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"
@@ -1,9 +1,9 @@
 {
   "best_metric": 11.5,
   "best_model_checkpoint": "miner_id_24/checkpoint-1000",
-  "epoch": 1.800720288115246,
+  "epoch": 2.4009603841536613,
   "eval_steps": 1000,
-  "global_step": 3000,
+  "global_step": 4000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -21039,6 +21039,7014 @@
       "eval_samples_per_second": 261.522,
       "eval_steps_per_second": 32.807,
       "step": 3000
+    },
+    {
+      "epoch": 1.8013205282112845,
+      "grad_norm": 0.0011961712734773755,
+      "learning_rate": 0.00018465877693180072,
+      "loss": 23.0,
+      "step": 3001
+    },
+    {
+      "epoch": 1.801920768307323,
+      "grad_norm": 0.0015885218745097518,
+      "learning_rate": 0.00018464872064075577,
+      "loss": 23.0,
+      "step": 3002
+    },
+    {
+      "epoch": 1.8025210084033614,
+      "grad_norm": 0.0019562295638024807,
+      "learning_rate": 0.00018463866132881578,
+      "loss": 23.0,
+      "step": 3003
+    },
+    {
+      "epoch": 1.8031212484993997,
+      "grad_norm": 0.003253971692174673,
+      "learning_rate": 0.00018462859899633967,
+      "loss": 23.0,
+      "step": 3004
+    },
+    {
+      "epoch": 1.8037214885954382,
+      "grad_norm": 0.000821275869384408,
+      "learning_rate": 0.00018461853364368653,
+      "loss": 23.0,
+      "step": 3005
+    },
+    {
+      "epoch": 1.8043217286914766,
+      "grad_norm": 0.0012803277932107449,
+      "learning_rate": 0.0001846084652712156,
+      "loss": 23.0,
+      "step": 3006
+    },
+    {
+      "epoch": 1.8049219687875149,
+      "grad_norm": 0.0007911855354905128,
+      "learning_rate": 0.00018459839387928619,
+      "loss": 23.0,
+      "step": 3007
+    },
+    {
+      "epoch": 1.8055222088835534,
+      "grad_norm": 0.0007396016153506935,
+      "learning_rate": 0.0001845883194682577,
+      "loss": 23.0,
+      "step": 3008
+    },
+    {
+      "epoch": 1.806122448979592,
+      "grad_norm": 0.0014897441724315286,
+      "learning_rate": 0.0001845782420384897,
+      "loss": 23.0,
+      "step": 3009
+    },
+    {
+      "epoch": 1.8067226890756303,
+      "grad_norm": 0.0004990187590010464,
+      "learning_rate": 0.0001845681615903418,
+      "loss": 23.0,
+      "step": 3010
+    },
+    {
+      "epoch": 1.8073229291716686,
+      "grad_norm": 0.0008598102140240371,
+      "learning_rate": 0.00018455807812417372,
+      "loss": 23.0,
+      "step": 3011
+    },
+    {
+      "epoch": 1.8079231692677071,
+      "grad_norm": 0.0011542581487447023,
+      "learning_rate": 0.00018454799164034538,
+      "loss": 23.0,
+      "step": 3012
+    },
+    {
+      "epoch": 1.8085234093637454,
+      "grad_norm": 0.0013645098078995943,
+      "learning_rate": 0.00018453790213921665,
+      "loss": 23.0,
+      "step": 3013
+    },
+    {
+      "epoch": 1.8091236494597838,
+      "grad_norm": 0.0008868533186614513,
+      "learning_rate": 0.00018452780962114772,
+      "loss": 23.0,
+      "step": 3014
+    },
+    {
+      "epoch": 1.8097238895558223,
+      "grad_norm": 0.0018087849020957947,
+      "learning_rate": 0.00018451771408649865,
+      "loss": 23.0,
+      "step": 3015
+    },
+    {
+      "epoch": 1.8103241296518608,
+      "grad_norm": 0.00035064207622781396,
+      "learning_rate": 0.00018450761553562975,
+      "loss": 23.0,
+      "step": 3016
+    },
+    {
+      "epoch": 1.8109243697478992,
+      "grad_norm": 0.0014904963318258524,
+      "learning_rate": 0.00018449751396890148,
+      "loss": 23.0,
+      "step": 3017
+    },
+    {
+      "epoch": 1.8115246098439375,
+      "grad_norm": 0.0019730026833713055,
+      "learning_rate": 0.00018448740938667427,
+      "loss": 23.0,
+      "step": 3018
+    },
+    {
+      "epoch": 1.812124849939976,
+      "grad_norm": 0.0017877138452604413,
+      "learning_rate": 0.00018447730178930874,
+      "loss": 23.0,
+      "step": 3019
+    },
+    {
+      "epoch": 1.8127250900360146,
+      "grad_norm": 0.0008809540886431932,
+      "learning_rate": 0.0001844671911771656,
+      "loss": 23.0,
+      "step": 3020
+    },
+    {
+      "epoch": 1.8133253301320527,
+      "grad_norm": 0.0015870918286964297,
+      "learning_rate": 0.0001844570775506057,
+      "loss": 23.0,
+      "step": 3021
+    },
+    {
+      "epoch": 1.8139255702280912,
+      "grad_norm": 0.001148207113146782,
+      "learning_rate": 0.00018444696090998995,
+      "loss": 23.0,
+      "step": 3022
+    },
+    {
+      "epoch": 1.8145258103241297,
+      "grad_norm": 0.0025478298775851727,
+      "learning_rate": 0.00018443684125567938,
+      "loss": 23.0,
+      "step": 3023
+    },
+    {
+      "epoch": 1.815126050420168,
+      "grad_norm": 0.0015544770285487175,
+      "learning_rate": 0.00018442671858803514,
+      "loss": 23.0,
+      "step": 3024
+    },
+    {
+      "epoch": 1.8157262905162064,
+      "grad_norm": 0.001434922218322754,
+      "learning_rate": 0.00018441659290741852,
+      "loss": 23.0,
+      "step": 3025
+    },
+    {
+      "epoch": 1.816326530612245,
+      "grad_norm": 0.0009723508264869452,
+      "learning_rate": 0.00018440646421419085,
+      "loss": 23.0,
+      "step": 3026
+    },
+    {
+      "epoch": 1.8169267707082835,
+      "grad_norm": 0.0012221380602568388,
+      "learning_rate": 0.00018439633250871354,
+      "loss": 23.0,
+      "step": 3027
+    },
+    {
+      "epoch": 1.8175270108043218,
+      "grad_norm": 0.0012381580891087651,
+      "learning_rate": 0.00018438619779134822,
+      "loss": 23.0,
+      "step": 3028
+    },
+    {
+      "epoch": 1.81812725090036,
+      "grad_norm": 0.0011028312146663666,
+      "learning_rate": 0.0001843760600624566,
+      "loss": 23.0,
+      "step": 3029
+    },
+    {
+      "epoch": 1.8187274909963986,
+      "grad_norm": 0.0011208378709852695,
+      "learning_rate": 0.0001843659193224004,
+      "loss": 23.0,
+      "step": 3030
+    },
+    {
+      "epoch": 1.819327731092437,
+      "grad_norm": 0.001310601131990552,
+      "learning_rate": 0.0001843557755715416,
+      "loss": 23.0,
+      "step": 3031
+    },
+    {
+      "epoch": 1.8199279711884753,
+      "grad_norm": 0.0017411686712875962,
+      "learning_rate": 0.00018434562881024214,
+      "loss": 23.0,
+      "step": 3032
+    },
+    {
+      "epoch": 1.8205282112845138,
+      "grad_norm": 0.0012776413932442665,
+      "learning_rate": 0.00018433547903886415,
+      "loss": 23.0,
+      "step": 3033
+    },
+    {
+      "epoch": 1.8211284513805523,
+      "grad_norm": 0.00228596362285316,
+      "learning_rate": 0.00018432532625776987,
+      "loss": 23.0,
+      "step": 3034
+    },
+    {
+      "epoch": 1.8217286914765907,
+      "grad_norm": 0.0009363253484480083,
+      "learning_rate": 0.00018431517046732157,
+      "loss": 23.0,
+      "step": 3035
+    },
+    {
+      "epoch": 1.822328931572629,
+      "grad_norm": 0.0019296659156680107,
+      "learning_rate": 0.00018430501166788173,
+      "loss": 23.0,
+      "step": 3036
+    },
+    {
+      "epoch": 1.8229291716686675,
+      "grad_norm": 0.0037253268528729677,
+      "learning_rate": 0.0001842948498598129,
+      "loss": 23.0,
+      "step": 3037
+    },
+    {
+      "epoch": 1.8235294117647058,
+      "grad_norm": 0.0003991881385445595,
+      "learning_rate": 0.00018428468504347767,
+      "loss": 23.0,
+      "step": 3038
+    },
+    {
+      "epoch": 1.8241296518607442,
+      "grad_norm": 0.004006813745945692,
+      "learning_rate": 0.00018427451721923888,
+      "loss": 23.0,
+      "step": 3039
+    },
+    {
+      "epoch": 1.8247298919567827,
+      "grad_norm": 0.0013116990448907018,
+      "learning_rate": 0.00018426434638745933,
+      "loss": 23.0,
+      "step": 3040
+    },
+    {
+      "epoch": 1.8253301320528212,
+      "grad_norm": 0.0044224620796740055,
+      "learning_rate": 0.000184254172548502,
+      "loss": 23.0,
+      "step": 3041
+    },
+    {
+      "epoch": 1.8259303721488596,
+      "grad_norm": 0.0010624263668432832,
+      "learning_rate": 0.00018424399570273002,
+      "loss": 23.0,
+      "step": 3042
+    },
+    {
+      "epoch": 1.8265306122448979,
+      "grad_norm": 0.00032893690513446927,
+      "learning_rate": 0.0001842338158505065,
+      "loss": 23.0,
+      "step": 3043
+    },
+    {
+      "epoch": 1.8271308523409364,
+      "grad_norm": 0.0007519045611843467,
+      "learning_rate": 0.00018422363299219478,
+      "loss": 23.0,
+      "step": 3044
+    },
+    {
+      "epoch": 1.8277310924369747,
+      "grad_norm": 0.0011156450491398573,
+      "learning_rate": 0.00018421344712815825,
+      "loss": 23.0,
+      "step": 3045
+    },
+    {
+      "epoch": 1.828331332533013,
+      "grad_norm": 0.0012068160576745868,
+      "learning_rate": 0.0001842032582587604,
+      "loss": 23.0,
+      "step": 3046
+    },
+    {
+      "epoch": 1.8289315726290516,
+      "grad_norm": 0.00150686118286103,
+      "learning_rate": 0.00018419306638436486,
+      "loss": 23.0,
+      "step": 3047
+    },
+    {
+      "epoch": 1.8295318127250901,
+      "grad_norm": 0.0015496856067329645,
+      "learning_rate": 0.00018418287150533537,
+      "loss": 23.0,
+      "step": 3048
+    },
+    {
+      "epoch": 1.8301320528211285,
+      "grad_norm": 0.0019011717522516847,
+      "learning_rate": 0.00018417267362203575,
+      "loss": 23.0,
+      "step": 3049
+    },
+    {
+      "epoch": 1.8307322929171668,
+      "grad_norm": 0.001324790413491428,
+      "learning_rate": 0.00018416247273482988,
+      "loss": 23.0,
+      "step": 3050
+    },
+    {
+      "epoch": 1.8313325330132053,
+      "grad_norm": 0.0008508468163199723,
+      "learning_rate": 0.0001841522688440819,
+      "loss": 23.0,
+      "step": 3051
+    },
+    {
+      "epoch": 1.8319327731092439,
+      "grad_norm": 0.0004634323122445494,
+      "learning_rate": 0.00018414206195015587,
+      "loss": 23.0,
+      "step": 3052
+    },
+    {
+      "epoch": 1.832533013205282,
+      "grad_norm": 0.003283916972577572,
+      "learning_rate": 0.0001841318520534161,
+      "loss": 23.0,
+      "step": 3053
+    },
+    {
+      "epoch": 1.8331332533013205,
+      "grad_norm": 0.0021354639902710915,
+      "learning_rate": 0.00018412163915422695,
+      "loss": 23.0,
+      "step": 3054
+    },
+    {
+      "epoch": 1.833733493397359,
+      "grad_norm": 0.0010375322308391333,
+      "learning_rate": 0.0001841114232529529,
+      "loss": 23.0,
+      "step": 3055
+    },
+    {
+      "epoch": 1.8343337334933973,
+      "grad_norm": 0.0026274758856743574,
+      "learning_rate": 0.0001841012043499585,
+      "loss": 23.0,
+      "step": 3056
+    },
+    {
+      "epoch": 1.8349339735894357,
+      "grad_norm": 0.0008885942515917122,
+      "learning_rate": 0.00018409098244560844,
+      "loss": 23.0,
+      "step": 3057
+    },
+    {
+      "epoch": 1.8355342136854742,
+      "grad_norm": 0.0019456876907497644,
+      "learning_rate": 0.00018408075754026753,
+      "loss": 23.0,
+      "step": 3058
+    },
+    {
+      "epoch": 1.8361344537815127,
+      "grad_norm": 0.002114507369697094,
+      "learning_rate": 0.00018407052963430068,
+      "loss": 23.0,
+      "step": 3059
+    },
+    {
+      "epoch": 1.836734693877551,
+      "grad_norm": 0.0008166294428519905,
+      "learning_rate": 0.00018406029872807287,
+      "loss": 23.0,
+      "step": 3060
+    },
+    {
+      "epoch": 1.8373349339735894,
+      "grad_norm": 0.0016927891410887241,
+      "learning_rate": 0.00018405006482194921,
+      "loss": 23.0,
+      "step": 3061
+    },
+    {
+      "epoch": 1.837935174069628,
+      "grad_norm": 0.0014067868469282985,
+      "learning_rate": 0.00018403982791629497,
+      "loss": 23.0,
+      "step": 3062
+    },
+    {
+      "epoch": 1.8385354141656662,
+      "grad_norm": 0.0017834990285336971,
+      "learning_rate": 0.00018402958801147542,
+      "loss": 23.0,
+      "step": 3063
+    },
+    {
+      "epoch": 1.8391356542617046,
+      "grad_norm": 0.0011745449155569077,
+      "learning_rate": 0.00018401934510785606,
+      "loss": 23.0,
+      "step": 3064
+    },
+    {
+      "epoch": 1.839735894357743,
+      "grad_norm": 0.0009545880602672696,
+      "learning_rate": 0.00018400909920580238,
+      "loss": 23.0,
+      "step": 3065
+    },
+    {
+      "epoch": 1.8403361344537816,
+      "grad_norm": 0.003715930739417672,
+      "learning_rate": 0.00018399885030568003,
+      "loss": 23.0,
+      "step": 3066
+    },
+    {
+      "epoch": 1.84093637454982,
+      "grad_norm": 0.0006399541744031012,
+      "learning_rate": 0.0001839885984078548,
+      "loss": 23.0,
+      "step": 3067
+    },
+    {
+      "epoch": 1.8415366146458583,
+      "grad_norm": 0.001763959531672299,
+      "learning_rate": 0.00018397834351269255,
+      "loss": 23.0,
+      "step": 3068
+    },
+    {
+      "epoch": 1.8421368547418968,
+      "grad_norm": 0.0012634745799005032,
+      "learning_rate": 0.0001839680856205592,
+      "loss": 23.0,
+      "step": 3069
+    },
+    {
+      "epoch": 1.8427370948379351,
+      "grad_norm": 0.0015287197893485427,
+      "learning_rate": 0.0001839578247318209,
+      "loss": 23.0,
+      "step": 3070
+    },
+    {
+      "epoch": 1.8433373349339734,
+      "grad_norm": 0.0009792933706194162,
+      "learning_rate": 0.00018394756084684377,
+      "loss": 23.0,
+      "step": 3071
+    },
+    {
+      "epoch": 1.843937575030012,
+      "grad_norm": 0.0009839729173108935,
+      "learning_rate": 0.00018393729396599416,
+      "loss": 23.0,
+      "step": 3072
+    },
+    {
+      "epoch": 1.8445378151260505,
+      "grad_norm": 0.0008178153657354414,
+      "learning_rate": 0.00018392702408963842,
+      "loss": 23.0,
+      "step": 3073
+    },
+    {
+      "epoch": 1.8451380552220888,
+      "grad_norm": 0.0018251874716952443,
+      "learning_rate": 0.0001839167512181431,
+      "loss": 23.0,
+      "step": 3074
+    },
+    {
+      "epoch": 1.8457382953181272,
+      "grad_norm": 0.002089198911562562,
+      "learning_rate": 0.00018390647535187476,
+      "loss": 23.0,
+      "step": 3075
+    },
+    {
+      "epoch": 1.8463385354141657,
+      "grad_norm": 0.0021562918554991484,
+      "learning_rate": 0.00018389619649120017,
+      "loss": 23.0,
+      "step": 3076
+    },
+    {
+      "epoch": 1.8469387755102042,
+      "grad_norm": 0.0017833629390224814,
+      "learning_rate": 0.00018388591463648616,
+      "loss": 23.0,
+      "step": 3077
+    },
+    {
+      "epoch": 1.8475390156062423,
+      "grad_norm": 0.000439735857071355,
+      "learning_rate": 0.00018387562978809958,
+      "loss": 23.0,
+      "step": 3078
+    },
+    {
+      "epoch": 1.8481392557022809,
+      "grad_norm": 0.001050430117174983,
+      "learning_rate": 0.00018386534194640754,
+      "loss": 23.0,
+      "step": 3079
+    },
+    {
+      "epoch": 1.8487394957983194,
+      "grad_norm": 0.001189610455185175,
+      "learning_rate": 0.00018385505111177719,
+      "loss": 23.0,
+      "step": 3080
+    },
+    {
+      "epoch": 1.8493397358943577,
+      "grad_norm": 0.0010707725305110216,
+      "learning_rate": 0.00018384475728457574,
+      "loss": 23.0,
+      "step": 3081
+    },
+    {
+      "epoch": 1.849939975990396,
+      "grad_norm": 0.002398673677816987,
+      "learning_rate": 0.00018383446046517062,
+      "loss": 23.0,
+      "step": 3082
+    },
+    {
+      "epoch": 1.8505402160864346,
+      "grad_norm": 0.0018957408610731363,
+      "learning_rate": 0.00018382416065392923,
+      "loss": 23.0,
+      "step": 3083
+    },
+    {
+      "epoch": 1.8511404561824731,
+      "grad_norm": 0.0022743563167750835,
+      "learning_rate": 0.0001838138578512192,
+      "loss": 23.0,
+      "step": 3084
+    },
+    {
+      "epoch": 1.8517406962785115,
+      "grad_norm": 0.0014944793656468391,
+      "learning_rate": 0.00018380355205740813,
+      "loss": 23.0,
+      "step": 3085
+    },
+    {
+      "epoch": 1.8523409363745498,
+      "grad_norm": 0.0011577775003388524,
+      "learning_rate": 0.00018379324327286387,
+      "loss": 23.0,
+      "step": 3086
+    },
+    {
+      "epoch": 1.8529411764705883,
+      "grad_norm": 0.0023493997287005186,
+      "learning_rate": 0.0001837829314979543,
+      "loss": 23.0,
+      "step": 3087
+    },
+    {
+      "epoch": 1.8535414165666266,
+      "grad_norm": 0.0017023790860548615,
+      "learning_rate": 0.00018377261673304743,
+      "loss": 23.0,
+      "step": 3088
+    },
+    {
+      "epoch": 1.854141656662665,
+      "grad_norm": 0.001627809484489262,
+      "learning_rate": 0.00018376229897851135,
+      "loss": 23.0,
+      "step": 3089
+    },
+    {
+      "epoch": 1.8547418967587035,
+      "grad_norm": 0.0016393931582570076,
+      "learning_rate": 0.00018375197823471427,
+      "loss": 23.0,
+      "step": 3090
+    },
+    {
+      "epoch": 1.855342136854742,
+      "grad_norm": 0.0018370094476267695,
+      "learning_rate": 0.00018374165450202456,
+      "loss": 23.0,
+      "step": 3091
+    },
+    {
+      "epoch": 1.8559423769507803,
+      "grad_norm": 0.0014342045178636909,
+      "learning_rate": 0.0001837313277808106,
+      "loss": 23.0,
+      "step": 3092
+    },
+    {
+      "epoch": 1.8565426170468187,
+      "grad_norm": 0.0029582458082586527,
+      "learning_rate": 0.0001837209980714409,
+      "loss": 23.0,
+      "step": 3093
+    },
+    {
+      "epoch": 1.8571428571428572,
+      "grad_norm": 0.0012816624948754907,
+      "learning_rate": 0.00018371066537428417,
+      "loss": 23.0,
+      "step": 3094
+    },
+    {
+      "epoch": 1.8577430972388955,
+      "grad_norm": 0.002697428921237588,
+      "learning_rate": 0.0001837003296897091,
+      "loss": 23.0,
+      "step": 3095
+    },
+    {
+      "epoch": 1.8583433373349338,
+      "grad_norm": 0.0037919096648693085,
+      "learning_rate": 0.00018368999101808458,
+      "loss": 23.0,
+      "step": 3096
+    },
+    {
+      "epoch": 1.8589435774309724,
+      "grad_norm": 0.003335389541462064,
+      "learning_rate": 0.00018367964935977953,
+      "loss": 23.0,
+      "step": 3097
+    },
+    {
+      "epoch": 1.859543817527011,
+      "grad_norm": 0.0011625085026025772,
+      "learning_rate": 0.00018366930471516306,
+      "loss": 23.0,
+      "step": 3098
+    },
+    {
+      "epoch": 1.8601440576230492,
+      "grad_norm": 0.001312834327109158,
+      "learning_rate": 0.00018365895708460434,
+      "loss": 23.0,
+      "step": 3099
+    },
+    {
+      "epoch": 1.8607442977190876,
+      "grad_norm": 0.0009217692422680557,
+      "learning_rate": 0.00018364860646847265,
+      "loss": 23.0,
+      "step": 3100
+    },
+    {
+      "epoch": 1.861344537815126,
+      "grad_norm": 0.001049682148732245,
+      "learning_rate": 0.00018363825286713735,
+      "loss": 23.0,
+      "step": 3101
+    },
+    {
+      "epoch": 1.8619447779111644,
+      "grad_norm": 0.0009720840607769787,
+      "learning_rate": 0.00018362789628096794,
+      "loss": 23.0,
+      "step": 3102
+    },
+    {
+      "epoch": 1.8625450180072027,
+      "grad_norm": 0.0013249212643131614,
+      "learning_rate": 0.000183617536710334,
+      "loss": 23.0,
+      "step": 3103
+    },
+    {
+      "epoch": 1.8631452581032413,
+      "grad_norm": 0.0006475619738921523,
+      "learning_rate": 0.00018360717415560532,
+      "loss": 23.0,
+      "step": 3104
+    },
+    {
+      "epoch": 1.8637454981992798,
+      "grad_norm": 0.0009137241286225617,
+      "learning_rate": 0.00018359680861715163,
+      "loss": 23.0,
+      "step": 3105
+    },
+    {
+      "epoch": 1.8643457382953181,
+      "grad_norm": 0.001901013427414,
+      "learning_rate": 0.00018358644009534286,
+      "loss": 23.0,
+      "step": 3106
+    },
+    {
+      "epoch": 1.8649459783913565,
+      "grad_norm": 0.0014280707109719515,
+      "learning_rate": 0.0001835760685905491,
+      "loss": 23.0,
+      "step": 3107
+    },
+    {
+      "epoch": 1.865546218487395,
+      "grad_norm": 0.0019309140043333173,
+      "learning_rate": 0.00018356569410314042,
+      "loss": 23.0,
+      "step": 3108
+    },
+    {
+      "epoch": 1.8661464585834335,
+      "grad_norm": 0.0008051837212406099,
+      "learning_rate": 0.00018355531663348705,
+      "loss": 23.0,
+      "step": 3109
+    },
+    {
+      "epoch": 1.8667466986794716,
+      "grad_norm": 0.0013398687588050961,
+      "learning_rate": 0.00018354493618195935,
+      "loss": 23.0,
+      "step": 3110
+    },
+    {
+      "epoch": 1.8673469387755102,
+      "grad_norm": 0.000537229236215353,
+      "learning_rate": 0.00018353455274892783,
+      "loss": 23.0,
+      "step": 3111
+    },
+    {
+      "epoch": 1.8679471788715487,
+      "grad_norm": 0.00452477065846324,
+      "learning_rate": 0.00018352416633476298,
+      "loss": 23.0,
+      "step": 3112
+    },
+    {
+      "epoch": 1.868547418967587,
+      "grad_norm": 0.002200663322582841,
+      "learning_rate": 0.0001835137769398355,
+      "loss": 23.0,
+      "step": 3113
+    },
+    {
+      "epoch": 1.8691476590636253,
+      "grad_norm": 0.0013940661447122693,
+      "learning_rate": 0.0001835033845645161,
+      "loss": 23.0,
+      "step": 3114
+    },
+    {
+      "epoch": 1.8697478991596639,
+      "grad_norm": 0.002113936236128211,
+      "learning_rate": 0.00018349298920917572,
+      "loss": 23.0,
+      "step": 3115
+    },
+    {
+      "epoch": 1.8703481392557024,
+      "grad_norm": 0.0012045325711369514,
+      "learning_rate": 0.00018348259087418533,
+      "loss": 23.0,
+      "step": 3116
+    },
+    {
+      "epoch": 1.8709483793517407,
+      "grad_norm": 0.0033185603097081184,
+      "learning_rate": 0.000183472189559916,
+      "loss": 23.0,
+      "step": 3117
+    },
+    {
+      "epoch": 1.871548619447779,
+      "grad_norm": 0.0011681788600981236,
+      "learning_rate": 0.00018346178526673897,
+      "loss": 23.0,
+      "step": 3118
+    },
+    {
+      "epoch": 1.8721488595438176,
+      "grad_norm": 0.0022086468525230885,
+      "learning_rate": 0.0001834513779950255,
+      "loss": 23.0,
+      "step": 3119
+    },
+    {
+      "epoch": 1.872749099639856,
+      "grad_norm": 0.0005794914322905242,
+      "learning_rate": 0.000183440967745147,
+      "loss": 23.0,
+      "step": 3120
+    },
+    {
+      "epoch": 1.8733493397358942,
+      "grad_norm": 0.0011907683219760656,
+      "learning_rate": 0.00018343055451747498,
+      "loss": 23.0,
+      "step": 3121
+    },
+    {
+      "epoch": 1.8739495798319328,
+      "grad_norm": 0.0005777979386039078,
+      "learning_rate": 0.0001834201383123811,
+      "loss": 23.0,
+      "step": 3122
+    },
+    {
+      "epoch": 1.8745498199279713,
+      "grad_norm": 0.004292360506951809,
+      "learning_rate": 0.00018340971913023707,
+      "loss": 23.0,
+      "step": 3123
+    },
+    {
+      "epoch": 1.8751500600240096,
+      "grad_norm": 0.0035806214436888695,
+      "learning_rate": 0.00018339929697141474,
+      "loss": 23.0,
+      "step": 3124
+    },
+    {
+      "epoch": 1.875750300120048,
+      "grad_norm": 0.0015884626191109419,
+      "learning_rate": 0.00018338887183628599,
+      "loss": 23.0,
+      "step": 3125
+    },
+    {
+      "epoch": 1.8763505402160865,
+      "grad_norm": 0.0011369887506589293,
+      "learning_rate": 0.00018337844372522292,
+      "loss": 23.0,
+      "step": 3126
+    },
+    {
+      "epoch": 1.8769507803121248,
+      "grad_norm": 0.005162140820175409,
+      "learning_rate": 0.00018336801263859768,
+      "loss": 23.0,
+      "step": 3127
+    },
+    {
+      "epoch": 1.8775510204081631,
+      "grad_norm": 0.0024860259145498276,
+      "learning_rate": 0.00018335757857678253,
+      "loss": 23.0,
+      "step": 3128
+    },
+    {
+      "epoch": 1.8781512605042017,
+      "grad_norm": 0.0012651464203372598,
+      "learning_rate": 0.0001833471415401498,
+      "loss": 23.0,
+      "step": 3129
+    },
+    {
+      "epoch": 1.8787515006002402,
+      "grad_norm": 0.00254456396214664,
+      "learning_rate": 0.00018333670152907198,
+      "loss": 23.0,
+      "step": 3130
+    },
+    {
+      "epoch": 1.8793517406962785,
+      "grad_norm": 0.0015262291999533772,
+      "learning_rate": 0.00018332625854392164,
+      "loss": 23.0,
+      "step": 3131
+    },
+    {
+      "epoch": 1.8799519807923168,
+      "grad_norm": 0.002018503611907363,
+      "learning_rate": 0.0001833158125850715,
+      "loss": 23.0,
+      "step": 3132
+    },
+    {
+      "epoch": 1.8805522208883554,
+      "grad_norm": 0.00178581103682518,
+      "learning_rate": 0.0001833053636528943,
+      "loss": 23.0,
+      "step": 3133
+    },
+    {
+      "epoch": 1.8811524609843937,
+      "grad_norm": 0.002688503125682473,
+      "learning_rate": 0.00018329491174776295,
+      "loss": 23.0,
+      "step": 3134
+    },
+    {
+      "epoch": 1.881752701080432,
+      "grad_norm": 0.0027273113373667,
+      "learning_rate": 0.0001832844568700505,
+      "loss": 23.0,
+      "step": 3135
+    },
+    {
+      "epoch": 1.8823529411764706,
+      "grad_norm": 0.0010961064836010337,
+      "learning_rate": 0.00018327399902013,
+      "loss": 23.0,
+      "step": 3136
+    },
+    {
+      "epoch": 1.882953181272509,
+      "grad_norm": 0.0035878592170774937,
+      "learning_rate": 0.00018326353819837467,
+      "loss": 23.0,
+      "step": 3137
+    },
+    {
+      "epoch": 1.8835534213685474,
+      "grad_norm": 0.0022766878828406334,
+      "learning_rate": 0.00018325307440515785,
+      "loss": 23.0,
+      "step": 3138
+    },
+    {
+      "epoch": 1.8841536614645857,
+      "grad_norm": 0.0014452520990744233,
+      "learning_rate": 0.00018324260764085294,
+      "loss": 23.0,
+      "step": 3139
+    },
+    {
+      "epoch": 1.8847539015606243,
+      "grad_norm": 0.001565837417729199,
+      "learning_rate": 0.00018323213790583353,
+      "loss": 23.0,
+      "step": 3140
+    },
+    {
+      "epoch": 1.8853541416566628,
+      "grad_norm": 0.002488331636413932,
+      "learning_rate": 0.0001832216652004732,
+      "loss": 23.0,
+      "step": 3141
+    },
+    {
+      "epoch": 1.885954381752701,
+      "grad_norm": 0.0023180756252259016,
+      "learning_rate": 0.00018321118952514568,
+      "loss": 23.0,
+      "step": 3142
+    },
+    {
+      "epoch": 1.8865546218487395,
+      "grad_norm": 0.0013001611223444343,
+      "learning_rate": 0.0001832007108802249,
+      "loss": 23.0,
+      "step": 3143
+    },
+    {
+      "epoch": 1.887154861944778,
+      "grad_norm": 0.00461295573040843,
+      "learning_rate": 0.0001831902292660847,
+      "loss": 23.0,
+      "step": 3144
+    },
+    {
+      "epoch": 1.8877551020408163,
+      "grad_norm": 0.0016194535419344902,
+      "learning_rate": 0.00018317974468309926,
+      "loss": 23.0,
+      "step": 3145
+    },
+    {
+      "epoch": 1.8883553421368546,
+      "grad_norm": 0.002495100488886237,
+      "learning_rate": 0.00018316925713164267,
+      "loss": 23.0,
+      "step": 3146
+    },
+    {
+      "epoch": 1.8889555822328932,
+      "grad_norm": 0.0009624058147892356,
+      "learning_rate": 0.00018315876661208926,
+      "loss": 23.0,
+      "step": 3147
+    },
+    {
+      "epoch": 1.8895558223289317,
+      "grad_norm": 0.002280160551890731,
+      "learning_rate": 0.00018314827312481335,
+      "loss": 23.0,
+      "step": 3148
+    },
+    {
+      "epoch": 1.89015606242497,
+      "grad_norm": 0.001309713814407587,
+      "learning_rate": 0.00018313777667018947,
+      "loss": 23.0,
+      "step": 3149
+    },
+    {
+      "epoch": 1.8907563025210083,
+      "grad_norm": 0.00454369792714715,
+      "learning_rate": 0.0001831272772485922,
+      "loss": 23.0,
+      "step": 3150
+    },
+    {
+      "epoch": 1.8913565426170469,
+      "grad_norm": 0.0026175181847065687,
+      "learning_rate": 0.0001831167748603962,
+      "loss": 23.0,
+      "step": 3151
+    },
+    {
+      "epoch": 1.8919567827130852,
+      "grad_norm": 0.0012701862724497914,
+      "learning_rate": 0.00018310626950597637,
+      "loss": 23.0,
+      "step": 3152
+    },
+    {
+      "epoch": 1.8925570228091235,
+      "grad_norm": 0.0033725642133504152,
+      "learning_rate": 0.0001830957611857075,
+      "loss": 23.0,
+      "step": 3153
+    },
+    {
+      "epoch": 1.893157262905162,
+      "grad_norm": 0.005127619951963425,
+      "learning_rate": 0.0001830852498999647,
+      "loss": 23.0,
+      "step": 3154
+    },
+    {
+      "epoch": 1.8937575030012006,
+      "grad_norm": 0.0004975322517566383,
+      "learning_rate": 0.00018307473564912303,
+      "loss": 23.0,
+      "step": 3155
+    },
+    {
+      "epoch": 1.894357743097239,
+      "grad_norm": 0.002670707879588008,
+      "learning_rate": 0.00018306421843355773,
+      "loss": 23.0,
+      "step": 3156
+    },
+    {
+      "epoch": 1.8949579831932772,
+      "grad_norm": 0.003962348680943251,
+      "learning_rate": 0.00018305369825364417,
+      "loss": 23.0,
+      "step": 3157
+    },
+    {
+      "epoch": 1.8955582232893158,
+      "grad_norm": 0.000557166466023773,
+      "learning_rate": 0.00018304317510975775,
+      "loss": 23.0,
+      "step": 3158
+    },
+    {
+      "epoch": 1.896158463385354,
+      "grad_norm": 0.0021320811938494444,
+      "learning_rate": 0.00018303264900227398,
+      "loss": 23.0,
+      "step": 3159
+    },
+    {
+      "epoch": 1.8967587034813924,
+      "grad_norm": 0.003195348661392927,
+      "learning_rate": 0.0001830221199315686,
+      "loss": 23.0,
+      "step": 3160
+    },
+    {
+      "epoch": 1.897358943577431,
+      "grad_norm": 0.0013635127106681466,
+      "learning_rate": 0.0001830115878980173,
+      "loss": 23.0,
+      "step": 3161
+    },
+    {
+      "epoch": 1.8979591836734695,
+      "grad_norm": 0.002584997797384858,
+      "learning_rate": 0.00018300105290199595,
+      "loss": 23.0,
+      "step": 3162
+    },
+    {
+      "epoch": 1.8985594237695078,
+      "grad_norm": 0.0012520054588094354,
+      "learning_rate": 0.00018299051494388054,
+      "loss": 23.0,
+      "step": 3163
+    },
+    {
+      "epoch": 1.8991596638655461,
+      "grad_norm": 0.0018940556328743696,
+      "learning_rate": 0.00018297997402404715,
+      "loss": 23.0,
+      "step": 3164
+    },
+    {
+      "epoch": 1.8997599039615847,
+      "grad_norm": 0.0019348510541021824,
+      "learning_rate": 0.00018296943014287193,
+      "loss": 23.0,
+      "step": 3165
+    },
+    {
+      "epoch": 1.9003601440576232,
+      "grad_norm": 0.0018156596925109625,
+      "learning_rate": 0.00018295888330073113,
+      "loss": 23.0,
+      "step": 3166
+    },
+    {
+      "epoch": 1.9009603841536613,
+      "grad_norm": 0.0035436146426945925,
+      "learning_rate": 0.0001829483334980012,
+      "loss": 23.0,
+      "step": 3167
+    },
+    {
+      "epoch": 1.9015606242496998,
+      "grad_norm": 0.0012572517152875662,
+      "learning_rate": 0.00018293778073505865,
+      "loss": 23.0,
+      "step": 3168
+    },
+    {
+      "epoch": 1.9021608643457384,
+      "grad_norm": 0.0008448131266050041,
+      "learning_rate": 0.00018292722501228004,
+      "loss": 23.0,
+      "step": 3169
+    },
+    {
+      "epoch": 1.9027611044417767,
+      "grad_norm": 0.0037472823169082403,
+      "learning_rate": 0.00018291666633004204,
+      "loss": 23.0,
+      "step": 3170
+    },
+    {
+      "epoch": 1.903361344537815,
+      "grad_norm": 0.0018287114799022675,
+      "learning_rate": 0.00018290610468872155,
+      "loss": 23.0,
+      "step": 3171
+    },
+    {
+      "epoch": 1.9039615846338536,
+      "grad_norm": 0.0022742743603885174,
+      "learning_rate": 0.0001828955400886954,
+      "loss": 23.0,
+      "step": 3172
+    },
+    {
+      "epoch": 1.904561824729892,
+      "grad_norm": 0.0018581197364255786,
+      "learning_rate": 0.0001828849725303407,
+      "loss": 23.0,
+      "step": 3173
+    },
+    {
+      "epoch": 1.9051620648259304,
+      "grad_norm": 0.001101133180782199,
+      "learning_rate": 0.0001828744020140345,
+      "loss": 23.0,
+      "step": 3174
+    },
+    {
+      "epoch": 1.9057623049219687,
+      "grad_norm": 0.0020204149186611176,
+      "learning_rate": 0.0001828638285401541,
+      "loss": 23.0,
+      "step": 3175
+    },
+    {
+      "epoch": 1.9063625450180073,
+      "grad_norm": 0.0028773958329111338,
+      "learning_rate": 0.0001828532521090768,
+      "loss": 23.0,
+      "step": 3176
+    },
+    {
+      "epoch": 1.9069627851140456,
+      "grad_norm": 0.0018713538302108645,
+      "learning_rate": 0.00018284267272118004,
+      "loss": 23.0,
+      "step": 3177
+    },
+    {
+      "epoch": 1.907563025210084,
+      "grad_norm": 0.0018324162811040878,
+      "learning_rate": 0.00018283209037684141,
+      "loss": 23.0,
+      "step": 3178
+    },
+    {
+      "epoch": 1.9081632653061225,
+      "grad_norm": 0.002509085228666663,
+      "learning_rate": 0.00018282150507643855,
+      "loss": 23.0,
+      "step": 3179
+    },
+    {
+      "epoch": 1.908763505402161,
+      "grad_norm": 0.002151481108739972,
+      "learning_rate": 0.0001828109168203492,
+      "loss": 23.0,
+      "step": 3180
+    },
+    {
+      "epoch": 1.9093637454981993,
+      "grad_norm": 0.0014471776084974408,
+      "learning_rate": 0.00018280032560895123,
+      "loss": 23.0,
+      "step": 3181
+    },
+    {
+      "epoch": 1.9099639855942376,
+      "grad_norm": 0.0023641546722501516,
+      "learning_rate": 0.00018278973144262266,
+      "loss": 23.0,
+      "step": 3182
+    },
+    {
+      "epoch": 1.9105642256902762,
+      "grad_norm": 0.003392763901501894,
+      "learning_rate": 0.0001827791343217415,
+      "loss": 23.0,
+      "step": 3183
+    },
+    {
+      "epoch": 1.9111644657863145,
+      "grad_norm": 0.003529018722474575,
+      "learning_rate": 0.00018276853424668602,
+      "loss": 23.0,
+      "step": 3184
+    },
+    {
+      "epoch": 1.9117647058823528,
+      "grad_norm": 0.003656035289168358,
+      "learning_rate": 0.0001827579312178344,
+      "loss": 23.0,
+      "step": 3185
+    },
+    {
+      "epoch": 1.9123649459783914,
+      "grad_norm": 0.0014189134817570448,
+      "learning_rate": 0.00018274732523556513,
+      "loss": 23.0,
+      "step": 3186
+    },
+    {
+      "epoch": 1.91296518607443,
+      "grad_norm": 0.0013264564331620932,
+      "learning_rate": 0.00018273671630025665,
+      "loss": 23.0,
+      "step": 3187
+    },
+    {
+      "epoch": 1.9135654261704682,
+      "grad_norm": 0.004117329139262438,
+      "learning_rate": 0.00018272610441228763,
+      "loss": 23.0,
+      "step": 3188
+    },
+    {
+      "epoch": 1.9141656662665065,
+      "grad_norm": 0.0024092942476272583,
+      "learning_rate": 0.0001827154895720367,
+      "loss": 23.0,
+      "step": 3189
+    },
+    {
+      "epoch": 1.914765906362545,
+      "grad_norm": 0.001247992506250739,
+      "learning_rate": 0.00018270487177988272,
+      "loss": 23.0,
+      "step": 3190
+    },
+    {
+      "epoch": 1.9153661464585834,
+      "grad_norm": 0.006278992164880037,
+      "learning_rate": 0.00018269425103620464,
+      "loss": 23.0,
+      "step": 3191
+    },
+    {
+      "epoch": 1.9159663865546217,
+      "grad_norm": 0.001128924312070012,
+      "learning_rate": 0.00018268362734138142,
+      "loss": 23.0,
+      "step": 3192
+    },
+    {
+      "epoch": 1.9165666266506602,
+      "grad_norm": 0.004643549211323261,
+      "learning_rate": 0.00018267300069579222,
+      "loss": 23.0,
+      "step": 3193
+    },
+    {
+      "epoch": 1.9171668667466988,
+      "grad_norm": 0.002494530752301216,
+      "learning_rate": 0.00018266237109981632,
+      "loss": 23.0,
+      "step": 3194
+    },
+    {
+      "epoch": 1.917767106842737,
+      "grad_norm": 0.00441815285012126,
+      "learning_rate": 0.000182651738553833,
+      "loss": 23.0,
+      "step": 3195
+    },
+    {
+      "epoch": 1.9183673469387754,
+      "grad_norm": 0.0028012660332024097,
+      "learning_rate": 0.00018264110305822178,
+      "loss": 23.0,
+      "step": 3196
+    },
+    {
+      "epoch": 1.918967587034814,
+      "grad_norm": 0.0017156752292066813,
+      "learning_rate": 0.00018263046461336214,
+      "loss": 23.0,
+      "step": 3197
+    },
+    {
+      "epoch": 1.9195678271308525,
+      "grad_norm": 0.002733529545366764,
+      "learning_rate": 0.00018261982321963379,
+      "loss": 23.0,
+      "step": 3198
+    },
+    {
+      "epoch": 1.9201680672268906,
+      "grad_norm": 0.0011137155815958977,
+      "learning_rate": 0.00018260917887741643,
+      "loss": 23.0,
+      "step": 3199
+    },
+    {
+      "epoch": 1.9207683073229291,
+      "grad_norm": 0.0014791876310482621,
+      "learning_rate": 0.00018259853158708998,
+      "loss": 23.0,
+      "step": 3200
+    },
+    {
+      "epoch": 1.9213685474189677,
+      "grad_norm": 0.0009261185768991709,
+      "learning_rate": 0.00018258788134903445,
+      "loss": 23.0,
+      "step": 3201
+    },
+    {
+      "epoch": 1.921968787515006,
+      "grad_norm": 0.0030771689489483833,
+      "learning_rate": 0.00018257722816362985,
+      "loss": 23.0,
+      "step": 3202
+    },
+    {
+      "epoch": 1.9225690276110443,
+      "grad_norm": 0.003100763075053692,
+      "learning_rate": 0.00018256657203125637,
+      "loss": 23.0,
+      "step": 3203
+    },
+    {
+      "epoch": 1.9231692677070829,
+      "grad_norm": 0.002674756571650505,
+      "learning_rate": 0.00018255591295229437,
+      "loss": 23.0,
+      "step": 3204
+    },
+    {
+      "epoch": 1.9237695078031214,
+      "grad_norm": 0.0022848814260214567,
+      "learning_rate": 0.00018254525092712415,
+      "loss": 23.0,
+      "step": 3205
+    },
+    {
+      "epoch": 1.9243697478991597,
+      "grad_norm": 0.0018523391336202621,
+      "learning_rate": 0.00018253458595612627,
+      "loss": 23.0,
+      "step": 3206
+    },
+    {
+      "epoch": 1.924969987995198,
+      "grad_norm": 0.0009723806870169938,
+      "learning_rate": 0.00018252391803968133,
+      "loss": 23.0,
+      "step": 3207
+    },
+    {
+      "epoch": 1.9255702280912366,
+      "grad_norm": 0.001458384795114398,
+      "learning_rate": 0.00018251324717817005,
+      "loss": 23.0,
+      "step": 3208
+    },
+    {
+      "epoch": 1.9261704681872749,
+      "grad_norm": 0.0013265210436657071,
+      "learning_rate": 0.0001825025733719732,
+      "loss": 23.0,
+      "step": 3209
+    },
+    {
+      "epoch": 1.9267707082833132,
+      "grad_norm": 0.0010744367027655244,
+      "learning_rate": 0.00018249189662147171,
+      "loss": 23.0,
+      "step": 3210
+    },
+    {
+      "epoch": 1.9273709483793517,
+      "grad_norm": 0.0026470967568457127,
+      "learning_rate": 0.00018248121692704668,
+      "loss": 23.0,
+      "step": 3211
+    },
+    {
+      "epoch": 1.9279711884753903,
+      "grad_norm": 0.0016062292270362377,
+      "learning_rate": 0.00018247053428907913,
+      "loss": 23.0,
+      "step": 3212
+    },
+    {
+      "epoch": 1.9285714285714286,
+      "grad_norm": 0.000824375543743372,
+      "learning_rate": 0.00018245984870795038,
+      "loss": 23.0,
+      "step": 3213
+    },
+    {
+      "epoch": 1.929171668667467,
+      "grad_norm": 0.002928749890998006,
+      "learning_rate": 0.00018244916018404173,
+      "loss": 23.0,
+      "step": 3214
+    },
+    {
+      "epoch": 1.9297719087635055,
+      "grad_norm": 0.002207791665568948,
+      "learning_rate": 0.00018243846871773465,
+      "loss": 23.0,
+      "step": 3215
+    },
+    {
+      "epoch": 1.9303721488595438,
+      "grad_norm": 0.0028601568192243576,
+      "learning_rate": 0.00018242777430941068,
+      "loss": 23.0,
+      "step": 3216
+    },
+    {
+      "epoch": 1.930972388955582,
+      "grad_norm": 0.0021290185395628214,
+      "learning_rate": 0.00018241707695945144,
+      "loss": 23.0,
+      "step": 3217
+    },
+    {
+      "epoch": 1.9315726290516206,
+      "grad_norm": 0.000839258951600641,
+      "learning_rate": 0.00018240637666823876,
+      "loss": 23.0,
+      "step": 3218
+    },
+    {
+      "epoch": 1.9321728691476592,
+      "grad_norm": 0.002701072720810771,
+      "learning_rate": 0.00018239567343615443,
+      "loss": 23.0,
+      "step": 3219
+    },
+    {
+      "epoch": 1.9327731092436975,
+      "grad_norm": 0.0020356529857963324,
+      "learning_rate": 0.00018238496726358052,
+      "loss": 23.0,
+      "step": 3220
+    },
+    {
+      "epoch": 1.9333733493397358,
+      "grad_norm": 0.0014873469481244683,
+      "learning_rate": 0.00018237425815089903,
+      "loss": 23.0,
+      "step": 3221
+    },
+    {
+      "epoch": 1.9339735894357744,
+      "grad_norm": 0.002182344440370798,
+      "learning_rate": 0.00018236354609849214,
+      "loss": 23.0,
+      "step": 3222
+    },
+    {
+      "epoch": 1.9345738295318127,
+      "grad_norm": 0.0016528373816981912,
+      "learning_rate": 0.00018235283110674218,
+      "loss": 23.0,
+      "step": 3223
+    },
+    {
+      "epoch": 1.935174069627851,
+      "grad_norm": 0.0012866542674601078,
+      "learning_rate": 0.0001823421131760315,
+      "loss": 23.0,
+      "step": 3224
+    },
+    {
+      "epoch": 1.9357743097238895,
+      "grad_norm": 0.0019892966374754906,
+      "learning_rate": 0.00018233139230674258,
+      "loss": 23.0,
+      "step": 3225
+    },
+    {
+      "epoch": 1.936374549819928,
+      "grad_norm": 0.002182098338380456,
+      "learning_rate": 0.0001823206684992581,
+      "loss": 23.0,
+      "step": 3226
+    },
+    {
+      "epoch": 1.9369747899159664,
+      "grad_norm": 0.0021431762725114822,
+      "learning_rate": 0.0001823099417539607,
+      "loss": 23.0,
+      "step": 3227
+    },
+    {
+      "epoch": 1.9375750300120047,
+      "grad_norm": 0.0024950047954916954,
+      "learning_rate": 0.00018229921207123318,
+      "loss": 23.0,
+      "step": 3228
+    },
+    {
+      "epoch": 1.9381752701080432,
+      "grad_norm": 0.0017660352168604732,
+      "learning_rate": 0.0001822884794514585,
+      "loss": 23.0,
+      "step": 3229
+    },
+    {
+      "epoch": 1.9387755102040818,
+      "grad_norm": 0.0065404558554291725,
+      "learning_rate": 0.00018227774389501967,
+      "loss": 23.0,
+      "step": 3230
+    },
+    {
+      "epoch": 1.9393757503001199,
+      "grad_norm": 0.0011474026832729578,
+      "learning_rate": 0.00018226700540229976,
+      "loss": 23.0,
+      "step": 3231
+    },
+    {
+      "epoch": 1.9399759903961584,
+      "grad_norm": 0.00047521217493340373,
+      "learning_rate": 0.0001822562639736821,
+      "loss": 23.0,
+      "step": 3232
+    },
+    {
+      "epoch": 1.940576230492197,
+      "grad_norm": 0.0019478441681712866,
+      "learning_rate": 0.00018224551960954995,
+      "loss": 23.0,
+      "step": 3233
+    },
+    {
+      "epoch": 1.9411764705882353,
+      "grad_norm": 0.0010433909483253956,
+      "learning_rate": 0.00018223477231028677,
+      "loss": 23.0,
+      "step": 3234
+    },
+    {
+      "epoch": 1.9417767106842736,
+      "grad_norm": 0.0018290705047547817,
+      "learning_rate": 0.00018222402207627606,
+      "loss": 23.0,
+      "step": 3235
+    },
+    {
+      "epoch": 1.9423769507803121,
+      "grad_norm": 0.0017392985755577683,
+      "learning_rate": 0.00018221326890790153,
+      "loss": 23.0,
+      "step": 3236
+    },
+    {
+      "epoch": 1.9429771908763507,
+      "grad_norm": 0.002610771218314767,
+      "learning_rate": 0.00018220251280554696,
+      "loss": 23.0,
+      "step": 3237
+    },
+    {
+      "epoch": 1.943577430972389,
+      "grad_norm": 0.0016300787683576345,
+      "learning_rate": 0.00018219175376959606,
+      "loss": 23.0,
+      "step": 3238
+    },
+    {
+      "epoch": 1.9441776710684273,
+      "grad_norm": 0.0022878015879541636,
+      "learning_rate": 0.00018218099180043295,
+      "loss": 23.0,
+      "step": 3239
+    },
+    {
+      "epoch": 1.9447779111644659,
+      "grad_norm": 0.0036289305426180363,
+      "learning_rate": 0.00018217022689844163,
+      "loss": 23.0,
+      "step": 3240
+    },
+    {
+      "epoch": 1.9453781512605042,
+      "grad_norm": 0.0009557732264511287,
+      "learning_rate": 0.00018215945906400628,
+      "loss": 23.0,
+      "step": 3241
+    },
+    {
+      "epoch": 1.9459783913565425,
+      "grad_norm": 0.0023918780498206615,
+      "learning_rate": 0.0001821486882975112,
+      "loss": 23.0,
+      "step": 3242
+    },
+    {
+      "epoch": 1.946578631452581,
+      "grad_norm": 0.001407388597726822,
+      "learning_rate": 0.0001821379145993407,
+      "loss": 23.0,
+      "step": 3243
+    },
+    {
+      "epoch": 1.9471788715486196,
+      "grad_norm": 0.0011877256911247969,
+      "learning_rate": 0.00018212713796987936,
+      "loss": 23.0,
+      "step": 3244
+    },
+    {
+      "epoch": 1.947779111644658,
+      "grad_norm": 0.001021384377963841,
+      "learning_rate": 0.0001821163584095117,
+      "loss": 23.0,
+      "step": 3245
+    },
+    {
+      "epoch": 1.9483793517406962,
+      "grad_norm": 0.0026115647051483393,
+      "learning_rate": 0.0001821055759186224,
+      "loss": 23.0,
+      "step": 3246
+    },
+    {
+      "epoch": 1.9489795918367347,
+      "grad_norm": 0.001517315162345767,
+      "learning_rate": 0.00018209479049759636,
+      "loss": 23.0,
+      "step": 3247
+    },
+    {
+      "epoch": 1.949579831932773,
+      "grad_norm": 0.003172236494719982,
+      "learning_rate": 0.00018208400214681837,
+      "loss": 23.0,
+      "step": 3248
+    },
+    {
+      "epoch": 1.9501800720288114,
+      "grad_norm": 0.001906992169097066,
+      "learning_rate": 0.00018207321086667353,
+      "loss": 23.0,
+      "step": 3249
+    },
+    {
+      "epoch": 1.95078031212485,
+      "grad_norm": 0.003104976611211896,
+      "learning_rate": 0.00018206241665754688,
+      "loss": 23.0,
+      "step": 3250
+    },
+    {
+      "epoch": 1.9513805522208885,
+      "grad_norm": 0.0015054944669827819,
+      "learning_rate": 0.00018205161951982367,
+      "loss": 23.0,
+      "step": 3251
+    },
+    {
+      "epoch": 1.9519807923169268,
+      "grad_norm": 0.0020006780978292227,
+      "learning_rate": 0.00018204081945388922,
+      "loss": 23.0,
+      "step": 3252
+    },
+    {
+      "epoch": 1.952581032412965,
+      "grad_norm": 0.0018055624095723033,
+      "learning_rate": 0.00018203001646012899,
+      "loss": 23.0,
+      "step": 3253
+    },
+    {
+      "epoch": 1.9531812725090036,
+      "grad_norm": 0.0017827142728492618,
+      "learning_rate": 0.00018201921053892845,
+      "loss": 23.0,
+      "step": 3254
+    },
+    {
+      "epoch": 1.9537815126050422,
+      "grad_norm": 0.0020725359208881855,
+      "learning_rate": 0.0001820084016906733,
+      "loss": 23.0,
+      "step": 3255
+    },
+    {
+      "epoch": 1.9543817527010803,
+      "grad_norm": 0.001384815899655223,
+      "learning_rate": 0.0001819975899157492,
+      "loss": 23.0,
+      "step": 3256
+    },
+    {
+      "epoch": 1.9549819927971188,
+      "grad_norm": 0.002811410464346409,
+      "learning_rate": 0.00018198677521454206,
+      "loss": 23.0,
+      "step": 3257
+    },
+    {
+      "epoch": 1.9555822328931574,
+      "grad_norm": 0.0008216790156438947,
+      "learning_rate": 0.0001819759575874378,
+      "loss": 23.0,
+      "step": 3258
+    },
+    {
+      "epoch": 1.9561824729891957,
+      "grad_norm": 0.0014283070340752602,
+      "learning_rate": 0.00018196513703482248,
+      "loss": 23.0,
+      "step": 3259
+    },
+    {
+      "epoch": 1.956782713085234,
+      "grad_norm": 0.0016851375112310052,
+      "learning_rate": 0.00018195431355708225,
+      "loss": 23.0,
+      "step": 3260
+    },
+    {
+      "epoch": 1.9573829531812725,
+      "grad_norm": 0.0019182129763066769,
+      "learning_rate": 0.0001819434871546034,
+      "loss": 23.0,
+      "step": 3261
+    },
+    {
+      "epoch": 1.957983193277311,
+      "grad_norm": 0.0008913344354368746,
+      "learning_rate": 0.00018193265782777228,
+      "loss": 23.0,
+      "step": 3262
+    },
+    {
+      "epoch": 1.9585834333733494,
+      "grad_norm": 0.0013015334261581302,
+      "learning_rate": 0.00018192182557697534,
+      "loss": 23.0,
+      "step": 3263
+    },
+    {
+      "epoch": 1.9591836734693877,
+      "grad_norm": 0.0009710061131045222,
+      "learning_rate": 0.00018191099040259918,
+      "loss": 23.0,
+      "step": 3264
+    },
+    {
+      "epoch": 1.9597839135654262,
+      "grad_norm": 0.0013321270234882832,
+      "learning_rate": 0.0001819001523050305,
+      "loss": 23.0,
+      "step": 3265
+    },
+    {
+      "epoch": 1.9603841536614646,
+      "grad_norm": 0.004230269230902195,
+      "learning_rate": 0.00018188931128465603,
+      "loss": 23.0,
+      "step": 3266
+    },
+    {
+      "epoch": 1.9609843937575029,
+      "grad_norm": 0.0016738398699089885,
+      "learning_rate": 0.00018187846734186267,
+      "loss": 23.0,
+      "step": 3267
+    },
+    {
+      "epoch": 1.9615846338535414,
+      "grad_norm": 0.0009718580404296517,
+      "learning_rate": 0.00018186762047703745,
+      "loss": 23.0,
+      "step": 3268
+    },
+    {
+      "epoch": 1.96218487394958,
+      "grad_norm": 0.0022842460311949253,
+      "learning_rate": 0.00018185677069056742,
+      "loss": 23.0,
+      "step": 3269
+    },
+    {
+      "epoch": 1.9627851140456183,
+      "grad_norm": 0.0014329022960737348,
+      "learning_rate": 0.00018184591798283984,
+      "loss": 23.0,
+      "step": 3270
+    },
+    {
+      "epoch": 1.9633853541416566,
+      "grad_norm": 0.0033265934325754642,
+      "learning_rate": 0.00018183506235424198,
+      "loss": 23.0,
+      "step": 3271
+    },
+    {
+      "epoch": 1.9639855942376951,
+      "grad_norm": 0.0007821978651918471,
+      "learning_rate": 0.0001818242038051612,
+      "loss": 23.0,
+      "step": 3272
+    },
+    {
+      "epoch": 1.9645858343337335,
+      "grad_norm": 0.0015688884304836392,
+      "learning_rate": 0.00018181334233598507,
+      "loss": 23.0,
+      "step": 3273
+    },
+    {
+      "epoch": 1.9651860744297718,
+      "grad_norm": 0.0008193565299734473,
+      "learning_rate": 0.00018180247794710125,
+      "loss": 23.0,
+      "step": 3274
+    },
+    {
+      "epoch": 1.9657863145258103,
+      "grad_norm": 0.002213628264144063,
+      "learning_rate": 0.0001817916106388974,
+      "loss": 23.0,
+      "step": 3275
+    },
+    {
+      "epoch": 1.9663865546218489,
+      "grad_norm": 0.0014024798292666674,
+      "learning_rate": 0.00018178074041176135,
+      "loss": 23.0,
+      "step": 3276
+    },
+    {
+      "epoch": 1.9669867947178872,
+      "grad_norm": 0.0037618144415318966,
+      "learning_rate": 0.00018176986726608103,
+      "loss": 23.0,
+      "step": 3277
+    },
+    {
+      "epoch": 1.9675870348139255,
+      "grad_norm": 0.0024038751143962145,
+      "learning_rate": 0.00018175899120224449,
+      "loss": 23.0,
+      "step": 3278
+    },
+    {
+      "epoch": 1.968187274909964,
+      "grad_norm": 0.0033458671532571316,
+      "learning_rate": 0.00018174811222063987,
+      "loss": 23.0,
+      "step": 3279
+    },
+    {
+      "epoch": 1.9687875150060024,
+      "grad_norm": 0.0019863531924784184,
+      "learning_rate": 0.0001817372303216554,
+      "loss": 23.0,
+      "step": 3280
+    },
+    {
+      "epoch": 1.9693877551020407,
+      "grad_norm": 0.0015641896752640605,
+      "learning_rate": 0.0001817263455056794,
+      "loss": 23.0,
+      "step": 3281
+    },
+    {
+      "epoch": 1.9699879951980792,
+      "grad_norm": 0.0012357387458905578,
+      "learning_rate": 0.00018171545777310042,
+      "loss": 23.0,
+      "step": 3282
+    },
+    {
+      "epoch": 1.9705882352941178,
+      "grad_norm": 0.002113505732268095,
+      "learning_rate": 0.00018170456712430691,
+      "loss": 23.0,
+      "step": 3283
+    },
+    {
+      "epoch": 1.971188475390156,
+      "grad_norm": 0.002042030915617943,
+      "learning_rate": 0.0001816936735596876,
+      "loss": 23.0,
+      "step": 3284
+    },
+    {
+      "epoch": 1.9717887154861944,
+      "grad_norm": 0.0012755439383909106,
+      "learning_rate": 0.00018168277707963118,
+      "loss": 23.0,
+      "step": 3285
+    },
+    {
+      "epoch": 1.972388955582233,
+      "grad_norm": 0.0019094778690487146,
+      "learning_rate": 0.0001816718776845266,
+      "loss": 23.0,
+      "step": 3286
+    },
+    {
+      "epoch": 1.9729891956782715,
+      "grad_norm": 0.00413705175742507,
+      "learning_rate": 0.00018166097537476277,
+      "loss": 23.0,
+      "step": 3287
+    },
+    {
+      "epoch": 1.9735894357743096,
+      "grad_norm": 0.0020690206438302994,
+      "learning_rate": 0.00018165007015072876,
+      "loss": 23.0,
+      "step": 3288
+    },
+    {
+      "epoch": 1.974189675870348,
+      "grad_norm": 0.00169268692843616,
+      "learning_rate": 0.00018163916201281382,
+      "loss": 23.0,
+      "step": 3289
+    },
+    {
+      "epoch": 1.9747899159663866,
+      "grad_norm": 0.001092193415388465,
+      "learning_rate": 0.00018162825096140718,
+      "loss": 23.0,
+      "step": 3290
+    },
+    {
+      "epoch": 1.975390156062425,
+      "grad_norm": 0.0015564619097858667,
+      "learning_rate": 0.0001816173369968982,
+      "loss": 23.0,
+      "step": 3291
+    },
+    {
+      "epoch": 1.9759903961584633,
+      "grad_norm": 0.0038381817284971476,
+      "learning_rate": 0.00018160642011967644,
+      "loss": 23.0,
+      "step": 3292
+    },
+    {
+      "epoch": 1.9765906362545018,
+      "grad_norm": 0.0013099567731842399,
+      "learning_rate": 0.00018159550033013144,
+      "loss": 23.0,
+      "step": 3293
+    },
+    {
+      "epoch": 1.9771908763505404,
+      "grad_norm": 0.00048246965161524713,
+      "learning_rate": 0.00018158457762865296,
+      "loss": 23.0,
+      "step": 3294
+    },
+    {
+      "epoch": 1.9777911164465787,
+      "grad_norm": 0.0024292506277561188,
+      "learning_rate": 0.00018157365201563073,
+      "loss": 23.0,
+      "step": 3295
+    },
+    {
+      "epoch": 1.978391356542617,
+      "grad_norm": 0.0020980066619813442,
+      "learning_rate": 0.0001815627234914547,
+      "loss": 23.0,
+      "step": 3296
+    },
+    {
+      "epoch": 1.9789915966386555,
+      "grad_norm": 0.0009206788381561637,
+      "learning_rate": 0.00018155179205651487,
+      "loss": 23.0,
+      "step": 3297
+    },
+    {
+      "epoch": 1.9795918367346939,
+      "grad_norm": 0.0014182684244588017,
+      "learning_rate": 0.00018154085771120138,
+      "loss": 23.0,
+      "step": 3298
+    },
+    {
+      "epoch": 1.9801920768307322,
+      "grad_norm": 0.0010655620135366917,
+      "learning_rate": 0.0001815299204559044,
+      "loss": 23.0,
+      "step": 3299
+    },
+    {
+      "epoch": 1.9807923169267707,
+      "grad_norm": 0.0018564078491181135,
+      "learning_rate": 0.0001815189802910143,
+      "loss": 23.0,
+      "step": 3300
+    },
+    {
+      "epoch": 1.9813925570228093,
+      "grad_norm": 0.0016347630880773067,
+      "learning_rate": 0.00018150803721692148,
+      "loss": 23.0,
+      "step": 3301
+    },
+    {
+      "epoch": 1.9819927971188476,
+      "grad_norm": 0.0021132517140358686,
+      "learning_rate": 0.00018149709123401646,
+      "loss": 23.0,
+      "step": 3302
+    },
+    {
+      "epoch": 1.982593037214886,
+      "grad_norm": 0.002489116508513689,
+      "learning_rate": 0.0001814861423426899,
+      "loss": 23.0,
+      "step": 3303
+    },
+    {
+      "epoch": 1.9831932773109244,
+      "grad_norm": 0.0014195957919582725,
+      "learning_rate": 0.0001814751905433325,
+      "loss": 23.0,
+      "step": 3304
+    },
+    {
+      "epoch": 1.9837935174069627,
+      "grad_norm": 0.0011736804153770208,
+      "learning_rate": 0.00018146423583633518,
+      "loss": 23.0,
+      "step": 3305
+    },
+    {
+      "epoch": 1.984393757503001,
+      "grad_norm": 0.002241331385448575,
+      "learning_rate": 0.00018145327822208883,
+      "loss": 23.0,
+      "step": 3306
+    },
+    {
+      "epoch": 1.9849939975990396,
+      "grad_norm": 0.0017535320948809385,
+      "learning_rate": 0.00018144231770098447,
+      "loss": 23.0,
+      "step": 3307
+    },
+    {
+      "epoch": 1.9855942376950781,
+      "grad_norm": 0.0021708435378968716,
+      "learning_rate": 0.00018143135427341328,
+      "loss": 23.0,
+      "step": 3308
+    },
+    {
+      "epoch": 1.9861944777911165,
+      "grad_norm": 0.0023300815373659134,
+      "learning_rate": 0.00018142038793976655,
+      "loss": 23.0,
+      "step": 3309
+    },
+    {
+      "epoch": 1.9867947178871548,
+      "grad_norm": 0.0014577904948964715,
+      "learning_rate": 0.0001814094187004356,
+      "loss": 23.0,
+      "step": 3310
+    },
+    {
+      "epoch": 1.9873949579831933,
+      "grad_norm": 0.0024585998617112637,
+      "learning_rate": 0.00018139844655581193,
+      "loss": 23.0,
+      "step": 3311
+    },
+    {
+      "epoch": 1.9879951980792316,
+      "grad_norm": 0.0038247143384069204,
+      "learning_rate": 0.00018138747150628706,
+      "loss": 23.0,
+      "step": 3312
+    },
+    {
+      "epoch": 1.98859543817527,
+      "grad_norm": 0.0011696047149598598,
+      "learning_rate": 0.0001813764935522527,
+      "loss": 23.0,
+      "step": 3313
+    },
+    {
+      "epoch": 1.9891956782713085,
+      "grad_norm": 0.0028556687757372856,
+      "learning_rate": 0.00018136551269410062,
+      "loss": 23.0,
+      "step": 3314
+    },
+    {
+      "epoch": 1.989795918367347,
+      "grad_norm": 0.0023186022881418467,
+      "learning_rate": 0.00018135452893222262,
+      "loss": 23.0,
+      "step": 3315
+    },
+    {
+      "epoch": 1.9903961584633854,
+      "grad_norm": 0.005861006677150726,
+      "learning_rate": 0.0001813435422670108,
+      "loss": 23.0,
+      "step": 3316
+    },
+    {
+      "epoch": 1.9909963985594237,
+      "grad_norm": 0.0006705807754769921,
+      "learning_rate": 0.00018133255269885723,
+      "loss": 23.0,
+      "step": 3317
+    },
+    {
+      "epoch": 1.9915966386554622,
+      "grad_norm": 0.0016647104639559984,
+      "learning_rate": 0.00018132156022815404,
+      "loss": 23.0,
+      "step": 3318
+    },
+    {
+      "epoch": 1.9921968787515008,
+      "grad_norm": 0.0013882304774597287,
+      "learning_rate": 0.00018131056485529352,
+      "loss": 23.0,
+      "step": 3319
+    },
+    {
+      "epoch": 1.9927971188475389,
+      "grad_norm": 0.0024001754354685545,
+      "learning_rate": 0.00018129956658066812,
+      "loss": 23.0,
+      "step": 3320
+    },
+    {
+      "epoch": 1.9933973589435774,
+      "grad_norm": 0.0018285515252500772,
+      "learning_rate": 0.0001812885654046703,
+      "loss": 23.0,
+      "step": 3321
+    },
+    {
+      "epoch": 1.993997599039616,
+      "grad_norm": 0.0015243416419252753,
+      "learning_rate": 0.00018127756132769272,
+      "loss": 23.0,
+      "step": 3322
+    },
+    {
+      "epoch": 1.9945978391356542,
+      "grad_norm": 0.005187567323446274,
+      "learning_rate": 0.00018126655435012803,
+      "loss": 23.0,
+      "step": 3323
+    },
+    {
+      "epoch": 1.9951980792316926,
+      "grad_norm": 0.0017248830990865827,
+      "learning_rate": 0.00018125554447236906,
+      "loss": 23.0,
+      "step": 3324
+    },
+    {
+      "epoch": 1.995798319327731,
+      "grad_norm": 0.0024234713055193424,
+      "learning_rate": 0.00018124453169480868,
+      "loss": 23.0,
+      "step": 3325
+    },
+    {
+      "epoch": 1.9963985594237696,
+      "grad_norm": 0.0025638325605541468,
+      "learning_rate": 0.00018123351601784,
+      "loss": 23.0,
+      "step": 3326
+    },
+    {
+      "epoch": 1.996998799519808,
+      "grad_norm": 0.002679846016690135,
+      "learning_rate": 0.00018122249744185605,
+      "loss": 23.0,
+      "step": 3327
+    },
+    {
+      "epoch": 1.9975990396158463,
+      "grad_norm": 0.003047887934371829,
+      "learning_rate": 0.00018121147596725015,
+      "loss": 23.0,
+      "step": 3328
+    },
+    {
+      "epoch": 1.9981992797118848,
+      "grad_norm": 0.0017722995253279805,
+      "learning_rate": 0.00018120045159441555,
+      "loss": 23.0,
+      "step": 3329
+    },
+    {
+      "epoch": 1.9987995198079231,
+      "grad_norm": 0.0012083728797733784,
+      "learning_rate": 0.00018118942432374569,
+      "loss": 23.0,
+      "step": 3330
+    },
+    {
+      "epoch": 1.9993997599039615,
+      "grad_norm": 0.0022785505279898643,
+      "learning_rate": 0.00018117839415563412,
+      "loss": 23.0,
+      "step": 3331
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.0014194271061569452,
+      "learning_rate": 0.0001811673610904745,
+      "loss": 23.0,
+      "step": 3332
+    },
+    {
+      "epoch": 2.0006002400960385,
+      "grad_norm": 0.0032136687077581882,
+      "learning_rate": 0.0001811563251286605,
+      "loss": 23.0,
+      "step": 3333
+    },
+    {
+      "epoch": 2.0012004801920766,
+      "grad_norm": 0.00217724172398448,
+      "learning_rate": 0.00018114528627058605,
+      "loss": 23.0,
+      "step": 3334
+    },
+    {
+      "epoch": 2.001800720288115,
+      "grad_norm": 0.0017727838130667806,
+      "learning_rate": 0.00018113424451664508,
+      "loss": 23.0,
+      "step": 3335
+    },
+    {
+      "epoch": 2.0024009603841537,
+      "grad_norm": 0.00423310836777091,
+      "learning_rate": 0.00018112319986723162,
+      "loss": 23.0,
+      "step": 3336
+    },
+    {
+      "epoch": 2.0030012004801923,
+      "grad_norm": 0.0028058600146323442,
+      "learning_rate": 0.00018111215232273983,
+      "loss": 23.0,
+      "step": 3337
+    },
+    {
+      "epoch": 2.0036014405762304,
+      "grad_norm": 0.0015503312461078167,
+      "learning_rate": 0.00018110110188356395,
+      "loss": 23.0,
+      "step": 3338
+    },
+    {
+      "epoch": 2.004201680672269,
+      "grad_norm": 0.0017939833924174309,
+      "learning_rate": 0.00018109004855009837,
+      "loss": 23.0,
+      "step": 3339
+    },
+    {
+      "epoch": 2.0048019207683074,
+      "grad_norm": 0.0033577047288417816,
+      "learning_rate": 0.00018107899232273756,
+      "loss": 23.0,
+      "step": 3340
+    },
+    {
+      "epoch": 2.0054021608643455,
+      "grad_norm": 0.0008726062951609492,
+      "learning_rate": 0.00018106793320187605,
+      "loss": 23.0,
+      "step": 3341
+    },
+    {
+      "epoch": 2.006002400960384,
+      "grad_norm": 0.0011513312347233295,
+      "learning_rate": 0.00018105687118790854,
+      "loss": 23.0,
+      "step": 3342
+    },
+    {
+      "epoch": 2.0066026410564226,
+      "grad_norm": 0.001895427005365491,
+      "learning_rate": 0.00018104580628122984,
+      "loss": 23.0,
+      "step": 3343
+    },
+    {
+      "epoch": 2.007202881152461,
+      "grad_norm": 0.0009158776374533772,
+      "learning_rate": 0.0001810347384822347,
+      "loss": 23.0,
+      "step": 3344
+    },
+    {
+      "epoch": 2.0078031212484992,
+      "grad_norm": 0.0008634175173938274,
+      "learning_rate": 0.00018102366779131827,
+      "loss": 23.0,
+      "step": 3345
+    },
+    {
+      "epoch": 2.008403361344538,
+      "grad_norm": 0.0019585357513278723,
+      "learning_rate": 0.00018101259420887553,
+      "loss": 23.0,
+      "step": 3346
+    },
+    {
+      "epoch": 2.0090036014405763,
+      "grad_norm": 0.005161345470696688,
+      "learning_rate": 0.00018100151773530168,
+      "loss": 23.0,
+      "step": 3347
+    },
+    {
+      "epoch": 2.009603841536615,
+      "grad_norm": 0.003942587878555059,
+      "learning_rate": 0.00018099043837099203,
+      "loss": 23.0,
+      "step": 3348
+    },
+    {
+      "epoch": 2.010204081632653,
+      "grad_norm": 0.0013700701529160142,
+      "learning_rate": 0.00018097935611634198,
+      "loss": 23.0,
+      "step": 3349
+    },
+    {
+      "epoch": 2.0108043217286915,
+      "grad_norm": 0.0007138294167816639,
+      "learning_rate": 0.000180968270971747,
+      "loss": 23.0,
+      "step": 3350
+    },
+    {
+      "epoch": 2.01140456182473,
+      "grad_norm": 0.005530905444175005,
+      "learning_rate": 0.0001809571829376027,
+      "loss": 23.0,
+      "step": 3351
+    },
+    {
+      "epoch": 2.012004801920768,
+      "grad_norm": 0.0032923435792326927,
+      "learning_rate": 0.0001809460920143048,
+      "loss": 23.0,
+      "step": 3352
+    },
+    {
+      "epoch": 2.0126050420168067,
+      "grad_norm": 0.0005104127340018749,
+      "learning_rate": 0.0001809349982022491,
+      "loss": 23.0,
+      "step": 3353
+    },
+    {
+      "epoch": 2.013205282112845,
+      "grad_norm": 0.001574002206325531,
+      "learning_rate": 0.0001809239015018315,
+      "loss": 23.0,
+      "step": 3354
+    },
+    {
+      "epoch": 2.0138055222088838,
+      "grad_norm": 0.002212009159848094,
+      "learning_rate": 0.000180912801913448,
+      "loss": 23.0,
+      "step": 3355
+    },
+    {
+      "epoch": 2.014405762304922,
+      "grad_norm": 0.0011559105478227139,
+      "learning_rate": 0.00018090169943749476,
+      "loss": 23.0,
+      "step": 3356
+    },
+    {
+      "epoch": 2.0150060024009604,
+      "grad_norm": 0.0033464916050434113,
+      "learning_rate": 0.00018089059407436794,
+      "loss": 23.0,
+      "step": 3357
+    },
+    {
+      "epoch": 2.015606242496999,
+      "grad_norm": 0.0011934455251321197,
+      "learning_rate": 0.0001808794858244639,
+      "loss": 23.0,
+      "step": 3358
+    },
+    {
+      "epoch": 2.016206482593037,
+      "grad_norm": 0.0006701741367578506,
+      "learning_rate": 0.00018086837468817907,
+      "loss": 23.0,
+      "step": 3359
+    },
+    {
+      "epoch": 2.0168067226890756,
+      "grad_norm": 0.0014794671442359686,
+      "learning_rate": 0.00018085726066590998,
+      "loss": 23.0,
+      "step": 3360
+    },
+    {
+      "epoch": 2.017406962785114,
+      "grad_norm": 0.00047398850438185036,
+      "learning_rate": 0.00018084614375805323,
+      "loss": 23.0,
+      "step": 3361
+    },
+    {
+      "epoch": 2.0180072028811527,
+      "grad_norm": 0.0023361060302704573,
+      "learning_rate": 0.00018083502396500556,
+      "loss": 23.0,
+      "step": 3362
+    },
+    {
+      "epoch": 2.0186074429771907,
+      "grad_norm": 0.0012610971461981535,
+      "learning_rate": 0.0001808239012871638,
+      "loss": 23.0,
+      "step": 3363
+    },
+    {
+      "epoch": 2.0192076830732293,
+      "grad_norm": 0.004134600982069969,
+      "learning_rate": 0.00018081277572492494,
+      "loss": 23.0,
+      "step": 3364
+    },
+    {
+      "epoch": 2.019807923169268,
+      "grad_norm": 0.000841438421048224,
+      "learning_rate": 0.000180801647278686,
+      "loss": 23.0,
+      "step": 3365
+    },
+    {
+      "epoch": 2.020408163265306,
+      "grad_norm": 0.002190615748986602,
+      "learning_rate": 0.0001807905159488441,
+      "loss": 23.0,
+      "step": 3366
+    },
+    {
+      "epoch": 2.0210084033613445,
+      "grad_norm": 0.0028436193242669106,
+      "learning_rate": 0.00018077938173579648,
+      "loss": 23.0,
+      "step": 3367
+    },
+    {
+      "epoch": 2.021608643457383,
+      "grad_norm": 0.003500116989016533,
+      "learning_rate": 0.00018076824463994056,
+      "loss": 23.0,
+      "step": 3368
+    },
+    {
+      "epoch": 2.0222088835534215,
+      "grad_norm": 0.0026991087943315506,
+      "learning_rate": 0.0001807571046616737,
+      "loss": 23.0,
+      "step": 3369
+    },
+    {
+      "epoch": 2.0228091236494596,
+      "grad_norm": 0.0010631488403305411,
+      "learning_rate": 0.00018074596180139353,
+      "loss": 23.0,
+      "step": 3370
+    },
+    {
+      "epoch": 2.023409363745498,
+      "grad_norm": 0.0020518836099654436,
+      "learning_rate": 0.00018073481605949772,
+      "loss": 23.0,
+      "step": 3371
+    },
+    {
+      "epoch": 2.0240096038415367,
+      "grad_norm": 0.006052622105926275,
+      "learning_rate": 0.00018072366743638395,
+      "loss": 23.0,
+      "step": 3372
+    },
+    {
+      "epoch": 2.024609843937575,
+      "grad_norm": 0.0021911882795393467,
+      "learning_rate": 0.00018071251593245016,
+      "loss": 23.0,
+      "step": 3373
+    },
+    {
+      "epoch": 2.0252100840336134,
+      "grad_norm": 0.0032586664892733097,
+      "learning_rate": 0.00018070136154809427,
+      "loss": 23.0,
+      "step": 3374
+    },
+    {
+      "epoch": 2.025810324129652,
+      "grad_norm": 0.000785050040576607,
+      "learning_rate": 0.0001806902042837144,
+      "loss": 23.0,
+      "step": 3375
+    },
+    {
+      "epoch": 2.0264105642256904,
+      "grad_norm": 0.0007825555512681603,
+      "learning_rate": 0.00018067904413970868,
+      "loss": 23.0,
+      "step": 3376
+    },
+    {
+      "epoch": 2.0270108043217285,
+      "grad_norm": 0.003356239991262555,
+      "learning_rate": 0.00018066788111647543,
+      "loss": 23.0,
+      "step": 3377
+    },
+    {
+      "epoch": 2.027611044417767,
+      "grad_norm": 0.0014766575768589973,
+      "learning_rate": 0.00018065671521441302,
+      "loss": 23.0,
+      "step": 3378
+    },
+    {
+      "epoch": 2.0282112845138056,
+      "grad_norm": 0.0009158303728327155,
+      "learning_rate": 0.00018064554643391987,
+      "loss": 23.0,
+      "step": 3379
+    },
+    {
+      "epoch": 2.028811524609844,
+      "grad_norm": 0.0009348996100015938,
+      "learning_rate": 0.00018063437477539466,
+      "loss": 23.0,
+      "step": 3380
+    },
+    {
+      "epoch": 2.0294117647058822,
+      "grad_norm": 0.0008702090126462281,
+      "learning_rate": 0.00018062320023923598,
+      "loss": 23.0,
+      "step": 3381
+    },
+    {
+      "epoch": 2.030012004801921,
+      "grad_norm": 0.002282114699482918,
+      "learning_rate": 0.0001806120228258427,
+      "loss": 23.0,
+      "step": 3382
+    },
+    {
+      "epoch": 2.0306122448979593,
+      "grad_norm": 0.005010796710848808,
+      "learning_rate": 0.00018060084253561368,
+      "loss": 23.0,
+      "step": 3383
+    },
+    {
+      "epoch": 2.0312124849939974,
+      "grad_norm": 0.0017042094841599464,
+      "learning_rate": 0.0001805896593689479,
+      "loss": 23.0,
+      "step": 3384
+    },
+    {
+      "epoch": 2.031812725090036,
+      "grad_norm": 0.0037521664053201675,
+      "learning_rate": 0.00018057847332624452,
+      "loss": 23.0,
+      "step": 3385
+    },
+    {
+      "epoch": 2.0324129651860745,
+      "grad_norm": 0.0015214249724522233,
+      "learning_rate": 0.00018056728440790266,
+      "loss": 23.0,
+      "step": 3386
+    },
+    {
+      "epoch": 2.033013205282113,
+      "grad_norm": 0.0004935787292197347,
+      "learning_rate": 0.00018055609261432169,
+      "loss": 23.0,
+      "step": 3387
+    },
+    {
+      "epoch": 2.033613445378151,
+      "grad_norm": 0.003463397966697812,
+      "learning_rate": 0.00018054489794590097,
+      "loss": 23.0,
+      "step": 3388
+    },
+    {
+      "epoch": 2.0342136854741897,
+      "grad_norm": 0.002584519563242793,
+      "learning_rate": 0.00018053370040304004,
+      "loss": 23.0,
+      "step": 3389
+    },
+    {
+      "epoch": 2.034813925570228,
+      "grad_norm": 0.0012157384771853685,
+      "learning_rate": 0.0001805224999861385,
+      "loss": 23.0,
+      "step": 3390
+    },
+    {
+      "epoch": 2.0354141656662663,
+      "grad_norm": 0.0029177102260291576,
+      "learning_rate": 0.00018051129669559605,
+      "loss": 23.0,
+      "step": 3391
+    },
+    {
+      "epoch": 2.036014405762305,
+      "grad_norm": 0.0015428938204422593,
+      "learning_rate": 0.00018050009053181257,
+      "loss": 23.0,
+      "step": 3392
+    },
+    {
+      "epoch": 2.0366146458583434,
+      "grad_norm": 0.0007340401643887162,
+      "learning_rate": 0.00018048888149518788,
+      "loss": 23.0,
+      "step": 3393
+    },
+    {
+      "epoch": 2.037214885954382,
+      "grad_norm": 0.0011957536917179823,
+      "learning_rate": 0.00018047766958612208,
+      "loss": 23.0,
+      "step": 3394
+    },
+    {
+      "epoch": 2.03781512605042,
+      "grad_norm": 0.0017125644953921437,
+      "learning_rate": 0.00018046645480501524,
+      "loss": 23.0,
+      "step": 3395
+    },
+    {
+      "epoch": 2.0384153661464586,
+      "grad_norm": 0.0023938282392919064,
+      "learning_rate": 0.0001804552371522676,
+      "loss": 23.0,
+      "step": 3396
+    },
+    {
+      "epoch": 2.039015606242497,
+      "grad_norm": 0.0033557454589754343,
+      "learning_rate": 0.00018044401662827955,
+      "loss": 23.0,
+      "step": 3397
+    },
+    {
+      "epoch": 2.039615846338535,
+      "grad_norm": 0.0025200413074344397,
+      "learning_rate": 0.00018043279323345143,
+      "loss": 23.0,
+      "step": 3398
+    },
+    {
+      "epoch": 2.0402160864345738,
+      "grad_norm": 0.0025761276483535767,
+      "learning_rate": 0.00018042156696818385,
+      "loss": 23.0,
+      "step": 3399
+    },
+    {
+      "epoch": 2.0408163265306123,
+      "grad_norm": 0.0009423838928341866,
+      "learning_rate": 0.00018041033783287738,
+      "loss": 23.0,
+      "step": 3400
+    },
+    {
+      "epoch": 2.041416566626651,
+      "grad_norm": 0.004442538134753704,
+      "learning_rate": 0.00018039910582793282,
+      "loss": 23.0,
+      "step": 3401
+    },
+    {
+      "epoch": 2.042016806722689,
+      "grad_norm": 0.002732986817136407,
+      "learning_rate": 0.000180387870953751,
+      "loss": 23.0,
+      "step": 3402
+    },
+    {
+      "epoch": 2.0426170468187275,
+      "grad_norm": 0.003854157403111458,
+      "learning_rate": 0.0001803766332107328,
+      "loss": 23.0,
+      "step": 3403
+    },
+    {
+      "epoch": 2.043217286914766,
+      "grad_norm": 0.00363258458673954,
+      "learning_rate": 0.00018036539259927932,
+      "loss": 23.0,
+      "step": 3404
+    },
+    {
+      "epoch": 2.0438175270108045,
+      "grad_norm": 0.004020926542580128,
+      "learning_rate": 0.00018035414911979177,
+      "loss": 23.0,
+      "step": 3405
+    },
+    {
+      "epoch": 2.0444177671068426,
+      "grad_norm": 0.0015982519835233688,
+      "learning_rate": 0.00018034290277267126,
+      "loss": 23.0,
+      "step": 3406
+    },
+    {
+      "epoch": 2.045018007202881,
+      "grad_norm": 0.0038941625971347094,
+      "learning_rate": 0.00018033165355831924,
+      "loss": 23.0,
+      "step": 3407
+    },
+    {
+      "epoch": 2.0456182472989197,
+      "grad_norm": 0.0026739637833088636,
+      "learning_rate": 0.00018032040147713716,
+      "loss": 23.0,
+      "step": 3408
+    },
+    {
+      "epoch": 2.046218487394958,
+      "grad_norm": 0.0016816748538985848,
+      "learning_rate": 0.00018030914652952652,
+      "loss": 23.0,
+      "step": 3409
+    },
+    {
+      "epoch": 2.0468187274909964,
+      "grad_norm": 0.001945959753356874,
+      "learning_rate": 0.00018029788871588905,
+      "loss": 23.0,
+      "step": 3410
+    },
+    {
+      "epoch": 2.047418967587035,
+      "grad_norm": 0.002473194617778063,
+      "learning_rate": 0.0001802866280366265,
+      "loss": 23.0,
+      "step": 3411
+    },
+    {
+      "epoch": 2.0480192076830734,
+      "grad_norm": 0.003011916531249881,
+      "learning_rate": 0.00018027536449214066,
+      "loss": 23.0,
+      "step": 3412
+    },
+    {
+      "epoch": 2.0486194477791115,
+      "grad_norm": 0.0023058606311678886,
+      "learning_rate": 0.0001802640980828336,
+      "loss": 23.0,
+      "step": 3413
+    },
+    {
+      "epoch": 2.04921968787515,
+      "grad_norm": 0.0021154494024813175,
+      "learning_rate": 0.00018025282880910733,
+      "loss": 23.0,
+      "step": 3414
+    },
+    {
+      "epoch": 2.0498199279711886,
+      "grad_norm": 0.000943707418628037,
+      "learning_rate": 0.00018024155667136405,
+      "loss": 23.0,
+      "step": 3415
+    },
+    {
+      "epoch": 2.0504201680672267,
+      "grad_norm": 0.0005733675789088011,
+      "learning_rate": 0.000180230281670006,
+      "loss": 23.0,
+      "step": 3416
+    },
+    {
+      "epoch": 2.0510204081632653,
+      "grad_norm": 0.0012774687493219972,
+      "learning_rate": 0.00018021900380543558,
+      "loss": 23.0,
+      "step": 3417
+    },
+    {
+      "epoch": 2.051620648259304,
+      "grad_norm": 0.0015681550139561296,
+      "learning_rate": 0.00018020772307805524,
+      "loss": 23.0,
+      "step": 3418
+    },
+    {
+      "epoch": 2.0522208883553423,
+      "grad_norm": 0.002922586165368557,
+      "learning_rate": 0.00018019643948826762,
+      "loss": 23.0,
+      "step": 3419
+    },
+    {
+      "epoch": 2.0528211284513804,
+      "grad_norm": 0.000941049656830728,
+      "learning_rate": 0.00018018515303647533,
+      "loss": 23.0,
+      "step": 3420
+    },
+    {
+      "epoch": 2.053421368547419,
+      "grad_norm": 0.0013102737721055746,
+      "learning_rate": 0.0001801738637230812,
+      "loss": 23.0,
+      "step": 3421
+    },
+    {
+      "epoch": 2.0540216086434575,
+      "grad_norm": 0.0010620380053296685,
+      "learning_rate": 0.00018016257154848813,
+      "loss": 23.0,
+      "step": 3422
+    },
+    {
+      "epoch": 2.0546218487394956,
+      "grad_norm": 0.001468815840780735,
+      "learning_rate": 0.00018015127651309907,
+      "loss": 23.0,
+      "step": 3423
+    },
+    {
+      "epoch": 2.055222088835534,
+      "grad_norm": 0.002112776506692171,
+      "learning_rate": 0.0001801399786173171,
+      "loss": 23.0,
+      "step": 3424
+    },
+    {
+      "epoch": 2.0558223289315727,
+      "grad_norm": 0.0043845935724675655,
+      "learning_rate": 0.00018012867786154545,
+      "loss": 23.0,
+      "step": 3425
+    },
+    {
+      "epoch": 2.0564225690276112,
+      "grad_norm": 0.0023064163979142904,
+      "learning_rate": 0.0001801173742461874,
+      "loss": 23.0,
+      "step": 3426
+    },
+    {
+      "epoch": 2.0570228091236493,
+      "grad_norm": 0.0015078387223184109,
+      "learning_rate": 0.0001801060677716464,
+      "loss": 23.0,
+      "step": 3427
+    },
+    {
+      "epoch": 2.057623049219688,
+      "grad_norm": 0.0014458864461630583,
+      "learning_rate": 0.00018009475843832587,
+      "loss": 23.0,
+      "step": 3428
+    },
+    {
+      "epoch": 2.0582232893157264,
+      "grad_norm": 0.0032603242434561253,
+      "learning_rate": 0.00018008344624662943,
+      "loss": 23.0,
+      "step": 3429
+    },
+    {
+      "epoch": 2.0588235294117645,
+      "grad_norm": 0.0012629261473193765,
+      "learning_rate": 0.00018007213119696077,
+      "loss": 23.0,
+      "step": 3430
+    },
+    {
+      "epoch": 2.059423769507803,
+      "grad_norm": 0.0016698999097570777,
+      "learning_rate": 0.00018006081328972374,
+      "loss": 23.0,
+      "step": 3431
+    },
+    {
+      "epoch": 2.0600240096038416,
+      "grad_norm": 0.001395948464050889,
+      "learning_rate": 0.00018004949252532223,
+      "loss": 23.0,
+      "step": 3432
+    },
+    {
+      "epoch": 2.06062424969988,
+      "grad_norm": 0.004593064077198505,
+      "learning_rate": 0.00018003816890416023,
+      "loss": 23.0,
+      "step": 3433
+    },
+    {
+      "epoch": 2.061224489795918,
+      "grad_norm": 0.0021219372283667326,
+      "learning_rate": 0.00018002684242664186,
+      "loss": 23.0,
+      "step": 3434
+    },
+    {
+      "epoch": 2.0618247298919568,
+      "grad_norm": 0.0006292082252912223,
+      "learning_rate": 0.00018001551309317136,
+      "loss": 23.0,
+      "step": 3435
+    },
+    {
+      "epoch": 2.0624249699879953,
+      "grad_norm": 0.0024436218664050102,
+      "learning_rate": 0.000180004180904153,
+      "loss": 23.0,
+      "step": 3436
+    },
+    {
+      "epoch": 2.0630252100840334,
+      "grad_norm": 0.0022101947106420994,
+      "learning_rate": 0.00017999284585999126,
+      "loss": 23.0,
+      "step": 3437
+    },
+    {
+      "epoch": 2.063625450180072,
+      "grad_norm": 0.0024126351345330477,
+      "learning_rate": 0.00017998150796109057,
+      "loss": 23.0,
+      "step": 3438
+    },
+    {
+      "epoch": 2.0642256902761105,
+      "grad_norm": 0.00167564966250211,
+      "learning_rate": 0.00017997016720785563,
+      "loss": 23.0,
+      "step": 3439
+    },
+    {
+      "epoch": 2.064825930372149,
+      "grad_norm": 0.001377441338263452,
+      "learning_rate": 0.0001799588236006911,
+      "loss": 23.0,
+      "step": 3440
+    },
+    {
+      "epoch": 2.065426170468187,
+      "grad_norm": 0.002041054656729102,
+      "learning_rate": 0.00017994747714000187,
+      "loss": 23.0,
+      "step": 3441
+    },
+    {
+      "epoch": 2.0660264105642256,
+      "grad_norm": 0.003891473403200507,
+      "learning_rate": 0.0001799361278261928,
+      "loss": 23.0,
+      "step": 3442
+    },
+    {
+      "epoch": 2.066626650660264,
+      "grad_norm": 0.0010193546768277884,
+      "learning_rate": 0.00017992477565966896,
+      "loss": 23.0,
+      "step": 3443
+    },
+    {
+      "epoch": 2.0672268907563027,
+      "grad_norm": 0.0011679233284667134,
+      "learning_rate": 0.00017991342064083547,
+      "loss": 23.0,
+      "step": 3444
+    },
+    {
+      "epoch": 2.067827130852341,
+      "grad_norm": 0.0009892381494864821,
+      "learning_rate": 0.00017990206277009756,
+      "loss": 23.0,
+      "step": 3445
+    },
+    {
+      "epoch": 2.0684273709483794,
+      "grad_norm": 0.0028641563840210438,
+      "learning_rate": 0.00017989070204786053,
+      "loss": 23.0,
+      "step": 3446
+    },
+    {
+      "epoch": 2.069027611044418,
+      "grad_norm": 0.002391669899225235,
+      "learning_rate": 0.0001798793384745299,
+      "loss": 23.0,
+      "step": 3447
+    },
+    {
+      "epoch": 2.069627851140456,
+      "grad_norm": 0.0008025117567740381,
+      "learning_rate": 0.0001798679720505111,
+      "loss": 23.0,
+      "step": 3448
+    },
+    {
+      "epoch": 2.0702280912364945,
+      "grad_norm": 0.0011154050007462502,
+      "learning_rate": 0.00017985660277620985,
+      "loss": 23.0,
+      "step": 3449
+    },
+    {
+      "epoch": 2.070828331332533,
+      "grad_norm": 0.001934365602210164,
+      "learning_rate": 0.00017984523065203188,
+      "loss": 23.0,
+      "step": 3450
+    },
+    {
+      "epoch": 2.0714285714285716,
+      "grad_norm": 0.0007151182508096099,
+      "learning_rate": 0.000179833855678383,
+      "loss": 23.0,
+      "step": 3451
+    },
+    {
+      "epoch": 2.0720288115246097,
+      "grad_norm": 0.001716459752060473,
+      "learning_rate": 0.00017982247785566918,
+      "loss": 23.0,
+      "step": 3452
+    },
+    {
+      "epoch": 2.0726290516206483,
+      "grad_norm": 0.0010225954465568066,
+      "learning_rate": 0.00017981109718429643,
+      "loss": 23.0,
+      "step": 3453
+    },
+    {
+      "epoch": 2.073229291716687,
+      "grad_norm": 0.0027839874383062124,
+      "learning_rate": 0.00017979971366467094,
+      "loss": 23.0,
+      "step": 3454
+    },
+    {
+      "epoch": 2.073829531812725,
+      "grad_norm": 0.0011828261194750667,
+      "learning_rate": 0.00017978832729719893,
+      "loss": 23.0,
+      "step": 3455
+    },
+    {
+      "epoch": 2.0744297719087634,
+      "grad_norm": 0.00159098906442523,
+      "learning_rate": 0.00017977693808228677,
+      "loss": 23.0,
+      "step": 3456
+    },
+    {
+      "epoch": 2.075030012004802,
+      "grad_norm": 0.0014060731045901775,
+      "learning_rate": 0.0001797655460203409,
+      "loss": 23.0,
+      "step": 3457
+    },
+    {
+      "epoch": 2.0756302521008405,
+      "grad_norm": 0.001484134467318654,
+      "learning_rate": 0.00017975415111176783,
+      "loss": 23.0,
+      "step": 3458
+    },
+    {
+      "epoch": 2.0762304921968786,
+      "grad_norm": 0.0017137357499450445,
+      "learning_rate": 0.0001797427533569743,
+      "loss": 23.0,
+      "step": 3459
+    },
+    {
+      "epoch": 2.076830732292917,
+      "grad_norm": 0.0030945942271500826,
+      "learning_rate": 0.000179731352756367,
+      "loss": 23.0,
+      "step": 3460
+    },
+    {
+      "epoch": 2.0774309723889557,
+      "grad_norm": 0.0011171615915372968,
+      "learning_rate": 0.00017971994931035283,
+      "loss": 23.0,
+      "step": 3461
+    },
+    {
+      "epoch": 2.078031212484994,
+      "grad_norm": 0.001267662737518549,
+      "learning_rate": 0.00017970854301933874,
+      "loss": 23.0,
+      "step": 3462
+    },
+    {
+      "epoch": 2.0786314525810323,
+      "grad_norm": 0.0021587491501122713,
+      "learning_rate": 0.0001796971338837318,
+      "loss": 23.0,
+      "step": 3463
+    },
+    {
+      "epoch": 2.079231692677071,
+      "grad_norm": 0.002123309997841716,
+      "learning_rate": 0.00017968572190393912,
+      "loss": 23.0,
+      "step": 3464
+    },
+    {
+      "epoch": 2.0798319327731094,
+      "grad_norm": 0.0012843715958297253,
+      "learning_rate": 0.000179674307080368,
+      "loss": 23.0,
+      "step": 3465
+    },
+    {
+      "epoch": 2.0804321728691475,
+      "grad_norm": 0.0021665177773684263,
+      "learning_rate": 0.00017966288941342583,
+      "loss": 23.0,
+      "step": 3466
+    },
+    {
+      "epoch": 2.081032412965186,
+      "grad_norm": 0.0019804276525974274,
+      "learning_rate": 0.00017965146890352006,
+      "loss": 23.0,
+      "step": 3467
+    },
+    {
+      "epoch": 2.0816326530612246,
+      "grad_norm": 0.0025594120379537344,
+      "learning_rate": 0.00017964004555105824,
+      "loss": 23.0,
+      "step": 3468
+    },
+    {
+      "epoch": 2.082232893157263,
+      "grad_norm": 0.002886942122131586,
+      "learning_rate": 0.00017962861935644805,
+      "loss": 23.0,
+      "step": 3469
+    },
+    {
+      "epoch": 2.082833133253301,
+      "grad_norm": 0.001857664086855948,
+      "learning_rate": 0.0001796171903200973,
+      "loss": 23.0,
+      "step": 3470
+    },
+    {
+      "epoch": 2.0834333733493398,
+      "grad_norm": 0.002572208410128951,
+      "learning_rate": 0.0001796057584424138,
+      "loss": 23.0,
+      "step": 3471
+    },
+    {
+      "epoch": 2.0840336134453783,
+      "grad_norm": 0.002026068978011608,
+      "learning_rate": 0.00017959432372380554,
+      "loss": 23.0,
+      "step": 3472
+    },
+    {
+      "epoch": 2.0846338535414164,
+      "grad_norm": 0.0016905996017158031,
+      "learning_rate": 0.0001795828861646806,
+      "loss": 23.0,
+      "step": 3473
+    },
+    {
+      "epoch": 2.085234093637455,
+      "grad_norm": 0.0011786259710788727,
+      "learning_rate": 0.00017957144576544718,
+      "loss": 23.0,
+      "step": 3474
+    },
+    {
+      "epoch": 2.0858343337334935,
+      "grad_norm": 0.0007971758022904396,
+      "learning_rate": 0.00017956000252651353,
+      "loss": 23.0,
+      "step": 3475
+    },
+    {
+      "epoch": 2.086434573829532,
+      "grad_norm": 0.0009448099881410599,
+      "learning_rate": 0.00017954855644828808,
+      "loss": 23.0,
+      "step": 3476
+    },
+    {
+      "epoch": 2.08703481392557,
+      "grad_norm": 0.002794134197756648,
+      "learning_rate": 0.00017953710753117923,
+      "loss": 23.0,
+      "step": 3477
+    },
+    {
+      "epoch": 2.0876350540216086,
+      "grad_norm": 0.0017918311059474945,
+      "learning_rate": 0.00017952565577559562,
+      "loss": 23.0,
+      "step": 3478
+    },
+    {
+      "epoch": 2.088235294117647,
+      "grad_norm": 0.0032223749440163374,
+      "learning_rate": 0.00017951420118194594,
+      "loss": 23.0,
+      "step": 3479
+    },
+    {
+      "epoch": 2.0888355342136853,
+      "grad_norm": 0.0009358105598948896,
+      "learning_rate": 0.00017950274375063893,
+      "loss": 23.0,
+      "step": 3480
+    },
+    {
+      "epoch": 2.089435774309724,
+      "grad_norm": 0.004243810195475817,
+      "learning_rate": 0.00017949128348208353,
+      "loss": 23.0,
+      "step": 3481
+    },
+    {
+      "epoch": 2.0900360144057624,
+      "grad_norm": 0.002074790420010686,
+      "learning_rate": 0.00017947982037668867,
+      "loss": 23.0,
+      "step": 3482
+    },
+    {
+      "epoch": 2.090636254501801,
+      "grad_norm": 0.0004455752205103636,
+      "learning_rate": 0.00017946835443486347,
+      "loss": 23.0,
+      "step": 3483
+    },
+    {
+      "epoch": 2.091236494597839,
+      "grad_norm": 0.005418194457888603,
+      "learning_rate": 0.00017945688565701714,
+      "loss": 23.0,
+      "step": 3484
+    },
+    {
+      "epoch": 2.0918367346938775,
+      "grad_norm": 0.0012525448109954596,
+      "learning_rate": 0.0001794454140435589,
+      "loss": 23.0,
+      "step": 3485
+    },
+    {
+      "epoch": 2.092436974789916,
+      "grad_norm": 0.0017173114465549588,
+      "learning_rate": 0.00017943393959489825,
+      "loss": 23.0,
+      "step": 3486
+    },
+    {
+      "epoch": 2.093037214885954,
+      "grad_norm": 0.0008452086476609111,
+      "learning_rate": 0.00017942246231144457,
+      "loss": 23.0,
+      "step": 3487
+    },
+    {
+      "epoch": 2.0936374549819927,
+      "grad_norm": 0.0011205211048945785,
+      "learning_rate": 0.00017941098219360755,
+      "loss": 23.0,
+      "step": 3488
+    },
+    {
+      "epoch": 2.0942376950780313,
+      "grad_norm": 0.0008696285076439381,
+      "learning_rate": 0.00017939949924179684,
+      "loss": 23.0,
+      "step": 3489
+    },
+    {
+      "epoch": 2.09483793517407,
+      "grad_norm": 0.0013399510644376278,
+      "learning_rate": 0.00017938801345642223,
+      "loss": 23.0,
+      "step": 3490
+    },
+    {
+      "epoch": 2.095438175270108,
+      "grad_norm": 0.0006139983888715506,
+      "learning_rate": 0.00017937652483789363,
+      "loss": 23.0,
+      "step": 3491
+    },
+    {
+      "epoch": 2.0960384153661464,
+      "grad_norm": 0.0014349283883348107,
+      "learning_rate": 0.00017936503338662104,
+      "loss": 23.0,
+      "step": 3492
+    },
+    {
+      "epoch": 2.096638655462185,
+      "grad_norm": 0.0010579722002148628,
+      "learning_rate": 0.00017935353910301455,
+      "loss": 23.0,
+      "step": 3493
+    },
+    {
+      "epoch": 2.097238895558223,
+      "grad_norm": 0.0010757379932329059,
+      "learning_rate": 0.0001793420419874844,
+      "loss": 23.0,
+      "step": 3494
+    },
+    {
+      "epoch": 2.0978391356542616,
+      "grad_norm": 0.0016692588105797768,
+      "learning_rate": 0.00017933054204044083,
+      "loss": 23.0,
+      "step": 3495
+    },
+    {
+      "epoch": 2.0984393757503,
+      "grad_norm": 0.002716292394325137,
+      "learning_rate": 0.00017931903926229428,
+      "loss": 23.0,
+      "step": 3496
+    },
+    {
+      "epoch": 2.0990396158463387,
+      "grad_norm": 0.0011049581225961447,
+      "learning_rate": 0.00017930753365345527,
+      "loss": 23.0,
+      "step": 3497
+    },
+    {
+      "epoch": 2.099639855942377,
+      "grad_norm": 0.001968407304957509,
+      "learning_rate": 0.0001792960252143344,
+      "loss": 23.0,
+      "step": 3498
+    },
+    {
+      "epoch": 2.1002400960384153,
+      "grad_norm": 0.0006715644267387688,
+      "learning_rate": 0.00017928451394534233,
+      "loss": 23.0,
+      "step": 3499
+    },
+    {
+      "epoch": 2.100840336134454,
+      "grad_norm": 0.0011009502923116088,
+      "learning_rate": 0.00017927299984688992,
+      "loss": 23.0,
+      "step": 3500
+    },
+    {
+      "epoch": 2.1014405762304924,
+      "grad_norm": 0.002423146739602089,
+      "learning_rate": 0.00017926148291938808,
+      "loss": 23.0,
+      "step": 3501
+    },
+    {
+      "epoch": 2.1020408163265305,
+      "grad_norm": 0.004653971176594496,
+      "learning_rate": 0.00017924996316324778,
+      "loss": 23.0,
+      "step": 3502
+    },
+    {
+      "epoch": 2.102641056422569,
+      "grad_norm": 0.002340496750548482,
+      "learning_rate": 0.00017923844057888012,
+      "loss": 23.0,
+      "step": 3503
+    },
+    {
+      "epoch": 2.1032412965186076,
+      "grad_norm": 0.0018698208732530475,
+      "learning_rate": 0.00017922691516669635,
+      "loss": 23.0,
+      "step": 3504
+    },
+    {
+      "epoch": 2.1038415366146457,
+      "grad_norm": 0.0017265928909182549,
+      "learning_rate": 0.00017921538692710782,
+      "loss": 23.0,
+      "step": 3505
+    },
+    {
+      "epoch": 2.104441776710684,
+      "grad_norm": 0.0035198975820094347,
+      "learning_rate": 0.00017920385586052588,
+      "loss": 23.0,
+      "step": 3506
+    },
+    {
+      "epoch": 2.1050420168067228,
+      "grad_norm": 0.005279972683638334,
+      "learning_rate": 0.00017919232196736203,
+      "loss": 23.0,
+      "step": 3507
+    },
+    {
+      "epoch": 2.1056422569027613,
+      "grad_norm": 0.0019218252273276448,
+      "learning_rate": 0.00017918078524802795,
+      "loss": 23.0,
+      "step": 3508
+    },
+    {
+      "epoch": 2.1062424969987994,
+      "grad_norm": 0.0037660326343029737,
+      "learning_rate": 0.00017916924570293528,
+      "loss": 23.0,
+      "step": 3509
+    },
+    {
+      "epoch": 2.106842737094838,
+      "grad_norm": 0.0016380378510802984,
+      "learning_rate": 0.00017915770333249594,
+      "loss": 23.0,
+      "step": 3510
+    },
+    {
+      "epoch": 2.1074429771908765,
+      "grad_norm": 0.0015107912477105856,
+      "learning_rate": 0.00017914615813712173,
+      "loss": 23.0,
+      "step": 3511
+    },
+    {
+      "epoch": 2.1080432172869146,
+      "grad_norm": 0.0029202804435044527,
+      "learning_rate": 0.00017913461011722474,
+      "loss": 23.0,
+      "step": 3512
+    },
+    {
+      "epoch": 2.108643457382953,
+      "grad_norm": 0.0013868259266018867,
+      "learning_rate": 0.0001791230592732171,
+      "loss": 23.0,
+      "step": 3513
+    },
+    {
+      "epoch": 2.1092436974789917,
+      "grad_norm": 0.0014497030060738325,
+      "learning_rate": 0.000179111505605511,
+      "loss": 23.0,
+      "step": 3514
+    },
+    {
+      "epoch": 2.10984393757503,
+      "grad_norm": 0.0015206874813884497,
+      "learning_rate": 0.00017909994911451872,
+      "loss": 23.0,
+      "step": 3515
+    },
+    {
+      "epoch": 2.1104441776710683,
+      "grad_norm": 0.003641763236373663,
+      "learning_rate": 0.00017908838980065274,
+      "loss": 23.0,
+      "step": 3516
+    },
+    {
+      "epoch": 2.111044417767107,
+      "grad_norm": 0.0021101627498865128,
+      "learning_rate": 0.0001790768276643256,
+      "loss": 23.0,
+      "step": 3517
+    },
+    {
+      "epoch": 2.1116446578631454,
+      "grad_norm": 0.0026833494193851948,
+      "learning_rate": 0.00017906526270594988,
+      "loss": 23.0,
+      "step": 3518
+    },
+    {
+      "epoch": 2.1122448979591835,
+      "grad_norm": 0.0010312756057828665,
+      "learning_rate": 0.0001790536949259383,
+      "loss": 23.0,
+      "step": 3519
+    },
+    {
+      "epoch": 2.112845138055222,
+      "grad_norm": 0.0033930472563952208,
+      "learning_rate": 0.00017904212432470371,
+      "loss": 23.0,
+      "step": 3520
+    },
+    {
+      "epoch": 2.1134453781512605,
+      "grad_norm": 0.0020478384103626013,
+      "learning_rate": 0.00017903055090265902,
+      "loss": 23.0,
+      "step": 3521
+    },
+    {
+      "epoch": 2.114045618247299,
+      "grad_norm": 0.0017036633798852563,
+      "learning_rate": 0.00017901897466021725,
+      "loss": 23.0,
+      "step": 3522
+    },
+    {
+      "epoch": 2.114645858343337,
+      "grad_norm": 0.0015527663053944707,
+      "learning_rate": 0.00017900739559779155,
+      "loss": 23.0,
+      "step": 3523
+    },
+    {
+      "epoch": 2.1152460984393757,
+      "grad_norm": 0.001124212285503745,
+      "learning_rate": 0.00017899581371579512,
+      "loss": 23.0,
+      "step": 3524
+    },
+    {
+      "epoch": 2.1158463385354143,
+      "grad_norm": 0.0014042783295735717,
+      "learning_rate": 0.0001789842290146413,
+      "loss": 23.0,
+      "step": 3525
+    },
+    {
+      "epoch": 2.116446578631453,
+      "grad_norm": 0.003444352652877569,
+      "learning_rate": 0.0001789726414947435,
+      "loss": 23.0,
+      "step": 3526
+    },
+    {
+      "epoch": 2.117046818727491,
+      "grad_norm": 0.003073784988373518,
+      "learning_rate": 0.00017896105115651533,
+      "loss": 23.0,
+      "step": 3527
+    },
+    {
+      "epoch": 2.1176470588235294,
+      "grad_norm": 0.0012429555645212531,
+      "learning_rate": 0.0001789494580003703,
+      "loss": 23.0,
+      "step": 3528
+    },
+    {
+      "epoch": 2.118247298919568,
+      "grad_norm": 0.004864856600761414,
+      "learning_rate": 0.0001789378620267222,
+      "loss": 23.0,
+      "step": 3529
+    },
+    {
+      "epoch": 2.118847539015606,
+      "grad_norm": 0.0022995832841843367,
+      "learning_rate": 0.00017892626323598486,
+      "loss": 23.0,
+      "step": 3530
+    },
+    {
+      "epoch": 2.1194477791116446,
+      "grad_norm": 0.00140533153899014,
+      "learning_rate": 0.00017891466162857223,
+      "loss": 23.0,
+      "step": 3531
+    },
+    {
+      "epoch": 2.120048019207683,
+      "grad_norm": 0.0009306234424002469,
+      "learning_rate": 0.0001789030572048983,
+      "loss": 23.0,
+      "step": 3532
+    },
+    {
+      "epoch": 2.1206482593037217,
+      "grad_norm": 0.003682507202029228,
+      "learning_rate": 0.00017889144996537725,
+      "loss": 23.0,
+      "step": 3533
+    },
+    {
+      "epoch": 2.12124849939976,
+      "grad_norm": 0.0009380745468661189,
+      "learning_rate": 0.00017887983991042323,
+      "loss": 23.0,
+      "step": 3534
+    },
+    {
+      "epoch": 2.1218487394957983,
+      "grad_norm": 0.0013799188891425729,
+      "learning_rate": 0.00017886822704045064,
+      "loss": 23.0,
+      "step": 3535
+    },
+    {
+      "epoch": 2.122448979591837,
+      "grad_norm": 0.0009243764216080308,
+      "learning_rate": 0.00017885661135587393,
+      "loss": 23.0,
+      "step": 3536
+    },
+    {
+      "epoch": 2.123049219687875,
+      "grad_norm": 0.0008515262743458152,
+      "learning_rate": 0.00017884499285710762,
+      "loss": 23.0,
+      "step": 3537
+    },
+    {
+      "epoch": 2.1236494597839135,
+      "grad_norm": 0.0015884919557720423,
+      "learning_rate": 0.00017883337154456629,
+      "loss": 23.0,
+      "step": 3538
+    },
+    {
+      "epoch": 2.124249699879952,
+      "grad_norm": 0.0032588981557637453,
+      "learning_rate": 0.0001788217474186647,
+      "loss": 23.0,
+      "step": 3539
+    },
+    {
+      "epoch": 2.1248499399759906,
+      "grad_norm": 0.0024941442534327507,
+      "learning_rate": 0.00017881012047981772,
+      "loss": 23.0,
+      "step": 3540
+    },
+    {
+      "epoch": 2.1254501800720287,
+      "grad_norm": 0.0017060820246115327,
+      "learning_rate": 0.00017879849072844028,
+      "loss": 23.0,
+      "step": 3541
+    },
+    {
+      "epoch": 2.1260504201680672,
+      "grad_norm": 0.003131990320980549,
+      "learning_rate": 0.0001787868581649474,
+      "loss": 23.0,
+      "step": 3542
+    },
+    {
+      "epoch": 2.1266506602641058,
+      "grad_norm": 0.0024005863815546036,
+      "learning_rate": 0.00017877522278975417,
+      "loss": 23.0,
+      "step": 3543
+    },
+    {
+      "epoch": 2.127250900360144,
+      "grad_norm": 0.0006045355694368482,
+      "learning_rate": 0.0001787635846032759,
+      "loss": 23.0,
+      "step": 3544
+    },
+    {
+      "epoch": 2.1278511404561824,
+      "grad_norm": 0.0035146684385836124,
+      "learning_rate": 0.0001787519436059279,
+      "loss": 23.0,
+      "step": 3545
+    },
+    {
+      "epoch": 2.128451380552221,
+      "grad_norm": 0.0014111412456259131,
+      "learning_rate": 0.00017874029979812563,
+      "loss": 23.0,
+      "step": 3546
+    },
+    {
+      "epoch": 2.1290516206482595,
+      "grad_norm": 0.0018674216698855162,
+      "learning_rate": 0.00017872865318028459,
+      "loss": 23.0,
+      "step": 3547
+    },
+    {
+      "epoch": 2.1296518607442976,
+      "grad_norm": 0.0012715421617031097,
+      "learning_rate": 0.00017871700375282043,
+      "loss": 23.0,
+      "step": 3548
+    },
+    {
+      "epoch": 2.130252100840336,
+      "grad_norm": 0.003209817223250866,
+      "learning_rate": 0.0001787053515161489,
+      "loss": 23.0,
+      "step": 3549
+    },
+    {
+      "epoch": 2.1308523409363747,
+      "grad_norm": 0.001961090601980686,
+      "learning_rate": 0.0001786936964706858,
+      "loss": 23.0,
+      "step": 3550
+    },
+    {
+      "epoch": 2.1314525810324128,
+      "grad_norm": 0.0013860109029337764,
+      "learning_rate": 0.0001786820386168471,
+      "loss": 23.0,
+      "step": 3551
+    },
+    {
+      "epoch": 2.1320528211284513,
+      "grad_norm": 0.00047536773490719497,
+      "learning_rate": 0.00017867037795504887,
+      "loss": 23.0,
+      "step": 3552
+    },
+    {
+      "epoch": 2.13265306122449,
+      "grad_norm": 0.0019673663191497326,
+      "learning_rate": 0.00017865871448570717,
+      "loss": 23.0,
+      "step": 3553
+    },
+    {
+      "epoch": 2.1332533013205284,
+      "grad_norm": 0.0021188133396208286,
+      "learning_rate": 0.00017864704820923832,
+      "loss": 23.0,
+      "step": 3554
+    },
+    {
+      "epoch": 2.1338535414165665,
+      "grad_norm": 0.0027213292196393013,
+      "learning_rate": 0.0001786353791260586,
+      "loss": 23.0,
+      "step": 3555
+    },
+    {
+      "epoch": 2.134453781512605,
+      "grad_norm": 0.0027741962112486362,
+      "learning_rate": 0.0001786237072365845,
+      "loss": 23.0,
+      "step": 3556
+    },
+    {
+      "epoch": 2.1350540216086435,
+      "grad_norm": 0.002745026256889105,
+      "learning_rate": 0.00017861203254123252,
+      "loss": 23.0,
+      "step": 3557
+    },
+    {
+      "epoch": 2.1356542617046816,
+      "grad_norm": 0.0020450023002922535,
+      "learning_rate": 0.00017860035504041932,
+      "loss": 23.0,
+      "step": 3558
+    },
+    {
+      "epoch": 2.13625450180072,
+      "grad_norm": 0.0010032516438513994,
+      "learning_rate": 0.00017858867473456163,
+      "loss": 23.0,
+      "step": 3559
+    },
+    {
+      "epoch": 2.1368547418967587,
+      "grad_norm": 0.0014476818032562733,
+      "learning_rate": 0.00017857699162407626,
+      "loss": 23.0,
+      "step": 3560
+    },
+    {
+      "epoch": 2.1374549819927973,
+      "grad_norm": 0.001177955069579184,
+      "learning_rate": 0.00017856530570938022,
+      "loss": 23.0,
+      "step": 3561
+    },
+    {
+      "epoch": 2.1380552220888354,
+      "grad_norm": 0.002862164517864585,
+      "learning_rate": 0.0001785536169908905,
+      "loss": 23.0,
+      "step": 3562
+    },
+    {
+      "epoch": 2.138655462184874,
+      "grad_norm": 0.0010038964683189988,
+      "learning_rate": 0.00017854192546902427,
+      "loss": 23.0,
+      "step": 3563
+    },
+    {
+      "epoch": 2.1392557022809124,
+      "grad_norm": 0.0017746913945302367,
+      "learning_rate": 0.00017853023114419877,
+      "loss": 23.0,
+      "step": 3564
+    },
+    {
+      "epoch": 2.139855942376951,
+      "grad_norm": 0.0028854040428996086,
+      "learning_rate": 0.00017851853401683126,
+      "loss": 23.0,
+      "step": 3565
+    },
+    {
+      "epoch": 2.140456182472989,
+      "grad_norm": 0.001586511847563088,
+      "learning_rate": 0.00017850683408733928,
+      "loss": 23.0,
+      "step": 3566
+    },
+    {
+      "epoch": 2.1410564225690276,
+      "grad_norm": 0.003254214534536004,
+      "learning_rate": 0.00017849513135614035,
+      "loss": 23.0,
+      "step": 3567
+    },
+    {
+      "epoch": 2.141656662665066,
+      "grad_norm": 0.0012385440059006214,
+      "learning_rate": 0.00017848342582365206,
+      "loss": 23.0,
+      "step": 3568
+    },
+    {
+      "epoch": 2.1422569027611043,
+      "grad_norm": 0.0018727433634921908,
+      "learning_rate": 0.0001784717174902922,
+      "loss": 23.0,
+      "step": 3569
+    },
+    {
+      "epoch": 2.142857142857143,
+      "grad_norm": 0.004589486867189407,
+      "learning_rate": 0.00017846000635647863,
+      "loss": 23.0,
+      "step": 3570
+    },
+    {
+      "epoch": 2.1434573829531813,
+      "grad_norm": 0.0021497050765901804,
+      "learning_rate": 0.0001784482924226292,
+      "loss": 23.0,
+      "step": 3571
+    },
+    {
+      "epoch": 2.14405762304922,
+      "grad_norm": 0.0012627036776393652,
+      "learning_rate": 0.00017843657568916206,
+      "loss": 23.0,
+      "step": 3572
+    },
+    {
+      "epoch": 2.144657863145258,
+      "grad_norm": 0.001609223079867661,
+      "learning_rate": 0.00017842485615649528,
+      "loss": 23.0,
+      "step": 3573
+    },
+    {
+      "epoch": 2.1452581032412965,
+      "grad_norm": 0.001351055921986699,
+      "learning_rate": 0.00017841313382504711,
+      "loss": 23.0,
+      "step": 3574
+    },
+    {
+      "epoch": 2.145858343337335,
+      "grad_norm": 0.004179363604635,
+      "learning_rate": 0.00017840140869523587,
+      "loss": 23.0,
+      "step": 3575
+    },
+    {
+      "epoch": 2.146458583433373,
+      "grad_norm": 0.0016952658770605922,
+      "learning_rate": 0.00017838968076748008,
+      "loss": 23.0,
+      "step": 3576
+    },
+    {
+      "epoch": 2.1470588235294117,
+      "grad_norm": 0.0015167961828410625,
+      "learning_rate": 0.00017837795004219817,
+      "loss": 23.0,
+      "step": 3577
+    },
+    {
+      "epoch": 2.1476590636254502,
+      "grad_norm": 0.003133446676656604,
+      "learning_rate": 0.00017836621651980887,
+      "loss": 23.0,
+      "step": 3578
+    },
+    {
+      "epoch": 2.1482593037214888,
+      "grad_norm": 0.00226042908616364,
+      "learning_rate": 0.00017835448020073088,
+      "loss": 23.0,
+      "step": 3579
+    },
+    {
+      "epoch": 2.148859543817527,
+      "grad_norm": 0.0013810255331918597,
+      "learning_rate": 0.00017834274108538304,
+      "loss": 23.0,
+      "step": 3580
+    },
+    {
+      "epoch": 2.1494597839135654,
+      "grad_norm": 0.001641986076720059,
+      "learning_rate": 0.0001783309991741843,
+      "loss": 23.0,
+      "step": 3581
+    },
+    {
+      "epoch": 2.150060024009604,
+      "grad_norm": 0.002287675393745303,
+      "learning_rate": 0.0001783192544675537,
+      "loss": 23.0,
+      "step": 3582
+    },
+    {
+      "epoch": 2.1506602641056425,
+      "grad_norm": 0.0009976879227906466,
+      "learning_rate": 0.00017830750696591033,
+      "loss": 23.0,
+      "step": 3583
+    },
+    {
+      "epoch": 2.1512605042016806,
+      "grad_norm": 0.0018353014020249248,
+      "learning_rate": 0.0001782957566696735,
+      "loss": 23.0,
+      "step": 3584
+    },
+    {
+      "epoch": 2.151860744297719,
+      "grad_norm": 0.0019918165635317564,
+      "learning_rate": 0.00017828400357926253,
+      "loss": 23.0,
+      "step": 3585
+    },
+    {
+      "epoch": 2.1524609843937577,
+      "grad_norm": 0.0026910703163594007,
+      "learning_rate": 0.0001782722476950968,
+      "loss": 23.0,
+      "step": 3586
+    },
+    {
+      "epoch": 2.1530612244897958,
+      "grad_norm": 0.002473343163728714,
+      "learning_rate": 0.00017826048901759588,
+      "loss": 23.0,
+      "step": 3587
+    },
+    {
+      "epoch": 2.1536614645858343,
+      "grad_norm": 0.002732550958171487,
+      "learning_rate": 0.00017824872754717945,
+      "loss": 23.0,
+      "step": 3588
+    },
+    {
+      "epoch": 2.154261704681873,
+      "grad_norm": 0.0010042079957202077,
+      "learning_rate": 0.00017823696328426723,
+      "loss": 23.0,
+      "step": 3589
+    },
+    {
+      "epoch": 2.1548619447779114,
+      "grad_norm": 0.0027955600526183844,
+      "learning_rate": 0.000178225196229279,
+      "loss": 23.0,
+      "step": 3590
+    },
+    {
+      "epoch": 2.1554621848739495,
+      "grad_norm": 0.004130184184759855,
+      "learning_rate": 0.00017821342638263478,
+      "loss": 23.0,
+      "step": 3591
+    },
+    {
+      "epoch": 2.156062424969988,
+      "grad_norm": 0.003868487663567066,
+      "learning_rate": 0.00017820165374475452,
+      "loss": 23.0,
+      "step": 3592
+    },
+    {
+      "epoch": 2.1566626650660266,
+      "grad_norm": 0.0011650120140984654,
+      "learning_rate": 0.00017818987831605843,
+      "loss": 23.0,
+      "step": 3593
+    },
+    {
+      "epoch": 2.1572629051620646,
+      "grad_norm": 0.002393965143710375,
+      "learning_rate": 0.00017817810009696666,
+      "loss": 23.0,
+      "step": 3594
+    },
+    {
+      "epoch": 2.157863145258103,
+      "grad_norm": 0.0035790205001831055,
+      "learning_rate": 0.00017816631908789966,
+      "loss": 23.0,
+      "step": 3595
+    },
+    {
+      "epoch": 2.1584633853541417,
+      "grad_norm": 0.003235345706343651,
+      "learning_rate": 0.00017815453528927778,
+      "loss": 23.0,
+      "step": 3596
+    },
+    {
+      "epoch": 2.1590636254501803,
+      "grad_norm": 0.0013717797119170427,
+      "learning_rate": 0.00017814274870152155,
+      "loss": 23.0,
+      "step": 3597
+    },
+    {
+      "epoch": 2.1596638655462184,
+      "grad_norm": 0.0011841889936476946,
+      "learning_rate": 0.0001781309593250516,
+      "loss": 23.0,
+      "step": 3598
+    },
+    {
+      "epoch": 2.160264105642257,
+      "grad_norm": 0.0035572301130741835,
+      "learning_rate": 0.00017811916716028872,
+      "loss": 23.0,
+      "step": 3599
+    },
+    {
+      "epoch": 2.1608643457382954,
+      "grad_norm": 0.0009041174198500812,
+      "learning_rate": 0.00017810737220765374,
+      "loss": 23.0,
+      "step": 3600
+    },
+    {
+      "epoch": 2.1614645858343335,
+      "grad_norm": 0.0011678171576932073,
+      "learning_rate": 0.00017809557446756756,
+      "loss": 23.0,
+      "step": 3601
+    },
+    {
+      "epoch": 2.162064825930372,
+      "grad_norm": 0.0014870171435177326,
+      "learning_rate": 0.0001780837739404512,
+      "loss": 23.0,
+      "step": 3602
+    },
+    {
+      "epoch": 2.1626650660264106,
+      "grad_norm": 0.0015179289039224386,
+      "learning_rate": 0.00017807197062672577,
+      "loss": 23.0,
+      "step": 3603
+    },
+    {
+      "epoch": 2.163265306122449,
+      "grad_norm": 0.0020413794554769993,
+      "learning_rate": 0.00017806016452681254,
+      "loss": 23.0,
+      "step": 3604
+    },
+    {
+      "epoch": 2.1638655462184873,
+      "grad_norm": 0.0032616599928587675,
+      "learning_rate": 0.00017804835564113288,
+      "loss": 23.0,
+      "step": 3605
+    },
+    {
+      "epoch": 2.164465786314526,
+      "grad_norm": 0.0011516850208863616,
+      "learning_rate": 0.00017803654397010815,
+      "loss": 23.0,
+      "step": 3606
+    },
+    {
+      "epoch": 2.1650660264105643,
+      "grad_norm": 0.003489278955385089,
+      "learning_rate": 0.0001780247295141599,
+      "loss": 23.0,
+      "step": 3607
+    },
+    {
+      "epoch": 2.1656662665066024,
+      "grad_norm": 0.003842744044959545,
+      "learning_rate": 0.00017801291227370974,
+      "loss": 23.0,
+      "step": 3608
+    },
+    {
+      "epoch": 2.166266506602641,
+      "grad_norm": 0.001340657938271761,
+      "learning_rate": 0.0001780010922491794,
+      "loss": 23.0,
+      "step": 3609
+    },
+    {
+      "epoch": 2.1668667466986795,
+      "grad_norm": 0.0008930175099521875,
+      "learning_rate": 0.00017798926944099077,
+      "loss": 23.0,
+      "step": 3610
+    },
+    {
+      "epoch": 2.167466986794718,
+      "grad_norm": 0.0036189949605613947,
+      "learning_rate": 0.0001779774438495657,
+      "loss": 23.0,
+      "step": 3611
+    },
+    {
+      "epoch": 2.168067226890756,
+      "grad_norm": 0.003450029995292425,
+      "learning_rate": 0.00017796561547532623,
+      "loss": 23.0,
+      "step": 3612
+    },
+    {
+      "epoch": 2.1686674669867947,
+      "grad_norm": 0.0013816222781315446,
+      "learning_rate": 0.00017795378431869453,
+      "loss": 23.0,
+      "step": 3613
+    },
+    {
+      "epoch": 2.1692677070828332,
+      "grad_norm": 0.003153825644403696,
+      "learning_rate": 0.00017794195038009275,
+      "loss": 23.0,
+      "step": 3614
+    },
+    {
+      "epoch": 2.1698679471788713,
+      "grad_norm": 0.001024477300234139,
+      "learning_rate": 0.00017793011365994325,
+      "loss": 23.0,
+      "step": 3615
+    },
+    {
+      "epoch": 2.17046818727491,
+      "grad_norm": 0.003891037544235587,
+      "learning_rate": 0.00017791827415866848,
+      "loss": 23.0,
+      "step": 3616
+    },
+    {
+      "epoch": 2.1710684273709484,
+      "grad_norm": 0.002924687694758177,
+      "learning_rate": 0.00017790643187669093,
+      "loss": 23.0,
+      "step": 3617
+    },
+    {
+      "epoch": 2.171668667466987,
+      "grad_norm": 0.009289652109146118,
+      "learning_rate": 0.00017789458681443323,
+      "loss": 23.0,
+      "step": 3618
+    },
+    {
+      "epoch": 2.172268907563025,
+      "grad_norm": 0.0021330954041332006,
+      "learning_rate": 0.00017788273897231807,
+      "loss": 23.0,
+      "step": 3619
+    },
+    {
+      "epoch": 2.1728691476590636,
+      "grad_norm": 0.0019930656999349594,
+      "learning_rate": 0.0001778708883507683,
+      "loss": 23.0,
+      "step": 3620
+    },
+    {
+      "epoch": 2.173469387755102,
+      "grad_norm": 0.00517197186127305,
+      "learning_rate": 0.00017785903495020686,
+      "loss": 23.0,
+      "step": 3621
+    },
+    {
+      "epoch": 2.1740696278511407,
+      "grad_norm": 0.0015679982025176287,
+      "learning_rate": 0.00017784717877105673,
+      "loss": 23.0,
+      "step": 3622
+    },
+    {
+      "epoch": 2.1746698679471788,
+      "grad_norm": 0.004141742363572121,
+      "learning_rate": 0.00017783531981374102,
+      "loss": 23.0,
+      "step": 3623
+    },
+    {
+      "epoch": 2.1752701080432173,
+      "grad_norm": 0.00065791723318398,
+      "learning_rate": 0.00017782345807868297,
+      "loss": 23.0,
+      "step": 3624
+    },
+    {
+      "epoch": 2.175870348139256,
+      "grad_norm": 0.004961501341313124,
+      "learning_rate": 0.0001778115935663059,
+      "loss": 23.0,
+      "step": 3625
+    },
+    {
+      "epoch": 2.176470588235294,
+      "grad_norm": 0.0023005243856459856,
+      "learning_rate": 0.0001777997262770332,
+      "loss": 23.0,
+      "step": 3626
+    },
+    {
+      "epoch": 2.1770708283313325,
+      "grad_norm": 0.00899723544716835,
+      "learning_rate": 0.00017778785621128836,
+      "loss": 23.0,
+      "step": 3627
+    },
+    {
+      "epoch": 2.177671068427371,
+      "grad_norm": 0.0029707015492022038,
+      "learning_rate": 0.0001777759833694951,
+      "loss": 23.0,
+      "step": 3628
+    },
+    {
+      "epoch": 2.1782713085234096,
+      "grad_norm": 0.0018353236373513937,
+      "learning_rate": 0.000177764107752077,
+      "loss": 23.0,
+      "step": 3629
+    },
+    {
+      "epoch": 2.1788715486194477,
+      "grad_norm": 0.0010108323767781258,
+      "learning_rate": 0.00017775222935945796,
+      "loss": 23.0,
+      "step": 3630
+    },
+    {
+      "epoch": 2.179471788715486,
+      "grad_norm": 0.0008860941743478179,
+      "learning_rate": 0.00017774034819206181,
+      "loss": 23.0,
+      "step": 3631
+    },
+    {
+      "epoch": 2.1800720288115247,
+      "grad_norm": 0.003606445388868451,
+      "learning_rate": 0.00017772846425031266,
+      "loss": 23.0,
+      "step": 3632
+    },
+    {
+      "epoch": 2.180672268907563,
+      "grad_norm": 0.0016308276681229472,
+      "learning_rate": 0.00017771657753463453,
+      "loss": 23.0,
+      "step": 3633
+    },
+    {
+      "epoch": 2.1812725090036014,
+      "grad_norm": 0.0007045452948659658,
+      "learning_rate": 0.00017770468804545169,
+      "loss": 23.0,
+      "step": 3634
+    },
+    {
+      "epoch": 2.18187274909964,
+      "grad_norm": 0.0013048743130639195,
+      "learning_rate": 0.0001776927957831884,
+      "loss": 23.0,
+      "step": 3635
+    },
+    {
+      "epoch": 2.1824729891956784,
+      "grad_norm": 0.0012544161872938275,
+      "learning_rate": 0.00017768090074826904,
+      "loss": 23.0,
+      "step": 3636
+    },
+    {
+      "epoch": 2.1830732292917165,
+      "grad_norm": 0.00237638084217906,
+      "learning_rate": 0.0001776690029411182,
+      "loss": 23.0,
+      "step": 3637
+    },
+    {
+      "epoch": 2.183673469387755,
+      "grad_norm": 0.0048915608786046505,
+      "learning_rate": 0.0001776571023621604,
+      "loss": 23.0,
+      "step": 3638
+    },
+    {
+      "epoch": 2.1842737094837936,
+      "grad_norm": 0.000844002643134445,
+      "learning_rate": 0.0001776451990118204,
+      "loss": 23.0,
+      "step": 3639
+    },
+    {
+      "epoch": 2.184873949579832,
+      "grad_norm": 0.00372367724776268,
+      "learning_rate": 0.00017763329289052296,
+      "loss": 23.0,
+      "step": 3640
+    },
+    {
+      "epoch": 2.1854741896758703,
+      "grad_norm": 0.0023307260125875473,
+      "learning_rate": 0.000177621383998693,
+      "loss": 23.0,
+      "step": 3641
+    },
+    {
+      "epoch": 2.186074429771909,
+      "grad_norm": 0.0026479533407837152,
+      "learning_rate": 0.0001776094723367555,
+      "loss": 23.0,
+      "step": 3642
+    },
+    {
+      "epoch": 2.1866746698679473,
+      "grad_norm": 0.00475787278264761,
+      "learning_rate": 0.00017759755790513557,
+      "loss": 23.0,
+      "step": 3643
+    },
+    {
+      "epoch": 2.1872749099639854,
+      "grad_norm": 0.0008646330097690225,
+      "learning_rate": 0.00017758564070425843,
+      "loss": 23.0,
+      "step": 3644
+    },
+    {
+      "epoch": 2.187875150060024,
+      "grad_norm": 0.000730241765268147,
+      "learning_rate": 0.00017757372073454931,
+      "loss": 23.0,
+      "step": 3645
+    },
+    {
+      "epoch": 2.1884753901560625,
+      "grad_norm": 0.0020450761076062918,
+      "learning_rate": 0.00017756179799643368,
+      "loss": 23.0,
+      "step": 3646
+    },
+    {
+      "epoch": 2.189075630252101,
+      "grad_norm": 0.001998509978875518,
+      "learning_rate": 0.00017754987249033694,
+      "loss": 23.0,
+      "step": 3647
+    },
+    {
+      "epoch": 2.189675870348139,
+      "grad_norm": 0.0017022277461364865,
+      "learning_rate": 0.00017753794421668478,
+      "loss": 23.0,
+      "step": 3648
+    },
+    {
+      "epoch": 2.1902761104441777,
+      "grad_norm": 0.0010228764731436968,
+      "learning_rate": 0.0001775260131759028,
+      "loss": 23.0,
+      "step": 3649
+    },
+    {
+      "epoch": 2.1908763505402162,
+      "grad_norm": 0.0021690605208277702,
+      "learning_rate": 0.00017751407936841688,
+      "loss": 23.0,
+      "step": 3650
+    },
+    {
+      "epoch": 2.1914765906362543,
+      "grad_norm": 0.0016400478780269623,
+      "learning_rate": 0.00017750214279465282,
+      "loss": 23.0,
+      "step": 3651
+    },
+    {
+      "epoch": 2.192076830732293,
+      "grad_norm": 0.002070733578875661,
+      "learning_rate": 0.00017749020345503666,
+      "loss": 23.0,
+      "step": 3652
+    },
+    {
+      "epoch": 2.1926770708283314,
+      "grad_norm": 0.0014095326187089086,
+      "learning_rate": 0.00017747826134999445,
+      "loss": 23.0,
+      "step": 3653
+    },
+    {
+      "epoch": 2.19327731092437,
+      "grad_norm": 0.0019122892990708351,
+      "learning_rate": 0.00017746631647995242,
+      "loss": 23.0,
+      "step": 3654
+    },
+    {
+      "epoch": 2.193877551020408,
+      "grad_norm": 0.0029271540697664022,
+      "learning_rate": 0.00017745436884533683,
+      "loss": 23.0,
+      "step": 3655
+    },
+    {
+      "epoch": 2.1944777911164466,
+      "grad_norm": 0.002827174263074994,
+      "learning_rate": 0.00017744241844657398,
+      "loss": 23.0,
+      "step": 3656
+    },
+    {
+      "epoch": 2.195078031212485,
+      "grad_norm": 0.0012498522410169244,
+      "learning_rate": 0.00017743046528409049,
+      "loss": 23.0,
+      "step": 3657
+    },
+    {
+      "epoch": 2.195678271308523,
+      "grad_norm": 0.003829399822279811,
+      "learning_rate": 0.00017741850935831287,
+      "loss": 23.0,
+      "step": 3658
+    },
+    {
+      "epoch": 2.1962785114045618,
+      "grad_norm": 0.0021660744678229094,
+      "learning_rate": 0.00017740655066966773,
+      "loss": 23.0,
+      "step": 3659
+    },
+    {
+      "epoch": 2.1968787515006003,
+      "grad_norm": 0.002248313743621111,
+      "learning_rate": 0.00017739458921858198,
+      "loss": 23.0,
+      "step": 3660
+    },
+    {
+      "epoch": 2.197478991596639,
+      "grad_norm": 0.0018800124526023865,
+      "learning_rate": 0.00017738262500548236,
+      "loss": 23.0,
+      "step": 3661
+    },
+    {
+      "epoch": 2.198079231692677,
+      "grad_norm": 0.0030018959660083055,
+      "learning_rate": 0.00017737065803079597,
+      "loss": 23.0,
+      "step": 3662
+    },
+    {
+      "epoch": 2.1986794717887155,
+      "grad_norm": 0.0009485651389695704,
+      "learning_rate": 0.00017735868829494978,
+      "loss": 23.0,
+      "step": 3663
+    },
+    {
+      "epoch": 2.199279711884754,
+      "grad_norm": 0.002609929535537958,
+      "learning_rate": 0.000177346715798371,
+      "loss": 23.0,
+      "step": 3664
+    },
+    {
+      "epoch": 2.199879951980792,
+      "grad_norm": 0.0013200175017118454,
+      "learning_rate": 0.0001773347405414869,
+      "loss": 23.0,
+      "step": 3665
+    },
+    {
+      "epoch": 2.2004801920768307,
+      "grad_norm": 0.0007020602351985872,
+      "learning_rate": 0.00017732276252472483,
+      "loss": 23.0,
+      "step": 3666
+    },
+    {
+      "epoch": 2.201080432172869,
+      "grad_norm": 0.0026193074882030487,
+      "learning_rate": 0.00017731078174851227,
+      "loss": 23.0,
+      "step": 3667
+    },
+    {
+      "epoch": 2.2016806722689077,
+      "grad_norm": 0.001026253798045218,
+      "learning_rate": 0.00017729879821327678,
+      "loss": 23.0,
+      "step": 3668
+    },
+    {
+      "epoch": 2.202280912364946,
+      "grad_norm": 0.0017351701389998198,
+      "learning_rate": 0.000177286811919446,
+      "loss": 23.0,
+      "step": 3669
+    },
+    {
+      "epoch": 2.2028811524609844,
+      "grad_norm": 0.0014646050985902548,
+      "learning_rate": 0.0001772748228674477,
+      "loss": 23.0,
+      "step": 3670
+    },
+    {
+      "epoch": 2.203481392557023,
+      "grad_norm": 0.0030125817283988,
+      "learning_rate": 0.0001772628310577098,
+      "loss": 23.0,
+      "step": 3671
+    },
+    {
+      "epoch": 2.204081632653061,
+      "grad_norm": 0.004487418103963137,
+      "learning_rate": 0.00017725083649066016,
+      "loss": 23.0,
+      "step": 3672
+    },
+    {
+      "epoch": 2.2046818727490995,
+      "grad_norm": 0.0021111650858074427,
+      "learning_rate": 0.0001772388391667269,
+      "loss": 23.0,
+      "step": 3673
+    },
+    {
+      "epoch": 2.205282112845138,
+      "grad_norm": 0.00048810808220878243,
+      "learning_rate": 0.00017722683908633812,
+      "loss": 23.0,
+      "step": 3674
+    },
+    {
+      "epoch": 2.2058823529411766,
+      "grad_norm": 0.003283157479017973,
+      "learning_rate": 0.00017721483624992215,
+      "loss": 23.0,
+      "step": 3675
+    },
+    {
+      "epoch": 2.2064825930372147,
+      "grad_norm": 0.0016707901377230883,
+      "learning_rate": 0.00017720283065790726,
+      "loss": 23.0,
+      "step": 3676
+    },
+    {
+      "epoch": 2.2070828331332533,
+      "grad_norm": 0.001358469482511282,
+      "learning_rate": 0.00017719082231072193,
+      "loss": 23.0,
+      "step": 3677
+    },
+    {
+      "epoch": 2.207683073229292,
+      "grad_norm": 0.0013346460182219744,
+      "learning_rate": 0.00017717881120879475,
+      "loss": 23.0,
+      "step": 3678
+    },
+    {
+      "epoch": 2.20828331332533,
+      "grad_norm": 0.001013239729218185,
+      "learning_rate": 0.0001771667973525543,
+      "loss": 23.0,
+      "step": 3679
+    },
+    {
+      "epoch": 2.2088835534213684,
+      "grad_norm": 0.001471264404244721,
+      "learning_rate": 0.00017715478074242932,
+      "loss": 23.0,
+      "step": 3680
+    },
+    {
+      "epoch": 2.209483793517407,
+      "grad_norm": 0.0015335743082687259,
+      "learning_rate": 0.00017714276137884873,
+      "loss": 23.0,
+      "step": 3681
+    },
+    {
+      "epoch": 2.2100840336134455,
+      "grad_norm": 0.005060260649770498,
+      "learning_rate": 0.0001771307392622414,
+      "loss": 23.0,
+      "step": 3682
+    },
+    {
+      "epoch": 2.2106842737094836,
+      "grad_norm": 0.0011684778146445751,
+      "learning_rate": 0.00017711871439303638,
+      "loss": 23.0,
+      "step": 3683
+    },
+    {
+      "epoch": 2.211284513805522,
+      "grad_norm": 0.0035212927032262087,
+      "learning_rate": 0.00017710668677166282,
+      "loss": 23.0,
+      "step": 3684
+    },
+    {
+      "epoch": 2.2118847539015607,
+      "grad_norm": 0.0023291881661862135,
+      "learning_rate": 0.00017709465639854997,
+      "loss": 23.0,
+      "step": 3685
+    },
+    {
+      "epoch": 2.2124849939975992,
+      "grad_norm": 0.0015391027554869652,
+      "learning_rate": 0.00017708262327412712,
+      "loss": 23.0,
+      "step": 3686
+    },
+    {
+      "epoch": 2.2130852340936373,
+      "grad_norm": 0.001002484350465238,
+      "learning_rate": 0.00017707058739882374,
+      "loss": 23.0,
+      "step": 3687
+    },
+    {
+      "epoch": 2.213685474189676,
+      "grad_norm": 0.0014180615544319153,
+      "learning_rate": 0.00017705854877306934,
+      "loss": 23.0,
+      "step": 3688
+    },
+    {
+      "epoch": 2.2142857142857144,
+      "grad_norm": 0.0008763477671891451,
+      "learning_rate": 0.00017704650739729357,
+      "loss": 23.0,
+      "step": 3689
+    },
+    {
+      "epoch": 2.2148859543817525,
+      "grad_norm": 0.0022361702285706997,
+      "learning_rate": 0.00017703446327192612,
+      "loss": 23.0,
+      "step": 3690
+    },
+    {
+      "epoch": 2.215486194477791,
+      "grad_norm": 0.0015827770112082362,
+      "learning_rate": 0.0001770224163973968,
+      "loss": 23.0,
+      "step": 3691
+    },
+    {
+      "epoch": 2.2160864345738296,
+      "grad_norm": 0.001904232893139124,
+      "learning_rate": 0.00017701036677413563,
+      "loss": 23.0,
+      "step": 3692
+    },
+    {
+      "epoch": 2.216686674669868,
+      "grad_norm": 0.0013081013457849622,
+      "learning_rate": 0.00017699831440257251,
+      "loss": 23.0,
+      "step": 3693
+    },
+    {
+      "epoch": 2.2172869147659062,
+      "grad_norm": 0.004817170090973377,
+      "learning_rate": 0.00017698625928313765,
+      "loss": 23.0,
+      "step": 3694
+    },
+    {
+      "epoch": 2.2178871548619448,
+      "grad_norm": 0.004000441636890173,
+      "learning_rate": 0.0001769742014162612,
+      "loss": 23.0,
+      "step": 3695
+    },
+    {
+      "epoch": 2.2184873949579833,
+      "grad_norm": 0.0027444609440863132,
+      "learning_rate": 0.00017696214080237352,
+      "loss": 23.0,
+      "step": 3696
+    },
+    {
+      "epoch": 2.2190876350540214,
+      "grad_norm": 0.002896189922466874,
+      "learning_rate": 0.000176950077441905,
+      "loss": 23.0,
+      "step": 3697
+    },
+    {
+      "epoch": 2.21968787515006,
+      "grad_norm": 0.002128873486071825,
+      "learning_rate": 0.00017693801133528616,
+      "loss": 23.0,
+      "step": 3698
+    },
+    {
+      "epoch": 2.2202881152460985,
+      "grad_norm": 0.0011179738212376833,
+      "learning_rate": 0.0001769259424829476,
+      "loss": 23.0,
+      "step": 3699
+    },
+    {
+      "epoch": 2.220888355342137,
+      "grad_norm": 0.002058587037026882,
+      "learning_rate": 0.00017691387088532003,
+      "loss": 23.0,
+      "step": 3700
+    },
+    {
+      "epoch": 2.221488595438175,
+      "grad_norm": 0.0022081804927438498,
+      "learning_rate": 0.00017690179654283425,
+      "loss": 23.0,
+      "step": 3701
+    },
+    {
+      "epoch": 2.2220888355342137,
+      "grad_norm": 0.0023553965147584677,
+      "learning_rate": 0.0001768897194559212,
+      "loss": 23.0,
+      "step": 3702
+    },
+    {
+      "epoch": 2.222689075630252,
+      "grad_norm": 0.001018383540213108,
+      "learning_rate": 0.00017687763962501182,
+      "loss": 23.0,
+      "step": 3703
+    },
+    {
+      "epoch": 2.2232893157262907,
+      "grad_norm": 0.0007036729366518557,
+      "learning_rate": 0.0001768655570505372,
+      "loss": 23.0,
+      "step": 3704
+    },
+    {
+      "epoch": 2.223889555822329,
+      "grad_norm": 0.0008182197925634682,
+      "learning_rate": 0.0001768534717329286,
+      "loss": 23.0,
+      "step": 3705
+    },
+    {
+      "epoch": 2.2244897959183674,
+      "grad_norm": 0.0010957723716273904,
+      "learning_rate": 0.00017684138367261731,
+      "loss": 23.0,
+      "step": 3706
+    },
+    {
+      "epoch": 2.225090036014406,
+      "grad_norm": 0.0017919333186000586,
+      "learning_rate": 0.00017682929287003466,
+      "loss": 23.0,
+      "step": 3707
+    },
+    {
+      "epoch": 2.225690276110444,
+      "grad_norm": 0.003724980168044567,
+      "learning_rate": 0.0001768171993256122,
+      "loss": 23.0,
+      "step": 3708
+    },
+    {
+      "epoch": 2.2262905162064826,
+      "grad_norm": 0.0013054338051006198,
+      "learning_rate": 0.00017680510303978147,
+      "loss": 23.0,
+      "step": 3709
+    },
+    {
+      "epoch": 2.226890756302521,
+      "grad_norm": 0.0018630026606842875,
+      "learning_rate": 0.0001767930040129742,
+      "loss": 23.0,
+      "step": 3710
+    },
+    {
+      "epoch": 2.2274909963985596,
+      "grad_norm": 0.0014727768721058965,
+      "learning_rate": 0.00017678090224562213,
+      "loss": 23.0,
+      "step": 3711
+    },
+    {
+      "epoch": 2.2280912364945977,
+      "grad_norm": 0.0010810722596943378,
+      "learning_rate": 0.0001767687977381572,
+      "loss": 23.0,
+      "step": 3712
+    },
+    {
+      "epoch": 2.2286914765906363,
+      "grad_norm": 0.001895120949484408,
+      "learning_rate": 0.00017675669049101132,
+      "loss": 23.0,
+      "step": 3713
+    },
+    {
+      "epoch": 2.229291716686675,
+      "grad_norm": 0.0021297503262758255,
+      "learning_rate": 0.0001767445805046166,
+      "loss": 23.0,
+      "step": 3714
+    },
+    {
+      "epoch": 2.229891956782713,
+      "grad_norm": 0.0009311452740803361,
+      "learning_rate": 0.00017673246777940522,
+      "loss": 23.0,
+      "step": 3715
+    },
+    {
+      "epoch": 2.2304921968787514,
+      "grad_norm": 0.0026380294002592564,
+      "learning_rate": 0.00017672035231580942,
+      "loss": 23.0,
+      "step": 3716
+    },
+    {
+      "epoch": 2.23109243697479,
+      "grad_norm": 0.0012547700898721814,
+      "learning_rate": 0.00017670823411426162,
+      "loss": 23.0,
+      "step": 3717
+    },
+    {
+      "epoch": 2.2316926770708285,
+      "grad_norm": 0.0012008714256808162,
+      "learning_rate": 0.0001766961131751943,
+      "loss": 23.0,
+      "step": 3718
+    },
+    {
+      "epoch": 2.2322929171668666,
+      "grad_norm": 0.0007726665935479105,
+      "learning_rate": 0.00017668398949903993,
+      "loss": 23.0,
+      "step": 3719
+    },
+    {
+      "epoch": 2.232893157262905,
+      "grad_norm": 0.003182122251018882,
+      "learning_rate": 0.00017667186308623122,
+      "loss": 23.0,
+      "step": 3720
+    },
+    {
+      "epoch": 2.2334933973589437,
+      "grad_norm": 0.0025740358978509903,
+      "learning_rate": 0.000176659733937201,
+      "loss": 23.0,
+      "step": 3721
+    },
+    {
+      "epoch": 2.234093637454982,
+      "grad_norm": 0.0025537547189742327,
+      "learning_rate": 0.00017664760205238203,
+      "loss": 23.0,
+      "step": 3722
+    },
+    {
+      "epoch": 2.2346938775510203,
+      "grad_norm": 0.0030393393244594336,
+      "learning_rate": 0.00017663546743220733,
+      "loss": 23.0,
+      "step": 3723
+    },
+    {
+      "epoch": 2.235294117647059,
+      "grad_norm": 0.0024942343588918447,
+      "learning_rate": 0.00017662333007710987,
+      "loss": 23.0,
+      "step": 3724
+    },
+    {
+      "epoch": 2.2358943577430974,
+      "grad_norm": 0.0016120598884299397,
+      "learning_rate": 0.0001766111899875229,
+      "loss": 23.0,
+      "step": 3725
+    },
+    {
+      "epoch": 2.2364945978391355,
+      "grad_norm": 0.00033892018836922944,
+      "learning_rate": 0.0001765990471638796,
+      "loss": 23.0,
+      "step": 3726
+    },
+    {
+      "epoch": 2.237094837935174,
+      "grad_norm": 0.002562824171036482,
+      "learning_rate": 0.00017658690160661338,
+      "loss": 23.0,
+      "step": 3727
+    },
+    {
+      "epoch": 2.2376950780312126,
+      "grad_norm": 0.0022175584454089403,
+      "learning_rate": 0.00017657475331615762,
+      "loss": 23.0,
+      "step": 3728
+    },
+    {
+      "epoch": 2.2382953181272507,
+      "grad_norm": 0.001756917918100953,
+      "learning_rate": 0.0001765626022929459,
+      "loss": 23.0,
+      "step": 3729
+    },
+    {
+      "epoch": 2.2388955582232892,
+      "grad_norm": 0.003884568577632308,
+      "learning_rate": 0.00017655044853741184,
+      "loss": 23.0,
+      "step": 3730
+    },
+    {
+      "epoch": 2.2394957983193278,
+      "grad_norm": 0.001167623675428331,
+      "learning_rate": 0.00017653829204998918,
+      "loss": 23.0,
+      "step": 3731
+    },
+    {
+      "epoch": 2.2400960384153663,
+      "grad_norm": 0.0010216180235147476,
+      "learning_rate": 0.00017652613283111176,
+      "loss": 23.0,
+      "step": 3732
+    },
+    {
+      "epoch": 2.2406962785114044,
+      "grad_norm": 0.003365274053066969,
+      "learning_rate": 0.00017651397088121353,
+      "loss": 23.0,
+      "step": 3733
+    },
+    {
+      "epoch": 2.241296518607443,
+      "grad_norm": 0.0018816734664142132,
+      "learning_rate": 0.00017650180620072846,
+      "loss": 23.0,
+      "step": 3734
+    },
+    {
+      "epoch": 2.2418967587034815,
+      "grad_norm": 0.0029328542295843363,
+      "learning_rate": 0.00017648963879009075,
+      "loss": 23.0,
+      "step": 3735
+    },
+    {
+      "epoch": 2.2424969987995196,
+      "grad_norm": 0.0007874789298512042,
+      "learning_rate": 0.00017647746864973458,
+      "loss": 23.0,
+      "step": 3736
+    },
+    {
+      "epoch": 2.243097238895558,
+      "grad_norm": 0.0030753707978874445,
+      "learning_rate": 0.00017646529578009426,
+      "loss": 23.0,
+      "step": 3737
+    },
+    {
+      "epoch": 2.2436974789915967,
+      "grad_norm": 0.0010473758447915316,
+      "learning_rate": 0.00017645312018160422,
+      "loss": 23.0,
+      "step": 3738
+    },
+    {
+      "epoch": 2.244297719087635,
+      "grad_norm": 0.0018478642450645566,
+      "learning_rate": 0.00017644094185469898,
+      "loss": 23.0,
+      "step": 3739
+    },
+    {
+      "epoch": 2.2448979591836733,
+      "grad_norm": 0.0013350150547921658,
+      "learning_rate": 0.0001764287607998132,
+      "loss": 23.0,
+      "step": 3740
+    },
+    {
+      "epoch": 2.245498199279712,
+      "grad_norm": 0.0012342286063358188,
+      "learning_rate": 0.00017641657701738148,
+      "loss": 23.0,
+      "step": 3741
+    },
+    {
+      "epoch": 2.2460984393757504,
+      "grad_norm": 0.0013594189658761024,
+      "learning_rate": 0.00017640439050783875,
+      "loss": 23.0,
+      "step": 3742
+    },
+    {
+      "epoch": 2.246698679471789,
+      "grad_norm": 0.0033450396731495857,
+      "learning_rate": 0.00017639220127161983,
+      "loss": 23.0,
+      "step": 3743
+    },
+    {
+      "epoch": 2.247298919567827,
+      "grad_norm": 0.002889930969104171,
+      "learning_rate": 0.00017638000930915974,
+      "loss": 23.0,
+      "step": 3744
+    },
+    {
+      "epoch": 2.2478991596638656,
+      "grad_norm": 0.0019997726194560528,
+      "learning_rate": 0.00017636781462089364,
+      "loss": 23.0,
+      "step": 3745
+    },
+    {
+      "epoch": 2.248499399759904,
+      "grad_norm": 0.0012661672662943602,
+      "learning_rate": 0.0001763556172072566,
+      "loss": 23.0,
+      "step": 3746
+    },
+    {
+      "epoch": 2.249099639855942,
+      "grad_norm": 0.007048753555864096,
+      "learning_rate": 0.00017634341706868403,
+      "loss": 23.0,
+      "step": 3747
+    },
+    {
+      "epoch": 2.2496998799519807,
+      "grad_norm": 0.0015775463543832302,
+      "learning_rate": 0.00017633121420561131,
+      "loss": 23.0,
+      "step": 3748
+    },
+    {
+      "epoch": 2.2503001200480193,
+      "grad_norm": 0.0020231371745467186,
+      "learning_rate": 0.00017631900861847389,
+      "loss": 23.0,
+      "step": 3749
+    },
+    {
+      "epoch": 2.250900360144058,
+      "grad_norm": 0.0012863476295024157,
+      "learning_rate": 0.00017630680030770735,
+      "loss": 23.0,
+      "step": 3750
+    },
+    {
+      "epoch": 2.251500600240096,
+      "grad_norm": 0.0016621587565168738,
+      "learning_rate": 0.0001762945892737474,
+      "loss": 23.0,
+      "step": 3751
+    },
+    {
+      "epoch": 2.2521008403361344,
+      "grad_norm": 0.00226738303899765,
+      "learning_rate": 0.00017628237551702982,
+      "loss": 23.0,
+      "step": 3752
+    },
+    {
+      "epoch": 2.252701080432173,
+      "grad_norm": 0.0014520992990583181,
+      "learning_rate": 0.00017627015903799048,
+      "loss": 23.0,
+      "step": 3753
+    },
+    {
+      "epoch": 2.2533013205282115,
+      "grad_norm": 0.0013434399152174592,
+      "learning_rate": 0.00017625793983706533,
+      "loss": 23.0,
+      "step": 3754
+    },
+    {
+      "epoch": 2.2539015606242496,
+      "grad_norm": 0.0007136189378798008,
+      "learning_rate": 0.00017624571791469048,
+      "loss": 23.0,
+      "step": 3755
+    },
+    {
+      "epoch": 2.254501800720288,
+      "grad_norm": 0.002455267123878002,
+      "learning_rate": 0.00017623349327130206,
+      "loss": 23.0,
+      "step": 3756
+    },
+    {
+      "epoch": 2.2551020408163267,
+      "grad_norm": 0.0009659494389779866,
+      "learning_rate": 0.00017622126590733644,
+      "loss": 23.0,
+      "step": 3757
+    },
+    {
+      "epoch": 2.255702280912365,
+      "grad_norm": 0.002808001358062029,
+      "learning_rate": 0.00017620903582322983,
+      "loss": 23.0,
+      "step": 3758
+    },
+    {
+      "epoch": 2.2563025210084033,
+      "grad_norm": 0.0018349566962569952,
+      "learning_rate": 0.0001761968030194188,
+      "loss": 23.0,
+      "step": 3759
+    },
+    {
+      "epoch": 2.256902761104442,
+      "grad_norm": 0.002876591170206666,
+      "learning_rate": 0.00017618456749633987,
+      "loss": 23.0,
+      "step": 3760
+    },
+    {
+      "epoch": 2.2575030012004804,
+      "grad_norm": 0.0012843102449551225,
+      "learning_rate": 0.00017617232925442968,
+      "loss": 23.0,
+      "step": 3761
+    },
+    {
+      "epoch": 2.2581032412965185,
+      "grad_norm": 0.0011905563296750188,
+      "learning_rate": 0.00017616008829412502,
+      "loss": 23.0,
+      "step": 3762
+    },
+    {
+      "epoch": 2.258703481392557,
+      "grad_norm": 0.000979992444626987,
+      "learning_rate": 0.00017614784461586266,
+      "loss": 23.0,
+      "step": 3763
+    },
+    {
+      "epoch": 2.2593037214885956,
+      "grad_norm": 0.002167720813304186,
+      "learning_rate": 0.00017613559822007963,
+      "loss": 23.0,
+      "step": 3764
+    },
+    {
+      "epoch": 2.2599039615846337,
+      "grad_norm": 0.0008691643597558141,
+      "learning_rate": 0.00017612334910721298,
+      "loss": 23.0,
+      "step": 3765
+    },
+    {
+      "epoch": 2.2605042016806722,
+      "grad_norm": 0.0019571988377720118,
+      "learning_rate": 0.00017611109727769977,
+      "loss": 23.0,
+      "step": 3766
+    },
+    {
+      "epoch": 2.2611044417767108,
+      "grad_norm": 0.0012904416071251035,
+      "learning_rate": 0.00017609884273197732,
+      "loss": 23.0,
+      "step": 3767
+    },
+    {
+      "epoch": 2.2617046818727493,
+      "grad_norm": 0.0017439661314710975,
+      "learning_rate": 0.00017608658547048288,
+      "loss": 23.0,
+      "step": 3768
+    },
+    {
+      "epoch": 2.2623049219687874,
+      "grad_norm": 0.003602192969992757,
+      "learning_rate": 0.00017607432549365394,
+      "loss": 23.0,
+      "step": 3769
+    },
+    {
+      "epoch": 2.262905162064826,
+      "grad_norm": 0.002050132257863879,
+      "learning_rate": 0.00017606206280192801,
+      "loss": 23.0,
+      "step": 3770
+    },
+    {
+      "epoch": 2.2635054021608645,
+      "grad_norm": 0.0030350079759955406,
+      "learning_rate": 0.00017604979739574273,
+      "loss": 23.0,
+      "step": 3771
+    },
+    {
+      "epoch": 2.2641056422569026,
+      "grad_norm": 0.0011385931866243482,
+      "learning_rate": 0.0001760375292755358,
+      "loss": 23.0,
+      "step": 3772
+    },
+    {
+      "epoch": 2.264705882352941,
+      "grad_norm": 0.0015834002988412976,
+      "learning_rate": 0.000176025258441745,
+      "loss": 23.0,
+      "step": 3773
+    },
+    {
+      "epoch": 2.2653061224489797,
+      "grad_norm": 0.002169420477002859,
+      "learning_rate": 0.00017601298489480834,
+      "loss": 23.0,
+      "step": 3774
+    },
+    {
+      "epoch": 2.265906362545018,
+      "grad_norm": 0.0020891001913696527,
+      "learning_rate": 0.00017600070863516373,
+      "loss": 23.0,
+      "step": 3775
+    },
+    {
+      "epoch": 2.2665066026410563,
+      "grad_norm": 0.0015407392056658864,
+      "learning_rate": 0.00017598842966324937,
+      "loss": 23.0,
+      "step": 3776
+    },
+    {
+      "epoch": 2.267106842737095,
+      "grad_norm": 0.0012298728106543422,
+      "learning_rate": 0.0001759761479795034,
+      "loss": 23.0,
+      "step": 3777
+    },
+    {
+      "epoch": 2.2677070828331334,
+      "grad_norm": 0.0022670759353786707,
+      "learning_rate": 0.00017596386358436413,
+      "loss": 23.0,
+      "step": 3778
+    },
+    {
+      "epoch": 2.2683073229291715,
+      "grad_norm": 0.0009981478797271848,
+      "learning_rate": 0.00017595157647827,
+      "loss": 23.0,
+      "step": 3779
+    },
+    {
+      "epoch": 2.26890756302521,
+      "grad_norm": 0.0004938762285746634,
+      "learning_rate": 0.00017593928666165945,
+      "loss": 23.0,
+      "step": 3780
+    },
+    {
+      "epoch": 2.2695078031212486,
+      "grad_norm": 0.0011169624049216509,
+      "learning_rate": 0.00017592699413497114,
+      "loss": 23.0,
+      "step": 3781
+    },
+    {
+      "epoch": 2.270108043217287,
+      "grad_norm": 0.003080305876210332,
+      "learning_rate": 0.00017591469889864368,
+      "loss": 23.0,
+      "step": 3782
+    },
+    {
+      "epoch": 2.270708283313325,
+      "grad_norm": 0.004471518564969301,
+      "learning_rate": 0.00017590240095311587,
+      "loss": 23.0,
+      "step": 3783
+    },
+    {
+      "epoch": 2.2713085234093637,
+      "grad_norm": 0.0011711474508047104,
+      "learning_rate": 0.00017589010029882664,
+      "loss": 23.0,
+      "step": 3784
+    },
+    {
+      "epoch": 2.2719087635054023,
+      "grad_norm": 0.0011285408399999142,
+      "learning_rate": 0.00017587779693621495,
+      "loss": 23.0,
+      "step": 3785
+    },
+    {
+      "epoch": 2.2725090036014404,
+      "grad_norm": 0.0020412777084857225,
+      "learning_rate": 0.00017586549086571988,
+      "loss": 23.0,
+      "step": 3786
+    },
+    {
+      "epoch": 2.273109243697479,
+      "grad_norm": 0.001386416843160987,
+      "learning_rate": 0.00017585318208778058,
+      "loss": 23.0,
+      "step": 3787
+    },
+    {
+      "epoch": 2.2737094837935174,
+      "grad_norm": 0.0021522578317672014,
+      "learning_rate": 0.00017584087060283632,
+      "loss": 23.0,
+      "step": 3788
+    },
+    {
+      "epoch": 2.274309723889556,
+      "grad_norm": 0.0017842737725004554,
+      "learning_rate": 0.00017582855641132646,
+      "loss": 23.0,
+      "step": 3789
+    },
+    {
+      "epoch": 2.274909963985594,
+      "grad_norm": 0.0008319435291923583,
+      "learning_rate": 0.0001758162395136905,
+      "loss": 23.0,
+      "step": 3790
+    },
+    {
+      "epoch": 2.2755102040816326,
+      "grad_norm": 0.0022286258172243834,
+      "learning_rate": 0.000175803919910368,
+      "loss": 23.0,
+      "step": 3791
+    },
+    {
+      "epoch": 2.276110444177671,
+      "grad_norm": 0.0015450834762305021,
+      "learning_rate": 0.00017579159760179854,
+      "loss": 23.0,
+      "step": 3792
+    },
+    {
+      "epoch": 2.2767106842737093,
+      "grad_norm": 0.0017403220990672708,
+      "learning_rate": 0.00017577927258842193,
+      "loss": 23.0,
+      "step": 3793
+    },
+    {
+      "epoch": 2.277310924369748,
+      "grad_norm": 0.0021773509215563536,
+      "learning_rate": 0.000175766944870678,
+      "loss": 23.0,
+      "step": 3794
+    },
+    {
+      "epoch": 2.2779111644657863,
+      "grad_norm": 0.0007699204725213349,
+      "learning_rate": 0.00017575461444900676,
+      "loss": 23.0,
+      "step": 3795
+    },
+    {
+      "epoch": 2.278511404561825,
+      "grad_norm": 0.001452354365028441,
+      "learning_rate": 0.00017574228132384814,
+      "loss": 23.0,
+      "step": 3796
+    },
+    {
+      "epoch": 2.279111644657863,
+      "grad_norm": 0.0027013334911316633,
+      "learning_rate": 0.00017572994549564235,
+      "loss": 23.0,
+      "step": 3797
+    },
+    {
+      "epoch": 2.2797118847539015,
+      "grad_norm": 0.0016655053477734327,
+      "learning_rate": 0.0001757176069648296,
+      "loss": 23.0,
+      "step": 3798
+    },
+    {
+      "epoch": 2.28031212484994,
+      "grad_norm": 0.0017022525426000357,
+      "learning_rate": 0.00017570526573185022,
+      "loss": 23.0,
+      "step": 3799
+    },
+    {
+      "epoch": 2.280912364945978,
+      "grad_norm": 0.0011267677182331681,
+      "learning_rate": 0.00017569292179714466,
+      "loss": 23.0,
+      "step": 3800
+    },
+    {
+      "epoch": 2.2815126050420167,
+      "grad_norm": 0.0012624160153791308,
+      "learning_rate": 0.00017568057516115343,
+      "loss": 23.0,
+      "step": 3801
+    },
+    {
+      "epoch": 2.2821128451380552,
+      "grad_norm": 0.0012623511720448732,
+      "learning_rate": 0.00017566822582431714,
+      "loss": 23.0,
+      "step": 3802
+    },
+    {
+      "epoch": 2.2827130852340938,
+      "grad_norm": 0.002993621863424778,
+      "learning_rate": 0.0001756558737870765,
+      "loss": 23.0,
+      "step": 3803
+    },
+    {
+      "epoch": 2.283313325330132,
+      "grad_norm": 0.001764954999089241,
+      "learning_rate": 0.00017564351904987232,
+      "loss": 23.0,
+      "step": 3804
+    },
+    {
+      "epoch": 2.2839135654261704,
+      "grad_norm": 0.0033577526919543743,
+      "learning_rate": 0.00017563116161314557,
+      "loss": 23.0,
+      "step": 3805
+    },
+    {
+      "epoch": 2.284513805522209,
+      "grad_norm": 0.0030459025874733925,
+      "learning_rate": 0.0001756188014773372,
+      "loss": 23.0,
+      "step": 3806
+    },
+    {
+      "epoch": 2.2851140456182475,
+      "grad_norm": 0.0020036788191646338,
+      "learning_rate": 0.00017560643864288831,
+      "loss": 23.0,
+      "step": 3807
+    },
+    {
+      "epoch": 2.2857142857142856,
+      "grad_norm": 0.003118950640782714,
+      "learning_rate": 0.00017559407311024008,
+      "loss": 23.0,
+      "step": 3808
+    },
+    {
+      "epoch": 2.286314525810324,
+      "grad_norm": 0.0023817415349185467,
+      "learning_rate": 0.00017558170487983388,
+      "loss": 23.0,
+      "step": 3809
+    },
+    {
+      "epoch": 2.2869147659063627,
+      "grad_norm": 0.0020011558663100004,
+      "learning_rate": 0.00017556933395211099,
+      "loss": 23.0,
+      "step": 3810
+    },
+    {
+      "epoch": 2.287515006002401,
+      "grad_norm": 0.0019172454485669732,
+      "learning_rate": 0.000175556960327513,
+      "loss": 23.0,
+      "step": 3811
+    },
+    {
+      "epoch": 2.2881152460984393,
+      "grad_norm": 0.005320819094777107,
+      "learning_rate": 0.00017554458400648146,
+      "loss": 23.0,
+      "step": 3812
+    },
+    {
+      "epoch": 2.288715486194478,
+      "grad_norm": 0.0011504385620355606,
+      "learning_rate": 0.00017553220498945802,
+      "loss": 23.0,
+      "step": 3813
+    },
+    {
+      "epoch": 2.2893157262905164,
+      "grad_norm": 0.004282000940293074,
+      "learning_rate": 0.00017551982327688447,
+      "loss": 23.0,
+      "step": 3814
+    },
+    {
+      "epoch": 2.2899159663865545,
+      "grad_norm": 0.0016033421270549297,
+      "learning_rate": 0.0001755074388692027,
+      "loss": 23.0,
+      "step": 3815
+    },
+    {
+      "epoch": 2.290516206482593,
+      "grad_norm": 0.001392611302435398,
+      "learning_rate": 0.00017549505176685468,
+      "loss": 23.0,
+      "step": 3816
+    },
+    {
+      "epoch": 2.2911164465786316,
+      "grad_norm": 0.0024148111697286367,
+      "learning_rate": 0.00017548266197028243,
+      "loss": 23.0,
+      "step": 3817
+    },
+    {
+      "epoch": 2.29171668667467,
+      "grad_norm": 0.0022317797411233187,
+      "learning_rate": 0.00017547026947992817,
+      "loss": 23.0,
+      "step": 3818
+    },
+    {
+      "epoch": 2.292316926770708,
+      "grad_norm": 0.00245701358653605,
+      "learning_rate": 0.0001754578742962341,
+      "loss": 23.0,
+      "step": 3819
+    },
+    {
+      "epoch": 2.2929171668667467,
+      "grad_norm": 0.0017128176987171173,
+      "learning_rate": 0.00017544547641964263,
+      "loss": 23.0,
+      "step": 3820
+    },
+    {
+      "epoch": 2.2935174069627853,
+      "grad_norm": 0.0031775077804923058,
+      "learning_rate": 0.00017543307585059613,
+      "loss": 23.0,
+      "step": 3821
+    },
+    {
+      "epoch": 2.2941176470588234,
+      "grad_norm": 0.002050775336101651,
+      "learning_rate": 0.00017542067258953723,
+      "loss": 23.0,
+      "step": 3822
+    },
+    {
+      "epoch": 2.294717887154862,
+      "grad_norm": 0.0014890647726133466,
+      "learning_rate": 0.00017540826663690851,
+      "loss": 23.0,
+      "step": 3823
+    },
+    {
+      "epoch": 2.2953181272509005,
+      "grad_norm": 0.004480647388845682,
+      "learning_rate": 0.00017539585799315272,
+      "loss": 23.0,
+      "step": 3824
+    },
+    {
+      "epoch": 2.295918367346939,
+      "grad_norm": 0.0015886913752183318,
+      "learning_rate": 0.0001753834466587127,
+      "loss": 23.0,
+      "step": 3825
+    },
+    {
+      "epoch": 2.296518607442977,
+      "grad_norm": 0.0019695230294018984,
+      "learning_rate": 0.0001753710326340314,
+      "loss": 23.0,
+      "step": 3826
+    },
+    {
+      "epoch": 2.2971188475390156,
+      "grad_norm": 0.0018076825654134154,
+      "learning_rate": 0.0001753586159195518,
+      "loss": 23.0,
+      "step": 3827
+    },
+    {
+      "epoch": 2.297719087635054,
+      "grad_norm": 0.002728229621425271,
+      "learning_rate": 0.00017534619651571705,
+      "loss": 23.0,
+      "step": 3828
+    },
+    {
+      "epoch": 2.2983193277310923,
+      "grad_norm": 0.0017432133900001645,
+      "learning_rate": 0.0001753337744229704,
+      "loss": 23.0,
+      "step": 3829
+    },
+    {
+      "epoch": 2.298919567827131,
+      "grad_norm": 0.0018603568896651268,
+      "learning_rate": 0.00017532134964175504,
+      "loss": 23.0,
+      "step": 3830
+    },
+    {
+      "epoch": 2.2995198079231693,
+      "grad_norm": 0.002048965310677886,
+      "learning_rate": 0.0001753089221725145,
+      "loss": 23.0,
+      "step": 3831
+    },
+    {
+      "epoch": 2.300120048019208,
+      "grad_norm": 0.001226543914526701,
+      "learning_rate": 0.00017529649201569226,
+      "loss": 23.0,
+      "step": 3832
+    },
+    {
+      "epoch": 2.300720288115246,
+      "grad_norm": 0.004013392608612776,
+      "learning_rate": 0.0001752840591717319,
+      "loss": 23.0,
+      "step": 3833
+    },
+    {
+      "epoch": 2.3013205282112845,
+      "grad_norm": 0.0008342837681993842,
+      "learning_rate": 0.00017527162364107713,
+      "loss": 23.0,
+      "step": 3834
+    },
+    {
+      "epoch": 2.301920768307323,
+      "grad_norm": 0.0007289492059499025,
+      "learning_rate": 0.00017525918542417171,
+      "loss": 23.0,
+      "step": 3835
+    },
+    {
+      "epoch": 2.302521008403361,
+      "grad_norm": 0.0029447299893945456,
+      "learning_rate": 0.00017524674452145956,
+      "loss": 23.0,
+      "step": 3836
+    },
+    {
+      "epoch": 2.3031212484993997,
+      "grad_norm": 0.002568695694208145,
+      "learning_rate": 0.00017523430093338468,
+      "loss": 23.0,
+      "step": 3837
+    },
+    {
+      "epoch": 2.3037214885954382,
+      "grad_norm": 0.0012608342804014683,
+      "learning_rate": 0.0001752218546603911,
+      "loss": 23.0,
+      "step": 3838
+    },
+    {
+      "epoch": 2.304321728691477,
+      "grad_norm": 0.00195410312153399,
+      "learning_rate": 0.00017520940570292303,
+      "loss": 23.0,
+      "step": 3839
+    },
+    {
+      "epoch": 2.304921968787515,
+      "grad_norm": 0.003413090016692877,
+      "learning_rate": 0.00017519695406142472,
+      "loss": 23.0,
+      "step": 3840
+    },
+    {
+      "epoch": 2.3055222088835534,
+      "grad_norm": 0.002465988975018263,
+      "learning_rate": 0.00017518449973634055,
+      "loss": 23.0,
+      "step": 3841
+    },
+    {
+      "epoch": 2.306122448979592,
+      "grad_norm": 0.00117593037430197,
+      "learning_rate": 0.00017517204272811501,
+      "loss": 23.0,
+      "step": 3842
+    },
+    {
+      "epoch": 2.30672268907563,
+      "grad_norm": 0.00045529394992627203,
+      "learning_rate": 0.0001751595830371926,
+      "loss": 23.0,
+      "step": 3843
+    },
+    {
+      "epoch": 2.3073229291716686,
+      "grad_norm": 0.0013716558460146189,
+      "learning_rate": 0.000175147120664018,
+      "loss": 23.0,
+      "step": 3844
+    },
+    {
+      "epoch": 2.307923169267707,
+      "grad_norm": 0.0009694201871752739,
+      "learning_rate": 0.000175134655609036,
+      "loss": 23.0,
+      "step": 3845
+    },
+    {
+      "epoch": 2.3085234093637457,
+      "grad_norm": 0.002793167717754841,
+      "learning_rate": 0.00017512218787269138,
+      "loss": 23.0,
+      "step": 3846
+    },
+    {
+      "epoch": 2.3091236494597838,
+      "grad_norm": 0.0012160228798165917,
+      "learning_rate": 0.0001751097174554291,
+      "loss": 23.0,
+      "step": 3847
+    },
+    {
+      "epoch": 2.3097238895558223,
+      "grad_norm": 0.0041273715905845165,
+      "learning_rate": 0.00017509724435769427,
+      "loss": 23.0,
+      "step": 3848
+    },
+    {
+      "epoch": 2.310324129651861,
+      "grad_norm": 0.0019413263071328402,
+      "learning_rate": 0.00017508476857993192,
+      "loss": 23.0,
+      "step": 3849
+    },
+    {
+      "epoch": 2.310924369747899,
+      "grad_norm": 0.000679299992043525,
+      "learning_rate": 0.00017507229012258732,
+      "loss": 23.0,
+      "step": 3850
+    },
+    {
+      "epoch": 2.3115246098439375,
+      "grad_norm": 0.0014612323138862848,
+      "learning_rate": 0.0001750598089861058,
+      "loss": 23.0,
+      "step": 3851
+    },
+    {
+      "epoch": 2.312124849939976,
+      "grad_norm": 0.002256182488054037,
+      "learning_rate": 0.00017504732517093278,
+      "loss": 23.0,
+      "step": 3852
+    },
+    {
+      "epoch": 2.3127250900360146,
+      "grad_norm": 0.0009005808969959617,
+      "learning_rate": 0.00017503483867751377,
+      "loss": 23.0,
+      "step": 3853
+    },
+    {
+      "epoch": 2.3133253301320527,
+      "grad_norm": 0.0015893650706857443,
+      "learning_rate": 0.00017502234950629437,
+      "loss": 23.0,
+      "step": 3854
+    },
+    {
+      "epoch": 2.313925570228091,
+      "grad_norm": 0.001747640548273921,
+      "learning_rate": 0.0001750098576577203,
+      "loss": 23.0,
+      "step": 3855
+    },
+    {
+      "epoch": 2.3145258103241297,
+      "grad_norm": 0.0005708754761144519,
+      "learning_rate": 0.00017499736313223737,
+      "loss": 23.0,
+      "step": 3856
+    },
+    {
+      "epoch": 2.315126050420168,
+      "grad_norm": 0.0028467399533838034,
+      "learning_rate": 0.00017498486593029144,
+      "loss": 23.0,
+      "step": 3857
+    },
+    {
+      "epoch": 2.3157262905162064,
+      "grad_norm": 0.0014178903074935079,
+      "learning_rate": 0.00017497236605232855,
+      "loss": 23.0,
+      "step": 3858
+    },
+    {
+      "epoch": 2.316326530612245,
+      "grad_norm": 0.005164925009012222,
+      "learning_rate": 0.00017495986349879477,
+      "loss": 23.0,
+      "step": 3859
+    },
+    {
+      "epoch": 2.3169267707082835,
+      "grad_norm": 0.0027939407154917717,
+      "learning_rate": 0.00017494735827013627,
+      "loss": 23.0,
+      "step": 3860
+    },
+    {
+      "epoch": 2.3175270108043216,
+      "grad_norm": 0.002403461141511798,
+      "learning_rate": 0.00017493485036679935,
+      "loss": 23.0,
+      "step": 3861
+    },
+    {
+      "epoch": 2.31812725090036,
+      "grad_norm": 0.003365919226780534,
+      "learning_rate": 0.00017492233978923035,
+      "loss": 23.0,
+      "step": 3862
+    },
+    {
+      "epoch": 2.3187274909963986,
+      "grad_norm": 0.001767502399161458,
+      "learning_rate": 0.00017490982653787577,
+      "loss": 23.0,
+      "step": 3863
+    },
+    {
+      "epoch": 2.3193277310924367,
+      "grad_norm": 0.0037461065221577883,
+      "learning_rate": 0.0001748973106131822,
+      "loss": 23.0,
+      "step": 3864
+    },
+    {
+      "epoch": 2.3199279711884753,
+      "grad_norm": 0.0054776170291006565,
+      "learning_rate": 0.00017488479201559627,
+      "loss": 23.0,
+      "step": 3865
+    },
+    {
+      "epoch": 2.320528211284514,
+      "grad_norm": 0.0020852498710155487,
+      "learning_rate": 0.00017487227074556474,
+      "loss": 23.0,
+      "step": 3866
+    },
+    {
+      "epoch": 2.3211284513805523,
+      "grad_norm": 0.001896170899271965,
+      "learning_rate": 0.00017485974680353446,
+      "loss": 23.0,
+      "step": 3867
+    },
+    {
+      "epoch": 2.3217286914765904,
+      "grad_norm": 0.0022559871431440115,
+      "learning_rate": 0.00017484722018995235,
+      "loss": 23.0,
+      "step": 3868
+    },
+    {
+      "epoch": 2.322328931572629,
+      "grad_norm": 0.0008233110420405865,
+      "learning_rate": 0.00017483469090526552,
+      "loss": 23.0,
+      "step": 3869
+    },
+    {
+      "epoch": 2.3229291716686675,
+      "grad_norm": 0.003072496736422181,
+      "learning_rate": 0.00017482215894992106,
+      "loss": 23.0,
+      "step": 3870
+    },
+    {
+      "epoch": 2.323529411764706,
+      "grad_norm": 0.0034991884604096413,
+      "learning_rate": 0.00017480962432436618,
+      "loss": 23.0,
+      "step": 3871
+    },
+    {
+      "epoch": 2.324129651860744,
+      "grad_norm": 0.002439623698592186,
+      "learning_rate": 0.00017479708702904827,
+      "loss": 23.0,
+      "step": 3872
+    },
+    {
+      "epoch": 2.3247298919567827,
+      "grad_norm": 0.0018866292666643858,
+      "learning_rate": 0.00017478454706441472,
+      "loss": 23.0,
+      "step": 3873
+    },
+    {
+      "epoch": 2.3253301320528212,
+      "grad_norm": 0.00198563514277339,
+      "learning_rate": 0.00017477200443091306,
+      "loss": 23.0,
+      "step": 3874
+    },
+    {
+      "epoch": 2.32593037214886,
+      "grad_norm": 0.005812112707644701,
+      "learning_rate": 0.0001747594591289909,
+      "loss": 23.0,
+      "step": 3875
+    },
+    {
+      "epoch": 2.326530612244898,
+      "grad_norm": 0.0013709774939343333,
+      "learning_rate": 0.00017474691115909594,
+      "loss": 23.0,
+      "step": 3876
+    },
+    {
+      "epoch": 2.3271308523409364,
+      "grad_norm": 0.0018185654189437628,
+      "learning_rate": 0.00017473436052167597,
+      "loss": 23.0,
+      "step": 3877
+    },
+    {
+      "epoch": 2.327731092436975,
+      "grad_norm": 0.002282187808305025,
+      "learning_rate": 0.00017472180721717895,
+      "loss": 23.0,
+      "step": 3878
+    },
+    {
+      "epoch": 2.328331332533013,
+      "grad_norm": 0.0013887138338759542,
+      "learning_rate": 0.00017470925124605282,
+      "loss": 23.0,
+      "step": 3879
+    },
+    {
+      "epoch": 2.3289315726290516,
+      "grad_norm": 0.00287288217805326,
+      "learning_rate": 0.0001746966926087457,
+      "loss": 23.0,
+      "step": 3880
+    },
+    {
+      "epoch": 2.32953181272509,
+      "grad_norm": 0.0033778510987758636,
+      "learning_rate": 0.00017468413130570577,
+      "loss": 23.0,
+      "step": 3881
+    },
+    {
+      "epoch": 2.3301320528211287,
+      "grad_norm": 0.0016036947490647435,
+      "learning_rate": 0.00017467156733738128,
+      "loss": 23.0,
+      "step": 3882
+    },
+    {
+      "epoch": 2.3307322929171668,
+      "grad_norm": 0.0013769504148513079,
+      "learning_rate": 0.00017465900070422064,
+      "loss": 23.0,
+      "step": 3883
+    },
+    {
+      "epoch": 2.3313325330132053,
+      "grad_norm": 0.00230945716612041,
+      "learning_rate": 0.0001746464314066723,
+      "loss": 23.0,
+      "step": 3884
+    },
+    {
+      "epoch": 2.331932773109244,
+      "grad_norm": 0.0031504861544817686,
+      "learning_rate": 0.00017463385944518485,
+      "loss": 23.0,
+      "step": 3885
+    },
+    {
+      "epoch": 2.332533013205282,
+      "grad_norm": 0.0008833253523334861,
+      "learning_rate": 0.00017462128482020694,
+      "loss": 23.0,
+      "step": 3886
+    },
+    {
+      "epoch": 2.3331332533013205,
+      "grad_norm": 0.002419440308585763,
+      "learning_rate": 0.00017460870753218733,
+      "loss": 23.0,
+      "step": 3887
+    },
+    {
+      "epoch": 2.333733493397359,
+      "grad_norm": 0.00045076129026710987,
+      "learning_rate": 0.00017459612758157484,
+      "loss": 23.0,
+      "step": 3888
+    },
+    {
+      "epoch": 2.3343337334933976,
+      "grad_norm": 0.0007832158589735627,
+      "learning_rate": 0.00017458354496881846,
+      "loss": 23.0,
+      "step": 3889
+    },
+    {
+      "epoch": 2.3349339735894357,
+      "grad_norm": 0.0010271031642332673,
+      "learning_rate": 0.00017457095969436722,
+      "loss": 23.0,
+      "step": 3890
+    },
+    {
+      "epoch": 2.335534213685474,
+      "grad_norm": 0.002968541579321027,
+      "learning_rate": 0.0001745583717586702,
+      "loss": 23.0,
+      "step": 3891
+    },
+    {
+      "epoch": 2.3361344537815127,
+      "grad_norm": 0.005150226876139641,
+      "learning_rate": 0.00017454578116217674,
+      "loss": 23.0,
+      "step": 3892
+    },
+    {
+      "epoch": 2.336734693877551,
+      "grad_norm": 0.0014614604879170656,
+      "learning_rate": 0.00017453318790533606,
+      "loss": 23.0,
+      "step": 3893
+    },
+    {
+      "epoch": 2.3373349339735894,
+      "grad_norm": 0.003210582537576556,
+      "learning_rate": 0.00017452059198859766,
+      "loss": 23.0,
+      "step": 3894
+    },
+    {
+      "epoch": 2.337935174069628,
+      "grad_norm": 0.0043360283598303795,
+      "learning_rate": 0.000174507993412411,
+      "loss": 23.0,
+      "step": 3895
+    },
+    {
+      "epoch": 2.3385354141656665,
+      "grad_norm": 0.0008345422684215009,
+      "learning_rate": 0.00017449539217722571,
+      "loss": 23.0,
+      "step": 3896
+    },
+    {
+      "epoch": 2.3391356542617046,
+      "grad_norm": 0.003543426748365164,
+      "learning_rate": 0.00017448278828349152,
+      "loss": 23.0,
+      "step": 3897
+    },
+    {
+      "epoch": 2.339735894357743,
+      "grad_norm": 0.0006024944595992565,
+      "learning_rate": 0.00017447018173165818,
+      "loss": 23.0,
+      "step": 3898
+    },
+    {
+      "epoch": 2.3403361344537816,
+      "grad_norm": 0.002525531454011798,
+      "learning_rate": 0.00017445757252217562,
+      "loss": 23.0,
+      "step": 3899
+    },
+    {
+      "epoch": 2.3409363745498197,
+      "grad_norm": 0.0018678716151043773,
+      "learning_rate": 0.00017444496065549385,
+      "loss": 23.0,
+      "step": 3900
+    },
+    {
+      "epoch": 2.3415366146458583,
+      "grad_norm": 0.002660698490217328,
+      "learning_rate": 0.00017443234613206293,
+      "loss": 23.0,
+      "step": 3901
+    },
+    {
+      "epoch": 2.342136854741897,
+      "grad_norm": 0.0027779710944741964,
+      "learning_rate": 0.000174419728952333,
+      "loss": 23.0,
+      "step": 3902
+    },
+    {
+      "epoch": 2.3427370948379354,
+      "grad_norm": 0.0036743993405252695,
+      "learning_rate": 0.0001744071091167544,
+      "loss": 23.0,
+      "step": 3903
+    },
+    {
+      "epoch": 2.3433373349339734,
+      "grad_norm": 0.002380775986239314,
+      "learning_rate": 0.00017439448662577749,
+      "loss": 23.0,
+      "step": 3904
+    },
+    {
+      "epoch": 2.343937575030012,
+      "grad_norm": 0.0030045825988054276,
+      "learning_rate": 0.0001743818614798527,
+      "loss": 23.0,
+      "step": 3905
+    },
+    {
+      "epoch": 2.3445378151260505,
+      "grad_norm": 0.0010804989142343402,
+      "learning_rate": 0.00017436923367943058,
+      "loss": 23.0,
+      "step": 3906
+    },
+    {
+      "epoch": 2.3451380552220886,
+      "grad_norm": 0.0031437580473721027,
+      "learning_rate": 0.00017435660322496187,
+      "loss": 23.0,
+      "step": 3907
+    },
+    {
+      "epoch": 2.345738295318127,
+      "grad_norm": 0.002179528819397092,
+      "learning_rate": 0.0001743439701168972,
+      "loss": 23.0,
+      "step": 3908
+    },
+    {
+      "epoch": 2.3463385354141657,
+      "grad_norm": 0.0047058421187102795,
+      "learning_rate": 0.0001743313343556875,
+      "loss": 23.0,
+      "step": 3909
+    },
+    {
+      "epoch": 2.3469387755102042,
+      "grad_norm": 0.0017768297111615539,
+      "learning_rate": 0.00017431869594178366,
+      "loss": 23.0,
+      "step": 3910
+    },
+    {
+      "epoch": 2.3475390156062423,
+      "grad_norm": 0.0016148180002346635,
+      "learning_rate": 0.00017430605487563677,
+      "loss": 23.0,
+      "step": 3911
+    },
+    {
+      "epoch": 2.348139255702281,
+      "grad_norm": 0.001844774466007948,
+      "learning_rate": 0.00017429341115769787,
+      "loss": 23.0,
+      "step": 3912
+    },
+    {
+      "epoch": 2.3487394957983194,
+      "grad_norm": 0.001695389044471085,
+      "learning_rate": 0.0001742807647884183,
+      "loss": 23.0,
+      "step": 3913
+    },
+    {
+      "epoch": 2.3493397358943575,
+      "grad_norm": 0.0021679711062461138,
+      "learning_rate": 0.00017426811576824923,
+      "loss": 23.0,
+      "step": 3914
+    },
+    {
+      "epoch": 2.349939975990396,
+      "grad_norm": 0.0033461309503763914,
+      "learning_rate": 0.00017425546409764218,
+      "loss": 23.0,
+      "step": 3915
+    },
+    {
+      "epoch": 2.3505402160864346,
+      "grad_norm": 0.0019258579704910517,
+      "learning_rate": 0.00017424280977704863,
+      "loss": 23.0,
+      "step": 3916
+    },
+    {
+      "epoch": 2.351140456182473,
+      "grad_norm": 0.0012793492060154676,
+      "learning_rate": 0.0001742301528069202,
+      "loss": 23.0,
+      "step": 3917
+    },
+    {
+      "epoch": 2.3517406962785112,
+      "grad_norm": 0.0014299890026450157,
+      "learning_rate": 0.00017421749318770853,
+      "loss": 23.0,
+      "step": 3918
+    },
+    {
+      "epoch": 2.3523409363745498,
+      "grad_norm": 0.0012088280636817217,
+      "learning_rate": 0.00017420483091986543,
+      "loss": 23.0,
+      "step": 3919
+    },
+    {
+      "epoch": 2.3529411764705883,
+      "grad_norm": 0.0010071590077131987,
+      "learning_rate": 0.0001741921660038428,
+      "loss": 23.0,
+      "step": 3920
+    },
+    {
+      "epoch": 2.3535414165666264,
+      "grad_norm": 0.0018569778185337782,
+      "learning_rate": 0.0001741794984400926,
+      "loss": 23.0,
+      "step": 3921
+    },
+    {
+      "epoch": 2.354141656662665,
+      "grad_norm": 0.0018122204346582294,
+      "learning_rate": 0.00017416682822906694,
+      "loss": 23.0,
+      "step": 3922
+    },
+    {
+      "epoch": 2.3547418967587035,
+      "grad_norm": 0.0012169646797701716,
+      "learning_rate": 0.00017415415537121797,
+      "loss": 23.0,
+      "step": 3923
+    },
+    {
+      "epoch": 2.355342136854742,
+      "grad_norm": 0.0032757974695414305,
+      "learning_rate": 0.00017414147986699792,
+      "loss": 23.0,
+      "step": 3924
+    },
+    {
+      "epoch": 2.35594237695078,
+      "grad_norm": 0.0015983044868335128,
+      "learning_rate": 0.00017412880171685914,
+      "loss": 23.0,
+      "step": 3925
+    },
+    {
+      "epoch": 2.3565426170468187,
+      "grad_norm": 0.000568926683627069,
+      "learning_rate": 0.00017411612092125417,
+      "loss": 23.0,
+      "step": 3926
+    },
+    {
+      "epoch": 2.357142857142857,
+      "grad_norm": 0.0018514172406867146,
+      "learning_rate": 0.00017410343748063545,
+      "loss": 23.0,
+      "step": 3927
+    },
+    {
+      "epoch": 2.3577430972388957,
+      "grad_norm": 0.0015337765216827393,
+      "learning_rate": 0.00017409075139545568,
+      "loss": 23.0,
+      "step": 3928
+    },
+    {
+      "epoch": 2.358343337334934,
+      "grad_norm": 0.0009689840371720493,
+      "learning_rate": 0.00017407806266616754,
+      "loss": 23.0,
+      "step": 3929
+    },
+    {
+      "epoch": 2.3589435774309724,
+      "grad_norm": 0.002194127067923546,
+      "learning_rate": 0.00017406537129322394,
+      "loss": 23.0,
+      "step": 3930
+    },
+    {
+      "epoch": 2.359543817527011,
+      "grad_norm": 0.0015414162771776319,
+      "learning_rate": 0.00017405267727707774,
+      "loss": 23.0,
+      "step": 3931
+    },
+    {
+      "epoch": 2.3601440576230495,
+      "grad_norm": 0.00167924037668854,
+      "learning_rate": 0.000174039980618182,
+      "loss": 23.0,
+      "step": 3932
+    },
+    {
+      "epoch": 2.3607442977190876,
+      "grad_norm": 0.001501681748777628,
+      "learning_rate": 0.00017402728131698977,
+      "loss": 23.0,
+      "step": 3933
+    },
+    {
+      "epoch": 2.361344537815126,
+      "grad_norm": 0.0010487270774319768,
+      "learning_rate": 0.0001740145793739543,
+      "loss": 23.0,
+      "step": 3934
+    },
+    {
+      "epoch": 2.3619447779111646,
+      "grad_norm": 0.0017754529835656285,
+      "learning_rate": 0.0001740018747895289,
+      "loss": 23.0,
+      "step": 3935
+    },
+    {
+      "epoch": 2.3625450180072027,
+      "grad_norm": 0.00277283089235425,
+      "learning_rate": 0.0001739891675641669,
+      "loss": 23.0,
+      "step": 3936
+    },
+    {
+      "epoch": 2.3631452581032413,
+      "grad_norm": 0.0016696704551577568,
+      "learning_rate": 0.00017397645769832187,
+      "loss": 23.0,
+      "step": 3937
+    },
+    {
+      "epoch": 2.36374549819928,
+      "grad_norm": 0.0018485120963305235,
+      "learning_rate": 0.00017396374519244737,
+      "loss": 23.0,
+      "step": 3938
+    },
+    {
+      "epoch": 2.3643457382953184,
+      "grad_norm": 0.0054781329818069935,
+      "learning_rate": 0.000173951030046997,
+      "loss": 23.0,
+      "step": 3939
+    },
+    {
+      "epoch": 2.3649459783913565,
+      "grad_norm": 0.0019205682910978794,
+      "learning_rate": 0.00017393831226242466,
+      "loss": 23.0,
+      "step": 3940
+    },
+    {
+      "epoch": 2.365546218487395,
+      "grad_norm": 0.003620882984250784,
+      "learning_rate": 0.0001739255918391841,
+      "loss": 23.0,
+      "step": 3941
+    },
+    {
+      "epoch": 2.3661464585834335,
+      "grad_norm": 0.0012869867496192455,
+      "learning_rate": 0.00017391286877772933,
+      "loss": 23.0,
+      "step": 3942
+    },
+    {
+      "epoch": 2.3667466986794716,
+      "grad_norm": 0.002010358963161707,
+      "learning_rate": 0.00017390014307851442,
+      "loss": 23.0,
+      "step": 3943
+    },
+    {
+      "epoch": 2.36734693877551,
+      "grad_norm": 0.0021779662929475307,
+      "learning_rate": 0.0001738874147419935,
+      "loss": 23.0,
+      "step": 3944
+    },
+    {
+      "epoch": 2.3679471788715487,
+      "grad_norm": 0.0021605216898024082,
+      "learning_rate": 0.0001738746837686208,
+      "loss": 23.0,
+      "step": 3945
+    },
+    {
+      "epoch": 2.3685474189675872,
+      "grad_norm": 0.0006803186261095107,
+      "learning_rate": 0.00017386195015885064,
+      "loss": 23.0,
+      "step": 3946
+    },
+    {
+      "epoch": 2.3691476590636253,
+      "grad_norm": 0.0011268210364505649,
+      "learning_rate": 0.0001738492139131375,
+      "loss": 23.0,
+      "step": 3947
+    },
+    {
+      "epoch": 2.369747899159664,
+      "grad_norm": 0.0007069764542393386,
+      "learning_rate": 0.0001738364750319359,
+      "loss": 23.0,
+      "step": 3948
+    },
+    {
+      "epoch": 2.3703481392557024,
+      "grad_norm": 0.007430940866470337,
+      "learning_rate": 0.0001738237335157004,
+      "loss": 23.0,
+      "step": 3949
+    },
+    {
+      "epoch": 2.3709483793517405,
+      "grad_norm": 0.003919681068509817,
+      "learning_rate": 0.00017381098936488574,
+      "loss": 23.0,
+      "step": 3950
+    },
+    {
+      "epoch": 2.371548619447779,
+      "grad_norm": 0.0008578869164921343,
+      "learning_rate": 0.0001737982425799468,
+      "loss": 23.0,
+      "step": 3951
+    },
+    {
+      "epoch": 2.3721488595438176,
+      "grad_norm": 0.0018030635546892881,
+      "learning_rate": 0.00017378549316133835,
+      "loss": 23.0,
+      "step": 3952
+    },
+    {
+      "epoch": 2.372749099639856,
+      "grad_norm": 0.0018941910238936543,
+      "learning_rate": 0.00017377274110951543,
+      "loss": 23.0,
+      "step": 3953
+    },
+    {
+      "epoch": 2.3733493397358942,
+      "grad_norm": 0.0007153527112677693,
+      "learning_rate": 0.00017375998642493322,
+      "loss": 23.0,
+      "step": 3954
+    },
+    {
+      "epoch": 2.3739495798319328,
+      "grad_norm": 0.0009323756094090641,
+      "learning_rate": 0.00017374722910804677,
+      "loss": 23.0,
+      "step": 3955
+    },
+    {
+      "epoch": 2.3745498199279713,
+      "grad_norm": 0.0036982784513384104,
+      "learning_rate": 0.00017373446915931143,
+      "loss": 23.0,
+      "step": 3956
+    },
+    {
+      "epoch": 2.3751500600240094,
+      "grad_norm": 0.0014185566687956452,
+      "learning_rate": 0.00017372170657918256,
+      "loss": 23.0,
+      "step": 3957
+    },
+    {
+      "epoch": 2.375750300120048,
+      "grad_norm": 0.002080840989947319,
+      "learning_rate": 0.00017370894136811563,
+      "loss": 23.0,
+      "step": 3958
+    },
+    {
+      "epoch": 2.3763505402160865,
+      "grad_norm": 0.0006996280862949789,
+      "learning_rate": 0.00017369617352656618,
+      "loss": 23.0,
+      "step": 3959
+    },
+    {
+      "epoch": 2.376950780312125,
+      "grad_norm": 0.0018821045523509383,
+      "learning_rate": 0.00017368340305498985,
+      "loss": 23.0,
+      "step": 3960
+    },
+    {
+      "epoch": 2.377551020408163,
+      "grad_norm": 0.0013187207514420152,
+      "learning_rate": 0.0001736706299538424,
+      "loss": 23.0,
+      "step": 3961
+    },
+    {
+      "epoch": 2.3781512605042017,
+      "grad_norm": 0.001449449686333537,
+      "learning_rate": 0.0001736578542235797,
+      "loss": 23.0,
+      "step": 3962
+    },
+    {
+      "epoch": 2.37875150060024,
+      "grad_norm": 0.002266083611175418,
+      "learning_rate": 0.00017364507586465762,
+      "loss": 23.0,
+      "step": 3963
+    },
+    {
+      "epoch": 2.3793517406962783,
+      "grad_norm": 0.002582366345450282,
+      "learning_rate": 0.00017363229487753223,
+      "loss": 23.0,
+      "step": 3964
+    },
+    {
+      "epoch": 2.379951980792317,
+      "grad_norm": 0.0003937427536584437,
+      "learning_rate": 0.00017361951126265964,
+      "loss": 23.0,
+      "step": 3965
+    },
+    {
+      "epoch": 2.3805522208883554,
+      "grad_norm": 0.0017639965517446399,
+      "learning_rate": 0.00017360672502049606,
+      "loss": 23.0,
+      "step": 3966
+    },
+    {
+      "epoch": 2.381152460984394,
+      "grad_norm": 0.004503778647631407,
+      "learning_rate": 0.0001735939361514978,
+      "loss": 23.0,
+      "step": 3967
+    },
+    {
+      "epoch": 2.381752701080432,
+      "grad_norm": 0.0036358425859361887,
+      "learning_rate": 0.0001735811446561213,
+      "loss": 23.0,
+      "step": 3968
+    },
+    {
+      "epoch": 2.3823529411764706,
+      "grad_norm": 0.0023689253721386194,
+      "learning_rate": 0.00017356835053482299,
+      "loss": 23.0,
+      "step": 3969
+    },
+    {
+      "epoch": 2.382953181272509,
+      "grad_norm": 0.002466338686645031,
+      "learning_rate": 0.0001735555537880595,
+      "loss": 23.0,
+      "step": 3970
+    },
+    {
+      "epoch": 2.383553421368547,
+      "grad_norm": 0.0021039212588220835,
+      "learning_rate": 0.0001735427544162875,
+      "loss": 23.0,
+      "step": 3971
+    },
+    {
+      "epoch": 2.3841536614645857,
+      "grad_norm": 0.0027853043284267187,
+      "learning_rate": 0.00017352995241996374,
+      "loss": 23.0,
+      "step": 3972
+    },
+    {
+      "epoch": 2.3847539015606243,
+      "grad_norm": 0.0033636244479566813,
+      "learning_rate": 0.00017351714779954516,
+      "loss": 23.0,
+      "step": 3973
+    },
+    {
+      "epoch": 2.385354141656663,
+      "grad_norm": 0.002105874475091696,
+      "learning_rate": 0.00017350434055548867,
+      "loss": 23.0,
+      "step": 3974
+    },
+    {
+      "epoch": 2.385954381752701,
+      "grad_norm": 0.0009028427302837372,
+      "learning_rate": 0.00017349153068825133,
+      "loss": 23.0,
+      "step": 3975
+    },
+    {
+      "epoch": 2.3865546218487395,
+      "grad_norm": 0.0030058857519179583,
+      "learning_rate": 0.00017347871819829033,
+      "loss": 23.0,
+      "step": 3976
+    },
+    {
+      "epoch": 2.387154861944778,
+      "grad_norm": 0.0017001151572912931,
+      "learning_rate": 0.00017346590308606286,
+      "loss": 23.0,
+      "step": 3977
+    },
+    {
+      "epoch": 2.387755102040816,
+      "grad_norm": 0.0008253180421888828,
+      "learning_rate": 0.0001734530853520263,
+      "loss": 23.0,
+      "step": 3978
+    },
+    {
+      "epoch": 2.3883553421368546,
+      "grad_norm": 0.0019861324690282345,
+      "learning_rate": 0.00017344026499663807,
+      "loss": 23.0,
+      "step": 3979
+    },
+    {
+      "epoch": 2.388955582232893,
+      "grad_norm": 0.002510624472051859,
+      "learning_rate": 0.0001734274420203557,
+      "loss": 23.0,
+      "step": 3980
+    },
+    {
+      "epoch": 2.3895558223289317,
+      "grad_norm": 0.0011581958970054984,
+      "learning_rate": 0.00017341461642363676,
+      "loss": 23.0,
+      "step": 3981
+    },
+    {
+      "epoch": 2.39015606242497,
+      "grad_norm": 0.002012391574680805,
+      "learning_rate": 0.00017340178820693906,
+      "loss": 23.0,
+      "step": 3982
+    },
+    {
+      "epoch": 2.3907563025210083,
+      "grad_norm": 0.002154824323952198,
+      "learning_rate": 0.0001733889573707203,
+      "loss": 23.0,
+      "step": 3983
+    },
+    {
+      "epoch": 2.391356542617047,
+      "grad_norm": 0.0009000212885439396,
+      "learning_rate": 0.00017337612391543847,
+      "loss": 23.0,
+      "step": 3984
+    },
+    {
+      "epoch": 2.3919567827130854,
+      "grad_norm": 0.003189866431057453,
+      "learning_rate": 0.0001733632878415515,
+      "loss": 23.0,
+      "step": 3985
+    },
+    {
+      "epoch": 2.3925570228091235,
+      "grad_norm": 0.0023710643872618675,
+      "learning_rate": 0.00017335044914951753,
+      "loss": 23.0,
+      "step": 3986
+    },
+    {
+      "epoch": 2.393157262905162,
+      "grad_norm": 0.0008908773888833821,
+      "learning_rate": 0.00017333760783979466,
+      "loss": 23.0,
+      "step": 3987
+    },
+    {
+      "epoch": 2.3937575030012006,
+      "grad_norm": 0.0027625104412436485,
+      "learning_rate": 0.00017332476391284128,
+      "loss": 23.0,
+      "step": 3988
+    },
+    {
+      "epoch": 2.394357743097239,
+      "grad_norm": 0.002077084966003895,
+      "learning_rate": 0.00017331191736911564,
+      "loss": 23.0,
+      "step": 3989
+    },
+    {
+      "epoch": 2.3949579831932772,
+      "grad_norm": 0.0020401370711624622,
+      "learning_rate": 0.00017329906820907627,
+      "loss": 23.0,
+      "step": 3990
+    },
+    {
+      "epoch": 2.395558223289316,
+      "grad_norm": 0.00204063905403018,
+      "learning_rate": 0.0001732862164331817,
+      "loss": 23.0,
+      "step": 3991
+    },
+    {
+      "epoch": 2.3961584633853543,
+      "grad_norm": 0.0018513710238039494,
+      "learning_rate": 0.00017327336204189055,
+      "loss": 23.0,
+      "step": 3992
+    },
+    {
+      "epoch": 2.3967587034813924,
+      "grad_norm": 0.0022755179088562727,
+      "learning_rate": 0.00017326050503566162,
+      "loss": 23.0,
+      "step": 3993
+    },
+    {
+      "epoch": 2.397358943577431,
+      "grad_norm": 0.0027569227386265993,
+      "learning_rate": 0.00017324764541495373,
+      "loss": 23.0,
+      "step": 3994
+    },
+    {
+      "epoch": 2.3979591836734695,
+      "grad_norm": 0.0031000536400824785,
+      "learning_rate": 0.00017323478318022578,
+      "loss": 23.0,
+      "step": 3995
+    },
+    {
+      "epoch": 2.398559423769508,
+      "grad_norm": 0.0016642792616039515,
+      "learning_rate": 0.00017322191833193674,
+      "loss": 23.0,
+      "step": 3996
+    },
+    {
+      "epoch": 2.399159663865546,
+      "grad_norm": 0.0015426806639879942,
+      "learning_rate": 0.00017320905087054585,
+      "loss": 23.0,
+      "step": 3997
+    },
+    {
+      "epoch": 2.3997599039615847,
+      "grad_norm": 0.0016276773530989885,
+      "learning_rate": 0.00017319618079651226,
+      "loss": 23.0,
+      "step": 3998
+    },
+    {
+      "epoch": 2.400360144057623,
+      "grad_norm": 0.00199799588881433,
+      "learning_rate": 0.0001731833081102952,
+      "loss": 23.0,
+      "step": 3999
+    },
+    {
+      "epoch": 2.4009603841536613,
+      "grad_norm": 0.001323958276771009,
+      "learning_rate": 0.00017317043281235418,
+      "loss": 23.0,
+      "step": 4000
+    },
+    {
+      "epoch": 2.4009603841536613,
+      "eval_loss": 11.5,
+      "eval_runtime": 5.4819,
+      "eval_samples_per_second": 255.931,
+      "eval_steps_per_second": 32.105,
+      "step": 4000
     }
   ],
   "logging_steps": 1,
@@ -21053,7 +28061,7 @@
         "early_stopping_threshold": 0.0
       },
       "attributes": {
-        "early_stopping_patience_counter": 2
+        "early_stopping_patience_counter": 3
       }
     },
     "TrainerControl": {
@@ -21062,12 +28070,12 @@
         "should_evaluate": false,
         "should_log": false,
         "should_save": true,
-        "should_training_stop": false
+        "should_training_stop": true
       },
       "attributes": {}
     }
   },
-  "total_flos": 60013876285440.0,
+  "total_flos": 80021001510912.0,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null