{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 1662,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0012040939193257074,
      "grad_norm": 0.3228262662887573,
      "learning_rate": 0.0,
      "loss": 1.1978780031204224,
      "step": 1
    },
    {
      "epoch": 0.002408187838651415,
      "grad_norm": 0.2775137722492218,
      "learning_rate": 1.1904761904761904e-06,
      "loss": 1.16929292678833,
      "step": 2
    },
    {
      "epoch": 0.003612281757977122,
      "grad_norm": 0.2929379642009735,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 1.4060571193695068,
      "step": 3
    },
    {
      "epoch": 0.00481637567730283,
      "grad_norm": 0.2869538962841034,
      "learning_rate": 3.5714285714285714e-06,
      "loss": 1.1399140357971191,
      "step": 4
    },
    {
      "epoch": 0.006020469596628537,
      "grad_norm": 0.31194964051246643,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 1.1719307899475098,
      "step": 5
    },
    {
      "epoch": 0.007224563515954244,
      "grad_norm": 0.3013097941875458,
      "learning_rate": 5.9523809523809525e-06,
      "loss": 1.178497314453125,
      "step": 6
    },
    {
      "epoch": 0.008428657435279952,
      "grad_norm": 0.3038809895515442,
      "learning_rate": 7.142857142857143e-06,
      "loss": 1.085801362991333,
      "step": 7
    },
    {
      "epoch": 0.00963275135460566,
      "grad_norm": 0.3142983019351959,
      "learning_rate": 8.333333333333334e-06,
      "loss": 1.3388087749481201,
      "step": 8
    },
    {
      "epoch": 0.010836845273931367,
      "grad_norm": 0.36046934127807617,
      "learning_rate": 9.523809523809523e-06,
      "loss": 1.0800137519836426,
      "step": 9
    },
    {
      "epoch": 0.012040939193257074,
      "grad_norm": 0.32254818081855774,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 1.2095422744750977,
      "step": 10
    },
    {
      "epoch": 0.013245033112582781,
      "grad_norm": 0.26548296213150024,
      "learning_rate": 1.1904761904761905e-05,
      "loss": 1.1551485061645508,
      "step": 11
    },
    {
      "epoch": 0.014449127031908489,
      "grad_norm": 0.24723590910434723,
      "learning_rate": 1.3095238095238096e-05,
      "loss": 1.0877487659454346,
      "step": 12
    },
    {
      "epoch": 0.015653220951234198,
      "grad_norm": 0.21321748197078705,
      "learning_rate": 1.4285714285714285e-05,
      "loss": 1.072063684463501,
      "step": 13
    },
    {
      "epoch": 0.016857314870559904,
      "grad_norm": 0.19131672382354736,
      "learning_rate": 1.5476190476190476e-05,
      "loss": 1.131504774093628,
      "step": 14
    },
    {
      "epoch": 0.018061408789885613,
      "grad_norm": 0.19355866312980652,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 1.1454461812973022,
      "step": 15
    },
    {
      "epoch": 0.01926550270921132,
      "grad_norm": 0.18354736268520355,
      "learning_rate": 1.785714285714286e-05,
      "loss": 1.086208701133728,
      "step": 16
    },
    {
      "epoch": 0.020469596628537028,
      "grad_norm": 0.16947612166404724,
      "learning_rate": 1.9047619047619046e-05,
      "loss": 1.1802699565887451,
      "step": 17
    },
    {
      "epoch": 0.021673690547862733,
      "grad_norm": 0.16902878880500793,
      "learning_rate": 2.023809523809524e-05,
      "loss": 1.0732470750808716,
      "step": 18
    },
    {
      "epoch": 0.022877784467188442,
      "grad_norm": 0.1820913702249527,
      "learning_rate": 2.1428571428571428e-05,
      "loss": 1.142699956893921,
      "step": 19
    },
    {
      "epoch": 0.024081878386514148,
      "grad_norm": 0.19912295043468475,
      "learning_rate": 2.261904761904762e-05,
      "loss": 1.0827000141143799,
      "step": 20
    },
    {
      "epoch": 0.025285972305839857,
      "grad_norm": 0.1683034896850586,
      "learning_rate": 2.380952380952381e-05,
      "loss": 1.008629322052002,
      "step": 21
    },
    {
      "epoch": 0.026490066225165563,
      "grad_norm": 0.16864903271198273,
      "learning_rate": 2.5e-05,
      "loss": 0.943488359451294,
      "step": 22
    },
    {
      "epoch": 0.027694160144491272,
      "grad_norm": 0.18769897520542145,
      "learning_rate": 2.6190476190476192e-05,
      "loss": 0.8735959529876709,
      "step": 23
    },
    {
      "epoch": 0.028898254063816978,
      "grad_norm": 0.18208791315555573,
      "learning_rate": 2.7380952380952383e-05,
      "loss": 1.0424199104309082,
      "step": 24
    },
    {
      "epoch": 0.030102347983142687,
      "grad_norm": 0.2056523561477661,
      "learning_rate": 2.857142857142857e-05,
      "loss": 1.0286569595336914,
      "step": 25
    },
    {
      "epoch": 0.031306441902468396,
      "grad_norm": 0.2174437940120697,
      "learning_rate": 2.9761904761904762e-05,
      "loss": 1.1197870969772339,
      "step": 26
    },
    {
      "epoch": 0.0325105358217941,
      "grad_norm": 0.19437749683856964,
      "learning_rate": 3.095238095238095e-05,
      "loss": 1.0514827966690063,
      "step": 27
    },
    {
      "epoch": 0.03371462974111981,
      "grad_norm": 0.16864313185214996,
      "learning_rate": 3.2142857142857144e-05,
      "loss": 1.0640960931777954,
      "step": 28
    },
    {
      "epoch": 0.034918723660445516,
      "grad_norm": 0.16689112782478333,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 0.9558049440383911,
      "step": 29
    },
    {
      "epoch": 0.036122817579771226,
      "grad_norm": 0.15765565633773804,
      "learning_rate": 3.4523809523809526e-05,
      "loss": 0.8983189463615417,
      "step": 30
    },
    {
      "epoch": 0.03732691149909693,
      "grad_norm": 0.1815180480480194,
      "learning_rate": 3.571428571428572e-05,
      "loss": 0.9899806976318359,
      "step": 31
    },
    {
      "epoch": 0.03853100541842264,
      "grad_norm": 0.16196444630622864,
      "learning_rate": 3.690476190476191e-05,
      "loss": 0.9349527359008789,
      "step": 32
    },
    {
      "epoch": 0.039735099337748346,
      "grad_norm": 0.1598745584487915,
      "learning_rate": 3.809523809523809e-05,
      "loss": 1.024754285812378,
      "step": 33
    },
    {
      "epoch": 0.040939193257074055,
      "grad_norm": 0.15310589969158173,
      "learning_rate": 3.928571428571429e-05,
      "loss": 0.9699112176895142,
      "step": 34
    },
    {
      "epoch": 0.04214328717639976,
      "grad_norm": 0.16002674400806427,
      "learning_rate": 4.047619047619048e-05,
      "loss": 0.9108308553695679,
      "step": 35
    },
    {
      "epoch": 0.04334738109572547,
      "grad_norm": 0.17263197898864746,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.9592783451080322,
      "step": 36
    },
    {
      "epoch": 0.044551475015051176,
      "grad_norm": 0.1654745191335678,
      "learning_rate": 4.2857142857142856e-05,
      "loss": 0.9511594772338867,
      "step": 37
    },
    {
      "epoch": 0.045755568934376885,
      "grad_norm": 0.18950365483760834,
      "learning_rate": 4.404761904761905e-05,
      "loss": 1.0216031074523926,
      "step": 38
    },
    {
      "epoch": 0.04695966285370259,
      "grad_norm": 0.13648928701877594,
      "learning_rate": 4.523809523809524e-05,
      "loss": 0.8888837099075317,
      "step": 39
    },
    {
      "epoch": 0.048163756773028296,
      "grad_norm": 0.1466016322374344,
      "learning_rate": 4.642857142857143e-05,
      "loss": 0.8743679523468018,
      "step": 40
    },
    {
      "epoch": 0.049367850692354005,
      "grad_norm": 0.16377733647823334,
      "learning_rate": 4.761904761904762e-05,
      "loss": 0.9161079525947571,
      "step": 41
    },
    {
      "epoch": 0.050571944611679714,
      "grad_norm": 0.1535230427980423,
      "learning_rate": 4.880952380952381e-05,
      "loss": 0.9315277338027954,
      "step": 42
    },
    {
      "epoch": 0.05177603853100542,
      "grad_norm": 0.16339963674545288,
      "learning_rate": 5e-05,
      "loss": 0.9525159597396851,
      "step": 43
    },
    {
      "epoch": 0.052980132450331126,
      "grad_norm": 0.16096949577331543,
      "learning_rate": 5.119047619047619e-05,
      "loss": 0.9592565298080444,
      "step": 44
    },
    {
      "epoch": 0.054184226369656835,
      "grad_norm": 0.15226922929286957,
      "learning_rate": 5.2380952380952384e-05,
      "loss": 0.9001655578613281,
      "step": 45
    },
    {
      "epoch": 0.055388320288982544,
      "grad_norm": 0.1582118719816208,
      "learning_rate": 5.3571428571428575e-05,
      "loss": 0.9657700657844543,
      "step": 46
    },
    {
      "epoch": 0.056592414208308246,
      "grad_norm": 0.18544508516788483,
      "learning_rate": 5.4761904761904766e-05,
      "loss": 0.9573442935943604,
      "step": 47
    },
    {
      "epoch": 0.057796508127633955,
      "grad_norm": 0.16871900856494904,
      "learning_rate": 5.595238095238096e-05,
      "loss": 0.8906192779541016,
      "step": 48
    },
    {
      "epoch": 0.059000602046959665,
      "grad_norm": 0.14332857728004456,
      "learning_rate": 5.714285714285714e-05,
      "loss": 0.8418011665344238,
      "step": 49
    },
    {
      "epoch": 0.060204695966285374,
      "grad_norm": 0.15429189801216125,
      "learning_rate": 5.833333333333334e-05,
      "loss": 0.7943311929702759,
      "step": 50
    },
    {
      "epoch": 0.061408789885611076,
      "grad_norm": 0.18795368075370789,
      "learning_rate": 5.9523809523809524e-05,
      "loss": 0.8945696353912354,
      "step": 51
    },
    {
      "epoch": 0.06261288380493679,
      "grad_norm": 0.16643480956554413,
      "learning_rate": 6.0714285714285715e-05,
      "loss": 0.9036577939987183,
      "step": 52
    },
    {
      "epoch": 0.0638169777242625,
      "grad_norm": 0.19310061633586884,
      "learning_rate": 6.19047619047619e-05,
      "loss": 0.9542645812034607,
      "step": 53
    },
    {
      "epoch": 0.0650210716435882,
      "grad_norm": 0.19109384715557098,
      "learning_rate": 6.30952380952381e-05,
      "loss": 0.9353885650634766,
      "step": 54
    },
    {
      "epoch": 0.06622516556291391,
      "grad_norm": 0.1707829236984253,
      "learning_rate": 6.428571428571429e-05,
      "loss": 1.0063107013702393,
      "step": 55
    },
    {
      "epoch": 0.06742925948223961,
      "grad_norm": 0.18073849380016327,
      "learning_rate": 6.547619047619048e-05,
      "loss": 0.8856950998306274,
      "step": 56
    },
    {
      "epoch": 0.06863335340156532,
      "grad_norm": 0.16882732510566711,
      "learning_rate": 6.666666666666667e-05,
      "loss": 0.9654364585876465,
      "step": 57
    },
    {
      "epoch": 0.06983744732089103,
      "grad_norm": 0.18666990101337433,
      "learning_rate": 6.785714285714286e-05,
      "loss": 0.877626359462738,
      "step": 58
    },
    {
      "epoch": 0.07104154124021674,
      "grad_norm": 0.17450827360153198,
      "learning_rate": 6.904761904761905e-05,
      "loss": 0.9396389126777649,
      "step": 59
    },
    {
      "epoch": 0.07224563515954245,
      "grad_norm": 0.16580334305763245,
      "learning_rate": 7.023809523809524e-05,
      "loss": 0.9616518020629883,
      "step": 60
    },
    {
      "epoch": 0.07344972907886815,
      "grad_norm": 0.17088699340820312,
      "learning_rate": 7.142857142857143e-05,
      "loss": 0.8649123907089233,
      "step": 61
    },
    {
      "epoch": 0.07465382299819386,
      "grad_norm": 0.17133638262748718,
      "learning_rate": 7.261904761904762e-05,
      "loss": 0.9410029649734497,
      "step": 62
    },
    {
      "epoch": 0.07585791691751957,
      "grad_norm": 0.1581082046031952,
      "learning_rate": 7.380952380952382e-05,
      "loss": 0.9109706878662109,
      "step": 63
    },
    {
      "epoch": 0.07706201083684527,
      "grad_norm": 0.1860303431749344,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.9495095610618591,
      "step": 64
    },
    {
      "epoch": 0.07826610475617098,
      "grad_norm": 0.16582341492176056,
      "learning_rate": 7.619047619047618e-05,
      "loss": 0.8647099733352661,
      "step": 65
    },
    {
      "epoch": 0.07947019867549669,
      "grad_norm": 0.17522723972797394,
      "learning_rate": 7.738095238095239e-05,
      "loss": 0.9968418478965759,
      "step": 66
    },
    {
      "epoch": 0.0806742925948224,
      "grad_norm": 0.21471670269966125,
      "learning_rate": 7.857142857142858e-05,
      "loss": 0.9037712216377258,
      "step": 67
    },
    {
      "epoch": 0.08187838651414811,
      "grad_norm": 0.17316053807735443,
      "learning_rate": 7.976190476190477e-05,
      "loss": 0.9034141898155212,
      "step": 68
    },
    {
      "epoch": 0.08308248043347381,
      "grad_norm": 0.18495336174964905,
      "learning_rate": 8.095238095238096e-05,
      "loss": 0.8833793997764587,
      "step": 69
    },
    {
      "epoch": 0.08428657435279951,
      "grad_norm": 0.1660023331642151,
      "learning_rate": 8.214285714285714e-05,
      "loss": 0.8803894519805908,
      "step": 70
    },
    {
      "epoch": 0.08549066827212523,
      "grad_norm": 0.18014533817768097,
      "learning_rate": 8.333333333333334e-05,
      "loss": 0.9559789299964905,
      "step": 71
    },
    {
      "epoch": 0.08669476219145093,
      "grad_norm": 0.16337937116622925,
      "learning_rate": 8.452380952380952e-05,
      "loss": 0.9799549579620361,
      "step": 72
    },
    {
      "epoch": 0.08789885611077664,
      "grad_norm": 0.15280377864837646,
      "learning_rate": 8.571428571428571e-05,
      "loss": 0.9033361673355103,
      "step": 73
    },
    {
      "epoch": 0.08910295003010235,
      "grad_norm": 0.15562526881694794,
      "learning_rate": 8.690476190476192e-05,
      "loss": 0.8616568446159363,
      "step": 74
    },
    {
      "epoch": 0.09030704394942805,
      "grad_norm": 0.17806018888950348,
      "learning_rate": 8.80952380952381e-05,
      "loss": 0.9578632116317749,
      "step": 75
    },
    {
      "epoch": 0.09151113786875377,
      "grad_norm": 0.1533149778842926,
      "learning_rate": 8.92857142857143e-05,
      "loss": 0.845272958278656,
      "step": 76
    },
    {
      "epoch": 0.09271523178807947,
      "grad_norm": 0.181115984916687,
      "learning_rate": 9.047619047619048e-05,
      "loss": 0.8684731721878052,
      "step": 77
    },
    {
      "epoch": 0.09391932570740517,
      "grad_norm": 0.18638168275356293,
      "learning_rate": 9.166666666666667e-05,
      "loss": 0.9435790181159973,
      "step": 78
    },
    {
      "epoch": 0.09512341962673089,
      "grad_norm": 0.16785700619220734,
      "learning_rate": 9.285714285714286e-05,
      "loss": 0.9153584241867065,
      "step": 79
    },
    {
      "epoch": 0.09632751354605659,
      "grad_norm": 0.16825895011425018,
      "learning_rate": 9.404761904761905e-05,
      "loss": 0.9684340953826904,
      "step": 80
    },
    {
      "epoch": 0.0975316074653823,
      "grad_norm": 0.16373522579669952,
      "learning_rate": 9.523809523809524e-05,
      "loss": 0.8836486339569092,
      "step": 81
    },
    {
      "epoch": 0.09873570138470801,
      "grad_norm": 0.1769053041934967,
      "learning_rate": 9.642857142857143e-05,
      "loss": 0.9499003887176514,
      "step": 82
    },
    {
      "epoch": 0.09993979530403371,
      "grad_norm": 0.14187361299991608,
      "learning_rate": 9.761904761904762e-05,
      "loss": 0.8693102598190308,
      "step": 83
    },
    {
      "epoch": 0.10114388922335943,
      "grad_norm": 0.15836596488952637,
      "learning_rate": 9.880952380952381e-05,
      "loss": 0.886056125164032,
      "step": 84
    },
    {
      "epoch": 0.10234798314268513,
      "grad_norm": 0.1684958040714264,
      "learning_rate": 0.0001,
      "loss": 0.8352786302566528,
      "step": 85
    },
    {
      "epoch": 0.10355207706201083,
      "grad_norm": 0.16528484225273132,
      "learning_rate": 9.999990091096183e-05,
      "loss": 1.0620546340942383,
      "step": 86
    },
    {
      "epoch": 0.10475617098133655,
      "grad_norm": 0.2082303911447525,
      "learning_rate": 9.999960364424007e-05,
      "loss": 0.8344641923904419,
      "step": 87
    },
    {
      "epoch": 0.10596026490066225,
      "grad_norm": 0.15356576442718506,
      "learning_rate": 9.999910820101293e-05,
      "loss": 0.7941401600837708,
      "step": 88
    },
    {
      "epoch": 0.10716435881998795,
      "grad_norm": 0.1841215193271637,
      "learning_rate": 9.999841458324413e-05,
      "loss": 0.8641438484191895,
      "step": 89
    },
    {
      "epoch": 0.10836845273931367,
      "grad_norm": 0.18341666460037231,
      "learning_rate": 9.999752279368291e-05,
      "loss": 0.9263696670532227,
      "step": 90
    },
    {
      "epoch": 0.10957254665863937,
      "grad_norm": 0.17533987760543823,
      "learning_rate": 9.999643283586388e-05,
      "loss": 0.9713923931121826,
      "step": 91
    },
    {
      "epoch": 0.11077664057796509,
      "grad_norm": 0.17124471068382263,
      "learning_rate": 9.999514471410719e-05,
      "loss": 0.8557345867156982,
      "step": 92
    },
    {
      "epoch": 0.11198073449729079,
      "grad_norm": 0.1648399978876114,
      "learning_rate": 9.999365843351838e-05,
      "loss": 0.8542652726173401,
      "step": 93
    },
    {
      "epoch": 0.11318482841661649,
      "grad_norm": 0.178237184882164,
      "learning_rate": 9.999197399998841e-05,
      "loss": 0.8339974284172058,
      "step": 94
    },
    {
      "epoch": 0.11438892233594221,
      "grad_norm": 0.18585343658924103,
      "learning_rate": 9.999009142019364e-05,
      "loss": 0.9198064804077148,
      "step": 95
    },
    {
      "epoch": 0.11559301625526791,
      "grad_norm": 0.17155712842941284,
      "learning_rate": 9.99880107015958e-05,
      "loss": 0.9260415434837341,
      "step": 96
    },
    {
      "epoch": 0.11679711017459361,
      "grad_norm": 0.16283343732357025,
      "learning_rate": 9.998573185244192e-05,
      "loss": 0.8462975025177002,
      "step": 97
    },
    {
      "epoch": 0.11800120409391933,
      "grad_norm": 0.14888884127140045,
      "learning_rate": 9.998325488176437e-05,
      "loss": 0.8500807285308838,
      "step": 98
    },
    {
      "epoch": 0.11920529801324503,
      "grad_norm": 0.1418602615594864,
      "learning_rate": 9.99805797993808e-05,
      "loss": 0.8023315072059631,
      "step": 99
    },
    {
      "epoch": 0.12040939193257075,
      "grad_norm": 0.1508619487285614,
      "learning_rate": 9.997770661589403e-05,
      "loss": 0.8490456938743591,
      "step": 100
    },
    {
      "epoch": 0.12161348585189645,
      "grad_norm": 0.1628970205783844,
      "learning_rate": 9.997463534269213e-05,
      "loss": 0.8226766586303711,
      "step": 101
    },
    {
      "epoch": 0.12281757977122215,
      "grad_norm": 0.17291021347045898,
      "learning_rate": 9.997136599194825e-05,
      "loss": 0.8973463773727417,
      "step": 102
    },
    {
      "epoch": 0.12402167369054787,
      "grad_norm": 0.1452181488275528,
      "learning_rate": 9.996789857662068e-05,
      "loss": 0.8462653160095215,
      "step": 103
    },
    {
      "epoch": 0.12522576760987358,
      "grad_norm": 0.14783787727355957,
      "learning_rate": 9.996423311045273e-05,
      "loss": 0.8368968963623047,
      "step": 104
    },
    {
      "epoch": 0.12642986152919927,
      "grad_norm": 0.15426383912563324,
      "learning_rate": 9.99603696079727e-05,
      "loss": 0.8936138153076172,
      "step": 105
    },
    {
      "epoch": 0.127633955448525,
      "grad_norm": 0.13673485815525055,
      "learning_rate": 9.995630808449383e-05,
      "loss": 0.7002023458480835,
      "step": 106
    },
    {
      "epoch": 0.1288380493678507,
      "grad_norm": 0.17347267270088196,
      "learning_rate": 9.99520485561142e-05,
      "loss": 0.856157660484314,
      "step": 107
    },
    {
      "epoch": 0.1300421432871764,
      "grad_norm": 0.14660269021987915,
      "learning_rate": 9.994759103971673e-05,
      "loss": 0.8079285621643066,
      "step": 108
    },
    {
      "epoch": 0.1312462372065021,
      "grad_norm": 0.15511545538902283,
      "learning_rate": 9.994293555296904e-05,
      "loss": 0.6953619718551636,
      "step": 109
    },
    {
      "epoch": 0.13245033112582782,
      "grad_norm": 0.15847957134246826,
      "learning_rate": 9.993808211432346e-05,
      "loss": 0.8541761040687561,
      "step": 110
    },
    {
      "epoch": 0.1336544250451535,
      "grad_norm": 0.16614878177642822,
      "learning_rate": 9.993303074301689e-05,
      "loss": 1.0016850233078003,
      "step": 111
    },
    {
      "epoch": 0.13485851896447923,
      "grad_norm": 0.15124846994876862,
      "learning_rate": 9.992778145907073e-05,
      "loss": 0.8165380954742432,
      "step": 112
    },
    {
      "epoch": 0.13606261288380495,
      "grad_norm": 0.17631462216377258,
      "learning_rate": 9.992233428329088e-05,
      "loss": 0.8048033714294434,
      "step": 113
    },
    {
      "epoch": 0.13726670680313063,
      "grad_norm": 0.16587044298648834,
      "learning_rate": 9.99166892372675e-05,
      "loss": 0.8249691724777222,
      "step": 114
    },
    {
      "epoch": 0.13847080072245635,
      "grad_norm": 0.18603375554084778,
      "learning_rate": 9.991084634337511e-05,
      "loss": 0.9583686590194702,
      "step": 115
    },
    {
      "epoch": 0.13967489464178207,
      "grad_norm": 0.15955089032649994,
      "learning_rate": 9.99048056247724e-05,
      "loss": 0.8414074182510376,
      "step": 116
    },
    {
      "epoch": 0.14087898856110775,
      "grad_norm": 0.1664114147424698,
      "learning_rate": 9.989856710540209e-05,
      "loss": 0.8285802006721497,
      "step": 117
    },
    {
      "epoch": 0.14208308248043347,
      "grad_norm": 0.16105720400810242,
      "learning_rate": 9.989213080999097e-05,
      "loss": 0.8771381974220276,
      "step": 118
    },
    {
      "epoch": 0.1432871763997592,
      "grad_norm": 0.17184105515480042,
      "learning_rate": 9.988549676404965e-05,
      "loss": 0.9105034470558167,
      "step": 119
    },
    {
      "epoch": 0.1444912703190849,
      "grad_norm": 0.20621643960475922,
      "learning_rate": 9.987866499387264e-05,
      "loss": 0.8696529865264893,
      "step": 120
    },
    {
      "epoch": 0.1456953642384106,
      "grad_norm": 0.1607387661933899,
      "learning_rate": 9.987163552653802e-05,
      "loss": 0.88026362657547,
      "step": 121
    },
    {
      "epoch": 0.1468994581577363,
      "grad_norm": 0.19336579740047455,
      "learning_rate": 9.986440838990755e-05,
      "loss": 0.8111530542373657,
      "step": 122
    },
    {
      "epoch": 0.14810355207706202,
      "grad_norm": 0.1640203297138214,
      "learning_rate": 9.985698361262642e-05,
      "loss": 0.9487682580947876,
      "step": 123
    },
    {
      "epoch": 0.1493076459963877,
      "grad_norm": 0.16107501089572906,
      "learning_rate": 9.984936122412319e-05,
      "loss": 0.9055774807929993,
      "step": 124
    },
    {
      "epoch": 0.15051173991571343,
      "grad_norm": 0.1338699460029602,
      "learning_rate": 9.984154125460969e-05,
      "loss": 0.7875896096229553,
      "step": 125
    },
    {
      "epoch": 0.15171583383503914,
      "grad_norm": 0.16280344128608704,
      "learning_rate": 9.983352373508081e-05,
      "loss": 0.7907180190086365,
      "step": 126
    },
    {
      "epoch": 0.15291992775436483,
      "grad_norm": 0.14342188835144043,
      "learning_rate": 9.982530869731451e-05,
      "loss": 0.8160648345947266,
      "step": 127
    },
    {
      "epoch": 0.15412402167369055,
      "grad_norm": 0.15245693922042847,
      "learning_rate": 9.98168961738716e-05,
      "loss": 0.8188925981521606,
      "step": 128
    },
    {
      "epoch": 0.15532811559301626,
      "grad_norm": 0.16324079036712646,
      "learning_rate": 9.980828619809561e-05,
      "loss": 0.79526686668396,
      "step": 129
    },
    {
      "epoch": 0.15653220951234195,
      "grad_norm": 0.13547760248184204,
      "learning_rate": 9.979947880411273e-05,
      "loss": 0.8501705527305603,
      "step": 130
    },
    {
      "epoch": 0.15773630343166767,
      "grad_norm": 0.15600715577602386,
      "learning_rate": 9.97904740268316e-05,
      "loss": 0.8191217184066772,
      "step": 131
    },
    {
      "epoch": 0.15894039735099338,
      "grad_norm": 0.168145552277565,
      "learning_rate": 9.978127190194322e-05,
      "loss": 0.8244034647941589,
      "step": 132
    },
    {
      "epoch": 0.16014449127031907,
      "grad_norm": 0.1524210423231125,
      "learning_rate": 9.977187246592076e-05,
      "loss": 0.766787052154541,
      "step": 133
    },
    {
      "epoch": 0.1613485851896448,
      "grad_norm": 0.17103226482868195,
      "learning_rate": 9.976227575601947e-05,
      "loss": 0.920512318611145,
      "step": 134
    },
    {
      "epoch": 0.1625526791089705,
      "grad_norm": 0.15900297462940216,
      "learning_rate": 9.97524818102765e-05,
      "loss": 0.8282946348190308,
      "step": 135
    },
    {
      "epoch": 0.16375677302829622,
      "grad_norm": 0.17010007798671722,
      "learning_rate": 9.974249066751077e-05,
      "loss": 0.9069222211837769,
      "step": 136
    },
    {
      "epoch": 0.1649608669476219,
      "grad_norm": 0.16585490107536316,
      "learning_rate": 9.973230236732276e-05,
      "loss": 0.8059027194976807,
      "step": 137
    },
    {
      "epoch": 0.16616496086694763,
      "grad_norm": 0.15215714275836945,
      "learning_rate": 9.972191695009446e-05,
      "loss": 0.7079100012779236,
      "step": 138
    },
    {
      "epoch": 0.16736905478627334,
      "grad_norm": 0.16049662232398987,
      "learning_rate": 9.971133445698908e-05,
      "loss": 0.7887976169586182,
      "step": 139
    },
    {
      "epoch": 0.16857314870559903,
      "grad_norm": 0.15005750954151154,
      "learning_rate": 9.9700554929951e-05,
      "loss": 0.892037034034729,
      "step": 140
    },
    {
      "epoch": 0.16977724262492475,
      "grad_norm": 0.15534451603889465,
      "learning_rate": 9.968957841170554e-05,
      "loss": 0.8484441041946411,
      "step": 141
    },
    {
      "epoch": 0.17098133654425046,
      "grad_norm": 0.18025758862495422,
      "learning_rate": 9.967840494575879e-05,
      "loss": 0.8157638907432556,
      "step": 142
    },
    {
      "epoch": 0.17218543046357615,
      "grad_norm": 0.15553221106529236,
      "learning_rate": 9.966703457639748e-05,
      "loss": 0.7651965618133545,
      "step": 143
    },
    {
      "epoch": 0.17338952438290187,
      "grad_norm": 0.1625150442123413,
      "learning_rate": 9.965546734868875e-05,
      "loss": 0.7751995921134949,
      "step": 144
    },
    {
      "epoch": 0.17459361830222758,
      "grad_norm": 0.22047273814678192,
      "learning_rate": 9.964370330848005e-05,
      "loss": 0.7471305131912231,
      "step": 145
    },
    {
      "epoch": 0.17579771222155327,
      "grad_norm": 0.16728994250297546,
      "learning_rate": 9.963174250239888e-05,
      "loss": 0.8730856776237488,
      "step": 146
    },
    {
      "epoch": 0.177001806140879,
      "grad_norm": 0.16989463567733765,
      "learning_rate": 9.96195849778526e-05,
      "loss": 0.6675388813018799,
      "step": 147
    },
    {
      "epoch": 0.1782059000602047,
      "grad_norm": 0.23268599808216095,
      "learning_rate": 9.960723078302832e-05,
      "loss": 0.8339875936508179,
      "step": 148
    },
    {
      "epoch": 0.1794099939795304,
      "grad_norm": 0.15975242853164673,
      "learning_rate": 9.959467996689264e-05,
      "loss": 0.8011846542358398,
      "step": 149
    },
    {
      "epoch": 0.1806140878988561,
      "grad_norm": 0.1567070484161377,
      "learning_rate": 9.95819325791915e-05,
      "loss": 0.7133753895759583,
      "step": 150
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 0.1732303500175476,
      "learning_rate": 9.956898867044999e-05,
      "loss": 0.7509013414382935,
      "step": 151
    },
    {
      "epoch": 0.18302227573750754,
      "grad_norm": 0.17202892899513245,
      "learning_rate": 9.955584829197203e-05,
      "loss": 0.7432981133460999,
      "step": 152
    },
    {
      "epoch": 0.18422636965683323,
      "grad_norm": 0.17465098202228546,
      "learning_rate": 9.954251149584036e-05,
      "loss": 0.738294243812561,
      "step": 153
    },
    {
      "epoch": 0.18543046357615894,
      "grad_norm": 0.15495821833610535,
      "learning_rate": 9.952897833491617e-05,
      "loss": 0.8399054408073425,
      "step": 154
    },
    {
      "epoch": 0.18663455749548466,
      "grad_norm": 0.17095357179641724,
      "learning_rate": 9.951524886283899e-05,
      "loss": 0.8912245631217957,
      "step": 155
    },
    {
      "epoch": 0.18783865141481035,
      "grad_norm": 0.1855614334344864,
      "learning_rate": 9.950132313402642e-05,
      "loss": 0.8787100315093994,
      "step": 156
    },
    {
      "epoch": 0.18904274533413606,
      "grad_norm": 0.15850698947906494,
      "learning_rate": 9.948720120367394e-05,
      "loss": 0.8694636225700378,
      "step": 157
    },
    {
      "epoch": 0.19024683925346178,
      "grad_norm": 0.17447032034397125,
      "learning_rate": 9.947288312775471e-05,
      "loss": 0.8903257846832275,
      "step": 158
    },
    {
      "epoch": 0.19145093317278747,
      "grad_norm": 0.16062240302562714,
      "learning_rate": 9.945836896301927e-05,
      "loss": 0.824906051158905,
      "step": 159
    },
    {
      "epoch": 0.19265502709211318,
      "grad_norm": 0.19191031157970428,
      "learning_rate": 9.944365876699544e-05,
      "loss": 0.8124850988388062,
      "step": 160
    },
    {
      "epoch": 0.1938591210114389,
      "grad_norm": 0.15191487967967987,
      "learning_rate": 9.942875259798796e-05,
      "loss": 0.7818014621734619,
      "step": 161
    },
    {
      "epoch": 0.1950632149307646,
      "grad_norm": 0.16724996268749237,
      "learning_rate": 9.941365051507836e-05,
      "loss": 0.8735147714614868,
      "step": 162
    },
    {
      "epoch": 0.1962673088500903,
      "grad_norm": 0.15072351694107056,
      "learning_rate": 9.939835257812468e-05,
      "loss": 0.7777504920959473,
      "step": 163
    },
    {
      "epoch": 0.19747140276941602,
      "grad_norm": 0.1557697355747223,
      "learning_rate": 9.938285884776121e-05,
      "loss": 0.8688787221908569,
      "step": 164
    },
    {
      "epoch": 0.1986754966887417,
      "grad_norm": 0.15751446783542633,
      "learning_rate": 9.936716938539834e-05,
      "loss": 0.9143678545951843,
      "step": 165
    },
    {
      "epoch": 0.19987959060806743,
      "grad_norm": 0.16234248876571655,
      "learning_rate": 9.93512842532222e-05,
      "loss": 0.8578541874885559,
      "step": 166
    },
    {
      "epoch": 0.20108368452739314,
      "grad_norm": 0.15054866671562195,
      "learning_rate": 9.93352035141945e-05,
      "loss": 0.7462522387504578,
      "step": 167
    },
    {
      "epoch": 0.20228777844671886,
      "grad_norm": 0.15021702647209167,
      "learning_rate": 9.931892723205221e-05,
      "loss": 0.753405749797821,
      "step": 168
    },
    {
      "epoch": 0.20349187236604455,
      "grad_norm": 0.16552209854125977,
      "learning_rate": 9.93024554713074e-05,
      "loss": 0.7950703501701355,
      "step": 169
    },
    {
      "epoch": 0.20469596628537026,
      "grad_norm": 0.16023050248622894,
      "learning_rate": 9.92857882972469e-05,
      "loss": 0.8190892338752747,
      "step": 170
    },
    {
      "epoch": 0.20590006020469598,
      "grad_norm": 0.14788106083869934,
      "learning_rate": 9.926892577593208e-05,
      "loss": 0.7566771507263184,
      "step": 171
    },
    {
      "epoch": 0.20710415412402167,
      "grad_norm": 0.1426115483045578,
      "learning_rate": 9.925186797419858e-05,
      "loss": 0.7555860877037048,
      "step": 172
    },
    {
      "epoch": 0.20830824804334738,
      "grad_norm": 0.15389211475849152,
      "learning_rate": 9.923461495965607e-05,
      "loss": 0.780582845211029,
      "step": 173
    },
    {
      "epoch": 0.2095123419626731,
      "grad_norm": 0.15657348930835724,
      "learning_rate": 9.92171668006879e-05,
      "loss": 0.7064439058303833,
      "step": 174
    },
    {
      "epoch": 0.2107164358819988,
      "grad_norm": 0.1489691436290741,
      "learning_rate": 9.919952356645092e-05,
      "loss": 0.8028382658958435,
      "step": 175
    },
    {
      "epoch": 0.2119205298013245,
      "grad_norm": 0.15523386001586914,
      "learning_rate": 9.91816853268752e-05,
      "loss": 0.8292052745819092,
      "step": 176
    },
    {
      "epoch": 0.21312462372065022,
      "grad_norm": 0.17593027651309967,
      "learning_rate": 9.91636521526637e-05,
      "loss": 0.7599755525588989,
      "step": 177
    },
    {
      "epoch": 0.2143287176399759,
      "grad_norm": 0.1579512059688568,
      "learning_rate": 9.9145424115292e-05,
      "loss": 0.8224988579750061,
      "step": 178
    },
    {
      "epoch": 0.21553281155930162,
      "grad_norm": 0.13823267817497253,
      "learning_rate": 9.912700128700805e-05,
      "loss": 0.7569247484207153,
      "step": 179
    },
    {
      "epoch": 0.21673690547862734,
      "grad_norm": 0.1548740416765213,
      "learning_rate": 9.910838374083185e-05,
      "loss": 0.8395345211029053,
      "step": 180
    },
    {
      "epoch": 0.21794099939795303,
      "grad_norm": 0.15824519097805023,
      "learning_rate": 9.908957155055523e-05,
      "loss": 0.8274417519569397,
      "step": 181
    },
    {
      "epoch": 0.21914509331727874,
      "grad_norm": 0.22959287464618683,
      "learning_rate": 9.907056479074142e-05,
      "loss": 0.782774031162262,
      "step": 182
    },
    {
      "epoch": 0.22034918723660446,
      "grad_norm": 0.16858088970184326,
      "learning_rate": 9.905136353672492e-05,
      "loss": 0.7703776359558105,
      "step": 183
    },
    {
      "epoch": 0.22155328115593018,
      "grad_norm": 0.1553880274295807,
      "learning_rate": 9.903196786461106e-05,
      "loss": 0.7607191801071167,
      "step": 184
    },
    {
      "epoch": 0.22275737507525586,
      "grad_norm": 0.13512618839740753,
      "learning_rate": 9.901237785127577e-05,
      "loss": 0.7218681573867798,
      "step": 185
    },
    {
      "epoch": 0.22396146899458158,
      "grad_norm": 0.1593676060438156,
      "learning_rate": 9.89925935743653e-05,
      "loss": 0.8243957757949829,
      "step": 186
    },
    {
      "epoch": 0.2251655629139073,
      "grad_norm": 0.14236581325531006,
      "learning_rate": 9.897261511229583e-05,
      "loss": 0.8230786919593811,
      "step": 187
    },
    {
      "epoch": 0.22636965683323299,
      "grad_norm": 0.1548842340707779,
      "learning_rate": 9.895244254425322e-05,
      "loss": 0.7949791550636292,
      "step": 188
    },
    {
      "epoch": 0.2275737507525587,
      "grad_norm": 0.16853167116641998,
      "learning_rate": 9.893207595019269e-05,
      "loss": 0.8834179639816284,
      "step": 189
    },
    {
      "epoch": 0.22877784467188442,
      "grad_norm": 0.1502329260110855,
      "learning_rate": 9.891151541083852e-05,
      "loss": 0.8546383380889893,
      "step": 190
    },
    {
      "epoch": 0.2299819385912101,
      "grad_norm": 0.15839482843875885,
      "learning_rate": 9.889076100768362e-05,
      "loss": 0.8507488965988159,
      "step": 191
    },
    {
      "epoch": 0.23118603251053582,
      "grad_norm": 0.161220982670784,
      "learning_rate": 9.886981282298937e-05,
      "loss": 0.7661045789718628,
      "step": 192
    },
    {
      "epoch": 0.23239012642986154,
      "grad_norm": 0.16198964416980743,
      "learning_rate": 9.884867093978519e-05,
      "loss": 0.9086393117904663,
      "step": 193
    },
    {
      "epoch": 0.23359422034918723,
      "grad_norm": 0.16452190279960632,
      "learning_rate": 9.882733544186825e-05,
      "loss": 0.8532902598381042,
      "step": 194
    },
    {
      "epoch": 0.23479831426851294,
      "grad_norm": 0.1316503882408142,
      "learning_rate": 9.880580641380306e-05,
      "loss": 0.6958003044128418,
      "step": 195
    },
    {
      "epoch": 0.23600240818783866,
      "grad_norm": 0.14050406217575073,
      "learning_rate": 9.878408394092129e-05,
      "loss": 0.8419471979141235,
      "step": 196
    },
    {
      "epoch": 0.23720650210716435,
      "grad_norm": 0.13732720911502838,
      "learning_rate": 9.876216810932129e-05,
      "loss": 0.7769860625267029,
      "step": 197
    },
    {
      "epoch": 0.23841059602649006,
      "grad_norm": 0.13452723622322083,
      "learning_rate": 9.87400590058678e-05,
      "loss": 0.6639009118080139,
      "step": 198
    },
    {
      "epoch": 0.23961468994581578,
      "grad_norm": 0.1409616768360138,
      "learning_rate": 9.871775671819162e-05,
      "loss": 0.8293867111206055,
      "step": 199
    },
    {
      "epoch": 0.2408187838651415,
      "grad_norm": 0.14871157705783844,
      "learning_rate": 9.869526133468923e-05,
      "loss": 0.7684170007705688,
      "step": 200
    },
    {
      "epoch": 0.24202287778446718,
      "grad_norm": 0.13409508764743805,
      "learning_rate": 9.867257294452245e-05,
      "loss": 0.7892642617225647,
      "step": 201
    },
    {
      "epoch": 0.2432269717037929,
      "grad_norm": 0.14893606305122375,
      "learning_rate": 9.864969163761817e-05,
      "loss": 0.7485178709030151,
      "step": 202
    },
    {
      "epoch": 0.24443106562311862,
      "grad_norm": 0.15578876435756683,
      "learning_rate": 9.86266175046678e-05,
      "loss": 0.8133466243743896,
      "step": 203
    },
    {
      "epoch": 0.2456351595424443,
      "grad_norm": 0.16745084524154663,
      "learning_rate": 9.86033506371271e-05,
      "loss": 0.8918555974960327,
      "step": 204
    },
    {
      "epoch": 0.24683925346177002,
      "grad_norm": 0.15875816345214844,
      "learning_rate": 9.857989112721574e-05,
      "loss": 0.9448569416999817,
      "step": 205
    },
    {
      "epoch": 0.24804334738109574,
      "grad_norm": 0.15180985629558563,
      "learning_rate": 9.855623906791693e-05,
      "loss": 0.7574045062065125,
      "step": 206
    },
    {
      "epoch": 0.24924744130042142,
      "grad_norm": 0.1434941291809082,
      "learning_rate": 9.853239455297705e-05,
      "loss": 0.7993340492248535,
      "step": 207
    },
    {
      "epoch": 0.25045153521974717,
      "grad_norm": 0.16140271723270416,
      "learning_rate": 9.850835767690532e-05,
      "loss": 0.7958835959434509,
      "step": 208
    },
    {
      "epoch": 0.25165562913907286,
      "grad_norm": 0.1590467393398285,
      "learning_rate": 9.848412853497337e-05,
      "loss": 0.774200975894928,
      "step": 209
    },
    {
      "epoch": 0.25285972305839854,
      "grad_norm": 0.16585272550582886,
      "learning_rate": 9.845970722321489e-05,
      "loss": 0.9030719995498657,
      "step": 210
    },
    {
      "epoch": 0.2540638169777243,
      "grad_norm": 0.16352301836013794,
      "learning_rate": 9.843509383842525e-05,
      "loss": 0.8171011209487915,
      "step": 211
    },
    {
      "epoch": 0.25526791089705,
      "grad_norm": 0.14271050691604614,
      "learning_rate": 9.841028847816112e-05,
      "loss": 0.8035110831260681,
      "step": 212
    },
    {
      "epoch": 0.25647200481637566,
      "grad_norm": 0.15654458105564117,
      "learning_rate": 9.838529124074006e-05,
      "loss": 0.8152414560317993,
      "step": 213
    },
    {
      "epoch": 0.2576760987357014,
      "grad_norm": 0.14926882088184357,
      "learning_rate": 9.836010222524018e-05,
      "loss": 0.7910839915275574,
      "step": 214
    },
    {
      "epoch": 0.2588801926550271,
      "grad_norm": 0.14828084409236908,
      "learning_rate": 9.833472153149968e-05,
      "loss": 0.8133422136306763,
      "step": 215
    },
    {
      "epoch": 0.2600842865743528,
      "grad_norm": 0.16326478123664856,
      "learning_rate": 9.830914926011651e-05,
      "loss": 0.7649755477905273,
      "step": 216
    },
    {
      "epoch": 0.26128838049367853,
      "grad_norm": 0.14547421038150787,
      "learning_rate": 9.828338551244794e-05,
      "loss": 0.7777073979377747,
      "step": 217
    },
    {
      "epoch": 0.2624924744130042,
      "grad_norm": 0.13898834586143494,
      "learning_rate": 9.825743039061015e-05,
      "loss": 0.7566221952438354,
      "step": 218
    },
    {
      "epoch": 0.2636965683323299,
      "grad_norm": 0.1541513353586197,
      "learning_rate": 9.823128399747788e-05,
      "loss": 0.868842601776123,
      "step": 219
    },
    {
      "epoch": 0.26490066225165565,
      "grad_norm": 0.1467703878879547,
      "learning_rate": 9.820494643668396e-05,
      "loss": 0.8981370329856873,
      "step": 220
    },
    {
      "epoch": 0.26610475617098134,
      "grad_norm": 0.14359429478645325,
      "learning_rate": 9.817841781261894e-05,
      "loss": 0.8630922436714172,
      "step": 221
    },
    {
      "epoch": 0.267308850090307,
      "grad_norm": 0.15819276869297028,
      "learning_rate": 9.815169823043066e-05,
      "loss": 0.9258560538291931,
      "step": 222
    },
    {
      "epoch": 0.26851294400963277,
      "grad_norm": 0.15252874791622162,
      "learning_rate": 9.812478779602381e-05,
      "loss": 0.8332533836364746,
      "step": 223
    },
    {
      "epoch": 0.26971703792895846,
      "grad_norm": 0.13327279686927795,
      "learning_rate": 9.809768661605956e-05,
      "loss": 0.7024679183959961,
      "step": 224
    },
    {
      "epoch": 0.27092113184828415,
      "grad_norm": 0.18224233388900757,
      "learning_rate": 9.80703947979551e-05,
      "loss": 0.9371347427368164,
      "step": 225
    },
    {
      "epoch": 0.2721252257676099,
      "grad_norm": 0.15701915323734283,
      "learning_rate": 9.804291244988324e-05,
      "loss": 0.8759557008743286,
      "step": 226
    },
    {
      "epoch": 0.2733293196869356,
      "grad_norm": 0.15363524854183197,
      "learning_rate": 9.801523968077195e-05,
      "loss": 0.7347216606140137,
      "step": 227
    },
    {
      "epoch": 0.27453341360626127,
      "grad_norm": 0.14150449633598328,
      "learning_rate": 9.798737660030397e-05,
      "loss": 0.7347065210342407,
      "step": 228
    },
    {
      "epoch": 0.275737507525587,
      "grad_norm": 0.16662348806858063,
      "learning_rate": 9.79593233189163e-05,
      "loss": 0.9140637516975403,
      "step": 229
    },
    {
      "epoch": 0.2769416014449127,
      "grad_norm": 0.14806142449378967,
      "learning_rate": 9.793107994779988e-05,
      "loss": 0.9076550006866455,
      "step": 230
    },
    {
      "epoch": 0.2781456953642384,
      "grad_norm": 0.1513150930404663,
      "learning_rate": 9.790264659889903e-05,
      "loss": 0.88444584608078,
      "step": 231
    },
    {
      "epoch": 0.27934978928356413,
      "grad_norm": 0.12769709527492523,
      "learning_rate": 9.787402338491108e-05,
      "loss": 0.7734633088111877,
      "step": 232
    },
    {
      "epoch": 0.2805538832028898,
      "grad_norm": 0.14580336213111877,
      "learning_rate": 9.78452104192859e-05,
      "loss": 0.8569942712783813,
      "step": 233
    },
    {
      "epoch": 0.2817579771222155,
      "grad_norm": 0.14663909375667572,
      "learning_rate": 9.781620781622546e-05,
      "loss": 0.7928228378295898,
      "step": 234
    },
    {
      "epoch": 0.28296207104154125,
      "grad_norm": 0.14392022788524628,
      "learning_rate": 9.778701569068336e-05,
      "loss": 0.8053445816040039,
      "step": 235
    },
    {
      "epoch": 0.28416616496086694,
      "grad_norm": 0.13541586697101593,
      "learning_rate": 9.775763415836439e-05,
      "loss": 0.8910754919052124,
      "step": 236
    },
    {
      "epoch": 0.28537025888019263,
      "grad_norm": 0.14465011656284332,
      "learning_rate": 9.772806333572405e-05,
      "loss": 0.8262342214584351,
      "step": 237
    },
    {
      "epoch": 0.2865743527995184,
      "grad_norm": 0.1482345163822174,
      "learning_rate": 9.76983033399681e-05,
      "loss": 0.7707383632659912,
      "step": 238
    },
    {
      "epoch": 0.28777844671884406,
      "grad_norm": 0.1594555675983429,
      "learning_rate": 9.766835428905213e-05,
      "loss": 0.813245952129364,
      "step": 239
    },
    {
      "epoch": 0.2889825406381698,
      "grad_norm": 0.16228754818439484,
      "learning_rate": 9.763821630168106e-05,
      "loss": 0.8050968647003174,
      "step": 240
    },
    {
      "epoch": 0.2901866345574955,
      "grad_norm": 0.14761123061180115,
      "learning_rate": 9.760788949730866e-05,
      "loss": 0.7867326140403748,
      "step": 241
    },
    {
      "epoch": 0.2913907284768212,
      "grad_norm": 0.14403793215751648,
      "learning_rate": 9.757737399613706e-05,
      "loss": 0.8061637282371521,
      "step": 242
    },
    {
      "epoch": 0.2925948223961469,
      "grad_norm": 0.15832845866680145,
      "learning_rate": 9.754666991911633e-05,
      "loss": 0.8290490508079529,
      "step": 243
    },
    {
      "epoch": 0.2937989163154726,
      "grad_norm": 0.16302058100700378,
      "learning_rate": 9.751577738794398e-05,
      "loss": 0.8388375043869019,
      "step": 244
    },
    {
      "epoch": 0.2950030102347983,
      "grad_norm": 0.15200331807136536,
      "learning_rate": 9.748469652506446e-05,
      "loss": 0.8658491969108582,
      "step": 245
    },
    {
      "epoch": 0.29620710415412405,
      "grad_norm": 0.14951592683792114,
      "learning_rate": 9.745342745366867e-05,
      "loss": 0.794253945350647,
      "step": 246
    },
    {
      "epoch": 0.29741119807344973,
      "grad_norm": 0.15377479791641235,
      "learning_rate": 9.74219702976935e-05,
      "loss": 0.7936686277389526,
      "step": 247
    },
    {
      "epoch": 0.2986152919927754,
      "grad_norm": 0.15265917778015137,
      "learning_rate": 9.739032518182134e-05,
      "loss": 0.7467873692512512,
      "step": 248
    },
    {
      "epoch": 0.29981938591210117,
      "grad_norm": 0.161261186003685,
      "learning_rate": 9.735849223147952e-05,
      "loss": 0.8245280981063843,
      "step": 249
    },
    {
      "epoch": 0.30102347983142685,
      "grad_norm": 0.1292843073606491,
      "learning_rate": 9.732647157283994e-05,
      "loss": 0.660576581954956,
      "step": 250
    },
    {
      "epoch": 0.30222757375075254,
      "grad_norm": 0.14917084574699402,
      "learning_rate": 9.729426333281841e-05,
      "loss": 0.7710285186767578,
      "step": 251
    },
    {
      "epoch": 0.3034316676700783,
      "grad_norm": 0.12222438305616379,
      "learning_rate": 9.726186763907432e-05,
      "loss": 0.7755141854286194,
      "step": 252
    },
    {
      "epoch": 0.304635761589404,
      "grad_norm": 0.1457638442516327,
      "learning_rate": 9.722928462000995e-05,
      "loss": 0.8170596361160278,
      "step": 253
    },
    {
      "epoch": 0.30583985550872966,
      "grad_norm": 0.14369265735149384,
      "learning_rate": 9.71965144047701e-05,
      "loss": 0.8565990924835205,
      "step": 254
    },
    {
      "epoch": 0.3070439494280554,
      "grad_norm": 0.17217011749744415,
      "learning_rate": 9.716355712324156e-05,
      "loss": 0.8595105409622192,
      "step": 255
    },
    {
      "epoch": 0.3082480433473811,
      "grad_norm": 0.14087100327014923,
      "learning_rate": 9.713041290605254e-05,
      "loss": 0.7477849125862122,
      "step": 256
    },
    {
      "epoch": 0.3094521372667068,
      "grad_norm": 0.15932901203632355,
      "learning_rate": 9.709708188457218e-05,
      "loss": 0.7864524126052856,
      "step": 257
    },
    {
      "epoch": 0.3106562311860325,
      "grad_norm": 0.14455156028270721,
      "learning_rate": 9.706356419091003e-05,
      "loss": 0.7899723649024963,
      "step": 258
    },
    {
      "epoch": 0.3118603251053582,
      "grad_norm": 0.15004439651966095,
      "learning_rate": 9.702985995791554e-05,
      "loss": 0.837183952331543,
      "step": 259
    },
    {
      "epoch": 0.3130644190246839,
      "grad_norm": 0.14573058485984802,
      "learning_rate": 9.69959693191775e-05,
      "loss": 0.7990862727165222,
      "step": 260
    },
    {
      "epoch": 0.31426851294400965,
      "grad_norm": 0.14866943657398224,
      "learning_rate": 9.696189240902353e-05,
      "loss": 0.8250412940979004,
      "step": 261
    },
    {
      "epoch": 0.31547260686333534,
      "grad_norm": 0.15827123820781708,
      "learning_rate": 9.69276293625196e-05,
      "loss": 0.8502783179283142,
      "step": 262
    },
    {
      "epoch": 0.316676700782661,
      "grad_norm": 0.15928424894809723,
      "learning_rate": 9.689318031546935e-05,
      "loss": 0.864741325378418,
      "step": 263
    },
    {
      "epoch": 0.31788079470198677,
      "grad_norm": 0.14357474446296692,
      "learning_rate": 9.685854540441374e-05,
      "loss": 0.8982298374176025,
      "step": 264
    },
    {
      "epoch": 0.31908488862131246,
      "grad_norm": 0.14526790380477905,
      "learning_rate": 9.682372476663037e-05,
      "loss": 0.6971644163131714,
      "step": 265
    },
    {
      "epoch": 0.32028898254063815,
      "grad_norm": 0.14952056109905243,
      "learning_rate": 9.678871854013296e-05,
      "loss": 0.7566766142845154,
      "step": 266
    },
    {
      "epoch": 0.3214930764599639,
      "grad_norm": 0.17004022002220154,
      "learning_rate": 9.675352686367086e-05,
      "loss": 0.8731152415275574,
      "step": 267
    },
    {
      "epoch": 0.3226971703792896,
      "grad_norm": 0.15685567259788513,
      "learning_rate": 9.671814987672842e-05,
      "loss": 0.7922989130020142,
      "step": 268
    },
    {
      "epoch": 0.32390126429861527,
      "grad_norm": 0.157073974609375,
      "learning_rate": 9.668258771952453e-05,
      "loss": 0.7555691599845886,
      "step": 269
    },
    {
      "epoch": 0.325105358217941,
      "grad_norm": 0.1426512449979782,
      "learning_rate": 9.664684053301199e-05,
      "loss": 0.774568498134613,
      "step": 270
    },
    {
      "epoch": 0.3263094521372667,
      "grad_norm": 0.13459919393062592,
      "learning_rate": 9.661090845887693e-05,
      "loss": 0.7332807183265686,
      "step": 271
    },
    {
      "epoch": 0.32751354605659244,
      "grad_norm": 0.14334864914417267,
      "learning_rate": 9.657479163953839e-05,
      "loss": 0.7398945689201355,
      "step": 272
    },
    {
      "epoch": 0.32871763997591813,
      "grad_norm": 0.14441382884979248,
      "learning_rate": 9.653849021814759e-05,
      "loss": 0.7343026995658875,
      "step": 273
    },
    {
      "epoch": 0.3299217338952438,
      "grad_norm": 0.16644948720932007,
      "learning_rate": 9.650200433858741e-05,
      "loss": 0.8654240369796753,
      "step": 274
    },
    {
      "epoch": 0.33112582781456956,
      "grad_norm": 0.1455552726984024,
      "learning_rate": 9.646533414547193e-05,
      "loss": 0.80277019739151,
      "step": 275
    },
    {
      "epoch": 0.33232992173389525,
      "grad_norm": 0.14409396052360535,
      "learning_rate": 9.642847978414569e-05,
      "loss": 0.8152948021888733,
      "step": 276
    },
    {
      "epoch": 0.33353401565322094,
      "grad_norm": 0.15238624811172485,
      "learning_rate": 9.639144140068324e-05,
      "loss": 0.8617125153541565,
      "step": 277
    },
    {
      "epoch": 0.3347381095725467,
      "grad_norm": 0.14458949863910675,
      "learning_rate": 9.635421914188845e-05,
      "loss": 0.8282997608184814,
      "step": 278
    },
    {
      "epoch": 0.33594220349187237,
      "grad_norm": 0.1437564641237259,
      "learning_rate": 9.631681315529408e-05,
      "loss": 0.7632128000259399,
      "step": 279
    },
    {
      "epoch": 0.33714629741119806,
      "grad_norm": 0.15033510327339172,
      "learning_rate": 9.627922358916102e-05,
      "loss": 0.7594578862190247,
      "step": 280
    },
    {
      "epoch": 0.3383503913305238,
      "grad_norm": 0.14697270095348358,
      "learning_rate": 9.624145059247787e-05,
      "loss": 0.7671185731887817,
      "step": 281
    },
    {
      "epoch": 0.3395544852498495,
      "grad_norm": 0.16469521820545197,
      "learning_rate": 9.620349431496018e-05,
      "loss": 0.9038112759590149,
      "step": 282
    },
    {
      "epoch": 0.3407585791691752,
      "grad_norm": 0.15169623494148254,
      "learning_rate": 9.616535490705004e-05,
      "loss": 0.7705464363098145,
      "step": 283
    },
    {
      "epoch": 0.3419626730885009,
      "grad_norm": 0.13130046427249908,
      "learning_rate": 9.612703251991528e-05,
      "loss": 0.6947124004364014,
      "step": 284
    },
    {
      "epoch": 0.3431667670078266,
      "grad_norm": 0.15084530413150787,
      "learning_rate": 9.60885273054491e-05,
      "loss": 0.7870749235153198,
      "step": 285
    },
    {
      "epoch": 0.3443708609271523,
      "grad_norm": 0.16112524271011353,
      "learning_rate": 9.604983941626924e-05,
      "loss": 0.7660699486732483,
      "step": 286
    },
    {
      "epoch": 0.34557495484647804,
      "grad_norm": 0.17407681047916412,
      "learning_rate": 9.601096900571757e-05,
      "loss": 0.7593162059783936,
      "step": 287
    },
    {
      "epoch": 0.34677904876580373,
      "grad_norm": 0.13153290748596191,
      "learning_rate": 9.597191622785932e-05,
      "loss": 0.7584017515182495,
      "step": 288
    },
    {
      "epoch": 0.3479831426851294,
      "grad_norm": 0.15213483572006226,
      "learning_rate": 9.593268123748259e-05,
      "loss": 0.7646967172622681,
      "step": 289
    },
    {
      "epoch": 0.34918723660445516,
      "grad_norm": 0.16267520189285278,
      "learning_rate": 9.589326419009768e-05,
      "loss": 0.8550824522972107,
      "step": 290
    },
    {
      "epoch": 0.35039133052378085,
      "grad_norm": 0.14083021879196167,
      "learning_rate": 9.585366524193648e-05,
      "loss": 0.8166961073875427,
      "step": 291
    },
    {
      "epoch": 0.35159542444310654,
      "grad_norm": 0.15070311725139618,
      "learning_rate": 9.581388454995187e-05,
      "loss": 0.7632104754447937,
      "step": 292
    },
    {
      "epoch": 0.3527995183624323,
      "grad_norm": 0.133270263671875,
      "learning_rate": 9.577392227181706e-05,
      "loss": 0.7489652037620544,
      "step": 293
    },
    {
      "epoch": 0.354003612281758,
      "grad_norm": 0.1417430192232132,
      "learning_rate": 9.573377856592499e-05,
      "loss": 0.7489148378372192,
      "step": 294
    },
    {
      "epoch": 0.35520770620108366,
      "grad_norm": 0.1535170078277588,
      "learning_rate": 9.569345359138771e-05,
      "loss": 0.824393630027771,
      "step": 295
    },
    {
      "epoch": 0.3564118001204094,
      "grad_norm": 0.14139412343502045,
      "learning_rate": 9.565294750803575e-05,
      "loss": 0.8096657991409302,
      "step": 296
    },
    {
      "epoch": 0.3576158940397351,
      "grad_norm": 0.14470556378364563,
      "learning_rate": 9.561226047641745e-05,
      "loss": 0.8006949424743652,
      "step": 297
    },
    {
      "epoch": 0.3588199879590608,
      "grad_norm": 0.15374445915222168,
      "learning_rate": 9.557139265779838e-05,
      "loss": 0.8048600554466248,
      "step": 298
    },
    {
      "epoch": 0.3600240818783865,
      "grad_norm": 0.15500962734222412,
      "learning_rate": 9.553034421416062e-05,
      "loss": 0.7799292802810669,
      "step": 299
    },
    {
      "epoch": 0.3612281757977122,
      "grad_norm": 0.1318802386522293,
      "learning_rate": 9.548911530820226e-05,
      "loss": 0.7221542000770569,
      "step": 300
    },
    {
      "epoch": 0.3624322697170379,
      "grad_norm": 0.16511353850364685,
      "learning_rate": 9.544770610333655e-05,
      "loss": 0.8232554197311401,
      "step": 301
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 0.13170106709003448,
      "learning_rate": 9.540611676369145e-05,
      "loss": 0.721223771572113,
      "step": 302
    },
    {
      "epoch": 0.36484045755568933,
      "grad_norm": 0.14849360287189484,
      "learning_rate": 9.536434745410885e-05,
      "loss": 0.8091297149658203,
      "step": 303
    },
    {
      "epoch": 0.3660445514750151,
      "grad_norm": 0.156411811709404,
      "learning_rate": 9.5322398340144e-05,
      "loss": 0.6771798729896545,
      "step": 304
    },
    {
      "epoch": 0.36724864539434077,
      "grad_norm": 0.1678953915834427,
      "learning_rate": 9.528026958806477e-05,
      "loss": 0.7591527700424194,
      "step": 305
    },
    {
      "epoch": 0.36845273931366646,
      "grad_norm": 0.15519757568836212,
      "learning_rate": 9.523796136485109e-05,
      "loss": 0.7611120343208313,
      "step": 306
    },
    {
      "epoch": 0.3696568332329922,
      "grad_norm": 0.17793579399585724,
      "learning_rate": 9.519547383819416e-05,
      "loss": 0.8761749267578125,
      "step": 307
    },
    {
      "epoch": 0.3708609271523179,
      "grad_norm": 0.22375215590000153,
      "learning_rate": 9.515280717649594e-05,
      "loss": 0.9409160017967224,
      "step": 308
    },
    {
      "epoch": 0.3720650210716436,
      "grad_norm": 0.13814345002174377,
      "learning_rate": 9.510996154886839e-05,
      "loss": 0.7814185619354248,
      "step": 309
    },
    {
      "epoch": 0.3732691149909693,
      "grad_norm": 0.12590160965919495,
      "learning_rate": 9.506693712513274e-05,
      "loss": 0.7038348913192749,
      "step": 310
    },
    {
      "epoch": 0.374473208910295,
      "grad_norm": 0.1413682997226715,
      "learning_rate": 9.502373407581897e-05,
      "loss": 0.8262088298797607,
      "step": 311
    },
    {
      "epoch": 0.3756773028296207,
      "grad_norm": 0.17772600054740906,
      "learning_rate": 9.498035257216501e-05,
      "loss": 0.9201569557189941,
      "step": 312
    },
    {
      "epoch": 0.37688139674894644,
      "grad_norm": 0.1873147040605545,
      "learning_rate": 9.493679278611616e-05,
      "loss": 0.7338141202926636,
      "step": 313
    },
    {
      "epoch": 0.37808549066827213,
      "grad_norm": 0.13902495801448822,
      "learning_rate": 9.489305489032425e-05,
      "loss": 0.7284054756164551,
      "step": 314
    },
    {
      "epoch": 0.3792895845875978,
      "grad_norm": 0.1337212324142456,
      "learning_rate": 9.484913905814716e-05,
      "loss": 0.7779532670974731,
      "step": 315
    },
    {
      "epoch": 0.38049367850692356,
      "grad_norm": 0.14821310341358185,
      "learning_rate": 9.4805045463648e-05,
      "loss": 0.8427436947822571,
      "step": 316
    },
    {
      "epoch": 0.38169777242624925,
      "grad_norm": 0.13520045578479767,
      "learning_rate": 9.47607742815944e-05,
      "loss": 0.7337285876274109,
      "step": 317
    },
    {
      "epoch": 0.38290186634557494,
      "grad_norm": 0.15708714723587036,
      "learning_rate": 9.471632568745794e-05,
      "loss": 0.8182603716850281,
      "step": 318
    },
    {
      "epoch": 0.3841059602649007,
      "grad_norm": 0.14449602365493774,
      "learning_rate": 9.467169985741337e-05,
      "loss": 0.8103877902030945,
      "step": 319
    },
    {
      "epoch": 0.38531005418422637,
      "grad_norm": 0.1497582048177719,
      "learning_rate": 9.462689696833791e-05,
      "loss": 0.76885586977005,
      "step": 320
    },
    {
      "epoch": 0.38651414810355206,
      "grad_norm": 0.1406441330909729,
      "learning_rate": 9.458191719781056e-05,
      "loss": 0.7220385670661926,
      "step": 321
    },
    {
      "epoch": 0.3877182420228778,
      "grad_norm": 0.14791478216648102,
      "learning_rate": 9.453676072411142e-05,
      "loss": 0.7174328565597534,
      "step": 322
    },
    {
      "epoch": 0.3889223359422035,
      "grad_norm": 0.15194010734558105,
      "learning_rate": 9.449142772622092e-05,
      "loss": 0.8088712096214294,
      "step": 323
    },
    {
      "epoch": 0.3901264298615292,
      "grad_norm": 0.16500987112522125,
      "learning_rate": 9.444591838381922e-05,
      "loss": 0.7606682777404785,
      "step": 324
    },
    {
      "epoch": 0.3913305237808549,
      "grad_norm": 0.15634052455425262,
      "learning_rate": 9.440023287728537e-05,
      "loss": 0.8200355768203735,
      "step": 325
    },
    {
      "epoch": 0.3925346177001806,
      "grad_norm": 0.14405246078968048,
      "learning_rate": 9.435437138769672e-05,
      "loss": 0.7506157159805298,
      "step": 326
    },
    {
      "epoch": 0.3937387116195063,
      "grad_norm": 0.16040895879268646,
      "learning_rate": 9.430833409682806e-05,
      "loss": 0.728818416595459,
      "step": 327
    },
    {
      "epoch": 0.39494280553883204,
      "grad_norm": 0.1454458236694336,
      "learning_rate": 9.426212118715108e-05,
      "loss": 0.7636417150497437,
      "step": 328
    },
    {
      "epoch": 0.39614689945815773,
      "grad_norm": 0.16521288454532623,
      "learning_rate": 9.421573284183345e-05,
      "loss": 0.8812077045440674,
      "step": 329
    },
    {
      "epoch": 0.3973509933774834,
      "grad_norm": 0.15389931201934814,
      "learning_rate": 9.416916924473825e-05,
      "loss": 0.8367751836776733,
      "step": 330
    },
    {
      "epoch": 0.39855508729680916,
      "grad_norm": 0.1578996181488037,
      "learning_rate": 9.412243058042315e-05,
      "loss": 0.8190886974334717,
      "step": 331
    },
    {
      "epoch": 0.39975918121613485,
      "grad_norm": 0.14078624546527863,
      "learning_rate": 9.407551703413973e-05,
      "loss": 0.787665605545044,
      "step": 332
    },
    {
      "epoch": 0.40096327513546054,
      "grad_norm": 0.16844964027404785,
      "learning_rate": 9.402842879183272e-05,
      "loss": 0.8270130157470703,
      "step": 333
    },
    {
      "epoch": 0.4021673690547863,
      "grad_norm": 0.13521817326545715,
      "learning_rate": 9.398116604013925e-05,
      "loss": 0.7942067384719849,
      "step": 334
    },
    {
      "epoch": 0.40337146297411197,
      "grad_norm": 0.1386251449584961,
      "learning_rate": 9.393372896638816e-05,
      "loss": 0.7909543514251709,
      "step": 335
    },
    {
      "epoch": 0.4045755568934377,
      "grad_norm": 0.14085575938224792,
      "learning_rate": 9.38861177585992e-05,
      "loss": 0.7976891398429871,
      "step": 336
    },
    {
      "epoch": 0.4057796508127634,
      "grad_norm": 0.15985777974128723,
      "learning_rate": 9.383833260548233e-05,
      "loss": 0.9223812222480774,
      "step": 337
    },
    {
      "epoch": 0.4069837447320891,
      "grad_norm": 0.15927723050117493,
      "learning_rate": 9.379037369643694e-05,
      "loss": 0.7600573897361755,
      "step": 338
    },
    {
      "epoch": 0.40818783865141484,
      "grad_norm": 0.14136390388011932,
      "learning_rate": 9.374224122155111e-05,
      "loss": 0.6745304465293884,
      "step": 339
    },
    {
      "epoch": 0.4093919325707405,
      "grad_norm": 0.13713400065898895,
      "learning_rate": 9.369393537160089e-05,
      "loss": 0.7461698651313782,
      "step": 340
    },
    {
      "epoch": 0.4105960264900662,
      "grad_norm": 0.1508476287126541,
      "learning_rate": 9.364545633804946e-05,
      "loss": 0.7970109581947327,
      "step": 341
    },
    {
      "epoch": 0.41180012040939196,
      "grad_norm": 0.15379178524017334,
      "learning_rate": 9.359680431304647e-05,
      "loss": 0.8115739226341248,
      "step": 342
    },
    {
      "epoch": 0.41300421432871764,
      "grad_norm": 0.14935919642448425,
      "learning_rate": 9.354797948942719e-05,
      "loss": 0.8235081434249878,
      "step": 343
    },
    {
      "epoch": 0.41420830824804333,
      "grad_norm": 0.15869762003421783,
      "learning_rate": 9.349898206071186e-05,
      "loss": 0.7955063581466675,
      "step": 344
    },
    {
      "epoch": 0.4154124021673691,
      "grad_norm": 0.1530236154794693,
      "learning_rate": 9.344981222110476e-05,
      "loss": 0.8023476600646973,
      "step": 345
    },
    {
      "epoch": 0.41661649608669477,
      "grad_norm": 0.13725599646568298,
      "learning_rate": 9.340047016549358e-05,
      "loss": 0.8545880317687988,
      "step": 346
    },
    {
      "epoch": 0.41782059000602045,
      "grad_norm": 0.15515175461769104,
      "learning_rate": 9.335095608944861e-05,
      "loss": 0.9410490989685059,
      "step": 347
    },
    {
      "epoch": 0.4190246839253462,
      "grad_norm": 0.15557458996772766,
      "learning_rate": 9.330127018922194e-05,
      "loss": 0.7643197178840637,
      "step": 348
    },
    {
      "epoch": 0.4202287778446719,
      "grad_norm": 0.146572083234787,
      "learning_rate": 9.325141266174666e-05,
      "loss": 0.779821515083313,
      "step": 349
    },
    {
      "epoch": 0.4214328717639976,
      "grad_norm": 0.13859672844409943,
      "learning_rate": 9.320138370463618e-05,
      "loss": 0.8529562950134277,
      "step": 350
    },
    {
      "epoch": 0.4226369656833233,
      "grad_norm": 0.17474834620952606,
      "learning_rate": 9.315118351618335e-05,
      "loss": 0.8432419896125793,
      "step": 351
    },
    {
      "epoch": 0.423841059602649,
      "grad_norm": 0.15198786556720734,
      "learning_rate": 9.310081229535968e-05,
      "loss": 0.7468613386154175,
      "step": 352
    },
    {
      "epoch": 0.4250451535219747,
      "grad_norm": 0.14015290141105652,
      "learning_rate": 9.305027024181462e-05,
      "loss": 0.7275055646896362,
      "step": 353
    },
    {
      "epoch": 0.42624924744130044,
      "grad_norm": 0.14974236488342285,
      "learning_rate": 9.29995575558747e-05,
      "loss": 0.7844899296760559,
      "step": 354
    },
    {
      "epoch": 0.4274533413606261,
      "grad_norm": 0.14732293784618378,
      "learning_rate": 9.294867443854278e-05,
      "loss": 0.7087262272834778,
      "step": 355
    },
    {
      "epoch": 0.4286574352799518,
      "grad_norm": 0.1540423035621643,
      "learning_rate": 9.289762109149723e-05,
      "loss": 0.8284270763397217,
      "step": 356
    },
    {
      "epoch": 0.42986152919927756,
      "grad_norm": 0.15612474083900452,
      "learning_rate": 9.284639771709112e-05,
      "loss": 0.8184201717376709,
      "step": 357
    },
    {
      "epoch": 0.43106562311860325,
      "grad_norm": 0.13649208843708038,
      "learning_rate": 9.279500451835145e-05,
      "loss": 0.795914888381958,
      "step": 358
    },
    {
      "epoch": 0.43226971703792894,
      "grad_norm": 0.14697588980197906,
      "learning_rate": 9.274344169897834e-05,
      "loss": 0.7656745314598083,
      "step": 359
    },
    {
      "epoch": 0.4334738109572547,
      "grad_norm": 0.16912950575351715,
      "learning_rate": 9.269170946334418e-05,
      "loss": 0.8664093613624573,
      "step": 360
    },
    {
      "epoch": 0.43467790487658037,
      "grad_norm": 0.1677756905555725,
      "learning_rate": 9.263980801649286e-05,
      "loss": 0.8193706274032593,
      "step": 361
    },
    {
      "epoch": 0.43588199879590606,
      "grad_norm": 0.15780210494995117,
      "learning_rate": 9.258773756413898e-05,
      "loss": 0.715253472328186,
      "step": 362
    },
    {
      "epoch": 0.4370860927152318,
      "grad_norm": 0.1534217894077301,
      "learning_rate": 9.253549831266696e-05,
      "loss": 0.8745909333229065,
      "step": 363
    },
    {
      "epoch": 0.4382901866345575,
      "grad_norm": 0.13887542486190796,
      "learning_rate": 9.248309046913032e-05,
      "loss": 0.730850875377655,
      "step": 364
    },
    {
      "epoch": 0.4394942805538832,
      "grad_norm": 0.1386643797159195,
      "learning_rate": 9.243051424125075e-05,
      "loss": 0.8557745814323425,
      "step": 365
    },
    {
      "epoch": 0.4406983744732089,
      "grad_norm": 0.1456262469291687,
      "learning_rate": 9.237776983741735e-05,
      "loss": 0.7595902681350708,
      "step": 366
    },
    {
      "epoch": 0.4419024683925346,
      "grad_norm": 0.14748980104923248,
      "learning_rate": 9.232485746668584e-05,
      "loss": 0.7671077251434326,
      "step": 367
    },
    {
      "epoch": 0.44310656231186035,
      "grad_norm": 0.1645089089870453,
      "learning_rate": 9.227177733877763e-05,
      "loss": 0.7938269376754761,
      "step": 368
    },
    {
      "epoch": 0.44431065623118604,
      "grad_norm": 0.1592482030391693,
      "learning_rate": 9.221852966407909e-05,
      "loss": 0.812677264213562,
      "step": 369
    },
    {
      "epoch": 0.44551475015051173,
      "grad_norm": 0.1517871916294098,
      "learning_rate": 9.216511465364066e-05,
      "loss": 0.6703803539276123,
      "step": 370
    },
    {
      "epoch": 0.4467188440698375,
      "grad_norm": 0.14634014666080475,
      "learning_rate": 9.2111532519176e-05,
      "loss": 0.6993438005447388,
      "step": 371
    },
    {
      "epoch": 0.44792293798916316,
      "grad_norm": 0.16064521670341492,
      "learning_rate": 9.205778347306121e-05,
      "loss": 0.8705127239227295,
      "step": 372
    },
    {
      "epoch": 0.44912703190848885,
      "grad_norm": 0.15612183511257172,
      "learning_rate": 9.200386772833394e-05,
      "loss": 0.7602788209915161,
      "step": 373
    },
    {
      "epoch": 0.4503311258278146,
      "grad_norm": 0.14914312958717346,
      "learning_rate": 9.194978549869256e-05,
      "loss": 0.8395487070083618,
      "step": 374
    },
    {
      "epoch": 0.4515352197471403,
      "grad_norm": 0.13432413339614868,
      "learning_rate": 9.189553699849531e-05,
      "loss": 0.75290846824646,
      "step": 375
    },
    {
      "epoch": 0.45273931366646597,
      "grad_norm": 0.17705515027046204,
      "learning_rate": 9.184112244275948e-05,
      "loss": 0.9329785704612732,
      "step": 376
    },
    {
      "epoch": 0.4539434075857917,
      "grad_norm": 0.13686609268188477,
      "learning_rate": 9.178654204716048e-05,
      "loss": 0.7564486265182495,
      "step": 377
    },
    {
      "epoch": 0.4551475015051174,
      "grad_norm": 0.12022358179092407,
      "learning_rate": 9.173179602803108e-05,
      "loss": 0.6906337738037109,
      "step": 378
    },
    {
      "epoch": 0.4563515954244431,
      "grad_norm": 0.13804689049720764,
      "learning_rate": 9.167688460236049e-05,
      "loss": 0.851130485534668,
      "step": 379
    },
    {
      "epoch": 0.45755568934376883,
      "grad_norm": 0.17003387212753296,
      "learning_rate": 9.162180798779354e-05,
      "loss": 0.8722536563873291,
      "step": 380
    },
    {
      "epoch": 0.4587597832630945,
      "grad_norm": 0.15241733193397522,
      "learning_rate": 9.156656640262975e-05,
      "loss": 0.8923465013504028,
      "step": 381
    },
    {
      "epoch": 0.4599638771824202,
      "grad_norm": 0.1540643274784088,
      "learning_rate": 9.151116006582259e-05,
      "loss": 0.7983347177505493,
      "step": 382
    },
    {
      "epoch": 0.46116797110174595,
      "grad_norm": 0.14970754086971283,
      "learning_rate": 9.145558919697844e-05,
      "loss": 0.8543246388435364,
      "step": 383
    },
    {
      "epoch": 0.46237206502107164,
      "grad_norm": 0.15691834688186646,
      "learning_rate": 9.139985401635587e-05,
      "loss": 0.7405463457107544,
      "step": 384
    },
    {
      "epoch": 0.46357615894039733,
      "grad_norm": 0.14189770817756653,
      "learning_rate": 9.13439547448647e-05,
      "loss": 0.8395057916641235,
      "step": 385
    },
    {
      "epoch": 0.4647802528597231,
      "grad_norm": 0.14295047521591187,
      "learning_rate": 9.128789160406512e-05,
      "loss": 0.7399108409881592,
      "step": 386
    },
    {
      "epoch": 0.46598434677904876,
      "grad_norm": 0.1451527625322342,
      "learning_rate": 9.123166481616687e-05,
      "loss": 0.7674145698547363,
      "step": 387
    },
    {
      "epoch": 0.46718844069837445,
      "grad_norm": 0.1433471143245697,
      "learning_rate": 9.117527460402826e-05,
      "loss": 0.7844995856285095,
      "step": 388
    },
    {
      "epoch": 0.4683925346177002,
      "grad_norm": 0.1410779505968094,
      "learning_rate": 9.111872119115537e-05,
      "loss": 0.7607631683349609,
      "step": 389
    },
    {
      "epoch": 0.4695966285370259,
      "grad_norm": 0.14412206411361694,
      "learning_rate": 9.106200480170113e-05,
      "loss": 0.8115881681442261,
      "step": 390
    },
    {
      "epoch": 0.4708007224563516,
      "grad_norm": 0.15132735669612885,
      "learning_rate": 9.100512566046444e-05,
      "loss": 0.6876002550125122,
      "step": 391
    },
    {
      "epoch": 0.4720048163756773,
      "grad_norm": 0.13577982783317566,
      "learning_rate": 9.094808399288927e-05,
      "loss": 0.7821998596191406,
      "step": 392
    },
    {
      "epoch": 0.473208910295003,
      "grad_norm": 0.14433644711971283,
      "learning_rate": 9.089088002506379e-05,
      "loss": 0.7749229669570923,
      "step": 393
    },
    {
      "epoch": 0.4744130042143287,
      "grad_norm": 0.1616220474243164,
      "learning_rate": 9.083351398371944e-05,
      "loss": 0.8197864294052124,
      "step": 394
    },
    {
      "epoch": 0.47561709813365444,
      "grad_norm": 0.13242773711681366,
      "learning_rate": 9.077598609623006e-05,
      "loss": 0.7858530282974243,
      "step": 395
    },
    {
      "epoch": 0.4768211920529801,
      "grad_norm": 0.16815578937530518,
      "learning_rate": 9.071829659061097e-05,
      "loss": 0.8673710227012634,
      "step": 396
    },
    {
      "epoch": 0.4780252859723058,
      "grad_norm": 0.1420876383781433,
      "learning_rate": 9.066044569551807e-05,
      "loss": 0.7124256491661072,
      "step": 397
    },
    {
      "epoch": 0.47922937989163156,
      "grad_norm": 0.16514599323272705,
      "learning_rate": 9.060243364024692e-05,
      "loss": 0.7791928052902222,
      "step": 398
    },
    {
      "epoch": 0.48043347381095725,
      "grad_norm": 0.1764056235551834,
      "learning_rate": 9.054426065473192e-05,
      "loss": 0.9173452258110046,
      "step": 399
    },
    {
      "epoch": 0.481637567730283,
      "grad_norm": 0.1389186829328537,
      "learning_rate": 9.048592696954524e-05,
      "loss": 0.8092312216758728,
      "step": 400
    },
    {
      "epoch": 0.4828416616496087,
      "grad_norm": 0.1505022943019867,
      "learning_rate": 9.042743281589605e-05,
      "loss": 0.797105073928833,
      "step": 401
    },
    {
      "epoch": 0.48404575556893437,
      "grad_norm": 0.1351250410079956,
      "learning_rate": 9.036877842562952e-05,
      "loss": 0.6983296871185303,
      "step": 402
    },
    {
      "epoch": 0.4852498494882601,
      "grad_norm": 0.1606484204530716,
      "learning_rate": 9.030996403122592e-05,
      "loss": 0.8399434685707092,
      "step": 403
    },
    {
      "epoch": 0.4864539434075858,
      "grad_norm": 0.1389196366071701,
      "learning_rate": 9.025098986579975e-05,
      "loss": 0.8033434152603149,
      "step": 404
    },
    {
      "epoch": 0.4876580373269115,
      "grad_norm": 0.13865266740322113,
      "learning_rate": 9.019185616309871e-05,
      "loss": 0.8269985318183899,
      "step": 405
    },
    {
      "epoch": 0.48886213124623723,
      "grad_norm": 0.15112046897411346,
      "learning_rate": 9.013256315750291e-05,
      "loss": 0.7290336489677429,
      "step": 406
    },
    {
      "epoch": 0.4900662251655629,
      "grad_norm": 0.15697422623634338,
      "learning_rate": 9.007311108402379e-05,
      "loss": 0.8027770519256592,
      "step": 407
    },
    {
      "epoch": 0.4912703190848886,
      "grad_norm": 0.15798939764499664,
      "learning_rate": 9.001350017830331e-05,
      "loss": 0.7963067889213562,
      "step": 408
    },
    {
      "epoch": 0.49247441300421435,
      "grad_norm": 0.1397654265165329,
      "learning_rate": 8.995373067661296e-05,
      "loss": 0.8707777261734009,
      "step": 409
    },
    {
      "epoch": 0.49367850692354004,
      "grad_norm": 0.1369328796863556,
      "learning_rate": 8.989380281585287e-05,
      "loss": 0.7579842805862427,
      "step": 410
    },
    {
      "epoch": 0.4948826008428657,
      "grad_norm": 0.16840219497680664,
      "learning_rate": 8.983371683355075e-05,
      "loss": 0.7257000207901001,
      "step": 411
    },
    {
      "epoch": 0.49608669476219147,
      "grad_norm": 0.14521774649620056,
      "learning_rate": 8.977347296786113e-05,
      "loss": 0.8197977542877197,
      "step": 412
    },
    {
      "epoch": 0.49729078868151716,
      "grad_norm": 0.14849089086055756,
      "learning_rate": 8.971307145756426e-05,
      "loss": 0.7103527188301086,
      "step": 413
    },
    {
      "epoch": 0.49849488260084285,
      "grad_norm": 0.14931897819042206,
      "learning_rate": 8.965251254206524e-05,
      "loss": 0.7112723588943481,
      "step": 414
    },
    {
      "epoch": 0.4996989765201686,
      "grad_norm": 0.12863224744796753,
      "learning_rate": 8.959179646139306e-05,
      "loss": 0.8152425289154053,
      "step": 415
    },
    {
      "epoch": 0.5009030704394943,
      "grad_norm": 0.13744914531707764,
      "learning_rate": 8.953092345619964e-05,
      "loss": 0.7117719650268555,
      "step": 416
    },
    {
      "epoch": 0.50210716435882,
      "grad_norm": 0.15159831941127777,
      "learning_rate": 8.94698937677589e-05,
      "loss": 0.8456029891967773,
      "step": 417
    },
    {
      "epoch": 0.5033112582781457,
      "grad_norm": 0.129548579454422,
      "learning_rate": 8.940870763796574e-05,
      "loss": 0.7555068731307983,
      "step": 418
    },
    {
      "epoch": 0.5045153521974715,
      "grad_norm": 0.14241458475589752,
      "learning_rate": 8.934736530933519e-05,
      "loss": 0.7059701085090637,
      "step": 419
    },
    {
      "epoch": 0.5057194461167971,
      "grad_norm": 0.1404544860124588,
      "learning_rate": 8.928586702500128e-05,
      "loss": 0.8070000410079956,
      "step": 420
    },
    {
      "epoch": 0.5069235400361228,
      "grad_norm": 0.13401378691196442,
      "learning_rate": 8.922421302871629e-05,
      "loss": 0.7877148389816284,
      "step": 421
    },
    {
      "epoch": 0.5081276339554486,
      "grad_norm": 0.13424450159072876,
      "learning_rate": 8.916240356484961e-05,
      "loss": 0.8055455088615417,
      "step": 422
    },
    {
      "epoch": 0.5093317278747742,
      "grad_norm": 0.13702525198459625,
      "learning_rate": 8.910043887838686e-05,
      "loss": 0.7678868174552917,
      "step": 423
    },
    {
      "epoch": 0.5105358217941,
      "grad_norm": 0.1551496833562851,
      "learning_rate": 8.903831921492889e-05,
      "loss": 0.7496639490127563,
      "step": 424
    },
    {
      "epoch": 0.5117399157134257,
      "grad_norm": 0.13463352620601654,
      "learning_rate": 8.89760448206908e-05,
      "loss": 0.8483864068984985,
      "step": 425
    },
    {
      "epoch": 0.5129440096327513,
      "grad_norm": 0.13580475747585297,
      "learning_rate": 8.8913615942501e-05,
      "loss": 0.7934477925300598,
      "step": 426
    },
    {
      "epoch": 0.5141481035520771,
      "grad_norm": 0.14507661759853363,
      "learning_rate": 8.885103282780016e-05,
      "loss": 0.7932281494140625,
      "step": 427
    },
    {
      "epoch": 0.5153521974714028,
      "grad_norm": 0.11664706468582153,
      "learning_rate": 8.878829572464034e-05,
      "loss": 0.7825329303741455,
      "step": 428
    },
    {
      "epoch": 0.5165562913907285,
      "grad_norm": 0.12552881240844727,
      "learning_rate": 8.872540488168389e-05,
      "loss": 0.6748529672622681,
      "step": 429
    },
    {
      "epoch": 0.5177603853100542,
      "grad_norm": 0.15718883275985718,
      "learning_rate": 8.866236054820251e-05,
      "loss": 0.8103749752044678,
      "step": 430
    },
    {
      "epoch": 0.5189644792293799,
      "grad_norm": 0.14597220718860626,
      "learning_rate": 8.859916297407636e-05,
      "loss": 0.7614364624023438,
      "step": 431
    },
    {
      "epoch": 0.5201685731487056,
      "grad_norm": 0.15009157359600067,
      "learning_rate": 8.853581240979285e-05,
      "loss": 0.7507392168045044,
      "step": 432
    },
    {
      "epoch": 0.5213726670680313,
      "grad_norm": 0.15521551668643951,
      "learning_rate": 8.847230910644586e-05,
      "loss": 0.7126198410987854,
      "step": 433
    },
    {
      "epoch": 0.5225767609873571,
      "grad_norm": 0.13860978186130524,
      "learning_rate": 8.840865331573465e-05,
      "loss": 0.8197992444038391,
      "step": 434
    },
    {
      "epoch": 0.5237808549066827,
      "grad_norm": 0.14406178891658783,
      "learning_rate": 8.834484528996287e-05,
      "loss": 0.7747809290885925,
      "step": 435
    },
    {
      "epoch": 0.5249849488260084,
      "grad_norm": 0.1502278745174408,
      "learning_rate": 8.828088528203753e-05,
      "loss": 0.8083539605140686,
      "step": 436
    },
    {
      "epoch": 0.5261890427453342,
      "grad_norm": 0.1459406465291977,
      "learning_rate": 8.821677354546807e-05,
      "loss": 0.8128182291984558,
      "step": 437
    },
    {
      "epoch": 0.5273931366646598,
      "grad_norm": 0.1293898969888687,
      "learning_rate": 8.815251033436531e-05,
      "loss": 0.7243235111236572,
      "step": 438
    },
    {
      "epoch": 0.5285972305839856,
      "grad_norm": 0.1493912786245346,
      "learning_rate": 8.808809590344042e-05,
      "loss": 0.8678902983665466,
      "step": 439
    },
    {
      "epoch": 0.5298013245033113,
      "grad_norm": 0.1284581869840622,
      "learning_rate": 8.802353050800398e-05,
      "loss": 0.8014649748802185,
      "step": 440
    },
    {
      "epoch": 0.5310054184226369,
      "grad_norm": 0.14924846589565277,
      "learning_rate": 8.795881440396491e-05,
      "loss": 0.7606640458106995,
      "step": 441
    },
    {
      "epoch": 0.5322095123419627,
      "grad_norm": 0.13209116458892822,
      "learning_rate": 8.789394784782945e-05,
      "loss": 0.7208225727081299,
      "step": 442
    },
    {
      "epoch": 0.5334136062612884,
      "grad_norm": 0.13858436048030853,
      "learning_rate": 8.78289310967002e-05,
      "loss": 0.7206466794013977,
      "step": 443
    },
    {
      "epoch": 0.534617700180614,
      "grad_norm": 0.13447819650173187,
      "learning_rate": 8.776376440827505e-05,
      "loss": 0.7764282822608948,
      "step": 444
    },
    {
      "epoch": 0.5358217940999398,
      "grad_norm": 0.15937335789203644,
      "learning_rate": 8.769844804084619e-05,
      "loss": 0.8614116311073303,
      "step": 445
    },
    {
      "epoch": 0.5370258880192655,
      "grad_norm": 0.17174158990383148,
      "learning_rate": 8.763298225329903e-05,
      "loss": 0.80394446849823,
      "step": 446
    },
    {
      "epoch": 0.5382299819385912,
      "grad_norm": 0.13603968918323517,
      "learning_rate": 8.756736730511128e-05,
      "loss": 0.7360084056854248,
      "step": 447
    },
    {
      "epoch": 0.5394340758579169,
      "grad_norm": 0.14448818564414978,
      "learning_rate": 8.750160345635183e-05,
      "loss": 0.7574860453605652,
      "step": 448
    },
    {
      "epoch": 0.5406381697772427,
      "grad_norm": 0.13352635502815247,
      "learning_rate": 8.74356909676797e-05,
      "loss": 0.6965577006340027,
      "step": 449
    },
    {
      "epoch": 0.5418422636965683,
      "grad_norm": 0.13442601263523102,
      "learning_rate": 8.736963010034311e-05,
      "loss": 0.7972019910812378,
      "step": 450
    },
    {
      "epoch": 0.543046357615894,
      "grad_norm": 0.1476263701915741,
      "learning_rate": 8.73034211161784e-05,
      "loss": 0.8003101348876953,
      "step": 451
    },
    {
      "epoch": 0.5442504515352198,
      "grad_norm": 0.14945708215236664,
      "learning_rate": 8.723706427760892e-05,
      "loss": 0.7748737931251526,
      "step": 452
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.13502340018749237,
      "learning_rate": 8.717055984764411e-05,
      "loss": 0.8290547132492065,
      "step": 453
    },
    {
      "epoch": 0.5466586393738712,
      "grad_norm": 0.12881049513816833,
      "learning_rate": 8.710390808987833e-05,
      "loss": 0.7508219480514526,
      "step": 454
    },
    {
      "epoch": 0.5478627332931969,
      "grad_norm": 0.13485009968280792,
      "learning_rate": 8.703710926848995e-05,
      "loss": 0.8000583648681641,
      "step": 455
    },
    {
      "epoch": 0.5490668272125225,
      "grad_norm": 0.14547601342201233,
      "learning_rate": 8.697016364824023e-05,
      "loss": 0.8134283423423767,
      "step": 456
    },
    {
      "epoch": 0.5502709211318483,
      "grad_norm": 0.1349562108516693,
      "learning_rate": 8.690307149447221e-05,
      "loss": 0.7571486830711365,
      "step": 457
    },
    {
      "epoch": 0.551475015051174,
      "grad_norm": 0.14580993354320526,
      "learning_rate": 8.683583307310978e-05,
      "loss": 0.8517851829528809,
      "step": 458
    },
    {
      "epoch": 0.5526791089704997,
      "grad_norm": 0.15511661767959595,
      "learning_rate": 8.676844865065659e-05,
      "loss": 0.8604599237442017,
      "step": 459
    },
    {
      "epoch": 0.5538832028898254,
      "grad_norm": 0.15549787878990173,
      "learning_rate": 8.67009184941949e-05,
      "loss": 0.9273145198822021,
      "step": 460
    },
    {
      "epoch": 0.5550872968091511,
      "grad_norm": 0.14486557245254517,
      "learning_rate": 8.663324287138469e-05,
      "loss": 0.8311535120010376,
      "step": 461
    },
    {
      "epoch": 0.5562913907284768,
      "grad_norm": 0.13686466217041016,
      "learning_rate": 8.656542205046243e-05,
      "loss": 0.783361554145813,
      "step": 462
    },
    {
      "epoch": 0.5574954846478025,
      "grad_norm": 0.13392846286296844,
      "learning_rate": 8.64974563002401e-05,
      "loss": 0.7674474716186523,
      "step": 463
    },
    {
      "epoch": 0.5586995785671283,
      "grad_norm": 0.14726831018924713,
      "learning_rate": 8.642934589010414e-05,
      "loss": 0.8609716296195984,
      "step": 464
    },
    {
      "epoch": 0.5599036724864539,
      "grad_norm": 0.14644522964954376,
      "learning_rate": 8.636109109001438e-05,
      "loss": 0.8058178424835205,
      "step": 465
    },
    {
      "epoch": 0.5611077664057796,
      "grad_norm": 0.132773756980896,
      "learning_rate": 8.629269217050289e-05,
      "loss": 0.7636709809303284,
      "step": 466
    },
    {
      "epoch": 0.5623118603251054,
      "grad_norm": 0.15270403027534485,
      "learning_rate": 8.6224149402673e-05,
      "loss": 0.7716991901397705,
      "step": 467
    },
    {
      "epoch": 0.563515954244431,
      "grad_norm": 0.13487881422042847,
      "learning_rate": 8.61554630581982e-05,
      "loss": 0.7246546745300293,
      "step": 468
    },
    {
      "epoch": 0.5647200481637568,
      "grad_norm": 0.14720039069652557,
      "learning_rate": 8.608663340932104e-05,
      "loss": 0.7644535899162292,
      "step": 469
    },
    {
      "epoch": 0.5659241420830825,
      "grad_norm": 0.12816502153873444,
      "learning_rate": 8.601766072885204e-05,
      "loss": 0.6405297517776489,
      "step": 470
    },
    {
      "epoch": 0.5671282360024081,
      "grad_norm": 0.12482790648937225,
      "learning_rate": 8.594854529016872e-05,
      "loss": 0.6063294410705566,
      "step": 471
    },
    {
      "epoch": 0.5683323299217339,
      "grad_norm": 0.13587307929992676,
      "learning_rate": 8.587928736721432e-05,
      "loss": 0.7394362688064575,
      "step": 472
    },
    {
      "epoch": 0.5695364238410596,
      "grad_norm": 0.15476983785629272,
      "learning_rate": 8.580988723449688e-05,
      "loss": 0.879747211933136,
      "step": 473
    },
    {
      "epoch": 0.5707405177603853,
      "grad_norm": 0.1372615545988083,
      "learning_rate": 8.574034516708814e-05,
      "loss": 0.7705991864204407,
      "step": 474
    },
    {
      "epoch": 0.571944611679711,
      "grad_norm": 0.1265069842338562,
      "learning_rate": 8.567066144062232e-05,
      "loss": 0.7046324014663696,
      "step": 475
    },
    {
      "epoch": 0.5731487055990367,
      "grad_norm": 0.13162341713905334,
      "learning_rate": 8.560083633129519e-05,
      "loss": 0.7208070158958435,
      "step": 476
    },
    {
      "epoch": 0.5743527995183624,
      "grad_norm": 0.12545783817768097,
      "learning_rate": 8.553087011586284e-05,
      "loss": 0.6994925737380981,
      "step": 477
    },
    {
      "epoch": 0.5755568934376881,
      "grad_norm": 0.13866853713989258,
      "learning_rate": 8.546076307164068e-05,
      "loss": 0.7362884879112244,
      "step": 478
    },
    {
      "epoch": 0.5767609873570139,
      "grad_norm": 0.1489093005657196,
      "learning_rate": 8.53905154765023e-05,
      "loss": 0.6578172445297241,
      "step": 479
    },
    {
      "epoch": 0.5779650812763396,
      "grad_norm": 0.1568404883146286,
      "learning_rate": 8.532012760887837e-05,
      "loss": 0.8208804130554199,
      "step": 480
    },
    {
      "epoch": 0.5791691751956652,
      "grad_norm": 0.15127775073051453,
      "learning_rate": 8.524959974775551e-05,
      "loss": 0.7533735036849976,
      "step": 481
    },
    {
      "epoch": 0.580373269114991,
      "grad_norm": 0.14071667194366455,
      "learning_rate": 8.517893217267525e-05,
      "loss": 0.6773391962051392,
      "step": 482
    },
    {
      "epoch": 0.5815773630343167,
      "grad_norm": 0.14453256130218506,
      "learning_rate": 8.510812516373288e-05,
      "loss": 0.8153808116912842,
      "step": 483
    },
    {
      "epoch": 0.5827814569536424,
      "grad_norm": 0.1420130878686905,
      "learning_rate": 8.503717900157632e-05,
      "loss": 0.7941230535507202,
      "step": 484
    },
    {
      "epoch": 0.5839855508729681,
      "grad_norm": 0.14127445220947266,
      "learning_rate": 8.496609396740506e-05,
      "loss": 0.7883418202400208,
      "step": 485
    },
    {
      "epoch": 0.5851896447922939,
      "grad_norm": 0.14173413813114166,
      "learning_rate": 8.489487034296902e-05,
      "loss": 0.7391270995140076,
      "step": 486
    },
    {
      "epoch": 0.5863937387116195,
      "grad_norm": 0.1412261575460434,
      "learning_rate": 8.482350841056737e-05,
      "loss": 0.6304709315299988,
      "step": 487
    },
    {
      "epoch": 0.5875978326309452,
      "grad_norm": 0.14295165240764618,
      "learning_rate": 8.475200845304758e-05,
      "loss": 0.8438637256622314,
      "step": 488
    },
    {
      "epoch": 0.588801926550271,
      "grad_norm": 0.1608818620443344,
      "learning_rate": 8.468037075380408e-05,
      "loss": 0.7637495994567871,
      "step": 489
    },
    {
      "epoch": 0.5900060204695966,
      "grad_norm": 0.1327633112668991,
      "learning_rate": 8.460859559677734e-05,
      "loss": 0.647986888885498,
      "step": 490
    },
    {
      "epoch": 0.5912101143889223,
      "grad_norm": 0.13329662382602692,
      "learning_rate": 8.453668326645259e-05,
      "loss": 0.6748461723327637,
      "step": 491
    },
    {
      "epoch": 0.5924142083082481,
      "grad_norm": 0.1708861142396927,
      "learning_rate": 8.446463404785875e-05,
      "loss": 0.8186511993408203,
      "step": 492
    },
    {
      "epoch": 0.5936183022275737,
      "grad_norm": 0.1555318981409073,
      "learning_rate": 8.43924482265674e-05,
      "loss": 0.7217735052108765,
      "step": 493
    },
    {
      "epoch": 0.5948223961468995,
      "grad_norm": 0.12814928591251373,
      "learning_rate": 8.432012608869141e-05,
      "loss": 0.7544658780097961,
      "step": 494
    },
    {
      "epoch": 0.5960264900662252,
      "grad_norm": 0.12696655094623566,
      "learning_rate": 8.424766792088408e-05,
      "loss": 0.6882718205451965,
      "step": 495
    },
    {
      "epoch": 0.5972305839855508,
      "grad_norm": 0.13328474760055542,
      "learning_rate": 8.417507401033779e-05,
      "loss": 0.7377568483352661,
      "step": 496
    },
    {
      "epoch": 0.5984346779048766,
      "grad_norm": 0.13095304369926453,
      "learning_rate": 8.410234464478297e-05,
      "loss": 0.6852051615715027,
      "step": 497
    },
    {
      "epoch": 0.5996387718242023,
      "grad_norm": 0.1368059366941452,
      "learning_rate": 8.402948011248692e-05,
      "loss": 0.7555135488510132,
      "step": 498
    },
    {
      "epoch": 0.600842865743528,
      "grad_norm": 0.1365233212709427,
      "learning_rate": 8.395648070225272e-05,
      "loss": 0.7157353758811951,
      "step": 499
    },
    {
      "epoch": 0.6020469596628537,
      "grad_norm": 0.13838793337345123,
      "learning_rate": 8.388334670341805e-05,
      "loss": 0.7702412605285645,
      "step": 500
    },
    {
      "epoch": 0.6032510535821795,
      "grad_norm": 0.14599739015102386,
      "learning_rate": 8.381007840585395e-05,
      "loss": 0.7846642732620239,
      "step": 501
    },
    {
      "epoch": 0.6044551475015051,
      "grad_norm": 0.17089492082595825,
      "learning_rate": 8.373667609996387e-05,
      "loss": 0.8437250852584839,
      "step": 502
    },
    {
      "epoch": 0.6056592414208308,
      "grad_norm": 0.15564851462841034,
      "learning_rate": 8.366314007668235e-05,
      "loss": 0.7621622085571289,
      "step": 503
    },
    {
      "epoch": 0.6068633353401566,
      "grad_norm": 0.12179785966873169,
      "learning_rate": 8.358947062747397e-05,
      "loss": 0.6772765517234802,
      "step": 504
    },
    {
      "epoch": 0.6080674292594822,
      "grad_norm": 0.14322376251220703,
      "learning_rate": 8.351566804433207e-05,
      "loss": 0.7505850791931152,
      "step": 505
    },
    {
      "epoch": 0.609271523178808,
      "grad_norm": 0.129380002617836,
      "learning_rate": 8.344173261977777e-05,
      "loss": 0.7710934281349182,
      "step": 506
    },
    {
      "epoch": 0.6104756170981337,
      "grad_norm": 0.15720881521701813,
      "learning_rate": 8.336766464685869e-05,
      "loss": 0.7477296590805054,
      "step": 507
    },
    {
      "epoch": 0.6116797110174593,
      "grad_norm": 0.14132221043109894,
      "learning_rate": 8.329346441914774e-05,
      "loss": 0.7790762186050415,
      "step": 508
    },
    {
      "epoch": 0.6128838049367851,
      "grad_norm": 0.18391427397727966,
      "learning_rate": 8.321913223074212e-05,
      "loss": 0.8408628702163696,
      "step": 509
    },
    {
      "epoch": 0.6140878988561108,
      "grad_norm": 0.13065968453884125,
      "learning_rate": 8.314466837626205e-05,
      "loss": 0.7032425403594971,
      "step": 510
    },
    {
      "epoch": 0.6152919927754364,
      "grad_norm": 0.1441313624382019,
      "learning_rate": 8.307007315084958e-05,
      "loss": 0.7481244802474976,
      "step": 511
    },
    {
      "epoch": 0.6164960866947622,
      "grad_norm": 0.13757000863552094,
      "learning_rate": 8.299534685016747e-05,
      "loss": 0.7431373000144958,
      "step": 512
    },
    {
      "epoch": 0.6177001806140879,
      "grad_norm": 0.15404686331748962,
      "learning_rate": 8.292048977039801e-05,
      "loss": 0.7886311411857605,
      "step": 513
    },
    {
      "epoch": 0.6189042745334136,
      "grad_norm": 0.14478828012943268,
      "learning_rate": 8.284550220824187e-05,
      "loss": 0.6625759601593018,
      "step": 514
    },
    {
      "epoch": 0.6201083684527393,
      "grad_norm": 0.13734887540340424,
      "learning_rate": 8.277038446091683e-05,
      "loss": 0.7376775741577148,
      "step": 515
    },
    {
      "epoch": 0.621312462372065,
      "grad_norm": 0.1305343210697174,
      "learning_rate": 8.269513682615672e-05,
      "loss": 0.7062985897064209,
      "step": 516
    },
    {
      "epoch": 0.6225165562913907,
      "grad_norm": 0.14777213335037231,
      "learning_rate": 8.261975960221017e-05,
      "loss": 0.7180648446083069,
      "step": 517
    },
    {
      "epoch": 0.6237206502107164,
      "grad_norm": 0.1526557207107544,
      "learning_rate": 8.254425308783944e-05,
      "loss": 0.805204451084137,
      "step": 518
    },
    {
      "epoch": 0.6249247441300422,
      "grad_norm": 0.13665392994880676,
      "learning_rate": 8.246861758231925e-05,
      "loss": 0.7313157916069031,
      "step": 519
    },
    {
      "epoch": 0.6261288380493678,
      "grad_norm": 0.14072732627391815,
      "learning_rate": 8.239285338543558e-05,
      "loss": 0.8550055027008057,
      "step": 520
    },
    {
      "epoch": 0.6273329319686936,
      "grad_norm": 0.14398051798343658,
      "learning_rate": 8.23169607974845e-05,
      "loss": 0.7382502555847168,
      "step": 521
    },
    {
      "epoch": 0.6285370258880193,
      "grad_norm": 0.14732129871845245,
      "learning_rate": 8.224094011927091e-05,
      "loss": 0.7357527613639832,
      "step": 522
    },
    {
      "epoch": 0.6297411198073449,
      "grad_norm": 0.134531170129776,
      "learning_rate": 8.216479165210748e-05,
      "loss": 0.669085681438446,
      "step": 523
    },
    {
      "epoch": 0.6309452137266707,
      "grad_norm": 0.13648062944412231,
      "learning_rate": 8.208851569781335e-05,
      "loss": 0.8549784421920776,
      "step": 524
    },
    {
      "epoch": 0.6321493076459964,
      "grad_norm": 0.14021635055541992,
      "learning_rate": 8.201211255871293e-05,
      "loss": 0.7222084999084473,
      "step": 525
    },
    {
      "epoch": 0.633353401565322,
      "grad_norm": 0.13175466656684875,
      "learning_rate": 8.193558253763478e-05,
      "loss": 0.727872371673584,
      "step": 526
    },
    {
      "epoch": 0.6345574954846478,
      "grad_norm": 0.1462666392326355,
      "learning_rate": 8.185892593791034e-05,
      "loss": 0.8319973945617676,
      "step": 527
    },
    {
      "epoch": 0.6357615894039735,
      "grad_norm": 0.14199678599834442,
      "learning_rate": 8.178214306337278e-05,
      "loss": 0.7619800567626953,
      "step": 528
    },
    {
      "epoch": 0.6369656833232992,
      "grad_norm": 0.17002788186073303,
      "learning_rate": 8.170523421835572e-05,
      "loss": 0.8425567746162415,
      "step": 529
    },
    {
      "epoch": 0.6381697772426249,
      "grad_norm": 0.1549702137708664,
      "learning_rate": 8.162819970769211e-05,
      "loss": 0.8674654960632324,
      "step": 530
    },
    {
      "epoch": 0.6393738711619507,
      "grad_norm": 0.13054436445236206,
      "learning_rate": 8.155103983671297e-05,
      "loss": 0.7390493154525757,
      "step": 531
    },
    {
      "epoch": 0.6405779650812763,
      "grad_norm": 0.12799249589443207,
      "learning_rate": 8.14737549112462e-05,
      "loss": 0.6426804661750793,
      "step": 532
    },
    {
      "epoch": 0.641782059000602,
      "grad_norm": 0.12660059332847595,
      "learning_rate": 8.139634523761537e-05,
      "loss": 0.682824969291687,
      "step": 533
    },
    {
      "epoch": 0.6429861529199278,
      "grad_norm": 0.14024198055267334,
      "learning_rate": 8.131881112263845e-05,
      "loss": 0.7971470952033997,
      "step": 534
    },
    {
      "epoch": 0.6441902468392534,
      "grad_norm": 0.14972057938575745,
      "learning_rate": 8.12411528736267e-05,
      "loss": 0.7102527022361755,
      "step": 535
    },
    {
      "epoch": 0.6453943407585792,
      "grad_norm": 0.13412761688232422,
      "learning_rate": 8.116337079838337e-05,
      "loss": 0.7489075064659119,
      "step": 536
    },
    {
      "epoch": 0.6465984346779049,
      "grad_norm": 0.14248909056186676,
      "learning_rate": 8.10854652052025e-05,
      "loss": 0.7161577343940735,
      "step": 537
    },
    {
      "epoch": 0.6478025285972305,
      "grad_norm": 0.13242509961128235,
      "learning_rate": 8.100743640286768e-05,
      "loss": 0.762588381767273,
      "step": 538
    },
    {
      "epoch": 0.6490066225165563,
      "grad_norm": 0.13860951364040375,
      "learning_rate": 8.092928470065091e-05,
      "loss": 0.7201557159423828,
      "step": 539
    },
    {
      "epoch": 0.650210716435882,
      "grad_norm": 0.13951687514781952,
      "learning_rate": 8.085101040831122e-05,
      "loss": 0.6891500949859619,
      "step": 540
    },
    {
      "epoch": 0.6514148103552077,
      "grad_norm": 0.14506259560585022,
      "learning_rate": 8.077261383609363e-05,
      "loss": 0.7638989686965942,
      "step": 541
    },
    {
      "epoch": 0.6526189042745334,
      "grad_norm": 0.13475656509399414,
      "learning_rate": 8.069409529472774e-05,
      "loss": 0.7189249992370605,
      "step": 542
    },
    {
      "epoch": 0.6538229981938591,
      "grad_norm": 0.1373392790555954,
      "learning_rate": 8.061545509542663e-05,
      "loss": 0.7654851675033569,
      "step": 543
    },
    {
      "epoch": 0.6550270921131849,
      "grad_norm": 0.14753283560276031,
      "learning_rate": 8.05366935498856e-05,
      "loss": 0.8151734471321106,
      "step": 544
    },
    {
      "epoch": 0.6562311860325105,
      "grad_norm": 0.1436505764722824,
      "learning_rate": 8.045781097028083e-05,
      "loss": 0.7153245210647583,
      "step": 545
    },
    {
      "epoch": 0.6574352799518363,
      "grad_norm": 0.1477065086364746,
      "learning_rate": 8.037880766926833e-05,
      "loss": 0.7415351867675781,
      "step": 546
    },
    {
      "epoch": 0.658639373871162,
      "grad_norm": 0.14270779490470886,
      "learning_rate": 8.02996839599825e-05,
      "loss": 0.8138120770454407,
      "step": 547
    },
    {
      "epoch": 0.6598434677904876,
      "grad_norm": 0.13764041662216187,
      "learning_rate": 8.022044015603505e-05,
      "loss": 0.673293948173523,
      "step": 548
    },
    {
      "epoch": 0.6610475617098134,
      "grad_norm": 0.135722815990448,
      "learning_rate": 8.014107657151369e-05,
      "loss": 0.7463434934616089,
      "step": 549
    },
    {
      "epoch": 0.6622516556291391,
      "grad_norm": 0.13560885190963745,
      "learning_rate": 8.006159352098082e-05,
      "loss": 0.7464152574539185,
      "step": 550
    },
    {
      "epoch": 0.6634557495484648,
      "grad_norm": 0.14212897419929504,
      "learning_rate": 7.998199131947247e-05,
      "loss": 0.7562774419784546,
      "step": 551
    },
    {
      "epoch": 0.6646598434677905,
      "grad_norm": 0.14250150322914124,
      "learning_rate": 7.990227028249678e-05,
      "loss": 0.8172465562820435,
      "step": 552
    },
    {
      "epoch": 0.6658639373871162,
      "grad_norm": 0.1538594663143158,
      "learning_rate": 7.982243072603306e-05,
      "loss": 0.8376189470291138,
      "step": 553
    },
    {
      "epoch": 0.6670680313064419,
      "grad_norm": 0.14822985231876373,
      "learning_rate": 7.974247296653028e-05,
      "loss": 0.8548381328582764,
      "step": 554
    },
    {
      "epoch": 0.6682721252257676,
      "grad_norm": 0.13167864084243774,
      "learning_rate": 7.966239732090592e-05,
      "loss": 0.6964908242225647,
      "step": 555
    },
    {
      "epoch": 0.6694762191450934,
      "grad_norm": 0.13585121929645538,
      "learning_rate": 7.958220410654475e-05,
      "loss": 0.7265467643737793,
      "step": 556
    },
    {
      "epoch": 0.670680313064419,
      "grad_norm": 0.143823504447937,
      "learning_rate": 7.95018936412975e-05,
      "loss": 0.7484427094459534,
      "step": 557
    },
    {
      "epoch": 0.6718844069837447,
      "grad_norm": 0.1312553733587265,
      "learning_rate": 7.942146624347964e-05,
      "loss": 0.6955775618553162,
      "step": 558
    },
    {
      "epoch": 0.6730885009030705,
      "grad_norm": 0.15476994216442108,
      "learning_rate": 7.93409222318701e-05,
      "loss": 0.8015362024307251,
      "step": 559
    },
    {
      "epoch": 0.6742925948223961,
      "grad_norm": 0.1393594741821289,
      "learning_rate": 7.926026192571007e-05,
      "loss": 0.710114061832428,
      "step": 560
    },
    {
      "epoch": 0.6754966887417219,
      "grad_norm": 0.14550964534282684,
      "learning_rate": 7.917948564470157e-05,
      "loss": 0.8316333889961243,
      "step": 561
    },
    {
      "epoch": 0.6767007826610476,
      "grad_norm": 0.1332012265920639,
      "learning_rate": 7.909859370900642e-05,
      "loss": 0.7794836163520813,
      "step": 562
    },
    {
      "epoch": 0.6779048765803732,
      "grad_norm": 0.16359443962574005,
      "learning_rate": 7.901758643924475e-05,
      "loss": 0.7346463203430176,
      "step": 563
    },
    {
      "epoch": 0.679108970499699,
      "grad_norm": 0.13705454766750336,
      "learning_rate": 7.893646415649384e-05,
      "loss": 0.8340646624565125,
      "step": 564
    },
    {
      "epoch": 0.6803130644190247,
      "grad_norm": 0.14088058471679688,
      "learning_rate": 7.88552271822869e-05,
      "loss": 0.8018200993537903,
      "step": 565
    },
    {
      "epoch": 0.6815171583383504,
      "grad_norm": 0.13394062221050262,
      "learning_rate": 7.877387583861165e-05,
      "loss": 0.6355478763580322,
      "step": 566
    },
    {
      "epoch": 0.6827212522576761,
      "grad_norm": 0.14850552380084991,
      "learning_rate": 7.869241044790915e-05,
      "loss": 0.8230655193328857,
      "step": 567
    },
    {
      "epoch": 0.6839253461770018,
      "grad_norm": 0.14765743911266327,
      "learning_rate": 7.861083133307247e-05,
      "loss": 0.7551670670509338,
      "step": 568
    },
    {
      "epoch": 0.6851294400963275,
      "grad_norm": 0.14147378504276276,
      "learning_rate": 7.852913881744547e-05,
      "loss": 0.8793714046478271,
      "step": 569
    },
    {
      "epoch": 0.6863335340156532,
      "grad_norm": 0.14339742064476013,
      "learning_rate": 7.844733322482145e-05,
      "loss": 0.7733831405639648,
      "step": 570
    },
    {
      "epoch": 0.687537627934979,
      "grad_norm": 0.14398925006389618,
      "learning_rate": 7.836541487944193e-05,
      "loss": 0.8752420544624329,
      "step": 571
    },
    {
      "epoch": 0.6887417218543046,
      "grad_norm": 0.16347667574882507,
      "learning_rate": 7.82833841059953e-05,
      "loss": 0.7307330965995789,
      "step": 572
    },
    {
      "epoch": 0.6899458157736303,
      "grad_norm": 0.13709548115730286,
      "learning_rate": 7.820124122961557e-05,
      "loss": 0.8230308890342712,
      "step": 573
    },
    {
      "epoch": 0.6911499096929561,
      "grad_norm": 0.1505584865808487,
      "learning_rate": 7.811898657588109e-05,
      "loss": 0.7051193714141846,
      "step": 574
    },
    {
      "epoch": 0.6923540036122817,
      "grad_norm": 0.13604536652565002,
      "learning_rate": 7.803662047081323e-05,
      "loss": 0.7320842742919922,
      "step": 575
    },
    {
      "epoch": 0.6935580975316075,
      "grad_norm": 0.140181764960289,
      "learning_rate": 7.795414324087515e-05,
      "loss": 0.7327534556388855,
      "step": 576
    },
    {
      "epoch": 0.6947621914509332,
      "grad_norm": 0.15296891331672668,
      "learning_rate": 7.78715552129704e-05,
      "loss": 0.7624315023422241,
      "step": 577
    },
    {
      "epoch": 0.6959662853702588,
      "grad_norm": 0.1467852145433426,
      "learning_rate": 7.778885671444169e-05,
      "loss": 0.7331960201263428,
      "step": 578
    },
    {
      "epoch": 0.6971703792895846,
      "grad_norm": 0.13914917409420013,
      "learning_rate": 7.770604807306966e-05,
      "loss": 0.7070205211639404,
      "step": 579
    },
    {
      "epoch": 0.6983744732089103,
      "grad_norm": 0.14825184643268585,
      "learning_rate": 7.762312961707141e-05,
      "loss": 0.745339035987854,
      "step": 580
    },
    {
      "epoch": 0.699578567128236,
      "grad_norm": 0.12418396770954132,
      "learning_rate": 7.754010167509935e-05,
      "loss": 0.6360680460929871,
      "step": 581
    },
    {
      "epoch": 0.7007826610475617,
      "grad_norm": 0.14344526827335358,
      "learning_rate": 7.745696457623986e-05,
      "loss": 0.6916585564613342,
      "step": 582
    },
    {
      "epoch": 0.7019867549668874,
      "grad_norm": 0.13355498015880585,
      "learning_rate": 7.73737186500119e-05,
      "loss": 0.7929323315620422,
      "step": 583
    },
    {
      "epoch": 0.7031908488862131,
      "grad_norm": 0.14853526651859283,
      "learning_rate": 7.729036422636589e-05,
      "loss": 0.7886224389076233,
      "step": 584
    },
    {
      "epoch": 0.7043949428055388,
      "grad_norm": 0.15538610517978668,
      "learning_rate": 7.720690163568214e-05,
      "loss": 0.6876072287559509,
      "step": 585
    },
    {
      "epoch": 0.7055990367248646,
      "grad_norm": 0.1566692739725113,
      "learning_rate": 7.712333120876983e-05,
      "loss": 0.7971206903457642,
      "step": 586
    },
    {
      "epoch": 0.7068031306441902,
      "grad_norm": 0.13929766416549683,
      "learning_rate": 7.703965327686544e-05,
      "loss": 0.7881407141685486,
      "step": 587
    },
    {
      "epoch": 0.708007224563516,
      "grad_norm": 0.1471068412065506,
      "learning_rate": 7.695586817163163e-05,
      "loss": 0.6721351146697998,
      "step": 588
    },
    {
      "epoch": 0.7092113184828417,
      "grad_norm": 0.14538784325122833,
      "learning_rate": 7.68719762251558e-05,
      "loss": 0.7565207481384277,
      "step": 589
    },
    {
      "epoch": 0.7104154124021673,
      "grad_norm": 0.1496374011039734,
      "learning_rate": 7.678797776994886e-05,
      "loss": 0.8598500490188599,
      "step": 590
    },
    {
      "epoch": 0.7116195063214931,
      "grad_norm": 0.1530589610338211,
      "learning_rate": 7.670387313894384e-05,
      "loss": 0.7822431325912476,
      "step": 591
    },
    {
      "epoch": 0.7128236002408188,
      "grad_norm": 0.1517655998468399,
      "learning_rate": 7.661966266549463e-05,
      "loss": 0.702894389629364,
      "step": 592
    },
    {
      "epoch": 0.7140276941601444,
      "grad_norm": 0.1469973623752594,
      "learning_rate": 7.653534668337463e-05,
      "loss": 0.8086850643157959,
      "step": 593
    },
    {
      "epoch": 0.7152317880794702,
      "grad_norm": 0.13878528773784637,
      "learning_rate": 7.645092552677539e-05,
      "loss": 0.8444103002548218,
      "step": 594
    },
    {
      "epoch": 0.7164358819987959,
      "grad_norm": 0.14931461215019226,
      "learning_rate": 7.636639953030541e-05,
      "loss": 0.9278316497802734,
      "step": 595
    },
    {
      "epoch": 0.7176399759181216,
      "grad_norm": 0.13261747360229492,
      "learning_rate": 7.628176902898863e-05,
      "loss": 0.7018995881080627,
      "step": 596
    },
    {
      "epoch": 0.7188440698374473,
      "grad_norm": 0.1461964100599289,
      "learning_rate": 7.619703435826328e-05,
      "loss": 0.7035484313964844,
      "step": 597
    },
    {
      "epoch": 0.720048163756773,
      "grad_norm": 0.15646134316921234,
      "learning_rate": 7.61121958539804e-05,
      "loss": 0.7596480250358582,
      "step": 598
    },
    {
      "epoch": 0.7212522576760987,
      "grad_norm": 0.14584432542324066,
      "learning_rate": 7.602725385240268e-05,
      "loss": 0.7328644394874573,
      "step": 599
    },
    {
      "epoch": 0.7224563515954244,
      "grad_norm": 0.1254493147134781,
      "learning_rate": 7.594220869020293e-05,
      "loss": 0.6355226039886475,
      "step": 600
    },
    {
      "epoch": 0.7236604455147502,
      "grad_norm": 0.13658897578716278,
      "learning_rate": 7.585706070446288e-05,
      "loss": 0.6962606906890869,
      "step": 601
    },
    {
      "epoch": 0.7248645394340758,
      "grad_norm": 0.13184867799282074,
      "learning_rate": 7.577181023267185e-05,
      "loss": 0.7538126707077026,
      "step": 602
    },
    {
      "epoch": 0.7260686333534015,
      "grad_norm": 0.12889988720417023,
      "learning_rate": 7.568645761272527e-05,
      "loss": 0.6959149837493896,
      "step": 603
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 0.1360941231250763,
      "learning_rate": 7.560100318292355e-05,
      "loss": 0.7438724040985107,
      "step": 604
    },
    {
      "epoch": 0.7284768211920529,
      "grad_norm": 0.1455586701631546,
      "learning_rate": 7.551544728197057e-05,
      "loss": 0.7490986585617065,
      "step": 605
    },
    {
      "epoch": 0.7296809151113787,
      "grad_norm": 0.12847721576690674,
      "learning_rate": 7.542979024897239e-05,
      "loss": 0.7193471789360046,
      "step": 606
    },
    {
      "epoch": 0.7308850090307044,
      "grad_norm": 0.14494414627552032,
      "learning_rate": 7.534403242343595e-05,
      "loss": 0.7355188727378845,
      "step": 607
    },
    {
      "epoch": 0.7320891029500302,
      "grad_norm": 0.1086283028125763,
      "learning_rate": 7.525817414526764e-05,
      "loss": 0.6301099061965942,
      "step": 608
    },
    {
      "epoch": 0.7332931968693558,
      "grad_norm": 0.13148364424705505,
      "learning_rate": 7.517221575477209e-05,
      "loss": 0.6432584524154663,
      "step": 609
    },
    {
      "epoch": 0.7344972907886815,
      "grad_norm": 0.1335676610469818,
      "learning_rate": 7.508615759265059e-05,
      "loss": 0.7239419221878052,
      "step": 610
    },
    {
      "epoch": 0.7357013847080073,
      "grad_norm": 0.13746167719364166,
      "learning_rate": 7.500000000000001e-05,
      "loss": 0.8151872754096985,
      "step": 611
    },
    {
      "epoch": 0.7369054786273329,
      "grad_norm": 0.1268731653690338,
      "learning_rate": 7.491374331831125e-05,
      "loss": 0.6599137783050537,
      "step": 612
    },
    {
      "epoch": 0.7381095725466587,
      "grad_norm": 0.12436234951019287,
      "learning_rate": 7.482738788946799e-05,
      "loss": 0.7022063136100769,
      "step": 613
    },
    {
      "epoch": 0.7393136664659844,
      "grad_norm": 0.135104238986969,
      "learning_rate": 7.474093405574527e-05,
      "loss": 0.8015887141227722,
      "step": 614
    },
    {
      "epoch": 0.74051776038531,
      "grad_norm": 0.15988576412200928,
      "learning_rate": 7.465438215980819e-05,
      "loss": 0.8476065397262573,
      "step": 615
    },
    {
      "epoch": 0.7417218543046358,
      "grad_norm": 0.13814352452754974,
      "learning_rate": 7.456773254471053e-05,
      "loss": 0.6568721532821655,
      "step": 616
    },
    {
      "epoch": 0.7429259482239615,
      "grad_norm": 0.13201281428337097,
      "learning_rate": 7.448098555389333e-05,
      "loss": 0.7030305862426758,
      "step": 617
    },
    {
      "epoch": 0.7441300421432872,
      "grad_norm": 0.1357838213443756,
      "learning_rate": 7.439414153118364e-05,
      "loss": 0.7653726935386658,
      "step": 618
    },
    {
      "epoch": 0.7453341360626129,
      "grad_norm": 0.13093659281730652,
      "learning_rate": 7.43072008207931e-05,
      "loss": 0.6980884075164795,
      "step": 619
    },
    {
      "epoch": 0.7465382299819386,
      "grad_norm": 0.13680848479270935,
      "learning_rate": 7.422016376731658e-05,
      "loss": 0.7077144384384155,
      "step": 620
    },
    {
      "epoch": 0.7477423239012643,
      "grad_norm": 0.1408122479915619,
      "learning_rate": 7.413303071573077e-05,
      "loss": 0.7148746252059937,
      "step": 621
    },
    {
      "epoch": 0.74894641782059,
      "grad_norm": 0.13702018558979034,
      "learning_rate": 7.404580201139286e-05,
      "loss": 0.7091426849365234,
      "step": 622
    },
    {
      "epoch": 0.7501505117399158,
      "grad_norm": 0.13043171167373657,
      "learning_rate": 7.395847800003925e-05,
      "loss": 0.7373806238174438,
      "step": 623
    },
    {
      "epoch": 0.7513546056592414,
      "grad_norm": 0.15487802028656006,
      "learning_rate": 7.387105902778397e-05,
      "loss": 0.7793884873390198,
      "step": 624
    },
    {
      "epoch": 0.7525586995785671,
      "grad_norm": 0.13877488672733307,
      "learning_rate": 7.378354544111755e-05,
      "loss": 0.8008089661598206,
      "step": 625
    },
    {
      "epoch": 0.7537627934978929,
      "grad_norm": 0.13346640765666962,
      "learning_rate": 7.36959375869054e-05,
      "loss": 0.8154155611991882,
      "step": 626
    },
    {
      "epoch": 0.7549668874172185,
      "grad_norm": 0.13805145025253296,
      "learning_rate": 7.360823581238672e-05,
      "loss": 0.8011735081672668,
      "step": 627
    },
    {
      "epoch": 0.7561709813365443,
      "grad_norm": 0.14190644025802612,
      "learning_rate": 7.352044046517285e-05,
      "loss": 0.7732957601547241,
      "step": 628
    },
    {
      "epoch": 0.75737507525587,
      "grad_norm": 0.12676766514778137,
      "learning_rate": 7.343255189324605e-05,
      "loss": 0.7489777207374573,
      "step": 629
    },
    {
      "epoch": 0.7585791691751956,
      "grad_norm": 0.13424651324748993,
      "learning_rate": 7.334457044495811e-05,
      "loss": 0.7267791032791138,
      "step": 630
    },
    {
      "epoch": 0.7597832630945214,
      "grad_norm": 0.13618718087673187,
      "learning_rate": 7.325649646902887e-05,
      "loss": 0.7499865293502808,
      "step": 631
    },
    {
      "epoch": 0.7609873570138471,
      "grad_norm": 0.12215069681406021,
      "learning_rate": 7.316833031454498e-05,
      "loss": 0.752011239528656,
      "step": 632
    },
    {
      "epoch": 0.7621914509331728,
      "grad_norm": 0.13151827454566956,
      "learning_rate": 7.30800723309584e-05,
      "loss": 0.7771649956703186,
      "step": 633
    },
    {
      "epoch": 0.7633955448524985,
      "grad_norm": 0.14986532926559448,
      "learning_rate": 7.299172286808511e-05,
      "loss": 0.7088158130645752,
      "step": 634
    },
    {
      "epoch": 0.7645996387718242,
      "grad_norm": 0.14447058737277985,
      "learning_rate": 7.290328227610362e-05,
      "loss": 0.7139779329299927,
      "step": 635
    },
    {
      "epoch": 0.7658037326911499,
      "grad_norm": 0.1423000544309616,
      "learning_rate": 7.281475090555365e-05,
      "loss": 0.7641744613647461,
      "step": 636
    },
    {
      "epoch": 0.7670078266104756,
      "grad_norm": 0.1317109316587448,
      "learning_rate": 7.272612910733475e-05,
      "loss": 0.6884832382202148,
      "step": 637
    },
    {
      "epoch": 0.7682119205298014,
      "grad_norm": 0.13702678680419922,
      "learning_rate": 7.263741723270486e-05,
      "loss": 0.7158911228179932,
      "step": 638
    },
    {
      "epoch": 0.769416014449127,
      "grad_norm": 0.12467742711305618,
      "learning_rate": 7.254861563327896e-05,
      "loss": 0.7161774039268494,
      "step": 639
    },
    {
      "epoch": 0.7706201083684527,
      "grad_norm": 0.13597269356250763,
      "learning_rate": 7.245972466102766e-05,
      "loss": 0.7500175833702087,
      "step": 640
    },
    {
      "epoch": 0.7718242022877785,
      "grad_norm": 0.1343757063150406,
      "learning_rate": 7.237074466827579e-05,
      "loss": 0.7022717595100403,
      "step": 641
    },
    {
      "epoch": 0.7730282962071041,
      "grad_norm": 0.14996503293514252,
      "learning_rate": 7.228167600770101e-05,
      "loss": 0.7420458793640137,
      "step": 642
    },
    {
      "epoch": 0.7742323901264299,
      "grad_norm": 0.14641951024532318,
      "learning_rate": 7.219251903233246e-05,
      "loss": 0.7393530011177063,
      "step": 643
    },
    {
      "epoch": 0.7754364840457556,
      "grad_norm": 0.1334446370601654,
      "learning_rate": 7.210327409554926e-05,
      "loss": 0.6367544531822205,
      "step": 644
    },
    {
      "epoch": 0.7766405779650812,
      "grad_norm": 0.14800557494163513,
      "learning_rate": 7.201394155107928e-05,
      "loss": 0.7870895266532898,
      "step": 645
    },
    {
      "epoch": 0.777844671884407,
      "grad_norm": 0.15482763946056366,
      "learning_rate": 7.192452175299748e-05,
      "loss": 0.8093364238739014,
      "step": 646
    },
    {
      "epoch": 0.7790487658037327,
      "grad_norm": 0.15954411029815674,
      "learning_rate": 7.183501505572478e-05,
      "loss": 0.7867874503135681,
      "step": 647
    },
    {
      "epoch": 0.7802528597230584,
      "grad_norm": 0.14859770238399506,
      "learning_rate": 7.174542181402646e-05,
      "loss": 0.6899842023849487,
      "step": 648
    },
    {
      "epoch": 0.7814569536423841,
      "grad_norm": 0.13135534524917603,
      "learning_rate": 7.165574238301085e-05,
      "loss": 0.6626449823379517,
      "step": 649
    },
    {
      "epoch": 0.7826610475617098,
      "grad_norm": 0.1352769434452057,
      "learning_rate": 7.15659771181279e-05,
      "loss": 0.7246488332748413,
      "step": 650
    },
    {
      "epoch": 0.7838651414810355,
      "grad_norm": 0.14771783351898193,
      "learning_rate": 7.147612637516775e-05,
      "loss": 0.8368797302246094,
      "step": 651
    },
    {
      "epoch": 0.7850692354003612,
      "grad_norm": 0.14814864099025726,
      "learning_rate": 7.138619051025935e-05,
      "loss": 0.8104938268661499,
      "step": 652
    },
    {
      "epoch": 0.786273329319687,
      "grad_norm": 0.14024882018566132,
      "learning_rate": 7.129616987986905e-05,
      "loss": 0.7284850478172302,
      "step": 653
    },
    {
      "epoch": 0.7874774232390126,
      "grad_norm": 0.12419838458299637,
      "learning_rate": 7.120606484079912e-05,
      "loss": 0.8031896352767944,
      "step": 654
    },
    {
      "epoch": 0.7886815171583383,
      "grad_norm": 0.14694742858409882,
      "learning_rate": 7.111587575018648e-05,
      "loss": 0.7721980214118958,
      "step": 655
    },
    {
      "epoch": 0.7898856110776641,
      "grad_norm": 0.14434750378131866,
      "learning_rate": 7.102560296550109e-05,
      "loss": 0.8467543721199036,
      "step": 656
    },
    {
      "epoch": 0.7910897049969897,
      "grad_norm": 0.13904523849487305,
      "learning_rate": 7.093524684454471e-05,
      "loss": 0.7543468475341797,
      "step": 657
    },
    {
      "epoch": 0.7922937989163155,
      "grad_norm": 0.1310669481754303,
      "learning_rate": 7.084480774544937e-05,
      "loss": 0.7381628751754761,
      "step": 658
    },
    {
      "epoch": 0.7934978928356412,
      "grad_norm": 0.13590441644191742,
      "learning_rate": 7.075428602667602e-05,
      "loss": 0.7622985243797302,
      "step": 659
    },
    {
      "epoch": 0.7947019867549668,
      "grad_norm": 0.1434076726436615,
      "learning_rate": 7.066368204701306e-05,
      "loss": 0.7465397119522095,
      "step": 660
    },
    {
      "epoch": 0.7959060806742926,
      "grad_norm": 0.11832129210233688,
      "learning_rate": 7.057299616557493e-05,
      "loss": 0.710469126701355,
      "step": 661
    },
    {
      "epoch": 0.7971101745936183,
      "grad_norm": 0.13439592719078064,
      "learning_rate": 7.048222874180072e-05,
      "loss": 0.674365758895874,
      "step": 662
    },
    {
      "epoch": 0.798314268512944,
      "grad_norm": 0.1394403874874115,
      "learning_rate": 7.039138013545265e-05,
      "loss": 0.8113305568695068,
      "step": 663
    },
    {
      "epoch": 0.7995183624322697,
      "grad_norm": 0.14216552674770355,
      "learning_rate": 7.030045070661484e-05,
      "loss": 0.7501247525215149,
      "step": 664
    },
    {
      "epoch": 0.8007224563515954,
      "grad_norm": 0.12801142036914825,
      "learning_rate": 7.020944081569158e-05,
      "loss": 0.680537223815918,
      "step": 665
    },
    {
      "epoch": 0.8019265502709211,
      "grad_norm": 0.14361411333084106,
      "learning_rate": 7.011835082340625e-05,
      "loss": 0.7748258113861084,
      "step": 666
    },
    {
      "epoch": 0.8031306441902468,
      "grad_norm": 0.2736600637435913,
      "learning_rate": 7.002718109079964e-05,
      "loss": 0.7710071802139282,
      "step": 667
    },
    {
      "epoch": 0.8043347381095726,
      "grad_norm": 0.13474589586257935,
      "learning_rate": 6.993593197922852e-05,
      "loss": 0.6965528726577759,
      "step": 668
    },
    {
      "epoch": 0.8055388320288982,
      "grad_norm": 0.12602798640727997,
      "learning_rate": 6.984460385036442e-05,
      "loss": 0.818771243095398,
      "step": 669
    },
    {
      "epoch": 0.8067429259482239,
      "grad_norm": 0.13726601004600525,
      "learning_rate": 6.975319706619197e-05,
      "loss": 0.7508562803268433,
      "step": 670
    },
    {
      "epoch": 0.8079470198675497,
      "grad_norm": 0.1314438283443451,
      "learning_rate": 6.966171198900761e-05,
      "loss": 0.7460685968399048,
      "step": 671
    },
    {
      "epoch": 0.8091511137868754,
      "grad_norm": 0.17150211334228516,
      "learning_rate": 6.957014898141805e-05,
      "loss": 0.7301174402236938,
      "step": 672
    },
    {
      "epoch": 0.8103552077062011,
      "grad_norm": 0.125514417886734,
      "learning_rate": 6.947850840633892e-05,
      "loss": 0.6950365304946899,
      "step": 673
    },
    {
      "epoch": 0.8115593016255268,
      "grad_norm": 0.1376231461763382,
      "learning_rate": 6.938679062699327e-05,
      "loss": 0.7544848918914795,
      "step": 674
    },
    {
      "epoch": 0.8127633955448526,
      "grad_norm": 0.13513053953647614,
      "learning_rate": 6.929499600691014e-05,
      "loss": 0.8200382590293884,
      "step": 675
    },
    {
      "epoch": 0.8139674894641782,
      "grad_norm": 0.13901929557323456,
      "learning_rate": 6.92031249099232e-05,
      "loss": 0.7404852509498596,
      "step": 676
    },
    {
      "epoch": 0.8151715833835039,
      "grad_norm": 0.12839938700199127,
      "learning_rate": 6.911117770016915e-05,
      "loss": 0.6891971826553345,
      "step": 677
    },
    {
      "epoch": 0.8163756773028297,
      "grad_norm": 0.1373388171195984,
      "learning_rate": 6.901915474208644e-05,
      "loss": 0.773501992225647,
      "step": 678
    },
    {
      "epoch": 0.8175797712221553,
      "grad_norm": 0.14600759744644165,
      "learning_rate": 6.892705640041373e-05,
      "loss": 0.7567825317382812,
      "step": 679
    },
    {
      "epoch": 0.818783865141481,
      "grad_norm": 0.15081800520420074,
      "learning_rate": 6.883488304018844e-05,
      "loss": 0.740884006023407,
      "step": 680
    },
    {
      "epoch": 0.8199879590608068,
      "grad_norm": 0.13229772448539734,
      "learning_rate": 6.874263502674538e-05,
      "loss": 0.6893157958984375,
      "step": 681
    },
    {
      "epoch": 0.8211920529801324,
      "grad_norm": 0.13728021085262299,
      "learning_rate": 6.86503127257152e-05,
      "loss": 0.7654533386230469,
      "step": 682
    },
    {
      "epoch": 0.8223961468994582,
      "grad_norm": 0.15406455099582672,
      "learning_rate": 6.855791650302305e-05,
      "loss": 0.8069047331809998,
      "step": 683
    },
    {
      "epoch": 0.8236002408187839,
      "grad_norm": 0.14553096890449524,
      "learning_rate": 6.846544672488701e-05,
      "loss": 0.8268608450889587,
      "step": 684
    },
    {
      "epoch": 0.8248043347381095,
      "grad_norm": 0.13753806054592133,
      "learning_rate": 6.837290375781678e-05,
      "loss": 0.7510198950767517,
      "step": 685
    },
    {
      "epoch": 0.8260084286574353,
      "grad_norm": 0.1292235553264618,
      "learning_rate": 6.828028796861207e-05,
      "loss": 0.7316303849220276,
      "step": 686
    },
    {
      "epoch": 0.827212522576761,
      "grad_norm": 0.14600948989391327,
      "learning_rate": 6.818759972436125e-05,
      "loss": 0.8135302662849426,
      "step": 687
    },
    {
      "epoch": 0.8284166164960867,
      "grad_norm": 0.132733553647995,
      "learning_rate": 6.809483939243992e-05,
      "loss": 0.7760171890258789,
      "step": 688
    },
    {
      "epoch": 0.8296207104154124,
      "grad_norm": 0.1248960867524147,
      "learning_rate": 6.800200734050931e-05,
      "loss": 0.7128435969352722,
      "step": 689
    },
    {
      "epoch": 0.8308248043347382,
      "grad_norm": 0.13273389637470245,
      "learning_rate": 6.790910393651502e-05,
      "loss": 0.7495109438896179,
      "step": 690
    },
    {
      "epoch": 0.8320288982540638,
      "grad_norm": 0.13568413257598877,
      "learning_rate": 6.781612954868538e-05,
      "loss": 0.6649609804153442,
      "step": 691
    },
    {
      "epoch": 0.8332329921733895,
      "grad_norm": 0.13244058191776276,
      "learning_rate": 6.77230845455301e-05,
      "loss": 0.6891012787818909,
      "step": 692
    },
    {
      "epoch": 0.8344370860927153,
      "grad_norm": 0.13847336173057556,
      "learning_rate": 6.762996929583878e-05,
      "loss": 0.7848809957504272,
      "step": 693
    },
    {
      "epoch": 0.8356411800120409,
      "grad_norm": 0.1425885260105133,
      "learning_rate": 6.753678416867944e-05,
      "loss": 0.6814917922019958,
      "step": 694
    },
    {
      "epoch": 0.8368452739313667,
      "grad_norm": 0.12745307385921478,
      "learning_rate": 6.744352953339706e-05,
      "loss": 0.7613441348075867,
      "step": 695
    },
    {
      "epoch": 0.8380493678506924,
      "grad_norm": 0.15960842370986938,
      "learning_rate": 6.735020575961213e-05,
      "loss": 0.7680386304855347,
      "step": 696
    },
    {
      "epoch": 0.839253461770018,
      "grad_norm": 0.13327401876449585,
      "learning_rate": 6.725681321721916e-05,
      "loss": 0.8069470524787903,
      "step": 697
    },
    {
      "epoch": 0.8404575556893438,
      "grad_norm": 0.13140256702899933,
      "learning_rate": 6.716335227638525e-05,
      "loss": 0.6358769536018372,
      "step": 698
    },
    {
      "epoch": 0.8416616496086695,
      "grad_norm": 0.1473989188671112,
      "learning_rate": 6.706982330754858e-05,
      "loss": 0.7711199522018433,
      "step": 699
    },
    {
      "epoch": 0.8428657435279951,
      "grad_norm": 0.12495903670787811,
      "learning_rate": 6.697622668141698e-05,
      "loss": 0.6610642671585083,
      "step": 700
    },
    {
      "epoch": 0.8440698374473209,
      "grad_norm": 0.1274222880601883,
      "learning_rate": 6.688256276896643e-05,
      "loss": 0.7276433110237122,
      "step": 701
    },
    {
      "epoch": 0.8452739313666466,
      "grad_norm": 0.14475814998149872,
      "learning_rate": 6.678883194143962e-05,
      "loss": 0.8190998435020447,
      "step": 702
    },
    {
      "epoch": 0.8464780252859723,
      "grad_norm": 0.14313237369060516,
      "learning_rate": 6.669503457034446e-05,
      "loss": 0.7752524018287659,
      "step": 703
    },
    {
      "epoch": 0.847682119205298,
      "grad_norm": 0.1357709914445877,
      "learning_rate": 6.660117102745256e-05,
      "loss": 0.8446824550628662,
      "step": 704
    },
    {
      "epoch": 0.8488862131246238,
      "grad_norm": 0.12758085131645203,
      "learning_rate": 6.650724168479789e-05,
      "loss": 0.763249397277832,
      "step": 705
    },
    {
      "epoch": 0.8500903070439494,
      "grad_norm": 0.14276817440986633,
      "learning_rate": 6.641324691467514e-05,
      "loss": 0.7540005445480347,
      "step": 706
    },
    {
      "epoch": 0.8512944009632751,
      "grad_norm": 0.12354648858308792,
      "learning_rate": 6.63191870896384e-05,
      "loss": 0.6784765124320984,
      "step": 707
    },
    {
      "epoch": 0.8524984948826009,
      "grad_norm": 0.14203065633773804,
      "learning_rate": 6.622506258249956e-05,
      "loss": 0.746931791305542,
      "step": 708
    },
    {
      "epoch": 0.8537025888019265,
      "grad_norm": 0.13627232611179352,
      "learning_rate": 6.61308737663269e-05,
      "loss": 0.7391974925994873,
      "step": 709
    },
    {
      "epoch": 0.8549066827212523,
      "grad_norm": 0.1430472731590271,
      "learning_rate": 6.60366210144436e-05,
      "loss": 0.691170334815979,
      "step": 710
    },
    {
      "epoch": 0.856110776640578,
      "grad_norm": 0.140480637550354,
      "learning_rate": 6.59423047004262e-05,
      "loss": 0.6575626134872437,
      "step": 711
    },
    {
      "epoch": 0.8573148705599036,
      "grad_norm": 0.148088276386261,
      "learning_rate": 6.584792519810325e-05,
      "loss": 0.7172805666923523,
      "step": 712
    },
    {
      "epoch": 0.8585189644792294,
      "grad_norm": 0.14462804794311523,
      "learning_rate": 6.57534828815537e-05,
      "loss": 0.8729466795921326,
      "step": 713
    },
    {
      "epoch": 0.8597230583985551,
      "grad_norm": 0.13183382153511047,
      "learning_rate": 6.565897812510549e-05,
      "loss": 0.814098060131073,
      "step": 714
    },
    {
      "epoch": 0.8609271523178808,
      "grad_norm": 0.139236718416214,
      "learning_rate": 6.556441130333403e-05,
      "loss": 0.7661739587783813,
      "step": 715
    },
    {
      "epoch": 0.8621312462372065,
      "grad_norm": 0.1398772895336151,
      "learning_rate": 6.546978279106074e-05,
      "loss": 0.7995048761367798,
      "step": 716
    },
    {
      "epoch": 0.8633353401565322,
      "grad_norm": 0.1288633644580841,
      "learning_rate": 6.537509296335155e-05,
      "loss": 0.6718826293945312,
      "step": 717
    },
    {
      "epoch": 0.8645394340758579,
      "grad_norm": 0.1483159214258194,
      "learning_rate": 6.528034219551543e-05,
      "loss": 0.7817126512527466,
      "step": 718
    },
    {
      "epoch": 0.8657435279951836,
      "grad_norm": 0.15745297074317932,
      "learning_rate": 6.518553086310285e-05,
      "loss": 0.7215754985809326,
      "step": 719
    },
    {
      "epoch": 0.8669476219145094,
      "grad_norm": 0.12937229871749878,
      "learning_rate": 6.509065934190437e-05,
      "loss": 0.5938048958778381,
      "step": 720
    },
    {
      "epoch": 0.868151715833835,
      "grad_norm": 0.13402795791625977,
      "learning_rate": 6.499572800794911e-05,
      "loss": 0.7180178761482239,
      "step": 721
    },
    {
      "epoch": 0.8693558097531607,
      "grad_norm": 0.14314861595630646,
      "learning_rate": 6.490073723750326e-05,
      "loss": 0.7653616070747375,
      "step": 722
    },
    {
      "epoch": 0.8705599036724865,
      "grad_norm": 0.1314808875322342,
      "learning_rate": 6.480568740706856e-05,
      "loss": 0.7359954714775085,
      "step": 723
    },
    {
      "epoch": 0.8717639975918121,
      "grad_norm": 0.13052570819854736,
      "learning_rate": 6.471057889338089e-05,
      "loss": 0.7510445713996887,
      "step": 724
    },
    {
      "epoch": 0.8729680915111379,
      "grad_norm": 0.1357552707195282,
      "learning_rate": 6.461541207340866e-05,
      "loss": 0.8839564323425293,
      "step": 725
    },
    {
      "epoch": 0.8741721854304636,
      "grad_norm": 0.14918017387390137,
      "learning_rate": 6.452018732435145e-05,
      "loss": 0.8053610324859619,
      "step": 726
    },
    {
      "epoch": 0.8753762793497892,
      "grad_norm": 0.14189930260181427,
      "learning_rate": 6.442490502363838e-05,
      "loss": 0.7528836727142334,
      "step": 727
    },
    {
      "epoch": 0.876580373269115,
      "grad_norm": 0.16059346497058868,
      "learning_rate": 6.432956554892675e-05,
      "loss": 0.7770289778709412,
      "step": 728
    },
    {
      "epoch": 0.8777844671884407,
      "grad_norm": 0.1507798135280609,
      "learning_rate": 6.42341692781004e-05,
      "loss": 0.8498313426971436,
      "step": 729
    },
    {
      "epoch": 0.8789885611077664,
      "grad_norm": 0.1525810956954956,
      "learning_rate": 6.413871658926833e-05,
      "loss": 0.7812171578407288,
      "step": 730
    },
    {
      "epoch": 0.8801926550270921,
      "grad_norm": 0.1425810307264328,
      "learning_rate": 6.404320786076317e-05,
      "loss": 0.7805963158607483,
      "step": 731
    },
    {
      "epoch": 0.8813967489464178,
      "grad_norm": 0.13486258685588837,
      "learning_rate": 6.39476434711396e-05,
      "loss": 0.770559549331665,
      "step": 732
    },
    {
      "epoch": 0.8826008428657435,
      "grad_norm": 0.14722631871700287,
      "learning_rate": 6.385202379917297e-05,
      "loss": 0.7062653303146362,
      "step": 733
    },
    {
      "epoch": 0.8838049367850692,
      "grad_norm": 0.19616195559501648,
      "learning_rate": 6.375634922385775e-05,
      "loss": 0.7493513822555542,
      "step": 734
    },
    {
      "epoch": 0.885009030704395,
      "grad_norm": 0.1783919632434845,
      "learning_rate": 6.366062012440599e-05,
      "loss": 0.7614865303039551,
      "step": 735
    },
    {
      "epoch": 0.8862131246237207,
      "grad_norm": 0.13586460053920746,
      "learning_rate": 6.356483688024588e-05,
      "loss": 0.6685861349105835,
      "step": 736
    },
    {
      "epoch": 0.8874172185430463,
      "grad_norm": 0.13347139954566956,
      "learning_rate": 6.346899987102019e-05,
      "loss": 0.7626739740371704,
      "step": 737
    },
    {
      "epoch": 0.8886213124623721,
      "grad_norm": 0.13073810935020447,
      "learning_rate": 6.337310947658478e-05,
      "loss": 0.7991435527801514,
      "step": 738
    },
    {
      "epoch": 0.8898254063816978,
      "grad_norm": 0.1310962438583374,
      "learning_rate": 6.327716607700719e-05,
      "loss": 0.8112761974334717,
      "step": 739
    },
    {
      "epoch": 0.8910295003010235,
      "grad_norm": 0.13366377353668213,
      "learning_rate": 6.318117005256494e-05,
      "loss": 0.7123982906341553,
      "step": 740
    },
    {
      "epoch": 0.8922335942203492,
      "grad_norm": 0.13902029395103455,
      "learning_rate": 6.308512178374419e-05,
      "loss": 0.7494668364524841,
      "step": 741
    },
    {
      "epoch": 0.893437688139675,
      "grad_norm": 0.1480409801006317,
      "learning_rate": 6.298902165123815e-05,
      "loss": 0.7715325355529785,
      "step": 742
    },
    {
      "epoch": 0.8946417820590006,
      "grad_norm": 0.12304423004388809,
      "learning_rate": 6.289287003594564e-05,
      "loss": 0.633497416973114,
      "step": 743
    },
    {
      "epoch": 0.8958458759783263,
      "grad_norm": 0.12992003560066223,
      "learning_rate": 6.279666731896946e-05,
      "loss": 0.608023464679718,
      "step": 744
    },
    {
      "epoch": 0.8970499698976521,
      "grad_norm": 0.14757700264453888,
      "learning_rate": 6.270041388161503e-05,
      "loss": 0.804186224937439,
      "step": 745
    },
    {
      "epoch": 0.8982540638169777,
      "grad_norm": 0.12987659871578217,
      "learning_rate": 6.26041101053888e-05,
      "loss": 0.7719977498054504,
      "step": 746
    },
    {
      "epoch": 0.8994581577363034,
      "grad_norm": 0.14930501580238342,
      "learning_rate": 6.250775637199661e-05,
      "loss": 0.6917624473571777,
      "step": 747
    },
    {
      "epoch": 0.9006622516556292,
      "grad_norm": 0.1476936638355255,
      "learning_rate": 6.241135306334254e-05,
      "loss": 0.7304196953773499,
      "step": 748
    },
    {
      "epoch": 0.9018663455749548,
      "grad_norm": 0.14694079756736755,
      "learning_rate": 6.231490056152692e-05,
      "loss": 0.7570903897285461,
      "step": 749
    },
    {
      "epoch": 0.9030704394942806,
      "grad_norm": 0.1306212842464447,
      "learning_rate": 6.221839924884527e-05,
      "loss": 0.7606673836708069,
      "step": 750
    },
    {
      "epoch": 0.9042745334136063,
      "grad_norm": 0.14507654309272766,
      "learning_rate": 6.21218495077864e-05,
      "loss": 0.7652634382247925,
      "step": 751
    },
    {
      "epoch": 0.9054786273329319,
      "grad_norm": 0.17458806931972504,
      "learning_rate": 6.20252517210312e-05,
      "loss": 0.8033649921417236,
      "step": 752
    },
    {
      "epoch": 0.9066827212522577,
      "grad_norm": 0.12614940106868744,
      "learning_rate": 6.192860627145094e-05,
      "loss": 0.631454348564148,
      "step": 753
    },
    {
      "epoch": 0.9078868151715834,
      "grad_norm": 0.139654740691185,
      "learning_rate": 6.183191354210577e-05,
      "loss": 0.8121630549430847,
      "step": 754
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 0.141023188829422,
      "learning_rate": 6.173517391624331e-05,
      "loss": 0.7503607273101807,
      "step": 755
    },
    {
      "epoch": 0.9102950030102348,
      "grad_norm": 0.1344241499900818,
      "learning_rate": 6.163838777729699e-05,
      "loss": 0.7310186624526978,
      "step": 756
    },
    {
      "epoch": 0.9114990969295605,
      "grad_norm": 0.14049741625785828,
      "learning_rate": 6.154155550888466e-05,
      "loss": 0.6904723048210144,
      "step": 757
    },
    {
      "epoch": 0.9127031908488862,
      "grad_norm": 0.14592325687408447,
      "learning_rate": 6.144467749480695e-05,
      "loss": 0.7393869757652283,
      "step": 758
    },
    {
      "epoch": 0.9139072847682119,
      "grad_norm": 0.1442546546459198,
      "learning_rate": 6.134775411904581e-05,
      "loss": 0.6929787993431091,
      "step": 759
    },
    {
      "epoch": 0.9151113786875377,
      "grad_norm": 0.13714499771595,
      "learning_rate": 6.125078576576306e-05,
      "loss": 0.6989441514015198,
      "step": 760
    },
    {
      "epoch": 0.9163154726068633,
      "grad_norm": 0.14847946166992188,
      "learning_rate": 6.115377281929867e-05,
      "loss": 0.7599726915359497,
      "step": 761
    },
    {
      "epoch": 0.917519566526189,
      "grad_norm": 0.1352458894252777,
      "learning_rate": 6.105671566416947e-05,
      "loss": 0.7713584899902344,
      "step": 762
    },
    {
      "epoch": 0.9187236604455148,
      "grad_norm": 0.13632886111736298,
      "learning_rate": 6.0959614685067444e-05,
      "loss": 0.7323758006095886,
      "step": 763
    },
    {
      "epoch": 0.9199277543648404,
      "grad_norm": 0.15880534052848816,
      "learning_rate": 6.086247026685832e-05,
      "loss": 0.7737868428230286,
      "step": 764
    },
    {
      "epoch": 0.9211318482841662,
      "grad_norm": 0.12512539327144623,
      "learning_rate": 6.0765282794579934e-05,
      "loss": 0.6763508915901184,
      "step": 765
    },
    {
      "epoch": 0.9223359422034919,
      "grad_norm": 0.14834412932395935,
      "learning_rate": 6.066805265344084e-05,
      "loss": 0.8439559936523438,
      "step": 766
    },
    {
      "epoch": 0.9235400361228175,
      "grad_norm": 0.14139895141124725,
      "learning_rate": 6.0570780228818705e-05,
      "loss": 0.6648552417755127,
      "step": 767
    },
    {
      "epoch": 0.9247441300421433,
      "grad_norm": 0.13584230840206146,
      "learning_rate": 6.047346590625872e-05,
      "loss": 0.7966293096542358,
      "step": 768
    },
    {
      "epoch": 0.925948223961469,
      "grad_norm": 0.13365517556667328,
      "learning_rate": 6.0376110071472234e-05,
      "loss": 0.6662843227386475,
      "step": 769
    },
    {
      "epoch": 0.9271523178807947,
      "grad_norm": 0.14853501319885254,
      "learning_rate": 6.027871311033506e-05,
      "loss": 0.7631108164787292,
      "step": 770
    },
    {
      "epoch": 0.9283564118001204,
      "grad_norm": 0.12813806533813477,
      "learning_rate": 6.0181275408886064e-05,
      "loss": 0.6767235994338989,
      "step": 771
    },
    {
      "epoch": 0.9295605057194462,
      "grad_norm": 0.1400189995765686,
      "learning_rate": 6.008379735332556e-05,
      "loss": 0.7986948490142822,
      "step": 772
    },
    {
      "epoch": 0.9307645996387718,
      "grad_norm": 0.12961047887802124,
      "learning_rate": 5.998627933001381e-05,
      "loss": 0.7983364462852478,
      "step": 773
    },
    {
      "epoch": 0.9319686935580975,
      "grad_norm": 0.13788118958473206,
      "learning_rate": 5.988872172546952e-05,
      "loss": 0.751894474029541,
      "step": 774
    },
    {
      "epoch": 0.9331727874774233,
      "grad_norm": 0.15322254598140717,
      "learning_rate": 5.979112492636824e-05,
      "loss": 0.8806198835372925,
      "step": 775
    },
    {
      "epoch": 0.9343768813967489,
      "grad_norm": 0.15146565437316895,
      "learning_rate": 5.9693489319540906e-05,
      "loss": 0.8116838335990906,
      "step": 776
    },
    {
      "epoch": 0.9355809753160746,
      "grad_norm": 0.15707086026668549,
      "learning_rate": 5.959581529197225e-05,
      "loss": 0.7261554002761841,
      "step": 777
    },
    {
      "epoch": 0.9367850692354004,
      "grad_norm": 0.14480994641780853,
      "learning_rate": 5.9498103230799274e-05,
      "loss": 0.7364398241043091,
      "step": 778
    },
    {
      "epoch": 0.937989163154726,
      "grad_norm": 0.14421702921390533,
      "learning_rate": 5.940035352330975e-05,
      "loss": 0.7090728282928467,
      "step": 779
    },
    {
      "epoch": 0.9391932570740518,
      "grad_norm": 0.12545166909694672,
      "learning_rate": 5.9302566556940654e-05,
      "loss": 0.6562091708183289,
      "step": 780
    },
    {
      "epoch": 0.9403973509933775,
      "grad_norm": 0.15133622288703918,
      "learning_rate": 5.9204742719276676e-05,
      "loss": 0.8031641840934753,
      "step": 781
    },
    {
      "epoch": 0.9416014449127031,
      "grad_norm": 0.12315740436315536,
      "learning_rate": 5.910688239804857e-05,
      "loss": 0.691980242729187,
      "step": 782
    },
    {
      "epoch": 0.9428055388320289,
      "grad_norm": 0.14063894748687744,
      "learning_rate": 5.900898598113177e-05,
      "loss": 0.6285038590431213,
      "step": 783
    },
    {
      "epoch": 0.9440096327513546,
      "grad_norm": 0.12904803454875946,
      "learning_rate": 5.891105385654474e-05,
      "loss": 0.8334552049636841,
      "step": 784
    },
    {
      "epoch": 0.9452137266706803,
      "grad_norm": 0.14659355580806732,
      "learning_rate": 5.881308641244747e-05,
      "loss": 0.796967625617981,
      "step": 785
    },
    {
      "epoch": 0.946417820590006,
      "grad_norm": 0.13811902701854706,
      "learning_rate": 5.871508403713997e-05,
      "loss": 0.7257508635520935,
      "step": 786
    },
    {
      "epoch": 0.9476219145093318,
      "grad_norm": 0.13612543046474457,
      "learning_rate": 5.861704711906067e-05,
      "loss": 0.6375232934951782,
      "step": 787
    },
    {
      "epoch": 0.9488260084286574,
      "grad_norm": 0.13595229387283325,
      "learning_rate": 5.8518976046784944e-05,
      "loss": 0.7213425636291504,
      "step": 788
    },
    {
      "epoch": 0.9500301023479831,
      "grad_norm": 0.1368950754404068,
      "learning_rate": 5.842087120902351e-05,
      "loss": 0.7742677330970764,
      "step": 789
    },
    {
      "epoch": 0.9512341962673089,
      "grad_norm": 0.14558850228786469,
      "learning_rate": 5.832273299462092e-05,
      "loss": 0.82569420337677,
      "step": 790
    },
    {
      "epoch": 0.9524382901866345,
      "grad_norm": 0.13086587190628052,
      "learning_rate": 5.8224561792554036e-05,
      "loss": 0.6587938070297241,
      "step": 791
    },
    {
      "epoch": 0.9536423841059603,
      "grad_norm": 0.12629231810569763,
      "learning_rate": 5.812635799193046e-05,
      "loss": 0.7619765400886536,
      "step": 792
    },
    {
      "epoch": 0.954846478025286,
      "grad_norm": 0.13679230213165283,
      "learning_rate": 5.802812198198699e-05,
      "loss": 0.7186375856399536,
      "step": 793
    },
    {
      "epoch": 0.9560505719446116,
      "grad_norm": 0.14019349217414856,
      "learning_rate": 5.792985415208809e-05,
      "loss": 0.7573325037956238,
      "step": 794
    },
    {
      "epoch": 0.9572546658639374,
      "grad_norm": 0.1353059709072113,
      "learning_rate": 5.783155489172437e-05,
      "loss": 0.665139377117157,
      "step": 795
    },
    {
      "epoch": 0.9584587597832631,
      "grad_norm": 0.1335773766040802,
      "learning_rate": 5.773322459051098e-05,
      "loss": 0.79070645570755,
      "step": 796
    },
    {
      "epoch": 0.9596628537025887,
      "grad_norm": 0.1414458453655243,
      "learning_rate": 5.763486363818613e-05,
      "loss": 0.7078026533126831,
      "step": 797
    },
    {
      "epoch": 0.9608669476219145,
      "grad_norm": 0.15831796824932098,
      "learning_rate": 5.75364724246095e-05,
      "loss": 0.7539790868759155,
      "step": 798
    },
    {
      "epoch": 0.9620710415412402,
      "grad_norm": 0.1731119304895401,
      "learning_rate": 5.743805133976071e-05,
      "loss": 0.7221494317054749,
      "step": 799
    },
    {
      "epoch": 0.963275135460566,
      "grad_norm": 0.13116182386875153,
      "learning_rate": 5.733960077373781e-05,
      "loss": 0.712371289730072,
      "step": 800
    },
    {
      "epoch": 0.9644792293798916,
      "grad_norm": 0.15551303327083588,
      "learning_rate": 5.7241121116755646e-05,
      "loss": 0.6562230587005615,
      "step": 801
    },
    {
      "epoch": 0.9656833232992174,
      "grad_norm": 0.1384788304567337,
      "learning_rate": 5.714261275914442e-05,
      "loss": 0.7449044585227966,
      "step": 802
    },
    {
      "epoch": 0.9668874172185431,
      "grad_norm": 0.15467505156993866,
      "learning_rate": 5.704407609134805e-05,
      "loss": 0.7772901058197021,
      "step": 803
    },
    {
      "epoch": 0.9680915111378687,
      "grad_norm": 0.14512059092521667,
      "learning_rate": 5.694551150392271e-05,
      "loss": 0.7369959354400635,
      "step": 804
    },
    {
      "epoch": 0.9692956050571945,
      "grad_norm": 0.14831583201885223,
      "learning_rate": 5.684691938753517e-05,
      "loss": 0.7524558901786804,
      "step": 805
    },
    {
      "epoch": 0.9704996989765202,
      "grad_norm": 0.128688782453537,
      "learning_rate": 5.674830013296137e-05,
      "loss": 0.7282066941261292,
      "step": 806
    },
    {
      "epoch": 0.9717037928958459,
      "grad_norm": 0.1558089554309845,
      "learning_rate": 5.664965413108481e-05,
      "loss": 0.8398785591125488,
      "step": 807
    },
    {
      "epoch": 0.9729078868151716,
      "grad_norm": 0.1662004441022873,
      "learning_rate": 5.655098177289496e-05,
      "loss": 0.7244765758514404,
      "step": 808
    },
    {
      "epoch": 0.9741119807344973,
      "grad_norm": 0.15641339123249054,
      "learning_rate": 5.6452283449485774e-05,
      "loss": 0.7376512289047241,
      "step": 809
    },
    {
      "epoch": 0.975316074653823,
      "grad_norm": 0.15008556842803955,
      "learning_rate": 5.635355955205416e-05,
      "loss": 0.6745847463607788,
      "step": 810
    },
    {
      "epoch": 0.9765201685731487,
      "grad_norm": 0.1481301188468933,
      "learning_rate": 5.625481047189835e-05,
      "loss": 0.6975297331809998,
      "step": 811
    },
    {
      "epoch": 0.9777242624924745,
      "grad_norm": 0.1616453230381012,
      "learning_rate": 5.6156036600416385e-05,
      "loss": 0.7198654413223267,
      "step": 812
    },
    {
      "epoch": 0.9789283564118001,
      "grad_norm": 0.14123551547527313,
      "learning_rate": 5.60572383291046e-05,
      "loss": 0.7763234972953796,
      "step": 813
    },
    {
      "epoch": 0.9801324503311258,
      "grad_norm": 0.1352493166923523,
      "learning_rate": 5.595841604955601e-05,
      "loss": 0.5468838810920715,
      "step": 814
    },
    {
      "epoch": 0.9813365442504516,
      "grad_norm": 0.1380176842212677,
      "learning_rate": 5.585957015345882e-05,
      "loss": 0.7065597772598267,
      "step": 815
    },
    {
      "epoch": 0.9825406381697772,
      "grad_norm": 0.24529443681240082,
      "learning_rate": 5.5760701032594775e-05,
      "loss": 0.707173228263855,
      "step": 816
    },
    {
      "epoch": 0.983744732089103,
      "grad_norm": 0.1512458771467209,
      "learning_rate": 5.566180907883777e-05,
      "loss": 0.7358670234680176,
      "step": 817
    },
    {
      "epoch": 0.9849488260084287,
      "grad_norm": 0.14294596016407013,
      "learning_rate": 5.556289468415212e-05,
      "loss": 0.787729024887085,
      "step": 818
    },
    {
      "epoch": 0.9861529199277543,
      "grad_norm": 0.13133664429187775,
      "learning_rate": 5.546395824059113e-05,
      "loss": 0.7464638948440552,
      "step": 819
    },
    {
      "epoch": 0.9873570138470801,
      "grad_norm": 0.14197419583797455,
      "learning_rate": 5.536500014029547e-05,
      "loss": 0.6847254633903503,
      "step": 820
    },
    {
      "epoch": 0.9885611077664058,
      "grad_norm": 0.14456631243228912,
      "learning_rate": 5.5266020775491654e-05,
      "loss": 0.9427334070205688,
      "step": 821
    },
    {
      "epoch": 0.9897652016857315,
      "grad_norm": 0.13878577947616577,
      "learning_rate": 5.5167020538490514e-05,
      "loss": 0.8490236401557922,
      "step": 822
    },
    {
      "epoch": 0.9909692956050572,
      "grad_norm": 0.13749754428863525,
      "learning_rate": 5.506799982168553e-05,
      "loss": 0.7198784351348877,
      "step": 823
    },
    {
      "epoch": 0.9921733895243829,
      "grad_norm": 0.13495425879955292,
      "learning_rate": 5.496895901755145e-05,
      "loss": 0.7533031702041626,
      "step": 824
    },
    {
      "epoch": 0.9933774834437086,
      "grad_norm": 0.12389883399009705,
      "learning_rate": 5.486989851864258e-05,
      "loss": 0.6607186794281006,
      "step": 825
    },
    {
      "epoch": 0.9945815773630343,
      "grad_norm": 0.13767609000205994,
      "learning_rate": 5.47708187175913e-05,
      "loss": 0.6844433546066284,
      "step": 826
    },
    {
      "epoch": 0.9957856712823601,
      "grad_norm": 0.12802822887897491,
      "learning_rate": 5.4671720007106507e-05,
      "loss": 0.7333580255508423,
      "step": 827
    },
    {
      "epoch": 0.9969897652016857,
      "grad_norm": 0.14050599932670593,
      "learning_rate": 5.4572602779972006e-05,
      "loss": 0.6959406733512878,
      "step": 828
    },
    {
      "epoch": 0.9981938591210114,
      "grad_norm": 0.1342303454875946,
      "learning_rate": 5.447346742904508e-05,
      "loss": 0.7508161664009094,
      "step": 829
    },
    {
      "epoch": 0.9993979530403372,
      "grad_norm": 0.12283643335103989,
      "learning_rate": 5.437431434725473e-05,
      "loss": 0.8068416714668274,
      "step": 830
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.17149929702281952,
      "learning_rate": 5.427514392760034e-05,
      "loss": 0.7839486002922058,
      "step": 831
    },
    {
      "epoch": 1.0012040939193256,
      "grad_norm": 0.1212155669927597,
      "learning_rate": 5.417595656314997e-05,
      "loss": 0.6534155011177063,
      "step": 832
    },
    {
      "epoch": 1.0024081878386515,
      "grad_norm": 0.11349339783191681,
      "learning_rate": 5.40767526470388e-05,
      "loss": 0.6328350305557251,
      "step": 833
    },
    {
      "epoch": 1.0036122817579771,
      "grad_norm": 0.12745831906795502,
      "learning_rate": 5.39775325724677e-05,
      "loss": 0.7121830582618713,
      "step": 834
    },
    {
      "epoch": 1.0048163756773028,
      "grad_norm": 0.13753186166286469,
      "learning_rate": 5.3878296732701515e-05,
      "loss": 0.5889167785644531,
      "step": 835
    },
    {
      "epoch": 1.0060204695966286,
      "grad_norm": 0.12029321491718292,
      "learning_rate": 5.377904552106763e-05,
      "loss": 0.6463069915771484,
      "step": 836
    },
    {
      "epoch": 1.0072245635159542,
      "grad_norm": 0.12784896790981293,
      "learning_rate": 5.367977933095428e-05,
      "loss": 0.6505364179611206,
      "step": 837
    },
    {
      "epoch": 1.0084286574352799,
      "grad_norm": 0.16782256960868835,
      "learning_rate": 5.3580498555809163e-05,
      "loss": 0.6237580180168152,
      "step": 838
    },
    {
      "epoch": 1.0096327513546057,
      "grad_norm": 0.13288913667201996,
      "learning_rate": 5.348120358913773e-05,
      "loss": 0.5774434208869934,
      "step": 839
    },
    {
      "epoch": 1.0108368452739314,
      "grad_norm": 0.2107757329940796,
      "learning_rate": 5.338189482450167e-05,
      "loss": 0.6808729767799377,
      "step": 840
    },
    {
      "epoch": 1.012040939193257,
      "grad_norm": 0.14137078821659088,
      "learning_rate": 5.3282572655517416e-05,
      "loss": 0.6233557462692261,
      "step": 841
    },
    {
      "epoch": 1.0132450331125828,
      "grad_norm": 0.14869503676891327,
      "learning_rate": 5.318323747585444e-05,
      "loss": 0.6059342622756958,
      "step": 842
    },
    {
      "epoch": 1.0144491270319085,
      "grad_norm": 0.14882372319698334,
      "learning_rate": 5.308388967923391e-05,
      "loss": 0.5985732078552246,
      "step": 843
    },
    {
      "epoch": 1.0156532209512341,
      "grad_norm": 0.1362399160861969,
      "learning_rate": 5.298452965942687e-05,
      "loss": 0.6568828821182251,
      "step": 844
    },
    {
      "epoch": 1.01685731487056,
      "grad_norm": 0.13402271270751953,
      "learning_rate": 5.2885157810252915e-05,
      "loss": 0.548881471157074,
      "step": 845
    },
    {
      "epoch": 1.0180614087898856,
      "grad_norm": 0.1451653391122818,
      "learning_rate": 5.278577452557845e-05,
      "loss": 0.5865733623504639,
      "step": 846
    },
    {
      "epoch": 1.0192655027092112,
      "grad_norm": 0.13090397417545319,
      "learning_rate": 5.2686380199315244e-05,
      "loss": 0.5792367458343506,
      "step": 847
    },
    {
      "epoch": 1.020469596628537,
      "grad_norm": 0.1603151559829712,
      "learning_rate": 5.2586975225418854e-05,
      "loss": 0.6294496059417725,
      "step": 848
    },
    {
      "epoch": 1.0216736905478627,
      "grad_norm": 0.14944049715995789,
      "learning_rate": 5.248755999788699e-05,
      "loss": 0.6562313437461853,
      "step": 849
    },
    {
      "epoch": 1.0228777844671884,
      "grad_norm": 0.13818524777889252,
      "learning_rate": 5.2388134910758015e-05,
      "loss": 0.6064617037773132,
      "step": 850
    },
    {
      "epoch": 1.0240818783865142,
      "grad_norm": 0.14991815388202667,
      "learning_rate": 5.2288700358109375e-05,
      "loss": 0.6686738133430481,
      "step": 851
    },
    {
      "epoch": 1.0252859723058398,
      "grad_norm": 0.13863402605056763,
      "learning_rate": 5.218925673405607e-05,
      "loss": 0.648642897605896,
      "step": 852
    },
    {
      "epoch": 1.0264900662251655,
      "grad_norm": 0.1415686011314392,
      "learning_rate": 5.208980443274899e-05,
      "loss": 0.6485585570335388,
      "step": 853
    },
    {
      "epoch": 1.0276941601444913,
      "grad_norm": 0.15413688123226166,
      "learning_rate": 5.199034384837345e-05,
      "loss": 0.6191515326499939,
      "step": 854
    },
    {
      "epoch": 1.028898254063817,
      "grad_norm": 0.13257445394992828,
      "learning_rate": 5.189087537514763e-05,
      "loss": 0.5325754880905151,
      "step": 855
    },
    {
      "epoch": 1.0301023479831426,
      "grad_norm": 0.14374899864196777,
      "learning_rate": 5.179139940732091e-05,
      "loss": 0.5836261510848999,
      "step": 856
    },
    {
      "epoch": 1.0313064419024685,
      "grad_norm": 0.16513516008853912,
      "learning_rate": 5.169191633917242e-05,
      "loss": 0.6007570028305054,
      "step": 857
    },
    {
      "epoch": 1.032510535821794,
      "grad_norm": 0.14810235798358917,
      "learning_rate": 5.159242656500942e-05,
      "loss": 0.611772358417511,
      "step": 858
    },
    {
      "epoch": 1.0337146297411197,
      "grad_norm": 0.15233954787254333,
      "learning_rate": 5.149293047916576e-05,
      "loss": 0.5861673355102539,
      "step": 859
    },
    {
      "epoch": 1.0349187236604456,
      "grad_norm": 0.16954539716243744,
      "learning_rate": 5.139342847600028e-05,
      "loss": 0.6227894425392151,
      "step": 860
    },
    {
      "epoch": 1.0361228175797712,
      "grad_norm": 0.14000533521175385,
      "learning_rate": 5.12939209498953e-05,
      "loss": 0.5907000303268433,
      "step": 861
    },
    {
      "epoch": 1.0373269114990968,
      "grad_norm": 0.14107802510261536,
      "learning_rate": 5.119440829525504e-05,
      "loss": 0.5909730792045593,
      "step": 862
    },
    {
      "epoch": 1.0385310054184227,
      "grad_norm": 0.14568990468978882,
      "learning_rate": 5.1094890906504e-05,
      "loss": 0.651166558265686,
      "step": 863
    },
    {
      "epoch": 1.0397350993377483,
      "grad_norm": 0.1641809046268463,
      "learning_rate": 5.0995369178085484e-05,
      "loss": 0.6219112277030945,
      "step": 864
    },
    {
      "epoch": 1.040939193257074,
      "grad_norm": 0.14424622058868408,
      "learning_rate": 5.0895843504460005e-05,
      "loss": 0.5498909950256348,
      "step": 865
    },
    {
      "epoch": 1.0421432871763998,
      "grad_norm": 0.14483071863651276,
      "learning_rate": 5.0796314280103664e-05,
      "loss": 0.600116491317749,
      "step": 866
    },
    {
      "epoch": 1.0433473810957254,
      "grad_norm": 0.17019861936569214,
      "learning_rate": 5.0696781899506686e-05,
      "loss": 0.6473695039749146,
      "step": 867
    },
    {
      "epoch": 1.044551475015051,
      "grad_norm": 0.15227004885673523,
      "learning_rate": 5.059724675717177e-05,
      "loss": 0.6332957744598389,
      "step": 868
    },
    {
      "epoch": 1.045755568934377,
      "grad_norm": 0.15956717729568481,
      "learning_rate": 5.049770924761259e-05,
      "loss": 0.6524595022201538,
      "step": 869
    },
    {
      "epoch": 1.0469596628537026,
      "grad_norm": 0.14208130538463593,
      "learning_rate": 5.039816976535219e-05,
      "loss": 0.6076785326004028,
      "step": 870
    },
    {
      "epoch": 1.0481637567730282,
      "grad_norm": 0.14891913533210754,
      "learning_rate": 5.029862870492142e-05,
      "loss": 0.6748784780502319,
      "step": 871
    },
    {
      "epoch": 1.049367850692354,
      "grad_norm": 0.15910445153713226,
      "learning_rate": 5.0199086460857406e-05,
      "loss": 0.6313496828079224,
      "step": 872
    },
    {
      "epoch": 1.0505719446116797,
      "grad_norm": 0.1477075070142746,
      "learning_rate": 5.0099543427701956e-05,
      "loss": 0.660576581954956,
      "step": 873
    },
    {
      "epoch": 1.0517760385310053,
      "grad_norm": 0.15786121785640717,
      "learning_rate": 5e-05,
      "loss": 0.5102140307426453,
      "step": 874
    },
    {
      "epoch": 1.0529801324503312,
      "grad_norm": 0.1523813009262085,
      "learning_rate": 4.9900456572298055e-05,
      "loss": 0.549045979976654,
      "step": 875
    },
    {
      "epoch": 1.0541842263696568,
      "grad_norm": 0.15490423142910004,
      "learning_rate": 4.980091353914259e-05,
      "loss": 0.6267224550247192,
      "step": 876
    },
    {
      "epoch": 1.0553883202889824,
      "grad_norm": 0.16186775267124176,
      "learning_rate": 4.9701371295078603e-05,
      "loss": 0.5967670679092407,
      "step": 877
    },
    {
      "epoch": 1.0565924142083083,
      "grad_norm": 0.15865488350391388,
      "learning_rate": 4.9601830234647824e-05,
      "loss": 0.610004186630249,
      "step": 878
    },
    {
      "epoch": 1.057796508127634,
      "grad_norm": 0.16221114993095398,
      "learning_rate": 4.950229075238742e-05,
      "loss": 0.6145402789115906,
      "step": 879
    },
    {
      "epoch": 1.0590006020469596,
      "grad_norm": 0.15233594179153442,
      "learning_rate": 4.940275324282824e-05,
      "loss": 0.5986229181289673,
      "step": 880
    },
    {
      "epoch": 1.0602046959662854,
      "grad_norm": 0.14293724298477173,
      "learning_rate": 4.930321810049334e-05,
      "loss": 0.5472497940063477,
      "step": 881
    },
    {
      "epoch": 1.061408789885611,
      "grad_norm": 0.148305743932724,
      "learning_rate": 4.920368571989636e-05,
      "loss": 0.5126625895500183,
      "step": 882
    },
    {
      "epoch": 1.062612883804937,
      "grad_norm": 0.16439023613929749,
      "learning_rate": 4.910415649554001e-05,
      "loss": 0.6607561111450195,
      "step": 883
    },
    {
      "epoch": 1.0638169777242625,
      "grad_norm": 0.1524340808391571,
      "learning_rate": 4.900463082191452e-05,
      "loss": 0.5643805265426636,
      "step": 884
    },
    {
      "epoch": 1.0650210716435882,
      "grad_norm": 0.15595945715904236,
      "learning_rate": 4.890510909349602e-05,
      "loss": 0.6397435665130615,
      "step": 885
    },
    {
      "epoch": 1.0662251655629138,
      "grad_norm": 0.16699078679084778,
      "learning_rate": 4.880559170474499e-05,
      "loss": 0.585167646408081,
      "step": 886
    },
    {
      "epoch": 1.0674292594822397,
      "grad_norm": 0.1592654436826706,
      "learning_rate": 4.870607905010471e-05,
      "loss": 0.662108302116394,
      "step": 887
    },
    {
      "epoch": 1.0686333534015653,
      "grad_norm": 0.1400386393070221,
      "learning_rate": 4.860657152399973e-05,
      "loss": 0.5310679078102112,
      "step": 888
    },
    {
      "epoch": 1.0698374473208911,
      "grad_norm": 0.17089571058750153,
      "learning_rate": 4.850706952083426e-05,
      "loss": 0.646551251411438,
      "step": 889
    },
    {
      "epoch": 1.0710415412402168,
      "grad_norm": 0.1379624754190445,
      "learning_rate": 4.840757343499059e-05,
      "loss": 0.6057286858558655,
      "step": 890
    },
    {
      "epoch": 1.0722456351595424,
      "grad_norm": 0.14356563985347748,
      "learning_rate": 4.83080836608276e-05,
      "loss": 0.6515355706214905,
      "step": 891
    },
    {
      "epoch": 1.073449729078868,
      "grad_norm": 0.16751691699028015,
      "learning_rate": 4.82086005926791e-05,
      "loss": 0.5371620059013367,
      "step": 892
    },
    {
      "epoch": 1.074653822998194,
      "grad_norm": 0.1702519655227661,
      "learning_rate": 4.8109124624852386e-05,
      "loss": 0.6395319104194641,
      "step": 893
    },
    {
      "epoch": 1.0758579169175195,
      "grad_norm": 0.14478567242622375,
      "learning_rate": 4.800965615162655e-05,
      "loss": 0.5439417958259583,
      "step": 894
    },
    {
      "epoch": 1.0770620108368454,
      "grad_norm": 0.14434917271137238,
      "learning_rate": 4.791019556725104e-05,
      "loss": 0.6112794876098633,
      "step": 895
    },
    {
      "epoch": 1.078266104756171,
      "grad_norm": 0.14645694196224213,
      "learning_rate": 4.7810743265943955e-05,
      "loss": 0.5768594145774841,
      "step": 896
    },
    {
      "epoch": 1.0794701986754967,
      "grad_norm": 0.1523243635892868,
      "learning_rate": 4.771129964189063e-05,
      "loss": 0.6814373731613159,
      "step": 897
    },
    {
      "epoch": 1.0806742925948223,
      "grad_norm": 0.16345615684986115,
      "learning_rate": 4.7611865089242004e-05,
      "loss": 0.6606166362762451,
      "step": 898
    },
    {
      "epoch": 1.0818783865141481,
      "grad_norm": 0.14699672162532806,
      "learning_rate": 4.751244000211302e-05,
      "loss": 0.6406362652778625,
      "step": 899
    },
    {
      "epoch": 1.0830824804334738,
      "grad_norm": 0.13665667176246643,
      "learning_rate": 4.741302477458116e-05,
      "loss": 0.4699036777019501,
      "step": 900
    },
    {
      "epoch": 1.0842865743527996,
      "grad_norm": 0.1709156632423401,
      "learning_rate": 4.731361980068476e-05,
      "loss": 0.5594170093536377,
      "step": 901
    },
    {
      "epoch": 1.0854906682721253,
      "grad_norm": 0.1679125875234604,
      "learning_rate": 4.7214225474421556e-05,
      "loss": 0.5294967889785767,
      "step": 902
    },
    {
      "epoch": 1.086694762191451,
      "grad_norm": 0.14729350805282593,
      "learning_rate": 4.7114842189747096e-05,
      "loss": 0.6543394327163696,
      "step": 903
    },
    {
      "epoch": 1.0878988561107765,
      "grad_norm": 0.1523282527923584,
      "learning_rate": 4.701547034057313e-05,
      "loss": 0.5641987323760986,
      "step": 904
    },
    {
      "epoch": 1.0891029500301024,
      "grad_norm": 0.15066470205783844,
      "learning_rate": 4.691611032076611e-05,
      "loss": 0.5678690075874329,
      "step": 905
    },
    {
      "epoch": 1.090307043949428,
      "grad_norm": 0.15151198208332062,
      "learning_rate": 4.6816762524145565e-05,
      "loss": 0.6115108132362366,
      "step": 906
    },
    {
      "epoch": 1.0915111378687539,
      "grad_norm": 0.16325950622558594,
      "learning_rate": 4.67174273444826e-05,
      "loss": 0.6259532570838928,
      "step": 907
    },
    {
      "epoch": 1.0927152317880795,
      "grad_norm": 0.16196483373641968,
      "learning_rate": 4.6618105175498336e-05,
      "loss": 0.5948746800422668,
      "step": 908
    },
    {
      "epoch": 1.0939193257074051,
      "grad_norm": 0.15428294241428375,
      "learning_rate": 4.6518796410862284e-05,
      "loss": 0.536793053150177,
      "step": 909
    },
    {
      "epoch": 1.095123419626731,
      "grad_norm": 0.1517171710729599,
      "learning_rate": 4.641950144419085e-05,
      "loss": 0.6297614574432373,
      "step": 910
    },
    {
      "epoch": 1.0963275135460566,
      "grad_norm": 0.157099649310112,
      "learning_rate": 4.632022066904573e-05,
      "loss": 0.6112175583839417,
      "step": 911
    },
    {
      "epoch": 1.0975316074653823,
      "grad_norm": 0.15802069008350372,
      "learning_rate": 4.622095447893238e-05,
      "loss": 0.6322358250617981,
      "step": 912
    },
    {
      "epoch": 1.098735701384708,
      "grad_norm": 0.14938995242118835,
      "learning_rate": 4.612170326729849e-05,
      "loss": 0.5891709923744202,
      "step": 913
    },
    {
      "epoch": 1.0999397953040337,
      "grad_norm": 0.1618727594614029,
      "learning_rate": 4.6022467427532316e-05,
      "loss": 0.6967465281486511,
      "step": 914
    },
    {
      "epoch": 1.1011438892233594,
      "grad_norm": 0.14040106534957886,
      "learning_rate": 4.592324735296122e-05,
      "loss": 0.5867524147033691,
      "step": 915
    },
    {
      "epoch": 1.1023479831426852,
      "grad_norm": 0.1565229445695877,
      "learning_rate": 4.582404343685005e-05,
      "loss": 0.6546963453292847,
      "step": 916
    },
    {
      "epoch": 1.1035520770620109,
      "grad_norm": 0.15965211391448975,
      "learning_rate": 4.5724856072399666e-05,
      "loss": 0.6052303314208984,
      "step": 917
    },
    {
      "epoch": 1.1047561709813365,
      "grad_norm": 0.15058669447898865,
      "learning_rate": 4.562568565274528e-05,
      "loss": 0.548164963722229,
      "step": 918
    },
    {
      "epoch": 1.1059602649006623,
      "grad_norm": 0.1625698208808899,
      "learning_rate": 4.552653257095495e-05,
      "loss": 0.5786932706832886,
      "step": 919
    },
    {
      "epoch": 1.107164358819988,
      "grad_norm": 0.15275244414806366,
      "learning_rate": 4.5427397220028006e-05,
      "loss": 0.5595954656600952,
      "step": 920
    },
    {
      "epoch": 1.1083684527393136,
      "grad_norm": 0.15977346897125244,
      "learning_rate": 4.5328279992893505e-05,
      "loss": 0.6239919662475586,
      "step": 921
    },
    {
      "epoch": 1.1095725466586395,
      "grad_norm": 0.14655499160289764,
      "learning_rate": 4.5229181282408705e-05,
      "loss": 0.5107656717300415,
      "step": 922
    },
    {
      "epoch": 1.110776640577965,
      "grad_norm": 0.17171819508075714,
      "learning_rate": 4.513010148135743e-05,
      "loss": 0.7030598521232605,
      "step": 923
    },
    {
      "epoch": 1.1119807344972907,
      "grad_norm": 0.1356562376022339,
      "learning_rate": 4.5031040982448564e-05,
      "loss": 0.5616992115974426,
      "step": 924
    },
    {
      "epoch": 1.1131848284166166,
      "grad_norm": 0.15307123959064484,
      "learning_rate": 4.493200017831448e-05,
      "loss": 0.5258873105049133,
      "step": 925
    },
    {
      "epoch": 1.1143889223359422,
      "grad_norm": 0.16134271025657654,
      "learning_rate": 4.48329794615095e-05,
      "loss": 0.6148583292961121,
      "step": 926
    },
    {
      "epoch": 1.1155930162552679,
      "grad_norm": 0.15891937911510468,
      "learning_rate": 4.4733979224508344e-05,
      "loss": 0.624730110168457,
      "step": 927
    },
    {
      "epoch": 1.1167971101745937,
      "grad_norm": 0.16219955682754517,
      "learning_rate": 4.4634999859704546e-05,
      "loss": 0.605157732963562,
      "step": 928
    },
    {
      "epoch": 1.1180012040939193,
      "grad_norm": 0.15462738275527954,
      "learning_rate": 4.453604175940888e-05,
      "loss": 0.5247471332550049,
      "step": 929
    },
    {
      "epoch": 1.119205298013245,
      "grad_norm": 0.1656009703874588,
      "learning_rate": 4.443710531584789e-05,
      "loss": 0.6055982112884521,
      "step": 930
    },
    {
      "epoch": 1.1204093919325708,
      "grad_norm": 0.17724959552288055,
      "learning_rate": 4.433819092116223e-05,
      "loss": 0.7452636957168579,
      "step": 931
    },
    {
      "epoch": 1.1216134858518965,
      "grad_norm": 0.1799347698688507,
      "learning_rate": 4.423929896740522e-05,
      "loss": 0.6251393556594849,
      "step": 932
    },
    {
      "epoch": 1.122817579771222,
      "grad_norm": 0.17604993283748627,
      "learning_rate": 4.41404298465412e-05,
      "loss": 0.5560930967330933,
      "step": 933
    },
    {
      "epoch": 1.124021673690548,
      "grad_norm": 0.1689262092113495,
      "learning_rate": 4.4041583950444004e-05,
      "loss": 0.5756097435951233,
      "step": 934
    },
    {
      "epoch": 1.1252257676098736,
      "grad_norm": 0.16565203666687012,
      "learning_rate": 4.3942761670895416e-05,
      "loss": 0.548450231552124,
      "step": 935
    },
    {
      "epoch": 1.1264298615291992,
      "grad_norm": 0.15794895589351654,
      "learning_rate": 4.384396339958361e-05,
      "loss": 0.616039514541626,
      "step": 936
    },
    {
      "epoch": 1.127633955448525,
      "grad_norm": 0.1657445877790451,
      "learning_rate": 4.374518952810167e-05,
      "loss": 0.7053273916244507,
      "step": 937
    },
    {
      "epoch": 1.1288380493678507,
      "grad_norm": 0.1536550521850586,
      "learning_rate": 4.364644044794585e-05,
      "loss": 0.562394917011261,
      "step": 938
    },
    {
      "epoch": 1.1300421432871763,
      "grad_norm": 0.15293379127979279,
      "learning_rate": 4.354771655051424e-05,
      "loss": 0.5577017068862915,
      "step": 939
    },
    {
      "epoch": 1.1312462372065022,
      "grad_norm": 0.17534448206424713,
      "learning_rate": 4.344901822710506e-05,
      "loss": 0.6501519680023193,
      "step": 940
    },
    {
      "epoch": 1.1324503311258278,
      "grad_norm": 0.15314526855945587,
      "learning_rate": 4.33503458689152e-05,
      "loss": 0.6502077579498291,
      "step": 941
    },
    {
      "epoch": 1.1336544250451535,
      "grad_norm": 0.1568373590707779,
      "learning_rate": 4.3251699867038634e-05,
      "loss": 0.5606850385665894,
      "step": 942
    },
    {
      "epoch": 1.1348585189644793,
      "grad_norm": 0.17886261641979218,
      "learning_rate": 4.3153080612464835e-05,
      "loss": 0.6628350019454956,
      "step": 943
    },
    {
      "epoch": 1.136062612883805,
      "grad_norm": 0.1440589278936386,
      "learning_rate": 4.305448849607731e-05,
      "loss": 0.5645763874053955,
      "step": 944
    },
    {
      "epoch": 1.1372667068031306,
      "grad_norm": 0.14748460054397583,
      "learning_rate": 4.295592390865194e-05,
      "loss": 0.6695147752761841,
      "step": 945
    },
    {
      "epoch": 1.1384708007224564,
      "grad_norm": 0.15702413022518158,
      "learning_rate": 4.28573872408556e-05,
      "loss": 0.6074323058128357,
      "step": 946
    },
    {
      "epoch": 1.139674894641782,
      "grad_norm": 0.16561400890350342,
      "learning_rate": 4.2758878883244366e-05,
      "loss": 0.6456789970397949,
      "step": 947
    },
    {
      "epoch": 1.1408789885611077,
      "grad_norm": 0.18494848906993866,
      "learning_rate": 4.2660399226262205e-05,
      "loss": 0.6043604612350464,
      "step": 948
    },
    {
      "epoch": 1.1420830824804336,
      "grad_norm": 0.1635720580816269,
      "learning_rate": 4.256194866023929e-05,
      "loss": 0.6773695349693298,
      "step": 949
    },
    {
      "epoch": 1.1432871763997592,
      "grad_norm": 0.1827755570411682,
      "learning_rate": 4.24635275753905e-05,
      "loss": 0.5373539328575134,
      "step": 950
    },
    {
      "epoch": 1.1444912703190848,
      "grad_norm": 0.15528537333011627,
      "learning_rate": 4.236513636181389e-05,
      "loss": 0.599889874458313,
      "step": 951
    },
    {
      "epoch": 1.1456953642384107,
      "grad_norm": 0.1533524990081787,
      "learning_rate": 4.2266775409489023e-05,
      "loss": 0.573198676109314,
      "step": 952
    },
    {
      "epoch": 1.1468994581577363,
      "grad_norm": 0.14608602225780487,
      "learning_rate": 4.2168445108275636e-05,
      "loss": 0.6073895692825317,
      "step": 953
    },
    {
      "epoch": 1.148103552077062,
      "grad_norm": 0.18135611712932587,
      "learning_rate": 4.2070145847911915e-05,
      "loss": 0.6508622765541077,
      "step": 954
    },
    {
      "epoch": 1.1493076459963878,
      "grad_norm": 0.15927289426326752,
      "learning_rate": 4.197187801801301e-05,
      "loss": 0.6691917181015015,
      "step": 955
    },
    {
      "epoch": 1.1505117399157134,
      "grad_norm": 0.197689488530159,
      "learning_rate": 4.187364200806956e-05,
      "loss": 0.7441767454147339,
      "step": 956
    },
    {
      "epoch": 1.151715833835039,
      "grad_norm": 0.1447354555130005,
      "learning_rate": 4.177543820744597e-05,
      "loss": 0.593164324760437,
      "step": 957
    },
    {
      "epoch": 1.152919927754365,
      "grad_norm": 0.16338500380516052,
      "learning_rate": 4.167726700537909e-05,
      "loss": 0.6521214246749878,
      "step": 958
    },
    {
      "epoch": 1.1541240216736905,
      "grad_norm": 0.14394867420196533,
      "learning_rate": 4.157912879097651e-05,
      "loss": 0.6071224808692932,
      "step": 959
    },
    {
      "epoch": 1.1553281155930162,
      "grad_norm": 0.16275353729724884,
      "learning_rate": 4.1481023953215074e-05,
      "loss": 0.6252530813217163,
      "step": 960
    },
    {
      "epoch": 1.156532209512342,
      "grad_norm": 0.16491104662418365,
      "learning_rate": 4.1382952880939346e-05,
      "loss": 0.6386094093322754,
      "step": 961
    },
    {
      "epoch": 1.1577363034316677,
      "grad_norm": 0.15932697057724,
      "learning_rate": 4.128491596286004e-05,
      "loss": 0.6520111560821533,
      "step": 962
    },
    {
      "epoch": 1.1589403973509933,
      "grad_norm": 0.16653047502040863,
      "learning_rate": 4.118691358755254e-05,
      "loss": 0.6260276436805725,
      "step": 963
    },
    {
      "epoch": 1.1601444912703192,
      "grad_norm": 0.14379431307315826,
      "learning_rate": 4.108894614345527e-05,
      "loss": 0.5896302461624146,
      "step": 964
    },
    {
      "epoch": 1.1613485851896448,
      "grad_norm": 0.1580624133348465,
      "learning_rate": 4.0991014018868255e-05,
      "loss": 0.5671614408493042,
      "step": 965
    },
    {
      "epoch": 1.1625526791089704,
      "grad_norm": 0.15322285890579224,
      "learning_rate": 4.0893117601951434e-05,
      "loss": 0.6120231747627258,
      "step": 966
    },
    {
      "epoch": 1.1637567730282963,
      "grad_norm": 0.14445200562477112,
      "learning_rate": 4.079525728072334e-05,
      "loss": 0.6091594696044922,
      "step": 967
    },
    {
      "epoch": 1.164960866947622,
      "grad_norm": 0.1552407145500183,
      "learning_rate": 4.0697433443059344e-05,
      "loss": 0.5857088565826416,
      "step": 968
    },
    {
      "epoch": 1.1661649608669475,
      "grad_norm": 0.16211766004562378,
      "learning_rate": 4.059964647669025e-05,
      "loss": 0.6573500037193298,
      "step": 969
    },
    {
      "epoch": 1.1673690547862734,
      "grad_norm": 0.17158940434455872,
      "learning_rate": 4.050189676920075e-05,
      "loss": 0.6244156956672668,
      "step": 970
    },
    {
      "epoch": 1.168573148705599,
      "grad_norm": 0.15571577847003937,
      "learning_rate": 4.0404184708027764e-05,
      "loss": 0.5679243803024292,
      "step": 971
    },
    {
      "epoch": 1.1697772426249247,
      "grad_norm": 0.15378053486347198,
      "learning_rate": 4.03065106804591e-05,
      "loss": 0.5513820052146912,
      "step": 972
    },
    {
      "epoch": 1.1709813365442505,
      "grad_norm": 0.16843657195568085,
      "learning_rate": 4.0208875073631767e-05,
      "loss": 0.6658234596252441,
      "step": 973
    },
    {
      "epoch": 1.1721854304635762,
      "grad_norm": 0.16118177771568298,
      "learning_rate": 4.01112782745305e-05,
      "loss": 0.5040348172187805,
      "step": 974
    },
    {
      "epoch": 1.1733895243829018,
      "grad_norm": 0.1389022022485733,
      "learning_rate": 4.001372066998621e-05,
      "loss": 0.6135053634643555,
      "step": 975
    },
    {
      "epoch": 1.1745936183022276,
      "grad_norm": 0.17904014885425568,
      "learning_rate": 3.9916202646674454e-05,
      "loss": 0.615231454372406,
      "step": 976
    },
    {
      "epoch": 1.1757977122215533,
      "grad_norm": 0.16800546646118164,
      "learning_rate": 3.981872459111394e-05,
      "loss": 0.49969616532325745,
      "step": 977
    },
    {
      "epoch": 1.177001806140879,
      "grad_norm": 0.17903780937194824,
      "learning_rate": 3.9721286889664946e-05,
      "loss": 0.5987198352813721,
      "step": 978
    },
    {
      "epoch": 1.1782059000602048,
      "grad_norm": 0.14727278053760529,
      "learning_rate": 3.962388992852778e-05,
      "loss": 0.44832926988601685,
      "step": 979
    },
    {
      "epoch": 1.1794099939795304,
      "grad_norm": 0.1618252694606781,
      "learning_rate": 3.9526534093741294e-05,
      "loss": 0.6353049874305725,
      "step": 980
    },
    {
      "epoch": 1.180614087898856,
      "grad_norm": 0.15937864780426025,
      "learning_rate": 3.94292197711813e-05,
      "loss": 0.5794386863708496,
      "step": 981
    },
    {
      "epoch": 1.1818181818181819,
      "grad_norm": 0.18125776946544647,
      "learning_rate": 3.933194734655916e-05,
      "loss": 0.6997560262680054,
      "step": 982
    },
    {
      "epoch": 1.1830222757375075,
      "grad_norm": 0.16208913922309875,
      "learning_rate": 3.923471720542007e-05,
      "loss": 0.5903565287590027,
      "step": 983
    },
    {
      "epoch": 1.1842263696568331,
      "grad_norm": 0.15385866165161133,
      "learning_rate": 3.9137529733141705e-05,
      "loss": 0.5572262406349182,
      "step": 984
    },
    {
      "epoch": 1.185430463576159,
      "grad_norm": 0.16204695403575897,
      "learning_rate": 3.904038531493257e-05,
      "loss": 0.6065976619720459,
      "step": 985
    },
    {
      "epoch": 1.1866345574954846,
      "grad_norm": 0.1647799164056778,
      "learning_rate": 3.894328433583053e-05,
      "loss": 0.6103823781013489,
      "step": 986
    },
    {
      "epoch": 1.1878386514148103,
      "grad_norm": 0.15922331809997559,
      "learning_rate": 3.8846227180701335e-05,
      "loss": 0.5625864267349243,
      "step": 987
    },
    {
      "epoch": 1.1890427453341361,
      "grad_norm": 0.15828777849674225,
      "learning_rate": 3.874921423423697e-05,
      "loss": 0.5728390216827393,
      "step": 988
    },
    {
      "epoch": 1.1902468392534618,
      "grad_norm": 0.1705407053232193,
      "learning_rate": 3.86522458809542e-05,
      "loss": 0.6246944069862366,
      "step": 989
    },
    {
      "epoch": 1.1914509331727874,
      "grad_norm": 0.15679261088371277,
      "learning_rate": 3.855532250519307e-05,
      "loss": 0.5869783759117126,
      "step": 990
    },
    {
      "epoch": 1.1926550270921132,
      "grad_norm": 0.16110317409038544,
      "learning_rate": 3.845844449111535e-05,
      "loss": 0.5512909889221191,
      "step": 991
    },
    {
      "epoch": 1.1938591210114389,
      "grad_norm": 0.14724299311637878,
      "learning_rate": 3.8361612222703015e-05,
      "loss": 0.5556501150131226,
      "step": 992
    },
    {
      "epoch": 1.1950632149307645,
      "grad_norm": 0.14060966670513153,
      "learning_rate": 3.826482608375671e-05,
      "loss": 0.5251903533935547,
      "step": 993
    },
    {
      "epoch": 1.1962673088500904,
      "grad_norm": 0.17426644265651703,
      "learning_rate": 3.816808645789425e-05,
      "loss": 0.6150487065315247,
      "step": 994
    },
    {
      "epoch": 1.197471402769416,
      "grad_norm": 0.19017861783504486,
      "learning_rate": 3.8071393728549074e-05,
      "loss": 0.6459817290306091,
      "step": 995
    },
    {
      "epoch": 1.1986754966887416,
      "grad_norm": 0.16790057718753815,
      "learning_rate": 3.79747482789688e-05,
      "loss": 0.585645318031311,
      "step": 996
    },
    {
      "epoch": 1.1998795906080675,
      "grad_norm": 0.169486865401268,
      "learning_rate": 3.787815049221361e-05,
      "loss": 0.6063693761825562,
      "step": 997
    },
    {
      "epoch": 1.2010836845273931,
      "grad_norm": 0.17052797973155975,
      "learning_rate": 3.778160075115476e-05,
      "loss": 0.543434739112854,
      "step": 998
    },
    {
      "epoch": 1.2022877784467187,
      "grad_norm": 0.16562561690807343,
      "learning_rate": 3.768509943847309e-05,
      "loss": 0.590913712978363,
      "step": 999
    },
    {
      "epoch": 1.2034918723660446,
      "grad_norm": 0.1593073159456253,
      "learning_rate": 3.758864693665748e-05,
      "loss": 0.6726359128952026,
      "step": 1000
    },
    {
      "epoch": 1.2046959662853702,
      "grad_norm": 0.1681635081768036,
      "learning_rate": 3.749224362800338e-05,
      "loss": 0.667739987373352,
      "step": 1001
    },
    {
      "epoch": 1.205900060204696,
      "grad_norm": 0.1677180677652359,
      "learning_rate": 3.739588989461123e-05,
      "loss": 0.6548608541488647,
      "step": 1002
    },
    {
      "epoch": 1.2071041541240217,
      "grad_norm": 0.16399335861206055,
      "learning_rate": 3.729958611838496e-05,
      "loss": 0.5780633687973022,
      "step": 1003
    },
    {
      "epoch": 1.2083082480433474,
      "grad_norm": 0.1599387675523758,
      "learning_rate": 3.720333268103055e-05,
      "loss": 0.6349557042121887,
      "step": 1004
    },
    {
      "epoch": 1.209512341962673,
      "grad_norm": 0.15925081074237823,
      "learning_rate": 3.7107129964054366e-05,
      "loss": 0.5969941020011902,
      "step": 1005
    },
    {
      "epoch": 1.2107164358819988,
      "grad_norm": 0.154674232006073,
      "learning_rate": 3.701097834876185e-05,
      "loss": 0.6194593906402588,
      "step": 1006
    },
    {
      "epoch": 1.2119205298013245,
      "grad_norm": 0.16653570532798767,
      "learning_rate": 3.691487821625583e-05,
      "loss": 0.718635082244873,
      "step": 1007
    },
    {
      "epoch": 1.2131246237206503,
      "grad_norm": 0.1615322083234787,
      "learning_rate": 3.6818829947435076e-05,
      "loss": 0.6117541193962097,
      "step": 1008
    },
    {
      "epoch": 1.214328717639976,
      "grad_norm": 0.1630868911743164,
      "learning_rate": 3.672283392299282e-05,
      "loss": 0.595018208026886,
      "step": 1009
    },
    {
      "epoch": 1.2155328115593016,
      "grad_norm": 0.16575488448143005,
      "learning_rate": 3.66268905234152e-05,
      "loss": 0.6463031768798828,
      "step": 1010
    },
    {
      "epoch": 1.2167369054786272,
      "grad_norm": 0.1551463007926941,
      "learning_rate": 3.653100012897983e-05,
      "loss": 0.6781764626502991,
      "step": 1011
    },
    {
      "epoch": 1.217940999397953,
      "grad_norm": 0.1669500768184662,
      "learning_rate": 3.643516311975413e-05,
      "loss": 0.6090478897094727,
      "step": 1012
    },
    {
      "epoch": 1.2191450933172787,
      "grad_norm": 0.16930589079856873,
      "learning_rate": 3.633937987559402e-05,
      "loss": 0.6901058554649353,
      "step": 1013
    },
    {
      "epoch": 1.2203491872366046,
      "grad_norm": 0.17781247198581696,
      "learning_rate": 3.624365077614226e-05,
      "loss": 0.6221113204956055,
      "step": 1014
    },
    {
      "epoch": 1.2215532811559302,
      "grad_norm": 0.1621992439031601,
      "learning_rate": 3.614797620082703e-05,
      "loss": 0.6003121733665466,
      "step": 1015
    },
    {
      "epoch": 1.2227573750752558,
      "grad_norm": 0.14853785932064056,
      "learning_rate": 3.605235652886042e-05,
      "loss": 0.6390659809112549,
      "step": 1016
    },
    {
      "epoch": 1.2239614689945815,
      "grad_norm": 0.16178292036056519,
      "learning_rate": 3.595679213923685e-05,
      "loss": 0.5156100392341614,
      "step": 1017
    },
    {
      "epoch": 1.2251655629139073,
      "grad_norm": 0.16475273668766022,
      "learning_rate": 3.586128341073167e-05,
      "loss": 0.5032063126564026,
      "step": 1018
    },
    {
      "epoch": 1.226369656833233,
      "grad_norm": 0.15835539996623993,
      "learning_rate": 3.57658307218996e-05,
      "loss": 0.7338312268257141,
      "step": 1019
    },
    {
      "epoch": 1.2275737507525588,
      "grad_norm": 0.15580782294273376,
      "learning_rate": 3.567043445107326e-05,
      "loss": 0.6075632572174072,
      "step": 1020
    },
    {
      "epoch": 1.2287778446718844,
      "grad_norm": 0.17988945543766022,
      "learning_rate": 3.5575094976361625e-05,
      "loss": 0.5785654783248901,
      "step": 1021
    },
    {
      "epoch": 1.22998193859121,
      "grad_norm": 0.17855846881866455,
      "learning_rate": 3.5479812675648575e-05,
      "loss": 0.544749915599823,
      "step": 1022
    },
    {
      "epoch": 1.2311860325105357,
      "grad_norm": 0.16329751908779144,
      "learning_rate": 3.5384587926591355e-05,
      "loss": 0.6018604636192322,
      "step": 1023
    },
    {
      "epoch": 1.2323901264298616,
      "grad_norm": 0.18654924631118774,
      "learning_rate": 3.5289421106619126e-05,
      "loss": 0.6640027761459351,
      "step": 1024
    },
    {
      "epoch": 1.2335942203491872,
      "grad_norm": 0.15990373492240906,
      "learning_rate": 3.519431259293147e-05,
      "loss": 0.5555522441864014,
      "step": 1025
    },
    {
      "epoch": 1.234798314268513,
      "grad_norm": 0.15645863115787506,
      "learning_rate": 3.509926276249676e-05,
      "loss": 0.5377998948097229,
      "step": 1026
    },
    {
      "epoch": 1.2360024081878387,
      "grad_norm": 0.1740531176328659,
      "learning_rate": 3.500427199205091e-05,
      "loss": 0.670097827911377,
      "step": 1027
    },
    {
      "epoch": 1.2372065021071643,
      "grad_norm": 0.15555530786514282,
      "learning_rate": 3.4909340658095646e-05,
      "loss": 0.582030177116394,
      "step": 1028
    },
    {
      "epoch": 1.23841059602649,
      "grad_norm": 0.15726661682128906,
      "learning_rate": 3.4814469136897165e-05,
      "loss": 0.6469033360481262,
      "step": 1029
    },
    {
      "epoch": 1.2396146899458158,
      "grad_norm": 0.17427633702754974,
      "learning_rate": 3.471965780448461e-05,
      "loss": 0.5809903740882874,
      "step": 1030
    },
    {
      "epoch": 1.2408187838651414,
      "grad_norm": 0.15497168898582458,
      "learning_rate": 3.462490703664846e-05,
      "loss": 0.5504792332649231,
      "step": 1031
    },
    {
      "epoch": 1.2420228777844673,
      "grad_norm": 0.1575293093919754,
      "learning_rate": 3.4530217208939274e-05,
      "loss": 0.5221585035324097,
      "step": 1032
    },
    {
      "epoch": 1.243226971703793,
      "grad_norm": 0.19641974568367004,
      "learning_rate": 3.443558869666598e-05,
      "loss": 0.6650092601776123,
      "step": 1033
    },
    {
      "epoch": 1.2444310656231186,
      "grad_norm": 0.19074885547161102,
      "learning_rate": 3.434102187489451e-05,
      "loss": 0.6131666898727417,
      "step": 1034
    },
    {
      "epoch": 1.2456351595424442,
      "grad_norm": 0.1775917112827301,
      "learning_rate": 3.424651711844632e-05,
      "loss": 0.7188624143600464,
      "step": 1035
    },
    {
      "epoch": 1.24683925346177,
      "grad_norm": 0.1702939122915268,
      "learning_rate": 3.415207480189676e-05,
      "loss": 0.5875617265701294,
      "step": 1036
    },
    {
      "epoch": 1.2480433473810957,
      "grad_norm": 0.1662929505109787,
      "learning_rate": 3.405769529957381e-05,
      "loss": 0.6017499566078186,
      "step": 1037
    },
    {
      "epoch": 1.2492474413004215,
      "grad_norm": 0.18800735473632812,
      "learning_rate": 3.396337898555642e-05,
      "loss": 0.6093910932540894,
      "step": 1038
    },
    {
      "epoch": 1.2504515352197472,
      "grad_norm": 0.16465304791927338,
      "learning_rate": 3.386912623367311e-05,
      "loss": 0.566482424736023,
      "step": 1039
    },
    {
      "epoch": 1.2516556291390728,
      "grad_norm": 0.17017149925231934,
      "learning_rate": 3.377493741750044e-05,
      "loss": 0.5644276142120361,
      "step": 1040
    },
    {
      "epoch": 1.2528597230583984,
      "grad_norm": 0.17133264243602753,
      "learning_rate": 3.36808129103616e-05,
      "loss": 0.5747494697570801,
      "step": 1041
    },
    {
      "epoch": 1.2540638169777243,
      "grad_norm": 0.15970660746097565,
      "learning_rate": 3.358675308532486e-05,
      "loss": 0.6045863628387451,
      "step": 1042
    },
    {
      "epoch": 1.25526791089705,
      "grad_norm": 0.1526801735162735,
      "learning_rate": 3.349275831520212e-05,
      "loss": 0.6288525462150574,
      "step": 1043
    },
    {
      "epoch": 1.2564720048163758,
      "grad_norm": 0.16733554005622864,
      "learning_rate": 3.339882897254746e-05,
      "loss": 0.5904152989387512,
      "step": 1044
    },
    {
      "epoch": 1.2576760987357014,
      "grad_norm": 0.15646928548812866,
      "learning_rate": 3.330496542965556e-05,
      "loss": 0.6009266376495361,
      "step": 1045
    },
    {
      "epoch": 1.258880192655027,
      "grad_norm": 0.16449110209941864,
      "learning_rate": 3.3211168058560386e-05,
      "loss": 0.6029863357543945,
      "step": 1046
    },
    {
      "epoch": 1.2600842865743527,
      "grad_norm": 0.16384822130203247,
      "learning_rate": 3.311743723103357e-05,
      "loss": 0.5534104704856873,
      "step": 1047
    },
    {
      "epoch": 1.2612883804936785,
      "grad_norm": 0.17358887195587158,
      "learning_rate": 3.302377331858302e-05,
      "loss": 0.5713550448417664,
      "step": 1048
    },
    {
      "epoch": 1.2624924744130042,
      "grad_norm": 0.17000506818294525,
      "learning_rate": 3.293017669245144e-05,
      "loss": 0.5923644304275513,
      "step": 1049
    },
    {
      "epoch": 1.26369656833233,
      "grad_norm": 0.16530491411685944,
      "learning_rate": 3.283664772361476e-05,
      "loss": 0.5657535195350647,
      "step": 1050
    },
    {
      "epoch": 1.2649006622516556,
      "grad_norm": 0.1605950891971588,
      "learning_rate": 3.274318678278085e-05,
      "loss": 0.5940046310424805,
      "step": 1051
    },
    {
      "epoch": 1.2661047561709813,
      "grad_norm": 0.16642048954963684,
      "learning_rate": 3.264979424038789e-05,
      "loss": 0.5894283056259155,
      "step": 1052
    },
    {
      "epoch": 1.267308850090307,
      "grad_norm": 0.16590026021003723,
      "learning_rate": 3.255647046660296e-05,
      "loss": 0.6208168268203735,
      "step": 1053
    },
    {
      "epoch": 1.2685129440096328,
      "grad_norm": 0.16934767365455627,
      "learning_rate": 3.246321583132058e-05,
      "loss": 0.6506465673446655,
      "step": 1054
    },
    {
      "epoch": 1.2697170379289584,
      "grad_norm": 0.1705813705921173,
      "learning_rate": 3.237003070416122e-05,
      "loss": 0.6065971255302429,
      "step": 1055
    },
    {
      "epoch": 1.2709211318482843,
      "grad_norm": 0.1804744452238083,
      "learning_rate": 3.22769154544699e-05,
      "loss": 0.6065521836280823,
      "step": 1056
    },
    {
      "epoch": 1.27212522576761,
      "grad_norm": 0.17240937054157257,
      "learning_rate": 3.2183870451314624e-05,
      "loss": 0.561596155166626,
      "step": 1057
    },
    {
      "epoch": 1.2733293196869355,
      "grad_norm": 0.1627223938703537,
      "learning_rate": 3.2090896063485e-05,
      "loss": 0.5732601284980774,
      "step": 1058
    },
    {
      "epoch": 1.2745334136062612,
      "grad_norm": 0.16261062026023865,
      "learning_rate": 3.19979926594907e-05,
      "loss": 0.6336162090301514,
      "step": 1059
    },
    {
      "epoch": 1.275737507525587,
      "grad_norm": 0.16265533864498138,
      "learning_rate": 3.190516060756009e-05,
      "loss": 0.47382524609565735,
      "step": 1060
    },
    {
      "epoch": 1.2769416014449126,
      "grad_norm": 0.17050917446613312,
      "learning_rate": 3.181240027563875e-05,
      "loss": 0.6433200240135193,
      "step": 1061
    },
    {
      "epoch": 1.2781456953642385,
      "grad_norm": 0.16491582989692688,
      "learning_rate": 3.171971203138795e-05,
      "loss": 0.5382289886474609,
      "step": 1062
    },
    {
      "epoch": 1.2793497892835641,
      "grad_norm": 0.18545588850975037,
      "learning_rate": 3.1627096242183243e-05,
      "loss": 0.6317499279975891,
      "step": 1063
    },
    {
      "epoch": 1.2805538832028898,
      "grad_norm": 0.16492144763469696,
      "learning_rate": 3.1534553275112994e-05,
      "loss": 0.5808607339859009,
      "step": 1064
    },
    {
      "epoch": 1.2817579771222154,
      "grad_norm": 0.15888823568820953,
      "learning_rate": 3.144208349697696e-05,
      "loss": 0.6033707857131958,
      "step": 1065
    },
    {
      "epoch": 1.2829620710415413,
      "grad_norm": 0.18718279898166656,
      "learning_rate": 3.13496872742848e-05,
      "loss": 0.6049675345420837,
      "step": 1066
    },
    {
      "epoch": 1.2841661649608669,
      "grad_norm": 0.5164883732795715,
      "learning_rate": 3.125736497325464e-05,
      "loss": 0.6893006563186646,
      "step": 1067
    },
    {
      "epoch": 1.2853702588801927,
      "grad_norm": 0.16807810962200165,
      "learning_rate": 3.116511695981157e-05,
      "loss": 0.6553009152412415,
      "step": 1068
    },
    {
      "epoch": 1.2865743527995184,
      "grad_norm": 0.17083129286766052,
      "learning_rate": 3.107294359958628e-05,
      "loss": 0.6303476095199585,
      "step": 1069
    },
    {
      "epoch": 1.287778446718844,
      "grad_norm": 0.15924455225467682,
      "learning_rate": 3.098084525791356e-05,
      "loss": 0.6455898880958557,
      "step": 1070
    },
    {
      "epoch": 1.2889825406381699,
      "grad_norm": 0.15850618481636047,
      "learning_rate": 3.0888822299830854e-05,
      "loss": 0.5427078604698181,
      "step": 1071
    },
    {
      "epoch": 1.2901866345574955,
      "grad_norm": 0.1786697655916214,
      "learning_rate": 3.079687509007682e-05,
      "loss": 0.6142567992210388,
      "step": 1072
    },
    {
      "epoch": 1.2913907284768211,
      "grad_norm": 0.1683788001537323,
      "learning_rate": 3.070500399308987e-05,
      "loss": 0.5809274315834045,
      "step": 1073
    },
    {
      "epoch": 1.292594822396147,
      "grad_norm": 0.16226041316986084,
      "learning_rate": 3.0613209373006745e-05,
      "loss": 0.529127836227417,
      "step": 1074
    },
    {
      "epoch": 1.2937989163154726,
      "grad_norm": 0.15721242129802704,
      "learning_rate": 3.052149159366109e-05,
      "loss": 0.6804173588752747,
      "step": 1075
    },
    {
      "epoch": 1.2950030102347982,
      "grad_norm": 0.16457633674144745,
      "learning_rate": 3.0429851018581955e-05,
      "loss": 0.5480682253837585,
      "step": 1076
    },
    {
      "epoch": 1.296207104154124,
      "grad_norm": 0.1629006713628769,
      "learning_rate": 3.0338288010992398e-05,
      "loss": 0.5215630531311035,
      "step": 1077
    },
    {
      "epoch": 1.2974111980734497,
      "grad_norm": 0.1551741510629654,
      "learning_rate": 3.024680293380804e-05,
      "loss": 0.5615051984786987,
      "step": 1078
    },
    {
      "epoch": 1.2986152919927754,
      "grad_norm": 0.1624448001384735,
      "learning_rate": 3.0155396149635585e-05,
      "loss": 0.5636274218559265,
      "step": 1079
    },
    {
      "epoch": 1.2998193859121012,
      "grad_norm": 0.17864665389060974,
      "learning_rate": 3.0064068020771486e-05,
      "loss": 0.7295652031898499,
      "step": 1080
    },
    {
      "epoch": 1.3010234798314269,
      "grad_norm": 0.16876764595508575,
      "learning_rate": 2.9972818909200396e-05,
      "loss": 0.6045938730239868,
      "step": 1081
    },
    {
      "epoch": 1.3022275737507525,
      "grad_norm": 0.15641742944717407,
      "learning_rate": 2.988164917659375e-05,
      "loss": 0.6161600351333618,
      "step": 1082
    },
    {
      "epoch": 1.3034316676700783,
      "grad_norm": 0.17526626586914062,
      "learning_rate": 2.979055918430842e-05,
      "loss": 0.6593427658081055,
      "step": 1083
    },
    {
      "epoch": 1.304635761589404,
      "grad_norm": 0.17632213234901428,
      "learning_rate": 2.9699549293385176e-05,
      "loss": 0.7346253395080566,
      "step": 1084
    },
    {
      "epoch": 1.3058398555087296,
      "grad_norm": 0.16998739540576935,
      "learning_rate": 2.9608619864547337e-05,
      "loss": 0.6805224418640137,
      "step": 1085
    },
    {
      "epoch": 1.3070439494280555,
      "grad_norm": 0.15366403758525848,
      "learning_rate": 2.95177712581993e-05,
      "loss": 0.5345244407653809,
      "step": 1086
    },
    {
      "epoch": 1.308248043347381,
      "grad_norm": 0.15834107995033264,
      "learning_rate": 2.9427003834425075e-05,
      "loss": 0.5288026928901672,
      "step": 1087
    },
    {
      "epoch": 1.3094521372667067,
      "grad_norm": 0.1679813116788864,
      "learning_rate": 2.9336317952986946e-05,
      "loss": 0.5174233913421631,
      "step": 1088
    },
    {
      "epoch": 1.3106562311860326,
      "grad_norm": 0.15127837657928467,
      "learning_rate": 2.924571397332398e-05,
      "loss": 0.5448307394981384,
      "step": 1089
    },
    {
      "epoch": 1.3118603251053582,
      "grad_norm": 0.17347003519535065,
      "learning_rate": 2.915519225455065e-05,
      "loss": 0.6627212166786194,
      "step": 1090
    },
    {
      "epoch": 1.3130644190246838,
      "grad_norm": 0.17600205540657043,
      "learning_rate": 2.906475315545532e-05,
      "loss": 0.5408031940460205,
      "step": 1091
    },
    {
      "epoch": 1.3142685129440097,
      "grad_norm": 0.1831415295600891,
      "learning_rate": 2.8974397034498917e-05,
      "loss": 0.5786025524139404,
      "step": 1092
    },
    {
      "epoch": 1.3154726068633353,
      "grad_norm": 0.27714836597442627,
      "learning_rate": 2.8884124249813526e-05,
      "loss": 0.5641752481460571,
      "step": 1093
    },
    {
      "epoch": 1.316676700782661,
      "grad_norm": 0.15373986959457397,
      "learning_rate": 2.879393515920087e-05,
      "loss": 0.5789990425109863,
      "step": 1094
    },
    {
      "epoch": 1.3178807947019868,
      "grad_norm": 0.17779627442359924,
      "learning_rate": 2.8703830120130976e-05,
      "loss": 0.6139446496963501,
      "step": 1095
    },
    {
      "epoch": 1.3190848886213125,
      "grad_norm": 0.16855160892009735,
      "learning_rate": 2.8613809489740662e-05,
      "loss": 0.7243695259094238,
      "step": 1096
    },
    {
      "epoch": 1.320288982540638,
      "grad_norm": 0.17640021443367004,
      "learning_rate": 2.8523873624832247e-05,
      "loss": 0.6266144514083862,
      "step": 1097
    },
    {
      "epoch": 1.321493076459964,
      "grad_norm": 0.1791626214981079,
      "learning_rate": 2.8434022881872103e-05,
      "loss": 0.5299221277236938,
      "step": 1098
    },
    {
      "epoch": 1.3226971703792896,
      "grad_norm": 0.17264147102832794,
      "learning_rate": 2.8344257616989144e-05,
      "loss": 0.5940788388252258,
      "step": 1099
    },
    {
      "epoch": 1.3239012642986152,
      "grad_norm": 0.16167393326759338,
      "learning_rate": 2.8254578185973556e-05,
      "loss": 0.6222594976425171,
      "step": 1100
    },
    {
      "epoch": 1.325105358217941,
      "grad_norm": 0.1643231064081192,
      "learning_rate": 2.8164984944275242e-05,
      "loss": 0.6413952112197876,
      "step": 1101
    },
    {
      "epoch": 1.3263094521372667,
      "grad_norm": 0.1838713437318802,
      "learning_rate": 2.8075478247002518e-05,
      "loss": 0.5988833904266357,
      "step": 1102
    },
    {
      "epoch": 1.3275135460565926,
      "grad_norm": 0.16674576699733734,
      "learning_rate": 2.7986058448920728e-05,
      "loss": 0.5993229150772095,
      "step": 1103
    },
    {
      "epoch": 1.3287176399759182,
      "grad_norm": 0.16011007130146027,
      "learning_rate": 2.7896725904450748e-05,
      "loss": 0.680863618850708,
      "step": 1104
    },
    {
      "epoch": 1.3299217338952438,
      "grad_norm": 0.19493143260478973,
      "learning_rate": 2.7807480967667576e-05,
      "loss": 0.7075426578521729,
      "step": 1105
    },
    {
      "epoch": 1.3311258278145695,
      "grad_norm": 0.16794447600841522,
      "learning_rate": 2.7718323992299e-05,
      "loss": 0.6083692908287048,
      "step": 1106
    },
    {
      "epoch": 1.3323299217338953,
      "grad_norm": 0.17588356137275696,
      "learning_rate": 2.7629255331724225e-05,
      "loss": 0.6508508324623108,
      "step": 1107
    },
    {
      "epoch": 1.333534015653221,
      "grad_norm": 0.16961358487606049,
      "learning_rate": 2.7540275338972343e-05,
      "loss": 0.594325065612793,
      "step": 1108
    },
    {
      "epoch": 1.3347381095725468,
      "grad_norm": 0.1566169708967209,
      "learning_rate": 2.7451384366721057e-05,
      "loss": 0.6136887073516846,
      "step": 1109
    },
    {
      "epoch": 1.3359422034918724,
      "grad_norm": 0.1754075437784195,
      "learning_rate": 2.7362582767295158e-05,
      "loss": 0.5982072949409485,
      "step": 1110
    },
    {
      "epoch": 1.337146297411198,
      "grad_norm": 0.15479500591754913,
      "learning_rate": 2.7273870892665253e-05,
      "loss": 0.6467243432998657,
      "step": 1111
    },
    {
      "epoch": 1.3383503913305237,
      "grad_norm": 0.1562364399433136,
      "learning_rate": 2.718524909444635e-05,
      "loss": 0.5602697730064392,
      "step": 1112
    },
    {
      "epoch": 1.3395544852498495,
      "grad_norm": 0.17557063698768616,
      "learning_rate": 2.709671772389639e-05,
      "loss": 0.6857227087020874,
      "step": 1113
    },
    {
      "epoch": 1.3407585791691752,
      "grad_norm": 0.17871588468551636,
      "learning_rate": 2.7008277131914916e-05,
      "loss": 0.6147187352180481,
      "step": 1114
    },
    {
      "epoch": 1.341962673088501,
      "grad_norm": 0.15592212975025177,
      "learning_rate": 2.691992766904161e-05,
      "loss": 0.6028821468353271,
      "step": 1115
    },
    {
      "epoch": 1.3431667670078267,
      "grad_norm": 0.15563534200191498,
      "learning_rate": 2.683166968545503e-05,
      "loss": 0.5705239772796631,
      "step": 1116
    },
    {
      "epoch": 1.3443708609271523,
      "grad_norm": 0.15587513148784637,
      "learning_rate": 2.6743503530971138e-05,
      "loss": 0.580167829990387,
      "step": 1117
    },
    {
      "epoch": 1.345574954846478,
      "grad_norm": 0.17696598172187805,
      "learning_rate": 2.6655429555041922e-05,
      "loss": 0.6762676239013672,
      "step": 1118
    },
    {
      "epoch": 1.3467790487658038,
      "grad_norm": 0.16402024030685425,
      "learning_rate": 2.656744810675397e-05,
      "loss": 0.6196094751358032,
      "step": 1119
    },
    {
      "epoch": 1.3479831426851294,
      "grad_norm": 0.16355285048484802,
      "learning_rate": 2.6479559534827168e-05,
      "loss": 0.6322900652885437,
      "step": 1120
    },
    {
      "epoch": 1.3491872366044553,
      "grad_norm": 0.1462378352880478,
      "learning_rate": 2.6391764187613277e-05,
      "loss": 0.5070129036903381,
      "step": 1121
    },
    {
      "epoch": 1.350391330523781,
      "grad_norm": 0.15422296524047852,
      "learning_rate": 2.6304062413094588e-05,
      "loss": 0.6439323425292969,
      "step": 1122
    },
    {
      "epoch": 1.3515954244431065,
      "grad_norm": 0.1549481451511383,
      "learning_rate": 2.6216454558882486e-05,
      "loss": 0.5427043437957764,
      "step": 1123
    },
    {
      "epoch": 1.3527995183624322,
      "grad_norm": 0.1677609384059906,
      "learning_rate": 2.6128940972216044e-05,
      "loss": 0.5514862537384033,
      "step": 1124
    },
    {
      "epoch": 1.354003612281758,
      "grad_norm": 0.14602091908454895,
      "learning_rate": 2.604152199996077e-05,
      "loss": 0.6449571251869202,
      "step": 1125
    },
    {
      "epoch": 1.3552077062010837,
      "grad_norm": 0.1636328250169754,
      "learning_rate": 2.5954197988607133e-05,
      "loss": 0.5468547940254211,
      "step": 1126
    },
    {
      "epoch": 1.3564118001204095,
      "grad_norm": 0.16185049712657928,
      "learning_rate": 2.586696928426926e-05,
      "loss": 0.6048194169998169,
      "step": 1127
    },
    {
      "epoch": 1.3576158940397351,
      "grad_norm": 0.15281720459461212,
      "learning_rate": 2.5779836232683442e-05,
      "loss": 0.5126120448112488,
      "step": 1128
    },
    {
      "epoch": 1.3588199879590608,
      "grad_norm": 0.16885338723659515,
      "learning_rate": 2.5692799179206906e-05,
      "loss": 0.5917194485664368,
      "step": 1129
    },
    {
      "epoch": 1.3600240818783864,
      "grad_norm": 0.1634187251329422,
      "learning_rate": 2.5605858468816358e-05,
      "loss": 0.5365522503852844,
      "step": 1130
    },
    {
      "epoch": 1.3612281757977123,
      "grad_norm": 0.1805935651063919,
      "learning_rate": 2.5519014446106682e-05,
      "loss": 0.6086816787719727,
      "step": 1131
    },
    {
      "epoch": 1.362432269717038,
      "grad_norm": 0.15620438754558563,
      "learning_rate": 2.5432267455289503e-05,
      "loss": 0.6138871312141418,
      "step": 1132
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 0.1816980391740799,
      "learning_rate": 2.5345617840191828e-05,
      "loss": 0.6277750134468079,
      "step": 1133
    },
    {
      "epoch": 1.3648404575556894,
      "grad_norm": 0.18459297716617584,
      "learning_rate": 2.5259065944254746e-05,
      "loss": 0.5710020065307617,
      "step": 1134
    },
    {
      "epoch": 1.366044551475015,
      "grad_norm": 0.16930045187473297,
      "learning_rate": 2.5172612110532012e-05,
      "loss": 0.5848778486251831,
      "step": 1135
    },
    {
      "epoch": 1.3672486453943407,
      "grad_norm": 0.16714534163475037,
      "learning_rate": 2.5086256681688745e-05,
      "loss": 0.5663998126983643,
      "step": 1136
    },
    {
      "epoch": 1.3684527393136665,
      "grad_norm": 0.16956114768981934,
      "learning_rate": 2.500000000000001e-05,
      "loss": 0.6081545352935791,
      "step": 1137
    },
    {
      "epoch": 1.3696568332329921,
      "grad_norm": 0.16294512152671814,
      "learning_rate": 2.491384240734943e-05,
      "loss": 0.6211032867431641,
      "step": 1138
    },
    {
      "epoch": 1.370860927152318,
      "grad_norm": 0.1613115519285202,
      "learning_rate": 2.4827784245227937e-05,
      "loss": 0.5988506078720093,
      "step": 1139
    },
    {
      "epoch": 1.3720650210716436,
      "grad_norm": 0.15139776468276978,
      "learning_rate": 2.4741825854732344e-05,
      "loss": 0.6277011036872864,
      "step": 1140
    },
    {
      "epoch": 1.3732691149909693,
      "grad_norm": 0.16650590300559998,
      "learning_rate": 2.4655967576564064e-05,
      "loss": 0.5168682336807251,
      "step": 1141
    },
    {
      "epoch": 1.374473208910295,
      "grad_norm": 0.1660197675228119,
      "learning_rate": 2.4570209751027623e-05,
      "loss": 0.6027352809906006,
      "step": 1142
    },
    {
      "epoch": 1.3756773028296208,
      "grad_norm": 0.16958127915859222,
      "learning_rate": 2.4484552718029448e-05,
      "loss": 0.5519788265228271,
      "step": 1143
    },
    {
      "epoch": 1.3768813967489464,
      "grad_norm": 0.1609538495540619,
      "learning_rate": 2.439899681707646e-05,
      "loss": 0.5937942266464233,
      "step": 1144
    },
    {
      "epoch": 1.3780854906682722,
      "grad_norm": 0.20702077448368073,
      "learning_rate": 2.4313542387274717e-05,
      "loss": 0.6151435375213623,
      "step": 1145
    },
    {
      "epoch": 1.3792895845875979,
      "grad_norm": 0.17944055795669556,
      "learning_rate": 2.4228189767328173e-05,
      "loss": 0.6017966270446777,
      "step": 1146
    },
    {
      "epoch": 1.3804936785069235,
      "grad_norm": 0.162122443318367,
      "learning_rate": 2.4142939295537126e-05,
      "loss": 0.5728113651275635,
      "step": 1147
    },
    {
      "epoch": 1.3816977724262491,
      "grad_norm": 0.16300708055496216,
      "learning_rate": 2.405779130979709e-05,
      "loss": 0.5014937520027161,
      "step": 1148
    },
    {
      "epoch": 1.382901866345575,
      "grad_norm": 0.1317259818315506,
      "learning_rate": 2.3972746147597335e-05,
      "loss": 0.47494035959243774,
      "step": 1149
    },
    {
      "epoch": 1.3841059602649006,
      "grad_norm": 0.1612226814031601,
      "learning_rate": 2.388780414601959e-05,
      "loss": 0.5716497898101807,
      "step": 1150
    },
    {
      "epoch": 1.3853100541842265,
      "grad_norm": 0.16332001984119415,
      "learning_rate": 2.380296564173674e-05,
      "loss": 0.582008421421051,
      "step": 1151
    },
    {
      "epoch": 1.3865141481035521,
      "grad_norm": 0.17107804119586945,
      "learning_rate": 2.3718230971011386e-05,
      "loss": 0.5154463648796082,
      "step": 1152
    },
    {
      "epoch": 1.3877182420228777,
      "grad_norm": 0.15683507919311523,
      "learning_rate": 2.3633600469694606e-05,
      "loss": 0.5298494100570679,
      "step": 1153
    },
    {
      "epoch": 1.3889223359422034,
      "grad_norm": 0.17860683798789978,
      "learning_rate": 2.3549074473224612e-05,
      "loss": 0.6552767753601074,
      "step": 1154
    },
    {
      "epoch": 1.3901264298615292,
      "grad_norm": 0.1721535623073578,
      "learning_rate": 2.3464653316625385e-05,
      "loss": 0.7259576320648193,
      "step": 1155
    },
    {
      "epoch": 1.3913305237808549,
      "grad_norm": 0.1654064804315567,
      "learning_rate": 2.3380337334505374e-05,
      "loss": 0.5280846357345581,
      "step": 1156
    },
    {
      "epoch": 1.3925346177001807,
      "grad_norm": 0.16835972666740417,
      "learning_rate": 2.3296126861056167e-05,
      "loss": 0.6780106425285339,
      "step": 1157
    },
    {
      "epoch": 1.3937387116195064,
      "grad_norm": 0.16164492070674896,
      "learning_rate": 2.3212022230051154e-05,
      "loss": 0.6554681062698364,
      "step": 1158
    },
    {
      "epoch": 1.394942805538832,
      "grad_norm": 0.17035424709320068,
      "learning_rate": 2.3128023774844194e-05,
      "loss": 0.604200541973114,
      "step": 1159
    },
    {
      "epoch": 1.3961468994581576,
      "grad_norm": 0.15810167789459229,
      "learning_rate": 2.3044131828368386e-05,
      "loss": 0.5685842633247375,
      "step": 1160
    },
    {
      "epoch": 1.3973509933774835,
      "grad_norm": 0.18553194403648376,
      "learning_rate": 2.2960346723134575e-05,
      "loss": 0.6021862030029297,
      "step": 1161
    },
    {
      "epoch": 1.398555087296809,
      "grad_norm": 0.1716788113117218,
      "learning_rate": 2.2876668791230193e-05,
      "loss": 0.5697296857833862,
      "step": 1162
    },
    {
      "epoch": 1.399759181216135,
      "grad_norm": 0.1573840081691742,
      "learning_rate": 2.2793098364317868e-05,
      "loss": 0.5595786571502686,
      "step": 1163
    },
    {
      "epoch": 1.4009632751354606,
      "grad_norm": 0.1667952537536621,
      "learning_rate": 2.2709635773634113e-05,
      "loss": 0.6001214981079102,
      "step": 1164
    },
    {
      "epoch": 1.4021673690547862,
      "grad_norm": 0.17569540441036224,
      "learning_rate": 2.2626281349988103e-05,
      "loss": 0.6388605237007141,
      "step": 1165
    },
    {
      "epoch": 1.4033714629741119,
      "grad_norm": 0.19071106612682343,
      "learning_rate": 2.254303542376016e-05,
      "loss": 0.6901451945304871,
      "step": 1166
    },
    {
      "epoch": 1.4045755568934377,
      "grad_norm": 0.17054232954978943,
      "learning_rate": 2.2459898324900662e-05,
      "loss": 0.5873799324035645,
      "step": 1167
    },
    {
      "epoch": 1.4057796508127633,
      "grad_norm": 0.15402156114578247,
      "learning_rate": 2.2376870382928607e-05,
      "loss": 0.5346029996871948,
      "step": 1168
    },
    {
      "epoch": 1.4069837447320892,
      "grad_norm": 0.16029417514801025,
      "learning_rate": 2.2293951926930356e-05,
      "loss": 0.6451740264892578,
      "step": 1169
    },
    {
      "epoch": 1.4081878386514148,
      "grad_norm": 0.1675398200750351,
      "learning_rate": 2.2211143285558312e-05,
      "loss": 0.6542309522628784,
      "step": 1170
    },
    {
      "epoch": 1.4093919325707405,
      "grad_norm": 0.16078129410743713,
      "learning_rate": 2.2128444787029618e-05,
      "loss": 0.6347694396972656,
      "step": 1171
    },
    {
      "epoch": 1.410596026490066,
      "grad_norm": 0.17774108052253723,
      "learning_rate": 2.204585675912486e-05,
      "loss": 0.5897160768508911,
      "step": 1172
    },
    {
      "epoch": 1.411800120409392,
      "grad_norm": 0.16636426746845245,
      "learning_rate": 2.1963379529186768e-05,
      "loss": 0.5917282104492188,
      "step": 1173
    },
    {
      "epoch": 1.4130042143287176,
      "grad_norm": 0.15956087410449982,
      "learning_rate": 2.1881013424118922e-05,
      "loss": 0.591199517250061,
      "step": 1174
    },
    {
      "epoch": 1.4142083082480434,
      "grad_norm": 0.1708986610174179,
      "learning_rate": 2.1798758770384442e-05,
      "loss": 0.547240674495697,
      "step": 1175
    },
    {
      "epoch": 1.415412402167369,
      "grad_norm": 0.16554829478263855,
      "learning_rate": 2.171661589400471e-05,
      "loss": 0.6129944920539856,
      "step": 1176
    },
    {
      "epoch": 1.4166164960866947,
      "grad_norm": 0.16216960549354553,
      "learning_rate": 2.1634585120558078e-05,
      "loss": 0.6224234104156494,
      "step": 1177
    },
    {
      "epoch": 1.4178205900060203,
      "grad_norm": 0.17306363582611084,
      "learning_rate": 2.1552666775178548e-05,
      "loss": 0.6343253254890442,
      "step": 1178
    },
    {
      "epoch": 1.4190246839253462,
      "grad_norm": 0.17951203882694244,
      "learning_rate": 2.1470861182554534e-05,
      "loss": 0.5904716849327087,
      "step": 1179
    },
    {
      "epoch": 1.4202287778446718,
      "grad_norm": 0.16520901024341583,
      "learning_rate": 2.138916866692754e-05,
      "loss": 0.5932174921035767,
      "step": 1180
    },
    {
      "epoch": 1.4214328717639977,
      "grad_norm": 0.17519637942314148,
      "learning_rate": 2.1307589552090866e-05,
      "loss": 0.662615954875946,
      "step": 1181
    },
    {
      "epoch": 1.4226369656833233,
      "grad_norm": 0.14822949469089508,
      "learning_rate": 2.1226124161388354e-05,
      "loss": 0.5252461433410645,
      "step": 1182
    },
    {
      "epoch": 1.423841059602649,
      "grad_norm": 0.16894812881946564,
      "learning_rate": 2.1144772817713103e-05,
      "loss": 0.5238260626792908,
      "step": 1183
    },
    {
      "epoch": 1.4250451535219746,
      "grad_norm": 0.1621467024087906,
      "learning_rate": 2.106353584350616e-05,
      "loss": 0.5356015563011169,
      "step": 1184
    },
    {
      "epoch": 1.4262492474413004,
      "grad_norm": 0.15989820659160614,
      "learning_rate": 2.0982413560755272e-05,
      "loss": 0.5089754462242126,
      "step": 1185
    },
    {
      "epoch": 1.427453341360626,
      "grad_norm": 0.17008116841316223,
      "learning_rate": 2.0901406290993598e-05,
      "loss": 0.5833384990692139,
      "step": 1186
    },
    {
      "epoch": 1.428657435279952,
      "grad_norm": 0.16225847601890564,
      "learning_rate": 2.0820514355298432e-05,
      "loss": 0.5259945392608643,
      "step": 1187
    },
    {
      "epoch": 1.4298615291992776,
      "grad_norm": 0.16823600232601166,
      "learning_rate": 2.073973807428995e-05,
      "loss": 0.5302376747131348,
      "step": 1188
    },
    {
      "epoch": 1.4310656231186032,
      "grad_norm": 0.1717822253704071,
      "learning_rate": 2.0659077768129898e-05,
      "loss": 0.5729860067367554,
      "step": 1189
    },
    {
      "epoch": 1.4322697170379288,
      "grad_norm": 0.16328737139701843,
      "learning_rate": 2.0578533756520375e-05,
      "loss": 0.526997447013855,
      "step": 1190
    },
    {
      "epoch": 1.4334738109572547,
      "grad_norm": 0.1617710292339325,
      "learning_rate": 2.049810635870251e-05,
      "loss": 0.5945241451263428,
      "step": 1191
    },
    {
      "epoch": 1.4346779048765803,
      "grad_norm": 0.1537000685930252,
      "learning_rate": 2.0417795893455265e-05,
      "loss": 0.553534209728241,
      "step": 1192
    },
    {
      "epoch": 1.4358819987959062,
      "grad_norm": 0.1759205013513565,
      "learning_rate": 2.0337602679094092e-05,
      "loss": 0.6230572462081909,
      "step": 1193
    },
    {
      "epoch": 1.4370860927152318,
      "grad_norm": 0.17249514162540436,
      "learning_rate": 2.0257527033469735e-05,
      "loss": 0.6353952884674072,
      "step": 1194
    },
    {
      "epoch": 1.4382901866345574,
      "grad_norm": 0.1703486442565918,
      "learning_rate": 2.0177569273966945e-05,
      "loss": 0.6504722237586975,
      "step": 1195
    },
    {
      "epoch": 1.439494280553883,
      "grad_norm": 0.17596854269504547,
      "learning_rate": 2.009772971750322e-05,
      "loss": 0.6102354526519775,
      "step": 1196
    },
    {
      "epoch": 1.440698374473209,
      "grad_norm": 0.17083287239074707,
      "learning_rate": 2.0018008680527556e-05,
      "loss": 0.5687814950942993,
      "step": 1197
    },
    {
      "epoch": 1.4419024683925346,
      "grad_norm": 0.16367903351783752,
      "learning_rate": 1.9938406479019183e-05,
      "loss": 0.6123007535934448,
      "step": 1198
    },
    {
      "epoch": 1.4431065623118604,
      "grad_norm": 0.1769656538963318,
      "learning_rate": 1.985892342848633e-05,
      "loss": 0.6357004642486572,
      "step": 1199
    },
    {
      "epoch": 1.444310656231186,
      "grad_norm": 0.8626880049705505,
      "learning_rate": 1.9779559843964957e-05,
      "loss": 0.6029178500175476,
      "step": 1200
    },
    {
      "epoch": 1.4455147501505117,
      "grad_norm": 0.1776512712240219,
      "learning_rate": 1.9700316040017515e-05,
      "loss": 0.5986659526824951,
      "step": 1201
    },
    {
      "epoch": 1.4467188440698375,
      "grad_norm": 0.19223180413246155,
      "learning_rate": 1.9621192330731684e-05,
      "loss": 0.6336244344711304,
      "step": 1202
    },
    {
      "epoch": 1.4479229379891632,
      "grad_norm": 0.17094776034355164,
      "learning_rate": 1.954218902971917e-05,
      "loss": 0.6061995625495911,
      "step": 1203
    },
    {
      "epoch": 1.4491270319084888,
      "grad_norm": 0.16348566114902496,
      "learning_rate": 1.9463306450114416e-05,
      "loss": 0.562362790107727,
      "step": 1204
    },
    {
      "epoch": 1.4503311258278146,
      "grad_norm": 0.15452855825424194,
      "learning_rate": 1.938454490457337e-05,
      "loss": 0.5266240835189819,
      "step": 1205
    },
    {
      "epoch": 1.4515352197471403,
      "grad_norm": 0.17398321628570557,
      "learning_rate": 1.9305904705272275e-05,
      "loss": 0.5572274923324585,
      "step": 1206
    },
    {
      "epoch": 1.452739313666466,
      "grad_norm": 0.16937664151191711,
      "learning_rate": 1.922738616390639e-05,
      "loss": 0.6406630277633667,
      "step": 1207
    },
    {
      "epoch": 1.4539434075857918,
      "grad_norm": 0.16251932084560394,
      "learning_rate": 1.9148989591688786e-05,
      "loss": 0.48134297132492065,
      "step": 1208
    },
    {
      "epoch": 1.4551475015051174,
      "grad_norm": 0.16470041871070862,
      "learning_rate": 1.9070715299349105e-05,
      "loss": 0.5903266668319702,
      "step": 1209
    },
    {
      "epoch": 1.456351595424443,
      "grad_norm": 0.16918903589248657,
      "learning_rate": 1.8992563597132323e-05,
      "loss": 0.6756315231323242,
      "step": 1210
    },
    {
      "epoch": 1.457555689343769,
      "grad_norm": 0.1502673327922821,
      "learning_rate": 1.8914534794797513e-05,
      "loss": 0.5706387758255005,
      "step": 1211
    },
    {
      "epoch": 1.4587597832630945,
      "grad_norm": 0.16120700538158417,
      "learning_rate": 1.8836629201616635e-05,
      "loss": 0.6284589767456055,
      "step": 1212
    },
    {
      "epoch": 1.4599638771824202,
      "grad_norm": 0.1616784632205963,
      "learning_rate": 1.8758847126373303e-05,
      "loss": 0.6258765459060669,
      "step": 1213
    },
    {
      "epoch": 1.461167971101746,
      "grad_norm": 0.15989549458026886,
      "learning_rate": 1.8681188877361555e-05,
      "loss": 0.5628883838653564,
      "step": 1214
    },
    {
      "epoch": 1.4623720650210716,
      "grad_norm": 0.16660171747207642,
      "learning_rate": 1.8603654762384642e-05,
      "loss": 0.521628737449646,
      "step": 1215
    },
    {
      "epoch": 1.4635761589403973,
      "grad_norm": 0.1649349331855774,
      "learning_rate": 1.85262450887538e-05,
      "loss": 0.558128833770752,
      "step": 1216
    },
    {
      "epoch": 1.4647802528597231,
      "grad_norm": 0.1684584766626358,
      "learning_rate": 1.8448960163287034e-05,
      "loss": 0.6911593079566956,
      "step": 1217
    },
    {
      "epoch": 1.4659843467790488,
      "grad_norm": 0.1754569113254547,
      "learning_rate": 1.83718002923079e-05,
      "loss": 0.5982605814933777,
      "step": 1218
    },
    {
      "epoch": 1.4671884406983744,
      "grad_norm": 0.19743938744068146,
      "learning_rate": 1.8294765781644285e-05,
      "loss": 0.5992114543914795,
      "step": 1219
    },
    {
      "epoch": 1.4683925346177003,
      "grad_norm": 0.1823793202638626,
      "learning_rate": 1.821785693662724e-05,
      "loss": 0.6889064311981201,
      "step": 1220
    },
    {
      "epoch": 1.4695966285370259,
      "grad_norm": 0.21644377708435059,
      "learning_rate": 1.8141074062089657e-05,
      "loss": 0.6728036403656006,
      "step": 1221
    },
    {
      "epoch": 1.4708007224563515,
      "grad_norm": 0.17821632325649261,
      "learning_rate": 1.8064417462365226e-05,
      "loss": 0.5144159197807312,
      "step": 1222
    },
    {
      "epoch": 1.4720048163756774,
      "grad_norm": 0.15794169902801514,
      "learning_rate": 1.798788744128707e-05,
      "loss": 0.6287606954574585,
      "step": 1223
    },
    {
      "epoch": 1.473208910295003,
      "grad_norm": 0.16721273958683014,
      "learning_rate": 1.791148430218666e-05,
      "loss": 0.6314302682876587,
      "step": 1224
    },
    {
      "epoch": 1.4744130042143286,
      "grad_norm": 0.17019809782505035,
      "learning_rate": 1.7835208347892535e-05,
      "loss": 0.562951385974884,
      "step": 1225
    },
    {
      "epoch": 1.4756170981336545,
      "grad_norm": 0.15808868408203125,
      "learning_rate": 1.775905988072909e-05,
      "loss": 0.666721522808075,
      "step": 1226
    },
    {
      "epoch": 1.4768211920529801,
      "grad_norm": 0.17738115787506104,
      "learning_rate": 1.768303920251551e-05,
      "loss": 0.6059695482254028,
      "step": 1227
    },
    {
      "epoch": 1.4780252859723058,
      "grad_norm": 0.17433080077171326,
      "learning_rate": 1.7607146614564418e-05,
      "loss": 0.7393290996551514,
      "step": 1228
    },
    {
      "epoch": 1.4792293798916316,
      "grad_norm": 0.15435297787189484,
      "learning_rate": 1.753138241768075e-05,
      "loss": 0.5769613981246948,
      "step": 1229
    },
    {
      "epoch": 1.4804334738109572,
      "grad_norm": 0.1618458777666092,
      "learning_rate": 1.7455746912160564e-05,
      "loss": 0.5489301085472107,
      "step": 1230
    },
    {
      "epoch": 1.4816375677302829,
      "grad_norm": 0.16177277266979218,
      "learning_rate": 1.7380240397789836e-05,
      "loss": 0.6001340746879578,
      "step": 1231
    },
    {
      "epoch": 1.4828416616496087,
      "grad_norm": 0.16781818866729736,
      "learning_rate": 1.7304863173843283e-05,
      "loss": 0.6215763092041016,
      "step": 1232
    },
    {
      "epoch": 1.4840457555689344,
      "grad_norm": 0.16190500557422638,
      "learning_rate": 1.722961553908318e-05,
      "loss": 0.5031696557998657,
      "step": 1233
    },
    {
      "epoch": 1.4852498494882602,
      "grad_norm": 0.1518600583076477,
      "learning_rate": 1.7154497791758157e-05,
      "loss": 0.5363492369651794,
      "step": 1234
    },
    {
      "epoch": 1.4864539434075859,
      "grad_norm": 0.16558556258678436,
      "learning_rate": 1.707951022960199e-05,
      "loss": 0.6088852882385254,
      "step": 1235
    },
    {
      "epoch": 1.4876580373269115,
      "grad_norm": 0.16319610178470612,
      "learning_rate": 1.700465314983254e-05,
      "loss": 0.5996498465538025,
      "step": 1236
    },
    {
      "epoch": 1.4888621312462371,
      "grad_norm": 0.17455993592739105,
      "learning_rate": 1.6929926849150428e-05,
      "loss": 0.5898240804672241,
      "step": 1237
    },
    {
      "epoch": 1.490066225165563,
      "grad_norm": 0.16749973595142365,
      "learning_rate": 1.6855331623737958e-05,
      "loss": 0.5993600487709045,
      "step": 1238
    },
    {
      "epoch": 1.4912703190848886,
      "grad_norm": 0.18684302270412445,
      "learning_rate": 1.6780867769257898e-05,
      "loss": 0.6024582982063293,
      "step": 1239
    },
    {
      "epoch": 1.4924744130042145,
      "grad_norm": 0.16553924977779388,
      "learning_rate": 1.6706535580852267e-05,
      "loss": 0.5867906808853149,
      "step": 1240
    },
    {
      "epoch": 1.49367850692354,
      "grad_norm": 0.1652764230966568,
      "learning_rate": 1.6632335353141333e-05,
      "loss": 0.5839992165565491,
      "step": 1241
    },
    {
      "epoch": 1.4948826008428657,
      "grad_norm": 0.16593536734580994,
      "learning_rate": 1.6558267380222224e-05,
      "loss": 0.5716294646263123,
      "step": 1242
    },
    {
      "epoch": 1.4960866947621914,
      "grad_norm": 0.16340181231498718,
      "learning_rate": 1.6484331955667947e-05,
      "loss": 0.6093785762786865,
      "step": 1243
    },
    {
      "epoch": 1.4972907886815172,
      "grad_norm": 0.16942591965198517,
      "learning_rate": 1.6410529372526057e-05,
      "loss": 0.5457426309585571,
      "step": 1244
    },
    {
      "epoch": 1.4984948826008428,
      "grad_norm": 0.17523625493049622,
      "learning_rate": 1.6336859923317643e-05,
      "loss": 0.5633116960525513,
      "step": 1245
    },
    {
      "epoch": 1.4996989765201687,
      "grad_norm": 0.17758457362651825,
      "learning_rate": 1.6263323900036126e-05,
      "loss": 0.5521470904350281,
      "step": 1246
    },
    {
      "epoch": 1.5009030704394943,
      "grad_norm": 0.1720777451992035,
      "learning_rate": 1.6189921594146046e-05,
      "loss": 0.6299923658370972,
      "step": 1247
    },
    {
      "epoch": 1.50210716435882,
      "grad_norm": 0.1923622190952301,
      "learning_rate": 1.6116653296581975e-05,
      "loss": 0.6659127473831177,
      "step": 1248
    },
    {
      "epoch": 1.5033112582781456,
      "grad_norm": 0.1628662794828415,
      "learning_rate": 1.6043519297747285e-05,
      "loss": 0.5213074088096619,
      "step": 1249
    },
    {
      "epoch": 1.5045153521974715,
      "grad_norm": 0.18480616807937622,
      "learning_rate": 1.5970519887513073e-05,
      "loss": 0.6727724671363831,
      "step": 1250
    },
    {
      "epoch": 1.505719446116797,
      "grad_norm": 0.1616045981645584,
      "learning_rate": 1.5897655355217038e-05,
      "loss": 0.5364658832550049,
      "step": 1251
    },
    {
      "epoch": 1.506923540036123,
      "grad_norm": 0.19679591059684753,
      "learning_rate": 1.5824925989662216e-05,
      "loss": 0.5936850309371948,
      "step": 1252
    },
    {
      "epoch": 1.5081276339554486,
      "grad_norm": 0.1580863893032074,
      "learning_rate": 1.5752332079115932e-05,
      "loss": 0.5735425353050232,
      "step": 1253
    },
    {
      "epoch": 1.5093317278747742,
      "grad_norm": 0.16985823214054108,
      "learning_rate": 1.5679873911308597e-05,
      "loss": 0.6173081994056702,
      "step": 1254
    },
    {
      "epoch": 1.5105358217940998,
      "grad_norm": 0.16111443936824799,
      "learning_rate": 1.56075517734326e-05,
      "loss": 0.46890902519226074,
      "step": 1255
    },
    {
      "epoch": 1.5117399157134257,
      "grad_norm": 0.16507288813591003,
      "learning_rate": 1.5535365952141233e-05,
      "loss": 0.5570113658905029,
      "step": 1256
    },
    {
      "epoch": 1.5129440096327513,
      "grad_norm": 0.18061162531375885,
      "learning_rate": 1.5463316733547434e-05,
      "loss": 0.5872465372085571,
      "step": 1257
    },
    {
      "epoch": 1.5141481035520772,
      "grad_norm": 0.21220402419567108,
      "learning_rate": 1.5391404403222676e-05,
      "loss": 0.6409997940063477,
      "step": 1258
    },
    {
      "epoch": 1.5153521974714028,
      "grad_norm": 0.1598881185054779,
      "learning_rate": 1.5319629246195917e-05,
      "loss": 0.5302499532699585,
      "step": 1259
    },
    {
      "epoch": 1.5165562913907285,
      "grad_norm": 0.17412401735782623,
      "learning_rate": 1.5247991546952428e-05,
      "loss": 0.6735079288482666,
      "step": 1260
    },
    {
      "epoch": 1.517760385310054,
      "grad_norm": 0.19888164103031158,
      "learning_rate": 1.5176491589432628e-05,
      "loss": 0.6130843162536621,
      "step": 1261
    },
    {
      "epoch": 1.51896447922938,
      "grad_norm": 0.16864970326423645,
      "learning_rate": 1.5105129657031009e-05,
      "loss": 0.5657567977905273,
      "step": 1262
    },
    {
      "epoch": 1.5201685731487056,
      "grad_norm": 0.17937639355659485,
      "learning_rate": 1.5033906032594958e-05,
      "loss": 0.5417755842208862,
      "step": 1263
    },
    {
      "epoch": 1.5213726670680314,
      "grad_norm": 0.16613496840000153,
      "learning_rate": 1.4962820998423683e-05,
      "loss": 0.5459073781967163,
      "step": 1264
    },
    {
      "epoch": 1.522576760987357,
      "grad_norm": 0.1830260306596756,
      "learning_rate": 1.4891874836267127e-05,
      "loss": 0.5660001039505005,
      "step": 1265
    },
    {
      "epoch": 1.5237808549066827,
      "grad_norm": 0.18154628574848175,
      "learning_rate": 1.4821067827324753e-05,
      "loss": 0.5635465383529663,
      "step": 1266
    },
    {
      "epoch": 1.5249849488260083,
      "grad_norm": 0.1634484976530075,
      "learning_rate": 1.4750400252244511e-05,
      "loss": 0.4771401286125183,
      "step": 1267
    },
    {
      "epoch": 1.5261890427453342,
      "grad_norm": 0.17339955270290375,
      "learning_rate": 1.4679872391121651e-05,
      "loss": 0.5483553409576416,
      "step": 1268
    },
    {
      "epoch": 1.5273931366646598,
      "grad_norm": 0.16898883879184723,
      "learning_rate": 1.4609484523497696e-05,
      "loss": 0.6612141132354736,
      "step": 1269
    },
    {
      "epoch": 1.5285972305839857,
      "grad_norm": 0.1576375812292099,
      "learning_rate": 1.4539236928359318e-05,
      "loss": 0.6176823973655701,
      "step": 1270
    },
    {
      "epoch": 1.5298013245033113,
      "grad_norm": 0.18051175773143768,
      "learning_rate": 1.4469129884137177e-05,
      "loss": 0.6495242714881897,
      "step": 1271
    },
    {
      "epoch": 1.531005418422637,
      "grad_norm": 0.18644917011260986,
      "learning_rate": 1.4399163668704829e-05,
      "loss": 0.6112721562385559,
      "step": 1272
    },
    {
      "epoch": 1.5322095123419626,
      "grad_norm": 0.17363107204437256,
      "learning_rate": 1.4329338559377691e-05,
      "loss": 0.6350094079971313,
      "step": 1273
    },
    {
      "epoch": 1.5334136062612884,
      "grad_norm": 0.17279848456382751,
      "learning_rate": 1.4259654832911867e-05,
      "loss": 0.6157035231590271,
      "step": 1274
    },
    {
      "epoch": 1.534617700180614,
      "grad_norm": 0.16541044414043427,
      "learning_rate": 1.4190112765503117e-05,
      "loss": 0.5796827077865601,
      "step": 1275
    },
    {
      "epoch": 1.53582179409994,
      "grad_norm": 0.18048810958862305,
      "learning_rate": 1.412071263278571e-05,
      "loss": 0.6246907114982605,
      "step": 1276
    },
    {
      "epoch": 1.5370258880192655,
      "grad_norm": 0.18979230523109436,
      "learning_rate": 1.4051454709831308e-05,
      "loss": 0.6179551482200623,
      "step": 1277
    },
    {
      "epoch": 1.5382299819385912,
      "grad_norm": 0.1621571034193039,
      "learning_rate": 1.398233927114797e-05,
      "loss": 0.6340515613555908,
      "step": 1278
    },
    {
      "epoch": 1.5394340758579168,
      "grad_norm": 0.16793611645698547,
      "learning_rate": 1.3913366590678966e-05,
      "loss": 0.5309664011001587,
      "step": 1279
    },
    {
      "epoch": 1.5406381697772427,
      "grad_norm": 0.18016891181468964,
      "learning_rate": 1.38445369418018e-05,
      "loss": 0.6927658319473267,
      "step": 1280
    },
    {
      "epoch": 1.5418422636965683,
      "grad_norm": 0.1707635521888733,
      "learning_rate": 1.377585059732701e-05,
      "loss": 0.5956138372421265,
      "step": 1281
    },
    {
      "epoch": 1.5430463576158941,
      "grad_norm": 0.1465148776769638,
      "learning_rate": 1.370730782949713e-05,
      "loss": 0.5477831363677979,
      "step": 1282
    },
    {
      "epoch": 1.5442504515352198,
      "grad_norm": 0.16733048856258392,
      "learning_rate": 1.3638908909985621e-05,
      "loss": 0.5932857990264893,
      "step": 1283
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 0.16134332120418549,
      "learning_rate": 1.3570654109895853e-05,
      "loss": 0.5714342594146729,
      "step": 1284
    },
    {
      "epoch": 1.546658639373871,
      "grad_norm": 0.16964860260486603,
      "learning_rate": 1.3502543699759917e-05,
      "loss": 0.5700778365135193,
      "step": 1285
    },
    {
      "epoch": 1.547862733293197,
      "grad_norm": 0.13966421782970428,
      "learning_rate": 1.3434577949537592e-05,
      "loss": 0.5588657259941101,
      "step": 1286
    },
    {
      "epoch": 1.5490668272125225,
      "grad_norm": 0.17091353237628937,
      "learning_rate": 1.336675712861532e-05,
      "loss": 0.5641943216323853,
      "step": 1287
    },
    {
      "epoch": 1.5502709211318484,
      "grad_norm": 0.16251222789287567,
      "learning_rate": 1.3299081505805088e-05,
      "loss": 0.6409082412719727,
      "step": 1288
    },
    {
      "epoch": 1.551475015051174,
      "grad_norm": 0.16648387908935547,
      "learning_rate": 1.3231551349343413e-05,
      "loss": 0.49409541487693787,
      "step": 1289
    },
    {
      "epoch": 1.5526791089704997,
      "grad_norm": 0.1589309424161911,
      "learning_rate": 1.3164166926890226e-05,
      "loss": 0.5277682542800903,
      "step": 1290
    },
    {
      "epoch": 1.5538832028898253,
      "grad_norm": 0.18723368644714355,
      "learning_rate": 1.3096928505527811e-05,
      "loss": 0.5979494452476501,
      "step": 1291
    },
    {
      "epoch": 1.5550872968091511,
      "grad_norm": 0.16589628159999847,
      "learning_rate": 1.3029836351759795e-05,
      "loss": 0.6154754757881165,
      "step": 1292
    },
    {
      "epoch": 1.5562913907284768,
      "grad_norm": 0.2500206530094147,
      "learning_rate": 1.296289073151004e-05,
      "loss": 0.5692221522331238,
      "step": 1293
    },
    {
      "epoch": 1.5574954846478026,
      "grad_norm": 0.1658240109682083,
      "learning_rate": 1.2896091910121666e-05,
      "loss": 0.7039614915847778,
      "step": 1294
    },
    {
      "epoch": 1.5586995785671283,
      "grad_norm": 0.1783086359500885,
      "learning_rate": 1.2829440152355915e-05,
      "loss": 0.6289134621620178,
      "step": 1295
    },
    {
      "epoch": 1.559903672486454,
      "grad_norm": 0.1739896982908249,
      "learning_rate": 1.2762935722391089e-05,
      "loss": 0.612827479839325,
      "step": 1296
    },
    {
      "epoch": 1.5611077664057795,
      "grad_norm": 0.17161191999912262,
      "learning_rate": 1.2696578883821614e-05,
      "loss": 0.6256206035614014,
      "step": 1297
    },
    {
      "epoch": 1.5623118603251054,
      "grad_norm": 0.18163631856441498,
      "learning_rate": 1.263036989965688e-05,
      "loss": 0.7100282907485962,
      "step": 1298
    },
    {
      "epoch": 1.563515954244431,
      "grad_norm": 0.1552843153476715,
      "learning_rate": 1.2564309032320315e-05,
      "loss": 0.5413612127304077,
      "step": 1299
    },
    {
      "epoch": 1.5647200481637569,
      "grad_norm": 0.17517521977424622,
      "learning_rate": 1.2498396543648195e-05,
      "loss": 0.539400577545166,
      "step": 1300
    },
    {
      "epoch": 1.5659241420830825,
      "grad_norm": 0.17437493801116943,
      "learning_rate": 1.2432632694888724e-05,
      "loss": 0.5721178650856018,
      "step": 1301
    },
    {
      "epoch": 1.5671282360024081,
      "grad_norm": 0.1776028871536255,
      "learning_rate": 1.2367017746700977e-05,
      "loss": 0.5741942524909973,
      "step": 1302
    },
    {
      "epoch": 1.5683323299217338,
      "grad_norm": 0.17527516186237335,
      "learning_rate": 1.2301551959153813e-05,
      "loss": 0.58126300573349,
      "step": 1303
    },
    {
      "epoch": 1.5695364238410596,
      "grad_norm": 0.1610412746667862,
      "learning_rate": 1.223623559172496e-05,
      "loss": 0.5186402797698975,
      "step": 1304
    },
    {
      "epoch": 1.5707405177603853,
      "grad_norm": 0.1678943783044815,
      "learning_rate": 1.217106890329981e-05,
      "loss": 0.5822352170944214,
      "step": 1305
    },
    {
      "epoch": 1.5719446116797111,
      "grad_norm": 0.1677101105451584,
      "learning_rate": 1.2106052152170561e-05,
      "loss": 0.6485540866851807,
      "step": 1306
    },
    {
      "epoch": 1.5731487055990367,
      "grad_norm": 0.1607329100370407,
      "learning_rate": 1.204118559603511e-05,
      "loss": 0.5938149690628052,
      "step": 1307
    },
    {
      "epoch": 1.5743527995183624,
      "grad_norm": 0.18100596964359283,
      "learning_rate": 1.1976469491996028e-05,
      "loss": 0.679480254650116,
      "step": 1308
    },
    {
      "epoch": 1.575556893437688,
      "grad_norm": 0.18429264426231384,
      "learning_rate": 1.1911904096559589e-05,
      "loss": 0.5611011385917664,
      "step": 1309
    },
    {
      "epoch": 1.5767609873570139,
      "grad_norm": 0.1655278354883194,
      "learning_rate": 1.1847489665634709e-05,
      "loss": 0.5671539306640625,
      "step": 1310
    },
    {
      "epoch": 1.5779650812763397,
      "grad_norm": 0.18665340542793274,
      "learning_rate": 1.1783226454531942e-05,
      "loss": 0.5724003911018372,
      "step": 1311
    },
    {
      "epoch": 1.5791691751956654,
      "grad_norm": 0.15790997445583344,
      "learning_rate": 1.1719114717962476e-05,
      "loss": 0.5717609524726868,
      "step": 1312
    },
    {
      "epoch": 1.580373269114991,
      "grad_norm": 0.16969580948352814,
      "learning_rate": 1.1655154710037153e-05,
      "loss": 0.5796958208084106,
      "step": 1313
    },
    {
      "epoch": 1.5815773630343166,
      "grad_norm": 0.1928783804178238,
      "learning_rate": 1.1591346684265359e-05,
      "loss": 0.5869431495666504,
      "step": 1314
    },
    {
      "epoch": 1.5827814569536423,
      "grad_norm": 0.16961272060871124,
      "learning_rate": 1.1527690893554156e-05,
      "loss": 0.5898787975311279,
      "step": 1315
    },
    {
      "epoch": 1.583985550872968,
      "grad_norm": 0.17396080493927002,
      "learning_rate": 1.1464187590207171e-05,
      "loss": 0.6148181557655334,
      "step": 1316
    },
    {
      "epoch": 1.585189644792294,
      "grad_norm": 0.16814124584197998,
      "learning_rate": 1.1400837025923649e-05,
      "loss": 0.6253941059112549,
      "step": 1317
    },
    {
      "epoch": 1.5863937387116196,
      "grad_norm": 0.16475097835063934,
      "learning_rate": 1.1337639451797494e-05,
      "loss": 0.582787036895752,
      "step": 1318
    },
    {
      "epoch": 1.5875978326309452,
      "grad_norm": 0.19830529391765594,
      "learning_rate": 1.1274595118316134e-05,
      "loss": 0.6686637997627258,
      "step": 1319
    },
    {
      "epoch": 1.5888019265502709,
      "grad_norm": 0.17821630835533142,
      "learning_rate": 1.1211704275359675e-05,
      "loss": 0.6269924640655518,
      "step": 1320
    },
    {
      "epoch": 1.5900060204695965,
      "grad_norm": 0.18093682825565338,
      "learning_rate": 1.1148967172199848e-05,
      "loss": 0.5708060264587402,
      "step": 1321
    },
    {
      "epoch": 1.5912101143889223,
      "grad_norm": 0.16417568922042847,
      "learning_rate": 1.1086384057499011e-05,
      "loss": 0.5943193435668945,
      "step": 1322
    },
    {
      "epoch": 1.5924142083082482,
      "grad_norm": 0.19033674895763397,
      "learning_rate": 1.1023955179309203e-05,
      "loss": 0.6637415885925293,
      "step": 1323
    },
    {
      "epoch": 1.5936183022275738,
      "grad_norm": 0.17734594643115997,
      "learning_rate": 1.0961680785071116e-05,
      "loss": 0.5631223917007446,
      "step": 1324
    },
    {
      "epoch": 1.5948223961468995,
      "grad_norm": 0.16428886353969574,
      "learning_rate": 1.0899561121613145e-05,
      "loss": 0.5043269395828247,
      "step": 1325
    },
    {
      "epoch": 1.596026490066225,
      "grad_norm": 0.17552150785923004,
      "learning_rate": 1.0837596435150398e-05,
      "loss": 0.5466732382774353,
      "step": 1326
    },
    {
      "epoch": 1.5972305839855507,
      "grad_norm": 0.17015375196933746,
      "learning_rate": 1.0775786971283725e-05,
      "loss": 0.6014482378959656,
      "step": 1327
    },
    {
      "epoch": 1.5984346779048766,
      "grad_norm": 0.16729521751403809,
      "learning_rate": 1.0714132974998731e-05,
      "loss": 0.6966091990470886,
      "step": 1328
    },
    {
      "epoch": 1.5996387718242024,
      "grad_norm": 0.176364466547966,
      "learning_rate": 1.065263469066483e-05,
      "loss": 0.578704833984375,
      "step": 1329
    },
    {
      "epoch": 1.600842865743528,
      "grad_norm": 0.16971707344055176,
      "learning_rate": 1.0591292362034255e-05,
      "loss": 0.6056995391845703,
      "step": 1330
    },
    {
      "epoch": 1.6020469596628537,
      "grad_norm": 0.15724879503250122,
      "learning_rate": 1.0530106232241099e-05,
      "loss": 0.5634128451347351,
      "step": 1331
    },
    {
      "epoch": 1.6032510535821793,
      "grad_norm": 0.18894138932228088,
      "learning_rate": 1.0469076543800355e-05,
      "loss": 0.6138325929641724,
      "step": 1332
    },
    {
      "epoch": 1.604455147501505,
      "grad_norm": 0.20011907815933228,
      "learning_rate": 1.0408203538606948e-05,
      "loss": 0.5783783793449402,
      "step": 1333
    },
    {
      "epoch": 1.6056592414208308,
      "grad_norm": 0.17266902327537537,
      "learning_rate": 1.0347487457934768e-05,
      "loss": 0.5669416189193726,
      "step": 1334
    },
    {
      "epoch": 1.6068633353401567,
      "grad_norm": 0.17447730898857117,
      "learning_rate": 1.0286928542435754e-05,
      "loss": 0.4800940454006195,
      "step": 1335
    },
    {
      "epoch": 1.6080674292594823,
      "grad_norm": 0.1672305464744568,
      "learning_rate": 1.0226527032138878e-05,
      "loss": 0.6234422326087952,
      "step": 1336
    },
    {
      "epoch": 1.609271523178808,
      "grad_norm": 0.16313524544239044,
      "learning_rate": 1.0166283166449253e-05,
      "loss": 0.5977500081062317,
      "step": 1337
    },
    {
      "epoch": 1.6104756170981336,
      "grad_norm": 0.14919257164001465,
      "learning_rate": 1.0106197184147143e-05,
      "loss": 0.4973722994327545,
      "step": 1338
    },
    {
      "epoch": 1.6116797110174592,
      "grad_norm": 0.16079452633857727,
      "learning_rate": 1.0046269323387036e-05,
      "loss": 0.5992749929428101,
      "step": 1339
    },
    {
      "epoch": 1.612883804936785,
      "grad_norm": 0.17574235796928406,
      "learning_rate": 9.986499821696693e-06,
      "loss": 0.6141721606254578,
      "step": 1340
    },
    {
      "epoch": 1.614087898856111,
      "grad_norm": 0.17646431922912598,
      "learning_rate": 9.926888915976219e-06,
      "loss": 0.5338242053985596,
      "step": 1341
    },
    {
      "epoch": 1.6152919927754366,
      "grad_norm": 0.1584014743566513,
      "learning_rate": 9.867436842497103e-06,
      "loss": 0.5542763471603394,
      "step": 1342
    },
    {
      "epoch": 1.6164960866947622,
      "grad_norm": 0.16947726905345917,
      "learning_rate": 9.808143836901284e-06,
      "loss": 0.6113898158073425,
      "step": 1343
    },
    {
      "epoch": 1.6177001806140878,
      "grad_norm": 0.16287527978420258,
      "learning_rate": 9.74901013420026e-06,
      "loss": 0.5720081925392151,
      "step": 1344
    },
    {
      "epoch": 1.6189042745334135,
      "grad_norm": 0.1791754812002182,
      "learning_rate": 9.69003596877408e-06,
      "loss": 0.6065548062324524,
      "step": 1345
    },
    {
      "epoch": 1.6201083684527393,
      "grad_norm": 0.18168096244335175,
      "learning_rate": 9.631221574370491e-06,
      "loss": 0.5888320207595825,
      "step": 1346
    },
    {
      "epoch": 1.6213124623720652,
      "grad_norm": 0.16367362439632416,
      "learning_rate": 9.572567184103958e-06,
      "loss": 0.5836629867553711,
      "step": 1347
    },
    {
      "epoch": 1.6225165562913908,
      "grad_norm": 0.15913835167884827,
      "learning_rate": 9.514073030454762e-06,
      "loss": 0.6174272894859314,
      "step": 1348
    },
    {
      "epoch": 1.6237206502107164,
      "grad_norm": 0.17368106544017792,
      "learning_rate": 9.455739345268088e-06,
      "loss": 0.6067365407943726,
      "step": 1349
    },
    {
      "epoch": 1.624924744130042,
      "grad_norm": 0.1788904368877411,
      "learning_rate": 9.39756635975308e-06,
      "loss": 0.5914356708526611,
      "step": 1350
    },
    {
      "epoch": 1.6261288380493677,
      "grad_norm": 0.17372700572013855,
      "learning_rate": 9.339554304481951e-06,
      "loss": 0.5700035095214844,
      "step": 1351
    },
    {
      "epoch": 1.6273329319686936,
      "grad_norm": 0.17488062381744385,
      "learning_rate": 9.281703409389043e-06,
      "loss": 0.7085583806037903,
      "step": 1352
    },
    {
      "epoch": 1.6285370258880194,
      "grad_norm": 0.17074020206928253,
      "learning_rate": 9.224013903769946e-06,
      "loss": 0.5492441058158875,
      "step": 1353
    },
    {
      "epoch": 1.629741119807345,
      "grad_norm": 0.1883363276720047,
      "learning_rate": 9.166486016280562e-06,
      "loss": 0.6559622287750244,
      "step": 1354
    },
    {
      "epoch": 1.6309452137266707,
      "grad_norm": 0.16982145607471466,
      "learning_rate": 9.109119974936214e-06,
      "loss": 0.543138861656189,
      "step": 1355
    },
    {
      "epoch": 1.6321493076459963,
      "grad_norm": 0.20051883161067963,
      "learning_rate": 9.051916007110734e-06,
      "loss": 0.6901743412017822,
      "step": 1356
    },
    {
      "epoch": 1.633353401565322,
      "grad_norm": 0.17479044198989868,
      "learning_rate": 8.994874339535569e-06,
      "loss": 0.5976893305778503,
      "step": 1357
    },
    {
      "epoch": 1.6345574954846478,
      "grad_norm": 0.1779562085866928,
      "learning_rate": 8.937995198298876e-06,
      "loss": 0.6421647667884827,
      "step": 1358
    },
    {
      "epoch": 1.6357615894039736,
      "grad_norm": 0.18827801942825317,
      "learning_rate": 8.881278808844634e-06,
      "loss": 0.600217342376709,
      "step": 1359
    },
    {
      "epoch": 1.6369656833232993,
      "grad_norm": 0.17657113075256348,
      "learning_rate": 8.824725395971745e-06,
      "loss": 0.5228882431983948,
      "step": 1360
    },
    {
      "epoch": 1.638169777242625,
      "grad_norm": 0.16930431127548218,
      "learning_rate": 8.768335183833132e-06,
      "loss": 0.5633640885353088,
      "step": 1361
    },
    {
      "epoch": 1.6393738711619505,
      "grad_norm": 0.18676616251468658,
      "learning_rate": 8.712108395934881e-06,
      "loss": 0.604053795337677,
      "step": 1362
    },
    {
      "epoch": 1.6405779650812762,
      "grad_norm": 0.1758546233177185,
      "learning_rate": 8.656045255135314e-06,
      "loss": 0.6599376797676086,
      "step": 1363
    },
    {
      "epoch": 1.641782059000602,
      "grad_norm": 0.171214759349823,
      "learning_rate": 8.600145983644148e-06,
      "loss": 0.5752171277999878,
      "step": 1364
    },
    {
      "epoch": 1.642986152919928,
      "grad_norm": 0.17174966633319855,
      "learning_rate": 8.544410803021574e-06,
      "loss": 0.5965902805328369,
      "step": 1365
    },
    {
      "epoch": 1.6441902468392535,
      "grad_norm": 0.1809854954481125,
      "learning_rate": 8.488839934177422e-06,
      "loss": 0.5568514466285706,
      "step": 1366
    },
    {
      "epoch": 1.6453943407585792,
      "grad_norm": 0.1726883351802826,
      "learning_rate": 8.43343359737025e-06,
      "loss": 0.5705770254135132,
      "step": 1367
    },
    {
      "epoch": 1.6465984346779048,
      "grad_norm": 0.17551152408123016,
      "learning_rate": 8.37819201220647e-06,
      "loss": 0.6413697004318237,
      "step": 1368
    },
    {
      "epoch": 1.6478025285972304,
      "grad_norm": 0.1715322583913803,
      "learning_rate": 8.323115397639513e-06,
      "loss": 0.632416307926178,
      "step": 1369
    },
    {
      "epoch": 1.6490066225165563,
      "grad_norm": 0.16529041528701782,
      "learning_rate": 8.268203971968924e-06,
      "loss": 0.6020644903182983,
      "step": 1370
    },
    {
      "epoch": 1.6502107164358821,
      "grad_norm": 0.14709921181201935,
      "learning_rate": 8.213457952839527e-06,
      "loss": 0.5302895307540894,
      "step": 1371
    },
    {
      "epoch": 1.6514148103552078,
      "grad_norm": 0.16793395578861237,
      "learning_rate": 8.158877557240529e-06,
      "loss": 0.5995750427246094,
      "step": 1372
    },
    {
      "epoch": 1.6526189042745334,
      "grad_norm": 0.1651146113872528,
      "learning_rate": 8.104463001504697e-06,
      "loss": 0.6138071417808533,
      "step": 1373
    },
    {
      "epoch": 1.653822998193859,
      "grad_norm": 0.17747002840042114,
      "learning_rate": 8.05021450130744e-06,
      "loss": 0.6624958515167236,
      "step": 1374
    },
    {
      "epoch": 1.6550270921131849,
      "grad_norm": 0.17120714485645294,
      "learning_rate": 7.996132271666062e-06,
      "loss": 0.6298037767410278,
      "step": 1375
    },
    {
      "epoch": 1.6562311860325105,
      "grad_norm": 0.17241822183132172,
      "learning_rate": 7.942216526938795e-06,
      "loss": 0.6754534244537354,
      "step": 1376
    },
    {
      "epoch": 1.6574352799518364,
      "grad_norm": 0.1633012294769287,
      "learning_rate": 7.888467480824003e-06,
      "loss": 0.576261043548584,
      "step": 1377
    },
    {
      "epoch": 1.658639373871162,
      "grad_norm": 0.17508645355701447,
      "learning_rate": 7.83488534635936e-06,
      "loss": 0.5186232924461365,
      "step": 1378
    },
    {
      "epoch": 1.6598434677904876,
      "grad_norm": 0.1668899953365326,
      "learning_rate": 7.78147033592091e-06,
      "loss": 0.5184745788574219,
      "step": 1379
    },
    {
      "epoch": 1.6610475617098133,
      "grad_norm": 0.16185078024864197,
      "learning_rate": 7.728222661222372e-06,
      "loss": 0.5718276500701904,
      "step": 1380
    },
    {
      "epoch": 1.6622516556291391,
      "grad_norm": 0.17697584629058838,
      "learning_rate": 7.675142533314172e-06,
      "loss": 0.6737019419670105,
      "step": 1381
    },
    {
      "epoch": 1.6634557495484648,
      "grad_norm": 0.1830935925245285,
      "learning_rate": 7.622230162582656e-06,
      "loss": 0.5729782581329346,
      "step": 1382
    },
    {
      "epoch": 1.6646598434677906,
      "grad_norm": 0.1729450821876526,
      "learning_rate": 7.569485758749262e-06,
      "loss": 0.552348792552948,
      "step": 1383
    },
    {
      "epoch": 1.6658639373871162,
      "grad_norm": 0.1686101257801056,
      "learning_rate": 7.5169095308696865e-06,
      "loss": 0.5171111226081848,
      "step": 1384
    },
    {
      "epoch": 1.6670680313064419,
      "grad_norm": 0.17723223567008972,
      "learning_rate": 7.4645016873330366e-06,
      "loss": 0.5126849412918091,
      "step": 1385
    },
    {
      "epoch": 1.6682721252257675,
      "grad_norm": 0.17451541125774384,
      "learning_rate": 7.412262435861023e-06,
      "loss": 0.611954927444458,
      "step": 1386
    },
    {
      "epoch": 1.6694762191450934,
      "grad_norm": 0.15248830616474152,
      "learning_rate": 7.360191983507153e-06,
      "loss": 0.5664346814155579,
      "step": 1387
    },
    {
      "epoch": 1.670680313064419,
      "grad_norm": 0.16013115644454956,
      "learning_rate": 7.308290536655832e-06,
      "loss": 0.5460264086723328,
      "step": 1388
    },
    {
      "epoch": 1.6718844069837449,
      "grad_norm": 0.17455200850963593,
      "learning_rate": 7.256558301021665e-06,
      "loss": 0.6276478171348572,
      "step": 1389
    },
    {
      "epoch": 1.6730885009030705,
      "grad_norm": 0.1620652973651886,
      "learning_rate": 7.2049954816485465e-06,
      "loss": 0.5149453282356262,
      "step": 1390
    },
    {
      "epoch": 1.6742925948223961,
      "grad_norm": 0.1981610357761383,
      "learning_rate": 7.153602282908877e-06,
      "loss": 0.6605783700942993,
      "step": 1391
    },
    {
      "epoch": 1.6754966887417218,
      "grad_norm": 0.1599699854850769,
      "learning_rate": 7.102378908502783e-06,
      "loss": 0.5460922718048096,
      "step": 1392
    },
    {
      "epoch": 1.6767007826610476,
      "grad_norm": 0.17770592868328094,
      "learning_rate": 7.051325561457217e-06,
      "loss": 0.5630290508270264,
      "step": 1393
    },
    {
      "epoch": 1.6779048765803732,
      "grad_norm": 0.16870608925819397,
      "learning_rate": 7.000442444125299e-06,
      "loss": 0.5493181943893433,
      "step": 1394
    },
    {
      "epoch": 1.679108970499699,
      "grad_norm": 0.1710635870695114,
      "learning_rate": 6.949729758185386e-06,
      "loss": 0.528740406036377,
      "step": 1395
    },
    {
      "epoch": 1.6803130644190247,
      "grad_norm": 0.1684509664773941,
      "learning_rate": 6.899187704640325e-06,
      "loss": 0.5733117461204529,
      "step": 1396
    },
    {
      "epoch": 1.6815171583383504,
      "grad_norm": 0.17383374273777008,
      "learning_rate": 6.848816483816672e-06,
      "loss": 0.5148810148239136,
      "step": 1397
    },
    {
      "epoch": 1.682721252257676,
      "grad_norm": 0.18023723363876343,
      "learning_rate": 6.798616295363819e-06,
      "loss": 0.4807356297969818,
      "step": 1398
    },
    {
      "epoch": 1.6839253461770018,
      "grad_norm": 0.21825659275054932,
      "learning_rate": 6.748587338253337e-06,
      "loss": 0.6548990607261658,
      "step": 1399
    },
    {
      "epoch": 1.6851294400963275,
      "grad_norm": 0.18581441044807434,
      "learning_rate": 6.698729810778065e-06,
      "loss": 0.6097226738929749,
      "step": 1400
    },
    {
      "epoch": 1.6863335340156533,
      "grad_norm": 0.1781386137008667,
      "learning_rate": 6.649043910551394e-06,
      "loss": 0.5588247179985046,
      "step": 1401
    },
    {
      "epoch": 1.687537627934979,
      "grad_norm": 0.17311859130859375,
      "learning_rate": 6.59952983450643e-06,
      "loss": 0.5745834112167358,
      "step": 1402
    },
    {
      "epoch": 1.6887417218543046,
      "grad_norm": 0.1612737774848938,
      "learning_rate": 6.550187778895245e-06,
      "loss": 0.5151218175888062,
      "step": 1403
    },
    {
      "epoch": 1.6899458157736302,
      "grad_norm": 0.1691303700208664,
      "learning_rate": 6.501017939288145e-06,
      "loss": 0.5005168318748474,
      "step": 1404
    },
    {
      "epoch": 1.691149909692956,
      "grad_norm": 0.17320071160793304,
      "learning_rate": 6.452020510572798e-06,
      "loss": 0.6099838614463806,
      "step": 1405
    },
    {
      "epoch": 1.6923540036122817,
      "grad_norm": 0.16700418293476105,
      "learning_rate": 6.4031956869535446e-06,
      "loss": 0.5465252995491028,
      "step": 1406
    },
    {
      "epoch": 1.6935580975316076,
      "grad_norm": 0.18560250103473663,
      "learning_rate": 6.354543661950552e-06,
      "loss": 0.6228649616241455,
      "step": 1407
    },
    {
      "epoch": 1.6947621914509332,
      "grad_norm": 0.1689670830965042,
      "learning_rate": 6.3060646283991106e-06,
      "loss": 0.4893586039543152,
      "step": 1408
    },
    {
      "epoch": 1.6959662853702588,
      "grad_norm": 0.20757900178432465,
      "learning_rate": 6.257758778448886e-06,
      "loss": 0.6186881065368652,
      "step": 1409
    },
    {
      "epoch": 1.6971703792895845,
      "grad_norm": 0.17712172865867615,
      "learning_rate": 6.20962630356306e-06,
      "loss": 0.5862473845481873,
      "step": 1410
    },
    {
      "epoch": 1.6983744732089103,
      "grad_norm": 0.18376494944095612,
      "learning_rate": 6.1616673945176836e-06,
      "loss": 0.6037924289703369,
      "step": 1411
    },
    {
      "epoch": 1.699578567128236,
      "grad_norm": 0.17721505463123322,
      "learning_rate": 6.113882241400804e-06,
      "loss": 0.5376719236373901,
      "step": 1412
    },
    {
      "epoch": 1.7007826610475618,
      "grad_norm": 0.16330905258655548,
      "learning_rate": 6.0662710336118425e-06,
      "loss": 0.5823795795440674,
      "step": 1413
    },
    {
      "epoch": 1.7019867549668874,
      "grad_norm": 0.16761308908462524,
      "learning_rate": 6.018833959860753e-06,
      "loss": 0.5702024698257446,
      "step": 1414
    },
    {
      "epoch": 1.703190848886213,
      "grad_norm": 0.15750226378440857,
      "learning_rate": 5.971571208167298e-06,
      "loss": 0.6319963932037354,
      "step": 1415
    },
    {
      "epoch": 1.7043949428055387,
      "grad_norm": 0.18519818782806396,
      "learning_rate": 5.9244829658602845e-06,
      "loss": 0.49184101819992065,
      "step": 1416
    },
    {
      "epoch": 1.7055990367248646,
      "grad_norm": 0.17171220481395721,
      "learning_rate": 5.87756941957685e-06,
      "loss": 0.6053203344345093,
      "step": 1417
    },
    {
      "epoch": 1.7068031306441902,
      "grad_norm": 0.1618029624223709,
      "learning_rate": 5.830830755261751e-06,
      "loss": 0.562938928604126,
      "step": 1418
    },
    {
      "epoch": 1.708007224563516,
      "grad_norm": 0.17781466245651245,
      "learning_rate": 5.784267158166551e-06,
      "loss": 0.6935231685638428,
      "step": 1419
    },
    {
      "epoch": 1.7092113184828417,
      "grad_norm": 0.19096331298351288,
      "learning_rate": 5.737878812848929e-06,
      "loss": 0.6321725249290466,
      "step": 1420
    },
    {
      "epoch": 1.7104154124021673,
      "grad_norm": 0.18068662285804749,
      "learning_rate": 5.691665903171939e-06,
      "loss": 0.6156484484672546,
      "step": 1421
    },
    {
      "epoch": 1.711619506321493,
      "grad_norm": 0.17065894603729248,
      "learning_rate": 5.6456286123032884e-06,
      "loss": 0.5775973200798035,
      "step": 1422
    },
    {
      "epoch": 1.7128236002408188,
      "grad_norm": 0.17004157602787018,
      "learning_rate": 5.599767122714627e-06,
      "loss": 0.5303872227668762,
      "step": 1423
    },
    {
      "epoch": 1.7140276941601444,
      "grad_norm": 0.19186696410179138,
      "learning_rate": 5.554081616180801e-06,
      "loss": 0.48998844623565674,
      "step": 1424
    },
    {
      "epoch": 1.7152317880794703,
      "grad_norm": 0.16569344699382782,
      "learning_rate": 5.508572273779089e-06,
      "loss": 0.6217376589775085,
      "step": 1425
    },
    {
      "epoch": 1.716435881998796,
      "grad_norm": 0.17863419651985168,
      "learning_rate": 5.4632392758885985e-06,
      "loss": 0.5908452272415161,
      "step": 1426
    },
    {
      "epoch": 1.7176399759181216,
      "grad_norm": 0.20405490696430206,
      "learning_rate": 5.418082802189434e-06,
      "loss": 0.5521903038024902,
      "step": 1427
    },
    {
      "epoch": 1.7188440698374472,
      "grad_norm": 0.16790297627449036,
      "learning_rate": 5.37310303166208e-06,
      "loss": 0.529878556728363,
      "step": 1428
    },
    {
      "epoch": 1.720048163756773,
      "grad_norm": 0.16328345239162445,
      "learning_rate": 5.328300142586629e-06,
      "loss": 0.5641664862632751,
      "step": 1429
    },
    {
      "epoch": 1.7212522576760987,
      "grad_norm": 0.19397473335266113,
      "learning_rate": 5.283674312542064e-06,
      "loss": 0.6154850721359253,
      "step": 1430
    },
    {
      "epoch": 1.7224563515954245,
      "grad_norm": 0.16560742259025574,
      "learning_rate": 5.239225718405616e-06,
      "loss": 0.5652956962585449,
      "step": 1431
    },
    {
      "epoch": 1.7236604455147502,
      "grad_norm": 0.1641513556241989,
      "learning_rate": 5.194954536352021e-06,
      "loss": 0.5688132047653198,
      "step": 1432
    },
    {
      "epoch": 1.7248645394340758,
      "grad_norm": 0.17349234223365784,
      "learning_rate": 5.150860941852842e-06,
      "loss": 0.5766156315803528,
      "step": 1433
    },
    {
      "epoch": 1.7260686333534014,
      "grad_norm": 0.1737147718667984,
      "learning_rate": 5.106945109675759e-06,
      "loss": 0.6204696893692017,
      "step": 1434
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 0.18704469501972198,
      "learning_rate": 5.0632072138838584e-06,
      "loss": 0.6103050708770752,
      "step": 1435
    },
    {
      "epoch": 1.728476821192053,
      "grad_norm": 0.17506591975688934,
      "learning_rate": 5.019647427834978e-06,
      "loss": 0.6423869132995605,
      "step": 1436
    },
    {
      "epoch": 1.7296809151113788,
      "grad_norm": 0.17806795239448547,
      "learning_rate": 4.9762659241810285e-06,
      "loss": 0.6385177373886108,
      "step": 1437
    },
    {
      "epoch": 1.7308850090307044,
      "grad_norm": 0.18682622909545898,
      "learning_rate": 4.933062874867267e-06,
      "loss": 0.6592196822166443,
      "step": 1438
    },
    {
      "epoch": 1.73208910295003,
      "grad_norm": 0.15755236148834229,
      "learning_rate": 4.8900384511316234e-06,
      "loss": 0.6020253300666809,
      "step": 1439
    },
    {
      "epoch": 1.7332931968693557,
      "grad_norm": 0.16945621371269226,
      "learning_rate": 4.847192823504054e-06,
      "loss": 0.5768476724624634,
      "step": 1440
    },
    {
      "epoch": 1.7344972907886815,
      "grad_norm": 0.19501367211341858,
      "learning_rate": 4.804526161805833e-06,
      "loss": 0.6353685855865479,
      "step": 1441
    },
    {
      "epoch": 1.7357013847080074,
      "grad_norm": 0.16163913905620575,
      "learning_rate": 4.7620386351489185e-06,
      "loss": 0.5788745284080505,
      "step": 1442
    },
    {
      "epoch": 1.736905478627333,
      "grad_norm": 0.1651899367570877,
      "learning_rate": 4.7197304119352355e-06,
      "loss": 0.6376733779907227,
      "step": 1443
    },
    {
      "epoch": 1.7381095725466587,
      "grad_norm": 0.17699390649795532,
      "learning_rate": 4.6776016598560124e-06,
      "loss": 0.5289449095726013,
      "step": 1444
    },
    {
      "epoch": 1.7393136664659843,
      "grad_norm": 0.18723583221435547,
      "learning_rate": 4.635652545891156e-06,
      "loss": 0.6893998980522156,
      "step": 1445
    },
    {
      "epoch": 1.74051776038531,
      "grad_norm": 0.189725860953331,
      "learning_rate": 4.5938832363085495e-06,
      "loss": 0.7069261074066162,
      "step": 1446
    },
    {
      "epoch": 1.7417218543046358,
      "grad_norm": 0.17144793272018433,
      "learning_rate": 4.552293896663451e-06,
      "loss": 0.5559759140014648,
      "step": 1447
    },
    {
      "epoch": 1.7429259482239616,
      "grad_norm": 0.17889682948589325,
      "learning_rate": 4.510884691797751e-06,
      "loss": 0.5811923146247864,
      "step": 1448
    },
    {
      "epoch": 1.7441300421432873,
      "grad_norm": 0.1701122224330902,
      "learning_rate": 4.469655785839377e-06,
      "loss": 0.5842764377593994,
      "step": 1449
    },
    {
      "epoch": 1.745334136062613,
      "grad_norm": 0.17281197011470795,
      "learning_rate": 4.428607342201635e-06,
      "loss": 0.522236704826355,
      "step": 1450
    },
    {
      "epoch": 1.7465382299819385,
      "grad_norm": 0.16742025315761566,
      "learning_rate": 4.387739523582551e-06,
      "loss": 0.5697470307350159,
      "step": 1451
    },
    {
      "epoch": 1.7477423239012642,
      "grad_norm": 0.17915286123752594,
      "learning_rate": 4.347052491964265e-06,
      "loss": 0.5438842177391052,
      "step": 1452
    },
    {
      "epoch": 1.74894641782059,
      "grad_norm": 0.18038634955883026,
      "learning_rate": 4.306546408612306e-06,
      "loss": 0.5878657698631287,
      "step": 1453
    },
    {
      "epoch": 1.7501505117399159,
      "grad_norm": 0.1685640811920166,
      "learning_rate": 4.26622143407503e-06,
      "loss": 0.5175699591636658,
      "step": 1454
    },
    {
      "epoch": 1.7513546056592415,
      "grad_norm": 0.15979155898094177,
      "learning_rate": 4.226077728182959e-06,
      "loss": 0.5842319130897522,
      "step": 1455
    },
    {
      "epoch": 1.7525586995785671,
      "grad_norm": 0.18487407267093658,
      "learning_rate": 4.186115450048128e-06,
      "loss": 0.5384514331817627,
      "step": 1456
    },
    {
      "epoch": 1.7537627934978928,
      "grad_norm": 0.1843617856502533,
      "learning_rate": 4.1463347580635195e-06,
      "loss": 0.4496367275714874,
      "step": 1457
    },
    {
      "epoch": 1.7549668874172184,
      "grad_norm": 0.17069736123085022,
      "learning_rate": 4.106735809902324e-06,
      "loss": 0.5590825080871582,
      "step": 1458
    },
    {
      "epoch": 1.7561709813365443,
      "grad_norm": 0.19505466520786285,
      "learning_rate": 4.0673187625174195e-06,
      "loss": 0.670661211013794,
      "step": 1459
    },
    {
      "epoch": 1.7573750752558701,
      "grad_norm": 0.17665988206863403,
      "learning_rate": 4.028083772140689e-06,
      "loss": 0.6377481818199158,
      "step": 1460
    },
    {
      "epoch": 1.7585791691751957,
      "grad_norm": 0.16847488284111023,
      "learning_rate": 3.989030994282434e-06,
      "loss": 0.5547840595245361,
      "step": 1461
    },
    {
      "epoch": 1.7597832630945214,
      "grad_norm": 0.18270504474639893,
      "learning_rate": 3.950160583730761e-06,
      "loss": 0.5961182117462158,
      "step": 1462
    },
    {
      "epoch": 1.760987357013847,
      "grad_norm": 0.16627801954746246,
      "learning_rate": 3.911472694550916e-06,
      "loss": 0.5199828743934631,
      "step": 1463
    },
    {
      "epoch": 1.7621914509331726,
      "grad_norm": 0.17543698847293854,
      "learning_rate": 3.872967480084727e-06,
      "loss": 0.5772284269332886,
      "step": 1464
    },
    {
      "epoch": 1.7633955448524985,
      "grad_norm": 0.19409753382205963,
      "learning_rate": 3.834645092949973e-06,
      "loss": 0.7199985980987549,
      "step": 1465
    },
    {
      "epoch": 1.7645996387718244,
      "grad_norm": 0.19312088191509247,
      "learning_rate": 3.796505685039825e-06,
      "loss": 0.7305468916893005,
      "step": 1466
    },
    {
      "epoch": 1.76580373269115,
      "grad_norm": 0.16722188889980316,
      "learning_rate": 3.758549407522144e-06,
      "loss": 0.5545473694801331,
      "step": 1467
    },
    {
      "epoch": 1.7670078266104756,
      "grad_norm": 0.1744573563337326,
      "learning_rate": 3.720776410838983e-06,
      "loss": 0.6022785305976868,
      "step": 1468
    },
    {
      "epoch": 1.7682119205298013,
      "grad_norm": 0.1630740761756897,
      "learning_rate": 3.6831868447059324e-06,
      "loss": 0.5020629167556763,
      "step": 1469
    },
    {
      "epoch": 1.7694160144491269,
      "grad_norm": 0.1899329125881195,
      "learning_rate": 3.645780858111547e-06,
      "loss": 0.6314529776573181,
      "step": 1470
    },
    {
      "epoch": 1.7706201083684527,
      "grad_norm": 0.16368846595287323,
      "learning_rate": 3.6085585993167805e-06,
      "loss": 0.5994744300842285,
      "step": 1471
    },
    {
      "epoch": 1.7718242022877786,
      "grad_norm": 0.18946246802806854,
      "learning_rate": 3.5715202158543125e-06,
      "loss": 0.6071475744247437,
      "step": 1472
    },
    {
      "epoch": 1.7730282962071042,
      "grad_norm": 0.19678771495819092,
      "learning_rate": 3.5346658545280795e-06,
      "loss": 0.6201621890068054,
      "step": 1473
    },
    {
      "epoch": 1.7742323901264299,
      "grad_norm": 0.1572706550359726,
      "learning_rate": 3.4979956614125953e-06,
      "loss": 0.574611485004425,
      "step": 1474
    },
    {
      "epoch": 1.7754364840457555,
      "grad_norm": 0.1965961456298828,
      "learning_rate": 3.4615097818524235e-06,
      "loss": 0.6616047024726868,
      "step": 1475
    },
    {
      "epoch": 1.7766405779650811,
      "grad_norm": 0.17455288767814636,
      "learning_rate": 3.4252083604616182e-06,
      "loss": 0.6152704954147339,
      "step": 1476
    },
    {
      "epoch": 1.777844671884407,
      "grad_norm": 0.19565220177173615,
      "learning_rate": 3.389091541123074e-06,
      "loss": 0.7469292283058167,
      "step": 1477
    },
    {
      "epoch": 1.7790487658037328,
      "grad_norm": 0.17200477421283722,
      "learning_rate": 3.353159466988032e-06,
      "loss": 0.5917014479637146,
      "step": 1478
    },
    {
      "epoch": 1.7802528597230585,
      "grad_norm": 0.1737939417362213,
      "learning_rate": 3.3174122804754738e-06,
      "loss": 0.5610493421554565,
      "step": 1479
    },
    {
      "epoch": 1.781456953642384,
      "grad_norm": 0.15580341219902039,
      "learning_rate": 3.2818501232715794e-06,
      "loss": 0.5798150897026062,
      "step": 1480
    },
    {
      "epoch": 1.7826610475617097,
      "grad_norm": 0.16556191444396973,
      "learning_rate": 3.246473136329148e-06,
      "loss": 0.6512298583984375,
      "step": 1481
    },
    {
      "epoch": 1.7838651414810354,
      "grad_norm": 0.16518540680408478,
      "learning_rate": 3.211281459867038e-06,
      "loss": 0.5852155685424805,
      "step": 1482
    },
    {
      "epoch": 1.7850692354003612,
      "grad_norm": 0.17119024693965912,
      "learning_rate": 3.1762752333696297e-06,
      "loss": 0.5942105054855347,
      "step": 1483
    },
    {
      "epoch": 1.786273329319687,
      "grad_norm": 0.18802329897880554,
      "learning_rate": 3.141454595586252e-06,
      "loss": 0.649821400642395,
      "step": 1484
    },
    {
      "epoch": 1.7874774232390127,
      "grad_norm": 0.16373737156391144,
      "learning_rate": 3.1068196845306487e-06,
      "loss": 0.5035237073898315,
      "step": 1485
    },
    {
      "epoch": 1.7886815171583383,
      "grad_norm": 0.16998307406902313,
      "learning_rate": 3.072370637480415e-06,
      "loss": 0.6510118842124939,
      "step": 1486
    },
    {
      "epoch": 1.789885611077664,
      "grad_norm": 0.18051420152187347,
      "learning_rate": 3.0381075909764744e-06,
      "loss": 0.64877849817276,
      "step": 1487
    },
    {
      "epoch": 1.7910897049969896,
      "grad_norm": 0.16519369184970856,
      "learning_rate": 3.004030680822517e-06,
      "loss": 0.6371068954467773,
      "step": 1488
    },
    {
      "epoch": 1.7922937989163155,
      "grad_norm": 0.17515403032302856,
      "learning_rate": 2.9701400420844737e-06,
      "loss": 0.5964099168777466,
      "step": 1489
    },
    {
      "epoch": 1.7934978928356413,
      "grad_norm": 0.18329456448554993,
      "learning_rate": 2.9364358090899766e-06,
      "loss": 0.5721734762191772,
      "step": 1490
    },
    {
      "epoch": 1.794701986754967,
      "grad_norm": 0.15678924322128296,
      "learning_rate": 2.9029181154278274e-06,
      "loss": 0.5435581207275391,
      "step": 1491
    },
    {
      "epoch": 1.7959060806742926,
      "grad_norm": 0.1719670593738556,
      "learning_rate": 2.8695870939474624e-06,
      "loss": 0.6054028272628784,
      "step": 1492
    },
    {
      "epoch": 1.7971101745936182,
      "grad_norm": 0.18294009566307068,
      "learning_rate": 2.836442876758438e-06,
      "loss": 0.6650201678276062,
      "step": 1493
    },
    {
      "epoch": 1.7983142685129438,
      "grad_norm": 0.16223953664302826,
      "learning_rate": 2.8034855952299045e-06,
      "loss": 0.5323062539100647,
      "step": 1494
    },
    {
      "epoch": 1.7995183624322697,
      "grad_norm": 0.16355708241462708,
      "learning_rate": 2.770715379990069e-06,
      "loss": 0.6287299394607544,
      "step": 1495
    },
    {
      "epoch": 1.8007224563515956,
      "grad_norm": 0.1636963039636612,
      "learning_rate": 2.7381323609256937e-06,
      "loss": 0.549708366394043,
      "step": 1496
    },
    {
      "epoch": 1.8019265502709212,
      "grad_norm": 0.19313931465148926,
      "learning_rate": 2.7057366671815856e-06,
      "loss": 0.6183158755302429,
      "step": 1497
    },
    {
      "epoch": 1.8031306441902468,
      "grad_norm": 0.17596328258514404,
      "learning_rate": 2.6735284271600657e-06,
      "loss": 0.6525415182113647,
      "step": 1498
    },
    {
      "epoch": 1.8043347381095725,
      "grad_norm": 0.1715078502893448,
      "learning_rate": 2.641507768520479e-06,
      "loss": 0.5351325273513794,
      "step": 1499
    },
    {
      "epoch": 1.805538832028898,
      "grad_norm": 0.18794047832489014,
      "learning_rate": 2.6096748181786758e-06,
      "loss": 0.5899437069892883,
      "step": 1500
    },
    {
      "epoch": 1.806742925948224,
      "grad_norm": 0.1812296360731125,
      "learning_rate": 2.5780297023065057e-06,
      "loss": 0.6455264091491699,
      "step": 1501
    },
    {
      "epoch": 1.8079470198675498,
      "grad_norm": 0.1602133959531784,
      "learning_rate": 2.546572546331338e-06,
      "loss": 0.5433736443519592,
      "step": 1502
    },
    {
      "epoch": 1.8091511137868754,
      "grad_norm": 0.17856872081756592,
      "learning_rate": 2.5153034749355487e-06,
      "loss": 0.585578978061676,
      "step": 1503
    },
    {
      "epoch": 1.810355207706201,
      "grad_norm": 0.17193603515625,
      "learning_rate": 2.4842226120560255e-06,
      "loss": 0.54914790391922,
      "step": 1504
    },
    {
      "epoch": 1.8115593016255267,
      "grad_norm": 0.17232759296894073,
      "learning_rate": 2.4533300808836757e-06,
      "loss": 0.6378284692764282,
      "step": 1505
    },
    {
      "epoch": 1.8127633955448526,
      "grad_norm": 0.17345201969146729,
      "learning_rate": 2.4226260038629545e-06,
      "loss": 0.5778675079345703,
      "step": 1506
    },
    {
      "epoch": 1.8139674894641782,
      "grad_norm": 0.165665864944458,
      "learning_rate": 2.3921105026913527e-06,
      "loss": 0.5345752835273743,
      "step": 1507
    },
    {
      "epoch": 1.815171583383504,
      "grad_norm": 0.16123533248901367,
      "learning_rate": 2.3617836983189366e-06,
      "loss": 0.5712841749191284,
      "step": 1508
    },
    {
      "epoch": 1.8163756773028297,
      "grad_norm": 0.16117937862873077,
      "learning_rate": 2.3316457109478716e-06,
      "loss": 0.48529744148254395,
      "step": 1509
    },
    {
      "epoch": 1.8175797712221553,
      "grad_norm": 0.16783325374126434,
      "learning_rate": 2.3016966600319154e-06,
      "loss": 0.6714291572570801,
      "step": 1510
    },
    {
      "epoch": 1.818783865141481,
      "grad_norm": 0.18452955782413483,
      "learning_rate": 2.2719366642759754e-06,
      "loss": 0.6201899647712708,
      "step": 1511
    },
    {
      "epoch": 1.8199879590608068,
      "grad_norm": 0.17982904613018036,
      "learning_rate": 2.2423658416356296e-06,
      "loss": 0.5961569547653198,
      "step": 1512
    },
    {
      "epoch": 1.8211920529801324,
      "grad_norm": 0.20002754032611847,
      "learning_rate": 2.212984309316646e-06,
      "loss": 0.6063302159309387,
      "step": 1513
    },
    {
      "epoch": 1.8223961468994583,
      "grad_norm": 0.16794919967651367,
      "learning_rate": 2.183792183774541e-06,
      "loss": 0.5876860618591309,
      "step": 1514
    },
    {
      "epoch": 1.823600240818784,
      "grad_norm": 0.1582144796848297,
      "learning_rate": 2.1547895807141004e-06,
      "loss": 0.5998432636260986,
      "step": 1515
    },
    {
      "epoch": 1.8248043347381095,
      "grad_norm": 0.16906720399856567,
      "learning_rate": 2.125976615088926e-06,
      "loss": 0.6220269203186035,
      "step": 1516
    },
    {
      "epoch": 1.8260084286574352,
      "grad_norm": 0.15978339314460754,
      "learning_rate": 2.0973534011009823e-06,
      "loss": 0.5583670735359192,
      "step": 1517
    },
    {
      "epoch": 1.827212522576761,
      "grad_norm": 0.17373648285865784,
      "learning_rate": 2.0689200522001294e-06,
      "loss": 0.5534298419952393,
      "step": 1518
    },
    {
      "epoch": 1.8284166164960867,
      "grad_norm": 0.17821577191352844,
      "learning_rate": 2.040676681083703e-06,
      "loss": 0.5730056762695312,
      "step": 1519
    },
    {
      "epoch": 1.8296207104154125,
      "grad_norm": 0.16129855811595917,
      "learning_rate": 2.01262339969604e-06,
      "loss": 0.5167095065116882,
      "step": 1520
    },
    {
      "epoch": 1.8308248043347382,
      "grad_norm": 0.1761590540409088,
      "learning_rate": 1.9847603192280515e-06,
      "loss": 0.638519287109375,
      "step": 1521
    },
    {
      "epoch": 1.8320288982540638,
      "grad_norm": 0.15942449867725372,
      "learning_rate": 1.957087550116765e-06,
      "loss": 0.5271167159080505,
      "step": 1522
    },
    {
      "epoch": 1.8332329921733894,
      "grad_norm": 0.1646331399679184,
      "learning_rate": 1.929605202044904e-06,
      "loss": 0.574680507183075,
      "step": 1523
    },
    {
      "epoch": 1.8344370860927153,
      "grad_norm": 0.19252583384513855,
      "learning_rate": 1.9023133839404517e-06,
      "loss": 0.5918468832969666,
      "step": 1524
    },
    {
      "epoch": 1.835641180012041,
      "grad_norm": 0.15829242765903473,
      "learning_rate": 1.875212203976201e-06,
      "loss": 0.5070363283157349,
      "step": 1525
    },
    {
      "epoch": 1.8368452739313668,
      "grad_norm": 0.1709422916173935,
      "learning_rate": 1.8483017695693494e-06,
      "loss": 0.6618005037307739,
      "step": 1526
    },
    {
      "epoch": 1.8380493678506924,
      "grad_norm": 0.16142213344573975,
      "learning_rate": 1.8215821873810601e-06,
      "loss": 0.5864096283912659,
      "step": 1527
    },
    {
      "epoch": 1.839253461770018,
      "grad_norm": 0.1600547432899475,
      "learning_rate": 1.7950535633160403e-06,
      "loss": 0.5859323740005493,
      "step": 1528
    },
    {
      "epoch": 1.8404575556893437,
      "grad_norm": 0.20026294887065887,
      "learning_rate": 1.768716002522125e-06,
      "loss": 0.6432352662086487,
      "step": 1529
    },
    {
      "epoch": 1.8416616496086695,
      "grad_norm": 0.17253966629505157,
      "learning_rate": 1.7425696093898548e-06,
      "loss": 0.5920822024345398,
      "step": 1530
    },
    {
      "epoch": 1.8428657435279951,
      "grad_norm": 0.16520057618618011,
      "learning_rate": 1.7166144875520763e-06,
      "loss": 0.6401042938232422,
      "step": 1531
    },
    {
      "epoch": 1.844069837447321,
      "grad_norm": 0.17522233724594116,
      "learning_rate": 1.6908507398834927e-06,
      "loss": 0.5717169046401978,
      "step": 1532
    },
    {
      "epoch": 1.8452739313666466,
      "grad_norm": 0.18461531400680542,
      "learning_rate": 1.6652784685003197e-06,
      "loss": 0.6172612905502319,
      "step": 1533
    },
    {
      "epoch": 1.8464780252859723,
      "grad_norm": 0.16957464814186096,
      "learning_rate": 1.6398977747598243e-06,
      "loss": 0.5045825242996216,
      "step": 1534
    },
    {
      "epoch": 1.847682119205298,
      "grad_norm": 0.155729278922081,
      "learning_rate": 1.614708759259942e-06,
      "loss": 0.5045057535171509,
      "step": 1535
    },
    {
      "epoch": 1.8488862131246238,
      "grad_norm": 0.17569220066070557,
      "learning_rate": 1.5897115218388936e-06,
      "loss": 0.567057192325592,
      "step": 1536
    },
    {
      "epoch": 1.8500903070439494,
      "grad_norm": 0.15697702765464783,
      "learning_rate": 1.564906161574764e-06,
      "loss": 0.5586735010147095,
      "step": 1537
    },
    {
      "epoch": 1.8512944009632752,
      "grad_norm": 0.18077059090137482,
      "learning_rate": 1.5402927767851239e-06,
      "loss": 0.5339556336402893,
      "step": 1538
    },
    {
      "epoch": 1.8524984948826009,
      "grad_norm": 0.15517669916152954,
      "learning_rate": 1.5158714650266414e-06,
      "loss": 0.5625603795051575,
      "step": 1539
    },
    {
      "epoch": 1.8537025888019265,
      "grad_norm": 0.18605338037014008,
      "learning_rate": 1.4916423230946885e-06,
      "loss": 0.6815199255943298,
      "step": 1540
    },
    {
      "epoch": 1.8549066827212521,
      "grad_norm": 0.16735219955444336,
      "learning_rate": 1.4676054470229517e-06,
      "loss": 0.5845962762832642,
      "step": 1541
    },
    {
      "epoch": 1.856110776640578,
      "grad_norm": 0.17093312740325928,
      "learning_rate": 1.443760932083077e-06,
      "loss": 0.5695465803146362,
      "step": 1542
    },
    {
      "epoch": 1.8573148705599036,
      "grad_norm": 0.1818591058254242,
      "learning_rate": 1.4201088727842648e-06,
      "loss": 0.5898520946502686,
      "step": 1543
    },
    {
      "epoch": 1.8585189644792295,
      "grad_norm": 0.18692384660243988,
      "learning_rate": 1.3966493628729039e-06,
      "loss": 0.6374169588088989,
      "step": 1544
    },
    {
      "epoch": 1.8597230583985551,
      "grad_norm": 0.1623840481042862,
      "learning_rate": 1.373382495332215e-06,
      "loss": 0.6356952786445618,
      "step": 1545
    },
    {
      "epoch": 1.8609271523178808,
      "grad_norm": 0.1605909913778305,
      "learning_rate": 1.3503083623818412e-06,
      "loss": 0.5187837481498718,
      "step": 1546
    },
    {
      "epoch": 1.8621312462372064,
      "grad_norm": 0.17687231302261353,
      "learning_rate": 1.3274270554775425e-06,
      "loss": 0.5791580080986023,
      "step": 1547
    },
    {
      "epoch": 1.8633353401565322,
      "grad_norm": 0.17903019487857819,
      "learning_rate": 1.3047386653107784e-06,
      "loss": 0.5597528219223022,
      "step": 1548
    },
    {
      "epoch": 1.8645394340758579,
      "grad_norm": 0.19061800837516785,
      "learning_rate": 1.282243281808393e-06,
      "loss": 0.6321817636489868,
      "step": 1549
    },
    {
      "epoch": 1.8657435279951837,
      "grad_norm": 0.17408376932144165,
      "learning_rate": 1.2599409941322081e-06,
      "loss": 0.5055710077285767,
      "step": 1550
    },
    {
      "epoch": 1.8669476219145094,
      "grad_norm": 0.19276069104671478,
      "learning_rate": 1.2378318906787145e-06,
      "loss": 0.6349813938140869,
      "step": 1551
    },
    {
      "epoch": 1.868151715833835,
      "grad_norm": 0.1691667139530182,
      "learning_rate": 1.2159160590787143e-06,
      "loss": 0.5730689764022827,
      "step": 1552
    },
    {
      "epoch": 1.8693558097531606,
      "grad_norm": 0.19586029648780823,
      "learning_rate": 1.1941935861969455e-06,
      "loss": 0.5993601083755493,
      "step": 1553
    },
    {
      "epoch": 1.8705599036724865,
      "grad_norm": 0.18812040984630585,
      "learning_rate": 1.1726645581317752e-06,
      "loss": 0.6908564567565918,
      "step": 1554
    },
    {
      "epoch": 1.8717639975918121,
      "grad_norm": 0.18055526912212372,
      "learning_rate": 1.1513290602148174e-06,
      "loss": 0.6268129348754883,
      "step": 1555
    },
    {
      "epoch": 1.872968091511138,
      "grad_norm": 0.17199203372001648,
      "learning_rate": 1.1301871770106332e-06,
      "loss": 0.5379941463470459,
      "step": 1556
    },
    {
      "epoch": 1.8741721854304636,
      "grad_norm": 0.17254063487052917,
      "learning_rate": 1.1092389923163915e-06,
      "loss": 0.5519986748695374,
      "step": 1557
    },
    {
      "epoch": 1.8753762793497892,
      "grad_norm": 0.15900661051273346,
      "learning_rate": 1.0884845891614925e-06,
      "loss": 0.5397247076034546,
      "step": 1558
    },
    {
      "epoch": 1.8765803732691149,
      "grad_norm": 0.17740990221500397,
      "learning_rate": 1.0679240498073118e-06,
      "loss": 0.5967949032783508,
      "step": 1559
    },
    {
      "epoch": 1.8777844671884407,
      "grad_norm": 0.1513747274875641,
      "learning_rate": 1.0475574557467837e-06,
      "loss": 0.4645845890045166,
      "step": 1560
    },
    {
      "epoch": 1.8789885611077664,
      "grad_norm": 0.16508939862251282,
      "learning_rate": 1.0273848877041802e-06,
      "loss": 0.610875129699707,
      "step": 1561
    },
    {
      "epoch": 1.8801926550270922,
      "grad_norm": 0.16647538542747498,
      "learning_rate": 1.0074064256347104e-06,
      "loss": 0.575761616230011,
      "step": 1562
    },
    {
      "epoch": 1.8813967489464178,
      "grad_norm": 0.17891721427440643,
      "learning_rate": 9.876221487242322e-07,
      "loss": 0.6897559762001038,
      "step": 1563
    },
    {
      "epoch": 1.8826008428657435,
      "grad_norm": 0.1780918836593628,
      "learning_rate": 9.680321353889576e-07,
      "loss": 0.5767819285392761,
      "step": 1564
    },
    {
      "epoch": 1.883804936785069,
      "grad_norm": 0.1640281230211258,
      "learning_rate": 9.486364632750878e-07,
      "loss": 0.5813727974891663,
      "step": 1565
    },
    {
      "epoch": 1.885009030704395,
      "grad_norm": 0.21059930324554443,
      "learning_rate": 9.294352092585779e-07,
      "loss": 0.5976967811584473,
      "step": 1566
    },
    {
      "epoch": 1.8862131246237208,
      "grad_norm": 0.1780235916376114,
      "learning_rate": 9.104284494447779e-07,
      "loss": 0.6003977656364441,
      "step": 1567
    },
    {
      "epoch": 1.8874172185430464,
      "grad_norm": 0.16711755096912384,
      "learning_rate": 8.916162591681543e-07,
      "loss": 0.5505768656730652,
      "step": 1568
    },
    {
      "epoch": 1.888621312462372,
      "grad_norm": 0.18486176431179047,
      "learning_rate": 8.729987129919682e-07,
      "loss": 0.6029719114303589,
      "step": 1569
    },
    {
      "epoch": 1.8898254063816977,
      "grad_norm": 0.16008959710597992,
      "learning_rate": 8.545758847080143e-07,
      "loss": 0.5083392858505249,
      "step": 1570
    },
    {
      "epoch": 1.8910295003010233,
      "grad_norm": 0.16204681992530823,
      "learning_rate": 8.363478473363107e-07,
      "loss": 0.5530649423599243,
      "step": 1571
    },
    {
      "epoch": 1.8922335942203492,
      "grad_norm": 0.17961721122264862,
      "learning_rate": 8.183146731247982e-07,
      "loss": 0.6466346383094788,
      "step": 1572
    },
    {
      "epoch": 1.893437688139675,
      "grad_norm": 0.17810463905334473,
      "learning_rate": 8.004764335490856e-07,
      "loss": 0.5068295001983643,
      "step": 1573
    },
    {
      "epoch": 1.8946417820590007,
      "grad_norm": 0.17782258987426758,
      "learning_rate": 7.828331993121163e-07,
      "loss": 0.5920776724815369,
      "step": 1574
    },
    {
      "epoch": 1.8958458759783263,
      "grad_norm": 0.1609419733285904,
      "learning_rate": 7.653850403439411e-07,
      "loss": 0.49203193187713623,
      "step": 1575
    },
    {
      "epoch": 1.897049969897652,
      "grad_norm": 0.17676517367362976,
      "learning_rate": 7.481320258014124e-07,
      "loss": 0.5533766150474548,
      "step": 1576
    },
    {
      "epoch": 1.8982540638169776,
      "grad_norm": 0.1633012592792511,
      "learning_rate": 7.31074224067918e-07,
      "loss": 0.5492685437202454,
      "step": 1577
    },
    {
      "epoch": 1.8994581577363034,
      "grad_norm": 0.15624099969863892,
      "learning_rate": 7.142117027531092e-07,
      "loss": 0.5692270994186401,
      "step": 1578
    },
    {
      "epoch": 1.9006622516556293,
      "grad_norm": 0.17841790616512299,
      "learning_rate": 6.975445286926063e-07,
      "loss": 0.5900111198425293,
      "step": 1579
    },
    {
      "epoch": 1.901866345574955,
      "grad_norm": 0.1768088936805725,
      "learning_rate": 6.810727679477935e-07,
      "loss": 0.5334895849227905,
      "step": 1580
    },
    {
      "epoch": 1.9030704394942806,
      "grad_norm": 0.16499072313308716,
      "learning_rate": 6.647964858055133e-07,
      "loss": 0.5377838611602783,
      "step": 1581
    },
    {
      "epoch": 1.9042745334136062,
      "grad_norm": 0.1927908957004547,
      "learning_rate": 6.48715746777806e-07,
      "loss": 0.5937684774398804,
      "step": 1582
    },
    {
      "epoch": 1.9054786273329318,
      "grad_norm": 0.17469263076782227,
      "learning_rate": 6.328306146016593e-07,
      "loss": 0.6178199648857117,
      "step": 1583
    },
    {
      "epoch": 1.9066827212522577,
      "grad_norm": 0.1746751219034195,
      "learning_rate": 6.171411522387871e-07,
      "loss": 0.6327543258666992,
      "step": 1584
    },
    {
      "epoch": 1.9078868151715835,
      "grad_norm": 0.1598159521818161,
      "learning_rate": 6.016474218753288e-07,
      "loss": 0.5312304496765137,
      "step": 1585
    },
    {
      "epoch": 1.9090909090909092,
      "grad_norm": 0.16450975835323334,
      "learning_rate": 5.863494849216444e-07,
      "loss": 0.48702770471572876,
      "step": 1586
    },
    {
      "epoch": 1.9102950030102348,
      "grad_norm": 0.18231473863124847,
      "learning_rate": 5.712474020120484e-07,
      "loss": 0.6207761168479919,
      "step": 1587
    },
    {
      "epoch": 1.9114990969295604,
      "grad_norm": 0.18410496413707733,
      "learning_rate": 5.563412330045758e-07,
      "loss": 0.606268584728241,
      "step": 1588
    },
    {
      "epoch": 1.912703190848886,
      "grad_norm": 0.17159083485603333,
      "learning_rate": 5.416310369807331e-07,
      "loss": 0.5566038489341736,
      "step": 1589
    },
    {
      "epoch": 1.913907284768212,
      "grad_norm": 0.15807877480983734,
      "learning_rate": 5.271168722453035e-07,
      "loss": 0.5376920700073242,
      "step": 1590
    },
    {
      "epoch": 1.9151113786875378,
      "grad_norm": 0.17678941786289215,
      "learning_rate": 5.127987963260583e-07,
      "loss": 0.6632184386253357,
      "step": 1591
    },
    {
      "epoch": 1.9163154726068634,
      "grad_norm": 0.17437922954559326,
      "learning_rate": 4.986768659735852e-07,
      "loss": 0.5931605100631714,
      "step": 1592
    },
    {
      "epoch": 1.917519566526189,
      "grad_norm": 0.17323121428489685,
      "learning_rate": 4.847511371610159e-07,
      "loss": 0.5931187868118286,
      "step": 1593
    },
    {
      "epoch": 1.9187236604455147,
      "grad_norm": 0.1626964658498764,
      "learning_rate": 4.710216650838317e-07,
      "loss": 0.550021231174469,
      "step": 1594
    },
    {
      "epoch": 1.9199277543648403,
      "grad_norm": 0.1690921187400818,
      "learning_rate": 4.5748850415964774e-07,
      "loss": 0.562748908996582,
      "step": 1595
    },
    {
      "epoch": 1.9211318482841662,
      "grad_norm": 0.19784128665924072,
      "learning_rate": 4.4415170802797333e-07,
      "loss": 0.5627641677856445,
      "step": 1596
    },
    {
      "epoch": 1.922335942203492,
      "grad_norm": 0.16546784341335297,
      "learning_rate": 4.3101132955002396e-07,
      "loss": 0.5471245050430298,
      "step": 1597
    },
    {
      "epoch": 1.9235400361228177,
      "grad_norm": 0.1499754637479782,
      "learning_rate": 4.180674208084989e-07,
      "loss": 0.4810486435890198,
      "step": 1598
    },
    {
      "epoch": 1.9247441300421433,
      "grad_norm": 0.18828915059566498,
      "learning_rate": 4.0532003310736475e-07,
      "loss": 0.6407335996627808,
      "step": 1599
    },
    {
      "epoch": 1.925948223961469,
      "grad_norm": 0.1771274209022522,
      "learning_rate": 3.9276921697169455e-07,
      "loss": 0.6203795671463013,
      "step": 1600
    },
    {
      "epoch": 1.9271523178807946,
      "grad_norm": 0.15607336163520813,
      "learning_rate": 3.804150221474179e-07,
      "loss": 0.4951472878456116,
      "step": 1601
    },
    {
      "epoch": 1.9283564118001204,
      "grad_norm": 0.16553087532520294,
      "learning_rate": 3.6825749760113215e-07,
      "loss": 0.6408381462097168,
      "step": 1602
    },
    {
      "epoch": 1.9295605057194463,
      "grad_norm": 0.16196313500404358,
      "learning_rate": 3.5629669151994725e-07,
      "loss": 0.6005507707595825,
      "step": 1603
    },
    {
      "epoch": 1.930764599638772,
      "grad_norm": 0.17648328840732574,
      "learning_rate": 3.4453265131124677e-07,
      "loss": 0.6115264296531677,
      "step": 1604
    },
    {
      "epoch": 1.9319686935580975,
      "grad_norm": 0.17745113372802734,
      "learning_rate": 3.3296542360253256e-07,
      "loss": 0.652868390083313,
      "step": 1605
    },
    {
      "epoch": 1.9331727874774232,
      "grad_norm": 0.1785673350095749,
      "learning_rate": 3.2159505424122495e-07,
      "loss": 0.6854658126831055,
      "step": 1606
    },
    {
      "epoch": 1.9343768813967488,
      "grad_norm": 0.16135065257549286,
      "learning_rate": 3.1042158829447385e-07,
      "loss": 0.5672754049301147,
      "step": 1607
    },
    {
      "epoch": 1.9355809753160746,
      "grad_norm": 0.16168314218521118,
      "learning_rate": 2.9944507004900367e-07,
      "loss": 0.5310696959495544,
      "step": 1608
    },
    {
      "epoch": 1.9367850692354005,
      "grad_norm": 0.1800370216369629,
      "learning_rate": 2.8866554301091866e-07,
      "loss": 0.6367923021316528,
      "step": 1609
    },
    {
      "epoch": 1.9379891631547261,
      "grad_norm": 0.191808819770813,
      "learning_rate": 2.780830499055476e-07,
      "loss": 0.5861743688583374,
      "step": 1610
    },
    {
      "epoch": 1.9391932570740518,
      "grad_norm": 0.16959501802921295,
      "learning_rate": 2.6769763267723845e-07,
      "loss": 0.5837535262107849,
      "step": 1611
    },
    {
      "epoch": 1.9403973509933774,
      "grad_norm": 0.1673300415277481,
      "learning_rate": 2.575093324892364e-07,
      "loss": 0.5709747672080994,
      "step": 1612
    },
    {
      "epoch": 1.941601444912703,
      "grad_norm": 0.16310818493366241,
      "learning_rate": 2.4751818972350016e-07,
      "loss": 0.5027601718902588,
      "step": 1613
    },
    {
      "epoch": 1.9428055388320289,
      "grad_norm": 0.19135311245918274,
      "learning_rate": 2.377242439805305e-07,
      "loss": 0.6039505004882812,
      "step": 1614
    },
    {
      "epoch": 1.9440096327513547,
      "grad_norm": 0.1754077821969986,
      "learning_rate": 2.281275340792477e-07,
      "loss": 0.5850051045417786,
      "step": 1615
    },
    {
      "epoch": 1.9452137266706804,
      "grad_norm": 0.17281068861484528,
      "learning_rate": 2.187280980567863e-07,
      "loss": 0.5468301773071289,
      "step": 1616
    },
    {
      "epoch": 1.946417820590006,
      "grad_norm": 0.17825911939144135,
      "learning_rate": 2.095259731684007e-07,
      "loss": 0.6005904674530029,
      "step": 1617
    },
    {
      "epoch": 1.9476219145093316,
      "grad_norm": 0.17072688043117523,
      "learning_rate": 2.0052119588727103e-07,
      "loss": 0.5770457983016968,
      "step": 1618
    },
    {
      "epoch": 1.9488260084286573,
      "grad_norm": 0.1511860340833664,
      "learning_rate": 1.917138019043918e-07,
      "loss": 0.5325096249580383,
      "step": 1619
    },
    {
      "epoch": 1.9500301023479831,
      "grad_norm": 0.17599450051784515,
      "learning_rate": 1.8310382612841125e-07,
      "loss": 0.553093433380127,
      "step": 1620
    },
    {
      "epoch": 1.951234196267309,
      "grad_norm": 0.17477749288082123,
      "learning_rate": 1.7469130268549238e-07,
      "loss": 0.5234755873680115,
      "step": 1621
    },
    {
      "epoch": 1.9524382901866346,
      "grad_norm": 0.15659305453300476,
      "learning_rate": 1.6647626491919088e-07,
      "loss": 0.5138795375823975,
      "step": 1622
    },
    {
      "epoch": 1.9536423841059603,
      "grad_norm": 0.18101339042186737,
      "learning_rate": 1.5845874539032192e-07,
      "loss": 0.6855711936950684,
      "step": 1623
    },
    {
      "epoch": 1.9548464780252859,
      "grad_norm": 0.20194146037101746,
      "learning_rate": 1.5063877587681019e-07,
      "loss": 0.6403281688690186,
      "step": 1624
    },
    {
      "epoch": 1.9560505719446115,
      "grad_norm": 0.1765589565038681,
      "learning_rate": 1.4301638737358459e-07,
      "loss": 0.6540019512176514,
      "step": 1625
    },
    {
      "epoch": 1.9572546658639374,
      "grad_norm": 0.17175477743148804,
      "learning_rate": 1.3559161009246146e-07,
      "loss": 0.6006916165351868,
      "step": 1626
    },
    {
      "epoch": 1.9584587597832632,
      "grad_norm": 0.1576770544052124,
      "learning_rate": 1.283644734619893e-07,
      "loss": 0.4710315465927124,
      "step": 1627
    },
    {
      "epoch": 1.9596628537025889,
      "grad_norm": 0.16778403520584106,
      "learning_rate": 1.2133500612737103e-07,
      "loss": 0.46124938130378723,
      "step": 1628
    },
    {
      "epoch": 1.9608669476219145,
      "grad_norm": 0.16817563772201538,
      "learning_rate": 1.1450323595034174e-07,
      "loss": 0.5780761241912842,
      "step": 1629
    },
    {
      "epoch": 1.9620710415412401,
      "grad_norm": 0.18651093542575836,
      "learning_rate": 1.0786919000903562e-07,
      "loss": 0.6273614168167114,
      "step": 1630
    },
    {
      "epoch": 1.963275135460566,
      "grad_norm": 0.15057410299777985,
      "learning_rate": 1.0143289459790816e-07,
      "loss": 0.5283514857292175,
      "step": 1631
    },
    {
      "epoch": 1.9644792293798916,
      "grad_norm": 0.16399194300174713,
      "learning_rate": 9.519437522760299e-08,
      "loss": 0.5393345952033997,
      "step": 1632
    },
    {
      "epoch": 1.9656833232992175,
      "grad_norm": 0.1684487760066986,
      "learning_rate": 8.915365662488518e-08,
      "loss": 0.5280832052230835,
      "step": 1633
    },
    {
      "epoch": 1.966887417218543,
      "grad_norm": 0.1924428939819336,
      "learning_rate": 8.331076273250249e-08,
      "loss": 0.6634703874588013,
      "step": 1634
    },
    {
      "epoch": 1.9680915111378687,
      "grad_norm": 0.16864493489265442,
      "learning_rate": 7.766571670913547e-08,
      "loss": 0.5633054375648499,
      "step": 1635
    },
    {
      "epoch": 1.9692956050571944,
      "grad_norm": 0.1691860407590866,
      "learning_rate": 7.221854092926971e-08,
      "loss": 0.5685979127883911,
      "step": 1636
    },
    {
      "epoch": 1.9704996989765202,
      "grad_norm": 0.16321423649787903,
      "learning_rate": 6.696925698311817e-08,
      "loss": 0.5267124176025391,
      "step": 1637
    },
    {
      "epoch": 1.9717037928958459,
      "grad_norm": 0.16299042105674744,
      "learning_rate": 6.191788567654344e-08,
      "loss": 0.5181851983070374,
      "step": 1638
    },
    {
      "epoch": 1.9729078868151717,
      "grad_norm": 0.17853769659996033,
      "learning_rate": 5.706444703096336e-08,
      "loss": 0.5656585693359375,
      "step": 1639
    },
    {
      "epoch": 1.9741119807344973,
      "grad_norm": 0.17220546305179596,
      "learning_rate": 5.240896028327891e-08,
      "loss": 0.6099276542663574,
      "step": 1640
    },
    {
      "epoch": 1.975316074653823,
      "grad_norm": 0.16959227621555328,
      "learning_rate": 4.7951443885807524e-08,
      "loss": 0.5535591244697571,
      "step": 1641
    },
    {
      "epoch": 1.9765201685731486,
      "grad_norm": 0.1898435652256012,
      "learning_rate": 4.3691915506177686e-08,
      "loss": 0.6471256017684937,
      "step": 1642
    },
    {
      "epoch": 1.9777242624924745,
      "grad_norm": 0.20082776248455048,
      "learning_rate": 3.963039202730112e-08,
      "loss": 0.5444841384887695,
      "step": 1643
    },
    {
      "epoch": 1.9789283564118,
      "grad_norm": 0.17552979290485382,
      "learning_rate": 3.576688954727847e-08,
      "loss": 0.5809622406959534,
      "step": 1644
    },
    {
      "epoch": 1.980132450331126,
      "grad_norm": 0.17236486077308655,
      "learning_rate": 3.210142337932709e-08,
      "loss": 0.5722782611846924,
      "step": 1645
    },
    {
      "epoch": 1.9813365442504516,
      "grad_norm": 0.1635494828224182,
      "learning_rate": 2.8634008051758864e-08,
      "loss": 0.5646048784255981,
      "step": 1646
    },
    {
      "epoch": 1.9825406381697772,
      "grad_norm": 0.17788147926330566,
      "learning_rate": 2.536465730788029e-08,
      "loss": 0.5577666163444519,
      "step": 1647
    },
    {
      "epoch": 1.9837447320891028,
      "grad_norm": 0.1660120040178299,
      "learning_rate": 2.229338410597026e-08,
      "loss": 0.567434549331665,
      "step": 1648
    },
    {
      "epoch": 1.9849488260084287,
      "grad_norm": 0.17396770417690277,
      "learning_rate": 1.9420200619207907e-08,
      "loss": 0.5758538842201233,
      "step": 1649
    },
    {
      "epoch": 1.9861529199277543,
      "grad_norm": 0.20760847628116608,
      "learning_rate": 1.6745118235628188e-08,
      "loss": 0.6495552659034729,
      "step": 1650
    },
    {
      "epoch": 1.9873570138470802,
      "grad_norm": 0.15991172194480896,
      "learning_rate": 1.4268147558088585e-08,
      "loss": 0.4884168207645416,
      "step": 1651
    },
    {
      "epoch": 1.9885611077664058,
      "grad_norm": 0.17651431262493134,
      "learning_rate": 1.1989298404213588e-08,
      "loss": 0.6884113550186157,
      "step": 1652
    },
    {
      "epoch": 1.9897652016857315,
      "grad_norm": 0.1671612560749054,
      "learning_rate": 9.908579806361396e-09,
      "loss": 0.6421818137168884,
      "step": 1653
    },
    {
      "epoch": 1.990969295605057,
      "grad_norm": 0.18387183547019958,
      "learning_rate": 8.026000011596146e-09,
      "loss": 0.5787399411201477,
      "step": 1654
    },
    {
      "epoch": 1.992173389524383,
      "grad_norm": 0.19110684096813202,
      "learning_rate": 6.341566481626871e-09,
      "loss": 0.6162360310554504,
      "step": 1655
    },
    {
      "epoch": 1.9933774834437086,
      "grad_norm": 0.16275465488433838,
      "learning_rate": 4.855285892813033e-09,
      "loss": 0.5244137048721313,
      "step": 1656
    },
    {
      "epoch": 1.9945815773630344,
      "grad_norm": 0.16620507836341858,
      "learning_rate": 3.567164136120127e-09,
      "loss": 0.5226976275444031,
      "step": 1657
    },
    {
      "epoch": 1.99578567128236,
      "grad_norm": 0.15561197698116302,
      "learning_rate": 2.47720631710302e-09,
      "loss": 0.5414796471595764,
      "step": 1658
    },
    {
      "epoch": 1.9969897652016857,
      "grad_norm": 0.17942088842391968,
      "learning_rate": 1.5854167558670975e-09,
      "loss": 0.5349199175834656,
      "step": 1659
    },
    {
      "epoch": 1.9981938591210113,
      "grad_norm": 0.17291858792304993,
      "learning_rate": 8.917989870849131e-10,
      "loss": 0.5907130241394043,
      "step": 1660
    },
    {
      "epoch": 1.9993979530403372,
      "grad_norm": 0.18802672624588013,
      "learning_rate": 3.9635575994623196e-10,
      "loss": 0.6180111765861511,
      "step": 1661
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.28912901878356934,
      "learning_rate": 9.908903817468229e-11,
      "loss": 0.6963255405426025,
      "step": 1662
    },
    {
      "epoch": 2.0,
      "step": 1662,
      "total_flos": 2.3540766095452406e+18,
      "train_loss": 0.6956572735998174,
      "train_runtime": 14497.215,
      "train_samples_per_second": 0.917,
      "train_steps_per_second": 0.115
    }
  ],
  "logging_steps": 1,
  "max_steps": 1662,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.3540766095452406e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}