diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4293 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.98769987699877,
+  "eval_steps": 500,
+  "global_step": 6090,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.016400164001640016,
+      "grad_norm": 0.49903953075408936,
+      "learning_rate": 0.00019999866944080257,
+      "loss": 1.1198,
+      "step": 10
+    },
+    {
+      "epoch": 0.03280032800328003,
+      "grad_norm": 0.38635629415512085,
+      "learning_rate": 0.00019999467779861793,
+      "loss": 0.957,
+      "step": 20
+    },
+    {
+      "epoch": 0.04920049200492005,
+      "grad_norm": 0.3574012815952301,
+      "learning_rate": 0.0001999880251796685,
+      "loss": 0.8963,
+      "step": 30
+    },
+    {
+      "epoch": 0.06560065600656007,
+      "grad_norm": 0.36165040731430054,
+      "learning_rate": 0.00019997871176098827,
+      "loss": 0.8893,
+      "step": 40
+    },
+    {
+      "epoch": 0.08200082000820008,
+      "grad_norm": 0.3633221983909607,
+      "learning_rate": 0.0001999667377904184,
+      "loss": 0.8788,
+      "step": 50
+    },
+    {
+      "epoch": 0.0984009840098401,
+      "grad_norm": 0.3308263123035431,
+      "learning_rate": 0.00019995210358660038,
+      "loss": 0.8597,
+      "step": 60
+    },
+    {
+      "epoch": 0.11480114801148011,
+      "grad_norm": 0.37404152750968933,
+      "learning_rate": 0.0001999348095389677,
+      "loss": 0.8394,
+      "step": 70
+    },
+    {
+      "epoch": 0.13120131201312013,
+      "grad_norm": 0.36840370297431946,
+      "learning_rate": 0.0001999148561077355,
+      "loss": 0.8305,
+      "step": 80
+    },
+    {
+      "epoch": 0.14760147601476015,
+      "grad_norm": 0.33561620116233826,
+      "learning_rate": 0.00019989224382388813,
+      "loss": 0.836,
+      "step": 90
+    },
+    {
+      "epoch": 0.16400164001640016,
+      "grad_norm": 0.34898167848587036,
+      "learning_rate": 0.00019986697328916526,
+      "loss": 0.8259,
+      "step": 100
+    },
+    {
+      "epoch": 0.18040180401804018,
+      "grad_norm": 0.32518813014030457,
+      "learning_rate": 0.00019983904517604576,
+      "loss": 0.8255,
+      "step": 110
+    },
+    {
+      "epoch": 0.1968019680196802,
+      "grad_norm": 0.34965017437934875,
+      "learning_rate": 0.00019980846022772978,
+      "loss": 0.8192,
+      "step": 120
+    },
+    {
+      "epoch": 0.2132021320213202,
+      "grad_norm": 0.3208518326282501,
+      "learning_rate": 0.00019977521925811903,
+      "loss": 0.8177,
+      "step": 130
+    },
+    {
+      "epoch": 0.22960229602296023,
+      "grad_norm": 0.3215318024158478,
+      "learning_rate": 0.000199739323151795,
+      "loss": 0.8046,
+      "step": 140
+    },
+    {
+      "epoch": 0.24600246002460024,
+      "grad_norm": 0.3197932541370392,
+      "learning_rate": 0.0001997007728639956,
+      "loss": 0.8031,
+      "step": 150
+    },
+    {
+      "epoch": 0.26240262402624026,
+      "grad_norm": 0.3127509355545044,
+      "learning_rate": 0.0001996595694205897,
+      "loss": 0.7911,
+      "step": 160
+    },
+    {
+      "epoch": 0.2788027880278803,
+      "grad_norm": 0.3898150324821472,
+      "learning_rate": 0.00019961571391804962,
+      "loss": 0.8013,
+      "step": 170
+    },
+    {
+      "epoch": 0.2952029520295203,
+      "grad_norm": 0.32089415192604065,
+      "learning_rate": 0.00019956920752342225,
+      "loss": 0.7959,
+      "step": 180
+    },
+    {
+      "epoch": 0.3116031160311603,
+      "grad_norm": 0.3503674864768982,
+      "learning_rate": 0.0001995200514742978,
+      "loss": 0.7869,
+      "step": 190
+    },
+    {
+      "epoch": 0.3280032800328003,
+      "grad_norm": 0.36325424909591675,
+      "learning_rate": 0.00019946824707877693,
+      "loss": 0.7813,
+      "step": 200
+    },
+    {
+      "epoch": 0.34440344403444034,
+      "grad_norm": 0.3318350613117218,
+      "learning_rate": 0.00019941379571543596,
+      "loss": 0.796,
+      "step": 210
+    },
+    {
+      "epoch": 0.36080360803608036,
+      "grad_norm": 0.3133643865585327,
+      "learning_rate": 0.00019935669883329013,
+      "loss": 0.7832,
+      "step": 220
+    },
+    {
+      "epoch": 0.3772037720377204,
+      "grad_norm": 0.3414137363433838,
+      "learning_rate": 0.00019929695795175507,
+      "loss": 0.7907,
+      "step": 230
+    },
+    {
+      "epoch": 0.3936039360393604,
+      "grad_norm": 0.32717064023017883,
+      "learning_rate": 0.00019923457466060636,
+      "loss": 0.7824,
+      "step": 240
+    },
+    {
+      "epoch": 0.4100041000410004,
+      "grad_norm": 0.3150332272052765,
+      "learning_rate": 0.00019916955061993725,
+      "loss": 0.7992,
+      "step": 250
+    },
+    {
+      "epoch": 0.4264042640426404,
+      "grad_norm": 0.3353354334831238,
+      "learning_rate": 0.00019910188756011446,
+      "loss": 0.7782,
+      "step": 260
+    },
+    {
+      "epoch": 0.44280442804428044,
+      "grad_norm": 0.2986792325973511,
+      "learning_rate": 0.00019903158728173205,
+      "loss": 0.7609,
+      "step": 270
+    },
+    {
+      "epoch": 0.45920459204592046,
+      "grad_norm": 0.3223719596862793,
+      "learning_rate": 0.00019895865165556377,
+      "loss": 0.7785,
+      "step": 280
+    },
+    {
+      "epoch": 0.4756047560475605,
+      "grad_norm": 0.3108189105987549,
+      "learning_rate": 0.00019888308262251285,
+      "loss": 0.7762,
+      "step": 290
+    },
+    {
+      "epoch": 0.4920049200492005,
+      "grad_norm": 0.3153323829174042,
+      "learning_rate": 0.00019880488219356087,
+      "loss": 0.7797,
+      "step": 300
+    },
+    {
+      "epoch": 0.5084050840508405,
+      "grad_norm": 0.33102768659591675,
+      "learning_rate": 0.00019872405244971374,
+      "loss": 0.7731,
+      "step": 310
+    },
+    {
+      "epoch": 0.5248052480524805,
+      "grad_norm": 0.29496216773986816,
+      "learning_rate": 0.00019864059554194669,
+      "loss": 0.7648,
+      "step": 320
+    },
+    {
+      "epoch": 0.5412054120541205,
+      "grad_norm": 0.3221481740474701,
+      "learning_rate": 0.00019855451369114676,
+      "loss": 0.7762,
+      "step": 330
+    },
+    {
+      "epoch": 0.5576055760557606,
+      "grad_norm": 0.3474864065647125,
+      "learning_rate": 0.000198465809188054,
+      "loss": 0.7691,
+      "step": 340
+    },
+    {
+      "epoch": 0.5740057400574006,
+      "grad_norm": 0.36715272068977356,
+      "learning_rate": 0.00019837448439320027,
+      "loss": 0.7751,
+      "step": 350
+    },
+    {
+      "epoch": 0.5904059040590406,
+      "grad_norm": 0.33955511450767517,
+      "learning_rate": 0.00019828054173684644,
+      "loss": 0.7791,
+      "step": 360
+    },
+    {
+      "epoch": 0.6068060680606806,
+      "grad_norm": 0.3461228311061859,
+      "learning_rate": 0.00019818398371891784,
+      "loss": 0.7728,
+      "step": 370
+    },
+    {
+      "epoch": 0.6232062320623206,
+      "grad_norm": 0.32821011543273926,
+      "learning_rate": 0.0001980848129089376,
+      "loss": 0.7736,
+      "step": 380
+    },
+    {
+      "epoch": 0.6396063960639606,
+      "grad_norm": 0.3576536476612091,
+      "learning_rate": 0.00019798303194595846,
+      "loss": 0.756,
+      "step": 390
+    },
+    {
+      "epoch": 0.6560065600656007,
+      "grad_norm": 0.3349688649177551,
+      "learning_rate": 0.00019787864353849232,
+      "loss": 0.765,
+      "step": 400
+    },
+    {
+      "epoch": 0.6724067240672407,
+      "grad_norm": 0.2960722744464874,
+      "learning_rate": 0.00019777165046443824,
+      "loss": 0.7676,
+      "step": 410
+    },
+    {
+      "epoch": 0.6888068880688807,
+      "grad_norm": 0.31184637546539307,
+      "learning_rate": 0.00019766205557100868,
+      "loss": 0.7366,
+      "step": 420
+    },
+    {
+      "epoch": 0.7052070520705207,
+      "grad_norm": 0.3365538418292999,
+      "learning_rate": 0.00019754986177465344,
+      "loss": 0.7647,
+      "step": 430
+    },
+    {
+      "epoch": 0.7216072160721607,
+      "grad_norm": 0.3039809763431549,
+      "learning_rate": 0.00019743507206098233,
+      "loss": 0.741,
+      "step": 440
+    },
+    {
+      "epoch": 0.7380073800738007,
+      "grad_norm": 0.32564470171928406,
+      "learning_rate": 0.00019731768948468549,
+      "loss": 0.7576,
+      "step": 450
+    },
+    {
+      "epoch": 0.7544075440754408,
+      "grad_norm": 0.33124250173568726,
+      "learning_rate": 0.00019719771716945227,
+      "loss": 0.7392,
+      "step": 460
+    },
+    {
+      "epoch": 0.7708077080770808,
+      "grad_norm": 0.3306678235530853,
+      "learning_rate": 0.00019707515830788802,
+      "loss": 0.7513,
+      "step": 470
+    },
+    {
+      "epoch": 0.7872078720787208,
+      "grad_norm": 0.31717410683631897,
+      "learning_rate": 0.00019695001616142915,
+      "loss": 0.7419,
+      "step": 480
+    },
+    {
+      "epoch": 0.8036080360803608,
+      "grad_norm": 0.3117675483226776,
+      "learning_rate": 0.00019682229406025635,
+      "loss": 0.7543,
+      "step": 490
+    },
+    {
+      "epoch": 0.8200082000820008,
+      "grad_norm": 0.3408834636211395,
+      "learning_rate": 0.0001966919954032059,
+      "loss": 0.7356,
+      "step": 500
+    },
+    {
+      "epoch": 0.8364083640836408,
+      "grad_norm": 0.3456558883190155,
+      "learning_rate": 0.0001965591236576794,
+      "loss": 0.7509,
+      "step": 510
+    },
+    {
+      "epoch": 0.8528085280852808,
+      "grad_norm": 0.30214881896972656,
+      "learning_rate": 0.00019642368235955125,
+      "loss": 0.7483,
+      "step": 520
+    },
+    {
+      "epoch": 0.8692086920869209,
+      "grad_norm": 0.3334808945655823,
+      "learning_rate": 0.0001962856751130748,
+      "loss": 0.7532,
+      "step": 530
+    },
+    {
+      "epoch": 0.8856088560885609,
+      "grad_norm": 0.3257792890071869,
+      "learning_rate": 0.00019614510559078625,
+      "loss": 0.7465,
+      "step": 540
+    },
+    {
+      "epoch": 0.9020090200902009,
+      "grad_norm": 0.4038065969944,
+      "learning_rate": 0.00019600197753340699,
+      "loss": 0.7555,
+      "step": 550
+    },
+    {
+      "epoch": 0.9184091840918409,
+      "grad_norm": 0.32042616605758667,
+      "learning_rate": 0.00019585629474974415,
+      "loss": 0.7395,
+      "step": 560
+    },
+    {
+      "epoch": 0.9348093480934809,
+      "grad_norm": 0.3661757707595825,
+      "learning_rate": 0.00019570806111658898,
+      "loss": 0.7447,
+      "step": 570
+    },
+    {
+      "epoch": 0.951209512095121,
+      "grad_norm": 0.31112104654312134,
+      "learning_rate": 0.0001955572805786141,
+      "loss": 0.7333,
+      "step": 580
+    },
+    {
+      "epoch": 0.967609676096761,
+      "grad_norm": 0.31758418679237366,
+      "learning_rate": 0.000195403957148268,
+      "loss": 0.7243,
+      "step": 590
+    },
+    {
+      "epoch": 0.984009840098401,
+      "grad_norm": 0.3203936219215393,
+      "learning_rate": 0.00019524809490566877,
+      "loss": 0.744,
+      "step": 600
+    },
+    {
+      "epoch": 1.000410004100041,
+      "grad_norm": 0.3630532920360565,
+      "learning_rate": 0.00019508969799849523,
+      "loss": 0.7249,
+      "step": 610
+    },
+    {
+      "epoch": 1.016810168101681,
+      "grad_norm": 0.31734269857406616,
+      "learning_rate": 0.00019492877064187654,
+      "loss": 0.7088,
+      "step": 620
+    },
+    {
+      "epoch": 1.033210332103321,
+      "grad_norm": 0.3637007176876068,
+      "learning_rate": 0.00019476531711828027,
+      "loss": 0.7139,
+      "step": 630
+    },
+    {
+      "epoch": 1.049610496104961,
+      "grad_norm": 0.3162935972213745,
+      "learning_rate": 0.00019459934177739813,
+      "loss": 0.7064,
+      "step": 640
+    },
+    {
+      "epoch": 1.066010660106601,
+      "grad_norm": 0.32640552520751953,
+      "learning_rate": 0.0001944308490360305,
+      "loss": 0.7161,
+      "step": 650
+    },
+    {
+      "epoch": 1.082410824108241,
+      "grad_norm": 0.34453850984573364,
+      "learning_rate": 0.0001942598433779687,
+      "loss": 0.7095,
+      "step": 660
+    },
+    {
+      "epoch": 1.098810988109881,
+      "grad_norm": 0.3382568061351776,
+      "learning_rate": 0.00019408632935387577,
+      "loss": 0.7092,
+      "step": 670
+    },
+    {
+      "epoch": 1.115211152111521,
+      "grad_norm": 0.3298267722129822,
+      "learning_rate": 0.0001939103115811653,
+      "loss": 0.7154,
+      "step": 680
+    },
+    {
+      "epoch": 1.1316113161131611,
+      "grad_norm": 0.33924031257629395,
+      "learning_rate": 0.00019373179474387858,
+      "loss": 0.6945,
+      "step": 690
+    },
+    {
+      "epoch": 1.1480114801148011,
+      "grad_norm": 0.3305610418319702,
+      "learning_rate": 0.0001935507835925601,
+      "loss": 0.7156,
+      "step": 700
+    },
+    {
+      "epoch": 1.1644116441164412,
+      "grad_norm": 0.3400373160839081,
+      "learning_rate": 0.00019336728294413083,
+      "loss": 0.7091,
+      "step": 710
+    },
+    {
+      "epoch": 1.1808118081180812,
+      "grad_norm": 0.3354150056838989,
+      "learning_rate": 0.00019318129768176032,
+      "loss": 0.7019,
+      "step": 720
+    },
+    {
+      "epoch": 1.1972119721197212,
+      "grad_norm": 0.3427719175815582,
+      "learning_rate": 0.00019299283275473657,
+      "loss": 0.7157,
+      "step": 730
+    },
+    {
+      "epoch": 1.2136121361213612,
+      "grad_norm": 0.3088740408420563,
+      "learning_rate": 0.00019280189317833445,
+      "loss": 0.7115,
+      "step": 740
+    },
+    {
+      "epoch": 1.2300123001230012,
+      "grad_norm": 0.34402570128440857,
+      "learning_rate": 0.0001926084840336821,
+      "loss": 0.7197,
+      "step": 750
+    },
+    {
+      "epoch": 1.2464124641246412,
+      "grad_norm": 0.3542528450489044,
+      "learning_rate": 0.0001924126104676259,
+      "loss": 0.7103,
+      "step": 760
+    },
+    {
+      "epoch": 1.2628126281262813,
+      "grad_norm": 0.3533455431461334,
+      "learning_rate": 0.00019221427769259333,
+      "loss": 0.7209,
+      "step": 770
+    },
+    {
+      "epoch": 1.2792127921279213,
+      "grad_norm": 0.3470947742462158,
+      "learning_rate": 0.00019201349098645434,
+      "loss": 0.7137,
+      "step": 780
+    },
+    {
+      "epoch": 1.2956129561295613,
+      "grad_norm": 0.3277159035205841,
+      "learning_rate": 0.0001918102556923809,
+      "loss": 0.7036,
+      "step": 790
+    },
+    {
+      "epoch": 1.3120131201312013,
+      "grad_norm": 0.35741642117500305,
+      "learning_rate": 0.0001916045772187048,
+      "loss": 0.7013,
+      "step": 800
+    },
+    {
+      "epoch": 1.3284132841328413,
+      "grad_norm": 0.33222687244415283,
+      "learning_rate": 0.0001913964610387738,
+      "loss": 0.7052,
+      "step": 810
+    },
+    {
+      "epoch": 1.3448134481344813,
+      "grad_norm": 0.3761126399040222,
+      "learning_rate": 0.00019118591269080574,
+      "loss": 0.695,
+      "step": 820
+    },
+    {
+      "epoch": 1.3612136121361214,
+      "grad_norm": 0.34687796235084534,
+      "learning_rate": 0.00019097293777774153,
+      "loss": 0.7053,
+      "step": 830
+    },
+    {
+      "epoch": 1.3776137761377614,
+      "grad_norm": 0.3391115367412567,
+      "learning_rate": 0.00019075754196709572,
+      "loss": 0.6842,
+      "step": 840
+    },
+    {
+      "epoch": 1.3940139401394014,
+      "grad_norm": 0.35560715198516846,
+      "learning_rate": 0.00019053973099080585,
+      "loss": 0.6925,
+      "step": 850
+    },
+    {
+      "epoch": 1.4104141041410414,
+      "grad_norm": 0.35144075751304626,
+      "learning_rate": 0.00019031951064507987,
+      "loss": 0.7063,
+      "step": 860
+    },
+    {
+      "epoch": 1.4268142681426814,
+      "grad_norm": 0.3776865005493164,
+      "learning_rate": 0.0001900968867902419,
+      "loss": 0.7157,
+      "step": 870
+    },
+    {
+      "epoch": 1.4432144321443214,
+      "grad_norm": 0.3723245859146118,
+      "learning_rate": 0.00018987186535057634,
+      "loss": 0.7064,
+      "step": 880
+    },
+    {
+      "epoch": 1.4596145961459615,
+      "grad_norm": 0.3623642921447754,
+      "learning_rate": 0.0001896444523141701,
+      "loss": 0.7056,
+      "step": 890
+    },
+    {
+      "epoch": 1.4760147601476015,
+      "grad_norm": 0.34806951880455017,
+      "learning_rate": 0.0001894146537327533,
+      "loss": 0.7086,
+      "step": 900
+    },
+    {
+      "epoch": 1.4924149241492415,
+      "grad_norm": 0.36614471673965454,
+      "learning_rate": 0.00018918247572153823,
+      "loss": 0.7032,
+      "step": 910
+    },
+    {
+      "epoch": 1.5088150881508815,
+      "grad_norm": 0.351053923368454,
+      "learning_rate": 0.00018894792445905674,
+      "loss": 0.6988,
+      "step": 920
+    },
+    {
+      "epoch": 1.5252152521525215,
+      "grad_norm": 0.37270405888557434,
+      "learning_rate": 0.00018871100618699554,
+      "loss": 0.6926,
+      "step": 930
+    },
+    {
+      "epoch": 1.5416154161541615,
+      "grad_norm": 0.3521835207939148,
+      "learning_rate": 0.00018847172721003043,
+      "loss": 0.7004,
+      "step": 940
+    },
+    {
+      "epoch": 1.5580155801558015,
+      "grad_norm": 0.37083443999290466,
+      "learning_rate": 0.00018823009389565818,
+      "loss": 0.7141,
+      "step": 950
+    },
+    {
+      "epoch": 1.5744157441574416,
+      "grad_norm": 0.36266541481018066,
+      "learning_rate": 0.00018798611267402746,
+      "loss": 0.7029,
+      "step": 960
+    },
+    {
+      "epoch": 1.5908159081590816,
+      "grad_norm": 0.34403783082962036,
+      "learning_rate": 0.0001877397900377674,
+      "loss": 0.7022,
+      "step": 970
+    },
+    {
+      "epoch": 1.6072160721607216,
+      "grad_norm": 0.37788769602775574,
+      "learning_rate": 0.00018749113254181498,
+      "loss": 0.7049,
+      "step": 980
+    },
+    {
+      "epoch": 1.6236162361623616,
+      "grad_norm": 0.35607653856277466,
+      "learning_rate": 0.00018724014680324057,
+      "loss": 0.6957,
+      "step": 990
+    },
+    {
+      "epoch": 1.6400164001640016,
+      "grad_norm": 0.34468725323677063,
+      "learning_rate": 0.00018698683950107184,
+      "loss": 0.7148,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6564165641656416,
+      "grad_norm": 0.36872032284736633,
+      "learning_rate": 0.00018673121737611598,
+      "loss": 0.6922,
+      "step": 1010
+    },
+    {
+      "epoch": 1.6728167281672817,
+      "grad_norm": 0.34786421060562134,
+      "learning_rate": 0.00018647328723078038,
+      "loss": 0.7021,
+      "step": 1020
+    },
+    {
+      "epoch": 1.6892168921689217,
+      "grad_norm": 0.370241641998291,
+      "learning_rate": 0.00018621305592889163,
+      "loss": 0.6789,
+      "step": 1030
+    },
+    {
+      "epoch": 1.7056170561705617,
+      "grad_norm": 0.3378838300704956,
+      "learning_rate": 0.00018595053039551274,
+      "loss": 0.6741,
+      "step": 1040
+    },
+    {
+      "epoch": 1.7220172201722017,
+      "grad_norm": 0.3661263883113861,
+      "learning_rate": 0.00018568571761675893,
+      "loss": 0.6808,
+      "step": 1050
+    },
+    {
+      "epoch": 1.7384173841738417,
+      "grad_norm": 0.3620483875274658,
+      "learning_rate": 0.0001854186246396118,
+      "loss": 0.6895,
+      "step": 1060
+    },
+    {
+      "epoch": 1.7548175481754817,
+      "grad_norm": 0.3466210663318634,
+      "learning_rate": 0.00018514925857173177,
+      "loss": 0.6876,
+      "step": 1070
+    },
+    {
+      "epoch": 1.7712177121771218,
+      "grad_norm": 0.36353030800819397,
+      "learning_rate": 0.0001848776265812687,
+      "loss": 0.6876,
+      "step": 1080
+    },
+    {
+      "epoch": 1.7876178761787618,
+      "grad_norm": 0.3688850998878479,
+      "learning_rate": 0.00018460373589667154,
+      "loss": 0.6881,
+      "step": 1090
+    },
+    {
+      "epoch": 1.8040180401804018,
+      "grad_norm": 0.35844072699546814,
+      "learning_rate": 0.00018432759380649562,
+      "loss": 0.7038,
+      "step": 1100
+    },
+    {
+      "epoch": 1.8204182041820418,
+      "grad_norm": 0.38838833570480347,
+      "learning_rate": 0.00018404920765920896,
+      "loss": 0.6961,
+      "step": 1110
+    },
+    {
+      "epoch": 1.8368183681836818,
+      "grad_norm": 0.34356170892715454,
+      "learning_rate": 0.00018376858486299647,
+      "loss": 0.6849,
+      "step": 1120
+    },
+    {
+      "epoch": 1.8532185321853218,
+      "grad_norm": 0.3590626120567322,
+      "learning_rate": 0.00018348573288556308,
+      "loss": 0.68,
+      "step": 1130
+    },
+    {
+      "epoch": 1.8696186961869619,
+      "grad_norm": 0.3639477491378784,
+      "learning_rate": 0.00018320065925393468,
+      "loss": 0.6996,
+      "step": 1140
+    },
+    {
+      "epoch": 1.8860188601886019,
+      "grad_norm": 0.39745190739631653,
+      "learning_rate": 0.00018291337155425821,
+      "loss": 0.6901,
+      "step": 1150
+    },
+    {
+      "epoch": 1.902419024190242,
+      "grad_norm": 0.37302690744400024,
+      "learning_rate": 0.0001826238774315995,
+      "loss": 0.6784,
+      "step": 1160
+    },
+    {
+      "epoch": 1.918819188191882,
+      "grad_norm": 0.37480470538139343,
+      "learning_rate": 0.00018233218458973984,
+      "loss": 0.6806,
+      "step": 1170
+    },
+    {
+      "epoch": 1.935219352193522,
+      "grad_norm": 0.37153083086013794,
+      "learning_rate": 0.00018203830079097113,
+      "loss": 0.6853,
+      "step": 1180
+    },
+    {
+      "epoch": 1.951619516195162,
+      "grad_norm": 0.3530969023704529,
+      "learning_rate": 0.00018174223385588917,
+      "loss": 0.6785,
+      "step": 1190
+    },
+    {
+      "epoch": 1.968019680196802,
+      "grad_norm": 0.37324872612953186,
+      "learning_rate": 0.00018144399166318572,
+      "loss": 0.6733,
+      "step": 1200
+    },
+    {
+      "epoch": 1.984419844198442,
+      "grad_norm": 0.3680344223976135,
+      "learning_rate": 0.00018114358214943853,
+      "loss": 0.6885,
+      "step": 1210
+    },
+    {
+      "epoch": 2.000820008200082,
+      "grad_norm": 0.42146405577659607,
+      "learning_rate": 0.00018084101330890048,
+      "loss": 0.6759,
+      "step": 1220
+    },
+    {
+      "epoch": 2.0172201722017222,
+      "grad_norm": 0.3808436989784241,
+      "learning_rate": 0.00018053629319328662,
+      "loss": 0.655,
+      "step": 1230
+    },
+    {
+      "epoch": 2.033620336203362,
+      "grad_norm": 0.39610400795936584,
+      "learning_rate": 0.00018022942991156,
+      "loss": 0.6675,
+      "step": 1240
+    },
+    {
+      "epoch": 2.0500205002050023,
+      "grad_norm": 0.3906943202018738,
+      "learning_rate": 0.00017992043162971588,
+      "loss": 0.6451,
+      "step": 1250
+    },
+    {
+      "epoch": 2.066420664206642,
+      "grad_norm": 0.4022831916809082,
+      "learning_rate": 0.00017960930657056438,
+      "loss": 0.6699,
+      "step": 1260
+    },
+    {
+      "epoch": 2.0828208282082823,
+      "grad_norm": 0.41200950741767883,
+      "learning_rate": 0.00017929606301351168,
+      "loss": 0.6559,
+      "step": 1270
+    },
+    {
+      "epoch": 2.099220992209922,
+      "grad_norm": 0.37226754426956177,
+      "learning_rate": 0.00017898070929433965,
+      "loss": 0.655,
+      "step": 1280
+    },
+    {
+      "epoch": 2.1156211562115623,
+      "grad_norm": 0.3640751540660858,
+      "learning_rate": 0.00017866325380498416,
+      "loss": 0.6471,
+      "step": 1290
+    },
+    {
+      "epoch": 2.132021320213202,
+      "grad_norm": 0.37838858366012573,
+      "learning_rate": 0.00017834370499331166,
+      "loss": 0.651,
+      "step": 1300
+    },
+    {
+      "epoch": 2.1484214842148424,
+      "grad_norm": 0.3831733167171478,
+      "learning_rate": 0.0001780220713628943,
+      "loss": 0.6504,
+      "step": 1310
+    },
+    {
+      "epoch": 2.164821648216482,
+      "grad_norm": 0.39556068181991577,
+      "learning_rate": 0.0001776983614727838,
+      "loss": 0.6501,
+      "step": 1320
+    },
+    {
+      "epoch": 2.181221812218122,
+      "grad_norm": 0.3833105266094208,
+      "learning_rate": 0.00017737258393728364,
+      "loss": 0.6598,
+      "step": 1330
+    },
+    {
+      "epoch": 2.197621976219762,
+      "grad_norm": 0.40265655517578125,
+      "learning_rate": 0.00017704474742571969,
+      "loss": 0.663,
+      "step": 1340
+    },
+    {
+      "epoch": 2.2140221402214024,
+      "grad_norm": 0.4098280072212219,
+      "learning_rate": 0.00017671486066220965,
+      "loss": 0.6567,
+      "step": 1350
+    },
+    {
+      "epoch": 2.230422304223042,
+      "grad_norm": 0.4019940197467804,
+      "learning_rate": 0.0001763829324254309,
+      "loss": 0.6717,
+      "step": 1360
+    },
+    {
+      "epoch": 2.246822468224682,
+      "grad_norm": 0.3930509388446808,
+      "learning_rate": 0.00017604897154838685,
+      "loss": 0.6544,
+      "step": 1370
+    },
+    {
+      "epoch": 2.2632226322263223,
+      "grad_norm": 0.40629473328590393,
+      "learning_rate": 0.00017571298691817177,
+      "loss": 0.65,
+      "step": 1380
+    },
+    {
+      "epoch": 2.2796227962279625,
+      "grad_norm": 0.41943445801734924,
+      "learning_rate": 0.00017537498747573443,
+      "loss": 0.6612,
+      "step": 1390
+    },
+    {
+      "epoch": 2.2960229602296023,
+      "grad_norm": 0.3832157552242279,
+      "learning_rate": 0.00017503498221564025,
+      "loss": 0.6599,
+      "step": 1400
+    },
+    {
+      "epoch": 2.312423124231242,
+      "grad_norm": 0.3786883056163788,
+      "learning_rate": 0.0001746929801858317,
+      "loss": 0.6636,
+      "step": 1410
+    },
+    {
+      "epoch": 2.3288232882328823,
+      "grad_norm": 0.40903323888778687,
+      "learning_rate": 0.0001743489904873877,
+      "loss": 0.6462,
+      "step": 1420
+    },
+    {
+      "epoch": 2.3452234522345226,
+      "grad_norm": 0.3858684003353119,
+      "learning_rate": 0.00017400302227428143,
+      "loss": 0.6556,
+      "step": 1430
+    },
+    {
+      "epoch": 2.3616236162361623,
+      "grad_norm": 0.43149909377098083,
+      "learning_rate": 0.0001736550847531366,
+      "loss": 0.6591,
+      "step": 1440
+    },
+    {
+      "epoch": 2.378023780237802,
+      "grad_norm": 0.43920159339904785,
+      "learning_rate": 0.00017330518718298264,
+      "loss": 0.6609,
+      "step": 1450
+    },
+    {
+      "epoch": 2.3944239442394424,
+      "grad_norm": 0.39907246828079224,
+      "learning_rate": 0.0001729533388750081,
+      "loss": 0.6572,
+      "step": 1460
+    },
+    {
+      "epoch": 2.4108241082410826,
+      "grad_norm": 0.3920808434486389,
+      "learning_rate": 0.0001725995491923131,
+      "loss": 0.6497,
+      "step": 1470
+    },
+    {
+      "epoch": 2.4272242722427224,
+      "grad_norm": 0.39895495772361755,
+      "learning_rate": 0.00017224382754965989,
+      "loss": 0.646,
+      "step": 1480
+    },
+    {
+      "epoch": 2.443624436244362,
+      "grad_norm": 0.41222187876701355,
+      "learning_rate": 0.00017188618341322254,
+      "loss": 0.6458,
+      "step": 1490
+    },
+    {
+      "epoch": 2.4600246002460024,
+      "grad_norm": 0.41394343972206116,
+      "learning_rate": 0.00017152662630033505,
+      "loss": 0.6528,
+      "step": 1500
+    },
+    {
+      "epoch": 2.4764247642476427,
+      "grad_norm": 0.4161871373653412,
+      "learning_rate": 0.00017116516577923775,
+      "loss": 0.6414,
+      "step": 1510
+    },
+    {
+      "epoch": 2.4928249282492825,
+      "grad_norm": 0.43303370475769043,
+      "learning_rate": 0.00017080181146882317,
+      "loss": 0.6573,
+      "step": 1520
+    },
+    {
+      "epoch": 2.5092250922509223,
+      "grad_norm": 0.3973419964313507,
+      "learning_rate": 0.00017043657303837963,
+      "loss": 0.6541,
+      "step": 1530
+    },
+    {
+      "epoch": 2.5256252562525625,
+      "grad_norm": 0.3919370472431183,
+      "learning_rate": 0.00017006946020733425,
+      "loss": 0.6626,
+      "step": 1540
+    },
+    {
+      "epoch": 2.5420254202542027,
+      "grad_norm": 0.4177446663379669,
+      "learning_rate": 0.00016970048274499408,
+      "loss": 0.6478,
+      "step": 1550
+    },
+    {
+      "epoch": 2.5584255842558425,
+      "grad_norm": 0.39967086911201477,
+      "learning_rate": 0.0001693296504702862,
+      "loss": 0.6679,
+      "step": 1560
+    },
+    {
+      "epoch": 2.5748257482574823,
+      "grad_norm": 0.4124162495136261,
+      "learning_rate": 0.00016895697325149657,
+      "loss": 0.6597,
+      "step": 1570
+    },
+    {
+      "epoch": 2.5912259122591226,
+      "grad_norm": 0.37739571928977966,
+      "learning_rate": 0.00016858246100600715,
+      "loss": 0.6578,
+      "step": 1580
+    },
+    {
+      "epoch": 2.607626076260763,
+      "grad_norm": 0.4335273802280426,
+      "learning_rate": 0.00016820612370003221,
+      "loss": 0.6608,
+      "step": 1590
+    },
+    {
+      "epoch": 2.6240262402624026,
+      "grad_norm": 0.3718789517879486,
+      "learning_rate": 0.00016782797134835305,
+      "loss": 0.6447,
+      "step": 1600
+    },
+    {
+      "epoch": 2.6404264042640424,
+      "grad_norm": 0.41410380601882935,
+      "learning_rate": 0.0001674480140140514,
+      "loss": 0.6562,
+      "step": 1610
+    },
+    {
+      "epoch": 2.6568265682656826,
+      "grad_norm": 0.4115227460861206,
+      "learning_rate": 0.00016706626180824186,
+      "loss": 0.6451,
+      "step": 1620
+    },
+    {
+      "epoch": 2.673226732267323,
+      "grad_norm": 0.4145969748497009,
+      "learning_rate": 0.00016668272488980254,
+      "loss": 0.6326,
+      "step": 1630
+    },
+    {
+      "epoch": 2.6896268962689627,
+      "grad_norm": 0.42359670996665955,
+      "learning_rate": 0.00016629741346510496,
+      "loss": 0.6324,
+      "step": 1640
+    },
+    {
+      "epoch": 2.7060270602706025,
+      "grad_norm": 0.38336479663848877,
+      "learning_rate": 0.0001659103377877423,
+      "loss": 0.6508,
+      "step": 1650
+    },
+    {
+      "epoch": 2.7224272242722427,
+      "grad_norm": 0.4128510057926178,
+      "learning_rate": 0.00016552150815825662,
+      "loss": 0.6538,
+      "step": 1660
+    },
+    {
+      "epoch": 2.738827388273883,
+      "grad_norm": 0.47073811292648315,
+      "learning_rate": 0.0001651309349238647,
+      "loss": 0.6574,
+      "step": 1670
+    },
+    {
+      "epoch": 2.7552275522755227,
+      "grad_norm": 0.4234805703163147,
+      "learning_rate": 0.00016473862847818277,
+      "loss": 0.6643,
+      "step": 1680
+    },
+    {
+      "epoch": 2.7716277162771625,
+      "grad_norm": 0.4076490104198456,
+      "learning_rate": 0.0001643445992609498,
+      "loss": 0.6509,
+      "step": 1690
+    },
+    {
+      "epoch": 2.7880278802788028,
+      "grad_norm": 0.4320470988750458,
+      "learning_rate": 0.00016394885775774975,
+      "loss": 0.643,
+      "step": 1700
+    },
+    {
+      "epoch": 2.804428044280443,
+      "grad_norm": 0.409150630235672,
+      "learning_rate": 0.00016355141449973256,
+      "loss": 0.6452,
+      "step": 1710
+    },
+    {
+      "epoch": 2.820828208282083,
+      "grad_norm": 0.46668359637260437,
+      "learning_rate": 0.0001631522800633339,
+      "loss": 0.645,
+      "step": 1720
+    },
+    {
+      "epoch": 2.8372283722837226,
+      "grad_norm": 0.4314485192298889,
+      "learning_rate": 0.00016275146506999365,
+      "loss": 0.6374,
+      "step": 1730
+    },
+    {
+      "epoch": 2.853628536285363,
+      "grad_norm": 0.41864487528800964,
+      "learning_rate": 0.00016234898018587337,
+      "loss": 0.6343,
+      "step": 1740
+    },
+    {
+      "epoch": 2.870028700287003,
+      "grad_norm": 0.4255831241607666,
+      "learning_rate": 0.0001619448361215723,
+      "loss": 0.6614,
+      "step": 1750
+    },
+    {
+      "epoch": 2.886428864288643,
+      "grad_norm": 0.411807656288147,
+      "learning_rate": 0.0001615390436318425,
+      "loss": 0.6372,
+      "step": 1760
+    },
+    {
+      "epoch": 2.9028290282902827,
+      "grad_norm": 0.41166943311691284,
+      "learning_rate": 0.0001611316135153026,
+      "loss": 0.6505,
+      "step": 1770
+    },
+    {
+      "epoch": 2.919229192291923,
+      "grad_norm": 0.41985031962394714,
+      "learning_rate": 0.0001607225566141503,
+      "loss": 0.6515,
+      "step": 1780
+    },
+    {
+      "epoch": 2.935629356293563,
+      "grad_norm": 0.40023133158683777,
+      "learning_rate": 0.0001603118838138741,
+      "loss": 0.6466,
+      "step": 1790
+    },
+    {
+      "epoch": 2.952029520295203,
+      "grad_norm": 0.4138244688510895,
+      "learning_rate": 0.0001598996060429634,
+      "loss": 0.6475,
+      "step": 1800
+    },
+    {
+      "epoch": 2.9684296842968427,
+      "grad_norm": 0.39823102951049805,
+      "learning_rate": 0.0001594857342726178,
+      "loss": 0.6419,
+      "step": 1810
+    },
+    {
+      "epoch": 2.984829848298483,
+      "grad_norm": 0.3888455033302307,
+      "learning_rate": 0.0001590702795164551,
+      "loss": 0.6416,
+      "step": 1820
+    },
+    {
+      "epoch": 3.001230012300123,
+      "grad_norm": 0.5585771799087524,
+      "learning_rate": 0.0001586532528302183,
+      "loss": 0.6565,
+      "step": 1830
+    },
+    {
+      "epoch": 3.017630176301763,
+      "grad_norm": 0.4332481026649475,
+      "learning_rate": 0.00015823466531148124,
+      "loss": 0.619,
+      "step": 1840
+    },
+    {
+      "epoch": 3.0340303403034032,
+      "grad_norm": 0.44493553042411804,
+      "learning_rate": 0.0001578145280993533,
+      "loss": 0.6057,
+      "step": 1850
+    },
+    {
+      "epoch": 3.050430504305043,
+      "grad_norm": 0.43917590379714966,
+      "learning_rate": 0.0001573928523741832,
+      "loss": 0.6058,
+      "step": 1860
+    },
+    {
+      "epoch": 3.0668306683066833,
+      "grad_norm": 0.40770649909973145,
+      "learning_rate": 0.0001569696493572612,
+      "loss": 0.6092,
+      "step": 1870
+    },
+    {
+      "epoch": 3.083230832308323,
+      "grad_norm": 0.44579923152923584,
+      "learning_rate": 0.0001565449303105207,
+      "loss": 0.6272,
+      "step": 1880
+    },
+    {
+      "epoch": 3.0996309963099633,
+      "grad_norm": 0.4329695403575897,
+      "learning_rate": 0.00015611870653623825,
+      "loss": 0.6238,
+      "step": 1890
+    },
+    {
+      "epoch": 3.116031160311603,
+      "grad_norm": 0.44702407717704773,
+      "learning_rate": 0.00015569098937673318,
+      "loss": 0.6124,
+      "step": 1900
+    },
+    {
+      "epoch": 3.1324313243132433,
+      "grad_norm": 0.43535304069519043,
+      "learning_rate": 0.00015526179021406553,
+      "loss": 0.6179,
+      "step": 1910
+    },
+    {
+      "epoch": 3.148831488314883,
+      "grad_norm": 0.4820752739906311,
+      "learning_rate": 0.0001548311204697331,
+      "loss": 0.6228,
+      "step": 1920
+    },
+    {
+      "epoch": 3.1652316523165234,
+      "grad_norm": 0.4629499614238739,
+      "learning_rate": 0.00015439899160436772,
+      "loss": 0.6189,
+      "step": 1930
+    },
+    {
+      "epoch": 3.181631816318163,
+      "grad_norm": 0.4845089316368103,
+      "learning_rate": 0.00015396541511743012,
+      "loss": 0.6229,
+      "step": 1940
+    },
+    {
+      "epoch": 3.1980319803198034,
+      "grad_norm": 0.46592435240745544,
+      "learning_rate": 0.00015353040254690393,
+      "loss": 0.6178,
+      "step": 1950
+    },
+    {
+      "epoch": 3.214432144321443,
+      "grad_norm": 0.44851964712142944,
+      "learning_rate": 0.0001530939654689887,
+      "loss": 0.6246,
+      "step": 1960
+    },
+    {
+      "epoch": 3.2308323083230834,
+      "grad_norm": 0.4483705461025238,
+      "learning_rate": 0.00015265611549779176,
+      "loss": 0.6204,
+      "step": 1970
+    },
+    {
+      "epoch": 3.2472324723247232,
+      "grad_norm": 0.45693984627723694,
+      "learning_rate": 0.00015221686428501928,
+      "loss": 0.6061,
+      "step": 1980
+    },
+    {
+      "epoch": 3.2636326363263635,
+      "grad_norm": 0.46255984902381897,
+      "learning_rate": 0.0001517762235196661,
+      "loss": 0.6058,
+      "step": 1990
+    },
+    {
+      "epoch": 3.2800328003280033,
+      "grad_norm": 0.471983939409256,
+      "learning_rate": 0.00015133420492770462,
+      "loss": 0.6157,
+      "step": 2000
+    },
+    {
+      "epoch": 3.296432964329643,
+      "grad_norm": 0.49640896916389465,
+      "learning_rate": 0.0001508908202717729,
+      "loss": 0.6237,
+      "step": 2010
+    },
+    {
+      "epoch": 3.3128331283312833,
+      "grad_norm": 0.4903806149959564,
+      "learning_rate": 0.00015044608135086164,
+      "loss": 0.6339,
+      "step": 2020
+    },
+    {
+      "epoch": 3.3292332923329235,
+      "grad_norm": 0.47123804688453674,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.6319,
+      "step": 2030
+    },
+    {
+      "epoch": 3.3456334563345633,
+      "grad_norm": 0.47124335169792175,
+      "learning_rate": 0.00014955258808994096,
+      "loss": 0.6119,
+      "step": 2040
+    },
+    {
+      "epoch": 3.362033620336203,
+      "grad_norm": 0.4579004645347595,
+      "learning_rate": 0.00014910385752684506,
+      "loss": 0.6042,
+      "step": 2050
+    },
+    {
+      "epoch": 3.3784337843378434,
+      "grad_norm": 0.4778992235660553,
+      "learning_rate": 0.0001486538202519639,
+      "loss": 0.6187,
+      "step": 2060
+    },
+    {
+      "epoch": 3.3948339483394836,
+      "grad_norm": 0.4603955149650574,
+      "learning_rate": 0.0001482024882413222,
+      "loss": 0.6166,
+      "step": 2070
+    },
+    {
+      "epoch": 3.4112341123411234,
+      "grad_norm": 0.4642026126384735,
+      "learning_rate": 0.00014774987350539911,
+      "loss": 0.6183,
+      "step": 2080
+    },
+    {
+      "epoch": 3.427634276342763,
+      "grad_norm": 0.5194045305252075,
+      "learning_rate": 0.00014729598808880861,
+      "loss": 0.617,
+      "step": 2090
+    },
+    {
+      "epoch": 3.4440344403444034,
+      "grad_norm": 0.4535214900970459,
+      "learning_rate": 0.00014684084406997903,
+      "loss": 0.6301,
+      "step": 2100
+    },
+    {
+      "epoch": 3.4604346043460437,
+      "grad_norm": 0.47860169410705566,
+      "learning_rate": 0.00014638445356083155,
+      "loss": 0.6113,
+      "step": 2110
+    },
+    {
+      "epoch": 3.4768347683476835,
+      "grad_norm": 0.4803871810436249,
+      "learning_rate": 0.000145926828706458,
+      "loss": 0.6122,
+      "step": 2120
+    },
+    {
+      "epoch": 3.4932349323493233,
+      "grad_norm": 0.47383010387420654,
+      "learning_rate": 0.00014546798168479756,
+      "loss": 0.6137,
+      "step": 2130
+    },
+    {
+      "epoch": 3.5096350963509635,
+      "grad_norm": 0.47791990637779236,
+      "learning_rate": 0.0001450079247063127,
+      "loss": 0.6215,
+      "step": 2140
+    },
+    {
+      "epoch": 3.5260352603526037,
+      "grad_norm": 0.4606546461582184,
+      "learning_rate": 0.0001445466700136643,
+      "loss": 0.6022,
+      "step": 2150
+    },
+    {
+      "epoch": 3.5424354243542435,
+      "grad_norm": 0.4335190951824188,
+      "learning_rate": 0.00014408422988138584,
+      "loss": 0.6089,
+      "step": 2160
+    },
+    {
+      "epoch": 3.5588355883558833,
+      "grad_norm": 0.45156562328338623,
+      "learning_rate": 0.00014362061661555675,
+      "loss": 0.6251,
+      "step": 2170
+    },
+    {
+      "epoch": 3.5752357523575236,
+      "grad_norm": 0.43693843483924866,
+      "learning_rate": 0.00014315584255347494,
+      "loss": 0.62,
+      "step": 2180
+    },
+    {
+      "epoch": 3.591635916359164,
+      "grad_norm": 0.46648937463760376,
+      "learning_rate": 0.00014268992006332846,
+      "loss": 0.6336,
+      "step": 2190
+    },
+    {
+      "epoch": 3.6080360803608036,
+      "grad_norm": 0.47864094376564026,
+      "learning_rate": 0.00014222286154386642,
+      "loss": 0.6079,
+      "step": 2200
+    },
+    {
+      "epoch": 3.6244362443624434,
+      "grad_norm": 0.43990063667297363,
+      "learning_rate": 0.00014175467942406894,
+      "loss": 0.6321,
+      "step": 2210
+    },
+    {
+      "epoch": 3.6408364083640836,
+      "grad_norm": 0.4700396656990051,
+      "learning_rate": 0.0001412853861628166,
+      "loss": 0.6195,
+      "step": 2220
+    },
+    {
+      "epoch": 3.657236572365724,
+      "grad_norm": 0.45286816358566284,
+      "learning_rate": 0.00014081499424855863,
+      "loss": 0.6236,
+      "step": 2230
+    },
+    {
+      "epoch": 3.6736367363673637,
+      "grad_norm": 0.4623110294342041,
+      "learning_rate": 0.00014034351619898088,
+      "loss": 0.6269,
+      "step": 2240
+    },
+    {
+      "epoch": 3.6900369003690034,
+      "grad_norm": 0.5022940039634705,
+      "learning_rate": 0.00013987096456067236,
+      "loss": 0.6107,
+      "step": 2250
+    },
+    {
+      "epoch": 3.7064370643706437,
+      "grad_norm": 0.49338841438293457,
+      "learning_rate": 0.00013939735190879172,
+      "loss": 0.5996,
+      "step": 2260
+    },
+    {
+      "epoch": 3.722837228372284,
+      "grad_norm": 0.4769245982170105,
+      "learning_rate": 0.0001389226908467323,
+      "loss": 0.6074,
+      "step": 2270
+    },
+    {
+      "epoch": 3.7392373923739237,
+      "grad_norm": 0.48450419306755066,
+      "learning_rate": 0.00013844699400578696,
+      "loss": 0.6092,
+      "step": 2280
+    },
+    {
+      "epoch": 3.7556375563755635,
+      "grad_norm": 0.46382635831832886,
+      "learning_rate": 0.00013797027404481184,
+      "loss": 0.6154,
+      "step": 2290
+    },
+    {
+      "epoch": 3.7720377203772038,
+      "grad_norm": 0.4822761118412018,
+      "learning_rate": 0.00013749254364988956,
+      "loss": 0.612,
+      "step": 2300
+    },
+    {
+      "epoch": 3.788437884378844,
+      "grad_norm": 0.4664019048213959,
+      "learning_rate": 0.00013701381553399145,
+      "loss": 0.6003,
+      "step": 2310
+    },
+    {
+      "epoch": 3.804838048380484,
+      "grad_norm": 0.4592795670032501,
+      "learning_rate": 0.00013653410243663952,
+      "loss": 0.6122,
+      "step": 2320
+    },
+    {
+      "epoch": 3.8212382123821236,
+      "grad_norm": 0.4780454635620117,
+      "learning_rate": 0.00013605341712356723,
+      "loss": 0.6216,
+      "step": 2330
+    },
+    {
+      "epoch": 3.837638376383764,
+      "grad_norm": 0.48119214177131653,
+      "learning_rate": 0.00013557177238637986,
+      "loss": 0.6119,
+      "step": 2340
+    },
+    {
+      "epoch": 3.854038540385404,
+      "grad_norm": 0.44965630769729614,
+      "learning_rate": 0.00013508918104221412,
+      "loss": 0.619,
+      "step": 2350
+    },
+    {
+      "epoch": 3.870438704387044,
+      "grad_norm": 0.4915274977684021,
+      "learning_rate": 0.00013460565593339705,
+      "loss": 0.6101,
+      "step": 2360
+    },
+    {
+      "epoch": 3.8868388683886836,
+      "grad_norm": 0.4430049955844879,
+      "learning_rate": 0.00013412120992710425,
+      "loss": 0.6125,
+      "step": 2370
+    },
+    {
+      "epoch": 3.903239032390324,
+      "grad_norm": 0.4661041796207428,
+      "learning_rate": 0.0001336358559150175,
+      "loss": 0.6177,
+      "step": 2380
+    },
+    {
+      "epoch": 3.919639196391964,
+      "grad_norm": 0.48300737142562866,
+      "learning_rate": 0.00013314960681298175,
+      "loss": 0.6039,
+      "step": 2390
+    },
+    {
+      "epoch": 3.936039360393604,
+      "grad_norm": 0.45341914892196655,
+      "learning_rate": 0.00013266247556066122,
+      "loss": 0.5989,
+      "step": 2400
+    },
+    {
+      "epoch": 3.9524395243952437,
+      "grad_norm": 0.46126899123191833,
+      "learning_rate": 0.00013217447512119533,
+      "loss": 0.6188,
+      "step": 2410
+    },
+    {
+      "epoch": 3.968839688396884,
+      "grad_norm": 0.4723737835884094,
+      "learning_rate": 0.0001316856184808535,
+      "loss": 0.6091,
+      "step": 2420
+    },
+    {
+      "epoch": 3.985239852398524,
+      "grad_norm": 0.4723500907421112,
+      "learning_rate": 0.0001311959186486898,
+      "loss": 0.6178,
+      "step": 2430
+    },
+    {
+      "epoch": 4.001640016400164,
+      "grad_norm": 0.4641455113887787,
+      "learning_rate": 0.00013070538865619642,
+      "loss": 0.6144,
+      "step": 2440
+    },
+    {
+      "epoch": 4.018040180401804,
+      "grad_norm": 0.5465942621231079,
+      "learning_rate": 0.00013021404155695725,
+      "loss": 0.5762,
+      "step": 2450
+    },
+    {
+      "epoch": 4.0344403444034445,
+      "grad_norm": 0.5195038318634033,
+      "learning_rate": 0.00012972189042630044,
+      "loss": 0.5936,
+      "step": 2460
+    },
+    {
+      "epoch": 4.050840508405084,
+      "grad_norm": 0.5215120911598206,
+      "learning_rate": 0.00012922894836095013,
+      "loss": 0.5797,
+      "step": 2470
+    },
+    {
+      "epoch": 4.067240672406724,
+      "grad_norm": 0.5008783340454102,
+      "learning_rate": 0.00012873522847867835,
+      "loss": 0.5887,
+      "step": 2480
+    },
+    {
+      "epoch": 4.083640836408364,
+      "grad_norm": 0.47933313250541687,
+      "learning_rate": 0.0001282407439179557,
+      "loss": 0.5831,
+      "step": 2490
+    },
+    {
+      "epoch": 4.1000410004100045,
+      "grad_norm": 0.5141870379447937,
+      "learning_rate": 0.00012774550783760182,
+      "loss": 0.5765,
+      "step": 2500
+    },
+    {
+      "epoch": 4.116441164411644,
+      "grad_norm": 0.4889928996562958,
+      "learning_rate": 0.0001272495334164351,
+      "loss": 0.5834,
+      "step": 2510
+    },
+    {
+      "epoch": 4.132841328413284,
+      "grad_norm": 0.5281891822814941,
+      "learning_rate": 0.00012675283385292212,
+      "loss": 0.5778,
+      "step": 2520
+    },
+    {
+      "epoch": 4.149241492414924,
+      "grad_norm": 0.5182332396507263,
+      "learning_rate": 0.00012625542236482628,
+      "loss": 0.5882,
+      "step": 2530
+    },
+    {
+      "epoch": 4.165641656416565,
+      "grad_norm": 0.554658055305481,
+      "learning_rate": 0.00012575731218885625,
+      "loss": 0.5871,
+      "step": 2540
+    },
+    {
+      "epoch": 4.182041820418204,
+      "grad_norm": 0.5192410349845886,
+      "learning_rate": 0.00012525851658031352,
+      "loss": 0.5768,
+      "step": 2550
+    },
+    {
+      "epoch": 4.198441984419844,
+      "grad_norm": 0.5177326202392578,
+      "learning_rate": 0.0001247590488127398,
+      "loss": 0.5849,
+      "step": 2560
+    },
+    {
+      "epoch": 4.214842148421484,
+      "grad_norm": 0.5142077803611755,
+      "learning_rate": 0.0001242589221775637,
+      "loss": 0.5744,
+      "step": 2570
+    },
+    {
+      "epoch": 4.231242312423125,
+      "grad_norm": 0.48389384150505066,
+      "learning_rate": 0.00012375814998374712,
+      "loss": 0.5977,
+      "step": 2580
+    },
+    {
+      "epoch": 4.2476424764247644,
+      "grad_norm": 0.5631837248802185,
+      "learning_rate": 0.00012325674555743106,
+      "loss": 0.5827,
+      "step": 2590
+    },
+    {
+      "epoch": 4.264042640426404,
+      "grad_norm": 0.5213803052902222,
+      "learning_rate": 0.0001227547222415809,
+      "loss": 0.5815,
+      "step": 2600
+    },
+    {
+      "epoch": 4.280442804428044,
+      "grad_norm": 0.5063258409500122,
+      "learning_rate": 0.00012225209339563145,
+      "loss": 0.591,
+      "step": 2610
+    },
+    {
+      "epoch": 4.296842968429685,
+      "grad_norm": 0.503495454788208,
+      "learning_rate": 0.0001217488723951314,
+      "loss": 0.5872,
+      "step": 2620
+    },
+    {
+      "epoch": 4.3132431324313245,
+      "grad_norm": 0.5165495276451111,
+      "learning_rate": 0.00012124507263138736,
+      "loss": 0.5739,
+      "step": 2630
+    },
+    {
+      "epoch": 4.329643296432964,
+      "grad_norm": 0.4980062246322632,
+      "learning_rate": 0.00012074070751110751,
+      "loss": 0.5795,
+      "step": 2640
+    },
+    {
+      "epoch": 4.346043460434604,
+      "grad_norm": 0.5419421792030334,
+      "learning_rate": 0.00012023579045604485,
+      "loss": 0.5844,
+      "step": 2650
+    },
+    {
+      "epoch": 4.362443624436244,
+      "grad_norm": 0.5244302749633789,
+      "learning_rate": 0.00011973033490264001,
+      "loss": 0.5777,
+      "step": 2660
+    },
+    {
+      "epoch": 4.378843788437885,
+      "grad_norm": 0.5156043171882629,
+      "learning_rate": 0.0001192243543016637,
+      "loss": 0.5832,
+      "step": 2670
+    },
+    {
+      "epoch": 4.395243952439524,
+      "grad_norm": 0.5119035840034485,
+      "learning_rate": 0.00011871786211785876,
+      "loss": 0.59,
+      "step": 2680
+    },
+    {
+      "epoch": 4.411644116441164,
+      "grad_norm": 0.5079758167266846,
+      "learning_rate": 0.00011821087182958186,
+      "loss": 0.5652,
+      "step": 2690
+    },
+    {
+      "epoch": 4.428044280442805,
+      "grad_norm": 0.5077067613601685,
+      "learning_rate": 0.00011770339692844483,
+      "loss": 0.6014,
+      "step": 2700
+    },
+    {
+      "epoch": 4.444444444444445,
+      "grad_norm": 0.5312401056289673,
+      "learning_rate": 0.00011719545091895564,
+      "loss": 0.594,
+      "step": 2710
+    },
+    {
+      "epoch": 4.460844608446084,
+      "grad_norm": 0.5242477655410767,
+      "learning_rate": 0.00011668704731815892,
+      "loss": 0.5914,
+      "step": 2720
+    },
+    {
+      "epoch": 4.477244772447724,
+      "grad_norm": 0.48659422993659973,
+      "learning_rate": 0.0001161781996552765,
+      "loss": 0.5927,
+      "step": 2730
+    },
+    {
+      "epoch": 4.493644936449364,
+      "grad_norm": 0.5395858287811279,
+      "learning_rate": 0.00011566892147134705,
+      "loss": 0.5961,
+      "step": 2740
+    },
+    {
+      "epoch": 4.510045100451005,
+      "grad_norm": 0.503237783908844,
+      "learning_rate": 0.00011515922631886605,
+      "loss": 0.5763,
+      "step": 2750
+    },
+    {
+      "epoch": 4.5264452644526445,
+      "grad_norm": 0.5137144923210144,
+      "learning_rate": 0.00011464912776142494,
+      "loss": 0.5912,
+      "step": 2760
+    },
+    {
+      "epoch": 4.542845428454284,
+      "grad_norm": 0.5336122512817383,
+      "learning_rate": 0.00011413863937335028,
+      "loss": 0.5857,
+      "step": 2770
+    },
+    {
+      "epoch": 4.559245592455925,
+      "grad_norm": 0.5271728038787842,
+      "learning_rate": 0.00011362777473934248,
+      "loss": 0.5915,
+      "step": 2780
+    },
+    {
+      "epoch": 4.575645756457565,
+      "grad_norm": 0.5467897057533264,
+      "learning_rate": 0.00011311654745411425,
+      "loss": 0.5871,
+      "step": 2790
+    },
+    {
+      "epoch": 4.592045920459205,
+      "grad_norm": 0.5065737962722778,
+      "learning_rate": 0.00011260497112202895,
+      "loss": 0.5867,
+      "step": 2800
+    },
+    {
+      "epoch": 4.608446084460844,
+      "grad_norm": 0.5274237990379333,
+      "learning_rate": 0.00011209305935673844,
+      "loss": 0.5885,
+      "step": 2810
+    },
+    {
+      "epoch": 4.624846248462484,
+      "grad_norm": 0.5007864832878113,
+      "learning_rate": 0.00011158082578082089,
+      "loss": 0.5965,
+      "step": 2820
+    },
+    {
+      "epoch": 4.641246412464125,
+      "grad_norm": 0.48857831954956055,
+      "learning_rate": 0.0001110682840254182,
+      "loss": 0.5898,
+      "step": 2830
+    },
+    {
+      "epoch": 4.657646576465765,
+      "grad_norm": 0.5346599221229553,
+      "learning_rate": 0.00011055544772987335,
+      "loss": 0.5941,
+      "step": 2840
+    },
+    {
+      "epoch": 4.674046740467404,
+      "grad_norm": 0.4968061149120331,
+      "learning_rate": 0.00011004233054136725,
+      "loss": 0.5852,
+      "step": 2850
+    },
+    {
+      "epoch": 4.690446904469045,
+      "grad_norm": 0.5415840744972229,
+      "learning_rate": 0.00010952894611455584,
+      "loss": 0.5891,
+      "step": 2860
+    },
+    {
+      "epoch": 4.706847068470685,
+      "grad_norm": 0.5450642704963684,
+      "learning_rate": 0.00010901530811120655,
+      "loss": 0.5758,
+      "step": 2870
+    },
+    {
+      "epoch": 4.723247232472325,
+      "grad_norm": 0.5052106976509094,
+      "learning_rate": 0.00010850143019983474,
+      "loss": 0.5736,
+      "step": 2880
+    },
+    {
+      "epoch": 4.7396473964739645,
+      "grad_norm": 0.5136608481407166,
+      "learning_rate": 0.00010798732605534006,
+      "loss": 0.5769,
+      "step": 2890
+    },
+    {
+      "epoch": 4.756047560475604,
+      "grad_norm": 0.5369486808776855,
+      "learning_rate": 0.00010747300935864243,
+      "loss": 0.5762,
+      "step": 2900
+    },
+    {
+      "epoch": 4.772447724477245,
+      "grad_norm": 0.5404406189918518,
+      "learning_rate": 0.00010695849379631813,
+      "loss": 0.5862,
+      "step": 2910
+    },
+    {
+      "epoch": 4.788847888478885,
+      "grad_norm": 0.5491173267364502,
+      "learning_rate": 0.0001064437930602354,
+      "loss": 0.5783,
+      "step": 2920
+    },
+    {
+      "epoch": 4.805248052480525,
+      "grad_norm": 0.5794305801391602,
+      "learning_rate": 0.00010592892084719024,
+      "loss": 0.596,
+      "step": 2930
+    },
+    {
+      "epoch": 4.821648216482165,
+      "grad_norm": 0.5437578558921814,
+      "learning_rate": 0.00010541389085854176,
+      "loss": 0.5892,
+      "step": 2940
+    },
+    {
+      "epoch": 4.838048380483805,
+      "grad_norm": 0.5468413233757019,
+      "learning_rate": 0.00010489871679984777,
+      "loss": 0.57,
+      "step": 2950
+    },
+    {
+      "epoch": 4.854448544485445,
+      "grad_norm": 0.5444806814193726,
+      "learning_rate": 0.00010438341238049991,
+      "loss": 0.5754,
+      "step": 2960
+    },
+    {
+      "epoch": 4.870848708487085,
+      "grad_norm": 0.5279709696769714,
+      "learning_rate": 0.00010386799131335889,
+      "loss": 0.5757,
+      "step": 2970
+    },
+    {
+      "epoch": 4.887248872488724,
+      "grad_norm": 0.5229681134223938,
+      "learning_rate": 0.00010335246731438948,
+      "loss": 0.5851,
+      "step": 2980
+    },
+    {
+      "epoch": 4.903649036490365,
+      "grad_norm": 0.5356857180595398,
+      "learning_rate": 0.00010283685410229571,
+      "loss": 0.5846,
+      "step": 2990
+    },
+    {
+      "epoch": 4.920049200492005,
+      "grad_norm": 0.5306787490844727,
+      "learning_rate": 0.00010232116539815558,
+      "loss": 0.5595,
+      "step": 3000
+    },
+    {
+      "epoch": 4.936449364493645,
+      "grad_norm": 0.5103681683540344,
+      "learning_rate": 0.00010180541492505604,
+      "loss": 0.5878,
+      "step": 3010
+    },
+    {
+      "epoch": 4.952849528495285,
+      "grad_norm": 0.528273344039917,
+      "learning_rate": 0.00010128961640772785,
+      "loss": 0.5684,
+      "step": 3020
+    },
+    {
+      "epoch": 4.969249692496925,
+      "grad_norm": 0.5309355854988098,
+      "learning_rate": 0.00010077378357218021,
+      "loss": 0.5949,
+      "step": 3030
+    },
+    {
+      "epoch": 4.985649856498565,
+      "grad_norm": 0.5428647994995117,
+      "learning_rate": 0.00010025793014533558,
+      "loss": 0.5893,
+      "step": 3040
+    },
+    {
+      "epoch": 5.002050020500205,
+      "grad_norm": 0.519814133644104,
+      "learning_rate": 9.974206985466442e-05,
+      "loss": 0.5898,
+      "step": 3050
+    },
+    {
+      "epoch": 5.018450184501845,
+      "grad_norm": 0.5455579161643982,
+      "learning_rate": 9.92262164278198e-05,
+      "loss": 0.5495,
+      "step": 3060
+    },
+    {
+      "epoch": 5.034850348503485,
+      "grad_norm": 0.5572001338005066,
+      "learning_rate": 9.871038359227214e-05,
+      "loss": 0.543,
+      "step": 3070
+    },
+    {
+      "epoch": 5.051250512505125,
+      "grad_norm": 0.5723180174827576,
+      "learning_rate": 9.819458507494394e-05,
+      "loss": 0.5557,
+      "step": 3080
+    },
+    {
+      "epoch": 5.067650676506765,
+      "grad_norm": 0.5615038275718689,
+      "learning_rate": 9.767883460184443e-05,
+      "loss": 0.5505,
+      "step": 3090
+    },
+    {
+      "epoch": 5.0840508405084055,
+      "grad_norm": 0.5597648024559021,
+      "learning_rate": 9.71631458977043e-05,
+      "loss": 0.5599,
+      "step": 3100
+    },
+    {
+      "epoch": 5.100451004510045,
+      "grad_norm": 0.5660073161125183,
+      "learning_rate": 9.66475326856105e-05,
+      "loss": 0.5609,
+      "step": 3110
+    },
+    {
+      "epoch": 5.116851168511685,
+      "grad_norm": 0.5740461945533752,
+      "learning_rate": 9.613200868664112e-05,
+      "loss": 0.5505,
+      "step": 3120
+    },
+    {
+      "epoch": 5.133251332513325,
+      "grad_norm": 0.5641244649887085,
+      "learning_rate": 9.561658761950007e-05,
+      "loss": 0.5725,
+      "step": 3130
+    },
+    {
+      "epoch": 5.149651496514966,
+      "grad_norm": 0.5780383944511414,
+      "learning_rate": 9.510128320015224e-05,
+      "loss": 0.5584,
+      "step": 3140
+    },
+    {
+      "epoch": 5.166051660516605,
+      "grad_norm": 0.5865011215209961,
+      "learning_rate": 9.458610914145826e-05,
+      "loss": 0.555,
+      "step": 3150
+    },
+    {
+      "epoch": 5.182451824518245,
+      "grad_norm": 0.5646878480911255,
+      "learning_rate": 9.40710791528098e-05,
+      "loss": 0.5524,
+      "step": 3160
+    },
+    {
+      "epoch": 5.198851988519885,
+      "grad_norm": 0.6092641353607178,
+      "learning_rate": 9.355620693976461e-05,
+      "loss": 0.5704,
+      "step": 3170
+    },
+    {
+      "epoch": 5.215252152521526,
+      "grad_norm": 0.5585269927978516,
+      "learning_rate": 9.304150620368188e-05,
+      "loss": 0.5652,
+      "step": 3180
+    },
+    {
+      "epoch": 5.231652316523165,
+      "grad_norm": 0.5374350547790527,
+      "learning_rate": 9.252699064135758e-05,
+      "loss": 0.5591,
+      "step": 3190
+    },
+    {
+      "epoch": 5.248052480524805,
+      "grad_norm": 0.5892452001571655,
+      "learning_rate": 9.201267394465998e-05,
+      "loss": 0.568,
+      "step": 3200
+    },
+    {
+      "epoch": 5.264452644526445,
+      "grad_norm": 0.5627169609069824,
+      "learning_rate": 9.149856980016529e-05,
+      "loss": 0.5573,
+      "step": 3210
+    },
+    {
+      "epoch": 5.280852808528086,
+      "grad_norm": 0.6352980732917786,
+      "learning_rate": 9.098469188879349e-05,
+      "loss": 0.5548,
+      "step": 3220
+    },
+    {
+      "epoch": 5.2972529725297255,
+      "grad_norm": 0.5747233629226685,
+      "learning_rate": 9.047105388544417e-05,
+      "loss": 0.5802,
+      "step": 3230
+    },
+    {
+      "epoch": 5.313653136531365,
+      "grad_norm": 0.66510009765625,
+      "learning_rate": 8.995766945863277e-05,
+      "loss": 0.5583,
+      "step": 3240
+    },
+    {
+      "epoch": 5.330053300533005,
+      "grad_norm": 0.6103816628456116,
+      "learning_rate": 8.944455227012666e-05,
+      "loss": 0.5622,
+      "step": 3250
+    },
+    {
+      "epoch": 5.346453464534646,
+      "grad_norm": 0.53193598985672,
+      "learning_rate": 8.89317159745818e-05,
+      "loss": 0.5622,
+      "step": 3260
+    },
+    {
+      "epoch": 5.3628536285362856,
+      "grad_norm": 0.5971400737762451,
+      "learning_rate": 8.841917421917912e-05,
+      "loss": 0.5407,
+      "step": 3270
+    },
+    {
+      "epoch": 5.379253792537925,
+      "grad_norm": 0.5562443733215332,
+      "learning_rate": 8.790694064326157e-05,
+      "loss": 0.5544,
+      "step": 3280
+    },
+    {
+      "epoch": 5.395653956539565,
+      "grad_norm": 0.6092881560325623,
+      "learning_rate": 8.739502887797107e-05,
+      "loss": 0.5543,
+      "step": 3290
+    },
+    {
+      "epoch": 5.412054120541206,
+      "grad_norm": 0.5391745567321777,
+      "learning_rate": 8.688345254588578e-05,
+      "loss": 0.5553,
+      "step": 3300
+    },
+    {
+      "epoch": 5.428454284542846,
+      "grad_norm": 0.558480978012085,
+      "learning_rate": 8.637222526065756e-05,
+      "loss": 0.5614,
+      "step": 3310
+    },
+    {
+      "epoch": 5.444854448544485,
+      "grad_norm": 0.5935754776000977,
+      "learning_rate": 8.586136062664974e-05,
+      "loss": 0.5569,
+      "step": 3320
+    },
+    {
+      "epoch": 5.461254612546125,
+      "grad_norm": 0.5984821319580078,
+      "learning_rate": 8.535087223857508e-05,
+      "loss": 0.5625,
+      "step": 3330
+    },
+    {
+      "epoch": 5.477654776547766,
+      "grad_norm": 0.5659565925598145,
+      "learning_rate": 8.484077368113399e-05,
+      "loss": 0.5483,
+      "step": 3340
+    },
+    {
+      "epoch": 5.494054940549406,
+      "grad_norm": 0.5952328443527222,
+      "learning_rate": 8.433107852865298e-05,
+      "loss": 0.5546,
+      "step": 3350
+    },
+    {
+      "epoch": 5.5104551045510455,
+      "grad_norm": 0.5910452604293823,
+      "learning_rate": 8.382180034472353e-05,
+      "loss": 0.5371,
+      "step": 3360
+    },
+    {
+      "epoch": 5.526855268552685,
+      "grad_norm": 0.5623068809509277,
+      "learning_rate": 8.33129526818411e-05,
+      "loss": 0.5463,
+      "step": 3370
+    },
+    {
+      "epoch": 5.543255432554325,
+      "grad_norm": 0.557976484298706,
+      "learning_rate": 8.280454908104439e-05,
+      "loss": 0.555,
+      "step": 3380
+    },
+    {
+      "epoch": 5.559655596555966,
+      "grad_norm": 0.5730746984481812,
+      "learning_rate": 8.229660307155518e-05,
+      "loss": 0.5612,
+      "step": 3390
+    },
+    {
+      "epoch": 5.5760557605576055,
+      "grad_norm": 0.5673303604125977,
+      "learning_rate": 8.178912817041817e-05,
+      "loss": 0.5583,
+      "step": 3400
+    },
+    {
+      "epoch": 5.592455924559245,
+      "grad_norm": 0.5584287047386169,
+      "learning_rate": 8.128213788214126e-05,
+      "loss": 0.5471,
+      "step": 3410
+    },
+    {
+      "epoch": 5.608856088560886,
+      "grad_norm": 0.5875297784805298,
+      "learning_rate": 8.077564569833632e-05,
+      "loss": 0.5576,
+      "step": 3420
+    },
+    {
+      "epoch": 5.625256252562526,
+      "grad_norm": 0.5899044275283813,
+      "learning_rate": 8.026966509736001e-05,
+      "loss": 0.5437,
+      "step": 3430
+    },
+    {
+      "epoch": 5.641656416564166,
+      "grad_norm": 0.6080020666122437,
+      "learning_rate": 7.976420954395518e-05,
+      "loss": 0.5552,
+      "step": 3440
+    },
+    {
+      "epoch": 5.658056580565805,
+      "grad_norm": 0.594315230846405,
+      "learning_rate": 7.92592924888925e-05,
+      "loss": 0.5587,
+      "step": 3450
+    },
+    {
+      "epoch": 5.674456744567445,
+      "grad_norm": 0.5834687948226929,
+      "learning_rate": 7.875492736861266e-05,
+      "loss": 0.57,
+      "step": 3460
+    },
+    {
+      "epoch": 5.690856908569086,
+      "grad_norm": 0.5709730982780457,
+      "learning_rate": 7.825112760486861e-05,
+      "loss": 0.5737,
+      "step": 3470
+    },
+    {
+      "epoch": 5.707257072570726,
+      "grad_norm": 0.5856568217277527,
+      "learning_rate": 7.774790660436858e-05,
+      "loss": 0.5484,
+      "step": 3480
+    },
+    {
+      "epoch": 5.7236572365723655,
+      "grad_norm": 0.5535131692886353,
+      "learning_rate": 7.724527775841914e-05,
+      "loss": 0.5483,
+      "step": 3490
+    },
+    {
+      "epoch": 5.740057400574006,
+      "grad_norm": 0.5895147323608398,
+      "learning_rate": 7.674325444256899e-05,
+      "loss": 0.5463,
+      "step": 3500
+    },
+    {
+      "epoch": 5.756457564575646,
+      "grad_norm": 0.57289719581604,
+      "learning_rate": 7.624185001625292e-05,
+      "loss": 0.5569,
+      "step": 3510
+    },
+    {
+      "epoch": 5.772857728577286,
+      "grad_norm": 0.5502966046333313,
+      "learning_rate": 7.574107782243634e-05,
+      "loss": 0.5696,
+      "step": 3520
+    },
+    {
+      "epoch": 5.7892578925789255,
+      "grad_norm": 0.5804471969604492,
+      "learning_rate": 7.524095118726025e-05,
+      "loss": 0.5506,
+      "step": 3530
+    },
+    {
+      "epoch": 5.805658056580565,
+      "grad_norm": 0.5519583225250244,
+      "learning_rate": 7.474148341968652e-05,
+      "loss": 0.5571,
+      "step": 3540
+    },
+    {
+      "epoch": 5.822058220582206,
+      "grad_norm": 0.5806939005851746,
+      "learning_rate": 7.42426878111438e-05,
+      "loss": 0.5556,
+      "step": 3550
+    },
+    {
+      "epoch": 5.838458384583846,
+      "grad_norm": 0.5681191682815552,
+      "learning_rate": 7.374457763517376e-05,
+      "loss": 0.5557,
+      "step": 3560
+    },
+    {
+      "epoch": 5.854858548585486,
+      "grad_norm": 0.6004670262336731,
+      "learning_rate": 7.324716614707793e-05,
+      "loss": 0.5686,
+      "step": 3570
+    },
+    {
+      "epoch": 5.871258712587126,
+      "grad_norm": 0.5577644109725952,
+      "learning_rate": 7.275046658356494e-05,
+      "loss": 0.5619,
+      "step": 3580
+    },
+    {
+      "epoch": 5.887658876588766,
+      "grad_norm": 0.5934208631515503,
+      "learning_rate": 7.225449216239821e-05,
+      "loss": 0.5593,
+      "step": 3590
+    },
+    {
+      "epoch": 5.904059040590406,
+      "grad_norm": 0.5490043759346008,
+      "learning_rate": 7.175925608204428e-05,
+      "loss": 0.5404,
+      "step": 3600
+    },
+    {
+      "epoch": 5.920459204592046,
+      "grad_norm": 0.662482500076294,
+      "learning_rate": 7.126477152132164e-05,
+      "loss": 0.5525,
+      "step": 3610
+    },
+    {
+      "epoch": 5.9368593685936855,
+      "grad_norm": 0.5766283869743347,
+      "learning_rate": 7.077105163904987e-05,
+      "loss": 0.5706,
+      "step": 3620
+    },
+    {
+      "epoch": 5.953259532595326,
+      "grad_norm": 0.5751728415489197,
+      "learning_rate": 7.027810957369957e-05,
+      "loss": 0.5626,
+      "step": 3630
+    },
+    {
+      "epoch": 5.969659696596966,
+      "grad_norm": 0.5690221190452576,
+      "learning_rate": 6.978595844304271e-05,
+      "loss": 0.558,
+      "step": 3640
+    },
+    {
+      "epoch": 5.986059860598606,
+      "grad_norm": 0.5637174248695374,
+      "learning_rate": 6.92946113438036e-05,
+      "loss": 0.5665,
+      "step": 3650
+    },
+    {
+      "epoch": 6.002460024600246,
+      "grad_norm": 0.5749226808547974,
+      "learning_rate": 6.880408135131022e-05,
+      "loss": 0.5687,
+      "step": 3660
+    },
+    {
+      "epoch": 6.018860188601886,
+      "grad_norm": 0.581462562084198,
+      "learning_rate": 6.831438151914649e-05,
+      "loss": 0.5272,
+      "step": 3670
+    },
+    {
+      "epoch": 6.035260352603526,
+      "grad_norm": 0.585290253162384,
+      "learning_rate": 6.782552487880468e-05,
+      "loss": 0.5347,
+      "step": 3680
+    },
+    {
+      "epoch": 6.051660516605166,
+      "grad_norm": 0.6168552041053772,
+      "learning_rate": 6.733752443933878e-05,
+      "loss": 0.5224,
+      "step": 3690
+    },
+    {
+      "epoch": 6.0680606806068065,
+      "grad_norm": 0.5952621698379517,
+      "learning_rate": 6.685039318701826e-05,
+      "loss": 0.5267,
+      "step": 3700
+    },
+    {
+      "epoch": 6.084460844608446,
+      "grad_norm": 0.708752453327179,
+      "learning_rate": 6.636414408498249e-05,
+      "loss": 0.5323,
+      "step": 3710
+    },
+    {
+      "epoch": 6.100861008610086,
+      "grad_norm": 0.6072937250137329,
+      "learning_rate": 6.587879007289576e-05,
+      "loss": 0.5361,
+      "step": 3720
+    },
+    {
+      "epoch": 6.117261172611726,
+      "grad_norm": 0.6426242589950562,
+      "learning_rate": 6.539434406660296e-05,
+      "loss": 0.5439,
+      "step": 3730
+    },
+    {
+      "epoch": 6.1336613366133665,
+      "grad_norm": 0.6066299080848694,
+      "learning_rate": 6.491081895778588e-05,
+      "loss": 0.5256,
+      "step": 3740
+    },
+    {
+      "epoch": 6.150061500615006,
+      "grad_norm": 0.578899621963501,
+      "learning_rate": 6.442822761362015e-05,
+      "loss": 0.5373,
+      "step": 3750
+    },
+    {
+      "epoch": 6.166461664616646,
+      "grad_norm": 0.6230922341346741,
+      "learning_rate": 6.394658287643278e-05,
+      "loss": 0.5249,
+      "step": 3760
+    },
+    {
+      "epoch": 6.182861828618286,
+      "grad_norm": 0.610723614692688,
+      "learning_rate": 6.34658975633605e-05,
+      "loss": 0.5417,
+      "step": 3770
+    },
+    {
+      "epoch": 6.199261992619927,
+      "grad_norm": 0.6284223198890686,
+      "learning_rate": 6.298618446600856e-05,
+      "loss": 0.5361,
+      "step": 3780
+    },
+    {
+      "epoch": 6.215662156621566,
+      "grad_norm": 0.6320038437843323,
+      "learning_rate": 6.250745635011048e-05,
+      "loss": 0.5344,
+      "step": 3790
+    },
+    {
+      "epoch": 6.232062320623206,
+      "grad_norm": 0.5748194456100464,
+      "learning_rate": 6.202972595518817e-05,
+      "loss": 0.5285,
+      "step": 3800
+    },
+    {
+      "epoch": 6.248462484624846,
+      "grad_norm": 0.6720646023750305,
+      "learning_rate": 6.155300599421306e-05,
+      "loss": 0.5399,
+      "step": 3810
+    },
+    {
+      "epoch": 6.264862648626487,
+      "grad_norm": 0.6088182330131531,
+      "learning_rate": 6.107730915326772e-05,
+      "loss": 0.5387,
+      "step": 3820
+    },
+    {
+      "epoch": 6.2812628126281265,
+      "grad_norm": 0.6509405970573425,
+      "learning_rate": 6.0602648091208324e-05,
+      "loss": 0.5356,
+      "step": 3830
+    },
+    {
+      "epoch": 6.297662976629766,
+      "grad_norm": 0.6377142667770386,
+      "learning_rate": 6.012903543932766e-05,
+      "loss": 0.5328,
+      "step": 3840
+    },
+    {
+      "epoch": 6.314063140631406,
+      "grad_norm": 0.638660728931427,
+      "learning_rate": 5.965648380101916e-05,
+      "loss": 0.5403,
+      "step": 3850
+    },
+    {
+      "epoch": 6.330463304633047,
+      "grad_norm": 0.6343023180961609,
+      "learning_rate": 5.918500575144138e-05,
+      "loss": 0.5316,
+      "step": 3860
+    },
+    {
+      "epoch": 6.3468634686346865,
+      "grad_norm": 0.6120122075080872,
+      "learning_rate": 5.871461383718344e-05,
+      "loss": 0.5475,
+      "step": 3870
+    },
+    {
+      "epoch": 6.363263632636326,
+      "grad_norm": 0.664526104927063,
+      "learning_rate": 5.8245320575931085e-05,
+      "loss": 0.5539,
+      "step": 3880
+    },
+    {
+      "epoch": 6.379663796637966,
+      "grad_norm": 0.6195924878120422,
+      "learning_rate": 5.777713845613364e-05,
+      "loss": 0.5314,
+      "step": 3890
+    },
+    {
+      "epoch": 6.396063960639607,
+      "grad_norm": 0.6664077639579773,
+      "learning_rate": 5.7310079936671545e-05,
+      "loss": 0.5309,
+      "step": 3900
+    },
+    {
+      "epoch": 6.412464124641247,
+      "grad_norm": 0.6408997178077698,
+      "learning_rate": 5.684415744652509e-05,
+      "loss": 0.5346,
+      "step": 3910
+    },
+    {
+      "epoch": 6.428864288642886,
+      "grad_norm": 0.596552312374115,
+      "learning_rate": 5.6379383384443255e-05,
+      "loss": 0.5424,
+      "step": 3920
+    },
+    {
+      "epoch": 6.445264452644526,
+      "grad_norm": 0.6852882504463196,
+      "learning_rate": 5.59157701186142e-05,
+      "loss": 0.5367,
+      "step": 3930
+    },
+    {
+      "epoch": 6.461664616646167,
+      "grad_norm": 0.6832631826400757,
+      "learning_rate": 5.545332998633572e-05,
+      "loss": 0.5311,
+      "step": 3940
+    },
+    {
+      "epoch": 6.478064780647807,
+      "grad_norm": 0.6201120018959045,
+      "learning_rate": 5.499207529368734e-05,
+      "loss": 0.5398,
+      "step": 3950
+    },
+    {
+      "epoch": 6.4944649446494465,
+      "grad_norm": 0.6140998601913452,
+      "learning_rate": 5.453201831520245e-05,
+      "loss": 0.5388,
+      "step": 3960
+    },
+    {
+      "epoch": 6.510865108651086,
+      "grad_norm": 0.6650799512863159,
+      "learning_rate": 5.4073171293542016e-05,
+      "loss": 0.5419,
+      "step": 3970
+    },
+    {
+      "epoch": 6.527265272652727,
+      "grad_norm": 0.6716285347938538,
+      "learning_rate": 5.3615546439168485e-05,
+      "loss": 0.5234,
+      "step": 3980
+    },
+    {
+      "epoch": 6.543665436654367,
+      "grad_norm": 0.6595677733421326,
+      "learning_rate": 5.3159155930021e-05,
+      "loss": 0.5194,
+      "step": 3990
+    },
+    {
+      "epoch": 6.5600656006560065,
+      "grad_norm": 0.5784692168235779,
+      "learning_rate": 5.270401191119143e-05,
+      "loss": 0.5222,
+      "step": 4000
+    },
+    {
+      "epoch": 6.576465764657646,
+      "grad_norm": 0.6419820189476013,
+      "learning_rate": 5.2250126494600916e-05,
+      "loss": 0.5352,
+      "step": 4010
+    },
+    {
+      "epoch": 6.592865928659286,
+      "grad_norm": 0.6142451763153076,
+      "learning_rate": 5.179751175867784e-05,
+      "loss": 0.5357,
+      "step": 4020
+    },
+    {
+      "epoch": 6.609266092660927,
+      "grad_norm": 0.6860802173614502,
+      "learning_rate": 5.1346179748036116e-05,
+      "loss": 0.531,
+      "step": 4030
+    },
+    {
+      "epoch": 6.625666256662567,
+      "grad_norm": 0.6106059551239014,
+      "learning_rate": 5.0896142473154987e-05,
+      "loss": 0.5333,
+      "step": 4040
+    },
+    {
+      "epoch": 6.642066420664206,
+      "grad_norm": 0.6644913554191589,
+      "learning_rate": 5.044741191005908e-05,
+      "loss": 0.5403,
+      "step": 4050
+    },
+    {
+      "epoch": 6.658466584665847,
+      "grad_norm": 0.5942371487617493,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.5467,
+      "step": 4060
+    },
+    {
+      "epoch": 6.674866748667487,
+      "grad_norm": 0.617970883846283,
+      "learning_rate": 4.9553918649138386e-05,
+      "loss": 0.5325,
+      "step": 4070
+    },
+    {
+      "epoch": 6.691266912669127,
+      "grad_norm": 0.6382017135620117,
+      "learning_rate": 4.910917972822713e-05,
+      "loss": 0.525,
+      "step": 4080
+    },
+    {
+      "epoch": 6.7076670766707664,
+      "grad_norm": 0.6390166282653809,
+      "learning_rate": 4.866579507229545e-05,
+      "loss": 0.5356,
+      "step": 4090
+    },
+    {
+      "epoch": 6.724067240672406,
+      "grad_norm": 0.6630433201789856,
+      "learning_rate": 4.822377648033394e-05,
+      "loss": 0.541,
+      "step": 4100
+    },
+    {
+      "epoch": 6.740467404674047,
+      "grad_norm": 0.6468001008033752,
+      "learning_rate": 4.7783135714980744e-05,
+      "loss": 0.5399,
+      "step": 4110
+    },
+    {
+      "epoch": 6.756867568675687,
+      "grad_norm": 0.620833694934845,
+      "learning_rate": 4.734388450220825e-05,
+      "loss": 0.5265,
+      "step": 4120
+    },
+    {
+      "epoch": 6.7732677326773265,
+      "grad_norm": 0.61692875623703,
+      "learning_rate": 4.6906034531011346e-05,
+      "loss": 0.5303,
+      "step": 4130
+    },
+    {
+      "epoch": 6.789667896678967,
+      "grad_norm": 0.6430515646934509,
+      "learning_rate": 4.646959745309609e-05,
+      "loss": 0.543,
+      "step": 4140
+    },
+    {
+      "epoch": 6.806068060680607,
+      "grad_norm": 0.653596043586731,
+      "learning_rate": 4.603458488256992e-05,
+      "loss": 0.5384,
+      "step": 4150
+    },
+    {
+      "epoch": 6.822468224682247,
+      "grad_norm": 0.6125525832176208,
+      "learning_rate": 4.560100839563229e-05,
+      "loss": 0.5417,
+      "step": 4160
+    },
+    {
+      "epoch": 6.838868388683887,
+      "grad_norm": 0.5968248844146729,
+      "learning_rate": 4.516887953026691e-05,
+      "loss": 0.5326,
+      "step": 4170
+    },
+    {
+      "epoch": 6.855268552685526,
+      "grad_norm": 0.5980390906333923,
+      "learning_rate": 4.4738209785934505e-05,
+      "loss": 0.5313,
+      "step": 4180
+    },
+    {
+      "epoch": 6.871668716687167,
+      "grad_norm": 0.614548921585083,
+      "learning_rate": 4.430901062326681e-05,
+      "loss": 0.5321,
+      "step": 4190
+    },
+    {
+      "epoch": 6.888068880688807,
+      "grad_norm": 0.5848116278648376,
+      "learning_rate": 4.388129346376178e-05,
+      "loss": 0.542,
+      "step": 4200
+    },
+    {
+      "epoch": 6.904469044690447,
+      "grad_norm": 0.6172247529029846,
+      "learning_rate": 4.345506968947931e-05,
+      "loss": 0.5439,
+      "step": 4210
+    },
+    {
+      "epoch": 6.920869208692087,
+      "grad_norm": 0.6525529026985168,
+      "learning_rate": 4.303035064273878e-05,
+      "loss": 0.5319,
+      "step": 4220
+    },
+    {
+      "epoch": 6.937269372693727,
+      "grad_norm": 0.614452064037323,
+      "learning_rate": 4.260714762581677e-05,
+      "loss": 0.5287,
+      "step": 4230
+    },
+    {
+      "epoch": 6.953669536695367,
+      "grad_norm": 0.6268100142478943,
+      "learning_rate": 4.21854719006467e-05,
+      "loss": 0.5337,
+      "step": 4240
+    },
+    {
+      "epoch": 6.970069700697007,
+      "grad_norm": 0.6285504698753357,
+      "learning_rate": 4.1765334688518766e-05,
+      "loss": 0.5402,
+      "step": 4250
+    },
+    {
+      "epoch": 6.9864698646986465,
+      "grad_norm": 0.6280767917633057,
+      "learning_rate": 4.13467471697817e-05,
+      "loss": 0.533,
+      "step": 4260
+    },
+    {
+      "epoch": 7.002870028700287,
+      "grad_norm": 0.5814547538757324,
+      "learning_rate": 4.092972048354491e-05,
+      "loss": 0.5346,
+      "step": 4270
+    },
+    {
+      "epoch": 7.019270192701927,
+      "grad_norm": 0.6583351492881775,
+      "learning_rate": 4.0514265727382215e-05,
+      "loss": 0.5261,
+      "step": 4280
+    },
+    {
+      "epoch": 7.035670356703567,
+      "grad_norm": 0.6630005240440369,
+      "learning_rate": 4.010039395703664e-05,
+      "loss": 0.5196,
+      "step": 4290
+    },
+    {
+      "epoch": 7.0520705207052075,
+      "grad_norm": 0.6557034254074097,
+      "learning_rate": 3.968811618612592e-05,
+      "loss": 0.5161,
+      "step": 4300
+    },
+    {
+      "epoch": 7.068470684706847,
+      "grad_norm": 0.6600791215896606,
+      "learning_rate": 3.927744338584972e-05,
+      "loss": 0.5066,
+      "step": 4310
+    },
+    {
+      "epoch": 7.084870848708487,
+      "grad_norm": 0.6123402118682861,
+      "learning_rate": 3.8868386484697417e-05,
+      "loss": 0.5195,
+      "step": 4320
+    },
+    {
+      "epoch": 7.101271012710127,
+      "grad_norm": 0.6454523801803589,
+      "learning_rate": 3.84609563681575e-05,
+      "loss": 0.5113,
+      "step": 4330
+    },
+    {
+      "epoch": 7.1176711767117675,
+      "grad_norm": 0.6421491503715515,
+      "learning_rate": 3.80551638784277e-05,
+      "loss": 0.5134,
+      "step": 4340
+    },
+    {
+      "epoch": 7.134071340713407,
+      "grad_norm": 0.6418097615242004,
+      "learning_rate": 3.7651019814126654e-05,
+      "loss": 0.5185,
+      "step": 4350
+    },
+    {
+      "epoch": 7.150471504715047,
+      "grad_norm": 0.6425509452819824,
+      "learning_rate": 3.724853493000635e-05,
+      "loss": 0.5251,
+      "step": 4360
+    },
+    {
+      "epoch": 7.166871668716687,
+      "grad_norm": 0.6572188138961792,
+      "learning_rate": 3.6847719936666124e-05,
+      "loss": 0.5124,
+      "step": 4370
+    },
+    {
+      "epoch": 7.183271832718328,
+      "grad_norm": 0.6694904565811157,
+      "learning_rate": 3.6448585500267485e-05,
+      "loss": 0.5174,
+      "step": 4380
+    },
+    {
+      "epoch": 7.199671996719967,
+      "grad_norm": 0.6733431816101074,
+      "learning_rate": 3.605114224225028e-05,
+      "loss": 0.5227,
+      "step": 4390
+    },
+    {
+      "epoch": 7.216072160721607,
+      "grad_norm": 0.6532800793647766,
+      "learning_rate": 3.565540073905025e-05,
+      "loss": 0.5128,
+      "step": 4400
+    },
+    {
+      "epoch": 7.232472324723247,
+      "grad_norm": 0.6475611329078674,
+      "learning_rate": 3.5261371521817244e-05,
+      "loss": 0.5282,
+      "step": 4410
+    },
+    {
+      "epoch": 7.248872488724888,
+      "grad_norm": 0.6134998202323914,
+      "learning_rate": 3.486906507613531e-05,
+      "loss": 0.5118,
+      "step": 4420
+    },
+    {
+      "epoch": 7.2652726527265274,
+      "grad_norm": 0.648802638053894,
+      "learning_rate": 3.4478491841743397e-05,
+      "loss": 0.5253,
+      "step": 4430
+    },
+    {
+      "epoch": 7.281672816728167,
+      "grad_norm": 0.6617629528045654,
+      "learning_rate": 3.408966221225773e-05,
+      "loss": 0.5177,
+      "step": 4440
+    },
+    {
+      "epoch": 7.298072980729807,
+      "grad_norm": 0.6465151906013489,
+      "learning_rate": 3.370258653489505e-05,
+      "loss": 0.5133,
+      "step": 4450
+    },
+    {
+      "epoch": 7.314473144731448,
+      "grad_norm": 0.6327465176582336,
+      "learning_rate": 3.331727511019749e-05,
+      "loss": 0.5155,
+      "step": 4460
+    },
+    {
+      "epoch": 7.3308733087330875,
+      "grad_norm": 0.6604763865470886,
+      "learning_rate": 3.293373819175816e-05,
+      "loss": 0.519,
+      "step": 4470
+    },
+    {
+      "epoch": 7.347273472734727,
+      "grad_norm": 0.681736171245575,
+      "learning_rate": 3.2551985985948616e-05,
+      "loss": 0.5243,
+      "step": 4480
+    },
+    {
+      "epoch": 7.363673636736367,
+      "grad_norm": 0.6531623601913452,
+      "learning_rate": 3.217202865164697e-05,
+      "loss": 0.5201,
+      "step": 4490
+    },
+    {
+      "epoch": 7.380073800738008,
+      "grad_norm": 0.674201250076294,
+      "learning_rate": 3.1793876299967816e-05,
+      "loss": 0.5252,
+      "step": 4500
+    },
+    {
+      "epoch": 7.396473964739648,
+      "grad_norm": 0.6714997291564941,
+      "learning_rate": 3.141753899399289e-05,
+      "loss": 0.5253,
+      "step": 4510
+    },
+    {
+      "epoch": 7.412874128741287,
+      "grad_norm": 0.6872825622558594,
+      "learning_rate": 3.104302674850346e-05,
+      "loss": 0.5311,
+      "step": 4520
+    },
+    {
+      "epoch": 7.429274292742927,
+      "grad_norm": 0.7338685989379883,
+      "learning_rate": 3.0670349529713816e-05,
+      "loss": 0.515,
+      "step": 4530
+    },
+    {
+      "epoch": 7.445674456744568,
+      "grad_norm": 0.6552906036376953,
+      "learning_rate": 3.0299517255005937e-05,
+      "loss": 0.5077,
+      "step": 4540
+    },
+    {
+      "epoch": 7.462074620746208,
+      "grad_norm": 0.6388695240020752,
+      "learning_rate": 2.993053979266577e-05,
+      "loss": 0.5296,
+      "step": 4550
+    },
+    {
+      "epoch": 7.478474784747847,
+      "grad_norm": 0.6324784159660339,
+      "learning_rate": 2.9563426961620367e-05,
+      "loss": 0.5249,
+      "step": 4560
+    },
+    {
+      "epoch": 7.494874948749487,
+      "grad_norm": 0.6526088714599609,
+      "learning_rate": 2.9198188531176863e-05,
+      "loss": 0.5164,
+      "step": 4570
+    },
+    {
+      "epoch": 7.511275112751127,
+      "grad_norm": 0.6632060408592224,
+      "learning_rate": 2.883483422076225e-05,
+      "loss": 0.5185,
+      "step": 4580
+    },
+    {
+      "epoch": 7.527675276752768,
+      "grad_norm": 0.643341600894928,
+      "learning_rate": 2.8473373699664997e-05,
+      "loss": 0.5162,
+      "step": 4590
+    },
+    {
+      "epoch": 7.5440754407544075,
+      "grad_norm": 0.6466885805130005,
+      "learning_rate": 2.811381658677744e-05,
+      "loss": 0.5103,
+      "step": 4600
+    },
+    {
+      "epoch": 7.560475604756047,
+      "grad_norm": 0.6679097414016724,
+      "learning_rate": 2.7756172450340134e-05,
+      "loss": 0.5238,
+      "step": 4610
+    },
+    {
+      "epoch": 7.576875768757688,
+      "grad_norm": 0.6607965230941772,
+      "learning_rate": 2.7400450807686938e-05,
+      "loss": 0.5104,
+      "step": 4620
+    },
+    {
+      "epoch": 7.593275932759328,
+      "grad_norm": 0.64109867811203,
+      "learning_rate": 2.70466611249919e-05,
+      "loss": 0.5199,
+      "step": 4630
+    },
+    {
+      "epoch": 7.609676096760968,
+      "grad_norm": 0.6699521541595459,
+      "learning_rate": 2.669481281701739e-05,
+      "loss": 0.5339,
+      "step": 4640
+    },
+    {
+      "epoch": 7.626076260762607,
+      "grad_norm": 0.6698071360588074,
+      "learning_rate": 2.6344915246863412e-05,
+      "loss": 0.5128,
+      "step": 4650
+    },
+    {
+      "epoch": 7.642476424764247,
+      "grad_norm": 0.6424985527992249,
+      "learning_rate": 2.5996977725718607e-05,
+      "loss": 0.5053,
+      "step": 4660
+    },
+    {
+      "epoch": 7.658876588765888,
+      "grad_norm": 0.7101566195487976,
+      "learning_rate": 2.5651009512612312e-05,
+      "loss": 0.5311,
+      "step": 4670
+    },
+    {
+      "epoch": 7.675276752767528,
+      "grad_norm": 0.6432877779006958,
+      "learning_rate": 2.5307019814168342e-05,
+      "loss": 0.5216,
+      "step": 4680
+    },
+    {
+      "epoch": 7.691676916769167,
+      "grad_norm": 0.6850785613059998,
+      "learning_rate": 2.496501778435977e-05,
+      "loss": 0.5108,
+      "step": 4690
+    },
+    {
+      "epoch": 7.708077080770808,
+      "grad_norm": 0.6734909415245056,
+      "learning_rate": 2.462501252426559e-05,
+      "loss": 0.5186,
+      "step": 4700
+    },
+    {
+      "epoch": 7.724477244772448,
+      "grad_norm": 0.7187743782997131,
+      "learning_rate": 2.4287013081828257e-05,
+      "loss": 0.5182,
+      "step": 4710
+    },
+    {
+      "epoch": 7.740877408774088,
+      "grad_norm": 0.7147987484931946,
+      "learning_rate": 2.3951028451613144e-05,
+      "loss": 0.5193,
+      "step": 4720
+    },
+    {
+      "epoch": 7.7572775727757275,
+      "grad_norm": 0.7066251635551453,
+      "learning_rate": 2.3617067574569087e-05,
+      "loss": 0.5175,
+      "step": 4730
+    },
+    {
+      "epoch": 7.773677736777367,
+      "grad_norm": 0.7206938862800598,
+      "learning_rate": 2.328513933779034e-05,
+      "loss": 0.5255,
+      "step": 4740
+    },
+    {
+      "epoch": 7.790077900779008,
+      "grad_norm": 0.6903772354125977,
+      "learning_rate": 2.2955252574280328e-05,
+      "loss": 0.512,
+      "step": 4750
+    },
+    {
+      "epoch": 7.806478064780648,
+      "grad_norm": 0.6938795447349548,
+      "learning_rate": 2.2627416062716366e-05,
+      "loss": 0.513,
+      "step": 4760
+    },
+    {
+      "epoch": 7.822878228782288,
+      "grad_norm": 0.6563030481338501,
+      "learning_rate": 2.2301638527216194e-05,
+      "loss": 0.5255,
+      "step": 4770
+    },
+    {
+      "epoch": 7.839278392783928,
+      "grad_norm": 0.676189124584198,
+      "learning_rate": 2.1977928637105692e-05,
+      "loss": 0.5331,
+      "step": 4780
+    },
+    {
+      "epoch": 7.855678556785568,
+      "grad_norm": 0.6740292906761169,
+      "learning_rate": 2.1656295006688353e-05,
+      "loss": 0.5161,
+      "step": 4790
+    },
+    {
+      "epoch": 7.872078720787208,
+      "grad_norm": 0.6368678212165833,
+      "learning_rate": 2.1336746195015846e-05,
+      "loss": 0.5181,
+      "step": 4800
+    },
+    {
+      "epoch": 7.888478884788848,
+      "grad_norm": 0.6666802763938904,
+      "learning_rate": 2.1019290705660356e-05,
+      "loss": 0.5248,
+      "step": 4810
+    },
+    {
+      "epoch": 7.904879048790487,
+      "grad_norm": 0.7021058797836304,
+      "learning_rate": 2.070393698648836e-05,
+      "loss": 0.5076,
+      "step": 4820
+    },
+    {
+      "epoch": 7.921279212792128,
+      "grad_norm": 0.6414440870285034,
+      "learning_rate": 2.0390693429435627e-05,
+      "loss": 0.5091,
+      "step": 4830
+    },
+    {
+      "epoch": 7.937679376793768,
+      "grad_norm": 0.6494581699371338,
+      "learning_rate": 2.0079568370284128e-05,
+      "loss": 0.5205,
+      "step": 4840
+    },
+    {
+      "epoch": 7.954079540795408,
+      "grad_norm": 0.6568921208381653,
+      "learning_rate": 1.977057008844e-05,
+      "loss": 0.5161,
+      "step": 4850
+    },
+    {
+      "epoch": 7.970479704797048,
+      "grad_norm": 0.6623067855834961,
+      "learning_rate": 1.946370680671341e-05,
+      "loss": 0.5134,
+      "step": 4860
+    },
+    {
+      "epoch": 7.986879868798688,
+      "grad_norm": 0.6465177536010742,
+      "learning_rate": 1.9158986691099544e-05,
+      "loss": 0.5097,
+      "step": 4870
+    },
+    {
+      "epoch": 8.003280032800328,
+      "grad_norm": 0.6603899002075195,
+      "learning_rate": 1.885641785056149e-05,
+      "loss": 0.5152,
+      "step": 4880
+    },
+    {
+      "epoch": 8.019680196801968,
+      "grad_norm": 0.6793326735496521,
+      "learning_rate": 1.85560083368143e-05,
+      "loss": 0.5201,
+      "step": 4890
+    },
+    {
+      "epoch": 8.036080360803608,
+      "grad_norm": 0.6705737709999084,
+      "learning_rate": 1.825776614411082e-05,
+      "loss": 0.4945,
+      "step": 4900
+    },
+    {
+      "epoch": 8.052480524805247,
+      "grad_norm": 0.6642212271690369,
+      "learning_rate": 1.7961699209028905e-05,
+      "loss": 0.5019,
+      "step": 4910
+    },
+    {
+      "epoch": 8.068880688806889,
+      "grad_norm": 0.6646486520767212,
+      "learning_rate": 1.766781541026018e-05,
+      "loss": 0.5245,
+      "step": 4920
+    },
+    {
+      "epoch": 8.085280852808529,
+      "grad_norm": 0.6514326930046082,
+      "learning_rate": 1.7376122568400532e-05,
+      "loss": 0.5083,
+      "step": 4930
+    },
+    {
+      "epoch": 8.101681016810169,
+      "grad_norm": 0.6762531995773315,
+      "learning_rate": 1.708662844574178e-05,
+      "loss": 0.5144,
+      "step": 4940
+    },
+    {
+      "epoch": 8.118081180811808,
+      "grad_norm": 0.6511978507041931,
+      "learning_rate": 1.679934074606533e-05,
+      "loss": 0.5045,
+      "step": 4950
+    },
+    {
+      "epoch": 8.134481344813448,
+      "grad_norm": 0.6896832585334778,
+      "learning_rate": 1.6514267114436945e-05,
+      "loss": 0.5089,
+      "step": 4960
+    },
+    {
+      "epoch": 8.150881508815088,
+      "grad_norm": 0.6686916947364807,
+      "learning_rate": 1.6231415137003537e-05,
+      "loss": 0.5048,
+      "step": 4970
+    },
+    {
+      "epoch": 8.167281672816728,
+      "grad_norm": 0.6836830377578735,
+      "learning_rate": 1.5950792340791043e-05,
+      "loss": 0.5072,
+      "step": 4980
+    },
+    {
+      "epoch": 8.183681836818367,
+      "grad_norm": 0.6605989336967468,
+      "learning_rate": 1.5672406193504384e-05,
+      "loss": 0.5115,
+      "step": 4990
+    },
+    {
+      "epoch": 8.200082000820009,
+      "grad_norm": 0.6973133087158203,
+      "learning_rate": 1.5396264103328474e-05,
+      "loss": 0.5132,
+      "step": 5000
+    },
+    {
+      "epoch": 8.216482164821649,
+      "grad_norm": 0.7265208959579468,
+      "learning_rate": 1.5122373418731306e-05,
+      "loss": 0.5036,
+      "step": 5010
+    },
+    {
+      "epoch": 8.232882328823289,
+      "grad_norm": 0.6954894661903381,
+      "learning_rate": 1.4850741428268244e-05,
+      "loss": 0.5037,
+      "step": 5020
+    },
+    {
+      "epoch": 8.249282492824928,
+      "grad_norm": 0.655786395072937,
+      "learning_rate": 1.4581375360388183e-05,
+      "loss": 0.515,
+      "step": 5030
+    },
+    {
+      "epoch": 8.265682656826568,
+      "grad_norm": 0.674171507358551,
+      "learning_rate": 1.4314282383241096e-05,
+      "loss": 0.4967,
+      "step": 5040
+    },
+    {
+      "epoch": 8.282082820828208,
+      "grad_norm": 0.7670585513114929,
+      "learning_rate": 1.4049469604487297e-05,
+      "loss": 0.4991,
+      "step": 5050
+    },
+    {
+      "epoch": 8.298482984829848,
+      "grad_norm": 0.6614052057266235,
+      "learning_rate": 1.3786944071108398e-05,
+      "loss": 0.5169,
+      "step": 5060
+    },
+    {
+      "epoch": 8.314883148831488,
+      "grad_norm": 0.7165561318397522,
+      "learning_rate": 1.3526712769219618e-05,
+      "loss": 0.502,
+      "step": 5070
+    },
+    {
+      "epoch": 8.33128331283313,
+      "grad_norm": 0.7127593159675598,
+      "learning_rate": 1.3268782623884047e-05,
+      "loss": 0.5091,
+      "step": 5080
+    },
+    {
+      "epoch": 8.347683476834769,
+      "grad_norm": 0.6743724346160889,
+      "learning_rate": 1.301316049892818e-05,
+      "loss": 0.5019,
+      "step": 5090
+    },
+    {
+      "epoch": 8.364083640836409,
+      "grad_norm": 0.6609333157539368,
+      "learning_rate": 1.2759853196759453e-05,
+      "loss": 0.5052,
+      "step": 5100
+    },
+    {
+      "epoch": 8.380483804838049,
+      "grad_norm": 0.7156481146812439,
+      "learning_rate": 1.2508867458185037e-05,
+      "loss": 0.51,
+      "step": 5110
+    },
+    {
+      "epoch": 8.396883968839688,
+      "grad_norm": 0.7323744893074036,
+      "learning_rate": 1.2260209962232628e-05,
+      "loss": 0.5092,
+      "step": 5120
+    },
+    {
+      "epoch": 8.413284132841328,
+      "grad_norm": 0.6540841460227966,
+      "learning_rate": 1.201388732597255e-05,
+      "loss": 0.507,
+      "step": 5130
+    },
+    {
+      "epoch": 8.429684296842968,
+      "grad_norm": 0.6970362663269043,
+      "learning_rate": 1.1769906104341832e-05,
+      "loss": 0.5216,
+      "step": 5140
+    },
+    {
+      "epoch": 8.446084460844608,
+      "grad_norm": 0.6555099487304688,
+      "learning_rate": 1.1528272789969618e-05,
+      "loss": 0.5001,
+      "step": 5150
+    },
+    {
+      "epoch": 8.46248462484625,
+      "grad_norm": 0.6940327286720276,
+      "learning_rate": 1.1288993813004467e-05,
+      "loss": 0.5025,
+      "step": 5160
+    },
+    {
+      "epoch": 8.478884788847889,
+      "grad_norm": 0.6825557351112366,
+      "learning_rate": 1.1052075540943296e-05,
+      "loss": 0.5089,
+      "step": 5170
+    },
+    {
+      "epoch": 8.495284952849529,
+      "grad_norm": 0.6553847789764404,
+      "learning_rate": 1.0817524278461776e-05,
+      "loss": 0.5052,
+      "step": 5180
+    },
+    {
+      "epoch": 8.511685116851169,
+      "grad_norm": 0.6892216205596924,
+      "learning_rate": 1.0585346267246743e-05,
+      "loss": 0.5158,
+      "step": 5190
+    },
+    {
+      "epoch": 8.528085280852808,
+      "grad_norm": 0.6756864786148071,
+      "learning_rate": 1.0355547685829926e-05,
+      "loss": 0.5133,
+      "step": 5200
+    },
+    {
+      "epoch": 8.544485444854448,
+      "grad_norm": 0.7741680145263672,
+      "learning_rate": 1.0128134649423671e-05,
+      "loss": 0.5167,
+      "step": 5210
+    },
+    {
+      "epoch": 8.560885608856088,
+      "grad_norm": 0.6552737355232239,
+      "learning_rate": 9.903113209758096e-06,
+      "loss": 0.5094,
+      "step": 5220
+    },
+    {
+      "epoch": 8.577285772857728,
+      "grad_norm": 0.6564160585403442,
+      "learning_rate": 9.680489354920152e-06,
+      "loss": 0.5048,
+      "step": 5230
+    },
+    {
+      "epoch": 8.59368593685937,
+      "grad_norm": 0.7316927313804626,
+      "learning_rate": 9.460269009194167e-06,
+      "loss": 0.499,
+      "step": 5240
+    },
+    {
+      "epoch": 8.61008610086101,
+      "grad_norm": 0.7389256358146667,
+      "learning_rate": 9.242458032904311e-06,
+      "loss": 0.5022,
+      "step": 5250
+    },
+    {
+      "epoch": 8.626486264862649,
+      "grad_norm": 0.6633957624435425,
+      "learning_rate": 9.027062222258487e-06,
+      "loss": 0.5042,
+      "step": 5260
+    },
+    {
+      "epoch": 8.642886428864289,
+      "grad_norm": 0.640006422996521,
+      "learning_rate": 8.814087309194251e-06,
+      "loss": 0.5068,
+      "step": 5270
+    },
+    {
+      "epoch": 8.659286592865929,
+      "grad_norm": 0.6830674409866333,
+      "learning_rate": 8.603538961226232e-06,
+      "loss": 0.502,
+      "step": 5280
+    },
+    {
+      "epoch": 8.675686756867568,
+      "grad_norm": 0.6639739871025085,
+      "learning_rate": 8.395422781295192e-06,
+      "loss": 0.5177,
+      "step": 5290
+    },
+    {
+      "epoch": 8.692086920869208,
+      "grad_norm": 0.6661989092826843,
+      "learning_rate": 8.189744307619118e-06,
+      "loss": 0.5014,
+      "step": 5300
+    },
+    {
+      "epoch": 8.708487084870848,
+      "grad_norm": 0.6645560264587402,
+      "learning_rate": 7.986509013545673e-06,
+      "loss": 0.5017,
+      "step": 5310
+    },
+    {
+      "epoch": 8.724887248872488,
+      "grad_norm": 0.7042533159255981,
+      "learning_rate": 7.785722307406684e-06,
+      "loss": 0.5163,
+      "step": 5320
+    },
+    {
+      "epoch": 8.74128741287413,
+      "grad_norm": 0.698277473449707,
+      "learning_rate": 7.587389532374123e-06,
+      "loss": 0.5054,
+      "step": 5330
+    },
+    {
+      "epoch": 8.75768757687577,
+      "grad_norm": 0.6191078424453735,
+      "learning_rate": 7.3915159663179075e-06,
+      "loss": 0.4992,
+      "step": 5340
+    },
+    {
+      "epoch": 8.774087740877409,
+      "grad_norm": 0.6769982576370239,
+      "learning_rate": 7.198106821665585e-06,
+      "loss": 0.5147,
+      "step": 5350
+    },
+    {
+      "epoch": 8.790487904879049,
+      "grad_norm": 0.6643761396408081,
+      "learning_rate": 7.007167245263435e-06,
+      "loss": 0.5105,
+      "step": 5360
+    },
+    {
+      "epoch": 8.806888068880689,
+      "grad_norm": 0.6664229035377502,
+      "learning_rate": 6.818702318239689e-06,
+      "loss": 0.5021,
+      "step": 5370
+    },
+    {
+      "epoch": 8.823288232882328,
+      "grad_norm": 0.6411841511726379,
+      "learning_rate": 6.632717055869164e-06,
+      "loss": 0.5076,
+      "step": 5380
+    },
+    {
+      "epoch": 8.839688396883968,
+      "grad_norm": 0.6643224954605103,
+      "learning_rate": 6.4492164074399065e-06,
+      "loss": 0.5044,
+      "step": 5390
+    },
+    {
+      "epoch": 8.85608856088561,
+      "grad_norm": 0.6722341775894165,
+      "learning_rate": 6.268205256121396e-06,
+      "loss": 0.5092,
+      "step": 5400
+    },
+    {
+      "epoch": 8.87248872488725,
+      "grad_norm": 0.6829484105110168,
+      "learning_rate": 6.089688418834727e-06,
+      "loss": 0.5164,
+      "step": 5410
+    },
+    {
+      "epoch": 8.88888888888889,
+      "grad_norm": 0.7037432789802551,
+      "learning_rate": 5.913670646124236e-06,
+      "loss": 0.5142,
+      "step": 5420
+    },
+    {
+      "epoch": 8.905289052890529,
+      "grad_norm": 0.6822744607925415,
+      "learning_rate": 5.7401566220313005e-06,
+      "loss": 0.5018,
+      "step": 5430
+    },
+    {
+      "epoch": 8.921689216892169,
+      "grad_norm": 0.6738746166229248,
+      "learning_rate": 5.569150963969494e-06,
+      "loss": 0.5013,
+      "step": 5440
+    },
+    {
+      "epoch": 8.938089380893809,
+      "grad_norm": 0.6965447664260864,
+      "learning_rate": 5.400658222601873e-06,
+      "loss": 0.5014,
+      "step": 5450
+    },
+    {
+      "epoch": 8.954489544895448,
+      "grad_norm": 0.7140269875526428,
+      "learning_rate": 5.2346828817197655e-06,
+      "loss": 0.5119,
+      "step": 5460
+    },
+    {
+      "epoch": 8.970889708897088,
+      "grad_norm": 0.6803216934204102,
+      "learning_rate": 5.071229358123464e-06,
+      "loss": 0.5021,
+      "step": 5470
+    },
+    {
+      "epoch": 8.987289872898728,
+      "grad_norm": 0.6827109456062317,
+      "learning_rate": 4.910302001504807e-06,
+      "loss": 0.5157,
+      "step": 5480
+    },
+    {
+      "epoch": 9.00369003690037,
+      "grad_norm": 0.6659403443336487,
+      "learning_rate": 4.7519050943312325e-06,
+      "loss": 0.4975,
+      "step": 5490
+    },
+    {
+      "epoch": 9.02009020090201,
+      "grad_norm": 0.7039642333984375,
+      "learning_rate": 4.596042851732008e-06,
+      "loss": 0.4949,
+      "step": 5500
+    },
+    {
+      "epoch": 9.03649036490365,
+      "grad_norm": 0.6756150722503662,
+      "learning_rate": 4.442719421385922e-06,
+      "loss": 0.4925,
+      "step": 5510
+    },
+    {
+      "epoch": 9.052890528905289,
+      "grad_norm": 0.684483528137207,
+      "learning_rate": 4.291938883411007e-06,
+      "loss": 0.4997,
+      "step": 5520
+    },
+    {
+      "epoch": 9.069290692906929,
+      "grad_norm": 0.6780720353126526,
+      "learning_rate": 4.143705250255869e-06,
+      "loss": 0.5104,
+      "step": 5530
+    },
+    {
+      "epoch": 9.085690856908569,
+      "grad_norm": 0.6675730347633362,
+      "learning_rate": 3.99802246659301e-06,
+      "loss": 0.5181,
+      "step": 5540
+    },
+    {
+      "epoch": 9.102091020910208,
+      "grad_norm": 0.6699404120445251,
+      "learning_rate": 3.85489440921376e-06,
+      "loss": 0.506,
+      "step": 5550
+    },
+    {
+      "epoch": 9.11849118491185,
+      "grad_norm": 0.6987513899803162,
+      "learning_rate": 3.7143248869252022e-06,
+      "loss": 0.5058,
+      "step": 5560
+    },
+    {
+      "epoch": 9.13489134891349,
+      "grad_norm": 0.6789569854736328,
+      "learning_rate": 3.5763176404487564e-06,
+      "loss": 0.4977,
+      "step": 5570
+    },
+    {
+      "epoch": 9.15129151291513,
+      "grad_norm": 0.6613333821296692,
+      "learning_rate": 3.440876342320609e-06,
+      "loss": 0.5096,
+      "step": 5580
+    },
+    {
+      "epoch": 9.16769167691677,
+      "grad_norm": 0.689217746257782,
+      "learning_rate": 3.308004596794101e-06,
+      "loss": 0.4949,
+      "step": 5590
+    },
+    {
+      "epoch": 9.18409184091841,
+      "grad_norm": 0.7100328803062439,
+      "learning_rate": 3.1777059397436692e-06,
+      "loss": 0.5044,
+      "step": 5600
+    },
+    {
+      "epoch": 9.200492004920049,
+      "grad_norm": 0.6680495738983154,
+      "learning_rate": 3.049983838570858e-06,
+      "loss": 0.5037,
+      "step": 5610
+    },
+    {
+      "epoch": 9.216892168921689,
+      "grad_norm": 0.6641035676002502,
+      "learning_rate": 2.9248416921119794e-06,
+      "loss": 0.4964,
+      "step": 5620
+    },
+    {
+      "epoch": 9.233292332923329,
+      "grad_norm": 0.7248251438140869,
+      "learning_rate": 2.8022828305477423e-06,
+      "loss": 0.5017,
+      "step": 5630
+    },
+    {
+      "epoch": 9.24969249692497,
+      "grad_norm": 0.7348821759223938,
+      "learning_rate": 2.682310515314512e-06,
+      "loss": 0.5077,
+      "step": 5640
+    },
+    {
+      "epoch": 9.26609266092661,
+      "grad_norm": 0.6882705688476562,
+      "learning_rate": 2.5649279390176806e-06,
+      "loss": 0.4926,
+      "step": 5650
+    },
+    {
+      "epoch": 9.28249282492825,
+      "grad_norm": 0.6873205900192261,
+      "learning_rate": 2.4501382253465543e-06,
+      "loss": 0.5009,
+      "step": 5660
+    },
+    {
+      "epoch": 9.29889298892989,
+      "grad_norm": 0.703940212726593,
+      "learning_rate": 2.3379444289913342e-06,
+      "loss": 0.5011,
+      "step": 5670
+    },
+    {
+      "epoch": 9.31529315293153,
+      "grad_norm": 0.6873738169670105,
+      "learning_rate": 2.228349535561769e-06,
+      "loss": 0.4991,
+      "step": 5680
+    },
+    {
+      "epoch": 9.331693316933169,
+      "grad_norm": 0.6344056725502014,
+      "learning_rate": 2.1213564615077065e-06,
+      "loss": 0.5099,
+      "step": 5690
+    },
+    {
+      "epoch": 9.348093480934809,
+      "grad_norm": 0.6923158168792725,
+      "learning_rate": 2.016968054041546e-06,
+      "loss": 0.5068,
+      "step": 5700
+    },
+    {
+      "epoch": 9.364493644936449,
+      "grad_norm": 0.6982558965682983,
+      "learning_rate": 1.915187091062387e-06,
+      "loss": 0.5052,
+      "step": 5710
+    },
+    {
+      "epoch": 9.38089380893809,
+      "grad_norm": 0.6965045928955078,
+      "learning_rate": 1.816016281082178e-06,
+      "loss": 0.4963,
+      "step": 5720
+    },
+    {
+      "epoch": 9.39729397293973,
+      "grad_norm": 0.704952597618103,
+      "learning_rate": 1.7194582631535617e-06,
+      "loss": 0.4922,
+      "step": 5730
+    },
+    {
+      "epoch": 9.41369413694137,
+      "grad_norm": 0.673637866973877,
+      "learning_rate": 1.6255156067997323e-06,
+      "loss": 0.4943,
+      "step": 5740
+    },
+    {
+      "epoch": 9.43009430094301,
+      "grad_norm": 0.6612069010734558,
+      "learning_rate": 1.5341908119459792e-06,
+      "loss": 0.4973,
+      "step": 5750
+    },
+    {
+      "epoch": 9.44649446494465,
+      "grad_norm": 0.7053012847900391,
+      "learning_rate": 1.4454863088532388e-06,
+      "loss": 0.5049,
+      "step": 5760
+    },
+    {
+      "epoch": 9.46289462894629,
+      "grad_norm": 0.6800163388252258,
+      "learning_rate": 1.3594044580533482e-06,
+      "loss": 0.5009,
+      "step": 5770
+    },
+    {
+      "epoch": 9.479294792947929,
+      "grad_norm": 0.7071276307106018,
+      "learning_rate": 1.2759475502862828e-06,
+      "loss": 0.5016,
+      "step": 5780
+    },
+    {
+      "epoch": 9.495694956949569,
+      "grad_norm": 0.693181037902832,
+      "learning_rate": 1.19511780643915e-06,
+      "loss": 0.5006,
+      "step": 5790
+    },
+    {
+      "epoch": 9.512095120951209,
+      "grad_norm": 0.7130532264709473,
+      "learning_rate": 1.1169173774871478e-06,
+      "loss": 0.5072,
+      "step": 5800
+    },
+    {
+      "epoch": 9.52849528495285,
+      "grad_norm": 0.6810155510902405,
+      "learning_rate": 1.0413483444362771e-06,
+      "loss": 0.5014,
+      "step": 5810
+    },
+    {
+      "epoch": 9.54489544895449,
+      "grad_norm": 0.6402135491371155,
+      "learning_rate": 9.684127182679526e-07,
+      "loss": 0.4956,
+      "step": 5820
+    },
+    {
+      "epoch": 9.56129561295613,
+      "grad_norm": 0.6540482640266418,
+      "learning_rate": 8.981124398855678e-07,
+      "loss": 0.5035,
+      "step": 5830
+    },
+    {
+      "epoch": 9.57769577695777,
+      "grad_norm": 0.6759128570556641,
+      "learning_rate": 8.304493800627589e-07,
+      "loss": 0.4896,
+      "step": 5840
+    },
+    {
+      "epoch": 9.59409594095941,
+      "grad_norm": 0.67359459400177,
+      "learning_rate": 7.654253393936439e-07,
+      "loss": 0.5071,
+      "step": 5850
+    },
+    {
+      "epoch": 9.61049610496105,
+      "grad_norm": 0.740897536277771,
+      "learning_rate": 7.030420482449395e-07,
+      "loss": 0.5013,
+      "step": 5860
+    },
+    {
+      "epoch": 9.626896268962689,
+      "grad_norm": 0.687887966632843,
+      "learning_rate": 6.433011667098754e-07,
+      "loss": 0.5017,
+      "step": 5870
+    },
+    {
+      "epoch": 9.64329643296433,
+      "grad_norm": 0.6390769481658936,
+      "learning_rate": 5.862042845640403e-07,
+      "loss": 0.5032,
+      "step": 5880
+    },
+    {
+      "epoch": 9.65969659696597,
+      "grad_norm": 0.6661211848258972,
+      "learning_rate": 5.317529212230721e-07,
+      "loss": 0.5026,
+      "step": 5890
+    },
+    {
+      "epoch": 9.67609676096761,
+      "grad_norm": 0.7113086581230164,
+      "learning_rate": 4.799485257022118e-07,
+      "loss": 0.4996,
+      "step": 5900
+    },
+    {
+      "epoch": 9.69249692496925,
+      "grad_norm": 0.6506823301315308,
+      "learning_rate": 4.307924765777682e-07,
+      "loss": 0.5032,
+      "step": 5910
+    },
+    {
+      "epoch": 9.70889708897089,
+      "grad_norm": 0.6711443066596985,
+      "learning_rate": 3.842860819504024e-07,
+      "loss": 0.502,
+      "step": 5920
+    },
+    {
+      "epoch": 9.72529725297253,
+      "grad_norm": 0.7337656021118164,
+      "learning_rate": 3.404305794103224e-07,
+      "loss": 0.5138,
+      "step": 5930
+    },
+    {
+      "epoch": 9.74169741697417,
+      "grad_norm": 0.6719791889190674,
+      "learning_rate": 2.9922713600439854e-07,
+      "loss": 0.5015,
+      "step": 5940
+    },
+    {
+      "epoch": 9.758097580975809,
+      "grad_norm": 0.6952204704284668,
+      "learning_rate": 2.606768482050215e-07,
+      "loss": 0.5161,
+      "step": 5950
+    },
+    {
+      "epoch": 9.774497744977449,
+      "grad_norm": 0.7182480096817017,
+      "learning_rate": 2.2478074188099219e-07,
+      "loss": 0.5056,
+      "step": 5960
+    },
+    {
+      "epoch": 9.79089790897909,
+      "grad_norm": 0.7192881107330322,
+      "learning_rate": 1.915397722702217e-07,
+      "loss": 0.4972,
+      "step": 5970
+    },
+    {
+      "epoch": 9.80729807298073,
+      "grad_norm": 0.6714368462562561,
+      "learning_rate": 1.609548239542402e-07,
+      "loss": 0.4958,
+      "step": 5980
+    },
+    {
+      "epoch": 9.82369823698237,
+      "grad_norm": 0.6657147407531738,
+      "learning_rate": 1.3302671083474938e-07,
+      "loss": 0.4943,
+      "step": 5990
+    },
+    {
+      "epoch": 9.84009840098401,
+      "grad_norm": 0.6431368589401245,
+      "learning_rate": 1.0775617611189503e-07,
+      "loss": 0.5079,
+      "step": 6000
+    },
+    {
+      "epoch": 9.85649856498565,
+      "grad_norm": 0.7010581493377686,
+      "learning_rate": 8.514389226452757e-08,
+      "loss": 0.52,
+      "step": 6010
+    },
+    {
+      "epoch": 9.87289872898729,
+      "grad_norm": 0.6842231750488281,
+      "learning_rate": 6.519046103230508e-08,
+      "loss": 0.5011,
+      "step": 6020
+    },
+    {
+      "epoch": 9.88929889298893,
+      "grad_norm": 0.6943760514259338,
+      "learning_rate": 4.789641339963957e-08,
+      "loss": 0.5041,
+      "step": 6030
+    },
+    {
+      "epoch": 9.90569905699057,
+      "grad_norm": 0.6967430710792542,
+      "learning_rate": 3.3262209581619297e-08,
+      "loss": 0.5096,
+      "step": 6040
+    },
+    {
+      "epoch": 9.92209922099221,
+      "grad_norm": 0.6971690058708191,
+      "learning_rate": 2.1288239011729717e-08,
+      "loss": 0.5076,
+      "step": 6050
+    },
+    {
+      "epoch": 9.93849938499385,
+      "grad_norm": 0.668303906917572,
+      "learning_rate": 1.1974820331517312e-08,
+      "loss": 0.497,
+      "step": 6060
+    },
+    {
+      "epoch": 9.95489954899549,
+      "grad_norm": 0.6784859895706177,
+      "learning_rate": 5.3222013820741765e-09,
+      "loss": 0.5058,
+      "step": 6070
+    },
+    {
+      "epoch": 9.97129971299713,
+      "grad_norm": 0.6811095476150513,
+      "learning_rate": 1.3305591974543953e-09,
+      "loss": 0.501,
+      "step": 6080
+    },
+    {
+      "epoch": 9.98769987699877,
+      "grad_norm": 0.6882724165916443,
+      "learning_rate": 0.0,
+      "loss": 0.5079,
+      "step": 6090
+    },
+    {
+      "epoch": 9.98769987699877,
+      "step": 6090,
+      "total_flos": 4.6640782508241715e+17,
+      "train_loss": 0.5962841084046513,
+      "train_runtime": 32440.4443,
+      "train_samples_per_second": 2.255,
+      "train_steps_per_second": 0.188
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6090,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 500,
+  "total_flos": 4.6640782508241715e+17,
+  "train_batch_size": 3,
+  "trial_name": null,
+  "trial_params": null
+}