diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,7849 +10,7849 @@
   "log_history": [
     {
       "epoch": 0.0001786831055123738,
-      "grad_norm": 0.6767252087593079,
+      "grad_norm": 0.677041232585907,
       "learning_rate": 3.5714285714285716e-07,
       "loss": 2.5858,
       "step": 1
     },
     {
       "epoch": 0.000893415527561869,
-      "grad_norm": 0.8736157417297363,
+      "grad_norm": 0.87285977602005,
       "learning_rate": 1.7857142857142857e-06,
-      "loss": 2.8563,
+      "loss": 2.8564,
       "step": 5
     },
     {
       "epoch": 0.001786831055123738,
-      "grad_norm": 0.9124715328216553,
+      "grad_norm": 0.91197270154953,
       "learning_rate": 3.5714285714285714e-06,
-      "loss": 2.7741,
+      "loss": 2.7743,
       "step": 10
     },
     {
       "epoch": 0.002680246582685607,
-      "grad_norm": 0.8725903034210205,
+      "grad_norm": 0.8713774681091309,
       "learning_rate": 5.357142857142857e-06,
-      "loss": 2.7188,
+      "loss": 2.7189,
       "step": 15
     },
     {
       "epoch": 0.003573662110247476,
-      "grad_norm": 0.7513923645019531,
+      "grad_norm": 0.7496689558029175,
       "learning_rate": 7.142857142857143e-06,
-      "loss": 2.7556,
+      "loss": 2.7557,
       "step": 20
     },
     {
       "epoch": 0.004467077637809345,
-      "grad_norm": 0.7444078326225281,
+      "grad_norm": 0.7549262642860413,
       "learning_rate": 8.92857142857143e-06,
       "loss": 2.7256,
       "step": 25
     },
     {
       "epoch": 0.005360493165371214,
-      "grad_norm": 0.8215367794036865,
+      "grad_norm": 0.8223145008087158,
       "learning_rate": 1.0714285714285714e-05,
       "loss": 2.7319,
       "step": 30
     },
     {
       "epoch": 0.006253908692933083,
-      "grad_norm": 0.7591757774353027,
+      "grad_norm": 0.7578197121620178,
       "learning_rate": 1.25e-05,
-      "loss": 2.6611,
+      "loss": 2.661,
       "step": 35
     },
     {
       "epoch": 0.007147324220494952,
-      "grad_norm": 0.6609904766082764,
+      "grad_norm": 0.6663674712181091,
       "learning_rate": 1.4285714285714285e-05,
       "loss": 2.6396,
       "step": 40
     },
     {
       "epoch": 0.008040739748056821,
-      "grad_norm": 0.523094117641449,
+      "grad_norm": 0.5226909518241882,
       "learning_rate": 1.6071428571428572e-05,
       "loss": 2.5739,
       "step": 45
     },
     {
       "epoch": 0.00893415527561869,
-      "grad_norm": 0.4964945614337921,
+      "grad_norm": 0.4958447515964508,
       "learning_rate": 1.785714285714286e-05,
-      "loss": 2.4766,
+      "loss": 2.4765,
       "step": 50
     },
     {
       "epoch": 0.00982757080318056,
-      "grad_norm": 0.4061848819255829,
+      "grad_norm": 0.40554603934288025,
       "learning_rate": 1.9642857142857145e-05,
-      "loss": 2.4273,
+      "loss": 2.4275,
       "step": 55
     },
     {
       "epoch": 0.010720986330742428,
-      "grad_norm": 0.438088059425354,
+      "grad_norm": 0.43960413336753845,
       "learning_rate": 2.1428571428571428e-05,
-      "loss": 2.3819,
+      "loss": 2.3822,
       "step": 60
     },
     {
       "epoch": 0.011614401858304297,
-      "grad_norm": 0.406583696603775,
+      "grad_norm": 0.40701353549957275,
       "learning_rate": 2.3214285714285715e-05,
       "loss": 2.3326,
       "step": 65
     },
     {
       "epoch": 0.012507817385866166,
-      "grad_norm": 0.38678309321403503,
+      "grad_norm": 0.3899136483669281,
       "learning_rate": 2.5e-05,
-      "loss": 2.3171,
+      "loss": 2.3172,
       "step": 70
     },
     {
       "epoch": 0.013401232913428035,
-      "grad_norm": 0.4419669210910797,
+      "grad_norm": 0.444366455078125,
       "learning_rate": 2.6785714285714288e-05,
-      "loss": 2.2798,
+      "loss": 2.2801,
       "step": 75
     },
     {
       "epoch": 0.014294648440989904,
-      "grad_norm": 0.43436020612716675,
+      "grad_norm": 0.43577924370765686,
       "learning_rate": 2.857142857142857e-05,
       "loss": 2.23,
       "step": 80
     },
     {
       "epoch": 0.015188063968551773,
-      "grad_norm": 0.4241899251937866,
+      "grad_norm": 0.42386895418167114,
       "learning_rate": 3.0357142857142857e-05,
-      "loss": 2.2421,
+      "loss": 2.2422,
       "step": 85
     },
     {
       "epoch": 0.016081479496113642,
-      "grad_norm": 0.4129183888435364,
+      "grad_norm": 0.4131146967411041,
       "learning_rate": 3.2142857142857144e-05,
-      "loss": 2.182,
+      "loss": 2.1822,
       "step": 90
     },
     {
       "epoch": 0.016974895023675513,
-      "grad_norm": 0.429864764213562,
+      "grad_norm": 0.43191808462142944,
       "learning_rate": 3.392857142857143e-05,
-      "loss": 2.1112,
+      "loss": 2.1113,
       "step": 95
     },
     {
       "epoch": 0.01786831055123738,
-      "grad_norm": 0.4998682141304016,
+      "grad_norm": 0.49910837411880493,
       "learning_rate": 3.571428571428572e-05,
-      "loss": 2.1752,
+      "loss": 2.1751,
       "step": 100
     },
     {
       "epoch": 0.01876172607879925,
-      "grad_norm": 0.4673575162887573,
+      "grad_norm": 0.4678761065006256,
       "learning_rate": 3.7500000000000003e-05,
-      "loss": 2.1021,
+      "loss": 2.1023,
       "step": 105
     },
     {
       "epoch": 0.01965514160636112,
-      "grad_norm": 0.5105161070823669,
+      "grad_norm": 0.5106930136680603,
       "learning_rate": 3.928571428571429e-05,
-      "loss": 2.075,
+      "loss": 2.0749,
       "step": 110
     },
     {
       "epoch": 0.02054855713392299,
-      "grad_norm": 0.5451802015304565,
+      "grad_norm": 0.5173302292823792,
       "learning_rate": 4.107142857142857e-05,
-      "loss": 2.0945,
+      "loss": 2.0946,
       "step": 115
     },
     {
       "epoch": 0.021441972661484857,
-      "grad_norm": 0.5659247040748596,
+      "grad_norm": 0.5672593712806702,
       "learning_rate": 4.2857142857142856e-05,
-      "loss": 2.0086,
+      "loss": 2.0089,
       "step": 120
     },
     {
       "epoch": 0.022335388189046727,
-      "grad_norm": 0.5502987504005432,
+      "grad_norm": 0.5483328700065613,
       "learning_rate": 4.464285714285715e-05,
-      "loss": 2.0306,
+      "loss": 2.0303,
       "step": 125
     },
     {
       "epoch": 0.023228803716608595,
-      "grad_norm": 0.5537978410720825,
+      "grad_norm": 0.5528935790061951,
       "learning_rate": 4.642857142857143e-05,
-      "loss": 1.995,
+      "loss": 1.9952,
       "step": 130
     },
     {
       "epoch": 0.024122219244170465,
-      "grad_norm": 0.6560796499252319,
+      "grad_norm": 0.6560689210891724,
       "learning_rate": 4.8214285714285716e-05,
       "loss": 1.9854,
       "step": 135
     },
     {
       "epoch": 0.025015634771732333,
-      "grad_norm": 0.6709268689155579,
+      "grad_norm": 0.6699485182762146,
       "learning_rate": 5e-05,
-      "loss": 2.0174,
+      "loss": 2.0175,
       "step": 140
     },
     {
       "epoch": 0.025909050299294203,
-      "grad_norm": 0.6651104092597961,
+      "grad_norm": 0.6667202711105347,
       "learning_rate": 5.1785714285714296e-05,
-      "loss": 1.9972,
+      "loss": 1.9973,
       "step": 145
     },
     {
       "epoch": 0.02680246582685607,
-      "grad_norm": 0.6255117058753967,
+      "grad_norm": 0.624862790107727,
       "learning_rate": 5.3571428571428575e-05,
-      "loss": 1.9597,
+      "loss": 1.9596,
       "step": 150
     },
     {
       "epoch": 0.02769588135441794,
-      "grad_norm": 0.7806227207183838,
+      "grad_norm": 0.7823134064674377,
       "learning_rate": 5.535714285714286e-05,
       "loss": 1.9863,
       "step": 155
     },
     {
       "epoch": 0.02858929688197981,
-      "grad_norm": 0.7957138419151306,
+      "grad_norm": 0.7916033864021301,
       "learning_rate": 5.714285714285714e-05,
-      "loss": 1.965,
+      "loss": 1.9652,
       "step": 160
     },
     {
       "epoch": 0.02948271240954168,
-      "grad_norm": 0.8026310205459595,
+      "grad_norm": 0.8077523708343506,
       "learning_rate": 5.8928571428571435e-05,
-      "loss": 1.9605,
+      "loss": 1.961,
       "step": 165
     },
     {
       "epoch": 0.030376127937103547,
-      "grad_norm": 0.724997878074646,
+      "grad_norm": 0.7293629050254822,
       "learning_rate": 6.0714285714285715e-05,
-      "loss": 1.932,
+      "loss": 1.9324,
       "step": 170
     },
     {
       "epoch": 0.031269543464665414,
-      "grad_norm": 0.7658754587173462,
+      "grad_norm": 0.7660003304481506,
       "learning_rate": 6.25e-05,
-      "loss": 1.8975,
+      "loss": 1.8978,
       "step": 175
     },
     {
       "epoch": 0.032162958992227285,
-      "grad_norm": 0.7666235566139221,
+      "grad_norm": 0.7684458494186401,
       "learning_rate": 6.428571428571429e-05,
-      "loss": 1.888,
+      "loss": 1.8884,
       "step": 180
     },
     {
       "epoch": 0.033056374519789156,
-      "grad_norm": 0.8629304766654968,
+      "grad_norm": 0.8633774518966675,
       "learning_rate": 6.607142857142857e-05,
-      "loss": 1.8824,
+      "loss": 1.8827,
       "step": 185
     },
     {
       "epoch": 0.033949790047351026,
-      "grad_norm": 0.912053108215332,
+      "grad_norm": 0.918566882610321,
       "learning_rate": 6.785714285714286e-05,
-      "loss": 1.8766,
+      "loss": 1.8765,
       "step": 190
     },
     {
       "epoch": 0.03484320557491289,
-      "grad_norm": 0.8080089688301086,
+      "grad_norm": 0.8094245791435242,
       "learning_rate": 6.964285714285715e-05,
       "loss": 1.8944,
       "step": 195
     },
     {
       "epoch": 0.03573662110247476,
-      "grad_norm": 0.8864442110061646,
+      "grad_norm": 0.8820126056671143,
       "learning_rate": 7.142857142857143e-05,
-      "loss": 1.8932,
+      "loss": 1.8934,
       "step": 200
     },
     {
       "epoch": 0.03663003663003663,
-      "grad_norm": 0.9291841387748718,
+      "grad_norm": 0.9279444813728333,
       "learning_rate": 7.321428571428571e-05,
-      "loss": 1.86,
+      "loss": 1.8604,
       "step": 205
     },
     {
       "epoch": 0.0375234521575985,
-      "grad_norm": 0.8211831450462341,
+      "grad_norm": 0.8216668963432312,
       "learning_rate": 7.500000000000001e-05,
-      "loss": 1.8467,
+      "loss": 1.8466,
       "step": 210
     },
     {
       "epoch": 0.038416867685160366,
-      "grad_norm": 0.9697086215019226,
+      "grad_norm": 0.9710394740104675,
       "learning_rate": 7.67857142857143e-05,
       "loss": 1.8802,
       "step": 215
     },
     {
       "epoch": 0.03931028321272224,
-      "grad_norm": 1.0050380229949951,
+      "grad_norm": 1.00575852394104,
       "learning_rate": 7.857142857142858e-05,
-      "loss": 1.8544,
+      "loss": 1.8543,
       "step": 220
     },
     {
       "epoch": 0.04020369874028411,
-      "grad_norm": 0.9131980538368225,
+      "grad_norm": 0.9068517088890076,
       "learning_rate": 8.035714285714287e-05,
-      "loss": 1.8207,
+      "loss": 1.8208,
       "step": 225
     },
     {
       "epoch": 0.04109711426784598,
-      "grad_norm": 0.9117507338523865,
+      "grad_norm": 0.9185478091239929,
       "learning_rate": 8.214285714285714e-05,
-      "loss": 1.8391,
+      "loss": 1.8395,
       "step": 230
     },
     {
       "epoch": 0.04199052979540784,
-      "grad_norm": 0.8625417351722717,
+      "grad_norm": 0.8614683747291565,
       "learning_rate": 8.392857142857144e-05,
       "loss": 1.7931,
       "step": 235
     },
     {
       "epoch": 0.04288394532296971,
-      "grad_norm": 1.025675654411316,
+      "grad_norm": 1.0321149826049805,
       "learning_rate": 8.571428571428571e-05,
-      "loss": 1.8135,
+      "loss": 1.8134,
       "step": 240
     },
     {
       "epoch": 0.043777360850531584,
-      "grad_norm": 0.9845486879348755,
+      "grad_norm": 0.997043788433075,
       "learning_rate": 8.75e-05,
-      "loss": 1.8192,
+      "loss": 1.8195,
       "step": 245
     },
     {
       "epoch": 0.044670776378093455,
-      "grad_norm": 0.9469249844551086,
+      "grad_norm": 0.9486415386199951,
       "learning_rate": 8.92857142857143e-05,
-      "loss": 1.8107,
+      "loss": 1.811,
       "step": 250
     },
     {
       "epoch": 0.04556419190565532,
-      "grad_norm": 0.8901212215423584,
+      "grad_norm": 0.8874170780181885,
       "learning_rate": 9.107142857142857e-05,
-      "loss": 1.8072,
+      "loss": 1.808,
       "step": 255
     },
     {
       "epoch": 0.04645760743321719,
-      "grad_norm": 0.9586057066917419,
+      "grad_norm": 0.9631446003913879,
       "learning_rate": 9.285714285714286e-05,
-      "loss": 1.7973,
+      "loss": 1.7976,
       "step": 260
     },
     {
       "epoch": 0.04735102296077906,
-      "grad_norm": 0.9514407515525818,
+      "grad_norm": 0.9562223553657532,
       "learning_rate": 9.464285714285715e-05,
-      "loss": 1.7815,
+      "loss": 1.7817,
       "step": 265
     },
     {
       "epoch": 0.04824443848834093,
-      "grad_norm": 0.8750002980232239,
+      "grad_norm": 0.8817927241325378,
       "learning_rate": 9.642857142857143e-05,
-      "loss": 1.765,
+      "loss": 1.7646,
       "step": 270
     },
     {
       "epoch": 0.049137854015902795,
-      "grad_norm": 0.966252863407135,
+      "grad_norm": 0.9726014137268066,
       "learning_rate": 9.821428571428572e-05,
       "loss": 1.8197,
       "step": 275
     },
     {
       "epoch": 0.050031269543464665,
-      "grad_norm": 0.9129286408424377,
+      "grad_norm": 0.914252758026123,
       "learning_rate": 0.0001,
-      "loss": 1.7548,
+      "loss": 1.7556,
       "step": 280
     },
     {
       "epoch": 0.050924685071026536,
-      "grad_norm": 0.9409673810005188,
+      "grad_norm": 0.9449843764305115,
       "learning_rate": 0.00010178571428571428,
-      "loss": 1.7109,
+      "loss": 1.7112,
       "step": 285
     },
     {
       "epoch": 0.05181810059858841,
-      "grad_norm": 0.9348879456520081,
+      "grad_norm": 0.9346893429756165,
       "learning_rate": 0.00010357142857142859,
-      "loss": 1.7618,
+      "loss": 1.7619,
       "step": 290
     },
     {
       "epoch": 0.05271151612615027,
-      "grad_norm": 0.8221137523651123,
+      "grad_norm": 0.8253093957901001,
       "learning_rate": 0.00010535714285714286,
-      "loss": 1.734,
+      "loss": 1.7342,
       "step": 295
     },
     {
       "epoch": 0.05360493165371214,
-      "grad_norm": 0.8188982009887695,
+      "grad_norm": 0.8210521936416626,
       "learning_rate": 0.00010714285714285715,
-      "loss": 1.7522,
+      "loss": 1.7528,
       "step": 300
     },
     {
       "epoch": 0.05449834718127401,
-      "grad_norm": 0.7834224700927734,
+      "grad_norm": 0.7791708707809448,
       "learning_rate": 0.00010892857142857142,
-      "loss": 1.7878,
+      "loss": 1.7881,
       "step": 305
     },
     {
       "epoch": 0.05539176270883588,
-      "grad_norm": 0.9102511405944824,
+      "grad_norm": 0.9075311422348022,
       "learning_rate": 0.00011071428571428572,
-      "loss": 1.7572,
+      "loss": 1.7569,
       "step": 310
     },
     {
       "epoch": 0.05628517823639775,
-      "grad_norm": 0.8769788146018982,
+      "grad_norm": 0.8817685842514038,
       "learning_rate": 0.00011250000000000001,
-      "loss": 1.7443,
+      "loss": 1.7444,
       "step": 315
     },
     {
       "epoch": 0.05717859376395962,
-      "grad_norm": 0.9550459980964661,
+      "grad_norm": 0.9592775106430054,
       "learning_rate": 0.00011428571428571428,
-      "loss": 1.7346,
+      "loss": 1.7344,
       "step": 320
     },
     {
       "epoch": 0.05807200929152149,
-      "grad_norm": 0.8972039222717285,
+      "grad_norm": 0.8894655704498291,
       "learning_rate": 0.00011607142857142858,
-      "loss": 1.7377,
+      "loss": 1.738,
       "step": 325
     },
     {
       "epoch": 0.05896542481908336,
-      "grad_norm": 0.9489789009094238,
+      "grad_norm": 0.9590314030647278,
       "learning_rate": 0.00011785714285714287,
-      "loss": 1.7335,
+      "loss": 1.7336,
       "step": 330
     },
     {
       "epoch": 0.05985884034664522,
-      "grad_norm": 0.9783198833465576,
+      "grad_norm": 0.9731721878051758,
       "learning_rate": 0.00011964285714285714,
-      "loss": 1.7243,
+      "loss": 1.7242,
       "step": 335
     },
     {
       "epoch": 0.060752255874207094,
-      "grad_norm": 1.0081501007080078,
+      "grad_norm": 1.0074090957641602,
       "learning_rate": 0.00012142857142857143,
-      "loss": 1.7717,
+      "loss": 1.7719,
       "step": 340
     },
     {
       "epoch": 0.061645671401768964,
-      "grad_norm": 0.9436238408088684,
+      "grad_norm": 0.9376741051673889,
       "learning_rate": 0.00012321428571428572,
-      "loss": 1.7301,
+      "loss": 1.7299,
       "step": 345
     },
     {
       "epoch": 0.06253908692933083,
-      "grad_norm": 0.8638908863067627,
+      "grad_norm": 0.8597344756126404,
       "learning_rate": 0.000125,
-      "loss": 1.738,
+      "loss": 1.7378,
       "step": 350
     },
     {
       "epoch": 0.0634325024568927,
-      "grad_norm": 0.8834109306335449,
+      "grad_norm": 0.8836060762405396,
       "learning_rate": 0.0001267857142857143,
       "loss": 1.7379,
       "step": 355
     },
     {
       "epoch": 0.06432591798445457,
-      "grad_norm": 1.045885682106018,
+      "grad_norm": 1.0358092784881592,
       "learning_rate": 0.00012857142857142858,
       "loss": 1.7175,
       "step": 360
     },
     {
       "epoch": 0.06521933351201643,
-      "grad_norm": 0.8442100286483765,
+      "grad_norm": 0.8426574468612671,
       "learning_rate": 0.00013035714285714286,
-      "loss": 1.6629,
+      "loss": 1.6632,
       "step": 365
     },
     {
       "epoch": 0.06611274903957831,
-      "grad_norm": 0.948798656463623,
+      "grad_norm": 0.9615597128868103,
       "learning_rate": 0.00013214285714285715,
-      "loss": 1.6661,
+      "loss": 1.6667,
       "step": 370
     },
     {
       "epoch": 0.06700616456714018,
-      "grad_norm": 0.9104481935501099,
+      "grad_norm": 0.9134716391563416,
       "learning_rate": 0.00013392857142857144,
-      "loss": 1.7239,
+      "loss": 1.7246,
       "step": 375
     },
     {
       "epoch": 0.06789958009470205,
-      "grad_norm": 0.8660710453987122,
+      "grad_norm": 0.8629338145256042,
       "learning_rate": 0.00013571428571428572,
       "loss": 1.6696,
       "step": 380
     },
     {
       "epoch": 0.06879299562226392,
-      "grad_norm": 0.8465679287910461,
+      "grad_norm": 0.8393704295158386,
       "learning_rate": 0.0001375,
       "loss": 1.6838,
       "step": 385
     },
     {
       "epoch": 0.06968641114982578,
-      "grad_norm": 0.9764780402183533,
+      "grad_norm": 0.9799148440361023,
       "learning_rate": 0.0001392857142857143,
-      "loss": 1.7024,
+      "loss": 1.7022,
       "step": 390
     },
     {
       "epoch": 0.07057982667738766,
-      "grad_norm": 0.8730722665786743,
+      "grad_norm": 0.8769751191139221,
       "learning_rate": 0.00014107142857142858,
-      "loss": 1.678,
+      "loss": 1.6783,
       "step": 395
     },
     {
       "epoch": 0.07147324220494952,
-      "grad_norm": 0.8504787087440491,
+      "grad_norm": 0.8460381627082825,
       "learning_rate": 0.00014285714285714287,
-      "loss": 1.6513,
+      "loss": 1.6509,
       "step": 400
     },
     {
       "epoch": 0.07236665773251139,
-      "grad_norm": 0.7620259523391724,
+      "grad_norm": 0.7672830820083618,
       "learning_rate": 0.00014464285714285715,
-      "loss": 1.6478,
+      "loss": 1.6485,
       "step": 405
     },
     {
       "epoch": 0.07326007326007326,
-      "grad_norm": 0.8745117783546448,
+      "grad_norm": 0.8759172558784485,
       "learning_rate": 0.00014642857142857141,
-      "loss": 1.6717,
+      "loss": 1.6723,
       "step": 410
     },
     {
       "epoch": 0.07415348878763513,
-      "grad_norm": 0.8818298578262329,
+      "grad_norm": 0.8765602707862854,
       "learning_rate": 0.00014821428571428573,
-      "loss": 1.5922,
+      "loss": 1.5925,
       "step": 415
     },
     {
       "epoch": 0.075046904315197,
-      "grad_norm": 0.8444076180458069,
+      "grad_norm": 0.8540732264518738,
       "learning_rate": 0.00015000000000000001,
-      "loss": 1.6568,
+      "loss": 1.6583,
       "step": 420
     },
     {
       "epoch": 0.07594031984275887,
-      "grad_norm": 0.7880132794380188,
+      "grad_norm": 0.7945284843444824,
       "learning_rate": 0.00015178571428571427,
-      "loss": 1.6642,
+      "loss": 1.6644,
       "step": 425
     },
     {
       "epoch": 0.07683373537032073,
-      "grad_norm": 1.0402840375900269,
+      "grad_norm": 1.0458513498306274,
       "learning_rate": 0.0001535714285714286,
-      "loss": 1.6942,
+      "loss": 1.6946,
       "step": 430
     },
     {
       "epoch": 0.07772715089788261,
-      "grad_norm": 0.8606407046318054,
+      "grad_norm": 0.8617852330207825,
       "learning_rate": 0.00015535714285714287,
-      "loss": 1.6216,
+      "loss": 1.6218,
       "step": 435
     },
     {
       "epoch": 0.07862056642544447,
-      "grad_norm": 0.8538222312927246,
+      "grad_norm": 0.8525850176811218,
       "learning_rate": 0.00015714285714285716,
-      "loss": 1.6577,
+      "loss": 1.6579,
       "step": 440
     },
     {
       "epoch": 0.07951398195300634,
-      "grad_norm": 0.8017067909240723,
+      "grad_norm": 0.7932422161102295,
       "learning_rate": 0.00015892857142857142,
-      "loss": 1.6588,
+      "loss": 1.6593,
       "step": 445
     },
     {
       "epoch": 0.08040739748056822,
-      "grad_norm": 0.8014382123947144,
+      "grad_norm": 0.795437216758728,
       "learning_rate": 0.00016071428571428573,
       "loss": 1.6661,
       "step": 450
     },
     {
       "epoch": 0.08130081300813008,
-      "grad_norm": 0.8568883538246155,
+      "grad_norm": 0.8602275848388672,
       "learning_rate": 0.00016250000000000002,
-      "loss": 1.6668,
+      "loss": 1.6667,
       "step": 455
     },
     {
       "epoch": 0.08219422853569196,
-      "grad_norm": 0.7566272020339966,
+      "grad_norm": 0.7548096776008606,
       "learning_rate": 0.00016428571428571428,
       "loss": 1.6403,
       "step": 460
     },
     {
       "epoch": 0.08308764406325382,
-      "grad_norm": 0.8573461174964905,
+      "grad_norm": 0.8513688445091248,
       "learning_rate": 0.0001660714285714286,
-      "loss": 1.6548,
+      "loss": 1.6543,
       "step": 465
     },
     {
       "epoch": 0.08398105959081568,
-      "grad_norm": 0.8354015350341797,
+      "grad_norm": 0.836286187171936,
       "learning_rate": 0.00016785714285714288,
-      "loss": 1.6219,
+      "loss": 1.6227,
       "step": 470
     },
     {
       "epoch": 0.08487447511837756,
-      "grad_norm": 0.790770947933197,
+      "grad_norm": 0.7904291749000549,
       "learning_rate": 0.00016964285714285714,
-      "loss": 1.6428,
+      "loss": 1.6433,
       "step": 475
     },
     {
       "epoch": 0.08576789064593943,
-      "grad_norm": 0.7817040085792542,
+      "grad_norm": 0.7850791215896606,
       "learning_rate": 0.00017142857142857143,
-      "loss": 1.6103,
+      "loss": 1.6113,
       "step": 480
     },
     {
       "epoch": 0.08666130617350129,
-      "grad_norm": 0.6936588287353516,
+      "grad_norm": 0.6912005543708801,
       "learning_rate": 0.00017321428571428574,
-      "loss": 1.6043,
+      "loss": 1.6047,
       "step": 485
     },
     {
       "epoch": 0.08755472170106317,
-      "grad_norm": 0.8221834897994995,
+      "grad_norm": 0.8217949867248535,
       "learning_rate": 0.000175,
-      "loss": 1.6187,
+      "loss": 1.6206,
       "step": 490
     },
     {
       "epoch": 0.08844813722862503,
-      "grad_norm": 0.7943176627159119,
+      "grad_norm": 0.7975768446922302,
       "learning_rate": 0.00017678571428571428,
-      "loss": 1.6695,
+      "loss": 1.6711,
       "step": 495
     },
     {
       "epoch": 0.08934155275618691,
-      "grad_norm": 0.7854962944984436,
+      "grad_norm": 0.7934837937355042,
       "learning_rate": 0.0001785714285714286,
-      "loss": 1.6191,
+      "loss": 1.6195,
       "step": 500
     },
     {
       "epoch": 0.09023496828374877,
-      "grad_norm": 0.8033050298690796,
+      "grad_norm": 0.8391872048377991,
       "learning_rate": 0.00018035714285714286,
-      "loss": 1.6066,
+      "loss": 1.6068,
       "step": 505
     },
     {
       "epoch": 0.09112838381131064,
-      "grad_norm": 0.8427201509475708,
+      "grad_norm": 0.8512648344039917,
       "learning_rate": 0.00018214285714285714,
-      "loss": 1.6486,
+      "loss": 1.6489,
       "step": 510
     },
     {
       "epoch": 0.09202179933887251,
-      "grad_norm": 0.7360148429870605,
+      "grad_norm": 0.7523573637008667,
       "learning_rate": 0.00018392857142857143,
-      "loss": 1.6022,
+      "loss": 1.6035,
       "step": 515
     },
     {
       "epoch": 0.09291521486643438,
-      "grad_norm": 0.6847764253616333,
+      "grad_norm": 0.6890606880187988,
       "learning_rate": 0.00018571428571428572,
-      "loss": 1.6158,
+      "loss": 1.6162,
       "step": 520
     },
     {
       "epoch": 0.09380863039399624,
-      "grad_norm": 0.8051613569259644,
+      "grad_norm": 0.8024640679359436,
       "learning_rate": 0.0001875,
-      "loss": 1.5828,
+      "loss": 1.5836,
       "step": 525
     },
     {
       "epoch": 0.09470204592155812,
-      "grad_norm": 0.7814632058143616,
+      "grad_norm": 0.778583824634552,
       "learning_rate": 0.0001892857142857143,
-      "loss": 1.6537,
+      "loss": 1.6539,
       "step": 530
     },
     {
       "epoch": 0.09559546144911998,
-      "grad_norm": 0.6943553686141968,
+      "grad_norm": 0.6993560194969177,
       "learning_rate": 0.00019107142857142858,
-      "loss": 1.5729,
+      "loss": 1.5736,
       "step": 535
     },
     {
       "epoch": 0.09648887697668186,
-      "grad_norm": 0.6946085095405579,
+      "grad_norm": 0.6990346908569336,
       "learning_rate": 0.00019285714285714286,
-      "loss": 1.5977,
+      "loss": 1.5987,
       "step": 540
     },
     {
       "epoch": 0.09738229250424373,
-      "grad_norm": 0.7925629019737244,
+      "grad_norm": 0.7906347513198853,
       "learning_rate": 0.00019464285714285715,
-      "loss": 1.6249,
+      "loss": 1.625,
       "step": 545
     },
     {
       "epoch": 0.09827570803180559,
-      "grad_norm": 0.8013836741447449,
+      "grad_norm": 0.8062006831169128,
       "learning_rate": 0.00019642857142857144,
-      "loss": 1.5448,
+      "loss": 1.5445,
       "step": 550
     },
     {
       "epoch": 0.09916912355936747,
-      "grad_norm": 0.7135029435157776,
+      "grad_norm": 0.7047358751296997,
       "learning_rate": 0.00019821428571428572,
-      "loss": 1.5866,
+      "loss": 1.5868,
       "step": 555
     },
     {
       "epoch": 0.10006253908692933,
-      "grad_norm": 0.7435690760612488,
+      "grad_norm": 0.7451456785202026,
       "learning_rate": 0.0002,
-      "loss": 1.6324,
+      "loss": 1.632,
       "step": 560
     },
     {
       "epoch": 0.1009559546144912,
-      "grad_norm": 0.73707115650177,
+      "grad_norm": 0.7321707606315613,
       "learning_rate": 0.00019999951355027364,
-      "loss": 1.5756,
+      "loss": 1.5762,
       "step": 565
     },
     {
       "epoch": 0.10184937014205307,
-      "grad_norm": 0.7932726144790649,
+      "grad_norm": 0.790035605430603,
       "learning_rate": 0.00019999805420582728,
-      "loss": 1.5724,
+      "loss": 1.5721,
       "step": 570
     },
     {
       "epoch": 0.10274278566961494,
-      "grad_norm": 0.7718573808670044,
+      "grad_norm": 0.7690011858940125,
       "learning_rate": 0.00019999562198085878,
-      "loss": 1.5735,
+      "loss": 1.5736,
       "step": 575
     },
     {
       "epoch": 0.10363620119717681,
-      "grad_norm": 0.7420961856842041,
+      "grad_norm": 0.7484734058380127,
       "learning_rate": 0.00019999221689903133,
-      "loss": 1.5687,
+      "loss": 1.5685,
       "step": 580
     },
     {
       "epoch": 0.10452961672473868,
-      "grad_norm": 0.6695738434791565,
+      "grad_norm": 0.6638728976249695,
       "learning_rate": 0.00019998783899347294,
-      "loss": 1.5685,
+      "loss": 1.5683,
       "step": 585
     },
     {
       "epoch": 0.10542303225230054,
-      "grad_norm": 0.6004559397697449,
+      "grad_norm": 0.6009637117385864,
       "learning_rate": 0.0001999824883067762,
-      "loss": 1.625,
+      "loss": 1.6239,
       "step": 590
     },
     {
       "epoch": 0.10631644777986242,
-      "grad_norm": 0.6196651458740234,
+      "grad_norm": 0.6219035387039185,
       "learning_rate": 0.00019997616489099792,
-      "loss": 1.5445,
+      "loss": 1.5447,
       "step": 595
     },
     {
       "epoch": 0.10720986330742428,
-      "grad_norm": 0.6544579267501831,
+      "grad_norm": 0.6466681957244873,
       "learning_rate": 0.00019996886880765854,
-      "loss": 1.5167,
+      "loss": 1.5165,
       "step": 600
     },
     {
       "epoch": 0.10810327883498615,
-      "grad_norm": 0.7319378852844238,
+      "grad_norm": 0.7301130890846252,
       "learning_rate": 0.0001999606001277417,
-      "loss": 1.5423,
+      "loss": 1.5419,
       "step": 605
     },
     {
       "epoch": 0.10899669436254802,
-      "grad_norm": 0.6340135335922241,
+      "grad_norm": 0.635531485080719,
       "learning_rate": 0.0001999513589316933,
-      "loss": 1.5131,
+      "loss": 1.5127,
       "step": 610
     },
     {
       "epoch": 0.10989010989010989,
-      "grad_norm": 0.719735324382782,
+      "grad_norm": 0.7266018986701965,
       "learning_rate": 0.00019994114530942088,
-      "loss": 1.5784,
+      "loss": 1.5785,
       "step": 615
     },
     {
       "epoch": 0.11078352541767177,
-      "grad_norm": 0.7871695756912231,
+      "grad_norm": 0.7803109288215637,
       "learning_rate": 0.0001999299593602927,
-      "loss": 1.5561,
+      "loss": 1.5557,
       "step": 620
     },
     {
       "epoch": 0.11167694094523363,
-      "grad_norm": 0.7577756643295288,
+      "grad_norm": 0.7546414136886597,
       "learning_rate": 0.00019991780119313682,
-      "loss": 1.5863,
+      "loss": 1.5858,
       "step": 625
     },
     {
       "epoch": 0.1125703564727955,
-      "grad_norm": 0.7437707185745239,
+      "grad_norm": 0.7428862452507019,
       "learning_rate": 0.00019990467092623998,
-      "loss": 1.5244,
+      "loss": 1.5246,
       "step": 630
     },
     {
       "epoch": 0.11346377200035737,
-      "grad_norm": 0.6263313889503479,
+      "grad_norm": 0.6253552436828613,
       "learning_rate": 0.00019989056868734647,
-      "loss": 1.5438,
+      "loss": 1.5436,
       "step": 635
     },
     {
       "epoch": 0.11435718752791924,
-      "grad_norm": 0.6760973930358887,
+      "grad_norm": 0.6777628064155579,
       "learning_rate": 0.0001998754946136569,
-      "loss": 1.5808,
+      "loss": 1.5812,
       "step": 640
     },
     {
       "epoch": 0.1152506030554811,
-      "grad_norm": 0.7239100337028503,
+      "grad_norm": 0.7340919971466064,
       "learning_rate": 0.00019985944885182687,
-      "loss": 1.5129,
+      "loss": 1.5125,
       "step": 645
     },
     {
       "epoch": 0.11614401858304298,
-      "grad_norm": 0.6674967408180237,
+      "grad_norm": 0.6683153510093689,
       "learning_rate": 0.00019984243155796546,
-      "loss": 1.5238,
+      "loss": 1.5239,
       "step": 650
     },
     {
       "epoch": 0.11703743411060484,
-      "grad_norm": 0.6619219779968262,
+      "grad_norm": 0.6631706357002258,
       "learning_rate": 0.00019982444289763388,
-      "loss": 1.4989,
+      "loss": 1.4991,
       "step": 655
     },
     {
       "epoch": 0.11793084963816672,
-      "grad_norm": 0.6065383553504944,
+      "grad_norm": 0.6204883456230164,
       "learning_rate": 0.00019980548304584364,
-      "loss": 1.4831,
+      "loss": 1.4839,
       "step": 660
     },
     {
       "epoch": 0.11882426516572858,
-      "grad_norm": 0.6437244415283203,
+      "grad_norm": 0.6456660032272339,
       "learning_rate": 0.00019978555218705513,
-      "loss": 1.5404,
+      "loss": 1.5401,
       "step": 665
     },
     {
       "epoch": 0.11971768069329045,
-      "grad_norm": 0.6297422647476196,
+      "grad_norm": 0.6324872970581055,
       "learning_rate": 0.00019976465051517548,
-      "loss": 1.5232,
+      "loss": 1.5242,
       "step": 670
     },
     {
       "epoch": 0.12061109622085232,
-      "grad_norm": 0.660554051399231,
+      "grad_norm": 0.6695961952209473,
       "learning_rate": 0.00019974277823355698,
-      "loss": 1.5417,
+      "loss": 1.5427,
       "step": 675
     },
     {
       "epoch": 0.12150451174841419,
-      "grad_norm": 0.6322402358055115,
+      "grad_norm": 0.6281846165657043,
       "learning_rate": 0.00019971993555499494,
-      "loss": 1.5161,
+      "loss": 1.5168,
       "step": 680
     },
     {
       "epoch": 0.12239792727597605,
-      "grad_norm": 0.6935272216796875,
+      "grad_norm": 0.6930037140846252,
       "learning_rate": 0.00019969612270172567,
-      "loss": 1.5089,
+      "loss": 1.5093,
       "step": 685
     },
     {
       "epoch": 0.12329134280353793,
-      "grad_norm": 0.7320016026496887,
+      "grad_norm": 0.7301099300384521,
       "learning_rate": 0.00019967133990542423,
-      "loss": 1.5479,
+      "loss": 1.5472,
       "step": 690
     },
     {
       "epoch": 0.12418475833109979,
-      "grad_norm": 0.6796984672546387,
+      "grad_norm": 0.6746507287025452,
       "learning_rate": 0.0001996455874072024,
-      "loss": 1.5242,
+      "loss": 1.5237,
       "step": 695
     },
     {
       "epoch": 0.12507817385866166,
-      "grad_norm": 0.6992778778076172,
+      "grad_norm": 0.7017915844917297,
       "learning_rate": 0.00019961886545760598,
-      "loss": 1.5287,
+      "loss": 1.5288,
       "step": 700
     },
     {
       "epoch": 0.12597158938622352,
-      "grad_norm": 0.6733859777450562,
+      "grad_norm": 0.6702309846878052,
       "learning_rate": 0.00019959117431661273,
-      "loss": 1.5488,
+      "loss": 1.5484,
       "step": 705
     },
     {
       "epoch": 0.1268650049137854,
-      "grad_norm": 0.7480871081352234,
+      "grad_norm": 0.7421427369117737,
       "learning_rate": 0.00019956251425362967,
-      "loss": 1.4335,
+      "loss": 1.4334,
       "step": 710
     },
     {
       "epoch": 0.12775842044134728,
-      "grad_norm": 0.6139786243438721,
+      "grad_norm": 0.6117368340492249,
       "learning_rate": 0.0001995328855474903,
-      "loss": 1.5344,
+      "loss": 1.5339,
       "step": 715
     },
     {
       "epoch": 0.12865183596890914,
-      "grad_norm": 0.6919850707054138,
+      "grad_norm": 0.6836646199226379,
       "learning_rate": 0.00019950228848645218,
-      "loss": 1.5349,
+      "loss": 1.5339,
       "step": 720
     },
     {
       "epoch": 0.129545251496471,
-      "grad_norm": 0.6435360908508301,
+      "grad_norm": 0.6464391350746155,
       "learning_rate": 0.00019947072336819397,
-      "loss": 1.52,
+      "loss": 1.5194,
       "step": 725
     },
     {
       "epoch": 0.13043866702403287,
-      "grad_norm": 0.7430850267410278,
+      "grad_norm": 0.7467032670974731,
       "learning_rate": 0.00019943819049981248,
-      "loss": 1.4731,
+      "loss": 1.4732,
       "step": 730
     },
     {
       "epoch": 0.13133208255159476,
-      "grad_norm": 0.6160067319869995,
+      "grad_norm": 0.6169785857200623,
       "learning_rate": 0.00019940469019781985,
-      "loss": 1.5028,
+      "loss": 1.503,
       "step": 735
     },
     {
       "epoch": 0.13222549807915662,
-      "grad_norm": 0.6640142798423767,
+      "grad_norm": 0.6561746001243591,
       "learning_rate": 0.00019937022278814032,
       "loss": 1.5135,
       "step": 740
     },
     {
       "epoch": 0.1331189136067185,
-      "grad_norm": 0.680387556552887,
+      "grad_norm": 0.6765868663787842,
       "learning_rate": 0.00019933478860610713,
-      "loss": 1.4975,
+      "loss": 1.498,
       "step": 745
     },
     {
       "epoch": 0.13401232913428035,
-      "grad_norm": 0.5649276971817017,
+      "grad_norm": 0.5688114762306213,
       "learning_rate": 0.00019929838799645925,
-      "loss": 1.4241,
+      "loss": 1.424,
       "step": 750
     },
     {
       "epoch": 0.13490574466184221,
-      "grad_norm": 0.69096440076828,
+      "grad_norm": 0.6881040334701538,
       "learning_rate": 0.00019926102131333803,
-      "loss": 1.4581,
+      "loss": 1.4588,
       "step": 755
     },
     {
       "epoch": 0.1357991601894041,
-      "grad_norm": 0.7214106917381287,
+      "grad_norm": 0.7207677960395813,
       "learning_rate": 0.00019922268892028368,
-      "loss": 1.4663,
+      "loss": 1.4658,
       "step": 760
     },
     {
       "epoch": 0.13669257571696597,
-      "grad_norm": 0.6388475894927979,
+      "grad_norm": 0.645916759967804,
       "learning_rate": 0.0001991833911902319,
-      "loss": 1.4789,
+      "loss": 1.4797,
       "step": 765
     },
     {
       "epoch": 0.13758599124452783,
-      "grad_norm": 0.653536319732666,
+      "grad_norm": 0.6428040266036987,
       "learning_rate": 0.00019914312850551,
-      "loss": 1.5142,
+      "loss": 1.5143,
       "step": 770
     },
     {
       "epoch": 0.1384794067720897,
-      "grad_norm": 0.6236264109611511,
+      "grad_norm": 0.6166855096817017,
       "learning_rate": 0.0001991019012578335,
-      "loss": 1.4822,
+      "loss": 1.4823,
       "step": 775
     },
     {
       "epoch": 0.13937282229965156,
-      "grad_norm": 0.6071485280990601,
+      "grad_norm": 0.6110146641731262,
       "learning_rate": 0.00019905970984830204,
-      "loss": 1.4656,
+      "loss": 1.4651,
       "step": 780
     },
     {
       "epoch": 0.14026623782721342,
-      "grad_norm": 0.6202497482299805,
+      "grad_norm": 0.6262656450271606,
       "learning_rate": 0.00019901655468739562,
-      "loss": 1.4984,
+      "loss": 1.4983,
       "step": 785
     },
     {
       "epoch": 0.14115965335477532,
-      "grad_norm": 0.6519142389297485,
+      "grad_norm": 0.6442315578460693,
       "learning_rate": 0.00019897243619497056,
-      "loss": 1.473,
+      "loss": 1.4735,
       "step": 790
     },
     {
       "epoch": 0.14205306888233718,
-      "grad_norm": 0.5959755182266235,
+      "grad_norm": 0.5954890251159668,
       "learning_rate": 0.00019892735480025545,
-      "loss": 1.4976,
+      "loss": 1.4986,
       "step": 795
     },
     {
       "epoch": 0.14294648440989904,
-      "grad_norm": 0.6568565368652344,
+      "grad_norm": 0.656102180480957,
       "learning_rate": 0.0001988813109418469,
-      "loss": 1.4929,
+      "loss": 1.4934,
       "step": 800
     },
     {
       "epoch": 0.1438398999374609,
-      "grad_norm": 0.6304877996444702,
+      "grad_norm": 0.630328893661499,
       "learning_rate": 0.00019883430506770536,
-      "loss": 1.4768,
+      "loss": 1.4769,
       "step": 805
     },
     {
       "epoch": 0.14473331546502277,
-      "grad_norm": 0.6260554194450378,
+      "grad_norm": 0.6247830390930176,
       "learning_rate": 0.00019878633763515074,
-      "loss": 1.4449,
+      "loss": 1.4446,
       "step": 810
     },
     {
       "epoch": 0.14562673099258466,
-      "grad_norm": 0.6394698619842529,
+      "grad_norm": 0.6472021341323853,
       "learning_rate": 0.00019873740911085792,
-      "loss": 1.4705,
+      "loss": 1.4711,
       "step": 815
     },
     {
       "epoch": 0.14652014652014653,
-      "grad_norm": 0.6746828556060791,
+      "grad_norm": 0.6813840270042419,
       "learning_rate": 0.00019868751997085225,
-      "loss": 1.4527,
+      "loss": 1.4533,
       "step": 820
     },
     {
       "epoch": 0.1474135620477084,
-      "grad_norm": 0.6241781115531921,
+      "grad_norm": 0.6286069750785828,
       "learning_rate": 0.0001986366707005049,
-      "loss": 1.4807,
+      "loss": 1.4806,
       "step": 825
     },
     {
       "epoch": 0.14830697757527025,
-      "grad_norm": 0.6119701266288757,
+      "grad_norm": 0.6147023439407349,
       "learning_rate": 0.00019858486179452812,
-      "loss": 1.478,
+      "loss": 1.4779,
       "step": 830
     },
     {
       "epoch": 0.14920039310283212,
-      "grad_norm": 0.6817257404327393,
+      "grad_norm": 0.6876225471496582,
       "learning_rate": 0.0001985320937569705,
-      "loss": 1.507,
+      "loss": 1.5072,
       "step": 835
     },
     {
       "epoch": 0.150093808630394,
-      "grad_norm": 0.6211469173431396,
+      "grad_norm": 0.6138557195663452,
       "learning_rate": 0.00019847836710121198,
-      "loss": 1.4612,
+      "loss": 1.4605,
       "step": 840
     },
     {
       "epoch": 0.15098722415795587,
-      "grad_norm": 0.6433590054512024,
+      "grad_norm": 0.642737865447998,
       "learning_rate": 0.0001984236823499589,
-      "loss": 1.4689,
+      "loss": 1.4692,
       "step": 845
     },
     {
       "epoch": 0.15188063968551774,
-      "grad_norm": 0.6386223435401917,
+      "grad_norm": 0.6422574520111084,
       "learning_rate": 0.0001983680400352389,
-      "loss": 1.4681,
+      "loss": 1.4684,
       "step": 850
     },
     {
       "epoch": 0.1527740552130796,
-      "grad_norm": 0.6272917985916138,
+      "grad_norm": 0.6301836967468262,
       "learning_rate": 0.00019831144069839578,
-      "loss": 1.4372,
+      "loss": 1.4375,
       "step": 855
     },
     {
       "epoch": 0.15366747074064147,
-      "grad_norm": 0.5687587857246399,
+      "grad_norm": 0.5667446255683899,
       "learning_rate": 0.00019825388489008415,
-      "loss": 1.4271,
+      "loss": 1.427,
       "step": 860
     },
     {
       "epoch": 0.15456088626820333,
-      "grad_norm": 0.6058815121650696,
+      "grad_norm": 0.6058502197265625,
       "learning_rate": 0.0001981953731702642,
-      "loss": 1.4664,
+      "loss": 1.4666,
       "step": 865
     },
     {
       "epoch": 0.15545430179576522,
-      "grad_norm": 0.5973011255264282,
+      "grad_norm": 0.6004565358161926,
       "learning_rate": 0.00019813590610819604,
-      "loss": 1.4698,
+      "loss": 1.4691,
       "step": 870
     },
     {
       "epoch": 0.15634771732332708,
-      "grad_norm": 0.6365277767181396,
+      "grad_norm": 0.6401507258415222,
       "learning_rate": 0.00019807548428243447,
-      "loss": 1.4151,
+      "loss": 1.4149,
       "step": 875
     },
     {
       "epoch": 0.15724113285088895,
-      "grad_norm": 0.6562153100967407,
+      "grad_norm": 0.6585138440132141,
       "learning_rate": 0.00019801410828082307,
-      "loss": 1.4769,
+      "loss": 1.478,
       "step": 880
     },
     {
       "epoch": 0.1581345483784508,
-      "grad_norm": 0.6231607794761658,
+      "grad_norm": 0.6155353784561157,
       "learning_rate": 0.00019795177870048864,
-      "loss": 1.4464,
+      "loss": 1.446,
       "step": 885
     },
     {
       "epoch": 0.15902796390601268,
-      "grad_norm": 0.6296908855438232,
+      "grad_norm": 0.6294763684272766,
       "learning_rate": 0.00019788849614783534,
-      "loss": 1.4202,
+      "loss": 1.4208,
       "step": 890
     },
     {
       "epoch": 0.15992137943357457,
-      "grad_norm": 0.5952068567276001,
+      "grad_norm": 0.5941749811172485,
       "learning_rate": 0.00019782426123853873,
-      "loss": 1.4916,
+      "loss": 1.4909,
       "step": 895
     },
     {
       "epoch": 0.16081479496113643,
-      "grad_norm": 0.659608006477356,
+      "grad_norm": 0.6547942161560059,
       "learning_rate": 0.0001977590745975399,
-      "loss": 1.3604,
+      "loss": 1.3606,
       "step": 900
     },
     {
       "epoch": 0.1617082104886983,
-      "grad_norm": 0.6710678339004517,
+      "grad_norm": 0.6711387634277344,
       "learning_rate": 0.00019769293685903937,
-      "loss": 1.426,
+      "loss": 1.4263,
       "step": 905
     },
     {
       "epoch": 0.16260162601626016,
-      "grad_norm": 0.6199014782905579,
+      "grad_norm": 0.6281640529632568,
       "learning_rate": 0.0001976258486664908,
-      "loss": 1.4376,
+      "loss": 1.4387,
       "step": 910
     },
     {
       "epoch": 0.16349504154382202,
-      "grad_norm": 0.5568641424179077,
+      "grad_norm": 0.5495440363883972,
       "learning_rate": 0.00019755781067259487,
-      "loss": 1.4233,
+      "loss": 1.4227,
       "step": 915
     },
     {
       "epoch": 0.16438845707138391,
-      "grad_norm": 0.6075164675712585,
+      "grad_norm": 0.6133942008018494,
       "learning_rate": 0.00019748882353929283,
-      "loss": 1.3773,
+      "loss": 1.3777,
       "step": 920
     },
     {
       "epoch": 0.16528187259894578,
-      "grad_norm": 0.6208435297012329,
+      "grad_norm": 0.6217373013496399,
       "learning_rate": 0.00019741888793776012,
-      "loss": 1.4342,
+      "loss": 1.4337,
       "step": 925
     },
     {
       "epoch": 0.16617528812650764,
-      "grad_norm": 0.5977674126625061,
+      "grad_norm": 0.5960021018981934,
       "learning_rate": 0.00019734800454839985,
       "loss": 1.4001,
       "step": 930
     },
     {
       "epoch": 0.1670687036540695,
-      "grad_norm": 0.6107041835784912,
+      "grad_norm": 0.6111594438552856,
       "learning_rate": 0.00019727617406083608,
-      "loss": 1.4483,
+      "loss": 1.4485,
       "step": 935
     },
     {
       "epoch": 0.16796211918163137,
-      "grad_norm": 0.6070581078529358,
+      "grad_norm": 0.6061040163040161,
       "learning_rate": 0.00019720339717390725,
-      "loss": 1.434,
+      "loss": 1.4341,
       "step": 940
     },
     {
       "epoch": 0.16885553470919323,
-      "grad_norm": 0.5824504494667053,
+      "grad_norm": 0.5827285051345825,
       "learning_rate": 0.00019712967459565935,
-      "loss": 1.4214,
+      "loss": 1.4212,
       "step": 945
     },
     {
       "epoch": 0.16974895023675513,
-      "grad_norm": 0.6274870038032532,
+      "grad_norm": 0.625501275062561,
       "learning_rate": 0.00019705500704333888,
-      "loss": 1.3898,
+      "loss": 1.3902,
       "step": 950
     },
     {
       "epoch": 0.170642365764317,
-      "grad_norm": 0.6038565039634705,
+      "grad_norm": 0.6008924245834351,
       "learning_rate": 0.00019697939524338605,
-      "loss": 1.441,
+      "loss": 1.4401,
       "step": 955
     },
     {
       "epoch": 0.17153578129187885,
-      "grad_norm": 0.5851924419403076,
+      "grad_norm": 0.5836713314056396,
       "learning_rate": 0.00019690283993142768,
-      "loss": 1.4293,
+      "loss": 1.4285,
       "step": 960
     },
     {
       "epoch": 0.17242919681944072,
-      "grad_norm": 0.6128916144371033,
+      "grad_norm": 0.6134169101715088,
       "learning_rate": 0.00019682534185226996,
-      "loss": 1.4526,
+      "loss": 1.4527,
       "step": 965
     },
     {
       "epoch": 0.17332261234700258,
-      "grad_norm": 0.5964568853378296,
+      "grad_norm": 0.593845784664154,
       "learning_rate": 0.0001967469017598913,
-      "loss": 1.4286,
+      "loss": 1.4293,
       "step": 970
     },
     {
       "epoch": 0.17421602787456447,
-      "grad_norm": 0.6289203763008118,
+      "grad_norm": 0.6239144802093506,
       "learning_rate": 0.00019666752041743485,
-      "loss": 1.3942,
+      "loss": 1.394,
       "step": 975
     },
     {
       "epoch": 0.17510944340212634,
-      "grad_norm": 0.538835883140564,
+      "grad_norm": 0.5358985066413879,
       "learning_rate": 0.00019658719859720137,
-      "loss": 1.4165,
+      "loss": 1.4162,
       "step": 980
     },
     {
       "epoch": 0.1760028589296882,
-      "grad_norm": 0.571048378944397,
+      "grad_norm": 0.5680092573165894,
       "learning_rate": 0.00019650593708064133,
-      "loss": 1.4293,
+      "loss": 1.4292,
       "step": 985
     },
     {
       "epoch": 0.17689627445725006,
-      "grad_norm": 0.6304472088813782,
+      "grad_norm": 0.6288170218467712,
       "learning_rate": 0.0001964237366583476,
-      "loss": 1.4069,
+      "loss": 1.4065,
       "step": 990
     },
     {
       "epoch": 0.17778968998481193,
-      "grad_norm": 0.6265087723731995,
+      "grad_norm": 0.6214203238487244,
       "learning_rate": 0.00019634059813004767,
-      "loss": 1.3956,
+      "loss": 1.3958,
       "step": 995
     },
     {
       "epoch": 0.17868310551237382,
-      "grad_norm": 0.5952944159507751,
+      "grad_norm": 0.5966234803199768,
       "learning_rate": 0.00019625652230459577,
-      "loss": 1.4243,
+      "loss": 1.4245,
       "step": 1000
     },
     {
       "epoch": 0.17957652103993568,
-      "grad_norm": 0.5611996650695801,
+      "grad_norm": 0.5608589053153992,
       "learning_rate": 0.00019617150999996522,
-      "loss": 1.3769,
+      "loss": 1.3759,
       "step": 1005
     },
     {
       "epoch": 0.18046993656749755,
-      "grad_norm": 0.6139466166496277,
+      "grad_norm": 0.6149086952209473,
       "learning_rate": 0.00019608556204324016,
-      "loss": 1.3937,
+      "loss": 1.3928,
       "step": 1010
     },
     {
       "epoch": 0.1813633520950594,
-      "grad_norm": 0.59721839427948,
+      "grad_norm": 0.5998888611793518,
       "learning_rate": 0.00019599867927060788,
-      "loss": 1.4536,
+      "loss": 1.4535,
       "step": 1015
     },
     {
       "epoch": 0.18225676762262127,
-      "grad_norm": 0.6089359521865845,
+      "grad_norm": 0.6067023873329163,
       "learning_rate": 0.0001959108625273504,
-      "loss": 1.4217,
+      "loss": 1.4224,
       "step": 1020
     },
     {
       "epoch": 0.18315018315018314,
-      "grad_norm": 0.6171666383743286,
+      "grad_norm": 0.6208594441413879,
       "learning_rate": 0.0001958221126678363,
-      "loss": 1.3777,
+      "loss": 1.3772,
       "step": 1025
     },
     {
       "epoch": 0.18404359867774503,
-      "grad_norm": 0.5585091710090637,
+      "grad_norm": 0.555916428565979,
       "learning_rate": 0.00019573243055551247,
-      "loss": 1.4085,
+      "loss": 1.4087,
       "step": 1030
     },
     {
       "epoch": 0.1849370142053069,
-      "grad_norm": 0.5734081268310547,
+      "grad_norm": 0.5708112716674805,
       "learning_rate": 0.0001956418170628957,
-      "loss": 1.3873,
+      "loss": 1.3879,
       "step": 1035
     },
     {
       "epoch": 0.18583042973286876,
-      "grad_norm": 0.6407363414764404,
+      "grad_norm": 0.6426623463630676,
       "learning_rate": 0.0001955502730715642,
-      "loss": 1.4274,
+      "loss": 1.4281,
       "step": 1040
     },
     {
       "epoch": 0.18672384526043062,
-      "grad_norm": 0.6202085614204407,
+      "grad_norm": 0.6203565001487732,
       "learning_rate": 0.0001954577994721489,
-      "loss": 1.4569,
+      "loss": 1.457,
       "step": 1045
     },
     {
       "epoch": 0.18761726078799248,
-      "grad_norm": 0.5608704090118408,
+      "grad_norm": 0.5595524907112122,
       "learning_rate": 0.00019536439716432496,
-      "loss": 1.3569,
+      "loss": 1.3567,
       "step": 1050
     },
     {
       "epoch": 0.18851067631555438,
-      "grad_norm": 0.5941877365112305,
+      "grad_norm": 0.5930672287940979,
       "learning_rate": 0.00019527006705680297,
-      "loss": 1.3997,
+      "loss": 1.3994,
       "step": 1055
     },
     {
       "epoch": 0.18940409184311624,
-      "grad_norm": 0.5690048336982727,
+      "grad_norm": 0.5650060176849365,
       "learning_rate": 0.00019517481006731997,
-      "loss": 1.4205,
+      "loss": 1.4198,
       "step": 1060
     },
     {
       "epoch": 0.1902975073706781,
-      "grad_norm": 0.5559670925140381,
+      "grad_norm": 0.558860719203949,
       "learning_rate": 0.0001950786271226307,
-      "loss": 1.3996,
+      "loss": 1.399,
       "step": 1065
     },
     {
       "epoch": 0.19119092289823997,
-      "grad_norm": 0.633906364440918,
+      "grad_norm": 0.6348270177841187,
       "learning_rate": 0.00019498151915849855,
-      "loss": 1.4224,
+      "loss": 1.4218,
       "step": 1070
     },
     {
       "epoch": 0.19208433842580183,
-      "grad_norm": 0.6395050287246704,
+      "grad_norm": 0.63995760679245,
       "learning_rate": 0.00019488348711968633,
-      "loss": 1.3711,
+      "loss": 1.3706,
       "step": 1075
     },
     {
       "epoch": 0.19297775395336372,
-      "grad_norm": 0.5923237800598145,
+      "grad_norm": 0.5952381491661072,
       "learning_rate": 0.00019478453195994719,
-      "loss": 1.4317,
+      "loss": 1.4318,
       "step": 1080
     },
     {
       "epoch": 0.1938711694809256,
-      "grad_norm": 0.6085279583930969,
+      "grad_norm": 0.6112159490585327,
       "learning_rate": 0.0001946846546420154,
-      "loss": 1.3991,
+      "loss": 1.3999,
       "step": 1085
     },
     {
       "epoch": 0.19476458500848745,
-      "grad_norm": 0.5692582726478577,
+      "grad_norm": 0.5743675231933594,
       "learning_rate": 0.0001945838561375968,
-      "loss": 1.4099,
+      "loss": 1.4095,
       "step": 1090
     },
     {
       "epoch": 0.19565800053604931,
-      "grad_norm": 0.5419508218765259,
+      "grad_norm": 0.542060911655426,
       "learning_rate": 0.00019448213742735942,
-      "loss": 1.3943,
+      "loss": 1.3946,
       "step": 1095
     },
     {
       "epoch": 0.19655141606361118,
-      "grad_norm": 0.5866132378578186,
+      "grad_norm": 0.5920471549034119,
       "learning_rate": 0.0001943794995009242,
-      "loss": 1.3711,
+      "loss": 1.3708,
       "step": 1100
     },
     {
       "epoch": 0.19744483159117304,
-      "grad_norm": 0.5647941827774048,
+      "grad_norm": 0.5597534775733948,
       "learning_rate": 0.00019427594335685478,
-      "loss": 1.3508,
+      "loss": 1.35,
       "step": 1105
     },
     {
       "epoch": 0.19833824711873493,
-      "grad_norm": 0.5306203365325928,
+      "grad_norm": 0.531980037689209,
       "learning_rate": 0.00019417147000264852,
-      "loss": 1.3501,
+      "loss": 1.3503,
       "step": 1110
     },
     {
       "epoch": 0.1992316626462968,
-      "grad_norm": 0.5783630609512329,
+      "grad_norm": 0.5867334604263306,
       "learning_rate": 0.0001940660804547259,
-      "loss": 1.367,
+      "loss": 1.3673,
       "step": 1115
     },
     {
       "epoch": 0.20012507817385866,
-      "grad_norm": 0.5690432786941528,
+      "grad_norm": 0.5700910091400146,
       "learning_rate": 0.00019395977573842142,
-      "loss": 1.3662,
+      "loss": 1.3664,
       "step": 1120
     },
     {
       "epoch": 0.20101849370142053,
-      "grad_norm": 0.5380032658576965,
+      "grad_norm": 0.5407332181930542,
       "learning_rate": 0.000193852556887973,
-      "loss": 1.3674,
+      "loss": 1.3677,
       "step": 1125
     },
     {
       "epoch": 0.2019119092289824,
-      "grad_norm": 0.5310468077659607,
+      "grad_norm": 0.5347940921783447,
       "learning_rate": 0.00019374442494651223,
-      "loss": 1.3827,
+      "loss": 1.3819,
       "step": 1130
     },
     {
       "epoch": 0.20280532475654428,
-      "grad_norm": 0.5655437707901001,
+      "grad_norm": 0.5672542452812195,
       "learning_rate": 0.00019363538096605427,
-      "loss": 1.3837,
+      "loss": 1.3838,
       "step": 1135
     },
     {
       "epoch": 0.20369874028410614,
-      "grad_norm": 0.5816426873207092,
+      "grad_norm": 0.5784164667129517,
       "learning_rate": 0.00019352542600748734,
-      "loss": 1.3857,
+      "loss": 1.3848,
       "step": 1140
     },
     {
       "epoch": 0.204592155811668,
-      "grad_norm": 0.5484604835510254,
+      "grad_norm": 0.5457838177680969,
       "learning_rate": 0.00019341456114056263,
-      "loss": 1.3561,
+      "loss": 1.3567,
       "step": 1145
     },
     {
       "epoch": 0.20548557133922987,
-      "grad_norm": 0.5980105996131897,
+      "grad_norm": 0.6002548336982727,
       "learning_rate": 0.00019330278744388385,
-      "loss": 1.3571,
+      "loss": 1.357,
       "step": 1150
     },
     {
       "epoch": 0.20637898686679174,
-      "grad_norm": 0.5934194922447205,
+      "grad_norm": 0.58780437707901,
       "learning_rate": 0.00019319010600489663,
-      "loss": 1.3629,
+      "loss": 1.3618,
       "step": 1155
     },
     {
       "epoch": 0.20727240239435363,
-      "grad_norm": 0.5532252192497253,
+      "grad_norm": 0.5515660047531128,
       "learning_rate": 0.00019307651791987816,
-      "loss": 1.3442,
+      "loss": 1.344,
       "step": 1160
     },
     {
       "epoch": 0.2081658179219155,
-      "grad_norm": 0.5868908762931824,
+      "grad_norm": 0.5853777527809143,
       "learning_rate": 0.00019296202429392622,
-      "loss": 1.3697,
+      "loss": 1.3699,
       "step": 1165
     },
     {
       "epoch": 0.20905923344947736,
-      "grad_norm": 0.557306170463562,
+      "grad_norm": 0.558180034160614,
       "learning_rate": 0.00019284662624094874,
-      "loss": 1.3245,
+      "loss": 1.325,
       "step": 1170
     },
     {
       "epoch": 0.20995264897703922,
-      "grad_norm": 0.5710069537162781,
+      "grad_norm": 0.5694774389266968,
       "learning_rate": 0.00019273032488365267,
-      "loss": 1.3606,
+      "loss": 1.3608,
       "step": 1175
     },
     {
       "epoch": 0.21084606450460108,
-      "grad_norm": 0.5752673745155334,
+      "grad_norm": 0.5699239373207092,
       "learning_rate": 0.00019261312135353332,
       "loss": 1.3384,
       "step": 1180
     },
     {
       "epoch": 0.21173948003216295,
-      "grad_norm": 0.5679814219474792,
+      "grad_norm": 0.5678359270095825,
       "learning_rate": 0.0001924950167908632,
-      "loss": 1.3272,
+      "loss": 1.3271,
       "step": 1185
     },
     {
       "epoch": 0.21263289555972484,
-      "grad_norm": 0.5580770373344421,
+      "grad_norm": 0.560370147228241,
       "learning_rate": 0.00019237601234468096,
-      "loss": 1.3557,
+      "loss": 1.3563,
       "step": 1190
     },
     {
       "epoch": 0.2135263110872867,
-      "grad_norm": 0.5821225047111511,
+      "grad_norm": 0.5844460129737854,
       "learning_rate": 0.0001922561091727802,
-      "loss": 1.326,
+      "loss": 1.3256,
       "step": 1195
     },
     {
       "epoch": 0.21441972661484857,
-      "grad_norm": 0.593561053276062,
+      "grad_norm": 0.5936894416809082,
       "learning_rate": 0.00019213530844169817,
-      "loss": 1.3584,
+      "loss": 1.3581,
       "step": 1200
     },
     {
       "epoch": 0.21531314214241043,
-      "grad_norm": 0.6168695688247681,
+      "grad_norm": 0.6150529384613037,
       "learning_rate": 0.00019201361132670456,
-      "loss": 1.3307,
+      "loss": 1.3306,
       "step": 1205
     },
     {
       "epoch": 0.2162065576699723,
-      "grad_norm": 0.5816683173179626,
+      "grad_norm": 0.5778645873069763,
       "learning_rate": 0.00019189101901178997,
-      "loss": 1.3807,
+      "loss": 1.381,
       "step": 1210
     },
     {
       "epoch": 0.21709997319753419,
-      "grad_norm": 0.5499683022499084,
+      "grad_norm": 0.5521702170372009,
       "learning_rate": 0.00019176753268965432,
-      "loss": 1.3177,
+      "loss": 1.3191,
       "step": 1215
     },
     {
       "epoch": 0.21799338872509605,
-      "grad_norm": 0.5521628260612488,
+      "grad_norm": 0.550817608833313,
       "learning_rate": 0.00019164315356169536,
-      "loss": 1.3386,
+      "loss": 1.3381,
       "step": 1220
     },
     {
       "epoch": 0.2188868042526579,
-      "grad_norm": 0.6149545907974243,
+      "grad_norm": 0.6147415041923523,
       "learning_rate": 0.00019151788283799698,
-      "loss": 1.3589,
+      "loss": 1.3591,
       "step": 1225
     },
     {
       "epoch": 0.21978021978021978,
-      "grad_norm": 0.5420519113540649,
+      "grad_norm": 0.5430126786231995,
       "learning_rate": 0.00019139172173731733,
-      "loss": 1.3446,
+      "loss": 1.344,
       "step": 1230
     },
     {
       "epoch": 0.22067363530778164,
-      "grad_norm": 0.5895097851753235,
+      "grad_norm": 0.5844172835350037,
       "learning_rate": 0.0001912646714870771,
-      "loss": 1.3714,
+      "loss": 1.371,
       "step": 1235
     },
     {
       "epoch": 0.22156705083534353,
-      "grad_norm": 0.5893450975418091,
+      "grad_norm": 0.5882902145385742,
       "learning_rate": 0.0001911367333233474,
-      "loss": 1.3723,
+      "loss": 1.3712,
       "step": 1240
     },
     {
       "epoch": 0.2224604663629054,
-      "grad_norm": 0.5816531181335449,
+      "grad_norm": 0.5897734761238098,
       "learning_rate": 0.00019100790849083804,
-      "loss": 1.3002,
+      "loss": 1.3004,
       "step": 1245
     },
     {
       "epoch": 0.22335388189046726,
-      "grad_norm": 0.557987630367279,
+      "grad_norm": 0.554033637046814,
       "learning_rate": 0.00019087819824288504,
-      "loss": 1.3805,
+      "loss": 1.3803,
       "step": 1250
     },
     {
       "epoch": 0.22424729741802912,
-      "grad_norm": 0.5602419972419739,
+      "grad_norm": 0.5678560733795166,
       "learning_rate": 0.0001907476038414387,
-      "loss": 1.3822,
+      "loss": 1.3817,
       "step": 1255
     },
     {
       "epoch": 0.225140712945591,
-      "grad_norm": 0.5873110294342041,
+      "grad_norm": 0.5871914029121399,
       "learning_rate": 0.00019061612655705128,
-      "loss": 1.3457,
+      "loss": 1.3456,
       "step": 1260
     },
     {
       "epoch": 0.22603412847315285,
-      "grad_norm": 0.5846714973449707,
+      "grad_norm": 0.587515115737915,
       "learning_rate": 0.00019048376766886448,
-      "loss": 1.3483,
+      "loss": 1.3484,
       "step": 1265
     },
     {
       "epoch": 0.22692754400071474,
-      "grad_norm": 0.5950532555580139,
+      "grad_norm": 0.5953535437583923,
       "learning_rate": 0.00019035052846459727,
-      "loss": 1.3358,
+      "loss": 1.3351,
       "step": 1270
     },
     {
       "epoch": 0.2278209595282766,
-      "grad_norm": 0.5122111439704895,
+      "grad_norm": 0.5097110271453857,
       "learning_rate": 0.00019021641024053308,
-      "loss": 1.3126,
+      "loss": 1.3122,
       "step": 1275
     },
     {
       "epoch": 0.22871437505583847,
-      "grad_norm": 0.5635678172111511,
+      "grad_norm": 0.563342273235321,
       "learning_rate": 0.00019008141430150745,
-      "loss": 1.3349,
+      "loss": 1.3345,
       "step": 1280
     },
     {
       "epoch": 0.22960779058340033,
-      "grad_norm": 0.5411239266395569,
+      "grad_norm": 0.5420668721199036,
       "learning_rate": 0.00018994554196089506,
-      "loss": 1.3382,
+      "loss": 1.3379,
       "step": 1285
     },
     {
       "epoch": 0.2305012061109622,
-      "grad_norm": 0.5744121670722961,
+      "grad_norm": 0.5748648047447205,
       "learning_rate": 0.0001898087945405972,
-      "loss": 1.2789,
+      "loss": 1.2791,
       "step": 1290
     },
     {
       "epoch": 0.2313946216385241,
-      "grad_norm": 0.578059732913971,
+      "grad_norm": 0.5751796364784241,
       "learning_rate": 0.00018967117337102883,
-      "loss": 1.3034,
+      "loss": 1.3039,
       "step": 1295
     },
     {
       "epoch": 0.23228803716608595,
-      "grad_norm": 0.5708723664283752,
+      "grad_norm": 0.5670933127403259,
       "learning_rate": 0.00018953267979110545,
-      "loss": 1.3583,
+      "loss": 1.358,
       "step": 1300
     },
     {
       "epoch": 0.23318145269364782,
-      "grad_norm": 0.6082606315612793,
+      "grad_norm": 0.6038489937782288,
       "learning_rate": 0.0001893933151482304,
-      "loss": 1.3753,
+      "loss": 1.3747,
       "step": 1305
     },
     {
       "epoch": 0.23407486822120968,
-      "grad_norm": 0.57098788022995,
+      "grad_norm": 0.567458987236023,
       "learning_rate": 0.00018925308079828152,
       "loss": 1.3369,
       "step": 1310
     },
     {
       "epoch": 0.23496828374877154,
-      "grad_norm": 0.5236451625823975,
+      "grad_norm": 0.5277537107467651,
       "learning_rate": 0.00018911197810559803,
-      "loss": 1.3468,
+      "loss": 1.3464,
       "step": 1315
     },
     {
       "epoch": 0.23586169927633344,
-      "grad_norm": 0.5678128600120544,
+      "grad_norm": 0.5678889155387878,
       "learning_rate": 0.00018897000844296727,
-      "loss": 1.342,
+      "loss": 1.3414,
       "step": 1320
     },
     {
       "epoch": 0.2367551148038953,
-      "grad_norm": 0.544394314289093,
+      "grad_norm": 0.548362135887146,
       "learning_rate": 0.00018882717319161128,
-      "loss": 1.3434,
+      "loss": 1.3431,
       "step": 1325
     },
     {
       "epoch": 0.23764853033145716,
-      "grad_norm": 0.5973039865493774,
+      "grad_norm": 0.5988355278968811,
       "learning_rate": 0.00018868347374117344,
-      "loss": 1.2866,
+      "loss": 1.287,
       "step": 1330
     },
     {
       "epoch": 0.23854194585901903,
-      "grad_norm": 0.6183381080627441,
+      "grad_norm": 0.6163780093193054,
       "learning_rate": 0.00018853891148970498,
-      "loss": 1.3513,
+      "loss": 1.3503,
       "step": 1335
     },
     {
       "epoch": 0.2394353613865809,
-      "grad_norm": 0.5657194256782532,
+      "grad_norm": 0.5658857822418213,
       "learning_rate": 0.00018839348784365116,
-      "loss": 1.3293,
+      "loss": 1.3291,
       "step": 1340
     },
     {
       "epoch": 0.24032877691414276,
-      "grad_norm": 0.5699062347412109,
+      "grad_norm": 0.5733643770217896,
       "learning_rate": 0.0001882472042178379,
-      "loss": 1.3529,
+      "loss": 1.3533,
       "step": 1345
     },
     {
       "epoch": 0.24122219244170465,
-      "grad_norm": 0.5497627258300781,
+      "grad_norm": 0.5487879514694214,
       "learning_rate": 0.0001881000620354578,
-      "loss": 1.3395,
+      "loss": 1.3396,
       "step": 1350
     },
     {
       "epoch": 0.2421156079692665,
-      "grad_norm": 0.5238006711006165,
+      "grad_norm": 0.5239369869232178,
       "learning_rate": 0.0001879520627280563,
-      "loss": 1.3195,
+      "loss": 1.3193,
       "step": 1355
     },
     {
       "epoch": 0.24300902349682837,
-      "grad_norm": 0.5889202356338501,
+      "grad_norm": 0.5878532528877258,
       "learning_rate": 0.0001878032077355179,
-      "loss": 1.3201,
+      "loss": 1.3203,
       "step": 1360
     },
     {
       "epoch": 0.24390243902439024,
-      "grad_norm": 0.543110191822052,
+      "grad_norm": 0.5400727391242981,
       "learning_rate": 0.00018765349850605195,
-      "loss": 1.3083,
+      "loss": 1.3079,
       "step": 1365
     },
     {
       "epoch": 0.2447958545519521,
-      "grad_norm": 0.580144464969635,
+      "grad_norm": 0.5751279592514038,
       "learning_rate": 0.0001875029364961788,
       "loss": 1.2999,
       "step": 1370
     },
     {
       "epoch": 0.245689270079514,
-      "grad_norm": 0.5746851563453674,
+      "grad_norm": 0.5734187960624695,
       "learning_rate": 0.00018735152317071534,
-      "loss": 1.3282,
+      "loss": 1.3283,
       "step": 1375
     },
     {
       "epoch": 0.24658268560707586,
-      "grad_norm": 0.5605974793434143,
+      "grad_norm": 0.5625242590904236,
       "learning_rate": 0.00018719926000276106,
-      "loss": 1.3357,
+      "loss": 1.3355,
       "step": 1380
     },
     {
       "epoch": 0.24747610113463772,
-      "grad_norm": 0.556448757648468,
+      "grad_norm": 0.5596960186958313,
       "learning_rate": 0.0001870461484736834,
-      "loss": 1.3159,
+      "loss": 1.3156,
       "step": 1385
     },
     {
       "epoch": 0.24836951666219959,
-      "grad_norm": 0.5685500502586365,
+      "grad_norm": 0.5672683119773865,
       "learning_rate": 0.00018689219007310369,
-      "loss": 1.3422,
+      "loss": 1.342,
       "step": 1390
     },
     {
       "epoch": 0.24926293218976145,
-      "grad_norm": 0.5622739195823669,
+      "grad_norm": 0.5605707764625549,
       "learning_rate": 0.00018673738629888226,
-      "loss": 1.3162,
+      "loss": 1.3166,
       "step": 1395
     },
     {
       "epoch": 0.2501563477173233,
-      "grad_norm": 0.5816725492477417,
+      "grad_norm": 0.5812191367149353,
       "learning_rate": 0.0001865817386571043,
-      "loss": 1.348,
+      "loss": 1.3484,
       "step": 1400
     },
     {
       "epoch": 0.2510497632448852,
-      "grad_norm": 0.5116124153137207,
+      "grad_norm": 0.5138891935348511,
       "learning_rate": 0.00018642524866206475,
-      "loss": 1.3196,
+      "loss": 1.3201,
       "step": 1405
     },
     {
       "epoch": 0.25194317877244704,
-      "grad_norm": 0.5634516477584839,
+      "grad_norm": 0.5659753084182739,
       "learning_rate": 0.000186267917836254,
       "loss": 1.3117,
       "step": 1410
     },
     {
       "epoch": 0.25283659430000893,
-      "grad_norm": 0.5746950507164001,
+      "grad_norm": 0.5799670815467834,
       "learning_rate": 0.00018610974771034275,
-      "loss": 1.316,
+      "loss": 1.3161,
       "step": 1415
     },
     {
       "epoch": 0.2537300098275708,
-      "grad_norm": 0.5387247800827026,
+      "grad_norm": 0.537071943283081,
       "learning_rate": 0.00018595073982316732,
-      "loss": 1.3182,
+      "loss": 1.3183,
       "step": 1420
     },
     {
       "epoch": 0.25462342535513266,
-      "grad_norm": 0.5491217374801636,
+      "grad_norm": 0.5561192631721497,
       "learning_rate": 0.00018579089572171454,
-      "loss": 1.3214,
+      "loss": 1.321,
       "step": 1425
     },
     {
       "epoch": 0.25551684088269455,
-      "grad_norm": 0.5725581049919128,
+      "grad_norm": 0.5795418620109558,
       "learning_rate": 0.00018563021696110682,
-      "loss": 1.3304,
+      "loss": 1.3301,
       "step": 1430
     },
     {
       "epoch": 0.2564102564102564,
-      "grad_norm": 0.5422765016555786,
+      "grad_norm": 0.537104070186615,
       "learning_rate": 0.0001854687051045869,
-      "loss": 1.2947,
+      "loss": 1.2937,
       "step": 1435
     },
     {
       "epoch": 0.2573036719378183,
-      "grad_norm": 0.5965689420700073,
+      "grad_norm": 0.6000511050224304,
       "learning_rate": 0.00018530636172350287,
-      "loss": 1.2988,
+      "loss": 1.2987,
       "step": 1440
     },
     {
       "epoch": 0.25819708746538017,
-      "grad_norm": 0.5194745659828186,
+      "grad_norm": 0.5180054903030396,
       "learning_rate": 0.00018514318839729242,
-      "loss": 1.29,
+      "loss": 1.2903,
       "step": 1445
     },
     {
       "epoch": 0.259090502992942,
-      "grad_norm": 0.5547758936882019,
+      "grad_norm": 0.5556187629699707,
       "learning_rate": 0.00018497918671346808,
       "loss": 1.3159,
       "step": 1450
     },
     {
       "epoch": 0.2599839185205039,
-      "grad_norm": 0.5594758987426758,
+      "grad_norm": 0.5588898062705994,
       "learning_rate": 0.0001848143582676013,
-      "loss": 1.2949,
+      "loss": 1.295,
       "step": 1455
     },
     {
       "epoch": 0.26087733404806573,
-      "grad_norm": 0.5487962961196899,
+      "grad_norm": 0.5521529316902161,
       "learning_rate": 0.0001846487046633071,
       "loss": 1.2851,
       "step": 1460
     },
     {
       "epoch": 0.2617707495756276,
-      "grad_norm": 0.5708212852478027,
+      "grad_norm": 0.5741902589797974,
       "learning_rate": 0.0001844822275122285,
-      "loss": 1.3092,
+      "loss": 1.3095,
       "step": 1465
     },
     {
       "epoch": 0.2626641651031895,
-      "grad_norm": 0.5744941234588623,
+      "grad_norm": 0.5776512026786804,
       "learning_rate": 0.00018431492843402084,
-      "loss": 1.32,
+      "loss": 1.3194,
       "step": 1470
     },
     {
       "epoch": 0.26355758063075135,
-      "grad_norm": 0.5791029930114746,
+      "grad_norm": 0.5787376165390015,
       "learning_rate": 0.00018414680905633586,
-      "loss": 1.3131,
+      "loss": 1.3132,
       "step": 1475
     },
     {
       "epoch": 0.26445099615831325,
-      "grad_norm": 0.5608724355697632,
+      "grad_norm": 0.551239013671875,
       "learning_rate": 0.00018397787101480612,
-      "loss": 1.2954,
+      "loss": 1.2952,
       "step": 1480
     },
     {
       "epoch": 0.2653444116858751,
-      "grad_norm": 0.5400808453559875,
+      "grad_norm": 0.535438060760498,
       "learning_rate": 0.0001838081159530289,
-      "loss": 1.2775,
+      "loss": 1.277,
       "step": 1485
     },
     {
       "epoch": 0.266237827213437,
-      "grad_norm": 0.5288774371147156,
+      "grad_norm": 0.5279135704040527,
       "learning_rate": 0.0001836375455225502,
-      "loss": 1.3106,
+      "loss": 1.3103,
       "step": 1490
     },
     {
       "epoch": 0.26713124274099886,
-      "grad_norm": 0.5479869842529297,
+      "grad_norm": 0.54332035779953,
       "learning_rate": 0.00018346616138284892,
-      "loss": 1.3533,
+      "loss": 1.3535,
       "step": 1495
     },
     {
       "epoch": 0.2680246582685607,
-      "grad_norm": 0.5189379453659058,
+      "grad_norm": 0.519645631313324,
       "learning_rate": 0.0001832939652013203,
-      "loss": 1.3082,
+      "loss": 1.3083,
       "step": 1500
     },
     {
       "epoch": 0.2689180737961226,
-      "grad_norm": 0.5478213429450989,
+      "grad_norm": 0.5469849705696106,
       "learning_rate": 0.00018312095865326012,
-      "loss": 1.273,
+      "loss": 1.2726,
       "step": 1505
     },
     {
       "epoch": 0.26981148932368443,
-      "grad_norm": 0.5622948408126831,
+      "grad_norm": 0.5668709874153137,
       "learning_rate": 0.0001829471434218481,
-      "loss": 1.2966,
+      "loss": 1.2965,
       "step": 1510
     },
     {
       "epoch": 0.2707049048512463,
-      "grad_norm": 0.5554372072219849,
+      "grad_norm": 0.5535402894020081,
       "learning_rate": 0.00018277252119813176,
-      "loss": 1.274,
+      "loss": 1.2742,
       "step": 1515
     },
     {
       "epoch": 0.2715983203788082,
-      "grad_norm": 0.5189542770385742,
+      "grad_norm": 0.520740807056427,
       "learning_rate": 0.00018259709368100962,
-      "loss": 1.2845,
+      "loss": 1.2838,
       "step": 1520
     },
     {
       "epoch": 0.27249173590637005,
-      "grad_norm": 0.5948411226272583,
+      "grad_norm": 0.5902200937271118,
       "learning_rate": 0.00018242086257721515,
-      "loss": 1.2779,
+      "loss": 1.278,
       "step": 1525
     },
     {
       "epoch": 0.27338515143393194,
-      "grad_norm": 0.5104305744171143,
+      "grad_norm": 0.5141775608062744,
       "learning_rate": 0.00018224382960129972,
       "loss": 1.2815,
       "step": 1530
     },
     {
       "epoch": 0.2742785669614938,
-      "grad_norm": 0.5303378701210022,
+      "grad_norm": 0.5365716218948364,
       "learning_rate": 0.00018206599647561627,
-      "loss": 1.3096,
+      "loss": 1.3087,
       "step": 1535
     },
     {
       "epoch": 0.27517198248905567,
-      "grad_norm": 0.5319021344184875,
+      "grad_norm": 0.5323083400726318,
       "learning_rate": 0.00018188736493030222,
-      "loss": 1.2779,
+      "loss": 1.2783,
       "step": 1540
     },
     {
       "epoch": 0.2760653980166175,
-      "grad_norm": 0.5153783559799194,
+      "grad_norm": 0.5199279189109802,
       "learning_rate": 0.00018170793670326292,
-      "loss": 1.2889,
+      "loss": 1.2883,
       "step": 1545
     },
     {
       "epoch": 0.2769588135441794,
-      "grad_norm": 0.5285100936889648,
+      "grad_norm": 0.5308305025100708,
       "learning_rate": 0.0001815277135401546,
-      "loss": 1.2865,
+      "loss": 1.2868,
       "step": 1550
     },
     {
       "epoch": 0.2778522290717413,
-      "grad_norm": 0.5145849585533142,
+      "grad_norm": 0.5134423971176147,
       "learning_rate": 0.00018134669719436749,
-      "loss": 1.2755,
+      "loss": 1.2758,
       "step": 1555
     },
     {
       "epoch": 0.2787456445993031,
-      "grad_norm": 0.541465163230896,
+      "grad_norm": 0.5420312881469727,
       "learning_rate": 0.00018116488942700857,
-      "loss": 1.2888,
+      "loss": 1.2882,
       "step": 1560
     },
     {
       "epoch": 0.279639060126865,
-      "grad_norm": 0.5526989698410034,
+      "grad_norm": 0.5505488514900208,
       "learning_rate": 0.00018098229200688462,
-      "loss": 1.3123,
+      "loss": 1.3117,
       "step": 1565
     },
     {
       "epoch": 0.28053247565442685,
-      "grad_norm": 0.5341644287109375,
+      "grad_norm": 0.5319713950157166,
       "learning_rate": 0.00018079890671048493,
-      "loss": 1.2959,
+      "loss": 1.2958,
       "step": 1570
     },
     {
       "epoch": 0.28142589118198874,
-      "grad_norm": 0.5949883460998535,
+      "grad_norm": 0.5935749411582947,
       "learning_rate": 0.0001806147353219641,
-      "loss": 1.2757,
+      "loss": 1.2753,
       "step": 1575
     },
     {
       "epoch": 0.28231930670955063,
-      "grad_norm": 0.5318828821182251,
+      "grad_norm": 0.5329294204711914,
       "learning_rate": 0.00018042977963312451,
-      "loss": 1.319,
+      "loss": 1.3191,
       "step": 1580
     },
     {
       "epoch": 0.28321272223711247,
-      "grad_norm": 0.514118492603302,
+      "grad_norm": 0.5126677751541138,
       "learning_rate": 0.00018024404144339906,
       "loss": 1.2924,
       "step": 1585
     },
     {
       "epoch": 0.28410613776467436,
-      "grad_norm": 0.5623351335525513,
+      "grad_norm": 0.554897129535675,
       "learning_rate": 0.00018005752255983355,
-      "loss": 1.2712,
+      "loss": 1.2715,
       "step": 1590
     },
     {
       "epoch": 0.2849995532922362,
-      "grad_norm": 0.5316884517669678,
+      "grad_norm": 0.5345707535743713,
       "learning_rate": 0.00017987022479706922,
-      "loss": 1.2994,
+      "loss": 1.2999,
       "step": 1595
     },
     {
       "epoch": 0.2858929688197981,
-      "grad_norm": 0.5517479777336121,
+      "grad_norm": 0.5505250096321106,
       "learning_rate": 0.000179682149977325,
       "loss": 1.3081,
       "step": 1600
     },
     {
       "epoch": 0.28678638434736,
-      "grad_norm": 0.5413886904716492,
+      "grad_norm": 0.5407662987709045,
       "learning_rate": 0.0001794932999303797,
-      "loss": 1.2911,
+      "loss": 1.2921,
       "step": 1605
     },
     {
       "epoch": 0.2876797998749218,
-      "grad_norm": 0.5298170447349548,
+      "grad_norm": 0.5246217846870422,
       "learning_rate": 0.00017930367649355447,
-      "loss": 1.304,
+      "loss": 1.3035,
       "step": 1610
     },
     {
       "epoch": 0.2885732154024837,
-      "grad_norm": 0.6161289215087891,
+      "grad_norm": 0.5454705953598022,
       "learning_rate": 0.00017911328151169466,
-      "loss": 1.3018,
+      "loss": 1.302,
       "step": 1615
     },
     {
       "epoch": 0.28946663093004554,
-      "grad_norm": 0.5742626190185547,
+      "grad_norm": 0.571116030216217,
       "learning_rate": 0.000178922116837152,
-      "loss": 1.2513,
+      "loss": 1.2515,
       "step": 1620
     },
     {
       "epoch": 0.29036004645760743,
-      "grad_norm": 0.5081664323806763,
+      "grad_norm": 0.5103644728660583,
       "learning_rate": 0.00017873018432976658,
-      "loss": 1.2688,
+      "loss": 1.2696,
       "step": 1625
     },
     {
       "epoch": 0.2912534619851693,
-      "grad_norm": 0.550042450428009,
+      "grad_norm": 0.5452694296836853,
       "learning_rate": 0.00017853748585684867,
-      "loss": 1.2895,
+      "loss": 1.2893,
       "step": 1630
     },
     {
       "epoch": 0.29214687751273116,
-      "grad_norm": 0.5372338891029358,
+      "grad_norm": 0.5386608242988586,
       "learning_rate": 0.0001783440232931607,
-      "loss": 1.2525,
+      "loss": 1.2519,
       "step": 1635
     },
     {
       "epoch": 0.29304029304029305,
-      "grad_norm": 0.5075910687446594,
+      "grad_norm": 0.5036607384681702,
       "learning_rate": 0.0001781497985208989,
-      "loss": 1.2852,
+      "loss": 1.2865,
       "step": 1640
     },
     {
       "epoch": 0.2939337085678549,
-      "grad_norm": 0.4970715641975403,
+      "grad_norm": 0.4949742555618286,
       "learning_rate": 0.00017795481342967501,
-      "loss": 1.2175,
+      "loss": 1.218,
       "step": 1645
     },
     {
       "epoch": 0.2948271240954168,
-      "grad_norm": 0.510547399520874,
+      "grad_norm": 0.5089282989501953,
       "learning_rate": 0.0001777590699164979,
-      "loss": 1.3002,
+      "loss": 1.2993,
       "step": 1650
     },
     {
       "epoch": 0.2957205396229787,
-      "grad_norm": 0.5301088094711304,
+      "grad_norm": 0.5262947678565979,
       "learning_rate": 0.00017756256988575513,
-      "loss": 1.2885,
+      "loss": 1.2875,
       "step": 1655
     },
     {
       "epoch": 0.2966139551505405,
-      "grad_norm": 0.5963689684867859,
+      "grad_norm": 0.598348081111908,
       "learning_rate": 0.00017736531524919445,
-      "loss": 1.3018,
+      "loss": 1.3016,
       "step": 1660
     },
     {
       "epoch": 0.2975073706781024,
-      "grad_norm": 0.5285865068435669,
+      "grad_norm": 0.5289075374603271,
       "learning_rate": 0.00017716730792590512,
-      "loss": 1.3158,
+      "loss": 1.3164,
       "step": 1665
     },
     {
       "epoch": 0.29840078620566424,
-      "grad_norm": 0.5783317685127258,
+      "grad_norm": 0.5724149942398071,
       "learning_rate": 0.00017696854984229933,
       "loss": 1.299,
       "step": 1670
     },
     {
       "epoch": 0.29929420173322613,
-      "grad_norm": 0.5652763843536377,
+      "grad_norm": 0.5685451626777649,
       "learning_rate": 0.00017676904293209336,
-      "loss": 1.303,
+      "loss": 1.3032,
       "step": 1675
     },
     {
       "epoch": 0.300187617260788,
-      "grad_norm": 0.5288776755332947,
+      "grad_norm": 0.5210792422294617,
       "learning_rate": 0.00017656878913628883,
-      "loss": 1.2839,
+      "loss": 1.2844,
       "step": 1680
     },
     {
       "epoch": 0.30108103278834986,
-      "grad_norm": 0.5509206056594849,
+      "grad_norm": 0.5431951880455017,
       "learning_rate": 0.0001763677904031539,
-      "loss": 1.2739,
+      "loss": 1.274,
       "step": 1685
     },
     {
       "epoch": 0.30197444831591175,
-      "grad_norm": 0.5611214637756348,
+      "grad_norm": 0.551347553730011,
       "learning_rate": 0.00017616604868820406,
-      "loss": 1.3148,
+      "loss": 1.3147,
       "step": 1690
     },
     {
       "epoch": 0.3028678638434736,
-      "grad_norm": 0.5442464351654053,
+      "grad_norm": 0.54395991563797,
       "learning_rate": 0.0001759635659541834,
-      "loss": 1.2931,
+      "loss": 1.2935,
       "step": 1695
     },
     {
       "epoch": 0.3037612793710355,
-      "grad_norm": 0.5243226885795593,
+      "grad_norm": 0.519129753112793,
       "learning_rate": 0.0001757603441710453,
-      "loss": 1.2174,
+      "loss": 1.2166,
       "step": 1700
     },
     {
       "epoch": 0.3046546948985973,
-      "grad_norm": 0.5281797647476196,
+      "grad_norm": 0.5352473258972168,
       "learning_rate": 0.0001755563853159334,
-      "loss": 1.239,
+      "loss": 1.2378,
       "step": 1705
     },
     {
       "epoch": 0.3055481104261592,
-      "grad_norm": 0.5556020736694336,
+      "grad_norm": 0.5558743476867676,
       "learning_rate": 0.00017535169137316227,
-      "loss": 1.2909,
+      "loss": 1.2911,
       "step": 1710
     },
     {
       "epoch": 0.3064415259537211,
-      "grad_norm": 0.5657395124435425,
+      "grad_norm": 0.5650675892829895,
       "learning_rate": 0.0001751462643341982,
-      "loss": 1.2534,
+      "loss": 1.254,
       "step": 1715
     },
     {
       "epoch": 0.30733494148128293,
-      "grad_norm": 0.5573301315307617,
+      "grad_norm": 0.5582188367843628,
       "learning_rate": 0.0001749401061976397,
-      "loss": 1.2457,
+      "loss": 1.246,
       "step": 1720
     },
     {
       "epoch": 0.3082283570088448,
-      "grad_norm": 0.5255555510520935,
+      "grad_norm": 0.5283738970756531,
       "learning_rate": 0.0001747332189691981,
-      "loss": 1.2453,
+      "loss": 1.246,
       "step": 1725
     },
     {
       "epoch": 0.30912177253640666,
-      "grad_norm": 0.5263189673423767,
+      "grad_norm": 0.5312855839729309,
       "learning_rate": 0.00017452560466167818,
-      "loss": 1.2899,
+      "loss": 1.2898,
       "step": 1730
     },
     {
       "epoch": 0.31001518806396855,
-      "grad_norm": 0.5250783562660217,
+      "grad_norm": 0.5232421159744263,
       "learning_rate": 0.00017431726529495837,
-      "loss": 1.2823,
+      "loss": 1.282,
       "step": 1735
     },
     {
       "epoch": 0.31090860359153044,
-      "grad_norm": 0.5130211114883423,
+      "grad_norm": 0.5114389061927795,
       "learning_rate": 0.00017410820289597126,
-      "loss": 1.2815,
+      "loss": 1.2814,
       "step": 1740
     },
     {
       "epoch": 0.3118020191190923,
-      "grad_norm": 0.4937721788883209,
+      "grad_norm": 0.49061334133148193,
       "learning_rate": 0.00017389841949868378,
-      "loss": 1.2589,
+      "loss": 1.2584,
       "step": 1745
     },
     {
       "epoch": 0.31269543464665417,
-      "grad_norm": 0.5342549681663513,
+      "grad_norm": 0.531428873538971,
       "learning_rate": 0.0001736879171440774,
-      "loss": 1.2599,
+      "loss": 1.2596,
       "step": 1750
     },
     {
       "epoch": 0.313588850174216,
-      "grad_norm": 0.5212100148200989,
+      "grad_norm": 0.5237516164779663,
       "learning_rate": 0.00017347669788012846,
-      "loss": 1.221,
+      "loss": 1.2213,
       "step": 1755
     },
     {
       "epoch": 0.3144822657017779,
-      "grad_norm": 0.4915301203727722,
+      "grad_norm": 0.4916597306728363,
       "learning_rate": 0.00017326476376178796,
-      "loss": 1.2569,
+      "loss": 1.2573,
       "step": 1760
     },
     {
       "epoch": 0.3153756812293398,
-      "grad_norm": 0.5691230297088623,
+      "grad_norm": 0.5719464421272278,
       "learning_rate": 0.00017305211685096178,
-      "loss": 1.2304,
+      "loss": 1.2306,
       "step": 1765
     },
     {
       "epoch": 0.3162690967569016,
-      "grad_norm": 0.5791122317314148,
+      "grad_norm": 0.5756252408027649,
       "learning_rate": 0.00017283875921649057,
-      "loss": 1.284,
+      "loss": 1.2842,
       "step": 1770
     },
     {
       "epoch": 0.3171625122844635,
-      "grad_norm": 0.5447010397911072,
+      "grad_norm": 0.544870913028717,
       "learning_rate": 0.0001726246929341296,
-      "loss": 1.2642,
+      "loss": 1.264,
       "step": 1775
     },
     {
       "epoch": 0.31805592781202535,
-      "grad_norm": 0.49899354577064514,
+      "grad_norm": 0.4986020028591156,
       "learning_rate": 0.00017240992008652855,
-      "loss": 1.2926,
+      "loss": 1.2919,
       "step": 1780
     },
     {
       "epoch": 0.31894934333958724,
-      "grad_norm": 0.541107714176178,
+      "grad_norm": 0.5383627414703369,
       "learning_rate": 0.00017219444276321127,
-      "loss": 1.2677,
+      "loss": 1.2679,
       "step": 1785
     },
     {
       "epoch": 0.31984275886714914,
-      "grad_norm": 0.563939094543457,
+      "grad_norm": 0.5600801706314087,
       "learning_rate": 0.00017197826306055547,
-      "loss": 1.2592,
+      "loss": 1.2589,
       "step": 1790
     },
     {
       "epoch": 0.32073617439471097,
-      "grad_norm": 0.538004994392395,
+      "grad_norm": 0.5386105179786682,
       "learning_rate": 0.0001717613830817723,
-      "loss": 1.2802,
+      "loss": 1.2812,
       "step": 1795
     },
     {
       "epoch": 0.32162958992227286,
-      "grad_norm": 0.5084322690963745,
+      "grad_norm": 0.5082394480705261,
       "learning_rate": 0.00017154380493688583,
-      "loss": 1.2718,
+      "loss": 1.2717,
       "step": 1800
     },
     {
       "epoch": 0.3225230054498347,
-      "grad_norm": 0.5454220175743103,
+      "grad_norm": 0.5480278134346008,
       "learning_rate": 0.00017132553074271272,
-      "loss": 1.2522,
+      "loss": 1.2521,
       "step": 1805
     },
     {
       "epoch": 0.3234164209773966,
-      "grad_norm": 0.5236728191375732,
+      "grad_norm": 0.522881269454956,
       "learning_rate": 0.00017110656262284135,
-      "loss": 1.2325,
+      "loss": 1.233,
       "step": 1810
     },
     {
       "epoch": 0.3243098365049585,
-      "grad_norm": 0.5248743295669556,
+      "grad_norm": 0.5211268663406372,
       "learning_rate": 0.0001708869027076114,
-      "loss": 1.2443,
+      "loss": 1.2448,
       "step": 1815
     },
     {
       "epoch": 0.3252032520325203,
-      "grad_norm": 0.5419241786003113,
+      "grad_norm": 0.5418733954429626,
       "learning_rate": 0.00017066655313409295,
-      "loss": 1.2359,
+      "loss": 1.2354,
       "step": 1820
     },
     {
       "epoch": 0.3260966675600822,
-      "grad_norm": 0.5280671715736389,
+      "grad_norm": 0.524122416973114,
       "learning_rate": 0.00017044551604606585,
-      "loss": 1.2639,
+      "loss": 1.2641,
       "step": 1825
     },
     {
       "epoch": 0.32699008308764405,
-      "grad_norm": 0.5182398557662964,
+      "grad_norm": 0.5205156207084656,
       "learning_rate": 0.00017022379359399868,
-      "loss": 1.2303,
+      "loss": 1.2312,
       "step": 1830
     },
     {
       "epoch": 0.32788349861520594,
-      "grad_norm": 0.5220509171485901,
+      "grad_norm": 0.5240566730499268,
       "learning_rate": 0.00017000138793502796,
-      "loss": 1.2605,
+      "loss": 1.2603,
       "step": 1835
     },
     {
       "epoch": 0.32877691414276783,
-      "grad_norm": 0.5644056797027588,
+      "grad_norm": 0.5625032186508179,
       "learning_rate": 0.00016977830123293713,
-      "loss": 1.2321,
+      "loss": 1.2308,
       "step": 1840
     },
     {
       "epoch": 0.32967032967032966,
-      "grad_norm": 0.5633097887039185,
+      "grad_norm": 0.5633267164230347,
       "learning_rate": 0.00016955453565813548,
-      "loss": 1.2882,
+      "loss": 1.287,
       "step": 1845
     },
     {
       "epoch": 0.33056374519789156,
-      "grad_norm": 0.5087893605232239,
+      "grad_norm": 0.5072932243347168,
       "learning_rate": 0.0001693300933876371,
-      "loss": 1.2588,
+      "loss": 1.259,
       "step": 1850
     },
     {
       "epoch": 0.3314571607254534,
-      "grad_norm": 0.5339887738227844,
+      "grad_norm": 0.532630205154419,
       "learning_rate": 0.00016910497660503957,
-      "loss": 1.2356,
+      "loss": 1.2352,
       "step": 1855
     },
     {
       "epoch": 0.3323505762530153,
-      "grad_norm": 0.5123136043548584,
+      "grad_norm": 0.5094119310379028,
       "learning_rate": 0.00016887918750050292,
       "loss": 1.2672,
       "step": 1860
     },
     {
       "epoch": 0.3332439917805771,
-      "grad_norm": 0.5174533128738403,
+      "grad_norm": 0.5230539441108704,
       "learning_rate": 0.00016865272827072797,
       "loss": 1.2506,
       "step": 1865
     },
     {
       "epoch": 0.334137407308139,
-      "grad_norm": 0.4912176728248596,
+      "grad_norm": 0.4917963445186615,
       "learning_rate": 0.00016842560111893543,
-      "loss": 1.2689,
+      "loss": 1.2685,
       "step": 1870
     },
     {
       "epoch": 0.3350308228357009,
-      "grad_norm": 0.5270506739616394,
+      "grad_norm": 0.5256093144416809,
       "learning_rate": 0.00016819780825484413,
-      "loss": 1.2343,
+      "loss": 1.2344,
       "step": 1875
     },
     {
       "epoch": 0.33592423836326274,
-      "grad_norm": 0.568572461605072,
+      "grad_norm": 0.5623132586479187,
       "learning_rate": 0.00016796935189464956,
-      "loss": 1.2454,
+      "loss": 1.246,
       "step": 1880
     },
     {
       "epoch": 0.33681765389082463,
-      "grad_norm": 0.5541931390762329,
+      "grad_norm": 0.5536134839057922,
       "learning_rate": 0.00016774023426100238,
-      "loss": 1.2434,
+      "loss": 1.2428,
       "step": 1885
     },
     {
       "epoch": 0.33771106941838647,
-      "grad_norm": 0.5010518431663513,
+      "grad_norm": 0.4987887442111969,
       "learning_rate": 0.0001675104575829868,
-      "loss": 1.2402,
+      "loss": 1.2414,
       "step": 1890
     },
     {
       "epoch": 0.33860448494594836,
-      "grad_norm": 0.5561944246292114,
+      "grad_norm": 0.5539235472679138,
       "learning_rate": 0.0001672800240960989,
-      "loss": 1.2616,
+      "loss": 1.262,
       "step": 1895
     },
     {
       "epoch": 0.33949790047351025,
-      "grad_norm": 0.5340306758880615,
+      "grad_norm": 0.5383912920951843,
       "learning_rate": 0.00016704893604222476,
-      "loss": 1.2949,
+      "loss": 1.2957,
       "step": 1900
     },
     {
       "epoch": 0.3403913160010721,
-      "grad_norm": 0.5372459888458252,
+      "grad_norm": 0.5472669005393982,
       "learning_rate": 0.00016681719566961887,
-      "loss": 1.2721,
+      "loss": 1.2723,
       "step": 1905
     },
     {
       "epoch": 0.341284731528634,
-      "grad_norm": 0.5558496117591858,
+      "grad_norm": 0.5573126673698425,
       "learning_rate": 0.00016658480523288195,
-      "loss": 1.2146,
+      "loss": 1.214,
       "step": 1910
     },
     {
       "epoch": 0.3421781470561958,
-      "grad_norm": 0.523130476474762,
+      "grad_norm": 0.5227344632148743,
       "learning_rate": 0.00016635176699293934,
-      "loss": 1.2607,
+      "loss": 1.2613,
       "step": 1915
     },
     {
       "epoch": 0.3430715625837577,
-      "grad_norm": 0.5020752549171448,
+      "grad_norm": 0.5032819509506226,
       "learning_rate": 0.00016611808321701882,
-      "loss": 1.2572,
+      "loss": 1.2575,
       "step": 1920
     },
     {
       "epoch": 0.3439649781113196,
-      "grad_norm": 0.52286297082901,
+      "grad_norm": 0.5216634273529053,
       "learning_rate": 0.00016588375617862858,
-      "loss": 1.2249,
+      "loss": 1.2253,
       "step": 1925
     },
     {
       "epoch": 0.34485839363888143,
-      "grad_norm": 0.5461264848709106,
+      "grad_norm": 0.5421084761619568,
       "learning_rate": 0.000165648788157535,
-      "loss": 1.2896,
+      "loss": 1.2899,
       "step": 1930
     },
     {
       "epoch": 0.3457518091664433,
-      "grad_norm": 0.5276172161102295,
+      "grad_norm": 0.5299046039581299,
       "learning_rate": 0.00016541318143974075,
       "loss": 1.263,
       "step": 1935
     },
     {
       "epoch": 0.34664522469400516,
-      "grad_norm": 0.5360847115516663,
+      "grad_norm": 0.5358394980430603,
       "learning_rate": 0.00016517693831746225,
-      "loss": 1.2618,
+      "loss": 1.2614,
       "step": 1940
     },
     {
       "epoch": 0.34753864022156705,
-      "grad_norm": 0.5737685561180115,
+      "grad_norm": 0.570563018321991,
       "learning_rate": 0.00016494006108910757,
-      "loss": 1.2572,
+      "loss": 1.257,
       "step": 1945
     },
     {
       "epoch": 0.34843205574912894,
-      "grad_norm": 0.4999217987060547,
+      "grad_norm": 0.4944029450416565,
       "learning_rate": 0.000164702552059254,
-      "loss": 1.2388,
+      "loss": 1.2376,
       "step": 1950
     },
     {
       "epoch": 0.3493254712766908,
-      "grad_norm": 0.5562167763710022,
+      "grad_norm": 0.5532763004302979,
       "learning_rate": 0.00016446441353862556,
-      "loss": 1.3043,
+      "loss": 1.3048,
       "step": 1955
     },
     {
       "epoch": 0.35021888680425267,
-      "grad_norm": 0.542860209941864,
+      "grad_norm": 0.5413596034049988,
       "learning_rate": 0.0001642256478440706,
-      "loss": 1.2583,
+      "loss": 1.2578,
       "step": 1960
     },
     {
       "epoch": 0.3511123023318145,
-      "grad_norm": 0.5292645692825317,
+      "grad_norm": 0.5195985436439514,
       "learning_rate": 0.00016398625729853924,
-      "loss": 1.2556,
+      "loss": 1.2549,
       "step": 1965
     },
     {
       "epoch": 0.3520057178593764,
-      "grad_norm": 0.5102574229240417,
+      "grad_norm": 0.5096262097358704,
       "learning_rate": 0.00016374624423106087,
-      "loss": 1.2536,
+      "loss": 1.2538,
       "step": 1970
     },
     {
       "epoch": 0.3528991333869383,
-      "grad_norm": 0.5271995067596436,
+      "grad_norm": 0.5297495126724243,
       "learning_rate": 0.00016350561097672122,
-      "loss": 1.2147,
+      "loss": 1.2156,
       "step": 1975
     },
     {
       "epoch": 0.3537925489145001,
-      "grad_norm": 0.5228263735771179,
+      "grad_norm": 0.523977518081665,
       "learning_rate": 0.00016326435987663995,
-      "loss": 1.247,
+      "loss": 1.2471,
       "step": 1980
     },
     {
       "epoch": 0.354685964442062,
-      "grad_norm": 0.5432056188583374,
+      "grad_norm": 0.5540353059768677,
       "learning_rate": 0.0001630224932779477,
-      "loss": 1.2328,
+      "loss": 1.232,
       "step": 1985
     },
     {
       "epoch": 0.35557937996962385,
-      "grad_norm": 0.5798959732055664,
+      "grad_norm": 0.5836973190307617,
       "learning_rate": 0.00016278001353376323,
-      "loss": 1.2105,
+      "loss": 1.2102,
       "step": 1990
     },
     {
       "epoch": 0.35647279549718575,
-      "grad_norm": 0.5255203247070312,
+      "grad_norm": 0.52564537525177,
       "learning_rate": 0.0001625369230031707,
-      "loss": 1.2412,
+      "loss": 1.2415,
       "step": 1995
     },
     {
       "epoch": 0.35736621102474764,
-      "grad_norm": 0.5129863619804382,
+      "grad_norm": 0.512956976890564,
       "learning_rate": 0.00016229322405119655,
-      "loss": 1.2291,
+      "loss": 1.2297,
       "step": 2000
     },
     {
       "epoch": 0.3582596265523095,
-      "grad_norm": 0.5471330881118774,
+      "grad_norm": 0.5465174317359924,
       "learning_rate": 0.00016204891904878657,
-      "loss": 1.2397,
+      "loss": 1.2394,
       "step": 2005
     },
     {
       "epoch": 0.35915304207987137,
-      "grad_norm": 0.4979896545410156,
+      "grad_norm": 0.502322793006897,
       "learning_rate": 0.0001618040103727827,
-      "loss": 1.2312,
+      "loss": 1.2317,
       "step": 2010
     },
     {
       "epoch": 0.3600464576074332,
-      "grad_norm": 0.5342873930931091,
+      "grad_norm": 0.5260788202285767,
       "learning_rate": 0.00016155850040590016,
-      "loss": 1.2454,
+      "loss": 1.2446,
       "step": 2015
     },
     {
       "epoch": 0.3609398731349951,
-      "grad_norm": 0.5229947566986084,
+      "grad_norm": 0.5212415456771851,
       "learning_rate": 0.0001613123915367041,
-      "loss": 1.2417,
+      "loss": 1.2422,
       "step": 2020
     },
     {
       "epoch": 0.36183328866255693,
-      "grad_norm": 0.5110452175140381,
+      "grad_norm": 0.5056548714637756,
       "learning_rate": 0.00016106568615958632,
-      "loss": 1.1992,
+      "loss": 1.1996,
       "step": 2025
     },
     {
       "epoch": 0.3627267041901188,
-      "grad_norm": 0.5229007601737976,
+      "grad_norm": 0.52322918176651,
       "learning_rate": 0.00016081838667474213,
-      "loss": 1.2174,
+      "loss": 1.2173,
       "step": 2030
     },
     {
       "epoch": 0.3636201197176807,
-      "grad_norm": 0.5256268978118896,
+      "grad_norm": 0.5279656052589417,
       "learning_rate": 0.0001605704954881468,
-      "loss": 1.2365,
+      "loss": 1.2358,
       "step": 2035
     },
     {
       "epoch": 0.36451353524524255,
-      "grad_norm": 0.5547985434532166,
+      "grad_norm": 0.5513635277748108,
       "learning_rate": 0.00016032201501153242,
-      "loss": 1.2282,
+      "loss": 1.2278,
       "step": 2040
     },
     {
       "epoch": 0.36540695077280444,
-      "grad_norm": 0.5028260946273804,
+      "grad_norm": 0.5040432214736938,
       "learning_rate": 0.00016007294766236406,
-      "loss": 1.2256,
+      "loss": 1.2253,
       "step": 2045
     },
     {
       "epoch": 0.3663003663003663,
-      "grad_norm": 0.48005393147468567,
+      "grad_norm": 0.4764319062232971,
       "learning_rate": 0.00015982329586381675,
-      "loss": 1.2313,
+      "loss": 1.2304,
       "step": 2050
     },
     {
       "epoch": 0.36719378182792817,
-      "grad_norm": 0.48138970136642456,
+      "grad_norm": 0.4786640703678131,
       "learning_rate": 0.00015957306204475132,
-      "loss": 1.2547,
+      "loss": 1.2555,
       "step": 2055
     },
     {
       "epoch": 0.36808719735549006,
-      "grad_norm": 0.5655422210693359,
+      "grad_norm": 0.5668711066246033,
       "learning_rate": 0.00015932224863969135,
-      "loss": 1.2304,
+      "loss": 1.2294,
       "step": 2060
     },
     {
       "epoch": 0.3689806128830519,
-      "grad_norm": 0.514876127243042,
+      "grad_norm": 0.5136781930923462,
       "learning_rate": 0.000159070858088799,
-      "loss": 1.1753,
+      "loss": 1.1752,
       "step": 2065
     },
     {
       "epoch": 0.3698740284106138,
-      "grad_norm": 0.5176100134849548,
+      "grad_norm": 0.5189619660377502,
       "learning_rate": 0.0001588188928378516,
-      "loss": 1.2343,
+      "loss": 1.2351,
       "step": 2070
     },
     {
       "epoch": 0.3707674439381756,
-      "grad_norm": 0.5761798024177551,
+      "grad_norm": 0.5765734910964966,
       "learning_rate": 0.00015856635533821774,
-      "loss": 1.2001,
+      "loss": 1.2,
       "step": 2075
     },
     {
       "epoch": 0.3716608594657375,
-      "grad_norm": 0.5333663821220398,
+      "grad_norm": 0.5305128693580627,
       "learning_rate": 0.00015831324804683328,
-      "loss": 1.2414,
+      "loss": 1.241,
       "step": 2080
     },
     {
       "epoch": 0.3725542749932994,
-      "grad_norm": 0.5275282859802246,
+      "grad_norm": 0.5297000408172607,
       "learning_rate": 0.0001580595734261777,
-      "loss": 1.1956,
+      "loss": 1.1959,
       "step": 2085
     },
     {
       "epoch": 0.37344769052086124,
-      "grad_norm": 0.5324888825416565,
+      "grad_norm": 0.5325464010238647,
       "learning_rate": 0.00015780533394425006,
-      "loss": 1.2218,
+      "loss": 1.2221,
       "step": 2090
     },
     {
       "epoch": 0.37434110604842313,
-      "grad_norm": 0.5113807916641235,
+      "grad_norm": 0.5105712413787842,
       "learning_rate": 0.00015755053207454483,
-      "loss": 1.2431,
+      "loss": 1.2422,
       "step": 2095
     },
     {
       "epoch": 0.37523452157598497,
-      "grad_norm": 0.5269024968147278,
+      "grad_norm": 0.5255334973335266,
       "learning_rate": 0.00015729517029602802,
-      "loss": 1.2333,
+      "loss": 1.2326,
       "step": 2100
     },
     {
       "epoch": 0.37612793710354686,
-      "grad_norm": 0.538032054901123,
+      "grad_norm": 0.5344464778900146,
       "learning_rate": 0.00015703925109311295,
-      "loss": 1.2178,
+      "loss": 1.2188,
       "step": 2105
     },
     {
       "epoch": 0.37702135263110875,
-      "grad_norm": 0.5196306109428406,
+      "grad_norm": 0.5103366374969482,
       "learning_rate": 0.00015678277695563617,
-      "loss": 1.2357,
+      "loss": 1.2354,
       "step": 2110
     },
     {
       "epoch": 0.3779147681586706,
-      "grad_norm": 0.543264627456665,
+      "grad_norm": 0.5386320948600769,
       "learning_rate": 0.00015652575037883318,
-      "loss": 1.1962,
+      "loss": 1.1967,
       "step": 2115
     },
     {
       "epoch": 0.3788081836862325,
-      "grad_norm": 0.5403855443000793,
+      "grad_norm": 0.5370935201644897,
       "learning_rate": 0.0001562681738633141,
-      "loss": 1.2599,
+      "loss": 1.2594,
       "step": 2120
     },
     {
       "epoch": 0.3797015992137943,
-      "grad_norm": 0.5239453911781311,
+      "grad_norm": 0.5193549990653992,
       "learning_rate": 0.00015601004991503946,
       "loss": 1.2144,
       "step": 2125
     },
     {
       "epoch": 0.3805950147413562,
-      "grad_norm": 0.5659239292144775,
+      "grad_norm": 0.5521916151046753,
       "learning_rate": 0.00015575138104529577,
-      "loss": 1.201,
+      "loss": 1.2009,
       "step": 2130
     },
     {
       "epoch": 0.3814884302689181,
-      "grad_norm": 0.4925094246864319,
+      "grad_norm": 0.4977095425128937,
       "learning_rate": 0.00015549216977067099,
-      "loss": 1.2219,
+      "loss": 1.2222,
       "step": 2135
     },
     {
       "epoch": 0.38238184579647994,
-      "grad_norm": 0.4957289695739746,
+      "grad_norm": 0.4930327832698822,
       "learning_rate": 0.0001552324186130302,
-      "loss": 1.2577,
+      "loss": 1.2569,
       "step": 2140
     },
     {
       "epoch": 0.3832752613240418,
-      "grad_norm": 0.5230002999305725,
+      "grad_norm": 0.5264571905136108,
       "learning_rate": 0.00015497213009949104,
-      "loss": 1.2017,
+      "loss": 1.2016,
       "step": 2145
     },
     {
       "epoch": 0.38416867685160366,
-      "grad_norm": 0.5107108354568481,
+      "grad_norm": 0.505418598651886,
       "learning_rate": 0.000154711306762399,
       "loss": 1.2115,
       "step": 2150
     },
     {
       "epoch": 0.38506209237916555,
-      "grad_norm": 0.5467280745506287,
+      "grad_norm": 0.5414779186248779,
       "learning_rate": 0.00015444995113930287,
-      "loss": 1.2443,
+      "loss": 1.2452,
       "step": 2155
     },
     {
       "epoch": 0.38595550790672745,
-      "grad_norm": 0.5325524806976318,
+      "grad_norm": 0.534766435623169,
       "learning_rate": 0.00015418806577293013,
-      "loss": 1.2103,
+      "loss": 1.2098,
       "step": 2160
     },
     {
       "epoch": 0.3868489234342893,
-      "grad_norm": 0.5066697001457214,
+      "grad_norm": 0.5061149001121521,
       "learning_rate": 0.00015392565321116207,
-      "loss": 1.2157,
+      "loss": 1.2156,
       "step": 2165
     },
     {
       "epoch": 0.3877423389618512,
-      "grad_norm": 0.5199021100997925,
+      "grad_norm": 0.5177295207977295,
       "learning_rate": 0.00015366271600700902,
-      "loss": 1.2284,
+      "loss": 1.2279,
       "step": 2170
     },
     {
       "epoch": 0.388635754489413,
-      "grad_norm": 0.557767391204834,
+      "grad_norm": 0.5544948577880859,
       "learning_rate": 0.00015339925671858563,
-      "loss": 1.2144,
+      "loss": 1.2138,
       "step": 2175
     },
     {
       "epoch": 0.3895291700169749,
-      "grad_norm": 0.5319749116897583,
+      "grad_norm": 0.531894862651825,
       "learning_rate": 0.0001531352779090859,
       "loss": 1.1939,
       "step": 2180
     },
     {
       "epoch": 0.39042258554453674,
-      "grad_norm": 0.5209249258041382,
+      "grad_norm": 0.5233749151229858,
       "learning_rate": 0.00015287078214675819,
-      "loss": 1.2352,
+      "loss": 1.2357,
       "step": 2185
     },
     {
       "epoch": 0.39131600107209863,
-      "grad_norm": 0.5308728218078613,
+      "grad_norm": 0.5275238156318665,
       "learning_rate": 0.00015260577200488034,
-      "loss": 1.232,
+      "loss": 1.2322,
       "step": 2190
     },
     {
       "epoch": 0.3922094165996605,
-      "grad_norm": 0.5475233793258667,
+      "grad_norm": 0.5474215745925903,
       "learning_rate": 0.00015234025006173452,
-      "loss": 1.2255,
+      "loss": 1.2265,
       "step": 2195
     },
     {
       "epoch": 0.39310283212722236,
-      "grad_norm": 0.5130513310432434,
+      "grad_norm": 0.5120342373847961,
       "learning_rate": 0.00015207421890058237,
-      "loss": 1.2169,
+      "loss": 1.216,
       "step": 2200
     },
     {
       "epoch": 0.39399624765478425,
-      "grad_norm": 0.5458916425704956,
+      "grad_norm": 0.5480685830116272,
       "learning_rate": 0.0001518076811096395,
-      "loss": 1.206,
+      "loss": 1.2065,
       "step": 2205
     },
     {
       "epoch": 0.3948896631823461,
-      "grad_norm": 0.4863106608390808,
+      "grad_norm": 0.47449925541877747,
       "learning_rate": 0.00015154063928205067,
-      "loss": 1.1518,
+      "loss": 1.1511,
       "step": 2210
     },
     {
       "epoch": 0.395783078709908,
-      "grad_norm": 0.494357168674469,
+      "grad_norm": 0.49476173520088196,
       "learning_rate": 0.00015127309601586434,
-      "loss": 1.1735,
+      "loss": 1.1736,
       "step": 2215
     },
     {
       "epoch": 0.39667649423746987,
-      "grad_norm": 0.5145966410636902,
+      "grad_norm": 0.513217568397522,
       "learning_rate": 0.0001510050539140075,
-      "loss": 1.2215,
+      "loss": 1.2213,
       "step": 2220
     },
     {
       "epoch": 0.3975699097650317,
-      "grad_norm": 0.5166826844215393,
+      "grad_norm": 0.5108741521835327,
       "learning_rate": 0.00015073651558426026,
-      "loss": 1.2375,
+      "loss": 1.2367,
       "step": 2225
     },
     {
       "epoch": 0.3984633252925936,
-      "grad_norm": 0.5195355415344238,
+      "grad_norm": 0.5116496682167053,
       "learning_rate": 0.00015046748363923057,
-      "loss": 1.2458,
+      "loss": 1.2457,
       "step": 2230
     },
     {
       "epoch": 0.39935674082015543,
-      "grad_norm": 0.5170833468437195,
+      "grad_norm": 0.5216470956802368,
       "learning_rate": 0.00015019796069632878,
       "loss": 1.2356,
       "step": 2235
     },
     {
       "epoch": 0.4002501563477173,
-      "grad_norm": 0.5438719987869263,
+      "grad_norm": 0.5476000308990479,
       "learning_rate": 0.00014992794937774211,
-      "loss": 1.2214,
+      "loss": 1.2221,
       "step": 2240
     },
     {
       "epoch": 0.4011435718752792,
-      "grad_norm": 0.510884165763855,
+      "grad_norm": 0.5119903683662415,
       "learning_rate": 0.00014965745231040919,
-      "loss": 1.2497,
+      "loss": 1.2492,
       "step": 2245
     },
     {
       "epoch": 0.40203698740284105,
-      "grad_norm": 0.5298115611076355,
+      "grad_norm": 0.5329665541648865,
       "learning_rate": 0.00014938647212599452,
-      "loss": 1.2133,
+      "loss": 1.2128,
       "step": 2250
     },
     {
       "epoch": 0.40293040293040294,
-      "grad_norm": 0.5149961709976196,
+      "grad_norm": 0.5195185542106628,
       "learning_rate": 0.00014911501146086281,
-      "loss": 1.217,
+      "loss": 1.2172,
       "step": 2255
     },
     {
       "epoch": 0.4038238184579648,
-      "grad_norm": 0.49967697262763977,
+      "grad_norm": 0.4988158345222473,
       "learning_rate": 0.00014884307295605343,
-      "loss": 1.2235,
+      "loss": 1.2229,
       "step": 2260
     },
     {
       "epoch": 0.40471723398552667,
-      "grad_norm": 0.5198853015899658,
+      "grad_norm": 0.5173177719116211,
       "learning_rate": 0.00014857065925725452,
-      "loss": 1.197,
+      "loss": 1.1982,
       "step": 2265
     },
     {
       "epoch": 0.40561064951308856,
-      "grad_norm": 0.5316833853721619,
+      "grad_norm": 0.5261069536209106,
       "learning_rate": 0.0001482977730147776,
-      "loss": 1.2051,
+      "loss": 1.2049,
       "step": 2270
     },
     {
       "epoch": 0.4065040650406504,
-      "grad_norm": 0.5476366877555847,
+      "grad_norm": 0.5446577072143555,
       "learning_rate": 0.00014802441688353127,
-      "loss": 1.2368,
+      "loss": 1.2377,
       "step": 2275
     },
     {
       "epoch": 0.4073974805682123,
-      "grad_norm": 0.523506224155426,
+      "grad_norm": 0.5223033428192139,
       "learning_rate": 0.00014775059352299598,
-      "loss": 1.1973,
+      "loss": 1.1983,
       "step": 2280
     },
     {
       "epoch": 0.4082908960957741,
-      "grad_norm": 0.5155140161514282,
+      "grad_norm": 0.5107192397117615,
       "learning_rate": 0.00014747630559719762,
-      "loss": 1.2026,
+      "loss": 1.2022,
       "step": 2285
     },
     {
       "epoch": 0.409184311623336,
-      "grad_norm": 0.5298252701759338,
+      "grad_norm": 0.5348856449127197,
       "learning_rate": 0.00014720155577468193,
-      "loss": 1.1871,
+      "loss": 1.1886,
       "step": 2290
     },
     {
       "epoch": 0.4100777271508979,
-      "grad_norm": 0.5185467004776001,
+      "grad_norm": 0.5229494571685791,
       "learning_rate": 0.00014692634672848847,
-      "loss": 1.208,
+      "loss": 1.2079,
       "step": 2295
     },
     {
       "epoch": 0.41097114267845974,
-      "grad_norm": 0.4993501603603363,
+      "grad_norm": 0.49456527829170227,
       "learning_rate": 0.00014665068113612449,
-      "loss": 1.213,
+      "loss": 1.2135,
       "step": 2300
     },
     {
       "epoch": 0.41186455820602164,
-      "grad_norm": 0.5336927175521851,
+      "grad_norm": 0.5290020108222961,
       "learning_rate": 0.00014637456167953907,
-      "loss": 1.2232,
+      "loss": 1.2236,
       "step": 2305
     },
     {
       "epoch": 0.41275797373358347,
-      "grad_norm": 0.4852610230445862,
+      "grad_norm": 0.4855465888977051,
       "learning_rate": 0.00014609799104509685,
-      "loss": 1.2055,
+      "loss": 1.2053,
       "step": 2310
     },
     {
       "epoch": 0.41365138926114536,
-      "grad_norm": 0.5042843222618103,
+      "grad_norm": 0.5055922269821167,
       "learning_rate": 0.00014582097192355207,
-      "loss": 1.1969,
+      "loss": 1.1958,
       "step": 2315
     },
     {
       "epoch": 0.41454480478870726,
-      "grad_norm": 0.5343486070632935,
+      "grad_norm": 0.5334281325340271,
       "learning_rate": 0.00014554350701002222,
-      "loss": 1.1814,
+      "loss": 1.1807,
       "step": 2320
     },
     {
       "epoch": 0.4154382203162691,
-      "grad_norm": 0.5457481741905212,
+      "grad_norm": 0.5472596287727356,
       "learning_rate": 0.00014526559900396188,
-      "loss": 1.1891,
+      "loss": 1.1889,
       "step": 2325
     },
     {
       "epoch": 0.416331635843831,
-      "grad_norm": 0.5025014281272888,
+      "grad_norm": 0.5031034350395203,
       "learning_rate": 0.00014498725060913662,
       "loss": 1.2334,
       "step": 2330
     },
     {
       "epoch": 0.4172250513713928,
-      "grad_norm": 0.5562976002693176,
+      "grad_norm": 0.5575581789016724,
       "learning_rate": 0.00014470846453359636,
-      "loss": 1.1992,
+      "loss": 1.199,
       "step": 2335
     },
     {
       "epoch": 0.4181184668989547,
-      "grad_norm": 0.5212603807449341,
+      "grad_norm": 0.5209872126579285,
       "learning_rate": 0.00014442924348964938,
-      "loss": 1.2133,
+      "loss": 1.2144,
       "step": 2340
     },
     {
       "epoch": 0.41901188242651655,
-      "grad_norm": 0.49604693055152893,
+      "grad_norm": 0.5038667321205139,
       "learning_rate": 0.00014414959019383564,
-      "loss": 1.2171,
+      "loss": 1.2162,
       "step": 2345
     },
     {
       "epoch": 0.41990529795407844,
-      "grad_norm": 0.5236498713493347,
+      "grad_norm": 0.5263863801956177,
       "learning_rate": 0.00014386950736690053,
-      "loss": 1.2163,
+      "loss": 1.2165,
       "step": 2350
     },
     {
       "epoch": 0.42079871348164033,
-      "grad_norm": 0.5673661828041077,
+      "grad_norm": 0.5653106570243835,
       "learning_rate": 0.00014358899773376832,
-      "loss": 1.1914,
+      "loss": 1.1906,
       "step": 2355
     },
     {
       "epoch": 0.42169212900920217,
-      "grad_norm": 0.49621349573135376,
+      "grad_norm": 0.49144411087036133,
       "learning_rate": 0.00014330806402351574,
-      "loss": 1.188,
+      "loss": 1.1886,
       "step": 2360
     },
     {
       "epoch": 0.42258554453676406,
-      "grad_norm": 0.5138354897499084,
+      "grad_norm": 0.5124874114990234,
       "learning_rate": 0.00014302670896934532,
-      "loss": 1.2169,
+      "loss": 1.2159,
       "step": 2365
     },
     {
       "epoch": 0.4234789600643259,
-      "grad_norm": 0.5443947315216064,
+      "grad_norm": 0.5377556085586548,
       "learning_rate": 0.00014274493530855878,
-      "loss": 1.2049,
+      "loss": 1.2051,
       "step": 2370
     },
     {
       "epoch": 0.4243723755918878,
-      "grad_norm": 0.5496159195899963,
+      "grad_norm": 0.5524542331695557,
       "learning_rate": 0.00014246274578253059,
-      "loss": 1.2084,
+      "loss": 1.2096,
       "step": 2375
     },
     {
       "epoch": 0.4252657911194497,
-      "grad_norm": 0.5288496017456055,
+      "grad_norm": 0.5271286368370056,
       "learning_rate": 0.00014218014313668105,
-      "loss": 1.2225,
+      "loss": 1.222,
       "step": 2380
     },
     {
       "epoch": 0.4261592066470115,
-      "grad_norm": 0.5264514088630676,
+      "grad_norm": 0.5286676287651062,
       "learning_rate": 0.00014189713012044977,
-      "loss": 1.2164,
+      "loss": 1.2163,
       "step": 2385
     },
     {
       "epoch": 0.4270526221745734,
-      "grad_norm": 0.5251997113227844,
+      "grad_norm": 0.5305603742599487,
       "learning_rate": 0.00014161370948726894,
-      "loss": 1.1923,
+      "loss": 1.1911,
       "step": 2390
     },
     {
       "epoch": 0.42794603770213524,
-      "grad_norm": 0.5069969296455383,
+      "grad_norm": 0.5002890825271606,
       "learning_rate": 0.00014132988399453617,
-      "loss": 1.2421,
+      "loss": 1.242,
       "step": 2395
     },
     {
       "epoch": 0.42883945322969713,
-      "grad_norm": 0.5222852826118469,
+      "grad_norm": 0.5160467028617859,
       "learning_rate": 0.00014104565640358824,
-      "loss": 1.1921,
+      "loss": 1.1924,
       "step": 2400
     },
     {
       "epoch": 0.429732868757259,
-      "grad_norm": 0.5466008186340332,
+      "grad_norm": 0.5422557592391968,
       "learning_rate": 0.0001407610294796738,
-      "loss": 1.1934,
+      "loss": 1.193,
       "step": 2405
     },
     {
       "epoch": 0.43062628428482086,
-      "grad_norm": 0.5061824917793274,
+      "grad_norm": 0.5061042904853821,
       "learning_rate": 0.00014047600599192666,
-      "loss": 1.1836,
+      "loss": 1.1835,
       "step": 2410
     },
     {
       "epoch": 0.43151969981238275,
-      "grad_norm": 0.5115388631820679,
+      "grad_norm": 0.507078230381012,
       "learning_rate": 0.0001401905887133387,
-      "loss": 1.1785,
+      "loss": 1.1788,
       "step": 2415
     },
     {
       "epoch": 0.4324131153399446,
-      "grad_norm": 0.5375398397445679,
+      "grad_norm": 0.5291483402252197,
       "learning_rate": 0.00013990478042073313,
-      "loss": 1.1939,
+      "loss": 1.1941,
       "step": 2420
     },
     {
       "epoch": 0.4333065308675065,
-      "grad_norm": 0.510578989982605,
+      "grad_norm": 0.5093267560005188,
       "learning_rate": 0.00013961858389473727,
-      "loss": 1.1948,
+      "loss": 1.1938,
       "step": 2425
     },
     {
       "epoch": 0.43419994639506837,
-      "grad_norm": 0.526443362236023,
+      "grad_norm": 0.5296744108200073,
       "learning_rate": 0.0001393320019197555,
       "loss": 1.1978,
       "step": 2430
     },
     {
       "epoch": 0.4350933619226302,
-      "grad_norm": 0.4990314841270447,
+      "grad_norm": 0.49533629417419434,
       "learning_rate": 0.00013904503728394234,
-      "loss": 1.2009,
+      "loss": 1.2015,
       "step": 2435
     },
     {
       "epoch": 0.4359867774501921,
-      "grad_norm": 0.5471546053886414,
+      "grad_norm": 0.5509293079376221,
       "learning_rate": 0.00013875769277917513,
-      "loss": 1.1962,
+      "loss": 1.1963,
       "step": 2440
     },
     {
       "epoch": 0.43688019297775393,
-      "grad_norm": 0.4955877959728241,
+      "grad_norm": 0.49859216809272766,
       "learning_rate": 0.000138469971201027,
-      "loss": 1.1761,
+      "loss": 1.1748,
       "step": 2445
     },
     {
       "epoch": 0.4377736085053158,
-      "grad_norm": 0.5647017359733582,
+      "grad_norm": 0.5663615465164185,
       "learning_rate": 0.00013818187534873954,
-      "loss": 1.2009,
+      "loss": 1.2001,
       "step": 2450
     },
     {
       "epoch": 0.4386670240328777,
-      "grad_norm": 0.5101616978645325,
+      "grad_norm": 0.5102608799934387,
       "learning_rate": 0.00013789340802519581,
-      "loss": 1.2066,
+      "loss": 1.2063,
       "step": 2455
     },
     {
       "epoch": 0.43956043956043955,
-      "grad_norm": 0.5000629425048828,
+      "grad_norm": 0.5029070973396301,
       "learning_rate": 0.0001376045720368928,
-      "loss": 1.2066,
+      "loss": 1.2071,
       "step": 2460
     },
     {
       "epoch": 0.44045385508800144,
-      "grad_norm": 0.5091837048530579,
+      "grad_norm": 0.5075359344482422,
       "learning_rate": 0.00013731537019391428,
       "loss": 1.152,
       "step": 2465
     },
     {
       "epoch": 0.4413472706155633,
-      "grad_norm": 0.5215714573860168,
+      "grad_norm": 0.5264108777046204,
       "learning_rate": 0.00013702580530990335,
-      "loss": 1.202,
+      "loss": 1.2019,
       "step": 2470
     },
     {
       "epoch": 0.4422406861431252,
-      "grad_norm": 0.5080455541610718,
+      "grad_norm": 0.5055813789367676,
       "learning_rate": 0.00013673588020203517,
-      "loss": 1.1798,
+      "loss": 1.1786,
       "step": 2475
     },
     {
       "epoch": 0.44313410167068706,
-      "grad_norm": 0.5180327892303467,
+      "grad_norm": 0.5127353668212891,
       "learning_rate": 0.0001364455976909896,
-      "loss": 1.1654,
+      "loss": 1.1649,
       "step": 2480
     },
     {
       "epoch": 0.4440275171982489,
-      "grad_norm": 0.5066566467285156,
+      "grad_norm": 0.5117409825325012,
       "learning_rate": 0.00013615496060092355,
-      "loss": 1.1768,
+      "loss": 1.1762,
       "step": 2485
     },
     {
       "epoch": 0.4449209327258108,
-      "grad_norm": 0.5126291513442993,
+      "grad_norm": 0.5126619935035706,
       "learning_rate": 0.00013586397175944368,
-      "loss": 1.2028,
+      "loss": 1.2026,
       "step": 2490
     },
     {
       "epoch": 0.4458143482533726,
-      "grad_norm": 0.5379718542098999,
+      "grad_norm": 0.5334925055503845,
       "learning_rate": 0.0001355726339975788,
-      "loss": 1.2208,
+      "loss": 1.2209,
       "step": 2495
     },
     {
       "epoch": 0.4467077637809345,
-      "grad_norm": 0.5048497915267944,
+      "grad_norm": 0.5059501528739929,
       "learning_rate": 0.00013528095014975252,
-      "loss": 1.1812,
+      "loss": 1.1809,
       "step": 2500
     },
     {
       "epoch": 0.44760117930849636,
-      "grad_norm": 0.5384891033172607,
+      "grad_norm": 0.5418409705162048,
       "learning_rate": 0.0001349889230537553,
-      "loss": 1.2319,
+      "loss": 1.2321,
       "step": 2505
     },
     {
       "epoch": 0.44849459483605825,
-      "grad_norm": 0.541824221611023,
+      "grad_norm": 0.5434291362762451,
       "learning_rate": 0.00013469655555071715,
-      "loss": 1.1513,
+      "loss": 1.1511,
       "step": 2510
     },
     {
       "epoch": 0.44938801036362014,
-      "grad_norm": 0.5238659977912903,
+      "grad_norm": 0.523363471031189,
       "learning_rate": 0.00013440385048507997,
-      "loss": 1.209,
+      "loss": 1.2086,
       "step": 2515
     },
     {
       "epoch": 0.450281425891182,
-      "grad_norm": 0.529753565788269,
+      "grad_norm": 0.5296551585197449,
       "learning_rate": 0.0001341108107045697,
-      "loss": 1.1841,
+      "loss": 1.1834,
       "step": 2520
     },
     {
       "epoch": 0.45117484141874387,
-      "grad_norm": 0.5439874529838562,
+      "grad_norm": 0.5411925911903381,
       "learning_rate": 0.00013381743906016878,
-      "loss": 1.1729,
+      "loss": 1.173,
       "step": 2525
     },
     {
       "epoch": 0.4520682569463057,
-      "grad_norm": 0.5199108719825745,
+      "grad_norm": 0.524066150188446,
       "learning_rate": 0.00013352373840608834,
-      "loss": 1.226,
+      "loss": 1.2259,
       "step": 2530
     },
     {
       "epoch": 0.4529616724738676,
-      "grad_norm": 0.529483437538147,
+      "grad_norm": 0.5258145928382874,
       "learning_rate": 0.00013322971159974043,
-      "loss": 1.172,
+      "loss": 1.1724,
       "step": 2535
     },
     {
       "epoch": 0.4538550880014295,
-      "grad_norm": 0.5473654270172119,
+      "grad_norm": 0.5472233295440674,
       "learning_rate": 0.0001329353615017102,
-      "loss": 1.1524,
+      "loss": 1.1523,
       "step": 2540
     },
     {
       "epoch": 0.4547485035289913,
-      "grad_norm": 0.5226883292198181,
+      "grad_norm": 0.5172765254974365,
       "learning_rate": 0.00013264069097572816,
-      "loss": 1.1901,
+      "loss": 1.1904,
       "step": 2545
     },
     {
       "epoch": 0.4556419190565532,
-      "grad_norm": 0.49766266345977783,
+      "grad_norm": 0.49758392572402954,
       "learning_rate": 0.00013234570288864228,
-      "loss": 1.1731,
+      "loss": 1.1737,
       "step": 2550
     },
     {
       "epoch": 0.45653533458411505,
-      "grad_norm": 0.5254089832305908,
+      "grad_norm": 0.5258619785308838,
       "learning_rate": 0.00013205040011039004,
-      "loss": 1.1752,
+      "loss": 1.1746,
       "step": 2555
     },
     {
       "epoch": 0.45742875011167694,
-      "grad_norm": 0.4883306324481964,
+      "grad_norm": 0.4878065586090088,
       "learning_rate": 0.0001317547855139705,
-      "loss": 1.1997,
+      "loss": 1.1987,
       "step": 2560
     },
     {
       "epoch": 0.45832216563923883,
-      "grad_norm": 0.48100265860557556,
+      "grad_norm": 0.4854235053062439,
       "learning_rate": 0.00013145886197541651,
-      "loss": 1.1615,
+      "loss": 1.1605,
       "step": 2565
     },
     {
       "epoch": 0.45921558116680067,
-      "grad_norm": 0.4742489159107208,
+      "grad_norm": 0.48126521706581116,
       "learning_rate": 0.0001311626323737665,
-      "loss": 1.2034,
+      "loss": 1.2038,
       "step": 2570
     },
     {
       "epoch": 0.46010899669436256,
-      "grad_norm": 0.5177445411682129,
+      "grad_norm": 0.5124824643135071,
       "learning_rate": 0.00013086609959103672,
-      "loss": 1.1795,
+      "loss": 1.1791,
       "step": 2575
     },
     {
       "epoch": 0.4610024122219244,
-      "grad_norm": 0.548886775970459,
+      "grad_norm": 0.5470382571220398,
       "learning_rate": 0.00013056926651219293,
-      "loss": 1.2038,
+      "loss": 1.2034,
       "step": 2580
     },
     {
       "epoch": 0.4618958277494863,
-      "grad_norm": 0.5247954726219177,
+      "grad_norm": 0.5247116684913635,
       "learning_rate": 0.00013027213602512258,
-      "loss": 1.1983,
+      "loss": 1.1979,
       "step": 2585
     },
     {
       "epoch": 0.4627892432770482,
-      "grad_norm": 0.513616681098938,
+      "grad_norm": 0.5126326680183411,
       "learning_rate": 0.00012997471102060647,
-      "loss": 1.2036,
+      "loss": 1.2037,
       "step": 2590
     },
     {
       "epoch": 0.46368265880461,
-      "grad_norm": 0.5180881023406982,
+      "grad_norm": 0.5170397162437439,
       "learning_rate": 0.00012967699439229093,
-      "loss": 1.1869,
+      "loss": 1.186,
       "step": 2595
     },
     {
       "epoch": 0.4645760743321719,
-      "grad_norm": 0.513308048248291,
+      "grad_norm": 0.5169094800949097,
       "learning_rate": 0.00012937898903665935,
-      "loss": 1.1945,
+      "loss": 1.1941,
       "step": 2600
     },
     {
       "epoch": 0.46546948985973374,
-      "grad_norm": 0.5470111966133118,
+      "grad_norm": 0.537554144859314,
       "learning_rate": 0.0001290806978530042,
-      "loss": 1.184,
+      "loss": 1.1847,
       "step": 2605
     },
     {
       "epoch": 0.46636290538729563,
-      "grad_norm": 0.5049375891685486,
+      "grad_norm": 0.49902471899986267,
       "learning_rate": 0.00012878212374339883,
-      "loss": 1.1989,
+      "loss": 1.1983,
       "step": 2610
     },
     {
       "epoch": 0.4672563209148575,
-      "grad_norm": 0.5457560420036316,
+      "grad_norm": 0.5386156439781189,
       "learning_rate": 0.0001284832696126691,
-      "loss": 1.183,
+      "loss": 1.1829,
       "step": 2615
     },
     {
       "epoch": 0.46814973644241936,
-      "grad_norm": 0.5425926446914673,
+      "grad_norm": 0.5420889854431152,
       "learning_rate": 0.00012818413836836515,
-      "loss": 1.1818,
+      "loss": 1.1821,
       "step": 2620
     },
     {
       "epoch": 0.46904315196998125,
-      "grad_norm": 0.5093770027160645,
+      "grad_norm": 0.5098642110824585,
       "learning_rate": 0.00012788473292073328,
-      "loss": 1.1977,
+      "loss": 1.1982,
       "step": 2625
     },
     {
       "epoch": 0.4699365674975431,
-      "grad_norm": 0.47960811853408813,
+      "grad_norm": 0.4834945499897003,
       "learning_rate": 0.00012758505618268743,
-      "loss": 1.1948,
+      "loss": 1.1951,
       "step": 2630
     },
     {
       "epoch": 0.470829983025105,
-      "grad_norm": 0.5090500116348267,
+      "grad_norm": 0.5107752680778503,
       "learning_rate": 0.000127285111069781,
-      "loss": 1.1457,
+      "loss": 1.1451,
       "step": 2635
     },
     {
       "epoch": 0.4717233985526669,
-      "grad_norm": 0.5354019403457642,
+      "grad_norm": 0.5363967418670654,
       "learning_rate": 0.00012698490050017824,
-      "loss": 1.184,
+      "loss": 1.1832,
       "step": 2640
     },
     {
       "epoch": 0.4726168140802287,
-      "grad_norm": 0.46892282366752625,
+      "grad_norm": 0.469596266746521,
       "learning_rate": 0.0001266844273946262,
-      "loss": 1.1394,
+      "loss": 1.1397,
       "step": 2645
     },
     {
       "epoch": 0.4735102296077906,
-      "grad_norm": 0.5120964646339417,
+      "grad_norm": 0.5100436210632324,
       "learning_rate": 0.0001263836946764261,
-      "loss": 1.1829,
+      "loss": 1.1823,
       "step": 2650
     },
     {
       "epoch": 0.47440364513535244,
-      "grad_norm": 0.5253159403800964,
+      "grad_norm": 0.5279127359390259,
       "learning_rate": 0.0001260827052714049,
-      "loss": 1.1393,
+      "loss": 1.1407,
       "step": 2655
     },
     {
       "epoch": 0.47529706066291433,
-      "grad_norm": 0.49876195192337036,
+      "grad_norm": 0.49943119287490845,
       "learning_rate": 0.00012578146210788686,
-      "loss": 1.1603,
+      "loss": 1.16,
       "step": 2660
     },
     {
       "epoch": 0.47619047619047616,
-      "grad_norm": 0.542789876461029,
+      "grad_norm": 0.5424295663833618,
       "learning_rate": 0.0001254799681166651,
-      "loss": 1.197,
+      "loss": 1.1971,
       "step": 2665
     },
     {
       "epoch": 0.47708389171803806,
-      "grad_norm": 0.5235465168952942,
+      "grad_norm": 0.5269612669944763,
       "learning_rate": 0.00012517822623097296,
-      "loss": 1.1878,
+      "loss": 1.1883,
       "step": 2670
     },
     {
       "epoch": 0.47797730724559995,
-      "grad_norm": 0.4984067678451538,
+      "grad_norm": 0.4970255196094513,
       "learning_rate": 0.0001248762393864556,
-      "loss": 1.1687,
+      "loss": 1.169,
       "step": 2675
     },
     {
       "epoch": 0.4788707227731618,
-      "grad_norm": 0.4617389142513275,
+      "grad_norm": 0.46272140741348267,
       "learning_rate": 0.0001245740105211414,
-      "loss": 1.1608,
+      "loss": 1.1619,
       "step": 2680
     },
     {
       "epoch": 0.4797641383007237,
-      "grad_norm": 0.5108579993247986,
+      "grad_norm": 0.5113199949264526,
       "learning_rate": 0.00012427154257541333,
-      "loss": 1.1825,
+      "loss": 1.183,
       "step": 2685
     },
     {
       "epoch": 0.4806575538282855,
-      "grad_norm": 0.535814642906189,
+      "grad_norm": 0.5340253114700317,
       "learning_rate": 0.0001239688384919804,
-      "loss": 1.1533,
+      "loss": 1.1534,
       "step": 2690
     },
     {
       "epoch": 0.4815509693558474,
-      "grad_norm": 0.5493077635765076,
+      "grad_norm": 0.5490040183067322,
       "learning_rate": 0.00012366590121584895,
-      "loss": 1.1494,
+      "loss": 1.1496,
       "step": 2695
     },
     {
       "epoch": 0.4824443848834093,
-      "grad_norm": 0.5428511500358582,
+      "grad_norm": 0.5425564646720886,
       "learning_rate": 0.0001233627336942941,
-      "loss": 1.1999,
+      "loss": 1.2009,
       "step": 2700
     },
     {
       "epoch": 0.48333780041097113,
-      "grad_norm": 0.4960634708404541,
+      "grad_norm": 0.4985598027706146,
       "learning_rate": 0.00012305933887683102,
-      "loss": 1.1713,
+      "loss": 1.1716,
       "step": 2705
     },
     {
       "epoch": 0.484231215938533,
-      "grad_norm": 0.49134474992752075,
+      "grad_norm": 0.48820316791534424,
       "learning_rate": 0.00012275571971518616,
-      "loss": 1.1895,
+      "loss": 1.189,
       "step": 2710
     },
     {
       "epoch": 0.48512463146609486,
-      "grad_norm": 0.5415772795677185,
+      "grad_norm": 0.5365601181983948,
       "learning_rate": 0.00012245187916326878,
-      "loss": 1.1975,
+      "loss": 1.1969,
       "step": 2715
     },
     {
       "epoch": 0.48601804699365675,
-      "grad_norm": 0.5128939747810364,
+      "grad_norm": 0.5123448371887207,
       "learning_rate": 0.00012214782017714185,
-      "loss": 1.1426,
+      "loss": 1.1424,
       "step": 2720
     },
     {
       "epoch": 0.48691146252121864,
-      "grad_norm": 0.5035459399223328,
+      "grad_norm": 0.5060442686080933,
       "learning_rate": 0.00012184354571499365,
-      "loss": 1.1667,
+      "loss": 1.1682,
       "step": 2725
     },
     {
       "epoch": 0.4878048780487805,
-      "grad_norm": 0.5528464317321777,
+      "grad_norm": 0.5641975998878479,
       "learning_rate": 0.00012153905873710878,
-      "loss": 1.1943,
+      "loss": 1.1952,
       "step": 2730
     },
     {
       "epoch": 0.48869829357634237,
-      "grad_norm": 0.44781333208084106,
+      "grad_norm": 0.4496929347515106,
       "learning_rate": 0.00012123436220583931,
-      "loss": 1.178,
+      "loss": 1.1788,
       "step": 2735
     },
     {
       "epoch": 0.4895917091039042,
-      "grad_norm": 0.5722351670265198,
+      "grad_norm": 0.5671601295471191,
       "learning_rate": 0.00012092945908557616,
-      "loss": 1.1797,
+      "loss": 1.1792,
       "step": 2740
     },
     {
       "epoch": 0.4904851246314661,
-      "grad_norm": 0.5207979083061218,
+      "grad_norm": 0.5188336968421936,
       "learning_rate": 0.00012062435234272007,
-      "loss": 1.1919,
+      "loss": 1.1921,
       "step": 2745
     },
     {
       "epoch": 0.491378540159028,
-      "grad_norm": 0.5260438323020935,
+      "grad_norm": 0.5338768362998962,
       "learning_rate": 0.00012031904494565296,
-      "loss": 1.1874,
+      "loss": 1.1879,
       "step": 2750
     },
     {
       "epoch": 0.4922719556865898,
-      "grad_norm": 0.5548182725906372,
+      "grad_norm": 0.5573875308036804,
       "learning_rate": 0.00012001353986470878,
-      "loss": 1.1995,
+      "loss": 1.2,
       "step": 2755
     },
     {
       "epoch": 0.4931653712141517,
-      "grad_norm": 0.5213192701339722,
+      "grad_norm": 0.523671567440033,
       "learning_rate": 0.00011970784007214477,
-      "loss": 1.1892,
+      "loss": 1.1887,
       "step": 2760
     },
     {
       "epoch": 0.49405878674171355,
-      "grad_norm": 0.5301674604415894,
+      "grad_norm": 0.5382546186447144,
       "learning_rate": 0.00011940194854211258,
-      "loss": 1.1679,
+      "loss": 1.1676,
       "step": 2765
     },
     {
       "epoch": 0.49495220226927544,
-      "grad_norm": 0.5306346416473389,
+      "grad_norm": 0.5233120918273926,
       "learning_rate": 0.00011909586825062917,
-      "loss": 1.1775,
+      "loss": 1.178,
       "step": 2770
     },
     {
       "epoch": 0.49584561779683733,
-      "grad_norm": 0.5511103272438049,
+      "grad_norm": 0.5520097017288208,
       "learning_rate": 0.00011878960217554809,
-      "loss": 1.1754,
+      "loss": 1.1757,
       "step": 2775
     },
     {
       "epoch": 0.49673903332439917,
-      "grad_norm": 0.5091948509216309,
+      "grad_norm": 0.5060712695121765,
       "learning_rate": 0.00011848315329653028,
-      "loss": 1.1693,
+      "loss": 1.169,
       "step": 2780
     },
     {
       "epoch": 0.49763244885196106,
-      "grad_norm": 0.5281689167022705,
+      "grad_norm": 0.5251619815826416,
       "learning_rate": 0.0001181765245950152,
-      "loss": 1.1817,
+      "loss": 1.1813,
       "step": 2785
     },
     {
       "epoch": 0.4985258643795229,
-      "grad_norm": 0.48319005966186523,
+      "grad_norm": 0.48611682653427124,
       "learning_rate": 0.00011786971905419179,
       "loss": 1.1437,
       "step": 2790
     },
     {
       "epoch": 0.4994192799070848,
-      "grad_norm": 0.5014276504516602,
+      "grad_norm": 0.5000553727149963,
       "learning_rate": 0.00011756273965896953,
-      "loss": 1.1665,
+      "loss": 1.1668,
       "step": 2795
     },
     {
       "epoch": 0.5003126954346466,
-      "grad_norm": 0.5470197796821594,
+      "grad_norm": 0.548265278339386,
       "learning_rate": 0.00011725558939594924,
-      "loss": 1.1852,
+      "loss": 1.1859,
       "step": 2800
     },
     {
       "epoch": 0.5012061109622086,
-      "grad_norm": 0.5165842771530151,
+      "grad_norm": 0.5142680406570435,
       "learning_rate": 0.00011694827125339418,
-      "loss": 1.1756,
+      "loss": 1.176,
       "step": 2805
     },
     {
       "epoch": 0.5020995264897704,
-      "grad_norm": 0.5374709963798523,
+      "grad_norm": 0.5385056138038635,
       "learning_rate": 0.00011664078822120084,
-      "loss": 1.1946,
+      "loss": 1.195,
       "step": 2810
     },
     {
       "epoch": 0.5029929420173322,
-      "grad_norm": 0.48660171031951904,
+      "grad_norm": 0.4849853217601776,
       "learning_rate": 0.00011633314329086993,
-      "loss": 1.1606,
+      "loss": 1.1613,
       "step": 2815
     },
     {
       "epoch": 0.5038863575448941,
-      "grad_norm": 0.541174590587616,
+      "grad_norm": 0.5373867154121399,
       "learning_rate": 0.00011602533945547737,
-      "loss": 1.1393,
+      "loss": 1.1389,
       "step": 2820
     },
     {
       "epoch": 0.504779773072456,
-      "grad_norm": 0.5074775815010071,
+      "grad_norm": 0.5037809014320374,
       "learning_rate": 0.00011571737970964496,
-      "loss": 1.1719,
+      "loss": 1.1717,
       "step": 2825
     },
     {
       "epoch": 0.5056731886000179,
-      "grad_norm": 0.49670732021331787,
+      "grad_norm": 0.4938678741455078,
       "learning_rate": 0.00011540926704951136,
-      "loss": 1.1632,
+      "loss": 1.1625,
       "step": 2830
     },
     {
       "epoch": 0.5065666041275797,
-      "grad_norm": 0.4699111580848694,
+      "grad_norm": 0.47168177366256714,
       "learning_rate": 0.000115101004472703,
-      "loss": 1.137,
+      "loss": 1.1373,
       "step": 2835
     },
     {
       "epoch": 0.5074600196551416,
-      "grad_norm": 0.4844904839992523,
+      "grad_norm": 0.4792795479297638,
       "learning_rate": 0.00011479259497830472,
-      "loss": 1.1636,
+      "loss": 1.1635,
       "step": 2840
     },
     {
       "epoch": 0.5083534351827035,
-      "grad_norm": 0.5194590091705322,
+      "grad_norm": 0.5276928544044495,
       "learning_rate": 0.00011448404156683088,
-      "loss": 1.1817,
+      "loss": 1.1809,
       "step": 2845
     },
     {
       "epoch": 0.5092468507102653,
-      "grad_norm": 0.49438104033470154,
+      "grad_norm": 0.4980674684047699,
       "learning_rate": 0.00011417534724019592,
       "loss": 1.1669,
       "step": 2850
     },
     {
       "epoch": 0.5101402662378273,
-      "grad_norm": 0.5134628415107727,
+      "grad_norm": 0.5144380927085876,
       "learning_rate": 0.00011386651500168524,
-      "loss": 1.1938,
+      "loss": 1.1944,
       "step": 2855
     },
     {
       "epoch": 0.5110336817653891,
-      "grad_norm": 0.5145233869552612,
+      "grad_norm": 0.5162400007247925,
       "learning_rate": 0.00011355754785592596,
-      "loss": 1.1529,
+      "loss": 1.1526,
       "step": 2860
     },
     {
       "epoch": 0.5119270972929509,
-      "grad_norm": 0.5291239619255066,
+      "grad_norm": 0.5225709080696106,
       "learning_rate": 0.00011324844880885783,
-      "loss": 1.1737,
+      "loss": 1.1731,
       "step": 2865
     },
     {
       "epoch": 0.5128205128205128,
-      "grad_norm": 0.5408050417900085,
+      "grad_norm": 0.5327043533325195,
       "learning_rate": 0.00011293922086770375,
-      "loss": 1.148,
+      "loss": 1.1483,
       "step": 2870
     },
     {
       "epoch": 0.5137139283480747,
-      "grad_norm": 0.622309684753418,
+      "grad_norm": 0.60153728723526,
       "learning_rate": 0.00011262986704094065,
-      "loss": 1.1984,
+      "loss": 1.1987,
       "step": 2875
     },
     {
       "epoch": 0.5146073438756366,
-      "grad_norm": 0.5061530470848083,
+      "grad_norm": 0.5078974962234497,
       "learning_rate": 0.00011232039033827025,
-      "loss": 1.1492,
+      "loss": 1.15,
       "step": 2880
     },
     {
       "epoch": 0.5155007594031984,
-      "grad_norm": 0.49988749623298645,
+      "grad_norm": 0.5076168775558472,
       "learning_rate": 0.00011201079377058963,
-      "loss": 1.1392,
+      "loss": 1.14,
       "step": 2885
     },
     {
       "epoch": 0.5163941749307603,
-      "grad_norm": 0.5244172811508179,
+      "grad_norm": 0.5268626809120178,
       "learning_rate": 0.00011170108034996218,
-      "loss": 1.1682,
+      "loss": 1.1686,
       "step": 2890
     },
     {
       "epoch": 0.5172875904583222,
-      "grad_norm": 0.5669301748275757,
+      "grad_norm": 0.5654718279838562,
       "learning_rate": 0.00011139125308958804,
-      "loss": 1.1655,
+      "loss": 1.1656,
       "step": 2895
     },
     {
       "epoch": 0.518181005985884,
-      "grad_norm": 0.5045320987701416,
+      "grad_norm": 0.505577027797699,
       "learning_rate": 0.00011108131500377494,
-      "loss": 1.172,
+      "loss": 1.1724,
       "step": 2900
     },
     {
       "epoch": 0.5190744215134458,
-      "grad_norm": 0.5101925730705261,
+      "grad_norm": 0.50611811876297,
       "learning_rate": 0.00011077126910790882,
-      "loss": 1.1566,
+      "loss": 1.1562,
       "step": 2905
     },
     {
       "epoch": 0.5199678370410078,
-      "grad_norm": 0.5170836448669434,
+      "grad_norm": 0.5131402015686035,
       "learning_rate": 0.0001104611184184245,
-      "loss": 1.1494,
+      "loss": 1.1485,
       "step": 2910
     },
     {
       "epoch": 0.5208612525685696,
-      "grad_norm": 0.49804747104644775,
+      "grad_norm": 0.498565137386322,
       "learning_rate": 0.00011015086595277633,
-      "loss": 1.1615,
+      "loss": 1.1614,
       "step": 2915
     },
     {
       "epoch": 0.5217546680961315,
-      "grad_norm": 0.494862824678421,
+      "grad_norm": 0.4939316511154175,
       "learning_rate": 0.00010984051472940885,
-      "loss": 1.1393,
+      "loss": 1.1401,
       "step": 2920
     },
     {
       "epoch": 0.5226480836236934,
-      "grad_norm": 0.5359513759613037,
+      "grad_norm": 0.5283201336860657,
       "learning_rate": 0.00010953006776772747,
-      "loss": 1.1552,
+      "loss": 1.1557,
       "step": 2925
     },
     {
       "epoch": 0.5235414991512553,
-      "grad_norm": 0.49183422327041626,
+      "grad_norm": 0.492447167634964,
       "learning_rate": 0.00010921952808806888,
-      "loss": 1.1521,
+      "loss": 1.1519,
       "step": 2930
     },
     {
       "epoch": 0.5244349146788171,
-      "grad_norm": 0.5181066989898682,
+      "grad_norm": 0.5183736681938171,
       "learning_rate": 0.00010890889871167203,
-      "loss": 1.1544,
+      "loss": 1.1543,
       "step": 2935
     },
     {
       "epoch": 0.525328330206379,
-      "grad_norm": 0.5055237412452698,
+      "grad_norm": 0.5172650218009949,
       "learning_rate": 0.00010859818266064835,
-      "loss": 1.1722,
+      "loss": 1.1731,
       "step": 2940
     },
     {
       "epoch": 0.5262217457339409,
-      "grad_norm": 0.521922767162323,
+      "grad_norm": 0.5232465267181396,
       "learning_rate": 0.00010828738295795262,
-      "loss": 1.1637,
+      "loss": 1.1628,
       "step": 2945
     },
     {
       "epoch": 0.5271151612615027,
-      "grad_norm": 0.5190510153770447,
+      "grad_norm": 0.5208895206451416,
       "learning_rate": 0.00010797650262735346,
-      "loss": 1.1692,
+      "loss": 1.1682,
       "step": 2950
     },
     {
       "epoch": 0.5280085767890645,
-      "grad_norm": 0.5434553623199463,
+      "grad_norm": 0.5442891716957092,
       "learning_rate": 0.00010766554469340386,
-      "loss": 1.1801,
+      "loss": 1.1796,
       "step": 2955
     },
     {
       "epoch": 0.5289019923166265,
-      "grad_norm": 0.502286970615387,
+      "grad_norm": 0.5013934373855591,
       "learning_rate": 0.00010735451218141191,
-      "loss": 1.1753,
+      "loss": 1.1758,
       "step": 2960
     },
     {
       "epoch": 0.5297954078441883,
-      "grad_norm": 0.5568491220474243,
+      "grad_norm": 0.560184895992279,
       "learning_rate": 0.0001070434081174112,
-      "loss": 1.1603,
+      "loss": 1.1602,
       "step": 2965
     },
     {
       "epoch": 0.5306888233717502,
-      "grad_norm": 0.470877468585968,
+      "grad_norm": 0.4728075861930847,
       "learning_rate": 0.00010673223552813147,
-      "loss": 1.1581,
+      "loss": 1.1579,
       "step": 2970
     },
     {
       "epoch": 0.5315822388993121,
-      "grad_norm": 0.5503152012825012,
+      "grad_norm": 0.547065794467926,
       "learning_rate": 0.00010642099744096914,
-      "loss": 1.1617,
+      "loss": 1.1606,
       "step": 2975
     },
     {
       "epoch": 0.532475654426874,
-      "grad_norm": 0.5499110221862793,
+      "grad_norm": 0.5507860779762268,
       "learning_rate": 0.00010610969688395782,
-      "loss": 1.1525,
+      "loss": 1.1524,
       "step": 2980
     },
     {
       "epoch": 0.5333690699544358,
-      "grad_norm": 0.5412406325340271,
+      "grad_norm": 0.5366077423095703,
       "learning_rate": 0.00010579833688573897,
-      "loss": 1.1664,
+      "loss": 1.1668,
       "step": 2985
     },
     {
       "epoch": 0.5342624854819977,
-      "grad_norm": 0.4897433817386627,
+      "grad_norm": 0.49021753668785095,
       "learning_rate": 0.00010548692047553227,
-      "loss": 1.156,
+      "loss": 1.1554,
       "step": 2990
     },
     {
       "epoch": 0.5351559010095596,
-      "grad_norm": 0.5212523341178894,
+      "grad_norm": 0.5239299535751343,
       "learning_rate": 0.00010517545068310635,
-      "loss": 1.1502,
+      "loss": 1.1501,
       "step": 2995
     },
     {
       "epoch": 0.5360493165371214,
-      "grad_norm": 0.48938560485839844,
+      "grad_norm": 0.4920014441013336,
       "learning_rate": 0.00010486393053874902,
-      "loss": 1.1703,
+      "loss": 1.1705,
       "step": 3000
     },
     {
       "epoch": 0.5369427320646832,
-      "grad_norm": 0.5584519505500793,
+      "grad_norm": 0.5581287741661072,
       "learning_rate": 0.0001045523630732381,
-      "loss": 1.1739,
+      "loss": 1.1742,
       "step": 3005
     },
     {
       "epoch": 0.5378361475922452,
-      "grad_norm": 0.5129048228263855,
+      "grad_norm": 0.5146217346191406,
       "learning_rate": 0.00010424075131781178,
-      "loss": 1.1852,
+      "loss": 1.185,
       "step": 3010
     },
     {
       "epoch": 0.538729563119807,
-      "grad_norm": 0.5111370086669922,
+      "grad_norm": 0.5112408995628357,
       "learning_rate": 0.00010392909830413904,
-      "loss": 1.1613,
+      "loss": 1.162,
       "step": 3015
     },
     {
       "epoch": 0.5396229786473689,
-      "grad_norm": 0.5235916376113892,
+      "grad_norm": 0.526857316493988,
       "learning_rate": 0.00010361740706429046,
-      "loss": 1.1644,
+      "loss": 1.164,
       "step": 3020
     },
     {
       "epoch": 0.5405163941749308,
-      "grad_norm": 0.523694634437561,
+      "grad_norm": 0.5244843363761902,
       "learning_rate": 0.00010330568063070832,
-      "loss": 1.1947,
+      "loss": 1.1951,
       "step": 3025
     },
     {
       "epoch": 0.5414098097024926,
-      "grad_norm": 0.4990216791629791,
+      "grad_norm": 0.5005354881286621,
       "learning_rate": 0.00010299392203617744,
       "loss": 1.1552,
       "step": 3030
     },
     {
       "epoch": 0.5423032252300545,
-      "grad_norm": 0.5227737426757812,
+      "grad_norm": 0.5191918015480042,
       "learning_rate": 0.00010268213431379543,
-      "loss": 1.187,
+      "loss": 1.1868,
       "step": 3035
     },
     {
       "epoch": 0.5431966407576164,
-      "grad_norm": 0.49936816096305847,
+      "grad_norm": 0.4955528974533081,
       "learning_rate": 0.00010237032049694335,
       "loss": 1.1584,
       "step": 3040
     },
     {
       "epoch": 0.5440900562851783,
-      "grad_norm": 0.5005204081535339,
+      "grad_norm": 0.4983779489994049,
       "learning_rate": 0.00010205848361925618,
-      "loss": 1.1528,
+      "loss": 1.1532,
       "step": 3045
     },
     {
       "epoch": 0.5449834718127401,
-      "grad_norm": 0.553865909576416,
+      "grad_norm": 0.5516743659973145,
       "learning_rate": 0.0001017466267145931,
-      "loss": 1.1696,
+      "loss": 1.1702,
       "step": 3050
     },
     {
       "epoch": 0.5458768873403019,
-      "grad_norm": 0.5369734168052673,
+      "grad_norm": 0.5391404032707214,
       "learning_rate": 0.0001014347528170083,
-      "loss": 1.1576,
+      "loss": 1.1573,
       "step": 3055
     },
     {
       "epoch": 0.5467703028678639,
-      "grad_norm": 0.5343630909919739,
+      "grad_norm": 0.5334770083427429,
       "learning_rate": 0.00010112286496072117,
-      "loss": 1.1848,
+      "loss": 1.1847,
       "step": 3060
     },
     {
       "epoch": 0.5476637183954257,
-      "grad_norm": 0.535750687122345,
+      "grad_norm": 0.5390617251396179,
       "learning_rate": 0.00010081096618008699,
-      "loss": 1.1436,
+      "loss": 1.1428,
       "step": 3065
     },
     {
       "epoch": 0.5485571339229875,
-      "grad_norm": 0.551857590675354,
+      "grad_norm": 0.5461019277572632,
       "learning_rate": 0.00010049905950956728,
-      "loss": 1.1477,
+      "loss": 1.1484,
       "step": 3070
     },
     {
       "epoch": 0.5494505494505495,
-      "grad_norm": 0.5209478139877319,
+      "grad_norm": 0.5262424945831299,
       "learning_rate": 0.00010018714798370035,
-      "loss": 1.1672,
+      "loss": 1.1679,
       "step": 3075
     },
     {
       "epoch": 0.5503439649781113,
-      "grad_norm": 0.47555601596832275,
+      "grad_norm": 0.4764571189880371,
       "learning_rate": 9.98752346370717e-05,
       "loss": 1.1585,
       "step": 3080
     },
     {
       "epoch": 0.5512373805056732,
-      "grad_norm": 0.4957531988620758,
+      "grad_norm": 0.49840331077575684,
       "learning_rate": 9.956332250428457e-05,
-      "loss": 1.1574,
+      "loss": 1.1572,
       "step": 3085
     },
     {
       "epoch": 0.552130796033235,
-      "grad_norm": 0.4895710051059723,
+      "grad_norm": 0.48860788345336914,
       "learning_rate": 9.925141461993043e-05,
-      "loss": 1.1262,
+      "loss": 1.126,
       "step": 3090
     },
     {
       "epoch": 0.553024211560797,
-      "grad_norm": 0.5204643607139587,
+      "grad_norm": 0.5242288708686829,
       "learning_rate": 9.893951401855932e-05,
-      "loss": 1.1567,
+      "loss": 1.1558,
       "step": 3095
     },
     {
       "epoch": 0.5539176270883588,
-      "grad_norm": 0.4921712279319763,
+      "grad_norm": 0.49476343393325806,
       "learning_rate": 9.862762373465055e-05,
-      "loss": 1.1659,
+      "loss": 1.1653,
       "step": 3100
     },
     {
       "epoch": 0.5548110426159206,
-      "grad_norm": 0.5866613388061523,
+      "grad_norm": 0.583516538143158,
       "learning_rate": 9.831574680258297e-05,
-      "loss": 1.141,
+      "loss": 1.1412,
       "step": 3105
     },
     {
       "epoch": 0.5557044581434826,
-      "grad_norm": 0.5556049346923828,
+      "grad_norm": 0.5565661191940308,
       "learning_rate": 9.800388625660553e-05,
-      "loss": 1.1803,
+      "loss": 1.1802,
       "step": 3110
     },
     {
       "epoch": 0.5565978736710444,
-      "grad_norm": 0.5409796833992004,
+      "grad_norm": 0.5394544005393982,
       "learning_rate": 9.769204513080775e-05,
-      "loss": 1.1225,
+      "loss": 1.1222,
       "step": 3115
     },
     {
       "epoch": 0.5574912891986062,
-      "grad_norm": 0.5227869749069214,
+      "grad_norm": 0.5191746354103088,
       "learning_rate": 9.738022645909026e-05,
-      "loss": 1.2076,
+      "loss": 1.2078,
       "step": 3120
     },
     {
       "epoch": 0.5583847047261682,
-      "grad_norm": 0.5156333446502686,
+      "grad_norm": 0.5176223516464233,
       "learning_rate": 9.706843327513521e-05,
-      "loss": 1.1229,
+      "loss": 1.1232,
       "step": 3125
     },
     {
       "epoch": 0.55927812025373,
-      "grad_norm": 0.5258979201316833,
+      "grad_norm": 0.512442409992218,
       "learning_rate": 9.675666861237677e-05,
-      "loss": 1.1495,
+      "loss": 1.1489,
       "step": 3130
     },
     {
       "epoch": 0.5601715357812919,
-      "grad_norm": 0.5187394022941589,
+      "grad_norm": 0.5076617002487183,
       "learning_rate": 9.644493550397168e-05,
-      "loss": 1.1477,
+      "loss": 1.1471,
       "step": 3135
     },
     {
       "epoch": 0.5610649513088537,
-      "grad_norm": 0.4999004304409027,
+      "grad_norm": 0.5024195909500122,
       "learning_rate": 9.61332369827696e-05,
-      "loss": 1.1449,
+      "loss": 1.1461,
       "step": 3140
     },
     {
       "epoch": 0.5619583668364156,
-      "grad_norm": 0.5169167518615723,
+      "grad_norm": 0.5137572884559631,
       "learning_rate": 9.582157608128374e-05,
-      "loss": 1.1284,
+      "loss": 1.1283,
       "step": 3145
     },
     {
       "epoch": 0.5628517823639775,
-      "grad_norm": 0.5010549426078796,
+      "grad_norm": 0.5018545985221863,
       "learning_rate": 9.550995583166133e-05,
-      "loss": 1.1526,
+      "loss": 1.1518,
       "step": 3150
     },
     {
       "epoch": 0.5637451978915393,
-      "grad_norm": 0.4774199426174164,
+      "grad_norm": 0.4772183299064636,
       "learning_rate": 9.519837926565409e-05,
-      "loss": 1.1382,
+      "loss": 1.1379,
       "step": 3155
     },
     {
       "epoch": 0.5646386134191013,
-      "grad_norm": 0.5373679995536804,
+      "grad_norm": 0.5322695970535278,
       "learning_rate": 9.488684941458867e-05,
-      "loss": 1.1759,
+      "loss": 1.1748,
       "step": 3160
     },
     {
       "epoch": 0.5655320289466631,
-      "grad_norm": 0.514975368976593,
+      "grad_norm": 0.5145288109779358,
       "learning_rate": 9.45753693093373e-05,
-      "loss": 1.1553,
+      "loss": 1.1556,
       "step": 3165
     },
     {
       "epoch": 0.5664254444742249,
-      "grad_norm": 0.5256127119064331,
+      "grad_norm": 0.5218089818954468,
       "learning_rate": 9.426394198028823e-05,
-      "loss": 1.1217,
+      "loss": 1.1221,
       "step": 3170
     },
     {
       "epoch": 0.5673188600017869,
-      "grad_norm": 0.49551287293434143,
+      "grad_norm": 0.49951624870300293,
       "learning_rate": 9.395257045731627e-05,
-      "loss": 1.1269,
+      "loss": 1.1255,
       "step": 3175
     },
     {
       "epoch": 0.5682122755293487,
-      "grad_norm": 0.5360212922096252,
+      "grad_norm": 0.5349987745285034,
       "learning_rate": 9.364125776975318e-05,
-      "loss": 1.1636,
+      "loss": 1.1647,
       "step": 3180
     },
     {
       "epoch": 0.5691056910569106,
-      "grad_norm": 0.5216602087020874,
+      "grad_norm": 0.5180804133415222,
       "learning_rate": 9.333000694635849e-05,
-      "loss": 1.1649,
+      "loss": 1.1653,
       "step": 3185
     },
     {
       "epoch": 0.5699991065844724,
-      "grad_norm": 0.4999989867210388,
+      "grad_norm": 0.49906909465789795,
       "learning_rate": 9.30188210152897e-05,
-      "loss": 1.1135,
+      "loss": 1.1138,
       "step": 3190
     },
     {
       "epoch": 0.5708925221120343,
-      "grad_norm": 0.5036138892173767,
+      "grad_norm": 0.5072324872016907,
       "learning_rate": 9.270770300407305e-05,
-      "loss": 1.1549,
+      "loss": 1.1542,
       "step": 3195
     },
     {
       "epoch": 0.5717859376395962,
-      "grad_norm": 0.5282987356185913,
+      "grad_norm": 0.52315354347229,
       "learning_rate": 9.239665593957398e-05,
-      "loss": 1.1426,
+      "loss": 1.1425,
       "step": 3200
     },
     {
       "epoch": 0.572679353167158,
-      "grad_norm": 0.5551292300224304,
+      "grad_norm": 0.5507954359054565,
       "learning_rate": 9.208568284796766e-05,
-      "loss": 1.147,
+      "loss": 1.1462,
       "step": 3205
     },
     {
       "epoch": 0.57357276869472,
-      "grad_norm": 0.5044068098068237,
+      "grad_norm": 0.5008817911148071,
       "learning_rate": 9.177478675470956e-05,
       "loss": 1.1307,
       "step": 3210
     },
     {
       "epoch": 0.5744661842222818,
-      "grad_norm": 0.4707668125629425,
+      "grad_norm": 0.47194069623947144,
       "learning_rate": 9.146397068450612e-05,
-      "loss": 1.1298,
+      "loss": 1.13,
       "step": 3215
     },
     {
       "epoch": 0.5753595997498436,
-      "grad_norm": 0.5296815037727356,
+      "grad_norm": 0.5233185887336731,
       "learning_rate": 9.11532376612852e-05,
-      "loss": 1.1601,
+      "loss": 1.1596,
       "step": 3220
     },
     {
       "epoch": 0.5762530152774055,
-      "grad_norm": 0.508043646812439,
+      "grad_norm": 0.5047646164894104,
       "learning_rate": 9.084259070816662e-05,
-      "loss": 1.1773,
+      "loss": 1.1768,
       "step": 3225
     },
     {
       "epoch": 0.5771464308049674,
-      "grad_norm": 0.5130497813224792,
+      "grad_norm": 0.5104073286056519,
       "learning_rate": 9.053203284743294e-05,
-      "loss": 1.1281,
+      "loss": 1.1276,
       "step": 3230
     },
     {
       "epoch": 0.5780398463325293,
-      "grad_norm": 0.48072364926338196,
+      "grad_norm": 0.48400700092315674,
       "learning_rate": 9.022156710049985e-05,
-      "loss": 1.1401,
+      "loss": 1.1409,
       "step": 3235
     },
     {
       "epoch": 0.5789332618600911,
-      "grad_norm": 0.5246201753616333,
+      "grad_norm": 0.5238465666770935,
       "learning_rate": 8.991119648788696e-05,
-      "loss": 1.1313,
+      "loss": 1.1316,
       "step": 3240
     },
     {
       "epoch": 0.579826677387653,
-      "grad_norm": 0.5002977252006531,
+      "grad_norm": 0.49904054403305054,
       "learning_rate": 8.960092402918819e-05,
       "loss": 1.1225,
       "step": 3245
     },
     {
       "epoch": 0.5807200929152149,
-      "grad_norm": 0.5016835927963257,
+      "grad_norm": 0.5016510486602783,
       "learning_rate": 8.929075274304267e-05,
-      "loss": 1.1257,
+      "loss": 1.1262,
       "step": 3250
     },
     {
       "epoch": 0.5816135084427767,
-      "grad_norm": 0.5389193296432495,
+      "grad_norm": 0.5361919403076172,
       "learning_rate": 8.898068564710508e-05,
-      "loss": 1.1445,
+      "loss": 1.1453,
       "step": 3255
     },
     {
       "epoch": 0.5825069239703387,
-      "grad_norm": 0.5444117188453674,
+      "grad_norm": 0.5483645796775818,
       "learning_rate": 8.86707257580165e-05,
-      "loss": 1.1449,
+      "loss": 1.1445,
       "step": 3260
     },
     {
       "epoch": 0.5834003394979005,
-      "grad_norm": 0.501939594745636,
+      "grad_norm": 0.49957215785980225,
       "learning_rate": 8.836087609137502e-05,
-      "loss": 1.1912,
+      "loss": 1.1917,
       "step": 3265
     },
     {
       "epoch": 0.5842937550254623,
-      "grad_norm": 0.5322566628456116,
+      "grad_norm": 0.5269734859466553,
       "learning_rate": 8.805113966170635e-05,
-      "loss": 1.128,
+      "loss": 1.1277,
       "step": 3270
     },
     {
       "epoch": 0.5851871705530242,
-      "grad_norm": 0.5263496041297913,
+      "grad_norm": 0.5271036028862,
       "learning_rate": 8.774151948243453e-05,
-      "loss": 1.1253,
+      "loss": 1.1263,
       "step": 3275
     },
     {
       "epoch": 0.5860805860805861,
-      "grad_norm": 0.5112000703811646,
+      "grad_norm": 0.5119196772575378,
       "learning_rate": 8.743201856585256e-05,
-      "loss": 1.1595,
+      "loss": 1.1596,
       "step": 3280
     },
     {
       "epoch": 0.5869740016081479,
-      "grad_norm": 0.5330345630645752,
+      "grad_norm": 0.5136052370071411,
       "learning_rate": 8.712263992309318e-05,
       "loss": 1.1191,
       "step": 3285
     },
     {
       "epoch": 0.5878674171357098,
-      "grad_norm": 0.49791669845581055,
+      "grad_norm": 0.500639796257019,
       "learning_rate": 8.681338656409951e-05,
-      "loss": 1.1125,
+      "loss": 1.1134,
       "step": 3290
     },
     {
       "epoch": 0.5887608326632717,
-      "grad_norm": 0.5101556181907654,
+      "grad_norm": 0.5073699951171875,
       "learning_rate": 8.650426149759578e-05,
-      "loss": 1.1345,
+      "loss": 1.1346,
       "step": 3295
     },
     {
       "epoch": 0.5896542481908336,
-      "grad_norm": 0.521859347820282,
+      "grad_norm": 0.5222281813621521,
       "learning_rate": 8.619526773105813e-05,
-      "loss": 1.1581,
+      "loss": 1.1575,
       "step": 3300
     },
     {
       "epoch": 0.5905476637183954,
-      "grad_norm": 0.5242456197738647,
+      "grad_norm": 0.5326797366142273,
       "learning_rate": 8.588640827068512e-05,
-      "loss": 1.1512,
+      "loss": 1.1509,
       "step": 3305
     },
     {
       "epoch": 0.5914410792459573,
-      "grad_norm": 0.489236444234848,
+      "grad_norm": 0.49078214168548584,
       "learning_rate": 8.557768612136882e-05,
-      "loss": 1.099,
+      "loss": 1.0998,
       "step": 3310
     },
     {
       "epoch": 0.5923344947735192,
-      "grad_norm": 0.520885169506073,
+      "grad_norm": 0.5205763578414917,
       "learning_rate": 8.52691042866653e-05,
-      "loss": 1.1243,
+      "loss": 1.1247,
       "step": 3315
     },
     {
       "epoch": 0.593227910301081,
-      "grad_norm": 0.495631605386734,
+      "grad_norm": 0.49795159697532654,
       "learning_rate": 8.496066576876556e-05,
-      "loss": 1.1235,
+      "loss": 1.1234,
       "step": 3320
     },
     {
       "epoch": 0.5941213258286429,
-      "grad_norm": 0.5357411503791809,
+      "grad_norm": 0.5376335382461548,
       "learning_rate": 8.465237356846624e-05,
       "loss": 1.1647,
       "step": 3325
     },
     {
       "epoch": 0.5950147413562048,
-      "grad_norm": 0.543991208076477,
+      "grad_norm": 0.5440948009490967,
       "learning_rate": 8.434423068514048e-05,
-      "loss": 1.1379,
+      "loss": 1.1377,
       "step": 3330
     },
     {
       "epoch": 0.5959081568837666,
-      "grad_norm": 0.5474541187286377,
+      "grad_norm": 0.5351749062538147,
       "learning_rate": 8.403624011670871e-05,
-      "loss": 1.1423,
+      "loss": 1.1429,
       "step": 3335
     },
     {
       "epoch": 0.5968015724113285,
-      "grad_norm": 0.5549032092094421,
+      "grad_norm": 0.5545377731323242,
       "learning_rate": 8.372840485960947e-05,
-      "loss": 1.1872,
+      "loss": 1.1869,
       "step": 3340
     },
     {
       "epoch": 0.5976949879388904,
-      "grad_norm": 0.5237979888916016,
+      "grad_norm": 0.5247533917427063,
       "learning_rate": 8.342072790877033e-05,
-      "loss": 1.1388,
+      "loss": 1.138,
       "step": 3345
     },
     {
       "epoch": 0.5985884034664523,
-      "grad_norm": 0.5107823610305786,
+      "grad_norm": 0.5115054845809937,
       "learning_rate": 8.311321225757858e-05,
-      "loss": 1.1151,
+      "loss": 1.1147,
       "step": 3350
     },
     {
       "epoch": 0.5994818189940141,
-      "grad_norm": 0.4889257252216339,
+      "grad_norm": 0.49062785506248474,
       "learning_rate": 8.280586089785236e-05,
-      "loss": 1.1501,
+      "loss": 1.1498,
       "step": 3355
     },
     {
       "epoch": 0.600375234521576,
-      "grad_norm": 0.5312927961349487,
+      "grad_norm": 0.5327452421188354,
       "learning_rate": 8.249867681981139e-05,
-      "loss": 1.1124,
+      "loss": 1.113,
       "step": 3360
     },
     {
       "epoch": 0.6012686500491379,
-      "grad_norm": 0.5377475619316101,
+      "grad_norm": 0.5428333878517151,
       "learning_rate": 8.219166301204781e-05,
       "loss": 1.1632,
       "step": 3365
     },
     {
       "epoch": 0.6021620655766997,
-      "grad_norm": 0.5036457777023315,
+      "grad_norm": 0.5033100247383118,
       "learning_rate": 8.188482246149736e-05,
-      "loss": 1.1485,
+      "loss": 1.1484,
       "step": 3370
     },
     {
       "epoch": 0.6030554811042615,
-      "grad_norm": 0.5324269533157349,
+      "grad_norm": 0.5400857329368591,
       "learning_rate": 8.157815815341002e-05,
-      "loss": 1.1563,
+      "loss": 1.1556,
       "step": 3375
     },
     {
       "epoch": 0.6039488966318235,
-      "grad_norm": 0.5336183309555054,
+      "grad_norm": 0.5341722369194031,
       "learning_rate": 8.127167307132119e-05,
       "loss": 1.1387,
       "step": 3380
     },
     {
       "epoch": 0.6048423121593853,
-      "grad_norm": 0.5451260805130005,
+      "grad_norm": 0.5450566411018372,
       "learning_rate": 8.096537019702255e-05,
-      "loss": 1.1599,
+      "loss": 1.1598,
       "step": 3385
     },
     {
       "epoch": 0.6057357276869472,
-      "grad_norm": 0.4915177524089813,
+      "grad_norm": 0.496113121509552,
       "learning_rate": 8.065925251053307e-05,
       "loss": 1.1664,
       "step": 3390
     },
     {
       "epoch": 0.6066291432145091,
-      "grad_norm": 0.5154266357421875,
+      "grad_norm": 0.5090356469154358,
       "learning_rate": 8.035332299007014e-05,
-      "loss": 1.1343,
+      "loss": 1.1338,
       "step": 3395
     },
     {
       "epoch": 0.607522558742071,
-      "grad_norm": 0.48472052812576294,
+      "grad_norm": 0.48691728711128235,
       "learning_rate": 8.004758461202023e-05,
-      "loss": 1.1482,
+      "loss": 1.1481,
       "step": 3400
     },
     {
       "epoch": 0.6084159742696328,
-      "grad_norm": 0.48825910687446594,
+      "grad_norm": 0.4871600270271301,
       "learning_rate": 7.974204035091046e-05,
-      "loss": 1.1233,
+      "loss": 1.1227,
       "step": 3405
     },
     {
       "epoch": 0.6093093897971946,
-      "grad_norm": 0.5200708508491516,
+      "grad_norm": 0.5205426216125488,
       "learning_rate": 7.943669317937923e-05,
-      "loss": 1.1334,
+      "loss": 1.1338,
       "step": 3410
     },
     {
       "epoch": 0.6102028053247566,
-      "grad_norm": 0.5124212503433228,
+      "grad_norm": 0.5116109848022461,
       "learning_rate": 7.913154606814753e-05,
-      "loss": 1.1336,
+      "loss": 1.1329,
       "step": 3415
     },
     {
       "epoch": 0.6110962208523184,
-      "grad_norm": 0.5567544102668762,
+      "grad_norm": 0.5520155429840088,
       "learning_rate": 7.882660198598993e-05,
-      "loss": 1.1524,
+      "loss": 1.1526,
       "step": 3420
     },
     {
       "epoch": 0.6119896363798802,
-      "grad_norm": 0.5360432267189026,
+      "grad_norm": 0.5377830266952515,
       "learning_rate": 7.852186389970571e-05,
-      "loss": 1.1671,
+      "loss": 1.1676,
       "step": 3425
     },
     {
       "epoch": 0.6128830519074422,
-      "grad_norm": 0.5166192650794983,
+      "grad_norm": 0.5186334252357483,
       "learning_rate": 7.821733477409005e-05,
-      "loss": 1.1043,
+      "loss": 1.1049,
       "step": 3430
     },
     {
       "epoch": 0.613776467435004,
-      "grad_norm": 0.5590426921844482,
+      "grad_norm": 0.5580330491065979,
       "learning_rate": 7.791301757190516e-05,
-      "loss": 1.1091,
+      "loss": 1.1103,
       "step": 3435
     },
     {
       "epoch": 0.6146698829625659,
-      "grad_norm": 0.5402864813804626,
+      "grad_norm": 0.5414636135101318,
       "learning_rate": 7.760891525385146e-05,
-      "loss": 1.125,
+      "loss": 1.1243,
       "step": 3440
     },
     {
       "epoch": 0.6155632984901278,
-      "grad_norm": 0.4936152696609497,
+      "grad_norm": 0.494886577129364,
       "learning_rate": 7.730503077853873e-05,
-      "loss": 1.1471,
+      "loss": 1.1476,
       "step": 3445
     },
     {
       "epoch": 0.6164567140176896,
-      "grad_norm": 0.48089155554771423,
+      "grad_norm": 0.4783135950565338,
       "learning_rate": 7.700136710245731e-05,
-      "loss": 1.0873,
+      "loss": 1.0874,
       "step": 3450
     },
     {
       "epoch": 0.6173501295452515,
-      "grad_norm": 0.517113447189331,
+      "grad_norm": 0.5152401924133301,
       "learning_rate": 7.669792717994946e-05,
-      "loss": 1.1164,
+      "loss": 1.1166,
       "step": 3455
     },
     {
       "epoch": 0.6182435450728133,
-      "grad_norm": 0.5420798063278198,
+      "grad_norm": 0.5359469056129456,
       "learning_rate": 7.639471396318057e-05,
-      "loss": 1.1514,
+      "loss": 1.1519,
       "step": 3460
     },
     {
       "epoch": 0.6191369606003753,
-      "grad_norm": 0.5132094025611877,
+      "grad_norm": 0.5105694532394409,
       "learning_rate": 7.609173040211035e-05,
-      "loss": 1.1526,
+      "loss": 1.1516,
       "step": 3465
     },
     {
       "epoch": 0.6200303761279371,
-      "grad_norm": 0.5790542364120483,
+      "grad_norm": 0.5673738718032837,
       "learning_rate": 7.578897944446417e-05,
-      "loss": 1.1685,
+      "loss": 1.168,
       "step": 3470
     },
     {
       "epoch": 0.6209237916554989,
-      "grad_norm": 0.5307674407958984,
+      "grad_norm": 0.5316708087921143,
       "learning_rate": 7.548646403570449e-05,
-      "loss": 1.1245,
+      "loss": 1.1237,
       "step": 3475
     },
     {
       "epoch": 0.6218172071830609,
-      "grad_norm": 0.5037259459495544,
+      "grad_norm": 0.5048865079879761,
       "learning_rate": 7.518418711900206e-05,
-      "loss": 1.1549,
+      "loss": 1.1546,
       "step": 3480
     },
     {
       "epoch": 0.6227106227106227,
-      "grad_norm": 0.49981221556663513,
+      "grad_norm": 0.5034053921699524,
       "learning_rate": 7.488215163520729e-05,
-      "loss": 1.1136,
+      "loss": 1.114,
       "step": 3485
     },
     {
       "epoch": 0.6236040382381846,
-      "grad_norm": 0.5257123708724976,
+      "grad_norm": 0.5447829365730286,
       "learning_rate": 7.458036052282185e-05,
-      "loss": 1.0913,
+      "loss": 1.0924,
       "step": 3490
     },
     {
       "epoch": 0.6244974537657465,
-      "grad_norm": 0.5649625062942505,
+      "grad_norm": 0.5631914138793945,
       "learning_rate": 7.427881671796973e-05,
-      "loss": 1.129,
+      "loss": 1.1281,
       "step": 3495
     },
     {
       "epoch": 0.6253908692933083,
-      "grad_norm": 0.5401619672775269,
+      "grad_norm": 0.5400162935256958,
       "learning_rate": 7.3977523154369e-05,
-      "loss": 1.1526,
+      "loss": 1.1529,
       "step": 3500
     },
     {
       "epoch": 0.6262842848208702,
-      "grad_norm": 0.5306874513626099,
+      "grad_norm": 0.5325872302055359,
       "learning_rate": 7.367648276330305e-05,
-      "loss": 1.1597,
+      "loss": 1.1596,
       "step": 3505
     },
     {
       "epoch": 0.627177700348432,
-      "grad_norm": 0.5311725735664368,
+      "grad_norm": 0.5262747406959534,
       "learning_rate": 7.337569847359226e-05,
-      "loss": 1.1308,
+      "loss": 1.1311,
       "step": 3510
     },
     {
       "epoch": 0.628071115875994,
-      "grad_norm": 0.5326215028762817,
+      "grad_norm": 0.530774712562561,
       "learning_rate": 7.307517321156528e-05,
-      "loss": 1.1142,
+      "loss": 1.1147,
       "step": 3515
     },
     {
       "epoch": 0.6289645314035558,
-      "grad_norm": 0.5113494992256165,
+      "grad_norm": 0.5109870433807373,
       "learning_rate": 7.277490990103079e-05,
-      "loss": 1.1321,
+      "loss": 1.132,
       "step": 3520
     },
     {
       "epoch": 0.6298579469311176,
-      "grad_norm": 0.5225013494491577,
+      "grad_norm": 0.5231294631958008,
       "learning_rate": 7.247491146324887e-05,
-      "loss": 1.1103,
+      "loss": 1.1106,
       "step": 3525
     },
     {
       "epoch": 0.6307513624586796,
-      "grad_norm": 0.522226095199585,
+      "grad_norm": 0.5209240913391113,
       "learning_rate": 7.217518081690265e-05,
-      "loss": 1.1611,
+      "loss": 1.1614,
       "step": 3530
     },
     {
       "epoch": 0.6316447779862414,
-      "grad_norm": 0.5471552014350891,
+      "grad_norm": 0.5387827157974243,
       "learning_rate": 7.187572087807e-05,
-      "loss": 1.1654,
+      "loss": 1.1656,
       "step": 3535
     },
     {
       "epoch": 0.6325381935138032,
-      "grad_norm": 0.5112610459327698,
+      "grad_norm": 0.5082910060882568,
       "learning_rate": 7.157653456019504e-05,
-      "loss": 1.1273,
+      "loss": 1.1274,
       "step": 3540
     },
     {
       "epoch": 0.6334316090413651,
-      "grad_norm": 0.5094234347343445,
+      "grad_norm": 0.5118305683135986,
       "learning_rate": 7.127762477405976e-05,
-      "loss": 1.1483,
+      "loss": 1.1485,
       "step": 3545
     },
     {
       "epoch": 0.634325024568927,
-      "grad_norm": 0.5493167638778687,
+      "grad_norm": 0.5459609031677246,
       "learning_rate": 7.097899442775584e-05,
-      "loss": 1.1254,
+      "loss": 1.1253,
       "step": 3550
     },
     {
       "epoch": 0.6352184400964889,
-      "grad_norm": 0.5173693299293518,
+      "grad_norm": 0.5178401470184326,
       "learning_rate": 7.068064642665631e-05,
-      "loss": 1.1468,
+      "loss": 1.1465,
       "step": 3555
     },
     {
       "epoch": 0.6361118556240507,
-      "grad_norm": 0.5034879446029663,
+      "grad_norm": 0.5017228722572327,
       "learning_rate": 7.038258367338723e-05,
-      "loss": 1.1349,
+      "loss": 1.1346,
       "step": 3560
     },
     {
       "epoch": 0.6370052711516127,
-      "grad_norm": 0.546248197555542,
+      "grad_norm": 0.5513466000556946,
       "learning_rate": 7.008480906779948e-05,
       "loss": 1.1423,
       "step": 3565
     },
     {
       "epoch": 0.6378986866791745,
-      "grad_norm": 0.49636945128440857,
+      "grad_norm": 0.49276652932167053,
       "learning_rate": 6.97873255069406e-05,
-      "loss": 1.1527,
+      "loss": 1.1522,
       "step": 3570
     },
     {
       "epoch": 0.6387921022067363,
-      "grad_norm": 0.5306327939033508,
+      "grad_norm": 0.5311774015426636,
       "learning_rate": 6.949013588502651e-05,
-      "loss": 1.1298,
+      "loss": 1.1291,
       "step": 3575
     },
     {
       "epoch": 0.6396855177342983,
-      "grad_norm": 0.4831233024597168,
+      "grad_norm": 0.4871313273906708,
       "learning_rate": 6.919324309341341e-05,
-      "loss": 1.1189,
+      "loss": 1.1199,
       "step": 3580
     },
     {
       "epoch": 0.6405789332618601,
-      "grad_norm": 0.553893506526947,
+      "grad_norm": 0.5568504333496094,
       "learning_rate": 6.889665002056966e-05,
-      "loss": 1.118,
+      "loss": 1.1184,
       "step": 3585
     },
     {
       "epoch": 0.6414723487894219,
-      "grad_norm": 0.4722842872142792,
+      "grad_norm": 0.4711260199546814,
       "learning_rate": 6.860035955204767e-05,
-      "loss": 1.1195,
+      "loss": 1.1204,
       "step": 3590
     },
     {
       "epoch": 0.6423657643169838,
-      "grad_norm": 0.5145003795623779,
+      "grad_norm": 0.5173743963241577,
       "learning_rate": 6.830437457045568e-05,
       "loss": 1.1491,
       "step": 3595
     },
     {
       "epoch": 0.6432591798445457,
-      "grad_norm": 0.5275436639785767,
+      "grad_norm": 0.5250765085220337,
       "learning_rate": 6.800869795543007e-05,
       "loss": 1.1086,
       "step": 3600
     },
     {
       "epoch": 0.6441525953721076,
-      "grad_norm": 0.5202250480651855,
+      "grad_norm": 0.5202131867408752,
       "learning_rate": 6.771333258360694e-05,
-      "loss": 1.1357,
+      "loss": 1.1355,
       "step": 3605
     },
     {
       "epoch": 0.6450460108996694,
-      "grad_norm": 0.5156890749931335,
+      "grad_norm": 0.5153080821037292,
       "learning_rate": 6.74182813285944e-05,
-      "loss": 1.118,
+      "loss": 1.1181,
       "step": 3610
     },
     {
       "epoch": 0.6459394264272313,
-      "grad_norm": 0.5047606229782104,
+      "grad_norm": 0.5040249228477478,
       "learning_rate": 6.712354706094452e-05,
-      "loss": 1.1066,
+      "loss": 1.1061,
       "step": 3615
     },
     {
       "epoch": 0.6468328419547932,
-      "grad_norm": 0.5137758851051331,
+      "grad_norm": 0.511741042137146,
       "learning_rate": 6.682913264812533e-05,
-      "loss": 1.1402,
+      "loss": 1.14,
       "step": 3620
     },
     {
       "epoch": 0.647726257482355,
-      "grad_norm": 0.5158393383026123,
+      "grad_norm": 0.5170619487762451,
       "learning_rate": 6.653504095449305e-05,
-      "loss": 1.1016,
+      "loss": 1.1019,
       "step": 3625
     },
     {
       "epoch": 0.648619673009917,
-      "grad_norm": 0.5306075811386108,
+      "grad_norm": 0.5282120704650879,
       "learning_rate": 6.624127484126421e-05,
-      "loss": 1.1417,
+      "loss": 1.1406,
       "step": 3630
     },
     {
       "epoch": 0.6495130885374788,
-      "grad_norm": 0.5093557238578796,
+      "grad_norm": 0.5090754628181458,
       "learning_rate": 6.594783716648769e-05,
-      "loss": 1.1397,
+      "loss": 1.1396,
       "step": 3635
     },
     {
       "epoch": 0.6504065040650406,
-      "grad_norm": 0.4910166561603546,
+      "grad_norm": 0.48920685052871704,
       "learning_rate": 6.565473078501698e-05,
-      "loss": 1.1187,
+      "loss": 1.1186,
       "step": 3640
     },
     {
       "epoch": 0.6512999195926025,
-      "grad_norm": 0.5265088677406311,
+      "grad_norm": 0.5252284407615662,
       "learning_rate": 6.536195854848248e-05,
-      "loss": 1.0743,
+      "loss": 1.0744,
       "step": 3645
     },
     {
       "epoch": 0.6521933351201644,
-      "grad_norm": 0.518035888671875,
+      "grad_norm": 0.5125118494033813,
       "learning_rate": 6.506952330526373e-05,
-      "loss": 1.1224,
+      "loss": 1.1227,
       "step": 3650
     },
     {
       "epoch": 0.6530867506477263,
-      "grad_norm": 0.5047338604927063,
+      "grad_norm": 0.5054253339767456,
       "learning_rate": 6.477742790046152e-05,
-      "loss": 1.1183,
+      "loss": 1.1191,
       "step": 3655
     },
     {
       "epoch": 0.6539801661752881,
-      "grad_norm": 0.5210195779800415,
+      "grad_norm": 0.5262144804000854,
       "learning_rate": 6.448567517587052e-05,
-      "loss": 1.1546,
+      "loss": 1.1542,
       "step": 3660
     },
     {
       "epoch": 0.65487358170285,
-      "grad_norm": 0.5376661419868469,
+      "grad_norm": 0.5490496158599854,
       "learning_rate": 6.419426796995137e-05,
-      "loss": 1.147,
+      "loss": 1.1468,
       "step": 3665
     },
     {
       "epoch": 0.6557669972304119,
-      "grad_norm": 0.5363953709602356,
+      "grad_norm": 0.5373784303665161,
       "learning_rate": 6.390320911780317e-05,
-      "loss": 1.1399,
+      "loss": 1.1393,
       "step": 3670
     },
     {
       "epoch": 0.6566604127579737,
-      "grad_norm": 0.5107960104942322,
+      "grad_norm": 0.5126230716705322,
       "learning_rate": 6.36125014511359e-05,
-      "loss": 1.1382,
+      "loss": 1.1385,
       "step": 3675
     },
     {
       "epoch": 0.6575538282855357,
-      "grad_norm": 0.511021614074707,
+      "grad_norm": 0.5112528204917908,
       "learning_rate": 6.332214779824288e-05,
-      "loss": 1.1503,
+      "loss": 1.1505,
       "step": 3680
     },
     {
       "epoch": 0.6584472438130975,
-      "grad_norm": 0.5405541062355042,
+      "grad_norm": 0.539935827255249,
       "learning_rate": 6.303215098397321e-05,
-      "loss": 1.1043,
+      "loss": 1.1044,
       "step": 3685
     },
     {
       "epoch": 0.6593406593406593,
-      "grad_norm": 0.542253851890564,
+      "grad_norm": 0.5397330522537231,
       "learning_rate": 6.274251382970427e-05,
-      "loss": 1.1344,
+      "loss": 1.1336,
       "step": 3690
     },
     {
       "epoch": 0.6602340748682212,
-      "grad_norm": 0.5661068558692932,
+      "grad_norm": 0.5695128440856934,
       "learning_rate": 6.245323915331439e-05,
-      "loss": 1.1463,
+      "loss": 1.1468,
       "step": 3695
     },
     {
       "epoch": 0.6611274903957831,
-      "grad_norm": 0.5561456680297852,
+      "grad_norm": 0.5581777095794678,
       "learning_rate": 6.216432976915527e-05,
-      "loss": 1.1655,
+      "loss": 1.1658,
       "step": 3700
     },
     {
       "epoch": 0.662020905923345,
-      "grad_norm": 0.5392094254493713,
+      "grad_norm": 0.5400391817092896,
       "learning_rate": 6.187578848802475e-05,
-      "loss": 1.1092,
+      "loss": 1.1089,
       "step": 3705
     },
     {
       "epoch": 0.6629143214509068,
-      "grad_norm": 0.54625403881073,
+      "grad_norm": 0.5398588180541992,
       "learning_rate": 6.15876181171394e-05,
-      "loss": 1.1417,
+      "loss": 1.1412,
       "step": 3710
     },
     {
       "epoch": 0.6638077369784687,
-      "grad_norm": 0.5339346528053284,
+      "grad_norm": 0.5350549817085266,
       "learning_rate": 6.129982146010713e-05,
-      "loss": 1.1493,
+      "loss": 1.1489,
       "step": 3715
     },
     {
       "epoch": 0.6647011525060306,
-      "grad_norm": 0.4904159605503082,
+      "grad_norm": 0.49134427309036255,
       "learning_rate": 6.101240131690009e-05,
-      "loss": 1.1063,
+      "loss": 1.1069,
       "step": 3720
     },
     {
       "epoch": 0.6655945680335924,
-      "grad_norm": 0.5268272161483765,
+      "grad_norm": 0.526871383190155,
       "learning_rate": 6.072536048382726e-05,
-      "loss": 1.1261,
+      "loss": 1.1264,
       "step": 3725
     },
     {
       "epoch": 0.6664879835611542,
-      "grad_norm": 0.5363719463348389,
+      "grad_norm": 0.539218008518219,
       "learning_rate": 6.043870175350732e-05,
-      "loss": 1.1581,
+      "loss": 1.1582,
       "step": 3730
     },
     {
       "epoch": 0.6673813990887162,
-      "grad_norm": 0.5529670715332031,
+      "grad_norm": 0.5474951267242432,
       "learning_rate": 6.0152427914841544e-05,
       "loss": 1.112,
       "step": 3735
     },
     {
       "epoch": 0.668274814616278,
-      "grad_norm": 0.49713894724845886,
+      "grad_norm": 0.49627193808555603,
       "learning_rate": 5.9866541752986485e-05,
-      "loss": 1.1057,
+      "loss": 1.1059,
       "step": 3740
     },
     {
       "epoch": 0.6691682301438399,
-      "grad_norm": 0.5073032975196838,
+      "grad_norm": 0.5210728645324707,
       "learning_rate": 5.958104604932706e-05,
-      "loss": 1.1555,
+      "loss": 1.1543,
       "step": 3745
     },
     {
       "epoch": 0.6700616456714018,
-      "grad_norm": 0.5101709961891174,
+      "grad_norm": 0.510959267616272,
       "learning_rate": 5.9295943581449385e-05,
-      "loss": 1.1366,
+      "loss": 1.1369,
       "step": 3750
     },
     {
       "epoch": 0.6709550611989636,
-      "grad_norm": 0.5283429026603699,
+      "grad_norm": 0.5323443412780762,
       "learning_rate": 5.901123712311385e-05,
-      "loss": 1.1178,
+      "loss": 1.1186,
       "step": 3755
     },
     {
       "epoch": 0.6718484767265255,
-      "grad_norm": 0.5022781491279602,
+      "grad_norm": 0.5052646398544312,
       "learning_rate": 5.8726929444228016e-05,
-      "loss": 1.1079,
+      "loss": 1.1084,
       "step": 3760
     },
     {
       "epoch": 0.6727418922540874,
-      "grad_norm": 0.5120312571525574,
+      "grad_norm": 0.5128947496414185,
       "learning_rate": 5.844302331081972e-05,
       "loss": 1.1218,
       "step": 3765
     },
     {
       "epoch": 0.6736353077816493,
-      "grad_norm": 0.48626720905303955,
+      "grad_norm": 0.48404374718666077,
       "learning_rate": 5.8159521485010214e-05,
-      "loss": 1.1053,
+      "loss": 1.1055,
       "step": 3770
     },
     {
       "epoch": 0.6745287233092111,
-      "grad_norm": 0.5270091891288757,
+      "grad_norm": 0.5284962058067322,
       "learning_rate": 5.787642672498719e-05,
-      "loss": 1.1096,
+      "loss": 1.1089,
       "step": 3775
     },
     {
       "epoch": 0.6754221388367729,
-      "grad_norm": 0.5076459050178528,
+      "grad_norm": 0.49954989552497864,
       "learning_rate": 5.759374178497801e-05,
-      "loss": 1.0824,
+      "loss": 1.0823,
       "step": 3780
     },
     {
       "epoch": 0.6763155543643349,
-      "grad_norm": 0.5193483829498291,
+      "grad_norm": 0.5159726142883301,
       "learning_rate": 5.731146941522292e-05,
-      "loss": 1.1143,
+      "loss": 1.1137,
       "step": 3785
     },
     {
       "epoch": 0.6772089698918967,
-      "grad_norm": 0.5338672399520874,
+      "grad_norm": 0.5402414202690125,
       "learning_rate": 5.702961236194826e-05,
-      "loss": 1.1068,
+      "loss": 1.1067,
       "step": 3790
     },
     {
       "epoch": 0.6781023854194586,
-      "grad_norm": 0.5404512286186218,
+      "grad_norm": 0.5399303436279297,
       "learning_rate": 5.674817336733975e-05,
-      "loss": 1.107,
+      "loss": 1.106,
       "step": 3795
     },
     {
       "epoch": 0.6789958009470205,
-      "grad_norm": 0.5492522120475769,
+      "grad_norm": 0.557214081287384,
       "learning_rate": 5.646715516951584e-05,
-      "loss": 1.1108,
+      "loss": 1.111,
       "step": 3800
     },
     {
       "epoch": 0.6798892164745823,
-      "grad_norm": 0.561931848526001,
+      "grad_norm": 0.5638577342033386,
       "learning_rate": 5.618656050250099e-05,
-      "loss": 1.1257,
+      "loss": 1.1255,
       "step": 3805
     },
     {
       "epoch": 0.6807826320021442,
-      "grad_norm": 0.5259913206100464,
+      "grad_norm": 0.5255394577980042,
       "learning_rate": 5.5906392096199255e-05,
       "loss": 1.0976,
       "step": 3810
     },
     {
       "epoch": 0.6816760475297061,
-      "grad_norm": 0.5238481163978577,
+      "grad_norm": 0.5245288610458374,
       "learning_rate": 5.562665267636751e-05,
-      "loss": 1.1711,
+      "loss": 1.1709,
       "step": 3815
     },
     {
       "epoch": 0.682569463057268,
-      "grad_norm": 0.5467326045036316,
+      "grad_norm": 0.5460274815559387,
       "learning_rate": 5.5347344964588996e-05,
-      "loss": 1.0727,
+      "loss": 1.0736,
       "step": 3820
     },
     {
       "epoch": 0.6834628785848298,
-      "grad_norm": 0.5232309699058533,
+      "grad_norm": 0.5235292911529541,
       "learning_rate": 5.506847167824696e-05,
-      "loss": 1.1064,
+      "loss": 1.1055,
       "step": 3825
     },
     {
       "epoch": 0.6843562941123916,
-      "grad_norm": 0.5298530459403992,
+      "grad_norm": 0.5296208262443542,
       "learning_rate": 5.479003553049806e-05,
-      "loss": 1.1082,
+      "loss": 1.1081,
       "step": 3830
     },
     {
       "epoch": 0.6852497096399536,
-      "grad_norm": 0.506592333316803,
+      "grad_norm": 0.48945724964141846,
       "learning_rate": 5.4512039230246035e-05,
-      "loss": 1.0942,
+      "loss": 1.0935,
       "step": 3835
     },
     {
       "epoch": 0.6861431251675154,
-      "grad_norm": 0.5119116306304932,
+      "grad_norm": 0.5146022439002991,
       "learning_rate": 5.42344854821154e-05,
-      "loss": 1.1278,
+      "loss": 1.128,
       "step": 3840
     },
     {
       "epoch": 0.6870365406950772,
-      "grad_norm": 0.5258318185806274,
+      "grad_norm": 0.5302464962005615,
       "learning_rate": 5.395737698642503e-05,
-      "loss": 1.1161,
+      "loss": 1.1157,
       "step": 3845
     },
     {
       "epoch": 0.6879299562226392,
-      "grad_norm": 0.5019605159759521,
+      "grad_norm": 0.502776026725769,
       "learning_rate": 5.368071643916194e-05,
-      "loss": 1.1363,
+      "loss": 1.1372,
       "step": 3850
     },
     {
       "epoch": 0.688823371750201,
-      "grad_norm": 0.47688642144203186,
+      "grad_norm": 0.47847309708595276,
       "learning_rate": 5.3404506531955146e-05,
-      "loss": 1.0935,
+      "loss": 1.0927,
       "step": 3855
     },
     {
       "epoch": 0.6897167872777629,
-      "grad_norm": 0.5404648184776306,
+      "grad_norm": 0.5378836989402771,
       "learning_rate": 5.3128749952049284e-05,
-      "loss": 1.1134,
+      "loss": 1.1137,
       "step": 3860
     },
     {
       "epoch": 0.6906102028053247,
-      "grad_norm": 0.5106511116027832,
+      "grad_norm": 0.5127781629562378,
       "learning_rate": 5.2853449382278605e-05,
-      "loss": 1.1165,
+      "loss": 1.1158,
       "step": 3865
     },
     {
       "epoch": 0.6915036183328866,
-      "grad_norm": 0.5181242227554321,
+      "grad_norm": 0.5223949551582336,
       "learning_rate": 5.2578607501040863e-05,
-      "loss": 1.104,
+      "loss": 1.1043,
       "step": 3870
     },
     {
       "epoch": 0.6923970338604485,
-      "grad_norm": 0.5107550024986267,
+      "grad_norm": 0.5131104588508606,
       "learning_rate": 5.2304226982271174e-05,
-      "loss": 1.1438,
+      "loss": 1.1444,
       "step": 3875
     },
     {
       "epoch": 0.6932904493880103,
-      "grad_norm": 0.5244885683059692,
+      "grad_norm": 0.5290147662162781,
       "learning_rate": 5.203031049541621e-05,
-      "loss": 1.1284,
+      "loss": 1.1286,
       "step": 3880
     },
     {
       "epoch": 0.6941838649155723,
-      "grad_norm": 0.519130289554596,
+      "grad_norm": 0.5196212530136108,
       "learning_rate": 5.175686070540786e-05,
-      "loss": 1.1188,
+      "loss": 1.1189,
       "step": 3885
     },
     {
       "epoch": 0.6950772804431341,
-      "grad_norm": 0.5156662464141846,
+      "grad_norm": 0.517926037311554,
       "learning_rate": 5.148388027263769e-05,
       "loss": 1.1648,
       "step": 3890
     },
     {
       "epoch": 0.6959706959706959,
-      "grad_norm": 0.5092237591743469,
+      "grad_norm": 0.5101890563964844,
       "learning_rate": 5.1211371852930766e-05,
-      "loss": 1.0715,
+      "loss": 1.0711,
       "step": 3895
     },
     {
       "epoch": 0.6968641114982579,
-      "grad_norm": 0.544964075088501,
+      "grad_norm": 0.5416128039360046,
       "learning_rate": 5.0939338097520095e-05,
-      "loss": 1.114,
+      "loss": 1.1141,
       "step": 3900
     },
     {
       "epoch": 0.6977575270258197,
-      "grad_norm": 0.5449029207229614,
+      "grad_norm": 0.5422283411026001,
       "learning_rate": 5.0667781653020584e-05,
-      "loss": 1.0786,
+      "loss": 1.0783,
       "step": 3905
     },
     {
       "epoch": 0.6986509425533816,
-      "grad_norm": 0.5234874486923218,
+      "grad_norm": 0.5199434161186218,
       "learning_rate": 5.039670516140338e-05,
-      "loss": 1.1419,
+      "loss": 1.1424,
       "step": 3910
     },
     {
       "epoch": 0.6995443580809434,
-      "grad_norm": 0.5106183290481567,
+      "grad_norm": 0.5065299868583679,
       "learning_rate": 5.012611125997018e-05,
-      "loss": 1.0891,
+      "loss": 1.0881,
       "step": 3915
     },
     {
       "epoch": 0.7004377736085053,
-      "grad_norm": 0.49707773327827454,
+      "grad_norm": 0.49652373790740967,
       "learning_rate": 4.9856002581327565e-05,
-      "loss": 1.1272,
+      "loss": 1.1281,
       "step": 3920
     },
     {
       "epoch": 0.7013311891360672,
-      "grad_norm": 0.5646072626113892,
+      "grad_norm": 0.5640977621078491,
       "learning_rate": 4.958638175336137e-05,
-      "loss": 1.1029,
+      "loss": 1.1016,
       "step": 3925
     },
     {
       "epoch": 0.702224604663629,
-      "grad_norm": 0.5331147313117981,
+      "grad_norm": 0.5382977724075317,
       "learning_rate": 4.931725139921126e-05,
       "loss": 1.1271,
       "step": 3930
     },
     {
       "epoch": 0.703118020191191,
-      "grad_norm": 0.521113932132721,
+      "grad_norm": 0.5227957963943481,
       "learning_rate": 4.9048614137244865e-05,
-      "loss": 1.0927,
+      "loss": 1.0931,
       "step": 3935
     },
     {
       "epoch": 0.7040114357187528,
-      "grad_norm": 0.5384479761123657,
+      "grad_norm": 0.5395195484161377,
       "learning_rate": 4.878047258103267e-05,
-      "loss": 1.1262,
+      "loss": 1.1264,
       "step": 3940
     },
     {
       "epoch": 0.7049048512463146,
-      "grad_norm": 0.5266872644424438,
+      "grad_norm": 0.5258668065071106,
       "learning_rate": 4.8512829339322375e-05,
-      "loss": 1.1521,
+      "loss": 1.1525,
       "step": 3945
     },
     {
       "epoch": 0.7057982667738766,
-      "grad_norm": 0.5195939540863037,
+      "grad_norm": 0.5186535716056824,
       "learning_rate": 4.8245687016013696e-05,
-      "loss": 1.1201,
+      "loss": 1.119,
       "step": 3950
     },
     {
       "epoch": 0.7066916823014384,
-      "grad_norm": 0.49455180764198303,
+      "grad_norm": 0.4964883327484131,
       "learning_rate": 4.797904821013278e-05,
       "loss": 1.1118,
       "step": 3955
     },
     {
       "epoch": 0.7075850978290003,
-      "grad_norm": 0.49879488348960876,
+      "grad_norm": 0.4993223547935486,
       "learning_rate": 4.771291551580712e-05,
-      "loss": 1.1106,
+      "loss": 1.1115,
       "step": 3960
     },
     {
       "epoch": 0.7084785133565621,
-      "grad_norm": 0.5107743740081787,
+      "grad_norm": 0.5137820243835449,
       "learning_rate": 4.744729152224024e-05,
-      "loss": 1.0964,
+      "loss": 1.0969,
       "step": 3965
     },
     {
       "epoch": 0.709371928884124,
-      "grad_norm": 0.4998965263366699,
+      "grad_norm": 0.49753323197364807,
       "learning_rate": 4.71821788136865e-05,
-      "loss": 1.0842,
+      "loss": 1.0846,
       "step": 3970
     },
     {
       "epoch": 0.7102653444116859,
-      "grad_norm": 0.5038910508155823,
+      "grad_norm": 0.5026699900627136,
       "learning_rate": 4.691757996942607e-05,
-      "loss": 1.1108,
+      "loss": 1.1105,
       "step": 3975
     },
     {
       "epoch": 0.7111587599392477,
-      "grad_norm": 0.5343025326728821,
+      "grad_norm": 0.5345607995986938,
       "learning_rate": 4.665349756373957e-05,
-      "loss": 1.1084,
+      "loss": 1.1079,
       "step": 3980
     },
     {
       "epoch": 0.7120521754668097,
-      "grad_norm": 0.5399016737937927,
+      "grad_norm": 0.5438992977142334,
       "learning_rate": 4.6389934165883306e-05,
-      "loss": 1.1055,
+      "loss": 1.1048,
       "step": 3985
     },
     {
       "epoch": 0.7129455909943715,
-      "grad_norm": 0.49205487966537476,
+      "grad_norm": 0.48879095911979675,
       "learning_rate": 4.6126892340064096e-05,
-      "loss": 1.1041,
+      "loss": 1.1043,
       "step": 3990
     },
     {
       "epoch": 0.7138390065219333,
-      "grad_norm": 0.5059479475021362,
+      "grad_norm": 0.5040327310562134,
       "learning_rate": 4.586437464541451e-05,
-      "loss": 1.1028,
+      "loss": 1.1026,
       "step": 3995
     },
     {
       "epoch": 0.7147324220494953,
-      "grad_norm": 0.520707368850708,
+      "grad_norm": 0.5185865759849548,
       "learning_rate": 4.560238363596771e-05,
-      "loss": 1.064,
+      "loss": 1.0629,
       "step": 4000
     },
     {
       "epoch": 0.7156258375770571,
-      "grad_norm": 0.46862202882766724,
+      "grad_norm": 0.4688703119754791,
       "learning_rate": 4.53409218606328e-05,
-      "loss": 1.116,
+      "loss": 1.1158,
       "step": 4005
     },
     {
       "epoch": 0.716519253104619,
-      "grad_norm": 0.5046384930610657,
+      "grad_norm": 0.5067079663276672,
       "learning_rate": 4.5079991863169966e-05,
-      "loss": 1.0911,
+      "loss": 1.0912,
       "step": 4010
     },
     {
       "epoch": 0.7174126686321808,
-      "grad_norm": 0.5321259498596191,
+      "grad_norm": 0.5316655039787292,
       "learning_rate": 4.481959618216568e-05,
-      "loss": 1.0903,
+      "loss": 1.0901,
       "step": 4015
     },
     {
       "epoch": 0.7183060841597427,
-      "grad_norm": 0.5124063491821289,
+      "grad_norm": 0.5138748288154602,
       "learning_rate": 4.455973735100818e-05,
-      "loss": 1.1261,
+      "loss": 1.1262,
       "step": 4020
     },
     {
       "epoch": 0.7191994996873046,
-      "grad_norm": 0.5123804211616516,
+      "grad_norm": 0.5119633674621582,
       "learning_rate": 4.43004178978626e-05,
-      "loss": 1.1248,
+      "loss": 1.125,
       "step": 4025
     },
     {
       "epoch": 0.7200929152148664,
-      "grad_norm": 0.5785008668899536,
+      "grad_norm": 0.5808505415916443,
       "learning_rate": 4.404164034564641e-05,
       "loss": 1.1447,
       "step": 4030
     },
     {
       "epoch": 0.7209863307424283,
-      "grad_norm": 0.537517786026001,
+      "grad_norm": 0.5414739847183228,
       "learning_rate": 4.378340721200501e-05,
-      "loss": 1.1067,
+      "loss": 1.1063,
       "step": 4035
     },
     {
       "epoch": 0.7218797462699902,
-      "grad_norm": 0.5579099655151367,
+      "grad_norm": 0.5619556903839111,
       "learning_rate": 4.3525721009287215e-05,
-      "loss": 1.1124,
+      "loss": 1.1123,
       "step": 4040
     },
     {
       "epoch": 0.722773161797552,
-      "grad_norm": 0.5302057266235352,
+      "grad_norm": 0.5197097063064575,
       "learning_rate": 4.326858424452063e-05,
       "loss": 1.0999,
       "step": 4045
     },
     {
       "epoch": 0.7236665773251139,
-      "grad_norm": 0.5399320125579834,
+      "grad_norm": 0.5409570336341858,
       "learning_rate": 4.301199941938744e-05,
-      "loss": 1.0849,
+      "loss": 1.0848,
       "step": 4050
     },
     {
       "epoch": 0.7245599928526758,
-      "grad_norm": 0.5739198923110962,
+      "grad_norm": 0.580436110496521,
       "learning_rate": 4.275596903020001e-05,
-      "loss": 1.118,
+      "loss": 1.1184,
       "step": 4055
     },
     {
       "epoch": 0.7254534083802376,
-      "grad_norm": 0.5343251824378967,
+      "grad_norm": 0.5301364660263062,
       "learning_rate": 4.250049556787655e-05,
-      "loss": 1.1116,
+      "loss": 1.1118,
       "step": 4060
     },
     {
       "epoch": 0.7263468239077995,
-      "grad_norm": 0.5469746589660645,
+      "grad_norm": 0.5562632083892822,
       "learning_rate": 4.2245581517917065e-05,
-      "loss": 1.0922,
+      "loss": 1.0929,
       "step": 4065
     },
     {
       "epoch": 0.7272402394353614,
-      "grad_norm": 0.5611655712127686,
+      "grad_norm": 0.5643694996833801,
       "learning_rate": 4.199122936037889e-05,
-      "loss": 1.1282,
+      "loss": 1.128,
       "step": 4070
     },
     {
       "epoch": 0.7281336549629233,
-      "grad_norm": 0.5936520099639893,
+      "grad_norm": 0.5907604098320007,
       "learning_rate": 4.173744156985283e-05,
-      "loss": 1.135,
+      "loss": 1.1354,
       "step": 4075
     },
     {
       "epoch": 0.7290270704904851,
-      "grad_norm": 0.5532947778701782,
+      "grad_norm": 0.5513320565223694,
       "learning_rate": 4.148422061543884e-05,
-      "loss": 1.123,
+      "loss": 1.1238,
       "step": 4080
     },
     {
       "epoch": 0.729920486018047,
-      "grad_norm": 0.5437267422676086,
+      "grad_norm": 0.5430623292922974,
       "learning_rate": 4.123156896072219e-05,
-      "loss": 1.084,
+      "loss": 1.0839,
       "step": 4085
     },
     {
       "epoch": 0.7308139015456089,
-      "grad_norm": 0.5226598978042603,
+      "grad_norm": 0.5164535045623779,
       "learning_rate": 4.097948906374951e-05,
       "loss": 1.1311,
       "step": 4090
     },
     {
       "epoch": 0.7317073170731707,
-      "grad_norm": 0.5470917820930481,
+      "grad_norm": 0.5333124399185181,
       "learning_rate": 4.0727983377004716e-05,
-      "loss": 1.1086,
+      "loss": 1.1088,
       "step": 4095
     },
     {
       "epoch": 0.7326007326007326,
-      "grad_norm": 0.5484129190444946,
+      "grad_norm": 0.5494021773338318,
       "learning_rate": 4.047705434738527e-05,
-      "loss": 1.0744,
+      "loss": 1.0739,
       "step": 4100
     },
     {
       "epoch": 0.7334941481282945,
-      "grad_norm": 0.5352585911750793,
+      "grad_norm": 0.5382459759712219,
       "learning_rate": 4.02267044161783e-05,
-      "loss": 1.1017,
+      "loss": 1.102,
       "step": 4105
     },
     {
       "epoch": 0.7343875636558563,
-      "grad_norm": 0.5201607942581177,
+      "grad_norm": 0.5144065022468567,
       "learning_rate": 3.997693601903688e-05,
       "loss": 1.1283,
       "step": 4110
     },
     {
       "epoch": 0.7352809791834182,
-      "grad_norm": 0.5287265181541443,
+      "grad_norm": 0.5298936367034912,
       "learning_rate": 3.9727751585956477e-05,
-      "loss": 1.1334,
+      "loss": 1.1328,
       "step": 4115
     },
     {
       "epoch": 0.7361743947109801,
-      "grad_norm": 0.5448318123817444,
+      "grad_norm": 0.5441993474960327,
       "learning_rate": 3.9479153541251056e-05,
-      "loss": 1.1403,
+      "loss": 1.1399,
       "step": 4120
     },
     {
       "epoch": 0.737067810238542,
-      "grad_norm": 0.5017204880714417,
+      "grad_norm": 0.5025814175605774,
       "learning_rate": 3.923114430352958e-05,
       "loss": 1.126,
       "step": 4125
     },
     {
       "epoch": 0.7379612257661038,
-      "grad_norm": 0.5083391666412354,
+      "grad_norm": 0.512550950050354,
       "learning_rate": 3.8983726285672536e-05,
-      "loss": 1.1325,
+      "loss": 1.1322,
       "step": 4130
     },
     {
       "epoch": 0.7388546412936657,
-      "grad_norm": 0.4848039150238037,
+      "grad_norm": 0.48689234256744385,
       "learning_rate": 3.8736901894808575e-05,
-      "loss": 1.0555,
+      "loss": 1.0553,
       "step": 4135
     },
     {
       "epoch": 0.7397480568212276,
-      "grad_norm": 0.5136265158653259,
+      "grad_norm": 0.5119283199310303,
       "learning_rate": 3.849067353229078e-05,
-      "loss": 1.1228,
+      "loss": 1.1224,
       "step": 4140
     },
     {
       "epoch": 0.7406414723487894,
-      "grad_norm": 0.5346769690513611,
+      "grad_norm": 0.5378916263580322,
       "learning_rate": 3.824504359367355e-05,
       "loss": 1.1229,
       "step": 4145
     },
     {
       "epoch": 0.7415348878763512,
-      "grad_norm": 0.5281243324279785,
+      "grad_norm": 0.524563193321228,
       "learning_rate": 3.80000144686892e-05,
-      "loss": 1.1327,
+      "loss": 1.1329,
       "step": 4150
     },
     {
       "epoch": 0.7424283034039132,
-      "grad_norm": 0.5599030256271362,
+      "grad_norm": 0.5638777017593384,
       "learning_rate": 3.775558854122475e-05,
-      "loss": 1.0789,
+      "loss": 1.0796,
       "step": 4155
     },
     {
       "epoch": 0.743321718931475,
-      "grad_norm": 0.5306710600852966,
+      "grad_norm": 0.5309896469116211,
       "learning_rate": 3.7511768189298746e-05,
-      "loss": 1.0885,
+      "loss": 1.088,
       "step": 4160
     },
     {
       "epoch": 0.7442151344590369,
-      "grad_norm": 0.5410204529762268,
+      "grad_norm": 0.5442638397216797,
       "learning_rate": 3.726855578503804e-05,
-      "loss": 1.0959,
+      "loss": 1.0958,
       "step": 4165
     },
     {
       "epoch": 0.7451085499865988,
-      "grad_norm": 0.5430173873901367,
+      "grad_norm": 0.5414631962776184,
       "learning_rate": 3.70259536946548e-05,
-      "loss": 1.1033,
+      "loss": 1.103,
       "step": 4170
     },
     {
       "epoch": 0.7460019655141606,
-      "grad_norm": 0.5353587865829468,
+      "grad_norm": 0.538829505443573,
       "learning_rate": 3.678396427842334e-05,
-      "loss": 1.121,
+      "loss": 1.1218,
       "step": 4175
     },
     {
       "epoch": 0.7468953810417225,
-      "grad_norm": 0.5129430890083313,
+      "grad_norm": 0.516124963760376,
       "learning_rate": 3.6542589890657476e-05,
-      "loss": 1.0958,
+      "loss": 1.0967,
       "step": 4180
     },
     {
       "epoch": 0.7477887965692843,
-      "grad_norm": 0.5174726843833923,
+      "grad_norm": 0.5191131234169006,
       "learning_rate": 3.630183287968727e-05,
-      "loss": 1.1203,
+      "loss": 1.1197,
       "step": 4185
     },
     {
       "epoch": 0.7486822120968463,
-      "grad_norm": 0.5343511700630188,
+      "grad_norm": 0.5277869701385498,
       "learning_rate": 3.606169558783635e-05,
       "loss": 1.0792,
       "step": 4190
     },
     {
       "epoch": 0.7495756276244081,
-      "grad_norm": 0.5478770136833191,
+      "grad_norm": 0.5471330285072327,
       "learning_rate": 3.5822180351399136e-05,
-      "loss": 1.0975,
+      "loss": 1.0985,
       "step": 4195
     },
     {
       "epoch": 0.7504690431519699,
-      "grad_norm": 0.4997379183769226,
+      "grad_norm": 0.5002536773681641,
       "learning_rate": 3.5583289500618e-05,
       "loss": 1.0913,
       "step": 4200
     },
     {
       "epoch": 0.7513624586795319,
-      "grad_norm": 0.5360389351844788,
+      "grad_norm": 0.5431557297706604,
       "learning_rate": 3.53450253596608e-05,
-      "loss": 1.081,
+      "loss": 1.0816,
       "step": 4205
     },
     {
       "epoch": 0.7522558742070937,
-      "grad_norm": 0.5671184062957764,
+      "grad_norm": 0.5701403617858887,
       "learning_rate": 3.510739024659802e-05,
-      "loss": 1.1355,
+      "loss": 1.1358,
       "step": 4210
     },
     {
       "epoch": 0.7531492897346556,
-      "grad_norm": 0.5735521912574768,
+      "grad_norm": 0.5729346871376038,
       "learning_rate": 3.487038647338038e-05,
-      "loss": 1.0998,
+      "loss": 1.0999,
       "step": 4215
     },
     {
       "epoch": 0.7540427052622175,
-      "grad_norm": 0.5425437092781067,
+      "grad_norm": 0.5427317023277283,
       "learning_rate": 3.463401634581631e-05,
-      "loss": 1.1192,
+      "loss": 1.1184,
       "step": 4220
     },
     {
       "epoch": 0.7549361207897793,
-      "grad_norm": 0.5446605086326599,
+      "grad_norm": 0.5459471344947815,
       "learning_rate": 3.4398282163549414e-05,
-      "loss": 1.1166,
+      "loss": 1.1179,
       "step": 4225
     },
     {
       "epoch": 0.7558295363173412,
-      "grad_norm": 0.5220611691474915,
+      "grad_norm": 0.5358048677444458,
       "learning_rate": 3.416318622003634e-05,
-      "loss": 1.0814,
+      "loss": 1.0815,
       "step": 4230
     },
     {
       "epoch": 0.756722951844903,
-      "grad_norm": 0.513034999370575,
+      "grad_norm": 0.5081732869148254,
       "learning_rate": 3.39287308025242e-05,
-      "loss": 1.1026,
+      "loss": 1.1027,
       "step": 4235
     },
     {
       "epoch": 0.757616367372465,
-      "grad_norm": 0.5283447504043579,
+      "grad_norm": 0.5418072938919067,
       "learning_rate": 3.369491819202849e-05,
-      "loss": 1.1125,
+      "loss": 1.112,
       "step": 4240
     },
     {
       "epoch": 0.7585097829000268,
-      "grad_norm": 0.5065436363220215,
+      "grad_norm": 0.5014515519142151,
       "learning_rate": 3.34617506633108e-05,
-      "loss": 1.1106,
+      "loss": 1.1101,
       "step": 4245
     },
     {
       "epoch": 0.7594031984275886,
-      "grad_norm": 0.5315711498260498,
+      "grad_norm": 0.5339967608451843,
       "learning_rate": 3.322923048485672e-05,
-      "loss": 1.0821,
+      "loss": 1.0815,
       "step": 4250
     },
     {
       "epoch": 0.7602966139551506,
-      "grad_norm": 0.5503782033920288,
+      "grad_norm": 0.5515729188919067,
       "learning_rate": 3.2997359918853845e-05,
-      "loss": 1.1076,
+      "loss": 1.1077,
       "step": 4255
     },
     {
       "epoch": 0.7611900294827124,
-      "grad_norm": 0.5607456564903259,
+      "grad_norm": 0.5655061602592468,
       "learning_rate": 3.276614122116962e-05,
-      "loss": 1.0913,
+      "loss": 1.0921,
       "step": 4260
     },
     {
       "epoch": 0.7620834450102743,
-      "grad_norm": 0.536376953125,
+      "grad_norm": 0.5367369055747986,
       "learning_rate": 3.2535576641329514e-05,
-      "loss": 1.0911,
+      "loss": 1.091,
       "step": 4265
     },
     {
       "epoch": 0.7629768605378362,
-      "grad_norm": 0.5168653726577759,
+      "grad_norm": 0.5201234817504883,
       "learning_rate": 3.230566842249497e-05,
-      "loss": 1.0845,
+      "loss": 1.0846,
       "step": 4270
     },
     {
       "epoch": 0.763870276065398,
-      "grad_norm": 0.5393795371055603,
+      "grad_norm": 0.5395834445953369,
       "learning_rate": 3.2076418801441886e-05,
-      "loss": 1.098,
+      "loss": 1.0972,
       "step": 4275
     },
     {
       "epoch": 0.7647636915929599,
-      "grad_norm": 0.538724422454834,
+      "grad_norm": 0.5469521284103394,
       "learning_rate": 3.1847830008538545e-05,
-      "loss": 1.096,
+      "loss": 1.0966,
       "step": 4280
     },
     {
       "epoch": 0.7656571071205217,
-      "grad_norm": 0.527229905128479,
+      "grad_norm": 0.5283955931663513,
       "learning_rate": 3.1619904267724065e-05,
-      "loss": 1.0845,
+      "loss": 1.084,
       "step": 4285
     },
     {
       "epoch": 0.7665505226480837,
-      "grad_norm": 0.5256169438362122,
+      "grad_norm": 0.5325857996940613,
       "learning_rate": 3.139264379648671e-05,
-      "loss": 1.0796,
+      "loss": 1.0797,
       "step": 4290
     },
     {
       "epoch": 0.7674439381756455,
-      "grad_norm": 0.540058434009552,
+      "grad_norm": 0.5412785410881042,
       "learning_rate": 3.116605080584235e-05,
-      "loss": 1.0909,
+      "loss": 1.0911,
       "step": 4295
     },
     {
       "epoch": 0.7683373537032073,
-      "grad_norm": 0.5083367824554443,
+      "grad_norm": 0.5118827819824219,
       "learning_rate": 3.0940127500313e-05,
-      "loss": 1.1013,
+      "loss": 1.1005,
       "step": 4300
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 0.5150957703590393,
+      "grad_norm": 0.5167849063873291,
       "learning_rate": 3.071487607790524e-05,
-      "loss": 1.1422,
+      "loss": 1.143,
       "step": 4305
     },
     {
       "epoch": 0.7701241847583311,
-      "grad_norm": 0.5053666234016418,
+      "grad_norm": 0.5016911625862122,
       "learning_rate": 3.049029873008893e-05,
-      "loss": 1.0878,
+      "loss": 1.0874,
       "step": 4310
     },
     {
       "epoch": 0.771017600285893,
-      "grad_norm": 0.5635530352592468,
+      "grad_norm": 0.5407574772834778,
       "learning_rate": 3.0266397641775835e-05,
       "loss": 1.106,
       "step": 4315
     },
     {
       "epoch": 0.7719110158134549,
-      "grad_norm": 0.5005871653556824,
+      "grad_norm": 0.5007250905036926,
       "learning_rate": 3.004317499129845e-05,
-      "loss": 1.1155,
+      "loss": 1.1162,
       "step": 4320
     },
     {
       "epoch": 0.7728044313410167,
-      "grad_norm": 0.5285431742668152,
+      "grad_norm": 0.5197978019714355,
       "learning_rate": 2.9820632950388695e-05,
       "loss": 1.0865,
       "step": 4325
     },
     {
       "epoch": 0.7736978468685786,
-      "grad_norm": 0.5447906851768494,
+      "grad_norm": 0.5517399311065674,
       "learning_rate": 2.9598773684156878e-05,
-      "loss": 1.1002,
+      "loss": 1.1008,
       "step": 4330
     },
     {
       "epoch": 0.7745912623961404,
-      "grad_norm": 0.5144314765930176,
+      "grad_norm": 0.5117592215538025,
       "learning_rate": 2.9377599351070595e-05,
-      "loss": 1.1234,
+      "loss": 1.1244,
       "step": 4335
     },
     {
       "epoch": 0.7754846779237023,
-      "grad_norm": 0.5474838614463806,
+      "grad_norm": 0.5437847971916199,
       "learning_rate": 2.915711210293367e-05,
-      "loss": 1.1577,
+      "loss": 1.1578,
       "step": 4340
     },
     {
       "epoch": 0.7763780934512642,
-      "grad_norm": 0.5402074456214905,
+      "grad_norm": 0.5368949770927429,
       "learning_rate": 2.8937314084865407e-05,
-      "loss": 1.119,
+      "loss": 1.1185,
       "step": 4345
     },
     {
       "epoch": 0.777271508978826,
-      "grad_norm": 0.5311411023139954,
+      "grad_norm": 0.5332816243171692,
       "learning_rate": 2.8718207435279486e-05,
-      "loss": 1.1003,
+      "loss": 1.1008,
       "step": 4350
     },
     {
       "epoch": 0.778164924506388,
-      "grad_norm": 0.5577319264411926,
+      "grad_norm": 0.5594480633735657,
       "learning_rate": 2.849979428586331e-05,
       "loss": 1.0855,
       "step": 4355
     },
     {
       "epoch": 0.7790583400339498,
-      "grad_norm": 0.5216962099075317,
+      "grad_norm": 0.5220233201980591,
       "learning_rate": 2.828207676155722e-05,
-      "loss": 1.0935,
+      "loss": 1.0946,
       "step": 4360
     },
     {
       "epoch": 0.7799517555615116,
-      "grad_norm": 0.508258044719696,
+      "grad_norm": 0.5092322826385498,
       "learning_rate": 2.80650569805338e-05,
-      "loss": 1.0922,
+      "loss": 1.0925,
       "step": 4365
     },
     {
       "epoch": 0.7808451710890735,
-      "grad_norm": 0.5772315263748169,
+      "grad_norm": 0.5742801427841187,
       "learning_rate": 2.784873705417731e-05,
-      "loss": 1.0749,
+      "loss": 1.0748,
       "step": 4370
     },
     {
       "epoch": 0.7817385866166354,
-      "grad_norm": 0.5212002992630005,
+      "grad_norm": 0.5175203680992126,
       "learning_rate": 2.7633119087063152e-05,
       "loss": 1.1221,
       "step": 4375
     },
     {
       "epoch": 0.7826320021441973,
-      "grad_norm": 0.5161262154579163,
+      "grad_norm": 0.5183749794960022,
       "learning_rate": 2.7418205176937327e-05,
-      "loss": 1.1418,
+      "loss": 1.1412,
       "step": 4380
     },
     {
       "epoch": 0.7835254176717591,
-      "grad_norm": 0.5041150450706482,
+      "grad_norm": 0.5045965909957886,
       "learning_rate": 2.7203997414696104e-05,
       "loss": 1.1163,
       "step": 4385
     },
     {
       "epoch": 0.784418833199321,
-      "grad_norm": 0.5260159373283386,
+      "grad_norm": 0.5237119793891907,
       "learning_rate": 2.6990497884365586e-05,
-      "loss": 1.0596,
+      "loss": 1.0593,
       "step": 4390
     },
     {
       "epoch": 0.7853122487268829,
-      "grad_norm": 0.5398173332214355,
+      "grad_norm": 0.5403294563293457,
       "learning_rate": 2.67777086630816e-05,
-      "loss": 1.1323,
+      "loss": 1.1331,
       "step": 4395
     },
     {
       "epoch": 0.7862056642544447,
-      "grad_norm": 0.529534101486206,
+      "grad_norm": 0.5277557373046875,
       "learning_rate": 2.6565631821069304e-05,
       "loss": 1.0992,
       "step": 4400
     },
     {
       "epoch": 0.7870990797820067,
-      "grad_norm": 0.5247671008110046,
+      "grad_norm": 0.5190049409866333,
       "learning_rate": 2.6354269421623112e-05,
-      "loss": 1.1322,
+      "loss": 1.131,
       "step": 4405
     },
     {
       "epoch": 0.7879924953095685,
-      "grad_norm": 0.4856671690940857,
+      "grad_norm": 0.4882740080356598,
       "learning_rate": 2.6143623521086647e-05,
-      "loss": 1.0655,
+      "loss": 1.0654,
       "step": 4410
     },
     {
       "epoch": 0.7888859108371303,
-      "grad_norm": 0.5172522664070129,
+      "grad_norm": 0.5139947533607483,
       "learning_rate": 2.5933696168832743e-05,
-      "loss": 1.0842,
+      "loss": 1.0843,
       "step": 4415
     },
     {
       "epoch": 0.7897793263646922,
-      "grad_norm": 0.4878483712673187,
+      "grad_norm": 0.48566868901252747,
       "learning_rate": 2.5724489407243447e-05,
-      "loss": 1.1015,
+      "loss": 1.1022,
       "step": 4420
     },
     {
       "epoch": 0.7906727418922541,
-      "grad_norm": 0.5132545828819275,
+      "grad_norm": 0.5145934224128723,
       "learning_rate": 2.5516005271690203e-05,
-      "loss": 1.1243,
+      "loss": 1.1237,
       "step": 4425
     },
     {
       "epoch": 0.791566157419816,
-      "grad_norm": 0.5201311111450195,
+      "grad_norm": 0.51795893907547,
       "learning_rate": 2.530824579051403e-05,
-      "loss": 1.0829,
+      "loss": 1.0836,
       "step": 4430
     },
     {
       "epoch": 0.7924595729473778,
-      "grad_norm": 0.554908275604248,
+      "grad_norm": 0.5562942624092102,
       "learning_rate": 2.510121298500573e-05,
-      "loss": 1.1092,
+      "loss": 1.1093,
       "step": 4435
     },
     {
       "epoch": 0.7933529884749397,
-      "grad_norm": 0.5134171843528748,
+      "grad_norm": 0.5125170350074768,
       "learning_rate": 2.4894908869386424e-05,
-      "loss": 1.1063,
+      "loss": 1.1064,
       "step": 4440
     },
     {
       "epoch": 0.7942464040025016,
-      "grad_norm": 0.5175535678863525,
+      "grad_norm": 0.5201259851455688,
       "learning_rate": 2.4689335450787675e-05,
-      "loss": 1.1117,
+      "loss": 1.1108,
       "step": 4445
     },
     {
       "epoch": 0.7951398195300634,
-      "grad_norm": 0.543436586856842,
+      "grad_norm": 0.5387787222862244,
       "learning_rate": 2.4484494729232155e-05,
-      "loss": 1.0985,
+      "loss": 1.0988,
       "step": 4450
     },
     {
       "epoch": 0.7960332350576254,
-      "grad_norm": 0.5105295181274414,
+      "grad_norm": 0.5109546184539795,
       "learning_rate": 2.428038869761412e-05,
-      "loss": 1.0921,
+      "loss": 1.0924,
       "step": 4455
     },
     {
       "epoch": 0.7969266505851872,
-      "grad_norm": 0.524043619632721,
+      "grad_norm": 0.524856448173523,
       "learning_rate": 2.4077019341680042e-05,
-      "loss": 1.0728,
+      "loss": 1.0735,
       "step": 4460
     },
     {
       "epoch": 0.797820066112749,
-      "grad_norm": 0.5553478002548218,
+      "grad_norm": 0.5550493001937866,
       "learning_rate": 2.387438864000926e-05,
-      "loss": 1.1036,
+      "loss": 1.1041,
       "step": 4465
     },
     {
       "epoch": 0.7987134816403109,
-      "grad_norm": 0.4928396940231323,
+      "grad_norm": 0.49559786915779114,
       "learning_rate": 2.3672498563994762e-05,
-      "loss": 1.1025,
+      "loss": 1.1026,
       "step": 4470
     },
     {
       "epoch": 0.7996068971678728,
-      "grad_norm": 0.5613781213760376,
+      "grad_norm": 0.559374213218689,
       "learning_rate": 2.3471351077824e-05,
-      "loss": 1.1173,
+      "loss": 1.1167,
       "step": 4475
     },
     {
       "epoch": 0.8005003126954346,
-      "grad_norm": 0.5897471904754639,
+      "grad_norm": 0.5809065699577332,
       "learning_rate": 2.3270948138459735e-05,
-      "loss": 1.1225,
+      "loss": 1.1228,
       "step": 4480
     },
     {
       "epoch": 0.8013937282229965,
-      "grad_norm": 0.5354331135749817,
+      "grad_norm": 0.5337952971458435,
       "learning_rate": 2.3071291695621135e-05,
-      "loss": 1.0784,
+      "loss": 1.0787,
       "step": 4485
     },
     {
       "epoch": 0.8022871437505584,
-      "grad_norm": 0.5370863676071167,
+      "grad_norm": 0.5353108048439026,
       "learning_rate": 2.2872383691764586e-05,
-      "loss": 1.0601,
+      "loss": 1.0598,
       "step": 4490
     },
     {
       "epoch": 0.8031805592781203,
-      "grad_norm": 0.5611812472343445,
+      "grad_norm": 0.5580396056175232,
       "learning_rate": 2.2674226062064996e-05,
-      "loss": 1.0762,
+      "loss": 1.0763,
       "step": 4495
     },
     {
       "epoch": 0.8040739748056821,
-      "grad_norm": 0.5540429353713989,
+      "grad_norm": 0.5546059608459473,
       "learning_rate": 2.2476820734396843e-05,
-      "loss": 1.0666,
+      "loss": 1.067,
       "step": 4500
     },
     {
       "epoch": 0.8049673903332439,
-      "grad_norm": 0.5697487592697144,
+      "grad_norm": 0.5651270151138306,
       "learning_rate": 2.2280169629315484e-05,
-      "loss": 1.1338,
+      "loss": 1.134,
       "step": 4505
     },
     {
       "epoch": 0.8058608058608059,
-      "grad_norm": 0.517309308052063,
+      "grad_norm": 0.5143851637840271,
       "learning_rate": 2.208427466003844e-05,
-      "loss": 1.0976,
+      "loss": 1.0975,
       "step": 4510
     },
     {
       "epoch": 0.8067542213883677,
-      "grad_norm": 0.5207148790359497,
+      "grad_norm": 0.5212395191192627,
       "learning_rate": 2.1889137732426802e-05,
-      "loss": 1.13,
+      "loss": 1.1308,
       "step": 4515
     },
     {
       "epoch": 0.8076476369159296,
-      "grad_norm": 0.5412813425064087,
+      "grad_norm": 0.5373924374580383,
       "learning_rate": 2.1694760744966668e-05,
-      "loss": 1.1341,
+      "loss": 1.134,
       "step": 4520
     },
     {
       "epoch": 0.8085410524434915,
-      "grad_norm": 0.5917104482650757,
+      "grad_norm": 0.5924869179725647,
       "learning_rate": 2.1501145588750694e-05,
-      "loss": 1.0628,
+      "loss": 1.0626,
       "step": 4525
     },
     {
       "epoch": 0.8094344679710533,
-      "grad_norm": 0.5434411764144897,
+      "grad_norm": 0.5431603193283081,
       "learning_rate": 2.1308294147459628e-05,
-      "loss": 1.1121,
+      "loss": 1.1114,
       "step": 4530
     },
     {
       "epoch": 0.8103278834986152,
-      "grad_norm": 0.5469930171966553,
+      "grad_norm": 0.548326849937439,
       "learning_rate": 2.1116208297344155e-05,
-      "loss": 1.1057,
+      "loss": 1.1051,
       "step": 4535
     },
     {
       "epoch": 0.8112212990261771,
-      "grad_norm": 0.5234044194221497,
+      "grad_norm": 0.5223492383956909,
       "learning_rate": 2.0924889907206425e-05,
-      "loss": 1.0711,
+      "loss": 1.071,
       "step": 4540
     },
     {
       "epoch": 0.812114714553739,
-      "grad_norm": 0.5449784398078918,
+      "grad_norm": 0.5413901805877686,
       "learning_rate": 2.0734340838382015e-05,
-      "loss": 1.0914,
+      "loss": 1.0917,
       "step": 4545
     },
     {
       "epoch": 0.8130081300813008,
-      "grad_norm": 0.5194562077522278,
+      "grad_norm": 0.5205683708190918,
       "learning_rate": 2.0544562944721778e-05,
-      "loss": 1.0878,
+      "loss": 1.0875,
       "step": 4550
     },
     {
       "epoch": 0.8139015456088626,
-      "grad_norm": 0.646867573261261,
+      "grad_norm": 0.6497357487678528,
       "learning_rate": 2.0355558072573787e-05,
-      "loss": 1.121,
+      "loss": 1.1216,
       "step": 4555
     },
     {
       "epoch": 0.8147949611364246,
-      "grad_norm": 0.530295729637146,
+      "grad_norm": 0.5288578867912292,
       "learning_rate": 2.016732806076542e-05,
-      "loss": 1.0924,
+      "loss": 1.0919,
       "step": 4560
     },
     {
       "epoch": 0.8156883766639864,
-      "grad_norm": 0.5485954880714417,
+      "grad_norm": 0.5458259582519531,
       "learning_rate": 1.9979874740585426e-05,
-      "loss": 1.1192,
+      "loss": 1.1187,
       "step": 4565
     },
     {
       "epoch": 0.8165817921915483,
-      "grad_norm": 0.549314022064209,
+      "grad_norm": 0.5482889413833618,
       "learning_rate": 1.979319993576614e-05,
       "loss": 1.083,
       "step": 4570
     },
     {
       "epoch": 0.8174752077191102,
-      "grad_norm": 0.5288789868354797,
+      "grad_norm": 0.535325288772583,
       "learning_rate": 1.9607305462465686e-05,
-      "loss": 1.0898,
+      "loss": 1.0897,
       "step": 4575
     },
     {
       "epoch": 0.818368623246672,
-      "grad_norm": 0.5372732281684875,
+      "grad_norm": 0.5331825017929077,
       "learning_rate": 1.942219312925042e-05,
-      "loss": 1.1002,
+      "loss": 1.1001,
       "step": 4580
     },
     {
       "epoch": 0.8192620387742339,
-      "grad_norm": 0.5401791930198669,
+      "grad_norm": 0.541584312915802,
       "learning_rate": 1.9237864737077204e-05,
-      "loss": 1.0702,
+      "loss": 1.07,
       "step": 4585
     },
     {
       "epoch": 0.8201554543017958,
-      "grad_norm": 0.49033990502357483,
+      "grad_norm": 0.49262598156929016,
       "learning_rate": 1.9054322079275953e-05,
-      "loss": 1.1151,
+      "loss": 1.1154,
       "step": 4590
     },
     {
       "epoch": 0.8210488698293577,
-      "grad_norm": 0.5073211193084717,
+      "grad_norm": 0.510992705821991,
       "learning_rate": 1.8871566941532182e-05,
       "loss": 1.0934,
       "step": 4595
     },
     {
       "epoch": 0.8219422853569195,
-      "grad_norm": 0.48106008768081665,
+      "grad_norm": 0.47639334201812744,
       "learning_rate": 1.8689601101869604e-05,
-      "loss": 1.0789,
+      "loss": 1.0779,
       "step": 4600
     },
     {
       "epoch": 0.8228357008844813,
-      "grad_norm": 0.5558092594146729,
+      "grad_norm": 0.5555627346038818,
       "learning_rate": 1.8508426330632933e-05,
-      "loss": 1.0921,
+      "loss": 1.0919,
       "step": 4605
     },
     {
       "epoch": 0.8237291164120433,
-      "grad_norm": 0.5238335728645325,
+      "grad_norm": 0.5235790014266968,
       "learning_rate": 1.8328044390470478e-05,
-      "loss": 1.0684,
+      "loss": 1.0678,
       "step": 4610
     },
     {
       "epoch": 0.8246225319396051,
-      "grad_norm": 0.5261583924293518,
+      "grad_norm": 0.5292229652404785,
       "learning_rate": 1.8148457036317157e-05,
-      "loss": 1.1005,
+      "loss": 1.0999,
       "step": 4615
     },
     {
       "epoch": 0.8255159474671669,
-      "grad_norm": 0.5787078738212585,
+      "grad_norm": 0.5767297744750977,
       "learning_rate": 1.796966601537734e-05,
-      "loss": 1.0957,
+      "loss": 1.0964,
       "step": 4620
     },
     {
       "epoch": 0.8264093629947289,
-      "grad_norm": 0.5445564985275269,
+      "grad_norm": 0.5446600317955017,
       "learning_rate": 1.7791673067107927e-05,
-      "loss": 1.1059,
+      "loss": 1.1066,
       "step": 4625
     },
     {
       "epoch": 0.8273027785222907,
-      "grad_norm": 0.4829619228839874,
+      "grad_norm": 0.48151037096977234,
       "learning_rate": 1.7614479923201333e-05,
       "loss": 1.0794,
       "step": 4630
     },
     {
       "epoch": 0.8281961940498526,
-      "grad_norm": 0.5207480192184448,
+      "grad_norm": 0.5238900780677795,
       "learning_rate": 1.7438088307568667e-05,
-      "loss": 1.0725,
+      "loss": 1.073,
       "step": 4635
     },
     {
       "epoch": 0.8290896095774145,
-      "grad_norm": 0.5238359570503235,
+      "grad_norm": 0.5254760384559631,
       "learning_rate": 1.7262499936322997e-05,
-      "loss": 1.1395,
+      "loss": 1.139,
       "step": 4640
     },
     {
       "epoch": 0.8299830251049763,
-      "grad_norm": 0.5396834015846252,
+      "grad_norm": 0.5375151038169861,
       "learning_rate": 1.708771651776263e-05,
-      "loss": 1.1148,
+      "loss": 1.1143,
       "step": 4645
     },
     {
       "epoch": 0.8308764406325382,
-      "grad_norm": 0.5577458143234253,
+      "grad_norm": 0.5599473118782043,
       "learning_rate": 1.6913739752354464e-05,
-      "loss": 1.104,
+      "loss": 1.1043,
       "step": 4650
     },
     {
       "epoch": 0.8317698561601,
-      "grad_norm": 0.5378464460372925,
+      "grad_norm": 0.5429250597953796,
       "learning_rate": 1.6740571332717558e-05,
-      "loss": 1.0964,
+      "loss": 1.096,
       "step": 4655
     },
     {
       "epoch": 0.832663271687662,
-      "grad_norm": 0.5314394235610962,
+      "grad_norm": 0.5355926752090454,
       "learning_rate": 1.6568212943606465e-05,
-      "loss": 1.1034,
+      "loss": 1.103,
       "step": 4660
     },
     {
       "epoch": 0.8335566872152238,
-      "grad_norm": 0.519146203994751,
+      "grad_norm": 0.5205219388008118,
       "learning_rate": 1.6396666261895034e-05,
-      "loss": 1.076,
+      "loss": 1.0765,
       "step": 4665
     },
     {
       "epoch": 0.8344501027427856,
-      "grad_norm": 0.5421656966209412,
+      "grad_norm": 0.5437434315681458,
       "learning_rate": 1.6225932956559943e-05,
-      "loss": 1.0792,
+      "loss": 1.0793,
       "step": 4670
     },
     {
       "epoch": 0.8353435182703476,
-      "grad_norm": 0.5136523842811584,
+      "grad_norm": 0.5111892223358154,
       "learning_rate": 1.6056014688664656e-05,
-      "loss": 1.1073,
+      "loss": 1.1069,
       "step": 4675
     },
     {
       "epoch": 0.8362369337979094,
-      "grad_norm": 0.5214201807975769,
+      "grad_norm": 0.5251837968826294,
       "learning_rate": 1.5886913111343037e-05,
-      "loss": 1.0946,
+      "loss": 1.0945,
       "step": 4680
     },
     {
       "epoch": 0.8371303493254713,
-      "grad_norm": 0.5285434722900391,
+      "grad_norm": 0.5311691164970398,
       "learning_rate": 1.571862986978342e-05,
       "loss": 1.1,
       "step": 4685
     },
     {
       "epoch": 0.8380237648530331,
-      "grad_norm": 0.5122743844985962,
+      "grad_norm": 0.515058696269989,
       "learning_rate": 1.555116660121253e-05,
-      "loss": 1.1146,
+      "loss": 1.1141,
       "step": 4690
     },
     {
       "epoch": 0.838917180380595,
-      "grad_norm": 0.5290825963020325,
+      "grad_norm": 0.5322168469429016,
       "learning_rate": 1.538452493487956e-05,
-      "loss": 1.1022,
+      "loss": 1.1025,
       "step": 4695
     },
     {
       "epoch": 0.8398105959081569,
-      "grad_norm": 0.5569957494735718,
+      "grad_norm": 0.5564744472503662,
       "learning_rate": 1.5218706492040435e-05,
-      "loss": 1.068,
+      "loss": 1.0679,
       "step": 4700
     },
     {
       "epoch": 0.8407040114357187,
-      "grad_norm": 0.5273507833480835,
+      "grad_norm": 0.52479088306427,
       "learning_rate": 1.5053712885941862e-05,
-      "loss": 1.0983,
+      "loss": 1.0985,
       "step": 4705
     },
     {
       "epoch": 0.8415974269632807,
-      "grad_norm": 0.5313451886177063,
+      "grad_norm": 0.5248088240623474,
       "learning_rate": 1.4889545721805687e-05,
-      "loss": 1.051,
+      "loss": 1.0503,
       "step": 4710
     },
     {
       "epoch": 0.8424908424908425,
-      "grad_norm": 0.5266393423080444,
+      "grad_norm": 0.525884747505188,
       "learning_rate": 1.4726206596813363e-05,
-      "loss": 1.1074,
+      "loss": 1.1073,
       "step": 4715
     },
     {
       "epoch": 0.8433842580184043,
-      "grad_norm": 0.5717965364456177,
+      "grad_norm": 0.5724270343780518,
       "learning_rate": 1.456369710009038e-05,
-      "loss": 1.1266,
+      "loss": 1.126,
       "step": 4720
     },
     {
       "epoch": 0.8442776735459663,
-      "grad_norm": 0.4887787699699402,
+      "grad_norm": 0.4863172173500061,
       "learning_rate": 1.4402018812690721e-05,
-      "loss": 1.0694,
+      "loss": 1.0695,
       "step": 4725
     },
     {
       "epoch": 0.8451710890735281,
-      "grad_norm": 0.5646076202392578,
+      "grad_norm": 0.5609614849090576,
       "learning_rate": 1.4241173307581558e-05,
-      "loss": 1.0688,
+      "loss": 1.0684,
       "step": 4730
     },
     {
       "epoch": 0.84606450460109,
-      "grad_norm": 0.5532607436180115,
+      "grad_norm": 0.5545142292976379,
       "learning_rate": 1.4081162149627936e-05,
-      "loss": 1.0364,
+      "loss": 1.0362,
       "step": 4735
     },
     {
       "epoch": 0.8469579201286518,
-      "grad_norm": 0.5489101409912109,
+      "grad_norm": 0.5475736856460571,
       "learning_rate": 1.39219868955775e-05,
-      "loss": 1.0935,
+      "loss": 1.0926,
       "step": 4740
     },
     {
       "epoch": 0.8478513356562137,
-      "grad_norm": 0.48812875151634216,
+      "grad_norm": 0.4896586537361145,
       "learning_rate": 1.3763649094045483e-05,
-      "loss": 1.0678,
+      "loss": 1.0675,
       "step": 4745
     },
     {
       "epoch": 0.8487447511837756,
-      "grad_norm": 0.4934766888618469,
+      "grad_norm": 0.49385780096054077,
       "learning_rate": 1.3606150285499475e-05,
-      "loss": 1.0741,
+      "loss": 1.0739,
       "step": 4750
     },
     {
       "epoch": 0.8496381667113374,
-      "grad_norm": 0.5471161603927612,
+      "grad_norm": 0.5389235019683838,
       "learning_rate": 1.3449492002244502e-05,
-      "loss": 1.0799,
+      "loss": 1.0808,
       "step": 4755
     },
     {
       "epoch": 0.8505315822388994,
-      "grad_norm": 0.520369827747345,
+      "grad_norm": 0.5244671702384949,
       "learning_rate": 1.329367576840812e-05,
-      "loss": 1.0963,
+      "loss": 1.0967,
       "step": 4760
     },
     {
       "epoch": 0.8514249977664612,
-      "grad_norm": 0.5638504028320312,
+      "grad_norm": 0.5618093013763428,
       "learning_rate": 1.3138703099925676e-05,
-      "loss": 1.0745,
+      "loss": 1.0743,
       "step": 4765
     },
     {
       "epoch": 0.852318413294023,
-      "grad_norm": 0.5431874990463257,
+      "grad_norm": 0.5471380949020386,
       "learning_rate": 1.2984575504525376e-05,
       "loss": 1.1035,
       "step": 4770
     },
     {
       "epoch": 0.853211828821585,
-      "grad_norm": 0.49290958046913147,
+      "grad_norm": 0.5013295412063599,
       "learning_rate": 1.2831294481713763e-05,
-      "loss": 1.0742,
+      "loss": 1.0736,
       "step": 4775
     },
     {
       "epoch": 0.8541052443491468,
-      "grad_norm": 0.5271280407905579,
+      "grad_norm": 0.5280576348304749,
       "learning_rate": 1.2678861522761066e-05,
-      "loss": 1.0823,
+      "loss": 1.0825,
       "step": 4780
     },
     {
       "epoch": 0.8549986598767086,
-      "grad_norm": 0.5476307272911072,
+      "grad_norm": 0.5448872447013855,
       "learning_rate": 1.2527278110686712e-05,
-      "loss": 1.0886,
+      "loss": 1.0895,
       "step": 4785
     },
     {
       "epoch": 0.8558920754042705,
-      "grad_norm": 0.5493554472923279,
+      "grad_norm": 0.5476821660995483,
       "learning_rate": 1.237654572024487e-05,
-      "loss": 1.109,
+      "loss": 1.1082,
       "step": 4790
     },
     {
       "epoch": 0.8567854909318324,
-      "grad_norm": 0.5390406847000122,
+      "grad_norm": 0.5306336283683777,
       "learning_rate": 1.2226665817910166e-05,
-      "loss": 1.0723,
+      "loss": 1.0715,
       "step": 4795
     },
     {
       "epoch": 0.8576789064593943,
-      "grad_norm": 0.5032559037208557,
+      "grad_norm": 0.5052401423454285,
       "learning_rate": 1.2077639861863365e-05,
-      "loss": 1.0737,
+      "loss": 1.0731,
       "step": 4800
     },
     {
       "epoch": 0.8585723219869561,
-      "grad_norm": 0.5364784598350525,
+      "grad_norm": 0.5403242707252502,
       "learning_rate": 1.1929469301977136e-05,
-      "loss": 1.0947,
+      "loss": 1.0941,
       "step": 4805
     },
     {
       "epoch": 0.859465737514518,
-      "grad_norm": 0.519612729549408,
+      "grad_norm": 0.5217272639274597,
       "learning_rate": 1.1782155579802034e-05,
-      "loss": 1.0627,
+      "loss": 1.0628,
       "step": 4810
     },
     {
       "epoch": 0.8603591530420799,
-      "grad_norm": 0.5701001286506653,
+      "grad_norm": 0.570672869682312,
       "learning_rate": 1.1635700128552508e-05,
-      "loss": 1.1162,
+      "loss": 1.1157,
       "step": 4815
     },
     {
       "epoch": 0.8612525685696417,
-      "grad_norm": 0.5443189740180969,
+      "grad_norm": 0.5487679839134216,
       "learning_rate": 1.1490104373092825e-05,
-      "loss": 1.0975,
+      "loss": 1.0981,
       "step": 4820
     },
     {
       "epoch": 0.8621459840972036,
-      "grad_norm": 0.5358694195747375,
+      "grad_norm": 0.5410107970237732,
       "learning_rate": 1.1345369729923271e-05,
-      "loss": 1.0922,
+      "loss": 1.0917,
       "step": 4825
     },
     {
       "epoch": 0.8630393996247655,
-      "grad_norm": 0.5254837274551392,
+      "grad_norm": 0.5344003438949585,
       "learning_rate": 1.1201497607166423e-05,
-      "loss": 1.106,
+      "loss": 1.1058,
       "step": 4830
     },
     {
       "epoch": 0.8639328151523273,
-      "grad_norm": 0.5468859672546387,
+      "grad_norm": 0.548433780670166,
       "learning_rate": 1.105848940455334e-05,
-      "loss": 1.0725,
+      "loss": 1.0724,
       "step": 4835
     },
     {
       "epoch": 0.8648262306798892,
-      "grad_norm": 0.5670140385627747,
+      "grad_norm": 0.573233962059021,
       "learning_rate": 1.0916346513410081e-05,
-      "loss": 1.0736,
+      "loss": 1.0733,
       "step": 4840
     },
     {
       "epoch": 0.8657196462074511,
-      "grad_norm": 0.5354899168014526,
+      "grad_norm": 0.5367952585220337,
       "learning_rate": 1.0775070316644042e-05,
-      "loss": 1.0831,
+      "loss": 1.0829,
       "step": 4845
     },
     {
       "epoch": 0.866613061735013,
-      "grad_norm": 0.5412014126777649,
+      "grad_norm": 0.5424239635467529,
       "learning_rate": 1.0634662188730604e-05,
-      "loss": 1.1105,
+      "loss": 1.11,
       "step": 4850
     },
     {
       "epoch": 0.8675064772625748,
-      "grad_norm": 0.5652616620063782,
+      "grad_norm": 0.5632114410400391,
       "learning_rate": 1.0495123495699588e-05,
-      "loss": 1.092,
+      "loss": 1.0926,
       "step": 4855
     },
     {
       "epoch": 0.8683998927901367,
-      "grad_norm": 0.5190427303314209,
+      "grad_norm": 0.517599880695343,
       "learning_rate": 1.0356455595122239e-05,
-      "loss": 1.1216,
+      "loss": 1.1214,
       "step": 4860
     },
     {
       "epoch": 0.8692933083176986,
-      "grad_norm": 0.5157451629638672,
+      "grad_norm": 0.5182797908782959,
       "learning_rate": 1.02186598360978e-05,
-      "loss": 1.0765,
+      "loss": 1.0759,
       "step": 4865
     },
     {
       "epoch": 0.8701867238452604,
-      "grad_norm": 0.532561182975769,
+      "grad_norm": 0.5280758142471313,
       "learning_rate": 1.0081737559240445e-05,
-      "loss": 1.1269,
+      "loss": 1.1264,
       "step": 4870
     },
     {
       "epoch": 0.8710801393728222,
-      "grad_norm": 0.5337056517601013,
+      "grad_norm": 0.534575343132019,
       "learning_rate": 9.945690096666249e-06,
-      "loss": 1.0813,
+      "loss": 1.0818,
       "step": 4875
     },
     {
       "epoch": 0.8719735549003842,
-      "grad_norm": 0.5260433554649353,
+      "grad_norm": 0.5310872793197632,
       "learning_rate": 9.810518771980225e-06,
-      "loss": 1.1078,
+      "loss": 1.1092,
       "step": 4880
     },
     {
       "epoch": 0.872866970427946,
-      "grad_norm": 0.5416473746299744,
+      "grad_norm": 0.5414004921913147,
       "learning_rate": 9.676224900263497e-06,
       "loss": 1.1235,
       "step": 4885
     },
     {
       "epoch": 0.8737603859555079,
-      "grad_norm": 0.5500181913375854,
+      "grad_norm": 0.5505749583244324,
       "learning_rate": 9.542809788060358e-06,
-      "loss": 1.1465,
+      "loss": 1.1467,
       "step": 4890
     },
     {
       "epoch": 0.8746538014830698,
-      "grad_norm": 0.5110030770301819,
+      "grad_norm": 0.507029116153717,
       "learning_rate": 9.410274733365753e-06,
-      "loss": 1.0819,
+      "loss": 1.0818,
       "step": 4895
     },
     {
       "epoch": 0.8755472170106317,
-      "grad_norm": 0.5659358501434326,
+      "grad_norm": 0.5631771683692932,
       "learning_rate": 9.278621025612434e-06,
-      "loss": 1.0976,
+      "loss": 1.0975,
       "step": 4900
     },
     {
       "epoch": 0.8764406325381935,
-      "grad_norm": 0.5617207884788513,
+      "grad_norm": 0.6328200697898865,
       "learning_rate": 9.147849945658648e-06,
-      "loss": 1.0852,
+      "loss": 1.0853,
       "step": 4905
     },
     {
       "epoch": 0.8773340480657554,
-      "grad_norm": 0.5818383693695068,
+      "grad_norm": 0.5625194907188416,
       "learning_rate": 9.017962765775523e-06,
-      "loss": 1.1086,
+      "loss": 1.1083,
       "step": 4910
     },
     {
       "epoch": 0.8782274635933173,
-      "grad_norm": 0.5794780850410461,
+      "grad_norm": 0.5809071063995361,
       "learning_rate": 8.888960749634712e-06,
-      "loss": 1.0831,
+      "loss": 1.0833,
       "step": 4915
     },
     {
       "epoch": 0.8791208791208791,
-      "grad_norm": 0.5406253933906555,
+      "grad_norm": 0.5429468154907227,
       "learning_rate": 8.760845152296116e-06,
-      "loss": 1.0816,
+      "loss": 1.0814,
       "step": 4920
     },
     {
       "epoch": 0.8800142946484409,
-      "grad_norm": 0.5111543536186218,
+      "grad_norm": 0.5104892253875732,
       "learning_rate": 8.63361722019569e-06,
-      "loss": 1.0647,
+      "loss": 1.0644,
       "step": 4925
     },
     {
       "epoch": 0.8809077101760029,
-      "grad_norm": 0.5049743056297302,
+      "grad_norm": 0.5056647658348083,
       "learning_rate": 8.507278191133261e-06,
-      "loss": 1.0666,
+      "loss": 1.0679,
       "step": 4930
     },
     {
       "epoch": 0.8818011257035647,
-      "grad_norm": 0.5689327716827393,
+      "grad_norm": 0.5692788362503052,
       "learning_rate": 8.38182929426059e-06,
-      "loss": 1.1112,
+      "loss": 1.112,
       "step": 4935
     },
     {
       "epoch": 0.8826945412311266,
-      "grad_norm": 0.5630576014518738,
+      "grad_norm": 0.560130774974823,
       "learning_rate": 8.257271750069295e-06,
-      "loss": 1.1307,
+      "loss": 1.1309,
       "step": 4940
     },
     {
       "epoch": 0.8835879567586885,
-      "grad_norm": 0.518915593624115,
+      "grad_norm": 0.5144816040992737,
       "learning_rate": 8.133606770379055e-06,
-      "loss": 1.0815,
+      "loss": 1.0819,
       "step": 4945
     },
     {
       "epoch": 0.8844813722862503,
-      "grad_norm": 0.4844837188720703,
+      "grad_norm": 0.484380304813385,
       "learning_rate": 8.010835558325735e-06,
-      "loss": 1.0847,
+      "loss": 1.0844,
       "step": 4950
     },
     {
       "epoch": 0.8853747878138122,
-      "grad_norm": 0.5525776147842407,
+      "grad_norm": 0.5501832365989685,
       "learning_rate": 7.88895930834983e-06,
-      "loss": 1.101,
+      "loss": 1.1013,
       "step": 4955
     },
     {
       "epoch": 0.8862682033413741,
-      "grad_norm": 0.49364981055259705,
+      "grad_norm": 0.49795544147491455,
       "learning_rate": 7.767979206184694e-06,
-      "loss": 1.0628,
+      "loss": 1.0627,
       "step": 4960
     },
     {
       "epoch": 0.887161618868936,
-      "grad_norm": 0.5372298359870911,
+      "grad_norm": 0.5382062792778015,
       "learning_rate": 7.64789642884508e-06,
-      "loss": 1.0762,
+      "loss": 1.0758,
       "step": 4965
     },
     {
       "epoch": 0.8880550343964978,
-      "grad_norm": 0.5336244106292725,
+      "grad_norm": 0.5314489603042603,
       "learning_rate": 7.528712144615679e-06,
-      "loss": 1.051,
+      "loss": 1.0501,
       "step": 4970
     },
     {
       "epoch": 0.8889484499240596,
-      "grad_norm": 0.510352611541748,
+      "grad_norm": 0.5101165175437927,
       "learning_rate": 7.4104275130397085e-06,
-      "loss": 1.0813,
+      "loss": 1.0807,
       "step": 4975
     },
     {
       "epoch": 0.8898418654516216,
-      "grad_norm": 0.5422449707984924,
+      "grad_norm": 0.5354017019271851,
       "learning_rate": 7.29304368490773e-06,
-      "loss": 1.0806,
+      "loss": 1.0803,
       "step": 4980
     },
     {
       "epoch": 0.8907352809791834,
-      "grad_norm": 0.5817996263504028,
+      "grad_norm": 0.5763533711433411,
       "learning_rate": 7.176561802246373e-06,
-      "loss": 1.1025,
+      "loss": 1.1026,
       "step": 4985
     },
     {
       "epoch": 0.8916286965067453,
-      "grad_norm": 0.5407490730285645,
+      "grad_norm": 0.537377655506134,
       "learning_rate": 7.06098299830722e-06,
-      "loss": 1.0885,
+      "loss": 1.0888,
       "step": 4990
     },
     {
       "epoch": 0.8925221120343072,
-      "grad_norm": 0.5575851202011108,
+      "grad_norm": 0.5555933713912964,
       "learning_rate": 6.946308397555823e-06,
-      "loss": 1.135,
+      "loss": 1.1351,
       "step": 4995
     },
     {
       "epoch": 0.893415527561869,
-      "grad_norm": 0.5409137010574341,
+      "grad_norm": 0.5403766632080078,
       "learning_rate": 6.832539115660752e-06,
-      "loss": 1.0889,
+      "loss": 1.0891,
       "step": 5000
     },
     {
       "epoch": 0.8943089430894309,
-      "grad_norm": 0.515312910079956,
+      "grad_norm": 0.517660915851593,
       "learning_rate": 6.719676259482721e-06,
-      "loss": 1.0601,
+      "loss": 1.0607,
       "step": 5005
     },
     {
       "epoch": 0.8952023586169927,
-      "grad_norm": 0.5435192584991455,
+      "grad_norm": 0.5434561371803284,
       "learning_rate": 6.607720927063843e-06,
-      "loss": 1.0978,
+      "loss": 1.0982,
       "step": 5010
     },
     {
       "epoch": 0.8960957741445547,
-      "grad_norm": 0.5604092478752136,
+      "grad_norm": 0.563258707523346,
       "learning_rate": 6.496674207616926e-06,
-      "loss": 1.0699,
+      "loss": 1.0701,
       "step": 5015
     },
     {
       "epoch": 0.8969891896721165,
-      "grad_norm": 0.5125716924667358,
+      "grad_norm": 0.511574387550354,
       "learning_rate": 6.386537181514896e-06,
-      "loss": 1.0807,
+      "loss": 1.0808,
       "step": 5020
     },
     {
       "epoch": 0.8978826051996783,
-      "grad_norm": 0.5005190372467041,
+      "grad_norm": 0.49810346961021423,
       "learning_rate": 6.277310920280299e-06,
-      "loss": 1.066,
+      "loss": 1.0664,
       "step": 5025
     },
     {
       "epoch": 0.8987760207272403,
-      "grad_norm": 0.5269173979759216,
+      "grad_norm": 0.526667058467865,
       "learning_rate": 6.1689964865748185e-06,
-      "loss": 1.0719,
+      "loss": 1.0714,
       "step": 5030
     },
     {
       "epoch": 0.8996694362548021,
-      "grad_norm": 0.5614027380943298,
+      "grad_norm": 0.5600285530090332,
       "learning_rate": 6.061594934188985e-06,
-      "loss": 1.0712,
+      "loss": 1.0706,
       "step": 5035
     },
     {
       "epoch": 0.900562851782364,
-      "grad_norm": 0.5723266005516052,
+      "grad_norm": 0.5716047883033752,
       "learning_rate": 5.955107308031915e-06,
-      "loss": 1.0802,
+      "loss": 1.0806,
       "step": 5040
     },
     {
       "epoch": 0.9014562673099259,
-      "grad_norm": 0.5639398694038391,
+      "grad_norm": 0.5603742599487305,
       "learning_rate": 5.849534644121146e-06,
-      "loss": 1.0757,
+      "loss": 1.0755,
       "step": 5045
     },
     {
       "epoch": 0.9023496828374877,
-      "grad_norm": 0.522843599319458,
+      "grad_norm": 0.5195134878158569,
       "learning_rate": 5.744877969572537e-06,
-      "loss": 1.0538,
+      "loss": 1.0537,
       "step": 5050
     },
     {
       "epoch": 0.9032430983650496,
-      "grad_norm": 0.5338366031646729,
+      "grad_norm": 0.5341048240661621,
       "learning_rate": 5.6411383025903205e-06,
-      "loss": 1.0957,
+      "loss": 1.0951,
       "step": 5055
     },
     {
       "epoch": 0.9041365138926114,
-      "grad_norm": 0.55523681640625,
+      "grad_norm": 0.55425626039505,
       "learning_rate": 5.538316652457121e-06,
-      "loss": 1.0752,
+      "loss": 1.0755,
       "step": 5060
     },
     {
       "epoch": 0.9050299294201734,
-      "grad_norm": 0.5244052410125732,
+      "grad_norm": 0.520702600479126,
       "learning_rate": 5.436414019524216e-06,
-      "loss": 1.1156,
+      "loss": 1.1157,
       "step": 5065
     },
     {
       "epoch": 0.9059233449477352,
-      "grad_norm": 0.5795427560806274,
+      "grad_norm": 0.5679965019226074,
       "learning_rate": 5.335431395201784e-06,
-      "loss": 1.0806,
+      "loss": 1.0801,
       "step": 5070
     },
     {
       "epoch": 0.906816760475297,
-      "grad_norm": 0.5322124361991882,
+      "grad_norm": 0.5292661190032959,
       "learning_rate": 5.235369761949216e-06,
-      "loss": 1.0903,
+      "loss": 1.0908,
       "step": 5075
     },
     {
       "epoch": 0.907710176002859,
-      "grad_norm": 0.5396597981452942,
+      "grad_norm": 0.5407772660255432,
       "learning_rate": 5.136230093265593e-06,
       "loss": 1.0639,
       "step": 5080
     },
     {
       "epoch": 0.9086035915304208,
-      "grad_norm": 0.5385281443595886,
+      "grad_norm": 0.5404497981071472,
       "learning_rate": 5.038013353680204e-06,
-      "loss": 1.1,
+      "loss": 1.1004,
       "step": 5085
     },
     {
       "epoch": 0.9094970070579826,
-      "grad_norm": 0.5123850107192993,
+      "grad_norm": 0.5132922530174255,
       "learning_rate": 4.940720498743179e-06,
-      "loss": 1.1072,
+      "loss": 1.107,
       "step": 5090
     },
     {
       "epoch": 0.9103904225855446,
-      "grad_norm": 0.4937289357185364,
+      "grad_norm": 0.49404093623161316,
       "learning_rate": 4.8443524750161676e-06,
-      "loss": 1.1047,
+      "loss": 1.1043,
       "step": 5095
     },
     {
       "epoch": 0.9112838381131064,
-      "grad_norm": 0.5448057055473328,
+      "grad_norm": 0.5417544841766357,
       "learning_rate": 4.74891022006313e-06,
-      "loss": 1.091,
+      "loss": 1.0917,
       "step": 5100
     },
     {
       "epoch": 0.9121772536406683,
-      "grad_norm": 0.5475412011146545,
+      "grad_norm": 0.5480242371559143,
       "learning_rate": 4.654394662441264e-06,
-      "loss": 1.0841,
+      "loss": 1.0834,
       "step": 5105
     },
     {
       "epoch": 0.9130706691682301,
-      "grad_norm": 0.5547966361045837,
+      "grad_norm": 0.555147647857666,
       "learning_rate": 4.560806721691913e-06,
-      "loss": 1.089,
+      "loss": 1.0887,
       "step": 5110
     },
     {
       "epoch": 0.913964084695792,
-      "grad_norm": 0.5370575785636902,
+      "grad_norm": 0.5397526621818542,
       "learning_rate": 4.468147308331605e-06,
-      "loss": 1.1023,
+      "loss": 1.1017,
       "step": 5115
     },
     {
       "epoch": 0.9148575002233539,
-      "grad_norm": 0.5097227096557617,
+      "grad_norm": 0.510307252407074,
       "learning_rate": 4.376417323843318e-06,
-      "loss": 1.0737,
+      "loss": 1.0739,
       "step": 5120
     },
     {
       "epoch": 0.9157509157509157,
-      "grad_norm": 0.5403570532798767,
+      "grad_norm": 0.5409055948257446,
       "learning_rate": 4.28561766066754e-06,
-      "loss": 1.1012,
+      "loss": 1.101,
       "step": 5125
     },
     {
       "epoch": 0.9166443312784777,
-      "grad_norm": 0.5509471893310547,
+      "grad_norm": 0.552873969078064,
       "learning_rate": 4.195749202193699e-06,
-      "loss": 1.091,
+      "loss": 1.0913,
       "step": 5130
     },
     {
       "epoch": 0.9175377468060395,
-      "grad_norm": 0.5664262175559998,
+      "grad_norm": 0.5643073916435242,
       "learning_rate": 4.106812822751538e-06,
-      "loss": 1.0778,
+      "loss": 1.0779,
       "step": 5135
     },
     {
       "epoch": 0.9184311623336013,
-      "grad_norm": 0.5238862633705139,
+      "grad_norm": 0.523567259311676,
       "learning_rate": 4.018809387602596e-06,
-      "loss": 1.087,
+      "loss": 1.0872,
       "step": 5140
     },
     {
       "epoch": 0.9193245778611632,
-      "grad_norm": 0.5326477289199829,
+      "grad_norm": 0.5306380391120911,
       "learning_rate": 3.931739752931829e-06,
-      "loss": 1.0912,
+      "loss": 1.0917,
       "step": 5145
     },
     {
       "epoch": 0.9202179933887251,
-      "grad_norm": 0.5700786709785461,
+      "grad_norm": 0.5733050107955933,
       "learning_rate": 3.845604765839228e-06,
-      "loss": 1.1111,
+      "loss": 1.111,
       "step": 5150
     },
     {
       "epoch": 0.921111408916287,
-      "grad_norm": 0.5622351765632629,
+      "grad_norm": 0.5619934797286987,
       "learning_rate": 3.760405264331612e-06,
-      "loss": 1.0649,
+      "loss": 1.065,
       "step": 5155
     },
     {
       "epoch": 0.9220048244438488,
-      "grad_norm": 0.5550172328948975,
+      "grad_norm": 0.5529298186302185,
       "learning_rate": 3.676142077314448e-06,
-      "loss": 1.0898,
+      "loss": 1.0899,
       "step": 5160
     },
     {
       "epoch": 0.9228982399714107,
-      "grad_norm": 0.5153648853302002,
+      "grad_norm": 0.5153440237045288,
       "learning_rate": 3.592816024583856e-06,
-      "loss": 1.0997,
+      "loss": 1.0995,
       "step": 5165
     },
     {
       "epoch": 0.9237916554989726,
-      "grad_norm": 0.5297582745552063,
+      "grad_norm": 0.5353826880455017,
       "learning_rate": 3.510427916818526e-06,
-      "loss": 1.0909,
+      "loss": 1.0906,
       "step": 5170
     },
     {
       "epoch": 0.9246850710265344,
-      "grad_norm": 0.550730288028717,
+      "grad_norm": 0.5490701794624329,
       "learning_rate": 3.4289785555719157e-06,
-      "loss": 1.1237,
+      "loss": 1.1228,
       "step": 5175
     },
     {
       "epoch": 0.9255784865540964,
-      "grad_norm": 0.5354406237602234,
+      "grad_norm": 0.5379487872123718,
       "learning_rate": 3.348468733264398e-06,
-      "loss": 1.058,
+      "loss": 1.0582,
       "step": 5180
     },
     {
       "epoch": 0.9264719020816582,
-      "grad_norm": 0.5400915145874023,
+      "grad_norm": 0.5375136733055115,
       "learning_rate": 3.268899233175604e-06,
-      "loss": 1.0929,
+      "loss": 1.0934,
       "step": 5185
     },
     {
       "epoch": 0.92736531760922,
-      "grad_norm": 0.5201057195663452,
+      "grad_norm": 0.51712965965271,
       "learning_rate": 3.19027082943677e-06,
       "loss": 1.0704,
       "step": 5190
     },
     {
       "epoch": 0.9282587331367819,
-      "grad_norm": 0.5121946930885315,
+      "grad_norm": 0.512758195400238,
       "learning_rate": 3.1125842870232014e-06,
-      "loss": 1.1248,
+      "loss": 1.124,
       "step": 5195
     },
     {
       "epoch": 0.9291521486643438,
-      "grad_norm": 0.5835002064704895,
+      "grad_norm": 0.58779376745224,
       "learning_rate": 3.0358403617468446e-06,
-      "loss": 1.14,
+      "loss": 1.1403,
       "step": 5200
     },
     {
       "epoch": 0.9300455641919056,
-      "grad_norm": 0.5196322798728943,
+      "grad_norm": 0.5201031565666199,
       "learning_rate": 2.960039800248915e-06,
       "loss": 1.0889,
       "step": 5205
     },
     {
       "epoch": 0.9309389797194675,
-      "grad_norm": 0.4990968108177185,
+      "grad_norm": 0.49719128012657166,
       "learning_rate": 2.885183339992692e-06,
-      "loss": 1.0883,
+      "loss": 1.0881,
       "step": 5210
     },
     {
       "epoch": 0.9318323952470294,
-      "grad_norm": 0.5504352450370789,
+      "grad_norm": 0.5498881340026855,
       "learning_rate": 2.8112717092562358e-06,
-      "loss": 1.0471,
+      "loss": 1.0472,
       "step": 5215
     },
     {
       "epoch": 0.9327258107745913,
-      "grad_norm": 0.5271458625793457,
+      "grad_norm": 0.5250979661941528,
       "learning_rate": 2.738305627125415e-06,
-      "loss": 1.0967,
+      "loss": 1.0962,
       "step": 5220
     },
     {
       "epoch": 0.9336192263021531,
-      "grad_norm": 0.5700637698173523,
+      "grad_norm": 0.5712759494781494,
       "learning_rate": 2.6662858034868454e-06,
-      "loss": 1.1138,
+      "loss": 1.1149,
       "step": 5225
     },
     {
       "epoch": 0.934512641829715,
-      "grad_norm": 0.502876341342926,
+      "grad_norm": 0.5062239170074463,
       "learning_rate": 2.5952129390209854e-06,
-      "loss": 1.0897,
+      "loss": 1.0905,
       "step": 5230
     },
     {
       "epoch": 0.9354060573572769,
-      "grad_norm": 0.5170863270759583,
+      "grad_norm": 0.5184369683265686,
       "learning_rate": 2.525087725195352e-06,
-      "loss": 1.0711,
+      "loss": 1.0709,
       "step": 5235
     },
     {
       "epoch": 0.9362994728848387,
-      "grad_norm": 0.5183836817741394,
+      "grad_norm": 0.5225459933280945,
       "learning_rate": 2.4559108442577585e-06,
-      "loss": 1.0735,
+      "loss": 1.0738,
       "step": 5240
     },
     {
       "epoch": 0.9371928884124006,
-      "grad_norm": 0.5455399751663208,
+      "grad_norm": 0.5459494590759277,
       "learning_rate": 2.38768296922971e-06,
-      "loss": 1.0648,
+      "loss": 1.064,
       "step": 5245
     },
     {
       "epoch": 0.9380863039399625,
-      "grad_norm": 0.47233831882476807,
+      "grad_norm": 0.47445785999298096,
       "learning_rate": 2.3204047638998195e-06,
-      "loss": 1.0621,
+      "loss": 1.0625,
       "step": 5250
     },
     {
       "epoch": 0.9389797194675243,
-      "grad_norm": 0.5074595808982849,
+      "grad_norm": 0.5076881051063538,
       "learning_rate": 2.2540768828173795e-06,
       "loss": 1.0736,
       "step": 5255
     },
     {
       "epoch": 0.9398731349950862,
-      "grad_norm": 0.5757531523704529,
+      "grad_norm": 0.5733228921890259,
       "learning_rate": 2.1886999712860014e-06,
-      "loss": 1.1304,
+      "loss": 1.1301,
       "step": 5260
     },
     {
       "epoch": 0.9407665505226481,
-      "grad_norm": 0.5505633354187012,
+      "grad_norm": 0.5537548065185547,
       "learning_rate": 2.1242746653572845e-06,
-      "loss": 1.1001,
+      "loss": 1.1,
       "step": 5265
     },
     {
       "epoch": 0.94165996605021,
-      "grad_norm": 0.5664975047111511,
+      "grad_norm": 0.5685492753982544,
       "learning_rate": 2.060801591824668e-06,
-      "loss": 1.1198,
+      "loss": 1.1201,
       "step": 5270
     },
     {
       "epoch": 0.9425533815777718,
-      "grad_norm": 0.5190711617469788,
+      "grad_norm": 0.5240213871002197,
       "learning_rate": 1.9982813682173586e-06,
-      "loss": 1.0804,
+      "loss": 1.0814,
       "step": 5275
     },
     {
       "epoch": 0.9434467971053337,
-      "grad_norm": 0.5266507863998413,
+      "grad_norm": 0.5279719829559326,
       "learning_rate": 1.936714602794254e-06,
-      "loss": 1.0819,
+      "loss": 1.0812,
       "step": 5280
     },
     {
       "epoch": 0.9443402126328956,
-      "grad_norm": 0.5362054705619812,
+      "grad_norm": 0.5382205247879028,
       "learning_rate": 1.8761018945380849e-06,
-      "loss": 1.0607,
+      "loss": 1.06,
       "step": 5285
     },
     {
       "epoch": 0.9452336281604574,
-      "grad_norm": 0.50426185131073,
+      "grad_norm": 0.503291666507721,
       "learning_rate": 1.8164438331495614e-06,
-      "loss": 1.1005,
+      "loss": 1.1014,
       "step": 5290
     },
     {
       "epoch": 0.9461270436880193,
-      "grad_norm": 0.5616719722747803,
+      "grad_norm": 0.5597132444381714,
       "learning_rate": 1.7577409990416237e-06,
-      "loss": 1.0772,
+      "loss": 1.0768,
       "step": 5295
     },
     {
       "epoch": 0.9470204592155812,
-      "grad_norm": 0.5646582841873169,
+      "grad_norm": 0.5632822513580322,
       "learning_rate": 1.6999939633338236e-06,
-      "loss": 1.1123,
+      "loss": 1.1121,
       "step": 5300
     },
     {
       "epoch": 0.947913874743143,
-      "grad_norm": 0.5808379650115967,
+      "grad_norm": 0.5778818130493164,
       "learning_rate": 1.6432032878467729e-06,
       "loss": 1.0763,
       "step": 5305
     },
     {
       "epoch": 0.9488072902707049,
-      "grad_norm": 0.51576167345047,
+      "grad_norm": 0.5174503922462463,
       "learning_rate": 1.587369525096627e-06,
-      "loss": 1.0801,
+      "loss": 1.0799,
       "step": 5310
     },
     {
       "epoch": 0.9497007057982668,
-      "grad_norm": 0.5430741906166077,
+      "grad_norm": 0.5436321496963501,
       "learning_rate": 1.5324932182897656e-06,
-      "loss": 1.0978,
+      "loss": 1.0975,
       "step": 5315
     },
     {
       "epoch": 0.9505941213258287,
-      "grad_norm": 0.5385932922363281,
+      "grad_norm": 0.5373766422271729,
       "learning_rate": 1.4785749013174754e-06,
-      "loss": 1.0807,
+      "loss": 1.0808,
       "step": 5320
     },
     {
       "epoch": 0.9514875368533905,
-      "grad_norm": 0.5400928854942322,
+      "grad_norm": 0.5411360859870911,
       "learning_rate": 1.4256150987507544e-06,
-      "loss": 1.1189,
+      "loss": 1.1184,
       "step": 5325
     },
     {
       "epoch": 0.9523809523809523,
-      "grad_norm": 0.5451071262359619,
+      "grad_norm": 0.5489704608917236,
       "learning_rate": 1.3736143258352707e-06,
-      "loss": 1.073,
+      "loss": 1.0733,
       "step": 5330
     },
     {
       "epoch": 0.9532743679085143,
-      "grad_norm": 0.5353686213493347,
+      "grad_norm": 0.5345862507820129,
       "learning_rate": 1.322573088486212e-06,
-      "loss": 1.1045,
+      "loss": 1.1047,
       "step": 5335
     },
     {
       "epoch": 0.9541677834360761,
-      "grad_norm": 0.50026935338974,
+      "grad_norm": 0.4998050630092621,
       "learning_rate": 1.272491883283533e-06,
       "loss": 1.1046,
       "step": 5340
     },
     {
       "epoch": 0.955061198963638,
-      "grad_norm": 0.5251268148422241,
+      "grad_norm": 0.5264286994934082,
       "learning_rate": 1.2233711974669714e-06,
       "loss": 1.0887,
       "step": 5345
     },
     {
       "epoch": 0.9559546144911999,
-      "grad_norm": 0.5508390069007874,
+      "grad_norm": 0.5480632185935974,
       "learning_rate": 1.1752115089314398e-06,
-      "loss": 1.1005,
+      "loss": 1.1008,
       "step": 5350
     },
     {
       "epoch": 0.9568480300187617,
-      "grad_norm": 0.5888670086860657,
+      "grad_norm": 0.5870484113693237,
       "learning_rate": 1.1280132862222737e-06,
-      "loss": 1.095,
+      "loss": 1.0944,
       "step": 5355
     },
     {
       "epoch": 0.9577414455463236,
-      "grad_norm": 0.5448806285858154,
+      "grad_norm": 0.5417635440826416,
       "learning_rate": 1.081776988530725e-06,
-      "loss": 1.094,
+      "loss": 1.0934,
       "step": 5360
     },
     {
       "epoch": 0.9586348610738855,
-      "grad_norm": 0.5379595160484314,
+      "grad_norm": 0.5396819114685059,
       "learning_rate": 1.0365030656894759e-06,
-      "loss": 1.1017,
+      "loss": 1.1019,
       "step": 5365
     },
     {
       "epoch": 0.9595282766014473,
-      "grad_norm": 0.520027756690979,
+      "grad_norm": 0.521092414855957,
       "learning_rate": 9.921919581682759e-07,
-      "loss": 1.0904,
+      "loss": 1.0899,
       "step": 5370
     },
     {
       "epoch": 0.9604216921290092,
-      "grad_norm": 0.5289562940597534,
+      "grad_norm": 0.5292440056800842,
       "learning_rate": 9.488440970696566e-07,
-      "loss": 1.1229,
+      "loss": 1.1232,
       "step": 5375
     },
     {
       "epoch": 0.961315107656571,
-      "grad_norm": 0.5180892944335938,
+      "grad_norm": 0.5165191292762756,
       "learning_rate": 9.064599041247124e-07,
-      "loss": 1.1399,
+      "loss": 1.1398,
       "step": 5380
     },
     {
       "epoch": 0.962208523184133,
-      "grad_norm": 0.5140677094459534,
+      "grad_norm": 0.5134493708610535,
       "learning_rate": 8.650397916890263e-07,
-      "loss": 1.0734,
+      "loss": 1.0735,
       "step": 5385
     },
     {
       "epoch": 0.9631019387116948,
-      "grad_norm": 0.5384522080421448,
+      "grad_norm": 0.5425078868865967,
       "learning_rate": 8.245841627386397e-07,
       "loss": 1.1054,
       "step": 5390
     },
     {
       "epoch": 0.9639953542392566,
-      "grad_norm": 0.5593591928482056,
+      "grad_norm": 0.556138277053833,
       "learning_rate": 7.850934108661556e-07,
-      "loss": 1.0937,
+      "loss": 1.0938,
       "step": 5395
     },
     {
       "epoch": 0.9648887697668186,
-      "grad_norm": 0.5436766743659973,
+      "grad_norm": 0.5389893054962158,
       "learning_rate": 7.465679202768749e-07,
-      "loss": 1.1055,
+      "loss": 1.1052,
       "step": 5400
     },
     {
       "epoch": 0.9657821852943804,
-      "grad_norm": 0.47669553756713867,
+      "grad_norm": 0.47252729535102844,
       "learning_rate": 7.090080657850884e-07,
-      "loss": 1.0633,
+      "loss": 1.063,
       "step": 5405
     },
     {
       "epoch": 0.9666756008219423,
-      "grad_norm": 0.49469995498657227,
+      "grad_norm": 0.4975879490375519,
       "learning_rate": 6.724142128104239e-07,
-      "loss": 1.0978,
+      "loss": 1.0983,
       "step": 5410
     },
     {
       "epoch": 0.9675690163495042,
-      "grad_norm": 0.565022349357605,
+      "grad_norm": 0.5652892589569092,
       "learning_rate": 6.367867173742603e-07,
-      "loss": 1.0757,
+      "loss": 1.075,
       "step": 5415
     },
     {
       "epoch": 0.968462431877066,
-      "grad_norm": 0.5273885726928711,
+      "grad_norm": 0.5284938216209412,
       "learning_rate": 6.021259260963085e-07,
-      "loss": 1.0793,
+      "loss": 1.0788,
       "step": 5420
     },
     {
       "epoch": 0.9693558474046279,
-      "grad_norm": 0.47813916206359863,
+      "grad_norm": 0.4806806445121765,
       "learning_rate": 5.684321761912247e-07,
       "loss": 1.0439,
       "step": 5425
     },
     {
       "epoch": 0.9702492629321897,
-      "grad_norm": 0.4988810122013092,
+      "grad_norm": 0.5021379590034485,
       "learning_rate": 5.357057954653244e-07,
-      "loss": 1.0706,
+      "loss": 1.0704,
       "step": 5430
     },
     {
       "epoch": 0.9711426784597517,
-      "grad_norm": 0.5139685869216919,
+      "grad_norm": 0.5142560005187988,
       "learning_rate": 5.039471023133624e-07,
-      "loss": 1.1125,
+      "loss": 1.1126,
       "step": 5435
     },
     {
       "epoch": 0.9720360939873135,
-      "grad_norm": 0.5338658690452576,
+      "grad_norm": 0.5299791693687439,
       "learning_rate": 4.7315640571550246e-07,
       "loss": 1.1047,
       "step": 5440
     },
     {
       "epoch": 0.9729295095148753,
-      "grad_norm": 0.49352288246154785,
+      "grad_norm": 0.49153584241867065,
       "learning_rate": 4.433340052342749e-07,
-      "loss": 1.0723,
+      "loss": 1.072,
       "step": 5445
     },
     {
       "epoch": 0.9738229250424373,
-      "grad_norm": 0.545481264591217,
+      "grad_norm": 0.5386480093002319,
       "learning_rate": 4.1448019101163473e-07,
-      "loss": 1.0697,
+      "loss": 1.0694,
       "step": 5450
     },
     {
       "epoch": 0.9747163405699991,
-      "grad_norm": 0.5542700886726379,
+      "grad_norm": 0.5521134734153748,
       "learning_rate": 3.865952437661968e-07,
-      "loss": 1.1145,
+      "loss": 1.1142,
       "step": 5455
     },
     {
       "epoch": 0.975609756097561,
-      "grad_norm": 0.5232776999473572,
+      "grad_norm": 0.5171558856964111,
       "learning_rate": 3.5967943479043867e-07,
-      "loss": 1.1051,
+      "loss": 1.1047,
       "step": 5460
     },
     {
       "epoch": 0.9765031716251228,
-      "grad_norm": 0.5434113144874573,
+      "grad_norm": 0.5417544841766357,
       "learning_rate": 3.3373302594814637e-07,
-      "loss": 1.1001,
+      "loss": 1.0995,
       "step": 5465
     },
     {
       "epoch": 0.9773965871526847,
-      "grad_norm": 0.5470346212387085,
+      "grad_norm": 0.5449820160865784,
       "learning_rate": 3.0875626967176165e-07,
-      "loss": 1.1008,
+      "loss": 1.1006,
       "step": 5470
     },
     {
       "epoch": 0.9782900026802466,
-      "grad_norm": 0.5273741483688354,
+      "grad_norm": 0.5323817133903503,
       "learning_rate": 2.8474940896003887e-07,
-      "loss": 1.1393,
+      "loss": 1.139,
       "step": 5475
     },
     {
       "epoch": 0.9791834182078084,
-      "grad_norm": 0.5102143287658691,
+      "grad_norm": 0.5092723369598389,
       "learning_rate": 2.617126773755696e-07,
-      "loss": 1.0718,
+      "loss": 1.0711,
       "step": 5480
     },
     {
       "epoch": 0.9800768337353704,
-      "grad_norm": 0.5177452564239502,
+      "grad_norm": 0.5194780230522156,
       "learning_rate": 2.3964629904259514e-07,
-      "loss": 1.1154,
+      "loss": 1.1153,
       "step": 5485
     },
     {
       "epoch": 0.9809702492629322,
-      "grad_norm": 0.5169101357460022,
+      "grad_norm": 0.5183728933334351,
       "learning_rate": 2.1855048864479754e-07,
-      "loss": 1.0984,
+      "loss": 1.0979,
       "step": 5490
     },
     {
       "epoch": 0.981863664790494,
-      "grad_norm": 0.5187830924987793,
+      "grad_norm": 0.5222030878067017,
       "learning_rate": 1.984254514232009e-07,
-      "loss": 1.0837,
+      "loss": 1.0832,
       "step": 5495
     },
     {
       "epoch": 0.982757080318056,
-      "grad_norm": 0.5481850504875183,
+      "grad_norm": 0.5462630987167358,
       "learning_rate": 1.7927138317417324e-07,
-      "loss": 1.0616,
+      "loss": 1.0618,
       "step": 5500
     },
     {
       "epoch": 0.9836504958456178,
-      "grad_norm": 0.5242062211036682,
+      "grad_norm": 0.5219123959541321,
       "learning_rate": 1.6108847024755015e-07,
       "loss": 1.1152,
       "step": 5505
     },
     {
       "epoch": 0.9845439113731796,
-      "grad_norm": 0.5562185645103455,
+      "grad_norm": 0.5546027421951294,
       "learning_rate": 1.4387688954478063e-07,
-      "loss": 1.1404,
+      "loss": 1.1397,
       "step": 5510
     },
     {
       "epoch": 0.9854373269007415,
-      "grad_norm": 0.490465372800827,
+      "grad_norm": 0.49054640531539917,
       "learning_rate": 1.276368085172397e-07,
-      "loss": 1.0926,
+      "loss": 1.0925,
       "step": 5515
     },
     {
       "epoch": 0.9863307424283034,
-      "grad_norm": 0.5415327548980713,
+      "grad_norm": 0.5430747270584106,
       "learning_rate": 1.1236838516459625e-07,
-      "loss": 1.0913,
+      "loss": 1.091,
       "step": 5520
     },
     {
       "epoch": 0.9872241579558653,
-      "grad_norm": 0.536961555480957,
+      "grad_norm": 0.5430759191513062,
       "learning_rate": 9.807176803325879e-08,
-      "loss": 1.1042,
+      "loss": 1.1034,
       "step": 5525
     },
     {
       "epoch": 0.9881175734834271,
-      "grad_norm": 0.5555692315101624,
+      "grad_norm": 0.554993212223053,
       "learning_rate": 8.474709621492105e-08,
-      "loss": 1.106,
+      "loss": 1.105,
       "step": 5530
     },
     {
       "epoch": 0.989010989010989,
-      "grad_norm": 0.5068987011909485,
+      "grad_norm": 0.5086091160774231,
       "learning_rate": 7.239449934525189e-08,
-      "loss": 1.0631,
+      "loss": 1.0637,
       "step": 5535
     },
     {
       "epoch": 0.9899044045385509,
-      "grad_norm": 0.5715160369873047,
+      "grad_norm": 0.5667893290519714,
       "learning_rate": 6.101409760260746e-08,
-      "loss": 1.0986,
+      "loss": 1.0987,
       "step": 5540
     },
     {
       "epoch": 0.9907978200661127,
-      "grad_norm": 0.5271539092063904,
+      "grad_norm": 0.5301327705383301,
       "learning_rate": 5.0606001706843264e-08,
       "loss": 1.0638,
       "step": 5545
     },
     {
       "epoch": 0.9916912355936747,
-      "grad_norm": 0.580915093421936,
+      "grad_norm": 0.5770794153213501,
       "learning_rate": 4.1170312918259456e-08,
-      "loss": 1.0884,
+      "loss": 1.089,
       "step": 5550
     },
     {
       "epoch": 0.9925846511212365,
-      "grad_norm": 0.5607919692993164,
+      "grad_norm": 0.5558438897132874,
       "learning_rate": 3.2707123036646026e-08,
-      "loss": 1.0791,
+      "loss": 1.0793,
       "step": 5555
     },
     {
       "epoch": 0.9934780666487983,
-      "grad_norm": 0.5168312788009644,
+      "grad_norm": 0.5180490016937256,
       "learning_rate": 2.5216514400305813e-08,
-      "loss": 1.1157,
+      "loss": 1.1154,
       "step": 5560
     },
     {
       "epoch": 0.9943714821763602,
-      "grad_norm": 0.5568081736564636,
+      "grad_norm": 0.5610867142677307,
       "learning_rate": 1.869855988534397e-08,
-      "loss": 1.0988,
+      "loss": 1.0993,
       "step": 5565
     },
     {
       "epoch": 0.9952648977039221,
-      "grad_norm": 0.5483577251434326,
+      "grad_norm": 0.5481888651847839,
       "learning_rate": 1.31533229049019e-08,
-      "loss": 1.1063,
+      "loss": 1.1059,
       "step": 5570
     },
     {
       "epoch": 0.996158313231484,
-      "grad_norm": 0.5298967957496643,
+      "grad_norm": 0.5289728045463562,
       "learning_rate": 8.580857408546639e-09,
-      "loss": 1.0713,
+      "loss": 1.0708,
       "step": 5575
     },
     {
       "epoch": 0.9970517287590458,
-      "grad_norm": 0.5338754057884216,
+      "grad_norm": 0.5380046963691711,
       "learning_rate": 4.9812078817934596e-09,
       "loss": 1.0486,
       "step": 5580
     },
     {
       "epoch": 0.9979451442866077,
-      "grad_norm": 0.5403038263320923,
+      "grad_norm": 0.5321928858757019,
       "learning_rate": 2.3544093455951654e-09,
-      "loss": 1.063,
+      "loss": 1.0629,
       "step": 5585
     },
     {
       "epoch": 0.9988385598141696,
-      "grad_norm": 0.5284872055053711,
+      "grad_norm": 0.5315663814544678,
       "learning_rate": 7.004873560645387e-10,
-      "loss": 1.099,
+      "loss": 1.0994,
       "step": 5590
     },
     {
       "epoch": 0.9997319753417314,
-      "grad_norm": 0.5430343747138977,
+      "grad_norm": 0.5456522703170776,
       "learning_rate": 1.9458004196781787e-11,
-      "loss": 1.0962,
+      "loss": 1.0959,
       "step": 5595
     },
     {
       "epoch": 0.9999106584472438,
-      "eval_loss": 1.0571058988571167,
-      "eval_runtime": 871.0685,
-      "eval_samples_per_second": 5.121,
+      "eval_loss": 1.0571272373199463,
+      "eval_runtime": 870.817,
+      "eval_samples_per_second": 5.123,
       "eval_steps_per_second": 0.641,
       "step": 5596
     },
@@ -7860,10 +7860,10 @@
       "epoch": 0.9999106584472438,
       "step": 5596,
       "total_flos": 4.151601388859687e+18,
-      "train_loss": 1.2674157931559251,
-      "train_runtime": 31909.1388,
-      "train_samples_per_second": 1.403,
-      "train_steps_per_second": 0.175
+      "train_loss": 1.267410275018411,
+      "train_runtime": 31869.1034,
+      "train_samples_per_second": 1.405,
+      "train_steps_per_second": 0.176
     }
   ],
   "logging_steps": 5,