0637134313583374,
      "learning_rate": 1.8227958745950255e-05,
      "loss": 0.631,
      "step": 3474
    },
    {
      "epoch": 0.43976208554796253,
      "grad_norm": 1.0363184213638306,
      "learning_rate": 1.822694008367686e-05,
      "loss": 0.6695,
      "step": 3475
    },
    {
      "epoch": 0.439888635788408,
      "grad_norm": 1.10686194896698,
      "learning_rate": 1.8225921157178255e-05,
      "loss": 0.7027,
      "step": 3476
    },
    {
      "epoch": 0.4400151860288535,
      "grad_norm": 0.9854366779327393,
      "learning_rate": 1.8224901966487173e-05,
      "loss": 0.6923,
      "step": 3477
    },
    {
      "epoch": 0.4401417362692989,
      "grad_norm": 1.1008752584457397,
      "learning_rate": 1.8223882511636338e-05,
      "loss": 0.6544,
      "step": 3478
    },
    {
      "epoch": 0.44026828650974437,
      "grad_norm": 1.027188777923584,
      "learning_rate": 1.8222862792658495e-05,
      "loss": 0.6473,
      "step": 3479
    },
    {
      "epoch": 0.4403948367501898,
      "grad_norm": 1.1654289960861206,
      "learning_rate": 1.8221842809586397e-05,
      "loss": 0.6875,
      "step": 3480
    },
    {
      "epoch": 0.44052138699063526,
      "grad_norm": 1.0651624202728271,
      "learning_rate": 1.82208225624528e-05,
      "loss": 0.6525,
      "step": 3481
    },
    {
      "epoch": 0.44064793723108076,
      "grad_norm": 1.1145713329315186,
      "learning_rate": 1.821980205129048e-05,
      "loss": 0.686,
      "step": 3482
    },
    {
      "epoch": 0.4407744874715262,
      "grad_norm": 0.9889352321624756,
      "learning_rate": 1.82187812761322e-05,
      "loss": 0.5814,
      "step": 3483
    },
    {
      "epoch": 0.44090103771197164,
      "grad_norm": 1.082283616065979,
      "learning_rate": 1.8217760237010753e-05,
      "loss": 0.6683,
      "step": 3484
    },
    {
      "epoch": 0.4410275879524171,
      "grad_norm": 1.2264595031738281,
      "learning_rate": 1.821673893395893e-05,
      "loss": 0.6538,
      "step": 3485
    },
    {
      "epoch": 0.4411541381928626,
      "grad_norm": 1.121049165725708,
      "learning_rate": 1.8215717367009526e-05,
      "loss": 0.6317,
      "step": 3486
    },
    {
      "epoch": 0.44128068843330803,
      "grad_norm": 1.1513875722885132,
      "learning_rate": 1.8214695536195362e-05,
      "loss": 0.684,
      "step": 3487
    },
    {
      "epoch": 0.4414072386737535,
      "grad_norm": 1.1192246675491333,
      "learning_rate": 1.8213673441549247e-05,
      "loss": 0.6034,
      "step": 3488
    },
    {
      "epoch": 0.4415337889141989,
      "grad_norm": 1.0721633434295654,
      "learning_rate": 1.821265108310401e-05,
      "loss": 0.6415,
      "step": 3489
    },
    {
      "epoch": 0.4416603391546444,
      "grad_norm": 1.081709384918213,
      "learning_rate": 1.8211628460892487e-05,
      "loss": 0.6285,
      "step": 3490
    },
    {
      "epoch": 0.44178688939508987,
      "grad_norm": 1.1203186511993408,
      "learning_rate": 1.8210605574947525e-05,
      "loss": 0.6916,
      "step": 3491
    },
    {
      "epoch": 0.4419134396355353,
      "grad_norm": 1.0343222618103027,
      "learning_rate": 1.820958242530197e-05,
      "loss": 0.6565,
      "step": 3492
    },
    {
      "epoch": 0.44203998987598075,
      "grad_norm": 1.0234657526016235,
      "learning_rate": 1.8208559011988686e-05,
      "loss": 0.6125,
      "step": 3493
    },
    {
      "epoch": 0.4421665401164262,
      "grad_norm": 1.0168932676315308,
      "learning_rate": 1.8207535335040542e-05,
      "loss": 0.6639,
      "step": 3494
    },
    {
      "epoch": 0.4422930903568717,
      "grad_norm": 1.1451226472854614,
      "learning_rate": 1.8206511394490416e-05,
      "loss": 0.6516,
      "step": 3495
    },
    {
      "epoch": 0.44241964059731714,
      "grad_norm": 1.1177213191986084,
      "learning_rate": 1.820548719037119e-05,
      "loss": 0.6431,
      "step": 3496
    },
    {
      "epoch": 0.4425461908377626,
      "grad_norm": 1.0095741748809814,
      "learning_rate": 1.8204462722715762e-05,
      "loss": 0.6511,
      "step": 3497
    },
    {
      "epoch": 0.44267274107820803,
      "grad_norm": 1.0313482284545898,
      "learning_rate": 1.820343799155704e-05,
      "loss": 0.6882,
      "step": 3498
    },
    {
      "epoch": 0.44279929131865353,
      "grad_norm": 1.0760951042175293,
      "learning_rate": 1.820241299692792e-05,
      "loss": 0.7004,
      "step": 3499
    },
    {
      "epoch": 0.442925841559099,
      "grad_norm": 1.0424748659133911,
      "learning_rate": 1.8201387738861337e-05,
      "loss": 0.6163,
      "step": 3500
    },
    {
      "epoch": 0.4430523917995444,
      "grad_norm": 1.0922242403030396,
      "learning_rate": 1.8200362217390212e-05,
      "loss": 0.6746,
      "step": 3501
    },
    {
      "epoch": 0.44317894203998986,
      "grad_norm": 1.1185551881790161,
      "learning_rate": 1.8199336432547487e-05,
      "loss": 0.6724,
      "step": 3502
    },
    {
      "epoch": 0.4433054922804353,
      "grad_norm": 1.0910882949829102,
      "learning_rate": 1.8198310384366098e-05,
      "loss": 0.6439,
      "step": 3503
    },
    {
      "epoch": 0.4434320425208808,
      "grad_norm": 1.0714894533157349,
      "learning_rate": 1.819728407287901e-05,
      "loss": 0.6697,
      "step": 3504
    },
    {
      "epoch": 0.44355859276132625,
      "grad_norm": 1.1411677598953247,
      "learning_rate": 1.8196257498119175e-05,
      "loss": 0.6627,
      "step": 3505
    },
    {
      "epoch": 0.4436851430017717,
      "grad_norm": 1.0620627403259277,
      "learning_rate": 1.8195230660119566e-05,
      "loss": 0.7037,
      "step": 3506
    },
    {
      "epoch": 0.44381169324221714,
      "grad_norm": 1.184406042098999,
      "learning_rate": 1.819420355891317e-05,
      "loss": 0.6495,
      "step": 3507
    },
    {
      "epoch": 0.44393824348266264,
      "grad_norm": 1.1063004732131958,
      "learning_rate": 1.8193176194532966e-05,
      "loss": 0.6659,
      "step": 3508
    },
    {
      "epoch": 0.4440647937231081,
      "grad_norm": 1.4074866771697998,
      "learning_rate": 1.8192148567011952e-05,
      "loss": 0.6781,
      "step": 3509
    },
    {
      "epoch": 0.44419134396355353,
      "grad_norm": 1.0621230602264404,
      "learning_rate": 1.8191120676383134e-05,
      "loss": 0.6339,
      "step": 3510
    },
    {
      "epoch": 0.444317894203999,
      "grad_norm": 1.132812738418579,
      "learning_rate": 1.8190092522679517e-05,
      "loss": 0.6338,
      "step": 3511
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 1.0819836854934692,
      "learning_rate": 1.8189064105934138e-05,
      "loss": 0.6476,
      "step": 3512
    },
    {
      "epoch": 0.4445709946848899,
      "grad_norm": 1.0814540386199951,
      "learning_rate": 1.8188035426180015e-05,
      "loss": 0.6815,
      "step": 3513
    },
    {
      "epoch": 0.44469754492533536,
      "grad_norm": 1.1612694263458252,
      "learning_rate": 1.818700648345019e-05,
      "loss": 0.6787,
      "step": 3514
    },
    {
      "epoch": 0.4448240951657808,
      "grad_norm": 0.9869551658630371,
      "learning_rate": 1.81859772777777e-05,
      "loss": 0.6362,
      "step": 3515
    },
    {
      "epoch": 0.44495064540622625,
      "grad_norm": 1.0606868267059326,
      "learning_rate": 1.8184947809195617e-05,
      "loss": 0.6334,
      "step": 3516
    },
    {
      "epoch": 0.44507719564667175,
      "grad_norm": 1.0048929452896118,
      "learning_rate": 1.8183918077736997e-05,
      "loss": 0.6262,
      "step": 3517
    },
    {
      "epoch": 0.4452037458871172,
      "grad_norm": 1.0983184576034546,
      "learning_rate": 1.8182888083434906e-05,
      "loss": 0.7134,
      "step": 3518
    },
    {
      "epoch": 0.44533029612756264,
      "grad_norm": 1.1194813251495361,
      "learning_rate": 1.8181857826322435e-05,
      "loss": 0.6625,
      "step": 3519
    },
    {
      "epoch": 0.4454568463680081,
      "grad_norm": 0.9751540422439575,
      "learning_rate": 1.818082730643266e-05,
      "loss": 0.6635,
      "step": 3520
    },
    {
      "epoch": 0.44558339660845353,
      "grad_norm": 1.1299583911895752,
      "learning_rate": 1.817979652379869e-05,
      "loss": 0.6679,
      "step": 3521
    },
    {
      "epoch": 0.44570994684889903,
      "grad_norm": 0.9962847232818604,
      "learning_rate": 1.8178765478453625e-05,
      "loss": 0.5961,
      "step": 3522
    },
    {
      "epoch": 0.4458364970893445,
      "grad_norm": 1.0900784730911255,
      "learning_rate": 1.817773417043058e-05,
      "loss": 0.6837,
      "step": 3523
    },
    {
      "epoch": 0.4459630473297899,
      "grad_norm": 1.0024892091751099,
      "learning_rate": 1.817670259976268e-05,
      "loss": 0.6669,
      "step": 3524
    },
    {
      "epoch": 0.44608959757023536,
      "grad_norm": 1.056160807609558,
      "learning_rate": 1.8175670766483052e-05,
      "loss": 0.6541,
      "step": 3525
    },
    {
      "epoch": 0.44621614781068086,
      "grad_norm": 1.035613775253296,
      "learning_rate": 1.817463867062484e-05,
      "loss": 0.6425,
      "step": 3526
    },
    {
      "epoch": 0.4463426980511263,
      "grad_norm": 1.123332142829895,
      "learning_rate": 1.8173606312221187e-05,
      "loss": 0.6647,
      "step": 3527
    },
    {
      "epoch": 0.44646924829157175,
      "grad_norm": 1.0141135454177856,
      "learning_rate": 1.817257369130525e-05,
      "loss": 0.6743,
      "step": 3528
    },
    {
      "epoch": 0.4465957985320172,
      "grad_norm": 1.0412395000457764,
      "learning_rate": 1.81715408079102e-05,
      "loss": 0.657,
      "step": 3529
    },
    {
      "epoch": 0.44672234877246264,
      "grad_norm": 1.0328269004821777,
      "learning_rate": 1.8170507662069203e-05,
      "loss": 0.6492,
      "step": 3530
    },
    {
      "epoch": 0.44684889901290814,
      "grad_norm": 0.9959753751754761,
      "learning_rate": 1.8169474253815444e-05,
      "loss": 0.6802,
      "step": 3531
    },
    {
      "epoch": 0.4469754492533536,
      "grad_norm": 1.052844762802124,
      "learning_rate": 1.816844058318211e-05,
      "loss": 0.664,
      "step": 3532
    },
    {
      "epoch": 0.44710199949379903,
      "grad_norm": 1.1197655200958252,
      "learning_rate": 1.8167406650202407e-05,
      "loss": 0.6007,
      "step": 3533
    },
    {
      "epoch": 0.4472285497342445,
      "grad_norm": 1.1233497858047485,
      "learning_rate": 1.816637245490953e-05,
      "loss": 0.7034,
      "step": 3534
    },
    {
      "epoch": 0.44735509997469,
      "grad_norm": 1.1005915403366089,
      "learning_rate": 1.8165337997336706e-05,
      "loss": 0.6742,
      "step": 3535
    },
    {
      "epoch": 0.4474816502151354,
      "grad_norm": 1.226944923400879,
      "learning_rate": 1.816430327751715e-05,
      "loss": 0.639,
      "step": 3536
    },
    {
      "epoch": 0.44760820045558086,
      "grad_norm": 1.111226201057434,
      "learning_rate": 1.81632682954841e-05,
      "loss": 0.6748,
      "step": 3537
    },
    {
      "epoch": 0.4477347506960263,
      "grad_norm": 1.1305862665176392,
      "learning_rate": 1.8162233051270797e-05,
      "loss": 0.6717,
      "step": 3538
    },
    {
      "epoch": 0.4478613009364718,
      "grad_norm": 0.9864685535430908,
      "learning_rate": 1.8161197544910484e-05,
      "loss": 0.6268,
      "step": 3539
    },
    {
      "epoch": 0.44798785117691725,
      "grad_norm": 1.0797038078308105,
      "learning_rate": 1.8160161776436423e-05,
      "loss": 0.6412,
      "step": 3540
    },
    {
      "epoch": 0.4481144014173627,
      "grad_norm": 1.069161295890808,
      "learning_rate": 1.8159125745881877e-05,
      "loss": 0.666,
      "step": 3541
    },
    {
      "epoch": 0.44824095165780814,
      "grad_norm": 1.0330842733383179,
      "learning_rate": 1.8158089453280125e-05,
      "loss": 0.7055,
      "step": 3542
    },
    {
      "epoch": 0.4483675018982536,
      "grad_norm": 0.9706534147262573,
      "learning_rate": 1.815705289866445e-05,
      "loss": 0.6062,
      "step": 3543
    },
    {
      "epoch": 0.4484940521386991,
      "grad_norm": 1.1269224882125854,
      "learning_rate": 1.8156016082068132e-05,
      "loss": 0.6585,
      "step": 3544
    },
    {
      "epoch": 0.44862060237914453,
      "grad_norm": 1.01727294921875,
      "learning_rate": 1.815497900352448e-05,
      "loss": 0.6431,
      "step": 3545
    },
    {
      "epoch": 0.44874715261959,
      "grad_norm": 1.0578693151474,
      "learning_rate": 1.8153941663066805e-05,
      "loss": 0.6249,
      "step": 3546
    },
    {
      "epoch": 0.4488737028600354,
      "grad_norm": 1.0534111261367798,
      "learning_rate": 1.8152904060728415e-05,
      "loss": 0.6458,
      "step": 3547
    },
    {
      "epoch": 0.4490002531004809,
      "grad_norm": 1.0226811170578003,
      "learning_rate": 1.8151866196542637e-05,
      "loss": 0.6703,
      "step": 3548
    },
    {
      "epoch": 0.44912680334092636,
      "grad_norm": 1.0495796203613281,
      "learning_rate": 1.8150828070542807e-05,
      "loss": 0.6323,
      "step": 3549
    },
    {
      "epoch": 0.4492533535813718,
      "grad_norm": 1.1461201906204224,
      "learning_rate": 1.8149789682762264e-05,
      "loss": 0.6694,
      "step": 3550
    },
    {
      "epoch": 0.44937990382181725,
      "grad_norm": 1.0103824138641357,
      "learning_rate": 1.814875103323436e-05,
      "loss": 0.6488,
      "step": 3551
    },
    {
      "epoch": 0.4495064540622627,
      "grad_norm": 1.459146499633789,
      "learning_rate": 1.8147712121992455e-05,
      "loss": 0.6445,
      "step": 3552
    },
    {
      "epoch": 0.4496330043027082,
      "grad_norm": 1.093222737312317,
      "learning_rate": 1.814667294906991e-05,
      "loss": 0.6463,
      "step": 3553
    },
    {
      "epoch": 0.44975955454315364,
      "grad_norm": 1.0491993427276611,
      "learning_rate": 1.8145633514500106e-05,
      "loss": 0.6232,
      "step": 3554
    },
    {
      "epoch": 0.4498861047835991,
      "grad_norm": 1.1006333827972412,
      "learning_rate": 1.8144593818316423e-05,
      "loss": 0.652,
      "step": 3555
    },
    {
      "epoch": 0.45001265502404453,
      "grad_norm": 1.117205023765564,
      "learning_rate": 1.8143553860552252e-05,
      "loss": 0.7063,
      "step": 3556
    },
    {
      "epoch": 0.45013920526449,
      "grad_norm": 1.0848149061203003,
      "learning_rate": 1.8142513641240994e-05,
      "loss": 0.6519,
      "step": 3557
    },
    {
      "epoch": 0.45026575550493547,
      "grad_norm": 1.0518403053283691,
      "learning_rate": 1.8141473160416065e-05,
      "loss": 0.6868,
      "step": 3558
    },
    {
      "epoch": 0.4503923057453809,
      "grad_norm": 1.017738938331604,
      "learning_rate": 1.8140432418110873e-05,
      "loss": 0.6527,
      "step": 3559
    },
    {
      "epoch": 0.45051885598582636,
      "grad_norm": 1.0254367589950562,
      "learning_rate": 1.8139391414358848e-05,
      "loss": 0.6166,
      "step": 3560
    },
    {
      "epoch": 0.4506454062262718,
      "grad_norm": 1.173999547958374,
      "learning_rate": 1.813835014919342e-05,
      "loss": 0.6582,
      "step": 3561
    },
    {
      "epoch": 0.4507719564667173,
      "grad_norm": 1.0616523027420044,
      "learning_rate": 1.8137308622648036e-05,
      "loss": 0.6386,
      "step": 3562
    },
    {
      "epoch": 0.45089850670716275,
      "grad_norm": 0.9934318661689758,
      "learning_rate": 1.8136266834756148e-05,
      "loss": 0.5983,
      "step": 3563
    },
    {
      "epoch": 0.4510250569476082,
      "grad_norm": 1.1364760398864746,
      "learning_rate": 1.8135224785551208e-05,
      "loss": 0.6381,
      "step": 3564
    },
    {
      "epoch": 0.45115160718805364,
      "grad_norm": 1.0245304107666016,
      "learning_rate": 1.813418247506669e-05,
      "loss": 0.6313,
      "step": 3565
    },
    {
      "epoch": 0.45127815742849914,
      "grad_norm": 0.980806291103363,
      "learning_rate": 1.813313990333607e-05,
      "loss": 0.6618,
      "step": 3566
    },
    {
      "epoch": 0.4514047076689446,
      "grad_norm": 1.054916262626648,
      "learning_rate": 1.8132097070392826e-05,
      "loss": 0.6811,
      "step": 3567
    },
    {
      "epoch": 0.45153125790939,
      "grad_norm": 1.0657973289489746,
      "learning_rate": 1.8131053976270454e-05,
      "loss": 0.6519,
      "step": 3568
    },
    {
      "epoch": 0.45165780814983547,
      "grad_norm": 1.0923433303833008,
      "learning_rate": 1.8130010621002458e-05,
      "loss": 0.6828,
      "step": 3569
    },
    {
      "epoch": 0.4517843583902809,
      "grad_norm": 1.1006027460098267,
      "learning_rate": 1.812896700462235e-05,
      "loss": 0.6686,
      "step": 3570
    },
    {
      "epoch": 0.4519109086307264,
      "grad_norm": 1.0015718936920166,
      "learning_rate": 1.812792312716364e-05,
      "loss": 0.6313,
      "step": 3571
    },
    {
      "epoch": 0.45203745887117186,
      "grad_norm": 1.0375021696090698,
      "learning_rate": 1.8126878988659852e-05,
      "loss": 0.6581,
      "step": 3572
    },
    {
      "epoch": 0.4521640091116173,
      "grad_norm": 1.0229820013046265,
      "learning_rate": 1.8125834589144532e-05,
      "loss": 0.6317,
      "step": 3573
    },
    {
      "epoch": 0.45229055935206275,
      "grad_norm": 0.9992733597755432,
      "learning_rate": 1.812478992865122e-05,
      "loss": 0.6462,
      "step": 3574
    },
    {
      "epoch": 0.45241710959250825,
      "grad_norm": 1.3156284093856812,
      "learning_rate": 1.8123745007213462e-05,
      "loss": 0.6466,
      "step": 3575
    },
    {
      "epoch": 0.4525436598329537,
      "grad_norm": 1.1572543382644653,
      "learning_rate": 1.812269982486482e-05,
      "loss": 0.6868,
      "step": 3576
    },
    {
      "epoch": 0.45267021007339914,
      "grad_norm": 1.312527060508728,
      "learning_rate": 1.8121654381638864e-05,
      "loss": 0.6054,
      "step": 3577
    },
    {
      "epoch": 0.4527967603138446,
      "grad_norm": 1.1473982334136963,
      "learning_rate": 1.812060867756917e-05,
      "loss": 0.655,
      "step": 3578
    },
    {
      "epoch": 0.45292331055429,
      "grad_norm": 1.1483732461929321,
      "learning_rate": 1.811956271268932e-05,
      "loss": 0.6689,
      "step": 3579
    },
    {
      "epoch": 0.4530498607947355,
      "grad_norm": 1.140423059463501,
      "learning_rate": 1.8118516487032913e-05,
      "loss": 0.6771,
      "step": 3580
    },
    {
      "epoch": 0.45317641103518097,
      "grad_norm": 1.0209412574768066,
      "learning_rate": 1.8117470000633542e-05,
      "loss": 0.6425,
      "step": 3581
    },
    {
      "epoch": 0.4533029612756264,
      "grad_norm": 0.9916905760765076,
      "learning_rate": 1.8116423253524826e-05,
      "loss": 0.6538,
      "step": 3582
    },
    {
      "epoch": 0.45342951151607186,
      "grad_norm": 0.9872011542320251,
      "learning_rate": 1.811537624574038e-05,
      "loss": 0.6365,
      "step": 3583
    },
    {
      "epoch": 0.45355606175651736,
      "grad_norm": 1.0883829593658447,
      "learning_rate": 1.811432897731383e-05,
      "loss": 0.6739,
      "step": 3584
    },
    {
      "epoch": 0.4536826119969628,
      "grad_norm": 1.0848579406738281,
      "learning_rate": 1.8113281448278814e-05,
      "loss": 0.6286,
      "step": 3585
    },
    {
      "epoch": 0.45380916223740825,
      "grad_norm": 1.1425601243972778,
      "learning_rate": 1.811223365866897e-05,
      "loss": 0.6052,
      "step": 3586
    },
    {
      "epoch": 0.4539357124778537,
      "grad_norm": 1.1536409854888916,
      "learning_rate": 1.811118560851796e-05,
      "loss": 0.6184,
      "step": 3587
    },
    {
      "epoch": 0.4540622627182992,
      "grad_norm": 1.0391196012496948,
      "learning_rate": 1.811013729785943e-05,
      "loss": 0.6358,
      "step": 3588
    },
    {
      "epoch": 0.45418881295874464,
      "grad_norm": 1.0676082372665405,
      "learning_rate": 1.810908872672706e-05,
      "loss": 0.6436,
      "step": 3589
    },
    {
      "epoch": 0.4543153631991901,
      "grad_norm": 0.991287112236023,
      "learning_rate": 1.8108039895154522e-05,
      "loss": 0.6615,
      "step": 3590
    },
    {
      "epoch": 0.4544419134396355,
      "grad_norm": 1.0168933868408203,
      "learning_rate": 1.8106990803175506e-05,
      "loss": 0.6211,
      "step": 3591
    },
    {
      "epoch": 0.45456846368008097,
      "grad_norm": 1.0483753681182861,
      "learning_rate": 1.8105941450823702e-05,
      "loss": 0.6472,
      "step": 3592
    },
    {
      "epoch": 0.45469501392052647,
      "grad_norm": 1.095287799835205,
      "learning_rate": 1.8104891838132812e-05,
      "loss": 0.668,
      "step": 3593
    },
    {
      "epoch": 0.4548215641609719,
      "grad_norm": 1.0517196655273438,
      "learning_rate": 1.8103841965136547e-05,
      "loss": 0.6779,
      "step": 3594
    },
    {
      "epoch": 0.45494811440141736,
      "grad_norm": 0.9930626153945923,
      "learning_rate": 1.8102791831868626e-05,
      "loss": 0.6367,
      "step": 3595
    },
    {
      "epoch": 0.4550746646418628,
      "grad_norm": 1.3757139444351196,
      "learning_rate": 1.8101741438362778e-05,
      "loss": 0.6659,
      "step": 3596
    },
    {
      "epoch": 0.4552012148823083,
      "grad_norm": 1.1530219316482544,
      "learning_rate": 1.8100690784652732e-05,
      "loss": 0.718,
      "step": 3597
    },
    {
      "epoch": 0.45532776512275375,
      "grad_norm": 1.0358517169952393,
      "learning_rate": 1.809963987077224e-05,
      "loss": 0.6528,
      "step": 3598
    },
    {
      "epoch": 0.4554543153631992,
      "grad_norm": 1.0223804712295532,
      "learning_rate": 1.8098588696755052e-05,
      "loss": 0.5636,
      "step": 3599
    },
    {
      "epoch": 0.45558086560364464,
      "grad_norm": 1.2212520837783813,
      "learning_rate": 1.8097537262634924e-05,
      "loss": 0.6541,
      "step": 3600
    },
    {
      "epoch": 0.4557074158440901,
      "grad_norm": 1.0999711751937866,
      "learning_rate": 1.809648556844563e-05,
      "loss": 0.6522,
      "step": 3601
    },
    {
      "epoch": 0.4558339660845356,
      "grad_norm": 1.1833720207214355,
      "learning_rate": 1.809543361422095e-05,
      "loss": 0.6568,
      "step": 3602
    },
    {
      "epoch": 0.455960516324981,
      "grad_norm": 1.0488357543945312,
      "learning_rate": 1.8094381399994662e-05,
      "loss": 0.6303,
      "step": 3603
    },
    {
      "epoch": 0.45608706656542647,
      "grad_norm": 1.1194939613342285,
      "learning_rate": 1.8093328925800565e-05,
      "loss": 0.602,
      "step": 3604
    },
    {
      "epoch": 0.4562136168058719,
      "grad_norm": 1.2816318273544312,
      "learning_rate": 1.809227619167246e-05,
      "loss": 0.6813,
      "step": 3605
    },
    {
      "epoch": 0.4563401670463174,
      "grad_norm": 0.9907235503196716,
      "learning_rate": 1.8091223197644157e-05,
      "loss": 0.6007,
      "step": 3606
    },
    {
      "epoch": 0.45646671728676286,
      "grad_norm": 1.2371647357940674,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.6771,
      "step": 3607
    },
    {
      "epoch": 0.4565932675272083,
      "grad_norm": 1.1338722705841064,
      "learning_rate": 1.8089116430022245e-05,
      "loss": 0.6927,
      "step": 3608
    },
    {
      "epoch": 0.45671981776765375,
      "grad_norm": 1.0808212757110596,
      "learning_rate": 1.8088062656496298e-05,
      "loss": 0.7012,
      "step": 3609
    },
    {
      "epoch": 0.4568463680080992,
      "grad_norm": 1.0904604196548462,
      "learning_rate": 1.808700862320548e-05,
      "loss": 0.6663,
      "step": 3610
    },
    {
      "epoch": 0.4569729182485447,
      "grad_norm": 1.1190259456634521,
      "learning_rate": 1.8085954330183645e-05,
      "loss": 0.6577,
      "step": 3611
    },
    {
      "epoch": 0.45709946848899013,
      "grad_norm": 0.9832779765129089,
      "learning_rate": 1.808489977746465e-05,
      "loss": 0.6488,
      "step": 3612
    },
    {
      "epoch": 0.4572260187294356,
      "grad_norm": 1.0147607326507568,
      "learning_rate": 1.8083844965082372e-05,
      "loss": 0.6319,
      "step": 3613
    },
    {
      "epoch": 0.457352568969881,
      "grad_norm": 1.1013182401657104,
      "learning_rate": 1.808278989307068e-05,
      "loss": 0.6422,
      "step": 3614
    },
    {
      "epoch": 0.4574791192103265,
      "grad_norm": 1.168721079826355,
      "learning_rate": 1.8081734561463463e-05,
      "loss": 0.6538,
      "step": 3615
    },
    {
      "epoch": 0.45760566945077197,
      "grad_norm": 1.2125072479248047,
      "learning_rate": 1.8080678970294613e-05,
      "loss": 0.6411,
      "step": 3616
    },
    {
      "epoch": 0.4577322196912174,
      "grad_norm": 1.1858141422271729,
      "learning_rate": 1.807962311959804e-05,
      "loss": 0.6769,
      "step": 3617
    },
    {
      "epoch": 0.45785876993166286,
      "grad_norm": 1.093203067779541,
      "learning_rate": 1.8078567009407642e-05,
      "loss": 0.6321,
      "step": 3618
    },
    {
      "epoch": 0.4579853201721083,
      "grad_norm": 1.0675889253616333,
      "learning_rate": 1.8077510639757353e-05,
      "loss": 0.6212,
      "step": 3619
    },
    {
      "epoch": 0.4581118704125538,
      "grad_norm": 1.1632144451141357,
      "learning_rate": 1.807645401068109e-05,
      "loss": 0.6452,
      "step": 3620
    },
    {
      "epoch": 0.45823842065299925,
      "grad_norm": 1.049445390701294,
      "learning_rate": 1.807539712221279e-05,
      "loss": 0.6234,
      "step": 3621
    },
    {
      "epoch": 0.4583649708934447,
      "grad_norm": 1.1429351568222046,
      "learning_rate": 1.8074339974386398e-05,
      "loss": 0.6769,
      "step": 3622
    },
    {
      "epoch": 0.45849152113389013,
      "grad_norm": 1.0403567552566528,
      "learning_rate": 1.807328256723587e-05,
      "loss": 0.6381,
      "step": 3623
    },
    {
      "epoch": 0.45861807137433563,
      "grad_norm": 1.1595245599746704,
      "learning_rate": 1.8072224900795167e-05,
      "loss": 0.6504,
      "step": 3624
    },
    {
      "epoch": 0.4587446216147811,
      "grad_norm": 0.9873105883598328,
      "learning_rate": 1.807116697509825e-05,
      "loss": 0.673,
      "step": 3625
    },
    {
      "epoch": 0.4588711718552265,
      "grad_norm": 1.044687032699585,
      "learning_rate": 1.8070108790179107e-05,
      "loss": 0.6354,
      "step": 3626
    },
    {
      "epoch": 0.45899772209567197,
      "grad_norm": 0.987978994846344,
      "learning_rate": 1.806905034607172e-05,
      "loss": 0.6113,
      "step": 3627
    },
    {
      "epoch": 0.4591242723361174,
      "grad_norm": 1.0583484172821045,
      "learning_rate": 1.8067991642810074e-05,
      "loss": 0.6457,
      "step": 3628
    },
    {
      "epoch": 0.4592508225765629,
      "grad_norm": 0.9957163333892822,
      "learning_rate": 1.8066932680428183e-05,
      "loss": 0.6904,
      "step": 3629
    },
    {
      "epoch": 0.45937737281700836,
      "grad_norm": 0.9921567440032959,
      "learning_rate": 1.8065873458960058e-05,
      "loss": 0.6554,
      "step": 3630
    },
    {
      "epoch": 0.4595039230574538,
      "grad_norm": 1.0227787494659424,
      "learning_rate": 1.806481397843971e-05,
      "loss": 0.6788,
      "step": 3631
    },
    {
      "epoch": 0.45963047329789924,
      "grad_norm": 1.0668398141860962,
      "learning_rate": 1.8063754238901173e-05,
      "loss": 0.6445,
      "step": 3632
    },
    {
      "epoch": 0.45975702353834474,
      "grad_norm": 1.0613397359848022,
      "learning_rate": 1.806269424037848e-05,
      "loss": 0.6798,
      "step": 3633
    },
    {
      "epoch": 0.4598835737787902,
      "grad_norm": 1.1696535348892212,
      "learning_rate": 1.8061633982905675e-05,
      "loss": 0.6796,
      "step": 3634
    },
    {
      "epoch": 0.46001012401923563,
      "grad_norm": 1.125043272972107,
      "learning_rate": 1.806057346651681e-05,
      "loss": 0.6545,
      "step": 3635
    },
    {
      "epoch": 0.4601366742596811,
      "grad_norm": 1.0323021411895752,
      "learning_rate": 1.8059512691245945e-05,
      "loss": 0.5998,
      "step": 3636
    },
    {
      "epoch": 0.4602632245001266,
      "grad_norm": 0.9998145699501038,
      "learning_rate": 1.8058451657127153e-05,
      "loss": 0.6562,
      "step": 3637
    },
    {
      "epoch": 0.460389774740572,
      "grad_norm": 1.1538581848144531,
      "learning_rate": 1.805739036419451e-05,
      "loss": 0.647,
      "step": 3638
    },
    {
      "epoch": 0.46051632498101747,
      "grad_norm": 1.1390964984893799,
      "learning_rate": 1.80563288124821e-05,
      "loss": 0.6691,
      "step": 3639
    },
    {
      "epoch": 0.4606428752214629,
      "grad_norm": 1.0369409322738647,
      "learning_rate": 1.8055267002024015e-05,
      "loss": 0.6631,
      "step": 3640
    },
    {
      "epoch": 0.46076942546190836,
      "grad_norm": 1.0259534120559692,
      "learning_rate": 1.805420493285436e-05,
      "loss": 0.637,
      "step": 3641
    },
    {
      "epoch": 0.46089597570235386,
      "grad_norm": 1.1352763175964355,
      "learning_rate": 1.805314260500725e-05,
      "loss": 0.6526,
      "step": 3642
    },
    {
      "epoch": 0.4610225259427993,
      "grad_norm": 1.1179221868515015,
      "learning_rate": 1.8052080018516792e-05,
      "loss": 0.6685,
      "step": 3643
    },
    {
      "epoch": 0.46114907618324474,
      "grad_norm": 1.0240534543991089,
      "learning_rate": 1.805101717341712e-05,
      "loss": 0.6586,
      "step": 3644
    },
    {
      "epoch": 0.4612756264236902,
      "grad_norm": 1.0620770454406738,
      "learning_rate": 1.8049954069742375e-05,
      "loss": 0.6524,
      "step": 3645
    },
    {
      "epoch": 0.4614021766641357,
      "grad_norm": 1.0840580463409424,
      "learning_rate": 1.804889070752669e-05,
      "loss": 0.67,
      "step": 3646
    },
    {
      "epoch": 0.46152872690458113,
      "grad_norm": 1.3824074268341064,
      "learning_rate": 1.8047827086804225e-05,
      "loss": 0.6654,
      "step": 3647
    },
    {
      "epoch": 0.4616552771450266,
      "grad_norm": 1.2531386613845825,
      "learning_rate": 1.8046763207609138e-05,
      "loss": 0.7056,
      "step": 3648
    },
    {
      "epoch": 0.461781827385472,
      "grad_norm": 1.0755720138549805,
      "learning_rate": 1.8045699069975597e-05,
      "loss": 0.6511,
      "step": 3649
    },
    {
      "epoch": 0.46190837762591747,
      "grad_norm": 1.156458854675293,
      "learning_rate": 1.8044634673937783e-05,
      "loss": 0.629,
      "step": 3650
    },
    {
      "epoch": 0.46203492786636297,
      "grad_norm": 0.9980881214141846,
      "learning_rate": 1.8043570019529872e-05,
      "loss": 0.6749,
      "step": 3651
    },
    {
      "epoch": 0.4621614781068084,
      "grad_norm": 1.0202831029891968,
      "learning_rate": 1.804250510678607e-05,
      "loss": 0.6465,
      "step": 3652
    },
    {
      "epoch": 0.46228802834725385,
      "grad_norm": 1.0503383874893188,
      "learning_rate": 1.8041439935740563e-05,
      "loss": 0.679,
      "step": 3653
    },
    {
      "epoch": 0.4624145785876993,
      "grad_norm": 0.9722995162010193,
      "learning_rate": 1.804037450642758e-05,
      "loss": 0.6141,
      "step": 3654
    },
    {
      "epoch": 0.4625411288281448,
      "grad_norm": 1.045229434967041,
      "learning_rate": 1.8039308818881323e-05,
      "loss": 0.6322,
      "step": 3655
    },
    {
      "epoch": 0.46266767906859024,
      "grad_norm": 1.0873316526412964,
      "learning_rate": 1.803824287313603e-05,
      "loss": 0.6051,
      "step": 3656
    },
    {
      "epoch": 0.4627942293090357,
      "grad_norm": 0.9817551374435425,
      "learning_rate": 1.8037176669225927e-05,
      "loss": 0.6329,
      "step": 3657
    },
    {
      "epoch": 0.46292077954948113,
      "grad_norm": 1.0568116903305054,
      "learning_rate": 1.8036110207185268e-05,
      "loss": 0.6135,
      "step": 3658
    },
    {
      "epoch": 0.4630473297899266,
      "grad_norm": 1.0470303297042847,
      "learning_rate": 1.8035043487048295e-05,
      "loss": 0.6395,
      "step": 3659
    },
    {
      "epoch": 0.4631738800303721,
      "grad_norm": 1.088196873664856,
      "learning_rate": 1.8033976508849268e-05,
      "loss": 0.6389,
      "step": 3660
    },
    {
      "epoch": 0.4633004302708175,
      "grad_norm": 6.04470157623291,
      "learning_rate": 1.8032909272622463e-05,
      "loss": 0.6812,
      "step": 3661
    },
    {
      "epoch": 0.46342698051126296,
      "grad_norm": 1.1282955408096313,
      "learning_rate": 1.8031841778402155e-05,
      "loss": 0.6594,
      "step": 3662
    },
    {
      "epoch": 0.4635535307517084,
      "grad_norm": 1.0260168313980103,
      "learning_rate": 1.803077402622262e-05,
      "loss": 0.6437,
      "step": 3663
    },
    {
      "epoch": 0.4636800809921539,
      "grad_norm": 1.0516149997711182,
      "learning_rate": 1.8029706016118164e-05,
      "loss": 0.656,
      "step": 3664
    },
    {
      "epoch": 0.46380663123259935,
      "grad_norm": 1.1204456090927124,
      "learning_rate": 1.8028637748123077e-05,
      "loss": 0.6966,
      "step": 3665
    },
    {
      "epoch": 0.4639331814730448,
      "grad_norm": 1.0282396078109741,
      "learning_rate": 1.8027569222271673e-05,
      "loss": 0.621,
      "step": 3666
    },
    {
      "epoch": 0.46405973171349024,
      "grad_norm": 1.008209228515625,
      "learning_rate": 1.802650043859827e-05,
      "loss": 0.677,
      "step": 3667
    },
    {
      "epoch": 0.4641862819539357,
      "grad_norm": 1.0832899808883667,
      "learning_rate": 1.8025431397137194e-05,
      "loss": 0.6547,
      "step": 3668
    },
    {
      "epoch": 0.4643128321943812,
      "grad_norm": 1.1290203332901,
      "learning_rate": 1.8024362097922785e-05,
      "loss": 0.6379,
      "step": 3669
    },
    {
      "epoch": 0.46443938243482663,
      "grad_norm": 1.017605185508728,
      "learning_rate": 1.8023292540989375e-05,
      "loss": 0.6269,
      "step": 3670
    },
    {
      "epoch": 0.4645659326752721,
      "grad_norm": 1.1318848133087158,
      "learning_rate": 1.802222272637132e-05,
      "loss": 0.6842,
      "step": 3671
    },
    {
      "epoch": 0.4646924829157175,
      "grad_norm": 1.039520502090454,
      "learning_rate": 1.8021152654102984e-05,
      "loss": 0.6316,
      "step": 3672
    },
    {
      "epoch": 0.464819033156163,
      "grad_norm": 1.0573430061340332,
      "learning_rate": 1.8020082324218725e-05,
      "loss": 0.6492,
      "step": 3673
    },
    {
      "epoch": 0.46494558339660846,
      "grad_norm": 1.017120361328125,
      "learning_rate": 1.801901173675293e-05,
      "loss": 0.6002,
      "step": 3674
    },
    {
      "epoch": 0.4650721336370539,
      "grad_norm": 1.1046580076217651,
      "learning_rate": 1.8017940891739975e-05,
      "loss": 0.6566,
      "step": 3675
    },
    {
      "epoch": 0.46519868387749935,
      "grad_norm": 0.9912790060043335,
      "learning_rate": 1.8016869789214257e-05,
      "loss": 0.6588,
      "step": 3676
    },
    {
      "epoch": 0.4653252341179448,
      "grad_norm": 1.0375900268554688,
      "learning_rate": 1.801579842921017e-05,
      "loss": 0.6428,
      "step": 3677
    },
    {
      "epoch": 0.4654517843583903,
      "grad_norm": 0.9967744946479797,
      "learning_rate": 1.801472681176213e-05,
      "loss": 0.6633,
      "step": 3678
    },
    {
      "epoch": 0.46557833459883574,
      "grad_norm": 1.0542765855789185,
      "learning_rate": 1.801365493690455e-05,
      "loss": 0.642,
      "step": 3679
    },
    {
      "epoch": 0.4657048848392812,
      "grad_norm": 1.140801191329956,
      "learning_rate": 1.801258280467186e-05,
      "loss": 0.6352,
      "step": 3680
    },
    {
      "epoch": 0.46583143507972663,
      "grad_norm": 1.0095024108886719,
      "learning_rate": 1.801151041509849e-05,
      "loss": 0.6643,
      "step": 3681
    },
    {
      "epoch": 0.46595798532017213,
      "grad_norm": 1.191558837890625,
      "learning_rate": 1.801043776821888e-05,
      "loss": 0.6793,
      "step": 3682
    },
    {
      "epoch": 0.4660845355606176,
      "grad_norm": 1.068209171295166,
      "learning_rate": 1.8009364864067487e-05,
      "loss": 0.6545,
      "step": 3683
    },
    {
      "epoch": 0.466211085801063,
      "grad_norm": 1.0273059606552124,
      "learning_rate": 1.800829170267876e-05,
      "loss": 0.6995,
      "step": 3684
    },
    {
      "epoch": 0.46633763604150846,
      "grad_norm": 0.9314380884170532,
      "learning_rate": 1.8007218284087182e-05,
      "loss": 0.6441,
      "step": 3685
    },
    {
      "epoch": 0.46646418628195396,
      "grad_norm": 1.0287450551986694,
      "learning_rate": 1.8006144608327208e-05,
      "loss": 0.6108,
      "step": 3686
    },
    {
      "epoch": 0.4665907365223994,
      "grad_norm": 1.0485386848449707,
      "learning_rate": 1.8005070675433335e-05,
      "loss": 0.6393,
      "step": 3687
    },
    {
      "epoch": 0.46671728676284485,
      "grad_norm": 1.107762336730957,
      "learning_rate": 1.800399648544005e-05,
      "loss": 0.6586,
      "step": 3688
    },
    {
      "epoch": 0.4668438370032903,
      "grad_norm": 2.058061361312866,
      "learning_rate": 1.800292203838185e-05,
      "loss": 0.617,
      "step": 3689
    },
    {
      "epoch": 0.46697038724373574,
      "grad_norm": 1.235780954360962,
      "learning_rate": 1.800184733429325e-05,
      "loss": 0.6679,
      "step": 3690
    },
    {
      "epoch": 0.46709693748418124,
      "grad_norm": 0.983381450176239,
      "learning_rate": 1.8000772373208763e-05,
      "loss": 0.6177,
      "step": 3691
    },
    {
      "epoch": 0.4672234877246267,
      "grad_norm": 1.13225519657135,
      "learning_rate": 1.7999697155162912e-05,
      "loss": 0.588,
      "step": 3692
    },
    {
      "epoch": 0.46735003796507213,
      "grad_norm": 1.1247869729995728,
      "learning_rate": 1.799862168019023e-05,
      "loss": 0.6683,
      "step": 3693
    },
    {
      "epoch": 0.4674765882055176,
      "grad_norm": 1.0096794366836548,
      "learning_rate": 1.7997545948325262e-05,
      "loss": 0.6506,
      "step": 3694
    },
    {
      "epoch": 0.4676031384459631,
      "grad_norm": 1.075408935546875,
      "learning_rate": 1.7996469959602553e-05,
      "loss": 0.6582,
      "step": 3695
    },
    {
      "epoch": 0.4677296886864085,
      "grad_norm": 1.0867985486984253,
      "learning_rate": 1.799539371405666e-05,
      "loss": 0.6798,
      "step": 3696
    },
    {
      "epoch": 0.46785623892685396,
      "grad_norm": 1.0105688571929932,
      "learning_rate": 1.7994317211722158e-05,
      "loss": 0.6258,
      "step": 3697
    },
    {
      "epoch": 0.4679827891672994,
      "grad_norm": 1.0116173028945923,
      "learning_rate": 1.7993240452633606e-05,
      "loss": 0.6605,
      "step": 3698
    },
    {
      "epoch": 0.46810933940774485,
      "grad_norm": 1.055088758468628,
      "learning_rate": 1.79921634368256e-05,
      "loss": 0.6127,
      "step": 3699
    },
    {
      "epoch": 0.46823588964819035,
      "grad_norm": 1.0256116390228271,
      "learning_rate": 1.7991086164332726e-05,
      "loss": 0.6483,
      "step": 3700
    },
    {
      "epoch": 0.4683624398886358,
      "grad_norm": 1.7982838153839111,
      "learning_rate": 1.7990008635189578e-05,
      "loss": 0.6366,
      "step": 3701
    },
    {
      "epoch": 0.46848899012908124,
      "grad_norm": 1.1619462966918945,
      "learning_rate": 1.7988930849430767e-05,
      "loss": 0.6266,
      "step": 3702
    },
    {
      "epoch": 0.4686155403695267,
      "grad_norm": 1.0153419971466064,
      "learning_rate": 1.7987852807090912e-05,
      "loss": 0.6543,
      "step": 3703
    },
    {
      "epoch": 0.4687420906099722,
      "grad_norm": 1.2192738056182861,
      "learning_rate": 1.798677450820463e-05,
      "loss": 0.6291,
      "step": 3704
    },
    {
      "epoch": 0.46886864085041763,
      "grad_norm": 1.0449861288070679,
      "learning_rate": 1.7985695952806555e-05,
      "loss": 0.6893,
      "step": 3705
    },
    {
      "epoch": 0.4689951910908631,
      "grad_norm": 1.0337568521499634,
      "learning_rate": 1.798461714093133e-05,
      "loss": 0.6232,
      "step": 3706
    },
    {
      "epoch": 0.4691217413313085,
      "grad_norm": 0.9969984889030457,
      "learning_rate": 1.7983538072613602e-05,
      "loss": 0.5754,
      "step": 3707
    },
    {
      "epoch": 0.46924829157175396,
      "grad_norm": 1.0287301540374756,
      "learning_rate": 1.7982458747888023e-05,
      "loss": 0.5901,
      "step": 3708
    },
    {
      "epoch": 0.46937484181219946,
      "grad_norm": 0.9762424230575562,
      "learning_rate": 1.7981379166789264e-05,
      "loss": 0.6538,
      "step": 3709
    },
    {
      "epoch": 0.4695013920526449,
      "grad_norm": 1.1191840171813965,
      "learning_rate": 1.7980299329351996e-05,
      "loss": 0.6501,
      "step": 3710
    },
    {
      "epoch": 0.46962794229309035,
      "grad_norm": 1.154434323310852,
      "learning_rate": 1.7979219235610896e-05,
      "loss": 0.6131,
      "step": 3711
    },
    {
      "epoch": 0.4697544925335358,
      "grad_norm": 1.2005712985992432,
      "learning_rate": 1.797813888560066e-05,
      "loss": 0.6615,
      "step": 3712
    },
    {
      "epoch": 0.4698810427739813,
      "grad_norm": 1.0186296701431274,
      "learning_rate": 1.7977058279355982e-05,
      "loss": 0.6241,
      "step": 3713
    },
    {
      "epoch": 0.47000759301442674,
      "grad_norm": 1.0912730693817139,
      "learning_rate": 1.7975977416911567e-05,
      "loss": 0.6494,
      "step": 3714
    },
    {
      "epoch": 0.4701341432548722,
      "grad_norm": 1.028948187828064,
      "learning_rate": 1.7974896298302133e-05,
      "loss": 0.5686,
      "step": 3715
    },
    {
      "epoch": 0.4702606934953176,
      "grad_norm": 1.0385874509811401,
      "learning_rate": 1.7973814923562397e-05,
      "loss": 0.6406,
      "step": 3716
    },
    {
      "epoch": 0.47038724373576307,
      "grad_norm": 1.054891586303711,
      "learning_rate": 1.7972733292727094e-05,
      "loss": 0.6454,
      "step": 3717
    },
    {
      "epoch": 0.47051379397620857,
      "grad_norm": 1.0894874334335327,
      "learning_rate": 1.7971651405830963e-05,
      "loss": 0.6515,
      "step": 3718
    },
    {
      "epoch": 0.470640344216654,
      "grad_norm": 1.0492730140686035,
      "learning_rate": 1.797056926290875e-05,
      "loss": 0.639,
      "step": 3719
    },
    {
      "epoch": 0.47076689445709946,
      "grad_norm": 1.0039489269256592,
      "learning_rate": 1.7969486863995205e-05,
      "loss": 0.6398,
      "step": 3720
    },
    {
      "epoch": 0.4708934446975449,
      "grad_norm": 1.0987361669540405,
      "learning_rate": 1.79684042091251e-05,
      "loss": 0.7106,
      "step": 3721
    },
    {
      "epoch": 0.4710199949379904,
      "grad_norm": 1.0469807386398315,
      "learning_rate": 1.7967321298333204e-05,
      "loss": 0.6733,
      "step": 3722
    },
    {
      "epoch": 0.47114654517843585,
      "grad_norm": 1.0638114213943481,
      "learning_rate": 1.7966238131654292e-05,
      "loss": 0.6041,
      "step": 3723
    },
    {
      "epoch": 0.4712730954188813,
      "grad_norm": 0.9833846092224121,
      "learning_rate": 1.796515470912316e-05,
      "loss": 0.6301,
      "step": 3724
    },
    {
      "epoch": 0.47139964565932674,
      "grad_norm": 1.029477834701538,
      "learning_rate": 1.7964071030774595e-05,
      "loss": 0.6381,
      "step": 3725
    },
    {
      "epoch": 0.4715261958997722,
      "grad_norm": 0.9825719594955444,
      "learning_rate": 1.7962987096643412e-05,
      "loss": 0.6808,
      "step": 3726
    },
    {
      "epoch": 0.4716527461402177,
      "grad_norm": 1.0845991373062134,
      "learning_rate": 1.7961902906764413e-05,
      "loss": 0.6909,
      "step": 3727
    },
    {
      "epoch": 0.4717792963806631,
      "grad_norm": 1.032916784286499,
      "learning_rate": 1.796081846117243e-05,
      "loss": 0.6241,
      "step": 3728
    },
    {
      "epoch": 0.47190584662110857,
      "grad_norm": 0.9982900619506836,
      "learning_rate": 1.7959733759902287e-05,
      "loss": 0.6089,
      "step": 3729
    },
    {
      "epoch": 0.472032396861554,
      "grad_norm": 2.1682004928588867,
      "learning_rate": 1.795864880298882e-05,
      "loss": 0.6545,
      "step": 3730
    },
    {
      "epoch": 0.4721589471019995,
      "grad_norm": 1.1291199922561646,
      "learning_rate": 1.7957563590466875e-05,
      "loss": 0.626,
      "step": 3731
    },
    {
      "epoch": 0.47228549734244496,
      "grad_norm": 1.0523627996444702,
      "learning_rate": 1.7956478122371307e-05,
      "loss": 0.6209,
      "step": 3732
    },
    {
      "epoch": 0.4724120475828904,
      "grad_norm": 1.0699446201324463,
      "learning_rate": 1.7955392398736978e-05,
      "loss": 0.6491,
      "step": 3733
    },
    {
      "epoch": 0.47253859782333585,
      "grad_norm": 1.044511079788208,
      "learning_rate": 1.7954306419598758e-05,
      "loss": 0.6541,
      "step": 3734
    },
    {
      "epoch": 0.47266514806378135,
      "grad_norm": 1.0194731950759888,
      "learning_rate": 1.7953220184991526e-05,
      "loss": 0.6447,
      "step": 3735
    },
    {
      "epoch": 0.4727916983042268,
      "grad_norm": 1.029425859451294,
      "learning_rate": 1.7952133694950168e-05,
      "loss": 0.6729,
      "step": 3736
    },
    {
      "epoch": 0.47291824854467224,
      "grad_norm": 1.0398201942443848,
      "learning_rate": 1.7951046949509582e-05,
      "loss": 0.6533,
      "step": 3737
    },
    {
      "epoch": 0.4730447987851177,
      "grad_norm": 1.0641241073608398,
      "learning_rate": 1.7949959948704664e-05,
      "loss": 0.645,
      "step": 3738
    },
    {
      "epoch": 0.4731713490255631,
      "grad_norm": 1.0475564002990723,
      "learning_rate": 1.794887269257033e-05,
      "loss": 0.636,
      "step": 3739
    },
    {
      "epoch": 0.4732978992660086,
      "grad_norm": 1.0004771947860718,
      "learning_rate": 1.7947785181141503e-05,
      "loss": 0.6294,
      "step": 3740
    },
    {
      "epoch": 0.47342444950645407,
      "grad_norm": 1.0546979904174805,
      "learning_rate": 1.79466974144531e-05,
      "loss": 0.6632,
      "step": 3741
    },
    {
      "epoch": 0.4735509997468995,
      "grad_norm": 1.0715694427490234,
      "learning_rate": 1.794560939254007e-05,
      "loss": 0.6326,
      "step": 3742
    },
    {
      "epoch": 0.47367754998734496,
      "grad_norm": 1.0361989736557007,
      "learning_rate": 1.7944521115437348e-05,
      "loss": 0.6601,
      "step": 3743
    },
    {
      "epoch": 0.47380410022779046,
      "grad_norm": 0.9859742522239685,
      "learning_rate": 1.794343258317989e-05,
      "loss": 0.6417,
      "step": 3744
    },
    {
      "epoch": 0.4739306504682359,
      "grad_norm": 0.955078661441803,
      "learning_rate": 1.794234379580265e-05,
      "loss": 0.6142,
      "step": 3745
    },
    {
      "epoch": 0.47405720070868135,
      "grad_norm": 1.000733733177185,
      "learning_rate": 1.7941254753340608e-05,
      "loss": 0.6379,
      "step": 3746
    },
    {
      "epoch": 0.4741837509491268,
      "grad_norm": 1.4369515180587769,
      "learning_rate": 1.7940165455828733e-05,
      "loss": 0.655,
      "step": 3747
    },
    {
      "epoch": 0.47431030118957224,
      "grad_norm": 1.142285704612732,
      "learning_rate": 1.793907590330201e-05,
      "loss": 0.694,
      "step": 3748
    },
    {
      "epoch": 0.47443685143001774,
      "grad_norm": 0.9758923053741455,
      "learning_rate": 1.793798609579544e-05,
      "loss": 0.6334,
      "step": 3749
    },
    {
      "epoch": 0.4745634016704632,
      "grad_norm": 1.070786952972412,
      "learning_rate": 1.7936896033344014e-05,
      "loss": 0.6511,
      "step": 3750
    },
    {
      "epoch": 0.4746899519109086,
      "grad_norm": 1.0575228929519653,
      "learning_rate": 1.7935805715982746e-05,
      "loss": 0.6215,
      "step": 3751
    },
    {
      "epoch": 0.47481650215135407,
      "grad_norm": 1.117210865020752,
      "learning_rate": 1.7934715143746655e-05,
      "loss": 0.6826,
      "step": 3752
    },
    {
      "epoch": 0.47494305239179957,
      "grad_norm": 1.029540777206421,
      "learning_rate": 1.7933624316670765e-05,
      "loss": 0.6549,
      "step": 3753
    },
    {
      "epoch": 0.475069602632245,
      "grad_norm": 1.0943797826766968,
      "learning_rate": 1.793253323479011e-05,
      "loss": 0.6625,
      "step": 3754
    },
    {
      "epoch": 0.47519615287269046,
      "grad_norm": 1.0995146036148071,
      "learning_rate": 1.7931441898139735e-05,
      "loss": 0.6427,
      "step": 3755
    },
    {
      "epoch": 0.4753227031131359,
      "grad_norm": 1.032387137413025,
      "learning_rate": 1.793035030675469e-05,
      "loss": 0.5986,
      "step": 3756
    },
    {
      "epoch": 0.47544925335358135,
      "grad_norm": 1.1464791297912598,
      "learning_rate": 1.7929258460670035e-05,
      "loss": 0.6678,
      "step": 3757
    },
    {
      "epoch": 0.47557580359402685,
      "grad_norm": 1.0263363122940063,
      "learning_rate": 1.792816635992083e-05,
      "loss": 0.6877,
      "step": 3758
    },
    {
      "epoch": 0.4757023538344723,
      "grad_norm": 1.036238431930542,
      "learning_rate": 1.7927074004542155e-05,
      "loss": 0.6706,
      "step": 3759
    },
    {
      "epoch": 0.47582890407491774,
      "grad_norm": 1.027527928352356,
      "learning_rate": 1.7925981394569098e-05,
      "loss": 0.6924,
      "step": 3760
    },
    {
      "epoch": 0.4759554543153632,
      "grad_norm": 1.0540354251861572,
      "learning_rate": 1.792488853003674e-05,
      "loss": 0.7028,
      "step": 3761
    },
    {
      "epoch": 0.4760820045558087,
      "grad_norm": 0.9683442115783691,
      "learning_rate": 1.792379541098019e-05,
      "loss": 0.6565,
      "step": 3762
    },
    {
      "epoch": 0.4762085547962541,
      "grad_norm": 0.9778546094894409,
      "learning_rate": 1.7922702037434546e-05,
      "loss": 0.6328,
      "step": 3763
    },
    {
      "epoch": 0.47633510503669957,
      "grad_norm": 1.2453725337982178,
      "learning_rate": 1.792160840943493e-05,
      "loss": 0.6393,
      "step": 3764
    },
    {
      "epoch": 0.476461655277145,
      "grad_norm": 1.0561354160308838,
      "learning_rate": 1.7920514527016474e-05,
      "loss": 0.6688,
      "step": 3765
    },
    {
      "epoch": 0.47658820551759046,
      "grad_norm": 1.1852314472198486,
      "learning_rate": 1.7919420390214294e-05,
      "loss": 0.6791,
      "step": 3766
    },
    {
      "epoch": 0.47671475575803596,
      "grad_norm": 1.1038740873336792,
      "learning_rate": 1.7918325999063543e-05,
      "loss": 0.6576,
      "step": 3767
    },
    {
      "epoch": 0.4768413059984814,
      "grad_norm": 1.132628321647644,
      "learning_rate": 1.7917231353599365e-05,
      "loss": 0.6286,
      "step": 3768
    },
    {
      "epoch": 0.47696785623892685,
      "grad_norm": 0.9138455390930176,
      "learning_rate": 1.7916136453856917e-05,
      "loss": 0.5715,
      "step": 3769
    },
    {
      "epoch": 0.4770944064793723,
      "grad_norm": 1.0144712924957275,
      "learning_rate": 1.791504129987136e-05,
      "loss": 0.6403,
      "step": 3770
    },
    {
      "epoch": 0.4772209567198178,
      "grad_norm": 1.0161722898483276,
      "learning_rate": 1.7913945891677876e-05,
      "loss": 0.6418,
      "step": 3771
    },
    {
      "epoch": 0.47734750696026323,
      "grad_norm": 1.0591785907745361,
      "learning_rate": 1.7912850229311642e-05,
      "loss": 0.6751,
      "step": 3772
    },
    {
      "epoch": 0.4774740572007087,
      "grad_norm": 1.0113210678100586,
      "learning_rate": 1.7911754312807842e-05,
      "loss": 0.6332,
      "step": 3773
    },
    {
      "epoch": 0.4776006074411541,
      "grad_norm": 1.1487067937850952,
      "learning_rate": 1.791065814220168e-05,
      "loss": 0.684,
      "step": 3774
    },
    {
      "epoch": 0.47772715768159957,
      "grad_norm": 0.9914533495903015,
      "learning_rate": 1.7909561717528364e-05,
      "loss": 0.6538,
      "step": 3775
    },
    {
      "epoch": 0.47785370792204507,
      "grad_norm": 1.1379528045654297,
      "learning_rate": 1.79084650388231e-05,
      "loss": 0.5998,
      "step": 3776
    },
    {
      "epoch": 0.4779802581624905,
      "grad_norm": 1.0110656023025513,
      "learning_rate": 1.7907368106121116e-05,
      "loss": 0.6605,
      "step": 3777
    },
    {
      "epoch": 0.47810680840293596,
      "grad_norm": 1.1098994016647339,
      "learning_rate": 1.7906270919457642e-05,
      "loss": 0.6169,
      "step": 3778
    },
    {
      "epoch": 0.4782333586433814,
      "grad_norm": 1.0036983489990234,
      "learning_rate": 1.7905173478867914e-05,
      "loss": 0.6095,
      "step": 3779
    },
    {
      "epoch": 0.4783599088838269,
      "grad_norm": 1.0021899938583374,
      "learning_rate": 1.7904075784387182e-05,
      "loss": 0.643,
      "step": 3780
    },
    {
      "epoch": 0.47848645912427235,
      "grad_norm": 1.0346513986587524,
      "learning_rate": 1.7902977836050696e-05,
      "loss": 0.6192,
      "step": 3781
    },
    {
      "epoch": 0.4786130093647178,
      "grad_norm": 1.0931202173233032,
      "learning_rate": 1.790187963389372e-05,
      "loss": 0.6568,
      "step": 3782
    },
    {
      "epoch": 0.47873955960516323,
      "grad_norm": 1.0547969341278076,
      "learning_rate": 1.7900781177951527e-05,
      "loss": 0.6653,
      "step": 3783
    },
    {
      "epoch": 0.47886610984560873,
      "grad_norm": 1.0672423839569092,
      "learning_rate": 1.7899682468259398e-05,
      "loss": 0.6326,
      "step": 3784
    },
    {
      "epoch": 0.4789926600860542,
      "grad_norm": 1.1085673570632935,
      "learning_rate": 1.7898583504852614e-05,
      "loss": 0.6349,
      "step": 3785
    },
    {
      "epoch": 0.4791192103264996,
      "grad_norm": 1.1780064105987549,
      "learning_rate": 1.7897484287766476e-05,
      "loss": 0.6635,
      "step": 3786
    },
    {
      "epoch": 0.47924576056694507,
      "grad_norm": 1.1199493408203125,
      "learning_rate": 1.789638481703629e-05,
      "loss": 0.6394,
      "step": 3787
    },
    {
      "epoch": 0.4793723108073905,
      "grad_norm": 1.0795456171035767,
      "learning_rate": 1.7895285092697358e-05,
      "loss": 0.6596,
      "step": 3788
    },
    {
      "epoch": 0.479498861047836,
      "grad_norm": 1.1757549047470093,
      "learning_rate": 1.7894185114785007e-05,
      "loss": 0.6773,
      "step": 3789
    },
    {
      "epoch": 0.47962541128828146,
      "grad_norm": 1.230859398841858,
      "learning_rate": 1.7893084883334566e-05,
      "loss": 0.6388,
      "step": 3790
    },
    {
      "epoch": 0.4797519615287269,
      "grad_norm": 1.2665318250656128,
      "learning_rate": 1.7891984398381364e-05,
      "loss": 0.6424,
      "step": 3791
    },
    {
      "epoch": 0.47987851176917234,
      "grad_norm": 1.335686445236206,
      "learning_rate": 1.7890883659960756e-05,
      "loss": 0.643,
      "step": 3792
    },
    {
      "epoch": 0.48000506200961784,
      "grad_norm": 1.2512445449829102,
      "learning_rate": 1.7889782668108086e-05,
      "loss": 0.647,
      "step": 3793
    },
    {
      "epoch": 0.4801316122500633,
      "grad_norm": 1.1429073810577393,
      "learning_rate": 1.7888681422858717e-05,
      "loss": 0.6528,
      "step": 3794
    },
    {
      "epoch": 0.48025816249050873,
      "grad_norm": 1.2045130729675293,
      "learning_rate": 1.788757992424802e-05,
      "loss": 0.6164,
      "step": 3795
    },
    {
      "epoch": 0.4803847127309542,
      "grad_norm": 1.1011337041854858,
      "learning_rate": 1.788647817231137e-05,
      "loss": 0.6637,
      "step": 3796
    },
    {
      "epoch": 0.4805112629713996,
      "grad_norm": 1.144209861755371,
      "learning_rate": 1.7885376167084148e-05,
      "loss": 0.697,
      "step": 3797
    },
    {
      "epoch": 0.4806378132118451,
      "grad_norm": 1.0024828910827637,
      "learning_rate": 1.7884273908601757e-05,
      "loss": 0.6275,
      "step": 3798
    },
    {
      "epoch": 0.48076436345229057,
      "grad_norm": 1.0077712535858154,
      "learning_rate": 1.788317139689959e-05,
      "loss": 0.6266,
      "step": 3799
    },
    {
      "epoch": 0.480890913692736,
      "grad_norm": 1.0106184482574463,
      "learning_rate": 1.788206863201306e-05,
      "loss": 0.6378,
      "step": 3800
    },
    {
      "epoch": 0.48101746393318146,
      "grad_norm": 1.07358717918396,
      "learning_rate": 1.788096561397758e-05,
      "loss": 0.6942,
      "step": 3801
    },
    {
      "epoch": 0.48114401417362695,
      "grad_norm": 0.9711339473724365,
      "learning_rate": 1.7879862342828583e-05,
      "loss": 0.6419,
      "step": 3802
    },
    {
      "epoch": 0.4812705644140724,
      "grad_norm": 0.9597119092941284,
      "learning_rate": 1.78787588186015e-05,
      "loss": 0.6338,
      "step": 3803
    },
    {
      "epoch": 0.48139711465451784,
      "grad_norm": 1.0565906763076782,
      "learning_rate": 1.787765504133177e-05,
      "loss": 0.6384,
      "step": 3804
    },
    {
      "epoch": 0.4815236648949633,
      "grad_norm": 1.0552022457122803,
      "learning_rate": 1.7876551011054844e-05,
      "loss": 0.6574,
      "step": 3805
    },
    {
      "epoch": 0.48165021513540873,
      "grad_norm": 1.095504879951477,
      "learning_rate": 1.7875446727806186e-05,
      "loss": 0.6886,
      "step": 3806
    },
    {
      "epoch": 0.48177676537585423,
      "grad_norm": 0.9999192357063293,
      "learning_rate": 1.7874342191621252e-05,
      "loss": 0.6393,
      "step": 3807
    },
    {
      "epoch": 0.4819033156162997,
      "grad_norm": 1.1025633811950684,
      "learning_rate": 1.7873237402535527e-05,
      "loss": 0.6166,
      "step": 3808
    },
    {
      "epoch": 0.4820298658567451,
      "grad_norm": 1.1233006715774536,
      "learning_rate": 1.7872132360584486e-05,
      "loss": 0.6645,
      "step": 3809
    },
    {
      "epoch": 0.48215641609719057,
      "grad_norm": 1.1535712480545044,
      "learning_rate": 1.7871027065803624e-05,
      "loss": 0.6348,
      "step": 3810
    },
    {
      "epoch": 0.48228296633763607,
      "grad_norm": 1.0836893320083618,
      "learning_rate": 1.786992151822844e-05,
      "loss": 0.7173,
      "step": 3811
    },
    {
      "epoch": 0.4824095165780815,
      "grad_norm": 1.0691277980804443,
      "learning_rate": 1.7868815717894438e-05,
      "loss": 0.6223,
      "step": 3812
    },
    {
      "epoch": 0.48253606681852695,
      "grad_norm": 1.0641636848449707,
      "learning_rate": 1.786770966483713e-05,
      "loss": 0.6496,
      "step": 3813
    },
    {
      "epoch": 0.4826626170589724,
      "grad_norm": 1.002548336982727,
      "learning_rate": 1.7866603359092046e-05,
      "loss": 0.5983,
      "step": 3814
    },
    {
      "epoch": 0.48278916729941784,
      "grad_norm": 0.9886831641197205,
      "learning_rate": 1.786549680069472e-05,
      "loss": 0.6318,
      "step": 3815
    },
    {
      "epoch": 0.48291571753986334,
      "grad_norm": 1.004212737083435,
      "learning_rate": 1.786438998968068e-05,
      "loss": 0.6329,
      "step": 3816
    },
    {
      "epoch": 0.4830422677803088,
      "grad_norm": 1.0373752117156982,
      "learning_rate": 1.7863282926085485e-05,
      "loss": 0.6274,
      "step": 3817
    },
    {
      "epoch": 0.48316881802075423,
      "grad_norm": 1.5390236377716064,
      "learning_rate": 1.786217560994468e-05,
      "loss": 0.6008,
      "step": 3818
    },
    {
      "epoch": 0.4832953682611997,
      "grad_norm": 1.0182251930236816,
      "learning_rate": 1.786106804129384e-05,
      "loss": 0.6509,
      "step": 3819
    },
    {
      "epoch": 0.4834219185016452,
      "grad_norm": 0.9404624700546265,
      "learning_rate": 1.7859960220168524e-05,
      "loss": 0.5849,
      "step": 3820
    },
    {
      "epoch": 0.4835484687420906,
      "grad_norm": 1.0375182628631592,
      "learning_rate": 1.7858852146604325e-05,
      "loss": 0.6866,
      "step": 3821
    },
    {
      "epoch": 0.48367501898253606,
      "grad_norm": 0.9943538308143616,
      "learning_rate": 1.785774382063682e-05,
      "loss": 0.6366,
      "step": 3822
    },
    {
      "epoch": 0.4838015692229815,
      "grad_norm": 1.0644606351852417,
      "learning_rate": 1.7856635242301615e-05,
      "loss": 0.657,
      "step": 3823
    },
    {
      "epoch": 0.48392811946342695,
      "grad_norm": 0.9338289499282837,
      "learning_rate": 1.785552641163431e-05,
      "loss": 0.6564,
      "step": 3824
    },
    {
      "epoch": 0.48405466970387245,
      "grad_norm": 0.9913939833641052,
      "learning_rate": 1.785441732867051e-05,
      "loss": 0.6121,
      "step": 3825
    },
    {
      "epoch": 0.4841812199443179,
      "grad_norm": 1.0165678262710571,
      "learning_rate": 1.785330799344585e-05,
      "loss": 0.5981,
      "step": 3826
    },
    {
      "epoch": 0.48430777018476334,
      "grad_norm": 0.9773701429367065,
      "learning_rate": 1.7852198405995952e-05,
      "loss": 0.6799,
      "step": 3827
    },
    {
      "epoch": 0.4844343204252088,
      "grad_norm": 0.9775345325469971,
      "learning_rate": 1.7851088566356447e-05,
      "loss": 0.6352,
      "step": 3828
    },
    {
      "epoch": 0.4845608706656543,
      "grad_norm": 1.0114829540252686,
      "learning_rate": 1.7849978474562985e-05,
      "loss": 0.6516,
      "step": 3829
    },
    {
      "epoch": 0.48468742090609973,
      "grad_norm": 0.9941225051879883,
      "learning_rate": 1.784886813065122e-05,
      "loss": 0.6492,
      "step": 3830
    },
    {
      "epoch": 0.4848139711465452,
      "grad_norm": 1.0661499500274658,
      "learning_rate": 1.784775753465681e-05,
      "loss": 0.6552,
      "step": 3831
    },
    {
      "epoch": 0.4849405213869906,
      "grad_norm": 0.9994024634361267,
      "learning_rate": 1.7846646686615426e-05,
      "loss": 0.5872,
      "step": 3832
    },
    {
      "epoch": 0.4850670716274361,
      "grad_norm": 1.1061444282531738,
      "learning_rate": 1.7845535586562746e-05,
      "loss": 0.647,
      "step": 3833
    },
    {
      "epoch": 0.48519362186788156,
      "grad_norm": 1.0368949174880981,
      "learning_rate": 1.7844424234534452e-05,
      "loss": 0.6482,
      "step": 3834
    },
    {
      "epoch": 0.485320172108327,
      "grad_norm": 1.0388342142105103,
      "learning_rate": 1.784331263056624e-05,
      "loss": 0.6141,
      "step": 3835
    },
    {
      "epoch": 0.48544672234877245,
      "grad_norm": 1.0480427742004395,
      "learning_rate": 1.7842200774693813e-05,
      "loss": 0.652,
      "step": 3836
    },
    {
      "epoch": 0.4855732725892179,
      "grad_norm": 0.9877181053161621,
      "learning_rate": 1.7841088666952873e-05,
      "loss": 0.6317,
      "step": 3837
    },
    {
      "epoch": 0.4856998228296634,
      "grad_norm": 1.0595611333847046,
      "learning_rate": 1.783997630737915e-05,
      "loss": 0.7159,
      "step": 3838
    },
    {
      "epoch": 0.48582637307010884,
      "grad_norm": 1.0294756889343262,
      "learning_rate": 1.7838863696008357e-05,
      "loss": 0.655,
      "step": 3839
    },
    {
      "epoch": 0.4859529233105543,
      "grad_norm": 1.114383578300476,
      "learning_rate": 1.783775083287624e-05,
      "loss": 0.6512,
      "step": 3840
    },
    {
      "epoch": 0.48607947355099973,
      "grad_norm": 1.0972636938095093,
      "learning_rate": 1.783663771801853e-05,
      "loss": 0.5967,
      "step": 3841
    },
    {
      "epoch": 0.48620602379144523,
      "grad_norm": 1.2182155847549438,
      "learning_rate": 1.783552435147098e-05,
      "loss": 0.6751,
      "step": 3842
    },
    {
      "epoch": 0.4863325740318907,
      "grad_norm": 0.9985271096229553,
      "learning_rate": 1.7834410733269352e-05,
      "loss": 0.6491,
      "step": 3843
    },
    {
      "epoch": 0.4864591242723361,
      "grad_norm": 1.095234751701355,
      "learning_rate": 1.7833296863449408e-05,
      "loss": 0.6594,
      "step": 3844
    },
    {
      "epoch": 0.48658567451278156,
      "grad_norm": 1.1570243835449219,
      "learning_rate": 1.7832182742046926e-05,
      "loss": 0.6664,
      "step": 3845
    },
    {
      "epoch": 0.486712224753227,
      "grad_norm": 1.063307523727417,
      "learning_rate": 1.7831068369097685e-05,
      "loss": 0.6756,
      "step": 3846
    },
    {
      "epoch": 0.4868387749936725,
      "grad_norm": 1.036694049835205,
      "learning_rate": 1.7829953744637477e-05,
      "loss": 0.6424,
      "step": 3847
    },
    {
      "epoch": 0.48696532523411795,
      "grad_norm": 1.0423901081085205,
      "learning_rate": 1.7828838868702098e-05,
      "loss": 0.6456,
      "step": 3848
    },
    {
      "epoch": 0.4870918754745634,
      "grad_norm": 1.0674798488616943,
      "learning_rate": 1.782772374132736e-05,
      "loss": 0.6203,
      "step": 3849
    },
    {
      "epoch": 0.48721842571500884,
      "grad_norm": 0.9412623047828674,
      "learning_rate": 1.7826608362549073e-05,
      "loss": 0.6585,
      "step": 3850
    },
    {
      "epoch": 0.48734497595545434,
      "grad_norm": 1.2141387462615967,
      "learning_rate": 1.782549273240306e-05,
      "loss": 0.658,
      "step": 3851
    },
    {
      "epoch": 0.4874715261958998,
      "grad_norm": 0.9492089748382568,
      "learning_rate": 1.7824376850925154e-05,
      "loss": 0.6105,
      "step": 3852
    },
    {
      "epoch": 0.48759807643634523,
      "grad_norm": 1.2181345224380493,
      "learning_rate": 1.7823260718151192e-05,
      "loss": 0.6494,
      "step": 3853
    },
    {
      "epoch": 0.4877246266767907,
      "grad_norm": 0.9877932667732239,
      "learning_rate": 1.7822144334117022e-05,
      "loss": 0.6264,
      "step": 3854
    },
    {
      "epoch": 0.4878511769172361,
      "grad_norm": 1.342203974723816,
      "learning_rate": 1.7821027698858497e-05,
      "loss": 0.6775,
      "step": 3855
    },
    {
      "epoch": 0.4879777271576816,
      "grad_norm": 1.143002986907959,
      "learning_rate": 1.7819910812411484e-05,
      "loss": 0.6478,
      "step": 3856
    },
    {
      "epoch": 0.48810427739812706,
      "grad_norm": 1.363236665725708,
      "learning_rate": 1.781879367481185e-05,
      "loss": 0.69,
      "step": 3857
    },
    {
      "epoch": 0.4882308276385725,
      "grad_norm": 1.129618525505066,
      "learning_rate": 1.7817676286095476e-05,
      "loss": 0.6333,
      "step": 3858
    },
    {
      "epoch": 0.48835737787901795,
      "grad_norm": 1.130828857421875,
      "learning_rate": 1.781655864629825e-05,
      "loss": 0.6342,
      "step": 3859
    },
    {
      "epoch": 0.48848392811946345,
      "grad_norm": 1.0913642644882202,
      "learning_rate": 1.7815440755456063e-05,
      "loss": 0.6475,
      "step": 3860
    },
    {
      "epoch": 0.4886104783599089,
      "grad_norm": 1.0129860639572144,
      "learning_rate": 1.7814322613604826e-05,
      "loss": 0.5779,
      "step": 3861
    },
    {
      "epoch": 0.48873702860035434,
      "grad_norm": 1.0795023441314697,
      "learning_rate": 1.7813204220780444e-05,
      "loss": 0.6277,
      "step": 3862
    },
    {
      "epoch": 0.4888635788407998,
      "grad_norm": 1.068575382232666,
      "learning_rate": 1.7812085577018838e-05,
      "loss": 0.6431,
      "step": 3863
    },
    {
      "epoch": 0.48899012908124523,
      "grad_norm": 1.05008864402771,
      "learning_rate": 1.781096668235594e-05,
      "loss": 0.6436,
      "step": 3864
    },
    {
      "epoch": 0.48911667932169073,
      "grad_norm": 1.12154221534729,
      "learning_rate": 1.780984753682768e-05,
      "loss": 0.6593,
      "step": 3865
    },
    {
      "epoch": 0.4892432295621362,
      "grad_norm": 1.117016077041626,
      "learning_rate": 1.780872814047e-05,
      "loss": 0.6786,
      "step": 3866
    },
    {
      "epoch": 0.4893697798025816,
      "grad_norm": 1.1129906177520752,
      "learning_rate": 1.7807608493318864e-05,
      "loss": 0.6524,
      "step": 3867
    },
    {
      "epoch": 0.48949633004302706,
      "grad_norm": 1.13564932346344,
      "learning_rate": 1.7806488595410213e-05,
      "loss": 0.6137,
      "step": 3868
    },
    {
      "epoch": 0.48962288028347256,
      "grad_norm": 1.0551520586013794,
      "learning_rate": 1.780536844678003e-05,
      "loss": 0.6208,
      "step": 3869
    },
    {
      "epoch": 0.489749430523918,
      "grad_norm": 1.2829041481018066,
      "learning_rate": 1.7804248047464287e-05,
      "loss": 0.6645,
      "step": 3870
    },
    {
      "epoch": 0.48987598076436345,
      "grad_norm": 1.038827896118164,
      "learning_rate": 1.7803127397498965e-05,
      "loss": 0.6912,
      "step": 3871
    },
    {
      "epoch": 0.4900025310048089,
      "grad_norm": 1.0587564706802368,
      "learning_rate": 1.7802006496920054e-05,
      "loss": 0.6381,
      "step": 3872
    },
    {
      "epoch": 0.49012908124525434,
      "grad_norm": 0.9919149279594421,
      "learning_rate": 1.7800885345763562e-05,
      "loss": 0.6362,
      "step": 3873
    },
    {
      "epoch": 0.49025563148569984,
      "grad_norm": 0.9560357928276062,
      "learning_rate": 1.7799763944065495e-05,
      "loss": 0.604,
      "step": 3874
    },
    {
      "epoch": 0.4903821817261453,
      "grad_norm": 0.9955530166625977,
      "learning_rate": 1.7798642291861862e-05,
      "loss": 0.6218,
      "step": 3875
    },
    {
      "epoch": 0.4905087319665907,
      "grad_norm": 1.2107939720153809,
      "learning_rate": 1.7797520389188697e-05,
      "loss": 0.6991,
      "step": 3876
    },
    {
      "epoch": 0.49063528220703617,
      "grad_norm": 1.0238230228424072,
      "learning_rate": 1.7796398236082024e-05,
      "loss": 0.6318,
      "step": 3877
    },
    {
      "epoch": 0.49076183244748167,
      "grad_norm": 0.9812856912612915,
      "learning_rate": 1.7795275832577887e-05,
      "loss": 0.6415,
      "step": 3878
    },
    {
      "epoch": 0.4908883826879271,
      "grad_norm": 0.9449880719184875,
      "learning_rate": 1.7794153178712335e-05,
      "loss": 0.6277,
      "step": 3879
    },
    {
      "epoch": 0.49101493292837256,
      "grad_norm": 0.9706348180770874,
      "learning_rate": 1.7793030274521423e-05,
      "loss": 0.5983,
      "step": 3880
    },
    {
      "epoch": 0.491141483168818,
      "grad_norm": 1.028849482536316,
      "learning_rate": 1.7791907120041215e-05,
      "loss": 0.6835,
      "step": 3881
    },
    {
      "epoch": 0.4912680334092635,
      "grad_norm": 0.9806727766990662,
      "learning_rate": 1.7790783715307783e-05,
      "loss": 0.658,
      "step": 3882
    },
    {
      "epoch": 0.49139458364970895,
      "grad_norm": 0.9839315414428711,
      "learning_rate": 1.7789660060357214e-05,
      "loss": 0.6114,
      "step": 3883
    },
    {
      "epoch": 0.4915211338901544,
      "grad_norm": 0.9537076354026794,
      "learning_rate": 1.778853615522559e-05,
      "loss": 0.6208,
      "step": 3884
    },
    {
      "epoch": 0.49164768413059984,
      "grad_norm": 0.96515291929245,
      "learning_rate": 1.7787411999949003e-05,
      "loss": 0.6318,
      "step": 3885
    },
    {
      "epoch": 0.4917742343710453,
      "grad_norm": 0.9368592500686646,
      "learning_rate": 1.778628759456357e-05,
      "loss": 0.5983,
      "step": 3886
    },
    {
      "epoch": 0.4919007846114908,
      "grad_norm": 1.0348082780838013,
      "learning_rate": 1.7785162939105395e-05,
      "loss": 0.6666,
      "step": 3887
    },
    {
      "epoch": 0.4920273348519362,
      "grad_norm": 1.0622079372406006,
      "learning_rate": 1.7784038033610594e-05,
      "loss": 0.6298,
      "step": 3888
    },
    {
      "epoch": 0.49215388509238167,
      "grad_norm": 0.9998353123664856,
      "learning_rate": 1.778291287811531e-05,
      "loss": 0.6237,
      "step": 3889
    },
    {
      "epoch": 0.4922804353328271,
      "grad_norm": 1.0289629697799683,
      "learning_rate": 1.7781787472655668e-05,
      "loss": 0.6435,
      "step": 3890
    },
    {
      "epoch": 0.4924069855732726,
      "grad_norm": 1.1081048250198364,
      "learning_rate": 1.778066181726782e-05,
      "loss": 0.6614,
      "step": 3891
    },
    {
      "epoch": 0.49253353581371806,
      "grad_norm": 1.0124738216400146,
      "learning_rate": 1.777953591198791e-05,
      "loss": 0.6779,
      "step": 3892
    },
    {
      "epoch": 0.4926600860541635,
      "grad_norm": 1.3178901672363281,
      "learning_rate": 1.7778409756852108e-05,
      "loss": 0.6462,
      "step": 3893
    },
    {
      "epoch": 0.49278663629460895,
      "grad_norm": 1.0380499362945557,
      "learning_rate": 1.777728335189658e-05,
      "loss": 0.67,
      "step": 3894
    },
    {
      "epoch": 0.4929131865350544,
      "grad_norm": 1.1356054544448853,
      "learning_rate": 1.77761566971575e-05,
      "loss": 0.6452,
      "step": 3895
    },
    {
      "epoch": 0.4930397367754999,
      "grad_norm": 1.132164716720581,
      "learning_rate": 1.7775029792671055e-05,
      "loss": 0.6792,
      "step": 3896
    },
    {
      "epoch": 0.49316628701594534,
      "grad_norm": 1.110201358795166,
      "learning_rate": 1.7773902638473437e-05,
      "loss": 0.616,
      "step": 3897
    },
    {
      "epoch": 0.4932928372563908,
      "grad_norm": 1.0754413604736328,
      "learning_rate": 1.7772775234600847e-05,
      "loss": 0.6155,
      "step": 3898
    },
    {
      "epoch": 0.4934193874968362,
      "grad_norm": 1.115430235862732,
      "learning_rate": 1.7771647581089494e-05,
      "loss": 0.6295,
      "step": 3899
    },
    {
      "epoch": 0.4935459377372817,
      "grad_norm": 1.1807103157043457,
      "learning_rate": 1.7770519677975597e-05,
      "loss": 0.6707,
      "step": 3900
    },
    {
      "epoch": 0.49367248797772717,
      "grad_norm": 1.1192072629928589,
      "learning_rate": 1.776939152529538e-05,
      "loss": 0.6569,
      "step": 3901
    },
    {
      "epoch": 0.4937990382181726,
      "grad_norm": 1.09604012966156,
      "learning_rate": 1.776826312308507e-05,
      "loss": 0.6316,
      "step": 3902
    },
    {
      "epoch": 0.49392558845861806,
      "grad_norm": 1.041379690170288,
      "learning_rate": 1.7767134471380917e-05,
      "loss": 0.6573,
      "step": 3903
    },
    {
      "epoch": 0.4940521386990635,
      "grad_norm": 1.0621203184127808,
      "learning_rate": 1.7766005570219168e-05,
      "loss": 0.6195,
      "step": 3904
    },
    {
      "epoch": 0.494178688939509,
      "grad_norm": 1.1248962879180908,
      "learning_rate": 1.776487641963607e-05,
      "loss": 0.6242,
      "step": 3905
    },
    {
      "epoch": 0.49430523917995445,
      "grad_norm": 1.1207420825958252,
      "learning_rate": 1.7763747019667904e-05,
      "loss": 0.6233,
      "step": 3906
    },
    {
      "epoch": 0.4944317894203999,
      "grad_norm": 1.0553897619247437,
      "learning_rate": 1.7762617370350934e-05,
      "loss": 0.6497,
      "step": 3907
    },
    {
      "epoch": 0.49455833966084534,
      "grad_norm": 0.978480339050293,
      "learning_rate": 1.7761487471721438e-05,
      "loss": 0.6604,
      "step": 3908
    },
    {
      "epoch": 0.49468488990129084,
      "grad_norm": 1.060784935951233,
      "learning_rate": 1.7760357323815714e-05,
      "loss": 0.6396,
      "step": 3909
    },
    {
      "epoch": 0.4948114401417363,
      "grad_norm": 1.036144495010376,
      "learning_rate": 1.7759226926670052e-05,
      "loss": 0.6523,
      "step": 3910
    },
    {
      "epoch": 0.4949379903821817,
      "grad_norm": 1.0469727516174316,
      "learning_rate": 1.775809628032076e-05,
      "loss": 0.6287,
      "step": 3911
    },
    {
      "epoch": 0.49506454062262717,
      "grad_norm": 1.0111886262893677,
      "learning_rate": 1.775696538480415e-05,
      "loss": 0.6813,
      "step": 3912
    },
    {
      "epoch": 0.4951910908630726,
      "grad_norm": 1.0299443006515503,
      "learning_rate": 1.775583424015654e-05,
      "loss": 0.6222,
      "step": 3913
    },
    {
      "epoch": 0.4953176411035181,
      "grad_norm": 1.042903184890747,
      "learning_rate": 1.7754702846414268e-05,
      "loss": 0.6671,
      "step": 3914
    },
    {
      "epoch": 0.49544419134396356,
      "grad_norm": 1.230303168296814,
      "learning_rate": 1.7753571203613662e-05,
      "loss": 0.6913,
      "step": 3915
    },
    {
      "epoch": 0.495570741584409,
      "grad_norm": 1.057818055152893,
      "learning_rate": 1.7752439311791072e-05,
      "loss": 0.6489,
      "step": 3916
    },
    {
      "epoch": 0.49569729182485445,
      "grad_norm": 1.0760383605957031,
      "learning_rate": 1.7751307170982846e-05,
      "loss": 0.6121,
      "step": 3917
    },
    {
      "epoch": 0.49582384206529995,
      "grad_norm": 0.9713461995124817,
      "learning_rate": 1.7750174781225352e-05,
      "loss": 0.6209,
      "step": 3918
    },
    {
      "epoch": 0.4959503923057454,
      "grad_norm": 1.058954119682312,
      "learning_rate": 1.7749042142554954e-05,
      "loss": 0.6306,
      "step": 3919
    },
    {
      "epoch": 0.49607694254619084,
      "grad_norm": 6.495849609375,
      "learning_rate": 1.7747909255008032e-05,
      "loss": 0.6705,
      "step": 3920
    },
    {
      "epoch": 0.4962034927866363,
      "grad_norm": 1.069602608680725,
      "learning_rate": 1.7746776118620973e-05,
      "loss": 0.6097,
      "step": 3921
    },
    {
      "epoch": 0.4963300430270817,
      "grad_norm": 0.983797550201416,
      "learning_rate": 1.7745642733430162e-05,
      "loss": 0.6289,
      "step": 3922
    },
    {
      "epoch": 0.4964565932675272,
      "grad_norm": 1.0864920616149902,
      "learning_rate": 1.7744509099472006e-05,
      "loss": 0.6631,
      "step": 3923
    },
    {
      "epoch": 0.49658314350797267,
      "grad_norm": 1.2831703424453735,
      "learning_rate": 1.774337521678291e-05,
      "loss": 0.6679,
      "step": 3924
    },
    {
      "epoch": 0.4967096937484181,
      "grad_norm": 1.0591236352920532,
      "learning_rate": 1.7742241085399298e-05,
      "loss": 0.6161,
      "step": 3925
    },
    {
      "epoch": 0.49683624398886356,
      "grad_norm": 1.322955846786499,
      "learning_rate": 1.7741106705357588e-05,
      "loss": 0.6892,
      "step": 3926
    },
    {
      "epoch": 0.49696279422930906,
      "grad_norm": 1.0011287927627563,
      "learning_rate": 1.7739972076694215e-05,
      "loss": 0.6893,
      "step": 3927
    },
    {
      "epoch": 0.4970893444697545,
      "grad_norm": 0.9903163909912109,
      "learning_rate": 1.7738837199445625e-05,
      "loss": 0.6392,
      "step": 3928
    },
    {
      "epoch": 0.49721589471019995,
      "grad_norm": 1.0063953399658203,
      "learning_rate": 1.7737702073648258e-05,
      "loss": 0.6362,
      "step": 3929
    },
    {
      "epoch": 0.4973424449506454,
      "grad_norm": 1.031295657157898,
      "learning_rate": 1.7736566699338576e-05,
      "loss": 0.7325,
      "step": 3930
    },
    {
      "epoch": 0.4974689951910909,
      "grad_norm": 1.1711994409561157,
      "learning_rate": 1.7735431076553043e-05,
      "loss": 0.6907,
      "step": 3931
    },
    {
      "epoch": 0.49759554543153633,
      "grad_norm": 1.0036011934280396,
      "learning_rate": 1.7734295205328136e-05,
      "loss": 0.6446,
      "step": 3932
    },
    {
      "epoch": 0.4977220956719818,
      "grad_norm": 1.0261836051940918,
      "learning_rate": 1.773315908570033e-05,
      "loss": 0.6365,
      "step": 3933
    },
    {
      "epoch": 0.4978486459124272,
      "grad_norm": 1.0146286487579346,
      "learning_rate": 1.7732022717706113e-05,
      "loss": 0.6775,
      "step": 3934
    },
    {
      "epoch": 0.49797519615287267,
      "grad_norm": 1.0480272769927979,
      "learning_rate": 1.773088610138199e-05,
      "loss": 0.6668,
      "step": 3935
    },
    {
      "epoch": 0.49810174639331817,
      "grad_norm": 1.0465948581695557,
      "learning_rate": 1.7729749236764457e-05,
      "loss": 0.6231,
      "step": 3936
    },
    {
      "epoch": 0.4982282966337636,
      "grad_norm": 1.020687222480774,
      "learning_rate": 1.772861212389003e-05,
      "loss": 0.6255,
      "step": 3937
    },
    {
      "epoch": 0.49835484687420906,
      "grad_norm": 0.9659907221794128,
      "learning_rate": 1.772747476279523e-05,
      "loss": 0.6182,
      "step": 3938
    },
    {
      "epoch": 0.4984813971146545,
      "grad_norm": 1.04002845287323,
      "learning_rate": 1.7726337153516585e-05,
      "loss": 0.7147,
      "step": 3939
    },
    {
      "epoch": 0.4986079473551,
      "grad_norm": 1.052437424659729,
      "learning_rate": 1.7725199296090638e-05,
      "loss": 0.6882,
      "step": 3940
    },
    {
      "epoch": 0.49873449759554545,
      "grad_norm": 1.0293093919754028,
      "learning_rate": 1.772406119055392e-05,
      "loss": 0.6514,
      "step": 3941
    },
    {
      "epoch": 0.4988610478359909,
      "grad_norm": 0.9575158357620239,
      "learning_rate": 1.7722922836942995e-05,
      "loss": 0.6253,
      "step": 3942
    },
    {
      "epoch": 0.49898759807643633,
      "grad_norm": 0.9818516373634338,
      "learning_rate": 1.7721784235294423e-05,
      "loss": 0.6501,
      "step": 3943
    },
    {
      "epoch": 0.4991141483168818,
      "grad_norm": 1.309751033782959,
      "learning_rate": 1.7720645385644767e-05,
      "loss": 0.6476,
      "step": 3944
    },
    {
      "epoch": 0.4992406985573273,
      "grad_norm": 0.991097092628479,
      "learning_rate": 1.7719506288030604e-05,
      "loss": 0.6614,
      "step": 3945
    },
    {
      "epoch": 0.4993672487977727,
      "grad_norm": 1.0519996881484985,
      "learning_rate": 1.7718366942488526e-05,
      "loss": 0.6337,
      "step": 3946
    },
    {
      "epoch": 0.49949379903821817,
      "grad_norm": 1.0081238746643066,
      "learning_rate": 1.7717227349055116e-05,
      "loss": 0.6267,
      "step": 3947
    },
    {
      "epoch": 0.4996203492786636,
      "grad_norm": 1.0478965044021606,
      "learning_rate": 1.7716087507766978e-05,
      "loss": 0.6405,
      "step": 3948
    },
    {
      "epoch": 0.4997468995191091,
      "grad_norm": 1.0374400615692139,
      "learning_rate": 1.7714947418660722e-05,
      "loss": 0.6298,
      "step": 3949
    },
    {
      "epoch": 0.49987344975955456,
      "grad_norm": 1.0132006406784058,
      "learning_rate": 1.7713807081772966e-05,
      "loss": 0.6439,
      "step": 3950
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9857183694839478,
      "learning_rate": 1.771266649714033e-05,
      "loss": 0.6249,
      "step": 3951
    },
    {
      "epoch": 0.5001265502404455,
      "grad_norm": 1.0264167785644531,
      "learning_rate": 1.771152566479945e-05,
      "loss": 0.608,
      "step": 3952
    },
    {
      "epoch": 0.5002531004808909,
      "grad_norm": 1.0356078147888184,
      "learning_rate": 1.771038458478696e-05,
      "loss": 0.6598,
      "step": 3953
    },
    {
      "epoch": 0.5003796507213364,
      "grad_norm": 1.1028711795806885,
      "learning_rate": 1.7709243257139515e-05,
      "loss": 0.6477,
      "step": 3954
    },
    {
      "epoch": 0.5005062009617818,
      "grad_norm": 1.0042120218276978,
      "learning_rate": 1.7708101681893768e-05,
      "loss": 0.641,
      "step": 3955
    },
    {
      "epoch": 0.5006327512022273,
      "grad_norm": 0.9904719591140747,
      "learning_rate": 1.770695985908638e-05,
      "loss": 0.6471,
      "step": 3956
    },
    {
      "epoch": 0.5007593014426728,
      "grad_norm": 1.0007102489471436,
      "learning_rate": 1.770581778875403e-05,
      "loss": 0.6091,
      "step": 3957
    },
    {
      "epoch": 0.5008858516831182,
      "grad_norm": 1.0277128219604492,
      "learning_rate": 1.7704675470933393e-05,
      "loss": 0.6528,
      "step": 3958
    },
    {
      "epoch": 0.5010124019235637,
      "grad_norm": 1.0390955209732056,
      "learning_rate": 1.7703532905661158e-05,
      "loss": 0.6703,
      "step": 3959
    },
    {
      "epoch": 0.5011389521640092,
      "grad_norm": 1.0644727945327759,
      "learning_rate": 1.770239009297402e-05,
      "loss": 0.6263,
      "step": 3960
    },
    {
      "epoch": 0.5012655024044546,
      "grad_norm": 0.9924448132514954,
      "learning_rate": 1.7701247032908685e-05,
      "loss": 0.6487,
      "step": 3961
    },
    {
      "epoch": 0.5013920526449,
      "grad_norm": 1.141964316368103,
      "learning_rate": 1.770010372550186e-05,
      "loss": 0.646,
      "step": 3962
    },
    {
      "epoch": 0.5015186028853454,
      "grad_norm": 1.4920915365219116,
      "learning_rate": 1.7698960170790274e-05,
      "loss": 0.6524,
      "step": 3963
    },
    {
      "epoch": 0.5016451531257909,
      "grad_norm": 1.0045795440673828,
      "learning_rate": 1.7697816368810646e-05,
      "loss": 0.6371,
      "step": 3964
    },
    {
      "epoch": 0.5017717033662364,
      "grad_norm": 1.2878732681274414,
      "learning_rate": 1.7696672319599713e-05,
      "loss": 0.6611,
      "step": 3965
    },
    {
      "epoch": 0.5018982536066818,
      "grad_norm": 0.9712731838226318,
      "learning_rate": 1.7695528023194222e-05,
      "loss": 0.6469,
      "step": 3966
    },
    {
      "epoch": 0.5020248038471273,
      "grad_norm": 1.0403873920440674,
      "learning_rate": 1.7694383479630924e-05,
      "loss": 0.6673,
      "step": 3967
    },
    {
      "epoch": 0.5021513540875727,
      "grad_norm": 1.1344172954559326,
      "learning_rate": 1.7693238688946573e-05,
      "loss": 0.6394,
      "step": 3968
    },
    {
      "epoch": 0.5022779043280182,
      "grad_norm": 1.0912680625915527,
      "learning_rate": 1.769209365117794e-05,
      "loss": 0.6852,
      "step": 3969
    },
    {
      "epoch": 0.5024044545684637,
      "grad_norm": 1.0667967796325684,
      "learning_rate": 1.7690948366361806e-05,
      "loss": 0.6786,
      "step": 3970
    },
    {
      "epoch": 0.5025310048089091,
      "grad_norm": 1.0249662399291992,
      "learning_rate": 1.7689802834534942e-05,
      "loss": 0.5874,
      "step": 3971
    },
    {
      "epoch": 0.5026575550493546,
      "grad_norm": 1.0334569215774536,
      "learning_rate": 1.768865705573415e-05,
      "loss": 0.6349,
      "step": 3972
    },
    {
      "epoch": 0.5027841052898,
      "grad_norm": 1.1030828952789307,
      "learning_rate": 1.768751102999622e-05,
      "loss": 0.6357,
      "step": 3973
    },
    {
      "epoch": 0.5029106555302455,
      "grad_norm": 1.0779871940612793,
      "learning_rate": 1.7686364757357964e-05,
      "loss": 0.6019,
      "step": 3974
    },
    {
      "epoch": 0.503037205770691,
      "grad_norm": 1.0224939584732056,
      "learning_rate": 1.76852182378562e-05,
      "loss": 0.7015,
      "step": 3975
    },
    {
      "epoch": 0.5031637560111364,
      "grad_norm": 1.0538674592971802,
      "learning_rate": 1.7684071471527747e-05,
      "loss": 0.6582,
      "step": 3976
    },
    {
      "epoch": 0.5032903062515819,
      "grad_norm": 1.1277375221252441,
      "learning_rate": 1.7682924458409436e-05,
      "loss": 0.6311,
      "step": 3977
    },
    {
      "epoch": 0.5034168564920274,
      "grad_norm": 1.6715831756591797,
      "learning_rate": 1.7681777198538107e-05,
      "loss": 0.6325,
      "step": 3978
    },
    {
      "epoch": 0.5035434067324728,
      "grad_norm": 1.04865300655365,
      "learning_rate": 1.7680629691950604e-05,
      "loss": 0.6129,
      "step": 3979
    },
    {
      "epoch": 0.5036699569729183,
      "grad_norm": 1.0875663757324219,
      "learning_rate": 1.7679481938683783e-05,
      "loss": 0.6277,
      "step": 3980
    },
    {
      "epoch": 0.5037965072133637,
      "grad_norm": 1.1046762466430664,
      "learning_rate": 1.7678333938774506e-05,
      "loss": 0.6448,
      "step": 3981
    },
    {
      "epoch": 0.5039230574538092,
      "grad_norm": 1.0247490406036377,
      "learning_rate": 1.7677185692259645e-05,
      "loss": 0.6255,
      "step": 3982
    },
    {
      "epoch": 0.5040496076942547,
      "grad_norm": 1.0225920677185059,
      "learning_rate": 1.7676037199176077e-05,
      "loss": 0.6374,
      "step": 3983
    },
    {
      "epoch": 0.5041761579347,
      "grad_norm": 1.040948748588562,
      "learning_rate": 1.7674888459560686e-05,
      "loss": 0.5943,
      "step": 3984
    },
    {
      "epoch": 0.5043027081751456,
      "grad_norm": 1.210566520690918,
      "learning_rate": 1.7673739473450374e-05,
      "loss": 0.6406,
      "step": 3985
    },
    {
      "epoch": 0.5044292584155909,
      "grad_norm": 1.01771879196167,
      "learning_rate": 1.767259024088203e-05,
      "loss": 0.6294,
      "step": 3986
    },
    {
      "epoch": 0.5045558086560364,
      "grad_norm": 1.050480842590332,
      "learning_rate": 1.7671440761892578e-05,
      "loss": 0.6154,
      "step": 3987
    },
    {
      "epoch": 0.5046823588964819,
      "grad_norm": 1.0369288921356201,
      "learning_rate": 1.767029103651893e-05,
      "loss": 0.6591,
      "step": 3988
    },
    {
      "epoch": 0.5048089091369273,
      "grad_norm": 1.1465014219284058,
      "learning_rate": 1.766914106479801e-05,
      "loss": 0.6886,
      "step": 3989
    },
    {
      "epoch": 0.5049354593773728,
      "grad_norm": 1.0286684036254883,
      "learning_rate": 1.766799084676675e-05,
      "loss": 0.6764,
      "step": 3990
    },
    {
      "epoch": 0.5050620096178182,
      "grad_norm": 1.0720863342285156,
      "learning_rate": 1.76668403824621e-05,
      "loss": 0.6335,
      "step": 3991
    },
    {
      "epoch": 0.5051885598582637,
      "grad_norm": 1.1876827478408813,
      "learning_rate": 1.7665689671920997e-05,
      "loss": 0.6401,
      "step": 3992
    },
    {
      "epoch": 0.5053151100987092,
      "grad_norm": 1.0811212062835693,
      "learning_rate": 1.766453871518041e-05,
      "loss": 0.6588,
      "step": 3993
    },
    {
      "epoch": 0.5054416603391546,
      "grad_norm": 0.9825507998466492,
      "learning_rate": 1.7663387512277302e-05,
      "loss": 0.6547,
      "step": 3994
    },
    {
      "epoch": 0.5055682105796001,
      "grad_norm": 1.1250131130218506,
      "learning_rate": 1.7662236063248645e-05,
      "loss": 0.6256,
      "step": 3995
    },
    {
      "epoch": 0.5056947608200456,
      "grad_norm": 1.0172539949417114,
      "learning_rate": 1.7661084368131417e-05,
      "loss": 0.6827,
      "step": 3996
    },
    {
      "epoch": 0.505821311060491,
      "grad_norm": 1.0780781507492065,
      "learning_rate": 1.7659932426962608e-05,
      "loss": 0.6389,
      "step": 3997
    },
    {
      "epoch": 0.5059478613009365,
      "grad_norm": 0.9641669392585754,
      "learning_rate": 1.7658780239779218e-05,
      "loss": 0.5976,
      "step": 3998
    },
    {
      "epoch": 0.5060744115413819,
      "grad_norm": 1.1431715488433838,
      "learning_rate": 1.7657627806618253e-05,
      "loss": 0.6301,
      "step": 3999
    },
    {
      "epoch": 0.5062009617818274,
      "grad_norm": 1.02279531955719,
      "learning_rate": 1.765647512751672e-05,
      "loss": 0.6204,
      "step": 4000
    },
    {
      "epoch": 0.5063275120222729,
      "grad_norm": 0.946200966835022,
      "learning_rate": 1.7655322202511646e-05,
      "loss": 0.6168,
      "step": 4001
    },
    {
      "epoch": 0.5064540622627183,
      "grad_norm": 1.0325901508331299,
      "learning_rate": 1.765416903164005e-05,
      "loss": 0.677,
      "step": 4002
    },
    {
      "epoch": 0.5065806125031638,
      "grad_norm": 0.9984400272369385,
      "learning_rate": 1.765301561493898e-05,
      "loss": 0.6558,
      "step": 4003
    },
    {
      "epoch": 0.5067071627436092,
      "grad_norm": 1.034259557723999,
      "learning_rate": 1.765186195244548e-05,
      "loss": 0.6642,
      "step": 4004
    },
    {
      "epoch": 0.5068337129840547,
      "grad_norm": 0.9729730486869812,
      "learning_rate": 1.765070804419659e-05,
      "loss": 0.6172,
      "step": 4005
    },
    {
      "epoch": 0.5069602632245002,
      "grad_norm": 1.0778754949569702,
      "learning_rate": 1.7649553890229377e-05,
      "loss": 0.6419,
      "step": 4006
    },
    {
      "epoch": 0.5070868134649456,
      "grad_norm": 0.9771596789360046,
      "learning_rate": 1.7648399490580916e-05,
      "loss": 0.6388,
      "step": 4007
    },
    {
      "epoch": 0.507213363705391,
      "grad_norm": 0.9950843453407288,
      "learning_rate": 1.7647244845288273e-05,
      "loss": 0.6148,
      "step": 4008
    },
    {
      "epoch": 0.5073399139458366,
      "grad_norm": 1.1565368175506592,
      "learning_rate": 1.7646089954388536e-05,
      "loss": 0.5984,
      "step": 4009
    },
    {
      "epoch": 0.5074664641862819,
      "grad_norm": 1.1046972274780273,
      "learning_rate": 1.7644934817918796e-05,
      "loss": 0.6548,
      "step": 4010
    },
    {
      "epoch": 0.5075930144267274,
      "grad_norm": 1.0679678916931152,
      "learning_rate": 1.7643779435916153e-05,
      "loss": 0.6805,
      "step": 4011
    },
    {
      "epoch": 0.5077195646671728,
      "grad_norm": 1.0441012382507324,
      "learning_rate": 1.7642623808417714e-05,
      "loss": 0.6263,
      "step": 4012
    },
    {
      "epoch": 0.5078461149076183,
      "grad_norm": 1.0686031579971313,
      "learning_rate": 1.7641467935460595e-05,
      "loss": 0.6724,
      "step": 4013
    },
    {
      "epoch": 0.5079726651480638,
      "grad_norm": 1.152392029762268,
      "learning_rate": 1.7640311817081918e-05,
      "loss": 0.6288,
      "step": 4014
    },
    {
      "epoch": 0.5080992153885092,
      "grad_norm": 0.9437813758850098,
      "learning_rate": 1.7639155453318815e-05,
      "loss": 0.6128,
      "step": 4015
    },
    {
      "epoch": 0.5082257656289547,
      "grad_norm": 0.9891296625137329,
      "learning_rate": 1.763799884420843e-05,
      "loss": 0.6326,
      "step": 4016
    },
    {
      "epoch": 0.5083523158694001,
      "grad_norm": 1.0738584995269775,
      "learning_rate": 1.76368419897879e-05,
      "loss": 0.6513,
      "step": 4017
    },
    {
      "epoch": 0.5084788661098456,
      "grad_norm": 1.0206869840621948,
      "learning_rate": 1.7635684890094385e-05,
      "loss": 0.6346,
      "step": 4018
    },
    {
      "epoch": 0.5086054163502911,
      "grad_norm": 1.2019275426864624,
      "learning_rate": 1.763452754516505e-05,
      "loss": 0.658,
      "step": 4019
    },
    {
      "epoch": 0.5087319665907365,
      "grad_norm": 0.9918675422668457,
      "learning_rate": 1.763336995503706e-05,
      "loss": 0.6236,
      "step": 4020
    },
    {
      "epoch": 0.508858516831182,
      "grad_norm": 1.0451573133468628,
      "learning_rate": 1.76322121197476e-05,
      "loss": 0.6419,
      "step": 4021
    },
    {
      "epoch": 0.5089850670716274,
      "grad_norm": 1.0236995220184326,
      "learning_rate": 1.7631054039333847e-05,
      "loss": 0.6635,
      "step": 4022
    },
    {
      "epoch": 0.5091116173120729,
      "grad_norm": 0.9794477820396423,
      "learning_rate": 1.7629895713833006e-05,
      "loss": 0.6258,
      "step": 4023
    },
    {
      "epoch": 0.5092381675525184,
      "grad_norm": 0.9860031604766846,
      "learning_rate": 1.7628737143282268e-05,
      "loss": 0.6101,
      "step": 4024
    },
    {
      "epoch": 0.5093647177929638,
      "grad_norm": 1.0803897380828857,
      "learning_rate": 1.7627578327718853e-05,
      "loss": 0.6725,
      "step": 4025
    },
    {
      "epoch": 0.5094912680334093,
      "grad_norm": 1.098114013671875,
      "learning_rate": 1.7626419267179975e-05,
      "loss": 0.7012,
      "step": 4026
    },
    {
      "epoch": 0.5096178182738548,
      "grad_norm": 1.0326793193817139,
      "learning_rate": 1.7625259961702858e-05,
      "loss": 0.6409,
      "step": 4027
    },
    {
      "epoch": 0.5097443685143002,
      "grad_norm": 0.9709293842315674,
      "learning_rate": 1.7624100411324735e-05,
      "loss": 0.6535,
      "step": 4028
    },
    {
      "epoch": 0.5098709187547457,
      "grad_norm": 1.0319955348968506,
      "learning_rate": 1.762294061608285e-05,
      "loss": 0.6472,
      "step": 4029
    },
    {
      "epoch": 0.509997468995191,
      "grad_norm": 1.110695481300354,
      "learning_rate": 1.7621780576014452e-05,
      "loss": 0.6765,
      "step": 4030
    },
    {
      "epoch": 0.5101240192356365,
      "grad_norm": 1.0636708736419678,
      "learning_rate": 1.76206202911568e-05,
      "loss": 0.6197,
      "step": 4031
    },
    {
      "epoch": 0.510250569476082,
      "grad_norm": 2.356776237487793,
      "learning_rate": 1.7619459761547154e-05,
      "loss": 0.6743,
      "step": 4032
    },
    {
      "epoch": 0.5103771197165274,
      "grad_norm": 1.0347853899002075,
      "learning_rate": 1.7618298987222787e-05,
      "loss": 0.6374,
      "step": 4033
    },
    {
      "epoch": 0.5105036699569729,
      "grad_norm": 1.0734267234802246,
      "learning_rate": 1.7617137968220982e-05,
      "loss": 0.6083,
      "step": 4034
    },
    {
      "epoch": 0.5106302201974183,
      "grad_norm": 1.032182216644287,
      "learning_rate": 1.7615976704579027e-05,
      "loss": 0.6817,
      "step": 4035
    },
    {
      "epoch": 0.5107567704378638,
      "grad_norm": 0.9722873568534851,
      "learning_rate": 1.7614815196334218e-05,
      "loss": 0.6423,
      "step": 4036
    },
    {
      "epoch": 0.5108833206783093,
      "grad_norm": 0.9874975085258484,
      "learning_rate": 1.7613653443523865e-05,
      "loss": 0.6219,
      "step": 4037
    },
    {
      "epoch": 0.5110098709187547,
      "grad_norm": 0.9979323148727417,
      "learning_rate": 1.761249144618527e-05,
      "loss": 0.652,
      "step": 4038
    },
    {
      "epoch": 0.5111364211592002,
      "grad_norm": 0.9665987491607666,
      "learning_rate": 1.761132920435576e-05,
      "loss": 0.6453,
      "step": 4039
    },
    {
      "epoch": 0.5112629713996456,
      "grad_norm": 1.1587885618209839,
      "learning_rate": 1.7610166718072655e-05,
      "loss": 0.6559,
      "step": 4040
    },
    {
      "epoch": 0.5113895216400911,
      "grad_norm": 1.0246306657791138,
      "learning_rate": 1.7609003987373305e-05,
      "loss": 0.6539,
      "step": 4041
    },
    {
      "epoch": 0.5115160718805366,
      "grad_norm": 1.1339266300201416,
      "learning_rate": 1.760784101229504e-05,
      "loss": 0.62,
      "step": 4042
    },
    {
      "epoch": 0.511642622120982,
      "grad_norm": 0.9948379397392273,
      "learning_rate": 1.7606677792875215e-05,
      "loss": 0.643,
      "step": 4043
    },
    {
      "epoch": 0.5117691723614275,
      "grad_norm": 1.0471094846725464,
      "learning_rate": 1.7605514329151194e-05,
      "loss": 0.6726,
      "step": 4044
    },
    {
      "epoch": 0.511895722601873,
      "grad_norm": 1.0291378498077393,
      "learning_rate": 1.7604350621160336e-05,
      "loss": 0.6297,
      "step": 4045
    },
    {
      "epoch": 0.5120222728423184,
      "grad_norm": 1.0274906158447266,
      "learning_rate": 1.7603186668940024e-05,
      "loss": 0.6289,
      "step": 4046
    },
    {
      "epoch": 0.5121488230827639,
      "grad_norm": 0.9676612019538879,
      "learning_rate": 1.760202247252764e-05,
      "loss": 0.627,
      "step": 4047
    },
    {
      "epoch": 0.5122753733232093,
      "grad_norm": 0.9797534942626953,
      "learning_rate": 1.7600858031960567e-05,
      "loss": 0.6456,
      "step": 4048
    },
    {
      "epoch": 0.5124019235636548,
      "grad_norm": 1.0896323919296265,
      "learning_rate": 1.7599693347276205e-05,
      "loss": 0.6742,
      "step": 4049
    },
    {
      "epoch": 0.5125284738041003,
      "grad_norm": 0.9752399325370789,
      "learning_rate": 1.759852841851197e-05,
      "loss": 0.6187,
      "step": 4050
    },
    {
      "epoch": 0.5126550240445457,
      "grad_norm": 1.4677338600158691,
      "learning_rate": 1.7597363245705265e-05,
      "loss": 0.6382,
      "step": 4051
    },
    {
      "epoch": 0.5127815742849912,
      "grad_norm": 1.0490316152572632,
      "learning_rate": 1.7596197828893514e-05,
      "loss": 0.6255,
      "step": 4052
    },
    {
      "epoch": 0.5129081245254365,
      "grad_norm": 1.0418723821640015,
      "learning_rate": 1.7595032168114157e-05,
      "loss": 0.6951,
      "step": 4053
    },
    {
      "epoch": 0.513034674765882,
      "grad_norm": 1.158613920211792,
      "learning_rate": 1.759386626340462e-05,
      "loss": 0.7191,
      "step": 4054
    },
    {
      "epoch": 0.5131612250063275,
      "grad_norm": 1.018190622329712,
      "learning_rate": 1.759270011480235e-05,
      "loss": 0.6386,
      "step": 4055
    },
    {
      "epoch": 0.5132877752467729,
      "grad_norm": 1.0581557750701904,
      "learning_rate": 1.75915337223448e-05,
      "loss": 0.6676,
      "step": 4056
    },
    {
      "epoch": 0.5134143254872184,
      "grad_norm": 1.045455813407898,
      "learning_rate": 1.7590367086069444e-05,
      "loss": 0.6231,
      "step": 4057
    },
    {
      "epoch": 0.5135408757276639,
      "grad_norm": 0.9925214648246765,
      "learning_rate": 1.758920020601373e-05,
      "loss": 0.6231,
      "step": 4058
    },
    {
      "epoch": 0.5136674259681093,
      "grad_norm": 1.0668964385986328,
      "learning_rate": 1.758803308221515e-05,
      "loss": 0.6535,
      "step": 4059
    },
    {
      "epoch": 0.5137939762085548,
      "grad_norm": 1.0059781074523926,
      "learning_rate": 1.7586865714711185e-05,
      "loss": 0.6552,
      "step": 4060
    },
    {
      "epoch": 0.5139205264490002,
      "grad_norm": 1.0289661884307861,
      "learning_rate": 1.758569810353932e-05,
      "loss": 0.6415,
      "step": 4061
    },
    {
      "epoch": 0.5140470766894457,
      "grad_norm": 1.018892765045166,
      "learning_rate": 1.758453024873707e-05,
      "loss": 0.6382,
      "step": 4062
    },
    {
      "epoch": 0.5141736269298912,
      "grad_norm": 1.0264326333999634,
      "learning_rate": 1.758336215034193e-05,
      "loss": 0.6638,
      "step": 4063
    },
    {
      "epoch": 0.5143001771703366,
      "grad_norm": 1.0447402000427246,
      "learning_rate": 1.7582193808391423e-05,
      "loss": 0.6324,
      "step": 4064
    },
    {
      "epoch": 0.5144267274107821,
      "grad_norm": 1.0533779859542847,
      "learning_rate": 1.7581025222923064e-05,
      "loss": 0.6591,
      "step": 4065
    },
    {
      "epoch": 0.5145532776512275,
      "grad_norm": 0.9965164661407471,
      "learning_rate": 1.7579856393974398e-05,
      "loss": 0.6431,
      "step": 4066
    },
    {
      "epoch": 0.514679827891673,
      "grad_norm": 1.0754468441009521,
      "learning_rate": 1.7578687321582953e-05,
      "loss": 0.6561,
      "step": 4067
    },
    {
      "epoch": 0.5148063781321185,
      "grad_norm": 1.0177758932113647,
      "learning_rate": 1.757751800578628e-05,
      "loss": 0.6427,
      "step": 4068
    },
    {
      "epoch": 0.5149329283725639,
      "grad_norm": 1.0353988409042358,
      "learning_rate": 1.7576348446621934e-05,
      "loss": 0.6434,
      "step": 4069
    },
    {
      "epoch": 0.5150594786130094,
      "grad_norm": 1.0258400440216064,
      "learning_rate": 1.7575178644127483e-05,
      "loss": 0.6519,
      "step": 4070
    },
    {
      "epoch": 0.5151860288534548,
      "grad_norm": 0.982403576374054,
      "learning_rate": 1.757400859834049e-05,
      "loss": 0.6431,
      "step": 4071
    },
    {
      "epoch": 0.5153125790939003,
      "grad_norm": 1.181771159172058,
      "learning_rate": 1.7572838309298532e-05,
      "loss": 0.6865,
      "step": 4072
    },
    {
      "epoch": 0.5154391293343458,
      "grad_norm": 1.0117573738098145,
      "learning_rate": 1.7571667777039203e-05,
      "loss": 0.668,
      "step": 4073
    },
    {
      "epoch": 0.5155656795747912,
      "grad_norm": 1.1187008619308472,
      "learning_rate": 1.7570497001600095e-05,
      "loss": 0.6638,
      "step": 4074
    },
    {
      "epoch": 0.5156922298152367,
      "grad_norm": 1.0650858879089355,
      "learning_rate": 1.7569325983018805e-05,
      "loss": 0.6299,
      "step": 4075
    },
    {
      "epoch": 0.5158187800556822,
      "grad_norm": 1.0408228635787964,
      "learning_rate": 1.756815472133295e-05,
      "loss": 0.6676,
      "step": 4076
    },
    {
      "epoch": 0.5159453302961275,
      "grad_norm": 1.0633037090301514,
      "learning_rate": 1.756698321658014e-05,
      "loss": 0.6608,
      "step": 4077
    },
    {
      "epoch": 0.516071880536573,
      "grad_norm": 0.9380621910095215,
      "learning_rate": 1.7565811468798005e-05,
      "loss": 0.6367,
      "step": 4078
    },
    {
      "epoch": 0.5161984307770184,
      "grad_norm": 1.1553171873092651,
      "learning_rate": 1.7564639478024177e-05,
      "loss": 0.6821,
      "step": 4079
    },
    {
      "epoch": 0.5163249810174639,
      "grad_norm": 1.0154623985290527,
      "learning_rate": 1.7563467244296298e-05,
      "loss": 0.637,
      "step": 4080
    },
    {
      "epoch": 0.5164515312579094,
      "grad_norm": 1.058924674987793,
      "learning_rate": 1.7562294767652014e-05,
      "loss": 0.6532,
      "step": 4081
    },
    {
      "epoch": 0.5165780814983548,
      "grad_norm": 0.9908797740936279,
      "learning_rate": 1.7561122048128982e-05,
      "loss": 0.6335,
      "step": 4082
    },
    {
      "epoch": 0.5167046317388003,
      "grad_norm": 1.0255070924758911,
      "learning_rate": 1.7559949085764873e-05,
      "loss": 0.6533,
      "step": 4083
    },
    {
      "epoch": 0.5168311819792457,
      "grad_norm": 1.4627783298492432,
      "learning_rate": 1.7558775880597346e-05,
      "loss": 0.6427,
      "step": 4084
    },
    {
      "epoch": 0.5169577322196912,
      "grad_norm": 0.9592077732086182,
      "learning_rate": 1.7557602432664092e-05,
      "loss": 0.6755,
      "step": 4085
    },
    {
      "epoch": 0.5170842824601367,
      "grad_norm": 1.02475905418396,
      "learning_rate": 1.7556428742002797e-05,
      "loss": 0.5915,
      "step": 4086
    },
    {
      "epoch": 0.5172108327005821,
      "grad_norm": 1.0013190507888794,
      "learning_rate": 1.7555254808651152e-05,
      "loss": 0.6481,
      "step": 4087
    },
    {
      "epoch": 0.5173373829410276,
      "grad_norm": 1.1943860054016113,
      "learning_rate": 1.7554080632646862e-05,
      "loss": 0.6304,
      "step": 4088
    },
    {
      "epoch": 0.517463933181473,
      "grad_norm": 0.9842087626457214,
      "learning_rate": 1.755290621402764e-05,
      "loss": 0.6413,
      "step": 4089
    },
    {
      "epoch": 0.5175904834219185,
      "grad_norm": 1.0801136493682861,
      "learning_rate": 1.7551731552831206e-05,
      "loss": 0.6476,
      "step": 4090
    },
    {
      "epoch": 0.517717033662364,
      "grad_norm": 1.1444674730300903,
      "learning_rate": 1.7550556649095283e-05,
      "loss": 0.6868,
      "step": 4091
    },
    {
      "epoch": 0.5178435839028094,
      "grad_norm": 1.1257901191711426,
      "learning_rate": 1.754938150285761e-05,
      "loss": 0.6888,
      "step": 4092
    },
    {
      "epoch": 0.5179701341432549,
      "grad_norm": 1.0015796422958374,
      "learning_rate": 1.754820611415592e-05,
      "loss": 0.6369,
      "step": 4093
    },
    {
      "epoch": 0.5180966843837004,
      "grad_norm": 1.0603851079940796,
      "learning_rate": 1.7547030483027977e-05,
      "loss": 0.6576,
      "step": 4094
    },
    {
      "epoch": 0.5182232346241458,
      "grad_norm": 1.0558322668075562,
      "learning_rate": 1.7545854609511524e-05,
      "loss": 0.6267,
      "step": 4095
    },
    {
      "epoch": 0.5183497848645913,
      "grad_norm": 0.9814608693122864,
      "learning_rate": 1.7544678493644335e-05,
      "loss": 0.6523,
      "step": 4096
    },
    {
      "epoch": 0.5184763351050367,
      "grad_norm": 0.9936235547065735,
      "learning_rate": 1.7543502135464185e-05,
      "loss": 0.6397,
      "step": 4097
    },
    {
      "epoch": 0.5186028853454822,
      "grad_norm": 1.0889614820480347,
      "learning_rate": 1.7542325535008854e-05,
      "loss": 0.6974,
      "step": 4098
    },
    {
      "epoch": 0.5187294355859277,
      "grad_norm": 0.998439610004425,
      "learning_rate": 1.7541148692316127e-05,
      "loss": 0.6116,
      "step": 4099
    },
    {
      "epoch": 0.518855985826373,
      "grad_norm": 0.9729573130607605,
      "learning_rate": 1.7539971607423802e-05,
      "loss": 0.6416,
      "step": 4100
    },
    {
      "epoch": 0.5189825360668185,
      "grad_norm": 1.1325976848602295,
      "learning_rate": 1.7538794280369683e-05,
      "loss": 0.6359,
      "step": 4101
    },
    {
      "epoch": 0.5191090863072639,
      "grad_norm": 0.966227650642395,
      "learning_rate": 1.7537616711191587e-05,
      "loss": 0.6364,
      "step": 4102
    },
    {
      "epoch": 0.5192356365477094,
      "grad_norm": 1.0804615020751953,
      "learning_rate": 1.7536438899927327e-05,
      "loss": 0.6018,
      "step": 4103
    },
    {
      "epoch": 0.5193621867881549,
      "grad_norm": 1.0882880687713623,
      "learning_rate": 1.753526084661474e-05,
      "loss": 0.6703,
      "step": 4104
    },
    {
      "epoch": 0.5194887370286003,
      "grad_norm": 1.2032662630081177,
      "learning_rate": 1.7534082551291654e-05,
      "loss": 0.6781,
      "step": 4105
    },
    {
      "epoch": 0.5196152872690458,
      "grad_norm": 1.0169636011123657,
      "learning_rate": 1.7532904013995912e-05,
      "loss": 0.6821,
      "step": 4106
    },
    {
      "epoch": 0.5197418375094913,
      "grad_norm": 2.5788092613220215,
      "learning_rate": 1.7531725234765375e-05,
      "loss": 0.6836,
      "step": 4107
    },
    {
      "epoch": 0.5198683877499367,
      "grad_norm": 1.149085283279419,
      "learning_rate": 1.7530546213637888e-05,
      "loss": 0.6501,
      "step": 4108
    },
    {
      "epoch": 0.5199949379903822,
      "grad_norm": 0.9908795952796936,
      "learning_rate": 1.7529366950651327e-05,
      "loss": 0.6411,
      "step": 4109
    },
    {
      "epoch": 0.5201214882308276,
      "grad_norm": 1.0843958854675293,
      "learning_rate": 1.7528187445843565e-05,
      "loss": 0.6772,
      "step": 4110
    },
    {
      "epoch": 0.5202480384712731,
      "grad_norm": 1.0328060388565063,
      "learning_rate": 1.7527007699252484e-05,
      "loss": 0.66,
      "step": 4111
    },
    {
      "epoch": 0.5203745887117186,
      "grad_norm": 1.0095032453536987,
      "learning_rate": 1.7525827710915974e-05,
      "loss": 0.5934,
      "step": 4112
    },
    {
      "epoch": 0.520501138952164,
      "grad_norm": 1.0873081684112549,
      "learning_rate": 1.7524647480871927e-05,
      "loss": 0.6888,
      "step": 4113
    },
    {
      "epoch": 0.5206276891926095,
      "grad_norm": 1.0730372667312622,
      "learning_rate": 1.7523467009158258e-05,
      "loss": 0.632,
      "step": 4114
    },
    {
      "epoch": 0.5207542394330549,
      "grad_norm": 0.9972754120826721,
      "learning_rate": 1.7522286295812872e-05,
      "loss": 0.6107,
      "step": 4115
    },
    {
      "epoch": 0.5208807896735004,
      "grad_norm": 1.177927017211914,
      "learning_rate": 1.7521105340873696e-05,
      "loss": 0.6305,
      "step": 4116
    },
    {
      "epoch": 0.5210073399139459,
      "grad_norm": 1.0081144571304321,
      "learning_rate": 1.751992414437866e-05,
      "loss": 0.6287,
      "step": 4117
    },
    {
      "epoch": 0.5211338901543913,
      "grad_norm": 1.1203069686889648,
      "learning_rate": 1.751874270636569e-05,
      "loss": 0.6713,
      "step": 4118
    },
    {
      "epoch": 0.5212604403948368,
      "grad_norm": 0.9485503435134888,
      "learning_rate": 1.7517561026872743e-05,
      "loss": 0.6243,
      "step": 4119
    },
    {
      "epoch": 0.5213869906352822,
      "grad_norm": 1.0120116472244263,
      "learning_rate": 1.7516379105937767e-05,
      "loss": 0.6237,
      "step": 4120
    },
    {
      "epoch": 0.5215135408757277,
      "grad_norm": 1.007546305656433,
      "learning_rate": 1.751519694359872e-05,
      "loss": 0.6515,
      "step": 4121
    },
    {
      "epoch": 0.5216400911161732,
      "grad_norm": 0.9701309204101562,
      "learning_rate": 1.7514014539893566e-05,
      "loss": 0.6434,
      "step": 4122
    },
    {
      "epoch": 0.5217666413566185,
      "grad_norm": 0.9924671649932861,
      "learning_rate": 1.7512831894860287e-05,
      "loss": 0.6141,
      "step": 4123
    },
    {
      "epoch": 0.521893191597064,
      "grad_norm": 1.1942921876907349,
      "learning_rate": 1.7511649008536867e-05,
      "loss": 0.6219,
      "step": 4124
    },
    {
      "epoch": 0.5220197418375095,
      "grad_norm": 1.0537793636322021,
      "learning_rate": 1.7510465880961288e-05,
      "loss": 0.635,
      "step": 4125
    },
    {
      "epoch": 0.5221462920779549,
      "grad_norm": 1.0758732557296753,
      "learning_rate": 1.7509282512171556e-05,
      "loss": 0.6594,
      "step": 4126
    },
    {
      "epoch": 0.5222728423184004,
      "grad_norm": 1.089646339416504,
      "learning_rate": 1.7508098902205675e-05,
      "loss": 0.5853,
      "step": 4127
    },
    {
      "epoch": 0.5223993925588458,
      "grad_norm": 0.9996501207351685,
      "learning_rate": 1.750691505110166e-05,
      "loss": 0.6357,
      "step": 4128
    },
    {
      "epoch": 0.5225259427992913,
      "grad_norm": 1.0089170932769775,
      "learning_rate": 1.7505730958897532e-05,
      "loss": 0.6515,
      "step": 4129
    },
    {
      "epoch": 0.5226524930397368,
      "grad_norm": 1.0203502178192139,
      "learning_rate": 1.7504546625631326e-05,
      "loss": 0.6646,
      "step": 4130
    },
    {
      "epoch": 0.5227790432801822,
      "grad_norm": 1.0809646844863892,
      "learning_rate": 1.7503362051341068e-05,
      "loss": 0.6448,
      "step": 4131
    },
    {
      "epoch": 0.5229055935206277,
      "grad_norm": 1.0677824020385742,
      "learning_rate": 1.7502177236064806e-05,
      "loss": 0.6453,
      "step": 4132
    },
    {
      "epoch": 0.5230321437610731,
      "grad_norm": 1.0316158533096313,
      "learning_rate": 1.7500992179840604e-05,
      "loss": 0.6633,
      "step": 4133
    },
    {
      "epoch": 0.5231586940015186,
      "grad_norm": 1.0779893398284912,
      "learning_rate": 1.749980688270651e-05,
      "loss": 0.6544,
      "step": 4134
    },
    {
      "epoch": 0.5232852442419641,
      "grad_norm": 0.9542427659034729,
      "learning_rate": 1.74986213447006e-05,
      "loss": 0.6283,
      "step": 4135
    },
    {
      "epoch": 0.5234117944824095,
      "grad_norm": 1.0441476106643677,
      "learning_rate": 1.7497435565860942e-05,
      "loss": 0.6559,
      "step": 4136
    },
    {
      "epoch": 0.523538344722855,
      "grad_norm": 0.978478729724884,
      "learning_rate": 1.7496249546225626e-05,
      "loss": 0.6166,
      "step": 4137
    },
    {
      "epoch": 0.5236648949633004,
      "grad_norm": 0.953277587890625,
      "learning_rate": 1.749506328583275e-05,
      "loss": 0.6008,
      "step": 4138
    },
    {
      "epoch": 0.5237914452037459,
      "grad_norm": 1.0038237571716309,
      "learning_rate": 1.7493876784720396e-05,
      "loss": 0.6561,
      "step": 4139
    },
    {
      "epoch": 0.5239179954441914,
      "grad_norm": 0.9781518578529358,
      "learning_rate": 1.749269004292668e-05,
      "loss": 0.5981,
      "step": 4140
    },
    {
      "epoch": 0.5240445456846368,
      "grad_norm": 0.9607418775558472,
      "learning_rate": 1.749150306048972e-05,
      "loss": 0.63,
      "step": 4141
    },
    {
      "epoch": 0.5241710959250823,
      "grad_norm": 1.0322695970535278,
      "learning_rate": 1.7490315837447636e-05,
      "loss": 0.6736,
      "step": 4142
    },
    {
      "epoch": 0.5242976461655278,
      "grad_norm": 1.0242177248001099,
      "learning_rate": 1.748912837383856e-05,
      "loss": 0.6589,
      "step": 4143
    },
    {
      "epoch": 0.5244241964059732,
      "grad_norm": 1.1075679063796997,
      "learning_rate": 1.748794066970062e-05,
      "loss": 0.662,
      "step": 4144
    },
    {
      "epoch": 0.5245507466464187,
      "grad_norm": 1.0073598623275757,
      "learning_rate": 1.7486752725071975e-05,
      "loss": 0.6731,
      "step": 4145
    },
    {
      "epoch": 0.524677296886864,
      "grad_norm": 1.0181692838668823,
      "learning_rate": 1.7485564539990767e-05,
      "loss": 0.6474,
      "step": 4146
    },
    {
      "epoch": 0.5248038471273095,
      "grad_norm": 1.0700197219848633,
      "learning_rate": 1.7484376114495164e-05,
      "loss": 0.6472,
      "step": 4147
    },
    {
      "epoch": 0.524930397367755,
      "grad_norm": 1.0823767185211182,
      "learning_rate": 1.7483187448623337e-05,
      "loss": 0.7049,
      "step": 4148
    },
    {
      "epoch": 0.5250569476082004,
      "grad_norm": 1.0510648488998413,
      "learning_rate": 1.7481998542413453e-05,
      "loss": 0.6288,
      "step": 4149
    },
    {
      "epoch": 0.5251834978486459,
      "grad_norm": 0.9992852210998535,
      "learning_rate": 1.7480809395903703e-05,
      "loss": 0.5897,
      "step": 4150
    },
    {
      "epoch": 0.5253100480890913,
      "grad_norm": 1.1938040256500244,
      "learning_rate": 1.7479620009132278e-05,
      "loss": 0.6588,
      "step": 4151
    },
    {
      "epoch": 0.5254365983295368,
      "grad_norm": 0.9712004065513611,
      "learning_rate": 1.7478430382137377e-05,
      "loss": 0.6237,
      "step": 4152
    },
    {
      "epoch": 0.5255631485699823,
      "grad_norm": 1.165401816368103,
      "learning_rate": 1.7477240514957206e-05,
      "loss": 0.6717,
      "step": 4153
    },
    {
      "epoch": 0.5256896988104277,
      "grad_norm": 1.0211997032165527,
      "learning_rate": 1.7476050407629985e-05,
      "loss": 0.6354,
      "step": 4154
    },
    {
      "epoch": 0.5258162490508732,
      "grad_norm": 2.1672487258911133,
      "learning_rate": 1.747486006019393e-05,
      "loss": 0.6491,
      "step": 4155
    },
    {
      "epoch": 0.5259427992913187,
      "grad_norm": 1.099382996559143,
      "learning_rate": 1.7473669472687275e-05,
      "loss": 0.6335,
      "step": 4156
    },
    {
      "epoch": 0.5260693495317641,
      "grad_norm": 1.2175146341323853,
      "learning_rate": 1.7472478645148257e-05,
      "loss": 0.6268,
      "step": 4157
    },
    {
      "epoch": 0.5261958997722096,
      "grad_norm": 1.0211302042007446,
      "learning_rate": 1.7471287577615122e-05,
      "loss": 0.6259,
      "step": 4158
    },
    {
      "epoch": 0.526322450012655,
      "grad_norm": 1.0331388711929321,
      "learning_rate": 1.747009627012613e-05,
      "loss": 0.6285,
      "step": 4159
    },
    {
      "epoch": 0.5264490002531005,
      "grad_norm": 1.0780794620513916,
      "learning_rate": 1.746890472271953e-05,
      "loss": 0.6614,
      "step": 4160
    },
    {
      "epoch": 0.526575550493546,
      "grad_norm": 1.0267137289047241,
      "learning_rate": 1.7467712935433603e-05,
      "loss": 0.6613,
      "step": 4161
    },
    {
      "epoch": 0.5267021007339914,
      "grad_norm": 1.0777196884155273,
      "learning_rate": 1.7466520908306617e-05,
      "loss": 0.6241,
      "step": 4162
    },
    {
      "epoch": 0.5268286509744369,
      "grad_norm": 0.9902595281600952,
      "learning_rate": 1.746532864137686e-05,
      "loss": 0.6429,
      "step": 4163
    },
    {
      "epoch": 0.5269552012148823,
      "grad_norm": 0.9747466444969177,
      "learning_rate": 1.7464136134682626e-05,
      "loss": 0.6161,
      "step": 4164
    },
    {
      "epoch": 0.5270817514553278,
      "grad_norm": 1.0182294845581055,
      "learning_rate": 1.746294338826221e-05,
      "loss": 0.6718,
      "step": 4165
    },
    {
      "epoch": 0.5272083016957733,
      "grad_norm": 0.9990819692611694,
      "learning_rate": 1.7461750402153926e-05,
      "loss": 0.6171,
      "step": 4166
    },
    {
      "epoch": 0.5273348519362187,
      "grad_norm": 1.0082306861877441,
      "learning_rate": 1.746055717639608e-05,
      "loss": 0.6606,
      "step": 4167
    },
    {
      "epoch": 0.5274614021766642,
      "grad_norm": 1.0941110849380493,
      "learning_rate": 1.7459363711027008e-05,
      "loss": 0.6778,
      "step": 4168
    },
    {
      "epoch": 0.5275879524171095,
      "grad_norm": 0.9241145849227905,
      "learning_rate": 1.7458170006085023e-05,
      "loss": 0.624,
      "step": 4169
    },
    {
      "epoch": 0.527714502657555,
      "grad_norm": 0.9949114322662354,
      "learning_rate": 1.745697606160848e-05,
      "loss": 0.6332,
      "step": 4170
    },
    {
      "epoch": 0.5278410528980005,
      "grad_norm": 1.0167045593261719,
      "learning_rate": 1.7455781877635717e-05,
      "loss": 0.6438,
      "step": 4171
    },
    {
      "epoch": 0.5279676031384459,
      "grad_norm": 0.9726533889770508,
      "learning_rate": 1.745458745420509e-05,
      "loss": 0.6077,
      "step": 4172
    },
    {
      "epoch": 0.5280941533788914,
      "grad_norm": 1.0147241353988647,
      "learning_rate": 1.7453392791354957e-05,
      "loss": 0.6727,
      "step": 4173
    },
    {
      "epoch": 0.5282207036193369,
      "grad_norm": 1.0096521377563477,
      "learning_rate": 1.7452197889123692e-05,
      "loss": 0.6479,
      "step": 4174
    },
    {
      "epoch": 0.5283472538597823,
      "grad_norm": 1.0104920864105225,
      "learning_rate": 1.7451002747549668e-05,
      "loss": 0.6342,
      "step": 4175
    },
    {
      "epoch": 0.5284738041002278,
      "grad_norm": 1.0449905395507812,
      "learning_rate": 1.744980736667127e-05,
      "loss": 0.6615,
      "step": 4176
    },
    {
      "epoch": 0.5286003543406732,
      "grad_norm": 0.9867274761199951,
      "learning_rate": 1.7448611746526892e-05,
      "loss": 0.6936,
      "step": 4177
    },
    {
      "epoch": 0.5287269045811187,
      "grad_norm": 0.9455195665359497,
      "learning_rate": 1.7447415887154932e-05,
      "loss": 0.6461,
      "step": 4178
    },
    {
      "epoch": 0.5288534548215642,
      "grad_norm": 1.5031814575195312,
      "learning_rate": 1.7446219788593797e-05,
      "loss": 0.6485,
      "step": 4179
    },
    {
      "epoch": 0.5289800050620096,
      "grad_norm": 1.039778470993042,
      "learning_rate": 1.7445023450881905e-05,
      "loss": 0.6453,
      "step": 4180
    },
    {
      "epoch": 0.5291065553024551,
      "grad_norm": 0.9857110977172852,
      "learning_rate": 1.7443826874057674e-05,
      "loss": 0.6586,
      "step": 4181
    },
    {
      "epoch": 0.5292331055429005,
      "grad_norm": 1.0555042028427124,
      "learning_rate": 1.7442630058159542e-05,
      "loss": 0.6604,
      "step": 4182
    },
    {
      "epoch": 0.529359655783346,
      "grad_norm": 0.9818982481956482,
      "learning_rate": 1.744143300322594e-05,
      "loss": 0.6498,
      "step": 4183
    },
    {
      "epoch": 0.5294862060237915,
      "grad_norm": 1.0391827821731567,
      "learning_rate": 1.7440235709295316e-05,
      "loss": 0.6779,
      "step": 4184
    },
    {
      "epoch": 0.5296127562642369,
      "grad_norm": 0.9882618188858032,
      "learning_rate": 1.743903817640613e-05,
      "loss": 0.6419,
      "step": 4185
    },
    {
      "epoch": 0.5297393065046824,
      "grad_norm": 1.0270352363586426,
      "learning_rate": 1.743784040459683e-05,
      "loss": 0.6516,
      "step": 4186
    },
    {
      "epoch": 0.5298658567451278,
      "grad_norm": 1.0378817319869995,
      "learning_rate": 1.7436642393905896e-05,
      "loss": 0.6511,
      "step": 4187
    },
    {
      "epoch": 0.5299924069855733,
      "grad_norm": 0.9759880304336548,
      "learning_rate": 1.7435444144371802e-05,
      "loss": 0.6003,
      "step": 4188
    },
    {
      "epoch": 0.5301189572260188,
      "grad_norm": 1.0460172891616821,
      "learning_rate": 1.743424565603303e-05,
      "loss": 0.6169,
      "step": 4189
    },
    {
      "epoch": 0.5302455074664641,
      "grad_norm": 0.9411864280700684,
      "learning_rate": 1.7433046928928073e-05,
      "loss": 0.6165,
      "step": 4190
    },
    {
      "epoch": 0.5303720577069096,
      "grad_norm": 1.0169235467910767,
      "learning_rate": 1.743184796309543e-05,
      "loss": 0.6572,
      "step": 4191
    },
    {
      "epoch": 0.5304986079473551,
      "grad_norm": 1.028469443321228,
      "learning_rate": 1.7430648758573614e-05,
      "loss": 0.6063,
      "step": 4192
    },
    {
      "epoch": 0.5306251581878005,
      "grad_norm": 1.0244840383529663,
      "learning_rate": 1.742944931540113e-05,
      "loss": 0.6575,
      "step": 4193
    },
    {
      "epoch": 0.530751708428246,
      "grad_norm": 1.005407452583313,
      "learning_rate": 1.742824963361651e-05,
      "loss": 0.6145,
      "step": 4194
    },
    {
      "epoch": 0.5308782586686914,
      "grad_norm": 1.0682952404022217,
      "learning_rate": 1.7427049713258278e-05,
      "loss": 0.6363,
      "step": 4195
    },
    {
      "epoch": 0.5310048089091369,
      "grad_norm": 1.0462723970413208,
      "learning_rate": 1.742584955436497e-05,
      "loss": 0.6575,
      "step": 4196
    },
    {
      "epoch": 0.5311313591495824,
      "grad_norm": 0.9545847177505493,
      "learning_rate": 1.742464915697514e-05,
      "loss": 0.6419,
      "step": 4197
    },
    {
      "epoch": 0.5312579093900278,
      "grad_norm": 1.0268778800964355,
      "learning_rate": 1.7423448521127335e-05,
      "loss": 0.7075,
      "step": 4198
    },
    {
      "epoch": 0.5313844596304733,
      "grad_norm": 1.0220296382904053,
      "learning_rate": 1.742224764686012e-05,
      "loss": 0.6651,
      "step": 4199
    },
    {
      "epoch": 0.5315110098709187,
      "grad_norm": 1.0258550643920898,
      "learning_rate": 1.742104653421206e-05,
      "loss": 0.6113,
      "step": 4200
    },
    {
      "epoch": 0.5316375601113642,
      "grad_norm": 0.9885451197624207,
      "learning_rate": 1.741984518322173e-05,
      "loss": 0.6297,
      "step": 4201
    },
    {
      "epoch": 0.5317641103518097,
      "grad_norm": 1.0152106285095215,
      "learning_rate": 1.741864359392772e-05,
      "loss": 0.663,
      "step": 4202
    },
    {
      "epoch": 0.5318906605922551,
      "grad_norm": 0.9949878454208374,
      "learning_rate": 1.7417441766368613e-05,
      "loss": 0.648,
      "step": 4203
    },
    {
      "epoch": 0.5320172108327006,
      "grad_norm": 0.9759773015975952,
      "learning_rate": 1.741623970058302e-05,
      "loss": 0.6655,
      "step": 4204
    },
    {
      "epoch": 0.5321437610731461,
      "grad_norm": 0.9379879832267761,
      "learning_rate": 1.7415037396609533e-05,
      "loss": 0.6327,
      "step": 4205
    },
    {
      "epoch": 0.5322703113135915,
      "grad_norm": 1.0020771026611328,
      "learning_rate": 1.741383485448678e-05,
      "loss": 0.6103,
      "step": 4206
    },
    {
      "epoch": 0.532396861554037,
      "grad_norm": 0.9837918281555176,
      "learning_rate": 1.7412632074253376e-05,
      "loss": 0.6526,
      "step": 4207
    },
    {
      "epoch": 0.5325234117944824,
      "grad_norm": 0.9479527473449707,
      "learning_rate": 1.741142905594795e-05,
      "loss": 0.626,
      "step": 4208
    },
    {
      "epoch": 0.5326499620349279,
      "grad_norm": 1.0170249938964844,
      "learning_rate": 1.7410225799609144e-05,
      "loss": 0.6252,
      "step": 4209
    },
    {
      "epoch": 0.5327765122753734,
      "grad_norm": 0.9742327928543091,
      "learning_rate": 1.74090223052756e-05,
      "loss": 0.6231,
      "step": 4210
    },
    {
      "epoch": 0.5329030625158188,
      "grad_norm": 1.0283945798873901,
      "learning_rate": 1.7407818572985973e-05,
      "loss": 0.6721,
      "step": 4211
    },
    {
      "epoch": 0.5330296127562643,
      "grad_norm": 0.9785340428352356,
      "learning_rate": 1.7406614602778917e-05,
      "loss": 0.625,
      "step": 4212
    },
    {
      "epoch": 0.5331561629967096,
      "grad_norm": 1.024965524673462,
      "learning_rate": 1.740541039469311e-05,
      "loss": 0.6062,
      "step": 4213
    },
    {
      "epoch": 0.5332827132371551,
      "grad_norm": 1.0058246850967407,
      "learning_rate": 1.7404205948767217e-05,
      "loss": 0.6688,
      "step": 4214
    },
    {
      "epoch": 0.5334092634776006,
      "grad_norm": 1.0200526714324951,
      "learning_rate": 1.740300126503993e-05,
      "loss": 0.6312,
      "step": 4215
    },
    {
      "epoch": 0.533535813718046,
      "grad_norm": 1.010882019996643,
      "learning_rate": 1.740179634354994e-05,
      "loss": 0.6174,
      "step": 4216
    },
    {
      "epoch": 0.5336623639584915,
      "grad_norm": 0.9999744296073914,
      "learning_rate": 1.7400591184335938e-05,
      "loss": 0.6579,
      "step": 4217
    },
    {
      "epoch": 0.5337889141989369,
      "grad_norm": 1.2087819576263428,
      "learning_rate": 1.7399385787436634e-05,
      "loss": 0.6918,
      "step": 4218
    },
    {
      "epoch": 0.5339154644393824,
      "grad_norm": 0.9872794151306152,
      "learning_rate": 1.7398180152890742e-05,
      "loss": 0.6028,
      "step": 4219
    },
    {
      "epoch": 0.5340420146798279,
      "grad_norm": 1.0556533336639404,
      "learning_rate": 1.7396974280736983e-05,
      "loss": 0.673,
      "step": 4220
    },
    {
      "epoch": 0.5341685649202733,
      "grad_norm": 1.0516022443771362,
      "learning_rate": 1.739576817101409e-05,
      "loss": 0.6264,
      "step": 4221
    },
    {
      "epoch": 0.5342951151607188,
      "grad_norm": 1.008940577507019,
      "learning_rate": 1.73945618237608e-05,
      "loss": 0.6393,
      "step": 4222
    },
    {
      "epoch": 0.5344216654011643,
      "grad_norm": 0.9825328588485718,
      "learning_rate": 1.7393355239015844e-05,
      "loss": 0.6421,
      "step": 4223
    },
    {
      "epoch": 0.5345482156416097,
      "grad_norm": 1.0351592302322388,
      "learning_rate": 1.739214841681799e-05,
      "loss": 0.6063,
      "step": 4224
    },
    {
      "epoch": 0.5346747658820552,
      "grad_norm": 0.9489290714263916,
      "learning_rate": 1.739094135720599e-05,
      "loss": 0.6095,
      "step": 4225
    },
    {
      "epoch": 0.5348013161225006,
      "grad_norm": 1.032175064086914,
      "learning_rate": 1.738973406021861e-05,
      "loss": 0.6773,
      "step": 4226
    },
    {
      "epoch": 0.5349278663629461,
      "grad_norm": 1.0176976919174194,
      "learning_rate": 1.7388526525894628e-05,
      "loss": 0.6148,
      "step": 4227
    },
    {
      "epoch": 0.5350544166033916,
      "grad_norm": 1.254349946975708,
      "learning_rate": 1.738731875427283e-05,
      "loss": 0.6709,
      "step": 4228
    },
    {
      "epoch": 0.535180966843837,
      "grad_norm": 1.019317865371704,
      "learning_rate": 1.7386110745392e-05,
      "loss": 0.6571,
      "step": 4229
    },
    {
      "epoch": 0.5353075170842825,
      "grad_norm": 1.0633893013000488,
      "learning_rate": 1.7384902499290932e-05,
      "loss": 0.5934,
      "step": 4230
    },
    {
      "epoch": 0.5354340673247279,
      "grad_norm": 1.016862154006958,
      "learning_rate": 1.7383694016008443e-05,
      "loss": 0.6119,
      "step": 4231
    },
    {
      "epoch": 0.5355606175651734,
      "grad_norm": 1.0499321222305298,
      "learning_rate": 1.738248529558334e-05,
      "loss": 0.631,
      "step": 4232
    },
    {
      "epoch": 0.5356871678056189,
      "grad_norm": 1.029733657836914,
      "learning_rate": 1.7381276338054437e-05,
      "loss": 0.6377,
      "step": 4233
    },
    {
      "epoch": 0.5358137180460643,
      "grad_norm": 0.9839189052581787,
      "learning_rate": 1.7380067143460574e-05,
      "loss": 0.6235,
      "step": 4234
    },
    {
      "epoch": 0.5359402682865098,
      "grad_norm": 1.0920039415359497,
      "learning_rate": 1.737885771184058e-05,
      "loss": 0.6135,
      "step": 4235
    },
    {
      "epoch": 0.5360668185269551,
      "grad_norm": 1.1447553634643555,
      "learning_rate": 1.7377648043233303e-05,
      "loss": 0.6644,
      "step": 4236
    },
    {
      "epoch": 0.5361933687674006,
      "grad_norm": 1.102704405784607,
      "learning_rate": 1.7376438137677585e-05,
      "loss": 0.6306,
      "step": 4237
    },
    {
      "epoch": 0.5363199190078461,
      "grad_norm": 0.9686986207962036,
      "learning_rate": 1.7375227995212295e-05,
      "loss": 0.6409,
      "step": 4238
    },
    {
      "epoch": 0.5364464692482915,
      "grad_norm": 1.2450947761535645,
      "learning_rate": 1.737401761587629e-05,
      "loss": 0.6701,
      "step": 4239
    },
    {
      "epoch": 0.536573019488737,
      "grad_norm": 1.0547044277191162,
      "learning_rate": 1.737280699970845e-05,
      "loss": 0.6636,
      "step": 4240
    },
    {
      "epoch": 0.5366995697291825,
      "grad_norm": 1.0385633707046509,
      "learning_rate": 1.737159614674766e-05,
      "loss": 0.5976,
      "step": 4241
    },
    {
      "epoch": 0.5368261199696279,
      "grad_norm": 1.2448227405548096,
      "learning_rate": 1.73703850570328e-05,
      "loss": 0.6359,
      "step": 4242
    },
    {
      "epoch": 0.5369526702100734,
      "grad_norm": 0.9976712465286255,
      "learning_rate": 1.7369173730602772e-05,
      "loss": 0.6231,
      "step": 4243
    },
    {
      "epoch": 0.5370792204505188,
      "grad_norm": 1.0194441080093384,
      "learning_rate": 1.7367962167496476e-05,
      "loss": 0.6058,
      "step": 4244
    },
    {
      "epoch": 0.5372057706909643,
      "grad_norm": 1.0495917797088623,
      "learning_rate": 1.7366750367752826e-05,
      "loss": 0.6321,
      "step": 4245
    },
    {
      "epoch": 0.5373323209314098,
      "grad_norm": 0.9572970867156982,
      "learning_rate": 1.7365538331410747e-05,
      "loss": 0.6035,
      "step": 4246
    },
    {
      "epoch": 0.5374588711718552,
      "grad_norm": 1.039796233177185,
      "learning_rate": 1.7364326058509157e-05,
      "loss": 0.6252,
      "step": 4247
    },
    {
      "epoch": 0.5375854214123007,
      "grad_norm": 0.997391939163208,
      "learning_rate": 1.7363113549086996e-05,
      "loss": 0.6232,
      "step": 4248
    },
    {
      "epoch": 0.5377119716527461,
      "grad_norm": 1.0066618919372559,
      "learning_rate": 1.7361900803183208e-05,
      "loss": 0.6173,
      "step": 4249
    },
    {
      "epoch": 0.5378385218931916,
      "grad_norm": 1.1529701948165894,
      "learning_rate": 1.7360687820836735e-05,
      "loss": 0.6377,
      "step": 4250
    },
    {
      "epoch": 0.5379650721336371,
      "grad_norm": 1.0151913166046143,
      "learning_rate": 1.7359474602086544e-05,
      "loss": 0.612,
      "step": 4251
    },
    {
      "epoch": 0.5380916223740825,
      "grad_norm": 1.3823285102844238,
      "learning_rate": 1.7358261146971593e-05,
      "loss": 0.6588,
      "step": 4252
    },
    {
      "epoch": 0.538218172614528,
      "grad_norm": 1.1742604970932007,
      "learning_rate": 1.7357047455530858e-05,
      "loss": 0.5914,
      "step": 4253
    },
    {
      "epoch": 0.5383447228549735,
      "grad_norm": 1.0306345224380493,
      "learning_rate": 1.7355833527803317e-05,
      "loss": 0.6377,
      "step": 4254
    },
    {
      "epoch": 0.5384712730954189,
      "grad_norm": 1.049048900604248,
      "learning_rate": 1.735461936382796e-05,
      "loss": 0.6329,
      "step": 4255
    },
    {
      "epoch": 0.5385978233358644,
      "grad_norm": 1.0615248680114746,
      "learning_rate": 1.735340496364378e-05,
      "loss": 0.6229,
      "step": 4256
    },
    {
      "epoch": 0.5387243735763098,
      "grad_norm": 1.0655988454818726,
      "learning_rate": 1.7352190327289782e-05,
      "loss": 0.6354,
      "step": 4257
    },
    {
      "epoch": 0.5388509238167553,
      "grad_norm": 0.9441892504692078,
      "learning_rate": 1.7350975454804978e-05,
      "loss": 0.6245,
      "step": 4258
    },
    {
      "epoch": 0.5389774740572008,
      "grad_norm": 1.0006119012832642,
      "learning_rate": 1.7349760346228385e-05,
      "loss": 0.6063,
      "step": 4259
    },
    {
      "epoch": 0.5391040242976461,
      "grad_norm": 1.1174741983413696,
      "learning_rate": 1.7348545001599023e-05,
      "loss": 0.6319,
      "step": 4260
    },
    {
      "epoch": 0.5392305745380916,
      "grad_norm": 1.1947648525238037,
      "learning_rate": 1.7347329420955935e-05,
      "loss": 0.6785,
      "step": 4261
    },
    {
      "epoch": 0.539357124778537,
      "grad_norm": 0.9751371145248413,
      "learning_rate": 1.7346113604338156e-05,
      "loss": 0.6206,
      "step": 4262
    },
    {
      "epoch": 0.5394836750189825,
      "grad_norm": 1.1937344074249268,
      "learning_rate": 1.7344897551784732e-05,
      "loss": 0.6755,
      "step": 4263
    },
    {
      "epoch": 0.539610225259428,
      "grad_norm": 1.070923089981079,
      "learning_rate": 1.7343681263334726e-05,
      "loss": 0.6334,
      "step": 4264
    },
    {
      "epoch": 0.5397367754998734,
      "grad_norm": 1.038317322731018,
      "learning_rate": 1.7342464739027197e-05,
      "loss": 0.6527,
      "step": 4265
    },
    {
      "epoch": 0.5398633257403189,
      "grad_norm": 1.0184755325317383,
      "learning_rate": 1.7341247978901216e-05,
      "loss": 0.6606,
      "step": 4266
    },
    {
      "epoch": 0.5399898759807643,
      "grad_norm": 0.9915881156921387,
      "learning_rate": 1.734003098299587e-05,
      "loss": 0.6265,
      "step": 4267
    },
    {
      "epoch": 0.5401164262212098,
      "grad_norm": 0.9702232480049133,
      "learning_rate": 1.7338813751350228e-05,
      "loss": 0.6063,
      "step": 4268
    },
    {
      "epoch": 0.5402429764616553,
      "grad_norm": 1.045401692390442,
      "learning_rate": 1.73375962840034e-05,
      "loss": 0.6305,
      "step": 4269
    },
    {
      "epoch": 0.5403695267021007,
      "grad_norm": 1.0007450580596924,
      "learning_rate": 1.7336378580994483e-05,
      "loss": 0.6589,
      "step": 4270
    },
    {
      "epoch": 0.5404960769425462,
      "grad_norm": 1.0007117986679077,
      "learning_rate": 1.7335160642362584e-05,
      "loss": 0.6671,
      "step": 4271
    },
    {
      "epoch": 0.5406226271829917,
      "grad_norm": 1.0208916664123535,
      "learning_rate": 1.7333942468146818e-05,
      "loss": 0.632,
      "step": 4272
    },
    {
      "epoch": 0.5407491774234371,
      "grad_norm": 0.9346381425857544,
      "learning_rate": 1.7332724058386316e-05,
      "loss": 0.6451,
      "step": 4273
    },
    {
      "epoch": 0.5408757276638826,
      "grad_norm": 1.006056547164917,
      "learning_rate": 1.7331505413120203e-05,
      "loss": 0.6341,
      "step": 4274
    },
    {
      "epoch": 0.541002277904328,
      "grad_norm": 1.0303527116775513,
      "learning_rate": 1.733028653238762e-05,
      "loss": 0.6378,
      "step": 4275
    },
    {
      "epoch": 0.5411288281447735,
      "grad_norm": 0.9717079401016235,
      "learning_rate": 1.7329067416227717e-05,
      "loss": 0.6291,
      "step": 4276
    },
    {
      "epoch": 0.541255378385219,
      "grad_norm": 1.0111148357391357,
      "learning_rate": 1.7327848064679648e-05,
      "loss": 0.6582,
      "step": 4277
    },
    {
      "epoch": 0.5413819286256644,
      "grad_norm": 0.942504346370697,
      "learning_rate": 1.732662847778257e-05,
      "loss": 0.6225,
      "step": 4278
    },
    {
      "epoch": 0.5415084788661099,
      "grad_norm": 0.9549508094787598,
      "learning_rate": 1.7325408655575655e-05,
      "loss": 0.6107,
      "step": 4279
    },
    {
      "epoch": 0.5416350291065553,
      "grad_norm": 1.0201317071914673,
      "learning_rate": 1.732418859809808e-05,
      "loss": 0.7017,
      "step": 4280
    },
    {
      "epoch": 0.5417615793470008,
      "grad_norm": 1.014092206954956,
      "learning_rate": 1.732296830538903e-05,
      "loss": 0.6245,
      "step": 4281
    },
    {
      "epoch": 0.5418881295874463,
      "grad_norm": 1.0299259424209595,
      "learning_rate": 1.73217477774877e-05,
      "loss": 0.6583,
      "step": 4282
    },
    {
      "epoch": 0.5420146798278916,
      "grad_norm": 0.929051399230957,
      "learning_rate": 1.7320527014433287e-05,
      "loss": 0.5889,
      "step": 4283
    },
    {
      "epoch": 0.5421412300683371,
      "grad_norm": 1.0093138217926025,
      "learning_rate": 1.7319306016265e-05,
      "loss": 0.6184,
      "step": 4284
    },
    {
      "epoch": 0.5422677803087825,
      "grad_norm": 1.0220023393630981,
      "learning_rate": 1.7318084783022053e-05,
      "loss": 0.6586,
      "step": 4285
    },
    {
      "epoch": 0.542394330549228,
      "grad_norm": 0.9981477856636047,
      "learning_rate": 1.7316863314743666e-05,
      "loss": 0.6604,
      "step": 4286
    },
    {
      "epoch": 0.5425208807896735,
      "grad_norm": 1.080320954322815,
      "learning_rate": 1.7315641611469073e-05,
      "loss": 0.659,
      "step": 4287
    },
    {
      "epoch": 0.5426474310301189,
      "grad_norm": 1.0447622537612915,
      "learning_rate": 1.7314419673237507e-05,
      "loss": 0.6419,
      "step": 4288
    },
    {
      "epoch": 0.5427739812705644,
      "grad_norm": 1.1379972696304321,
      "learning_rate": 1.731319750008822e-05,
      "loss": 0.6664,
      "step": 4289
    },
    {
      "epoch": 0.5429005315110099,
      "grad_norm": 0.9462612271308899,
      "learning_rate": 1.7311975092060458e-05,
      "loss": 0.637,
      "step": 4290
    },
    {
      "epoch": 0.5430270817514553,
      "grad_norm": 1.0280081033706665,
      "learning_rate": 1.7310752449193484e-05,
      "loss": 0.6279,
      "step": 4291
    },
    {
      "epoch": 0.5431536319919008,
      "grad_norm": 0.9899301528930664,
      "learning_rate": 1.730952957152656e-05,
      "loss": 0.6601,
      "step": 4292
    },
    {
      "epoch": 0.5432801822323462,
      "grad_norm": 1.0561381578445435,
      "learning_rate": 1.7308306459098974e-05,
      "loss": 0.6594,
      "step": 4293
    },
    {
      "epoch": 0.5434067324727917,
      "grad_norm": 0.9528133869171143,
      "learning_rate": 1.730708311195e-05,
      "loss": 0.6189,
      "step": 4294
    },
    {
      "epoch": 0.5435332827132372,
      "grad_norm": 1.0806382894515991,
      "learning_rate": 1.7305859530118925e-05,
      "loss": 0.6679,
      "step": 4295
    },
    {
      "epoch": 0.5436598329536826,
      "grad_norm": 1.0443060398101807,
      "learning_rate": 1.7304635713645054e-05,
      "loss": 0.6543,
      "step": 4296
    },
    {
      "epoch": 0.5437863831941281,
      "grad_norm": 0.9775519371032715,
      "learning_rate": 1.7303411662567692e-05,
      "loss": 0.5936,
      "step": 4297
    },
    {
      "epoch": 0.5439129334345735,
      "grad_norm": 1.0442317724227905,
      "learning_rate": 1.7302187376926148e-05,
      "loss": 0.6261,
      "step": 4298
    },
    {
      "epoch": 0.544039483675019,
      "grad_norm": 1.035428762435913,
      "learning_rate": 1.7300962856759744e-05,
      "loss": 0.6366,
      "step": 4299
    },
    {
      "epoch": 0.5441660339154645,
      "grad_norm": 1.0607532262802124,
      "learning_rate": 1.7299738102107806e-05,
      "loss": 0.6448,
      "step": 4300
    },
    {
      "epoch": 0.5442925841559099,
      "grad_norm": 1.041333794593811,
      "learning_rate": 1.7298513113009675e-05,
      "loss": 0.6394,
      "step": 4301
    },
    {
      "epoch": 0.5444191343963554,
      "grad_norm": 1.0173429250717163,
      "learning_rate": 1.729728788950469e-05,
      "loss": 0.6348,
      "step": 4302
    },
    {
      "epoch": 0.5445456846368009,
      "grad_norm": 1.1349537372589111,
      "learning_rate": 1.7296062431632203e-05,
      "loss": 0.6791,
      "step": 4303
    },
    {
      "epoch": 0.5446722348772463,
      "grad_norm": 1.0249390602111816,
      "learning_rate": 1.729483673943157e-05,
      "loss": 0.6274,
      "step": 4304
    },
    {
      "epoch": 0.5447987851176918,
      "grad_norm": 0.9776830673217773,
      "learning_rate": 1.729361081294216e-05,
      "loss": 0.6412,
      "step": 4305
    },
    {
      "epoch": 0.5449253353581371,
      "grad_norm": 1.0973002910614014,
      "learning_rate": 1.7292384652203342e-05,
      "loss": 0.6168,
      "step": 4306
    },
    {
      "epoch": 0.5450518855985826,
      "grad_norm": 1.0782252550125122,
      "learning_rate": 1.72911582572545e-05,
      "loss": 0.649,
      "step": 4307
    },
    {
      "epoch": 0.5451784358390281,
      "grad_norm": 1.0829696655273438,
      "learning_rate": 1.7289931628135022e-05,
      "loss": 0.6402,
      "step": 4308
    },
    {
      "epoch": 0.5453049860794735,
      "grad_norm": 1.012529730796814,
      "learning_rate": 1.7288704764884304e-05,
      "loss": 0.638,
      "step": 4309
    },
    {
      "epoch": 0.545431536319919,
      "grad_norm": 1.0851984024047852,
      "learning_rate": 1.7287477667541747e-05,
      "loss": 0.6455,
      "step": 4310
    },
    {
      "epoch": 0.5455580865603644,
      "grad_norm": 1.0075401067733765,
      "learning_rate": 1.728625033614676e-05,
      "loss": 0.652,
      "step": 4311
    },
    {
      "epoch": 0.5456846368008099,
      "grad_norm": 0.9303786754608154,
      "learning_rate": 1.7285022770738767e-05,
      "loss": 0.5853,
      "step": 4312
    },
    {
      "epoch": 0.5458111870412554,
      "grad_norm": 0.9662461280822754,
      "learning_rate": 1.7283794971357193e-05,
      "loss": 0.6086,
      "step": 4313
    },
    {
      "epoch": 0.5459377372817008,
      "grad_norm": 1.0990700721740723,
      "learning_rate": 1.7282566938041467e-05,
      "loss": 0.6377,
      "step": 4314
    },
    {
      "epoch": 0.5460642875221463,
      "grad_norm": 1.0279593467712402,
      "learning_rate": 1.7281338670831033e-05,
      "loss": 0.6064,
      "step": 4315
    },
    {
      "epoch": 0.5461908377625917,
      "grad_norm": 1.0001680850982666,
      "learning_rate": 1.728011016976534e-05,
      "loss": 0.615,
      "step": 4316
    },
    {
      "epoch": 0.5463173880030372,
      "grad_norm": 1.0127393007278442,
      "learning_rate": 1.7278881434883838e-05,
      "loss": 0.6349,
      "step": 4317
    },
    {
      "epoch": 0.5464439382434827,
      "grad_norm": 0.9560511708259583,
      "learning_rate": 1.7277652466225997e-05,
      "loss": 0.6056,
      "step": 4318
    },
    {
      "epoch": 0.5465704884839281,
      "grad_norm": 0.9794509410858154,
      "learning_rate": 1.7276423263831287e-05,
      "loss": 0.6364,
      "step": 4319
    },
    {
      "epoch": 0.5466970387243736,
      "grad_norm": 0.9459718465805054,
      "learning_rate": 1.7275193827739184e-05,
      "loss": 0.6003,
      "step": 4320
    },
    {
      "epoch": 0.5468235889648191,
      "grad_norm": 0.9729351997375488,
      "learning_rate": 1.7273964157989174e-05,
      "loss": 0.6664,
      "step": 4321
    },
    {
      "epoch": 0.5469501392052645,
      "grad_norm": 0.9947158694267273,
      "learning_rate": 1.7272734254620755e-05,
      "loss": 0.6564,
      "step": 4322
    },
    {
      "epoch": 0.54707668944571,
      "grad_norm": 1.0118085145950317,
      "learning_rate": 1.7271504117673422e-05,
      "loss": 0.5833,
      "step": 4323
    },
    {
      "epoch": 0.5472032396861554,
      "grad_norm": 0.9825928807258606,
      "learning_rate": 1.7270273747186685e-05,
      "loss": 0.6402,
      "step": 4324
    },
    {
      "epoch": 0.5473297899266009,
      "grad_norm": 1.0456112623214722,
      "learning_rate": 1.726904314320006e-05,
      "loss": 0.6303,
      "step": 4325
    },
    {
      "epoch": 0.5474563401670464,
      "grad_norm": 1.0163284540176392,
      "learning_rate": 1.726781230575307e-05,
      "loss": 0.629,
      "step": 4326
    },
    {
      "epoch": 0.5475828904074918,
      "grad_norm": 0.9309964776039124,
      "learning_rate": 1.7266581234885248e-05,
      "loss": 0.5792,
      "step": 4327
    },
    {
      "epoch": 0.5477094406479373,
      "grad_norm": 1.117426872253418,
      "learning_rate": 1.7265349930636132e-05,
      "loss": 0.6921,
      "step": 4328
    },
    {
      "epoch": 0.5478359908883826,
      "grad_norm": 0.9928696155548096,
      "learning_rate": 1.7264118393045265e-05,
      "loss": 0.6331,
      "step": 4329
    },
    {
      "epoch": 0.5479625411288281,
      "grad_norm": 0.9325777292251587,
      "learning_rate": 1.7262886622152207e-05,
      "loss": 0.5733,
      "step": 4330
    },
    {
      "epoch": 0.5480890913692736,
      "grad_norm": 1.013215184211731,
      "learning_rate": 1.726165461799651e-05,
      "loss": 0.6725,
      "step": 4331
    },
    {
      "epoch": 0.548215641609719,
      "grad_norm": 0.9439978003501892,
      "learning_rate": 1.726042238061775e-05,
      "loss": 0.6264,
      "step": 4332
    },
    {
      "epoch": 0.5483421918501645,
      "grad_norm": 1.009045124053955,
      "learning_rate": 1.7259189910055497e-05,
      "loss": 0.6204,
      "step": 4333
    },
    {
      "epoch": 0.5484687420906099,
      "grad_norm": 1.075164794921875,
      "learning_rate": 1.7257957206349337e-05,
      "loss": 0.6597,
      "step": 4334
    },
    {
      "epoch": 0.5485952923310554,
      "grad_norm": 1.052014946937561,
      "learning_rate": 1.725672426953886e-05,
      "loss": 0.6534,
      "step": 4335
    },
    {
      "epoch": 0.5487218425715009,
      "grad_norm": 0.9876077771186829,
      "learning_rate": 1.725549109966367e-05,
      "loss": 0.6539,
      "step": 4336
    },
    {
      "epoch": 0.5488483928119463,
      "grad_norm": 0.9819099307060242,
      "learning_rate": 1.725425769676336e-05,
      "loss": 0.6112,
      "step": 4337
    },
    {
      "epoch": 0.5489749430523918,
      "grad_norm": 1.0281052589416504,
      "learning_rate": 1.7253024060877554e-05,
      "loss": 0.6591,
      "step": 4338
    },
    {
      "epoch": 0.5491014932928373,
      "grad_norm": 0.9508469700813293,
      "learning_rate": 1.725179019204587e-05,
      "loss": 0.6138,
      "step": 4339
    },
    {
      "epoch": 0.5492280435332827,
      "grad_norm": 1.0973596572875977,
      "learning_rate": 1.725055609030794e-05,
      "loss": 0.7166,
      "step": 4340
    },
    {
      "epoch": 0.5493545937737282,
      "grad_norm": 1.0051034688949585,
      "learning_rate": 1.7249321755703394e-05,
      "loss": 0.6412,
      "step": 4341
    },
    {
      "epoch": 0.5494811440141736,
      "grad_norm": 0.9728878736495972,
      "learning_rate": 1.7248087188271873e-05,
      "loss": 0.6348,
      "step": 4342
    },
    {
      "epoch": 0.5496076942546191,
      "grad_norm": 1.0005637407302856,
      "learning_rate": 1.724685238805304e-05,
      "loss": 0.6453,
      "step": 4343
    },
    {
      "epoch": 0.5497342444950646,
      "grad_norm": 1.0050474405288696,
      "learning_rate": 1.7245617355086538e-05,
      "loss": 0.6791,
      "step": 4344
    },
    {
      "epoch": 0.54986079473551,
      "grad_norm": 0.9850348830223083,
      "learning_rate": 1.7244382089412037e-05,
      "loss": 0.6328,
      "step": 4345
    },
    {
      "epoch": 0.5499873449759555,
      "grad_norm": 0.9690372943878174,
      "learning_rate": 1.7243146591069216e-05,
      "loss": 0.6186,
      "step": 4346
    },
    {
      "epoch": 0.5501138952164009,
      "grad_norm": 0.9676150679588318,
      "learning_rate": 1.7241910860097755e-05,
      "loss": 0.6286,
      "step": 4347
    },
    {
      "epoch": 0.5502404454568464,
      "grad_norm": 0.9730240702629089,
      "learning_rate": 1.724067489653734e-05,
      "loss": 0.6125,
      "step": 4348
    },
    {
      "epoch": 0.5503669956972919,
      "grad_norm": 1.035199761390686,
      "learning_rate": 1.723943870042766e-05,
      "loss": 0.6732,
      "step": 4349
    },
    {
      "epoch": 0.5504935459377372,
      "grad_norm": 0.9495027661323547,
      "learning_rate": 1.7238202271808427e-05,
      "loss": 0.6248,
      "step": 4350
    },
    {
      "epoch": 0.5506200961781827,
      "grad_norm": 1.1011266708374023,
      "learning_rate": 1.7236965610719346e-05,
      "loss": 0.6775,
      "step": 4351
    },
    {
      "epoch": 0.5507466464186282,
      "grad_norm": 1.0237390995025635,
      "learning_rate": 1.7235728717200145e-05,
      "loss": 0.6671,
      "step": 4352
    },
    {
      "epoch": 0.5508731966590736,
      "grad_norm": 1.027785062789917,
      "learning_rate": 1.7234491591290534e-05,
      "loss": 0.646,
      "step": 4353
    },
    {
      "epoch": 0.5509997468995191,
      "grad_norm": 0.9674209952354431,
      "learning_rate": 1.7233254233030257e-05,
      "loss": 0.6526,
      "step": 4354
    },
    {
      "epoch": 0.5511262971399645,
      "grad_norm": 1.0223479270935059,
      "learning_rate": 1.7232016642459045e-05,
      "loss": 0.636,
      "step": 4355
    },
    {
      "epoch": 0.55125284738041,
      "grad_norm": 0.9786264300346375,
      "learning_rate": 1.723077881961666e-05,
      "loss": 0.634,
      "step": 4356
    },
    {
      "epoch": 0.5513793976208555,
      "grad_norm": 1.0285663604736328,
      "learning_rate": 1.7229540764542846e-05,
      "loss": 0.6827,
      "step": 4357
    },
    {
      "epoch": 0.5515059478613009,
      "grad_norm": 1.0006968975067139,
      "learning_rate": 1.7228302477277365e-05,
      "loss": 0.7147,
      "step": 4358
    },
    {
      "epoch": 0.5516324981017464,
      "grad_norm": 1.1044254302978516,
      "learning_rate": 1.7227063957859994e-05,
      "loss": 0.6548,
      "step": 4359
    },
    {
      "epoch": 0.5517590483421918,
      "grad_norm": 1.0224539041519165,
      "learning_rate": 1.7225825206330506e-05,
      "loss": 0.6605,
      "step": 4360
    },
    {
      "epoch": 0.5518855985826373,
      "grad_norm": 1.8790887594223022,
      "learning_rate": 1.722458622272869e-05,
      "loss": 0.7239,
      "step": 4361
    },
    {
      "epoch": 0.5520121488230828,
      "grad_norm": 1.080468773841858,
      "learning_rate": 1.722334700709433e-05,
      "loss": 0.6306,
      "step": 4362
    },
    {
      "epoch": 0.5521386990635282,
      "grad_norm": 1.1019065380096436,
      "learning_rate": 1.7222107559467235e-05,
      "loss": 0.6353,
      "step": 4363
    },
    {
      "epoch": 0.5522652493039737,
      "grad_norm": 1.1491663455963135,
      "learning_rate": 1.722086787988721e-05,
      "loss": 0.6707,
      "step": 4364
    },
    {
      "epoch": 0.5523917995444191,
      "grad_norm": 1.0888620615005493,
      "learning_rate": 1.721962796839407e-05,
      "loss": 0.6385,
      "step": 4365
    },
    {
      "epoch": 0.5525183497848646,
      "grad_norm": 1.1717225313186646,
      "learning_rate": 1.7218387825027637e-05,
      "loss": 0.6788,
      "step": 4366
    },
    {
      "epoch": 0.5526449000253101,
      "grad_norm": 0.9537553191184998,
      "learning_rate": 1.721714744982774e-05,
      "loss": 0.6123,
      "step": 4367
    },
    {
      "epoch": 0.5527714502657555,
      "grad_norm": 1.1662622690200806,
      "learning_rate": 1.7215906842834214e-05,
      "loss": 0.6034,
      "step": 4368
    },
    {
      "epoch": 0.552898000506201,
      "grad_norm": 0.9631742835044861,
      "learning_rate": 1.7214666004086907e-05,
      "loss": 0.6346,
      "step": 4369
    },
    {
      "epoch": 0.5530245507466465,
      "grad_norm": 1.0683790445327759,
      "learning_rate": 1.721342493362567e-05,
      "loss": 0.5821,
      "step": 4370
    },
    {
      "epoch": 0.5531511009870919,
      "grad_norm": 1.020707368850708,
      "learning_rate": 1.721218363149036e-05,
      "loss": 0.6121,
      "step": 4371
    },
    {
      "epoch": 0.5532776512275374,
      "grad_norm": 1.1349570751190186,
      "learning_rate": 1.7210942097720856e-05,
      "loss": 0.6821,
      "step": 4372
    },
    {
      "epoch": 0.5534042014679827,
      "grad_norm": 1.048203945159912,
      "learning_rate": 1.7209700332357015e-05,
      "loss": 0.6501,
      "step": 4373
    },
    {
      "epoch": 0.5535307517084282,
      "grad_norm": 1.031558632850647,
      "learning_rate": 1.720845833543873e-05,
      "loss": 0.5944,
      "step": 4374
    },
    {
      "epoch": 0.5536573019488737,
      "grad_norm": 1.0765753984451294,
      "learning_rate": 1.7207216107005886e-05,
      "loss": 0.671,
      "step": 4375
    },
    {
      "epoch": 0.5537838521893191,
      "grad_norm": 0.9966520071029663,
      "learning_rate": 1.7205973647098383e-05,
      "loss": 0.6007,
      "step": 4376
    },
    {
      "epoch": 0.5539104024297646,
      "grad_norm": 1.1757100820541382,
      "learning_rate": 1.720473095575612e-05,
      "loss": 0.6504,
      "step": 4377
    },
    {
      "epoch": 0.55403695267021,
      "grad_norm": 1.059820294380188,
      "learning_rate": 1.7203488033019017e-05,
      "loss": 0.641,
      "step": 4378
    },
    {
      "epoch": 0.5541635029106555,
      "grad_norm": 1.189070701599121,
      "learning_rate": 1.7202244878926983e-05,
      "loss": 0.5742,
      "step": 4379
    },
    {
      "epoch": 0.554290053151101,
      "grad_norm": 0.9728901386260986,
      "learning_rate": 1.720100149351995e-05,
      "loss": 0.6111,
      "step": 4380
    },
    {
      "epoch": 0.5544166033915464,
      "grad_norm": 1.226865530014038,
      "learning_rate": 1.7199757876837855e-05,
      "loss": 0.6574,
      "step": 4381
    },
    {
      "epoch": 0.5545431536319919,
      "grad_norm": 1.1676024198532104,
      "learning_rate": 1.7198514028920632e-05,
      "loss": 0.6505,
      "step": 4382
    },
    {
      "epoch": 0.5546697038724373,
      "grad_norm": 0.9869944453239441,
      "learning_rate": 1.7197269949808236e-05,
      "loss": 0.5751,
      "step": 4383
    },
    {
      "epoch": 0.5547962541128828,
      "grad_norm": 1.086470603942871,
      "learning_rate": 1.719602563954062e-05,
      "loss": 0.6134,
      "step": 4384
    },
    {
      "epoch": 0.5549228043533283,
      "grad_norm": 0.9762349128723145,
      "learning_rate": 1.7194781098157747e-05,
      "loss": 0.6441,
      "step": 4385
    },
    {
      "epoch": 0.5550493545937737,
      "grad_norm": 1.0464707612991333,
      "learning_rate": 1.719353632569959e-05,
      "loss": 0.6923,
      "step": 4386
    },
    {
      "epoch": 0.5551759048342192,
      "grad_norm": 1.0478343963623047,
      "learning_rate": 1.7192291322206126e-05,
      "loss": 0.6419,
      "step": 4387
    },
    {
      "epoch": 0.5553024550746647,
      "grad_norm": 1.0286532640457153,
      "learning_rate": 1.7191046087717344e-05,
      "loss": 0.6416,
      "step": 4388
    },
    {
      "epoch": 0.5554290053151101,
      "grad_norm": 0.9784203767776489,
      "learning_rate": 1.718980062227323e-05,
      "loss": 0.6398,
      "step": 4389
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 1.1716669797897339,
      "learning_rate": 1.7188554925913793e-05,
      "loss": 0.6627,
      "step": 4390
    },
    {
      "epoch": 0.555682105796001,
      "grad_norm": 1.0678017139434814,
      "learning_rate": 1.7187308998679034e-05,
      "loss": 0.626,
      "step": 4391
    },
    {
      "epoch": 0.5558086560364465,
      "grad_norm": 1.1025689840316772,
      "learning_rate": 1.7186062840608976e-05,
      "loss": 0.6296,
      "step": 4392
    },
    {
      "epoch": 0.555935206276892,
      "grad_norm": 1.13590407371521,
      "learning_rate": 1.7184816451743635e-05,
      "loss": 0.6398,
      "step": 4393
    },
    {
      "epoch": 0.5560617565173374,
      "grad_norm": 1.0162352323532104,
      "learning_rate": 1.7183569832123048e-05,
      "loss": 0.6731,
      "step": 4394
    },
    {
      "epoch": 0.5561883067577829,
      "grad_norm": 0.9907544255256653,
      "learning_rate": 1.7182322981787246e-05,
      "loss": 0.6504,
      "step": 4395
    },
    {
      "epoch": 0.5563148569982282,
      "grad_norm": 1.026537537574768,
      "learning_rate": 1.718107590077628e-05,
      "loss": 0.6152,
      "step": 4396
    },
    {
      "epoch": 0.5564414072386737,
      "grad_norm": 1.0087366104125977,
      "learning_rate": 1.7179828589130196e-05,
      "loss": 0.6513,
      "step": 4397
    },
    {
      "epoch": 0.5565679574791192,
      "grad_norm": 1.0135691165924072,
      "learning_rate": 1.7178581046889057e-05,
      "loss": 0.6672,
      "step": 4398
    },
    {
      "epoch": 0.5566945077195646,
      "grad_norm": 0.971059262752533,
      "learning_rate": 1.717733327409294e-05,
      "loss": 0.6317,
      "step": 4399
    },
    {
      "epoch": 0.5568210579600101,
      "grad_norm": 0.9351518750190735,
      "learning_rate": 1.7176085270781904e-05,
      "loss": 0.6162,
      "step": 4400
    },
    {
      "epoch": 0.5569476082004556,
      "grad_norm": 0.9957442283630371,
      "learning_rate": 1.717483703699604e-05,
      "loss": 0.6539,
      "step": 4401
    },
    {
      "epoch": 0.557074158440901,
      "grad_norm": 1.0595225095748901,
      "learning_rate": 1.7173588572775437e-05,
      "loss": 0.6347,
      "step": 4402
    },
    {
      "epoch": 0.5572007086813465,
      "grad_norm": 1.2181692123413086,
      "learning_rate": 1.7172339878160187e-05,
      "loss": 0.6441,
      "step": 4403
    },
    {
      "epoch": 0.5573272589217919,
      "grad_norm": 1.1953729391098022,
      "learning_rate": 1.7171090953190402e-05,
      "loss": 0.6566,
      "step": 4404
    },
    {
      "epoch": 0.5574538091622374,
      "grad_norm": 0.9494336247444153,
      "learning_rate": 1.716984179790619e-05,
      "loss": 0.6147,
      "step": 4405
    },
    {
      "epoch": 0.5575803594026829,
      "grad_norm": 1.1100208759307861,
      "learning_rate": 1.716859241234767e-05,
      "loss": 0.5561,
      "step": 4406
    },
    {
      "epoch": 0.5577069096431283,
      "grad_norm": 1.1267021894454956,
      "learning_rate": 1.716734279655497e-05,
      "loss": 0.6422,
      "step": 4407
    },
    {
      "epoch": 0.5578334598835738,
      "grad_norm": 1.039607286453247,
      "learning_rate": 1.716609295056822e-05,
      "loss": 0.6327,
      "step": 4408
    },
    {
      "epoch": 0.5579600101240192,
      "grad_norm": 1.0122809410095215,
      "learning_rate": 1.716484287442757e-05,
      "loss": 0.6185,
      "step": 4409
    },
    {
      "epoch": 0.5580865603644647,
      "grad_norm": 1.0677322149276733,
      "learning_rate": 1.7163592568173154e-05,
      "loss": 0.64,
      "step": 4410
    },
    {
      "epoch": 0.5582131106049102,
      "grad_norm": 0.9799858927726746,
      "learning_rate": 1.7162342031845144e-05,
      "loss": 0.6539,
      "step": 4411
    },
    {
      "epoch": 0.5583396608453556,
      "grad_norm": 0.972794234752655,
      "learning_rate": 1.7161091265483696e-05,
      "loss": 0.6384,
      "step": 4412
    },
    {
      "epoch": 0.5584662110858011,
      "grad_norm": 1.3091113567352295,
      "learning_rate": 1.7159840269128982e-05,
      "loss": 0.6626,
      "step": 4413
    },
    {
      "epoch": 0.5585927613262465,
      "grad_norm": 0.9989495277404785,
      "learning_rate": 1.715858904282118e-05,
      "loss": 0.6035,
      "step": 4414
    },
    {
      "epoch": 0.558719311566692,
      "grad_norm": 1.0493096113204956,
      "learning_rate": 1.7157337586600474e-05,
      "loss": 0.6714,
      "step": 4415
    },
    {
      "epoch": 0.5588458618071375,
      "grad_norm": 0.9908995628356934,
      "learning_rate": 1.715608590050706e-05,
      "loss": 0.6643,
      "step": 4416
    },
    {
      "epoch": 0.5589724120475829,
      "grad_norm": 1.1452399492263794,
      "learning_rate": 1.715483398458114e-05,
      "loss": 0.6585,
      "step": 4417
    },
    {
      "epoch": 0.5590989622880284,
      "grad_norm": 1.002698540687561,
      "learning_rate": 1.7153581838862917e-05,
      "loss": 0.579,
      "step": 4418
    },
    {
      "epoch": 0.5592255125284739,
      "grad_norm": 1.0588665008544922,
      "learning_rate": 1.715232946339261e-05,
      "loss": 0.6187,
      "step": 4419
    },
    {
      "epoch": 0.5593520627689192,
      "grad_norm": 0.9906606078147888,
      "learning_rate": 1.7151076858210438e-05,
      "loss": 0.6262,
      "step": 4420
    },
    {
      "epoch": 0.5594786130093647,
      "grad_norm": 1.0087342262268066,
      "learning_rate": 1.7149824023356637e-05,
      "loss": 0.6428,
      "step": 4421
    },
    {
      "epoch": 0.5596051632498101,
      "grad_norm": 1.046163558959961,
      "learning_rate": 1.7148570958871435e-05,
      "loss": 0.6155,
      "step": 4422
    },
    {
      "epoch": 0.5597317134902556,
      "grad_norm": 0.9368391036987305,
      "learning_rate": 1.714731766479509e-05,
      "loss": 0.6244,
      "step": 4423
    },
    {
      "epoch": 0.5598582637307011,
      "grad_norm": 1.0520635843276978,
      "learning_rate": 1.714606414116784e-05,
      "loss": 0.6079,
      "step": 4424
    },
    {
      "epoch": 0.5599848139711465,
      "grad_norm": 1.0234299898147583,
      "learning_rate": 1.7144810388029953e-05,
      "loss": 0.6285,
      "step": 4425
    },
    {
      "epoch": 0.560111364211592,
      "grad_norm": 1.0192737579345703,
      "learning_rate": 1.7143556405421694e-05,
      "loss": 0.6253,
      "step": 4426
    },
    {
      "epoch": 0.5602379144520374,
      "grad_norm": 1.0066919326782227,
      "learning_rate": 1.7142302193383337e-05,
      "loss": 0.5623,
      "step": 4427
    },
    {
      "epoch": 0.5603644646924829,
      "grad_norm": 1.0715866088867188,
      "learning_rate": 1.7141047751955165e-05,
      "loss": 0.5977,
      "step": 4428
    },
    {
      "epoch": 0.5604910149329284,
      "grad_norm": 0.9749395251274109,
      "learning_rate": 1.7139793081177465e-05,
      "loss": 0.6395,
      "step": 4429
    },
    {
      "epoch": 0.5606175651733738,
      "grad_norm": 0.9352781176567078,
      "learning_rate": 1.7138538181090534e-05,
      "loss": 0.5898,
      "step": 4430
    },
    {
      "epoch": 0.5607441154138193,
      "grad_norm": 0.9846215844154358,
      "learning_rate": 1.7137283051734678e-05,
      "loss": 0.6163,
      "step": 4431
    },
    {
      "epoch": 0.5608706656542647,
      "grad_norm": 1.078647494316101,
      "learning_rate": 1.7136027693150202e-05,
      "loss": 0.6701,
      "step": 4432
    },
    {
      "epoch": 0.5609972158947102,
      "grad_norm": 0.9772288203239441,
      "learning_rate": 1.7134772105377432e-05,
      "loss": 0.6441,
      "step": 4433
    },
    {
      "epoch": 0.5611237661351557,
      "grad_norm": 0.9968873262405396,
      "learning_rate": 1.7133516288456686e-05,
      "loss": 0.6378,
      "step": 4434
    },
    {
      "epoch": 0.5612503163756011,
      "grad_norm": 1.1313047409057617,
      "learning_rate": 1.7132260242428303e-05,
      "loss": 0.6317,
      "step": 4435
    },
    {
      "epoch": 0.5613768666160466,
      "grad_norm": 1.0410264730453491,
      "learning_rate": 1.7131003967332623e-05,
      "loss": 0.6464,
      "step": 4436
    },
    {
      "epoch": 0.5615034168564921,
      "grad_norm": 1.0075217485427856,
      "learning_rate": 1.7129747463209994e-05,
      "loss": 0.6758,
      "step": 4437
    },
    {
      "epoch": 0.5616299670969375,
      "grad_norm": 1.025356411933899,
      "learning_rate": 1.7128490730100767e-05,
      "loss": 0.6311,
      "step": 4438
    },
    {
      "epoch": 0.561756517337383,
      "grad_norm": 1.046890377998352,
      "learning_rate": 1.7127233768045307e-05,
      "loss": 0.6134,
      "step": 4439
    },
    {
      "epoch": 0.5618830675778284,
      "grad_norm": 1.002550482749939,
      "learning_rate": 1.7125976577083987e-05,
      "loss": 0.5875,
      "step": 4440
    },
    {
      "epoch": 0.5620096178182739,
      "grad_norm": 1.0215506553649902,
      "learning_rate": 1.712471915725718e-05,
      "loss": 0.6229,
      "step": 4441
    },
    {
      "epoch": 0.5621361680587194,
      "grad_norm": 1.0398216247558594,
      "learning_rate": 1.7123461508605276e-05,
      "loss": 0.6089,
      "step": 4442
    },
    {
      "epoch": 0.5622627182991647,
      "grad_norm": 0.980120837688446,
      "learning_rate": 1.712220363116866e-05,
      "loss": 0.6593,
      "step": 4443
    },
    {
      "epoch": 0.5623892685396102,
      "grad_norm": 0.9477469325065613,
      "learning_rate": 1.7120945524987735e-05,
      "loss": 0.6504,
      "step": 4444
    },
    {
      "epoch": 0.5625158187800556,
      "grad_norm": 1.0421550273895264,
      "learning_rate": 1.7119687190102907e-05,
      "loss": 0.6385,
      "step": 4445
    },
    {
      "epoch": 0.5626423690205011,
      "grad_norm": 1.0257822275161743,
      "learning_rate": 1.711842862655459e-05,
      "loss": 0.6817,
      "step": 4446
    },
    {
      "epoch": 0.5627689192609466,
      "grad_norm": 1.0361385345458984,
      "learning_rate": 1.7117169834383206e-05,
      "loss": 0.646,
      "step": 4447
    },
    {
      "epoch": 0.562895469501392,
      "grad_norm": 1.0193579196929932,
      "learning_rate": 1.7115910813629187e-05,
      "loss": 0.6658,
      "step": 4448
    },
    {
      "epoch": 0.5630220197418375,
      "grad_norm": 1.1174753904342651,
      "learning_rate": 1.7114651564332963e-05,
      "loss": 0.6313,
      "step": 4449
    },
    {
      "epoch": 0.563148569982283,
      "grad_norm": 1.1309151649475098,
      "learning_rate": 1.711339208653498e-05,
      "loss": 0.6412,
      "step": 4450
    },
    {
      "epoch": 0.5632751202227284,
      "grad_norm": 1.1628214120864868,
      "learning_rate": 1.7112132380275688e-05,
      "loss": 0.6636,
      "step": 4451
    },
    {
      "epoch": 0.5634016704631739,
      "grad_norm": 1.16172194480896,
      "learning_rate": 1.7110872445595545e-05,
      "loss": 0.6688,
      "step": 4452
    },
    {
      "epoch": 0.5635282207036193,
      "grad_norm": 0.9950689673423767,
      "learning_rate": 1.7109612282535018e-05,
      "loss": 0.6348,
      "step": 4453
    },
    {
      "epoch": 0.5636547709440648,
      "grad_norm": 1.2346309423446655,
      "learning_rate": 1.710835189113458e-05,
      "loss": 0.6529,
      "step": 4454
    },
    {
      "epoch": 0.5637813211845103,
      "grad_norm": 0.9963491559028625,
      "learning_rate": 1.710709127143471e-05,
      "loss": 0.642,
      "step": 4455
    },
    {
      "epoch": 0.5639078714249557,
      "grad_norm": 0.9569481611251831,
      "learning_rate": 1.7105830423475894e-05,
      "loss": 0.6327,
      "step": 4456
    },
    {
      "epoch": 0.5640344216654012,
      "grad_norm": 1.0271151065826416,
      "learning_rate": 1.710456934729863e-05,
      "loss": 0.6343,
      "step": 4457
    },
    {
      "epoch": 0.5641609719058466,
      "grad_norm": 1.1024787425994873,
      "learning_rate": 1.710330804294342e-05,
      "loss": 0.6459,
      "step": 4458
    },
    {
      "epoch": 0.5642875221462921,
      "grad_norm": 1.030055046081543,
      "learning_rate": 1.7102046510450767e-05,
      "loss": 0.64,
      "step": 4459
    },
    {
      "epoch": 0.5644140723867376,
      "grad_norm": 1.0623797178268433,
      "learning_rate": 1.7100784749861192e-05,
      "loss": 0.6045,
      "step": 4460
    },
    {
      "epoch": 0.564540622627183,
      "grad_norm": 1.0164778232574463,
      "learning_rate": 1.7099522761215223e-05,
      "loss": 0.6685,
      "step": 4461
    },
    {
      "epoch": 0.5646671728676285,
      "grad_norm": 1.0192677974700928,
      "learning_rate": 1.7098260544553388e-05,
      "loss": 0.6259,
      "step": 4462
    },
    {
      "epoch": 0.5647937231080739,
      "grad_norm": 1.0955913066864014,
      "learning_rate": 1.7096998099916224e-05,
      "loss": 0.6412,
      "step": 4463
    },
    {
      "epoch": 0.5649202733485194,
      "grad_norm": 0.9351078271865845,
      "learning_rate": 1.7095735427344275e-05,
      "loss": 0.5741,
      "step": 4464
    },
    {
      "epoch": 0.5650468235889649,
      "grad_norm": 1.3807083368301392,
      "learning_rate": 1.7094472526878102e-05,
      "loss": 0.6684,
      "step": 4465
    },
    {
      "epoch": 0.5651733738294102,
      "grad_norm": 0.9096883535385132,
      "learning_rate": 1.709320939855826e-05,
      "loss": 0.618,
      "step": 4466
    },
    {
      "epoch": 0.5652999240698557,
      "grad_norm": 1.01005220413208,
      "learning_rate": 1.709194604242532e-05,
      "loss": 0.6488,
      "step": 4467
    },
    {
      "epoch": 0.5654264743103012,
      "grad_norm": 0.952096700668335,
      "learning_rate": 1.7090682458519856e-05,
      "loss": 0.6764,
      "step": 4468
    },
    {
      "epoch": 0.5655530245507466,
      "grad_norm": 1.0432758331298828,
      "learning_rate": 1.708941864688245e-05,
      "loss": 0.6327,
      "step": 4469
    },
    {
      "epoch": 0.5656795747911921,
      "grad_norm": 1.042814016342163,
      "learning_rate": 1.708815460755369e-05,
      "loss": 0.6547,
      "step": 4470
    },
    {
      "epoch": 0.5658061250316375,
      "grad_norm": 1.3224154710769653,
      "learning_rate": 1.708689034057418e-05,
      "loss": 0.6082,
      "step": 4471
    },
    {
      "epoch": 0.565932675272083,
      "grad_norm": 1.0976248979568481,
      "learning_rate": 1.7085625845984517e-05,
      "loss": 0.66,
      "step": 4472
    },
    {
      "epoch": 0.5660592255125285,
      "grad_norm": 1.2413684129714966,
      "learning_rate": 1.7084361123825316e-05,
      "loss": 0.7167,
      "step": 4473
    },
    {
      "epoch": 0.5661857757529739,
      "grad_norm": 1.1345349550247192,
      "learning_rate": 1.7083096174137195e-05,
      "loss": 0.6332,
      "step": 4474
    },
    {
      "epoch": 0.5663123259934194,
      "grad_norm": 0.9562109112739563,
      "learning_rate": 1.7081830996960786e-05,
      "loss": 0.622,
      "step": 4475
    },
    {
      "epoch": 0.5664388762338648,
      "grad_norm": 1.1085622310638428,
      "learning_rate": 1.7080565592336714e-05,
      "loss": 0.5859,
      "step": 4476
    },
    {
      "epoch": 0.5665654264743103,
      "grad_norm": 1.0570464134216309,
      "learning_rate": 1.7079299960305628e-05,
      "loss": 0.6527,
      "step": 4477
    },
    {
      "epoch": 0.5666919767147558,
      "grad_norm": 1.3536789417266846,
      "learning_rate": 1.707803410090817e-05,
      "loss": 0.6247,
      "step": 4478
    },
    {
      "epoch": 0.5668185269552012,
      "grad_norm": 1.125487208366394,
      "learning_rate": 1.7076768014184998e-05,
      "loss": 0.6006,
      "step": 4479
    },
    {
      "epoch": 0.5669450771956467,
      "grad_norm": 0.9939672350883484,
      "learning_rate": 1.707550170017678e-05,
      "loss": 0.6402,
      "step": 4480
    },
    {
      "epoch": 0.5670716274360921,
      "grad_norm": 1.0981162786483765,
      "learning_rate": 1.7074235158924176e-05,
      "loss": 0.6376,
      "step": 4481
    },
    {
      "epoch": 0.5671981776765376,
      "grad_norm": 0.9770739078521729,
      "learning_rate": 1.7072968390467874e-05,
      "loss": 0.5818,
      "step": 4482
    },
    {
      "epoch": 0.5673247279169831,
      "grad_norm": 1.0756548643112183,
      "learning_rate": 1.7071701394848556e-05,
      "loss": 0.6588,
      "step": 4483
    },
    {
      "epoch": 0.5674512781574285,
      "grad_norm": 0.9181696772575378,
      "learning_rate": 1.7070434172106908e-05,
      "loss": 0.6107,
      "step": 4484
    },
    {
      "epoch": 0.567577828397874,
      "grad_norm": 0.9676690697669983,
      "learning_rate": 1.7069166722283635e-05,
      "loss": 0.6203,
      "step": 4485
    },
    {
      "epoch": 0.5677043786383195,
      "grad_norm": 1.0053426027297974,
      "learning_rate": 1.7067899045419442e-05,
      "loss": 0.6234,
      "step": 4486
    },
    {
      "epoch": 0.5678309288787649,
      "grad_norm": 0.9963805079460144,
      "learning_rate": 1.7066631141555047e-05,
      "loss": 0.6749,
      "step": 4487
    },
    {
      "epoch": 0.5679574791192104,
      "grad_norm": 1.1691746711730957,
      "learning_rate": 1.7065363010731165e-05,
      "loss": 0.6694,
      "step": 4488
    },
    {
      "epoch": 0.5680840293596557,
      "grad_norm": 0.9708976149559021,
      "learning_rate": 1.706409465298853e-05,
      "loss": 0.5993,
      "step": 4489
    },
    {
      "epoch": 0.5682105796001012,
      "grad_norm": 1.0692623853683472,
      "learning_rate": 1.7062826068367874e-05,
      "loss": 0.6247,
      "step": 4490
    },
    {
      "epoch": 0.5683371298405467,
      "grad_norm": 1.0893223285675049,
      "learning_rate": 1.7061557256909943e-05,
      "loss": 0.5966,
      "step": 4491
    },
    {
      "epoch": 0.5684636800809921,
      "grad_norm": 1.043317198753357,
      "learning_rate": 1.7060288218655484e-05,
      "loss": 0.6536,
      "step": 4492
    },
    {
      "epoch": 0.5685902303214376,
      "grad_norm": 0.9517982006072998,
      "learning_rate": 1.705901895364526e-05,
      "loss": 0.6593,
      "step": 4493
    },
    {
      "epoch": 0.568716780561883,
      "grad_norm": 0.9267576932907104,
      "learning_rate": 1.7057749461920034e-05,
      "loss": 0.6379,
      "step": 4494
    },
    {
      "epoch": 0.5688433308023285,
      "grad_norm": 0.9855020046234131,
      "learning_rate": 1.705647974352058e-05,
      "loss": 0.5829,
      "step": 4495
    },
    {
      "epoch": 0.568969881042774,
      "grad_norm": 1.025022268295288,
      "learning_rate": 1.705520979848767e-05,
      "loss": 0.6268,
      "step": 4496
    },
    {
      "epoch": 0.5690964312832194,
      "grad_norm": 0.9540415406227112,
      "learning_rate": 1.70539396268621e-05,
      "loss": 0.6353,
      "step": 4497
    },
    {
      "epoch": 0.5692229815236649,
      "grad_norm": 0.9788616895675659,
      "learning_rate": 1.7052669228684658e-05,
      "loss": 0.6205,
      "step": 4498
    },
    {
      "epoch": 0.5693495317641104,
      "grad_norm": 1.1064794063568115,
      "learning_rate": 1.705139860399615e-05,
      "loss": 0.6874,
      "step": 4499
    },
    {
      "epoch": 0.5694760820045558,
      "grad_norm": 1.0142152309417725,
      "learning_rate": 1.7050127752837383e-05,
      "loss": 0.6346,
      "step": 4500
    },
    {
      "epoch": 0.5696026322450013,
      "grad_norm": 0.981065034866333,
      "learning_rate": 1.704885667524917e-05,
      "loss": 0.6226,
      "step": 4501
    },
    {
      "epoch": 0.5697291824854467,
      "grad_norm": 0.9810211658477783,
      "learning_rate": 1.704758537127234e-05,
      "loss": 0.6265,
      "step": 4502
    },
    {
      "epoch": 0.5698557327258922,
      "grad_norm": 1.0564448833465576,
      "learning_rate": 1.7046313840947715e-05,
      "loss": 0.6108,
      "step": 4503
    },
    {
      "epoch": 0.5699822829663377,
      "grad_norm": 1.0710499286651611,
      "learning_rate": 1.7045042084316145e-05,
      "loss": 0.6261,
      "step": 4504
    },
    {
      "epoch": 0.5701088332067831,
      "grad_norm": 0.9845359921455383,
      "learning_rate": 1.7043770101418466e-05,
      "loss": 0.6038,
      "step": 4505
    },
    {
      "epoch": 0.5702353834472286,
      "grad_norm": 0.9840255975723267,
      "learning_rate": 1.7042497892295532e-05,
      "loss": 0.626,
      "step": 4506
    },
    {
      "epoch": 0.570361933687674,
      "grad_norm": 0.9479249715805054,
      "learning_rate": 1.7041225456988204e-05,
      "loss": 0.6248,
      "step": 4507
    },
    {
      "epoch": 0.5704884839281195,
      "grad_norm": 0.9910328388214111,
      "learning_rate": 1.703995279553735e-05,
      "loss": 0.5933,
      "step": 4508
    },
    {
      "epoch": 0.570615034168565,
      "grad_norm": 1.0415199995040894,
      "learning_rate": 1.703867990798384e-05,
      "loss": 0.6034,
      "step": 4509
    },
    {
      "epoch": 0.5707415844090103,
      "grad_norm": 0.9912620186805725,
      "learning_rate": 1.7037406794368562e-05,
      "loss": 0.6566,
      "step": 4510
    },
    {
      "epoch": 0.5708681346494558,
      "grad_norm": 1.078033208847046,
      "learning_rate": 1.7036133454732394e-05,
      "loss": 0.6268,
      "step": 4511
    },
    {
      "epoch": 0.5709946848899012,
      "grad_norm": 1.0096882581710815,
      "learning_rate": 1.7034859889116244e-05,
      "loss": 0.6203,
      "step": 4512
    },
    {
      "epoch": 0.5711212351303467,
      "grad_norm": 0.9665147662162781,
      "learning_rate": 1.7033586097561008e-05,
      "loss": 0.6214,
      "step": 4513
    },
    {
      "epoch": 0.5712477853707922,
      "grad_norm": 1.0038434267044067,
      "learning_rate": 1.7032312080107598e-05,
      "loss": 0.6698,
      "step": 4514
    },
    {
      "epoch": 0.5713743356112376,
      "grad_norm": 1.0176647901535034,
      "learning_rate": 1.703103783679693e-05,
      "loss": 0.6947,
      "step": 4515
    },
    {
      "epoch": 0.5715008858516831,
      "grad_norm": 0.9229211211204529,
      "learning_rate": 1.7029763367669934e-05,
      "loss": 0.6071,
      "step": 4516
    },
    {
      "epoch": 0.5716274360921286,
      "grad_norm": 0.9994952082633972,
      "learning_rate": 1.7028488672767538e-05,
      "loss": 0.6525,
      "step": 4517
    },
    {
      "epoch": 0.571753986332574,
      "grad_norm": 0.9967939257621765,
      "learning_rate": 1.7027213752130682e-05,
      "loss": 0.6234,
      "step": 4518
    },
    {
      "epoch": 0.5718805365730195,
      "grad_norm": 0.9175492525100708,
      "learning_rate": 1.7025938605800315e-05,
      "loss": 0.6013,
      "step": 4519
    },
    {
      "epoch": 0.5720070868134649,
      "grad_norm": 0.9440793395042419,
      "learning_rate": 1.7024663233817386e-05,
      "loss": 0.6225,
      "step": 4520
    },
    {
      "epoch": 0.5721336370539104,
      "grad_norm": 0.9530991315841675,
      "learning_rate": 1.702338763622286e-05,
      "loss": 0.6422,
      "step": 4521
    },
    {
      "epoch": 0.5722601872943559,
      "grad_norm": 0.9865661263465881,
      "learning_rate": 1.7022111813057708e-05,
      "loss": 0.666,
      "step": 4522
    },
    {
      "epoch": 0.5723867375348013,
      "grad_norm": 0.9985664486885071,
      "learning_rate": 1.70208357643629e-05,
      "loss": 0.6516,
      "step": 4523
    },
    {
      "epoch": 0.5725132877752468,
      "grad_norm": 0.9142382740974426,
      "learning_rate": 1.7019559490179423e-05,
      "loss": 0.6153,
      "step": 4524
    },
    {
      "epoch": 0.5726398380156922,
      "grad_norm": 0.9869629144668579,
      "learning_rate": 1.7018282990548267e-05,
      "loss": 0.6354,
      "step": 4525
    },
    {
      "epoch": 0.5727663882561377,
      "grad_norm": 1.0132321119308472,
      "learning_rate": 1.7017006265510425e-05,
      "loss": 0.6398,
      "step": 4526
    },
    {
      "epoch": 0.5728929384965832,
      "grad_norm": 0.9964350461959839,
      "learning_rate": 1.7015729315106905e-05,
      "loss": 0.645,
      "step": 4527
    },
    {
      "epoch": 0.5730194887370286,
      "grad_norm": 0.9556720852851868,
      "learning_rate": 1.7014452139378722e-05,
      "loss": 0.6149,
      "step": 4528
    },
    {
      "epoch": 0.5731460389774741,
      "grad_norm": 0.933444619178772,
      "learning_rate": 1.7013174738366888e-05,
      "loss": 0.6113,
      "step": 4529
    },
    {
      "epoch": 0.5732725892179195,
      "grad_norm": 0.9476785063743591,
      "learning_rate": 1.7011897112112438e-05,
      "loss": 0.5709,
      "step": 4530
    },
    {
      "epoch": 0.573399139458365,
      "grad_norm": 0.992634117603302,
      "learning_rate": 1.7010619260656397e-05,
      "loss": 0.6229,
      "step": 4531
    },
    {
      "epoch": 0.5735256896988105,
      "grad_norm": 0.9734261631965637,
      "learning_rate": 1.700934118403981e-05,
      "loss": 0.6683,
      "step": 4532
    },
    {
      "epoch": 0.5736522399392558,
      "grad_norm": 0.9342517852783203,
      "learning_rate": 1.7008062882303724e-05,
      "loss": 0.6081,
      "step": 4533
    },
    {
      "epoch": 0.5737787901797013,
      "grad_norm": 0.9744494557380676,
      "learning_rate": 1.7006784355489195e-05,
      "loss": 0.6237,
      "step": 4534
    },
    {
      "epoch": 0.5739053404201468,
      "grad_norm": 0.9565207362174988,
      "learning_rate": 1.700550560363729e-05,
      "loss": 0.5943,
      "step": 4535
    },
    {
      "epoch": 0.5740318906605922,
      "grad_norm": 1.017101526260376,
      "learning_rate": 1.700422662678907e-05,
      "loss": 0.659,
      "step": 4536
    },
    {
      "epoch": 0.5741584409010377,
      "grad_norm": 0.9707169532775879,
      "learning_rate": 1.700294742498562e-05,
      "loss": 0.6406,
      "step": 4537
    },
    {
      "epoch": 0.5742849911414831,
      "grad_norm": 1.0428005456924438,
      "learning_rate": 1.7001667998268017e-05,
      "loss": 0.6431,
      "step": 4538
    },
    {
      "epoch": 0.5744115413819286,
      "grad_norm": 0.9508194327354431,
      "learning_rate": 1.7000388346677356e-05,
      "loss": 0.6466,
      "step": 4539
    },
    {
      "epoch": 0.5745380916223741,
      "grad_norm": 0.9433960914611816,
      "learning_rate": 1.699910847025474e-05,
      "loss": 0.574,
      "step": 4540
    },
    {
      "epoch": 0.5746646418628195,
      "grad_norm": 1.0046008825302124,
      "learning_rate": 1.6997828369041266e-05,
      "loss": 0.6152,
      "step": 4541
    },
    {
      "epoch": 0.574791192103265,
      "grad_norm": 1.0230684280395508,
      "learning_rate": 1.6996548043078053e-05,
      "loss": 0.625,
      "step": 4542
    },
    {
      "epoch": 0.5749177423437104,
      "grad_norm": 1.0168534517288208,
      "learning_rate": 1.699526749240622e-05,
      "loss": 0.6341,
      "step": 4543
    },
    {
      "epoch": 0.5750442925841559,
      "grad_norm": 0.9866940975189209,
      "learning_rate": 1.6993986717066894e-05,
      "loss": 0.6173,
      "step": 4544
    },
    {
      "epoch": 0.5751708428246014,
      "grad_norm": 0.9577439427375793,
      "learning_rate": 1.699270571710121e-05,
      "loss": 0.6395,
      "step": 4545
    },
    {
      "epoch": 0.5752973930650468,
      "grad_norm": 1.0087182521820068,
      "learning_rate": 1.699142449255031e-05,
      "loss": 0.6528,
      "step": 4546
    },
    {
      "epoch": 0.5754239433054923,
      "grad_norm": 0.9650278687477112,
      "learning_rate": 1.6990143043455348e-05,
      "loss": 0.6148,
      "step": 4547
    },
    {
      "epoch": 0.5755504935459378,
      "grad_norm": 0.9704685807228088,
      "learning_rate": 1.698886136985747e-05,
      "loss": 0.6183,
      "step": 4548
    },
    {
      "epoch": 0.5756770437863832,
      "grad_norm": 0.9676848649978638,
      "learning_rate": 1.698757947179785e-05,
      "loss": 0.6219,
      "step": 4549
    },
    {
      "epoch": 0.5758035940268287,
      "grad_norm": 0.9982961416244507,
      "learning_rate": 1.698629734931765e-05,
      "loss": 0.6749,
      "step": 4550
    },
    {
      "epoch": 0.5759301442672741,
      "grad_norm": 1.0066721439361572,
      "learning_rate": 1.698501500245805e-05,
      "loss": 0.6446,
      "step": 4551
    },
    {
      "epoch": 0.5760566945077196,
      "grad_norm": 0.9544116258621216,
      "learning_rate": 1.6983732431260244e-05,
      "loss": 0.6285,
      "step": 4552
    },
    {
      "epoch": 0.5761832447481651,
      "grad_norm": 0.9703861474990845,
      "learning_rate": 1.6982449635765416e-05,
      "loss": 0.6375,
      "step": 4553
    },
    {
      "epoch": 0.5763097949886105,
      "grad_norm": 1.0092310905456543,
      "learning_rate": 1.6981166616014766e-05,
      "loss": 0.6306,
      "step": 4554
    },
    {
      "epoch": 0.576436345229056,
      "grad_norm": 0.997493326663971,
      "learning_rate": 1.6979883372049505e-05,
      "loss": 0.6629,
      "step": 4555
    },
    {
      "epoch": 0.5765628954695013,
      "grad_norm": 0.9903466105461121,
      "learning_rate": 1.697859990391084e-05,
      "loss": 0.6547,
      "step": 4556
    },
    {
      "epoch": 0.5766894457099468,
      "grad_norm": 0.9717198014259338,
      "learning_rate": 1.697731621164e-05,
      "loss": 0.6463,
      "step": 4557
    },
    {
      "epoch": 0.5768159959503923,
      "grad_norm": 0.9767163395881653,
      "learning_rate": 1.697603229527821e-05,
      "loss": 0.6378,
      "step": 4558
    },
    {
      "epoch": 0.5769425461908377,
      "grad_norm": 1.0397379398345947,
      "learning_rate": 1.6974748154866706e-05,
      "loss": 0.7091,
      "step": 4559
    },
    {
      "epoch": 0.5770690964312832,
      "grad_norm": 1.0153272151947021,
      "learning_rate": 1.6973463790446726e-05,
      "loss": 0.63,
      "step": 4560
    },
    {
      "epoch": 0.5771956466717286,
      "grad_norm": 1.0787832736968994,
      "learning_rate": 1.6972179202059532e-05,
      "loss": 0.6442,
      "step": 4561
    },
    {
      "epoch": 0.5773221969121741,
      "grad_norm": 1.0410041809082031,
      "learning_rate": 1.697089438974637e-05,
      "loss": 0.6409,
      "step": 4562
    },
    {
      "epoch": 0.5774487471526196,
      "grad_norm": 0.9646349549293518,
      "learning_rate": 1.6969609353548505e-05,
      "loss": 0.6055,
      "step": 4563
    },
    {
      "epoch": 0.577575297393065,
      "grad_norm": 1.0764778852462769,
      "learning_rate": 1.6968324093507213e-05,
      "loss": 0.6397,
      "step": 4564
    },
    {
      "epoch": 0.5777018476335105,
      "grad_norm": 0.9947677850723267,
      "learning_rate": 1.6967038609663774e-05,
      "loss": 0.6517,
      "step": 4565
    },
    {
      "epoch": 0.577828397873956,
      "grad_norm": 1.0629146099090576,
      "learning_rate": 1.6965752902059472e-05,
      "loss": 0.644,
      "step": 4566
    },
    {
      "epoch": 0.5779549481144014,
      "grad_norm": 1.0688378810882568,
      "learning_rate": 1.6964466970735597e-05,
      "loss": 0.6569,
      "step": 4567
    },
    {
      "epoch": 0.5780814983548469,
      "grad_norm": 1.0131865739822388,
      "learning_rate": 1.6963180815733453e-05,
      "loss": 0.6229,
      "step": 4568
    },
    {
      "epoch": 0.5782080485952923,
      "grad_norm": 0.9698796272277832,
      "learning_rate": 1.6961894437094345e-05,
      "loss": 0.5702,
      "step": 4569
    },
    {
      "epoch": 0.5783345988357378,
      "grad_norm": 1.2299432754516602,
      "learning_rate": 1.6960607834859594e-05,
      "loss": 0.6777,
      "step": 4570
    },
    {
      "epoch": 0.5784611490761833,
      "grad_norm": 1.1523182392120361,
      "learning_rate": 1.6959321009070513e-05,
      "loss": 0.6559,
      "step": 4571
    },
    {
      "epoch": 0.5785876993166287,
      "grad_norm": 1.0355820655822754,
      "learning_rate": 1.6958033959768438e-05,
      "loss": 0.6462,
      "step": 4572
    },
    {
      "epoch": 0.5787142495570742,
      "grad_norm": 1.1199853420257568,
      "learning_rate": 1.69567466869947e-05,
      "loss": 0.6215,
      "step": 4573
    },
    {
      "epoch": 0.5788407997975196,
      "grad_norm": 0.9873904585838318,
      "learning_rate": 1.6955459190790643e-05,
      "loss": 0.6396,
      "step": 4574
    },
    {
      "epoch": 0.5789673500379651,
      "grad_norm": 0.987755298614502,
      "learning_rate": 1.6954171471197624e-05,
      "loss": 0.6413,
      "step": 4575
    },
    {
      "epoch": 0.5790939002784106,
      "grad_norm": 1.0501543283462524,
      "learning_rate": 1.6952883528256995e-05,
      "loss": 0.6286,
      "step": 4576
    },
    {
      "epoch": 0.579220450518856,
      "grad_norm": 1.0330153703689575,
      "learning_rate": 1.695159536201012e-05,
      "loss": 0.626,
      "step": 4577
    },
    {
      "epoch": 0.5793470007593015,
      "grad_norm": 0.9823784828186035,
      "learning_rate": 1.6950306972498377e-05,
      "loss": 0.6211,
      "step": 4578
    },
    {
      "epoch": 0.5794735509997468,
      "grad_norm": 1.00816011428833,
      "learning_rate": 1.694901835976314e-05,
      "loss": 0.6873,
      "step": 4579
    },
    {
      "epoch": 0.5796001012401923,
      "grad_norm": 1.0862774848937988,
      "learning_rate": 1.694772952384579e-05,
      "loss": 0.7046,
      "step": 4580
    },
    {
      "epoch": 0.5797266514806378,
      "grad_norm": 0.989403247833252,
      "learning_rate": 1.6946440464787738e-05,
      "loss": 0.6301,
      "step": 4581
    },
    {
      "epoch": 0.5798532017210832,
      "grad_norm": 1.041321039199829,
      "learning_rate": 1.6945151182630374e-05,
      "loss": 0.6609,
      "step": 4582
    },
    {
      "epoch": 0.5799797519615287,
      "grad_norm": 1.0074390172958374,
      "learning_rate": 1.69438616774151e-05,
      "loss": 0.6343,
      "step": 4583
    },
    {
      "epoch": 0.5801063022019742,
      "grad_norm": 1.0424922704696655,
      "learning_rate": 1.6942571949183342e-05,
      "loss": 0.6637,
      "step": 4584
    },
    {
      "epoch": 0.5802328524424196,
      "grad_norm": 0.9726418256759644,
      "learning_rate": 1.6941281997976517e-05,
      "loss": 0.6253,
      "step": 4585
    },
    {
      "epoch": 0.5803594026828651,
      "grad_norm": 0.958712637424469,
      "learning_rate": 1.6939991823836055e-05,
      "loss": 0.5961,
      "step": 4586
    },
    {
      "epoch": 0.5804859529233105,
      "grad_norm": 1.007611632347107,
      "learning_rate": 1.6938701426803392e-05,
      "loss": 0.6232,
      "step": 4587
    },
    {
      "epoch": 0.580612503163756,
      "grad_norm": 1.0996941328048706,
      "learning_rate": 1.693741080691998e-05,
      "loss": 0.679,
      "step": 4588
    },
    {
      "epoch": 0.5807390534042015,
      "grad_norm": 0.9582080245018005,
      "learning_rate": 1.6936119964227254e-05,
      "loss": 0.6196,
      "step": 4589
    },
    {
      "epoch": 0.5808656036446469,
      "grad_norm": 1.0576802492141724,
      "learning_rate": 1.6934828898766683e-05,
      "loss": 0.6389,
      "step": 4590
    },
    {
      "epoch": 0.5809921538850924,
      "grad_norm": 0.9927720427513123,
      "learning_rate": 1.693353761057973e-05,
      "loss": 0.6485,
      "step": 4591
    },
    {
      "epoch": 0.5811187041255378,
      "grad_norm": 0.8905054926872253,
      "learning_rate": 1.6932246099707866e-05,
      "loss": 0.5941,
      "step": 4592
    },
    {
      "epoch": 0.5812452543659833,
      "grad_norm": 1.0080022811889648,
      "learning_rate": 1.693095436619258e-05,
      "loss": 0.6058,
      "step": 4593
    },
    {
      "epoch": 0.5813718046064288,
      "grad_norm": 0.9897368550300598,
      "learning_rate": 1.6929662410075342e-05,
      "loss": 0.6517,
      "step": 4594
    },
    {
      "epoch": 0.5814983548468742,
      "grad_norm": 0.9862417578697205,
      "learning_rate": 1.6928370231397658e-05,
      "loss": 0.6204,
      "step": 4595
    },
    {
      "epoch": 0.5816249050873197,
      "grad_norm": 0.9930499792098999,
      "learning_rate": 1.6927077830201023e-05,
      "loss": 0.5957,
      "step": 4596
    },
    {
      "epoch": 0.5817514553277652,
      "grad_norm": 0.969685971736908,
      "learning_rate": 1.692578520652695e-05,
      "loss": 0.6438,
      "step": 4597
    },
    {
      "epoch": 0.5818780055682106,
      "grad_norm": 1.0098382234573364,
      "learning_rate": 1.6924492360416953e-05,
      "loss": 0.6145,
      "step": 4598
    },
    {
      "epoch": 0.5820045558086561,
      "grad_norm": 1.1395081281661987,
      "learning_rate": 1.692319929191255e-05,
      "loss": 0.6812,
      "step": 4599
    },
    {
      "epoch": 0.5821311060491015,
      "grad_norm": 0.9741681814193726,
      "learning_rate": 1.6921906001055275e-05,
      "loss": 0.6297,
      "step": 4600
    },
    {
      "epoch": 0.582257656289547,
      "grad_norm": 0.995086133480072,
      "learning_rate": 1.6920612487886666e-05,
      "loss": 0.5966,
      "step": 4601
    },
    {
      "epoch": 0.5823842065299925,
      "grad_norm": 1.0202003717422485,
      "learning_rate": 1.691931875244826e-05,
      "loss": 0.6134,
      "step": 4602
    },
    {
      "epoch": 0.5825107567704378,
      "grad_norm": 0.9814889430999756,
      "learning_rate": 1.6918024794781614e-05,
      "loss": 0.6066,
      "step": 4603
    },
    {
      "epoch": 0.5826373070108833,
      "grad_norm": 0.9860258102416992,
      "learning_rate": 1.6916730614928283e-05,
      "loss": 0.6192,
      "step": 4604
    },
    {
      "epoch": 0.5827638572513287,
      "grad_norm": 1.0130950212478638,
      "learning_rate": 1.6915436212929842e-05,
      "loss": 0.646,
      "step": 4605
    },
    {
      "epoch": 0.5828904074917742,
      "grad_norm": 0.9307775497436523,
      "learning_rate": 1.691414158882785e-05,
      "loss": 0.5949,
      "step": 4606
    },
    {
      "epoch": 0.5830169577322197,
      "grad_norm": 1.0142015218734741,
      "learning_rate": 1.691284674266389e-05,
      "loss": 0.6705,
      "step": 4607
    },
    {
      "epoch": 0.5831435079726651,
      "grad_norm": 1.0592799186706543,
      "learning_rate": 1.691155167447955e-05,
      "loss": 0.6313,
      "step": 4608
    },
    {
      "epoch": 0.5832700582131106,
      "grad_norm": 1.0018006563186646,
      "learning_rate": 1.691025638431643e-05,
      "loss": 0.6621,
      "step": 4609
    },
    {
      "epoch": 0.583396608453556,
      "grad_norm": 1.0330278873443604,
      "learning_rate": 1.690896087221612e-05,
      "loss": 0.5913,
      "step": 4610
    },
    {
      "epoch": 0.5835231586940015,
      "grad_norm": 4.508431911468506,
      "learning_rate": 1.6907665138220234e-05,
      "loss": 0.5996,
      "step": 4611
    },
    {
      "epoch": 0.583649708934447,
      "grad_norm": 1.0343027114868164,
      "learning_rate": 1.6906369182370384e-05,
      "loss": 0.6338,
      "step": 4612
    },
    {
      "epoch": 0.5837762591748924,
      "grad_norm": 0.9435905814170837,
      "learning_rate": 1.69050730047082e-05,
      "loss": 0.5987,
      "step": 4613
    },
    {
      "epoch": 0.5839028094153379,
      "grad_norm": 0.9831632971763611,
      "learning_rate": 1.6903776605275304e-05,
      "loss": 0.6395,
      "step": 4614
    },
    {
      "epoch": 0.5840293596557834,
      "grad_norm": 0.986305832862854,
      "learning_rate": 1.6902479984113336e-05,
      "loss": 0.6415,
      "step": 4615
    },
    {
      "epoch": 0.5841559098962288,
      "grad_norm": 1.0204402208328247,
      "learning_rate": 1.6901183141263934e-05,
      "loss": 0.6323,
      "step": 4616
    },
    {
      "epoch": 0.5842824601366743,
      "grad_norm": 1.0024161338806152,
      "learning_rate": 1.689988607676876e-05,
      "loss": 0.6576,
      "step": 4617
    },
    {
      "epoch": 0.5844090103771197,
      "grad_norm": 0.9526926279067993,
      "learning_rate": 1.689858879066946e-05,
      "loss": 0.6275,
      "step": 4618
    },
    {
      "epoch": 0.5845355606175652,
      "grad_norm": 1.0449203252792358,
      "learning_rate": 1.6897291283007704e-05,
      "loss": 0.6356,
      "step": 4619
    },
    {
      "epoch": 0.5846621108580107,
      "grad_norm": 1.050522804260254,
      "learning_rate": 1.6895993553825164e-05,
      "loss": 0.6681,
      "step": 4620
    },
    {
      "epoch": 0.5847886610984561,
      "grad_norm": 1.0921432971954346,
      "learning_rate": 1.689469560316352e-05,
      "loss": 0.6575,
      "step": 4621
    },
    {
      "epoch": 0.5849152113389016,
      "grad_norm": 1.1313753128051758,
      "learning_rate": 1.689339743106446e-05,
      "loss": 0.6709,
      "step": 4622
    },
    {
      "epoch": 0.585041761579347,
      "grad_norm": 1.1128360033035278,
      "learning_rate": 1.6892099037569676e-05,
      "loss": 0.6709,
      "step": 4623
    },
    {
      "epoch": 0.5851683118197925,
      "grad_norm": 0.9635475277900696,
      "learning_rate": 1.6890800422720867e-05,
      "loss": 0.6358,
      "step": 4624
    },
    {
      "epoch": 0.585294862060238,
      "grad_norm": 0.9821682572364807,
      "learning_rate": 1.688950158655974e-05,
      "loss": 0.6254,
      "step": 4625
    },
    {
      "epoch": 0.5854214123006833,
      "grad_norm": 1.024868130683899,
      "learning_rate": 1.6888202529128012e-05,
      "loss": 0.6506,
      "step": 4626
    },
    {
      "epoch": 0.5855479625411288,
      "grad_norm": 1.0012316703796387,
      "learning_rate": 1.6886903250467403e-05,
      "loss": 0.6467,
      "step": 4627
    },
    {
      "epoch": 0.5856745127815742,
      "grad_norm": 0.9731863141059875,
      "learning_rate": 1.6885603750619647e-05,
      "loss": 0.6624,
      "step": 4628
    },
    {
      "epoch": 0.5858010630220197,
      "grad_norm": 0.9703306555747986,
      "learning_rate": 1.6884304029626473e-05,
      "loss": 0.6244,
      "step": 4629
    },
    {
      "epoch": 0.5859276132624652,
      "grad_norm": 0.9990056753158569,
      "learning_rate": 1.6883004087529632e-05,
      "loss": 0.649,
      "step": 4630
    },
    {
      "epoch": 0.5860541635029106,
      "grad_norm": 1.0198856592178345,
      "learning_rate": 1.6881703924370863e-05,
      "loss": 0.6497,
      "step": 4631
    },
    {
      "epoch": 0.5861807137433561,
      "grad_norm": 0.9558019638061523,
      "learning_rate": 1.6880403540191935e-05,
      "loss": 0.643,
      "step": 4632
    },
    {
      "epoch": 0.5863072639838016,
      "grad_norm": 0.9619436264038086,
      "learning_rate": 1.6879102935034606e-05,
      "loss": 0.6121,
      "step": 4633
    },
    {
      "epoch": 0.586433814224247,
      "grad_norm": 0.9764634370803833,
      "learning_rate": 1.6877802108940656e-05,
      "loss": 0.6308,
      "step": 4634
    },
    {
      "epoch": 0.5865603644646925,
      "grad_norm": 1.020692229270935,
      "learning_rate": 1.6876501061951854e-05,
      "loss": 0.6274,
      "step": 4635
    },
    {
      "epoch": 0.5866869147051379,
      "grad_norm": 0.9773028492927551,
      "learning_rate": 1.6875199794109988e-05,
      "loss": 0.649,
      "step": 4636
    },
    {
      "epoch": 0.5868134649455834,
      "grad_norm": 1.0829226970672607,
      "learning_rate": 1.687389830545685e-05,
      "loss": 0.6717,
      "step": 4637
    },
    {
      "epoch": 0.5869400151860289,
      "grad_norm": 1.0102622509002686,
      "learning_rate": 1.6872596596034246e-05,
      "loss": 0.6163,
      "step": 4638
    },
    {
      "epoch": 0.5870665654264743,
      "grad_norm": 0.9699188470840454,
      "learning_rate": 1.6871294665883976e-05,
      "loss": 0.6456,
      "step": 4639
    },
    {
      "epoch": 0.5871931156669198,
      "grad_norm": 1.0223515033721924,
      "learning_rate": 1.686999251504786e-05,
      "loss": 0.6333,
      "step": 4640
    },
    {
      "epoch": 0.5873196659073652,
      "grad_norm": 0.9757845997810364,
      "learning_rate": 1.6868690143567717e-05,
      "loss": 0.6142,
      "step": 4641
    },
    {
      "epoch": 0.5874462161478107,
      "grad_norm": 1.0009205341339111,
      "learning_rate": 1.6867387551485373e-05,
      "loss": 0.6489,
      "step": 4642
    },
    {
      "epoch": 0.5875727663882562,
      "grad_norm": 0.9710200428962708,
      "learning_rate": 1.6866084738842666e-05,
      "loss": 0.5779,
      "step": 4643
    },
    {
      "epoch": 0.5876993166287016,
      "grad_norm": 1.0013483762741089,
      "learning_rate": 1.6864781705681437e-05,
      "loss": 0.6397,
      "step": 4644
    },
    {
      "epoch": 0.5878258668691471,
      "grad_norm": 1.0376827716827393,
      "learning_rate": 1.686347845204354e-05,
      "loss": 0.6249,
      "step": 4645
    },
    {
      "epoch": 0.5879524171095926,
      "grad_norm": 1.016176462173462,
      "learning_rate": 1.6862174977970825e-05,
      "loss": 0.6532,
      "step": 4646
    },
    {
      "epoch": 0.588078967350038,
      "grad_norm": 1.0411306619644165,
      "learning_rate": 1.6860871283505157e-05,
      "loss": 0.6415,
      "step": 4647
    },
    {
      "epoch": 0.5882055175904835,
      "grad_norm": 1.0157209634780884,
      "learning_rate": 1.6859567368688414e-05,
      "loss": 0.627,
      "step": 4648
    },
    {
      "epoch": 0.5883320678309288,
      "grad_norm": 0.950716495513916,
      "learning_rate": 1.6858263233562467e-05,
      "loss": 0.5981,
      "step": 4649
    },
    {
      "epoch": 0.5884586180713743,
      "grad_norm": 0.9843599200248718,
      "learning_rate": 1.68569588781692e-05,
      "loss": 0.5918,
      "step": 4650
    },
    {
      "epoch": 0.5885851683118198,
      "grad_norm": 0.994611382484436,
      "learning_rate": 1.6855654302550512e-05,
      "loss": 0.6357,
      "step": 4651
    },
    {
      "epoch": 0.5887117185522652,
      "grad_norm": 1.0334490537643433,
      "learning_rate": 1.6854349506748296e-05,
      "loss": 0.6571,
      "step": 4652
    },
    {
      "epoch": 0.5888382687927107,
      "grad_norm": 0.9836839437484741,
      "learning_rate": 1.6853044490804462e-05,
      "loss": 0.6009,
      "step": 4653
    },
    {
      "epoch": 0.5889648190331561,
      "grad_norm": 1.0088367462158203,
      "learning_rate": 1.685173925476092e-05,
      "loss": 0.6444,
      "step": 4654
    },
    {
      "epoch": 0.5890913692736016,
      "grad_norm": 0.9689052700996399,
      "learning_rate": 1.685043379865959e-05,
      "loss": 0.6139,
      "step": 4655
    },
    {
      "epoch": 0.5892179195140471,
      "grad_norm": 0.981314480304718,
      "learning_rate": 1.6849128122542406e-05,
      "loss": 0.6371,
      "step": 4656
    },
    {
      "epoch": 0.5893444697544925,
      "grad_norm": 1.0500767230987549,
      "learning_rate": 1.6847822226451292e-05,
      "loss": 0.6416,
      "step": 4657
    },
    {
      "epoch": 0.589471019994938,
      "grad_norm": 0.9812139272689819,
      "learning_rate": 1.6846516110428203e-05,
      "loss": 0.5842,
      "step": 4658
    },
    {
      "epoch": 0.5895975702353834,
      "grad_norm": 1.0033787488937378,
      "learning_rate": 1.6845209774515076e-05,
      "loss": 0.6769,
      "step": 4659
    },
    {
      "epoch": 0.5897241204758289,
      "grad_norm": 1.010302186012268,
      "learning_rate": 1.684390321875387e-05,
      "loss": 0.6183,
      "step": 4660
    },
    {
      "epoch": 0.5898506707162744,
      "grad_norm": 0.9948256015777588,
      "learning_rate": 1.6842596443186547e-05,
      "loss": 0.6264,
      "step": 4661
    },
    {
      "epoch": 0.5899772209567198,
      "grad_norm": 0.9975658059120178,
      "learning_rate": 1.6841289447855082e-05,
      "loss": 0.6031,
      "step": 4662
    },
    {
      "epoch": 0.5901037711971653,
      "grad_norm": 0.9472527503967285,
      "learning_rate": 1.6839982232801444e-05,
      "loss": 0.6304,
      "step": 4663
    },
    {
      "epoch": 0.5902303214376108,
      "grad_norm": 0.9558452367782593,
      "learning_rate": 1.6838674798067623e-05,
      "loss": 0.6324,
      "step": 4664
    },
    {
      "epoch": 0.5903568716780562,
      "grad_norm": 1.0188488960266113,
      "learning_rate": 1.6837367143695608e-05,
      "loss": 0.6517,
      "step": 4665
    },
    {
      "epoch": 0.5904834219185017,
      "grad_norm": 1.2372242212295532,
      "learning_rate": 1.6836059269727397e-05,
      "loss": 0.6561,
      "step": 4666
    },
    {
      "epoch": 0.5906099721589471,
      "grad_norm": 0.9374577403068542,
      "learning_rate": 1.6834751176204994e-05,
      "loss": 0.6223,
      "step": 4667
    },
    {
      "epoch": 0.5907365223993926,
      "grad_norm": 0.9998795986175537,
      "learning_rate": 1.6833442863170414e-05,
      "loss": 0.6474,
      "step": 4668
    },
    {
      "epoch": 0.5908630726398381,
      "grad_norm": 0.948992908000946,
      "learning_rate": 1.6832134330665675e-05,
      "loss": 0.6247,
      "step": 4669
    },
    {
      "epoch": 0.5909896228802834,
      "grad_norm": 1.048763632774353,
      "learning_rate": 1.6830825578732798e-05,
      "loss": 0.6544,
      "step": 4670
    },
    {
      "epoch": 0.591116173120729,
      "grad_norm": 1.1603635549545288,
      "learning_rate": 1.6829516607413824e-05,
      "loss": 0.6094,
      "step": 4671
    },
    {
      "epoch": 0.5912427233611743,
      "grad_norm": 0.9884189367294312,
      "learning_rate": 1.6828207416750793e-05,
      "loss": 0.6034,
      "step": 4672
    },
    {
      "epoch": 0.5913692736016198,
      "grad_norm": 0.9401545524597168,
      "learning_rate": 1.6826898006785745e-05,
      "loss": 0.6,
      "step": 4673
    },
    {
      "epoch": 0.5914958238420653,
      "grad_norm": 0.9643544554710388,
      "learning_rate": 1.6825588377560743e-05,
      "loss": 0.6239,
      "step": 4674
    },
    {
      "epoch": 0.5916223740825107,
      "grad_norm": 1.0084402561187744,
      "learning_rate": 1.682427852911784e-05,
      "loss": 0.5917,
      "step": 4675
    },
    {
      "epoch": 0.5917489243229562,
      "grad_norm": 1.269783616065979,
      "learning_rate": 1.6822968461499112e-05,
      "loss": 0.6134,
      "step": 4676
    },
    {
      "epoch": 0.5918754745634016,
      "grad_norm": 0.9998228549957275,
      "learning_rate": 1.6821658174746628e-05,
      "loss": 0.6532,
      "step": 4677
    },
    {
      "epoch": 0.5920020248038471,
      "grad_norm": 0.9631361365318298,
      "learning_rate": 1.6820347668902478e-05,
      "loss": 0.6411,
      "step": 4678
    },
    {
      "epoch": 0.5921285750442926,
      "grad_norm": 1.02515709400177,
      "learning_rate": 1.6819036944008744e-05,
      "loss": 0.6443,
      "step": 4679
    },
    {
      "epoch": 0.592255125284738,
      "grad_norm": 0.9845661520957947,
      "learning_rate": 1.6817726000107534e-05,
      "loss": 0.6363,
      "step": 4680
    },
    {
      "epoch": 0.5923816755251835,
      "grad_norm": 1.161564588546753,
      "learning_rate": 1.6816414837240937e-05,
      "loss": 0.6688,
      "step": 4681
    },
    {
      "epoch": 0.592508225765629,
      "grad_norm": 1.0247256755828857,
      "learning_rate": 1.681510345545107e-05,
      "loss": 0.6132,
      "step": 4682
    },
    {
      "epoch": 0.5926347760060744,
      "grad_norm": 1.003604531288147,
      "learning_rate": 1.6813791854780053e-05,
      "loss": 0.6356,
      "step": 4683
    },
    {
      "epoch": 0.5927613262465199,
      "grad_norm": 1.0740771293640137,
      "learning_rate": 1.6812480035270012e-05,
      "loss": 0.6895,
      "step": 4684
    },
    {
      "epoch": 0.5928878764869653,
      "grad_norm": 0.9730736017227173,
      "learning_rate": 1.6811167996963074e-05,
      "loss": 0.5948,
      "step": 4685
    },
    {
      "epoch": 0.5930144267274108,
      "grad_norm": 0.9874363541603088,
      "learning_rate": 1.680985573990138e-05,
      "loss": 0.6466,
      "step": 4686
    },
    {
      "epoch": 0.5931409769678563,
      "grad_norm": 1.0544052124023438,
      "learning_rate": 1.6808543264127073e-05,
      "loss": 0.6523,
      "step": 4687
    },
    {
      "epoch": 0.5932675272083017,
      "grad_norm": 1.0170836448669434,
      "learning_rate": 1.680723056968231e-05,
      "loss": 0.6099,
      "step": 4688
    },
    {
      "epoch": 0.5933940774487472,
      "grad_norm": 0.9825134873390198,
      "learning_rate": 1.680591765660925e-05,
      "loss": 0.6455,
      "step": 4689
    },
    {
      "epoch": 0.5935206276891926,
      "grad_norm": 1.0418766736984253,
      "learning_rate": 1.680460452495006e-05,
      "loss": 0.6287,
      "step": 4690
    },
    {
      "epoch": 0.5936471779296381,
      "grad_norm": 0.9646141529083252,
      "learning_rate": 1.6803291174746913e-05,
      "loss": 0.6015,
      "step": 4691
    },
    {
      "epoch": 0.5937737281700836,
      "grad_norm": 9.958036422729492,
      "learning_rate": 1.680197760604199e-05,
      "loss": 0.6058,
      "step": 4692
    },
    {
      "epoch": 0.593900278410529,
      "grad_norm": 1.1330965757369995,
      "learning_rate": 1.680066381887748e-05,
      "loss": 0.5747,
      "step": 4693
    },
    {
      "epoch": 0.5940268286509744,
      "grad_norm": 0.9384613633155823,
      "learning_rate": 1.6799349813295575e-05,
      "loss": 0.6399,
      "step": 4694
    },
    {
      "epoch": 0.59415337889142,
      "grad_norm": 1.00328528881073,
      "learning_rate": 1.679803558933848e-05,
      "loss": 0.6288,
      "step": 4695
    },
    {
      "epoch": 0.5942799291318653,
      "grad_norm": 1.0050256252288818,
      "learning_rate": 1.6796721147048405e-05,
      "loss": 0.6457,
      "step": 4696
    },
    {
      "epoch": 0.5944064793723108,
      "grad_norm": 0.9676522612571716,
      "learning_rate": 1.6795406486467564e-05,
      "loss": 0.5901,
      "step": 4697
    },
    {
      "epoch": 0.5945330296127562,
      "grad_norm": 1.1216456890106201,
      "learning_rate": 1.6794091607638178e-05,
      "loss": 0.6651,
      "step": 4698
    },
    {
      "epoch": 0.5946595798532017,
      "grad_norm": 0.9091808199882507,
      "learning_rate": 1.679277651060248e-05,
      "loss": 0.6108,
      "step": 4699
    },
    {
      "epoch": 0.5947861300936472,
      "grad_norm": 1.018638014793396,
      "learning_rate": 1.6791461195402708e-05,
      "loss": 0.6175,
      "step": 4700
    },
    {
      "epoch": 0.5949126803340926,
      "grad_norm": 0.9438992142677307,
      "learning_rate": 1.6790145662081104e-05,
      "loss": 0.6288,
      "step": 4701
    },
    {
      "epoch": 0.5950392305745381,
      "grad_norm": 1.789417028427124,
      "learning_rate": 1.678882991067992e-05,
      "loss": 0.6507,
      "step": 4702
    },
    {
      "epoch": 0.5951657808149835,
      "grad_norm": 1.1440414190292358,
      "learning_rate": 1.6787513941241412e-05,
      "loss": 0.6926,
      "step": 4703
    },
    {
      "epoch": 0.595292331055429,
      "grad_norm": 0.9804644584655762,
      "learning_rate": 1.6786197753807848e-05,
      "loss": 0.6353,
      "step": 4704
    },
    {
      "epoch": 0.5954188812958745,
      "grad_norm": 0.9389517307281494,
      "learning_rate": 1.6784881348421498e-05,
      "loss": 0.5781,
      "step": 4705
    },
    {
      "epoch": 0.5955454315363199,
      "grad_norm": 0.9989356994628906,
      "learning_rate": 1.6783564725124643e-05,
      "loss": 0.676,
      "step": 4706
    },
    {
      "epoch": 0.5956719817767654,
      "grad_norm": 0.9631972908973694,
      "learning_rate": 1.6782247883959567e-05,
      "loss": 0.6513,
      "step": 4707
    },
    {
      "epoch": 0.5957985320172108,
      "grad_norm": 1.0232806205749512,
      "learning_rate": 1.6780930824968565e-05,
      "loss": 0.5914,
      "step": 4708
    },
    {
      "epoch": 0.5959250822576563,
      "grad_norm": 1.0711759328842163,
      "learning_rate": 1.6779613548193937e-05,
      "loss": 0.6378,
      "step": 4709
    },
    {
      "epoch": 0.5960516324981018,
      "grad_norm": 1.1483837366104126,
      "learning_rate": 1.6778296053677986e-05,
      "loss": 0.6462,
      "step": 4710
    },
    {
      "epoch": 0.5961781827385472,
      "grad_norm": 1.0427535772323608,
      "learning_rate": 1.6776978341463035e-05,
      "loss": 0.6371,
      "step": 4711
    },
    {
      "epoch": 0.5963047329789927,
      "grad_norm": 1.0654315948486328,
      "learning_rate": 1.6775660411591395e-05,
      "loss": 0.6189,
      "step": 4712
    },
    {
      "epoch": 0.5964312832194382,
      "grad_norm": 1.038988471031189,
      "learning_rate": 1.6774342264105396e-05,
      "loss": 0.6763,
      "step": 4713
    },
    {
      "epoch": 0.5965578334598836,
      "grad_norm": 0.9667231440544128,
      "learning_rate": 1.677302389904738e-05,
      "loss": 0.6641,
      "step": 4714
    },
    {
      "epoch": 0.5966843837003291,
      "grad_norm": 1.0320703983306885,
      "learning_rate": 1.6771705316459686e-05,
      "loss": 0.6144,
      "step": 4715
    },
    {
      "epoch": 0.5968109339407744,
      "grad_norm": 0.9645806550979614,
      "learning_rate": 1.6770386516384657e-05,
      "loss": 0.6262,
      "step": 4716
    },
    {
      "epoch": 0.59693748418122,
      "grad_norm": 0.9570954442024231,
      "learning_rate": 1.6769067498864653e-05,
      "loss": 0.5871,
      "step": 4717
    },
    {
      "epoch": 0.5970640344216654,
      "grad_norm": 1.0553431510925293,
      "learning_rate": 1.676774826394204e-05,
      "loss": 0.6273,
      "step": 4718
    },
    {
      "epoch": 0.5971905846621108,
      "grad_norm": 1.0170890092849731,
      "learning_rate": 1.6766428811659184e-05,
      "loss": 0.6325,
      "step": 4719
    },
    {
      "epoch": 0.5973171349025563,
      "grad_norm": 0.9743509888648987,
      "learning_rate": 1.6765109142058468e-05,
      "loss": 0.597,
      "step": 4720
    },
    {
      "epoch": 0.5974436851430017,
      "grad_norm": 1.0172795057296753,
      "learning_rate": 1.6763789255182262e-05,
      "loss": 0.6018,
      "step": 4721
    },
    {
      "epoch": 0.5975702353834472,
      "grad_norm": 0.9658122062683105,
      "learning_rate": 1.6762469151072974e-05,
      "loss": 0.6314,
      "step": 4722
    },
    {
      "epoch": 0.5976967856238927,
      "grad_norm": 1.085983395576477,
      "learning_rate": 1.676114882977299e-05,
      "loss": 0.6148,
      "step": 4723
    },
    {
      "epoch": 0.5978233358643381,
      "grad_norm": 1.0298150777816772,
      "learning_rate": 1.675982829132472e-05,
      "loss": 0.6042,
      "step": 4724
    },
    {
      "epoch": 0.5979498861047836,
      "grad_norm": 1.0496424436569214,
      "learning_rate": 1.6758507535770573e-05,
      "loss": 0.6481,
      "step": 4725
    },
    {
      "epoch": 0.598076436345229,
      "grad_norm": 1.1450345516204834,
      "learning_rate": 1.6757186563152977e-05,
      "loss": 0.6098,
      "step": 4726
    },
    {
      "epoch": 0.5982029865856745,
      "grad_norm": 0.9805414080619812,
      "learning_rate": 1.675586537351434e-05,
      "loss": 0.572,
      "step": 4727
    },
    {
      "epoch": 0.59832953682612,
      "grad_norm": 1.0748146772384644,
      "learning_rate": 1.6754543966897113e-05,
      "loss": 0.6428,
      "step": 4728
    },
    {
      "epoch": 0.5984560870665654,
      "grad_norm": 0.9672859907150269,
      "learning_rate": 1.6753222343343725e-05,
      "loss": 0.6581,
      "step": 4729
    },
    {
      "epoch": 0.5985826373070109,
      "grad_norm": 1.2098219394683838,
      "learning_rate": 1.675190050289662e-05,
      "loss": 0.6465,
      "step": 4730
    },
    {
      "epoch": 0.5987091875474564,
      "grad_norm": 1.0028499364852905,
      "learning_rate": 1.6750578445598265e-05,
      "loss": 0.6399,
      "step": 4731
    },
    {
      "epoch": 0.5988357377879018,
      "grad_norm": 1.228824496269226,
      "learning_rate": 1.674925617149111e-05,
      "loss": 0.6344,
      "step": 4732
    },
    {
      "epoch": 0.5989622880283473,
      "grad_norm": 1.1075236797332764,
      "learning_rate": 1.6747933680617624e-05,
      "loss": 0.6409,
      "step": 4733
    },
    {
      "epoch": 0.5990888382687927,
      "grad_norm": 0.9976152181625366,
      "learning_rate": 1.6746610973020283e-05,
      "loss": 0.596,
      "step": 4734
    },
    {
      "epoch": 0.5992153885092382,
      "grad_norm": 1.0799096822738647,
      "learning_rate": 1.674528804874157e-05,
      "loss": 0.6633,
      "step": 4735
    },
    {
      "epoch": 0.5993419387496837,
      "grad_norm": 1.0091720819473267,
      "learning_rate": 1.674396490782397e-05,
      "loss": 0.6854,
      "step": 4736
    },
    {
      "epoch": 0.5994684889901291,
      "grad_norm": 0.9992642402648926,
      "learning_rate": 1.674264155030998e-05,
      "loss": 0.5748,
      "step": 4737
    },
    {
      "epoch": 0.5995950392305746,
      "grad_norm": 0.9530751705169678,
      "learning_rate": 1.6741317976242102e-05,
      "loss": 0.6247,
      "step": 4738
    },
    {
      "epoch": 0.59972158947102,
      "grad_norm": 0.9815991520881653,
      "learning_rate": 1.673999418566285e-05,
      "loss": 0.6054,
      "step": 4739
    },
    {
      "epoch": 0.5998481397114654,
      "grad_norm": 0.9296523928642273,
      "learning_rate": 1.673867017861473e-05,
      "loss": 0.6152,
      "step": 4740
    },
    {
      "epoch": 0.5999746899519109,
      "grad_norm": 1.072317123413086,
      "learning_rate": 1.6737345955140277e-05,
      "loss": 0.6101,
      "step": 4741
    },
    {
      "epoch": 0.6001012401923563,
      "grad_norm": 1.1234105825424194,
      "learning_rate": 1.6736021515282012e-05,
      "loss": 0.6333,
      "step": 4742
    },
    {
      "epoch": 0.6002277904328018,
      "grad_norm": 1.1006985902786255,
      "learning_rate": 1.6734696859082475e-05,
      "loss": 0.6294,
      "step": 4743
    },
    {
      "epoch": 0.6003543406732473,
      "grad_norm": 1.0339913368225098,
      "learning_rate": 1.673337198658421e-05,
      "loss": 0.6426,
      "step": 4744
    },
    {
      "epoch": 0.6004808909136927,
      "grad_norm": 1.0489493608474731,
      "learning_rate": 1.673204689782977e-05,
      "loss": 0.6614,
      "step": 4745
    },
    {
      "epoch": 0.6006074411541382,
      "grad_norm": 1.0218915939331055,
      "learning_rate": 1.6730721592861714e-05,
      "loss": 0.6544,
      "step": 4746
    },
    {
      "epoch": 0.6007339913945836,
      "grad_norm": 0.8841778635978699,
      "learning_rate": 1.67293960717226e-05,
      "loss": 0.5648,
      "step": 4747
    },
    {
      "epoch": 0.6008605416350291,
      "grad_norm": 1.0486931800842285,
      "learning_rate": 1.672807033445501e-05,
      "loss": 0.6521,
      "step": 4748
    },
    {
      "epoch": 0.6009870918754746,
      "grad_norm": 1.0169973373413086,
      "learning_rate": 1.6726744381101513e-05,
      "loss": 0.6359,
      "step": 4749
    },
    {
      "epoch": 0.60111364211592,
      "grad_norm": 1.0021164417266846,
      "learning_rate": 1.67254182117047e-05,
      "loss": 0.6479,
      "step": 4750
    },
    {
      "epoch": 0.6012401923563655,
      "grad_norm": 0.977283775806427,
      "learning_rate": 1.6724091826307162e-05,
      "loss": 0.6523,
      "step": 4751
    },
    {
      "epoch": 0.6013667425968109,
      "grad_norm": 1.2997983694076538,
      "learning_rate": 1.67227652249515e-05,
      "loss": 0.6306,
      "step": 4752
    },
    {
      "epoch": 0.6014932928372564,
      "grad_norm": 1.0262815952301025,
      "learning_rate": 1.6721438407680318e-05,
      "loss": 0.603,
      "step": 4753
    },
    {
      "epoch": 0.6016198430777019,
      "grad_norm": 1.0270181894302368,
      "learning_rate": 1.6720111374536237e-05,
      "loss": 0.6399,
      "step": 4754
    },
    {
      "epoch": 0.6017463933181473,
      "grad_norm": 0.9594701528549194,
      "learning_rate": 1.6718784125561866e-05,
      "loss": 0.6153,
      "step": 4755
    },
    {
      "epoch": 0.6018729435585928,
      "grad_norm": 1.0037094354629517,
      "learning_rate": 1.6717456660799842e-05,
      "loss": 0.6799,
      "step": 4756
    },
    {
      "epoch": 0.6019994937990382,
      "grad_norm": 1.022078275680542,
      "learning_rate": 1.6716128980292793e-05,
      "loss": 0.6511,
      "step": 4757
    },
    {
      "epoch": 0.6021260440394837,
      "grad_norm": 1.107797384262085,
      "learning_rate": 1.6714801084083362e-05,
      "loss": 0.7013,
      "step": 4758
    },
    {
      "epoch": 0.6022525942799292,
      "grad_norm": 0.8909764289855957,
      "learning_rate": 1.6713472972214197e-05,
      "loss": 0.5939,
      "step": 4759
    },
    {
      "epoch": 0.6023791445203746,
      "grad_norm": 1.0383964776992798,
      "learning_rate": 1.6712144644727953e-05,
      "loss": 0.6503,
      "step": 4760
    },
    {
      "epoch": 0.60250569476082,
      "grad_norm": 1.0433731079101562,
      "learning_rate": 1.6710816101667298e-05,
      "loss": 0.6843,
      "step": 4761
    },
    {
      "epoch": 0.6026322450012656,
      "grad_norm": 0.9835220575332642,
      "learning_rate": 1.6709487343074887e-05,
      "loss": 0.6167,
      "step": 4762
    },
    {
      "epoch": 0.6027587952417109,
      "grad_norm": 1.0392268896102905,
      "learning_rate": 1.6708158368993412e-05,
      "loss": 0.6598,
      "step": 4763
    },
    {
      "epoch": 0.6028853454821564,
      "grad_norm": 1.0205026865005493,
      "learning_rate": 1.6706829179465546e-05,
      "loss": 0.6612,
      "step": 4764
    },
    {
      "epoch": 0.6030118957226018,
      "grad_norm": 1.6720499992370605,
      "learning_rate": 1.6705499774533983e-05,
      "loss": 0.6135,
      "step": 4765
    },
    {
      "epoch": 0.6031384459630473,
      "grad_norm": 0.9761236310005188,
      "learning_rate": 1.6704170154241413e-05,
      "loss": 0.6254,
      "step": 4766
    },
    {
      "epoch": 0.6032649962034928,
      "grad_norm": 0.9958686232566833,
      "learning_rate": 1.6702840318630547e-05,
      "loss": 0.6698,
      "step": 4767
    },
    {
      "epoch": 0.6033915464439382,
      "grad_norm": 1.054896593093872,
      "learning_rate": 1.670151026774409e-05,
      "loss": 0.67,
      "step": 4768
    },
    {
      "epoch": 0.6035180966843837,
      "grad_norm": 1.0733072757720947,
      "learning_rate": 1.670018000162476e-05,
      "loss": 0.6646,
      "step": 4769
    },
    {
      "epoch": 0.6036446469248291,
      "grad_norm": 1.0170780420303345,
      "learning_rate": 1.669884952031529e-05,
      "loss": 0.6719,
      "step": 4770
    },
    {
      "epoch": 0.6037711971652746,
      "grad_norm": 1.0625476837158203,
      "learning_rate": 1.66975188238584e-05,
      "loss": 0.6687,
      "step": 4771
    },
    {
      "epoch": 0.6038977474057201,
      "grad_norm": 0.9420751929283142,
      "learning_rate": 1.669618791229683e-05,
      "loss": 0.6156,
      "step": 4772
    },
    {
      "epoch": 0.6040242976461655,
      "grad_norm": 0.9414976239204407,
      "learning_rate": 1.669485678567333e-05,
      "loss": 0.6444,
      "step": 4773
    },
    {
      "epoch": 0.604150847886611,
      "grad_norm": 0.9412595629692078,
      "learning_rate": 1.669352544403065e-05,
      "loss": 0.6335,
      "step": 4774
    },
    {
      "epoch": 0.6042773981270564,
      "grad_norm": 0.9801754355430603,
      "learning_rate": 1.6692193887411547e-05,
      "loss": 0.621,
      "step": 4775
    },
    {
      "epoch": 0.6044039483675019,
      "grad_norm": 0.9452688694000244,
      "learning_rate": 1.669086211585879e-05,
      "loss": 0.6052,
      "step": 4776
    },
    {
      "epoch": 0.6045304986079474,
      "grad_norm": 0.9858344793319702,
      "learning_rate": 1.6689530129415144e-05,
      "loss": 0.6518,
      "step": 4777
    },
    {
      "epoch": 0.6046570488483928,
      "grad_norm": 0.9401479959487915,
      "learning_rate": 1.6688197928123395e-05,
      "loss": 0.6438,
      "step": 4778
    },
    {
      "epoch": 0.6047835990888383,
      "grad_norm": 0.9964611530303955,
      "learning_rate": 1.668686551202633e-05,
      "loss": 0.6508,
      "step": 4779
    },
    {
      "epoch": 0.6049101493292838,
      "grad_norm": 1.0031343698501587,
      "learning_rate": 1.6685532881166737e-05,
      "loss": 0.6146,
      "step": 4780
    },
    {
      "epoch": 0.6050366995697292,
      "grad_norm": 0.996946394443512,
      "learning_rate": 1.668420003558742e-05,
      "loss": 0.6194,
      "step": 4781
    },
    {
      "epoch": 0.6051632498101747,
      "grad_norm": 1.1352951526641846,
      "learning_rate": 1.668286697533119e-05,
      "loss": 0.6709,
      "step": 4782
    },
    {
      "epoch": 0.60528980005062,
      "grad_norm": 1.0673720836639404,
      "learning_rate": 1.6681533700440855e-05,
      "loss": 0.6256,
      "step": 4783
    },
    {
      "epoch": 0.6054163502910656,
      "grad_norm": 1.1185474395751953,
      "learning_rate": 1.6680200210959237e-05,
      "loss": 0.5854,
      "step": 4784
    },
    {
      "epoch": 0.605542900531511,
      "grad_norm": 0.9990734457969666,
      "learning_rate": 1.6678866506929162e-05,
      "loss": 0.6236,
      "step": 4785
    },
    {
      "epoch": 0.6056694507719564,
      "grad_norm": 1.0021679401397705,
      "learning_rate": 1.6677532588393468e-05,
      "loss": 0.6393,
      "step": 4786
    },
    {
      "epoch": 0.6057960010124019,
      "grad_norm": 0.9878848791122437,
      "learning_rate": 1.6676198455394995e-05,
      "loss": 0.6315,
      "step": 4787
    },
    {
      "epoch": 0.6059225512528473,
      "grad_norm": 1.018678903579712,
      "learning_rate": 1.6674864107976592e-05,
      "loss": 0.5971,
      "step": 4788
    },
    {
      "epoch": 0.6060491014932928,
      "grad_norm": 0.9773340821266174,
      "learning_rate": 1.667352954618112e-05,
      "loss": 0.618,
      "step": 4789
    },
    {
      "epoch": 0.6061756517337383,
      "grad_norm": 1.0371479988098145,
      "learning_rate": 1.667219477005143e-05,
      "loss": 0.6435,
      "step": 4790
    },
    {
      "epoch": 0.6063022019741837,
      "grad_norm": 1.6560168266296387,
      "learning_rate": 1.6670859779630396e-05,
      "loss": 0.6254,
      "step": 4791
    },
    {
      "epoch": 0.6064287522146292,
      "grad_norm": 1.021414875984192,
      "learning_rate": 1.66695245749609e-05,
      "loss": 0.6073,
      "step": 4792
    },
    {
      "epoch": 0.6065553024550747,
      "grad_norm": 1.0684680938720703,
      "learning_rate": 1.666818915608581e-05,
      "loss": 0.6485,
      "step": 4793
    },
    {
      "epoch": 0.6066818526955201,
      "grad_norm": 0.9575248956680298,
      "learning_rate": 1.6666853523048035e-05,
      "loss": 0.6055,
      "step": 4794
    },
    {
      "epoch": 0.6068084029359656,
      "grad_norm": 0.8927161693572998,
      "learning_rate": 1.6665517675890457e-05,
      "loss": 0.6119,
      "step": 4795
    },
    {
      "epoch": 0.606934953176411,
      "grad_norm": 0.9903932809829712,
      "learning_rate": 1.6664181614655984e-05,
      "loss": 0.6275,
      "step": 4796
    },
    {
      "epoch": 0.6070615034168565,
      "grad_norm": 1.0043087005615234,
      "learning_rate": 1.666284533938753e-05,
      "loss": 0.6906,
      "step": 4797
    },
    {
      "epoch": 0.607188053657302,
      "grad_norm": 1.0462186336517334,
      "learning_rate": 1.6661508850128004e-05,
      "loss": 0.6591,
      "step": 4798
    },
    {
      "epoch": 0.6073146038977474,
      "grad_norm": 0.9362214207649231,
      "learning_rate": 1.6660172146920338e-05,
      "loss": 0.623,
      "step": 4799
    },
    {
      "epoch": 0.6074411541381929,
      "grad_norm": 1.0390158891677856,
      "learning_rate": 1.6658835229807458e-05,
      "loss": 0.653,
      "step": 4800
    },
    {
      "epoch": 0.6075677043786383,
      "grad_norm": 0.9599921107292175,
      "learning_rate": 1.6657498098832308e-05,
      "loss": 0.6188,
      "step": 4801
    },
    {
      "epoch": 0.6076942546190838,
      "grad_norm": 1.022635579109192,
      "learning_rate": 1.6656160754037826e-05,
      "loss": 0.6451,
      "step": 4802
    },
    {
      "epoch": 0.6078208048595293,
      "grad_norm": 1.0708481073379517,
      "learning_rate": 1.6654823195466966e-05,
      "loss": 0.6474,
      "step": 4803
    },
    {
      "epoch": 0.6079473550999747,
      "grad_norm": 1.0205283164978027,
      "learning_rate": 1.6653485423162686e-05,
      "loss": 0.5969,
      "step": 4804
    },
    {
      "epoch": 0.6080739053404202,
      "grad_norm": 1.030654788017273,
      "learning_rate": 1.6652147437167952e-05,
      "loss": 0.6338,
      "step": 4805
    },
    {
      "epoch": 0.6082004555808656,
      "grad_norm": 0.9406358003616333,
      "learning_rate": 1.665080923752574e-05,
      "loss": 0.6109,
      "step": 4806
    },
    {
      "epoch": 0.608327005821311,
      "grad_norm": 1.0316277742385864,
      "learning_rate": 1.664947082427902e-05,
      "loss": 0.5988,
      "step": 4807
    },
    {
      "epoch": 0.6084535560617566,
      "grad_norm": 0.9710107445716858,
      "learning_rate": 1.6648132197470785e-05,
      "loss": 0.6366,
      "step": 4808
    },
    {
      "epoch": 0.6085801063022019,
      "grad_norm": 1.026525855064392,
      "learning_rate": 1.6646793357144028e-05,
      "loss": 0.6091,
      "step": 4809
    },
    {
      "epoch": 0.6087066565426474,
      "grad_norm": 0.9457966089248657,
      "learning_rate": 1.6645454303341742e-05,
      "loss": 0.5986,
      "step": 4810
    },
    {
      "epoch": 0.6088332067830929,
      "grad_norm": 1.0668805837631226,
      "learning_rate": 1.6644115036106944e-05,
      "loss": 0.7079,
      "step": 4811
    },
    {
      "epoch": 0.6089597570235383,
      "grad_norm": 0.9903569221496582,
      "learning_rate": 1.664277555548264e-05,
      "loss": 0.6011,
      "step": 4812
    },
    {
      "epoch": 0.6090863072639838,
      "grad_norm": 0.9936301112174988,
      "learning_rate": 1.664143586151185e-05,
      "loss": 0.615,
      "step": 4813
    },
    {
      "epoch": 0.6092128575044292,
      "grad_norm": 0.9646633267402649,
      "learning_rate": 1.6640095954237597e-05,
      "loss": 0.6666,
      "step": 4814
    },
    {
      "epoch": 0.6093394077448747,
      "grad_norm": 1.0340068340301514,
      "learning_rate": 1.663875583370293e-05,
      "loss": 0.6645,
      "step": 4815
    },
    {
      "epoch": 0.6094659579853202,
      "grad_norm": 0.9950717091560364,
      "learning_rate": 1.6637415499950872e-05,
      "loss": 0.5809,
      "step": 4816
    },
    {
      "epoch": 0.6095925082257656,
      "grad_norm": 1.0122852325439453,
      "learning_rate": 1.6636074953024484e-05,
      "loss": 0.5978,
      "step": 4817
    },
    {
      "epoch": 0.6097190584662111,
      "grad_norm": 0.9762634038925171,
      "learning_rate": 1.6634734192966816e-05,
      "loss": 0.6067,
      "step": 4818
    },
    {
      "epoch": 0.6098456087066565,
      "grad_norm": 0.940963864326477,
      "learning_rate": 1.6633393219820926e-05,
      "loss": 0.6435,
      "step": 4819
    },
    {
      "epoch": 0.609972158947102,
      "grad_norm": 1.0982623100280762,
      "learning_rate": 1.6632052033629886e-05,
      "loss": 0.5948,
      "step": 4820
    },
    {
      "epoch": 0.6100987091875475,
      "grad_norm": 0.9806229472160339,
      "learning_rate": 1.663071063443677e-05,
      "loss": 0.6401,
      "step": 4821
    },
    {
      "epoch": 0.6102252594279929,
      "grad_norm": 1.2599190473556519,
      "learning_rate": 1.6629369022284656e-05,
      "loss": 0.658,
      "step": 4822
    },
    {
      "epoch": 0.6103518096684384,
      "grad_norm": 1.013060212135315,
      "learning_rate": 1.6628027197216637e-05,
      "loss": 0.6092,
      "step": 4823
    },
    {
      "epoch": 0.6104783599088838,
      "grad_norm": 1.0160080194473267,
      "learning_rate": 1.662668515927581e-05,
      "loss": 0.6027,
      "step": 4824
    },
    {
      "epoch": 0.6106049101493293,
      "grad_norm": 1.1324126720428467,
      "learning_rate": 1.662534290850528e-05,
      "loss": 0.601,
      "step": 4825
    },
    {
      "epoch": 0.6107314603897748,
      "grad_norm": 0.9979002475738525,
      "learning_rate": 1.662400044494814e-05,
      "loss": 0.6425,
      "step": 4826
    },
    {
      "epoch": 0.6108580106302202,
      "grad_norm": 1.0338596105575562,
      "learning_rate": 1.6622657768647524e-05,
      "loss": 0.6262,
      "step": 4827
    },
    {
      "epoch": 0.6109845608706657,
      "grad_norm": 1.0009208917617798,
      "learning_rate": 1.6621314879646546e-05,
      "loss": 0.6433,
      "step": 4828
    },
    {
      "epoch": 0.6111111111111112,
      "grad_norm": 0.9916245341300964,
      "learning_rate": 1.6619971777988338e-05,
      "loss": 0.6536,
      "step": 4829
    },
    {
      "epoch": 0.6112376613515565,
      "grad_norm": 1.0608025789260864,
      "learning_rate": 1.6618628463716038e-05,
      "loss": 0.6263,
      "step": 4830
    },
    {
      "epoch": 0.611364211592002,
      "grad_norm": 0.9879071712493896,
      "learning_rate": 1.6617284936872782e-05,
      "loss": 0.6093,
      "step": 4831
    },
    {
      "epoch": 0.6114907618324474,
      "grad_norm": 0.9867640733718872,
      "learning_rate": 1.661594119750173e-05,
      "loss": 0.6201,
      "step": 4832
    },
    {
      "epoch": 0.6116173120728929,
      "grad_norm": 0.959924042224884,
      "learning_rate": 1.6614597245646033e-05,
      "loss": 0.6092,
      "step": 4833
    },
    {
      "epoch": 0.6117438623133384,
      "grad_norm": 0.9721276760101318,
      "learning_rate": 1.661325308134886e-05,
      "loss": 0.6102,
      "step": 4834
    },
    {
      "epoch": 0.6118704125537838,
      "grad_norm": 0.9666483998298645,
      "learning_rate": 1.661190870465337e-05,
      "loss": 0.6226,
      "step": 4835
    },
    {
      "epoch": 0.6119969627942293,
      "grad_norm": 0.9747925996780396,
      "learning_rate": 1.6610564115602756e-05,
      "loss": 0.6761,
      "step": 4836
    },
    {
      "epoch": 0.6121235130346747,
      "grad_norm": 0.9537906050682068,
      "learning_rate": 1.660921931424019e-05,
      "loss": 0.6701,
      "step": 4837
    },
    {
      "epoch": 0.6122500632751202,
      "grad_norm": 1.061625361442566,
      "learning_rate": 1.6607874300608872e-05,
      "loss": 0.6086,
      "step": 4838
    },
    {
      "epoch": 0.6123766135155657,
      "grad_norm": 0.9405450224876404,
      "learning_rate": 1.660652907475199e-05,
      "loss": 0.6479,
      "step": 4839
    },
    {
      "epoch": 0.6125031637560111,
      "grad_norm": 1.0495742559432983,
      "learning_rate": 1.6605183636712757e-05,
      "loss": 0.6409,
      "step": 4840
    },
    {
      "epoch": 0.6126297139964566,
      "grad_norm": 1.1838898658752441,
      "learning_rate": 1.660383798653438e-05,
      "loss": 0.6564,
      "step": 4841
    },
    {
      "epoch": 0.6127562642369021,
      "grad_norm": 0.979337751865387,
      "learning_rate": 1.6602492124260083e-05,
      "loss": 0.6248,
      "step": 4842
    },
    {
      "epoch": 0.6128828144773475,
      "grad_norm": 0.9912594556808472,
      "learning_rate": 1.6601146049933085e-05,
      "loss": 0.6515,
      "step": 4843
    },
    {
      "epoch": 0.613009364717793,
      "grad_norm": 0.9925404191017151,
      "learning_rate": 1.659979976359662e-05,
      "loss": 0.6367,
      "step": 4844
    },
    {
      "epoch": 0.6131359149582384,
      "grad_norm": 0.9923914074897766,
      "learning_rate": 1.659845326529393e-05,
      "loss": 0.6279,
      "step": 4845
    },
    {
      "epoch": 0.6132624651986839,
      "grad_norm": 0.912603497505188,
      "learning_rate": 1.6597106555068253e-05,
      "loss": 0.6675,
      "step": 4846
    },
    {
      "epoch": 0.6133890154391294,
      "grad_norm": 1.0055608749389648,
      "learning_rate": 1.6595759632962848e-05,
      "loss": 0.6294,
      "step": 4847
    },
    {
      "epoch": 0.6135155656795748,
      "grad_norm": 0.9515992403030396,
      "learning_rate": 1.6594412499020975e-05,
      "loss": 0.6347,
      "step": 4848
    },
    {
      "epoch": 0.6136421159200203,
      "grad_norm": 0.988907516002655,
      "learning_rate": 1.6593065153285892e-05,
      "loss": 0.6448,
      "step": 4849
    },
    {
      "epoch": 0.6137686661604657,
      "grad_norm": 1.0385364294052124,
      "learning_rate": 1.659171759580088e-05,
      "loss": 0.6781,
      "step": 4850
    },
    {
      "epoch": 0.6138952164009112,
      "grad_norm": 0.9847390055656433,
      "learning_rate": 1.6590369826609216e-05,
      "loss": 0.6221,
      "step": 4851
    },
    {
      "epoch": 0.6140217666413567,
      "grad_norm": 0.9737639427185059,
      "learning_rate": 1.6589021845754186e-05,
      "loss": 0.6044,
      "step": 4852
    },
    {
      "epoch": 0.614148316881802,
      "grad_norm": 0.9844863414764404,
      "learning_rate": 1.658767365327908e-05,
      "loss": 0.6551,
      "step": 4853
    },
    {
      "epoch": 0.6142748671222475,
      "grad_norm": 0.9980952143669128,
      "learning_rate": 1.6586325249227203e-05,
      "loss": 0.599,
      "step": 4854
    },
    {
      "epoch": 0.6144014173626929,
      "grad_norm": 0.9885138869285583,
      "learning_rate": 1.6584976633641864e-05,
      "loss": 0.5866,
      "step": 4855
    },
    {
      "epoch": 0.6145279676031384,
      "grad_norm": 0.9742646217346191,
      "learning_rate": 1.6583627806566367e-05,
      "loss": 0.6148,
      "step": 4856
    },
    {
      "epoch": 0.6146545178435839,
      "grad_norm": 0.974466860294342,
      "learning_rate": 1.658227876804404e-05,
      "loss": 0.6157,
      "step": 4857
    },
    {
      "epoch": 0.6147810680840293,
      "grad_norm": 1.0539014339447021,
      "learning_rate": 1.658092951811821e-05,
      "loss": 0.6108,
      "step": 4858
    },
    {
      "epoch": 0.6149076183244748,
      "grad_norm": 1.0300445556640625,
      "learning_rate": 1.6579580056832206e-05,
      "loss": 0.6241,
      "step": 4859
    },
    {
      "epoch": 0.6150341685649203,
      "grad_norm": 1.019544243812561,
      "learning_rate": 1.6578230384229373e-05,
      "loss": 0.6187,
      "step": 4860
    },
    {
      "epoch": 0.6151607188053657,
      "grad_norm": 1.2183488607406616,
      "learning_rate": 1.6576880500353057e-05,
      "loss": 0.6171,
      "step": 4861
    },
    {
      "epoch": 0.6152872690458112,
      "grad_norm": 0.930145800113678,
      "learning_rate": 1.6575530405246613e-05,
      "loss": 0.5949,
      "step": 4862
    },
    {
      "epoch": 0.6154138192862566,
      "grad_norm": 1.2037031650543213,
      "learning_rate": 1.65741800989534e-05,
      "loss": 0.5986,
      "step": 4863
    },
    {
      "epoch": 0.6155403695267021,
      "grad_norm": 1.1049782037734985,
      "learning_rate": 1.6572829581516793e-05,
      "loss": 0.6098,
      "step": 4864
    },
    {
      "epoch": 0.6156669197671476,
      "grad_norm": 0.9650065302848816,
      "learning_rate": 1.6571478852980157e-05,
      "loss": 0.5961,
      "step": 4865
    },
    {
      "epoch": 0.615793470007593,
      "grad_norm": 1.1871880292892456,
      "learning_rate": 1.657012791338688e-05,
      "loss": 0.6341,
      "step": 4866
    },
    {
      "epoch": 0.6159200202480385,
      "grad_norm": 1.514570951461792,
      "learning_rate": 1.6568776762780343e-05,
      "loss": 0.6199,
      "step": 4867
    },
    {
      "epoch": 0.6160465704884839,
      "grad_norm": 0.9736329317092896,
      "learning_rate": 1.656742540120395e-05,
      "loss": 0.6165,
      "step": 4868
    },
    {
      "epoch": 0.6161731207289294,
      "grad_norm": 0.9051958918571472,
      "learning_rate": 1.6566073828701098e-05,
      "loss": 0.6144,
      "step": 4869
    },
    {
      "epoch": 0.6162996709693749,
      "grad_norm": 0.9515475630760193,
      "learning_rate": 1.6564722045315196e-05,
      "loss": 0.6416,
      "step": 4870
    },
    {
      "epoch": 0.6164262212098203,
      "grad_norm": 1.0310500860214233,
      "learning_rate": 1.656337005108966e-05,
      "loss": 0.6364,
      "step": 4871
    },
    {
      "epoch": 0.6165527714502658,
      "grad_norm": 1.036171793937683,
      "learning_rate": 1.656201784606791e-05,
      "loss": 0.6213,
      "step": 4872
    },
    {
      "epoch": 0.6166793216907112,
      "grad_norm": 1.5313293933868408,
      "learning_rate": 1.6560665430293377e-05,
      "loss": 0.6318,
      "step": 4873
    },
    {
      "epoch": 0.6168058719311567,
      "grad_norm": 0.9640253782272339,
      "learning_rate": 1.65593128038095e-05,
      "loss": 0.6274,
      "step": 4874
    },
    {
      "epoch": 0.6169324221716022,
      "grad_norm": 1.0137228965759277,
      "learning_rate": 1.6557959966659713e-05,
      "loss": 0.6546,
      "step": 4875
    },
    {
      "epoch": 0.6170589724120475,
      "grad_norm": 0.9771240949630737,
      "learning_rate": 1.655660691888747e-05,
      "loss": 0.6551,
      "step": 4876
    },
    {
      "epoch": 0.617185522652493,
      "grad_norm": 1.0374138355255127,
      "learning_rate": 1.6555253660536227e-05,
      "loss": 0.6564,
      "step": 4877
    },
    {
      "epoch": 0.6173120728929385,
      "grad_norm": 0.9126228094100952,
      "learning_rate": 1.655390019164945e-05,
      "loss": 0.6268,
      "step": 4878
    },
    {
      "epoch": 0.6174386231333839,
      "grad_norm": 0.9785920977592468,
      "learning_rate": 1.65525465122706e-05,
      "loss": 0.6123,
      "step": 4879
    },
    {
      "epoch": 0.6175651733738294,
      "grad_norm": 1.027802586555481,
      "learning_rate": 1.655119262244316e-05,
      "loss": 0.6331,
      "step": 4880
    },
    {
      "epoch": 0.6176917236142748,
      "grad_norm": 0.9339941143989563,
      "learning_rate": 1.654983852221061e-05,
      "loss": 0.6602,
      "step": 4881
    },
    {
      "epoch": 0.6178182738547203,
      "grad_norm": 0.9431333541870117,
      "learning_rate": 1.6548484211616442e-05,
      "loss": 0.5962,
      "step": 4882
    },
    {
      "epoch": 0.6179448240951658,
      "grad_norm": 0.9352684617042542,
      "learning_rate": 1.6547129690704152e-05,
      "loss": 0.6069,
      "step": 4883
    },
    {
      "epoch": 0.6180713743356112,
      "grad_norm": 0.9719752073287964,
      "learning_rate": 1.654577495951724e-05,
      "loss": 0.6242,
      "step": 4884
    },
    {
      "epoch": 0.6181979245760567,
      "grad_norm": 1.0175151824951172,
      "learning_rate": 1.6544420018099222e-05,
      "loss": 0.5881,
      "step": 4885
    },
    {
      "epoch": 0.6183244748165021,
      "grad_norm": 1.1137593984603882,
      "learning_rate": 1.6543064866493606e-05,
      "loss": 0.6543,
      "step": 4886
    },
    {
      "epoch": 0.6184510250569476,
      "grad_norm": 0.9607265591621399,
      "learning_rate": 1.6541709504743924e-05,
      "loss": 0.5663,
      "step": 4887
    },
    {
      "epoch": 0.6185775752973931,
      "grad_norm": 1.0422559976577759,
      "learning_rate": 1.6540353932893706e-05,
      "loss": 0.6589,
      "step": 4888
    },
    {
      "epoch": 0.6187041255378385,
      "grad_norm": 1.0261383056640625,
      "learning_rate": 1.653899815098648e-05,
      "loss": 0.6198,
      "step": 4889
    },
    {
      "epoch": 0.618830675778284,
      "grad_norm": 0.979456901550293,
      "learning_rate": 1.6537642159065795e-05,
      "loss": 0.5826,
      "step": 4890
    },
    {
      "epoch": 0.6189572260187295,
      "grad_norm": 1.0943048000335693,
      "learning_rate": 1.6536285957175208e-05,
      "loss": 0.6398,
      "step": 4891
    },
    {
      "epoch": 0.6190837762591749,
      "grad_norm": 1.0625675916671753,
      "learning_rate": 1.653492954535827e-05,
      "loss": 0.6506,
      "step": 4892
    },
    {
      "epoch": 0.6192103264996204,
      "grad_norm": 0.9924008846282959,
      "learning_rate": 1.653357292365854e-05,
      "loss": 0.627,
      "step": 4893
    },
    {
      "epoch": 0.6193368767400658,
      "grad_norm": 1.1224912405014038,
      "learning_rate": 1.6532216092119598e-05,
      "loss": 0.6104,
      "step": 4894
    },
    {
      "epoch": 0.6194634269805113,
      "grad_norm": 0.9155873656272888,
      "learning_rate": 1.6530859050785015e-05,
      "loss": 0.6453,
      "step": 4895
    },
    {
      "epoch": 0.6195899772209568,
      "grad_norm": 1.0228205919265747,
      "learning_rate": 1.652950179969838e-05,
      "loss": 0.6173,
      "step": 4896
    },
    {
      "epoch": 0.6197165274614022,
      "grad_norm": 0.9285111427307129,
      "learning_rate": 1.6528144338903282e-05,
      "loss": 0.6051,
      "step": 4897
    },
    {
      "epoch": 0.6198430777018477,
      "grad_norm": 0.9602589011192322,
      "learning_rate": 1.652678666844332e-05,
      "loss": 0.6518,
      "step": 4898
    },
    {
      "epoch": 0.619969627942293,
      "grad_norm": 0.9714701771736145,
      "learning_rate": 1.652542878836209e-05,
      "loss": 0.5982,
      "step": 4899
    },
    {
      "epoch": 0.6200961781827385,
      "grad_norm": 1.0412648916244507,
      "learning_rate": 1.6524070698703215e-05,
      "loss": 0.6696,
      "step": 4900
    },
    {
      "epoch": 0.620222728423184,
      "grad_norm": 0.9910541772842407,
      "learning_rate": 1.6522712399510306e-05,
      "loss": 0.6408,
      "step": 4901
    },
    {
      "epoch": 0.6203492786636294,
      "grad_norm": 1.0025726556777954,
      "learning_rate": 1.652135389082699e-05,
      "loss": 0.6173,
      "step": 4902
    },
    {
      "epoch": 0.6204758289040749,
      "grad_norm": 1.0898683071136475,
      "learning_rate": 1.65199951726969e-05,
      "loss": 0.6306,
      "step": 4903
    },
    {
      "epoch": 0.6206023791445203,
      "grad_norm": 0.941952645778656,
      "learning_rate": 1.6518636245163668e-05,
      "loss": 0.6127,
      "step": 4904
    },
    {
      "epoch": 0.6207289293849658,
      "grad_norm": 0.9697741270065308,
      "learning_rate": 1.6517277108270945e-05,
      "loss": 0.5803,
      "step": 4905
    },
    {
      "epoch": 0.6208554796254113,
      "grad_norm": 1.0207839012145996,
      "learning_rate": 1.6515917762062375e-05,
      "loss": 0.6714,
      "step": 4906
    },
    {
      "epoch": 0.6209820298658567,
      "grad_norm": 0.9884211421012878,
      "learning_rate": 1.6514558206581627e-05,
      "loss": 0.6506,
      "step": 4907
    },
    {
      "epoch": 0.6211085801063022,
      "grad_norm": 1.0799264907836914,
      "learning_rate": 1.6513198441872356e-05,
      "loss": 0.6813,
      "step": 4908
    },
    {
      "epoch": 0.6212351303467477,
      "grad_norm": 1.0320783853530884,
      "learning_rate": 1.651183846797824e-05,
      "loss": 0.6063,
      "step": 4909
    },
    {
      "epoch": 0.6213616805871931,
      "grad_norm": 1.1446582078933716,
      "learning_rate": 1.6510478284942954e-05,
      "loss": 0.6778,
      "step": 4910
    },
    {
      "epoch": 0.6214882308276386,
      "grad_norm": 1.1068090200424194,
      "learning_rate": 1.6509117892810186e-05,
      "loss": 0.6341,
      "step": 4911
    },
    {
      "epoch": 0.621614781068084,
      "grad_norm": 1.2374937534332275,
      "learning_rate": 1.6507757291623622e-05,
      "loss": 0.6405,
      "step": 4912
    },
    {
      "epoch": 0.6217413313085295,
      "grad_norm": 0.9411082863807678,
      "learning_rate": 1.650639648142697e-05,
      "loss": 0.6341,
      "step": 4913
    },
    {
      "epoch": 0.621867881548975,
      "grad_norm": 1.1429007053375244,
      "learning_rate": 1.6505035462263922e-05,
      "loss": 0.6673,
      "step": 4914
    },
    {
      "epoch": 0.6219944317894204,
      "grad_norm": 1.0249228477478027,
      "learning_rate": 1.6503674234178204e-05,
      "loss": 0.5886,
      "step": 4915
    },
    {
      "epoch": 0.6221209820298659,
      "grad_norm": 1.0630288124084473,
      "learning_rate": 1.6502312797213528e-05,
      "loss": 0.5954,
      "step": 4916
    },
    {
      "epoch": 0.6222475322703113,
      "grad_norm": 1.0934251546859741,
      "learning_rate": 1.6500951151413617e-05,
      "loss": 0.6005,
      "step": 4917
    },
    {
      "epoch": 0.6223740825107568,
      "grad_norm": 0.9513339996337891,
      "learning_rate": 1.649958929682221e-05,
      "loss": 0.6242,
      "step": 4918
    },
    {
      "epoch": 0.6225006327512023,
      "grad_norm": 0.9983115792274475,
      "learning_rate": 1.649822723348304e-05,
      "loss": 0.604,
      "step": 4919
    },
    {
      "epoch": 0.6226271829916477,
      "grad_norm": 1.1461880207061768,
      "learning_rate": 1.649686496143985e-05,
      "loss": 0.6089,
      "step": 4920
    },
    {
      "epoch": 0.6227537332320932,
      "grad_norm": 1.0438355207443237,
      "learning_rate": 1.64955024807364e-05,
      "loss": 0.6345,
      "step": 4921
    },
    {
      "epoch": 0.6228802834725385,
      "grad_norm": 1.0786542892456055,
      "learning_rate": 1.6494139791416447e-05,
      "loss": 0.6257,
      "step": 4922
    },
    {
      "epoch": 0.623006833712984,
      "grad_norm": 0.9620309472084045,
      "learning_rate": 1.649277689352375e-05,
      "loss": 0.6038,
      "step": 4923
    },
    {
      "epoch": 0.6231333839534295,
      "grad_norm": 1.31256902217865,
      "learning_rate": 1.6491413787102088e-05,
      "loss": 0.6325,
      "step": 4924
    },
    {
      "epoch": 0.6232599341938749,
      "grad_norm": 1.0138219594955444,
      "learning_rate": 1.6490050472195237e-05,
      "loss": 0.6485,
      "step": 4925
    },
    {
      "epoch": 0.6233864844343204,
      "grad_norm": 1.1632968187332153,
      "learning_rate": 1.6488686948846987e-05,
      "loss": 0.5969,
      "step": 4926
    },
    {
      "epoch": 0.6235130346747659,
      "grad_norm": 1.0883063077926636,
      "learning_rate": 1.6487323217101124e-05,
      "loss": 0.6101,
      "step": 4927
    },
    {
      "epoch": 0.6236395849152113,
      "grad_norm": 0.9873179793357849,
      "learning_rate": 1.648595927700145e-05,
      "loss": 0.6346,
      "step": 4928
    },
    {
      "epoch": 0.6237661351556568,
      "grad_norm": 1.1546977758407593,
      "learning_rate": 1.6484595128591773e-05,
      "loss": 0.6686,
      "step": 4929
    },
    {
      "epoch": 0.6238926853961022,
      "grad_norm": 0.9527599811553955,
      "learning_rate": 1.6483230771915904e-05,
      "loss": 0.6039,
      "step": 4930
    },
    {
      "epoch": 0.6240192356365477,
      "grad_norm": 1.1728280782699585,
      "learning_rate": 1.6481866207017657e-05,
      "loss": 0.6268,
      "step": 4931
    },
    {
      "epoch": 0.6241457858769932,
      "grad_norm": 1.0418123006820679,
      "learning_rate": 1.6480501433940866e-05,
      "loss": 0.5875,
      "step": 4932
    },
    {
      "epoch": 0.6242723361174386,
      "grad_norm": 1.0595701932907104,
      "learning_rate": 1.647913645272936e-05,
      "loss": 0.6043,
      "step": 4933
    },
    {
      "epoch": 0.6243988863578841,
      "grad_norm": 1.156853437423706,
      "learning_rate": 1.6477771263426976e-05,
      "loss": 0.6249,
      "step": 4934
    },
    {
      "epoch": 0.6245254365983295,
      "grad_norm": 1.0073710680007935,
      "learning_rate": 1.6476405866077562e-05,
      "loss": 0.6222,
      "step": 4935
    },
    {
      "epoch": 0.624651986838775,
      "grad_norm": 1.067483901977539,
      "learning_rate": 1.647504026072497e-05,
      "loss": 0.6112,
      "step": 4936
    },
    {
      "epoch": 0.6247785370792205,
      "grad_norm": 1.1318429708480835,
      "learning_rate": 1.6473674447413062e-05,
      "loss": 0.6495,
      "step": 4937
    },
    {
      "epoch": 0.6249050873196659,
      "grad_norm": 0.9736796021461487,
      "learning_rate": 1.64723084261857e-05,
      "loss": 0.5963,
      "step": 4938
    },
    {
      "epoch": 0.6250316375601114,
      "grad_norm": 0.952562689781189,
      "learning_rate": 1.647094219708676e-05,
      "loss": 0.6089,
      "step": 4939
    },
    {
      "epoch": 0.6251581878005569,
      "grad_norm": 0.967103123664856,
      "learning_rate": 1.646957576016012e-05,
      "loss": 0.6254,
      "step": 4940
    },
    {
      "epoch": 0.6252847380410023,
      "grad_norm": 0.9723150730133057,
      "learning_rate": 1.646820911544966e-05,
      "loss": 0.6095,
      "step": 4941
    },
    {
      "epoch": 0.6254112882814478,
      "grad_norm": 0.9614007472991943,
      "learning_rate": 1.6466842262999283e-05,
      "loss": 0.6289,
      "step": 4942
    },
    {
      "epoch": 0.6255378385218932,
      "grad_norm": 1.0257490873336792,
      "learning_rate": 1.646547520285288e-05,
      "loss": 0.6199,
      "step": 4943
    },
    {
      "epoch": 0.6256643887623387,
      "grad_norm": 0.961565375328064,
      "learning_rate": 1.6464107935054364e-05,
      "loss": 0.6481,
      "step": 4944
    },
    {
      "epoch": 0.6257909390027842,
      "grad_norm": 1.0731682777404785,
      "learning_rate": 1.646274045964764e-05,
      "loss": 0.6375,
      "step": 4945
    },
    {
      "epoch": 0.6259174892432295,
      "grad_norm": 0.9477475881576538,
      "learning_rate": 1.6461372776676637e-05,
      "loss": 0.6142,
      "step": 4946
    },
    {
      "epoch": 0.626044039483675,
      "grad_norm": 0.9621178507804871,
      "learning_rate": 1.646000488618527e-05,
      "loss": 0.6047,
      "step": 4947
    },
    {
      "epoch": 0.6261705897241204,
      "grad_norm": 1.0654377937316895,
      "learning_rate": 1.6458636788217477e-05,
      "loss": 0.6506,
      "step": 4948
    },
    {
      "epoch": 0.6262971399645659,
      "grad_norm": 0.9613418579101562,
      "learning_rate": 1.6457268482817198e-05,
      "loss": 0.5901,
      "step": 4949
    },
    {
      "epoch": 0.6264236902050114,
      "grad_norm": 0.98826003074646,
      "learning_rate": 1.6455899970028382e-05,
      "loss": 0.6075,
      "step": 4950
    },
    {
      "epoch": 0.6265502404454568,
      "grad_norm": 0.9565483331680298,
      "learning_rate": 1.6454531249894975e-05,
      "loss": 0.6323,
      "step": 4951
    },
    {
      "epoch": 0.6266767906859023,
      "grad_norm": 0.9204974174499512,
      "learning_rate": 1.6453162322460936e-05,
      "loss": 0.583,
      "step": 4952
    },
    {
      "epoch": 0.6268033409263477,
      "grad_norm": 0.984189510345459,
      "learning_rate": 1.6451793187770238e-05,
      "loss": 0.6043,
      "step": 4953
    },
    {
      "epoch": 0.6269298911667932,
      "grad_norm": 0.9975201487541199,
      "learning_rate": 1.6450423845866844e-05,
      "loss": 0.6144,
      "step": 4954
    },
    {
      "epoch": 0.6270564414072387,
      "grad_norm": 0.9301570653915405,
      "learning_rate": 1.6449054296794743e-05,
      "loss": 0.606,
      "step": 4955
    },
    {
      "epoch": 0.6271829916476841,
      "grad_norm": 0.9953912496566772,
      "learning_rate": 1.644768454059792e-05,
      "loss": 0.6791,
      "step": 4956
    },
    {
      "epoch": 0.6273095418881296,
      "grad_norm": 1.0647956132888794,
      "learning_rate": 1.644631457732036e-05,
      "loss": 0.6295,
      "step": 4957
    },
    {
      "epoch": 0.6274360921285751,
      "grad_norm": 0.9678139090538025,
      "learning_rate": 1.6444944407006066e-05,
      "loss": 0.6289,
      "step": 4958
    },
    {
      "epoch": 0.6275626423690205,
      "grad_norm": 0.9781757593154907,
      "learning_rate": 1.6443574029699044e-05,
      "loss": 0.6647,
      "step": 4959
    },
    {
      "epoch": 0.627689192609466,
      "grad_norm": 0.9246792793273926,
      "learning_rate": 1.644220344544331e-05,
      "loss": 0.6353,
      "step": 4960
    },
    {
      "epoch": 0.6278157428499114,
      "grad_norm": 0.9527065753936768,
      "learning_rate": 1.6440832654282876e-05,
      "loss": 0.6182,
      "step": 4961
    },
    {
      "epoch": 0.6279422930903569,
      "grad_norm": 1.0459609031677246,
      "learning_rate": 1.6439461656261775e-05,
      "loss": 0.6622,
      "step": 4962
    },
    {
      "epoch": 0.6280688433308024,
      "grad_norm": 1.0513672828674316,
      "learning_rate": 1.6438090451424034e-05,
      "loss": 0.6397,
      "step": 4963
    },
    {
      "epoch": 0.6281953935712478,
      "grad_norm": 1.0179442167282104,
      "learning_rate": 1.6436719039813696e-05,
      "loss": 0.5961,
      "step": 4964
    },
    {
      "epoch": 0.6283219438116933,
      "grad_norm": 0.9708465337753296,
      "learning_rate": 1.6435347421474806e-05,
      "loss": 0.6282,
      "step": 4965
    },
    {
      "epoch": 0.6284484940521387,
      "grad_norm": 0.94657963514328,
      "learning_rate": 1.6433975596451413e-05,
      "loss": 0.6156,
      "step": 4966
    },
    {
      "epoch": 0.6285750442925842,
      "grad_norm": 1.013396978378296,
      "learning_rate": 1.643260356478758e-05,
      "loss": 0.6368,
      "step": 4967
    },
    {
      "epoch": 0.6287015945330297,
      "grad_norm": 1.05673348903656,
      "learning_rate": 1.643123132652737e-05,
      "loss": 0.6745,
      "step": 4968
    },
    {
      "epoch": 0.628828144773475,
      "grad_norm": 1.023739218711853,
      "learning_rate": 1.642985888171486e-05,
      "loss": 0.6367,
      "step": 4969
    },
    {
      "epoch": 0.6289546950139205,
      "grad_norm": 1.0001230239868164,
      "learning_rate": 1.6428486230394116e-05,
      "loss": 0.6191,
      "step": 4970
    },
    {
      "epoch": 0.6290812452543659,
      "grad_norm": 0.9385541081428528,
      "learning_rate": 1.642711337260924e-05,
      "loss": 0.6049,
      "step": 4971
    },
    {
      "epoch": 0.6292077954948114,
      "grad_norm": 0.9703039526939392,
      "learning_rate": 1.6425740308404314e-05,
      "loss": 0.653,
      "step": 4972
    },
    {
      "epoch": 0.6293343457352569,
      "grad_norm": 0.9904257655143738,
      "learning_rate": 1.642436703782344e-05,
      "loss": 0.6057,
      "step": 4973
    },
    {
      "epoch": 0.6294608959757023,
      "grad_norm": 1.020111083984375,
      "learning_rate": 1.6422993560910726e-05,
      "loss": 0.6445,
      "step": 4974
    },
    {
      "epoch": 0.6295874462161478,
      "grad_norm": 1.1077265739440918,
      "learning_rate": 1.642161987771028e-05,
      "loss": 0.6405,
      "step": 4975
    },
    {
      "epoch": 0.6297139964565933,
      "grad_norm": 0.9491071105003357,
      "learning_rate": 1.642024598826622e-05,
      "loss": 0.6201,
      "step": 4976
    },
    {
      "epoch": 0.6298405466970387,
      "grad_norm": 0.9532394409179688,
      "learning_rate": 1.6418871892622673e-05,
      "loss": 0.6335,
      "step": 4977
    },
    {
      "epoch": 0.6299670969374842,
      "grad_norm": 0.9948004484176636,
      "learning_rate": 1.641749759082377e-05,
      "loss": 0.6572,
      "step": 4978
    },
    {
      "epoch": 0.6300936471779296,
      "grad_norm": 0.9757208228111267,
      "learning_rate": 1.6416123082913653e-05,
      "loss": 0.6128,
      "step": 4979
    },
    {
      "epoch": 0.6302201974183751,
      "grad_norm": 0.9197485446929932,
      "learning_rate": 1.6414748368936468e-05,
      "loss": 0.5892,
      "step": 4980
    },
    {
      "epoch": 0.6303467476588206,
      "grad_norm": 0.9384541511535645,
      "learning_rate": 1.641337344893636e-05,
      "loss": 0.6169,
      "step": 4981
    },
    {
      "epoch": 0.630473297899266,
      "grad_norm": 0.9695776104927063,
      "learning_rate": 1.6411998322957487e-05,
      "loss": 0.588,
      "step": 4982
    },
    {
      "epoch": 0.6305998481397115,
      "grad_norm": 0.8986698389053345,
      "learning_rate": 1.6410622991044024e-05,
      "loss": 0.6016,
      "step": 4983
    },
    {
      "epoch": 0.6307263983801569,
      "grad_norm": 0.98868328332901,
      "learning_rate": 1.6409247453240134e-05,
      "loss": 0.6106,
      "step": 4984
    },
    {
      "epoch": 0.6308529486206024,
      "grad_norm": 0.9197222590446472,
      "learning_rate": 1.640787170959e-05,
      "loss": 0.6054,
      "step": 4985
    },
    {
      "epoch": 0.6309794988610479,
      "grad_norm": 0.9788580536842346,
      "learning_rate": 1.64064957601378e-05,
      "loss": 0.6016,
      "step": 4986
    },
    {
      "epoch": 0.6311060491014933,
      "grad_norm": 0.9268537163734436,
      "learning_rate": 1.6405119604927734e-05,
      "loss": 0.6403,
      "step": 4987
    },
    {
      "epoch": 0.6312325993419388,
      "grad_norm": 0.9704577326774597,
      "learning_rate": 1.6403743244003996e-05,
      "loss": 0.621,
      "step": 4988
    },
    {
      "epoch": 0.6313591495823843,
      "grad_norm": 0.9384557008743286,
      "learning_rate": 1.640236667741079e-05,
      "loss": 0.6102,
      "step": 4989
    },
    {
      "epoch": 0.6314856998228296,
      "grad_norm": 0.9007760882377625,
      "learning_rate": 1.640098990519233e-05,
      "loss": 0.5662,
      "step": 4990
    },
    {
      "epoch": 0.6316122500632751,
      "grad_norm": 1.1845450401306152,
      "learning_rate": 1.639961292739283e-05,
      "loss": 0.6253,
      "step": 4991
    },
    {
      "epoch": 0.6317388003037205,
      "grad_norm": 1.0037262439727783,
      "learning_rate": 1.639823574405652e-05,
      "loss": 0.6564,
      "step": 4992
    },
    {
      "epoch": 0.631865350544166,
      "grad_norm": 2.152132034301758,
      "learning_rate": 1.6396858355227624e-05,
      "loss": 0.6417,
      "step": 4993
    },
    {
      "epoch": 0.6319919007846115,
      "grad_norm": 0.9781987071037292,
      "learning_rate": 1.6395480760950382e-05,
      "loss": 0.623,
      "step": 4994
    },
    {
      "epoch": 0.6321184510250569,
      "grad_norm": 0.9567299485206604,
      "learning_rate": 1.6394102961269043e-05,
      "loss": 0.5847,
      "step": 4995
    },
    {
      "epoch": 0.6322450012655024,
      "grad_norm": 0.897968590259552,
      "learning_rate": 1.6392724956227856e-05,
      "loss": 0.6062,
      "step": 4996
    },
    {
      "epoch": 0.6323715515059478,
      "grad_norm": 1.0893832445144653,
      "learning_rate": 1.6391346745871073e-05,
      "loss": 0.5734,
      "step": 4997
    },
    {
      "epoch": 0.6324981017463933,
      "grad_norm": 0.9354502558708191,
      "learning_rate": 1.6389968330242964e-05,
      "loss": 0.6051,
      "step": 4998
    },
    {
      "epoch": 0.6326246519868388,
      "grad_norm": 0.881625235080719,
      "learning_rate": 1.6388589709387797e-05,
      "loss": 0.5814,
      "step": 4999
    },
    {
      "epoch": 0.6327512022272842,
      "grad_norm": 0.8853029012680054,
      "learning_rate": 1.6387210883349853e-05,
      "loss": 0.5852,
      "step": 5000
    },
    {
      "epoch": 0.6328777524677297,
      "grad_norm": 0.9159303307533264,
      "learning_rate": 1.6385831852173407e-05,
      "loss": 0.6473,
      "step": 5001
    },
    {
      "epoch": 0.6330043027081751,
      "grad_norm": 0.9547086954116821,
      "learning_rate": 1.638445261590276e-05,
      "loss": 0.6449,
      "step": 5002
    },
    {
      "epoch": 0.6331308529486206,
      "grad_norm": 0.9424626231193542,
      "learning_rate": 1.6383073174582203e-05,
      "loss": 0.61,
      "step": 5003
    },
    {
      "epoch": 0.6332574031890661,
      "grad_norm": 0.9707544445991516,
      "learning_rate": 1.638169352825604e-05,
      "loss": 0.5803,
      "step": 5004
    },
    {
      "epoch": 0.6333839534295115,
      "grad_norm": 0.8798820376396179,
      "learning_rate": 1.6380313676968584e-05,
      "loss": 0.5921,
      "step": 5005
    },
    {
      "epoch": 0.633510503669957,
      "grad_norm": 0.9221363067626953,
      "learning_rate": 1.637893362076415e-05,
      "loss": 0.5797,
      "step": 5006
    },
    {
      "epoch": 0.6336370539104025,
      "grad_norm": 0.9708017110824585,
      "learning_rate": 1.637755335968706e-05,
      "loss": 0.601,
      "step": 5007
    },
    {
      "epoch": 0.6337636041508479,
      "grad_norm": 1.019362211227417,
      "learning_rate": 1.637617289378165e-05,
      "loss": 0.6366,
      "step": 5008
    },
    {
      "epoch": 0.6338901543912934,
      "grad_norm": 0.9699429273605347,
      "learning_rate": 1.6374792223092245e-05,
      "loss": 0.6175,
      "step": 5009
    },
    {
      "epoch": 0.6340167046317388,
      "grad_norm": 0.8911601901054382,
      "learning_rate": 1.6373411347663198e-05,
      "loss": 0.6007,
      "step": 5010
    },
    {
      "epoch": 0.6341432548721843,
      "grad_norm": 0.958130955696106,
      "learning_rate": 1.6372030267538853e-05,
      "loss": 0.6139,
      "step": 5011
    },
    {
      "epoch": 0.6342698051126298,
      "grad_norm": 0.886265218257904,
      "learning_rate": 1.637064898276357e-05,
      "loss": 0.5761,
      "step": 5012
    },
    {
      "epoch": 0.6343963553530751,
      "grad_norm": 1.0017881393432617,
      "learning_rate": 1.6369267493381712e-05,
      "loss": 0.6802,
      "step": 5013
    },
    {
      "epoch": 0.6345229055935206,
      "grad_norm": 0.972873330116272,
      "learning_rate": 1.636788579943765e-05,
      "loss": 0.6241,
      "step": 5014
    },
    {
      "epoch": 0.634649455833966,
      "grad_norm": 1.0263009071350098,
      "learning_rate": 1.636650390097575e-05,
      "loss": 0.5858,
      "step": 5015
    },
    {
      "epoch": 0.6347760060744115,
      "grad_norm": 1.051133394241333,
      "learning_rate": 1.6365121798040407e-05,
      "loss": 0.7027,
      "step": 5016
    },
    {
      "epoch": 0.634902556314857,
      "grad_norm": 0.984431266784668,
      "learning_rate": 1.6363739490676002e-05,
      "loss": 0.6334,
      "step": 5017
    },
    {
      "epoch": 0.6350291065553024,
      "grad_norm": 1.1289761066436768,
      "learning_rate": 1.6362356978926932e-05,
      "loss": 0.6646,
      "step": 5018
    },
    {
      "epoch": 0.6351556567957479,
      "grad_norm": 0.9182308912277222,
      "learning_rate": 1.6360974262837604e-05,
      "loss": 0.5851,
      "step": 5019
    },
    {
      "epoch": 0.6352822070361933,
      "grad_norm": 1.0278699398040771,
      "learning_rate": 1.6359591342452422e-05,
      "loss": 0.6093,
      "step": 5020
    },
    {
      "epoch": 0.6354087572766388,
      "grad_norm": 0.9683732390403748,
      "learning_rate": 1.63582082178158e-05,
      "loss": 0.591,
      "step": 5021
    },
    {
      "epoch": 0.6355353075170843,
      "grad_norm": 0.9945572018623352,
      "learning_rate": 1.6356824888972165e-05,
      "loss": 0.5881,
      "step": 5022
    },
    {
      "epoch": 0.6356618577575297,
      "grad_norm": 0.906052827835083,
      "learning_rate": 1.6355441355965943e-05,
      "loss": 0.6338,
      "step": 5023
    },
    {
      "epoch": 0.6357884079979752,
      "grad_norm": 1.0265772342681885,
      "learning_rate": 1.635405761884157e-05,
      "loss": 0.6501,
      "step": 5024
    },
    {
      "epoch": 0.6359149582384207,
      "grad_norm": 0.9602616429328918,
      "learning_rate": 1.6352673677643484e-05,
      "loss": 0.6322,
      "step": 5025
    },
    {
      "epoch": 0.6360415084788661,
      "grad_norm": 1.028283953666687,
      "learning_rate": 1.6351289532416136e-05,
      "loss": 0.6566,
      "step": 5026
    },
    {
      "epoch": 0.6361680587193116,
      "grad_norm": 0.9199585914611816,
      "learning_rate": 1.634990518320398e-05,
      "loss": 0.5912,
      "step": 5027
    },
    {
      "epoch": 0.636294608959757,
      "grad_norm": 0.9507606029510498,
      "learning_rate": 1.6348520630051484e-05,
      "loss": 0.6341,
      "step": 5028
    },
    {
      "epoch": 0.6364211592002025,
      "grad_norm": 1.0823040008544922,
      "learning_rate": 1.6347135873003103e-05,
      "loss": 0.6842,
      "step": 5029
    },
    {
      "epoch": 0.636547709440648,
      "grad_norm": 0.9330976605415344,
      "learning_rate": 1.6345750912103318e-05,
      "loss": 0.6197,
      "step": 5030
    },
    {
      "epoch": 0.6366742596810934,
      "grad_norm": 1.0301353931427002,
      "learning_rate": 1.6344365747396614e-05,
      "loss": 0.6223,
      "step": 5031
    },
    {
      "epoch": 0.6368008099215389,
      "grad_norm": 0.9504181146621704,
      "learning_rate": 1.6342980378927468e-05,
      "loss": 0.6177,
      "step": 5032
    },
    {
      "epoch": 0.6369273601619843,
      "grad_norm": 0.9613906145095825,
      "learning_rate": 1.6341594806740383e-05,
      "loss": 0.6439,
      "step": 5033
    },
    {
      "epoch": 0.6370539104024298,
      "grad_norm": 0.9241666197776794,
      "learning_rate": 1.6340209030879858e-05,
      "loss": 0.5995,
      "step": 5034
    },
    {
      "epoch": 0.6371804606428753,
      "grad_norm": 0.965772271156311,
      "learning_rate": 1.6338823051390397e-05,
      "loss": 0.631,
      "step": 5035
    },
    {
      "epoch": 0.6373070108833206,
      "grad_norm": 1.0086679458618164,
      "learning_rate": 1.6337436868316516e-05,
      "loss": 0.6355,
      "step": 5036
    },
    {
      "epoch": 0.6374335611237661,
      "grad_norm": 0.8680964112281799,
      "learning_rate": 1.6336050481702735e-05,
      "loss": 0.5652,
      "step": 5037
    },
    {
      "epoch": 0.6375601113642116,
      "grad_norm": 0.9615563750267029,
      "learning_rate": 1.633466389159358e-05,
      "loss": 0.5985,
      "step": 5038
    },
    {
      "epoch": 0.637686661604657,
      "grad_norm": 1.02921462059021,
      "learning_rate": 1.6333277098033583e-05,
      "loss": 0.603,
      "step": 5039
    },
    {
      "epoch": 0.6378132118451025,
      "grad_norm": 0.8931372165679932,
      "learning_rate": 1.633189010106728e-05,
      "loss": 0.6183,
      "step": 5040
    },
    {
      "epoch": 0.6379397620855479,
      "grad_norm": 1.3016127347946167,
      "learning_rate": 1.633050290073923e-05,
      "loss": 0.6377,
      "step": 5041
    },
    {
      "epoch": 0.6380663123259934,
      "grad_norm": 0.9837984442710876,
      "learning_rate": 1.632911549709398e-05,
      "loss": 0.6335,
      "step": 5042
    },
    {
      "epoch": 0.6381928625664389,
      "grad_norm": 0.9335854053497314,
      "learning_rate": 1.6327727890176083e-05,
      "loss": 0.6004,
      "step": 5043
    },
    {
      "epoch": 0.6383194128068843,
      "grad_norm": 1.0143975019454956,
      "learning_rate": 1.6326340080030108e-05,
      "loss": 0.6279,
      "step": 5044
    },
    {
      "epoch": 0.6384459630473298,
      "grad_norm": 1.027974009513855,
      "learning_rate": 1.6324952066700632e-05,
      "loss": 0.6344,
      "step": 5045
    },
    {
      "epoch": 0.6385725132877752,
      "grad_norm": 0.9762600660324097,
      "learning_rate": 1.6323563850232227e-05,
      "loss": 0.6372,
      "step": 5046
    },
    {
      "epoch": 0.6386990635282207,
      "grad_norm": 0.9693305492401123,
      "learning_rate": 1.6322175430669487e-05,
      "loss": 0.621,
      "step": 5047
    },
    {
      "epoch": 0.6388256137686662,
      "grad_norm": 0.9663785696029663,
      "learning_rate": 1.6320786808056997e-05,
      "loss": 0.599,
      "step": 5048
    },
    {
      "epoch": 0.6389521640091116,
      "grad_norm": 1.1238471269607544,
      "learning_rate": 1.6319397982439356e-05,
      "loss": 0.705,
      "step": 5049
    },
    {
      "epoch": 0.6390787142495571,
      "grad_norm": 1.0468418598175049,
      "learning_rate": 1.631800895386117e-05,
      "loss": 0.6422,
      "step": 5050
    },
    {
      "epoch": 0.6392052644900025,
      "grad_norm": 0.9663745760917664,
      "learning_rate": 1.6316619722367052e-05,
      "loss": 0.6355,
      "step": 5051
    },
    {
      "epoch": 0.639331814730448,
      "grad_norm": 0.9869068264961243,
      "learning_rate": 1.631523028800162e-05,
      "loss": 0.6095,
      "step": 5052
    },
    {
      "epoch": 0.6394583649708935,
      "grad_norm": 1.0664825439453125,
      "learning_rate": 1.6313840650809497e-05,
      "loss": 0.6144,
      "step": 5053
    },
    {
      "epoch": 0.6395849152113389,
      "grad_norm": 0.9860745668411255,
      "learning_rate": 1.6312450810835316e-05,
      "loss": 0.5838,
      "step": 5054
    },
    {
      "epoch": 0.6397114654517844,
      "grad_norm": 1.9176839590072632,
      "learning_rate": 1.631106076812371e-05,
      "loss": 0.6884,
      "step": 5055
    },
    {
      "epoch": 0.6398380156922299,
      "grad_norm": 1.4636203050613403,
      "learning_rate": 1.630967052271933e-05,
      "loss": 0.6785,
      "step": 5056
    },
    {
      "epoch": 0.6399645659326753,
      "grad_norm": 1.0812125205993652,
      "learning_rate": 1.630828007466682e-05,
      "loss": 0.6059,
      "step": 5057
    },
    {
      "epoch": 0.6400911161731208,
      "grad_norm": 1.1192023754119873,
      "learning_rate": 1.6306889424010844e-05,
      "loss": 0.6502,
      "step": 5058
    },
    {
      "epoch": 0.6402176664135661,
      "grad_norm": 1.109489917755127,
      "learning_rate": 1.6305498570796058e-05,
      "loss": 0.6438,
      "step": 5059
    },
    {
      "epoch": 0.6403442166540116,
      "grad_norm": 1.049373984336853,
      "learning_rate": 1.6304107515067143e-05,
      "loss": 0.6334,
      "step": 5060
    },
    {
      "epoch": 0.6404707668944571,
      "grad_norm": 1.1471835374832153,
      "learning_rate": 1.630271625686876e-05,
      "loss": 0.6086,
      "step": 5061
    },
    {
      "epoch": 0.6405973171349025,
      "grad_norm": 1.0139096975326538,
      "learning_rate": 1.6301324796245606e-05,
      "loss": 0.6339,
      "step": 5062
    },
    {
      "epoch": 0.640723867375348,
      "grad_norm": 1.0581527948379517,
      "learning_rate": 1.6299933133242362e-05,
      "loss": 0.6432,
      "step": 5063
    },
    {
      "epoch": 0.6408504176157934,
      "grad_norm": 0.9573229551315308,
      "learning_rate": 1.629854126790373e-05,
      "loss": 0.6185,
      "step": 5064
    },
    {
      "epoch": 0.6409769678562389,
      "grad_norm": 1.0065054893493652,
      "learning_rate": 1.629714920027441e-05,
      "loss": 0.6576,
      "step": 5065
    },
    {
      "epoch": 0.6411035180966844,
      "grad_norm": 0.9911777377128601,
      "learning_rate": 1.629575693039911e-05,
      "loss": 0.5982,
      "step": 5066
    },
    {
      "epoch": 0.6412300683371298,
      "grad_norm": 1.0355792045593262,
      "learning_rate": 1.6294364458322544e-05,
      "loss": 0.6397,
      "step": 5067
    },
    {
      "epoch": 0.6413566185775753,
      "grad_norm": 1.1592377424240112,
      "learning_rate": 1.6292971784089443e-05,
      "loss": 0.6867,
      "step": 5068
    },
    {
      "epoch": 0.6414831688180207,
      "grad_norm": 0.9971093535423279,
      "learning_rate": 1.6291578907744527e-05,
      "loss": 0.6782,
      "step": 5069
    },
    {
      "epoch": 0.6416097190584662,
      "grad_norm": 1.505778431892395,
      "learning_rate": 1.629018582933253e-05,
      "loss": 0.6511,
      "step": 5070
    },
    {
      "epoch": 0.6417362692989117,
      "grad_norm": 1.1326725482940674,
      "learning_rate": 1.62887925488982e-05,
      "loss": 0.6663,
      "step": 5071
    },
    {
      "epoch": 0.6418628195393571,
      "grad_norm": 1.010561466217041,
      "learning_rate": 1.6287399066486286e-05,
      "loss": 0.6277,
      "step": 5072
    },
    {
      "epoch": 0.6419893697798026,
      "grad_norm": 1.2300052642822266,
      "learning_rate": 1.6286005382141533e-05,
      "loss": 0.6287,
      "step": 5073
    },
    {
      "epoch": 0.6421159200202481,
      "grad_norm": 1.0259004831314087,
      "learning_rate": 1.6284611495908714e-05,
      "loss": 0.63,
      "step": 5074
    },
    {
      "epoch": 0.6422424702606935,
      "grad_norm": 0.9959113001823425,
      "learning_rate": 1.6283217407832585e-05,
      "loss": 0.5882,
      "step": 5075
    },
    {
      "epoch": 0.642369020501139,
      "grad_norm": 1.0398824214935303,
      "learning_rate": 1.628182311795793e-05,
      "loss": 0.6016,
      "step": 5076
    },
    {
      "epoch": 0.6424955707415844,
      "grad_norm": 0.9924778938293457,
      "learning_rate": 1.6280428626329525e-05,
      "loss": 0.6343,
      "step": 5077
    },
    {
      "epoch": 0.6426221209820299,
      "grad_norm": 2.387824296951294,
      "learning_rate": 1.6279033932992153e-05,
      "loss": 0.6109,
      "step": 5078
    },
    {
      "epoch": 0.6427486712224754,
      "grad_norm": 1.0437318086624146,
      "learning_rate": 1.6277639037990615e-05,
      "loss": 0.6037,
      "step": 5079
    },
    {
      "epoch": 0.6428752214629208,
      "grad_norm": 1.015927791595459,
      "learning_rate": 1.6276243941369706e-05,
      "loss": 0.6365,
      "step": 5080
    },
    {
      "epoch": 0.6430017717033663,
      "grad_norm": 0.998635470867157,
      "learning_rate": 1.6274848643174235e-05,
      "loss": 0.6236,
      "step": 5081
    },
    {
      "epoch": 0.6431283219438116,
      "grad_norm": 0.9458276033401489,
      "learning_rate": 1.627345314344901e-05,
      "loss": 0.6158,
      "step": 5082
    },
    {
      "epoch": 0.6432548721842571,
      "grad_norm": 1.0810710191726685,
      "learning_rate": 1.627205744223886e-05,
      "loss": 0.6455,
      "step": 5083
    },
    {
      "epoch": 0.6433814224247026,
      "grad_norm": 1.0446990728378296,
      "learning_rate": 1.62706615395886e-05,
      "loss": 0.6333,
      "step": 5084
    },
    {
      "epoch": 0.643507972665148,
      "grad_norm": 1.2553350925445557,
      "learning_rate": 1.6269265435543072e-05,
      "loss": 0.6991,
      "step": 5085
    },
    {
      "epoch": 0.6436345229055935,
      "grad_norm": 1.0162614583969116,
      "learning_rate": 1.6267869130147106e-05,
      "loss": 0.6215,
      "step": 5086
    },
    {
      "epoch": 0.643761073146039,
      "grad_norm": 0.9273417592048645,
      "learning_rate": 1.6266472623445554e-05,
      "loss": 0.6371,
      "step": 5087
    },
    {
      "epoch": 0.6438876233864844,
      "grad_norm": 0.9674442410469055,
      "learning_rate": 1.6265075915483265e-05,
      "loss": 0.6101,
      "step": 5088
    },
    {
      "epoch": 0.6440141736269299,
      "grad_norm": 0.9408568143844604,
      "learning_rate": 1.6263679006305094e-05,
      "loss": 0.616,
      "step": 5089
    },
    {
      "epoch": 0.6441407238673753,
      "grad_norm": 0.9410844445228577,
      "learning_rate": 1.626228189595591e-05,
      "loss": 0.6,
      "step": 5090
    },
    {
      "epoch": 0.6442672741078208,
      "grad_norm": 0.9492683410644531,
      "learning_rate": 1.6260884584480586e-05,
      "loss": 0.6091,
      "step": 5091
    },
    {
      "epoch": 0.6443938243482663,
      "grad_norm": 0.9017401933670044,
      "learning_rate": 1.625948707192399e-05,
      "loss": 0.5775,
      "step": 5092
    },
    {
      "epoch": 0.6445203745887117,
      "grad_norm": 0.9406362771987915,
      "learning_rate": 1.6258089358331016e-05,
      "loss": 0.6345,
      "step": 5093
    },
    {
      "epoch": 0.6446469248291572,
      "grad_norm": 1.0144526958465576,
      "learning_rate": 1.6256691443746553e-05,
      "loss": 0.6127,
      "step": 5094
    },
    {
      "epoch": 0.6447734750696026,
      "grad_norm": 1.1452200412750244,
      "learning_rate": 1.6255293328215497e-05,
      "loss": 0.6073,
      "step": 5095
    },
    {
      "epoch": 0.6449000253100481,
      "grad_norm": 1.0850183963775635,
      "learning_rate": 1.6253895011782747e-05,
      "loss": 0.6744,
      "step": 5096
    },
    {
      "epoch": 0.6450265755504936,
      "grad_norm": 0.9833073616027832,
      "learning_rate": 1.6252496494493216e-05,
      "loss": 0.5994,
      "step": 5097
    },
    {
      "epoch": 0.645153125790939,
      "grad_norm": 0.8974035978317261,
      "learning_rate": 1.625109777639182e-05,
      "loss": 0.5951,
      "step": 5098
    },
    {
      "epoch": 0.6452796760313845,
      "grad_norm": 1.0513089895248413,
      "learning_rate": 1.624969885752348e-05,
      "loss": 0.6457,
      "step": 5099
    },
    {
      "epoch": 0.6454062262718299,
      "grad_norm": 1.0283472537994385,
      "learning_rate": 1.6248299737933134e-05,
      "loss": 0.6683,
      "step": 5100
    },
    {
      "epoch": 0.6455327765122754,
      "grad_norm": 0.8837717175483704,
      "learning_rate": 1.6246900417665704e-05,
      "loss": 0.5986,
      "step": 5101
    },
    {
      "epoch": 0.6456593267527209,
      "grad_norm": 1.027660846710205,
      "learning_rate": 1.6245500896766143e-05,
      "loss": 0.6553,
      "step": 5102
    },
    {
      "epoch": 0.6457858769931663,
      "grad_norm": 0.9639826416969299,
      "learning_rate": 1.6244101175279393e-05,
      "loss": 0.6122,
      "step": 5103
    },
    {
      "epoch": 0.6459124272336118,
      "grad_norm": 0.9352954626083374,
      "learning_rate": 1.6242701253250412e-05,
      "loss": 0.629,
      "step": 5104
    },
    {
      "epoch": 0.6460389774740573,
      "grad_norm": 1.0007667541503906,
      "learning_rate": 1.624130113072416e-05,
      "loss": 0.6026,
      "step": 5105
    },
    {
      "epoch": 0.6461655277145026,
      "grad_norm": 0.9105656147003174,
      "learning_rate": 1.623990080774561e-05,
      "loss": 0.5887,
      "step": 5106
    },
    {
      "epoch": 0.6462920779549481,
      "grad_norm": 0.9242458939552307,
      "learning_rate": 1.6238500284359725e-05,
      "loss": 0.6206,
      "step": 5107
    },
    {
      "epoch": 0.6464186281953935,
      "grad_norm": 0.9997619986534119,
      "learning_rate": 1.62370995606115e-05,
      "loss": 0.632,
      "step": 5108
    },
    {
      "epoch": 0.646545178435839,
      "grad_norm": 0.927789568901062,
      "learning_rate": 1.623569863654591e-05,
      "loss": 0.6059,
      "step": 5109
    },
    {
      "epoch": 0.6466717286762845,
      "grad_norm": 1.0071221590042114,
      "learning_rate": 1.623429751220795e-05,
      "loss": 0.6292,
      "step": 5110
    },
    {
      "epoch": 0.6467982789167299,
      "grad_norm": 0.9423531889915466,
      "learning_rate": 1.623289618764263e-05,
      "loss": 0.6704,
      "step": 5111
    },
    {
      "epoch": 0.6469248291571754,
      "grad_norm": 0.9155585169792175,
      "learning_rate": 1.6231494662894946e-05,
      "loss": 0.6443,
      "step": 5112
    },
    {
      "epoch": 0.6470513793976208,
      "grad_norm": 0.9178339838981628,
      "learning_rate": 1.6230092938009917e-05,
      "loss": 0.5987,
      "step": 5113
    },
    {
      "epoch": 0.6471779296380663,
      "grad_norm": 0.977591335773468,
      "learning_rate": 1.622869101303256e-05,
      "loss": 0.6171,
      "step": 5114
    },
    {
      "epoch": 0.6473044798785118,
      "grad_norm": 0.9716110229492188,
      "learning_rate": 1.6227288888007895e-05,
      "loss": 0.6417,
      "step": 5115
    },
    {
      "epoch": 0.6474310301189572,
      "grad_norm": 0.9754112362861633,
      "learning_rate": 1.6225886562980965e-05,
      "loss": 0.5924,
      "step": 5116
    },
    {
      "epoch": 0.6475575803594027,
      "grad_norm": 0.9821953773498535,
      "learning_rate": 1.6224484037996803e-05,
      "loss": 0.5861,
      "step": 5117
    },
    {
      "epoch": 0.6476841305998481,
      "grad_norm": 0.9757218956947327,
      "learning_rate": 1.6223081313100452e-05,
      "loss": 0.654,
      "step": 5118
    },
    {
      "epoch": 0.6478106808402936,
      "grad_norm": 0.9611251354217529,
      "learning_rate": 1.6221678388336967e-05,
      "loss": 0.6021,
      "step": 5119
    },
    {
      "epoch": 0.6479372310807391,
      "grad_norm": 0.9887414574623108,
      "learning_rate": 1.6220275263751406e-05,
      "loss": 0.635,
      "step": 5120
    },
    {
      "epoch": 0.6480637813211845,
      "grad_norm": 0.9453829526901245,
      "learning_rate": 1.6218871939388828e-05,
      "loss": 0.6273,
      "step": 5121
    },
    {
      "epoch": 0.64819033156163,
      "grad_norm": 0.9413347840309143,
      "learning_rate": 1.6217468415294312e-05,
      "loss": 0.6518,
      "step": 5122
    },
    {
      "epoch": 0.6483168818020755,
      "grad_norm": 0.9504764080047607,
      "learning_rate": 1.6216064691512928e-05,
      "loss": 0.6009,
      "step": 5123
    },
    {
      "epoch": 0.6484434320425209,
      "grad_norm": 0.9773117899894714,
      "learning_rate": 1.6214660768089764e-05,
      "loss": 0.6058,
      "step": 5124
    },
    {
      "epoch": 0.6485699822829664,
      "grad_norm": 0.9987741708755493,
      "learning_rate": 1.6213256645069907e-05,
      "loss": 0.6355,
      "step": 5125
    },
    {
      "epoch": 0.6486965325234118,
      "grad_norm": 0.8680945038795471,
      "learning_rate": 1.6211852322498452e-05,
      "loss": 0.5479,
      "step": 5126
    },
    {
      "epoch": 0.6488230827638573,
      "grad_norm": 1.0049455165863037,
      "learning_rate": 1.6210447800420506e-05,
      "loss": 0.637,
      "step": 5127
    },
    {
      "epoch": 0.6489496330043028,
      "grad_norm": 1.034554123878479,
      "learning_rate": 1.6209043078881174e-05,
      "loss": 0.6696,
      "step": 5128
    },
    {
      "epoch": 0.6490761832447481,
      "grad_norm": 1.029884934425354,
      "learning_rate": 1.6207638157925577e-05,
      "loss": 0.6297,
      "step": 5129
    },
    {
      "epoch": 0.6492027334851936,
      "grad_norm": 0.909457266330719,
      "learning_rate": 1.620623303759883e-05,
      "loss": 0.5901,
      "step": 5130
    },
    {
      "epoch": 0.649329283725639,
      "grad_norm": 1.0195457935333252,
      "learning_rate": 1.620482771794607e-05,
      "loss": 0.6101,
      "step": 5131
    },
    {
      "epoch": 0.6494558339660845,
      "grad_norm": 0.9752768874168396,
      "learning_rate": 1.6203422199012422e-05,
      "loss": 0.6317,
      "step": 5132
    },
    {
      "epoch": 0.64958238420653,
      "grad_norm": 1.0955638885498047,
      "learning_rate": 1.6202016480843035e-05,
      "loss": 0.5925,
      "step": 5133
    },
    {
      "epoch": 0.6497089344469754,
      "grad_norm": 0.973827064037323,
      "learning_rate": 1.620061056348305e-05,
      "loss": 0.6269,
      "step": 5134
    },
    {
      "epoch": 0.6498354846874209,
      "grad_norm": 0.965758204460144,
      "learning_rate": 1.619920444697763e-05,
      "loss": 0.6086,
      "step": 5135
    },
    {
      "epoch": 0.6499620349278664,
      "grad_norm": 0.9895248413085938,
      "learning_rate": 1.6197798131371924e-05,
      "loss": 0.6373,
      "step": 5136
    },
    {
      "epoch": 0.6500885851683118,
      "grad_norm": 0.9985294342041016,
      "learning_rate": 1.6196391616711108e-05,
      "loss": 0.6157,
      "step": 5137
    },
    {
      "epoch": 0.6502151354087573,
      "grad_norm": 1.1325457096099854,
      "learning_rate": 1.6194984903040347e-05,
      "loss": 0.6412,
      "step": 5138
    },
    {
      "epoch": 0.6503416856492027,
      "grad_norm": 1.0644410848617554,
      "learning_rate": 1.619357799040483e-05,
      "loss": 0.6663,
      "step": 5139
    },
    {
      "epoch": 0.6504682358896482,
      "grad_norm": 1.0624569654464722,
      "learning_rate": 1.6192170878849736e-05,
      "loss": 0.6057,
      "step": 5140
    },
    {
      "epoch": 0.6505947861300937,
      "grad_norm": 0.977685809135437,
      "learning_rate": 1.619076356842026e-05,
      "loss": 0.6048,
      "step": 5141
    },
    {
      "epoch": 0.6507213363705391,
      "grad_norm": 0.9481033682823181,
      "learning_rate": 1.61893560591616e-05,
      "loss": 0.6227,
      "step": 5142
    },
    {
      "epoch": 0.6508478866109846,
      "grad_norm": 0.9664007425308228,
      "learning_rate": 1.6187948351118958e-05,
      "loss": 0.6323,
      "step": 5143
    },
    {
      "epoch": 0.65097443685143,
      "grad_norm": 0.9695121049880981,
      "learning_rate": 1.618654044433755e-05,
      "loss": 0.645,
      "step": 5144
    },
    {
      "epoch": 0.6511009870918755,
      "grad_norm": 1.0100144147872925,
      "learning_rate": 1.6185132338862594e-05,
      "loss": 0.6507,
      "step": 5145
    },
    {
      "epoch": 0.651227537332321,
      "grad_norm": 0.9527573585510254,
      "learning_rate": 1.6183724034739313e-05,
      "loss": 0.6151,
      "step": 5146
    },
    {
      "epoch": 0.6513540875727664,
      "grad_norm": 0.9349608421325684,
      "learning_rate": 1.6182315532012934e-05,
      "loss": 0.6539,
      "step": 5147
    },
    {
      "epoch": 0.6514806378132119,
      "grad_norm": 1.107710599899292,
      "learning_rate": 1.61809068307287e-05,
      "loss": 0.6877,
      "step": 5148
    },
    {
      "epoch": 0.6516071880536572,
      "grad_norm": 0.9643529057502747,
      "learning_rate": 1.617949793093185e-05,
      "loss": 0.6156,
      "step": 5149
    },
    {
      "epoch": 0.6517337382941027,
      "grad_norm": 0.9383097887039185,
      "learning_rate": 1.6178088832667635e-05,
      "loss": 0.6136,
      "step": 5150
    },
    {
      "epoch": 0.6518602885345482,
      "grad_norm": 0.9287600517272949,
      "learning_rate": 1.6176679535981314e-05,
      "loss": 0.6085,
      "step": 5151
    },
    {
      "epoch": 0.6519868387749936,
      "grad_norm": 0.9559696912765503,
      "learning_rate": 1.6175270040918144e-05,
      "loss": 0.6256,
      "step": 5152
    },
    {
      "epoch": 0.6521133890154391,
      "grad_norm": 0.9530015587806702,
      "learning_rate": 1.61738603475234e-05,
      "loss": 0.6053,
      "step": 5153
    },
    {
      "epoch": 0.6522399392558846,
      "grad_norm": 0.9585109949111938,
      "learning_rate": 1.6172450455842352e-05,
      "loss": 0.6335,
      "step": 5154
    },
    {
      "epoch": 0.65236648949633,
      "grad_norm": 0.918837308883667,
      "learning_rate": 1.617104036592028e-05,
      "loss": 0.6344,
      "step": 5155
    },
    {
      "epoch": 0.6524930397367755,
      "grad_norm": 0.956230103969574,
      "learning_rate": 1.6169630077802477e-05,
      "loss": 0.5907,
      "step": 5156
    },
    {
      "epoch": 0.6526195899772209,
      "grad_norm": 0.9641780257225037,
      "learning_rate": 1.616821959153424e-05,
      "loss": 0.6059,
      "step": 5157
    },
    {
      "epoch": 0.6527461402176664,
      "grad_norm": 0.9506321549415588,
      "learning_rate": 1.6166808907160866e-05,
      "loss": 0.6319,
      "step": 5158
    },
    {
      "epoch": 0.6528726904581119,
      "grad_norm": 0.9552674889564514,
      "learning_rate": 1.6165398024727657e-05,
      "loss": 0.6523,
      "step": 5159
    },
    {
      "epoch": 0.6529992406985573,
      "grad_norm": 0.9442626237869263,
      "learning_rate": 1.6163986944279934e-05,
      "loss": 0.6036,
      "step": 5160
    },
    {
      "epoch": 0.6531257909390028,
      "grad_norm": 0.9150153994560242,
      "learning_rate": 1.6162575665863006e-05,
      "loss": 0.58,
      "step": 5161
    },
    {
      "epoch": 0.6532523411794482,
      "grad_norm": 0.9207025170326233,
      "learning_rate": 1.6161164189522214e-05,
      "loss": 0.6037,
      "step": 5162
    },
    {
      "epoch": 0.6533788914198937,
      "grad_norm": 0.9848641157150269,
      "learning_rate": 1.6159752515302884e-05,
      "loss": 0.6398,
      "step": 5163
    },
    {
      "epoch": 0.6535054416603392,
      "grad_norm": 1.0186164379119873,
      "learning_rate": 1.6158340643250353e-05,
      "loss": 0.635,
      "step": 5164
    },
    {
      "epoch": 0.6536319919007846,
      "grad_norm": 0.9343981146812439,
      "learning_rate": 1.6156928573409966e-05,
      "loss": 0.6449,
      "step": 5165
    },
    {
      "epoch": 0.6537585421412301,
      "grad_norm": 0.9631854891777039,
      "learning_rate": 1.6155516305827075e-05,
      "loss": 0.6209,
      "step": 5166
    },
    {
      "epoch": 0.6538850923816755,
      "grad_norm": 0.9451952576637268,
      "learning_rate": 1.615410384054704e-05,
      "loss": 0.5809,
      "step": 5167
    },
    {
      "epoch": 0.654011642622121,
      "grad_norm": 0.9745913743972778,
      "learning_rate": 1.6152691177615224e-05,
      "loss": 0.6215,
      "step": 5168
    },
    {
      "epoch": 0.6541381928625665,
      "grad_norm": 2.3778791427612305,
      "learning_rate": 1.6151278317076998e-05,
      "loss": 0.6534,
      "step": 5169
    },
    {
      "epoch": 0.6542647431030119,
      "grad_norm": 0.9401265978813171,
      "learning_rate": 1.614986525897774e-05,
      "loss": 0.6142,
      "step": 5170
    },
    {
      "epoch": 0.6543912933434574,
      "grad_norm": 0.937778890132904,
      "learning_rate": 1.6148452003362828e-05,
      "loss": 0.608,
      "step": 5171
    },
    {
      "epoch": 0.6545178435839029,
      "grad_norm": 0.9007929563522339,
      "learning_rate": 1.6147038550277655e-05,
      "loss": 0.5772,
      "step": 5172
    },
    {
      "epoch": 0.6546443938243482,
      "grad_norm": 0.9613137245178223,
      "learning_rate": 1.6145624899767622e-05,
      "loss": 0.6327,
      "step": 5173
    },
    {
      "epoch": 0.6547709440647937,
      "grad_norm": 0.9593843817710876,
      "learning_rate": 1.6144211051878126e-05,
      "loss": 0.5888,
      "step": 5174
    },
    {
      "epoch": 0.6548974943052391,
      "grad_norm": 1.055247187614441,
      "learning_rate": 1.6142797006654575e-05,
      "loss": 0.6326,
      "step": 5175
    },
    {
      "epoch": 0.6550240445456846,
      "grad_norm": 1.0433217287063599,
      "learning_rate": 1.6141382764142383e-05,
      "loss": 0.6469,
      "step": 5176
    },
    {
      "epoch": 0.6551505947861301,
      "grad_norm": 0.9826193451881409,
      "learning_rate": 1.613996832438698e-05,
      "loss": 0.6038,
      "step": 5177
    },
    {
      "epoch": 0.6552771450265755,
      "grad_norm": 0.9999788403511047,
      "learning_rate": 1.613855368743378e-05,
      "loss": 0.6924,
      "step": 5178
    },
    {
      "epoch": 0.655403695267021,
      "grad_norm": 0.926266074180603,
      "learning_rate": 1.6137138853328227e-05,
      "loss": 0.594,
      "step": 5179
    },
    {
      "epoch": 0.6555302455074664,
      "grad_norm": 1.071482539176941,
      "learning_rate": 1.613572382211576e-05,
      "loss": 0.6528,
      "step": 5180
    },
    {
      "epoch": 0.6556567957479119,
      "grad_norm": 0.9910255670547485,
      "learning_rate": 1.6134308593841824e-05,
      "loss": 0.6239,
      "step": 5181
    },
    {
      "epoch": 0.6557833459883574,
      "grad_norm": 0.9433602690696716,
      "learning_rate": 1.6132893168551872e-05,
      "loss": 0.6355,
      "step": 5182
    },
    {
      "epoch": 0.6559098962288028,
      "grad_norm": 0.9825139045715332,
      "learning_rate": 1.613147754629136e-05,
      "loss": 0.6159,
      "step": 5183
    },
    {
      "epoch": 0.6560364464692483,
      "grad_norm": 0.9918792843818665,
      "learning_rate": 1.6130061727105767e-05,
      "loss": 0.6588,
      "step": 5184
    },
    {
      "epoch": 0.6561629967096938,
      "grad_norm": 0.9615283012390137,
      "learning_rate": 1.6128645711040544e-05,
      "loss": 0.6228,
      "step": 5185
    },
    {
      "epoch": 0.6562895469501392,
      "grad_norm": 0.9779450297355652,
      "learning_rate": 1.6127229498141187e-05,
      "loss": 0.6498,
      "step": 5186
    },
    {
      "epoch": 0.6564160971905847,
      "grad_norm": 1.035597562789917,
      "learning_rate": 1.6125813088453174e-05,
      "loss": 0.6496,
      "step": 5187
    },
    {
      "epoch": 0.6565426474310301,
      "grad_norm": 0.9472329020500183,
      "learning_rate": 1.6124396482021995e-05,
      "loss": 0.6057,
      "step": 5188
    },
    {
      "epoch": 0.6566691976714756,
      "grad_norm": 0.9126632213592529,
      "learning_rate": 1.612297967889315e-05,
      "loss": 0.6192,
      "step": 5189
    },
    {
      "epoch": 0.6567957479119211,
      "grad_norm": 0.9300206899642944,
      "learning_rate": 1.6121562679112136e-05,
      "loss": 0.5918,
      "step": 5190
    },
    {
      "epoch": 0.6569222981523665,
      "grad_norm": 0.9767802953720093,
      "learning_rate": 1.612014548272447e-05,
      "loss": 0.5874,
      "step": 5191
    },
    {
      "epoch": 0.657048848392812,
      "grad_norm": 1.0115207433700562,
      "learning_rate": 1.611872808977567e-05,
      "loss": 0.6386,
      "step": 5192
    },
    {
      "epoch": 0.6571753986332574,
      "grad_norm": 0.9762291312217712,
      "learning_rate": 1.611731050031125e-05,
      "loss": 0.6035,
      "step": 5193
    },
    {
      "epoch": 0.6573019488737029,
      "grad_norm": 1.0164153575897217,
      "learning_rate": 1.6115892714376743e-05,
      "loss": 0.6231,
      "step": 5194
    },
    {
      "epoch": 0.6574284991141484,
      "grad_norm": 0.9561837315559387,
      "learning_rate": 1.6114474732017686e-05,
      "loss": 0.643,
      "step": 5195
    },
    {
      "epoch": 0.6575550493545937,
      "grad_norm": 0.9498550891876221,
      "learning_rate": 1.611305655327962e-05,
      "loss": 0.6559,
      "step": 5196
    },
    {
      "epoch": 0.6576815995950392,
      "grad_norm": 0.9827994108200073,
      "learning_rate": 1.611163817820809e-05,
      "loss": 0.6243,
      "step": 5197
    },
    {
      "epoch": 0.6578081498354846,
      "grad_norm": 0.9426862597465515,
      "learning_rate": 1.6110219606848653e-05,
      "loss": 0.6188,
      "step": 5198
    },
    {
      "epoch": 0.6579347000759301,
      "grad_norm": 0.9941657781600952,
      "learning_rate": 1.610880083924687e-05,
      "loss": 0.6697,
      "step": 5199
    },
    {
      "epoch": 0.6580612503163756,
      "grad_norm": 0.9473764300346375,
      "learning_rate": 1.61073818754483e-05,
      "loss": 0.5887,
      "step": 5200
    },
    {
      "epoch": 0.658187800556821,
      "grad_norm": 0.9052305221557617,
      "learning_rate": 1.6105962715498523e-05,
      "loss": 0.5975,
      "step": 5201
    },
    {
      "epoch": 0.6583143507972665,
      "grad_norm": 0.9544158577919006,
      "learning_rate": 1.610454335944312e-05,
      "loss": 0.6139,
      "step": 5202
    },
    {
      "epoch": 0.658440901037712,
      "grad_norm": 1.0932941436767578,
      "learning_rate": 1.6103123807327673e-05,
      "loss": 0.6684,
      "step": 5203
    },
    {
      "epoch": 0.6585674512781574,
      "grad_norm": 1.0642390251159668,
      "learning_rate": 1.610170405919777e-05,
      "loss": 0.6365,
      "step": 5204
    },
    {
      "epoch": 0.6586940015186029,
      "grad_norm": 0.9770675897598267,
      "learning_rate": 1.6100284115099016e-05,
      "loss": 0.6144,
      "step": 5205
    },
    {
      "epoch": 0.6588205517590483,
      "grad_norm": 0.9462794661521912,
      "learning_rate": 1.6098863975077012e-05,
      "loss": 0.6491,
      "step": 5206
    },
    {
      "epoch": 0.6589471019994938,
      "grad_norm": 0.9223582744598389,
      "learning_rate": 1.609744363917737e-05,
      "loss": 0.5977,
      "step": 5207
    },
    {
      "epoch": 0.6590736522399393,
      "grad_norm": 1.110593557357788,
      "learning_rate": 1.609602310744571e-05,
      "loss": 0.6196,
      "step": 5208
    },
    {
      "epoch": 0.6592002024803847,
      "grad_norm": 0.9175549149513245,
      "learning_rate": 1.609460237992765e-05,
      "loss": 0.5811,
      "step": 5209
    },
    {
      "epoch": 0.6593267527208302,
      "grad_norm": 0.9692025184631348,
      "learning_rate": 1.609318145666882e-05,
      "loss": 0.6305,
      "step": 5210
    },
    {
      "epoch": 0.6594533029612756,
      "grad_norm": 1.0278936624526978,
      "learning_rate": 1.6091760337714856e-05,
      "loss": 0.661,
      "step": 5211
    },
    {
      "epoch": 0.6595798532017211,
      "grad_norm": 0.990783154964447,
      "learning_rate": 1.6090339023111403e-05,
      "loss": 0.6032,
      "step": 5212
    },
    {
      "epoch": 0.6597064034421666,
      "grad_norm": 0.9878718256950378,
      "learning_rate": 1.6088917512904112e-05,
      "loss": 0.6482,
      "step": 5213
    },
    {
      "epoch": 0.659832953682612,
      "grad_norm": 1.033773422241211,
      "learning_rate": 1.6087495807138628e-05,
      "loss": 0.6626,
      "step": 5214
    },
    {
      "epoch": 0.6599595039230575,
      "grad_norm": 0.9637693762779236,
      "learning_rate": 1.6086073905860623e-05,
      "loss": 0.6135,
      "step": 5215
    },
    {
      "epoch": 0.6600860541635029,
      "grad_norm": 1.0459917783737183,
      "learning_rate": 1.6084651809115756e-05,
      "loss": 0.599,
      "step": 5216
    },
    {
      "epoch": 0.6602126044039484,
      "grad_norm": 0.9807189702987671,
      "learning_rate": 1.6083229516949705e-05,
      "loss": 0.6211,
      "step": 5217
    },
    {
      "epoch": 0.6603391546443939,
      "grad_norm": 0.9471689462661743,
      "learning_rate": 1.6081807029408145e-05,
      "loss": 0.6034,
      "step": 5218
    },
    {
      "epoch": 0.6604657048848392,
      "grad_norm": 1.0066721439361572,
      "learning_rate": 1.608038434653677e-05,
      "loss": 0.6026,
      "step": 5219
    },
    {
      "epoch": 0.6605922551252847,
      "grad_norm": 0.8815466165542603,
      "learning_rate": 1.6078961468381267e-05,
      "loss": 0.5822,
      "step": 5220
    },
    {
      "epoch": 0.6607188053657302,
      "grad_norm": 0.9153020977973938,
      "learning_rate": 1.6077538394987334e-05,
      "loss": 0.5972,
      "step": 5221
    },
    {
      "epoch": 0.6608453556061756,
      "grad_norm": 0.9406945705413818,
      "learning_rate": 1.607611512640068e-05,
      "loss": 0.6322,
      "step": 5222
    },
    {
      "epoch": 0.6609719058466211,
      "grad_norm": 0.9544447064399719,
      "learning_rate": 1.6074691662667013e-05,
      "loss": 0.6143,
      "step": 5223
    },
    {
      "epoch": 0.6610984560870665,
      "grad_norm": 0.9941664934158325,
      "learning_rate": 1.607326800383205e-05,
      "loss": 0.6276,
      "step": 5224
    },
    {
      "epoch": 0.661225006327512,
      "grad_norm": 1.0631906986236572,
      "learning_rate": 1.6071844149941523e-05,
      "loss": 0.6482,
      "step": 5225
    },
    {
      "epoch": 0.6613515565679575,
      "grad_norm": 1.6522129774093628,
      "learning_rate": 1.607042010104115e-05,
      "loss": 0.6624,
      "step": 5226
    },
    {
      "epoch": 0.6614781068084029,
      "grad_norm": 2.49702525138855,
      "learning_rate": 1.6068995857176673e-05,
      "loss": 0.5968,
      "step": 5227
    },
    {
      "epoch": 0.6616046570488484,
      "grad_norm": 0.9569133520126343,
      "learning_rate": 1.606757141839383e-05,
      "loss": 0.5842,
      "step": 5228
    },
    {
      "epoch": 0.6617312072892938,
      "grad_norm": 0.9164731502532959,
      "learning_rate": 1.6066146784738378e-05,
      "loss": 0.5789,
      "step": 5229
    },
    {
      "epoch": 0.6618577575297393,
      "grad_norm": 0.9924995303153992,
      "learning_rate": 1.606472195625607e-05,
      "loss": 0.6364,
      "step": 5230
    },
    {
      "epoch": 0.6619843077701848,
      "grad_norm": 1.0087382793426514,
      "learning_rate": 1.606329693299266e-05,
      "loss": 0.6513,
      "step": 5231
    },
    {
      "epoch": 0.6621108580106302,
      "grad_norm": 1.0233440399169922,
      "learning_rate": 1.6061871714993925e-05,
      "loss": 0.6479,
      "step": 5232
    },
    {
      "epoch": 0.6622374082510757,
      "grad_norm": 0.9415802359580994,
      "learning_rate": 1.606044630230563e-05,
      "loss": 0.658,
      "step": 5233
    },
    {
      "epoch": 0.6623639584915212,
      "grad_norm": 0.9755691289901733,
      "learning_rate": 1.6059020694973567e-05,
      "loss": 0.5955,
      "step": 5234
    },
    {
      "epoch": 0.6624905087319666,
      "grad_norm": 0.9160376191139221,
      "learning_rate": 1.605759489304351e-05,
      "loss": 0.6296,
      "step": 5235
    },
    {
      "epoch": 0.6626170589724121,
      "grad_norm": 0.9136980175971985,
      "learning_rate": 1.6056168896561257e-05,
      "loss": 0.6266,
      "step": 5236
    },
    {
      "epoch": 0.6627436092128575,
      "grad_norm": 1.0022133588790894,
      "learning_rate": 1.6054742705572606e-05,
      "loss": 0.6362,
      "step": 5237
    },
    {
      "epoch": 0.662870159453303,
      "grad_norm": 0.9620835781097412,
      "learning_rate": 1.6053316320123364e-05,
      "loss": 0.5895,
      "step": 5238
    },
    {
      "epoch": 0.6629967096937485,
      "grad_norm": 0.9167860150337219,
      "learning_rate": 1.605188974025934e-05,
      "loss": 0.5922,
      "step": 5239
    },
    {
      "epoch": 0.6631232599341939,
      "grad_norm": 1.0833830833435059,
      "learning_rate": 1.6050462966026352e-05,
      "loss": 0.6242,
      "step": 5240
    },
    {
      "epoch": 0.6632498101746394,
      "grad_norm": 0.9066812992095947,
      "learning_rate": 1.6049035997470224e-05,
      "loss": 0.583,
      "step": 5241
    },
    {
      "epoch": 0.6633763604150847,
      "grad_norm": 1.0579631328582764,
      "learning_rate": 1.6047608834636787e-05,
      "loss": 0.5871,
      "step": 5242
    },
    {
      "epoch": 0.6635029106555302,
      "grad_norm": 1.0766416788101196,
      "learning_rate": 1.6046181477571874e-05,
      "loss": 0.6028,
      "step": 5243
    },
    {
      "epoch": 0.6636294608959757,
      "grad_norm": 1.0191105604171753,
      "learning_rate": 1.6044753926321334e-05,
      "loss": 0.6375,
      "step": 5244
    },
    {
      "epoch": 0.6637560111364211,
      "grad_norm": 0.9993996024131775,
      "learning_rate": 1.6043326180931012e-05,
      "loss": 0.5984,
      "step": 5245
    },
    {
      "epoch": 0.6638825613768666,
      "grad_norm": 1.0812190771102905,
      "learning_rate": 1.604189824144676e-05,
      "loss": 0.6498,
      "step": 5246
    },
    {
      "epoch": 0.664009111617312,
      "grad_norm": 1.002073049545288,
      "learning_rate": 1.604047010791444e-05,
      "loss": 0.6229,
      "step": 5247
    },
    {
      "epoch": 0.6641356618577575,
      "grad_norm": 1.0245617628097534,
      "learning_rate": 1.6039041780379924e-05,
      "loss": 0.6107,
      "step": 5248
    },
    {
      "epoch": 0.664262212098203,
      "grad_norm": 0.9481391906738281,
      "learning_rate": 1.6037613258889087e-05,
      "loss": 0.6412,
      "step": 5249
    },
    {
      "epoch": 0.6643887623386484,
      "grad_norm": 0.9692586660385132,
      "learning_rate": 1.60361845434878e-05,
      "loss": 0.6374,
      "step": 5250
    },
    {
      "epoch": 0.6645153125790939,
      "grad_norm": 1.0159797668457031,
      "learning_rate": 1.6034755634221958e-05,
      "loss": 0.6227,
      "step": 5251
    },
    {
      "epoch": 0.6646418628195394,
      "grad_norm": 1.0123363733291626,
      "learning_rate": 1.6033326531137446e-05,
      "loss": 0.6656,
      "step": 5252
    },
    {
      "epoch": 0.6647684130599848,
      "grad_norm": 0.9163931608200073,
      "learning_rate": 1.603189723428017e-05,
      "loss": 0.5913,
      "step": 5253
    },
    {
      "epoch": 0.6648949633004303,
      "grad_norm": 0.9409859776496887,
      "learning_rate": 1.6030467743696027e-05,
      "loss": 0.646,
      "step": 5254
    },
    {
      "epoch": 0.6650215135408757,
      "grad_norm": 0.9656674861907959,
      "learning_rate": 1.602903805943093e-05,
      "loss": 0.5961,
      "step": 5255
    },
    {
      "epoch": 0.6651480637813212,
      "grad_norm": 0.9278133511543274,
      "learning_rate": 1.6027608181530806e-05,
      "loss": 0.6568,
      "step": 5256
    },
    {
      "epoch": 0.6652746140217667,
      "grad_norm": 1.1497169733047485,
      "learning_rate": 1.602617811004157e-05,
      "loss": 0.6678,
      "step": 5257
    },
    {
      "epoch": 0.6654011642622121,
      "grad_norm": 0.8867719173431396,
      "learning_rate": 1.6024747845009146e-05,
      "loss": 0.5886,
      "step": 5258
    },
    {
      "epoch": 0.6655277145026576,
      "grad_norm": 0.9848606586456299,
      "learning_rate": 1.6023317386479483e-05,
      "loss": 0.614,
      "step": 5259
    },
    {
      "epoch": 0.665654264743103,
      "grad_norm": 0.9285809397697449,
      "learning_rate": 1.602188673449851e-05,
      "loss": 0.6095,
      "step": 5260
    },
    {
      "epoch": 0.6657808149835485,
      "grad_norm": 0.9423862099647522,
      "learning_rate": 1.602045588911219e-05,
      "loss": 0.662,
      "step": 5261
    },
    {
      "epoch": 0.665907365223994,
      "grad_norm": 0.983855128288269,
      "learning_rate": 1.6019024850366465e-05,
      "loss": 0.633,
      "step": 5262
    },
    {
      "epoch": 0.6660339154644394,
      "grad_norm": 0.918409526348114,
      "learning_rate": 1.60175936183073e-05,
      "loss": 0.6043,
      "step": 5263
    },
    {
      "epoch": 0.6661604657048849,
      "grad_norm": 0.9694800972938538,
      "learning_rate": 1.6016162192980665e-05,
      "loss": 0.6055,
      "step": 5264
    },
    {
      "epoch": 0.6662870159453302,
      "grad_norm": 0.9152687191963196,
      "learning_rate": 1.6014730574432527e-05,
      "loss": 0.5501,
      "step": 5265
    },
    {
      "epoch": 0.6664135661857757,
      "grad_norm": 0.9440381526947021,
      "learning_rate": 1.601329876270887e-05,
      "loss": 0.6162,
      "step": 5266
    },
    {
      "epoch": 0.6665401164262212,
      "grad_norm": 1.0131142139434814,
      "learning_rate": 1.6011866757855677e-05,
      "loss": 0.5797,
      "step": 5267
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.9417364597320557,
      "learning_rate": 1.6010434559918947e-05,
      "loss": 0.6782,
      "step": 5268
    },
    {
      "epoch": 0.6667932169071121,
      "grad_norm": 0.8907631635665894,
      "learning_rate": 1.6009002168944666e-05,
      "loss": 0.5828,
      "step": 5269
    },
    {
      "epoch": 0.6669197671475576,
      "grad_norm": 0.9532276391983032,
      "learning_rate": 1.6007569584978847e-05,
      "loss": 0.6308,
      "step": 5270
    },
    {
      "epoch": 0.667046317388003,
      "grad_norm": 0.9265908598899841,
      "learning_rate": 1.6006136808067498e-05,
      "loss": 0.6076,
      "step": 5271
    },
    {
      "epoch": 0.6671728676284485,
      "grad_norm": 0.9233419895172119,
      "learning_rate": 1.600470383825664e-05,
      "loss": 0.6,
      "step": 5272
    },
    {
      "epoch": 0.6672994178688939,
      "grad_norm": 0.9218330383300781,
      "learning_rate": 1.6003270675592284e-05,
      "loss": 0.6405,
      "step": 5273
    },
    {
      "epoch": 0.6674259681093394,
      "grad_norm": 0.9238069653511047,
      "learning_rate": 1.600183732012047e-05,
      "loss": 0.5749,
      "step": 5274
    },
    {
      "epoch": 0.6675525183497849,
      "grad_norm": 0.9717869162559509,
      "learning_rate": 1.600040377188723e-05,
      "loss": 0.6096,
      "step": 5275
    },
    {
      "epoch": 0.6676790685902303,
      "grad_norm": 0.899998128414154,
      "learning_rate": 1.59989700309386e-05,
      "loss": 0.6032,
      "step": 5276
    },
    {
      "epoch": 0.6678056188306758,
      "grad_norm": 1.0193510055541992,
      "learning_rate": 1.5997536097320636e-05,
      "loss": 0.6003,
      "step": 5277
    },
    {
      "epoch": 0.6679321690711212,
      "grad_norm": 0.950679361820221,
      "learning_rate": 1.5996101971079388e-05,
      "loss": 0.6385,
      "step": 5278
    },
    {
      "epoch": 0.6680587193115667,
      "grad_norm": 0.9370666742324829,
      "learning_rate": 1.5994667652260917e-05,
      "loss": 0.6348,
      "step": 5279
    },
    {
      "epoch": 0.6681852695520122,
      "grad_norm": 0.8968284726142883,
      "learning_rate": 1.5993233140911285e-05,
      "loss": 0.5875,
      "step": 5280
    },
    {
      "epoch": 0.6683118197924576,
      "grad_norm": 0.9388942122459412,
      "learning_rate": 1.5991798437076573e-05,
      "loss": 0.6424,
      "step": 5281
    },
    {
      "epoch": 0.6684383700329031,
      "grad_norm": 0.9384039044380188,
      "learning_rate": 1.5990363540802846e-05,
      "loss": 0.6005,
      "step": 5282
    },
    {
      "epoch": 0.6685649202733486,
      "grad_norm": 0.9807649254798889,
      "learning_rate": 1.59889284521362e-05,
      "loss": 0.6487,
      "step": 5283
    },
    {
      "epoch": 0.668691470513794,
      "grad_norm": 0.89345383644104,
      "learning_rate": 1.5987493171122725e-05,
      "loss": 0.5833,
      "step": 5284
    },
    {
      "epoch": 0.6688180207542395,
      "grad_norm": 0.9611429572105408,
      "learning_rate": 1.5986057697808512e-05,
      "loss": 0.5564,
      "step": 5285
    },
    {
      "epoch": 0.6689445709946849,
      "grad_norm": 0.9021902680397034,
      "learning_rate": 1.5984622032239673e-05,
      "loss": 0.6011,
      "step": 5286
    },
    {
      "epoch": 0.6690711212351304,
      "grad_norm": 0.9246148467063904,
      "learning_rate": 1.5983186174462307e-05,
      "loss": 0.5449,
      "step": 5287
    },
    {
      "epoch": 0.6691976714755759,
      "grad_norm": 1.0585423707962036,
      "learning_rate": 1.5981750124522534e-05,
      "loss": 0.6114,
      "step": 5288
    },
    {
      "epoch": 0.6693242217160212,
      "grad_norm": 0.9459659457206726,
      "learning_rate": 1.598031388246648e-05,
      "loss": 0.5698,
      "step": 5289
    },
    {
      "epoch": 0.6694507719564667,
      "grad_norm": 0.9649491906166077,
      "learning_rate": 1.5978877448340266e-05,
      "loss": 0.6183,
      "step": 5290
    },
    {
      "epoch": 0.6695773221969121,
      "grad_norm": 0.9959087371826172,
      "learning_rate": 1.5977440822190033e-05,
      "loss": 0.6228,
      "step": 5291
    },
    {
      "epoch": 0.6697038724373576,
      "grad_norm": 1.024117350578308,
      "learning_rate": 1.5976004004061914e-05,
      "loss": 0.6658,
      "step": 5292
    },
    {
      "epoch": 0.6698304226778031,
      "grad_norm": 0.9860383868217468,
      "learning_rate": 1.5974566994002062e-05,
      "loss": 0.6453,
      "step": 5293
    },
    {
      "epoch": 0.6699569729182485,
      "grad_norm": 0.9421555399894714,
      "learning_rate": 1.597312979205662e-05,
      "loss": 0.5985,
      "step": 5294
    },
    {
      "epoch": 0.670083523158694,
      "grad_norm": 1.0028185844421387,
      "learning_rate": 1.597169239827176e-05,
      "loss": 0.6749,
      "step": 5295
    },
    {
      "epoch": 0.6702100733991394,
      "grad_norm": 0.960588276386261,
      "learning_rate": 1.5970254812693638e-05,
      "loss": 0.5992,
      "step": 5296
    },
    {
      "epoch": 0.6703366236395849,
      "grad_norm": 0.9220970869064331,
      "learning_rate": 1.5968817035368425e-05,
      "loss": 0.595,
      "step": 5297
    },
    {
      "epoch": 0.6704631738800304,
      "grad_norm": 0.9018083214759827,
      "learning_rate": 1.5967379066342302e-05,
      "loss": 0.5605,
      "step": 5298
    },
    {
      "epoch": 0.6705897241204758,
      "grad_norm": 0.9537651538848877,
      "learning_rate": 1.596594090566145e-05,
      "loss": 0.6215,
      "step": 5299
    },
    {
      "epoch": 0.6707162743609213,
      "grad_norm": 0.963990330696106,
      "learning_rate": 1.5964502553372057e-05,
      "loss": 0.6525,
      "step": 5300
    },
    {
      "epoch": 0.6708428246013668,
      "grad_norm": 0.9002560973167419,
      "learning_rate": 1.5963064009520326e-05,
      "loss": 0.6011,
      "step": 5301
    },
    {
      "epoch": 0.6709693748418122,
      "grad_norm": 1.024066686630249,
      "learning_rate": 1.596162527415245e-05,
      "loss": 0.6415,
      "step": 5302
    },
    {
      "epoch": 0.6710959250822577,
      "grad_norm": 0.9723648428916931,
      "learning_rate": 1.596018634731464e-05,
      "loss": 0.6146,
      "step": 5303
    },
    {
      "epoch": 0.6712224753227031,
      "grad_norm": 0.9938017725944519,
      "learning_rate": 1.595874722905311e-05,
      "loss": 0.6737,
      "step": 5304
    },
    {
      "epoch": 0.6713490255631486,
      "grad_norm": 0.9475978016853333,
      "learning_rate": 1.5957307919414087e-05,
      "loss": 0.6321,
      "step": 5305
    },
    {
      "epoch": 0.6714755758035941,
      "grad_norm": 0.9769773483276367,
      "learning_rate": 1.5955868418443786e-05,
      "loss": 0.6522,
      "step": 5306
    },
    {
      "epoch": 0.6716021260440395,
      "grad_norm": 1.0093694925308228,
      "learning_rate": 1.5954428726188448e-05,
      "loss": 0.6487,
      "step": 5307
    },
    {
      "epoch": 0.671728676284485,
      "grad_norm": 0.9129651784896851,
      "learning_rate": 1.5952988842694304e-05,
      "loss": 0.5654,
      "step": 5308
    },
    {
      "epoch": 0.6718552265249303,
      "grad_norm": 0.9330417513847351,
      "learning_rate": 1.5951548768007605e-05,
      "loss": 0.6382,
      "step": 5309
    },
    {
      "epoch": 0.6719817767653758,
      "grad_norm": 0.9713588356971741,
      "learning_rate": 1.59501085021746e-05,
      "loss": 0.6475,
      "step": 5310
    },
    {
      "epoch": 0.6721083270058213,
      "grad_norm": 0.9999588131904602,
      "learning_rate": 1.5948668045241545e-05,
      "loss": 0.6642,
      "step": 5311
    },
    {
      "epoch": 0.6722348772462667,
      "grad_norm": 0.9931942820549011,
      "learning_rate": 1.594722739725471e-05,
      "loss": 0.6088,
      "step": 5312
    },
    {
      "epoch": 0.6723614274867122,
      "grad_norm": 1.128387689590454,
      "learning_rate": 1.5945786558260358e-05,
      "loss": 0.5883,
      "step": 5313
    },
    {
      "epoch": 0.6724879777271576,
      "grad_norm": 0.9262527227401733,
      "learning_rate": 1.594434552830476e-05,
      "loss": 0.5888,
      "step": 5314
    },
    {
      "epoch": 0.6726145279676031,
      "grad_norm": 0.9159889817237854,
      "learning_rate": 1.594290430743421e-05,
      "loss": 0.6267,
      "step": 5315
    },
    {
      "epoch": 0.6727410782080486,
      "grad_norm": 0.9373523592948914,
      "learning_rate": 1.5941462895694986e-05,
      "loss": 0.6213,
      "step": 5316
    },
    {
      "epoch": 0.672867628448494,
      "grad_norm": 1.0297967195510864,
      "learning_rate": 1.5940021293133384e-05,
      "loss": 0.6017,
      "step": 5317
    },
    {
      "epoch": 0.6729941786889395,
      "grad_norm": 1.039692997932434,
      "learning_rate": 1.5938579499795707e-05,
      "loss": 0.6023,
      "step": 5318
    },
    {
      "epoch": 0.673120728929385,
      "grad_norm": 1.0025602579116821,
      "learning_rate": 1.593713751572826e-05,
      "loss": 0.5976,
      "step": 5319
    },
    {
      "epoch": 0.6732472791698304,
      "grad_norm": 0.9724187254905701,
      "learning_rate": 1.5935695340977352e-05,
      "loss": 0.6226,
      "step": 5320
    },
    {
      "epoch": 0.6733738294102759,
      "grad_norm": 0.9571942687034607,
      "learning_rate": 1.5934252975589302e-05,
      "loss": 0.634,
      "step": 5321
    },
    {
      "epoch": 0.6735003796507213,
      "grad_norm": 0.9679285883903503,
      "learning_rate": 1.5932810419610444e-05,
      "loss": 0.5912,
      "step": 5322
    },
    {
      "epoch": 0.6736269298911668,
      "grad_norm": 0.9523757696151733,
      "learning_rate": 1.5931367673087096e-05,
      "loss": 0.5792,
      "step": 5323
    },
    {
      "epoch": 0.6737534801316123,
      "grad_norm": 1.0088167190551758,
      "learning_rate": 1.59299247360656e-05,
      "loss": 0.6121,
      "step": 5324
    },
    {
      "epoch": 0.6738800303720577,
      "grad_norm": 0.9067131280899048,
      "learning_rate": 1.59284816085923e-05,
      "loss": 0.5731,
      "step": 5325
    },
    {
      "epoch": 0.6740065806125032,
      "grad_norm": 0.9318097233772278,
      "learning_rate": 1.5927038290713545e-05,
      "loss": 0.5875,
      "step": 5326
    },
    {
      "epoch": 0.6741331308529486,
      "grad_norm": 0.9412702918052673,
      "learning_rate": 1.5925594782475686e-05,
      "loss": 0.6097,
      "step": 5327
    },
    {
      "epoch": 0.6742596810933941,
      "grad_norm": 0.9022665619850159,
      "learning_rate": 1.5924151083925087e-05,
      "loss": 0.5793,
      "step": 5328
    },
    {
      "epoch": 0.6743862313338396,
      "grad_norm": 0.9219616651535034,
      "learning_rate": 1.592270719510812e-05,
      "loss": 0.5776,
      "step": 5329
    },
    {
      "epoch": 0.674512781574285,
      "grad_norm": 1.0205466747283936,
      "learning_rate": 1.5921263116071152e-05,
      "loss": 0.6242,
      "step": 5330
    },
    {
      "epoch": 0.6746393318147305,
      "grad_norm": 0.9720457196235657,
      "learning_rate": 1.5919818846860565e-05,
      "loss": 0.6414,
      "step": 5331
    },
    {
      "epoch": 0.674765882055176,
      "grad_norm": 0.9499323964118958,
      "learning_rate": 1.5918374387522742e-05,
      "loss": 0.5966,
      "step": 5332
    },
    {
      "epoch": 0.6748924322956213,
      "grad_norm": 1.0356323719024658,
      "learning_rate": 1.5916929738104083e-05,
      "loss": 0.6299,
      "step": 5333
    },
    {
      "epoch": 0.6750189825360668,
      "grad_norm": 1.8094673156738281,
      "learning_rate": 1.5915484898650976e-05,
      "loss": 0.6188,
      "step": 5334
    },
    {
      "epoch": 0.6751455327765122,
      "grad_norm": 1.0068401098251343,
      "learning_rate": 1.591403986920983e-05,
      "loss": 0.5988,
      "step": 5335
    },
    {
      "epoch": 0.6752720830169577,
      "grad_norm": 0.9453700184822083,
      "learning_rate": 1.5912594649827053e-05,
      "loss": 0.6143,
      "step": 5336
    },
    {
      "epoch": 0.6753986332574032,
      "grad_norm": 0.8903030157089233,
      "learning_rate": 1.591114924054906e-05,
      "loss": 0.5802,
      "step": 5337
    },
    {
      "epoch": 0.6755251834978486,
      "grad_norm": 0.9467645287513733,
      "learning_rate": 1.5909703641422282e-05,
      "loss": 0.5936,
      "step": 5338
    },
    {
      "epoch": 0.6756517337382941,
      "grad_norm": 1.0296049118041992,
      "learning_rate": 1.590825785249314e-05,
      "loss": 0.6186,
      "step": 5339
    },
    {
      "epoch": 0.6757782839787395,
      "grad_norm": 0.9386209845542908,
      "learning_rate": 1.5906811873808067e-05,
      "loss": 0.6086,
      "step": 5340
    },
    {
      "epoch": 0.675904834219185,
      "grad_norm": 0.9179689288139343,
      "learning_rate": 1.5905365705413503e-05,
      "loss": 0.6044,
      "step": 5341
    },
    {
      "epoch": 0.6760313844596305,
      "grad_norm": 0.9375709295272827,
      "learning_rate": 1.5903919347355907e-05,
      "loss": 0.6056,
      "step": 5342
    },
    {
      "epoch": 0.6761579347000759,
      "grad_norm": 1.028432846069336,
      "learning_rate": 1.5902472799681717e-05,
      "loss": 0.6418,
      "step": 5343
    },
    {
      "epoch": 0.6762844849405214,
      "grad_norm": 0.9514113664627075,
      "learning_rate": 1.5901026062437397e-05,
      "loss": 0.6266,
      "step": 5344
    },
    {
      "epoch": 0.6764110351809668,
      "grad_norm": 1.0102382898330688,
      "learning_rate": 1.5899579135669415e-05,
      "loss": 0.6226,
      "step": 5345
    },
    {
      "epoch": 0.6765375854214123,
      "grad_norm": 1.005203127861023,
      "learning_rate": 1.5898132019424238e-05,
      "loss": 0.6493,
      "step": 5346
    },
    {
      "epoch": 0.6766641356618578,
      "grad_norm": 1.022031545639038,
      "learning_rate": 1.5896684713748348e-05,
      "loss": 0.6131,
      "step": 5347
    },
    {
      "epoch": 0.6767906859023032,
      "grad_norm": 0.9757088422775269,
      "learning_rate": 1.589523721868822e-05,
      "loss": 0.6712,
      "step": 5348
    },
    {
      "epoch": 0.6769172361427487,
      "grad_norm": 0.9594284892082214,
      "learning_rate": 1.5893789534290355e-05,
      "loss": 0.6301,
      "step": 5349
    },
    {
      "epoch": 0.6770437863831942,
      "grad_norm": 1.0156570672988892,
      "learning_rate": 1.5892341660601235e-05,
      "loss": 0.6632,
      "step": 5350
    },
    {
      "epoch": 0.6771703366236396,
      "grad_norm": 0.9374982118606567,
      "learning_rate": 1.589089359766737e-05,
      "loss": 0.6203,
      "step": 5351
    },
    {
      "epoch": 0.6772968868640851,
      "grad_norm": 0.9233638048171997,
      "learning_rate": 1.5889445345535267e-05,
      "loss": 0.6474,
      "step": 5352
    },
    {
      "epoch": 0.6774234371045305,
      "grad_norm": 0.9850400686264038,
      "learning_rate": 1.588799690425144e-05,
      "loss": 0.6433,
      "step": 5353
    },
    {
      "epoch": 0.677549987344976,
      "grad_norm": 0.9652509689331055,
      "learning_rate": 1.5886548273862404e-05,
      "loss": 0.6131,
      "step": 5354
    },
    {
      "epoch": 0.6776765375854215,
      "grad_norm": 0.9618252515792847,
      "learning_rate": 1.5885099454414686e-05,
      "loss": 0.6197,
      "step": 5355
    },
    {
      "epoch": 0.6778030878258668,
      "grad_norm": 0.9758585691452026,
      "learning_rate": 1.5883650445954822e-05,
      "loss": 0.598,
      "step": 5356
    },
    {
      "epoch": 0.6779296380663123,
      "grad_norm": 1.0235531330108643,
      "learning_rate": 1.588220124852935e-05,
      "loss": 0.5803,
      "step": 5357
    },
    {
      "epoch": 0.6780561883067577,
      "grad_norm": 1.058462142944336,
      "learning_rate": 1.5880751862184806e-05,
      "loss": 0.6489,
      "step": 5358
    },
    {
      "epoch": 0.6781827385472032,
      "grad_norm": 0.9343885779380798,
      "learning_rate": 1.587930228696775e-05,
      "loss": 0.6049,
      "step": 5359
    },
    {
      "epoch": 0.6783092887876487,
      "grad_norm": 0.9685532450675964,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 0.5781,
      "step": 5360
    },
    {
      "epoch": 0.6784358390280941,
      "grad_norm": 1.023666262626648,
      "learning_rate": 1.587640257010232e-05,
      "loss": 0.6386,
      "step": 5361
    },
    {
      "epoch": 0.6785623892685396,
      "grad_norm": 0.9609639048576355,
      "learning_rate": 1.5874952428547072e-05,
      "loss": 0.6135,
      "step": 5362
    },
    {
      "epoch": 0.678688939508985,
      "grad_norm": 0.9123840928077698,
      "learning_rate": 1.587350209830557e-05,
      "loss": 0.6118,
      "step": 5363
    },
    {
      "epoch": 0.6788154897494305,
      "grad_norm": 1.0190818309783936,
      "learning_rate": 1.5872051579424393e-05,
      "loss": 0.6525,
      "step": 5364
    },
    {
      "epoch": 0.678942039989876,
      "grad_norm": 0.9352638721466064,
      "learning_rate": 1.5870600871950126e-05,
      "loss": 0.5961,
      "step": 5365
    },
    {
      "epoch": 0.6790685902303214,
      "grad_norm": 0.9974629282951355,
      "learning_rate": 1.5869149975929364e-05,
      "loss": 0.6232,
      "step": 5366
    },
    {
      "epoch": 0.6791951404707669,
      "grad_norm": 1.0189381837844849,
      "learning_rate": 1.5867698891408704e-05,
      "loss": 0.6378,
      "step": 5367
    },
    {
      "epoch": 0.6793216907112124,
      "grad_norm": 0.9407010078430176,
      "learning_rate": 1.586624761843475e-05,
      "loss": 0.5897,
      "step": 5368
    },
    {
      "epoch": 0.6794482409516578,
      "grad_norm": 1.0377521514892578,
      "learning_rate": 1.5864796157054115e-05,
      "loss": 0.588,
      "step": 5369
    },
    {
      "epoch": 0.6795747911921033,
      "grad_norm": 0.9413378238677979,
      "learning_rate": 1.586334450731341e-05,
      "loss": 0.6029,
      "step": 5370
    },
    {
      "epoch": 0.6797013414325487,
      "grad_norm": 0.8946295380592346,
      "learning_rate": 1.5861892669259264e-05,
      "loss": 0.5947,
      "step": 5371
    },
    {
      "epoch": 0.6798278916729942,
      "grad_norm": 0.9970521330833435,
      "learning_rate": 1.5860440642938306e-05,
      "loss": 0.6368,
      "step": 5372
    },
    {
      "epoch": 0.6799544419134397,
      "grad_norm": 0.9343540072441101,
      "learning_rate": 1.5858988428397168e-05,
      "loss": 0.6258,
      "step": 5373
    },
    {
      "epoch": 0.6800809921538851,
      "grad_norm": 0.9481945037841797,
      "learning_rate": 1.5857536025682487e-05,
      "loss": 0.649,
      "step": 5374
    },
    {
      "epoch": 0.6802075423943306,
      "grad_norm": 0.8952786922454834,
      "learning_rate": 1.5856083434840917e-05,
      "loss": 0.5971,
      "step": 5375
    },
    {
      "epoch": 0.680334092634776,
      "grad_norm": 0.9548031091690063,
      "learning_rate": 1.5854630655919107e-05,
      "loss": 0.6015,
      "step": 5376
    },
    {
      "epoch": 0.6804606428752215,
      "grad_norm": 0.9302863478660583,
      "learning_rate": 1.585317768896372e-05,
      "loss": 0.6665,
      "step": 5377
    },
    {
      "epoch": 0.680587193115667,
      "grad_norm": 1.0701757669448853,
      "learning_rate": 1.5851724534021413e-05,
      "loss": 0.5544,
      "step": 5378
    },
    {
      "epoch": 0.6807137433561123,
      "grad_norm": 0.9196771383285522,
      "learning_rate": 1.5850271191138867e-05,
      "loss": 0.6362,
      "step": 5379
    },
    {
      "epoch": 0.6808402935965578,
      "grad_norm": 1.0138314962387085,
      "learning_rate": 1.5848817660362752e-05,
      "loss": 0.6198,
      "step": 5380
    },
    {
      "epoch": 0.6809668438370033,
      "grad_norm": 0.9624118804931641,
      "learning_rate": 1.5847363941739754e-05,
      "loss": 0.6133,
      "step": 5381
    },
    {
      "epoch": 0.6810933940774487,
      "grad_norm": 0.9930436015129089,
      "learning_rate": 1.5845910035316562e-05,
      "loss": 0.6312,
      "step": 5382
    },
    {
      "epoch": 0.6812199443178942,
      "grad_norm": 0.9415516257286072,
      "learning_rate": 1.584445594113987e-05,
      "loss": 0.6099,
      "step": 5383
    },
    {
      "epoch": 0.6813464945583396,
      "grad_norm": 0.9418936371803284,
      "learning_rate": 1.584300165925638e-05,
      "loss": 0.6152,
      "step": 5384
    },
    {
      "epoch": 0.6814730447987851,
      "grad_norm": 0.9635658860206604,
      "learning_rate": 1.58415471897128e-05,
      "loss": 0.6028,
      "step": 5385
    },
    {
      "epoch": 0.6815995950392306,
      "grad_norm": 0.9788090586662292,
      "learning_rate": 1.584009253255584e-05,
      "loss": 0.6468,
      "step": 5386
    },
    {
      "epoch": 0.681726145279676,
      "grad_norm": 0.9856061339378357,
      "learning_rate": 1.5838637687832226e-05,
      "loss": 0.6416,
      "step": 5387
    },
    {
      "epoch": 0.6818526955201215,
      "grad_norm": 0.9150157570838928,
      "learning_rate": 1.583718265558868e-05,
      "loss": 0.6145,
      "step": 5388
    },
    {
      "epoch": 0.6819792457605669,
      "grad_norm": 0.9979318380355835,
      "learning_rate": 1.5835727435871932e-05,
      "loss": 0.6121,
      "step": 5389
    },
    {
      "epoch": 0.6821057960010124,
      "grad_norm": 1.0161608457565308,
      "learning_rate": 1.5834272028728717e-05,
      "loss": 0.6302,
      "step": 5390
    },
    {
      "epoch": 0.6822323462414579,
      "grad_norm": 1.000433087348938,
      "learning_rate": 1.5832816434205784e-05,
      "loss": 0.6261,
      "step": 5391
    },
    {
      "epoch": 0.6823588964819033,
      "grad_norm": 0.9305796027183533,
      "learning_rate": 1.5831360652349878e-05,
      "loss": 0.6108,
      "step": 5392
    },
    {
      "epoch": 0.6824854467223488,
      "grad_norm": 0.8969127535820007,
      "learning_rate": 1.582990468320776e-05,
      "loss": 0.6075,
      "step": 5393
    },
    {
      "epoch": 0.6826119969627942,
      "grad_norm": 0.9757826924324036,
      "learning_rate": 1.5828448526826186e-05,
      "loss": 0.6187,
      "step": 5394
    },
    {
      "epoch": 0.6827385472032397,
      "grad_norm": 0.9165109992027283,
      "learning_rate": 1.5826992183251927e-05,
      "loss": 0.5812,
      "step": 5395
    },
    {
      "epoch": 0.6828650974436852,
      "grad_norm": 1.0136733055114746,
      "learning_rate": 1.582553565253175e-05,
      "loss": 0.5909,
      "step": 5396
    },
    {
      "epoch": 0.6829916476841306,
      "grad_norm": 0.9819451570510864,
      "learning_rate": 1.582407893471244e-05,
      "loss": 0.6444,
      "step": 5397
    },
    {
      "epoch": 0.6831181979245761,
      "grad_norm": 0.9772393703460693,
      "learning_rate": 1.5822622029840788e-05,
      "loss": 0.6273,
      "step": 5398
    },
    {
      "epoch": 0.6832447481650216,
      "grad_norm": 0.9273337125778198,
      "learning_rate": 1.5821164937963572e-05,
      "loss": 0.6175,
      "step": 5399
    },
    {
      "epoch": 0.683371298405467,
      "grad_norm": 0.9519692659378052,
      "learning_rate": 1.5819707659127605e-05,
      "loss": 0.6193,
      "step": 5400
    },
    {
      "epoch": 0.6834978486459125,
      "grad_norm": 0.9015218019485474,
      "learning_rate": 1.5818250193379676e-05,
      "loss": 0.5941,
      "step": 5401
    },
    {
      "epoch": 0.6836243988863578,
      "grad_norm": 1.2322144508361816,
      "learning_rate": 1.58167925407666e-05,
      "loss": 0.7019,
      "step": 5402
    },
    {
      "epoch": 0.6837509491268033,
      "grad_norm": 0.9350510835647583,
      "learning_rate": 1.58153347013352e-05,
      "loss": 0.5809,
      "step": 5403
    },
    {
      "epoch": 0.6838774993672488,
      "grad_norm": 0.9795088768005371,
      "learning_rate": 1.581387667513229e-05,
      "loss": 0.6306,
      "step": 5404
    },
    {
      "epoch": 0.6840040496076942,
      "grad_norm": 1.0177541971206665,
      "learning_rate": 1.5812418462204694e-05,
      "loss": 0.6224,
      "step": 5405
    },
    {
      "epoch": 0.6841305998481397,
      "grad_norm": 0.8842471241950989,
      "learning_rate": 1.581096006259925e-05,
      "loss": 0.5873,
      "step": 5406
    },
    {
      "epoch": 0.6842571500885851,
      "grad_norm": 0.9116804003715515,
      "learning_rate": 1.58095014763628e-05,
      "loss": 0.5884,
      "step": 5407
    },
    {
      "epoch": 0.6843837003290306,
      "grad_norm": 0.9691157937049866,
      "learning_rate": 1.5808042703542185e-05,
      "loss": 0.5985,
      "step": 5408
    },
    {
      "epoch": 0.6845102505694761,
      "grad_norm": 0.8915805220603943,
      "learning_rate": 1.5806583744184258e-05,
      "loss": 0.6218,
      "step": 5409
    },
    {
      "epoch": 0.6846368008099215,
      "grad_norm": 0.964680552482605,
      "learning_rate": 1.5805124598335883e-05,
      "loss": 0.6191,
      "step": 5410
    },
    {
      "epoch": 0.684763351050367,
      "grad_norm": 0.9292550683021545,
      "learning_rate": 1.5803665266043914e-05,
      "loss": 0.6316,
      "step": 5411
    },
    {
      "epoch": 0.6848899012908124,
      "grad_norm": 0.9587753415107727,
      "learning_rate": 1.580220574735522e-05,
      "loss": 0.601,
      "step": 5412
    },
    {
      "epoch": 0.6850164515312579,
      "grad_norm": 0.8788020014762878,
      "learning_rate": 1.5800746042316687e-05,
      "loss": 0.6017,
      "step": 5413
    },
    {
      "epoch": 0.6851430017717034,
      "grad_norm": 0.9049255847930908,
      "learning_rate": 1.5799286150975184e-05,
      "loss": 0.6177,
      "step": 5414
    },
    {
      "epoch": 0.6852695520121488,
      "grad_norm": 0.8756937384605408,
      "learning_rate": 1.5797826073377608e-05,
      "loss": 0.6112,
      "step": 5415
    },
    {
      "epoch": 0.6853961022525943,
      "grad_norm": 0.9349949955940247,
      "learning_rate": 1.5796365809570853e-05,
      "loss": 0.5856,
      "step": 5416
    },
    {
      "epoch": 0.6855226524930398,
      "grad_norm": 1.007418155670166,
      "learning_rate": 1.579490535960181e-05,
      "loss": 0.5971,
      "step": 5417
    },
    {
      "epoch": 0.6856492027334852,
      "grad_norm": 0.9928464889526367,
      "learning_rate": 1.5793444723517387e-05,
      "loss": 0.6341,
      "step": 5418
    },
    {
      "epoch": 0.6857757529739307,
      "grad_norm": 0.9606293439865112,
      "learning_rate": 1.5791983901364497e-05,
      "loss": 0.6575,
      "step": 5419
    },
    {
      "epoch": 0.6859023032143761,
      "grad_norm": 0.9934487342834473,
      "learning_rate": 1.579052289319006e-05,
      "loss": 0.6091,
      "step": 5420
    },
    {
      "epoch": 0.6860288534548216,
      "grad_norm": 0.9301649928092957,
      "learning_rate": 1.5789061699040995e-05,
      "loss": 0.5675,
      "step": 5421
    },
    {
      "epoch": 0.6861554036952671,
      "grad_norm": 0.9354473948478699,
      "learning_rate": 1.5787600318964233e-05,
      "loss": 0.6054,
      "step": 5422
    },
    {
      "epoch": 0.6862819539357125,
      "grad_norm": 1.0151543617248535,
      "learning_rate": 1.578613875300671e-05,
      "loss": 0.5994,
      "step": 5423
    },
    {
      "epoch": 0.686408504176158,
      "grad_norm": 1.0561214685440063,
      "learning_rate": 1.5784677001215365e-05,
      "loss": 0.6825,
      "step": 5424
    },
    {
      "epoch": 0.6865350544166033,
      "grad_norm": 0.9135820865631104,
      "learning_rate": 1.5783215063637148e-05,
      "loss": 0.6152,
      "step": 5425
    },
    {
      "epoch": 0.6866616046570488,
      "grad_norm": 1.359537959098816,
      "learning_rate": 1.5781752940319007e-05,
      "loss": 0.6464,
      "step": 5426
    },
    {
      "epoch": 0.6867881548974943,
      "grad_norm": 0.9982617497444153,
      "learning_rate": 1.5780290631307907e-05,
      "loss": 0.6717,
      "step": 5427
    },
    {
      "epoch": 0.6869147051379397,
      "grad_norm": 0.9932838082313538,
      "learning_rate": 1.5778828136650814e-05,
      "loss": 0.6181,
      "step": 5428
    },
    {
      "epoch": 0.6870412553783852,
      "grad_norm": 0.976707398891449,
      "learning_rate": 1.5777365456394694e-05,
      "loss": 0.6212,
      "step": 5429
    },
    {
      "epoch": 0.6871678056188307,
      "grad_norm": 1.0642836093902588,
      "learning_rate": 1.5775902590586524e-05,
      "loss": 0.5895,
      "step": 5430
    },
    {
      "epoch": 0.6872943558592761,
      "grad_norm": 1.085463523864746,
      "learning_rate": 1.577443953927329e-05,
      "loss": 0.6061,
      "step": 5431
    },
    {
      "epoch": 0.6874209060997216,
      "grad_norm": 1.0771642923355103,
      "learning_rate": 1.5772976302501978e-05,
      "loss": 0.6171,
      "step": 5432
    },
    {
      "epoch": 0.687547456340167,
      "grad_norm": 1.0711747407913208,
      "learning_rate": 1.5771512880319586e-05,
      "loss": 0.6252,
      "step": 5433
    },
    {
      "epoch": 0.6876740065806125,
      "grad_norm": 0.9088315963745117,
      "learning_rate": 1.5770049272773117e-05,
      "loss": 0.5887,
      "step": 5434
    },
    {
      "epoch": 0.687800556821058,
      "grad_norm": 1.0136953592300415,
      "learning_rate": 1.576858547990957e-05,
      "loss": 0.6055,
      "step": 5435
    },
    {
      "epoch": 0.6879271070615034,
      "grad_norm": 0.9462160468101501,
      "learning_rate": 1.5767121501775963e-05,
      "loss": 0.608,
      "step": 5436
    },
    {
      "epoch": 0.6880536573019489,
      "grad_norm": 0.9472612738609314,
      "learning_rate": 1.5765657338419315e-05,
      "loss": 0.5992,
      "step": 5437
    },
    {
      "epoch": 0.6881802075423943,
      "grad_norm": 1.163434624671936,
      "learning_rate": 1.5764192989886645e-05,
      "loss": 0.6054,
      "step": 5438
    },
    {
      "epoch": 0.6883067577828398,
      "grad_norm": 0.9214063882827759,
      "learning_rate": 1.5762728456224994e-05,
      "loss": 0.5914,
      "step": 5439
    },
    {
      "epoch": 0.6884333080232853,
      "grad_norm": 0.9602842926979065,
      "learning_rate": 1.576126373748139e-05,
      "loss": 0.6022,
      "step": 5440
    },
    {
      "epoch": 0.6885598582637307,
      "grad_norm": 0.9245358109474182,
      "learning_rate": 1.5759798833702874e-05,
      "loss": 0.5811,
      "step": 5441
    },
    {
      "epoch": 0.6886864085041762,
      "grad_norm": 0.961592972278595,
      "learning_rate": 1.5758333744936503e-05,
      "loss": 0.6374,
      "step": 5442
    },
    {
      "epoch": 0.6888129587446216,
      "grad_norm": 0.9201575517654419,
      "learning_rate": 1.5756868471229325e-05,
      "loss": 0.6086,
      "step": 5443
    },
    {
      "epoch": 0.6889395089850671,
      "grad_norm": 0.9470493793487549,
      "learning_rate": 1.5755403012628404e-05,
      "loss": 0.6561,
      "step": 5444
    },
    {
      "epoch": 0.6890660592255126,
      "grad_norm": 1.0054875612258911,
      "learning_rate": 1.57539373691808e-05,
      "loss": 0.585,
      "step": 5445
    },
    {
      "epoch": 0.689192609465958,
      "grad_norm": 0.9847509264945984,
      "learning_rate": 1.5752471540933593e-05,
      "loss": 0.6633,
      "step": 5446
    },
    {
      "epoch": 0.6893191597064035,
      "grad_norm": 1.0020802021026611,
      "learning_rate": 1.5751005527933856e-05,
      "loss": 0.592,
      "step": 5447
    },
    {
      "epoch": 0.689445709946849,
      "grad_norm": 0.9534604549407959,
      "learning_rate": 1.5749539330228675e-05,
      "loss": 0.6148,
      "step": 5448
    },
    {
      "epoch": 0.6895722601872943,
      "grad_norm": 0.9739737510681152,
      "learning_rate": 1.5748072947865138e-05,
      "loss": 0.6353,
      "step": 5449
    },
    {
      "epoch": 0.6896988104277398,
      "grad_norm": 1.0082886219024658,
      "learning_rate": 1.5746606380890343e-05,
      "loss": 0.6148,
      "step": 5450
    },
    {
      "epoch": 0.6898253606681852,
      "grad_norm": 0.9699578881263733,
      "learning_rate": 1.5745139629351395e-05,
      "loss": 0.5886,
      "step": 5451
    },
    {
      "epoch": 0.6899519109086307,
      "grad_norm": 1.1883091926574707,
      "learning_rate": 1.5743672693295396e-05,
      "loss": 0.5681,
      "step": 5452
    },
    {
      "epoch": 0.6900784611490762,
      "grad_norm": 1.0565482378005981,
      "learning_rate": 1.574220557276946e-05,
      "loss": 0.6052,
      "step": 5453
    },
    {
      "epoch": 0.6902050113895216,
      "grad_norm": 0.9792020320892334,
      "learning_rate": 1.574073826782071e-05,
      "loss": 0.5731,
      "step": 5454
    },
    {
      "epoch": 0.6903315616299671,
      "grad_norm": 0.9265932440757751,
      "learning_rate": 1.5739270778496267e-05,
      "loss": 0.6056,
      "step": 5455
    },
    {
      "epoch": 0.6904581118704125,
      "grad_norm": 0.9476988315582275,
      "learning_rate": 1.5737803104843267e-05,
      "loss": 0.6172,
      "step": 5456
    },
    {
      "epoch": 0.690584662110858,
      "grad_norm": 0.9744604229927063,
      "learning_rate": 1.5736335246908847e-05,
      "loss": 0.6515,
      "step": 5457
    },
    {
      "epoch": 0.6907112123513035,
      "grad_norm": 0.9429535269737244,
      "learning_rate": 1.5734867204740146e-05,
      "loss": 0.5902,
      "step": 5458
    },
    {
      "epoch": 0.6908377625917489,
      "grad_norm": 0.8966645002365112,
      "learning_rate": 1.573339897838432e-05,
      "loss": 0.5944,
      "step": 5459
    },
    {
      "epoch": 0.6909643128321944,
      "grad_norm": 1.0380401611328125,
      "learning_rate": 1.573193056788852e-05,
      "loss": 0.6709,
      "step": 5460
    },
    {
      "epoch": 0.6910908630726398,
      "grad_norm": 0.9342979192733765,
      "learning_rate": 1.5730461973299903e-05,
      "loss": 0.5634,
      "step": 5461
    },
    {
      "epoch": 0.6912174133130853,
      "grad_norm": 0.9407649040222168,
      "learning_rate": 1.5728993194665646e-05,
      "loss": 0.6274,
      "step": 5462
    },
    {
      "epoch": 0.6913439635535308,
      "grad_norm": 1.0451548099517822,
      "learning_rate": 1.5727524232032913e-05,
      "loss": 0.7018,
      "step": 5463
    },
    {
      "epoch": 0.6914705137939762,
      "grad_norm": 0.9321942925453186,
      "learning_rate": 1.5726055085448886e-05,
      "loss": 0.5924,
      "step": 5464
    },
    {
      "epoch": 0.6915970640344217,
      "grad_norm": 0.9364402890205383,
      "learning_rate": 1.572458575496075e-05,
      "loss": 0.5848,
      "step": 5465
    },
    {
      "epoch": 0.6917236142748672,
      "grad_norm": 0.9547738432884216,
      "learning_rate": 1.5723116240615693e-05,
      "loss": 0.6136,
      "step": 5466
    },
    {
      "epoch": 0.6918501645153126,
      "grad_norm": 0.9030520915985107,
      "learning_rate": 1.5721646542460914e-05,
      "loss": 0.5853,
      "step": 5467
    },
    {
      "epoch": 0.6919767147557581,
      "grad_norm": 0.9260213971138,
      "learning_rate": 1.572017666054362e-05,
      "loss": 0.6145,
      "step": 5468
    },
    {
      "epoch": 0.6921032649962034,
      "grad_norm": 0.9943070411682129,
      "learning_rate": 1.5718706594911014e-05,
      "loss": 0.5947,
      "step": 5469
    },
    {
      "epoch": 0.692229815236649,
      "grad_norm": 0.9167776703834534,
      "learning_rate": 1.5717236345610304e-05,
      "loss": 0.5619,
      "step": 5470
    },
    {
      "epoch": 0.6923563654770944,
      "grad_norm": 1.0175435543060303,
      "learning_rate": 1.571576591268872e-05,
      "loss": 0.6916,
      "step": 5471
    },
    {
      "epoch": 0.6924829157175398,
      "grad_norm": 0.9570114016532898,
      "learning_rate": 1.5714295296193486e-05,
      "loss": 0.6498,
      "step": 5472
    },
    {
      "epoch": 0.6926094659579853,
      "grad_norm": 0.9894493222236633,
      "learning_rate": 1.571282449617183e-05,
      "loss": 0.6035,
      "step": 5473
    },
    {
      "epoch": 0.6927360161984307,
      "grad_norm": 1.0277701616287231,
      "learning_rate": 1.5711353512671e-05,
      "loss": 0.6052,
      "step": 5474
    },
    {
      "epoch": 0.6928625664388762,
      "grad_norm": 0.9351342916488647,
      "learning_rate": 1.5709882345738226e-05,
      "loss": 0.6056,
      "step": 5475
    },
    {
      "epoch": 0.6929891166793217,
      "grad_norm": 0.9632000923156738,
      "learning_rate": 1.570841099542076e-05,
      "loss": 0.6339,
      "step": 5476
    },
    {
      "epoch": 0.6931156669197671,
      "grad_norm": 0.8976531624794006,
      "learning_rate": 1.570693946176587e-05,
      "loss": 0.5992,
      "step": 5477
    },
    {
      "epoch": 0.6932422171602126,
      "grad_norm": 0.9575624465942383,
      "learning_rate": 1.57054677448208e-05,
      "loss": 0.6564,
      "step": 5478
    },
    {
      "epoch": 0.6933687674006581,
      "grad_norm": 1.0146814584732056,
      "learning_rate": 1.5703995844632832e-05,
      "loss": 0.6527,
      "step": 5479
    },
    {
      "epoch": 0.6934953176411035,
      "grad_norm": 0.9479714035987854,
      "learning_rate": 1.5702523761249232e-05,
      "loss": 0.6236,
      "step": 5480
    },
    {
      "epoch": 0.693621867881549,
      "grad_norm": 0.9229686260223389,
      "learning_rate": 1.5701051494717278e-05,
      "loss": 0.5911,
      "step": 5481
    },
    {
      "epoch": 0.6937484181219944,
      "grad_norm": 1.10174560546875,
      "learning_rate": 1.5699579045084254e-05,
      "loss": 0.6653,
      "step": 5482
    },
    {
      "epoch": 0.6938749683624399,
      "grad_norm": 0.9616940021514893,
      "learning_rate": 1.569810641239746e-05,
      "loss": 0.6172,
      "step": 5483
    },
    {
      "epoch": 0.6940015186028854,
      "grad_norm": 0.9165851473808289,
      "learning_rate": 1.569663359670418e-05,
      "loss": 0.6045,
      "step": 5484
    },
    {
      "epoch": 0.6941280688433308,
      "grad_norm": 0.9087163805961609,
      "learning_rate": 1.5695160598051728e-05,
      "loss": 0.5914,
      "step": 5485
    },
    {
      "epoch": 0.6942546190837763,
      "grad_norm": 1.1711386442184448,
      "learning_rate": 1.56936874164874e-05,
      "loss": 0.6078,
      "step": 5486
    },
    {
      "epoch": 0.6943811693242217,
      "grad_norm": 0.9599922299385071,
      "learning_rate": 1.5692214052058523e-05,
      "loss": 0.6393,
      "step": 5487
    },
    {
      "epoch": 0.6945077195646672,
      "grad_norm": 1.6775507926940918,
      "learning_rate": 1.5690740504812407e-05,
      "loss": 0.6113,
      "step": 5488
    },
    {
      "epoch": 0.6946342698051127,
      "grad_norm": 0.9649134874343872,
      "learning_rate": 1.5689266774796384e-05,
      "loss": 0.6097,
      "step": 5489
    },
    {
      "epoch": 0.6947608200455581,
      "grad_norm": 0.9809866547584534,
      "learning_rate": 1.5687792862057788e-05,
      "loss": 0.6026,
      "step": 5490
    },
    {
      "epoch": 0.6948873702860036,
      "grad_norm": 0.9997078776359558,
      "learning_rate": 1.5686318766643952e-05,
      "loss": 0.6354,
      "step": 5491
    },
    {
      "epoch": 0.695013920526449,
      "grad_norm": 0.982336699962616,
      "learning_rate": 1.5684844488602217e-05,
      "loss": 0.6171,
      "step": 5492
    },
    {
      "epoch": 0.6951404707668944,
      "grad_norm": 1.003830909729004,
      "learning_rate": 1.5683370027979942e-05,
      "loss": 0.6392,
      "step": 5493
    },
    {
      "epoch": 0.69526702100734,
      "grad_norm": 1.0004825592041016,
      "learning_rate": 1.568189538482447e-05,
      "loss": 0.6072,
      "step": 5494
    },
    {
      "epoch": 0.6953935712477853,
      "grad_norm": 0.9406870603561401,
      "learning_rate": 1.5680420559183174e-05,
      "loss": 0.6086,
      "step": 5495
    },
    {
      "epoch": 0.6955201214882308,
      "grad_norm": 0.8984605669975281,
      "learning_rate": 1.5678945551103413e-05,
      "loss": 0.5994,
      "step": 5496
    },
    {
      "epoch": 0.6956466717286763,
      "grad_norm": 1.0025242567062378,
      "learning_rate": 1.5677470360632566e-05,
      "loss": 0.6761,
      "step": 5497
    },
    {
      "epoch": 0.6957732219691217,
      "grad_norm": 0.8563474416732788,
      "learning_rate": 1.5675994987818003e-05,
      "loss": 0.6058,
      "step": 5498
    },
    {
      "epoch": 0.6958997722095672,
      "grad_norm": 0.9012765884399414,
      "learning_rate": 1.567451943270712e-05,
      "loss": 0.6285,
      "step": 5499
    },
    {
      "epoch": 0.6960263224500126,
      "grad_norm": 1.01656174659729,
      "learning_rate": 1.5673043695347298e-05,
      "loss": 0.6161,
      "step": 5500
    },
    {
      "epoch": 0.6961528726904581,
      "grad_norm": 0.9303178787231445,
      "learning_rate": 1.5671567775785937e-05,
      "loss": 0.6159,
      "step": 5501
    },
    {
      "epoch": 0.6962794229309036,
      "grad_norm": 0.8925102353096008,
      "learning_rate": 1.5670091674070445e-05,
      "loss": 0.6101,
      "step": 5502
    },
    {
      "epoch": 0.696405973171349,
      "grad_norm": 0.9721536636352539,
      "learning_rate": 1.566861539024822e-05,
      "loss": 0.6216,
      "step": 5503
    },
    {
      "epoch": 0.6965325234117945,
      "grad_norm": 0.9243012070655823,
      "learning_rate": 1.566713892436668e-05,
      "loss": 0.624,
      "step": 5504
    },
    {
      "epoch": 0.6966590736522399,
      "grad_norm": 0.9331715106964111,
      "learning_rate": 1.5665662276473246e-05,
      "loss": 0.6359,
      "step": 5505
    },
    {
      "epoch": 0.6967856238926854,
      "grad_norm": 0.9074288010597229,
      "learning_rate": 1.5664185446615344e-05,
      "loss": 0.6265,
      "step": 5506
    },
    {
      "epoch": 0.6969121741331309,
      "grad_norm": 0.9562578797340393,
      "learning_rate": 1.56627084348404e-05,
      "loss": 0.6209,
      "step": 5507
    },
    {
      "epoch": 0.6970387243735763,
      "grad_norm": 0.903859555721283,
      "learning_rate": 1.566123124119586e-05,
      "loss": 0.5995,
      "step": 5508
    },
    {
      "epoch": 0.6971652746140218,
      "grad_norm": 0.90989750623703,
      "learning_rate": 1.565975386572916e-05,
      "loss": 0.5766,
      "step": 5509
    },
    {
      "epoch": 0.6972918248544672,
      "grad_norm": 1.569305658340454,
      "learning_rate": 1.5658276308487753e-05,
      "loss": 0.5857,
      "step": 5510
    },
    {
      "epoch": 0.6974183750949127,
      "grad_norm": 1.2913739681243896,
      "learning_rate": 1.565679856951909e-05,
      "loss": 0.5944,
      "step": 5511
    },
    {
      "epoch": 0.6975449253353582,
      "grad_norm": 1.0047142505645752,
      "learning_rate": 1.5655320648870637e-05,
      "loss": 0.6144,
      "step": 5512
    },
    {
      "epoch": 0.6976714755758036,
      "grad_norm": 0.9662113785743713,
      "learning_rate": 1.5653842546589853e-05,
      "loss": 0.5595,
      "step": 5513
    },
    {
      "epoch": 0.6977980258162491,
      "grad_norm": 1.0931588411331177,
      "learning_rate": 1.565236426272422e-05,
      "loss": 0.625,
      "step": 5514
    },
    {
      "epoch": 0.6979245760566946,
      "grad_norm": 1.0513335466384888,
      "learning_rate": 1.565088579732121e-05,
      "loss": 0.6374,
      "step": 5515
    },
    {
      "epoch": 0.69805112629714,
      "grad_norm": 0.9801070690155029,
      "learning_rate": 1.5649407150428303e-05,
      "loss": 0.6151,
      "step": 5516
    },
    {
      "epoch": 0.6981776765375854,
      "grad_norm": 0.9863511919975281,
      "learning_rate": 1.5647928322093e-05,
      "loss": 0.597,
      "step": 5517
    },
    {
      "epoch": 0.6983042267780308,
      "grad_norm": 0.9558197259902954,
      "learning_rate": 1.5646449312362787e-05,
      "loss": 0.6512,
      "step": 5518
    },
    {
      "epoch": 0.6984307770184763,
      "grad_norm": 1.1030064821243286,
      "learning_rate": 1.564497012128517e-05,
      "loss": 0.6404,
      "step": 5519
    },
    {
      "epoch": 0.6985573272589218,
      "grad_norm": 0.9518397450447083,
      "learning_rate": 1.5643490748907653e-05,
      "loss": 0.6308,
      "step": 5520
    },
    {
      "epoch": 0.6986838774993672,
      "grad_norm": 0.9875779151916504,
      "learning_rate": 1.564201119527775e-05,
      "loss": 0.6201,
      "step": 5521
    },
    {
      "epoch": 0.6988104277398127,
      "grad_norm": 0.9571218490600586,
      "learning_rate": 1.5640531460442985e-05,
      "loss": 0.6121,
      "step": 5522
    },
    {
      "epoch": 0.6989369779802581,
      "grad_norm": 0.9713514447212219,
      "learning_rate": 1.5639051544450878e-05,
      "loss": 0.6478,
      "step": 5523
    },
    {
      "epoch": 0.6990635282207036,
      "grad_norm": 0.9697023034095764,
      "learning_rate": 1.563757144734896e-05,
      "loss": 0.5913,
      "step": 5524
    },
    {
      "epoch": 0.6991900784611491,
      "grad_norm": 0.904420793056488,
      "learning_rate": 1.5636091169184768e-05,
      "loss": 0.6259,
      "step": 5525
    },
    {
      "epoch": 0.6993166287015945,
      "grad_norm": 0.9413772225379944,
      "learning_rate": 1.5634610710005844e-05,
      "loss": 0.6332,
      "step": 5526
    },
    {
      "epoch": 0.69944317894204,
      "grad_norm": 1.0228631496429443,
      "learning_rate": 1.5633130069859737e-05,
      "loss": 0.5841,
      "step": 5527
    },
    {
      "epoch": 0.6995697291824855,
      "grad_norm": 0.9115861058235168,
      "learning_rate": 1.5631649248794e-05,
      "loss": 0.6092,
      "step": 5528
    },
    {
      "epoch": 0.6996962794229309,
      "grad_norm": 0.962841272354126,
      "learning_rate": 1.5630168246856186e-05,
      "loss": 0.6255,
      "step": 5529
    },
    {
      "epoch": 0.6998228296633764,
      "grad_norm": 1.0558180809020996,
      "learning_rate": 1.562868706409387e-05,
      "loss": 0.6187,
      "step": 5530
    },
    {
      "epoch": 0.6999493799038218,
      "grad_norm": 1.0552526712417603,
      "learning_rate": 1.5627205700554627e-05,
      "loss": 0.6458,
      "step": 5531
    },
    {
      "epoch": 0.7000759301442673,
      "grad_norm": 0.9335843920707703,
      "learning_rate": 1.5625724156286023e-05,
      "loss": 0.5916,
      "step": 5532
    },
    {
      "epoch": 0.7002024803847128,
      "grad_norm": 0.9723650217056274,
      "learning_rate": 1.562424243133564e-05,
      "loss": 0.6058,
      "step": 5533
    },
    {
      "epoch": 0.7003290306251582,
      "grad_norm": 1.1410678625106812,
      "learning_rate": 1.5622760525751076e-05,
      "loss": 0.627,
      "step": 5534
    },
    {
      "epoch": 0.7004555808656037,
      "grad_norm": 0.9524587392807007,
      "learning_rate": 1.5621278439579924e-05,
      "loss": 0.5813,
      "step": 5535
    },
    {
      "epoch": 0.700582131106049,
      "grad_norm": 1.2749344110488892,
      "learning_rate": 1.5619796172869777e-05,
      "loss": 0.6253,
      "step": 5536
    },
    {
      "epoch": 0.7007086813464946,
      "grad_norm": 0.9010131359100342,
      "learning_rate": 1.561831372566825e-05,
      "loss": 0.5708,
      "step": 5537
    },
    {
      "epoch": 0.70083523158694,
      "grad_norm": 1.0691391229629517,
      "learning_rate": 1.5616831098022943e-05,
      "loss": 0.6356,
      "step": 5538
    },
    {
      "epoch": 0.7009617818273854,
      "grad_norm": 1.0113544464111328,
      "learning_rate": 1.561534828998149e-05,
      "loss": 0.6335,
      "step": 5539
    },
    {
      "epoch": 0.7010883320678309,
      "grad_norm": 1.1017500162124634,
      "learning_rate": 1.56138653015915e-05,
      "loss": 0.6091,
      "step": 5540
    },
    {
      "epoch": 0.7012148823082763,
      "grad_norm": 1.0787967443466187,
      "learning_rate": 1.561238213290061e-05,
      "loss": 0.5846,
      "step": 5541
    },
    {
      "epoch": 0.7013414325487218,
      "grad_norm": 0.9687598347663879,
      "learning_rate": 1.5610898783956448e-05,
      "loss": 0.6445,
      "step": 5542
    },
    {
      "epoch": 0.7014679827891673,
      "grad_norm": 1.0326303243637085,
      "learning_rate": 1.5609415254806665e-05,
      "loss": 0.6123,
      "step": 5543
    },
    {
      "epoch": 0.7015945330296127,
      "grad_norm": 0.9340121746063232,
      "learning_rate": 1.56079315454989e-05,
      "loss": 0.5907,
      "step": 5544
    },
    {
      "epoch": 0.7017210832700582,
      "grad_norm": 0.9866454005241394,
      "learning_rate": 1.5606447656080806e-05,
      "loss": 0.6235,
      "step": 5545
    },
    {
      "epoch": 0.7018476335105037,
      "grad_norm": 0.9342936277389526,
      "learning_rate": 1.5604963586600046e-05,
      "loss": 0.6195,
      "step": 5546
    },
    {
      "epoch": 0.7019741837509491,
      "grad_norm": 1.1159203052520752,
      "learning_rate": 1.560347933710428e-05,
      "loss": 0.6399,
      "step": 5547
    },
    {
      "epoch": 0.7021007339913946,
      "grad_norm": 1.3253896236419678,
      "learning_rate": 1.5601994907641178e-05,
      "loss": 0.6087,
      "step": 5548
    },
    {
      "epoch": 0.70222728423184,
      "grad_norm": 1.243253469467163,
      "learning_rate": 1.5600510298258415e-05,
      "loss": 0.6448,
      "step": 5549
    },
    {
      "epoch": 0.7023538344722855,
      "grad_norm": 0.9990518689155579,
      "learning_rate": 1.5599025509003673e-05,
      "loss": 0.6018,
      "step": 5550
    },
    {
      "epoch": 0.702480384712731,
      "grad_norm": 0.9094237089157104,
      "learning_rate": 1.5597540539924642e-05,
      "loss": 0.6027,
      "step": 5551
    },
    {
      "epoch": 0.7026069349531764,
      "grad_norm": 1.0450786352157593,
      "learning_rate": 1.559605539106901e-05,
      "loss": 0.6183,
      "step": 5552
    },
    {
      "epoch": 0.7027334851936219,
      "grad_norm": 0.971417248249054,
      "learning_rate": 1.559457006248448e-05,
      "loss": 0.6292,
      "step": 5553
    },
    {
      "epoch": 0.7028600354340673,
      "grad_norm": 0.923369288444519,
      "learning_rate": 1.5593084554218754e-05,
      "loss": 0.6049,
      "step": 5554
    },
    {
      "epoch": 0.7029865856745128,
      "grad_norm": 0.97089684009552,
      "learning_rate": 1.5591598866319538e-05,
      "loss": 0.6757,
      "step": 5555
    },
    {
      "epoch": 0.7031131359149583,
      "grad_norm": 0.9762014746665955,
      "learning_rate": 1.5590112998834558e-05,
      "loss": 0.6364,
      "step": 5556
    },
    {
      "epoch": 0.7032396861554037,
      "grad_norm": 0.8814507126808167,
      "learning_rate": 1.558862695181153e-05,
      "loss": 0.5935,
      "step": 5557
    },
    {
      "epoch": 0.7033662363958492,
      "grad_norm": 0.945438027381897,
      "learning_rate": 1.5587140725298176e-05,
      "loss": 0.6013,
      "step": 5558
    },
    {
      "epoch": 0.7034927866362946,
      "grad_norm": 0.9374029040336609,
      "learning_rate": 1.558565431934224e-05,
      "loss": 0.6066,
      "step": 5559
    },
    {
      "epoch": 0.70361933687674,
      "grad_norm": 0.8872487545013428,
      "learning_rate": 1.5584167733991456e-05,
      "loss": 0.587,
      "step": 5560
    },
    {
      "epoch": 0.7037458871171856,
      "grad_norm": 0.9674831628799438,
      "learning_rate": 1.5582680969293565e-05,
      "loss": 0.6259,
      "step": 5561
    },
    {
      "epoch": 0.7038724373576309,
      "grad_norm": 0.9545443654060364,
      "learning_rate": 1.558119402529632e-05,
      "loss": 0.6326,
      "step": 5562
    },
    {
      "epoch": 0.7039989875980764,
      "grad_norm": 0.898814857006073,
      "learning_rate": 1.5579706902047482e-05,
      "loss": 0.5977,
      "step": 5563
    },
    {
      "epoch": 0.7041255378385219,
      "grad_norm": 0.9773504137992859,
      "learning_rate": 1.557821959959481e-05,
      "loss": 0.5981,
      "step": 5564
    },
    {
      "epoch": 0.7042520880789673,
      "grad_norm": 0.9446035623550415,
      "learning_rate": 1.5576732117986066e-05,
      "loss": 0.5957,
      "step": 5565
    },
    {
      "epoch": 0.7043786383194128,
      "grad_norm": 0.961270272731781,
      "learning_rate": 1.5575244457269036e-05,
      "loss": 0.5566,
      "step": 5566
    },
    {
      "epoch": 0.7045051885598582,
      "grad_norm": 0.9900641441345215,
      "learning_rate": 1.557375661749149e-05,
      "loss": 0.6172,
      "step": 5567
    },
    {
      "epoch": 0.7046317388003037,
      "grad_norm": 0.9471518993377686,
      "learning_rate": 1.557226859870121e-05,
      "loss": 0.5654,
      "step": 5568
    },
    {
      "epoch": 0.7047582890407492,
      "grad_norm": 0.9116830825805664,
      "learning_rate": 1.5570780400945995e-05,
      "loss": 0.571,
      "step": 5569
    },
    {
      "epoch": 0.7048848392811946,
      "grad_norm": 0.9603677988052368,
      "learning_rate": 1.556929202427364e-05,
      "loss": 0.6296,
      "step": 5570
    },
    {
      "epoch": 0.7050113895216401,
      "grad_norm": 0.9584797620773315,
      "learning_rate": 1.5567803468731945e-05,
      "loss": 0.6735,
      "step": 5571
    },
    {
      "epoch": 0.7051379397620855,
      "grad_norm": 0.9485217332839966,
      "learning_rate": 1.556631473436872e-05,
      "loss": 0.6013,
      "step": 5572
    },
    {
      "epoch": 0.705264490002531,
      "grad_norm": 0.9578256011009216,
      "learning_rate": 1.5564825821231777e-05,
      "loss": 0.6063,
      "step": 5573
    },
    {
      "epoch": 0.7053910402429765,
      "grad_norm": 0.8889157176017761,
      "learning_rate": 1.5563336729368933e-05,
      "loss": 0.5657,
      "step": 5574
    },
    {
      "epoch": 0.7055175904834219,
      "grad_norm": 0.9594210982322693,
      "learning_rate": 1.556184745882802e-05,
      "loss": 0.6061,
      "step": 5575
    },
    {
      "epoch": 0.7056441407238674,
      "grad_norm": 0.9280291795730591,
      "learning_rate": 1.5560358009656863e-05,
      "loss": 0.5984,
      "step": 5576
    },
    {
      "epoch": 0.7057706909643129,
      "grad_norm": 0.9601249098777771,
      "learning_rate": 1.5558868381903304e-05,
      "loss": 0.6287,
      "step": 5577
    },
    {
      "epoch": 0.7058972412047583,
      "grad_norm": 0.9922426342964172,
      "learning_rate": 1.555737857561518e-05,
      "loss": 0.6198,
      "step": 5578
    },
    {
      "epoch": 0.7060237914452038,
      "grad_norm": 0.9663091897964478,
      "learning_rate": 1.5555888590840343e-05,
      "loss": 0.6246,
      "step": 5579
    },
    {
      "epoch": 0.7061503416856492,
      "grad_norm": 1.0186715126037598,
      "learning_rate": 1.5554398427626646e-05,
      "loss": 0.616,
      "step": 5580
    },
    {
      "epoch": 0.7062768919260947,
      "grad_norm": 1.0495104789733887,
      "learning_rate": 1.5552908086021945e-05,
      "loss": 0.5864,
      "step": 5581
    },
    {
      "epoch": 0.7064034421665402,
      "grad_norm": 0.9383636116981506,
      "learning_rate": 1.5551417566074116e-05,
      "loss": 0.6203,
      "step": 5582
    },
    {
      "epoch": 0.7065299924069856,
      "grad_norm": 1.0231828689575195,
      "learning_rate": 1.5549926867831014e-05,
      "loss": 0.5941,
      "step": 5583
    },
    {
      "epoch": 0.706656542647431,
      "grad_norm": 1.0672622919082642,
      "learning_rate": 1.554843599134053e-05,
      "loss": 0.6395,
      "step": 5584
    },
    {
      "epoch": 0.7067830928878764,
      "grad_norm": 0.9469766616821289,
      "learning_rate": 1.5546944936650543e-05,
      "loss": 0.556,
      "step": 5585
    },
    {
      "epoch": 0.7069096431283219,
      "grad_norm": 0.9598663449287415,
      "learning_rate": 1.5545453703808936e-05,
      "loss": 0.63,
      "step": 5586
    },
    {
      "epoch": 0.7070361933687674,
      "grad_norm": 1.0688140392303467,
      "learning_rate": 1.554396229286361e-05,
      "loss": 0.6113,
      "step": 5587
    },
    {
      "epoch": 0.7071627436092128,
      "grad_norm": 0.9486138224601746,
      "learning_rate": 1.554247070386246e-05,
      "loss": 0.613,
      "step": 5588
    },
    {
      "epoch": 0.7072892938496583,
      "grad_norm": 0.9637340307235718,
      "learning_rate": 1.5540978936853394e-05,
      "loss": 0.6118,
      "step": 5589
    },
    {
      "epoch": 0.7074158440901037,
      "grad_norm": 1.005053162574768,
      "learning_rate": 1.5539486991884324e-05,
      "loss": 0.6211,
      "step": 5590
    },
    {
      "epoch": 0.7075423943305492,
      "grad_norm": 0.9197508096694946,
      "learning_rate": 1.553799486900316e-05,
      "loss": 0.5738,
      "step": 5591
    },
    {
      "epoch": 0.7076689445709947,
      "grad_norm": 0.9787551164627075,
      "learning_rate": 1.553650256825783e-05,
      "loss": 0.646,
      "step": 5592
    },
    {
      "epoch": 0.7077954948114401,
      "grad_norm": 0.9482073187828064,
      "learning_rate": 1.5535010089696266e-05,
      "loss": 0.6404,
      "step": 5593
    },
    {
      "epoch": 0.7079220450518856,
      "grad_norm": 0.9631180167198181,
      "learning_rate": 1.5533517433366396e-05,
      "loss": 0.5943,
      "step": 5594
    },
    {
      "epoch": 0.7080485952923311,
      "grad_norm": 1.0173008441925049,
      "learning_rate": 1.553202459931616e-05,
      "loss": 0.5819,
      "step": 5595
    },
    {
      "epoch": 0.7081751455327765,
      "grad_norm": 1.0429314374923706,
      "learning_rate": 1.553053158759351e-05,
      "loss": 0.5855,
      "step": 5596
    },
    {
      "epoch": 0.708301695773222,
      "grad_norm": 0.9790948629379272,
      "learning_rate": 1.552903839824639e-05,
      "loss": 0.5951,
      "step": 5597
    },
    {
      "epoch": 0.7084282460136674,
      "grad_norm": 0.9428908228874207,
      "learning_rate": 1.5527545031322756e-05,
      "loss": 0.6117,
      "step": 5598
    },
    {
      "epoch": 0.7085547962541129,
      "grad_norm": 0.9551436901092529,
      "learning_rate": 1.5526051486870577e-05,
      "loss": 0.6135,
      "step": 5599
    },
    {
      "epoch": 0.7086813464945584,
      "grad_norm": 0.9765203595161438,
      "learning_rate": 1.552455776493782e-05,
      "loss": 0.6109,
      "step": 5600
    },
    {
      "epoch": 0.7088078967350038,
      "grad_norm": 0.9069428443908691,
      "learning_rate": 1.552306386557245e-05,
      "loss": 0.5742,
      "step": 5601
    },
    {
      "epoch": 0.7089344469754493,
      "grad_norm": 0.9519375562667847,
      "learning_rate": 1.552156978882246e-05,
      "loss": 0.6037,
      "step": 5602
    },
    {
      "epoch": 0.7090609972158947,
      "grad_norm": 0.907823920249939,
      "learning_rate": 1.552007553473583e-05,
      "loss": 0.5714,
      "step": 5603
    },
    {
      "epoch": 0.7091875474563402,
      "grad_norm": 0.9536734819412231,
      "learning_rate": 1.551858110336055e-05,
      "loss": 0.6384,
      "step": 5604
    },
    {
      "epoch": 0.7093140976967857,
      "grad_norm": 0.9797875881195068,
      "learning_rate": 1.5517086494744612e-05,
      "loss": 0.6322,
      "step": 5605
    },
    {
      "epoch": 0.709440647937231,
      "grad_norm": 0.9825603365898132,
      "learning_rate": 1.5515591708936026e-05,
      "loss": 0.594,
      "step": 5606
    },
    {
      "epoch": 0.7095671981776766,
      "grad_norm": 1.0025030374526978,
      "learning_rate": 1.5514096745982798e-05,
      "loss": 0.603,
      "step": 5607
    },
    {
      "epoch": 0.7096937484181219,
      "grad_norm": 0.9682935476303101,
      "learning_rate": 1.5512601605932944e-05,
      "loss": 0.6158,
      "step": 5608
    },
    {
      "epoch": 0.7098202986585674,
      "grad_norm": 1.038206696510315,
      "learning_rate": 1.5511106288834475e-05,
      "loss": 0.6369,
      "step": 5609
    },
    {
      "epoch": 0.7099468488990129,
      "grad_norm": 1.053033709526062,
      "learning_rate": 1.5509610794735424e-05,
      "loss": 0.6755,
      "step": 5610
    },
    {
      "epoch": 0.7100733991394583,
      "grad_norm": 0.94826340675354,
      "learning_rate": 1.550811512368382e-05,
      "loss": 0.557,
      "step": 5611
    },
    {
      "epoch": 0.7101999493799038,
      "grad_norm": 1.0036588907241821,
      "learning_rate": 1.5506619275727707e-05,
      "loss": 0.6361,
      "step": 5612
    },
    {
      "epoch": 0.7103264996203493,
      "grad_norm": 0.9580093026161194,
      "learning_rate": 1.550512325091511e-05,
      "loss": 0.6062,
      "step": 5613
    },
    {
      "epoch": 0.7104530498607947,
      "grad_norm": 0.9331609010696411,
      "learning_rate": 1.550362704929409e-05,
      "loss": 0.6197,
      "step": 5614
    },
    {
      "epoch": 0.7105796001012402,
      "grad_norm": 0.9580181837081909,
      "learning_rate": 1.5502130670912697e-05,
      "loss": 0.6661,
      "step": 5615
    },
    {
      "epoch": 0.7107061503416856,
      "grad_norm": 0.9442073702812195,
      "learning_rate": 1.550063411581899e-05,
      "loss": 0.6541,
      "step": 5616
    },
    {
      "epoch": 0.7108327005821311,
      "grad_norm": 0.9172679781913757,
      "learning_rate": 1.549913738406104e-05,
      "loss": 0.6322,
      "step": 5617
    },
    {
      "epoch": 0.7109592508225766,
      "grad_norm": 0.9327040910720825,
      "learning_rate": 1.5497640475686905e-05,
      "loss": 0.5824,
      "step": 5618
    },
    {
      "epoch": 0.711085801063022,
      "grad_norm": 0.9640412926673889,
      "learning_rate": 1.5496143390744674e-05,
      "loss": 0.6491,
      "step": 5619
    },
    {
      "epoch": 0.7112123513034675,
      "grad_norm": 0.9854453802108765,
      "learning_rate": 1.549464612928242e-05,
      "loss": 0.6127,
      "step": 5620
    },
    {
      "epoch": 0.7113389015439129,
      "grad_norm": 0.9895620346069336,
      "learning_rate": 1.5493148691348238e-05,
      "loss": 0.6034,
      "step": 5621
    },
    {
      "epoch": 0.7114654517843584,
      "grad_norm": 1.1098405122756958,
      "learning_rate": 1.549165107699021e-05,
      "loss": 0.6168,
      "step": 5622
    },
    {
      "epoch": 0.7115920020248039,
      "grad_norm": 1.0279395580291748,
      "learning_rate": 1.549015328625645e-05,
      "loss": 0.5488,
      "step": 5623
    },
    {
      "epoch": 0.7117185522652493,
      "grad_norm": 1.1854515075683594,
      "learning_rate": 1.5488655319195052e-05,
      "loss": 0.6151,
      "step": 5624
    },
    {
      "epoch": 0.7118451025056948,
      "grad_norm": 0.9513448476791382,
      "learning_rate": 1.5487157175854132e-05,
      "loss": 0.6414,
      "step": 5625
    },
    {
      "epoch": 0.7119716527461403,
      "grad_norm": 1.0309170484542847,
      "learning_rate": 1.5485658856281796e-05,
      "loss": 0.6256,
      "step": 5626
    },
    {
      "epoch": 0.7120982029865857,
      "grad_norm": 1.0945243835449219,
      "learning_rate": 1.548416036052618e-05,
      "loss": 0.6462,
      "step": 5627
    },
    {
      "epoch": 0.7122247532270312,
      "grad_norm": 1.0275708436965942,
      "learning_rate": 1.5482661688635404e-05,
      "loss": 0.6136,
      "step": 5628
    },
    {
      "epoch": 0.7123513034674765,
      "grad_norm": 0.9922972321510315,
      "learning_rate": 1.54811628406576e-05,
      "loss": 0.6302,
      "step": 5629
    },
    {
      "epoch": 0.712477853707922,
      "grad_norm": 1.038603663444519,
      "learning_rate": 1.547966381664091e-05,
      "loss": 0.6358,
      "step": 5630
    },
    {
      "epoch": 0.7126044039483675,
      "grad_norm": 1.1616027355194092,
      "learning_rate": 1.5478164616633472e-05,
      "loss": 0.6833,
      "step": 5631
    },
    {
      "epoch": 0.7127309541888129,
      "grad_norm": 0.9977415800094604,
      "learning_rate": 1.547666524068344e-05,
      "loss": 0.6111,
      "step": 5632
    },
    {
      "epoch": 0.7128575044292584,
      "grad_norm": 0.9167080521583557,
      "learning_rate": 1.5475165688838975e-05,
      "loss": 0.6145,
      "step": 5633
    },
    {
      "epoch": 0.7129840546697038,
      "grad_norm": 0.9629576802253723,
      "learning_rate": 1.5473665961148226e-05,
      "loss": 0.5891,
      "step": 5634
    },
    {
      "epoch": 0.7131106049101493,
      "grad_norm": 1.020159125328064,
      "learning_rate": 1.5472166057659374e-05,
      "loss": 0.6279,
      "step": 5635
    },
    {
      "epoch": 0.7132371551505948,
      "grad_norm": 1.0581097602844238,
      "learning_rate": 1.547066597842058e-05,
      "loss": 0.6331,
      "step": 5636
    },
    {
      "epoch": 0.7133637053910402,
      "grad_norm": 0.9531593322753906,
      "learning_rate": 1.5469165723480024e-05,
      "loss": 0.6371,
      "step": 5637
    },
    {
      "epoch": 0.7134902556314857,
      "grad_norm": 0.9894569516181946,
      "learning_rate": 1.5467665292885896e-05,
      "loss": 0.5931,
      "step": 5638
    },
    {
      "epoch": 0.7136168058719311,
      "grad_norm": 0.9526286125183105,
      "learning_rate": 1.546616468668638e-05,
      "loss": 0.579,
      "step": 5639
    },
    {
      "epoch": 0.7137433561123766,
      "grad_norm": 0.9440636038780212,
      "learning_rate": 1.5464663904929672e-05,
      "loss": 0.6095,
      "step": 5640
    },
    {
      "epoch": 0.7138699063528221,
      "grad_norm": 1.0200581550598145,
      "learning_rate": 1.5463162947663976e-05,
      "loss": 0.5697,
      "step": 5641
    },
    {
      "epoch": 0.7139964565932675,
      "grad_norm": 1.0387682914733887,
      "learning_rate": 1.5461661814937495e-05,
      "loss": 0.6446,
      "step": 5642
    },
    {
      "epoch": 0.714123006833713,
      "grad_norm": 0.9554896950721741,
      "learning_rate": 1.546016050679844e-05,
      "loss": 0.6196,
      "step": 5643
    },
    {
      "epoch": 0.7142495570741585,
      "grad_norm": 1.0927995443344116,
      "learning_rate": 1.5458659023295032e-05,
      "loss": 0.6119,
      "step": 5644
    },
    {
      "epoch": 0.7143761073146039,
      "grad_norm": 0.9848804473876953,
      "learning_rate": 1.545715736447549e-05,
      "loss": 0.6185,
      "step": 5645
    },
    {
      "epoch": 0.7145026575550494,
      "grad_norm": 1.085891842842102,
      "learning_rate": 1.5455655530388046e-05,
      "loss": 0.6236,
      "step": 5646
    },
    {
      "epoch": 0.7146292077954948,
      "grad_norm": 1.1609258651733398,
      "learning_rate": 1.5454153521080937e-05,
      "loss": 0.6192,
      "step": 5647
    },
    {
      "epoch": 0.7147557580359403,
      "grad_norm": 0.9651893973350525,
      "learning_rate": 1.5452651336602397e-05,
      "loss": 0.617,
      "step": 5648
    },
    {
      "epoch": 0.7148823082763858,
      "grad_norm": 1.1989319324493408,
      "learning_rate": 1.5451148977000676e-05,
      "loss": 0.6269,
      "step": 5649
    },
    {
      "epoch": 0.7150088585168312,
      "grad_norm": 0.9238415360450745,
      "learning_rate": 1.5449646442324025e-05,
      "loss": 0.6198,
      "step": 5650
    },
    {
      "epoch": 0.7151354087572767,
      "grad_norm": 0.9846417903900146,
      "learning_rate": 1.54481437326207e-05,
      "loss": 0.6019,
      "step": 5651
    },
    {
      "epoch": 0.715261958997722,
      "grad_norm": 1.0169790983200073,
      "learning_rate": 1.5446640847938966e-05,
      "loss": 0.5982,
      "step": 5652
    },
    {
      "epoch": 0.7153885092381675,
      "grad_norm": 0.945046603679657,
      "learning_rate": 1.5445137788327086e-05,
      "loss": 0.6015,
      "step": 5653
    },
    {
      "epoch": 0.715515059478613,
      "grad_norm": 0.962907612323761,
      "learning_rate": 1.5443634553833335e-05,
      "loss": 0.5929,
      "step": 5654
    },
    {
      "epoch": 0.7156416097190584,
      "grad_norm": 1.0719630718231201,
      "learning_rate": 1.5442131144505997e-05,
      "loss": 0.6388,
      "step": 5655
    },
    {
      "epoch": 0.7157681599595039,
      "grad_norm": 0.9600449204444885,
      "learning_rate": 1.5440627560393355e-05,
      "loss": 0.5854,
      "step": 5656
    },
    {
      "epoch": 0.7158947101999493,
      "grad_norm": 1.0231739282608032,
      "learning_rate": 1.5439123801543696e-05,
      "loss": 0.6753,
      "step": 5657
    },
    {
      "epoch": 0.7160212604403948,
      "grad_norm": 0.9286175966262817,
      "learning_rate": 1.5437619868005325e-05,
      "loss": 0.5661,
      "step": 5658
    },
    {
      "epoch": 0.7161478106808403,
      "grad_norm": 1.013429045677185,
      "learning_rate": 1.5436115759826535e-05,
      "loss": 0.5915,
      "step": 5659
    },
    {
      "epoch": 0.7162743609212857,
      "grad_norm": 0.96408611536026,
      "learning_rate": 1.543461147705564e-05,
      "loss": 0.635,
      "step": 5660
    },
    {
      "epoch": 0.7164009111617312,
      "grad_norm": 1.0884208679199219,
      "learning_rate": 1.543310701974095e-05,
      "loss": 0.6432,
      "step": 5661
    },
    {
      "epoch": 0.7165274614021767,
      "grad_norm": 0.9147074818611145,
      "learning_rate": 1.543160238793078e-05,
      "loss": 0.6574,
      "step": 5662
    },
    {
      "epoch": 0.7166540116426221,
      "grad_norm": 1.0730208158493042,
      "learning_rate": 1.543009758167346e-05,
      "loss": 0.6556,
      "step": 5663
    },
    {
      "epoch": 0.7167805618830676,
      "grad_norm": 0.9247079491615295,
      "learning_rate": 1.542859260101732e-05,
      "loss": 0.6055,
      "step": 5664
    },
    {
      "epoch": 0.716907112123513,
      "grad_norm": 1.0644992589950562,
      "learning_rate": 1.5427087446010695e-05,
      "loss": 0.6214,
      "step": 5665
    },
    {
      "epoch": 0.7170336623639585,
      "grad_norm": 1.053755521774292,
      "learning_rate": 1.5425582116701924e-05,
      "loss": 0.662,
      "step": 5666
    },
    {
      "epoch": 0.717160212604404,
      "grad_norm": 0.9940082430839539,
      "learning_rate": 1.5424076613139352e-05,
      "loss": 0.605,
      "step": 5667
    },
    {
      "epoch": 0.7172867628448494,
      "grad_norm": 1.1117749214172363,
      "learning_rate": 1.5422570935371336e-05,
      "loss": 0.6071,
      "step": 5668
    },
    {
      "epoch": 0.7174133130852949,
      "grad_norm": 1.047428846359253,
      "learning_rate": 1.5421065083446235e-05,
      "loss": 0.6375,
      "step": 5669
    },
    {
      "epoch": 0.7175398633257403,
      "grad_norm": 1.0305734872817993,
      "learning_rate": 1.5419559057412407e-05,
      "loss": 0.5748,
      "step": 5670
    },
    {
      "epoch": 0.7176664135661858,
      "grad_norm": 0.96730637550354,
      "learning_rate": 1.5418052857318224e-05,
      "loss": 0.6312,
      "step": 5671
    },
    {
      "epoch": 0.7177929638066313,
      "grad_norm": 0.9380179047584534,
      "learning_rate": 1.541654648321206e-05,
      "loss": 0.5885,
      "step": 5672
    },
    {
      "epoch": 0.7179195140470767,
      "grad_norm": 0.9471980333328247,
      "learning_rate": 1.5415039935142298e-05,
      "loss": 0.6271,
      "step": 5673
    },
    {
      "epoch": 0.7180460642875222,
      "grad_norm": 0.9320921301841736,
      "learning_rate": 1.541353321315732e-05,
      "loss": 0.602,
      "step": 5674
    },
    {
      "epoch": 0.7181726145279677,
      "grad_norm": 0.9538108110427856,
      "learning_rate": 1.541202631730552e-05,
      "loss": 0.6225,
      "step": 5675
    },
    {
      "epoch": 0.718299164768413,
      "grad_norm": 0.9531292915344238,
      "learning_rate": 1.5410519247635293e-05,
      "loss": 0.6292,
      "step": 5676
    },
    {
      "epoch": 0.7184257150088585,
      "grad_norm": 0.9198936820030212,
      "learning_rate": 1.5409012004195046e-05,
      "loss": 0.5719,
      "step": 5677
    },
    {
      "epoch": 0.7185522652493039,
      "grad_norm": 0.969600260257721,
      "learning_rate": 1.540750458703318e-05,
      "loss": 0.5929,
      "step": 5678
    },
    {
      "epoch": 0.7186788154897494,
      "grad_norm": 0.9367181658744812,
      "learning_rate": 1.540599699619812e-05,
      "loss": 0.6219,
      "step": 5679
    },
    {
      "epoch": 0.7188053657301949,
      "grad_norm": 0.9684308767318726,
      "learning_rate": 1.5404489231738275e-05,
      "loss": 0.6496,
      "step": 5680
    },
    {
      "epoch": 0.7189319159706403,
      "grad_norm": 0.9589041471481323,
      "learning_rate": 1.540298129370207e-05,
      "loss": 0.5854,
      "step": 5681
    },
    {
      "epoch": 0.7190584662110858,
      "grad_norm": 0.9616936445236206,
      "learning_rate": 1.5401473182137943e-05,
      "loss": 0.6285,
      "step": 5682
    },
    {
      "epoch": 0.7191850164515312,
      "grad_norm": 0.9641441106796265,
      "learning_rate": 1.5399964897094322e-05,
      "loss": 0.5943,
      "step": 5683
    },
    {
      "epoch": 0.7193115666919767,
      "grad_norm": 0.9326357245445251,
      "learning_rate": 1.539845643861966e-05,
      "loss": 0.6356,
      "step": 5684
    },
    {
      "epoch": 0.7194381169324222,
      "grad_norm": 1.0204890966415405,
      "learning_rate": 1.5396947806762392e-05,
      "loss": 0.5745,
      "step": 5685
    },
    {
      "epoch": 0.7195646671728676,
      "grad_norm": 0.9786368608474731,
      "learning_rate": 1.5395439001570977e-05,
      "loss": 0.6129,
      "step": 5686
    },
    {
      "epoch": 0.7196912174133131,
      "grad_norm": 0.8876911997795105,
      "learning_rate": 1.5393930023093876e-05,
      "loss": 0.6096,
      "step": 5687
    },
    {
      "epoch": 0.7198177676537585,
      "grad_norm": 0.9658730030059814,
      "learning_rate": 1.5392420871379544e-05,
      "loss": 0.5964,
      "step": 5688
    },
    {
      "epoch": 0.719944317894204,
      "grad_norm": 0.9092459678649902,
      "learning_rate": 1.539091154647646e-05,
      "loss": 0.6187,
      "step": 5689
    },
    {
      "epoch": 0.7200708681346495,
      "grad_norm": 0.9632468223571777,
      "learning_rate": 1.538940204843309e-05,
      "loss": 0.6136,
      "step": 5690
    },
    {
      "epoch": 0.7201974183750949,
      "grad_norm": 0.9357160925865173,
      "learning_rate": 1.5387892377297925e-05,
      "loss": 0.609,
      "step": 5691
    },
    {
      "epoch": 0.7203239686155404,
      "grad_norm": 0.9179970026016235,
      "learning_rate": 1.538638253311944e-05,
      "loss": 0.6485,
      "step": 5692
    },
    {
      "epoch": 0.7204505188559859,
      "grad_norm": 0.9847589731216431,
      "learning_rate": 1.5384872515946138e-05,
      "loss": 0.6093,
      "step": 5693
    },
    {
      "epoch": 0.7205770690964313,
      "grad_norm": 0.9390530586242676,
      "learning_rate": 1.538336232582651e-05,
      "loss": 0.6061,
      "step": 5694
    },
    {
      "epoch": 0.7207036193368768,
      "grad_norm": 0.9294517040252686,
      "learning_rate": 1.5381851962809058e-05,
      "loss": 0.6289,
      "step": 5695
    },
    {
      "epoch": 0.7208301695773222,
      "grad_norm": 0.9180437326431274,
      "learning_rate": 1.5380341426942293e-05,
      "loss": 0.6095,
      "step": 5696
    },
    {
      "epoch": 0.7209567198177677,
      "grad_norm": 0.9638039469718933,
      "learning_rate": 1.5378830718274728e-05,
      "loss": 0.6427,
      "step": 5697
    },
    {
      "epoch": 0.7210832700582132,
      "grad_norm": 1.0686906576156616,
      "learning_rate": 1.5377319836854885e-05,
      "loss": 0.6041,
      "step": 5698
    },
    {
      "epoch": 0.7212098202986585,
      "grad_norm": 0.9114986062049866,
      "learning_rate": 1.5375808782731283e-05,
      "loss": 0.5651,
      "step": 5699
    },
    {
      "epoch": 0.721336370539104,
      "grad_norm": 0.9114232659339905,
      "learning_rate": 1.537429755595246e-05,
      "loss": 0.6307,
      "step": 5700
    },
    {
      "epoch": 0.7214629207795494,
      "grad_norm": 0.9194821119308472,
      "learning_rate": 1.5372786156566947e-05,
      "loss": 0.6073,
      "step": 5701
    },
    {
      "epoch": 0.7215894710199949,
      "grad_norm": 0.9788965582847595,
      "learning_rate": 1.5371274584623287e-05,
      "loss": 0.6239,
      "step": 5702
    },
    {
      "epoch": 0.7217160212604404,
      "grad_norm": 0.9043117761611938,
      "learning_rate": 1.5369762840170025e-05,
      "loss": 0.6155,
      "step": 5703
    },
    {
      "epoch": 0.7218425715008858,
      "grad_norm": 0.9251746535301208,
      "learning_rate": 1.5368250923255724e-05,
      "loss": 0.6301,
      "step": 5704
    },
    {
      "epoch": 0.7219691217413313,
      "grad_norm": 0.8913114666938782,
      "learning_rate": 1.5366738833928928e-05,
      "loss": 0.57,
      "step": 5705
    },
    {
      "epoch": 0.7220956719817767,
      "grad_norm": 0.9351162314414978,
      "learning_rate": 1.536522657223821e-05,
      "loss": 0.6241,
      "step": 5706
    },
    {
      "epoch": 0.7222222222222222,
      "grad_norm": 0.9440637230873108,
      "learning_rate": 1.5363714138232136e-05,
      "loss": 0.5844,
      "step": 5707
    },
    {
      "epoch": 0.7223487724626677,
      "grad_norm": 0.9789125323295593,
      "learning_rate": 1.5362201531959286e-05,
      "loss": 0.6428,
      "step": 5708
    },
    {
      "epoch": 0.7224753227031131,
      "grad_norm": 0.9133086800575256,
      "learning_rate": 1.5360688753468235e-05,
      "loss": 0.5756,
      "step": 5709
    },
    {
      "epoch": 0.7226018729435586,
      "grad_norm": 0.9055839776992798,
      "learning_rate": 1.535917580280757e-05,
      "loss": 0.5862,
      "step": 5710
    },
    {
      "epoch": 0.7227284231840041,
      "grad_norm": 0.900937020778656,
      "learning_rate": 1.535766268002588e-05,
      "loss": 0.5923,
      "step": 5711
    },
    {
      "epoch": 0.7228549734244495,
      "grad_norm": 0.9140967726707458,
      "learning_rate": 1.5356149385171767e-05,
      "loss": 0.587,
      "step": 5712
    },
    {
      "epoch": 0.722981523664895,
      "grad_norm": 1.066474437713623,
      "learning_rate": 1.5354635918293835e-05,
      "loss": 0.6084,
      "step": 5713
    },
    {
      "epoch": 0.7231080739053404,
      "grad_norm": 0.9724334478378296,
      "learning_rate": 1.5353122279440683e-05,
      "loss": 0.6137,
      "step": 5714
    },
    {
      "epoch": 0.7232346241457859,
      "grad_norm": 0.9532704949378967,
      "learning_rate": 1.5351608468660933e-05,
      "loss": 0.6558,
      "step": 5715
    },
    {
      "epoch": 0.7233611743862314,
      "grad_norm": 0.9758806228637695,
      "learning_rate": 1.5350094486003203e-05,
      "loss": 0.6119,
      "step": 5716
    },
    {
      "epoch": 0.7234877246266768,
      "grad_norm": 0.9156096577644348,
      "learning_rate": 1.5348580331516115e-05,
      "loss": 0.6131,
      "step": 5717
    },
    {
      "epoch": 0.7236142748671223,
      "grad_norm": 0.9400560855865479,
      "learning_rate": 1.5347066005248302e-05,
      "loss": 0.5969,
      "step": 5718
    },
    {
      "epoch": 0.7237408251075677,
      "grad_norm": 0.8818110227584839,
      "learning_rate": 1.5345551507248398e-05,
      "loss": 0.6008,
      "step": 5719
    },
    {
      "epoch": 0.7238673753480132,
      "grad_norm": 0.9418174624443054,
      "learning_rate": 1.5344036837565046e-05,
      "loss": 0.6204,
      "step": 5720
    },
    {
      "epoch": 0.7239939255884587,
      "grad_norm": 0.937606930732727,
      "learning_rate": 1.534252199624689e-05,
      "loss": 0.608,
      "step": 5721
    },
    {
      "epoch": 0.724120475828904,
      "grad_norm": 0.9540881514549255,
      "learning_rate": 1.5341006983342582e-05,
      "loss": 0.6139,
      "step": 5722
    },
    {
      "epoch": 0.7242470260693495,
      "grad_norm": 1.020067572593689,
      "learning_rate": 1.5339491798900784e-05,
      "loss": 0.6393,
      "step": 5723
    },
    {
      "epoch": 0.724373576309795,
      "grad_norm": 1.0643911361694336,
      "learning_rate": 1.5337976442970153e-05,
      "loss": 0.6264,
      "step": 5724
    },
    {
      "epoch": 0.7245001265502404,
      "grad_norm": 0.9301388263702393,
      "learning_rate": 1.533646091559936e-05,
      "loss": 0.617,
      "step": 5725
    },
    {
      "epoch": 0.7246266767906859,
      "grad_norm": 1.084900140762329,
      "learning_rate": 1.5334945216837085e-05,
      "loss": 0.6214,
      "step": 5726
    },
    {
      "epoch": 0.7247532270311313,
      "grad_norm": 0.9406861066818237,
      "learning_rate": 1.5333429346732002e-05,
      "loss": 0.6012,
      "step": 5727
    },
    {
      "epoch": 0.7248797772715768,
      "grad_norm": 1.1854970455169678,
      "learning_rate": 1.53319133053328e-05,
      "loss": 0.6278,
      "step": 5728
    },
    {
      "epoch": 0.7250063275120223,
      "grad_norm": 0.9256955981254578,
      "learning_rate": 1.5330397092688162e-05,
      "loss": 0.6245,
      "step": 5729
    },
    {
      "epoch": 0.7251328777524677,
      "grad_norm": 0.9275783896446228,
      "learning_rate": 1.5328880708846793e-05,
      "loss": 0.5781,
      "step": 5730
    },
    {
      "epoch": 0.7252594279929132,
      "grad_norm": 1.0211137533187866,
      "learning_rate": 1.5327364153857392e-05,
      "loss": 0.6303,
      "step": 5731
    },
    {
      "epoch": 0.7253859782333586,
      "grad_norm": 1.0086843967437744,
      "learning_rate": 1.532584742776867e-05,
      "loss": 0.6439,
      "step": 5732
    },
    {
      "epoch": 0.7255125284738041,
      "grad_norm": 1.0618396997451782,
      "learning_rate": 1.532433053062933e-05,
      "loss": 0.6526,
      "step": 5733
    },
    {
      "epoch": 0.7256390787142496,
      "grad_norm": 0.9424541592597961,
      "learning_rate": 1.53228134624881e-05,
      "loss": 0.5978,
      "step": 5734
    },
    {
      "epoch": 0.725765628954695,
      "grad_norm": 0.9987956285476685,
      "learning_rate": 1.53212962233937e-05,
      "loss": 0.6335,
      "step": 5735
    },
    {
      "epoch": 0.7258921791951405,
      "grad_norm": 0.95489102602005,
      "learning_rate": 1.5319778813394855e-05,
      "loss": 0.5994,
      "step": 5736
    },
    {
      "epoch": 0.7260187294355859,
      "grad_norm": 0.9512636661529541,
      "learning_rate": 1.5318261232540307e-05,
      "loss": 0.5924,
      "step": 5737
    },
    {
      "epoch": 0.7261452796760314,
      "grad_norm": 0.9316332340240479,
      "learning_rate": 1.531674348087879e-05,
      "loss": 0.6054,
      "step": 5738
    },
    {
      "epoch": 0.7262718299164769,
      "grad_norm": 0.9618790149688721,
      "learning_rate": 1.5315225558459058e-05,
      "loss": 0.5753,
      "step": 5739
    },
    {
      "epoch": 0.7263983801569223,
      "grad_norm": 0.9082068204879761,
      "learning_rate": 1.5313707465329853e-05,
      "loss": 0.6126,
      "step": 5740
    },
    {
      "epoch": 0.7265249303973678,
      "grad_norm": 1.051428198814392,
      "learning_rate": 1.5312189201539935e-05,
      "loss": 0.6174,
      "step": 5741
    },
    {
      "epoch": 0.7266514806378133,
      "grad_norm": 0.8977088332176208,
      "learning_rate": 1.5310670767138072e-05,
      "loss": 0.5555,
      "step": 5742
    },
    {
      "epoch": 0.7267780308782587,
      "grad_norm": 0.9013815522193909,
      "learning_rate": 1.530915216217302e-05,
      "loss": 0.6239,
      "step": 5743
    },
    {
      "epoch": 0.7269045811187042,
      "grad_norm": 0.9530035257339478,
      "learning_rate": 1.5307633386693565e-05,
      "loss": 0.6144,
      "step": 5744
    },
    {
      "epoch": 0.7270311313591495,
      "grad_norm": 0.8963824510574341,
      "learning_rate": 1.5306114440748475e-05,
      "loss": 0.5756,
      "step": 5745
    },
    {
      "epoch": 0.727157681599595,
      "grad_norm": 0.8896440267562866,
      "learning_rate": 1.5304595324386543e-05,
      "loss": 0.5774,
      "step": 5746
    },
    {
      "epoch": 0.7272842318400405,
      "grad_norm": 0.9971113204956055,
      "learning_rate": 1.530307603765655e-05,
      "loss": 0.6783,
      "step": 5747
    },
    {
      "epoch": 0.7274107820804859,
      "grad_norm": 0.8847318291664124,
      "learning_rate": 1.5301556580607298e-05,
      "loss": 0.5706,
      "step": 5748
    },
    {
      "epoch": 0.7275373323209314,
      "grad_norm": 0.931384801864624,
      "learning_rate": 1.5300036953287584e-05,
      "loss": 0.6012,
      "step": 5749
    },
    {
      "epoch": 0.7276638825613768,
      "grad_norm": 0.9728636145591736,
      "learning_rate": 1.5298517155746214e-05,
      "loss": 0.5676,
      "step": 5750
    },
    {
      "epoch": 0.7277904328018223,
      "grad_norm": 0.9335664510726929,
      "learning_rate": 1.5296997188032e-05,
      "loss": 0.5777,
      "step": 5751
    },
    {
      "epoch": 0.7279169830422678,
      "grad_norm": 0.9796631932258606,
      "learning_rate": 1.529547705019376e-05,
      "loss": 0.6084,
      "step": 5752
    },
    {
      "epoch": 0.7280435332827132,
      "grad_norm": 0.9044491648674011,
      "learning_rate": 1.5293956742280314e-05,
      "loss": 0.5627,
      "step": 5753
    },
    {
      "epoch": 0.7281700835231587,
      "grad_norm": 0.9312912225723267,
      "learning_rate": 1.5292436264340492e-05,
      "loss": 0.6067,
      "step": 5754
    },
    {
      "epoch": 0.7282966337636041,
      "grad_norm": 0.9704937934875488,
      "learning_rate": 1.5290915616423125e-05,
      "loss": 0.6428,
      "step": 5755
    },
    {
      "epoch": 0.7284231840040496,
      "grad_norm": 0.8916208744049072,
      "learning_rate": 1.5289394798577055e-05,
      "loss": 0.6088,
      "step": 5756
    },
    {
      "epoch": 0.7285497342444951,
      "grad_norm": 0.8932051062583923,
      "learning_rate": 1.5287873810851124e-05,
      "loss": 0.6175,
      "step": 5757
    },
    {
      "epoch": 0.7286762844849405,
      "grad_norm": 0.9843328595161438,
      "learning_rate": 1.528635265329418e-05,
      "loss": 0.5951,
      "step": 5758
    },
    {
      "epoch": 0.728802834725386,
      "grad_norm": 0.9366129636764526,
      "learning_rate": 1.5284831325955084e-05,
      "loss": 0.6373,
      "step": 5759
    },
    {
      "epoch": 0.7289293849658315,
      "grad_norm": 0.8604097366333008,
      "learning_rate": 1.5283309828882693e-05,
      "loss": 0.538,
      "step": 5760
    },
    {
      "epoch": 0.7290559352062769,
      "grad_norm": 0.9770821928977966,
      "learning_rate": 1.5281788162125867e-05,
      "loss": 0.6366,
      "step": 5761
    },
    {
      "epoch": 0.7291824854467224,
      "grad_norm": 0.924847424030304,
      "learning_rate": 1.528026632573349e-05,
      "loss": 0.6403,
      "step": 5762
    },
    {
      "epoch": 0.7293090356871678,
      "grad_norm": 0.932874321937561,
      "learning_rate": 1.527874431975443e-05,
      "loss": 0.6424,
      "step": 5763
    },
    {
      "epoch": 0.7294355859276133,
      "grad_norm": 0.8930548429489136,
      "learning_rate": 1.5277222144237568e-05,
      "loss": 0.6233,
      "step": 5764
    },
    {
      "epoch": 0.7295621361680588,
      "grad_norm": 0.8842646479606628,
      "learning_rate": 1.52756997992318e-05,
      "loss": 0.5569,
      "step": 5765
    },
    {
      "epoch": 0.7296886864085042,
      "grad_norm": 0.9183261394500732,
      "learning_rate": 1.527417728478601e-05,
      "loss": 0.6412,
      "step": 5766
    },
    {
      "epoch": 0.7298152366489497,
      "grad_norm": 0.87581467628479,
      "learning_rate": 1.5272654600949105e-05,
      "loss": 0.5896,
      "step": 5767
    },
    {
      "epoch": 0.729941786889395,
      "grad_norm": 0.8773422837257385,
      "learning_rate": 1.5271131747769988e-05,
      "loss": 0.5434,
      "step": 5768
    },
    {
      "epoch": 0.7300683371298405,
      "grad_norm": 0.9937372803688049,
      "learning_rate": 1.526960872529756e-05,
      "loss": 0.663,
      "step": 5769
    },
    {
      "epoch": 0.730194887370286,
      "grad_norm": 0.9207173585891724,
      "learning_rate": 1.5268085533580744e-05,
      "loss": 0.5962,
      "step": 5770
    },
    {
      "epoch": 0.7303214376107314,
      "grad_norm": 0.9160839915275574,
      "learning_rate": 1.5266562172668462e-05,
      "loss": 0.6188,
      "step": 5771
    },
    {
      "epoch": 0.7304479878511769,
      "grad_norm": 0.9198680520057678,
      "learning_rate": 1.5265038642609634e-05,
      "loss": 0.5408,
      "step": 5772
    },
    {
      "epoch": 0.7305745380916224,
      "grad_norm": 0.9570396542549133,
      "learning_rate": 1.5263514943453193e-05,
      "loss": 0.6256,
      "step": 5773
    },
    {
      "epoch": 0.7307010883320678,
      "grad_norm": 0.9040963053703308,
      "learning_rate": 1.526199107524808e-05,
      "loss": 0.6114,
      "step": 5774
    },
    {
      "epoch": 0.7308276385725133,
      "grad_norm": 0.9688183665275574,
      "learning_rate": 1.5260467038043228e-05,
      "loss": 0.6359,
      "step": 5775
    },
    {
      "epoch": 0.7309541888129587,
      "grad_norm": 1.009284257888794,
      "learning_rate": 1.5258942831887594e-05,
      "loss": 0.5917,
      "step": 5776
    },
    {
      "epoch": 0.7310807390534042,
      "grad_norm": 0.9893501996994019,
      "learning_rate": 1.5257418456830127e-05,
      "loss": 0.5232,
      "step": 5777
    },
    {
      "epoch": 0.7312072892938497,
      "grad_norm": 0.9047680497169495,
      "learning_rate": 1.5255893912919786e-05,
      "loss": 0.5742,
      "step": 5778
    },
    {
      "epoch": 0.7313338395342951,
      "grad_norm": 0.8619433641433716,
      "learning_rate": 1.5254369200205537e-05,
      "loss": 0.5413,
      "step": 5779
    },
    {
      "epoch": 0.7314603897747406,
      "grad_norm": 0.9859150648117065,
      "learning_rate": 1.5252844318736341e-05,
      "loss": 0.6008,
      "step": 5780
    },
    {
      "epoch": 0.731586940015186,
      "grad_norm": 0.940841555595398,
      "learning_rate": 1.5251319268561182e-05,
      "loss": 0.5995,
      "step": 5781
    },
    {
      "epoch": 0.7317134902556315,
      "grad_norm": 0.9463821053504944,
      "learning_rate": 1.5249794049729038e-05,
      "loss": 0.6048,
      "step": 5782
    },
    {
      "epoch": 0.731840040496077,
      "grad_norm": 0.9296199083328247,
      "learning_rate": 1.5248268662288892e-05,
      "loss": 0.6531,
      "step": 5783
    },
    {
      "epoch": 0.7319665907365224,
      "grad_norm": 0.9478632807731628,
      "learning_rate": 1.5246743106289738e-05,
      "loss": 0.6045,
      "step": 5784
    },
    {
      "epoch": 0.7320931409769679,
      "grad_norm": 0.9138583540916443,
      "learning_rate": 1.5245217381780571e-05,
      "loss": 0.6183,
      "step": 5785
    },
    {
      "epoch": 0.7322196912174133,
      "grad_norm": 0.9717731475830078,
      "learning_rate": 1.5243691488810391e-05,
      "loss": 0.635,
      "step": 5786
    },
    {
      "epoch": 0.7323462414578588,
      "grad_norm": 0.88224858045578,
      "learning_rate": 1.5242165427428206e-05,
      "loss": 0.6012,
      "step": 5787
    },
    {
      "epoch": 0.7324727916983043,
      "grad_norm": 0.9590405821800232,
      "learning_rate": 1.5240639197683029e-05,
      "loss": 0.6046,
      "step": 5788
    },
    {
      "epoch": 0.7325993419387496,
      "grad_norm": 0.9168108105659485,
      "learning_rate": 1.523911279962388e-05,
      "loss": 0.5611,
      "step": 5789
    },
    {
      "epoch": 0.7327258921791951,
      "grad_norm": 1.051866888999939,
      "learning_rate": 1.5237586233299784e-05,
      "loss": 0.6582,
      "step": 5790
    },
    {
      "epoch": 0.7328524424196406,
      "grad_norm": 1.0312330722808838,
      "learning_rate": 1.5236059498759761e-05,
      "loss": 0.632,
      "step": 5791
    },
    {
      "epoch": 0.732978992660086,
      "grad_norm": 0.895278811454773,
      "learning_rate": 1.5234532596052854e-05,
      "loss": 0.5742,
      "step": 5792
    },
    {
      "epoch": 0.7331055429005315,
      "grad_norm": 0.9628463387489319,
      "learning_rate": 1.52330055252281e-05,
      "loss": 0.6224,
      "step": 5793
    },
    {
      "epoch": 0.7332320931409769,
      "grad_norm": 0.9528704881668091,
      "learning_rate": 1.5231478286334542e-05,
      "loss": 0.5919,
      "step": 5794
    },
    {
      "epoch": 0.7333586433814224,
      "grad_norm": 0.9220967292785645,
      "learning_rate": 1.5229950879421234e-05,
      "loss": 0.5891,
      "step": 5795
    },
    {
      "epoch": 0.7334851936218679,
      "grad_norm": 0.8946312069892883,
      "learning_rate": 1.522842330453723e-05,
      "loss": 0.6327,
      "step": 5796
    },
    {
      "epoch": 0.7336117438623133,
      "grad_norm": 1.0240442752838135,
      "learning_rate": 1.522689556173159e-05,
      "loss": 0.6208,
      "step": 5797
    },
    {
      "epoch": 0.7337382941027588,
      "grad_norm": 0.9281057715415955,
      "learning_rate": 1.522536765105338e-05,
      "loss": 0.6171,
      "step": 5798
    },
    {
      "epoch": 0.7338648443432042,
      "grad_norm": 1.0296717882156372,
      "learning_rate": 1.5223839572551677e-05,
      "loss": 0.6183,
      "step": 5799
    },
    {
      "epoch": 0.7339913945836497,
      "grad_norm": 0.9326429963111877,
      "learning_rate": 1.5222311326275555e-05,
      "loss": 0.6027,
      "step": 5800
    },
    {
      "epoch": 0.7341179448240952,
      "grad_norm": 0.8794600963592529,
      "learning_rate": 1.5220782912274092e-05,
      "loss": 0.628,
      "step": 5801
    },
    {
      "epoch": 0.7342444950645406,
      "grad_norm": 0.9334079027175903,
      "learning_rate": 1.5219254330596388e-05,
      "loss": 0.602,
      "step": 5802
    },
    {
      "epoch": 0.7343710453049861,
      "grad_norm": 0.9002235531806946,
      "learning_rate": 1.5217725581291525e-05,
      "loss": 0.5873,
      "step": 5803
    },
    {
      "epoch": 0.7344975955454315,
      "grad_norm": 0.9221592545509338,
      "learning_rate": 1.5216196664408606e-05,
      "loss": 0.6198,
      "step": 5804
    },
    {
      "epoch": 0.734624145785877,
      "grad_norm": 0.9956276416778564,
      "learning_rate": 1.521466757999674e-05,
      "loss": 0.637,
      "step": 5805
    },
    {
      "epoch": 0.7347506960263225,
      "grad_norm": 1.0644464492797852,
      "learning_rate": 1.521313832810503e-05,
      "loss": 0.6399,
      "step": 5806
    },
    {
      "epoch": 0.7348772462667679,
      "grad_norm": 0.9371156096458435,
      "learning_rate": 1.5211608908782598e-05,
      "loss": 0.5994,
      "step": 5807
    },
    {
      "epoch": 0.7350037965072134,
      "grad_norm": 0.9344585537910461,
      "learning_rate": 1.5210079322078557e-05,
      "loss": 0.6171,
      "step": 5808
    },
    {
      "epoch": 0.7351303467476589,
      "grad_norm": 1.0317984819412231,
      "learning_rate": 1.5208549568042035e-05,
      "loss": 0.6091,
      "step": 5809
    },
    {
      "epoch": 0.7352568969881043,
      "grad_norm": 0.9513420462608337,
      "learning_rate": 1.5207019646722168e-05,
      "loss": 0.6338,
      "step": 5810
    },
    {
      "epoch": 0.7353834472285498,
      "grad_norm": 1.0340709686279297,
      "learning_rate": 1.5205489558168083e-05,
      "loss": 0.6586,
      "step": 5811
    },
    {
      "epoch": 0.7355099974689951,
      "grad_norm": 0.9259512424468994,
      "learning_rate": 1.5203959302428933e-05,
      "loss": 0.534,
      "step": 5812
    },
    {
      "epoch": 0.7356365477094406,
      "grad_norm": 0.9885875582695007,
      "learning_rate": 1.520242887955386e-05,
      "loss": 0.614,
      "step": 5813
    },
    {
      "epoch": 0.7357630979498861,
      "grad_norm": 0.9177362322807312,
      "learning_rate": 1.5200898289592015e-05,
      "loss": 0.6311,
      "step": 5814
    },
    {
      "epoch": 0.7358896481903315,
      "grad_norm": 8.925389289855957,
      "learning_rate": 1.5199367532592556e-05,
      "loss": 0.6101,
      "step": 5815
    },
    {
      "epoch": 0.736016198430777,
      "grad_norm": 1.0397080183029175,
      "learning_rate": 1.519783660860465e-05,
      "loss": 0.6678,
      "step": 5816
    },
    {
      "epoch": 0.7361427486712224,
      "grad_norm": 0.8899138569831848,
      "learning_rate": 1.5196305517677465e-05,
      "loss": 0.572,
      "step": 5817
    },
    {
      "epoch": 0.7362692989116679,
      "grad_norm": 0.9922070503234863,
      "learning_rate": 1.5194774259860175e-05,
      "loss": 0.6866,
      "step": 5818
    },
    {
      "epoch": 0.7363958491521134,
      "grad_norm": 0.9549830555915833,
      "learning_rate": 1.519324283520196e-05,
      "loss": 0.6196,
      "step": 5819
    },
    {
      "epoch": 0.7365223993925588,
      "grad_norm": 0.92320317029953,
      "learning_rate": 1.5191711243752e-05,
      "loss": 0.5875,
      "step": 5820
    },
    {
      "epoch": 0.7366489496330043,
      "grad_norm": 1.1452521085739136,
      "learning_rate": 1.5190179485559492e-05,
      "loss": 0.6334,
      "step": 5821
    },
    {
      "epoch": 0.7367754998734498,
      "grad_norm": 0.9853760600090027,
      "learning_rate": 1.5188647560673628e-05,
      "loss": 0.6123,
      "step": 5822
    },
    {
      "epoch": 0.7369020501138952,
      "grad_norm": 0.9608046412467957,
      "learning_rate": 1.5187115469143608e-05,
      "loss": 0.6288,
      "step": 5823
    },
    {
      "epoch": 0.7370286003543407,
      "grad_norm": 1.0083184242248535,
      "learning_rate": 1.5185583211018639e-05,
      "loss": 0.625,
      "step": 5824
    },
    {
      "epoch": 0.7371551505947861,
      "grad_norm": 0.9231647849082947,
      "learning_rate": 1.518405078634794e-05,
      "loss": 0.5858,
      "step": 5825
    },
    {
      "epoch": 0.7372817008352316,
      "grad_norm": 1.0206218957901,
      "learning_rate": 1.5182518195180713e-05,
      "loss": 0.6259,
      "step": 5826
    },
    {
      "epoch": 0.7374082510756771,
      "grad_norm": 0.9200266599655151,
      "learning_rate": 1.5180985437566192e-05,
      "loss": 0.587,
      "step": 5827
    },
    {
      "epoch": 0.7375348013161225,
      "grad_norm": 1.6251521110534668,
      "learning_rate": 1.51794525135536e-05,
      "loss": 0.6159,
      "step": 5828
    },
    {
      "epoch": 0.737661351556568,
      "grad_norm": 0.947180449962616,
      "learning_rate": 1.5177919423192173e-05,
      "loss": 0.6131,
      "step": 5829
    },
    {
      "epoch": 0.7377879017970134,
      "grad_norm": 1.0297932624816895,
      "learning_rate": 1.5176386166531147e-05,
      "loss": 0.6807,
      "step": 5830
    },
    {
      "epoch": 0.7379144520374589,
      "grad_norm": 0.9341084361076355,
      "learning_rate": 1.517485274361977e-05,
      "loss": 0.6198,
      "step": 5831
    },
    {
      "epoch": 0.7380410022779044,
      "grad_norm": 0.9735609889030457,
      "learning_rate": 1.5173319154507286e-05,
      "loss": 0.5821,
      "step": 5832
    },
    {
      "epoch": 0.7381675525183498,
      "grad_norm": 1.1074793338775635,
      "learning_rate": 1.517178539924295e-05,
      "loss": 0.6065,
      "step": 5833
    },
    {
      "epoch": 0.7382941027587953,
      "grad_norm": 1.000594973564148,
      "learning_rate": 1.5170251477876021e-05,
      "loss": 0.6412,
      "step": 5834
    },
    {
      "epoch": 0.7384206529992406,
      "grad_norm": 1.183289647102356,
      "learning_rate": 1.5168717390455769e-05,
      "loss": 0.6109,
      "step": 5835
    },
    {
      "epoch": 0.7385472032396861,
      "grad_norm": 1.1034361124038696,
      "learning_rate": 1.516718313703146e-05,
      "loss": 0.5892,
      "step": 5836
    },
    {
      "epoch": 0.7386737534801316,
      "grad_norm": 1.2564928531646729,
      "learning_rate": 1.516564871765237e-05,
      "loss": 0.6155,
      "step": 5837
    },
    {
      "epoch": 0.738800303720577,
      "grad_norm": 0.942898154258728,
      "learning_rate": 1.516411413236778e-05,
      "loss": 0.607,
      "step": 5838
    },
    {
      "epoch": 0.7389268539610225,
      "grad_norm": 1.1261687278747559,
      "learning_rate": 1.5162579381226979e-05,
      "loss": 0.6112,
      "step": 5839
    },
    {
      "epoch": 0.739053404201468,
      "grad_norm": 0.960727334022522,
      "learning_rate": 1.5161044464279258e-05,
      "loss": 0.6096,
      "step": 5840
    },
    {
      "epoch": 0.7391799544419134,
      "grad_norm": 0.9568994641304016,
      "learning_rate": 1.5159509381573908e-05,
      "loss": 0.6211,
      "step": 5841
    },
    {
      "epoch": 0.7393065046823589,
      "grad_norm": 0.9183349013328552,
      "learning_rate": 1.5157974133160242e-05,
      "loss": 0.6103,
      "step": 5842
    },
    {
      "epoch": 0.7394330549228043,
      "grad_norm": 0.9724486470222473,
      "learning_rate": 1.515643871908756e-05,
      "loss": 0.6001,
      "step": 5843
    },
    {
      "epoch": 0.7395596051632498,
      "grad_norm": 0.9086158871650696,
      "learning_rate": 1.515490313940518e-05,
      "loss": 0.5922,
      "step": 5844
    },
    {
      "epoch": 0.7396861554036953,
      "grad_norm": 1.0249860286712646,
      "learning_rate": 1.5153367394162412e-05,
      "loss": 0.6287,
      "step": 5845
    },
    {
      "epoch": 0.7398127056441407,
      "grad_norm": 0.9310980439186096,
      "learning_rate": 1.5151831483408587e-05,
      "loss": 0.6007,
      "step": 5846
    },
    {
      "epoch": 0.7399392558845862,
      "grad_norm": 1.0808359384536743,
      "learning_rate": 1.5150295407193035e-05,
      "loss": 0.6429,
      "step": 5847
    },
    {
      "epoch": 0.7400658061250316,
      "grad_norm": 1.0476443767547607,
      "learning_rate": 1.5148759165565086e-05,
      "loss": 0.6326,
      "step": 5848
    },
    {
      "epoch": 0.7401923563654771,
      "grad_norm": 0.9356460571289062,
      "learning_rate": 1.5147222758574081e-05,
      "loss": 0.6118,
      "step": 5849
    },
    {
      "epoch": 0.7403189066059226,
      "grad_norm": 0.9681439399719238,
      "learning_rate": 1.5145686186269362e-05,
      "loss": 0.6163,
      "step": 5850
    },
    {
      "epoch": 0.740445456846368,
      "grad_norm": 0.965099573135376,
      "learning_rate": 1.5144149448700283e-05,
      "loss": 0.5682,
      "step": 5851
    },
    {
      "epoch": 0.7405720070868135,
      "grad_norm": 0.9631316661834717,
      "learning_rate": 1.5142612545916198e-05,
      "loss": 0.5994,
      "step": 5852
    },
    {
      "epoch": 0.7406985573272589,
      "grad_norm": 0.8940647840499878,
      "learning_rate": 1.5141075477966468e-05,
      "loss": 0.5936,
      "step": 5853
    },
    {
      "epoch": 0.7408251075677044,
      "grad_norm": 0.9514415860176086,
      "learning_rate": 1.5139538244900463e-05,
      "loss": 0.5826,
      "step": 5854
    },
    {
      "epoch": 0.7409516578081499,
      "grad_norm": 1.0306177139282227,
      "learning_rate": 1.513800084676755e-05,
      "loss": 0.6677,
      "step": 5855
    },
    {
      "epoch": 0.7410782080485953,
      "grad_norm": 1.0049474239349365,
      "learning_rate": 1.5136463283617102e-05,
      "loss": 0.6333,
      "step": 5856
    },
    {
      "epoch": 0.7412047582890408,
      "grad_norm": 1.0005826950073242,
      "learning_rate": 1.513492555549851e-05,
      "loss": 0.6086,
      "step": 5857
    },
    {
      "epoch": 0.7413313085294863,
      "grad_norm": 0.9245320558547974,
      "learning_rate": 1.5133387662461154e-05,
      "loss": 0.5865,
      "step": 5858
    },
    {
      "epoch": 0.7414578587699316,
      "grad_norm": 0.9376662969589233,
      "learning_rate": 1.5131849604554432e-05,
      "loss": 0.575,
      "step": 5859
    },
    {
      "epoch": 0.7415844090103771,
      "grad_norm": 1.0297659635543823,
      "learning_rate": 1.5130311381827738e-05,
      "loss": 0.6418,
      "step": 5860
    },
    {
      "epoch": 0.7417109592508225,
      "grad_norm": 0.9565175175666809,
      "learning_rate": 1.5128772994330476e-05,
      "loss": 0.6119,
      "step": 5861
    },
    {
      "epoch": 0.741837509491268,
      "grad_norm": 1.0433294773101807,
      "learning_rate": 1.5127234442112055e-05,
      "loss": 0.6541,
      "step": 5862
    },
    {
      "epoch": 0.7419640597317135,
      "grad_norm": 0.9246670007705688,
      "learning_rate": 1.5125695725221889e-05,
      "loss": 0.5934,
      "step": 5863
    },
    {
      "epoch": 0.7420906099721589,
      "grad_norm": 0.946749746799469,
      "learning_rate": 1.51241568437094e-05,
      "loss": 0.5918,
      "step": 5864
    },
    {
      "epoch": 0.7422171602126044,
      "grad_norm": 0.9453434348106384,
      "learning_rate": 1.5122617797624006e-05,
      "loss": 0.546,
      "step": 5865
    },
    {
      "epoch": 0.7423437104530498,
      "grad_norm": 0.9623180031776428,
      "learning_rate": 1.5121078587015142e-05,
      "loss": 0.5944,
      "step": 5866
    },
    {
      "epoch": 0.7424702606934953,
      "grad_norm": 0.9396483898162842,
      "learning_rate": 1.5119539211932238e-05,
      "loss": 0.6059,
      "step": 5867
    },
    {
      "epoch": 0.7425968109339408,
      "grad_norm": 0.9625547528266907,
      "learning_rate": 1.511799967242474e-05,
      "loss": 0.5973,
      "step": 5868
    },
    {
      "epoch": 0.7427233611743862,
      "grad_norm": 1.0170809030532837,
      "learning_rate": 1.511645996854209e-05,
      "loss": 0.6664,
      "step": 5869
    },
    {
      "epoch": 0.7428499114148317,
      "grad_norm": 0.9134652018547058,
      "learning_rate": 1.5114920100333739e-05,
      "loss": 0.6314,
      "step": 5870
    },
    {
      "epoch": 0.7429764616552772,
      "grad_norm": 0.9899610877037048,
      "learning_rate": 1.5113380067849144e-05,
      "loss": 0.6237,
      "step": 5871
    },
    {
      "epoch": 0.7431030118957226,
      "grad_norm": 1.0009429454803467,
      "learning_rate": 1.5111839871137765e-05,
      "loss": 0.632,
      "step": 5872
    },
    {
      "epoch": 0.7432295621361681,
      "grad_norm": 0.9408573508262634,
      "learning_rate": 1.5110299510249068e-05,
      "loss": 0.6272,
      "step": 5873
    },
    {
      "epoch": 0.7433561123766135,
      "grad_norm": 0.9648882150650024,
      "learning_rate": 1.5108758985232527e-05,
      "loss": 0.6507,
      "step": 5874
    },
    {
      "epoch": 0.743482662617059,
      "grad_norm": 0.9087377786636353,
      "learning_rate": 1.510721829613762e-05,
      "loss": 0.585,
      "step": 5875
    },
    {
      "epoch": 0.7436092128575045,
      "grad_norm": 0.9441981911659241,
      "learning_rate": 1.5105677443013826e-05,
      "loss": 0.6166,
      "step": 5876
    },
    {
      "epoch": 0.7437357630979499,
      "grad_norm": 0.9420744776725769,
      "learning_rate": 1.5104136425910638e-05,
      "loss": 0.5971,
      "step": 5877
    },
    {
      "epoch": 0.7438623133383954,
      "grad_norm": 0.9420576691627502,
      "learning_rate": 1.5102595244877543e-05,
      "loss": 0.6442,
      "step": 5878
    },
    {
      "epoch": 0.7439888635788408,
      "grad_norm": 0.906653106212616,
      "learning_rate": 1.510105389996404e-05,
      "loss": 0.6046,
      "step": 5879
    },
    {
      "epoch": 0.7441154138192863,
      "grad_norm": 0.9583439826965332,
      "learning_rate": 1.5099512391219637e-05,
      "loss": 0.6349,
      "step": 5880
    },
    {
      "epoch": 0.7442419640597318,
      "grad_norm": 0.9358474016189575,
      "learning_rate": 1.509797071869384e-05,
      "loss": 0.5947,
      "step": 5881
    },
    {
      "epoch": 0.7443685143001771,
      "grad_norm": 0.8836809992790222,
      "learning_rate": 1.509642888243616e-05,
      "loss": 0.6215,
      "step": 5882
    },
    {
      "epoch": 0.7444950645406226,
      "grad_norm": 0.9043595194816589,
      "learning_rate": 1.5094886882496123e-05,
      "loss": 0.6185,
      "step": 5883
    },
    {
      "epoch": 0.744621614781068,
      "grad_norm": 0.889555811882019,
      "learning_rate": 1.5093344718923248e-05,
      "loss": 0.5669,
      "step": 5884
    },
    {
      "epoch": 0.7447481650215135,
      "grad_norm": 1.0318716764450073,
      "learning_rate": 1.5091802391767065e-05,
      "loss": 0.6022,
      "step": 5885
    },
    {
      "epoch": 0.744874715261959,
      "grad_norm": 0.9283406734466553,
      "learning_rate": 1.5090259901077112e-05,
      "loss": 0.5919,
      "step": 5886
    },
    {
      "epoch": 0.7450012655024044,
      "grad_norm": 0.9266576766967773,
      "learning_rate": 1.5088717246902926e-05,
      "loss": 0.5929,
      "step": 5887
    },
    {
      "epoch": 0.7451278157428499,
      "grad_norm": 0.8593734502792358,
      "learning_rate": 1.5087174429294056e-05,
      "loss": 0.5756,
      "step": 5888
    },
    {
      "epoch": 0.7452543659832954,
      "grad_norm": 0.9074184894561768,
      "learning_rate": 1.5085631448300051e-05,
      "loss": 0.5974,
      "step": 5889
    },
    {
      "epoch": 0.7453809162237408,
      "grad_norm": 0.9138680100440979,
      "learning_rate": 1.5084088303970467e-05,
      "loss": 0.5993,
      "step": 5890
    },
    {
      "epoch": 0.7455074664641863,
      "grad_norm": 0.9471439719200134,
      "learning_rate": 1.5082544996354865e-05,
      "loss": 0.6005,
      "step": 5891
    },
    {
      "epoch": 0.7456340167046317,
      "grad_norm": 0.8921017646789551,
      "learning_rate": 1.5081001525502813e-05,
      "loss": 0.6005,
      "step": 5892
    },
    {
      "epoch": 0.7457605669450772,
      "grad_norm": 0.8919736742973328,
      "learning_rate": 1.5079457891463882e-05,
      "loss": 0.5939,
      "step": 5893
    },
    {
      "epoch": 0.7458871171855227,
      "grad_norm": 0.8899592161178589,
      "learning_rate": 1.5077914094287647e-05,
      "loss": 0.6155,
      "step": 5894
    },
    {
      "epoch": 0.7460136674259681,
      "grad_norm": 0.8949479460716248,
      "learning_rate": 1.5076370134023695e-05,
      "loss": 0.6241,
      "step": 5895
    },
    {
      "epoch": 0.7461402176664136,
      "grad_norm": 0.9334869384765625,
      "learning_rate": 1.5074826010721605e-05,
      "loss": 0.6077,
      "step": 5896
    },
    {
      "epoch": 0.746266767906859,
      "grad_norm": 0.9509887099266052,
      "learning_rate": 1.507328172443098e-05,
      "loss": 0.6364,
      "step": 5897
    },
    {
      "epoch": 0.7463933181473045,
      "grad_norm": 0.9222272634506226,
      "learning_rate": 1.507173727520141e-05,
      "loss": 0.6159,
      "step": 5898
    },
    {
      "epoch": 0.74651986838775,
      "grad_norm": 0.9428218007087708,
      "learning_rate": 1.50701926630825e-05,
      "loss": 0.5756,
      "step": 5899
    },
    {
      "epoch": 0.7466464186281954,
      "grad_norm": 0.8918053507804871,
      "learning_rate": 1.5068647888123864e-05,
      "loss": 0.5882,
      "step": 5900
    },
    {
      "epoch": 0.7467729688686409,
      "grad_norm": 0.9503786563873291,
      "learning_rate": 1.506710295037511e-05,
      "loss": 0.6213,
      "step": 5901
    },
    {
      "epoch": 0.7468995191090863,
      "grad_norm": 0.9467563033103943,
      "learning_rate": 1.5065557849885857e-05,
      "loss": 0.6013,
      "step": 5902
    },
    {
      "epoch": 0.7470260693495318,
      "grad_norm": 0.8743632435798645,
      "learning_rate": 1.5064012586705729e-05,
      "loss": 0.6154,
      "step": 5903
    },
    {
      "epoch": 0.7471526195899773,
      "grad_norm": 0.9086523056030273,
      "learning_rate": 1.5062467160884357e-05,
      "loss": 0.6033,
      "step": 5904
    },
    {
      "epoch": 0.7472791698304226,
      "grad_norm": 0.9773529767990112,
      "learning_rate": 1.5060921572471379e-05,
      "loss": 0.5967,
      "step": 5905
    },
    {
      "epoch": 0.7474057200708681,
      "grad_norm": 0.9855648875236511,
      "learning_rate": 1.5059375821516429e-05,
      "loss": 0.5981,
      "step": 5906
    },
    {
      "epoch": 0.7475322703113136,
      "grad_norm": 0.9144850373268127,
      "learning_rate": 1.5057829908069155e-05,
      "loss": 0.6141,
      "step": 5907
    },
    {
      "epoch": 0.747658820551759,
      "grad_norm": 0.9168948531150818,
      "learning_rate": 1.5056283832179203e-05,
      "loss": 0.6155,
      "step": 5908
    },
    {
      "epoch": 0.7477853707922045,
      "grad_norm": 0.8822616934776306,
      "learning_rate": 1.5054737593896234e-05,
      "loss": 0.5407,
      "step": 5909
    },
    {
      "epoch": 0.7479119210326499,
      "grad_norm": 1.0263229608535767,
      "learning_rate": 1.5053191193269905e-05,
      "loss": 0.6558,
      "step": 5910
    },
    {
      "epoch": 0.7480384712730954,
      "grad_norm": 0.9491477608680725,
      "learning_rate": 1.5051644630349883e-05,
      "loss": 0.6325,
      "step": 5911
    },
    {
      "epoch": 0.7481650215135409,
      "grad_norm": 0.9656940698623657,
      "learning_rate": 1.505009790518584e-05,
      "loss": 0.6056,
      "step": 5912
    },
    {
      "epoch": 0.7482915717539863,
      "grad_norm": 0.9474108219146729,
      "learning_rate": 1.5048551017827451e-05,
      "loss": 0.5864,
      "step": 5913
    },
    {
      "epoch": 0.7484181219944318,
      "grad_norm": 0.9400244355201721,
      "learning_rate": 1.5047003968324396e-05,
      "loss": 0.6025,
      "step": 5914
    },
    {
      "epoch": 0.7485446722348772,
      "grad_norm": 0.9152452945709229,
      "learning_rate": 1.5045456756726367e-05,
      "loss": 0.6234,
      "step": 5915
    },
    {
      "epoch": 0.7486712224753227,
      "grad_norm": 0.9400520324707031,
      "learning_rate": 1.5043909383083049e-05,
      "loss": 0.6145,
      "step": 5916
    },
    {
      "epoch": 0.7487977727157682,
      "grad_norm": 0.8930346965789795,
      "learning_rate": 1.5042361847444146e-05,
      "loss": 0.6067,
      "step": 5917
    },
    {
      "epoch": 0.7489243229562136,
      "grad_norm": 0.9638873934745789,
      "learning_rate": 1.5040814149859354e-05,
      "loss": 0.6282,
      "step": 5918
    },
    {
      "epoch": 0.7490508731966591,
      "grad_norm": 0.9220654964447021,
      "learning_rate": 1.5039266290378387e-05,
      "loss": 0.5931,
      "step": 5919
    },
    {
      "epoch": 0.7491774234371046,
      "grad_norm": 0.8781309127807617,
      "learning_rate": 1.5037718269050952e-05,
      "loss": 0.5471,
      "step": 5920
    },
    {
      "epoch": 0.74930397367755,
      "grad_norm": 1.0029293298721313,
      "learning_rate": 1.503617008592677e-05,
      "loss": 0.5745,
      "step": 5921
    },
    {
      "epoch": 0.7494305239179955,
      "grad_norm": 0.9157075881958008,
      "learning_rate": 1.5034621741055562e-05,
      "loss": 0.6102,
      "step": 5922
    },
    {
      "epoch": 0.7495570741584409,
      "grad_norm": 0.9322646260261536,
      "learning_rate": 1.5033073234487056e-05,
      "loss": 0.5916,
      "step": 5923
    },
    {
      "epoch": 0.7496836243988864,
      "grad_norm": 0.9843838810920715,
      "learning_rate": 1.5031524566270989e-05,
      "loss": 0.6581,
      "step": 5924
    },
    {
      "epoch": 0.7498101746393319,
      "grad_norm": 0.9308920502662659,
      "learning_rate": 1.5029975736457096e-05,
      "loss": 0.5403,
      "step": 5925
    },
    {
      "epoch": 0.7499367248797773,
      "grad_norm": 0.919053316116333,
      "learning_rate": 1.5028426745095123e-05,
      "loss": 0.6552,
      "step": 5926
    },
    {
      "epoch": 0.7500632751202227,
      "grad_norm": 0.9102080464363098,
      "learning_rate": 1.5026877592234819e-05,
      "loss": 0.5812,
      "step": 5927
    },
    {
      "epoch": 0.7501898253606681,
      "grad_norm": 0.9064688086509705,
      "learning_rate": 1.5025328277925937e-05,
      "loss": 0.5851,
      "step": 5928
    },
    {
      "epoch": 0.7503163756011136,
      "grad_norm": 0.8989754915237427,
      "learning_rate": 1.5023778802218241e-05,
      "loss": 0.5841,
      "step": 5929
    },
    {
      "epoch": 0.7504429258415591,
      "grad_norm": 0.987586259841919,
      "learning_rate": 1.5022229165161488e-05,
      "loss": 0.5944,
      "step": 5930
    },
    {
      "epoch": 0.7505694760820045,
      "grad_norm": 0.9969536066055298,
      "learning_rate": 1.5020679366805455e-05,
      "loss": 0.6199,
      "step": 5931
    },
    {
      "epoch": 0.75069602632245,
      "grad_norm": 1.0332653522491455,
      "learning_rate": 1.5019129407199911e-05,
      "loss": 0.5941,
      "step": 5932
    },
    {
      "epoch": 0.7508225765628954,
      "grad_norm": 0.9379153251647949,
      "learning_rate": 1.5017579286394637e-05,
      "loss": 0.6218,
      "step": 5933
    },
    {
      "epoch": 0.7509491268033409,
      "grad_norm": 1.019877314567566,
      "learning_rate": 1.5016029004439423e-05,
      "loss": 0.6158,
      "step": 5934
    },
    {
      "epoch": 0.7510756770437864,
      "grad_norm": 1.015566110610962,
      "learning_rate": 1.5014478561384059e-05,
      "loss": 0.6291,
      "step": 5935
    },
    {
      "epoch": 0.7512022272842318,
      "grad_norm": 1.0194694995880127,
      "learning_rate": 1.5012927957278335e-05,
      "loss": 0.5915,
      "step": 5936
    },
    {
      "epoch": 0.7513287775246773,
      "grad_norm": 0.8852446675300598,
      "learning_rate": 1.5011377192172054e-05,
      "loss": 0.5931,
      "step": 5937
    },
    {
      "epoch": 0.7514553277651228,
      "grad_norm": 1.0330933332443237,
      "learning_rate": 1.5009826266115024e-05,
      "loss": 0.6541,
      "step": 5938
    },
    {
      "epoch": 0.7515818780055682,
      "grad_norm": 0.9223150014877319,
      "learning_rate": 1.5008275179157055e-05,
      "loss": 0.6068,
      "step": 5939
    },
    {
      "epoch": 0.7517084282460137,
      "grad_norm": 0.8879669904708862,
      "learning_rate": 1.5006723931347967e-05,
      "loss": 0.5706,
      "step": 5940
    },
    {
      "epoch": 0.7518349784864591,
      "grad_norm": 0.9278859496116638,
      "learning_rate": 1.5005172522737574e-05,
      "loss": 0.6057,
      "step": 5941
    },
    {
      "epoch": 0.7519615287269046,
      "grad_norm": 0.9172214865684509,
      "learning_rate": 1.500362095337571e-05,
      "loss": 0.6197,
      "step": 5942
    },
    {
      "epoch": 0.7520880789673501,
      "grad_norm": 0.9467888474464417,
      "learning_rate": 1.5002069223312203e-05,
      "loss": 0.6168,
      "step": 5943
    },
    {
      "epoch": 0.7522146292077955,
      "grad_norm": 0.9515306949615479,
      "learning_rate": 1.5000517332596892e-05,
      "loss": 0.6407,
      "step": 5944
    },
    {
      "epoch": 0.752341179448241,
      "grad_norm": 0.9014859199523926,
      "learning_rate": 1.4998965281279615e-05,
      "loss": 0.6133,
      "step": 5945
    },
    {
      "epoch": 0.7524677296886864,
      "grad_norm": 0.9269716143608093,
      "learning_rate": 1.4997413069410227e-05,
      "loss": 0.6225,
      "step": 5946
    },
    {
      "epoch": 0.7525942799291319,
      "grad_norm": 1.0150110721588135,
      "learning_rate": 1.4995860697038573e-05,
      "loss": 0.6374,
      "step": 5947
    },
    {
      "epoch": 0.7527208301695774,
      "grad_norm": 0.947364866733551,
      "learning_rate": 1.4994308164214515e-05,
      "loss": 0.6079,
      "step": 5948
    },
    {
      "epoch": 0.7528473804100227,
      "grad_norm": 0.9676277041435242,
      "learning_rate": 1.4992755470987912e-05,
      "loss": 0.586,
      "step": 5949
    },
    {
      "epoch": 0.7529739306504682,
      "grad_norm": 0.9146112203598022,
      "learning_rate": 1.4991202617408638e-05,
      "loss": 0.5906,
      "step": 5950
    },
    {
      "epoch": 0.7531004808909137,
      "grad_norm": 0.9252833724021912,
      "learning_rate": 1.4989649603526559e-05,
      "loss": 0.5888,
      "step": 5951
    },
    {
      "epoch": 0.7532270311313591,
      "grad_norm": 0.9593004584312439,
      "learning_rate": 1.4988096429391559e-05,
      "loss": 0.6333,
      "step": 5952
    },
    {
      "epoch": 0.7533535813718046,
      "grad_norm": 0.964630126953125,
      "learning_rate": 1.4986543095053518e-05,
      "loss": 0.6648,
      "step": 5953
    },
    {
      "epoch": 0.75348013161225,
      "grad_norm": 0.9148818850517273,
      "learning_rate": 1.4984989600562326e-05,
      "loss": 0.6373,
      "step": 5954
    },
    {
      "epoch": 0.7536066818526955,
      "grad_norm": 0.9832289814949036,
      "learning_rate": 1.4983435945967876e-05,
      "loss": 0.6034,
      "step": 5955
    },
    {
      "epoch": 0.753733232093141,
      "grad_norm": 0.9155037999153137,
      "learning_rate": 1.4981882131320067e-05,
      "loss": 0.6016,
      "step": 5956
    },
    {
      "epoch": 0.7538597823335864,
      "grad_norm": 0.963983416557312,
      "learning_rate": 1.4980328156668806e-05,
      "loss": 0.5691,
      "step": 5957
    },
    {
      "epoch": 0.7539863325740319,
      "grad_norm": 0.8843207359313965,
      "learning_rate": 1.4978774022063996e-05,
      "loss": 0.608,
      "step": 5958
    },
    {
      "epoch": 0.7541128828144773,
      "grad_norm": 0.9107388257980347,
      "learning_rate": 1.4977219727555555e-05,
      "loss": 0.5884,
      "step": 5959
    },
    {
      "epoch": 0.7542394330549228,
      "grad_norm": 0.9321625828742981,
      "learning_rate": 1.4975665273193403e-05,
      "loss": 0.6307,
      "step": 5960
    },
    {
      "epoch": 0.7543659832953683,
      "grad_norm": 0.9541513919830322,
      "learning_rate": 1.4974110659027461e-05,
      "loss": 0.6036,
      "step": 5961
    },
    {
      "epoch": 0.7544925335358137,
      "grad_norm": 0.9926546216011047,
      "learning_rate": 1.4972555885107663e-05,
      "loss": 0.6236,
      "step": 5962
    },
    {
      "epoch": 0.7546190837762592,
      "grad_norm": 0.9488472938537598,
      "learning_rate": 1.4971000951483942e-05,
      "loss": 0.594,
      "step": 5963
    },
    {
      "epoch": 0.7547456340167046,
      "grad_norm": 0.909123957157135,
      "learning_rate": 1.4969445858206238e-05,
      "loss": 0.5927,
      "step": 5964
    },
    {
      "epoch": 0.7548721842571501,
      "grad_norm": 0.9481947422027588,
      "learning_rate": 1.4967890605324493e-05,
      "loss": 0.6376,
      "step": 5965
    },
    {
      "epoch": 0.7549987344975956,
      "grad_norm": 0.9394279718399048,
      "learning_rate": 1.4966335192888664e-05,
      "loss": 0.583,
      "step": 5966
    },
    {
      "epoch": 0.755125284738041,
      "grad_norm": 0.8736150860786438,
      "learning_rate": 1.49647796209487e-05,
      "loss": 0.5875,
      "step": 5967
    },
    {
      "epoch": 0.7552518349784865,
      "grad_norm": 0.971824586391449,
      "learning_rate": 1.4963223889554564e-05,
      "loss": 0.6508,
      "step": 5968
    },
    {
      "epoch": 0.755378385218932,
      "grad_norm": 0.9317954778671265,
      "learning_rate": 1.4961667998756222e-05,
      "loss": 0.5992,
      "step": 5969
    },
    {
      "epoch": 0.7555049354593774,
      "grad_norm": 0.8944098949432373,
      "learning_rate": 1.4960111948603642e-05,
      "loss": 0.6004,
      "step": 5970
    },
    {
      "epoch": 0.7556314856998229,
      "grad_norm": 0.9627091288566589,
      "learning_rate": 1.4958555739146803e-05,
      "loss": 0.6085,
      "step": 5971
    },
    {
      "epoch": 0.7557580359402682,
      "grad_norm": 0.9102935791015625,
      "learning_rate": 1.4956999370435683e-05,
      "loss": 0.5856,
      "step": 5972
    },
    {
      "epoch": 0.7558845861807137,
      "grad_norm": 0.9382615089416504,
      "learning_rate": 1.4955442842520272e-05,
      "loss": 0.646,
      "step": 5973
    },
    {
      "epoch": 0.7560111364211592,
      "grad_norm": 1.005528450012207,
      "learning_rate": 1.4953886155450557e-05,
      "loss": 0.5778,
      "step": 5974
    },
    {
      "epoch": 0.7561376866616046,
      "grad_norm": 1.0415844917297363,
      "learning_rate": 1.4952329309276537e-05,
      "loss": 0.6439,
      "step": 5975
    },
    {
      "epoch": 0.7562642369020501,
      "grad_norm": 1.133527159690857,
      "learning_rate": 1.4950772304048212e-05,
      "loss": 0.6535,
      "step": 5976
    },
    {
      "epoch": 0.7563907871424955,
      "grad_norm": 0.9037154912948608,
      "learning_rate": 1.4949215139815585e-05,
      "loss": 0.6065,
      "step": 5977
    },
    {
      "epoch": 0.756517337382941,
      "grad_norm": 0.8851611018180847,
      "learning_rate": 1.4947657816628676e-05,
      "loss": 0.5786,
      "step": 5978
    },
    {
      "epoch": 0.7566438876233865,
      "grad_norm": 1.0043294429779053,
      "learning_rate": 1.4946100334537495e-05,
      "loss": 0.575,
      "step": 5979
    },
    {
      "epoch": 0.7567704378638319,
      "grad_norm": 0.9228445887565613,
      "learning_rate": 1.4944542693592065e-05,
      "loss": 0.6217,
      "step": 5980
    },
    {
      "epoch": 0.7568969881042774,
      "grad_norm": 0.9068300127983093,
      "learning_rate": 1.4942984893842416e-05,
      "loss": 0.5681,
      "step": 5981
    },
    {
      "epoch": 0.7570235383447228,
      "grad_norm": 0.9096628427505493,
      "learning_rate": 1.4941426935338577e-05,
      "loss": 0.5965,
      "step": 5982
    },
    {
      "epoch": 0.7571500885851683,
      "grad_norm": 0.9674122929573059,
      "learning_rate": 1.4939868818130582e-05,
      "loss": 0.64,
      "step": 5983
    },
    {
      "epoch": 0.7572766388256138,
      "grad_norm": 0.9392924308776855,
      "learning_rate": 1.4938310542268477e-05,
      "loss": 0.5882,
      "step": 5984
    },
    {
      "epoch": 0.7574031890660592,
      "grad_norm": 0.965067446231842,
      "learning_rate": 1.4936752107802313e-05,
      "loss": 0.6511,
      "step": 5985
    },
    {
      "epoch": 0.7575297393065047,
      "grad_norm": 0.9150431752204895,
      "learning_rate": 1.4935193514782136e-05,
      "loss": 0.6518,
      "step": 5986
    },
    {
      "epoch": 0.7576562895469502,
      "grad_norm": 0.876345157623291,
      "learning_rate": 1.4933634763258004e-05,
      "loss": 0.5628,
      "step": 5987
    },
    {
      "epoch": 0.7577828397873956,
      "grad_norm": 0.901504635810852,
      "learning_rate": 1.4932075853279983e-05,
      "loss": 0.5843,
      "step": 5988
    },
    {
      "epoch": 0.7579093900278411,
      "grad_norm": 0.973301112651825,
      "learning_rate": 1.493051678489814e-05,
      "loss": 0.5929,
      "step": 5989
    },
    {
      "epoch": 0.7580359402682865,
      "grad_norm": 1.2280701398849487,
      "learning_rate": 1.4928957558162545e-05,
      "loss": 0.5809,
      "step": 5990
    },
    {
      "epoch": 0.758162490508732,
      "grad_norm": 0.9240809679031372,
      "learning_rate": 1.4927398173123277e-05,
      "loss": 0.6179,
      "step": 5991
    },
    {
      "epoch": 0.7582890407491775,
      "grad_norm": 0.9410731792449951,
      "learning_rate": 1.492583862983042e-05,
      "loss": 0.6139,
      "step": 5992
    },
    {
      "epoch": 0.7584155909896229,
      "grad_norm": 0.8890549540519714,
      "learning_rate": 1.4924278928334062e-05,
      "loss": 0.5623,
      "step": 5993
    },
    {
      "epoch": 0.7585421412300684,
      "grad_norm": 0.9266168475151062,
      "learning_rate": 1.4922719068684295e-05,
      "loss": 0.5899,
      "step": 5994
    },
    {
      "epoch": 0.7586686914705137,
      "grad_norm": 0.909533679485321,
      "learning_rate": 1.4921159050931215e-05,
      "loss": 0.5832,
      "step": 5995
    },
    {
      "epoch": 0.7587952417109592,
      "grad_norm": 0.9227083325386047,
      "learning_rate": 1.491959887512493e-05,
      "loss": 0.6333,
      "step": 5996
    },
    {
      "epoch": 0.7589217919514047,
      "grad_norm": 0.927035391330719,
      "learning_rate": 1.4918038541315542e-05,
      "loss": 0.5974,
      "step": 5997
    },
    {
      "epoch": 0.7590483421918501,
      "grad_norm": 0.9472408890724182,
      "learning_rate": 1.4916478049553174e-05,
      "loss": 0.6351,
      "step": 5998
    },
    {
      "epoch": 0.7591748924322956,
      "grad_norm": 0.8791247010231018,
      "learning_rate": 1.4914917399887936e-05,
      "loss": 0.6235,
      "step": 5999
    },
    {
      "epoch": 0.7593014426727411,
      "grad_norm": 0.9060704708099365,
      "learning_rate": 1.4913356592369952e-05,
      "loss": 0.5924,
      "step": 6000
    },
    {
      "epoch": 0.7594279929131865,
      "grad_norm": 1.0501688718795776,
      "learning_rate": 1.4911795627049355e-05,
      "loss": 0.6082,
      "step": 6001
    },
    {
      "epoch": 0.759554543153632,
      "grad_norm": 0.8845449090003967,
      "learning_rate": 1.4910234503976275e-05,
      "loss": 0.5779,
      "step": 6002
    },
    {
      "epoch": 0.7596810933940774,
      "grad_norm": 0.8487492799758911,
      "learning_rate": 1.4908673223200854e-05,
      "loss": 0.5888,
      "step": 6003
    },
    {
      "epoch": 0.7598076436345229,
      "grad_norm": 0.906827986240387,
      "learning_rate": 1.4907111784773232e-05,
      "loss": 0.6084,
      "step": 6004
    },
    {
      "epoch": 0.7599341938749684,
      "grad_norm": 0.8989441394805908,
      "learning_rate": 1.490555018874356e-05,
      "loss": 0.584,
      "step": 6005
    },
    {
      "epoch": 0.7600607441154138,
      "grad_norm": 0.925911545753479,
      "learning_rate": 1.4903988435161993e-05,
      "loss": 0.6022,
      "step": 6006
    },
    {
      "epoch": 0.7601872943558593,
      "grad_norm": 0.9291368126869202,
      "learning_rate": 1.4902426524078686e-05,
      "loss": 0.6023,
      "step": 6007
    },
    {
      "epoch": 0.7603138445963047,
      "grad_norm": 0.9382019639015198,
      "learning_rate": 1.4900864455543807e-05,
      "loss": 0.6145,
      "step": 6008
    },
    {
      "epoch": 0.7604403948367502,
      "grad_norm": 0.8580788969993591,
      "learning_rate": 1.4899302229607524e-05,
      "loss": 0.5898,
      "step": 6009
    },
    {
      "epoch": 0.7605669450771957,
      "grad_norm": 0.9332236647605896,
      "learning_rate": 1.489773984632001e-05,
      "loss": 0.6249,
      "step": 6010
    },
    {
      "epoch": 0.7606934953176411,
      "grad_norm": 0.9118776321411133,
      "learning_rate": 1.4896177305731443e-05,
      "loss": 0.6097,
      "step": 6011
    },
    {
      "epoch": 0.7608200455580866,
      "grad_norm": 0.9004005789756775,
      "learning_rate": 1.4894614607892009e-05,
      "loss": 0.6119,
      "step": 6012
    },
    {
      "epoch": 0.760946595798532,
      "grad_norm": 1.0255110263824463,
      "learning_rate": 1.4893051752851898e-05,
      "loss": 0.6263,
      "step": 6013
    },
    {
      "epoch": 0.7610731460389775,
      "grad_norm": 0.8949034214019775,
      "learning_rate": 1.4891488740661305e-05,
      "loss": 0.6063,
      "step": 6014
    },
    {
      "epoch": 0.761199696279423,
      "grad_norm": 0.9510985612869263,
      "learning_rate": 1.4889925571370428e-05,
      "loss": 0.6274,
      "step": 6015
    },
    {
      "epoch": 0.7613262465198684,
      "grad_norm": 0.9662999510765076,
      "learning_rate": 1.4888362245029468e-05,
      "loss": 0.6162,
      "step": 6016
    },
    {
      "epoch": 0.7614527967603139,
      "grad_norm": 0.9620106220245361,
      "learning_rate": 1.4886798761688644e-05,
      "loss": 0.6426,
      "step": 6017
    },
    {
      "epoch": 0.7615793470007594,
      "grad_norm": 0.942406177520752,
      "learning_rate": 1.4885235121398158e-05,
      "loss": 0.5998,
      "step": 6018
    },
    {
      "epoch": 0.7617058972412047,
      "grad_norm": 0.9218757748603821,
      "learning_rate": 1.488367132420824e-05,
      "loss": 0.6451,
      "step": 6019
    },
    {
      "epoch": 0.7618324474816502,
      "grad_norm": 0.8636212348937988,
      "learning_rate": 1.4882107370169107e-05,
      "loss": 0.5713,
      "step": 6020
    },
    {
      "epoch": 0.7619589977220956,
      "grad_norm": 0.8813939094543457,
      "learning_rate": 1.4880543259330997e-05,
      "loss": 0.5628,
      "step": 6021
    },
    {
      "epoch": 0.7620855479625411,
      "grad_norm": 0.9282249212265015,
      "learning_rate": 1.4878978991744134e-05,
      "loss": 0.6042,
      "step": 6022
    },
    {
      "epoch": 0.7622120982029866,
      "grad_norm": 0.9046787023544312,
      "learning_rate": 1.4877414567458768e-05,
      "loss": 0.6234,
      "step": 6023
    },
    {
      "epoch": 0.762338648443432,
      "grad_norm": 0.9034560918807983,
      "learning_rate": 1.4875849986525137e-05,
      "loss": 0.6015,
      "step": 6024
    },
    {
      "epoch": 0.7624651986838775,
      "grad_norm": 0.8644375205039978,
      "learning_rate": 1.4874285248993492e-05,
      "loss": 0.5815,
      "step": 6025
    },
    {
      "epoch": 0.7625917489243229,
      "grad_norm": 0.9431485533714294,
      "learning_rate": 1.4872720354914092e-05,
      "loss": 0.6077,
      "step": 6026
    },
    {
      "epoch": 0.7627182991647684,
      "grad_norm": 0.9055364727973938,
      "learning_rate": 1.487115530433719e-05,
      "loss": 0.5471,
      "step": 6027
    },
    {
      "epoch": 0.7628448494052139,
      "grad_norm": 0.9131219387054443,
      "learning_rate": 1.4869590097313056e-05,
      "loss": 0.6437,
      "step": 6028
    },
    {
      "epoch": 0.7629713996456593,
      "grad_norm": 0.8683010339736938,
      "learning_rate": 1.486802473389196e-05,
      "loss": 0.5883,
      "step": 6029
    },
    {
      "epoch": 0.7630979498861048,
      "grad_norm": 0.9517945051193237,
      "learning_rate": 1.4866459214124173e-05,
      "loss": 0.5925,
      "step": 6030
    },
    {
      "epoch": 0.7632245001265502,
      "grad_norm": 0.931640088558197,
      "learning_rate": 1.4864893538059977e-05,
      "loss": 0.6425,
      "step": 6031
    },
    {
      "epoch": 0.7633510503669957,
      "grad_norm": 0.913373589515686,
      "learning_rate": 1.4863327705749657e-05,
      "loss": 0.5804,
      "step": 6032
    },
    {
      "epoch": 0.7634776006074412,
      "grad_norm": 1.0619914531707764,
      "learning_rate": 1.4861761717243504e-05,
      "loss": 0.6067,
      "step": 6033
    },
    {
      "epoch": 0.7636041508478866,
      "grad_norm": 0.9327672719955444,
      "learning_rate": 1.486019557259181e-05,
      "loss": 0.5908,
      "step": 6034
    },
    {
      "epoch": 0.7637307010883321,
      "grad_norm": 0.9435297846794128,
      "learning_rate": 1.4858629271844878e-05,
      "loss": 0.6486,
      "step": 6035
    },
    {
      "epoch": 0.7638572513287776,
      "grad_norm": 0.9263685941696167,
      "learning_rate": 1.485706281505301e-05,
      "loss": 0.6055,
      "step": 6036
    },
    {
      "epoch": 0.763983801569223,
      "grad_norm": 0.9238411784172058,
      "learning_rate": 1.4855496202266519e-05,
      "loss": 0.5906,
      "step": 6037
    },
    {
      "epoch": 0.7641103518096685,
      "grad_norm": 0.884492039680481,
      "learning_rate": 1.4853929433535721e-05,
      "loss": 0.5911,
      "step": 6038
    },
    {
      "epoch": 0.7642369020501139,
      "grad_norm": 0.9243160486221313,
      "learning_rate": 1.4852362508910932e-05,
      "loss": 0.6031,
      "step": 6039
    },
    {
      "epoch": 0.7643634522905594,
      "grad_norm": 0.9252090454101562,
      "learning_rate": 1.4850795428442477e-05,
      "loss": 0.5812,
      "step": 6040
    },
    {
      "epoch": 0.7644900025310049,
      "grad_norm": 1.0017650127410889,
      "learning_rate": 1.4849228192180691e-05,
      "loss": 0.6431,
      "step": 6041
    },
    {
      "epoch": 0.7646165527714502,
      "grad_norm": 0.9766731262207031,
      "learning_rate": 1.4847660800175905e-05,
      "loss": 0.6391,
      "step": 6042
    },
    {
      "epoch": 0.7647431030118957,
      "grad_norm": 2.5104141235351562,
      "learning_rate": 1.4846093252478459e-05,
      "loss": 0.6253,
      "step": 6043
    },
    {
      "epoch": 0.7648696532523411,
      "grad_norm": 0.9868268370628357,
      "learning_rate": 1.4844525549138702e-05,
      "loss": 0.6032,
      "step": 6044
    },
    {
      "epoch": 0.7649962034927866,
      "grad_norm": 0.9621540307998657,
      "learning_rate": 1.4842957690206978e-05,
      "loss": 0.6091,
      "step": 6045
    },
    {
      "epoch": 0.7651227537332321,
      "grad_norm": 0.9562923908233643,
      "learning_rate": 1.4841389675733646e-05,
      "loss": 0.6329,
      "step": 6046
    },
    {
      "epoch": 0.7652493039736775,
      "grad_norm": 0.9270918369293213,
      "learning_rate": 1.4839821505769065e-05,
      "loss": 0.6047,
      "step": 6047
    },
    {
      "epoch": 0.765375854214123,
      "grad_norm": 0.9437395334243774,
      "learning_rate": 1.48382531803636e-05,
      "loss": 0.616,
      "step": 6048
    },
    {
      "epoch": 0.7655024044545685,
      "grad_norm": 0.9317203760147095,
      "learning_rate": 1.4836684699567622e-05,
      "loss": 0.643,
      "step": 6049
    },
    {
      "epoch": 0.7656289546950139,
      "grad_norm": 0.9565442800521851,
      "learning_rate": 1.4835116063431506e-05,
      "loss": 0.6398,
      "step": 6050
    },
    {
      "epoch": 0.7657555049354594,
      "grad_norm": 1.0311962366104126,
      "learning_rate": 1.4833547272005628e-05,
      "loss": 0.62,
      "step": 6051
    },
    {
      "epoch": 0.7658820551759048,
      "grad_norm": 1.0203384160995483,
      "learning_rate": 1.4831978325340377e-05,
      "loss": 0.6023,
      "step": 6052
    },
    {
      "epoch": 0.7660086054163503,
      "grad_norm": 0.9524819850921631,
      "learning_rate": 1.4830409223486144e-05,
      "loss": 0.5997,
      "step": 6053
    },
    {
      "epoch": 0.7661351556567958,
      "grad_norm": 0.9036087393760681,
      "learning_rate": 1.482883996649332e-05,
      "loss": 0.5635,
      "step": 6054
    },
    {
      "epoch": 0.7662617058972412,
      "grad_norm": 1.1096762418746948,
      "learning_rate": 1.482727055441231e-05,
      "loss": 0.5915,
      "step": 6055
    },
    {
      "epoch": 0.7663882561376867,
      "grad_norm": 0.9346973896026611,
      "learning_rate": 1.4825700987293514e-05,
      "loss": 0.6275,
      "step": 6056
    },
    {
      "epoch": 0.7665148063781321,
      "grad_norm": 0.9824437499046326,
      "learning_rate": 1.4824131265187343e-05,
      "loss": 0.5871,
      "step": 6057
    },
    {
      "epoch": 0.7666413566185776,
      "grad_norm": 0.9755717515945435,
      "learning_rate": 1.4822561388144214e-05,
      "loss": 0.6089,
      "step": 6058
    },
    {
      "epoch": 0.7667679068590231,
      "grad_norm": 1.0068045854568481,
      "learning_rate": 1.4820991356214545e-05,
      "loss": 0.5893,
      "step": 6059
    },
    {
      "epoch": 0.7668944570994685,
      "grad_norm": 0.9153152704238892,
      "learning_rate": 1.481942116944876e-05,
      "loss": 0.6019,
      "step": 6060
    },
    {
      "epoch": 0.767021007339914,
      "grad_norm": 0.9821732044219971,
      "learning_rate": 1.4817850827897293e-05,
      "loss": 0.6455,
      "step": 6061
    },
    {
      "epoch": 0.7671475575803594,
      "grad_norm": 1.0551267862319946,
      "learning_rate": 1.4816280331610573e-05,
      "loss": 0.6393,
      "step": 6062
    },
    {
      "epoch": 0.7672741078208049,
      "grad_norm": 0.9674519300460815,
      "learning_rate": 1.4814709680639044e-05,
      "loss": 0.6052,
      "step": 6063
    },
    {
      "epoch": 0.7674006580612504,
      "grad_norm": 0.8936464786529541,
      "learning_rate": 1.4813138875033149e-05,
      "loss": 0.623,
      "step": 6064
    },
    {
      "epoch": 0.7675272083016957,
      "grad_norm": 0.9872289299964905,
      "learning_rate": 1.481156791484334e-05,
      "loss": 0.629,
      "step": 6065
    },
    {
      "epoch": 0.7676537585421412,
      "grad_norm": 0.8923141360282898,
      "learning_rate": 1.4809996800120068e-05,
      "loss": 0.613,
      "step": 6066
    },
    {
      "epoch": 0.7677803087825867,
      "grad_norm": 0.9206587076187134,
      "learning_rate": 1.4808425530913794e-05,
      "loss": 0.6018,
      "step": 6067
    },
    {
      "epoch": 0.7679068590230321,
      "grad_norm": 0.9640328288078308,
      "learning_rate": 1.4806854107274986e-05,
      "loss": 0.6155,
      "step": 6068
    },
    {
      "epoch": 0.7680334092634776,
      "grad_norm": 0.9368852376937866,
      "learning_rate": 1.4805282529254106e-05,
      "loss": 0.5903,
      "step": 6069
    },
    {
      "epoch": 0.768159959503923,
      "grad_norm": 0.9458762407302856,
      "learning_rate": 1.4803710796901634e-05,
      "loss": 0.6228,
      "step": 6070
    },
    {
      "epoch": 0.7682865097443685,
      "grad_norm": 0.9534546136856079,
      "learning_rate": 1.4802138910268049e-05,
      "loss": 0.6268,
      "step": 6071
    },
    {
      "epoch": 0.768413059984814,
      "grad_norm": 0.9751471281051636,
      "learning_rate": 1.4800566869403837e-05,
      "loss": 0.6268,
      "step": 6072
    },
    {
      "epoch": 0.7685396102252594,
      "grad_norm": 0.9165608286857605,
      "learning_rate": 1.479899467435948e-05,
      "loss": 0.5744,
      "step": 6073
    },
    {
      "epoch": 0.7686661604657049,
      "grad_norm": 0.9047976136207581,
      "learning_rate": 1.479742232518548e-05,
      "loss": 0.5812,
      "step": 6074
    },
    {
      "epoch": 0.7687927107061503,
      "grad_norm": 0.952853262424469,
      "learning_rate": 1.4795849821932333e-05,
      "loss": 0.5948,
      "step": 6075
    },
    {
      "epoch": 0.7689192609465958,
      "grad_norm": 0.9950709939002991,
      "learning_rate": 1.4794277164650544e-05,
      "loss": 0.6254,
      "step": 6076
    },
    {
      "epoch": 0.7690458111870413,
      "grad_norm": 0.9479635953903198,
      "learning_rate": 1.4792704353390623e-05,
      "loss": 0.6031,
      "step": 6077
    },
    {
      "epoch": 0.7691723614274867,
      "grad_norm": 1.1336109638214111,
      "learning_rate": 1.4791131388203082e-05,
      "loss": 0.6212,
      "step": 6078
    },
    {
      "epoch": 0.7692989116679322,
      "grad_norm": 0.8952705264091492,
      "learning_rate": 1.4789558269138442e-05,
      "loss": 0.614,
      "step": 6079
    },
    {
      "epoch": 0.7694254619083776,
      "grad_norm": 0.9946603178977966,
      "learning_rate": 1.4787984996247226e-05,
      "loss": 0.5964,
      "step": 6080
    },
    {
      "epoch": 0.7695520121488231,
      "grad_norm": 0.8944501280784607,
      "learning_rate": 1.4786411569579961e-05,
      "loss": 0.5917,
      "step": 6081
    },
    {
      "epoch": 0.7696785623892686,
      "grad_norm": 0.8832550048828125,
      "learning_rate": 1.4784837989187184e-05,
      "loss": 0.5364,
      "step": 6082
    },
    {
      "epoch": 0.769805112629714,
      "grad_norm": 0.9194617867469788,
      "learning_rate": 1.4783264255119432e-05,
      "loss": 0.6056,
      "step": 6083
    },
    {
      "epoch": 0.7699316628701595,
      "grad_norm": 0.96201491355896,
      "learning_rate": 1.4781690367427253e-05,
      "loss": 0.6433,
      "step": 6084
    },
    {
      "epoch": 0.770058213110605,
      "grad_norm": 0.8929730653762817,
      "learning_rate": 1.4780116326161186e-05,
      "loss": 0.5608,
      "step": 6085
    },
    {
      "epoch": 0.7701847633510503,
      "grad_norm": 0.9570783376693726,
      "learning_rate": 1.4778542131371794e-05,
      "loss": 0.6155,
      "step": 6086
    },
    {
      "epoch": 0.7703113135914958,
      "grad_norm": 0.9269987344741821,
      "learning_rate": 1.477696778310963e-05,
      "loss": 0.651,
      "step": 6087
    },
    {
      "epoch": 0.7704378638319412,
      "grad_norm": 0.9783715605735779,
      "learning_rate": 1.477539328142526e-05,
      "loss": 0.5802,
      "step": 6088
    },
    {
      "epoch": 0.7705644140723867,
      "grad_norm": 0.8493004441261292,
      "learning_rate": 1.4773818626369253e-05,
      "loss": 0.5321,
      "step": 6089
    },
    {
      "epoch": 0.7706909643128322,
      "grad_norm": 0.9257528781890869,
      "learning_rate": 1.4772243817992182e-05,
      "loss": 0.6443,
      "step": 6090
    },
    {
      "epoch": 0.7708175145532776,
      "grad_norm": 0.9605647921562195,
      "learning_rate": 1.4770668856344626e-05,
      "loss": 0.5877,
      "step": 6091
    },
    {
      "epoch": 0.7709440647937231,
      "grad_norm": 0.9918435215950012,
      "learning_rate": 1.4769093741477163e-05,
      "loss": 0.6363,
      "step": 6092
    },
    {
      "epoch": 0.7710706150341685,
      "grad_norm": 1.0762412548065186,
      "learning_rate": 1.4767518473440385e-05,
      "loss": 0.6404,
      "step": 6093
    },
    {
      "epoch": 0.771197165274614,
      "grad_norm": 0.9425109028816223,
      "learning_rate": 1.4765943052284885e-05,
      "loss": 0.5987,
      "step": 6094
    },
    {
      "epoch": 0.7713237155150595,
      "grad_norm": 1.0025279521942139,
      "learning_rate": 1.4764367478061264e-05,
      "loss": 0.6289,
      "step": 6095
    },
    {
      "epoch": 0.7714502657555049,
      "grad_norm": 0.8842350244522095,
      "learning_rate": 1.4762791750820115e-05,
      "loss": 0.5921,
      "step": 6096
    },
    {
      "epoch": 0.7715768159959504,
      "grad_norm": 0.922878086566925,
      "learning_rate": 1.4761215870612055e-05,
      "loss": 0.5856,
      "step": 6097
    },
    {
      "epoch": 0.7717033662363959,
      "grad_norm": 1.0045175552368164,
      "learning_rate": 1.4759639837487696e-05,
      "loss": 0.601,
      "step": 6098
    },
    {
      "epoch": 0.7718299164768413,
      "grad_norm": 0.9672038555145264,
      "learning_rate": 1.4758063651497652e-05,
      "loss": 0.6196,
      "step": 6099
    },
    {
      "epoch": 0.7719564667172868,
      "grad_norm": 0.9807994365692139,
      "learning_rate": 1.4756487312692544e-05,
      "loss": 0.5922,
      "step": 6100
    },
    {
      "epoch": 0.7720830169577322,
      "grad_norm": 0.9354923963546753,
      "learning_rate": 1.4754910821123008e-05,
      "loss": 0.6169,
      "step": 6101
    },
    {
      "epoch": 0.7722095671981777,
      "grad_norm": 1.0201919078826904,
      "learning_rate": 1.4753334176839667e-05,
      "loss": 0.6041,
      "step": 6102
    },
    {
      "epoch": 0.7723361174386232,
      "grad_norm": 1.1235626935958862,
      "learning_rate": 1.4751757379893161e-05,
      "loss": 0.6189,
      "step": 6103
    },
    {
      "epoch": 0.7724626676790686,
      "grad_norm": 0.936603844165802,
      "learning_rate": 1.4750180430334134e-05,
      "loss": 0.6209,
      "step": 6104
    },
    {
      "epoch": 0.7725892179195141,
      "grad_norm": 0.9884146451950073,
      "learning_rate": 1.4748603328213232e-05,
      "loss": 0.5684,
      "step": 6105
    },
    {
      "epoch": 0.7727157681599595,
      "grad_norm": 0.9018115997314453,
      "learning_rate": 1.4747026073581106e-05,
      "loss": 0.6102,
      "step": 6106
    },
    {
      "epoch": 0.772842318400405,
      "grad_norm": 0.9730303287506104,
      "learning_rate": 1.4745448666488414e-05,
      "loss": 0.6074,
      "step": 6107
    },
    {
      "epoch": 0.7729688686408505,
      "grad_norm": 0.919784426689148,
      "learning_rate": 1.4743871106985816e-05,
      "loss": 0.5978,
      "step": 6108
    },
    {
      "epoch": 0.7730954188812958,
      "grad_norm": 0.9644774198532104,
      "learning_rate": 1.474229339512398e-05,
      "loss": 0.5866,
      "step": 6109
    },
    {
      "epoch": 0.7732219691217413,
      "grad_norm": 0.9482862949371338,
      "learning_rate": 1.4740715530953577e-05,
      "loss": 0.6055,
      "step": 6110
    },
    {
      "epoch": 0.7733485193621867,
      "grad_norm": 0.926561176776886,
      "learning_rate": 1.4739137514525285e-05,
      "loss": 0.6154,
      "step": 6111
    },
    {
      "epoch": 0.7734750696026322,
      "grad_norm": 0.951707124710083,
      "learning_rate": 1.4737559345889782e-05,
      "loss": 0.6007,
      "step": 6112
    },
    {
      "epoch": 0.7736016198430777,
      "grad_norm": 0.8724009394645691,
      "learning_rate": 1.4735981025097758e-05,
      "loss": 0.5558,
      "step": 6113
    },
    {
      "epoch": 0.7737281700835231,
      "grad_norm": 0.9237924218177795,
      "learning_rate": 1.4734402552199902e-05,
      "loss": 0.5949,
      "step": 6114
    },
    {
      "epoch": 0.7738547203239686,
      "grad_norm": 0.9388823509216309,
      "learning_rate": 1.4732823927246913e-05,
      "loss": 0.5911,
      "step": 6115
    },
    {
      "epoch": 0.7739812705644141,
      "grad_norm": 0.9512797594070435,
      "learning_rate": 1.4731245150289483e-05,
      "loss": 0.5888,
      "step": 6116
    },
    {
      "epoch": 0.7741078208048595,
      "grad_norm": 0.9612288475036621,
      "learning_rate": 1.4729666221378325e-05,
      "loss": 0.5934,
      "step": 6117
    },
    {
      "epoch": 0.774234371045305,
      "grad_norm": 0.9499718546867371,
      "learning_rate": 1.4728087140564152e-05,
      "loss": 0.5603,
      "step": 6118
    },
    {
      "epoch": 0.7743609212857504,
      "grad_norm": 0.9170525074005127,
      "learning_rate": 1.4726507907897671e-05,
      "loss": 0.654,
      "step": 6119
    },
    {
      "epoch": 0.7744874715261959,
      "grad_norm": 0.9130842685699463,
      "learning_rate": 1.4724928523429609e-05,
      "loss": 0.6019,
      "step": 6120
    },
    {
      "epoch": 0.7746140217666414,
      "grad_norm": 0.8705264925956726,
      "learning_rate": 1.4723348987210688e-05,
      "loss": 0.5597,
      "step": 6121
    },
    {
      "epoch": 0.7747405720070868,
      "grad_norm": 1.1325709819793701,
      "learning_rate": 1.4721769299291637e-05,
      "loss": 0.5989,
      "step": 6122
    },
    {
      "epoch": 0.7748671222475323,
      "grad_norm": 0.9184576869010925,
      "learning_rate": 1.4720189459723195e-05,
      "loss": 0.6163,
      "step": 6123
    },
    {
      "epoch": 0.7749936724879777,
      "grad_norm": 0.9615244269371033,
      "learning_rate": 1.4718609468556105e-05,
      "loss": 0.6144,
      "step": 6124
    },
    {
      "epoch": 0.7751202227284232,
      "grad_norm": 1.0301487445831299,
      "learning_rate": 1.4717029325841099e-05,
      "loss": 0.6447,
      "step": 6125
    },
    {
      "epoch": 0.7752467729688687,
      "grad_norm": 0.9617226123809814,
      "learning_rate": 1.4715449031628937e-05,
      "loss": 0.6086,
      "step": 6126
    },
    {
      "epoch": 0.7753733232093141,
      "grad_norm": 0.8934260010719299,
      "learning_rate": 1.4713868585970372e-05,
      "loss": 0.5586,
      "step": 6127
    },
    {
      "epoch": 0.7754998734497596,
      "grad_norm": 1.0555118322372437,
      "learning_rate": 1.471228798891616e-05,
      "loss": 0.6234,
      "step": 6128
    },
    {
      "epoch": 0.775626423690205,
      "grad_norm": 0.9119481444358826,
      "learning_rate": 1.4710707240517067e-05,
      "loss": 0.6043,
      "step": 6129
    },
    {
      "epoch": 0.7757529739306505,
      "grad_norm": 0.9579862952232361,
      "learning_rate": 1.4709126340823865e-05,
      "loss": 0.5995,
      "step": 6130
    },
    {
      "epoch": 0.775879524171096,
      "grad_norm": 0.9751715064048767,
      "learning_rate": 1.4707545289887322e-05,
      "loss": 0.5648,
      "step": 6131
    },
    {
      "epoch": 0.7760060744115413,
      "grad_norm": 0.9747804999351501,
      "learning_rate": 1.4705964087758218e-05,
      "loss": 0.6266,
      "step": 6132
    },
    {
      "epoch": 0.7761326246519868,
      "grad_norm": 0.9379604458808899,
      "learning_rate": 1.4704382734487341e-05,
      "loss": 0.6166,
      "step": 6133
    },
    {
      "epoch": 0.7762591748924323,
      "grad_norm": 0.9518353343009949,
      "learning_rate": 1.4702801230125474e-05,
      "loss": 0.6465,
      "step": 6134
    },
    {
      "epoch": 0.7763857251328777,
      "grad_norm": 0.9930849075317383,
      "learning_rate": 1.4701219574723413e-05,
      "loss": 0.6213,
      "step": 6135
    },
    {
      "epoch": 0.7765122753733232,
      "grad_norm": 1.1050559282302856,
      "learning_rate": 1.469963776833196e-05,
      "loss": 0.6237,
      "step": 6136
    },
    {
      "epoch": 0.7766388256137686,
      "grad_norm": 0.9240737557411194,
      "learning_rate": 1.4698055811001912e-05,
      "loss": 0.5785,
      "step": 6137
    },
    {
      "epoch": 0.7767653758542141,
      "grad_norm": 1.0403649806976318,
      "learning_rate": 1.4696473702784078e-05,
      "loss": 0.5978,
      "step": 6138
    },
    {
      "epoch": 0.7768919260946596,
      "grad_norm": 1.4536856412887573,
      "learning_rate": 1.4694891443729273e-05,
      "loss": 0.6445,
      "step": 6139
    },
    {
      "epoch": 0.777018476335105,
      "grad_norm": 0.957747757434845,
      "learning_rate": 1.469330903388831e-05,
      "loss": 0.5955,
      "step": 6140
    },
    {
      "epoch": 0.7771450265755505,
      "grad_norm": 1.018406629562378,
      "learning_rate": 1.4691726473312017e-05,
      "loss": 0.6043,
      "step": 6141
    },
    {
      "epoch": 0.7772715768159959,
      "grad_norm": 0.9549241662025452,
      "learning_rate": 1.4690143762051219e-05,
      "loss": 0.5857,
      "step": 6142
    },
    {
      "epoch": 0.7773981270564414,
      "grad_norm": 0.9258644580841064,
      "learning_rate": 1.4688560900156745e-05,
      "loss": 0.6136,
      "step": 6143
    },
    {
      "epoch": 0.7775246772968869,
      "grad_norm": 0.9569690823554993,
      "learning_rate": 1.4686977887679437e-05,
      "loss": 0.6016,
      "step": 6144
    },
    {
      "epoch": 0.7776512275373323,
      "grad_norm": 0.9303547739982605,
      "learning_rate": 1.468539472467013e-05,
      "loss": 0.6124,
      "step": 6145
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 0.9194175601005554,
      "learning_rate": 1.4683811411179678e-05,
      "loss": 0.6214,
      "step": 6146
    },
    {
      "epoch": 0.7779043280182233,
      "grad_norm": 1.0015445947647095,
      "learning_rate": 1.468222794725893e-05,
      "loss": 0.6134,
      "step": 6147
    },
    {
      "epoch": 0.7780308782586687,
      "grad_norm": 0.9506818652153015,
      "learning_rate": 1.4680644332958743e-05,
      "loss": 0.61,
      "step": 6148
    },
    {
      "epoch": 0.7781574284991142,
      "grad_norm": 0.9843763709068298,
      "learning_rate": 1.4679060568329972e-05,
      "loss": 0.5978,
      "step": 6149
    },
    {
      "epoch": 0.7782839787395596,
      "grad_norm": 0.9954895377159119,
      "learning_rate": 1.467747665342349e-05,
      "loss": 0.5974,
      "step": 6150
    },
    {
      "epoch": 0.7784105289800051,
      "grad_norm": 1.0021827220916748,
      "learning_rate": 1.4675892588290167e-05,
      "loss": 0.6509,
      "step": 6151
    },
    {
      "epoch": 0.7785370792204506,
      "grad_norm": 0.9135122299194336,
      "learning_rate": 1.4674308372980871e-05,
      "loss": 0.5681,
      "step": 6152
    },
    {
      "epoch": 0.778663629460896,
      "grad_norm": 0.9001660943031311,
      "learning_rate": 1.4672724007546493e-05,
      "loss": 0.6053,
      "step": 6153
    },
    {
      "epoch": 0.7787901797013415,
      "grad_norm": 0.9350323677062988,
      "learning_rate": 1.4671139492037911e-05,
      "loss": 0.5864,
      "step": 6154
    },
    {
      "epoch": 0.7789167299417868,
      "grad_norm": 0.9283379316329956,
      "learning_rate": 1.4669554826506015e-05,
      "loss": 0.5692,
      "step": 6155
    },
    {
      "epoch": 0.7790432801822323,
      "grad_norm": 0.9663082957267761,
      "learning_rate": 1.4667970011001705e-05,
      "loss": 0.6021,
      "step": 6156
    },
    {
      "epoch": 0.7791698304226778,
      "grad_norm": 0.9172666668891907,
      "learning_rate": 1.4666385045575878e-05,
      "loss": 0.5689,
      "step": 6157
    },
    {
      "epoch": 0.7792963806631232,
      "grad_norm": 0.9352624416351318,
      "learning_rate": 1.4664799930279435e-05,
      "loss": 0.5978,
      "step": 6158
    },
    {
      "epoch": 0.7794229309035687,
      "grad_norm": 1.0303301811218262,
      "learning_rate": 1.466321466516329e-05,
      "loss": 0.6262,
      "step": 6159
    },
    {
      "epoch": 0.7795494811440141,
      "grad_norm": 0.8887065649032593,
      "learning_rate": 1.4661629250278356e-05,
      "loss": 0.5659,
      "step": 6160
    },
    {
      "epoch": 0.7796760313844596,
      "grad_norm": 0.9899806976318359,
      "learning_rate": 1.466004368567555e-05,
      "loss": 0.6705,
      "step": 6161
    },
    {
      "epoch": 0.7798025816249051,
      "grad_norm": 0.9425326585769653,
      "learning_rate": 1.4658457971405798e-05,
      "loss": 0.5925,
      "step": 6162
    },
    {
      "epoch": 0.7799291318653505,
      "grad_norm": 0.9455769658088684,
      "learning_rate": 1.4656872107520027e-05,
      "loss": 0.6211,
      "step": 6163
    },
    {
      "epoch": 0.780055682105796,
      "grad_norm": 0.9566508531570435,
      "learning_rate": 1.465528609406917e-05,
      "loss": 0.6352,
      "step": 6164
    },
    {
      "epoch": 0.7801822323462415,
      "grad_norm": 0.91495680809021,
      "learning_rate": 1.4653699931104168e-05,
      "loss": 0.6041,
      "step": 6165
    },
    {
      "epoch": 0.7803087825866869,
      "grad_norm": 0.9557334184646606,
      "learning_rate": 1.465211361867596e-05,
      "loss": 0.5995,
      "step": 6166
    },
    {
      "epoch": 0.7804353328271324,
      "grad_norm": 0.873988151550293,
      "learning_rate": 1.4650527156835494e-05,
      "loss": 0.5831,
      "step": 6167
    },
    {
      "epoch": 0.7805618830675778,
      "grad_norm": 0.9283025860786438,
      "learning_rate": 1.4648940545633726e-05,
      "loss": 0.6054,
      "step": 6168
    },
    {
      "epoch": 0.7806884333080233,
      "grad_norm": 1.24228835105896,
      "learning_rate": 1.464735378512161e-05,
      "loss": 0.6207,
      "step": 6169
    },
    {
      "epoch": 0.7808149835484688,
      "grad_norm": 0.9734742641448975,
      "learning_rate": 1.464576687535011e-05,
      "loss": 0.5445,
      "step": 6170
    },
    {
      "epoch": 0.7809415337889142,
      "grad_norm": 0.8949166536331177,
      "learning_rate": 1.4644179816370192e-05,
      "loss": 0.5709,
      "step": 6171
    },
    {
      "epoch": 0.7810680840293597,
      "grad_norm": 0.9401496052742004,
      "learning_rate": 1.4642592608232829e-05,
      "loss": 0.6184,
      "step": 6172
    },
    {
      "epoch": 0.7811946342698051,
      "grad_norm": 0.9417930245399475,
      "learning_rate": 1.4641005250988993e-05,
      "loss": 0.5596,
      "step": 6173
    },
    {
      "epoch": 0.7813211845102506,
      "grad_norm": 0.976580798625946,
      "learning_rate": 1.463941774468967e-05,
      "loss": 0.6121,
      "step": 6174
    },
    {
      "epoch": 0.7814477347506961,
      "grad_norm": 0.8847119808197021,
      "learning_rate": 1.4637830089385845e-05,
      "loss": 0.6077,
      "step": 6175
    },
    {
      "epoch": 0.7815742849911415,
      "grad_norm": 0.9549226760864258,
      "learning_rate": 1.4636242285128512e-05,
      "loss": 0.6438,
      "step": 6176
    },
    {
      "epoch": 0.781700835231587,
      "grad_norm": 0.9858276844024658,
      "learning_rate": 1.4634654331968661e-05,
      "loss": 0.6341,
      "step": 6177
    },
    {
      "epoch": 0.7818273854720323,
      "grad_norm": 0.9182687997817993,
      "learning_rate": 1.4633066229957293e-05,
      "loss": 0.5926,
      "step": 6178
    },
    {
      "epoch": 0.7819539357124778,
      "grad_norm": 1.0140999555587769,
      "learning_rate": 1.4631477979145414e-05,
      "loss": 0.5844,
      "step": 6179
    },
    {
      "epoch": 0.7820804859529233,
      "grad_norm": 0.9471853971481323,
      "learning_rate": 1.4629889579584036e-05,
      "loss": 0.6348,
      "step": 6180
    },
    {
      "epoch": 0.7822070361933687,
      "grad_norm": 0.993915855884552,
      "learning_rate": 1.4628301031324172e-05,
      "loss": 0.6155,
      "step": 6181
    },
    {
      "epoch": 0.7823335864338142,
      "grad_norm": 0.9002034068107605,
      "learning_rate": 1.4626712334416845e-05,
      "loss": 0.6073,
      "step": 6182
    },
    {
      "epoch": 0.7824601366742597,
      "grad_norm": 1.0047938823699951,
      "learning_rate": 1.4625123488913073e-05,
      "loss": 0.6089,
      "step": 6183
    },
    {
      "epoch": 0.7825866869147051,
      "grad_norm": 0.9621512293815613,
      "learning_rate": 1.462353449486389e-05,
      "loss": 0.6059,
      "step": 6184
    },
    {
      "epoch": 0.7827132371551506,
      "grad_norm": 0.9285721778869629,
      "learning_rate": 1.4621945352320327e-05,
      "loss": 0.6035,
      "step": 6185
    },
    {
      "epoch": 0.782839787395596,
      "grad_norm": 0.8984435796737671,
      "learning_rate": 1.4620356061333421e-05,
      "loss": 0.6325,
      "step": 6186
    },
    {
      "epoch": 0.7829663376360415,
      "grad_norm": 0.8708816170692444,
      "learning_rate": 1.4618766621954221e-05,
      "loss": 0.6085,
      "step": 6187
    },
    {
      "epoch": 0.783092887876487,
      "grad_norm": 0.9270763397216797,
      "learning_rate": 1.4617177034233773e-05,
      "loss": 0.6223,
      "step": 6188
    },
    {
      "epoch": 0.7832194381169324,
      "grad_norm": 0.9046269655227661,
      "learning_rate": 1.461558729822313e-05,
      "loss": 0.6179,
      "step": 6189
    },
    {
      "epoch": 0.7833459883573779,
      "grad_norm": 0.9482342004776001,
      "learning_rate": 1.4613997413973346e-05,
      "loss": 0.6731,
      "step": 6190
    },
    {
      "epoch": 0.7834725385978233,
      "grad_norm": 0.9555659294128418,
      "learning_rate": 1.4612407381535488e-05,
      "loss": 0.6072,
      "step": 6191
    },
    {
      "epoch": 0.7835990888382688,
      "grad_norm": 0.8920866250991821,
      "learning_rate": 1.4610817200960619e-05,
      "loss": 0.6099,
      "step": 6192
    },
    {
      "epoch": 0.7837256390787143,
      "grad_norm": 0.9488208889961243,
      "learning_rate": 1.4609226872299815e-05,
      "loss": 0.5886,
      "step": 6193
    },
    {
      "epoch": 0.7838521893191597,
      "grad_norm": 0.9251211285591125,
      "learning_rate": 1.4607636395604152e-05,
      "loss": 0.6361,
      "step": 6194
    },
    {
      "epoch": 0.7839787395596052,
      "grad_norm": 0.9608500599861145,
      "learning_rate": 1.4606045770924708e-05,
      "loss": 0.6622,
      "step": 6195
    },
    {
      "epoch": 0.7841052898000507,
      "grad_norm": 0.8893974423408508,
      "learning_rate": 1.4604454998312574e-05,
      "loss": 0.6218,
      "step": 6196
    },
    {
      "epoch": 0.7842318400404961,
      "grad_norm": 0.9200258851051331,
      "learning_rate": 1.4602864077818836e-05,
      "loss": 0.6301,
      "step": 6197
    },
    {
      "epoch": 0.7843583902809416,
      "grad_norm": 0.9607974886894226,
      "learning_rate": 1.4601273009494594e-05,
      "loss": 0.6334,
      "step": 6198
    },
    {
      "epoch": 0.784484940521387,
      "grad_norm": 0.9603307247161865,
      "learning_rate": 1.4599681793390947e-05,
      "loss": 0.5663,
      "step": 6199
    },
    {
      "epoch": 0.7846114907618325,
      "grad_norm": 0.930243194103241,
      "learning_rate": 1.4598090429559003e-05,
      "loss": 0.5935,
      "step": 6200
    },
    {
      "epoch": 0.784738041002278,
      "grad_norm": 1.0662604570388794,
      "learning_rate": 1.4596498918049866e-05,
      "loss": 0.6368,
      "step": 6201
    },
    {
      "epoch": 0.7848645912427233,
      "grad_norm": 0.9701881408691406,
      "learning_rate": 1.4594907258914652e-05,
      "loss": 0.6079,
      "step": 6202
    },
    {
      "epoch": 0.7849911414831688,
      "grad_norm": 1.0555129051208496,
      "learning_rate": 1.4593315452204482e-05,
      "loss": 0.6286,
      "step": 6203
    },
    {
      "epoch": 0.7851176917236142,
      "grad_norm": 0.9873868227005005,
      "learning_rate": 1.4591723497970482e-05,
      "loss": 0.6406,
      "step": 6204
    },
    {
      "epoch": 0.7852442419640597,
      "grad_norm": 0.95042884349823,
      "learning_rate": 1.459013139626378e-05,
      "loss": 0.6326,
      "step": 6205
    },
    {
      "epoch": 0.7853707922045052,
      "grad_norm": 0.9758216142654419,
      "learning_rate": 1.4588539147135507e-05,
      "loss": 0.6018,
      "step": 6206
    },
    {
      "epoch": 0.7854973424449506,
      "grad_norm": 0.9637975096702576,
      "learning_rate": 1.4586946750636803e-05,
      "loss": 0.6029,
      "step": 6207
    },
    {
      "epoch": 0.7856238926853961,
      "grad_norm": 0.8603747487068176,
      "learning_rate": 1.4585354206818812e-05,
      "loss": 0.6439,
      "step": 6208
    },
    {
      "epoch": 0.7857504429258415,
      "grad_norm": 0.8725047707557678,
      "learning_rate": 1.4583761515732679e-05,
      "loss": 0.5642,
      "step": 6209
    },
    {
      "epoch": 0.785876993166287,
      "grad_norm": 1.0069574117660522,
      "learning_rate": 1.458216867742956e-05,
      "loss": 0.5959,
      "step": 6210
    },
    {
      "epoch": 0.7860035434067325,
      "grad_norm": 0.8959051966667175,
      "learning_rate": 1.4580575691960614e-05,
      "loss": 0.6321,
      "step": 6211
    },
    {
      "epoch": 0.7861300936471779,
      "grad_norm": 0.9798262119293213,
      "learning_rate": 1.4578982559376997e-05,
      "loss": 0.5725,
      "step": 6212
    },
    {
      "epoch": 0.7862566438876234,
      "grad_norm": 0.886082112789154,
      "learning_rate": 1.4577389279729878e-05,
      "loss": 0.6122,
      "step": 6213
    },
    {
      "epoch": 0.7863831941280689,
      "grad_norm": 0.9623590111732483,
      "learning_rate": 1.4575795853070434e-05,
      "loss": 0.6139,
      "step": 6214
    },
    {
      "epoch": 0.7865097443685143,
      "grad_norm": 0.9479607343673706,
      "learning_rate": 1.4574202279449831e-05,
      "loss": 0.644,
      "step": 6215
    },
    {
      "epoch": 0.7866362946089598,
      "grad_norm": 0.9021344780921936,
      "learning_rate": 1.457260855891926e-05,
      "loss": 0.6052,
      "step": 6216
    },
    {
      "epoch": 0.7867628448494052,
      "grad_norm": 0.9313355088233948,
      "learning_rate": 1.4571014691529897e-05,
      "loss": 0.6134,
      "step": 6217
    },
    {
      "epoch": 0.7868893950898507,
      "grad_norm": 0.9083965420722961,
      "learning_rate": 1.456942067733294e-05,
      "loss": 0.6306,
      "step": 6218
    },
    {
      "epoch": 0.7870159453302962,
      "grad_norm": 0.9143850803375244,
      "learning_rate": 1.456782651637958e-05,
      "loss": 0.6174,
      "step": 6219
    },
    {
      "epoch": 0.7871424955707416,
      "grad_norm": 0.9872404336929321,
      "learning_rate": 1.4566232208721016e-05,
      "loss": 0.5903,
      "step": 6220
    },
    {
      "epoch": 0.7872690458111871,
      "grad_norm": 0.9639718532562256,
      "learning_rate": 1.4564637754408457e-05,
      "loss": 0.5893,
      "step": 6221
    },
    {
      "epoch": 0.7873955960516325,
      "grad_norm": 0.8829008340835571,
      "learning_rate": 1.4563043153493112e-05,
      "loss": 0.5614,
      "step": 6222
    },
    {
      "epoch": 0.787522146292078,
      "grad_norm": 0.9445704817771912,
      "learning_rate": 1.4561448406026192e-05,
      "loss": 0.5718,
      "step": 6223
    },
    {
      "epoch": 0.7876486965325235,
      "grad_norm": 0.9382862448692322,
      "learning_rate": 1.4559853512058915e-05,
      "loss": 0.6561,
      "step": 6224
    },
    {
      "epoch": 0.7877752467729688,
      "grad_norm": 0.9346461892127991,
      "learning_rate": 1.4558258471642507e-05,
      "loss": 0.6084,
      "step": 6225
    },
    {
      "epoch": 0.7879017970134143,
      "grad_norm": 0.9248414039611816,
      "learning_rate": 1.4556663284828193e-05,
      "loss": 0.5786,
      "step": 6226
    },
    {
      "epoch": 0.7880283472538597,
      "grad_norm": 0.8851456046104431,
      "learning_rate": 1.4555067951667207e-05,
      "loss": 0.5857,
      "step": 6227
    },
    {
      "epoch": 0.7881548974943052,
      "grad_norm": 0.8754459619522095,
      "learning_rate": 1.455347247221079e-05,
      "loss": 0.5982,
      "step": 6228
    },
    {
      "epoch": 0.7882814477347507,
      "grad_norm": 0.8694011569023132,
      "learning_rate": 1.4551876846510179e-05,
      "loss": 0.5784,
      "step": 6229
    },
    {
      "epoch": 0.7884079979751961,
      "grad_norm": 0.914068341255188,
      "learning_rate": 1.4550281074616621e-05,
      "loss": 0.6108,
      "step": 6230
    },
    {
      "epoch": 0.7885345482156416,
      "grad_norm": 0.9207755327224731,
      "learning_rate": 1.454868515658137e-05,
      "loss": 0.5852,
      "step": 6231
    },
    {
      "epoch": 0.7886610984560871,
      "grad_norm": 0.9458100199699402,
      "learning_rate": 1.4547089092455684e-05,
      "loss": 0.575,
      "step": 6232
    },
    {
      "epoch": 0.7887876486965325,
      "grad_norm": 0.9008004665374756,
      "learning_rate": 1.4545492882290819e-05,
      "loss": 0.6414,
      "step": 6233
    },
    {
      "epoch": 0.788914198936978,
      "grad_norm": 1.0233995914459229,
      "learning_rate": 1.4543896526138044e-05,
      "loss": 0.6154,
      "step": 6234
    },
    {
      "epoch": 0.7890407491774234,
      "grad_norm": 1.01310133934021,
      "learning_rate": 1.454230002404863e-05,
      "loss": 0.6348,
      "step": 6235
    },
    {
      "epoch": 0.7891672994178689,
      "grad_norm": 0.9172359704971313,
      "learning_rate": 1.4540703376073846e-05,
      "loss": 0.5831,
      "step": 6236
    },
    {
      "epoch": 0.7892938496583144,
      "grad_norm": 0.9686781764030457,
      "learning_rate": 1.4539106582264979e-05,
      "loss": 0.5536,
      "step": 6237
    },
    {
      "epoch": 0.7894203998987598,
      "grad_norm": 0.950230062007904,
      "learning_rate": 1.453750964267331e-05,
      "loss": 0.6444,
      "step": 6238
    },
    {
      "epoch": 0.7895469501392053,
      "grad_norm": 0.9419342279434204,
      "learning_rate": 1.4535912557350131e-05,
      "loss": 0.6124,
      "step": 6239
    },
    {
      "epoch": 0.7896735003796507,
      "grad_norm": 1.0289037227630615,
      "learning_rate": 1.4534315326346728e-05,
      "loss": 0.6358,
      "step": 6240
    },
    {
      "epoch": 0.7898000506200962,
      "grad_norm": 0.9505602121353149,
      "learning_rate": 1.4532717949714408e-05,
      "loss": 0.5965,
      "step": 6241
    },
    {
      "epoch": 0.7899266008605417,
      "grad_norm": 0.9413037896156311,
      "learning_rate": 1.4531120427504468e-05,
      "loss": 0.5982,
      "step": 6242
    },
    {
      "epoch": 0.7900531511009871,
      "grad_norm": 0.9812368750572205,
      "learning_rate": 1.4529522759768217e-05,
      "loss": 0.6049,
      "step": 6243
    },
    {
      "epoch": 0.7901797013414326,
      "grad_norm": 0.948453426361084,
      "learning_rate": 1.452792494655697e-05,
      "loss": 0.6044,
      "step": 6244
    },
    {
      "epoch": 0.7903062515818781,
      "grad_norm": 0.9239628911018372,
      "learning_rate": 1.4526326987922044e-05,
      "loss": 0.6004,
      "step": 6245
    },
    {
      "epoch": 0.7904328018223234,
      "grad_norm": 0.9566352367401123,
      "learning_rate": 1.4524728883914759e-05,
      "loss": 0.6268,
      "step": 6246
    },
    {
      "epoch": 0.790559352062769,
      "grad_norm": 1.0354541540145874,
      "learning_rate": 1.452313063458644e-05,
      "loss": 0.6236,
      "step": 6247
    },
    {
      "epoch": 0.7906859023032143,
      "grad_norm": 0.9319937229156494,
      "learning_rate": 1.4521532239988421e-05,
      "loss": 0.5763,
      "step": 6248
    },
    {
      "epoch": 0.7908124525436598,
      "grad_norm": 0.953366219997406,
      "learning_rate": 1.4519933700172035e-05,
      "loss": 0.5949,
      "step": 6249
    },
    {
      "epoch": 0.7909390027841053,
      "grad_norm": 0.9830202460289001,
      "learning_rate": 1.4518335015188628e-05,
      "loss": 0.6008,
      "step": 6250
    },
    {
      "epoch": 0.7910655530245507,
      "grad_norm": 0.9060303568840027,
      "learning_rate": 1.451673618508954e-05,
      "loss": 0.6102,
      "step": 6251
    },
    {
      "epoch": 0.7911921032649962,
      "grad_norm": 0.9881319403648376,
      "learning_rate": 1.451513720992612e-05,
      "loss": 0.6209,
      "step": 6252
    },
    {
      "epoch": 0.7913186535054416,
      "grad_norm": 0.9688220620155334,
      "learning_rate": 1.4513538089749725e-05,
      "loss": 0.5926,
      "step": 6253
    },
    {
      "epoch": 0.7914452037458871,
      "grad_norm": 1.0664409399032593,
      "learning_rate": 1.4511938824611714e-05,
      "loss": 0.6419,
      "step": 6254
    },
    {
      "epoch": 0.7915717539863326,
      "grad_norm": 0.92633455991745,
      "learning_rate": 1.4510339414563448e-05,
      "loss": 0.6221,
      "step": 6255
    },
    {
      "epoch": 0.791698304226778,
      "grad_norm": 0.895087480545044,
      "learning_rate": 1.4508739859656298e-05,
      "loss": 0.5903,
      "step": 6256
    },
    {
      "epoch": 0.7918248544672235,
      "grad_norm": 0.9413992762565613,
      "learning_rate": 1.4507140159941636e-05,
      "loss": 0.5943,
      "step": 6257
    },
    {
      "epoch": 0.7919514047076689,
      "grad_norm": 0.8789564371109009,
      "learning_rate": 1.4505540315470843e-05,
      "loss": 0.5752,
      "step": 6258
    },
    {
      "epoch": 0.7920779549481144,
      "grad_norm": 0.9040088057518005,
      "learning_rate": 1.4503940326295295e-05,
      "loss": 0.5887,
      "step": 6259
    },
    {
      "epoch": 0.7922045051885599,
      "grad_norm": 0.903374195098877,
      "learning_rate": 1.4502340192466385e-05,
      "loss": 0.6035,
      "step": 6260
    },
    {
      "epoch": 0.7923310554290053,
      "grad_norm": 1.1363320350646973,
      "learning_rate": 1.4500739914035498e-05,
      "loss": 0.6185,
      "step": 6261
    },
    {
      "epoch": 0.7924576056694508,
      "grad_norm": 0.8959009051322937,
      "learning_rate": 1.4499139491054039e-05,
      "loss": 0.6054,
      "step": 6262
    },
    {
      "epoch": 0.7925841559098963,
      "grad_norm": 2.2825751304626465,
      "learning_rate": 1.4497538923573403e-05,
      "loss": 0.6091,
      "step": 6263
    },
    {
      "epoch": 0.7927107061503417,
      "grad_norm": 0.9061213135719299,
      "learning_rate": 1.4495938211644995e-05,
      "loss": 0.622,
      "step": 6264
    },
    {
      "epoch": 0.7928372563907872,
      "grad_norm": 0.9348456859588623,
      "learning_rate": 1.4494337355320229e-05,
      "loss": 0.6218,
      "step": 6265
    },
    {
      "epoch": 0.7929638066312326,
      "grad_norm": 0.9612807631492615,
      "learning_rate": 1.4492736354650514e-05,
      "loss": 0.6196,
      "step": 6266
    },
    {
      "epoch": 0.7930903568716781,
      "grad_norm": 0.9181503653526306,
      "learning_rate": 1.4491135209687275e-05,
      "loss": 0.588,
      "step": 6267
    },
    {
      "epoch": 0.7932169071121236,
      "grad_norm": 0.9864458441734314,
      "learning_rate": 1.4489533920481935e-05,
      "loss": 0.5976,
      "step": 6268
    },
    {
      "epoch": 0.793343457352569,
      "grad_norm": 1.1457419395446777,
      "learning_rate": 1.4487932487085923e-05,
      "loss": 0.6029,
      "step": 6269
    },
    {
      "epoch": 0.7934700075930144,
      "grad_norm": 0.9095690846443176,
      "learning_rate": 1.4486330909550668e-05,
      "loss": 0.5961,
      "step": 6270
    },
    {
      "epoch": 0.7935965578334598,
      "grad_norm": 0.9470913410186768,
      "learning_rate": 1.4484729187927613e-05,
      "loss": 0.593,
      "step": 6271
    },
    {
      "epoch": 0.7937231080739053,
      "grad_norm": 1.1205883026123047,
      "learning_rate": 1.4483127322268199e-05,
      "loss": 0.6235,
      "step": 6272
    },
    {
      "epoch": 0.7938496583143508,
      "grad_norm": 0.9238761067390442,
      "learning_rate": 1.4481525312623869e-05,
      "loss": 0.5966,
      "step": 6273
    },
    {
      "epoch": 0.7939762085547962,
      "grad_norm": 0.9553199410438538,
      "learning_rate": 1.4479923159046085e-05,
      "loss": 0.5975,
      "step": 6274
    },
    {
      "epoch": 0.7941027587952417,
      "grad_norm": 1.1415817737579346,
      "learning_rate": 1.4478320861586294e-05,
      "loss": 0.5944,
      "step": 6275
    },
    {
      "epoch": 0.7942293090356871,
      "grad_norm": 0.9100820422172546,
      "learning_rate": 1.4476718420295964e-05,
      "loss": 0.6094,
      "step": 6276
    },
    {
      "epoch": 0.7943558592761326,
      "grad_norm": 0.9712570309638977,
      "learning_rate": 1.4475115835226552e-05,
      "loss": 0.581,
      "step": 6277
    },
    {
      "epoch": 0.7944824095165781,
      "grad_norm": 0.9473804831504822,
      "learning_rate": 1.4473513106429536e-05,
      "loss": 0.6026,
      "step": 6278
    },
    {
      "epoch": 0.7946089597570235,
      "grad_norm": 0.9712766408920288,
      "learning_rate": 1.4471910233956387e-05,
      "loss": 0.5862,
      "step": 6279
    },
    {
      "epoch": 0.794735509997469,
      "grad_norm": 0.8500210046768188,
      "learning_rate": 1.447030721785859e-05,
      "loss": 0.5484,
      "step": 6280
    },
    {
      "epoch": 0.7948620602379145,
      "grad_norm": 0.9410779476165771,
      "learning_rate": 1.4468704058187623e-05,
      "loss": 0.5881,
      "step": 6281
    },
    {
      "epoch": 0.7949886104783599,
      "grad_norm": 1.016420602798462,
      "learning_rate": 1.4467100754994976e-05,
      "loss": 0.6438,
      "step": 6282
    },
    {
      "epoch": 0.7951151607188054,
      "grad_norm": 0.9046664834022522,
      "learning_rate": 1.4465497308332143e-05,
      "loss": 0.6129,
      "step": 6283
    },
    {
      "epoch": 0.7952417109592508,
      "grad_norm": 0.9442148804664612,
      "learning_rate": 1.4463893718250623e-05,
      "loss": 0.619,
      "step": 6284
    },
    {
      "epoch": 0.7953682611996963,
      "grad_norm": 1.3010650873184204,
      "learning_rate": 1.446228998480192e-05,
      "loss": 0.6193,
      "step": 6285
    },
    {
      "epoch": 0.7954948114401418,
      "grad_norm": 0.8344910740852356,
      "learning_rate": 1.4460686108037538e-05,
      "loss": 0.564,
      "step": 6286
    },
    {
      "epoch": 0.7956213616805872,
      "grad_norm": 0.9112892150878906,
      "learning_rate": 1.4459082088008989e-05,
      "loss": 0.5882,
      "step": 6287
    },
    {
      "epoch": 0.7957479119210327,
      "grad_norm": 0.9050635099411011,
      "learning_rate": 1.4457477924767794e-05,
      "loss": 0.6083,
      "step": 6288
    },
    {
      "epoch": 0.7958744621614781,
      "grad_norm": 0.8898835182189941,
      "learning_rate": 1.4455873618365467e-05,
      "loss": 0.6049,
      "step": 6289
    },
    {
      "epoch": 0.7960010124019236,
      "grad_norm": 0.8959856629371643,
      "learning_rate": 1.4454269168853538e-05,
      "loss": 0.6057,
      "step": 6290
    },
    {
      "epoch": 0.7961275626423691,
      "grad_norm": 1.0034390687942505,
      "learning_rate": 1.4452664576283537e-05,
      "loss": 0.617,
      "step": 6291
    },
    {
      "epoch": 0.7962541128828144,
      "grad_norm": 0.9472471475601196,
      "learning_rate": 1.4451059840706998e-05,
      "loss": 0.5761,
      "step": 6292
    },
    {
      "epoch": 0.79638066312326,
      "grad_norm": 0.9760855436325073,
      "learning_rate": 1.444945496217546e-05,
      "loss": 0.6417,
      "step": 6293
    },
    {
      "epoch": 0.7965072133637054,
      "grad_norm": 0.9006792306900024,
      "learning_rate": 1.4447849940740466e-05,
      "loss": 0.6131,
      "step": 6294
    },
    {
      "epoch": 0.7966337636041508,
      "grad_norm": 0.9180368185043335,
      "learning_rate": 1.4446244776453569e-05,
      "loss": 0.6369,
      "step": 6295
    },
    {
      "epoch": 0.7967603138445963,
      "grad_norm": 0.8832228779792786,
      "learning_rate": 1.4444639469366317e-05,
      "loss": 0.5833,
      "step": 6296
    },
    {
      "epoch": 0.7968868640850417,
      "grad_norm": 0.881236732006073,
      "learning_rate": 1.4443034019530273e-05,
      "loss": 0.6164,
      "step": 6297
    },
    {
      "epoch": 0.7970134143254872,
      "grad_norm": 0.8793712258338928,
      "learning_rate": 1.4441428426996993e-05,
      "loss": 0.572,
      "step": 6298
    },
    {
      "epoch": 0.7971399645659327,
      "grad_norm": 0.9510046243667603,
      "learning_rate": 1.4439822691818051e-05,
      "loss": 0.6329,
      "step": 6299
    },
    {
      "epoch": 0.7972665148063781,
      "grad_norm": 0.9322739243507385,
      "learning_rate": 1.4438216814045012e-05,
      "loss": 0.618,
      "step": 6300
    },
    {
      "epoch": 0.7973930650468236,
      "grad_norm": 0.963421106338501,
      "learning_rate": 1.4436610793729458e-05,
      "loss": 0.6117,
      "step": 6301
    },
    {
      "epoch": 0.797519615287269,
      "grad_norm": 0.8717143535614014,
      "learning_rate": 1.4435004630922963e-05,
      "loss": 0.5631,
      "step": 6302
    },
    {
      "epoch": 0.7976461655277145,
      "grad_norm": 0.8536733388900757,
      "learning_rate": 1.4433398325677117e-05,
      "loss": 0.5845,
      "step": 6303
    },
    {
      "epoch": 0.79777271576816,
      "grad_norm": 0.9431373476982117,
      "learning_rate": 1.4431791878043511e-05,
      "loss": 0.587,
      "step": 6304
    },
    {
      "epoch": 0.7978992660086054,
      "grad_norm": 0.9664932489395142,
      "learning_rate": 1.4430185288073735e-05,
      "loss": 0.6136,
      "step": 6305
    },
    {
      "epoch": 0.7980258162490509,
      "grad_norm": 0.9715459942817688,
      "learning_rate": 1.4428578555819389e-05,
      "loss": 0.6437,
      "step": 6306
    },
    {
      "epoch": 0.7981523664894963,
      "grad_norm": 0.9799660444259644,
      "learning_rate": 1.4426971681332078e-05,
      "loss": 0.573,
      "step": 6307
    },
    {
      "epoch": 0.7982789167299418,
      "grad_norm": 0.9325512051582336,
      "learning_rate": 1.4425364664663412e-05,
      "loss": 0.5448,
      "step": 6308
    },
    {
      "epoch": 0.7984054669703873,
      "grad_norm": 0.9103249311447144,
      "learning_rate": 1.4423757505865e-05,
      "loss": 0.5889,
      "step": 6309
    },
    {
      "epoch": 0.7985320172108327,
      "grad_norm": 0.8913708925247192,
      "learning_rate": 1.4422150204988459e-05,
      "loss": 0.5857,
      "step": 6310
    },
    {
      "epoch": 0.7986585674512782,
      "grad_norm": 0.9321138262748718,
      "learning_rate": 1.4420542762085412e-05,
      "loss": 0.5952,
      "step": 6311
    },
    {
      "epoch": 0.7987851176917237,
      "grad_norm": 0.9274806976318359,
      "learning_rate": 1.441893517720749e-05,
      "loss": 0.5824,
      "step": 6312
    },
    {
      "epoch": 0.7989116679321691,
      "grad_norm": 0.9367141127586365,
      "learning_rate": 1.4417327450406316e-05,
      "loss": 0.6098,
      "step": 6313
    },
    {
      "epoch": 0.7990382181726146,
      "grad_norm": 0.993442714214325,
      "learning_rate": 1.4415719581733531e-05,
      "loss": 0.6045,
      "step": 6314
    },
    {
      "epoch": 0.79916476841306,
      "grad_norm": 0.9246801137924194,
      "learning_rate": 1.4414111571240771e-05,
      "loss": 0.5884,
      "step": 6315
    },
    {
      "epoch": 0.7992913186535054,
      "grad_norm": 0.91633141040802,
      "learning_rate": 1.4412503418979683e-05,
      "loss": 0.5979,
      "step": 6316
    },
    {
      "epoch": 0.799417868893951,
      "grad_norm": 0.9085030555725098,
      "learning_rate": 1.4410895125001918e-05,
      "loss": 0.6313,
      "step": 6317
    },
    {
      "epoch": 0.7995444191343963,
      "grad_norm": 0.9443812370300293,
      "learning_rate": 1.4409286689359126e-05,
      "loss": 0.6246,
      "step": 6318
    },
    {
      "epoch": 0.7996709693748418,
      "grad_norm": 0.9073315262794495,
      "learning_rate": 1.4407678112102968e-05,
      "loss": 0.5977,
      "step": 6319
    },
    {
      "epoch": 0.7997975196152872,
      "grad_norm": 0.9278984665870667,
      "learning_rate": 1.4406069393285105e-05,
      "loss": 0.6134,
      "step": 6320
    },
    {
      "epoch": 0.7999240698557327,
      "grad_norm": 0.9954926371574402,
      "learning_rate": 1.4404460532957206e-05,
      "loss": 0.6481,
      "step": 6321
    },
    {
      "epoch": 0.8000506200961782,
      "grad_norm": 0.9039890170097351,
      "learning_rate": 1.4402851531170937e-05,
      "loss": 0.5808,
      "step": 6322
    },
    {
      "epoch": 0.8001771703366236,
      "grad_norm": 1.1715699434280396,
      "learning_rate": 1.4401242387977983e-05,
      "loss": 0.5931,
      "step": 6323
    },
    {
      "epoch": 0.8003037205770691,
      "grad_norm": 0.9607008695602417,
      "learning_rate": 1.4399633103430021e-05,
      "loss": 0.5826,
      "step": 6324
    },
    {
      "epoch": 0.8004302708175145,
      "grad_norm": 0.8829991221427917,
      "learning_rate": 1.4398023677578739e-05,
      "loss": 0.5835,
      "step": 6325
    },
    {
      "epoch": 0.80055682105796,
      "grad_norm": 0.928693413734436,
      "learning_rate": 1.4396414110475822e-05,
      "loss": 0.5809,
      "step": 6326
    },
    {
      "epoch": 0.8006833712984055,
      "grad_norm": 0.8579795360565186,
      "learning_rate": 1.4394804402172965e-05,
      "loss": 0.5995,
      "step": 6327
    },
    {
      "epoch": 0.8008099215388509,
      "grad_norm": 0.8858914971351624,
      "learning_rate": 1.4393194552721875e-05,
      "loss": 0.5732,
      "step": 6328
    },
    {
      "epoch": 0.8009364717792964,
      "grad_norm": 0.8624101877212524,
      "learning_rate": 1.4391584562174243e-05,
      "loss": 0.6281,
      "step": 6329
    },
    {
      "epoch": 0.8010630220197419,
      "grad_norm": 0.9241023063659668,
      "learning_rate": 1.4389974430581787e-05,
      "loss": 0.6534,
      "step": 6330
    },
    {
      "epoch": 0.8011895722601873,
      "grad_norm": 0.9107518792152405,
      "learning_rate": 1.4388364157996221e-05,
      "loss": 0.5936,
      "step": 6331
    },
    {
      "epoch": 0.8013161225006328,
      "grad_norm": 0.9313393235206604,
      "learning_rate": 1.4386753744469255e-05,
      "loss": 0.627,
      "step": 6332
    },
    {
      "epoch": 0.8014426727410782,
      "grad_norm": 0.8930660486221313,
      "learning_rate": 1.4385143190052616e-05,
      "loss": 0.6252,
      "step": 6333
    },
    {
      "epoch": 0.8015692229815237,
      "grad_norm": 0.911413848400116,
      "learning_rate": 1.4383532494798024e-05,
      "loss": 0.6211,
      "step": 6334
    },
    {
      "epoch": 0.8016957732219692,
      "grad_norm": 0.9269979596138,
      "learning_rate": 1.4381921658757216e-05,
      "loss": 0.6497,
      "step": 6335
    },
    {
      "epoch": 0.8018223234624146,
      "grad_norm": 0.9511229395866394,
      "learning_rate": 1.4380310681981927e-05,
      "loss": 0.6838,
      "step": 6336
    },
    {
      "epoch": 0.80194887370286,
      "grad_norm": 0.8654604554176331,
      "learning_rate": 1.4378699564523894e-05,
      "loss": 0.5811,
      "step": 6337
    },
    {
      "epoch": 0.8020754239433054,
      "grad_norm": 0.9529143571853638,
      "learning_rate": 1.4377088306434864e-05,
      "loss": 0.6175,
      "step": 6338
    },
    {
      "epoch": 0.8022019741837509,
      "grad_norm": 0.978054940700531,
      "learning_rate": 1.4375476907766583e-05,
      "loss": 0.6472,
      "step": 6339
    },
    {
      "epoch": 0.8023285244241964,
      "grad_norm": 0.9188019633293152,
      "learning_rate": 1.4373865368570806e-05,
      "loss": 0.6055,
      "step": 6340
    },
    {
      "epoch": 0.8024550746646418,
      "grad_norm": 1.014862298965454,
      "learning_rate": 1.437225368889929e-05,
      "loss": 0.6571,
      "step": 6341
    },
    {
      "epoch": 0.8025816249050873,
      "grad_norm": 1.08307683467865,
      "learning_rate": 1.4370641868803798e-05,
      "loss": 0.6008,
      "step": 6342
    },
    {
      "epoch": 0.8027081751455328,
      "grad_norm": 0.97823166847229,
      "learning_rate": 1.4369029908336098e-05,
      "loss": 0.624,
      "step": 6343
    },
    {
      "epoch": 0.8028347253859782,
      "grad_norm": 0.8880308866500854,
      "learning_rate": 1.4367417807547961e-05,
      "loss": 0.5876,
      "step": 6344
    },
    {
      "epoch": 0.8029612756264237,
      "grad_norm": 0.921758234500885,
      "learning_rate": 1.4365805566491164e-05,
      "loss": 0.6313,
      "step": 6345
    },
    {
      "epoch": 0.8030878258668691,
      "grad_norm": 0.9230298399925232,
      "learning_rate": 1.4364193185217484e-05,
      "loss": 0.5954,
      "step": 6346
    },
    {
      "epoch": 0.8032143761073146,
      "grad_norm": 0.926233172416687,
      "learning_rate": 1.4362580663778706e-05,
      "loss": 0.6044,
      "step": 6347
    },
    {
      "epoch": 0.8033409263477601,
      "grad_norm": 0.8785331845283508,
      "learning_rate": 1.4360968002226625e-05,
      "loss": 0.6047,
      "step": 6348
    },
    {
      "epoch": 0.8034674765882055,
      "grad_norm": 0.947014570236206,
      "learning_rate": 1.4359355200613032e-05,
      "loss": 0.5776,
      "step": 6349
    },
    {
      "epoch": 0.803594026828651,
      "grad_norm": 0.9300737380981445,
      "learning_rate": 1.4357742258989724e-05,
      "loss": 0.6007,
      "step": 6350
    },
    {
      "epoch": 0.8037205770690964,
      "grad_norm": 0.9002818465232849,
      "learning_rate": 1.4356129177408504e-05,
      "loss": 0.5576,
      "step": 6351
    },
    {
      "epoch": 0.8038471273095419,
      "grad_norm": 0.8858178853988647,
      "learning_rate": 1.435451595592118e-05,
      "loss": 0.5772,
      "step": 6352
    },
    {
      "epoch": 0.8039736775499874,
      "grad_norm": 0.9789788722991943,
      "learning_rate": 1.4352902594579563e-05,
      "loss": 0.6054,
      "step": 6353
    },
    {
      "epoch": 0.8041002277904328,
      "grad_norm": 0.8935126662254333,
      "learning_rate": 1.4351289093435473e-05,
      "loss": 0.5556,
      "step": 6354
    },
    {
      "epoch": 0.8042267780308783,
      "grad_norm": 0.9790074825286865,
      "learning_rate": 1.4349675452540728e-05,
      "loss": 0.5853,
      "step": 6355
    },
    {
      "epoch": 0.8043533282713237,
      "grad_norm": 0.9387352466583252,
      "learning_rate": 1.4348061671947153e-05,
      "loss": 0.5934,
      "step": 6356
    },
    {
      "epoch": 0.8044798785117692,
      "grad_norm": 1.0166525840759277,
      "learning_rate": 1.4346447751706581e-05,
      "loss": 0.6084,
      "step": 6357
    },
    {
      "epoch": 0.8046064287522147,
      "grad_norm": 0.9970707297325134,
      "learning_rate": 1.434483369187084e-05,
      "loss": 0.6078,
      "step": 6358
    },
    {
      "epoch": 0.80473297899266,
      "grad_norm": 1.0166873931884766,
      "learning_rate": 1.4343219492491778e-05,
      "loss": 0.6278,
      "step": 6359
    },
    {
      "epoch": 0.8048595292331056,
      "grad_norm": 0.9228894710540771,
      "learning_rate": 1.4341605153621231e-05,
      "loss": 0.5537,
      "step": 6360
    },
    {
      "epoch": 0.804986079473551,
      "grad_norm": 0.9494940638542175,
      "learning_rate": 1.433999067531105e-05,
      "loss": 0.6204,
      "step": 6361
    },
    {
      "epoch": 0.8051126297139964,
      "grad_norm": 1.155038595199585,
      "learning_rate": 1.4338376057613088e-05,
      "loss": 0.6001,
      "step": 6362
    },
    {
      "epoch": 0.8052391799544419,
      "grad_norm": 0.9772529006004333,
      "learning_rate": 1.43367613005792e-05,
      "loss": 0.5914,
      "step": 6363
    },
    {
      "epoch": 0.8053657301948873,
      "grad_norm": 0.9030422568321228,
      "learning_rate": 1.4335146404261247e-05,
      "loss": 0.5589,
      "step": 6364
    },
    {
      "epoch": 0.8054922804353328,
      "grad_norm": 1.1063188314437866,
      "learning_rate": 1.4333531368711095e-05,
      "loss": 0.6037,
      "step": 6365
    },
    {
      "epoch": 0.8056188306757783,
      "grad_norm": 0.9129397869110107,
      "learning_rate": 1.4331916193980617e-05,
      "loss": 0.5754,
      "step": 6366
    },
    {
      "epoch": 0.8057453809162237,
      "grad_norm": 0.9845894575119019,
      "learning_rate": 1.4330300880121682e-05,
      "loss": 0.6107,
      "step": 6367
    },
    {
      "epoch": 0.8058719311566692,
      "grad_norm": 0.954099178314209,
      "learning_rate": 1.4328685427186172e-05,
      "loss": 0.6129,
      "step": 6368
    },
    {
      "epoch": 0.8059984813971146,
      "grad_norm": 0.9589537978172302,
      "learning_rate": 1.4327069835225974e-05,
      "loss": 0.6141,
      "step": 6369
    },
    {
      "epoch": 0.8061250316375601,
      "grad_norm": 0.9817461371421814,
      "learning_rate": 1.4325454104292972e-05,
      "loss": 0.6261,
      "step": 6370
    },
    {
      "epoch": 0.8062515818780056,
      "grad_norm": 0.984397828578949,
      "learning_rate": 1.432383823443906e-05,
      "loss": 0.586,
      "step": 6371
    },
    {
      "epoch": 0.806378132118451,
      "grad_norm": 0.9746671915054321,
      "learning_rate": 1.4322222225716135e-05,
      "loss": 0.6148,
      "step": 6372
    },
    {
      "epoch": 0.8065046823588965,
      "grad_norm": 1.0283358097076416,
      "learning_rate": 1.43206060781761e-05,
      "loss": 0.586,
      "step": 6373
    },
    {
      "epoch": 0.8066312325993419,
      "grad_norm": 0.8750171065330505,
      "learning_rate": 1.4318989791870856e-05,
      "loss": 0.5868,
      "step": 6374
    },
    {
      "epoch": 0.8067577828397874,
      "grad_norm": 0.9284573197364807,
      "learning_rate": 1.4317373366852318e-05,
      "loss": 0.5514,
      "step": 6375
    },
    {
      "epoch": 0.8068843330802329,
      "grad_norm": 0.972199559211731,
      "learning_rate": 1.4315756803172401e-05,
      "loss": 0.6346,
      "step": 6376
    },
    {
      "epoch": 0.8070108833206783,
      "grad_norm": 0.9801551699638367,
      "learning_rate": 1.431414010088302e-05,
      "loss": 0.6049,
      "step": 6377
    },
    {
      "epoch": 0.8071374335611238,
      "grad_norm": 0.9814848899841309,
      "learning_rate": 1.4312523260036101e-05,
      "loss": 0.6008,
      "step": 6378
    },
    {
      "epoch": 0.8072639838015693,
      "grad_norm": 0.9750253558158875,
      "learning_rate": 1.4310906280683575e-05,
      "loss": 0.5971,
      "step": 6379
    },
    {
      "epoch": 0.8073905340420147,
      "grad_norm": 0.9484819769859314,
      "learning_rate": 1.4309289162877372e-05,
      "loss": 0.6206,
      "step": 6380
    },
    {
      "epoch": 0.8075170842824602,
      "grad_norm": 0.9095343351364136,
      "learning_rate": 1.4307671906669428e-05,
      "loss": 0.5877,
      "step": 6381
    },
    {
      "epoch": 0.8076436345229056,
      "grad_norm": 0.9046811461448669,
      "learning_rate": 1.4306054512111682e-05,
      "loss": 0.592,
      "step": 6382
    },
    {
      "epoch": 0.807770184763351,
      "grad_norm": 0.9082105755805969,
      "learning_rate": 1.4304436979256091e-05,
      "loss": 0.6129,
      "step": 6383
    },
    {
      "epoch": 0.8078967350037966,
      "grad_norm": 0.9595736265182495,
      "learning_rate": 1.4302819308154594e-05,
      "loss": 0.5853,
      "step": 6384
    },
    {
      "epoch": 0.8080232852442419,
      "grad_norm": 0.9010822772979736,
      "learning_rate": 1.430120149885915e-05,
      "loss": 0.5808,
      "step": 6385
    },
    {
      "epoch": 0.8081498354846874,
      "grad_norm": 0.9035583734512329,
      "learning_rate": 1.4299583551421721e-05,
      "loss": 0.606,
      "step": 6386
    },
    {
      "epoch": 0.8082763857251328,
      "grad_norm": 0.850709080696106,
      "learning_rate": 1.4297965465894266e-05,
      "loss": 0.6058,
      "step": 6387
    },
    {
      "epoch": 0.8084029359655783,
      "grad_norm": 0.9624572396278381,
      "learning_rate": 1.4296347242328756e-05,
      "loss": 0.5918,
      "step": 6388
    },
    {
      "epoch": 0.8085294862060238,
      "grad_norm": 0.9293954968452454,
      "learning_rate": 1.4294728880777165e-05,
      "loss": 0.603,
      "step": 6389
    },
    {
      "epoch": 0.8086560364464692,
      "grad_norm": 0.9368062019348145,
      "learning_rate": 1.4293110381291464e-05,
      "loss": 0.6419,
      "step": 6390
    },
    {
      "epoch": 0.8087825866869147,
      "grad_norm": 0.9614192843437195,
      "learning_rate": 1.429149174392364e-05,
      "loss": 0.6246,
      "step": 6391
    },
    {
      "epoch": 0.8089091369273602,
      "grad_norm": 0.8920786380767822,
      "learning_rate": 1.4289872968725675e-05,
      "loss": 0.5628,
      "step": 6392
    },
    {
      "epoch": 0.8090356871678056,
      "grad_norm": 0.9787629842758179,
      "learning_rate": 1.4288254055749565e-05,
      "loss": 0.5956,
      "step": 6393
    },
    {
      "epoch": 0.8091622374082511,
      "grad_norm": 0.9335732460021973,
      "learning_rate": 1.4286635005047299e-05,
      "loss": 0.5986,
      "step": 6394
    },
    {
      "epoch": 0.8092887876486965,
      "grad_norm": 1.0022039413452148,
      "learning_rate": 1.4285015816670879e-05,
      "loss": 0.5705,
      "step": 6395
    },
    {
      "epoch": 0.809415337889142,
      "grad_norm": 0.9103634357452393,
      "learning_rate": 1.428339649067231e-05,
      "loss": 0.6321,
      "step": 6396
    },
    {
      "epoch": 0.8095418881295875,
      "grad_norm": 0.9750420451164246,
      "learning_rate": 1.4281777027103596e-05,
      "loss": 0.6123,
      "step": 6397
    },
    {
      "epoch": 0.8096684383700329,
      "grad_norm": 0.8970471620559692,
      "learning_rate": 1.4280157426016752e-05,
      "loss": 0.647,
      "step": 6398
    },
    {
      "epoch": 0.8097949886104784,
      "grad_norm": 0.9844629764556885,
      "learning_rate": 1.4278537687463794e-05,
      "loss": 0.6039,
      "step": 6399
    },
    {
      "epoch": 0.8099215388509238,
      "grad_norm": 0.9092326760292053,
      "learning_rate": 1.4276917811496746e-05,
      "loss": 0.5852,
      "step": 6400
    },
    {
      "epoch": 0.8100480890913693,
      "grad_norm": 0.8720864057540894,
      "learning_rate": 1.4275297798167629e-05,
      "loss": 0.5867,
      "step": 6401
    },
    {
      "epoch": 0.8101746393318148,
      "grad_norm": 0.912558376789093,
      "learning_rate": 1.4273677647528474e-05,
      "loss": 0.6059,
      "step": 6402
    },
    {
      "epoch": 0.8103011895722602,
      "grad_norm": 0.9211778044700623,
      "learning_rate": 1.4272057359631317e-05,
      "loss": 0.5559,
      "step": 6403
    },
    {
      "epoch": 0.8104277398127057,
      "grad_norm": 0.9043623208999634,
      "learning_rate": 1.4270436934528198e-05,
      "loss": 0.5747,
      "step": 6404
    },
    {
      "epoch": 0.810554290053151,
      "grad_norm": 0.912589430809021,
      "learning_rate": 1.426881637227116e-05,
      "loss": 0.5653,
      "step": 6405
    },
    {
      "epoch": 0.8106808402935965,
      "grad_norm": 0.8705000877380371,
      "learning_rate": 1.426719567291225e-05,
      "loss": 0.5775,
      "step": 6406
    },
    {
      "epoch": 0.810807390534042,
      "grad_norm": 1.013471007347107,
      "learning_rate": 1.4265574836503518e-05,
      "loss": 0.6062,
      "step": 6407
    },
    {
      "epoch": 0.8109339407744874,
      "grad_norm": 0.8858911395072937,
      "learning_rate": 1.4263953863097023e-05,
      "loss": 0.5806,
      "step": 6408
    },
    {
      "epoch": 0.8110604910149329,
      "grad_norm": 1.0077632665634155,
      "learning_rate": 1.4262332752744822e-05,
      "loss": 0.6329,
      "step": 6409
    },
    {
      "epoch": 0.8111870412553784,
      "grad_norm": 0.9351118803024292,
      "learning_rate": 1.4260711505498989e-05,
      "loss": 0.6045,
      "step": 6410
    },
    {
      "epoch": 0.8113135914958238,
      "grad_norm": 0.8859950304031372,
      "learning_rate": 1.4259090121411585e-05,
      "loss": 0.6061,
      "step": 6411
    },
    {
      "epoch": 0.8114401417362693,
      "grad_norm": 0.9111003875732422,
      "learning_rate": 1.4257468600534692e-05,
      "loss": 0.6033,
      "step": 6412
    },
    {
      "epoch": 0.8115666919767147,
      "grad_norm": 0.907008171081543,
      "learning_rate": 1.425584694292038e-05,
      "loss": 0.6048,
      "step": 6413
    },
    {
      "epoch": 0.8116932422171602,
      "grad_norm": 0.9780369997024536,
      "learning_rate": 1.4254225148620733e-05,
      "loss": 0.595,
      "step": 6414
    },
    {
      "epoch": 0.8118197924576057,
      "grad_norm": 0.9151084423065186,
      "learning_rate": 1.4252603217687843e-05,
      "loss": 0.5869,
      "step": 6415
    },
    {
      "epoch": 0.8119463426980511,
      "grad_norm": 0.9488850831985474,
      "learning_rate": 1.4250981150173801e-05,
      "loss": 0.6272,
      "step": 6416
    },
    {
      "epoch": 0.8120728929384966,
      "grad_norm": 0.995065450668335,
      "learning_rate": 1.4249358946130703e-05,
      "loss": 0.5915,
      "step": 6417
    },
    {
      "epoch": 0.812199443178942,
      "grad_norm": 0.937324583530426,
      "learning_rate": 1.4247736605610647e-05,
      "loss": 0.6028,
      "step": 6418
    },
    {
      "epoch": 0.8123259934193875,
      "grad_norm": 1.006866693496704,
      "learning_rate": 1.424611412866574e-05,
      "loss": 0.6155,
      "step": 6419
    },
    {
      "epoch": 0.812452543659833,
      "grad_norm": 0.9842099547386169,
      "learning_rate": 1.4244491515348087e-05,
      "loss": 0.622,
      "step": 6420
    },
    {
      "epoch": 0.8125790939002784,
      "grad_norm": 0.9486444592475891,
      "learning_rate": 1.4242868765709808e-05,
      "loss": 0.6117,
      "step": 6421
    },
    {
      "epoch": 0.8127056441407239,
      "grad_norm": 1.017253041267395,
      "learning_rate": 1.4241245879803019e-05,
      "loss": 0.6459,
      "step": 6422
    },
    {
      "epoch": 0.8128321943811693,
      "grad_norm": 0.9450559020042419,
      "learning_rate": 1.4239622857679836e-05,
      "loss": 0.6234,
      "step": 6423
    },
    {
      "epoch": 0.8129587446216148,
      "grad_norm": 0.9005565643310547,
      "learning_rate": 1.4237999699392395e-05,
      "loss": 0.5884,
      "step": 6424
    },
    {
      "epoch": 0.8130852948620603,
      "grad_norm": 0.8711721897125244,
      "learning_rate": 1.4236376404992823e-05,
      "loss": 0.5911,
      "step": 6425
    },
    {
      "epoch": 0.8132118451025057,
      "grad_norm": 0.9904502034187317,
      "learning_rate": 1.4234752974533253e-05,
      "loss": 0.621,
      "step": 6426
    },
    {
      "epoch": 0.8133383953429512,
      "grad_norm": 0.8388975262641907,
      "learning_rate": 1.4233129408065829e-05,
      "loss": 0.6011,
      "step": 6427
    },
    {
      "epoch": 0.8134649455833967,
      "grad_norm": 0.9215306639671326,
      "learning_rate": 1.4231505705642692e-05,
      "loss": 0.6076,
      "step": 6428
    },
    {
      "epoch": 0.813591495823842,
      "grad_norm": 0.9335918426513672,
      "learning_rate": 1.4229881867315994e-05,
      "loss": 0.5897,
      "step": 6429
    },
    {
      "epoch": 0.8137180460642875,
      "grad_norm": 0.9080913662910461,
      "learning_rate": 1.4228257893137885e-05,
      "loss": 0.5884,
      "step": 6430
    },
    {
      "epoch": 0.8138445963047329,
      "grad_norm": 0.865234911441803,
      "learning_rate": 1.4226633783160521e-05,
      "loss": 0.5976,
      "step": 6431
    },
    {
      "epoch": 0.8139711465451784,
      "grad_norm": 0.9401648044586182,
      "learning_rate": 1.4225009537436067e-05,
      "loss": 0.5952,
      "step": 6432
    },
    {
      "epoch": 0.8140976967856239,
      "grad_norm": 0.8934062123298645,
      "learning_rate": 1.422338515601669e-05,
      "loss": 0.6162,
      "step": 6433
    },
    {
      "epoch": 0.8142242470260693,
      "grad_norm": 0.9562647938728333,
      "learning_rate": 1.4221760638954558e-05,
      "loss": 0.5908,
      "step": 6434
    },
    {
      "epoch": 0.8143507972665148,
      "grad_norm": 0.9224563837051392,
      "learning_rate": 1.4220135986301844e-05,
      "loss": 0.6291,
      "step": 6435
    },
    {
      "epoch": 0.8144773475069602,
      "grad_norm": 0.9142544269561768,
      "learning_rate": 1.4218511198110731e-05,
      "loss": 0.5896,
      "step": 6436
    },
    {
      "epoch": 0.8146038977474057,
      "grad_norm": 0.9223670363426208,
      "learning_rate": 1.42168862744334e-05,
      "loss": 0.5785,
      "step": 6437
    },
    {
      "epoch": 0.8147304479878512,
      "grad_norm": 0.9265197515487671,
      "learning_rate": 1.4215261215322037e-05,
      "loss": 0.5941,
      "step": 6438
    },
    {
      "epoch": 0.8148569982282966,
      "grad_norm": 0.8888291716575623,
      "learning_rate": 1.421363602082884e-05,
      "loss": 0.5903,
      "step": 6439
    },
    {
      "epoch": 0.8149835484687421,
      "grad_norm": 0.948273241519928,
      "learning_rate": 1.4212010691006e-05,
      "loss": 0.6128,
      "step": 6440
    },
    {
      "epoch": 0.8151100987091876,
      "grad_norm": 0.8962889909744263,
      "learning_rate": 1.4210385225905721e-05,
      "loss": 0.5784,
      "step": 6441
    },
    {
      "epoch": 0.815236648949633,
      "grad_norm": 0.8624507784843445,
      "learning_rate": 1.4208759625580205e-05,
      "loss": 0.5652,
      "step": 6442
    },
    {
      "epoch": 0.8153631991900785,
      "grad_norm": 0.888399600982666,
      "learning_rate": 1.4207133890081664e-05,
      "loss": 0.6042,
      "step": 6443
    },
    {
      "epoch": 0.8154897494305239,
      "grad_norm": 0.8926485180854797,
      "learning_rate": 1.420550801946231e-05,
      "loss": 0.5888,
      "step": 6444
    },
    {
      "epoch": 0.8156162996709694,
      "grad_norm": 0.8995068669319153,
      "learning_rate": 1.4203882013774364e-05,
      "loss": 0.5861,
      "step": 6445
    },
    {
      "epoch": 0.8157428499114149,
      "grad_norm": 0.8781632781028748,
      "learning_rate": 1.4202255873070047e-05,
      "loss": 0.5715,
      "step": 6446
    },
    {
      "epoch": 0.8158694001518603,
      "grad_norm": 0.93213951587677,
      "learning_rate": 1.4200629597401587e-05,
      "loss": 0.6048,
      "step": 6447
    },
    {
      "epoch": 0.8159959503923058,
      "grad_norm": 0.874436616897583,
      "learning_rate": 1.4199003186821215e-05,
      "loss": 0.5909,
      "step": 6448
    },
    {
      "epoch": 0.8161225006327512,
      "grad_norm": 0.9158216118812561,
      "learning_rate": 1.4197376641381162e-05,
      "loss": 0.6372,
      "step": 6449
    },
    {
      "epoch": 0.8162490508731967,
      "grad_norm": 0.9008409976959229,
      "learning_rate": 1.4195749961133673e-05,
      "loss": 0.6084,
      "step": 6450
    },
    {
      "epoch": 0.8163756011136422,
      "grad_norm": 0.9118524193763733,
      "learning_rate": 1.4194123146130991e-05,
      "loss": 0.596,
      "step": 6451
    },
    {
      "epoch": 0.8165021513540875,
      "grad_norm": 0.8870657086372375,
      "learning_rate": 1.4192496196425367e-05,
      "loss": 0.594,
      "step": 6452
    },
    {
      "epoch": 0.816628701594533,
      "grad_norm": 0.8611319661140442,
      "learning_rate": 1.4190869112069047e-05,
      "loss": 0.5869,
      "step": 6453
    },
    {
      "epoch": 0.8167552518349784,
      "grad_norm": 0.9008516669273376,
      "learning_rate": 1.4189241893114294e-05,
      "loss": 0.5779,
      "step": 6454
    },
    {
      "epoch": 0.8168818020754239,
      "grad_norm": 1.0328487157821655,
      "learning_rate": 1.4187614539613365e-05,
      "loss": 0.6299,
      "step": 6455
    },
    {
      "epoch": 0.8170083523158694,
      "grad_norm": 0.8874877691268921,
      "learning_rate": 1.4185987051618532e-05,
      "loss": 0.6251,
      "step": 6456
    },
    {
      "epoch": 0.8171349025563148,
      "grad_norm": 1.0362622737884521,
      "learning_rate": 1.4184359429182059e-05,
      "loss": 0.6146,
      "step": 6457
    },
    {
      "epoch": 0.8172614527967603,
      "grad_norm": 1.000121831893921,
      "learning_rate": 1.418273167235623e-05,
      "loss": 0.5521,
      "step": 6458
    },
    {
      "epoch": 0.8173880030372058,
      "grad_norm": 0.9252048134803772,
      "learning_rate": 1.418110378119331e-05,
      "loss": 0.5819,
      "step": 6459
    },
    {
      "epoch": 0.8175145532776512,
      "grad_norm": 1.0834931135177612,
      "learning_rate": 1.4179475755745593e-05,
      "loss": 0.5892,
      "step": 6460
    },
    {
      "epoch": 0.8176411035180967,
      "grad_norm": 0.9172561764717102,
      "learning_rate": 1.4177847596065359e-05,
      "loss": 0.6146,
      "step": 6461
    },
    {
      "epoch": 0.8177676537585421,
      "grad_norm": 0.9491912126541138,
      "learning_rate": 1.4176219302204905e-05,
      "loss": 0.5713,
      "step": 6462
    },
    {
      "epoch": 0.8178942039989876,
      "grad_norm": 0.9565256237983704,
      "learning_rate": 1.4174590874216526e-05,
      "loss": 0.6435,
      "step": 6463
    },
    {
      "epoch": 0.8180207542394331,
      "grad_norm": 0.9339431524276733,
      "learning_rate": 1.4172962312152521e-05,
      "loss": 0.6175,
      "step": 6464
    },
    {
      "epoch": 0.8181473044798785,
      "grad_norm": 0.963414192199707,
      "learning_rate": 1.4171333616065195e-05,
      "loss": 0.6112,
      "step": 6465
    },
    {
      "epoch": 0.818273854720324,
      "grad_norm": 0.870279848575592,
      "learning_rate": 1.4169704786006858e-05,
      "loss": 0.5867,
      "step": 6466
    },
    {
      "epoch": 0.8184004049607694,
      "grad_norm": 1.0005258321762085,
      "learning_rate": 1.416807582202982e-05,
      "loss": 0.6507,
      "step": 6467
    },
    {
      "epoch": 0.8185269552012149,
      "grad_norm": 0.932662308216095,
      "learning_rate": 1.4166446724186402e-05,
      "loss": 0.6006,
      "step": 6468
    },
    {
      "epoch": 0.8186535054416604,
      "grad_norm": 0.9404528141021729,
      "learning_rate": 1.4164817492528924e-05,
      "loss": 0.5671,
      "step": 6469
    },
    {
      "epoch": 0.8187800556821058,
      "grad_norm": 0.9521636962890625,
      "learning_rate": 1.4163188127109717e-05,
      "loss": 0.5903,
      "step": 6470
    },
    {
      "epoch": 0.8189066059225513,
      "grad_norm": 0.9380059242248535,
      "learning_rate": 1.4161558627981105e-05,
      "loss": 0.5968,
      "step": 6471
    },
    {
      "epoch": 0.8190331561629967,
      "grad_norm": 1.0389957427978516,
      "learning_rate": 1.4159928995195427e-05,
      "loss": 0.6255,
      "step": 6472
    },
    {
      "epoch": 0.8191597064034422,
      "grad_norm": 0.875787615776062,
      "learning_rate": 1.4158299228805017e-05,
      "loss": 0.5675,
      "step": 6473
    },
    {
      "epoch": 0.8192862566438877,
      "grad_norm": 0.9468932151794434,
      "learning_rate": 1.4156669328862222e-05,
      "loss": 0.6211,
      "step": 6474
    },
    {
      "epoch": 0.819412806884333,
      "grad_norm": 0.9698768258094788,
      "learning_rate": 1.4155039295419392e-05,
      "loss": 0.6027,
      "step": 6475
    },
    {
      "epoch": 0.8195393571247785,
      "grad_norm": 0.8679444789886475,
      "learning_rate": 1.4153409128528873e-05,
      "loss": 0.6032,
      "step": 6476
    },
    {
      "epoch": 0.819665907365224,
      "grad_norm": 0.9615499377250671,
      "learning_rate": 1.4151778828243026e-05,
      "loss": 0.595,
      "step": 6477
    },
    {
      "epoch": 0.8197924576056694,
      "grad_norm": 0.9280304908752441,
      "learning_rate": 1.4150148394614207e-05,
      "loss": 0.6269,
      "step": 6478
    },
    {
      "epoch": 0.8199190078461149,
      "grad_norm": 0.9884299039840698,
      "learning_rate": 1.4148517827694784e-05,
      "loss": 0.6423,
      "step": 6479
    },
    {
      "epoch": 0.8200455580865603,
      "grad_norm": 0.940523087978363,
      "learning_rate": 1.4146887127537125e-05,
      "loss": 0.6193,
      "step": 6480
    },
    {
      "epoch": 0.8201721083270058,
      "grad_norm": 0.8788697123527527,
      "learning_rate": 1.4145256294193607e-05,
      "loss": 0.6183,
      "step": 6481
    },
    {
      "epoch": 0.8202986585674513,
      "grad_norm": 0.8964776992797852,
      "learning_rate": 1.4143625327716603e-05,
      "loss": 0.6305,
      "step": 6482
    },
    {
      "epoch": 0.8204252088078967,
      "grad_norm": 0.9146429300308228,
      "learning_rate": 1.4141994228158495e-05,
      "loss": 0.6291,
      "step": 6483
    },
    {
      "epoch": 0.8205517590483422,
      "grad_norm": 0.9559044241905212,
      "learning_rate": 1.4140362995571674e-05,
      "loss": 0.5865,
      "step": 6484
    },
    {
      "epoch": 0.8206783092887876,
      "grad_norm": 1.0748506784439087,
      "learning_rate": 1.4138731630008524e-05,
      "loss": 0.6156,
      "step": 6485
    },
    {
      "epoch": 0.8208048595292331,
      "grad_norm": 0.8939298987388611,
      "learning_rate": 1.4137100131521442e-05,
      "loss": 0.5882,
      "step": 6486
    },
    {
      "epoch": 0.8209314097696786,
      "grad_norm": 1.1453793048858643,
      "learning_rate": 1.413546850016283e-05,
      "loss": 0.5866,
      "step": 6487
    },
    {
      "epoch": 0.821057960010124,
      "grad_norm": 0.8544284105300903,
      "learning_rate": 1.4133836735985085e-05,
      "loss": 0.5823,
      "step": 6488
    },
    {
      "epoch": 0.8211845102505695,
      "grad_norm": 0.9207281470298767,
      "learning_rate": 1.4132204839040618e-05,
      "loss": 0.6166,
      "step": 6489
    },
    {
      "epoch": 0.821311060491015,
      "grad_norm": 0.9532732963562012,
      "learning_rate": 1.4130572809381841e-05,
      "loss": 0.5796,
      "step": 6490
    },
    {
      "epoch": 0.8214376107314604,
      "grad_norm": 0.8768398761749268,
      "learning_rate": 1.4128940647061171e-05,
      "loss": 0.581,
      "step": 6491
    },
    {
      "epoch": 0.8215641609719059,
      "grad_norm": 0.9231427311897278,
      "learning_rate": 1.4127308352131028e-05,
      "loss": 0.5757,
      "step": 6492
    },
    {
      "epoch": 0.8216907112123513,
      "grad_norm": 0.9452505111694336,
      "learning_rate": 1.4125675924643836e-05,
      "loss": 0.5454,
      "step": 6493
    },
    {
      "epoch": 0.8218172614527968,
      "grad_norm": 0.9169554114341736,
      "learning_rate": 1.4124043364652022e-05,
      "loss": 0.613,
      "step": 6494
    },
    {
      "epoch": 0.8219438116932423,
      "grad_norm": 0.9397392868995667,
      "learning_rate": 1.4122410672208023e-05,
      "loss": 0.5858,
      "step": 6495
    },
    {
      "epoch": 0.8220703619336877,
      "grad_norm": 0.9969725608825684,
      "learning_rate": 1.412077784736427e-05,
      "loss": 0.6253,
      "step": 6496
    },
    {
      "epoch": 0.8221969121741332,
      "grad_norm": 0.9251161813735962,
      "learning_rate": 1.4119144890173212e-05,
      "loss": 0.5844,
      "step": 6497
    },
    {
      "epoch": 0.8223234624145785,
      "grad_norm": 0.8685432076454163,
      "learning_rate": 1.4117511800687289e-05,
      "loss": 0.6249,
      "step": 6498
    },
    {
      "epoch": 0.822450012655024,
      "grad_norm": 0.9376678466796875,
      "learning_rate": 1.4115878578958952e-05,
      "loss": 0.5616,
      "step": 6499
    },
    {
      "epoch": 0.8225765628954695,
      "grad_norm": 0.8716501593589783,
      "learning_rate": 1.4114245225040656e-05,
      "loss": 0.5651,
      "step": 6500
    },
    {
      "epoch": 0.8227031131359149,
      "grad_norm": 0.9369723796844482,
      "learning_rate": 1.4112611738984865e-05,
      "loss": 0.6174,
      "step": 6501
    },
    {
      "epoch": 0.8228296633763604,
      "grad_norm": 0.8807774186134338,
      "learning_rate": 1.4110978120844031e-05,
      "loss": 0.6212,
      "step": 6502
    },
    {
      "epoch": 0.8229562136168058,
      "grad_norm": 0.9283157587051392,
      "learning_rate": 1.410934437067063e-05,
      "loss": 0.619,
      "step": 6503
    },
    {
      "epoch": 0.8230827638572513,
      "grad_norm": 0.995728075504303,
      "learning_rate": 1.410771048851713e-05,
      "loss": 0.5668,
      "step": 6504
    },
    {
      "epoch": 0.8232093140976968,
      "grad_norm": 1.0210673809051514,
      "learning_rate": 1.4106076474436006e-05,
      "loss": 0.5838,
      "step": 6505
    },
    {
      "epoch": 0.8233358643381422,
      "grad_norm": 0.9775877594947815,
      "learning_rate": 1.4104442328479738e-05,
      "loss": 0.6081,
      "step": 6506
    },
    {
      "epoch": 0.8234624145785877,
      "grad_norm": 0.9052392840385437,
      "learning_rate": 1.410280805070081e-05,
      "loss": 0.5638,
      "step": 6507
    },
    {
      "epoch": 0.8235889648190332,
      "grad_norm": 1.0898590087890625,
      "learning_rate": 1.4101173641151712e-05,
      "loss": 0.6133,
      "step": 6508
    },
    {
      "epoch": 0.8237155150594786,
      "grad_norm": 0.9187296032905579,
      "learning_rate": 1.4099539099884933e-05,
      "loss": 0.5851,
      "step": 6509
    },
    {
      "epoch": 0.8238420652999241,
      "grad_norm": 0.9866913557052612,
      "learning_rate": 1.4097904426952975e-05,
      "loss": 0.6008,
      "step": 6510
    },
    {
      "epoch": 0.8239686155403695,
      "grad_norm": 0.9709261655807495,
      "learning_rate": 1.4096269622408334e-05,
      "loss": 0.6596,
      "step": 6511
    },
    {
      "epoch": 0.824095165780815,
      "grad_norm": 0.9196149110794067,
      "learning_rate": 1.4094634686303516e-05,
      "loss": 0.5827,
      "step": 6512
    },
    {
      "epoch": 0.8242217160212605,
      "grad_norm": 0.9282026886940002,
      "learning_rate": 1.409299961869103e-05,
      "loss": 0.654,
      "step": 6513
    },
    {
      "epoch": 0.8243482662617059,
      "grad_norm": 0.8490349650382996,
      "learning_rate": 1.409136441962339e-05,
      "loss": 0.5269,
      "step": 6514
    },
    {
      "epoch": 0.8244748165021514,
      "grad_norm": 0.8940841555595398,
      "learning_rate": 1.4089729089153117e-05,
      "loss": 0.5423,
      "step": 6515
    },
    {
      "epoch": 0.8246013667425968,
      "grad_norm": 0.9247065186500549,
      "learning_rate": 1.408809362733273e-05,
      "loss": 0.6169,
      "step": 6516
    },
    {
      "epoch": 0.8247279169830423,
      "grad_norm": 1.0178979635238647,
      "learning_rate": 1.4086458034214756e-05,
      "loss": 0.641,
      "step": 6517
    },
    {
      "epoch": 0.8248544672234878,
      "grad_norm": 0.89583420753479,
      "learning_rate": 1.4084822309851724e-05,
      "loss": 0.5427,
      "step": 6518
    },
    {
      "epoch": 0.8249810174639332,
      "grad_norm": 0.8285821080207825,
      "learning_rate": 1.408318645429617e-05,
      "loss": 0.5114,
      "step": 6519
    },
    {
      "epoch": 0.8251075677043787,
      "grad_norm": 0.8868833780288696,
      "learning_rate": 1.4081550467600633e-05,
      "loss": 0.5746,
      "step": 6520
    },
    {
      "epoch": 0.825234117944824,
      "grad_norm": 0.9151160717010498,
      "learning_rate": 1.4079914349817655e-05,
      "loss": 0.5835,
      "step": 6521
    },
    {
      "epoch": 0.8253606681852695,
      "grad_norm": 0.9413664937019348,
      "learning_rate": 1.4078278100999789e-05,
      "loss": 0.6135,
      "step": 6522
    },
    {
      "epoch": 0.825487218425715,
      "grad_norm": 0.9340777397155762,
      "learning_rate": 1.4076641721199575e-05,
      "loss": 0.6382,
      "step": 6523
    },
    {
      "epoch": 0.8256137686661604,
      "grad_norm": 0.937332808971405,
      "learning_rate": 1.4075005210469578e-05,
      "loss": 0.6532,
      "step": 6524
    },
    {
      "epoch": 0.8257403189066059,
      "grad_norm": 0.9027602076530457,
      "learning_rate": 1.4073368568862356e-05,
      "loss": 0.6101,
      "step": 6525
    },
    {
      "epoch": 0.8258668691470514,
      "grad_norm": 0.9086715579032898,
      "learning_rate": 1.407173179643047e-05,
      "loss": 0.5728,
      "step": 6526
    },
    {
      "epoch": 0.8259934193874968,
      "grad_norm": 0.9102026224136353,
      "learning_rate": 1.4070094893226493e-05,
      "loss": 0.6048,
      "step": 6527
    },
    {
      "epoch": 0.8261199696279423,
      "grad_norm": 0.9421181082725525,
      "learning_rate": 1.4068457859302995e-05,
      "loss": 0.6119,
      "step": 6528
    },
    {
      "epoch": 0.8262465198683877,
      "grad_norm": 0.952704131603241,
      "learning_rate": 1.4066820694712553e-05,
      "loss": 0.5907,
      "step": 6529
    },
    {
      "epoch": 0.8263730701088332,
      "grad_norm": 0.930475115776062,
      "learning_rate": 1.4065183399507747e-05,
      "loss": 0.6376,
      "step": 6530
    },
    {
      "epoch": 0.8264996203492787,
      "grad_norm": 0.9588854312896729,
      "learning_rate": 1.4063545973741165e-05,
      "loss": 0.6028,
      "step": 6531
    },
    {
      "epoch": 0.8266261705897241,
      "grad_norm": 1.0159461498260498,
      "learning_rate": 1.406190841746539e-05,
      "loss": 0.6126,
      "step": 6532
    },
    {
      "epoch": 0.8267527208301696,
      "grad_norm": 0.8490450382232666,
      "learning_rate": 1.4060270730733026e-05,
      "loss": 0.5521,
      "step": 6533
    },
    {
      "epoch": 0.826879271070615,
      "grad_norm": 0.9232289791107178,
      "learning_rate": 1.405863291359666e-05,
      "loss": 0.6481,
      "step": 6534
    },
    {
      "epoch": 0.8270058213110605,
      "grad_norm": 0.8720576167106628,
      "learning_rate": 1.4056994966108901e-05,
      "loss": 0.5938,
      "step": 6535
    },
    {
      "epoch": 0.827132371551506,
      "grad_norm": 0.9398362636566162,
      "learning_rate": 1.4055356888322352e-05,
      "loss": 0.5965,
      "step": 6536
    },
    {
      "epoch": 0.8272589217919514,
      "grad_norm": 0.9207731485366821,
      "learning_rate": 1.4053718680289622e-05,
      "loss": 0.5858,
      "step": 6537
    },
    {
      "epoch": 0.8273854720323969,
      "grad_norm": 0.8753196597099304,
      "learning_rate": 1.4052080342063328e-05,
      "loss": 0.5717,
      "step": 6538
    },
    {
      "epoch": 0.8275120222728424,
      "grad_norm": 0.9273605942726135,
      "learning_rate": 1.405044187369609e-05,
      "loss": 0.6006,
      "step": 6539
    },
    {
      "epoch": 0.8276385725132878,
      "grad_norm": 0.9414604306221008,
      "learning_rate": 1.4048803275240527e-05,
      "loss": 0.6167,
      "step": 6540
    },
    {
      "epoch": 0.8277651227537333,
      "grad_norm": 0.9019355177879333,
      "learning_rate": 1.4047164546749267e-05,
      "loss": 0.5754,
      "step": 6541
    },
    {
      "epoch": 0.8278916729941787,
      "grad_norm": 0.9055572748184204,
      "learning_rate": 1.4045525688274944e-05,
      "loss": 0.5827,
      "step": 6542
    },
    {
      "epoch": 0.8280182232346242,
      "grad_norm": 0.9065368175506592,
      "learning_rate": 1.4043886699870188e-05,
      "loss": 0.5952,
      "step": 6543
    },
    {
      "epoch": 0.8281447734750697,
      "grad_norm": 0.9130824208259583,
      "learning_rate": 1.4042247581587644e-05,
      "loss": 0.5841,
      "step": 6544
    },
    {
      "epoch": 0.828271323715515,
      "grad_norm": 0.9248535633087158,
      "learning_rate": 1.4040608333479952e-05,
      "loss": 0.5736,
      "step": 6545
    },
    {
      "epoch": 0.8283978739559605,
      "grad_norm": 1.0054833889007568,
      "learning_rate": 1.4038968955599765e-05,
      "loss": 0.6044,
      "step": 6546
    },
    {
      "epoch": 0.8285244241964059,
      "grad_norm": 0.9023453593254089,
      "learning_rate": 1.4037329447999728e-05,
      "loss": 0.6151,
      "step": 6547
    },
    {
      "epoch": 0.8286509744368514,
      "grad_norm": 0.958000659942627,
      "learning_rate": 1.40356898107325e-05,
      "loss": 0.6182,
      "step": 6548
    },
    {
      "epoch": 0.8287775246772969,
      "grad_norm": 0.9645642638206482,
      "learning_rate": 1.4034050043850741e-05,
      "loss": 0.613,
      "step": 6549
    },
    {
      "epoch": 0.8289040749177423,
      "grad_norm": 0.9120882749557495,
      "learning_rate": 1.4032410147407118e-05,
      "loss": 0.6459,
      "step": 6550
    },
    {
      "epoch": 0.8290306251581878,
      "grad_norm": 0.9407641291618347,
      "learning_rate": 1.4030770121454298e-05,
      "loss": 0.5731,
      "step": 6551
    },
    {
      "epoch": 0.8291571753986332,
      "grad_norm": 0.8991252183914185,
      "learning_rate": 1.4029129966044953e-05,
      "loss": 0.5874,
      "step": 6552
    },
    {
      "epoch": 0.8292837256390787,
      "grad_norm": 0.903893768787384,
      "learning_rate": 1.4027489681231761e-05,
      "loss": 0.588,
      "step": 6553
    },
    {
      "epoch": 0.8294102758795242,
      "grad_norm": 0.8757995367050171,
      "learning_rate": 1.4025849267067402e-05,
      "loss": 0.5806,
      "step": 6554
    },
    {
      "epoch": 0.8295368261199696,
      "grad_norm": 0.944688618183136,
      "learning_rate": 1.4024208723604562e-05,
      "loss": 0.6289,
      "step": 6555
    },
    {
      "epoch": 0.8296633763604151,
      "grad_norm": 0.8632702231407166,
      "learning_rate": 1.4022568050895932e-05,
      "loss": 0.5881,
      "step": 6556
    },
    {
      "epoch": 0.8297899266008606,
      "grad_norm": 0.9123443961143494,
      "learning_rate": 1.4020927248994202e-05,
      "loss": 0.6012,
      "step": 6557
    },
    {
      "epoch": 0.829916476841306,
      "grad_norm": 0.8977423310279846,
      "learning_rate": 1.4019286317952077e-05,
      "loss": 0.5965,
      "step": 6558
    },
    {
      "epoch": 0.8300430270817515,
      "grad_norm": 0.9769346117973328,
      "learning_rate": 1.401764525782225e-05,
      "loss": 0.5871,
      "step": 6559
    },
    {
      "epoch": 0.8301695773221969,
      "grad_norm": 0.9189664721488953,
      "learning_rate": 1.401600406865743e-05,
      "loss": 0.6038,
      "step": 6560
    },
    {
      "epoch": 0.8302961275626424,
      "grad_norm": 0.9289467930793762,
      "learning_rate": 1.4014362750510328e-05,
      "loss": 0.601,
      "step": 6561
    },
    {
      "epoch": 0.8304226778030879,
      "grad_norm": 0.8412683606147766,
      "learning_rate": 1.4012721303433661e-05,
      "loss": 0.5758,
      "step": 6562
    },
    {
      "epoch": 0.8305492280435333,
      "grad_norm": 0.9764110445976257,
      "learning_rate": 1.4011079727480143e-05,
      "loss": 0.5464,
      "step": 6563
    },
    {
      "epoch": 0.8306757782839788,
      "grad_norm": 0.8863469958305359,
      "learning_rate": 1.40094380227025e-05,
      "loss": 0.5996,
      "step": 6564
    },
    {
      "epoch": 0.8308023285244241,
      "grad_norm": 0.9061722755432129,
      "learning_rate": 1.4007796189153452e-05,
      "loss": 0.5928,
      "step": 6565
    },
    {
      "epoch": 0.8309288787648696,
      "grad_norm": 0.9366016983985901,
      "learning_rate": 1.4006154226885743e-05,
      "loss": 0.6041,
      "step": 6566
    },
    {
      "epoch": 0.8310554290053151,
      "grad_norm": 1.014243483543396,
      "learning_rate": 1.4004512135952096e-05,
      "loss": 0.6113,
      "step": 6567
    },
    {
      "epoch": 0.8311819792457605,
      "grad_norm": 0.9108131527900696,
      "learning_rate": 1.4002869916405254e-05,
      "loss": 0.6049,
      "step": 6568
    },
    {
      "epoch": 0.831308529486206,
      "grad_norm": 0.9068788886070251,
      "learning_rate": 1.4001227568297962e-05,
      "loss": 0.5688,
      "step": 6569
    },
    {
      "epoch": 0.8314350797266514,
      "grad_norm": 0.9378610849380493,
      "learning_rate": 1.3999585091682966e-05,
      "loss": 0.6204,
      "step": 6570
    },
    {
      "epoch": 0.8315616299670969,
      "grad_norm": 1.0140725374221802,
      "learning_rate": 1.3997942486613013e-05,
      "loss": 0.6044,
      "step": 6571
    },
    {
      "epoch": 0.8316881802075424,
      "grad_norm": 0.9073925614356995,
      "learning_rate": 1.3996299753140869e-05,
      "loss": 0.5683,
      "step": 6572
    },
    {
      "epoch": 0.8318147304479878,
      "grad_norm": 0.9852568507194519,
      "learning_rate": 1.3994656891319287e-05,
      "loss": 0.5917,
      "step": 6573
    },
    {
      "epoch": 0.8319412806884333,
      "grad_norm": 0.935731053352356,
      "learning_rate": 1.399301390120103e-05,
      "loss": 0.5923,
      "step": 6574
    },
    {
      "epoch": 0.8320678309288788,
      "grad_norm": 0.9881905913352966,
      "learning_rate": 1.3991370782838867e-05,
      "loss": 0.6004,
      "step": 6575
    },
    {
      "epoch": 0.8321943811693242,
      "grad_norm": 0.9966711401939392,
      "learning_rate": 1.3989727536285571e-05,
      "loss": 0.5866,
      "step": 6576
    },
    {
      "epoch": 0.8323209314097697,
      "grad_norm": 1.0151251554489136,
      "learning_rate": 1.398808416159392e-05,
      "loss": 0.6317,
      "step": 6577
    },
    {
      "epoch": 0.8324474816502151,
      "grad_norm": 0.9266090989112854,
      "learning_rate": 1.3986440658816692e-05,
      "loss": 0.5603,
      "step": 6578
    },
    {
      "epoch": 0.8325740318906606,
      "grad_norm": 0.9630829095840454,
      "learning_rate": 1.3984797028006674e-05,
      "loss": 0.5681,
      "step": 6579
    },
    {
      "epoch": 0.8327005821311061,
      "grad_norm": 0.8727161884307861,
      "learning_rate": 1.3983153269216652e-05,
      "loss": 0.5559,
      "step": 6580
    },
    {
      "epoch": 0.8328271323715515,
      "grad_norm": 1.0061904191970825,
      "learning_rate": 1.3981509382499416e-05,
      "loss": 0.6225,
      "step": 6581
    },
    {
      "epoch": 0.832953682611997,
      "grad_norm": 1.045216679573059,
      "learning_rate": 1.397986536790777e-05,
      "loss": 0.6203,
      "step": 6582
    },
    {
      "epoch": 0.8330802328524424,
      "grad_norm": 0.8592104315757751,
      "learning_rate": 1.3978221225494513e-05,
      "loss": 0.5934,
      "step": 6583
    },
    {
      "epoch": 0.8332067830928879,
      "grad_norm": 1.0400985479354858,
      "learning_rate": 1.3976576955312444e-05,
      "loss": 0.632,
      "step": 6584
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.9176501035690308,
      "learning_rate": 1.3974932557414378e-05,
      "loss": 0.5734,
      "step": 6585
    },
    {
      "epoch": 0.8334598835737788,
      "grad_norm": 0.9213414192199707,
      "learning_rate": 1.3973288031853127e-05,
      "loss": 0.6051,
      "step": 6586
    },
    {
      "epoch": 0.8335864338142243,
      "grad_norm": 1.0577563047409058,
      "learning_rate": 1.3971643378681505e-05,
      "loss": 0.5958,
      "step": 6587
    },
    {
      "epoch": 0.8337129840546698,
      "grad_norm": 0.9799981713294983,
      "learning_rate": 1.396999859795234e-05,
      "loss": 0.5704,
      "step": 6588
    },
    {
      "epoch": 0.8338395342951151,
      "grad_norm": 0.9956713914871216,
      "learning_rate": 1.3968353689718452e-05,
      "loss": 0.5904,
      "step": 6589
    },
    {
      "epoch": 0.8339660845355606,
      "grad_norm": 0.9873670935630798,
      "learning_rate": 1.3966708654032675e-05,
      "loss": 0.5406,
      "step": 6590
    },
    {
      "epoch": 0.834092634776006,
      "grad_norm": 0.8917446732521057,
      "learning_rate": 1.3965063490947838e-05,
      "loss": 0.5824,
      "step": 6591
    },
    {
      "epoch": 0.8342191850164515,
      "grad_norm": 0.987467348575592,
      "learning_rate": 1.3963418200516783e-05,
      "loss": 0.6115,
      "step": 6592
    },
    {
      "epoch": 0.834345735256897,
      "grad_norm": 0.8841798305511475,
      "learning_rate": 1.3961772782792346e-05,
      "loss": 0.599,
      "step": 6593
    },
    {
      "epoch": 0.8344722854973424,
      "grad_norm": 1.0179975032806396,
      "learning_rate": 1.3960127237827381e-05,
      "loss": 0.5965,
      "step": 6594
    },
    {
      "epoch": 0.8345988357377879,
      "grad_norm": 0.8760759830474854,
      "learning_rate": 1.3958481565674733e-05,
      "loss": 0.6323,
      "step": 6595
    },
    {
      "epoch": 0.8347253859782333,
      "grad_norm": 0.9018712043762207,
      "learning_rate": 1.395683576638726e-05,
      "loss": 0.6085,
      "step": 6596
    },
    {
      "epoch": 0.8348519362186788,
      "grad_norm": 1.0670146942138672,
      "learning_rate": 1.3955189840017815e-05,
      "loss": 0.6226,
      "step": 6597
    },
    {
      "epoch": 0.8349784864591243,
      "grad_norm": 0.8869194984436035,
      "learning_rate": 1.3953543786619261e-05,
      "loss": 0.5937,
      "step": 6598
    },
    {
      "epoch": 0.8351050366995697,
      "grad_norm": 1.0860238075256348,
      "learning_rate": 1.3951897606244468e-05,
      "loss": 0.5766,
      "step": 6599
    },
    {
      "epoch": 0.8352315869400152,
      "grad_norm": 0.9979828596115112,
      "learning_rate": 1.3950251298946305e-05,
      "loss": 0.5857,
      "step": 6600
    },
    {
      "epoch": 0.8353581371804606,
      "grad_norm": 0.9298877716064453,
      "learning_rate": 1.3948604864777647e-05,
      "loss": 0.5755,
      "step": 6601
    },
    {
      "epoch": 0.8354846874209061,
      "grad_norm": 0.9724204540252686,
      "learning_rate": 1.394695830379137e-05,
      "loss": 0.5702,
      "step": 6602
    },
    {
      "epoch": 0.8356112376613516,
      "grad_norm": 0.9046828150749207,
      "learning_rate": 1.394531161604036e-05,
      "loss": 0.581,
      "step": 6603
    },
    {
      "epoch": 0.835737787901797,
      "grad_norm": 0.9041823148727417,
      "learning_rate": 1.3943664801577503e-05,
      "loss": 0.6019,
      "step": 6604
    },
    {
      "epoch": 0.8358643381422425,
      "grad_norm": 0.9012951850891113,
      "learning_rate": 1.3942017860455688e-05,
      "loss": 0.5927,
      "step": 6605
    },
    {
      "epoch": 0.835990888382688,
      "grad_norm": 0.9492535591125488,
      "learning_rate": 1.394037079272781e-05,
      "loss": 0.6621,
      "step": 6606
    },
    {
      "epoch": 0.8361174386231334,
      "grad_norm": 0.866436779499054,
      "learning_rate": 1.3938723598446772e-05,
      "loss": 0.5635,
      "step": 6607
    },
    {
      "epoch": 0.8362439888635789,
      "grad_norm": 0.9052554368972778,
      "learning_rate": 1.3937076277665478e-05,
      "loss": 0.5756,
      "step": 6608
    },
    {
      "epoch": 0.8363705391040243,
      "grad_norm": 0.8832292556762695,
      "learning_rate": 1.3935428830436823e-05,
      "loss": 0.5927,
      "step": 6609
    },
    {
      "epoch": 0.8364970893444698,
      "grad_norm": 1.014780044555664,
      "learning_rate": 1.3933781256813732e-05,
      "loss": 0.6534,
      "step": 6610
    },
    {
      "epoch": 0.8366236395849153,
      "grad_norm": 0.924275279045105,
      "learning_rate": 1.3932133556849113e-05,
      "loss": 0.6252,
      "step": 6611
    },
    {
      "epoch": 0.8367501898253606,
      "grad_norm": 0.9812994599342346,
      "learning_rate": 1.3930485730595886e-05,
      "loss": 0.6129,
      "step": 6612
    },
    {
      "epoch": 0.8368767400658061,
      "grad_norm": 0.9713335633277893,
      "learning_rate": 1.3928837778106979e-05,
      "loss": 0.5521,
      "step": 6613
    },
    {
      "epoch": 0.8370032903062515,
      "grad_norm": 0.8653222918510437,
      "learning_rate": 1.3927189699435313e-05,
      "loss": 0.5749,
      "step": 6614
    },
    {
      "epoch": 0.837129840546697,
      "grad_norm": 0.9625530242919922,
      "learning_rate": 1.3925541494633824e-05,
      "loss": 0.602,
      "step": 6615
    },
    {
      "epoch": 0.8372563907871425,
      "grad_norm": 0.9331874251365662,
      "learning_rate": 1.3923893163755441e-05,
      "loss": 0.6214,
      "step": 6616
    },
    {
      "epoch": 0.8373829410275879,
      "grad_norm": 0.9515397548675537,
      "learning_rate": 1.3922244706853113e-05,
      "loss": 0.6273,
      "step": 6617
    },
    {
      "epoch": 0.8375094912680334,
      "grad_norm": 0.9141596555709839,
      "learning_rate": 1.392059612397978e-05,
      "loss": 0.5942,
      "step": 6618
    },
    {
      "epoch": 0.8376360415084788,
      "grad_norm": 0.8657425045967102,
      "learning_rate": 1.3918947415188386e-05,
      "loss": 0.6092,
      "step": 6619
    },
    {
      "epoch": 0.8377625917489243,
      "grad_norm": 0.8978883028030396,
      "learning_rate": 1.3917298580531887e-05,
      "loss": 0.5998,
      "step": 6620
    },
    {
      "epoch": 0.8378891419893698,
      "grad_norm": 0.8875873684883118,
      "learning_rate": 1.3915649620063239e-05,
      "loss": 0.6237,
      "step": 6621
    },
    {
      "epoch": 0.8380156922298152,
      "grad_norm": 0.903781533241272,
      "learning_rate": 1.3914000533835394e-05,
      "loss": 0.6149,
      "step": 6622
    },
    {
      "epoch": 0.8381422424702607,
      "grad_norm": 0.8990657925605774,
      "learning_rate": 1.3912351321901324e-05,
      "loss": 0.5682,
      "step": 6623
    },
    {
      "epoch": 0.8382687927107062,
      "grad_norm": 0.8951390385627747,
      "learning_rate": 1.3910701984313996e-05,
      "loss": 0.5912,
      "step": 6624
    },
    {
      "epoch": 0.8383953429511516,
      "grad_norm": 0.8940503001213074,
      "learning_rate": 1.3909052521126381e-05,
      "loss": 0.5788,
      "step": 6625
    },
    {
      "epoch": 0.8385218931915971,
      "grad_norm": 1.0894330739974976,
      "learning_rate": 1.3907402932391455e-05,
      "loss": 0.6179,
      "step": 6626
    },
    {
      "epoch": 0.8386484434320425,
      "grad_norm": 0.9243391752243042,
      "learning_rate": 1.3905753218162192e-05,
      "loss": 0.5736,
      "step": 6627
    },
    {
      "epoch": 0.838774993672488,
      "grad_norm": 0.8849332928657532,
      "learning_rate": 1.3904103378491586e-05,
      "loss": 0.5723,
      "step": 6628
    },
    {
      "epoch": 0.8389015439129335,
      "grad_norm": 0.9170801043510437,
      "learning_rate": 1.390245341343262e-05,
      "loss": 0.5739,
      "step": 6629
    },
    {
      "epoch": 0.8390280941533789,
      "grad_norm": 0.889288067817688,
      "learning_rate": 1.3900803323038284e-05,
      "loss": 0.5905,
      "step": 6630
    },
    {
      "epoch": 0.8391546443938244,
      "grad_norm": 0.9082183241844177,
      "learning_rate": 1.3899153107361579e-05,
      "loss": 0.5522,
      "step": 6631
    },
    {
      "epoch": 0.8392811946342698,
      "grad_norm": 0.9619323015213013,
      "learning_rate": 1.38975027664555e-05,
      "loss": 0.6165,
      "step": 6632
    },
    {
      "epoch": 0.8394077448747153,
      "grad_norm": 0.9458474516868591,
      "learning_rate": 1.3895852300373056e-05,
      "loss": 0.6434,
      "step": 6633
    },
    {
      "epoch": 0.8395342951151608,
      "grad_norm": 0.9920973777770996,
      "learning_rate": 1.3894201709167252e-05,
      "loss": 0.6163,
      "step": 6634
    },
    {
      "epoch": 0.8396608453556061,
      "grad_norm": 0.9139273166656494,
      "learning_rate": 1.38925509928911e-05,
      "loss": 0.6097,
      "step": 6635
    },
    {
      "epoch": 0.8397873955960516,
      "grad_norm": 0.8826106190681458,
      "learning_rate": 1.389090015159762e-05,
      "loss": 0.5595,
      "step": 6636
    },
    {
      "epoch": 0.8399139458364971,
      "grad_norm": 1.0075355768203735,
      "learning_rate": 1.3889249185339826e-05,
      "loss": 0.6182,
      "step": 6637
    },
    {
      "epoch": 0.8400404960769425,
      "grad_norm": 0.9495134949684143,
      "learning_rate": 1.3887598094170746e-05,
      "loss": 0.5858,
      "step": 6638
    },
    {
      "epoch": 0.840167046317388,
      "grad_norm": 0.9145963788032532,
      "learning_rate": 1.3885946878143408e-05,
      "loss": 0.6346,
      "step": 6639
    },
    {
      "epoch": 0.8402935965578334,
      "grad_norm": 0.9055142402648926,
      "learning_rate": 1.3884295537310844e-05,
      "loss": 0.5847,
      "step": 6640
    },
    {
      "epoch": 0.8404201467982789,
      "grad_norm": 0.9301719069480896,
      "learning_rate": 1.388264407172609e-05,
      "loss": 0.6094,
      "step": 6641
    },
    {
      "epoch": 0.8405466970387244,
      "grad_norm": 0.948216438293457,
      "learning_rate": 1.3880992481442187e-05,
      "loss": 0.6312,
      "step": 6642
    },
    {
      "epoch": 0.8406732472791698,
      "grad_norm": 0.8449627757072449,
      "learning_rate": 1.3879340766512176e-05,
      "loss": 0.59,
      "step": 6643
    },
    {
      "epoch": 0.8407997975196153,
      "grad_norm": 0.8609544634819031,
      "learning_rate": 1.387768892698911e-05,
      "loss": 0.5412,
      "step": 6644
    },
    {
      "epoch": 0.8409263477600607,
      "grad_norm": 0.8950684666633606,
      "learning_rate": 1.387603696292604e-05,
      "loss": 0.579,
      "step": 6645
    },
    {
      "epoch": 0.8410528980005062,
      "grad_norm": 1.0423763990402222,
      "learning_rate": 1.3874384874376021e-05,
      "loss": 0.6172,
      "step": 6646
    },
    {
      "epoch": 0.8411794482409517,
      "grad_norm": 0.9503747820854187,
      "learning_rate": 1.3872732661392112e-05,
      "loss": 0.6086,
      "step": 6647
    },
    {
      "epoch": 0.8413059984813971,
      "grad_norm": 0.8499541282653809,
      "learning_rate": 1.3871080324027382e-05,
      "loss": 0.5929,
      "step": 6648
    },
    {
      "epoch": 0.8414325487218426,
      "grad_norm": 0.9976781010627747,
      "learning_rate": 1.3869427862334894e-05,
      "loss": 0.5815,
      "step": 6649
    },
    {
      "epoch": 0.841559098962288,
      "grad_norm": 0.9125748872756958,
      "learning_rate": 1.386777527636772e-05,
      "loss": 0.565,
      "step": 6650
    },
    {
      "epoch": 0.8416856492027335,
      "grad_norm": 0.9535832405090332,
      "learning_rate": 1.386612256617894e-05,
      "loss": 0.6259,
      "step": 6651
    },
    {
      "epoch": 0.841812199443179,
      "grad_norm": 0.9705054759979248,
      "learning_rate": 1.3864469731821633e-05,
      "loss": 0.6392,
      "step": 6652
    },
    {
      "epoch": 0.8419387496836244,
      "grad_norm": 0.9105415940284729,
      "learning_rate": 1.3862816773348881e-05,
      "loss": 0.5695,
      "step": 6653
    },
    {
      "epoch": 0.8420652999240699,
      "grad_norm": 0.9371193647384644,
      "learning_rate": 1.3861163690813779e-05,
      "loss": 0.6109,
      "step": 6654
    },
    {
      "epoch": 0.8421918501645154,
      "grad_norm": 0.918563187122345,
      "learning_rate": 1.3859510484269411e-05,
      "loss": 0.5867,
      "step": 6655
    },
    {
      "epoch": 0.8423184004049608,
      "grad_norm": 0.9843639731407166,
      "learning_rate": 1.3857857153768876e-05,
      "loss": 0.6334,
      "step": 6656
    },
    {
      "epoch": 0.8424449506454063,
      "grad_norm": 0.9117849469184875,
      "learning_rate": 1.3856203699365278e-05,
      "loss": 0.5905,
      "step": 6657
    },
    {
      "epoch": 0.8425715008858516,
      "grad_norm": 1.018241047859192,
      "learning_rate": 1.3854550121111714e-05,
      "loss": 0.6126,
      "step": 6658
    },
    {
      "epoch": 0.8426980511262971,
      "grad_norm": 1.0152233839035034,
      "learning_rate": 1.3852896419061299e-05,
      "loss": 0.5815,
      "step": 6659
    },
    {
      "epoch": 0.8428246013667426,
      "grad_norm": 0.8697996139526367,
      "learning_rate": 1.3851242593267138e-05,
      "loss": 0.6046,
      "step": 6660
    },
    {
      "epoch": 0.842951151607188,
      "grad_norm": 0.8425203561782837,
      "learning_rate": 1.3849588643782353e-05,
      "loss": 0.5607,
      "step": 6661
    },
    {
      "epoch": 0.8430777018476335,
      "grad_norm": 0.9375426769256592,
      "learning_rate": 1.384793457066006e-05,
      "loss": 0.615,
      "step": 6662
    },
    {
      "epoch": 0.8432042520880789,
      "grad_norm": 0.9226163029670715,
      "learning_rate": 1.3846280373953384e-05,
      "loss": 0.6089,
      "step": 6663
    },
    {
      "epoch": 0.8433308023285244,
      "grad_norm": 0.8831661343574524,
      "learning_rate": 1.3844626053715456e-05,
      "loss": 0.6048,
      "step": 6664
    },
    {
      "epoch": 0.8434573525689699,
      "grad_norm": 0.966916561126709,
      "learning_rate": 1.3842971609999407e-05,
      "loss": 0.6464,
      "step": 6665
    },
    {
      "epoch": 0.8435839028094153,
      "grad_norm": 0.870904803276062,
      "learning_rate": 1.3841317042858369e-05,
      "loss": 0.5646,
      "step": 6666
    },
    {
      "epoch": 0.8437104530498608,
      "grad_norm": 0.9219532608985901,
      "learning_rate": 1.3839662352345484e-05,
      "loss": 0.5363,
      "step": 6667
    },
    {
      "epoch": 0.8438370032903062,
      "grad_norm": 0.9170980453491211,
      "learning_rate": 1.3838007538513898e-05,
      "loss": 0.6361,
      "step": 6668
    },
    {
      "epoch": 0.8439635535307517,
      "grad_norm": 0.933588445186615,
      "learning_rate": 1.3836352601416758e-05,
      "loss": 0.6323,
      "step": 6669
    },
    {
      "epoch": 0.8440901037711972,
      "grad_norm": 0.9092592000961304,
      "learning_rate": 1.3834697541107213e-05,
      "loss": 0.5419,
      "step": 6670
    },
    {
      "epoch": 0.8442166540116426,
      "grad_norm": 0.9245826601982117,
      "learning_rate": 1.3833042357638423e-05,
      "loss": 0.634,
      "step": 6671
    },
    {
      "epoch": 0.8443432042520881,
      "grad_norm": 0.8656516075134277,
      "learning_rate": 1.3831387051063541e-05,
      "loss": 0.5897,
      "step": 6672
    },
    {
      "epoch": 0.8444697544925336,
      "grad_norm": 0.9016159176826477,
      "learning_rate": 1.3829731621435736e-05,
      "loss": 0.6254,
      "step": 6673
    },
    {
      "epoch": 0.844596304732979,
      "grad_norm": 0.9076156616210938,
      "learning_rate": 1.3828076068808174e-05,
      "loss": 0.58,
      "step": 6674
    },
    {
      "epoch": 0.8447228549734245,
      "grad_norm": 0.9383162260055542,
      "learning_rate": 1.3826420393234028e-05,
      "loss": 0.5931,
      "step": 6675
    },
    {
      "epoch": 0.8448494052138699,
      "grad_norm": 0.8998589515686035,
      "learning_rate": 1.3824764594766471e-05,
      "loss": 0.6015,
      "step": 6676
    },
    {
      "epoch": 0.8449759554543154,
      "grad_norm": 0.9332043528556824,
      "learning_rate": 1.3823108673458685e-05,
      "loss": 0.6094,
      "step": 6677
    },
    {
      "epoch": 0.8451025056947609,
      "grad_norm": 1.1361567974090576,
      "learning_rate": 1.382145262936385e-05,
      "loss": 0.6178,
      "step": 6678
    },
    {
      "epoch": 0.8452290559352063,
      "grad_norm": 0.9585675597190857,
      "learning_rate": 1.3819796462535155e-05,
      "loss": 0.5866,
      "step": 6679
    },
    {
      "epoch": 0.8453556061756518,
      "grad_norm": 0.8896278142929077,
      "learning_rate": 1.3818140173025794e-05,
      "loss": 0.6021,
      "step": 6680
    },
    {
      "epoch": 0.8454821564160971,
      "grad_norm": 0.9524937272071838,
      "learning_rate": 1.3816483760888959e-05,
      "loss": 0.6413,
      "step": 6681
    },
    {
      "epoch": 0.8456087066565426,
      "grad_norm": 0.924495279788971,
      "learning_rate": 1.3814827226177847e-05,
      "loss": 0.6175,
      "step": 6682
    },
    {
      "epoch": 0.8457352568969881,
      "grad_norm": 0.953271210193634,
      "learning_rate": 1.3813170568945666e-05,
      "loss": 0.5895,
      "step": 6683
    },
    {
      "epoch": 0.8458618071374335,
      "grad_norm": 0.9360235929489136,
      "learning_rate": 1.381151378924562e-05,
      "loss": 0.6354,
      "step": 6684
    },
    {
      "epoch": 0.845988357377879,
      "grad_norm": 0.9695387482643127,
      "learning_rate": 1.380985688713092e-05,
      "loss": 0.5776,
      "step": 6685
    },
    {
      "epoch": 0.8461149076183245,
      "grad_norm": 0.9678358435630798,
      "learning_rate": 1.3808199862654783e-05,
      "loss": 0.611,
      "step": 6686
    },
    {
      "epoch": 0.8462414578587699,
      "grad_norm": 0.9493866562843323,
      "learning_rate": 1.3806542715870423e-05,
      "loss": 0.5809,
      "step": 6687
    },
    {
      "epoch": 0.8463680080992154,
      "grad_norm": 0.9286646246910095,
      "learning_rate": 1.380488544683107e-05,
      "loss": 0.5878,
      "step": 6688
    },
    {
      "epoch": 0.8464945583396608,
      "grad_norm": 0.9000338912010193,
      "learning_rate": 1.3803228055589944e-05,
      "loss": 0.641,
      "step": 6689
    },
    {
      "epoch": 0.8466211085801063,
      "grad_norm": 1.12400484085083,
      "learning_rate": 1.3801570542200277e-05,
      "loss": 0.6983,
      "step": 6690
    },
    {
      "epoch": 0.8467476588205518,
      "grad_norm": 0.913135290145874,
      "learning_rate": 1.3799912906715306e-05,
      "loss": 0.5934,
      "step": 6691
    },
    {
      "epoch": 0.8468742090609972,
      "grad_norm": 0.8889170289039612,
      "learning_rate": 1.3798255149188269e-05,
      "loss": 0.5652,
      "step": 6692
    },
    {
      "epoch": 0.8470007593014427,
      "grad_norm": 0.9389752745628357,
      "learning_rate": 1.379659726967241e-05,
      "loss": 0.598,
      "step": 6693
    },
    {
      "epoch": 0.8471273095418881,
      "grad_norm": 0.9070677161216736,
      "learning_rate": 1.3794939268220966e-05,
      "loss": 0.5612,
      "step": 6694
    },
    {
      "epoch": 0.8472538597823336,
      "grad_norm": 0.9298079013824463,
      "learning_rate": 1.3793281144887198e-05,
      "loss": 0.602,
      "step": 6695
    },
    {
      "epoch": 0.8473804100227791,
      "grad_norm": 1.1552815437316895,
      "learning_rate": 1.3791622899724355e-05,
      "loss": 0.6177,
      "step": 6696
    },
    {
      "epoch": 0.8475069602632245,
      "grad_norm": 0.8929833173751831,
      "learning_rate": 1.3789964532785694e-05,
      "loss": 0.5537,
      "step": 6697
    },
    {
      "epoch": 0.84763351050367,
      "grad_norm": 0.9052556753158569,
      "learning_rate": 1.378830604412448e-05,
      "loss": 0.5953,
      "step": 6698
    },
    {
      "epoch": 0.8477600607441154,
      "grad_norm": 0.9424183964729309,
      "learning_rate": 1.3786647433793973e-05,
      "loss": 0.5638,
      "step": 6699
    },
    {
      "epoch": 0.8478866109845609,
      "grad_norm": 0.8961161375045776,
      "learning_rate": 1.3784988701847453e-05,
      "loss": 0.5889,
      "step": 6700
    },
    {
      "epoch": 0.8480131612250064,
      "grad_norm": 0.8978267908096313,
      "learning_rate": 1.3783329848338184e-05,
      "loss": 0.5968,
      "step": 6701
    },
    {
      "epoch": 0.8481397114654518,
      "grad_norm": 0.8892055749893188,
      "learning_rate": 1.3781670873319451e-05,
      "loss": 0.5726,
      "step": 6702
    },
    {
      "epoch": 0.8482662617058973,
      "grad_norm": 0.9103731513023376,
      "learning_rate": 1.3780011776844528e-05,
      "loss": 0.5801,
      "step": 6703
    },
    {
      "epoch": 0.8483928119463428,
      "grad_norm": 0.8960819244384766,
      "learning_rate": 1.3778352558966704e-05,
      "loss": 0.5346,
      "step": 6704
    },
    {
      "epoch": 0.8485193621867881,
      "grad_norm": 0.9089915752410889,
      "learning_rate": 1.3776693219739272e-05,
      "loss": 0.6356,
      "step": 6705
    },
    {
      "epoch": 0.8486459124272336,
      "grad_norm": 0.9044176936149597,
      "learning_rate": 1.3775033759215517e-05,
      "loss": 0.6148,
      "step": 6706
    },
    {
      "epoch": 0.848772462667679,
      "grad_norm": 0.9265152215957642,
      "learning_rate": 1.3773374177448744e-05,
      "loss": 0.6284,
      "step": 6707
    },
    {
      "epoch": 0.8488990129081245,
      "grad_norm": 0.9547927379608154,
      "learning_rate": 1.3771714474492245e-05,
      "loss": 0.6377,
      "step": 6708
    },
    {
      "epoch": 0.84902556314857,
      "grad_norm": 0.8847858309745789,
      "learning_rate": 1.3770054650399334e-05,
      "loss": 0.5675,
      "step": 6709
    },
    {
      "epoch": 0.8491521133890154,
      "grad_norm": 0.8794218897819519,
      "learning_rate": 1.3768394705223313e-05,
      "loss": 0.5827,
      "step": 6710
    },
    {
      "epoch": 0.8492786636294609,
      "grad_norm": 0.8964936137199402,
      "learning_rate": 1.3766734639017501e-05,
      "loss": 0.5837,
      "step": 6711
    },
    {
      "epoch": 0.8494052138699063,
      "grad_norm": 0.8930639624595642,
      "learning_rate": 1.3765074451835208e-05,
      "loss": 0.6037,
      "step": 6712
    },
    {
      "epoch": 0.8495317641103518,
      "grad_norm": 0.959267795085907,
      "learning_rate": 1.3763414143729756e-05,
      "loss": 0.6587,
      "step": 6713
    },
    {
      "epoch": 0.8496583143507973,
      "grad_norm": 0.8852070569992065,
      "learning_rate": 1.376175371475447e-05,
      "loss": 0.6006,
      "step": 6714
    },
    {
      "epoch": 0.8497848645912427,
      "grad_norm": 0.9192984700202942,
      "learning_rate": 1.3760093164962682e-05,
      "loss": 0.6001,
      "step": 6715
    },
    {
      "epoch": 0.8499114148316882,
      "grad_norm": 0.9136788249015808,
      "learning_rate": 1.3758432494407717e-05,
      "loss": 0.5883,
      "step": 6716
    },
    {
      "epoch": 0.8500379650721336,
      "grad_norm": 1.01317298412323,
      "learning_rate": 1.3756771703142918e-05,
      "loss": 0.5983,
      "step": 6717
    },
    {
      "epoch": 0.8501645153125791,
      "grad_norm": 0.9490894675254822,
      "learning_rate": 1.3755110791221618e-05,
      "loss": 0.6078,
      "step": 6718
    },
    {
      "epoch": 0.8502910655530246,
      "grad_norm": 0.922404408454895,
      "learning_rate": 1.3753449758697164e-05,
      "loss": 0.5469,
      "step": 6719
    },
    {
      "epoch": 0.85041761579347,
      "grad_norm": 0.9005807042121887,
      "learning_rate": 1.3751788605622904e-05,
      "loss": 0.6055,
      "step": 6720
    },
    {
      "epoch": 0.8505441660339155,
      "grad_norm": 0.9213549494743347,
      "learning_rate": 1.3750127332052185e-05,
      "loss": 0.5499,
      "step": 6721
    },
    {
      "epoch": 0.850670716274361,
      "grad_norm": 0.9589576721191406,
      "learning_rate": 1.3748465938038369e-05,
      "loss": 0.5955,
      "step": 6722
    },
    {
      "epoch": 0.8507972665148064,
      "grad_norm": 0.9246467351913452,
      "learning_rate": 1.3746804423634812e-05,
      "loss": 0.5986,
      "step": 6723
    },
    {
      "epoch": 0.8509238167552519,
      "grad_norm": 0.8969545364379883,
      "learning_rate": 1.3745142788894876e-05,
      "loss": 0.5723,
      "step": 6724
    },
    {
      "epoch": 0.8510503669956972,
      "grad_norm": 0.88646000623703,
      "learning_rate": 1.3743481033871929e-05,
      "loss": 0.5628,
      "step": 6725
    },
    {
      "epoch": 0.8511769172361427,
      "grad_norm": 0.869911253452301,
      "learning_rate": 1.3741819158619338e-05,
      "loss": 0.567,
      "step": 6726
    },
    {
      "epoch": 0.8513034674765882,
      "grad_norm": 0.9179938435554504,
      "learning_rate": 1.3740157163190484e-05,
      "loss": 0.6416,
      "step": 6727
    },
    {
      "epoch": 0.8514300177170336,
      "grad_norm": 0.9170773029327393,
      "learning_rate": 1.3738495047638743e-05,
      "loss": 0.6481,
      "step": 6728
    },
    {
      "epoch": 0.8515565679574791,
      "grad_norm": 0.9107789397239685,
      "learning_rate": 1.3736832812017496e-05,
      "loss": 0.6029,
      "step": 6729
    },
    {
      "epoch": 0.8516831181979245,
      "grad_norm": 1.9601500034332275,
      "learning_rate": 1.3735170456380129e-05,
      "loss": 0.5816,
      "step": 6730
    },
    {
      "epoch": 0.85180966843837,
      "grad_norm": 0.9314422607421875,
      "learning_rate": 1.3733507980780035e-05,
      "loss": 0.5732,
      "step": 6731
    },
    {
      "epoch": 0.8519362186788155,
      "grad_norm": 0.9280622005462646,
      "learning_rate": 1.3731845385270601e-05,
      "loss": 0.6195,
      "step": 6732
    },
    {
      "epoch": 0.8520627689192609,
      "grad_norm": 1.1211754083633423,
      "learning_rate": 1.3730182669905233e-05,
      "loss": 0.6112,
      "step": 6733
    },
    {
      "epoch": 0.8521893191597064,
      "grad_norm": 0.8829653263092041,
      "learning_rate": 1.3728519834737328e-05,
      "loss": 0.6061,
      "step": 6734
    },
    {
      "epoch": 0.8523158694001519,
      "grad_norm": 0.9792661070823669,
      "learning_rate": 1.3726856879820292e-05,
      "loss": 0.5915,
      "step": 6735
    },
    {
      "epoch": 0.8524424196405973,
      "grad_norm": 0.9516897201538086,
      "learning_rate": 1.3725193805207532e-05,
      "loss": 0.6107,
      "step": 6736
    },
    {
      "epoch": 0.8525689698810428,
      "grad_norm": 0.8952273726463318,
      "learning_rate": 1.3723530610952466e-05,
      "loss": 0.6049,
      "step": 6737
    },
    {
      "epoch": 0.8526955201214882,
      "grad_norm": 0.9422025680541992,
      "learning_rate": 1.3721867297108508e-05,
      "loss": 0.5819,
      "step": 6738
    },
    {
      "epoch": 0.8528220703619337,
      "grad_norm": 0.9305140376091003,
      "learning_rate": 1.372020386372908e-05,
      "loss": 0.628,
      "step": 6739
    },
    {
      "epoch": 0.8529486206023792,
      "grad_norm": 1.0342588424682617,
      "learning_rate": 1.3718540310867607e-05,
      "loss": 0.6124,
      "step": 6740
    },
    {
      "epoch": 0.8530751708428246,
      "grad_norm": 0.9049599170684814,
      "learning_rate": 1.3716876638577512e-05,
      "loss": 0.6459,
      "step": 6741
    },
    {
      "epoch": 0.8532017210832701,
      "grad_norm": 1.0146366357803345,
      "learning_rate": 1.3715212846912233e-05,
      "loss": 0.6537,
      "step": 6742
    },
    {
      "epoch": 0.8533282713237155,
      "grad_norm": 0.9891189336776733,
      "learning_rate": 1.371354893592521e-05,
      "loss": 0.5813,
      "step": 6743
    },
    {
      "epoch": 0.853454821564161,
      "grad_norm": 0.9185702800750732,
      "learning_rate": 1.3711884905669871e-05,
      "loss": 0.5909,
      "step": 6744
    },
    {
      "epoch": 0.8535813718046065,
      "grad_norm": 0.9187882542610168,
      "learning_rate": 1.3710220756199668e-05,
      "loss": 0.56,
      "step": 6745
    },
    {
      "epoch": 0.8537079220450519,
      "grad_norm": 0.8716415762901306,
      "learning_rate": 1.370855648756805e-05,
      "loss": 0.5857,
      "step": 6746
    },
    {
      "epoch": 0.8538344722854974,
      "grad_norm": 0.9105336666107178,
      "learning_rate": 1.3706892099828461e-05,
      "loss": 0.6215,
      "step": 6747
    },
    {
      "epoch": 0.8539610225259427,
      "grad_norm": 0.8906025290489197,
      "learning_rate": 1.3705227593034363e-05,
      "loss": 0.5579,
      "step": 6748
    },
    {
      "epoch": 0.8540875727663882,
      "grad_norm": 0.8580873608589172,
      "learning_rate": 1.3703562967239213e-05,
      "loss": 0.6057,
      "step": 6749
    },
    {
      "epoch": 0.8542141230068337,
      "grad_norm": 0.9363141059875488,
      "learning_rate": 1.370189822249647e-05,
      "loss": 0.604,
      "step": 6750
    },
    {
      "epoch": 0.8543406732472791,
      "grad_norm": 0.9504256844520569,
      "learning_rate": 1.3700233358859608e-05,
      "loss": 0.6133,
      "step": 6751
    },
    {
      "epoch": 0.8544672234877246,
      "grad_norm": 1.0939087867736816,
      "learning_rate": 1.3698568376382096e-05,
      "loss": 0.587,
      "step": 6752
    },
    {
      "epoch": 0.8545937737281701,
      "grad_norm": 0.955476701259613,
      "learning_rate": 1.3696903275117402e-05,
      "loss": 0.5873,
      "step": 6753
    },
    {
      "epoch": 0.8547203239686155,
      "grad_norm": 0.949418842792511,
      "learning_rate": 1.3695238055119008e-05,
      "loss": 0.6027,
      "step": 6754
    },
    {
      "epoch": 0.854846874209061,
      "grad_norm": 0.931566059589386,
      "learning_rate": 1.3693572716440401e-05,
      "loss": 0.5926,
      "step": 6755
    },
    {
      "epoch": 0.8549734244495064,
      "grad_norm": 0.9447817206382751,
      "learning_rate": 1.369190725913506e-05,
      "loss": 0.6442,
      "step": 6756
    },
    {
      "epoch": 0.8550999746899519,
      "grad_norm": 0.9000686407089233,
      "learning_rate": 1.3690241683256478e-05,
      "loss": 0.5953,
      "step": 6757
    },
    {
      "epoch": 0.8552265249303974,
      "grad_norm": 0.8443261981010437,
      "learning_rate": 1.3688575988858145e-05,
      "loss": 0.5565,
      "step": 6758
    },
    {
      "epoch": 0.8553530751708428,
      "grad_norm": 0.9201819896697998,
      "learning_rate": 1.368691017599356e-05,
      "loss": 0.6114,
      "step": 6759
    },
    {
      "epoch": 0.8554796254112883,
      "grad_norm": 0.8948781490325928,
      "learning_rate": 1.3685244244716225e-05,
      "loss": 0.5715,
      "step": 6760
    },
    {
      "epoch": 0.8556061756517337,
      "grad_norm": 1.088076114654541,
      "learning_rate": 1.3683578195079643e-05,
      "loss": 0.6455,
      "step": 6761
    },
    {
      "epoch": 0.8557327258921792,
      "grad_norm": 0.9577755331993103,
      "learning_rate": 1.3681912027137328e-05,
      "loss": 0.6505,
      "step": 6762
    },
    {
      "epoch": 0.8558592761326247,
      "grad_norm": 0.8942970037460327,
      "learning_rate": 1.3680245740942785e-05,
      "loss": 0.5732,
      "step": 6763
    },
    {
      "epoch": 0.8559858263730701,
      "grad_norm": 0.8798139095306396,
      "learning_rate": 1.3678579336549535e-05,
      "loss": 0.5733,
      "step": 6764
    },
    {
      "epoch": 0.8561123766135156,
      "grad_norm": 0.8711024522781372,
      "learning_rate": 1.3676912814011096e-05,
      "loss": 0.5846,
      "step": 6765
    },
    {
      "epoch": 0.856238926853961,
      "grad_norm": 0.9483872056007385,
      "learning_rate": 1.3675246173380993e-05,
      "loss": 0.6207,
      "step": 6766
    },
    {
      "epoch": 0.8563654770944065,
      "grad_norm": 1.0720140933990479,
      "learning_rate": 1.3673579414712754e-05,
      "loss": 0.6201,
      "step": 6767
    },
    {
      "epoch": 0.856492027334852,
      "grad_norm": 0.906495213508606,
      "learning_rate": 1.3671912538059909e-05,
      "loss": 0.6016,
      "step": 6768
    },
    {
      "epoch": 0.8566185775752974,
      "grad_norm": 0.9568055272102356,
      "learning_rate": 1.3670245543475995e-05,
      "loss": 0.6149,
      "step": 6769
    },
    {
      "epoch": 0.8567451278157429,
      "grad_norm": 0.927063524723053,
      "learning_rate": 1.3668578431014548e-05,
      "loss": 0.5963,
      "step": 6770
    },
    {
      "epoch": 0.8568716780561884,
      "grad_norm": 0.8909215927124023,
      "learning_rate": 1.3666911200729112e-05,
      "loss": 0.6094,
      "step": 6771
    },
    {
      "epoch": 0.8569982282966337,
      "grad_norm": 0.8974158763885498,
      "learning_rate": 1.3665243852673233e-05,
      "loss": 0.6042,
      "step": 6772
    },
    {
      "epoch": 0.8571247785370792,
      "grad_norm": 0.88932204246521,
      "learning_rate": 1.3663576386900464e-05,
      "loss": 0.5829,
      "step": 6773
    },
    {
      "epoch": 0.8572513287775246,
      "grad_norm": 0.9010241627693176,
      "learning_rate": 1.3661908803464355e-05,
      "loss": 0.5933,
      "step": 6774
    },
    {
      "epoch": 0.8573778790179701,
      "grad_norm": 0.8664101362228394,
      "learning_rate": 1.366024110241847e-05,
      "loss": 0.5935,
      "step": 6775
    },
    {
      "epoch": 0.8575044292584156,
      "grad_norm": 0.923099935054779,
      "learning_rate": 1.3658573283816366e-05,
      "loss": 0.6231,
      "step": 6776
    },
    {
      "epoch": 0.857630979498861,
      "grad_norm": 0.8806052207946777,
      "learning_rate": 1.3656905347711607e-05,
      "loss": 0.5541,
      "step": 6777
    },
    {
      "epoch": 0.8577575297393065,
      "grad_norm": 0.898094892501831,
      "learning_rate": 1.3655237294157767e-05,
      "loss": 0.6051,
      "step": 6778
    },
    {
      "epoch": 0.8578840799797519,
      "grad_norm": 0.8852434158325195,
      "learning_rate": 1.3653569123208417e-05,
      "loss": 0.5823,
      "step": 6779
    },
    {
      "epoch": 0.8580106302201974,
      "grad_norm": 0.8233832716941833,
      "learning_rate": 1.365190083491713e-05,
      "loss": 0.5389,
      "step": 6780
    },
    {
      "epoch": 0.8581371804606429,
      "grad_norm": 0.895520806312561,
      "learning_rate": 1.3650232429337495e-05,
      "loss": 0.6071,
      "step": 6781
    },
    {
      "epoch": 0.8582637307010883,
      "grad_norm": 0.9160528182983398,
      "learning_rate": 1.3648563906523089e-05,
      "loss": 0.5811,
      "step": 6782
    },
    {
      "epoch": 0.8583902809415338,
      "grad_norm": 0.8465480804443359,
      "learning_rate": 1.36468952665275e-05,
      "loss": 0.613,
      "step": 6783
    },
    {
      "epoch": 0.8585168311819793,
      "grad_norm": 0.9157769680023193,
      "learning_rate": 1.3645226509404324e-05,
      "loss": 0.6273,
      "step": 6784
    },
    {
      "epoch": 0.8586433814224247,
      "grad_norm": 0.956445574760437,
      "learning_rate": 1.3643557635207154e-05,
      "loss": 0.6054,
      "step": 6785
    },
    {
      "epoch": 0.8587699316628702,
      "grad_norm": 0.9156041145324707,
      "learning_rate": 1.3641888643989593e-05,
      "loss": 0.5323,
      "step": 6786
    },
    {
      "epoch": 0.8588964819033156,
      "grad_norm": 0.8941291570663452,
      "learning_rate": 1.3640219535805238e-05,
      "loss": 0.5574,
      "step": 6787
    },
    {
      "epoch": 0.8590230321437611,
      "grad_norm": 0.8939699530601501,
      "learning_rate": 1.36385503107077e-05,
      "loss": 0.6055,
      "step": 6788
    },
    {
      "epoch": 0.8591495823842066,
      "grad_norm": 0.9600785970687866,
      "learning_rate": 1.3636880968750587e-05,
      "loss": 0.5783,
      "step": 6789
    },
    {
      "epoch": 0.859276132624652,
      "grad_norm": 0.8976497650146484,
      "learning_rate": 1.3635211509987519e-05,
      "loss": 0.6048,
      "step": 6790
    },
    {
      "epoch": 0.8594026828650975,
      "grad_norm": 0.9011789560317993,
      "learning_rate": 1.3633541934472107e-05,
      "loss": 0.604,
      "step": 6791
    },
    {
      "epoch": 0.8595292331055429,
      "grad_norm": 0.9283462762832642,
      "learning_rate": 1.3631872242257978e-05,
      "loss": 0.6327,
      "step": 6792
    },
    {
      "epoch": 0.8596557833459884,
      "grad_norm": 0.8818456530570984,
      "learning_rate": 1.3630202433398755e-05,
      "loss": 0.6248,
      "step": 6793
    },
    {
      "epoch": 0.8597823335864339,
      "grad_norm": 0.9216594099998474,
      "learning_rate": 1.3628532507948067e-05,
      "loss": 0.619,
      "step": 6794
    },
    {
      "epoch": 0.8599088838268792,
      "grad_norm": 0.9366409778594971,
      "learning_rate": 1.3626862465959547e-05,
      "loss": 0.6244,
      "step": 6795
    },
    {
      "epoch": 0.8600354340673247,
      "grad_norm": 0.8931125402450562,
      "learning_rate": 1.3625192307486834e-05,
      "loss": 0.5984,
      "step": 6796
    },
    {
      "epoch": 0.8601619843077701,
      "grad_norm": 0.8965520858764648,
      "learning_rate": 1.362352203258357e-05,
      "loss": 0.6049,
      "step": 6797
    },
    {
      "epoch": 0.8602885345482156,
      "grad_norm": 0.8706507682800293,
      "learning_rate": 1.3621851641303395e-05,
      "loss": 0.5886,
      "step": 6798
    },
    {
      "epoch": 0.8604150847886611,
      "grad_norm": 0.8895033597946167,
      "learning_rate": 1.362018113369996e-05,
      "loss": 0.6052,
      "step": 6799
    },
    {
      "epoch": 0.8605416350291065,
      "grad_norm": 0.92757648229599,
      "learning_rate": 1.3618510509826913e-05,
      "loss": 0.5854,
      "step": 6800
    },
    {
      "epoch": 0.860668185269552,
      "grad_norm": 0.8873517513275146,
      "learning_rate": 1.3616839769737913e-05,
      "loss": 0.5719,
      "step": 6801
    },
    {
      "epoch": 0.8607947355099975,
      "grad_norm": 0.9619740843772888,
      "learning_rate": 1.3615168913486621e-05,
      "loss": 0.5355,
      "step": 6802
    },
    {
      "epoch": 0.8609212857504429,
      "grad_norm": 0.8827283978462219,
      "learning_rate": 1.3613497941126696e-05,
      "loss": 0.5596,
      "step": 6803
    },
    {
      "epoch": 0.8610478359908884,
      "grad_norm": 0.8840576410293579,
      "learning_rate": 1.3611826852711807e-05,
      "loss": 0.5753,
      "step": 6804
    },
    {
      "epoch": 0.8611743862313338,
      "grad_norm": 0.9362415671348572,
      "learning_rate": 1.3610155648295624e-05,
      "loss": 0.6141,
      "step": 6805
    },
    {
      "epoch": 0.8613009364717793,
      "grad_norm": 0.9291157126426697,
      "learning_rate": 1.360848432793182e-05,
      "loss": 0.5728,
      "step": 6806
    },
    {
      "epoch": 0.8614274867122248,
      "grad_norm": 0.9398753643035889,
      "learning_rate": 1.3606812891674075e-05,
      "loss": 0.5596,
      "step": 6807
    },
    {
      "epoch": 0.8615540369526702,
      "grad_norm": 1.1135869026184082,
      "learning_rate": 1.3605141339576066e-05,
      "loss": 0.5886,
      "step": 6808
    },
    {
      "epoch": 0.8616805871931157,
      "grad_norm": 0.9347293376922607,
      "learning_rate": 1.3603469671691489e-05,
      "loss": 0.6129,
      "step": 6809
    },
    {
      "epoch": 0.8618071374335611,
      "grad_norm": 0.9541815519332886,
      "learning_rate": 1.3601797888074022e-05,
      "loss": 0.5728,
      "step": 6810
    },
    {
      "epoch": 0.8619336876740066,
      "grad_norm": 0.9245452284812927,
      "learning_rate": 1.3600125988777359e-05,
      "loss": 0.5931,
      "step": 6811
    },
    {
      "epoch": 0.8620602379144521,
      "grad_norm": 0.9040193557739258,
      "learning_rate": 1.3598453973855201e-05,
      "loss": 0.5885,
      "step": 6812
    },
    {
      "epoch": 0.8621867881548975,
      "grad_norm": 0.9486054182052612,
      "learning_rate": 1.3596781843361249e-05,
      "loss": 0.6372,
      "step": 6813
    },
    {
      "epoch": 0.862313338395343,
      "grad_norm": 0.9245076775550842,
      "learning_rate": 1.35951095973492e-05,
      "loss": 0.5998,
      "step": 6814
    },
    {
      "epoch": 0.8624398886357884,
      "grad_norm": 0.9260420203208923,
      "learning_rate": 1.3593437235872772e-05,
      "loss": 0.621,
      "step": 6815
    },
    {
      "epoch": 0.8625664388762339,
      "grad_norm": 0.9878348708152771,
      "learning_rate": 1.3591764758985665e-05,
      "loss": 0.5881,
      "step": 6816
    },
    {
      "epoch": 0.8626929891166794,
      "grad_norm": 0.9184291362762451,
      "learning_rate": 1.3590092166741604e-05,
      "loss": 0.6295,
      "step": 6817
    },
    {
      "epoch": 0.8628195393571247,
      "grad_norm": 0.9397498965263367,
      "learning_rate": 1.35884194591943e-05,
      "loss": 0.5824,
      "step": 6818
    },
    {
      "epoch": 0.8629460895975702,
      "grad_norm": 1.0467835664749146,
      "learning_rate": 1.358674663639748e-05,
      "loss": 0.5998,
      "step": 6819
    },
    {
      "epoch": 0.8630726398380157,
      "grad_norm": 0.8750253319740295,
      "learning_rate": 1.3585073698404866e-05,
      "loss": 0.5957,
      "step": 6820
    },
    {
      "epoch": 0.8631991900784611,
      "grad_norm": 0.8705596327781677,
      "learning_rate": 1.3583400645270197e-05,
      "loss": 0.6221,
      "step": 6821
    },
    {
      "epoch": 0.8633257403189066,
      "grad_norm": 0.9502031207084656,
      "learning_rate": 1.3581727477047195e-05,
      "loss": 0.569,
      "step": 6822
    },
    {
      "epoch": 0.863452290559352,
      "grad_norm": 0.9155302047729492,
      "learning_rate": 1.3580054193789605e-05,
      "loss": 0.6005,
      "step": 6823
    },
    {
      "epoch": 0.8635788407997975,
      "grad_norm": 0.8453879952430725,
      "learning_rate": 1.3578380795551165e-05,
      "loss": 0.5397,
      "step": 6824
    },
    {
      "epoch": 0.863705391040243,
      "grad_norm": 0.8977921605110168,
      "learning_rate": 1.3576707282385617e-05,
      "loss": 0.6273,
      "step": 6825
    },
    {
      "epoch": 0.8638319412806884,
      "grad_norm": 50.30351257324219,
      "learning_rate": 1.357503365434672e-05,
      "loss": 0.7444,
      "step": 6826
    },
    {
      "epoch": 0.8639584915211339,
      "grad_norm": 0.9121528267860413,
      "learning_rate": 1.3573359911488211e-05,
      "loss": 0.5761,
      "step": 6827
    },
    {
      "epoch": 0.8640850417615793,
      "grad_norm": 1.2540892362594604,
      "learning_rate": 1.3571686053863854e-05,
      "loss": 0.5838,
      "step": 6828
    },
    {
      "epoch": 0.8642115920020248,
      "grad_norm": 0.9676316976547241,
      "learning_rate": 1.3570012081527412e-05,
      "loss": 0.6282,
      "step": 6829
    },
    {
      "epoch": 0.8643381422424703,
      "grad_norm": 0.9285439848899841,
      "learning_rate": 1.356833799453264e-05,
      "loss": 0.5552,
      "step": 6830
    },
    {
      "epoch": 0.8644646924829157,
      "grad_norm": 1.0312094688415527,
      "learning_rate": 1.3566663792933311e-05,
      "loss": 0.5808,
      "step": 6831
    },
    {
      "epoch": 0.8645912427233612,
      "grad_norm": 0.9537996649742126,
      "learning_rate": 1.3564989476783192e-05,
      "loss": 0.605,
      "step": 6832
    },
    {
      "epoch": 0.8647177929638067,
      "grad_norm": 0.9478954672813416,
      "learning_rate": 1.3563315046136059e-05,
      "loss": 0.5912,
      "step": 6833
    },
    {
      "epoch": 0.8648443432042521,
      "grad_norm": 0.8719847202301025,
      "learning_rate": 1.3561640501045686e-05,
      "loss": 0.596,
      "step": 6834
    },
    {
      "epoch": 0.8649708934446976,
      "grad_norm": 1.0829942226409912,
      "learning_rate": 1.355996584156586e-05,
      "loss": 0.5876,
      "step": 6835
    },
    {
      "epoch": 0.865097443685143,
      "grad_norm": 0.9623666405677795,
      "learning_rate": 1.355829106775036e-05,
      "loss": 0.5769,
      "step": 6836
    },
    {
      "epoch": 0.8652239939255885,
      "grad_norm": 0.8932594656944275,
      "learning_rate": 1.3556616179652981e-05,
      "loss": 0.6102,
      "step": 6837
    },
    {
      "epoch": 0.865350544166034,
      "grad_norm": 0.9865716099739075,
      "learning_rate": 1.3554941177327515e-05,
      "loss": 0.5934,
      "step": 6838
    },
    {
      "epoch": 0.8654770944064794,
      "grad_norm": 0.9180856347084045,
      "learning_rate": 1.3553266060827752e-05,
      "loss": 0.5535,
      "step": 6839
    },
    {
      "epoch": 0.8656036446469249,
      "grad_norm": 1.0101784467697144,
      "learning_rate": 1.3551590830207497e-05,
      "loss": 0.6218,
      "step": 6840
    },
    {
      "epoch": 0.8657301948873702,
      "grad_norm": 0.8620776534080505,
      "learning_rate": 1.3549915485520556e-05,
      "loss": 0.6238,
      "step": 6841
    },
    {
      "epoch": 0.8658567451278157,
      "grad_norm": 0.913325309753418,
      "learning_rate": 1.3548240026820727e-05,
      "loss": 0.5562,
      "step": 6842
    },
    {
      "epoch": 0.8659832953682612,
      "grad_norm": 0.907250702381134,
      "learning_rate": 1.354656445416183e-05,
      "loss": 0.5531,
      "step": 6843
    },
    {
      "epoch": 0.8661098456087066,
      "grad_norm": 0.8880519866943359,
      "learning_rate": 1.3544888767597674e-05,
      "loss": 0.5458,
      "step": 6844
    },
    {
      "epoch": 0.8662363958491521,
      "grad_norm": 0.8813243508338928,
      "learning_rate": 1.354321296718208e-05,
      "loss": 0.5867,
      "step": 6845
    },
    {
      "epoch": 0.8663629460895975,
      "grad_norm": 0.8956414461135864,
      "learning_rate": 1.3541537052968867e-05,
      "loss": 0.5903,
      "step": 6846
    },
    {
      "epoch": 0.866489496330043,
      "grad_norm": 0.9343301057815552,
      "learning_rate": 1.3539861025011864e-05,
      "loss": 0.6034,
      "step": 6847
    },
    {
      "epoch": 0.8666160465704885,
      "grad_norm": 0.957200288772583,
      "learning_rate": 1.3538184883364897e-05,
      "loss": 0.5711,
      "step": 6848
    },
    {
      "epoch": 0.8667425968109339,
      "grad_norm": 0.9404920935630798,
      "learning_rate": 1.3536508628081802e-05,
      "loss": 0.6048,
      "step": 6849
    },
    {
      "epoch": 0.8668691470513794,
      "grad_norm": 0.9149640798568726,
      "learning_rate": 1.3534832259216412e-05,
      "loss": 0.5775,
      "step": 6850
    },
    {
      "epoch": 0.8669956972918249,
      "grad_norm": 0.8606804609298706,
      "learning_rate": 1.3533155776822569e-05,
      "loss": 0.5857,
      "step": 6851
    },
    {
      "epoch": 0.8671222475322703,
      "grad_norm": 0.8740735054016113,
      "learning_rate": 1.3531479180954115e-05,
      "loss": 0.611,
      "step": 6852
    },
    {
      "epoch": 0.8672487977727158,
      "grad_norm": 0.9076464772224426,
      "learning_rate": 1.3529802471664902e-05,
      "loss": 0.6217,
      "step": 6853
    },
    {
      "epoch": 0.8673753480131612,
      "grad_norm": 0.8895063996315002,
      "learning_rate": 1.3528125649008774e-05,
      "loss": 0.5983,
      "step": 6854
    },
    {
      "epoch": 0.8675018982536067,
      "grad_norm": 0.8997427225112915,
      "learning_rate": 1.352644871303959e-05,
      "loss": 0.5756,
      "step": 6855
    },
    {
      "epoch": 0.8676284484940522,
      "grad_norm": 0.9342398643493652,
      "learning_rate": 1.3524771663811208e-05,
      "loss": 0.5694,
      "step": 6856
    },
    {
      "epoch": 0.8677549987344976,
      "grad_norm": 0.9000945687294006,
      "learning_rate": 1.352309450137749e-05,
      "loss": 0.6214,
      "step": 6857
    },
    {
      "epoch": 0.8678815489749431,
      "grad_norm": 1.0556256771087646,
      "learning_rate": 1.35214172257923e-05,
      "loss": 0.6498,
      "step": 6858
    },
    {
      "epoch": 0.8680080992153885,
      "grad_norm": 0.9716585874557495,
      "learning_rate": 1.3519739837109506e-05,
      "loss": 0.6186,
      "step": 6859
    },
    {
      "epoch": 0.868134649455834,
      "grad_norm": 0.9741637706756592,
      "learning_rate": 1.3518062335382985e-05,
      "loss": 0.626,
      "step": 6860
    },
    {
      "epoch": 0.8682611996962795,
      "grad_norm": 0.905069887638092,
      "learning_rate": 1.3516384720666614e-05,
      "loss": 0.5735,
      "step": 6861
    },
    {
      "epoch": 0.8683877499367249,
      "grad_norm": 0.9064026474952698,
      "learning_rate": 1.3514706993014266e-05,
      "loss": 0.576,
      "step": 6862
    },
    {
      "epoch": 0.8685143001771704,
      "grad_norm": 0.8842785954475403,
      "learning_rate": 1.3513029152479829e-05,
      "loss": 0.5798,
      "step": 6863
    },
    {
      "epoch": 0.8686408504176157,
      "grad_norm": 0.9297350645065308,
      "learning_rate": 1.3511351199117193e-05,
      "loss": 0.6388,
      "step": 6864
    },
    {
      "epoch": 0.8687674006580612,
      "grad_norm": 0.9806543588638306,
      "learning_rate": 1.3509673132980247e-05,
      "loss": 0.5943,
      "step": 6865
    },
    {
      "epoch": 0.8688939508985067,
      "grad_norm": 0.8561622500419617,
      "learning_rate": 1.3507994954122883e-05,
      "loss": 0.6243,
      "step": 6866
    },
    {
      "epoch": 0.8690205011389521,
      "grad_norm": 0.8920688033103943,
      "learning_rate": 1.3506316662599002e-05,
      "loss": 0.6023,
      "step": 6867
    },
    {
      "epoch": 0.8691470513793976,
      "grad_norm": 0.9612197279930115,
      "learning_rate": 1.3504638258462505e-05,
      "loss": 0.6124,
      "step": 6868
    },
    {
      "epoch": 0.8692736016198431,
      "grad_norm": 0.9513689875602722,
      "learning_rate": 1.3502959741767296e-05,
      "loss": 0.5874,
      "step": 6869
    },
    {
      "epoch": 0.8694001518602885,
      "grad_norm": 0.8658301830291748,
      "learning_rate": 1.3501281112567286e-05,
      "loss": 0.5807,
      "step": 6870
    },
    {
      "epoch": 0.869526702100734,
      "grad_norm": 0.9120857119560242,
      "learning_rate": 1.3499602370916387e-05,
      "loss": 0.6104,
      "step": 6871
    },
    {
      "epoch": 0.8696532523411794,
      "grad_norm": 0.9075548648834229,
      "learning_rate": 1.3497923516868516e-05,
      "loss": 0.5982,
      "step": 6872
    },
    {
      "epoch": 0.8697798025816249,
      "grad_norm": 0.8847180008888245,
      "learning_rate": 1.3496244550477593e-05,
      "loss": 0.5587,
      "step": 6873
    },
    {
      "epoch": 0.8699063528220704,
      "grad_norm": 0.9303979277610779,
      "learning_rate": 1.3494565471797538e-05,
      "loss": 0.5732,
      "step": 6874
    },
    {
      "epoch": 0.8700329030625158,
      "grad_norm": 1.0360990762710571,
      "learning_rate": 1.3492886280882282e-05,
      "loss": 0.5987,
      "step": 6875
    },
    {
      "epoch": 0.8701594533029613,
      "grad_norm": 0.8988347053527832,
      "learning_rate": 1.3491206977785754e-05,
      "loss": 0.6326,
      "step": 6876
    },
    {
      "epoch": 0.8702860035434067,
      "grad_norm": 0.9303744435310364,
      "learning_rate": 1.3489527562561891e-05,
      "loss": 0.6203,
      "step": 6877
    },
    {
      "epoch": 0.8704125537838522,
      "grad_norm": 0.9930063486099243,
      "learning_rate": 1.3487848035264626e-05,
      "loss": 0.5437,
      "step": 6878
    },
    {
      "epoch": 0.8705391040242977,
      "grad_norm": 0.8410153388977051,
      "learning_rate": 1.3486168395947905e-05,
      "loss": 0.5544,
      "step": 6879
    },
    {
      "epoch": 0.8706656542647431,
      "grad_norm": 1.0593781471252441,
      "learning_rate": 1.348448864466567e-05,
      "loss": 0.5935,
      "step": 6880
    },
    {
      "epoch": 0.8707922045051886,
      "grad_norm": 0.9328171610832214,
      "learning_rate": 1.348280878147187e-05,
      "loss": 0.624,
      "step": 6881
    },
    {
      "epoch": 0.8709187547456341,
      "grad_norm": 0.9236932992935181,
      "learning_rate": 1.348112880642046e-05,
      "loss": 0.6286,
      "step": 6882
    },
    {
      "epoch": 0.8710453049860795,
      "grad_norm": 0.9105655550956726,
      "learning_rate": 1.3479448719565393e-05,
      "loss": 0.5723,
      "step": 6883
    },
    {
      "epoch": 0.871171855226525,
      "grad_norm": 0.8979597091674805,
      "learning_rate": 1.3477768520960632e-05,
      "loss": 0.6053,
      "step": 6884
    },
    {
      "epoch": 0.8712984054669703,
      "grad_norm": 0.8759354948997498,
      "learning_rate": 1.3476088210660135e-05,
      "loss": 0.5691,
      "step": 6885
    },
    {
      "epoch": 0.8714249557074158,
      "grad_norm": 0.8514079451560974,
      "learning_rate": 1.347440778871787e-05,
      "loss": 0.5739,
      "step": 6886
    },
    {
      "epoch": 0.8715515059478613,
      "grad_norm": 0.8438194990158081,
      "learning_rate": 1.347272725518781e-05,
      "loss": 0.5251,
      "step": 6887
    },
    {
      "epoch": 0.8716780561883067,
      "grad_norm": 0.869636058807373,
      "learning_rate": 1.3471046610123928e-05,
      "loss": 0.5933,
      "step": 6888
    },
    {
      "epoch": 0.8718046064287522,
      "grad_norm": 0.8809413313865662,
      "learning_rate": 1.3469365853580201e-05,
      "loss": 0.5797,
      "step": 6889
    },
    {
      "epoch": 0.8719311566691976,
      "grad_norm": 0.8818023204803467,
      "learning_rate": 1.3467684985610608e-05,
      "loss": 0.5965,
      "step": 6890
    },
    {
      "epoch": 0.8720577069096431,
      "grad_norm": 0.911974310874939,
      "learning_rate": 1.3466004006269135e-05,
      "loss": 0.5811,
      "step": 6891
    },
    {
      "epoch": 0.8721842571500886,
      "grad_norm": 0.965225875377655,
      "learning_rate": 1.3464322915609772e-05,
      "loss": 0.6383,
      "step": 6892
    },
    {
      "epoch": 0.872310807390534,
      "grad_norm": 0.9100854396820068,
      "learning_rate": 1.3462641713686508e-05,
      "loss": 0.5885,
      "step": 6893
    },
    {
      "epoch": 0.8724373576309795,
      "grad_norm": 0.9006154537200928,
      "learning_rate": 1.346096040055334e-05,
      "loss": 0.5723,
      "step": 6894
    },
    {
      "epoch": 0.8725639078714249,
      "grad_norm": 0.9389773011207581,
      "learning_rate": 1.3459278976264268e-05,
      "loss": 0.6112,
      "step": 6895
    },
    {
      "epoch": 0.8726904581118704,
      "grad_norm": 0.8757954239845276,
      "learning_rate": 1.345759744087329e-05,
      "loss": 0.5556,
      "step": 6896
    },
    {
      "epoch": 0.8728170083523159,
      "grad_norm": 0.8929532170295715,
      "learning_rate": 1.345591579443441e-05,
      "loss": 0.6291,
      "step": 6897
    },
    {
      "epoch": 0.8729435585927613,
      "grad_norm": 0.8774446249008179,
      "learning_rate": 1.3454234037001647e-05,
      "loss": 0.5935,
      "step": 6898
    },
    {
      "epoch": 0.8730701088332068,
      "grad_norm": 0.9590854644775391,
      "learning_rate": 1.3452552168629009e-05,
      "loss": 0.657,
      "step": 6899
    },
    {
      "epoch": 0.8731966590736523,
      "grad_norm": 0.8627186417579651,
      "learning_rate": 1.3450870189370512e-05,
      "loss": 0.6078,
      "step": 6900
    },
    {
      "epoch": 0.8733232093140977,
      "grad_norm": 0.9658176302909851,
      "learning_rate": 1.344918809928018e-05,
      "loss": 0.5717,
      "step": 6901
    },
    {
      "epoch": 0.8734497595545432,
      "grad_norm": 0.8521448969841003,
      "learning_rate": 1.3447505898412033e-05,
      "loss": 0.5483,
      "step": 6902
    },
    {
      "epoch": 0.8735763097949886,
      "grad_norm": 0.839915931224823,
      "learning_rate": 1.34458235868201e-05,
      "loss": 0.6013,
      "step": 6903
    },
    {
      "epoch": 0.8737028600354341,
      "grad_norm": 0.9174504280090332,
      "learning_rate": 1.3444141164558408e-05,
      "loss": 0.5894,
      "step": 6904
    },
    {
      "epoch": 0.8738294102758796,
      "grad_norm": 0.8846542835235596,
      "learning_rate": 1.3442458631680998e-05,
      "loss": 0.6168,
      "step": 6905
    },
    {
      "epoch": 0.873955960516325,
      "grad_norm": 0.9211921095848083,
      "learning_rate": 1.3440775988241901e-05,
      "loss": 0.6028,
      "step": 6906
    },
    {
      "epoch": 0.8740825107567705,
      "grad_norm": 0.8262376189231873,
      "learning_rate": 1.3439093234295167e-05,
      "loss": 0.5616,
      "step": 6907
    },
    {
      "epoch": 0.8742090609972158,
      "grad_norm": 0.9595574140548706,
      "learning_rate": 1.3437410369894833e-05,
      "loss": 0.5478,
      "step": 6908
    },
    {
      "epoch": 0.8743356112376613,
      "grad_norm": 0.8900925517082214,
      "learning_rate": 1.343572739509495e-05,
      "loss": 0.6261,
      "step": 6909
    },
    {
      "epoch": 0.8744621614781068,
      "grad_norm": 0.854499340057373,
      "learning_rate": 1.3434044309949572e-05,
      "loss": 0.5575,
      "step": 6910
    },
    {
      "epoch": 0.8745887117185522,
      "grad_norm": 0.9344044923782349,
      "learning_rate": 1.3432361114512754e-05,
      "loss": 0.5737,
      "step": 6911
    },
    {
      "epoch": 0.8747152619589977,
      "grad_norm": 0.8973875045776367,
      "learning_rate": 1.343067780883856e-05,
      "loss": 0.5778,
      "step": 6912
    },
    {
      "epoch": 0.8748418121994431,
      "grad_norm": 0.9075842499732971,
      "learning_rate": 1.3428994392981046e-05,
      "loss": 0.5847,
      "step": 6913
    },
    {
      "epoch": 0.8749683624398886,
      "grad_norm": 0.9361830949783325,
      "learning_rate": 1.3427310866994278e-05,
      "loss": 0.6589,
      "step": 6914
    },
    {
      "epoch": 0.8750949126803341,
      "grad_norm": 0.8723381757736206,
      "learning_rate": 1.3425627230932331e-05,
      "loss": 0.5785,
      "step": 6915
    },
    {
      "epoch": 0.8752214629207795,
      "grad_norm": 0.9712938070297241,
      "learning_rate": 1.3423943484849275e-05,
      "loss": 0.5873,
      "step": 6916
    },
    {
      "epoch": 0.875348013161225,
      "grad_norm": 0.8481249809265137,
      "learning_rate": 1.3422259628799187e-05,
      "loss": 0.5827,
      "step": 6917
    },
    {
      "epoch": 0.8754745634016705,
      "grad_norm": 0.9411253333091736,
      "learning_rate": 1.3420575662836153e-05,
      "loss": 0.5881,
      "step": 6918
    },
    {
      "epoch": 0.8756011136421159,
      "grad_norm": 0.9306680560112,
      "learning_rate": 1.341889158701425e-05,
      "loss": 0.5807,
      "step": 6919
    },
    {
      "epoch": 0.8757276638825614,
      "grad_norm": 0.9018140435218811,
      "learning_rate": 1.3417207401387567e-05,
      "loss": 0.5825,
      "step": 6920
    },
    {
      "epoch": 0.8758542141230068,
      "grad_norm": 0.9472603797912598,
      "learning_rate": 1.3415523106010195e-05,
      "loss": 0.5949,
      "step": 6921
    },
    {
      "epoch": 0.8759807643634523,
      "grad_norm": 3.371274709701538,
      "learning_rate": 1.3413838700936231e-05,
      "loss": 0.6497,
      "step": 6922
    },
    {
      "epoch": 0.8761073146038978,
      "grad_norm": 0.8509463667869568,
      "learning_rate": 1.3412154186219773e-05,
      "loss": 0.5704,
      "step": 6923
    },
    {
      "epoch": 0.8762338648443432,
      "grad_norm": 0.9340940117835999,
      "learning_rate": 1.3410469561914923e-05,
      "loss": 0.6351,
      "step": 6924
    },
    {
      "epoch": 0.8763604150847887,
      "grad_norm": 0.9187849164009094,
      "learning_rate": 1.340878482807578e-05,
      "loss": 0.5798,
      "step": 6925
    },
    {
      "epoch": 0.8764869653252341,
      "grad_norm": 1.0131099224090576,
      "learning_rate": 1.3407099984756462e-05,
      "loss": 0.6446,
      "step": 6926
    },
    {
      "epoch": 0.8766135155656796,
      "grad_norm": 0.9402414560317993,
      "learning_rate": 1.3405415032011077e-05,
      "loss": 0.6085,
      "step": 6927
    },
    {
      "epoch": 0.8767400658061251,
      "grad_norm": 0.8618337512016296,
      "learning_rate": 1.3403729969893739e-05,
      "loss": 0.5796,
      "step": 6928
    },
    {
      "epoch": 0.8768666160465705,
      "grad_norm": 0.8856440186500549,
      "learning_rate": 1.3402044798458569e-05,
      "loss": 0.6095,
      "step": 6929
    },
    {
      "epoch": 0.876993166287016,
      "grad_norm": 0.8898863196372986,
      "learning_rate": 1.3400359517759691e-05,
      "loss": 0.5447,
      "step": 6930
    },
    {
      "epoch": 0.8771197165274615,
      "grad_norm": 0.8841405510902405,
      "learning_rate": 1.339867412785123e-05,
      "loss": 0.5613,
      "step": 6931
    },
    {
      "epoch": 0.8772462667679068,
      "grad_norm": 0.9026646018028259,
      "learning_rate": 1.3396988628787313e-05,
      "loss": 0.6118,
      "step": 6932
    },
    {
      "epoch": 0.8773728170083523,
      "grad_norm": 0.8938359022140503,
      "learning_rate": 1.3395303020622076e-05,
      "loss": 0.576,
      "step": 6933
    },
    {
      "epoch": 0.8774993672487977,
      "grad_norm": 0.8731948733329773,
      "learning_rate": 1.3393617303409658e-05,
      "loss": 0.5856,
      "step": 6934
    },
    {
      "epoch": 0.8776259174892432,
      "grad_norm": 0.8989064693450928,
      "learning_rate": 1.3391931477204196e-05,
      "loss": 0.6239,
      "step": 6935
    },
    {
      "epoch": 0.8777524677296887,
      "grad_norm": 0.9122359752655029,
      "learning_rate": 1.3390245542059836e-05,
      "loss": 0.5619,
      "step": 6936
    },
    {
      "epoch": 0.8778790179701341,
      "grad_norm": 0.8526920676231384,
      "learning_rate": 1.3388559498030725e-05,
      "loss": 0.5621,
      "step": 6937
    },
    {
      "epoch": 0.8780055682105796,
      "grad_norm": 0.892700731754303,
      "learning_rate": 1.338687334517101e-05,
      "loss": 0.5972,
      "step": 6938
    },
    {
      "epoch": 0.878132118451025,
      "grad_norm": 0.9504854083061218,
      "learning_rate": 1.338518708353485e-05,
      "loss": 0.5421,
      "step": 6939
    },
    {
      "epoch": 0.8782586686914705,
      "grad_norm": 0.8618983626365662,
      "learning_rate": 1.33835007131764e-05,
      "loss": 0.57,
      "step": 6940
    },
    {
      "epoch": 0.878385218931916,
      "grad_norm": 0.9318660497665405,
      "learning_rate": 1.3381814234149825e-05,
      "loss": 0.601,
      "step": 6941
    },
    {
      "epoch": 0.8785117691723614,
      "grad_norm": 0.8806367516517639,
      "learning_rate": 1.3380127646509284e-05,
      "loss": 0.6207,
      "step": 6942
    },
    {
      "epoch": 0.8786383194128069,
      "grad_norm": 0.8976816534996033,
      "learning_rate": 1.3378440950308949e-05,
      "loss": 0.6146,
      "step": 6943
    },
    {
      "epoch": 0.8787648696532523,
      "grad_norm": 0.9295952320098877,
      "learning_rate": 1.3376754145602993e-05,
      "loss": 0.6138,
      "step": 6944
    },
    {
      "epoch": 0.8788914198936978,
      "grad_norm": 0.936373770236969,
      "learning_rate": 1.3375067232445587e-05,
      "loss": 0.5745,
      "step": 6945
    },
    {
      "epoch": 0.8790179701341433,
      "grad_norm": 0.9395208358764648,
      "learning_rate": 1.3373380210890912e-05,
      "loss": 0.5962,
      "step": 6946
    },
    {
      "epoch": 0.8791445203745887,
      "grad_norm": 0.9344805479049683,
      "learning_rate": 1.3371693080993152e-05,
      "loss": 0.6312,
      "step": 6947
    },
    {
      "epoch": 0.8792710706150342,
      "grad_norm": 0.8605906367301941,
      "learning_rate": 1.3370005842806488e-05,
      "loss": 0.6276,
      "step": 6948
    },
    {
      "epoch": 0.8793976208554797,
      "grad_norm": 0.9032329320907593,
      "learning_rate": 1.3368318496385113e-05,
      "loss": 0.6046,
      "step": 6949
    },
    {
      "epoch": 0.8795241710959251,
      "grad_norm": 0.9584810137748718,
      "learning_rate": 1.336663104178322e-05,
      "loss": 0.63,
      "step": 6950
    },
    {
      "epoch": 0.8796507213363706,
      "grad_norm": 0.8707498908042908,
      "learning_rate": 1.3364943479055003e-05,
      "loss": 0.5743,
      "step": 6951
    },
    {
      "epoch": 0.879777271576816,
      "grad_norm": 0.9281973242759705,
      "learning_rate": 1.3363255808254664e-05,
      "loss": 0.5785,
      "step": 6952
    },
    {
      "epoch": 0.8799038218172615,
      "grad_norm": 0.9249032735824585,
      "learning_rate": 1.3361568029436404e-05,
      "loss": 0.6016,
      "step": 6953
    },
    {
      "epoch": 0.880030372057707,
      "grad_norm": 0.9458953738212585,
      "learning_rate": 1.3359880142654429e-05,
      "loss": 0.5981,
      "step": 6954
    },
    {
      "epoch": 0.8801569222981523,
      "grad_norm": 1.0022562742233276,
      "learning_rate": 1.335819214796295e-05,
      "loss": 0.6182,
      "step": 6955
    },
    {
      "epoch": 0.8802834725385978,
      "grad_norm": 0.8881837129592896,
      "learning_rate": 1.3356504045416176e-05,
      "loss": 0.6055,
      "step": 6956
    },
    {
      "epoch": 0.8804100227790432,
      "grad_norm": 0.879956841468811,
      "learning_rate": 1.3354815835068333e-05,
      "loss": 0.5915,
      "step": 6957
    },
    {
      "epoch": 0.8805365730194887,
      "grad_norm": 0.9036775827407837,
      "learning_rate": 1.3353127516973633e-05,
      "loss": 0.589,
      "step": 6958
    },
    {
      "epoch": 0.8806631232599342,
      "grad_norm": 0.8960497975349426,
      "learning_rate": 1.335143909118631e-05,
      "loss": 0.5659,
      "step": 6959
    },
    {
      "epoch": 0.8807896735003796,
      "grad_norm": 0.9235580563545227,
      "learning_rate": 1.334975055776058e-05,
      "loss": 0.6085,
      "step": 6960
    },
    {
      "epoch": 0.8809162237408251,
      "grad_norm": 0.888612687587738,
      "learning_rate": 1.3348061916750676e-05,
      "loss": 0.6411,
      "step": 6961
    },
    {
      "epoch": 0.8810427739812705,
      "grad_norm": 1.0372538566589355,
      "learning_rate": 1.3346373168210837e-05,
      "loss": 0.6864,
      "step": 6962
    },
    {
      "epoch": 0.881169324221716,
      "grad_norm": 0.8823118805885315,
      "learning_rate": 1.33446843121953e-05,
      "loss": 0.5917,
      "step": 6963
    },
    {
      "epoch": 0.8812958744621615,
      "grad_norm": 0.906293511390686,
      "learning_rate": 1.3342995348758303e-05,
      "loss": 0.5881,
      "step": 6964
    },
    {
      "epoch": 0.8814224247026069,
      "grad_norm": 0.9038301110267639,
      "learning_rate": 1.3341306277954094e-05,
      "loss": 0.5481,
      "step": 6965
    },
    {
      "epoch": 0.8815489749430524,
      "grad_norm": 0.8643226027488708,
      "learning_rate": 1.3339617099836917e-05,
      "loss": 0.5444,
      "step": 6966
    },
    {
      "epoch": 0.8816755251834979,
      "grad_norm": 0.9167434573173523,
      "learning_rate": 1.3337927814461024e-05,
      "loss": 0.5751,
      "step": 6967
    },
    {
      "epoch": 0.8818020754239433,
      "grad_norm": 0.9468004107475281,
      "learning_rate": 1.3336238421880672e-05,
      "loss": 0.632,
      "step": 6968
    },
    {
      "epoch": 0.8819286256643888,
      "grad_norm": 1.036281943321228,
      "learning_rate": 1.3334548922150118e-05,
      "loss": 0.5597,
      "step": 6969
    },
    {
      "epoch": 0.8820551759048342,
      "grad_norm": 0.933056116104126,
      "learning_rate": 1.3332859315323627e-05,
      "loss": 0.6067,
      "step": 6970
    },
    {
      "epoch": 0.8821817261452797,
      "grad_norm": 0.8724732995033264,
      "learning_rate": 1.333116960145546e-05,
      "loss": 0.6081,
      "step": 6971
    },
    {
      "epoch": 0.8823082763857252,
      "grad_norm": 0.8720096349716187,
      "learning_rate": 1.3329479780599885e-05,
      "loss": 0.6043,
      "step": 6972
    },
    {
      "epoch": 0.8824348266261706,
      "grad_norm": 0.8813140392303467,
      "learning_rate": 1.3327789852811179e-05,
      "loss": 0.5878,
      "step": 6973
    },
    {
      "epoch": 0.8825613768666161,
      "grad_norm": 0.8896889686584473,
      "learning_rate": 1.3326099818143616e-05,
      "loss": 0.5881,
      "step": 6974
    },
    {
      "epoch": 0.8826879271070615,
      "grad_norm": 0.8804824948310852,
      "learning_rate": 1.332440967665147e-05,
      "loss": 0.5493,
      "step": 6975
    },
    {
      "epoch": 0.882814477347507,
      "grad_norm": 0.9095078706741333,
      "learning_rate": 1.3322719428389033e-05,
      "loss": 0.6498,
      "step": 6976
    },
    {
      "epoch": 0.8829410275879525,
      "grad_norm": 0.9055818915367126,
      "learning_rate": 1.3321029073410583e-05,
      "loss": 0.5931,
      "step": 6977
    },
    {
      "epoch": 0.8830675778283978,
      "grad_norm": 0.9405645728111267,
      "learning_rate": 1.3319338611770409e-05,
      "loss": 0.6034,
      "step": 6978
    },
    {
      "epoch": 0.8831941280688433,
      "grad_norm": 1.1085094213485718,
      "learning_rate": 1.3317648043522807e-05,
      "loss": 0.6098,
      "step": 6979
    },
    {
      "epoch": 0.8833206783092888,
      "grad_norm": 0.912726640701294,
      "learning_rate": 1.331595736872207e-05,
      "loss": 0.5915,
      "step": 6980
    },
    {
      "epoch": 0.8834472285497342,
      "grad_norm": 1.0023043155670166,
      "learning_rate": 1.3314266587422501e-05,
      "loss": 0.5831,
      "step": 6981
    },
    {
      "epoch": 0.8835737787901797,
      "grad_norm": 0.9022034406661987,
      "learning_rate": 1.3312575699678405e-05,
      "loss": 0.558,
      "step": 6982
    },
    {
      "epoch": 0.8837003290306251,
      "grad_norm": 0.9010526537895203,
      "learning_rate": 1.3310884705544082e-05,
      "loss": 0.6382,
      "step": 6983
    },
    {
      "epoch": 0.8838268792710706,
      "grad_norm": 0.9393845796585083,
      "learning_rate": 1.3309193605073844e-05,
      "loss": 0.6133,
      "step": 6984
    },
    {
      "epoch": 0.8839534295115161,
      "grad_norm": 0.8661372661590576,
      "learning_rate": 1.3307502398322005e-05,
      "loss": 0.5883,
      "step": 6985
    },
    {
      "epoch": 0.8840799797519615,
      "grad_norm": 0.8806837797164917,
      "learning_rate": 1.3305811085342882e-05,
      "loss": 0.555,
      "step": 6986
    },
    {
      "epoch": 0.884206529992407,
      "grad_norm": 0.9142630696296692,
      "learning_rate": 1.3304119666190795e-05,
      "loss": 0.5601,
      "step": 6987
    },
    {
      "epoch": 0.8843330802328524,
      "grad_norm": 1.0132005214691162,
      "learning_rate": 1.3302428140920067e-05,
      "loss": 0.618,
      "step": 6988
    },
    {
      "epoch": 0.8844596304732979,
      "grad_norm": 0.8400798439979553,
      "learning_rate": 1.3300736509585024e-05,
      "loss": 0.5801,
      "step": 6989
    },
    {
      "epoch": 0.8845861807137434,
      "grad_norm": 0.8414878249168396,
      "learning_rate": 1.329904477224e-05,
      "loss": 0.5481,
      "step": 6990
    },
    {
      "epoch": 0.8847127309541888,
      "grad_norm": 0.8505609631538391,
      "learning_rate": 1.3297352928939322e-05,
      "loss": 0.5634,
      "step": 6991
    },
    {
      "epoch": 0.8848392811946343,
      "grad_norm": 0.8871234059333801,
      "learning_rate": 1.329566097973733e-05,
      "loss": 0.5876,
      "step": 6992
    },
    {
      "epoch": 0.8849658314350797,
      "grad_norm": 0.8716427683830261,
      "learning_rate": 1.3293968924688368e-05,
      "loss": 0.6134,
      "step": 6993
    },
    {
      "epoch": 0.8850923816755252,
      "grad_norm": 0.9150274395942688,
      "learning_rate": 1.3292276763846776e-05,
      "loss": 0.5761,
      "step": 6994
    },
    {
      "epoch": 0.8852189319159707,
      "grad_norm": 0.8536589741706848,
      "learning_rate": 1.3290584497266901e-05,
      "loss": 0.5468,
      "step": 6995
    },
    {
      "epoch": 0.8853454821564161,
      "grad_norm": 0.926173746585846,
      "learning_rate": 1.3288892125003097e-05,
      "loss": 0.5853,
      "step": 6996
    },
    {
      "epoch": 0.8854720323968616,
      "grad_norm": 0.9393620491027832,
      "learning_rate": 1.3287199647109712e-05,
      "loss": 0.5908,
      "step": 6997
    },
    {
      "epoch": 0.8855985826373071,
      "grad_norm": 0.8517267107963562,
      "learning_rate": 1.328550706364111e-05,
      "loss": 0.5452,
      "step": 6998
    },
    {
      "epoch": 0.8857251328777525,
      "grad_norm": 0.8782621026039124,
      "learning_rate": 1.3283814374651651e-05,
      "loss": 0.5586,
      "step": 6999
    },
    {
      "epoch": 0.885851683118198,
      "grad_norm": 0.8844263553619385,
      "learning_rate": 1.3282121580195695e-05,
      "loss": 0.5864,
      "step": 7000
    },
    {
      "epoch": 0.8859782333586433,
      "grad_norm": 0.9874921441078186,
      "learning_rate": 1.3280428680327613e-05,
      "loss": 0.6188,
      "step": 7001
    },
    {
      "epoch": 0.8861047835990888,
      "grad_norm": 0.9243314266204834,
      "learning_rate": 1.3278735675101776e-05,
      "loss": 0.6053,
      "step": 7002
    },
    {
      "epoch": 0.8862313338395343,
      "grad_norm": 0.9977458119392395,
      "learning_rate": 1.3277042564572555e-05,
      "loss": 0.5937,
      "step": 7003
    },
    {
      "epoch": 0.8863578840799797,
      "grad_norm": 0.9828252196311951,
      "learning_rate": 1.3275349348794331e-05,
      "loss": 0.6117,
      "step": 7004
    },
    {
      "epoch": 0.8864844343204252,
      "grad_norm": 0.8953370451927185,
      "learning_rate": 1.3273656027821486e-05,
      "loss": 0.5997,
      "step": 7005
    },
    {
      "epoch": 0.8866109845608706,
      "grad_norm": 0.8930144906044006,
      "learning_rate": 1.32719626017084e-05,
      "loss": 0.5739,
      "step": 7006
    },
    {
      "epoch": 0.8867375348013161,
      "grad_norm": 0.946709394454956,
      "learning_rate": 1.3270269070509463e-05,
      "loss": 0.5973,
      "step": 7007
    },
    {
      "epoch": 0.8868640850417616,
      "grad_norm": 0.9389736652374268,
      "learning_rate": 1.3268575434279068e-05,
      "loss": 0.5673,
      "step": 7008
    },
    {
      "epoch": 0.886990635282207,
      "grad_norm": 0.8613573908805847,
      "learning_rate": 1.3266881693071609e-05,
      "loss": 0.5397,
      "step": 7009
    },
    {
      "epoch": 0.8871171855226525,
      "grad_norm": 1.110939383506775,
      "learning_rate": 1.3265187846941485e-05,
      "loss": 0.5991,
      "step": 7010
    },
    {
      "epoch": 0.8872437357630979,
      "grad_norm": 0.8764998316764832,
      "learning_rate": 1.3263493895943093e-05,
      "loss": 0.5675,
      "step": 7011
    },
    {
      "epoch": 0.8873702860035434,
      "grad_norm": 0.9009690284729004,
      "learning_rate": 1.3261799840130843e-05,
      "loss": 0.5504,
      "step": 7012
    },
    {
      "epoch": 0.8874968362439889,
      "grad_norm": 0.8946569561958313,
      "learning_rate": 1.3260105679559137e-05,
      "loss": 0.6016,
      "step": 7013
    },
    {
      "epoch": 0.8876233864844343,
      "grad_norm": 0.92336106300354,
      "learning_rate": 1.3258411414282395e-05,
      "loss": 0.6114,
      "step": 7014
    },
    {
      "epoch": 0.8877499367248798,
      "grad_norm": 0.9179095029830933,
      "learning_rate": 1.3256717044355024e-05,
      "loss": 0.5766,
      "step": 7015
    },
    {
      "epoch": 0.8878764869653253,
      "grad_norm": 0.9988757371902466,
      "learning_rate": 1.3255022569831449e-05,
      "loss": 0.6033,
      "step": 7016
    },
    {
      "epoch": 0.8880030372057707,
      "grad_norm": 0.8900525569915771,
      "learning_rate": 1.3253327990766083e-05,
      "loss": 0.6017,
      "step": 7017
    },
    {
      "epoch": 0.8881295874462162,
      "grad_norm": 0.9486972689628601,
      "learning_rate": 1.3251633307213357e-05,
      "loss": 0.6242,
      "step": 7018
    },
    {
      "epoch": 0.8882561376866616,
      "grad_norm": 0.9246190786361694,
      "learning_rate": 1.3249938519227699e-05,
      "loss": 0.5605,
      "step": 7019
    },
    {
      "epoch": 0.8883826879271071,
      "grad_norm": 0.9730463027954102,
      "learning_rate": 1.3248243626863537e-05,
      "loss": 0.591,
      "step": 7020
    },
    {
      "epoch": 0.8885092381675526,
      "grad_norm": 0.9320659041404724,
      "learning_rate": 1.3246548630175308e-05,
      "loss": 0.6313,
      "step": 7021
    },
    {
      "epoch": 0.888635788407998,
      "grad_norm": 0.9263740181922913,
      "learning_rate": 1.3244853529217457e-05,
      "loss": 0.5962,
      "step": 7022
    },
    {
      "epoch": 0.8887623386484435,
      "grad_norm": 0.9537425637245178,
      "learning_rate": 1.3243158324044415e-05,
      "loss": 0.6106,
      "step": 7023
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.9392647743225098,
      "learning_rate": 1.3241463014710632e-05,
      "loss": 0.5822,
      "step": 7024
    },
    {
      "epoch": 0.8890154391293343,
      "grad_norm": 0.9033863544464111,
      "learning_rate": 1.3239767601270557e-05,
      "loss": 0.569,
      "step": 7025
    },
    {
      "epoch": 0.8891419893697798,
      "grad_norm": 0.870108962059021,
      "learning_rate": 1.323807208377864e-05,
      "loss": 0.5704,
      "step": 7026
    },
    {
      "epoch": 0.8892685396102252,
      "grad_norm": 0.9220842123031616,
      "learning_rate": 1.3236376462289338e-05,
      "loss": 0.6363,
      "step": 7027
    },
    {
      "epoch": 0.8893950898506707,
      "grad_norm": 0.9733474254608154,
      "learning_rate": 1.3234680736857106e-05,
      "loss": 0.55,
      "step": 7028
    },
    {
      "epoch": 0.8895216400911162,
      "grad_norm": 0.8777604103088379,
      "learning_rate": 1.323298490753641e-05,
      "loss": 0.5691,
      "step": 7029
    },
    {
      "epoch": 0.8896481903315616,
      "grad_norm": 0.8875998258590698,
      "learning_rate": 1.323128897438171e-05,
      "loss": 0.5624,
      "step": 7030
    },
    {
      "epoch": 0.8897747405720071,
      "grad_norm": 0.8837996125221252,
      "learning_rate": 1.322959293744748e-05,
      "loss": 0.576,
      "step": 7031
    },
    {
      "epoch": 0.8899012908124525,
      "grad_norm": 0.8948081731796265,
      "learning_rate": 1.3227896796788186e-05,
      "loss": 0.6037,
      "step": 7032
    },
    {
      "epoch": 0.890027841052898,
      "grad_norm": 0.9175015687942505,
      "learning_rate": 1.3226200552458307e-05,
      "loss": 0.5763,
      "step": 7033
    },
    {
      "epoch": 0.8901543912933435,
      "grad_norm": 0.9139240980148315,
      "learning_rate": 1.3224504204512325e-05,
      "loss": 0.5814,
      "step": 7034
    },
    {
      "epoch": 0.8902809415337889,
      "grad_norm": 0.9329380393028259,
      "learning_rate": 1.3222807753004712e-05,
      "loss": 0.6355,
      "step": 7035
    },
    {
      "epoch": 0.8904074917742344,
      "grad_norm": 0.9937695860862732,
      "learning_rate": 1.322111119798996e-05,
      "loss": 0.6102,
      "step": 7036
    },
    {
      "epoch": 0.8905340420146798,
      "grad_norm": 0.8977766633033752,
      "learning_rate": 1.3219414539522558e-05,
      "loss": 0.6123,
      "step": 7037
    },
    {
      "epoch": 0.8906605922551253,
      "grad_norm": 0.8977183103561401,
      "learning_rate": 1.3217717777656994e-05,
      "loss": 0.5827,
      "step": 7038
    },
    {
      "epoch": 0.8907871424955708,
      "grad_norm": 0.9108182191848755,
      "learning_rate": 1.3216020912447762e-05,
      "loss": 0.5902,
      "step": 7039
    },
    {
      "epoch": 0.8909136927360162,
      "grad_norm": 0.9141498804092407,
      "learning_rate": 1.3214323943949366e-05,
      "loss": 0.6045,
      "step": 7040
    },
    {
      "epoch": 0.8910402429764617,
      "grad_norm": 0.9278892874717712,
      "learning_rate": 1.3212626872216301e-05,
      "loss": 0.6143,
      "step": 7041
    },
    {
      "epoch": 0.8911667932169071,
      "grad_norm": 0.9022103548049927,
      "learning_rate": 1.3210929697303077e-05,
      "loss": 0.5734,
      "step": 7042
    },
    {
      "epoch": 0.8912933434573526,
      "grad_norm": 0.881424605846405,
      "learning_rate": 1.32092324192642e-05,
      "loss": 0.5768,
      "step": 7043
    },
    {
      "epoch": 0.8914198936977981,
      "grad_norm": 0.8540716767311096,
      "learning_rate": 1.3207535038154182e-05,
      "loss": 0.5493,
      "step": 7044
    },
    {
      "epoch": 0.8915464439382434,
      "grad_norm": 0.8945510387420654,
      "learning_rate": 1.320583755402754e-05,
      "loss": 0.5734,
      "step": 7045
    },
    {
      "epoch": 0.891672994178689,
      "grad_norm": 0.8714374303817749,
      "learning_rate": 1.3204139966938788e-05,
      "loss": 0.562,
      "step": 7046
    },
    {
      "epoch": 0.8917995444191344,
      "grad_norm": 0.8839296102523804,
      "learning_rate": 1.3202442276942453e-05,
      "loss": 0.5743,
      "step": 7047
    },
    {
      "epoch": 0.8919260946595798,
      "grad_norm": 0.9477389454841614,
      "learning_rate": 1.3200744484093054e-05,
      "loss": 0.6056,
      "step": 7048
    },
    {
      "epoch": 0.8920526449000253,
      "grad_norm": 0.9175111055374146,
      "learning_rate": 1.3199046588445121e-05,
      "loss": 0.5655,
      "step": 7049
    },
    {
      "epoch": 0.8921791951404707,
      "grad_norm": 0.8856128454208374,
      "learning_rate": 1.319734859005319e-05,
      "loss": 0.5842,
      "step": 7050
    },
    {
      "epoch": 0.8923057453809162,
      "grad_norm": 0.8671517372131348,
      "learning_rate": 1.319565048897179e-05,
      "loss": 0.5761,
      "step": 7051
    },
    {
      "epoch": 0.8924322956213617,
      "grad_norm": 0.8799173831939697,
      "learning_rate": 1.3193952285255463e-05,
      "loss": 0.5987,
      "step": 7052
    },
    {
      "epoch": 0.8925588458618071,
      "grad_norm": 0.9036600589752197,
      "learning_rate": 1.3192253978958744e-05,
      "loss": 0.5888,
      "step": 7053
    },
    {
      "epoch": 0.8926853961022526,
      "grad_norm": 0.8887011408805847,
      "learning_rate": 1.3190555570136185e-05,
      "loss": 0.6001,
      "step": 7054
    },
    {
      "epoch": 0.892811946342698,
      "grad_norm": 0.9727654457092285,
      "learning_rate": 1.3188857058842331e-05,
      "loss": 0.5951,
      "step": 7055
    },
    {
      "epoch": 0.8929384965831435,
      "grad_norm": 0.9735031127929688,
      "learning_rate": 1.3187158445131731e-05,
      "loss": 0.6097,
      "step": 7056
    },
    {
      "epoch": 0.893065046823589,
      "grad_norm": 0.8912603259086609,
      "learning_rate": 1.3185459729058944e-05,
      "loss": 0.5513,
      "step": 7057
    },
    {
      "epoch": 0.8931915970640344,
      "grad_norm": 0.8848540782928467,
      "learning_rate": 1.3183760910678525e-05,
      "loss": 0.6014,
      "step": 7058
    },
    {
      "epoch": 0.8933181473044799,
      "grad_norm": 0.9405155777931213,
      "learning_rate": 1.3182061990045032e-05,
      "loss": 0.5668,
      "step": 7059
    },
    {
      "epoch": 0.8934446975449253,
      "grad_norm": 0.8952250480651855,
      "learning_rate": 1.3180362967213036e-05,
      "loss": 0.5704,
      "step": 7060
    },
    {
      "epoch": 0.8935712477853708,
      "grad_norm": 0.8598820567131042,
      "learning_rate": 1.3178663842237101e-05,
      "loss": 0.5372,
      "step": 7061
    },
    {
      "epoch": 0.8936977980258163,
      "grad_norm": 0.9365651607513428,
      "learning_rate": 1.3176964615171799e-05,
      "loss": 0.6186,
      "step": 7062
    },
    {
      "epoch": 0.8938243482662617,
      "grad_norm": 0.8341464400291443,
      "learning_rate": 1.3175265286071702e-05,
      "loss": 0.5298,
      "step": 7063
    },
    {
      "epoch": 0.8939508985067072,
      "grad_norm": 1.0132391452789307,
      "learning_rate": 1.3173565854991392e-05,
      "loss": 0.6134,
      "step": 7064
    },
    {
      "epoch": 0.8940774487471527,
      "grad_norm": 0.8730557560920715,
      "learning_rate": 1.3171866321985443e-05,
      "loss": 0.6031,
      "step": 7065
    },
    {
      "epoch": 0.8942039989875981,
      "grad_norm": 0.9077618718147278,
      "learning_rate": 1.3170166687108443e-05,
      "loss": 0.6121,
      "step": 7066
    },
    {
      "epoch": 0.8943305492280436,
      "grad_norm": 0.8992863893508911,
      "learning_rate": 1.3168466950414982e-05,
      "loss": 0.5674,
      "step": 7067
    },
    {
      "epoch": 0.894457099468489,
      "grad_norm": 0.8536291122436523,
      "learning_rate": 1.3166767111959647e-05,
      "loss": 0.5852,
      "step": 7068
    },
    {
      "epoch": 0.8945836497089344,
      "grad_norm": 0.88123619556427,
      "learning_rate": 1.316506717179703e-05,
      "loss": 0.5807,
      "step": 7069
    },
    {
      "epoch": 0.89471019994938,
      "grad_norm": 0.865717887878418,
      "learning_rate": 1.3163367129981732e-05,
      "loss": 0.5777,
      "step": 7070
    },
    {
      "epoch": 0.8948367501898253,
      "grad_norm": 0.9106322526931763,
      "learning_rate": 1.316166698656835e-05,
      "loss": 0.6394,
      "step": 7071
    },
    {
      "epoch": 0.8949633004302708,
      "grad_norm": 0.8796997666358948,
      "learning_rate": 1.3159966741611492e-05,
      "loss": 0.62,
      "step": 7072
    },
    {
      "epoch": 0.8950898506707162,
      "grad_norm": 0.8440794348716736,
      "learning_rate": 1.3158266395165762e-05,
      "loss": 0.5738,
      "step": 7073
    },
    {
      "epoch": 0.8952164009111617,
      "grad_norm": 0.8987999558448792,
      "learning_rate": 1.3156565947285773e-05,
      "loss": 0.5832,
      "step": 7074
    },
    {
      "epoch": 0.8953429511516072,
      "grad_norm": 0.95954829454422,
      "learning_rate": 1.3154865398026131e-05,
      "loss": 0.6363,
      "step": 7075
    },
    {
      "epoch": 0.8954695013920526,
      "grad_norm": 0.9219143390655518,
      "learning_rate": 1.3153164747441462e-05,
      "loss": 0.6047,
      "step": 7076
    },
    {
      "epoch": 0.8955960516324981,
      "grad_norm": 0.9356088042259216,
      "learning_rate": 1.315146399558638e-05,
      "loss": 0.5976,
      "step": 7077
    },
    {
      "epoch": 0.8957226018729436,
      "grad_norm": 1.006813883781433,
      "learning_rate": 1.3149763142515509e-05,
      "loss": 0.5926,
      "step": 7078
    },
    {
      "epoch": 0.895849152113389,
      "grad_norm": 0.9240928292274475,
      "learning_rate": 1.3148062188283474e-05,
      "loss": 0.5692,
      "step": 7079
    },
    {
      "epoch": 0.8959757023538345,
      "grad_norm": 0.8926593661308289,
      "learning_rate": 1.3146361132944911e-05,
      "loss": 0.5689,
      "step": 7080
    },
    {
      "epoch": 0.8961022525942799,
      "grad_norm": 1.0029243230819702,
      "learning_rate": 1.3144659976554448e-05,
      "loss": 0.6146,
      "step": 7081
    },
    {
      "epoch": 0.8962288028347254,
      "grad_norm": 0.8569334745407104,
      "learning_rate": 1.3142958719166718e-05,
      "loss": 0.5427,
      "step": 7082
    },
    {
      "epoch": 0.8963553530751709,
      "grad_norm": 0.8978208899497986,
      "learning_rate": 1.3141257360836367e-05,
      "loss": 0.5707,
      "step": 7083
    },
    {
      "epoch": 0.8964819033156163,
      "grad_norm": 0.9711762070655823,
      "learning_rate": 1.3139555901618035e-05,
      "loss": 0.5926,
      "step": 7084
    },
    {
      "epoch": 0.8966084535560618,
      "grad_norm": 0.9115219712257385,
      "learning_rate": 1.3137854341566369e-05,
      "loss": 0.6036,
      "step": 7085
    },
    {
      "epoch": 0.8967350037965072,
      "grad_norm": 0.8900178074836731,
      "learning_rate": 1.3136152680736016e-05,
      "loss": 0.6017,
      "step": 7086
    },
    {
      "epoch": 0.8968615540369527,
      "grad_norm": 7.356264114379883,
      "learning_rate": 1.3134450919181627e-05,
      "loss": 0.5952,
      "step": 7087
    },
    {
      "epoch": 0.8969881042773982,
      "grad_norm": 0.8709291815757751,
      "learning_rate": 1.3132749056957863e-05,
      "loss": 0.568,
      "step": 7088
    },
    {
      "epoch": 0.8971146545178436,
      "grad_norm": 0.966749906539917,
      "learning_rate": 1.3131047094119378e-05,
      "loss": 0.5965,
      "step": 7089
    },
    {
      "epoch": 0.8972412047582891,
      "grad_norm": 0.9349532127380371,
      "learning_rate": 1.3129345030720834e-05,
      "loss": 0.5732,
      "step": 7090
    },
    {
      "epoch": 0.8973677549987344,
      "grad_norm": 0.8970251083374023,
      "learning_rate": 1.3127642866816904e-05,
      "loss": 0.5839,
      "step": 7091
    },
    {
      "epoch": 0.89749430523918,
      "grad_norm": 0.9072021245956421,
      "learning_rate": 1.3125940602462247e-05,
      "loss": 0.6224,
      "step": 7092
    },
    {
      "epoch": 0.8976208554796254,
      "grad_norm": 0.9072549343109131,
      "learning_rate": 1.3124238237711536e-05,
      "loss": 0.5698,
      "step": 7093
    },
    {
      "epoch": 0.8977474057200708,
      "grad_norm": 0.9013631343841553,
      "learning_rate": 1.3122535772619452e-05,
      "loss": 0.5814,
      "step": 7094
    },
    {
      "epoch": 0.8978739559605163,
      "grad_norm": 0.8589306473731995,
      "learning_rate": 1.3120833207240665e-05,
      "loss": 0.549,
      "step": 7095
    },
    {
      "epoch": 0.8980005062009618,
      "grad_norm": 0.8765511512756348,
      "learning_rate": 1.3119130541629863e-05,
      "loss": 0.5825,
      "step": 7096
    },
    {
      "epoch": 0.8981270564414072,
      "grad_norm": 0.9225287437438965,
      "learning_rate": 1.3117427775841732e-05,
      "loss": 0.5935,
      "step": 7097
    },
    {
      "epoch": 0.8982536066818527,
      "grad_norm": 0.880448043346405,
      "learning_rate": 1.3115724909930952e-05,
      "loss": 0.5605,
      "step": 7098
    },
    {
      "epoch": 0.8983801569222981,
      "grad_norm": 0.8609302043914795,
      "learning_rate": 1.311402194395222e-05,
      "loss": 0.5785,
      "step": 7099
    },
    {
      "epoch": 0.8985067071627436,
      "grad_norm": 0.8964793086051941,
      "learning_rate": 1.3112318877960228e-05,
      "loss": 0.5792,
      "step": 7100
    },
    {
      "epoch": 0.8986332574031891,
      "grad_norm": 0.8225686550140381,
      "learning_rate": 1.3110615712009675e-05,
      "loss": 0.6171,
      "step": 7101
    },
    {
      "epoch": 0.8987598076436345,
      "grad_norm": 0.925268828868866,
      "learning_rate": 1.3108912446155259e-05,
      "loss": 0.593,
      "step": 7102
    },
    {
      "epoch": 0.89888635788408,
      "grad_norm": 0.8330813646316528,
      "learning_rate": 1.3107209080451691e-05,
      "loss": 0.578,
      "step": 7103
    },
    {
      "epoch": 0.8990129081245254,
      "grad_norm": 0.9429606199264526,
      "learning_rate": 1.3105505614953669e-05,
      "loss": 0.6031,
      "step": 7104
    },
    {
      "epoch": 0.8991394583649709,
      "grad_norm": 0.9213603734970093,
      "learning_rate": 1.3103802049715906e-05,
      "loss": 0.6446,
      "step": 7105
    },
    {
      "epoch": 0.8992660086054164,
      "grad_norm": 0.8906048536300659,
      "learning_rate": 1.3102098384793118e-05,
      "loss": 0.6053,
      "step": 7106
    },
    {
      "epoch": 0.8993925588458618,
      "grad_norm": 0.9643265008926392,
      "learning_rate": 1.310039462024002e-05,
      "loss": 0.5934,
      "step": 7107
    },
    {
      "epoch": 0.8995191090863073,
      "grad_norm": 0.8995770812034607,
      "learning_rate": 1.3098690756111334e-05,
      "loss": 0.5885,
      "step": 7108
    },
    {
      "epoch": 0.8996456593267527,
      "grad_norm": 0.8826399445533752,
      "learning_rate": 1.309698679246178e-05,
      "loss": 0.5878,
      "step": 7109
    },
    {
      "epoch": 0.8997722095671982,
      "grad_norm": 0.9398763179779053,
      "learning_rate": 1.3095282729346087e-05,
      "loss": 0.6139,
      "step": 7110
    },
    {
      "epoch": 0.8998987598076437,
      "grad_norm": 0.9306251406669617,
      "learning_rate": 1.309357856681898e-05,
      "loss": 0.591,
      "step": 7111
    },
    {
      "epoch": 0.9000253100480891,
      "grad_norm": 0.8772594928741455,
      "learning_rate": 1.3091874304935201e-05,
      "loss": 0.6029,
      "step": 7112
    },
    {
      "epoch": 0.9001518602885346,
      "grad_norm": 0.8605245351791382,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.5801,
      "step": 7113
    },
    {
      "epoch": 0.90027841052898,
      "grad_norm": 0.8117199540138245,
      "learning_rate": 1.308846548331655e-05,
      "loss": 0.5744,
      "step": 7114
    },
    {
      "epoch": 0.9004049607694254,
      "grad_norm": 0.9047157764434814,
      "learning_rate": 1.3086760923691162e-05,
      "loss": 0.5923,
      "step": 7115
    },
    {
      "epoch": 0.9005315110098709,
      "grad_norm": 0.8982356190681458,
      "learning_rate": 1.3085056264928058e-05,
      "loss": 0.5753,
      "step": 7116
    },
    {
      "epoch": 0.9006580612503163,
      "grad_norm": 0.9118571281433105,
      "learning_rate": 1.3083351507081988e-05,
      "loss": 0.5742,
      "step": 7117
    },
    {
      "epoch": 0.9007846114907618,
      "grad_norm": 0.8941091895103455,
      "learning_rate": 1.3081646650207702e-05,
      "loss": 0.5733,
      "step": 7118
    },
    {
      "epoch": 0.9009111617312073,
      "grad_norm": 0.885737419128418,
      "learning_rate": 1.3079941694359957e-05,
      "loss": 0.5978,
      "step": 7119
    },
    {
      "epoch": 0.9010377119716527,
      "grad_norm": 0.8960573077201843,
      "learning_rate": 1.3078236639593513e-05,
      "loss": 0.555,
      "step": 7120
    },
    {
      "epoch": 0.9011642622120982,
      "grad_norm": 0.8773680925369263,
      "learning_rate": 1.3076531485963126e-05,
      "loss": 0.6003,
      "step": 7121
    },
    {
      "epoch": 0.9012908124525436,
      "grad_norm": 0.9479233026504517,
      "learning_rate": 1.3074826233523563e-05,
      "loss": 0.5674,
      "step": 7122
    },
    {
      "epoch": 0.9014173626929891,
      "grad_norm": 0.9092199206352234,
      "learning_rate": 1.3073120882329593e-05,
      "loss": 0.6042,
      "step": 7123
    },
    {
      "epoch": 0.9015439129334346,
      "grad_norm": 0.9335364699363708,
      "learning_rate": 1.3071415432435987e-05,
      "loss": 0.6131,
      "step": 7124
    },
    {
      "epoch": 0.90167046317388,
      "grad_norm": 0.9448356032371521,
      "learning_rate": 1.3069709883897516e-05,
      "loss": 0.6103,
      "step": 7125
    },
    {
      "epoch": 0.9017970134143255,
      "grad_norm": 0.8676756024360657,
      "learning_rate": 1.3068004236768963e-05,
      "loss": 0.5623,
      "step": 7126
    },
    {
      "epoch": 0.901923563654771,
      "grad_norm": 0.9341461062431335,
      "learning_rate": 1.3066298491105104e-05,
      "loss": 0.5693,
      "step": 7127
    },
    {
      "epoch": 0.9020501138952164,
      "grad_norm": 0.8888821601867676,
      "learning_rate": 1.306459264696072e-05,
      "loss": 0.5815,
      "step": 7128
    },
    {
      "epoch": 0.9021766641356619,
      "grad_norm": 0.8691679835319519,
      "learning_rate": 1.30628867043906e-05,
      "loss": 0.6183,
      "step": 7129
    },
    {
      "epoch": 0.9023032143761073,
      "grad_norm": 0.888267457485199,
      "learning_rate": 1.3061180663449537e-05,
      "loss": 0.6176,
      "step": 7130
    },
    {
      "epoch": 0.9024297646165528,
      "grad_norm": 0.9056199193000793,
      "learning_rate": 1.3059474524192321e-05,
      "loss": 0.601,
      "step": 7131
    },
    {
      "epoch": 0.9025563148569983,
      "grad_norm": 0.913936972618103,
      "learning_rate": 1.305776828667375e-05,
      "loss": 0.6209,
      "step": 7132
    },
    {
      "epoch": 0.9026828650974437,
      "grad_norm": 0.8723173141479492,
      "learning_rate": 1.3056061950948621e-05,
      "loss": 0.6068,
      "step": 7133
    },
    {
      "epoch": 0.9028094153378892,
      "grad_norm": 0.8844301104545593,
      "learning_rate": 1.3054355517071737e-05,
      "loss": 0.6614,
      "step": 7134
    },
    {
      "epoch": 0.9029359655783346,
      "grad_norm": 0.8996499180793762,
      "learning_rate": 1.3052648985097907e-05,
      "loss": 0.5864,
      "step": 7135
    },
    {
      "epoch": 0.90306251581878,
      "grad_norm": 0.9028494358062744,
      "learning_rate": 1.3050942355081936e-05,
      "loss": 0.5607,
      "step": 7136
    },
    {
      "epoch": 0.9031890660592256,
      "grad_norm": 0.882596492767334,
      "learning_rate": 1.3049235627078638e-05,
      "loss": 0.5443,
      "step": 7137
    },
    {
      "epoch": 0.9033156162996709,
      "grad_norm": 0.8829905986785889,
      "learning_rate": 1.3047528801142828e-05,
      "loss": 0.5735,
      "step": 7138
    },
    {
      "epoch": 0.9034421665401164,
      "grad_norm": 0.9511678814888,
      "learning_rate": 1.304582187732932e-05,
      "loss": 0.6491,
      "step": 7139
    },
    {
      "epoch": 0.9035687167805618,
      "grad_norm": 0.9021119475364685,
      "learning_rate": 1.3044114855692942e-05,
      "loss": 0.6019,
      "step": 7140
    },
    {
      "epoch": 0.9036952670210073,
      "grad_norm": 0.9276257157325745,
      "learning_rate": 1.3042407736288514e-05,
      "loss": 0.5869,
      "step": 7141
    },
    {
      "epoch": 0.9038218172614528,
      "grad_norm": 0.9473169445991516,
      "learning_rate": 1.3040700519170866e-05,
      "loss": 0.6267,
      "step": 7142
    },
    {
      "epoch": 0.9039483675018982,
      "grad_norm": 0.8994737863540649,
      "learning_rate": 1.303899320439483e-05,
      "loss": 0.5998,
      "step": 7143
    },
    {
      "epoch": 0.9040749177423437,
      "grad_norm": 0.851489782333374,
      "learning_rate": 1.3037285792015235e-05,
      "loss": 0.5935,
      "step": 7144
    },
    {
      "epoch": 0.9042014679827892,
      "grad_norm": 0.9754672050476074,
      "learning_rate": 1.303557828208692e-05,
      "loss": 0.5772,
      "step": 7145
    },
    {
      "epoch": 0.9043280182232346,
      "grad_norm": 0.9062927961349487,
      "learning_rate": 1.3033870674664728e-05,
      "loss": 0.6211,
      "step": 7146
    },
    {
      "epoch": 0.9044545684636801,
      "grad_norm": 0.9192376136779785,
      "learning_rate": 1.3032162969803502e-05,
      "loss": 0.6076,
      "step": 7147
    },
    {
      "epoch": 0.9045811187041255,
      "grad_norm": 0.8636511564254761,
      "learning_rate": 1.3030455167558084e-05,
      "loss": 0.5543,
      "step": 7148
    },
    {
      "epoch": 0.904707668944571,
      "grad_norm": 0.910205066204071,
      "learning_rate": 1.302874726798333e-05,
      "loss": 0.5764,
      "step": 7149
    },
    {
      "epoch": 0.9048342191850165,
      "grad_norm": 0.9564871788024902,
      "learning_rate": 1.302703927113409e-05,
      "loss": 0.6402,
      "step": 7150
    },
    {
      "epoch": 0.9049607694254619,
      "grad_norm": 0.9283785223960876,
      "learning_rate": 1.3025331177065214e-05,
      "loss": 0.6189,
      "step": 7151
    },
    {
      "epoch": 0.9050873196659074,
      "grad_norm": 0.9734298586845398,
      "learning_rate": 1.302362298583157e-05,
      "loss": 0.611,
      "step": 7152
    },
    {
      "epoch": 0.9052138699063528,
      "grad_norm": 0.9142728447914124,
      "learning_rate": 1.3021914697488015e-05,
      "loss": 0.5878,
      "step": 7153
    },
    {
      "epoch": 0.9053404201467983,
      "grad_norm": 0.9572013020515442,
      "learning_rate": 1.3020206312089419e-05,
      "loss": 0.6061,
      "step": 7154
    },
    {
      "epoch": 0.9054669703872438,
      "grad_norm": 1.0409823656082153,
      "learning_rate": 1.3018497829690641e-05,
      "loss": 0.6218,
      "step": 7155
    },
    {
      "epoch": 0.9055935206276892,
      "grad_norm": 0.903965950012207,
      "learning_rate": 1.301678925034656e-05,
      "loss": 0.5356,
      "step": 7156
    },
    {
      "epoch": 0.9057200708681347,
      "grad_norm": 0.9763181209564209,
      "learning_rate": 1.301508057411205e-05,
      "loss": 0.5453,
      "step": 7157
    },
    {
      "epoch": 0.90584662110858,
      "grad_norm": 0.8888749480247498,
      "learning_rate": 1.3013371801041988e-05,
      "loss": 0.6179,
      "step": 7158
    },
    {
      "epoch": 0.9059731713490256,
      "grad_norm": 0.9119275212287903,
      "learning_rate": 1.3011662931191253e-05,
      "loss": 0.5505,
      "step": 7159
    },
    {
      "epoch": 0.906099721589471,
      "grad_norm": 0.8908205628395081,
      "learning_rate": 1.3009953964614733e-05,
      "loss": 0.5657,
      "step": 7160
    },
    {
      "epoch": 0.9062262718299164,
      "grad_norm": 0.911276638507843,
      "learning_rate": 1.3008244901367308e-05,
      "loss": 0.6142,
      "step": 7161
    },
    {
      "epoch": 0.9063528220703619,
      "grad_norm": 0.9278319478034973,
      "learning_rate": 1.3006535741503878e-05,
      "loss": 0.552,
      "step": 7162
    },
    {
      "epoch": 0.9064793723108074,
      "grad_norm": 0.8742029666900635,
      "learning_rate": 1.3004826485079325e-05,
      "loss": 0.5914,
      "step": 7163
    },
    {
      "epoch": 0.9066059225512528,
      "grad_norm": 0.8478991389274597,
      "learning_rate": 1.3003117132148553e-05,
      "loss": 0.5368,
      "step": 7164
    },
    {
      "epoch": 0.9067324727916983,
      "grad_norm": 0.9242635369300842,
      "learning_rate": 1.3001407682766458e-05,
      "loss": 0.5899,
      "step": 7165
    },
    {
      "epoch": 0.9068590230321437,
      "grad_norm": 0.9354316592216492,
      "learning_rate": 1.2999698136987946e-05,
      "loss": 0.6181,
      "step": 7166
    },
    {
      "epoch": 0.9069855732725892,
      "grad_norm": 0.8887894153594971,
      "learning_rate": 1.2997988494867919e-05,
      "loss": 0.6016,
      "step": 7167
    },
    {
      "epoch": 0.9071121235130347,
      "grad_norm": 0.9061421155929565,
      "learning_rate": 1.2996278756461287e-05,
      "loss": 0.6025,
      "step": 7168
    },
    {
      "epoch": 0.9072386737534801,
      "grad_norm": 0.8580600023269653,
      "learning_rate": 1.2994568921822963e-05,
      "loss": 0.5634,
      "step": 7169
    },
    {
      "epoch": 0.9073652239939256,
      "grad_norm": 0.8036961555480957,
      "learning_rate": 1.2992858991007858e-05,
      "loss": 0.5537,
      "step": 7170
    },
    {
      "epoch": 0.907491774234371,
      "grad_norm": 0.9132328033447266,
      "learning_rate": 1.2991148964070894e-05,
      "loss": 0.5901,
      "step": 7171
    },
    {
      "epoch": 0.9076183244748165,
      "grad_norm": 0.8809610605239868,
      "learning_rate": 1.2989438841066993e-05,
      "loss": 0.5828,
      "step": 7172
    },
    {
      "epoch": 0.907744874715262,
      "grad_norm": 1.4278874397277832,
      "learning_rate": 1.2987728622051075e-05,
      "loss": 0.623,
      "step": 7173
    },
    {
      "epoch": 0.9078714249557074,
      "grad_norm": 0.9473563432693481,
      "learning_rate": 1.298601830707807e-05,
      "loss": 0.6217,
      "step": 7174
    },
    {
      "epoch": 0.9079979751961529,
      "grad_norm": 0.868340790271759,
      "learning_rate": 1.2984307896202906e-05,
      "loss": 0.5833,
      "step": 7175
    },
    {
      "epoch": 0.9081245254365984,
      "grad_norm": 0.9654658436775208,
      "learning_rate": 1.2982597389480518e-05,
      "loss": 0.6244,
      "step": 7176
    },
    {
      "epoch": 0.9082510756770438,
      "grad_norm": 0.8917039632797241,
      "learning_rate": 1.2980886786965843e-05,
      "loss": 0.6025,
      "step": 7177
    },
    {
      "epoch": 0.9083776259174893,
      "grad_norm": 0.91196209192276,
      "learning_rate": 1.2979176088713818e-05,
      "loss": 0.605,
      "step": 7178
    },
    {
      "epoch": 0.9085041761579347,
      "grad_norm": 1.0902588367462158,
      "learning_rate": 1.297746529477939e-05,
      "loss": 0.6007,
      "step": 7179
    },
    {
      "epoch": 0.9086307263983802,
      "grad_norm": 0.9204440116882324,
      "learning_rate": 1.29757544052175e-05,
      "loss": 0.5826,
      "step": 7180
    },
    {
      "epoch": 0.9087572766388257,
      "grad_norm": 0.866417646408081,
      "learning_rate": 1.29740434200831e-05,
      "loss": 0.5766,
      "step": 7181
    },
    {
      "epoch": 0.908883826879271,
      "grad_norm": 0.8759669065475464,
      "learning_rate": 1.2972332339431137e-05,
      "loss": 0.5818,
      "step": 7182
    },
    {
      "epoch": 0.9090103771197166,
      "grad_norm": 0.8623671531677246,
      "learning_rate": 1.2970621163316577e-05,
      "loss": 0.5781,
      "step": 7183
    },
    {
      "epoch": 0.9091369273601619,
      "grad_norm": 0.8464904427528381,
      "learning_rate": 1.2968909891794366e-05,
      "loss": 0.5658,
      "step": 7184
    },
    {
      "epoch": 0.9092634776006074,
      "grad_norm": 0.8809431195259094,
      "learning_rate": 1.2967198524919468e-05,
      "loss": 0.5885,
      "step": 7185
    },
    {
      "epoch": 0.9093900278410529,
      "grad_norm": 0.8770003914833069,
      "learning_rate": 1.2965487062746851e-05,
      "loss": 0.5785,
      "step": 7186
    },
    {
      "epoch": 0.9095165780814983,
      "grad_norm": 0.8600886464118958,
      "learning_rate": 1.296377550533148e-05,
      "loss": 0.5682,
      "step": 7187
    },
    {
      "epoch": 0.9096431283219438,
      "grad_norm": 0.8844076991081238,
      "learning_rate": 1.2962063852728324e-05,
      "loss": 0.6367,
      "step": 7188
    },
    {
      "epoch": 0.9097696785623892,
      "grad_norm": 0.8673045039176941,
      "learning_rate": 1.2960352104992357e-05,
      "loss": 0.6263,
      "step": 7189
    },
    {
      "epoch": 0.9098962288028347,
      "grad_norm": 1.0533798933029175,
      "learning_rate": 1.2958640262178557e-05,
      "loss": 0.5606,
      "step": 7190
    },
    {
      "epoch": 0.9100227790432802,
      "grad_norm": 0.9230180978775024,
      "learning_rate": 1.29569283243419e-05,
      "loss": 0.6015,
      "step": 7191
    },
    {
      "epoch": 0.9101493292837256,
      "grad_norm": 0.928135335445404,
      "learning_rate": 1.2955216291537371e-05,
      "loss": 0.5818,
      "step": 7192
    },
    {
      "epoch": 0.9102758795241711,
      "grad_norm": 0.9012940526008606,
      "learning_rate": 1.2953504163819956e-05,
      "loss": 0.5608,
      "step": 7193
    },
    {
      "epoch": 0.9104024297646166,
      "grad_norm": 0.894477128982544,
      "learning_rate": 1.295179194124464e-05,
      "loss": 0.5996,
      "step": 7194
    },
    {
      "epoch": 0.910528980005062,
      "grad_norm": 0.8935933113098145,
      "learning_rate": 1.2950079623866421e-05,
      "loss": 0.5477,
      "step": 7195
    },
    {
      "epoch": 0.9106555302455075,
      "grad_norm": 0.8913790583610535,
      "learning_rate": 1.2948367211740287e-05,
      "loss": 0.6035,
      "step": 7196
    },
    {
      "epoch": 0.9107820804859529,
      "grad_norm": 0.873952329158783,
      "learning_rate": 1.2946654704921237e-05,
      "loss": 0.5665,
      "step": 7197
    },
    {
      "epoch": 0.9109086307263984,
      "grad_norm": 0.924275279045105,
      "learning_rate": 1.2944942103464277e-05,
      "loss": 0.6064,
      "step": 7198
    },
    {
      "epoch": 0.9110351809668439,
      "grad_norm": 0.8571138381958008,
      "learning_rate": 1.2943229407424402e-05,
      "loss": 0.5396,
      "step": 7199
    },
    {
      "epoch": 0.9111617312072893,
      "grad_norm": 0.9203358292579651,
      "learning_rate": 1.2941516616856628e-05,
      "loss": 0.6529,
      "step": 7200
    },
    {
      "epoch": 0.9112882814477348,
      "grad_norm": 0.9573639631271362,
      "learning_rate": 1.2939803731815959e-05,
      "loss": 0.6084,
      "step": 7201
    },
    {
      "epoch": 0.9114148316881802,
      "grad_norm": 0.8832917809486389,
      "learning_rate": 1.2938090752357407e-05,
      "loss": 0.5765,
      "step": 7202
    },
    {
      "epoch": 0.9115413819286257,
      "grad_norm": 0.8387922048568726,
      "learning_rate": 1.2936377678535989e-05,
      "loss": 0.5459,
      "step": 7203
    },
    {
      "epoch": 0.9116679321690712,
      "grad_norm": 0.8683356046676636,
      "learning_rate": 1.2934664510406728e-05,
      "loss": 0.6199,
      "step": 7204
    },
    {
      "epoch": 0.9117944824095165,
      "grad_norm": 0.8913978934288025,
      "learning_rate": 1.2932951248024641e-05,
      "loss": 0.5758,
      "step": 7205
    },
    {
      "epoch": 0.911921032649962,
      "grad_norm": 0.8837693929672241,
      "learning_rate": 1.2931237891444757e-05,
      "loss": 0.5894,
      "step": 7206
    },
    {
      "epoch": 0.9120475828904074,
      "grad_norm": 0.9464014768600464,
      "learning_rate": 1.29295244407221e-05,
      "loss": 0.5693,
      "step": 7207
    },
    {
      "epoch": 0.9121741331308529,
      "grad_norm": 0.8436591625213623,
      "learning_rate": 1.2927810895911701e-05,
      "loss": 0.559,
      "step": 7208
    },
    {
      "epoch": 0.9123006833712984,
      "grad_norm": 0.8947300910949707,
      "learning_rate": 1.2926097257068597e-05,
      "loss": 0.5412,
      "step": 7209
    },
    {
      "epoch": 0.9124272336117438,
      "grad_norm": 0.9400574564933777,
      "learning_rate": 1.2924383524247829e-05,
      "loss": 0.5957,
      "step": 7210
    },
    {
      "epoch": 0.9125537838521893,
      "grad_norm": 0.9456593990325928,
      "learning_rate": 1.2922669697504426e-05,
      "loss": 0.6203,
      "step": 7211
    },
    {
      "epoch": 0.9126803340926348,
      "grad_norm": 0.9126949906349182,
      "learning_rate": 1.292095577689344e-05,
      "loss": 0.5671,
      "step": 7212
    },
    {
      "epoch": 0.9128068843330802,
      "grad_norm": 0.9387266635894775,
      "learning_rate": 1.2919241762469911e-05,
      "loss": 0.6289,
      "step": 7213
    },
    {
      "epoch": 0.9129334345735257,
      "grad_norm": 0.8765755891799927,
      "learning_rate": 1.2917527654288894e-05,
      "loss": 0.5707,
      "step": 7214
    },
    {
      "epoch": 0.9130599848139711,
      "grad_norm": 0.9277388453483582,
      "learning_rate": 1.2915813452405438e-05,
      "loss": 0.6724,
      "step": 7215
    },
    {
      "epoch": 0.9131865350544166,
      "grad_norm": 0.8611820936203003,
      "learning_rate": 1.2914099156874598e-05,
      "loss": 0.5181,
      "step": 7216
    },
    {
      "epoch": 0.9133130852948621,
      "grad_norm": 0.9395997524261475,
      "learning_rate": 1.2912384767751434e-05,
      "loss": 0.6093,
      "step": 7217
    },
    {
      "epoch": 0.9134396355353075,
      "grad_norm": 0.8679150938987732,
      "learning_rate": 1.2910670285091005e-05,
      "loss": 0.5689,
      "step": 7218
    },
    {
      "epoch": 0.913566185775753,
      "grad_norm": 0.9109736680984497,
      "learning_rate": 1.2908955708948376e-05,
      "loss": 0.6034,
      "step": 7219
    },
    {
      "epoch": 0.9136927360161984,
      "grad_norm": 0.8676458597183228,
      "learning_rate": 1.2907241039378615e-05,
      "loss": 0.5914,
      "step": 7220
    },
    {
      "epoch": 0.9138192862566439,
      "grad_norm": 0.9793232083320618,
      "learning_rate": 1.2905526276436788e-05,
      "loss": 0.6169,
      "step": 7221
    },
    {
      "epoch": 0.9139458364970894,
      "grad_norm": 0.8748593330383301,
      "learning_rate": 1.2903811420177976e-05,
      "loss": 0.5497,
      "step": 7222
    },
    {
      "epoch": 0.9140723867375348,
      "grad_norm": 1.0206120014190674,
      "learning_rate": 1.290209647065725e-05,
      "loss": 0.5654,
      "step": 7223
    },
    {
      "epoch": 0.9141989369779803,
      "grad_norm": 0.8930026888847351,
      "learning_rate": 1.290038142792969e-05,
      "loss": 0.6196,
      "step": 7224
    },
    {
      "epoch": 0.9143254872184258,
      "grad_norm": 0.9263008236885071,
      "learning_rate": 1.2898666292050376e-05,
      "loss": 0.5858,
      "step": 7225
    },
    {
      "epoch": 0.9144520374588712,
      "grad_norm": 0.9419111609458923,
      "learning_rate": 1.2896951063074396e-05,
      "loss": 0.5824,
      "step": 7226
    },
    {
      "epoch": 0.9145785876993167,
      "grad_norm": 0.9514508247375488,
      "learning_rate": 1.2895235741056837e-05,
      "loss": 0.6062,
      "step": 7227
    },
    {
      "epoch": 0.914705137939762,
      "grad_norm": 0.9188080430030823,
      "learning_rate": 1.2893520326052791e-05,
      "loss": 0.6105,
      "step": 7228
    },
    {
      "epoch": 0.9148316881802075,
      "grad_norm": 0.9366318583488464,
      "learning_rate": 1.289180481811735e-05,
      "loss": 0.5882,
      "step": 7229
    },
    {
      "epoch": 0.914958238420653,
      "grad_norm": 0.8870358467102051,
      "learning_rate": 1.2890089217305615e-05,
      "loss": 0.5996,
      "step": 7230
    },
    {
      "epoch": 0.9150847886610984,
      "grad_norm": 0.9157900810241699,
      "learning_rate": 1.2888373523672681e-05,
      "loss": 0.589,
      "step": 7231
    },
    {
      "epoch": 0.9152113389015439,
      "grad_norm": 0.9502314925193787,
      "learning_rate": 1.2886657737273653e-05,
      "loss": 0.6259,
      "step": 7232
    },
    {
      "epoch": 0.9153378891419893,
      "grad_norm": 0.9153730273246765,
      "learning_rate": 1.2884941858163643e-05,
      "loss": 0.6315,
      "step": 7233
    },
    {
      "epoch": 0.9154644393824348,
      "grad_norm": 0.9755611419677734,
      "learning_rate": 1.288322588639775e-05,
      "loss": 0.5886,
      "step": 7234
    },
    {
      "epoch": 0.9155909896228803,
      "grad_norm": 0.8705446720123291,
      "learning_rate": 1.2881509822031093e-05,
      "loss": 0.5742,
      "step": 7235
    },
    {
      "epoch": 0.9157175398633257,
      "grad_norm": 0.9584953188896179,
      "learning_rate": 1.2879793665118785e-05,
      "loss": 0.6576,
      "step": 7236
    },
    {
      "epoch": 0.9158440901037712,
      "grad_norm": 0.861957848072052,
      "learning_rate": 1.2878077415715942e-05,
      "loss": 0.6102,
      "step": 7237
    },
    {
      "epoch": 0.9159706403442166,
      "grad_norm": 1.4319528341293335,
      "learning_rate": 1.2876361073877686e-05,
      "loss": 0.5647,
      "step": 7238
    },
    {
      "epoch": 0.9160971905846621,
      "grad_norm": 0.8976839184761047,
      "learning_rate": 1.2874644639659142e-05,
      "loss": 0.6145,
      "step": 7239
    },
    {
      "epoch": 0.9162237408251076,
      "grad_norm": 0.8713439106941223,
      "learning_rate": 1.2872928113115434e-05,
      "loss": 0.5464,
      "step": 7240
    },
    {
      "epoch": 0.916350291065553,
      "grad_norm": 0.8615956902503967,
      "learning_rate": 1.28712114943017e-05,
      "loss": 0.5335,
      "step": 7241
    },
    {
      "epoch": 0.9164768413059985,
      "grad_norm": 0.8789289593696594,
      "learning_rate": 1.2869494783273062e-05,
      "loss": 0.5884,
      "step": 7242
    },
    {
      "epoch": 0.916603391546444,
      "grad_norm": 0.9033557176589966,
      "learning_rate": 1.2867777980084662e-05,
      "loss": 0.6187,
      "step": 7243
    },
    {
      "epoch": 0.9167299417868894,
      "grad_norm": 0.8265066146850586,
      "learning_rate": 1.2866061084791634e-05,
      "loss": 0.5743,
      "step": 7244
    },
    {
      "epoch": 0.9168564920273349,
      "grad_norm": 0.8937229514122009,
      "learning_rate": 1.2864344097449127e-05,
      "loss": 0.5459,
      "step": 7245
    },
    {
      "epoch": 0.9169830422677803,
      "grad_norm": 0.8720482587814331,
      "learning_rate": 1.2862627018112282e-05,
      "loss": 0.618,
      "step": 7246
    },
    {
      "epoch": 0.9171095925082258,
      "grad_norm": 0.9078347682952881,
      "learning_rate": 1.2860909846836243e-05,
      "loss": 0.5682,
      "step": 7247
    },
    {
      "epoch": 0.9172361427486713,
      "grad_norm": 0.8770893216133118,
      "learning_rate": 1.2859192583676166e-05,
      "loss": 0.5745,
      "step": 7248
    },
    {
      "epoch": 0.9173626929891167,
      "grad_norm": 0.8955170512199402,
      "learning_rate": 1.2857475228687202e-05,
      "loss": 0.6152,
      "step": 7249
    },
    {
      "epoch": 0.9174892432295622,
      "grad_norm": 0.9087849855422974,
      "learning_rate": 1.2855757781924505e-05,
      "loss": 0.5865,
      "step": 7250
    },
    {
      "epoch": 0.9176157934700075,
      "grad_norm": 1.1877385377883911,
      "learning_rate": 1.2854040243443238e-05,
      "loss": 0.5739,
      "step": 7251
    },
    {
      "epoch": 0.917742343710453,
      "grad_norm": 0.9314249753952026,
      "learning_rate": 1.2852322613298562e-05,
      "loss": 0.6155,
      "step": 7252
    },
    {
      "epoch": 0.9178688939508985,
      "grad_norm": 0.9034152626991272,
      "learning_rate": 1.2850604891545641e-05,
      "loss": 0.583,
      "step": 7253
    },
    {
      "epoch": 0.9179954441913439,
      "grad_norm": 0.9114294052124023,
      "learning_rate": 1.2848887078239645e-05,
      "loss": 0.5976,
      "step": 7254
    },
    {
      "epoch": 0.9181219944317894,
      "grad_norm": 0.8619153499603271,
      "learning_rate": 1.2847169173435745e-05,
      "loss": 0.5569,
      "step": 7255
    },
    {
      "epoch": 0.9182485446722348,
      "grad_norm": 0.8832245469093323,
      "learning_rate": 1.2845451177189115e-05,
      "loss": 0.5814,
      "step": 7256
    },
    {
      "epoch": 0.9183750949126803,
      "grad_norm": 0.8708439469337463,
      "learning_rate": 1.284373308955493e-05,
      "loss": 0.5754,
      "step": 7257
    },
    {
      "epoch": 0.9185016451531258,
      "grad_norm": 0.8654161691665649,
      "learning_rate": 1.2842014910588375e-05,
      "loss": 0.5559,
      "step": 7258
    },
    {
      "epoch": 0.9186281953935712,
      "grad_norm": 0.9329561591148376,
      "learning_rate": 1.2840296640344627e-05,
      "loss": 0.539,
      "step": 7259
    },
    {
      "epoch": 0.9187547456340167,
      "grad_norm": 0.881659209728241,
      "learning_rate": 1.2838578278878877e-05,
      "loss": 0.5766,
      "step": 7260
    },
    {
      "epoch": 0.9188812958744622,
      "grad_norm": 0.898004949092865,
      "learning_rate": 1.2836859826246308e-05,
      "loss": 0.6317,
      "step": 7261
    },
    {
      "epoch": 0.9190078461149076,
      "grad_norm": 0.8923391103744507,
      "learning_rate": 1.2835141282502114e-05,
      "loss": 0.5919,
      "step": 7262
    },
    {
      "epoch": 0.9191343963553531,
      "grad_norm": 0.9424705505371094,
      "learning_rate": 1.2833422647701492e-05,
      "loss": 0.5845,
      "step": 7263
    },
    {
      "epoch": 0.9192609465957985,
      "grad_norm": 1.0212689638137817,
      "learning_rate": 1.283170392189964e-05,
      "loss": 0.5723,
      "step": 7264
    },
    {
      "epoch": 0.919387496836244,
      "grad_norm": 0.9284647703170776,
      "learning_rate": 1.2829985105151752e-05,
      "loss": 0.6202,
      "step": 7265
    },
    {
      "epoch": 0.9195140470766895,
      "grad_norm": 0.8910000324249268,
      "learning_rate": 1.2828266197513037e-05,
      "loss": 0.6344,
      "step": 7266
    },
    {
      "epoch": 0.9196405973171349,
      "grad_norm": 0.8907369375228882,
      "learning_rate": 1.28265471990387e-05,
      "loss": 0.561,
      "step": 7267
    },
    {
      "epoch": 0.9197671475575804,
      "grad_norm": 0.9341697096824646,
      "learning_rate": 1.2824828109783952e-05,
      "loss": 0.604,
      "step": 7268
    },
    {
      "epoch": 0.9198936977980258,
      "grad_norm": 0.8799750208854675,
      "learning_rate": 1.2823108929804002e-05,
      "loss": 0.5984,
      "step": 7269
    },
    {
      "epoch": 0.9200202480384713,
      "grad_norm": 0.9774530529975891,
      "learning_rate": 1.2821389659154067e-05,
      "loss": 0.6028,
      "step": 7270
    },
    {
      "epoch": 0.9201467982789168,
      "grad_norm": 0.8611702919006348,
      "learning_rate": 1.2819670297889361e-05,
      "loss": 0.5736,
      "step": 7271
    },
    {
      "epoch": 0.9202733485193622,
      "grad_norm": 0.9027135372161865,
      "learning_rate": 1.281795084606511e-05,
      "loss": 0.5744,
      "step": 7272
    },
    {
      "epoch": 0.9203998987598077,
      "grad_norm": 0.9201870560646057,
      "learning_rate": 1.2816231303736534e-05,
      "loss": 0.6036,
      "step": 7273
    },
    {
      "epoch": 0.9205264490002532,
      "grad_norm": 0.9763070940971375,
      "learning_rate": 1.2814511670958864e-05,
      "loss": 0.5844,
      "step": 7274
    },
    {
      "epoch": 0.9206529992406985,
      "grad_norm": 0.9838351011276245,
      "learning_rate": 1.2812791947787329e-05,
      "loss": 0.6458,
      "step": 7275
    },
    {
      "epoch": 0.920779549481144,
      "grad_norm": 0.9045073390007019,
      "learning_rate": 1.2811072134277153e-05,
      "loss": 0.6039,
      "step": 7276
    },
    {
      "epoch": 0.9209060997215894,
      "grad_norm": 0.8684774041175842,
      "learning_rate": 1.2809352230483582e-05,
      "loss": 0.5925,
      "step": 7277
    },
    {
      "epoch": 0.9210326499620349,
      "grad_norm": 0.9912064075469971,
      "learning_rate": 1.280763223646185e-05,
      "loss": 0.6095,
      "step": 7278
    },
    {
      "epoch": 0.9211592002024804,
      "grad_norm": 0.8448657393455505,
      "learning_rate": 1.2805912152267195e-05,
      "loss": 0.5317,
      "step": 7279
    },
    {
      "epoch": 0.9212857504429258,
      "grad_norm": 0.906173825263977,
      "learning_rate": 1.2804191977954868e-05,
      "loss": 0.5888,
      "step": 7280
    },
    {
      "epoch": 0.9214123006833713,
      "grad_norm": 0.9469031095504761,
      "learning_rate": 1.2802471713580112e-05,
      "loss": 0.5999,
      "step": 7281
    },
    {
      "epoch": 0.9215388509238167,
      "grad_norm": 0.8872721195220947,
      "learning_rate": 1.2800751359198174e-05,
      "loss": 0.5685,
      "step": 7282
    },
    {
      "epoch": 0.9216654011642622,
      "grad_norm": 0.9381269812583923,
      "learning_rate": 1.2799030914864312e-05,
      "loss": 0.5756,
      "step": 7283
    },
    {
      "epoch": 0.9217919514047077,
      "grad_norm": 19.669082641601562,
      "learning_rate": 1.279731038063378e-05,
      "loss": 0.581,
      "step": 7284
    },
    {
      "epoch": 0.9219185016451531,
      "grad_norm": 0.9039819836616516,
      "learning_rate": 1.2795589756561834e-05,
      "loss": 0.6063,
      "step": 7285
    },
    {
      "epoch": 0.9220450518855986,
      "grad_norm": 0.9374739527702332,
      "learning_rate": 1.279386904270374e-05,
      "loss": 0.6093,
      "step": 7286
    },
    {
      "epoch": 0.922171602126044,
      "grad_norm": 0.9562804698944092,
      "learning_rate": 1.2792148239114759e-05,
      "loss": 0.6222,
      "step": 7287
    },
    {
      "epoch": 0.9222981523664895,
      "grad_norm": 0.9336996078491211,
      "learning_rate": 1.2790427345850158e-05,
      "loss": 0.6074,
      "step": 7288
    },
    {
      "epoch": 0.922424702606935,
      "grad_norm": 0.9798687696456909,
      "learning_rate": 1.2788706362965208e-05,
      "loss": 0.6355,
      "step": 7289
    },
    {
      "epoch": 0.9225512528473804,
      "grad_norm": 0.9602746963500977,
      "learning_rate": 1.2786985290515181e-05,
      "loss": 0.5769,
      "step": 7290
    },
    {
      "epoch": 0.9226778030878259,
      "grad_norm": 0.8941152691841125,
      "learning_rate": 1.2785264128555355e-05,
      "loss": 0.598,
      "step": 7291
    },
    {
      "epoch": 0.9228043533282714,
      "grad_norm": 1.3285576105117798,
      "learning_rate": 1.2783542877141007e-05,
      "loss": 0.5931,
      "step": 7292
    },
    {
      "epoch": 0.9229309035687168,
      "grad_norm": 0.9407840967178345,
      "learning_rate": 1.278182153632742e-05,
      "loss": 0.5526,
      "step": 7293
    },
    {
      "epoch": 0.9230574538091623,
      "grad_norm": 0.8951389193534851,
      "learning_rate": 1.2780100106169874e-05,
      "loss": 0.5829,
      "step": 7294
    },
    {
      "epoch": 0.9231840040496077,
      "grad_norm": 0.9867222309112549,
      "learning_rate": 1.277837858672366e-05,
      "loss": 0.5701,
      "step": 7295
    },
    {
      "epoch": 0.9233105542900532,
      "grad_norm": 0.9328342080116272,
      "learning_rate": 1.277665697804407e-05,
      "loss": 0.5562,
      "step": 7296
    },
    {
      "epoch": 0.9234371045304987,
      "grad_norm": 0.904831051826477,
      "learning_rate": 1.2774935280186394e-05,
      "loss": 0.5649,
      "step": 7297
    },
    {
      "epoch": 0.923563654770944,
      "grad_norm": 0.9086433053016663,
      "learning_rate": 1.2773213493205932e-05,
      "loss": 0.5776,
      "step": 7298
    },
    {
      "epoch": 0.9236902050113895,
      "grad_norm": 0.8945661187171936,
      "learning_rate": 1.2771491617157973e-05,
      "loss": 0.6022,
      "step": 7299
    },
    {
      "epoch": 0.9238167552518349,
      "grad_norm": 12.213249206542969,
      "learning_rate": 1.2769769652097828e-05,
      "loss": 0.6089,
      "step": 7300
    },
    {
      "epoch": 0.9239433054922804,
      "grad_norm": 0.972781777381897,
      "learning_rate": 1.27680475980808e-05,
      "loss": 0.5996,
      "step": 7301
    },
    {
      "epoch": 0.9240698557327259,
      "grad_norm": 0.8978752493858337,
      "learning_rate": 1.2766325455162192e-05,
      "loss": 0.5901,
      "step": 7302
    },
    {
      "epoch": 0.9241964059731713,
      "grad_norm": 0.9731771349906921,
      "learning_rate": 1.2764603223397316e-05,
      "loss": 0.6064,
      "step": 7303
    },
    {
      "epoch": 0.9243229562136168,
      "grad_norm": 1.45345938205719,
      "learning_rate": 1.2762880902841493e-05,
      "loss": 0.6161,
      "step": 7304
    },
    {
      "epoch": 0.9244495064540622,
      "grad_norm": 0.9333415031433105,
      "learning_rate": 1.2761158493550026e-05,
      "loss": 0.5838,
      "step": 7305
    },
    {
      "epoch": 0.9245760566945077,
      "grad_norm": 0.9375561475753784,
      "learning_rate": 1.275943599557824e-05,
      "loss": 0.5837,
      "step": 7306
    },
    {
      "epoch": 0.9247026069349532,
      "grad_norm": 0.8850058317184448,
      "learning_rate": 1.2757713408981456e-05,
      "loss": 0.6021,
      "step": 7307
    },
    {
      "epoch": 0.9248291571753986,
      "grad_norm": 0.8683674931526184,
      "learning_rate": 1.2755990733815002e-05,
      "loss": 0.6126,
      "step": 7308
    },
    {
      "epoch": 0.9249557074158441,
      "grad_norm": 0.9108549952507019,
      "learning_rate": 1.2754267970134198e-05,
      "loss": 0.5884,
      "step": 7309
    },
    {
      "epoch": 0.9250822576562896,
      "grad_norm": 0.8513150811195374,
      "learning_rate": 1.2752545117994381e-05,
      "loss": 0.5647,
      "step": 7310
    },
    {
      "epoch": 0.925208807896735,
      "grad_norm": 0.9137847423553467,
      "learning_rate": 1.2750822177450882e-05,
      "loss": 0.5733,
      "step": 7311
    },
    {
      "epoch": 0.9253353581371805,
      "grad_norm": 0.9339722394943237,
      "learning_rate": 1.2749099148559032e-05,
      "loss": 0.587,
      "step": 7312
    },
    {
      "epoch": 0.9254619083776259,
      "grad_norm": 0.8843140006065369,
      "learning_rate": 1.2747376031374175e-05,
      "loss": 0.5892,
      "step": 7313
    },
    {
      "epoch": 0.9255884586180714,
      "grad_norm": 0.8794586658477783,
      "learning_rate": 1.2745652825951648e-05,
      "loss": 0.563,
      "step": 7314
    },
    {
      "epoch": 0.9257150088585169,
      "grad_norm": 0.9083543419837952,
      "learning_rate": 1.2743929532346802e-05,
      "loss": 0.566,
      "step": 7315
    },
    {
      "epoch": 0.9258415590989623,
      "grad_norm": 0.9424682855606079,
      "learning_rate": 1.2742206150614982e-05,
      "loss": 0.6297,
      "step": 7316
    },
    {
      "epoch": 0.9259681093394078,
      "grad_norm": 0.911194384098053,
      "learning_rate": 1.2740482680811532e-05,
      "loss": 0.6,
      "step": 7317
    },
    {
      "epoch": 0.9260946595798532,
      "grad_norm": 0.9460294842720032,
      "learning_rate": 1.273875912299181e-05,
      "loss": 0.6065,
      "step": 7318
    },
    {
      "epoch": 0.9262212098202987,
      "grad_norm": 0.8553239107131958,
      "learning_rate": 1.2737035477211172e-05,
      "loss": 0.5643,
      "step": 7319
    },
    {
      "epoch": 0.9263477600607442,
      "grad_norm": 0.8869155049324036,
      "learning_rate": 1.2735311743524976e-05,
      "loss": 0.6133,
      "step": 7320
    },
    {
      "epoch": 0.9264743103011895,
      "grad_norm": 0.9473394751548767,
      "learning_rate": 1.2733587921988584e-05,
      "loss": 0.6085,
      "step": 7321
    },
    {
      "epoch": 0.926600860541635,
      "grad_norm": 0.8219423294067383,
      "learning_rate": 1.2731864012657358e-05,
      "loss": 0.5347,
      "step": 7322
    },
    {
      "epoch": 0.9267274107820805,
      "grad_norm": 0.9575338959693909,
      "learning_rate": 1.2730140015586663e-05,
      "loss": 0.6112,
      "step": 7323
    },
    {
      "epoch": 0.9268539610225259,
      "grad_norm": 0.900327742099762,
      "learning_rate": 1.2728415930831873e-05,
      "loss": 0.5671,
      "step": 7324
    },
    {
      "epoch": 0.9269805112629714,
      "grad_norm": 0.8788415789604187,
      "learning_rate": 1.2726691758448358e-05,
      "loss": 0.5451,
      "step": 7325
    },
    {
      "epoch": 0.9271070615034168,
      "grad_norm": 0.9185836315155029,
      "learning_rate": 1.2724967498491492e-05,
      "loss": 0.5884,
      "step": 7326
    },
    {
      "epoch": 0.9272336117438623,
      "grad_norm": 0.9388731718063354,
      "learning_rate": 1.272324315101666e-05,
      "loss": 0.5942,
      "step": 7327
    },
    {
      "epoch": 0.9273601619843078,
      "grad_norm": 0.8665038347244263,
      "learning_rate": 1.2721518716079235e-05,
      "loss": 0.5726,
      "step": 7328
    },
    {
      "epoch": 0.9274867122247532,
      "grad_norm": 0.8740907311439514,
      "learning_rate": 1.2719794193734606e-05,
      "loss": 0.5485,
      "step": 7329
    },
    {
      "epoch": 0.9276132624651987,
      "grad_norm": 0.8856884837150574,
      "learning_rate": 1.2718069584038155e-05,
      "loss": 0.6168,
      "step": 7330
    },
    {
      "epoch": 0.9277398127056441,
      "grad_norm": 0.8744152188301086,
      "learning_rate": 1.2716344887045275e-05,
      "loss": 0.5574,
      "step": 7331
    },
    {
      "epoch": 0.9278663629460896,
      "grad_norm": 1.0901861190795898,
      "learning_rate": 1.271462010281136e-05,
      "loss": 0.624,
      "step": 7332
    },
    {
      "epoch": 0.9279929131865351,
      "grad_norm": 0.9094693064689636,
      "learning_rate": 1.27128952313918e-05,
      "loss": 0.6275,
      "step": 7333
    },
    {
      "epoch": 0.9281194634269805,
      "grad_norm": 0.8766735196113586,
      "learning_rate": 1.2711170272841997e-05,
      "loss": 0.5554,
      "step": 7334
    },
    {
      "epoch": 0.928246013667426,
      "grad_norm": 0.9370375275611877,
      "learning_rate": 1.270944522721735e-05,
      "loss": 0.5824,
      "step": 7335
    },
    {
      "epoch": 0.9283725639078714,
      "grad_norm": 0.822529673576355,
      "learning_rate": 1.2707720094573261e-05,
      "loss": 0.5392,
      "step": 7336
    },
    {
      "epoch": 0.9284991141483169,
      "grad_norm": 1.057883620262146,
      "learning_rate": 1.2705994874965138e-05,
      "loss": 0.557,
      "step": 7337
    },
    {
      "epoch": 0.9286256643887624,
      "grad_norm": 1.0382095575332642,
      "learning_rate": 1.2704269568448386e-05,
      "loss": 0.5501,
      "step": 7338
    },
    {
      "epoch": 0.9287522146292078,
      "grad_norm": 0.979487955570221,
      "learning_rate": 1.2702544175078427e-05,
      "loss": 0.5783,
      "step": 7339
    },
    {
      "epoch": 0.9288787648696533,
      "grad_norm": 0.9708924293518066,
      "learning_rate": 1.2700818694910661e-05,
      "loss": 0.5745,
      "step": 7340
    },
    {
      "epoch": 0.9290053151100988,
      "grad_norm": 0.9522074460983276,
      "learning_rate": 1.269909312800052e-05,
      "loss": 0.584,
      "step": 7341
    },
    {
      "epoch": 0.9291318653505442,
      "grad_norm": 0.9241313934326172,
      "learning_rate": 1.2697367474403412e-05,
      "loss": 0.5809,
      "step": 7342
    },
    {
      "epoch": 0.9292584155909897,
      "grad_norm": 0.9131771922111511,
      "learning_rate": 1.2695641734174768e-05,
      "loss": 0.6222,
      "step": 7343
    },
    {
      "epoch": 0.929384965831435,
      "grad_norm": 0.9981878399848938,
      "learning_rate": 1.2693915907370011e-05,
      "loss": 0.5402,
      "step": 7344
    },
    {
      "epoch": 0.9295115160718805,
      "grad_norm": 0.8872877955436707,
      "learning_rate": 1.269218999404457e-05,
      "loss": 0.5531,
      "step": 7345
    },
    {
      "epoch": 0.929638066312326,
      "grad_norm": 0.8670530319213867,
      "learning_rate": 1.2690463994253874e-05,
      "loss": 0.5589,
      "step": 7346
    },
    {
      "epoch": 0.9297646165527714,
      "grad_norm": 0.9552625417709351,
      "learning_rate": 1.268873790805336e-05,
      "loss": 0.6093,
      "step": 7347
    },
    {
      "epoch": 0.9298911667932169,
      "grad_norm": 0.9246053099632263,
      "learning_rate": 1.2687011735498462e-05,
      "loss": 0.6005,
      "step": 7348
    },
    {
      "epoch": 0.9300177170336623,
      "grad_norm": 0.9342123866081238,
      "learning_rate": 1.2685285476644622e-05,
      "loss": 0.5941,
      "step": 7349
    },
    {
      "epoch": 0.9301442672741078,
      "grad_norm": 0.8378381729125977,
      "learning_rate": 1.2683559131547286e-05,
      "loss": 0.5514,
      "step": 7350
    },
    {
      "epoch": 0.9302708175145533,
      "grad_norm": 0.8867259621620178,
      "learning_rate": 1.2681832700261889e-05,
      "loss": 0.6159,
      "step": 7351
    },
    {
      "epoch": 0.9303973677549987,
      "grad_norm": 1.0791122913360596,
      "learning_rate": 1.2680106182843885e-05,
      "loss": 0.6225,
      "step": 7352
    },
    {
      "epoch": 0.9305239179954442,
      "grad_norm": 0.906002938747406,
      "learning_rate": 1.2678379579348728e-05,
      "loss": 0.5759,
      "step": 7353
    },
    {
      "epoch": 0.9306504682358896,
      "grad_norm": 0.9238664507865906,
      "learning_rate": 1.2676652889831864e-05,
      "loss": 0.5996,
      "step": 7354
    },
    {
      "epoch": 0.9307770184763351,
      "grad_norm": 0.8640145659446716,
      "learning_rate": 1.2674926114348754e-05,
      "loss": 0.5735,
      "step": 7355
    },
    {
      "epoch": 0.9309035687167806,
      "grad_norm": 0.8763598799705505,
      "learning_rate": 1.2673199252954858e-05,
      "loss": 0.6036,
      "step": 7356
    },
    {
      "epoch": 0.931030118957226,
      "grad_norm": 0.93541419506073,
      "learning_rate": 1.2671472305705635e-05,
      "loss": 0.5939,
      "step": 7357
    },
    {
      "epoch": 0.9311566691976715,
      "grad_norm": 0.8698449730873108,
      "learning_rate": 1.2669745272656548e-05,
      "loss": 0.6132,
      "step": 7358
    },
    {
      "epoch": 0.931283219438117,
      "grad_norm": 0.9330861568450928,
      "learning_rate": 1.2668018153863071e-05,
      "loss": 0.5944,
      "step": 7359
    },
    {
      "epoch": 0.9314097696785624,
      "grad_norm": 0.8973618149757385,
      "learning_rate": 1.2666290949380664e-05,
      "loss": 0.5665,
      "step": 7360
    },
    {
      "epoch": 0.9315363199190079,
      "grad_norm": 0.8731130361557007,
      "learning_rate": 1.2664563659264807e-05,
      "loss": 0.5494,
      "step": 7361
    },
    {
      "epoch": 0.9316628701594533,
      "grad_norm": 1.0151034593582153,
      "learning_rate": 1.2662836283570976e-05,
      "loss": 0.6057,
      "step": 7362
    },
    {
      "epoch": 0.9317894203998988,
      "grad_norm": 0.8730753660202026,
      "learning_rate": 1.2661108822354646e-05,
      "loss": 0.5474,
      "step": 7363
    },
    {
      "epoch": 0.9319159706403443,
      "grad_norm": 0.9198098182678223,
      "learning_rate": 1.2659381275671297e-05,
      "loss": 0.6227,
      "step": 7364
    },
    {
      "epoch": 0.9320425208807896,
      "grad_norm": 1.6037930250167847,
      "learning_rate": 1.2657653643576415e-05,
      "loss": 0.5823,
      "step": 7365
    },
    {
      "epoch": 0.9321690711212351,
      "grad_norm": 0.8897778987884521,
      "learning_rate": 1.2655925926125488e-05,
      "loss": 0.5664,
      "step": 7366
    },
    {
      "epoch": 0.9322956213616805,
      "grad_norm": 0.8638305068016052,
      "learning_rate": 1.2654198123374002e-05,
      "loss": 0.5432,
      "step": 7367
    },
    {
      "epoch": 0.932422171602126,
      "grad_norm": 0.8737266063690186,
      "learning_rate": 1.265247023537745e-05,
      "loss": 0.5688,
      "step": 7368
    },
    {
      "epoch": 0.9325487218425715,
      "grad_norm": 0.9548583626747131,
      "learning_rate": 1.2650742262191326e-05,
      "loss": 0.6006,
      "step": 7369
    },
    {
      "epoch": 0.9326752720830169,
      "grad_norm": 0.8709455728530884,
      "learning_rate": 1.264901420387113e-05,
      "loss": 0.5795,
      "step": 7370
    },
    {
      "epoch": 0.9328018223234624,
      "grad_norm": 0.9130673408508301,
      "learning_rate": 1.2647286060472364e-05,
      "loss": 0.6035,
      "step": 7371
    },
    {
      "epoch": 0.9329283725639079,
      "grad_norm": 0.9084006547927856,
      "learning_rate": 1.264555783205052e-05,
      "loss": 0.5829,
      "step": 7372
    },
    {
      "epoch": 0.9330549228043533,
      "grad_norm": 1.0041700601577759,
      "learning_rate": 1.264382951866112e-05,
      "loss": 0.5625,
      "step": 7373
    },
    {
      "epoch": 0.9331814730447988,
      "grad_norm": 1.0332499742507935,
      "learning_rate": 1.2642101120359658e-05,
      "loss": 0.5663,
      "step": 7374
    },
    {
      "epoch": 0.9333080232852442,
      "grad_norm": 1.090165615081787,
      "learning_rate": 1.264037263720165e-05,
      "loss": 0.5911,
      "step": 7375
    },
    {
      "epoch": 0.9334345735256897,
      "grad_norm": 0.9901909828186035,
      "learning_rate": 1.2638644069242612e-05,
      "loss": 0.6145,
      "step": 7376
    },
    {
      "epoch": 0.9335611237661352,
      "grad_norm": 0.9382426738739014,
      "learning_rate": 1.2636915416538056e-05,
      "loss": 0.6114,
      "step": 7377
    },
    {
      "epoch": 0.9336876740065806,
      "grad_norm": 0.8841835260391235,
      "learning_rate": 1.2635186679143508e-05,
      "loss": 0.6302,
      "step": 7378
    },
    {
      "epoch": 0.9338142242470261,
      "grad_norm": 0.9200120568275452,
      "learning_rate": 1.2633457857114487e-05,
      "loss": 0.6201,
      "step": 7379
    },
    {
      "epoch": 0.9339407744874715,
      "grad_norm": 0.8929511904716492,
      "learning_rate": 1.2631728950506514e-05,
      "loss": 0.5905,
      "step": 7380
    },
    {
      "epoch": 0.934067324727917,
      "grad_norm": 0.8834142088890076,
      "learning_rate": 1.2629999959375118e-05,
      "loss": 0.5566,
      "step": 7381
    },
    {
      "epoch": 0.9341938749683625,
      "grad_norm": 0.9297951459884644,
      "learning_rate": 1.2628270883775832e-05,
      "loss": 0.6395,
      "step": 7382
    },
    {
      "epoch": 0.9343204252088079,
      "grad_norm": 0.9358946084976196,
      "learning_rate": 1.2626541723764189e-05,
      "loss": 0.6189,
      "step": 7383
    },
    {
      "epoch": 0.9344469754492534,
      "grad_norm": 0.8587300181388855,
      "learning_rate": 1.2624812479395723e-05,
      "loss": 0.5989,
      "step": 7384
    },
    {
      "epoch": 0.9345735256896988,
      "grad_norm": 0.9011515974998474,
      "learning_rate": 1.2623083150725972e-05,
      "loss": 0.571,
      "step": 7385
    },
    {
      "epoch": 0.9347000759301443,
      "grad_norm": 0.8395063877105713,
      "learning_rate": 1.2621353737810475e-05,
      "loss": 0.5188,
      "step": 7386
    },
    {
      "epoch": 0.9348266261705898,
      "grad_norm": 0.8623970746994019,
      "learning_rate": 1.2619624240704777e-05,
      "loss": 0.5611,
      "step": 7387
    },
    {
      "epoch": 0.9349531764110351,
      "grad_norm": 0.8870042562484741,
      "learning_rate": 1.2617894659464427e-05,
      "loss": 0.5886,
      "step": 7388
    },
    {
      "epoch": 0.9350797266514806,
      "grad_norm": 0.8967263698577881,
      "learning_rate": 1.261616499414497e-05,
      "loss": 0.5515,
      "step": 7389
    },
    {
      "epoch": 0.9352062768919261,
      "grad_norm": 0.871242880821228,
      "learning_rate": 1.2614435244801964e-05,
      "loss": 0.5385,
      "step": 7390
    },
    {
      "epoch": 0.9353328271323715,
      "grad_norm": 0.9068472981452942,
      "learning_rate": 1.2612705411490958e-05,
      "loss": 0.6181,
      "step": 7391
    },
    {
      "epoch": 0.935459377372817,
      "grad_norm": 0.8572544455528259,
      "learning_rate": 1.2610975494267508e-05,
      "loss": 0.5882,
      "step": 7392
    },
    {
      "epoch": 0.9355859276132624,
      "grad_norm": 0.8853620886802673,
      "learning_rate": 1.2609245493187179e-05,
      "loss": 0.5757,
      "step": 7393
    },
    {
      "epoch": 0.9357124778537079,
      "grad_norm": 0.9183065295219421,
      "learning_rate": 1.2607515408305529e-05,
      "loss": 0.6201,
      "step": 7394
    },
    {
      "epoch": 0.9358390280941534,
      "grad_norm": 0.9437170028686523,
      "learning_rate": 1.2605785239678128e-05,
      "loss": 0.6118,
      "step": 7395
    },
    {
      "epoch": 0.9359655783345988,
      "grad_norm": 0.9058578014373779,
      "learning_rate": 1.260405498736054e-05,
      "loss": 0.6365,
      "step": 7396
    },
    {
      "epoch": 0.9360921285750443,
      "grad_norm": 0.9491403698921204,
      "learning_rate": 1.2602324651408337e-05,
      "loss": 0.5941,
      "step": 7397
    },
    {
      "epoch": 0.9362186788154897,
      "grad_norm": 0.8923918008804321,
      "learning_rate": 1.2600594231877092e-05,
      "loss": 0.568,
      "step": 7398
    },
    {
      "epoch": 0.9363452290559352,
      "grad_norm": 0.9068533778190613,
      "learning_rate": 1.2598863728822379e-05,
      "loss": 0.6192,
      "step": 7399
    },
    {
      "epoch": 0.9364717792963807,
      "grad_norm": 1.0132005214691162,
      "learning_rate": 1.2597133142299782e-05,
      "loss": 0.6469,
      "step": 7400
    },
    {
      "epoch": 0.9365983295368261,
      "grad_norm": 0.9065911769866943,
      "learning_rate": 1.2595402472364877e-05,
      "loss": 0.6058,
      "step": 7401
    },
    {
      "epoch": 0.9367248797772716,
      "grad_norm": 0.8653408885002136,
      "learning_rate": 1.2593671719073255e-05,
      "loss": 0.5713,
      "step": 7402
    },
    {
      "epoch": 0.936851430017717,
      "grad_norm": 1.0130679607391357,
      "learning_rate": 1.2591940882480493e-05,
      "loss": 0.6135,
      "step": 7403
    },
    {
      "epoch": 0.9369779802581625,
      "grad_norm": 0.8937441110610962,
      "learning_rate": 1.2590209962642185e-05,
      "loss": 0.6165,
      "step": 7404
    },
    {
      "epoch": 0.937104530498608,
      "grad_norm": 0.9320510625839233,
      "learning_rate": 1.2588478959613926e-05,
      "loss": 0.6252,
      "step": 7405
    },
    {
      "epoch": 0.9372310807390534,
      "grad_norm": 0.9559963345527649,
      "learning_rate": 1.2586747873451308e-05,
      "loss": 0.5896,
      "step": 7406
    },
    {
      "epoch": 0.9373576309794989,
      "grad_norm": 0.8599642515182495,
      "learning_rate": 1.2585016704209929e-05,
      "loss": 0.5712,
      "step": 7407
    },
    {
      "epoch": 0.9374841812199444,
      "grad_norm": 0.9260480999946594,
      "learning_rate": 1.2583285451945389e-05,
      "loss": 0.5941,
      "step": 7408
    },
    {
      "epoch": 0.9376107314603898,
      "grad_norm": 0.8962565064430237,
      "learning_rate": 1.258155411671329e-05,
      "loss": 0.5528,
      "step": 7409
    },
    {
      "epoch": 0.9377372817008353,
      "grad_norm": 0.9080045223236084,
      "learning_rate": 1.2579822698569237e-05,
      "loss": 0.5697,
      "step": 7410
    },
    {
      "epoch": 0.9378638319412806,
      "grad_norm": 0.8880090713500977,
      "learning_rate": 1.257809119756884e-05,
      "loss": 0.5928,
      "step": 7411
    },
    {
      "epoch": 0.9379903821817261,
      "grad_norm": 0.9150490164756775,
      "learning_rate": 1.2576359613767709e-05,
      "loss": 0.5863,
      "step": 7412
    },
    {
      "epoch": 0.9381169324221716,
      "grad_norm": 0.9170653820037842,
      "learning_rate": 1.2574627947221458e-05,
      "loss": 0.57,
      "step": 7413
    },
    {
      "epoch": 0.938243482662617,
      "grad_norm": 0.8432600498199463,
      "learning_rate": 1.25728961979857e-05,
      "loss": 0.548,
      "step": 7414
    },
    {
      "epoch": 0.9383700329030625,
      "grad_norm": 0.9151769280433655,
      "learning_rate": 1.2571164366116055e-05,
      "loss": 0.6322,
      "step": 7415
    },
    {
      "epoch": 0.9384965831435079,
      "grad_norm": 1.0014894008636475,
      "learning_rate": 1.2569432451668147e-05,
      "loss": 0.5718,
      "step": 7416
    },
    {
      "epoch": 0.9386231333839534,
      "grad_norm": 1.3230270147323608,
      "learning_rate": 1.2567700454697598e-05,
      "loss": 0.6294,
      "step": 7417
    },
    {
      "epoch": 0.9387496836243989,
      "grad_norm": 0.8497980833053589,
      "learning_rate": 1.2565968375260033e-05,
      "loss": 0.5854,
      "step": 7418
    },
    {
      "epoch": 0.9388762338648443,
      "grad_norm": 0.908061146736145,
      "learning_rate": 1.2564236213411087e-05,
      "loss": 0.565,
      "step": 7419
    },
    {
      "epoch": 0.9390027841052898,
      "grad_norm": 0.9078131318092346,
      "learning_rate": 1.2562503969206388e-05,
      "loss": 0.5837,
      "step": 7420
    },
    {
      "epoch": 0.9391293343457353,
      "grad_norm": 0.9532117247581482,
      "learning_rate": 1.256077164270157e-05,
      "loss": 0.5908,
      "step": 7421
    },
    {
      "epoch": 0.9392558845861807,
      "grad_norm": 0.8799686431884766,
      "learning_rate": 1.255903923395227e-05,
      "loss": 0.5853,
      "step": 7422
    },
    {
      "epoch": 0.9393824348266262,
      "grad_norm": 0.9286990761756897,
      "learning_rate": 1.2557306743014129e-05,
      "loss": 0.5776,
      "step": 7423
    },
    {
      "epoch": 0.9395089850670716,
      "grad_norm": 0.9385512471199036,
      "learning_rate": 1.255557416994279e-05,
      "loss": 0.6169,
      "step": 7424
    },
    {
      "epoch": 0.9396355353075171,
      "grad_norm": 0.959147572517395,
      "learning_rate": 1.2553841514793898e-05,
      "loss": 0.6296,
      "step": 7425
    },
    {
      "epoch": 0.9397620855479626,
      "grad_norm": 0.9566076993942261,
      "learning_rate": 1.25521087776231e-05,
      "loss": 0.5936,
      "step": 7426
    },
    {
      "epoch": 0.939888635788408,
      "grad_norm": 0.891991376876831,
      "learning_rate": 1.2550375958486048e-05,
      "loss": 0.5716,
      "step": 7427
    },
    {
      "epoch": 0.9400151860288535,
      "grad_norm": 0.9928742051124573,
      "learning_rate": 1.2548643057438391e-05,
      "loss": 0.592,
      "step": 7428
    },
    {
      "epoch": 0.9401417362692989,
      "grad_norm": 1.1238821744918823,
      "learning_rate": 1.254691007453579e-05,
      "loss": 0.6455,
      "step": 7429
    },
    {
      "epoch": 0.9402682865097444,
      "grad_norm": 0.8256356716156006,
      "learning_rate": 1.25451770098339e-05,
      "loss": 0.5816,
      "step": 7430
    },
    {
      "epoch": 0.9403948367501899,
      "grad_norm": 0.9147263169288635,
      "learning_rate": 1.2543443863388385e-05,
      "loss": 0.5729,
      "step": 7431
    },
    {
      "epoch": 0.9405213869906353,
      "grad_norm": 0.8885672092437744,
      "learning_rate": 1.2541710635254905e-05,
      "loss": 0.5928,
      "step": 7432
    },
    {
      "epoch": 0.9406479372310808,
      "grad_norm": 0.8850177526473999,
      "learning_rate": 1.253997732548913e-05,
      "loss": 0.5526,
      "step": 7433
    },
    {
      "epoch": 0.9407744874715261,
      "grad_norm": 0.871637225151062,
      "learning_rate": 1.2538243934146724e-05,
      "loss": 0.5935,
      "step": 7434
    },
    {
      "epoch": 0.9409010377119716,
      "grad_norm": 0.9013028740882874,
      "learning_rate": 1.253651046128336e-05,
      "loss": 0.5881,
      "step": 7435
    },
    {
      "epoch": 0.9410275879524171,
      "grad_norm": 0.9015315175056458,
      "learning_rate": 1.2534776906954718e-05,
      "loss": 0.5983,
      "step": 7436
    },
    {
      "epoch": 0.9411541381928625,
      "grad_norm": 0.8670211434364319,
      "learning_rate": 1.2533043271216464e-05,
      "loss": 0.5977,
      "step": 7437
    },
    {
      "epoch": 0.941280688433308,
      "grad_norm": 0.9188063144683838,
      "learning_rate": 1.2531309554124286e-05,
      "loss": 0.5596,
      "step": 7438
    },
    {
      "epoch": 0.9414072386737535,
      "grad_norm": 1.1337196826934814,
      "learning_rate": 1.2529575755733862e-05,
      "loss": 0.6166,
      "step": 7439
    },
    {
      "epoch": 0.9415337889141989,
      "grad_norm": 0.9465997219085693,
      "learning_rate": 1.252784187610088e-05,
      "loss": 0.545,
      "step": 7440
    },
    {
      "epoch": 0.9416603391546444,
      "grad_norm": 0.9375872611999512,
      "learning_rate": 1.252610791528102e-05,
      "loss": 0.5723,
      "step": 7441
    },
    {
      "epoch": 0.9417868893950898,
      "grad_norm": 0.888317883014679,
      "learning_rate": 1.2524373873329982e-05,
      "loss": 0.5928,
      "step": 7442
    },
    {
      "epoch": 0.9419134396355353,
      "grad_norm": 1.0016491413116455,
      "learning_rate": 1.2522639750303451e-05,
      "loss": 0.6128,
      "step": 7443
    },
    {
      "epoch": 0.9420399898759808,
      "grad_norm": 0.9099216461181641,
      "learning_rate": 1.252090554625712e-05,
      "loss": 0.6079,
      "step": 7444
    },
    {
      "epoch": 0.9421665401164262,
      "grad_norm": 1.1047391891479492,
      "learning_rate": 1.2519171261246697e-05,
      "loss": 0.6111,
      "step": 7445
    },
    {
      "epoch": 0.9422930903568717,
      "grad_norm": 0.8933492302894592,
      "learning_rate": 1.251743689532787e-05,
      "loss": 0.544,
      "step": 7446
    },
    {
      "epoch": 0.9424196405973171,
      "grad_norm": 0.8433326482772827,
      "learning_rate": 1.2515702448556348e-05,
      "loss": 0.5708,
      "step": 7447
    },
    {
      "epoch": 0.9425461908377626,
      "grad_norm": 0.9337080717086792,
      "learning_rate": 1.251396792098784e-05,
      "loss": 0.6104,
      "step": 7448
    },
    {
      "epoch": 0.9426727410782081,
      "grad_norm": 1.0388940572738647,
      "learning_rate": 1.2512233312678044e-05,
      "loss": 0.6194,
      "step": 7449
    },
    {
      "epoch": 0.9427992913186535,
      "grad_norm": 0.9423564672470093,
      "learning_rate": 1.2510498623682676e-05,
      "loss": 0.5955,
      "step": 7450
    },
    {
      "epoch": 0.942925841559099,
      "grad_norm": 0.9651192426681519,
      "learning_rate": 1.250876385405745e-05,
      "loss": 0.5985,
      "step": 7451
    },
    {
      "epoch": 0.9430523917995444,
      "grad_norm": 0.8382248878479004,
      "learning_rate": 1.2507029003858083e-05,
      "loss": 0.5612,
      "step": 7452
    },
    {
      "epoch": 0.9431789420399899,
      "grad_norm": 0.8848263621330261,
      "learning_rate": 1.2505294073140291e-05,
      "loss": 0.5743,
      "step": 7453
    },
    {
      "epoch": 0.9433054922804354,
      "grad_norm": 0.8764775395393372,
      "learning_rate": 1.2503559061959794e-05,
      "loss": 0.5926,
      "step": 7454
    },
    {
      "epoch": 0.9434320425208808,
      "grad_norm": 0.8708125352859497,
      "learning_rate": 1.2501823970372316e-05,
      "loss": 0.5667,
      "step": 7455
    },
    {
      "epoch": 0.9435585927613263,
      "grad_norm": 0.9903971552848816,
      "learning_rate": 1.2500088798433583e-05,
      "loss": 0.6138,
      "step": 7456
    },
    {
      "epoch": 0.9436851430017718,
      "grad_norm": 0.8256422281265259,
      "learning_rate": 1.2498353546199327e-05,
      "loss": 0.5833,
      "step": 7457
    },
    {
      "epoch": 0.9438116932422171,
      "grad_norm": 0.9379015564918518,
      "learning_rate": 1.2496618213725273e-05,
      "loss": 0.5687,
      "step": 7458
    },
    {
      "epoch": 0.9439382434826626,
      "grad_norm": 0.8920388221740723,
      "learning_rate": 1.2494882801067163e-05,
      "loss": 0.5774,
      "step": 7459
    },
    {
      "epoch": 0.944064793723108,
      "grad_norm": 0.9215178489685059,
      "learning_rate": 1.2493147308280724e-05,
      "loss": 0.6034,
      "step": 7460
    },
    {
      "epoch": 0.9441913439635535,
      "grad_norm": 0.8974180221557617,
      "learning_rate": 1.2491411735421702e-05,
      "loss": 0.5992,
      "step": 7461
    },
    {
      "epoch": 0.944317894203999,
      "grad_norm": 0.9229962229728699,
      "learning_rate": 1.2489676082545833e-05,
      "loss": 0.6263,
      "step": 7462
    },
    {
      "epoch": 0.9444444444444444,
      "grad_norm": 0.8790378570556641,
      "learning_rate": 1.2487940349708868e-05,
      "loss": 0.5585,
      "step": 7463
    },
    {
      "epoch": 0.9445709946848899,
      "grad_norm": 0.8361641764640808,
      "learning_rate": 1.2486204536966547e-05,
      "loss": 0.5276,
      "step": 7464
    },
    {
      "epoch": 0.9446975449253353,
      "grad_norm": 0.9665467143058777,
      "learning_rate": 1.2484468644374624e-05,
      "loss": 0.6103,
      "step": 7465
    },
    {
      "epoch": 0.9448240951657808,
      "grad_norm": 0.8910186886787415,
      "learning_rate": 1.2482732671988846e-05,
      "loss": 0.6467,
      "step": 7466
    },
    {
      "epoch": 0.9449506454062263,
      "grad_norm": 0.8751964569091797,
      "learning_rate": 1.248099661986497e-05,
      "loss": 0.5589,
      "step": 7467
    },
    {
      "epoch": 0.9450771956466717,
      "grad_norm": 0.8556379675865173,
      "learning_rate": 1.2479260488058756e-05,
      "loss": 0.5696,
      "step": 7468
    },
    {
      "epoch": 0.9452037458871172,
      "grad_norm": 0.9024743437767029,
      "learning_rate": 1.247752427662596e-05,
      "loss": 0.5773,
      "step": 7469
    },
    {
      "epoch": 0.9453302961275627,
      "grad_norm": 0.8644241690635681,
      "learning_rate": 1.2475787985622345e-05,
      "loss": 0.5683,
      "step": 7470
    },
    {
      "epoch": 0.9454568463680081,
      "grad_norm": 0.8488756418228149,
      "learning_rate": 1.2474051615103675e-05,
      "loss": 0.5331,
      "step": 7471
    },
    {
      "epoch": 0.9455833966084536,
      "grad_norm": 0.9615640640258789,
      "learning_rate": 1.2472315165125717e-05,
      "loss": 0.5871,
      "step": 7472
    },
    {
      "epoch": 0.945709946848899,
      "grad_norm": 0.8821071982383728,
      "learning_rate": 1.247057863574424e-05,
      "loss": 0.5365,
      "step": 7473
    },
    {
      "epoch": 0.9458364970893445,
      "grad_norm": 0.859182596206665,
      "learning_rate": 1.2468842027015016e-05,
      "loss": 0.5879,
      "step": 7474
    },
    {
      "epoch": 0.94596304732979,
      "grad_norm": 0.8772773742675781,
      "learning_rate": 1.246710533899382e-05,
      "loss": 0.568,
      "step": 7475
    },
    {
      "epoch": 0.9460895975702354,
      "grad_norm": 0.8491340279579163,
      "learning_rate": 1.2465368571736433e-05,
      "loss": 0.5459,
      "step": 7476
    },
    {
      "epoch": 0.9462161478106809,
      "grad_norm": 0.9315661787986755,
      "learning_rate": 1.2463631725298633e-05,
      "loss": 0.6138,
      "step": 7477
    },
    {
      "epoch": 0.9463426980511263,
      "grad_norm": 0.909738302230835,
      "learning_rate": 1.24618947997362e-05,
      "loss": 0.5591,
      "step": 7478
    },
    {
      "epoch": 0.9464692482915718,
      "grad_norm": 0.9052975177764893,
      "learning_rate": 1.2460157795104922e-05,
      "loss": 0.6489,
      "step": 7479
    },
    {
      "epoch": 0.9465957985320173,
      "grad_norm": 0.8451615571975708,
      "learning_rate": 1.2458420711460581e-05,
      "loss": 0.5774,
      "step": 7480
    },
    {
      "epoch": 0.9467223487724626,
      "grad_norm": 0.8562792539596558,
      "learning_rate": 1.2456683548858977e-05,
      "loss": 0.5992,
      "step": 7481
    },
    {
      "epoch": 0.9468488990129081,
      "grad_norm": 0.8831309080123901,
      "learning_rate": 1.2454946307355892e-05,
      "loss": 0.5796,
      "step": 7482
    },
    {
      "epoch": 0.9469754492533535,
      "grad_norm": 0.8853042125701904,
      "learning_rate": 1.2453208987007129e-05,
      "loss": 0.6044,
      "step": 7483
    },
    {
      "epoch": 0.947101999493799,
      "grad_norm": 0.9199261665344238,
      "learning_rate": 1.245147158786848e-05,
      "loss": 0.6019,
      "step": 7484
    },
    {
      "epoch": 0.9472285497342445,
      "grad_norm": 0.9342353343963623,
      "learning_rate": 1.2449734109995748e-05,
      "loss": 0.5943,
      "step": 7485
    },
    {
      "epoch": 0.9473550999746899,
      "grad_norm": 0.895170271396637,
      "learning_rate": 1.2447996553444735e-05,
      "loss": 0.5935,
      "step": 7486
    },
    {
      "epoch": 0.9474816502151354,
      "grad_norm": 0.8344471454620361,
      "learning_rate": 1.2446258918271245e-05,
      "loss": 0.5863,
      "step": 7487
    },
    {
      "epoch": 0.9476082004555809,
      "grad_norm": 0.9129809141159058,
      "learning_rate": 1.244452120453109e-05,
      "loss": 0.586,
      "step": 7488
    },
    {
      "epoch": 0.9477347506960263,
      "grad_norm": 0.8784977793693542,
      "learning_rate": 1.2442783412280077e-05,
      "loss": 0.5863,
      "step": 7489
    },
    {
      "epoch": 0.9478613009364718,
      "grad_norm": 0.846532940864563,
      "learning_rate": 1.244104554157402e-05,
      "loss": 0.5807,
      "step": 7490
    },
    {
      "epoch": 0.9479878511769172,
      "grad_norm": 0.8482547998428345,
      "learning_rate": 1.243930759246873e-05,
      "loss": 0.54,
      "step": 7491
    },
    {
      "epoch": 0.9481144014173627,
      "grad_norm": 0.8746203780174255,
      "learning_rate": 1.243756956502003e-05,
      "loss": 0.598,
      "step": 7492
    },
    {
      "epoch": 0.9482409516578082,
      "grad_norm": 0.8758103251457214,
      "learning_rate": 1.2435831459283741e-05,
      "loss": 0.6115,
      "step": 7493
    },
    {
      "epoch": 0.9483675018982536,
      "grad_norm": 0.8517273664474487,
      "learning_rate": 1.2434093275315682e-05,
      "loss": 0.5914,
      "step": 7494
    },
    {
      "epoch": 0.9484940521386991,
      "grad_norm": 0.8622626662254333,
      "learning_rate": 1.2432355013171683e-05,
      "loss": 0.5839,
      "step": 7495
    },
    {
      "epoch": 0.9486206023791445,
      "grad_norm": 0.8622676134109497,
      "learning_rate": 1.2430616672907566e-05,
      "loss": 0.5521,
      "step": 7496
    },
    {
      "epoch": 0.94874715261959,
      "grad_norm": 0.923340916633606,
      "learning_rate": 1.2428878254579163e-05,
      "loss": 0.5959,
      "step": 7497
    },
    {
      "epoch": 0.9488737028600355,
      "grad_norm": 0.9104349613189697,
      "learning_rate": 1.2427139758242309e-05,
      "loss": 0.6009,
      "step": 7498
    },
    {
      "epoch": 0.9490002531004809,
      "grad_norm": 0.8392719030380249,
      "learning_rate": 1.242540118395284e-05,
      "loss": 0.5704,
      "step": 7499
    },
    {
      "epoch": 0.9491268033409264,
      "grad_norm": 0.8744332194328308,
      "learning_rate": 1.2423662531766592e-05,
      "loss": 0.5305,
      "step": 7500
    },
    {
      "epoch": 0.9492533535813718,
      "grad_norm": 0.8848696947097778,
      "learning_rate": 1.2421923801739404e-05,
      "loss": 0.6077,
      "step": 7501
    },
    {
      "epoch": 0.9493799038218173,
      "grad_norm": 0.876089870929718,
      "learning_rate": 1.2420184993927122e-05,
      "loss": 0.5813,
      "step": 7502
    },
    {
      "epoch": 0.9495064540622628,
      "grad_norm": 0.8567616939544678,
      "learning_rate": 1.2418446108385591e-05,
      "loss": 0.5688,
      "step": 7503
    },
    {
      "epoch": 0.9496330043027081,
      "grad_norm": 0.895741879940033,
      "learning_rate": 1.2416707145170656e-05,
      "loss": 0.6042,
      "step": 7504
    },
    {
      "epoch": 0.9497595545431536,
      "grad_norm": 0.8758161664009094,
      "learning_rate": 1.2414968104338174e-05,
      "loss": 0.619,
      "step": 7505
    },
    {
      "epoch": 0.9498861047835991,
      "grad_norm": 0.8950083255767822,
      "learning_rate": 1.241322898594399e-05,
      "loss": 0.599,
      "step": 7506
    },
    {
      "epoch": 0.9500126550240445,
      "grad_norm": 0.9140344858169556,
      "learning_rate": 1.2411489790043967e-05,
      "loss": 0.602,
      "step": 7507
    },
    {
      "epoch": 0.95013920526449,
      "grad_norm": 0.8676614761352539,
      "learning_rate": 1.2409750516693954e-05,
      "loss": 0.5817,
      "step": 7508
    },
    {
      "epoch": 0.9502657555049354,
      "grad_norm": 0.8872243762016296,
      "learning_rate": 1.2408011165949816e-05,
      "loss": 0.5778,
      "step": 7509
    },
    {
      "epoch": 0.9503923057453809,
      "grad_norm": 0.937654435634613,
      "learning_rate": 1.240627173786742e-05,
      "loss": 0.5825,
      "step": 7510
    },
    {
      "epoch": 0.9505188559858264,
      "grad_norm": 0.9820090532302856,
      "learning_rate": 1.2404532232502625e-05,
      "loss": 0.5976,
      "step": 7511
    },
    {
      "epoch": 0.9506454062262718,
      "grad_norm": 0.8843904733657837,
      "learning_rate": 1.24027926499113e-05,
      "loss": 0.5585,
      "step": 7512
    },
    {
      "epoch": 0.9507719564667173,
      "grad_norm": 1.0023643970489502,
      "learning_rate": 1.2401052990149315e-05,
      "loss": 0.5862,
      "step": 7513
    },
    {
      "epoch": 0.9508985067071627,
      "grad_norm": 1.0110279321670532,
      "learning_rate": 1.2399313253272547e-05,
      "loss": 0.62,
      "step": 7514
    },
    {
      "epoch": 0.9510250569476082,
      "grad_norm": 0.8873618245124817,
      "learning_rate": 1.2397573439336868e-05,
      "loss": 0.5815,
      "step": 7515
    },
    {
      "epoch": 0.9511516071880537,
      "grad_norm": 0.946996808052063,
      "learning_rate": 1.2395833548398156e-05,
      "loss": 0.572,
      "step": 7516
    },
    {
      "epoch": 0.9512781574284991,
      "grad_norm": 0.8815979957580566,
      "learning_rate": 1.2394093580512292e-05,
      "loss": 0.5763,
      "step": 7517
    },
    {
      "epoch": 0.9514047076689446,
      "grad_norm": 0.8564516305923462,
      "learning_rate": 1.2392353535735158e-05,
      "loss": 0.5583,
      "step": 7518
    },
    {
      "epoch": 0.9515312579093901,
      "grad_norm": 0.847163200378418,
      "learning_rate": 1.2390613414122638e-05,
      "loss": 0.5775,
      "step": 7519
    },
    {
      "epoch": 0.9516578081498355,
      "grad_norm": 0.8686860203742981,
      "learning_rate": 1.2388873215730622e-05,
      "loss": 0.5679,
      "step": 7520
    },
    {
      "epoch": 0.951784358390281,
      "grad_norm": 0.8822721242904663,
      "learning_rate": 1.2387132940614997e-05,
      "loss": 0.5672,
      "step": 7521
    },
    {
      "epoch": 0.9519109086307264,
      "grad_norm": 0.9150294065475464,
      "learning_rate": 1.2385392588831657e-05,
      "loss": 0.5958,
      "step": 7522
    },
    {
      "epoch": 0.9520374588711719,
      "grad_norm": 0.9498109221458435,
      "learning_rate": 1.23836521604365e-05,
      "loss": 0.6003,
      "step": 7523
    },
    {
      "epoch": 0.9521640091116174,
      "grad_norm": 0.8238849639892578,
      "learning_rate": 1.2381911655485419e-05,
      "loss": 0.5792,
      "step": 7524
    },
    {
      "epoch": 0.9522905593520627,
      "grad_norm": 0.862092912197113,
      "learning_rate": 1.2380171074034315e-05,
      "loss": 0.5672,
      "step": 7525
    },
    {
      "epoch": 0.9524171095925082,
      "grad_norm": 0.9008827209472656,
      "learning_rate": 1.2378430416139091e-05,
      "loss": 0.5725,
      "step": 7526
    },
    {
      "epoch": 0.9525436598329536,
      "grad_norm": 0.9074214100837708,
      "learning_rate": 1.2376689681855654e-05,
      "loss": 0.6087,
      "step": 7527
    },
    {
      "epoch": 0.9526702100733991,
      "grad_norm": 0.8644921183586121,
      "learning_rate": 1.237494887123991e-05,
      "loss": 0.575,
      "step": 7528
    },
    {
      "epoch": 0.9527967603138446,
      "grad_norm": 0.8743188977241516,
      "learning_rate": 1.2373207984347765e-05,
      "loss": 0.5883,
      "step": 7529
    },
    {
      "epoch": 0.95292331055429,
      "grad_norm": 0.858798086643219,
      "learning_rate": 1.2371467021235134e-05,
      "loss": 0.57,
      "step": 7530
    },
    {
      "epoch": 0.9530498607947355,
      "grad_norm": 1.0050075054168701,
      "learning_rate": 1.2369725981957936e-05,
      "loss": 0.6304,
      "step": 7531
    },
    {
      "epoch": 0.9531764110351809,
      "grad_norm": 0.9630895853042603,
      "learning_rate": 1.236798486657208e-05,
      "loss": 0.5999,
      "step": 7532
    },
    {
      "epoch": 0.9533029612756264,
      "grad_norm": 0.8628865480422974,
      "learning_rate": 1.2366243675133492e-05,
      "loss": 0.5807,
      "step": 7533
    },
    {
      "epoch": 0.9534295115160719,
      "grad_norm": 0.9301726222038269,
      "learning_rate": 1.2364502407698091e-05,
      "loss": 0.6039,
      "step": 7534
    },
    {
      "epoch": 0.9535560617565173,
      "grad_norm": 0.9060618281364441,
      "learning_rate": 1.23627610643218e-05,
      "loss": 0.6162,
      "step": 7535
    },
    {
      "epoch": 0.9536826119969628,
      "grad_norm": 0.8781424760818481,
      "learning_rate": 1.2361019645060548e-05,
      "loss": 0.5868,
      "step": 7536
    },
    {
      "epoch": 0.9538091622374083,
      "grad_norm": 0.9497135877609253,
      "learning_rate": 1.2359278149970262e-05,
      "loss": 0.5647,
      "step": 7537
    },
    {
      "epoch": 0.9539357124778537,
      "grad_norm": 0.8606073260307312,
      "learning_rate": 1.2357536579106877e-05,
      "loss": 0.5977,
      "step": 7538
    },
    {
      "epoch": 0.9540622627182992,
      "grad_norm": 0.9087311029434204,
      "learning_rate": 1.2355794932526325e-05,
      "loss": 0.567,
      "step": 7539
    },
    {
      "epoch": 0.9541888129587446,
      "grad_norm": 1.0798110961914062,
      "learning_rate": 1.2354053210284547e-05,
      "loss": 0.5541,
      "step": 7540
    },
    {
      "epoch": 0.9543153631991901,
      "grad_norm": 0.8887361288070679,
      "learning_rate": 1.2352311412437476e-05,
      "loss": 0.584,
      "step": 7541
    },
    {
      "epoch": 0.9544419134396356,
      "grad_norm": 0.9766591787338257,
      "learning_rate": 1.2350569539041055e-05,
      "loss": 0.6305,
      "step": 7542
    },
    {
      "epoch": 0.954568463680081,
      "grad_norm": 0.895287811756134,
      "learning_rate": 1.234882759015123e-05,
      "loss": 0.5908,
      "step": 7543
    },
    {
      "epoch": 0.9546950139205265,
      "grad_norm": 0.8996095061302185,
      "learning_rate": 1.2347085565823946e-05,
      "loss": 0.6254,
      "step": 7544
    },
    {
      "epoch": 0.9548215641609719,
      "grad_norm": 0.9523221254348755,
      "learning_rate": 1.2345343466115151e-05,
      "loss": 0.6092,
      "step": 7545
    },
    {
      "epoch": 0.9549481144014174,
      "grad_norm": 0.8779715299606323,
      "learning_rate": 1.2343601291080798e-05,
      "loss": 0.6224,
      "step": 7546
    },
    {
      "epoch": 0.9550746646418629,
      "grad_norm": 0.9207479357719421,
      "learning_rate": 1.2341859040776838e-05,
      "loss": 0.5784,
      "step": 7547
    },
    {
      "epoch": 0.9552012148823082,
      "grad_norm": 0.8776651620864868,
      "learning_rate": 1.2340116715259226e-05,
      "loss": 0.5467,
      "step": 7548
    },
    {
      "epoch": 0.9553277651227537,
      "grad_norm": 0.8607171177864075,
      "learning_rate": 1.2338374314583928e-05,
      "loss": 0.5984,
      "step": 7549
    },
    {
      "epoch": 0.9554543153631991,
      "grad_norm": 0.9702886343002319,
      "learning_rate": 1.2336631838806897e-05,
      "loss": 0.5661,
      "step": 7550
    },
    {
      "epoch": 0.9555808656036446,
      "grad_norm": 0.8991416096687317,
      "learning_rate": 1.2334889287984098e-05,
      "loss": 0.6007,
      "step": 7551
    },
    {
      "epoch": 0.9557074158440901,
      "grad_norm": 0.8746899962425232,
      "learning_rate": 1.2333146662171498e-05,
      "loss": 0.568,
      "step": 7552
    },
    {
      "epoch": 0.9558339660845355,
      "grad_norm": 1.0007355213165283,
      "learning_rate": 1.2331403961425064e-05,
      "loss": 0.5687,
      "step": 7553
    },
    {
      "epoch": 0.955960516324981,
      "grad_norm": 0.9026124477386475,
      "learning_rate": 1.2329661185800768e-05,
      "loss": 0.5871,
      "step": 7554
    },
    {
      "epoch": 0.9560870665654265,
      "grad_norm": 0.9686092734336853,
      "learning_rate": 1.2327918335354581e-05,
      "loss": 0.5822,
      "step": 7555
    },
    {
      "epoch": 0.9562136168058719,
      "grad_norm": 0.853834331035614,
      "learning_rate": 1.2326175410142478e-05,
      "loss": 0.547,
      "step": 7556
    },
    {
      "epoch": 0.9563401670463174,
      "grad_norm": 0.8614591360092163,
      "learning_rate": 1.2324432410220441e-05,
      "loss": 0.5935,
      "step": 7557
    },
    {
      "epoch": 0.9564667172867628,
      "grad_norm": 0.8975144028663635,
      "learning_rate": 1.2322689335644444e-05,
      "loss": 0.6041,
      "step": 7558
    },
    {
      "epoch": 0.9565932675272083,
      "grad_norm": 0.8652198910713196,
      "learning_rate": 1.232094618647047e-05,
      "loss": 0.5935,
      "step": 7559
    },
    {
      "epoch": 0.9567198177676538,
      "grad_norm": 0.886644721031189,
      "learning_rate": 1.2319202962754508e-05,
      "loss": 0.6066,
      "step": 7560
    },
    {
      "epoch": 0.9568463680080992,
      "grad_norm": 0.8960562348365784,
      "learning_rate": 1.2317459664552543e-05,
      "loss": 0.5839,
      "step": 7561
    },
    {
      "epoch": 0.9569729182485447,
      "grad_norm": 0.9364986419677734,
      "learning_rate": 1.2315716291920566e-05,
      "loss": 0.5682,
      "step": 7562
    },
    {
      "epoch": 0.9570994684889901,
      "grad_norm": 0.8747580051422119,
      "learning_rate": 1.2313972844914568e-05,
      "loss": 0.549,
      "step": 7563
    },
    {
      "epoch": 0.9572260187294356,
      "grad_norm": 0.925470232963562,
      "learning_rate": 1.2312229323590541e-05,
      "loss": 0.5949,
      "step": 7564
    },
    {
      "epoch": 0.9573525689698811,
      "grad_norm": 0.9899237751960754,
      "learning_rate": 1.2310485728004485e-05,
      "loss": 0.5619,
      "step": 7565
    },
    {
      "epoch": 0.9574791192103265,
      "grad_norm": 0.9679496884346008,
      "learning_rate": 1.2308742058212398e-05,
      "loss": 0.5851,
      "step": 7566
    },
    {
      "epoch": 0.957605669450772,
      "grad_norm": 0.8789438009262085,
      "learning_rate": 1.2306998314270283e-05,
      "loss": 0.575,
      "step": 7567
    },
    {
      "epoch": 0.9577322196912175,
      "grad_norm": 0.8681644201278687,
      "learning_rate": 1.2305254496234142e-05,
      "loss": 0.5952,
      "step": 7568
    },
    {
      "epoch": 0.9578587699316629,
      "grad_norm": 0.8862312436103821,
      "learning_rate": 1.2303510604159982e-05,
      "loss": 0.5761,
      "step": 7569
    },
    {
      "epoch": 0.9579853201721084,
      "grad_norm": 0.8330593109130859,
      "learning_rate": 1.230176663810381e-05,
      "loss": 0.5691,
      "step": 7570
    },
    {
      "epoch": 0.9581118704125537,
      "grad_norm": 0.8774002194404602,
      "learning_rate": 1.230002259812164e-05,
      "loss": 0.5675,
      "step": 7571
    },
    {
      "epoch": 0.9582384206529992,
      "grad_norm": 0.8568604588508606,
      "learning_rate": 1.2298278484269485e-05,
      "loss": 0.5589,
      "step": 7572
    },
    {
      "epoch": 0.9583649708934447,
      "grad_norm": 0.8644806146621704,
      "learning_rate": 1.229653429660336e-05,
      "loss": 0.5772,
      "step": 7573
    },
    {
      "epoch": 0.9584915211338901,
      "grad_norm": 1.193619966506958,
      "learning_rate": 1.2294790035179285e-05,
      "loss": 0.5812,
      "step": 7574
    },
    {
      "epoch": 0.9586180713743356,
      "grad_norm": 0.8888121843338013,
      "learning_rate": 1.2293045700053277e-05,
      "loss": 0.5922,
      "step": 7575
    },
    {
      "epoch": 0.958744621614781,
      "grad_norm": 0.8808746933937073,
      "learning_rate": 1.2291301291281359e-05,
      "loss": 0.5696,
      "step": 7576
    },
    {
      "epoch": 0.9588711718552265,
      "grad_norm": 0.9046097993850708,
      "learning_rate": 1.2289556808919561e-05,
      "loss": 0.5718,
      "step": 7577
    },
    {
      "epoch": 0.958997722095672,
      "grad_norm": 0.8720799088478088,
      "learning_rate": 1.2287812253023906e-05,
      "loss": 0.5696,
      "step": 7578
    },
    {
      "epoch": 0.9591242723361174,
      "grad_norm": 0.908298671245575,
      "learning_rate": 1.2286067623650429e-05,
      "loss": 0.6176,
      "step": 7579
    },
    {
      "epoch": 0.9592508225765629,
      "grad_norm": 0.8357524275779724,
      "learning_rate": 1.2284322920855154e-05,
      "loss": 0.5607,
      "step": 7580
    },
    {
      "epoch": 0.9593773728170083,
      "grad_norm": 0.9587050676345825,
      "learning_rate": 1.2282578144694127e-05,
      "loss": 0.6225,
      "step": 7581
    },
    {
      "epoch": 0.9595039230574538,
      "grad_norm": 0.9387618899345398,
      "learning_rate": 1.2280833295223374e-05,
      "loss": 0.6055,
      "step": 7582
    },
    {
      "epoch": 0.9596304732978993,
      "grad_norm": 0.8999502062797546,
      "learning_rate": 1.227908837249894e-05,
      "loss": 0.6091,
      "step": 7583
    },
    {
      "epoch": 0.9597570235383447,
      "grad_norm": 0.9267122745513916,
      "learning_rate": 1.2277343376576867e-05,
      "loss": 0.5466,
      "step": 7584
    },
    {
      "epoch": 0.9598835737787902,
      "grad_norm": 0.8783663511276245,
      "learning_rate": 1.2275598307513197e-05,
      "loss": 0.5893,
      "step": 7585
    },
    {
      "epoch": 0.9600101240192357,
      "grad_norm": 0.927107036113739,
      "learning_rate": 1.2273853165363983e-05,
      "loss": 0.5927,
      "step": 7586
    },
    {
      "epoch": 0.9601366742596811,
      "grad_norm": 0.8814966678619385,
      "learning_rate": 1.2272107950185263e-05,
      "loss": 0.5503,
      "step": 7587
    },
    {
      "epoch": 0.9602632245001266,
      "grad_norm": 0.9119658470153809,
      "learning_rate": 1.2270362662033098e-05,
      "loss": 0.5784,
      "step": 7588
    },
    {
      "epoch": 0.960389774740572,
      "grad_norm": 0.9756544828414917,
      "learning_rate": 1.2268617300963532e-05,
      "loss": 0.5574,
      "step": 7589
    },
    {
      "epoch": 0.9605163249810175,
      "grad_norm": 0.8645396828651428,
      "learning_rate": 1.226687186703263e-05,
      "loss": 0.5277,
      "step": 7590
    },
    {
      "epoch": 0.960642875221463,
      "grad_norm": 0.8697201013565063,
      "learning_rate": 1.2265126360296448e-05,
      "loss": 0.553,
      "step": 7591
    },
    {
      "epoch": 0.9607694254619084,
      "grad_norm": 0.8776320219039917,
      "learning_rate": 1.2263380780811045e-05,
      "loss": 0.5472,
      "step": 7592
    },
    {
      "epoch": 0.9608959757023539,
      "grad_norm": 0.8698025941848755,
      "learning_rate": 1.2261635128632485e-05,
      "loss": 0.5896,
      "step": 7593
    },
    {
      "epoch": 0.9610225259427992,
      "grad_norm": 0.9272088408470154,
      "learning_rate": 1.225988940381683e-05,
      "loss": 0.633,
      "step": 7594
    },
    {
      "epoch": 0.9611490761832447,
      "grad_norm": 0.8944891691207886,
      "learning_rate": 1.225814360642015e-05,
      "loss": 0.6127,
      "step": 7595
    },
    {
      "epoch": 0.9612756264236902,
      "grad_norm": 0.9470350742340088,
      "learning_rate": 1.2256397736498514e-05,
      "loss": 0.5625,
      "step": 7596
    },
    {
      "epoch": 0.9614021766641356,
      "grad_norm": 0.876390814781189,
      "learning_rate": 1.2254651794107995e-05,
      "loss": 0.5765,
      "step": 7597
    },
    {
      "epoch": 0.9615287269045811,
      "grad_norm": 0.9403905272483826,
      "learning_rate": 1.2252905779304668e-05,
      "loss": 0.5581,
      "step": 7598
    },
    {
      "epoch": 0.9616552771450265,
      "grad_norm": 0.9202782511711121,
      "learning_rate": 1.2251159692144613e-05,
      "loss": 0.5953,
      "step": 7599
    },
    {
      "epoch": 0.961781827385472,
      "grad_norm": 0.9152198433876038,
      "learning_rate": 1.22494135326839e-05,
      "loss": 0.5661,
      "step": 7600
    },
    {
      "epoch": 0.9619083776259175,
      "grad_norm": 0.9098156690597534,
      "learning_rate": 1.2247667300978619e-05,
      "loss": 0.5614,
      "step": 7601
    },
    {
      "epoch": 0.9620349278663629,
      "grad_norm": 0.8701578974723816,
      "learning_rate": 1.224592099708485e-05,
      "loss": 0.5787,
      "step": 7602
    },
    {
      "epoch": 0.9621614781068084,
      "grad_norm": 0.8852734565734863,
      "learning_rate": 1.2244174621058681e-05,
      "loss": 0.6138,
      "step": 7603
    },
    {
      "epoch": 0.9622880283472539,
      "grad_norm": 0.9226470589637756,
      "learning_rate": 1.22424281729562e-05,
      "loss": 0.583,
      "step": 7604
    },
    {
      "epoch": 0.9624145785876993,
      "grad_norm": 0.885880708694458,
      "learning_rate": 1.2240681652833498e-05,
      "loss": 0.5612,
      "step": 7605
    },
    {
      "epoch": 0.9625411288281448,
      "grad_norm": 0.9011273384094238,
      "learning_rate": 1.2238935060746667e-05,
      "loss": 0.5772,
      "step": 7606
    },
    {
      "epoch": 0.9626676790685902,
      "grad_norm": 0.9082784056663513,
      "learning_rate": 1.22371883967518e-05,
      "loss": 0.5872,
      "step": 7607
    },
    {
      "epoch": 0.9627942293090357,
      "grad_norm": 0.9086839556694031,
      "learning_rate": 1.2235441660905001e-05,
      "loss": 0.5866,
      "step": 7608
    },
    {
      "epoch": 0.9629207795494812,
      "grad_norm": 0.9514667987823486,
      "learning_rate": 1.2233694853262366e-05,
      "loss": 0.6214,
      "step": 7609
    },
    {
      "epoch": 0.9630473297899266,
      "grad_norm": 0.8958361744880676,
      "learning_rate": 1.2231947973879998e-05,
      "loss": 0.5794,
      "step": 7610
    },
    {
      "epoch": 0.9631738800303721,
      "grad_norm": 0.8826557993888855,
      "learning_rate": 1.2230201022814002e-05,
      "loss": 0.5271,
      "step": 7611
    },
    {
      "epoch": 0.9633004302708175,
      "grad_norm": 0.9311051368713379,
      "learning_rate": 1.2228454000120483e-05,
      "loss": 0.5574,
      "step": 7612
    },
    {
      "epoch": 0.963426980511263,
      "grad_norm": 0.9701524376869202,
      "learning_rate": 1.2226706905855553e-05,
      "loss": 0.6315,
      "step": 7613
    },
    {
      "epoch": 0.9635535307517085,
      "grad_norm": 0.8438827395439148,
      "learning_rate": 1.2224959740075322e-05,
      "loss": 0.5424,
      "step": 7614
    },
    {
      "epoch": 0.9636800809921539,
      "grad_norm": 0.9425041079521179,
      "learning_rate": 1.2223212502835906e-05,
      "loss": 0.6273,
      "step": 7615
    },
    {
      "epoch": 0.9638066312325994,
      "grad_norm": 0.9427367448806763,
      "learning_rate": 1.2221465194193419e-05,
      "loss": 0.6113,
      "step": 7616
    },
    {
      "epoch": 0.9639331814730449,
      "grad_norm": 0.9251165390014648,
      "learning_rate": 1.2219717814203979e-05,
      "loss": 0.5396,
      "step": 7617
    },
    {
      "epoch": 0.9640597317134902,
      "grad_norm": 0.9176473617553711,
      "learning_rate": 1.2217970362923708e-05,
      "loss": 0.5906,
      "step": 7618
    },
    {
      "epoch": 0.9641862819539357,
      "grad_norm": 0.8891769647598267,
      "learning_rate": 1.2216222840408726e-05,
      "loss": 0.6174,
      "step": 7619
    },
    {
      "epoch": 0.9643128321943811,
      "grad_norm": 1.0642396211624146,
      "learning_rate": 1.2214475246715164e-05,
      "loss": 0.6462,
      "step": 7620
    },
    {
      "epoch": 0.9644393824348266,
      "grad_norm": 0.8939231038093567,
      "learning_rate": 1.221272758189915e-05,
      "loss": 0.5807,
      "step": 7621
    },
    {
      "epoch": 0.9645659326752721,
      "grad_norm": 0.8706002831459045,
      "learning_rate": 1.2210979846016804e-05,
      "loss": 0.5885,
      "step": 7622
    },
    {
      "epoch": 0.9646924829157175,
      "grad_norm": 0.8470277786254883,
      "learning_rate": 1.2209232039124266e-05,
      "loss": 0.5824,
      "step": 7623
    },
    {
      "epoch": 0.964819033156163,
      "grad_norm": 0.9789892435073853,
      "learning_rate": 1.220748416127767e-05,
      "loss": 0.6014,
      "step": 7624
    },
    {
      "epoch": 0.9649455833966084,
      "grad_norm": 0.8829630017280579,
      "learning_rate": 1.2205736212533151e-05,
      "loss": 0.5938,
      "step": 7625
    },
    {
      "epoch": 0.9650721336370539,
      "grad_norm": 0.8562055230140686,
      "learning_rate": 1.2203988192946849e-05,
      "loss": 0.591,
      "step": 7626
    },
    {
      "epoch": 0.9651986838774994,
      "grad_norm": 0.9347453117370605,
      "learning_rate": 1.2202240102574905e-05,
      "loss": 0.5784,
      "step": 7627
    },
    {
      "epoch": 0.9653252341179448,
      "grad_norm": 0.8661239743232727,
      "learning_rate": 1.2200491941473462e-05,
      "loss": 0.581,
      "step": 7628
    },
    {
      "epoch": 0.9654517843583903,
      "grad_norm": 0.9044026732444763,
      "learning_rate": 1.2198743709698669e-05,
      "loss": 0.5701,
      "step": 7629
    },
    {
      "epoch": 0.9655783345988357,
      "grad_norm": 0.87742018699646,
      "learning_rate": 1.2196995407306669e-05,
      "loss": 0.5958,
      "step": 7630
    },
    {
      "epoch": 0.9657048848392812,
      "grad_norm": 0.8366647958755493,
      "learning_rate": 1.2195247034353615e-05,
      "loss": 0.5882,
      "step": 7631
    },
    {
      "epoch": 0.9658314350797267,
      "grad_norm": 0.8705164194107056,
      "learning_rate": 1.219349859089566e-05,
      "loss": 0.5599,
      "step": 7632
    },
    {
      "epoch": 0.9659579853201721,
      "grad_norm": 1.0280089378356934,
      "learning_rate": 1.2191750076988958e-05,
      "loss": 0.5934,
      "step": 7633
    },
    {
      "epoch": 0.9660845355606176,
      "grad_norm": 0.8804489970207214,
      "learning_rate": 1.2190001492689664e-05,
      "loss": 0.6163,
      "step": 7634
    },
    {
      "epoch": 0.9662110858010631,
      "grad_norm": 0.8995026350021362,
      "learning_rate": 1.2188252838053943e-05,
      "loss": 0.5957,
      "step": 7635
    },
    {
      "epoch": 0.9663376360415085,
      "grad_norm": 0.8782361745834351,
      "learning_rate": 1.2186504113137952e-05,
      "loss": 0.5874,
      "step": 7636
    },
    {
      "epoch": 0.966464186281954,
      "grad_norm": 0.8594604730606079,
      "learning_rate": 1.2184755317997853e-05,
      "loss": 0.5801,
      "step": 7637
    },
    {
      "epoch": 0.9665907365223994,
      "grad_norm": 0.9391226172447205,
      "learning_rate": 1.2183006452689823e-05,
      "loss": 0.6192,
      "step": 7638
    },
    {
      "epoch": 0.9667172867628449,
      "grad_norm": 0.8396593928337097,
      "learning_rate": 1.2181257517270018e-05,
      "loss": 0.5871,
      "step": 7639
    },
    {
      "epoch": 0.9668438370032904,
      "grad_norm": 0.893108069896698,
      "learning_rate": 1.2179508511794613e-05,
      "loss": 0.5978,
      "step": 7640
    },
    {
      "epoch": 0.9669703872437357,
      "grad_norm": 0.8584726452827454,
      "learning_rate": 1.2177759436319785e-05,
      "loss": 0.5896,
      "step": 7641
    },
    {
      "epoch": 0.9670969374841812,
      "grad_norm": 0.8733391165733337,
      "learning_rate": 1.2176010290901703e-05,
      "loss": 0.6072,
      "step": 7642
    },
    {
      "epoch": 0.9672234877246266,
      "grad_norm": 0.8547091484069824,
      "learning_rate": 1.2174261075596547e-05,
      "loss": 0.5595,
      "step": 7643
    },
    {
      "epoch": 0.9673500379650721,
      "grad_norm": 0.8806403279304504,
      "learning_rate": 1.2172511790460502e-05,
      "loss": 0.5649,
      "step": 7644
    },
    {
      "epoch": 0.9674765882055176,
      "grad_norm": 0.8423972129821777,
      "learning_rate": 1.2170762435549739e-05,
      "loss": 0.5581,
      "step": 7645
    },
    {
      "epoch": 0.967603138445963,
      "grad_norm": 0.8579310178756714,
      "learning_rate": 1.2169013010920449e-05,
      "loss": 0.5792,
      "step": 7646
    },
    {
      "epoch": 0.9677296886864085,
      "grad_norm": 0.9088007211685181,
      "learning_rate": 1.2167263516628818e-05,
      "loss": 0.6184,
      "step": 7647
    },
    {
      "epoch": 0.9678562389268539,
      "grad_norm": 0.9065024852752686,
      "learning_rate": 1.2165513952731032e-05,
      "loss": 0.6291,
      "step": 7648
    },
    {
      "epoch": 0.9679827891672994,
      "grad_norm": 0.924679160118103,
      "learning_rate": 1.2163764319283286e-05,
      "loss": 0.5716,
      "step": 7649
    },
    {
      "epoch": 0.9681093394077449,
      "grad_norm": 0.908970832824707,
      "learning_rate": 1.216201461634177e-05,
      "loss": 0.598,
      "step": 7650
    },
    {
      "epoch": 0.9682358896481903,
      "grad_norm": 0.8680769801139832,
      "learning_rate": 1.216026484396268e-05,
      "loss": 0.5698,
      "step": 7651
    },
    {
      "epoch": 0.9683624398886358,
      "grad_norm": 0.9226531982421875,
      "learning_rate": 1.2158515002202216e-05,
      "loss": 0.5492,
      "step": 7652
    },
    {
      "epoch": 0.9684889901290813,
      "grad_norm": 0.8971876502037048,
      "learning_rate": 1.2156765091116575e-05,
      "loss": 0.5818,
      "step": 7653
    },
    {
      "epoch": 0.9686155403695267,
      "grad_norm": 0.939102292060852,
      "learning_rate": 1.2155015110761959e-05,
      "loss": 0.6153,
      "step": 7654
    },
    {
      "epoch": 0.9687420906099722,
      "grad_norm": 0.9017421007156372,
      "learning_rate": 1.2153265061194573e-05,
      "loss": 0.6318,
      "step": 7655
    },
    {
      "epoch": 0.9688686408504176,
      "grad_norm": 0.8473058342933655,
      "learning_rate": 1.2151514942470623e-05,
      "loss": 0.5404,
      "step": 7656
    },
    {
      "epoch": 0.9689951910908631,
      "grad_norm": 0.8757271766662598,
      "learning_rate": 1.2149764754646318e-05,
      "loss": 0.5508,
      "step": 7657
    },
    {
      "epoch": 0.9691217413313086,
      "grad_norm": 0.863818347454071,
      "learning_rate": 1.2148014497777868e-05,
      "loss": 0.5665,
      "step": 7658
    },
    {
      "epoch": 0.969248291571754,
      "grad_norm": 0.8559112548828125,
      "learning_rate": 1.2146264171921487e-05,
      "loss": 0.5542,
      "step": 7659
    },
    {
      "epoch": 0.9693748418121995,
      "grad_norm": 0.8316748738288879,
      "learning_rate": 1.2144513777133389e-05,
      "loss": 0.5666,
      "step": 7660
    },
    {
      "epoch": 0.9695013920526449,
      "grad_norm": 1.072278618812561,
      "learning_rate": 1.2142763313469796e-05,
      "loss": 0.6183,
      "step": 7661
    },
    {
      "epoch": 0.9696279422930904,
      "grad_norm": 0.894563615322113,
      "learning_rate": 1.2141012780986924e-05,
      "loss": 0.5762,
      "step": 7662
    },
    {
      "epoch": 0.9697544925335359,
      "grad_norm": 0.8891991972923279,
      "learning_rate": 1.2139262179740993e-05,
      "loss": 0.5766,
      "step": 7663
    },
    {
      "epoch": 0.9698810427739812,
      "grad_norm": 0.857686460018158,
      "learning_rate": 1.2137511509788233e-05,
      "loss": 0.5529,
      "step": 7664
    },
    {
      "epoch": 0.9700075930144267,
      "grad_norm": 0.851616382598877,
      "learning_rate": 1.2135760771184868e-05,
      "loss": 0.5744,
      "step": 7665
    },
    {
      "epoch": 0.9701341432548722,
      "grad_norm": 0.8730929493904114,
      "learning_rate": 1.2134009963987124e-05,
      "loss": 0.593,
      "step": 7666
    },
    {
      "epoch": 0.9702606934953176,
      "grad_norm": 0.8859708309173584,
      "learning_rate": 1.2132259088251235e-05,
      "loss": 0.6098,
      "step": 7667
    },
    {
      "epoch": 0.9703872437357631,
      "grad_norm": 0.9330201745033264,
      "learning_rate": 1.213050814403343e-05,
      "loss": 0.5855,
      "step": 7668
    },
    {
      "epoch": 0.9705137939762085,
      "grad_norm": 0.8877742886543274,
      "learning_rate": 1.2128757131389947e-05,
      "loss": 0.5638,
      "step": 7669
    },
    {
      "epoch": 0.970640344216654,
      "grad_norm": 0.8961247205734253,
      "learning_rate": 1.2127006050377024e-05,
      "loss": 0.6054,
      "step": 7670
    },
    {
      "epoch": 0.9707668944570995,
      "grad_norm": 0.9505306482315063,
      "learning_rate": 1.21252549010509e-05,
      "loss": 0.6141,
      "step": 7671
    },
    {
      "epoch": 0.9708934446975449,
      "grad_norm": 0.9406991600990295,
      "learning_rate": 1.2123503683467817e-05,
      "loss": 0.6326,
      "step": 7672
    },
    {
      "epoch": 0.9710199949379904,
      "grad_norm": 0.8540436029434204,
      "learning_rate": 1.2121752397684017e-05,
      "loss": 0.5736,
      "step": 7673
    },
    {
      "epoch": 0.9711465451784358,
      "grad_norm": 0.8945413827896118,
      "learning_rate": 1.2120001043755749e-05,
      "loss": 0.5867,
      "step": 7674
    },
    {
      "epoch": 0.9712730954188813,
      "grad_norm": 0.9277662634849548,
      "learning_rate": 1.211824962173926e-05,
      "loss": 0.5857,
      "step": 7675
    },
    {
      "epoch": 0.9713996456593268,
      "grad_norm": 0.9180772304534912,
      "learning_rate": 1.2116498131690799e-05,
      "loss": 0.5744,
      "step": 7676
    },
    {
      "epoch": 0.9715261958997722,
      "grad_norm": 0.916350781917572,
      "learning_rate": 1.2114746573666623e-05,
      "loss": 0.6146,
      "step": 7677
    },
    {
      "epoch": 0.9716527461402177,
      "grad_norm": 0.9099380373954773,
      "learning_rate": 1.2112994947722982e-05,
      "loss": 0.549,
      "step": 7678
    },
    {
      "epoch": 0.9717792963806631,
      "grad_norm": 0.9400988817214966,
      "learning_rate": 1.2111243253916138e-05,
      "loss": 0.6048,
      "step": 7679
    },
    {
      "epoch": 0.9719058466211086,
      "grad_norm": 0.9117671251296997,
      "learning_rate": 1.2109491492302346e-05,
      "loss": 0.6093,
      "step": 7680
    },
    {
      "epoch": 0.9720323968615541,
      "grad_norm": 0.8584964275360107,
      "learning_rate": 1.2107739662937867e-05,
      "loss": 0.5983,
      "step": 7681
    },
    {
      "epoch": 0.9721589471019995,
      "grad_norm": 0.8385818004608154,
      "learning_rate": 1.2105987765878968e-05,
      "loss": 0.5462,
      "step": 7682
    },
    {
      "epoch": 0.972285497342445,
      "grad_norm": 0.8815106153488159,
      "learning_rate": 1.2104235801181915e-05,
      "loss": 0.5362,
      "step": 7683
    },
    {
      "epoch": 0.9724120475828905,
      "grad_norm": 0.8935019373893738,
      "learning_rate": 1.2102483768902975e-05,
      "loss": 0.5816,
      "step": 7684
    },
    {
      "epoch": 0.9725385978233358,
      "grad_norm": 0.9151484370231628,
      "learning_rate": 1.2100731669098416e-05,
      "loss": 0.6056,
      "step": 7685
    },
    {
      "epoch": 0.9726651480637813,
      "grad_norm": 0.8929476737976074,
      "learning_rate": 1.209897950182451e-05,
      "loss": 0.5943,
      "step": 7686
    },
    {
      "epoch": 0.9727916983042267,
      "grad_norm": 0.9204092025756836,
      "learning_rate": 1.2097227267137534e-05,
      "loss": 0.5963,
      "step": 7687
    },
    {
      "epoch": 0.9729182485446722,
      "grad_norm": 0.8680403828620911,
      "learning_rate": 1.2095474965093765e-05,
      "loss": 0.6076,
      "step": 7688
    },
    {
      "epoch": 0.9730447987851177,
      "grad_norm": 0.8868405222892761,
      "learning_rate": 1.2093722595749483e-05,
      "loss": 0.5352,
      "step": 7689
    },
    {
      "epoch": 0.9731713490255631,
      "grad_norm": 0.8778827786445618,
      "learning_rate": 1.2091970159160964e-05,
      "loss": 0.5753,
      "step": 7690
    },
    {
      "epoch": 0.9732978992660086,
      "grad_norm": 0.901732325553894,
      "learning_rate": 1.2090217655384497e-05,
      "loss": 0.5997,
      "step": 7691
    },
    {
      "epoch": 0.973424449506454,
      "grad_norm": 1.0393649339675903,
      "learning_rate": 1.2088465084476362e-05,
      "loss": 0.6029,
      "step": 7692
    },
    {
      "epoch": 0.9735509997468995,
      "grad_norm": 0.8817983269691467,
      "learning_rate": 1.2086712446492847e-05,
      "loss": 0.5925,
      "step": 7693
    },
    {
      "epoch": 0.973677549987345,
      "grad_norm": 0.9719743132591248,
      "learning_rate": 1.2084959741490245e-05,
      "loss": 0.6261,
      "step": 7694
    },
    {
      "epoch": 0.9738041002277904,
      "grad_norm": 0.8690062165260315,
      "learning_rate": 1.2083206969524848e-05,
      "loss": 0.5516,
      "step": 7695
    },
    {
      "epoch": 0.9739306504682359,
      "grad_norm": 0.8782529234886169,
      "learning_rate": 1.2081454130652945e-05,
      "loss": 0.6013,
      "step": 7696
    },
    {
      "epoch": 0.9740572007086813,
      "grad_norm": 0.9111871719360352,
      "learning_rate": 1.2079701224930836e-05,
      "loss": 0.5654,
      "step": 7697
    },
    {
      "epoch": 0.9741837509491268,
      "grad_norm": 0.8905799388885498,
      "learning_rate": 1.2077948252414817e-05,
      "loss": 0.5561,
      "step": 7698
    },
    {
      "epoch": 0.9743103011895723,
      "grad_norm": 0.930957019329071,
      "learning_rate": 1.207619521316119e-05,
      "loss": 0.6277,
      "step": 7699
    },
    {
      "epoch": 0.9744368514300177,
      "grad_norm": 0.9447407722473145,
      "learning_rate": 1.2074442107226256e-05,
      "loss": 0.5617,
      "step": 7700
    },
    {
      "epoch": 0.9745634016704632,
      "grad_norm": 2.666163444519043,
      "learning_rate": 1.2072688934666326e-05,
      "loss": 0.5857,
      "step": 7701
    },
    {
      "epoch": 0.9746899519109087,
      "grad_norm": 0.8735400438308716,
      "learning_rate": 1.2070935695537695e-05,
      "loss": 0.5708,
      "step": 7702
    },
    {
      "epoch": 0.9748165021513541,
      "grad_norm": 0.8892156481742859,
      "learning_rate": 1.2069182389896683e-05,
      "loss": 0.5528,
      "step": 7703
    },
    {
      "epoch": 0.9749430523917996,
      "grad_norm": 0.893377959728241,
      "learning_rate": 1.2067429017799593e-05,
      "loss": 0.5943,
      "step": 7704
    },
    {
      "epoch": 0.975069602632245,
      "grad_norm": 0.9038506150245667,
      "learning_rate": 1.2065675579302743e-05,
      "loss": 0.5889,
      "step": 7705
    },
    {
      "epoch": 0.9751961528726905,
      "grad_norm": 0.8380857706069946,
      "learning_rate": 1.2063922074462444e-05,
      "loss": 0.5268,
      "step": 7706
    },
    {
      "epoch": 0.975322703113136,
      "grad_norm": 0.8876428008079529,
      "learning_rate": 1.2062168503335021e-05,
      "loss": 0.611,
      "step": 7707
    },
    {
      "epoch": 0.9754492533535813,
      "grad_norm": 0.8472273349761963,
      "learning_rate": 1.2060414865976785e-05,
      "loss": 0.5689,
      "step": 7708
    },
    {
      "epoch": 0.9755758035940268,
      "grad_norm": 0.8777064085006714,
      "learning_rate": 1.2058661162444064e-05,
      "loss": 0.5597,
      "step": 7709
    },
    {
      "epoch": 0.9757023538344722,
      "grad_norm": 0.9504784941673279,
      "learning_rate": 1.2056907392793176e-05,
      "loss": 0.6077,
      "step": 7710
    },
    {
      "epoch": 0.9758289040749177,
      "grad_norm": 0.983430027961731,
      "learning_rate": 1.205515355708045e-05,
      "loss": 0.6202,
      "step": 7711
    },
    {
      "epoch": 0.9759554543153632,
      "grad_norm": 0.9237971305847168,
      "learning_rate": 1.2053399655362216e-05,
      "loss": 0.582,
      "step": 7712
    },
    {
      "epoch": 0.9760820045558086,
      "grad_norm": 0.8885115385055542,
      "learning_rate": 1.2051645687694806e-05,
      "loss": 0.5693,
      "step": 7713
    },
    {
      "epoch": 0.9762085547962541,
      "grad_norm": 0.8646411299705505,
      "learning_rate": 1.2049891654134544e-05,
      "loss": 0.5428,
      "step": 7714
    },
    {
      "epoch": 0.9763351050366996,
      "grad_norm": 0.9262856245040894,
      "learning_rate": 1.2048137554737773e-05,
      "loss": 0.5552,
      "step": 7715
    },
    {
      "epoch": 0.976461655277145,
      "grad_norm": 0.9466032981872559,
      "learning_rate": 1.2046383389560822e-05,
      "loss": 0.6342,
      "step": 7716
    },
    {
      "epoch": 0.9765882055175905,
      "grad_norm": 0.9023475050926208,
      "learning_rate": 1.2044629158660034e-05,
      "loss": 0.6005,
      "step": 7717
    },
    {
      "epoch": 0.9767147557580359,
      "grad_norm": 0.8907579183578491,
      "learning_rate": 1.2042874862091753e-05,
      "loss": 0.6127,
      "step": 7718
    },
    {
      "epoch": 0.9768413059984814,
      "grad_norm": 0.869663417339325,
      "learning_rate": 1.2041120499912315e-05,
      "loss": 0.605,
      "step": 7719
    },
    {
      "epoch": 0.9769678562389269,
      "grad_norm": 0.8857529163360596,
      "learning_rate": 1.2039366072178067e-05,
      "loss": 0.5792,
      "step": 7720
    },
    {
      "epoch": 0.9770944064793723,
      "grad_norm": 0.9135351181030273,
      "learning_rate": 1.2037611578945359e-05,
      "loss": 0.6083,
      "step": 7721
    },
    {
      "epoch": 0.9772209567198178,
      "grad_norm": 0.8699597716331482,
      "learning_rate": 1.2035857020270535e-05,
      "loss": 0.5575,
      "step": 7722
    },
    {
      "epoch": 0.9773475069602632,
      "grad_norm": 0.8541949391365051,
      "learning_rate": 1.2034102396209952e-05,
      "loss": 0.5978,
      "step": 7723
    },
    {
      "epoch": 0.9774740572007087,
      "grad_norm": 0.9245918989181519,
      "learning_rate": 1.2032347706819962e-05,
      "loss": 0.6149,
      "step": 7724
    },
    {
      "epoch": 0.9776006074411542,
      "grad_norm": 0.8520073294639587,
      "learning_rate": 1.2030592952156918e-05,
      "loss": 0.6015,
      "step": 7725
    },
    {
      "epoch": 0.9777271576815996,
      "grad_norm": 0.9236282110214233,
      "learning_rate": 1.2028838132277178e-05,
      "loss": 0.6046,
      "step": 7726
    },
    {
      "epoch": 0.9778537079220451,
      "grad_norm": 0.892780601978302,
      "learning_rate": 1.2027083247237103e-05,
      "loss": 0.6049,
      "step": 7727
    },
    {
      "epoch": 0.9779802581624905,
      "grad_norm": 0.8171454668045044,
      "learning_rate": 1.2025328297093057e-05,
      "loss": 0.5442,
      "step": 7728
    },
    {
      "epoch": 0.978106808402936,
      "grad_norm": 0.8270429372787476,
      "learning_rate": 1.20235732819014e-05,
      "loss": 0.5421,
      "step": 7729
    },
    {
      "epoch": 0.9782333586433815,
      "grad_norm": 0.901929497718811,
      "learning_rate": 1.2021818201718501e-05,
      "loss": 0.5519,
      "step": 7730
    },
    {
      "epoch": 0.9783599088838268,
      "grad_norm": 0.8695142865180969,
      "learning_rate": 1.2020063056600721e-05,
      "loss": 0.5745,
      "step": 7731
    },
    {
      "epoch": 0.9784864591242723,
      "grad_norm": 0.8937426209449768,
      "learning_rate": 1.2018307846604437e-05,
      "loss": 0.5847,
      "step": 7732
    },
    {
      "epoch": 0.9786130093647178,
      "grad_norm": 0.8828654289245605,
      "learning_rate": 1.2016552571786021e-05,
      "loss": 0.599,
      "step": 7733
    },
    {
      "epoch": 0.9787395596051632,
      "grad_norm": 0.8322235941886902,
      "learning_rate": 1.2014797232201845e-05,
      "loss": 0.5642,
      "step": 7734
    },
    {
      "epoch": 0.9788661098456087,
      "grad_norm": 0.9071614146232605,
      "learning_rate": 1.2013041827908287e-05,
      "loss": 0.5951,
      "step": 7735
    },
    {
      "epoch": 0.9789926600860541,
      "grad_norm": 0.8735289573669434,
      "learning_rate": 1.2011286358961724e-05,
      "loss": 0.5654,
      "step": 7736
    },
    {
      "epoch": 0.9791192103264996,
      "grad_norm": 0.9082298278808594,
      "learning_rate": 1.2009530825418538e-05,
      "loss": 0.5834,
      "step": 7737
    },
    {
      "epoch": 0.9792457605669451,
      "grad_norm": 0.9889516830444336,
      "learning_rate": 1.200777522733511e-05,
      "loss": 0.6146,
      "step": 7738
    },
    {
      "epoch": 0.9793723108073905,
      "grad_norm": 0.8731065392494202,
      "learning_rate": 1.2006019564767824e-05,
      "loss": 0.5645,
      "step": 7739
    },
    {
      "epoch": 0.979498861047836,
      "grad_norm": 0.8962516784667969,
      "learning_rate": 1.2004263837773071e-05,
      "loss": 0.5663,
      "step": 7740
    },
    {
      "epoch": 0.9796254112882814,
      "grad_norm": 0.8394458889961243,
      "learning_rate": 1.2002508046407237e-05,
      "loss": 0.5798,
      "step": 7741
    },
    {
      "epoch": 0.9797519615287269,
      "grad_norm": 0.8685465455055237,
      "learning_rate": 1.2000752190726713e-05,
      "loss": 0.5749,
      "step": 7742
    },
    {
      "epoch": 0.9798785117691724,
      "grad_norm": 0.8842285871505737,
      "learning_rate": 1.1998996270787889e-05,
      "loss": 0.5941,
      "step": 7743
    },
    {
      "epoch": 0.9800050620096178,
      "grad_norm": 0.8671888113021851,
      "learning_rate": 1.1997240286647163e-05,
      "loss": 0.6004,
      "step": 7744
    },
    {
      "epoch": 0.9801316122500633,
      "grad_norm": 0.9153568744659424,
      "learning_rate": 1.1995484238360933e-05,
      "loss": 0.5473,
      "step": 7745
    },
    {
      "epoch": 0.9802581624905087,
      "grad_norm": 0.8869165778160095,
      "learning_rate": 1.1993728125985595e-05,
      "loss": 0.5523,
      "step": 7746
    },
    {
      "epoch": 0.9803847127309542,
      "grad_norm": 0.8620353937149048,
      "learning_rate": 1.1991971949577557e-05,
      "loss": 0.5284,
      "step": 7747
    },
    {
      "epoch": 0.9805112629713997,
      "grad_norm": 0.8861702084541321,
      "learning_rate": 1.1990215709193214e-05,
      "loss": 0.5881,
      "step": 7748
    },
    {
      "epoch": 0.9806378132118451,
      "grad_norm": 0.8386902809143066,
      "learning_rate": 1.1988459404888972e-05,
      "loss": 0.5824,
      "step": 7749
    },
    {
      "epoch": 0.9807643634522906,
      "grad_norm": 0.8535817265510559,
      "learning_rate": 1.1986703036721243e-05,
      "loss": 0.5525,
      "step": 7750
    },
    {
      "epoch": 0.9808909136927361,
      "grad_norm": 0.8539744019508362,
      "learning_rate": 1.1984946604746437e-05,
      "loss": 0.547,
      "step": 7751
    },
    {
      "epoch": 0.9810174639331815,
      "grad_norm": 0.8219632506370544,
      "learning_rate": 1.1983190109020962e-05,
      "loss": 0.5619,
      "step": 7752
    },
    {
      "epoch": 0.981144014173627,
      "grad_norm": 0.8554999232292175,
      "learning_rate": 1.1981433549601233e-05,
      "loss": 0.5361,
      "step": 7753
    },
    {
      "epoch": 0.9812705644140723,
      "grad_norm": 0.9082462787628174,
      "learning_rate": 1.1979676926543665e-05,
      "loss": 0.5867,
      "step": 7754
    },
    {
      "epoch": 0.9813971146545178,
      "grad_norm": 0.8701650500297546,
      "learning_rate": 1.1977920239904673e-05,
      "loss": 0.5518,
      "step": 7755
    },
    {
      "epoch": 0.9815236648949633,
      "grad_norm": 1.0912797451019287,
      "learning_rate": 1.1976163489740677e-05,
      "loss": 0.6325,
      "step": 7756
    },
    {
      "epoch": 0.9816502151354087,
      "grad_norm": 0.8394995331764221,
      "learning_rate": 1.1974406676108105e-05,
      "loss": 0.5741,
      "step": 7757
    },
    {
      "epoch": 0.9817767653758542,
      "grad_norm": 0.8934289216995239,
      "learning_rate": 1.1972649799063374e-05,
      "loss": 0.5934,
      "step": 7758
    },
    {
      "epoch": 0.9819033156162996,
      "grad_norm": 0.8761329650878906,
      "learning_rate": 1.1970892858662915e-05,
      "loss": 0.5318,
      "step": 7759
    },
    {
      "epoch": 0.9820298658567451,
      "grad_norm": 0.8718777298927307,
      "learning_rate": 1.1969135854963152e-05,
      "loss": 0.5634,
      "step": 7760
    },
    {
      "epoch": 0.9821564160971906,
      "grad_norm": 0.9330469965934753,
      "learning_rate": 1.1967378788020514e-05,
      "loss": 0.6097,
      "step": 7761
    },
    {
      "epoch": 0.982282966337636,
      "grad_norm": 0.8955352902412415,
      "learning_rate": 1.1965621657891436e-05,
      "loss": 0.5839,
      "step": 7762
    },
    {
      "epoch": 0.9824095165780815,
      "grad_norm": 0.8827700018882751,
      "learning_rate": 1.1963864464632348e-05,
      "loss": 0.5708,
      "step": 7763
    },
    {
      "epoch": 0.982536066818527,
      "grad_norm": 0.9014087319374084,
      "learning_rate": 1.1962107208299694e-05,
      "loss": 0.5963,
      "step": 7764
    },
    {
      "epoch": 0.9826626170589724,
      "grad_norm": 0.8955312967300415,
      "learning_rate": 1.1960349888949903e-05,
      "loss": 0.5677,
      "step": 7765
    },
    {
      "epoch": 0.9827891672994179,
      "grad_norm": 0.8598860502243042,
      "learning_rate": 1.195859250663942e-05,
      "loss": 0.6016,
      "step": 7766
    },
    {
      "epoch": 0.9829157175398633,
      "grad_norm": 0.9024737477302551,
      "learning_rate": 1.1956835061424682e-05,
      "loss": 0.6209,
      "step": 7767
    },
    {
      "epoch": 0.9830422677803088,
      "grad_norm": 0.8673641681671143,
      "learning_rate": 1.1955077553362139e-05,
      "loss": 0.5533,
      "step": 7768
    },
    {
      "epoch": 0.9831688180207543,
      "grad_norm": 0.8713016510009766,
      "learning_rate": 1.1953319982508234e-05,
      "loss": 0.5888,
      "step": 7769
    },
    {
      "epoch": 0.9832953682611997,
      "grad_norm": 0.9806578755378723,
      "learning_rate": 1.1951562348919417e-05,
      "loss": 0.5945,
      "step": 7770
    },
    {
      "epoch": 0.9834219185016452,
      "grad_norm": 0.862619161605835,
      "learning_rate": 1.1949804652652133e-05,
      "loss": 0.5558,
      "step": 7771
    },
    {
      "epoch": 0.9835484687420906,
      "grad_norm": 0.9064843654632568,
      "learning_rate": 1.1948046893762837e-05,
      "loss": 0.5653,
      "step": 7772
    },
    {
      "epoch": 0.9836750189825361,
      "grad_norm": 0.858784019947052,
      "learning_rate": 1.1946289072307983e-05,
      "loss": 0.5778,
      "step": 7773
    },
    {
      "epoch": 0.9838015692229816,
      "grad_norm": 0.8779910206794739,
      "learning_rate": 1.194453118834403e-05,
      "loss": 0.564,
      "step": 7774
    },
    {
      "epoch": 0.983928119463427,
      "grad_norm": 0.8357840180397034,
      "learning_rate": 1.1942773241927433e-05,
      "loss": 0.5121,
      "step": 7775
    },
    {
      "epoch": 0.9840546697038725,
      "grad_norm": 0.8941797018051147,
      "learning_rate": 1.1941015233114654e-05,
      "loss": 0.5593,
      "step": 7776
    },
    {
      "epoch": 0.9841812199443178,
      "grad_norm": 0.8933176398277283,
      "learning_rate": 1.1939257161962152e-05,
      "loss": 0.59,
      "step": 7777
    },
    {
      "epoch": 0.9843077701847633,
      "grad_norm": 0.8774940967559814,
      "learning_rate": 1.1937499028526395e-05,
      "loss": 0.5969,
      "step": 7778
    },
    {
      "epoch": 0.9844343204252088,
      "grad_norm": 0.9014822840690613,
      "learning_rate": 1.1935740832863844e-05,
      "loss": 0.5788,
      "step": 7779
    },
    {
      "epoch": 0.9845608706656542,
      "grad_norm": 0.9460915327072144,
      "learning_rate": 1.1933982575030971e-05,
      "loss": 0.5847,
      "step": 7780
    },
    {
      "epoch": 0.9846874209060997,
      "grad_norm": 0.9063306450843811,
      "learning_rate": 1.1932224255084246e-05,
      "loss": 0.5846,
      "step": 7781
    },
    {
      "epoch": 0.9848139711465452,
      "grad_norm": 0.9036730527877808,
      "learning_rate": 1.1930465873080142e-05,
      "loss": 0.5807,
      "step": 7782
    },
    {
      "epoch": 0.9849405213869906,
      "grad_norm": 0.894768238067627,
      "learning_rate": 1.1928707429075128e-05,
      "loss": 0.5808,
      "step": 7783
    },
    {
      "epoch": 0.9850670716274361,
      "grad_norm": 0.9313892126083374,
      "learning_rate": 1.1926948923125683e-05,
      "loss": 0.5853,
      "step": 7784
    },
    {
      "epoch": 0.9851936218678815,
      "grad_norm": 1.2579535245895386,
      "learning_rate": 1.1925190355288288e-05,
      "loss": 0.6158,
      "step": 7785
    },
    {
      "epoch": 0.985320172108327,
      "grad_norm": 0.858120858669281,
      "learning_rate": 1.1923431725619419e-05,
      "loss": 0.5979,
      "step": 7786
    },
    {
      "epoch": 0.9854467223487725,
      "grad_norm": 0.9135092496871948,
      "learning_rate": 1.1921673034175561e-05,
      "loss": 0.5737,
      "step": 7787
    },
    {
      "epoch": 0.9855732725892179,
      "grad_norm": 0.8873236775398254,
      "learning_rate": 1.1919914281013195e-05,
      "loss": 0.5732,
      "step": 7788
    },
    {
      "epoch": 0.9856998228296634,
      "grad_norm": 0.8671883344650269,
      "learning_rate": 1.1918155466188808e-05,
      "loss": 0.6017,
      "step": 7789
    },
    {
      "epoch": 0.9858263730701088,
      "grad_norm": 0.8487978577613831,
      "learning_rate": 1.1916396589758893e-05,
      "loss": 0.5815,
      "step": 7790
    },
    {
      "epoch": 0.9859529233105543,
      "grad_norm": 0.932255208492279,
      "learning_rate": 1.191463765177993e-05,
      "loss": 0.5774,
      "step": 7791
    },
    {
      "epoch": 0.9860794735509998,
      "grad_norm": 0.9038334488868713,
      "learning_rate": 1.1912878652308417e-05,
      "loss": 0.6011,
      "step": 7792
    },
    {
      "epoch": 0.9862060237914452,
      "grad_norm": 0.8738823533058167,
      "learning_rate": 1.191111959140085e-05,
      "loss": 0.5631,
      "step": 7793
    },
    {
      "epoch": 0.9863325740318907,
      "grad_norm": 0.8529459238052368,
      "learning_rate": 1.190936046911372e-05,
      "loss": 0.5844,
      "step": 7794
    },
    {
      "epoch": 0.9864591242723361,
      "grad_norm": 1.0206470489501953,
      "learning_rate": 1.1907601285503528e-05,
      "loss": 0.5404,
      "step": 7795
    },
    {
      "epoch": 0.9865856745127816,
      "grad_norm": 0.9095285534858704,
      "learning_rate": 1.1905842040626772e-05,
      "loss": 0.568,
      "step": 7796
    },
    {
      "epoch": 0.9867122247532271,
      "grad_norm": 0.8839933276176453,
      "learning_rate": 1.1904082734539954e-05,
      "loss": 0.5816,
      "step": 7797
    },
    {
      "epoch": 0.9868387749936725,
      "grad_norm": 0.8498429656028748,
      "learning_rate": 1.190232336729958e-05,
      "loss": 0.5599,
      "step": 7798
    },
    {
      "epoch": 0.986965325234118,
      "grad_norm": 0.8206489682197571,
      "learning_rate": 1.1900563938962156e-05,
      "loss": 0.554,
      "step": 7799
    },
    {
      "epoch": 0.9870918754745635,
      "grad_norm": 0.9225293397903442,
      "learning_rate": 1.1898804449584185e-05,
      "loss": 0.6083,
      "step": 7800
    },
    {
      "epoch": 0.9872184257150088,
      "grad_norm": 0.8857840299606323,
      "learning_rate": 1.189704489922218e-05,
      "loss": 0.539,
      "step": 7801
    },
    {
      "epoch": 0.9873449759554543,
      "grad_norm": 0.9003739953041077,
      "learning_rate": 1.1895285287932654e-05,
      "loss": 0.585,
      "step": 7802
    },
    {
      "epoch": 0.9874715261958997,
      "grad_norm": 0.8708470463752747,
      "learning_rate": 1.1893525615772115e-05,
      "loss": 0.5926,
      "step": 7803
    },
    {
      "epoch": 0.9875980764363452,
      "grad_norm": 0.9587939977645874,
      "learning_rate": 1.1891765882797083e-05,
      "loss": 0.6031,
      "step": 7804
    },
    {
      "epoch": 0.9877246266767907,
      "grad_norm": 0.8813052177429199,
      "learning_rate": 1.1890006089064079e-05,
      "loss": 0.5702,
      "step": 7805
    },
    {
      "epoch": 0.9878511769172361,
      "grad_norm": 0.9034405946731567,
      "learning_rate": 1.1888246234629612e-05,
      "loss": 0.5754,
      "step": 7806
    },
    {
      "epoch": 0.9879777271576816,
      "grad_norm": 0.8647516369819641,
      "learning_rate": 1.1886486319550212e-05,
      "loss": 0.5307,
      "step": 7807
    },
    {
      "epoch": 0.988104277398127,
      "grad_norm": 0.9303377270698547,
      "learning_rate": 1.18847263438824e-05,
      "loss": 0.5889,
      "step": 7808
    },
    {
      "epoch": 0.9882308276385725,
      "grad_norm": 0.8905583620071411,
      "learning_rate": 1.18829663076827e-05,
      "loss": 0.5635,
      "step": 7809
    },
    {
      "epoch": 0.988357377879018,
      "grad_norm": 0.968963086605072,
      "learning_rate": 1.1881206211007642e-05,
      "loss": 0.5568,
      "step": 7810
    },
    {
      "epoch": 0.9884839281194634,
      "grad_norm": 0.9137853384017944,
      "learning_rate": 1.1879446053913753e-05,
      "loss": 0.6164,
      "step": 7811
    },
    {
      "epoch": 0.9886104783599089,
      "grad_norm": 0.8613213896751404,
      "learning_rate": 1.1877685836457562e-05,
      "loss": 0.5595,
      "step": 7812
    },
    {
      "epoch": 0.9887370286003544,
      "grad_norm": 0.8865113258361816,
      "learning_rate": 1.1875925558695604e-05,
      "loss": 0.6019,
      "step": 7813
    },
    {
      "epoch": 0.9888635788407998,
      "grad_norm": 0.9234564304351807,
      "learning_rate": 1.1874165220684419e-05,
      "loss": 0.6008,
      "step": 7814
    },
    {
      "epoch": 0.9889901290812453,
      "grad_norm": 0.9121437072753906,
      "learning_rate": 1.1872404822480536e-05,
      "loss": 0.5815,
      "step": 7815
    },
    {
      "epoch": 0.9891166793216907,
      "grad_norm": 0.8762125968933105,
      "learning_rate": 1.1870644364140497e-05,
      "loss": 0.575,
      "step": 7816
    },
    {
      "epoch": 0.9892432295621362,
      "grad_norm": 0.8587542772293091,
      "learning_rate": 1.1868883845720845e-05,
      "loss": 0.5392,
      "step": 7817
    },
    {
      "epoch": 0.9893697798025817,
      "grad_norm": 0.8864484429359436,
      "learning_rate": 1.1867123267278117e-05,
      "loss": 0.5211,
      "step": 7818
    },
    {
      "epoch": 0.9894963300430271,
      "grad_norm": 0.8683136701583862,
      "learning_rate": 1.1865362628868864e-05,
      "loss": 0.5875,
      "step": 7819
    },
    {
      "epoch": 0.9896228802834726,
      "grad_norm": 0.8717551231384277,
      "learning_rate": 1.1863601930549629e-05,
      "loss": 0.5775,
      "step": 7820
    },
    {
      "epoch": 0.989749430523918,
      "grad_norm": 0.8659651875495911,
      "learning_rate": 1.186184117237696e-05,
      "loss": 0.5576,
      "step": 7821
    },
    {
      "epoch": 0.9898759807643635,
      "grad_norm": 0.8705272674560547,
      "learning_rate": 1.186008035440741e-05,
      "loss": 0.584,
      "step": 7822
    },
    {
      "epoch": 0.990002531004809,
      "grad_norm": 0.8767131567001343,
      "learning_rate": 1.185831947669753e-05,
      "loss": 0.5664,
      "step": 7823
    },
    {
      "epoch": 0.9901290812452543,
      "grad_norm": 0.8562760949134827,
      "learning_rate": 1.1856558539303873e-05,
      "loss": 0.5935,
      "step": 7824
    },
    {
      "epoch": 0.9902556314856998,
      "grad_norm": 0.9079773426055908,
      "learning_rate": 1.1854797542282996e-05,
      "loss": 0.5644,
      "step": 7825
    },
    {
      "epoch": 0.9903821817261452,
      "grad_norm": 0.9269507527351379,
      "learning_rate": 1.1853036485691462e-05,
      "loss": 0.5684,
      "step": 7826
    },
    {
      "epoch": 0.9905087319665907,
      "grad_norm": 0.8631138801574707,
      "learning_rate": 1.1851275369585823e-05,
      "loss": 0.5565,
      "step": 7827
    },
    {
      "epoch": 0.9906352822070362,
      "grad_norm": 0.84872967004776,
      "learning_rate": 1.1849514194022645e-05,
      "loss": 0.5529,
      "step": 7828
    },
    {
      "epoch": 0.9907618324474816,
      "grad_norm": 0.9420230984687805,
      "learning_rate": 1.184775295905849e-05,
      "loss": 0.5476,
      "step": 7829
    },
    {
      "epoch": 0.9908883826879271,
      "grad_norm": 0.8650606274604797,
      "learning_rate": 1.1845991664749922e-05,
      "loss": 0.5906,
      "step": 7830
    },
    {
      "epoch": 0.9910149329283726,
      "grad_norm": 1.0181430578231812,
      "learning_rate": 1.1844230311153516e-05,
      "loss": 0.6482,
      "step": 7831
    },
    {
      "epoch": 0.991141483168818,
      "grad_norm": 1.0008189678192139,
      "learning_rate": 1.1842468898325832e-05,
      "loss": 0.5562,
      "step": 7832
    },
    {
      "epoch": 0.9912680334092635,
      "grad_norm": 0.8728497624397278,
      "learning_rate": 1.1840707426323452e-05,
      "loss": 0.5943,
      "step": 7833
    },
    {
      "epoch": 0.9913945836497089,
      "grad_norm": 0.9698101282119751,
      "learning_rate": 1.183894589520294e-05,
      "loss": 0.5676,
      "step": 7834
    },
    {
      "epoch": 0.9915211338901544,
      "grad_norm": 0.897786021232605,
      "learning_rate": 1.1837184305020874e-05,
      "loss": 0.5714,
      "step": 7835
    },
    {
      "epoch": 0.9916476841305999,
      "grad_norm": 0.8817154765129089,
      "learning_rate": 1.1835422655833835e-05,
      "loss": 0.6014,
      "step": 7836
    },
    {
      "epoch": 0.9917742343710453,
      "grad_norm": 1.0272743701934814,
      "learning_rate": 1.1833660947698395e-05,
      "loss": 0.5866,
      "step": 7837
    },
    {
      "epoch": 0.9919007846114908,
      "grad_norm": 0.8758074045181274,
      "learning_rate": 1.1831899180671145e-05,
      "loss": 0.5833,
      "step": 7838
    },
    {
      "epoch": 0.9920273348519362,
      "grad_norm": 0.9564797878265381,
      "learning_rate": 1.1830137354808656e-05,
      "loss": 0.6197,
      "step": 7839
    },
    {
      "epoch": 0.9921538850923817,
      "grad_norm": 0.8928477764129639,
      "learning_rate": 1.1828375470167521e-05,
      "loss": 0.5919,
      "step": 7840
    },
    {
      "epoch": 0.9922804353328272,
      "grad_norm": 0.8170707821846008,
      "learning_rate": 1.1826613526804325e-05,
      "loss": 0.5611,
      "step": 7841
    },
    {
      "epoch": 0.9924069855732726,
      "grad_norm": 0.8988536596298218,
      "learning_rate": 1.1824851524775651e-05,
      "loss": 0.5642,
      "step": 7842
    },
    {
      "epoch": 0.9925335358137181,
      "grad_norm": 0.9240908026695251,
      "learning_rate": 1.1823089464138095e-05,
      "loss": 0.5775,
      "step": 7843
    },
    {
      "epoch": 0.9926600860541634,
      "grad_norm": 0.8627564311027527,
      "learning_rate": 1.182132734494825e-05,
      "loss": 0.5689,
      "step": 7844
    },
    {
      "epoch": 0.992786636294609,
      "grad_norm": 0.8826587796211243,
      "learning_rate": 1.1819565167262707e-05,
      "loss": 0.5577,
      "step": 7845
    },
    {
      "epoch": 0.9929131865350544,
      "grad_norm": 0.8691292405128479,
      "learning_rate": 1.1817802931138065e-05,
      "loss": 0.5586,
      "step": 7846
    },
    {
      "epoch": 0.9930397367754998,
      "grad_norm": 0.943533718585968,
      "learning_rate": 1.1816040636630916e-05,
      "loss": 0.5772,
      "step": 7847
    },
    {
      "epoch": 0.9931662870159453,
      "grad_norm": 0.854979932308197,
      "learning_rate": 1.1814278283797865e-05,
      "loss": 0.5778,
      "step": 7848
    },
    {
      "epoch": 0.9932928372563908,
      "grad_norm": 0.8603362441062927,
      "learning_rate": 1.1812515872695511e-05,
      "loss": 0.6179,
      "step": 7849
    },
    {
      "epoch": 0.9934193874968362,
      "grad_norm": 0.9245336055755615,
      "learning_rate": 1.1810753403380464e-05,
      "loss": 0.5709,
      "step": 7850
    },
    {
      "epoch": 0.9935459377372817,
      "grad_norm": 0.8523623943328857,
      "learning_rate": 1.1808990875909318e-05,
      "loss": 0.5748,
      "step": 7851
    },
    {
      "epoch": 0.9936724879777271,
      "grad_norm": 0.9364871978759766,
      "learning_rate": 1.180722829033869e-05,
      "loss": 0.5947,
      "step": 7852
    },
    {
      "epoch": 0.9937990382181726,
      "grad_norm": 0.9354009628295898,
      "learning_rate": 1.1805465646725185e-05,
      "loss": 0.5947,
      "step": 7853
    },
    {
      "epoch": 0.9939255884586181,
      "grad_norm": 0.8380885124206543,
      "learning_rate": 1.1803702945125413e-05,
      "loss": 0.5397,
      "step": 7854
    },
    {
      "epoch": 0.9940521386990635,
      "grad_norm": 0.9262405037879944,
      "learning_rate": 1.1801940185595985e-05,
      "loss": 0.5621,
      "step": 7855
    },
    {
      "epoch": 0.994178688939509,
      "grad_norm": 0.8516844511032104,
      "learning_rate": 1.1800177368193526e-05,
      "loss": 0.5794,
      "step": 7856
    },
    {
      "epoch": 0.9943052391799544,
      "grad_norm": 0.8300899267196655,
      "learning_rate": 1.1798414492974641e-05,
      "loss": 0.5823,
      "step": 7857
    },
    {
      "epoch": 0.9944317894203999,
      "grad_norm": 0.8852511048316956,
      "learning_rate": 1.1796651559995951e-05,
      "loss": 0.5864,
      "step": 7858
    },
    {
      "epoch": 0.9945583396608454,
      "grad_norm": 0.9219977259635925,
      "learning_rate": 1.179488856931408e-05,
      "loss": 0.5676,
      "step": 7859
    },
    {
      "epoch": 0.9946848899012908,
      "grad_norm": 0.8785005211830139,
      "learning_rate": 1.1793125520985647e-05,
      "loss": 0.5494,
      "step": 7860
    },
    {
      "epoch": 0.9948114401417363,
      "grad_norm": 0.8741841316223145,
      "learning_rate": 1.1791362415067277e-05,
      "loss": 0.594,
      "step": 7861
    },
    {
      "epoch": 0.9949379903821818,
      "grad_norm": 0.8946130275726318,
      "learning_rate": 1.17895992516156e-05,
      "loss": 0.5826,
      "step": 7862
    },
    {
      "epoch": 0.9950645406226272,
      "grad_norm": 0.9019899368286133,
      "learning_rate": 1.1787836030687234e-05,
      "loss": 0.5922,
      "step": 7863
    },
    {
      "epoch": 0.9951910908630727,
      "grad_norm": 0.9249155521392822,
      "learning_rate": 1.1786072752338818e-05,
      "loss": 0.6015,
      "step": 7864
    },
    {
      "epoch": 0.9953176411035181,
      "grad_norm": 0.870521068572998,
      "learning_rate": 1.1784309416626978e-05,
      "loss": 0.6104,
      "step": 7865
    },
    {
      "epoch": 0.9954441913439636,
      "grad_norm": 0.8442793488502502,
      "learning_rate": 1.1782546023608345e-05,
      "loss": 0.5389,
      "step": 7866
    },
    {
      "epoch": 0.9955707415844091,
      "grad_norm": 0.8566998839378357,
      "learning_rate": 1.178078257333956e-05,
      "loss": 0.5336,
      "step": 7867
    },
    {
      "epoch": 0.9956972918248544,
      "grad_norm": 0.8733071684837341,
      "learning_rate": 1.177901906587726e-05,
      "loss": 0.5837,
      "step": 7868
    },
    {
      "epoch": 0.9958238420653,
      "grad_norm": 0.8973745703697205,
      "learning_rate": 1.1777255501278076e-05,
      "loss": 0.5835,
      "step": 7869
    },
    {
      "epoch": 0.9959503923057453,
      "grad_norm": 0.8898856043815613,
      "learning_rate": 1.1775491879598655e-05,
      "loss": 0.5726,
      "step": 7870
    },
    {
      "epoch": 0.9960769425461908,
      "grad_norm": 0.875329315662384,
      "learning_rate": 1.1773728200895638e-05,
      "loss": 0.5812,
      "step": 7871
    },
    {
      "epoch": 0.9962034927866363,
      "grad_norm": 0.9060443639755249,
      "learning_rate": 1.1771964465225668e-05,
      "loss": 0.5799,
      "step": 7872
    },
    {
      "epoch": 0.9963300430270817,
      "grad_norm": 0.9631041288375854,
      "learning_rate": 1.1770200672645394e-05,
      "loss": 0.578,
      "step": 7873
    },
    {
      "epoch": 0.9964565932675272,
      "grad_norm": 0.8919326663017273,
      "learning_rate": 1.1768436823211463e-05,
      "loss": 0.5459,
      "step": 7874
    },
    {
      "epoch": 0.9965831435079726,
      "grad_norm": 0.9805166125297546,
      "learning_rate": 1.1766672916980522e-05,
      "loss": 0.6,
      "step": 7875
    },
    {
      "epoch": 0.9967096937484181,
      "grad_norm": 0.860995352268219,
      "learning_rate": 1.1764908954009226e-05,
      "loss": 0.5499,
      "step": 7876
    },
    {
      "epoch": 0.9968362439888636,
      "grad_norm": 0.8994447588920593,
      "learning_rate": 1.1763144934354222e-05,
      "loss": 0.597,
      "step": 7877
    },
    {
      "epoch": 0.996962794229309,
      "grad_norm": 0.8069507479667664,
      "learning_rate": 1.1761380858072173e-05,
      "loss": 0.5409,
      "step": 7878
    },
    {
      "epoch": 0.9970893444697545,
      "grad_norm": 0.8956016898155212,
      "learning_rate": 1.1759616725219732e-05,
      "loss": 0.6056,
      "step": 7879
    },
    {
      "epoch": 0.9972158947102,
      "grad_norm": 0.8541343808174133,
      "learning_rate": 1.1757852535853558e-05,
      "loss": 0.5774,
      "step": 7880
    },
    {
      "epoch": 0.9973424449506454,
      "grad_norm": 0.8626531362533569,
      "learning_rate": 1.1756088290030312e-05,
      "loss": 0.5266,
      "step": 7881
    },
    {
      "epoch": 0.9974689951910909,
      "grad_norm": 0.9067010283470154,
      "learning_rate": 1.1754323987806655e-05,
      "loss": 0.5858,
      "step": 7882
    },
    {
      "epoch": 0.9975955454315363,
      "grad_norm": 0.8713178634643555,
      "learning_rate": 1.1752559629239253e-05,
      "loss": 0.5252,
      "step": 7883
    },
    {
      "epoch": 0.9977220956719818,
      "grad_norm": 0.9568624496459961,
      "learning_rate": 1.1750795214384772e-05,
      "loss": 0.6032,
      "step": 7884
    },
    {
      "epoch": 0.9978486459124273,
      "grad_norm": 0.8612202405929565,
      "learning_rate": 1.174903074329988e-05,
      "loss": 0.5588,
      "step": 7885
    },
    {
      "epoch": 0.9979751961528727,
      "grad_norm": 0.8668959140777588,
      "learning_rate": 1.1747266216041244e-05,
      "loss": 0.5592,
      "step": 7886
    },
    {
      "epoch": 0.9981017463933182,
      "grad_norm": 0.9580898284912109,
      "learning_rate": 1.1745501632665536e-05,
      "loss": 0.5806,
      "step": 7887
    },
    {
      "epoch": 0.9982282966337636,
      "grad_norm": 0.874991238117218,
      "learning_rate": 1.1743736993229437e-05,
      "loss": 0.5511,
      "step": 7888
    },
    {
      "epoch": 0.9983548468742091,
      "grad_norm": 0.9254584908485413,
      "learning_rate": 1.1741972297789612e-05,
      "loss": 0.5644,
      "step": 7889
    },
    {
      "epoch": 0.9984813971146546,
      "grad_norm": 1.7413911819458008,
      "learning_rate": 1.174020754640274e-05,
      "loss": 0.614,
      "step": 7890
    },
    {
      "epoch": 0.9986079473551,
      "grad_norm": 1.0011004209518433,
      "learning_rate": 1.1738442739125504e-05,
      "loss": 0.5438,
      "step": 7891
    },
    {
      "epoch": 0.9987344975955454,
      "grad_norm": 0.9333179593086243,
      "learning_rate": 1.1736677876014582e-05,
      "loss": 0.6025,
      "step": 7892
    },
    {
      "epoch": 0.9988610478359908,
      "grad_norm": 0.8875114917755127,
      "learning_rate": 1.1734912957126654e-05,
      "loss": 0.5769,
      "step": 7893
    },
    {
      "epoch": 0.9989875980764363,
      "grad_norm": 0.8530983328819275,
      "learning_rate": 1.1733147982518405e-05,
      "loss": 0.525,
      "step": 7894
    },
    {
      "epoch": 0.9991141483168818,
      "grad_norm": 0.9123764634132385,
      "learning_rate": 1.1731382952246521e-05,
      "loss": 0.5841,
      "step": 7895
    },
    {
      "epoch": 0.9992406985573272,
      "grad_norm": 0.927833616733551,
      "learning_rate": 1.1729617866367692e-05,
      "loss": 0.5846,
      "step": 7896
    },
    {
      "epoch": 0.9993672487977727,
      "grad_norm": 0.826422929763794,
      "learning_rate": 1.1727852724938608e-05,
      "loss": 0.5461,
      "step": 7897
    },
    {
      "epoch": 0.9994937990382182,
      "grad_norm": 0.9142796397209167,
      "learning_rate": 1.1726087528015957e-05,
      "loss": 0.6083,
      "step": 7898
    },
    {
      "epoch": 0.9996203492786636,
      "grad_norm": 0.8775556087493896,
      "learning_rate": 1.172432227565643e-05,
      "loss": 0.5564,
      "step": 7899
    },
    {
      "epoch": 0.9997468995191091,
      "grad_norm": 0.8468303680419922,
      "learning_rate": 1.1722556967916729e-05,
      "loss": 0.5606,
      "step": 7900
    },
    {
      "epoch": 0.9998734497595545,
      "grad_norm": 0.8511351346969604,
      "learning_rate": 1.1720791604853543e-05,
      "loss": 0.5326,
      "step": 7901
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9015161395072937,
      "learning_rate": 1.1719026186523572e-05,
      "loss": 0.6021,
      "step": 7902
    }
  ],
  "logging_steps": 1,
  "max_steps": 17630,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 2.8463520601919088e+20,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}