{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 288,
  "global_step": 1152,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0008680555555555555,
      "grad_norm": 1.8690676370169967e-05,
      "learning_rate": 2e-05,
      "loss": 11.5,
      "step": 1
    },
    {
      "epoch": 0.001736111111111111,
      "grad_norm": 3.8940950616961345e-05,
      "learning_rate": 4e-05,
      "loss": 11.5,
      "step": 2
    },
    {
      "epoch": 0.0026041666666666665,
      "grad_norm": 2.2233934942050837e-05,
      "learning_rate": 6e-05,
      "loss": 11.5,
      "step": 3
    },
    {
      "epoch": 0.003472222222222222,
      "grad_norm": 3.2008359994506463e-05,
      "learning_rate": 8e-05,
      "loss": 11.5,
      "step": 4
    },
    {
      "epoch": 0.004340277777777778,
      "grad_norm": 4.118507058592513e-05,
      "learning_rate": 0.0001,
      "loss": 11.5,
      "step": 5
    },
    {
      "epoch": 0.005208333333333333,
      "grad_norm": 5.2076458814553916e-05,
      "learning_rate": 0.00012,
      "loss": 11.5,
      "step": 6
    },
    {
      "epoch": 0.006076388888888889,
      "grad_norm": 5.457341467263177e-05,
      "learning_rate": 0.00014,
      "loss": 11.5,
      "step": 7
    },
    {
      "epoch": 0.006944444444444444,
      "grad_norm": 8.313016587635502e-05,
      "learning_rate": 0.00016,
      "loss": 11.5,
      "step": 8
    },
    {
      "epoch": 0.0078125,
      "grad_norm": 7.593315240228549e-05,
      "learning_rate": 0.00018,
      "loss": 11.5,
      "step": 9
    },
    {
      "epoch": 0.008680555555555556,
      "grad_norm": 7.41191252018325e-05,
      "learning_rate": 0.0002,
      "loss": 11.5,
      "step": 10
    },
    {
      "epoch": 0.009548611111111112,
      "grad_norm": 4.545129195321351e-05,
      "learning_rate": 0.00019999962161207577,
      "loss": 11.5,
      "step": 11
    },
    {
      "epoch": 0.010416666666666666,
      "grad_norm": 6.200841016834602e-05,
      "learning_rate": 0.00019999848645116654,
      "loss": 11.5,
      "step": 12
    },
    {
      "epoch": 0.011284722222222222,
      "grad_norm": 6.061939711798914e-05,
      "learning_rate": 0.000199996594525863,
      "loss": 11.5,
      "step": 13
    },
    {
      "epoch": 0.012152777777777778,
      "grad_norm": 6.343455606838688e-05,
      "learning_rate": 0.0001999939458504828,
      "loss": 11.5,
      "step": 14
    },
    {
      "epoch": 0.013020833333333334,
      "grad_norm": 4.285885734134354e-05,
      "learning_rate": 0.00019999054044507038,
      "loss": 11.5,
      "step": 15
    },
    {
      "epoch": 0.013888888888888888,
      "grad_norm": 5.143031376064755e-05,
      "learning_rate": 0.00019998637833539714,
      "loss": 11.5,
      "step": 16
    },
    {
      "epoch": 0.014756944444444444,
      "grad_norm": 6.0782280343119055e-05,
      "learning_rate": 0.00019998145955296083,
      "loss": 11.5,
      "step": 17
    },
    {
      "epoch": 0.015625,
      "grad_norm": 8.231138781411573e-05,
      "learning_rate": 0.00019997578413498564,
      "loss": 11.5,
      "step": 18
    },
    {
      "epoch": 0.016493055555555556,
      "grad_norm": 4.770286614075303e-05,
      "learning_rate": 0.0001999693521244218,
      "loss": 11.5,
      "step": 19
    },
    {
      "epoch": 0.017361111111111112,
      "grad_norm": 2.215989661635831e-05,
      "learning_rate": 0.00019996216356994517,
      "loss": 11.5,
      "step": 20
    },
    {
      "epoch": 0.018229166666666668,
      "grad_norm": 3.4150350984418765e-05,
      "learning_rate": 0.000199954218525957,
      "loss": 11.5,
      "step": 21
    },
    {
      "epoch": 0.019097222222222224,
      "grad_norm": 4.7818444727454334e-05,
      "learning_rate": 0.00019994551705258343,
      "loss": 11.5,
      "step": 22
    },
    {
      "epoch": 0.019965277777777776,
      "grad_norm": 5.774742749053985e-05,
      "learning_rate": 0.0001999360592156752,
      "loss": 11.5,
      "step": 23
    },
    {
      "epoch": 0.020833333333333332,
      "grad_norm": 5.2880528528476134e-05,
      "learning_rate": 0.00019992584508680684,
      "loss": 11.5,
      "step": 24
    },
    {
      "epoch": 0.021701388888888888,
      "grad_norm": 7.131279562599957e-05,
      "learning_rate": 0.00019991487474327647,
      "loss": 11.5,
      "step": 25
    },
    {
      "epoch": 0.022569444444444444,
      "grad_norm": 3.6745255783898756e-05,
      "learning_rate": 0.000199903148268105,
      "loss": 11.5,
      "step": 26
    },
    {
      "epoch": 0.0234375,
      "grad_norm": 0.00010852243576664478,
      "learning_rate": 0.0001998906657500355,
      "loss": 11.5,
      "step": 27
    },
    {
      "epoch": 0.024305555555555556,
      "grad_norm": 5.074693763162941e-05,
      "learning_rate": 0.00019987742728353274,
      "loss": 11.5,
      "step": 28
    },
    {
      "epoch": 0.025173611111111112,
      "grad_norm": 7.016357267275453e-05,
      "learning_rate": 0.00019986343296878216,
      "loss": 11.5,
      "step": 29
    },
    {
      "epoch": 0.026041666666666668,
      "grad_norm": 4.338881262810901e-05,
      "learning_rate": 0.0001998486829116894,
      "loss": 11.5,
      "step": 30
    },
    {
      "epoch": 0.026909722222222224,
      "grad_norm": 8.875480125425383e-05,
      "learning_rate": 0.0001998331772238793,
      "loss": 11.5,
      "step": 31
    },
    {
      "epoch": 0.027777777777777776,
      "grad_norm": 9.680647781351581e-05,
      "learning_rate": 0.0001998169160226952,
      "loss": 11.5,
      "step": 32
    },
    {
      "epoch": 0.028645833333333332,
      "grad_norm": 0.00010232284694211558,
      "learning_rate": 0.0001997998994311979,
      "loss": 11.5,
      "step": 33
    },
    {
      "epoch": 0.029513888888888888,
      "grad_norm": 9.061591845238581e-05,
      "learning_rate": 0.00019978212757816493,
      "loss": 11.5,
      "step": 34
    },
    {
      "epoch": 0.030381944444444444,
      "grad_norm": 7.143637776607648e-05,
      "learning_rate": 0.00019976360059808927,
      "loss": 11.5,
      "step": 35
    },
    {
      "epoch": 0.03125,
      "grad_norm": 5.950556806055829e-05,
      "learning_rate": 0.0001997443186311787,
      "loss": 11.5,
      "step": 36
    },
    {
      "epoch": 0.03211805555555555,
      "grad_norm": 0.00010689818009268492,
      "learning_rate": 0.0001997242818233545,
      "loss": 11.5,
      "step": 37
    },
    {
      "epoch": 0.03298611111111111,
      "grad_norm": 9.630536078475416e-05,
      "learning_rate": 0.00019970349032625035,
      "loss": 11.5,
      "step": 38
    },
    {
      "epoch": 0.033854166666666664,
      "grad_norm": 9.237516496796161e-05,
      "learning_rate": 0.0001996819442972113,
      "loss": 11.5,
      "step": 39
    },
    {
      "epoch": 0.034722222222222224,
      "grad_norm": 7.252427167259157e-05,
      "learning_rate": 0.0001996596438992925,
      "loss": 11.5,
      "step": 40
    },
    {
      "epoch": 0.035590277777777776,
      "grad_norm": 8.173917012754828e-05,
      "learning_rate": 0.00019963658930125794,
      "loss": 11.5,
      "step": 41
    },
    {
      "epoch": 0.036458333333333336,
      "grad_norm": 0.00022069711121730506,
      "learning_rate": 0.0001996127806775793,
      "loss": 11.5,
      "step": 42
    },
    {
      "epoch": 0.03732638888888889,
      "grad_norm": 0.00016364461043849587,
      "learning_rate": 0.00019958821820843448,
      "loss": 11.5,
      "step": 43
    },
    {
      "epoch": 0.03819444444444445,
      "grad_norm": 9.40857280511409e-05,
      "learning_rate": 0.0001995629020797063,
      "loss": 11.5,
      "step": 44
    },
    {
      "epoch": 0.0390625,
      "grad_norm": 8.06432799436152e-05,
      "learning_rate": 0.00019953683248298106,
      "loss": 11.5,
      "step": 45
    },
    {
      "epoch": 0.03993055555555555,
      "grad_norm": 0.00011379068746464327,
      "learning_rate": 0.00019951000961554728,
      "loss": 11.5,
      "step": 46
    },
    {
      "epoch": 0.04079861111111111,
      "grad_norm": 0.00021569308592006564,
      "learning_rate": 0.0001994824336803939,
      "loss": 11.5,
      "step": 47
    },
    {
      "epoch": 0.041666666666666664,
      "grad_norm": 0.00011612941307248548,
      "learning_rate": 0.0001994541048862089,
      "loss": 11.5,
      "step": 48
    },
    {
      "epoch": 0.042534722222222224,
      "grad_norm": 0.00018939947767648846,
      "learning_rate": 0.00019942502344737778,
      "loss": 11.5,
      "step": 49
    },
    {
      "epoch": 0.043402777777777776,
      "grad_norm": 0.0001262818550458178,
      "learning_rate": 0.00019939518958398184,
      "loss": 11.5,
      "step": 50
    },
    {
      "epoch": 0.044270833333333336,
      "grad_norm": 3.728344745468348e-05,
      "learning_rate": 0.00019936460352179655,
      "loss": 11.5,
      "step": 51
    },
    {
      "epoch": 0.04513888888888889,
      "grad_norm": 0.0001446466048946604,
      "learning_rate": 0.00019933326549228988,
      "loss": 11.5,
      "step": 52
    },
    {
      "epoch": 0.04600694444444445,
      "grad_norm": 0.00018098857253789902,
      "learning_rate": 0.0001993011757326204,
      "loss": 11.5,
      "step": 53
    },
    {
      "epoch": 0.046875,
      "grad_norm": 8.264760253950953e-05,
      "learning_rate": 0.00019926833448563573,
      "loss": 11.5,
      "step": 54
    },
    {
      "epoch": 0.04774305555555555,
      "grad_norm": 9.048637730302289e-05,
      "learning_rate": 0.00019923474199987044,
      "loss": 11.5,
      "step": 55
    },
    {
      "epoch": 0.04861111111111111,
      "grad_norm": 7.499189086956903e-05,
      "learning_rate": 0.00019920039852954434,
      "loss": 11.5,
      "step": 56
    },
    {
      "epoch": 0.049479166666666664,
      "grad_norm": 0.0002214812848251313,
      "learning_rate": 0.00019916530433456063,
      "loss": 11.5,
      "step": 57
    },
    {
      "epoch": 0.050347222222222224,
      "grad_norm": 7.582152466056868e-05,
      "learning_rate": 0.00019912945968050355,
      "loss": 11.5,
      "step": 58
    },
    {
      "epoch": 0.051215277777777776,
      "grad_norm": 7.347519567701966e-05,
      "learning_rate": 0.0001990928648386369,
      "loss": 11.5,
      "step": 59
    },
    {
      "epoch": 0.052083333333333336,
      "grad_norm": 9.305484854849055e-05,
      "learning_rate": 0.00019905552008590155,
      "loss": 11.5,
      "step": 60
    },
    {
      "epoch": 0.05295138888888889,
      "grad_norm": 0.00024667783873155713,
      "learning_rate": 0.00019901742570491358,
      "loss": 11.5,
      "step": 61
    },
    {
      "epoch": 0.05381944444444445,
      "grad_norm": 0.00019169499864801764,
      "learning_rate": 0.00019897858198396207,
      "loss": 11.5,
      "step": 62
    },
    {
      "epoch": 0.0546875,
      "grad_norm": 0.00013044146180618554,
      "learning_rate": 0.0001989389892170069,
      "loss": 11.5,
      "step": 63
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 0.00014805140381213278,
      "learning_rate": 0.00019889864770367661,
      "loss": 11.5,
      "step": 64
    },
    {
      "epoch": 0.05642361111111111,
      "grad_norm": 0.0002353356103412807,
      "learning_rate": 0.000198857557749266,
      "loss": 11.5,
      "step": 65
    },
    {
      "epoch": 0.057291666666666664,
      "grad_norm": 0.00017275268328376114,
      "learning_rate": 0.00019881571966473392,
      "loss": 11.5,
      "step": 66
    },
    {
      "epoch": 0.058159722222222224,
      "grad_norm": 0.00016553839668631554,
      "learning_rate": 0.00019877313376670088,
      "loss": 11.5,
      "step": 67
    },
    {
      "epoch": 0.059027777777777776,
      "grad_norm": 0.0001652859355090186,
      "learning_rate": 0.0001987298003774467,
      "loss": 11.5,
      "step": 68
    },
    {
      "epoch": 0.059895833333333336,
      "grad_norm": 0.0003071606042794883,
      "learning_rate": 0.00019868571982490797,
      "loss": 11.5,
      "step": 69
    },
    {
      "epoch": 0.06076388888888889,
      "grad_norm": 0.0003049922233913094,
      "learning_rate": 0.0001986408924426757,
      "loss": 11.5,
      "step": 70
    },
    {
      "epoch": 0.06163194444444445,
      "grad_norm": 0.00018270904547534883,
      "learning_rate": 0.00019859531856999268,
      "loss": 11.5,
      "step": 71
    },
    {
      "epoch": 0.0625,
      "grad_norm": 0.0003253938921261579,
      "learning_rate": 0.00019854899855175097,
      "loss": 11.5,
      "step": 72
    },
    {
      "epoch": 0.06336805555555555,
      "grad_norm": 0.00011453074694145471,
      "learning_rate": 0.00019850193273848927,
      "loss": 11.5,
      "step": 73
    },
    {
      "epoch": 0.0642361111111111,
      "grad_norm": 0.0002909557370003313,
      "learning_rate": 0.00019845412148639032,
      "loss": 11.5,
      "step": 74
    },
    {
      "epoch": 0.06510416666666667,
      "grad_norm": 0.0003043895121663809,
      "learning_rate": 0.0001984055651572781,
      "loss": 11.5,
      "step": 75
    },
    {
      "epoch": 0.06597222222222222,
      "grad_norm": 0.0001119500448112376,
      "learning_rate": 0.00019835626411861514,
      "loss": 11.5,
      "step": 76
    },
    {
      "epoch": 0.06684027777777778,
      "grad_norm": 0.0003007218474522233,
      "learning_rate": 0.0001983062187434999,
      "loss": 11.5,
      "step": 77
    },
    {
      "epoch": 0.06770833333333333,
      "grad_norm": 0.0003525039937812835,
      "learning_rate": 0.0001982554294106636,
      "loss": 11.5,
      "step": 78
    },
    {
      "epoch": 0.0685763888888889,
      "grad_norm": 0.00017316255252808332,
      "learning_rate": 0.00019820389650446768,
      "loss": 11.5,
      "step": 79
    },
    {
      "epoch": 0.06944444444444445,
      "grad_norm": 0.0001584248529979959,
      "learning_rate": 0.00019815162041490075,
      "loss": 11.5,
      "step": 80
    },
    {
      "epoch": 0.0703125,
      "grad_norm": 0.0002387677232036367,
      "learning_rate": 0.0001980986015375756,
      "loss": 11.5,
      "step": 81
    },
    {
      "epoch": 0.07118055555555555,
      "grad_norm": 0.0003504953347146511,
      "learning_rate": 0.0001980448402737263,
      "loss": 11.5,
      "step": 82
    },
    {
      "epoch": 0.0720486111111111,
      "grad_norm": 0.0003012304368894547,
      "learning_rate": 0.00019799033703020512,
      "loss": 11.5,
      "step": 83
    },
    {
      "epoch": 0.07291666666666667,
      "grad_norm": 0.00016139281797222793,
      "learning_rate": 0.00019793509221947936,
      "loss": 11.5,
      "step": 84
    },
    {
      "epoch": 0.07378472222222222,
      "grad_norm": 0.00024013794609345496,
      "learning_rate": 0.00019787910625962854,
      "loss": 11.5,
      "step": 85
    },
    {
      "epoch": 0.07465277777777778,
      "grad_norm": 0.0004409114771988243,
      "learning_rate": 0.0001978223795743408,
      "loss": 11.5,
      "step": 86
    },
    {
      "epoch": 0.07552083333333333,
      "grad_norm": 0.00014997906691860408,
      "learning_rate": 0.00019776491259291,
      "loss": 11.5,
      "step": 87
    },
    {
      "epoch": 0.0763888888888889,
      "grad_norm": 0.0002702058118302375,
      "learning_rate": 0.00019770670575023242,
      "loss": 11.5,
      "step": 88
    },
    {
      "epoch": 0.07725694444444445,
      "grad_norm": 0.0004157458897680044,
      "learning_rate": 0.00019764775948680332,
      "loss": 11.5,
      "step": 89
    },
    {
      "epoch": 0.078125,
      "grad_norm": 0.0004937223857268691,
      "learning_rate": 0.00019758807424871385,
      "loss": 11.5,
      "step": 90
    },
    {
      "epoch": 0.07899305555555555,
      "grad_norm": 0.0004576531646307558,
      "learning_rate": 0.0001975276504876474,
      "loss": 11.5,
      "step": 91
    },
    {
      "epoch": 0.0798611111111111,
      "grad_norm": 0.00024630161351524293,
      "learning_rate": 0.0001974664886608765,
      "loss": 11.5,
      "step": 92
    },
    {
      "epoch": 0.08072916666666667,
      "grad_norm": 0.0002799352223519236,
      "learning_rate": 0.000197404589231259,
      "loss": 11.5,
      "step": 93
    },
    {
      "epoch": 0.08159722222222222,
      "grad_norm": 0.00042265214142389596,
      "learning_rate": 0.0001973419526672349,
      "loss": 11.5,
      "step": 94
    },
    {
      "epoch": 0.08246527777777778,
      "grad_norm": 0.000869172450620681,
      "learning_rate": 0.00019727857944282253,
      "loss": 11.5,
      "step": 95
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 0.00021286625997163355,
      "learning_rate": 0.00019721447003761517,
      "loss": 11.5,
      "step": 96
    },
    {
      "epoch": 0.0842013888888889,
      "grad_norm": 0.00027908102492801845,
      "learning_rate": 0.00019714962493677733,
      "loss": 11.5,
      "step": 97
    },
    {
      "epoch": 0.08506944444444445,
      "grad_norm": 0.00037845943006686866,
      "learning_rate": 0.00019708404463104104,
      "loss": 11.5,
      "step": 98
    },
    {
      "epoch": 0.0859375,
      "grad_norm": 0.0003541351470630616,
      "learning_rate": 0.00019701772961670225,
      "loss": 11.5,
      "step": 99
    },
    {
      "epoch": 0.08680555555555555,
      "grad_norm": 0.0012007280020043254,
      "learning_rate": 0.00019695068039561695,
      "loss": 11.5,
      "step": 100
    },
    {
      "epoch": 0.0876736111111111,
      "grad_norm": 0.00035098547232337296,
      "learning_rate": 0.00019688289747519746,
      "loss": 11.5,
      "step": 101
    },
    {
      "epoch": 0.08854166666666667,
      "grad_norm": 0.0005831103189848363,
      "learning_rate": 0.00019681438136840852,
      "loss": 11.5,
      "step": 102
    },
    {
      "epoch": 0.08940972222222222,
      "grad_norm": 0.00026166639872826636,
      "learning_rate": 0.00019674513259376354,
      "loss": 11.5,
      "step": 103
    },
    {
      "epoch": 0.09027777777777778,
      "grad_norm": 0.0003540641046129167,
      "learning_rate": 0.0001966751516753205,
      "loss": 11.5,
      "step": 104
    },
    {
      "epoch": 0.09114583333333333,
      "grad_norm": 0.00016458670143038034,
      "learning_rate": 0.00019660443914267805,
      "loss": 11.5,
      "step": 105
    },
    {
      "epoch": 0.0920138888888889,
      "grad_norm": 0.0005392400198616087,
      "learning_rate": 0.0001965329955309716,
      "loss": 11.5,
      "step": 106
    },
    {
      "epoch": 0.09288194444444445,
      "grad_norm": 0.00011072640336351469,
      "learning_rate": 0.00019646082138086914,
      "loss": 11.5,
      "step": 107
    },
    {
      "epoch": 0.09375,
      "grad_norm": 0.0006778779206797481,
      "learning_rate": 0.0001963879172385672,
      "loss": 11.5,
      "step": 108
    },
    {
      "epoch": 0.09461805555555555,
      "grad_norm": 0.0005206135683692992,
      "learning_rate": 0.0001963142836557867,
      "loss": 11.5,
      "step": 109
    },
    {
      "epoch": 0.0954861111111111,
      "grad_norm": 0.00024275887699332088,
      "learning_rate": 0.00019623992118976888,
      "loss": 11.5,
      "step": 110
    },
    {
      "epoch": 0.09635416666666667,
      "grad_norm": 0.0006468957872129977,
      "learning_rate": 0.00019616483040327083,
      "loss": 11.5,
      "step": 111
    },
    {
      "epoch": 0.09722222222222222,
      "grad_norm": 0.00020624508033506572,
      "learning_rate": 0.00019608901186456156,
      "loss": 11.5,
      "step": 112
    },
    {
      "epoch": 0.09809027777777778,
      "grad_norm": 0.00035258455318398774,
      "learning_rate": 0.00019601246614741745,
      "loss": 11.5,
      "step": 113
    },
    {
      "epoch": 0.09895833333333333,
      "grad_norm": 0.000752267602365464,
      "learning_rate": 0.00019593519383111799,
      "loss": 11.5,
      "step": 114
    },
    {
      "epoch": 0.0998263888888889,
      "grad_norm": 0.0009282658575102687,
      "learning_rate": 0.00019585719550044138,
      "loss": 11.5,
      "step": 115
    },
    {
      "epoch": 0.10069444444444445,
      "grad_norm": 0.00031870201928541064,
      "learning_rate": 0.00019577847174566016,
      "loss": 11.5,
      "step": 116
    },
    {
      "epoch": 0.1015625,
      "grad_norm": 0.0005301000201143324,
      "learning_rate": 0.00019569902316253672,
      "loss": 11.5,
      "step": 117
    },
    {
      "epoch": 0.10243055555555555,
      "grad_norm": 0.0005146575858816504,
      "learning_rate": 0.00019561885035231872,
      "loss": 11.5,
      "step": 118
    },
    {
      "epoch": 0.1032986111111111,
      "grad_norm": 0.0005217585130594671,
      "learning_rate": 0.00019553795392173468,
      "loss": 11.5,
      "step": 119
    },
    {
      "epoch": 0.10416666666666667,
      "grad_norm": 0.0009250346920453012,
      "learning_rate": 0.00019545633448298918,
      "loss": 11.5,
      "step": 120
    },
    {
      "epoch": 0.10503472222222222,
      "grad_norm": 0.000493133848067373,
      "learning_rate": 0.00019537399265375844,
      "loss": 11.5,
      "step": 121
    },
    {
      "epoch": 0.10590277777777778,
      "grad_norm": 0.0005411256570369005,
      "learning_rate": 0.00019529092905718555,
      "loss": 11.5,
      "step": 122
    },
    {
      "epoch": 0.10677083333333333,
      "grad_norm": 0.0003347770543769002,
      "learning_rate": 0.00019520714432187574,
      "loss": 11.5,
      "step": 123
    },
    {
      "epoch": 0.1076388888888889,
      "grad_norm": 0.0007514807512052357,
      "learning_rate": 0.00019512263908189165,
      "loss": 11.5,
      "step": 124
    },
    {
      "epoch": 0.10850694444444445,
      "grad_norm": 0.00037369472556747496,
      "learning_rate": 0.00019503741397674856,
      "loss": 11.5,
      "step": 125
    },
    {
      "epoch": 0.109375,
      "grad_norm": 0.0021254736930131912,
      "learning_rate": 0.00019495146965140943,
      "loss": 11.5,
      "step": 126
    },
    {
      "epoch": 0.11024305555555555,
      "grad_norm": 0.0008453225018456578,
      "learning_rate": 0.00019486480675628019,
      "loss": 11.5,
      "step": 127
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 0.00043887304491363466,
      "learning_rate": 0.00019477742594720468,
      "loss": 11.5,
      "step": 128
    },
    {
      "epoch": 0.11197916666666667,
      "grad_norm": 0.0009173875441774726,
      "learning_rate": 0.00019468932788545978,
      "loss": 11.5,
      "step": 129
    },
    {
      "epoch": 0.11284722222222222,
      "grad_norm": 0.00040986991371028125,
      "learning_rate": 0.00019460051323775033,
      "loss": 11.5,
      "step": 130
    },
    {
      "epoch": 0.11371527777777778,
      "grad_norm": 0.00020309908722992986,
      "learning_rate": 0.00019451098267620411,
      "loss": 11.5,
      "step": 131
    },
    {
      "epoch": 0.11458333333333333,
      "grad_norm": 0.000617681653238833,
      "learning_rate": 0.00019442073687836686,
      "loss": 11.5,
      "step": 132
    },
    {
      "epoch": 0.1154513888888889,
      "grad_norm": 0.00087716244161129,
      "learning_rate": 0.00019432977652719693,
      "loss": 11.5,
      "step": 133
    },
    {
      "epoch": 0.11631944444444445,
      "grad_norm": 0.0005619502626359463,
      "learning_rate": 0.00019423810231106026,
      "loss": 11.5,
      "step": 134
    },
    {
      "epoch": 0.1171875,
      "grad_norm": 0.0005967443576082587,
      "learning_rate": 0.00019414571492372523,
      "loss": 11.5,
      "step": 135
    },
    {
      "epoch": 0.11805555555555555,
      "grad_norm": 0.00042016865336336195,
      "learning_rate": 0.00019405261506435728,
      "loss": 11.5,
      "step": 136
    },
    {
      "epoch": 0.1189236111111111,
      "grad_norm": 0.0006284159026108682,
      "learning_rate": 0.00019395880343751364,
      "loss": 11.5,
      "step": 137
    },
    {
      "epoch": 0.11979166666666667,
      "grad_norm": 0.0007687513716518879,
      "learning_rate": 0.000193864280753138,
      "loss": 11.5,
      "step": 138
    },
    {
      "epoch": 0.12065972222222222,
      "grad_norm": 0.0004984052502550185,
      "learning_rate": 0.0001937690477265553,
      "loss": 11.5,
      "step": 139
    },
    {
      "epoch": 0.12152777777777778,
      "grad_norm": 0.000553861609660089,
      "learning_rate": 0.000193673105078466,
      "loss": 11.5,
      "step": 140
    },
    {
      "epoch": 0.12239583333333333,
      "grad_norm": 0.0007645233999937773,
      "learning_rate": 0.00019357645353494093,
      "loss": 11.5,
      "step": 141
    },
    {
      "epoch": 0.1232638888888889,
      "grad_norm": 0.000750528066419065,
      "learning_rate": 0.00019347909382741563,
      "loss": 11.5,
      "step": 142
    },
    {
      "epoch": 0.12413194444444445,
      "grad_norm": 0.0009470049990341067,
      "learning_rate": 0.00019338102669268486,
      "loss": 11.5,
      "step": 143
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.0006648892303928733,
      "learning_rate": 0.000193282252872897,
      "loss": 11.5,
      "step": 144
    },
    {
      "epoch": 0.12586805555555555,
      "grad_norm": 0.0016906660748645663,
      "learning_rate": 0.0001931827731155484,
      "loss": 11.5,
      "step": 145
    },
    {
      "epoch": 0.1267361111111111,
      "grad_norm": 0.0005745351081714034,
      "learning_rate": 0.00019308258817347796,
      "loss": 11.5,
      "step": 146
    },
    {
      "epoch": 0.12760416666666666,
      "grad_norm": 0.0017070550238713622,
      "learning_rate": 0.00019298169880486104,
      "loss": 11.5,
      "step": 147
    },
    {
      "epoch": 0.1284722222222222,
      "grad_norm": 0.0003711479075718671,
      "learning_rate": 0.00019288010577320406,
      "loss": 11.5,
      "step": 148
    },
    {
      "epoch": 0.1293402777777778,
      "grad_norm": 0.0007744630565866828,
      "learning_rate": 0.00019277780984733848,
      "loss": 11.5,
      "step": 149
    },
    {
      "epoch": 0.13020833333333334,
      "grad_norm": 0.000989002757705748,
      "learning_rate": 0.00019267481180141525,
      "loss": 11.5,
      "step": 150
    },
    {
      "epoch": 0.1310763888888889,
      "grad_norm": 0.0002649259986355901,
      "learning_rate": 0.00019257111241489865,
      "loss": 11.5,
      "step": 151
    },
    {
      "epoch": 0.13194444444444445,
      "grad_norm": 0.0016404542839154601,
      "learning_rate": 0.00019246671247256057,
      "loss": 11.5,
      "step": 152
    },
    {
      "epoch": 0.1328125,
      "grad_norm": 0.0008670961833558977,
      "learning_rate": 0.00019236161276447463,
      "loss": 11.5,
      "step": 153
    },
    {
      "epoch": 0.13368055555555555,
      "grad_norm": 0.0009681170340627432,
      "learning_rate": 0.00019225581408601,
      "loss": 11.5,
      "step": 154
    },
    {
      "epoch": 0.1345486111111111,
      "grad_norm": 0.001822664518840611,
      "learning_rate": 0.00019214931723782552,
      "loss": 11.5,
      "step": 155
    },
    {
      "epoch": 0.13541666666666666,
      "grad_norm": 0.0010668959002941847,
      "learning_rate": 0.00019204212302586363,
      "loss": 11.5,
      "step": 156
    },
    {
      "epoch": 0.1362847222222222,
      "grad_norm": 0.0015830661868676543,
      "learning_rate": 0.00019193423226134424,
      "loss": 11.5,
      "step": 157
    },
    {
      "epoch": 0.1371527777777778,
      "grad_norm": 0.00042740401113405824,
      "learning_rate": 0.0001918256457607586,
      "loss": 11.5,
      "step": 158
    },
    {
      "epoch": 0.13802083333333334,
      "grad_norm": 0.00036738254129886627,
      "learning_rate": 0.00019171636434586308,
      "loss": 11.5,
      "step": 159
    },
    {
      "epoch": 0.1388888888888889,
      "grad_norm": 0.0023373544681817293,
      "learning_rate": 0.0001916063888436731,
      "loss": 11.5,
      "step": 160
    },
    {
      "epoch": 0.13975694444444445,
      "grad_norm": 0.0014293011045083404,
      "learning_rate": 0.00019149572008645663,
      "loss": 11.5,
      "step": 161
    },
    {
      "epoch": 0.140625,
      "grad_norm": 0.0004194263310637325,
      "learning_rate": 0.00019138435891172812,
      "loss": 11.5,
      "step": 162
    },
    {
      "epoch": 0.14149305555555555,
      "grad_norm": 0.0020428544376045465,
      "learning_rate": 0.00019127230616224213,
      "loss": 11.5,
      "step": 163
    },
    {
      "epoch": 0.1423611111111111,
      "grad_norm": 0.0009024456376209855,
      "learning_rate": 0.00019115956268598664,
      "loss": 11.5,
      "step": 164
    },
    {
      "epoch": 0.14322916666666666,
      "grad_norm": 0.0011780044296756387,
      "learning_rate": 0.0001910461293361772,
      "loss": 11.5,
      "step": 165
    },
    {
      "epoch": 0.1440972222222222,
      "grad_norm": 0.0018483465537428856,
      "learning_rate": 0.00019093200697124996,
      "loss": 11.5,
      "step": 166
    },
    {
      "epoch": 0.1449652777777778,
      "grad_norm": 0.0009278951911255717,
      "learning_rate": 0.0001908171964548554,
      "loss": 11.5,
      "step": 167
    },
    {
      "epoch": 0.14583333333333334,
      "grad_norm": 0.0012078545987606049,
      "learning_rate": 0.00019070169865585175,
      "loss": 11.5,
      "step": 168
    },
    {
      "epoch": 0.1467013888888889,
      "grad_norm": 0.0028224762063473463,
      "learning_rate": 0.0001905855144482985,
      "loss": 11.5,
      "step": 169
    },
    {
      "epoch": 0.14756944444444445,
      "grad_norm": 0.002220297697931528,
      "learning_rate": 0.0001904686447114497,
      "loss": 11.5,
      "step": 170
    },
    {
      "epoch": 0.1484375,
      "grad_norm": 0.0015452924417331815,
      "learning_rate": 0.00019035109032974722,
      "loss": 11.5,
      "step": 171
    },
    {
      "epoch": 0.14930555555555555,
      "grad_norm": 0.0026909364387392998,
      "learning_rate": 0.00019023285219281429,
      "loss": 11.5,
      "step": 172
    },
    {
      "epoch": 0.1501736111111111,
      "grad_norm": 0.00165748770814389,
      "learning_rate": 0.00019011393119544852,
      "loss": 11.5,
      "step": 173
    },
    {
      "epoch": 0.15104166666666666,
      "grad_norm": 0.001105487230233848,
      "learning_rate": 0.00018999432823761534,
      "loss": 11.5,
      "step": 174
    },
    {
      "epoch": 0.1519097222222222,
      "grad_norm": 0.0010610170429572463,
      "learning_rate": 0.00018987404422444104,
      "loss": 11.5,
      "step": 175
    },
    {
      "epoch": 0.1527777777777778,
      "grad_norm": 0.001609691185876727,
      "learning_rate": 0.00018975308006620597,
      "loss": 11.5,
      "step": 176
    },
    {
      "epoch": 0.15364583333333334,
      "grad_norm": 0.0023760476615279913,
      "learning_rate": 0.0001896314366783377,
      "loss": 11.5,
      "step": 177
    },
    {
      "epoch": 0.1545138888888889,
      "grad_norm": 0.0015683915698900819,
      "learning_rate": 0.00018950911498140396,
      "loss": 11.5,
      "step": 178
    },
    {
      "epoch": 0.15538194444444445,
      "grad_norm": 0.00172493327409029,
      "learning_rate": 0.00018938611590110584,
      "loss": 11.5,
      "step": 179
    },
    {
      "epoch": 0.15625,
      "grad_norm": 0.0014707579975947738,
      "learning_rate": 0.00018926244036827065,
      "loss": 11.5,
      "step": 180
    },
    {
      "epoch": 0.15711805555555555,
      "grad_norm": 0.00042890908662229776,
      "learning_rate": 0.00018913808931884498,
      "loss": 11.5,
      "step": 181
    },
    {
      "epoch": 0.1579861111111111,
      "grad_norm": 0.0008432182366959751,
      "learning_rate": 0.00018901306369388754,
      "loss": 11.5,
      "step": 182
    },
    {
      "epoch": 0.15885416666666666,
      "grad_norm": 0.0010861698538064957,
      "learning_rate": 0.00018888736443956202,
      "loss": 11.5,
      "step": 183
    },
    {
      "epoch": 0.1597222222222222,
      "grad_norm": 0.0022869682870805264,
      "learning_rate": 0.00018876099250713011,
      "loss": 11.5,
      "step": 184
    },
    {
      "epoch": 0.1605902777777778,
      "grad_norm": 0.0013702702708542347,
      "learning_rate": 0.00018863394885294398,
      "loss": 11.5,
      "step": 185
    },
    {
      "epoch": 0.16145833333333334,
      "grad_norm": 0.001422468340024352,
      "learning_rate": 0.00018850623443843935,
      "loss": 11.5,
      "step": 186
    },
    {
      "epoch": 0.1623263888888889,
      "grad_norm": 0.0011830556904897094,
      "learning_rate": 0.00018837785023012807,
      "loss": 11.5,
      "step": 187
    },
    {
      "epoch": 0.16319444444444445,
      "grad_norm": 0.0010504068341106176,
      "learning_rate": 0.00018824879719959086,
      "loss": 11.5,
      "step": 188
    },
    {
      "epoch": 0.1640625,
      "grad_norm": 0.0007689615595154464,
      "learning_rate": 0.0001881190763234698,
      "loss": 11.5,
      "step": 189
    },
    {
      "epoch": 0.16493055555555555,
      "grad_norm": 0.0009464145405218005,
      "learning_rate": 0.00018798868858346124,
      "loss": 11.5,
      "step": 190
    },
    {
      "epoch": 0.1657986111111111,
      "grad_norm": 0.0015288802096620202,
      "learning_rate": 0.00018785763496630806,
      "loss": 11.5,
      "step": 191
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.0012186490930616856,
      "learning_rate": 0.00018772591646379238,
      "loss": 11.5,
      "step": 192
    },
    {
      "epoch": 0.1675347222222222,
      "grad_norm": 0.0012458726996555924,
      "learning_rate": 0.00018759353407272805,
      "loss": 11.5,
      "step": 193
    },
    {
      "epoch": 0.1684027777777778,
      "grad_norm": 0.0017216532723978162,
      "learning_rate": 0.00018746048879495297,
      "loss": 11.5,
      "step": 194
    },
    {
      "epoch": 0.16927083333333334,
      "grad_norm": 0.0011988511541858315,
      "learning_rate": 0.0001873267816373217,
      "loss": 11.5,
      "step": 195
    },
    {
      "epoch": 0.1701388888888889,
      "grad_norm": 0.0017185450997203588,
      "learning_rate": 0.00018719241361169777,
      "loss": 11.5,
      "step": 196
    },
    {
      "epoch": 0.17100694444444445,
      "grad_norm": 0.002013278426602483,
      "learning_rate": 0.00018705738573494588,
      "loss": 11.5,
      "step": 197
    },
    {
      "epoch": 0.171875,
      "grad_norm": 0.002573801204562187,
      "learning_rate": 0.00018692169902892442,
      "loss": 11.5,
      "step": 198
    },
    {
      "epoch": 0.17274305555555555,
      "grad_norm": 0.0012108241207897663,
      "learning_rate": 0.00018678535452047759,
      "loss": 11.5,
      "step": 199
    },
    {
      "epoch": 0.1736111111111111,
      "grad_norm": 0.0010977673809975386,
      "learning_rate": 0.0001866483532414277,
      "loss": 11.5,
      "step": 200
    },
    {
      "epoch": 0.17447916666666666,
      "grad_norm": 0.002036850433796644,
      "learning_rate": 0.0001865106962285674,
      "loss": 11.5,
      "step": 201
    },
    {
      "epoch": 0.1753472222222222,
      "grad_norm": 0.0010651119519025087,
      "learning_rate": 0.00018637238452365163,
      "loss": 11.5,
      "step": 202
    },
    {
      "epoch": 0.1762152777777778,
      "grad_norm": 0.0016260347329080105,
      "learning_rate": 0.00018623341917339,
      "loss": 11.5,
      "step": 203
    },
    {
      "epoch": 0.17708333333333334,
      "grad_norm": 0.002315690042451024,
      "learning_rate": 0.00018609380122943875,
      "loss": 11.5,
      "step": 204
    },
    {
      "epoch": 0.1779513888888889,
      "grad_norm": 0.0019365769112482667,
      "learning_rate": 0.00018595353174839273,
      "loss": 11.5,
      "step": 205
    },
    {
      "epoch": 0.17881944444444445,
      "grad_norm": 0.002083905739709735,
      "learning_rate": 0.00018581261179177752,
      "loss": 11.5,
      "step": 206
    },
    {
      "epoch": 0.1796875,
      "grad_norm": 0.002875610487535596,
      "learning_rate": 0.0001856710424260413,
      "loss": 11.5,
      "step": 207
    },
    {
      "epoch": 0.18055555555555555,
      "grad_norm": 0.001979454653337598,
      "learning_rate": 0.00018552882472254684,
      "loss": 11.5,
      "step": 208
    },
    {
      "epoch": 0.1814236111111111,
      "grad_norm": 0.0018453949596732855,
      "learning_rate": 0.00018538595975756334,
      "loss": 11.5,
      "step": 209
    },
    {
      "epoch": 0.18229166666666666,
      "grad_norm": 0.0016776544507592916,
      "learning_rate": 0.0001852424486122584,
      "loss": 11.5,
      "step": 210
    },
    {
      "epoch": 0.1831597222222222,
      "grad_norm": 0.0013891557464376092,
      "learning_rate": 0.00018509829237268975,
      "loss": 11.5,
      "step": 211
    },
    {
      "epoch": 0.1840277777777778,
      "grad_norm": 0.002156435512006283,
      "learning_rate": 0.00018495349212979684,
      "loss": 11.5,
      "step": 212
    },
    {
      "epoch": 0.18489583333333334,
      "grad_norm": 0.002287535695359111,
      "learning_rate": 0.0001848080489793931,
      "loss": 11.5,
      "step": 213
    },
    {
      "epoch": 0.1857638888888889,
      "grad_norm": 0.0013429473619908094,
      "learning_rate": 0.00018466196402215708,
      "loss": 11.5,
      "step": 214
    },
    {
      "epoch": 0.18663194444444445,
      "grad_norm": 0.0012768427841365337,
      "learning_rate": 0.00018451523836362447,
      "loss": 11.5,
      "step": 215
    },
    {
      "epoch": 0.1875,
      "grad_norm": 0.0007630002219229937,
      "learning_rate": 0.00018436787311417962,
      "loss": 11.5,
      "step": 216
    },
    {
      "epoch": 0.18836805555555555,
      "grad_norm": 0.0021615992300212383,
      "learning_rate": 0.00018421986938904714,
      "loss": 11.5,
      "step": 217
    },
    {
      "epoch": 0.1892361111111111,
      "grad_norm": 0.00131071952637285,
      "learning_rate": 0.0001840712283082835,
      "loss": 11.5,
      "step": 218
    },
    {
      "epoch": 0.19010416666666666,
      "grad_norm": 0.0032089310698211193,
      "learning_rate": 0.00018392195099676845,
      "loss": 11.5,
      "step": 219
    },
    {
      "epoch": 0.1909722222222222,
      "grad_norm": 0.0016311652725562453,
      "learning_rate": 0.0001837720385841967,
      "loss": 11.5,
      "step": 220
    },
    {
      "epoch": 0.1918402777777778,
      "grad_norm": 0.0013007306260988116,
      "learning_rate": 0.00018362149220506915,
      "loss": 11.5,
      "step": 221
    },
    {
      "epoch": 0.19270833333333334,
      "grad_norm": 0.0016422068001702428,
      "learning_rate": 0.00018347031299868443,
      "loss": 11.5,
      "step": 222
    },
    {
      "epoch": 0.1935763888888889,
      "grad_norm": 0.0008372614393010736,
      "learning_rate": 0.00018331850210913023,
      "loss": 11.5,
      "step": 223
    },
    {
      "epoch": 0.19444444444444445,
      "grad_norm": 0.0019389386288821697,
      "learning_rate": 0.00018316606068527477,
      "loss": 11.5,
      "step": 224
    },
    {
      "epoch": 0.1953125,
      "grad_norm": 0.0011536555830389261,
      "learning_rate": 0.0001830129898807579,
      "loss": 11.5,
      "step": 225
    },
    {
      "epoch": 0.19618055555555555,
      "grad_norm": 0.0015878123231232166,
      "learning_rate": 0.00018285929085398248,
      "loss": 11.5,
      "step": 226
    },
    {
      "epoch": 0.1970486111111111,
      "grad_norm": 0.001472399802878499,
      "learning_rate": 0.00018270496476810563,
      "loss": 11.5,
      "step": 227
    },
    {
      "epoch": 0.19791666666666666,
      "grad_norm": 0.0018204131629317999,
      "learning_rate": 0.00018255001279102992,
      "loss": 11.5,
      "step": 228
    },
    {
      "epoch": 0.1987847222222222,
      "grad_norm": 0.002056293422356248,
      "learning_rate": 0.00018239443609539447,
      "loss": 11.5,
      "step": 229
    },
    {
      "epoch": 0.1996527777777778,
      "grad_norm": 0.002107437467202544,
      "learning_rate": 0.00018223823585856614,
      "loss": 11.5,
      "step": 230
    },
    {
      "epoch": 0.20052083333333334,
      "grad_norm": 0.0024665426462888718,
      "learning_rate": 0.00018208141326263059,
      "loss": 11.5,
      "step": 231
    },
    {
      "epoch": 0.2013888888888889,
      "grad_norm": 0.0012031730730086565,
      "learning_rate": 0.00018192396949438336,
      "loss": 11.5,
      "step": 232
    },
    {
      "epoch": 0.20225694444444445,
      "grad_norm": 0.0017438753275200725,
      "learning_rate": 0.00018176590574532083,
      "loss": 11.5,
      "step": 233
    },
    {
      "epoch": 0.203125,
      "grad_norm": 0.0008248971425928175,
      "learning_rate": 0.00018160722321163136,
      "loss": 11.5,
      "step": 234
    },
    {
      "epoch": 0.20399305555555555,
      "grad_norm": 0.0013439902104437351,
      "learning_rate": 0.000181447923094186,
      "loss": 11.5,
      "step": 235
    },
    {
      "epoch": 0.2048611111111111,
      "grad_norm": 0.002117580734193325,
      "learning_rate": 0.00018128800659852952,
      "loss": 11.5,
      "step": 236
    },
    {
      "epoch": 0.20572916666666666,
      "grad_norm": 0.0016160148661583662,
      "learning_rate": 0.00018112747493487138,
      "loss": 11.5,
      "step": 237
    },
    {
      "epoch": 0.2065972222222222,
      "grad_norm": 0.001116547267884016,
      "learning_rate": 0.00018096632931807643,
      "loss": 11.5,
      "step": 238
    },
    {
      "epoch": 0.2074652777777778,
      "grad_norm": 0.0014612178783863783,
      "learning_rate": 0.0001808045709676558,
      "loss": 11.5,
      "step": 239
    },
    {
      "epoch": 0.20833333333333334,
      "grad_norm": 0.0021290243603289127,
      "learning_rate": 0.0001806422011077576,
      "loss": 11.5,
      "step": 240
    },
    {
      "epoch": 0.2092013888888889,
      "grad_norm": 0.002343961503356695,
      "learning_rate": 0.0001804792209671577,
      "loss": 11.5,
      "step": 241
    },
    {
      "epoch": 0.21006944444444445,
      "grad_norm": 0.0014112481148913503,
      "learning_rate": 0.00018031563177925048,
      "loss": 11.5,
      "step": 242
    },
    {
      "epoch": 0.2109375,
      "grad_norm": 0.0021090067457407713,
      "learning_rate": 0.00018015143478203937,
      "loss": 11.5,
      "step": 243
    },
    {
      "epoch": 0.21180555555555555,
      "grad_norm": 0.0018467740155756474,
      "learning_rate": 0.00017998663121812763,
      "loss": 11.5,
      "step": 244
    },
    {
      "epoch": 0.2126736111111111,
      "grad_norm": 0.0012689410941675305,
      "learning_rate": 0.00017982122233470876,
      "loss": 11.5,
      "step": 245
    },
    {
      "epoch": 0.21354166666666666,
      "grad_norm": 0.0022510108537971973,
      "learning_rate": 0.00017965520938355729,
      "loss": 11.5,
      "step": 246
    },
    {
      "epoch": 0.2144097222222222,
      "grad_norm": 0.0016390194650739431,
      "learning_rate": 0.00017948859362101915,
      "loss": 11.5,
      "step": 247
    },
    {
      "epoch": 0.2152777777777778,
      "grad_norm": 0.0006908321520313621,
      "learning_rate": 0.00017932137630800217,
      "loss": 11.5,
      "step": 248
    },
    {
      "epoch": 0.21614583333333334,
      "grad_norm": 0.0029407439287751913,
      "learning_rate": 0.00017915355870996654,
      "loss": 11.5,
      "step": 249
    },
    {
      "epoch": 0.2170138888888889,
      "grad_norm": 0.0011984260054305196,
      "learning_rate": 0.00017898514209691545,
      "loss": 11.5,
      "step": 250
    },
    {
      "epoch": 0.21788194444444445,
      "grad_norm": 0.0019082878716289997,
      "learning_rate": 0.000178816127743385,
      "loss": 11.5,
      "step": 251
    },
    {
      "epoch": 0.21875,
      "grad_norm": 0.003266829065978527,
      "learning_rate": 0.00017864651692843509,
      "loss": 11.5,
      "step": 252
    },
    {
      "epoch": 0.21961805555555555,
      "grad_norm": 0.0011768423719331622,
      "learning_rate": 0.00017847631093563938,
      "loss": 11.5,
      "step": 253
    },
    {
      "epoch": 0.2204861111111111,
      "grad_norm": 0.001366365235298872,
      "learning_rate": 0.00017830551105307572,
      "loss": 11.5,
      "step": 254
    },
    {
      "epoch": 0.22135416666666666,
      "grad_norm": 0.0025889321696013212,
      "learning_rate": 0.00017813411857331635,
      "loss": 11.5,
      "step": 255
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.001713864621706307,
      "learning_rate": 0.00017796213479341817,
      "loss": 11.5,
      "step": 256
    },
    {
      "epoch": 0.2230902777777778,
      "grad_norm": 0.0021035310346633196,
      "learning_rate": 0.0001777895610149129,
      "loss": 11.5,
      "step": 257
    },
    {
      "epoch": 0.22395833333333334,
      "grad_norm": 0.0010583502007648349,
      "learning_rate": 0.0001776163985437972,
      "loss": 11.5,
      "step": 258
    },
    {
      "epoch": 0.2248263888888889,
      "grad_norm": 0.001230224734172225,
      "learning_rate": 0.00017744264869052286,
      "loss": 11.5,
      "step": 259
    },
    {
      "epoch": 0.22569444444444445,
      "grad_norm": 0.0014014148619025946,
      "learning_rate": 0.00017726831276998681,
      "loss": 11.5,
      "step": 260
    },
    {
      "epoch": 0.2265625,
      "grad_norm": 0.00228292029350996,
      "learning_rate": 0.00017709339210152117,
      "loss": 11.5,
      "step": 261
    },
    {
      "epoch": 0.22743055555555555,
      "grad_norm": 0.0007677875109948218,
      "learning_rate": 0.00017691788800888326,
      "loss": 11.5,
      "step": 262
    },
    {
      "epoch": 0.2282986111111111,
      "grad_norm": 0.0013327487977221608,
      "learning_rate": 0.00017674180182024579,
      "loss": 11.5,
      "step": 263
    },
    {
      "epoch": 0.22916666666666666,
      "grad_norm": 0.0012688027927652001,
      "learning_rate": 0.0001765651348681864,
      "loss": 11.5,
      "step": 264
    },
    {
      "epoch": 0.2300347222222222,
      "grad_norm": 0.0012343355920165777,
      "learning_rate": 0.00017638788848967795,
      "loss": 11.5,
      "step": 265
    },
    {
      "epoch": 0.2309027777777778,
      "grad_norm": 0.0013646919978782535,
      "learning_rate": 0.00017621006402607827,
      "loss": 11.5,
      "step": 266
    },
    {
      "epoch": 0.23177083333333334,
      "grad_norm": 0.002846561837941408,
      "learning_rate": 0.00017603166282311987,
      "loss": 11.5,
      "step": 267
    },
    {
      "epoch": 0.2326388888888889,
      "grad_norm": 0.0011637487914413214,
      "learning_rate": 0.00017585268623090004,
      "loss": 11.5,
      "step": 268
    },
    {
      "epoch": 0.23350694444444445,
      "grad_norm": 0.0020906825084239244,
      "learning_rate": 0.0001756731356038704,
      "loss": 11.5,
      "step": 269
    },
    {
      "epoch": 0.234375,
      "grad_norm": 0.0006968987290747464,
      "learning_rate": 0.00017549301230082663,
      "loss": 11.5,
      "step": 270
    },
    {
      "epoch": 0.23524305555555555,
      "grad_norm": 0.0015382887795567513,
      "learning_rate": 0.0001753123176848985,
      "loss": 11.5,
      "step": 271
    },
    {
      "epoch": 0.2361111111111111,
      "grad_norm": 0.001084989751689136,
      "learning_rate": 0.00017513105312353918,
      "loss": 11.5,
      "step": 272
    },
    {
      "epoch": 0.23697916666666666,
      "grad_norm": 0.0012109302915632725,
      "learning_rate": 0.00017494921998851507,
      "loss": 11.5,
      "step": 273
    },
    {
      "epoch": 0.2378472222222222,
      "grad_norm": 0.0020071074832230806,
      "learning_rate": 0.00017476681965589548,
      "loss": 11.5,
      "step": 274
    },
    {
      "epoch": 0.2387152777777778,
      "grad_norm": 0.0008635442354716361,
      "learning_rate": 0.00017458385350604204,
      "loss": 11.5,
      "step": 275
    },
    {
      "epoch": 0.23958333333333334,
      "grad_norm": 0.0016295452369377017,
      "learning_rate": 0.0001744003229235984,
      "loss": 11.5,
      "step": 276
    },
    {
      "epoch": 0.2404513888888889,
      "grad_norm": 0.001285732607357204,
      "learning_rate": 0.00017421622929747964,
      "loss": 11.5,
      "step": 277
    },
    {
      "epoch": 0.24131944444444445,
      "grad_norm": 0.002172011649236083,
      "learning_rate": 0.00017403157402086193,
      "loss": 11.5,
      "step": 278
    },
    {
      "epoch": 0.2421875,
      "grad_norm": 0.0017907707951962948,
      "learning_rate": 0.00017384635849117171,
      "loss": 11.5,
      "step": 279
    },
    {
      "epoch": 0.24305555555555555,
      "grad_norm": 0.0008966317982412875,
      "learning_rate": 0.00017366058411007545,
      "loss": 11.5,
      "step": 280
    },
    {
      "epoch": 0.2439236111111111,
      "grad_norm": 0.0013841426698490977,
      "learning_rate": 0.0001734742522834688,
      "loss": 11.5,
      "step": 281
    },
    {
      "epoch": 0.24479166666666666,
      "grad_norm": 0.0019019154133275151,
      "learning_rate": 0.00017328736442146595,
      "loss": 11.5,
      "step": 282
    },
    {
      "epoch": 0.2456597222222222,
      "grad_norm": 0.0021870718337595463,
      "learning_rate": 0.0001730999219383892,
      "loss": 11.5,
      "step": 283
    },
    {
      "epoch": 0.2465277777777778,
      "grad_norm": 0.0010026195086538792,
      "learning_rate": 0.0001729119262527579,
      "loss": 11.5,
      "step": 284
    },
    {
      "epoch": 0.24739583333333334,
      "grad_norm": 0.0006980384350754321,
      "learning_rate": 0.00017272337878727808,
      "loss": 11.5,
      "step": 285
    },
    {
      "epoch": 0.2482638888888889,
      "grad_norm": 0.001652008737437427,
      "learning_rate": 0.00017253428096883138,
      "loss": 11.5,
      "step": 286
    },
    {
      "epoch": 0.24913194444444445,
      "grad_norm": 0.0022391362581402063,
      "learning_rate": 0.00017234463422846443,
      "loss": 11.5,
      "step": 287
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.0020901914685964584,
      "learning_rate": 0.00017215444000137796,
      "loss": 11.5,
      "step": 288
    },
    {
      "epoch": 0.25,
      "eval_loss": 11.5,
      "eval_runtime": 3.9706,
      "eval_samples_per_second": 122.148,
      "eval_steps_per_second": 61.2,
      "step": 288
    },
    {
      "epoch": 0.2508680555555556,
      "grad_norm": 0.0019298329716548324,
      "learning_rate": 0.00017196369972691592,
      "loss": 11.5,
      "step": 289
    },
    {
      "epoch": 0.2517361111111111,
      "grad_norm": 0.0019217822700738907,
      "learning_rate": 0.00017177241484855467,
      "loss": 11.5,
      "step": 290
    },
    {
      "epoch": 0.2526041666666667,
      "grad_norm": 0.0021822769194841385,
      "learning_rate": 0.00017158058681389192,
      "loss": 11.5,
      "step": 291
    },
    {
      "epoch": 0.2534722222222222,
      "grad_norm": 0.0018070607911795378,
      "learning_rate": 0.00017138821707463598,
      "loss": 11.5,
      "step": 292
    },
    {
      "epoch": 0.2543402777777778,
      "grad_norm": 0.0015647222753614187,
      "learning_rate": 0.00017119530708659456,
      "loss": 11.5,
      "step": 293
    },
    {
      "epoch": 0.2552083333333333,
      "grad_norm": 0.0014673898695036769,
      "learning_rate": 0.0001710018583096638,
      "loss": 11.5,
      "step": 294
    },
    {
      "epoch": 0.2560763888888889,
      "grad_norm": 0.002169738756492734,
      "learning_rate": 0.0001708078722078174,
      "loss": 11.5,
      "step": 295
    },
    {
      "epoch": 0.2569444444444444,
      "grad_norm": 0.003958852030336857,
      "learning_rate": 0.00017061335024909528,
      "loss": 11.5,
      "step": 296
    },
    {
      "epoch": 0.2578125,
      "grad_norm": 0.003060359740629792,
      "learning_rate": 0.00017041829390559263,
      "loss": 11.5,
      "step": 297
    },
    {
      "epoch": 0.2586805555555556,
      "grad_norm": 0.001239659497514367,
      "learning_rate": 0.00017022270465344883,
      "loss": 11.5,
      "step": 298
    },
    {
      "epoch": 0.2595486111111111,
      "grad_norm": 0.002644512802362442,
      "learning_rate": 0.000170026583972836,
      "loss": 11.5,
      "step": 299
    },
    {
      "epoch": 0.2604166666666667,
      "grad_norm": 0.00640529440715909,
      "learning_rate": 0.00016982993334794817,
      "loss": 11.5,
      "step": 300
    },
    {
      "epoch": 0.2612847222222222,
      "grad_norm": 0.0017492524348199368,
      "learning_rate": 0.00016963275426698973,
      "loss": 11.5,
      "step": 301
    },
    {
      "epoch": 0.2621527777777778,
      "grad_norm": 0.0015452845254912972,
      "learning_rate": 0.0001694350482221643,
      "loss": 11.5,
      "step": 302
    },
    {
      "epoch": 0.2630208333333333,
      "grad_norm": 0.0011449211742728949,
      "learning_rate": 0.0001692368167096636,
      "loss": 11.5,
      "step": 303
    },
    {
      "epoch": 0.2638888888888889,
      "grad_norm": 0.0007032388239167631,
      "learning_rate": 0.00016903806122965572,
      "loss": 11.5,
      "step": 304
    },
    {
      "epoch": 0.2647569444444444,
      "grad_norm": 0.001266799634322524,
      "learning_rate": 0.0001688387832862742,
      "loss": 11.5,
      "step": 305
    },
    {
      "epoch": 0.265625,
      "grad_norm": 0.001181058818474412,
      "learning_rate": 0.0001686389843876064,
      "loss": 11.5,
      "step": 306
    },
    {
      "epoch": 0.2664930555555556,
      "grad_norm": 0.0011826750123873353,
      "learning_rate": 0.00016843866604568204,
      "loss": 11.5,
      "step": 307
    },
    {
      "epoch": 0.2673611111111111,
      "grad_norm": 0.0017107395688071847,
      "learning_rate": 0.00016823782977646204,
      "loss": 11.5,
      "step": 308
    },
    {
      "epoch": 0.2682291666666667,
      "grad_norm": 0.0011749992845579982,
      "learning_rate": 0.00016803647709982676,
      "loss": 11.5,
      "step": 309
    },
    {
      "epoch": 0.2690972222222222,
      "grad_norm": 0.000666073348838836,
      "learning_rate": 0.00016783460953956458,
      "loss": 11.5,
      "step": 310
    },
    {
      "epoch": 0.2699652777777778,
      "grad_norm": 0.0011788930278271437,
      "learning_rate": 0.00016763222862336053,
      "loss": 11.5,
      "step": 311
    },
    {
      "epoch": 0.2708333333333333,
      "grad_norm": 0.001143649802543223,
      "learning_rate": 0.0001674293358827844,
      "loss": 11.5,
      "step": 312
    },
    {
      "epoch": 0.2717013888888889,
      "grad_norm": 0.0015109138330444694,
      "learning_rate": 0.00016722593285327953,
      "loss": 11.5,
      "step": 313
    },
    {
      "epoch": 0.2725694444444444,
      "grad_norm": 0.0013187967706471682,
      "learning_rate": 0.00016702202107415088,
      "loss": 11.5,
      "step": 314
    },
    {
      "epoch": 0.2734375,
      "grad_norm": 0.0016621032264083624,
      "learning_rate": 0.0001668176020885536,
      "loss": 11.5,
      "step": 315
    },
    {
      "epoch": 0.2743055555555556,
      "grad_norm": 0.0016569121507927775,
      "learning_rate": 0.00016661267744348112,
      "loss": 11.5,
      "step": 316
    },
    {
      "epoch": 0.2751736111111111,
      "grad_norm": 0.0009634601301513612,
      "learning_rate": 0.00016640724868975375,
      "loss": 11.5,
      "step": 317
    },
    {
      "epoch": 0.2760416666666667,
      "grad_norm": 0.0014234816189855337,
      "learning_rate": 0.00016620131738200663,
      "loss": 11.5,
      "step": 318
    },
    {
      "epoch": 0.2769097222222222,
      "grad_norm": 0.002640510443598032,
      "learning_rate": 0.00016599488507867816,
      "loss": 11.5,
      "step": 319
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 0.00093653000658378,
      "learning_rate": 0.00016578795334199818,
      "loss": 11.5,
      "step": 320
    },
    {
      "epoch": 0.2786458333333333,
      "grad_norm": 0.0017026716377586126,
      "learning_rate": 0.00016558052373797608,
      "loss": 11.5,
      "step": 321
    },
    {
      "epoch": 0.2795138888888889,
      "grad_norm": 0.0012330121826380491,
      "learning_rate": 0.000165372597836389,
      "loss": 11.5,
      "step": 322
    },
    {
      "epoch": 0.2803819444444444,
      "grad_norm": 0.0021235134918242693,
      "learning_rate": 0.00016516417721076996,
      "loss": 11.5,
      "step": 323
    },
    {
      "epoch": 0.28125,
      "grad_norm": 0.002146722050383687,
      "learning_rate": 0.0001649552634383959,
      "loss": 11.5,
      "step": 324
    },
    {
      "epoch": 0.2821180555555556,
      "grad_norm": 0.0016347194323316216,
      "learning_rate": 0.00016474585810027584,
      "loss": 11.5,
      "step": 325
    },
    {
      "epoch": 0.2829861111111111,
      "grad_norm": 0.0016562105156481266,
      "learning_rate": 0.00016453596278113877,
      "loss": 11.5,
      "step": 326
    },
    {
      "epoch": 0.2838541666666667,
      "grad_norm": 0.0012435222743079066,
      "learning_rate": 0.00016432557906942178,
      "loss": 11.5,
      "step": 327
    },
    {
      "epoch": 0.2847222222222222,
      "grad_norm": 0.001985622337087989,
      "learning_rate": 0.00016411470855725793,
      "loss": 11.5,
      "step": 328
    },
    {
      "epoch": 0.2855902777777778,
      "grad_norm": 0.0017628187779337168,
      "learning_rate": 0.0001639033528404644,
      "loss": 11.5,
      "step": 329
    },
    {
      "epoch": 0.2864583333333333,
      "grad_norm": 0.001387462136335671,
      "learning_rate": 0.00016369151351853023,
      "loss": 11.5,
      "step": 330
    },
    {
      "epoch": 0.2873263888888889,
      "grad_norm": 0.001489591202698648,
      "learning_rate": 0.00016347919219460416,
      "loss": 11.5,
      "step": 331
    },
    {
      "epoch": 0.2881944444444444,
      "grad_norm": 0.001772079966031015,
      "learning_rate": 0.0001632663904754828,
      "loss": 11.5,
      "step": 332
    },
    {
      "epoch": 0.2890625,
      "grad_norm": 0.0013231280026957393,
      "learning_rate": 0.00016305310997159804,
      "loss": 11.5,
      "step": 333
    },
    {
      "epoch": 0.2899305555555556,
      "grad_norm": 0.0023159291595220566,
      "learning_rate": 0.0001628393522970053,
      "loss": 11.5,
      "step": 334
    },
    {
      "epoch": 0.2907986111111111,
      "grad_norm": 0.0009754012571647763,
      "learning_rate": 0.00016262511906937104,
      "loss": 11.5,
      "step": 335
    },
    {
      "epoch": 0.2916666666666667,
      "grad_norm": 0.002182697644457221,
      "learning_rate": 0.00016241041190996055,
      "loss": 11.5,
      "step": 336
    },
    {
      "epoch": 0.2925347222222222,
      "grad_norm": 0.001281126169487834,
      "learning_rate": 0.0001621952324436258,
      "loss": 11.5,
      "step": 337
    },
    {
      "epoch": 0.2934027777777778,
      "grad_norm": 0.0026281271129846573,
      "learning_rate": 0.00016197958229879298,
      "loss": 11.5,
      "step": 338
    },
    {
      "epoch": 0.2942708333333333,
      "grad_norm": 0.0017629140056669712,
      "learning_rate": 0.00016176346310745033,
      "loss": 11.5,
      "step": 339
    },
    {
      "epoch": 0.2951388888888889,
      "grad_norm": 0.0013849474489688873,
      "learning_rate": 0.00016154687650513566,
      "loss": 11.5,
      "step": 340
    },
    {
      "epoch": 0.2960069444444444,
      "grad_norm": 0.0011748202377930284,
      "learning_rate": 0.0001613298241309241,
      "loss": 11.5,
      "step": 341
    },
    {
      "epoch": 0.296875,
      "grad_norm": 0.0012899189023301005,
      "learning_rate": 0.0001611123076274156,
      "loss": 11.5,
      "step": 342
    },
    {
      "epoch": 0.2977430555555556,
      "grad_norm": 0.0013365561608225107,
      "learning_rate": 0.0001608943286407225,
      "loss": 11.5,
      "step": 343
    },
    {
      "epoch": 0.2986111111111111,
      "grad_norm": 0.0031333351507782936,
      "learning_rate": 0.00016067588882045714,
      "loss": 11.5,
      "step": 344
    },
    {
      "epoch": 0.2994791666666667,
      "grad_norm": 0.0016387298237532377,
      "learning_rate": 0.00016045698981971934,
      "loss": 11.5,
      "step": 345
    },
    {
      "epoch": 0.3003472222222222,
      "grad_norm": 0.001601073774509132,
      "learning_rate": 0.00016023763329508386,
      "loss": 11.5,
      "step": 346
    },
    {
      "epoch": 0.3012152777777778,
      "grad_norm": 0.0020099536050111055,
      "learning_rate": 0.0001600178209065879,
      "loss": 11.5,
      "step": 347
    },
    {
      "epoch": 0.3020833333333333,
      "grad_norm": 0.0014927750453352928,
      "learning_rate": 0.00015979755431771848,
      "loss": 11.5,
      "step": 348
    },
    {
      "epoch": 0.3029513888888889,
      "grad_norm": 0.000984087004326284,
      "learning_rate": 0.0001595768351954,
      "loss": 11.5,
      "step": 349
    },
    {
      "epoch": 0.3038194444444444,
      "grad_norm": 0.0013183153932914138,
      "learning_rate": 0.00015935566520998147,
      "loss": 11.5,
      "step": 350
    },
    {
      "epoch": 0.3046875,
      "grad_norm": 0.0018050584476441145,
      "learning_rate": 0.0001591340460352239,
      "loss": 11.5,
      "step": 351
    },
    {
      "epoch": 0.3055555555555556,
      "grad_norm": 0.0020948173478245735,
      "learning_rate": 0.0001589119793482877,
      "loss": 11.5,
      "step": 352
    },
    {
      "epoch": 0.3064236111111111,
      "grad_norm": 0.000849515781737864,
      "learning_rate": 0.00015868946682971995,
      "loss": 11.5,
      "step": 353
    },
    {
      "epoch": 0.3072916666666667,
      "grad_norm": 0.0008011907339096069,
      "learning_rate": 0.00015846651016344159,
      "loss": 11.5,
      "step": 354
    },
    {
      "epoch": 0.3081597222222222,
      "grad_norm": 0.001497761346399784,
      "learning_rate": 0.00015824311103673484,
      "loss": 11.5,
      "step": 355
    },
    {
      "epoch": 0.3090277777777778,
      "grad_norm": 0.0009178329491987824,
      "learning_rate": 0.00015801927114023035,
      "loss": 11.5,
      "step": 356
    },
    {
      "epoch": 0.3098958333333333,
      "grad_norm": 0.00253950129263103,
      "learning_rate": 0.00015779499216789442,
      "loss": 11.5,
      "step": 357
    },
    {
      "epoch": 0.3107638888888889,
      "grad_norm": 0.0016381394816562533,
      "learning_rate": 0.00015757027581701612,
      "loss": 11.5,
      "step": 358
    },
    {
      "epoch": 0.3116319444444444,
      "grad_norm": 0.0018246745457872748,
      "learning_rate": 0.0001573451237881945,
      "loss": 11.5,
      "step": 359
    },
    {
      "epoch": 0.3125,
      "grad_norm": 0.001299652736634016,
      "learning_rate": 0.00015711953778532576,
      "loss": 11.5,
      "step": 360
    },
    {
      "epoch": 0.3133680555555556,
      "grad_norm": 0.001208305126056075,
      "learning_rate": 0.0001568935195155903,
      "loss": 11.5,
      "step": 361
    },
    {
      "epoch": 0.3142361111111111,
      "grad_norm": 0.0033954232931137085,
      "learning_rate": 0.00015666707068943974,
      "loss": 11.5,
      "step": 362
    },
    {
      "epoch": 0.3151041666666667,
      "grad_norm": 0.0019799405708909035,
      "learning_rate": 0.0001564401930205842,
      "loss": 11.5,
      "step": 363
    },
    {
      "epoch": 0.3159722222222222,
      "grad_norm": 0.0015467420453205705,
      "learning_rate": 0.00015621288822597902,
      "loss": 11.5,
      "step": 364
    },
    {
      "epoch": 0.3168402777777778,
      "grad_norm": 0.0009759374079294503,
      "learning_rate": 0.00015598515802581195,
      "loss": 11.5,
      "step": 365
    },
    {
      "epoch": 0.3177083333333333,
      "grad_norm": 0.0012391997734084725,
      "learning_rate": 0.00015575700414349023,
      "loss": 11.5,
      "step": 366
    },
    {
      "epoch": 0.3185763888888889,
      "grad_norm": 0.0012746209977194667,
      "learning_rate": 0.00015552842830562734,
      "loss": 11.5,
      "step": 367
    },
    {
      "epoch": 0.3194444444444444,
      "grad_norm": 0.0011782745132222772,
      "learning_rate": 0.00015529943224202995,
      "loss": 11.5,
      "step": 368
    },
    {
      "epoch": 0.3203125,
      "grad_norm": 0.0016793364193290472,
      "learning_rate": 0.00015507001768568502,
      "loss": 11.5,
      "step": 369
    },
    {
      "epoch": 0.3211805555555556,
      "grad_norm": 0.0011946028098464012,
      "learning_rate": 0.00015484018637274645,
      "loss": 11.5,
      "step": 370
    },
    {
      "epoch": 0.3220486111111111,
      "grad_norm": 0.001993674784898758,
      "learning_rate": 0.00015460994004252212,
      "loss": 11.5,
      "step": 371
    },
    {
      "epoch": 0.3229166666666667,
      "grad_norm": 0.0018555393908172846,
      "learning_rate": 0.0001543792804374607,
      "loss": 11.5,
      "step": 372
    },
    {
      "epoch": 0.3237847222222222,
      "grad_norm": 0.0015802582493051887,
      "learning_rate": 0.00015414820930313833,
      "loss": 11.5,
      "step": 373
    },
    {
      "epoch": 0.3246527777777778,
      "grad_norm": 0.0031402644235640764,
      "learning_rate": 0.00015391672838824553,
      "loss": 11.5,
      "step": 374
    },
    {
      "epoch": 0.3255208333333333,
      "grad_norm": 0.0017931369366124272,
      "learning_rate": 0.000153684839444574,
      "loss": 11.5,
      "step": 375
    },
    {
      "epoch": 0.3263888888888889,
      "grad_norm": 0.0012212598230689764,
      "learning_rate": 0.00015345254422700327,
      "loss": 11.5,
      "step": 376
    },
    {
      "epoch": 0.3272569444444444,
      "grad_norm": 0.0010161130921915174,
      "learning_rate": 0.0001532198444934874,
      "loss": 11.5,
      "step": 377
    },
    {
      "epoch": 0.328125,
      "grad_norm": 0.001056469976902008,
      "learning_rate": 0.00015298674200504183,
      "loss": 11.5,
      "step": 378
    },
    {
      "epoch": 0.3289930555555556,
      "grad_norm": 0.0006380357081070542,
      "learning_rate": 0.00015275323852572984,
      "loss": 11.5,
      "step": 379
    },
    {
      "epoch": 0.3298611111111111,
      "grad_norm": 0.0009562419727444649,
      "learning_rate": 0.00015251933582264936,
      "loss": 11.5,
      "step": 380
    },
    {
      "epoch": 0.3307291666666667,
      "grad_norm": 0.001869362429715693,
      "learning_rate": 0.0001522850356659196,
      "loss": 11.5,
      "step": 381
    },
    {
      "epoch": 0.3315972222222222,
      "grad_norm": 0.0009193017613142729,
      "learning_rate": 0.00015205033982866752,
      "loss": 11.5,
      "step": 382
    },
    {
      "epoch": 0.3324652777777778,
      "grad_norm": 0.0016343846218660474,
      "learning_rate": 0.00015181525008701457,
      "loss": 11.5,
      "step": 383
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.0019053410505875945,
      "learning_rate": 0.00015157976822006313,
      "loss": 11.5,
      "step": 384
    },
    {
      "epoch": 0.3342013888888889,
      "grad_norm": 0.0016181294340640306,
      "learning_rate": 0.0001513438960098831,
      "loss": 11.5,
      "step": 385
    },
    {
      "epoch": 0.3350694444444444,
      "grad_norm": 0.0011307490058243275,
      "learning_rate": 0.00015110763524149836,
      "loss": 11.5,
      "step": 386
    },
    {
      "epoch": 0.3359375,
      "grad_norm": 0.0013405167264863849,
      "learning_rate": 0.00015087098770287334,
      "loss": 11.5,
      "step": 387
    },
    {
      "epoch": 0.3368055555555556,
      "grad_norm": 0.001251003472134471,
      "learning_rate": 0.00015063395518489955,
      "loss": 11.5,
      "step": 388
    },
    {
      "epoch": 0.3376736111111111,
      "grad_norm": 0.0011935929069295526,
      "learning_rate": 0.00015039653948138176,
      "loss": 11.5,
      "step": 389
    },
    {
      "epoch": 0.3385416666666667,
      "grad_norm": 0.0010503452504053712,
      "learning_rate": 0.00015015874238902467,
      "loss": 11.5,
      "step": 390
    },
    {
      "epoch": 0.3394097222222222,
      "grad_norm": 0.0008192232344299555,
      "learning_rate": 0.0001499205657074193,
      "loss": 11.5,
      "step": 391
    },
    {
      "epoch": 0.3402777777777778,
      "grad_norm": 0.0026244446635246277,
      "learning_rate": 0.00014968201123902918,
      "loss": 11.5,
      "step": 392
    },
    {
      "epoch": 0.3411458333333333,
      "grad_norm": 0.0020733929704874754,
      "learning_rate": 0.00014944308078917702,
      "loss": 11.5,
      "step": 393
    },
    {
      "epoch": 0.3420138888888889,
      "grad_norm": 0.0019652587361633778,
      "learning_rate": 0.00014920377616603063,
      "loss": 11.5,
      "step": 394
    },
    {
      "epoch": 0.3428819444444444,
      "grad_norm": 0.0024613775312900543,
      "learning_rate": 0.00014896409918058968,
      "loss": 11.5,
      "step": 395
    },
    {
      "epoch": 0.34375,
      "grad_norm": 0.0014305724762380123,
      "learning_rate": 0.00014872405164667173,
      "loss": 11.5,
      "step": 396
    },
    {
      "epoch": 0.3446180555555556,
      "grad_norm": 0.0015622616047039628,
      "learning_rate": 0.00014848363538089846,
      "loss": 11.5,
      "step": 397
    },
    {
      "epoch": 0.3454861111111111,
      "grad_norm": 0.0018080153968185186,
      "learning_rate": 0.00014824285220268218,
      "loss": 11.5,
      "step": 398
    },
    {
      "epoch": 0.3463541666666667,
      "grad_norm": 0.0017224273178726435,
      "learning_rate": 0.0001480017039342118,
      "loss": 11.5,
      "step": 399
    },
    {
      "epoch": 0.3472222222222222,
      "grad_norm": 0.0019916968885809183,
      "learning_rate": 0.00014776019240043918,
      "loss": 11.5,
      "step": 400
    },
    {
      "epoch": 0.3480902777777778,
      "grad_norm": 0.001800294267013669,
      "learning_rate": 0.00014751831942906528,
      "loss": 11.5,
      "step": 401
    },
    {
      "epoch": 0.3489583333333333,
      "grad_norm": 0.001007885322906077,
      "learning_rate": 0.0001472760868505263,
      "loss": 11.5,
      "step": 402
    },
    {
      "epoch": 0.3498263888888889,
      "grad_norm": 0.0019039842300117016,
      "learning_rate": 0.0001470334964979799,
      "loss": 11.5,
      "step": 403
    },
    {
      "epoch": 0.3506944444444444,
      "grad_norm": 0.001605544239282608,
      "learning_rate": 0.00014679055020729134,
      "loss": 11.5,
      "step": 404
    },
    {
      "epoch": 0.3515625,
      "grad_norm": 0.0011697318404912949,
      "learning_rate": 0.0001465472498170194,
      "loss": 11.5,
      "step": 405
    },
    {
      "epoch": 0.3524305555555556,
      "grad_norm": 0.0013148639118298888,
      "learning_rate": 0.00014630359716840275,
      "loss": 11.5,
      "step": 406
    },
    {
      "epoch": 0.3532986111111111,
      "grad_norm": 0.0006943634361959994,
      "learning_rate": 0.0001460595941053457,
      "loss": 11.5,
      "step": 407
    },
    {
      "epoch": 0.3541666666666667,
      "grad_norm": 0.0010435678996145725,
      "learning_rate": 0.00014581524247440455,
      "loss": 11.5,
      "step": 408
    },
    {
      "epoch": 0.3550347222222222,
      "grad_norm": 0.002254709368571639,
      "learning_rate": 0.00014557054412477343,
      "loss": 11.5,
      "step": 409
    },
    {
      "epoch": 0.3559027777777778,
      "grad_norm": 0.0009712673490867019,
      "learning_rate": 0.00014532550090827034,
      "loss": 11.5,
      "step": 410
    },
    {
      "epoch": 0.3567708333333333,
      "grad_norm": 0.0015958211151883006,
      "learning_rate": 0.00014508011467932315,
      "loss": 11.5,
      "step": 411
    },
    {
      "epoch": 0.3576388888888889,
      "grad_norm": 0.0009054581169039011,
      "learning_rate": 0.00014483438729495562,
      "loss": 11.5,
      "step": 412
    },
    {
      "epoch": 0.3585069444444444,
      "grad_norm": 0.0014388291165232658,
      "learning_rate": 0.00014458832061477322,
      "loss": 11.5,
      "step": 413
    },
    {
      "epoch": 0.359375,
      "grad_norm": 0.0011277180165052414,
      "learning_rate": 0.00014434191650094911,
      "loss": 11.5,
      "step": 414
    },
    {
      "epoch": 0.3602430555555556,
      "grad_norm": 0.0007856336887925863,
      "learning_rate": 0.00014409517681821017,
      "loss": 11.5,
      "step": 415
    },
    {
      "epoch": 0.3611111111111111,
      "grad_norm": 0.000878229970112443,
      "learning_rate": 0.00014384810343382275,
      "loss": 11.5,
      "step": 416
    },
    {
      "epoch": 0.3619791666666667,
      "grad_norm": 0.0007663075812160969,
      "learning_rate": 0.0001436006982175785,
      "loss": 11.5,
      "step": 417
    },
    {
      "epoch": 0.3628472222222222,
      "grad_norm": 0.00184196000918746,
      "learning_rate": 0.00014335296304178035,
      "loss": 11.5,
      "step": 418
    },
    {
      "epoch": 0.3637152777777778,
      "grad_norm": 0.0012152615236118436,
      "learning_rate": 0.0001431048997812283,
      "loss": 11.5,
      "step": 419
    },
    {
      "epoch": 0.3645833333333333,
      "grad_norm": 0.0013130082515999675,
      "learning_rate": 0.00014285651031320517,
      "loss": 11.5,
      "step": 420
    },
    {
      "epoch": 0.3654513888888889,
      "grad_norm": 0.0009862114675343037,
      "learning_rate": 0.0001426077965174625,
      "loss": 11.5,
      "step": 421
    },
    {
      "epoch": 0.3663194444444444,
      "grad_norm": 0.001991483150050044,
      "learning_rate": 0.0001423587602762062,
      "loss": 11.5,
      "step": 422
    },
    {
      "epoch": 0.3671875,
      "grad_norm": 0.0012740781530737877,
      "learning_rate": 0.00014210940347408242,
      "loss": 11.5,
      "step": 423
    },
    {
      "epoch": 0.3680555555555556,
      "grad_norm": 0.001297510927543044,
      "learning_rate": 0.00014185972799816315,
      "loss": 11.5,
      "step": 424
    },
    {
      "epoch": 0.3689236111111111,
      "grad_norm": 0.0012677921913564205,
      "learning_rate": 0.0001416097357379322,
      "loss": 11.5,
      "step": 425
    },
    {
      "epoch": 0.3697916666666667,
      "grad_norm": 0.0010948481503874063,
      "learning_rate": 0.00014135942858527054,
      "loss": 11.5,
      "step": 426
    },
    {
      "epoch": 0.3706597222222222,
      "grad_norm": 0.0020884014666080475,
      "learning_rate": 0.00014110880843444225,
      "loss": 11.5,
      "step": 427
    },
    {
      "epoch": 0.3715277777777778,
      "grad_norm": 0.0008858646033331752,
      "learning_rate": 0.00014085787718208016,
      "loss": 11.5,
      "step": 428
    },
    {
      "epoch": 0.3723958333333333,
      "grad_norm": 0.0007041823118925095,
      "learning_rate": 0.0001406066367271713,
      "loss": 11.5,
      "step": 429
    },
    {
      "epoch": 0.3732638888888889,
      "grad_norm": 0.0025530520360916853,
      "learning_rate": 0.00014035508897104282,
      "loss": 11.5,
      "step": 430
    },
    {
      "epoch": 0.3741319444444444,
      "grad_norm": 0.001856551505625248,
      "learning_rate": 0.00014010323581734738,
      "loss": 11.5,
      "step": 431
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.001821168465539813,
      "learning_rate": 0.00013985107917204877,
      "loss": 11.5,
      "step": 432
    },
    {
      "epoch": 0.3758680555555556,
      "grad_norm": 0.0011784174712374806,
      "learning_rate": 0.00013959862094340764,
      "loss": 11.5,
      "step": 433
    },
    {
      "epoch": 0.3767361111111111,
      "grad_norm": 0.0008998655248433352,
      "learning_rate": 0.0001393458630419669,
      "loss": 11.5,
      "step": 434
    },
    {
      "epoch": 0.3776041666666667,
      "grad_norm": 0.0026559773832559586,
      "learning_rate": 0.0001390928073805372,
      "loss": 11.5,
      "step": 435
    },
    {
      "epoch": 0.3784722222222222,
      "grad_norm": 0.0011656692950055003,
      "learning_rate": 0.0001388394558741828,
      "loss": 11.5,
      "step": 436
    },
    {
      "epoch": 0.3793402777777778,
      "grad_norm": 0.0024249244015663862,
      "learning_rate": 0.00013858581044020663,
      "loss": 11.5,
      "step": 437
    },
    {
      "epoch": 0.3802083333333333,
      "grad_norm": 0.002038089092820883,
      "learning_rate": 0.00013833187299813613,
      "loss": 11.5,
      "step": 438
    },
    {
      "epoch": 0.3810763888888889,
      "grad_norm": 0.0038254428654909134,
      "learning_rate": 0.00013807764546970848,
      "loss": 11.5,
      "step": 439
    },
    {
      "epoch": 0.3819444444444444,
      "grad_norm": 0.0016774323303252459,
      "learning_rate": 0.00013782312977885623,
      "loss": 11.5,
      "step": 440
    },
    {
      "epoch": 0.3828125,
      "grad_norm": 0.0019626901485025883,
      "learning_rate": 0.0001375683278516927,
      "loss": 11.5,
      "step": 441
    },
    {
      "epoch": 0.3836805555555556,
      "grad_norm": 0.0015020320424810052,
      "learning_rate": 0.00013731324161649728,
      "loss": 11.5,
      "step": 442
    },
    {
      "epoch": 0.3845486111111111,
      "grad_norm": 0.002265243325382471,
      "learning_rate": 0.000137057873003701,
      "loss": 11.5,
      "step": 443
    },
    {
      "epoch": 0.3854166666666667,
      "grad_norm": 0.0017855282640084624,
      "learning_rate": 0.00013680222394587188,
      "loss": 11.5,
      "step": 444
    },
    {
      "epoch": 0.3862847222222222,
      "grad_norm": 0.0012480892473831773,
      "learning_rate": 0.0001365462963777002,
      "loss": 11.5,
      "step": 445
    },
    {
      "epoch": 0.3871527777777778,
      "grad_norm": 0.0013888552784919739,
      "learning_rate": 0.00013629009223598403,
      "loss": 11.5,
      "step": 446
    },
    {
      "epoch": 0.3880208333333333,
      "grad_norm": 0.0020385878160595894,
      "learning_rate": 0.00013603361345961444,
      "loss": 11.5,
      "step": 447
    },
    {
      "epoch": 0.3888888888888889,
      "grad_norm": 0.004183146171271801,
      "learning_rate": 0.0001357768619895608,
      "loss": 11.5,
      "step": 448
    },
    {
      "epoch": 0.3897569444444444,
      "grad_norm": 0.002523133996874094,
      "learning_rate": 0.00013551983976885633,
      "loss": 11.5,
      "step": 449
    },
    {
      "epoch": 0.390625,
      "grad_norm": 0.0016534990863874555,
      "learning_rate": 0.000135262548742583,
      "loss": 11.5,
      "step": 450
    },
    {
      "epoch": 0.3914930555555556,
      "grad_norm": 0.0011005090782418847,
      "learning_rate": 0.00013500499085785725,
      "loss": 11.5,
      "step": 451
    },
    {
      "epoch": 0.3923611111111111,
      "grad_norm": 0.000922717503271997,
      "learning_rate": 0.0001347471680638149,
      "loss": 11.5,
      "step": 452
    },
    {
      "epoch": 0.3932291666666667,
      "grad_norm": 0.0017686160281300545,
      "learning_rate": 0.00013448908231159662,
      "loss": 11.5,
      "step": 453
    },
    {
      "epoch": 0.3940972222222222,
      "grad_norm": 0.0011171618243679404,
      "learning_rate": 0.00013423073555433307,
      "loss": 11.5,
      "step": 454
    },
    {
      "epoch": 0.3949652777777778,
      "grad_norm": 0.0012590039987117052,
      "learning_rate": 0.00013397212974713006,
      "loss": 11.5,
      "step": 455
    },
    {
      "epoch": 0.3958333333333333,
      "grad_norm": 0.0006373192300088704,
      "learning_rate": 0.00013371326684705385,
      "loss": 11.5,
      "step": 456
    },
    {
      "epoch": 0.3967013888888889,
      "grad_norm": 0.0007318560965359211,
      "learning_rate": 0.00013345414881311645,
      "loss": 11.5,
      "step": 457
    },
    {
      "epoch": 0.3975694444444444,
      "grad_norm": 0.001518836710602045,
      "learning_rate": 0.0001331947776062605,
      "loss": 11.5,
      "step": 458
    },
    {
      "epoch": 0.3984375,
      "grad_norm": 0.00149513129144907,
      "learning_rate": 0.00013293515518934468,
      "loss": 11.5,
      "step": 459
    },
    {
      "epoch": 0.3993055555555556,
      "grad_norm": 0.0015671660657972097,
      "learning_rate": 0.00013267528352712868,
      "loss": 11.5,
      "step": 460
    },
    {
      "epoch": 0.4001736111111111,
      "grad_norm": 0.0013766905758529902,
      "learning_rate": 0.00013241516458625853,
      "loss": 11.5,
      "step": 461
    },
    {
      "epoch": 0.4010416666666667,
      "grad_norm": 0.0007458467734977603,
      "learning_rate": 0.00013215480033525157,
      "loss": 11.5,
      "step": 462
    },
    {
      "epoch": 0.4019097222222222,
      "grad_norm": 0.001208136323839426,
      "learning_rate": 0.0001318941927444815,
      "loss": 11.5,
      "step": 463
    },
    {
      "epoch": 0.4027777777777778,
      "grad_norm": 0.004407074768096209,
      "learning_rate": 0.00013163334378616369,
      "loss": 11.5,
      "step": 464
    },
    {
      "epoch": 0.4036458333333333,
      "grad_norm": 0.0015446634497493505,
      "learning_rate": 0.00013137225543434,
      "loss": 11.5,
      "step": 465
    },
    {
      "epoch": 0.4045138888888889,
      "grad_norm": 0.001385220093652606,
      "learning_rate": 0.0001311109296648641,
      "loss": 11.5,
      "step": 466
    },
    {
      "epoch": 0.4053819444444444,
      "grad_norm": 0.0008750546257942915,
      "learning_rate": 0.00013084936845538623,
      "loss": 11.5,
      "step": 467
    },
    {
      "epoch": 0.40625,
      "grad_norm": 0.0009986564982682467,
      "learning_rate": 0.00013058757378533846,
      "loss": 11.5,
      "step": 468
    },
    {
      "epoch": 0.4071180555555556,
      "grad_norm": 0.0009423622395843267,
      "learning_rate": 0.00013032554763591967,
      "loss": 11.5,
      "step": 469
    },
    {
      "epoch": 0.4079861111111111,
      "grad_norm": 0.0017997513059526682,
      "learning_rate": 0.00013006329199008042,
      "loss": 11.5,
      "step": 470
    },
    {
      "epoch": 0.4088541666666667,
      "grad_norm": 0.0009396991808898747,
      "learning_rate": 0.00012980080883250814,
      "loss": 11.5,
      "step": 471
    },
    {
      "epoch": 0.4097222222222222,
      "grad_norm": 0.0017715089488774538,
      "learning_rate": 0.00012953810014961194,
      "loss": 11.5,
      "step": 472
    },
    {
      "epoch": 0.4105902777777778,
      "grad_norm": 0.0008642635657452047,
      "learning_rate": 0.0001292751679295077,
      "loss": 11.5,
      "step": 473
    },
    {
      "epoch": 0.4114583333333333,
      "grad_norm": 0.00047178869135677814,
      "learning_rate": 0.000129012014162003,
      "loss": 11.5,
      "step": 474
    },
    {
      "epoch": 0.4123263888888889,
      "grad_norm": 0.00044494090252555907,
      "learning_rate": 0.00012874864083858194,
      "loss": 11.5,
      "step": 475
    },
    {
      "epoch": 0.4131944444444444,
      "grad_norm": 0.002891515614464879,
      "learning_rate": 0.00012848504995239022,
      "loss": 11.5,
      "step": 476
    },
    {
      "epoch": 0.4140625,
      "grad_norm": 0.002096660668030381,
      "learning_rate": 0.00012822124349822002,
      "loss": 11.5,
      "step": 477
    },
    {
      "epoch": 0.4149305555555556,
      "grad_norm": 0.0007149482262320817,
      "learning_rate": 0.0001279572234724949,
      "loss": 11.5,
      "step": 478
    },
    {
      "epoch": 0.4157986111111111,
      "grad_norm": 0.0014680104795843363,
      "learning_rate": 0.00012769299187325464,
      "loss": 11.5,
      "step": 479
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.0035200186539441347,
      "learning_rate": 0.00012742855070014015,
      "loss": 11.5,
      "step": 480
    },
    {
      "epoch": 0.4175347222222222,
      "grad_norm": 0.0012388027971610427,
      "learning_rate": 0.00012716390195437836,
      "loss": 11.5,
      "step": 481
    },
    {
      "epoch": 0.4184027777777778,
      "grad_norm": 0.0019148902501910925,
      "learning_rate": 0.00012689904763876707,
      "loss": 11.5,
      "step": 482
    },
    {
      "epoch": 0.4192708333333333,
      "grad_norm": 0.00093680719146505,
      "learning_rate": 0.00012663398975765975,
      "loss": 11.5,
      "step": 483
    },
    {
      "epoch": 0.4201388888888889,
      "grad_norm": 0.002589451614767313,
      "learning_rate": 0.0001263687303169505,
      "loss": 11.5,
      "step": 484
    },
    {
      "epoch": 0.4210069444444444,
      "grad_norm": 0.0013582732062786818,
      "learning_rate": 0.0001261032713240586,
      "loss": 11.5,
      "step": 485
    },
    {
      "epoch": 0.421875,
      "grad_norm": 0.0012645411770790815,
      "learning_rate": 0.00012583761478791367,
      "loss": 11.5,
      "step": 486
    },
    {
      "epoch": 0.4227430555555556,
      "grad_norm": 0.0031839171424508095,
      "learning_rate": 0.00012557176271894022,
      "loss": 11.5,
      "step": 487
    },
    {
      "epoch": 0.4236111111111111,
      "grad_norm": 0.0012117556761950254,
      "learning_rate": 0.00012530571712904244,
      "loss": 11.5,
      "step": 488
    },
    {
      "epoch": 0.4244791666666667,
      "grad_norm": 0.0015861785504966974,
      "learning_rate": 0.00012503948003158918,
      "loss": 11.5,
      "step": 489
    },
    {
      "epoch": 0.4253472222222222,
      "grad_norm": 0.003547539934515953,
      "learning_rate": 0.00012477305344139842,
      "loss": 11.5,
      "step": 490
    },
    {
      "epoch": 0.4262152777777778,
      "grad_norm": 0.0017221581656485796,
      "learning_rate": 0.0001245064393747223,
      "loss": 11.5,
      "step": 491
    },
    {
      "epoch": 0.4270833333333333,
      "grad_norm": 0.001912225503474474,
      "learning_rate": 0.00012423963984923165,
      "loss": 11.5,
      "step": 492
    },
    {
      "epoch": 0.4279513888888889,
      "grad_norm": 0.0016410641837865114,
      "learning_rate": 0.00012397265688400083,
      "loss": 11.5,
      "step": 493
    },
    {
      "epoch": 0.4288194444444444,
      "grad_norm": 0.0015870253555476665,
      "learning_rate": 0.00012370549249949252,
      "loss": 11.5,
      "step": 494
    },
    {
      "epoch": 0.4296875,
      "grad_norm": 0.002609993563964963,
      "learning_rate": 0.00012343814871754218,
      "loss": 11.5,
      "step": 495
    },
    {
      "epoch": 0.4305555555555556,
      "grad_norm": 0.001266726991161704,
      "learning_rate": 0.000123170627561343,
      "loss": 11.5,
      "step": 496
    },
    {
      "epoch": 0.4314236111111111,
      "grad_norm": 0.0021266017574816942,
      "learning_rate": 0.00012290293105543052,
      "loss": 11.5,
      "step": 497
    },
    {
      "epoch": 0.4322916666666667,
      "grad_norm": 0.0015497597632929683,
      "learning_rate": 0.0001226350612256672,
      "loss": 11.5,
      "step": 498
    },
    {
      "epoch": 0.4331597222222222,
      "grad_norm": 0.0016415165737271309,
      "learning_rate": 0.0001223670200992272,
      "loss": 11.5,
      "step": 499
    },
    {
      "epoch": 0.4340277777777778,
      "grad_norm": 0.0020661710295826197,
      "learning_rate": 0.0001220988097045811,
      "loss": 11.5,
      "step": 500
    },
    {
      "epoch": 0.4348958333333333,
      "grad_norm": 0.001812787726521492,
      "learning_rate": 0.00012183043207148032,
      "loss": 11.5,
      "step": 501
    },
    {
      "epoch": 0.4357638888888889,
      "grad_norm": 0.0023733002599328756,
      "learning_rate": 0.00012156188923094202,
      "loss": 11.5,
      "step": 502
    },
    {
      "epoch": 0.4366319444444444,
      "grad_norm": 0.0015286860289052129,
      "learning_rate": 0.00012129318321523353,
      "loss": 11.5,
      "step": 503
    },
    {
      "epoch": 0.4375,
      "grad_norm": 0.0012327119475230575,
      "learning_rate": 0.00012102431605785709,
      "loss": 11.5,
      "step": 504
    },
    {
      "epoch": 0.4383680555555556,
      "grad_norm": 0.0014177414122968912,
      "learning_rate": 0.00012075528979353442,
      "loss": 11.5,
      "step": 505
    },
    {
      "epoch": 0.4392361111111111,
      "grad_norm": 0.0008958170074038208,
      "learning_rate": 0.0001204861064581913,
      "loss": 11.5,
      "step": 506
    },
    {
      "epoch": 0.4401041666666667,
      "grad_norm": 0.0008167149499058723,
      "learning_rate": 0.00012021676808894222,
      "loss": 11.5,
      "step": 507
    },
    {
      "epoch": 0.4409722222222222,
      "grad_norm": 0.0006493402761407197,
      "learning_rate": 0.00011994727672407487,
      "loss": 11.5,
      "step": 508
    },
    {
      "epoch": 0.4418402777777778,
      "grad_norm": 0.0019143844256177545,
      "learning_rate": 0.00011967763440303485,
      "loss": 11.5,
      "step": 509
    },
    {
      "epoch": 0.4427083333333333,
      "grad_norm": 0.0008292369893752038,
      "learning_rate": 0.0001194078431664101,
      "loss": 11.5,
      "step": 510
    },
    {
      "epoch": 0.4435763888888889,
      "grad_norm": 0.0012759511591866612,
      "learning_rate": 0.00011913790505591556,
      "loss": 11.5,
      "step": 511
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.0020761843770742416,
      "learning_rate": 0.00011886782211437766,
      "loss": 11.5,
      "step": 512
    },
    {
      "epoch": 0.4453125,
      "grad_norm": 0.0010456456802785397,
      "learning_rate": 0.00011859759638571886,
      "loss": 11.5,
      "step": 513
    },
    {
      "epoch": 0.4461805555555556,
      "grad_norm": 0.0018422120483592153,
      "learning_rate": 0.00011832722991494218,
      "loss": 11.5,
      "step": 514
    },
    {
      "epoch": 0.4470486111111111,
      "grad_norm": 0.0010430898983031511,
      "learning_rate": 0.00011805672474811582,
      "loss": 11.5,
      "step": 515
    },
    {
      "epoch": 0.4479166666666667,
      "grad_norm": 0.0010992520255967975,
      "learning_rate": 0.00011778608293235756,
      "loss": 11.5,
      "step": 516
    },
    {
      "epoch": 0.4487847222222222,
      "grad_norm": 0.0015777871012687683,
      "learning_rate": 0.00011751530651581925,
      "loss": 11.5,
      "step": 517
    },
    {
      "epoch": 0.4496527777777778,
      "grad_norm": 0.0023814847227185965,
      "learning_rate": 0.00011724439754767145,
      "loss": 11.5,
      "step": 518
    },
    {
      "epoch": 0.4505208333333333,
      "grad_norm": 0.0015265214024111629,
      "learning_rate": 0.00011697335807808778,
      "loss": 11.5,
      "step": 519
    },
    {
      "epoch": 0.4513888888888889,
      "grad_norm": 0.0005157457198947668,
      "learning_rate": 0.00011670219015822949,
      "loss": 11.5,
      "step": 520
    },
    {
      "epoch": 0.4522569444444444,
      "grad_norm": 0.0012737824581563473,
      "learning_rate": 0.0001164308958402299,
      "loss": 11.5,
      "step": 521
    },
    {
      "epoch": 0.453125,
      "grad_norm": 0.0008698921883478761,
      "learning_rate": 0.00011615947717717894,
      "loss": 11.5,
      "step": 522
    },
    {
      "epoch": 0.4539930555555556,
      "grad_norm": 0.0011539001716300845,
      "learning_rate": 0.00011588793622310743,
      "loss": 11.5,
      "step": 523
    },
    {
      "epoch": 0.4548611111111111,
      "grad_norm": 0.0019196930807083845,
      "learning_rate": 0.00011561627503297176,
      "loss": 11.5,
      "step": 524
    },
    {
      "epoch": 0.4557291666666667,
      "grad_norm": 0.001891504623927176,
      "learning_rate": 0.00011534449566263821,
      "loss": 11.5,
      "step": 525
    },
    {
      "epoch": 0.4565972222222222,
      "grad_norm": 0.0021699066273868084,
      "learning_rate": 0.0001150726001688674,
      "loss": 11.5,
      "step": 526
    },
    {
      "epoch": 0.4574652777777778,
      "grad_norm": 0.0019159960793331265,
      "learning_rate": 0.00011480059060929882,
      "loss": 11.5,
      "step": 527
    },
    {
      "epoch": 0.4583333333333333,
      "grad_norm": 0.000828733027447015,
      "learning_rate": 0.00011452846904243502,
      "loss": 11.5,
      "step": 528
    },
    {
      "epoch": 0.4592013888888889,
      "grad_norm": 0.0010853870771825314,
      "learning_rate": 0.00011425623752762635,
      "loss": 11.5,
      "step": 529
    },
    {
      "epoch": 0.4600694444444444,
      "grad_norm": 0.0014166930923238397,
      "learning_rate": 0.00011398389812505521,
      "loss": 11.5,
      "step": 530
    },
    {
      "epoch": 0.4609375,
      "grad_norm": 0.0009068722720257938,
      "learning_rate": 0.00011371145289572031,
      "loss": 11.5,
      "step": 531
    },
    {
      "epoch": 0.4618055555555556,
      "grad_norm": 0.004488244187086821,
      "learning_rate": 0.00011343890390142149,
      "loss": 11.5,
      "step": 532
    },
    {
      "epoch": 0.4626736111111111,
      "grad_norm": 0.002014743397012353,
      "learning_rate": 0.00011316625320474358,
      "loss": 11.5,
      "step": 533
    },
    {
      "epoch": 0.4635416666666667,
      "grad_norm": 0.0014623699244111776,
      "learning_rate": 0.00011289350286904131,
      "loss": 11.5,
      "step": 534
    },
    {
      "epoch": 0.4644097222222222,
      "grad_norm": 0.0013415683060884476,
      "learning_rate": 0.00011262065495842329,
      "loss": 11.5,
      "step": 535
    },
    {
      "epoch": 0.4652777777777778,
      "grad_norm": 0.0015175221487879753,
      "learning_rate": 0.00011234771153773658,
      "loss": 11.5,
      "step": 536
    },
    {
      "epoch": 0.4661458333333333,
      "grad_norm": 0.0014910164754837751,
      "learning_rate": 0.00011207467467255117,
      "loss": 11.5,
      "step": 537
    },
    {
      "epoch": 0.4670138888888889,
      "grad_norm": 0.0024390704929828644,
      "learning_rate": 0.00011180154642914402,
      "loss": 11.5,
      "step": 538
    },
    {
      "epoch": 0.4678819444444444,
      "grad_norm": 0.001675798324868083,
      "learning_rate": 0.00011152832887448375,
      "loss": 11.5,
      "step": 539
    },
    {
      "epoch": 0.46875,
      "grad_norm": 0.0015164773212745786,
      "learning_rate": 0.00011125502407621484,
      "loss": 11.5,
      "step": 540
    },
    {
      "epoch": 0.4696180555555556,
      "grad_norm": 0.001736818696372211,
      "learning_rate": 0.00011098163410264194,
      "loss": 11.5,
      "step": 541
    },
    {
      "epoch": 0.4704861111111111,
      "grad_norm": 0.0009775893995538354,
      "learning_rate": 0.00011070816102271442,
      "loss": 11.5,
      "step": 542
    },
    {
      "epoch": 0.4713541666666667,
      "grad_norm": 0.0012841086136177182,
      "learning_rate": 0.00011043460690601044,
      "loss": 11.5,
      "step": 543
    },
    {
      "epoch": 0.4722222222222222,
      "grad_norm": 0.0030059965793043375,
      "learning_rate": 0.0001101609738227215,
      "loss": 11.5,
      "step": 544
    },
    {
      "epoch": 0.4730902777777778,
      "grad_norm": 0.0008778509800322354,
      "learning_rate": 0.00010988726384363672,
      "loss": 11.5,
      "step": 545
    },
    {
      "epoch": 0.4739583333333333,
      "grad_norm": 0.0019029427785426378,
      "learning_rate": 0.00010961347904012709,
      "loss": 11.5,
      "step": 546
    },
    {
      "epoch": 0.4748263888888889,
      "grad_norm": 0.001545676146633923,
      "learning_rate": 0.00010933962148412986,
      "loss": 11.5,
      "step": 547
    },
    {
      "epoch": 0.4756944444444444,
      "grad_norm": 0.0021841314155608416,
      "learning_rate": 0.00010906569324813292,
      "loss": 11.5,
      "step": 548
    },
    {
      "epoch": 0.4765625,
      "grad_norm": 0.0012104269117116928,
      "learning_rate": 0.00010879169640515896,
      "loss": 11.5,
      "step": 549
    },
    {
      "epoch": 0.4774305555555556,
      "grad_norm": 0.0042435056529939175,
      "learning_rate": 0.00010851763302874997,
      "loss": 11.5,
      "step": 550
    },
    {
      "epoch": 0.4782986111111111,
      "grad_norm": 0.0035029014106839895,
      "learning_rate": 0.00010824350519295133,
      "loss": 11.5,
      "step": 551
    },
    {
      "epoch": 0.4791666666666667,
      "grad_norm": 0.001896754838526249,
      "learning_rate": 0.0001079693149722963,
      "loss": 11.5,
      "step": 552
    },
    {
      "epoch": 0.4800347222222222,
      "grad_norm": 0.003470843192189932,
      "learning_rate": 0.00010769506444179029,
      "loss": 11.5,
      "step": 553
    },
    {
      "epoch": 0.4809027777777778,
      "grad_norm": 0.001424275105819106,
      "learning_rate": 0.00010742075567689504,
      "loss": 11.5,
      "step": 554
    },
    {
      "epoch": 0.4817708333333333,
      "grad_norm": 0.0010783561738207936,
      "learning_rate": 0.00010714639075351309,
      "loss": 11.5,
      "step": 555
    },
    {
      "epoch": 0.4826388888888889,
      "grad_norm": 0.0008930449839681387,
      "learning_rate": 0.00010687197174797186,
      "loss": 11.5,
      "step": 556
    },
    {
      "epoch": 0.4835069444444444,
      "grad_norm": 0.0009238430648110807,
      "learning_rate": 0.00010659750073700811,
      "loss": 11.5,
      "step": 557
    },
    {
      "epoch": 0.484375,
      "grad_norm": 0.0007530270959250629,
      "learning_rate": 0.00010632297979775219,
      "loss": 11.5,
      "step": 558
    },
    {
      "epoch": 0.4852430555555556,
      "grad_norm": 0.0013512952718883753,
      "learning_rate": 0.00010604841100771224,
      "loss": 11.5,
      "step": 559
    },
    {
      "epoch": 0.4861111111111111,
      "grad_norm": 0.0010359024163335562,
      "learning_rate": 0.0001057737964447586,
      "loss": 11.5,
      "step": 560
    },
    {
      "epoch": 0.4869791666666667,
      "grad_norm": 0.0016550614964216948,
      "learning_rate": 0.0001054991381871079,
      "loss": 11.5,
      "step": 561
    },
    {
      "epoch": 0.4878472222222222,
      "grad_norm": 0.0017867699498310685,
      "learning_rate": 0.0001052244383133075,
      "loss": 11.5,
      "step": 562
    },
    {
      "epoch": 0.4887152777777778,
      "grad_norm": 0.0008279422763735056,
      "learning_rate": 0.00010494969890221975,
      "loss": 11.5,
      "step": 563
    },
    {
      "epoch": 0.4895833333333333,
      "grad_norm": 0.0037701819092035294,
      "learning_rate": 0.00010467492203300615,
      "loss": 11.5,
      "step": 564
    },
    {
      "epoch": 0.4904513888888889,
      "grad_norm": 0.0010847491212189198,
      "learning_rate": 0.00010440010978511166,
      "loss": 11.5,
      "step": 565
    },
    {
      "epoch": 0.4913194444444444,
      "grad_norm": 0.0012644154485315084,
      "learning_rate": 0.000104125264238249,
      "loss": 11.5,
      "step": 566
    },
    {
      "epoch": 0.4921875,
      "grad_norm": 0.002801130060106516,
      "learning_rate": 0.00010385038747238289,
      "loss": 11.5,
      "step": 567
    },
    {
      "epoch": 0.4930555555555556,
      "grad_norm": 0.0009231338044628501,
      "learning_rate": 0.00010357548156771433,
      "loss": 11.5,
      "step": 568
    },
    {
      "epoch": 0.4939236111111111,
      "grad_norm": 0.0019018857274204493,
      "learning_rate": 0.00010330054860466478,
      "loss": 11.5,
      "step": 569
    },
    {
      "epoch": 0.4947916666666667,
      "grad_norm": 0.0012639490887522697,
      "learning_rate": 0.00010302559066386055,
      "loss": 11.5,
      "step": 570
    },
    {
      "epoch": 0.4956597222222222,
      "grad_norm": 0.001242935424670577,
      "learning_rate": 0.00010275060982611688,
      "loss": 11.5,
      "step": 571
    },
    {
      "epoch": 0.4965277777777778,
      "grad_norm": 0.0010218758834525943,
      "learning_rate": 0.00010247560817242237,
      "loss": 11.5,
      "step": 572
    },
    {
      "epoch": 0.4973958333333333,
      "grad_norm": 0.0006046834751032293,
      "learning_rate": 0.0001022005877839231,
      "loss": 11.5,
      "step": 573
    },
    {
      "epoch": 0.4982638888888889,
      "grad_norm": 0.0012121489271521568,
      "learning_rate": 0.00010192555074190695,
      "loss": 11.5,
      "step": 574
    },
    {
      "epoch": 0.4991319444444444,
      "grad_norm": 0.0012221458600834012,
      "learning_rate": 0.00010165049912778788,
      "loss": 11.5,
      "step": 575
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.000797056476585567,
      "learning_rate": 0.00010137543502308999,
      "loss": 11.5,
      "step": 576
    },
    {
      "epoch": 0.5,
      "eval_loss": 11.5,
      "eval_runtime": 3.9381,
      "eval_samples_per_second": 123.155,
      "eval_steps_per_second": 61.705,
      "step": 576
    },
    {
      "epoch": 0.5008680555555556,
      "grad_norm": 0.0023448157589882612,
      "learning_rate": 0.00010110036050943203,
      "loss": 11.5,
      "step": 577
    },
    {
      "epoch": 0.5017361111111112,
      "grad_norm": 0.0009461472509428859,
      "learning_rate": 0.00010082527766851151,
      "loss": 11.5,
      "step": 578
    },
    {
      "epoch": 0.5026041666666666,
      "grad_norm": 0.0019334993558004498,
      "learning_rate": 0.00010055018858208886,
      "loss": 11.5,
      "step": 579
    },
    {
      "epoch": 0.5034722222222222,
      "grad_norm": 0.00210960884578526,
      "learning_rate": 0.00010027509533197197,
      "loss": 11.5,
      "step": 580
    },
    {
      "epoch": 0.5043402777777778,
      "grad_norm": 0.001009828527458012,
      "learning_rate": 0.0001,
      "loss": 11.5,
      "step": 581
    },
    {
      "epoch": 0.5052083333333334,
      "grad_norm": 0.0008409370784647763,
      "learning_rate": 9.972490466802805e-05,
      "loss": 11.5,
      "step": 582
    },
    {
      "epoch": 0.5060763888888888,
      "grad_norm": 0.0015977314906194806,
      "learning_rate": 9.944981141791115e-05,
      "loss": 11.5,
      "step": 583
    },
    {
      "epoch": 0.5069444444444444,
      "grad_norm": 0.0010232948698103428,
      "learning_rate": 9.917472233148851e-05,
      "loss": 11.5,
      "step": 584
    },
    {
      "epoch": 0.5078125,
      "grad_norm": 0.0016299504786729813,
      "learning_rate": 9.889963949056799e-05,
      "loss": 11.5,
      "step": 585
    },
    {
      "epoch": 0.5086805555555556,
      "grad_norm": 0.0020702886395156384,
      "learning_rate": 9.862456497691003e-05,
      "loss": 11.5,
      "step": 586
    },
    {
      "epoch": 0.5095486111111112,
      "grad_norm": 0.0014128328766673803,
      "learning_rate": 9.834950087221216e-05,
      "loss": 11.5,
      "step": 587
    },
    {
      "epoch": 0.5104166666666666,
      "grad_norm": 0.0011754905572161078,
      "learning_rate": 9.807444925809306e-05,
      "loss": 11.5,
      "step": 588
    },
    {
      "epoch": 0.5112847222222222,
      "grad_norm": 0.0015927328495308757,
      "learning_rate": 9.77994122160769e-05,
      "loss": 11.5,
      "step": 589
    },
    {
      "epoch": 0.5121527777777778,
      "grad_norm": 0.0012330239405855536,
      "learning_rate": 9.752439182757764e-05,
      "loss": 11.5,
      "step": 590
    },
    {
      "epoch": 0.5130208333333334,
      "grad_norm": 0.0011607782216742635,
      "learning_rate": 9.724939017388313e-05,
      "loss": 11.5,
      "step": 591
    },
    {
      "epoch": 0.5138888888888888,
      "grad_norm": 0.002848805859684944,
      "learning_rate": 9.697440933613946e-05,
      "loss": 11.5,
      "step": 592
    },
    {
      "epoch": 0.5147569444444444,
      "grad_norm": 0.0025082037318497896,
      "learning_rate": 9.669945139533524e-05,
      "loss": 11.5,
      "step": 593
    },
    {
      "epoch": 0.515625,
      "grad_norm": 0.0012475978583097458,
      "learning_rate": 9.642451843228568e-05,
      "loss": 11.5,
      "step": 594
    },
    {
      "epoch": 0.5164930555555556,
      "grad_norm": 0.004725827369838953,
      "learning_rate": 9.614961252761713e-05,
      "loss": 11.5,
      "step": 595
    },
    {
      "epoch": 0.5173611111111112,
      "grad_norm": 0.001316457986831665,
      "learning_rate": 9.587473576175103e-05,
      "loss": 11.5,
      "step": 596
    },
    {
      "epoch": 0.5182291666666666,
      "grad_norm": 0.002363733947277069,
      "learning_rate": 9.559989021488838e-05,
      "loss": 11.5,
      "step": 597
    },
    {
      "epoch": 0.5190972222222222,
      "grad_norm": 0.002522528637200594,
      "learning_rate": 9.532507796699388e-05,
      "loss": 11.5,
      "step": 598
    },
    {
      "epoch": 0.5199652777777778,
      "grad_norm": 0.0013324473984539509,
      "learning_rate": 9.505030109778023e-05,
      "loss": 11.5,
      "step": 599
    },
    {
      "epoch": 0.5208333333333334,
      "grad_norm": 0.0027300387155264616,
      "learning_rate": 9.47755616866925e-05,
      "loss": 11.5,
      "step": 600
    },
    {
      "epoch": 0.5217013888888888,
      "grad_norm": 0.0027879918925464153,
      "learning_rate": 9.450086181289213e-05,
      "loss": 11.5,
      "step": 601
    },
    {
      "epoch": 0.5225694444444444,
      "grad_norm": 0.0010155023774132133,
      "learning_rate": 9.422620355524144e-05,
      "loss": 11.5,
      "step": 602
    },
    {
      "epoch": 0.5234375,
      "grad_norm": 0.0011581453727558255,
      "learning_rate": 9.395158899228777e-05,
      "loss": 11.5,
      "step": 603
    },
    {
      "epoch": 0.5243055555555556,
      "grad_norm": 0.001615899964235723,
      "learning_rate": 9.367702020224782e-05,
      "loss": 11.5,
      "step": 604
    },
    {
      "epoch": 0.5251736111111112,
      "grad_norm": 0.0021046611946076155,
      "learning_rate": 9.340249926299192e-05,
      "loss": 11.5,
      "step": 605
    },
    {
      "epoch": 0.5260416666666666,
      "grad_norm": 0.0012483590980991721,
      "learning_rate": 9.312802825202817e-05,
      "loss": 11.5,
      "step": 606
    },
    {
      "epoch": 0.5269097222222222,
      "grad_norm": 0.0006401913706213236,
      "learning_rate": 9.285360924648695e-05,
      "loss": 11.5,
      "step": 607
    },
    {
      "epoch": 0.5277777777777778,
      "grad_norm": 0.001484209788031876,
      "learning_rate": 9.257924432310497e-05,
      "loss": 11.5,
      "step": 608
    },
    {
      "epoch": 0.5286458333333334,
      "grad_norm": 0.0009637821931391954,
      "learning_rate": 9.230493555820971e-05,
      "loss": 11.5,
      "step": 609
    },
    {
      "epoch": 0.5295138888888888,
      "grad_norm": 0.002728967694565654,
      "learning_rate": 9.203068502770371e-05,
      "loss": 11.5,
      "step": 610
    },
    {
      "epoch": 0.5303819444444444,
      "grad_norm": 0.0013348457869142294,
      "learning_rate": 9.17564948070487e-05,
      "loss": 11.5,
      "step": 611
    },
    {
      "epoch": 0.53125,
      "grad_norm": 0.0017096330411732197,
      "learning_rate": 9.148236697125008e-05,
      "loss": 11.5,
      "step": 612
    },
    {
      "epoch": 0.5321180555555556,
      "grad_norm": 0.0010077767074108124,
      "learning_rate": 9.120830359484105e-05,
      "loss": 11.5,
      "step": 613
    },
    {
      "epoch": 0.5329861111111112,
      "grad_norm": 0.001199171063490212,
      "learning_rate": 9.093430675186708e-05,
      "loss": 11.5,
      "step": 614
    },
    {
      "epoch": 0.5338541666666666,
      "grad_norm": 0.0017377744661644101,
      "learning_rate": 9.066037851587015e-05,
      "loss": 11.5,
      "step": 615
    },
    {
      "epoch": 0.5347222222222222,
      "grad_norm": 0.002164931269362569,
      "learning_rate": 9.038652095987294e-05,
      "loss": 11.5,
      "step": 616
    },
    {
      "epoch": 0.5355902777777778,
      "grad_norm": 0.0014526225859299302,
      "learning_rate": 9.01127361563633e-05,
      "loss": 11.5,
      "step": 617
    },
    {
      "epoch": 0.5364583333333334,
      "grad_norm": 0.0021193870343267918,
      "learning_rate": 8.983902617727852e-05,
      "loss": 11.5,
      "step": 618
    },
    {
      "epoch": 0.5373263888888888,
      "grad_norm": 0.0013588265283033252,
      "learning_rate": 8.956539309398957e-05,
      "loss": 11.5,
      "step": 619
    },
    {
      "epoch": 0.5381944444444444,
      "grad_norm": 0.001437599305063486,
      "learning_rate": 8.92918389772856e-05,
      "loss": 11.5,
      "step": 620
    },
    {
      "epoch": 0.5390625,
      "grad_norm": 0.0006930895615369081,
      "learning_rate": 8.901836589735807e-05,
      "loss": 11.5,
      "step": 621
    },
    {
      "epoch": 0.5399305555555556,
      "grad_norm": 0.003981348592787981,
      "learning_rate": 8.87449759237852e-05,
      "loss": 11.5,
      "step": 622
    },
    {
      "epoch": 0.5407986111111112,
      "grad_norm": 0.001443542423658073,
      "learning_rate": 8.847167112551627e-05,
      "loss": 11.5,
      "step": 623
    },
    {
      "epoch": 0.5416666666666666,
      "grad_norm": 0.000959395372774452,
      "learning_rate": 8.819845357085598e-05,
      "loss": 11.5,
      "step": 624
    },
    {
      "epoch": 0.5425347222222222,
      "grad_norm": 0.001935778185725212,
      "learning_rate": 8.792532532744885e-05,
      "loss": 11.5,
      "step": 625
    },
    {
      "epoch": 0.5434027777777778,
      "grad_norm": 0.001356285996735096,
      "learning_rate": 8.765228846226341e-05,
      "loss": 11.5,
      "step": 626
    },
    {
      "epoch": 0.5442708333333334,
      "grad_norm": 0.007902055978775024,
      "learning_rate": 8.737934504157675e-05,
      "loss": 11.5,
      "step": 627
    },
    {
      "epoch": 0.5451388888888888,
      "grad_norm": 0.001759368577040732,
      "learning_rate": 8.710649713095872e-05,
      "loss": 11.5,
      "step": 628
    },
    {
      "epoch": 0.5460069444444444,
      "grad_norm": 0.0007872672867961228,
      "learning_rate": 8.683374679525644e-05,
      "loss": 11.5,
      "step": 629
    },
    {
      "epoch": 0.546875,
      "grad_norm": 0.000855692196637392,
      "learning_rate": 8.656109609857854e-05,
      "loss": 11.5,
      "step": 630
    },
    {
      "epoch": 0.5477430555555556,
      "grad_norm": 0.0013454173458740115,
      "learning_rate": 8.628854710427968e-05,
      "loss": 11.5,
      "step": 631
    },
    {
      "epoch": 0.5486111111111112,
      "grad_norm": 0.001836450886912644,
      "learning_rate": 8.601610187494484e-05,
      "loss": 11.5,
      "step": 632
    },
    {
      "epoch": 0.5494791666666666,
      "grad_norm": 0.001370997866615653,
      "learning_rate": 8.574376247237367e-05,
      "loss": 11.5,
      "step": 633
    },
    {
      "epoch": 0.5503472222222222,
      "grad_norm": 0.002842591842636466,
      "learning_rate": 8.547153095756503e-05,
      "loss": 11.5,
      "step": 634
    },
    {
      "epoch": 0.5512152777777778,
      "grad_norm": 0.0020862759556621313,
      "learning_rate": 8.519940939070121e-05,
      "loss": 11.5,
      "step": 635
    },
    {
      "epoch": 0.5520833333333334,
      "grad_norm": 0.0012913034297525883,
      "learning_rate": 8.49273998311326e-05,
      "loss": 11.5,
      "step": 636
    },
    {
      "epoch": 0.5529513888888888,
      "grad_norm": 0.001789560541510582,
      "learning_rate": 8.46555043373618e-05,
      "loss": 11.5,
      "step": 637
    },
    {
      "epoch": 0.5538194444444444,
      "grad_norm": 0.0020711093675345182,
      "learning_rate": 8.438372496702826e-05,
      "loss": 11.5,
      "step": 638
    },
    {
      "epoch": 0.5546875,
      "grad_norm": 0.008509905077517033,
      "learning_rate": 8.41120637768926e-05,
      "loss": 11.5,
      "step": 639
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 0.0013360590673983097,
      "learning_rate": 8.384052282282107e-05,
      "loss": 11.5,
      "step": 640
    },
    {
      "epoch": 0.5564236111111112,
      "grad_norm": 0.001760153565555811,
      "learning_rate": 8.356910415977008e-05,
      "loss": 11.5,
      "step": 641
    },
    {
      "epoch": 0.5572916666666666,
      "grad_norm": 0.0031870354432612658,
      "learning_rate": 8.329780984177053e-05,
      "loss": 11.5,
      "step": 642
    },
    {
      "epoch": 0.5581597222222222,
      "grad_norm": 0.0016672826604917645,
      "learning_rate": 8.302664192191225e-05,
      "loss": 11.5,
      "step": 643
    },
    {
      "epoch": 0.5590277777777778,
      "grad_norm": 0.0030133849941194057,
      "learning_rate": 8.27556024523286e-05,
      "loss": 11.5,
      "step": 644
    },
    {
      "epoch": 0.5598958333333334,
      "grad_norm": 0.001557863550260663,
      "learning_rate": 8.248469348418076e-05,
      "loss": 11.5,
      "step": 645
    },
    {
      "epoch": 0.5607638888888888,
      "grad_norm": 0.0017618224956095219,
      "learning_rate": 8.221391706764245e-05,
      "loss": 11.5,
      "step": 646
    },
    {
      "epoch": 0.5616319444444444,
      "grad_norm": 0.000878591206856072,
      "learning_rate": 8.194327525188419e-05,
      "loss": 11.5,
      "step": 647
    },
    {
      "epoch": 0.5625,
      "grad_norm": 0.0009552378323860466,
      "learning_rate": 8.167277008505783e-05,
      "loss": 11.5,
      "step": 648
    },
    {
      "epoch": 0.5633680555555556,
      "grad_norm": 0.0020112725906074047,
      "learning_rate": 8.140240361428119e-05,
      "loss": 11.5,
      "step": 649
    },
    {
      "epoch": 0.5642361111111112,
      "grad_norm": 0.007510410621762276,
      "learning_rate": 8.113217788562235e-05,
      "loss": 11.5,
      "step": 650
    },
    {
      "epoch": 0.5651041666666666,
      "grad_norm": 0.001489496324211359,
      "learning_rate": 8.086209494408443e-05,
      "loss": 11.5,
      "step": 651
    },
    {
      "epoch": 0.5659722222222222,
      "grad_norm": 0.0007072288426570594,
      "learning_rate": 8.059215683358991e-05,
      "loss": 11.5,
      "step": 652
    },
    {
      "epoch": 0.5668402777777778,
      "grad_norm": 0.0011473807971924543,
      "learning_rate": 8.032236559696517e-05,
      "loss": 11.5,
      "step": 653
    },
    {
      "epoch": 0.5677083333333334,
      "grad_norm": 0.0011727320961654186,
      "learning_rate": 8.005272327592517e-05,
      "loss": 11.5,
      "step": 654
    },
    {
      "epoch": 0.5685763888888888,
      "grad_norm": 0.0014338806504383683,
      "learning_rate": 7.978323191105781e-05,
      "loss": 11.5,
      "step": 655
    },
    {
      "epoch": 0.5694444444444444,
      "grad_norm": 0.000566482194699347,
      "learning_rate": 7.951389354180871e-05,
      "loss": 11.5,
      "step": 656
    },
    {
      "epoch": 0.5703125,
      "grad_norm": 0.0009298393270000815,
      "learning_rate": 7.924471020646559e-05,
      "loss": 11.5,
      "step": 657
    },
    {
      "epoch": 0.5711805555555556,
      "grad_norm": 0.0006682793609797955,
      "learning_rate": 7.897568394214292e-05,
      "loss": 11.5,
      "step": 658
    },
    {
      "epoch": 0.5720486111111112,
      "grad_norm": 0.0004757135175168514,
      "learning_rate": 7.87068167847665e-05,
      "loss": 11.5,
      "step": 659
    },
    {
      "epoch": 0.5729166666666666,
      "grad_norm": 0.0021236720494925976,
      "learning_rate": 7.8438110769058e-05,
      "loss": 11.5,
      "step": 660
    },
    {
      "epoch": 0.5737847222222222,
      "grad_norm": 0.0019081539940088987,
      "learning_rate": 7.81695679285197e-05,
      "loss": 11.5,
      "step": 661
    },
    {
      "epoch": 0.5746527777777778,
      "grad_norm": 0.0009791620541363955,
      "learning_rate": 7.790119029541893e-05,
      "loss": 11.5,
      "step": 662
    },
    {
      "epoch": 0.5755208333333334,
      "grad_norm": 0.0014489633031189442,
      "learning_rate": 7.76329799007728e-05,
      "loss": 11.5,
      "step": 663
    },
    {
      "epoch": 0.5763888888888888,
      "grad_norm": 0.0008113660151138902,
      "learning_rate": 7.736493877433285e-05,
      "loss": 11.5,
      "step": 664
    },
    {
      "epoch": 0.5772569444444444,
      "grad_norm": 0.0017051021568477154,
      "learning_rate": 7.709706894456949e-05,
      "loss": 11.5,
      "step": 665
    },
    {
      "epoch": 0.578125,
      "grad_norm": 0.0016557136550545692,
      "learning_rate": 7.682937243865701e-05,
      "loss": 11.5,
      "step": 666
    },
    {
      "epoch": 0.5789930555555556,
      "grad_norm": 0.0019885688088834286,
      "learning_rate": 7.656185128245786e-05,
      "loss": 11.5,
      "step": 667
    },
    {
      "epoch": 0.5798611111111112,
      "grad_norm": 0.0008550947532057762,
      "learning_rate": 7.62945075005075e-05,
      "loss": 11.5,
      "step": 668
    },
    {
      "epoch": 0.5807291666666666,
      "grad_norm": 0.001318315276876092,
      "learning_rate": 7.602734311599918e-05,
      "loss": 11.5,
      "step": 669
    },
    {
      "epoch": 0.5815972222222222,
      "grad_norm": 0.0005240674945525825,
      "learning_rate": 7.576036015076837e-05,
      "loss": 11.5,
      "step": 670
    },
    {
      "epoch": 0.5824652777777778,
      "grad_norm": 0.0018246195977553725,
      "learning_rate": 7.549356062527773e-05,
      "loss": 11.5,
      "step": 671
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 0.0021888245828449726,
      "learning_rate": 7.52269465586016e-05,
      "loss": 11.5,
      "step": 672
    },
    {
      "epoch": 0.5842013888888888,
      "grad_norm": 0.001586229307577014,
      "learning_rate": 7.496051996841084e-05,
      "loss": 11.5,
      "step": 673
    },
    {
      "epoch": 0.5850694444444444,
      "grad_norm": 0.0016963272355496883,
      "learning_rate": 7.469428287095757e-05,
      "loss": 11.5,
      "step": 674
    },
    {
      "epoch": 0.5859375,
      "grad_norm": 0.00124375952873379,
      "learning_rate": 7.442823728105979e-05,
      "loss": 11.5,
      "step": 675
    },
    {
      "epoch": 0.5868055555555556,
      "grad_norm": 0.0009501660242676735,
      "learning_rate": 7.416238521208634e-05,
      "loss": 11.5,
      "step": 676
    },
    {
      "epoch": 0.5876736111111112,
      "grad_norm": 0.0007508894195780158,
      "learning_rate": 7.389672867594142e-05,
      "loss": 11.5,
      "step": 677
    },
    {
      "epoch": 0.5885416666666666,
      "grad_norm": 0.0009574270225130022,
      "learning_rate": 7.363126968304955e-05,
      "loss": 11.5,
      "step": 678
    },
    {
      "epoch": 0.5894097222222222,
      "grad_norm": 0.001565096084959805,
      "learning_rate": 7.336601024234028e-05,
      "loss": 11.5,
      "step": 679
    },
    {
      "epoch": 0.5902777777777778,
      "grad_norm": 0.001496995915658772,
      "learning_rate": 7.310095236123295e-05,
      "loss": 11.5,
      "step": 680
    },
    {
      "epoch": 0.5911458333333334,
      "grad_norm": 0.0015568832168355584,
      "learning_rate": 7.283609804562168e-05,
      "loss": 11.5,
      "step": 681
    },
    {
      "epoch": 0.5920138888888888,
      "grad_norm": 0.002175402594730258,
      "learning_rate": 7.257144929985988e-05,
      "loss": 11.5,
      "step": 682
    },
    {
      "epoch": 0.5928819444444444,
      "grad_norm": 0.0047178422100842,
      "learning_rate": 7.230700812674538e-05,
      "loss": 11.5,
      "step": 683
    },
    {
      "epoch": 0.59375,
      "grad_norm": 0.0009569521062076092,
      "learning_rate": 7.204277652750512e-05,
      "loss": 11.5,
      "step": 684
    },
    {
      "epoch": 0.5946180555555556,
      "grad_norm": 0.0014703463530167937,
      "learning_rate": 7.177875650177998e-05,
      "loss": 11.5,
      "step": 685
    },
    {
      "epoch": 0.5954861111111112,
      "grad_norm": 0.0010557578643783927,
      "learning_rate": 7.15149500476098e-05,
      "loss": 11.5,
      "step": 686
    },
    {
      "epoch": 0.5963541666666666,
      "grad_norm": 0.0008688794914633036,
      "learning_rate": 7.12513591614181e-05,
      "loss": 11.5,
      "step": 687
    },
    {
      "epoch": 0.5972222222222222,
      "grad_norm": 0.0013972541783005,
      "learning_rate": 7.098798583799701e-05,
      "loss": 11.5,
      "step": 688
    },
    {
      "epoch": 0.5980902777777778,
      "grad_norm": 0.001837921212427318,
      "learning_rate": 7.072483207049232e-05,
      "loss": 11.5,
      "step": 689
    },
    {
      "epoch": 0.5989583333333334,
      "grad_norm": 0.0008326444076374173,
      "learning_rate": 7.046189985038807e-05,
      "loss": 11.5,
      "step": 690
    },
    {
      "epoch": 0.5998263888888888,
      "grad_norm": 0.0012137454468756914,
      "learning_rate": 7.01991911674919e-05,
      "loss": 11.5,
      "step": 691
    },
    {
      "epoch": 0.6006944444444444,
      "grad_norm": 0.002297397470101714,
      "learning_rate": 6.99367080099196e-05,
      "loss": 11.5,
      "step": 692
    },
    {
      "epoch": 0.6015625,
      "grad_norm": 0.0020339798647910357,
      "learning_rate": 6.967445236408037e-05,
      "loss": 11.5,
      "step": 693
    },
    {
      "epoch": 0.6024305555555556,
      "grad_norm": 0.0013982646632939577,
      "learning_rate": 6.941242621466157e-05,
      "loss": 11.5,
      "step": 694
    },
    {
      "epoch": 0.6032986111111112,
      "grad_norm": 0.0016188188455998898,
      "learning_rate": 6.915063154461379e-05,
      "loss": 11.5,
      "step": 695
    },
    {
      "epoch": 0.6041666666666666,
      "grad_norm": 0.0019961579237133265,
      "learning_rate": 6.888907033513592e-05,
      "loss": 11.5,
      "step": 696
    },
    {
      "epoch": 0.6050347222222222,
      "grad_norm": 0.0013090057764202356,
      "learning_rate": 6.862774456565999e-05,
      "loss": 11.5,
      "step": 697
    },
    {
      "epoch": 0.6059027777777778,
      "grad_norm": 0.0015911331865936518,
      "learning_rate": 6.836665621383635e-05,
      "loss": 11.5,
      "step": 698
    },
    {
      "epoch": 0.6067708333333334,
      "grad_norm": 0.0022514669690281153,
      "learning_rate": 6.810580725551853e-05,
      "loss": 11.5,
      "step": 699
    },
    {
      "epoch": 0.6076388888888888,
      "grad_norm": 0.002375276293605566,
      "learning_rate": 6.784519966474845e-05,
      "loss": 11.5,
      "step": 700
    },
    {
      "epoch": 0.6085069444444444,
      "grad_norm": 0.001514305593445897,
      "learning_rate": 6.758483541374148e-05,
      "loss": 11.5,
      "step": 701
    },
    {
      "epoch": 0.609375,
      "grad_norm": 0.0015389126492664218,
      "learning_rate": 6.732471647287133e-05,
      "loss": 11.5,
      "step": 702
    },
    {
      "epoch": 0.6102430555555556,
      "grad_norm": 0.0016499256016686559,
      "learning_rate": 6.706484481065539e-05,
      "loss": 11.5,
      "step": 703
    },
    {
      "epoch": 0.6111111111111112,
      "grad_norm": 0.004142056684941053,
      "learning_rate": 6.680522239373953e-05,
      "loss": 11.5,
      "step": 704
    },
    {
      "epoch": 0.6119791666666666,
      "grad_norm": 0.0009014815441332757,
      "learning_rate": 6.654585118688356e-05,
      "loss": 11.5,
      "step": 705
    },
    {
      "epoch": 0.6128472222222222,
      "grad_norm": 0.002275189384818077,
      "learning_rate": 6.628673315294616e-05,
      "loss": 11.5,
      "step": 706
    },
    {
      "epoch": 0.6137152777777778,
      "grad_norm": 0.0008502635755576193,
      "learning_rate": 6.602787025286998e-05,
      "loss": 11.5,
      "step": 707
    },
    {
      "epoch": 0.6145833333333334,
      "grad_norm": 0.0017711890395730734,
      "learning_rate": 6.576926444566698e-05,
      "loss": 11.5,
      "step": 708
    },
    {
      "epoch": 0.6154513888888888,
      "grad_norm": 0.0005183310131542385,
      "learning_rate": 6.551091768840338e-05,
      "loss": 11.5,
      "step": 709
    },
    {
      "epoch": 0.6163194444444444,
      "grad_norm": 0.0013101928634569049,
      "learning_rate": 6.525283193618508e-05,
      "loss": 11.5,
      "step": 710
    },
    {
      "epoch": 0.6171875,
      "grad_norm": 0.0026076992508023977,
      "learning_rate": 6.499500914214276e-05,
      "loss": 11.5,
      "step": 711
    },
    {
      "epoch": 0.6180555555555556,
      "grad_norm": 0.0014089919859543443,
      "learning_rate": 6.473745125741701e-05,
      "loss": 11.5,
      "step": 712
    },
    {
      "epoch": 0.6189236111111112,
      "grad_norm": 0.0005611629458144307,
      "learning_rate": 6.448016023114372e-05,
      "loss": 11.5,
      "step": 713
    },
    {
      "epoch": 0.6197916666666666,
      "grad_norm": 0.001678275759331882,
      "learning_rate": 6.422313801043922e-05,
      "loss": 11.5,
      "step": 714
    },
    {
      "epoch": 0.6206597222222222,
      "grad_norm": 0.0009339358657598495,
      "learning_rate": 6.396638654038557e-05,
      "loss": 11.5,
      "step": 715
    },
    {
      "epoch": 0.6215277777777778,
      "grad_norm": 0.000824684917461127,
      "learning_rate": 6.370990776401597e-05,
      "loss": 11.5,
      "step": 716
    },
    {
      "epoch": 0.6223958333333334,
      "grad_norm": 0.0014638352440670133,
      "learning_rate": 6.34537036222998e-05,
      "loss": 11.5,
      "step": 717
    },
    {
      "epoch": 0.6232638888888888,
      "grad_norm": 0.0010688347974792123,
      "learning_rate": 6.319777605412816e-05,
      "loss": 11.5,
      "step": 718
    },
    {
      "epoch": 0.6241319444444444,
      "grad_norm": 0.0009802256245166063,
      "learning_rate": 6.294212699629902e-05,
      "loss": 11.5,
      "step": 719
    },
    {
      "epoch": 0.625,
      "grad_norm": 0.002525707008317113,
      "learning_rate": 6.268675838350273e-05,
      "loss": 11.5,
      "step": 720
    },
    {
      "epoch": 0.6258680555555556,
      "grad_norm": 0.0027070611249655485,
      "learning_rate": 6.243167214830733e-05,
      "loss": 11.5,
      "step": 721
    },
    {
      "epoch": 0.6267361111111112,
      "grad_norm": 0.0007165224524214864,
      "learning_rate": 6.217687022114377e-05,
      "loss": 11.5,
      "step": 722
    },
    {
      "epoch": 0.6276041666666666,
      "grad_norm": 0.0014036680804565549,
      "learning_rate": 6.192235453029155e-05,
      "loss": 11.5,
      "step": 723
    },
    {
      "epoch": 0.6284722222222222,
      "grad_norm": 0.0024294545873999596,
      "learning_rate": 6.166812700186389e-05,
      "loss": 11.5,
      "step": 724
    },
    {
      "epoch": 0.6293402777777778,
      "grad_norm": 0.0022730554919689894,
      "learning_rate": 6.141418955979339e-05,
      "loss": 11.5,
      "step": 725
    },
    {
      "epoch": 0.6302083333333334,
      "grad_norm": 0.0037039376329630613,
      "learning_rate": 6.116054412581722e-05,
      "loss": 11.5,
      "step": 726
    },
    {
      "epoch": 0.6310763888888888,
      "grad_norm": 0.0015591325936838984,
      "learning_rate": 6.090719261946279e-05,
      "loss": 11.5,
      "step": 727
    },
    {
      "epoch": 0.6319444444444444,
      "grad_norm": 0.0016795331612229347,
      "learning_rate": 6.065413695803316e-05,
      "loss": 11.5,
      "step": 728
    },
    {
      "epoch": 0.6328125,
      "grad_norm": 0.0008462666883133352,
      "learning_rate": 6.0401379056592375e-05,
      "loss": 11.5,
      "step": 729
    },
    {
      "epoch": 0.6336805555555556,
      "grad_norm": 0.00047677994007244706,
      "learning_rate": 6.014892082795127e-05,
      "loss": 11.5,
      "step": 730
    },
    {
      "epoch": 0.6345486111111112,
      "grad_norm": 0.0008934232755564153,
      "learning_rate": 5.9896764182652654e-05,
      "loss": 11.5,
      "step": 731
    },
    {
      "epoch": 0.6354166666666666,
      "grad_norm": 0.002132834168151021,
      "learning_rate": 5.9644911028957196e-05,
      "loss": 11.5,
      "step": 732
    },
    {
      "epoch": 0.6362847222222222,
      "grad_norm": 0.0017735748551785946,
      "learning_rate": 5.939336327282874e-05,
      "loss": 11.5,
      "step": 733
    },
    {
      "epoch": 0.6371527777777778,
      "grad_norm": 0.0011407749261707067,
      "learning_rate": 5.914212281791989e-05,
      "loss": 11.5,
      "step": 734
    },
    {
      "epoch": 0.6380208333333334,
      "grad_norm": 0.0011274763382971287,
      "learning_rate": 5.889119156555778e-05,
      "loss": 11.5,
      "step": 735
    },
    {
      "epoch": 0.6388888888888888,
      "grad_norm": 0.0016217167722061276,
      "learning_rate": 5.8640571414729486e-05,
      "loss": 11.5,
      "step": 736
    },
    {
      "epoch": 0.6397569444444444,
      "grad_norm": 0.0017279143212363124,
      "learning_rate": 5.839026426206782e-05,
      "loss": 11.5,
      "step": 737
    },
    {
      "epoch": 0.640625,
      "grad_norm": 0.0018983453046530485,
      "learning_rate": 5.814027200183685e-05,
      "loss": 11.5,
      "step": 738
    },
    {
      "epoch": 0.6414930555555556,
      "grad_norm": 0.0019789356738328934,
      "learning_rate": 5.7890596525917593e-05,
      "loss": 11.5,
      "step": 739
    },
    {
      "epoch": 0.6423611111111112,
      "grad_norm": 0.0011423438554629683,
      "learning_rate": 5.7641239723793814e-05,
      "loss": 11.5,
      "step": 740
    },
    {
      "epoch": 0.6432291666666666,
      "grad_norm": 0.001913307118229568,
      "learning_rate": 5.739220348253753e-05,
      "loss": 11.5,
      "step": 741
    },
    {
      "epoch": 0.6440972222222222,
      "grad_norm": 0.0007154548075050116,
      "learning_rate": 5.714348968679483e-05,
      "loss": 11.5,
      "step": 742
    },
    {
      "epoch": 0.6449652777777778,
      "grad_norm": 0.0034100061748176813,
      "learning_rate": 5.689510021877173e-05,
      "loss": 11.5,
      "step": 743
    },
    {
      "epoch": 0.6458333333333334,
      "grad_norm": 0.0015718983486294746,
      "learning_rate": 5.664703695821965e-05,
      "loss": 11.5,
      "step": 744
    },
    {
      "epoch": 0.6467013888888888,
      "grad_norm": 0.0018214209703728557,
      "learning_rate": 5.639930178242152e-05,
      "loss": 11.5,
      "step": 745
    },
    {
      "epoch": 0.6475694444444444,
      "grad_norm": 0.005127399228513241,
      "learning_rate": 5.6151896566177275e-05,
      "loss": 11.5,
      "step": 746
    },
    {
      "epoch": 0.6484375,
      "grad_norm": 0.0019476967863738537,
      "learning_rate": 5.590482318178982e-05,
      "loss": 11.5,
      "step": 747
    },
    {
      "epoch": 0.6493055555555556,
      "grad_norm": 0.0021703115198761225,
      "learning_rate": 5.56580834990509e-05,
      "loss": 11.5,
      "step": 748
    },
    {
      "epoch": 0.6501736111111112,
      "grad_norm": 0.003434153739362955,
      "learning_rate": 5.54116793852268e-05,
      "loss": 11.5,
      "step": 749
    },
    {
      "epoch": 0.6510416666666666,
      "grad_norm": 0.0021836410742253065,
      "learning_rate": 5.5165612705044386e-05,
      "loss": 11.5,
      "step": 750
    },
    {
      "epoch": 0.6519097222222222,
      "grad_norm": 0.0010020373156294227,
      "learning_rate": 5.491988532067687e-05,
      "loss": 11.5,
      "step": 751
    },
    {
      "epoch": 0.6527777777777778,
      "grad_norm": 0.0019810206722468138,
      "learning_rate": 5.467449909172966e-05,
      "loss": 11.5,
      "step": 752
    },
    {
      "epoch": 0.6536458333333334,
      "grad_norm": 0.0008486639126203954,
      "learning_rate": 5.442945587522659e-05,
      "loss": 11.5,
      "step": 753
    },
    {
      "epoch": 0.6545138888888888,
      "grad_norm": 0.0007141575333662331,
      "learning_rate": 5.418475752559544e-05,
      "loss": 11.5,
      "step": 754
    },
    {
      "epoch": 0.6553819444444444,
      "grad_norm": 0.0011684248456731439,
      "learning_rate": 5.394040589465431e-05,
      "loss": 11.5,
      "step": 755
    },
    {
      "epoch": 0.65625,
      "grad_norm": 0.002104508690536022,
      "learning_rate": 5.369640283159729e-05,
      "loss": 11.5,
      "step": 756
    },
    {
      "epoch": 0.6571180555555556,
      "grad_norm": 0.0004082447267137468,
      "learning_rate": 5.3452750182980625e-05,
      "loss": 11.5,
      "step": 757
    },
    {
      "epoch": 0.6579861111111112,
      "grad_norm": 0.0008840294321998954,
      "learning_rate": 5.320944979270868e-05,
      "loss": 11.5,
      "step": 758
    },
    {
      "epoch": 0.6588541666666666,
      "grad_norm": 0.0017345227533951402,
      "learning_rate": 5.296650350202008e-05,
      "loss": 11.5,
      "step": 759
    },
    {
      "epoch": 0.6597222222222222,
      "grad_norm": 0.0013238615356385708,
      "learning_rate": 5.2723913149473737e-05,
      "loss": 11.5,
      "step": 760
    },
    {
      "epoch": 0.6605902777777778,
      "grad_norm": 0.0019203656120225787,
      "learning_rate": 5.248168057093478e-05,
      "loss": 11.5,
      "step": 761
    },
    {
      "epoch": 0.6614583333333334,
      "grad_norm": 0.0011490945471450686,
      "learning_rate": 5.223980759956089e-05,
      "loss": 11.5,
      "step": 762
    },
    {
      "epoch": 0.6623263888888888,
      "grad_norm": 0.001213114708662033,
      "learning_rate": 5.199829606578823e-05,
      "loss": 11.5,
      "step": 763
    },
    {
      "epoch": 0.6631944444444444,
      "grad_norm": 0.001129034673795104,
      "learning_rate": 5.175714779731782e-05,
      "loss": 11.5,
      "step": 764
    },
    {
      "epoch": 0.6640625,
      "grad_norm": 0.0007731254445388913,
      "learning_rate": 5.151636461910154e-05,
      "loss": 11.5,
      "step": 765
    },
    {
      "epoch": 0.6649305555555556,
      "grad_norm": 0.00102410523686558,
      "learning_rate": 5.127594835332831e-05,
      "loss": 11.5,
      "step": 766
    },
    {
      "epoch": 0.6657986111111112,
      "grad_norm": 0.0012864811578765512,
      "learning_rate": 5.1035900819410346e-05,
      "loss": 11.5,
      "step": 767
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.001206637010909617,
      "learning_rate": 5.079622383396938e-05,
      "loss": 11.5,
      "step": 768
    },
    {
      "epoch": 0.6675347222222222,
      "grad_norm": 0.0007971184677444398,
      "learning_rate": 5.055691921082299e-05,
      "loss": 11.5,
      "step": 769
    },
    {
      "epoch": 0.6684027777777778,
      "grad_norm": 0.001482365420088172,
      "learning_rate": 5.0317988760970804e-05,
      "loss": 11.5,
      "step": 770
    },
    {
      "epoch": 0.6692708333333334,
      "grad_norm": 0.0006629691924899817,
      "learning_rate": 5.007943429258073e-05,
      "loss": 11.5,
      "step": 771
    },
    {
      "epoch": 0.6701388888888888,
      "grad_norm": 0.0016475849552080035,
      "learning_rate": 4.9841257610975364e-05,
      "loss": 11.5,
      "step": 772
    },
    {
      "epoch": 0.6710069444444444,
      "grad_norm": 0.0017213310347869992,
      "learning_rate": 4.9603460518618274e-05,
      "loss": 11.5,
      "step": 773
    },
    {
      "epoch": 0.671875,
      "grad_norm": 0.0011116741225123405,
      "learning_rate": 4.936604481510045e-05,
      "loss": 11.5,
      "step": 774
    },
    {
      "epoch": 0.6727430555555556,
      "grad_norm": 0.0014985189773142338,
      "learning_rate": 4.912901229712665e-05,
      "loss": 11.5,
      "step": 775
    },
    {
      "epoch": 0.6736111111111112,
      "grad_norm": 0.0025598807260394096,
      "learning_rate": 4.889236475850169e-05,
      "loss": 11.5,
      "step": 776
    },
    {
      "epoch": 0.6744791666666666,
      "grad_norm": 0.001339590991847217,
      "learning_rate": 4.865610399011696e-05,
      "loss": 11.5,
      "step": 777
    },
    {
      "epoch": 0.6753472222222222,
      "grad_norm": 0.0009329714230261743,
      "learning_rate": 4.8420231779936885e-05,
      "loss": 11.5,
      "step": 778
    },
    {
      "epoch": 0.6762152777777778,
      "grad_norm": 0.0026437542401254177,
      "learning_rate": 4.818474991298541e-05,
      "loss": 11.5,
      "step": 779
    },
    {
      "epoch": 0.6770833333333334,
      "grad_norm": 0.001179800252430141,
      "learning_rate": 4.7949660171332475e-05,
      "loss": 11.5,
      "step": 780
    },
    {
      "epoch": 0.6779513888888888,
      "grad_norm": 0.0008856172789819539,
      "learning_rate": 4.771496433408043e-05,
      "loss": 11.5,
      "step": 781
    },
    {
      "epoch": 0.6788194444444444,
      "grad_norm": 0.0013951617293059826,
      "learning_rate": 4.748066417735068e-05,
      "loss": 11.5,
      "step": 782
    },
    {
      "epoch": 0.6796875,
      "grad_norm": 0.0011536451056599617,
      "learning_rate": 4.724676147427018e-05,
      "loss": 11.5,
      "step": 783
    },
    {
      "epoch": 0.6805555555555556,
      "grad_norm": 0.001306037069298327,
      "learning_rate": 4.701325799495817e-05,
      "loss": 11.5,
      "step": 784
    },
    {
      "epoch": 0.6814236111111112,
      "grad_norm": 0.0006856722175143659,
      "learning_rate": 4.678015550651258e-05,
      "loss": 11.5,
      "step": 785
    },
    {
      "epoch": 0.6822916666666666,
      "grad_norm": 0.002181056886911392,
      "learning_rate": 4.654745577299674e-05,
      "loss": 11.5,
      "step": 786
    },
    {
      "epoch": 0.6831597222222222,
      "grad_norm": 0.004753122106194496,
      "learning_rate": 4.631516055542601e-05,
      "loss": 11.5,
      "step": 787
    },
    {
      "epoch": 0.6840277777777778,
      "grad_norm": 0.0011703430209308863,
      "learning_rate": 4.6083271611754476e-05,
      "loss": 11.5,
      "step": 788
    },
    {
      "epoch": 0.6848958333333334,
      "grad_norm": 0.0038099815137684345,
      "learning_rate": 4.585179069686171e-05,
      "loss": 11.5,
      "step": 789
    },
    {
      "epoch": 0.6857638888888888,
      "grad_norm": 0.0016217122320085764,
      "learning_rate": 4.5620719562539314e-05,
      "loss": 11.5,
      "step": 790
    },
    {
      "epoch": 0.6866319444444444,
      "grad_norm": 0.0017002298263832927,
      "learning_rate": 4.5390059957477904e-05,
      "loss": 11.5,
      "step": 791
    },
    {
      "epoch": 0.6875,
      "grad_norm": 0.002357694087550044,
      "learning_rate": 4.515981362725361e-05,
      "loss": 11.5,
      "step": 792
    },
    {
      "epoch": 0.6883680555555556,
      "grad_norm": 0.0011604288592934608,
      "learning_rate": 4.4929982314315024e-05,
      "loss": 11.5,
      "step": 793
    },
    {
      "epoch": 0.6892361111111112,
      "grad_norm": 0.0008630178635939956,
      "learning_rate": 4.470056775797008e-05,
      "loss": 11.5,
      "step": 794
    },
    {
      "epoch": 0.6901041666666666,
      "grad_norm": 0.004088199231773615,
      "learning_rate": 4.447157169437267e-05,
      "loss": 11.5,
      "step": 795
    },
    {
      "epoch": 0.6909722222222222,
      "grad_norm": 0.0020213096868246794,
      "learning_rate": 4.424299585650977e-05,
      "loss": 11.5,
      "step": 796
    },
    {
      "epoch": 0.6918402777777778,
      "grad_norm": 0.004031007643789053,
      "learning_rate": 4.401484197418807e-05,
      "loss": 11.5,
      "step": 797
    },
    {
      "epoch": 0.6927083333333334,
      "grad_norm": 0.001632015686482191,
      "learning_rate": 4.3787111774021014e-05,
      "loss": 11.5,
      "step": 798
    },
    {
      "epoch": 0.6935763888888888,
      "grad_norm": 0.0012832857901230454,
      "learning_rate": 4.355980697941584e-05,
      "loss": 11.5,
      "step": 799
    },
    {
      "epoch": 0.6944444444444444,
      "grad_norm": 0.004337807185947895,
      "learning_rate": 4.3332929310560265e-05,
      "loss": 11.5,
      "step": 800
    },
    {
      "epoch": 0.6953125,
      "grad_norm": 0.000916890858206898,
      "learning_rate": 4.310648048440975e-05,
      "loss": 11.5,
      "step": 801
    },
    {
      "epoch": 0.6961805555555556,
      "grad_norm": 0.00043458392610773444,
      "learning_rate": 4.2880462214674287e-05,
      "loss": 11.5,
      "step": 802
    },
    {
      "epoch": 0.6970486111111112,
      "grad_norm": 0.0005981852882541716,
      "learning_rate": 4.2654876211805527e-05,
      "loss": 11.5,
      "step": 803
    },
    {
      "epoch": 0.6979166666666666,
      "grad_norm": 0.0010769993532449007,
      "learning_rate": 4.2429724182983935e-05,
      "loss": 11.5,
      "step": 804
    },
    {
      "epoch": 0.6987847222222222,
      "grad_norm": 0.00040476801223121583,
      "learning_rate": 4.220500783210559e-05,
      "loss": 11.5,
      "step": 805
    },
    {
      "epoch": 0.6996527777777778,
      "grad_norm": 0.0012501499149948359,
      "learning_rate": 4.198072885976966e-05,
      "loss": 11.5,
      "step": 806
    },
    {
      "epoch": 0.7005208333333334,
      "grad_norm": 0.0005873345071449876,
      "learning_rate": 4.1756888963265194e-05,
      "loss": 11.5,
      "step": 807
    },
    {
      "epoch": 0.7013888888888888,
      "grad_norm": 0.0005547106848098338,
      "learning_rate": 4.153348983655844e-05,
      "loss": 11.5,
      "step": 808
    },
    {
      "epoch": 0.7022569444444444,
      "grad_norm": 0.0006437045522034168,
      "learning_rate": 4.1310533170280084e-05,
      "loss": 11.5,
      "step": 809
    },
    {
      "epoch": 0.703125,
      "grad_norm": 0.0011192302918061614,
      "learning_rate": 4.108802065171229e-05,
      "loss": 11.5,
      "step": 810
    },
    {
      "epoch": 0.7039930555555556,
      "grad_norm": 0.0014064739225432277,
      "learning_rate": 4.086595396477611e-05,
      "loss": 11.5,
      "step": 811
    },
    {
      "epoch": 0.7048611111111112,
      "grad_norm": 0.0010803134646266699,
      "learning_rate": 4.064433479001857e-05,
      "loss": 11.5,
      "step": 812
    },
    {
      "epoch": 0.7057291666666666,
      "grad_norm": 0.002670304151251912,
      "learning_rate": 4.042316480460001e-05,
      "loss": 11.5,
      "step": 813
    },
    {
      "epoch": 0.7065972222222222,
      "grad_norm": 0.0009920287411659956,
      "learning_rate": 4.020244568228155e-05,
      "loss": 11.5,
      "step": 814
    },
    {
      "epoch": 0.7074652777777778,
      "grad_norm": 0.0012886729091405869,
      "learning_rate": 3.998217909341213e-05,
      "loss": 11.5,
      "step": 815
    },
    {
      "epoch": 0.7083333333333334,
      "grad_norm": 0.0018561139004305005,
      "learning_rate": 3.9762366704916165e-05,
      "loss": 11.5,
      "step": 816
    },
    {
      "epoch": 0.7092013888888888,
      "grad_norm": 0.0019951267167925835,
      "learning_rate": 3.954301018028068e-05,
      "loss": 11.5,
      "step": 817
    },
    {
      "epoch": 0.7100694444444444,
      "grad_norm": 0.0004654499643947929,
      "learning_rate": 3.932411117954286e-05,
      "loss": 11.5,
      "step": 818
    },
    {
      "epoch": 0.7109375,
      "grad_norm": 0.0012378923129290342,
      "learning_rate": 3.9105671359277517e-05,
      "loss": 11.5,
      "step": 819
    },
    {
      "epoch": 0.7118055555555556,
      "grad_norm": 0.0007940977229736745,
      "learning_rate": 3.8887692372584405e-05,
      "loss": 11.5,
      "step": 820
    },
    {
      "epoch": 0.7126736111111112,
      "grad_norm": 0.0008617853163741529,
      "learning_rate": 3.867017586907591e-05,
      "loss": 11.5,
      "step": 821
    },
    {
      "epoch": 0.7135416666666666,
      "grad_norm": 0.0011360191274434328,
      "learning_rate": 3.845312349486437e-05,
      "loss": 11.5,
      "step": 822
    },
    {
      "epoch": 0.7144097222222222,
      "grad_norm": 0.0004947633133269846,
      "learning_rate": 3.82365368925497e-05,
      "loss": 11.5,
      "step": 823
    },
    {
      "epoch": 0.7152777777777778,
      "grad_norm": 0.0016279894625768065,
      "learning_rate": 3.8020417701207056e-05,
      "loss": 11.5,
      "step": 824
    },
    {
      "epoch": 0.7161458333333334,
      "grad_norm": 0.0008642195607535541,
      "learning_rate": 3.780476755637421e-05,
      "loss": 11.5,
      "step": 825
    },
    {
      "epoch": 0.7170138888888888,
      "grad_norm": 0.002968879183754325,
      "learning_rate": 3.758958809003945e-05,
      "loss": 11.5,
      "step": 826
    },
    {
      "epoch": 0.7178819444444444,
      "grad_norm": 0.0009541690233163536,
      "learning_rate": 3.7374880930628986e-05,
      "loss": 11.5,
      "step": 827
    },
    {
      "epoch": 0.71875,
      "grad_norm": 0.0014237717259675264,
      "learning_rate": 3.71606477029947e-05,
      "loss": 11.5,
      "step": 828
    },
    {
      "epoch": 0.7196180555555556,
      "grad_norm": 0.00066854536999017,
      "learning_rate": 3.694689002840198e-05,
      "loss": 11.5,
      "step": 829
    },
    {
      "epoch": 0.7204861111111112,
      "grad_norm": 0.000623007130343467,
      "learning_rate": 3.673360952451722e-05,
      "loss": 11.5,
      "step": 830
    },
    {
      "epoch": 0.7213541666666666,
      "grad_norm": 0.0019452956039458513,
      "learning_rate": 3.652080780539584e-05,
      "loss": 11.5,
      "step": 831
    },
    {
      "epoch": 0.7222222222222222,
      "grad_norm": 0.0007634107605554163,
      "learning_rate": 3.6308486481469805e-05,
      "loss": 11.5,
      "step": 832
    },
    {
      "epoch": 0.7230902777777778,
      "grad_norm": 0.0009589885594323277,
      "learning_rate": 3.609664715953559e-05,
      "loss": 11.5,
      "step": 833
    },
    {
      "epoch": 0.7239583333333334,
      "grad_norm": 0.0016271749045699835,
      "learning_rate": 3.5885291442742084e-05,
      "loss": 11.5,
      "step": 834
    },
    {
      "epoch": 0.7248263888888888,
      "grad_norm": 0.001315324567258358,
      "learning_rate": 3.567442093057825e-05,
      "loss": 11.5,
      "step": 835
    },
    {
      "epoch": 0.7256944444444444,
      "grad_norm": 0.002419763244688511,
      "learning_rate": 3.546403721886126e-05,
      "loss": 11.5,
      "step": 836
    },
    {
      "epoch": 0.7265625,
      "grad_norm": 0.001359086949378252,
      "learning_rate": 3.5254141899724194e-05,
      "loss": 11.5,
      "step": 837
    },
    {
      "epoch": 0.7274305555555556,
      "grad_norm": 0.0016217930242419243,
      "learning_rate": 3.50447365616041e-05,
      "loss": 11.5,
      "step": 838
    },
    {
      "epoch": 0.7282986111111112,
      "grad_norm": 0.0007151829195208848,
      "learning_rate": 3.483582278923007e-05,
      "loss": 11.5,
      "step": 839
    },
    {
      "epoch": 0.7291666666666666,
      "grad_norm": 0.0013489589327946305,
      "learning_rate": 3.462740216361101e-05,
      "loss": 11.5,
      "step": 840
    },
    {
      "epoch": 0.7300347222222222,
      "grad_norm": 0.0011385135585442185,
      "learning_rate": 3.4419476262023956e-05,
      "loss": 11.5,
      "step": 841
    },
    {
      "epoch": 0.7309027777777778,
      "grad_norm": 0.0025621172972023487,
      "learning_rate": 3.421204665800186e-05,
      "loss": 11.5,
      "step": 842
    },
    {
      "epoch": 0.7317708333333334,
      "grad_norm": 0.0031880212482064962,
      "learning_rate": 3.400511492132186e-05,
      "loss": 11.5,
      "step": 843
    },
    {
      "epoch": 0.7326388888888888,
      "grad_norm": 0.0033356111962348223,
      "learning_rate": 3.3798682617993405e-05,
      "loss": 11.5,
      "step": 844
    },
    {
      "epoch": 0.7335069444444444,
      "grad_norm": 0.0022079595364630222,
      "learning_rate": 3.359275131024625e-05,
      "loss": 11.5,
      "step": 845
    },
    {
      "epoch": 0.734375,
      "grad_norm": 0.0009670985746197402,
      "learning_rate": 3.338732255651887e-05,
      "loss": 11.5,
      "step": 846
    },
    {
      "epoch": 0.7352430555555556,
      "grad_norm": 0.0011886127758771181,
      "learning_rate": 3.3182397911446436e-05,
      "loss": 11.5,
      "step": 847
    },
    {
      "epoch": 0.7361111111111112,
      "grad_norm": 0.0016066280659288168,
      "learning_rate": 3.297797892584912e-05,
      "loss": 11.5,
      "step": 848
    },
    {
      "epoch": 0.7369791666666666,
      "grad_norm": 0.001674711937084794,
      "learning_rate": 3.277406714672049e-05,
      "loss": 11.5,
      "step": 849
    },
    {
      "epoch": 0.7378472222222222,
      "grad_norm": 0.002176313428208232,
      "learning_rate": 3.25706641172156e-05,
      "loss": 11.5,
      "step": 850
    },
    {
      "epoch": 0.7387152777777778,
      "grad_norm": 0.004424555692821741,
      "learning_rate": 3.236777137663951e-05,
      "loss": 11.5,
      "step": 851
    },
    {
      "epoch": 0.7395833333333334,
      "grad_norm": 0.0009748575394041836,
      "learning_rate": 3.216539046043544e-05,
      "loss": 11.5,
      "step": 852
    },
    {
      "epoch": 0.7404513888888888,
      "grad_norm": 0.0022450436372309923,
      "learning_rate": 3.1963522900173294e-05,
      "loss": 11.5,
      "step": 853
    },
    {
      "epoch": 0.7413194444444444,
      "grad_norm": 0.001965005649253726,
      "learning_rate": 3.176217022353798e-05,
      "loss": 11.5,
      "step": 854
    },
    {
      "epoch": 0.7421875,
      "grad_norm": 0.004234009888023138,
      "learning_rate": 3.156133395431795e-05,
      "loss": 11.5,
      "step": 855
    },
    {
      "epoch": 0.7430555555555556,
      "grad_norm": 0.001011452404782176,
      "learning_rate": 3.136101561239363e-05,
      "loss": 11.5,
      "step": 856
    },
    {
      "epoch": 0.7439236111111112,
      "grad_norm": 0.001120606902986765,
      "learning_rate": 3.116121671372582e-05,
      "loss": 11.5,
      "step": 857
    },
    {
      "epoch": 0.7447916666666666,
      "grad_norm": 0.001175688928924501,
      "learning_rate": 3.09619387703443e-05,
      "loss": 11.5,
      "step": 858
    },
    {
      "epoch": 0.7456597222222222,
      "grad_norm": 0.0011202063178643584,
      "learning_rate": 3.076318329033643e-05,
      "loss": 11.5,
      "step": 859
    },
    {
      "epoch": 0.7465277777777778,
      "grad_norm": 0.001799761550500989,
      "learning_rate": 3.056495177783567e-05,
      "loss": 11.5,
      "step": 860
    },
    {
      "epoch": 0.7473958333333334,
      "grad_norm": 0.0012393156066536903,
      "learning_rate": 3.0367245733010298e-05,
      "loss": 11.5,
      "step": 861
    },
    {
      "epoch": 0.7482638888888888,
      "grad_norm": 0.001119028078392148,
      "learning_rate": 3.0170066652051866e-05,
      "loss": 11.5,
      "step": 862
    },
    {
      "epoch": 0.7491319444444444,
      "grad_norm": 0.000896016659680754,
      "learning_rate": 2.9973416027164025e-05,
      "loss": 11.5,
      "step": 863
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.0009167541866190732,
      "learning_rate": 2.9777295346551194e-05,
      "loss": 11.5,
      "step": 864
    },
    {
      "epoch": 0.75,
      "eval_loss": 11.5,
      "eval_runtime": 3.9722,
      "eval_samples_per_second": 122.098,
      "eval_steps_per_second": 61.175,
      "step": 864
    },
    {
      "epoch": 0.7508680555555556,
      "grad_norm": 0.0014070029137656093,
      "learning_rate": 2.9581706094407346e-05,
      "loss": 11.5,
      "step": 865
    },
    {
      "epoch": 0.7517361111111112,
      "grad_norm": 0.0013805751223117113,
      "learning_rate": 2.9386649750904738e-05,
      "loss": 11.5,
      "step": 866
    },
    {
      "epoch": 0.7526041666666666,
      "grad_norm": 0.0009950497187674046,
      "learning_rate": 2.9192127792182633e-05,
      "loss": 11.5,
      "step": 867
    },
    {
      "epoch": 0.7534722222222222,
      "grad_norm": 0.000892390962690115,
      "learning_rate": 2.899814169033622e-05,
      "loss": 11.5,
      "step": 868
    },
    {
      "epoch": 0.7543402777777778,
      "grad_norm": 0.0018002573633566499,
      "learning_rate": 2.8804692913405474e-05,
      "loss": 11.5,
      "step": 869
    },
    {
      "epoch": 0.7552083333333334,
      "grad_norm": 0.001591190230101347,
      "learning_rate": 2.8611782925364006e-05,
      "loss": 11.5,
      "step": 870
    },
    {
      "epoch": 0.7560763888888888,
      "grad_norm": 0.0007806799840182066,
      "learning_rate": 2.841941318610808e-05,
      "loss": 11.5,
      "step": 871
    },
    {
      "epoch": 0.7569444444444444,
      "grad_norm": 0.0008195408154278994,
      "learning_rate": 2.8227585151445367e-05,
      "loss": 11.5,
      "step": 872
    },
    {
      "epoch": 0.7578125,
      "grad_norm": 0.001267914893105626,
      "learning_rate": 2.8036300273084125e-05,
      "loss": 11.5,
      "step": 873
    },
    {
      "epoch": 0.7586805555555556,
      "grad_norm": 0.002437069546431303,
      "learning_rate": 2.784555999862206e-05,
      "loss": 11.5,
      "step": 874
    },
    {
      "epoch": 0.7595486111111112,
      "grad_norm": 0.0014965480659157038,
      "learning_rate": 2.7655365771535556e-05,
      "loss": 11.5,
      "step": 875
    },
    {
      "epoch": 0.7604166666666666,
      "grad_norm": 0.0010592719772830606,
      "learning_rate": 2.7465719031168612e-05,
      "loss": 11.5,
      "step": 876
    },
    {
      "epoch": 0.7612847222222222,
      "grad_norm": 0.002251991303637624,
      "learning_rate": 2.727662121272192e-05,
      "loss": 11.5,
      "step": 877
    },
    {
      "epoch": 0.7621527777777778,
      "grad_norm": 0.0018771494505926967,
      "learning_rate": 2.7088073747242104e-05,
      "loss": 11.5,
      "step": 878
    },
    {
      "epoch": 0.7630208333333334,
      "grad_norm": 0.00392155209556222,
      "learning_rate": 2.6900078061610822e-05,
      "loss": 11.5,
      "step": 879
    },
    {
      "epoch": 0.7638888888888888,
      "grad_norm": 0.001540718600153923,
      "learning_rate": 2.671263557853404e-05,
      "loss": 11.5,
      "step": 880
    },
    {
      "epoch": 0.7647569444444444,
      "grad_norm": 0.0007441625348292291,
      "learning_rate": 2.6525747716531236e-05,
      "loss": 11.5,
      "step": 881
    },
    {
      "epoch": 0.765625,
      "grad_norm": 0.0025893773417919874,
      "learning_rate": 2.633941588992458e-05,
      "loss": 11.5,
      "step": 882
    },
    {
      "epoch": 0.7664930555555556,
      "grad_norm": 0.0036958770360797644,
      "learning_rate": 2.615364150882833e-05,
      "loss": 11.5,
      "step": 883
    },
    {
      "epoch": 0.7673611111111112,
      "grad_norm": 0.0012324008857831359,
      "learning_rate": 2.5968425979138112e-05,
      "loss": 11.5,
      "step": 884
    },
    {
      "epoch": 0.7682291666666666,
      "grad_norm": 0.003916856832802296,
      "learning_rate": 2.578377070252038e-05,
      "loss": 11.5,
      "step": 885
    },
    {
      "epoch": 0.7690972222222222,
      "grad_norm": 0.0009747339645400643,
      "learning_rate": 2.5599677076401608e-05,
      "loss": 11.5,
      "step": 886
    },
    {
      "epoch": 0.7699652777777778,
      "grad_norm": 0.0031575867906212807,
      "learning_rate": 2.541614649395796e-05,
      "loss": 11.5,
      "step": 887
    },
    {
      "epoch": 0.7708333333333334,
      "grad_norm": 0.002601254265755415,
      "learning_rate": 2.5233180344104536e-05,
      "loss": 11.5,
      "step": 888
    },
    {
      "epoch": 0.7717013888888888,
      "grad_norm": 0.004009724594652653,
      "learning_rate": 2.505078001148492e-05,
      "loss": 11.5,
      "step": 889
    },
    {
      "epoch": 0.7725694444444444,
      "grad_norm": 0.0020184468012303114,
      "learning_rate": 2.486894687646085e-05,
      "loss": 11.5,
      "step": 890
    },
    {
      "epoch": 0.7734375,
      "grad_norm": 0.001276231836527586,
      "learning_rate": 2.4687682315101513e-05,
      "loss": 11.5,
      "step": 891
    },
    {
      "epoch": 0.7743055555555556,
      "grad_norm": 0.001120898174121976,
      "learning_rate": 2.450698769917339e-05,
      "loss": 11.5,
      "step": 892
    },
    {
      "epoch": 0.7751736111111112,
      "grad_norm": 0.0019128049025312066,
      "learning_rate": 2.4326864396129666e-05,
      "loss": 11.5,
      "step": 893
    },
    {
      "epoch": 0.7760416666666666,
      "grad_norm": 0.002192750573158264,
      "learning_rate": 2.4147313769099977e-05,
      "loss": 11.5,
      "step": 894
    },
    {
      "epoch": 0.7769097222222222,
      "grad_norm": 0.0014198143035173416,
      "learning_rate": 2.396833717688015e-05,
      "loss": 11.5,
      "step": 895
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 0.00313235423527658,
      "learning_rate": 2.3789935973921752e-05,
      "loss": 11.5,
      "step": 896
    },
    {
      "epoch": 0.7786458333333334,
      "grad_norm": 0.0008927979506552219,
      "learning_rate": 2.3612111510322054e-05,
      "loss": 11.5,
      "step": 897
    },
    {
      "epoch": 0.7795138888888888,
      "grad_norm": 0.0024561702739447355,
      "learning_rate": 2.3434865131813633e-05,
      "loss": 11.5,
      "step": 898
    },
    {
      "epoch": 0.7803819444444444,
      "grad_norm": 0.002858678577467799,
      "learning_rate": 2.325819817975423e-05,
      "loss": 11.5,
      "step": 899
    },
    {
      "epoch": 0.78125,
      "grad_norm": 0.005957089364528656,
      "learning_rate": 2.308211199111674e-05,
      "loss": 11.5,
      "step": 900
    },
    {
      "epoch": 0.7821180555555556,
      "grad_norm": 0.0005328590050339699,
      "learning_rate": 2.290660789847886e-05,
      "loss": 11.5,
      "step": 901
    },
    {
      "epoch": 0.7829861111111112,
      "grad_norm": 0.0012697110651060939,
      "learning_rate": 2.273168723001321e-05,
      "loss": 11.5,
      "step": 902
    },
    {
      "epoch": 0.7838541666666666,
      "grad_norm": 0.0009065602207556367,
      "learning_rate": 2.255735130947716e-05,
      "loss": 11.5,
      "step": 903
    },
    {
      "epoch": 0.7847222222222222,
      "grad_norm": 0.0030562197789549828,
      "learning_rate": 2.2383601456202797e-05,
      "loss": 11.5,
      "step": 904
    },
    {
      "epoch": 0.7855902777777778,
      "grad_norm": 0.0013023049104958773,
      "learning_rate": 2.2210438985087135e-05,
      "loss": 11.5,
      "step": 905
    },
    {
      "epoch": 0.7864583333333334,
      "grad_norm": 0.0026572714559733868,
      "learning_rate": 2.2037865206581855e-05,
      "loss": 11.5,
      "step": 906
    },
    {
      "epoch": 0.7873263888888888,
      "grad_norm": 0.0025171602610498667,
      "learning_rate": 2.1865881426683687e-05,
      "loss": 11.5,
      "step": 907
    },
    {
      "epoch": 0.7881944444444444,
      "grad_norm": 0.002224936615675688,
      "learning_rate": 2.169448894692433e-05,
      "loss": 11.5,
      "step": 908
    },
    {
      "epoch": 0.7890625,
      "grad_norm": 0.0030496774706989527,
      "learning_rate": 2.152368906436064e-05,
      "loss": 11.5,
      "step": 909
    },
    {
      "epoch": 0.7899305555555556,
      "grad_norm": 0.0012079485459253192,
      "learning_rate": 2.1353483071564938e-05,
      "loss": 11.5,
      "step": 910
    },
    {
      "epoch": 0.7907986111111112,
      "grad_norm": 0.0012827307218685746,
      "learning_rate": 2.1183872256615e-05,
      "loss": 11.5,
      "step": 911
    },
    {
      "epoch": 0.7916666666666666,
      "grad_norm": 0.0008129117195494473,
      "learning_rate": 2.1014857903084583e-05,
      "loss": 11.5,
      "step": 912
    },
    {
      "epoch": 0.7925347222222222,
      "grad_norm": 0.001060680253431201,
      "learning_rate": 2.084644129003346e-05,
      "loss": 11.5,
      "step": 913
    },
    {
      "epoch": 0.7934027777777778,
      "grad_norm": 0.0007513607852160931,
      "learning_rate": 2.0678623691997856e-05,
      "loss": 11.5,
      "step": 914
    },
    {
      "epoch": 0.7942708333333334,
      "grad_norm": 0.0009765429422259331,
      "learning_rate": 2.051140637898088e-05,
      "loss": 11.5,
      "step": 915
    },
    {
      "epoch": 0.7951388888888888,
      "grad_norm": 0.001668027718551457,
      "learning_rate": 2.034479061644271e-05,
      "loss": 11.5,
      "step": 916
    },
    {
      "epoch": 0.7960069444444444,
      "grad_norm": 0.0006954887066967785,
      "learning_rate": 2.017877766529126e-05,
      "loss": 11.5,
      "step": 917
    },
    {
      "epoch": 0.796875,
      "grad_norm": 0.0016365889459848404,
      "learning_rate": 2.001336878187241e-05,
      "loss": 11.5,
      "step": 918
    },
    {
      "epoch": 0.7977430555555556,
      "grad_norm": 0.001494184834882617,
      "learning_rate": 1.9848565217960635e-05,
      "loss": 11.5,
      "step": 919
    },
    {
      "epoch": 0.7986111111111112,
      "grad_norm": 0.002627990674227476,
      "learning_rate": 1.968436822074954e-05,
      "loss": 11.5,
      "step": 920
    },
    {
      "epoch": 0.7994791666666666,
      "grad_norm": 0.0005519287078641355,
      "learning_rate": 1.9520779032842283e-05,
      "loss": 11.5,
      "step": 921
    },
    {
      "epoch": 0.8003472222222222,
      "grad_norm": 0.0009910707594826818,
      "learning_rate": 1.9357798892242407e-05,
      "loss": 11.5,
      "step": 922
    },
    {
      "epoch": 0.8012152777777778,
      "grad_norm": 0.0007709738565608859,
      "learning_rate": 1.9195429032344216e-05,
      "loss": 11.5,
      "step": 923
    },
    {
      "epoch": 0.8020833333333334,
      "grad_norm": 0.0026251785457134247,
      "learning_rate": 1.9033670681923564e-05,
      "loss": 11.5,
      "step": 924
    },
    {
      "epoch": 0.8029513888888888,
      "grad_norm": 0.0024495297111570835,
      "learning_rate": 1.8872525065128632e-05,
      "loss": 11.5,
      "step": 925
    },
    {
      "epoch": 0.8038194444444444,
      "grad_norm": 0.0013684300938621163,
      "learning_rate": 1.8711993401470483e-05,
      "loss": 11.5,
      "step": 926
    },
    {
      "epoch": 0.8046875,
      "grad_norm": 0.0007330769440159202,
      "learning_rate": 1.855207690581403e-05,
      "loss": 11.5,
      "step": 927
    },
    {
      "epoch": 0.8055555555555556,
      "grad_norm": 0.0035512689501047134,
      "learning_rate": 1.839277678836866e-05,
      "loss": 11.5,
      "step": 928
    },
    {
      "epoch": 0.8064236111111112,
      "grad_norm": 0.0019437504233792424,
      "learning_rate": 1.8234094254679157e-05,
      "loss": 11.5,
      "step": 929
    },
    {
      "epoch": 0.8072916666666666,
      "grad_norm": 0.0017359075136482716,
      "learning_rate": 1.807603050561667e-05,
      "loss": 11.5,
      "step": 930
    },
    {
      "epoch": 0.8081597222222222,
      "grad_norm": 0.0014644009061157703,
      "learning_rate": 1.7918586737369424e-05,
      "loss": 11.5,
      "step": 931
    },
    {
      "epoch": 0.8090277777777778,
      "grad_norm": 0.0015440438874065876,
      "learning_rate": 1.776176414143389e-05,
      "loss": 11.5,
      "step": 932
    },
    {
      "epoch": 0.8098958333333334,
      "grad_norm": 0.004367981571704149,
      "learning_rate": 1.7605563904605558e-05,
      "loss": 11.5,
      "step": 933
    },
    {
      "epoch": 0.8107638888888888,
      "grad_norm": 0.0021727008279412985,
      "learning_rate": 1.7449987208970087e-05,
      "loss": 11.5,
      "step": 934
    },
    {
      "epoch": 0.8116319444444444,
      "grad_norm": 0.0016245286678895354,
      "learning_rate": 1.7295035231894384e-05,
      "loss": 11.5,
      "step": 935
    },
    {
      "epoch": 0.8125,
      "grad_norm": 0.0014990289928391576,
      "learning_rate": 1.714070914601752e-05,
      "loss": 11.5,
      "step": 936
    },
    {
      "epoch": 0.8133680555555556,
      "grad_norm": 0.0007343164761550725,
      "learning_rate": 1.698701011924212e-05,
      "loss": 11.5,
      "step": 937
    },
    {
      "epoch": 0.8142361111111112,
      "grad_norm": 0.0015765505377203226,
      "learning_rate": 1.6833939314725244e-05,
      "loss": 11.5,
      "step": 938
    },
    {
      "epoch": 0.8151041666666666,
      "grad_norm": 0.0012860825518146157,
      "learning_rate": 1.6681497890869756e-05,
      "loss": 11.5,
      "step": 939
    },
    {
      "epoch": 0.8159722222222222,
      "grad_norm": 0.0011854360345751047,
      "learning_rate": 1.6529687001315598e-05,
      "loss": 11.5,
      "step": 940
    },
    {
      "epoch": 0.8168402777777778,
      "grad_norm": 0.004952909890562296,
      "learning_rate": 1.6378507794930865e-05,
      "loss": 11.5,
      "step": 941
    },
    {
      "epoch": 0.8177083333333334,
      "grad_norm": 0.0016203427221626043,
      "learning_rate": 1.6227961415803305e-05,
      "loss": 11.5,
      "step": 942
    },
    {
      "epoch": 0.8185763888888888,
      "grad_norm": 0.003437288338318467,
      "learning_rate": 1.6078049003231554e-05,
      "loss": 11.5,
      "step": 943
    },
    {
      "epoch": 0.8194444444444444,
      "grad_norm": 0.0027985884808003902,
      "learning_rate": 1.5928771691716517e-05,
      "loss": 11.5,
      "step": 944
    },
    {
      "epoch": 0.8203125,
      "grad_norm": 0.0012315622298046947,
      "learning_rate": 1.578013061095288e-05,
      "loss": 11.5,
      "step": 945
    },
    {
      "epoch": 0.8211805555555556,
      "grad_norm": 0.0021417068783193827,
      "learning_rate": 1.5632126885820385e-05,
      "loss": 11.5,
      "step": 946
    },
    {
      "epoch": 0.8220486111111112,
      "grad_norm": 0.0027796742506325245,
      "learning_rate": 1.5484761636375543e-05,
      "loss": 11.5,
      "step": 947
    },
    {
      "epoch": 0.8229166666666666,
      "grad_norm": 0.0012023968156427145,
      "learning_rate": 1.5338035977842946e-05,
      "loss": 11.5,
      "step": 948
    },
    {
      "epoch": 0.8237847222222222,
      "grad_norm": 0.002301426138728857,
      "learning_rate": 1.5191951020606931e-05,
      "loss": 11.5,
      "step": 949
    },
    {
      "epoch": 0.8246527777777778,
      "grad_norm": 0.0010739682475104928,
      "learning_rate": 1.5046507870203152e-05,
      "loss": 11.5,
      "step": 950
    },
    {
      "epoch": 0.8255208333333334,
      "grad_norm": 0.001114601269364357,
      "learning_rate": 1.4901707627310279e-05,
      "loss": 11.5,
      "step": 951
    },
    {
      "epoch": 0.8263888888888888,
      "grad_norm": 0.0013304173480719328,
      "learning_rate": 1.4757551387741587e-05,
      "loss": 11.5,
      "step": 952
    },
    {
      "epoch": 0.8272569444444444,
      "grad_norm": 0.001954044448211789,
      "learning_rate": 1.4614040242436666e-05,
      "loss": 11.5,
      "step": 953
    },
    {
      "epoch": 0.828125,
      "grad_norm": 0.001330082188360393,
      "learning_rate": 1.4471175277453197e-05,
      "loss": 11.5,
      "step": 954
    },
    {
      "epoch": 0.8289930555555556,
      "grad_norm": 0.0008398760110139847,
      "learning_rate": 1.4328957573958712e-05,
      "loss": 11.5,
      "step": 955
    },
    {
      "epoch": 0.8298611111111112,
      "grad_norm": 0.0007177976076491177,
      "learning_rate": 1.4187388208222474e-05,
      "loss": 11.5,
      "step": 956
    },
    {
      "epoch": 0.8307291666666666,
      "grad_norm": 0.0007141125970520079,
      "learning_rate": 1.404646825160727e-05,
      "loss": 11.5,
      "step": 957
    },
    {
      "epoch": 0.8315972222222222,
      "grad_norm": 0.0010885123629122972,
      "learning_rate": 1.3906198770561263e-05,
      "loss": 11.5,
      "step": 958
    },
    {
      "epoch": 0.8324652777777778,
      "grad_norm": 0.0006587221287190914,
      "learning_rate": 1.376658082661002e-05,
      "loss": 11.5,
      "step": 959
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.0016354231629520655,
      "learning_rate": 1.36276154763484e-05,
      "loss": 11.5,
      "step": 960
    },
    {
      "epoch": 0.8342013888888888,
      "grad_norm": 0.0006749715539626777,
      "learning_rate": 1.3489303771432627e-05,
      "loss": 11.5,
      "step": 961
    },
    {
      "epoch": 0.8350694444444444,
      "grad_norm": 0.001312408596277237,
      "learning_rate": 1.3351646758572301e-05,
      "loss": 11.5,
      "step": 962
    },
    {
      "epoch": 0.8359375,
      "grad_norm": 0.0007768285577185452,
      "learning_rate": 1.3214645479522436e-05,
      "loss": 11.5,
      "step": 963
    },
    {
      "epoch": 0.8368055555555556,
      "grad_norm": 0.0005783441592939198,
      "learning_rate": 1.3078300971075618e-05,
      "loss": 11.5,
      "step": 964
    },
    {
      "epoch": 0.8376736111111112,
      "grad_norm": 0.0007800708408467472,
      "learning_rate": 1.2942614265054143e-05,
      "loss": 11.5,
      "step": 965
    },
    {
      "epoch": 0.8385416666666666,
      "grad_norm": 0.0033471109345555305,
      "learning_rate": 1.280758638830224e-05,
      "loss": 11.5,
      "step": 966
    },
    {
      "epoch": 0.8394097222222222,
      "grad_norm": 0.0010724832536652684,
      "learning_rate": 1.2673218362678297e-05,
      "loss": 11.5,
      "step": 967
    },
    {
      "epoch": 0.8402777777777778,
      "grad_norm": 0.0010352229001000524,
      "learning_rate": 1.253951120504706e-05,
      "loss": 11.5,
      "step": 968
    },
    {
      "epoch": 0.8411458333333334,
      "grad_norm": 0.0012645673705264926,
      "learning_rate": 1.2406465927272004e-05,
      "loss": 11.5,
      "step": 969
    },
    {
      "epoch": 0.8420138888888888,
      "grad_norm": 0.0010738532291725278,
      "learning_rate": 1.2274083536207636e-05,
      "loss": 11.5,
      "step": 970
    },
    {
      "epoch": 0.8428819444444444,
      "grad_norm": 0.002103982027620077,
      "learning_rate": 1.2142365033691949e-05,
      "loss": 11.5,
      "step": 971
    },
    {
      "epoch": 0.84375,
      "grad_norm": 0.0012553971027955413,
      "learning_rate": 1.2011311416538774e-05,
      "loss": 11.5,
      "step": 972
    },
    {
      "epoch": 0.8446180555555556,
      "grad_norm": 0.00196029432117939,
      "learning_rate": 1.1880923676530199e-05,
      "loss": 11.5,
      "step": 973
    },
    {
      "epoch": 0.8454861111111112,
      "grad_norm": 0.004547140095382929,
      "learning_rate": 1.1751202800409178e-05,
      "loss": 11.5,
      "step": 974
    },
    {
      "epoch": 0.8463541666666666,
      "grad_norm": 0.0010319905122742057,
      "learning_rate": 1.1622149769871926e-05,
      "loss": 11.5,
      "step": 975
    },
    {
      "epoch": 0.8472222222222222,
      "grad_norm": 0.0012119009625166655,
      "learning_rate": 1.1493765561560643e-05,
      "loss": 11.5,
      "step": 976
    },
    {
      "epoch": 0.8480902777777778,
      "grad_norm": 0.0012598815374076366,
      "learning_rate": 1.1366051147056033e-05,
      "loss": 11.5,
      "step": 977
    },
    {
      "epoch": 0.8489583333333334,
      "grad_norm": 0.002547974232584238,
      "learning_rate": 1.1239007492869913e-05,
      "loss": 11.5,
      "step": 978
    },
    {
      "epoch": 0.8498263888888888,
      "grad_norm": 0.0011588438646867871,
      "learning_rate": 1.1112635560437979e-05,
      "loss": 11.5,
      "step": 979
    },
    {
      "epoch": 0.8506944444444444,
      "grad_norm": 0.0016873128479346633,
      "learning_rate": 1.0986936306112472e-05,
      "loss": 11.5,
      "step": 980
    },
    {
      "epoch": 0.8515625,
      "grad_norm": 0.0013210588367655873,
      "learning_rate": 1.086191068115503e-05,
      "loss": 11.5,
      "step": 981
    },
    {
      "epoch": 0.8524305555555556,
      "grad_norm": 0.0009892116067931056,
      "learning_rate": 1.0737559631729354e-05,
      "loss": 11.5,
      "step": 982
    },
    {
      "epoch": 0.8532986111111112,
      "grad_norm": 0.0021760337986052036,
      "learning_rate": 1.061388409889419e-05,
      "loss": 11.5,
      "step": 983
    },
    {
      "epoch": 0.8541666666666666,
      "grad_norm": 0.0006816691602580249,
      "learning_rate": 1.0490885018596075e-05,
      "loss": 11.5,
      "step": 984
    },
    {
      "epoch": 0.8550347222222222,
      "grad_norm": 0.001448394381441176,
      "learning_rate": 1.0368563321662317e-05,
      "loss": 11.5,
      "step": 985
    },
    {
      "epoch": 0.8559027777777778,
      "grad_norm": 0.005412398371845484,
      "learning_rate": 1.0246919933794041e-05,
      "loss": 11.5,
      "step": 986
    },
    {
      "epoch": 0.8567708333333334,
      "grad_norm": 0.002708708168938756,
      "learning_rate": 1.012595577555897e-05,
      "loss": 11.5,
      "step": 987
    },
    {
      "epoch": 0.8576388888888888,
      "grad_norm": 0.0023373335134238005,
      "learning_rate": 1.0005671762384671e-05,
      "loss": 11.5,
      "step": 988
    },
    {
      "epoch": 0.8585069444444444,
      "grad_norm": 0.0011215019039809704,
      "learning_rate": 9.886068804551507e-06,
      "loss": 11.5,
      "step": 989
    },
    {
      "epoch": 0.859375,
      "grad_norm": 0.002261243760585785,
      "learning_rate": 9.76714780718574e-06,
      "loss": 11.5,
      "step": 990
    },
    {
      "epoch": 0.8602430555555556,
      "grad_norm": 0.002282239031046629,
      "learning_rate": 9.648909670252804e-06,
      "loss": 11.5,
      "step": 991
    },
    {
      "epoch": 0.8611111111111112,
      "grad_norm": 0.004859830718487501,
      "learning_rate": 9.531355288550325e-06,
      "loss": 11.5,
      "step": 992
    },
    {
      "epoch": 0.8619791666666666,
      "grad_norm": 0.0006296356441453099,
      "learning_rate": 9.414485551701502e-06,
      "loss": 11.5,
      "step": 993
    },
    {
      "epoch": 0.8628472222222222,
      "grad_norm": 0.0013137138448655605,
      "learning_rate": 9.298301344148275e-06,
      "loss": 11.5,
      "step": 994
    },
    {
      "epoch": 0.8637152777777778,
      "grad_norm": 0.0036577116698026657,
      "learning_rate": 9.182803545144624e-06,
      "loss": 11.5,
      "step": 995
    },
    {
      "epoch": 0.8645833333333334,
      "grad_norm": 0.0017342986539006233,
      "learning_rate": 9.067993028750066e-06,
      "loss": 11.5,
      "step": 996
    },
    {
      "epoch": 0.8654513888888888,
      "grad_norm": 0.0026329411193728447,
      "learning_rate": 8.953870663822794e-06,
      "loss": 11.5,
      "step": 997
    },
    {
      "epoch": 0.8663194444444444,
      "grad_norm": 0.002452557673677802,
      "learning_rate": 8.84043731401335e-06,
      "loss": 11.5,
      "step": 998
    },
    {
      "epoch": 0.8671875,
      "grad_norm": 0.004086048807948828,
      "learning_rate": 8.727693837757922e-06,
      "loss": 11.5,
      "step": 999
    },
    {
      "epoch": 0.8680555555555556,
      "grad_norm": 0.0037701039109379053,
      "learning_rate": 8.615641088271875e-06,
      "loss": 11.5,
      "step": 1000
    },
    {
      "epoch": 0.8689236111111112,
      "grad_norm": 0.00042145466431975365,
      "learning_rate": 8.5042799135434e-06,
      "loss": 11.5,
      "step": 1001
    },
    {
      "epoch": 0.8697916666666666,
      "grad_norm": 0.0008856959757395089,
      "learning_rate": 8.39361115632693e-06,
      "loss": 11.5,
      "step": 1002
    },
    {
      "epoch": 0.8706597222222222,
      "grad_norm": 0.0010345204500481486,
      "learning_rate": 8.283635654136945e-06,
      "loss": 11.5,
      "step": 1003
    },
    {
      "epoch": 0.8715277777777778,
      "grad_norm": 0.003807014087215066,
      "learning_rate": 8.174354239241444e-06,
      "loss": 11.5,
      "step": 1004
    },
    {
      "epoch": 0.8723958333333334,
      "grad_norm": 0.0011672809487208724,
      "learning_rate": 8.065767738655772e-06,
      "loss": 11.5,
      "step": 1005
    },
    {
      "epoch": 0.8732638888888888,
      "grad_norm": 0.000871351920068264,
      "learning_rate": 7.95787697413638e-06,
      "loss": 11.5,
      "step": 1006
    },
    {
      "epoch": 0.8741319444444444,
      "grad_norm": 0.001223891624249518,
      "learning_rate": 7.850682762174478e-06,
      "loss": 11.5,
      "step": 1007
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.0018293067114427686,
      "learning_rate": 7.744185913990009e-06,
      "loss": 11.5,
      "step": 1008
    },
    {
      "epoch": 0.8758680555555556,
      "grad_norm": 0.0011166062904521823,
      "learning_rate": 7.638387235525379e-06,
      "loss": 11.5,
      "step": 1009
    },
    {
      "epoch": 0.8767361111111112,
      "grad_norm": 0.0012625007657334208,
      "learning_rate": 7.533287527439425e-06,
      "loss": 11.5,
      "step": 1010
    },
    {
      "epoch": 0.8776041666666666,
      "grad_norm": 0.0016361350426450372,
      "learning_rate": 7.428887585101374e-06,
      "loss": 11.5,
      "step": 1011
    },
    {
      "epoch": 0.8784722222222222,
      "grad_norm": 0.005107779521495104,
      "learning_rate": 7.325188198584765e-06,
      "loss": 11.5,
      "step": 1012
    },
    {
      "epoch": 0.8793402777777778,
      "grad_norm": 0.006148005370050669,
      "learning_rate": 7.222190152661523e-06,
      "loss": 11.5,
      "step": 1013
    },
    {
      "epoch": 0.8802083333333334,
      "grad_norm": 0.0011766005773097277,
      "learning_rate": 7.119894226795976e-06,
      "loss": 11.5,
      "step": 1014
    },
    {
      "epoch": 0.8810763888888888,
      "grad_norm": 0.0047754766419529915,
      "learning_rate": 7.0183011951389655e-06,
      "loss": 11.5,
      "step": 1015
    },
    {
      "epoch": 0.8819444444444444,
      "grad_norm": 0.000929972913581878,
      "learning_rate": 6.917411826522058e-06,
      "loss": 11.5,
      "step": 1016
    },
    {
      "epoch": 0.8828125,
      "grad_norm": 0.002323335502296686,
      "learning_rate": 6.817226884451588e-06,
      "loss": 11.5,
      "step": 1017
    },
    {
      "epoch": 0.8836805555555556,
      "grad_norm": 0.003716230858117342,
      "learning_rate": 6.717747127103036e-06,
      "loss": 11.5,
      "step": 1018
    },
    {
      "epoch": 0.8845486111111112,
      "grad_norm": 0.0029551966581493616,
      "learning_rate": 6.618973307315169e-06,
      "loss": 11.5,
      "step": 1019
    },
    {
      "epoch": 0.8854166666666666,
      "grad_norm": 0.0009839937556535006,
      "learning_rate": 6.520906172584374e-06,
      "loss": 11.5,
      "step": 1020
    },
    {
      "epoch": 0.8862847222222222,
      "grad_norm": 0.0009133880375884473,
      "learning_rate": 6.423546465059083e-06,
      "loss": 11.5,
      "step": 1021
    },
    {
      "epoch": 0.8871527777777778,
      "grad_norm": 0.0013477804604917765,
      "learning_rate": 6.32689492153401e-06,
      "loss": 11.5,
      "step": 1022
    },
    {
      "epoch": 0.8880208333333334,
      "grad_norm": 0.00133781717158854,
      "learning_rate": 6.230952273444724e-06,
      "loss": 11.5,
      "step": 1023
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.0035183983854949474,
      "learning_rate": 6.135719246862015e-06,
      "loss": 11.5,
      "step": 1024
    },
    {
      "epoch": 0.8897569444444444,
      "grad_norm": 0.0021979487501084805,
      "learning_rate": 6.041196562486384e-06,
      "loss": 11.5,
      "step": 1025
    },
    {
      "epoch": 0.890625,
      "grad_norm": 0.0006728313164785504,
      "learning_rate": 5.947384935642741e-06,
      "loss": 11.5,
      "step": 1026
    },
    {
      "epoch": 0.8914930555555556,
      "grad_norm": 0.0014819198986515403,
      "learning_rate": 5.854285076274768e-06,
      "loss": 11.5,
      "step": 1027
    },
    {
      "epoch": 0.8923611111111112,
      "grad_norm": 0.0007111425511538982,
      "learning_rate": 5.761897688939743e-06,
      "loss": 11.5,
      "step": 1028
    },
    {
      "epoch": 0.8932291666666666,
      "grad_norm": 0.0018914854153990746,
      "learning_rate": 5.670223472803105e-06,
      "loss": 11.5,
      "step": 1029
    },
    {
      "epoch": 0.8940972222222222,
      "grad_norm": 0.0006797855603508651,
      "learning_rate": 5.57926312163316e-06,
      "loss": 11.5,
      "step": 1030
    },
    {
      "epoch": 0.8949652777777778,
      "grad_norm": 0.0008072173804976046,
      "learning_rate": 5.489017323795897e-06,
      "loss": 11.5,
      "step": 1031
    },
    {
      "epoch": 0.8958333333333334,
      "grad_norm": 0.001072912011295557,
      "learning_rate": 5.399486762249695e-06,
      "loss": 11.5,
      "step": 1032
    },
    {
      "epoch": 0.8967013888888888,
      "grad_norm": 0.002649150788784027,
      "learning_rate": 5.3106721145402386e-06,
      "loss": 11.5,
      "step": 1033
    },
    {
      "epoch": 0.8975694444444444,
      "grad_norm": 0.0014703402994200587,
      "learning_rate": 5.222574052795337e-06,
      "loss": 11.5,
      "step": 1034
    },
    {
      "epoch": 0.8984375,
      "grad_norm": 0.0012502596946433187,
      "learning_rate": 5.135193243719827e-06,
      "loss": 11.5,
      "step": 1035
    },
    {
      "epoch": 0.8993055555555556,
      "grad_norm": 0.0015454153763130307,
      "learning_rate": 5.048530348590586e-06,
      "loss": 11.5,
      "step": 1036
    },
    {
      "epoch": 0.9001736111111112,
      "grad_norm": 0.0013219431275501847,
      "learning_rate": 4.962586023251448e-06,
      "loss": 11.5,
      "step": 1037
    },
    {
      "epoch": 0.9010416666666666,
      "grad_norm": 0.0006762619013898075,
      "learning_rate": 4.877360918108342e-06,
      "loss": 11.5,
      "step": 1038
    },
    {
      "epoch": 0.9019097222222222,
      "grad_norm": 0.002098310040310025,
      "learning_rate": 4.792855678124275e-06,
      "loss": 11.5,
      "step": 1039
    },
    {
      "epoch": 0.9027777777777778,
      "grad_norm": 0.002462609438225627,
      "learning_rate": 4.709070942814453e-06,
      "loss": 11.5,
      "step": 1040
    },
    {
      "epoch": 0.9036458333333334,
      "grad_norm": 0.0010964088141918182,
      "learning_rate": 4.6260073462415785e-06,
      "loss": 11.5,
      "step": 1041
    },
    {
      "epoch": 0.9045138888888888,
      "grad_norm": 0.0016235937364399433,
      "learning_rate": 4.543665517010831e-06,
      "loss": 11.5,
      "step": 1042
    },
    {
      "epoch": 0.9053819444444444,
      "grad_norm": 0.0007968855788931251,
      "learning_rate": 4.462046078265336e-06,
      "loss": 11.5,
      "step": 1043
    },
    {
      "epoch": 0.90625,
      "grad_norm": 0.0018716808408498764,
      "learning_rate": 4.381149647681282e-06,
      "loss": 11.5,
      "step": 1044
    },
    {
      "epoch": 0.9071180555555556,
      "grad_norm": 0.0017487941076979041,
      "learning_rate": 4.3009768374633e-06,
      "loss": 11.5,
      "step": 1045
    },
    {
      "epoch": 0.9079861111111112,
      "grad_norm": 0.0011104134609922767,
      "learning_rate": 4.221528254339846e-06,
      "loss": 11.5,
      "step": 1046
    },
    {
      "epoch": 0.9088541666666666,
      "grad_norm": 0.0016517093172296882,
      "learning_rate": 4.142804499558639e-06,
      "loss": 11.5,
      "step": 1047
    },
    {
      "epoch": 0.9097222222222222,
      "grad_norm": 0.0013780518202111125,
      "learning_rate": 4.064806168882041e-06,
      "loss": 11.5,
      "step": 1048
    },
    {
      "epoch": 0.9105902777777778,
      "grad_norm": 0.0031211834866553545,
      "learning_rate": 3.98753385258257e-06,
      "loss": 11.5,
      "step": 1049
    },
    {
      "epoch": 0.9114583333333334,
      "grad_norm": 0.0017375049646943808,
      "learning_rate": 3.910988135438442e-06,
      "loss": 11.5,
      "step": 1050
    },
    {
      "epoch": 0.9123263888888888,
      "grad_norm": 0.0013293499359861016,
      "learning_rate": 3.835169596729171e-06,
      "loss": 11.5,
      "step": 1051
    },
    {
      "epoch": 0.9131944444444444,
      "grad_norm": 0.002293319208547473,
      "learning_rate": 3.760078810231138e-06,
      "loss": 11.5,
      "step": 1052
    },
    {
      "epoch": 0.9140625,
      "grad_norm": 0.001359097776003182,
      "learning_rate": 3.685716344213297e-06,
      "loss": 11.5,
      "step": 1053
    },
    {
      "epoch": 0.9149305555555556,
      "grad_norm": 0.0015647734981030226,
      "learning_rate": 3.612082761432822e-06,
      "loss": 11.5,
      "step": 1054
    },
    {
      "epoch": 0.9157986111111112,
      "grad_norm": 0.0014969836920499802,
      "learning_rate": 3.5391786191308894e-06,
      "loss": 11.5,
      "step": 1055
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 0.0006964701460674405,
      "learning_rate": 3.467004469028412e-06,
      "loss": 11.5,
      "step": 1056
    },
    {
      "epoch": 0.9175347222222222,
      "grad_norm": 0.0019835105631500483,
      "learning_rate": 3.3955608573219555e-06,
      "loss": 11.5,
      "step": 1057
    },
    {
      "epoch": 0.9184027777777778,
      "grad_norm": 0.0013506474206224084,
      "learning_rate": 3.3248483246795193e-06,
      "loss": 11.5,
      "step": 1058
    },
    {
      "epoch": 0.9192708333333334,
      "grad_norm": 0.0008903718553483486,
      "learning_rate": 3.2548674062364726e-06,
      "loss": 11.5,
      "step": 1059
    },
    {
      "epoch": 0.9201388888888888,
      "grad_norm": 0.0016759043792262673,
      "learning_rate": 3.1856186315914805e-06,
      "loss": 11.5,
      "step": 1060
    },
    {
      "epoch": 0.9210069444444444,
      "grad_norm": 0.0009094675770029426,
      "learning_rate": 3.117102524802562e-06,
      "loss": 11.5,
      "step": 1061
    },
    {
      "epoch": 0.921875,
      "grad_norm": 0.0004431757843121886,
      "learning_rate": 3.0493196043830606e-06,
      "loss": 11.5,
      "step": 1062
    },
    {
      "epoch": 0.9227430555555556,
      "grad_norm": 0.0015776895452290773,
      "learning_rate": 2.9822703832977694e-06,
      "loss": 11.5,
      "step": 1063
    },
    {
      "epoch": 0.9236111111111112,
      "grad_norm": 0.003403956536203623,
      "learning_rate": 2.915955368958967e-06,
      "loss": 11.5,
      "step": 1064
    },
    {
      "epoch": 0.9244791666666666,
      "grad_norm": 0.0015795637154951692,
      "learning_rate": 2.850375063222699e-06,
      "loss": 11.5,
      "step": 1065
    },
    {
      "epoch": 0.9253472222222222,
      "grad_norm": 0.0011268315138295293,
      "learning_rate": 2.785529962384836e-06,
      "loss": 11.5,
      "step": 1066
    },
    {
      "epoch": 0.9262152777777778,
      "grad_norm": 0.0008931608172133565,
      "learning_rate": 2.7214205571774786e-06,
      "loss": 11.5,
      "step": 1067
    },
    {
      "epoch": 0.9270833333333334,
      "grad_norm": 0.0008492959896102548,
      "learning_rate": 2.658047332765112e-06,
      "loss": 11.5,
      "step": 1068
    },
    {
      "epoch": 0.9279513888888888,
      "grad_norm": 0.0016730859642848372,
      "learning_rate": 2.595410768740991e-06,
      "loss": 11.5,
      "step": 1069
    },
    {
      "epoch": 0.9288194444444444,
      "grad_norm": 0.0011624651961028576,
      "learning_rate": 2.5335113391235064e-06,
      "loss": 11.5,
      "step": 1070
    },
    {
      "epoch": 0.9296875,
      "grad_norm": 0.0026385283563286066,
      "learning_rate": 2.47234951235259e-06,
      "loss": 11.5,
      "step": 1071
    },
    {
      "epoch": 0.9305555555555556,
      "grad_norm": 0.0012204269878566265,
      "learning_rate": 2.411925751286159e-06,
      "loss": 11.5,
      "step": 1072
    },
    {
      "epoch": 0.9314236111111112,
      "grad_norm": 0.0025933668948709965,
      "learning_rate": 2.3522405131966798e-06,
      "loss": 11.5,
      "step": 1073
    },
    {
      "epoch": 0.9322916666666666,
      "grad_norm": 0.0020303402561694384,
      "learning_rate": 2.293294249767597e-06,
      "loss": 11.5,
      "step": 1074
    },
    {
      "epoch": 0.9331597222222222,
      "grad_norm": 0.000987999141216278,
      "learning_rate": 2.2350874070900084e-06,
      "loss": 11.5,
      "step": 1075
    },
    {
      "epoch": 0.9340277777777778,
      "grad_norm": 0.002031516283750534,
      "learning_rate": 2.17762042565921e-06,
      "loss": 11.5,
      "step": 1076
    },
    {
      "epoch": 0.9348958333333334,
      "grad_norm": 0.0014038010267540812,
      "learning_rate": 2.120893740371477e-06,
      "loss": 11.5,
      "step": 1077
    },
    {
      "epoch": 0.9357638888888888,
      "grad_norm": 0.00222398666664958,
      "learning_rate": 2.0649077805206307e-06,
      "loss": 11.5,
      "step": 1078
    },
    {
      "epoch": 0.9366319444444444,
      "grad_norm": 0.002723652170971036,
      "learning_rate": 2.0096629697949233e-06,
      "loss": 11.5,
      "step": 1079
    },
    {
      "epoch": 0.9375,
      "grad_norm": 0.0005467318114824593,
      "learning_rate": 1.9551597262737252e-06,
      "loss": 11.5,
      "step": 1080
    },
    {
      "epoch": 0.9383680555555556,
      "grad_norm": 0.0028293549548834562,
      "learning_rate": 1.9013984624244085e-06,
      "loss": 11.5,
      "step": 1081
    },
    {
      "epoch": 0.9392361111111112,
      "grad_norm": 0.0006175352609716356,
      "learning_rate": 1.848379585099258e-06,
      "loss": 11.5,
      "step": 1082
    },
    {
      "epoch": 0.9401041666666666,
      "grad_norm": 0.0008081760606728494,
      "learning_rate": 1.7961034955323087e-06,
      "loss": 11.5,
      "step": 1083
    },
    {
      "epoch": 0.9409722222222222,
      "grad_norm": 0.0017990474589169025,
      "learning_rate": 1.7445705893364029e-06,
      "loss": 11.5,
      "step": 1084
    },
    {
      "epoch": 0.9418402777777778,
      "grad_norm": 0.0017763323849067092,
      "learning_rate": 1.6937812565001155e-06,
      "loss": 11.5,
      "step": 1085
    },
    {
      "epoch": 0.9427083333333334,
      "grad_norm": 0.0031769229099154472,
      "learning_rate": 1.6437358813848558e-06,
      "loss": 11.5,
      "step": 1086
    },
    {
      "epoch": 0.9435763888888888,
      "grad_norm": 0.0010171409230679274,
      "learning_rate": 1.5944348427219369e-06,
      "loss": 11.5,
      "step": 1087
    },
    {
      "epoch": 0.9444444444444444,
      "grad_norm": 0.0015496568521484733,
      "learning_rate": 1.5458785136096998e-06,
      "loss": 11.5,
      "step": 1088
    },
    {
      "epoch": 0.9453125,
      "grad_norm": 0.006778012495487928,
      "learning_rate": 1.4980672615107382e-06,
      "loss": 11.5,
      "step": 1089
    },
    {
      "epoch": 0.9461805555555556,
      "grad_norm": 0.0028673901688307524,
      "learning_rate": 1.4510014482490452e-06,
      "loss": 11.5,
      "step": 1090
    },
    {
      "epoch": 0.9470486111111112,
      "grad_norm": 0.0033669390249997377,
      "learning_rate": 1.4046814300073263e-06,
      "loss": 11.5,
      "step": 1091
    },
    {
      "epoch": 0.9479166666666666,
      "grad_norm": 0.001685646246187389,
      "learning_rate": 1.3591075573243129e-06,
      "loss": 11.5,
      "step": 1092
    },
    {
      "epoch": 0.9487847222222222,
      "grad_norm": 0.0011007391149178147,
      "learning_rate": 1.3142801750920308e-06,
      "loss": 11.5,
      "step": 1093
    },
    {
      "epoch": 0.9496527777777778,
      "grad_norm": 0.0020542109850794077,
      "learning_rate": 1.2701996225533252e-06,
      "loss": 11.5,
      "step": 1094
    },
    {
      "epoch": 0.9505208333333334,
      "grad_norm": 0.0012236145557835698,
      "learning_rate": 1.22686623329914e-06,
      "loss": 11.5,
      "step": 1095
    },
    {
      "epoch": 0.9513888888888888,
      "grad_norm": 0.0015340264653787017,
      "learning_rate": 1.184280335266108e-06,
      "loss": 11.5,
      "step": 1096
    },
    {
      "epoch": 0.9522569444444444,
      "grad_norm": 0.006743791978806257,
      "learning_rate": 1.1424422507340216e-06,
      "loss": 11.5,
      "step": 1097
    },
    {
      "epoch": 0.953125,
      "grad_norm": 0.005063168704509735,
      "learning_rate": 1.101352296323399e-06,
      "loss": 11.5,
      "step": 1098
    },
    {
      "epoch": 0.9539930555555556,
      "grad_norm": 0.005146476440131664,
      "learning_rate": 1.0610107829930994e-06,
      "loss": 11.5,
      "step": 1099
    },
    {
      "epoch": 0.9548611111111112,
      "grad_norm": 0.0010203375713899732,
      "learning_rate": 1.021418016037945e-06,
      "loss": 11.5,
      "step": 1100
    },
    {
      "epoch": 0.9557291666666666,
      "grad_norm": 0.0012535655405372381,
      "learning_rate": 9.825742950864359e-07,
      "loss": 11.5,
      "step": 1101
    },
    {
      "epoch": 0.9565972222222222,
      "grad_norm": 0.0027846684679389,
      "learning_rate": 9.444799140984617e-07,
      "loss": 11.5,
      "step": 1102
    },
    {
      "epoch": 0.9574652777777778,
      "grad_norm": 0.0016019606264308095,
      "learning_rate": 9.071351613631151e-07,
      "loss": 11.5,
      "step": 1103
    },
    {
      "epoch": 0.9583333333333334,
      "grad_norm": 0.0014748639659956098,
      "learning_rate": 8.70540319496449e-07,
      "loss": 11.5,
      "step": 1104
    },
    {
      "epoch": 0.9592013888888888,
      "grad_norm": 0.0007192360353656113,
      "learning_rate": 8.346956654394e-07,
      "loss": 11.5,
      "step": 1105
    },
    {
      "epoch": 0.9600694444444444,
      "grad_norm": 0.0008517622482031584,
      "learning_rate": 7.996014704556465e-07,
      "loss": 11.5,
      "step": 1106
    },
    {
      "epoch": 0.9609375,
      "grad_norm": 0.0018739933148026466,
      "learning_rate": 7.652580001295761e-07,
      "loss": 11.5,
      "step": 1107
    },
    {
      "epoch": 0.9618055555555556,
      "grad_norm": 0.0007520528743043542,
      "learning_rate": 7.316655143642881e-07,
      "loss": 11.5,
      "step": 1108
    },
    {
      "epoch": 0.9626736111111112,
      "grad_norm": 0.0007577742217108607,
      "learning_rate": 6.988242673796053e-07,
      "loss": 11.5,
      "step": 1109
    },
    {
      "epoch": 0.9635416666666666,
      "grad_norm": 0.0009282460669055581,
      "learning_rate": 6.667345077101428e-07,
      "loss": 11.5,
      "step": 1110
    },
    {
      "epoch": 0.9644097222222222,
      "grad_norm": 0.0008129182970151305,
      "learning_rate": 6.353964782034427e-07,
      "loss": 11.5,
      "step": 1111
    },
    {
      "epoch": 0.9652777777777778,
      "grad_norm": 0.0006060908199287951,
      "learning_rate": 6.048104160181644e-07,
      "loss": 11.5,
      "step": 1112
    },
    {
      "epoch": 0.9661458333333334,
      "grad_norm": 0.0012275254121050239,
      "learning_rate": 5.749765526222306e-07,
      "loss": 11.5,
      "step": 1113
    },
    {
      "epoch": 0.9670138888888888,
      "grad_norm": 0.0010882220230996609,
      "learning_rate": 5.458951137911061e-07,
      "loss": 11.5,
      "step": 1114
    },
    {
      "epoch": 0.9678819444444444,
      "grad_norm": 0.004293525125831366,
      "learning_rate": 5.175663196061109e-07,
      "loss": 11.5,
      "step": 1115
    },
    {
      "epoch": 0.96875,
      "grad_norm": 0.004516840446740389,
      "learning_rate": 4.899903844527098e-07,
      "loss": 11.5,
      "step": 1116
    },
    {
      "epoch": 0.9696180555555556,
      "grad_norm": 0.0040466878563165665,
      "learning_rate": 4.6316751701892534e-07,
      "loss": 11.5,
      "step": 1117
    },
    {
      "epoch": 0.9704861111111112,
      "grad_norm": 0.0007072569569572806,
      "learning_rate": 4.3709792029372756e-07,
      "loss": 11.5,
      "step": 1118
    },
    {
      "epoch": 0.9713541666666666,
      "grad_norm": 0.0008942272397689521,
      "learning_rate": 4.117817915655353e-07,
      "loss": 11.5,
      "step": 1119
    },
    {
      "epoch": 0.9722222222222222,
      "grad_norm": 0.0006828358164057136,
      "learning_rate": 3.8721932242070657e-07,
      "loss": 11.5,
      "step": 1120
    },
    {
      "epoch": 0.9730902777777778,
      "grad_norm": 0.0005529056070372462,
      "learning_rate": 3.6341069874206157e-07,
      "loss": 11.5,
      "step": 1121
    },
    {
      "epoch": 0.9739583333333334,
      "grad_norm": 0.003228049026802182,
      "learning_rate": 3.4035610070751736e-07,
      "loss": 11.5,
      "step": 1122
    },
    {
      "epoch": 0.9748263888888888,
      "grad_norm": 0.0009296713396906853,
      "learning_rate": 3.18055702788711e-07,
      "loss": 11.5,
      "step": 1123
    },
    {
      "epoch": 0.9756944444444444,
      "grad_norm": 0.001211634953506291,
      "learning_rate": 2.9650967374966755e-07,
      "loss": 11.5,
      "step": 1124
    },
    {
      "epoch": 0.9765625,
      "grad_norm": 0.0005990596837364137,
      "learning_rate": 2.7571817664551193e-07,
      "loss": 11.5,
      "step": 1125
    },
    {
      "epoch": 0.9774305555555556,
      "grad_norm": 0.001546860090456903,
      "learning_rate": 2.556813688212922e-07,
      "loss": 11.5,
      "step": 1126
    },
    {
      "epoch": 0.9782986111111112,
      "grad_norm": 0.0030194015707820654,
      "learning_rate": 2.3639940191073628e-07,
      "loss": 11.5,
      "step": 1127
    },
    {
      "epoch": 0.9791666666666666,
      "grad_norm": 0.001958158565685153,
      "learning_rate": 2.1787242183509693e-07,
      "loss": 11.5,
      "step": 1128
    },
    {
      "epoch": 0.9800347222222222,
      "grad_norm": 0.000692248169798404,
      "learning_rate": 2.0010056880208628e-07,
      "loss": 11.5,
      "step": 1129
    },
    {
      "epoch": 0.9809027777777778,
      "grad_norm": 0.0010920714121311903,
      "learning_rate": 1.8308397730479877e-07,
      "loss": 11.5,
      "step": 1130
    },
    {
      "epoch": 0.9817708333333334,
      "grad_norm": 0.0005377872148528695,
      "learning_rate": 1.6682277612070086e-07,
      "loss": 11.5,
      "step": 1131
    },
    {
      "epoch": 0.9826388888888888,
      "grad_norm": 0.0008873736951500177,
      "learning_rate": 1.5131708831062075e-07,
      "loss": 11.5,
      "step": 1132
    },
    {
      "epoch": 0.9835069444444444,
      "grad_norm": 0.0016772827366366982,
      "learning_rate": 1.3656703121784908e-07,
      "loss": 11.5,
      "step": 1133
    },
    {
      "epoch": 0.984375,
      "grad_norm": 0.0009485622867941856,
      "learning_rate": 1.2257271646728407e-07,
      "loss": 11.5,
      "step": 1134
    },
    {
      "epoch": 0.9852430555555556,
      "grad_norm": 0.000926626322325319,
      "learning_rate": 1.0933424996451003e-07,
      "loss": 11.5,
      "step": 1135
    },
    {
      "epoch": 0.9861111111111112,
      "grad_norm": 0.005970866419374943,
      "learning_rate": 9.685173189503128e-08,
      "loss": 11.5,
      "step": 1136
    },
    {
      "epoch": 0.9869791666666666,
      "grad_norm": 0.000818538130261004,
      "learning_rate": 8.512525672353944e-08,
      "loss": 11.5,
      "step": 1137
    },
    {
      "epoch": 0.9878472222222222,
      "grad_norm": 0.0073596760630607605,
      "learning_rate": 7.415491319316959e-08,
      "loss": 11.5,
      "step": 1138
    },
    {
      "epoch": 0.9887152777777778,
      "grad_norm": 0.0026570381596684456,
      "learning_rate": 6.394078432482297e-08,
      "loss": 11.5,
      "step": 1139
    },
    {
      "epoch": 0.9895833333333334,
      "grad_norm": 0.0015695440815761685,
      "learning_rate": 5.4482947416567566e-08,
      "loss": 11.5,
      "step": 1140
    },
    {
      "epoch": 0.9904513888888888,
      "grad_norm": 0.002003516536206007,
      "learning_rate": 4.578147404302735e-08,
      "loss": 11.5,
      "step": 1141
    },
    {
      "epoch": 0.9913194444444444,
      "grad_norm": 0.0015122609911486506,
      "learning_rate": 3.783643005484949e-08,
      "loss": 11.5,
      "step": 1142
    },
    {
      "epoch": 0.9921875,
      "grad_norm": 0.0016990628791972995,
      "learning_rate": 3.06478755782047e-08,
      "loss": 11.5,
      "step": 1143
    },
    {
      "epoch": 0.9930555555555556,
      "grad_norm": 0.0011751552810892463,
      "learning_rate": 2.4215865014354244e-08,
      "loss": 11.5,
      "step": 1144
    },
    {
      "epoch": 0.9939236111111112,
      "grad_norm": 0.0014843663666397333,
      "learning_rate": 1.854044703917257e-08,
      "loss": 11.5,
      "step": 1145
    },
    {
      "epoch": 0.9947916666666666,
      "grad_norm": 0.004292446654289961,
      "learning_rate": 1.3621664602880835e-08,
      "loss": 11.5,
      "step": 1146
    },
    {
      "epoch": 0.9956597222222222,
      "grad_norm": 0.0014173079980537295,
      "learning_rate": 9.459554929613923e-09,
      "loss": 11.5,
      "step": 1147
    },
    {
      "epoch": 0.9965277777777778,
      "grad_norm": 0.0027792402543127537,
      "learning_rate": 6.054149517220609e-09,
      "loss": 11.5,
      "step": 1148
    },
    {
      "epoch": 0.9973958333333334,
      "grad_norm": 0.0016082413494586945,
      "learning_rate": 3.4054741369859975e-09,
      "loss": 11.5,
      "step": 1149
    },
    {
      "epoch": 0.9982638888888888,
      "grad_norm": 0.003289499320089817,
      "learning_rate": 1.5135488334538928e-09,
      "loss": 11.5,
      "step": 1150
    },
    {
      "epoch": 0.9991319444444444,
      "grad_norm": 0.0012583371717482805,
      "learning_rate": 3.7838792424915723e-10,
      "loss": 11.5,
      "step": 1151
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.006324502173811197,
      "learning_rate": 0.0,
      "loss": 11.5,
      "step": 1152
    },
    {
      "epoch": 1.0,
      "eval_loss": 11.5,
      "eval_runtime": 3.9316,
      "eval_samples_per_second": 123.36,
      "eval_steps_per_second": 61.807,
      "step": 1152
    }
  ],
  "logging_steps": 1,
  "max_steps": 1152,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 288,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 47222202679296.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}