{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2181,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0004585052728106373,
      "grad_norm": 87.08877230333971,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 8.9274,
      "step": 1
    },
    {
      "epoch": 0.0009170105456212746,
      "grad_norm": 86.74486108504378,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 8.9055,
      "step": 2
    },
    {
      "epoch": 0.001375515818431912,
      "grad_norm": 59.165041824873434,
      "learning_rate": 4.545454545454546e-05,
      "loss": 8.3015,
      "step": 3
    },
    {
      "epoch": 0.0018340210912425492,
      "grad_norm": 38.304845971050625,
      "learning_rate": 6.060606060606061e-05,
      "loss": 7.339,
      "step": 4
    },
    {
      "epoch": 0.0022925263640531865,
      "grad_norm": 20.53634597585696,
      "learning_rate": 7.575757575757576e-05,
      "loss": 6.2547,
      "step": 5
    },
    {
      "epoch": 0.002751031636863824,
      "grad_norm": 12.92872990003818,
      "learning_rate": 9.090909090909092e-05,
      "loss": 5.5891,
      "step": 6
    },
    {
      "epoch": 0.0032095369096744614,
      "grad_norm": 10.077091946994813,
      "learning_rate": 0.00010606060606060606,
      "loss": 5.3297,
      "step": 7
    },
    {
      "epoch": 0.0036680421824850985,
      "grad_norm": 6.88021231566845,
      "learning_rate": 0.00012121212121212122,
      "loss": 5.033,
      "step": 8
    },
    {
      "epoch": 0.0041265474552957355,
      "grad_norm": 4.7103686240296465,
      "learning_rate": 0.00013636363636363637,
      "loss": 4.8718,
      "step": 9
    },
    {
      "epoch": 0.004585052728106373,
      "grad_norm": 3.0384267344873077,
      "learning_rate": 0.00015151515151515152,
      "loss": 4.7147,
      "step": 10
    },
    {
      "epoch": 0.0050435580009170105,
      "grad_norm": 3.0320321553638565,
      "learning_rate": 0.00016666666666666666,
      "loss": 4.6534,
      "step": 11
    },
    {
      "epoch": 0.005502063273727648,
      "grad_norm": 2.6013861492960104,
      "learning_rate": 0.00018181818181818183,
      "loss": 4.7432,
      "step": 12
    },
    {
      "epoch": 0.005960568546538285,
      "grad_norm": 2.0126693183445634,
      "learning_rate": 0.00019696969696969695,
      "loss": 4.6737,
      "step": 13
    },
    {
      "epoch": 0.006419073819348923,
      "grad_norm": 1.8443458478795265,
      "learning_rate": 0.00021212121212121213,
      "loss": 4.5847,
      "step": 14
    },
    {
      "epoch": 0.0068775790921595595,
      "grad_norm": 1.8944214406216793,
      "learning_rate": 0.00022727272727272727,
      "loss": 4.4907,
      "step": 15
    },
    {
      "epoch": 0.007336084364970197,
      "grad_norm": 1.5952256220286176,
      "learning_rate": 0.00024242424242424245,
      "loss": 4.4736,
      "step": 16
    },
    {
      "epoch": 0.007794589637780834,
      "grad_norm": 1.6894173958065297,
      "learning_rate": 0.00025757575757575756,
      "loss": 4.3523,
      "step": 17
    },
    {
      "epoch": 0.008253094910591471,
      "grad_norm": 1.4961214211226275,
      "learning_rate": 0.00027272727272727274,
      "loss": 4.3233,
      "step": 18
    },
    {
      "epoch": 0.00871160018340211,
      "grad_norm": 1.1267538116005982,
      "learning_rate": 0.0002878787878787879,
      "loss": 4.2767,
      "step": 19
    },
    {
      "epoch": 0.009170105456212746,
      "grad_norm": 1.198028678533528,
      "learning_rate": 0.00030303030303030303,
      "loss": 4.4438,
      "step": 20
    },
    {
      "epoch": 0.009628610729023384,
      "grad_norm": 1.0807052777107995,
      "learning_rate": 0.0003181818181818182,
      "loss": 4.3786,
      "step": 21
    },
    {
      "epoch": 0.010087116001834021,
      "grad_norm": 1.128516651156729,
      "learning_rate": 0.0003333333333333333,
      "loss": 4.2505,
      "step": 22
    },
    {
      "epoch": 0.01054562127464466,
      "grad_norm": 1.0429318688344307,
      "learning_rate": 0.0003484848484848485,
      "loss": 4.1966,
      "step": 23
    },
    {
      "epoch": 0.011004126547455296,
      "grad_norm": 1.0065809636616745,
      "learning_rate": 0.00036363636363636367,
      "loss": 4.3049,
      "step": 24
    },
    {
      "epoch": 0.011462631820265932,
      "grad_norm": 0.7693108774901681,
      "learning_rate": 0.0003787878787878788,
      "loss": 4.1811,
      "step": 25
    },
    {
      "epoch": 0.01192113709307657,
      "grad_norm": 0.6672323196406228,
      "learning_rate": 0.0003939393939393939,
      "loss": 4.1196,
      "step": 26
    },
    {
      "epoch": 0.012379642365887207,
      "grad_norm": 0.7029329355291158,
      "learning_rate": 0.00040909090909090913,
      "loss": 4.1835,
      "step": 27
    },
    {
      "epoch": 0.012838147638697846,
      "grad_norm": 0.7608684916334876,
      "learning_rate": 0.00042424242424242425,
      "loss": 4.1258,
      "step": 28
    },
    {
      "epoch": 0.013296652911508482,
      "grad_norm": 0.7391605836669172,
      "learning_rate": 0.0004393939393939394,
      "loss": 4.2266,
      "step": 29
    },
    {
      "epoch": 0.013755158184319119,
      "grad_norm": 0.5608907056915889,
      "learning_rate": 0.00045454545454545455,
      "loss": 4.1604,
      "step": 30
    },
    {
      "epoch": 0.014213663457129757,
      "grad_norm": 0.542599115773914,
      "learning_rate": 0.0004696969696969697,
      "loss": 4.1239,
      "step": 31
    },
    {
      "epoch": 0.014672168729940394,
      "grad_norm": 0.48097801765563425,
      "learning_rate": 0.0004848484848484849,
      "loss": 4.1458,
      "step": 32
    },
    {
      "epoch": 0.015130674002751032,
      "grad_norm": 0.5620939195837599,
      "learning_rate": 0.0005,
      "loss": 4.0925,
      "step": 33
    },
    {
      "epoch": 0.015589179275561669,
      "grad_norm": 0.4130774154098247,
      "learning_rate": 0.0005151515151515151,
      "loss": 4.1321,
      "step": 34
    },
    {
      "epoch": 0.016047684548372305,
      "grad_norm": 0.4506889478038943,
      "learning_rate": 0.0005303030303030302,
      "loss": 4.0939,
      "step": 35
    },
    {
      "epoch": 0.016506189821182942,
      "grad_norm": 0.3764017733124088,
      "learning_rate": 0.0005454545454545455,
      "loss": 3.9578,
      "step": 36
    },
    {
      "epoch": 0.016964695093993582,
      "grad_norm": 0.4250952485445014,
      "learning_rate": 0.0005606060606060606,
      "loss": 4.0497,
      "step": 37
    },
    {
      "epoch": 0.01742320036680422,
      "grad_norm": 0.40320682032887356,
      "learning_rate": 0.0005757575757575758,
      "loss": 4.0379,
      "step": 38
    },
    {
      "epoch": 0.017881705639614855,
      "grad_norm": 0.40094696025675963,
      "learning_rate": 0.0005909090909090909,
      "loss": 4.0667,
      "step": 39
    },
    {
      "epoch": 0.018340210912425492,
      "grad_norm": 0.41503311481228733,
      "learning_rate": 0.0006060606060606061,
      "loss": 4.0286,
      "step": 40
    },
    {
      "epoch": 0.01879871618523613,
      "grad_norm": 0.3250066768764527,
      "learning_rate": 0.0006212121212121212,
      "loss": 3.9436,
      "step": 41
    },
    {
      "epoch": 0.01925722145804677,
      "grad_norm": 0.3659954276205639,
      "learning_rate": 0.0006363636363636364,
      "loss": 3.9715,
      "step": 42
    },
    {
      "epoch": 0.019715726730857405,
      "grad_norm": 0.3575484433235927,
      "learning_rate": 0.0006515151515151515,
      "loss": 4.0161,
      "step": 43
    },
    {
      "epoch": 0.020174232003668042,
      "grad_norm": 0.36383222131379955,
      "learning_rate": 0.0006666666666666666,
      "loss": 3.9189,
      "step": 44
    },
    {
      "epoch": 0.02063273727647868,
      "grad_norm": 0.32822220228055654,
      "learning_rate": 0.0006818181818181818,
      "loss": 3.8938,
      "step": 45
    },
    {
      "epoch": 0.02109124254928932,
      "grad_norm": 0.2812227979406115,
      "learning_rate": 0.000696969696969697,
      "loss": 3.9709,
      "step": 46
    },
    {
      "epoch": 0.021549747822099955,
      "grad_norm": 0.3802989971175073,
      "learning_rate": 0.0007121212121212122,
      "loss": 3.9115,
      "step": 47
    },
    {
      "epoch": 0.02200825309491059,
      "grad_norm": 0.30225718209886104,
      "learning_rate": 0.0007272727272727273,
      "loss": 3.9185,
      "step": 48
    },
    {
      "epoch": 0.02246675836772123,
      "grad_norm": 0.30595555891161347,
      "learning_rate": 0.0007424242424242425,
      "loss": 4.0179,
      "step": 49
    },
    {
      "epoch": 0.022925263640531865,
      "grad_norm": 0.31127819072769947,
      "learning_rate": 0.0007575757575757576,
      "loss": 3.9604,
      "step": 50
    },
    {
      "epoch": 0.023383768913342505,
      "grad_norm": 0.33153102893570846,
      "learning_rate": 0.0007727272727272727,
      "loss": 3.9281,
      "step": 51
    },
    {
      "epoch": 0.02384227418615314,
      "grad_norm": 0.2969275024112839,
      "learning_rate": 0.0007878787878787878,
      "loss": 4.0265,
      "step": 52
    },
    {
      "epoch": 0.024300779458963778,
      "grad_norm": 0.3168749977684905,
      "learning_rate": 0.000803030303030303,
      "loss": 4.0416,
      "step": 53
    },
    {
      "epoch": 0.024759284731774415,
      "grad_norm": 0.28446243359504614,
      "learning_rate": 0.0008181818181818183,
      "loss": 3.9362,
      "step": 54
    },
    {
      "epoch": 0.02521779000458505,
      "grad_norm": 0.28186871317764356,
      "learning_rate": 0.0008333333333333334,
      "loss": 3.9391,
      "step": 55
    },
    {
      "epoch": 0.02567629527739569,
      "grad_norm": 0.29095199367290386,
      "learning_rate": 0.0008484848484848485,
      "loss": 4.0161,
      "step": 56
    },
    {
      "epoch": 0.026134800550206328,
      "grad_norm": 0.2991161831777244,
      "learning_rate": 0.0008636363636363636,
      "loss": 3.9581,
      "step": 57
    },
    {
      "epoch": 0.026593305823016965,
      "grad_norm": 0.30323830339128277,
      "learning_rate": 0.0008787878787878789,
      "loss": 4.0743,
      "step": 58
    },
    {
      "epoch": 0.0270518110958276,
      "grad_norm": 0.3603871067048101,
      "learning_rate": 0.000893939393939394,
      "loss": 3.8969,
      "step": 59
    },
    {
      "epoch": 0.027510316368638238,
      "grad_norm": 0.27772346797782793,
      "learning_rate": 0.0009090909090909091,
      "loss": 3.8931,
      "step": 60
    },
    {
      "epoch": 0.027968821641448878,
      "grad_norm": 0.27950483853539,
      "learning_rate": 0.0009242424242424242,
      "loss": 3.8017,
      "step": 61
    },
    {
      "epoch": 0.028427326914259515,
      "grad_norm": 0.371512221803555,
      "learning_rate": 0.0009393939393939394,
      "loss": 3.9196,
      "step": 62
    },
    {
      "epoch": 0.02888583218707015,
      "grad_norm": 0.31170234605922825,
      "learning_rate": 0.0009545454545454546,
      "loss": 3.8946,
      "step": 63
    },
    {
      "epoch": 0.029344337459880788,
      "grad_norm": 0.344881963944121,
      "learning_rate": 0.0009696969696969698,
      "loss": 4.0305,
      "step": 64
    },
    {
      "epoch": 0.029802842732691424,
      "grad_norm": 0.284808074057287,
      "learning_rate": 0.000984848484848485,
      "loss": 3.8221,
      "step": 65
    },
    {
      "epoch": 0.030261348005502064,
      "grad_norm": 0.3023529610067081,
      "learning_rate": 0.001,
      "loss": 3.9471,
      "step": 66
    },
    {
      "epoch": 0.0307198532783127,
      "grad_norm": 0.33305248590498987,
      "learning_rate": 0.0009999994484067654,
      "loss": 3.961,
      "step": 67
    },
    {
      "epoch": 0.031178358551123338,
      "grad_norm": 0.3225602211272884,
      "learning_rate": 0.0009999977936282788,
      "loss": 3.811,
      "step": 68
    },
    {
      "epoch": 0.03163686382393398,
      "grad_norm": 0.29214174452883823,
      "learning_rate": 0.0009999950356681913,
      "loss": 3.8573,
      "step": 69
    },
    {
      "epoch": 0.03209536909674461,
      "grad_norm": 0.3292349368868398,
      "learning_rate": 0.0009999911745325876,
      "loss": 3.8334,
      "step": 70
    },
    {
      "epoch": 0.03255387436955525,
      "grad_norm": 0.3226528129614394,
      "learning_rate": 0.0009999862102299873,
      "loss": 3.873,
      "step": 71
    },
    {
      "epoch": 0.033012379642365884,
      "grad_norm": 0.32353347687195616,
      "learning_rate": 0.0009999801427713433,
      "loss": 3.8825,
      "step": 72
    },
    {
      "epoch": 0.033470884915176524,
      "grad_norm": 0.3214164654484599,
      "learning_rate": 0.0009999729721700424,
      "loss": 3.7724,
      "step": 73
    },
    {
      "epoch": 0.033929390187987164,
      "grad_norm": 0.3215230349073946,
      "learning_rate": 0.000999964698441906,
      "loss": 3.8723,
      "step": 74
    },
    {
      "epoch": 0.0343878954607978,
      "grad_norm": 0.3668304209754661,
      "learning_rate": 0.0009999553216051892,
      "loss": 3.8944,
      "step": 75
    },
    {
      "epoch": 0.03484640073360844,
      "grad_norm": 0.34121582396134725,
      "learning_rate": 0.00099994484168058,
      "loss": 3.8291,
      "step": 76
    },
    {
      "epoch": 0.03530490600641907,
      "grad_norm": 0.34362207849426524,
      "learning_rate": 0.0009999332586912019,
      "loss": 3.8452,
      "step": 77
    },
    {
      "epoch": 0.03576341127922971,
      "grad_norm": 0.3572838676139915,
      "learning_rate": 0.0009999205726626108,
      "loss": 3.82,
      "step": 78
    },
    {
      "epoch": 0.03622191655204035,
      "grad_norm": 0.3548303184231157,
      "learning_rate": 0.000999906783622797,
      "loss": 3.7816,
      "step": 79
    },
    {
      "epoch": 0.036680421824850984,
      "grad_norm": 0.3895858629664817,
      "learning_rate": 0.0009998918916021842,
      "loss": 3.8703,
      "step": 80
    },
    {
      "epoch": 0.037138927097661624,
      "grad_norm": 0.41866014228471443,
      "learning_rate": 0.0009998758966336297,
      "loss": 3.7416,
      "step": 81
    },
    {
      "epoch": 0.03759743237047226,
      "grad_norm": 0.4128828886196255,
      "learning_rate": 0.0009998587987524242,
      "loss": 3.7861,
      "step": 82
    },
    {
      "epoch": 0.0380559376432829,
      "grad_norm": 0.3395402027191704,
      "learning_rate": 0.0009998405979962926,
      "loss": 3.8186,
      "step": 83
    },
    {
      "epoch": 0.03851444291609354,
      "grad_norm": 0.5413633730066792,
      "learning_rate": 0.000999821294405392,
      "loss": 3.7819,
      "step": 84
    },
    {
      "epoch": 0.03897294818890417,
      "grad_norm": 0.46323641003558813,
      "learning_rate": 0.0009998008880223134,
      "loss": 3.8454,
      "step": 85
    },
    {
      "epoch": 0.03943145346171481,
      "grad_norm": 0.5190903936291698,
      "learning_rate": 0.000999779378892081,
      "loss": 3.8391,
      "step": 86
    },
    {
      "epoch": 0.039889958734525444,
      "grad_norm": 0.5009166186297459,
      "learning_rate": 0.0009997567670621522,
      "loss": 3.6549,
      "step": 87
    },
    {
      "epoch": 0.040348464007336084,
      "grad_norm": 0.5158990099665065,
      "learning_rate": 0.0009997330525824165,
      "loss": 3.812,
      "step": 88
    },
    {
      "epoch": 0.040806969280146724,
      "grad_norm": 0.4983142124622741,
      "learning_rate": 0.0009997082355051976,
      "loss": 3.789,
      "step": 89
    },
    {
      "epoch": 0.04126547455295736,
      "grad_norm": 0.5147248336458968,
      "learning_rate": 0.000999682315885251,
      "loss": 3.8131,
      "step": 90
    },
    {
      "epoch": 0.041723979825768,
      "grad_norm": 0.5314783656397152,
      "learning_rate": 0.0009996552937797645,
      "loss": 3.6664,
      "step": 91
    },
    {
      "epoch": 0.04218248509857864,
      "grad_norm": 0.4644434625654812,
      "learning_rate": 0.0009996271692483596,
      "loss": 3.7076,
      "step": 92
    },
    {
      "epoch": 0.04264099037138927,
      "grad_norm": 0.6936687451733824,
      "learning_rate": 0.0009995979423530893,
      "loss": 3.6996,
      "step": 93
    },
    {
      "epoch": 0.04309949564419991,
      "grad_norm": 0.6473113123055435,
      "learning_rate": 0.000999567613158439,
      "loss": 3.6666,
      "step": 94
    },
    {
      "epoch": 0.04355800091701054,
      "grad_norm": 0.5880840099932867,
      "learning_rate": 0.0009995361817313263,
      "loss": 3.6633,
      "step": 95
    },
    {
      "epoch": 0.04401650618982118,
      "grad_norm": 0.6563861780714988,
      "learning_rate": 0.0009995036481411004,
      "loss": 3.7462,
      "step": 96
    },
    {
      "epoch": 0.044475011462631824,
      "grad_norm": 0.8841896102799615,
      "learning_rate": 0.0009994700124595429,
      "loss": 3.7864,
      "step": 97
    },
    {
      "epoch": 0.04493351673544246,
      "grad_norm": 0.6697703282113565,
      "learning_rate": 0.0009994352747608663,
      "loss": 3.6974,
      "step": 98
    },
    {
      "epoch": 0.0453920220082531,
      "grad_norm": 0.8201481741131788,
      "learning_rate": 0.0009993994351217151,
      "loss": 3.6893,
      "step": 99
    },
    {
      "epoch": 0.04585052728106373,
      "grad_norm": 0.6691876032691336,
      "learning_rate": 0.000999362493621165,
      "loss": 3.6743,
      "step": 100
    },
    {
      "epoch": 0.04630903255387437,
      "grad_norm": 0.7572530669456428,
      "learning_rate": 0.0009993244503407226,
      "loss": 3.69,
      "step": 101
    },
    {
      "epoch": 0.04676753782668501,
      "grad_norm": 0.6993425947995145,
      "learning_rate": 0.0009992853053643258,
      "loss": 3.5353,
      "step": 102
    },
    {
      "epoch": 0.04722604309949564,
      "grad_norm": 0.6270926150124078,
      "learning_rate": 0.0009992450587783426,
      "loss": 3.5452,
      "step": 103
    },
    {
      "epoch": 0.04768454837230628,
      "grad_norm": 0.8962510769966768,
      "learning_rate": 0.000999203710671572,
      "loss": 3.6187,
      "step": 104
    },
    {
      "epoch": 0.048143053645116916,
      "grad_norm": 0.7268635330239095,
      "learning_rate": 0.0009991612611352438,
      "loss": 3.582,
      "step": 105
    },
    {
      "epoch": 0.048601558917927556,
      "grad_norm": 0.8279963516917189,
      "learning_rate": 0.0009991177102630173,
      "loss": 3.65,
      "step": 106
    },
    {
      "epoch": 0.049060064190738197,
      "grad_norm": 0.6911227346799309,
      "learning_rate": 0.0009990730581509817,
      "loss": 3.5251,
      "step": 107
    },
    {
      "epoch": 0.04951856946354883,
      "grad_norm": 0.8263658590042374,
      "learning_rate": 0.0009990273048976566,
      "loss": 3.5625,
      "step": 108
    },
    {
      "epoch": 0.04997707473635947,
      "grad_norm": 0.6712532531553561,
      "learning_rate": 0.0009989804506039905,
      "loss": 3.5724,
      "step": 109
    },
    {
      "epoch": 0.0504355800091701,
      "grad_norm": 0.7831949535098771,
      "learning_rate": 0.0009989324953733614,
      "loss": 3.6032,
      "step": 110
    },
    {
      "epoch": 0.05089408528198074,
      "grad_norm": 0.7374426501872315,
      "learning_rate": 0.0009988834393115766,
      "loss": 3.5482,
      "step": 111
    },
    {
      "epoch": 0.05135259055479138,
      "grad_norm": 0.7604241664002718,
      "learning_rate": 0.000998833282526872,
      "loss": 3.5543,
      "step": 112
    },
    {
      "epoch": 0.051811095827602016,
      "grad_norm": 0.7144518743262781,
      "learning_rate": 0.0009987820251299122,
      "loss": 3.3951,
      "step": 113
    },
    {
      "epoch": 0.052269601100412656,
      "grad_norm": 0.6843486765219107,
      "learning_rate": 0.00099872966723379,
      "loss": 3.4749,
      "step": 114
    },
    {
      "epoch": 0.05272810637322329,
      "grad_norm": 0.8484108965864872,
      "learning_rate": 0.0009986762089540266,
      "loss": 3.5038,
      "step": 115
    },
    {
      "epoch": 0.05318661164603393,
      "grad_norm": 0.6502108423879875,
      "learning_rate": 0.0009986216504085709,
      "loss": 3.5346,
      "step": 116
    },
    {
      "epoch": 0.05364511691884457,
      "grad_norm": 0.6751161744357562,
      "learning_rate": 0.0009985659917177991,
      "loss": 3.5529,
      "step": 117
    },
    {
      "epoch": 0.0541036221916552,
      "grad_norm": 0.6720927887498935,
      "learning_rate": 0.0009985092330045155,
      "loss": 3.4656,
      "step": 118
    },
    {
      "epoch": 0.05456212746446584,
      "grad_norm": 0.6824888204303899,
      "learning_rate": 0.0009984513743939508,
      "loss": 3.4419,
      "step": 119
    },
    {
      "epoch": 0.055020632737276476,
      "grad_norm": 0.7376569162764148,
      "learning_rate": 0.0009983924160137626,
      "loss": 3.4546,
      "step": 120
    },
    {
      "epoch": 0.055479138010087116,
      "grad_norm": 0.7255995746418183,
      "learning_rate": 0.000998332357994035,
      "loss": 3.3955,
      "step": 121
    },
    {
      "epoch": 0.055937643282897756,
      "grad_norm": 0.7182223862252344,
      "learning_rate": 0.0009982712004672786,
      "loss": 3.4231,
      "step": 122
    },
    {
      "epoch": 0.05639614855570839,
      "grad_norm": 0.7090692068331048,
      "learning_rate": 0.0009982089435684295,
      "loss": 3.3949,
      "step": 123
    },
    {
      "epoch": 0.05685465382851903,
      "grad_norm": 0.7051401630283205,
      "learning_rate": 0.0009981455874348499,
      "loss": 3.3229,
      "step": 124
    },
    {
      "epoch": 0.05731315910132966,
      "grad_norm": 0.655086614619009,
      "learning_rate": 0.0009980811322063269,
      "loss": 3.3357,
      "step": 125
    },
    {
      "epoch": 0.0577716643741403,
      "grad_norm": 0.7289849518862422,
      "learning_rate": 0.0009980155780250728,
      "loss": 3.3516,
      "step": 126
    },
    {
      "epoch": 0.05823016964695094,
      "grad_norm": 0.6703489816417686,
      "learning_rate": 0.0009979489250357243,
      "loss": 3.2881,
      "step": 127
    },
    {
      "epoch": 0.058688674919761576,
      "grad_norm": 0.6647241552027786,
      "learning_rate": 0.0009978811733853431,
      "loss": 3.2691,
      "step": 128
    },
    {
      "epoch": 0.059147180192572216,
      "grad_norm": 0.6214068716191447,
      "learning_rate": 0.0009978123232234147,
      "loss": 3.3131,
      "step": 129
    },
    {
      "epoch": 0.05960568546538285,
      "grad_norm": 0.675840498058823,
      "learning_rate": 0.000997742374701848,
      "loss": 3.2456,
      "step": 130
    },
    {
      "epoch": 0.06006419073819349,
      "grad_norm": 0.6721896382456175,
      "learning_rate": 0.0009976713279749754,
      "loss": 3.2027,
      "step": 131
    },
    {
      "epoch": 0.06052269601100413,
      "grad_norm": 0.7009832827893479,
      "learning_rate": 0.0009975991831995528,
      "loss": 3.2775,
      "step": 132
    },
    {
      "epoch": 0.06098120128381476,
      "grad_norm": 0.714778650824546,
      "learning_rate": 0.0009975259405347581,
      "loss": 3.293,
      "step": 133
    },
    {
      "epoch": 0.0614397065566254,
      "grad_norm": 0.8125601364611585,
      "learning_rate": 0.0009974516001421926,
      "loss": 3.3028,
      "step": 134
    },
    {
      "epoch": 0.061898211829436035,
      "grad_norm": 0.8833622755838736,
      "learning_rate": 0.000997376162185878,
      "loss": 3.2577,
      "step": 135
    },
    {
      "epoch": 0.062356717102246675,
      "grad_norm": 0.71213484011175,
      "learning_rate": 0.0009972996268322594,
      "loss": 3.2005,
      "step": 136
    },
    {
      "epoch": 0.06281522237505731,
      "grad_norm": 0.7623680674014299,
      "learning_rate": 0.0009972219942502017,
      "loss": 3.2387,
      "step": 137
    },
    {
      "epoch": 0.06327372764786796,
      "grad_norm": 0.7545189989363856,
      "learning_rate": 0.0009971432646109918,
      "loss": 3.2384,
      "step": 138
    },
    {
      "epoch": 0.06373223292067859,
      "grad_norm": 0.7129665790718294,
      "learning_rate": 0.0009970634380883365,
      "loss": 3.1861,
      "step": 139
    },
    {
      "epoch": 0.06419073819348922,
      "grad_norm": 0.6712466160288836,
      "learning_rate": 0.0009969825148583627,
      "loss": 3.2555,
      "step": 140
    },
    {
      "epoch": 0.06464924346629987,
      "grad_norm": 0.742612122406992,
      "learning_rate": 0.0009969004950996173,
      "loss": 3.1753,
      "step": 141
    },
    {
      "epoch": 0.0651077487391105,
      "grad_norm": 0.6560609891144648,
      "learning_rate": 0.0009968173789930668,
      "loss": 3.2482,
      "step": 142
    },
    {
      "epoch": 0.06556625401192114,
      "grad_norm": 0.7374596454638317,
      "learning_rate": 0.0009967331667220958,
      "loss": 3.1604,
      "step": 143
    },
    {
      "epoch": 0.06602475928473177,
      "grad_norm": 0.669587965045298,
      "learning_rate": 0.0009966478584725086,
      "loss": 3.1173,
      "step": 144
    },
    {
      "epoch": 0.06648326455754242,
      "grad_norm": 0.6914345538606524,
      "learning_rate": 0.0009965614544325263,
      "loss": 3.2105,
      "step": 145
    },
    {
      "epoch": 0.06694176983035305,
      "grad_norm": 0.771797326751308,
      "learning_rate": 0.000996473954792789,
      "loss": 3.2182,
      "step": 146
    },
    {
      "epoch": 0.06740027510316368,
      "grad_norm": 0.6818983594459271,
      "learning_rate": 0.0009963853597463532,
      "loss": 3.164,
      "step": 147
    },
    {
      "epoch": 0.06785878037597433,
      "grad_norm": 0.7701879609981028,
      "learning_rate": 0.000996295669488693,
      "loss": 3.2179,
      "step": 148
    },
    {
      "epoch": 0.06831728564878496,
      "grad_norm": 0.7729375928199047,
      "learning_rate": 0.0009962048842176979,
      "loss": 3.2412,
      "step": 149
    },
    {
      "epoch": 0.0687757909215956,
      "grad_norm": 0.7222876705471246,
      "learning_rate": 0.0009961130041336748,
      "loss": 3.1502,
      "step": 150
    },
    {
      "epoch": 0.06923429619440624,
      "grad_norm": 0.7877832149491948,
      "learning_rate": 0.0009960200294393449,
      "loss": 3.1781,
      "step": 151
    },
    {
      "epoch": 0.06969280146721687,
      "grad_norm": 0.7520623214117006,
      "learning_rate": 0.0009959259603398453,
      "loss": 3.1627,
      "step": 152
    },
    {
      "epoch": 0.07015130674002751,
      "grad_norm": 0.7729330388377037,
      "learning_rate": 0.0009958307970427275,
      "loss": 3.0056,
      "step": 153
    },
    {
      "epoch": 0.07060981201283814,
      "grad_norm": 0.7306601842380531,
      "learning_rate": 0.0009957345397579572,
      "loss": 3.113,
      "step": 154
    },
    {
      "epoch": 0.07106831728564879,
      "grad_norm": 0.7252532871839816,
      "learning_rate": 0.0009956371886979138,
      "loss": 3.1247,
      "step": 155
    },
    {
      "epoch": 0.07152682255845942,
      "grad_norm": 0.7215338583103166,
      "learning_rate": 0.00099553874407739,
      "loss": 3.1658,
      "step": 156
    },
    {
      "epoch": 0.07198532783127005,
      "grad_norm": 0.8442910402265041,
      "learning_rate": 0.0009954392061135916,
      "loss": 3.0971,
      "step": 157
    },
    {
      "epoch": 0.0724438331040807,
      "grad_norm": 0.7439880826435179,
      "learning_rate": 0.0009953385750261364,
      "loss": 3.0981,
      "step": 158
    },
    {
      "epoch": 0.07290233837689133,
      "grad_norm": 0.7656973410737521,
      "learning_rate": 0.0009952368510370538,
      "loss": 3.0794,
      "step": 159
    },
    {
      "epoch": 0.07336084364970197,
      "grad_norm": 0.7641846350094631,
      "learning_rate": 0.0009951340343707852,
      "loss": 3.0527,
      "step": 160
    },
    {
      "epoch": 0.07381934892251261,
      "grad_norm": 0.7096038343972799,
      "learning_rate": 0.0009950301252541823,
      "loss": 3.1138,
      "step": 161
    },
    {
      "epoch": 0.07427785419532325,
      "grad_norm": 0.7040048522409736,
      "learning_rate": 0.0009949251239165075,
      "loss": 3.0835,
      "step": 162
    },
    {
      "epoch": 0.07473635946813388,
      "grad_norm": 0.6697833641407691,
      "learning_rate": 0.000994819030589433,
      "loss": 3.1248,
      "step": 163
    },
    {
      "epoch": 0.07519486474094451,
      "grad_norm": 0.6976930734502121,
      "learning_rate": 0.00099471184550704,
      "loss": 3.0154,
      "step": 164
    },
    {
      "epoch": 0.07565337001375516,
      "grad_norm": 0.8178385623824674,
      "learning_rate": 0.0009946035689058189,
      "loss": 3.0336,
      "step": 165
    },
    {
      "epoch": 0.0761118752865658,
      "grad_norm": 0.7695720307880182,
      "learning_rate": 0.0009944942010246681,
      "loss": 3.1134,
      "step": 166
    },
    {
      "epoch": 0.07657038055937643,
      "grad_norm": 0.8162000842991968,
      "learning_rate": 0.0009943837421048942,
      "loss": 3.1515,
      "step": 167
    },
    {
      "epoch": 0.07702888583218707,
      "grad_norm": 0.8169737289085561,
      "learning_rate": 0.0009942721923902106,
      "loss": 2.9624,
      "step": 168
    },
    {
      "epoch": 0.07748739110499771,
      "grad_norm": 0.6729562054558437,
      "learning_rate": 0.0009941595521267377,
      "loss": 2.9267,
      "step": 169
    },
    {
      "epoch": 0.07794589637780834,
      "grad_norm": 0.7625571097005902,
      "learning_rate": 0.0009940458215630017,
      "loss": 3.0617,
      "step": 170
    },
    {
      "epoch": 0.07840440165061899,
      "grad_norm": 0.7867257096682732,
      "learning_rate": 0.0009939310009499348,
      "loss": 3.0901,
      "step": 171
    },
    {
      "epoch": 0.07886290692342962,
      "grad_norm": 0.7366492539306999,
      "learning_rate": 0.000993815090540874,
      "loss": 3.0926,
      "step": 172
    },
    {
      "epoch": 0.07932141219624025,
      "grad_norm": 0.7037295373095799,
      "learning_rate": 0.000993698090591561,
      "loss": 3.1433,
      "step": 173
    },
    {
      "epoch": 0.07977991746905089,
      "grad_norm": 0.7852778349509771,
      "learning_rate": 0.0009935800013601416,
      "loss": 2.9805,
      "step": 174
    },
    {
      "epoch": 0.08023842274186153,
      "grad_norm": 0.6865973829570937,
      "learning_rate": 0.000993460823107164,
      "loss": 2.9824,
      "step": 175
    },
    {
      "epoch": 0.08069692801467217,
      "grad_norm": 0.7057453990223607,
      "learning_rate": 0.0009933405560955803,
      "loss": 2.9867,
      "step": 176
    },
    {
      "epoch": 0.0811554332874828,
      "grad_norm": 0.8292171604230166,
      "learning_rate": 0.0009932192005907446,
      "loss": 3.0199,
      "step": 177
    },
    {
      "epoch": 0.08161393856029345,
      "grad_norm": 0.7411295486218207,
      "learning_rate": 0.0009930967568604118,
      "loss": 3.0804,
      "step": 178
    },
    {
      "epoch": 0.08207244383310408,
      "grad_norm": 0.7791752347223339,
      "learning_rate": 0.000992973225174739,
      "loss": 2.941,
      "step": 179
    },
    {
      "epoch": 0.08253094910591471,
      "grad_norm": 0.8007665164685339,
      "learning_rate": 0.0009928486058062827,
      "loss": 2.9252,
      "step": 180
    },
    {
      "epoch": 0.08298945437872536,
      "grad_norm": 0.7663421264755373,
      "learning_rate": 0.0009927228990299999,
      "loss": 3.0739,
      "step": 181
    },
    {
      "epoch": 0.083447959651536,
      "grad_norm": 0.8240342317132103,
      "learning_rate": 0.0009925961051232468,
      "loss": 3.0219,
      "step": 182
    },
    {
      "epoch": 0.08390646492434663,
      "grad_norm": 0.7974077256218088,
      "learning_rate": 0.000992468224365778,
      "loss": 2.9478,
      "step": 183
    },
    {
      "epoch": 0.08436497019715727,
      "grad_norm": 0.7275469162696288,
      "learning_rate": 0.000992339257039746,
      "loss": 3.0061,
      "step": 184
    },
    {
      "epoch": 0.08482347546996791,
      "grad_norm": 0.7283103765761046,
      "learning_rate": 0.0009922092034297006,
      "loss": 2.9321,
      "step": 185
    },
    {
      "epoch": 0.08528198074277854,
      "grad_norm": 0.7805132332195546,
      "learning_rate": 0.0009920780638225891,
      "loss": 2.9032,
      "step": 186
    },
    {
      "epoch": 0.08574048601558917,
      "grad_norm": 0.729020727899942,
      "learning_rate": 0.0009919458385077538,
      "loss": 2.9023,
      "step": 187
    },
    {
      "epoch": 0.08619899128839982,
      "grad_norm": 0.7655028789698481,
      "learning_rate": 0.0009918125277769336,
      "loss": 2.8947,
      "step": 188
    },
    {
      "epoch": 0.08665749656121045,
      "grad_norm": 0.7347501451431111,
      "learning_rate": 0.0009916781319242614,
      "loss": 2.9794,
      "step": 189
    },
    {
      "epoch": 0.08711600183402109,
      "grad_norm": 0.7416427068420914,
      "learning_rate": 0.0009915426512462646,
      "loss": 2.95,
      "step": 190
    },
    {
      "epoch": 0.08757450710683173,
      "grad_norm": 0.756770244446837,
      "learning_rate": 0.0009914060860418644,
      "loss": 2.9937,
      "step": 191
    },
    {
      "epoch": 0.08803301237964237,
      "grad_norm": 0.7629813960943969,
      "learning_rate": 0.000991268436612374,
      "loss": 3.0554,
      "step": 192
    },
    {
      "epoch": 0.088491517652453,
      "grad_norm": 0.8732618716834192,
      "learning_rate": 0.0009911297032614997,
      "loss": 3.0286,
      "step": 193
    },
    {
      "epoch": 0.08895002292526365,
      "grad_norm": 0.8111465282968946,
      "learning_rate": 0.000990989886295339,
      "loss": 2.9466,
      "step": 194
    },
    {
      "epoch": 0.08940852819807428,
      "grad_norm": 0.7486305330410602,
      "learning_rate": 0.0009908489860223804,
      "loss": 2.9939,
      "step": 195
    },
    {
      "epoch": 0.08986703347088491,
      "grad_norm": 0.7464182948900351,
      "learning_rate": 0.000990707002753502,
      "loss": 2.9113,
      "step": 196
    },
    {
      "epoch": 0.09032553874369555,
      "grad_norm": 0.7411860527320553,
      "learning_rate": 0.0009905639368019724,
      "loss": 2.917,
      "step": 197
    },
    {
      "epoch": 0.0907840440165062,
      "grad_norm": 0.863076330734343,
      "learning_rate": 0.0009904197884834482,
      "loss": 2.9035,
      "step": 198
    },
    {
      "epoch": 0.09124254928931683,
      "grad_norm": 0.8084397820866335,
      "learning_rate": 0.0009902745581159742,
      "loss": 2.9539,
      "step": 199
    },
    {
      "epoch": 0.09170105456212746,
      "grad_norm": 0.7525505955839785,
      "learning_rate": 0.0009901282460199829,
      "loss": 3.0008,
      "step": 200
    },
    {
      "epoch": 0.0921595598349381,
      "grad_norm": 0.8299922959749281,
      "learning_rate": 0.0009899808525182935,
      "loss": 2.9834,
      "step": 201
    },
    {
      "epoch": 0.09261806510774874,
      "grad_norm": 0.8322203293233782,
      "learning_rate": 0.0009898323779361107,
      "loss": 2.8836,
      "step": 202
    },
    {
      "epoch": 0.09307657038055937,
      "grad_norm": 0.7945513745677251,
      "learning_rate": 0.000989682822601025,
      "loss": 2.8579,
      "step": 203
    },
    {
      "epoch": 0.09353507565337002,
      "grad_norm": 0.7725806687151319,
      "learning_rate": 0.0009895321868430113,
      "loss": 2.8837,
      "step": 204
    },
    {
      "epoch": 0.09399358092618065,
      "grad_norm": 0.7398905915510126,
      "learning_rate": 0.0009893804709944281,
      "loss": 2.8768,
      "step": 205
    },
    {
      "epoch": 0.09445208619899129,
      "grad_norm": 0.804482912623748,
      "learning_rate": 0.0009892276753900174,
      "loss": 2.9477,
      "step": 206
    },
    {
      "epoch": 0.09491059147180192,
      "grad_norm": 0.881280698834944,
      "learning_rate": 0.0009890738003669028,
      "loss": 2.8901,
      "step": 207
    },
    {
      "epoch": 0.09536909674461257,
      "grad_norm": 0.7629342577948108,
      "learning_rate": 0.0009889188462645904,
      "loss": 2.9711,
      "step": 208
    },
    {
      "epoch": 0.0958276020174232,
      "grad_norm": 0.8416734710292749,
      "learning_rate": 0.0009887628134249667,
      "loss": 2.9692,
      "step": 209
    },
    {
      "epoch": 0.09628610729023383,
      "grad_norm": 0.8302659542033772,
      "learning_rate": 0.0009886057021922983,
      "loss": 3.0161,
      "step": 210
    },
    {
      "epoch": 0.09674461256304448,
      "grad_norm": 0.8770912918864877,
      "learning_rate": 0.0009884475129132311,
      "loss": 2.8909,
      "step": 211
    },
    {
      "epoch": 0.09720311783585511,
      "grad_norm": 0.7949231001893708,
      "learning_rate": 0.0009882882459367897,
      "loss": 2.7386,
      "step": 212
    },
    {
      "epoch": 0.09766162310866575,
      "grad_norm": 0.8768515715229399,
      "learning_rate": 0.0009881279016143766,
      "loss": 2.7466,
      "step": 213
    },
    {
      "epoch": 0.09812012838147639,
      "grad_norm": 0.8040069126569066,
      "learning_rate": 0.0009879664802997707,
      "loss": 2.8883,
      "step": 214
    },
    {
      "epoch": 0.09857863365428703,
      "grad_norm": 0.7911447517867276,
      "learning_rate": 0.000987803982349128,
      "loss": 2.9602,
      "step": 215
    },
    {
      "epoch": 0.09903713892709766,
      "grad_norm": 0.8451763295947028,
      "learning_rate": 0.0009876404081209796,
      "loss": 2.8694,
      "step": 216
    },
    {
      "epoch": 0.09949564419990829,
      "grad_norm": 0.7386120138124803,
      "learning_rate": 0.000987475757976231,
      "loss": 2.8282,
      "step": 217
    },
    {
      "epoch": 0.09995414947271894,
      "grad_norm": 0.8103269289021655,
      "learning_rate": 0.000987310032278162,
      "loss": 2.8148,
      "step": 218
    },
    {
      "epoch": 0.10041265474552957,
      "grad_norm": 0.893796828102827,
      "learning_rate": 0.0009871432313924254,
      "loss": 2.837,
      "step": 219
    },
    {
      "epoch": 0.1008711600183402,
      "grad_norm": 0.887408421967769,
      "learning_rate": 0.000986975355687046,
      "loss": 2.8659,
      "step": 220
    },
    {
      "epoch": 0.10132966529115085,
      "grad_norm": 0.8275714302755509,
      "learning_rate": 0.0009868064055324204,
      "loss": 2.8079,
      "step": 221
    },
    {
      "epoch": 0.10178817056396149,
      "grad_norm": 0.9040797424248289,
      "learning_rate": 0.0009866363813013153,
      "loss": 2.8652,
      "step": 222
    },
    {
      "epoch": 0.10224667583677212,
      "grad_norm": 0.7504898198393671,
      "learning_rate": 0.0009864652833688676,
      "loss": 2.7899,
      "step": 223
    },
    {
      "epoch": 0.10270518110958277,
      "grad_norm": 0.8278147916606398,
      "learning_rate": 0.0009862931121125836,
      "loss": 2.8024,
      "step": 224
    },
    {
      "epoch": 0.1031636863823934,
      "grad_norm": 0.8973895327701829,
      "learning_rate": 0.000986119867912337,
      "loss": 2.8825,
      "step": 225
    },
    {
      "epoch": 0.10362219165520403,
      "grad_norm": 0.8846228085734674,
      "learning_rate": 0.000985945551150369,
      "loss": 2.7841,
      "step": 226
    },
    {
      "epoch": 0.10408069692801467,
      "grad_norm": 0.8015938743269606,
      "learning_rate": 0.0009857701622112876,
      "loss": 2.8099,
      "step": 227
    },
    {
      "epoch": 0.10453920220082531,
      "grad_norm": 0.7958510731429039,
      "learning_rate": 0.000985593701482066,
      "loss": 2.892,
      "step": 228
    },
    {
      "epoch": 0.10499770747363595,
      "grad_norm": 0.831208256445253,
      "learning_rate": 0.0009854161693520424,
      "loss": 2.831,
      "step": 229
    },
    {
      "epoch": 0.10545621274644658,
      "grad_norm": 0.8398552428057999,
      "learning_rate": 0.0009852375662129194,
      "loss": 2.85,
      "step": 230
    },
    {
      "epoch": 0.10591471801925723,
      "grad_norm": 0.7838008358201495,
      "learning_rate": 0.0009850578924587613,
      "loss": 2.755,
      "step": 231
    },
    {
      "epoch": 0.10637322329206786,
      "grad_norm": 0.7854723460877244,
      "learning_rate": 0.000984877148485996,
      "loss": 2.8465,
      "step": 232
    },
    {
      "epoch": 0.10683172856487849,
      "grad_norm": 0.8558687796268061,
      "learning_rate": 0.000984695334693412,
      "loss": 2.9196,
      "step": 233
    },
    {
      "epoch": 0.10729023383768914,
      "grad_norm": 0.7546588900935908,
      "learning_rate": 0.000984512451482158,
      "loss": 2.7997,
      "step": 234
    },
    {
      "epoch": 0.10774873911049977,
      "grad_norm": 0.7866130278053632,
      "learning_rate": 0.0009843284992557431,
      "loss": 2.7916,
      "step": 235
    },
    {
      "epoch": 0.1082072443833104,
      "grad_norm": 0.7488728129967709,
      "learning_rate": 0.000984143478420034,
      "loss": 2.8134,
      "step": 236
    },
    {
      "epoch": 0.10866574965612105,
      "grad_norm": 0.7872594588677201,
      "learning_rate": 0.0009839573893832563,
      "loss": 2.8224,
      "step": 237
    },
    {
      "epoch": 0.10912425492893169,
      "grad_norm": 0.8118121231376928,
      "learning_rate": 0.000983770232555991,
      "loss": 2.8771,
      "step": 238
    },
    {
      "epoch": 0.10958276020174232,
      "grad_norm": 0.8276971095903819,
      "learning_rate": 0.0009835820083511765,
      "loss": 2.7253,
      "step": 239
    },
    {
      "epoch": 0.11004126547455295,
      "grad_norm": 0.7869531694479154,
      "learning_rate": 0.0009833927171841055,
      "loss": 2.8045,
      "step": 240
    },
    {
      "epoch": 0.1104997707473636,
      "grad_norm": 0.8146051527903216,
      "learning_rate": 0.0009832023594724246,
      "loss": 2.7859,
      "step": 241
    },
    {
      "epoch": 0.11095827602017423,
      "grad_norm": 0.7883588197397442,
      "learning_rate": 0.0009830109356361344,
      "loss": 2.8018,
      "step": 242
    },
    {
      "epoch": 0.11141678129298486,
      "grad_norm": 0.7694969979732077,
      "learning_rate": 0.0009828184460975867,
      "loss": 2.7941,
      "step": 243
    },
    {
      "epoch": 0.11187528656579551,
      "grad_norm": 0.7444744777793517,
      "learning_rate": 0.0009826248912814855,
      "loss": 2.7385,
      "step": 244
    },
    {
      "epoch": 0.11233379183860615,
      "grad_norm": 0.7806090252725324,
      "learning_rate": 0.0009824302716148847,
      "loss": 2.7683,
      "step": 245
    },
    {
      "epoch": 0.11279229711141678,
      "grad_norm": 0.7719753974178828,
      "learning_rate": 0.0009822345875271884,
      "loss": 2.7219,
      "step": 246
    },
    {
      "epoch": 0.11325080238422743,
      "grad_norm": 0.8098563466852718,
      "learning_rate": 0.0009820378394501481,
      "loss": 2.7596,
      "step": 247
    },
    {
      "epoch": 0.11370930765703806,
      "grad_norm": 0.8207120359509854,
      "learning_rate": 0.0009818400278178636,
      "loss": 2.6587,
      "step": 248
    },
    {
      "epoch": 0.11416781292984869,
      "grad_norm": 0.8037714071791052,
      "learning_rate": 0.0009816411530667814,
      "loss": 2.6961,
      "step": 249
    },
    {
      "epoch": 0.11462631820265932,
      "grad_norm": 0.7676201749250869,
      "learning_rate": 0.000981441215635693,
      "loss": 2.8298,
      "step": 250
    },
    {
      "epoch": 0.11508482347546997,
      "grad_norm": 0.8479836973844477,
      "learning_rate": 0.0009812402159657353,
      "loss": 2.8496,
      "step": 251
    },
    {
      "epoch": 0.1155433287482806,
      "grad_norm": 0.7986124846537074,
      "learning_rate": 0.000981038154500388,
      "loss": 2.8155,
      "step": 252
    },
    {
      "epoch": 0.11600183402109124,
      "grad_norm": 0.8270226565415023,
      "learning_rate": 0.0009808350316854746,
      "loss": 2.7552,
      "step": 253
    },
    {
      "epoch": 0.11646033929390188,
      "grad_norm": 0.8418036887098906,
      "learning_rate": 0.0009806308479691594,
      "loss": 2.8152,
      "step": 254
    },
    {
      "epoch": 0.11691884456671252,
      "grad_norm": 0.8038866122250422,
      "learning_rate": 0.0009804256038019482,
      "loss": 2.7314,
      "step": 255
    },
    {
      "epoch": 0.11737734983952315,
      "grad_norm": 0.7991979872621745,
      "learning_rate": 0.0009802192996366857,
      "loss": 2.7575,
      "step": 256
    },
    {
      "epoch": 0.1178358551123338,
      "grad_norm": 0.8433751263727786,
      "learning_rate": 0.0009800119359285563,
      "loss": 2.6855,
      "step": 257
    },
    {
      "epoch": 0.11829436038514443,
      "grad_norm": 0.8347382459035291,
      "learning_rate": 0.0009798035131350813,
      "loss": 2.7341,
      "step": 258
    },
    {
      "epoch": 0.11875286565795506,
      "grad_norm": 0.8204001435933568,
      "learning_rate": 0.0009795940317161194,
      "loss": 2.6569,
      "step": 259
    },
    {
      "epoch": 0.1192113709307657,
      "grad_norm": 0.819965294285037,
      "learning_rate": 0.0009793834921338646,
      "loss": 2.8071,
      "step": 260
    },
    {
      "epoch": 0.11966987620357634,
      "grad_norm": 0.7781691012816704,
      "learning_rate": 0.0009791718948528457,
      "loss": 2.7438,
      "step": 261
    },
    {
      "epoch": 0.12012838147638698,
      "grad_norm": 0.8158899545140671,
      "learning_rate": 0.0009789592403399252,
      "loss": 2.6949,
      "step": 262
    },
    {
      "epoch": 0.12058688674919761,
      "grad_norm": 0.760364735898869,
      "learning_rate": 0.0009787455290642985,
      "loss": 2.7004,
      "step": 263
    },
    {
      "epoch": 0.12104539202200826,
      "grad_norm": 0.8687865839442547,
      "learning_rate": 0.000978530761497492,
      "loss": 2.8361,
      "step": 264
    },
    {
      "epoch": 0.12150389729481889,
      "grad_norm": 0.8199365655945232,
      "learning_rate": 0.0009783149381133633,
      "loss": 2.7372,
      "step": 265
    },
    {
      "epoch": 0.12196240256762952,
      "grad_norm": 0.7977641851123937,
      "learning_rate": 0.0009780980593880992,
      "loss": 2.712,
      "step": 266
    },
    {
      "epoch": 0.12242090784044017,
      "grad_norm": 0.7314399794166403,
      "learning_rate": 0.0009778801258002153,
      "loss": 2.7041,
      "step": 267
    },
    {
      "epoch": 0.1228794131132508,
      "grad_norm": 0.8812739344062223,
      "learning_rate": 0.000977661137830554,
      "loss": 2.6505,
      "step": 268
    },
    {
      "epoch": 0.12333791838606144,
      "grad_norm": 0.879907378929165,
      "learning_rate": 0.0009774410959622845,
      "loss": 2.6989,
      "step": 269
    },
    {
      "epoch": 0.12379642365887207,
      "grad_norm": 0.8000032999397415,
      "learning_rate": 0.000977220000680901,
      "loss": 2.5826,
      "step": 270
    },
    {
      "epoch": 0.12425492893168272,
      "grad_norm": 0.8768211258234881,
      "learning_rate": 0.000976997852474223,
      "loss": 2.6125,
      "step": 271
    },
    {
      "epoch": 0.12471343420449335,
      "grad_norm": 0.8370239913530501,
      "learning_rate": 0.0009767746518323914,
      "loss": 2.653,
      "step": 272
    },
    {
      "epoch": 0.12517193947730398,
      "grad_norm": 0.886063402496763,
      "learning_rate": 0.0009765503992478704,
      "loss": 2.6569,
      "step": 273
    },
    {
      "epoch": 0.12563044475011462,
      "grad_norm": 0.7963025748752814,
      "learning_rate": 0.0009763250952154449,
      "loss": 2.695,
      "step": 274
    },
    {
      "epoch": 0.12608895002292525,
      "grad_norm": 0.8618201312114302,
      "learning_rate": 0.0009760987402322195,
      "loss": 2.7004,
      "step": 275
    },
    {
      "epoch": 0.1265474552957359,
      "grad_norm": 0.7557161849195908,
      "learning_rate": 0.0009758713347976178,
      "loss": 2.6957,
      "step": 276
    },
    {
      "epoch": 0.12700596056854654,
      "grad_norm": 0.8062972641651752,
      "learning_rate": 0.000975642879413381,
      "loss": 2.7179,
      "step": 277
    },
    {
      "epoch": 0.12746446584135718,
      "grad_norm": 0.8176410488879629,
      "learning_rate": 0.0009754133745835665,
      "loss": 2.6662,
      "step": 278
    },
    {
      "epoch": 0.1279229711141678,
      "grad_norm": 0.8436616283128405,
      "learning_rate": 0.0009751828208145482,
      "loss": 2.6991,
      "step": 279
    },
    {
      "epoch": 0.12838147638697844,
      "grad_norm": 0.8826928549527348,
      "learning_rate": 0.0009749512186150131,
      "loss": 2.7612,
      "step": 280
    },
    {
      "epoch": 0.12883998165978908,
      "grad_norm": 0.8211600502682489,
      "learning_rate": 0.0009747185684959625,
      "loss": 2.7351,
      "step": 281
    },
    {
      "epoch": 0.12929848693259974,
      "grad_norm": 0.78809692336453,
      "learning_rate": 0.000974484870970709,
      "loss": 2.7142,
      "step": 282
    },
    {
      "epoch": 0.12975699220541037,
      "grad_norm": 0.8494687145499684,
      "learning_rate": 0.0009742501265548767,
      "loss": 2.6256,
      "step": 283
    },
    {
      "epoch": 0.130215497478221,
      "grad_norm": 0.8238540291331138,
      "learning_rate": 0.0009740143357663993,
      "loss": 2.7729,
      "step": 284
    },
    {
      "epoch": 0.13067400275103164,
      "grad_norm": 0.9834731785883287,
      "learning_rate": 0.000973777499125519,
      "loss": 2.6387,
      "step": 285
    },
    {
      "epoch": 0.13113250802384227,
      "grad_norm": 0.8980689144660529,
      "learning_rate": 0.0009735396171547859,
      "loss": 2.6566,
      "step": 286
    },
    {
      "epoch": 0.1315910132966529,
      "grad_norm": 0.8963311987540489,
      "learning_rate": 0.0009733006903790564,
      "loss": 2.7242,
      "step": 287
    },
    {
      "epoch": 0.13204951856946354,
      "grad_norm": 0.8386672294977313,
      "learning_rate": 0.0009730607193254922,
      "loss": 2.6764,
      "step": 288
    },
    {
      "epoch": 0.1325080238422742,
      "grad_norm": 0.8079881489290321,
      "learning_rate": 0.0009728197045235585,
      "loss": 2.7296,
      "step": 289
    },
    {
      "epoch": 0.13296652911508483,
      "grad_norm": 0.8948963652715817,
      "learning_rate": 0.0009725776465050242,
      "loss": 2.668,
      "step": 290
    },
    {
      "epoch": 0.13342503438789546,
      "grad_norm": 0.8347872523795483,
      "learning_rate": 0.0009723345458039594,
      "loss": 2.6313,
      "step": 291
    },
    {
      "epoch": 0.1338835396607061,
      "grad_norm": 0.812997931596397,
      "learning_rate": 0.000972090402956735,
      "loss": 2.6555,
      "step": 292
    },
    {
      "epoch": 0.13434204493351673,
      "grad_norm": 0.8333724680403146,
      "learning_rate": 0.0009718452185020212,
      "loss": 2.6156,
      "step": 293
    },
    {
      "epoch": 0.13480055020632736,
      "grad_norm": 0.8524865097590192,
      "learning_rate": 0.0009715989929807862,
      "loss": 2.6315,
      "step": 294
    },
    {
      "epoch": 0.13525905547913802,
      "grad_norm": 0.8100036207375677,
      "learning_rate": 0.0009713517269362955,
      "loss": 2.6952,
      "step": 295
    },
    {
      "epoch": 0.13571756075194866,
      "grad_norm": 0.8367568269863004,
      "learning_rate": 0.0009711034209141101,
      "loss": 2.6571,
      "step": 296
    },
    {
      "epoch": 0.1361760660247593,
      "grad_norm": 0.8328154596589041,
      "learning_rate": 0.0009708540754620856,
      "loss": 2.7102,
      "step": 297
    },
    {
      "epoch": 0.13663457129756992,
      "grad_norm": 0.807420535150202,
      "learning_rate": 0.0009706036911303713,
      "loss": 2.6966,
      "step": 298
    },
    {
      "epoch": 0.13709307657038056,
      "grad_norm": 0.8869319115501584,
      "learning_rate": 0.0009703522684714083,
      "loss": 2.5589,
      "step": 299
    },
    {
      "epoch": 0.1375515818431912,
      "grad_norm": 0.828507958522944,
      "learning_rate": 0.0009700998080399286,
      "loss": 2.6344,
      "step": 300
    },
    {
      "epoch": 0.13801008711600182,
      "grad_norm": 0.7865334677477023,
      "learning_rate": 0.0009698463103929542,
      "loss": 2.7052,
      "step": 301
    },
    {
      "epoch": 0.13846859238881248,
      "grad_norm": 0.9886952108544075,
      "learning_rate": 0.0009695917760897954,
      "loss": 2.6462,
      "step": 302
    },
    {
      "epoch": 0.13892709766162312,
      "grad_norm": 0.8115697820515401,
      "learning_rate": 0.0009693362056920501,
      "loss": 2.662,
      "step": 303
    },
    {
      "epoch": 0.13938560293443375,
      "grad_norm": 0.8524543146159462,
      "learning_rate": 0.0009690795997636015,
      "loss": 2.6153,
      "step": 304
    },
    {
      "epoch": 0.13984410820724438,
      "grad_norm": 0.8195007382120938,
      "learning_rate": 0.0009688219588706179,
      "loss": 2.6017,
      "step": 305
    },
    {
      "epoch": 0.14030261348005502,
      "grad_norm": 0.8751625194558688,
      "learning_rate": 0.0009685632835815518,
      "loss": 2.5996,
      "step": 306
    },
    {
      "epoch": 0.14076111875286565,
      "grad_norm": 0.8995505496280574,
      "learning_rate": 0.0009683035744671367,
      "loss": 2.5784,
      "step": 307
    },
    {
      "epoch": 0.14121962402567628,
      "grad_norm": 0.8925477160703733,
      "learning_rate": 0.0009680428321003883,
      "loss": 2.5505,
      "step": 308
    },
    {
      "epoch": 0.14167812929848694,
      "grad_norm": 0.9855144277407724,
      "learning_rate": 0.000967781057056601,
      "loss": 2.6354,
      "step": 309
    },
    {
      "epoch": 0.14213663457129758,
      "grad_norm": 0.9325930154748621,
      "learning_rate": 0.0009675182499133485,
      "loss": 2.6443,
      "step": 310
    },
    {
      "epoch": 0.1425951398441082,
      "grad_norm": 0.7952172292805091,
      "learning_rate": 0.0009672544112504813,
      "loss": 2.5989,
      "step": 311
    },
    {
      "epoch": 0.14305364511691884,
      "grad_norm": 0.8806074917190878,
      "learning_rate": 0.0009669895416501257,
      "loss": 2.5592,
      "step": 312
    },
    {
      "epoch": 0.14351215038972948,
      "grad_norm": 0.8688049016257948,
      "learning_rate": 0.0009667236416966833,
      "loss": 2.6452,
      "step": 313
    },
    {
      "epoch": 0.1439706556625401,
      "grad_norm": 0.8805710133934089,
      "learning_rate": 0.0009664567119768281,
      "loss": 2.6561,
      "step": 314
    },
    {
      "epoch": 0.14442916093535077,
      "grad_norm": 0.8749409030590802,
      "learning_rate": 0.0009661887530795067,
      "loss": 2.5688,
      "step": 315
    },
    {
      "epoch": 0.1448876662081614,
      "grad_norm": 0.8034733895816942,
      "learning_rate": 0.0009659197655959365,
      "loss": 2.5939,
      "step": 316
    },
    {
      "epoch": 0.14534617148097204,
      "grad_norm": 1.0007719655400986,
      "learning_rate": 0.000965649750119604,
      "loss": 2.6744,
      "step": 317
    },
    {
      "epoch": 0.14580467675378267,
      "grad_norm": 1.0058671117110065,
      "learning_rate": 0.0009653787072462643,
      "loss": 2.6472,
      "step": 318
    },
    {
      "epoch": 0.1462631820265933,
      "grad_norm": 0.8932748463188291,
      "learning_rate": 0.0009651066375739388,
      "loss": 2.6621,
      "step": 319
    },
    {
      "epoch": 0.14672168729940394,
      "grad_norm": 0.876154269444633,
      "learning_rate": 0.000964833541702915,
      "loss": 2.526,
      "step": 320
    },
    {
      "epoch": 0.14718019257221457,
      "grad_norm": 0.7845640827348987,
      "learning_rate": 0.0009645594202357438,
      "loss": 2.6271,
      "step": 321
    },
    {
      "epoch": 0.14763869784502523,
      "grad_norm": 0.8535158609634353,
      "learning_rate": 0.0009642842737772397,
      "loss": 2.5688,
      "step": 322
    },
    {
      "epoch": 0.14809720311783586,
      "grad_norm": 0.8553726532058795,
      "learning_rate": 0.0009640081029344782,
      "loss": 2.5946,
      "step": 323
    },
    {
      "epoch": 0.1485557083906465,
      "grad_norm": 0.9978699705241671,
      "learning_rate": 0.0009637309083167956,
      "loss": 2.6325,
      "step": 324
    },
    {
      "epoch": 0.14901421366345713,
      "grad_norm": 0.9019014427350965,
      "learning_rate": 0.0009634526905357859,
      "loss": 2.6304,
      "step": 325
    },
    {
      "epoch": 0.14947271893626776,
      "grad_norm": 0.7324906439683223,
      "learning_rate": 0.000963173450205302,
      "loss": 2.5971,
      "step": 326
    },
    {
      "epoch": 0.1499312242090784,
      "grad_norm": 0.7993542252852552,
      "learning_rate": 0.0009628931879414517,
      "loss": 2.51,
      "step": 327
    },
    {
      "epoch": 0.15038972948188903,
      "grad_norm": 0.7881714271214427,
      "learning_rate": 0.0009626119043625983,
      "loss": 2.5714,
      "step": 328
    },
    {
      "epoch": 0.1508482347546997,
      "grad_norm": 0.7991397525699102,
      "learning_rate": 0.0009623296000893582,
      "loss": 2.5165,
      "step": 329
    },
    {
      "epoch": 0.15130674002751032,
      "grad_norm": 0.9337400812774199,
      "learning_rate": 0.0009620462757446,
      "loss": 2.5875,
      "step": 330
    },
    {
      "epoch": 0.15176524530032096,
      "grad_norm": 0.8539692346591603,
      "learning_rate": 0.0009617619319534428,
      "loss": 2.5908,
      "step": 331
    },
    {
      "epoch": 0.1522237505731316,
      "grad_norm": 0.9403878286960008,
      "learning_rate": 0.000961476569343255,
      "loss": 2.5916,
      "step": 332
    },
    {
      "epoch": 0.15268225584594222,
      "grad_norm": 0.8118715378512626,
      "learning_rate": 0.0009611901885436529,
      "loss": 2.6172,
      "step": 333
    },
    {
      "epoch": 0.15314076111875286,
      "grad_norm": 0.8271797108396066,
      "learning_rate": 0.0009609027901864996,
      "loss": 2.4964,
      "step": 334
    },
    {
      "epoch": 0.15359926639156352,
      "grad_norm": 0.8758885260027227,
      "learning_rate": 0.0009606143749059029,
      "loss": 2.5411,
      "step": 335
    },
    {
      "epoch": 0.15405777166437415,
      "grad_norm": 0.8444198720054703,
      "learning_rate": 0.0009603249433382144,
      "loss": 2.6034,
      "step": 336
    },
    {
      "epoch": 0.15451627693718478,
      "grad_norm": 0.8641549900550156,
      "learning_rate": 0.0009600344961220282,
      "loss": 2.5762,
      "step": 337
    },
    {
      "epoch": 0.15497478220999542,
      "grad_norm": 0.9092580454087109,
      "learning_rate": 0.0009597430338981791,
      "loss": 2.605,
      "step": 338
    },
    {
      "epoch": 0.15543328748280605,
      "grad_norm": 0.8155912805243882,
      "learning_rate": 0.0009594505573097414,
      "loss": 2.5399,
      "step": 339
    },
    {
      "epoch": 0.15589179275561668,
      "grad_norm": 0.8083525711317538,
      "learning_rate": 0.0009591570670020277,
      "loss": 2.5998,
      "step": 340
    },
    {
      "epoch": 0.15635029802842731,
      "grad_norm": 0.9291872770721792,
      "learning_rate": 0.0009588625636225871,
      "loss": 2.5357,
      "step": 341
    },
    {
      "epoch": 0.15680880330123798,
      "grad_norm": 0.9541382456168027,
      "learning_rate": 0.0009585670478212036,
      "loss": 2.3913,
      "step": 342
    },
    {
      "epoch": 0.1572673085740486,
      "grad_norm": 0.9431911074957859,
      "learning_rate": 0.0009582705202498956,
      "loss": 2.5359,
      "step": 343
    },
    {
      "epoch": 0.15772581384685924,
      "grad_norm": 1.0240389313322378,
      "learning_rate": 0.0009579729815629133,
      "loss": 2.5889,
      "step": 344
    },
    {
      "epoch": 0.15818431911966988,
      "grad_norm": 0.8851979280585253,
      "learning_rate": 0.0009576744324167379,
      "loss": 2.5979,
      "step": 345
    },
    {
      "epoch": 0.1586428243924805,
      "grad_norm": 0.8519510072068981,
      "learning_rate": 0.0009573748734700804,
      "loss": 2.4751,
      "step": 346
    },
    {
      "epoch": 0.15910132966529114,
      "grad_norm": 0.8128911520956467,
      "learning_rate": 0.0009570743053838796,
      "loss": 2.5793,
      "step": 347
    },
    {
      "epoch": 0.15955983493810177,
      "grad_norm": 0.8261130139505831,
      "learning_rate": 0.0009567727288213005,
      "loss": 2.5248,
      "step": 348
    },
    {
      "epoch": 0.16001834021091244,
      "grad_norm": 0.8356169688136253,
      "learning_rate": 0.0009564701444477337,
      "loss": 2.5858,
      "step": 349
    },
    {
      "epoch": 0.16047684548372307,
      "grad_norm": 0.7728679288011396,
      "learning_rate": 0.000956166552930793,
      "loss": 2.5139,
      "step": 350
    },
    {
      "epoch": 0.1609353507565337,
      "grad_norm": 0.8325493289839105,
      "learning_rate": 0.0009558619549403147,
      "loss": 2.6345,
      "step": 351
    },
    {
      "epoch": 0.16139385602934433,
      "grad_norm": 0.8632555585998641,
      "learning_rate": 0.0009555563511483555,
      "loss": 2.5286,
      "step": 352
    },
    {
      "epoch": 0.16185236130215497,
      "grad_norm": 0.8166424217391273,
      "learning_rate": 0.0009552497422291912,
      "loss": 2.4922,
      "step": 353
    },
    {
      "epoch": 0.1623108665749656,
      "grad_norm": 0.7836144250943433,
      "learning_rate": 0.0009549421288593157,
      "loss": 2.5362,
      "step": 354
    },
    {
      "epoch": 0.16276937184777626,
      "grad_norm": 0.7784049971188488,
      "learning_rate": 0.0009546335117174385,
      "loss": 2.546,
      "step": 355
    },
    {
      "epoch": 0.1632278771205869,
      "grad_norm": 0.7832386157147156,
      "learning_rate": 0.0009543238914844843,
      "loss": 2.563,
      "step": 356
    },
    {
      "epoch": 0.16368638239339753,
      "grad_norm": 0.8508375555418638,
      "learning_rate": 0.0009540132688435907,
      "loss": 2.5331,
      "step": 357
    },
    {
      "epoch": 0.16414488766620816,
      "grad_norm": 0.8308992572858682,
      "learning_rate": 0.0009537016444801074,
      "loss": 2.4486,
      "step": 358
    },
    {
      "epoch": 0.1646033929390188,
      "grad_norm": 0.8783880010511094,
      "learning_rate": 0.0009533890190815935,
      "loss": 2.6236,
      "step": 359
    },
    {
      "epoch": 0.16506189821182943,
      "grad_norm": 0.86960349134162,
      "learning_rate": 0.0009530753933378173,
      "loss": 2.4985,
      "step": 360
    },
    {
      "epoch": 0.16552040348464006,
      "grad_norm": 0.8566401726460164,
      "learning_rate": 0.0009527607679407545,
      "loss": 2.5182,
      "step": 361
    },
    {
      "epoch": 0.16597890875745072,
      "grad_norm": 0.7955847865561575,
      "learning_rate": 0.0009524451435845857,
      "loss": 2.5669,
      "step": 362
    },
    {
      "epoch": 0.16643741403026135,
      "grad_norm": 0.856091653581431,
      "learning_rate": 0.0009521285209656963,
      "loss": 2.4964,
      "step": 363
    },
    {
      "epoch": 0.166895919303072,
      "grad_norm": 0.8020393561723776,
      "learning_rate": 0.0009518109007826734,
      "loss": 2.4125,
      "step": 364
    },
    {
      "epoch": 0.16735442457588262,
      "grad_norm": 0.8927413540238779,
      "learning_rate": 0.0009514922837363059,
      "loss": 2.5557,
      "step": 365
    },
    {
      "epoch": 0.16781292984869325,
      "grad_norm": 0.7578807470103401,
      "learning_rate": 0.0009511726705295817,
      "loss": 2.5046,
      "step": 366
    },
    {
      "epoch": 0.1682714351215039,
      "grad_norm": 0.8663694204880821,
      "learning_rate": 0.000950852061867687,
      "loss": 2.5502,
      "step": 367
    },
    {
      "epoch": 0.16872994039431455,
      "grad_norm": 0.8135842829996315,
      "learning_rate": 0.0009505304584580038,
      "loss": 2.4902,
      "step": 368
    },
    {
      "epoch": 0.16918844566712518,
      "grad_norm": 0.8354626076365095,
      "learning_rate": 0.0009502078610101092,
      "loss": 2.613,
      "step": 369
    },
    {
      "epoch": 0.16964695093993581,
      "grad_norm": 0.7769731973974763,
      "learning_rate": 0.0009498842702357736,
      "loss": 2.5799,
      "step": 370
    },
    {
      "epoch": 0.17010545621274645,
      "grad_norm": 0.8824597686295306,
      "learning_rate": 0.0009495596868489587,
      "loss": 2.5399,
      "step": 371
    },
    {
      "epoch": 0.17056396148555708,
      "grad_norm": 0.7809196636563762,
      "learning_rate": 0.0009492341115658167,
      "loss": 2.517,
      "step": 372
    },
    {
      "epoch": 0.1710224667583677,
      "grad_norm": 1.0809129483239044,
      "learning_rate": 0.0009489075451046879,
      "loss": 2.5141,
      "step": 373
    },
    {
      "epoch": 0.17148097203117835,
      "grad_norm": 0.8362069772097841,
      "learning_rate": 0.0009485799881861,
      "loss": 2.4329,
      "step": 374
    },
    {
      "epoch": 0.171939477303989,
      "grad_norm": 0.8683803595175134,
      "learning_rate": 0.0009482514415327654,
      "loss": 2.5353,
      "step": 375
    },
    {
      "epoch": 0.17239798257679964,
      "grad_norm": 0.7782012136440019,
      "learning_rate": 0.000947921905869581,
      "loss": 2.485,
      "step": 376
    },
    {
      "epoch": 0.17285648784961027,
      "grad_norm": 0.89436251565577,
      "learning_rate": 0.0009475913819236248,
      "loss": 2.5991,
      "step": 377
    },
    {
      "epoch": 0.1733149931224209,
      "grad_norm": 0.7614203207370723,
      "learning_rate": 0.0009472598704241561,
      "loss": 2.5567,
      "step": 378
    },
    {
      "epoch": 0.17377349839523154,
      "grad_norm": 0.7932741207792637,
      "learning_rate": 0.0009469273721026131,
      "loss": 2.4169,
      "step": 379
    },
    {
      "epoch": 0.17423200366804217,
      "grad_norm": 0.8493060851821493,
      "learning_rate": 0.0009465938876926111,
      "loss": 2.4354,
      "step": 380
    },
    {
      "epoch": 0.1746905089408528,
      "grad_norm": 0.790025557033006,
      "learning_rate": 0.0009462594179299406,
      "loss": 2.6392,
      "step": 381
    },
    {
      "epoch": 0.17514901421366347,
      "grad_norm": 0.8747291607446156,
      "learning_rate": 0.0009459239635525672,
      "loss": 2.4943,
      "step": 382
    },
    {
      "epoch": 0.1756075194864741,
      "grad_norm": 0.869813317256486,
      "learning_rate": 0.0009455875253006281,
      "loss": 2.4808,
      "step": 383
    },
    {
      "epoch": 0.17606602475928473,
      "grad_norm": 0.830913074900229,
      "learning_rate": 0.0009452501039164315,
      "loss": 2.6338,
      "step": 384
    },
    {
      "epoch": 0.17652453003209537,
      "grad_norm": 0.8182812657868653,
      "learning_rate": 0.0009449117001444549,
      "loss": 2.4197,
      "step": 385
    },
    {
      "epoch": 0.176983035304906,
      "grad_norm": 0.7751351927513044,
      "learning_rate": 0.0009445723147313433,
      "loss": 2.5391,
      "step": 386
    },
    {
      "epoch": 0.17744154057771663,
      "grad_norm": 0.8520025402228629,
      "learning_rate": 0.0009442319484259074,
      "loss": 2.4704,
      "step": 387
    },
    {
      "epoch": 0.1779000458505273,
      "grad_norm": 0.7781143912973151,
      "learning_rate": 0.0009438906019791222,
      "loss": 2.5553,
      "step": 388
    },
    {
      "epoch": 0.17835855112333793,
      "grad_norm": 0.7716340722167002,
      "learning_rate": 0.0009435482761441251,
      "loss": 2.5829,
      "step": 389
    },
    {
      "epoch": 0.17881705639614856,
      "grad_norm": 0.8050790567290881,
      "learning_rate": 0.000943204971676215,
      "loss": 2.5805,
      "step": 390
    },
    {
      "epoch": 0.1792755616689592,
      "grad_norm": 0.7637712089409552,
      "learning_rate": 0.0009428606893328493,
      "loss": 2.431,
      "step": 391
    },
    {
      "epoch": 0.17973406694176983,
      "grad_norm": 0.7712259412789307,
      "learning_rate": 0.0009425154298736432,
      "loss": 2.5019,
      "step": 392
    },
    {
      "epoch": 0.18019257221458046,
      "grad_norm": 0.8122610614983813,
      "learning_rate": 0.0009421691940603678,
      "loss": 2.4932,
      "step": 393
    },
    {
      "epoch": 0.1806510774873911,
      "grad_norm": 0.760494362284887,
      "learning_rate": 0.0009418219826569488,
      "loss": 2.6215,
      "step": 394
    },
    {
      "epoch": 0.18110958276020175,
      "grad_norm": 0.7968036848184551,
      "learning_rate": 0.0009414737964294635,
      "loss": 2.3897,
      "step": 395
    },
    {
      "epoch": 0.1815680880330124,
      "grad_norm": 0.7820327127701531,
      "learning_rate": 0.000941124636146141,
      "loss": 2.4505,
      "step": 396
    },
    {
      "epoch": 0.18202659330582302,
      "grad_norm": 0.8927633073338856,
      "learning_rate": 0.0009407745025773589,
      "loss": 2.3896,
      "step": 397
    },
    {
      "epoch": 0.18248509857863365,
      "grad_norm": 0.7879497168162005,
      "learning_rate": 0.0009404233964956423,
      "loss": 2.3223,
      "step": 398
    },
    {
      "epoch": 0.1829436038514443,
      "grad_norm": 0.8660506093511949,
      "learning_rate": 0.0009400713186756625,
      "loss": 2.4872,
      "step": 399
    },
    {
      "epoch": 0.18340210912425492,
      "grad_norm": 0.8220806961554811,
      "learning_rate": 0.0009397182698942342,
      "loss": 2.4557,
      "step": 400
    },
    {
      "epoch": 0.18386061439706555,
      "grad_norm": 0.7082158487201254,
      "learning_rate": 0.0009393642509303149,
      "loss": 2.4175,
      "step": 401
    },
    {
      "epoch": 0.1843191196698762,
      "grad_norm": 0.8088196684540367,
      "learning_rate": 0.0009390092625650023,
      "loss": 2.5124,
      "step": 402
    },
    {
      "epoch": 0.18477762494268685,
      "grad_norm": 0.7603070816867282,
      "learning_rate": 0.0009386533055815332,
      "loss": 2.4556,
      "step": 403
    },
    {
      "epoch": 0.18523613021549748,
      "grad_norm": 0.7582072389437988,
      "learning_rate": 0.0009382963807652813,
      "loss": 2.4405,
      "step": 404
    },
    {
      "epoch": 0.1856946354883081,
      "grad_norm": 0.7277167809247085,
      "learning_rate": 0.000937938488903756,
      "loss": 2.5005,
      "step": 405
    },
    {
      "epoch": 0.18615314076111875,
      "grad_norm": 0.8247351736198616,
      "learning_rate": 0.0009375796307866003,
      "loss": 2.5084,
      "step": 406
    },
    {
      "epoch": 0.18661164603392938,
      "grad_norm": 0.7369274401126619,
      "learning_rate": 0.0009372198072055888,
      "loss": 2.5215,
      "step": 407
    },
    {
      "epoch": 0.18707015130674004,
      "grad_norm": 0.771863027229834,
      "learning_rate": 0.0009368590189546268,
      "loss": 2.4161,
      "step": 408
    },
    {
      "epoch": 0.18752865657955067,
      "grad_norm": 0.7818918040170861,
      "learning_rate": 0.0009364972668297474,
      "loss": 2.5012,
      "step": 409
    },
    {
      "epoch": 0.1879871618523613,
      "grad_norm": 0.7520053927800081,
      "learning_rate": 0.0009361345516291111,
      "loss": 2.41,
      "step": 410
    },
    {
      "epoch": 0.18844566712517194,
      "grad_norm": 0.7962565102922315,
      "learning_rate": 0.0009357708741530024,
      "loss": 2.5009,
      "step": 411
    },
    {
      "epoch": 0.18890417239798257,
      "grad_norm": 0.7835300940572997,
      "learning_rate": 0.00093540623520383,
      "loss": 2.4258,
      "step": 412
    },
    {
      "epoch": 0.1893626776707932,
      "grad_norm": 0.7571898508895158,
      "learning_rate": 0.000935040635586123,
      "loss": 2.4366,
      "step": 413
    },
    {
      "epoch": 0.18982118294360384,
      "grad_norm": 0.7651806346636322,
      "learning_rate": 0.0009346740761065305,
      "loss": 2.3835,
      "step": 414
    },
    {
      "epoch": 0.1902796882164145,
      "grad_norm": 0.8112195736566856,
      "learning_rate": 0.0009343065575738197,
      "loss": 2.4179,
      "step": 415
    },
    {
      "epoch": 0.19073819348922513,
      "grad_norm": 0.7497204887058221,
      "learning_rate": 0.0009339380807988733,
      "loss": 2.4741,
      "step": 416
    },
    {
      "epoch": 0.19119669876203577,
      "grad_norm": 0.7049394326500392,
      "learning_rate": 0.0009335686465946887,
      "loss": 2.4693,
      "step": 417
    },
    {
      "epoch": 0.1916552040348464,
      "grad_norm": 0.8812583232816941,
      "learning_rate": 0.0009331982557763754,
      "loss": 2.5378,
      "step": 418
    },
    {
      "epoch": 0.19211370930765703,
      "grad_norm": 0.7551418024943931,
      "learning_rate": 0.0009328269091611537,
      "loss": 2.3332,
      "step": 419
    },
    {
      "epoch": 0.19257221458046767,
      "grad_norm": 0.8061706407603213,
      "learning_rate": 0.0009324546075683524,
      "loss": 2.5538,
      "step": 420
    },
    {
      "epoch": 0.19303071985327833,
      "grad_norm": 0.8223433341045504,
      "learning_rate": 0.0009320813518194083,
      "loss": 2.5058,
      "step": 421
    },
    {
      "epoch": 0.19348922512608896,
      "grad_norm": 0.7697888170732529,
      "learning_rate": 0.0009317071427378624,
      "loss": 2.4088,
      "step": 422
    },
    {
      "epoch": 0.1939477303988996,
      "grad_norm": 0.7695811370168018,
      "learning_rate": 0.0009313319811493594,
      "loss": 2.4141,
      "step": 423
    },
    {
      "epoch": 0.19440623567171023,
      "grad_norm": 0.8048797534799682,
      "learning_rate": 0.000930955867881646,
      "loss": 2.4963,
      "step": 424
    },
    {
      "epoch": 0.19486474094452086,
      "grad_norm": 0.7254753499018545,
      "learning_rate": 0.0009305788037645681,
      "loss": 2.3963,
      "step": 425
    },
    {
      "epoch": 0.1953232462173315,
      "grad_norm": 0.7798033971107133,
      "learning_rate": 0.0009302007896300697,
      "loss": 2.4889,
      "step": 426
    },
    {
      "epoch": 0.19578175149014213,
      "grad_norm": 0.7911012177532921,
      "learning_rate": 0.0009298218263121911,
      "loss": 2.4338,
      "step": 427
    },
    {
      "epoch": 0.19624025676295279,
      "grad_norm": 0.8075673778867486,
      "learning_rate": 0.0009294419146470668,
      "loss": 2.5079,
      "step": 428
    },
    {
      "epoch": 0.19669876203576342,
      "grad_norm": 0.7178599630237458,
      "learning_rate": 0.0009290610554729234,
      "loss": 2.4346,
      "step": 429
    },
    {
      "epoch": 0.19715726730857405,
      "grad_norm": 0.7431861104479859,
      "learning_rate": 0.0009286792496300784,
      "loss": 2.5153,
      "step": 430
    },
    {
      "epoch": 0.19761577258138469,
      "grad_norm": 0.7319904816079895,
      "learning_rate": 0.0009282964979609379,
      "loss": 2.4477,
      "step": 431
    },
    {
      "epoch": 0.19807427785419532,
      "grad_norm": 0.7548513598693645,
      "learning_rate": 0.0009279128013099947,
      "loss": 2.4276,
      "step": 432
    },
    {
      "epoch": 0.19853278312700595,
      "grad_norm": 0.7376110702681541,
      "learning_rate": 0.0009275281605238268,
      "loss": 2.4787,
      "step": 433
    },
    {
      "epoch": 0.19899128839981658,
      "grad_norm": 0.7702197744764998,
      "learning_rate": 0.0009271425764510953,
      "loss": 2.512,
      "step": 434
    },
    {
      "epoch": 0.19944979367262725,
      "grad_norm": 0.756230400207134,
      "learning_rate": 0.0009267560499425423,
      "loss": 2.3187,
      "step": 435
    },
    {
      "epoch": 0.19990829894543788,
      "grad_norm": 0.7775923954505322,
      "learning_rate": 0.0009263685818509895,
      "loss": 2.3962,
      "step": 436
    },
    {
      "epoch": 0.2003668042182485,
      "grad_norm": 0.7458802403436309,
      "learning_rate": 0.000925980173031336,
      "loss": 2.4409,
      "step": 437
    },
    {
      "epoch": 0.20082530949105915,
      "grad_norm": 0.7175180962656307,
      "learning_rate": 0.0009255908243405567,
      "loss": 2.4028,
      "step": 438
    },
    {
      "epoch": 0.20128381476386978,
      "grad_norm": 0.802860817310806,
      "learning_rate": 0.0009252005366376996,
      "loss": 2.5307,
      "step": 439
    },
    {
      "epoch": 0.2017423200366804,
      "grad_norm": 0.7649480055826658,
      "learning_rate": 0.0009248093107838852,
      "loss": 2.4441,
      "step": 440
    },
    {
      "epoch": 0.20220082530949107,
      "grad_norm": 0.7674659165904175,
      "learning_rate": 0.0009244171476423036,
      "loss": 2.463,
      "step": 441
    },
    {
      "epoch": 0.2026593305823017,
      "grad_norm": 0.7179326240974102,
      "learning_rate": 0.0009240240480782129,
      "loss": 2.3516,
      "step": 442
    },
    {
      "epoch": 0.20311783585511234,
      "grad_norm": 0.7166309992242168,
      "learning_rate": 0.0009236300129589376,
      "loss": 2.5359,
      "step": 443
    },
    {
      "epoch": 0.20357634112792297,
      "grad_norm": 0.7135406496979464,
      "learning_rate": 0.0009232350431538657,
      "loss": 2.5204,
      "step": 444
    },
    {
      "epoch": 0.2040348464007336,
      "grad_norm": 0.8000918463525477,
      "learning_rate": 0.0009228391395344482,
      "loss": 2.463,
      "step": 445
    },
    {
      "epoch": 0.20449335167354424,
      "grad_norm": 0.8018505445593266,
      "learning_rate": 0.000922442302974196,
      "loss": 2.524,
      "step": 446
    },
    {
      "epoch": 0.20495185694635487,
      "grad_norm": 0.7350313489510171,
      "learning_rate": 0.0009220445343486785,
      "loss": 2.462,
      "step": 447
    },
    {
      "epoch": 0.20541036221916553,
      "grad_norm": 0.7186027084309635,
      "learning_rate": 0.0009216458345355217,
      "loss": 2.461,
      "step": 448
    },
    {
      "epoch": 0.20586886749197617,
      "grad_norm": 0.7660464549565688,
      "learning_rate": 0.0009212462044144061,
      "loss": 2.5234,
      "step": 449
    },
    {
      "epoch": 0.2063273727647868,
      "grad_norm": 0.7598860200993713,
      "learning_rate": 0.0009208456448670648,
      "loss": 2.435,
      "step": 450
    },
    {
      "epoch": 0.20678587803759743,
      "grad_norm": 0.7498734862708083,
      "learning_rate": 0.0009204441567772816,
      "loss": 2.5197,
      "step": 451
    },
    {
      "epoch": 0.20724438331040806,
      "grad_norm": 0.7064824876721313,
      "learning_rate": 0.0009200417410308888,
      "loss": 2.4132,
      "step": 452
    },
    {
      "epoch": 0.2077028885832187,
      "grad_norm": 0.7253687513158725,
      "learning_rate": 0.0009196383985157656,
      "loss": 2.4877,
      "step": 453
    },
    {
      "epoch": 0.20816139385602933,
      "grad_norm": 0.6896311109177092,
      "learning_rate": 0.000919234130121836,
      "loss": 2.3802,
      "step": 454
    },
    {
      "epoch": 0.20861989912884,
      "grad_norm": 0.8113234117224787,
      "learning_rate": 0.0009188289367410672,
      "loss": 2.4151,
      "step": 455
    },
    {
      "epoch": 0.20907840440165062,
      "grad_norm": 0.726900226465735,
      "learning_rate": 0.0009184228192674666,
      "loss": 2.3754,
      "step": 456
    },
    {
      "epoch": 0.20953690967446126,
      "grad_norm": 0.7070867644017508,
      "learning_rate": 0.0009180157785970808,
      "loss": 2.5369,
      "step": 457
    },
    {
      "epoch": 0.2099954149472719,
      "grad_norm": 0.6915459891971876,
      "learning_rate": 0.0009176078156279932,
      "loss": 2.3762,
      "step": 458
    },
    {
      "epoch": 0.21045392022008252,
      "grad_norm": 0.7339223402861734,
      "learning_rate": 0.0009171989312603226,
      "loss": 2.4292,
      "step": 459
    },
    {
      "epoch": 0.21091242549289316,
      "grad_norm": 0.7765067431315216,
      "learning_rate": 0.0009167891263962202,
      "loss": 2.4498,
      "step": 460
    },
    {
      "epoch": 0.21137093076570382,
      "grad_norm": 0.8215780206985055,
      "learning_rate": 0.0009163784019398685,
      "loss": 2.442,
      "step": 461
    },
    {
      "epoch": 0.21182943603851445,
      "grad_norm": 0.7466907220561309,
      "learning_rate": 0.0009159667587974785,
      "loss": 2.5597,
      "step": 462
    },
    {
      "epoch": 0.21228794131132508,
      "grad_norm": 0.7647269799798667,
      "learning_rate": 0.0009155541978772887,
      "loss": 2.3649,
      "step": 463
    },
    {
      "epoch": 0.21274644658413572,
      "grad_norm": 0.7721184737809279,
      "learning_rate": 0.0009151407200895625,
      "loss": 2.3309,
      "step": 464
    },
    {
      "epoch": 0.21320495185694635,
      "grad_norm": 0.7435121607245686,
      "learning_rate": 0.000914726326346586,
      "loss": 2.3418,
      "step": 465
    },
    {
      "epoch": 0.21366345712975698,
      "grad_norm": 0.8435751032252214,
      "learning_rate": 0.0009143110175626661,
      "loss": 2.509,
      "step": 466
    },
    {
      "epoch": 0.21412196240256762,
      "grad_norm": 0.8162326892057103,
      "learning_rate": 0.0009138947946541291,
      "loss": 2.5614,
      "step": 467
    },
    {
      "epoch": 0.21458046767537828,
      "grad_norm": 0.7576924970929577,
      "learning_rate": 0.0009134776585393181,
      "loss": 2.4292,
      "step": 468
    },
    {
      "epoch": 0.2150389729481889,
      "grad_norm": 0.7806396448211431,
      "learning_rate": 0.0009130596101385906,
      "loss": 2.4884,
      "step": 469
    },
    {
      "epoch": 0.21549747822099954,
      "grad_norm": 0.7120552663719133,
      "learning_rate": 0.0009126406503743174,
      "loss": 2.4621,
      "step": 470
    },
    {
      "epoch": 0.21595598349381018,
      "grad_norm": 0.7380520229255588,
      "learning_rate": 0.0009122207801708802,
      "loss": 2.3346,
      "step": 471
    },
    {
      "epoch": 0.2164144887666208,
      "grad_norm": 0.8137535550827742,
      "learning_rate": 0.0009118000004546689,
      "loss": 2.4981,
      "step": 472
    },
    {
      "epoch": 0.21687299403943144,
      "grad_norm": 0.7294462805050542,
      "learning_rate": 0.0009113783121540807,
      "loss": 2.3717,
      "step": 473
    },
    {
      "epoch": 0.2173314993122421,
      "grad_norm": 0.7054271176070803,
      "learning_rate": 0.0009109557161995172,
      "loss": 2.496,
      "step": 474
    },
    {
      "epoch": 0.21779000458505274,
      "grad_norm": 0.7056402279998899,
      "learning_rate": 0.0009105322135233828,
      "loss": 2.466,
      "step": 475
    },
    {
      "epoch": 0.21824850985786337,
      "grad_norm": 0.7070566617037224,
      "learning_rate": 0.0009101078050600821,
      "loss": 2.4247,
      "step": 476
    },
    {
      "epoch": 0.218707015130674,
      "grad_norm": 0.7430113161663215,
      "learning_rate": 0.0009096824917460186,
      "loss": 2.4149,
      "step": 477
    },
    {
      "epoch": 0.21916552040348464,
      "grad_norm": 0.7386566686994099,
      "learning_rate": 0.0009092562745195921,
      "loss": 2.4195,
      "step": 478
    },
    {
      "epoch": 0.21962402567629527,
      "grad_norm": 0.7072920204751556,
      "learning_rate": 0.0009088291543211967,
      "loss": 2.4392,
      "step": 479
    },
    {
      "epoch": 0.2200825309491059,
      "grad_norm": 0.8095252787519912,
      "learning_rate": 0.0009084011320932188,
      "loss": 2.4231,
      "step": 480
    },
    {
      "epoch": 0.22054103622191656,
      "grad_norm": 0.71178602286616,
      "learning_rate": 0.0009079722087800352,
      "loss": 2.3605,
      "step": 481
    },
    {
      "epoch": 0.2209995414947272,
      "grad_norm": 0.8123351388204253,
      "learning_rate": 0.0009075423853280106,
      "loss": 2.4548,
      "step": 482
    },
    {
      "epoch": 0.22145804676753783,
      "grad_norm": 0.7033907542074143,
      "learning_rate": 0.0009071116626854958,
      "loss": 2.4309,
      "step": 483
    },
    {
      "epoch": 0.22191655204034846,
      "grad_norm": 0.8265026204217754,
      "learning_rate": 0.0009066800418028256,
      "loss": 2.5504,
      "step": 484
    },
    {
      "epoch": 0.2223750573131591,
      "grad_norm": 0.722826061361667,
      "learning_rate": 0.0009062475236323168,
      "loss": 2.4333,
      "step": 485
    },
    {
      "epoch": 0.22283356258596973,
      "grad_norm": 0.7526337566029124,
      "learning_rate": 0.0009058141091282656,
      "loss": 2.4198,
      "step": 486
    },
    {
      "epoch": 0.22329206785878036,
      "grad_norm": 0.6949257148636723,
      "learning_rate": 0.0009053797992469461,
      "loss": 2.4335,
      "step": 487
    },
    {
      "epoch": 0.22375057313159102,
      "grad_norm": 0.7365538030475173,
      "learning_rate": 0.0009049445949466078,
      "loss": 2.4147,
      "step": 488
    },
    {
      "epoch": 0.22420907840440166,
      "grad_norm": 0.7391937592981598,
      "learning_rate": 0.0009045084971874737,
      "loss": 2.4018,
      "step": 489
    },
    {
      "epoch": 0.2246675836772123,
      "grad_norm": 0.8858043961355212,
      "learning_rate": 0.0009040715069317382,
      "loss": 2.3544,
      "step": 490
    },
    {
      "epoch": 0.22512608895002292,
      "grad_norm": 0.7561547814644227,
      "learning_rate": 0.0009036336251435648,
      "loss": 2.4117,
      "step": 491
    },
    {
      "epoch": 0.22558459422283356,
      "grad_norm": 0.7905837773940351,
      "learning_rate": 0.0009031948527890839,
      "loss": 2.3704,
      "step": 492
    },
    {
      "epoch": 0.2260430994956442,
      "grad_norm": 0.75386768114088,
      "learning_rate": 0.000902755190836391,
      "loss": 2.4222,
      "step": 493
    },
    {
      "epoch": 0.22650160476845485,
      "grad_norm": 0.782868944699806,
      "learning_rate": 0.0009023146402555442,
      "loss": 2.4174,
      "step": 494
    },
    {
      "epoch": 0.22696011004126548,
      "grad_norm": 0.7484390413026653,
      "learning_rate": 0.0009018732020185624,
      "loss": 2.4072,
      "step": 495
    },
    {
      "epoch": 0.22741861531407612,
      "grad_norm": 0.7291616170137472,
      "learning_rate": 0.0009014308770994235,
      "loss": 2.4757,
      "step": 496
    },
    {
      "epoch": 0.22787712058688675,
      "grad_norm": 0.7372546839020957,
      "learning_rate": 0.0009009876664740605,
      "loss": 2.4996,
      "step": 497
    },
    {
      "epoch": 0.22833562585969738,
      "grad_norm": 0.7144532087104659,
      "learning_rate": 0.0009005435711203618,
      "loss": 2.529,
      "step": 498
    },
    {
      "epoch": 0.22879413113250802,
      "grad_norm": 0.7890303525979776,
      "learning_rate": 0.000900098592018167,
      "loss": 2.4343,
      "step": 499
    },
    {
      "epoch": 0.22925263640531865,
      "grad_norm": 0.6677197287937306,
      "learning_rate": 0.0008996527301492663,
      "loss": 2.3341,
      "step": 500
    },
    {
      "epoch": 0.2297111416781293,
      "grad_norm": 0.7553689267501004,
      "learning_rate": 0.0008992059864973972,
      "loss": 2.3808,
      "step": 501
    },
    {
      "epoch": 0.23016964695093994,
      "grad_norm": 0.7695644097775038,
      "learning_rate": 0.0008987583620482427,
      "loss": 2.3451,
      "step": 502
    },
    {
      "epoch": 0.23062815222375058,
      "grad_norm": 0.7787996478822583,
      "learning_rate": 0.0008983098577894292,
      "loss": 2.3807,
      "step": 503
    },
    {
      "epoch": 0.2310866574965612,
      "grad_norm": 0.7064137664848212,
      "learning_rate": 0.0008978604747105246,
      "loss": 2.3916,
      "step": 504
    },
    {
      "epoch": 0.23154516276937184,
      "grad_norm": 0.7072402798401647,
      "learning_rate": 0.0008974102138030354,
      "loss": 2.5351,
      "step": 505
    },
    {
      "epoch": 0.23200366804218248,
      "grad_norm": 0.7244942852766995,
      "learning_rate": 0.000896959076060405,
      "loss": 2.379,
      "step": 506
    },
    {
      "epoch": 0.2324621733149931,
      "grad_norm": 0.6470791709237697,
      "learning_rate": 0.0008965070624780116,
      "loss": 2.3881,
      "step": 507
    },
    {
      "epoch": 0.23292067858780377,
      "grad_norm": 0.7118664788652554,
      "learning_rate": 0.0008960541740531658,
      "loss": 2.4107,
      "step": 508
    },
    {
      "epoch": 0.2333791838606144,
      "grad_norm": 0.7373596674586275,
      "learning_rate": 0.0008956004117851083,
      "loss": 2.3445,
      "step": 509
    },
    {
      "epoch": 0.23383768913342504,
      "grad_norm": 0.8176189542457714,
      "learning_rate": 0.0008951457766750079,
      "loss": 2.5364,
      "step": 510
    },
    {
      "epoch": 0.23429619440623567,
      "grad_norm": 0.7593898857293397,
      "learning_rate": 0.0008946902697259593,
      "loss": 2.3295,
      "step": 511
    },
    {
      "epoch": 0.2347546996790463,
      "grad_norm": 0.7239338786496831,
      "learning_rate": 0.0008942338919429805,
      "loss": 2.4207,
      "step": 512
    },
    {
      "epoch": 0.23521320495185694,
      "grad_norm": 0.7500358057819337,
      "learning_rate": 0.0008937766443330113,
      "loss": 2.4091,
      "step": 513
    },
    {
      "epoch": 0.2356717102246676,
      "grad_norm": 0.6787965717082521,
      "learning_rate": 0.0008933185279049103,
      "loss": 2.44,
      "step": 514
    },
    {
      "epoch": 0.23613021549747823,
      "grad_norm": 0.7477710410400258,
      "learning_rate": 0.0008928595436694532,
      "loss": 2.3693,
      "step": 515
    },
    {
      "epoch": 0.23658872077028886,
      "grad_norm": 0.7506830010002995,
      "learning_rate": 0.0008923996926393305,
      "loss": 2.4416,
      "step": 516
    },
    {
      "epoch": 0.2370472260430995,
      "grad_norm": 0.6670619389591667,
      "learning_rate": 0.0008919389758291449,
      "loss": 2.3149,
      "step": 517
    },
    {
      "epoch": 0.23750573131591013,
      "grad_norm": 0.6419319443679041,
      "learning_rate": 0.0008914773942554098,
      "loss": 2.413,
      "step": 518
    },
    {
      "epoch": 0.23796423658872076,
      "grad_norm": 0.6641996866934272,
      "learning_rate": 0.000891014948936546,
      "loss": 2.4443,
      "step": 519
    },
    {
      "epoch": 0.2384227418615314,
      "grad_norm": 0.7300861940721249,
      "learning_rate": 0.0008905516408928804,
      "loss": 2.4332,
      "step": 520
    },
    {
      "epoch": 0.23888124713434206,
      "grad_norm": 0.7438824612084785,
      "learning_rate": 0.0008900874711466434,
      "loss": 2.477,
      "step": 521
    },
    {
      "epoch": 0.2393397524071527,
      "grad_norm": 0.7121802690070448,
      "learning_rate": 0.0008896224407219666,
      "loss": 2.3205,
      "step": 522
    },
    {
      "epoch": 0.23979825767996332,
      "grad_norm": 0.7586823373655908,
      "learning_rate": 0.0008891565506448804,
      "loss": 2.3618,
      "step": 523
    },
    {
      "epoch": 0.24025676295277396,
      "grad_norm": 0.740215026080773,
      "learning_rate": 0.0008886898019433122,
      "loss": 2.3321,
      "step": 524
    },
    {
      "epoch": 0.2407152682255846,
      "grad_norm": 0.7145263533648107,
      "learning_rate": 0.0008882221956470836,
      "loss": 2.4499,
      "step": 525
    },
    {
      "epoch": 0.24117377349839522,
      "grad_norm": 0.6940032265628713,
      "learning_rate": 0.0008877537327879086,
      "loss": 2.39,
      "step": 526
    },
    {
      "epoch": 0.24163227877120586,
      "grad_norm": 0.690551438642566,
      "learning_rate": 0.0008872844143993908,
      "loss": 2.3434,
      "step": 527
    },
    {
      "epoch": 0.24209078404401652,
      "grad_norm": 0.7027153948787254,
      "learning_rate": 0.0008868142415170218,
      "loss": 2.3127,
      "step": 528
    },
    {
      "epoch": 0.24254928931682715,
      "grad_norm": 0.6825459616909267,
      "learning_rate": 0.0008863432151781781,
      "loss": 2.4074,
      "step": 529
    },
    {
      "epoch": 0.24300779458963778,
      "grad_norm": 0.7548589401538661,
      "learning_rate": 0.0008858713364221195,
      "loss": 2.3916,
      "step": 530
    },
    {
      "epoch": 0.24346629986244842,
      "grad_norm": 0.7484927629275349,
      "learning_rate": 0.0008853986062899868,
      "loss": 2.4503,
      "step": 531
    },
    {
      "epoch": 0.24392480513525905,
      "grad_norm": 0.7257595730116329,
      "learning_rate": 0.0008849250258247986,
      "loss": 2.4751,
      "step": 532
    },
    {
      "epoch": 0.24438331040806968,
      "grad_norm": 0.6967883895684843,
      "learning_rate": 0.0008844505960714503,
      "loss": 2.3198,
      "step": 533
    },
    {
      "epoch": 0.24484181568088034,
      "grad_norm": 0.6983789416337913,
      "learning_rate": 0.0008839753180767108,
      "loss": 2.4165,
      "step": 534
    },
    {
      "epoch": 0.24530032095369098,
      "grad_norm": 0.6900150010266267,
      "learning_rate": 0.0008834991928892204,
      "loss": 2.509,
      "step": 535
    },
    {
      "epoch": 0.2457588262265016,
      "grad_norm": 0.6911632828128141,
      "learning_rate": 0.000883022221559489,
      "loss": 2.4282,
      "step": 536
    },
    {
      "epoch": 0.24621733149931224,
      "grad_norm": 0.698148177185482,
      "learning_rate": 0.0008825444051398934,
      "loss": 2.5008,
      "step": 537
    },
    {
      "epoch": 0.24667583677212288,
      "grad_norm": 0.6705314683363006,
      "learning_rate": 0.0008820657446846745,
      "loss": 2.4859,
      "step": 538
    },
    {
      "epoch": 0.2471343420449335,
      "grad_norm": 0.659443564756931,
      "learning_rate": 0.000881586241249936,
      "loss": 2.322,
      "step": 539
    },
    {
      "epoch": 0.24759284731774414,
      "grad_norm": 0.6611931664033653,
      "learning_rate": 0.0008811058958936411,
      "loss": 2.3883,
      "step": 540
    },
    {
      "epoch": 0.2480513525905548,
      "grad_norm": 0.7372830919867276,
      "learning_rate": 0.000880624709675611,
      "loss": 2.269,
      "step": 541
    },
    {
      "epoch": 0.24850985786336544,
      "grad_norm": 0.7381763189731727,
      "learning_rate": 0.000880142683657522,
      "loss": 2.4916,
      "step": 542
    },
    {
      "epoch": 0.24896836313617607,
      "grad_norm": 0.643630909457944,
      "learning_rate": 0.0008796598189029029,
      "loss": 2.3461,
      "step": 543
    },
    {
      "epoch": 0.2494268684089867,
      "grad_norm": 0.7132863217447835,
      "learning_rate": 0.0008791761164771338,
      "loss": 2.39,
      "step": 544
    },
    {
      "epoch": 0.24988537368179733,
      "grad_norm": 0.7149931313575791,
      "learning_rate": 0.0008786915774474424,
      "loss": 2.4294,
      "step": 545
    },
    {
      "epoch": 0.25034387895460797,
      "grad_norm": 0.7875538144647646,
      "learning_rate": 0.0008782062028829027,
      "loss": 2.547,
      "step": 546
    },
    {
      "epoch": 0.25080238422741863,
      "grad_norm": 0.7729170513026269,
      "learning_rate": 0.0008777199938544318,
      "loss": 2.3609,
      "step": 547
    },
    {
      "epoch": 0.25126088950022923,
      "grad_norm": 0.6888223724396384,
      "learning_rate": 0.0008772329514347883,
      "loss": 2.3418,
      "step": 548
    },
    {
      "epoch": 0.2517193947730399,
      "grad_norm": 0.7330218275874657,
      "learning_rate": 0.0008767450766985694,
      "loss": 2.4956,
      "step": 549
    },
    {
      "epoch": 0.2521779000458505,
      "grad_norm": 0.6599628753201839,
      "learning_rate": 0.0008762563707222086,
      "loss": 2.3531,
      "step": 550
    },
    {
      "epoch": 0.25263640531866116,
      "grad_norm": 0.697889764877089,
      "learning_rate": 0.0008757668345839738,
      "loss": 2.4478,
      "step": 551
    },
    {
      "epoch": 0.2530949105914718,
      "grad_norm": 0.6833236924688201,
      "learning_rate": 0.0008752764693639638,
      "loss": 2.4389,
      "step": 552
    },
    {
      "epoch": 0.2535534158642824,
      "grad_norm": 0.6889112035645604,
      "learning_rate": 0.0008747852761441078,
      "loss": 2.5055,
      "step": 553
    },
    {
      "epoch": 0.2540119211370931,
      "grad_norm": 0.6726810620975836,
      "learning_rate": 0.0008742932560081607,
      "loss": 2.4563,
      "step": 554
    },
    {
      "epoch": 0.2544704264099037,
      "grad_norm": 0.6629663117306965,
      "learning_rate": 0.0008738004100417025,
      "loss": 2.4245,
      "step": 555
    },
    {
      "epoch": 0.25492893168271435,
      "grad_norm": 0.6840385577038354,
      "learning_rate": 0.0008733067393321355,
      "loss": 2.4313,
      "step": 556
    },
    {
      "epoch": 0.255387436955525,
      "grad_norm": 0.6733387909643092,
      "learning_rate": 0.000872812244968681,
      "loss": 2.4157,
      "step": 557
    },
    {
      "epoch": 0.2558459422283356,
      "grad_norm": 0.6712354922774015,
      "learning_rate": 0.0008723169280423783,
      "loss": 2.4807,
      "step": 558
    },
    {
      "epoch": 0.2563044475011463,
      "grad_norm": 0.6663735543889616,
      "learning_rate": 0.0008718207896460811,
      "loss": 2.4349,
      "step": 559
    },
    {
      "epoch": 0.2567629527739569,
      "grad_norm": 0.6564837258886411,
      "learning_rate": 0.0008713238308744557,
      "loss": 2.3892,
      "step": 560
    },
    {
      "epoch": 0.25722145804676755,
      "grad_norm": 0.7522235704107716,
      "learning_rate": 0.0008708260528239789,
      "loss": 2.3342,
      "step": 561
    },
    {
      "epoch": 0.25767996331957815,
      "grad_norm": 0.690337929831646,
      "learning_rate": 0.000870327456592934,
      "loss": 2.3736,
      "step": 562
    },
    {
      "epoch": 0.2581384685923888,
      "grad_norm": 0.7050552456087895,
      "learning_rate": 0.0008698280432814107,
      "loss": 2.4486,
      "step": 563
    },
    {
      "epoch": 0.2585969738651995,
      "grad_norm": 0.6998544873783986,
      "learning_rate": 0.000869327813991301,
      "loss": 2.423,
      "step": 564
    },
    {
      "epoch": 0.2590554791380101,
      "grad_norm": 0.7169090286873294,
      "learning_rate": 0.0008688267698262971,
      "loss": 2.424,
      "step": 565
    },
    {
      "epoch": 0.25951398441082074,
      "grad_norm": 0.6738839506005453,
      "learning_rate": 0.0008683249118918894,
      "loss": 2.4175,
      "step": 566
    },
    {
      "epoch": 0.25997248968363135,
      "grad_norm": 0.7235923215136861,
      "learning_rate": 0.0008678222412953637,
      "loss": 2.3795,
      "step": 567
    },
    {
      "epoch": 0.260430994956442,
      "grad_norm": 0.6608036049574,
      "learning_rate": 0.0008673187591457987,
      "loss": 2.5111,
      "step": 568
    },
    {
      "epoch": 0.2608895002292526,
      "grad_norm": 0.7187927502410876,
      "learning_rate": 0.0008668144665540639,
      "loss": 2.3964,
      "step": 569
    },
    {
      "epoch": 0.2613480055020633,
      "grad_norm": 0.7493518518674204,
      "learning_rate": 0.0008663093646328167,
      "loss": 2.442,
      "step": 570
    },
    {
      "epoch": 0.26180651077487394,
      "grad_norm": 0.7149411890956157,
      "learning_rate": 0.0008658034544965003,
      "loss": 2.2823,
      "step": 571
    },
    {
      "epoch": 0.26226501604768454,
      "grad_norm": 0.7065676219872954,
      "learning_rate": 0.0008652967372613412,
      "loss": 2.3764,
      "step": 572
    },
    {
      "epoch": 0.2627235213204952,
      "grad_norm": 0.6687081689318802,
      "learning_rate": 0.0008647892140453466,
      "loss": 2.3294,
      "step": 573
    },
    {
      "epoch": 0.2631820265933058,
      "grad_norm": 0.6411015894630689,
      "learning_rate": 0.0008642808859683021,
      "loss": 2.3083,
      "step": 574
    },
    {
      "epoch": 0.26364053186611647,
      "grad_norm": 0.7013016570010037,
      "learning_rate": 0.0008637717541517689,
      "loss": 2.4245,
      "step": 575
    },
    {
      "epoch": 0.2640990371389271,
      "grad_norm": 0.7061103925207847,
      "learning_rate": 0.0008632618197190816,
      "loss": 2.4492,
      "step": 576
    },
    {
      "epoch": 0.26455754241173773,
      "grad_norm": 0.7594050351400723,
      "learning_rate": 0.0008627510837953458,
      "loss": 2.4518,
      "step": 577
    },
    {
      "epoch": 0.2650160476845484,
      "grad_norm": 0.6870447837898601,
      "learning_rate": 0.0008622395475074355,
      "loss": 2.4305,
      "step": 578
    },
    {
      "epoch": 0.265474552957359,
      "grad_norm": 0.7453271841702812,
      "learning_rate": 0.0008617272119839903,
      "loss": 2.4051,
      "step": 579
    },
    {
      "epoch": 0.26593305823016966,
      "grad_norm": 0.7050345647120914,
      "learning_rate": 0.0008612140783554136,
      "loss": 2.4765,
      "step": 580
    },
    {
      "epoch": 0.26639156350298027,
      "grad_norm": 0.6846642051012368,
      "learning_rate": 0.0008607001477538696,
      "loss": 2.3481,
      "step": 581
    },
    {
      "epoch": 0.2668500687757909,
      "grad_norm": 0.6507905537838528,
      "learning_rate": 0.0008601854213132807,
      "loss": 2.4568,
      "step": 582
    },
    {
      "epoch": 0.26730857404860153,
      "grad_norm": 0.7081638431033862,
      "learning_rate": 0.0008596699001693256,
      "loss": 2.4093,
      "step": 583
    },
    {
      "epoch": 0.2677670793214122,
      "grad_norm": 0.6550671399104782,
      "learning_rate": 0.000859153585459436,
      "loss": 2.369,
      "step": 584
    },
    {
      "epoch": 0.26822558459422285,
      "grad_norm": 0.6279547418543117,
      "learning_rate": 0.0008586364783227949,
      "loss": 2.3937,
      "step": 585
    },
    {
      "epoch": 0.26868408986703346,
      "grad_norm": 0.6544982565551759,
      "learning_rate": 0.0008581185799003332,
      "loss": 2.4621,
      "step": 586
    },
    {
      "epoch": 0.2691425951398441,
      "grad_norm": 0.7029659463043978,
      "learning_rate": 0.0008575998913347283,
      "loss": 2.4883,
      "step": 587
    },
    {
      "epoch": 0.2696011004126547,
      "grad_norm": 0.6642768167791873,
      "learning_rate": 0.0008570804137704004,
      "loss": 2.3909,
      "step": 588
    },
    {
      "epoch": 0.2700596056854654,
      "grad_norm": 0.6830000370392051,
      "learning_rate": 0.0008565601483535108,
      "loss": 2.3487,
      "step": 589
    },
    {
      "epoch": 0.27051811095827605,
      "grad_norm": 0.6714550661063899,
      "learning_rate": 0.0008560390962319591,
      "loss": 2.3708,
      "step": 590
    },
    {
      "epoch": 0.27097661623108665,
      "grad_norm": 0.6837376861518651,
      "learning_rate": 0.0008555172585553804,
      "loss": 2.4684,
      "step": 591
    },
    {
      "epoch": 0.2714351215038973,
      "grad_norm": 0.6624525015910365,
      "learning_rate": 0.0008549946364751435,
      "loss": 2.3425,
      "step": 592
    },
    {
      "epoch": 0.2718936267767079,
      "grad_norm": 0.6646674597778074,
      "learning_rate": 0.0008544712311443475,
      "loss": 2.407,
      "step": 593
    },
    {
      "epoch": 0.2723521320495186,
      "grad_norm": 0.704134268859165,
      "learning_rate": 0.0008539470437178196,
      "loss": 2.3471,
      "step": 594
    },
    {
      "epoch": 0.2728106373223292,
      "grad_norm": 0.6444402396092366,
      "learning_rate": 0.000853422075352113,
      "loss": 2.3878,
      "step": 595
    },
    {
      "epoch": 0.27326914259513985,
      "grad_norm": 0.7428487867603225,
      "learning_rate": 0.0008528963272055035,
      "loss": 2.4159,
      "step": 596
    },
    {
      "epoch": 0.2737276478679505,
      "grad_norm": 0.6491829097294289,
      "learning_rate": 0.0008523698004379877,
      "loss": 2.3634,
      "step": 597
    },
    {
      "epoch": 0.2741861531407611,
      "grad_norm": 0.7142451263267439,
      "learning_rate": 0.00085184249621128,
      "loss": 2.3089,
      "step": 598
    },
    {
      "epoch": 0.2746446584135718,
      "grad_norm": 0.686488793778934,
      "learning_rate": 0.0008513144156888101,
      "loss": 2.407,
      "step": 599
    },
    {
      "epoch": 0.2751031636863824,
      "grad_norm": 0.7756562487776282,
      "learning_rate": 0.0008507855600357207,
      "loss": 2.4999,
      "step": 600
    },
    {
      "epoch": 0.27556166895919304,
      "grad_norm": 0.7118407214207481,
      "learning_rate": 0.0008502559304188644,
      "loss": 2.3415,
      "step": 601
    },
    {
      "epoch": 0.27602017423200365,
      "grad_norm": 0.7308869090142528,
      "learning_rate": 0.0008497255280068019,
      "loss": 2.4508,
      "step": 602
    },
    {
      "epoch": 0.2764786795048143,
      "grad_norm": 0.6951525167085816,
      "learning_rate": 0.0008491943539697986,
      "loss": 2.4027,
      "step": 603
    },
    {
      "epoch": 0.27693718477762497,
      "grad_norm": 0.6989240992860168,
      "learning_rate": 0.0008486624094798226,
      "loss": 2.3426,
      "step": 604
    },
    {
      "epoch": 0.2773956900504356,
      "grad_norm": 0.6954543583126918,
      "learning_rate": 0.0008481296957105417,
      "loss": 2.4189,
      "step": 605
    },
    {
      "epoch": 0.27785419532324623,
      "grad_norm": 0.6786443854367697,
      "learning_rate": 0.0008475962138373213,
      "loss": 2.4579,
      "step": 606
    },
    {
      "epoch": 0.27831270059605684,
      "grad_norm": 0.6598307413502833,
      "learning_rate": 0.0008470619650372211,
      "loss": 2.3706,
      "step": 607
    },
    {
      "epoch": 0.2787712058688675,
      "grad_norm": 0.7116897089515539,
      "learning_rate": 0.0008465269504889934,
      "loss": 2.3433,
      "step": 608
    },
    {
      "epoch": 0.2792297111416781,
      "grad_norm": 0.6620669608927654,
      "learning_rate": 0.0008459911713730799,
      "loss": 2.3996,
      "step": 609
    },
    {
      "epoch": 0.27968821641448877,
      "grad_norm": 0.6892479979418136,
      "learning_rate": 0.0008454546288716089,
      "loss": 2.3905,
      "step": 610
    },
    {
      "epoch": 0.2801467216872994,
      "grad_norm": 0.6621767006057627,
      "learning_rate": 0.0008449173241683935,
      "loss": 2.3337,
      "step": 611
    },
    {
      "epoch": 0.28060522696011003,
      "grad_norm": 0.6921971059282274,
      "learning_rate": 0.0008443792584489281,
      "loss": 2.5179,
      "step": 612
    },
    {
      "epoch": 0.2810637322329207,
      "grad_norm": 0.639478150070234,
      "learning_rate": 0.0008438404329003863,
      "loss": 2.3805,
      "step": 613
    },
    {
      "epoch": 0.2815222375057313,
      "grad_norm": 0.6320414680083345,
      "learning_rate": 0.0008433008487116183,
      "loss": 2.3349,
      "step": 614
    },
    {
      "epoch": 0.28198074277854196,
      "grad_norm": 0.7003797928374847,
      "learning_rate": 0.0008427605070731481,
      "loss": 2.289,
      "step": 615
    },
    {
      "epoch": 0.28243924805135256,
      "grad_norm": 0.6431713627768156,
      "learning_rate": 0.0008422194091771708,
      "loss": 2.3898,
      "step": 616
    },
    {
      "epoch": 0.2828977533241632,
      "grad_norm": 0.7174540093181484,
      "learning_rate": 0.0008416775562175503,
      "loss": 2.4483,
      "step": 617
    },
    {
      "epoch": 0.2833562585969739,
      "grad_norm": 0.6899408204116008,
      "learning_rate": 0.000841134949389816,
      "loss": 2.3544,
      "step": 618
    },
    {
      "epoch": 0.2838147638697845,
      "grad_norm": 0.7090108980371626,
      "learning_rate": 0.0008405915898911611,
      "loss": 2.4419,
      "step": 619
    },
    {
      "epoch": 0.28427326914259515,
      "grad_norm": 0.6511546671870055,
      "learning_rate": 0.0008400474789204396,
      "loss": 2.4053,
      "step": 620
    },
    {
      "epoch": 0.28473177441540576,
      "grad_norm": 0.6407306988456571,
      "learning_rate": 0.0008395026176781626,
      "loss": 2.3532,
      "step": 621
    },
    {
      "epoch": 0.2851902796882164,
      "grad_norm": 0.6575545425465443,
      "learning_rate": 0.0008389570073664976,
      "loss": 2.4083,
      "step": 622
    },
    {
      "epoch": 0.285648784961027,
      "grad_norm": 0.6559982408500689,
      "learning_rate": 0.0008384106491892642,
      "loss": 2.4627,
      "step": 623
    },
    {
      "epoch": 0.2861072902338377,
      "grad_norm": 0.6389334194887178,
      "learning_rate": 0.0008378635443519327,
      "loss": 2.3873,
      "step": 624
    },
    {
      "epoch": 0.28656579550664835,
      "grad_norm": 0.6712244037444548,
      "learning_rate": 0.0008373156940616199,
      "loss": 2.3855,
      "step": 625
    },
    {
      "epoch": 0.28702430077945895,
      "grad_norm": 0.6370667542507263,
      "learning_rate": 0.0008367670995270882,
      "loss": 2.4334,
      "step": 626
    },
    {
      "epoch": 0.2874828060522696,
      "grad_norm": 0.6428676449578736,
      "learning_rate": 0.0008362177619587416,
      "loss": 2.3792,
      "step": 627
    },
    {
      "epoch": 0.2879413113250802,
      "grad_norm": 0.6474494829232641,
      "learning_rate": 0.0008356676825686238,
      "loss": 2.3558,
      "step": 628
    },
    {
      "epoch": 0.2883998165978909,
      "grad_norm": 0.6424891178684233,
      "learning_rate": 0.0008351168625704147,
      "loss": 2.4703,
      "step": 629
    },
    {
      "epoch": 0.28885832187070154,
      "grad_norm": 0.6723057986618843,
      "learning_rate": 0.0008345653031794292,
      "loss": 2.4034,
      "step": 630
    },
    {
      "epoch": 0.28931682714351215,
      "grad_norm": 0.6765500669849119,
      "learning_rate": 0.0008340130056126125,
      "loss": 2.3388,
      "step": 631
    },
    {
      "epoch": 0.2897753324163228,
      "grad_norm": 0.6482716122050285,
      "learning_rate": 0.0008334599710885394,
      "loss": 2.3833,
      "step": 632
    },
    {
      "epoch": 0.2902338376891334,
      "grad_norm": 0.648362180668662,
      "learning_rate": 0.0008329062008274098,
      "loss": 2.2705,
      "step": 633
    },
    {
      "epoch": 0.2906923429619441,
      "grad_norm": 0.671030427331317,
      "learning_rate": 0.000832351696051048,
      "loss": 2.3897,
      "step": 634
    },
    {
      "epoch": 0.2911508482347547,
      "grad_norm": 0.6173689161240775,
      "learning_rate": 0.000831796457982898,
      "loss": 2.3831,
      "step": 635
    },
    {
      "epoch": 0.29160935350756534,
      "grad_norm": 0.6421398527284494,
      "learning_rate": 0.0008312404878480222,
      "loss": 2.38,
      "step": 636
    },
    {
      "epoch": 0.292067858780376,
      "grad_norm": 0.6749159927116796,
      "learning_rate": 0.0008306837868730979,
      "loss": 2.3921,
      "step": 637
    },
    {
      "epoch": 0.2925263640531866,
      "grad_norm": 0.6771759796307572,
      "learning_rate": 0.0008301263562864152,
      "loss": 2.4066,
      "step": 638
    },
    {
      "epoch": 0.29298486932599727,
      "grad_norm": 0.6979967289001868,
      "learning_rate": 0.0008295681973178737,
      "loss": 2.4238,
      "step": 639
    },
    {
      "epoch": 0.29344337459880787,
      "grad_norm": 0.6584409586410245,
      "learning_rate": 0.0008290093111989804,
      "loss": 2.3571,
      "step": 640
    },
    {
      "epoch": 0.29390187987161853,
      "grad_norm": 0.6815671207817457,
      "learning_rate": 0.0008284496991628465,
      "loss": 2.3408,
      "step": 641
    },
    {
      "epoch": 0.29436038514442914,
      "grad_norm": 0.6730209073550302,
      "learning_rate": 0.0008278893624441847,
      "loss": 2.4082,
      "step": 642
    },
    {
      "epoch": 0.2948188904172398,
      "grad_norm": 0.6645347562100841,
      "learning_rate": 0.000827328302279307,
      "loss": 2.3274,
      "step": 643
    },
    {
      "epoch": 0.29527739569005046,
      "grad_norm": 0.673582301789298,
      "learning_rate": 0.0008267665199061211,
      "loss": 2.329,
      "step": 644
    },
    {
      "epoch": 0.29573590096286106,
      "grad_norm": 0.6680061972676621,
      "learning_rate": 0.0008262040165641288,
      "loss": 2.4158,
      "step": 645
    },
    {
      "epoch": 0.2961944062356717,
      "grad_norm": 0.6518586602918668,
      "learning_rate": 0.0008256407934944219,
      "loss": 2.3885,
      "step": 646
    },
    {
      "epoch": 0.29665291150848233,
      "grad_norm": 0.6431001396028199,
      "learning_rate": 0.0008250768519396807,
      "loss": 2.3109,
      "step": 647
    },
    {
      "epoch": 0.297111416781293,
      "grad_norm": 0.6165036509989333,
      "learning_rate": 0.0008245121931441706,
      "loss": 2.4072,
      "step": 648
    },
    {
      "epoch": 0.2975699220541036,
      "grad_norm": 0.6459997784598861,
      "learning_rate": 0.0008239468183537393,
      "loss": 2.4614,
      "step": 649
    },
    {
      "epoch": 0.29802842732691426,
      "grad_norm": 0.6623176096749874,
      "learning_rate": 0.0008233807288158146,
      "loss": 2.3537,
      "step": 650
    },
    {
      "epoch": 0.2984869325997249,
      "grad_norm": 0.6633501028334986,
      "learning_rate": 0.0008228139257794012,
      "loss": 2.4477,
      "step": 651
    },
    {
      "epoch": 0.2989454378725355,
      "grad_norm": 0.651372485142859,
      "learning_rate": 0.0008222464104950778,
      "loss": 2.3924,
      "step": 652
    },
    {
      "epoch": 0.2994039431453462,
      "grad_norm": 0.6445833743667585,
      "learning_rate": 0.000821678184214995,
      "loss": 2.3066,
      "step": 653
    },
    {
      "epoch": 0.2998624484181568,
      "grad_norm": 0.6626186976732255,
      "learning_rate": 0.0008211092481928716,
      "loss": 2.3312,
      "step": 654
    },
    {
      "epoch": 0.30032095369096745,
      "grad_norm": 0.7137729559995755,
      "learning_rate": 0.0008205396036839927,
      "loss": 2.4173,
      "step": 655
    },
    {
      "epoch": 0.30077945896377806,
      "grad_norm": 0.6967243529133988,
      "learning_rate": 0.0008199692519452069,
      "loss": 2.4942,
      "step": 656
    },
    {
      "epoch": 0.3012379642365887,
      "grad_norm": 0.7625554935935642,
      "learning_rate": 0.0008193981942349224,
      "loss": 2.3469,
      "step": 657
    },
    {
      "epoch": 0.3016964695093994,
      "grad_norm": 0.6724647757546426,
      "learning_rate": 0.0008188264318131056,
      "loss": 2.3636,
      "step": 658
    },
    {
      "epoch": 0.30215497478221,
      "grad_norm": 0.6653314725618608,
      "learning_rate": 0.0008182539659412776,
      "loss": 2.3255,
      "step": 659
    },
    {
      "epoch": 0.30261348005502064,
      "grad_norm": 0.7335091544532973,
      "learning_rate": 0.0008176807978825118,
      "loss": 2.3388,
      "step": 660
    },
    {
      "epoch": 0.30307198532783125,
      "grad_norm": 0.6981335116890897,
      "learning_rate": 0.0008171069289014306,
      "loss": 2.3325,
      "step": 661
    },
    {
      "epoch": 0.3035304906006419,
      "grad_norm": 0.6975086620306725,
      "learning_rate": 0.0008165323602642028,
      "loss": 2.4786,
      "step": 662
    },
    {
      "epoch": 0.30398899587345257,
      "grad_norm": 0.6432067124972036,
      "learning_rate": 0.0008159570932385414,
      "loss": 2.3554,
      "step": 663
    },
    {
      "epoch": 0.3044475011462632,
      "grad_norm": 0.6537826952637319,
      "learning_rate": 0.0008153811290936999,
      "loss": 2.4406,
      "step": 664
    },
    {
      "epoch": 0.30490600641907384,
      "grad_norm": 0.6826212477665592,
      "learning_rate": 0.0008148044691004698,
      "loss": 2.3984,
      "step": 665
    },
    {
      "epoch": 0.30536451169188444,
      "grad_norm": 0.6502304271403614,
      "learning_rate": 0.0008142271145311783,
      "loss": 2.4084,
      "step": 666
    },
    {
      "epoch": 0.3058230169646951,
      "grad_norm": 0.7000319146094501,
      "learning_rate": 0.000813649066659685,
      "loss": 2.4282,
      "step": 667
    },
    {
      "epoch": 0.3062815222375057,
      "grad_norm": 0.6913230609467859,
      "learning_rate": 0.0008130703267613787,
      "loss": 2.3331,
      "step": 668
    },
    {
      "epoch": 0.30674002751031637,
      "grad_norm": 0.6451423703209835,
      "learning_rate": 0.0008124908961131759,
      "loss": 2.3968,
      "step": 669
    },
    {
      "epoch": 0.30719853278312703,
      "grad_norm": 0.650050526439228,
      "learning_rate": 0.0008119107759935163,
      "loss": 2.2974,
      "step": 670
    },
    {
      "epoch": 0.30765703805593764,
      "grad_norm": 0.6749620102417362,
      "learning_rate": 0.0008113299676823615,
      "loss": 2.3997,
      "step": 671
    },
    {
      "epoch": 0.3081155433287483,
      "grad_norm": 0.6854439942483782,
      "learning_rate": 0.0008107484724611911,
      "loss": 2.3467,
      "step": 672
    },
    {
      "epoch": 0.3085740486015589,
      "grad_norm": 0.645676405666334,
      "learning_rate": 0.0008101662916130006,
      "loss": 2.3965,
      "step": 673
    },
    {
      "epoch": 0.30903255387436956,
      "grad_norm": 0.6736580412030302,
      "learning_rate": 0.0008095834264222979,
      "loss": 2.3882,
      "step": 674
    },
    {
      "epoch": 0.30949105914718017,
      "grad_norm": 0.6477522765439169,
      "learning_rate": 0.0008089998781751009,
      "loss": 2.4736,
      "step": 675
    },
    {
      "epoch": 0.30994956441999083,
      "grad_norm": 0.6296313587886688,
      "learning_rate": 0.0008084156481589349,
      "loss": 2.4171,
      "step": 676
    },
    {
      "epoch": 0.3104080696928015,
      "grad_norm": 0.6676318223598938,
      "learning_rate": 0.0008078307376628291,
      "loss": 2.36,
      "step": 677
    },
    {
      "epoch": 0.3108665749656121,
      "grad_norm": 0.6488852070236015,
      "learning_rate": 0.0008072451479773143,
      "loss": 2.4362,
      "step": 678
    },
    {
      "epoch": 0.31132508023842276,
      "grad_norm": 0.6297016268097562,
      "learning_rate": 0.0008066588803944195,
      "loss": 2.3427,
      "step": 679
    },
    {
      "epoch": 0.31178358551123336,
      "grad_norm": 0.6922131835517655,
      "learning_rate": 0.0008060719362076697,
      "loss": 2.4336,
      "step": 680
    },
    {
      "epoch": 0.312242090784044,
      "grad_norm": 0.6568731532438257,
      "learning_rate": 0.0008054843167120826,
      "loss": 2.3524,
      "step": 681
    },
    {
      "epoch": 0.31270059605685463,
      "grad_norm": 0.6373740714007508,
      "learning_rate": 0.0008048960232041663,
      "loss": 2.4572,
      "step": 682
    },
    {
      "epoch": 0.3131591013296653,
      "grad_norm": 0.7056224812677364,
      "learning_rate": 0.0008043070569819153,
      "loss": 2.3662,
      "step": 683
    },
    {
      "epoch": 0.31361760660247595,
      "grad_norm": 0.6455461205579698,
      "learning_rate": 0.0008037174193448089,
      "loss": 2.392,
      "step": 684
    },
    {
      "epoch": 0.31407611187528656,
      "grad_norm": 0.6597364366832446,
      "learning_rate": 0.0008031271115938077,
      "loss": 2.3794,
      "step": 685
    },
    {
      "epoch": 0.3145346171480972,
      "grad_norm": 0.6902553639336897,
      "learning_rate": 0.0008025361350313505,
      "loss": 2.4078,
      "step": 686
    },
    {
      "epoch": 0.3149931224209078,
      "grad_norm": 0.6693334026698443,
      "learning_rate": 0.0008019444909613523,
      "loss": 2.3959,
      "step": 687
    },
    {
      "epoch": 0.3154516276937185,
      "grad_norm": 0.7076315803436822,
      "learning_rate": 0.0008013521806892003,
      "loss": 2.3012,
      "step": 688
    },
    {
      "epoch": 0.3159101329665291,
      "grad_norm": 0.673404241608987,
      "learning_rate": 0.000800759205521752,
      "loss": 2.4591,
      "step": 689
    },
    {
      "epoch": 0.31636863823933975,
      "grad_norm": 0.6433395416305439,
      "learning_rate": 0.0008001655667673318,
      "loss": 2.4945,
      "step": 690
    },
    {
      "epoch": 0.3168271435121504,
      "grad_norm": 0.6386059383327098,
      "learning_rate": 0.0007995712657357279,
      "loss": 2.387,
      "step": 691
    },
    {
      "epoch": 0.317285648784961,
      "grad_norm": 0.6466733942576057,
      "learning_rate": 0.0007989763037381904,
      "loss": 2.555,
      "step": 692
    },
    {
      "epoch": 0.3177441540577717,
      "grad_norm": 0.7151126860936355,
      "learning_rate": 0.0007983806820874271,
      "loss": 2.3404,
      "step": 693
    },
    {
      "epoch": 0.3182026593305823,
      "grad_norm": 0.6756870811924842,
      "learning_rate": 0.0007977844020976016,
      "loss": 2.3657,
      "step": 694
    },
    {
      "epoch": 0.31866116460339294,
      "grad_norm": 0.684751373121012,
      "learning_rate": 0.00079718746508433,
      "loss": 2.4242,
      "step": 695
    },
    {
      "epoch": 0.31911966987620355,
      "grad_norm": 0.6698152176825337,
      "learning_rate": 0.0007965898723646776,
      "loss": 2.3926,
      "step": 696
    },
    {
      "epoch": 0.3195781751490142,
      "grad_norm": 0.6399788639142374,
      "learning_rate": 0.0007959916252571573,
      "loss": 2.3673,
      "step": 697
    },
    {
      "epoch": 0.32003668042182487,
      "grad_norm": 0.6240920625518914,
      "learning_rate": 0.000795392725081725,
      "loss": 2.286,
      "step": 698
    },
    {
      "epoch": 0.3204951856946355,
      "grad_norm": 0.6394219511674751,
      "learning_rate": 0.000794793173159778,
      "loss": 2.4378,
      "step": 699
    },
    {
      "epoch": 0.32095369096744614,
      "grad_norm": 0.6589106464592671,
      "learning_rate": 0.0007941929708141513,
      "loss": 2.5007,
      "step": 700
    },
    {
      "epoch": 0.32141219624025674,
      "grad_norm": 0.6446596810143197,
      "learning_rate": 0.0007935921193691153,
      "loss": 2.378,
      "step": 701
    },
    {
      "epoch": 0.3218707015130674,
      "grad_norm": 0.7166314718667169,
      "learning_rate": 0.0007929906201503722,
      "loss": 2.3498,
      "step": 702
    },
    {
      "epoch": 0.32232920678587806,
      "grad_norm": 0.7043711887836188,
      "learning_rate": 0.0007923884744850536,
      "loss": 2.4417,
      "step": 703
    },
    {
      "epoch": 0.32278771205868867,
      "grad_norm": 0.6840107911408047,
      "learning_rate": 0.0007917856837017176,
      "loss": 2.3358,
      "step": 704
    },
    {
      "epoch": 0.32324621733149933,
      "grad_norm": 0.7323061915670914,
      "learning_rate": 0.0007911822491303452,
      "loss": 2.4225,
      "step": 705
    },
    {
      "epoch": 0.32370472260430994,
      "grad_norm": 0.6604482910144658,
      "learning_rate": 0.0007905781721023382,
      "loss": 2.409,
      "step": 706
    },
    {
      "epoch": 0.3241632278771206,
      "grad_norm": 0.691519804192517,
      "learning_rate": 0.000789973453950516,
      "loss": 2.2642,
      "step": 707
    },
    {
      "epoch": 0.3246217331499312,
      "grad_norm": 0.6590444651738255,
      "learning_rate": 0.000789368096009112,
      "loss": 2.4089,
      "step": 708
    },
    {
      "epoch": 0.32508023842274186,
      "grad_norm": 0.6547845009541629,
      "learning_rate": 0.0007887620996137721,
      "loss": 2.3078,
      "step": 709
    },
    {
      "epoch": 0.3255387436955525,
      "grad_norm": 0.621555730062109,
      "learning_rate": 0.0007881554661015497,
      "loss": 2.2639,
      "step": 710
    },
    {
      "epoch": 0.32599724896836313,
      "grad_norm": 0.7168250519735687,
      "learning_rate": 0.0007875481968109051,
      "loss": 2.3816,
      "step": 711
    },
    {
      "epoch": 0.3264557542411738,
      "grad_norm": 0.6705462942396039,
      "learning_rate": 0.0007869402930817007,
      "loss": 2.4308,
      "step": 712
    },
    {
      "epoch": 0.3269142595139844,
      "grad_norm": 0.6830453848386918,
      "learning_rate": 0.0007863317562551987,
      "loss": 2.2725,
      "step": 713
    },
    {
      "epoch": 0.32737276478679506,
      "grad_norm": 0.6700246075883518,
      "learning_rate": 0.0007857225876740584,
      "loss": 2.3536,
      "step": 714
    },
    {
      "epoch": 0.32783127005960566,
      "grad_norm": 0.6641776633679097,
      "learning_rate": 0.0007851127886823327,
      "loss": 2.2645,
      "step": 715
    },
    {
      "epoch": 0.3282897753324163,
      "grad_norm": 0.6280589647315187,
      "learning_rate": 0.0007845023606254658,
      "loss": 2.3596,
      "step": 716
    },
    {
      "epoch": 0.328748280605227,
      "grad_norm": 0.685016347028729,
      "learning_rate": 0.0007838913048502894,
      "loss": 2.3147,
      "step": 717
    },
    {
      "epoch": 0.3292067858780376,
      "grad_norm": 0.7049733858744172,
      "learning_rate": 0.0007832796227050208,
      "loss": 2.5353,
      "step": 718
    },
    {
      "epoch": 0.32966529115084825,
      "grad_norm": 0.6710214757990399,
      "learning_rate": 0.0007826673155392587,
      "loss": 2.3852,
      "step": 719
    },
    {
      "epoch": 0.33012379642365886,
      "grad_norm": 0.6303947997012586,
      "learning_rate": 0.000782054384703981,
      "loss": 2.3184,
      "step": 720
    },
    {
      "epoch": 0.3305823016964695,
      "grad_norm": 0.6466604382724235,
      "learning_rate": 0.0007814408315515418,
      "loss": 2.3827,
      "step": 721
    },
    {
      "epoch": 0.3310408069692801,
      "grad_norm": 0.6489151349389152,
      "learning_rate": 0.0007808266574356683,
      "loss": 2.398,
      "step": 722
    },
    {
      "epoch": 0.3314993122420908,
      "grad_norm": 0.6591290325110795,
      "learning_rate": 0.0007802118637114573,
      "loss": 2.2873,
      "step": 723
    },
    {
      "epoch": 0.33195781751490144,
      "grad_norm": 0.6896786392701241,
      "learning_rate": 0.0007795964517353734,
      "loss": 2.5334,
      "step": 724
    },
    {
      "epoch": 0.33241632278771205,
      "grad_norm": 0.6828983011830053,
      "learning_rate": 0.0007789804228652449,
      "loss": 2.3545,
      "step": 725
    },
    {
      "epoch": 0.3328748280605227,
      "grad_norm": 0.6314230990068216,
      "learning_rate": 0.0007783637784602609,
      "loss": 2.3193,
      "step": 726
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.6215433627864178,
      "learning_rate": 0.0007777465198809692,
      "loss": 2.3871,
      "step": 727
    },
    {
      "epoch": 0.333791838606144,
      "grad_norm": 0.6821487260210978,
      "learning_rate": 0.0007771286484892722,
      "loss": 2.4792,
      "step": 728
    },
    {
      "epoch": 0.3342503438789546,
      "grad_norm": 0.6525204487547543,
      "learning_rate": 0.000776510165648425,
      "loss": 2.4222,
      "step": 729
    },
    {
      "epoch": 0.33470884915176524,
      "grad_norm": 0.6365661804888308,
      "learning_rate": 0.0007758910727230311,
      "loss": 2.3235,
      "step": 730
    },
    {
      "epoch": 0.3351673544245759,
      "grad_norm": 0.6579906280849055,
      "learning_rate": 0.0007752713710790404,
      "loss": 2.3619,
      "step": 731
    },
    {
      "epoch": 0.3356258596973865,
      "grad_norm": 0.6508148206345792,
      "learning_rate": 0.0007746510620837459,
      "loss": 2.3937,
      "step": 732
    },
    {
      "epoch": 0.33608436497019717,
      "grad_norm": 0.6867628944086733,
      "learning_rate": 0.0007740301471057807,
      "loss": 2.3524,
      "step": 733
    },
    {
      "epoch": 0.3365428702430078,
      "grad_norm": 0.6562134917160997,
      "learning_rate": 0.0007734086275151146,
      "loss": 2.3488,
      "step": 734
    },
    {
      "epoch": 0.33700137551581844,
      "grad_norm": 0.6393049806774187,
      "learning_rate": 0.0007727865046830517,
      "loss": 2.239,
      "step": 735
    },
    {
      "epoch": 0.3374598807886291,
      "grad_norm": 0.6161369224114513,
      "learning_rate": 0.0007721637799822269,
      "loss": 2.3852,
      "step": 736
    },
    {
      "epoch": 0.3379183860614397,
      "grad_norm": 0.6359427829815597,
      "learning_rate": 0.0007715404547866032,
      "loss": 2.3457,
      "step": 737
    },
    {
      "epoch": 0.33837689133425036,
      "grad_norm": 0.6432446063265741,
      "learning_rate": 0.0007709165304714685,
      "loss": 2.4442,
      "step": 738
    },
    {
      "epoch": 0.33883539660706097,
      "grad_norm": 0.648772438009759,
      "learning_rate": 0.0007702920084134324,
      "loss": 2.3896,
      "step": 739
    },
    {
      "epoch": 0.33929390187987163,
      "grad_norm": 0.6291734324751849,
      "learning_rate": 0.0007696668899904236,
      "loss": 2.29,
      "step": 740
    },
    {
      "epoch": 0.33975240715268223,
      "grad_norm": 0.6801660783747242,
      "learning_rate": 0.0007690411765816864,
      "loss": 2.3593,
      "step": 741
    },
    {
      "epoch": 0.3402109124254929,
      "grad_norm": 0.6403050519140824,
      "learning_rate": 0.0007684148695677778,
      "loss": 2.2652,
      "step": 742
    },
    {
      "epoch": 0.34066941769830356,
      "grad_norm": 0.6493766932563223,
      "learning_rate": 0.000767787970330565,
      "loss": 2.3191,
      "step": 743
    },
    {
      "epoch": 0.34112792297111416,
      "grad_norm": 0.6202377822057346,
      "learning_rate": 0.000767160480253221,
      "loss": 2.4669,
      "step": 744
    },
    {
      "epoch": 0.3415864282439248,
      "grad_norm": 0.6500070017062994,
      "learning_rate": 0.0007665324007202235,
      "loss": 2.2831,
      "step": 745
    },
    {
      "epoch": 0.3420449335167354,
      "grad_norm": 0.6842664291244414,
      "learning_rate": 0.0007659037331173498,
      "loss": 2.3373,
      "step": 746
    },
    {
      "epoch": 0.3425034387895461,
      "grad_norm": 0.6137797705502577,
      "learning_rate": 0.0007652744788316752,
      "loss": 2.28,
      "step": 747
    },
    {
      "epoch": 0.3429619440623567,
      "grad_norm": 0.6881895948483729,
      "learning_rate": 0.0007646446392515692,
      "loss": 2.3751,
      "step": 748
    },
    {
      "epoch": 0.34342044933516735,
      "grad_norm": 0.7034392311191875,
      "learning_rate": 0.000764014215766693,
      "loss": 2.3946,
      "step": 749
    },
    {
      "epoch": 0.343878954607978,
      "grad_norm": 0.7135664349804635,
      "learning_rate": 0.0007633832097679958,
      "loss": 2.3642,
      "step": 750
    },
    {
      "epoch": 0.3443374598807886,
      "grad_norm": 0.6955472686781569,
      "learning_rate": 0.0007627516226477122,
      "loss": 2.3725,
      "step": 751
    },
    {
      "epoch": 0.3447959651535993,
      "grad_norm": 0.6393847279541861,
      "learning_rate": 0.0007621194557993589,
      "loss": 2.4441,
      "step": 752
    },
    {
      "epoch": 0.3452544704264099,
      "grad_norm": 0.6369002602671016,
      "learning_rate": 0.0007614867106177319,
      "loss": 2.3613,
      "step": 753
    },
    {
      "epoch": 0.34571297569922055,
      "grad_norm": 0.6677473430058697,
      "learning_rate": 0.0007608533884989029,
      "loss": 2.408,
      "step": 754
    },
    {
      "epoch": 0.34617148097203115,
      "grad_norm": 0.6053366640531258,
      "learning_rate": 0.0007602194908402166,
      "loss": 2.3204,
      "step": 755
    },
    {
      "epoch": 0.3466299862448418,
      "grad_norm": 0.6147632205439276,
      "learning_rate": 0.0007595850190402877,
      "loss": 2.4102,
      "step": 756
    },
    {
      "epoch": 0.3470884915176525,
      "grad_norm": 0.658778142971114,
      "learning_rate": 0.0007589499744989976,
      "loss": 2.4871,
      "step": 757
    },
    {
      "epoch": 0.3475469967904631,
      "grad_norm": 0.6626182855927587,
      "learning_rate": 0.0007583143586174916,
      "loss": 2.3591,
      "step": 758
    },
    {
      "epoch": 0.34800550206327374,
      "grad_norm": 0.6449074631127794,
      "learning_rate": 0.000757678172798175,
      "loss": 2.3676,
      "step": 759
    },
    {
      "epoch": 0.34846400733608435,
      "grad_norm": 0.6501933572731315,
      "learning_rate": 0.0007570414184447112,
      "loss": 2.363,
      "step": 760
    },
    {
      "epoch": 0.348922512608895,
      "grad_norm": 0.6677451854966651,
      "learning_rate": 0.0007564040969620179,
      "loss": 2.3861,
      "step": 761
    },
    {
      "epoch": 0.3493810178817056,
      "grad_norm": 0.6233232765686981,
      "learning_rate": 0.0007557662097562636,
      "loss": 2.2327,
      "step": 762
    },
    {
      "epoch": 0.3498395231545163,
      "grad_norm": 0.6757412297224522,
      "learning_rate": 0.0007551277582348658,
      "loss": 2.4377,
      "step": 763
    },
    {
      "epoch": 0.35029802842732694,
      "grad_norm": 0.6275978671640042,
      "learning_rate": 0.0007544887438064862,
      "loss": 2.3278,
      "step": 764
    },
    {
      "epoch": 0.35075653370013754,
      "grad_norm": 0.6367533888181831,
      "learning_rate": 0.0007538491678810294,
      "loss": 2.233,
      "step": 765
    },
    {
      "epoch": 0.3512150389729482,
      "grad_norm": 0.6301592926992213,
      "learning_rate": 0.0007532090318696381,
      "loss": 2.3763,
      "step": 766
    },
    {
      "epoch": 0.3516735442457588,
      "grad_norm": 0.6364811523629422,
      "learning_rate": 0.0007525683371846913,
      "loss": 2.3939,
      "step": 767
    },
    {
      "epoch": 0.35213204951856947,
      "grad_norm": 0.6723362475111835,
      "learning_rate": 0.0007519270852398001,
      "loss": 2.4063,
      "step": 768
    },
    {
      "epoch": 0.35259055479138013,
      "grad_norm": 0.6094043426888228,
      "learning_rate": 0.000751285277449806,
      "loss": 2.3512,
      "step": 769
    },
    {
      "epoch": 0.35304906006419073,
      "grad_norm": 0.6711258302033779,
      "learning_rate": 0.0007506429152307756,
      "loss": 2.4044,
      "step": 770
    },
    {
      "epoch": 0.3535075653370014,
      "grad_norm": 0.656553822133271,
      "learning_rate": 0.00075,
      "loss": 2.3039,
      "step": 771
    },
    {
      "epoch": 0.353966070609812,
      "grad_norm": 0.6045490190609208,
      "learning_rate": 0.00074935653317599,
      "loss": 2.3695,
      "step": 772
    },
    {
      "epoch": 0.35442457588262266,
      "grad_norm": 0.6437961203008622,
      "learning_rate": 0.000748712516178473,
      "loss": 2.3577,
      "step": 773
    },
    {
      "epoch": 0.35488308115543327,
      "grad_norm": 0.646211321985262,
      "learning_rate": 0.0007480679504283911,
      "loss": 2.4003,
      "step": 774
    },
    {
      "epoch": 0.3553415864282439,
      "grad_norm": 0.6507395054354446,
      "learning_rate": 0.0007474228373478964,
      "loss": 2.3546,
      "step": 775
    },
    {
      "epoch": 0.3558000917010546,
      "grad_norm": 0.6620287423245884,
      "learning_rate": 0.0007467771783603492,
      "loss": 2.2855,
      "step": 776
    },
    {
      "epoch": 0.3562585969738652,
      "grad_norm": 0.5980096063927894,
      "learning_rate": 0.0007461309748903138,
      "loss": 2.3947,
      "step": 777
    },
    {
      "epoch": 0.35671710224667585,
      "grad_norm": 0.6155024646410623,
      "learning_rate": 0.0007454842283635562,
      "loss": 2.3588,
      "step": 778
    },
    {
      "epoch": 0.35717560751948646,
      "grad_norm": 0.6070770828109051,
      "learning_rate": 0.0007448369402070404,
      "loss": 2.2204,
      "step": 779
    },
    {
      "epoch": 0.3576341127922971,
      "grad_norm": 0.6574879253881969,
      "learning_rate": 0.0007441891118489254,
      "loss": 2.3952,
      "step": 780
    },
    {
      "epoch": 0.3580926180651077,
      "grad_norm": 0.6367168478593614,
      "learning_rate": 0.0007435407447185622,
      "loss": 2.3834,
      "step": 781
    },
    {
      "epoch": 0.3585511233379184,
      "grad_norm": 0.662597755210394,
      "learning_rate": 0.0007428918402464908,
      "loss": 2.3519,
      "step": 782
    },
    {
      "epoch": 0.35900962861072905,
      "grad_norm": 0.6237281386417163,
      "learning_rate": 0.0007422423998644359,
      "loss": 2.2769,
      "step": 783
    },
    {
      "epoch": 0.35946813388353965,
      "grad_norm": 0.5732701262171609,
      "learning_rate": 0.0007415924250053055,
      "loss": 2.3458,
      "step": 784
    },
    {
      "epoch": 0.3599266391563503,
      "grad_norm": 0.6485266476989069,
      "learning_rate": 0.0007409419171031865,
      "loss": 2.3588,
      "step": 785
    },
    {
      "epoch": 0.3603851444291609,
      "grad_norm": 0.6171655660440835,
      "learning_rate": 0.0007402908775933419,
      "loss": 2.2856,
      "step": 786
    },
    {
      "epoch": 0.3608436497019716,
      "grad_norm": 0.6644740198733657,
      "learning_rate": 0.0007396393079122077,
      "loss": 2.2359,
      "step": 787
    },
    {
      "epoch": 0.3613021549747822,
      "grad_norm": 0.6355449601127262,
      "learning_rate": 0.0007389872094973896,
      "loss": 2.3096,
      "step": 788
    },
    {
      "epoch": 0.36176066024759285,
      "grad_norm": 0.6184404364674315,
      "learning_rate": 0.00073833458378766,
      "loss": 2.3467,
      "step": 789
    },
    {
      "epoch": 0.3622191655204035,
      "grad_norm": 0.6830570218338579,
      "learning_rate": 0.0007376814322229544,
      "loss": 2.4438,
      "step": 790
    },
    {
      "epoch": 0.3626776707932141,
      "grad_norm": 0.6755774167236357,
      "learning_rate": 0.0007370277562443688,
      "loss": 2.2948,
      "step": 791
    },
    {
      "epoch": 0.3631361760660248,
      "grad_norm": 0.6783445380414667,
      "learning_rate": 0.0007363735572941564,
      "loss": 2.4091,
      "step": 792
    },
    {
      "epoch": 0.3635946813388354,
      "grad_norm": 0.6619933363906827,
      "learning_rate": 0.0007357188368157236,
      "loss": 2.4711,
      "step": 793
    },
    {
      "epoch": 0.36405318661164604,
      "grad_norm": 0.6712288434634871,
      "learning_rate": 0.0007350635962536284,
      "loss": 2.3187,
      "step": 794
    },
    {
      "epoch": 0.36451169188445665,
      "grad_norm": 0.6208818970488832,
      "learning_rate": 0.0007344078370535756,
      "loss": 2.2471,
      "step": 795
    },
    {
      "epoch": 0.3649701971572673,
      "grad_norm": 0.6054734054894049,
      "learning_rate": 0.0007337515606624148,
      "loss": 2.3983,
      "step": 796
    },
    {
      "epoch": 0.36542870243007797,
      "grad_norm": 0.6112779288775971,
      "learning_rate": 0.0007330947685281362,
      "loss": 2.2504,
      "step": 797
    },
    {
      "epoch": 0.3658872077028886,
      "grad_norm": 0.6486142802448305,
      "learning_rate": 0.0007324374620998682,
      "loss": 2.3774,
      "step": 798
    },
    {
      "epoch": 0.36634571297569923,
      "grad_norm": 0.6808080417892981,
      "learning_rate": 0.000731779642827874,
      "loss": 2.2808,
      "step": 799
    },
    {
      "epoch": 0.36680421824850984,
      "grad_norm": 0.658390690408349,
      "learning_rate": 0.0007311213121635483,
      "loss": 2.3084,
      "step": 800
    },
    {
      "epoch": 0.3672627235213205,
      "grad_norm": 0.6988432411341595,
      "learning_rate": 0.0007304624715594139,
      "loss": 2.3647,
      "step": 801
    },
    {
      "epoch": 0.3677212287941311,
      "grad_norm": 0.6717822275422306,
      "learning_rate": 0.0007298031224691193,
      "loss": 2.3739,
      "step": 802
    },
    {
      "epoch": 0.36817973406694177,
      "grad_norm": 0.6297840289993685,
      "learning_rate": 0.0007291432663474339,
      "loss": 2.3098,
      "step": 803
    },
    {
      "epoch": 0.3686382393397524,
      "grad_norm": 0.635431933618601,
      "learning_rate": 0.0007284829046502467,
      "loss": 2.3484,
      "step": 804
    },
    {
      "epoch": 0.36909674461256303,
      "grad_norm": 0.6235320614683441,
      "learning_rate": 0.0007278220388345619,
      "loss": 2.3198,
      "step": 805
    },
    {
      "epoch": 0.3695552498853737,
      "grad_norm": 0.6166198386123846,
      "learning_rate": 0.0007271606703584958,
      "loss": 2.3454,
      "step": 806
    },
    {
      "epoch": 0.3700137551581843,
      "grad_norm": 0.6090418843800038,
      "learning_rate": 0.000726498800681274,
      "loss": 2.3012,
      "step": 807
    },
    {
      "epoch": 0.37047226043099496,
      "grad_norm": 0.6809776439748013,
      "learning_rate": 0.0007258364312632279,
      "loss": 2.3332,
      "step": 808
    },
    {
      "epoch": 0.3709307657038056,
      "grad_norm": 0.6330196212194108,
      "learning_rate": 0.0007251735635657915,
      "loss": 2.3334,
      "step": 809
    },
    {
      "epoch": 0.3713892709766162,
      "grad_norm": 0.6324362367472803,
      "learning_rate": 0.000724510199051498,
      "loss": 2.3408,
      "step": 810
    },
    {
      "epoch": 0.3718477762494269,
      "grad_norm": 0.6850046301334559,
      "learning_rate": 0.0007238463391839769,
      "loss": 2.3551,
      "step": 811
    },
    {
      "epoch": 0.3723062815222375,
      "grad_norm": 0.6213557269828245,
      "learning_rate": 0.0007231819854279508,
      "loss": 2.4149,
      "step": 812
    },
    {
      "epoch": 0.37276478679504815,
      "grad_norm": 0.6539084908477354,
      "learning_rate": 0.0007225171392492316,
      "loss": 2.2542,
      "step": 813
    },
    {
      "epoch": 0.37322329206785876,
      "grad_norm": 0.7005333762688604,
      "learning_rate": 0.0007218518021147182,
      "loss": 2.3998,
      "step": 814
    },
    {
      "epoch": 0.3736817973406694,
      "grad_norm": 0.6688013738551166,
      "learning_rate": 0.0007211859754923923,
      "loss": 2.3551,
      "step": 815
    },
    {
      "epoch": 0.3741403026134801,
      "grad_norm": 0.6173452285363239,
      "learning_rate": 0.0007205196608513158,
      "loss": 2.3033,
      "step": 816
    },
    {
      "epoch": 0.3745988078862907,
      "grad_norm": 0.627454075472015,
      "learning_rate": 0.0007198528596616272,
      "loss": 2.4065,
      "step": 817
    },
    {
      "epoch": 0.37505731315910135,
      "grad_norm": 0.591191818221274,
      "learning_rate": 0.0007191855733945387,
      "loss": 2.276,
      "step": 818
    },
    {
      "epoch": 0.37551581843191195,
      "grad_norm": 0.6130003648943885,
      "learning_rate": 0.0007185178035223327,
      "loss": 2.3132,
      "step": 819
    },
    {
      "epoch": 0.3759743237047226,
      "grad_norm": 0.590302279598613,
      "learning_rate": 0.0007178495515183583,
      "loss": 2.3808,
      "step": 820
    },
    {
      "epoch": 0.3764328289775332,
      "grad_norm": 0.6204790154226676,
      "learning_rate": 0.000717180818857029,
      "loss": 2.3223,
      "step": 821
    },
    {
      "epoch": 0.3768913342503439,
      "grad_norm": 0.6386558992516045,
      "learning_rate": 0.0007165116070138182,
      "loss": 2.2343,
      "step": 822
    },
    {
      "epoch": 0.37734983952315454,
      "grad_norm": 0.6135311571871819,
      "learning_rate": 0.0007158419174652569,
      "loss": 2.2214,
      "step": 823
    },
    {
      "epoch": 0.37780834479596515,
      "grad_norm": 0.6554200836045881,
      "learning_rate": 0.00071517175168893,
      "loss": 2.3182,
      "step": 824
    },
    {
      "epoch": 0.3782668500687758,
      "grad_norm": 0.6314219910819017,
      "learning_rate": 0.0007145011111634732,
      "loss": 2.4148,
      "step": 825
    },
    {
      "epoch": 0.3787253553415864,
      "grad_norm": 0.6662139649250394,
      "learning_rate": 0.0007138299973685694,
      "loss": 2.4479,
      "step": 826
    },
    {
      "epoch": 0.3791838606143971,
      "grad_norm": 0.7372031010028899,
      "learning_rate": 0.0007131584117849459,
      "loss": 2.3444,
      "step": 827
    },
    {
      "epoch": 0.3796423658872077,
      "grad_norm": 0.6343095202191437,
      "learning_rate": 0.0007124863558943713,
      "loss": 2.3404,
      "step": 828
    },
    {
      "epoch": 0.38010087116001834,
      "grad_norm": 0.6271986320524525,
      "learning_rate": 0.0007118138311796514,
      "loss": 2.4345,
      "step": 829
    },
    {
      "epoch": 0.380559376432829,
      "grad_norm": 0.6553780024416881,
      "learning_rate": 0.0007111408391246262,
      "loss": 2.2853,
      "step": 830
    },
    {
      "epoch": 0.3810178817056396,
      "grad_norm": 0.6166912343659092,
      "learning_rate": 0.0007104673812141675,
      "loss": 2.4126,
      "step": 831
    },
    {
      "epoch": 0.38147638697845027,
      "grad_norm": 0.652118687179247,
      "learning_rate": 0.0007097934589341745,
      "loss": 2.3475,
      "step": 832
    },
    {
      "epoch": 0.38193489225126087,
      "grad_norm": 0.6279720583450172,
      "learning_rate": 0.0007091190737715711,
      "loss": 2.389,
      "step": 833
    },
    {
      "epoch": 0.38239339752407153,
      "grad_norm": 0.5819729247327663,
      "learning_rate": 0.0007084442272143026,
      "loss": 2.3599,
      "step": 834
    },
    {
      "epoch": 0.38285190279688214,
      "grad_norm": 0.6619160308415364,
      "learning_rate": 0.000707768920751332,
      "loss": 2.4204,
      "step": 835
    },
    {
      "epoch": 0.3833104080696928,
      "grad_norm": 0.6177788017609455,
      "learning_rate": 0.0007070931558726373,
      "loss": 2.3985,
      "step": 836
    },
    {
      "epoch": 0.38376891334250346,
      "grad_norm": 0.6255763999286063,
      "learning_rate": 0.0007064169340692076,
      "loss": 2.3194,
      "step": 837
    },
    {
      "epoch": 0.38422741861531406,
      "grad_norm": 0.6065673146791932,
      "learning_rate": 0.0007057402568330407,
      "loss": 2.2541,
      "step": 838
    },
    {
      "epoch": 0.3846859238881247,
      "grad_norm": 0.5997428967682963,
      "learning_rate": 0.0007050631256571389,
      "loss": 2.2655,
      "step": 839
    },
    {
      "epoch": 0.38514442916093533,
      "grad_norm": 0.6462517008378452,
      "learning_rate": 0.000704385542035506,
      "loss": 2.3134,
      "step": 840
    },
    {
      "epoch": 0.385602934433746,
      "grad_norm": 0.6231968293410592,
      "learning_rate": 0.000703707507463144,
      "loss": 2.2815,
      "step": 841
    },
    {
      "epoch": 0.38606143970655665,
      "grad_norm": 0.6484146075523692,
      "learning_rate": 0.0007030290234360505,
      "loss": 2.4547,
      "step": 842
    },
    {
      "epoch": 0.38651994497936726,
      "grad_norm": 0.6428272143081832,
      "learning_rate": 0.0007023500914512139,
      "loss": 2.3737,
      "step": 843
    },
    {
      "epoch": 0.3869784502521779,
      "grad_norm": 0.6007954360499461,
      "learning_rate": 0.0007016707130066116,
      "loss": 2.3024,
      "step": 844
    },
    {
      "epoch": 0.3874369555249885,
      "grad_norm": 0.652288770964336,
      "learning_rate": 0.0007009908896012055,
      "loss": 2.3039,
      "step": 845
    },
    {
      "epoch": 0.3878954607977992,
      "grad_norm": 0.6256840251248621,
      "learning_rate": 0.0007003106227349399,
      "loss": 2.4638,
      "step": 846
    },
    {
      "epoch": 0.3883539660706098,
      "grad_norm": 0.6744126054712565,
      "learning_rate": 0.000699629913908737,
      "loss": 2.3607,
      "step": 847
    },
    {
      "epoch": 0.38881247134342045,
      "grad_norm": 0.6341177243597828,
      "learning_rate": 0.0006989487646244943,
      "loss": 2.3677,
      "step": 848
    },
    {
      "epoch": 0.3892709766162311,
      "grad_norm": 0.6788097642398117,
      "learning_rate": 0.0006982671763850814,
      "loss": 2.3003,
      "step": 849
    },
    {
      "epoch": 0.3897294818890417,
      "grad_norm": 0.6348668449651265,
      "learning_rate": 0.0006975851506943359,
      "loss": 2.2969,
      "step": 850
    },
    {
      "epoch": 0.3901879871618524,
      "grad_norm": 0.7161490221024814,
      "learning_rate": 0.0006969026890570611,
      "loss": 2.3827,
      "step": 851
    },
    {
      "epoch": 0.390646492434663,
      "grad_norm": 0.6265029432926958,
      "learning_rate": 0.0006962197929790216,
      "loss": 2.2752,
      "step": 852
    },
    {
      "epoch": 0.39110499770747365,
      "grad_norm": 0.6057856735896819,
      "learning_rate": 0.0006955364639669409,
      "loss": 2.4431,
      "step": 853
    },
    {
      "epoch": 0.39156350298028425,
      "grad_norm": 0.6156420676234069,
      "learning_rate": 0.0006948527035284978,
      "loss": 2.3391,
      "step": 854
    },
    {
      "epoch": 0.3920220082530949,
      "grad_norm": 0.6115286563602929,
      "learning_rate": 0.0006941685131723225,
      "loss": 2.3689,
      "step": 855
    },
    {
      "epoch": 0.39248051352590557,
      "grad_norm": 0.6211734508694692,
      "learning_rate": 0.0006934838944079943,
      "loss": 2.364,
      "step": 856
    },
    {
      "epoch": 0.3929390187987162,
      "grad_norm": 0.6300254088348862,
      "learning_rate": 0.0006927988487460378,
      "loss": 2.3467,
      "step": 857
    },
    {
      "epoch": 0.39339752407152684,
      "grad_norm": 0.6129145702025217,
      "learning_rate": 0.0006921133776979186,
      "loss": 2.3511,
      "step": 858
    },
    {
      "epoch": 0.39385602934433744,
      "grad_norm": 0.6629276963770709,
      "learning_rate": 0.0006914274827760418,
      "loss": 2.2773,
      "step": 859
    },
    {
      "epoch": 0.3943145346171481,
      "grad_norm": 0.6605152076590137,
      "learning_rate": 0.0006907411654937475,
      "loss": 2.3767,
      "step": 860
    },
    {
      "epoch": 0.3947730398899587,
      "grad_norm": 0.658355884987859,
      "learning_rate": 0.0006900544273653075,
      "loss": 2.4332,
      "step": 861
    },
    {
      "epoch": 0.39523154516276937,
      "grad_norm": 0.6316671859152715,
      "learning_rate": 0.000689367269905922,
      "loss": 2.3662,
      "step": 862
    },
    {
      "epoch": 0.39569005043558003,
      "grad_norm": 0.647344414138551,
      "learning_rate": 0.0006886796946317168,
      "loss": 2.3287,
      "step": 863
    },
    {
      "epoch": 0.39614855570839064,
      "grad_norm": 0.6131513160483832,
      "learning_rate": 0.0006879917030597397,
      "loss": 2.3229,
      "step": 864
    },
    {
      "epoch": 0.3966070609812013,
      "grad_norm": 0.5957646695281302,
      "learning_rate": 0.0006873032967079561,
      "loss": 2.2602,
      "step": 865
    },
    {
      "epoch": 0.3970655662540119,
      "grad_norm": 0.5951965106631719,
      "learning_rate": 0.0006866144770952474,
      "loss": 2.3575,
      "step": 866
    },
    {
      "epoch": 0.39752407152682256,
      "grad_norm": 0.6184394960659765,
      "learning_rate": 0.0006859252457414067,
      "loss": 2.4153,
      "step": 867
    },
    {
      "epoch": 0.39798257679963317,
      "grad_norm": 0.6479119570948318,
      "learning_rate": 0.0006852356041671351,
      "loss": 2.241,
      "step": 868
    },
    {
      "epoch": 0.39844108207244383,
      "grad_norm": 0.6549970697191269,
      "learning_rate": 0.0006845455538940394,
      "loss": 2.391,
      "step": 869
    },
    {
      "epoch": 0.3988995873452545,
      "grad_norm": 0.6269384796799143,
      "learning_rate": 0.0006838550964446276,
      "loss": 2.303,
      "step": 870
    },
    {
      "epoch": 0.3993580926180651,
      "grad_norm": 0.6602760787597173,
      "learning_rate": 0.0006831642333423067,
      "loss": 2.419,
      "step": 871
    },
    {
      "epoch": 0.39981659789087576,
      "grad_norm": 0.7110401446017907,
      "learning_rate": 0.000682472966111378,
      "loss": 2.3121,
      "step": 872
    },
    {
      "epoch": 0.40027510316368636,
      "grad_norm": 0.6337512132469745,
      "learning_rate": 0.0006817812962770348,
      "loss": 2.3676,
      "step": 873
    },
    {
      "epoch": 0.400733608436497,
      "grad_norm": 0.6249053577015681,
      "learning_rate": 0.0006810892253653589,
      "loss": 2.3202,
      "step": 874
    },
    {
      "epoch": 0.40119211370930763,
      "grad_norm": 0.6387140190604292,
      "learning_rate": 0.0006803967549033167,
      "loss": 2.3414,
      "step": 875
    },
    {
      "epoch": 0.4016506189821183,
      "grad_norm": 0.619676582119536,
      "learning_rate": 0.0006797038864187564,
      "loss": 2.3628,
      "step": 876
    },
    {
      "epoch": 0.40210912425492895,
      "grad_norm": 0.6618349563697405,
      "learning_rate": 0.0006790106214404043,
      "loss": 2.4233,
      "step": 877
    },
    {
      "epoch": 0.40256762952773956,
      "grad_norm": 0.5712815384593491,
      "learning_rate": 0.0006783169614978614,
      "loss": 2.3403,
      "step": 878
    },
    {
      "epoch": 0.4030261348005502,
      "grad_norm": 0.6493101120156198,
      "learning_rate": 0.0006776229081216001,
      "loss": 2.3254,
      "step": 879
    },
    {
      "epoch": 0.4034846400733608,
      "grad_norm": 0.5760463871928262,
      "learning_rate": 0.0006769284628429611,
      "loss": 2.2833,
      "step": 880
    },
    {
      "epoch": 0.4039431453461715,
      "grad_norm": 0.6813019943601166,
      "learning_rate": 0.0006762336271941498,
      "loss": 2.355,
      "step": 881
    },
    {
      "epoch": 0.40440165061898214,
      "grad_norm": 0.6045595915881088,
      "learning_rate": 0.0006755384027082326,
      "loss": 2.3287,
      "step": 882
    },
    {
      "epoch": 0.40486015589179275,
      "grad_norm": 0.6357973639910502,
      "learning_rate": 0.0006748427909191342,
      "loss": 2.2681,
      "step": 883
    },
    {
      "epoch": 0.4053186611646034,
      "grad_norm": 0.6472096628499562,
      "learning_rate": 0.0006741467933616335,
      "loss": 2.3136,
      "step": 884
    },
    {
      "epoch": 0.405777166437414,
      "grad_norm": 0.6576223320741357,
      "learning_rate": 0.0006734504115713604,
      "loss": 2.3173,
      "step": 885
    },
    {
      "epoch": 0.4062356717102247,
      "grad_norm": 0.6463993498206633,
      "learning_rate": 0.0006727536470847932,
      "loss": 2.3289,
      "step": 886
    },
    {
      "epoch": 0.4066941769830353,
      "grad_norm": 0.6299025463184531,
      "learning_rate": 0.000672056501439254,
      "loss": 2.2807,
      "step": 887
    },
    {
      "epoch": 0.40715268225584594,
      "grad_norm": 0.629467530128995,
      "learning_rate": 0.0006713589761729063,
      "loss": 2.2951,
      "step": 888
    },
    {
      "epoch": 0.4076111875286566,
      "grad_norm": 0.6591426808071247,
      "learning_rate": 0.0006706610728247508,
      "loss": 2.3465,
      "step": 889
    },
    {
      "epoch": 0.4080696928014672,
      "grad_norm": 0.6944827940177551,
      "learning_rate": 0.0006699627929346227,
      "loss": 2.426,
      "step": 890
    },
    {
      "epoch": 0.40852819807427787,
      "grad_norm": 0.6299583565798872,
      "learning_rate": 0.0006692641380431879,
      "loss": 2.3226,
      "step": 891
    },
    {
      "epoch": 0.4089867033470885,
      "grad_norm": 0.6204461952056786,
      "learning_rate": 0.0006685651096919393,
      "loss": 2.3193,
      "step": 892
    },
    {
      "epoch": 0.40944520861989914,
      "grad_norm": 0.6680150483075629,
      "learning_rate": 0.0006678657094231944,
      "loss": 2.1314,
      "step": 893
    },
    {
      "epoch": 0.40990371389270974,
      "grad_norm": 0.5878891063667403,
      "learning_rate": 0.0006671659387800909,
      "loss": 2.4345,
      "step": 894
    },
    {
      "epoch": 0.4103622191655204,
      "grad_norm": 0.6091416063201586,
      "learning_rate": 0.000666465799306584,
      "loss": 2.3429,
      "step": 895
    },
    {
      "epoch": 0.41082072443833106,
      "grad_norm": 0.6093907162717401,
      "learning_rate": 0.0006657652925474423,
      "loss": 2.3043,
      "step": 896
    },
    {
      "epoch": 0.41127922971114167,
      "grad_norm": 0.6183667023103231,
      "learning_rate": 0.000665064420048245,
      "loss": 2.3856,
      "step": 897
    },
    {
      "epoch": 0.41173773498395233,
      "grad_norm": 0.6302571180726317,
      "learning_rate": 0.0006643631833553785,
      "loss": 2.2942,
      "step": 898
    },
    {
      "epoch": 0.41219624025676294,
      "grad_norm": 0.6027031206428978,
      "learning_rate": 0.000663661584016032,
      "loss": 2.287,
      "step": 899
    },
    {
      "epoch": 0.4126547455295736,
      "grad_norm": 0.6298320092738742,
      "learning_rate": 0.0006629596235781957,
      "loss": 2.4051,
      "step": 900
    },
    {
      "epoch": 0.4131132508023842,
      "grad_norm": 0.6319468538048978,
      "learning_rate": 0.0006622573035906556,
      "loss": 2.3218,
      "step": 901
    },
    {
      "epoch": 0.41357175607519486,
      "grad_norm": 0.6604327573794866,
      "learning_rate": 0.0006615546256029921,
      "loss": 2.3737,
      "step": 902
    },
    {
      "epoch": 0.4140302613480055,
      "grad_norm": 0.6260958003710608,
      "learning_rate": 0.0006608515911655743,
      "loss": 2.4781,
      "step": 903
    },
    {
      "epoch": 0.41448876662081613,
      "grad_norm": 0.6136078207068564,
      "learning_rate": 0.0006601482018295591,
      "loss": 2.38,
      "step": 904
    },
    {
      "epoch": 0.4149472718936268,
      "grad_norm": 0.5951679069332668,
      "learning_rate": 0.0006594444591468851,
      "loss": 2.3598,
      "step": 905
    },
    {
      "epoch": 0.4154057771664374,
      "grad_norm": 0.5847438099719648,
      "learning_rate": 0.0006587403646702713,
      "loss": 2.3576,
      "step": 906
    },
    {
      "epoch": 0.41586428243924806,
      "grad_norm": 0.6225081371916221,
      "learning_rate": 0.0006580359199532126,
      "loss": 2.2656,
      "step": 907
    },
    {
      "epoch": 0.41632278771205866,
      "grad_norm": 0.61039000733414,
      "learning_rate": 0.000657331126549977,
      "loss": 2.2625,
      "step": 908
    },
    {
      "epoch": 0.4167812929848693,
      "grad_norm": 0.6219703060200975,
      "learning_rate": 0.0006566259860156014,
      "loss": 2.3031,
      "step": 909
    },
    {
      "epoch": 0.41723979825768,
      "grad_norm": 0.6687221097732218,
      "learning_rate": 0.0006559204999058888,
      "loss": 2.3829,
      "step": 910
    },
    {
      "epoch": 0.4176983035304906,
      "grad_norm": 0.6612432419240112,
      "learning_rate": 0.0006552146697774049,
      "loss": 2.3807,
      "step": 911
    },
    {
      "epoch": 0.41815680880330125,
      "grad_norm": 0.6684367856532768,
      "learning_rate": 0.0006545084971874737,
      "loss": 2.3656,
      "step": 912
    },
    {
      "epoch": 0.41861531407611186,
      "grad_norm": 0.6511640039230993,
      "learning_rate": 0.0006538019836941758,
      "loss": 2.3276,
      "step": 913
    },
    {
      "epoch": 0.4190738193489225,
      "grad_norm": 0.652504973891663,
      "learning_rate": 0.0006530951308563431,
      "loss": 2.356,
      "step": 914
    },
    {
      "epoch": 0.4195323246217332,
      "grad_norm": 0.595426976089289,
      "learning_rate": 0.0006523879402335567,
      "loss": 2.3727,
      "step": 915
    },
    {
      "epoch": 0.4199908298945438,
      "grad_norm": 0.590061028318524,
      "learning_rate": 0.0006516804133861429,
      "loss": 2.2924,
      "step": 916
    },
    {
      "epoch": 0.42044933516735444,
      "grad_norm": 0.5874506482317866,
      "learning_rate": 0.0006509725518751698,
      "loss": 2.2755,
      "step": 917
    },
    {
      "epoch": 0.42090784044016505,
      "grad_norm": 0.654475355314294,
      "learning_rate": 0.0006502643572624438,
      "loss": 2.2996,
      "step": 918
    },
    {
      "epoch": 0.4213663457129757,
      "grad_norm": 0.6027938379579424,
      "learning_rate": 0.0006495558311105064,
      "loss": 2.2594,
      "step": 919
    },
    {
      "epoch": 0.4218248509857863,
      "grad_norm": 0.6144115003385453,
      "learning_rate": 0.0006488469749826305,
      "loss": 2.3211,
      "step": 920
    },
    {
      "epoch": 0.422283356258597,
      "grad_norm": 0.6236736972001795,
      "learning_rate": 0.000648137790442817,
      "loss": 2.3831,
      "step": 921
    },
    {
      "epoch": 0.42274186153140764,
      "grad_norm": 0.6582419947914452,
      "learning_rate": 0.0006474282790557916,
      "loss": 2.2704,
      "step": 922
    },
    {
      "epoch": 0.42320036680421824,
      "grad_norm": 0.6023436717659996,
      "learning_rate": 0.000646718442387001,
      "loss": 2.2874,
      "step": 923
    },
    {
      "epoch": 0.4236588720770289,
      "grad_norm": 0.6054938437084005,
      "learning_rate": 0.0006460082820026094,
      "loss": 2.2931,
      "step": 924
    },
    {
      "epoch": 0.4241173773498395,
      "grad_norm": 0.6135941729013356,
      "learning_rate": 0.0006452977994694959,
      "loss": 2.287,
      "step": 925
    },
    {
      "epoch": 0.42457588262265017,
      "grad_norm": 0.6206262990279261,
      "learning_rate": 0.0006445869963552496,
      "loss": 2.2883,
      "step": 926
    },
    {
      "epoch": 0.4250343878954608,
      "grad_norm": 0.6032705007890118,
      "learning_rate": 0.0006438758742281672,
      "loss": 2.3433,
      "step": 927
    },
    {
      "epoch": 0.42549289316827144,
      "grad_norm": 0.600580277018893,
      "learning_rate": 0.0006431644346572495,
      "loss": 2.3455,
      "step": 928
    },
    {
      "epoch": 0.4259513984410821,
      "grad_norm": 0.6278791584798313,
      "learning_rate": 0.0006424526792121974,
      "loss": 2.2128,
      "step": 929
    },
    {
      "epoch": 0.4264099037138927,
      "grad_norm": 0.6058862631000579,
      "learning_rate": 0.0006417406094634089,
      "loss": 2.4312,
      "step": 930
    },
    {
      "epoch": 0.42686840898670336,
      "grad_norm": 0.634520240221422,
      "learning_rate": 0.0006410282269819756,
      "loss": 2.3178,
      "step": 931
    },
    {
      "epoch": 0.42732691425951397,
      "grad_norm": 0.6414879212156869,
      "learning_rate": 0.0006403155333396787,
      "loss": 2.4011,
      "step": 932
    },
    {
      "epoch": 0.42778541953232463,
      "grad_norm": 0.6280127975253766,
      "learning_rate": 0.0006396025301089863,
      "loss": 2.3573,
      "step": 933
    },
    {
      "epoch": 0.42824392480513523,
      "grad_norm": 0.6297185668439333,
      "learning_rate": 0.0006388892188630493,
      "loss": 2.2458,
      "step": 934
    },
    {
      "epoch": 0.4287024300779459,
      "grad_norm": 0.6200622429122868,
      "learning_rate": 0.0006381756011756982,
      "loss": 2.3164,
      "step": 935
    },
    {
      "epoch": 0.42916093535075656,
      "grad_norm": 0.5767471751406218,
      "learning_rate": 0.0006374616786214403,
      "loss": 2.3881,
      "step": 936
    },
    {
      "epoch": 0.42961944062356716,
      "grad_norm": 0.5910268596419224,
      "learning_rate": 0.0006367474527754544,
      "loss": 2.3185,
      "step": 937
    },
    {
      "epoch": 0.4300779458963778,
      "grad_norm": 0.6177808743812254,
      "learning_rate": 0.0006360329252135894,
      "loss": 2.3483,
      "step": 938
    },
    {
      "epoch": 0.4305364511691884,
      "grad_norm": 0.6364988274222995,
      "learning_rate": 0.0006353180975123595,
      "loss": 2.3007,
      "step": 939
    },
    {
      "epoch": 0.4309949564419991,
      "grad_norm": 0.6321523257840606,
      "learning_rate": 0.0006346029712489413,
      "loss": 2.3402,
      "step": 940
    },
    {
      "epoch": 0.4314534617148097,
      "grad_norm": 0.6057839101165555,
      "learning_rate": 0.0006338875480011698,
      "loss": 2.4028,
      "step": 941
    },
    {
      "epoch": 0.43191196698762035,
      "grad_norm": 0.6235093394362694,
      "learning_rate": 0.0006331718293475357,
      "loss": 2.2031,
      "step": 942
    },
    {
      "epoch": 0.432370472260431,
      "grad_norm": 0.6121353249999769,
      "learning_rate": 0.0006324558168671811,
      "loss": 2.2914,
      "step": 943
    },
    {
      "epoch": 0.4328289775332416,
      "grad_norm": 0.6258197944167787,
      "learning_rate": 0.0006317395121398968,
      "loss": 2.382,
      "step": 944
    },
    {
      "epoch": 0.4332874828060523,
      "grad_norm": 0.6219274877729708,
      "learning_rate": 0.0006310229167461179,
      "loss": 2.3317,
      "step": 945
    },
    {
      "epoch": 0.4337459880788629,
      "grad_norm": 0.6592928424600399,
      "learning_rate": 0.0006303060322669214,
      "loss": 2.212,
      "step": 946
    },
    {
      "epoch": 0.43420449335167355,
      "grad_norm": 0.6702927080115592,
      "learning_rate": 0.0006295888602840214,
      "loss": 2.4652,
      "step": 947
    },
    {
      "epoch": 0.4346629986244842,
      "grad_norm": 0.6409490672633943,
      "learning_rate": 0.0006288714023797671,
      "loss": 2.3238,
      "step": 948
    },
    {
      "epoch": 0.4351215038972948,
      "grad_norm": 0.6469806610367961,
      "learning_rate": 0.000628153660137138,
      "loss": 2.4414,
      "step": 949
    },
    {
      "epoch": 0.4355800091701055,
      "grad_norm": 0.6160322429363382,
      "learning_rate": 0.0006274356351397413,
      "loss": 2.329,
      "step": 950
    },
    {
      "epoch": 0.4360385144429161,
      "grad_norm": 0.5669860296250903,
      "learning_rate": 0.0006267173289718079,
      "loss": 2.2481,
      "step": 951
    },
    {
      "epoch": 0.43649701971572674,
      "grad_norm": 0.6543796346406598,
      "learning_rate": 0.000625998743218189,
      "loss": 2.3803,
      "step": 952
    },
    {
      "epoch": 0.43695552498853735,
      "grad_norm": 0.6172871807187701,
      "learning_rate": 0.000625279879464353,
      "loss": 2.2929,
      "step": 953
    },
    {
      "epoch": 0.437414030261348,
      "grad_norm": 0.6638795115376179,
      "learning_rate": 0.000624560739296381,
      "loss": 2.3444,
      "step": 954
    },
    {
      "epoch": 0.43787253553415867,
      "grad_norm": 0.6383310613596909,
      "learning_rate": 0.0006238413243009648,
      "loss": 2.2231,
      "step": 955
    },
    {
      "epoch": 0.4383310408069693,
      "grad_norm": 0.6773816792835566,
      "learning_rate": 0.000623121636065402,
      "loss": 2.3413,
      "step": 956
    },
    {
      "epoch": 0.43878954607977994,
      "grad_norm": 0.6308804453052632,
      "learning_rate": 0.0006224016761775933,
      "loss": 2.3067,
      "step": 957
    },
    {
      "epoch": 0.43924805135259054,
      "grad_norm": 0.6683139179348452,
      "learning_rate": 0.0006216814462260386,
      "loss": 2.362,
      "step": 958
    },
    {
      "epoch": 0.4397065566254012,
      "grad_norm": 0.7008450124472028,
      "learning_rate": 0.0006209609477998338,
      "loss": 2.3302,
      "step": 959
    },
    {
      "epoch": 0.4401650618982118,
      "grad_norm": 0.6243222926046714,
      "learning_rate": 0.0006202401824886674,
      "loss": 2.3378,
      "step": 960
    },
    {
      "epoch": 0.44062356717102247,
      "grad_norm": 0.6207425467328255,
      "learning_rate": 0.0006195191518828162,
      "loss": 2.3315,
      "step": 961
    },
    {
      "epoch": 0.44108207244383313,
      "grad_norm": 0.6112026852003823,
      "learning_rate": 0.0006187978575731427,
      "loss": 2.3789,
      "step": 962
    },
    {
      "epoch": 0.44154057771664373,
      "grad_norm": 0.6119244275876462,
      "learning_rate": 0.0006180763011510911,
      "loss": 2.3627,
      "step": 963
    },
    {
      "epoch": 0.4419990829894544,
      "grad_norm": 0.6155406184511094,
      "learning_rate": 0.000617354484208684,
      "loss": 2.241,
      "step": 964
    },
    {
      "epoch": 0.442457588262265,
      "grad_norm": 0.6209182160854342,
      "learning_rate": 0.0006166324083385189,
      "loss": 2.307,
      "step": 965
    },
    {
      "epoch": 0.44291609353507566,
      "grad_norm": 0.6423841598299586,
      "learning_rate": 0.0006159100751337642,
      "loss": 2.3321,
      "step": 966
    },
    {
      "epoch": 0.44337459880788627,
      "grad_norm": 0.6528215292907604,
      "learning_rate": 0.0006151874861881565,
      "loss": 2.3795,
      "step": 967
    },
    {
      "epoch": 0.4438331040806969,
      "grad_norm": 0.6337161188863024,
      "learning_rate": 0.0006144646430959964,
      "loss": 2.298,
      "step": 968
    },
    {
      "epoch": 0.4442916093535076,
      "grad_norm": 0.6080259951946323,
      "learning_rate": 0.0006137415474521454,
      "loss": 2.2436,
      "step": 969
    },
    {
      "epoch": 0.4447501146263182,
      "grad_norm": 0.6252050995973073,
      "learning_rate": 0.0006130182008520222,
      "loss": 2.3164,
      "step": 970
    },
    {
      "epoch": 0.44520861989912885,
      "grad_norm": 0.5749357660825368,
      "learning_rate": 0.000612294604891599,
      "loss": 2.3111,
      "step": 971
    },
    {
      "epoch": 0.44566712517193946,
      "grad_norm": 0.6123462438360728,
      "learning_rate": 0.0006115707611673986,
      "loss": 2.2868,
      "step": 972
    },
    {
      "epoch": 0.4461256304447501,
      "grad_norm": 0.6119946925866714,
      "learning_rate": 0.0006108466712764902,
      "loss": 2.4374,
      "step": 973
    },
    {
      "epoch": 0.4465841357175607,
      "grad_norm": 0.603154357572585,
      "learning_rate": 0.0006101223368164858,
      "loss": 2.3712,
      "step": 974
    },
    {
      "epoch": 0.4470426409903714,
      "grad_norm": 0.5744985597944355,
      "learning_rate": 0.0006093977593855375,
      "loss": 2.4346,
      "step": 975
    },
    {
      "epoch": 0.44750114626318205,
      "grad_norm": 0.6190778260185676,
      "learning_rate": 0.0006086729405823335,
      "loss": 2.3174,
      "step": 976
    },
    {
      "epoch": 0.44795965153599265,
      "grad_norm": 0.6199362988013117,
      "learning_rate": 0.0006079478820060943,
      "loss": 2.3202,
      "step": 977
    },
    {
      "epoch": 0.4484181568088033,
      "grad_norm": 0.5796708770022393,
      "learning_rate": 0.0006072225852565695,
      "loss": 2.3159,
      "step": 978
    },
    {
      "epoch": 0.4488766620816139,
      "grad_norm": 0.6240334866316054,
      "learning_rate": 0.0006064970519340341,
      "loss": 2.31,
      "step": 979
    },
    {
      "epoch": 0.4493351673544246,
      "grad_norm": 0.6394352878082159,
      "learning_rate": 0.0006057712836392856,
      "loss": 2.3547,
      "step": 980
    },
    {
      "epoch": 0.4497936726272352,
      "grad_norm": 0.6181467749338779,
      "learning_rate": 0.0006050452819736389,
      "loss": 2.2915,
      "step": 981
    },
    {
      "epoch": 0.45025217790004585,
      "grad_norm": 0.667123425543737,
      "learning_rate": 0.000604319048538925,
      "loss": 2.3257,
      "step": 982
    },
    {
      "epoch": 0.4507106831728565,
      "grad_norm": 0.6305340509922304,
      "learning_rate": 0.0006035925849374855,
      "loss": 2.3259,
      "step": 983
    },
    {
      "epoch": 0.4511691884456671,
      "grad_norm": 0.5938665066308477,
      "learning_rate": 0.0006028658927721697,
      "loss": 2.3255,
      "step": 984
    },
    {
      "epoch": 0.4516276937184778,
      "grad_norm": 0.6176704190232546,
      "learning_rate": 0.0006021389736463321,
      "loss": 2.2898,
      "step": 985
    },
    {
      "epoch": 0.4520861989912884,
      "grad_norm": 0.5854511576625192,
      "learning_rate": 0.0006014118291638271,
      "loss": 2.2851,
      "step": 986
    },
    {
      "epoch": 0.45254470426409904,
      "grad_norm": 0.6262509873616877,
      "learning_rate": 0.0006006844609290065,
      "loss": 2.3717,
      "step": 987
    },
    {
      "epoch": 0.4530032095369097,
      "grad_norm": 0.6110415648273994,
      "learning_rate": 0.0005999568705467161,
      "loss": 2.3559,
      "step": 988
    },
    {
      "epoch": 0.4534617148097203,
      "grad_norm": 0.5504003775787095,
      "learning_rate": 0.0005992290596222915,
      "loss": 2.2723,
      "step": 989
    },
    {
      "epoch": 0.45392022008253097,
      "grad_norm": 0.5961516679100018,
      "learning_rate": 0.0005985010297615551,
      "loss": 2.223,
      "step": 990
    },
    {
      "epoch": 0.4543787253553416,
      "grad_norm": 0.6266551508257007,
      "learning_rate": 0.0005977727825708123,
      "loss": 2.3919,
      "step": 991
    },
    {
      "epoch": 0.45483723062815223,
      "grad_norm": 0.6223910267498735,
      "learning_rate": 0.0005970443196568478,
      "loss": 2.3773,
      "step": 992
    },
    {
      "epoch": 0.45529573590096284,
      "grad_norm": 0.7358165154590999,
      "learning_rate": 0.0005963156426269227,
      "loss": 2.3633,
      "step": 993
    },
    {
      "epoch": 0.4557542411737735,
      "grad_norm": 0.6160533207248625,
      "learning_rate": 0.0005955867530887702,
      "loss": 2.2881,
      "step": 994
    },
    {
      "epoch": 0.45621274644658416,
      "grad_norm": 0.6132494643350832,
      "learning_rate": 0.0005948576526505923,
      "loss": 2.3236,
      "step": 995
    },
    {
      "epoch": 0.45667125171939477,
      "grad_norm": 0.5893207538244488,
      "learning_rate": 0.0005941283429210568,
      "loss": 2.2033,
      "step": 996
    },
    {
      "epoch": 0.4571297569922054,
      "grad_norm": 0.6426722479718633,
      "learning_rate": 0.0005933988255092926,
      "loss": 2.3543,
      "step": 997
    },
    {
      "epoch": 0.45758826226501603,
      "grad_norm": 0.5985027238198615,
      "learning_rate": 0.0005926691020248874,
      "loss": 2.3387,
      "step": 998
    },
    {
      "epoch": 0.4580467675378267,
      "grad_norm": 0.6043601719545167,
      "learning_rate": 0.0005919391740778833,
      "loss": 2.2609,
      "step": 999
    },
    {
      "epoch": 0.4585052728106373,
      "grad_norm": 0.6054721052515216,
      "learning_rate": 0.0005912090432787736,
      "loss": 2.3489,
      "step": 1000
    },
    {
      "epoch": 0.45896377808344796,
      "grad_norm": 0.5936416777937311,
      "learning_rate": 0.000590478711238499,
      "loss": 2.3174,
      "step": 1001
    },
    {
      "epoch": 0.4594222833562586,
      "grad_norm": 0.5950173423595508,
      "learning_rate": 0.0005897481795684446,
      "loss": 2.303,
      "step": 1002
    },
    {
      "epoch": 0.4598807886290692,
      "grad_norm": 0.5947261590185035,
      "learning_rate": 0.0005890174498804355,
      "loss": 2.3054,
      "step": 1003
    },
    {
      "epoch": 0.4603392939018799,
      "grad_norm": 0.5813015507566608,
      "learning_rate": 0.0005882865237867339,
      "loss": 2.2595,
      "step": 1004
    },
    {
      "epoch": 0.4607977991746905,
      "grad_norm": 0.6137041130174968,
      "learning_rate": 0.0005875554029000353,
      "loss": 2.3201,
      "step": 1005
    },
    {
      "epoch": 0.46125630444750115,
      "grad_norm": 0.592576117654005,
      "learning_rate": 0.0005868240888334653,
      "loss": 2.3223,
      "step": 1006
    },
    {
      "epoch": 0.46171480972031176,
      "grad_norm": 0.6406655009844748,
      "learning_rate": 0.0005860925832005753,
      "loss": 2.3649,
      "step": 1007
    },
    {
      "epoch": 0.4621733149931224,
      "grad_norm": 0.6145076975461123,
      "learning_rate": 0.0005853608876153395,
      "loss": 2.3272,
      "step": 1008
    },
    {
      "epoch": 0.4626318202659331,
      "grad_norm": 0.6086828629806686,
      "learning_rate": 0.0005846290036921512,
      "loss": 2.2817,
      "step": 1009
    },
    {
      "epoch": 0.4630903255387437,
      "grad_norm": 0.6317181832170704,
      "learning_rate": 0.0005838969330458195,
      "loss": 2.288,
      "step": 1010
    },
    {
      "epoch": 0.46354883081155435,
      "grad_norm": 0.6322323277619806,
      "learning_rate": 0.0005831646772915651,
      "loss": 2.3009,
      "step": 1011
    },
    {
      "epoch": 0.46400733608436495,
      "grad_norm": 0.6765387603405646,
      "learning_rate": 0.0005824322380450173,
      "loss": 2.32,
      "step": 1012
    },
    {
      "epoch": 0.4644658413571756,
      "grad_norm": 0.6198504650597182,
      "learning_rate": 0.0005816996169222102,
      "loss": 2.3021,
      "step": 1013
    },
    {
      "epoch": 0.4649243466299862,
      "grad_norm": 0.5928553193471388,
      "learning_rate": 0.0005809668155395793,
      "loss": 2.3624,
      "step": 1014
    },
    {
      "epoch": 0.4653828519027969,
      "grad_norm": 0.6290558477793418,
      "learning_rate": 0.0005802338355139578,
      "loss": 2.2964,
      "step": 1015
    },
    {
      "epoch": 0.46584135717560754,
      "grad_norm": 0.6208744510064005,
      "learning_rate": 0.0005795006784625728,
      "loss": 2.3519,
      "step": 1016
    },
    {
      "epoch": 0.46629986244841815,
      "grad_norm": 0.6203135773605256,
      "learning_rate": 0.0005787673460030423,
      "loss": 2.3411,
      "step": 1017
    },
    {
      "epoch": 0.4667583677212288,
      "grad_norm": 0.6012867746177658,
      "learning_rate": 0.000578033839753371,
      "loss": 2.4145,
      "step": 1018
    },
    {
      "epoch": 0.4672168729940394,
      "grad_norm": 0.6302125166305977,
      "learning_rate": 0.0005773001613319476,
      "loss": 2.2089,
      "step": 1019
    },
    {
      "epoch": 0.4676753782668501,
      "grad_norm": 0.5989666951047617,
      "learning_rate": 0.00057656631235754,
      "loss": 2.3551,
      "step": 1020
    },
    {
      "epoch": 0.46813388353966073,
      "grad_norm": 0.6073858813525945,
      "learning_rate": 0.0005758322944492929,
      "loss": 2.3297,
      "step": 1021
    },
    {
      "epoch": 0.46859238881247134,
      "grad_norm": 0.6229455022346706,
      "learning_rate": 0.0005750981092267237,
      "loss": 2.3142,
      "step": 1022
    },
    {
      "epoch": 0.469050894085282,
      "grad_norm": 0.630396800418263,
      "learning_rate": 0.0005743637583097183,
      "loss": 2.2357,
      "step": 1023
    },
    {
      "epoch": 0.4695093993580926,
      "grad_norm": 0.6317543863579402,
      "learning_rate": 0.0005736292433185291,
      "loss": 2.2709,
      "step": 1024
    },
    {
      "epoch": 0.46996790463090327,
      "grad_norm": 0.6142863749218508,
      "learning_rate": 0.0005728945658737699,
      "loss": 2.4314,
      "step": 1025
    },
    {
      "epoch": 0.47042640990371387,
      "grad_norm": 0.5774596119431913,
      "learning_rate": 0.0005721597275964133,
      "loss": 2.3339,
      "step": 1026
    },
    {
      "epoch": 0.47088491517652453,
      "grad_norm": 0.5963176461155033,
      "learning_rate": 0.0005714247301077865,
      "loss": 2.2999,
      "step": 1027
    },
    {
      "epoch": 0.4713434204493352,
      "grad_norm": 0.6000582164665067,
      "learning_rate": 0.0005706895750295682,
      "loss": 2.3848,
      "step": 1028
    },
    {
      "epoch": 0.4718019257221458,
      "grad_norm": 0.6008497222934881,
      "learning_rate": 0.0005699542639837844,
      "loss": 2.302,
      "step": 1029
    },
    {
      "epoch": 0.47226043099495646,
      "grad_norm": 0.5882344986263912,
      "learning_rate": 0.0005692187985928055,
      "loss": 2.3918,
      "step": 1030
    },
    {
      "epoch": 0.47271893626776706,
      "grad_norm": 0.6125074559729995,
      "learning_rate": 0.0005684831804793427,
      "loss": 2.5825,
      "step": 1031
    },
    {
      "epoch": 0.4731774415405777,
      "grad_norm": 0.5737510998039831,
      "learning_rate": 0.0005677474112664438,
      "loss": 2.2857,
      "step": 1032
    },
    {
      "epoch": 0.47363594681338833,
      "grad_norm": 0.6576329629342521,
      "learning_rate": 0.0005670114925774899,
      "loss": 2.362,
      "step": 1033
    },
    {
      "epoch": 0.474094452086199,
      "grad_norm": 0.6100741985960022,
      "learning_rate": 0.0005662754260361924,
      "loss": 2.3095,
      "step": 1034
    },
    {
      "epoch": 0.47455295735900965,
      "grad_norm": 0.6029398692061133,
      "learning_rate": 0.0005655392132665884,
      "loss": 2.3079,
      "step": 1035
    },
    {
      "epoch": 0.47501146263182026,
      "grad_norm": 0.6733352953253742,
      "learning_rate": 0.000564802855893038,
      "loss": 2.3482,
      "step": 1036
    },
    {
      "epoch": 0.4754699679046309,
      "grad_norm": 0.6023458074023248,
      "learning_rate": 0.0005640663555402198,
      "loss": 2.4017,
      "step": 1037
    },
    {
      "epoch": 0.4759284731774415,
      "grad_norm": 0.6080102788769914,
      "learning_rate": 0.0005633297138331285,
      "loss": 2.3986,
      "step": 1038
    },
    {
      "epoch": 0.4763869784502522,
      "grad_norm": 0.6075889063174179,
      "learning_rate": 0.0005625929323970705,
      "loss": 2.4912,
      "step": 1039
    },
    {
      "epoch": 0.4768454837230628,
      "grad_norm": 0.5974768627194069,
      "learning_rate": 0.0005618560128576603,
      "loss": 2.3504,
      "step": 1040
    },
    {
      "epoch": 0.47730398899587345,
      "grad_norm": 0.6026823861719102,
      "learning_rate": 0.0005611189568408173,
      "loss": 2.3647,
      "step": 1041
    },
    {
      "epoch": 0.4777624942686841,
      "grad_norm": 0.6052736252497934,
      "learning_rate": 0.0005603817659727619,
      "loss": 2.2816,
      "step": 1042
    },
    {
      "epoch": 0.4782209995414947,
      "grad_norm": 0.5764875562487081,
      "learning_rate": 0.0005596444418800121,
      "loss": 2.2832,
      "step": 1043
    },
    {
      "epoch": 0.4786795048143054,
      "grad_norm": 0.6052545836859812,
      "learning_rate": 0.0005589069861893798,
      "loss": 2.2882,
      "step": 1044
    },
    {
      "epoch": 0.479138010087116,
      "grad_norm": 0.6136952749712792,
      "learning_rate": 0.0005581694005279673,
      "loss": 2.2862,
      "step": 1045
    },
    {
      "epoch": 0.47959651535992665,
      "grad_norm": 0.6085343445894976,
      "learning_rate": 0.0005574316865231637,
      "loss": 2.2461,
      "step": 1046
    },
    {
      "epoch": 0.48005502063273725,
      "grad_norm": 0.6230723142195075,
      "learning_rate": 0.0005566938458026411,
      "loss": 2.3111,
      "step": 1047
    },
    {
      "epoch": 0.4805135259055479,
      "grad_norm": 0.6390099016895102,
      "learning_rate": 0.0005559558799943514,
      "loss": 2.2696,
      "step": 1048
    },
    {
      "epoch": 0.48097203117835857,
      "grad_norm": 0.6270926475231282,
      "learning_rate": 0.0005552177907265223,
      "loss": 2.138,
      "step": 1049
    },
    {
      "epoch": 0.4814305364511692,
      "grad_norm": 0.612042610820908,
      "learning_rate": 0.000554479579627654,
      "loss": 2.2608,
      "step": 1050
    },
    {
      "epoch": 0.48188904172397984,
      "grad_norm": 0.6082996016223167,
      "learning_rate": 0.0005537412483265157,
      "loss": 2.3246,
      "step": 1051
    },
    {
      "epoch": 0.48234754699679044,
      "grad_norm": 0.6284006177697281,
      "learning_rate": 0.0005530027984521413,
      "loss": 2.2848,
      "step": 1052
    },
    {
      "epoch": 0.4828060522696011,
      "grad_norm": 0.619960147485842,
      "learning_rate": 0.0005522642316338268,
      "loss": 2.2364,
      "step": 1053
    },
    {
      "epoch": 0.4832645575424117,
      "grad_norm": 0.6451254067603307,
      "learning_rate": 0.0005515255495011259,
      "loss": 2.3308,
      "step": 1054
    },
    {
      "epoch": 0.48372306281522237,
      "grad_norm": 0.6369479223040997,
      "learning_rate": 0.0005507867536838472,
      "loss": 2.2724,
      "step": 1055
    },
    {
      "epoch": 0.48418156808803303,
      "grad_norm": 0.6232538252327935,
      "learning_rate": 0.0005500478458120492,
      "loss": 2.2252,
      "step": 1056
    },
    {
      "epoch": 0.48464007336084364,
      "grad_norm": 0.6063697366838137,
      "learning_rate": 0.0005493088275160387,
      "loss": 2.3002,
      "step": 1057
    },
    {
      "epoch": 0.4850985786336543,
      "grad_norm": 0.609500842977591,
      "learning_rate": 0.0005485697004263657,
      "loss": 2.3379,
      "step": 1058
    },
    {
      "epoch": 0.4855570839064649,
      "grad_norm": 0.5740812025999293,
      "learning_rate": 0.0005478304661738199,
      "loss": 2.3716,
      "step": 1059
    },
    {
      "epoch": 0.48601558917927556,
      "grad_norm": 0.5742262389504017,
      "learning_rate": 0.0005470911263894279,
      "loss": 2.2851,
      "step": 1060
    },
    {
      "epoch": 0.4864740944520862,
      "grad_norm": 0.5697684403224503,
      "learning_rate": 0.0005463516827044491,
      "loss": 2.3387,
      "step": 1061
    },
    {
      "epoch": 0.48693259972489683,
      "grad_norm": 0.56634013600934,
      "learning_rate": 0.000545612136750372,
      "loss": 2.2495,
      "step": 1062
    },
    {
      "epoch": 0.4873911049977075,
      "grad_norm": 0.6251730775629422,
      "learning_rate": 0.0005448724901589107,
      "loss": 2.4779,
      "step": 1063
    },
    {
      "epoch": 0.4878496102705181,
      "grad_norm": 0.5837574301739205,
      "learning_rate": 0.0005441327445620014,
      "loss": 2.3293,
      "step": 1064
    },
    {
      "epoch": 0.48830811554332876,
      "grad_norm": 0.6103317341444731,
      "learning_rate": 0.0005433929015917988,
      "loss": 2.3137,
      "step": 1065
    },
    {
      "epoch": 0.48876662081613936,
      "grad_norm": 0.5839937097553319,
      "learning_rate": 0.0005426529628806724,
      "loss": 2.261,
      "step": 1066
    },
    {
      "epoch": 0.48922512608895,
      "grad_norm": 0.602589702439264,
      "learning_rate": 0.0005419129300612029,
      "loss": 2.3329,
      "step": 1067
    },
    {
      "epoch": 0.4896836313617607,
      "grad_norm": 0.6201928385448374,
      "learning_rate": 0.000541172804766179,
      "loss": 2.4002,
      "step": 1068
    },
    {
      "epoch": 0.4901421366345713,
      "grad_norm": 0.5871032428397418,
      "learning_rate": 0.0005404325886285927,
      "loss": 2.2676,
      "step": 1069
    },
    {
      "epoch": 0.49060064190738195,
      "grad_norm": 0.6034271052731943,
      "learning_rate": 0.000539692283281637,
      "loss": 2.3676,
      "step": 1070
    },
    {
      "epoch": 0.49105914718019256,
      "grad_norm": 0.6121517818763319,
      "learning_rate": 0.0005389518903587017,
      "loss": 2.3288,
      "step": 1071
    },
    {
      "epoch": 0.4915176524530032,
      "grad_norm": 0.5877899863692208,
      "learning_rate": 0.0005382114114933695,
      "loss": 2.2509,
      "step": 1072
    },
    {
      "epoch": 0.4919761577258138,
      "grad_norm": 0.5989377302069921,
      "learning_rate": 0.0005374708483194132,
      "loss": 2.267,
      "step": 1073
    },
    {
      "epoch": 0.4924346629986245,
      "grad_norm": 0.6526333388662431,
      "learning_rate": 0.000536730202470791,
      "loss": 2.2903,
      "step": 1074
    },
    {
      "epoch": 0.49289316827143514,
      "grad_norm": 0.6403923873958121,
      "learning_rate": 0.0005359894755816443,
      "loss": 2.2654,
      "step": 1075
    },
    {
      "epoch": 0.49335167354424575,
      "grad_norm": 0.6029082091793515,
      "learning_rate": 0.0005352486692862926,
      "loss": 2.3134,
      "step": 1076
    },
    {
      "epoch": 0.4938101788170564,
      "grad_norm": 0.6387197782494327,
      "learning_rate": 0.0005345077852192307,
      "loss": 2.373,
      "step": 1077
    },
    {
      "epoch": 0.494268684089867,
      "grad_norm": 0.592567100240428,
      "learning_rate": 0.0005337668250151254,
      "loss": 2.2593,
      "step": 1078
    },
    {
      "epoch": 0.4947271893626777,
      "grad_norm": 0.6040111819721832,
      "learning_rate": 0.0005330257903088111,
      "loss": 2.2838,
      "step": 1079
    },
    {
      "epoch": 0.4951856946354883,
      "grad_norm": 0.6102745954145249,
      "learning_rate": 0.000532284682735287,
      "loss": 2.2131,
      "step": 1080
    },
    {
      "epoch": 0.49564419990829894,
      "grad_norm": 0.6004367482672044,
      "learning_rate": 0.0005315435039297124,
      "loss": 2.3403,
      "step": 1081
    },
    {
      "epoch": 0.4961027051811096,
      "grad_norm": 0.602255754140507,
      "learning_rate": 0.0005308022555274046,
      "loss": 2.2673,
      "step": 1082
    },
    {
      "epoch": 0.4965612104539202,
      "grad_norm": 0.5608849523065977,
      "learning_rate": 0.0005300609391638336,
      "loss": 2.2225,
      "step": 1083
    },
    {
      "epoch": 0.49701971572673087,
      "grad_norm": 0.5978026741987942,
      "learning_rate": 0.0005293195564746201,
      "loss": 2.2769,
      "step": 1084
    },
    {
      "epoch": 0.4974782209995415,
      "grad_norm": 0.6254846601061261,
      "learning_rate": 0.0005285781090955304,
      "loss": 2.3082,
      "step": 1085
    },
    {
      "epoch": 0.49793672627235214,
      "grad_norm": 0.5864662031711123,
      "learning_rate": 0.0005278365986624743,
      "loss": 2.2442,
      "step": 1086
    },
    {
      "epoch": 0.49839523154516274,
      "grad_norm": 0.5976520583276143,
      "learning_rate": 0.0005270950268115001,
      "loss": 2.2128,
      "step": 1087
    },
    {
      "epoch": 0.4988537368179734,
      "grad_norm": 0.5936226343985168,
      "learning_rate": 0.0005263533951787919,
      "loss": 2.3499,
      "step": 1088
    },
    {
      "epoch": 0.49931224209078406,
      "grad_norm": 0.5994495182541442,
      "learning_rate": 0.000525611705400666,
      "loss": 2.33,
      "step": 1089
    },
    {
      "epoch": 0.49977074736359467,
      "grad_norm": 0.5661782011666728,
      "learning_rate": 0.0005248699591135664,
      "loss": 2.3162,
      "step": 1090
    },
    {
      "epoch": 0.5002292526364053,
      "grad_norm": 0.6077718901137957,
      "learning_rate": 0.0005241281579540618,
      "loss": 2.2536,
      "step": 1091
    },
    {
      "epoch": 0.5006877579092159,
      "grad_norm": 0.6356976651168149,
      "learning_rate": 0.0005233863035588427,
      "loss": 2.3972,
      "step": 1092
    },
    {
      "epoch": 0.5011462631820266,
      "grad_norm": 0.6244793441536806,
      "learning_rate": 0.0005226443975647161,
      "loss": 2.3887,
      "step": 1093
    },
    {
      "epoch": 0.5016047684548373,
      "grad_norm": 0.6043475259909191,
      "learning_rate": 0.0005219024416086036,
      "loss": 2.2761,
      "step": 1094
    },
    {
      "epoch": 0.5020632737276479,
      "grad_norm": 0.5650110689662258,
      "learning_rate": 0.0005211604373275366,
      "loss": 2.3249,
      "step": 1095
    },
    {
      "epoch": 0.5025217790004585,
      "grad_norm": 0.6051775641178089,
      "learning_rate": 0.0005204183863586533,
      "loss": 2.2314,
      "step": 1096
    },
    {
      "epoch": 0.5029802842732691,
      "grad_norm": 0.6262410827019866,
      "learning_rate": 0.0005196762903391951,
      "loss": 2.3339,
      "step": 1097
    },
    {
      "epoch": 0.5034387895460798,
      "grad_norm": 0.5877150255707011,
      "learning_rate": 0.0005189341509065023,
      "loss": 2.2931,
      "step": 1098
    },
    {
      "epoch": 0.5038972948188905,
      "grad_norm": 0.5972870679590834,
      "learning_rate": 0.0005181919696980112,
      "loss": 2.316,
      "step": 1099
    },
    {
      "epoch": 0.504355800091701,
      "grad_norm": 0.6051600054560172,
      "learning_rate": 0.0005174497483512506,
      "loss": 2.3424,
      "step": 1100
    },
    {
      "epoch": 0.5048143053645117,
      "grad_norm": 0.6375472742555353,
      "learning_rate": 0.0005167074885038374,
      "loss": 2.22,
      "step": 1101
    },
    {
      "epoch": 0.5052728106373223,
      "grad_norm": 0.5750312108619777,
      "learning_rate": 0.0005159651917934735,
      "loss": 2.2515,
      "step": 1102
    },
    {
      "epoch": 0.505731315910133,
      "grad_norm": 0.6332884551009029,
      "learning_rate": 0.0005152228598579428,
      "loss": 2.3686,
      "step": 1103
    },
    {
      "epoch": 0.5061898211829436,
      "grad_norm": 0.6257637039269199,
      "learning_rate": 0.000514480494335106,
      "loss": 2.3049,
      "step": 1104
    },
    {
      "epoch": 0.5066483264557542,
      "grad_norm": 0.5828375454394569,
      "learning_rate": 0.0005137380968628983,
      "loss": 2.29,
      "step": 1105
    },
    {
      "epoch": 0.5071068317285649,
      "grad_norm": 0.6420777848239064,
      "learning_rate": 0.0005129956690793255,
      "loss": 2.2577,
      "step": 1106
    },
    {
      "epoch": 0.5075653370013755,
      "grad_norm": 0.5741245615639173,
      "learning_rate": 0.0005122532126224601,
      "loss": 2.3296,
      "step": 1107
    },
    {
      "epoch": 0.5080238422741862,
      "grad_norm": 0.6190597288790846,
      "learning_rate": 0.0005115107291304378,
      "loss": 2.3412,
      "step": 1108
    },
    {
      "epoch": 0.5084823475469968,
      "grad_norm": 0.6112704445970172,
      "learning_rate": 0.0005107682202414544,
      "loss": 2.3196,
      "step": 1109
    },
    {
      "epoch": 0.5089408528198074,
      "grad_norm": 0.5911329201490587,
      "learning_rate": 0.0005100256875937613,
      "loss": 2.2446,
      "step": 1110
    },
    {
      "epoch": 0.509399358092618,
      "grad_norm": 0.6380941056266427,
      "learning_rate": 0.0005092831328256625,
      "loss": 2.3576,
      "step": 1111
    },
    {
      "epoch": 0.5098578633654287,
      "grad_norm": 0.6106303403296834,
      "learning_rate": 0.0005085405575755105,
      "loss": 2.2955,
      "step": 1112
    },
    {
      "epoch": 0.5103163686382394,
      "grad_norm": 0.617999645233152,
      "learning_rate": 0.0005077979634817034,
      "loss": 2.2501,
      "step": 1113
    },
    {
      "epoch": 0.51077487391105,
      "grad_norm": 0.5928488798559758,
      "learning_rate": 0.0005070553521826808,
      "loss": 2.2918,
      "step": 1114
    },
    {
      "epoch": 0.5112333791838606,
      "grad_norm": 0.6008780244551577,
      "learning_rate": 0.00050631272531692,
      "loss": 2.3379,
      "step": 1115
    },
    {
      "epoch": 0.5116918844566712,
      "grad_norm": 0.5999412870256408,
      "learning_rate": 0.0005055700845229327,
      "loss": 2.2828,
      "step": 1116
    },
    {
      "epoch": 0.5121503897294819,
      "grad_norm": 0.6231903863631356,
      "learning_rate": 0.000504827431439262,
      "loss": 2.3821,
      "step": 1117
    },
    {
      "epoch": 0.5126088950022926,
      "grad_norm": 0.5738309718176878,
      "learning_rate": 0.000504084767704477,
      "loss": 2.3342,
      "step": 1118
    },
    {
      "epoch": 0.5130674002751031,
      "grad_norm": 0.6299818200009084,
      "learning_rate": 0.0005033420949571712,
      "loss": 2.2749,
      "step": 1119
    },
    {
      "epoch": 0.5135259055479138,
      "grad_norm": 0.5992963264733782,
      "learning_rate": 0.0005025994148359574,
      "loss": 2.2907,
      "step": 1120
    },
    {
      "epoch": 0.5139844108207244,
      "grad_norm": 0.6709166786592415,
      "learning_rate": 0.0005018567289794651,
      "loss": 2.2325,
      "step": 1121
    },
    {
      "epoch": 0.5144429160935351,
      "grad_norm": 0.6716611006306039,
      "learning_rate": 0.0005011140390263362,
      "loss": 2.3602,
      "step": 1122
    },
    {
      "epoch": 0.5149014213663458,
      "grad_norm": 0.6152432719696187,
      "learning_rate": 0.0005003713466152218,
      "loss": 2.271,
      "step": 1123
    },
    {
      "epoch": 0.5153599266391563,
      "grad_norm": 0.5872139818479444,
      "learning_rate": 0.0004996286533847783,
      "loss": 2.3272,
      "step": 1124
    },
    {
      "epoch": 0.515818431911967,
      "grad_norm": 0.6288456140206096,
      "learning_rate": 0.000498885960973664,
      "loss": 2.3037,
      "step": 1125
    },
    {
      "epoch": 0.5162769371847776,
      "grad_norm": 0.6171972378320196,
      "learning_rate": 0.000498143271020535,
      "loss": 2.2328,
      "step": 1126
    },
    {
      "epoch": 0.5167354424575883,
      "grad_norm": 0.6109367448307881,
      "learning_rate": 0.0004974005851640428,
      "loss": 2.3694,
      "step": 1127
    },
    {
      "epoch": 0.517193947730399,
      "grad_norm": 0.6055482650177088,
      "learning_rate": 0.000496657905042829,
      "loss": 2.3129,
      "step": 1128
    },
    {
      "epoch": 0.5176524530032095,
      "grad_norm": 0.6304175941844458,
      "learning_rate": 0.0004959152322955232,
      "loss": 2.253,
      "step": 1129
    },
    {
      "epoch": 0.5181109582760202,
      "grad_norm": 0.621016086618194,
      "learning_rate": 0.0004951725685607382,
      "loss": 2.3719,
      "step": 1130
    },
    {
      "epoch": 0.5185694635488308,
      "grad_norm": 0.6094454259822268,
      "learning_rate": 0.0004944299154770673,
      "loss": 2.2432,
      "step": 1131
    },
    {
      "epoch": 0.5190279688216415,
      "grad_norm": 0.6278843703443606,
      "learning_rate": 0.0004936872746830802,
      "loss": 2.2772,
      "step": 1132
    },
    {
      "epoch": 0.519486474094452,
      "grad_norm": 0.5847701281314913,
      "learning_rate": 0.0004929446478173195,
      "loss": 2.3516,
      "step": 1133
    },
    {
      "epoch": 0.5199449793672627,
      "grad_norm": 0.594693336044111,
      "learning_rate": 0.0004922020365182968,
      "loss": 2.2809,
      "step": 1134
    },
    {
      "epoch": 0.5204034846400734,
      "grad_norm": 0.6143423671929271,
      "learning_rate": 0.0004914594424244897,
      "loss": 2.3064,
      "step": 1135
    },
    {
      "epoch": 0.520861989912884,
      "grad_norm": 0.5865446384353833,
      "learning_rate": 0.0004907168671743376,
      "loss": 2.2778,
      "step": 1136
    },
    {
      "epoch": 0.5213204951856947,
      "grad_norm": 0.6126728574189834,
      "learning_rate": 0.0004899743124062387,
      "loss": 2.2986,
      "step": 1137
    },
    {
      "epoch": 0.5217790004585052,
      "grad_norm": 0.6728952397813074,
      "learning_rate": 0.0004892317797585456,
      "loss": 2.2453,
      "step": 1138
    },
    {
      "epoch": 0.5222375057313159,
      "grad_norm": 0.5906605121661976,
      "learning_rate": 0.0004884892708695623,
      "loss": 2.1728,
      "step": 1139
    },
    {
      "epoch": 0.5226960110041265,
      "grad_norm": 0.608219687335723,
      "learning_rate": 0.0004877467873775402,
      "loss": 2.3085,
      "step": 1140
    },
    {
      "epoch": 0.5231545162769372,
      "grad_norm": 0.6702472101682647,
      "learning_rate": 0.00048700433092067473,
      "loss": 2.3805,
      "step": 1141
    },
    {
      "epoch": 0.5236130215497479,
      "grad_norm": 0.6041127196888668,
      "learning_rate": 0.0004862619031371019,
      "loss": 2.3,
      "step": 1142
    },
    {
      "epoch": 0.5240715268225584,
      "grad_norm": 0.5894518733287804,
      "learning_rate": 0.0004855195056648942,
      "loss": 2.2342,
      "step": 1143
    },
    {
      "epoch": 0.5245300320953691,
      "grad_norm": 0.5905080464156753,
      "learning_rate": 0.00048477714014205734,
      "loss": 2.2813,
      "step": 1144
    },
    {
      "epoch": 0.5249885373681797,
      "grad_norm": 0.6331706902871318,
      "learning_rate": 0.00048403480820652644,
      "loss": 2.3637,
      "step": 1145
    },
    {
      "epoch": 0.5254470426409904,
      "grad_norm": 0.5693324479082843,
      "learning_rate": 0.0004832925114961629,
      "loss": 2.3005,
      "step": 1146
    },
    {
      "epoch": 0.5259055479138011,
      "grad_norm": 0.5928445954795357,
      "learning_rate": 0.0004825502516487497,
      "loss": 2.3011,
      "step": 1147
    },
    {
      "epoch": 0.5263640531866116,
      "grad_norm": 0.5984118217642289,
      "learning_rate": 0.00048180803030198896,
      "loss": 2.3965,
      "step": 1148
    },
    {
      "epoch": 0.5268225584594223,
      "grad_norm": 0.5865998415278872,
      "learning_rate": 0.0004810658490934979,
      "loss": 2.2943,
      "step": 1149
    },
    {
      "epoch": 0.5272810637322329,
      "grad_norm": 0.5988645325351258,
      "learning_rate": 0.000480323709660805,
      "loss": 2.4055,
      "step": 1150
    },
    {
      "epoch": 0.5277395690050436,
      "grad_norm": 0.5955876875639191,
      "learning_rate": 0.0004795816136413467,
      "loss": 2.2129,
      "step": 1151
    },
    {
      "epoch": 0.5281980742778541,
      "grad_norm": 0.5678522663652664,
      "learning_rate": 0.00047883956267246353,
      "loss": 2.3329,
      "step": 1152
    },
    {
      "epoch": 0.5286565795506648,
      "grad_norm": 0.5776007910968026,
      "learning_rate": 0.00047809755839139657,
      "loss": 2.3383,
      "step": 1153
    },
    {
      "epoch": 0.5291150848234755,
      "grad_norm": 0.6065214048130629,
      "learning_rate": 0.0004773556024352841,
      "loss": 2.3252,
      "step": 1154
    },
    {
      "epoch": 0.5295735900962861,
      "grad_norm": 0.5886967201319924,
      "learning_rate": 0.00047661369644115754,
      "loss": 2.1723,
      "step": 1155
    },
    {
      "epoch": 0.5300320953690968,
      "grad_norm": 0.5960337706659691,
      "learning_rate": 0.0004758718420459383,
      "loss": 2.2096,
      "step": 1156
    },
    {
      "epoch": 0.5304906006419073,
      "grad_norm": 0.6024018518135706,
      "learning_rate": 0.0004751300408864339,
      "loss": 2.2325,
      "step": 1157
    },
    {
      "epoch": 0.530949105914718,
      "grad_norm": 0.5835344560334369,
      "learning_rate": 0.00047438829459933414,
      "loss": 2.2995,
      "step": 1158
    },
    {
      "epoch": 0.5314076111875287,
      "grad_norm": 0.6199692842145512,
      "learning_rate": 0.0004736466048212082,
      "loss": 2.2964,
      "step": 1159
    },
    {
      "epoch": 0.5318661164603393,
      "grad_norm": 0.6088335410545618,
      "learning_rate": 0.0004729049731885002,
      "loss": 2.2831,
      "step": 1160
    },
    {
      "epoch": 0.53232462173315,
      "grad_norm": 0.5952062532847094,
      "learning_rate": 0.000472163401337526,
      "loss": 2.2537,
      "step": 1161
    },
    {
      "epoch": 0.5327831270059605,
      "grad_norm": 0.5775963639125902,
      "learning_rate": 0.00047142189090446985,
      "loss": 2.2179,
      "step": 1162
    },
    {
      "epoch": 0.5332416322787712,
      "grad_norm": 0.5725820029646824,
      "learning_rate": 0.0004706804435253802,
      "loss": 2.3937,
      "step": 1163
    },
    {
      "epoch": 0.5337001375515819,
      "grad_norm": 0.586312687976074,
      "learning_rate": 0.0004699390608361665,
      "loss": 2.4625,
      "step": 1164
    },
    {
      "epoch": 0.5341586428243925,
      "grad_norm": 0.5812764420318348,
      "learning_rate": 0.0004691977444725955,
      "loss": 2.2911,
      "step": 1165
    },
    {
      "epoch": 0.5346171480972031,
      "grad_norm": 0.5830770635929471,
      "learning_rate": 0.0004684564960702877,
      "loss": 2.3261,
      "step": 1166
    },
    {
      "epoch": 0.5350756533700137,
      "grad_norm": 0.574325132358433,
      "learning_rate": 0.0004677153172647131,
      "loss": 2.298,
      "step": 1167
    },
    {
      "epoch": 0.5355341586428244,
      "grad_norm": 0.5910568694254892,
      "learning_rate": 0.00046697420969118894,
      "loss": 2.3605,
      "step": 1168
    },
    {
      "epoch": 0.535992663915635,
      "grad_norm": 0.5836788011146854,
      "learning_rate": 0.00046623317498487466,
      "loss": 2.2542,
      "step": 1169
    },
    {
      "epoch": 0.5364511691884457,
      "grad_norm": 0.5983786066548055,
      "learning_rate": 0.0004654922147807694,
      "loss": 2.4187,
      "step": 1170
    },
    {
      "epoch": 0.5369096744612563,
      "grad_norm": 0.5504321890701663,
      "learning_rate": 0.00046475133071370757,
      "loss": 2.2695,
      "step": 1171
    },
    {
      "epoch": 0.5373681797340669,
      "grad_norm": 0.5892669700967379,
      "learning_rate": 0.00046401052441835574,
      "loss": 2.253,
      "step": 1172
    },
    {
      "epoch": 0.5378266850068776,
      "grad_norm": 0.5876454535381371,
      "learning_rate": 0.000463269797529209,
      "loss": 2.1869,
      "step": 1173
    },
    {
      "epoch": 0.5382851902796882,
      "grad_norm": 0.6203264076402072,
      "learning_rate": 0.00046252915168058697,
      "loss": 2.3277,
      "step": 1174
    },
    {
      "epoch": 0.5387436955524989,
      "grad_norm": 0.6114819334326105,
      "learning_rate": 0.0004617885885066305,
      "loss": 2.222,
      "step": 1175
    },
    {
      "epoch": 0.5392022008253095,
      "grad_norm": 0.5813225906418564,
      "learning_rate": 0.0004610481096412984,
      "loss": 2.286,
      "step": 1176
    },
    {
      "epoch": 0.5396607060981201,
      "grad_norm": 0.6327339827962377,
      "learning_rate": 0.000460307716718363,
      "loss": 2.2731,
      "step": 1177
    },
    {
      "epoch": 0.5401192113709308,
      "grad_norm": 0.6200039584214254,
      "learning_rate": 0.0004595674113714074,
      "loss": 2.2749,
      "step": 1178
    },
    {
      "epoch": 0.5405777166437414,
      "grad_norm": 0.6110017685843414,
      "learning_rate": 0.0004588271952338212,
      "loss": 2.3295,
      "step": 1179
    },
    {
      "epoch": 0.5410362219165521,
      "grad_norm": 0.6231383082156519,
      "learning_rate": 0.00045808706993879714,
      "loss": 2.4227,
      "step": 1180
    },
    {
      "epoch": 0.5414947271893626,
      "grad_norm": 0.5910344446128551,
      "learning_rate": 0.00045734703711932767,
      "loss": 2.2784,
      "step": 1181
    },
    {
      "epoch": 0.5419532324621733,
      "grad_norm": 0.5717745700658147,
      "learning_rate": 0.0004566070984082013,
      "loss": 2.1739,
      "step": 1182
    },
    {
      "epoch": 0.542411737734984,
      "grad_norm": 0.6078724158590614,
      "learning_rate": 0.00045586725543799865,
      "loss": 2.249,
      "step": 1183
    },
    {
      "epoch": 0.5428702430077946,
      "grad_norm": 0.6157017332461565,
      "learning_rate": 0.00045512750984108937,
      "loss": 2.317,
      "step": 1184
    },
    {
      "epoch": 0.5433287482806052,
      "grad_norm": 0.5915263852388473,
      "learning_rate": 0.000454387863249628,
      "loss": 2.3931,
      "step": 1185
    },
    {
      "epoch": 0.5437872535534158,
      "grad_norm": 0.6091492050022739,
      "learning_rate": 0.00045364831729555096,
      "loss": 2.2371,
      "step": 1186
    },
    {
      "epoch": 0.5442457588262265,
      "grad_norm": 0.6083825775784162,
      "learning_rate": 0.0004529088736105721,
      "loss": 2.3522,
      "step": 1187
    },
    {
      "epoch": 0.5447042640990372,
      "grad_norm": 0.615063714348859,
      "learning_rate": 0.0004521695338261802,
      "loss": 2.3529,
      "step": 1188
    },
    {
      "epoch": 0.5451627693718478,
      "grad_norm": 0.6161722464475398,
      "learning_rate": 0.0004514302995736344,
      "loss": 2.3345,
      "step": 1189
    },
    {
      "epoch": 0.5456212746446584,
      "grad_norm": 0.5754311089200006,
      "learning_rate": 0.0004506911724839613,
      "loss": 2.2989,
      "step": 1190
    },
    {
      "epoch": 0.546079779917469,
      "grad_norm": 0.5702997565580797,
      "learning_rate": 0.0004499521541879508,
      "loss": 2.2495,
      "step": 1191
    },
    {
      "epoch": 0.5465382851902797,
      "grad_norm": 0.6200386339279989,
      "learning_rate": 0.00044921324631615303,
      "loss": 2.2955,
      "step": 1192
    },
    {
      "epoch": 0.5469967904630904,
      "grad_norm": 0.585514513215187,
      "learning_rate": 0.0004484744504988742,
      "loss": 2.2849,
      "step": 1193
    },
    {
      "epoch": 0.547455295735901,
      "grad_norm": 0.6330661341738353,
      "learning_rate": 0.00044773576836617336,
      "loss": 2.2442,
      "step": 1194
    },
    {
      "epoch": 0.5479138010087116,
      "grad_norm": 0.5926935060136738,
      "learning_rate": 0.0004469972015478588,
      "loss": 2.2672,
      "step": 1195
    },
    {
      "epoch": 0.5483723062815222,
      "grad_norm": 0.5899889279628662,
      "learning_rate": 0.0004462587516734844,
      "loss": 2.2638,
      "step": 1196
    },
    {
      "epoch": 0.5488308115543329,
      "grad_norm": 0.5810834060111394,
      "learning_rate": 0.00044552042037234596,
      "loss": 2.359,
      "step": 1197
    },
    {
      "epoch": 0.5492893168271435,
      "grad_norm": 0.5851132409993337,
      "learning_rate": 0.00044478220927347774,
      "loss": 2.3189,
      "step": 1198
    },
    {
      "epoch": 0.5497478220999541,
      "grad_norm": 0.5591842462665795,
      "learning_rate": 0.00044404412000564875,
      "loss": 2.2322,
      "step": 1199
    },
    {
      "epoch": 0.5502063273727648,
      "grad_norm": 0.6009910572062622,
      "learning_rate": 0.000443306154197359,
      "loss": 2.2984,
      "step": 1200
    },
    {
      "epoch": 0.5506648326455754,
      "grad_norm": 0.5928974320441521,
      "learning_rate": 0.00044256831347683646,
      "loss": 2.2577,
      "step": 1201
    },
    {
      "epoch": 0.5511233379183861,
      "grad_norm": 0.6703591869581099,
      "learning_rate": 0.0004418305994720328,
      "loss": 2.3071,
      "step": 1202
    },
    {
      "epoch": 0.5515818431911967,
      "grad_norm": 0.5596868695182028,
      "learning_rate": 0.0004410930138106203,
      "loss": 2.2589,
      "step": 1203
    },
    {
      "epoch": 0.5520403484640073,
      "grad_norm": 0.5654189207294678,
      "learning_rate": 0.000440355558119988,
      "loss": 2.3485,
      "step": 1204
    },
    {
      "epoch": 0.552498853736818,
      "grad_norm": 0.6237551813726985,
      "learning_rate": 0.00043961823402723814,
      "loss": 2.3029,
      "step": 1205
    },
    {
      "epoch": 0.5529573590096286,
      "grad_norm": 0.6089929936387445,
      "learning_rate": 0.0004388810431591829,
      "loss": 2.3945,
      "step": 1206
    },
    {
      "epoch": 0.5534158642824393,
      "grad_norm": 0.6081091020016727,
      "learning_rate": 0.0004381439871423398,
      "loss": 2.2227,
      "step": 1207
    },
    {
      "epoch": 0.5538743695552499,
      "grad_norm": 0.5853906070013539,
      "learning_rate": 0.00043740706760292966,
      "loss": 2.2377,
      "step": 1208
    },
    {
      "epoch": 0.5543328748280605,
      "grad_norm": 0.5863006568585318,
      "learning_rate": 0.0004366702861668716,
      "loss": 2.3514,
      "step": 1209
    },
    {
      "epoch": 0.5547913801008711,
      "grad_norm": 0.5703582139097956,
      "learning_rate": 0.00043593364445978036,
      "loss": 2.3399,
      "step": 1210
    },
    {
      "epoch": 0.5552498853736818,
      "grad_norm": 0.6263848893071013,
      "learning_rate": 0.0004351971441069622,
      "loss": 2.2036,
      "step": 1211
    },
    {
      "epoch": 0.5557083906464925,
      "grad_norm": 0.6125153174723239,
      "learning_rate": 0.0004344607867334116,
      "loss": 2.2301,
      "step": 1212
    },
    {
      "epoch": 0.556166895919303,
      "grad_norm": 0.6183449807079889,
      "learning_rate": 0.00043372457396380766,
      "loss": 2.2114,
      "step": 1213
    },
    {
      "epoch": 0.5566254011921137,
      "grad_norm": 0.6292132469089465,
      "learning_rate": 0.00043298850742251013,
      "loss": 2.408,
      "step": 1214
    },
    {
      "epoch": 0.5570839064649243,
      "grad_norm": 0.6023914046684649,
      "learning_rate": 0.0004322525887335563,
      "loss": 2.2795,
      "step": 1215
    },
    {
      "epoch": 0.557542411737735,
      "grad_norm": 0.6305365584433974,
      "learning_rate": 0.00043151681952065734,
      "loss": 2.3323,
      "step": 1216
    },
    {
      "epoch": 0.5580009170105457,
      "grad_norm": 0.6344021128614875,
      "learning_rate": 0.00043078120140719456,
      "loss": 2.388,
      "step": 1217
    },
    {
      "epoch": 0.5584594222833562,
      "grad_norm": 0.5821059752765666,
      "learning_rate": 0.0004300457360162158,
      "loss": 2.2886,
      "step": 1218
    },
    {
      "epoch": 0.5589179275561669,
      "grad_norm": 0.5792937469454237,
      "learning_rate": 0.0004293104249704319,
      "loss": 2.2494,
      "step": 1219
    },
    {
      "epoch": 0.5593764328289775,
      "grad_norm": 0.5933838095642413,
      "learning_rate": 0.00042857526989221355,
      "loss": 2.3473,
      "step": 1220
    },
    {
      "epoch": 0.5598349381017882,
      "grad_norm": 0.6321953101284568,
      "learning_rate": 0.00042784027240358674,
      "loss": 2.3273,
      "step": 1221
    },
    {
      "epoch": 0.5602934433745989,
      "grad_norm": 0.611963267026813,
      "learning_rate": 0.0004271054341262301,
      "loss": 2.2637,
      "step": 1222
    },
    {
      "epoch": 0.5607519486474094,
      "grad_norm": 0.581552185877168,
      "learning_rate": 0.000426370756681471,
      "loss": 2.2744,
      "step": 1223
    },
    {
      "epoch": 0.5612104539202201,
      "grad_norm": 0.6073051715019443,
      "learning_rate": 0.0004256362416902817,
      "loss": 2.2982,
      "step": 1224
    },
    {
      "epoch": 0.5616689591930307,
      "grad_norm": 0.6259982188361063,
      "learning_rate": 0.00042490189077327637,
      "loss": 2.2687,
      "step": 1225
    },
    {
      "epoch": 0.5621274644658414,
      "grad_norm": 0.6284776350760171,
      "learning_rate": 0.00042416770555070703,
      "loss": 2.32,
      "step": 1226
    },
    {
      "epoch": 0.562585969738652,
      "grad_norm": 0.6014623795230621,
      "learning_rate": 0.00042343368764246,
      "loss": 2.1865,
      "step": 1227
    },
    {
      "epoch": 0.5630444750114626,
      "grad_norm": 0.5920611617013254,
      "learning_rate": 0.0004226998386680524,
      "loss": 2.3499,
      "step": 1228
    },
    {
      "epoch": 0.5635029802842733,
      "grad_norm": 0.588261858834969,
      "learning_rate": 0.000421966160246629,
      "loss": 2.3915,
      "step": 1229
    },
    {
      "epoch": 0.5639614855570839,
      "grad_norm": 0.612369044253157,
      "learning_rate": 0.00042123265399695783,
      "loss": 2.2822,
      "step": 1230
    },
    {
      "epoch": 0.5644199908298946,
      "grad_norm": 0.5906488243593299,
      "learning_rate": 0.0004204993215374273,
      "loss": 2.4266,
      "step": 1231
    },
    {
      "epoch": 0.5648784961027051,
      "grad_norm": 0.5995887815813583,
      "learning_rate": 0.00041976616448604226,
      "loss": 2.3545,
      "step": 1232
    },
    {
      "epoch": 0.5653370013755158,
      "grad_norm": 0.5683754139534649,
      "learning_rate": 0.00041903318446042076,
      "loss": 2.2711,
      "step": 1233
    },
    {
      "epoch": 0.5657955066483265,
      "grad_norm": 0.6167506614535563,
      "learning_rate": 0.00041830038307778984,
      "loss": 2.3484,
      "step": 1234
    },
    {
      "epoch": 0.5662540119211371,
      "grad_norm": 0.6510189227114996,
      "learning_rate": 0.0004175677619549828,
      "loss": 2.3408,
      "step": 1235
    },
    {
      "epoch": 0.5667125171939478,
      "grad_norm": 0.5937378703799082,
      "learning_rate": 0.000416835322708435,
      "loss": 2.1873,
      "step": 1236
    },
    {
      "epoch": 0.5671710224667583,
      "grad_norm": 0.5987148968102344,
      "learning_rate": 0.00041610306695418056,
      "loss": 2.2368,
      "step": 1237
    },
    {
      "epoch": 0.567629527739569,
      "grad_norm": 0.6679552415686504,
      "learning_rate": 0.0004153709963078488,
      "loss": 2.3354,
      "step": 1238
    },
    {
      "epoch": 0.5680880330123796,
      "grad_norm": 0.6045559781958659,
      "learning_rate": 0.0004146391123846606,
      "loss": 2.2922,
      "step": 1239
    },
    {
      "epoch": 0.5685465382851903,
      "grad_norm": 0.7125700357146528,
      "learning_rate": 0.0004139074167994249,
      "loss": 2.3071,
      "step": 1240
    },
    {
      "epoch": 0.569005043558001,
      "grad_norm": 0.6158500750548562,
      "learning_rate": 0.00041317591116653486,
      "loss": 2.3385,
      "step": 1241
    },
    {
      "epoch": 0.5694635488308115,
      "grad_norm": 0.5681443407264842,
      "learning_rate": 0.0004124445970999648,
      "loss": 2.3201,
      "step": 1242
    },
    {
      "epoch": 0.5699220541036222,
      "grad_norm": 0.6367483615754531,
      "learning_rate": 0.00041171347621326627,
      "loss": 2.3066,
      "step": 1243
    },
    {
      "epoch": 0.5703805593764328,
      "grad_norm": 0.6228831901681322,
      "learning_rate": 0.00041098255011956465,
      "loss": 2.3503,
      "step": 1244
    },
    {
      "epoch": 0.5708390646492435,
      "grad_norm": 0.5891464930748488,
      "learning_rate": 0.00041025182043155547,
      "loss": 2.3461,
      "step": 1245
    },
    {
      "epoch": 0.571297569922054,
      "grad_norm": 0.6569628686468844,
      "learning_rate": 0.000409521288761501,
      "loss": 2.3524,
      "step": 1246
    },
    {
      "epoch": 0.5717560751948647,
      "grad_norm": 0.5946842678793178,
      "learning_rate": 0.00040879095672122646,
      "loss": 2.2735,
      "step": 1247
    },
    {
      "epoch": 0.5722145804676754,
      "grad_norm": 0.6591309327179836,
      "learning_rate": 0.0004080608259221167,
      "loss": 2.309,
      "step": 1248
    },
    {
      "epoch": 0.572673085740486,
      "grad_norm": 0.5795875010366746,
      "learning_rate": 0.0004073308979751126,
      "loss": 2.2569,
      "step": 1249
    },
    {
      "epoch": 0.5731315910132967,
      "grad_norm": 0.5980188901961666,
      "learning_rate": 0.0004066011744907074,
      "loss": 2.2274,
      "step": 1250
    },
    {
      "epoch": 0.5735900962861072,
      "grad_norm": 0.6169573998240487,
      "learning_rate": 0.00040587165707894326,
      "loss": 2.3013,
      "step": 1251
    },
    {
      "epoch": 0.5740486015589179,
      "grad_norm": 0.6401017766533617,
      "learning_rate": 0.0004051423473494076,
      "loss": 2.2178,
      "step": 1252
    },
    {
      "epoch": 0.5745071068317286,
      "grad_norm": 0.5793004970781344,
      "learning_rate": 0.0004044132469112299,
      "loss": 2.2893,
      "step": 1253
    },
    {
      "epoch": 0.5749656121045392,
      "grad_norm": 0.5946711872069927,
      "learning_rate": 0.00040368435737307733,
      "loss": 2.3439,
      "step": 1254
    },
    {
      "epoch": 0.5754241173773499,
      "grad_norm": 0.5954871817141715,
      "learning_rate": 0.00040295568034315224,
      "loss": 2.2166,
      "step": 1255
    },
    {
      "epoch": 0.5758826226501604,
      "grad_norm": 0.5606112889669284,
      "learning_rate": 0.0004022272174291878,
      "loss": 2.363,
      "step": 1256
    },
    {
      "epoch": 0.5763411279229711,
      "grad_norm": 0.5874023264582118,
      "learning_rate": 0.0004014989702384449,
      "loss": 2.2367,
      "step": 1257
    },
    {
      "epoch": 0.5767996331957818,
      "grad_norm": 0.5820236073542624,
      "learning_rate": 0.00040077094037770843,
      "loss": 2.3274,
      "step": 1258
    },
    {
      "epoch": 0.5772581384685924,
      "grad_norm": 0.6145235730093623,
      "learning_rate": 0.0004000431294532838,
      "loss": 2.2552,
      "step": 1259
    },
    {
      "epoch": 0.5777166437414031,
      "grad_norm": 0.5831640404017642,
      "learning_rate": 0.0003993155390709935,
      "loss": 2.3151,
      "step": 1260
    },
    {
      "epoch": 0.5781751490142136,
      "grad_norm": 0.5617484832802628,
      "learning_rate": 0.0003985881708361729,
      "loss": 2.3646,
      "step": 1261
    },
    {
      "epoch": 0.5786336542870243,
      "grad_norm": 0.7167138730002593,
      "learning_rate": 0.00039786102635366784,
      "loss": 2.2061,
      "step": 1262
    },
    {
      "epoch": 0.579092159559835,
      "grad_norm": 0.5834083647796465,
      "learning_rate": 0.0003971341072278302,
      "loss": 2.2966,
      "step": 1263
    },
    {
      "epoch": 0.5795506648326456,
      "grad_norm": 0.5911492750593945,
      "learning_rate": 0.00039640741506251457,
      "loss": 2.3838,
      "step": 1264
    },
    {
      "epoch": 0.5800091701054562,
      "grad_norm": 0.6103184233252041,
      "learning_rate": 0.00039568095146107495,
      "loss": 2.2234,
      "step": 1265
    },
    {
      "epoch": 0.5804676753782668,
      "grad_norm": 0.6073072921348871,
      "learning_rate": 0.00039495471802636096,
      "loss": 2.4038,
      "step": 1266
    },
    {
      "epoch": 0.5809261806510775,
      "grad_norm": 0.5926417301368743,
      "learning_rate": 0.0003942287163607145,
      "loss": 2.2005,
      "step": 1267
    },
    {
      "epoch": 0.5813846859238881,
      "grad_norm": 0.5940335622649773,
      "learning_rate": 0.0003935029480659658,
      "loss": 2.3231,
      "step": 1268
    },
    {
      "epoch": 0.5818431911966988,
      "grad_norm": 0.575974868324918,
      "learning_rate": 0.00039277741474343054,
      "loss": 2.37,
      "step": 1269
    },
    {
      "epoch": 0.5823016964695094,
      "grad_norm": 0.5884240482353512,
      "learning_rate": 0.0003920521179939057,
      "loss": 2.2749,
      "step": 1270
    },
    {
      "epoch": 0.58276020174232,
      "grad_norm": 0.5980799531622462,
      "learning_rate": 0.00039132705941766644,
      "loss": 2.3793,
      "step": 1271
    },
    {
      "epoch": 0.5832187070151307,
      "grad_norm": 0.626354512759079,
      "learning_rate": 0.0003906022406144624,
      "loss": 2.205,
      "step": 1272
    },
    {
      "epoch": 0.5836772122879413,
      "grad_norm": 0.5929095525965009,
      "learning_rate": 0.0003898776631835143,
      "loss": 2.3512,
      "step": 1273
    },
    {
      "epoch": 0.584135717560752,
      "grad_norm": 0.6290098494259465,
      "learning_rate": 0.00038915332872350994,
      "loss": 2.2118,
      "step": 1274
    },
    {
      "epoch": 0.5845942228335625,
      "grad_norm": 0.55739364599535,
      "learning_rate": 0.00038842923883260135,
      "loss": 2.191,
      "step": 1275
    },
    {
      "epoch": 0.5850527281063732,
      "grad_norm": 0.616269402065547,
      "learning_rate": 0.00038770539510840093,
      "loss": 2.3498,
      "step": 1276
    },
    {
      "epoch": 0.5855112333791839,
      "grad_norm": 0.5994389718316675,
      "learning_rate": 0.00038698179914797783,
      "loss": 2.2802,
      "step": 1277
    },
    {
      "epoch": 0.5859697386519945,
      "grad_norm": 0.6326345923339509,
      "learning_rate": 0.0003862584525478545,
      "loss": 2.3424,
      "step": 1278
    },
    {
      "epoch": 0.5864282439248051,
      "grad_norm": 0.6140946379891526,
      "learning_rate": 0.00038553535690400353,
      "loss": 2.3678,
      "step": 1279
    },
    {
      "epoch": 0.5868867491976157,
      "grad_norm": 0.5850262461661917,
      "learning_rate": 0.00038481251381184355,
      "loss": 2.2949,
      "step": 1280
    },
    {
      "epoch": 0.5873452544704264,
      "grad_norm": 0.581711440021864,
      "learning_rate": 0.00038408992486623584,
      "loss": 2.2882,
      "step": 1281
    },
    {
      "epoch": 0.5878037597432371,
      "grad_norm": 0.5917125587644554,
      "learning_rate": 0.00038336759166148117,
      "loss": 2.3769,
      "step": 1282
    },
    {
      "epoch": 0.5882622650160477,
      "grad_norm": 0.5685598609303019,
      "learning_rate": 0.0003826455157913159,
      "loss": 2.1807,
      "step": 1283
    },
    {
      "epoch": 0.5887207702888583,
      "grad_norm": 0.5725870479852225,
      "learning_rate": 0.00038192369884890886,
      "loss": 2.321,
      "step": 1284
    },
    {
      "epoch": 0.5891792755616689,
      "grad_norm": 0.5880327964219259,
      "learning_rate": 0.00038120214242685723,
      "loss": 2.2242,
      "step": 1285
    },
    {
      "epoch": 0.5896377808344796,
      "grad_norm": 0.6204897446638993,
      "learning_rate": 0.00038048084811718373,
      "loss": 2.2711,
      "step": 1286
    },
    {
      "epoch": 0.5900962861072903,
      "grad_norm": 0.6179535882799805,
      "learning_rate": 0.0003797598175113327,
      "loss": 2.2685,
      "step": 1287
    },
    {
      "epoch": 0.5905547913801009,
      "grad_norm": 0.5927355134944444,
      "learning_rate": 0.0003790390522001662,
      "loss": 2.2372,
      "step": 1288
    },
    {
      "epoch": 0.5910132966529115,
      "grad_norm": 0.6005634036624582,
      "learning_rate": 0.0003783185537739615,
      "loss": 2.3695,
      "step": 1289
    },
    {
      "epoch": 0.5914718019257221,
      "grad_norm": 0.634573009959429,
      "learning_rate": 0.00037759832382240697,
      "loss": 2.3083,
      "step": 1290
    },
    {
      "epoch": 0.5919303071985328,
      "grad_norm": 0.6137727817098055,
      "learning_rate": 0.00037687836393459826,
      "loss": 2.3051,
      "step": 1291
    },
    {
      "epoch": 0.5923888124713435,
      "grad_norm": 0.646951971308209,
      "learning_rate": 0.0003761586756990354,
      "loss": 2.2664,
      "step": 1292
    },
    {
      "epoch": 0.5928473177441541,
      "grad_norm": 0.5790136060008407,
      "learning_rate": 0.0003754392607036191,
      "loss": 2.3521,
      "step": 1293
    },
    {
      "epoch": 0.5933058230169647,
      "grad_norm": 0.5897570368776711,
      "learning_rate": 0.0003747201205356472,
      "loss": 2.3149,
      "step": 1294
    },
    {
      "epoch": 0.5937643282897753,
      "grad_norm": 0.5920206424314559,
      "learning_rate": 0.0003740012567818111,
      "loss": 2.4262,
      "step": 1295
    },
    {
      "epoch": 0.594222833562586,
      "grad_norm": 0.5884492410669395,
      "learning_rate": 0.0003732826710281922,
      "loss": 2.2145,
      "step": 1296
    },
    {
      "epoch": 0.5946813388353966,
      "grad_norm": 0.5904349350494954,
      "learning_rate": 0.0003725643648602588,
      "loss": 2.2763,
      "step": 1297
    },
    {
      "epoch": 0.5951398441082072,
      "grad_norm": 0.6035427440700586,
      "learning_rate": 0.0003718463398628621,
      "loss": 2.3375,
      "step": 1298
    },
    {
      "epoch": 0.5955983493810179,
      "grad_norm": 0.5474952568744413,
      "learning_rate": 0.0003711285976202331,
      "loss": 2.3219,
      "step": 1299
    },
    {
      "epoch": 0.5960568546538285,
      "grad_norm": 0.5834781553192486,
      "learning_rate": 0.0003704111397159787,
      "loss": 2.249,
      "step": 1300
    },
    {
      "epoch": 0.5965153599266392,
      "grad_norm": 0.5384582596152956,
      "learning_rate": 0.0003696939677330788,
      "loss": 2.2684,
      "step": 1301
    },
    {
      "epoch": 0.5969738651994498,
      "grad_norm": 0.5565141754088955,
      "learning_rate": 0.00036897708325388213,
      "loss": 2.2319,
      "step": 1302
    },
    {
      "epoch": 0.5974323704722604,
      "grad_norm": 0.6095812261655434,
      "learning_rate": 0.0003682604878601034,
      "loss": 2.238,
      "step": 1303
    },
    {
      "epoch": 0.597890875745071,
      "grad_norm": 0.5914787808303502,
      "learning_rate": 0.000367544183132819,
      "loss": 2.2142,
      "step": 1304
    },
    {
      "epoch": 0.5983493810178817,
      "grad_norm": 0.5921332979052781,
      "learning_rate": 0.0003668281706524645,
      "loss": 2.1711,
      "step": 1305
    },
    {
      "epoch": 0.5988078862906924,
      "grad_norm": 0.5853067536724046,
      "learning_rate": 0.0003661124519988304,
      "loss": 2.2953,
      "step": 1306
    },
    {
      "epoch": 0.599266391563503,
      "grad_norm": 0.5937586207983536,
      "learning_rate": 0.00036539702875105893,
      "loss": 2.2632,
      "step": 1307
    },
    {
      "epoch": 0.5997248968363136,
      "grad_norm": 0.6062374773860594,
      "learning_rate": 0.0003646819024876406,
      "loss": 2.2785,
      "step": 1308
    },
    {
      "epoch": 0.6001834021091242,
      "grad_norm": 0.6300962680637766,
      "learning_rate": 0.0003639670747864107,
      "loss": 2.3635,
      "step": 1309
    },
    {
      "epoch": 0.6006419073819349,
      "grad_norm": 0.587538696415061,
      "learning_rate": 0.00036325254722454584,
      "loss": 2.3085,
      "step": 1310
    },
    {
      "epoch": 0.6011004126547456,
      "grad_norm": 0.6232567326960061,
      "learning_rate": 0.00036253832137855997,
      "loss": 2.272,
      "step": 1311
    },
    {
      "epoch": 0.6015589179275561,
      "grad_norm": 0.5893968124789836,
      "learning_rate": 0.00036182439882430183,
      "loss": 2.281,
      "step": 1312
    },
    {
      "epoch": 0.6020174232003668,
      "grad_norm": 0.5782293891176509,
      "learning_rate": 0.00036111078113695096,
      "loss": 2.1993,
      "step": 1313
    },
    {
      "epoch": 0.6024759284731774,
      "grad_norm": 0.6042471482530328,
      "learning_rate": 0.0003603974698910139,
      "loss": 2.1887,
      "step": 1314
    },
    {
      "epoch": 0.6029344337459881,
      "grad_norm": 0.5847653335981147,
      "learning_rate": 0.0003596844666603214,
      "loss": 2.1913,
      "step": 1315
    },
    {
      "epoch": 0.6033929390187988,
      "grad_norm": 0.5942628969686846,
      "learning_rate": 0.0003589717730180245,
      "loss": 2.2911,
      "step": 1316
    },
    {
      "epoch": 0.6038514442916093,
      "grad_norm": 0.5766296228235691,
      "learning_rate": 0.00035825939053659117,
      "loss": 2.3968,
      "step": 1317
    },
    {
      "epoch": 0.60430994956442,
      "grad_norm": 0.5712303236368929,
      "learning_rate": 0.00035754732078780273,
      "loss": 2.1681,
      "step": 1318
    },
    {
      "epoch": 0.6047684548372306,
      "grad_norm": 0.613392694865193,
      "learning_rate": 0.00035683556534275076,
      "loss": 2.4141,
      "step": 1319
    },
    {
      "epoch": 0.6052269601100413,
      "grad_norm": 0.6192060710135091,
      "learning_rate": 0.00035612412577183303,
      "loss": 2.2384,
      "step": 1320
    },
    {
      "epoch": 0.605685465382852,
      "grad_norm": 0.6105304895872194,
      "learning_rate": 0.00035541300364475063,
      "loss": 2.2229,
      "step": 1321
    },
    {
      "epoch": 0.6061439706556625,
      "grad_norm": 0.5790083082991705,
      "learning_rate": 0.0003547022005305043,
      "loss": 2.3091,
      "step": 1322
    },
    {
      "epoch": 0.6066024759284732,
      "grad_norm": 0.5918944687226493,
      "learning_rate": 0.0003539917179973907,
      "loss": 2.3233,
      "step": 1323
    },
    {
      "epoch": 0.6070609812012838,
      "grad_norm": 0.5623945157064689,
      "learning_rate": 0.00035328155761299917,
      "loss": 2.3121,
      "step": 1324
    },
    {
      "epoch": 0.6075194864740945,
      "grad_norm": 0.6076711150216336,
      "learning_rate": 0.0003525717209442085,
      "loss": 2.226,
      "step": 1325
    },
    {
      "epoch": 0.6079779917469051,
      "grad_norm": 0.5682293557411094,
      "learning_rate": 0.00035186220955718306,
      "loss": 2.1605,
      "step": 1326
    },
    {
      "epoch": 0.6084364970197157,
      "grad_norm": 0.5857419266744683,
      "learning_rate": 0.0003511530250173696,
      "loss": 2.303,
      "step": 1327
    },
    {
      "epoch": 0.6088950022925264,
      "grad_norm": 0.5957843784707653,
      "learning_rate": 0.00035044416888949364,
      "loss": 2.2414,
      "step": 1328
    },
    {
      "epoch": 0.609353507565337,
      "grad_norm": 0.6076893943248914,
      "learning_rate": 0.0003497356427375562,
      "loss": 2.2791,
      "step": 1329
    },
    {
      "epoch": 0.6098120128381477,
      "grad_norm": 0.6174261429264002,
      "learning_rate": 0.00034902744812483034,
      "loss": 2.3281,
      "step": 1330
    },
    {
      "epoch": 0.6102705181109582,
      "grad_norm": 0.5814351735934518,
      "learning_rate": 0.00034831958661385714,
      "loss": 2.3291,
      "step": 1331
    },
    {
      "epoch": 0.6107290233837689,
      "grad_norm": 0.5353767711995423,
      "learning_rate": 0.0003476120597664434,
      "loss": 2.2202,
      "step": 1332
    },
    {
      "epoch": 0.6111875286565795,
      "grad_norm": 0.601806303018394,
      "learning_rate": 0.00034690486914365704,
      "loss": 2.3021,
      "step": 1333
    },
    {
      "epoch": 0.6116460339293902,
      "grad_norm": 0.5789997343401071,
      "learning_rate": 0.00034619801630582435,
      "loss": 2.3663,
      "step": 1334
    },
    {
      "epoch": 0.6121045392022009,
      "grad_norm": 0.5857582194513566,
      "learning_rate": 0.00034549150281252633,
      "loss": 2.2045,
      "step": 1335
    },
    {
      "epoch": 0.6125630444750114,
      "grad_norm": 0.5912515839579194,
      "learning_rate": 0.0003447853302225952,
      "loss": 2.2527,
      "step": 1336
    },
    {
      "epoch": 0.6130215497478221,
      "grad_norm": 0.6102706001943474,
      "learning_rate": 0.00034407950009411126,
      "loss": 2.139,
      "step": 1337
    },
    {
      "epoch": 0.6134800550206327,
      "grad_norm": 0.5953799380418673,
      "learning_rate": 0.00034337401398439873,
      "loss": 2.2252,
      "step": 1338
    },
    {
      "epoch": 0.6139385602934434,
      "grad_norm": 0.6036464447529098,
      "learning_rate": 0.00034266887345002305,
      "loss": 2.2307,
      "step": 1339
    },
    {
      "epoch": 0.6143970655662541,
      "grad_norm": 0.5965095393901698,
      "learning_rate": 0.0003419640800467874,
      "loss": 2.3248,
      "step": 1340
    },
    {
      "epoch": 0.6148555708390646,
      "grad_norm": 0.6117955241936137,
      "learning_rate": 0.0003412596353297288,
      "loss": 2.3527,
      "step": 1341
    },
    {
      "epoch": 0.6153140761118753,
      "grad_norm": 0.6030001229250563,
      "learning_rate": 0.00034055554085311493,
      "loss": 2.2474,
      "step": 1342
    },
    {
      "epoch": 0.6157725813846859,
      "grad_norm": 0.5622381220159444,
      "learning_rate": 0.00033985179817044105,
      "loss": 2.297,
      "step": 1343
    },
    {
      "epoch": 0.6162310866574966,
      "grad_norm": 0.5756584963108302,
      "learning_rate": 0.0003391484088344257,
      "loss": 2.2679,
      "step": 1344
    },
    {
      "epoch": 0.6166895919303071,
      "grad_norm": 0.5657681169242593,
      "learning_rate": 0.00033844537439700807,
      "loss": 2.3839,
      "step": 1345
    },
    {
      "epoch": 0.6171480972031178,
      "grad_norm": 0.5756975112915997,
      "learning_rate": 0.00033774269640934445,
      "loss": 2.2936,
      "step": 1346
    },
    {
      "epoch": 0.6176066024759285,
      "grad_norm": 0.5919459193828116,
      "learning_rate": 0.0003370403764218045,
      "loss": 2.2337,
      "step": 1347
    },
    {
      "epoch": 0.6180651077487391,
      "grad_norm": 0.5995811995913277,
      "learning_rate": 0.000336338415983968,
      "loss": 2.2486,
      "step": 1348
    },
    {
      "epoch": 0.6185236130215498,
      "grad_norm": 0.5736715348391217,
      "learning_rate": 0.00033563681664462155,
      "loss": 2.255,
      "step": 1349
    },
    {
      "epoch": 0.6189821182943603,
      "grad_norm": 0.645549950210827,
      "learning_rate": 0.000334935579951755,
      "loss": 2.4509,
      "step": 1350
    },
    {
      "epoch": 0.619440623567171,
      "grad_norm": 0.6158948358559565,
      "learning_rate": 0.0003342347074525578,
      "loss": 2.2434,
      "step": 1351
    },
    {
      "epoch": 0.6198991288399817,
      "grad_norm": 0.5876506675314583,
      "learning_rate": 0.0003335342006934161,
      "loss": 2.3326,
      "step": 1352
    },
    {
      "epoch": 0.6203576341127923,
      "grad_norm": 0.5832289398889846,
      "learning_rate": 0.00033283406121990914,
      "loss": 2.3425,
      "step": 1353
    },
    {
      "epoch": 0.620816139385603,
      "grad_norm": 0.6284839606869922,
      "learning_rate": 0.0003321342905768057,
      "loss": 2.2137,
      "step": 1354
    },
    {
      "epoch": 0.6212746446584135,
      "grad_norm": 0.6381065378043375,
      "learning_rate": 0.00033143489030806086,
      "loss": 2.249,
      "step": 1355
    },
    {
      "epoch": 0.6217331499312242,
      "grad_norm": 0.5714416496775975,
      "learning_rate": 0.00033073586195681227,
      "loss": 2.2333,
      "step": 1356
    },
    {
      "epoch": 0.6221916552040349,
      "grad_norm": 0.6005834753869924,
      "learning_rate": 0.00033003720706537736,
      "loss": 2.1766,
      "step": 1357
    },
    {
      "epoch": 0.6226501604768455,
      "grad_norm": 0.7134430079003032,
      "learning_rate": 0.0003293389271752492,
      "loss": 2.2538,
      "step": 1358
    },
    {
      "epoch": 0.6231086657496562,
      "grad_norm": 0.5945921101021953,
      "learning_rate": 0.00032864102382709374,
      "loss": 2.3918,
      "step": 1359
    },
    {
      "epoch": 0.6235671710224667,
      "grad_norm": 0.6059458952133306,
      "learning_rate": 0.000327943498560746,
      "loss": 2.2751,
      "step": 1360
    },
    {
      "epoch": 0.6240256762952774,
      "grad_norm": 0.5996843107810945,
      "learning_rate": 0.00032724635291520694,
      "loss": 2.3271,
      "step": 1361
    },
    {
      "epoch": 0.624484181568088,
      "grad_norm": 0.5904057584790631,
      "learning_rate": 0.00032654958842863967,
      "loss": 2.2773,
      "step": 1362
    },
    {
      "epoch": 0.6249426868408987,
      "grad_norm": 0.5537034945433458,
      "learning_rate": 0.0003258532066383667,
      "loss": 2.1815,
      "step": 1363
    },
    {
      "epoch": 0.6254011921137093,
      "grad_norm": 0.5783756475084596,
      "learning_rate": 0.000325157209080866,
      "loss": 2.2267,
      "step": 1364
    },
    {
      "epoch": 0.6258596973865199,
      "grad_norm": 0.6234923758707516,
      "learning_rate": 0.00032446159729176743,
      "loss": 2.3211,
      "step": 1365
    },
    {
      "epoch": 0.6263182026593306,
      "grad_norm": 0.6008715219534501,
      "learning_rate": 0.0003237663728058502,
      "loss": 2.179,
      "step": 1366
    },
    {
      "epoch": 0.6267767079321412,
      "grad_norm": 0.5651792213835728,
      "learning_rate": 0.0003230715371570389,
      "loss": 2.3465,
      "step": 1367
    },
    {
      "epoch": 0.6272352132049519,
      "grad_norm": 0.6018924782008809,
      "learning_rate": 0.00032237709187839996,
      "loss": 2.2055,
      "step": 1368
    },
    {
      "epoch": 0.6276937184777625,
      "grad_norm": 0.5672786770147051,
      "learning_rate": 0.0003216830385021388,
      "loss": 2.2121,
      "step": 1369
    },
    {
      "epoch": 0.6281522237505731,
      "grad_norm": 0.6382228073850494,
      "learning_rate": 0.0003209893785595959,
      "loss": 2.4142,
      "step": 1370
    },
    {
      "epoch": 0.6286107290233838,
      "grad_norm": 0.590479204479416,
      "learning_rate": 0.00032029611358124366,
      "loss": 2.2381,
      "step": 1371
    },
    {
      "epoch": 0.6290692342961944,
      "grad_norm": 0.6301751855468772,
      "learning_rate": 0.00031960324509668336,
      "loss": 2.2203,
      "step": 1372
    },
    {
      "epoch": 0.6295277395690051,
      "grad_norm": 0.5877547854200119,
      "learning_rate": 0.0003189107746346412,
      "loss": 2.2931,
      "step": 1373
    },
    {
      "epoch": 0.6299862448418156,
      "grad_norm": 0.5917223674544857,
      "learning_rate": 0.0003182187037229653,
      "loss": 2.2616,
      "step": 1374
    },
    {
      "epoch": 0.6304447501146263,
      "grad_norm": 0.6032592389817995,
      "learning_rate": 0.0003175270338886221,
      "loss": 2.4095,
      "step": 1375
    },
    {
      "epoch": 0.630903255387437,
      "grad_norm": 0.6068082071032211,
      "learning_rate": 0.00031683576665769345,
      "loss": 2.3185,
      "step": 1376
    },
    {
      "epoch": 0.6313617606602476,
      "grad_norm": 0.5890624787470702,
      "learning_rate": 0.0003161449035553724,
      "loss": 2.186,
      "step": 1377
    },
    {
      "epoch": 0.6318202659330582,
      "grad_norm": 0.5719750427633006,
      "learning_rate": 0.00031545444610596077,
      "loss": 2.3191,
      "step": 1378
    },
    {
      "epoch": 0.6322787712058688,
      "grad_norm": 0.5895388479594507,
      "learning_rate": 0.000314764395832865,
      "loss": 2.2915,
      "step": 1379
    },
    {
      "epoch": 0.6327372764786795,
      "grad_norm": 0.5925756443414334,
      "learning_rate": 0.0003140747542585934,
      "loss": 2.2269,
      "step": 1380
    },
    {
      "epoch": 0.6331957817514902,
      "grad_norm": 0.5856977119140961,
      "learning_rate": 0.00031338552290475266,
      "loss": 2.2532,
      "step": 1381
    },
    {
      "epoch": 0.6336542870243008,
      "grad_norm": 0.5763979030515763,
      "learning_rate": 0.00031269670329204396,
      "loss": 2.234,
      "step": 1382
    },
    {
      "epoch": 0.6341127922971114,
      "grad_norm": 0.5837371220120813,
      "learning_rate": 0.0003120082969402604,
      "loss": 2.3276,
      "step": 1383
    },
    {
      "epoch": 0.634571297569922,
      "grad_norm": 0.6038629242334109,
      "learning_rate": 0.00031132030536828314,
      "loss": 2.3984,
      "step": 1384
    },
    {
      "epoch": 0.6350298028427327,
      "grad_norm": 0.5881535816369585,
      "learning_rate": 0.00031063273009407805,
      "loss": 2.2755,
      "step": 1385
    },
    {
      "epoch": 0.6354883081155434,
      "grad_norm": 0.6046846475587054,
      "learning_rate": 0.00030994557263469265,
      "loss": 2.2355,
      "step": 1386
    },
    {
      "epoch": 0.635946813388354,
      "grad_norm": 0.5839051060577486,
      "learning_rate": 0.0003092588345062526,
      "loss": 2.2656,
      "step": 1387
    },
    {
      "epoch": 0.6364053186611646,
      "grad_norm": 0.590660938676283,
      "learning_rate": 0.0003085725172239582,
      "loss": 2.3422,
      "step": 1388
    },
    {
      "epoch": 0.6368638239339752,
      "grad_norm": 0.5780222531525664,
      "learning_rate": 0.0003078866223020815,
      "loss": 2.2655,
      "step": 1389
    },
    {
      "epoch": 0.6373223292067859,
      "grad_norm": 0.5950592462436274,
      "learning_rate": 0.0003072011512539624,
      "loss": 2.257,
      "step": 1390
    },
    {
      "epoch": 0.6377808344795965,
      "grad_norm": 0.6016108736105069,
      "learning_rate": 0.00030651610559200574,
      "loss": 2.3284,
      "step": 1391
    },
    {
      "epoch": 0.6382393397524071,
      "grad_norm": 0.6569940553317023,
      "learning_rate": 0.00030583148682767757,
      "loss": 2.254,
      "step": 1392
    },
    {
      "epoch": 0.6386978450252178,
      "grad_norm": 0.6040795343645861,
      "learning_rate": 0.00030514729647150243,
      "loss": 2.2811,
      "step": 1393
    },
    {
      "epoch": 0.6391563502980284,
      "grad_norm": 0.5899062990444471,
      "learning_rate": 0.0003044635360330592,
      "loss": 2.2424,
      "step": 1394
    },
    {
      "epoch": 0.6396148555708391,
      "grad_norm": 0.5964927547823478,
      "learning_rate": 0.00030378020702097845,
      "loss": 2.2887,
      "step": 1395
    },
    {
      "epoch": 0.6400733608436497,
      "grad_norm": 0.6076446910126538,
      "learning_rate": 0.000303097310942939,
      "loss": 2.3608,
      "step": 1396
    },
    {
      "epoch": 0.6405318661164603,
      "grad_norm": 0.5822145643192888,
      "learning_rate": 0.0003024148493056641,
      "loss": 2.2771,
      "step": 1397
    },
    {
      "epoch": 0.640990371389271,
      "grad_norm": 0.6085693284311454,
      "learning_rate": 0.00030173282361491865,
      "loss": 2.2308,
      "step": 1398
    },
    {
      "epoch": 0.6414488766620816,
      "grad_norm": 0.6017560664857039,
      "learning_rate": 0.0003010512353755057,
      "loss": 2.2168,
      "step": 1399
    },
    {
      "epoch": 0.6419073819348923,
      "grad_norm": 0.6132704121568244,
      "learning_rate": 0.00030037008609126313,
      "loss": 2.2639,
      "step": 1400
    },
    {
      "epoch": 0.6423658872077029,
      "grad_norm": 0.6204410652422921,
      "learning_rate": 0.0002996893772650602,
      "loss": 2.2544,
      "step": 1401
    },
    {
      "epoch": 0.6428243924805135,
      "grad_norm": 0.584470861236187,
      "learning_rate": 0.0002990091103987945,
      "loss": 2.3757,
      "step": 1402
    },
    {
      "epoch": 0.6432828977533241,
      "grad_norm": 0.6183366099197916,
      "learning_rate": 0.0002983292869933886,
      "loss": 2.343,
      "step": 1403
    },
    {
      "epoch": 0.6437414030261348,
      "grad_norm": 0.5988974255733841,
      "learning_rate": 0.0002976499085487862,
      "loss": 2.2582,
      "step": 1404
    },
    {
      "epoch": 0.6441999082989455,
      "grad_norm": 0.5976724461740451,
      "learning_rate": 0.00029697097656394963,
      "loss": 2.2882,
      "step": 1405
    },
    {
      "epoch": 0.6446584135717561,
      "grad_norm": 0.6005130046081119,
      "learning_rate": 0.00029629249253685595,
      "loss": 2.2388,
      "step": 1406
    },
    {
      "epoch": 0.6451169188445667,
      "grad_norm": 0.6103026054298105,
      "learning_rate": 0.00029561445796449416,
      "loss": 2.3409,
      "step": 1407
    },
    {
      "epoch": 0.6455754241173773,
      "grad_norm": 0.6488620541179162,
      "learning_rate": 0.0002949368743428612,
      "loss": 2.216,
      "step": 1408
    },
    {
      "epoch": 0.646033929390188,
      "grad_norm": 0.5834784206802418,
      "learning_rate": 0.0002942597431669593,
      "loss": 2.2338,
      "step": 1409
    },
    {
      "epoch": 0.6464924346629987,
      "grad_norm": 0.6117318397288609,
      "learning_rate": 0.0002935830659307924,
      "loss": 2.2996,
      "step": 1410
    },
    {
      "epoch": 0.6469509399358092,
      "grad_norm": 0.6111420685912039,
      "learning_rate": 0.0002929068441273629,
      "loss": 2.2585,
      "step": 1411
    },
    {
      "epoch": 0.6474094452086199,
      "grad_norm": 0.6066521227747818,
      "learning_rate": 0.0002922310792486681,
      "loss": 2.2603,
      "step": 1412
    },
    {
      "epoch": 0.6478679504814305,
      "grad_norm": 0.605458726867267,
      "learning_rate": 0.00029155577278569745,
      "loss": 2.2564,
      "step": 1413
    },
    {
      "epoch": 0.6483264557542412,
      "grad_norm": 0.5959771668720394,
      "learning_rate": 0.00029088092622842895,
      "loss": 2.2797,
      "step": 1414
    },
    {
      "epoch": 0.6487849610270519,
      "grad_norm": 0.5969569161962697,
      "learning_rate": 0.00029020654106582544,
      "loss": 2.3284,
      "step": 1415
    },
    {
      "epoch": 0.6492434662998624,
      "grad_norm": 0.6007749739740442,
      "learning_rate": 0.0002895326187858326,
      "loss": 2.245,
      "step": 1416
    },
    {
      "epoch": 0.6497019715726731,
      "grad_norm": 0.6369052899665152,
      "learning_rate": 0.00028885916087537377,
      "loss": 2.2822,
      "step": 1417
    },
    {
      "epoch": 0.6501604768454837,
      "grad_norm": 0.6228009657197286,
      "learning_rate": 0.00028818616882034877,
      "loss": 2.3449,
      "step": 1418
    },
    {
      "epoch": 0.6506189821182944,
      "grad_norm": 0.5973433614606901,
      "learning_rate": 0.0002875136441056286,
      "loss": 2.1946,
      "step": 1419
    },
    {
      "epoch": 0.651077487391105,
      "grad_norm": 0.614788470008304,
      "learning_rate": 0.000286841588215054,
      "loss": 2.343,
      "step": 1420
    },
    {
      "epoch": 0.6515359926639156,
      "grad_norm": 0.614226686383826,
      "learning_rate": 0.0002861700026314308,
      "loss": 2.3149,
      "step": 1421
    },
    {
      "epoch": 0.6519944979367263,
      "grad_norm": 0.5814975412800109,
      "learning_rate": 0.00028549888883652686,
      "loss": 2.2533,
      "step": 1422
    },
    {
      "epoch": 0.6524530032095369,
      "grad_norm": 0.5815895460879652,
      "learning_rate": 0.00028482824831107,
      "loss": 2.3236,
      "step": 1423
    },
    {
      "epoch": 0.6529115084823476,
      "grad_norm": 0.6139865323680115,
      "learning_rate": 0.000284158082534743,
      "loss": 2.2857,
      "step": 1424
    },
    {
      "epoch": 0.6533700137551581,
      "grad_norm": 0.5960851068029447,
      "learning_rate": 0.00028348839298618177,
      "loss": 2.234,
      "step": 1425
    },
    {
      "epoch": 0.6538285190279688,
      "grad_norm": 0.5801472997420483,
      "learning_rate": 0.0002828191811429709,
      "loss": 2.2693,
      "step": 1426
    },
    {
      "epoch": 0.6542870243007795,
      "grad_norm": 0.5868842964153667,
      "learning_rate": 0.00028215044848164164,
      "loss": 2.3067,
      "step": 1427
    },
    {
      "epoch": 0.6547455295735901,
      "grad_norm": 0.605863826720443,
      "learning_rate": 0.00028148219647766747,
      "loss": 2.3709,
      "step": 1428
    },
    {
      "epoch": 0.6552040348464008,
      "grad_norm": 0.5975336851209517,
      "learning_rate": 0.00028081442660546124,
      "loss": 2.1982,
      "step": 1429
    },
    {
      "epoch": 0.6556625401192113,
      "grad_norm": 0.6167337738932567,
      "learning_rate": 0.0002801471403383728,
      "loss": 2.2072,
      "step": 1430
    },
    {
      "epoch": 0.656121045392022,
      "grad_norm": 0.5859943684657374,
      "learning_rate": 0.00027948033914868415,
      "loss": 2.2479,
      "step": 1431
    },
    {
      "epoch": 0.6565795506648326,
      "grad_norm": 0.5868275994452039,
      "learning_rate": 0.00027881402450760775,
      "loss": 2.29,
      "step": 1432
    },
    {
      "epoch": 0.6570380559376433,
      "grad_norm": 0.6030670274098038,
      "learning_rate": 0.00027814819788528165,
      "loss": 2.3379,
      "step": 1433
    },
    {
      "epoch": 0.657496561210454,
      "grad_norm": 0.6137598234401276,
      "learning_rate": 0.00027748286075076836,
      "loss": 2.2158,
      "step": 1434
    },
    {
      "epoch": 0.6579550664832645,
      "grad_norm": 0.5944494947209069,
      "learning_rate": 0.00027681801457204937,
      "loss": 2.2473,
      "step": 1435
    },
    {
      "epoch": 0.6584135717560752,
      "grad_norm": 0.5952115589822597,
      "learning_rate": 0.00027615366081602306,
      "loss": 2.3642,
      "step": 1436
    },
    {
      "epoch": 0.6588720770288858,
      "grad_norm": 0.5917959274722687,
      "learning_rate": 0.0002754898009485021,
      "loss": 2.1822,
      "step": 1437
    },
    {
      "epoch": 0.6593305823016965,
      "grad_norm": 0.5569298617551313,
      "learning_rate": 0.0002748264364342085,
      "loss": 2.2304,
      "step": 1438
    },
    {
      "epoch": 0.6597890875745072,
      "grad_norm": 0.6140678627850595,
      "learning_rate": 0.00027416356873677204,
      "loss": 2.1727,
      "step": 1439
    },
    {
      "epoch": 0.6602475928473177,
      "grad_norm": 0.5986641894879463,
      "learning_rate": 0.0002735011993187258,
      "loss": 2.2371,
      "step": 1440
    },
    {
      "epoch": 0.6607060981201284,
      "grad_norm": 0.5878239448588432,
      "learning_rate": 0.0002728393296415042,
      "loss": 2.346,
      "step": 1441
    },
    {
      "epoch": 0.661164603392939,
      "grad_norm": 0.6100046523790648,
      "learning_rate": 0.00027217796116543817,
      "loss": 2.2658,
      "step": 1442
    },
    {
      "epoch": 0.6616231086657497,
      "grad_norm": 0.5876721701220576,
      "learning_rate": 0.0002715170953497532,
      "loss": 2.1841,
      "step": 1443
    },
    {
      "epoch": 0.6620816139385602,
      "grad_norm": 0.5692487836755844,
      "learning_rate": 0.00027085673365256614,
      "loss": 2.1882,
      "step": 1444
    },
    {
      "epoch": 0.6625401192113709,
      "grad_norm": 0.5896017999588609,
      "learning_rate": 0.00027019687753088075,
      "loss": 2.3193,
      "step": 1445
    },
    {
      "epoch": 0.6629986244841816,
      "grad_norm": 0.5916631241277617,
      "learning_rate": 0.00026953752844058597,
      "loss": 2.3047,
      "step": 1446
    },
    {
      "epoch": 0.6634571297569922,
      "grad_norm": 0.5951406106061533,
      "learning_rate": 0.0002688786878364516,
      "loss": 2.3031,
      "step": 1447
    },
    {
      "epoch": 0.6639156350298029,
      "grad_norm": 0.606063470177889,
      "learning_rate": 0.00026822035717212597,
      "loss": 2.3472,
      "step": 1448
    },
    {
      "epoch": 0.6643741403026134,
      "grad_norm": 0.5913820730489273,
      "learning_rate": 0.00026756253790013193,
      "loss": 2.1613,
      "step": 1449
    },
    {
      "epoch": 0.6648326455754241,
      "grad_norm": 0.5956187196249855,
      "learning_rate": 0.0002669052314718641,
      "loss": 2.2345,
      "step": 1450
    },
    {
      "epoch": 0.6652911508482348,
      "grad_norm": 0.5615735812707039,
      "learning_rate": 0.0002662484393375855,
      "loss": 2.175,
      "step": 1451
    },
    {
      "epoch": 0.6657496561210454,
      "grad_norm": 0.6476963484233683,
      "learning_rate": 0.00026559216294642446,
      "loss": 2.2273,
      "step": 1452
    },
    {
      "epoch": 0.6662081613938561,
      "grad_norm": 0.5957131163157886,
      "learning_rate": 0.0002649364037463718,
      "loss": 2.2945,
      "step": 1453
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.6066012337555573,
      "learning_rate": 0.0002642811631842764,
      "loss": 2.2663,
      "step": 1454
    },
    {
      "epoch": 0.6671251719394773,
      "grad_norm": 0.5856346260033517,
      "learning_rate": 0.0002636264427058439,
      "loss": 2.2314,
      "step": 1455
    },
    {
      "epoch": 0.667583677212288,
      "grad_norm": 0.5630611630875972,
      "learning_rate": 0.00026297224375563123,
      "loss": 2.2081,
      "step": 1456
    },
    {
      "epoch": 0.6680421824850986,
      "grad_norm": 0.5887621022024006,
      "learning_rate": 0.00026231856777704575,
      "loss": 2.2817,
      "step": 1457
    },
    {
      "epoch": 0.6685006877579092,
      "grad_norm": 0.5861146243420942,
      "learning_rate": 0.00026166541621234026,
      "loss": 2.2043,
      "step": 1458
    },
    {
      "epoch": 0.6689591930307198,
      "grad_norm": 0.6022965676084563,
      "learning_rate": 0.00026101279050261045,
      "loss": 2.141,
      "step": 1459
    },
    {
      "epoch": 0.6694176983035305,
      "grad_norm": 0.603276887749449,
      "learning_rate": 0.00026036069208779247,
      "loss": 2.2744,
      "step": 1460
    },
    {
      "epoch": 0.6698762035763411,
      "grad_norm": 0.5902348098435336,
      "learning_rate": 0.0002597091224066581,
      "loss": 2.2404,
      "step": 1461
    },
    {
      "epoch": 0.6703347088491518,
      "grad_norm": 0.6055683814453174,
      "learning_rate": 0.00025905808289681365,
      "loss": 2.1989,
      "step": 1462
    },
    {
      "epoch": 0.6707932141219624,
      "grad_norm": 0.6080100247492677,
      "learning_rate": 0.0002584075749946946,
      "loss": 2.3177,
      "step": 1463
    },
    {
      "epoch": 0.671251719394773,
      "grad_norm": 0.6107608978471609,
      "learning_rate": 0.00025775760013556424,
      "loss": 2.2898,
      "step": 1464
    },
    {
      "epoch": 0.6717102246675837,
      "grad_norm": 0.6057381378671784,
      "learning_rate": 0.0002571081597535095,
      "loss": 2.3365,
      "step": 1465
    },
    {
      "epoch": 0.6721687299403943,
      "grad_norm": 0.6080544403925939,
      "learning_rate": 0.00025645925528143776,
      "loss": 2.3375,
      "step": 1466
    },
    {
      "epoch": 0.672627235213205,
      "grad_norm": 0.6733093034787225,
      "learning_rate": 0.0002558108881510747,
      "loss": 2.3976,
      "step": 1467
    },
    {
      "epoch": 0.6730857404860155,
      "grad_norm": 0.6021390418632149,
      "learning_rate": 0.00025516305979295963,
      "loss": 2.1831,
      "step": 1468
    },
    {
      "epoch": 0.6735442457588262,
      "grad_norm": 0.6354719129868343,
      "learning_rate": 0.0002545157716364439,
      "loss": 2.3006,
      "step": 1469
    },
    {
      "epoch": 0.6740027510316369,
      "grad_norm": 0.622183638023457,
      "learning_rate": 0.00025386902510968624,
      "loss": 2.1679,
      "step": 1470
    },
    {
      "epoch": 0.6744612563044475,
      "grad_norm": 0.5942328616105141,
      "learning_rate": 0.00025322282163965095,
      "loss": 2.3683,
      "step": 1471
    },
    {
      "epoch": 0.6749197615772582,
      "grad_norm": 0.6023159079979674,
      "learning_rate": 0.00025257716265210384,
      "loss": 2.2441,
      "step": 1472
    },
    {
      "epoch": 0.6753782668500687,
      "grad_norm": 0.5911853888435984,
      "learning_rate": 0.0002519320495716091,
      "loss": 2.1993,
      "step": 1473
    },
    {
      "epoch": 0.6758367721228794,
      "grad_norm": 0.6049855551359601,
      "learning_rate": 0.00025128748382152716,
      "loss": 2.357,
      "step": 1474
    },
    {
      "epoch": 0.6762952773956901,
      "grad_norm": 0.5484776927358127,
      "learning_rate": 0.00025064346682401016,
      "loss": 2.2476,
      "step": 1475
    },
    {
      "epoch": 0.6767537826685007,
      "grad_norm": 0.6152395615283393,
      "learning_rate": 0.0002500000000000001,
      "loss": 2.2539,
      "step": 1476
    },
    {
      "epoch": 0.6772122879413113,
      "grad_norm": 0.5964442049784661,
      "learning_rate": 0.0002493570847692246,
      "loss": 2.2463,
      "step": 1477
    },
    {
      "epoch": 0.6776707932141219,
      "grad_norm": 0.6171081800773115,
      "learning_rate": 0.00024871472255019424,
      "loss": 2.2364,
      "step": 1478
    },
    {
      "epoch": 0.6781292984869326,
      "grad_norm": 0.5727206487890428,
      "learning_rate": 0.00024807291476019994,
      "loss": 2.2694,
      "step": 1479
    },
    {
      "epoch": 0.6785878037597433,
      "grad_norm": 0.5817343700054762,
      "learning_rate": 0.00024743166281530877,
      "loss": 2.2209,
      "step": 1480
    },
    {
      "epoch": 0.6790463090325539,
      "grad_norm": 0.592603955007661,
      "learning_rate": 0.000246790968130362,
      "loss": 2.3305,
      "step": 1481
    },
    {
      "epoch": 0.6795048143053645,
      "grad_norm": 0.591180361493702,
      "learning_rate": 0.0002461508321189706,
      "loss": 2.2339,
      "step": 1482
    },
    {
      "epoch": 0.6799633195781751,
      "grad_norm": 0.5580503532690656,
      "learning_rate": 0.00024551125619351385,
      "loss": 2.2365,
      "step": 1483
    },
    {
      "epoch": 0.6804218248509858,
      "grad_norm": 0.5883211072849961,
      "learning_rate": 0.00024487224176513453,
      "loss": 2.2472,
      "step": 1484
    },
    {
      "epoch": 0.6808803301237965,
      "grad_norm": 0.571955260010129,
      "learning_rate": 0.0002442337902437365,
      "loss": 2.3093,
      "step": 1485
    },
    {
      "epoch": 0.6813388353966071,
      "grad_norm": 0.5762933459578785,
      "learning_rate": 0.0002435959030379824,
      "loss": 2.2112,
      "step": 1486
    },
    {
      "epoch": 0.6817973406694177,
      "grad_norm": 0.581107593429835,
      "learning_rate": 0.00024295858155528888,
      "loss": 2.2488,
      "step": 1487
    },
    {
      "epoch": 0.6822558459422283,
      "grad_norm": 0.6007863510895612,
      "learning_rate": 0.00024232182720182523,
      "loss": 2.2799,
      "step": 1488
    },
    {
      "epoch": 0.682714351215039,
      "grad_norm": 0.6118009213096199,
      "learning_rate": 0.00024168564138250855,
      "loss": 2.272,
      "step": 1489
    },
    {
      "epoch": 0.6831728564878496,
      "grad_norm": 0.6230971971787183,
      "learning_rate": 0.00024105002550100246,
      "loss": 2.337,
      "step": 1490
    },
    {
      "epoch": 0.6836313617606602,
      "grad_norm": 0.5937001548960426,
      "learning_rate": 0.00024041498095971254,
      "loss": 2.3085,
      "step": 1491
    },
    {
      "epoch": 0.6840898670334709,
      "grad_norm": 0.6347658923497004,
      "learning_rate": 0.0002397805091597835,
      "loss": 2.3384,
      "step": 1492
    },
    {
      "epoch": 0.6845483723062815,
      "grad_norm": 0.6355148713742135,
      "learning_rate": 0.0002391466115010973,
      "loss": 2.268,
      "step": 1493
    },
    {
      "epoch": 0.6850068775790922,
      "grad_norm": 0.5893280585685471,
      "learning_rate": 0.00023851328938226808,
      "loss": 2.2874,
      "step": 1494
    },
    {
      "epoch": 0.6854653828519028,
      "grad_norm": 0.6031108160876557,
      "learning_rate": 0.00023788054420064109,
      "loss": 2.2029,
      "step": 1495
    },
    {
      "epoch": 0.6859238881247134,
      "grad_norm": 0.5966465394217353,
      "learning_rate": 0.00023724837735228773,
      "loss": 2.2401,
      "step": 1496
    },
    {
      "epoch": 0.686382393397524,
      "grad_norm": 0.5939180000594488,
      "learning_rate": 0.00023661679023200422,
      "loss": 2.2795,
      "step": 1497
    },
    {
      "epoch": 0.6868408986703347,
      "grad_norm": 0.5994828686582088,
      "learning_rate": 0.00023598578423330714,
      "loss": 2.2134,
      "step": 1498
    },
    {
      "epoch": 0.6872994039431454,
      "grad_norm": 0.6177630392078973,
      "learning_rate": 0.00023535536074843083,
      "loss": 2.2122,
      "step": 1499
    },
    {
      "epoch": 0.687757909215956,
      "grad_norm": 0.601341988195703,
      "learning_rate": 0.00023472552116832502,
      "loss": 2.316,
      "step": 1500
    },
    {
      "epoch": 0.6882164144887666,
      "grad_norm": 0.5910632983054217,
      "learning_rate": 0.0002340962668826503,
      "loss": 2.2796,
      "step": 1501
    },
    {
      "epoch": 0.6886749197615772,
      "grad_norm": 0.6195893260456684,
      "learning_rate": 0.00023346759927977663,
      "loss": 2.3269,
      "step": 1502
    },
    {
      "epoch": 0.6891334250343879,
      "grad_norm": 0.5711002570737531,
      "learning_rate": 0.0002328395197467789,
      "loss": 2.259,
      "step": 1503
    },
    {
      "epoch": 0.6895919303071986,
      "grad_norm": 0.6038086798475198,
      "learning_rate": 0.00023221202966943515,
      "loss": 2.3153,
      "step": 1504
    },
    {
      "epoch": 0.6900504355800092,
      "grad_norm": 0.6274135841331396,
      "learning_rate": 0.0002315851304322223,
      "loss": 2.2883,
      "step": 1505
    },
    {
      "epoch": 0.6905089408528198,
      "grad_norm": 0.6061999576001827,
      "learning_rate": 0.0002309588234183137,
      "loss": 2.3207,
      "step": 1506
    },
    {
      "epoch": 0.6909674461256304,
      "grad_norm": 0.6275517758100625,
      "learning_rate": 0.00023033311000957653,
      "loss": 2.2038,
      "step": 1507
    },
    {
      "epoch": 0.6914259513984411,
      "grad_norm": 0.608071700428632,
      "learning_rate": 0.00022970799158656758,
      "loss": 2.3089,
      "step": 1508
    },
    {
      "epoch": 0.6918844566712518,
      "grad_norm": 0.6004871029767817,
      "learning_rate": 0.0002290834695285316,
      "loss": 2.2461,
      "step": 1509
    },
    {
      "epoch": 0.6923429619440623,
      "grad_norm": 0.6325180402873332,
      "learning_rate": 0.00022845954521339678,
      "loss": 2.3723,
      "step": 1510
    },
    {
      "epoch": 0.692801467216873,
      "grad_norm": 0.5860816212077076,
      "learning_rate": 0.0002278362200177732,
      "loss": 2.2586,
      "step": 1511
    },
    {
      "epoch": 0.6932599724896836,
      "grad_norm": 0.5735671421171183,
      "learning_rate": 0.00022721349531694852,
      "loss": 2.2271,
      "step": 1512
    },
    {
      "epoch": 0.6937184777624943,
      "grad_norm": 0.6267871219130468,
      "learning_rate": 0.0002265913724848855,
      "loss": 2.2902,
      "step": 1513
    },
    {
      "epoch": 0.694176983035305,
      "grad_norm": 0.5857354908649729,
      "learning_rate": 0.00022596985289421946,
      "loss": 2.2442,
      "step": 1514
    },
    {
      "epoch": 0.6946354883081155,
      "grad_norm": 0.5586692058427855,
      "learning_rate": 0.00022534893791625405,
      "loss": 2.2163,
      "step": 1515
    },
    {
      "epoch": 0.6950939935809262,
      "grad_norm": 0.606047621040995,
      "learning_rate": 0.00022472862892095968,
      "loss": 2.2743,
      "step": 1516
    },
    {
      "epoch": 0.6955524988537368,
      "grad_norm": 0.5833123652542777,
      "learning_rate": 0.00022410892727696896,
      "loss": 2.3741,
      "step": 1517
    },
    {
      "epoch": 0.6960110041265475,
      "grad_norm": 0.5725442814945333,
      "learning_rate": 0.0002234898343515751,
      "loss": 2.2822,
      "step": 1518
    },
    {
      "epoch": 0.6964695093993581,
      "grad_norm": 0.5816841639209663,
      "learning_rate": 0.00022287135151072792,
      "loss": 2.2276,
      "step": 1519
    },
    {
      "epoch": 0.6969280146721687,
      "grad_norm": 0.5902758302195801,
      "learning_rate": 0.00022225348011903096,
      "loss": 2.236,
      "step": 1520
    },
    {
      "epoch": 0.6973865199449794,
      "grad_norm": 0.6107364905102786,
      "learning_rate": 0.0002216362215397393,
      "loss": 2.1907,
      "step": 1521
    },
    {
      "epoch": 0.69784502521779,
      "grad_norm": 0.6080692760925578,
      "learning_rate": 0.00022101957713475522,
      "loss": 2.2487,
      "step": 1522
    },
    {
      "epoch": 0.6983035304906007,
      "grad_norm": 0.5675773979418313,
      "learning_rate": 0.00022040354826462666,
      "loss": 2.2395,
      "step": 1523
    },
    {
      "epoch": 0.6987620357634112,
      "grad_norm": 0.5893341975572952,
      "learning_rate": 0.0002197881362885426,
      "loss": 2.2335,
      "step": 1524
    },
    {
      "epoch": 0.6992205410362219,
      "grad_norm": 0.5826148089021594,
      "learning_rate": 0.0002191733425643318,
      "loss": 2.1526,
      "step": 1525
    },
    {
      "epoch": 0.6996790463090325,
      "grad_norm": 0.6274755613330035,
      "learning_rate": 0.00021855916844845826,
      "loss": 2.2319,
      "step": 1526
    },
    {
      "epoch": 0.7001375515818432,
      "grad_norm": 0.5780835479135575,
      "learning_rate": 0.00021794561529601898,
      "loss": 2.207,
      "step": 1527
    },
    {
      "epoch": 0.7005960568546539,
      "grad_norm": 0.6064633384619383,
      "learning_rate": 0.00021733268446074138,
      "loss": 2.2191,
      "step": 1528
    },
    {
      "epoch": 0.7010545621274644,
      "grad_norm": 0.6094327078409917,
      "learning_rate": 0.00021672037729497917,
      "loss": 2.2006,
      "step": 1529
    },
    {
      "epoch": 0.7015130674002751,
      "grad_norm": 0.6141115483621562,
      "learning_rate": 0.0002161086951497106,
      "loss": 2.3513,
      "step": 1530
    },
    {
      "epoch": 0.7019715726730857,
      "grad_norm": 0.5905783185253577,
      "learning_rate": 0.00021549763937453442,
      "loss": 2.3486,
      "step": 1531
    },
    {
      "epoch": 0.7024300779458964,
      "grad_norm": 0.5995002486230298,
      "learning_rate": 0.00021488721131766736,
      "loss": 2.3187,
      "step": 1532
    },
    {
      "epoch": 0.7028885832187071,
      "grad_norm": 0.6212177360442912,
      "learning_rate": 0.00021427741232594183,
      "loss": 2.2419,
      "step": 1533
    },
    {
      "epoch": 0.7033470884915176,
      "grad_norm": 0.5977352047231121,
      "learning_rate": 0.0002136682437448013,
      "loss": 2.2848,
      "step": 1534
    },
    {
      "epoch": 0.7038055937643283,
      "grad_norm": 0.6160254293682352,
      "learning_rate": 0.0002130597069182994,
      "loss": 2.1924,
      "step": 1535
    },
    {
      "epoch": 0.7042640990371389,
      "grad_norm": 0.5811525905498351,
      "learning_rate": 0.0002124518031890948,
      "loss": 2.2341,
      "step": 1536
    },
    {
      "epoch": 0.7047226043099496,
      "grad_norm": 0.5752607971747861,
      "learning_rate": 0.0002118445338984502,
      "loss": 2.2827,
      "step": 1537
    },
    {
      "epoch": 0.7051811095827603,
      "grad_norm": 0.6038911187684723,
      "learning_rate": 0.00021123790038622808,
      "loss": 2.3084,
      "step": 1538
    },
    {
      "epoch": 0.7056396148555708,
      "grad_norm": 0.5878750180789896,
      "learning_rate": 0.0002106319039908879,
      "loss": 2.3898,
      "step": 1539
    },
    {
      "epoch": 0.7060981201283815,
      "grad_norm": 0.5744384013839868,
      "learning_rate": 0.00021002654604948412,
      "loss": 2.267,
      "step": 1540
    },
    {
      "epoch": 0.7065566254011921,
      "grad_norm": 0.5808580684983876,
      "learning_rate": 0.00020942182789766172,
      "loss": 2.2406,
      "step": 1541
    },
    {
      "epoch": 0.7070151306740028,
      "grad_norm": 0.5910909385116244,
      "learning_rate": 0.00020881775086965492,
      "loss": 2.2224,
      "step": 1542
    },
    {
      "epoch": 0.7074736359468133,
      "grad_norm": 0.6053074695952517,
      "learning_rate": 0.00020821431629828246,
      "loss": 2.2614,
      "step": 1543
    },
    {
      "epoch": 0.707932141219624,
      "grad_norm": 0.5969545246785297,
      "learning_rate": 0.00020761152551494643,
      "loss": 2.2206,
      "step": 1544
    },
    {
      "epoch": 0.7083906464924347,
      "grad_norm": 0.5763709510857603,
      "learning_rate": 0.00020700937984962798,
      "loss": 2.2436,
      "step": 1545
    },
    {
      "epoch": 0.7088491517652453,
      "grad_norm": 0.6106364246804429,
      "learning_rate": 0.0002064078806308848,
      "loss": 2.3408,
      "step": 1546
    },
    {
      "epoch": 0.709307657038056,
      "grad_norm": 0.6113393869368154,
      "learning_rate": 0.00020580702918584882,
      "loss": 2.3551,
      "step": 1547
    },
    {
      "epoch": 0.7097661623108665,
      "grad_norm": 0.5883889256807447,
      "learning_rate": 0.000205206826840222,
      "loss": 2.1296,
      "step": 1548
    },
    {
      "epoch": 0.7102246675836772,
      "grad_norm": 0.6052869435348642,
      "learning_rate": 0.0002046072749182751,
      "loss": 2.3407,
      "step": 1549
    },
    {
      "epoch": 0.7106831728564879,
      "grad_norm": 0.5887814118494323,
      "learning_rate": 0.00020400837474284273,
      "loss": 2.2706,
      "step": 1550
    },
    {
      "epoch": 0.7111416781292985,
      "grad_norm": 0.6046197295254196,
      "learning_rate": 0.0002034101276353224,
      "loss": 2.2843,
      "step": 1551
    },
    {
      "epoch": 0.7116001834021092,
      "grad_norm": 0.5971348503115808,
      "learning_rate": 0.00020281253491567027,
      "loss": 2.3221,
      "step": 1552
    },
    {
      "epoch": 0.7120586886749197,
      "grad_norm": 0.5757041592095432,
      "learning_rate": 0.0002022155979023984,
      "loss": 2.2513,
      "step": 1553
    },
    {
      "epoch": 0.7125171939477304,
      "grad_norm": 0.5890330485139088,
      "learning_rate": 0.000201619317912573,
      "loss": 2.2728,
      "step": 1554
    },
    {
      "epoch": 0.712975699220541,
      "grad_norm": 0.5838598288775285,
      "learning_rate": 0.00020102369626180962,
      "loss": 2.2374,
      "step": 1555
    },
    {
      "epoch": 0.7134342044933517,
      "grad_norm": 0.5693311582711504,
      "learning_rate": 0.0002004287342642721,
      "loss": 2.2425,
      "step": 1556
    },
    {
      "epoch": 0.7138927097661623,
      "grad_norm": 0.604653490283188,
      "learning_rate": 0.00019983443323266824,
      "loss": 2.3779,
      "step": 1557
    },
    {
      "epoch": 0.7143512150389729,
      "grad_norm": 0.5828792191185371,
      "learning_rate": 0.00019924079447824805,
      "loss": 2.2484,
      "step": 1558
    },
    {
      "epoch": 0.7148097203117836,
      "grad_norm": 0.5886911586575414,
      "learning_rate": 0.00019864781931079977,
      "loss": 2.1911,
      "step": 1559
    },
    {
      "epoch": 0.7152682255845942,
      "grad_norm": 0.5614531353551595,
      "learning_rate": 0.00019805550903864773,
      "loss": 2.2768,
      "step": 1560
    },
    {
      "epoch": 0.7157267308574049,
      "grad_norm": 0.5800270425736807,
      "learning_rate": 0.00019746386496864948,
      "loss": 2.2428,
      "step": 1561
    },
    {
      "epoch": 0.7161852361302155,
      "grad_norm": 0.5892080663511231,
      "learning_rate": 0.00019687288840619226,
      "loss": 2.2218,
      "step": 1562
    },
    {
      "epoch": 0.7166437414030261,
      "grad_norm": 0.6132941698350487,
      "learning_rate": 0.0001962825806551911,
      "loss": 2.2025,
      "step": 1563
    },
    {
      "epoch": 0.7171022466758368,
      "grad_norm": 0.5993710741935635,
      "learning_rate": 0.0001956929430180846,
      "loss": 2.1702,
      "step": 1564
    },
    {
      "epoch": 0.7175607519486474,
      "grad_norm": 0.6040115972239735,
      "learning_rate": 0.00019510397679583374,
      "loss": 2.2671,
      "step": 1565
    },
    {
      "epoch": 0.7180192572214581,
      "grad_norm": 0.611114047475353,
      "learning_rate": 0.0001945156832879174,
      "loss": 2.1904,
      "step": 1566
    },
    {
      "epoch": 0.7184777624942686,
      "grad_norm": 0.6013539086832868,
      "learning_rate": 0.00019392806379233036,
      "loss": 2.2733,
      "step": 1567
    },
    {
      "epoch": 0.7189362677670793,
      "grad_norm": 0.6112361384481714,
      "learning_rate": 0.00019334111960558065,
      "loss": 2.146,
      "step": 1568
    },
    {
      "epoch": 0.71939477303989,
      "grad_norm": 0.5949362905609141,
      "learning_rate": 0.00019275485202268573,
      "loss": 2.3356,
      "step": 1569
    },
    {
      "epoch": 0.7198532783127006,
      "grad_norm": 0.6732902002844915,
      "learning_rate": 0.00019216926233717085,
      "loss": 2.1577,
      "step": 1570
    },
    {
      "epoch": 0.7203117835855112,
      "grad_norm": 0.5789930977711667,
      "learning_rate": 0.00019158435184106498,
      "loss": 2.2781,
      "step": 1571
    },
    {
      "epoch": 0.7207702888583218,
      "grad_norm": 0.6177219387758388,
      "learning_rate": 0.00019100012182489905,
      "loss": 2.2854,
      "step": 1572
    },
    {
      "epoch": 0.7212287941311325,
      "grad_norm": 0.5679584119926986,
      "learning_rate": 0.00019041657357770226,
      "loss": 2.2633,
      "step": 1573
    },
    {
      "epoch": 0.7216872994039432,
      "grad_norm": 0.5614076896248428,
      "learning_rate": 0.00018983370838699943,
      "loss": 2.2523,
      "step": 1574
    },
    {
      "epoch": 0.7221458046767538,
      "grad_norm": 0.6024041735861229,
      "learning_rate": 0.00018925152753880892,
      "loss": 2.1904,
      "step": 1575
    },
    {
      "epoch": 0.7226043099495644,
      "grad_norm": 0.5681449260870615,
      "learning_rate": 0.00018867003231763847,
      "loss": 2.3079,
      "step": 1576
    },
    {
      "epoch": 0.723062815222375,
      "grad_norm": 0.5775637320050182,
      "learning_rate": 0.00018808922400648375,
      "loss": 2.2348,
      "step": 1577
    },
    {
      "epoch": 0.7235213204951857,
      "grad_norm": 0.6038026047327376,
      "learning_rate": 0.00018750910388682428,
      "loss": 2.2536,
      "step": 1578
    },
    {
      "epoch": 0.7239798257679964,
      "grad_norm": 0.5906946423959513,
      "learning_rate": 0.00018692967323862125,
      "loss": 2.1908,
      "step": 1579
    },
    {
      "epoch": 0.724438331040807,
      "grad_norm": 0.6104659758619472,
      "learning_rate": 0.00018635093334031517,
      "loss": 2.2,
      "step": 1580
    },
    {
      "epoch": 0.7248968363136176,
      "grad_norm": 0.5835014332044178,
      "learning_rate": 0.00018577288546882165,
      "loss": 2.189,
      "step": 1581
    },
    {
      "epoch": 0.7253553415864282,
      "grad_norm": 0.6401915824517028,
      "learning_rate": 0.00018519553089953023,
      "loss": 2.2544,
      "step": 1582
    },
    {
      "epoch": 0.7258138468592389,
      "grad_norm": 0.6150088461228589,
      "learning_rate": 0.0001846188709063001,
      "loss": 2.3239,
      "step": 1583
    },
    {
      "epoch": 0.7262723521320495,
      "grad_norm": 0.6462763173935988,
      "learning_rate": 0.00018404290676145857,
      "loss": 2.3409,
      "step": 1584
    },
    {
      "epoch": 0.7267308574048602,
      "grad_norm": 0.6236484837399895,
      "learning_rate": 0.00018346763973579722,
      "loss": 2.2098,
      "step": 1585
    },
    {
      "epoch": 0.7271893626776708,
      "grad_norm": 0.6335441814447681,
      "learning_rate": 0.00018289307109856939,
      "loss": 2.1908,
      "step": 1586
    },
    {
      "epoch": 0.7276478679504814,
      "grad_norm": 0.5982890670689843,
      "learning_rate": 0.0001823192021174882,
      "loss": 2.2609,
      "step": 1587
    },
    {
      "epoch": 0.7281063732232921,
      "grad_norm": 0.5866709329455105,
      "learning_rate": 0.0001817460340587223,
      "loss": 2.284,
      "step": 1588
    },
    {
      "epoch": 0.7285648784961027,
      "grad_norm": 0.5633268188858578,
      "learning_rate": 0.00018117356818689445,
      "loss": 2.1453,
      "step": 1589
    },
    {
      "epoch": 0.7290233837689133,
      "grad_norm": 0.5965680496044397,
      "learning_rate": 0.00018060180576507756,
      "loss": 2.2689,
      "step": 1590
    },
    {
      "epoch": 0.729481889041724,
      "grad_norm": 0.6379342843571532,
      "learning_rate": 0.00018003074805479313,
      "loss": 2.3275,
      "step": 1591
    },
    {
      "epoch": 0.7299403943145346,
      "grad_norm": 0.606469573072782,
      "learning_rate": 0.00017946039631600724,
      "loss": 2.2055,
      "step": 1592
    },
    {
      "epoch": 0.7303988995873453,
      "grad_norm": 0.5898208260917053,
      "learning_rate": 0.00017889075180712837,
      "loss": 2.2824,
      "step": 1593
    },
    {
      "epoch": 0.7308574048601559,
      "grad_norm": 0.6160669521763493,
      "learning_rate": 0.00017832181578500512,
      "loss": 2.213,
      "step": 1594
    },
    {
      "epoch": 0.7313159101329665,
      "grad_norm": 0.5843811244328115,
      "learning_rate": 0.0001777535895049221,
      "loss": 2.1442,
      "step": 1595
    },
    {
      "epoch": 0.7317744154057771,
      "grad_norm": 0.589283425195204,
      "learning_rate": 0.0001771860742205988,
      "loss": 2.2921,
      "step": 1596
    },
    {
      "epoch": 0.7322329206785878,
      "grad_norm": 0.6191686718756209,
      "learning_rate": 0.00017661927118418525,
      "loss": 2.2005,
      "step": 1597
    },
    {
      "epoch": 0.7326914259513985,
      "grad_norm": 0.6293880680202208,
      "learning_rate": 0.00017605318164626066,
      "loss": 2.2404,
      "step": 1598
    },
    {
      "epoch": 0.7331499312242091,
      "grad_norm": 0.6191822322976878,
      "learning_rate": 0.00017548780685582949,
      "loss": 2.3319,
      "step": 1599
    },
    {
      "epoch": 0.7336084364970197,
      "grad_norm": 0.6215310182093967,
      "learning_rate": 0.00017492314806031922,
      "loss": 2.4053,
      "step": 1600
    },
    {
      "epoch": 0.7340669417698303,
      "grad_norm": 0.5906612913253897,
      "learning_rate": 0.00017435920650557806,
      "loss": 2.1769,
      "step": 1601
    },
    {
      "epoch": 0.734525447042641,
      "grad_norm": 0.588869028321832,
      "learning_rate": 0.00017379598343587112,
      "loss": 2.281,
      "step": 1602
    },
    {
      "epoch": 0.7349839523154517,
      "grad_norm": 0.5885501868494035,
      "learning_rate": 0.00017323348009387878,
      "loss": 2.3152,
      "step": 1603
    },
    {
      "epoch": 0.7354424575882622,
      "grad_norm": 0.5630023978102053,
      "learning_rate": 0.0001726716977206929,
      "loss": 2.2326,
      "step": 1604
    },
    {
      "epoch": 0.7359009628610729,
      "grad_norm": 0.5862600733525215,
      "learning_rate": 0.00017211063755581525,
      "loss": 2.2108,
      "step": 1605
    },
    {
      "epoch": 0.7363594681338835,
      "grad_norm": 0.5780687202565734,
      "learning_rate": 0.0001715503008371536,
      "loss": 2.26,
      "step": 1606
    },
    {
      "epoch": 0.7368179734066942,
      "grad_norm": 0.5769519031559501,
      "learning_rate": 0.0001709906888010196,
      "loss": 2.2575,
      "step": 1607
    },
    {
      "epoch": 0.7372764786795049,
      "grad_norm": 0.578678263262617,
      "learning_rate": 0.00017043180268212638,
      "loss": 2.2291,
      "step": 1608
    },
    {
      "epoch": 0.7377349839523154,
      "grad_norm": 0.6100816176938784,
      "learning_rate": 0.00016987364371358481,
      "loss": 2.2471,
      "step": 1609
    },
    {
      "epoch": 0.7381934892251261,
      "grad_norm": 0.6072071592312761,
      "learning_rate": 0.00016931621312690214,
      "loss": 2.306,
      "step": 1610
    },
    {
      "epoch": 0.7386519944979367,
      "grad_norm": 0.6043362557443366,
      "learning_rate": 0.00016875951215197777,
      "loss": 2.2948,
      "step": 1611
    },
    {
      "epoch": 0.7391104997707474,
      "grad_norm": 0.6229703693761873,
      "learning_rate": 0.00016820354201710214,
      "loss": 2.2156,
      "step": 1612
    },
    {
      "epoch": 0.739569005043558,
      "grad_norm": 0.5904726103321991,
      "learning_rate": 0.00016764830394895203,
      "loss": 2.2811,
      "step": 1613
    },
    {
      "epoch": 0.7400275103163686,
      "grad_norm": 0.6042727765407727,
      "learning_rate": 0.00016709379917259027,
      "loss": 2.264,
      "step": 1614
    },
    {
      "epoch": 0.7404860155891793,
      "grad_norm": 0.5851677844215986,
      "learning_rate": 0.00016654002891146091,
      "loss": 2.2128,
      "step": 1615
    },
    {
      "epoch": 0.7409445208619899,
      "grad_norm": 0.6144437997314076,
      "learning_rate": 0.00016598699438738764,
      "loss": 2.2183,
      "step": 1616
    },
    {
      "epoch": 0.7414030261348006,
      "grad_norm": 0.6101341082871893,
      "learning_rate": 0.00016543469682057105,
      "loss": 2.3135,
      "step": 1617
    },
    {
      "epoch": 0.7418615314076112,
      "grad_norm": 0.5828220465944232,
      "learning_rate": 0.00016488313742958526,
      "loss": 2.2752,
      "step": 1618
    },
    {
      "epoch": 0.7423200366804218,
      "grad_norm": 0.5928749931022954,
      "learning_rate": 0.00016433231743137646,
      "loss": 2.2698,
      "step": 1619
    },
    {
      "epoch": 0.7427785419532325,
      "grad_norm": 0.6172056528368892,
      "learning_rate": 0.0001637822380412584,
      "loss": 2.2246,
      "step": 1620
    },
    {
      "epoch": 0.7432370472260431,
      "grad_norm": 0.5961520597840547,
      "learning_rate": 0.00016323290047291195,
      "loss": 2.1843,
      "step": 1621
    },
    {
      "epoch": 0.7436955524988538,
      "grad_norm": 0.6163159474024973,
      "learning_rate": 0.0001626843059383803,
      "loss": 2.3094,
      "step": 1622
    },
    {
      "epoch": 0.7441540577716643,
      "grad_norm": 0.5904872030112631,
      "learning_rate": 0.00016213645564806752,
      "loss": 2.3333,
      "step": 1623
    },
    {
      "epoch": 0.744612563044475,
      "grad_norm": 0.6161989344623334,
      "learning_rate": 0.0001615893508107359,
      "loss": 2.1245,
      "step": 1624
    },
    {
      "epoch": 0.7450710683172856,
      "grad_norm": 0.5652402582011519,
      "learning_rate": 0.00016104299263350252,
      "loss": 2.2955,
      "step": 1625
    },
    {
      "epoch": 0.7455295735900963,
      "grad_norm": 0.559491097573223,
      "learning_rate": 0.00016049738232183758,
      "loss": 2.206,
      "step": 1626
    },
    {
      "epoch": 0.745988078862907,
      "grad_norm": 0.5688810819620383,
      "learning_rate": 0.0001599525210795606,
      "loss": 2.1378,
      "step": 1627
    },
    {
      "epoch": 0.7464465841357175,
      "grad_norm": 0.5652074059858746,
      "learning_rate": 0.00015940841010883889,
      "loss": 2.2516,
      "step": 1628
    },
    {
      "epoch": 0.7469050894085282,
      "grad_norm": 0.6141921114959948,
      "learning_rate": 0.00015886505061018413,
      "loss": 2.2447,
      "step": 1629
    },
    {
      "epoch": 0.7473635946813388,
      "grad_norm": 0.5805900775340352,
      "learning_rate": 0.0001583224437824498,
      "loss": 2.2506,
      "step": 1630
    },
    {
      "epoch": 0.7478220999541495,
      "grad_norm": 0.6425869789462878,
      "learning_rate": 0.0001577805908228293,
      "loss": 2.3039,
      "step": 1631
    },
    {
      "epoch": 0.7482806052269602,
      "grad_norm": 0.6132789172519977,
      "learning_rate": 0.00015723949292685191,
      "loss": 2.1922,
      "step": 1632
    },
    {
      "epoch": 0.7487391104997707,
      "grad_norm": 0.6127319382486002,
      "learning_rate": 0.0001566991512883818,
      "loss": 2.1919,
      "step": 1633
    },
    {
      "epoch": 0.7491976157725814,
      "grad_norm": 0.60177975883761,
      "learning_rate": 0.00015615956709961378,
      "loss": 2.373,
      "step": 1634
    },
    {
      "epoch": 0.749656121045392,
      "grad_norm": 0.5923493437312194,
      "learning_rate": 0.00015562074155107215,
      "loss": 2.2296,
      "step": 1635
    },
    {
      "epoch": 0.7501146263182027,
      "grad_norm": 0.635786875679161,
      "learning_rate": 0.0001550826758316068,
      "loss": 2.2905,
      "step": 1636
    },
    {
      "epoch": 0.7505731315910132,
      "grad_norm": 0.6206425101727069,
      "learning_rate": 0.00015454537112839122,
      "loss": 2.333,
      "step": 1637
    },
    {
      "epoch": 0.7510316368638239,
      "grad_norm": 0.6175423898863789,
      "learning_rate": 0.00015400882862692033,
      "loss": 2.3065,
      "step": 1638
    },
    {
      "epoch": 0.7514901421366346,
      "grad_norm": 0.5801592580052137,
      "learning_rate": 0.00015347304951100665,
      "loss": 2.2819,
      "step": 1639
    },
    {
      "epoch": 0.7519486474094452,
      "grad_norm": 0.5847534786875181,
      "learning_rate": 0.00015293803496277907,
      "loss": 2.3323,
      "step": 1640
    },
    {
      "epoch": 0.7524071526822559,
      "grad_norm": 0.6413469984876033,
      "learning_rate": 0.00015240378616267886,
      "loss": 2.304,
      "step": 1641
    },
    {
      "epoch": 0.7528656579550664,
      "grad_norm": 0.5789379114090617,
      "learning_rate": 0.00015187030428945843,
      "loss": 2.1407,
      "step": 1642
    },
    {
      "epoch": 0.7533241632278771,
      "grad_norm": 0.6100241630784641,
      "learning_rate": 0.0001513375905201776,
      "loss": 2.2932,
      "step": 1643
    },
    {
      "epoch": 0.7537826685006878,
      "grad_norm": 0.5759611129689683,
      "learning_rate": 0.00015080564603020142,
      "loss": 2.2395,
      "step": 1644
    },
    {
      "epoch": 0.7542411737734984,
      "grad_norm": 0.6212039461198502,
      "learning_rate": 0.0001502744719931982,
      "loss": 2.1521,
      "step": 1645
    },
    {
      "epoch": 0.7546996790463091,
      "grad_norm": 0.5814922250670004,
      "learning_rate": 0.00014974406958113558,
      "loss": 2.2731,
      "step": 1646
    },
    {
      "epoch": 0.7551581843191196,
      "grad_norm": 0.6527104653751928,
      "learning_rate": 0.00014921443996427947,
      "loss": 2.3028,
      "step": 1647
    },
    {
      "epoch": 0.7556166895919303,
      "grad_norm": 0.6174234929533297,
      "learning_rate": 0.0001486855843111901,
      "loss": 2.3198,
      "step": 1648
    },
    {
      "epoch": 0.756075194864741,
      "grad_norm": 0.5782546834502894,
      "learning_rate": 0.0001481575037887201,
      "loss": 2.3087,
      "step": 1649
    },
    {
      "epoch": 0.7565337001375516,
      "grad_norm": 0.5724240529656839,
      "learning_rate": 0.00014763019956201253,
      "loss": 2.3076,
      "step": 1650
    },
    {
      "epoch": 0.7569922054103623,
      "grad_norm": 0.6009637391105693,
      "learning_rate": 0.0001471036727944966,
      "loss": 2.2052,
      "step": 1651
    },
    {
      "epoch": 0.7574507106831728,
      "grad_norm": 0.5757679142949901,
      "learning_rate": 0.0001465779246478872,
      "loss": 2.1758,
      "step": 1652
    },
    {
      "epoch": 0.7579092159559835,
      "grad_norm": 0.5807615305468253,
      "learning_rate": 0.00014605295628218045,
      "loss": 2.1711,
      "step": 1653
    },
    {
      "epoch": 0.7583677212287941,
      "grad_norm": 0.596112566938182,
      "learning_rate": 0.0001455287688556527,
      "loss": 2.2573,
      "step": 1654
    },
    {
      "epoch": 0.7588262265016048,
      "grad_norm": 0.5876353920613575,
      "learning_rate": 0.00014500536352485673,
      "loss": 2.1985,
      "step": 1655
    },
    {
      "epoch": 0.7592847317744154,
      "grad_norm": 0.5624002227800754,
      "learning_rate": 0.00014448274144461965,
      "loss": 2.2259,
      "step": 1656
    },
    {
      "epoch": 0.759743237047226,
      "grad_norm": 0.571229808487486,
      "learning_rate": 0.00014396090376804112,
      "loss": 2.3,
      "step": 1657
    },
    {
      "epoch": 0.7602017423200367,
      "grad_norm": 0.5675142619854818,
      "learning_rate": 0.00014343985164648926,
      "loss": 2.1618,
      "step": 1658
    },
    {
      "epoch": 0.7606602475928473,
      "grad_norm": 0.5665168856605434,
      "learning_rate": 0.00014291958622959973,
      "loss": 2.4127,
      "step": 1659
    },
    {
      "epoch": 0.761118752865658,
      "grad_norm": 0.6068042825870393,
      "learning_rate": 0.00014240010866527176,
      "loss": 2.2472,
      "step": 1660
    },
    {
      "epoch": 0.7615772581384685,
      "grad_norm": 0.585092521447271,
      "learning_rate": 0.00014188142009966686,
      "loss": 2.2549,
      "step": 1661
    },
    {
      "epoch": 0.7620357634112792,
      "grad_norm": 0.6002849214600731,
      "learning_rate": 0.0001413635216772053,
      "loss": 2.1777,
      "step": 1662
    },
    {
      "epoch": 0.7624942686840899,
      "grad_norm": 0.5731352649206304,
      "learning_rate": 0.000140846414540564,
      "loss": 2.2789,
      "step": 1663
    },
    {
      "epoch": 0.7629527739569005,
      "grad_norm": 0.5771414881717902,
      "learning_rate": 0.00014033009983067452,
      "loss": 2.2569,
      "step": 1664
    },
    {
      "epoch": 0.7634112792297112,
      "grad_norm": 0.5593223375478731,
      "learning_rate": 0.00013981457868671927,
      "loss": 2.1983,
      "step": 1665
    },
    {
      "epoch": 0.7638697845025217,
      "grad_norm": 0.5821263855890129,
      "learning_rate": 0.0001392998522461305,
      "loss": 2.2646,
      "step": 1666
    },
    {
      "epoch": 0.7643282897753324,
      "grad_norm": 0.6143882843927165,
      "learning_rate": 0.00013878592164458635,
      "loss": 2.3248,
      "step": 1667
    },
    {
      "epoch": 0.7647867950481431,
      "grad_norm": 0.5930779487415347,
      "learning_rate": 0.00013827278801600978,
      "loss": 2.2252,
      "step": 1668
    },
    {
      "epoch": 0.7652453003209537,
      "grad_norm": 0.5728219987559503,
      "learning_rate": 0.0001377604524925647,
      "loss": 2.2326,
      "step": 1669
    },
    {
      "epoch": 0.7657038055937643,
      "grad_norm": 0.5834090919155499,
      "learning_rate": 0.00013724891620465424,
      "loss": 2.1837,
      "step": 1670
    },
    {
      "epoch": 0.7661623108665749,
      "grad_norm": 0.587786270282634,
      "learning_rate": 0.0001367381802809185,
      "loss": 2.2691,
      "step": 1671
    },
    {
      "epoch": 0.7666208161393856,
      "grad_norm": 0.6118883202363383,
      "learning_rate": 0.00013622824584823113,
      "loss": 2.2119,
      "step": 1672
    },
    {
      "epoch": 0.7670793214121963,
      "grad_norm": 0.5991293504180226,
      "learning_rate": 0.00013571911403169795,
      "loss": 2.2614,
      "step": 1673
    },
    {
      "epoch": 0.7675378266850069,
      "grad_norm": 0.6164156505987611,
      "learning_rate": 0.0001352107859546533,
      "loss": 2.2712,
      "step": 1674
    },
    {
      "epoch": 0.7679963319578175,
      "grad_norm": 0.5799783358331563,
      "learning_rate": 0.00013470326273865886,
      "loss": 2.2527,
      "step": 1675
    },
    {
      "epoch": 0.7684548372306281,
      "grad_norm": 0.5894932345129332,
      "learning_rate": 0.00013419654550349985,
      "loss": 2.2089,
      "step": 1676
    },
    {
      "epoch": 0.7689133425034388,
      "grad_norm": 0.5874873951548287,
      "learning_rate": 0.00013369063536718346,
      "loss": 2.1781,
      "step": 1677
    },
    {
      "epoch": 0.7693718477762495,
      "grad_norm": 0.5965211614484064,
      "learning_rate": 0.00013318553344593632,
      "loss": 2.2203,
      "step": 1678
    },
    {
      "epoch": 0.7698303530490601,
      "grad_norm": 0.5921688458979005,
      "learning_rate": 0.00013268124085420136,
      "loss": 2.218,
      "step": 1679
    },
    {
      "epoch": 0.7702888583218707,
      "grad_norm": 0.5859921082739302,
      "learning_rate": 0.0001321777587046364,
      "loss": 2.2317,
      "step": 1680
    },
    {
      "epoch": 0.7707473635946813,
      "grad_norm": 0.605478695385914,
      "learning_rate": 0.00013167508810811059,
      "loss": 2.215,
      "step": 1681
    },
    {
      "epoch": 0.771205868867492,
      "grad_norm": 0.6081833891799491,
      "learning_rate": 0.0001311732301737029,
      "loss": 2.3017,
      "step": 1682
    },
    {
      "epoch": 0.7716643741403026,
      "grad_norm": 0.601062448726836,
      "learning_rate": 0.0001306721860086991,
      "loss": 2.2519,
      "step": 1683
    },
    {
      "epoch": 0.7721228794131133,
      "grad_norm": 0.5732578551334039,
      "learning_rate": 0.00013017195671858928,
      "loss": 2.1683,
      "step": 1684
    },
    {
      "epoch": 0.7725813846859239,
      "grad_norm": 0.5691068912506582,
      "learning_rate": 0.0001296725434070661,
      "loss": 2.1912,
      "step": 1685
    },
    {
      "epoch": 0.7730398899587345,
      "grad_norm": 0.5718894475192622,
      "learning_rate": 0.00012917394717602121,
      "loss": 2.2352,
      "step": 1686
    },
    {
      "epoch": 0.7734983952315452,
      "grad_norm": 0.6031345946288902,
      "learning_rate": 0.00012867616912554426,
      "loss": 2.3291,
      "step": 1687
    },
    {
      "epoch": 0.7739569005043558,
      "grad_norm": 0.6090952552259253,
      "learning_rate": 0.00012817921035391882,
      "loss": 2.2158,
      "step": 1688
    },
    {
      "epoch": 0.7744154057771664,
      "grad_norm": 0.6254785620811277,
      "learning_rate": 0.00012768307195762168,
      "loss": 2.2749,
      "step": 1689
    },
    {
      "epoch": 0.774873911049977,
      "grad_norm": 0.6145078352251454,
      "learning_rate": 0.00012718775503131908,
      "loss": 2.2298,
      "step": 1690
    },
    {
      "epoch": 0.7753324163227877,
      "grad_norm": 0.5725352396513371,
      "learning_rate": 0.0001266932606678646,
      "loss": 2.1886,
      "step": 1691
    },
    {
      "epoch": 0.7757909215955984,
      "grad_norm": 0.625770568475586,
      "learning_rate": 0.00012619958995829756,
      "loss": 2.2447,
      "step": 1692
    },
    {
      "epoch": 0.776249426868409,
      "grad_norm": 0.6441576566285787,
      "learning_rate": 0.0001257067439918394,
      "loss": 2.185,
      "step": 1693
    },
    {
      "epoch": 0.7767079321412196,
      "grad_norm": 0.5824305667974592,
      "learning_rate": 0.00012521472385589234,
      "loss": 2.2697,
      "step": 1694
    },
    {
      "epoch": 0.7771664374140302,
      "grad_norm": 0.5883377354609788,
      "learning_rate": 0.00012472353063603626,
      "loss": 2.2152,
      "step": 1695
    },
    {
      "epoch": 0.7776249426868409,
      "grad_norm": 0.6061810891699849,
      "learning_rate": 0.0001242331654160263,
      "loss": 2.2017,
      "step": 1696
    },
    {
      "epoch": 0.7780834479596516,
      "grad_norm": 0.6111938651720559,
      "learning_rate": 0.0001237436292777914,
      "loss": 2.3119,
      "step": 1697
    },
    {
      "epoch": 0.7785419532324622,
      "grad_norm": 0.5860573810885317,
      "learning_rate": 0.00012325492330143061,
      "loss": 2.2787,
      "step": 1698
    },
    {
      "epoch": 0.7790004585052728,
      "grad_norm": 0.6163342254507269,
      "learning_rate": 0.00012276704856521175,
      "loss": 2.3297,
      "step": 1699
    },
    {
      "epoch": 0.7794589637780834,
      "grad_norm": 0.5927194340665577,
      "learning_rate": 0.00012228000614556816,
      "loss": 2.2005,
      "step": 1700
    },
    {
      "epoch": 0.7799174690508941,
      "grad_norm": 0.5781561275870598,
      "learning_rate": 0.00012179379711709738,
      "loss": 2.1057,
      "step": 1701
    },
    {
      "epoch": 0.7803759743237048,
      "grad_norm": 0.566929778164398,
      "learning_rate": 0.0001213084225525577,
      "loss": 2.3828,
      "step": 1702
    },
    {
      "epoch": 0.7808344795965153,
      "grad_norm": 0.5843361202204226,
      "learning_rate": 0.00012082388352286627,
      "loss": 2.262,
      "step": 1703
    },
    {
      "epoch": 0.781292984869326,
      "grad_norm": 0.5801783807310541,
      "learning_rate": 0.00012034018109709716,
      "loss": 2.2474,
      "step": 1704
    },
    {
      "epoch": 0.7817514901421366,
      "grad_norm": 0.5789005025895754,
      "learning_rate": 0.00011985731634247809,
      "loss": 2.3036,
      "step": 1705
    },
    {
      "epoch": 0.7822099954149473,
      "grad_norm": 0.5653141189086808,
      "learning_rate": 0.00011937529032438904,
      "loss": 2.204,
      "step": 1706
    },
    {
      "epoch": 0.782668500687758,
      "grad_norm": 0.569773618340014,
      "learning_rate": 0.00011889410410635887,
      "loss": 2.2206,
      "step": 1707
    },
    {
      "epoch": 0.7831270059605685,
      "grad_norm": 0.5630260611593052,
      "learning_rate": 0.0001184137587500641,
      "loss": 2.2831,
      "step": 1708
    },
    {
      "epoch": 0.7835855112333792,
      "grad_norm": 0.6020613407994116,
      "learning_rate": 0.00011793425531532564,
      "loss": 2.1892,
      "step": 1709
    },
    {
      "epoch": 0.7840440165061898,
      "grad_norm": 0.5788522093235321,
      "learning_rate": 0.00011745559486010671,
      "loss": 2.3636,
      "step": 1710
    },
    {
      "epoch": 0.7845025217790005,
      "grad_norm": 0.5932649401392818,
      "learning_rate": 0.00011697777844051105,
      "loss": 2.1691,
      "step": 1711
    },
    {
      "epoch": 0.7849610270518111,
      "grad_norm": 0.579460361344711,
      "learning_rate": 0.00011650080711077964,
      "loss": 2.259,
      "step": 1712
    },
    {
      "epoch": 0.7854195323246217,
      "grad_norm": 0.5964796128177372,
      "learning_rate": 0.00011602468192328936,
      "loss": 2.2205,
      "step": 1713
    },
    {
      "epoch": 0.7858780375974324,
      "grad_norm": 0.5976994656520636,
      "learning_rate": 0.00011554940392854973,
      "loss": 2.2477,
      "step": 1714
    },
    {
      "epoch": 0.786336542870243,
      "grad_norm": 0.618487320747336,
      "learning_rate": 0.00011507497417520146,
      "loss": 2.2983,
      "step": 1715
    },
    {
      "epoch": 0.7867950481430537,
      "grad_norm": 0.5993633027929348,
      "learning_rate": 0.00011460139371001339,
      "loss": 2.2896,
      "step": 1716
    },
    {
      "epoch": 0.7872535534158643,
      "grad_norm": 0.5665505971907712,
      "learning_rate": 0.00011412866357788049,
      "loss": 2.2184,
      "step": 1717
    },
    {
      "epoch": 0.7877120586886749,
      "grad_norm": 0.5931514455501788,
      "learning_rate": 0.00011365678482182207,
      "loss": 2.2784,
      "step": 1718
    },
    {
      "epoch": 0.7881705639614855,
      "grad_norm": 0.591652301962881,
      "learning_rate": 0.0001131857584829783,
      "loss": 2.2783,
      "step": 1719
    },
    {
      "epoch": 0.7886290692342962,
      "grad_norm": 0.6103971872443241,
      "learning_rate": 0.0001127155856006093,
      "loss": 2.2188,
      "step": 1720
    },
    {
      "epoch": 0.7890875745071069,
      "grad_norm": 0.5625799079064402,
      "learning_rate": 0.00011224626721209141,
      "loss": 2.2288,
      "step": 1721
    },
    {
      "epoch": 0.7895460797799174,
      "grad_norm": 0.6089860011436763,
      "learning_rate": 0.0001117778043529164,
      "loss": 2.2889,
      "step": 1722
    },
    {
      "epoch": 0.7900045850527281,
      "grad_norm": 0.5936348469237241,
      "learning_rate": 0.0001113101980566879,
      "loss": 2.2801,
      "step": 1723
    },
    {
      "epoch": 0.7904630903255387,
      "grad_norm": 0.6024908248582513,
      "learning_rate": 0.00011084344935511958,
      "loss": 2.1724,
      "step": 1724
    },
    {
      "epoch": 0.7909215955983494,
      "grad_norm": 0.5746413385685656,
      "learning_rate": 0.00011037755927803345,
      "loss": 2.2419,
      "step": 1725
    },
    {
      "epoch": 0.7913801008711601,
      "grad_norm": 0.5704872343057894,
      "learning_rate": 0.00010991252885335651,
      "loss": 2.2691,
      "step": 1726
    },
    {
      "epoch": 0.7918386061439706,
      "grad_norm": 0.6015524655807989,
      "learning_rate": 0.00010944835910711958,
      "loss": 2.2162,
      "step": 1727
    },
    {
      "epoch": 0.7922971114167813,
      "grad_norm": 0.6061444227162935,
      "learning_rate": 0.00010898505106345396,
      "loss": 2.2958,
      "step": 1728
    },
    {
      "epoch": 0.7927556166895919,
      "grad_norm": 0.5964605745427394,
      "learning_rate": 0.00010852260574459022,
      "loss": 2.1846,
      "step": 1729
    },
    {
      "epoch": 0.7932141219624026,
      "grad_norm": 0.6413946309747729,
      "learning_rate": 0.00010806102417085512,
      "loss": 2.2708,
      "step": 1730
    },
    {
      "epoch": 0.7936726272352133,
      "grad_norm": 0.6009262230178308,
      "learning_rate": 0.00010760030736066951,
      "loss": 2.3299,
      "step": 1731
    },
    {
      "epoch": 0.7941311325080238,
      "grad_norm": 0.5906738614507803,
      "learning_rate": 0.00010714045633054687,
      "loss": 2.225,
      "step": 1732
    },
    {
      "epoch": 0.7945896377808345,
      "grad_norm": 0.65467061054999,
      "learning_rate": 0.00010668147209508971,
      "loss": 2.1824,
      "step": 1733
    },
    {
      "epoch": 0.7950481430536451,
      "grad_norm": 0.6024883549751594,
      "learning_rate": 0.00010622335566698877,
      "loss": 2.3789,
      "step": 1734
    },
    {
      "epoch": 0.7955066483264558,
      "grad_norm": 0.5943761381492105,
      "learning_rate": 0.00010576610805701942,
      "loss": 2.2744,
      "step": 1735
    },
    {
      "epoch": 0.7959651535992663,
      "grad_norm": 0.6107311783904819,
      "learning_rate": 0.00010530973027404073,
      "loss": 2.224,
      "step": 1736
    },
    {
      "epoch": 0.796423658872077,
      "grad_norm": 0.6259158058668063,
      "learning_rate": 0.00010485422332499212,
      "loss": 2.2595,
      "step": 1737
    },
    {
      "epoch": 0.7968821641448877,
      "grad_norm": 0.6015657944538735,
      "learning_rate": 0.00010439958821489165,
      "loss": 2.2786,
      "step": 1738
    },
    {
      "epoch": 0.7973406694176983,
      "grad_norm": 0.5965685498012718,
      "learning_rate": 0.00010394582594683428,
      "loss": 2.3703,
      "step": 1739
    },
    {
      "epoch": 0.797799174690509,
      "grad_norm": 0.5722606178827904,
      "learning_rate": 0.0001034929375219884,
      "loss": 2.208,
      "step": 1740
    },
    {
      "epoch": 0.7982576799633195,
      "grad_norm": 0.5864496665929338,
      "learning_rate": 0.00010304092393959514,
      "loss": 2.1645,
      "step": 1741
    },
    {
      "epoch": 0.7987161852361302,
      "grad_norm": 0.5602584989960661,
      "learning_rate": 0.00010258978619696468,
      "loss": 2.2559,
      "step": 1742
    },
    {
      "epoch": 0.7991746905089409,
      "grad_norm": 0.5797008842492135,
      "learning_rate": 0.00010213952528947551,
      "loss": 2.2992,
      "step": 1743
    },
    {
      "epoch": 0.7996331957817515,
      "grad_norm": 0.5870903567771596,
      "learning_rate": 0.00010169014221057089,
      "loss": 2.1827,
      "step": 1744
    },
    {
      "epoch": 0.8000917010545622,
      "grad_norm": 0.6052062046042449,
      "learning_rate": 0.00010124163795175734,
      "loss": 2.237,
      "step": 1745
    },
    {
      "epoch": 0.8005502063273727,
      "grad_norm": 0.5974553891099214,
      "learning_rate": 0.00010079401350260287,
      "loss": 2.2764,
      "step": 1746
    },
    {
      "epoch": 0.8010087116001834,
      "grad_norm": 0.5839979522918485,
      "learning_rate": 0.00010034726985073362,
      "loss": 2.2896,
      "step": 1747
    },
    {
      "epoch": 0.801467216872994,
      "grad_norm": 0.5819671986346204,
      "learning_rate": 9.9901407981833e-05,
      "loss": 2.3206,
      "step": 1748
    },
    {
      "epoch": 0.8019257221458047,
      "grad_norm": 0.5790797204535041,
      "learning_rate": 9.94564288796384e-05,
      "loss": 2.1839,
      "step": 1749
    },
    {
      "epoch": 0.8023842274186153,
      "grad_norm": 0.5729045948152937,
      "learning_rate": 9.901233352593953e-05,
      "loss": 2.2536,
      "step": 1750
    },
    {
      "epoch": 0.8028427326914259,
      "grad_norm": 0.5991585042363958,
      "learning_rate": 9.856912290057668e-05,
      "loss": 2.2557,
      "step": 1751
    },
    {
      "epoch": 0.8033012379642366,
      "grad_norm": 0.6110474727603158,
      "learning_rate": 9.812679798143748e-05,
      "loss": 2.2539,
      "step": 1752
    },
    {
      "epoch": 0.8037597432370472,
      "grad_norm": 0.596991668933827,
      "learning_rate": 9.768535974445586e-05,
      "loss": 2.2759,
      "step": 1753
    },
    {
      "epoch": 0.8042182485098579,
      "grad_norm": 0.5880164759943804,
      "learning_rate": 9.724480916360906e-05,
      "loss": 2.2183,
      "step": 1754
    },
    {
      "epoch": 0.8046767537826685,
      "grad_norm": 0.6150172154973159,
      "learning_rate": 9.68051472109162e-05,
      "loss": 2.316,
      "step": 1755
    },
    {
      "epoch": 0.8051352590554791,
      "grad_norm": 0.5911022099966403,
      "learning_rate": 9.636637485643529e-05,
      "loss": 2.1779,
      "step": 1756
    },
    {
      "epoch": 0.8055937643282898,
      "grad_norm": 0.5928139737042365,
      "learning_rate": 9.592849306826174e-05,
      "loss": 2.2321,
      "step": 1757
    },
    {
      "epoch": 0.8060522696011004,
      "grad_norm": 0.6089273578672667,
      "learning_rate": 9.549150281252633e-05,
      "loss": 2.3158,
      "step": 1758
    },
    {
      "epoch": 0.8065107748739111,
      "grad_norm": 0.6080608958523754,
      "learning_rate": 9.505540505339223e-05,
      "loss": 2.15,
      "step": 1759
    },
    {
      "epoch": 0.8069692801467216,
      "grad_norm": 0.6275996891248407,
      "learning_rate": 9.4620200753054e-05,
      "loss": 2.2766,
      "step": 1760
    },
    {
      "epoch": 0.8074277854195323,
      "grad_norm": 0.6000410795118912,
      "learning_rate": 9.418589087173441e-05,
      "loss": 2.2134,
      "step": 1761
    },
    {
      "epoch": 0.807886290692343,
      "grad_norm": 0.5926100840269162,
      "learning_rate": 9.375247636768325e-05,
      "loss": 2.171,
      "step": 1762
    },
    {
      "epoch": 0.8083447959651536,
      "grad_norm": 0.6119492538149177,
      "learning_rate": 9.331995819717443e-05,
      "loss": 2.2596,
      "step": 1763
    },
    {
      "epoch": 0.8088033012379643,
      "grad_norm": 0.5966756175424495,
      "learning_rate": 9.288833731450419e-05,
      "loss": 2.2065,
      "step": 1764
    },
    {
      "epoch": 0.8092618065107748,
      "grad_norm": 0.5988184260851516,
      "learning_rate": 9.245761467198948e-05,
      "loss": 2.2147,
      "step": 1765
    },
    {
      "epoch": 0.8097203117835855,
      "grad_norm": 0.6144642359385452,
      "learning_rate": 9.20277912199648e-05,
      "loss": 2.16,
      "step": 1766
    },
    {
      "epoch": 0.8101788170563962,
      "grad_norm": 0.61940904099104,
      "learning_rate": 9.159886790678123e-05,
      "loss": 2.2541,
      "step": 1767
    },
    {
      "epoch": 0.8106373223292068,
      "grad_norm": 0.5726184788013966,
      "learning_rate": 9.11708456788033e-05,
      "loss": 2.1899,
      "step": 1768
    },
    {
      "epoch": 0.8110958276020174,
      "grad_norm": 0.5855061061588548,
      "learning_rate": 9.074372548040793e-05,
      "loss": 2.1887,
      "step": 1769
    },
    {
      "epoch": 0.811554332874828,
      "grad_norm": 0.5895810103592706,
      "learning_rate": 9.031750825398145e-05,
      "loss": 2.257,
      "step": 1770
    },
    {
      "epoch": 0.8120128381476387,
      "grad_norm": 0.5784874245523122,
      "learning_rate": 8.98921949399179e-05,
      "loss": 2.2593,
      "step": 1771
    },
    {
      "epoch": 0.8124713434204494,
      "grad_norm": 0.6057581168737523,
      "learning_rate": 8.94677864766173e-05,
      "loss": 2.1865,
      "step": 1772
    },
    {
      "epoch": 0.81292984869326,
      "grad_norm": 0.6458725789554516,
      "learning_rate": 8.904428380048269e-05,
      "loss": 2.2684,
      "step": 1773
    },
    {
      "epoch": 0.8133883539660706,
      "grad_norm": 0.6126537248861282,
      "learning_rate": 8.862168784591929e-05,
      "loss": 2.2514,
      "step": 1774
    },
    {
      "epoch": 0.8138468592388812,
      "grad_norm": 0.5961931715407234,
      "learning_rate": 8.819999954533115e-05,
      "loss": 2.1961,
      "step": 1775
    },
    {
      "epoch": 0.8143053645116919,
      "grad_norm": 0.5656255995573705,
      "learning_rate": 8.777921982911996e-05,
      "loss": 2.1254,
      "step": 1776
    },
    {
      "epoch": 0.8147638697845025,
      "grad_norm": 0.5905872844515515,
      "learning_rate": 8.735934962568253e-05,
      "loss": 2.2072,
      "step": 1777
    },
    {
      "epoch": 0.8152223750573132,
      "grad_norm": 0.5905546870528204,
      "learning_rate": 8.694038986140945e-05,
      "loss": 2.2604,
      "step": 1778
    },
    {
      "epoch": 0.8156808803301238,
      "grad_norm": 0.5981951010015996,
      "learning_rate": 8.652234146068206e-05,
      "loss": 2.1919,
      "step": 1779
    },
    {
      "epoch": 0.8161393856029344,
      "grad_norm": 0.6136834316361379,
      "learning_rate": 8.610520534587086e-05,
      "loss": 2.2008,
      "step": 1780
    },
    {
      "epoch": 0.8165978908757451,
      "grad_norm": 0.6147170175473826,
      "learning_rate": 8.568898243733397e-05,
      "loss": 2.1991,
      "step": 1781
    },
    {
      "epoch": 0.8170563961485557,
      "grad_norm": 0.6157560782811423,
      "learning_rate": 8.527367365341409e-05,
      "loss": 2.3099,
      "step": 1782
    },
    {
      "epoch": 0.8175149014213663,
      "grad_norm": 0.5970094452638811,
      "learning_rate": 8.485927991043757e-05,
      "loss": 2.3268,
      "step": 1783
    },
    {
      "epoch": 0.817973406694177,
      "grad_norm": 0.6128466817693672,
      "learning_rate": 8.444580212271125e-05,
      "loss": 2.3122,
      "step": 1784
    },
    {
      "epoch": 0.8184319119669876,
      "grad_norm": 0.5800233844815733,
      "learning_rate": 8.403324120252159e-05,
      "loss": 2.2799,
      "step": 1785
    },
    {
      "epoch": 0.8188904172397983,
      "grad_norm": 0.5808008757592229,
      "learning_rate": 8.362159806013175e-05,
      "loss": 2.1713,
      "step": 1786
    },
    {
      "epoch": 0.8193489225126089,
      "grad_norm": 0.575759813338383,
      "learning_rate": 8.321087360377988e-05,
      "loss": 2.1943,
      "step": 1787
    },
    {
      "epoch": 0.8198074277854195,
      "grad_norm": 0.6045340812287094,
      "learning_rate": 8.280106873967752e-05,
      "loss": 2.3479,
      "step": 1788
    },
    {
      "epoch": 0.8202659330582301,
      "grad_norm": 0.5880629045571827,
      "learning_rate": 8.239218437200679e-05,
      "loss": 2.235,
      "step": 1789
    },
    {
      "epoch": 0.8207244383310408,
      "grad_norm": 0.5831726219066014,
      "learning_rate": 8.198422140291939e-05,
      "loss": 2.1803,
      "step": 1790
    },
    {
      "epoch": 0.8211829436038515,
      "grad_norm": 0.5811639360997785,
      "learning_rate": 8.157718073253351e-05,
      "loss": 2.319,
      "step": 1791
    },
    {
      "epoch": 0.8216414488766621,
      "grad_norm": 0.5924039305012204,
      "learning_rate": 8.117106325893287e-05,
      "loss": 2.1864,
      "step": 1792
    },
    {
      "epoch": 0.8220999541494727,
      "grad_norm": 0.6316292567419228,
      "learning_rate": 8.076586987816404e-05,
      "loss": 2.2382,
      "step": 1793
    },
    {
      "epoch": 0.8225584594222833,
      "grad_norm": 0.6002922689229926,
      "learning_rate": 8.036160148423449e-05,
      "loss": 2.2258,
      "step": 1794
    },
    {
      "epoch": 0.823016964695094,
      "grad_norm": 0.5732967237142554,
      "learning_rate": 7.995825896911141e-05,
      "loss": 2.2306,
      "step": 1795
    },
    {
      "epoch": 0.8234754699679047,
      "grad_norm": 0.5863605948609267,
      "learning_rate": 7.955584322271853e-05,
      "loss": 2.2657,
      "step": 1796
    },
    {
      "epoch": 0.8239339752407153,
      "grad_norm": 0.5605429243092263,
      "learning_rate": 7.915435513293523e-05,
      "loss": 2.2345,
      "step": 1797
    },
    {
      "epoch": 0.8243924805135259,
      "grad_norm": 0.588794908181906,
      "learning_rate": 7.875379558559387e-05,
      "loss": 2.0931,
      "step": 1798
    },
    {
      "epoch": 0.8248509857863365,
      "grad_norm": 0.5997583017658218,
      "learning_rate": 7.835416546447838e-05,
      "loss": 2.2549,
      "step": 1799
    },
    {
      "epoch": 0.8253094910591472,
      "grad_norm": 0.6015537369890236,
      "learning_rate": 7.795546565132167e-05,
      "loss": 2.2356,
      "step": 1800
    },
    {
      "epoch": 0.8257679963319579,
      "grad_norm": 0.6205926167112517,
      "learning_rate": 7.755769702580412e-05,
      "loss": 2.2421,
      "step": 1801
    },
    {
      "epoch": 0.8262265016047684,
      "grad_norm": 0.582905594502652,
      "learning_rate": 7.716086046555193e-05,
      "loss": 2.2332,
      "step": 1802
    },
    {
      "epoch": 0.8266850068775791,
      "grad_norm": 0.5949990081276397,
      "learning_rate": 7.676495684613432e-05,
      "loss": 2.2428,
      "step": 1803
    },
    {
      "epoch": 0.8271435121503897,
      "grad_norm": 0.5958876955091899,
      "learning_rate": 7.636998704106252e-05,
      "loss": 2.1448,
      "step": 1804
    },
    {
      "epoch": 0.8276020174232004,
      "grad_norm": 0.5801372341924701,
      "learning_rate": 7.597595192178702e-05,
      "loss": 2.1743,
      "step": 1805
    },
    {
      "epoch": 0.828060522696011,
      "grad_norm": 0.598898366968248,
      "learning_rate": 7.558285235769646e-05,
      "loss": 2.2329,
      "step": 1806
    },
    {
      "epoch": 0.8285190279688216,
      "grad_norm": 0.5882937062969237,
      "learning_rate": 7.519068921611494e-05,
      "loss": 2.299,
      "step": 1807
    },
    {
      "epoch": 0.8289775332416323,
      "grad_norm": 0.5978151940628044,
      "learning_rate": 7.479946336230047e-05,
      "loss": 2.2347,
      "step": 1808
    },
    {
      "epoch": 0.8294360385144429,
      "grad_norm": 0.6317996645450951,
      "learning_rate": 7.440917565944349e-05,
      "loss": 2.3043,
      "step": 1809
    },
    {
      "epoch": 0.8298945437872536,
      "grad_norm": 0.5915589464107841,
      "learning_rate": 7.4019826968664e-05,
      "loss": 2.2509,
      "step": 1810
    },
    {
      "epoch": 0.8303530490600642,
      "grad_norm": 0.6176114781554587,
      "learning_rate": 7.363141814901053e-05,
      "loss": 2.2592,
      "step": 1811
    },
    {
      "epoch": 0.8308115543328748,
      "grad_norm": 0.5643648844601986,
      "learning_rate": 7.32439500574577e-05,
      "loss": 2.2502,
      "step": 1812
    },
    {
      "epoch": 0.8312700596056855,
      "grad_norm": 0.5725942101556069,
      "learning_rate": 7.285742354890473e-05,
      "loss": 2.2125,
      "step": 1813
    },
    {
      "epoch": 0.8317285648784961,
      "grad_norm": 0.6082688139653779,
      "learning_rate": 7.247183947617325e-05,
      "loss": 2.2084,
      "step": 1814
    },
    {
      "epoch": 0.8321870701513068,
      "grad_norm": 0.5572454156629482,
      "learning_rate": 7.20871986900053e-05,
      "loss": 2.1808,
      "step": 1815
    },
    {
      "epoch": 0.8326455754241173,
      "grad_norm": 0.5855385055156181,
      "learning_rate": 7.170350203906218e-05,
      "loss": 2.1708,
      "step": 1816
    },
    {
      "epoch": 0.833104080696928,
      "grad_norm": 0.5741851476079071,
      "learning_rate": 7.132075036992158e-05,
      "loss": 2.2462,
      "step": 1817
    },
    {
      "epoch": 0.8335625859697386,
      "grad_norm": 0.6205110913828076,
      "learning_rate": 7.093894452707666e-05,
      "loss": 2.1897,
      "step": 1818
    },
    {
      "epoch": 0.8340210912425493,
      "grad_norm": 0.5811044145290595,
      "learning_rate": 7.055808535293334e-05,
      "loss": 2.1653,
      "step": 1819
    },
    {
      "epoch": 0.83447959651536,
      "grad_norm": 0.5710420891820771,
      "learning_rate": 7.017817368780888e-05,
      "loss": 2.2645,
      "step": 1820
    },
    {
      "epoch": 0.8349381017881705,
      "grad_norm": 0.5724905582207316,
      "learning_rate": 6.979921036993042e-05,
      "loss": 2.226,
      "step": 1821
    },
    {
      "epoch": 0.8353966070609812,
      "grad_norm": 0.6052603652870521,
      "learning_rate": 6.942119623543202e-05,
      "loss": 2.3444,
      "step": 1822
    },
    {
      "epoch": 0.8358551123337918,
      "grad_norm": 0.5968494641022785,
      "learning_rate": 6.904413211835414e-05,
      "loss": 2.2404,
      "step": 1823
    },
    {
      "epoch": 0.8363136176066025,
      "grad_norm": 0.622061266470281,
      "learning_rate": 6.866801885064056e-05,
      "loss": 2.2359,
      "step": 1824
    },
    {
      "epoch": 0.8367721228794132,
      "grad_norm": 0.5940370085613376,
      "learning_rate": 6.829285726213769e-05,
      "loss": 2.1781,
      "step": 1825
    },
    {
      "epoch": 0.8372306281522237,
      "grad_norm": 0.594540659491894,
      "learning_rate": 6.79186481805918e-05,
      "loss": 2.234,
      "step": 1826
    },
    {
      "epoch": 0.8376891334250344,
      "grad_norm": 0.595442893787211,
      "learning_rate": 6.754539243164754e-05,
      "loss": 2.2219,
      "step": 1827
    },
    {
      "epoch": 0.838147638697845,
      "grad_norm": 0.5982025972227841,
      "learning_rate": 6.717309083884654e-05,
      "loss": 2.2626,
      "step": 1828
    },
    {
      "epoch": 0.8386061439706557,
      "grad_norm": 0.5962666554162032,
      "learning_rate": 6.680174422362468e-05,
      "loss": 2.1205,
      "step": 1829
    },
    {
      "epoch": 0.8390646492434664,
      "grad_norm": 0.6085270592423891,
      "learning_rate": 6.643135340531136e-05,
      "loss": 2.2676,
      "step": 1830
    },
    {
      "epoch": 0.8395231545162769,
      "grad_norm": 0.5888628099031301,
      "learning_rate": 6.606191920112664e-05,
      "loss": 2.3261,
      "step": 1831
    },
    {
      "epoch": 0.8399816597890876,
      "grad_norm": 0.6059022107962754,
      "learning_rate": 6.569344242618036e-05,
      "loss": 2.1734,
      "step": 1832
    },
    {
      "epoch": 0.8404401650618982,
      "grad_norm": 0.6284882987584378,
      "learning_rate": 6.532592389346958e-05,
      "loss": 2.3147,
      "step": 1833
    },
    {
      "epoch": 0.8408986703347089,
      "grad_norm": 0.5887604713563445,
      "learning_rate": 6.495936441387713e-05,
      "loss": 2.2119,
      "step": 1834
    },
    {
      "epoch": 0.8413571756075194,
      "grad_norm": 0.5961887788558476,
      "learning_rate": 6.459376479617013e-05,
      "loss": 2.2243,
      "step": 1835
    },
    {
      "epoch": 0.8418156808803301,
      "grad_norm": 0.5938420715968847,
      "learning_rate": 6.422912584699752e-05,
      "loss": 2.2765,
      "step": 1836
    },
    {
      "epoch": 0.8422741861531408,
      "grad_norm": 0.5746065939645355,
      "learning_rate": 6.386544837088904e-05,
      "loss": 2.1927,
      "step": 1837
    },
    {
      "epoch": 0.8427326914259514,
      "grad_norm": 0.5900508960164187,
      "learning_rate": 6.350273317025251e-05,
      "loss": 2.1477,
      "step": 1838
    },
    {
      "epoch": 0.8431911966987621,
      "grad_norm": 0.5944248907442472,
      "learning_rate": 6.314098104537324e-05,
      "loss": 2.2792,
      "step": 1839
    },
    {
      "epoch": 0.8436497019715726,
      "grad_norm": 0.6062944857385089,
      "learning_rate": 6.278019279441122e-05,
      "loss": 2.2525,
      "step": 1840
    },
    {
      "epoch": 0.8441082072443833,
      "grad_norm": 0.6117565463680034,
      "learning_rate": 6.242036921339972e-05,
      "loss": 2.2493,
      "step": 1841
    },
    {
      "epoch": 0.844566712517194,
      "grad_norm": 0.6154745620345589,
      "learning_rate": 6.206151109624402e-05,
      "loss": 2.2132,
      "step": 1842
    },
    {
      "epoch": 0.8450252177900046,
      "grad_norm": 0.5896911175423853,
      "learning_rate": 6.170361923471868e-05,
      "loss": 2.178,
      "step": 1843
    },
    {
      "epoch": 0.8454837230628153,
      "grad_norm": 0.620370646928066,
      "learning_rate": 6.134669441846691e-05,
      "loss": 2.2337,
      "step": 1844
    },
    {
      "epoch": 0.8459422283356258,
      "grad_norm": 0.5806099315494508,
      "learning_rate": 6.099073743499772e-05,
      "loss": 2.2494,
      "step": 1845
    },
    {
      "epoch": 0.8464007336084365,
      "grad_norm": 0.6123521789148821,
      "learning_rate": 6.063574906968511e-05,
      "loss": 2.1845,
      "step": 1846
    },
    {
      "epoch": 0.8468592388812471,
      "grad_norm": 0.602698613394643,
      "learning_rate": 6.028173010576582e-05,
      "loss": 2.1894,
      "step": 1847
    },
    {
      "epoch": 0.8473177441540578,
      "grad_norm": 0.5998699721034108,
      "learning_rate": 5.9928681324337544e-05,
      "loss": 2.2641,
      "step": 1848
    },
    {
      "epoch": 0.8477762494268684,
      "grad_norm": 0.5985677969776778,
      "learning_rate": 5.957660350435773e-05,
      "loss": 2.1854,
      "step": 1849
    },
    {
      "epoch": 0.848234754699679,
      "grad_norm": 0.5820622331898108,
      "learning_rate": 5.922549742264122e-05,
      "loss": 2.2141,
      "step": 1850
    },
    {
      "epoch": 0.8486932599724897,
      "grad_norm": 0.5902549580046063,
      "learning_rate": 5.8875363853859166e-05,
      "loss": 2.2154,
      "step": 1851
    },
    {
      "epoch": 0.8491517652453003,
      "grad_norm": 0.5761557575270855,
      "learning_rate": 5.852620357053651e-05,
      "loss": 2.2794,
      "step": 1852
    },
    {
      "epoch": 0.849610270518111,
      "grad_norm": 0.6360115180820416,
      "learning_rate": 5.8178017343051336e-05,
      "loss": 2.1768,
      "step": 1853
    },
    {
      "epoch": 0.8500687757909215,
      "grad_norm": 0.5908308195724649,
      "learning_rate": 5.783080593963219e-05,
      "loss": 2.2189,
      "step": 1854
    },
    {
      "epoch": 0.8505272810637322,
      "grad_norm": 0.6113673880132522,
      "learning_rate": 5.748457012635683e-05,
      "loss": 2.2515,
      "step": 1855
    },
    {
      "epoch": 0.8509857863365429,
      "grad_norm": 0.5758862952500144,
      "learning_rate": 5.713931066715078e-05,
      "loss": 2.1341,
      "step": 1856
    },
    {
      "epoch": 0.8514442916093535,
      "grad_norm": 0.5672926704979799,
      "learning_rate": 5.679502832378497e-05,
      "loss": 2.3032,
      "step": 1857
    },
    {
      "epoch": 0.8519027968821642,
      "grad_norm": 0.5913684878720692,
      "learning_rate": 5.645172385587482e-05,
      "loss": 2.1779,
      "step": 1858
    },
    {
      "epoch": 0.8523613021549747,
      "grad_norm": 0.5653993154412222,
      "learning_rate": 5.6109398020877834e-05,
      "loss": 2.1627,
      "step": 1859
    },
    {
      "epoch": 0.8528198074277854,
      "grad_norm": 0.6044883019194571,
      "learning_rate": 5.576805157409265e-05,
      "loss": 2.1802,
      "step": 1860
    },
    {
      "epoch": 0.8532783127005961,
      "grad_norm": 0.6123013316637461,
      "learning_rate": 5.542768526865677e-05,
      "loss": 2.187,
      "step": 1861
    },
    {
      "epoch": 0.8537368179734067,
      "grad_norm": 0.6148781622728293,
      "learning_rate": 5.508829985554509e-05,
      "loss": 2.3012,
      "step": 1862
    },
    {
      "epoch": 0.8541953232462174,
      "grad_norm": 0.5668173215486279,
      "learning_rate": 5.474989608356856e-05,
      "loss": 2.2924,
      "step": 1863
    },
    {
      "epoch": 0.8546538285190279,
      "grad_norm": 0.56054219134298,
      "learning_rate": 5.441247469937194e-05,
      "loss": 2.1634,
      "step": 1864
    },
    {
      "epoch": 0.8551123337918386,
      "grad_norm": 0.5745345522166201,
      "learning_rate": 5.407603644743286e-05,
      "loss": 2.1629,
      "step": 1865
    },
    {
      "epoch": 0.8555708390646493,
      "grad_norm": 0.5806969823849837,
      "learning_rate": 5.374058207005944e-05,
      "loss": 2.1115,
      "step": 1866
    },
    {
      "epoch": 0.8560293443374599,
      "grad_norm": 0.6139475815319624,
      "learning_rate": 5.3406112307389066e-05,
      "loss": 2.253,
      "step": 1867
    },
    {
      "epoch": 0.8564878496102705,
      "grad_norm": 0.5966234401598578,
      "learning_rate": 5.3072627897386926e-05,
      "loss": 2.2409,
      "step": 1868
    },
    {
      "epoch": 0.8569463548830811,
      "grad_norm": 0.7155132419769676,
      "learning_rate": 5.27401295758439e-05,
      "loss": 2.2006,
      "step": 1869
    },
    {
      "epoch": 0.8574048601558918,
      "grad_norm": 0.6032646600267031,
      "learning_rate": 5.2408618076375315e-05,
      "loss": 2.3254,
      "step": 1870
    },
    {
      "epoch": 0.8578633654287025,
      "grad_norm": 0.605315612646555,
      "learning_rate": 5.207809413041914e-05,
      "loss": 2.2961,
      "step": 1871
    },
    {
      "epoch": 0.8583218707015131,
      "grad_norm": 0.5828288683681316,
      "learning_rate": 5.174855846723459e-05,
      "loss": 2.1914,
      "step": 1872
    },
    {
      "epoch": 0.8587803759743237,
      "grad_norm": 0.5965523936170014,
      "learning_rate": 5.1420011813900104e-05,
      "loss": 2.2365,
      "step": 1873
    },
    {
      "epoch": 0.8592388812471343,
      "grad_norm": 0.6176111664825219,
      "learning_rate": 5.109245489531211e-05,
      "loss": 2.2513,
      "step": 1874
    },
    {
      "epoch": 0.859697386519945,
      "grad_norm": 0.5880945896335238,
      "learning_rate": 5.0765888434183446e-05,
      "loss": 2.1137,
      "step": 1875
    },
    {
      "epoch": 0.8601558917927556,
      "grad_norm": 0.595433004589854,
      "learning_rate": 5.0440313151041364e-05,
      "loss": 2.2577,
      "step": 1876
    },
    {
      "epoch": 0.8606143970655663,
      "grad_norm": 0.5915973515536221,
      "learning_rate": 5.011572976422657e-05,
      "loss": 2.1711,
      "step": 1877
    },
    {
      "epoch": 0.8610729023383769,
      "grad_norm": 0.579928177611397,
      "learning_rate": 4.9792138989890825e-05,
      "loss": 2.2167,
      "step": 1878
    },
    {
      "epoch": 0.8615314076111875,
      "grad_norm": 0.6145284977655803,
      "learning_rate": 4.9469541541996234e-05,
      "loss": 2.1636,
      "step": 1879
    },
    {
      "epoch": 0.8619899128839982,
      "grad_norm": 0.5820645283419708,
      "learning_rate": 4.914793813231305e-05,
      "loss": 2.2742,
      "step": 1880
    },
    {
      "epoch": 0.8624484181568088,
      "grad_norm": 0.5913303521458406,
      "learning_rate": 4.882732947041818e-05,
      "loss": 2.237,
      "step": 1881
    },
    {
      "epoch": 0.8629069234296194,
      "grad_norm": 0.5873575655774338,
      "learning_rate": 4.850771626369416e-05,
      "loss": 2.2606,
      "step": 1882
    },
    {
      "epoch": 0.86336542870243,
      "grad_norm": 0.603584667582422,
      "learning_rate": 4.818909921732662e-05,
      "loss": 2.2173,
      "step": 1883
    },
    {
      "epoch": 0.8638239339752407,
      "grad_norm": 0.6040510313516375,
      "learning_rate": 4.787147903430383e-05,
      "loss": 2.1937,
      "step": 1884
    },
    {
      "epoch": 0.8642824392480514,
      "grad_norm": 0.6039293524764934,
      "learning_rate": 4.755485641541424e-05,
      "loss": 2.2065,
      "step": 1885
    },
    {
      "epoch": 0.864740944520862,
      "grad_norm": 0.6039909201037811,
      "learning_rate": 4.723923205924557e-05,
      "loss": 2.1727,
      "step": 1886
    },
    {
      "epoch": 0.8651994497936726,
      "grad_norm": 0.5904204566716188,
      "learning_rate": 4.6924606662182736e-05,
      "loss": 2.2989,
      "step": 1887
    },
    {
      "epoch": 0.8656579550664832,
      "grad_norm": 0.5967346600582255,
      "learning_rate": 4.6610980918406596e-05,
      "loss": 2.2054,
      "step": 1888
    },
    {
      "epoch": 0.8661164603392939,
      "grad_norm": 0.5979933616288466,
      "learning_rate": 4.629835551989276e-05,
      "loss": 2.2543,
      "step": 1889
    },
    {
      "epoch": 0.8665749656121046,
      "grad_norm": 0.5954445713135842,
      "learning_rate": 4.5986731156409224e-05,
      "loss": 2.2535,
      "step": 1890
    },
    {
      "epoch": 0.8670334708849152,
      "grad_norm": 0.6066523202857538,
      "learning_rate": 4.567610851551568e-05,
      "loss": 2.3721,
      "step": 1891
    },
    {
      "epoch": 0.8674919761577258,
      "grad_norm": 0.5769112006810355,
      "learning_rate": 4.536648828256146e-05,
      "loss": 2.2207,
      "step": 1892
    },
    {
      "epoch": 0.8679504814305364,
      "grad_norm": 0.5937286117918483,
      "learning_rate": 4.505787114068433e-05,
      "loss": 2.1996,
      "step": 1893
    },
    {
      "epoch": 0.8684089867033471,
      "grad_norm": 0.585357079819607,
      "learning_rate": 4.4750257770808764e-05,
      "loss": 2.2569,
      "step": 1894
    },
    {
      "epoch": 0.8688674919761578,
      "grad_norm": 0.6174192964715872,
      "learning_rate": 4.444364885164448e-05,
      "loss": 2.195,
      "step": 1895
    },
    {
      "epoch": 0.8693259972489684,
      "grad_norm": 0.5943222784852443,
      "learning_rate": 4.413804505968533e-05,
      "loss": 2.2599,
      "step": 1896
    },
    {
      "epoch": 0.869784502521779,
      "grad_norm": 0.6050108725125566,
      "learning_rate": 4.3833447069206944e-05,
      "loss": 2.329,
      "step": 1897
    },
    {
      "epoch": 0.8702430077945896,
      "grad_norm": 0.6160414620625191,
      "learning_rate": 4.352985555226635e-05,
      "loss": 2.267,
      "step": 1898
    },
    {
      "epoch": 0.8707015130674003,
      "grad_norm": 0.589204993095767,
      "learning_rate": 4.322727117869951e-05,
      "loss": 2.1939,
      "step": 1899
    },
    {
      "epoch": 0.871160018340211,
      "grad_norm": 0.5768066292601052,
      "learning_rate": 4.29256946161205e-05,
      "loss": 2.1711,
      "step": 1900
    },
    {
      "epoch": 0.8716185236130215,
      "grad_norm": 0.58846920293021,
      "learning_rate": 4.262512652991968e-05,
      "loss": 2.1348,
      "step": 1901
    },
    {
      "epoch": 0.8720770288858322,
      "grad_norm": 0.6064534810168002,
      "learning_rate": 4.2325567583262113e-05,
      "loss": 2.1698,
      "step": 1902
    },
    {
      "epoch": 0.8725355341586428,
      "grad_norm": 0.5941180847762314,
      "learning_rate": 4.2027018437086895e-05,
      "loss": 2.2155,
      "step": 1903
    },
    {
      "epoch": 0.8729940394314535,
      "grad_norm": 0.6237309816483639,
      "learning_rate": 4.172947975010449e-05,
      "loss": 2.1811,
      "step": 1904
    },
    {
      "epoch": 0.8734525447042641,
      "grad_norm": 0.6315372658325775,
      "learning_rate": 4.143295217879645e-05,
      "loss": 2.3393,
      "step": 1905
    },
    {
      "epoch": 0.8739110499770747,
      "grad_norm": 0.5970964958232399,
      "learning_rate": 4.113743637741296e-05,
      "loss": 2.2492,
      "step": 1906
    },
    {
      "epoch": 0.8743695552498854,
      "grad_norm": 0.589706907717056,
      "learning_rate": 4.084293299797226e-05,
      "loss": 2.2622,
      "step": 1907
    },
    {
      "epoch": 0.874828060522696,
      "grad_norm": 0.6147159138377319,
      "learning_rate": 4.054944269025862e-05,
      "loss": 2.151,
      "step": 1908
    },
    {
      "epoch": 0.8752865657955067,
      "grad_norm": 0.6095565658562115,
      "learning_rate": 4.025696610182095e-05,
      "loss": 2.2138,
      "step": 1909
    },
    {
      "epoch": 0.8757450710683173,
      "grad_norm": 0.6457585325428716,
      "learning_rate": 3.996550387797187e-05,
      "loss": 2.1492,
      "step": 1910
    },
    {
      "epoch": 0.8762035763411279,
      "grad_norm": 0.6101766953785513,
      "learning_rate": 3.9675056661785556e-05,
      "loss": 2.0978,
      "step": 1911
    },
    {
      "epoch": 0.8766620816139385,
      "grad_norm": 0.5758772172609158,
      "learning_rate": 3.9385625094097154e-05,
      "loss": 2.2653,
      "step": 1912
    },
    {
      "epoch": 0.8771205868867492,
      "grad_norm": 0.5787504018602009,
      "learning_rate": 3.909720981350034e-05,
      "loss": 2.1293,
      "step": 1913
    },
    {
      "epoch": 0.8775790921595599,
      "grad_norm": 0.5943921452913169,
      "learning_rate": 3.880981145634704e-05,
      "loss": 2.2782,
      "step": 1914
    },
    {
      "epoch": 0.8780375974323704,
      "grad_norm": 0.6174314688580125,
      "learning_rate": 3.852343065674507e-05,
      "loss": 2.2711,
      "step": 1915
    },
    {
      "epoch": 0.8784961027051811,
      "grad_norm": 0.6209041631119023,
      "learning_rate": 3.8238068046557276e-05,
      "loss": 2.1966,
      "step": 1916
    },
    {
      "epoch": 0.8789546079779917,
      "grad_norm": 0.5870094398847924,
      "learning_rate": 3.795372425540006e-05,
      "loss": 2.1793,
      "step": 1917
    },
    {
      "epoch": 0.8794131132508024,
      "grad_norm": 0.5904757690614225,
      "learning_rate": 3.76703999106418e-05,
      "loss": 2.1642,
      "step": 1918
    },
    {
      "epoch": 0.8798716185236131,
      "grad_norm": 0.5623852595490635,
      "learning_rate": 3.7388095637401754e-05,
      "loss": 2.1673,
      "step": 1919
    },
    {
      "epoch": 0.8803301237964236,
      "grad_norm": 0.5715068809957012,
      "learning_rate": 3.7106812058548376e-05,
      "loss": 2.217,
      "step": 1920
    },
    {
      "epoch": 0.8807886290692343,
      "grad_norm": 0.6115807324716154,
      "learning_rate": 3.682654979469807e-05,
      "loss": 2.1936,
      "step": 1921
    },
    {
      "epoch": 0.8812471343420449,
      "grad_norm": 0.5680402429658851,
      "learning_rate": 3.654730946421403e-05,
      "loss": 2.2641,
      "step": 1922
    },
    {
      "epoch": 0.8817056396148556,
      "grad_norm": 0.5999301031167111,
      "learning_rate": 3.6269091683204466e-05,
      "loss": 2.2313,
      "step": 1923
    },
    {
      "epoch": 0.8821641448876663,
      "grad_norm": 0.607445425229815,
      "learning_rate": 3.5991897065521693e-05,
      "loss": 2.294,
      "step": 1924
    },
    {
      "epoch": 0.8826226501604768,
      "grad_norm": 0.611067332139173,
      "learning_rate": 3.571572622276026e-05,
      "loss": 2.2431,
      "step": 1925
    },
    {
      "epoch": 0.8830811554332875,
      "grad_norm": 0.6260431036902746,
      "learning_rate": 3.544057976425619e-05,
      "loss": 2.3116,
      "step": 1926
    },
    {
      "epoch": 0.8835396607060981,
      "grad_norm": 0.5798336164573706,
      "learning_rate": 3.5166458297085146e-05,
      "loss": 2.2666,
      "step": 1927
    },
    {
      "epoch": 0.8839981659789088,
      "grad_norm": 0.5945723359398125,
      "learning_rate": 3.489336242606111e-05,
      "loss": 2.227,
      "step": 1928
    },
    {
      "epoch": 0.8844566712517193,
      "grad_norm": 0.5940994041739167,
      "learning_rate": 3.462129275373577e-05,
      "loss": 2.1909,
      "step": 1929
    },
    {
      "epoch": 0.88491517652453,
      "grad_norm": 0.5650233324834317,
      "learning_rate": 3.4350249880395924e-05,
      "loss": 2.2874,
      "step": 1930
    },
    {
      "epoch": 0.8853736817973407,
      "grad_norm": 0.6031812687132124,
      "learning_rate": 3.408023440406355e-05,
      "loss": 2.1959,
      "step": 1931
    },
    {
      "epoch": 0.8858321870701513,
      "grad_norm": 0.5970151953698624,
      "learning_rate": 3.381124692049331e-05,
      "loss": 2.2577,
      "step": 1932
    },
    {
      "epoch": 0.886290692342962,
      "grad_norm": 0.5584178769142262,
      "learning_rate": 3.354328802317197e-05,
      "loss": 2.1859,
      "step": 1933
    },
    {
      "epoch": 0.8867491976157725,
      "grad_norm": 0.6068670914630158,
      "learning_rate": 3.327635830331677e-05,
      "loss": 2.171,
      "step": 1934
    },
    {
      "epoch": 0.8872077028885832,
      "grad_norm": 0.5861246588933362,
      "learning_rate": 3.3010458349874206e-05,
      "loss": 2.2375,
      "step": 1935
    },
    {
      "epoch": 0.8876662081613939,
      "grad_norm": 0.6145337206559797,
      "learning_rate": 3.2745588749518775e-05,
      "loss": 2.275,
      "step": 1936
    },
    {
      "epoch": 0.8881247134342045,
      "grad_norm": 0.5929688446999148,
      "learning_rate": 3.248175008665161e-05,
      "loss": 2.1893,
      "step": 1937
    },
    {
      "epoch": 0.8885832187070152,
      "grad_norm": 0.6079728887928932,
      "learning_rate": 3.221894294339911e-05,
      "loss": 2.2379,
      "step": 1938
    },
    {
      "epoch": 0.8890417239798257,
      "grad_norm": 0.600073626755313,
      "learning_rate": 3.1957167899611836e-05,
      "loss": 2.1737,
      "step": 1939
    },
    {
      "epoch": 0.8895002292526364,
      "grad_norm": 0.5952275693278651,
      "learning_rate": 3.169642553286334e-05,
      "loss": 2.2965,
      "step": 1940
    },
    {
      "epoch": 0.889958734525447,
      "grad_norm": 0.599095860682832,
      "learning_rate": 3.143671641844831e-05,
      "loss": 2.2272,
      "step": 1941
    },
    {
      "epoch": 0.8904172397982577,
      "grad_norm": 0.597038652779308,
      "learning_rate": 3.117804112938205e-05,
      "loss": 2.158,
      "step": 1942
    },
    {
      "epoch": 0.8908757450710684,
      "grad_norm": 0.6296430976718784,
      "learning_rate": 3.092040023639869e-05,
      "loss": 2.2653,
      "step": 1943
    },
    {
      "epoch": 0.8913342503438789,
      "grad_norm": 0.5986714638433944,
      "learning_rate": 3.066379430795002e-05,
      "loss": 2.2214,
      "step": 1944
    },
    {
      "epoch": 0.8917927556166896,
      "grad_norm": 0.5822662339300256,
      "learning_rate": 3.040822391020459e-05,
      "loss": 2.2419,
      "step": 1945
    },
    {
      "epoch": 0.8922512608895002,
      "grad_norm": 0.5735066460217177,
      "learning_rate": 3.0153689607045842e-05,
      "loss": 2.1722,
      "step": 1946
    },
    {
      "epoch": 0.8927097661623109,
      "grad_norm": 0.5676888804341053,
      "learning_rate": 2.9900191960071545e-05,
      "loss": 2.2018,
      "step": 1947
    },
    {
      "epoch": 0.8931682714351215,
      "grad_norm": 0.6253189407013929,
      "learning_rate": 2.9647731528591848e-05,
      "loss": 2.2347,
      "step": 1948
    },
    {
      "epoch": 0.8936267767079321,
      "grad_norm": 0.5941465847468409,
      "learning_rate": 2.9396308869628795e-05,
      "loss": 2.2041,
      "step": 1949
    },
    {
      "epoch": 0.8940852819807428,
      "grad_norm": 0.5882028522090555,
      "learning_rate": 2.914592453791448e-05,
      "loss": 2.2089,
      "step": 1950
    },
    {
      "epoch": 0.8945437872535534,
      "grad_norm": 0.5939835773043023,
      "learning_rate": 2.8896579085889994e-05,
      "loss": 2.2441,
      "step": 1951
    },
    {
      "epoch": 0.8950022925263641,
      "grad_norm": 0.5726572483748968,
      "learning_rate": 2.86482730637046e-05,
      "loss": 2.1903,
      "step": 1952
    },
    {
      "epoch": 0.8954607977991746,
      "grad_norm": 0.6089369256826278,
      "learning_rate": 2.840100701921383e-05,
      "loss": 2.2486,
      "step": 1953
    },
    {
      "epoch": 0.8959193030719853,
      "grad_norm": 0.6040374732365753,
      "learning_rate": 2.8154781497978898e-05,
      "loss": 2.282,
      "step": 1954
    },
    {
      "epoch": 0.896377808344796,
      "grad_norm": 0.6106283825167789,
      "learning_rate": 2.7909597043265013e-05,
      "loss": 2.1896,
      "step": 1955
    },
    {
      "epoch": 0.8968363136176066,
      "grad_norm": 0.5900343205595411,
      "learning_rate": 2.7665454196040662e-05,
      "loss": 2.2291,
      "step": 1956
    },
    {
      "epoch": 0.8972948188904173,
      "grad_norm": 0.5963790159120765,
      "learning_rate": 2.7422353494975905e-05,
      "loss": 2.2343,
      "step": 1957
    },
    {
      "epoch": 0.8977533241632278,
      "grad_norm": 0.6018650743269475,
      "learning_rate": 2.7180295476441573e-05,
      "loss": 2.2024,
      "step": 1958
    },
    {
      "epoch": 0.8982118294360385,
      "grad_norm": 0.608862118981205,
      "learning_rate": 2.6939280674508016e-05,
      "loss": 2.2036,
      "step": 1959
    },
    {
      "epoch": 0.8986703347088492,
      "grad_norm": 0.5641537897343564,
      "learning_rate": 2.669930962094358e-05,
      "loss": 2.2442,
      "step": 1960
    },
    {
      "epoch": 0.8991288399816598,
      "grad_norm": 0.6073184779186134,
      "learning_rate": 2.6460382845214126e-05,
      "loss": 2.3543,
      "step": 1961
    },
    {
      "epoch": 0.8995873452544704,
      "grad_norm": 0.5796515502536176,
      "learning_rate": 2.6222500874481025e-05,
      "loss": 2.2722,
      "step": 1962
    },
    {
      "epoch": 0.900045850527281,
      "grad_norm": 0.6207091653687039,
      "learning_rate": 2.5985664233600827e-05,
      "loss": 2.2368,
      "step": 1963
    },
    {
      "epoch": 0.9005043558000917,
      "grad_norm": 0.6167892913079599,
      "learning_rate": 2.574987344512336e-05,
      "loss": 2.1872,
      "step": 1964
    },
    {
      "epoch": 0.9009628610729024,
      "grad_norm": 0.5694232398537468,
      "learning_rate": 2.5515129029290984e-05,
      "loss": 2.2441,
      "step": 1965
    },
    {
      "epoch": 0.901421366345713,
      "grad_norm": 0.5885243638674613,
      "learning_rate": 2.5281431504037556e-05,
      "loss": 2.2108,
      "step": 1966
    },
    {
      "epoch": 0.9018798716185236,
      "grad_norm": 0.5649013351041455,
      "learning_rate": 2.504878138498684e-05,
      "loss": 2.1655,
      "step": 1967
    },
    {
      "epoch": 0.9023383768913342,
      "grad_norm": 0.6184461837419902,
      "learning_rate": 2.48171791854519e-05,
      "loss": 2.29,
      "step": 1968
    },
    {
      "epoch": 0.9027968821641449,
      "grad_norm": 0.5734673454696234,
      "learning_rate": 2.4586625416433473e-05,
      "loss": 2.224,
      "step": 1969
    },
    {
      "epoch": 0.9032553874369555,
      "grad_norm": 0.5947593598589026,
      "learning_rate": 2.435712058661921e-05,
      "loss": 2.2604,
      "step": 1970
    },
    {
      "epoch": 0.9037138927097662,
      "grad_norm": 0.6342957202978186,
      "learning_rate": 2.4128665202382327e-05,
      "loss": 2.1957,
      "step": 1971
    },
    {
      "epoch": 0.9041723979825768,
      "grad_norm": 0.5945502605740154,
      "learning_rate": 2.3901259767780515e-05,
      "loss": 2.1786,
      "step": 1972
    },
    {
      "epoch": 0.9046309032553874,
      "grad_norm": 0.5999178246433907,
      "learning_rate": 2.367490478455514e-05,
      "loss": 2.2857,
      "step": 1973
    },
    {
      "epoch": 0.9050894085281981,
      "grad_norm": 0.5825158940063397,
      "learning_rate": 2.3449600752129597e-05,
      "loss": 2.1997,
      "step": 1974
    },
    {
      "epoch": 0.9055479138010087,
      "grad_norm": 0.6129048901629183,
      "learning_rate": 2.3225348167608685e-05,
      "loss": 2.1738,
      "step": 1975
    },
    {
      "epoch": 0.9060064190738194,
      "grad_norm": 0.6047023582023044,
      "learning_rate": 2.3002147525777118e-05,
      "loss": 2.3102,
      "step": 1976
    },
    {
      "epoch": 0.90646492434663,
      "grad_norm": 0.5834124268217693,
      "learning_rate": 2.2779999319098856e-05,
      "loss": 2.1794,
      "step": 1977
    },
    {
      "epoch": 0.9069234296194406,
      "grad_norm": 0.5957084440068119,
      "learning_rate": 2.255890403771571e-05,
      "loss": 2.2168,
      "step": 1978
    },
    {
      "epoch": 0.9073819348922513,
      "grad_norm": 0.574696674938281,
      "learning_rate": 2.233886216944614e-05,
      "loss": 2.1575,
      "step": 1979
    },
    {
      "epoch": 0.9078404401650619,
      "grad_norm": 0.5917166513004455,
      "learning_rate": 2.211987419978484e-05,
      "loss": 2.2955,
      "step": 1980
    },
    {
      "epoch": 0.9082989454378725,
      "grad_norm": 0.6006838543281554,
      "learning_rate": 2.1901940611900705e-05,
      "loss": 2.2969,
      "step": 1981
    },
    {
      "epoch": 0.9087574507106831,
      "grad_norm": 0.6051507059230571,
      "learning_rate": 2.168506188663666e-05,
      "loss": 2.2504,
      "step": 1982
    },
    {
      "epoch": 0.9092159559834938,
      "grad_norm": 0.5628637579767456,
      "learning_rate": 2.1469238502507925e-05,
      "loss": 2.2246,
      "step": 1983
    },
    {
      "epoch": 0.9096744612563045,
      "grad_norm": 0.5860061742850657,
      "learning_rate": 2.125447093570154e-05,
      "loss": 2.1733,
      "step": 1984
    },
    {
      "epoch": 0.9101329665291151,
      "grad_norm": 0.6104972729322954,
      "learning_rate": 2.1040759660074793e-05,
      "loss": 2.2132,
      "step": 1985
    },
    {
      "epoch": 0.9105914718019257,
      "grad_norm": 0.6148154039847594,
      "learning_rate": 2.0828105147154273e-05,
      "loss": 2.2248,
      "step": 1986
    },
    {
      "epoch": 0.9110499770747363,
      "grad_norm": 0.6212882448346628,
      "learning_rate": 2.061650786613545e-05,
      "loss": 2.2138,
      "step": 1987
    },
    {
      "epoch": 0.911508482347547,
      "grad_norm": 0.6418313770828379,
      "learning_rate": 2.040596828388058e-05,
      "loss": 2.2389,
      "step": 1988
    },
    {
      "epoch": 0.9119669876203577,
      "grad_norm": 0.6481766822074002,
      "learning_rate": 2.019648686491865e-05,
      "loss": 2.2583,
      "step": 1989
    },
    {
      "epoch": 0.9124254928931683,
      "grad_norm": 0.6142741284151791,
      "learning_rate": 1.9988064071443767e-05,
      "loss": 2.1491,
      "step": 1990
    },
    {
      "epoch": 0.9128839981659789,
      "grad_norm": 0.6038671534734928,
      "learning_rate": 1.9780700363314253e-05,
      "loss": 2.233,
      "step": 1991
    },
    {
      "epoch": 0.9133425034387895,
      "grad_norm": 0.5966000737470799,
      "learning_rate": 1.957439619805196e-05,
      "loss": 2.2721,
      "step": 1992
    },
    {
      "epoch": 0.9138010087116002,
      "grad_norm": 0.6141152397430641,
      "learning_rate": 1.9369152030840554e-05,
      "loss": 2.1557,
      "step": 1993
    },
    {
      "epoch": 0.9142595139844109,
      "grad_norm": 0.6073570453083265,
      "learning_rate": 1.916496831452552e-05,
      "loss": 2.2659,
      "step": 1994
    },
    {
      "epoch": 0.9147180192572214,
      "grad_norm": 0.6079268123842073,
      "learning_rate": 1.8961845499611998e-05,
      "loss": 2.0827,
      "step": 1995
    },
    {
      "epoch": 0.9151765245300321,
      "grad_norm": 0.6241797261190638,
      "learning_rate": 1.8759784034264925e-05,
      "loss": 2.2592,
      "step": 1996
    },
    {
      "epoch": 0.9156350298028427,
      "grad_norm": 0.5932183173280962,
      "learning_rate": 1.855878436430708e-05,
      "loss": 2.3365,
      "step": 1997
    },
    {
      "epoch": 0.9160935350756534,
      "grad_norm": 0.5981630863990143,
      "learning_rate": 1.835884693321871e-05,
      "loss": 2.1886,
      "step": 1998
    },
    {
      "epoch": 0.916552040348464,
      "grad_norm": 0.600395659411338,
      "learning_rate": 1.8159972182136386e-05,
      "loss": 2.3394,
      "step": 1999
    },
    {
      "epoch": 0.9170105456212746,
      "grad_norm": 0.6020758218345509,
      "learning_rate": 1.7962160549851945e-05,
      "loss": 2.2142,
      "step": 2000
    },
    {
      "epoch": 0.9174690508940853,
      "grad_norm": 0.592794097366753,
      "learning_rate": 1.776541247281177e-05,
      "loss": 2.3084,
      "step": 2001
    },
    {
      "epoch": 0.9179275561668959,
      "grad_norm": 0.5702378318742825,
      "learning_rate": 1.7569728385115224e-05,
      "loss": 2.157,
      "step": 2002
    },
    {
      "epoch": 0.9183860614397066,
      "grad_norm": 0.5965100396848553,
      "learning_rate": 1.7375108718514665e-05,
      "loss": 2.2102,
      "step": 2003
    },
    {
      "epoch": 0.9188445667125172,
      "grad_norm": 0.6067080660868058,
      "learning_rate": 1.7181553902413438e-05,
      "loss": 2.2529,
      "step": 2004
    },
    {
      "epoch": 0.9193030719853278,
      "grad_norm": 0.5714825675382341,
      "learning_rate": 1.698906436386577e-05,
      "loss": 2.219,
      "step": 2005
    },
    {
      "epoch": 0.9197615772581385,
      "grad_norm": 0.5837273929851526,
      "learning_rate": 1.679764052757532e-05,
      "loss": 2.2948,
      "step": 2006
    },
    {
      "epoch": 0.9202200825309491,
      "grad_norm": 0.5825391788178951,
      "learning_rate": 1.6607282815894464e-05,
      "loss": 2.183,
      "step": 2007
    },
    {
      "epoch": 0.9206785878037598,
      "grad_norm": 0.5841745274626587,
      "learning_rate": 1.6417991648823405e-05,
      "loss": 2.2189,
      "step": 2008
    },
    {
      "epoch": 0.9211370930765704,
      "grad_norm": 0.5956785665007145,
      "learning_rate": 1.6229767444008835e-05,
      "loss": 2.2124,
      "step": 2009
    },
    {
      "epoch": 0.921595598349381,
      "grad_norm": 0.6167267065914307,
      "learning_rate": 1.604261061674378e-05,
      "loss": 2.2556,
      "step": 2010
    },
    {
      "epoch": 0.9220541036221916,
      "grad_norm": 0.6091376085901271,
      "learning_rate": 1.5856521579965865e-05,
      "loss": 2.1916,
      "step": 2011
    },
    {
      "epoch": 0.9225126088950023,
      "grad_norm": 0.6035051464510606,
      "learning_rate": 1.5671500744256938e-05,
      "loss": 2.2359,
      "step": 2012
    },
    {
      "epoch": 0.922971114167813,
      "grad_norm": 0.6078371599715308,
      "learning_rate": 1.5487548517841953e-05,
      "loss": 2.19,
      "step": 2013
    },
    {
      "epoch": 0.9234296194406235,
      "grad_norm": 0.5910987579631665,
      "learning_rate": 1.530466530658814e-05,
      "loss": 2.2634,
      "step": 2014
    },
    {
      "epoch": 0.9238881247134342,
      "grad_norm": 0.6253514024227311,
      "learning_rate": 1.5122851514004054e-05,
      "loss": 2.3235,
      "step": 2015
    },
    {
      "epoch": 0.9243466299862448,
      "grad_norm": 0.6552605802256022,
      "learning_rate": 1.4942107541238703e-05,
      "loss": 2.2658,
      "step": 2016
    },
    {
      "epoch": 0.9248051352590555,
      "grad_norm": 0.5831426130378176,
      "learning_rate": 1.4762433787080809e-05,
      "loss": 2.1825,
      "step": 2017
    },
    {
      "epoch": 0.9252636405318662,
      "grad_norm": 0.6188949197452187,
      "learning_rate": 1.4583830647957541e-05,
      "loss": 2.3114,
      "step": 2018
    },
    {
      "epoch": 0.9257221458046767,
      "grad_norm": 0.5717999614074785,
      "learning_rate": 1.4406298517934068e-05,
      "loss": 2.097,
      "step": 2019
    },
    {
      "epoch": 0.9261806510774874,
      "grad_norm": 0.5820716960966027,
      "learning_rate": 1.4229837788712562e-05,
      "loss": 2.1343,
      "step": 2020
    },
    {
      "epoch": 0.926639156350298,
      "grad_norm": 0.6029178599148091,
      "learning_rate": 1.4054448849631085e-05,
      "loss": 2.2928,
      "step": 2021
    },
    {
      "epoch": 0.9270976616231087,
      "grad_norm": 0.6115067773388972,
      "learning_rate": 1.3880132087663145e-05,
      "loss": 2.1738,
      "step": 2022
    },
    {
      "epoch": 0.9275561668959194,
      "grad_norm": 0.5856055035892648,
      "learning_rate": 1.3706887887416419e-05,
      "loss": 2.2085,
      "step": 2023
    },
    {
      "epoch": 0.9280146721687299,
      "grad_norm": 0.5783564932315128,
      "learning_rate": 1.3534716631132316e-05,
      "loss": 2.143,
      "step": 2024
    },
    {
      "epoch": 0.9284731774415406,
      "grad_norm": 0.5943792063585417,
      "learning_rate": 1.3363618698684853e-05,
      "loss": 2.1991,
      "step": 2025
    },
    {
      "epoch": 0.9289316827143512,
      "grad_norm": 0.5596031701300679,
      "learning_rate": 1.3193594467579728e-05,
      "loss": 2.2167,
      "step": 2026
    },
    {
      "epoch": 0.9293901879871619,
      "grad_norm": 0.5919268520635078,
      "learning_rate": 1.3024644312954026e-05,
      "loss": 2.2092,
      "step": 2027
    },
    {
      "epoch": 0.9298486932599724,
      "grad_norm": 0.5992286386145641,
      "learning_rate": 1.2856768607574564e-05,
      "loss": 2.2109,
      "step": 2028
    },
    {
      "epoch": 0.9303071985327831,
      "grad_norm": 0.6135733603134963,
      "learning_rate": 1.2689967721837947e-05,
      "loss": 2.307,
      "step": 2029
    },
    {
      "epoch": 0.9307657038055938,
      "grad_norm": 0.6375552989626662,
      "learning_rate": 1.2524242023769006e-05,
      "loss": 2.2991,
      "step": 2030
    },
    {
      "epoch": 0.9312242090784044,
      "grad_norm": 0.6145985375966521,
      "learning_rate": 1.2359591879020526e-05,
      "loss": 2.2461,
      "step": 2031
    },
    {
      "epoch": 0.9316827143512151,
      "grad_norm": 0.610632205767748,
      "learning_rate": 1.2196017650872081e-05,
      "loss": 2.2148,
      "step": 2032
    },
    {
      "epoch": 0.9321412196240256,
      "grad_norm": 0.5886260920192281,
      "learning_rate": 1.2033519700229367e-05,
      "loss": 2.0942,
      "step": 2033
    },
    {
      "epoch": 0.9325997248968363,
      "grad_norm": 0.6139975466516261,
      "learning_rate": 1.1872098385623586e-05,
      "loss": 2.2805,
      "step": 2034
    },
    {
      "epoch": 0.933058230169647,
      "grad_norm": 0.5718506369354385,
      "learning_rate": 1.1711754063210289e-05,
      "loss": 2.2396,
      "step": 2035
    },
    {
      "epoch": 0.9335167354424576,
      "grad_norm": 0.5967717488929729,
      "learning_rate": 1.155248708676887e-05,
      "loss": 2.1745,
      "step": 2036
    },
    {
      "epoch": 0.9339752407152683,
      "grad_norm": 0.5941058826254189,
      "learning_rate": 1.1394297807701737e-05,
      "loss": 2.1836,
      "step": 2037
    },
    {
      "epoch": 0.9344337459880788,
      "grad_norm": 0.589695826966411,
      "learning_rate": 1.1237186575033254e-05,
      "loss": 2.2794,
      "step": 2038
    },
    {
      "epoch": 0.9348922512608895,
      "grad_norm": 0.5949302949398788,
      "learning_rate": 1.1081153735409522e-05,
      "loss": 2.1612,
      "step": 2039
    },
    {
      "epoch": 0.9353507565337001,
      "grad_norm": 0.6113146059743578,
      "learning_rate": 1.0926199633097156e-05,
      "loss": 2.2748,
      "step": 2040
    },
    {
      "epoch": 0.9358092618065108,
      "grad_norm": 0.6014479380596317,
      "learning_rate": 1.0772324609982787e-05,
      "loss": 2.2507,
      "step": 2041
    },
    {
      "epoch": 0.9362677670793215,
      "grad_norm": 0.5963153324176408,
      "learning_rate": 1.0619529005571893e-05,
      "loss": 2.2278,
      "step": 2042
    },
    {
      "epoch": 0.936726272352132,
      "grad_norm": 0.6253943198236814,
      "learning_rate": 1.0467813156988748e-05,
      "loss": 2.2879,
      "step": 2043
    },
    {
      "epoch": 0.9371847776249427,
      "grad_norm": 0.5759170012446205,
      "learning_rate": 1.0317177398975031e-05,
      "loss": 2.2167,
      "step": 2044
    },
    {
      "epoch": 0.9376432828977533,
      "grad_norm": 0.568557206227818,
      "learning_rate": 1.0167622063889326e-05,
      "loss": 2.1459,
      "step": 2045
    },
    {
      "epoch": 0.938101788170564,
      "grad_norm": 0.6061232436471413,
      "learning_rate": 1.0019147481706625e-05,
      "loss": 2.2453,
      "step": 2046
    },
    {
      "epoch": 0.9385602934433745,
      "grad_norm": 0.6142548966509139,
      "learning_rate": 9.871753980017051e-06,
      "loss": 2.2053,
      "step": 2047
    },
    {
      "epoch": 0.9390187987161852,
      "grad_norm": 0.6007822234984665,
      "learning_rate": 9.725441884025855e-06,
      "loss": 2.2649,
      "step": 2048
    },
    {
      "epoch": 0.9394773039889959,
      "grad_norm": 0.5948975102336971,
      "learning_rate": 9.580211516551862e-06,
      "loss": 2.2024,
      "step": 2049
    },
    {
      "epoch": 0.9399358092618065,
      "grad_norm": 0.5745608808149021,
      "learning_rate": 9.436063198027589e-06,
      "loss": 2.205,
      "step": 2050
    },
    {
      "epoch": 0.9403943145346172,
      "grad_norm": 0.5811417741287898,
      "learning_rate": 9.292997246497959e-06,
      "loss": 2.2788,
      "step": 2051
    },
    {
      "epoch": 0.9408528198074277,
      "grad_norm": 0.5917127176836167,
      "learning_rate": 9.151013977619693e-06,
      "loss": 2.1127,
      "step": 2052
    },
    {
      "epoch": 0.9413113250802384,
      "grad_norm": 0.5909131939642609,
      "learning_rate": 9.010113704661038e-06,
      "loss": 2.1554,
      "step": 2053
    },
    {
      "epoch": 0.9417698303530491,
      "grad_norm": 0.5909465070800025,
      "learning_rate": 8.870296738500316e-06,
      "loss": 2.2111,
      "step": 2054
    },
    {
      "epoch": 0.9422283356258597,
      "grad_norm": 0.5775659160338358,
      "learning_rate": 8.731563387626095e-06,
      "loss": 2.2694,
      "step": 2055
    },
    {
      "epoch": 0.9426868408986704,
      "grad_norm": 0.5829118309374,
      "learning_rate": 8.59391395813569e-06,
      "loss": 2.2047,
      "step": 2056
    },
    {
      "epoch": 0.9431453461714809,
      "grad_norm": 0.5832841223291834,
      "learning_rate": 8.457348753735328e-06,
      "loss": 2.1913,
      "step": 2057
    },
    {
      "epoch": 0.9436038514442916,
      "grad_norm": 0.589206419920452,
      "learning_rate": 8.321868075738593e-06,
      "loss": 2.2438,
      "step": 2058
    },
    {
      "epoch": 0.9440623567171023,
      "grad_norm": 0.6084726164595207,
      "learning_rate": 8.187472223066371e-06,
      "loss": 2.2376,
      "step": 2059
    },
    {
      "epoch": 0.9445208619899129,
      "grad_norm": 0.605942474333704,
      "learning_rate": 8.054161492246136e-06,
      "loss": 2.2057,
      "step": 2060
    },
    {
      "epoch": 0.9449793672627235,
      "grad_norm": 0.7241332213247305,
      "learning_rate": 7.921936177411049e-06,
      "loss": 2.2983,
      "step": 2061
    },
    {
      "epoch": 0.9454378725355341,
      "grad_norm": 0.6209838179194426,
      "learning_rate": 7.790796570299463e-06,
      "loss": 2.2014,
      "step": 2062
    },
    {
      "epoch": 0.9458963778083448,
      "grad_norm": 0.5829170953097456,
      "learning_rate": 7.660742960254207e-06,
      "loss": 2.2343,
      "step": 2063
    },
    {
      "epoch": 0.9463548830811555,
      "grad_norm": 0.6096646297560612,
      "learning_rate": 7.531775634222138e-06,
      "loss": 2.1709,
      "step": 2064
    },
    {
      "epoch": 0.9468133883539661,
      "grad_norm": 0.6117939524400531,
      "learning_rate": 7.403894876753192e-06,
      "loss": 2.1662,
      "step": 2065
    },
    {
      "epoch": 0.9472718936267767,
      "grad_norm": 0.5769812949343294,
      "learning_rate": 7.277100970000061e-06,
      "loss": 2.152,
      "step": 2066
    },
    {
      "epoch": 0.9477303988995873,
      "grad_norm": 0.6169574752539277,
      "learning_rate": 7.151394193717408e-06,
      "loss": 2.2194,
      "step": 2067
    },
    {
      "epoch": 0.948188904172398,
      "grad_norm": 0.6289376738202695,
      "learning_rate": 7.026774825261151e-06,
      "loss": 2.2505,
      "step": 2068
    },
    {
      "epoch": 0.9486474094452086,
      "grad_norm": 0.5455753752045424,
      "learning_rate": 6.903243139588233e-06,
      "loss": 2.1536,
      "step": 2069
    },
    {
      "epoch": 0.9491059147180193,
      "grad_norm": 0.5892892049560431,
      "learning_rate": 6.780799409255522e-06,
      "loss": 2.2641,
      "step": 2070
    },
    {
      "epoch": 0.9495644199908299,
      "grad_norm": 0.6052551044011529,
      "learning_rate": 6.659443904419637e-06,
      "loss": 2.3686,
      "step": 2071
    },
    {
      "epoch": 0.9500229252636405,
      "grad_norm": 0.5952839160980128,
      "learning_rate": 6.539176892836008e-06,
      "loss": 2.1369,
      "step": 2072
    },
    {
      "epoch": 0.9504814305364512,
      "grad_norm": 0.5846864018474215,
      "learning_rate": 6.4199986398585375e-06,
      "loss": 2.2847,
      "step": 2073
    },
    {
      "epoch": 0.9509399358092618,
      "grad_norm": 0.6075732465702433,
      "learning_rate": 6.3019094084388884e-06,
      "loss": 2.2399,
      "step": 2074
    },
    {
      "epoch": 0.9513984410820725,
      "grad_norm": 0.597168502364852,
      "learning_rate": 6.18490945912592e-06,
      "loss": 2.1786,
      "step": 2075
    },
    {
      "epoch": 0.951856946354883,
      "grad_norm": 0.5778310815861538,
      "learning_rate": 6.068999050065249e-06,
      "loss": 2.1787,
      "step": 2076
    },
    {
      "epoch": 0.9523154516276937,
      "grad_norm": 0.581332855848827,
      "learning_rate": 5.9541784369983586e-06,
      "loss": 2.1692,
      "step": 2077
    },
    {
      "epoch": 0.9527739569005044,
      "grad_norm": 0.6040942123893562,
      "learning_rate": 5.840447873262433e-06,
      "loss": 2.199,
      "step": 2078
    },
    {
      "epoch": 0.953232462173315,
      "grad_norm": 0.5729035345066193,
      "learning_rate": 5.727807609789471e-06,
      "loss": 2.2102,
      "step": 2079
    },
    {
      "epoch": 0.9536909674461256,
      "grad_norm": 0.6002944074455714,
      "learning_rate": 5.616257895105892e-06,
      "loss": 2.193,
      "step": 2080
    },
    {
      "epoch": 0.9541494727189362,
      "grad_norm": 0.5997307338415329,
      "learning_rate": 5.505798975331933e-06,
      "loss": 2.2864,
      "step": 2081
    },
    {
      "epoch": 0.9546079779917469,
      "grad_norm": 0.6234894458401673,
      "learning_rate": 5.396431094181198e-06,
      "loss": 2.322,
      "step": 2082
    },
    {
      "epoch": 0.9550664832645576,
      "grad_norm": 0.5937604521412572,
      "learning_rate": 5.288154492960107e-06,
      "loss": 2.2199,
      "step": 2083
    },
    {
      "epoch": 0.9555249885373682,
      "grad_norm": 0.6076939825152823,
      "learning_rate": 5.1809694105671155e-06,
      "loss": 2.0623,
      "step": 2084
    },
    {
      "epoch": 0.9559834938101788,
      "grad_norm": 0.5742104495370645,
      "learning_rate": 5.074876083492441e-06,
      "loss": 2.1742,
      "step": 2085
    },
    {
      "epoch": 0.9564419990829894,
      "grad_norm": 0.5967158282330145,
      "learning_rate": 4.96987474581767e-06,
      "loss": 2.2674,
      "step": 2086
    },
    {
      "epoch": 0.9569005043558001,
      "grad_norm": 0.5932489528675744,
      "learning_rate": 4.865965629214819e-06,
      "loss": 2.1661,
      "step": 2087
    },
    {
      "epoch": 0.9573590096286108,
      "grad_norm": 0.5964639132781234,
      "learning_rate": 4.763148962946218e-06,
      "loss": 2.264,
      "step": 2088
    },
    {
      "epoch": 0.9578175149014214,
      "grad_norm": 0.6243410673287549,
      "learning_rate": 4.661424973863681e-06,
      "loss": 2.3105,
      "step": 2089
    },
    {
      "epoch": 0.958276020174232,
      "grad_norm": 0.6063510710979041,
      "learning_rate": 4.560793886408398e-06,
      "loss": 2.1955,
      "step": 2090
    },
    {
      "epoch": 0.9587345254470426,
      "grad_norm": 0.6045223925786759,
      "learning_rate": 4.461255922609986e-06,
      "loss": 2.2877,
      "step": 2091
    },
    {
      "epoch": 0.9591930307198533,
      "grad_norm": 0.5961892365588761,
      "learning_rate": 4.362811302086267e-06,
      "loss": 2.1196,
      "step": 2092
    },
    {
      "epoch": 0.959651535992664,
      "grad_norm": 0.5973576585498954,
      "learning_rate": 4.265460242042885e-06,
      "loss": 2.2601,
      "step": 2093
    },
    {
      "epoch": 0.9601100412654745,
      "grad_norm": 0.5895854816955586,
      "learning_rate": 4.169202957272522e-06,
      "loss": 2.2041,
      "step": 2094
    },
    {
      "epoch": 0.9605685465382852,
      "grad_norm": 0.5865309150148015,
      "learning_rate": 4.074039660154738e-06,
      "loss": 2.1694,
      "step": 2095
    },
    {
      "epoch": 0.9610270518110958,
      "grad_norm": 0.6037167324301282,
      "learning_rate": 3.9799705606551325e-06,
      "loss": 2.2721,
      "step": 2096
    },
    {
      "epoch": 0.9614855570839065,
      "grad_norm": 0.6136575162665873,
      "learning_rate": 3.886995866325294e-06,
      "loss": 2.2397,
      "step": 2097
    },
    {
      "epoch": 0.9619440623567171,
      "grad_norm": 0.6003714240882005,
      "learning_rate": 3.795115782302072e-06,
      "loss": 2.2337,
      "step": 2098
    },
    {
      "epoch": 0.9624025676295277,
      "grad_norm": 0.5737737589594759,
      "learning_rate": 3.704330511307197e-06,
      "loss": 2.1997,
      "step": 2099
    },
    {
      "epoch": 0.9628610729023384,
      "grad_norm": 0.5901808989449836,
      "learning_rate": 3.614640253646828e-06,
      "loss": 2.1947,
      "step": 2100
    },
    {
      "epoch": 0.963319578175149,
      "grad_norm": 0.5928226599571412,
      "learning_rate": 3.5260452072110594e-06,
      "loss": 2.2535,
      "step": 2101
    },
    {
      "epoch": 0.9637780834479597,
      "grad_norm": 0.6068453506425516,
      "learning_rate": 3.4385455674737498e-06,
      "loss": 2.2223,
      "step": 2102
    },
    {
      "epoch": 0.9642365887207703,
      "grad_norm": 0.6113722248955551,
      "learning_rate": 3.3521415274915256e-06,
      "loss": 2.1598,
      "step": 2103
    },
    {
      "epoch": 0.9646950939935809,
      "grad_norm": 0.6133070267587418,
      "learning_rate": 3.2668332779041133e-06,
      "loss": 2.1989,
      "step": 2104
    },
    {
      "epoch": 0.9651535992663915,
      "grad_norm": 0.61635099131859,
      "learning_rate": 3.1826210069332838e-06,
      "loss": 2.2533,
      "step": 2105
    },
    {
      "epoch": 0.9656121045392022,
      "grad_norm": 0.6131683542012144,
      "learning_rate": 3.0995049003826324e-06,
      "loss": 2.3074,
      "step": 2106
    },
    {
      "epoch": 0.9660706098120129,
      "grad_norm": 0.6001876390347106,
      "learning_rate": 3.017485141637355e-06,
      "loss": 2.1261,
      "step": 2107
    },
    {
      "epoch": 0.9665291150848234,
      "grad_norm": 0.5953508895194182,
      "learning_rate": 2.9365619116636376e-06,
      "loss": 2.1827,
      "step": 2108
    },
    {
      "epoch": 0.9669876203576341,
      "grad_norm": 0.5906987162653359,
      "learning_rate": 2.856735389008269e-06,
      "loss": 2.2094,
      "step": 2109
    },
    {
      "epoch": 0.9674461256304447,
      "grad_norm": 0.5916587799578885,
      "learning_rate": 2.778005749798307e-06,
      "loss": 2.1901,
      "step": 2110
    },
    {
      "epoch": 0.9679046309032554,
      "grad_norm": 0.5873725841261851,
      "learning_rate": 2.700373167740744e-06,
      "loss": 2.2446,
      "step": 2111
    },
    {
      "epoch": 0.9683631361760661,
      "grad_norm": 0.5966175610564024,
      "learning_rate": 2.62383781412201e-06,
      "loss": 2.1942,
      "step": 2112
    },
    {
      "epoch": 0.9688216414488766,
      "grad_norm": 0.6074739280836922,
      "learning_rate": 2.5483998578076373e-06,
      "loss": 2.2376,
      "step": 2113
    },
    {
      "epoch": 0.9692801467216873,
      "grad_norm": 0.5920158134090864,
      "learning_rate": 2.4740594652418736e-06,
      "loss": 2.2758,
      "step": 2114
    },
    {
      "epoch": 0.9697386519944979,
      "grad_norm": 0.5992122432065287,
      "learning_rate": 2.4008168004472917e-06,
      "loss": 2.2437,
      "step": 2115
    },
    {
      "epoch": 0.9701971572673086,
      "grad_norm": 0.5878381140321294,
      "learning_rate": 2.3286720250246253e-06,
      "loss": 2.2311,
      "step": 2116
    },
    {
      "epoch": 0.9706556625401193,
      "grad_norm": 0.6113664389218717,
      "learning_rate": 2.2576252981520994e-06,
      "loss": 2.2151,
      "step": 2117
    },
    {
      "epoch": 0.9711141678129298,
      "grad_norm": 0.5889059821736871,
      "learning_rate": 2.1876767765853233e-06,
      "loss": 2.1702,
      "step": 2118
    },
    {
      "epoch": 0.9715726730857405,
      "grad_norm": 0.5873075317149659,
      "learning_rate": 2.118826614656788e-06,
      "loss": 2.1948,
      "step": 2119
    },
    {
      "epoch": 0.9720311783585511,
      "grad_norm": 0.5850948375422849,
      "learning_rate": 2.051074964275701e-06,
      "loss": 2.176,
      "step": 2120
    },
    {
      "epoch": 0.9724896836313618,
      "grad_norm": 0.600976379664613,
      "learning_rate": 1.984421974927375e-06,
      "loss": 2.176,
      "step": 2121
    },
    {
      "epoch": 0.9729481889041725,
      "grad_norm": 0.5929144631855262,
      "learning_rate": 1.9188677936731734e-06,
      "loss": 2.2813,
      "step": 2122
    },
    {
      "epoch": 0.973406694176983,
      "grad_norm": 0.583615063465773,
      "learning_rate": 1.8544125651501208e-06,
      "loss": 2.2406,
      "step": 2123
    },
    {
      "epoch": 0.9738651994497937,
      "grad_norm": 0.6139820920244003,
      "learning_rate": 1.7910564315704035e-06,
      "loss": 2.2588,
      "step": 2124
    },
    {
      "epoch": 0.9743237047226043,
      "grad_norm": 0.5952814570388237,
      "learning_rate": 1.7287995327214257e-06,
      "loss": 2.1813,
      "step": 2125
    },
    {
      "epoch": 0.974782209995415,
      "grad_norm": 0.625039448741108,
      "learning_rate": 1.6676420059649754e-06,
      "loss": 2.2524,
      "step": 2126
    },
    {
      "epoch": 0.9752407152682255,
      "grad_norm": 0.5924013832396782,
      "learning_rate": 1.6075839862374485e-06,
      "loss": 2.1459,
      "step": 2127
    },
    {
      "epoch": 0.9756992205410362,
      "grad_norm": 0.6023441544591254,
      "learning_rate": 1.5486256060492366e-06,
      "loss": 2.1995,
      "step": 2128
    },
    {
      "epoch": 0.9761577258138469,
      "grad_norm": 0.6051898346078377,
      "learning_rate": 1.4907669954844495e-06,
      "loss": 2.1451,
      "step": 2129
    },
    {
      "epoch": 0.9766162310866575,
      "grad_norm": 0.5797407588301066,
      "learning_rate": 1.434008282200805e-06,
      "loss": 2.2321,
      "step": 2130
    },
    {
      "epoch": 0.9770747363594682,
      "grad_norm": 0.6556076524250202,
      "learning_rate": 1.3783495914291844e-06,
      "loss": 2.2817,
      "step": 2131
    },
    {
      "epoch": 0.9775332416322787,
      "grad_norm": 0.5938619213864081,
      "learning_rate": 1.3237910459734104e-06,
      "loss": 2.289,
      "step": 2132
    },
    {
      "epoch": 0.9779917469050894,
      "grad_norm": 0.6044559251226573,
      "learning_rate": 1.270332766210025e-06,
      "loss": 2.2942,
      "step": 2133
    },
    {
      "epoch": 0.9784502521779,
      "grad_norm": 0.5877653117897519,
      "learning_rate": 1.2179748700879012e-06,
      "loss": 2.2774,
      "step": 2134
    },
    {
      "epoch": 0.9789087574507107,
      "grad_norm": 0.6175411127505417,
      "learning_rate": 1.1667174731280205e-06,
      "loss": 2.286,
      "step": 2135
    },
    {
      "epoch": 0.9793672627235214,
      "grad_norm": 0.5949766229594072,
      "learning_rate": 1.1165606884234182e-06,
      "loss": 2.2315,
      "step": 2136
    },
    {
      "epoch": 0.9798257679963319,
      "grad_norm": 0.6112070681679934,
      "learning_rate": 1.0675046266386268e-06,
      "loss": 2.2447,
      "step": 2137
    },
    {
      "epoch": 0.9802842732691426,
      "grad_norm": 0.6103223014016338,
      "learning_rate": 1.019549396009567e-06,
      "loss": 2.3323,
      "step": 2138
    },
    {
      "epoch": 0.9807427785419532,
      "grad_norm": 0.5970451842537561,
      "learning_rate": 9.726951023434348e-07,
      "loss": 2.1925,
      "step": 2139
    },
    {
      "epoch": 0.9812012838147639,
      "grad_norm": 0.5923195238905639,
      "learning_rate": 9.269418490182591e-07,
      "loss": 2.3058,
      "step": 2140
    },
    {
      "epoch": 0.9816597890875745,
      "grad_norm": 0.5827518747803279,
      "learning_rate": 8.822897369827332e-07,
      "loss": 2.2167,
      "step": 2141
    },
    {
      "epoch": 0.9821182943603851,
      "grad_norm": 0.5731020333988741,
      "learning_rate": 8.387388647561611e-07,
      "loss": 2.1759,
      "step": 2142
    },
    {
      "epoch": 0.9825767996331958,
      "grad_norm": 0.5835749171897039,
      "learning_rate": 7.962893284279016e-07,
      "loss": 2.1196,
      "step": 2143
    },
    {
      "epoch": 0.9830353049060064,
      "grad_norm": 0.6415686126236099,
      "learning_rate": 7.549412216574791e-07,
      "loss": 2.3214,
      "step": 2144
    },
    {
      "epoch": 0.9834938101788171,
      "grad_norm": 0.6203426783230577,
      "learning_rate": 7.146946356743067e-07,
      "loss": 2.1403,
      "step": 2145
    },
    {
      "epoch": 0.9839523154516276,
      "grad_norm": 0.6033031745434978,
      "learning_rate": 6.755496592773524e-07,
      "loss": 2.2101,
      "step": 2146
    },
    {
      "epoch": 0.9844108207244383,
      "grad_norm": 0.615860520019511,
      "learning_rate": 6.375063788349733e-07,
      "loss": 2.2973,
      "step": 2147
    },
    {
      "epoch": 0.984869325997249,
      "grad_norm": 0.6381284186241466,
      "learning_rate": 6.005648782848594e-07,
      "loss": 2.1744,
      "step": 2148
    },
    {
      "epoch": 0.9853278312700596,
      "grad_norm": 0.5795639186145077,
      "learning_rate": 5.647252391337565e-07,
      "loss": 2.2489,
      "step": 2149
    },
    {
      "epoch": 0.9857863365428703,
      "grad_norm": 0.616795010981769,
      "learning_rate": 5.299875404572441e-07,
      "loss": 2.2345,
      "step": 2150
    },
    {
      "epoch": 0.9862448418156808,
      "grad_norm": 0.5828446753250427,
      "learning_rate": 4.963518588996796e-07,
      "loss": 2.1887,
      "step": 2151
    },
    {
      "epoch": 0.9867033470884915,
      "grad_norm": 0.5740923346027146,
      "learning_rate": 4.638182686738657e-07,
      "loss": 2.1878,
      "step": 2152
    },
    {
      "epoch": 0.9871618523613022,
      "grad_norm": 0.5917913443319522,
      "learning_rate": 4.3238684156110543e-07,
      "loss": 2.1747,
      "step": 2153
    },
    {
      "epoch": 0.9876203576341128,
      "grad_norm": 0.6138757789269582,
      "learning_rate": 4.020576469108139e-07,
      "loss": 2.1363,
      "step": 2154
    },
    {
      "epoch": 0.9880788629069235,
      "grad_norm": 0.6174263254421815,
      "learning_rate": 3.7283075164046274e-07,
      "loss": 2.2091,
      "step": 2155
    },
    {
      "epoch": 0.988537368179734,
      "grad_norm": 0.5954029654951943,
      "learning_rate": 3.4470622023557995e-07,
      "loss": 2.1855,
      "step": 2156
    },
    {
      "epoch": 0.9889958734525447,
      "grad_norm": 0.5692141821985357,
      "learning_rate": 3.176841147492504e-07,
      "loss": 2.1675,
      "step": 2157
    },
    {
      "epoch": 0.9894543787253554,
      "grad_norm": 0.5849067439192963,
      "learning_rate": 2.9176449480244895e-07,
      "loss": 2.1519,
      "step": 2158
    },
    {
      "epoch": 0.989912883998166,
      "grad_norm": 0.6049771603223657,
      "learning_rate": 2.6694741758342967e-07,
      "loss": 2.2734,
      "step": 2159
    },
    {
      "epoch": 0.9903713892709766,
      "grad_norm": 0.5879662718927848,
      "learning_rate": 2.432329378478926e-07,
      "loss": 2.2412,
      "step": 2160
    },
    {
      "epoch": 0.9908298945437872,
      "grad_norm": 0.5811369517138495,
      "learning_rate": 2.2062110791892798e-07,
      "loss": 2.1552,
      "step": 2161
    },
    {
      "epoch": 0.9912883998165979,
      "grad_norm": 0.6023864286142392,
      "learning_rate": 1.9911197768662792e-07,
      "loss": 2.2295,
      "step": 2162
    },
    {
      "epoch": 0.9917469050894085,
      "grad_norm": 0.5850375753957721,
      "learning_rate": 1.7870559460814173e-07,
      "loss": 2.3172,
      "step": 2163
    },
    {
      "epoch": 0.9922054103622192,
      "grad_norm": 0.5880266926972135,
      "learning_rate": 1.5940200370750947e-07,
      "loss": 2.2504,
      "step": 2164
    },
    {
      "epoch": 0.9926639156350298,
      "grad_norm": 0.6180466989469464,
      "learning_rate": 1.4120124757577291e-07,
      "loss": 2.3117,
      "step": 2165
    },
    {
      "epoch": 0.9931224209078404,
      "grad_norm": 0.620032699998332,
      "learning_rate": 1.2410336637047603e-07,
      "loss": 2.1308,
      "step": 2166
    },
    {
      "epoch": 0.9935809261806511,
      "grad_norm": 0.5716898615150566,
      "learning_rate": 1.081083978159425e-07,
      "loss": 2.2906,
      "step": 2167
    },
    {
      "epoch": 0.9940394314534617,
      "grad_norm": 0.5801556776004053,
      "learning_rate": 9.321637720310915e-08,
      "loss": 2.1712,
      "step": 2168
    },
    {
      "epoch": 0.9944979367262724,
      "grad_norm": 0.6276423632965349,
      "learning_rate": 7.942733738924845e-08,
      "loss": 2.299,
      "step": 2169
    },
    {
      "epoch": 0.994956441999083,
      "grad_norm": 0.5900074692312758,
      "learning_rate": 6.6741308798135e-08,
      "loss": 2.2172,
      "step": 2170
    },
    {
      "epoch": 0.9954149472718936,
      "grad_norm": 0.5913104096462739,
      "learning_rate": 5.5158319419934546e-08,
      "loss": 2.2489,
      "step": 2171
    },
    {
      "epoch": 0.9958734525447043,
      "grad_norm": 0.5963024027097789,
      "learning_rate": 4.4678394810981906e-08,
      "loss": 2.2175,
      "step": 2172
    },
    {
      "epoch": 0.9963319578175149,
      "grad_norm": 0.5791896395059174,
      "learning_rate": 3.5301558093947527e-08,
      "loss": 2.2948,
      "step": 2173
    },
    {
      "epoch": 0.9967904630903255,
      "grad_norm": 0.5749962760662819,
      "learning_rate": 2.7027829957559925e-08,
      "loss": 2.2321,
      "step": 2174
    },
    {
      "epoch": 0.9972489683631361,
      "grad_norm": 0.609413897905501,
      "learning_rate": 1.985722865682771e-08,
      "loss": 2.2003,
      "step": 2175
    },
    {
      "epoch": 0.9977074736359468,
      "grad_norm": 0.5784210497897095,
      "learning_rate": 1.3789770012762048e-08,
      "loss": 2.178,
      "step": 2176
    },
    {
      "epoch": 0.9981659789087575,
      "grad_norm": 0.5766649589209378,
      "learning_rate": 8.825467412376665e-09,
      "loss": 2.2511,
      "step": 2177
    },
    {
      "epoch": 0.9986244841815681,
      "grad_norm": 0.6169396940204713,
      "learning_rate": 4.96433180879885e-09,
      "loss": 2.2331,
      "step": 2178
    },
    {
      "epoch": 0.9990829894543787,
      "grad_norm": 0.6087936938397724,
      "learning_rate": 2.206371721158451e-09,
      "loss": 2.1983,
      "step": 2179
    },
    {
      "epoch": 0.9995414947271893,
      "grad_norm": 0.6016049160146183,
      "learning_rate": 5.515932345323549e-10,
      "loss": 2.1808,
      "step": 2180
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5732229675230027,
      "learning_rate": 0.0,
      "loss": 2.2346,
      "step": 2181
    },
    {
      "epoch": 1.0,
      "step": 2181,
      "total_flos": 6.576985065966797e+17,
      "train_loss": 2.462100135465654,
      "train_runtime": 6617.514,
      "train_samples_per_second": 84.341,
      "train_steps_per_second": 0.33
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2181,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 24000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.576985065966797e+17,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}