diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -11,5483 +11,5483 @@
   "log_history": [
     {
       "epoch": 0.00064,
-      "grad_norm": 2.353492879681207,
+      "grad_norm": 2.353485086238037,
       "learning_rate": 0.0,
       "loss": 0.5764,
       "step": 1
     },
     {
       "epoch": 0.00128,
-      "grad_norm": 2.567574488545934,
+      "grad_norm": 2.5676819953400525,
       "learning_rate": 6.329113924050633e-07,
       "loss": 0.6154,
       "step": 2
     },
     {
       "epoch": 0.00192,
-      "grad_norm": 2.452013107124523,
+      "grad_norm": 2.4406327433603328,
       "learning_rate": 1.2658227848101265e-06,
-      "loss": 0.5763,
+      "loss": 0.5765,
       "step": 3
     },
     {
       "epoch": 0.00256,
-      "grad_norm": 2.6431662747915237,
+      "grad_norm": 2.503604928634966,
       "learning_rate": 1.8987341772151901e-06,
-      "loss": 0.5897,
+      "loss": 0.5896,
       "step": 4
     },
     {
       "epoch": 0.0032,
-      "grad_norm": 2.3930737468544336,
+      "grad_norm": 2.3561613670472124,
       "learning_rate": 2.531645569620253e-06,
-      "loss": 0.5846,
+      "loss": 0.5848,
       "step": 5
     },
     {
       "epoch": 0.00384,
-      "grad_norm": 2.594961068226395,
+      "grad_norm": 2.487557281842397,
       "learning_rate": 3.1645569620253167e-06,
-      "loss": 0.6,
+      "loss": 0.6005,
       "step": 6
     },
     {
       "epoch": 0.00448,
-      "grad_norm": 1.8595975894553154,
+      "grad_norm": 1.8603603695583997,
       "learning_rate": 3.7974683544303802e-06,
-      "loss": 0.5579,
+      "loss": 0.558,
       "step": 7
     },
     {
       "epoch": 0.00512,
-      "grad_norm": 1.9840836213880872,
+      "grad_norm": 2.0016494810134997,
       "learning_rate": 4.430379746835443e-06,
-      "loss": 0.5861,
+      "loss": 0.586,
       "step": 8
     },
     {
       "epoch": 0.00576,
-      "grad_norm": 1.3806056222455356,
+      "grad_norm": 1.3866160295372776,
       "learning_rate": 5.063291139240506e-06,
-      "loss": 0.5483,
+      "loss": 0.5485,
       "step": 9
     },
     {
       "epoch": 0.0064,
-      "grad_norm": 1.2378254593133642,
+      "grad_norm": 1.2536265188961684,
       "learning_rate": 5.69620253164557e-06,
-      "loss": 0.5306,
+      "loss": 0.5304,
       "step": 10
     },
     {
       "epoch": 0.00704,
-      "grad_norm": 1.1874568518218973,
+      "grad_norm": 1.1405181059205818,
       "learning_rate": 6.329113924050633e-06,
-      "loss": 0.5569,
+      "loss": 0.5573,
       "step": 11
     },
     {
       "epoch": 0.00768,
-      "grad_norm": 1.4792517105132126,
+      "grad_norm": 1.4739138986648344,
       "learning_rate": 6.9620253164556965e-06,
-      "loss": 0.511,
+      "loss": 0.5111,
       "step": 12
     },
     {
       "epoch": 0.00832,
-      "grad_norm": 1.5972888909878988,
+      "grad_norm": 1.5987665650407654,
       "learning_rate": 7.5949367088607605e-06,
-      "loss": 0.4674,
+      "loss": 0.4672,
       "step": 13
     },
     {
       "epoch": 0.00896,
-      "grad_norm": 1.5705765751301897,
+      "grad_norm": 1.5801662763772257,
       "learning_rate": 8.227848101265822e-06,
-      "loss": 0.4798,
+      "loss": 0.4803,
       "step": 14
     },
     {
       "epoch": 0.0096,
-      "grad_norm": 1.2132734057876817,
+      "grad_norm": 1.1595227910276567,
       "learning_rate": 8.860759493670886e-06,
-      "loss": 0.4411,
+      "loss": 0.4413,
       "step": 15
     },
     {
       "epoch": 0.01024,
-      "grad_norm": 1.0220542665261685,
+      "grad_norm": 1.0232799236259744,
       "learning_rate": 9.49367088607595e-06,
-      "loss": 0.4258,
+      "loss": 0.426,
       "step": 16
     },
     {
       "epoch": 0.01088,
-      "grad_norm": 1.3769301736898738,
+      "grad_norm": 1.4200883806078057,
       "learning_rate": 1.0126582278481012e-05,
-      "loss": 0.4543,
+      "loss": 0.4544,
       "step": 17
     },
     {
       "epoch": 0.01152,
-      "grad_norm": 1.2058856706461065,
+      "grad_norm": 1.2500413307829548,
       "learning_rate": 1.0759493670886076e-05,
-      "loss": 0.4526,
+      "loss": 0.453,
       "step": 18
     },
     {
       "epoch": 0.01216,
-      "grad_norm": 0.9279758572161336,
+      "grad_norm": 0.9645673541147013,
       "learning_rate": 1.139240506329114e-05,
-      "loss": 0.4146,
+      "loss": 0.4149,
       "step": 19
     },
     {
       "epoch": 0.0128,
-      "grad_norm": 0.7302969117448893,
+      "grad_norm": 0.7619491956637449,
       "learning_rate": 1.2025316455696203e-05,
-      "loss": 0.4266,
+      "loss": 0.4268,
       "step": 20
     },
     {
       "epoch": 0.01344,
-      "grad_norm": 0.7558992094228546,
+      "grad_norm": 0.7950927664975571,
       "learning_rate": 1.2658227848101267e-05,
       "loss": 0.4315,
       "step": 21
     },
     {
       "epoch": 0.01408,
-      "grad_norm": 0.6498887321885588,
+      "grad_norm": 0.9130569306644377,
       "learning_rate": 1.3291139240506329e-05,
-      "loss": 0.3667,
+      "loss": 0.367,
       "step": 22
     },
     {
       "epoch": 0.01472,
-      "grad_norm": 0.7438110419135116,
+      "grad_norm": 0.6877425501125356,
       "learning_rate": 1.3924050632911393e-05,
-      "loss": 0.3827,
+      "loss": 0.3838,
       "step": 23
     },
     {
       "epoch": 0.01536,
-      "grad_norm": 0.6626828380488425,
+      "grad_norm": 0.678423749873734,
       "learning_rate": 1.4556962025316457e-05,
-      "loss": 0.3536,
+      "loss": 0.3542,
       "step": 24
     },
     {
       "epoch": 0.016,
-      "grad_norm": 0.7920276455155202,
+      "grad_norm": 0.8086215595785854,
       "learning_rate": 1.5189873417721521e-05,
-      "loss": 0.3942,
+      "loss": 0.3947,
       "step": 25
     },
     {
       "epoch": 0.01664,
-      "grad_norm": 0.6737828955016701,
+      "grad_norm": 0.7365716120703668,
       "learning_rate": 1.5822784810126583e-05,
-      "loss": 0.3749,
+      "loss": 0.3761,
       "step": 26
     },
     {
       "epoch": 0.01728,
-      "grad_norm": 0.5503753217412305,
+      "grad_norm": 0.57410441755097,
       "learning_rate": 1.6455696202531644e-05,
-      "loss": 0.3718,
+      "loss": 0.3722,
       "step": 27
     },
     {
       "epoch": 0.01792,
-      "grad_norm": 0.7894180890361427,
+      "grad_norm": 0.5067303990784363,
       "learning_rate": 1.7088607594936708e-05,
-      "loss": 0.3703,
+      "loss": 0.371,
       "step": 28
     },
     {
       "epoch": 0.01856,
-      "grad_norm": 0.52887713912012,
+      "grad_norm": 0.5387436191633763,
       "learning_rate": 1.7721518987341772e-05,
-      "loss": 0.3576,
+      "loss": 0.3577,
       "step": 29
     },
     {
       "epoch": 0.0192,
-      "grad_norm": 0.5935309461197494,
+      "grad_norm": 0.6059447121681072,
       "learning_rate": 1.8354430379746836e-05,
-      "loss": 0.384,
+      "loss": 0.3836,
       "step": 30
     },
     {
       "epoch": 0.01984,
-      "grad_norm": 0.4947385860590809,
+      "grad_norm": 0.5055998458711023,
       "learning_rate": 1.89873417721519e-05,
-      "loss": 0.3778,
+      "loss": 0.3783,
       "step": 31
     },
     {
       "epoch": 0.02048,
-      "grad_norm": 0.5605217572196567,
+      "grad_norm": 0.5708720902953378,
       "learning_rate": 1.962025316455696e-05,
-      "loss": 0.3838,
+      "loss": 0.3846,
       "step": 32
     },
     {
       "epoch": 0.02112,
-      "grad_norm": 0.6836657396769625,
+      "grad_norm": 0.557929165518282,
       "learning_rate": 2.0253164556962025e-05,
-      "loss": 0.3482,
+      "loss": 0.3484,
       "step": 33
     },
     {
       "epoch": 0.02176,
-      "grad_norm": 0.43798176485984003,
+      "grad_norm": 0.4519934652780253,
       "learning_rate": 2.088607594936709e-05,
-      "loss": 0.3612,
+      "loss": 0.3615,
       "step": 34
     },
     {
       "epoch": 0.0224,
-      "grad_norm": 0.4240945316235695,
+      "grad_norm": 0.43727165616564423,
       "learning_rate": 2.1518987341772153e-05,
       "loss": 0.3599,
       "step": 35
     },
     {
       "epoch": 0.02304,
-      "grad_norm": 0.4290568610982825,
+      "grad_norm": 0.44623181877706214,
       "learning_rate": 2.2151898734177217e-05,
       "loss": 0.3722,
       "step": 36
     },
     {
       "epoch": 0.02368,
-      "grad_norm": 0.46533348793486895,
+      "grad_norm": 0.4635506547428751,
       "learning_rate": 2.278481012658228e-05,
-      "loss": 0.3627,
+      "loss": 0.3628,
       "step": 37
     },
     {
       "epoch": 0.02432,
-      "grad_norm": 0.48639931562775274,
+      "grad_norm": 0.4583180814360715,
       "learning_rate": 2.341772151898734e-05,
-      "loss": 0.3582,
+      "loss": 0.3584,
       "step": 38
     },
     {
       "epoch": 0.02496,
-      "grad_norm": 0.4403975390170576,
+      "grad_norm": 0.40794109495076275,
       "learning_rate": 2.4050632911392405e-05,
-      "loss": 0.353,
+      "loss": 0.3528,
       "step": 39
     },
     {
       "epoch": 0.0256,
-      "grad_norm": 0.36233863650138515,
+      "grad_norm": 0.3709932525452395,
       "learning_rate": 2.468354430379747e-05,
-      "loss": 0.3451,
+      "loss": 0.345,
       "step": 40
     },
     {
       "epoch": 0.02624,
-      "grad_norm": 0.4175915746669695,
+      "grad_norm": 0.4210238800775235,
       "learning_rate": 2.5316455696202533e-05,
-      "loss": 0.3059,
+      "loss": 0.3056,
       "step": 41
     },
     {
       "epoch": 0.02688,
-      "grad_norm": 0.4694739579410676,
+      "grad_norm": 0.4786731092715819,
       "learning_rate": 2.5949367088607597e-05,
-      "loss": 0.3332,
+      "loss": 0.3329,
       "step": 42
     },
     {
       "epoch": 0.02752,
-      "grad_norm": 0.3958470601870478,
+      "grad_norm": 0.38803905635556746,
       "learning_rate": 2.6582278481012658e-05,
-      "loss": 0.3231,
+      "loss": 0.3227,
       "step": 43
     },
     {
       "epoch": 0.02816,
-      "grad_norm": 0.43940504584870443,
+      "grad_norm": 0.43575361580226557,
       "learning_rate": 2.7215189873417722e-05,
-      "loss": 0.3403,
+      "loss": 0.3398,
       "step": 44
     },
     {
       "epoch": 0.0288,
-      "grad_norm": 0.45387568054253924,
+      "grad_norm": 0.452207614606097,
       "learning_rate": 2.7848101265822786e-05,
-      "loss": 0.3522,
+      "loss": 0.3512,
       "step": 45
     },
     {
       "epoch": 0.02944,
-      "grad_norm": 0.47587749147936115,
+      "grad_norm": 0.4698411017252047,
       "learning_rate": 2.848101265822785e-05,
-      "loss": 0.344,
+      "loss": 0.3435,
       "step": 46
     },
     {
       "epoch": 0.03008,
-      "grad_norm": 0.39610819412760284,
+      "grad_norm": 0.40870853690109027,
       "learning_rate": 2.9113924050632914e-05,
-      "loss": 0.337,
+      "loss": 0.3367,
       "step": 47
     },
     {
       "epoch": 0.03072,
-      "grad_norm": 0.4363309227849157,
+      "grad_norm": 0.4405164959813624,
       "learning_rate": 2.9746835443037974e-05,
-      "loss": 0.3483,
+      "loss": 0.3476,
       "step": 48
     },
     {
       "epoch": 0.03136,
-      "grad_norm": 0.40862381905697387,
+      "grad_norm": 0.4259862667341005,
       "learning_rate": 3.0379746835443042e-05,
       "loss": 0.3305,
       "step": 49
     },
     {
       "epoch": 0.032,
-      "grad_norm": 0.39589715535465136,
+      "grad_norm": 0.39771125914492406,
       "learning_rate": 3.10126582278481e-05,
-      "loss": 0.3398,
+      "loss": 0.3395,
       "step": 50
     },
     {
       "epoch": 0.03264,
-      "grad_norm": 0.3989012389437649,
+      "grad_norm": 0.4111747463996418,
       "learning_rate": 3.1645569620253167e-05,
-      "loss": 0.3273,
+      "loss": 0.3274,
       "step": 51
     },
     {
       "epoch": 0.03328,
-      "grad_norm": 0.42021137918245616,
+      "grad_norm": 0.42586319667255845,
       "learning_rate": 3.227848101265823e-05,
-      "loss": 0.3267,
+      "loss": 0.3258,
       "step": 52
     },
     {
       "epoch": 0.03392,
-      "grad_norm": 0.4047725101285825,
+      "grad_norm": 0.40788830009535015,
       "learning_rate": 3.291139240506329e-05,
-      "loss": 0.2908,
+      "loss": 0.2907,
       "step": 53
     },
     {
       "epoch": 0.03456,
-      "grad_norm": 0.4172826076889594,
+      "grad_norm": 0.43296600863158063,
       "learning_rate": 3.354430379746836e-05,
-      "loss": 0.3348,
+      "loss": 0.3345,
       "step": 54
     },
     {
       "epoch": 0.0352,
-      "grad_norm": 0.42253047907814617,
+      "grad_norm": 0.42372120204803543,
       "learning_rate": 3.4177215189873416e-05,
-      "loss": 0.3292,
+      "loss": 0.3293,
       "step": 55
     },
     {
       "epoch": 0.03584,
-      "grad_norm": 0.42758322444932984,
+      "grad_norm": 0.4469884725915102,
       "learning_rate": 3.4810126582278487e-05,
-      "loss": 0.3289,
+      "loss": 0.3287,
       "step": 56
     },
     {
       "epoch": 0.03648,
-      "grad_norm": 0.4053738271148249,
+      "grad_norm": 0.41612674822365947,
       "learning_rate": 3.5443037974683544e-05,
-      "loss": 0.3276,
+      "loss": 0.3279,
       "step": 57
     },
     {
       "epoch": 0.03712,
-      "grad_norm": 0.43996040806002545,
+      "grad_norm": 0.45327242403181206,
       "learning_rate": 3.607594936708861e-05,
-      "loss": 0.3467,
+      "loss": 0.3468,
       "step": 58
     },
     {
       "epoch": 0.03776,
-      "grad_norm": 0.4254049766589256,
+      "grad_norm": 0.44353765655450733,
       "learning_rate": 3.670886075949367e-05,
-      "loss": 0.3338,
+      "loss": 0.3341,
       "step": 59
     },
     {
       "epoch": 0.0384,
-      "grad_norm": 0.43855409848587795,
+      "grad_norm": 0.44486424984007067,
       "learning_rate": 3.7341772151898736e-05,
-      "loss": 0.3274,
+      "loss": 0.3275,
       "step": 60
     },
     {
       "epoch": 0.03904,
-      "grad_norm": 0.4365281497934103,
+      "grad_norm": 0.4383809063442672,
       "learning_rate": 3.79746835443038e-05,
-      "loss": 0.3347,
+      "loss": 0.3342,
       "step": 61
     },
     {
       "epoch": 0.03968,
-      "grad_norm": 0.40792773541529964,
+      "grad_norm": 0.42608132493729306,
       "learning_rate": 3.8607594936708864e-05,
-      "loss": 0.3153,
+      "loss": 0.3155,
       "step": 62
     },
     {
       "epoch": 0.04032,
-      "grad_norm": 0.4494242243007552,
+      "grad_norm": 0.4555942982493556,
       "learning_rate": 3.924050632911392e-05,
-      "loss": 0.3203,
+      "loss": 0.3205,
       "step": 63
     },
     {
       "epoch": 0.04096,
-      "grad_norm": 0.4788676847095667,
+      "grad_norm": 0.494264400213869,
       "learning_rate": 3.987341772151899e-05,
-      "loss": 0.3539,
+      "loss": 0.3542,
       "step": 64
     },
     {
       "epoch": 0.0416,
-      "grad_norm": 0.40546778912791753,
+      "grad_norm": 0.3924518429856125,
       "learning_rate": 4.050632911392405e-05,
-      "loss": 0.3119,
+      "loss": 0.3116,
       "step": 65
     },
     {
       "epoch": 0.04224,
-      "grad_norm": 0.4700613066573691,
+      "grad_norm": 0.4658349233271844,
       "learning_rate": 4.113924050632912e-05,
-      "loss": 0.3464,
+      "loss": 0.3461,
       "step": 66
     },
     {
       "epoch": 0.04288,
-      "grad_norm": 0.4185455474650584,
+      "grad_norm": 0.4151003836242027,
       "learning_rate": 4.177215189873418e-05,
-      "loss": 0.3148,
+      "loss": 0.3152,
       "step": 67
     },
     {
       "epoch": 0.04352,
-      "grad_norm": 0.42213917822060293,
+      "grad_norm": 0.4328542434853278,
       "learning_rate": 4.240506329113924e-05,
-      "loss": 0.3264,
+      "loss": 0.3267,
       "step": 68
     },
     {
       "epoch": 0.04416,
-      "grad_norm": 0.4110186066327522,
+      "grad_norm": 0.4255328771601494,
       "learning_rate": 4.3037974683544305e-05,
-      "loss": 0.2942,
+      "loss": 0.2943,
       "step": 69
     },
     {
       "epoch": 0.0448,
-      "grad_norm": 0.3969987305256835,
+      "grad_norm": 0.3960781879337196,
       "learning_rate": 4.367088607594937e-05,
-      "loss": 0.3296,
+      "loss": 0.3295,
       "step": 70
     },
     {
       "epoch": 0.04544,
-      "grad_norm": 0.458340252055146,
+      "grad_norm": 0.42237083664040964,
       "learning_rate": 4.430379746835443e-05,
-      "loss": 0.3153,
+      "loss": 0.3156,
       "step": 71
     },
     {
       "epoch": 0.04608,
-      "grad_norm": 0.4460994424040082,
+      "grad_norm": 0.44252752029376236,
       "learning_rate": 4.49367088607595e-05,
-      "loss": 0.3237,
+      "loss": 0.3238,
       "step": 72
     },
     {
       "epoch": 0.04672,
-      "grad_norm": 0.48880157137172375,
+      "grad_norm": 0.4254192779309176,
       "learning_rate": 4.556962025316456e-05,
-      "loss": 0.3139,
+      "loss": 0.3137,
       "step": 73
     },
     {
       "epoch": 0.04736,
-      "grad_norm": 0.4187304633592487,
+      "grad_norm": 0.4068734829882739,
       "learning_rate": 4.6202531645569625e-05,
-      "loss": 0.308,
+      "loss": 0.3082,
       "step": 74
     },
     {
       "epoch": 0.048,
-      "grad_norm": 0.3977527819047683,
+      "grad_norm": 0.38540339351470876,
       "learning_rate": 4.683544303797468e-05,
-      "loss": 0.2984,
+      "loss": 0.298,
       "step": 75
     },
     {
       "epoch": 0.04864,
-      "grad_norm": 0.44371228034060706,
+      "grad_norm": 0.43228604399379483,
       "learning_rate": 4.7468354430379746e-05,
-      "loss": 0.3201,
+      "loss": 0.3202,
       "step": 76
     },
     {
       "epoch": 0.04928,
-      "grad_norm": 0.5076283851103198,
+      "grad_norm": 0.4662811772881026,
       "learning_rate": 4.810126582278481e-05,
       "loss": 0.3267,
       "step": 77
     },
     {
       "epoch": 0.04992,
-      "grad_norm": 0.3927986005300883,
+      "grad_norm": 0.4101396625445811,
       "learning_rate": 4.8734177215189874e-05,
-      "loss": 0.2903,
+      "loss": 0.2908,
       "step": 78
     },
     {
       "epoch": 0.05056,
-      "grad_norm": 0.46176244675521244,
+      "grad_norm": 0.4407899025941757,
       "learning_rate": 4.936708860759494e-05,
-      "loss": 0.3235,
+      "loss": 0.3234,
       "step": 79
     },
     {
       "epoch": 0.0512,
-      "grad_norm": 0.421631930666922,
+      "grad_norm": 0.42766797963707764,
       "learning_rate": 5e-05,
-      "loss": 0.3148,
+      "loss": 0.315,
       "step": 80
     },
     {
       "epoch": 0.05184,
-      "grad_norm": 0.4061280972034586,
+      "grad_norm": 0.39551732993824784,
       "learning_rate": 4.999994390459423e-05,
-      "loss": 0.3244,
+      "loss": 0.3241,
       "step": 81
     },
     {
       "epoch": 0.05248,
-      "grad_norm": 0.41908336407872915,
+      "grad_norm": 0.37307344185693914,
       "learning_rate": 4.999977561862863e-05,
-      "loss": 0.3172,
+      "loss": 0.3171,
       "step": 82
     },
     {
       "epoch": 0.05312,
-      "grad_norm": 0.3871831774034262,
+      "grad_norm": 0.47799834282771486,
       "learning_rate": 4.999949514285842e-05,
-      "loss": 0.3176,
+      "loss": 0.3174,
       "step": 83
     },
     {
       "epoch": 0.05376,
-      "grad_norm": 0.4569094416746116,
+      "grad_norm": 0.45391420249819664,
       "learning_rate": 4.999910247854227e-05,
-      "loss": 0.3302,
+      "loss": 0.3301,
       "step": 84
     },
     {
       "epoch": 0.0544,
-      "grad_norm": 0.47566240228622386,
+      "grad_norm": 0.49271462048539344,
       "learning_rate": 4.99985976274423e-05,
-      "loss": 0.3266,
+      "loss": 0.3271,
       "step": 85
     },
     {
       "epoch": 0.05504,
-      "grad_norm": 0.47728079070137763,
+      "grad_norm": 0.505091627450717,
       "learning_rate": 4.9997980591824126e-05,
-      "loss": 0.3072,
+      "loss": 0.308,
       "step": 86
     },
     {
       "epoch": 0.05568,
-      "grad_norm": 0.39221570157146257,
+      "grad_norm": 0.3949565660235225,
       "learning_rate": 4.999725137445674e-05,
       "loss": 0.3185,
       "step": 87
     },
     {
       "epoch": 0.05632,
-      "grad_norm": 0.5220845323376405,
+      "grad_norm": 0.5337665218445341,
       "learning_rate": 4.999640997861264e-05,
-      "loss": 0.3206,
+      "loss": 0.3211,
       "step": 88
     },
     {
       "epoch": 0.05696,
-      "grad_norm": 0.45138937219188247,
+      "grad_norm": 0.45754751119526205,
       "learning_rate": 4.999545640806766e-05,
-      "loss": 0.322,
+      "loss": 0.3228,
       "step": 89
     },
     {
       "epoch": 0.0576,
-      "grad_norm": 0.43231283770656975,
+      "grad_norm": 0.45385353033639203,
       "learning_rate": 4.999439066710111e-05,
-      "loss": 0.3413,
+      "loss": 0.3415,
       "step": 90
     },
     {
       "epoch": 0.05824,
-      "grad_norm": 0.5003943169755616,
+      "grad_norm": 0.49073962043136393,
       "learning_rate": 4.999321276049563e-05,
-      "loss": 0.3182,
+      "loss": 0.3179,
       "step": 91
     },
     {
       "epoch": 0.05888,
-      "grad_norm": 0.37424460362649736,
+      "grad_norm": 0.42999497815595844,
       "learning_rate": 4.9991922693537226e-05,
-      "loss": 0.3065,
+      "loss": 0.3067,
       "step": 92
     },
     {
       "epoch": 0.05952,
-      "grad_norm": 0.44875045802288094,
+      "grad_norm": 0.4625502305860712,
       "learning_rate": 4.9990520472015246e-05,
-      "loss": 0.2941,
+      "loss": 0.2949,
       "step": 93
     },
     {
       "epoch": 0.06016,
-      "grad_norm": 0.4021940194343882,
+      "grad_norm": 0.40226854319567656,
       "learning_rate": 4.998900610222236e-05,
-      "loss": 0.3047,
+      "loss": 0.3051,
       "step": 94
     },
     {
       "epoch": 0.0608,
-      "grad_norm": 0.3881453179250032,
+      "grad_norm": 0.39982821508119915,
       "learning_rate": 4.998737959095449e-05,
-      "loss": 0.2772,
+      "loss": 0.2778,
       "step": 95
     },
     {
       "epoch": 0.06144,
-      "grad_norm": 0.44699341616410987,
+      "grad_norm": 1.0464838609726888,
       "learning_rate": 4.998564094551081e-05,
-      "loss": 0.3147,
+      "loss": 0.3161,
       "step": 96
     },
     {
       "epoch": 0.06208,
-      "grad_norm": 0.43480749835897514,
+      "grad_norm": 0.4514203583299103,
       "learning_rate": 4.998379017369375e-05,
-      "loss": 0.3106,
+      "loss": 0.3119,
       "step": 97
     },
     {
       "epoch": 0.06272,
-      "grad_norm": 0.46102101762706826,
+      "grad_norm": 0.48265032091647747,
       "learning_rate": 4.9981827283808855e-05,
-      "loss": 0.3075,
+      "loss": 0.3083,
       "step": 98
     },
     {
       "epoch": 0.06336,
-      "grad_norm": 0.43677419770737613,
+      "grad_norm": 0.42207276430371016,
       "learning_rate": 4.997975228466489e-05,
-      "loss": 0.3346,
+      "loss": 0.335,
       "step": 99
     },
     {
       "epoch": 0.064,
-      "grad_norm": 0.3990810545801376,
+      "grad_norm": 0.4109136525662046,
       "learning_rate": 4.997756518557368e-05,
-      "loss": 0.2792,
+      "loss": 0.2801,
       "step": 100
     },
     {
       "epoch": 0.06464,
-      "grad_norm": 0.37423119672037614,
+      "grad_norm": 0.40459874238442717,
       "learning_rate": 4.997526599635011e-05,
-      "loss": 0.2843,
+      "loss": 0.2852,
       "step": 101
     },
     {
       "epoch": 0.06528,
-      "grad_norm": 0.4832295984241697,
+      "grad_norm": 0.48908720050476456,
       "learning_rate": 4.997285472731211e-05,
-      "loss": 0.3335,
+      "loss": 0.335,
       "step": 102
     },
     {
       "epoch": 0.06592,
-      "grad_norm": 0.42554025051835676,
+      "grad_norm": 0.42083268237680455,
       "learning_rate": 4.997033138928056e-05,
-      "loss": 0.3134,
+      "loss": 0.3143,
       "step": 103
     },
     {
       "epoch": 0.06656,
-      "grad_norm": 0.4527604445470986,
+      "grad_norm": 0.45257051558741634,
       "learning_rate": 4.996769599357927e-05,
-      "loss": 0.3082,
+      "loss": 0.309,
       "step": 104
     },
     {
       "epoch": 0.0672,
-      "grad_norm": 0.43328941260689235,
+      "grad_norm": 0.43169779281959547,
       "learning_rate": 4.996494855203493e-05,
-      "loss": 0.3233,
+      "loss": 0.3239,
       "step": 105
     },
     {
       "epoch": 0.06784,
-      "grad_norm": 0.5549824145577879,
+      "grad_norm": 0.5600681218655813,
       "learning_rate": 4.996208907697706e-05,
-      "loss": 0.3203,
+      "loss": 0.3218,
       "step": 106
     },
     {
       "epoch": 0.06848,
-      "grad_norm": 0.5226904312118057,
+      "grad_norm": 0.5148554799775189,
       "learning_rate": 4.995911758123791e-05,
-      "loss": 0.2832,
+      "loss": 0.2851,
       "step": 107
     },
     {
       "epoch": 0.06912,
-      "grad_norm": 0.4619131188070299,
+      "grad_norm": 1.6520154087150252,
       "learning_rate": 4.995603407815249e-05,
-      "loss": 0.2958,
+      "loss": 0.2973,
       "step": 108
     },
     {
       "epoch": 0.06976,
-      "grad_norm": 0.4694617073433174,
+      "grad_norm": 0.560865604631309,
       "learning_rate": 4.9952838581558406e-05,
-      "loss": 0.3022,
+      "loss": 0.3056,
       "step": 109
     },
     {
       "epoch": 0.0704,
-      "grad_norm": 0.4111272867245987,
+      "grad_norm": 0.45796131093813686,
       "learning_rate": 4.994953110579588e-05,
-      "loss": 0.3123,
+      "loss": 0.3151,
       "step": 110
     },
     {
       "epoch": 0.07104,
-      "grad_norm": 0.49830144949207156,
+      "grad_norm": 0.9637610478065832,
       "learning_rate": 4.9946111665707645e-05,
-      "loss": 0.3252,
+      "loss": 0.3279,
       "step": 111
     },
     {
       "epoch": 0.07168,
-      "grad_norm": 0.36298788411716454,
+      "grad_norm": 0.364055859969717,
       "learning_rate": 4.99425802766389e-05,
-      "loss": 0.2886,
+      "loss": 0.2899,
       "step": 112
     },
     {
       "epoch": 0.07232,
-      "grad_norm": 0.4553410362166831,
+      "grad_norm": 0.4976556238289915,
       "learning_rate": 4.993893695443721e-05,
-      "loss": 0.3014,
+      "loss": 0.3031,
       "step": 113
     },
     {
       "epoch": 0.07296,
-      "grad_norm": 0.3942254183560885,
+      "grad_norm": 0.43623766283866644,
       "learning_rate": 4.993518171545247e-05,
-      "loss": 0.2905,
+      "loss": 0.2915,
       "step": 114
     },
     {
       "epoch": 0.0736,
-      "grad_norm": 0.43886036816836405,
+      "grad_norm": 0.4725640071790753,
       "learning_rate": 4.993131457653682e-05,
-      "loss": 0.3016,
+      "loss": 0.3027,
       "step": 115
     },
     {
       "epoch": 0.07424,
-      "grad_norm": 0.4199519149927171,
+      "grad_norm": 0.4338862090037202,
       "learning_rate": 4.992733555504454e-05,
-      "loss": 0.3172,
+      "loss": 0.3179,
       "step": 116
     },
     {
       "epoch": 0.07488,
-      "grad_norm": 0.4190328574403923,
+      "grad_norm": 0.46363028669830225,
       "learning_rate": 4.9923244668832025e-05,
-      "loss": 0.3061,
+      "loss": 0.3075,
       "step": 117
     },
     {
       "epoch": 0.07552,
-      "grad_norm": 0.42026190311885486,
+      "grad_norm": 0.4285975451064772,
       "learning_rate": 4.991904193625768e-05,
-      "loss": 0.311,
+      "loss": 0.3117,
       "step": 118
     },
     {
       "epoch": 0.07616,
-      "grad_norm": 0.3651966339148085,
+      "grad_norm": 0.37832077577217255,
       "learning_rate": 4.991472737618181e-05,
-      "loss": 0.2732,
+      "loss": 0.2741,
       "step": 119
     },
     {
       "epoch": 0.0768,
-      "grad_norm": 0.40672045467505213,
+      "grad_norm": 0.41682487577691063,
       "learning_rate": 4.991030100796659e-05,
-      "loss": 0.3012,
+      "loss": 0.3019,
       "step": 120
     },
     {
       "epoch": 0.07744,
-      "grad_norm": 0.5292094384621548,
+      "grad_norm": 0.3716575375425735,
       "learning_rate": 4.9905762851475915e-05,
-      "loss": 0.3151,
+      "loss": 0.3154,
       "step": 121
     },
     {
       "epoch": 0.07808,
-      "grad_norm": 0.39692105389592514,
+      "grad_norm": 0.39573210518813595,
       "learning_rate": 4.990111292707537e-05,
       "loss": 0.3141,
       "step": 122
     },
     {
       "epoch": 0.07872,
-      "grad_norm": 0.39038544937881386,
+      "grad_norm": 0.4590220102268012,
       "learning_rate": 4.989635125563211e-05,
-      "loss": 0.2889,
+      "loss": 0.2911,
       "step": 123
     },
     {
       "epoch": 0.07936,
-      "grad_norm": 0.38869575358852054,
+      "grad_norm": 0.37422052213335605,
       "learning_rate": 4.989147785851477e-05,
-      "loss": 0.3083,
+      "loss": 0.3101,
       "step": 124
     },
     {
       "epoch": 0.08,
-      "grad_norm": 0.40766962966440357,
+      "grad_norm": 0.4283999831025601,
       "learning_rate": 4.988649275759335e-05,
-      "loss": 0.3186,
+      "loss": 0.32,
       "step": 125
     },
     {
       "epoch": 0.08064,
-      "grad_norm": 0.3818281612303487,
+      "grad_norm": 0.4133286791179184,
       "learning_rate": 4.988139597523916e-05,
-      "loss": 0.2913,
+      "loss": 0.2921,
       "step": 126
     },
     {
       "epoch": 0.08128,
-      "grad_norm": 0.4307784596567231,
+      "grad_norm": 0.4338853261587351,
       "learning_rate": 4.98761875343247e-05,
-      "loss": 0.2836,
+      "loss": 0.2846,
       "step": 127
     },
     {
       "epoch": 0.08192,
-      "grad_norm": 0.34729585966363985,
+      "grad_norm": 0.40905061401860165,
       "learning_rate": 4.987086745822352e-05,
-      "loss": 0.2844,
+      "loss": 0.2859,
       "step": 128
     },
     {
       "epoch": 0.08256,
-      "grad_norm": 0.3896516068942497,
+      "grad_norm": 0.39198335440260496,
       "learning_rate": 4.986543577081017e-05,
       "loss": 0.3214,
       "step": 129
     },
     {
       "epoch": 0.0832,
-      "grad_norm": 0.3605049005213821,
+      "grad_norm": 0.4072147961679465,
       "learning_rate": 4.985989249646007e-05,
-      "loss": 0.3018,
+      "loss": 0.303,
       "step": 130
     },
     {
       "epoch": 0.08384,
-      "grad_norm": 0.3728422361203703,
+      "grad_norm": 0.3960141589346821,
       "learning_rate": 4.985423766004939e-05,
-      "loss": 0.3097,
+      "loss": 0.3113,
       "step": 131
     },
     {
       "epoch": 0.08448,
-      "grad_norm": 0.4536755532298521,
+      "grad_norm": 0.4901660833242505,
       "learning_rate": 4.9848471286954965e-05,
-      "loss": 0.3266,
+      "loss": 0.3275,
       "step": 132
     },
     {
       "epoch": 0.08512,
-      "grad_norm": 0.3266252900478582,
+      "grad_norm": 0.3589215784625634,
       "learning_rate": 4.984259340305416e-05,
-      "loss": 0.2886,
+      "loss": 0.2894,
       "step": 133
     },
     {
       "epoch": 0.08576,
-      "grad_norm": 0.3884896844083972,
+      "grad_norm": 0.3998417320795738,
       "learning_rate": 4.9836604034724753e-05,
-      "loss": 0.2917,
+      "loss": 0.2926,
       "step": 134
     },
     {
       "epoch": 0.0864,
-      "grad_norm": 0.4147698757428519,
+      "grad_norm": 0.4296580917834302,
       "learning_rate": 4.9830503208844835e-05,
-      "loss": 0.2927,
+      "loss": 0.2946,
       "step": 135
     },
     {
       "epoch": 0.08704,
-      "grad_norm": 0.35299011989641577,
+      "grad_norm": 0.3544925944429582,
       "learning_rate": 4.982429095279266e-05,
-      "loss": 0.2941,
+      "loss": 0.2945,
       "step": 136
     },
     {
       "epoch": 0.08768,
-      "grad_norm": 0.43335162639635905,
+      "grad_norm": 0.4295052142302722,
       "learning_rate": 4.981796729444655e-05,
-      "loss": 0.3239,
+      "loss": 0.3251,
       "step": 137
     },
     {
       "epoch": 0.08832,
-      "grad_norm": 0.3940276865887743,
+      "grad_norm": 0.3810698917249912,
       "learning_rate": 4.981153226218477e-05,
-      "loss": 0.3076,
+      "loss": 0.3075,
       "step": 138
     },
     {
       "epoch": 0.08896,
-      "grad_norm": 0.3265746011811658,
+      "grad_norm": 0.3415977052001699,
       "learning_rate": 4.9804985884885366e-05,
-      "loss": 0.2868,
+      "loss": 0.2879,
       "step": 139
     },
     {
       "epoch": 0.0896,
-      "grad_norm": 0.38117660073360304,
+      "grad_norm": 0.3776850662597046,
       "learning_rate": 4.979832819192608e-05,
-      "loss": 0.3121,
+      "loss": 0.3129,
       "step": 140
     },
     {
       "epoch": 0.09024,
-      "grad_norm": 0.33607121469181583,
+      "grad_norm": 0.36338720848583483,
       "learning_rate": 4.97915592131842e-05,
-      "loss": 0.2878,
+      "loss": 0.2894,
       "step": 141
     },
     {
       "epoch": 0.09088,
-      "grad_norm": 0.3692306728222237,
+      "grad_norm": 0.39018083120154434,
       "learning_rate": 4.97846789790364e-05,
-      "loss": 0.3072,
+      "loss": 0.3073,
       "step": 142
     },
     {
       "epoch": 0.09152,
-      "grad_norm": 0.35801789239892756,
+      "grad_norm": 0.35656151957057863,
       "learning_rate": 4.977768752035865e-05,
-      "loss": 0.3115,
+      "loss": 0.3119,
       "step": 143
     },
     {
       "epoch": 0.09216,
-      "grad_norm": 0.3930347756960167,
+      "grad_norm": 0.3952926600732054,
       "learning_rate": 4.9770584868526046e-05,
-      "loss": 0.3116,
+      "loss": 0.312,
       "step": 144
     },
     {
       "epoch": 0.0928,
-      "grad_norm": 0.4285263500774063,
+      "grad_norm": 0.4288986528346642,
       "learning_rate": 4.9763371055412685e-05,
-      "loss": 0.3079,
+      "loss": 0.3084,
       "step": 145
     },
     {
       "epoch": 0.09344,
-      "grad_norm": 0.4008078130919119,
+      "grad_norm": 0.44696008744325466,
       "learning_rate": 4.9756046113391496e-05,
-      "loss": 0.3289,
+      "loss": 0.3292,
       "step": 146
     },
     {
       "epoch": 0.09408,
-      "grad_norm": 0.38210060332023665,
+      "grad_norm": 0.35163986650006296,
       "learning_rate": 4.974861007533413e-05,
-      "loss": 0.2861,
+      "loss": 0.2859,
       "step": 147
     },
     {
       "epoch": 0.09472,
-      "grad_norm": 0.4259001953879997,
+      "grad_norm": 0.42759675331208585,
       "learning_rate": 4.97410629746108e-05,
-      "loss": 0.2918,
+      "loss": 0.2919,
       "step": 148
     },
     {
       "epoch": 0.09536,
-      "grad_norm": 0.4076596626408926,
+      "grad_norm": 0.43529839553806815,
       "learning_rate": 4.9733404845090116e-05,
-      "loss": 0.2841,
+      "loss": 0.2842,
       "step": 149
     },
     {
       "epoch": 0.096,
-      "grad_norm": 0.40095369405114606,
+      "grad_norm": 0.41401104367991204,
       "learning_rate": 4.972563572113894e-05,
-      "loss": 0.2902,
+      "loss": 0.2904,
       "step": 150
     },
     {
       "epoch": 0.09664,
-      "grad_norm": 0.44874206133172484,
+      "grad_norm": 0.41056372455002155,
       "learning_rate": 4.971775563762226e-05,
-      "loss": 0.2912,
+      "loss": 0.291,
       "step": 151
     },
     {
       "epoch": 0.09728,
-      "grad_norm": 0.41878107643356327,
+      "grad_norm": 0.42450110637249844,
       "learning_rate": 4.970976462990299e-05,
-      "loss": 0.3106,
+      "loss": 0.3107,
       "step": 152
     },
     {
       "epoch": 0.09792,
-      "grad_norm": 0.3886961745648121,
+      "grad_norm": 0.36252875803102586,
       "learning_rate": 4.970166273384183e-05,
-      "loss": 0.2866,
+      "loss": 0.2867,
       "step": 153
     },
     {
       "epoch": 0.09856,
-      "grad_norm": 0.34534258751325136,
+      "grad_norm": 0.34724268049235696,
       "learning_rate": 4.969344998579712e-05,
-      "loss": 0.2657,
+      "loss": 0.2663,
       "step": 154
     },
     {
       "epoch": 0.0992,
-      "grad_norm": 0.3755269967559221,
+      "grad_norm": 0.4062042219778404,
       "learning_rate": 4.9685126422624644e-05,
-      "loss": 0.3146,
+      "loss": 0.3156,
       "step": 155
     },
     {
       "epoch": 0.09984,
-      "grad_norm": 0.40024019989003823,
+      "grad_norm": 0.38599214382639446,
       "learning_rate": 4.96766920816775e-05,
-      "loss": 0.3144,
+      "loss": 0.3147,
       "step": 156
     },
     {
       "epoch": 0.10048,
-      "grad_norm": 0.3528612408949507,
+      "grad_norm": 0.3662678782254152,
       "learning_rate": 4.966814700080592e-05,
-      "loss": 0.3175,
+      "loss": 0.3184,
       "step": 157
     },
     {
       "epoch": 0.10112,
-      "grad_norm": 0.37989058888383176,
+      "grad_norm": 0.37758974795094674,
       "learning_rate": 4.965949121835707e-05,
-      "loss": 0.3336,
+      "loss": 0.3341,
       "step": 158
     },
     {
       "epoch": 0.10176,
-      "grad_norm": 0.3935853059250065,
+      "grad_norm": 0.42335782183402376,
       "learning_rate": 4.965072477317493e-05,
-      "loss": 0.3093,
+      "loss": 0.3105,
       "step": 159
     },
     {
       "epoch": 0.1024,
-      "grad_norm": 0.3481727131458977,
+      "grad_norm": 0.3445716615812455,
       "learning_rate": 4.964184770460007e-05,
-      "loss": 0.2928,
+      "loss": 0.2935,
       "step": 160
     },
     {
       "epoch": 0.10304,
-      "grad_norm": 0.407095370695997,
+      "grad_norm": 0.40006390954972004,
       "learning_rate": 4.963286005246953e-05,
-      "loss": 0.2819,
+      "loss": 0.2834,
       "step": 161
     },
     {
       "epoch": 0.10368,
-      "grad_norm": 0.36659617800740035,
+      "grad_norm": 0.3664434351809571,
       "learning_rate": 4.9623761857116594e-05,
-      "loss": 0.3195,
+      "loss": 0.3191,
       "step": 162
     },
     {
       "epoch": 0.10432,
-      "grad_norm": 0.3734389648125262,
+      "grad_norm": 0.3746346471394553,
       "learning_rate": 4.961455315937059e-05,
-      "loss": 0.2812,
+      "loss": 0.282,
       "step": 163
     },
     {
       "epoch": 0.10496,
-      "grad_norm": 0.3413505309534122,
+      "grad_norm": 0.3543125314610225,
       "learning_rate": 4.960523400055679e-05,
-      "loss": 0.2934,
+      "loss": 0.2939,
       "step": 164
     },
     {
       "epoch": 0.1056,
-      "grad_norm": 0.3301887986531996,
+      "grad_norm": 0.34883234374623584,
       "learning_rate": 4.959580442249615e-05,
-      "loss": 0.3012,
+      "loss": 0.3011,
       "step": 165
     },
     {
       "epoch": 0.10624,
-      "grad_norm": 0.3123585093078612,
+      "grad_norm": 0.33879752208803143,
       "learning_rate": 4.958626446750515e-05,
-      "loss": 0.3202,
+      "loss": 0.3211,
       "step": 166
     },
     {
       "epoch": 0.10688,
-      "grad_norm": 0.35966543460940326,
+      "grad_norm": 0.37300311895352173,
       "learning_rate": 4.9576614178395595e-05,
-      "loss": 0.3006,
+      "loss": 0.301,
       "step": 167
     },
     {
       "epoch": 0.10752,
-      "grad_norm": 0.332623020117398,
+      "grad_norm": 0.3273073442908486,
       "learning_rate": 4.956685359847445e-05,
-      "loss": 0.297,
+      "loss": 0.2969,
       "step": 168
     },
     {
       "epoch": 0.10816,
-      "grad_norm": 0.3142621220673422,
+      "grad_norm": 0.3116171444679886,
       "learning_rate": 4.95569827715436e-05,
-      "loss": 0.2812,
+      "loss": 0.2817,
       "step": 169
     },
     {
       "epoch": 0.1088,
-      "grad_norm": 0.32468349021820947,
+      "grad_norm": 0.3361243539743496,
       "learning_rate": 4.9547001741899693e-05,
-      "loss": 0.2968,
+      "loss": 0.2977,
       "step": 170
     },
     {
       "epoch": 0.10944,
-      "grad_norm": 0.3408361426728115,
+      "grad_norm": 0.3418189113063408,
       "learning_rate": 4.953691055433391e-05,
-      "loss": 0.2903,
+      "loss": 0.2916,
       "step": 171
     },
     {
       "epoch": 0.11008,
-      "grad_norm": 0.36227080591435323,
+      "grad_norm": 0.3625913676240481,
       "learning_rate": 4.952670925413182e-05,
-      "loss": 0.3035,
+      "loss": 0.3038,
       "step": 172
     },
     {
       "epoch": 0.11072,
-      "grad_norm": 0.33583803873672646,
+      "grad_norm": 0.34636880625513666,
       "learning_rate": 4.951639788707307e-05,
-      "loss": 0.2845,
+      "loss": 0.2854,
       "step": 173
     },
     {
       "epoch": 0.11136,
-      "grad_norm": 0.34619921286834143,
+      "grad_norm": 0.35977242712508983,
       "learning_rate": 4.950597649943132e-05,
-      "loss": 0.278,
+      "loss": 0.2786,
       "step": 174
     },
     {
       "epoch": 0.112,
-      "grad_norm": 0.3404081978672373,
+      "grad_norm": 0.33680285828380363,
       "learning_rate": 4.9495445137973905e-05,
-      "loss": 0.2929,
+      "loss": 0.2934,
       "step": 175
     },
     {
       "epoch": 0.11264,
-      "grad_norm": 0.3310437000635691,
+      "grad_norm": 0.3270580360967154,
       "learning_rate": 4.948480384996172e-05,
-      "loss": 0.2872,
+      "loss": 0.2878,
       "step": 176
     },
     {
       "epoch": 0.11328,
-      "grad_norm": 0.35248099294249946,
+      "grad_norm": 0.3363028989990994,
       "learning_rate": 4.947405268314894e-05,
-      "loss": 0.3003,
+      "loss": 0.3014,
       "step": 177
     },
     {
       "epoch": 0.11392,
-      "grad_norm": 0.4133726433387471,
+      "grad_norm": 0.39424588001597727,
       "learning_rate": 4.9463191685782865e-05,
-      "loss": 0.307,
+      "loss": 0.3076,
       "step": 178
     },
     {
       "epoch": 0.11456,
-      "grad_norm": 0.31342845143199094,
+      "grad_norm": 0.31183590038965797,
       "learning_rate": 4.945222090660364e-05,
-      "loss": 0.2846,
+      "loss": 0.2859,
       "step": 179
     },
     {
       "epoch": 0.1152,
-      "grad_norm": 0.3518931188946702,
+      "grad_norm": 0.3487969513905697,
       "learning_rate": 4.944114039484411e-05,
-      "loss": 0.3046,
+      "loss": 0.3057,
       "step": 180
     },
     {
       "epoch": 0.11584,
-      "grad_norm": 0.3421314438441075,
+      "grad_norm": 0.3502275557092516,
       "learning_rate": 4.942995020022954e-05,
-      "loss": 0.3244,
+      "loss": 0.3255,
       "step": 181
     },
     {
       "epoch": 0.11648,
-      "grad_norm": 0.3267692222072987,
+      "grad_norm": 0.3051981568928157,
       "learning_rate": 4.9418650372977384e-05,
-      "loss": 0.2887,
+      "loss": 0.2881,
       "step": 182
     },
     {
       "epoch": 0.11712,
-      "grad_norm": 0.3150731393402504,
+      "grad_norm": 0.3213236160412315,
       "learning_rate": 4.940724096379714e-05,
-      "loss": 0.2826,
+      "loss": 0.2832,
       "step": 183
     },
     {
       "epoch": 0.11776,
-      "grad_norm": 0.28931360046418647,
+      "grad_norm": 0.2804638187646734,
       "learning_rate": 4.9395722023890035e-05,
-      "loss": 0.2806,
+      "loss": 0.2807,
       "step": 184
     },
     {
       "epoch": 0.1184,
-      "grad_norm": 0.3456286699266885,
+      "grad_norm": 0.32770405872721187,
       "learning_rate": 4.938409360494883e-05,
-      "loss": 0.2835,
+      "loss": 0.2843,
       "step": 185
     },
     {
       "epoch": 0.11904,
-      "grad_norm": 0.30464221501484345,
+      "grad_norm": 0.3196607887549686,
       "learning_rate": 4.937235575915761e-05,
-      "loss": 0.2917,
+      "loss": 0.2928,
       "step": 186
     },
     {
       "epoch": 0.11968,
-      "grad_norm": 0.3227096120306476,
+      "grad_norm": 0.31014707115252577,
       "learning_rate": 4.93605085391915e-05,
-      "loss": 0.2923,
+      "loss": 0.2926,
       "step": 187
     },
     {
       "epoch": 0.12032,
-      "grad_norm": 0.366171085820676,
+      "grad_norm": 0.3667018178707909,
       "learning_rate": 4.934855199821647e-05,
-      "loss": 0.2835,
+      "loss": 0.2837,
       "step": 188
     },
     {
       "epoch": 0.12096,
-      "grad_norm": 0.3304740847644211,
+      "grad_norm": 0.3370594761880159,
       "learning_rate": 4.93364861898891e-05,
-      "loss": 0.2763,
+      "loss": 0.2776,
       "step": 189
     },
     {
       "epoch": 0.1216,
-      "grad_norm": 0.35804683150654454,
+      "grad_norm": 0.3889677526233924,
       "learning_rate": 4.932431116835628e-05,
-      "loss": 0.297,
+      "loss": 0.2976,
       "step": 190
     },
     {
       "epoch": 0.12224,
-      "grad_norm": 0.3280992910010731,
+      "grad_norm": 0.32895998318993375,
       "learning_rate": 4.931202698825503e-05,
-      "loss": 0.2814,
+      "loss": 0.2823,
       "step": 191
     },
     {
       "epoch": 0.12288,
-      "grad_norm": 0.29841743502194873,
+      "grad_norm": 0.30268716856662087,
       "learning_rate": 4.929963370471226e-05,
-      "loss": 0.2878,
+      "loss": 0.2889,
       "step": 192
     },
     {
       "epoch": 0.12352,
-      "grad_norm": 0.2854325525422512,
+      "grad_norm": 0.2784816493323616,
       "learning_rate": 4.928713137334445e-05,
-      "loss": 0.2816,
+      "loss": 0.2819,
       "step": 193
     },
     {
       "epoch": 0.12416,
-      "grad_norm": 0.3265192973580523,
+      "grad_norm": 0.31652215898467684,
       "learning_rate": 4.927452005025748e-05,
-      "loss": 0.3019,
+      "loss": 0.3023,
       "step": 194
     },
     {
       "epoch": 0.1248,
-      "grad_norm": 0.31439513957912246,
+      "grad_norm": 0.30644827432002086,
       "learning_rate": 4.926179979204633e-05,
-      "loss": 0.3023,
+      "loss": 0.303,
       "step": 195
     },
     {
       "epoch": 0.12544,
-      "grad_norm": 0.3043558334720137,
+      "grad_norm": 0.298870893997602,
       "learning_rate": 4.9248970655794836e-05,
-      "loss": 0.2921,
+      "loss": 0.2924,
       "step": 196
     },
     {
       "epoch": 0.12608,
-      "grad_norm": 0.3083301719907164,
+      "grad_norm": 0.3019490103946191,
       "learning_rate": 4.923603269907545e-05,
-      "loss": 0.2746,
+      "loss": 0.2749,
       "step": 197
     },
     {
       "epoch": 0.12672,
-      "grad_norm": 0.3091270018423747,
+      "grad_norm": 0.34179704436331393,
       "learning_rate": 4.922298597994897e-05,
-      "loss": 0.2842,
+      "loss": 0.2837,
       "step": 198
     },
     {
       "epoch": 0.12736,
-      "grad_norm": 0.34382329346733104,
+      "grad_norm": 0.34714244369014874,
       "learning_rate": 4.920983055696428e-05,
-      "loss": 0.2937,
+      "loss": 0.2941,
       "step": 199
     },
     {
       "epoch": 0.128,
-      "grad_norm": 0.35936428325700825,
+      "grad_norm": 0.34514991924274774,
       "learning_rate": 4.919656648915807e-05,
-      "loss": 0.2847,
+      "loss": 0.2852,
       "step": 200
     },
     {
       "epoch": 0.12864,
-      "grad_norm": 0.3044935865997733,
+      "grad_norm": 0.2845599679176182,
       "learning_rate": 4.918319383605462e-05,
-      "loss": 0.2575,
+      "loss": 0.2583,
       "step": 201
     },
     {
       "epoch": 0.12928,
-      "grad_norm": 0.3547660540984471,
+      "grad_norm": 0.3683838687495882,
       "learning_rate": 4.916971265766547e-05,
-      "loss": 0.2949,
+      "loss": 0.2954,
       "step": 202
     },
     {
       "epoch": 0.12992,
-      "grad_norm": 0.36659141032868914,
+      "grad_norm": 0.35457325941860496,
       "learning_rate": 4.915612301448919e-05,
-      "loss": 0.2949,
+      "loss": 0.2957,
       "step": 203
     },
     {
       "epoch": 0.13056,
-      "grad_norm": 0.3603069106129497,
+      "grad_norm": 0.4102065495806397,
       "learning_rate": 4.914242496751111e-05,
-      "loss": 0.2882,
+      "loss": 0.289,
       "step": 204
     },
     {
       "epoch": 0.1312,
-      "grad_norm": 0.3963859179748674,
+      "grad_norm": 0.391931082619993,
       "learning_rate": 4.912861857820303e-05,
-      "loss": 0.3027,
+      "loss": 0.303,
       "step": 205
     },
     {
       "epoch": 0.13184,
-      "grad_norm": 0.37181091318644754,
+      "grad_norm": 0.37855393422147743,
       "learning_rate": 4.911470390852295e-05,
-      "loss": 0.3017,
+      "loss": 0.3013,
       "step": 206
     },
     {
       "epoch": 0.13248,
-      "grad_norm": 0.31409311742594026,
+      "grad_norm": 0.31160560780141033,
       "learning_rate": 4.91006810209148e-05,
-      "loss": 0.2884,
+      "loss": 0.2888,
       "step": 207
     },
     {
       "epoch": 0.13312,
-      "grad_norm": 0.3722479719556711,
+      "grad_norm": 0.38185424339716356,
       "learning_rate": 4.9086549978308125e-05,
-      "loss": 0.2991,
+      "loss": 0.2999,
       "step": 208
     },
     {
       "epoch": 0.13376,
-      "grad_norm": 0.3004507702000884,
+      "grad_norm": 0.30532616858387496,
       "learning_rate": 4.9072310844117875e-05,
-      "loss": 0.272,
+      "loss": 0.2732,
       "step": 209
     },
     {
       "epoch": 0.1344,
-      "grad_norm": 0.33911903822311734,
+      "grad_norm": 0.36508067170966735,
       "learning_rate": 4.9057963682244045e-05,
-      "loss": 0.2908,
+      "loss": 0.2909,
       "step": 210
     },
     {
       "epoch": 0.13504,
-      "grad_norm": 0.34057084273821486,
+      "grad_norm": 0.3148112484256156,
       "learning_rate": 4.904350855707142e-05,
-      "loss": 0.2869,
+      "loss": 0.2874,
       "step": 211
     },
     {
       "epoch": 0.13568,
-      "grad_norm": 0.31234744073094967,
+      "grad_norm": 0.3274325146184032,
       "learning_rate": 4.9028945533469276e-05,
-      "loss": 0.2653,
+      "loss": 0.2661,
       "step": 212
     },
     {
       "epoch": 0.13632,
-      "grad_norm": 0.3755970970195266,
+      "grad_norm": 0.3685996593265165,
       "learning_rate": 4.9014274676791134e-05,
-      "loss": 0.3031,
+      "loss": 0.3037,
       "step": 213
     },
     {
       "epoch": 0.13696,
-      "grad_norm": 0.32331976083334474,
+      "grad_norm": 0.34087816560391343,
       "learning_rate": 4.899949605287439e-05,
-      "loss": 0.2801,
+      "loss": 0.2797,
       "step": 214
     },
     {
       "epoch": 0.1376,
-      "grad_norm": 0.3755925580675204,
+      "grad_norm": 0.3613310322508522,
       "learning_rate": 4.898460972804009e-05,
-      "loss": 0.3092,
+      "loss": 0.3096,
       "step": 215
     },
     {
       "epoch": 0.13824,
-      "grad_norm": 0.36036624380998206,
+      "grad_norm": 0.375457846044839,
       "learning_rate": 4.896961576909258e-05,
-      "loss": 0.2817,
+      "loss": 0.2822,
       "step": 216
     },
     {
       "epoch": 0.13888,
-      "grad_norm": 0.36082215281448893,
+      "grad_norm": 0.3437437239754465,
       "learning_rate": 4.895451424331924e-05,
-      "loss": 0.2721,
+      "loss": 0.2725,
       "step": 217
     },
     {
       "epoch": 0.13952,
-      "grad_norm": 0.34106303407373795,
+      "grad_norm": 0.35840101574935934,
       "learning_rate": 4.893930521849016e-05,
-      "loss": 0.2675,
+      "loss": 0.268,
       "step": 218
     },
     {
       "epoch": 0.14016,
-      "grad_norm": 0.392247984251282,
+      "grad_norm": 0.3767989948671444,
       "learning_rate": 4.892398876285785e-05,
-      "loss": 0.2904,
+      "loss": 0.2901,
       "step": 219
     },
     {
       "epoch": 0.1408,
-      "grad_norm": 0.3855496076265292,
+      "grad_norm": 0.4065322507872538,
       "learning_rate": 4.890856494515696e-05,
-      "loss": 0.2887,
+      "loss": 0.2892,
       "step": 220
     },
     {
       "epoch": 0.14144,
-      "grad_norm": 0.35740557532357425,
+      "grad_norm": 0.3526522553070582,
       "learning_rate": 4.889303383460389e-05,
-      "loss": 0.2814,
+      "loss": 0.2818,
       "step": 221
     },
     {
       "epoch": 0.14208,
-      "grad_norm": 0.36474251567342575,
+      "grad_norm": 0.36687661566425206,
       "learning_rate": 4.887739550089656e-05,
       "loss": 0.2788,
       "step": 222
     },
     {
       "epoch": 0.14272,
-      "grad_norm": 0.36975719759429415,
+      "grad_norm": 0.36376035976937565,
       "learning_rate": 4.886165001421408e-05,
-      "loss": 0.3103,
+      "loss": 0.3102,
       "step": 223
     },
     {
       "epoch": 0.14336,
-      "grad_norm": 0.32122517879682544,
+      "grad_norm": 0.32605502970392203,
       "learning_rate": 4.884579744521638e-05,
-      "loss": 0.2846,
+      "loss": 0.2858,
       "step": 224
     },
     {
       "epoch": 0.144,
-      "grad_norm": 0.34843218274712867,
+      "grad_norm": 0.3473388829365213,
       "learning_rate": 4.8829837865043984e-05,
-      "loss": 0.2916,
+      "loss": 0.2927,
       "step": 225
     },
     {
       "epoch": 0.14464,
-      "grad_norm": 0.31168966766131073,
+      "grad_norm": 0.3161945789520979,
       "learning_rate": 4.881377134531762e-05,
-      "loss": 0.2678,
+      "loss": 0.2684,
       "step": 226
     },
     {
       "epoch": 0.14528,
-      "grad_norm": 0.32795585476788547,
+      "grad_norm": 0.3233058980574442,
       "learning_rate": 4.8797597958137914e-05,
-      "loss": 0.2804,
+      "loss": 0.2809,
       "step": 227
     },
     {
       "epoch": 0.14592,
-      "grad_norm": 0.366243634878421,
+      "grad_norm": 0.35628813657982605,
       "learning_rate": 4.878131777608509e-05,
-      "loss": 0.2848,
+      "loss": 0.2857,
       "step": 228
     },
     {
       "epoch": 0.14656,
-      "grad_norm": 0.32937437081275867,
+      "grad_norm": 0.34474317130830845,
       "learning_rate": 4.876493087221862e-05,
-      "loss": 0.2971,
+      "loss": 0.2981,
       "step": 229
     },
     {
       "epoch": 0.1472,
-      "grad_norm": 0.5263731374338324,
+      "grad_norm": 0.3332723948946639,
       "learning_rate": 4.87484373200769e-05,
-      "loss": 0.2902,
+      "loss": 0.2904,
       "step": 230
     },
     {
       "epoch": 0.14784,
-      "grad_norm": 0.3289646826837418,
+      "grad_norm": 0.36212328347879097,
       "learning_rate": 4.8731837193676944e-05,
-      "loss": 0.2792,
+      "loss": 0.2803,
       "step": 231
     },
     {
       "epoch": 0.14848,
-      "grad_norm": 0.3388926800791491,
+      "grad_norm": 0.32285084742216597,
       "learning_rate": 4.871513056751401e-05,
-      "loss": 0.2995,
+      "loss": 0.2991,
       "step": 232
     },
     {
       "epoch": 0.14912,
-      "grad_norm": 0.3530964158377889,
+      "grad_norm": 0.3757428283729143,
       "learning_rate": 4.8698317516561295e-05,
-      "loss": 0.3012,
+      "loss": 0.3007,
       "step": 233
     },
     {
       "epoch": 0.14976,
-      "grad_norm": 0.3661383465989981,
+      "grad_norm": 0.352925038021312,
       "learning_rate": 4.8681398116269596e-05,
-      "loss": 0.2867,
+      "loss": 0.2865,
       "step": 234
     },
     {
       "epoch": 0.1504,
-      "grad_norm": 0.33175219745607853,
+      "grad_norm": 0.33841490625531084,
       "learning_rate": 4.8664372442566954e-05,
-      "loss": 0.2935,
+      "loss": 0.294,
       "step": 235
     },
     {
       "epoch": 0.15104,
-      "grad_norm": 0.33548701697224265,
+      "grad_norm": 0.34183505785167,
       "learning_rate": 4.864724057185834e-05,
-      "loss": 0.2689,
+      "loss": 0.2691,
       "step": 236
     },
     {
       "epoch": 0.15168,
-      "grad_norm": 0.32268615702639075,
+      "grad_norm": 0.339769370877407,
       "learning_rate": 4.86300025810253e-05,
-      "loss": 0.2809,
+      "loss": 0.2818,
       "step": 237
     },
     {
       "epoch": 0.15232,
-      "grad_norm": 0.30024229483347803,
+      "grad_norm": 0.30700076377351593,
       "learning_rate": 4.861265854742559e-05,
-      "loss": 0.2744,
+      "loss": 0.2748,
       "step": 238
     },
     {
       "epoch": 0.15296,
-      "grad_norm": 0.34761571238874817,
+      "grad_norm": 0.3489951897607549,
       "learning_rate": 4.859520854889287e-05,
       "loss": 0.2937,
       "step": 239
     },
     {
       "epoch": 0.1536,
-      "grad_norm": 0.29306794724474977,
+      "grad_norm": 0.29156819501858466,
       "learning_rate": 4.857765266373631e-05,
-      "loss": 0.2962,
+      "loss": 0.2958,
       "step": 240
     },
     {
       "epoch": 0.15424,
-      "grad_norm": 0.3293199526625673,
+      "grad_norm": 0.3384643439513567,
       "learning_rate": 4.855999097074027e-05,
-      "loss": 0.2938,
+      "loss": 0.2939,
       "step": 241
     },
     {
       "epoch": 0.15488,
-      "grad_norm": 0.30974364555406575,
+      "grad_norm": 0.3077298938535667,
       "learning_rate": 4.8542223549163946e-05,
-      "loss": 0.2725,
+      "loss": 0.2724,
       "step": 242
     },
     {
       "epoch": 0.15552,
-      "grad_norm": 0.328770395916147,
+      "grad_norm": 0.31947348723786806,
       "learning_rate": 4.8524350478740985e-05,
-      "loss": 0.267,
+      "loss": 0.2666,
       "step": 243
     },
     {
       "epoch": 0.15616,
-      "grad_norm": 0.36960435497430866,
+      "grad_norm": 0.3774412893720739,
       "learning_rate": 4.850637183967917e-05,
       "loss": 0.2852,
       "step": 244
     },
     {
       "epoch": 0.1568,
-      "grad_norm": 0.3202372900146965,
+      "grad_norm": 0.3066835259607928,
       "learning_rate": 4.848828771266002e-05,
-      "loss": 0.2885,
+      "loss": 0.2883,
       "step": 245
     },
     {
       "epoch": 0.15744,
-      "grad_norm": 0.38632291902597493,
+      "grad_norm": 0.39290641044610064,
       "learning_rate": 4.847009817883844e-05,
-      "loss": 0.2833,
+      "loss": 0.284,
       "step": 246
     },
     {
       "epoch": 0.15808,
-      "grad_norm": 0.3039843721273191,
+      "grad_norm": 0.3087733867875181,
       "learning_rate": 4.845180331984239e-05,
-      "loss": 0.2724,
+      "loss": 0.2726,
       "step": 247
     },
     {
       "epoch": 0.15872,
-      "grad_norm": 0.37777270335325575,
+      "grad_norm": 0.3823066055668157,
       "learning_rate": 4.843340321777246e-05,
-      "loss": 0.2654,
+      "loss": 0.2656,
       "step": 248
     },
     {
       "epoch": 0.15936,
-      "grad_norm": 0.32765169609391825,
+      "grad_norm": 0.3211297194539539,
       "learning_rate": 4.8414897955201556e-05,
-      "loss": 0.2624,
+      "loss": 0.2628,
       "step": 249
     },
     {
       "epoch": 0.16,
-      "grad_norm": 0.30937134246453424,
+      "grad_norm": 0.3120564620889703,
       "learning_rate": 4.8396287615174484e-05,
-      "loss": 0.2588,
+      "loss": 0.2587,
       "step": 250
     },
     {
       "epoch": 0.16064,
-      "grad_norm": 0.3283849158663661,
+      "grad_norm": 0.3289936954141222,
       "learning_rate": 4.837757228120762e-05,
-      "loss": 0.2895,
+      "loss": 0.2896,
       "step": 251
     },
     {
       "epoch": 0.16128,
-      "grad_norm": 0.31535959823298476,
+      "grad_norm": 0.3188145095365673,
       "learning_rate": 4.835875203728849e-05,
-      "loss": 0.2552,
+      "loss": 0.2555,
       "step": 252
     },
     {
       "epoch": 0.16192,
-      "grad_norm": 0.473186549819399,
+      "grad_norm": 0.3889796632457812,
       "learning_rate": 4.8339826967875443e-05,
-      "loss": 0.288,
+      "loss": 0.2872,
       "step": 253
     },
     {
       "epoch": 0.16256,
-      "grad_norm": 0.36640618590983065,
+      "grad_norm": 0.381368538527566,
       "learning_rate": 4.832079715789724e-05,
-      "loss": 0.2991,
+      "loss": 0.2988,
       "step": 254
     },
     {
       "epoch": 0.1632,
-      "grad_norm": 0.40519416561462235,
+      "grad_norm": 0.36630965780444474,
       "learning_rate": 4.830166269275267e-05,
-      "loss": 0.2916,
+      "loss": 0.2906,
       "step": 255
     },
     {
       "epoch": 0.16384,
-      "grad_norm": 0.3537388526585215,
+      "grad_norm": 0.3552050245169036,
       "learning_rate": 4.828242365831016e-05,
-      "loss": 0.2757,
+      "loss": 0.2761,
       "step": 256
     },
     {
       "epoch": 0.16448,
-      "grad_norm": 0.3817071905765977,
+      "grad_norm": 0.37404437653877487,
       "learning_rate": 4.8263080140907455e-05,
-      "loss": 0.2916,
+      "loss": 0.2908,
       "step": 257
     },
     {
       "epoch": 0.16512,
-      "grad_norm": 0.361265701608818,
+      "grad_norm": 0.3507329450496866,
       "learning_rate": 4.824363222735114e-05,
-      "loss": 0.2755,
+      "loss": 0.275,
       "step": 258
     },
     {
       "epoch": 0.16576,
-      "grad_norm": 0.35887429797588116,
+      "grad_norm": 0.37299320025474214,
       "learning_rate": 4.82240800049163e-05,
       "loss": 0.2801,
       "step": 259
     },
     {
       "epoch": 0.1664,
-      "grad_norm": 0.4048340907079254,
+      "grad_norm": 0.38754434557373796,
       "learning_rate": 4.820442356134612e-05,
-      "loss": 0.2716,
+      "loss": 0.2709,
       "step": 260
     },
     {
       "epoch": 0.16704,
-      "grad_norm": 0.3554137106703619,
+      "grad_norm": 0.33798911863787995,
       "learning_rate": 4.8184662984851505e-05,
-      "loss": 0.3015,
+      "loss": 0.3009,
       "step": 261
     },
     {
       "epoch": 0.16768,
-      "grad_norm": 0.3540300438998987,
+      "grad_norm": 0.3559262440354945,
       "learning_rate": 4.816479836411065e-05,
-      "loss": 0.2833,
+      "loss": 0.2835,
       "step": 262
     },
     {
       "epoch": 0.16832,
-      "grad_norm": 0.4227691893247339,
+      "grad_norm": 0.4158079436380025,
       "learning_rate": 4.8144829788268674e-05,
-      "loss": 0.2958,
+      "loss": 0.2954,
       "step": 263
     },
     {
       "epoch": 0.16896,
-      "grad_norm": 0.35085282951596225,
+      "grad_norm": 0.3603676966517829,
       "learning_rate": 4.8124757346937214e-05,
-      "loss": 0.2633,
+      "loss": 0.2637,
       "step": 264
     },
     {
       "epoch": 0.1696,
-      "grad_norm": 0.3304398208876773,
+      "grad_norm": 0.3288136132861365,
       "learning_rate": 4.810458113019401e-05,
-      "loss": 0.275,
+      "loss": 0.2747,
       "step": 265
     },
     {
       "epoch": 0.17024,
-      "grad_norm": 0.32563241609865123,
+      "grad_norm": 0.3402933784236876,
       "learning_rate": 4.8084301228582485e-05,
-      "loss": 0.2576,
+      "loss": 0.2577,
       "step": 266
     },
     {
       "epoch": 0.17088,
-      "grad_norm": 0.38836797469119294,
+      "grad_norm": 0.36930227507381896,
       "learning_rate": 4.8063917733111406e-05,
-      "loss": 0.3099,
+      "loss": 0.3098,
       "step": 267
     },
     {
       "epoch": 0.17152,
-      "grad_norm": 0.3217274402475735,
+      "grad_norm": 0.309709429764299,
       "learning_rate": 4.8043430735254404e-05,
-      "loss": 0.2813,
+      "loss": 0.2808,
       "step": 268
     },
     {
       "epoch": 0.17216,
-      "grad_norm": 0.36813790588683976,
+      "grad_norm": 0.35813482945344854,
       "learning_rate": 4.8022840326949595e-05,
-      "loss": 0.2804,
+      "loss": 0.2812,
       "step": 269
     },
     {
       "epoch": 0.1728,
-      "grad_norm": 0.37938400264260286,
+      "grad_norm": 0.3631234651320861,
       "learning_rate": 4.800214660059916e-05,
-      "loss": 0.2805,
+      "loss": 0.2797,
       "step": 270
     },
     {
       "epoch": 0.17344,
-      "grad_norm": 0.3415094791341939,
+      "grad_norm": 0.3309005171733229,
       "learning_rate": 4.798134964906894e-05,
-      "loss": 0.2875,
+      "loss": 0.2867,
       "step": 271
     },
     {
       "epoch": 0.17408,
-      "grad_norm": 0.37458115660567415,
+      "grad_norm": 0.35715429838212853,
       "learning_rate": 4.7960449565688e-05,
-      "loss": 0.2635,
+      "loss": 0.2638,
       "step": 272
     },
     {
       "epoch": 0.17472,
-      "grad_norm": 0.3250425989019864,
+      "grad_norm": 0.3279629544159387,
       "learning_rate": 4.793944644424826e-05,
       "loss": 0.2855,
       "step": 273
     },
     {
       "epoch": 0.17536,
-      "grad_norm": 0.3766228590098436,
+      "grad_norm": 0.37196752238968245,
       "learning_rate": 4.7918340379003985e-05,
       "loss": 0.2875,
       "step": 274
     },
     {
       "epoch": 0.176,
-      "grad_norm": 0.32502593608424385,
+      "grad_norm": 0.31834632572626353,
       "learning_rate": 4.7897131464671444e-05,
-      "loss": 0.3004,
+      "loss": 0.3006,
       "step": 275
     },
     {
       "epoch": 0.17664,
-      "grad_norm": 0.36374996921029645,
+      "grad_norm": 0.34089694844336105,
       "learning_rate": 4.787581979642845e-05,
-      "loss": 0.2554,
+      "loss": 0.2557,
       "step": 276
     },
     {
       "epoch": 0.17728,
-      "grad_norm": 0.3271666456259006,
+      "grad_norm": 0.31522922339322745,
       "learning_rate": 4.7854405469913944e-05,
-      "loss": 0.274,
+      "loss": 0.2741,
       "step": 277
     },
     {
       "epoch": 0.17792,
-      "grad_norm": 0.3208313062202937,
+      "grad_norm": 0.31669044816633024,
       "learning_rate": 4.7832888581227545e-05,
-      "loss": 0.2765,
+      "loss": 0.2766,
       "step": 278
     },
     {
       "epoch": 0.17856,
-      "grad_norm": 0.33957962251078677,
+      "grad_norm": 0.3460333243052226,
       "learning_rate": 4.7811269226929135e-05,
-      "loss": 0.2801,
+      "loss": 0.2802,
       "step": 279
     },
     {
       "epoch": 0.1792,
-      "grad_norm": 0.3491584815817293,
+      "grad_norm": 0.33307222568871414,
       "learning_rate": 4.7789547504038443e-05,
-      "loss": 0.2692,
+      "loss": 0.2696,
       "step": 280
     },
     {
       "epoch": 0.17984,
-      "grad_norm": 0.34068986400037704,
+      "grad_norm": 0.3326671192080615,
       "learning_rate": 4.776772351003458e-05,
-      "loss": 0.281,
+      "loss": 0.2811,
       "step": 281
     },
     {
       "epoch": 0.18048,
-      "grad_norm": 0.32860891352594357,
+      "grad_norm": 0.32991380728873526,
       "learning_rate": 4.774579734285559e-05,
-      "loss": 0.2877,
+      "loss": 0.2878,
       "step": 282
     },
     {
       "epoch": 0.18112,
-      "grad_norm": 0.32161948162561343,
+      "grad_norm": 0.3247648689257416,
       "learning_rate": 4.772376910089808e-05,
-      "loss": 0.2791,
+      "loss": 0.2786,
       "step": 283
     },
     {
       "epoch": 0.18176,
-      "grad_norm": 0.29944358983259645,
+      "grad_norm": 0.3282388341197448,
       "learning_rate": 4.770163888301668e-05,
-      "loss": 0.2706,
+      "loss": 0.2708,
       "step": 284
     },
     {
       "epoch": 0.1824,
-      "grad_norm": 0.33259975876524245,
+      "grad_norm": 0.3314880902653025,
       "learning_rate": 4.7679406788523685e-05,
-      "loss": 0.2779,
+      "loss": 0.278,
       "step": 285
     },
     {
       "epoch": 0.18304,
-      "grad_norm": 0.3192766834044811,
+      "grad_norm": 0.3223575932382192,
       "learning_rate": 4.765707291718856e-05,
-      "loss": 0.2754,
+      "loss": 0.2752,
       "step": 286
     },
     {
       "epoch": 0.18368,
-      "grad_norm": 0.3333201828315899,
+      "grad_norm": 0.33183064717108757,
       "learning_rate": 4.7634637369237526e-05,
-      "loss": 0.2776,
+      "loss": 0.2772,
       "step": 287
     },
     {
       "epoch": 0.18432,
-      "grad_norm": 0.3504320913830548,
+      "grad_norm": 0.3408606960521383,
       "learning_rate": 4.7612100245353045e-05,
-      "loss": 0.2874,
+      "loss": 0.288,
       "step": 288
     },
     {
       "epoch": 0.18496,
-      "grad_norm": 0.34297542367165534,
+      "grad_norm": 0.35025830647742257,
       "learning_rate": 4.758946164667347e-05,
-      "loss": 0.2915,
+      "loss": 0.2908,
       "step": 289
     },
     {
       "epoch": 0.1856,
-      "grad_norm": 0.37229747511831845,
+      "grad_norm": 0.385301045406101,
       "learning_rate": 4.7566721674792515e-05,
-      "loss": 0.2782,
+      "loss": 0.279,
       "step": 290
     },
     {
       "epoch": 0.18624,
-      "grad_norm": 0.3317199247294434,
+      "grad_norm": 0.3365158470714505,
       "learning_rate": 4.7543880431758805e-05,
-      "loss": 0.2782,
+      "loss": 0.2785,
       "step": 291
     },
     {
       "epoch": 0.18688,
-      "grad_norm": 0.3375135597374568,
+      "grad_norm": 0.3520029833034493,
       "learning_rate": 4.752093802007543e-05,
-      "loss": 0.3017,
+      "loss": 0.3015,
       "step": 292
     },
     {
       "epoch": 0.18752,
-      "grad_norm": 0.3006592553400984,
+      "grad_norm": 0.30981353591353944,
       "learning_rate": 4.7497894542699534e-05,
       "loss": 0.2621,
       "step": 293
     },
     {
       "epoch": 0.18816,
-      "grad_norm": 0.3457799673936055,
+      "grad_norm": 0.33678488047130234,
       "learning_rate": 4.747475010304175e-05,
-      "loss": 0.2814,
+      "loss": 0.2809,
       "step": 294
     },
     {
       "epoch": 0.1888,
-      "grad_norm": 0.30139639958651177,
+      "grad_norm": 0.30138038658401767,
       "learning_rate": 4.745150480496583e-05,
-      "loss": 0.2963,
+      "loss": 0.2968,
       "step": 295
     },
     {
       "epoch": 0.18944,
-      "grad_norm": 0.3165095553552411,
+      "grad_norm": 0.32368771492417925,
       "learning_rate": 4.742815875278811e-05,
-      "loss": 0.295,
+      "loss": 0.2947,
       "step": 296
     },
     {
       "epoch": 0.19008,
-      "grad_norm": 0.49950391816814954,
+      "grad_norm": 0.3523596181548406,
       "learning_rate": 4.7404712051277115e-05,
-      "loss": 0.2564,
+      "loss": 0.2563,
       "step": 297
     },
     {
       "epoch": 0.19072,
-      "grad_norm": 0.4675033045283917,
+      "grad_norm": 0.3476373280918836,
       "learning_rate": 4.738116480565301e-05,
-      "loss": 0.2955,
+      "loss": 0.2954,
       "step": 298
     },
     {
       "epoch": 0.19136,
-      "grad_norm": 0.320627703534577,
+      "grad_norm": 0.3305958149949148,
       "learning_rate": 4.7357517121587183e-05,
-      "loss": 0.2793,
+      "loss": 0.2792,
       "step": 299
     },
     {
       "epoch": 0.192,
-      "grad_norm": 0.30638839622164654,
+      "grad_norm": 0.3260613594105097,
       "learning_rate": 4.7333769105201735e-05,
-      "loss": 0.2433,
+      "loss": 0.2435,
       "step": 300
     },
     {
       "epoch": 0.19264,
-      "grad_norm": 0.33993107017240015,
+      "grad_norm": 0.35007749460908505,
       "learning_rate": 4.730992086306906e-05,
       "loss": 0.2839,
       "step": 301
     },
     {
       "epoch": 0.19328,
-      "grad_norm": 0.31266754895882487,
+      "grad_norm": 0.3124453815680245,
       "learning_rate": 4.728597250221129e-05,
-      "loss": 0.2838,
+      "loss": 0.2835,
       "step": 302
     },
     {
       "epoch": 0.19392,
-      "grad_norm": 0.3333824096931821,
+      "grad_norm": 0.32999884470742785,
       "learning_rate": 4.726192413009986e-05,
-      "loss": 0.2956,
+      "loss": 0.2962,
       "step": 303
     },
     {
       "epoch": 0.19456,
-      "grad_norm": 0.31668134967973227,
+      "grad_norm": 0.3238134773271861,
       "learning_rate": 4.723777585465503e-05,
-      "loss": 0.2943,
+      "loss": 0.2938,
       "step": 304
     },
     {
       "epoch": 0.1952,
-      "grad_norm": 0.3516833949768645,
+      "grad_norm": 0.3561001066562547,
       "learning_rate": 4.7213527784245395e-05,
-      "loss": 0.2778,
+      "loss": 0.2772,
       "step": 305
     },
     {
       "epoch": 0.19584,
-      "grad_norm": 0.3221013371360776,
+      "grad_norm": 0.3207708243138028,
       "learning_rate": 4.718918002768738e-05,
-      "loss": 0.2775,
+      "loss": 0.2769,
       "step": 306
     },
     {
       "epoch": 0.19648,
-      "grad_norm": 0.2912869296710327,
+      "grad_norm": 0.2923323638292908,
       "learning_rate": 4.716473269424476e-05,
-      "loss": 0.2838,
+      "loss": 0.2842,
       "step": 307
     },
     {
       "epoch": 0.19712,
-      "grad_norm": 0.3063238267835587,
+      "grad_norm": 0.312740800108901,
       "learning_rate": 4.714018589362818e-05,
-      "loss": 0.3116,
+      "loss": 0.3113,
       "step": 308
     },
     {
       "epoch": 0.19776,
-      "grad_norm": 0.2678309458069726,
+      "grad_norm": 0.2764921423909185,
       "learning_rate": 4.7115539735994666e-05,
-      "loss": 0.2741,
+      "loss": 0.2735,
       "step": 309
     },
     {
       "epoch": 0.1984,
-      "grad_norm": 0.3203497385070914,
+      "grad_norm": 0.33245051095638717,
       "learning_rate": 4.709079433194712e-05,
-      "loss": 0.278,
+      "loss": 0.2783,
       "step": 310
     },
     {
       "epoch": 0.19904,
-      "grad_norm": 0.2886560865479684,
+      "grad_norm": 0.2883468898345436,
       "learning_rate": 4.7065949792533804e-05,
-      "loss": 0.2579,
+      "loss": 0.257,
       "step": 311
     },
     {
       "epoch": 0.19968,
-      "grad_norm": 0.34979254223314304,
+      "grad_norm": 0.3635737080055347,
       "learning_rate": 4.70410062292479e-05,
-      "loss": 0.2943,
+      "loss": 0.2933,
       "step": 312
     },
     {
       "epoch": 0.20032,
-      "grad_norm": 0.30129370691362123,
+      "grad_norm": 0.3254742482108127,
       "learning_rate": 4.701596375402694e-05,
-      "loss": 0.2935,
+      "loss": 0.2933,
       "step": 313
     },
     {
       "epoch": 0.20096,
-      "grad_norm": 0.3028194799255164,
+      "grad_norm": 0.30265278788416133,
       "learning_rate": 4.699082247925234e-05,
-      "loss": 0.3014,
+      "loss": 0.3004,
       "step": 314
     },
     {
       "epoch": 0.2016,
-      "grad_norm": 0.3245596740386509,
+      "grad_norm": 0.3229708540790062,
       "learning_rate": 4.6965582517748923e-05,
-      "loss": 0.2938,
+      "loss": 0.2935,
       "step": 315
     },
     {
       "epoch": 0.20224,
-      "grad_norm": 0.35740876844794395,
+      "grad_norm": 0.2955440929603968,
       "learning_rate": 4.694024398278434e-05,
-      "loss": 0.2888,
+      "loss": 0.2889,
       "step": 316
     },
     {
       "epoch": 0.20288,
-      "grad_norm": 0.3378708759886966,
+      "grad_norm": 0.32188591609611733,
       "learning_rate": 4.691480698806864e-05,
-      "loss": 0.3004,
+      "loss": 0.3002,
       "step": 317
     },
     {
       "epoch": 0.20352,
-      "grad_norm": 0.2824072245359557,
+      "grad_norm": 0.27836417438079064,
       "learning_rate": 4.6889271647753694e-05,
       "loss": 0.2641,
       "step": 318
     },
     {
       "epoch": 0.20416,
-      "grad_norm": 0.3174107262885623,
+      "grad_norm": 0.30503117060048834,
       "learning_rate": 4.686363807643272e-05,
-      "loss": 0.2558,
+      "loss": 0.2557,
       "step": 319
     },
     {
       "epoch": 0.2048,
-      "grad_norm": 0.3176431256834232,
+      "grad_norm": 0.31391016617770173,
       "learning_rate": 4.683790638913979e-05,
-      "loss": 0.301,
+      "loss": 0.3008,
       "step": 320
     },
     {
       "epoch": 0.20544,
-      "grad_norm": 0.3001231365631309,
+      "grad_norm": 0.30297458481431416,
       "learning_rate": 4.681207670134923e-05,
-      "loss": 0.285,
+      "loss": 0.2855,
       "step": 321
     },
     {
       "epoch": 0.20608,
-      "grad_norm": 0.2901884788933691,
+      "grad_norm": 0.31472600754542884,
       "learning_rate": 4.6786149128975196e-05,
-      "loss": 0.2787,
+      "loss": 0.2775,
       "step": 322
     },
     {
       "epoch": 0.20672,
-      "grad_norm": 0.2954537995473784,
+      "grad_norm": 0.2973831240453251,
       "learning_rate": 4.6760123788371114e-05,
-      "loss": 0.2827,
+      "loss": 0.2822,
       "step": 323
     },
     {
       "epoch": 0.20736,
-      "grad_norm": 0.3101653694159506,
+      "grad_norm": 0.3325176485448801,
       "learning_rate": 4.6734000796329125e-05,
-      "loss": 0.2848,
+      "loss": 0.2844,
       "step": 324
     },
     {
       "epoch": 0.208,
-      "grad_norm": 0.2874105687683901,
+      "grad_norm": 0.2781447146189205,
       "learning_rate": 4.670778027007964e-05,
-      "loss": 0.2973,
+      "loss": 0.2972,
       "step": 325
     },
     {
       "epoch": 0.20864,
-      "grad_norm": 0.3620198319608371,
+      "grad_norm": 0.35102418370987387,
       "learning_rate": 4.6681462327290725e-05,
-      "loss": 0.2675,
+      "loss": 0.2674,
       "step": 326
     },
     {
       "epoch": 0.20928,
-      "grad_norm": 0.3162114156351334,
+      "grad_norm": 0.3031132755291398,
       "learning_rate": 4.6655047086067636e-05,
-      "loss": 0.2788,
+      "loss": 0.2783,
       "step": 327
     },
     {
       "epoch": 0.20992,
-      "grad_norm": 0.30972198988187555,
+      "grad_norm": 0.2914541802538774,
       "learning_rate": 4.662853466495229e-05,
-      "loss": 0.2714,
+      "loss": 0.2718,
       "step": 328
     },
     {
       "epoch": 0.21056,
-      "grad_norm": 0.2972071351365883,
+      "grad_norm": 0.2845108849079057,
       "learning_rate": 4.660192518292266e-05,
-      "loss": 0.2704,
+      "loss": 0.2701,
       "step": 329
     },
     {
       "epoch": 0.2112,
-      "grad_norm": 0.4583399375330476,
+      "grad_norm": 0.30612752816767386,
       "learning_rate": 4.6575218759392336e-05,
       "loss": 0.2706,
       "step": 330
     },
     {
       "epoch": 0.21184,
-      "grad_norm": 0.29570973103502374,
+      "grad_norm": 0.3343001012996569,
       "learning_rate": 4.654841551420993e-05,
-      "loss": 0.2591,
+      "loss": 0.2592,
       "step": 331
     },
     {
       "epoch": 0.21248,
-      "grad_norm": 0.30593116338066634,
+      "grad_norm": 0.28779385954712927,
       "learning_rate": 4.6521515567658557e-05,
-      "loss": 0.268,
+      "loss": 0.2673,
       "step": 332
     },
     {
       "epoch": 0.21312,
-      "grad_norm": 0.3260999005002705,
+      "grad_norm": 0.3130712333478661,
       "learning_rate": 4.6494519040455275e-05,
-      "loss": 0.2799,
+      "loss": 0.2779,
       "step": 333
     },
     {
       "epoch": 0.21376,
-      "grad_norm": 0.2968856649158912,
+      "grad_norm": 0.29656018238823806,
       "learning_rate": 4.6467426053750596e-05,
-      "loss": 0.2642,
+      "loss": 0.2634,
       "step": 334
     },
     {
       "epoch": 0.2144,
-      "grad_norm": 0.29642619891446786,
+      "grad_norm": 0.29429809703015725,
       "learning_rate": 4.644023672912788e-05,
-      "loss": 0.2777,
+      "loss": 0.2771,
       "step": 335
     },
     {
       "epoch": 0.21504,
-      "grad_norm": 0.30350566844855414,
+      "grad_norm": 0.3092350735511281,
       "learning_rate": 4.641295118860281e-05,
-      "loss": 0.2879,
+      "loss": 0.2872,
       "step": 336
     },
     {
       "epoch": 0.21568,
-      "grad_norm": 0.26464623569777385,
+      "grad_norm": 0.25928563534355026,
       "learning_rate": 4.638556955462289e-05,
-      "loss": 0.2608,
+      "loss": 0.2602,
       "step": 337
     },
     {
       "epoch": 0.21632,
-      "grad_norm": 0.331526482181387,
+      "grad_norm": 0.32248922953258674,
       "learning_rate": 4.635809195006681e-05,
-      "loss": 0.2915,
+      "loss": 0.2909,
       "step": 338
     },
     {
       "epoch": 0.21696,
-      "grad_norm": 0.31249619783308275,
+      "grad_norm": 0.3066112074635602,
       "learning_rate": 4.633051849824396e-05,
-      "loss": 0.2587,
+      "loss": 0.2579,
       "step": 339
     },
     {
       "epoch": 0.2176,
-      "grad_norm": 0.3342213823905452,
+      "grad_norm": 0.34512498246286594,
       "learning_rate": 4.630284932289386e-05,
-      "loss": 0.2861,
+      "loss": 0.2853,
       "step": 340
     },
     {
       "epoch": 0.21824,
-      "grad_norm": 0.3443174294305122,
+      "grad_norm": 0.32772544741225995,
       "learning_rate": 4.6275084548185605e-05,
-      "loss": 0.2951,
+      "loss": 0.2936,
       "step": 341
     },
     {
       "epoch": 0.21888,
-      "grad_norm": 0.32706429001657567,
+      "grad_norm": 0.30638642844295744,
       "learning_rate": 4.6247224298717294e-05,
-      "loss": 0.2555,
+      "loss": 0.255,
       "step": 342
     },
     {
       "epoch": 0.21952,
-      "grad_norm": 0.3059927880992944,
+      "grad_norm": 0.2948939777755622,
       "learning_rate": 4.621926869951549e-05,
-      "loss": 0.246,
+      "loss": 0.2461,
       "step": 343
     },
     {
       "epoch": 0.22016,
-      "grad_norm": 0.3825724352839679,
+      "grad_norm": 0.3571814132315835,
       "learning_rate": 4.619121787603464e-05,
-      "loss": 0.2768,
+      "loss": 0.2749,
       "step": 344
     },
     {
       "epoch": 0.2208,
-      "grad_norm": 0.31461054559886714,
+      "grad_norm": 0.3019198830803027,
       "learning_rate": 4.6163071954156546e-05,
-      "loss": 0.2765,
+      "loss": 0.2764,
       "step": 345
     },
     {
       "epoch": 0.22144,
-      "grad_norm": 0.30279471004384817,
+      "grad_norm": 0.2947039259040435,
       "learning_rate": 4.6134831060189743e-05,
-      "loss": 0.283,
+      "loss": 0.2824,
       "step": 346
     },
     {
       "epoch": 0.22208,
-      "grad_norm": 0.3131415453093515,
+      "grad_norm": 0.311517126488146,
       "learning_rate": 4.610649532086899e-05,
-      "loss": 0.2738,
+      "loss": 0.273,
       "step": 347
     },
     {
       "epoch": 0.22272,
-      "grad_norm": 0.34626214360910446,
+      "grad_norm": 0.3202495311575204,
       "learning_rate": 4.607806486335468e-05,
-      "loss": 0.2899,
+      "loss": 0.2892,
       "step": 348
     },
     {
       "epoch": 0.22336,
-      "grad_norm": 0.3000462942424186,
+      "grad_norm": 0.29593870703206343,
       "learning_rate": 4.604953981523225e-05,
-      "loss": 0.288,
+      "loss": 0.2874,
       "step": 349
     },
     {
       "epoch": 0.224,
-      "grad_norm": 0.3076604165950532,
+      "grad_norm": 0.3046217268454308,
       "learning_rate": 4.602092030451162e-05,
-      "loss": 0.2707,
+      "loss": 0.2697,
       "step": 350
     },
     {
       "epoch": 0.22464,
-      "grad_norm": 0.3224147646178712,
+      "grad_norm": 0.32393845693805695,
       "learning_rate": 4.599220645962665e-05,
-      "loss": 0.2867,
+      "loss": 0.2857,
       "step": 351
     },
     {
       "epoch": 0.22528,
-      "grad_norm": 0.29505276608299713,
+      "grad_norm": 0.3054582480479615,
       "learning_rate": 4.596339840943452e-05,
-      "loss": 0.2688,
+      "loss": 0.2681,
       "step": 352
     },
     {
       "epoch": 0.22592,
-      "grad_norm": 0.2820029386723236,
+      "grad_norm": 0.28840335456316024,
       "learning_rate": 4.5934496283215164e-05,
-      "loss": 0.2675,
+      "loss": 0.2669,
       "step": 353
     },
     {
       "epoch": 0.22656,
-      "grad_norm": 0.2937811078323571,
+      "grad_norm": 0.2773362865048385,
       "learning_rate": 4.590550021067072e-05,
-      "loss": 0.2586,
+      "loss": 0.2579,
       "step": 354
     },
     {
       "epoch": 0.2272,
-      "grad_norm": 0.3089230579758364,
+      "grad_norm": 0.29887466103776417,
       "learning_rate": 4.587641032192488e-05,
-      "loss": 0.2678,
+      "loss": 0.2668,
       "step": 355
     },
     {
       "epoch": 0.22784,
-      "grad_norm": 0.3141525256276034,
+      "grad_norm": 0.3307188902703665,
       "learning_rate": 4.58472267475224e-05,
-      "loss": 0.2828,
+      "loss": 0.2819,
       "step": 356
     },
     {
       "epoch": 0.22848,
-      "grad_norm": 0.3221425300292192,
+      "grad_norm": 0.3390971905379334,
       "learning_rate": 4.5817949618428405e-05,
-      "loss": 0.2684,
+      "loss": 0.2678,
       "step": 357
     },
     {
       "epoch": 0.22912,
-      "grad_norm": 0.3081385245871579,
+      "grad_norm": 0.3013575293157085,
       "learning_rate": 4.578857906602791e-05,
-      "loss": 0.2686,
+      "loss": 0.2682,
       "step": 358
     },
     {
       "epoch": 0.22976,
-      "grad_norm": 0.29474067735994697,
+      "grad_norm": 0.3013139799493687,
       "learning_rate": 4.575911522212516e-05,
-      "loss": 0.2853,
+      "loss": 0.2854,
       "step": 359
     },
     {
       "epoch": 0.2304,
-      "grad_norm": 0.26872460388574704,
+      "grad_norm": 0.27853806891578586,
       "learning_rate": 4.572955821894305e-05,
-      "loss": 0.271,
+      "loss": 0.2708,
       "step": 360
     },
     {
       "epoch": 0.23104,
-      "grad_norm": 0.2795986689368969,
+      "grad_norm": 0.27288786971452156,
       "learning_rate": 4.5699908189122556e-05,
-      "loss": 0.237,
+      "loss": 0.2365,
       "step": 361
     },
     {
       "epoch": 0.23168,
-      "grad_norm": 0.3164595478830514,
+      "grad_norm": 0.33213885368239576,
       "learning_rate": 4.5670165265722096e-05,
-      "loss": 0.2575,
+      "loss": 0.2571,
       "step": 362
     },
     {
       "epoch": 0.23232,
-      "grad_norm": 0.3195893610809849,
+      "grad_norm": 0.3316342870811453,
       "learning_rate": 4.5640329582217004e-05,
-      "loss": 0.2857,
+      "loss": 0.2859,
       "step": 363
     },
     {
       "epoch": 0.23296,
-      "grad_norm": 0.31680454787494927,
+      "grad_norm": 0.31718513233843937,
       "learning_rate": 4.5610401272498836e-05,
-      "loss": 0.2554,
+      "loss": 0.255,
       "step": 364
     },
     {
       "epoch": 0.2336,
-      "grad_norm": 0.28146209486053303,
+      "grad_norm": 0.3003964869956768,
       "learning_rate": 4.558038047087486e-05,
-      "loss": 0.2643,
+      "loss": 0.2638,
       "step": 365
     },
     {
       "epoch": 0.23424,
-      "grad_norm": 0.30175920777877385,
+      "grad_norm": 0.3113573197668127,
       "learning_rate": 4.555026731206741e-05,
       "loss": 0.2492,
       "step": 366
     },
     {
       "epoch": 0.23488,
-      "grad_norm": 0.29436533992649466,
+      "grad_norm": 0.2870220952890737,
       "learning_rate": 4.5520061931213255e-05,
-      "loss": 0.2793,
+      "loss": 0.2789,
       "step": 367
     },
     {
       "epoch": 0.23552,
-      "grad_norm": 0.2829560308292762,
+      "grad_norm": 0.27167839521619286,
       "learning_rate": 4.548976446386305e-05,
-      "loss": 0.274,
+      "loss": 0.2735,
       "step": 368
     },
     {
       "epoch": 0.23616,
-      "grad_norm": 0.29408627346856103,
+      "grad_norm": 0.285331504835825,
       "learning_rate": 4.545937504598069e-05,
-      "loss": 0.2728,
+      "loss": 0.2723,
       "step": 369
     },
     {
       "epoch": 0.2368,
-      "grad_norm": 0.2889978012340055,
+      "grad_norm": 0.27918688536225067,
       "learning_rate": 4.5428893813942727e-05,
-      "loss": 0.2692,
+      "loss": 0.2685,
       "step": 370
     },
     {
       "epoch": 0.23744,
-      "grad_norm": 0.27148243923702725,
+      "grad_norm": 0.2668837009377814,
       "learning_rate": 4.539832090453771e-05,
       "loss": 0.2655,
       "step": 371
     },
     {
       "epoch": 0.23808,
-      "grad_norm": 0.31050739584394577,
+      "grad_norm": 0.3044254239966202,
       "learning_rate": 4.536765645496564e-05,
-      "loss": 0.262,
+      "loss": 0.2617,
       "step": 372
     },
     {
       "epoch": 0.23872,
-      "grad_norm": 0.2996079504786343,
+      "grad_norm": 0.284750804617016,
       "learning_rate": 4.533690060283727e-05,
-      "loss": 0.2662,
+      "loss": 0.2663,
       "step": 373
     },
     {
       "epoch": 0.23936,
-      "grad_norm": 0.2800563400226751,
+      "grad_norm": 0.28718270627176934,
       "learning_rate": 4.530605348617357e-05,
-      "loss": 0.275,
+      "loss": 0.2752,
       "step": 374
     },
     {
       "epoch": 0.24,
-      "grad_norm": 0.31562058411792526,
+      "grad_norm": 0.29634840394464135,
       "learning_rate": 4.527511524340508e-05,
-      "loss": 0.2656,
+      "loss": 0.2645,
       "step": 375
     },
     {
       "epoch": 0.24064,
-      "grad_norm": 0.3090111413770133,
+      "grad_norm": 0.3012232953009457,
       "learning_rate": 4.5244086013371246e-05,
-      "loss": 0.2646,
+      "loss": 0.2641,
       "step": 376
     },
     {
       "epoch": 0.24128,
-      "grad_norm": 0.30489397336811647,
+      "grad_norm": 0.3023357075993031,
       "learning_rate": 4.521296593531985e-05,
-      "loss": 0.2624,
+      "loss": 0.2618,
       "step": 377
     },
     {
       "epoch": 0.24192,
-      "grad_norm": 0.29221796413694495,
+      "grad_norm": 0.27441105527916904,
       "learning_rate": 4.518175514890636e-05,
-      "loss": 0.2656,
+      "loss": 0.2663,
       "step": 378
     },
     {
       "epoch": 0.24256,
-      "grad_norm": 0.33233722630021206,
+      "grad_norm": 0.3325569029148721,
       "learning_rate": 4.515045379419332e-05,
-      "loss": 0.2803,
+      "loss": 0.2795,
       "step": 379
     },
     {
       "epoch": 0.2432,
-      "grad_norm": 0.2999774916809297,
+      "grad_norm": 0.30652258454129583,
       "learning_rate": 4.5119062011649714e-05,
-      "loss": 0.275,
+      "loss": 0.2756,
       "step": 380
     },
     {
       "epoch": 0.24384,
-      "grad_norm": 0.3355386902700075,
+      "grad_norm": 0.33712951361660654,
       "learning_rate": 4.508757994215031e-05,
-      "loss": 0.2807,
+      "loss": 0.2806,
       "step": 381
     },
     {
       "epoch": 0.24448,
-      "grad_norm": 0.35909608718324937,
+      "grad_norm": 0.3426531131298772,
       "learning_rate": 4.505600772697507e-05,
-      "loss": 0.2789,
+      "loss": 0.2782,
       "step": 382
     },
     {
       "epoch": 0.24512,
-      "grad_norm": 0.3388616226576712,
+      "grad_norm": 0.33303701942677855,
       "learning_rate": 4.5024345507808495e-05,
-      "loss": 0.2665,
+      "loss": 0.2662,
       "step": 383
     },
     {
       "epoch": 0.24576,
-      "grad_norm": 0.3402302385955381,
+      "grad_norm": 0.31791526383279123,
       "learning_rate": 4.499259342673898e-05,
-      "loss": 0.2766,
+      "loss": 0.2768,
       "step": 384
     },
     {
       "epoch": 0.2464,
-      "grad_norm": 0.3134053278907894,
+      "grad_norm": 0.30503735620868655,
       "learning_rate": 4.4960751626258205e-05,
-      "loss": 0.2637,
+      "loss": 0.2642,
       "step": 385
     },
     {
       "epoch": 0.24704,
-      "grad_norm": 0.34108714934521905,
+      "grad_norm": 0.3312474346993656,
       "learning_rate": 4.492882024926046e-05,
       "loss": 0.2732,
       "step": 386
     },
     {
       "epoch": 0.24768,
-      "grad_norm": 0.322628980519524,
+      "grad_norm": 0.3133944170389108,
       "learning_rate": 4.489679943904203e-05,
-      "loss": 0.2942,
+      "loss": 0.2933,
       "step": 387
     },
     {
       "epoch": 0.24832,
-      "grad_norm": 0.2872634294452658,
+      "grad_norm": 0.2852787428119327,
       "learning_rate": 4.486468933930055e-05,
-      "loss": 0.2809,
+      "loss": 0.2806,
       "step": 388
     },
     {
       "epoch": 0.24896,
-      "grad_norm": 0.3112032329339079,
+      "grad_norm": 0.3092836612089024,
       "learning_rate": 4.483249009413433e-05,
-      "loss": 0.2596,
+      "loss": 0.2588,
       "step": 389
     },
     {
       "epoch": 0.2496,
-      "grad_norm": 0.3091772617757821,
+      "grad_norm": 0.3112336325754041,
       "learning_rate": 4.4800201848041765e-05,
-      "loss": 0.2895,
+      "loss": 0.2889,
       "step": 390
     },
     {
       "epoch": 0.25024,
-      "grad_norm": 0.3162533728810715,
+      "grad_norm": 0.3125491047833594,
       "learning_rate": 4.476782474592062e-05,
-      "loss": 0.2629,
+      "loss": 0.2628,
       "step": 391
     },
     {
       "epoch": 0.25088,
-      "grad_norm": 0.29776738506438605,
+      "grad_norm": 0.2912676631270001,
       "learning_rate": 4.473535893306744e-05,
-      "loss": 0.269,
+      "loss": 0.2696,
       "step": 392
     },
     {
       "epoch": 0.25152,
-      "grad_norm": 0.29061925034244496,
+      "grad_norm": 0.28628694727112053,
       "learning_rate": 4.470280455517687e-05,
-      "loss": 0.2621,
+      "loss": 0.2619,
       "step": 393
     },
     {
       "epoch": 0.25216,
-      "grad_norm": 0.32125404165323423,
+      "grad_norm": 0.3293686387916455,
       "learning_rate": 4.4670161758340966e-05,
       "loss": 0.2739,
       "step": 394
     },
     {
       "epoch": 0.2528,
-      "grad_norm": 0.32765191926742254,
+      "grad_norm": 0.31348173029121257,
       "learning_rate": 4.4637430689048634e-05,
-      "loss": 0.2603,
+      "loss": 0.2599,
       "step": 395
     },
     {
       "epoch": 0.25344,
-      "grad_norm": 0.3248426091429505,
+      "grad_norm": 0.32951196314360054,
       "learning_rate": 4.460461149418486e-05,
-      "loss": 0.2562,
+      "loss": 0.2567,
       "step": 396
     },
     {
       "epoch": 0.25408,
-      "grad_norm": 0.3293450138223353,
+      "grad_norm": 0.34019629325427525,
       "learning_rate": 4.4571704321030125e-05,
-      "loss": 0.2909,
+      "loss": 0.2913,
       "step": 397
     },
     {
       "epoch": 0.25472,
-      "grad_norm": 0.3268103816628839,
+      "grad_norm": 0.3301120643416949,
       "learning_rate": 4.4538709317259754e-05,
-      "loss": 0.2476,
+      "loss": 0.2472,
       "step": 398
     },
     {
       "epoch": 0.25536,
-      "grad_norm": 0.28789034131409996,
+      "grad_norm": 0.29742911532500216,
       "learning_rate": 4.4505626630943174e-05,
-      "loss": 0.2615,
+      "loss": 0.2617,
       "step": 399
     },
     {
       "epoch": 0.256,
-      "grad_norm": 0.3326450624021029,
+      "grad_norm": 0.3417634987932657,
       "learning_rate": 4.447245641054334e-05,
-      "loss": 0.2741,
+      "loss": 0.2742,
       "step": 400
     },
     {
       "epoch": 0.25664,
-      "grad_norm": 0.2849399630023341,
+      "grad_norm": 0.2826147129731043,
       "learning_rate": 4.443919880491598e-05,
-      "loss": 0.2526,
+      "loss": 0.2535,
       "step": 401
     },
     {
       "epoch": 0.25728,
-      "grad_norm": 0.28894563785283384,
+      "grad_norm": 0.29947144671717485,
       "learning_rate": 4.440585396330904e-05,
       "loss": 0.2688,
       "step": 402
     },
     {
       "epoch": 0.25792,
-      "grad_norm": 0.3197497488847442,
+      "grad_norm": 0.31610077864520053,
       "learning_rate": 4.437242203536189e-05,
-      "loss": 0.2728,
+      "loss": 0.2724,
       "step": 403
     },
     {
       "epoch": 0.25856,
-      "grad_norm": 0.3037447602492967,
+      "grad_norm": 0.3137992831884045,
       "learning_rate": 4.433890317110475e-05,
-      "loss": 0.2839,
+      "loss": 0.2848,
       "step": 404
     },
     {
       "epoch": 0.2592,
-      "grad_norm": 0.29041308751589046,
+      "grad_norm": 0.2796187619572129,
       "learning_rate": 4.430529752095795e-05,
-      "loss": 0.2712,
+      "loss": 0.2714,
       "step": 405
     },
     {
       "epoch": 0.25984,
-      "grad_norm": 0.29621099818039576,
+      "grad_norm": 0.2912248245493093,
       "learning_rate": 4.427160523573131e-05,
-      "loss": 0.2801,
+      "loss": 0.2795,
       "step": 406
     },
     {
       "epoch": 0.26048,
-      "grad_norm": 0.29524960348391543,
+      "grad_norm": 0.2980162603474383,
       "learning_rate": 4.4237826466623406e-05,
-      "loss": 0.2871,
+      "loss": 0.2867,
       "step": 407
     },
     {
       "epoch": 0.26112,
-      "grad_norm": 0.2966251537463695,
+      "grad_norm": 0.29652170821748175,
       "learning_rate": 4.420396136522095e-05,
-      "loss": 0.2728,
+      "loss": 0.2727,
       "step": 408
     },
     {
       "epoch": 0.26176,
-      "grad_norm": 0.2875766541626561,
+      "grad_norm": 0.2915563003617483,
       "learning_rate": 4.417001008349807e-05,
-      "loss": 0.2585,
+      "loss": 0.2583,
       "step": 409
     },
     {
       "epoch": 0.2624,
-      "grad_norm": 0.3113316229409783,
+      "grad_norm": 0.3225628807617719,
       "learning_rate": 4.413597277381564e-05,
-      "loss": 0.2707,
+      "loss": 0.271,
       "step": 410
     },
     {
       "epoch": 0.26304,
-      "grad_norm": 0.2811939335957826,
+      "grad_norm": 0.28870166282131143,
       "learning_rate": 4.410184958892059e-05,
-      "loss": 0.2655,
+      "loss": 0.2663,
       "step": 411
     },
     {
       "epoch": 0.26368,
-      "grad_norm": 0.29793389103823087,
+      "grad_norm": 0.29553450210347276,
       "learning_rate": 4.4067640681945244e-05,
-      "loss": 0.2637,
+      "loss": 0.2639,
       "step": 412
     },
     {
       "epoch": 0.26432,
-      "grad_norm": 0.3448593520224143,
+      "grad_norm": 0.3332862277579268,
       "learning_rate": 4.403334620640659e-05,
-      "loss": 0.2603,
+      "loss": 0.2609,
       "step": 413
     },
     {
       "epoch": 0.26496,
-      "grad_norm": 0.3031638665356794,
+      "grad_norm": 0.30107711275005455,
       "learning_rate": 4.3998966316205636e-05,
-      "loss": 0.2538,
+      "loss": 0.2541,
       "step": 414
     },
     {
       "epoch": 0.2656,
-      "grad_norm": 0.2532330310355455,
+      "grad_norm": 0.2591023620947492,
       "learning_rate": 4.396450116562669e-05,
-      "loss": 0.2453,
+      "loss": 0.2451,
       "step": 415
     },
     {
       "epoch": 0.26624,
-      "grad_norm": 0.27984325783985575,
+      "grad_norm": 0.27765150058129984,
       "learning_rate": 4.392995090933669e-05,
-      "loss": 0.2527,
+      "loss": 0.252,
       "step": 416
     },
     {
       "epoch": 0.26688,
-      "grad_norm": 0.2853666077902048,
+      "grad_norm": 0.27831462093368287,
       "learning_rate": 4.389531570238448e-05,
-      "loss": 0.2419,
+      "loss": 0.241,
       "step": 417
     },
     {
       "epoch": 0.26752,
-      "grad_norm": 0.34087513199677344,
+      "grad_norm": 0.3393673672608069,
       "learning_rate": 4.386059570020015e-05,
-      "loss": 0.2782,
+      "loss": 0.2781,
       "step": 418
     },
     {
       "epoch": 0.26816,
-      "grad_norm": 0.27330226400607055,
+      "grad_norm": 0.2709067486737786,
       "learning_rate": 4.382579105859428e-05,
-      "loss": 0.2739,
+      "loss": 0.2746,
       "step": 419
     },
     {
       "epoch": 0.2688,
-      "grad_norm": 0.31913954123535876,
+      "grad_norm": 0.31024807728799664,
       "learning_rate": 4.379090193375735e-05,
-      "loss": 0.2685,
+      "loss": 0.269,
       "step": 420
     },
     {
       "epoch": 0.26944,
-      "grad_norm": 0.28177893210943633,
+      "grad_norm": 0.29437773626601327,
       "learning_rate": 4.375592848225889e-05,
-      "loss": 0.2494,
+      "loss": 0.2497,
       "step": 421
     },
     {
       "epoch": 0.27008,
-      "grad_norm": 0.28222647654040717,
+      "grad_norm": 0.27249340717742593,
       "learning_rate": 4.372087086104692e-05,
       "loss": 0.2769,
       "step": 422
     },
     {
       "epoch": 0.27072,
-      "grad_norm": 0.30927952594934816,
+      "grad_norm": 0.3182175494652926,
       "learning_rate": 4.368572922744715e-05,
-      "loss": 0.2708,
+      "loss": 0.2709,
       "step": 423
     },
     {
       "epoch": 0.27136,
-      "grad_norm": 0.2949859599241957,
+      "grad_norm": 0.2906720843900486,
       "learning_rate": 4.365050373916233e-05,
-      "loss": 0.2753,
+      "loss": 0.275,
       "step": 424
     },
     {
       "epoch": 0.272,
-      "grad_norm": 0.29610748060111175,
+      "grad_norm": 0.3098393205417728,
       "learning_rate": 4.3615194554271485e-05,
-      "loss": 0.2335,
+      "loss": 0.234,
       "step": 425
     },
     {
       "epoch": 0.27264,
-      "grad_norm": 0.3034357877644478,
+      "grad_norm": 0.30226227188570165,
       "learning_rate": 4.357980183122927e-05,
-      "loss": 0.2726,
+      "loss": 0.2728,
       "step": 426
     },
     {
       "epoch": 0.27328,
-      "grad_norm": 0.3059352582010788,
+      "grad_norm": 0.3101806085164065,
       "learning_rate": 4.354432572886522e-05,
-      "loss": 0.271,
+      "loss": 0.272,
       "step": 427
     },
     {
       "epoch": 0.27392,
-      "grad_norm": 0.2771240245917802,
+      "grad_norm": 0.28245936335261307,
       "learning_rate": 4.3508766406383036e-05,
-      "loss": 0.2784,
+      "loss": 0.278,
       "step": 428
     },
     {
       "epoch": 0.27456,
-      "grad_norm": 0.27908141628363836,
+      "grad_norm": 0.27797054709506586,
       "learning_rate": 4.347312402335989e-05,
-      "loss": 0.2685,
+      "loss": 0.2684,
       "step": 429
     },
     {
       "epoch": 0.2752,
-      "grad_norm": 0.28707815382315055,
+      "grad_norm": 0.2990500240447701,
       "learning_rate": 4.34373987397457e-05,
       "loss": 0.278,
       "step": 430
     },
     {
       "epoch": 0.27584,
-      "grad_norm": 0.2752693747267436,
+      "grad_norm": 0.2727197517887389,
       "learning_rate": 4.340159071586241e-05,
-      "loss": 0.2447,
+      "loss": 0.2454,
       "step": 431
     },
     {
       "epoch": 0.27648,
-      "grad_norm": 0.28362003524505025,
+      "grad_norm": 0.30243924092224234,
       "learning_rate": 4.336570011240326e-05,
-      "loss": 0.2457,
+      "loss": 0.2455,
       "step": 432
     },
     {
       "epoch": 0.27712,
-      "grad_norm": 0.6135783826974505,
+      "grad_norm": 0.2803609027017541,
       "learning_rate": 4.33297270904321e-05,
-      "loss": 0.268,
+      "loss": 0.2681,
       "step": 433
     },
     {
       "epoch": 0.27776,
-      "grad_norm": 0.27762568110334607,
+      "grad_norm": 0.2920451450586205,
       "learning_rate": 4.329367181138262e-05,
-      "loss": 0.2765,
+      "loss": 0.2762,
       "step": 434
     },
     {
       "epoch": 0.2784,
-      "grad_norm": 0.2858205844663221,
+      "grad_norm": 0.30901997193481695,
       "learning_rate": 4.3257534437057674e-05,
-      "loss": 0.2569,
+      "loss": 0.2575,
       "step": 435
     },
     {
       "epoch": 0.27904,
-      "grad_norm": 0.30296957565612004,
+      "grad_norm": 0.28493311811209304,
       "learning_rate": 4.3221315129628506e-05,
-      "loss": 0.249,
+      "loss": 0.2493,
       "step": 436
     },
     {
       "epoch": 0.27968,
-      "grad_norm": 0.27957554300787335,
+      "grad_norm": 0.3039818748430415,
       "learning_rate": 4.318501405163406e-05,
-      "loss": 0.2628,
+      "loss": 0.2624,
       "step": 437
     },
     {
       "epoch": 0.28032,
-      "grad_norm": 0.26204789686898283,
+      "grad_norm": 0.26937915062885653,
       "learning_rate": 4.314863136598022e-05,
-      "loss": 0.2475,
+      "loss": 0.247,
       "step": 438
     },
     {
       "epoch": 0.28096,
-      "grad_norm": 0.3058255372708976,
+      "grad_norm": 0.33555541975858594,
       "learning_rate": 4.311216723593913e-05,
-      "loss": 0.2866,
+      "loss": 0.2867,
       "step": 439
     },
     {
       "epoch": 0.2816,
-      "grad_norm": 0.3251040999392898,
+      "grad_norm": 0.31925378826398826,
       "learning_rate": 4.307562182514838e-05,
-      "loss": 0.2622,
+      "loss": 0.2617,
       "step": 440
     },
     {
       "epoch": 0.28224,
-      "grad_norm": 0.30006956753828234,
+      "grad_norm": 0.32623010034841626,
       "learning_rate": 4.3038995297610364e-05,
-      "loss": 0.2488,
+      "loss": 0.249,
       "step": 441
     },
     {
       "epoch": 0.28288,
-      "grad_norm": 0.30274868902673485,
+      "grad_norm": 0.30358330160420666,
       "learning_rate": 4.3002287817691466e-05,
-      "loss": 0.2867,
+      "loss": 0.2865,
       "step": 442
     },
     {
       "epoch": 0.28352,
-      "grad_norm": 0.38145313435180433,
+      "grad_norm": 0.33915628852280005,
       "learning_rate": 4.2965499550121354e-05,
-      "loss": 0.2687,
+      "loss": 0.2685,
       "step": 443
     },
     {
       "epoch": 0.28416,
-      "grad_norm": 0.3094348741385053,
+      "grad_norm": 0.3547465820682451,
       "learning_rate": 4.2928630659992267e-05,
-      "loss": 0.2712,
+      "loss": 0.2711,
       "step": 444
     },
     {
       "epoch": 0.2848,
-      "grad_norm": 0.34366344981433733,
+      "grad_norm": 0.3336725360323176,
       "learning_rate": 4.2891681312758225e-05,
-      "loss": 0.2605,
+      "loss": 0.261,
       "step": 445
     },
     {
       "epoch": 0.28544,
-      "grad_norm": 0.2807715782879043,
+      "grad_norm": 0.30303539171283983,
       "learning_rate": 4.285465167423432e-05,
-      "loss": 0.272,
+      "loss": 0.2719,
       "step": 446
     },
     {
       "epoch": 0.28608,
-      "grad_norm": 0.3038994998195099,
+      "grad_norm": 0.311033741890672,
       "learning_rate": 4.281754191059596e-05,
-      "loss": 0.262,
+      "loss": 0.2616,
       "step": 447
     },
     {
       "epoch": 0.28672,
-      "grad_norm": 0.30453615561443753,
+      "grad_norm": 0.3125370811674668,
       "learning_rate": 4.278035218837813e-05,
       "loss": 0.2889,
       "step": 448
     },
     {
       "epoch": 0.28736,
-      "grad_norm": 0.291161967151285,
+      "grad_norm": 0.2756629291524462,
       "learning_rate": 4.274308267447463e-05,
-      "loss": 0.2519,
+      "loss": 0.2518,
       "step": 449
     },
     {
       "epoch": 0.288,
-      "grad_norm": 0.2956628354845414,
+      "grad_norm": 0.2984734533438556,
       "learning_rate": 4.270573353613734e-05,
-      "loss": 0.2709,
+      "loss": 0.2712,
       "step": 450
     },
     {
       "epoch": 0.28864,
-      "grad_norm": 0.2796475447880478,
+      "grad_norm": 0.29595878739145964,
       "learning_rate": 4.266830494097546e-05,
-      "loss": 0.2543,
+      "loss": 0.255,
       "step": 451
     },
     {
       "epoch": 0.28928,
-      "grad_norm": 0.3103279388881702,
+      "grad_norm": 0.3205021385414694,
       "learning_rate": 4.263079705695478e-05,
-      "loss": 0.2638,
+      "loss": 0.2636,
       "step": 452
     },
     {
       "epoch": 0.28992,
-      "grad_norm": 0.3546905046069225,
+      "grad_norm": 0.3600626165622021,
       "learning_rate": 4.2593210052396894e-05,
-      "loss": 0.2545,
+      "loss": 0.2541,
       "step": 453
     },
     {
       "epoch": 0.29056,
-      "grad_norm": 0.2790744270430645,
+      "grad_norm": 0.25969620151233935,
       "learning_rate": 4.255554409597846e-05,
-      "loss": 0.2405,
+      "loss": 0.2401,
       "step": 454
     },
     {
       "epoch": 0.2912,
-      "grad_norm": 0.3313654210870624,
+      "grad_norm": 0.33149623415612184,
       "learning_rate": 4.2517799356730445e-05,
-      "loss": 0.2692,
+      "loss": 0.2687,
       "step": 455
     },
     {
       "epoch": 0.29184,
-      "grad_norm": 0.3016384994176408,
+      "grad_norm": 0.30226474356531935,
       "learning_rate": 4.247997600403737e-05,
-      "loss": 0.2601,
+      "loss": 0.2594,
       "step": 456
     },
     {
       "epoch": 0.29248,
-      "grad_norm": 0.2993078346559434,
+      "grad_norm": 0.30007952734237353,
       "learning_rate": 4.244207420763655e-05,
-      "loss": 0.2513,
+      "loss": 0.2498,
       "step": 457
     },
     {
       "epoch": 0.29312,
-      "grad_norm": 0.31622969353803504,
+      "grad_norm": 0.28047639127931384,
       "learning_rate": 4.240409413761731e-05,
-      "loss": 0.2551,
+      "loss": 0.2556,
       "step": 458
     },
     {
       "epoch": 0.29376,
-      "grad_norm": 0.30524388925934887,
+      "grad_norm": 0.29832917410059595,
       "learning_rate": 4.236603596442023e-05,
-      "loss": 0.2631,
+      "loss": 0.2632,
       "step": 459
     },
     {
       "epoch": 0.2944,
-      "grad_norm": 0.30637282729194065,
+      "grad_norm": 0.29802598060016766,
       "learning_rate": 4.232789985883642e-05,
-      "loss": 0.2696,
+      "loss": 0.2693,
       "step": 460
     },
     {
       "epoch": 0.29504,
-      "grad_norm": 0.28616384975047443,
+      "grad_norm": 0.2911033726222457,
       "learning_rate": 4.2289685992006704e-05,
-      "loss": 0.277,
+      "loss": 0.2772,
       "step": 461
     },
     {
       "epoch": 0.29568,
-      "grad_norm": 0.3103128382558485,
+      "grad_norm": 0.2903001309362164,
       "learning_rate": 4.225139453542087e-05,
-      "loss": 0.2619,
+      "loss": 0.2607,
       "step": 462
     },
     {
       "epoch": 0.29632,
-      "grad_norm": 0.29703792526543216,
+      "grad_norm": 0.3078582069653545,
       "learning_rate": 4.2213025660916904e-05,
-      "loss": 0.2676,
+      "loss": 0.2667,
       "step": 463
     },
     {
       "epoch": 0.29696,
-      "grad_norm": 0.3168951655026096,
+      "grad_norm": 0.3030493045200669,
       "learning_rate": 4.21745795406802e-05,
-      "loss": 0.2473,
+      "loss": 0.2467,
       "step": 464
     },
     {
       "epoch": 0.2976,
-      "grad_norm": 0.2671842042378145,
+      "grad_norm": 0.26660015999204884,
       "learning_rate": 4.2136056347242835e-05,
-      "loss": 0.2505,
+      "loss": 0.25,
       "step": 465
     },
     {
       "epoch": 0.29824,
-      "grad_norm": 0.29683839331480205,
+      "grad_norm": 0.30144794273216646,
       "learning_rate": 4.209745625348273e-05,
       "loss": 0.284,
       "step": 466
     },
     {
       "epoch": 0.29888,
-      "grad_norm": 0.32464847892381105,
+      "grad_norm": 0.3284286693049362,
       "learning_rate": 4.205877943262291e-05,
-      "loss": 0.2462,
+      "loss": 0.2464,
       "step": 467
     },
     {
       "epoch": 0.29952,
-      "grad_norm": 0.25754594330485725,
+      "grad_norm": 0.25954840338063495,
       "learning_rate": 4.2020026058230756e-05,
-      "loss": 0.2688,
+      "loss": 0.2684,
       "step": 468
     },
     {
       "epoch": 0.30016,
-      "grad_norm": 0.31150290164451966,
+      "grad_norm": 0.2945590110420638,
       "learning_rate": 4.1981196304217154e-05,
-      "loss": 0.2641,
+      "loss": 0.264,
       "step": 469
     },
     {
       "epoch": 0.3008,
-      "grad_norm": 0.29116448915632004,
+      "grad_norm": 0.2927523824074909,
       "learning_rate": 4.1942290344835765e-05,
-      "loss": 0.2643,
+      "loss": 0.2645,
       "step": 470
     },
     {
       "epoch": 0.30144,
-      "grad_norm": 0.2990863020226855,
+      "grad_norm": 0.30771010182551334,
       "learning_rate": 4.190330835468224e-05,
-      "loss": 0.267,
+      "loss": 0.2671,
       "step": 471
     },
     {
       "epoch": 0.30208,
-      "grad_norm": 0.3072242093075254,
+      "grad_norm": 0.29384348361000195,
       "learning_rate": 4.1864250508693424e-05,
-      "loss": 0.2791,
+      "loss": 0.2794,
       "step": 472
     },
     {
       "epoch": 0.30272,
-      "grad_norm": 0.31058783224981473,
+      "grad_norm": 0.2896660319426316,
       "learning_rate": 4.182511698214657e-05,
-      "loss": 0.2712,
+      "loss": 0.2706,
       "step": 473
     },
     {
       "epoch": 0.30336,
-      "grad_norm": 0.2798304078793174,
+      "grad_norm": 0.28080256128136577,
       "learning_rate": 4.1785907950658573e-05,
-      "loss": 0.2593,
+      "loss": 0.259,
       "step": 474
     },
     {
       "epoch": 0.304,
-      "grad_norm": 0.31535291727654785,
+      "grad_norm": 0.30478773930609737,
       "learning_rate": 4.1746623590185155e-05,
-      "loss": 0.2536,
+      "loss": 0.2528,
       "step": 475
     },
     {
       "epoch": 0.30464,
-      "grad_norm": 0.2990140489891215,
+      "grad_norm": 0.2961976249198792,
       "learning_rate": 4.170726407702008e-05,
-      "loss": 0.247,
+      "loss": 0.2466,
       "step": 476
     },
     {
       "epoch": 0.30528,
-      "grad_norm": 0.33391702160348635,
+      "grad_norm": 0.31686233103136463,
       "learning_rate": 4.1667829587794374e-05,
-      "loss": 0.2835,
+      "loss": 0.283,
       "step": 477
     },
     {
       "epoch": 0.30592,
-      "grad_norm": 0.32213023722063483,
+      "grad_norm": 0.3159707315375394,
       "learning_rate": 4.162832029947553e-05,
-      "loss": 0.2745,
+      "loss": 0.2735,
       "step": 478
     },
     {
       "epoch": 0.30656,
-      "grad_norm": 0.27871899544394474,
+      "grad_norm": 0.27811512588290743,
       "learning_rate": 4.158873638936672e-05,
-      "loss": 0.2476,
+      "loss": 0.248,
       "step": 479
     },
     {
       "epoch": 0.3072,
-      "grad_norm": 0.32481480905586535,
+      "grad_norm": 0.45896640549255663,
       "learning_rate": 4.154907803510599e-05,
-      "loss": 0.2528,
+      "loss": 0.2525,
       "step": 480
     },
     {
       "epoch": 0.30784,
-      "grad_norm": 0.29044250985733133,
+      "grad_norm": 0.28996697916337744,
       "learning_rate": 4.150934541466545e-05,
-      "loss": 0.2626,
+      "loss": 0.2627,
       "step": 481
     },
     {
       "epoch": 0.30848,
-      "grad_norm": 0.3001465924251364,
+      "grad_norm": 0.3052139609989655,
       "learning_rate": 4.146953870635049e-05,
-      "loss": 0.2733,
+      "loss": 0.273,
       "step": 482
     },
     {
       "epoch": 0.30912,
-      "grad_norm": 0.2924668088023568,
+      "grad_norm": 0.32057984599943073,
       "learning_rate": 4.1429658088798996e-05,
-      "loss": 0.2639,
+      "loss": 0.2636,
       "step": 483
     },
     {
       "epoch": 0.30976,
-      "grad_norm": 0.2978329946137925,
+      "grad_norm": 0.29200883447827164,
       "learning_rate": 4.138970374098052e-05,
-      "loss": 0.2493,
+      "loss": 0.2498,
       "step": 484
     },
     {
       "epoch": 0.3104,
-      "grad_norm": 0.32039490633737067,
+      "grad_norm": 0.3211437061784034,
       "learning_rate": 4.1349675842195494e-05,
-      "loss": 0.2679,
+      "loss": 0.2688,
       "step": 485
     },
     {
       "epoch": 0.31104,
-      "grad_norm": 0.2784834783735675,
+      "grad_norm": 0.2692839546004917,
       "learning_rate": 4.130957457207441e-05,
-      "loss": 0.2511,
+      "loss": 0.2513,
       "step": 486
     },
     {
       "epoch": 0.31168,
-      "grad_norm": 0.3231096974927409,
+      "grad_norm": 0.29738376417175055,
       "learning_rate": 4.126940011057703e-05,
-      "loss": 0.2417,
+      "loss": 0.242,
       "step": 487
     },
     {
       "epoch": 0.31232,
-      "grad_norm": 0.29574649661084784,
+      "grad_norm": 0.2925125860829048,
       "learning_rate": 4.1229152637991576e-05,
-      "loss": 0.2683,
+      "loss": 0.269,
       "step": 488
     },
     {
       "epoch": 0.31296,
-      "grad_norm": 0.3198759705633704,
+      "grad_norm": 0.31841073786254037,
       "learning_rate": 4.1188832334933906e-05,
-      "loss": 0.2485,
+      "loss": 0.2491,
       "step": 489
     },
     {
       "epoch": 0.3136,
-      "grad_norm": 0.2743856200052568,
+      "grad_norm": 0.256917596278123,
       "learning_rate": 4.114843938234673e-05,
-      "loss": 0.2753,
+      "loss": 0.2759,
       "step": 490
     },
     {
       "epoch": 0.31424,
-      "grad_norm": 0.31005901539666614,
+      "grad_norm": 0.3063311474261881,
       "learning_rate": 4.1107973961498763e-05,
-      "loss": 0.2591,
+      "loss": 0.2598,
       "step": 491
     },
     {
       "epoch": 0.31488,
-      "grad_norm": 0.3300495689975318,
+      "grad_norm": 0.32855934285971683,
       "learning_rate": 4.106743625398395e-05,
-      "loss": 0.2675,
+      "loss": 0.2681,
       "step": 492
     },
     {
       "epoch": 0.31552,
-      "grad_norm": 0.3068927581792596,
+      "grad_norm": 0.36353433750679803,
       "learning_rate": 4.102682644172061e-05,
-      "loss": 0.2871,
+      "loss": 0.2873,
       "step": 493
     },
     {
       "epoch": 0.31616,
-      "grad_norm": 0.3264341456984584,
+      "grad_norm": 0.33622331775831543,
       "learning_rate": 4.0986144706950677e-05,
-      "loss": 0.2755,
+      "loss": 0.276,
       "step": 494
     },
     {
       "epoch": 0.3168,
-      "grad_norm": 0.31711346944454233,
+      "grad_norm": 0.31722692435482774,
       "learning_rate": 4.094539123223881e-05,
-      "loss": 0.2442,
+      "loss": 0.2444,
       "step": 495
     },
     {
       "epoch": 0.31744,
-      "grad_norm": 0.2862941880474931,
+      "grad_norm": 0.2780719831837547,
       "learning_rate": 4.090456620047162e-05,
-      "loss": 0.2447,
+      "loss": 0.2451,
       "step": 496
     },
     {
       "epoch": 0.31808,
-      "grad_norm": 0.3455300856074319,
+      "grad_norm": 0.35077422669493935,
       "learning_rate": 4.086366979485685e-05,
-      "loss": 0.2806,
+      "loss": 0.2816,
       "step": 497
     },
     {
       "epoch": 0.31872,
-      "grad_norm": 0.3038196785649234,
+      "grad_norm": 0.30473125946275736,
       "learning_rate": 4.082270219892254e-05,
-      "loss": 0.2643,
+      "loss": 0.2644,
       "step": 498
     },
     {
       "epoch": 0.31936,
-      "grad_norm": 0.27271340321003223,
+      "grad_norm": 0.27919846819675614,
       "learning_rate": 4.0781663596516206e-05,
-      "loss": 0.2433,
+      "loss": 0.244,
       "step": 499
     },
     {
       "epoch": 0.32,
-      "grad_norm": 0.29835333522383656,
+      "grad_norm": 0.30869545887631733,
       "learning_rate": 4.074055417180401e-05,
-      "loss": 0.2704,
+      "loss": 0.2709,
       "step": 500
     },
     {
       "epoch": 0.32064,
-      "grad_norm": 0.3196694498557124,
+      "grad_norm": 0.32401310700356356,
       "learning_rate": 4.069937410926994e-05,
-      "loss": 0.2549,
+      "loss": 0.2555,
       "step": 501
     },
     {
       "epoch": 0.32128,
-      "grad_norm": 0.28825300365435247,
+      "grad_norm": 0.28989905607846644,
       "learning_rate": 4.0658123593714984e-05,
-      "loss": 0.2647,
+      "loss": 0.2648,
       "step": 502
     },
     {
       "epoch": 0.32192,
-      "grad_norm": 0.2786545499138097,
+      "grad_norm": 0.2970056265621276,
       "learning_rate": 4.061680281025628e-05,
-      "loss": 0.2477,
+      "loss": 0.2491,
       "step": 503
     },
     {
       "epoch": 0.32256,
-      "grad_norm": 0.29978230649243376,
+      "grad_norm": 0.30095176466887164,
       "learning_rate": 4.057541194432634e-05,
-      "loss": 0.2421,
+      "loss": 0.2428,
       "step": 504
     },
     {
       "epoch": 0.3232,
-      "grad_norm": 0.3012784250324835,
+      "grad_norm": 0.307577464320209,
       "learning_rate": 4.053395118167214e-05,
-      "loss": 0.2581,
+      "loss": 0.2591,
       "step": 505
     },
     {
       "epoch": 0.32384,
-      "grad_norm": 0.29499478568673526,
+      "grad_norm": 0.28875399318294626,
       "learning_rate": 4.049242070835435e-05,
-      "loss": 0.2874,
+      "loss": 0.2876,
       "step": 506
     },
     {
       "epoch": 0.32448,
-      "grad_norm": 0.31674834634025634,
+      "grad_norm": 0.3372288637186776,
       "learning_rate": 4.0450820710746475e-05,
-      "loss": 0.2608,
+      "loss": 0.2614,
       "step": 507
     },
     {
       "epoch": 0.32512,
-      "grad_norm": 0.2865283632314673,
+      "grad_norm": 0.2893542015604074,
       "learning_rate": 4.0409151375534014e-05,
-      "loss": 0.2637,
+      "loss": 0.2646,
       "step": 508
     },
     {
       "epoch": 0.32576,
-      "grad_norm": 0.30827781105540003,
+      "grad_norm": 0.30517166216359926,
       "learning_rate": 4.036741288971362e-05,
-      "loss": 0.2558,
+      "loss": 0.2563,
       "step": 509
     },
     {
       "epoch": 0.3264,
-      "grad_norm": 0.31853194371996735,
+      "grad_norm": 0.3132371089607073,
       "learning_rate": 4.032560544059228e-05,
-      "loss": 0.255,
+      "loss": 0.2549,
       "step": 510
     },
     {
       "epoch": 0.32704,
-      "grad_norm": 0.38294307444540887,
+      "grad_norm": 0.2584564148069602,
       "learning_rate": 4.028372921578647e-05,
-      "loss": 0.2641,
+      "loss": 0.2642,
       "step": 511
     },
     {
       "epoch": 0.32768,
-      "grad_norm": 0.2864115669013345,
+      "grad_norm": 0.29056742266359853,
       "learning_rate": 4.024178440322128e-05,
-      "loss": 0.2507,
+      "loss": 0.2517,
       "step": 512
     },
     {
       "epoch": 0.32832,
-      "grad_norm": 0.2773279212490367,
+      "grad_norm": 0.2806424707934406,
       "learning_rate": 4.019977119112962e-05,
-      "loss": 0.2604,
+      "loss": 0.2612,
       "step": 513
     },
     {
       "epoch": 0.32896,
-      "grad_norm": 0.30350170146466277,
+      "grad_norm": 0.3213624904512003,
       "learning_rate": 4.015768976805135e-05,
-      "loss": 0.303,
+      "loss": 0.3041,
       "step": 514
     },
     {
       "epoch": 0.3296,
-      "grad_norm": 0.3019836637698821,
+      "grad_norm": 0.3196533973528481,
       "learning_rate": 4.011554032283242e-05,
-      "loss": 0.2342,
+      "loss": 0.2351,
       "step": 515
     },
     {
       "epoch": 0.33024,
-      "grad_norm": 0.33693079309207474,
+      "grad_norm": 0.3329565293585606,
       "learning_rate": 4.007332304462406e-05,
-      "loss": 0.2503,
+      "loss": 0.251,
       "step": 516
     },
     {
       "epoch": 0.33088,
-      "grad_norm": 0.301807035388646,
+      "grad_norm": 0.3109673121296247,
       "learning_rate": 4.003103812288189e-05,
-      "loss": 0.2681,
+      "loss": 0.2686,
       "step": 517
     },
     {
       "epoch": 0.33152,
-      "grad_norm": 0.27363500419504855,
+      "grad_norm": 0.2675208149986414,
       "learning_rate": 3.99886857473651e-05,
-      "loss": 0.2523,
+      "loss": 0.2525,
       "step": 518
     },
     {
       "epoch": 0.33216,
-      "grad_norm": 0.3288948783067086,
+      "grad_norm": 0.34229548303640794,
       "learning_rate": 3.9946266108135575e-05,
-      "loss": 0.2771,
+      "loss": 0.2772,
       "step": 519
     },
     {
       "epoch": 0.3328,
-      "grad_norm": 0.3116926537482921,
+      "grad_norm": 0.32006866438809456,
       "learning_rate": 3.9903779395557087e-05,
-      "loss": 0.261,
+      "loss": 0.2622,
       "step": 520
     },
     {
       "epoch": 0.33344,
-      "grad_norm": 0.2620289802478765,
+      "grad_norm": 0.26715648446115114,
       "learning_rate": 3.9861225800294366e-05,
-      "loss": 0.2295,
+      "loss": 0.2308,
       "step": 521
     },
     {
       "epoch": 0.33408,
-      "grad_norm": 0.32247525286268586,
+      "grad_norm": 0.3480585053707105,
       "learning_rate": 3.981860551331232e-05,
-      "loss": 0.2671,
+      "loss": 0.2669,
       "step": 522
     },
     {
       "epoch": 0.33472,
-      "grad_norm": 0.29485974202493387,
+      "grad_norm": 0.2922018702119113,
       "learning_rate": 3.977591872587512e-05,
-      "loss": 0.2559,
+      "loss": 0.2565,
       "step": 523
     },
     {
       "epoch": 0.33536,
-      "grad_norm": 0.2615445610289309,
+      "grad_norm": 0.2611827785563305,
       "learning_rate": 3.973316562954539e-05,
-      "loss": 0.2514,
+      "loss": 0.2516,
       "step": 524
     },
     {
       "epoch": 0.336,
-      "grad_norm": 0.28540916266337385,
+      "grad_norm": 0.2984964297936148,
       "learning_rate": 3.969034641618332e-05,
-      "loss": 0.2845,
+      "loss": 0.2852,
       "step": 525
     },
     {
       "epoch": 0.33664,
-      "grad_norm": 0.2973099909096069,
+      "grad_norm": 0.2892462655028235,
       "learning_rate": 3.964746127794579e-05,
-      "loss": 0.2627,
+      "loss": 0.2635,
       "step": 526
     },
     {
       "epoch": 0.33728,
-      "grad_norm": 0.2829998813648647,
+      "grad_norm": 0.28835600723222543,
       "learning_rate": 3.9604510407285524e-05,
-      "loss": 0.2502,
+      "loss": 0.2511,
       "step": 527
     },
     {
       "epoch": 0.33792,
-      "grad_norm": 0.2923882761100174,
+      "grad_norm": 0.2999026080900571,
       "learning_rate": 3.956149399695027e-05,
-      "loss": 0.2524,
+      "loss": 0.2529,
       "step": 528
     },
     {
       "epoch": 0.33856,
-      "grad_norm": 0.31889981262923656,
+      "grad_norm": 0.3174683122687009,
       "learning_rate": 3.951841223998186e-05,
-      "loss": 0.2802,
+      "loss": 0.2803,
       "step": 529
     },
     {
       "epoch": 0.3392,
-      "grad_norm": 0.27084760232555066,
+      "grad_norm": 0.26907149760511656,
       "learning_rate": 3.947526532971538e-05,
-      "loss": 0.2447,
+      "loss": 0.2452,
       "step": 530
     },
     {
       "epoch": 0.33984,
-      "grad_norm": 0.277167819276775,
+      "grad_norm": 0.2771670066399074,
       "learning_rate": 3.943205345977831e-05,
-      "loss": 0.2599,
+      "loss": 0.2598,
       "step": 531
     },
     {
       "epoch": 0.34048,
-      "grad_norm": 0.45847772602327513,
+      "grad_norm": 0.28877064076702313,
       "learning_rate": 3.9388776824089636e-05,
-      "loss": 0.2568,
+      "loss": 0.2559,
       "step": 532
     },
     {
       "epoch": 0.34112,
-      "grad_norm": 0.2646005627784305,
+      "grad_norm": 0.2649860095131696,
       "learning_rate": 3.9345435616858995e-05,
-      "loss": 0.2674,
+      "loss": 0.2673,
       "step": 533
     },
     {
       "epoch": 0.34176,
-      "grad_norm": 0.2704833186174627,
+      "grad_norm": 0.269531839762295,
       "learning_rate": 3.930203003258579e-05,
-      "loss": 0.2566,
+      "loss": 0.2569,
       "step": 534
     },
     {
       "epoch": 0.3424,
-      "grad_norm": 0.27320797496261323,
+      "grad_norm": 0.27891756261987266,
       "learning_rate": 3.9258560266058335e-05,
-      "loss": 0.2689,
+      "loss": 0.2702,
       "step": 535
     },
     {
       "epoch": 0.34304,
-      "grad_norm": 0.26245552780653786,
+      "grad_norm": 0.27053829922547884,
       "learning_rate": 3.921502651235297e-05,
-      "loss": 0.2533,
+      "loss": 0.2535,
       "step": 536
     },
     {
       "epoch": 0.34368,
-      "grad_norm": 0.2686721573275851,
+      "grad_norm": 0.28011073116796703,
       "learning_rate": 3.917142896683318e-05,
-      "loss": 0.2713,
+      "loss": 0.272,
       "step": 537
     },
     {
       "epoch": 0.34432,
-      "grad_norm": 0.27677482339556025,
+      "grad_norm": 0.270867487872639,
       "learning_rate": 3.912776782514872e-05,
-      "loss": 0.2541,
+      "loss": 0.2543,
       "step": 538
     },
     {
       "epoch": 0.34496,
-      "grad_norm": 0.2752780592634784,
+      "grad_norm": 0.2746991664504782,
       "learning_rate": 3.908404328323475e-05,
-      "loss": 0.2724,
+      "loss": 0.2728,
       "step": 539
     },
     {
       "epoch": 0.3456,
-      "grad_norm": 0.30938115723324516,
+      "grad_norm": 0.27939140218767883,
       "learning_rate": 3.904025553731094e-05,
-      "loss": 0.2413,
+      "loss": 0.2418,
       "step": 540
     },
     {
       "epoch": 0.34624,
-      "grad_norm": 0.26305059884444465,
+      "grad_norm": 0.2731998143935589,
       "learning_rate": 3.8996404783880606e-05,
-      "loss": 0.2429,
+      "loss": 0.2433,
       "step": 541
     },
     {
       "epoch": 0.34688,
-      "grad_norm": 0.27017911566688885,
+      "grad_norm": 0.2750465650133765,
       "learning_rate": 3.895249121972981e-05,
-      "loss": 0.2425,
+      "loss": 0.2422,
       "step": 542
     },
     {
       "epoch": 0.34752,
-      "grad_norm": 0.2807572440929959,
+      "grad_norm": 0.27336449849709915,
       "learning_rate": 3.890851504192649e-05,
-      "loss": 0.2692,
+      "loss": 0.2689,
       "step": 543
     },
     {
       "epoch": 0.34816,
-      "grad_norm": 0.2937755464668312,
+      "grad_norm": 0.29586657387061377,
       "learning_rate": 3.8864476447819565e-05,
-      "loss": 0.2471,
+      "loss": 0.2465,
       "step": 544
     },
     {
       "epoch": 0.3488,
-      "grad_norm": 0.26722639745381527,
+      "grad_norm": 0.270200591770297,
       "learning_rate": 3.8820375635038067e-05,
-      "loss": 0.257,
+      "loss": 0.2573,
       "step": 545
     },
     {
       "epoch": 0.34944,
-      "grad_norm": 0.32942353225222054,
+      "grad_norm": 0.3190425370253469,
       "learning_rate": 3.877621280149022e-05,
-      "loss": 0.2673,
+      "loss": 0.2678,
       "step": 546
     },
     {
       "epoch": 0.35008,
-      "grad_norm": 0.26071685079047663,
+      "grad_norm": 0.26043543170976835,
       "learning_rate": 3.87319881453626e-05,
-      "loss": 0.235,
+      "loss": 0.2354,
       "step": 547
     },
     {
       "epoch": 0.35072,
-      "grad_norm": 0.2703795884584656,
+      "grad_norm": 0.2704011594558705,
       "learning_rate": 3.868770186511921e-05,
-      "loss": 0.2618,
+      "loss": 0.2626,
       "step": 548
     },
     {
       "epoch": 0.35136,
-      "grad_norm": 0.3063325675005532,
+      "grad_norm": 0.29304403560572506,
       "learning_rate": 3.864335415950059e-05,
       "loss": 0.2437,
       "step": 549
     },
     {
       "epoch": 0.352,
-      "grad_norm": 0.25931696906001084,
+      "grad_norm": 0.2655479075710523,
       "learning_rate": 3.859894522752297e-05,
-      "loss": 0.2498,
+      "loss": 0.2503,
       "step": 550
     },
     {
       "epoch": 0.35264,
-      "grad_norm": 0.317347010295714,
+      "grad_norm": 0.312630229855102,
       "learning_rate": 3.8554475268477274e-05,
-      "loss": 0.2573,
+      "loss": 0.257,
       "step": 551
     },
     {
       "epoch": 0.35328,
-      "grad_norm": 0.2879966431448922,
+      "grad_norm": 0.274670625897055,
       "learning_rate": 3.850994448192836e-05,
-      "loss": 0.2601,
+      "loss": 0.2596,
       "step": 552
     },
     {
       "epoch": 0.35392,
-      "grad_norm": 0.2704993494133405,
+      "grad_norm": 0.2474262363659263,
       "learning_rate": 3.846535306771404e-05,
-      "loss": 0.2495,
+      "loss": 0.2497,
       "step": 553
     },
     {
       "epoch": 0.35456,
-      "grad_norm": 0.32753739181266783,
+      "grad_norm": 0.30813525842938966,
       "learning_rate": 3.842070122594417e-05,
-      "loss": 0.2647,
+      "loss": 0.2645,
       "step": 554
     },
     {
       "epoch": 0.3552,
-      "grad_norm": 0.27693874665061385,
+      "grad_norm": 0.27841535526184324,
       "learning_rate": 3.837598915699981e-05,
-      "loss": 0.2459,
+      "loss": 0.2464,
       "step": 555
     },
     {
       "epoch": 0.35584,
-      "grad_norm": 0.3033680637986832,
+      "grad_norm": 0.2938358535555391,
       "learning_rate": 3.83312170615323e-05,
-      "loss": 0.2384,
+      "loss": 0.2387,
       "step": 556
     },
     {
       "epoch": 0.35648,
-      "grad_norm": 0.3039697232443411,
+      "grad_norm": 0.29756211437074354,
       "learning_rate": 3.8286385140462334e-05,
-      "loss": 0.2473,
+      "loss": 0.2465,
       "step": 557
     },
     {
       "epoch": 0.35712,
-      "grad_norm": 0.3047824145474792,
+      "grad_norm": 0.30217380986318165,
       "learning_rate": 3.8241493594979114e-05,
-      "loss": 0.2635,
+      "loss": 0.264,
       "step": 558
     },
     {
       "epoch": 0.35776,
-      "grad_norm": 0.2991402315294844,
+      "grad_norm": 0.28961179785597074,
       "learning_rate": 3.8196542626539396e-05,
-      "loss": 0.2637,
+      "loss": 0.2641,
       "step": 559
     },
     {
       "epoch": 0.3584,
-      "grad_norm": 0.3092917498688086,
+      "grad_norm": 0.30695303659565276,
       "learning_rate": 3.815153243686658e-05,
-      "loss": 0.2648,
+      "loss": 0.2643,
       "step": 560
     },
     {
       "epoch": 0.35904,
-      "grad_norm": 0.30146699010235556,
+      "grad_norm": 0.29395099772794475,
       "learning_rate": 3.8106463227949885e-05,
-      "loss": 0.265,
+      "loss": 0.2643,
       "step": 561
     },
     {
       "epoch": 0.35968,
-      "grad_norm": 0.2769544706427565,
+      "grad_norm": 0.29219915519201384,
       "learning_rate": 3.806133520204335e-05,
-      "loss": 0.2352,
+      "loss": 0.2358,
       "step": 562
     },
     {
       "epoch": 0.36032,
-      "grad_norm": 0.35477609684955097,
+      "grad_norm": 0.2771159950154576,
       "learning_rate": 3.801614856166495e-05,
-      "loss": 0.2441,
+      "loss": 0.2444,
       "step": 563
     },
     {
       "epoch": 0.36096,
-      "grad_norm": 0.29044978422937073,
+      "grad_norm": 0.3140877640793322,
       "learning_rate": 3.797090350959574e-05,
-      "loss": 0.2478,
+      "loss": 0.2479,
       "step": 564
     },
     {
       "epoch": 0.3616,
-      "grad_norm": 0.2847992844131417,
+      "grad_norm": 0.28818249284565073,
       "learning_rate": 3.792560024887887e-05,
-      "loss": 0.253,
+      "loss": 0.2538,
       "step": 565
     },
     {
       "epoch": 0.36224,
-      "grad_norm": 0.3082206555429372,
+      "grad_norm": 0.2783892260751085,
       "learning_rate": 3.788023898281873e-05,
-      "loss": 0.252,
+      "loss": 0.2523,
       "step": 566
     },
     {
       "epoch": 0.36288,
-      "grad_norm": 0.29684039873581247,
+      "grad_norm": 0.3092513006484571,
       "learning_rate": 3.783481991498001e-05,
-      "loss": 0.2573,
+      "loss": 0.2575,
       "step": 567
     },
     {
       "epoch": 0.36352,
-      "grad_norm": 0.2799704203242341,
+      "grad_norm": 0.28561921777773985,
       "learning_rate": 3.7789343249186805e-05,
-      "loss": 0.2654,
+      "loss": 0.2657,
       "step": 568
     },
     {
       "epoch": 0.36416,
-      "grad_norm": 0.3072567979064305,
+      "grad_norm": 0.29745794354352423,
       "learning_rate": 3.774380918952165e-05,
-      "loss": 0.2749,
+      "loss": 0.2746,
       "step": 569
     },
     {
       "epoch": 0.3648,
-      "grad_norm": 0.2903629157466471,
+      "grad_norm": 0.28619753080969124,
       "learning_rate": 3.76982179403247e-05,
-      "loss": 0.2379,
+      "loss": 0.238,
       "step": 570
     },
     {
       "epoch": 0.36544,
-      "grad_norm": 0.27576649206877873,
+      "grad_norm": 0.2865777627431609,
       "learning_rate": 3.7652569706192694e-05,
-      "loss": 0.2682,
+      "loss": 0.2672,
       "step": 571
     },
     {
       "epoch": 0.36608,
-      "grad_norm": 0.27013938988147623,
+      "grad_norm": 0.2662051134951306,
       "learning_rate": 3.760686469197815e-05,
-      "loss": 0.2492,
+      "loss": 0.2493,
       "step": 572
     },
     {
       "epoch": 0.36672,
-      "grad_norm": 0.36320284367672434,
+      "grad_norm": 0.3122204747602635,
       "learning_rate": 3.756110310278835e-05,
       "loss": 0.249,
       "step": 573
     },
     {
       "epoch": 0.36736,
-      "grad_norm": 0.2455885509270789,
+      "grad_norm": 0.2551416288395675,
       "learning_rate": 3.7515285143984524e-05,
-      "loss": 0.2429,
+      "loss": 0.2426,
       "step": 574
     },
     {
       "epoch": 0.368,
-      "grad_norm": 0.2968349995967185,
+      "grad_norm": 0.2982227316906956,
       "learning_rate": 3.746941102118082e-05,
-      "loss": 0.2503,
+      "loss": 0.2509,
       "step": 575
     },
     {
       "epoch": 0.36864,
-      "grad_norm": 0.25701314511324636,
+      "grad_norm": 0.24907998630323142,
       "learning_rate": 3.742348094024342e-05,
-      "loss": 0.2721,
+      "loss": 0.2731,
       "step": 576
     },
     {
       "epoch": 0.36928,
-      "grad_norm": 0.3043146256349847,
+      "grad_norm": 0.308035846935991,
       "learning_rate": 3.7377495107289664e-05,
-      "loss": 0.2285,
+      "loss": 0.2294,
       "step": 577
     },
     {
       "epoch": 0.36992,
-      "grad_norm": 0.2977740197991099,
+      "grad_norm": 0.2981772132955674,
       "learning_rate": 3.7331453728687054e-05,
-      "loss": 0.2529,
+      "loss": 0.2521,
       "step": 578
     },
     {
       "epoch": 0.37056,
-      "grad_norm": 0.331830692934904,
+      "grad_norm": 0.31976646891891425,
       "learning_rate": 3.7285357011052394e-05,
-      "loss": 0.2391,
+      "loss": 0.2398,
       "step": 579
     },
     {
       "epoch": 0.3712,
-      "grad_norm": 0.29610508157500476,
+      "grad_norm": 0.30332985531986056,
       "learning_rate": 3.72392051612508e-05,
       "loss": 0.2692,
       "step": 580
     },
     {
       "epoch": 0.37184,
-      "grad_norm": 0.2695864248403297,
+      "grad_norm": 0.27711012195719126,
       "learning_rate": 3.719299838639481e-05,
-      "loss": 0.232,
+      "loss": 0.2317,
       "step": 581
     },
     {
       "epoch": 0.37248,
-      "grad_norm": 0.3011787829747353,
+      "grad_norm": 0.4275716591537019,
       "learning_rate": 3.714673689384346e-05,
-      "loss": 0.2734,
+      "loss": 0.2737,
       "step": 582
     },
     {
       "epoch": 0.37312,
-      "grad_norm": 0.29463703959958826,
+      "grad_norm": 0.3026702601663526,
       "learning_rate": 3.710042089120131e-05,
-      "loss": 0.2482,
+      "loss": 0.2483,
       "step": 583
     },
     {
       "epoch": 0.37376,
-      "grad_norm": 0.2891496575716541,
+      "grad_norm": 0.29631681543119553,
       "learning_rate": 3.705405058631756e-05,
-      "loss": 0.263,
+      "loss": 0.2632,
       "step": 584
     },
     {
       "epoch": 0.3744,
-      "grad_norm": 0.28223121574762366,
+      "grad_norm": 0.34902565005008385,
       "learning_rate": 3.700762618728508e-05,
-      "loss": 0.2566,
+      "loss": 0.2575,
       "step": 585
     },
     {
       "epoch": 0.37504,
-      "grad_norm": 0.30203997125282284,
+      "grad_norm": 0.30672309870176107,
       "learning_rate": 3.696114790243955e-05,
-      "loss": 0.2641,
+      "loss": 0.2645,
       "step": 586
     },
     {
       "epoch": 0.37568,
-      "grad_norm": 0.30418487005364714,
+      "grad_norm": 0.29876169974205874,
       "learning_rate": 3.691461594035841e-05,
-      "loss": 0.2788,
+      "loss": 0.2797,
       "step": 587
     },
     {
       "epoch": 0.37632,
-      "grad_norm": 0.27057259769384845,
+      "grad_norm": 0.2705265427737359,
       "learning_rate": 3.686803050986e-05,
-      "loss": 0.2475,
+      "loss": 0.2488,
       "step": 588
     },
     {
       "epoch": 0.37696,
-      "grad_norm": 0.32392640185038807,
+      "grad_norm": 0.36485954853209873,
       "learning_rate": 3.6821391820002596e-05,
-      "loss": 0.2435,
+      "loss": 0.2445,
       "step": 589
     },
     {
       "epoch": 0.3776,
-      "grad_norm": 0.2594129682407732,
+      "grad_norm": 0.2511551627718809,
       "learning_rate": 3.677470008008352e-05,
-      "loss": 0.2465,
+      "loss": 0.2461,
       "step": 590
     },
     {
       "epoch": 0.37824,
-      "grad_norm": 0.28387705453798534,
+      "grad_norm": 0.3221816057866106,
       "learning_rate": 3.672795549963814e-05,
-      "loss": 0.2456,
+      "loss": 0.2459,
       "step": 591
     },
     {
       "epoch": 0.37888,
-      "grad_norm": 0.2950074433710385,
+      "grad_norm": 0.29469731622849443,
       "learning_rate": 3.668115828843894e-05,
-      "loss": 0.2501,
+      "loss": 0.2507,
       "step": 592
     },
     {
       "epoch": 0.37952,
-      "grad_norm": 0.281455028183716,
+      "grad_norm": 0.2752002164016696,
       "learning_rate": 3.6634308656494605e-05,
-      "loss": 0.2507,
+      "loss": 0.2515,
       "step": 593
     },
     {
       "epoch": 0.38016,
-      "grad_norm": 0.29396350647979413,
+      "grad_norm": 0.2937245665619183,
       "learning_rate": 3.6587406814049075e-05,
-      "loss": 0.2408,
+      "loss": 0.2414,
       "step": 594
     },
     {
       "epoch": 0.3808,
-      "grad_norm": 0.27711962772675214,
+      "grad_norm": 0.2623481228772293,
       "learning_rate": 3.6540452971580566e-05,
-      "loss": 0.2586,
+      "loss": 0.2592,
       "step": 595
     },
     {
       "epoch": 0.38144,
-      "grad_norm": 0.294567573462565,
+      "grad_norm": 0.3278516786328869,
       "learning_rate": 3.649344733980068e-05,
-      "loss": 0.2552,
+      "loss": 0.2558,
       "step": 596
     },
     {
       "epoch": 0.38208,
-      "grad_norm": 0.27673346556360123,
+      "grad_norm": 0.30694577299986786,
       "learning_rate": 3.6446390129653405e-05,
-      "loss": 0.2498,
+      "loss": 0.2508,
       "step": 597
     },
     {
       "epoch": 0.38272,
-      "grad_norm": 0.2818818939618678,
+      "grad_norm": 0.2754762052588,
       "learning_rate": 3.63992815523142e-05,
       "loss": 0.2411,
       "step": 598
     },
     {
       "epoch": 0.38336,
-      "grad_norm": 0.29234462981277415,
+      "grad_norm": 0.2927492207692471,
       "learning_rate": 3.6352121819189075e-05,
-      "loss": 0.2425,
+      "loss": 0.2426,
       "step": 599
     },
     {
       "epoch": 0.384,
-      "grad_norm": 0.3269998641272275,
+      "grad_norm": 0.33037655660389065,
       "learning_rate": 3.630491114191354e-05,
-      "loss": 0.2405,
+      "loss": 0.2415,
       "step": 600
     },
     {
       "epoch": 0.38464,
-      "grad_norm": 0.3001209139639063,
+      "grad_norm": 0.29805367907913555,
       "learning_rate": 3.62576497323518e-05,
-      "loss": 0.2793,
+      "loss": 0.2803,
       "step": 601
     },
     {
       "epoch": 0.38528,
-      "grad_norm": 0.28134693156999974,
+      "grad_norm": 0.27588386664863995,
       "learning_rate": 3.621033780259566e-05,
-      "loss": 0.2388,
+      "loss": 0.2385,
       "step": 602
     },
     {
       "epoch": 0.38592,
-      "grad_norm": 0.3244703748522845,
+      "grad_norm": 0.32212522458208825,
       "learning_rate": 3.616297556496369e-05,
-      "loss": 0.238,
+      "loss": 0.2387,
       "step": 603
     },
     {
       "epoch": 0.38656,
-      "grad_norm": 0.2637566816735972,
+      "grad_norm": 0.29947124111938617,
       "learning_rate": 3.6115563232000206e-05,
-      "loss": 0.2354,
+      "loss": 0.2357,
       "step": 604
     },
     {
       "epoch": 0.3872,
-      "grad_norm": 0.28762941282849863,
+      "grad_norm": 0.27587166347065994,
       "learning_rate": 3.6068101016474314e-05,
-      "loss": 0.237,
+      "loss": 0.2386,
       "step": 605
     },
     {
       "epoch": 0.38784,
-      "grad_norm": 0.29578283278832357,
+      "grad_norm": 0.3128647806581742,
       "learning_rate": 3.602058913137902e-05,
-      "loss": 0.2536,
+      "loss": 0.2549,
       "step": 606
     },
     {
       "epoch": 0.38848,
-      "grad_norm": 0.25838120579268425,
+      "grad_norm": 0.2629559804900656,
       "learning_rate": 3.597302778993018e-05,
-      "loss": 0.2536,
+      "loss": 0.2544,
       "step": 607
     },
     {
       "epoch": 0.38912,
-      "grad_norm": 0.29178349766440004,
+      "grad_norm": 0.29509601479077685,
       "learning_rate": 3.592541720556561e-05,
-      "loss": 0.2468,
+      "loss": 0.247,
       "step": 608
     },
     {
       "epoch": 0.38976,
-      "grad_norm": 0.28538733473554107,
+      "grad_norm": 0.2750936321953339,
       "learning_rate": 3.5877757591944137e-05,
-      "loss": 0.2543,
+      "loss": 0.2551,
       "step": 609
     },
     {
       "epoch": 0.3904,
-      "grad_norm": 0.28910068675727285,
+      "grad_norm": 0.2794601458455836,
       "learning_rate": 3.583004916294458e-05,
       "loss": 0.2587,
       "step": 610
     },
     {
       "epoch": 0.39104,
-      "grad_norm": 0.27742239985535233,
+      "grad_norm": 0.27724909989488333,
       "learning_rate": 3.578229213266483e-05,
-      "loss": 0.2526,
+      "loss": 0.2532,
       "step": 611
     },
     {
       "epoch": 0.39168,
-      "grad_norm": 0.294115969883614,
+      "grad_norm": 0.2959448489747847,
       "learning_rate": 3.573448671542088e-05,
-      "loss": 0.2502,
+      "loss": 0.2507,
       "step": 612
     },
     {
       "epoch": 0.39232,
-      "grad_norm": 0.29258891729454395,
+      "grad_norm": 0.3149037372365962,
       "learning_rate": 3.568663312574589e-05,
-      "loss": 0.2351,
+      "loss": 0.2359,
       "step": 613
     },
     {
       "epoch": 0.39296,
-      "grad_norm": 0.2851564904354055,
+      "grad_norm": 0.28544817752125,
       "learning_rate": 3.563873157838917e-05,
-      "loss": 0.2585,
+      "loss": 0.2591,
       "step": 614
     },
     {
       "epoch": 0.3936,
-      "grad_norm": 0.2810424321596881,
+      "grad_norm": 0.2821951863213839,
       "learning_rate": 3.5590782288315266e-05,
-      "loss": 0.2703,
+      "loss": 0.271,
       "step": 615
     },
     {
       "epoch": 0.39424,
-      "grad_norm": 0.2963400750722823,
+      "grad_norm": 0.3062006780490601,
       "learning_rate": 3.5542785470702965e-05,
-      "loss": 0.2583,
+      "loss": 0.2584,
       "step": 616
     },
     {
       "epoch": 0.39488,
-      "grad_norm": 0.28554352127230415,
+      "grad_norm": 0.26982808623927396,
       "learning_rate": 3.549474134094434e-05,
       "loss": 0.2607,
       "step": 617
     },
     {
       "epoch": 0.39552,
-      "grad_norm": 0.2751785354484056,
+      "grad_norm": 0.28882535101383605,
       "learning_rate": 3.54466501146438e-05,
-      "loss": 0.2423,
+      "loss": 0.2439,
       "step": 618
     },
     {
       "epoch": 0.39616,
-      "grad_norm": 0.3109111333105493,
+      "grad_norm": 0.3054230159989776,
       "learning_rate": 3.539851200761707e-05,
-      "loss": 0.2826,
+      "loss": 0.283,
       "step": 619
     },
     {
       "epoch": 0.3968,
-      "grad_norm": 0.28380724524869966,
+      "grad_norm": 0.2803994922266339,
       "learning_rate": 3.535032723589031e-05,
-      "loss": 0.2382,
+      "loss": 0.2401,
       "step": 620
     },
     {
       "epoch": 0.39744,
-      "grad_norm": 0.29031803555316743,
+      "grad_norm": 0.2829193733985107,
       "learning_rate": 3.5302096015699035e-05,
-      "loss": 0.2529,
+      "loss": 0.2532,
       "step": 621
     },
     {
       "epoch": 0.39808,
-      "grad_norm": 0.2862319408791201,
+      "grad_norm": 0.29493309388688765,
       "learning_rate": 3.5253818563487266e-05,
-      "loss": 0.2497,
+      "loss": 0.2502,
       "step": 622
     },
     {
       "epoch": 0.39872,
-      "grad_norm": 0.2597359081273056,
+      "grad_norm": 0.2616815315666944,
       "learning_rate": 3.5205495095906444e-05,
-      "loss": 0.2502,
+      "loss": 0.2515,
       "step": 623
     },
     {
       "epoch": 0.39936,
-      "grad_norm": 0.26645746598257863,
+      "grad_norm": 0.2679431721589359,
       "learning_rate": 3.5157125829814534e-05,
-      "loss": 0.2427,
+      "loss": 0.2436,
       "step": 624
     },
     {
       "epoch": 0.4,
-      "grad_norm": 0.2589738737012416,
+      "grad_norm": 0.25554211659718934,
       "learning_rate": 3.510871098227503e-05,
-      "loss": 0.2314,
+      "loss": 0.2324,
       "step": 625
     },
     {
       "epoch": 0.40064,
-      "grad_norm": 0.2808606689504793,
+      "grad_norm": 0.2733276800637646,
       "learning_rate": 3.506025077055597e-05,
-      "loss": 0.2906,
+      "loss": 0.2914,
       "step": 626
     },
     {
       "epoch": 0.40128,
-      "grad_norm": 0.27935188272224154,
+      "grad_norm": 0.26328591986272304,
       "learning_rate": 3.501174541212898e-05,
-      "loss": 0.2419,
+      "loss": 0.2424,
       "step": 627
     },
     {
       "epoch": 0.40192,
-      "grad_norm": 0.2907671332799639,
+      "grad_norm": 0.2644108716838522,
       "learning_rate": 3.496319512466827e-05,
-      "loss": 0.2596,
+      "loss": 0.2555,
       "step": 628
     },
     {
       "epoch": 0.40256,
-      "grad_norm": 0.2844727991856023,
+      "grad_norm": 0.27240010844795565,
       "learning_rate": 3.4914600126049696e-05,
-      "loss": 0.2459,
+      "loss": 0.2458,
       "step": 629
     },
     {
       "epoch": 0.4032,
-      "grad_norm": 0.2719868034683634,
+      "grad_norm": 0.275592750641108,
       "learning_rate": 3.486596063434975e-05,
-      "loss": 0.2431,
+      "loss": 0.243,
       "step": 630
     },
     {
       "epoch": 0.40384,
-      "grad_norm": 0.2823824998273708,
+      "grad_norm": 0.28461986974911496,
       "learning_rate": 3.481727686784458e-05,
-      "loss": 0.2445,
+      "loss": 0.2442,
       "step": 631
     },
     {
       "epoch": 0.40448,
-      "grad_norm": 0.2756425905858028,
+      "grad_norm": 0.38358418504867753,
       "learning_rate": 3.476854904500905e-05,
-      "loss": 0.2463,
+      "loss": 0.2476,
       "step": 632
     },
     {
       "epoch": 0.40512,
-      "grad_norm": 0.2779852039855442,
+      "grad_norm": 0.27772995022705443,
       "learning_rate": 3.471977738451572e-05,
-      "loss": 0.2534,
+      "loss": 0.2532,
       "step": 633
     },
     {
       "epoch": 0.40576,
-      "grad_norm": 0.2950432225935468,
+      "grad_norm": 0.3063163796379174,
       "learning_rate": 3.467096210523387e-05,
-      "loss": 0.2532,
+      "loss": 0.2528,
       "step": 634
     },
     {
       "epoch": 0.4064,
-      "grad_norm": 0.24821255764789482,
+      "grad_norm": 0.2558103094662568,
       "learning_rate": 3.4622103426228534e-05,
-      "loss": 0.2616,
+      "loss": 0.2617,
       "step": 635
     },
     {
       "epoch": 0.40704,
-      "grad_norm": 0.2665965175038227,
+      "grad_norm": 0.27035171229910127,
       "learning_rate": 3.4573201566759515e-05,
-      "loss": 0.2451,
+      "loss": 0.2459,
       "step": 636
     },
     {
       "epoch": 0.40768,
-      "grad_norm": 0.2944539832776922,
+      "grad_norm": 0.2829589875667949,
       "learning_rate": 3.4524256746280374e-05,
-      "loss": 0.2546,
+      "loss": 0.2553,
       "step": 637
     },
     {
       "epoch": 0.40832,
-      "grad_norm": 0.26681868196558595,
+      "grad_norm": 0.276488759483485,
       "learning_rate": 3.447526918443748e-05,
-      "loss": 0.2489,
+      "loss": 0.2494,
       "step": 638
     },
     {
       "epoch": 0.40896,
-      "grad_norm": 0.2700230642277879,
+      "grad_norm": 0.2785085755001678,
       "learning_rate": 3.442623910106901e-05,
-      "loss": 0.2297,
+      "loss": 0.2308,
       "step": 639
     },
     {
       "epoch": 0.4096,
-      "grad_norm": 0.25839557803313135,
+      "grad_norm": 0.2582594146057901,
       "learning_rate": 3.4377166716203945e-05,
-      "loss": 0.2397,
+      "loss": 0.2401,
       "step": 640
     },
     {
       "epoch": 0.41024,
-      "grad_norm": 0.3021408976030671,
+      "grad_norm": 0.29532893078531347,
       "learning_rate": 3.4328052250061124e-05,
-      "loss": 0.2455,
+      "loss": 0.2464,
       "step": 641
     },
     {
       "epoch": 0.41088,
-      "grad_norm": 0.2859479339462905,
+      "grad_norm": 0.29066328186544355,
       "learning_rate": 3.4278895923048226e-05,
-      "loss": 0.2547,
+      "loss": 0.2556,
       "step": 642
     },
     {
       "epoch": 0.41152,
-      "grad_norm": 0.2745261932057034,
+      "grad_norm": 0.2687576511766121,
       "learning_rate": 3.4229697955760767e-05,
-      "loss": 0.2293,
+      "loss": 0.2287,
       "step": 643
     },
     {
       "epoch": 0.41216,
-      "grad_norm": 0.2909265259735638,
+      "grad_norm": 0.2835568843137537,
       "learning_rate": 3.4180458568981136e-05,
-      "loss": 0.2788,
+      "loss": 0.2797,
       "step": 644
     },
     {
       "epoch": 0.4128,
-      "grad_norm": 0.25620787622393093,
+      "grad_norm": 0.2629010441848825,
       "learning_rate": 3.4131177983677624e-05,
-      "loss": 0.2258,
+      "loss": 0.226,
       "step": 645
     },
     {
       "epoch": 0.41344,
-      "grad_norm": 0.2838977519379636,
+      "grad_norm": 0.28062825574605843,
       "learning_rate": 3.4081856421003364e-05,
-      "loss": 0.2364,
+      "loss": 0.2367,
       "step": 646
     },
     {
       "epoch": 0.41408,
-      "grad_norm": 0.2779231236669113,
+      "grad_norm": 0.270318316756242,
       "learning_rate": 3.403249410229542e-05,
-      "loss": 0.2465,
+      "loss": 0.2467,
       "step": 647
     },
     {
       "epoch": 0.41472,
-      "grad_norm": 0.33020342043739015,
+      "grad_norm": 0.316058725045681,
       "learning_rate": 3.398309124907372e-05,
-      "loss": 0.2562,
+      "loss": 0.2567,
       "step": 648
     },
     {
       "epoch": 0.41536,
-      "grad_norm": 0.24666782803318293,
+      "grad_norm": 0.24447817887936557,
       "learning_rate": 3.393364808304012e-05,
-      "loss": 0.2712,
+      "loss": 0.2718,
       "step": 649
     },
     {
       "epoch": 0.416,
-      "grad_norm": 0.30815484394945014,
+      "grad_norm": 0.2894191788618239,
       "learning_rate": 3.3884164826077384e-05,
-      "loss": 0.2456,
+      "loss": 0.2449,
       "step": 650
     },
     {
       "epoch": 0.41664,
-      "grad_norm": 0.2812813367796177,
+      "grad_norm": 0.2912762195649435,
       "learning_rate": 3.383464170024817e-05,
       "loss": 0.2441,
       "step": 651
     },
     {
       "epoch": 0.41728,
-      "grad_norm": 0.28607485450453224,
+      "grad_norm": 0.28705421045348156,
       "learning_rate": 3.378507892779408e-05,
-      "loss": 0.2701,
+      "loss": 0.2707,
       "step": 652
     },
     {
       "epoch": 0.41792,
-      "grad_norm": 0.2960472952173197,
+      "grad_norm": 0.29055146905014817,
       "learning_rate": 3.37354767311346e-05,
-      "loss": 0.2532,
+      "loss": 0.2536,
       "step": 653
     },
     {
       "epoch": 0.41856,
-      "grad_norm": 0.2832878309358752,
+      "grad_norm": 0.26018867708452015,
       "learning_rate": 3.368583533286617e-05,
-      "loss": 0.235,
+      "loss": 0.2355,
       "step": 654
     },
     {
       "epoch": 0.4192,
-      "grad_norm": 0.2906843275840211,
+      "grad_norm": 0.26817580454089857,
       "learning_rate": 3.363615495576115e-05,
-      "loss": 0.2564,
+      "loss": 0.2566,
       "step": 655
     },
     {
       "epoch": 0.41984,
-      "grad_norm": 0.2790824461789213,
+      "grad_norm": 0.2762550908141269,
       "learning_rate": 3.3586435822766785e-05,
-      "loss": 0.2568,
+      "loss": 0.2566,
       "step": 656
     },
     {
       "epoch": 0.42048,
-      "grad_norm": 0.28446065705330625,
+      "grad_norm": 0.27020213820169053,
       "learning_rate": 3.35366781570043e-05,
-      "loss": 0.2363,
+      "loss": 0.2359,
       "step": 657
     },
     {
       "epoch": 0.42112,
-      "grad_norm": 0.3157219415215847,
+      "grad_norm": 0.302439548213349,
       "learning_rate": 3.3486882181767785e-05,
-      "loss": 0.2606,
+      "loss": 0.2548,
       "step": 658
     },
     {
       "epoch": 0.42176,
-      "grad_norm": 0.3056369034770089,
+      "grad_norm": 0.300389758723506,
       "learning_rate": 3.34370481205233e-05,
-      "loss": 0.2617,
+      "loss": 0.2618,
       "step": 659
     },
     {
       "epoch": 0.4224,
-      "grad_norm": 0.2766010888311108,
+      "grad_norm": 0.2713457188069892,
       "learning_rate": 3.3387176196907776e-05,
-      "loss": 0.2609,
+      "loss": 0.2607,
       "step": 660
     },
     {
       "epoch": 0.42304,
-      "grad_norm": 0.2545434388799889,
+      "grad_norm": 0.29789023190014424,
       "learning_rate": 3.333726663472809e-05,
-      "loss": 0.2414,
+      "loss": 0.2415,
       "step": 661
     },
     {
       "epoch": 0.42368,
-      "grad_norm": 0.27376260910857747,
+      "grad_norm": 0.2773543670343778,
       "learning_rate": 3.328731965796e-05,
-      "loss": 0.244,
+      "loss": 0.2446,
       "step": 662
     },
     {
       "epoch": 0.42432,
-      "grad_norm": 0.2840566587736982,
+      "grad_norm": 0.2783765933442811,
       "learning_rate": 3.32373354907472e-05,
-      "loss": 0.2336,
+      "loss": 0.2339,
       "step": 663
     },
     {
       "epoch": 0.42496,
-      "grad_norm": 0.27188917531513446,
+      "grad_norm": 0.25339658241046037,
       "learning_rate": 3.3187314357400244e-05,
-      "loss": 0.2344,
+      "loss": 0.2346,
       "step": 664
     },
     {
       "epoch": 0.4256,
-      "grad_norm": 0.265226445811272,
+      "grad_norm": 0.2648778149152874,
       "learning_rate": 3.31372564823956e-05,
-      "loss": 0.2649,
+      "loss": 0.2644,
       "step": 665
     },
     {
       "epoch": 0.42624,
-      "grad_norm": 0.28854870363229795,
+      "grad_norm": 0.2862572466968127,
       "learning_rate": 3.308716209037462e-05,
-      "loss": 0.2538,
+      "loss": 0.2548,
       "step": 666
     },
     {
       "epoch": 0.42688,
-      "grad_norm": 0.2618928575977395,
+      "grad_norm": 0.2586423249267937,
       "learning_rate": 3.303703140614251e-05,
-      "loss": 0.2532,
+      "loss": 0.2531,
       "step": 667
     },
     {
       "epoch": 0.42752,
-      "grad_norm": 0.26410396546652154,
+      "grad_norm": 0.4114872279281381,
       "learning_rate": 3.2986864654667356e-05,
-      "loss": 0.2402,
+      "loss": 0.2405,
       "step": 668
     },
     {
       "epoch": 0.42816,
-      "grad_norm": 0.24094546777729578,
+      "grad_norm": 0.24446086855481872,
       "learning_rate": 3.293666206107911e-05,
-      "loss": 0.2516,
+      "loss": 0.2515,
       "step": 669
     },
     {
       "epoch": 0.4288,
-      "grad_norm": 0.28378779340674526,
+      "grad_norm": 0.2791407834679966,
       "learning_rate": 3.2886423850668555e-05,
-      "loss": 0.2487,
+      "loss": 0.2493,
       "step": 670
     },
     {
       "epoch": 0.42944,
-      "grad_norm": 0.2800453476199491,
+      "grad_norm": 0.2741780919630588,
       "learning_rate": 3.283615024888632e-05,
-      "loss": 0.2404,
+      "loss": 0.2402,
       "step": 671
     },
     {
       "epoch": 0.43008,
-      "grad_norm": 0.28987843294031523,
+      "grad_norm": 0.293985795428857,
       "learning_rate": 3.278584148134185e-05,
-      "loss": 0.2677,
+      "loss": 0.2682,
       "step": 672
     },
     {
       "epoch": 0.43072,
-      "grad_norm": 0.29824090133191716,
+      "grad_norm": 0.2915497388084258,
       "learning_rate": 3.2735497773802406e-05,
-      "loss": 0.2682,
+      "loss": 0.2686,
       "step": 673
     },
     {
       "epoch": 0.43136,
-      "grad_norm": 0.2718813071517211,
+      "grad_norm": 0.2794624861425458,
       "learning_rate": 3.268511935219204e-05,
-      "loss": 0.241,
+      "loss": 0.2418,
       "step": 674
     },
     {
       "epoch": 0.432,
-      "grad_norm": 0.3320650785911746,
+      "grad_norm": 0.2644683704239771,
       "learning_rate": 3.263470644259059e-05,
-      "loss": 0.2341,
+      "loss": 0.2225,
       "step": 675
     },
     {
       "epoch": 0.43264,
-      "grad_norm": 0.31100418326629736,
+      "grad_norm": 0.29839214475582065,
       "learning_rate": 3.258425927123266e-05,
-      "loss": 0.2663,
+      "loss": 0.2665,
       "step": 676
     },
     {
       "epoch": 0.43328,
-      "grad_norm": 0.2749981217880558,
+      "grad_norm": 0.2804235795912977,
       "learning_rate": 3.253377806450664e-05,
-      "loss": 0.2485,
+      "loss": 0.2491,
       "step": 677
     },
     {
       "epoch": 0.43392,
-      "grad_norm": 0.2491062102162234,
+      "grad_norm": 0.2552634500085142,
       "learning_rate": 3.248326304895361e-05,
-      "loss": 0.2258,
+      "loss": 0.2256,
       "step": 678
     },
     {
       "epoch": 0.43456,
-      "grad_norm": 0.27061105202173463,
+      "grad_norm": 0.27227200788700007,
       "learning_rate": 3.2432714451266395e-05,
       "loss": 0.2609,
       "step": 679
     },
     {
       "epoch": 0.4352,
-      "grad_norm": 0.2855069550319216,
+      "grad_norm": 0.27105089367623647,
       "learning_rate": 3.238213249828853e-05,
-      "loss": 0.2582,
+      "loss": 0.2587,
       "step": 680
     },
     {
       "epoch": 0.43584,
-      "grad_norm": 0.2705470532603105,
+      "grad_norm": 0.26509214505181616,
       "learning_rate": 3.233151741701322e-05,
-      "loss": 0.2501,
+      "loss": 0.2498,
       "step": 681
     },
     {
       "epoch": 0.43648,
-      "grad_norm": 0.25333939918230564,
+      "grad_norm": 0.2587788903976874,
       "learning_rate": 3.228086943458235e-05,
-      "loss": 0.2563,
+      "loss": 0.2562,
       "step": 682
     },
     {
       "epoch": 0.43712,
-      "grad_norm": 0.24493528767421877,
+      "grad_norm": 0.25091179033958944,
       "learning_rate": 3.223018877828546e-05,
-      "loss": 0.2207,
+      "loss": 0.2214,
       "step": 683
     },
     {
       "epoch": 0.43776,
-      "grad_norm": 0.27279546295316043,
+      "grad_norm": 0.2793253170166926,
       "learning_rate": 3.2179475675558694e-05,
-      "loss": 0.2449,
+      "loss": 0.2463,
       "step": 684
     },
     {
       "epoch": 0.4384,
-      "grad_norm": 0.2479820800381958,
+      "grad_norm": 0.2407880517047661,
       "learning_rate": 3.212873035398383e-05,
-      "loss": 0.249,
+      "loss": 0.2493,
       "step": 685
     },
     {
       "epoch": 0.43904,
-      "grad_norm": 0.2646928915495465,
+      "grad_norm": 0.26757070589299814,
       "learning_rate": 3.207795304128721e-05,
-      "loss": 0.2287,
+      "loss": 0.2293,
       "step": 686
     },
     {
       "epoch": 0.43968,
-      "grad_norm": 0.26738698042119885,
+      "grad_norm": 0.26723528784093614,
       "learning_rate": 3.2027143965338765e-05,
-      "loss": 0.2483,
+      "loss": 0.2484,
       "step": 687
     },
     {
       "epoch": 0.44032,
-      "grad_norm": 0.2741701894109012,
+      "grad_norm": 0.2656705855889354,
       "learning_rate": 3.1976303354150935e-05,
-      "loss": 0.2244,
+      "loss": 0.2236,
       "step": 688
     },
     {
       "epoch": 0.44096,
-      "grad_norm": 0.23871128665206856,
+      "grad_norm": 0.2439985371889559,
       "learning_rate": 3.192543143587771e-05,
-      "loss": 0.2331,
+      "loss": 0.2342,
       "step": 689
     },
     {
       "epoch": 0.4416,
-      "grad_norm": 0.24189532225791055,
+      "grad_norm": 0.26235220172661927,
       "learning_rate": 3.187452843881357e-05,
-      "loss": 0.2476,
+      "loss": 0.248,
       "step": 690
     },
     {
       "epoch": 0.44224,
-      "grad_norm": 0.25539624756502316,
+      "grad_norm": 0.25083797196574825,
       "learning_rate": 3.182359459139243e-05,
-      "loss": 0.2266,
+      "loss": 0.2262,
       "step": 691
     },
     {
       "epoch": 0.44288,
-      "grad_norm": 0.2756480505672449,
+      "grad_norm": 0.3227207880943149,
       "learning_rate": 3.1772630122186695e-05,
-      "loss": 0.2487,
+      "loss": 0.249,
       "step": 692
     },
     {
       "epoch": 0.44352,
-      "grad_norm": 0.2668909881136519,
+      "grad_norm": 0.2510582734287034,
       "learning_rate": 3.172163525990617e-05,
-      "loss": 0.2646,
+      "loss": 0.2642,
       "step": 693
     },
     {
       "epoch": 0.44416,
-      "grad_norm": 0.2371652011452226,
+      "grad_norm": 0.23405484365904264,
       "learning_rate": 3.167061023339706e-05,
-      "loss": 0.2497,
+      "loss": 0.25,
       "step": 694
     },
     {
       "epoch": 0.4448,
-      "grad_norm": 0.25390001109163235,
+      "grad_norm": 0.25693251630337804,
       "learning_rate": 3.161955527164092e-05,
-      "loss": 0.2508,
+      "loss": 0.2509,
       "step": 695
     },
     {
       "epoch": 0.44544,
-      "grad_norm": 0.2402366710016075,
+      "grad_norm": 0.24145057284019783,
       "learning_rate": 3.1568470603753666e-05,
-      "loss": 0.236,
+      "loss": 0.2369,
       "step": 696
     },
     {
       "epoch": 0.44608,
-      "grad_norm": 0.2784039809880136,
+      "grad_norm": 0.2656422680687307,
       "learning_rate": 3.1517356458984485e-05,
-      "loss": 0.2478,
+      "loss": 0.2483,
       "step": 697
     },
     {
       "epoch": 0.44672,
-      "grad_norm": 0.24727119521104232,
+      "grad_norm": 0.2511244867492243,
       "learning_rate": 3.14662130667149e-05,
-      "loss": 0.2217,
+      "loss": 0.2218,
       "step": 698
     },
     {
       "epoch": 0.44736,
-      "grad_norm": 0.27458593279014226,
+      "grad_norm": 0.27645698946904024,
       "learning_rate": 3.141504065645765e-05,
-      "loss": 0.2589,
+      "loss": 0.2597,
       "step": 699
     },
     {
       "epoch": 0.448,
-      "grad_norm": 0.24446850331658782,
+      "grad_norm": 0.2345290138656015,
       "learning_rate": 3.1363839457855704e-05,
-      "loss": 0.2475,
+      "loss": 0.2471,
       "step": 700
     },
     {
       "epoch": 0.44864,
-      "grad_norm": 0.2559445330730872,
+      "grad_norm": 0.2550749778653545,
       "learning_rate": 3.131260970068122e-05,
-      "loss": 0.2469,
+      "loss": 0.2474,
       "step": 701
     },
     {
       "epoch": 0.44928,
-      "grad_norm": 0.26623620144423177,
+      "grad_norm": 0.26631197327016853,
       "learning_rate": 3.1261351614834536e-05,
-      "loss": 0.2382,
+      "loss": 0.2383,
       "step": 702
     },
     {
       "epoch": 0.44992,
-      "grad_norm": 0.22387150133617711,
+      "grad_norm": 0.22169396341875647,
       "learning_rate": 3.121006543034307e-05,
       "loss": 0.2601,
       "step": 703
     },
     {
       "epoch": 0.45056,
-      "grad_norm": 0.23724739980041212,
+      "grad_norm": 0.2388958834756383,
       "learning_rate": 3.11587513773604e-05,
-      "loss": 0.222,
+      "loss": 0.2224,
       "step": 704
     },
     {
       "epoch": 0.4512,
-      "grad_norm": 0.25564336027468965,
+      "grad_norm": 0.24836043244890216,
       "learning_rate": 3.1107409686165105e-05,
       "loss": 0.2499,
       "step": 705
     },
     {
       "epoch": 0.45184,
-      "grad_norm": 0.2530073332441141,
+      "grad_norm": 0.2641275652959496,
       "learning_rate": 3.1056040587159856e-05,
-      "loss": 0.2266,
+      "loss": 0.2273,
       "step": 706
     },
     {
       "epoch": 0.45248,
-      "grad_norm": 0.2426770783855302,
+      "grad_norm": 0.2653747554412634,
       "learning_rate": 3.100464431087027e-05,
-      "loss": 0.235,
+      "loss": 0.2348,
       "step": 707
     },
     {
       "epoch": 0.45312,
-      "grad_norm": 0.24939202296967658,
+      "grad_norm": 0.2554298939523283,
       "learning_rate": 3.095322108794396e-05,
-      "loss": 0.2346,
+      "loss": 0.2352,
       "step": 708
     },
     {
       "epoch": 0.45376,
-      "grad_norm": 0.2670752936110549,
+      "grad_norm": 0.26559974477527737,
       "learning_rate": 3.0901771149149434e-05,
-      "loss": 0.2269,
+      "loss": 0.227,
       "step": 709
     },
     {
       "epoch": 0.4544,
-      "grad_norm": 0.24835502202214765,
+      "grad_norm": 0.24105288644141198,
       "learning_rate": 3.0850294725375105e-05,
-      "loss": 0.245,
+      "loss": 0.2456,
       "step": 710
     },
     {
       "epoch": 0.45504,
-      "grad_norm": 0.2398459646425662,
+      "grad_norm": 0.25639595036381535,
       "learning_rate": 3.079879204762827e-05,
-      "loss": 0.2329,
+      "loss": 0.2335,
       "step": 711
     },
     {
       "epoch": 0.45568,
-      "grad_norm": 0.2579451505404363,
+      "grad_norm": 0.2666175090966313,
       "learning_rate": 3.074726334703399e-05,
       "loss": 0.2375,
       "step": 712
     },
     {
       "epoch": 0.45632,
-      "grad_norm": 0.23992084217792944,
+      "grad_norm": 0.24609688887166636,
       "learning_rate": 3.069570885483414e-05,
-      "loss": 0.2493,
+      "loss": 0.2497,
       "step": 713
     },
     {
       "epoch": 0.45696,
-      "grad_norm": 0.26286173696168813,
+      "grad_norm": 0.26685315906241225,
       "learning_rate": 3.064412880238636e-05,
-      "loss": 0.249,
+      "loss": 0.2498,
       "step": 714
     },
     {
       "epoch": 0.4576,
-      "grad_norm": 0.26001195813349676,
+      "grad_norm": 0.2688506170038658,
       "learning_rate": 3.059252342116293e-05,
-      "loss": 0.248,
+      "loss": 0.2481,
       "step": 715
     },
     {
       "epoch": 0.45824,
-      "grad_norm": 0.2454394896408352,
+      "grad_norm": 0.2594010390645574,
       "learning_rate": 3.054089294274985e-05,
-      "loss": 0.2461,
+      "loss": 0.2468,
       "step": 716
     },
     {
       "epoch": 0.45888,
-      "grad_norm": 0.2740552510767714,
+      "grad_norm": 0.28084138121726687,
       "learning_rate": 3.048923759884573e-05,
-      "loss": 0.2363,
+      "loss": 0.2362,
       "step": 717
     },
     {
       "epoch": 0.45952,
-      "grad_norm": 0.28565496441503085,
+      "grad_norm": 0.2923747466796738,
       "learning_rate": 3.043755762126077e-05,
-      "loss": 0.2621,
+      "loss": 0.2619,
       "step": 718
     },
     {
       "epoch": 0.46016,
-      "grad_norm": 0.2558300796052746,
+      "grad_norm": 0.2575669206483763,
       "learning_rate": 3.038585324191571e-05,
-      "loss": 0.2526,
+      "loss": 0.2529,
       "step": 719
     },
     {
       "epoch": 0.4608,
-      "grad_norm": 0.2560629430046937,
+      "grad_norm": 0.24997801681198326,
       "learning_rate": 3.0334124692840815e-05,
       "loss": 0.2558,
       "step": 720
     },
     {
       "epoch": 0.46144,
-      "grad_norm": 0.2722189672194111,
+      "grad_norm": 0.28797967225037385,
       "learning_rate": 3.0282372206174785e-05,
-      "loss": 0.2496,
+      "loss": 0.2498,
       "step": 721
     },
     {
       "epoch": 0.46208,
-      "grad_norm": 0.3113674681390855,
+      "grad_norm": 0.29276072080394017,
       "learning_rate": 3.023059601416376e-05,
-      "loss": 0.2493,
+      "loss": 0.2492,
       "step": 722
     },
     {
       "epoch": 0.46272,
-      "grad_norm": 0.23312450422876216,
+      "grad_norm": 0.24024625168643718,
       "learning_rate": 3.017879634916027e-05,
-      "loss": 0.2394,
+      "loss": 0.24,
       "step": 723
     },
     {
       "epoch": 0.46336,
-      "grad_norm": 0.2473881664409505,
+      "grad_norm": 0.250522628591233,
       "learning_rate": 3.012697344362217e-05,
-      "loss": 0.2349,
+      "loss": 0.2354,
       "step": 724
     },
     {
       "epoch": 0.464,
-      "grad_norm": 0.2661149365212982,
+      "grad_norm": 0.28028684047213753,
       "learning_rate": 3.0075127530111606e-05,
-      "loss": 0.2242,
+      "loss": 0.2248,
       "step": 725
     },
     {
       "epoch": 0.46464,
-      "grad_norm": 0.2747099671427922,
+      "grad_norm": 0.2791381712788782,
       "learning_rate": 3.0023258841294e-05,
-      "loss": 0.242,
+      "loss": 0.2425,
       "step": 726
     },
     {
       "epoch": 0.46528,
-      "grad_norm": 0.257152761339917,
+      "grad_norm": 0.2659960218700461,
       "learning_rate": 2.9971367609936945e-05,
-      "loss": 0.2522,
+      "loss": 0.2526,
       "step": 727
     },
     {
       "epoch": 0.46592,
-      "grad_norm": 0.27242290443778083,
+      "grad_norm": 0.2772736125798831,
       "learning_rate": 2.9919454068909218e-05,
-      "loss": 0.2245,
+      "loss": 0.2247,
       "step": 728
     },
     {
       "epoch": 0.46656,
-      "grad_norm": 0.31224799275083753,
+      "grad_norm": 0.3124309146865046,
       "learning_rate": 2.986751845117972e-05,
-      "loss": 0.2562,
+      "loss": 0.2569,
       "step": 729
     },
     {
       "epoch": 0.4672,
-      "grad_norm": 0.2578554546888466,
+      "grad_norm": 0.2626897535417429,
       "learning_rate": 2.981556098981641e-05,
-      "loss": 0.2805,
+      "loss": 0.2806,
       "step": 730
     },
     {
       "epoch": 0.46784,
-      "grad_norm": 0.3024846857246908,
+      "grad_norm": 0.29956166216034524,
       "learning_rate": 2.9763581917985274e-05,
-      "loss": 0.2632,
+      "loss": 0.2636,
       "step": 731
     },
     {
       "epoch": 0.46848,
-      "grad_norm": 0.2824166641987713,
+      "grad_norm": 0.2911940048402578,
       "learning_rate": 2.9711581468949295e-05,
-      "loss": 0.2579,
+      "loss": 0.2581,
       "step": 732
     },
     {
       "epoch": 0.46912,
-      "grad_norm": 0.2944097917138229,
+      "grad_norm": 0.3066474268110775,
       "learning_rate": 2.965955987606736e-05,
-      "loss": 0.2716,
+      "loss": 0.2713,
       "step": 733
     },
     {
       "epoch": 0.46976,
-      "grad_norm": 0.2841807990688873,
+      "grad_norm": 0.2948399561745436,
       "learning_rate": 2.9607517372793268e-05,
-      "loss": 0.2498,
+      "loss": 0.2504,
       "step": 734
     },
     {
       "epoch": 0.4704,
-      "grad_norm": 0.3201603618025787,
+      "grad_norm": 0.29922556962835833,
       "learning_rate": 2.9555454192674638e-05,
-      "loss": 0.2498,
+      "loss": 0.2506,
       "step": 735
     },
     {
       "epoch": 0.47104,
-      "grad_norm": 0.29583253286832584,
+      "grad_norm": 0.3090688707981366,
       "learning_rate": 2.9503370569351895e-05,
-      "loss": 0.2611,
+      "loss": 0.2616,
       "step": 736
     },
     {
       "epoch": 0.47168,
-      "grad_norm": 0.28051105666181864,
+      "grad_norm": 0.2761500484620512,
       "learning_rate": 2.9451266736557202e-05,
-      "loss": 0.2575,
+      "loss": 0.2577,
       "step": 737
     },
     {
       "epoch": 0.47232,
-      "grad_norm": 0.24888601598003757,
+      "grad_norm": 0.254217878635331,
       "learning_rate": 2.9399142928113403e-05,
       "loss": 0.2189,
       "step": 738
     },
     {
       "epoch": 0.47296,
-      "grad_norm": 0.27307930235300004,
+      "grad_norm": 0.26296653857821706,
       "learning_rate": 2.9346999377932992e-05,
-      "loss": 0.2625,
+      "loss": 0.2634,
       "step": 739
     },
     {
       "epoch": 0.4736,
-      "grad_norm": 0.25703656850177853,
+      "grad_norm": 0.2572341383024019,
       "learning_rate": 2.929483632001706e-05,
-      "loss": 0.2418,
+      "loss": 0.2417,
       "step": 740
     },
     {
       "epoch": 0.47424,
-      "grad_norm": 0.2746230772665275,
+      "grad_norm": 0.28892208828624116,
       "learning_rate": 2.9242653988454237e-05,
-      "loss": 0.2436,
+      "loss": 0.2445,
       "step": 741
     },
     {
       "epoch": 0.47488,
-      "grad_norm": 0.27201723917797194,
+      "grad_norm": 0.28102463284648577,
       "learning_rate": 2.919045261741965e-05,
-      "loss": 0.2379,
+      "loss": 0.2388,
       "step": 742
     },
     {
       "epoch": 0.47552,
-      "grad_norm": 0.2907921867384871,
+      "grad_norm": 0.24662076667210664,
       "learning_rate": 2.9138232441173867e-05,
-      "loss": 0.2387,
+      "loss": 0.2383,
       "step": 743
     },
     {
       "epoch": 0.47616,
-      "grad_norm": 0.2981223972195515,
+      "grad_norm": 0.29423089365699157,
       "learning_rate": 2.9085993694061853e-05,
-      "loss": 0.2427,
+      "loss": 0.2425,
       "step": 744
     },
     {
       "epoch": 0.4768,
-      "grad_norm": 0.28193295528586837,
+      "grad_norm": 0.2957483091439502,
       "learning_rate": 2.9033736610511887e-05,
-      "loss": 0.2613,
+      "loss": 0.2617,
       "step": 745
     },
     {
       "epoch": 0.47744,
-      "grad_norm": 0.2757657774306609,
+      "grad_norm": 0.28663575693261895,
       "learning_rate": 2.898146142503456e-05,
-      "loss": 0.2627,
+      "loss": 0.2633,
       "step": 746
     },
     {
       "epoch": 0.47808,
-      "grad_norm": 0.2653548783721028,
+      "grad_norm": 0.2603089054365686,
       "learning_rate": 2.8929168372221693e-05,
       "loss": 0.2496,
       "step": 747
     },
     {
       "epoch": 0.47872,
-      "grad_norm": 0.29977110233394766,
+      "grad_norm": 0.29032157339351555,
       "learning_rate": 2.88768576867453e-05,
-      "loss": 0.2561,
+      "loss": 0.2562,
       "step": 748
     },
     {
       "epoch": 0.47936,
-      "grad_norm": 0.24436988805469104,
+      "grad_norm": 0.25757922984484416,
       "learning_rate": 2.8824529603356497e-05,
-      "loss": 0.2231,
+      "loss": 0.2233,
       "step": 749
     },
     {
       "epoch": 0.48,
-      "grad_norm": 0.29080555610569137,
+      "grad_norm": 0.29211519919338724,
       "learning_rate": 2.8772184356884496e-05,
       "loss": 0.2602,
       "step": 750
     },
     {
       "epoch": 0.48064,
-      "grad_norm": 0.2357009376696896,
+      "grad_norm": 0.26552159476837556,
       "learning_rate": 2.8719822182235516e-05,
-      "loss": 0.2552,
+      "loss": 0.2551,
       "step": 751
     },
     {
       "epoch": 0.48128,
-      "grad_norm": 0.2612000455514733,
+      "grad_norm": 0.2713416231519419,
       "learning_rate": 2.8667443314391762e-05,
-      "loss": 0.229,
+      "loss": 0.2291,
       "step": 752
     },
     {
       "epoch": 0.48192,
-      "grad_norm": 0.27847227931686186,
+      "grad_norm": 0.2688122633975986,
       "learning_rate": 2.8615047988410337e-05,
-      "loss": 0.2431,
+      "loss": 0.2426,
       "step": 753
     },
     {
       "epoch": 0.48256,
-      "grad_norm": 0.25403155179332854,
+      "grad_norm": 0.2552031893223486,
       "learning_rate": 2.8562636439422202e-05,
-      "loss": 0.2432,
+      "loss": 0.243,
       "step": 754
     },
     {
       "epoch": 0.4832,
-      "grad_norm": 0.2667691204165141,
+      "grad_norm": 0.25644480909672956,
       "learning_rate": 2.8510208902631135e-05,
-      "loss": 0.2509,
+      "loss": 0.2507,
       "step": 755
     },
     {
       "epoch": 0.48384,
-      "grad_norm": 0.2777759815907825,
+      "grad_norm": 0.2650303461926079,
       "learning_rate": 2.845776561331265e-05,
-      "loss": 0.253,
+      "loss": 0.2524,
       "step": 756
     },
     {
       "epoch": 0.48448,
-      "grad_norm": 0.2740727680932021,
+      "grad_norm": 0.2841325794413843,
       "learning_rate": 2.8405306806812947e-05,
-      "loss": 0.224,
+      "loss": 0.2246,
       "step": 757
     },
     {
       "epoch": 0.48512,
-      "grad_norm": 0.2817937944167678,
+      "grad_norm": 0.27963333968219384,
       "learning_rate": 2.835283271854788e-05,
-      "loss": 0.2628,
+      "loss": 0.2627,
       "step": 758
     },
     {
       "epoch": 0.48576,
-      "grad_norm": 0.24766172176538703,
+      "grad_norm": 0.2469872711729761,
       "learning_rate": 2.8300343584001853e-05,
       "loss": 0.2456,
       "step": 759
     },
     {
       "epoch": 0.4864,
-      "grad_norm": 0.24618641032586008,
+      "grad_norm": 0.2539236034985644,
       "learning_rate": 2.8247839638726825e-05,
-      "loss": 0.2587,
+      "loss": 0.2586,
       "step": 760
     },
     {
       "epoch": 0.48704,
-      "grad_norm": 0.2601684998720227,
+      "grad_norm": 0.25645384647167085,
       "learning_rate": 2.8195321118341204e-05,
-      "loss": 0.2397,
+      "loss": 0.2404,
       "step": 761
     },
     {
       "epoch": 0.48768,
-      "grad_norm": 0.25920726062087374,
+      "grad_norm": 0.3015394883426219,
       "learning_rate": 2.8142788258528813e-05,
-      "loss": 0.2572,
+      "loss": 0.2571,
       "step": 762
     },
     {
       "epoch": 0.48832,
-      "grad_norm": 0.2577373926665209,
+      "grad_norm": 0.26599432217072383,
       "learning_rate": 2.8090241295037804e-05,
-      "loss": 0.253,
+      "loss": 0.2534,
       "step": 763
     },
     {
       "epoch": 0.48896,
-      "grad_norm": 0.2769909768648415,
+      "grad_norm": 0.26116413484343565,
       "learning_rate": 2.8037680463679648e-05,
-      "loss": 0.2477,
+      "loss": 0.2473,
       "step": 764
     },
     {
       "epoch": 0.4896,
-      "grad_norm": 0.25021183963273425,
+      "grad_norm": 0.25013902597670157,
       "learning_rate": 2.7985106000328033e-05,
-      "loss": 0.2399,
+      "loss": 0.2406,
       "step": 765
     },
     {
       "epoch": 0.49024,
-      "grad_norm": 0.2751049696728052,
+      "grad_norm": 0.27203497182718117,
       "learning_rate": 2.793251814091782e-05,
-      "loss": 0.2365,
+      "loss": 0.2364,
       "step": 766
     },
     {
       "epoch": 0.49088,
-      "grad_norm": 0.2785539736743758,
+      "grad_norm": 0.2809161478373533,
       "learning_rate": 2.787991712144401e-05,
-      "loss": 0.2599,
+      "loss": 0.2598,
       "step": 767
     },
     {
       "epoch": 0.49152,
-      "grad_norm": 0.25754493785695,
+      "grad_norm": 0.2594480739530536,
       "learning_rate": 2.7827303177960638e-05,
-      "loss": 0.2654,
+      "loss": 0.2655,
       "step": 768
     },
     {
       "epoch": 0.49216,
-      "grad_norm": 0.2977071861983136,
+      "grad_norm": 0.2613564843549674,
       "learning_rate": 2.7774676546579737e-05,
-      "loss": 0.2513,
+      "loss": 0.2518,
       "step": 769
     },
     {
       "epoch": 0.4928,
-      "grad_norm": 0.2627055142139323,
+      "grad_norm": 0.2628312412666247,
       "learning_rate": 2.7722037463470297e-05,
       "loss": 0.2314,
       "step": 770
     },
     {
       "epoch": 0.49344,
-      "grad_norm": 0.24348337022176492,
+      "grad_norm": 0.24433922482979678,
       "learning_rate": 2.7669386164857174e-05,
-      "loss": 0.2159,
+      "loss": 0.216,
       "step": 771
     },
     {
       "epoch": 0.49408,
-      "grad_norm": 0.2567881766174943,
+      "grad_norm": 0.2811567735489263,
       "learning_rate": 2.7616722887020034e-05,
-      "loss": 0.2309,
+      "loss": 0.2317,
       "step": 772
     },
     {
       "epoch": 0.49472,
-      "grad_norm": 0.2960178723617407,
+      "grad_norm": 0.2988300663760754,
       "learning_rate": 2.7564047866292336e-05,
-      "loss": 0.2348,
+      "loss": 0.2353,
       "step": 773
     },
     {
       "epoch": 0.49536,
-      "grad_norm": 0.2573058308269504,
+      "grad_norm": 0.2462499230606909,
       "learning_rate": 2.7511361339060193e-05,
-      "loss": 0.2578,
+      "loss": 0.2579,
       "step": 774
     },
     {
       "epoch": 0.496,
-      "grad_norm": 0.23934502760321336,
+      "grad_norm": 0.23990263065434209,
       "learning_rate": 2.7458663541761376e-05,
-      "loss": 0.2539,
+      "loss": 0.2535,
       "step": 775
     },
     {
       "epoch": 0.49664,
-      "grad_norm": 0.2559326444177946,
+      "grad_norm": 0.2604844585372496,
       "learning_rate": 2.740595471088424e-05,
-      "loss": 0.233,
+      "loss": 0.2329,
       "step": 776
     },
     {
       "epoch": 0.49728,
-      "grad_norm": 0.29032921794905275,
+      "grad_norm": 0.5977257472918388,
       "learning_rate": 2.735323508296664e-05,
       "loss": 0.2542,
       "step": 777
     },
     {
       "epoch": 0.49792,
-      "grad_norm": 0.2575784084305343,
+      "grad_norm": 0.2584304116675002,
       "learning_rate": 2.7300504894594892e-05,
-      "loss": 0.242,
+      "loss": 0.2422,
       "step": 778
     },
     {
       "epoch": 0.49856,
-      "grad_norm": 0.2448048925081302,
+      "grad_norm": 0.25230670266535726,
       "learning_rate": 2.7247764382402696e-05,
-      "loss": 0.2691,
+      "loss": 0.2697,
       "step": 779
     },
     {
       "epoch": 0.4992,
-      "grad_norm": 0.25496846358680036,
+      "grad_norm": 0.25239364609932186,
       "learning_rate": 2.7195013783070093e-05,
-      "loss": 0.2622,
+      "loss": 0.2626,
       "step": 780
     },
     {
       "epoch": 0.49984,
-      "grad_norm": 0.2645276293483241,
+      "grad_norm": 0.26187905841301573,
       "learning_rate": 2.7142253333322387e-05,
-      "loss": 0.2483,
+      "loss": 0.2486,
       "step": 781
     },
     {
       "epoch": 0.50048,
-      "grad_norm": 0.24891173303191308,
+      "grad_norm": 0.24964884815156177,
       "learning_rate": 2.7089483269929074e-05,
-      "loss": 0.2511,
+      "loss": 0.2514,
       "step": 782
     },
     {
       "epoch": 0.50112,
-      "grad_norm": 0.24013450456492896,
+      "grad_norm": 0.238711792939048,
       "learning_rate": 2.703670382970281e-05,
-      "loss": 0.2351,
+      "loss": 0.2349,
       "step": 783
     }
   ],