diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -10,7010 +10,7010 @@
   "log_history": [
     {
       "epoch": 0.00010707784559374665,
-      "grad_norm": 0.4914730489253998,
+      "grad_norm": 0.4909955859184265,
       "learning_rate": 0.00019999950652018584,
       "loss": 1.0891,
       "step": 1
     },
     {
       "epoch": 0.0002141556911874933,
-      "grad_norm": 0.3680574297904968,
+      "grad_norm": 0.37118685245513916,
       "learning_rate": 0.0001999980260856137,
-      "loss": 1.0538,
+      "loss": 1.0543,
       "step": 2
     },
     {
       "epoch": 0.00032123353678124,
-      "grad_norm": 0.39417698979377747,
+      "grad_norm": 0.3935191035270691,
       "learning_rate": 0.000199995558710895,
-      "loss": 1.0477,
+      "loss": 1.0476,
       "step": 3
     },
     {
       "epoch": 0.0004283113823749866,
-      "grad_norm": 0.4732518792152405,
+      "grad_norm": 0.47593897581100464,
       "learning_rate": 0.00019999210442038162,
-      "loss": 0.9224,
+      "loss": 0.9218,
       "step": 4
     },
     {
       "epoch": 0.0005353892279687332,
-      "grad_norm": 0.41816651821136475,
+      "grad_norm": 0.4176044166088104,
       "learning_rate": 0.00019998766324816607,
-      "loss": 0.8018,
+      "loss": 0.8016,
       "step": 5
     },
     {
       "epoch": 0.00064246707356248,
-      "grad_norm": 0.45390111207962036,
+      "grad_norm": 0.45231443643569946,
       "learning_rate": 0.0001999822352380809,
-      "loss": 0.7244,
+      "loss": 0.7248,
       "step": 6
     },
     {
       "epoch": 0.0007495449191562266,
-      "grad_norm": 0.47834527492523193,
+      "grad_norm": 0.4732208251953125,
       "learning_rate": 0.00019997582044369843,
-      "loss": 0.6486,
+      "loss": 0.6483,
       "step": 7
     },
     {
       "epoch": 0.0008566227647499732,
-      "grad_norm": 0.4380565285682678,
+      "grad_norm": 0.4255562424659729,
       "learning_rate": 0.00019996841892833,
-      "loss": 0.5942,
+      "loss": 0.5931,
       "step": 8
     },
     {
       "epoch": 0.0009637006103437199,
-      "grad_norm": 0.5452256202697754,
+      "grad_norm": 0.5560129284858704,
       "learning_rate": 0.00019996003076502565,
-      "loss": 0.5054,
+      "loss": 0.5045,
       "step": 9
     },
     {
       "epoch": 0.0010707784559374665,
-      "grad_norm": 0.5409011244773865,
+      "grad_norm": 0.5253703594207764,
       "learning_rate": 0.00019995065603657316,
-      "loss": 0.3999,
+      "loss": 0.3965,
       "step": 10
     },
     {
       "epoch": 0.001177856301531213,
-      "grad_norm": 0.44740885496139526,
+      "grad_norm": 0.4471832513809204,
       "learning_rate": 0.0001999402948354973,
-      "loss": 0.5802,
+      "loss": 0.5783,
       "step": 11
     },
     {
       "epoch": 0.00128493414712496,
-      "grad_norm": 0.44699716567993164,
+      "grad_norm": 0.4470554292201996,
       "learning_rate": 0.00019992894726405893,
-      "loss": 0.4222,
+      "loss": 0.4207,
       "step": 12
     },
     {
       "epoch": 0.0013920119927187066,
-      "grad_norm": 0.4242172837257385,
+      "grad_norm": 0.42513272166252136,
       "learning_rate": 0.000199916613434254,
-      "loss": 0.4839,
+      "loss": 0.4833,
       "step": 13
     },
     {
       "epoch": 0.0014990898383124532,
-      "grad_norm": 0.3534027934074402,
+      "grad_norm": 0.34535884857177734,
       "learning_rate": 0.0001999032934678125,
-      "loss": 0.3729,
+      "loss": 0.3717,
       "step": 14
     },
     {
       "epoch": 0.0016061676839061998,
-      "grad_norm": 0.3338993191719055,
+      "grad_norm": 0.33447375893592834,
       "learning_rate": 0.00019988898749619702,
-      "loss": 0.3565,
+      "loss": 0.3569,
       "step": 15
     },
     {
       "epoch": 0.0017132455294999465,
-      "grad_norm": 0.5204335451126099,
+      "grad_norm": 0.50462406873703,
       "learning_rate": 0.00019987369566060176,
-      "loss": 0.3462,
+      "loss": 0.3444,
       "step": 16
     },
     {
       "epoch": 0.001820323375093693,
-      "grad_norm": 0.3927958905696869,
+      "grad_norm": 0.3608594834804535,
       "learning_rate": 0.00019985741811195097,
-      "loss": 0.455,
+      "loss": 0.4535,
       "step": 17
     },
     {
       "epoch": 0.0019274012206874397,
-      "grad_norm": 0.28747662901878357,
+      "grad_norm": 0.289742648601532,
       "learning_rate": 0.00019984015501089752,
-      "loss": 0.3915,
+      "loss": 0.3921,
       "step": 18
     },
     {
       "epoch": 0.0020344790662811863,
-      "grad_norm": 0.34313827753067017,
+      "grad_norm": 0.3383671045303345,
       "learning_rate": 0.0001998219065278212,
-      "loss": 0.4298,
+      "loss": 0.4304,
       "step": 19
     },
     {
       "epoch": 0.002141556911874933,
-      "grad_norm": 0.3621661067008972,
+      "grad_norm": 0.3464438319206238,
       "learning_rate": 0.00019980267284282717,
-      "loss": 0.3907,
+      "loss": 0.3887,
       "step": 20
     },
     {
       "epoch": 0.0022486347574686796,
-      "grad_norm": 0.3450513780117035,
+      "grad_norm": 0.3533989191055298,
       "learning_rate": 0.00019978245414574417,
-      "loss": 0.3854,
+      "loss": 0.3861,
       "step": 21
     },
     {
       "epoch": 0.002355712603062426,
-      "grad_norm": 0.3595326244831085,
+      "grad_norm": 0.3390829563140869,
       "learning_rate": 0.00019976125063612252,
-      "loss": 0.3459,
+      "loss": 0.3445,
       "step": 22
     },
     {
       "epoch": 0.002462790448656173,
-      "grad_norm": 0.7127284407615662,
+      "grad_norm": 0.23587378859519958,
       "learning_rate": 0.00019973906252323238,
-      "loss": 0.2965,
+      "loss": 0.2955,
       "step": 23
     },
     {
       "epoch": 0.00256986829424992,
-      "grad_norm": 0.40753620862960815,
+      "grad_norm": 0.42040207982063293,
       "learning_rate": 0.0001997158900260614,
-      "loss": 0.4976,
+      "loss": 0.4963,
       "step": 24
     },
     {
       "epoch": 0.0026769461398436665,
-      "grad_norm": 0.25613075494766235,
+      "grad_norm": 0.25572434067726135,
       "learning_rate": 0.0001996917333733128,
-      "loss": 0.3591,
+      "loss": 0.359,
       "step": 25
     },
     {
       "epoch": 0.002784023985437413,
-      "grad_norm": 0.42440053820610046,
+      "grad_norm": 0.44296184182167053,
       "learning_rate": 0.00019966659280340297,
-      "loss": 0.3673,
+      "loss": 0.366,
       "step": 26
     },
     {
       "epoch": 0.0028911018310311598,
-      "grad_norm": 0.2876407206058502,
+      "grad_norm": 0.2594900131225586,
       "learning_rate": 0.00019964046856445924,
-      "loss": 0.2941,
+      "loss": 0.2909,
       "step": 27
     },
     {
       "epoch": 0.0029981796766249064,
-      "grad_norm": 0.2591134011745453,
+      "grad_norm": 0.2581496238708496,
       "learning_rate": 0.00019961336091431727,
-      "loss": 0.297,
+      "loss": 0.2962,
       "step": 28
     },
     {
       "epoch": 0.003105257522218653,
-      "grad_norm": 0.25432032346725464,
+      "grad_norm": 0.29349207878112793,
       "learning_rate": 0.00019958527012051857,
-      "loss": 0.2986,
+      "loss": 0.2998,
       "step": 29
     },
     {
       "epoch": 0.0032123353678123997,
-      "grad_norm": 0.29807740449905396,
+      "grad_norm": 0.30551677942276,
       "learning_rate": 0.00019955619646030802,
-      "loss": 0.4165,
+      "loss": 0.4177,
       "step": 30
     },
     {
       "epoch": 0.0033194132134061463,
-      "grad_norm": 0.22817127406597137,
+      "grad_norm": 0.22254447638988495,
       "learning_rate": 0.00019952614022063084,
-      "loss": 0.3591,
+      "loss": 0.3558,
       "step": 31
     },
     {
       "epoch": 0.003426491058999893,
-      "grad_norm": 0.2147909253835678,
+      "grad_norm": 0.2150162309408188,
       "learning_rate": 0.00019949510169813003,
-      "loss": 0.3804,
+      "loss": 0.376,
       "step": 32
     },
     {
       "epoch": 0.0035335689045936395,
-      "grad_norm": 0.26951637864112854,
+      "grad_norm": 0.2759721577167511,
       "learning_rate": 0.00019946308119914323,
-      "loss": 0.3321,
+      "loss": 0.3295,
       "step": 33
     },
     {
       "epoch": 0.003640646750187386,
-      "grad_norm": 0.2519189119338989,
+      "grad_norm": 0.2623266875743866,
       "learning_rate": 0.0001994300790396999,
-      "loss": 0.3917,
+      "loss": 0.3923,
       "step": 34
     },
     {
       "epoch": 0.0037477245957811328,
-      "grad_norm": 0.2383698970079422,
+      "grad_norm": 0.24647563695907593,
       "learning_rate": 0.000199396095545518,
-      "loss": 0.3692,
+      "loss": 0.3683,
       "step": 35
     },
     {
       "epoch": 0.0038548024413748794,
-      "grad_norm": 0.1950429528951645,
+      "grad_norm": 0.20459140837192535,
       "learning_rate": 0.00019936113105200085,
-      "loss": 0.2132,
+      "loss": 0.2112,
       "step": 36
     },
     {
       "epoch": 0.003961880286968626,
-      "grad_norm": 0.24658283591270447,
+      "grad_norm": 0.23986490070819855,
       "learning_rate": 0.00019932518590423394,
-      "loss": 0.3589,
+      "loss": 0.3532,
       "step": 37
     },
     {
       "epoch": 0.004068958132562373,
-      "grad_norm": 0.21331077814102173,
+      "grad_norm": 0.21777169406414032,
       "learning_rate": 0.00019928826045698136,
-      "loss": 0.3155,
+      "loss": 0.3148,
       "step": 38
     },
     {
       "epoch": 0.004176035978156119,
-      "grad_norm": 0.25296568870544434,
+      "grad_norm": 0.2896265685558319,
       "learning_rate": 0.0001992503550746824,
-      "loss": 0.3137,
+      "loss": 0.3149,
       "step": 39
     },
     {
       "epoch": 0.004283113823749866,
-      "grad_norm": 0.23738770186901093,
+      "grad_norm": 0.24791480600833893,
       "learning_rate": 0.0001992114701314478,
-      "loss": 0.3445,
+      "loss": 0.3459,
       "step": 40
     },
     {
       "epoch": 0.0043901916693436125,
-      "grad_norm": 0.23128816485404968,
+      "grad_norm": 0.25446805357933044,
       "learning_rate": 0.0001991716060110563,
-      "loss": 0.3027,
+      "loss": 0.3023,
       "step": 41
     },
     {
       "epoch": 0.004497269514937359,
-      "grad_norm": 0.21762651205062866,
+      "grad_norm": 0.22087767720222473,
       "learning_rate": 0.00019913076310695068,
-      "loss": 0.3852,
+      "loss": 0.3853,
       "step": 42
     },
     {
       "epoch": 0.004604347360531106,
-      "grad_norm": 0.2557142972946167,
+      "grad_norm": 0.2717120945453644,
       "learning_rate": 0.00019908894182223388,
-      "loss": 0.3346,
+      "loss": 0.3354,
       "step": 43
     },
     {
       "epoch": 0.004711425206124852,
-      "grad_norm": 0.21737657487392426,
+      "grad_norm": 0.20693597197532654,
       "learning_rate": 0.00019904614256966512,
-      "loss": 0.3461,
+      "loss": 0.3419,
       "step": 44
     },
     {
       "epoch": 0.004818503051718599,
-      "grad_norm": 0.21647535264492035,
+      "grad_norm": 0.21952880918979645,
       "learning_rate": 0.00019900236577165576,
-      "loss": 0.2972,
+      "loss": 0.2945,
       "step": 45
     },
     {
       "epoch": 0.004925580897312346,
-      "grad_norm": 0.24528199434280396,
+      "grad_norm": 0.2796410620212555,
       "learning_rate": 0.0001989576118602651,
-      "loss": 0.349,
+      "loss": 0.3456,
       "step": 46
     },
     {
       "epoch": 0.005032658742906092,
-      "grad_norm": 0.22419703006744385,
+      "grad_norm": 0.21983903646469116,
       "learning_rate": 0.00019891188127719618,
-      "loss": 0.3198,
+      "loss": 0.3203,
       "step": 47
     },
     {
       "epoch": 0.00513973658849984,
-      "grad_norm": 0.18803644180297852,
+      "grad_norm": 0.1815391182899475,
       "learning_rate": 0.0001988651744737914,
-      "loss": 0.2826,
+      "loss": 0.2796,
       "step": 48
     },
     {
       "epoch": 0.005246814434093586,
-      "grad_norm": 0.23534414172172546,
+      "grad_norm": 0.2382379174232483,
       "learning_rate": 0.00019881749191102808,
-      "loss": 0.3813,
+      "loss": 0.3825,
       "step": 49
     },
     {
       "epoch": 0.005353892279687333,
-      "grad_norm": 0.1908506602048874,
+      "grad_norm": 0.19003333151340485,
       "learning_rate": 0.00019876883405951377,
-      "loss": 0.3349,
+      "loss": 0.3344,
       "step": 50
     },
     {
       "epoch": 0.00546097012528108,
-      "grad_norm": 0.22902971506118774,
+      "grad_norm": 0.2239263653755188,
       "learning_rate": 0.00019871920139948192,
       "loss": 0.4143,
       "step": 51
     },
     {
       "epoch": 0.005568047970874826,
-      "grad_norm": 0.19370073080062866,
+      "grad_norm": 0.1837572157382965,
       "learning_rate": 0.0001986685944207868,
-      "loss": 0.3182,
+      "loss": 0.3166,
       "step": 52
     },
     {
       "epoch": 0.005675125816468573,
-      "grad_norm": 0.22496962547302246,
+      "grad_norm": 0.22199811041355133,
       "learning_rate": 0.0001986170136228989,
-      "loss": 0.3622,
+      "loss": 0.3639,
       "step": 53
     },
     {
       "epoch": 0.0057822036620623195,
-      "grad_norm": 0.19907711446285248,
+      "grad_norm": 0.19471023976802826,
       "learning_rate": 0.00019856445951489982,
-      "loss": 0.3328,
+      "loss": 0.3314,
       "step": 54
     },
     {
       "epoch": 0.005889281507656066,
-      "grad_norm": 0.28724735975265503,
+      "grad_norm": 0.26663461327552795,
       "learning_rate": 0.0001985109326154774,
-      "loss": 0.4235,
+      "loss": 0.4204,
       "step": 55
     },
     {
       "epoch": 0.005996359353249813,
-      "grad_norm": 0.2073618769645691,
+      "grad_norm": 0.20248642563819885,
       "learning_rate": 0.00019845643345292054,
-      "loss": 0.3651,
+      "loss": 0.3668,
       "step": 56
     },
     {
       "epoch": 0.006103437198843559,
-      "grad_norm": 0.3525002896785736,
+      "grad_norm": 0.34333622455596924,
       "learning_rate": 0.00019840096256511398,
-      "loss": 0.4431,
+      "loss": 0.4426,
       "step": 57
     },
     {
       "epoch": 0.006210515044437306,
-      "grad_norm": 0.2288159877061844,
+      "grad_norm": 0.226885586977005,
       "learning_rate": 0.00019834452049953297,
-      "loss": 0.3314,
+      "loss": 0.3298,
       "step": 58
     },
     {
       "epoch": 0.006317592890031053,
-      "grad_norm": 0.27389129996299744,
+      "grad_norm": 0.2704862952232361,
       "learning_rate": 0.00019828710781323792,
-      "loss": 0.4399,
+      "loss": 0.4359,
       "step": 59
     },
     {
       "epoch": 0.006424670735624799,
-      "grad_norm": 0.37030190229415894,
+      "grad_norm": 0.24985526502132416,
       "learning_rate": 0.0001982287250728689,
-      "loss": 0.4215,
+      "loss": 0.4191,
       "step": 60
     },
     {
       "epoch": 0.006531748581218546,
-      "grad_norm": 0.19040270149707794,
+      "grad_norm": 0.1854284405708313,
       "learning_rate": 0.0001981693728546399,
-      "loss": 0.3156,
+      "loss": 0.3131,
       "step": 61
     },
     {
       "epoch": 0.0066388264268122926,
-      "grad_norm": 0.23044194281101227,
+      "grad_norm": 0.21722164750099182,
       "learning_rate": 0.0001981090517443334,
-      "loss": 0.3071,
+      "loss": 0.3055,
       "step": 62
     },
     {
       "epoch": 0.006745904272406039,
-      "grad_norm": 0.2296312153339386,
+      "grad_norm": 0.231235533952713,
       "learning_rate": 0.00019804776233729444,
-      "loss": 0.3353,
+      "loss": 0.3345,
       "step": 63
     },
     {
       "epoch": 0.006852982117999786,
-      "grad_norm": 0.2212522029876709,
+      "grad_norm": 0.22202789783477783,
       "learning_rate": 0.0001979855052384247,
-      "loss": 0.3534,
+      "loss": 0.3525,
       "step": 64
     },
     {
       "epoch": 0.0069600599635935324,
-      "grad_norm": 0.23206260800361633,
+      "grad_norm": 0.21439769864082336,
       "learning_rate": 0.00019792228106217658,
-      "loss": 0.3441,
+      "loss": 0.3413,
       "step": 65
     },
     {
       "epoch": 0.007067137809187279,
-      "grad_norm": 0.19569985568523407,
+      "grad_norm": 0.18203051388263702,
       "learning_rate": 0.00019785809043254722,
-      "loss": 0.335,
+      "loss": 0.3312,
       "step": 66
     },
     {
       "epoch": 0.007174215654781026,
-      "grad_norm": 0.23851607739925385,
+      "grad_norm": 0.22350089251995087,
       "learning_rate": 0.0001977929339830722,
-      "loss": 0.487,
+      "loss": 0.4871,
       "step": 67
     },
     {
       "epoch": 0.007281293500374772,
-      "grad_norm": 0.19546416401863098,
+      "grad_norm": 0.19005514681339264,
       "learning_rate": 0.00019772681235681936,
-      "loss": 0.44,
+      "loss": 0.4379,
       "step": 68
     },
     {
       "epoch": 0.007388371345968519,
-      "grad_norm": 0.20493975281715393,
+      "grad_norm": 0.18949486315250397,
       "learning_rate": 0.00019765972620638248,
-      "loss": 0.3202,
+      "loss": 0.3185,
       "step": 69
     },
     {
       "epoch": 0.0074954491915622656,
-      "grad_norm": 0.1694723516702652,
+      "grad_norm": 0.1703757345676422,
       "learning_rate": 0.00019759167619387476,
-      "loss": 0.3613,
+      "loss": 0.3595,
       "step": 70
     },
     {
       "epoch": 0.007602527037156012,
-      "grad_norm": 0.29382210969924927,
+      "grad_norm": 0.27864259481430054,
       "learning_rate": 0.00019752266299092236,
-      "loss": 0.3509,
+      "loss": 0.3437,
       "step": 71
     },
     {
       "epoch": 0.007709604882749759,
-      "grad_norm": 0.17875567078590393,
+      "grad_norm": 0.17867398262023926,
       "learning_rate": 0.00019745268727865774,
-      "loss": 0.332,
+      "loss": 0.3299,
       "step": 72
     },
     {
       "epoch": 0.007816682728343505,
-      "grad_norm": 0.1698933243751526,
+      "grad_norm": 0.1801465004682541,
       "learning_rate": 0.0001973817497477129,
-      "loss": 0.2699,
+      "loss": 0.2722,
       "step": 73
     },
     {
       "epoch": 0.007923760573937252,
-      "grad_norm": 0.17154669761657715,
+      "grad_norm": 0.16662195324897766,
       "learning_rate": 0.00019730985109821266,
-      "loss": 0.3798,
+      "loss": 0.3794,
       "step": 74
     },
     {
       "epoch": 0.008030838419530999,
-      "grad_norm": 0.19141319394111633,
+      "grad_norm": 0.1943478137254715,
       "learning_rate": 0.00019723699203976766,
-      "loss": 0.3966,
+      "loss": 0.3975,
       "step": 75
     },
     {
       "epoch": 0.008137916265124745,
-      "grad_norm": 0.1600220948457718,
+      "grad_norm": 0.1636917144060135,
       "learning_rate": 0.0001971631732914674,
-      "loss": 0.2993,
+      "loss": 0.3018,
       "step": 76
     },
     {
       "epoch": 0.008244994110718492,
-      "grad_norm": 0.1882285475730896,
+      "grad_norm": 0.1836482137441635,
       "learning_rate": 0.0001970883955818731,
-      "loss": 0.2683,
+      "loss": 0.2692,
       "step": 77
     },
     {
       "epoch": 0.008352071956312239,
-      "grad_norm": 0.17301051318645477,
+      "grad_norm": 0.17146596312522888,
       "learning_rate": 0.0001970126596490106,
-      "loss": 0.2705,
+      "loss": 0.271,
       "step": 78
     },
     {
       "epoch": 0.008459149801905985,
-      "grad_norm": 0.21556293964385986,
+      "grad_norm": 0.215361550450325,
       "learning_rate": 0.00019693596624036292,
-      "loss": 0.4031,
+      "loss": 0.4033,
       "step": 79
     },
     {
       "epoch": 0.008566227647499732,
-      "grad_norm": 0.1741161048412323,
+      "grad_norm": 0.16516317427158356,
       "learning_rate": 0.0001968583161128631,
-      "loss": 0.3607,
+      "loss": 0.3614,
       "step": 80
     },
     {
       "epoch": 0.008673305493093478,
-      "grad_norm": 0.25049349665641785,
+      "grad_norm": 0.24210558831691742,
       "learning_rate": 0.00019677971003288655,
-      "loss": 0.5397,
+      "loss": 0.5388,
       "step": 81
     },
     {
       "epoch": 0.008780383338687225,
-      "grad_norm": 0.16967053711414337,
+      "grad_norm": 0.16257977485656738,
       "learning_rate": 0.00019670014877624353,
-      "loss": 0.3115,
+      "loss": 0.3109,
       "step": 82
     },
     {
       "epoch": 0.008887461184280972,
-      "grad_norm": 0.26333314180374146,
+      "grad_norm": 0.26173606514930725,
       "learning_rate": 0.00019661963312817148,
-      "loss": 0.3876,
+      "loss": 0.389,
       "step": 83
     },
     {
       "epoch": 0.008994539029874718,
-      "grad_norm": 0.19522804021835327,
+      "grad_norm": 0.19191040098667145,
       "learning_rate": 0.0001965381638833274,
-      "loss": 0.3693,
+      "loss": 0.3695,
       "step": 84
     },
     {
       "epoch": 0.009101616875468465,
-      "grad_norm": 0.17835591733455658,
+      "grad_norm": 0.1762268841266632,
       "learning_rate": 0.00019645574184577982,
-      "loss": 0.3377,
+      "loss": 0.3374,
       "step": 85
     },
     {
       "epoch": 0.009208694721062212,
-      "grad_norm": 0.19591933488845825,
+      "grad_norm": 0.18683938682079315,
       "learning_rate": 0.000196372367829001,
-      "loss": 0.3446,
+      "loss": 0.3423,
       "step": 86
     },
     {
       "epoch": 0.009315772566655958,
-      "grad_norm": 0.17983509600162506,
+      "grad_norm": 0.17973066866397858,
       "learning_rate": 0.00019628804265585877,
-      "loss": 0.3993,
+      "loss": 0.398,
       "step": 87
     },
     {
       "epoch": 0.009422850412249705,
-      "grad_norm": 0.1638455092906952,
+      "grad_norm": 0.16043399274349213,
       "learning_rate": 0.0001962027671586086,
-      "loss": 0.2946,
+      "loss": 0.2939,
       "step": 88
     },
     {
       "epoch": 0.009529928257843451,
-      "grad_norm": 0.17286810278892517,
+      "grad_norm": 0.17222130298614502,
       "learning_rate": 0.0001961165421788852,
-      "loss": 0.3427,
+      "loss": 0.3419,
       "step": 89
     },
     {
       "epoch": 0.009637006103437198,
-      "grad_norm": 0.18723516166210175,
+      "grad_norm": 0.1862293928861618,
       "learning_rate": 0.0001960293685676943,
-      "loss": 0.2939,
+      "loss": 0.2944,
       "step": 90
     },
     {
       "epoch": 0.009744083949030945,
-      "grad_norm": 0.2359951138496399,
+      "grad_norm": 0.22960934042930603,
       "learning_rate": 0.0001959412471854043,
-      "loss": 0.4401,
+      "loss": 0.4391,
       "step": 91
     },
     {
       "epoch": 0.009851161794624691,
-      "grad_norm": 0.1666853427886963,
+      "grad_norm": 0.16783665120601654,
       "learning_rate": 0.0001958521789017376,
-      "loss": 0.3019,
+      "loss": 0.3055,
       "step": 92
     },
     {
       "epoch": 0.009958239640218438,
-      "grad_norm": 0.20216096937656403,
+      "grad_norm": 0.1998833566904068,
       "learning_rate": 0.00019576216459576222,
-      "loss": 0.3507,
+      "loss": 0.351,
       "step": 93
     },
     {
       "epoch": 0.010065317485812185,
-      "grad_norm": 0.1928347647190094,
+      "grad_norm": 0.19366595149040222,
       "learning_rate": 0.00019567120515588308,
-      "loss": 0.4894,
+      "loss": 0.4888,
       "step": 94
     },
     {
       "epoch": 0.010172395331405933,
-      "grad_norm": 0.2161427140235901,
+      "grad_norm": 0.20969517529010773,
       "learning_rate": 0.00019557930147983302,
-      "loss": 0.4311,
+      "loss": 0.429,
       "step": 95
     },
     {
       "epoch": 0.01027947317699968,
-      "grad_norm": 0.1798643171787262,
+      "grad_norm": 0.17586903274059296,
       "learning_rate": 0.00019548645447466431,
-      "loss": 0.2975,
+      "loss": 0.2981,
       "step": 96
     },
     {
       "epoch": 0.010386551022593426,
-      "grad_norm": 0.21056245267391205,
+      "grad_norm": 0.2091716229915619,
       "learning_rate": 0.00019539266505673938,
-      "loss": 0.3106,
+      "loss": 0.3118,
       "step": 97
     },
     {
       "epoch": 0.010493628868187173,
-      "grad_norm": 0.22675903141498566,
+      "grad_norm": 0.2315663993358612,
       "learning_rate": 0.00019529793415172192,
-      "loss": 0.4743,
+      "loss": 0.4724,
       "step": 98
     },
     {
       "epoch": 0.01060070671378092,
-      "grad_norm": 0.19647933542728424,
+      "grad_norm": 0.19753670692443848,
       "learning_rate": 0.00019520226269456768,
-      "loss": 0.319,
+      "loss": 0.3185,
       "step": 99
     },
     {
       "epoch": 0.010707784559374666,
-      "grad_norm": 0.15131019055843353,
+      "grad_norm": 0.1516932249069214,
       "learning_rate": 0.00019510565162951537,
-      "loss": 0.2944,
+      "loss": 0.2947,
       "step": 100
     },
     {
       "epoch": 0.010814862404968413,
-      "grad_norm": 0.18563662469387054,
+      "grad_norm": 0.1840781718492508,
       "learning_rate": 0.00019500810191007718,
-      "loss": 0.3675,
+      "loss": 0.3685,
       "step": 101
     },
     {
       "epoch": 0.01092194025056216,
-      "grad_norm": 0.203588604927063,
+      "grad_norm": 0.1962490826845169,
       "learning_rate": 0.00019490961449902946,
-      "loss": 0.3694,
+      "loss": 0.3657,
       "step": 102
     },
     {
       "epoch": 0.011029018096155906,
-      "grad_norm": 0.22488757967948914,
+      "grad_norm": 0.22264431416988373,
       "learning_rate": 0.0001948101903684032,
-      "loss": 0.4804,
+      "loss": 0.4791,
       "step": 103
     },
     {
       "epoch": 0.011136095941749653,
-      "grad_norm": 0.17686226963996887,
+      "grad_norm": 0.17540693283081055,
       "learning_rate": 0.00019470983049947444,
-      "loss": 0.3596,
+      "loss": 0.3603,
       "step": 104
     },
     {
       "epoch": 0.0112431737873434,
-      "grad_norm": 0.15883031487464905,
+      "grad_norm": 0.1619989275932312,
       "learning_rate": 0.00019460853588275454,
-      "loss": 0.3952,
+      "loss": 0.3965,
       "step": 105
     },
     {
       "epoch": 0.011350251632937146,
-      "grad_norm": 0.1682022213935852,
+      "grad_norm": 0.17258454859256744,
       "learning_rate": 0.00019450630751798048,
-      "loss": 0.2541,
+      "loss": 0.2544,
       "step": 106
     },
     {
       "epoch": 0.011457329478530892,
-      "grad_norm": 0.22802792489528656,
+      "grad_norm": 0.23875324428081512,
       "learning_rate": 0.000194403146414105,
-      "loss": 0.3886,
+      "loss": 0.3889,
       "step": 107
     },
     {
       "epoch": 0.011564407324124639,
-      "grad_norm": 0.17755688726902008,
+      "grad_norm": 0.17445383965969086,
       "learning_rate": 0.00019429905358928646,
-      "loss": 0.3412,
+      "loss": 0.3413,
       "step": 108
     },
     {
       "epoch": 0.011671485169718386,
-      "grad_norm": 0.20976732671260834,
+      "grad_norm": 0.21563485264778137,
       "learning_rate": 0.00019419403007087907,
-      "loss": 0.3449,
+      "loss": 0.3474,
       "step": 109
     },
     {
       "epoch": 0.011778563015312132,
-      "grad_norm": 0.1593596190214157,
+      "grad_norm": 0.15843790769577026,
       "learning_rate": 0.00019408807689542257,
-      "loss": 0.3573,
+      "loss": 0.3574,
       "step": 110
     },
     {
       "epoch": 0.011885640860905879,
-      "grad_norm": 0.20216397941112518,
+      "grad_norm": 0.2041575163602829,
       "learning_rate": 0.00019398119510863197,
       "loss": 0.4127,
       "step": 111
     },
     {
       "epoch": 0.011992718706499626,
-      "grad_norm": 0.1636137217283249,
+      "grad_norm": 0.16513045132160187,
       "learning_rate": 0.00019387338576538744,
-      "loss": 0.2631,
+      "loss": 0.2633,
       "step": 112
     },
     {
       "epoch": 0.012099796552093372,
-      "grad_norm": 0.18087638914585114,
+      "grad_norm": 0.18099965155124664,
       "learning_rate": 0.00019376464992972356,
-      "loss": 0.3848,
+      "loss": 0.3851,
       "step": 113
     },
     {
       "epoch": 0.012206874397687119,
-      "grad_norm": 0.22892820835113525,
+      "grad_norm": 0.22513428330421448,
       "learning_rate": 0.00019365498867481923,
-      "loss": 0.4325,
+      "loss": 0.4314,
       "step": 114
     },
     {
       "epoch": 0.012313952243280865,
-      "grad_norm": 0.16127265989780426,
+      "grad_norm": 0.15893103182315826,
       "learning_rate": 0.00019354440308298675,
-      "loss": 0.2838,
+      "loss": 0.2835,
       "step": 115
     },
     {
       "epoch": 0.012421030088874612,
-      "grad_norm": 0.17154908180236816,
+      "grad_norm": 0.17109234631061554,
       "learning_rate": 0.00019343289424566122,
-      "loss": 0.4173,
+      "loss": 0.4174,
       "step": 116
     },
     {
       "epoch": 0.012528107934468359,
-      "grad_norm": 0.13716156780719757,
+      "grad_norm": 0.13781225681304932,
       "learning_rate": 0.00019332046326338986,
-      "loss": 0.2642,
+      "loss": 0.2647,
       "step": 117
     },
     {
       "epoch": 0.012635185780062105,
-      "grad_norm": 0.16628596186637878,
+      "grad_norm": 0.16822676360607147,
       "learning_rate": 0.0001932071112458211,
-      "loss": 0.3342,
+      "loss": 0.3348,
       "step": 118
     },
     {
       "epoch": 0.012742263625655852,
-      "grad_norm": 0.17461515963077545,
+      "grad_norm": 0.17825409770011902,
       "learning_rate": 0.00019309283931169356,
-      "loss": 0.3829,
+      "loss": 0.3831,
       "step": 119
     },
     {
       "epoch": 0.012849341471249599,
-      "grad_norm": 0.20186425745487213,
+      "grad_norm": 0.2027072161436081,
       "learning_rate": 0.00019297764858882514,
-      "loss": 0.4213,
+      "loss": 0.4212,
       "step": 120
     },
     {
       "epoch": 0.012956419316843345,
-      "grad_norm": 0.18099796772003174,
+      "grad_norm": 0.17872022092342377,
       "learning_rate": 0.00019286154021410173,
-      "loss": 0.3205,
+      "loss": 0.32,
       "step": 121
     },
     {
       "epoch": 0.013063497162437092,
-      "grad_norm": 0.19527661800384521,
+      "grad_norm": 0.19515572488307953,
       "learning_rate": 0.00019274451533346615,
       "loss": 0.4162,
       "step": 122
     },
     {
       "epoch": 0.013170575008030838,
-      "grad_norm": 0.16075778007507324,
+      "grad_norm": 0.16127873957157135,
       "learning_rate": 0.00019262657510190666,
-      "loss": 0.2801,
+      "loss": 0.2789,
       "step": 123
     },
     {
       "epoch": 0.013277652853624585,
-      "grad_norm": 0.16372549533843994,
+      "grad_norm": 0.16778552532196045,
       "learning_rate": 0.0001925077206834458,
-      "loss": 0.3641,
+      "loss": 0.3651,
       "step": 124
     },
     {
       "epoch": 0.013384730699218332,
-      "grad_norm": 0.18662606179714203,
+      "grad_norm": 0.1893121302127838,
       "learning_rate": 0.0001923879532511287,
-      "loss": 0.3799,
+      "loss": 0.381,
       "step": 125
     },
     {
       "epoch": 0.013491808544812078,
-      "grad_norm": 0.1677158623933792,
+      "grad_norm": 0.17123256623744965,
       "learning_rate": 0.0001922672739870115,
-      "loss": 0.2898,
+      "loss": 0.2907,
       "step": 126
     },
     {
       "epoch": 0.013598886390405825,
-      "grad_norm": 0.19273678958415985,
+      "grad_norm": 0.19237865507602692,
       "learning_rate": 0.00019214568408214985,
-      "loss": 0.3585,
+      "loss": 0.3571,
       "step": 127
     },
     {
       "epoch": 0.013705964235999572,
-      "grad_norm": 0.20474790036678314,
+      "grad_norm": 0.20815183222293854,
       "learning_rate": 0.00019202318473658705,
-      "loss": 0.4194,
+      "loss": 0.4195,
       "step": 128
     },
     {
       "epoch": 0.013813042081593318,
-      "grad_norm": 0.19291414320468903,
+      "grad_norm": 0.19400858879089355,
       "learning_rate": 0.00019189977715934213,
-      "loss": 0.4429,
+      "loss": 0.443,
       "step": 129
     },
     {
       "epoch": 0.013920119927187065,
-      "grad_norm": 0.19790148735046387,
+      "grad_norm": 0.20479348301887512,
       "learning_rate": 0.00019177546256839812,
-      "loss": 0.3079,
+      "loss": 0.3082,
       "step": 130
     },
     {
       "epoch": 0.014027197772780811,
-      "grad_norm": 0.1970977485179901,
+      "grad_norm": 0.20510035753250122,
       "learning_rate": 0.0001916502421906898,
       "loss": 0.3948,
       "step": 131
     },
     {
       "epoch": 0.014134275618374558,
-      "grad_norm": 0.18109802901744843,
+      "grad_norm": 0.18292556703090668,
       "learning_rate": 0.00019152411726209176,
-      "loss": 0.2952,
+      "loss": 0.2958,
       "step": 132
     },
     {
       "epoch": 0.014241353463968305,
-      "grad_norm": 0.16982200741767883,
+      "grad_norm": 0.17402611672878265,
       "learning_rate": 0.00019139708902740613,
-      "loss": 0.3422,
+      "loss": 0.3426,
       "step": 133
     },
     {
       "epoch": 0.014348431309562051,
-      "grad_norm": 0.17093245685100555,
+      "grad_norm": 0.1712314933538437,
       "learning_rate": 0.0001912691587403503,
-      "loss": 0.3613,
+      "loss": 0.3618,
       "step": 134
     },
     {
       "epoch": 0.014455509155155798,
-      "grad_norm": 0.15467596054077148,
+      "grad_norm": 0.15368402004241943,
       "learning_rate": 0.00019114032766354453,
-      "loss": 0.2985,
+      "loss": 0.2978,
       "step": 135
     },
     {
       "epoch": 0.014562587000749545,
-      "grad_norm": 0.20507284998893738,
+      "grad_norm": 0.2001761794090271,
       "learning_rate": 0.00019101059706849957,
-      "loss": 0.3405,
+      "loss": 0.3388,
       "step": 136
     },
     {
       "epoch": 0.014669664846343291,
-      "grad_norm": 0.17259500920772552,
+      "grad_norm": 0.17164276540279388,
       "learning_rate": 0.00019087996823560402,
-      "loss": 0.4002,
+      "loss": 0.3974,
       "step": 137
     },
     {
       "epoch": 0.014776742691937038,
-      "grad_norm": 0.1541275680065155,
+      "grad_norm": 0.15010999143123627,
       "learning_rate": 0.0001907484424541117,
-      "loss": 0.2888,
+      "loss": 0.2875,
       "step": 138
     },
     {
       "epoch": 0.014883820537530785,
-      "grad_norm": 0.19659638404846191,
+      "grad_norm": 0.19477848708629608,
       "learning_rate": 0.00019061602102212898,
-      "loss": 0.4393,
+      "loss": 0.4388,
       "step": 139
     },
     {
       "epoch": 0.014990898383124531,
-      "grad_norm": 0.17995648086071014,
+      "grad_norm": 0.18081682920455933,
       "learning_rate": 0.00019048270524660196,
-      "loss": 0.333,
+      "loss": 0.3336,
       "step": 140
     },
     {
       "epoch": 0.015097976228718278,
-      "grad_norm": 0.22955287992954254,
+      "grad_norm": 0.22473369538784027,
       "learning_rate": 0.0001903484964433035,
-      "loss": 0.2958,
+      "loss": 0.294,
       "step": 141
     },
     {
       "epoch": 0.015205054074312024,
-      "grad_norm": 0.164543017745018,
+      "grad_norm": 0.16706225275993347,
       "learning_rate": 0.00019021339593682028,
-      "loss": 0.3002,
+      "loss": 0.302,
       "step": 142
     },
     {
       "epoch": 0.015312131919905771,
-      "grad_norm": 0.15457768738269806,
+      "grad_norm": 0.15441536903381348,
       "learning_rate": 0.00019007740506053983,
-      "loss": 0.3089,
+      "loss": 0.309,
       "step": 143
     },
     {
       "epoch": 0.015419209765499518,
-      "grad_norm": 0.17177622020244598,
+      "grad_norm": 0.1709998995065689,
       "learning_rate": 0.0001899405251566371,
-      "loss": 0.3263,
+      "loss": 0.3257,
       "step": 144
     },
     {
       "epoch": 0.015526287611093264,
-      "grad_norm": 0.1843390315771103,
+      "grad_norm": 0.18251051008701324,
       "learning_rate": 0.00018980275757606157,
-      "loss": 0.3417,
+      "loss": 0.3423,
       "step": 145
     },
     {
       "epoch": 0.01563336545668701,
-      "grad_norm": 0.15636736154556274,
+      "grad_norm": 0.155283123254776,
       "learning_rate": 0.00018966410367852362,
-      "loss": 0.2202,
+      "loss": 0.2204,
       "step": 146
     },
     {
       "epoch": 0.01574044330228076,
-      "grad_norm": 0.1969316303730011,
+      "grad_norm": 0.19766977429389954,
       "learning_rate": 0.00018952456483248119,
-      "loss": 0.4284,
+      "loss": 0.4286,
       "step": 147
     },
     {
       "epoch": 0.015847521147874504,
-      "grad_norm": 0.17580685019493103,
+      "grad_norm": 0.17703601717948914,
       "learning_rate": 0.0001893841424151264,
-      "loss": 0.2754,
+      "loss": 0.276,
       "step": 148
     },
     {
       "epoch": 0.015954598993468252,
-      "grad_norm": 0.1819283366203308,
+      "grad_norm": 0.178870290517807,
       "learning_rate": 0.0001892428378123718,
-      "loss": 0.3462,
+      "loss": 0.3468,
       "step": 149
     },
     {
       "epoch": 0.016061676839061997,
-      "grad_norm": 0.1740637719631195,
+      "grad_norm": 0.17410863935947418,
       "learning_rate": 0.0001891006524188368,
-      "loss": 0.3225,
+      "loss": 0.3231,
       "step": 150
     },
     {
       "epoch": 0.016168754684655746,
-      "grad_norm": 0.17211860418319702,
+      "grad_norm": 0.1725064218044281,
       "learning_rate": 0.00018895758763783383,
-      "loss": 0.3688,
+      "loss": 0.3679,
       "step": 151
     },
     {
       "epoch": 0.01627583253024949,
-      "grad_norm": 0.1924057900905609,
+      "grad_norm": 0.194014772772789,
       "learning_rate": 0.00018881364488135448,
-      "loss": 0.3207,
+      "loss": 0.3218,
       "step": 152
     },
     {
       "epoch": 0.01638291037584324,
-      "grad_norm": 0.16835831105709076,
+      "grad_norm": 0.16831275820732117,
       "learning_rate": 0.00018866882557005567,
-      "loss": 0.3377,
+      "loss": 0.338,
       "step": 153
     },
     {
       "epoch": 0.016489988221436984,
-      "grad_norm": 0.15114548802375793,
+      "grad_norm": 0.14952607452869415,
       "learning_rate": 0.00018852313113324552,
-      "loss": 0.2549,
+      "loss": 0.2548,
       "step": 154
     },
     {
       "epoch": 0.016597066067030732,
-      "grad_norm": 0.14319011569023132,
+      "grad_norm": 0.1442236751317978,
       "learning_rate": 0.00018837656300886937,
-      "loss": 0.2206,
+      "loss": 0.2202,
       "step": 155
     },
     {
       "epoch": 0.016704143912624477,
-      "grad_norm": 0.18031498789787292,
+      "grad_norm": 0.1802881807088852,
       "learning_rate": 0.00018822912264349534,
-      "loss": 0.4739,
+      "loss": 0.4736,
       "step": 156
     },
     {
       "epoch": 0.016811221758218226,
-      "grad_norm": 0.25163549184799194,
+      "grad_norm": 0.2469218522310257,
       "learning_rate": 0.00018808081149230036,
-      "loss": 0.4744,
+      "loss": 0.4753,
       "step": 157
     },
     {
       "epoch": 0.01691829960381197,
-      "grad_norm": 0.16316188871860504,
+      "grad_norm": 0.16127042472362518,
       "learning_rate": 0.00018793163101905563,
-      "loss": 0.3651,
+      "loss": 0.3649,
       "step": 158
     },
     {
       "epoch": 0.01702537744940572,
-      "grad_norm": 0.1717788577079773,
+      "grad_norm": 0.17643609642982483,
       "learning_rate": 0.00018778158269611218,
-      "loss": 0.31,
+      "loss": 0.311,
       "step": 159
     },
     {
       "epoch": 0.017132455294999464,
-      "grad_norm": 0.1477271467447281,
+      "grad_norm": 0.14460398256778717,
       "learning_rate": 0.00018763066800438636,
-      "loss": 0.2215,
+      "loss": 0.2213,
       "step": 160
     },
     {
       "epoch": 0.017239533140593212,
-      "grad_norm": 0.15727677941322327,
+      "grad_norm": 0.15388457477092743,
       "learning_rate": 0.0001874788884333453,
-      "loss": 0.3529,
+      "loss": 0.3528,
       "step": 161
     },
     {
       "epoch": 0.017346610986186957,
-      "grad_norm": 0.1828835904598236,
+      "grad_norm": 0.15638676285743713,
       "learning_rate": 0.00018732624548099204,
-      "loss": 0.3712,
+      "loss": 0.3714,
       "step": 162
     },
     {
       "epoch": 0.017453688831780705,
-      "grad_norm": 0.18454089760780334,
+      "grad_norm": 0.1833023875951767,
       "learning_rate": 0.0001871727406538509,
-      "loss": 0.3708,
+      "loss": 0.3702,
       "step": 163
     },
     {
       "epoch": 0.01756076667737445,
-      "grad_norm": 0.26658540964126587,
+      "grad_norm": 0.24797490239143372,
       "learning_rate": 0.0001870183754669526,
-      "loss": 0.3926,
+      "loss": 0.3922,
       "step": 164
     },
     {
       "epoch": 0.0176678445229682,
-      "grad_norm": 0.16646800935268402,
+      "grad_norm": 0.1474548578262329,
       "learning_rate": 0.00018686315144381913,
-      "loss": 0.2482,
+      "loss": 0.2486,
       "step": 165
     },
     {
       "epoch": 0.017774922368561943,
-      "grad_norm": 0.2011224329471588,
+      "grad_norm": 0.19909460842609406,
       "learning_rate": 0.000186707070116449,
-      "loss": 0.3693,
+      "loss": 0.3681,
       "step": 166
     },
     {
       "epoch": 0.017882000214155692,
-      "grad_norm": 0.15221862494945526,
+      "grad_norm": 0.15004006028175354,
       "learning_rate": 0.0001865501330253019,
-      "loss": 0.318,
+      "loss": 0.3179,
       "step": 167
     },
     {
       "epoch": 0.017989078059749437,
-      "grad_norm": 0.22113820910453796,
+      "grad_norm": 0.2146318256855011,
       "learning_rate": 0.00018639234171928353,
-      "loss": 0.3929,
+      "loss": 0.3937,
       "step": 168
     },
     {
       "epoch": 0.018096155905343185,
-      "grad_norm": 0.18122068047523499,
+      "grad_norm": 0.17869114875793457,
       "learning_rate": 0.0001862336977557304,
-      "loss": 0.4554,
+      "loss": 0.4543,
       "step": 169
     },
     {
       "epoch": 0.01820323375093693,
-      "grad_norm": 0.16834978759288788,
+      "grad_norm": 0.17866748571395874,
       "learning_rate": 0.0001860742027003944,
-      "loss": 0.3052,
+      "loss": 0.3069,
       "step": 170
     },
     {
       "epoch": 0.01831031159653068,
-      "grad_norm": 0.18118542432785034,
+      "grad_norm": 0.17818759381771088,
       "learning_rate": 0.00018591385812742725,
-      "loss": 0.2902,
+      "loss": 0.2896,
       "step": 171
     },
     {
       "epoch": 0.018417389442124423,
-      "grad_norm": 0.18435969948768616,
+      "grad_norm": 0.18158233165740967,
       "learning_rate": 0.00018575266561936523,
-      "loss": 0.3397,
+      "loss": 0.3399,
       "step": 172
     },
     {
       "epoch": 0.01852446728771817,
-      "grad_norm": 0.18012931942939758,
+      "grad_norm": 0.17146669328212738,
       "learning_rate": 0.00018559062676711332,
-      "loss": 0.2858,
+      "loss": 0.2842,
       "step": 173
     },
     {
       "epoch": 0.018631545133311916,
-      "grad_norm": 0.2357625663280487,
+      "grad_norm": 0.2015829235315323,
       "learning_rate": 0.0001854277431699295,
-      "loss": 0.3273,
+      "loss": 0.3274,
       "step": 174
     },
     {
       "epoch": 0.018738622978905665,
-      "grad_norm": 0.19773603975772858,
+      "grad_norm": 0.18138733506202698,
       "learning_rate": 0.00018526401643540922,
-      "loss": 0.3522,
+      "loss": 0.3509,
       "step": 175
     },
     {
       "epoch": 0.01884570082449941,
-      "grad_norm": 0.1897646188735962,
+      "grad_norm": 0.18845680356025696,
       "learning_rate": 0.00018509944817946922,
-      "loss": 0.422,
+      "loss": 0.4202,
       "step": 176
     },
     {
       "epoch": 0.018952778670093158,
-      "grad_norm": 0.2005687654018402,
+      "grad_norm": 0.19233529269695282,
       "learning_rate": 0.00018493404002633166,
-      "loss": 0.4571,
+      "loss": 0.4569,
       "step": 177
     },
     {
       "epoch": 0.019059856515686903,
-      "grad_norm": 0.18125112354755402,
+      "grad_norm": 0.1837352216243744,
       "learning_rate": 0.00018476779360850832,
-      "loss": 0.4359,
+      "loss": 0.4374,
       "step": 178
     },
     {
       "epoch": 0.01916693436128065,
-      "grad_norm": 0.2057882696390152,
+      "grad_norm": 0.19979973137378693,
       "learning_rate": 0.00018460071056678422,
-      "loss": 0.4458,
+      "loss": 0.4437,
       "step": 179
     },
     {
       "epoch": 0.019274012206874396,
-      "grad_norm": 0.3153612017631531,
+      "grad_norm": 0.3244735598564148,
       "learning_rate": 0.00018443279255020152,
-      "loss": 0.3945,
+      "loss": 0.3831,
       "step": 180
     },
     {
       "epoch": 0.019381090052468145,
-      "grad_norm": 0.21111486852169037,
+      "grad_norm": 0.21024329960346222,
       "learning_rate": 0.00018426404121604323,
-      "loss": 0.4705,
+      "loss": 0.4677,
       "step": 181
     },
     {
       "epoch": 0.01948816789806189,
-      "grad_norm": 0.2238568216562271,
+      "grad_norm": 0.22042858600616455,
       "learning_rate": 0.00018409445822981693,
-      "loss": 0.3776,
+      "loss": 0.3745,
       "step": 182
     },
     {
       "epoch": 0.019595245743655638,
-      "grad_norm": 0.2252410352230072,
+      "grad_norm": 0.17360983788967133,
       "learning_rate": 0.00018392404526523817,
-      "loss": 0.3556,
+      "loss": 0.3558,
       "step": 183
     },
     {
       "epoch": 0.019702323589249383,
-      "grad_norm": 0.1595439463853836,
+      "grad_norm": 0.1609887331724167,
       "learning_rate": 0.0001837528040042142,
-      "loss": 0.2908,
+      "loss": 0.2913,
       "step": 184
     },
     {
       "epoch": 0.01980940143484313,
-      "grad_norm": 0.17755721509456635,
+      "grad_norm": 0.17246180772781372,
       "learning_rate": 0.00018358073613682706,
-      "loss": 0.3642,
+      "loss": 0.3632,
       "step": 185
     },
     {
       "epoch": 0.019916479280436876,
-      "grad_norm": 0.24803121387958527,
+      "grad_norm": 0.24439960718154907,
       "learning_rate": 0.00018340784336131713,
-      "loss": 0.4246,
+      "loss": 0.423,
       "step": 186
     },
     {
       "epoch": 0.020023557126030624,
-      "grad_norm": 0.1807357370853424,
+      "grad_norm": 0.17695844173431396,
       "learning_rate": 0.00018323412738406635,
-      "loss": 0.3031,
+      "loss": 0.3018,
       "step": 187
     },
     {
       "epoch": 0.02013063497162437,
-      "grad_norm": 0.18305666744709015,
+      "grad_norm": 0.18329234421253204,
       "learning_rate": 0.00018305958991958127,
-      "loss": 0.4081,
+      "loss": 0.4078,
       "step": 188
     },
     {
       "epoch": 0.020237712817218118,
-      "grad_norm": 0.15840698778629303,
+      "grad_norm": 0.1552525758743286,
       "learning_rate": 0.0001828842326904762,
-      "loss": 0.2594,
+      "loss": 0.2596,
       "step": 189
     },
     {
       "epoch": 0.020344790662811866,
-      "grad_norm": 0.1579771190881729,
+      "grad_norm": 0.15624456107616425,
       "learning_rate": 0.00018270805742745617,
-      "loss": 0.2373,
+      "loss": 0.2375,
       "step": 190
     },
     {
       "epoch": 0.02045186850840561,
-      "grad_norm": 0.1966482698917389,
+      "grad_norm": 0.19527368247509003,
       "learning_rate": 0.00018253106586929997,
-      "loss": 0.4255,
+      "loss": 0.4228,
       "step": 191
     },
     {
       "epoch": 0.02055894635399936,
-      "grad_norm": 0.16462944447994232,
+      "grad_norm": 0.1665705144405365,
       "learning_rate": 0.00018235325976284275,
-      "loss": 0.2783,
+      "loss": 0.2786,
       "step": 192
     },
     {
       "epoch": 0.020666024199593104,
-      "grad_norm": 0.15610237419605255,
+      "grad_norm": 0.15713033080101013,
       "learning_rate": 0.00018217464086295904,
-      "loss": 0.2818,
+      "loss": 0.2833,
       "step": 193
     },
     {
       "epoch": 0.020773102045186852,
-      "grad_norm": 0.15715450048446655,
+      "grad_norm": 0.15957246720790863,
       "learning_rate": 0.00018199521093254523,
-      "loss": 0.2853,
+      "loss": 0.2854,
       "step": 194
     },
     {
       "epoch": 0.020880179890780597,
-      "grad_norm": 0.202264666557312,
+      "grad_norm": 0.20240135490894318,
       "learning_rate": 0.00018181497174250236,
-      "loss": 0.3436,
+      "loss": 0.342,
       "step": 195
     },
     {
       "epoch": 0.020987257736374346,
-      "grad_norm": 0.21248966455459595,
+      "grad_norm": 0.21210376918315887,
       "learning_rate": 0.00018163392507171842,
-      "loss": 0.3021,
+      "loss": 0.3008,
       "step": 196
     },
     {
       "epoch": 0.02109433558196809,
-      "grad_norm": 0.14210785925388336,
+      "grad_norm": 0.1445528119802475,
       "learning_rate": 0.00018145207270705096,
-      "loss": 0.246,
+      "loss": 0.2464,
       "step": 197
     },
     {
       "epoch": 0.02120141342756184,
-      "grad_norm": 0.15923386812210083,
+      "grad_norm": 0.1597689390182495,
       "learning_rate": 0.0001812694164433094,
-      "loss": 0.3208,
+      "loss": 0.3196,
       "step": 198
     },
     {
       "epoch": 0.021308491273155584,
-      "grad_norm": 0.16584694385528564,
+      "grad_norm": 0.1642187237739563,
       "learning_rate": 0.00018108595808323736,
-      "loss": 0.281,
+      "loss": 0.2816,
       "step": 199
     },
     {
       "epoch": 0.021415569118749332,
-      "grad_norm": 0.1751406490802765,
+      "grad_norm": 0.17551757395267487,
       "learning_rate": 0.00018090169943749476,
-      "loss": 0.3897,
+      "loss": 0.3887,
       "step": 200
     },
     {
       "epoch": 0.021522646964343077,
-      "grad_norm": 0.1961715668439865,
+      "grad_norm": 0.18966630101203918,
       "learning_rate": 0.00018071664232464002,
-      "loss": 0.4197,
+      "loss": 0.4184,
       "step": 201
     },
     {
       "epoch": 0.021629724809936825,
-      "grad_norm": 0.1973417103290558,
+      "grad_norm": 0.20549455285072327,
       "learning_rate": 0.0001805307885711122,
-      "loss": 0.3362,
+      "loss": 0.3353,
       "step": 202
     },
     {
       "epoch": 0.02173680265553057,
-      "grad_norm": 0.16562364995479584,
+      "grad_norm": 0.16554607450962067,
       "learning_rate": 0.00018034414001121278,
-      "loss": 0.4085,
+      "loss": 0.4082,
       "step": 203
     },
     {
       "epoch": 0.02184388050112432,
-      "grad_norm": 0.1984337717294693,
+      "grad_norm": 0.18681012094020844,
       "learning_rate": 0.00018015669848708767,
-      "loss": 0.2735,
+      "loss": 0.2714,
       "step": 204
     },
     {
       "epoch": 0.021950958346718064,
-      "grad_norm": 0.18170015513896942,
+      "grad_norm": 0.18294884264469147,
       "learning_rate": 0.00017996846584870908,
-      "loss": 0.3562,
+      "loss": 0.3567,
       "step": 205
     },
     {
       "epoch": 0.022058036192311812,
-      "grad_norm": 0.1938287615776062,
+      "grad_norm": 0.1881534308195114,
       "learning_rate": 0.0001797794439538571,
-      "loss": 0.3715,
+      "loss": 0.3717,
       "step": 206
     },
     {
       "epoch": 0.022165114037905557,
-      "grad_norm": 0.16263467073440552,
+      "grad_norm": 0.16362693905830383,
       "learning_rate": 0.0001795896346681016,
       "loss": 0.3797,
       "step": 207
     },
     {
       "epoch": 0.022272191883499305,
-      "grad_norm": 0.1593952178955078,
+      "grad_norm": 0.16024716198444366,
       "learning_rate": 0.00017939903986478355,
-      "loss": 0.2275,
+      "loss": 0.2268,
       "step": 208
     },
     {
       "epoch": 0.02237926972909305,
-      "grad_norm": 0.19579128921031952,
+      "grad_norm": 0.1983472853899002,
       "learning_rate": 0.00017920766142499672,
       "loss": 0.318,
       "step": 209
     },
     {
       "epoch": 0.0224863475746868,
-      "grad_norm": 0.18321353197097778,
+      "grad_norm": 0.1842176765203476,
       "learning_rate": 0.00017901550123756906,
-      "loss": 0.3672,
+      "loss": 0.3663,
       "step": 210
     },
     {
       "epoch": 0.022593425420280543,
-      "grad_norm": 0.20865146815776825,
+      "grad_norm": 0.21368902921676636,
       "learning_rate": 0.00017882256119904403,
-      "loss": 0.4745,
+      "loss": 0.4759,
       "step": 211
     },
     {
       "epoch": 0.02270050326587429,
-      "grad_norm": 0.15021853148937225,
+      "grad_norm": 0.1486794352531433,
       "learning_rate": 0.00017862884321366188,
-      "loss": 0.3397,
+      "loss": 0.3391,
       "step": 212
     },
     {
       "epoch": 0.022807581111468037,
-      "grad_norm": 0.19131535291671753,
+      "grad_norm": 0.19507035613059998,
       "learning_rate": 0.000178434349193341,
-      "loss": 0.3389,
+      "loss": 0.3386,
       "step": 213
     },
     {
       "epoch": 0.022914658957061785,
-      "grad_norm": 0.1574268788099289,
+      "grad_norm": 0.18279972672462463,
       "learning_rate": 0.0001782390810576588,
-      "loss": 0.3331,
+      "loss": 0.3325,
       "step": 214
     },
     {
       "epoch": 0.02302173680265553,
-      "grad_norm": 0.18308749794960022,
+      "grad_norm": 0.18135708570480347,
       "learning_rate": 0.000178043040733833,
-      "loss": 0.3105,
+      "loss": 0.3102,
       "step": 215
     },
     {
       "epoch": 0.023128814648249278,
-      "grad_norm": 0.2218441218137741,
+      "grad_norm": 0.2114657461643219,
       "learning_rate": 0.00017784623015670238,
-      "loss": 0.4791,
+      "loss": 0.4783,
       "step": 216
     },
     {
       "epoch": 0.023235892493843023,
-      "grad_norm": 0.18416514992713928,
+      "grad_norm": 0.1833896040916443,
       "learning_rate": 0.00017764865126870786,
-      "loss": 0.3551,
+      "loss": 0.3528,
       "step": 217
     },
     {
       "epoch": 0.02334297033943677,
-      "grad_norm": 0.15702709555625916,
+      "grad_norm": 0.15574641525745392,
       "learning_rate": 0.00017745030601987337,
-      "loss": 0.3354,
+      "loss": 0.3342,
       "step": 218
     },
     {
       "epoch": 0.023450048185030516,
-      "grad_norm": 0.12538433074951172,
+      "grad_norm": 0.14896464347839355,
       "learning_rate": 0.00017725119636778644,
-      "loss": 0.1759,
+      "loss": 0.1749,
       "step": 219
     },
     {
       "epoch": 0.023557126030624265,
-      "grad_norm": 0.1331990659236908,
+      "grad_norm": 0.13782522082328796,
       "learning_rate": 0.00017705132427757895,
-      "loss": 0.2875,
+      "loss": 0.2891,
       "step": 220
     },
     {
       "epoch": 0.02366420387621801,
-      "grad_norm": 0.18505120277404785,
+      "grad_norm": 0.18487310409545898,
       "learning_rate": 0.00017685069172190766,
-      "loss": 0.4,
+      "loss": 0.3989,
       "step": 221
     },
     {
       "epoch": 0.023771281721811758,
-      "grad_norm": 0.17060880362987518,
+      "grad_norm": 0.16647568345069885,
       "learning_rate": 0.00017664930068093498,
-      "loss": 0.4008,
+      "loss": 0.3998,
       "step": 222
     },
     {
       "epoch": 0.023878359567405503,
-      "grad_norm": 0.151236429810524,
+      "grad_norm": 0.1517525464296341,
       "learning_rate": 0.00017644715314230918,
-      "loss": 0.3216,
+      "loss": 0.3215,
       "step": 223
     },
     {
       "epoch": 0.02398543741299925,
-      "grad_norm": 0.16337113082408905,
+      "grad_norm": 0.16107575595378876,
       "learning_rate": 0.0001762442511011448,
-      "loss": 0.3807,
+      "loss": 0.3801,
       "step": 224
     },
     {
       "epoch": 0.024092515258592996,
-      "grad_norm": 0.18361254036426544,
+      "grad_norm": 0.18223942816257477,
       "learning_rate": 0.0001760405965600031,
-      "loss": 0.4019,
+      "loss": 0.4009,
       "step": 225
     },
     {
       "epoch": 0.024199593104186744,
-      "grad_norm": 0.1295214742422104,
+      "grad_norm": 0.12823620438575745,
       "learning_rate": 0.0001758361915288722,
-      "loss": 0.2674,
+      "loss": 0.2689,
       "step": 226
     },
     {
       "epoch": 0.02430667094978049,
-      "grad_norm": 0.1772903949022293,
+      "grad_norm": 0.1777333915233612,
       "learning_rate": 0.0001756310380251472,
-      "loss": 0.3025,
+      "loss": 0.3016,
       "step": 227
     },
     {
       "epoch": 0.024413748795374238,
-      "grad_norm": 0.20191480219364166,
+      "grad_norm": 0.20274615287780762,
       "learning_rate": 0.00017542513807361037,
-      "loss": 0.3619,
+      "loss": 0.3624,
       "step": 228
     },
     {
       "epoch": 0.024520826640967983,
-      "grad_norm": 0.20930209755897522,
+      "grad_norm": 0.4989624619483948,
       "learning_rate": 0.00017521849370641114,
-      "loss": 0.2344,
+      "loss": 0.237,
       "step": 229
     },
     {
       "epoch": 0.02462790448656173,
-      "grad_norm": 0.18147680163383484,
+      "grad_norm": 0.1811639964580536,
       "learning_rate": 0.00017501110696304596,
-      "loss": 0.3712,
+      "loss": 0.3738,
       "step": 230
     },
     {
       "epoch": 0.024734982332155476,
-      "grad_norm": 0.17919720709323883,
+      "grad_norm": 0.2014167159795761,
       "learning_rate": 0.00017480297989033825,
-      "loss": 0.3728,
+      "loss": 0.3726,
       "step": 231
     },
     {
       "epoch": 0.024842060177749224,
-      "grad_norm": 0.16749288141727448,
+      "grad_norm": 0.16710756719112396,
       "learning_rate": 0.00017459411454241822,
-      "loss": 0.3647,
+      "loss": 0.3655,
       "step": 232
     },
     {
       "epoch": 0.02494913802334297,
-      "grad_norm": 0.14664986729621887,
+      "grad_norm": 0.1497872918844223,
       "learning_rate": 0.00017438451298070252,
-      "loss": 0.1905,
+      "loss": 0.1916,
       "step": 233
     },
     {
       "epoch": 0.025056215868936717,
-      "grad_norm": 0.1727173924446106,
+      "grad_norm": 0.17602579295635223,
       "learning_rate": 0.00017417417727387394,
-      "loss": 0.3138,
+      "loss": 0.3154,
       "step": 234
     },
     {
       "epoch": 0.025163293714530462,
-      "grad_norm": 0.16467280685901642,
+      "grad_norm": 0.17509448528289795,
       "learning_rate": 0.000173963109497861,
-      "loss": 0.2621,
+      "loss": 0.2654,
       "step": 235
     },
     {
       "epoch": 0.02527037156012421,
-      "grad_norm": 0.21190643310546875,
+      "grad_norm": 0.22780412435531616,
       "learning_rate": 0.0001737513117358174,
-      "loss": 0.4417,
+      "loss": 0.4443,
       "step": 236
     },
     {
       "epoch": 0.025377449405717956,
-      "grad_norm": 0.191368967294693,
+      "grad_norm": 0.15670035779476166,
       "learning_rate": 0.0001735387860781016,
-      "loss": 0.3023,
+      "loss": 0.2997,
       "step": 237
     },
     {
       "epoch": 0.025484527251311704,
-      "grad_norm": 0.17699618637561798,
+      "grad_norm": 0.20101696252822876,
       "learning_rate": 0.00017332553462225602,
-      "loss": 0.3652,
+      "loss": 0.3639,
       "step": 238
     },
     {
       "epoch": 0.02559160509690545,
-      "grad_norm": 0.1797989159822464,
+      "grad_norm": 0.18168163299560547,
       "learning_rate": 0.00017311155947298643,
-      "loss": 0.4054,
+      "loss": 0.4077,
       "step": 239
     },
     {
       "epoch": 0.025698682942499197,
-      "grad_norm": 0.16934357583522797,
+      "grad_norm": 0.1704721599817276,
       "learning_rate": 0.00017289686274214118,
-      "loss": 0.2369,
+      "loss": 0.2374,
       "step": 240
     },
     {
       "epoch": 0.025805760788092942,
-      "grad_norm": 0.1936860829591751,
+      "grad_norm": 0.18674886226654053,
       "learning_rate": 0.0001726814465486903,
-      "loss": 0.3687,
+      "loss": 0.368,
       "step": 241
     },
     {
       "epoch": 0.02591283863368669,
-      "grad_norm": 0.19604068994522095,
+      "grad_norm": 0.19927389919757843,
       "learning_rate": 0.0001724653130187047,
-      "loss": 0.409,
+      "loss": 0.4098,
       "step": 242
     },
     {
       "epoch": 0.026019916479280435,
-      "grad_norm": 0.16098040342330933,
+      "grad_norm": 0.159666508436203,
       "learning_rate": 0.00017224846428533499,
-      "loss": 0.3661,
+      "loss": 0.3666,
       "step": 243
     },
     {
       "epoch": 0.026126994324874184,
-      "grad_norm": 0.1543160378932953,
+      "grad_norm": 0.15318766236305237,
       "learning_rate": 0.0001720309024887907,
-      "loss": 0.286,
+      "loss": 0.2852,
       "step": 244
     },
     {
       "epoch": 0.02623407217046793,
-      "grad_norm": 0.14484800398349762,
+      "grad_norm": 0.1432930827140808,
       "learning_rate": 0.00017181262977631888,
-      "loss": 0.277,
+      "loss": 0.2757,
       "step": 245
     },
     {
       "epoch": 0.026341150016061677,
-      "grad_norm": 0.18634939193725586,
+      "grad_norm": 0.18784955143928528,
       "learning_rate": 0.00017159364830218312,
-      "loss": 0.2708,
+      "loss": 0.2719,
       "step": 246
     },
     {
       "epoch": 0.026448227861655422,
-      "grad_norm": 0.14909416437149048,
+      "grad_norm": 0.15523239970207214,
       "learning_rate": 0.00017137396022764214,
-      "loss": 0.2739,
+      "loss": 0.2753,
       "step": 247
     },
     {
       "epoch": 0.02655530570724917,
-      "grad_norm": 0.17528975009918213,
+      "grad_norm": 0.17909985780715942,
       "learning_rate": 0.00017115356772092857,
-      "loss": 0.3052,
+      "loss": 0.3062,
       "step": 248
     },
     {
       "epoch": 0.026662383552842915,
-      "grad_norm": 0.17353974282741547,
+      "grad_norm": 0.17707036435604095,
       "learning_rate": 0.0001709324729572274,
-      "loss": 0.3065,
+      "loss": 0.3064,
       "step": 249
     },
     {
       "epoch": 0.026769461398436663,
-      "grad_norm": 0.15688441693782806,
+      "grad_norm": 0.15941859781742096,
       "learning_rate": 0.00017071067811865476,
-      "loss": 0.3088,
+      "loss": 0.3092,
       "step": 250
     },
     {
       "epoch": 0.026876539244030412,
-      "grad_norm": 0.18625454604625702,
+      "grad_norm": 0.18870612978935242,
       "learning_rate": 0.00017048818539423615,
-      "loss": 0.4439,
+      "loss": 0.4432,
       "step": 251
     },
     {
       "epoch": 0.026983617089624157,
-      "grad_norm": 0.17043183743953705,
+      "grad_norm": 0.1703352928161621,
       "learning_rate": 0.00017026499697988493,
-      "loss": 0.37,
+      "loss": 0.3708,
       "step": 252
     },
     {
       "epoch": 0.027090694935217905,
-      "grad_norm": 0.21933884918689728,
+      "grad_norm": 0.2195807695388794,
       "learning_rate": 0.00017004111507838064,
-      "loss": 0.4339,
+      "loss": 0.4352,
       "step": 253
     },
     {
       "epoch": 0.02719777278081165,
-      "grad_norm": 0.21618066728115082,
+      "grad_norm": 0.2179376184940338,
       "learning_rate": 0.00016981654189934727,
-      "loss": 0.3882,
+      "loss": 0.3906,
       "step": 254
     },
     {
       "epoch": 0.0273048506264054,
-      "grad_norm": 0.16221043467521667,
+      "grad_norm": 0.16640064120292664,
       "learning_rate": 0.00016959127965923142,
-      "loss": 0.2924,
+      "loss": 0.2906,
       "step": 255
     },
     {
       "epoch": 0.027411928471999143,
-      "grad_norm": 0.14445362985134125,
+      "grad_norm": 0.1458374410867691,
       "learning_rate": 0.0001693653305812805,
-      "loss": 0.2702,
+      "loss": 0.2691,
       "step": 256
     },
     {
       "epoch": 0.02751900631759289,
-      "grad_norm": 0.15952937304973602,
+      "grad_norm": 0.15912285447120667,
       "learning_rate": 0.00016913869689552064,
-      "loss": 0.3569,
+      "loss": 0.3578,
       "step": 257
     },
     {
       "epoch": 0.027626084163186636,
-      "grad_norm": 0.1786893904209137,
+      "grad_norm": 0.17868658900260925,
       "learning_rate": 0.00016891138083873487,
-      "loss": 0.4076,
+      "loss": 0.4068,
       "step": 258
     },
     {
       "epoch": 0.027733162008780385,
-      "grad_norm": 0.16461695730686188,
+      "grad_norm": 0.1656470149755478,
       "learning_rate": 0.00016868338465444085,
       "loss": 0.3617,
       "step": 259
     },
     {
       "epoch": 0.02784023985437413,
-      "grad_norm": 0.19281861186027527,
+      "grad_norm": 0.1929635852575302,
       "learning_rate": 0.00016845471059286887,
       "loss": 0.3202,
       "step": 260
     },
     {
       "epoch": 0.027947317699967878,
-      "grad_norm": 0.15104186534881592,
+      "grad_norm": 0.1525058150291443,
       "learning_rate": 0.00016822536091093965,
-      "loss": 0.3157,
+      "loss": 0.3161,
       "step": 261
     },
     {
       "epoch": 0.028054395545561623,
-      "grad_norm": 0.1911495327949524,
+      "grad_norm": 0.1779276728630066,
       "learning_rate": 0.00016799533787224192,
-      "loss": 0.385,
+      "loss": 0.3852,
       "step": 262
     },
     {
       "epoch": 0.02816147339115537,
-      "grad_norm": 0.17329667508602142,
+      "grad_norm": 0.16779017448425293,
       "learning_rate": 0.00016776464374701025,
-      "loss": 0.3613,
+      "loss": 0.3606,
       "step": 263
     },
     {
       "epoch": 0.028268551236749116,
-      "grad_norm": 0.15046873688697815,
+      "grad_norm": 0.14932067692279816,
       "learning_rate": 0.00016753328081210245,
-      "loss": 0.2792,
+      "loss": 0.2798,
       "step": 264
     },
     {
       "epoch": 0.028375629082342865,
-      "grad_norm": 0.1259920299053192,
+      "grad_norm": 0.12646813690662384,
       "learning_rate": 0.00016730125135097735,
-      "loss": 0.2741,
+      "loss": 0.2752,
       "step": 265
     },
     {
       "epoch": 0.02848270692793661,
-      "grad_norm": 0.19736653566360474,
+      "grad_norm": 0.1634105145931244,
       "learning_rate": 0.000167068557653672,
-      "loss": 0.3415,
+      "loss": 0.3424,
       "step": 266
     },
     {
       "epoch": 0.028589784773530358,
-      "grad_norm": 0.12615585327148438,
+      "grad_norm": 0.12478397786617279,
       "learning_rate": 0.0001668352020167793,
-      "loss": 0.2148,
+      "loss": 0.2156,
       "step": 267
     },
     {
       "epoch": 0.028696862619124103,
-      "grad_norm": 0.13925912976264954,
+      "grad_norm": 0.13762520253658295,
       "learning_rate": 0.00016660118674342517,
-      "loss": 0.2772,
+      "loss": 0.2768,
       "step": 268
     },
     {
       "epoch": 0.02880394046471785,
-      "grad_norm": 0.15701375901699066,
+      "grad_norm": 0.15719301998615265,
       "learning_rate": 0.00016636651414324587,
-      "loss": 0.3165,
+      "loss": 0.3163,
       "step": 269
     },
     {
       "epoch": 0.028911018310311596,
-      "grad_norm": 0.1971178501844406,
+      "grad_norm": 0.19167502224445343,
       "learning_rate": 0.00016613118653236518,
-      "loss": 0.336,
+      "loss": 0.3347,
       "step": 270
     },
     {
       "epoch": 0.029018096155905344,
-      "grad_norm": 0.19260185956954956,
+      "grad_norm": 0.218104287981987,
       "learning_rate": 0.0001658952062333717,
-      "loss": 0.3047,
+      "loss": 0.3021,
       "step": 271
     },
     {
       "epoch": 0.02912517400149909,
-      "grad_norm": 0.17433519661426544,
+      "grad_norm": 0.16969047486782074,
       "learning_rate": 0.00016565857557529566,
-      "loss": 0.3392,
+      "loss": 0.3387,
       "step": 272
     },
     {
       "epoch": 0.029232251847092838,
-      "grad_norm": 0.18723376095294952,
+      "grad_norm": 0.19966812431812286,
       "learning_rate": 0.00016542129689358612,
-      "loss": 0.3043,
+      "loss": 0.3035,
       "step": 273
     },
     {
       "epoch": 0.029339329692686583,
-      "grad_norm": 0.12574511766433716,
+      "grad_norm": 0.12486845999956131,
       "learning_rate": 0.0001651833725300879,
-      "loss": 0.2972,
+      "loss": 0.2951,
       "step": 274
     },
     {
       "epoch": 0.02944640753828033,
-      "grad_norm": 0.14964695274829865,
+      "grad_norm": 0.1498948633670807,
       "learning_rate": 0.00016494480483301836,
-      "loss": 0.3088,
+      "loss": 0.3087,
       "step": 275
     },
     {
       "epoch": 0.029553485383874076,
-      "grad_norm": 0.17207567393779755,
+      "grad_norm": 0.16795475780963898,
       "learning_rate": 0.00016470559615694446,
-      "loss": 0.3842,
+      "loss": 0.3827,
       "step": 276
     },
     {
       "epoch": 0.029660563229467824,
-      "grad_norm": 0.12991392612457275,
+      "grad_norm": 0.1276923269033432,
       "learning_rate": 0.00016446574886275913,
-      "loss": 0.2479,
+      "loss": 0.2476,
       "step": 277
     },
     {
       "epoch": 0.02976764107506157,
-      "grad_norm": 0.14477019011974335,
+      "grad_norm": 0.14436961710453033,
       "learning_rate": 0.00016422526531765846,
-      "loss": 0.3189,
+      "loss": 0.3199,
       "step": 278
     },
     {
       "epoch": 0.029874718920655317,
-      "grad_norm": 0.20534910261631012,
+      "grad_norm": 0.2020837962627411,
       "learning_rate": 0.00016398414789511786,
-      "loss": 0.3989,
+      "loss": 0.3996,
       "step": 279
     },
     {
       "epoch": 0.029981796766249062,
-      "grad_norm": 0.14957323670387268,
+      "grad_norm": 0.15087634325027466,
       "learning_rate": 0.000163742398974869,
-      "loss": 0.265,
+      "loss": 0.2648,
       "step": 280
     },
     {
       "epoch": 0.03008887461184281,
-      "grad_norm": 0.17124171555042267,
+      "grad_norm": 0.17405913770198822,
       "learning_rate": 0.00016350002094287609,
-      "loss": 0.3954,
+      "loss": 0.3969,
       "step": 281
     },
     {
       "epoch": 0.030195952457436556,
-      "grad_norm": 0.15740445256233215,
+      "grad_norm": 0.15879938006401062,
       "learning_rate": 0.00016325701619131246,
-      "loss": 0.3357,
+      "loss": 0.3349,
       "step": 282
     },
     {
       "epoch": 0.030303030303030304,
-      "grad_norm": 0.1666688174009323,
+      "grad_norm": 0.16552022099494934,
       "learning_rate": 0.00016301338711853693,
-      "loss": 0.3495,
+      "loss": 0.3489,
       "step": 283
     },
     {
       "epoch": 0.03041010814862405,
-      "grad_norm": 0.18019483983516693,
+      "grad_norm": 0.201536163687706,
       "learning_rate": 0.00016276913612907007,
-      "loss": 0.4245,
+      "loss": 0.4232,
       "step": 284
     },
     {
       "epoch": 0.030517185994217797,
-      "grad_norm": 0.21450522541999817,
+      "grad_norm": 0.22116144001483917,
       "learning_rate": 0.00016252426563357055,
-      "loss": 0.4582,
+      "loss": 0.4583,
       "step": 285
     },
     {
       "epoch": 0.030624263839811542,
-      "grad_norm": 0.21052783727645874,
+      "grad_norm": 0.21266256272792816,
       "learning_rate": 0.00016227877804881127,
       "loss": 0.4128,
       "step": 286
     },
     {
       "epoch": 0.03073134168540529,
-      "grad_norm": 0.1664145290851593,
+      "grad_norm": 0.16596044600009918,
       "learning_rate": 0.00016203267579765563,
-      "loss": 0.279,
+      "loss": 0.2805,
       "step": 287
     },
     {
       "epoch": 0.030838419530999035,
-      "grad_norm": 0.2330683022737503,
+      "grad_norm": 0.233629509806633,
       "learning_rate": 0.00016178596130903344,
-      "loss": 0.3742,
+      "loss": 0.3752,
       "step": 288
     },
     {
       "epoch": 0.030945497376592784,
-      "grad_norm": 0.18258881568908691,
+      "grad_norm": 0.182915598154068,
       "learning_rate": 0.00016153863701791717,
-      "loss": 0.3266,
+      "loss": 0.3252,
       "step": 289
     },
     {
       "epoch": 0.03105257522218653,
-      "grad_norm": 0.16531600058078766,
+      "grad_norm": 0.16230888664722443,
       "learning_rate": 0.00016129070536529766,
-      "loss": 0.2748,
+      "loss": 0.2728,
       "step": 290
     },
     {
       "epoch": 0.031159653067780277,
-      "grad_norm": 0.21928264200687408,
+      "grad_norm": 0.21524645388126373,
       "learning_rate": 0.00016104216879816026,
-      "loss": 0.3977,
+      "loss": 0.4,
       "step": 291
     },
     {
       "epoch": 0.03126673091337402,
-      "grad_norm": 0.1794746071100235,
+      "grad_norm": 0.18035411834716797,
       "learning_rate": 0.00016079302976946055,
-      "loss": 0.3939,
+      "loss": 0.3949,
       "step": 292
     },
     {
       "epoch": 0.03137380875896777,
-      "grad_norm": 0.14978636801242828,
+      "grad_norm": 0.15152864158153534,
       "learning_rate": 0.00016054329073810015,
-      "loss": 0.2803,
+      "loss": 0.2806,
       "step": 293
     },
     {
       "epoch": 0.03148088660456152,
-      "grad_norm": 0.1669725626707077,
+      "grad_norm": 0.16412824392318726,
       "learning_rate": 0.00016029295416890248,
-      "loss": 0.2965,
+      "loss": 0.2953,
       "step": 294
     },
     {
       "epoch": 0.03158796445015526,
-      "grad_norm": 0.15341994166374207,
+      "grad_norm": 0.155482217669487,
       "learning_rate": 0.00016004202253258842,
-      "loss": 0.2785,
+      "loss": 0.2799,
       "step": 295
     },
     {
       "epoch": 0.03169504229574901,
-      "grad_norm": 0.13306757807731628,
+      "grad_norm": 0.13629105687141418,
       "learning_rate": 0.0001597904983057519,
-      "loss": 0.2762,
+      "loss": 0.2761,
       "step": 296
     },
     {
       "epoch": 0.03180212014134275,
-      "grad_norm": 0.19124022126197815,
+      "grad_norm": 0.18656809628009796,
       "learning_rate": 0.00015953838397083552,
-      "loss": 0.2643,
+      "loss": 0.2649,
       "step": 297
     },
     {
       "epoch": 0.031909197986936505,
-      "grad_norm": 0.17234952747821808,
+      "grad_norm": 0.1716616302728653,
       "learning_rate": 0.00015928568201610595,
-      "loss": 0.2988,
+      "loss": 0.2967,
       "step": 298
     },
     {
       "epoch": 0.03201627583253025,
-      "grad_norm": 0.19267751276493073,
+      "grad_norm": 0.19273731112480164,
       "learning_rate": 0.00015903239493562948,
-      "loss": 0.4104,
+      "loss": 0.4114,
       "step": 299
     },
     {
       "epoch": 0.032123353678123995,
-      "grad_norm": 0.17110766470432281,
+      "grad_norm": 0.16990764439105988,
       "learning_rate": 0.00015877852522924732,
       "loss": 0.3486,
       "step": 300
     },
     {
       "epoch": 0.03223043152371774,
-      "grad_norm": 0.16783058643341064,
+      "grad_norm": 0.17067308723926544,
       "learning_rate": 0.00015852407540255104,
-      "loss": 0.3042,
+      "loss": 0.3064,
       "step": 301
     },
     {
       "epoch": 0.03233750936931149,
-      "grad_norm": 0.15925949811935425,
+      "grad_norm": 0.15841452777385712,
       "learning_rate": 0.00015826904796685762,
-      "loss": 0.3339,
+      "loss": 0.3338,
       "step": 302
     },
     {
       "epoch": 0.032444587214905236,
-      "grad_norm": 0.17646819353103638,
+      "grad_norm": 0.1738758087158203,
       "learning_rate": 0.00015801344543918495,
-      "loss": 0.3925,
+      "loss": 0.3908,
       "step": 303
     },
     {
       "epoch": 0.03255166506049898,
-      "grad_norm": 0.1683570295572281,
+      "grad_norm": 0.16537289321422577,
       "learning_rate": 0.00015775727034222675,
-      "loss": 0.4217,
+      "loss": 0.4218,
       "step": 304
     },
     {
       "epoch": 0.032658742906092726,
-      "grad_norm": 0.17200106382369995,
+      "grad_norm": 0.1708252876996994,
       "learning_rate": 0.00015750052520432787,
-      "loss": 0.3305,
+      "loss": 0.3284,
       "step": 305
     },
     {
       "epoch": 0.03276582075168648,
-      "grad_norm": 0.15256130695343018,
+      "grad_norm": 0.14976796507835388,
       "learning_rate": 0.0001572432125594591,
-      "loss": 0.288,
+      "loss": 0.2875,
       "step": 306
     },
     {
       "epoch": 0.03287289859728022,
-      "grad_norm": 0.16523954272270203,
+      "grad_norm": 0.16668827831745148,
       "learning_rate": 0.00015698533494719238,
-      "loss": 0.3902,
+      "loss": 0.3899,
       "step": 307
     },
     {
       "epoch": 0.03297997644287397,
-      "grad_norm": 0.12929733097553253,
+      "grad_norm": 0.12908723950386047,
       "learning_rate": 0.00015672689491267567,
-      "loss": 0.242,
+      "loss": 0.2418,
       "step": 308
     },
     {
       "epoch": 0.03308705428846771,
-      "grad_norm": 0.1565142422914505,
+      "grad_norm": 0.15876644849777222,
       "learning_rate": 0.00015646789500660773,
-      "loss": 0.3979,
+      "loss": 0.3975,
       "step": 309
     },
     {
       "epoch": 0.033194132134061465,
-      "grad_norm": 0.17319612205028534,
+      "grad_norm": 0.17148083448410034,
       "learning_rate": 0.00015620833778521307,
-      "loss": 0.4318,
+      "loss": 0.4312,
       "step": 310
     },
     {
       "epoch": 0.03330120997965521,
-      "grad_norm": 0.18638555705547333,
+      "grad_norm": 0.18296833336353302,
       "learning_rate": 0.0001559482258102167,
-      "loss": 0.3411,
+      "loss": 0.3408,
       "step": 311
     },
     {
       "epoch": 0.033408287825248954,
-      "grad_norm": 0.1663195937871933,
+      "grad_norm": 0.16649243235588074,
       "learning_rate": 0.00015568756164881882,
       "loss": 0.3898,
       "step": 312
     },
     {
       "epoch": 0.0335153656708427,
-      "grad_norm": 0.12741482257843018,
+      "grad_norm": 0.12311869859695435,
       "learning_rate": 0.00015542634787366942,
-      "loss": 0.2005,
+      "loss": 0.1999,
       "step": 313
     },
     {
       "epoch": 0.03362244351643645,
-      "grad_norm": 0.14058327674865723,
+      "grad_norm": 0.1413881629705429,
       "learning_rate": 0.00015516458706284303,
-      "loss": 0.2665,
+      "loss": 0.2685,
       "step": 314
     },
     {
       "epoch": 0.033729521362030196,
-      "grad_norm": 0.14359046518802643,
+      "grad_norm": 0.1433306783437729,
       "learning_rate": 0.0001549022817998132,
-      "loss": 0.3479,
+      "loss": 0.3465,
       "step": 315
     },
     {
       "epoch": 0.03383659920762394,
-      "grad_norm": 0.17742228507995605,
+      "grad_norm": 0.17652204632759094,
       "learning_rate": 0.00015463943467342693,
-      "loss": 0.4099,
+      "loss": 0.4089,
       "step": 316
     },
     {
       "epoch": 0.03394367705321769,
-      "grad_norm": 0.12765684723854065,
+      "grad_norm": 0.12554718554019928,
       "learning_rate": 0.00015437604827787927,
-      "loss": 0.2812,
+      "loss": 0.2809,
       "step": 317
     },
     {
       "epoch": 0.03405075489881144,
-      "grad_norm": 0.1804419457912445,
+      "grad_norm": 0.1779223084449768,
       "learning_rate": 0.00015411212521268758,
-      "loss": 0.2586,
+      "loss": 0.2576,
       "step": 318
     },
     {
       "epoch": 0.03415783274440518,
-      "grad_norm": 0.21663552522659302,
+      "grad_norm": 0.2108379304409027,
       "learning_rate": 0.00015384766808266602,
-      "loss": 0.3947,
+      "loss": 0.3919,
       "step": 319
     },
     {
       "epoch": 0.03426491058999893,
-      "grad_norm": 0.15343870222568512,
+      "grad_norm": 0.15554186701774597,
       "learning_rate": 0.00015358267949789966,
-      "loss": 0.3475,
+      "loss": 0.3479,
       "step": 320
     },
     {
       "epoch": 0.03437198843559268,
-      "grad_norm": 0.1694641411304474,
+      "grad_norm": 0.1714179813861847,
       "learning_rate": 0.00015331716207371888,
-      "loss": 0.36,
+      "loss": 0.3595,
       "step": 321
     },
     {
       "epoch": 0.034479066281186424,
-      "grad_norm": 0.24741993844509125,
+      "grad_norm": 0.2408221811056137,
       "learning_rate": 0.0001530511184306734,
-      "loss": 0.3511,
+      "loss": 0.3556,
       "step": 322
     },
     {
       "epoch": 0.03458614412678017,
-      "grad_norm": 0.17976830899715424,
+      "grad_norm": 0.17631877958774567,
       "learning_rate": 0.00015278455119450664,
-      "loss": 0.3648,
+      "loss": 0.3639,
       "step": 323
     },
     {
       "epoch": 0.034693221972373914,
-      "grad_norm": 0.1624085009098053,
+      "grad_norm": 0.16605743765830994,
       "learning_rate": 0.0001525174629961296,
-      "loss": 0.399,
+      "loss": 0.3989,
       "step": 324
     },
     {
       "epoch": 0.034800299817967666,
-      "grad_norm": 0.14293444156646729,
+      "grad_norm": 0.14433328807353973,
       "learning_rate": 0.0001522498564715949,
-      "loss": 0.3321,
+      "loss": 0.3316,
       "step": 325
     },
     {
       "epoch": 0.03490737766356141,
-      "grad_norm": 0.1629532128572464,
+      "grad_norm": 0.16017615795135498,
       "learning_rate": 0.00015198173426207094,
-      "loss": 0.3283,
+      "loss": 0.3285,
       "step": 326
     },
     {
       "epoch": 0.035014455509155155,
-      "grad_norm": 0.1322093904018402,
+      "grad_norm": 0.1323675662279129,
       "learning_rate": 0.00015171309901381572,
-      "loss": 0.2678,
+      "loss": 0.2674,
       "step": 327
     },
     {
       "epoch": 0.0351215333547489,
-      "grad_norm": 0.17138321697711945,
+      "grad_norm": 0.17207059264183044,
       "learning_rate": 0.00015144395337815064,
-      "loss": 0.2985,
+      "loss": 0.2994,
       "step": 328
     },
     {
       "epoch": 0.03522861120034265,
-      "grad_norm": 0.17492185533046722,
+      "grad_norm": 0.18844406306743622,
       "learning_rate": 0.00015117430001143452,
-      "loss": 0.3936,
+      "loss": 0.3926,
       "step": 329
     },
     {
       "epoch": 0.0353356890459364,
-      "grad_norm": 0.15431442856788635,
+      "grad_norm": 0.15473999083042145,
       "learning_rate": 0.00015090414157503714,
-      "loss": 0.2784,
+      "loss": 0.279,
       "step": 330
     },
     {
       "epoch": 0.03544276689153014,
-      "grad_norm": 0.14638526737689972,
+      "grad_norm": 0.1484527736902237,
       "learning_rate": 0.00015063348073531324,
-      "loss": 0.2342,
+      "loss": 0.2345,
       "step": 331
     },
     {
       "epoch": 0.03554984473712389,
-      "grad_norm": 0.14522488415241241,
+      "grad_norm": 0.14658762514591217,
       "learning_rate": 0.0001503623201635761,
-      "loss": 0.2721,
+      "loss": 0.2724,
       "step": 332
     },
     {
       "epoch": 0.03565692258271764,
-      "grad_norm": 0.1473628282546997,
+      "grad_norm": 0.14875711500644684,
       "learning_rate": 0.000150090662536071,
-      "loss": 0.2752,
+      "loss": 0.2749,
       "step": 333
     },
     {
       "epoch": 0.035764000428311384,
-      "grad_norm": 0.18698623776435852,
+      "grad_norm": 0.1902220994234085,
       "learning_rate": 0.0001498185105339491,
-      "loss": 0.3196,
+      "loss": 0.3198,
       "step": 334
     },
     {
       "epoch": 0.03587107827390513,
-      "grad_norm": 0.16111692786216736,
+      "grad_norm": 0.15855100750923157,
       "learning_rate": 0.00014954586684324078,
-      "loss": 0.3039,
+      "loss": 0.3031,
       "step": 335
     },
     {
       "epoch": 0.03597815611949887,
-      "grad_norm": 0.16252991557121277,
+      "grad_norm": 0.16661474108695984,
       "learning_rate": 0.00014927273415482915,
-      "loss": 0.2696,
+      "loss": 0.2691,
       "step": 336
     },
     {
       "epoch": 0.036085233965092625,
-      "grad_norm": 0.16255027055740356,
+      "grad_norm": 0.16410493850708008,
       "learning_rate": 0.00014899911516442365,
-      "loss": 0.2249,
+      "loss": 0.2248,
       "step": 337
     },
     {
       "epoch": 0.03619231181068637,
-      "grad_norm": 0.14544181525707245,
+      "grad_norm": 0.14755503833293915,
       "learning_rate": 0.00014872501257253323,
-      "loss": 0.2746,
+      "loss": 0.275,
       "step": 338
     },
     {
       "epoch": 0.036299389656280115,
-      "grad_norm": 0.13909004628658295,
+      "grad_norm": 0.1402604728937149,
       "learning_rate": 0.0001484504290844398,
-      "loss": 0.2982,
+      "loss": 0.2979,
       "step": 339
     },
     {
       "epoch": 0.03640646750187386,
-      "grad_norm": 0.20055408775806427,
+      "grad_norm": 0.20232371985912323,
       "learning_rate": 0.00014817536741017152,
-      "loss": 0.357,
+      "loss": 0.3584,
       "step": 340
     },
     {
       "epoch": 0.03651354534746761,
-      "grad_norm": 0.1298537254333496,
+      "grad_norm": 0.13137860596179962,
       "learning_rate": 0.00014789983026447612,
-      "loss": 0.2204,
+      "loss": 0.2216,
       "step": 341
     },
     {
       "epoch": 0.03662062319306136,
-      "grad_norm": 0.19047404825687408,
+      "grad_norm": 0.18769167363643646,
       "learning_rate": 0.0001476238203667939,
-      "loss": 0.2952,
+      "loss": 0.2955,
       "step": 342
     },
     {
       "epoch": 0.0367277010386551,
-      "grad_norm": 0.18302544951438904,
+      "grad_norm": 0.17783814668655396,
       "learning_rate": 0.0001473473404412312,
-      "loss": 0.2872,
+      "loss": 0.2859,
       "step": 343
     },
     {
       "epoch": 0.036834778884248846,
-      "grad_norm": 0.1702168881893158,
+      "grad_norm": 0.17168748378753662,
       "learning_rate": 0.0001470703932165333,
-      "loss": 0.332,
+      "loss": 0.3324,
       "step": 344
     },
     {
       "epoch": 0.0369418567298426,
-      "grad_norm": 0.20581865310668945,
+      "grad_norm": 0.2099820077419281,
       "learning_rate": 0.00014679298142605734,
-      "loss": 0.4375,
+      "loss": 0.4388,
       "step": 345
     },
     {
       "epoch": 0.03704893457543634,
-      "grad_norm": 0.17323224246501923,
+      "grad_norm": 0.17469075322151184,
       "learning_rate": 0.00014651510780774583,
-      "loss": 0.3551,
+      "loss": 0.3557,
       "step": 346
     },
     {
       "epoch": 0.03715601242103009,
-      "grad_norm": 0.149403378367424,
+      "grad_norm": 0.1499973088502884,
       "learning_rate": 0.00014623677510409918,
-      "loss": 0.2716,
+      "loss": 0.2724,
       "step": 347
     },
     {
       "epoch": 0.03726309026662383,
-      "grad_norm": 0.15081925690174103,
+      "grad_norm": 0.1490471214056015,
       "learning_rate": 0.00014595798606214882,
-      "loss": 0.311,
+      "loss": 0.3109,
       "step": 348
     },
     {
       "epoch": 0.037370168112217585,
-      "grad_norm": 0.18388235569000244,
+      "grad_norm": 0.18719898164272308,
       "learning_rate": 0.00014567874343342997,
-      "loss": 0.36,
+      "loss": 0.3609,
       "step": 349
     },
     {
       "epoch": 0.03747724595781133,
-      "grad_norm": 0.17369544506072998,
+      "grad_norm": 0.17404448986053467,
       "learning_rate": 0.00014539904997395468,
-      "loss": 0.3535,
+      "loss": 0.3542,
       "step": 350
     },
     {
       "epoch": 0.037584323803405074,
-      "grad_norm": 0.17372484505176544,
+      "grad_norm": 0.17870911955833435,
       "learning_rate": 0.00014511890844418453,
-      "loss": 0.2509,
+      "loss": 0.253,
       "step": 351
     },
     {
       "epoch": 0.03769140164899882,
-      "grad_norm": 0.1707952469587326,
+      "grad_norm": 0.1721985787153244,
       "learning_rate": 0.00014483832160900326,
-      "loss": 0.4087,
+      "loss": 0.4102,
       "step": 352
     },
     {
       "epoch": 0.03779847949459257,
-      "grad_norm": 0.18640413880348206,
+      "grad_norm": 0.1740548461675644,
       "learning_rate": 0.00014455729223768966,
-      "loss": 0.3737,
+      "loss": 0.3727,
       "step": 353
     },
     {
       "epoch": 0.037905557340186316,
-      "grad_norm": 0.1838582456111908,
+      "grad_norm": 0.18063552677631378,
       "learning_rate": 0.0001442758231038902,
-      "loss": 0.4719,
+      "loss": 0.4705,
       "step": 354
     },
     {
       "epoch": 0.03801263518578006,
-      "grad_norm": 0.16711768507957458,
+      "grad_norm": 0.16232942044734955,
       "learning_rate": 0.00014399391698559152,
-      "loss": 0.303,
+      "loss": 0.3015,
       "step": 355
     },
     {
       "epoch": 0.038119713031373806,
-      "grad_norm": 0.209504172205925,
+      "grad_norm": 0.2140200287103653,
       "learning_rate": 0.0001437115766650933,
-      "loss": 0.3234,
+      "loss": 0.3251,
       "step": 356
     },
     {
       "epoch": 0.03822679087696756,
-      "grad_norm": 0.19139063358306885,
+      "grad_norm": 0.19321036338806152,
       "learning_rate": 0.00014342880492898048,
-      "loss": 0.2605,
+      "loss": 0.2617,
       "step": 357
     },
     {
       "epoch": 0.0383338687225613,
-      "grad_norm": 0.14143365621566772,
+      "grad_norm": 0.14323490858078003,
       "learning_rate": 0.0001431456045680959,
-      "loss": 0.3509,
+      "loss": 0.3512,
       "step": 358
     },
     {
       "epoch": 0.03844094656815505,
-      "grad_norm": 0.14169888198375702,
+      "grad_norm": 0.14133870601654053,
       "learning_rate": 0.00014286197837751286,
-      "loss": 0.2654,
+      "loss": 0.2648,
       "step": 359
     },
     {
       "epoch": 0.03854802441374879,
-      "grad_norm": 0.14783132076263428,
+      "grad_norm": 0.1461154669523239,
       "learning_rate": 0.00014257792915650728,
-      "loss": 0.2976,
+      "loss": 0.2978,
       "step": 360
     },
     {
       "epoch": 0.038655102259342544,
-      "grad_norm": 0.14012813568115234,
+      "grad_norm": 0.13456670939922333,
       "learning_rate": 0.00014229345970853032,
-      "loss": 0.2406,
+      "loss": 0.2399,
       "step": 361
     },
     {
       "epoch": 0.03876218010493629,
-      "grad_norm": 0.20082809031009674,
+      "grad_norm": 0.19285790622234344,
       "learning_rate": 0.00014200857284118066,
-      "loss": 0.3533,
+      "loss": 0.3538,
       "step": 362
     },
     {
       "epoch": 0.038869257950530034,
-      "grad_norm": 0.1798025518655777,
+      "grad_norm": 0.17173908650875092,
       "learning_rate": 0.00014172327136617656,
-      "loss": 0.3245,
+      "loss": 0.325,
       "step": 363
     },
     {
       "epoch": 0.03897633579612378,
-      "grad_norm": 0.16510576009750366,
+      "grad_norm": 0.16437563300132751,
       "learning_rate": 0.00014143755809932845,
-      "loss": 0.2857,
+      "loss": 0.2856,
       "step": 364
     },
     {
       "epoch": 0.03908341364171753,
-      "grad_norm": 0.16456517577171326,
+      "grad_norm": 0.16356609761714935,
       "learning_rate": 0.00014115143586051088,
-      "loss": 0.3673,
+      "loss": 0.3665,
       "step": 365
     },
     {
       "epoch": 0.039190491487311276,
-      "grad_norm": 0.1842593401670456,
+      "grad_norm": 0.17940573394298553,
       "learning_rate": 0.00014086490747363493,
-      "loss": 0.3667,
+      "loss": 0.3673,
       "step": 366
     },
     {
       "epoch": 0.03929756933290502,
-      "grad_norm": 0.18094277381896973,
+      "grad_norm": 0.17950116097927094,
       "learning_rate": 0.00014057797576662,
-      "loss": 0.394,
+      "loss": 0.3944,
       "step": 367
     },
     {
       "epoch": 0.039404647178498765,
-      "grad_norm": 0.14859478175640106,
+      "grad_norm": 0.1494416892528534,
       "learning_rate": 0.00014029064357136628,
-      "loss": 0.2738,
+      "loss": 0.2745,
       "step": 368
     },
     {
       "epoch": 0.03951172502409252,
-      "grad_norm": 0.16508136689662933,
+      "grad_norm": 0.16802102327346802,
       "learning_rate": 0.00014000291372372647,
-      "loss": 0.3326,
+      "loss": 0.3329,
       "step": 369
     },
     {
       "epoch": 0.03961880286968626,
-      "grad_norm": 0.16775007545948029,
+      "grad_norm": 0.16510355472564697,
       "learning_rate": 0.00013971478906347806,
-      "loss": 0.4139,
+      "loss": 0.415,
       "step": 370
     },
     {
       "epoch": 0.03972588071528001,
-      "grad_norm": 0.1426904797554016,
+      "grad_norm": 0.14013710618019104,
       "learning_rate": 0.00013942627243429512,
-      "loss": 0.2177,
+      "loss": 0.2186,
       "step": 371
     },
     {
       "epoch": 0.03983295856087375,
-      "grad_norm": 0.20198000967502594,
+      "grad_norm": 0.20402979850769043,
       "learning_rate": 0.00013913736668372026,
-      "loss": 0.453,
+      "loss": 0.4516,
       "step": 372
     },
     {
       "epoch": 0.039940036406467504,
-      "grad_norm": 0.1864636093378067,
+      "grad_norm": 0.1825643926858902,
       "learning_rate": 0.00013884807466313663,
-      "loss": 0.3124,
+      "loss": 0.314,
       "step": 373
     },
     {
       "epoch": 0.04004711425206125,
-      "grad_norm": 0.13631337881088257,
+      "grad_norm": 0.13864319026470184,
       "learning_rate": 0.00013855839922773968,
-      "loss": 0.262,
+      "loss": 0.2619,
       "step": 374
     },
     {
       "epoch": 0.040154192097654993,
-      "grad_norm": 0.17490577697753906,
+      "grad_norm": 0.16547183692455292,
       "learning_rate": 0.000138268343236509,
-      "loss": 0.3139,
+      "loss": 0.3125,
       "step": 375
     },
     {
       "epoch": 0.04026126994324874,
-      "grad_norm": 0.1761641502380371,
+      "grad_norm": 0.17819920182228088,
       "learning_rate": 0.00013797790955218014,
-      "loss": 0.4065,
+      "loss": 0.4062,
       "step": 376
     },
     {
       "epoch": 0.04036834778884249,
-      "grad_norm": 0.156477689743042,
+      "grad_norm": 0.154698446393013,
       "learning_rate": 0.00013768710104121627,
-      "loss": 0.2875,
+      "loss": 0.2868,
       "step": 377
     },
     {
       "epoch": 0.040475425634436235,
-      "grad_norm": 0.14239205420017242,
+      "grad_norm": 0.14120882749557495,
       "learning_rate": 0.00013739592057378003,
-      "loss": 0.3407,
+      "loss": 0.3406,
       "step": 378
     },
     {
       "epoch": 0.04058250348002998,
-      "grad_norm": 0.18152830004692078,
+      "grad_norm": 0.17801566421985626,
       "learning_rate": 0.0001371043710237051,
-      "loss": 0.3821,
+      "loss": 0.3826,
       "step": 379
     },
     {
       "epoch": 0.04068958132562373,
-      "grad_norm": 0.14406077563762665,
+      "grad_norm": 0.1434149593114853,
       "learning_rate": 0.00013681245526846783,
-      "loss": 0.286,
+      "loss": 0.2858,
       "step": 380
     },
     {
       "epoch": 0.04079665917121748,
-      "grad_norm": 0.1428261250257492,
+      "grad_norm": 0.14281722903251648,
       "learning_rate": 0.0001365201761891588,
-      "loss": 0.2463,
+      "loss": 0.2466,
       "step": 381
     },
     {
       "epoch": 0.04090373701681122,
-      "grad_norm": 0.16200338304042816,
+      "grad_norm": 0.16065680980682373,
       "learning_rate": 0.00013622753667045457,
-      "loss": 0.3641,
+      "loss": 0.3634,
       "step": 382
     },
     {
       "epoch": 0.041010814862404966,
-      "grad_norm": 0.19177362322807312,
+      "grad_norm": 0.19151967763900757,
       "learning_rate": 0.00013593453960058908,
-      "loss": 0.3988,
+      "loss": 0.3986,
       "step": 383
     },
     {
       "epoch": 0.04111789270799872,
-      "grad_norm": 0.13588127493858337,
+      "grad_norm": 0.13544686138629913,
       "learning_rate": 0.00013564118787132506,
       "loss": 0.3097,
       "step": 384
     },
     {
       "epoch": 0.04122497055359246,
-      "grad_norm": 0.14050887525081635,
+      "grad_norm": 0.13807305693626404,
       "learning_rate": 0.00013534748437792573,
-      "loss": 0.2883,
+      "loss": 0.2874,
       "step": 385
     },
     {
       "epoch": 0.04133204839918621,
-      "grad_norm": 0.1991889923810959,
+      "grad_norm": 0.1964596062898636,
       "learning_rate": 0.0001350534320191259,
-      "loss": 0.3118,
+      "loss": 0.3109,
       "step": 386
     },
     {
       "epoch": 0.04143912624477995,
-      "grad_norm": 0.1473557949066162,
+      "grad_norm": 0.14541715383529663,
       "learning_rate": 0.0001347590336971037,
-      "loss": 0.363,
+      "loss": 0.3624,
       "step": 387
     },
     {
       "epoch": 0.041546204090373705,
-      "grad_norm": 0.20648524165153503,
+      "grad_norm": 0.20359724760055542,
       "learning_rate": 0.0001344642923174517,
-      "loss": 0.3746,
+      "loss": 0.3734,
       "step": 388
     },
     {
       "epoch": 0.04165328193596745,
-      "grad_norm": 0.12530086934566498,
+      "grad_norm": 0.12174426764249802,
       "learning_rate": 0.00013416921078914835,
-      "loss": 0.2218,
+      "loss": 0.2217,
       "step": 389
     },
     {
       "epoch": 0.041760359781561195,
-      "grad_norm": 0.1354837268590927,
+      "grad_norm": 0.13207176327705383,
       "learning_rate": 0.00013387379202452917,
-      "loss": 0.2357,
+      "loss": 0.2343,
       "step": 390
     },
     {
       "epoch": 0.04186743762715494,
-      "grad_norm": 0.13786575198173523,
+      "grad_norm": 0.1388169378042221,
       "learning_rate": 0.00013357803893925807,
-      "loss": 0.2601,
+      "loss": 0.2587,
       "step": 391
     },
     {
       "epoch": 0.04197451547274869,
-      "grad_norm": 0.14999490976333618,
+      "grad_norm": 0.1489415019750595,
       "learning_rate": 0.00013328195445229868,
-      "loss": 0.2587,
+      "loss": 0.2599,
       "step": 392
     },
     {
       "epoch": 0.042081593318342436,
-      "grad_norm": 0.20544356107711792,
+      "grad_norm": 0.20052435994148254,
       "learning_rate": 0.00013298554148588528,
-      "loss": 0.3826,
+      "loss": 0.3817,
       "step": 393
     },
     {
       "epoch": 0.04218867116393618,
-      "grad_norm": 0.18927697837352753,
+      "grad_norm": 0.19129063189029694,
       "learning_rate": 0.00013268880296549425,
-      "loss": 0.3572,
+      "loss": 0.3565,
       "step": 394
     },
     {
       "epoch": 0.042295749009529926,
-      "grad_norm": 0.18653343617916107,
+      "grad_norm": 0.18302471935749054,
       "learning_rate": 0.00013239174181981495,
-      "loss": 0.3651,
+      "loss": 0.3647,
       "step": 395
     },
     {
       "epoch": 0.04240282685512368,
-      "grad_norm": 0.2660796642303467,
+      "grad_norm": 0.23317281901836395,
       "learning_rate": 0.00013209436098072095,
-      "loss": 0.4001,
+      "loss": 0.3992,
       "step": 396
     },
     {
       "epoch": 0.04250990470071742,
-      "grad_norm": 0.1991802304983139,
+      "grad_norm": 0.20024320483207703,
       "learning_rate": 0.00013179666338324108,
-      "loss": 0.399,
+      "loss": 0.3974,
       "step": 397
     },
     {
       "epoch": 0.04261698254631117,
-      "grad_norm": 0.20211124420166016,
+      "grad_norm": 0.19934377074241638,
       "learning_rate": 0.0001314986519655305,
-      "loss": 0.452,
+      "loss": 0.449,
       "step": 398
     },
     {
       "epoch": 0.04272406039190491,
-      "grad_norm": 0.19349808990955353,
+      "grad_norm": 0.1918846070766449,
       "learning_rate": 0.0001312003296688415,
-      "loss": 0.5259,
+      "loss": 0.5255,
       "step": 399
     },
     {
       "epoch": 0.042831138237498664,
-      "grad_norm": 0.1513546109199524,
+      "grad_norm": 0.14945387840270996,
       "learning_rate": 0.00013090169943749476,
-      "loss": 0.2608,
+      "loss": 0.2609,
       "step": 400
     },
     {
       "epoch": 0.04293821608309241,
-      "grad_norm": 0.15879368782043457,
+      "grad_norm": 0.16165776550769806,
       "learning_rate": 0.0001306027642188501,
-      "loss": 0.3402,
+      "loss": 0.3412,
       "step": 401
     },
     {
       "epoch": 0.043045293928686154,
-      "grad_norm": 0.15345223248004913,
+      "grad_norm": 0.15505026280879974,
       "learning_rate": 0.00013030352696327742,
-      "loss": 0.311,
+      "loss": 0.3109,
       "step": 402
     },
     {
       "epoch": 0.0431523717742799,
-      "grad_norm": 0.17926479876041412,
+      "grad_norm": 0.17844939231872559,
       "learning_rate": 0.00013000399062412763,
-      "loss": 0.3439,
+      "loss": 0.3441,
       "step": 403
     },
     {
       "epoch": 0.04325944961987365,
-      "grad_norm": 0.1882927417755127,
+      "grad_norm": 0.19538049399852753,
       "learning_rate": 0.0001297041581577035,
-      "loss": 0.3147,
+      "loss": 0.3154,
       "step": 404
     },
     {
       "epoch": 0.043366527465467396,
-      "grad_norm": 0.12865504622459412,
+      "grad_norm": 0.12841279804706573,
       "learning_rate": 0.0001294040325232304,
-      "loss": 0.3,
+      "loss": 0.2992,
       "step": 405
     },
     {
       "epoch": 0.04347360531106114,
-      "grad_norm": 0.15574470162391663,
+      "grad_norm": 0.15678054094314575,
       "learning_rate": 0.00012910361668282719,
-      "loss": 0.3666,
+      "loss": 0.3667,
       "step": 406
     },
     {
       "epoch": 0.043580683156654886,
-      "grad_norm": 0.19106611609458923,
+      "grad_norm": 0.19353367388248444,
       "learning_rate": 0.00012880291360147693,
-      "loss": 0.3568,
+      "loss": 0.3586,
       "step": 407
     },
     {
       "epoch": 0.04368776100224864,
-      "grad_norm": 0.18762201070785522,
+      "grad_norm": 0.18881572782993317,
       "learning_rate": 0.0001285019262469976,
-      "loss": 0.3123,
+      "loss": 0.3125,
       "step": 408
     },
     {
       "epoch": 0.04379483884784238,
-      "grad_norm": 0.14002811908721924,
+      "grad_norm": 0.14315952360630035,
       "learning_rate": 0.00012820065759001293,
-      "loss": 0.2737,
+      "loss": 0.2743,
       "step": 409
     },
     {
       "epoch": 0.04390191669343613,
-      "grad_norm": 0.19170932471752167,
+      "grad_norm": 0.1901596039533615,
       "learning_rate": 0.00012789911060392294,
-      "loss": 0.4061,
+      "loss": 0.405,
       "step": 410
     },
     {
       "epoch": 0.04400899453902987,
-      "grad_norm": 0.17306004464626312,
+      "grad_norm": 0.17396406829357147,
       "learning_rate": 0.0001275972882648746,
-      "loss": 0.3492,
+      "loss": 0.3503,
       "step": 411
     },
     {
       "epoch": 0.044116072384623624,
-      "grad_norm": 0.1948164999485016,
+      "grad_norm": 0.19500721991062164,
       "learning_rate": 0.00012729519355173254,
-      "loss": 0.4394,
+      "loss": 0.4379,
       "step": 412
     },
     {
       "epoch": 0.04422315023021737,
-      "grad_norm": 0.15455390512943268,
+      "grad_norm": 0.15421026945114136,
       "learning_rate": 0.00012699282944604967,
-      "loss": 0.3778,
+      "loss": 0.3766,
       "step": 413
     },
     {
       "epoch": 0.044330228075811114,
-      "grad_norm": 0.1604427546262741,
+      "grad_norm": 0.16204634308815002,
       "learning_rate": 0.00012669019893203759,
-      "loss": 0.3684,
+      "loss": 0.3692,
       "step": 414
     },
     {
       "epoch": 0.04443730592140486,
-      "grad_norm": 0.16993975639343262,
+      "grad_norm": 0.16477806866168976,
       "learning_rate": 0.0001263873049965373,
-      "loss": 0.2551,
+      "loss": 0.2539,
       "step": 415
     },
     {
       "epoch": 0.04454438376699861,
-      "grad_norm": 0.15334467589855194,
+      "grad_norm": 0.15045298635959625,
       "learning_rate": 0.00012608415062898972,
-      "loss": 0.3069,
+      "loss": 0.3063,
       "step": 416
     },
     {
       "epoch": 0.044651461612592355,
-      "grad_norm": 0.1586979478597641,
+      "grad_norm": 0.1610919088125229,
       "learning_rate": 0.000125780738821406,
-      "loss": 0.3423,
+      "loss": 0.3427,
       "step": 417
     },
     {
       "epoch": 0.0447585394581861,
-      "grad_norm": 0.1945161521434784,
+      "grad_norm": 0.19420892000198364,
       "learning_rate": 0.00012547707256833823,
-      "loss": 0.3957,
+      "loss": 0.3954,
       "step": 418
     },
     {
       "epoch": 0.044865617303779845,
-      "grad_norm": 0.18735453486442566,
+      "grad_norm": 0.18012918531894684,
       "learning_rate": 0.00012517315486684972,
-      "loss": 0.3792,
+      "loss": 0.3788,
       "step": 419
     },
     {
       "epoch": 0.0449726951493736,
-      "grad_norm": 0.16127122938632965,
+      "grad_norm": 0.162590891122818,
       "learning_rate": 0.0001248689887164855,
-      "loss": 0.3564,
+      "loss": 0.357,
       "step": 420
     },
     {
       "epoch": 0.04507977299496734,
-      "grad_norm": 0.1495504528284073,
+      "grad_norm": 0.15058323740959167,
       "learning_rate": 0.00012456457711924266,
-      "loss": 0.2447,
+      "loss": 0.245,
       "step": 421
     },
     {
       "epoch": 0.04518685084056109,
-      "grad_norm": 0.12814819812774658,
+      "grad_norm": 0.12929019331932068,
       "learning_rate": 0.00012425992307954075,
-      "loss": 0.2419,
+      "loss": 0.2408,
       "step": 422
     },
     {
       "epoch": 0.04529392868615483,
-      "grad_norm": 0.13747601211071014,
+      "grad_norm": 0.1378922462463379,
       "learning_rate": 0.0001239550296041922,
-      "loss": 0.2881,
+      "loss": 0.2867,
       "step": 423
     },
     {
       "epoch": 0.04540100653174858,
-      "grad_norm": 0.14220485091209412,
+      "grad_norm": 0.14614133536815643,
       "learning_rate": 0.00012364989970237248,
-      "loss": 0.2969,
+      "loss": 0.2963,
       "step": 424
     },
     {
       "epoch": 0.04550808437734233,
-      "grad_norm": 0.15737488865852356,
+      "grad_norm": 0.15353389084339142,
       "learning_rate": 0.00012334453638559057,
-      "loss": 0.3217,
+      "loss": 0.3199,
       "step": 425
     },
     {
       "epoch": 0.04561516222293607,
-      "grad_norm": 0.13472171127796173,
+      "grad_norm": 0.1354091763496399,
       "learning_rate": 0.00012303894266765908,
-      "loss": 0.2394,
+      "loss": 0.2396,
       "step": 426
     },
     {
       "epoch": 0.04572224006852982,
-      "grad_norm": 0.16132071614265442,
+      "grad_norm": 0.1608564555644989,
       "learning_rate": 0.00012273312156466464,
-      "loss": 0.2786,
+      "loss": 0.2776,
       "step": 427
     },
     {
       "epoch": 0.04582931791412357,
-      "grad_norm": 0.17469573020935059,
+      "grad_norm": 0.17579632997512817,
       "learning_rate": 0.00012242707609493814,
-      "loss": 0.2979,
+      "loss": 0.3,
       "step": 428
     },
     {
       "epoch": 0.045936395759717315,
-      "grad_norm": 0.1818244755268097,
+      "grad_norm": 0.1844303011894226,
       "learning_rate": 0.00012212080927902474,
-      "loss": 0.3639,
+      "loss": 0.3654,
       "step": 429
     },
     {
       "epoch": 0.04604347360531106,
-      "grad_norm": 0.1542896330356598,
+      "grad_norm": 0.15440614521503448,
       "learning_rate": 0.00012181432413965428,
-      "loss": 0.2892,
+      "loss": 0.2888,
       "step": 430
     },
     {
       "epoch": 0.046150551450904805,
-      "grad_norm": 0.17613442242145538,
+      "grad_norm": 0.1817633956670761,
       "learning_rate": 0.00012150762370171136,
-      "loss": 0.424,
+      "loss": 0.4253,
       "step": 431
     },
     {
       "epoch": 0.046257629296498556,
-      "grad_norm": 0.17226651310920715,
+      "grad_norm": 0.17288000881671906,
       "learning_rate": 0.00012120071099220549,
-      "loss": 0.3764,
+      "loss": 0.3776,
       "step": 432
     },
     {
       "epoch": 0.0463647071420923,
-      "grad_norm": 0.16670876741409302,
+      "grad_norm": 0.18507465720176697,
       "learning_rate": 0.00012089358904024117,
-      "loss": 0.3949,
+      "loss": 0.3965,
       "step": 433
     },
     {
       "epoch": 0.046471784987686046,
-      "grad_norm": 0.137644961476326,
+      "grad_norm": 0.1360858529806137,
       "learning_rate": 0.00012058626087698814,
-      "loss": 0.203,
+      "loss": 0.2018,
       "step": 434
     },
     {
       "epoch": 0.04657886283327979,
-      "grad_norm": 0.18664462864398956,
+      "grad_norm": 0.18026049435138702,
       "learning_rate": 0.00012027872953565125,
-      "loss": 0.3211,
+      "loss": 0.3193,
       "step": 435
     },
     {
       "epoch": 0.04668594067887354,
-      "grad_norm": 0.2129397839307785,
+      "grad_norm": 0.2125885784626007,
       "learning_rate": 0.00011997099805144069,
-      "loss": 0.4424,
+      "loss": 0.4414,
       "step": 436
     },
     {
       "epoch": 0.04679301852446729,
-      "grad_norm": 0.1846884787082672,
+      "grad_norm": 0.18527942895889282,
       "learning_rate": 0.000119663069461542,
-      "loss": 0.2833,
+      "loss": 0.2822,
       "step": 437
     },
     {
       "epoch": 0.04690009637006103,
-      "grad_norm": 0.1526833325624466,
+      "grad_norm": 0.15211868286132812,
       "learning_rate": 0.00011935494680508606,
-      "loss": 0.3097,
+      "loss": 0.3093,
       "step": 438
     },
     {
       "epoch": 0.047007174215654784,
-      "grad_norm": 0.20358015596866608,
+      "grad_norm": 0.20865000784397125,
       "learning_rate": 0.00011904663312311901,
-      "loss": 0.3995,
+      "loss": 0.3999,
       "step": 439
     },
     {
       "epoch": 0.04711425206124853,
-      "grad_norm": 0.21734708547592163,
+      "grad_norm": 0.21300359070301056,
       "learning_rate": 0.00011873813145857249,
-      "loss": 0.3412,
+      "loss": 0.3397,
       "step": 440
     },
     {
       "epoch": 0.047221329906842274,
-      "grad_norm": 0.18342752754688263,
+      "grad_norm": 0.18172712624073029,
       "learning_rate": 0.00011842944485623335,
-      "loss": 0.336,
+      "loss": 0.3376,
       "step": 441
     },
     {
       "epoch": 0.04732840775243602,
-      "grad_norm": 0.16440509259700775,
+      "grad_norm": 0.16084377467632294,
       "learning_rate": 0.00011812057636271374,
-      "loss": 0.2243,
+      "loss": 0.2233,
       "step": 442
     },
     {
       "epoch": 0.04743548559802977,
-      "grad_norm": 0.16069766879081726,
+      "grad_norm": 0.1644655168056488,
       "learning_rate": 0.000117811529026421,
-      "loss": 0.3723,
+      "loss": 0.3709,
       "step": 443
     },
     {
       "epoch": 0.047542563443623516,
-      "grad_norm": 0.16657786071300507,
+      "grad_norm": 0.1645745038986206,
       "learning_rate": 0.00011750230589752762,
-      "loss": 0.3583,
+      "loss": 0.3592,
       "step": 444
     },
     {
       "epoch": 0.04764964128921726,
-      "grad_norm": 0.18991921842098236,
+      "grad_norm": 0.19533959031105042,
       "learning_rate": 0.00011719291002794096,
-      "loss": 0.3201,
+      "loss": 0.3205,
       "step": 445
     },
     {
       "epoch": 0.047756719134811006,
-      "grad_norm": 0.14543665945529938,
+      "grad_norm": 0.14579270780086517,
       "learning_rate": 0.00011688334447127338,
       "loss": 0.3762,
       "step": 446
     },
     {
       "epoch": 0.04786379698040476,
-      "grad_norm": 0.1752735674381256,
+      "grad_norm": 0.175794318318367,
       "learning_rate": 0.00011657361228281199,
-      "loss": 0.3603,
+      "loss": 0.361,
       "step": 447
     },
     {
       "epoch": 0.0479708748259985,
-      "grad_norm": 0.15264953672885895,
+      "grad_norm": 0.1505967080593109,
       "learning_rate": 0.00011626371651948838,
-      "loss": 0.2599,
+      "loss": 0.2582,
       "step": 448
     },
     {
       "epoch": 0.04807795267159225,
-      "grad_norm": 0.10569647699594498,
+      "grad_norm": 0.10504525899887085,
       "learning_rate": 0.00011595366023984864,
       "loss": 0.1823,
       "step": 449
     },
     {
       "epoch": 0.04818503051718599,
-      "grad_norm": 0.19080232083797455,
+      "grad_norm": 0.19151091575622559,
       "learning_rate": 0.0001156434465040231,
-      "loss": 0.3381,
+      "loss": 0.3385,
       "step": 450
     },
     {
       "epoch": 0.048292108362779744,
-      "grad_norm": 0.1927441954612732,
+      "grad_norm": 0.19403024017810822,
       "learning_rate": 0.00011533307837369607,
-      "loss": 0.3542,
+      "loss": 0.3538,
       "step": 451
     },
     {
       "epoch": 0.04839918620837349,
-      "grad_norm": 0.18672965466976166,
+      "grad_norm": 0.185320183634758,
       "learning_rate": 0.00011502255891207572,
-      "loss": 0.3396,
+      "loss": 0.3393,
       "step": 452
     },
     {
       "epoch": 0.048506264053967234,
-      "grad_norm": 0.20121031999588013,
+      "grad_norm": 0.20061573386192322,
       "learning_rate": 0.00011471189118386375,
-      "loss": 0.3582,
+      "loss": 0.3567,
       "step": 453
     },
     {
       "epoch": 0.04861334189956098,
-      "grad_norm": 0.14637677371501923,
+      "grad_norm": 0.15041832625865936,
       "learning_rate": 0.00011440107825522521,
-      "loss": 0.2784,
+      "loss": 0.2794,
       "step": 454
     },
     {
       "epoch": 0.04872041974515473,
-      "grad_norm": 0.16313035786151886,
+      "grad_norm": 0.16546788811683655,
       "learning_rate": 0.00011409012319375827,
-      "loss": 0.4364,
+      "loss": 0.4382,
       "step": 455
     },
     {
       "epoch": 0.048827497590748475,
-      "grad_norm": 0.17453531920909882,
+      "grad_norm": 0.17634829878807068,
       "learning_rate": 0.0001137790290684638,
-      "loss": 0.4113,
+      "loss": 0.4121,
       "step": 456
     },
     {
       "epoch": 0.04893457543634222,
-      "grad_norm": 0.19787636399269104,
+      "grad_norm": 0.2208283543586731,
       "learning_rate": 0.00011346779894971527,
-      "loss": 0.4537,
+      "loss": 0.454,
       "step": 457
     },
     {
       "epoch": 0.049041653281935965,
-      "grad_norm": 0.16840681433677673,
+      "grad_norm": 0.16910183429718018,
       "learning_rate": 0.00011315643590922827,
-      "loss": 0.3678,
+      "loss": 0.3679,
       "step": 458
     },
     {
       "epoch": 0.04914873112752972,
-      "grad_norm": 0.1894865781068802,
+      "grad_norm": 0.18956156075000763,
       "learning_rate": 0.0001128449430200303,
-      "loss": 0.4161,
+      "loss": 0.4173,
       "step": 459
     },
     {
       "epoch": 0.04925580897312346,
-      "grad_norm": 0.180477574467659,
+      "grad_norm": 0.1655164510011673,
       "learning_rate": 0.00011253332335643043,
-      "loss": 0.3242,
+      "loss": 0.3229,
       "step": 460
     },
     {
       "epoch": 0.04936288681871721,
-      "grad_norm": 0.17623566091060638,
+      "grad_norm": 0.1741032898426056,
       "learning_rate": 0.00011222157999398895,
-      "loss": 0.2212,
+      "loss": 0.2208,
       "step": 461
     },
     {
       "epoch": 0.04946996466431095,
-      "grad_norm": 0.14930705726146698,
+      "grad_norm": 0.14394614100456238,
       "learning_rate": 0.00011190971600948699,
-      "loss": 0.3156,
+      "loss": 0.3147,
       "step": 462
     },
     {
       "epoch": 0.049577042509904704,
-      "grad_norm": 0.19468611478805542,
+      "grad_norm": 0.19279611110687256,
       "learning_rate": 0.00011159773448089614,
-      "loss": 0.4011,
+      "loss": 0.4006,
       "step": 463
     },
     {
       "epoch": 0.04968412035549845,
-      "grad_norm": 0.17452605068683624,
+      "grad_norm": 0.17145444452762604,
       "learning_rate": 0.00011128563848734816,
       "loss": 0.3711,
       "step": 464
     },
     {
       "epoch": 0.04979119820109219,
-      "grad_norm": 0.15289977192878723,
+      "grad_norm": 0.15201549232006073,
       "learning_rate": 0.00011097343110910452,
-      "loss": 0.3943,
+      "loss": 0.3944,
       "step": 465
     },
     {
       "epoch": 0.04989827604668594,
-      "grad_norm": 0.19616980850696564,
+      "grad_norm": 0.19721528887748718,
       "learning_rate": 0.000110661115427526,
-      "loss": 0.3831,
+      "loss": 0.3839,
       "step": 466
     },
     {
       "epoch": 0.05000535389227969,
-      "grad_norm": 0.15868891775608063,
+      "grad_norm": 0.16234098374843597,
       "learning_rate": 0.00011034869452504226,
-      "loss": 0.3308,
+      "loss": 0.3303,
       "step": 467
     },
     {
       "epoch": 0.050112431737873435,
-      "grad_norm": 0.14652469754219055,
+      "grad_norm": 0.14313015341758728,
       "learning_rate": 0.00011003617148512149,
-      "loss": 0.2859,
+      "loss": 0.2862,
       "step": 468
     },
     {
       "epoch": 0.05021950958346718,
-      "grad_norm": 0.15360702574253082,
+      "grad_norm": 0.15435555577278137,
       "learning_rate": 0.00010972354939223996,
-      "loss": 0.2889,
+      "loss": 0.2875,
       "step": 469
     },
     {
       "epoch": 0.050326587429060925,
-      "grad_norm": 0.19247639179229736,
+      "grad_norm": 0.1893438845872879,
       "learning_rate": 0.00010941083133185146,
-      "loss": 0.3052,
+      "loss": 0.303,
       "step": 470
     },
     {
       "epoch": 0.050433665274654677,
-      "grad_norm": 0.17776168882846832,
+      "grad_norm": 0.17701826989650726,
       "learning_rate": 0.00010909802039035701,
-      "loss": 0.393,
+      "loss": 0.3937,
       "step": 471
     },
     {
       "epoch": 0.05054074312024842,
-      "grad_norm": 0.15174326300621033,
+      "grad_norm": 0.15118297934532166,
       "learning_rate": 0.00010878511965507434,
-      "loss": 0.2693,
+      "loss": 0.2687,
       "step": 472
     },
     {
       "epoch": 0.050647820965842166,
-      "grad_norm": 0.17235317826271057,
+      "grad_norm": 0.16342921555042267,
       "learning_rate": 0.00010847213221420736,
-      "loss": 0.2949,
+      "loss": 0.2945,
       "step": 473
     },
     {
       "epoch": 0.05075489881143591,
-      "grad_norm": 0.16213464736938477,
+      "grad_norm": 0.16562896966934204,
       "learning_rate": 0.00010815906115681578,
-      "loss": 0.3521,
+      "loss": 0.3528,
       "step": 474
     },
     {
       "epoch": 0.05086197665702966,
-      "grad_norm": 0.18122194707393646,
+      "grad_norm": 0.17907415330410004,
       "learning_rate": 0.0001078459095727845,
-      "loss": 0.2741,
+      "loss": 0.2738,
       "step": 475
     },
     {
       "epoch": 0.05096905450262341,
-      "grad_norm": 0.2096056491136551,
+      "grad_norm": 0.20665884017944336,
       "learning_rate": 0.00010753268055279329,
-      "loss": 0.3368,
+      "loss": 0.3369,
       "step": 476
     },
     {
       "epoch": 0.05107613234821715,
-      "grad_norm": 0.148423969745636,
+      "grad_norm": 0.14618425071239471,
       "learning_rate": 0.0001072193771882861,
-      "loss": 0.3058,
+      "loss": 0.3057,
       "step": 477
     },
     {
       "epoch": 0.0511832101938109,
-      "grad_norm": 0.1586843580007553,
+      "grad_norm": 0.15848585963249207,
       "learning_rate": 0.00010690600257144061,
-      "loss": 0.3129,
+      "loss": 0.3142,
       "step": 478
     },
     {
       "epoch": 0.05129028803940465,
-      "grad_norm": 0.15282008051872253,
+      "grad_norm": 0.1518888771533966,
       "learning_rate": 0.0001065925597951378,
-      "loss": 0.2693,
+      "loss": 0.2689,
       "step": 479
     },
     {
       "epoch": 0.051397365884998394,
-      "grad_norm": 0.16839781403541565,
+      "grad_norm": 0.1687459647655487,
       "learning_rate": 0.00010627905195293135,
-      "loss": 0.2877,
+      "loss": 0.2871,
       "step": 480
     },
     {
       "epoch": 0.05150444373059214,
-      "grad_norm": 0.16241709887981415,
+      "grad_norm": 0.16327998042106628,
       "learning_rate": 0.00010596548213901708,
-      "loss": 0.2732,
+      "loss": 0.2737,
       "step": 481
     },
     {
       "epoch": 0.051611521576185884,
-      "grad_norm": 0.17643004655838013,
+      "grad_norm": 0.17769210040569305,
       "learning_rate": 0.00010565185344820247,
-      "loss": 0.362,
+      "loss": 0.3618,
       "step": 482
     },
     {
       "epoch": 0.051718599421779636,
-      "grad_norm": 0.17319577932357788,
+      "grad_norm": 0.17134082317352295,
       "learning_rate": 0.00010533816897587606,
-      "loss": 0.3744,
+      "loss": 0.3736,
       "step": 483
     },
     {
       "epoch": 0.05182567726737338,
-      "grad_norm": 0.1710813194513321,
+      "grad_norm": 0.17243537306785583,
       "learning_rate": 0.00010502443181797697,
-      "loss": 0.3154,
+      "loss": 0.3158,
       "step": 484
     },
     {
       "epoch": 0.051932755112967126,
-      "grad_norm": 0.17228226363658905,
+      "grad_norm": 0.172983318567276,
       "learning_rate": 0.00010471064507096426,
-      "loss": 0.3267,
+      "loss": 0.3272,
       "step": 485
     },
     {
       "epoch": 0.05203983295856087,
-      "grad_norm": 0.16815848648548126,
+      "grad_norm": 0.16794148087501526,
       "learning_rate": 0.0001043968118317865,
-      "loss": 0.4157,
+      "loss": 0.4177,
       "step": 486
     },
     {
       "epoch": 0.05214691080415462,
-      "grad_norm": 0.14662803709506989,
+      "grad_norm": 0.14923620223999023,
       "learning_rate": 0.00010408293519785101,
       "loss": 0.2119,
       "step": 487
     },
     {
       "epoch": 0.05225398864974837,
-      "grad_norm": 0.14955608546733856,
+      "grad_norm": 0.148018017411232,
       "learning_rate": 0.00010376901826699348,
-      "loss": 0.3402,
+      "loss": 0.3396,
       "step": 488
     },
     {
       "epoch": 0.05236106649534211,
-      "grad_norm": 0.20436280965805054,
+      "grad_norm": 0.21457895636558533,
       "learning_rate": 0.00010345506413744726,
-      "loss": 0.2788,
+      "loss": 0.2789,
       "step": 489
     },
     {
       "epoch": 0.05246814434093586,
-      "grad_norm": 0.1706622689962387,
+      "grad_norm": 0.16679152846336365,
       "learning_rate": 0.00010314107590781284,
-      "loss": 0.3243,
+      "loss": 0.3228,
       "step": 490
     },
     {
       "epoch": 0.05257522218652961,
-      "grad_norm": 0.17119503021240234,
+      "grad_norm": 0.17177151143550873,
       "learning_rate": 0.00010282705667702734,
-      "loss": 0.2811,
+      "loss": 0.2813,
       "step": 491
     },
     {
       "epoch": 0.052682300032123354,
-      "grad_norm": 0.145518958568573,
+      "grad_norm": 0.1440981924533844,
       "learning_rate": 0.00010251300954433376,
-      "loss": 0.3067,
+      "loss": 0.3063,
       "step": 492
     },
     {
       "epoch": 0.0527893778777171,
-      "grad_norm": 0.1998199224472046,
+      "grad_norm": 0.1973811239004135,
       "learning_rate": 0.00010219893760925052,
-      "loss": 0.4282,
+      "loss": 0.4275,
       "step": 493
     },
     {
       "epoch": 0.052896455723310844,
-      "grad_norm": 0.15826267004013062,
+      "grad_norm": 0.15629740059375763,
       "learning_rate": 0.00010188484397154084,
-      "loss": 0.3698,
+      "loss": 0.3702,
       "step": 494
     },
     {
       "epoch": 0.053003533568904596,
-      "grad_norm": 0.15670500695705414,
+      "grad_norm": 0.15883494913578033,
       "learning_rate": 0.00010157073173118208,
-      "loss": 0.3849,
+      "loss": 0.3856,
       "step": 495
     },
     {
       "epoch": 0.05311061141449834,
-      "grad_norm": 0.214689239859581,
+      "grad_norm": 0.21182221174240112,
       "learning_rate": 0.00010125660398833528,
-      "loss": 0.4448,
+      "loss": 0.4464,
       "step": 496
     },
     {
       "epoch": 0.053217689260092085,
-      "grad_norm": 0.17683358490467072,
+      "grad_norm": 0.179497629404068,
       "learning_rate": 0.00010094246384331442,
-      "loss": 0.307,
+      "loss": 0.3077,
       "step": 497
     },
     {
       "epoch": 0.05332476710568583,
-      "grad_norm": 0.236779123544693,
+      "grad_norm": 0.2272624969482422,
       "learning_rate": 0.00010062831439655591,
-      "loss": 0.4571,
+      "loss": 0.4539,
       "step": 498
     },
     {
       "epoch": 0.05343184495127958,
-      "grad_norm": 0.19618524610996246,
+      "grad_norm": 0.19556404650211334,
       "learning_rate": 0.00010031415874858797,
-      "loss": 0.5339,
+      "loss": 0.5309,
       "step": 499
     },
     {
       "epoch": 0.05353892279687333,
-      "grad_norm": 0.14964862167835236,
+      "grad_norm": 0.14988619089126587,
       "learning_rate": 0.0001,
-      "loss": 0.246,
+      "loss": 0.2452,
       "step": 500
     },
     {
       "epoch": 0.05364600064246707,
-      "grad_norm": 0.15746654570102692,
+      "grad_norm": 0.15577977895736694,
       "learning_rate": 9.968584125141204e-05,
-      "loss": 0.3532,
+      "loss": 0.3528,
       "step": 501
     },
     {
       "epoch": 0.053753078488060824,
-      "grad_norm": 0.21315313875675201,
+      "grad_norm": 0.2019302248954773,
       "learning_rate": 9.937168560344412e-05,
-      "loss": 0.4048,
+      "loss": 0.4053,
       "step": 502
     },
     {
       "epoch": 0.05386015633365457,
-      "grad_norm": 0.17655318975448608,
+      "grad_norm": 0.17697198688983917,
       "learning_rate": 9.90575361566856e-05,
-      "loss": 0.319,
+      "loss": 0.3189,
       "step": 503
     },
     {
       "epoch": 0.05396723417924831,
-      "grad_norm": 0.15725268423557281,
+      "grad_norm": 0.15755611658096313,
       "learning_rate": 9.874339601166473e-05,
-      "loss": 0.2798,
+      "loss": 0.2804,
       "step": 504
     },
     {
       "epoch": 0.05407431202484206,
-      "grad_norm": 0.15933293104171753,
+      "grad_norm": 0.1574283093214035,
       "learning_rate": 9.842926826881796e-05,
-      "loss": 0.3104,
+      "loss": 0.3097,
       "step": 505
     },
     {
       "epoch": 0.05418138987043581,
-      "grad_norm": 0.11724121868610382,
+      "grad_norm": 0.11551764607429504,
       "learning_rate": 9.81151560284592e-05,
-      "loss": 0.2285,
+      "loss": 0.2281,
       "step": 506
     },
     {
       "epoch": 0.054288467716029555,
-      "grad_norm": 0.16910420358181,
+      "grad_norm": 0.17114503681659698,
       "learning_rate": 9.78010623907495e-05,
-      "loss": 0.3072,
+      "loss": 0.306,
       "step": 507
     },
     {
       "epoch": 0.0543955455616233,
-      "grad_norm": 0.15870051085948944,
+      "grad_norm": 0.15964044630527496,
       "learning_rate": 9.748699045566626e-05,
-      "loss": 0.3475,
+      "loss": 0.3484,
       "step": 508
     },
     {
       "epoch": 0.054502623407217045,
-      "grad_norm": 0.15897218883037567,
+      "grad_norm": 0.1551811248064041,
       "learning_rate": 9.717294332297268e-05,
-      "loss": 0.2984,
+      "loss": 0.2986,
       "step": 509
     },
     {
       "epoch": 0.0546097012528108,
-      "grad_norm": 0.15372000634670258,
+      "grad_norm": 0.15225476026535034,
       "learning_rate": 9.685892409218717e-05,
-      "loss": 0.3385,
+      "loss": 0.3387,
       "step": 510
     },
     {
       "epoch": 0.05471677909840454,
-      "grad_norm": 0.18505370616912842,
+      "grad_norm": 0.18411734700202942,
       "learning_rate": 9.654493586255278e-05,
-      "loss": 0.3844,
+      "loss": 0.384,
       "step": 511
     },
     {
       "epoch": 0.054823856943998286,
-      "grad_norm": 0.1893370896577835,
+      "grad_norm": 0.1901894211769104,
       "learning_rate": 9.623098173300654e-05,
-      "loss": 0.4111,
+      "loss": 0.4127,
       "step": 512
     },
     {
       "epoch": 0.05493093478959203,
-      "grad_norm": 0.1814618706703186,
+      "grad_norm": 0.18249692022800446,
       "learning_rate": 9.591706480214901e-05,
-      "loss": 0.2973,
+      "loss": 0.2972,
       "step": 513
     },
     {
       "epoch": 0.05503801263518578,
-      "grad_norm": 0.17390666902065277,
+      "grad_norm": 0.16920414566993713,
       "learning_rate": 9.560318816821353e-05,
-      "loss": 0.3329,
+      "loss": 0.3322,
       "step": 514
     },
     {
       "epoch": 0.05514509048077953,
-      "grad_norm": 0.14959359169006348,
+      "grad_norm": 0.1470497101545334,
       "learning_rate": 9.528935492903575e-05,
-      "loss": 0.2716,
+      "loss": 0.2718,
       "step": 515
     },
     {
       "epoch": 0.05525216832637327,
-      "grad_norm": 0.1462404578924179,
+      "grad_norm": 0.1421034187078476,
       "learning_rate": 9.497556818202306e-05,
-      "loss": 0.3095,
+      "loss": 0.3081,
       "step": 516
     },
     {
       "epoch": 0.05535924617196702,
-      "grad_norm": 0.1766779124736786,
+      "grad_norm": 0.17241626977920532,
       "learning_rate": 9.466183102412395e-05,
-      "loss": 0.3714,
+      "loss": 0.3708,
       "step": 517
     },
     {
       "epoch": 0.05546632401756077,
-      "grad_norm": 0.1642884612083435,
+      "grad_norm": 0.16179907321929932,
       "learning_rate": 9.434814655179755e-05,
-      "loss": 0.2852,
+      "loss": 0.2864,
       "step": 518
     },
     {
       "epoch": 0.055573401863154515,
-      "grad_norm": 0.16393111646175385,
+      "grad_norm": 0.16402095556259155,
       "learning_rate": 9.403451786098294e-05,
-      "loss": 0.3584,
+      "loss": 0.3596,
       "step": 519
     },
     {
       "epoch": 0.05568047970874826,
-      "grad_norm": 0.2010338306427002,
+      "grad_norm": 0.20050428807735443,
       "learning_rate": 9.372094804706867e-05,
-      "loss": 0.515,
+      "loss": 0.5162,
       "step": 520
     },
     {
       "epoch": 0.055787557554342004,
-      "grad_norm": 0.148441344499588,
+      "grad_norm": 0.15083353221416473,
       "learning_rate": 9.340744020486222e-05,
-      "loss": 0.3691,
+      "loss": 0.3693,
       "step": 521
     },
     {
       "epoch": 0.055894635399935756,
-      "grad_norm": 0.18052120506763458,
+      "grad_norm": 0.15941321849822998,
       "learning_rate": 9.309399742855942e-05,
-      "loss": 0.3099,
+      "loss": 0.3116,
       "step": 522
     },
     {
       "epoch": 0.0560017132455295,
-      "grad_norm": 0.13848088681697845,
+      "grad_norm": 0.13679879903793335,
       "learning_rate": 9.278062281171393e-05,
-      "loss": 0.2463,
+      "loss": 0.2461,
       "step": 523
     },
     {
       "epoch": 0.056108791091123246,
-      "grad_norm": 0.19286379218101501,
+      "grad_norm": 0.1818103790283203,
       "learning_rate": 9.246731944720675e-05,
-      "loss": 0.3067,
+      "loss": 0.3061,
       "step": 524
     },
     {
       "epoch": 0.05621586893671699,
-      "grad_norm": 0.16500720381736755,
+      "grad_norm": 0.15920798480510712,
       "learning_rate": 9.215409042721552e-05,
-      "loss": 0.3342,
+      "loss": 0.3344,
       "step": 525
     },
     {
       "epoch": 0.05632294678231074,
-      "grad_norm": 0.223133385181427,
+      "grad_norm": 0.2193126231431961,
       "learning_rate": 9.184093884318425e-05,
-      "loss": 0.5118,
+      "loss": 0.5115,
       "step": 526
     },
     {
       "epoch": 0.05643002462790449,
-      "grad_norm": 0.16499963402748108,
+      "grad_norm": 0.16002865135669708,
       "learning_rate": 9.152786778579267e-05,
-      "loss": 0.3624,
+      "loss": 0.3645,
       "step": 527
     },
     {
       "epoch": 0.05653710247349823,
-      "grad_norm": 0.14940662682056427,
+      "grad_norm": 0.14729884266853333,
       "learning_rate": 9.121488034492569e-05,
-      "loss": 0.307,
+      "loss": 0.3054,
       "step": 528
     },
     {
       "epoch": 0.05664418031909198,
-      "grad_norm": 0.15789055824279785,
+      "grad_norm": 0.15512928366661072,
       "learning_rate": 9.090197960964301e-05,
-      "loss": 0.3075,
+      "loss": 0.3074,
       "step": 529
     },
     {
       "epoch": 0.05675125816468573,
-      "grad_norm": 0.16344444453716278,
+      "grad_norm": 0.16242021322250366,
       "learning_rate": 9.058916866814858e-05,
-      "loss": 0.3722,
+      "loss": 0.3717,
       "step": 530
     },
     {
       "epoch": 0.056858336010279474,
-      "grad_norm": 0.18045423924922943,
+      "grad_norm": 0.17270362377166748,
       "learning_rate": 9.027645060776006e-05,
-      "loss": 0.4066,
+      "loss": 0.4058,
       "step": 531
     },
     {
       "epoch": 0.05696541385587322,
-      "grad_norm": 0.14639213681221008,
+      "grad_norm": 0.1453240066766739,
       "learning_rate": 8.99638285148785e-05,
-      "loss": 0.24,
+      "loss": 0.2409,
       "step": 532
     },
     {
       "epoch": 0.057072491701466964,
-      "grad_norm": 0.1606295108795166,
+      "grad_norm": 0.1585230678319931,
       "learning_rate": 8.965130547495776e-05,
-      "loss": 0.3495,
+      "loss": 0.3502,
       "step": 533
     },
     {
       "epoch": 0.057179569547060716,
-      "grad_norm": 0.17020860314369202,
+      "grad_norm": 0.16663610935211182,
       "learning_rate": 8.933888457247402e-05,
-      "loss": 0.3995,
+      "loss": 0.3985,
       "step": 534
     },
     {
       "epoch": 0.05728664739265446,
-      "grad_norm": 0.16404946148395538,
+      "grad_norm": 0.16193991899490356,
       "learning_rate": 8.902656889089548e-05,
-      "loss": 0.3087,
+      "loss": 0.308,
       "step": 535
     },
     {
       "epoch": 0.057393725238248205,
-      "grad_norm": 0.16885989904403687,
+      "grad_norm": 0.1661953181028366,
       "learning_rate": 8.871436151265184e-05,
-      "loss": 0.3451,
+      "loss": 0.3457,
       "step": 536
     },
     {
       "epoch": 0.05750080308384195,
-      "grad_norm": 0.16032719612121582,
+      "grad_norm": 0.15939539670944214,
       "learning_rate": 8.840226551910387e-05,
-      "loss": 0.3151,
+      "loss": 0.3158,
       "step": 537
     },
     {
       "epoch": 0.0576078809294357,
-      "grad_norm": 0.16192081570625305,
+      "grad_norm": 0.16002564132213593,
       "learning_rate": 8.809028399051302e-05,
-      "loss": 0.3284,
+      "loss": 0.3278,
       "step": 538
     },
     {
       "epoch": 0.05771495877502945,
-      "grad_norm": 0.16214251518249512,
+      "grad_norm": 0.1772623062133789,
       "learning_rate": 8.777842000601105e-05,
-      "loss": 0.2159,
+      "loss": 0.2166,
       "step": 539
     },
     {
       "epoch": 0.05782203662062319,
-      "grad_norm": 0.2248457819223404,
+      "grad_norm": 0.22258137166500092,
       "learning_rate": 8.746667664356956e-05,
-      "loss": 0.3886,
+      "loss": 0.3896,
       "step": 540
     },
     {
       "epoch": 0.05792911446621694,
-      "grad_norm": 0.17317667603492737,
+      "grad_norm": 0.1760905683040619,
       "learning_rate": 8.715505697996971e-05,
-      "loss": 0.2789,
+      "loss": 0.2794,
       "step": 541
     },
     {
       "epoch": 0.05803619231181069,
-      "grad_norm": 0.18882204592227936,
+      "grad_norm": 0.18258985877037048,
       "learning_rate": 8.684356409077176e-05,
-      "loss": 0.3398,
+      "loss": 0.3412,
       "step": 542
     },
     {
       "epoch": 0.058143270157404434,
-      "grad_norm": 0.2111193686723709,
+      "grad_norm": 0.20860691368579865,
       "learning_rate": 8.653220105028474e-05,
-      "loss": 0.4288,
+      "loss": 0.4274,
       "step": 543
     },
     {
       "epoch": 0.05825034800299818,
-      "grad_norm": 0.14698849618434906,
+      "grad_norm": 0.14682573080062866,
       "learning_rate": 8.62209709315362e-05,
-      "loss": 0.3265,
+      "loss": 0.3266,
       "step": 544
     },
     {
       "epoch": 0.05835742584859192,
-      "grad_norm": 0.15182967483997345,
+      "grad_norm": 0.15006284415721893,
       "learning_rate": 8.590987680624174e-05,
-      "loss": 0.2549,
+      "loss": 0.2551,
       "step": 545
     },
     {
       "epoch": 0.058464503694185675,
-      "grad_norm": 0.1783917099237442,
+      "grad_norm": 0.17763254046440125,
       "learning_rate": 8.559892174477479e-05,
-      "loss": 0.4651,
+      "loss": 0.4648,
       "step": 546
     },
     {
       "epoch": 0.05857158153977942,
-      "grad_norm": 0.2132243663072586,
+      "grad_norm": 0.21015936136245728,
       "learning_rate": 8.528810881613626e-05,
-      "loss": 0.3032,
+      "loss": 0.3025,
       "step": 547
     },
     {
       "epoch": 0.058678659385373165,
-      "grad_norm": 0.18143096566200256,
+      "grad_norm": 0.1679753214120865,
       "learning_rate": 8.497744108792429e-05,
-      "loss": 0.2871,
+      "loss": 0.2861,
       "step": 548
     },
     {
       "epoch": 0.05878573723096691,
-      "grad_norm": 0.1841660588979721,
+      "grad_norm": 0.18313933908939362,
       "learning_rate": 8.466692162630392e-05,
-      "loss": 0.3478,
+      "loss": 0.3483,
       "step": 549
     },
     {
       "epoch": 0.05889281507656066,
-      "grad_norm": 0.16989177465438843,
+      "grad_norm": 0.16596145927906036,
       "learning_rate": 8.435655349597689e-05,
-      "loss": 0.3273,
+      "loss": 0.3251,
       "step": 550
     },
     {
       "epoch": 0.05899989292215441,
-      "grad_norm": 0.1569872945547104,
+      "grad_norm": 0.18846194446086884,
       "learning_rate": 8.404633976015134e-05,
-      "loss": 0.2909,
+      "loss": 0.2898,
       "step": 551
     },
     {
       "epoch": 0.05910697076774815,
-      "grad_norm": 0.15851524472236633,
+      "grad_norm": 0.15795886516571045,
       "learning_rate": 8.373628348051165e-05,
-      "loss": 0.2183,
+      "loss": 0.2182,
       "step": 552
     },
     {
       "epoch": 0.059214048613341896,
-      "grad_norm": 0.16590699553489685,
+      "grad_norm": 0.14111833274364471,
       "learning_rate": 8.342638771718802e-05,
-      "loss": 0.215,
+      "loss": 0.2136,
       "step": 553
     },
     {
       "epoch": 0.05932112645893565,
-      "grad_norm": 0.2867710590362549,
+      "grad_norm": 0.2854287624359131,
       "learning_rate": 8.311665552872662e-05,
-      "loss": 0.4176,
+      "loss": 0.4156,
       "step": 554
     },
     {
       "epoch": 0.05942820430452939,
-      "grad_norm": 0.16975639760494232,
+      "grad_norm": 0.16893833875656128,
       "learning_rate": 8.280708997205904e-05,
-      "loss": 0.34,
+      "loss": 0.3394,
       "step": 555
     },
     {
       "epoch": 0.05953528215012314,
-      "grad_norm": 0.15163715183734894,
+      "grad_norm": 0.14959296584129333,
       "learning_rate": 8.249769410247239e-05,
-      "loss": 0.275,
+      "loss": 0.2749,
       "step": 556
     },
     {
       "epoch": 0.05964235999571688,
-      "grad_norm": 0.1733134686946869,
+      "grad_norm": 0.16701342165470123,
       "learning_rate": 8.218847097357898e-05,
-      "loss": 0.2584,
+      "loss": 0.257,
       "step": 557
     },
     {
       "epoch": 0.059749437841310635,
-      "grad_norm": 0.16540873050689697,
+      "grad_norm": 0.16094553470611572,
       "learning_rate": 8.187942363728625e-05,
-      "loss": 0.2495,
+      "loss": 0.2504,
       "step": 558
     },
     {
       "epoch": 0.05985651568690438,
-      "grad_norm": 0.18063515424728394,
+      "grad_norm": 0.1865304857492447,
       "learning_rate": 8.157055514376666e-05,
-      "loss": 0.3324,
+      "loss": 0.3331,
       "step": 559
     },
     {
       "epoch": 0.059963593532498125,
-      "grad_norm": 0.1537630259990692,
+      "grad_norm": 0.1482362002134323,
       "learning_rate": 8.126186854142752e-05,
-      "loss": 0.2322,
+      "loss": 0.2317,
       "step": 560
     },
     {
       "epoch": 0.06007067137809187,
-      "grad_norm": 0.14401783049106598,
+      "grad_norm": 0.14238731563091278,
       "learning_rate": 8.095336687688102e-05,
-      "loss": 0.2909,
+      "loss": 0.2891,
       "step": 561
     },
     {
       "epoch": 0.06017774922368562,
-      "grad_norm": 0.19324550032615662,
+      "grad_norm": 0.19169366359710693,
       "learning_rate": 8.064505319491398e-05,
-      "loss": 0.4552,
+      "loss": 0.455,
       "step": 562
     },
     {
       "epoch": 0.060284827069279366,
-      "grad_norm": 0.13768549263477325,
+      "grad_norm": 0.1366347372531891,
       "learning_rate": 8.033693053845801e-05,
-      "loss": 0.2493,
+      "loss": 0.2489,
       "step": 563
     },
     {
       "epoch": 0.06039190491487311,
-      "grad_norm": 0.17372079193592072,
+      "grad_norm": 0.168013796210289,
       "learning_rate": 8.002900194855932e-05,
-      "loss": 0.3261,
+      "loss": 0.3256,
       "step": 564
     },
     {
       "epoch": 0.06049898276046686,
-      "grad_norm": 0.18107670545578003,
+      "grad_norm": 0.18004128336906433,
       "learning_rate": 7.972127046434878e-05,
-      "loss": 0.3877,
+      "loss": 0.3891,
       "step": 565
     },
     {
       "epoch": 0.06060606060606061,
-      "grad_norm": 0.1908927708864212,
+      "grad_norm": 0.18658877909183502,
       "learning_rate": 7.941373912301189e-05,
-      "loss": 0.4814,
+      "loss": 0.4804,
       "step": 566
     },
     {
       "epoch": 0.06071313845165435,
-      "grad_norm": 0.20689429342746735,
+      "grad_norm": 0.20196396112442017,
       "learning_rate": 7.910641095975886e-05,
-      "loss": 0.432,
+      "loss": 0.4308,
       "step": 567
     },
     {
       "epoch": 0.0608202162972481,
-      "grad_norm": 0.18664167821407318,
+      "grad_norm": 0.18546605110168457,
       "learning_rate": 7.879928900779456e-05,
-      "loss": 0.3784,
+      "loss": 0.3777,
       "step": 568
     },
     {
       "epoch": 0.06092729414284185,
-      "grad_norm": 0.16357581317424774,
+      "grad_norm": 0.16258494555950165,
       "learning_rate": 7.849237629828869e-05,
-      "loss": 0.3061,
+      "loss": 0.3056,
       "step": 569
     },
     {
       "epoch": 0.061034371988435594,
-      "grad_norm": 0.20243744552135468,
+      "grad_norm": 0.207724466919899,
       "learning_rate": 7.818567586034577e-05,
-      "loss": 0.326,
+      "loss": 0.3235,
       "step": 570
     },
     {
       "epoch": 0.06114144983402934,
-      "grad_norm": 0.15626615285873413,
+      "grad_norm": 0.15525054931640625,
       "learning_rate": 7.787919072097531e-05,
-      "loss": 0.3051,
+      "loss": 0.3045,
       "step": 571
     },
     {
       "epoch": 0.061248527679623084,
-      "grad_norm": 0.16009844839572906,
+      "grad_norm": 0.1632733941078186,
       "learning_rate": 7.75729239050619e-05,
-      "loss": 0.3191,
+      "loss": 0.3197,
       "step": 572
     },
     {
       "epoch": 0.061355605525216836,
-      "grad_norm": 0.19337281584739685,
+      "grad_norm": 0.19123557209968567,
       "learning_rate": 7.726687843533538e-05,
-      "loss": 0.4369,
+      "loss": 0.4378,
       "step": 573
     },
     {
       "epoch": 0.06146268337081058,
-      "grad_norm": 0.17653042078018188,
+      "grad_norm": 0.1749892383813858,
       "learning_rate": 7.696105733234098e-05,
-      "loss": 0.3309,
+      "loss": 0.332,
       "step": 574
     },
     {
       "epoch": 0.061569761216404326,
-      "grad_norm": 0.16593199968338013,
+      "grad_norm": 0.16800446808338165,
       "learning_rate": 7.66554636144095e-05,
-      "loss": 0.44,
+      "loss": 0.4398,
       "step": 575
     },
     {
       "epoch": 0.06167683906199807,
-      "grad_norm": 0.16403137147426605,
+      "grad_norm": 0.164847269654274,
       "learning_rate": 7.635010029762756e-05,
-      "loss": 0.2433,
+      "loss": 0.2423,
       "step": 576
     },
     {
       "epoch": 0.06178391690759182,
-      "grad_norm": 0.13667182624340057,
+      "grad_norm": 0.13671499490737915,
       "learning_rate": 7.604497039580785e-05,
-      "loss": 0.2799,
+      "loss": 0.2792,
       "step": 577
     },
     {
       "epoch": 0.06189099475318557,
-      "grad_norm": 0.16916875541210175,
+      "grad_norm": 0.17302730679512024,
       "learning_rate": 7.574007692045928e-05,
-      "loss": 0.3701,
+      "loss": 0.3691,
       "step": 578
     },
     {
       "epoch": 0.06199807259877931,
-      "grad_norm": 0.16624587774276733,
+      "grad_norm": 0.16944162547588348,
       "learning_rate": 7.543542288075739e-05,
-      "loss": 0.2409,
+      "loss": 0.2405,
       "step": 579
     },
     {
       "epoch": 0.06210515044437306,
-      "grad_norm": 0.2018321007490158,
+      "grad_norm": 0.20139697194099426,
       "learning_rate": 7.513101128351454e-05,
-      "loss": 0.3169,
+      "loss": 0.3166,
       "step": 580
     },
     {
       "epoch": 0.06221222828996681,
-      "grad_norm": 0.16380462050437927,
+      "grad_norm": 0.16109436750411987,
       "learning_rate": 7.48268451331503e-05,
-      "loss": 0.4424,
+      "loss": 0.4416,
       "step": 581
     },
     {
       "epoch": 0.062319306135560554,
-      "grad_norm": 0.14434152841567993,
+      "grad_norm": 0.1434623748064041,
       "learning_rate": 7.45229274316618e-05,
-      "loss": 0.2669,
+      "loss": 0.266,
       "step": 582
     },
     {
       "epoch": 0.0624263839811543,
-      "grad_norm": 0.22366313636302948,
+      "grad_norm": 0.22343096137046814,
       "learning_rate": 7.421926117859403e-05,
-      "loss": 0.455,
+      "loss": 0.4556,
       "step": 583
     },
     {
       "epoch": 0.06253346182674804,
-      "grad_norm": 0.18707706034183502,
+      "grad_norm": 0.18569520115852356,
       "learning_rate": 7.391584937101033e-05,
-      "loss": 0.3651,
+      "loss": 0.3634,
       "step": 584
     },
     {
       "epoch": 0.06264053967234179,
-      "grad_norm": 0.15041711926460266,
+      "grad_norm": 0.1524548977613449,
       "learning_rate": 7.361269500346274e-05,
-      "loss": 0.3066,
+      "loss": 0.3065,
       "step": 585
     },
     {
       "epoch": 0.06274761751793553,
-      "grad_norm": 0.19988040626049042,
+      "grad_norm": 0.2060038298368454,
       "learning_rate": 7.330980106796246e-05,
-      "loss": 0.3106,
+      "loss": 0.3112,
       "step": 586
     },
     {
       "epoch": 0.06285469536352929,
-      "grad_norm": 0.18236958980560303,
+      "grad_norm": 0.1781323105096817,
       "learning_rate": 7.300717055395039e-05,
-      "loss": 0.3529,
+      "loss": 0.3519,
       "step": 587
     },
     {
       "epoch": 0.06296177320912304,
-      "grad_norm": 0.15908972918987274,
+      "grad_norm": 0.16407997906208038,
       "learning_rate": 7.270480644826749e-05,
-      "loss": 0.262,
+      "loss": 0.261,
       "step": 588
     },
     {
       "epoch": 0.06306885105471678,
-      "grad_norm": 0.16096021234989166,
+      "grad_norm": 0.15888680517673492,
       "learning_rate": 7.240271173512546e-05,
-      "loss": 0.3349,
+      "loss": 0.3337,
       "step": 589
     },
     {
       "epoch": 0.06317592890031053,
-      "grad_norm": 0.14579513669013977,
+      "grad_norm": 0.14741234481334686,
       "learning_rate": 7.210088939607708e-05,
-      "loss": 0.3535,
+      "loss": 0.3529,
       "step": 590
     },
     {
       "epoch": 0.06328300674590427,
-      "grad_norm": 0.1794016808271408,
+      "grad_norm": 0.17969931662082672,
       "learning_rate": 7.179934240998706e-05,
-      "loss": 0.3372,
+      "loss": 0.3359,
       "step": 591
     },
     {
       "epoch": 0.06339008459149802,
-      "grad_norm": 0.1510874330997467,
+      "grad_norm": 0.15772207081317902,
       "learning_rate": 7.149807375300239e-05,
-      "loss": 0.2272,
+      "loss": 0.2282,
       "step": 592
     },
     {
       "epoch": 0.06349716243709176,
-      "grad_norm": 0.23262274265289307,
+      "grad_norm": 0.23129960894584656,
       "learning_rate": 7.119708639852312e-05,
-      "loss": 0.4265,
+      "loss": 0.4251,
       "step": 593
     },
     {
       "epoch": 0.0636042402826855,
-      "grad_norm": 0.20220021903514862,
+      "grad_norm": 0.19827698171138763,
       "learning_rate": 7.089638331717284e-05,
-      "loss": 0.4163,
+      "loss": 0.4167,
       "step": 594
     },
     {
       "epoch": 0.06371131812827927,
-      "grad_norm": 0.1835254728794098,
+      "grad_norm": 0.18540829420089722,
       "learning_rate": 7.059596747676962e-05,
-      "loss": 0.3373,
+      "loss": 0.3376,
       "step": 595
     },
     {
       "epoch": 0.06381839597387301,
-      "grad_norm": 0.1331084668636322,
+      "grad_norm": 0.13156816363334656,
       "learning_rate": 7.029584184229653e-05,
-      "loss": 0.2568,
+      "loss": 0.2571,
       "step": 596
     },
     {
       "epoch": 0.06392547381946675,
-      "grad_norm": 0.1771346926689148,
+      "grad_norm": 0.1740744709968567,
       "learning_rate": 6.999600937587239e-05,
-      "loss": 0.3663,
+      "loss": 0.3639,
       "step": 597
     },
     {
       "epoch": 0.0640325516650605,
-      "grad_norm": 0.16743473708629608,
+      "grad_norm": 0.170998677611351,
       "learning_rate": 6.969647303672262e-05,
-      "loss": 0.292,
+      "loss": 0.2915,
       "step": 598
     },
     {
       "epoch": 0.06413962951065424,
-      "grad_norm": 0.15180084109306335,
+      "grad_norm": 0.15191666781902313,
       "learning_rate": 6.939723578114993e-05,
-      "loss": 0.2849,
+      "loss": 0.2864,
       "step": 599
     },
     {
       "epoch": 0.06424670735624799,
-      "grad_norm": 0.15079252421855927,
+      "grad_norm": 0.150756374001503,
       "learning_rate": 6.909830056250527e-05,
-      "loss": 0.3499,
+      "loss": 0.3503,
       "step": 600
     },
     {
       "epoch": 0.06435378520184173,
-      "grad_norm": 0.16857562959194183,
+      "grad_norm": 0.1681331992149353,
       "learning_rate": 6.879967033115853e-05,
-      "loss": 0.3369,
+      "loss": 0.336,
       "step": 601
     },
     {
       "epoch": 0.06446086304743548,
-      "grad_norm": 0.15610823035240173,
+      "grad_norm": 0.1576251983642578,
       "learning_rate": 6.850134803446954e-05,
-      "loss": 0.4078,
+      "loss": 0.4084,
       "step": 602
     },
     {
       "epoch": 0.06456794089302924,
-      "grad_norm": 0.16949012875556946,
+      "grad_norm": 0.17241467535495758,
       "learning_rate": 6.820333661675893e-05,
-      "loss": 0.326,
+      "loss": 0.3258,
       "step": 603
     },
     {
       "epoch": 0.06467501873862298,
-      "grad_norm": 0.15723907947540283,
+      "grad_norm": 0.15638768672943115,
       "learning_rate": 6.790563901927907e-05,
-      "loss": 0.3191,
+      "loss": 0.319,
       "step": 604
     },
     {
       "epoch": 0.06478209658421673,
-      "grad_norm": 0.1530887633562088,
+      "grad_norm": 0.15127749741077423,
       "learning_rate": 6.760825818018508e-05,
-      "loss": 0.3864,
+      "loss": 0.385,
       "step": 605
     },
     {
       "epoch": 0.06488917442981047,
-      "grad_norm": 0.1763346791267395,
+      "grad_norm": 0.18429574370384216,
       "learning_rate": 6.731119703450577e-05,
-      "loss": 0.3351,
+      "loss": 0.3338,
       "step": 606
     },
     {
       "epoch": 0.06499625227540422,
-      "grad_norm": 0.1836361289024353,
+      "grad_norm": 0.1837109476327896,
       "learning_rate": 6.701445851411472e-05,
-      "loss": 0.307,
+      "loss": 0.3073,
       "step": 607
     },
     {
       "epoch": 0.06510333012099796,
-      "grad_norm": 0.1673685908317566,
+      "grad_norm": 0.1658046543598175,
       "learning_rate": 6.671804554770135e-05,
-      "loss": 0.2753,
+      "loss": 0.2748,
       "step": 608
     },
     {
       "epoch": 0.06521040796659171,
-      "grad_norm": 0.1554986834526062,
+      "grad_norm": 0.15374763309955597,
       "learning_rate": 6.642196106074194e-05,
-      "loss": 0.3076,
+      "loss": 0.3071,
       "step": 609
     },
     {
       "epoch": 0.06531748581218545,
-      "grad_norm": 0.15603835880756378,
+      "grad_norm": 0.15231020748615265,
       "learning_rate": 6.612620797547087e-05,
-      "loss": 0.3019,
+      "loss": 0.3016,
       "step": 610
     },
     {
       "epoch": 0.06542456365777921,
-      "grad_norm": 0.1522570699453354,
+      "grad_norm": 0.15196466445922852,
       "learning_rate": 6.583078921085167e-05,
-      "loss": 0.2793,
+      "loss": 0.2787,
       "step": 611
     },
     {
       "epoch": 0.06553164150337296,
-      "grad_norm": 0.18546074628829956,
+      "grad_norm": 0.1837112009525299,
       "learning_rate": 6.55357076825483e-05,
-      "loss": 0.3955,
+      "loss": 0.3969,
       "step": 612
     },
     {
       "epoch": 0.0656387193489667,
-      "grad_norm": 0.14978669583797455,
+      "grad_norm": 0.14765381813049316,
       "learning_rate": 6.52409663028963e-05,
-      "loss": 0.3413,
+      "loss": 0.3398,
       "step": 613
     },
     {
       "epoch": 0.06574579719456045,
-      "grad_norm": 0.2138916552066803,
+      "grad_norm": 0.21580912172794342,
       "learning_rate": 6.494656798087412e-05,
-      "loss": 0.3169,
+      "loss": 0.3179,
       "step": 614
     },
     {
       "epoch": 0.06585287504015419,
-      "grad_norm": 0.1433931142091751,
+      "grad_norm": 0.14525169134140015,
       "learning_rate": 6.465251562207431e-05,
-      "loss": 0.2771,
+      "loss": 0.2758,
       "step": 615
     },
     {
       "epoch": 0.06595995288574794,
-      "grad_norm": 0.1670522540807724,
+      "grad_norm": 0.1643163412809372,
       "learning_rate": 6.435881212867493e-05,
-      "loss": 0.3817,
+      "loss": 0.3803,
       "step": 616
     },
     {
       "epoch": 0.06606703073134168,
-      "grad_norm": 0.17601914703845978,
+      "grad_norm": 0.18450172245502472,
       "learning_rate": 6.406546039941094e-05,
-      "loss": 0.3462,
+      "loss": 0.3461,
       "step": 617
     },
     {
       "epoch": 0.06617410857693543,
-      "grad_norm": 0.1717825084924698,
+      "grad_norm": 0.17202648520469666,
       "learning_rate": 6.377246332954544e-05,
-      "loss": 0.328,
+      "loss": 0.3287,
       "step": 618
     },
     {
       "epoch": 0.06628118642252918,
-      "grad_norm": 0.16034941375255585,
+      "grad_norm": 0.16204772889614105,
       "learning_rate": 6.347982381084123e-05,
-      "loss": 0.2962,
+      "loss": 0.2965,
       "step": 619
     },
     {
       "epoch": 0.06638826426812293,
-      "grad_norm": 0.1435386687517166,
+      "grad_norm": 0.14521515369415283,
       "learning_rate": 6.318754473153221e-05,
-      "loss": 0.2272,
+      "loss": 0.2263,
       "step": 620
     },
     {
       "epoch": 0.06649534211371667,
-      "grad_norm": 0.13801023364067078,
+      "grad_norm": 0.13584968447685242,
       "learning_rate": 6.289562897629492e-05,
-      "loss": 0.2799,
+      "loss": 0.2796,
       "step": 621
     },
     {
       "epoch": 0.06660241995931042,
-      "grad_norm": 0.15614642202854156,
+      "grad_norm": 0.15651963651180267,
       "learning_rate": 6.260407942621998e-05,
-      "loss": 0.441,
+      "loss": 0.4405,
       "step": 622
     },
     {
       "epoch": 0.06670949780490416,
-      "grad_norm": 0.16773013770580292,
+      "grad_norm": 0.1673763245344162,
       "learning_rate": 6.231289895878375e-05,
-      "loss": 0.2803,
+      "loss": 0.2816,
       "step": 623
     },
     {
       "epoch": 0.06681657565049791,
-      "grad_norm": 0.16002804040908813,
+      "grad_norm": 0.16071754693984985,
       "learning_rate": 6.20220904478199e-05,
-      "loss": 0.3051,
+      "loss": 0.3041,
       "step": 624
     },
     {
       "epoch": 0.06692365349609165,
-      "grad_norm": 0.15663942694664001,
+      "grad_norm": 0.15596891939640045,
       "learning_rate": 6.173165676349103e-05,
-      "loss": 0.3281,
+      "loss": 0.3273,
       "step": 625
     },
     {
       "epoch": 0.0670307313416854,
-      "grad_norm": 0.1416623592376709,
+      "grad_norm": 0.14081811904907227,
       "learning_rate": 6.144160077226036e-05,
-      "loss": 0.2683,
+      "loss": 0.2684,
       "step": 626
     },
     {
       "epoch": 0.06713780918727916,
-      "grad_norm": 0.15320400893688202,
+      "grad_norm": 0.15451277792453766,
       "learning_rate": 6.11519253368634e-05,
-      "loss": 0.2714,
+      "loss": 0.2725,
       "step": 627
     },
     {
       "epoch": 0.0672448870328729,
-      "grad_norm": 0.19521519541740417,
+      "grad_norm": 0.1893017739057541,
       "learning_rate": 6.086263331627976e-05,
-      "loss": 0.5039,
+      "loss": 0.5062,
       "step": 628
     },
     {
       "epoch": 0.06735196487846665,
-      "grad_norm": 0.182010218501091,
+      "grad_norm": 0.17904160916805267,
       "learning_rate": 6.05737275657049e-05,
-      "loss": 0.4331,
+      "loss": 0.4317,
       "step": 629
     },
     {
       "epoch": 0.06745904272406039,
-      "grad_norm": 0.20210574567317963,
+      "grad_norm": 0.20292861759662628,
       "learning_rate": 6.0285210936521955e-05,
-      "loss": 0.3858,
+      "loss": 0.3869,
       "step": 630
     },
     {
       "epoch": 0.06756612056965414,
-      "grad_norm": 0.15948592126369476,
+      "grad_norm": 0.1582748293876648,
       "learning_rate": 5.999708627627354e-05,
-      "loss": 0.2308,
+      "loss": 0.2296,
       "step": 631
     },
     {
       "epoch": 0.06767319841524788,
-      "grad_norm": 0.16815067827701569,
+      "grad_norm": 0.16759826242923737,
       "learning_rate": 5.9709356428633746e-05,
-      "loss": 0.2901,
+      "loss": 0.2913,
       "step": 632
     },
     {
       "epoch": 0.06778027626084163,
-      "grad_norm": 0.18218070268630981,
+      "grad_norm": 0.18039844930171967,
       "learning_rate": 5.9422024233380013e-05,
-      "loss": 0.3229,
+      "loss": 0.3225,
       "step": 633
     },
     {
       "epoch": 0.06788735410643539,
-      "grad_norm": 0.1760886311531067,
+      "grad_norm": 0.18086355924606323,
       "learning_rate": 5.913509252636511e-05,
-      "loss": 0.2355,
+      "loss": 0.2374,
       "step": 634
     },
     {
       "epoch": 0.06799443195202913,
-      "grad_norm": 0.14710214734077454,
+      "grad_norm": 0.1444912850856781,
       "learning_rate": 5.884856413948913e-05,
-      "loss": 0.3329,
+      "loss": 0.3337,
       "step": 635
     },
     {
       "epoch": 0.06810150979762288,
-      "grad_norm": 0.1451999992132187,
+      "grad_norm": 0.1468546986579895,
       "learning_rate": 5.856244190067159e-05,
-      "loss": 0.2894,
+      "loss": 0.2902,
       "step": 636
     },
     {
       "epoch": 0.06820858764321662,
-      "grad_norm": 0.17219161987304688,
+      "grad_norm": 0.17324474453926086,
       "learning_rate": 5.82767286338235e-05,
-      "loss": 0.3896,
+      "loss": 0.3888,
       "step": 637
     },
     {
       "epoch": 0.06831566548881036,
-      "grad_norm": 0.17436614632606506,
+      "grad_norm": 0.17539095878601074,
       "learning_rate": 5.799142715881938e-05,
-      "loss": 0.4378,
+      "loss": 0.439,
       "step": 638
     },
     {
       "epoch": 0.06842274333440411,
-      "grad_norm": 0.13370436429977417,
+      "grad_norm": 0.1325037181377411,
       "learning_rate": 5.770654029146969e-05,
-      "loss": 0.2642,
+      "loss": 0.2649,
       "step": 639
     },
     {
       "epoch": 0.06852982117999785,
-      "grad_norm": 0.1600145697593689,
+      "grad_norm": 0.1581648737192154,
       "learning_rate": 5.7422070843492734e-05,
-      "loss": 0.2838,
+      "loss": 0.2837,
       "step": 640
     },
     {
       "epoch": 0.0686368990255916,
-      "grad_norm": 0.18459884822368622,
+      "grad_norm": 0.17844368517398834,
       "learning_rate": 5.713802162248718e-05,
-      "loss": 0.3208,
+      "loss": 0.3192,
       "step": 641
     },
     {
       "epoch": 0.06874397687118536,
-      "grad_norm": 0.13782131671905518,
+      "grad_norm": 0.13828836381435394,
       "learning_rate": 5.6854395431904094e-05,
-      "loss": 0.3096,
+      "loss": 0.3098,
       "step": 642
     },
     {
       "epoch": 0.0688510547167791,
-      "grad_norm": 0.1642604023218155,
+      "grad_norm": 0.16136354207992554,
       "learning_rate": 5.657119507101954e-05,
-      "loss": 0.4043,
+      "loss": 0.4036,
       "step": 643
     },
     {
       "epoch": 0.06895813256237285,
-      "grad_norm": 0.17711035907268524,
+      "grad_norm": 0.17593613266944885,
       "learning_rate": 5.6288423334906735e-05,
-      "loss": 0.3593,
+      "loss": 0.3596,
       "step": 644
     },
     {
       "epoch": 0.06906521040796659,
-      "grad_norm": 0.17758657038211823,
+      "grad_norm": 0.1810455173254013,
       "learning_rate": 5.6006083014408484e-05,
-      "loss": 0.3003,
+      "loss": 0.3002,
       "step": 645
     },
     {
       "epoch": 0.06917228825356034,
-      "grad_norm": 0.18889419734477997,
+      "grad_norm": 0.18986423313617706,
       "learning_rate": 5.572417689610987e-05,
-      "loss": 0.3131,
+      "loss": 0.3121,
       "step": 646
     },
     {
       "epoch": 0.06927936609915408,
-      "grad_norm": 0.20904329419136047,
+      "grad_norm": 0.1990794688463211,
       "learning_rate": 5.544270776231038e-05,
-      "loss": 0.3291,
+      "loss": 0.3284,
       "step": 647
     },
     {
       "epoch": 0.06938644394474783,
-      "grad_norm": 0.15047985315322876,
+      "grad_norm": 0.15112511813640594,
       "learning_rate": 5.5161678390996796e-05,
-      "loss": 0.3102,
+      "loss": 0.3088,
       "step": 648
     },
     {
       "epoch": 0.06949352179034157,
-      "grad_norm": 0.23132561147212982,
+      "grad_norm": 0.23028168082237244,
       "learning_rate": 5.488109155581549e-05,
-      "loss": 0.4353,
+      "loss": 0.4338,
       "step": 649
     },
     {
       "epoch": 0.06960059963593533,
-      "grad_norm": 0.19946032762527466,
+      "grad_norm": 0.1991870403289795,
       "learning_rate": 5.4600950026045326e-05,
       "loss": 0.3372,
       "step": 650
     },
     {
       "epoch": 0.06970767748152908,
-      "grad_norm": 0.20680102705955505,
+      "grad_norm": 0.20563647150993347,
       "learning_rate": 5.4321256566570036e-05,
       "loss": 0.4628,
       "step": 651
     },
     {
       "epoch": 0.06981475532712282,
-      "grad_norm": 0.15579630434513092,
+      "grad_norm": 0.15499360859394073,
       "learning_rate": 5.404201393785122e-05,
       "loss": 0.2367,
       "step": 652
     },
     {
       "epoch": 0.06992183317271657,
-      "grad_norm": 0.20350880920886993,
+      "grad_norm": 0.20121176540851593,
       "learning_rate": 5.3763224895900846e-05,
-      "loss": 0.3568,
+      "loss": 0.3584,
       "step": 653
     },
     {
       "epoch": 0.07002891101831031,
-      "grad_norm": 0.17040583491325378,
+      "grad_norm": 0.16816240549087524,
       "learning_rate": 5.348489219225416e-05,
-      "loss": 0.3297,
+      "loss": 0.3287,
       "step": 654
     },
     {
       "epoch": 0.07013598886390406,
-      "grad_norm": 0.22072608768939972,
+      "grad_norm": 0.19340524077415466,
       "learning_rate": 5.320701857394268e-05,
-      "loss": 0.3243,
+      "loss": 0.3215,
       "step": 655
     },
     {
       "epoch": 0.0702430667094978,
-      "grad_norm": 0.16529580950737,
+      "grad_norm": 0.16680116951465607,
       "learning_rate": 5.292960678346675e-05,
-      "loss": 0.3504,
+      "loss": 0.3505,
       "step": 656
     },
     {
       "epoch": 0.07035014455509155,
-      "grad_norm": 0.17127272486686707,
+      "grad_norm": 0.1708446592092514,
       "learning_rate": 5.265265955876879e-05,
-      "loss": 0.3472,
+      "loss": 0.347,
       "step": 657
     },
     {
       "epoch": 0.0704572224006853,
-      "grad_norm": 0.18526853621006012,
+      "grad_norm": 0.1849069744348526,
       "learning_rate": 5.237617963320608e-05,
-      "loss": 0.302,
+      "loss": 0.3015,
       "step": 658
     },
     {
       "epoch": 0.07056430024627905,
-      "grad_norm": 0.18870258331298828,
+      "grad_norm": 0.18246212601661682,
       "learning_rate": 5.210016973552391e-05,
-      "loss": 0.3709,
+      "loss": 0.3702,
       "step": 659
     },
     {
       "epoch": 0.0706713780918728,
-      "grad_norm": 0.1274450570344925,
+      "grad_norm": 0.1265881061553955,
       "learning_rate": 5.182463258982846e-05,
-      "loss": 0.2597,
+      "loss": 0.2589,
       "step": 660
     },
     {
       "epoch": 0.07077845593746654,
-      "grad_norm": 0.17223431169986725,
+      "grad_norm": 0.1724609136581421,
       "learning_rate": 5.1549570915560206e-05,
-      "loss": 0.3575,
+      "loss": 0.3557,
       "step": 661
     },
     {
       "epoch": 0.07088553378306028,
-      "grad_norm": 0.16141627728939056,
+      "grad_norm": 0.15820424258708954,
       "learning_rate": 5.127498742746675e-05,
-      "loss": 0.3195,
+      "loss": 0.3181,
       "step": 662
     },
     {
       "epoch": 0.07099261162865403,
-      "grad_norm": 0.24255283176898956,
+      "grad_norm": 0.33776232600212097,
       "learning_rate": 5.100088483557634e-05,
-      "loss": 0.4154,
+      "loss": 0.4203,
       "step": 663
     },
     {
       "epoch": 0.07109968947424777,
-      "grad_norm": 0.18947787582874298,
+      "grad_norm": 0.19054165482521057,
       "learning_rate": 5.072726584517086e-05,
-      "loss": 0.3083,
+      "loss": 0.3089,
       "step": 664
     },
     {
       "epoch": 0.07120676731984152,
-      "grad_norm": 0.16710755228996277,
+      "grad_norm": 0.16495317220687866,
       "learning_rate": 5.045413315675924e-05,
-      "loss": 0.3242,
+      "loss": 0.3246,
       "step": 665
     },
     {
       "epoch": 0.07131384516543528,
-      "grad_norm": 0.16929441690444946,
+      "grad_norm": 0.16578929126262665,
       "learning_rate": 5.018148946605092e-05,
-      "loss": 0.2543,
+      "loss": 0.2536,
       "step": 666
     },
     {
       "epoch": 0.07142092301102902,
-      "grad_norm": 0.15913401544094086,
+      "grad_norm": 0.1600061058998108,
       "learning_rate": 4.990933746392899e-05,
-      "loss": 0.3979,
+      "loss": 0.3994,
       "step": 667
     },
     {
       "epoch": 0.07152800085662277,
-      "grad_norm": 0.2128509283065796,
+      "grad_norm": 0.2062436193227768,
       "learning_rate": 4.9637679836423924e-05,
-      "loss": 0.3802,
+      "loss": 0.3789,
       "step": 668
     },
     {
       "epoch": 0.07163507870221651,
-      "grad_norm": 0.19346699118614197,
+      "grad_norm": 0.197468101978302,
       "learning_rate": 4.9366519264686725e-05,
-      "loss": 0.3781,
+      "loss": 0.3784,
       "step": 669
     },
     {
       "epoch": 0.07174215654781026,
-      "grad_norm": 0.19327999651432037,
+      "grad_norm": 0.19289591908454895,
       "learning_rate": 4.909585842496287e-05,
-      "loss": 0.3583,
+      "loss": 0.3574,
       "step": 670
     },
     {
       "epoch": 0.071849234393404,
-      "grad_norm": 0.13450753688812256,
+      "grad_norm": 0.13273729383945465,
       "learning_rate": 4.8825699988565485e-05,
-      "loss": 0.2058,
+      "loss": 0.205,
       "step": 671
     },
     {
       "epoch": 0.07195631223899775,
-      "grad_norm": 0.16594289243221283,
+      "grad_norm": 0.16252535581588745,
       "learning_rate": 4.8556046621849346e-05,
-      "loss": 0.2914,
+      "loss": 0.2912,
       "step": 672
     },
     {
       "epoch": 0.07206339008459149,
-      "grad_norm": 0.17473144829273224,
+      "grad_norm": 0.17688702046871185,
       "learning_rate": 4.828690098618429e-05,
       "loss": 0.3397,
       "step": 673
     },
     {
       "epoch": 0.07217046793018525,
-      "grad_norm": 0.1864292323589325,
+      "grad_norm": 0.1794818937778473,
       "learning_rate": 4.8018265737929044e-05,
-      "loss": 0.4704,
+      "loss": 0.4702,
       "step": 674
     },
     {
       "epoch": 0.072277545775779,
-      "grad_norm": 0.17810168862342834,
+      "grad_norm": 0.17864678800106049,
       "learning_rate": 4.7750143528405126e-05,
-      "loss": 0.3505,
+      "loss": 0.3496,
       "step": 675
     },
     {
       "epoch": 0.07238462362137274,
-      "grad_norm": 0.15457181632518768,
+      "grad_norm": 0.15607388317584991,
       "learning_rate": 4.748253700387042e-05,
-      "loss": 0.323,
+      "loss": 0.3233,
       "step": 676
     },
     {
       "epoch": 0.07249170146696648,
-      "grad_norm": 0.1830655038356781,
+      "grad_norm": 0.18187350034713745,
       "learning_rate": 4.721544880549337e-05,
-      "loss": 0.332,
+      "loss": 0.3302,
       "step": 677
     },
     {
       "epoch": 0.07259877931256023,
-      "grad_norm": 0.1661139875650406,
+      "grad_norm": 0.16444875299930573,
       "learning_rate": 4.694888156932658e-05,
-      "loss": 0.2828,
+      "loss": 0.283,
       "step": 678
     },
     {
       "epoch": 0.07270585715815397,
-      "grad_norm": 0.1575031727552414,
+      "grad_norm": 0.15640141069889069,
       "learning_rate": 4.668283792628114e-05,
-      "loss": 0.3252,
+      "loss": 0.3251,
       "step": 679
     },
     {
       "epoch": 0.07281293500374772,
-      "grad_norm": 0.1568552553653717,
+      "grad_norm": 0.16079206764698029,
       "learning_rate": 4.6417320502100316e-05,
-      "loss": 0.2818,
+      "loss": 0.2819,
       "step": 680
     },
     {
       "epoch": 0.07292001284934146,
-      "grad_norm": 0.18243847787380219,
+      "grad_norm": 0.18440526723861694,
       "learning_rate": 4.615233191733398e-05,
-      "loss": 0.3775,
+      "loss": 0.3787,
       "step": 681
     },
     {
       "epoch": 0.07302709069493522,
-      "grad_norm": 0.14313600957393646,
+      "grad_norm": 0.1407841593027115,
       "learning_rate": 4.588787478731242e-05,
-      "loss": 0.2801,
+      "loss": 0.2777,
       "step": 682
     },
     {
       "epoch": 0.07313416854052897,
-      "grad_norm": 0.2518500089645386,
+      "grad_norm": 0.18285134434700012,
       "learning_rate": 4.5623951722120736e-05,
-      "loss": 0.3921,
+      "loss": 0.3911,
       "step": 683
     },
     {
       "epoch": 0.07324124638612271,
-      "grad_norm": 0.1808672994375229,
+      "grad_norm": 0.18212471902370453,
       "learning_rate": 4.5360565326573104e-05,
-      "loss": 0.3468,
+      "loss": 0.3466,
       "step": 684
     },
     {
       "epoch": 0.07334832423171646,
-      "grad_norm": 0.1666547656059265,
+      "grad_norm": 0.16741225123405457,
       "learning_rate": 4.5097718200186814e-05,
-      "loss": 0.3815,
+      "loss": 0.382,
       "step": 685
     },
     {
       "epoch": 0.0734554020773102,
-      "grad_norm": 0.1575447916984558,
+      "grad_norm": 0.160224050283432,
       "learning_rate": 4.483541293715698e-05,
-      "loss": 0.3646,
+      "loss": 0.3652,
       "step": 686
     },
     {
       "epoch": 0.07356247992290395,
-      "grad_norm": 0.17928656935691833,
+      "grad_norm": 0.1788359433412552,
       "learning_rate": 4.457365212633058e-05,
-      "loss": 0.3603,
+      "loss": 0.3589,
       "step": 687
     },
     {
       "epoch": 0.07366955776849769,
-      "grad_norm": 0.19365298748016357,
+      "grad_norm": 0.19194957613945007,
       "learning_rate": 4.431243835118124e-05,
-      "loss": 0.367,
+      "loss": 0.366,
       "step": 688
     },
     {
       "epoch": 0.07377663561409144,
-      "grad_norm": 0.1505446434020996,
+      "grad_norm": 0.15268036723136902,
       "learning_rate": 4.4051774189783315e-05,
-      "loss": 0.2705,
+      "loss": 0.2694,
       "step": 689
     },
     {
       "epoch": 0.0738837134596852,
-      "grad_norm": 0.18016034364700317,
+      "grad_norm": 0.1825036257505417,
       "learning_rate": 4.379166221478697e-05,
-      "loss": 0.2904,
+      "loss": 0.291,
       "step": 690
     },
     {
       "epoch": 0.07399079130527894,
-      "grad_norm": 0.14219722151756287,
+      "grad_norm": 0.14247457683086395,
       "learning_rate": 4.3532104993392306e-05,
-      "loss": 0.2767,
+      "loss": 0.277,
       "step": 691
     },
     {
       "epoch": 0.07409786915087269,
-      "grad_norm": 0.2010519653558731,
+      "grad_norm": 0.19379214942455292,
       "learning_rate": 4.327310508732437e-05,
-      "loss": 0.4859,
+      "loss": 0.4844,
       "step": 692
     },
     {
       "epoch": 0.07420494699646643,
-      "grad_norm": 0.15180367231369019,
+      "grad_norm": 0.1550220251083374,
       "learning_rate": 4.301466505280762e-05,
-      "loss": 0.3175,
+      "loss": 0.3188,
       "step": 693
     },
     {
       "epoch": 0.07431202484206018,
-      "grad_norm": 0.16391538083553314,
+      "grad_norm": 0.16109517216682434,
       "learning_rate": 4.2756787440540936e-05,
-      "loss": 0.3576,
+      "loss": 0.3566,
       "step": 694
     },
     {
       "epoch": 0.07441910268765392,
-      "grad_norm": 0.1550375074148178,
+      "grad_norm": 0.15639358758926392,
       "learning_rate": 4.249947479567218e-05,
-      "loss": 0.3563,
+      "loss": 0.3562,
       "step": 695
     },
     {
       "epoch": 0.07452618053324767,
-      "grad_norm": 0.16695566475391388,
+      "grad_norm": 0.17039039731025696,
       "learning_rate": 4.224272965777326e-05,
-      "loss": 0.2697,
+      "loss": 0.2699,
       "step": 696
     },
     {
       "epoch": 0.07463325837884142,
-      "grad_norm": 0.16913089156150818,
+      "grad_norm": 0.1703154742717743,
       "learning_rate": 4.1986554560815096e-05,
-      "loss": 0.3506,
+      "loss": 0.3501,
       "step": 697
     },
     {
       "epoch": 0.07474033622443517,
-      "grad_norm": 0.15745219588279724,
+      "grad_norm": 0.1571662575006485,
       "learning_rate": 4.173095203314241e-05,
-      "loss": 0.311,
+      "loss": 0.3103,
       "step": 698
     },
     {
       "epoch": 0.07484741407002891,
-      "grad_norm": 0.1870749592781067,
+      "grad_norm": 0.18659600615501404,
       "learning_rate": 4.1475924597449024e-05,
-      "loss": 0.3289,
+      "loss": 0.3283,
       "step": 699
     },
     {
       "epoch": 0.07495449191562266,
-      "grad_norm": 0.18129318952560425,
+      "grad_norm": 0.18269579112529755,
       "learning_rate": 4.12214747707527e-05,
-      "loss": 0.3735,
+      "loss": 0.3737,
       "step": 700
     },
     {
       "epoch": 0.0750615697612164,
-      "grad_norm": 0.1702040284872055,
+      "grad_norm": 0.17083945870399475,
       "learning_rate": 4.096760506437057e-05,
-      "loss": 0.3375,
+      "loss": 0.338,
       "step": 701
     },
     {
       "epoch": 0.07516864760681015,
-      "grad_norm": 0.16838037967681885,
+      "grad_norm": 0.16920264065265656,
       "learning_rate": 4.071431798389408e-05,
-      "loss": 0.3685,
+      "loss": 0.3694,
       "step": 702
     },
     {
       "epoch": 0.0752757254524039,
-      "grad_norm": 0.22011017799377441,
+      "grad_norm": 0.22974881529808044,
       "learning_rate": 4.0461616029164526e-05,
-      "loss": 0.3469,
+      "loss": 0.3473,
       "step": 703
     },
     {
       "epoch": 0.07538280329799764,
-      "grad_norm": 0.14272677898406982,
+      "grad_norm": 0.1423478126525879,
       "learning_rate": 4.020950169424815e-05,
-      "loss": 0.3497,
+      "loss": 0.3492,
       "step": 704
     },
     {
       "epoch": 0.0754898811435914,
-      "grad_norm": 0.15662071108818054,
+      "grad_norm": 0.15745437145233154,
       "learning_rate": 3.9957977467411615e-05,
-      "loss": 0.2359,
+      "loss": 0.2361,
       "step": 705
     },
     {
       "epoch": 0.07559695898918514,
-      "grad_norm": 0.12998908758163452,
+      "grad_norm": 0.13066445291042328,
       "learning_rate": 3.9707045831097555e-05,
-      "loss": 0.2384,
+      "loss": 0.2374,
       "step": 706
     },
     {
       "epoch": 0.07570403683477889,
-      "grad_norm": 0.18728865683078766,
+      "grad_norm": 0.18829156458377838,
       "learning_rate": 3.945670926189987e-05,
-      "loss": 0.3876,
+      "loss": 0.3874,
       "step": 707
     },
     {
       "epoch": 0.07581111468037263,
-      "grad_norm": 0.19201086461544037,
+      "grad_norm": 0.19448167085647583,
       "learning_rate": 3.920697023053949e-05,
-      "loss": 0.4129,
+      "loss": 0.4145,
       "step": 708
     },
     {
       "epoch": 0.07591819252596638,
-      "grad_norm": 0.15770968794822693,
+      "grad_norm": 0.15791544318199158,
       "learning_rate": 3.895783120183976e-05,
-      "loss": 0.3428,
+      "loss": 0.3423,
       "step": 709
     },
     {
       "epoch": 0.07602527037156012,
-      "grad_norm": 0.16312432289123535,
+      "grad_norm": 0.16304408013820648,
       "learning_rate": 3.8709294634702376e-05,
-      "loss": 0.3468,
+      "loss": 0.3469,
       "step": 710
     },
     {
       "epoch": 0.07613234821715387,
-      "grad_norm": 0.15018045902252197,
+      "grad_norm": 0.15038977563381195,
       "learning_rate": 3.846136298208285e-05,
-      "loss": 0.2924,
+      "loss": 0.2923,
       "step": 711
     },
     {
       "epoch": 0.07623942606274761,
-      "grad_norm": 0.150425523519516,
+      "grad_norm": 0.1510382443666458,
       "learning_rate": 3.821403869096658e-05,
-      "loss": 0.2774,
+      "loss": 0.2773,
       "step": 712
     },
     {
       "epoch": 0.07634650390834137,
-      "grad_norm": 0.2021748274564743,
+      "grad_norm": 0.2000676840543747,
       "learning_rate": 3.796732420234443e-05,
-      "loss": 0.3377,
+      "loss": 0.3363,
       "step": 713
     },
     {
       "epoch": 0.07645358175393512,
-      "grad_norm": 0.18799428641796112,
+      "grad_norm": 0.1835888922214508,
       "learning_rate": 3.7721221951188765e-05,
-      "loss": 0.3719,
+      "loss": 0.3695,
       "step": 714
     },
     {
       "epoch": 0.07656065959952886,
-      "grad_norm": 0.1765313595533371,
+      "grad_norm": 0.1762959063053131,
       "learning_rate": 3.747573436642951e-05,
-      "loss": 0.3921,
+      "loss": 0.3929,
       "step": 715
     },
     {
       "epoch": 0.0766677374451226,
-      "grad_norm": 0.15740084648132324,
+      "grad_norm": 0.15682466328144073,
       "learning_rate": 3.7230863870929964e-05,
-      "loss": 0.3102,
+      "loss": 0.3103,
       "step": 716
     },
     {
       "epoch": 0.07677481529071635,
-      "grad_norm": 0.17501920461654663,
+      "grad_norm": 0.17416490614414215,
       "learning_rate": 3.698661288146311e-05,
-      "loss": 0.4363,
+      "loss": 0.4379,
       "step": 717
     },
     {
       "epoch": 0.0768818931363101,
-      "grad_norm": 0.2245578169822693,
+      "grad_norm": 0.23010078072547913,
       "learning_rate": 3.674298380868756e-05,
-      "loss": 0.5983,
+      "loss": 0.5982,
       "step": 718
     },
     {
       "epoch": 0.07698897098190384,
-      "grad_norm": 0.18928749859333038,
+      "grad_norm": 0.1550900787115097,
       "learning_rate": 3.649997905712396e-05,
-      "loss": 0.2717,
+      "loss": 0.2699,
       "step": 719
     },
     {
       "epoch": 0.07709604882749758,
-      "grad_norm": 0.14683455228805542,
+      "grad_norm": 0.14471879601478577,
       "learning_rate": 3.6257601025131026e-05,
-      "loss": 0.2873,
+      "loss": 0.2861,
       "step": 720
     },
     {
       "epoch": 0.07720312667309134,
-      "grad_norm": 0.19283850491046906,
+      "grad_norm": 0.18543073534965515,
       "learning_rate": 3.601585210488218e-05,
-      "loss": 0.2882,
+      "loss": 0.288,
       "step": 721
     },
     {
       "epoch": 0.07731020451868509,
-      "grad_norm": 0.21515490114688873,
+      "grad_norm": 0.21810288727283478,
       "learning_rate": 3.577473468234156e-05,
-      "loss": 0.3484,
+      "loss": 0.3492,
       "step": 722
     },
     {
       "epoch": 0.07741728236427883,
-      "grad_norm": 0.16853037476539612,
+      "grad_norm": 0.18458350002765656,
       "learning_rate": 3.553425113724088e-05,
-      "loss": 0.3759,
+      "loss": 0.3745,
       "step": 723
     },
     {
       "epoch": 0.07752436020987258,
-      "grad_norm": 0.1734563559293747,
+      "grad_norm": 0.17252208292484283,
       "learning_rate": 3.52944038430556e-05,
-      "loss": 0.3769,
+      "loss": 0.376,
       "step": 724
     },
     {
       "epoch": 0.07763143805546632,
-      "grad_norm": 0.1489877551794052,
+      "grad_norm": 0.15418677031993866,
       "learning_rate": 3.5055195166981645e-05,
-      "loss": 0.3306,
+      "loss": 0.3308,
       "step": 725
     },
     {
       "epoch": 0.07773851590106007,
-      "grad_norm": 0.1704966127872467,
+      "grad_norm": 0.17200735211372375,
       "learning_rate": 3.481662746991214e-05,
-      "loss": 0.3635,
+      "loss": 0.3643,
       "step": 726
     },
     {
       "epoch": 0.07784559374665381,
-      "grad_norm": 0.13263505697250366,
+      "grad_norm": 0.13179121911525726,
       "learning_rate": 3.4578703106413904e-05,
-      "loss": 0.2811,
+      "loss": 0.2808,
       "step": 727
     },
     {
       "epoch": 0.07795267159224756,
-      "grad_norm": 0.1653466671705246,
+      "grad_norm": 0.16644497215747833,
       "learning_rate": 3.4341424424704375e-05,
-      "loss": 0.301,
+      "loss": 0.3007,
       "step": 728
     },
     {
       "epoch": 0.07805974943784132,
-      "grad_norm": 0.17807163298130035,
+      "grad_norm": 0.17268535494804382,
       "learning_rate": 3.4104793766628304e-05,
-      "loss": 0.3877,
+      "loss": 0.3881,
       "step": 729
     },
     {
       "epoch": 0.07816682728343506,
-      "grad_norm": 0.1522621512413025,
+      "grad_norm": 0.15279394388198853,
       "learning_rate": 3.386881346763483e-05,
-      "loss": 0.2858,
+      "loss": 0.2837,
       "step": 730
     },
     {
       "epoch": 0.0782739051290288,
-      "grad_norm": 0.15765105187892914,
+      "grad_norm": 0.155588760972023,
       "learning_rate": 3.363348585675414e-05,
-      "loss": 0.3006,
+      "loss": 0.2995,
       "step": 731
     },
     {
       "epoch": 0.07838098297462255,
-      "grad_norm": 0.16139338910579681,
+      "grad_norm": 0.1607223004102707,
       "learning_rate": 3.339881325657484e-05,
-      "loss": 0.3796,
+      "loss": 0.3795,
       "step": 732
     },
     {
       "epoch": 0.0784880608202163,
-      "grad_norm": 0.15585535764694214,
+      "grad_norm": 0.15511462092399597,
       "learning_rate": 3.316479798322072e-05,
-      "loss": 0.3781,
+      "loss": 0.3791,
       "step": 733
     },
     {
       "epoch": 0.07859513866581004,
-      "grad_norm": 0.14251768589019775,
+      "grad_norm": 0.13938754796981812,
       "learning_rate": 3.2931442346328004e-05,
-      "loss": 0.2142,
+      "loss": 0.2123,
       "step": 734
     },
     {
       "epoch": 0.07870221651140379,
-      "grad_norm": 0.14421778917312622,
+      "grad_norm": 0.18411380052566528,
       "learning_rate": 3.269874864902269e-05,
-      "loss": 0.282,
+      "loss": 0.2817,
       "step": 735
     },
     {
       "epoch": 0.07880929435699753,
-      "grad_norm": 0.1328769028186798,
+      "grad_norm": 0.13664402067661285,
       "learning_rate": 3.246671918789755e-05,
-      "loss": 0.2556,
+      "loss": 0.2564,
       "step": 736
     },
     {
       "epoch": 0.07891637220259129,
-      "grad_norm": 0.1606833040714264,
+      "grad_norm": 0.1579146832227707,
       "learning_rate": 3.223535625298979e-05,
-      "loss": 0.2804,
+      "loss": 0.2813,
       "step": 737
     },
     {
       "epoch": 0.07902345004818503,
-      "grad_norm": 0.1721002757549286,
+      "grad_norm": 0.17424800992012024,
       "learning_rate": 3.200466212775808e-05,
-      "loss": 0.2562,
+      "loss": 0.2554,
       "step": 738
     },
     {
       "epoch": 0.07913052789377878,
-      "grad_norm": 0.1502068191766739,
+      "grad_norm": 0.15112292766571045,
       "learning_rate": 3.1774639089060363e-05,
-      "loss": 0.319,
+      "loss": 0.3183,
       "step": 739
     },
     {
       "epoch": 0.07923760573937252,
-      "grad_norm": 0.20539775490760803,
+      "grad_norm": 0.2046094685792923,
       "learning_rate": 3.154528940713113e-05,
-      "loss": 0.3627,
+      "loss": 0.3629,
       "step": 740
     },
     {
       "epoch": 0.07934468358496627,
-      "grad_norm": 0.20355840027332306,
+      "grad_norm": 0.22456932067871094,
       "learning_rate": 3.1316615345559185e-05,
-      "loss": 0.3651,
+      "loss": 0.363,
       "step": 741
     },
     {
       "epoch": 0.07945176143056001,
-      "grad_norm": 0.18047836422920227,
+      "grad_norm": 0.17923475801944733,
       "learning_rate": 3.108861916126518e-05,
-      "loss": 0.4033,
+      "loss": 0.4022,
       "step": 742
     },
     {
       "epoch": 0.07955883927615376,
-      "grad_norm": 0.17290200293064117,
+      "grad_norm": 0.18033398687839508,
       "learning_rate": 3.086130310447937e-05,
-      "loss": 0.3729,
+      "loss": 0.3728,
       "step": 743
     },
     {
       "epoch": 0.0796659171217475,
-      "grad_norm": 0.1400056779384613,
+      "grad_norm": 0.14223916828632355,
       "learning_rate": 3.063466941871952e-05,
-      "loss": 0.2791,
+      "loss": 0.2802,
       "step": 744
     },
     {
       "epoch": 0.07977299496734126,
-      "grad_norm": 0.15376096963882446,
+      "grad_norm": 0.156300351023674,
       "learning_rate": 3.0408720340768572e-05,
-      "loss": 0.3535,
+      "loss": 0.3538,
       "step": 745
     },
     {
       "epoch": 0.07988007281293501,
-      "grad_norm": 0.13626393675804138,
+      "grad_norm": 0.13515561819076538,
       "learning_rate": 3.018345810065275e-05,
-      "loss": 0.2482,
+      "loss": 0.2488,
       "step": 746
     },
     {
       "epoch": 0.07998715065852875,
-      "grad_norm": 0.17024940252304077,
+      "grad_norm": 0.16864290833473206,
       "learning_rate": 2.9958884921619367e-05,
-      "loss": 0.3593,
+      "loss": 0.358,
       "step": 747
     },
     {
       "epoch": 0.0800942285041225,
-      "grad_norm": 0.20395411550998688,
+      "grad_norm": 0.20047788321971893,
       "learning_rate": 2.9735003020115092e-05,
       "loss": 0.4827,
       "step": 748
     },
     {
       "epoch": 0.08020130634971624,
-      "grad_norm": 0.18541432917118073,
+      "grad_norm": 0.18187908828258514,
       "learning_rate": 2.9511814605763855e-05,
-      "loss": 0.3223,
+      "loss": 0.322,
       "step": 749
     },
     {
       "epoch": 0.08030838419530999,
-      "grad_norm": 0.1652992069721222,
+      "grad_norm": 0.16408580541610718,
       "learning_rate": 2.9289321881345254e-05,
-      "loss": 0.3128,
+      "loss": 0.3125,
       "step": 750
     },
     {
       "epoch": 0.08041546204090373,
-      "grad_norm": 0.19549131393432617,
+      "grad_norm": 0.19623160362243652,
       "learning_rate": 2.9067527042772636e-05,
-      "loss": 0.4826,
+      "loss": 0.4812,
       "step": 751
     },
     {
       "epoch": 0.08052253988649748,
-      "grad_norm": 0.15394878387451172,
+      "grad_norm": 0.1534736603498459,
       "learning_rate": 2.8846432279071467e-05,
-      "loss": 0.3783,
+      "loss": 0.3785,
       "step": 752
     },
     {
       "epoch": 0.08062961773209124,
-      "grad_norm": 0.15814286470413208,
+      "grad_norm": 0.1560440957546234,
       "learning_rate": 2.8626039772357882e-05,
-      "loss": 0.324,
+      "loss": 0.3226,
       "step": 753
     },
     {
       "epoch": 0.08073669557768498,
-      "grad_norm": 0.21982432901859283,
+      "grad_norm": 0.21636521816253662,
       "learning_rate": 2.840635169781688e-05,
-      "loss": 0.45,
+      "loss": 0.4498,
       "step": 754
     },
     {
       "epoch": 0.08084377342327873,
-      "grad_norm": 0.16086405515670776,
+      "grad_norm": 0.1631963849067688,
       "learning_rate": 2.8187370223681132e-05,
-      "loss": 0.3717,
+      "loss": 0.3713,
       "step": 755
     },
     {
       "epoch": 0.08095085126887247,
-      "grad_norm": 0.13305172324180603,
+      "grad_norm": 0.13680961728096008,
       "learning_rate": 2.7969097511209308e-05,
-      "loss": 0.2966,
+      "loss": 0.2974,
       "step": 756
     },
     {
       "epoch": 0.08105792911446622,
-      "grad_norm": 0.1707354336977005,
+      "grad_norm": 0.17071770131587982,
       "learning_rate": 2.775153571466502e-05,
-      "loss": 0.2911,
+      "loss": 0.2913,
       "step": 757
     },
     {
       "epoch": 0.08116500696005996,
-      "grad_norm": 0.14525282382965088,
+      "grad_norm": 0.14643071591854095,
       "learning_rate": 2.753468698129533e-05,
-      "loss": 0.2842,
+      "loss": 0.2833,
       "step": 758
     },
     {
       "epoch": 0.0812720848056537,
-      "grad_norm": 0.17985416948795319,
+      "grad_norm": 0.1774536669254303,
       "learning_rate": 2.7318553451309726e-05,
-      "loss": 0.3126,
+      "loss": 0.3131,
       "step": 759
     },
     {
       "epoch": 0.08137916265124746,
-      "grad_norm": 0.16003213822841644,
+      "grad_norm": 0.1635051816701889,
       "learning_rate": 2.7103137257858868e-05,
-      "loss": 0.3547,
+      "loss": 0.354,
       "step": 760
     },
     {
       "epoch": 0.08148624049684121,
-      "grad_norm": 0.1999395340681076,
+      "grad_norm": 0.20249997079372406,
       "learning_rate": 2.688844052701359e-05,
-      "loss": 0.348,
+      "loss": 0.3487,
       "step": 761
     },
     {
       "epoch": 0.08159331834243495,
-      "grad_norm": 0.14586856961250305,
+      "grad_norm": 0.14786763489246368,
       "learning_rate": 2.6674465377744017e-05,
-      "loss": 0.3054,
+      "loss": 0.3071,
       "step": 762
     },
     {
       "epoch": 0.0817003961880287,
-      "grad_norm": 0.1372138261795044,
+      "grad_norm": 0.13844721019268036,
       "learning_rate": 2.646121392189841e-05,
-      "loss": 0.2372,
+      "loss": 0.2367,
       "step": 763
     },
     {
       "epoch": 0.08180747403362244,
-      "grad_norm": 0.15203608572483063,
+      "grad_norm": 0.1526380181312561,
       "learning_rate": 2.624868826418262e-05,
-      "loss": 0.2984,
+      "loss": 0.2976,
       "step": 764
     },
     {
       "epoch": 0.08191455187921619,
-      "grad_norm": 0.18717195093631744,
+      "grad_norm": 0.1874617487192154,
       "learning_rate": 2.603689050213902e-05,
-      "loss": 0.2993,
+      "loss": 0.2991,
       "step": 765
     },
     {
       "epoch": 0.08202162972480993,
-      "grad_norm": 0.17310731112957,
+      "grad_norm": 0.1745980829000473,
       "learning_rate": 2.582582272612609e-05,
-      "loss": 0.2612,
+      "loss": 0.2624,
       "step": 766
     },
     {
       "epoch": 0.08212870757040368,
-      "grad_norm": 0.1539638340473175,
+      "grad_norm": 0.15471521019935608,
       "learning_rate": 2.561548701929749e-05,
-      "loss": 0.3713,
+      "loss": 0.3706,
       "step": 767
     },
     {
       "epoch": 0.08223578541599744,
-      "grad_norm": 0.16853591799736023,
+      "grad_norm": 0.16713665425777435,
       "learning_rate": 2.540588545758179e-05,
-      "loss": 0.2724,
+      "loss": 0.2719,
       "step": 768
     },
     {
       "epoch": 0.08234286326159118,
-      "grad_norm": 0.1422484815120697,
+      "grad_norm": 0.14261217415332794,
       "learning_rate": 2.5197020109661772e-05,
-      "loss": 0.273,
+      "loss": 0.2737,
       "step": 769
     },
     {
       "epoch": 0.08244994110718493,
-      "grad_norm": 0.16871221363544464,
+      "grad_norm": 0.16853374242782593,
       "learning_rate": 2.4988893036954043e-05,
-      "loss": 0.3818,
+      "loss": 0.3817,
       "step": 770
     },
     {
       "epoch": 0.08255701895277867,
-      "grad_norm": 0.17967580258846283,
+      "grad_norm": 0.17833545804023743,
       "learning_rate": 2.4781506293588873e-05,
-      "loss": 0.3771,
+      "loss": 0.3775,
       "step": 771
     },
     {
       "epoch": 0.08266409679837242,
-      "grad_norm": 0.18379069864749908,
+      "grad_norm": 0.18436114490032196,
       "learning_rate": 2.4574861926389615e-05,
-      "loss": 0.2425,
+      "loss": 0.244,
       "step": 772
     },
     {
       "epoch": 0.08277117464396616,
-      "grad_norm": 0.17387156188488007,
+      "grad_norm": 0.17707830667495728,
       "learning_rate": 2.436896197485282e-05,
       "loss": 0.4478,
       "step": 773
     },
     {
       "epoch": 0.0828782524895599,
-      "grad_norm": 0.19421961903572083,
+      "grad_norm": 0.19500042498111725,
       "learning_rate": 2.4163808471127812e-05,
-      "loss": 0.3658,
+      "loss": 0.3646,
       "step": 774
     },
     {
       "epoch": 0.08298533033515365,
-      "grad_norm": 0.14912578463554382,
+      "grad_norm": 0.1503261774778366,
       "learning_rate": 2.3959403439996907e-05,
-      "loss": 0.338,
+      "loss": 0.3389,
       "step": 775
     },
     {
       "epoch": 0.08309240818074741,
-      "grad_norm": 0.17342334985733032,
+      "grad_norm": 0.1736772656440735,
       "learning_rate": 2.37557488988552e-05,
-      "loss": 0.3367,
+      "loss": 0.3369,
       "step": 776
     },
     {
       "epoch": 0.08319948602634115,
-      "grad_norm": 0.1985640674829483,
+      "grad_norm": 0.1980380415916443,
       "learning_rate": 2.3552846857690846e-05,
-      "loss": 0.3553,
+      "loss": 0.3544,
       "step": 777
     },
     {
       "epoch": 0.0833065638719349,
-      "grad_norm": 0.16898934543132782,
+      "grad_norm": 0.16803400218486786,
       "learning_rate": 2.3350699319065026e-05,
-      "loss": 0.3761,
+      "loss": 0.375,
       "step": 778
     },
     {
       "epoch": 0.08341364171752864,
-      "grad_norm": 0.17243239283561707,
+      "grad_norm": 0.17213231325149536,
       "learning_rate": 2.3149308278092342e-05,
       "loss": 0.3151,
       "step": 779
     },
     {
       "epoch": 0.08352071956312239,
-      "grad_norm": 0.19134309887886047,
+      "grad_norm": 0.18893983960151672,
       "learning_rate": 2.2948675722421086e-05,
-      "loss": 0.3636,
+      "loss": 0.3629,
       "step": 780
     },
     {
       "epoch": 0.08362779740871613,
-      "grad_norm": 0.20325163006782532,
+      "grad_norm": 0.20004568994045258,
       "learning_rate": 2.2748803632213557e-05,
-      "loss": 0.3778,
+      "loss": 0.3777,
       "step": 781
     },
     {
       "epoch": 0.08373487525430988,
-      "grad_norm": 0.14077632129192352,
+      "grad_norm": 0.1406610906124115,
       "learning_rate": 2.254969398012663e-05,
-      "loss": 0.2707,
+      "loss": 0.2712,
       "step": 782
     },
     {
       "epoch": 0.08384195309990362,
-      "grad_norm": 0.16174592077732086,
+      "grad_norm": 0.1625978797674179,
       "learning_rate": 2.235134873129213e-05,
-      "loss": 0.3096,
+      "loss": 0.3094,
       "step": 783
     },
     {
       "epoch": 0.08394903094549738,
-      "grad_norm": 0.1870279163122177,
+      "grad_norm": 0.1865498125553131,
       "learning_rate": 2.2153769843297667e-05,
-      "loss": 0.3855,
+      "loss": 0.3852,
       "step": 784
     },
     {
       "epoch": 0.08405610879109113,
-      "grad_norm": 0.1512727290391922,
+      "grad_norm": 0.16556480526924133,
       "learning_rate": 2.195695926616702e-05,
-      "loss": 0.2219,
+      "loss": 0.2227,
       "step": 785
     },
     {
       "epoch": 0.08416318663668487,
-      "grad_norm": 0.18673193454742432,
+      "grad_norm": 0.1859363615512848,
       "learning_rate": 2.1760918942341192e-05,
-      "loss": 0.4747,
+      "loss": 0.4757,
       "step": 786
     },
     {
       "epoch": 0.08427026448227862,
-      "grad_norm": 0.17171938717365265,
+      "grad_norm": 0.17114213109016418,
       "learning_rate": 2.1565650806658975e-05,
-      "loss": 0.3874,
+      "loss": 0.3861,
       "step": 787
     },
     {
       "epoch": 0.08437734232787236,
-      "grad_norm": 0.1813274323940277,
+      "grad_norm": 0.18057489395141602,
       "learning_rate": 2.137115678633811e-05,
-      "loss": 0.375,
+      "loss": 0.3755,
       "step": 788
     },
     {
       "epoch": 0.08448442017346611,
-      "grad_norm": 0.1640833467245102,
+      "grad_norm": 0.16163988411426544,
       "learning_rate": 2.1177438800956007e-05,
       "loss": 0.3012,
       "step": 789
     },
     {
       "epoch": 0.08459149801905985,
-      "grad_norm": 0.1751677691936493,
+      "grad_norm": 0.17569121718406677,
       "learning_rate": 2.098449876243096e-05,
-      "loss": 0.3478,
+      "loss": 0.3466,
       "step": 790
     },
     {
       "epoch": 0.0846985758646536,
-      "grad_norm": 0.14867408573627472,
+      "grad_norm": 0.14997471868991852,
       "learning_rate": 2.07923385750033e-05,
-      "loss": 0.2817,
+      "loss": 0.2821,
       "step": 791
     },
     {
       "epoch": 0.08480565371024736,
-      "grad_norm": 0.1769115924835205,
+      "grad_norm": 0.17677253484725952,
       "learning_rate": 2.0600960135216462e-05,
-      "loss": 0.3446,
+      "loss": 0.3436,
       "step": 792
     },
     {
       "epoch": 0.0849127315558411,
-      "grad_norm": 0.1462964117527008,
+      "grad_norm": 0.14548727869987488,
       "learning_rate": 2.0410365331898416e-05,
-      "loss": 0.2856,
+      "loss": 0.2875,
       "step": 793
     },
     {
       "epoch": 0.08501980940143485,
-      "grad_norm": 0.1861981898546219,
+      "grad_norm": 0.18589341640472412,
       "learning_rate": 2.0220556046142893e-05,
-      "loss": 0.3894,
+      "loss": 0.3885,
       "step": 794
     },
     {
       "epoch": 0.08512688724702859,
-      "grad_norm": 0.15652887523174286,
+      "grad_norm": 0.16286922991275787,
       "learning_rate": 2.0031534151290943e-05,
-      "loss": 0.245,
+      "loss": 0.2452,
       "step": 795
     },
     {
       "epoch": 0.08523396509262234,
-      "grad_norm": 0.14712314307689667,
+      "grad_norm": 0.14786767959594727,
       "learning_rate": 1.9843301512912327e-05,
-      "loss": 0.3458,
+      "loss": 0.3465,
       "step": 796
     },
     {
       "epoch": 0.08534104293821608,
-      "grad_norm": 0.15813574194908142,
+      "grad_norm": 0.15878774225711823,
       "learning_rate": 1.965585998878724e-05,
-      "loss": 0.3723,
+      "loss": 0.3714,
       "step": 797
     },
     {
       "epoch": 0.08544812078380983,
-      "grad_norm": 0.1604684740304947,
+      "grad_norm": 0.16029423475265503,
       "learning_rate": 1.946921142888781e-05,
-      "loss": 0.3116,
+      "loss": 0.3118,
       "step": 798
     },
     {
       "epoch": 0.08555519862940357,
-      "grad_norm": 0.14358116686344147,
+      "grad_norm": 0.11456546932458878,
       "learning_rate": 1.928335767535997e-05,
-      "loss": 0.2253,
+      "loss": 0.2231,
       "step": 799
     },
     {
       "epoch": 0.08566227647499733,
-      "grad_norm": 0.17999866604804993,
+      "grad_norm": 0.17691242694854736,
       "learning_rate": 1.9098300562505266e-05,
-      "loss": 0.3825,
+      "loss": 0.3831,
       "step": 800
     },
     {
       "epoch": 0.08576935432059107,
-      "grad_norm": 0.1776675432920456,
+      "grad_norm": 0.178320974111557,
       "learning_rate": 1.891404191676265e-05,
-      "loss": 0.3816,
+      "loss": 0.3814,
       "step": 801
     },
     {
       "epoch": 0.08587643216618482,
-      "grad_norm": 0.16066551208496094,
+      "grad_norm": 0.1586633324623108,
       "learning_rate": 1.8730583556690605e-05,
-      "loss": 0.2955,
+      "loss": 0.2953,
       "step": 802
     },
     {
       "epoch": 0.08598351001177856,
-      "grad_norm": 0.1738562434911728,
+      "grad_norm": 0.1737189143896103,
       "learning_rate": 1.854792729294905e-05,
-      "loss": 0.1947,
+      "loss": 0.1958,
       "step": 803
     },
     {
       "epoch": 0.08609058785737231,
-      "grad_norm": 0.18131086230278015,
+      "grad_norm": 0.17950400710105896,
       "learning_rate": 1.8366074928281607e-05,
-      "loss": 0.2306,
+      "loss": 0.2301,
       "step": 804
     },
     {
       "epoch": 0.08619766570296605,
-      "grad_norm": 0.17308153212070465,
+      "grad_norm": 0.1717943549156189,
       "learning_rate": 1.818502825749764e-05,
-      "loss": 0.3446,
+      "loss": 0.3452,
       "step": 805
     },
     {
       "epoch": 0.0863047435485598,
-      "grad_norm": 0.1967420130968094,
+      "grad_norm": 0.19653019309043884,
       "learning_rate": 1.8004789067454764e-05,
-      "loss": 0.3794,
+      "loss": 0.3789,
       "step": 806
     },
     {
       "epoch": 0.08641182139415354,
-      "grad_norm": 0.17925085127353668,
+      "grad_norm": 0.23031754791736603,
       "learning_rate": 1.7825359137040988e-05,
-      "loss": 0.3216,
+      "loss": 0.3241,
       "step": 807
     },
     {
       "epoch": 0.0865188992397473,
-      "grad_norm": 0.19170551002025604,
+      "grad_norm": 0.19152677059173584,
       "learning_rate": 1.7646740237157256e-05,
-      "loss": 0.3629,
+      "loss": 0.3634,
       "step": 808
     },
     {
       "epoch": 0.08662597708534105,
-      "grad_norm": 0.16575735807418823,
+      "grad_norm": 0.14082860946655273,
       "learning_rate": 1.7468934130700044e-05,
-      "loss": 0.204,
+      "loss": 0.2035,
       "step": 809
     },
     {
       "epoch": 0.08673305493093479,
-      "grad_norm": 0.15411527454853058,
+      "grad_norm": 0.15468376874923706,
       "learning_rate": 1.7291942572543807e-05,
-      "loss": 0.2983,
+      "loss": 0.299,
       "step": 810
     },
     {
       "epoch": 0.08684013277652854,
-      "grad_norm": 0.1592104583978653,
+      "grad_norm": 0.1572645902633667,
       "learning_rate": 1.7115767309523812e-05,
-      "loss": 0.3134,
+      "loss": 0.3126,
       "step": 811
     },
     {
       "epoch": 0.08694721062212228,
-      "grad_norm": 0.19162140786647797,
+      "grad_norm": 0.19454143941402435,
       "learning_rate": 1.6940410080418723e-05,
-      "loss": 0.4155,
+      "loss": 0.4161,
       "step": 812
     },
     {
       "epoch": 0.08705428846771603,
-      "grad_norm": 0.13956181704998016,
+      "grad_norm": 0.14077642560005188,
       "learning_rate": 1.6765872615933677e-05,
-      "loss": 0.248,
+      "loss": 0.249,
       "step": 813
     },
     {
       "epoch": 0.08716136631330977,
-      "grad_norm": 0.22904083132743835,
+      "grad_norm": 0.22318896651268005,
       "learning_rate": 1.6592156638682886e-05,
-      "loss": 0.3456,
+      "loss": 0.3444,
       "step": 814
     },
     {
       "epoch": 0.08726844415890353,
-      "grad_norm": 0.15575331449508667,
+      "grad_norm": 0.15059834718704224,
       "learning_rate": 1.6419263863172997e-05,
-      "loss": 0.2599,
+      "loss": 0.2598,
       "step": 815
     },
     {
       "epoch": 0.08737552200449727,
-      "grad_norm": 0.17155703902244568,
+      "grad_norm": 0.1700175404548645,
       "learning_rate": 1.6247195995785837e-05,
-      "loss": 0.3706,
+      "loss": 0.372,
       "step": 816
     },
     {
       "epoch": 0.08748259985009102,
-      "grad_norm": 0.1934448927640915,
+      "grad_norm": 0.19216172397136688,
       "learning_rate": 1.6075954734761845e-05,
       "loss": 0.3689,
       "step": 817
     },
     {
       "epoch": 0.08758967769568476,
-      "grad_norm": 0.14852175116539001,
+      "grad_norm": 0.1473630666732788,
       "learning_rate": 1.5905541770183096e-05,
-      "loss": 0.257,
+      "loss": 0.2568,
       "step": 818
     },
     {
       "epoch": 0.08769675554127851,
-      "grad_norm": 0.17280475795269012,
+      "grad_norm": 0.17223362624645233,
       "learning_rate": 1.5735958783956794e-05,
-      "loss": 0.3368,
+      "loss": 0.3367,
       "step": 819
     },
     {
       "epoch": 0.08780383338687225,
-      "grad_norm": 0.18640533089637756,
+      "grad_norm": 0.18466192483901978,
       "learning_rate": 1.5567207449798515e-05,
-      "loss": 0.3615,
+      "loss": 0.3607,
       "step": 820
     },
     {
       "epoch": 0.087910911232466,
-      "grad_norm": 0.19203537702560425,
+      "grad_norm": 0.1918594390153885,
       "learning_rate": 1.539928943321579e-05,
-      "loss": 0.4128,
+      "loss": 0.4122,
       "step": 821
     },
     {
       "epoch": 0.08801798907805974,
-      "grad_norm": 0.1681956797838211,
+      "grad_norm": 0.17056573927402496,
       "learning_rate": 1.5232206391491699e-05,
-      "loss": 0.3007,
+      "loss": 0.2995,
       "step": 822
     },
     {
       "epoch": 0.0881250669236535,
-      "grad_norm": 0.167117178440094,
+      "grad_norm": 0.16323423385620117,
       "learning_rate": 1.5065959973668353e-05,
-      "loss": 0.3348,
+      "loss": 0.3332,
       "step": 823
     },
     {
       "epoch": 0.08823214476924725,
-      "grad_norm": 0.145122230052948,
+      "grad_norm": 0.14421838521957397,
       "learning_rate": 1.4900551820530828e-05,
-      "loss": 0.3176,
+      "loss": 0.3181,
       "step": 824
     },
     {
       "epoch": 0.08833922261484099,
-      "grad_norm": 0.2096463143825531,
+      "grad_norm": 0.21069642901420593,
       "learning_rate": 1.4735983564590783e-05,
-      "loss": 0.4623,
+      "loss": 0.4617,
       "step": 825
     },
     {
       "epoch": 0.08844630046043474,
-      "grad_norm": 0.17497724294662476,
+      "grad_norm": 0.16968613862991333,
       "learning_rate": 1.4572256830070497e-05,
-      "loss": 0.4211,
+      "loss": 0.4204,
       "step": 826
     },
     {
       "epoch": 0.08855337830602848,
-      "grad_norm": 0.16783024370670319,
+      "grad_norm": 0.16650810837745667,
       "learning_rate": 1.4409373232886702e-05,
-      "loss": 0.3312,
+      "loss": 0.3315,
       "step": 827
     },
     {
       "epoch": 0.08866045615162223,
-      "grad_norm": 0.17440426349639893,
+      "grad_norm": 0.1736612319946289,
       "learning_rate": 1.4247334380634792e-05,
-      "loss": 0.3737,
+      "loss": 0.3739,
       "step": 828
     },
     {
       "epoch": 0.08876753399721597,
-      "grad_norm": 0.25179216265678406,
+      "grad_norm": 0.24995604157447815,
       "learning_rate": 1.4086141872572789e-05,
-      "loss": 0.4219,
+      "loss": 0.4213,
       "step": 829
     },
     {
       "epoch": 0.08887461184280972,
-      "grad_norm": 0.15539956092834473,
+      "grad_norm": 0.1580914556980133,
       "learning_rate": 1.3925797299605647e-05,
-      "loss": 0.2873,
+      "loss": 0.2874,
       "step": 830
     },
     {
       "epoch": 0.08898168968840348,
-      "grad_norm": 0.19273389875888824,
+      "grad_norm": 0.19272121787071228,
       "learning_rate": 1.3766302244269624e-05,
-      "loss": 0.4413,
+      "loss": 0.4418,
       "step": 831
     },
     {
       "epoch": 0.08908876753399722,
-      "grad_norm": 0.15699490904808044,
+      "grad_norm": 0.15801607072353363,
       "learning_rate": 1.3607658280716473e-05,
-      "loss": 0.3176,
+      "loss": 0.3175,
       "step": 832
     },
     {
       "epoch": 0.08919584537959097,
-      "grad_norm": 0.16093392670154572,
+      "grad_norm": 0.1594337821006775,
       "learning_rate": 1.3449866974698122e-05,
-      "loss": 0.2945,
+      "loss": 0.2932,
       "step": 833
     },
     {
       "epoch": 0.08930292322518471,
-      "grad_norm": 0.13421513140201569,
+      "grad_norm": 0.13377048075199127,
       "learning_rate": 1.3292929883550998e-05,
-      "loss": 0.2489,
+      "loss": 0.2479,
       "step": 834
     },
     {
       "epoch": 0.08941000107077846,
-      "grad_norm": 0.1671517938375473,
+      "grad_norm": 0.16444596648216248,
       "learning_rate": 1.3136848556180892e-05,
-      "loss": 0.3242,
+      "loss": 0.3236,
       "step": 835
     },
     {
       "epoch": 0.0895170789163722,
-      "grad_norm": 0.19687211513519287,
+      "grad_norm": 0.1959199160337448,
       "learning_rate": 1.2981624533047432e-05,
-      "loss": 0.4313,
+      "loss": 0.4308,
       "step": 836
     },
     {
       "epoch": 0.08962415676196595,
-      "grad_norm": 0.17762812972068787,
+      "grad_norm": 0.16385279595851898,
       "learning_rate": 1.2827259346149122e-05,
-      "loss": 0.3867,
+      "loss": 0.3871,
       "step": 837
     },
     {
       "epoch": 0.08973123460755969,
-      "grad_norm": 0.12915226817131042,
+      "grad_norm": 0.12947721779346466,
       "learning_rate": 1.2673754519008008e-05,
-      "loss": 0.2702,
+      "loss": 0.2712,
       "step": 838
     },
     {
       "epoch": 0.08983831245315345,
-      "grad_norm": 0.1644250750541687,
+      "grad_norm": 0.16591428220272064,
       "learning_rate": 1.2521111566654731e-05,
-      "loss": 0.2511,
+      "loss": 0.2512,
       "step": 839
     },
     {
       "epoch": 0.0899453902987472,
-      "grad_norm": 0.14260394871234894,
+      "grad_norm": 0.14401817321777344,
       "learning_rate": 1.2369331995613665e-05,
-      "loss": 0.2988,
+      "loss": 0.2989,
       "step": 840
     },
     {
       "epoch": 0.09005246814434094,
-      "grad_norm": 0.1875654011964798,
+      "grad_norm": 0.18844856321811676,
       "learning_rate": 1.2218417303887842e-05,
-      "loss": 0.3031,
+      "loss": 0.3036,
       "step": 841
     },
     {
       "epoch": 0.09015954598993468,
-      "grad_norm": 0.17389431595802307,
+      "grad_norm": 0.17315539717674255,
       "learning_rate": 1.206836898094439e-05,
-      "loss": 0.322,
+      "loss": 0.3203,
       "step": 842
     },
     {
       "epoch": 0.09026662383552843,
-      "grad_norm": 0.18567897379398346,
+      "grad_norm": 0.1839323788881302,
       "learning_rate": 1.191918850769964e-05,
       "loss": 0.3624,
       "step": 843
     },
     {
       "epoch": 0.09037370168112217,
-      "grad_norm": 0.23172374069690704,
+      "grad_norm": 0.19954797625541687,
       "learning_rate": 1.1770877356504683e-05,
-      "loss": 0.4064,
+      "loss": 0.4046,
       "step": 844
     },
     {
       "epoch": 0.09048077952671592,
-      "grad_norm": 0.17461536824703217,
+      "grad_norm": 0.17619667947292328,
       "learning_rate": 1.1623436991130654e-05,
-      "loss": 0.3287,
+      "loss": 0.328,
       "step": 845
     },
     {
       "epoch": 0.09058785737230966,
-      "grad_norm": 0.15354901552200317,
+      "grad_norm": 0.14954163134098053,
       "learning_rate": 1.1476868866754486e-05,
-      "loss": 0.2966,
+      "loss": 0.2961,
       "step": 846
     },
     {
       "epoch": 0.09069493521790342,
-      "grad_norm": 0.13990119099617004,
+      "grad_norm": 0.1400119662284851,
       "learning_rate": 1.1331174429944347e-05,
-      "loss": 0.2597,
+      "loss": 0.26,
       "step": 847
     },
     {
       "epoch": 0.09080201306349717,
-      "grad_norm": 0.15335947275161743,
+      "grad_norm": 0.15237313508987427,
       "learning_rate": 1.1186355118645554e-05,
-      "loss": 0.3011,
+      "loss": 0.3014,
       "step": 848
     },
     {
       "epoch": 0.09090909090909091,
-      "grad_norm": 0.164723739027977,
+      "grad_norm": 0.1653870940208435,
       "learning_rate": 1.1042412362166222e-05,
-      "loss": 0.2423,
+      "loss": 0.2426,
       "step": 849
     },
     {
       "epoch": 0.09101616875468466,
-      "grad_norm": 0.18542508780956268,
+      "grad_norm": 0.18718279898166656,
       "learning_rate": 1.0899347581163221e-05,
-      "loss": 0.372,
+      "loss": 0.3714,
       "step": 850
     },
     {
       "epoch": 0.0911232466002784,
-      "grad_norm": 0.16289161145687103,
+      "grad_norm": 0.1616310328245163,
       "learning_rate": 1.0757162187628222e-05,
-      "loss": 0.3027,
+      "loss": 0.3011,
       "step": 851
     },
     {
       "epoch": 0.09123032444587215,
-      "grad_norm": 0.17643985152244568,
+      "grad_norm": 0.17707855999469757,
       "learning_rate": 1.0615857584873623e-05,
-      "loss": 0.3172,
+      "loss": 0.3171,
       "step": 852
     },
     {
       "epoch": 0.09133740229146589,
-      "grad_norm": 0.1658363938331604,
+      "grad_norm": 0.16873344779014587,
       "learning_rate": 1.0475435167518843e-05,
-      "loss": 0.2663,
+      "loss": 0.2675,
       "step": 853
     },
     {
       "epoch": 0.09144448013705964,
-      "grad_norm": 0.17886339128017426,
+      "grad_norm": 0.17603348195552826,
       "learning_rate": 1.0335896321476413e-05,
-      "loss": 0.4144,
+      "loss": 0.4121,
       "step": 854
     },
     {
       "epoch": 0.0915515579826534,
-      "grad_norm": 0.15143504738807678,
+      "grad_norm": 0.1532546430826187,
       "learning_rate": 1.0197242423938446e-05,
-      "loss": 0.2608,
+      "loss": 0.2609,
       "step": 855
     },
     {
       "epoch": 0.09165863582824714,
-      "grad_norm": 0.2437451332807541,
+      "grad_norm": 0.23365063965320587,
       "learning_rate": 1.0059474843362892e-05,
-      "loss": 0.3112,
+      "loss": 0.3109,
       "step": 856
     },
     {
       "epoch": 0.09176571367384088,
-      "grad_norm": 0.2071426510810852,
+      "grad_norm": 0.20792338252067566,
       "learning_rate": 9.922594939460194e-06,
-      "loss": 0.434,
+      "loss": 0.4347,
       "step": 857
     },
     {
       "epoch": 0.09187279151943463,
-      "grad_norm": 0.15184633433818817,
+      "grad_norm": 0.15432211756706238,
       "learning_rate": 9.786604063179728e-06,
-      "loss": 0.3421,
+      "loss": 0.343,
       "step": 858
     },
     {
       "epoch": 0.09197986936502837,
-      "grad_norm": 0.20159287750720978,
+      "grad_norm": 0.1996994912624359,
       "learning_rate": 9.651503556696516e-06,
-      "loss": 0.3295,
+      "loss": 0.3304,
       "step": 859
     },
     {
       "epoch": 0.09208694721062212,
-      "grad_norm": 0.1514168232679367,
+      "grad_norm": 0.15036849677562714,
       "learning_rate": 9.517294753398064e-06,
-      "loss": 0.2855,
+      "loss": 0.2859,
       "step": 860
     },
     {
       "epoch": 0.09219402505621586,
-      "grad_norm": 0.1669357568025589,
+      "grad_norm": 0.163441002368927,
       "learning_rate": 9.383978977871021e-06,
-      "loss": 0.254,
+      "loss": 0.2546,
       "step": 861
     },
     {
       "epoch": 0.09230110290180961,
-      "grad_norm": 0.17768973112106323,
+      "grad_norm": 0.17922081053256989,
       "learning_rate": 9.251557545888312e-06,
-      "loss": 0.4355,
+      "loss": 0.435,
       "step": 862
     },
     {
       "epoch": 0.09240818074740337,
-      "grad_norm": 0.17250321805477142,
+      "grad_norm": 0.16963575780391693,
       "learning_rate": 9.120031764395987e-06,
-      "loss": 0.312,
+      "loss": 0.3112,
       "step": 863
     },
     {
       "epoch": 0.09251525859299711,
-      "grad_norm": 0.17513404786586761,
+      "grad_norm": 0.1739087551832199,
       "learning_rate": 8.989402931500434e-06,
-      "loss": 0.3356,
+      "loss": 0.3357,
       "step": 864
     },
     {
       "epoch": 0.09262233643859086,
-      "grad_norm": 0.15594890713691711,
+      "grad_norm": 0.15350165963172913,
       "learning_rate": 8.85967233645547e-06,
-      "loss": 0.2917,
+      "loss": 0.2912,
       "step": 865
     },
     {
       "epoch": 0.0927294142841846,
-      "grad_norm": 0.14464133977890015,
+      "grad_norm": 0.14361527562141418,
       "learning_rate": 8.730841259649725e-06,
-      "loss": 0.2637,
+      "loss": 0.2633,
       "step": 866
     },
     {
       "epoch": 0.09283649212977835,
-      "grad_norm": 0.16818121075630188,
+      "grad_norm": 0.16577735543251038,
       "learning_rate": 8.602910972593892e-06,
-      "loss": 0.3922,
+      "loss": 0.3942,
       "step": 867
     },
     {
       "epoch": 0.09294356997537209,
-      "grad_norm": 0.16253922879695892,
+      "grad_norm": 0.1609407663345337,
       "learning_rate": 8.475882737908248e-06,
-      "loss": 0.3469,
+      "loss": 0.3476,
       "step": 868
     },
     {
       "epoch": 0.09305064782096584,
-      "grad_norm": 0.18424832820892334,
+      "grad_norm": 0.17923296988010406,
       "learning_rate": 8.34975780931021e-06,
-      "loss": 0.3787,
+      "loss": 0.3774,
       "step": 869
     },
     {
       "epoch": 0.09315772566655958,
-      "grad_norm": 0.1491045355796814,
+      "grad_norm": 0.14731475710868835,
       "learning_rate": 8.224537431601886e-06,
-      "loss": 0.2689,
+      "loss": 0.2685,
       "step": 870
     },
     {
       "epoch": 0.09326480351215334,
-      "grad_norm": 0.19525621831417084,
+      "grad_norm": 0.1950971633195877,
       "learning_rate": 8.100222840657878e-06,
-      "loss": 0.3922,
+      "loss": 0.3933,
       "step": 871
     },
     {
       "epoch": 0.09337188135774709,
-      "grad_norm": 0.20885202288627625,
+      "grad_norm": 0.2104734629392624,
       "learning_rate": 7.976815263412963e-06,
-      "loss": 0.3052,
+      "loss": 0.305,
       "step": 872
     },
     {
       "epoch": 0.09347895920334083,
-      "grad_norm": 0.15215760469436646,
+      "grad_norm": 0.15162058174610138,
       "learning_rate": 7.854315917850163e-06,
-      "loss": 0.2968,
+      "loss": 0.296,
       "step": 873
     },
     {
       "epoch": 0.09358603704893458,
-      "grad_norm": 0.17333446443080902,
+      "grad_norm": 0.17476332187652588,
       "learning_rate": 7.73272601298851e-06,
-      "loss": 0.2817,
+      "loss": 0.2836,
       "step": 874
     },
     {
       "epoch": 0.09369311489452832,
-      "grad_norm": 0.17127728462219238,
+      "grad_norm": 0.1703297197818756,
       "learning_rate": 7.612046748871327e-06,
-      "loss": 0.2874,
+      "loss": 0.2873,
       "step": 875
     },
     {
       "epoch": 0.09380019274012207,
-      "grad_norm": 0.16041713953018188,
+      "grad_norm": 0.16354651749134064,
       "learning_rate": 7.492279316554207e-06,
-      "loss": 0.3136,
+      "loss": 0.3141,
       "step": 876
     },
     {
       "epoch": 0.09390727058571581,
-      "grad_norm": 0.1982324719429016,
+      "grad_norm": 0.19520509243011475,
       "learning_rate": 7.3734248980933395e-06,
-      "loss": 0.3975,
+      "loss": 0.397,
       "step": 877
     },
     {
       "epoch": 0.09401434843130957,
-      "grad_norm": 0.15895456075668335,
+      "grad_norm": 0.15677590668201447,
       "learning_rate": 7.255484666533874e-06,
-      "loss": 0.3546,
+      "loss": 0.3549,
       "step": 878
     },
     {
       "epoch": 0.09412142627690331,
-      "grad_norm": 0.19794389605522156,
+      "grad_norm": 0.19243308901786804,
       "learning_rate": 7.138459785898266e-06,
-      "loss": 0.4034,
+      "loss": 0.4026,
       "step": 879
     },
     {
       "epoch": 0.09422850412249706,
-      "grad_norm": 0.15700578689575195,
+      "grad_norm": 0.15591202676296234,
       "learning_rate": 7.022351411174866e-06,
-      "loss": 0.3068,
+      "loss": 0.3088,
       "step": 880
     },
     {
       "epoch": 0.0943355819680908,
-      "grad_norm": 0.17890901863574982,
+      "grad_norm": 0.17848993837833405,
       "learning_rate": 6.907160688306425e-06,
-      "loss": 0.2851,
+      "loss": 0.2836,
       "step": 881
     },
     {
       "epoch": 0.09444265981368455,
-      "grad_norm": 0.18140609562397003,
+      "grad_norm": 0.18109548091888428,
       "learning_rate": 6.7928887541789055e-06,
-      "loss": 0.3212,
+      "loss": 0.3225,
       "step": 882
     },
     {
       "epoch": 0.0945497376592783,
-      "grad_norm": 0.13674962520599365,
+      "grad_norm": 0.13605234026908875,
       "learning_rate": 6.679536736610137e-06,
       "loss": 0.2776,
       "step": 883
     },
     {
       "epoch": 0.09465681550487204,
-      "grad_norm": 0.13583561778068542,
+      "grad_norm": 0.12738342583179474,
       "learning_rate": 6.5671057543387985e-06,
-      "loss": 0.2168,
+      "loss": 0.2155,
       "step": 884
     },
     {
       "epoch": 0.09476389335046578,
-      "grad_norm": 0.19280678033828735,
+      "grad_norm": 0.19172219932079315,
       "learning_rate": 6.455596917013273e-06,
-      "loss": 0.2604,
+      "loss": 0.2594,
       "step": 885
     },
     {
       "epoch": 0.09487097119605954,
-      "grad_norm": 0.16224633157253265,
+      "grad_norm": 0.16190144419670105,
       "learning_rate": 6.345011325180772e-06,
-      "loss": 0.3116,
+      "loss": 0.311,
       "step": 886
     },
     {
       "epoch": 0.09497804904165329,
-      "grad_norm": 0.1368846297264099,
+      "grad_norm": 0.136839359998703,
       "learning_rate": 6.235350070276447e-06,
-      "loss": 0.252,
+      "loss": 0.2523,
       "step": 887
     },
     {
       "epoch": 0.09508512688724703,
-      "grad_norm": 0.12165465205907822,
+      "grad_norm": 0.12164784222841263,
       "learning_rate": 6.126614234612593e-06,
-      "loss": 0.2395,
+      "loss": 0.2399,
       "step": 888
     },
     {
       "epoch": 0.09519220473284078,
-      "grad_norm": 0.1612234264612198,
+      "grad_norm": 0.15871037542819977,
       "learning_rate": 6.018804891368035e-06,
-      "loss": 0.3441,
+      "loss": 0.3429,
       "step": 889
     },
     {
       "epoch": 0.09529928257843452,
-      "grad_norm": 0.15797550976276398,
+      "grad_norm": 0.15956546366214752,
       "learning_rate": 5.911923104577455e-06,
-      "loss": 0.3158,
+      "loss": 0.3148,
       "step": 890
     },
     {
       "epoch": 0.09540636042402827,
-      "grad_norm": 0.17879457771778107,
+      "grad_norm": 0.17873486876487732,
       "learning_rate": 5.805969929120947e-06,
-      "loss": 0.3497,
+      "loss": 0.3492,
       "step": 891
     },
     {
       "epoch": 0.09551343826962201,
-      "grad_norm": 0.1826862245798111,
+      "grad_norm": 0.18274949491024017,
       "learning_rate": 5.700946410713548e-06,
-      "loss": 0.4023,
+      "loss": 0.4022,
       "step": 892
     },
     {
       "epoch": 0.09562051611521576,
-      "grad_norm": 0.14438298344612122,
+      "grad_norm": 0.13471592962741852,
       "learning_rate": 5.5968535858950345e-06,
       "loss": 0.2356,
       "step": 893
     },
     {
       "epoch": 0.09572759396080951,
-      "grad_norm": 0.1626652181148529,
+      "grad_norm": 0.16178227961063385,
       "learning_rate": 5.49369248201953e-06,
-      "loss": 0.2726,
+      "loss": 0.2734,
       "step": 894
     },
     {
       "epoch": 0.09583467180640326,
-      "grad_norm": 0.235837921500206,
+      "grad_norm": 0.23086000978946686,
       "learning_rate": 5.39146411724547e-06,
-      "loss": 0.4375,
+      "loss": 0.4373,
       "step": 895
     },
     {
       "epoch": 0.095941749651997,
-      "grad_norm": 0.16252082586288452,
+      "grad_norm": 0.1594802439212799,
       "learning_rate": 5.290169500525577e-06,
-      "loss": 0.3127,
+      "loss": 0.3122,
       "step": 896
     },
     {
       "epoch": 0.09604882749759075,
-      "grad_norm": 0.20119427144527435,
+      "grad_norm": 0.1993282437324524,
       "learning_rate": 5.189809631596798e-06,
-      "loss": 0.4466,
+      "loss": 0.4464,
       "step": 897
     },
     {
       "epoch": 0.0961559053431845,
-      "grad_norm": 0.15925422310829163,
+      "grad_norm": 0.1615990251302719,
       "learning_rate": 5.0903855009705514e-06,
-      "loss": 0.2865,
+      "loss": 0.2872,
       "step": 898
     },
     {
       "epoch": 0.09626298318877824,
-      "grad_norm": 0.15317882597446442,
+      "grad_norm": 0.1525048166513443,
       "learning_rate": 4.991898089922819e-06,
-      "loss": 0.336,
+      "loss": 0.3361,
       "step": 899
     },
     {
       "epoch": 0.09637006103437198,
-      "grad_norm": 0.1327366828918457,
+      "grad_norm": 0.1328526735305786,
       "learning_rate": 4.8943483704846475e-06,
-      "loss": 0.2251,
+      "loss": 0.2255,
       "step": 900
     },
     {
       "epoch": 0.09647713887996573,
-      "grad_norm": 0.16071724891662598,
+      "grad_norm": 0.16284851729869843,
       "learning_rate": 4.797737305432337e-06,
-      "loss": 0.307,
+      "loss": 0.3079,
       "step": 901
     },
     {
       "epoch": 0.09658421672555949,
-      "grad_norm": 0.16794702410697937,
+      "grad_norm": 0.16765619814395905,
       "learning_rate": 4.702065848278126e-06,
-      "loss": 0.3508,
+      "loss": 0.3512,
       "step": 902
     },
     {
       "epoch": 0.09669129457115323,
-      "grad_norm": 0.13818074762821198,
+      "grad_norm": 0.136622354388237,
       "learning_rate": 4.607334943260655e-06,
-      "loss": 0.2423,
+      "loss": 0.2413,
       "step": 903
     },
     {
       "epoch": 0.09679837241674698,
-      "grad_norm": 0.20473352074623108,
+      "grad_norm": 0.22073903679847717,
       "learning_rate": 4.513545525335705e-06,
-      "loss": 0.4169,
+      "loss": 0.4166,
       "step": 904
     },
     {
       "epoch": 0.09690545026234072,
-      "grad_norm": 0.1584278792142868,
+      "grad_norm": 0.15773172676563263,
       "learning_rate": 4.420698520166988e-06,
-      "loss": 0.3027,
+      "loss": 0.303,
       "step": 905
     },
     {
       "epoch": 0.09701252810793447,
-      "grad_norm": 0.1746038943529129,
+      "grad_norm": 0.17100340127944946,
       "learning_rate": 4.328794844116946e-06,
-      "loss": 0.2906,
+      "loss": 0.2892,
       "step": 906
     },
     {
       "epoch": 0.09711960595352821,
-      "grad_norm": 0.17455877363681793,
+      "grad_norm": 0.17342492938041687,
       "learning_rate": 4.237835404237778e-06,
-      "loss": 0.2776,
+      "loss": 0.2773,
       "step": 907
     },
     {
       "epoch": 0.09722668379912196,
-      "grad_norm": 0.1757044494152069,
+      "grad_norm": 0.1771211475133896,
       "learning_rate": 4.147821098262405e-06,
-      "loss": 0.3792,
+      "loss": 0.3796,
       "step": 908
     },
     {
       "epoch": 0.0973337616447157,
-      "grad_norm": 0.1343322992324829,
+      "grad_norm": 0.13509005308151245,
       "learning_rate": 4.0587528145957235e-06,
-      "loss": 0.2821,
+      "loss": 0.2839,
       "step": 909
     },
     {
       "epoch": 0.09744083949030946,
-      "grad_norm": 0.17307552695274353,
+      "grad_norm": 0.17076431214809418,
       "learning_rate": 3.970631432305694e-06,
-      "loss": 0.3373,
+      "loss": 0.3372,
       "step": 910
     },
     {
       "epoch": 0.0975479173359032,
-      "grad_norm": 0.16109535098075867,
+      "grad_norm": 0.16045229136943817,
       "learning_rate": 3.883457821114811e-06,
-      "loss": 0.2397,
+      "loss": 0.2402,
       "step": 911
     },
     {
       "epoch": 0.09765499518149695,
-      "grad_norm": 0.17878471314907074,
+      "grad_norm": 0.18019679188728333,
       "learning_rate": 3.797232841391407e-06,
-      "loss": 0.4117,
+      "loss": 0.4094,
       "step": 912
     },
     {
       "epoch": 0.0977620730270907,
-      "grad_norm": 0.1923290491104126,
+      "grad_norm": 0.19246861338615417,
       "learning_rate": 3.711957344141237e-06,
-      "loss": 0.3319,
+      "loss": 0.3307,
       "step": 913
     },
     {
       "epoch": 0.09786915087268444,
-      "grad_norm": 0.2836756110191345,
+      "grad_norm": 0.1618802696466446,
       "learning_rate": 3.627632170999029e-06,
-      "loss": 0.2799,
+      "loss": 0.2785,
       "step": 914
     },
     {
       "epoch": 0.09797622871827819,
-      "grad_norm": 0.171551913022995,
+      "grad_norm": 0.17104919254779816,
       "learning_rate": 3.5442581542201923e-06,
-      "loss": 0.3929,
+      "loss": 0.3925,
       "step": 915
     },
     {
       "epoch": 0.09808330656387193,
-      "grad_norm": 0.15683524310588837,
+      "grad_norm": 0.15704232454299927,
       "learning_rate": 3.461836116672612e-06,
-      "loss": 0.28,
+      "loss": 0.2819,
       "step": 916
     },
     {
       "epoch": 0.09819038440946568,
-      "grad_norm": 0.17995594441890717,
+      "grad_norm": 0.17847512662410736,
       "learning_rate": 3.380366871828522e-06,
-      "loss": 0.4169,
+      "loss": 0.4173,
       "step": 917
     },
     {
       "epoch": 0.09829746225505943,
-      "grad_norm": 0.16825750470161438,
+      "grad_norm": 0.16614320874214172,
       "learning_rate": 3.2998512237565005e-06,
-      "loss": 0.2996,
+      "loss": 0.3005,
       "step": 918
     },
     {
       "epoch": 0.09840454010065318,
-      "grad_norm": 0.1549074649810791,
+      "grad_norm": 0.1553928405046463,
       "learning_rate": 3.2202899671134546e-06,
-      "loss": 0.3138,
+      "loss": 0.3145,
       "step": 919
     },
     {
       "epoch": 0.09851161794624692,
-      "grad_norm": 0.15720857679843903,
+      "grad_norm": 0.15356747806072235,
       "learning_rate": 3.1416838871368924e-06,
-      "loss": 0.2918,
+      "loss": 0.2905,
       "step": 920
     },
     {
       "epoch": 0.09861869579184067,
-      "grad_norm": 0.16377247869968414,
+      "grad_norm": 0.1633521169424057,
       "learning_rate": 3.064033759637064e-06,
-      "loss": 0.3749,
+      "loss": 0.3753,
       "step": 921
     },
     {
       "epoch": 0.09872577363743441,
-      "grad_norm": 0.18883401155471802,
+      "grad_norm": 0.16967448592185974,
       "learning_rate": 2.9873403509894203e-06,
-      "loss": 0.3754,
+      "loss": 0.3764,
       "step": 922
     },
     {
       "epoch": 0.09883285148302816,
-      "grad_norm": 0.19733136892318726,
+      "grad_norm": 0.1915234923362732,
       "learning_rate": 2.9116044181269007e-06,
-      "loss": 0.3307,
+      "loss": 0.3296,
       "step": 923
     },
     {
       "epoch": 0.0989399293286219,
-      "grad_norm": 0.20441798865795135,
+      "grad_norm": 0.2066269963979721,
       "learning_rate": 2.836826708532603e-06,
-      "loss": 0.3909,
+      "loss": 0.3912,
       "step": 924
     },
     {
       "epoch": 0.09904700717421565,
-      "grad_norm": 0.1249212846159935,
+      "grad_norm": 0.12314484268426895,
       "learning_rate": 2.7630079602323442e-06,
-      "loss": 0.2121,
+      "loss": 0.2109,
       "step": 925
     },
     {
       "epoch": 0.09915408501980941,
-      "grad_norm": 0.2168767899274826,
+      "grad_norm": 0.2099624127149582,
       "learning_rate": 2.690148901787337e-06,
-      "loss": 0.3815,
+      "loss": 0.3809,
       "step": 926
     },
     {
       "epoch": 0.09926116286540315,
-      "grad_norm": 0.15949676930904388,
+      "grad_norm": 0.16384465992450714,
       "learning_rate": 2.618250252287113e-06,
-      "loss": 0.3354,
+      "loss": 0.3344,
       "step": 927
     },
     {
       "epoch": 0.0993682407109969,
-      "grad_norm": 0.20438000559806824,
+      "grad_norm": 0.2039898931980133,
       "learning_rate": 2.5473127213422763e-06,
-      "loss": 0.3388,
+      "loss": 0.3398,
       "step": 928
     },
     {
       "epoch": 0.09947531855659064,
-      "grad_norm": 0.17350825667381287,
+      "grad_norm": 0.17251908779144287,
       "learning_rate": 2.4773370090776626e-06,
-      "loss": 0.2954,
+      "loss": 0.2963,
       "step": 929
     },
     {
       "epoch": 0.09958239640218439,
-      "grad_norm": 0.19200490415096283,
+      "grad_norm": 0.19402608275413513,
       "learning_rate": 2.4083238061252567e-06,
-      "loss": 0.5093,
+      "loss": 0.5081,
       "step": 930
     },
     {
       "epoch": 0.09968947424777813,
-      "grad_norm": 0.1551746129989624,
+      "grad_norm": 0.15272395312786102,
       "learning_rate": 2.3402737936175425e-06,
       "loss": 0.3675,
       "step": 931
     },
     {
       "epoch": 0.09979655209337188,
-      "grad_norm": 0.16438502073287964,
+      "grad_norm": 0.16620352864265442,
       "learning_rate": 2.273187643180652e-06,
-      "loss": 0.3649,
+      "loss": 0.3654,
       "step": 932
     },
     {
       "epoch": 0.09990362993896562,
-      "grad_norm": 0.2188139408826828,
+      "grad_norm": 0.2133980691432953,
       "learning_rate": 2.2070660169278166e-06,
-      "loss": 0.3949,
+      "loss": 0.3945,
       "step": 933
     },
     {
       "epoch": 0.10001070778455938,
-      "grad_norm": 0.14059962332248688,
+      "grad_norm": 0.13938206434249878,
       "learning_rate": 2.141909567452793e-06,
-      "loss": 0.256,
+      "loss": 0.2548,
       "step": 934
     },
     {
       "epoch": 0.10011778563015312,
-      "grad_norm": 0.15545494854450226,
+      "grad_norm": 0.1559179276227951,
       "learning_rate": 2.0777189378234143e-06,
-      "loss": 0.3074,
+      "loss": 0.3075,
       "step": 935
     },
     {
       "epoch": 0.10022486347574687,
-      "grad_norm": 0.1423916071653366,
+      "grad_norm": 0.14096151292324066,
       "learning_rate": 2.014494761575314e-06,
-      "loss": 0.2434,
+      "loss": 0.2429,
       "step": 936
     },
     {
       "epoch": 0.10033194132134061,
-      "grad_norm": 0.14946754276752472,
+      "grad_norm": 0.14772740006446838,
       "learning_rate": 1.9522376627055583e-06,
-      "loss": 0.3399,
+      "loss": 0.3397,
       "step": 937
     },
     {
       "epoch": 0.10043901916693436,
-      "grad_norm": 0.1648038625717163,
+      "grad_norm": 0.1633581519126892,
       "learning_rate": 1.8909482556666024e-06,
-      "loss": 0.3306,
+      "loss": 0.3296,
       "step": 938
     },
     {
       "epoch": 0.1005460970125281,
-      "grad_norm": 0.17612142860889435,
+      "grad_norm": 0.1756487637758255,
       "learning_rate": 1.8306271453601199e-06,
-      "loss": 0.3385,
+      "loss": 0.3387,
       "step": 939
     },
     {
       "epoch": 0.10065317485812185,
-      "grad_norm": 0.13343556225299835,
+      "grad_norm": 0.1336044818162918,
       "learning_rate": 1.771274927131139e-06,
       "loss": 0.2621,
       "step": 940
     },
     {
       "epoch": 0.10076025270371561,
-      "grad_norm": 0.1661439836025238,
+      "grad_norm": 0.1638735979795456,
       "learning_rate": 1.712892186762083e-06,
-      "loss": 0.3304,
+      "loss": 0.3286,
       "step": 941
     },
     {
       "epoch": 0.10086733054930935,
-      "grad_norm": 0.1674559861421585,
+      "grad_norm": 0.16432860493659973,
       "learning_rate": 1.6554795004670388e-06,
-      "loss": 0.4132,
+      "loss": 0.4136,
       "step": 942
     },
     {
       "epoch": 0.1009744083949031,
-      "grad_norm": 0.2064605951309204,
+      "grad_norm": 0.20629476010799408,
       "learning_rate": 1.5990374348860305e-06,
-      "loss": 0.3322,
+      "loss": 0.332,
       "step": 943
     },
     {
       "epoch": 0.10108148624049684,
-      "grad_norm": 0.1431376039981842,
+      "grad_norm": 0.14184875786304474,
       "learning_rate": 1.543566547079467e-06,
-      "loss": 0.2727,
+      "loss": 0.2722,
       "step": 944
     },
     {
       "epoch": 0.10118856408609059,
-      "grad_norm": 0.19648565351963043,
+      "grad_norm": 0.19237804412841797,
       "learning_rate": 1.4890673845226133e-06,
-      "loss": 0.4002,
+      "loss": 0.401,
       "step": 945
     },
     {
       "epoch": 0.10129564193168433,
-      "grad_norm": 0.1771528124809265,
+      "grad_norm": 0.17715074121952057,
       "learning_rate": 1.4355404851001952e-06,
-      "loss": 0.4174,
+      "loss": 0.4161,
       "step": 946
     },
     {
       "epoch": 0.10140271977727808,
-      "grad_norm": 0.16234669089317322,
+      "grad_norm": 0.16229744255542755,
       "learning_rate": 1.3829863771011253e-06,
-      "loss": 0.2401,
+      "loss": 0.2394,
       "step": 947
     },
     {
       "epoch": 0.10150979762287182,
-      "grad_norm": 0.17918367683887482,
+      "grad_norm": 0.2094830572605133,
       "learning_rate": 1.3314055792131964e-06,
-      "loss": 0.3435,
+      "loss": 0.3434,
       "step": 948
     },
     {
       "epoch": 0.10161687546846558,
-      "grad_norm": 0.17307892441749573,
+      "grad_norm": 0.1709808111190796,
       "learning_rate": 1.280798600518085e-06,
-      "loss": 0.3192,
+      "loss": 0.3196,
       "step": 949
     },
     {
       "epoch": 0.10172395331405933,
-      "grad_norm": 0.15263895690441132,
+      "grad_norm": 0.1492084413766861,
       "learning_rate": 1.231165940486234e-06,
-      "loss": 0.2911,
+      "loss": 0.2903,
       "step": 950
     },
     {
       "epoch": 0.10183103115965307,
-      "grad_norm": 0.1903427690267563,
+      "grad_norm": 0.1893269568681717,
       "learning_rate": 1.1825080889719563e-06,
-      "loss": 0.3734,
+      "loss": 0.3737,
       "step": 951
     },
     {
       "epoch": 0.10193810900524682,
-      "grad_norm": 0.14863133430480957,
+      "grad_norm": 0.15027467906475067,
       "learning_rate": 1.134825526208605e-06,
-      "loss": 0.2192,
+      "loss": 0.2198,
       "step": 952
     },
     {
       "epoch": 0.10204518685084056,
-      "grad_norm": 0.14327697455883026,
+      "grad_norm": 0.1478700190782547,
       "learning_rate": 1.0881187228038215e-06,
-      "loss": 0.2801,
+      "loss": 0.2796,
       "step": 953
     },
     {
       "epoch": 0.1021522646964343,
-      "grad_norm": 0.16685131192207336,
+      "grad_norm": 0.16411510109901428,
       "learning_rate": 1.0423881397349068e-06,
-      "loss": 0.3592,
+      "loss": 0.358,
       "step": 954
     },
     {
       "epoch": 0.10225934254202805,
-      "grad_norm": 0.16818654537200928,
+      "grad_norm": 0.1691845953464508,
       "learning_rate": 9.976342283442463e-07,
-      "loss": 0.323,
+      "loss": 0.3231,
       "step": 955
     },
     {
       "epoch": 0.1023664203876218,
-      "grad_norm": 0.20931144058704376,
+      "grad_norm": 0.21002455055713654,
       "learning_rate": 9.538574303348813e-07,
-      "loss": 0.3971,
+      "loss": 0.3972,
       "step": 956
     },
     {
       "epoch": 0.10247349823321555,
-      "grad_norm": 0.1491042673587799,
+      "grad_norm": 0.148806631565094,
       "learning_rate": 9.110581777661331e-07,
-      "loss": 0.2972,
+      "loss": 0.2966,
       "step": 957
     },
     {
       "epoch": 0.1025805760788093,
-      "grad_norm": 0.14343906939029694,
+      "grad_norm": 0.1439158171415329,
       "learning_rate": 8.692368930493521e-07,
-      "loss": 0.261,
+      "loss": 0.2605,
       "step": 958
     },
     {
       "epoch": 0.10268765392440304,
-      "grad_norm": 0.1593460738658905,
+      "grad_norm": 0.15770378708839417,
       "learning_rate": 8.283939889437209e-07,
-      "loss": 0.3349,
+      "loss": 0.3339,
       "step": 959
     },
     {
       "epoch": 0.10279473176999679,
-      "grad_norm": 0.16361872851848602,
+      "grad_norm": 0.16292540729045868,
       "learning_rate": 7.885298685522235e-07,
-      "loss": 0.3133,
+      "loss": 0.311,
       "step": 960
     },
     {
       "epoch": 0.10290180961559053,
-      "grad_norm": 0.20151600241661072,
+      "grad_norm": 0.20163190364837646,
       "learning_rate": 7.496449253176274e-07,
-      "loss": 0.3668,
+      "loss": 0.3665,
       "step": 961
     },
     {
       "epoch": 0.10300888746118428,
-      "grad_norm": 0.17242741584777832,
+      "grad_norm": 0.1703958362340927,
       "learning_rate": 7.117395430186414e-07,
-      "loss": 0.3549,
+      "loss": 0.3539,
       "step": 962
     },
     {
       "epoch": 0.10311596530677802,
-      "grad_norm": 0.21072721481323242,
+      "grad_norm": 0.2039760798215866,
       "learning_rate": 6.748140957660631e-07,
-      "loss": 0.3799,
+      "loss": 0.3793,
       "step": 963
     },
     {
       "epoch": 0.10322304315237177,
-      "grad_norm": 0.18324029445648193,
+      "grad_norm": 0.18234048783779144,
       "learning_rate": 6.388689479991605e-07,
-      "loss": 0.3466,
+      "loss": 0.3457,
       "step": 964
     },
     {
       "epoch": 0.10333012099796553,
-      "grad_norm": 0.13959814608097076,
+      "grad_norm": 0.14040732383728027,
       "learning_rate": 6.039044544820404e-07,
-      "loss": 0.2795,
+      "loss": 0.2796,
       "step": 965
     },
     {
       "epoch": 0.10343719884355927,
-      "grad_norm": 0.13971494138240814,
+      "grad_norm": 0.13894395530223846,
       "learning_rate": 5.699209603001076e-07,
-      "loss": 0.3022,
+      "loss": 0.3024,
       "step": 966
     },
     {
       "epoch": 0.10354427668915302,
-      "grad_norm": 0.1750052124261856,
+      "grad_norm": 0.17470873892307281,
       "learning_rate": 5.369188008567672e-07,
-      "loss": 0.3426,
+      "loss": 0.3419,
       "step": 967
     },
     {
       "epoch": 0.10365135453474676,
-      "grad_norm": 0.17160281538963318,
+      "grad_norm": 0.17506030201911926,
       "learning_rate": 5.048983018699827e-07,
-      "loss": 0.3439,
+      "loss": 0.3442,
       "step": 968
     },
     {
       "epoch": 0.1037584323803405,
-      "grad_norm": 0.17078298330307007,
+      "grad_norm": 0.17124956846237183,
       "learning_rate": 4.738597793691679e-07,
-      "loss": 0.34,
+      "loss": 0.3419,
       "step": 969
     },
     {
       "epoch": 0.10386551022593425,
-      "grad_norm": 0.1592579185962677,
+      "grad_norm": 0.15624070167541504,
       "learning_rate": 4.438035396920004e-07,
-      "loss": 0.2662,
+      "loss": 0.2658,
       "step": 970
     },
     {
       "epoch": 0.103972588071528,
-      "grad_norm": 0.23051397502422333,
+      "grad_norm": 0.2304699569940567,
       "learning_rate": 4.1472987948143473e-07,
-      "loss": 0.5427,
+      "loss": 0.5446,
       "step": 971
     },
     {
       "epoch": 0.10407966591712174,
-      "grad_norm": 0.14807277917861938,
+      "grad_norm": 0.14617536962032318,
       "learning_rate": 3.866390856827495e-07,
-      "loss": 0.3386,
+      "loss": 0.3385,
       "step": 972
     },
     {
       "epoch": 0.1041867437627155,
-      "grad_norm": 0.17786002159118652,
+      "grad_norm": 0.17763929069042206,
       "learning_rate": 3.595314355407609e-07,
-      "loss": 0.3552,
+      "loss": 0.3553,
       "step": 973
     },
     {
       "epoch": 0.10429382160830925,
-      "grad_norm": 0.14650464057922363,
+      "grad_norm": 0.14324170351028442,
       "learning_rate": 3.3340719659701313e-07,
-      "loss": 0.309,
+      "loss": 0.3078,
       "step": 974
     },
     {
       "epoch": 0.10440089945390299,
-      "grad_norm": 0.16224049031734467,
+      "grad_norm": 0.1634160429239273,
       "learning_rate": 3.0826662668720364e-07,
-      "loss": 0.3336,
+      "loss": 0.3353,
       "step": 975
     },
     {
       "epoch": 0.10450797729949673,
-      "grad_norm": 0.1791958510875702,
+      "grad_norm": 0.17448119819164276,
       "learning_rate": 2.841099739386066e-07,
-      "loss": 0.3769,
+      "loss": 0.3754,
       "step": 976
     },
     {
       "epoch": 0.10461505514509048,
-      "grad_norm": 0.154682457447052,
+      "grad_norm": 0.15392538905143738,
       "learning_rate": 2.609374767676309e-07,
-      "loss": 0.3388,
+      "loss": 0.338,
       "step": 977
     },
     {
       "epoch": 0.10472213299068422,
-      "grad_norm": 0.2175721973180771,
+      "grad_norm": 0.2160252183675766,
       "learning_rate": 2.387493638774774e-07,
       "loss": 0.4477,
       "step": 978
     },
     {
       "epoch": 0.10482921083627797,
-      "grad_norm": 0.16635477542877197,
+      "grad_norm": 0.16572636365890503,
       "learning_rate": 2.175458542558517e-07,
-      "loss": 0.3834,
+      "loss": 0.3833,
       "step": 979
     },
     {
       "epoch": 0.10493628868187171,
-      "grad_norm": 0.17459362745285034,
+      "grad_norm": 0.17389637231826782,
       "learning_rate": 1.973271571728441e-07,
-      "loss": 0.288,
+      "loss": 0.2881,
       "step": 980
     },
     {
       "epoch": 0.10504336652746547,
-      "grad_norm": 0.16596029698848724,
+      "grad_norm": 0.16537849605083466,
       "learning_rate": 1.7809347217881966e-07,
-      "loss": 0.3816,
+      "loss": 0.3809,
       "step": 981
     },
     {
       "epoch": 0.10515044437305922,
-      "grad_norm": 0.18052440881729126,
+      "grad_norm": 0.1813080906867981,
       "learning_rate": 1.598449891024978e-07,
-      "loss": 0.3266,
+      "loss": 0.3269,
       "step": 982
     },
     {
       "epoch": 0.10525752221865296,
-      "grad_norm": 0.17302225530147552,
+      "grad_norm": 0.1730542629957199,
       "learning_rate": 1.425818880490315e-07,
-      "loss": 0.3603,
+      "loss": 0.3606,
       "step": 983
     },
     {
       "epoch": 0.10536460006424671,
-      "grad_norm": 0.20757171511650085,
+      "grad_norm": 0.2064792513847351,
       "learning_rate": 1.2630433939825327e-07,
-      "loss": 0.3392,
+      "loss": 0.3393,
       "step": 984
     },
     {
       "epoch": 0.10547167790984045,
-      "grad_norm": 0.15444600582122803,
+      "grad_norm": 0.15513303875923157,
       "learning_rate": 1.1101250380300965e-07,
-      "loss": 0.3041,
+      "loss": 0.3034,
       "step": 985
     },
     {
       "epoch": 0.1055787557554342,
-      "grad_norm": 0.17258261144161224,
+      "grad_norm": 0.17312531173229218,
       "learning_rate": 9.670653218752934e-08,
-      "loss": 0.3568,
+      "loss": 0.3564,
       "step": 986
     },
     {
       "epoch": 0.10568583360102794,
-      "grad_norm": 0.1691325157880783,
+      "grad_norm": 0.16997037827968597,
       "learning_rate": 8.33865657459909e-08,
-      "loss": 0.4094,
+      "loss": 0.4089,
       "step": 987
     },
     {
       "epoch": 0.10579291144662169,
-      "grad_norm": 0.204833522439003,
+      "grad_norm": 0.2006881982088089,
       "learning_rate": 7.105273594107953e-08,
-      "loss": 0.3808,
+      "loss": 0.3803,
       "step": 988
     },
     {
       "epoch": 0.10589998929221545,
-      "grad_norm": 0.1730750948190689,
+      "grad_norm": 0.17012691497802734,
       "learning_rate": 5.970516450271025e-08,
-      "loss": 0.3734,
+      "loss": 0.3733,
       "step": 989
     },
     {
       "epoch": 0.10600706713780919,
-      "grad_norm": 0.16625729203224182,
+      "grad_norm": 0.1657508760690689,
       "learning_rate": 4.934396342684e-08,
-      "loss": 0.3676,
+      "loss": 0.3675,
       "step": 990
     },
     {
       "epoch": 0.10611414498340294,
-      "grad_norm": 0.15599259734153748,
+      "grad_norm": 0.15509915351867676,
       "learning_rate": 3.996923497434635e-08,
-      "loss": 0.3668,
+      "loss": 0.367,
       "step": 991
     },
     {
       "epoch": 0.10622122282899668,
-      "grad_norm": 0.1792498528957367,
+      "grad_norm": 0.1795806735754013,
       "learning_rate": 3.1581071670006015e-08,
-      "loss": 0.4077,
+      "loss": 0.4073,
       "step": 992
     },
     {
       "epoch": 0.10632830067459043,
-      "grad_norm": 0.18159109354019165,
+      "grad_norm": 0.1823895275592804,
       "learning_rate": 2.417955630159563e-08,
-      "loss": 0.3609,
+      "loss": 0.3606,
       "step": 993
     },
     {
       "epoch": 0.10643537852018417,
-      "grad_norm": 0.1742103546857834,
+      "grad_norm": 0.17259643971920013,
       "learning_rate": 1.7764761919103477e-08,
-      "loss": 0.3711,
+      "loss": 0.3723,
       "step": 994
     },
     {
       "epoch": 0.10654245636577792,
-      "grad_norm": 0.15315918624401093,
+      "grad_norm": 0.15075139701366425,
       "learning_rate": 1.2336751833941229e-08,
-      "loss": 0.2493,
+      "loss": 0.2497,
       "step": 995
     },
     {
       "epoch": 0.10664953421137166,
-      "grad_norm": 0.1537967324256897,
+      "grad_norm": 0.1535639613866806,
       "learning_rate": 7.895579618388827e-09,
-      "loss": 0.3098,
+      "loss": 0.3089,
       "step": 996
     },
     {
       "epoch": 0.10675661205696542,
-      "grad_norm": 0.17367656528949738,
+      "grad_norm": 0.17250818014144897,
       "learning_rate": 4.4412891050171765e-09,
-      "loss": 0.3143,
+      "loss": 0.3135,
       "step": 997
     },
     {
       "epoch": 0.10686368990255916,
-      "grad_norm": 0.14059410989284515,
+      "grad_norm": 0.1399751603603363,
       "learning_rate": 1.973914386288467e-09,
-      "loss": 0.2818,
+      "loss": 0.2826,
       "step": 998
     },
     {
       "epoch": 0.10697076774815291,
-      "grad_norm": 0.21730297803878784,
+      "grad_norm": 0.2154352217912674,
       "learning_rate": 4.934798141786879e-10,
-      "loss": 0.4292,
+      "loss": 0.4276,
       "step": 999
     },
     {
       "epoch": 0.10707784559374665,
-      "grad_norm": 0.21436914801597595,
+      "grad_norm": 0.21260492503643036,
       "learning_rate": 0.0,
-      "loss": 0.4271,
+      "loss": 0.4272,
       "step": 1000
     },
     {
       "epoch": 0.10707784559374665,
-      "eval_loss": 0.33198028802871704,
-      "eval_runtime": 367.0984,
-      "eval_samples_per_second": 5.358,
-      "eval_steps_per_second": 5.358,
+      "eval_loss": 0.3318476676940918,
+      "eval_runtime": 355.7741,
+      "eval_samples_per_second": 5.529,
+      "eval_steps_per_second": 5.529,
       "step": 1000
     }
   ],