diff --git "a/few_shot/checkpoint-15500/trainer_state.json" "b/few_shot/checkpoint-15500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/few_shot/checkpoint-15500/trainer_state.json"
@@ -0,0 +1,10883 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9415623861013243,
+  "eval_steps": 500,
+  "global_step": 15500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0006074596039363382,
+      "grad_norm": 2.236969232559204,
+      "learning_rate": 0.0002,
+      "loss": 2.2628,
+      "step": 10
+    },
+    {
+      "epoch": 0.0012149192078726764,
+      "grad_norm": 1.873207688331604,
+      "learning_rate": 0.0002,
+      "loss": 1.6589,
+      "step": 20
+    },
+    {
+      "epoch": 0.0018223788118090147,
+      "grad_norm": 1.7144545316696167,
+      "learning_rate": 0.0002,
+      "loss": 1.6129,
+      "step": 30
+    },
+    {
+      "epoch": 0.0024298384157453528,
+      "grad_norm": 1.6338379383087158,
+      "learning_rate": 0.0002,
+      "loss": 1.565,
+      "step": 40
+    },
+    {
+      "epoch": 0.003037298019681691,
+      "grad_norm": 1.733494758605957,
+      "learning_rate": 0.0002,
+      "loss": 1.5043,
+      "step": 50
+    },
+    {
+      "epoch": 0.0036447576236180294,
+      "grad_norm": 1.6118093729019165,
+      "learning_rate": 0.0002,
+      "loss": 1.4686,
+      "step": 60
+    },
+    {
+      "epoch": 0.004252217227554368,
+      "grad_norm": 1.5905429124832153,
+      "learning_rate": 0.0002,
+      "loss": 1.4902,
+      "step": 70
+    },
+    {
+      "epoch": 0.0048596768314907056,
+      "grad_norm": 1.667108416557312,
+      "learning_rate": 0.0002,
+      "loss": 1.4311,
+      "step": 80
+    },
+    {
+      "epoch": 0.005467136435427044,
+      "grad_norm": 1.813370943069458,
+      "learning_rate": 0.0002,
+      "loss": 1.3872,
+      "step": 90
+    },
+    {
+      "epoch": 0.006074596039363382,
+      "grad_norm": 1.6819970607757568,
+      "learning_rate": 0.0002,
+      "loss": 1.4375,
+      "step": 100
+    },
+    {
+      "epoch": 0.006682055643299721,
+      "grad_norm": 1.6235817670822144,
+      "learning_rate": 0.0002,
+      "loss": 1.4136,
+      "step": 110
+    },
+    {
+      "epoch": 0.007289515247236059,
+      "grad_norm": 1.6433429718017578,
+      "learning_rate": 0.0002,
+      "loss": 1.3771,
+      "step": 120
+    },
+    {
+      "epoch": 0.007896974851172398,
+      "grad_norm": 1.6093605756759644,
+      "learning_rate": 0.0002,
+      "loss": 1.4332,
+      "step": 130
+    },
+    {
+      "epoch": 0.008504434455108735,
+      "grad_norm": 1.63667893409729,
+      "learning_rate": 0.0002,
+      "loss": 1.4148,
+      "step": 140
+    },
+    {
+      "epoch": 0.009111894059045073,
+      "grad_norm": 1.5507344007492065,
+      "learning_rate": 0.0002,
+      "loss": 1.3764,
+      "step": 150
+    },
+    {
+      "epoch": 0.009719353662981411,
+      "grad_norm": 1.6159253120422363,
+      "learning_rate": 0.0002,
+      "loss": 1.3141,
+      "step": 160
+    },
+    {
+      "epoch": 0.01032681326691775,
+      "grad_norm": 1.6511636972427368,
+      "learning_rate": 0.0002,
+      "loss": 1.4633,
+      "step": 170
+    },
+    {
+      "epoch": 0.010934272870854089,
+      "grad_norm": 1.8297101259231567,
+      "learning_rate": 0.0002,
+      "loss": 1.3747,
+      "step": 180
+    },
+    {
+      "epoch": 0.011541732474790426,
+      "grad_norm": 1.8446108102798462,
+      "learning_rate": 0.0002,
+      "loss": 1.3742,
+      "step": 190
+    },
+    {
+      "epoch": 0.012149192078726764,
+      "grad_norm": 1.6419591903686523,
+      "learning_rate": 0.0002,
+      "loss": 1.329,
+      "step": 200
+    },
+    {
+      "epoch": 0.012756651682663102,
+      "grad_norm": 1.595116376876831,
+      "learning_rate": 0.0002,
+      "loss": 1.3658,
+      "step": 210
+    },
+    {
+      "epoch": 0.013364111286599442,
+      "grad_norm": 1.6085125207901,
+      "learning_rate": 0.0002,
+      "loss": 1.3062,
+      "step": 220
+    },
+    {
+      "epoch": 0.01397157089053578,
+      "grad_norm": 1.6018210649490356,
+      "learning_rate": 0.0002,
+      "loss": 1.3563,
+      "step": 230
+    },
+    {
+      "epoch": 0.014579030494472118,
+      "grad_norm": 1.5757384300231934,
+      "learning_rate": 0.0002,
+      "loss": 1.4252,
+      "step": 240
+    },
+    {
+      "epoch": 0.015186490098408455,
+      "grad_norm": 1.6396926641464233,
+      "learning_rate": 0.0002,
+      "loss": 1.3044,
+      "step": 250
+    },
+    {
+      "epoch": 0.015793949702344795,
+      "grad_norm": 1.7595162391662598,
+      "learning_rate": 0.0002,
+      "loss": 1.3885,
+      "step": 260
+    },
+    {
+      "epoch": 0.01640140930628113,
+      "grad_norm": 1.5528141260147095,
+      "learning_rate": 0.0002,
+      "loss": 1.2728,
+      "step": 270
+    },
+    {
+      "epoch": 0.01700886891021747,
+      "grad_norm": 1.536747932434082,
+      "learning_rate": 0.0002,
+      "loss": 1.3512,
+      "step": 280
+    },
+    {
+      "epoch": 0.01761632851415381,
+      "grad_norm": 1.6435034275054932,
+      "learning_rate": 0.0002,
+      "loss": 1.338,
+      "step": 290
+    },
+    {
+      "epoch": 0.018223788118090146,
+      "grad_norm": 1.5331226587295532,
+      "learning_rate": 0.0002,
+      "loss": 1.3519,
+      "step": 300
+    },
+    {
+      "epoch": 0.018831247722026486,
+      "grad_norm": 1.6326663494110107,
+      "learning_rate": 0.0002,
+      "loss": 1.3089,
+      "step": 310
+    },
+    {
+      "epoch": 0.019438707325962822,
+      "grad_norm": 1.567794919013977,
+      "learning_rate": 0.0002,
+      "loss": 1.253,
+      "step": 320
+    },
+    {
+      "epoch": 0.020046166929899162,
+      "grad_norm": 1.5589838027954102,
+      "learning_rate": 0.0002,
+      "loss": 1.33,
+      "step": 330
+    },
+    {
+      "epoch": 0.0206536265338355,
+      "grad_norm": 1.5770776271820068,
+      "learning_rate": 0.0002,
+      "loss": 1.3316,
+      "step": 340
+    },
+    {
+      "epoch": 0.021261086137771838,
+      "grad_norm": 1.4471594095230103,
+      "learning_rate": 0.0002,
+      "loss": 1.2864,
+      "step": 350
+    },
+    {
+      "epoch": 0.021868545741708177,
+      "grad_norm": 1.5740830898284912,
+      "learning_rate": 0.0002,
+      "loss": 1.2921,
+      "step": 360
+    },
+    {
+      "epoch": 0.022476005345644513,
+      "grad_norm": 1.7635418176651,
+      "learning_rate": 0.0002,
+      "loss": 1.2212,
+      "step": 370
+    },
+    {
+      "epoch": 0.023083464949580853,
+      "grad_norm": 1.5588957071304321,
+      "learning_rate": 0.0002,
+      "loss": 1.226,
+      "step": 380
+    },
+    {
+      "epoch": 0.023690924553517193,
+      "grad_norm": 1.6329195499420166,
+      "learning_rate": 0.0002,
+      "loss": 1.3694,
+      "step": 390
+    },
+    {
+      "epoch": 0.02429838415745353,
+      "grad_norm": 1.5749831199645996,
+      "learning_rate": 0.0002,
+      "loss": 1.3151,
+      "step": 400
+    },
+    {
+      "epoch": 0.024905843761389868,
+      "grad_norm": 1.776865839958191,
+      "learning_rate": 0.0002,
+      "loss": 1.2971,
+      "step": 410
+    },
+    {
+      "epoch": 0.025513303365326204,
+      "grad_norm": 1.4414085149765015,
+      "learning_rate": 0.0002,
+      "loss": 1.3434,
+      "step": 420
+    },
+    {
+      "epoch": 0.026120762969262544,
+      "grad_norm": 1.672532081604004,
+      "learning_rate": 0.0002,
+      "loss": 1.2647,
+      "step": 430
+    },
+    {
+      "epoch": 0.026728222573198884,
+      "grad_norm": 1.7707642316818237,
+      "learning_rate": 0.0002,
+      "loss": 1.2875,
+      "step": 440
+    },
+    {
+      "epoch": 0.02733568217713522,
+      "grad_norm": 1.510501503944397,
+      "learning_rate": 0.0002,
+      "loss": 1.2962,
+      "step": 450
+    },
+    {
+      "epoch": 0.02794314178107156,
+      "grad_norm": 1.5336624383926392,
+      "learning_rate": 0.0002,
+      "loss": 1.3686,
+      "step": 460
+    },
+    {
+      "epoch": 0.028550601385007895,
+      "grad_norm": 1.785957932472229,
+      "learning_rate": 0.0002,
+      "loss": 1.2887,
+      "step": 470
+    },
+    {
+      "epoch": 0.029158060988944235,
+      "grad_norm": 1.5779095888137817,
+      "learning_rate": 0.0002,
+      "loss": 1.3048,
+      "step": 480
+    },
+    {
+      "epoch": 0.029765520592880575,
+      "grad_norm": 1.9468932151794434,
+      "learning_rate": 0.0002,
+      "loss": 1.2892,
+      "step": 490
+    },
+    {
+      "epoch": 0.03037298019681691,
+      "grad_norm": 1.5167356729507446,
+      "learning_rate": 0.0002,
+      "loss": 1.3858,
+      "step": 500
+    },
+    {
+      "epoch": 0.03098043980075325,
+      "grad_norm": 1.3795329332351685,
+      "learning_rate": 0.0002,
+      "loss": 1.3023,
+      "step": 510
+    },
+    {
+      "epoch": 0.03158789940468959,
+      "grad_norm": 1.5920099020004272,
+      "learning_rate": 0.0002,
+      "loss": 1.3765,
+      "step": 520
+    },
+    {
+      "epoch": 0.032195359008625926,
+      "grad_norm": 1.462296724319458,
+      "learning_rate": 0.0002,
+      "loss": 1.2663,
+      "step": 530
+    },
+    {
+      "epoch": 0.03280281861256226,
+      "grad_norm": 1.3863338232040405,
+      "learning_rate": 0.0002,
+      "loss": 1.2386,
+      "step": 540
+    },
+    {
+      "epoch": 0.033410278216498605,
+      "grad_norm": 1.8175960779190063,
+      "learning_rate": 0.0002,
+      "loss": 1.3238,
+      "step": 550
+    },
+    {
+      "epoch": 0.03401773782043494,
+      "grad_norm": 1.7048077583312988,
+      "learning_rate": 0.0002,
+      "loss": 1.257,
+      "step": 560
+    },
+    {
+      "epoch": 0.03462519742437128,
+      "grad_norm": 1.5479084253311157,
+      "learning_rate": 0.0002,
+      "loss": 1.23,
+      "step": 570
+    },
+    {
+      "epoch": 0.03523265702830762,
+      "grad_norm": 1.515726089477539,
+      "learning_rate": 0.0002,
+      "loss": 1.2893,
+      "step": 580
+    },
+    {
+      "epoch": 0.03584011663224396,
+      "grad_norm": 1.7023965120315552,
+      "learning_rate": 0.0002,
+      "loss": 1.2783,
+      "step": 590
+    },
+    {
+      "epoch": 0.03644757623618029,
+      "grad_norm": 1.7180956602096558,
+      "learning_rate": 0.0002,
+      "loss": 1.2578,
+      "step": 600
+    },
+    {
+      "epoch": 0.03705503584011663,
+      "grad_norm": 1.3848283290863037,
+      "learning_rate": 0.0002,
+      "loss": 1.349,
+      "step": 610
+    },
+    {
+      "epoch": 0.03766249544405297,
+      "grad_norm": 1.4944247007369995,
+      "learning_rate": 0.0002,
+      "loss": 1.2718,
+      "step": 620
+    },
+    {
+      "epoch": 0.03826995504798931,
+      "grad_norm": 1.581346869468689,
+      "learning_rate": 0.0002,
+      "loss": 1.2798,
+      "step": 630
+    },
+    {
+      "epoch": 0.038877414651925644,
+      "grad_norm": 1.490548849105835,
+      "learning_rate": 0.0002,
+      "loss": 1.2878,
+      "step": 640
+    },
+    {
+      "epoch": 0.03948487425586199,
+      "grad_norm": 1.4963841438293457,
+      "learning_rate": 0.0002,
+      "loss": 1.2777,
+      "step": 650
+    },
+    {
+      "epoch": 0.040092333859798324,
+      "grad_norm": 1.729665994644165,
+      "learning_rate": 0.0002,
+      "loss": 1.2781,
+      "step": 660
+    },
+    {
+      "epoch": 0.04069979346373466,
+      "grad_norm": 1.5069007873535156,
+      "learning_rate": 0.0002,
+      "loss": 1.2978,
+      "step": 670
+    },
+    {
+      "epoch": 0.041307253067671,
+      "grad_norm": 4.421722412109375,
+      "learning_rate": 0.0002,
+      "loss": 1.2765,
+      "step": 680
+    },
+    {
+      "epoch": 0.04191471267160734,
+      "grad_norm": 1.6383577585220337,
+      "learning_rate": 0.0002,
+      "loss": 1.3364,
+      "step": 690
+    },
+    {
+      "epoch": 0.042522172275543675,
+      "grad_norm": 1.9025250673294067,
+      "learning_rate": 0.0002,
+      "loss": 1.3184,
+      "step": 700
+    },
+    {
+      "epoch": 0.04312963187948001,
+      "grad_norm": 1.4485440254211426,
+      "learning_rate": 0.0002,
+      "loss": 1.2749,
+      "step": 710
+    },
+    {
+      "epoch": 0.043737091483416354,
+      "grad_norm": 1.3731114864349365,
+      "learning_rate": 0.0002,
+      "loss": 1.2254,
+      "step": 720
+    },
+    {
+      "epoch": 0.04434455108735269,
+      "grad_norm": 2.1447882652282715,
+      "learning_rate": 0.0002,
+      "loss": 1.227,
+      "step": 730
+    },
+    {
+      "epoch": 0.04495201069128903,
+      "grad_norm": 1.4844560623168945,
+      "learning_rate": 0.0002,
+      "loss": 1.2656,
+      "step": 740
+    },
+    {
+      "epoch": 0.04555947029522537,
+      "grad_norm": 1.6620945930480957,
+      "learning_rate": 0.0002,
+      "loss": 1.299,
+      "step": 750
+    },
+    {
+      "epoch": 0.046166929899161706,
+      "grad_norm": 1.4194166660308838,
+      "learning_rate": 0.0002,
+      "loss": 1.2961,
+      "step": 760
+    },
+    {
+      "epoch": 0.04677438950309804,
+      "grad_norm": 1.4977781772613525,
+      "learning_rate": 0.0002,
+      "loss": 1.324,
+      "step": 770
+    },
+    {
+      "epoch": 0.047381849107034385,
+      "grad_norm": 1.5294965505599976,
+      "learning_rate": 0.0002,
+      "loss": 1.2344,
+      "step": 780
+    },
+    {
+      "epoch": 0.04798930871097072,
+      "grad_norm": 1.6185721158981323,
+      "learning_rate": 0.0002,
+      "loss": 1.3811,
+      "step": 790
+    },
+    {
+      "epoch": 0.04859676831490706,
+      "grad_norm": 1.8349323272705078,
+      "learning_rate": 0.0002,
+      "loss": 1.2636,
+      "step": 800
+    },
+    {
+      "epoch": 0.0492042279188434,
+      "grad_norm": 1.4457494020462036,
+      "learning_rate": 0.0002,
+      "loss": 1.278,
+      "step": 810
+    },
+    {
+      "epoch": 0.049811687522779736,
+      "grad_norm": 1.3572221994400024,
+      "learning_rate": 0.0002,
+      "loss": 1.2306,
+      "step": 820
+    },
+    {
+      "epoch": 0.05041914712671607,
+      "grad_norm": 1.3937549591064453,
+      "learning_rate": 0.0002,
+      "loss": 1.3023,
+      "step": 830
+    },
+    {
+      "epoch": 0.05102660673065241,
+      "grad_norm": 1.450095295906067,
+      "learning_rate": 0.0002,
+      "loss": 1.2393,
+      "step": 840
+    },
+    {
+      "epoch": 0.05163406633458875,
+      "grad_norm": 1.4043591022491455,
+      "learning_rate": 0.0002,
+      "loss": 1.2574,
+      "step": 850
+    },
+    {
+      "epoch": 0.05224152593852509,
+      "grad_norm": 1.4549307823181152,
+      "learning_rate": 0.0002,
+      "loss": 1.2938,
+      "step": 860
+    },
+    {
+      "epoch": 0.052848985542461424,
+      "grad_norm": 1.4602893590927124,
+      "learning_rate": 0.0002,
+      "loss": 1.1861,
+      "step": 870
+    },
+    {
+      "epoch": 0.05345644514639777,
+      "grad_norm": 1.867261290550232,
+      "learning_rate": 0.0002,
+      "loss": 1.3015,
+      "step": 880
+    },
+    {
+      "epoch": 0.0540639047503341,
+      "grad_norm": 1.6120606660842896,
+      "learning_rate": 0.0002,
+      "loss": 1.268,
+      "step": 890
+    },
+    {
+      "epoch": 0.05467136435427044,
+      "grad_norm": 1.6458678245544434,
+      "learning_rate": 0.0002,
+      "loss": 1.2661,
+      "step": 900
+    },
+    {
+      "epoch": 0.05527882395820678,
+      "grad_norm": 1.729551076889038,
+      "learning_rate": 0.0002,
+      "loss": 1.28,
+      "step": 910
+    },
+    {
+      "epoch": 0.05588628356214312,
+      "grad_norm": 1.5503125190734863,
+      "learning_rate": 0.0002,
+      "loss": 1.2612,
+      "step": 920
+    },
+    {
+      "epoch": 0.056493743166079455,
+      "grad_norm": 1.5638302564620972,
+      "learning_rate": 0.0002,
+      "loss": 1.3259,
+      "step": 930
+    },
+    {
+      "epoch": 0.05710120277001579,
+      "grad_norm": 1.4959323406219482,
+      "learning_rate": 0.0002,
+      "loss": 1.2489,
+      "step": 940
+    },
+    {
+      "epoch": 0.057708662373952134,
+      "grad_norm": 1.5417566299438477,
+      "learning_rate": 0.0002,
+      "loss": 1.2675,
+      "step": 950
+    },
+    {
+      "epoch": 0.05831612197788847,
+      "grad_norm": 1.5263869762420654,
+      "learning_rate": 0.0002,
+      "loss": 1.3438,
+      "step": 960
+    },
+    {
+      "epoch": 0.058923581581824806,
+      "grad_norm": 1.4479107856750488,
+      "learning_rate": 0.0002,
+      "loss": 1.2868,
+      "step": 970
+    },
+    {
+      "epoch": 0.05953104118576115,
+      "grad_norm": 1.678945541381836,
+      "learning_rate": 0.0002,
+      "loss": 1.2204,
+      "step": 980
+    },
+    {
+      "epoch": 0.060138500789697485,
+      "grad_norm": 1.431054949760437,
+      "learning_rate": 0.0002,
+      "loss": 1.2624,
+      "step": 990
+    },
+    {
+      "epoch": 0.06074596039363382,
+      "grad_norm": 1.6529697179794312,
+      "learning_rate": 0.0002,
+      "loss": 1.2867,
+      "step": 1000
+    },
+    {
+      "epoch": 0.061353419997570165,
+      "grad_norm": 1.4838560819625854,
+      "learning_rate": 0.0002,
+      "loss": 1.2556,
+      "step": 1010
+    },
+    {
+      "epoch": 0.0619608796015065,
+      "grad_norm": 1.303206443786621,
+      "learning_rate": 0.0002,
+      "loss": 1.1895,
+      "step": 1020
+    },
+    {
+      "epoch": 0.06256833920544284,
+      "grad_norm": 1.6381781101226807,
+      "learning_rate": 0.0002,
+      "loss": 1.2975,
+      "step": 1030
+    },
+    {
+      "epoch": 0.06317579880937918,
+      "grad_norm": 1.589048981666565,
+      "learning_rate": 0.0002,
+      "loss": 1.2776,
+      "step": 1040
+    },
+    {
+      "epoch": 0.06378325841331552,
+      "grad_norm": 1.71570885181427,
+      "learning_rate": 0.0002,
+      "loss": 1.2703,
+      "step": 1050
+    },
+    {
+      "epoch": 0.06439071801725185,
+      "grad_norm": 1.416674017906189,
+      "learning_rate": 0.0002,
+      "loss": 1.2465,
+      "step": 1060
+    },
+    {
+      "epoch": 0.06499817762118819,
+      "grad_norm": 1.773748517036438,
+      "learning_rate": 0.0002,
+      "loss": 1.339,
+      "step": 1070
+    },
+    {
+      "epoch": 0.06560563722512452,
+      "grad_norm": 1.5272996425628662,
+      "learning_rate": 0.0002,
+      "loss": 1.2973,
+      "step": 1080
+    },
+    {
+      "epoch": 0.06621309682906086,
+      "grad_norm": 1.3926866054534912,
+      "learning_rate": 0.0002,
+      "loss": 1.3862,
+      "step": 1090
+    },
+    {
+      "epoch": 0.06682055643299721,
+      "grad_norm": 1.576254963874817,
+      "learning_rate": 0.0002,
+      "loss": 1.2522,
+      "step": 1100
+    },
+    {
+      "epoch": 0.06742801603693355,
+      "grad_norm": 1.6619774103164673,
+      "learning_rate": 0.0002,
+      "loss": 1.3027,
+      "step": 1110
+    },
+    {
+      "epoch": 0.06803547564086988,
+      "grad_norm": 1.4636040925979614,
+      "learning_rate": 0.0002,
+      "loss": 1.3649,
+      "step": 1120
+    },
+    {
+      "epoch": 0.06864293524480622,
+      "grad_norm": 1.6307445764541626,
+      "learning_rate": 0.0002,
+      "loss": 1.2553,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06925039484874256,
+      "grad_norm": 1.531379222869873,
+      "learning_rate": 0.0002,
+      "loss": 1.2578,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06985785445267889,
+      "grad_norm": 1.5244723558425903,
+      "learning_rate": 0.0002,
+      "loss": 1.2745,
+      "step": 1150
+    },
+    {
+      "epoch": 0.07046531405661524,
+      "grad_norm": 1.7863515615463257,
+      "learning_rate": 0.0002,
+      "loss": 1.2835,
+      "step": 1160
+    },
+    {
+      "epoch": 0.07107277366055158,
+      "grad_norm": 1.5718461275100708,
+      "learning_rate": 0.0002,
+      "loss": 1.2352,
+      "step": 1170
+    },
+    {
+      "epoch": 0.07168023326448791,
+      "grad_norm": 1.5710781812667847,
+      "learning_rate": 0.0002,
+      "loss": 1.2357,
+      "step": 1180
+    },
+    {
+      "epoch": 0.07228769286842425,
+      "grad_norm": 1.4836217164993286,
+      "learning_rate": 0.0002,
+      "loss": 1.2811,
+      "step": 1190
+    },
+    {
+      "epoch": 0.07289515247236059,
+      "grad_norm": 1.5769175291061401,
+      "learning_rate": 0.0002,
+      "loss": 1.3826,
+      "step": 1200
+    },
+    {
+      "epoch": 0.07350261207629692,
+      "grad_norm": 1.552350640296936,
+      "learning_rate": 0.0002,
+      "loss": 1.2512,
+      "step": 1210
+    },
+    {
+      "epoch": 0.07411007168023326,
+      "grad_norm": 1.4635241031646729,
+      "learning_rate": 0.0002,
+      "loss": 1.3651,
+      "step": 1220
+    },
+    {
+      "epoch": 0.07471753128416961,
+      "grad_norm": 1.6063421964645386,
+      "learning_rate": 0.0002,
+      "loss": 1.3135,
+      "step": 1230
+    },
+    {
+      "epoch": 0.07532499088810594,
+      "grad_norm": 1.3370627164840698,
+      "learning_rate": 0.0002,
+      "loss": 1.3172,
+      "step": 1240
+    },
+    {
+      "epoch": 0.07593245049204228,
+      "grad_norm": 1.691219449043274,
+      "learning_rate": 0.0002,
+      "loss": 1.114,
+      "step": 1250
+    },
+    {
+      "epoch": 0.07653991009597862,
+      "grad_norm": 1.4442918300628662,
+      "learning_rate": 0.0002,
+      "loss": 1.2884,
+      "step": 1260
+    },
+    {
+      "epoch": 0.07714736969991495,
+      "grad_norm": 1.3976494073867798,
+      "learning_rate": 0.0002,
+      "loss": 1.2585,
+      "step": 1270
+    },
+    {
+      "epoch": 0.07775482930385129,
+      "grad_norm": 1.357978105545044,
+      "learning_rate": 0.0002,
+      "loss": 1.2449,
+      "step": 1280
+    },
+    {
+      "epoch": 0.07836228890778764,
+      "grad_norm": 1.4768654108047485,
+      "learning_rate": 0.0002,
+      "loss": 1.1673,
+      "step": 1290
+    },
+    {
+      "epoch": 0.07896974851172398,
+      "grad_norm": 1.3699668645858765,
+      "learning_rate": 0.0002,
+      "loss": 1.3029,
+      "step": 1300
+    },
+    {
+      "epoch": 0.07957720811566031,
+      "grad_norm": 1.494268774986267,
+      "learning_rate": 0.0002,
+      "loss": 1.2156,
+      "step": 1310
+    },
+    {
+      "epoch": 0.08018466771959665,
+      "grad_norm": 1.4966034889221191,
+      "learning_rate": 0.0002,
+      "loss": 1.2387,
+      "step": 1320
+    },
+    {
+      "epoch": 0.08079212732353298,
+      "grad_norm": 1.5984331369400024,
+      "learning_rate": 0.0002,
+      "loss": 1.2559,
+      "step": 1330
+    },
+    {
+      "epoch": 0.08139958692746932,
+      "grad_norm": 1.6565812826156616,
+      "learning_rate": 0.0002,
+      "loss": 1.2035,
+      "step": 1340
+    },
+    {
+      "epoch": 0.08200704653140566,
+      "grad_norm": 1.3882875442504883,
+      "learning_rate": 0.0002,
+      "loss": 1.2266,
+      "step": 1350
+    },
+    {
+      "epoch": 0.082614506135342,
+      "grad_norm": 1.4133611917495728,
+      "learning_rate": 0.0002,
+      "loss": 1.2908,
+      "step": 1360
+    },
+    {
+      "epoch": 0.08322196573927834,
+      "grad_norm": 1.5866031646728516,
+      "learning_rate": 0.0002,
+      "loss": 1.2541,
+      "step": 1370
+    },
+    {
+      "epoch": 0.08382942534321468,
+      "grad_norm": 1.7129892110824585,
+      "learning_rate": 0.0002,
+      "loss": 1.2653,
+      "step": 1380
+    },
+    {
+      "epoch": 0.08443688494715101,
+      "grad_norm": 1.821727991104126,
+      "learning_rate": 0.0002,
+      "loss": 1.2413,
+      "step": 1390
+    },
+    {
+      "epoch": 0.08504434455108735,
+      "grad_norm": 1.5216853618621826,
+      "learning_rate": 0.0002,
+      "loss": 1.2449,
+      "step": 1400
+    },
+    {
+      "epoch": 0.08565180415502369,
+      "grad_norm": 1.3088600635528564,
+      "learning_rate": 0.0002,
+      "loss": 1.1995,
+      "step": 1410
+    },
+    {
+      "epoch": 0.08625926375896002,
+      "grad_norm": 1.467633843421936,
+      "learning_rate": 0.0002,
+      "loss": 1.2901,
+      "step": 1420
+    },
+    {
+      "epoch": 0.08686672336289637,
+      "grad_norm": 1.467429518699646,
+      "learning_rate": 0.0002,
+      "loss": 1.2593,
+      "step": 1430
+    },
+    {
+      "epoch": 0.08747418296683271,
+      "grad_norm": 1.5163699388504028,
+      "learning_rate": 0.0002,
+      "loss": 1.3415,
+      "step": 1440
+    },
+    {
+      "epoch": 0.08808164257076904,
+      "grad_norm": 1.4762097597122192,
+      "learning_rate": 0.0002,
+      "loss": 1.2328,
+      "step": 1450
+    },
+    {
+      "epoch": 0.08868910217470538,
+      "grad_norm": 1.3106896877288818,
+      "learning_rate": 0.0002,
+      "loss": 1.2625,
+      "step": 1460
+    },
+    {
+      "epoch": 0.08929656177864172,
+      "grad_norm": 1.7591036558151245,
+      "learning_rate": 0.0002,
+      "loss": 1.2654,
+      "step": 1470
+    },
+    {
+      "epoch": 0.08990402138257805,
+      "grad_norm": 1.425759196281433,
+      "learning_rate": 0.0002,
+      "loss": 1.2492,
+      "step": 1480
+    },
+    {
+      "epoch": 0.0905114809865144,
+      "grad_norm": 1.4507300853729248,
+      "learning_rate": 0.0002,
+      "loss": 1.3173,
+      "step": 1490
+    },
+    {
+      "epoch": 0.09111894059045074,
+      "grad_norm": 1.564005970954895,
+      "learning_rate": 0.0002,
+      "loss": 1.2037,
+      "step": 1500
+    },
+    {
+      "epoch": 0.09172640019438708,
+      "grad_norm": 1.4253226518630981,
+      "learning_rate": 0.0002,
+      "loss": 1.2143,
+      "step": 1510
+    },
+    {
+      "epoch": 0.09233385979832341,
+      "grad_norm": 1.4880207777023315,
+      "learning_rate": 0.0002,
+      "loss": 1.2818,
+      "step": 1520
+    },
+    {
+      "epoch": 0.09294131940225975,
+      "grad_norm": 1.439846396446228,
+      "learning_rate": 0.0002,
+      "loss": 1.2082,
+      "step": 1530
+    },
+    {
+      "epoch": 0.09354877900619608,
+      "grad_norm": 1.5116918087005615,
+      "learning_rate": 0.0002,
+      "loss": 1.279,
+      "step": 1540
+    },
+    {
+      "epoch": 0.09415623861013242,
+      "grad_norm": 1.2679647207260132,
+      "learning_rate": 0.0002,
+      "loss": 1.2541,
+      "step": 1550
+    },
+    {
+      "epoch": 0.09476369821406877,
+      "grad_norm": 1.4966439008712769,
+      "learning_rate": 0.0002,
+      "loss": 1.2426,
+      "step": 1560
+    },
+    {
+      "epoch": 0.0953711578180051,
+      "grad_norm": 1.795647144317627,
+      "learning_rate": 0.0002,
+      "loss": 1.2582,
+      "step": 1570
+    },
+    {
+      "epoch": 0.09597861742194144,
+      "grad_norm": 1.485668659210205,
+      "learning_rate": 0.0002,
+      "loss": 1.2277,
+      "step": 1580
+    },
+    {
+      "epoch": 0.09658607702587778,
+      "grad_norm": 1.4666286706924438,
+      "learning_rate": 0.0002,
+      "loss": 1.1659,
+      "step": 1590
+    },
+    {
+      "epoch": 0.09719353662981411,
+      "grad_norm": 1.4714016914367676,
+      "learning_rate": 0.0002,
+      "loss": 1.2541,
+      "step": 1600
+    },
+    {
+      "epoch": 0.09780099623375045,
+      "grad_norm": 1.351139783859253,
+      "learning_rate": 0.0002,
+      "loss": 1.2047,
+      "step": 1610
+    },
+    {
+      "epoch": 0.0984084558376868,
+      "grad_norm": 1.4304152727127075,
+      "learning_rate": 0.0002,
+      "loss": 1.2028,
+      "step": 1620
+    },
+    {
+      "epoch": 0.09901591544162314,
+      "grad_norm": 1.3749319314956665,
+      "learning_rate": 0.0002,
+      "loss": 1.2509,
+      "step": 1630
+    },
+    {
+      "epoch": 0.09962337504555947,
+      "grad_norm": 1.5823308229446411,
+      "learning_rate": 0.0002,
+      "loss": 1.2278,
+      "step": 1640
+    },
+    {
+      "epoch": 0.10023083464949581,
+      "grad_norm": 1.3146113157272339,
+      "learning_rate": 0.0002,
+      "loss": 1.2092,
+      "step": 1650
+    },
+    {
+      "epoch": 0.10083829425343215,
+      "grad_norm": 1.5478814840316772,
+      "learning_rate": 0.0002,
+      "loss": 1.284,
+      "step": 1660
+    },
+    {
+      "epoch": 0.10144575385736848,
+      "grad_norm": 1.6172051429748535,
+      "learning_rate": 0.0002,
+      "loss": 1.2521,
+      "step": 1670
+    },
+    {
+      "epoch": 0.10205321346130482,
+      "grad_norm": 1.52104914188385,
+      "learning_rate": 0.0002,
+      "loss": 1.2477,
+      "step": 1680
+    },
+    {
+      "epoch": 0.10266067306524117,
+      "grad_norm": 1.5709279775619507,
+      "learning_rate": 0.0002,
+      "loss": 1.2424,
+      "step": 1690
+    },
+    {
+      "epoch": 0.1032681326691775,
+      "grad_norm": 1.4448177814483643,
+      "learning_rate": 0.0002,
+      "loss": 1.2296,
+      "step": 1700
+    },
+    {
+      "epoch": 0.10387559227311384,
+      "grad_norm": 1.3035105466842651,
+      "learning_rate": 0.0002,
+      "loss": 1.2177,
+      "step": 1710
+    },
+    {
+      "epoch": 0.10448305187705018,
+      "grad_norm": 1.6021480560302734,
+      "learning_rate": 0.0002,
+      "loss": 1.2391,
+      "step": 1720
+    },
+    {
+      "epoch": 0.10509051148098651,
+      "grad_norm": 1.5413893461227417,
+      "learning_rate": 0.0002,
+      "loss": 1.2627,
+      "step": 1730
+    },
+    {
+      "epoch": 0.10569797108492285,
+      "grad_norm": 1.582407832145691,
+      "learning_rate": 0.0002,
+      "loss": 1.2344,
+      "step": 1740
+    },
+    {
+      "epoch": 0.10630543068885918,
+      "grad_norm": 1.4765949249267578,
+      "learning_rate": 0.0002,
+      "loss": 1.2268,
+      "step": 1750
+    },
+    {
+      "epoch": 0.10691289029279553,
+      "grad_norm": 1.6818124055862427,
+      "learning_rate": 0.0002,
+      "loss": 1.229,
+      "step": 1760
+    },
+    {
+      "epoch": 0.10752034989673187,
+      "grad_norm": 1.570087194442749,
+      "learning_rate": 0.0002,
+      "loss": 1.322,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1081278095006682,
+      "grad_norm": 1.5548206567764282,
+      "learning_rate": 0.0002,
+      "loss": 1.1993,
+      "step": 1780
+    },
+    {
+      "epoch": 0.10873526910460454,
+      "grad_norm": 1.4159990549087524,
+      "learning_rate": 0.0002,
+      "loss": 1.2136,
+      "step": 1790
+    },
+    {
+      "epoch": 0.10934272870854088,
+      "grad_norm": 1.4259227514266968,
+      "learning_rate": 0.0002,
+      "loss": 1.2826,
+      "step": 1800
+    },
+    {
+      "epoch": 0.10995018831247722,
+      "grad_norm": 1.721754550933838,
+      "learning_rate": 0.0002,
+      "loss": 1.2958,
+      "step": 1810
+    },
+    {
+      "epoch": 0.11055764791641357,
+      "grad_norm": 1.5920616388320923,
+      "learning_rate": 0.0002,
+      "loss": 1.2152,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1111651075203499,
+      "grad_norm": 1.4243106842041016,
+      "learning_rate": 0.0002,
+      "loss": 1.2273,
+      "step": 1830
+    },
+    {
+      "epoch": 0.11177256712428624,
+      "grad_norm": 1.5096100568771362,
+      "learning_rate": 0.0002,
+      "loss": 1.2232,
+      "step": 1840
+    },
+    {
+      "epoch": 0.11238002672822257,
+      "grad_norm": 1.573041558265686,
+      "learning_rate": 0.0002,
+      "loss": 1.2303,
+      "step": 1850
+    },
+    {
+      "epoch": 0.11298748633215891,
+      "grad_norm": 1.546109676361084,
+      "learning_rate": 0.0002,
+      "loss": 1.2366,
+      "step": 1860
+    },
+    {
+      "epoch": 0.11359494593609525,
+      "grad_norm": 1.5966582298278809,
+      "learning_rate": 0.0002,
+      "loss": 1.174,
+      "step": 1870
+    },
+    {
+      "epoch": 0.11420240554003158,
+      "grad_norm": 1.5600007772445679,
+      "learning_rate": 0.0002,
+      "loss": 1.1639,
+      "step": 1880
+    },
+    {
+      "epoch": 0.11480986514396793,
+      "grad_norm": 1.7233026027679443,
+      "learning_rate": 0.0002,
+      "loss": 1.2161,
+      "step": 1890
+    },
+    {
+      "epoch": 0.11541732474790427,
+      "grad_norm": 1.5901647806167603,
+      "learning_rate": 0.0002,
+      "loss": 1.2153,
+      "step": 1900
+    },
+    {
+      "epoch": 0.1160247843518406,
+      "grad_norm": 1.376562237739563,
+      "learning_rate": 0.0002,
+      "loss": 1.266,
+      "step": 1910
+    },
+    {
+      "epoch": 0.11663224395577694,
+      "grad_norm": 1.656231164932251,
+      "learning_rate": 0.0002,
+      "loss": 1.2664,
+      "step": 1920
+    },
+    {
+      "epoch": 0.11723970355971328,
+      "grad_norm": 1.356429100036621,
+      "learning_rate": 0.0002,
+      "loss": 1.2244,
+      "step": 1930
+    },
+    {
+      "epoch": 0.11784716316364961,
+      "grad_norm": 1.5584162473678589,
+      "learning_rate": 0.0002,
+      "loss": 1.2545,
+      "step": 1940
+    },
+    {
+      "epoch": 0.11845462276758595,
+      "grad_norm": 1.5809051990509033,
+      "learning_rate": 0.0002,
+      "loss": 1.2847,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1190620823715223,
+      "grad_norm": 1.6206623315811157,
+      "learning_rate": 0.0002,
+      "loss": 1.1984,
+      "step": 1960
+    },
+    {
+      "epoch": 0.11966954197545863,
+      "grad_norm": 1.5814268589019775,
+      "learning_rate": 0.0002,
+      "loss": 1.2822,
+      "step": 1970
+    },
+    {
+      "epoch": 0.12027700157939497,
+      "grad_norm": 1.6698721647262573,
+      "learning_rate": 0.0002,
+      "loss": 1.2771,
+      "step": 1980
+    },
+    {
+      "epoch": 0.12088446118333131,
+      "grad_norm": 1.7568098306655884,
+      "learning_rate": 0.0002,
+      "loss": 1.2153,
+      "step": 1990
+    },
+    {
+      "epoch": 0.12149192078726764,
+      "grad_norm": 1.5525120496749878,
+      "learning_rate": 0.0002,
+      "loss": 1.1899,
+      "step": 2000
+    },
+    {
+      "epoch": 0.12209938039120398,
+      "grad_norm": 1.451067328453064,
+      "learning_rate": 0.0002,
+      "loss": 1.2241,
+      "step": 2010
+    },
+    {
+      "epoch": 0.12270683999514033,
+      "grad_norm": 1.6643682718276978,
+      "learning_rate": 0.0002,
+      "loss": 1.2726,
+      "step": 2020
+    },
+    {
+      "epoch": 0.12331429959907667,
+      "grad_norm": 1.7143374681472778,
+      "learning_rate": 0.0002,
+      "loss": 1.277,
+      "step": 2030
+    },
+    {
+      "epoch": 0.123921759203013,
+      "grad_norm": 1.7562227249145508,
+      "learning_rate": 0.0002,
+      "loss": 1.2066,
+      "step": 2040
+    },
+    {
+      "epoch": 0.12452921880694934,
+      "grad_norm": 1.5043004751205444,
+      "learning_rate": 0.0002,
+      "loss": 1.1832,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1251366784108857,
+      "grad_norm": 1.7817741632461548,
+      "learning_rate": 0.0002,
+      "loss": 1.1517,
+      "step": 2060
+    },
+    {
+      "epoch": 0.125744138014822,
+      "grad_norm": 1.7085005044937134,
+      "learning_rate": 0.0002,
+      "loss": 1.2548,
+      "step": 2070
+    },
+    {
+      "epoch": 0.12635159761875836,
+      "grad_norm": 1.3320608139038086,
+      "learning_rate": 0.0002,
+      "loss": 1.1689,
+      "step": 2080
+    },
+    {
+      "epoch": 0.12695905722269468,
+      "grad_norm": 1.2270746231079102,
+      "learning_rate": 0.0002,
+      "loss": 1.3156,
+      "step": 2090
+    },
+    {
+      "epoch": 0.12756651682663103,
+      "grad_norm": 1.4981920719146729,
+      "learning_rate": 0.0002,
+      "loss": 1.2243,
+      "step": 2100
+    },
+    {
+      "epoch": 0.12817397643056735,
+      "grad_norm": 1.4143863916397095,
+      "learning_rate": 0.0002,
+      "loss": 1.2765,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1287814360345037,
+      "grad_norm": 1.4701218605041504,
+      "learning_rate": 0.0002,
+      "loss": 1.1679,
+      "step": 2120
+    },
+    {
+      "epoch": 0.12938889563844005,
+      "grad_norm": 1.3249510526657104,
+      "learning_rate": 0.0002,
+      "loss": 1.2313,
+      "step": 2130
+    },
+    {
+      "epoch": 0.12999635524237638,
+      "grad_norm": 1.3991800546646118,
+      "learning_rate": 0.0002,
+      "loss": 1.2078,
+      "step": 2140
+    },
+    {
+      "epoch": 0.13060381484631273,
+      "grad_norm": 1.5967034101486206,
+      "learning_rate": 0.0002,
+      "loss": 1.1447,
+      "step": 2150
+    },
+    {
+      "epoch": 0.13121127445024905,
+      "grad_norm": 1.5604697465896606,
+      "learning_rate": 0.0002,
+      "loss": 1.247,
+      "step": 2160
+    },
+    {
+      "epoch": 0.1318187340541854,
+      "grad_norm": 1.3492066860198975,
+      "learning_rate": 0.0002,
+      "loss": 1.2734,
+      "step": 2170
+    },
+    {
+      "epoch": 0.13242619365812172,
+      "grad_norm": 1.6497987508773804,
+      "learning_rate": 0.0002,
+      "loss": 1.2067,
+      "step": 2180
+    },
+    {
+      "epoch": 0.13303365326205807,
+      "grad_norm": 1.4936901330947876,
+      "learning_rate": 0.0002,
+      "loss": 1.2062,
+      "step": 2190
+    },
+    {
+      "epoch": 0.13364111286599442,
+      "grad_norm": 1.4576996564865112,
+      "learning_rate": 0.0002,
+      "loss": 1.2225,
+      "step": 2200
+    },
+    {
+      "epoch": 0.13424857246993074,
+      "grad_norm": 1.3753255605697632,
+      "learning_rate": 0.0002,
+      "loss": 1.2431,
+      "step": 2210
+    },
+    {
+      "epoch": 0.1348560320738671,
+      "grad_norm": 1.563539743423462,
+      "learning_rate": 0.0002,
+      "loss": 1.2793,
+      "step": 2220
+    },
+    {
+      "epoch": 0.13546349167780342,
+      "grad_norm": 1.4935153722763062,
+      "learning_rate": 0.0002,
+      "loss": 1.2361,
+      "step": 2230
+    },
+    {
+      "epoch": 0.13607095128173977,
+      "grad_norm": 1.307816505432129,
+      "learning_rate": 0.0002,
+      "loss": 1.2314,
+      "step": 2240
+    },
+    {
+      "epoch": 0.13667841088567612,
+      "grad_norm": 1.535267949104309,
+      "learning_rate": 0.0002,
+      "loss": 1.224,
+      "step": 2250
+    },
+    {
+      "epoch": 0.13728587048961244,
+      "grad_norm": 1.3963550329208374,
+      "learning_rate": 0.0002,
+      "loss": 1.2803,
+      "step": 2260
+    },
+    {
+      "epoch": 0.1378933300935488,
+      "grad_norm": 1.527891755104065,
+      "learning_rate": 0.0002,
+      "loss": 1.1724,
+      "step": 2270
+    },
+    {
+      "epoch": 0.1385007896974851,
+      "grad_norm": 1.4339513778686523,
+      "learning_rate": 0.0002,
+      "loss": 1.2773,
+      "step": 2280
+    },
+    {
+      "epoch": 0.13910824930142146,
+      "grad_norm": 1.9729173183441162,
+      "learning_rate": 0.0002,
+      "loss": 1.2676,
+      "step": 2290
+    },
+    {
+      "epoch": 0.13971570890535778,
+      "grad_norm": 1.4557331800460815,
+      "learning_rate": 0.0002,
+      "loss": 1.1851,
+      "step": 2300
+    },
+    {
+      "epoch": 0.14032316850929413,
+      "grad_norm": 1.4285987615585327,
+      "learning_rate": 0.0002,
+      "loss": 1.2804,
+      "step": 2310
+    },
+    {
+      "epoch": 0.14093062811323048,
+      "grad_norm": 1.2934836149215698,
+      "learning_rate": 0.0002,
+      "loss": 1.2107,
+      "step": 2320
+    },
+    {
+      "epoch": 0.1415380877171668,
+      "grad_norm": 1.3619877099990845,
+      "learning_rate": 0.0002,
+      "loss": 1.1618,
+      "step": 2330
+    },
+    {
+      "epoch": 0.14214554732110316,
+      "grad_norm": 1.4161388874053955,
+      "learning_rate": 0.0002,
+      "loss": 1.228,
+      "step": 2340
+    },
+    {
+      "epoch": 0.14275300692503948,
+      "grad_norm": 1.3923609256744385,
+      "learning_rate": 0.0002,
+      "loss": 1.2526,
+      "step": 2350
+    },
+    {
+      "epoch": 0.14336046652897583,
+      "grad_norm": 1.3667365312576294,
+      "learning_rate": 0.0002,
+      "loss": 1.2691,
+      "step": 2360
+    },
+    {
+      "epoch": 0.14396792613291215,
+      "grad_norm": 1.420930027961731,
+      "learning_rate": 0.0002,
+      "loss": 1.2337,
+      "step": 2370
+    },
+    {
+      "epoch": 0.1445753857368485,
+      "grad_norm": 1.5273494720458984,
+      "learning_rate": 0.0002,
+      "loss": 1.2606,
+      "step": 2380
+    },
+    {
+      "epoch": 0.14518284534078485,
+      "grad_norm": 1.5331289768218994,
+      "learning_rate": 0.0002,
+      "loss": 1.1693,
+      "step": 2390
+    },
+    {
+      "epoch": 0.14579030494472117,
+      "grad_norm": 1.8542994260787964,
+      "learning_rate": 0.0002,
+      "loss": 1.2367,
+      "step": 2400
+    },
+    {
+      "epoch": 0.14639776454865752,
+      "grad_norm": 1.3973726034164429,
+      "learning_rate": 0.0002,
+      "loss": 1.27,
+      "step": 2410
+    },
+    {
+      "epoch": 0.14700522415259384,
+      "grad_norm": 1.5454223155975342,
+      "learning_rate": 0.0002,
+      "loss": 1.1725,
+      "step": 2420
+    },
+    {
+      "epoch": 0.1476126837565302,
+      "grad_norm": 1.3396209478378296,
+      "learning_rate": 0.0002,
+      "loss": 1.243,
+      "step": 2430
+    },
+    {
+      "epoch": 0.14822014336046652,
+      "grad_norm": 1.4007655382156372,
+      "learning_rate": 0.0002,
+      "loss": 1.2339,
+      "step": 2440
+    },
+    {
+      "epoch": 0.14882760296440287,
+      "grad_norm": 1.459521770477295,
+      "learning_rate": 0.0002,
+      "loss": 1.2409,
+      "step": 2450
+    },
+    {
+      "epoch": 0.14943506256833922,
+      "grad_norm": 1.4330353736877441,
+      "learning_rate": 0.0002,
+      "loss": 1.265,
+      "step": 2460
+    },
+    {
+      "epoch": 0.15004252217227554,
+      "grad_norm": 1.3534977436065674,
+      "learning_rate": 0.0002,
+      "loss": 1.1958,
+      "step": 2470
+    },
+    {
+      "epoch": 0.1506499817762119,
+      "grad_norm": 1.6496937274932861,
+      "learning_rate": 0.0002,
+      "loss": 1.2301,
+      "step": 2480
+    },
+    {
+      "epoch": 0.1512574413801482,
+      "grad_norm": 1.6388399600982666,
+      "learning_rate": 0.0002,
+      "loss": 1.2952,
+      "step": 2490
+    },
+    {
+      "epoch": 0.15186490098408456,
+      "grad_norm": 1.3567882776260376,
+      "learning_rate": 0.0002,
+      "loss": 1.202,
+      "step": 2500
+    },
+    {
+      "epoch": 0.15247236058802088,
+      "grad_norm": 1.297202706336975,
+      "learning_rate": 0.0002,
+      "loss": 1.1876,
+      "step": 2510
+    },
+    {
+      "epoch": 0.15307982019195723,
+      "grad_norm": 1.912382960319519,
+      "learning_rate": 0.0002,
+      "loss": 1.2745,
+      "step": 2520
+    },
+    {
+      "epoch": 0.15368727979589358,
+      "grad_norm": 1.6245383024215698,
+      "learning_rate": 0.0002,
+      "loss": 1.2009,
+      "step": 2530
+    },
+    {
+      "epoch": 0.1542947393998299,
+      "grad_norm": 1.4549682140350342,
+      "learning_rate": 0.0002,
+      "loss": 1.1817,
+      "step": 2540
+    },
+    {
+      "epoch": 0.15490219900376626,
+      "grad_norm": 1.6212128400802612,
+      "learning_rate": 0.0002,
+      "loss": 1.2409,
+      "step": 2550
+    },
+    {
+      "epoch": 0.15550965860770258,
+      "grad_norm": 1.49772310256958,
+      "learning_rate": 0.0002,
+      "loss": 1.1789,
+      "step": 2560
+    },
+    {
+      "epoch": 0.15611711821163893,
+      "grad_norm": 1.5688447952270508,
+      "learning_rate": 0.0002,
+      "loss": 1.2388,
+      "step": 2570
+    },
+    {
+      "epoch": 0.15672457781557528,
+      "grad_norm": 1.5437800884246826,
+      "learning_rate": 0.0002,
+      "loss": 1.2927,
+      "step": 2580
+    },
+    {
+      "epoch": 0.1573320374195116,
+      "grad_norm": 1.2287709712982178,
+      "learning_rate": 0.0002,
+      "loss": 1.2733,
+      "step": 2590
+    },
+    {
+      "epoch": 0.15793949702344795,
+      "grad_norm": 1.7542141675949097,
+      "learning_rate": 0.0002,
+      "loss": 1.2561,
+      "step": 2600
+    },
+    {
+      "epoch": 0.15854695662738427,
+      "grad_norm": 1.4796082973480225,
+      "learning_rate": 0.0002,
+      "loss": 1.224,
+      "step": 2610
+    },
+    {
+      "epoch": 0.15915441623132062,
+      "grad_norm": 1.5932079553604126,
+      "learning_rate": 0.0002,
+      "loss": 1.2596,
+      "step": 2620
+    },
+    {
+      "epoch": 0.15976187583525694,
+      "grad_norm": 1.3710395097732544,
+      "learning_rate": 0.0002,
+      "loss": 1.3019,
+      "step": 2630
+    },
+    {
+      "epoch": 0.1603693354391933,
+      "grad_norm": 1.393410325050354,
+      "learning_rate": 0.0002,
+      "loss": 1.1923,
+      "step": 2640
+    },
+    {
+      "epoch": 0.16097679504312964,
+      "grad_norm": 1.4769561290740967,
+      "learning_rate": 0.0002,
+      "loss": 1.2147,
+      "step": 2650
+    },
+    {
+      "epoch": 0.16158425464706597,
+      "grad_norm": 1.402292251586914,
+      "learning_rate": 0.0002,
+      "loss": 1.325,
+      "step": 2660
+    },
+    {
+      "epoch": 0.16219171425100232,
+      "grad_norm": 1.546512246131897,
+      "learning_rate": 0.0002,
+      "loss": 1.2197,
+      "step": 2670
+    },
+    {
+      "epoch": 0.16279917385493864,
+      "grad_norm": 1.724507451057434,
+      "learning_rate": 0.0002,
+      "loss": 1.2098,
+      "step": 2680
+    },
+    {
+      "epoch": 0.163406633458875,
+      "grad_norm": 1.4906692504882812,
+      "learning_rate": 0.0002,
+      "loss": 1.2176,
+      "step": 2690
+    },
+    {
+      "epoch": 0.1640140930628113,
+      "grad_norm": 1.510833501815796,
+      "learning_rate": 0.0002,
+      "loss": 1.2124,
+      "step": 2700
+    },
+    {
+      "epoch": 0.16462155266674766,
+      "grad_norm": 1.6878187656402588,
+      "learning_rate": 0.0002,
+      "loss": 1.1886,
+      "step": 2710
+    },
+    {
+      "epoch": 0.165229012270684,
+      "grad_norm": 1.2423148155212402,
+      "learning_rate": 0.0002,
+      "loss": 1.2933,
+      "step": 2720
+    },
+    {
+      "epoch": 0.16583647187462033,
+      "grad_norm": 1.377028465270996,
+      "learning_rate": 0.0002,
+      "loss": 1.196,
+      "step": 2730
+    },
+    {
+      "epoch": 0.16644393147855668,
+      "grad_norm": 1.6662096977233887,
+      "learning_rate": 0.0002,
+      "loss": 1.2769,
+      "step": 2740
+    },
+    {
+      "epoch": 0.167051391082493,
+      "grad_norm": 1.7372788190841675,
+      "learning_rate": 0.0002,
+      "loss": 1.1744,
+      "step": 2750
+    },
+    {
+      "epoch": 0.16765885068642936,
+      "grad_norm": 1.7153924703598022,
+      "learning_rate": 0.0002,
+      "loss": 1.2828,
+      "step": 2760
+    },
+    {
+      "epoch": 0.16826631029036568,
+      "grad_norm": 1.474407434463501,
+      "learning_rate": 0.0002,
+      "loss": 1.118,
+      "step": 2770
+    },
+    {
+      "epoch": 0.16887376989430203,
+      "grad_norm": 1.4870332479476929,
+      "learning_rate": 0.0002,
+      "loss": 1.1382,
+      "step": 2780
+    },
+    {
+      "epoch": 0.16948122949823838,
+      "grad_norm": 1.3868566751480103,
+      "learning_rate": 0.0002,
+      "loss": 1.1986,
+      "step": 2790
+    },
+    {
+      "epoch": 0.1700886891021747,
+      "grad_norm": 1.3732783794403076,
+      "learning_rate": 0.0002,
+      "loss": 1.2488,
+      "step": 2800
+    },
+    {
+      "epoch": 0.17069614870611105,
+      "grad_norm": 1.3792394399642944,
+      "learning_rate": 0.0002,
+      "loss": 1.2477,
+      "step": 2810
+    },
+    {
+      "epoch": 0.17130360831004737,
+      "grad_norm": 1.7015180587768555,
+      "learning_rate": 0.0002,
+      "loss": 1.2269,
+      "step": 2820
+    },
+    {
+      "epoch": 0.17191106791398372,
+      "grad_norm": 1.3385919332504272,
+      "learning_rate": 0.0002,
+      "loss": 1.2585,
+      "step": 2830
+    },
+    {
+      "epoch": 0.17251852751792005,
+      "grad_norm": 1.7190513610839844,
+      "learning_rate": 0.0002,
+      "loss": 1.1882,
+      "step": 2840
+    },
+    {
+      "epoch": 0.1731259871218564,
+      "grad_norm": 1.4032402038574219,
+      "learning_rate": 0.0002,
+      "loss": 1.0767,
+      "step": 2850
+    },
+    {
+      "epoch": 0.17373344672579275,
+      "grad_norm": 1.4926598072052002,
+      "learning_rate": 0.0002,
+      "loss": 1.2304,
+      "step": 2860
+    },
+    {
+      "epoch": 0.17434090632972907,
+      "grad_norm": 1.3792085647583008,
+      "learning_rate": 0.0002,
+      "loss": 1.209,
+      "step": 2870
+    },
+    {
+      "epoch": 0.17494836593366542,
+      "grad_norm": 1.5118519067764282,
+      "learning_rate": 0.0002,
+      "loss": 1.2282,
+      "step": 2880
+    },
+    {
+      "epoch": 0.17555582553760174,
+      "grad_norm": 1.3079215288162231,
+      "learning_rate": 0.0002,
+      "loss": 1.2303,
+      "step": 2890
+    },
+    {
+      "epoch": 0.1761632851415381,
+      "grad_norm": 1.4393481016159058,
+      "learning_rate": 0.0002,
+      "loss": 1.2233,
+      "step": 2900
+    },
+    {
+      "epoch": 0.17677074474547444,
+      "grad_norm": 1.6815558671951294,
+      "learning_rate": 0.0002,
+      "loss": 1.247,
+      "step": 2910
+    },
+    {
+      "epoch": 0.17737820434941076,
+      "grad_norm": 1.487924337387085,
+      "learning_rate": 0.0002,
+      "loss": 1.2744,
+      "step": 2920
+    },
+    {
+      "epoch": 0.1779856639533471,
+      "grad_norm": 1.4145201444625854,
+      "learning_rate": 0.0002,
+      "loss": 1.1545,
+      "step": 2930
+    },
+    {
+      "epoch": 0.17859312355728343,
+      "grad_norm": 1.705073595046997,
+      "learning_rate": 0.0002,
+      "loss": 1.2912,
+      "step": 2940
+    },
+    {
+      "epoch": 0.17920058316121978,
+      "grad_norm": 1.6041674613952637,
+      "learning_rate": 0.0002,
+      "loss": 1.2128,
+      "step": 2950
+    },
+    {
+      "epoch": 0.1798080427651561,
+      "grad_norm": 1.3653987646102905,
+      "learning_rate": 0.0002,
+      "loss": 1.259,
+      "step": 2960
+    },
+    {
+      "epoch": 0.18041550236909246,
+      "grad_norm": 1.7545150518417358,
+      "learning_rate": 0.0002,
+      "loss": 1.2867,
+      "step": 2970
+    },
+    {
+      "epoch": 0.1810229619730288,
+      "grad_norm": 1.3169398307800293,
+      "learning_rate": 0.0002,
+      "loss": 1.2008,
+      "step": 2980
+    },
+    {
+      "epoch": 0.18163042157696513,
+      "grad_norm": 1.5852036476135254,
+      "learning_rate": 0.0002,
+      "loss": 1.1975,
+      "step": 2990
+    },
+    {
+      "epoch": 0.18223788118090148,
+      "grad_norm": 1.4632734060287476,
+      "learning_rate": 0.0002,
+      "loss": 1.3247,
+      "step": 3000
+    },
+    {
+      "epoch": 0.1828453407848378,
+      "grad_norm": 1.486884593963623,
+      "learning_rate": 0.0002,
+      "loss": 1.2443,
+      "step": 3010
+    },
+    {
+      "epoch": 0.18345280038877415,
+      "grad_norm": 1.5297883749008179,
+      "learning_rate": 0.0002,
+      "loss": 1.2453,
+      "step": 3020
+    },
+    {
+      "epoch": 0.18406025999271047,
+      "grad_norm": 1.488099455833435,
+      "learning_rate": 0.0002,
+      "loss": 1.2177,
+      "step": 3030
+    },
+    {
+      "epoch": 0.18466771959664682,
+      "grad_norm": 1.335958480834961,
+      "learning_rate": 0.0002,
+      "loss": 1.2499,
+      "step": 3040
+    },
+    {
+      "epoch": 0.18527517920058317,
+      "grad_norm": 1.5462721586227417,
+      "learning_rate": 0.0002,
+      "loss": 1.2287,
+      "step": 3050
+    },
+    {
+      "epoch": 0.1858826388045195,
+      "grad_norm": 1.4703840017318726,
+      "learning_rate": 0.0002,
+      "loss": 1.2306,
+      "step": 3060
+    },
+    {
+      "epoch": 0.18649009840845585,
+      "grad_norm": 1.4151811599731445,
+      "learning_rate": 0.0002,
+      "loss": 1.2268,
+      "step": 3070
+    },
+    {
+      "epoch": 0.18709755801239217,
+      "grad_norm": 1.5566822290420532,
+      "learning_rate": 0.0002,
+      "loss": 1.2039,
+      "step": 3080
+    },
+    {
+      "epoch": 0.18770501761632852,
+      "grad_norm": 1.4005810022354126,
+      "learning_rate": 0.0002,
+      "loss": 1.1963,
+      "step": 3090
+    },
+    {
+      "epoch": 0.18831247722026484,
+      "grad_norm": 1.4320882558822632,
+      "learning_rate": 0.0002,
+      "loss": 1.2648,
+      "step": 3100
+    },
+    {
+      "epoch": 0.1889199368242012,
+      "grad_norm": 1.3715740442276,
+      "learning_rate": 0.0002,
+      "loss": 1.2178,
+      "step": 3110
+    },
+    {
+      "epoch": 0.18952739642813754,
+      "grad_norm": 1.670812964439392,
+      "learning_rate": 0.0002,
+      "loss": 1.1718,
+      "step": 3120
+    },
+    {
+      "epoch": 0.19013485603207386,
+      "grad_norm": 1.5320864915847778,
+      "learning_rate": 0.0002,
+      "loss": 1.1748,
+      "step": 3130
+    },
+    {
+      "epoch": 0.1907423156360102,
+      "grad_norm": 1.5839077234268188,
+      "learning_rate": 0.0002,
+      "loss": 1.2678,
+      "step": 3140
+    },
+    {
+      "epoch": 0.19134977523994653,
+      "grad_norm": 1.5966272354125977,
+      "learning_rate": 0.0002,
+      "loss": 1.24,
+      "step": 3150
+    },
+    {
+      "epoch": 0.19195723484388288,
+      "grad_norm": 1.4464362859725952,
+      "learning_rate": 0.0002,
+      "loss": 1.2149,
+      "step": 3160
+    },
+    {
+      "epoch": 0.1925646944478192,
+      "grad_norm": 1.392284631729126,
+      "learning_rate": 0.0002,
+      "loss": 1.2857,
+      "step": 3170
+    },
+    {
+      "epoch": 0.19317215405175556,
+      "grad_norm": 1.4531803131103516,
+      "learning_rate": 0.0002,
+      "loss": 1.1909,
+      "step": 3180
+    },
+    {
+      "epoch": 0.1937796136556919,
+      "grad_norm": 1.431835412979126,
+      "learning_rate": 0.0002,
+      "loss": 1.2081,
+      "step": 3190
+    },
+    {
+      "epoch": 0.19438707325962823,
+      "grad_norm": 1.4906381368637085,
+      "learning_rate": 0.0002,
+      "loss": 1.2071,
+      "step": 3200
+    },
+    {
+      "epoch": 0.19499453286356458,
+      "grad_norm": 1.5458217859268188,
+      "learning_rate": 0.0002,
+      "loss": 1.2037,
+      "step": 3210
+    },
+    {
+      "epoch": 0.1956019924675009,
+      "grad_norm": 1.4542298316955566,
+      "learning_rate": 0.0002,
+      "loss": 1.238,
+      "step": 3220
+    },
+    {
+      "epoch": 0.19620945207143725,
+      "grad_norm": 1.9179754257202148,
+      "learning_rate": 0.0002,
+      "loss": 1.2077,
+      "step": 3230
+    },
+    {
+      "epoch": 0.1968169116753736,
+      "grad_norm": 1.3963453769683838,
+      "learning_rate": 0.0002,
+      "loss": 1.2329,
+      "step": 3240
+    },
+    {
+      "epoch": 0.19742437127930992,
+      "grad_norm": 1.565443754196167,
+      "learning_rate": 0.0002,
+      "loss": 1.2155,
+      "step": 3250
+    },
+    {
+      "epoch": 0.19803183088324627,
+      "grad_norm": 1.4760220050811768,
+      "learning_rate": 0.0002,
+      "loss": 1.2112,
+      "step": 3260
+    },
+    {
+      "epoch": 0.1986392904871826,
+      "grad_norm": 1.324994444847107,
+      "learning_rate": 0.0002,
+      "loss": 1.2491,
+      "step": 3270
+    },
+    {
+      "epoch": 0.19924675009111895,
+      "grad_norm": 1.2665252685546875,
+      "learning_rate": 0.0002,
+      "loss": 1.2572,
+      "step": 3280
+    },
+    {
+      "epoch": 0.19985420969505527,
+      "grad_norm": 1.7519789934158325,
+      "learning_rate": 0.0002,
+      "loss": 1.2012,
+      "step": 3290
+    },
+    {
+      "epoch": 0.20046166929899162,
+      "grad_norm": 1.5129839181900024,
+      "learning_rate": 0.0002,
+      "loss": 1.2935,
+      "step": 3300
+    },
+    {
+      "epoch": 0.20106912890292797,
+      "grad_norm": 1.7140570878982544,
+      "learning_rate": 0.0002,
+      "loss": 1.1867,
+      "step": 3310
+    },
+    {
+      "epoch": 0.2016765885068643,
+      "grad_norm": 1.5850943326950073,
+      "learning_rate": 0.0002,
+      "loss": 1.214,
+      "step": 3320
+    },
+    {
+      "epoch": 0.20228404811080064,
+      "grad_norm": 1.4112968444824219,
+      "learning_rate": 0.0002,
+      "loss": 1.1865,
+      "step": 3330
+    },
+    {
+      "epoch": 0.20289150771473696,
+      "grad_norm": 1.5837844610214233,
+      "learning_rate": 0.0002,
+      "loss": 1.2948,
+      "step": 3340
+    },
+    {
+      "epoch": 0.2034989673186733,
+      "grad_norm": 1.4747986793518066,
+      "learning_rate": 0.0002,
+      "loss": 1.2132,
+      "step": 3350
+    },
+    {
+      "epoch": 0.20410642692260964,
+      "grad_norm": 1.5337995290756226,
+      "learning_rate": 0.0002,
+      "loss": 1.2326,
+      "step": 3360
+    },
+    {
+      "epoch": 0.20471388652654599,
+      "grad_norm": 1.5909236669540405,
+      "learning_rate": 0.0002,
+      "loss": 1.1964,
+      "step": 3370
+    },
+    {
+      "epoch": 0.20532134613048234,
+      "grad_norm": 1.4380537271499634,
+      "learning_rate": 0.0002,
+      "loss": 1.1912,
+      "step": 3380
+    },
+    {
+      "epoch": 0.20592880573441866,
+      "grad_norm": 1.317665934562683,
+      "learning_rate": 0.0002,
+      "loss": 1.2559,
+      "step": 3390
+    },
+    {
+      "epoch": 0.206536265338355,
+      "grad_norm": 1.4082776308059692,
+      "learning_rate": 0.0002,
+      "loss": 1.2575,
+      "step": 3400
+    },
+    {
+      "epoch": 0.20714372494229133,
+      "grad_norm": 1.5903863906860352,
+      "learning_rate": 0.0002,
+      "loss": 1.2513,
+      "step": 3410
+    },
+    {
+      "epoch": 0.20775118454622768,
+      "grad_norm": 1.2638167142868042,
+      "learning_rate": 0.0002,
+      "loss": 1.1693,
+      "step": 3420
+    },
+    {
+      "epoch": 0.208358644150164,
+      "grad_norm": 1.507632851600647,
+      "learning_rate": 0.0002,
+      "loss": 1.2523,
+      "step": 3430
+    },
+    {
+      "epoch": 0.20896610375410035,
+      "grad_norm": 1.4946129322052002,
+      "learning_rate": 0.0002,
+      "loss": 1.1957,
+      "step": 3440
+    },
+    {
+      "epoch": 0.2095735633580367,
+      "grad_norm": 1.4943633079528809,
+      "learning_rate": 0.0002,
+      "loss": 1.2256,
+      "step": 3450
+    },
+    {
+      "epoch": 0.21018102296197302,
+      "grad_norm": 1.8334934711456299,
+      "learning_rate": 0.0002,
+      "loss": 1.2509,
+      "step": 3460
+    },
+    {
+      "epoch": 0.21078848256590937,
+      "grad_norm": 1.2962932586669922,
+      "learning_rate": 0.0002,
+      "loss": 1.1625,
+      "step": 3470
+    },
+    {
+      "epoch": 0.2113959421698457,
+      "grad_norm": 1.3973207473754883,
+      "learning_rate": 0.0002,
+      "loss": 1.2567,
+      "step": 3480
+    },
+    {
+      "epoch": 0.21200340177378205,
+      "grad_norm": 1.420222520828247,
+      "learning_rate": 0.0002,
+      "loss": 1.2536,
+      "step": 3490
+    },
+    {
+      "epoch": 0.21261086137771837,
+      "grad_norm": 1.6628416776657104,
+      "learning_rate": 0.0002,
+      "loss": 1.3032,
+      "step": 3500
+    },
+    {
+      "epoch": 0.21321832098165472,
+      "grad_norm": 1.5589535236358643,
+      "learning_rate": 0.0002,
+      "loss": 1.2311,
+      "step": 3510
+    },
+    {
+      "epoch": 0.21382578058559107,
+      "grad_norm": 1.5037486553192139,
+      "learning_rate": 0.0002,
+      "loss": 1.1963,
+      "step": 3520
+    },
+    {
+      "epoch": 0.2144332401895274,
+      "grad_norm": 1.2449471950531006,
+      "learning_rate": 0.0002,
+      "loss": 1.1782,
+      "step": 3530
+    },
+    {
+      "epoch": 0.21504069979346374,
+      "grad_norm": 1.5314222574234009,
+      "learning_rate": 0.0002,
+      "loss": 1.1905,
+      "step": 3540
+    },
+    {
+      "epoch": 0.21564815939740006,
+      "grad_norm": 1.4641014337539673,
+      "learning_rate": 0.0002,
+      "loss": 1.2713,
+      "step": 3550
+    },
+    {
+      "epoch": 0.2162556190013364,
+      "grad_norm": 1.4204001426696777,
+      "learning_rate": 0.0002,
+      "loss": 1.1789,
+      "step": 3560
+    },
+    {
+      "epoch": 0.21686307860527274,
+      "grad_norm": 1.7577894926071167,
+      "learning_rate": 0.0002,
+      "loss": 1.1719,
+      "step": 3570
+    },
+    {
+      "epoch": 0.21747053820920909,
+      "grad_norm": 1.3610202074050903,
+      "learning_rate": 0.0002,
+      "loss": 1.2508,
+      "step": 3580
+    },
+    {
+      "epoch": 0.21807799781314544,
+      "grad_norm": 1.416024923324585,
+      "learning_rate": 0.0002,
+      "loss": 1.2606,
+      "step": 3590
+    },
+    {
+      "epoch": 0.21868545741708176,
+      "grad_norm": 1.3644566535949707,
+      "learning_rate": 0.0002,
+      "loss": 1.2205,
+      "step": 3600
+    },
+    {
+      "epoch": 0.2192929170210181,
+      "grad_norm": 1.3592606782913208,
+      "learning_rate": 0.0002,
+      "loss": 1.2563,
+      "step": 3610
+    },
+    {
+      "epoch": 0.21990037662495443,
+      "grad_norm": 1.3086717128753662,
+      "learning_rate": 0.0002,
+      "loss": 1.2059,
+      "step": 3620
+    },
+    {
+      "epoch": 0.22050783622889078,
+      "grad_norm": 1.4019612073898315,
+      "learning_rate": 0.0002,
+      "loss": 1.1962,
+      "step": 3630
+    },
+    {
+      "epoch": 0.22111529583282713,
+      "grad_norm": 1.3697004318237305,
+      "learning_rate": 0.0002,
+      "loss": 1.2538,
+      "step": 3640
+    },
+    {
+      "epoch": 0.22172275543676345,
+      "grad_norm": 1.4234853982925415,
+      "learning_rate": 0.0002,
+      "loss": 1.1667,
+      "step": 3650
+    },
+    {
+      "epoch": 0.2223302150406998,
+      "grad_norm": 1.3091281652450562,
+      "learning_rate": 0.0002,
+      "loss": 1.2034,
+      "step": 3660
+    },
+    {
+      "epoch": 0.22293767464463612,
+      "grad_norm": 1.3341374397277832,
+      "learning_rate": 0.0002,
+      "loss": 1.2284,
+      "step": 3670
+    },
+    {
+      "epoch": 0.22354513424857247,
+      "grad_norm": 1.5380843877792358,
+      "learning_rate": 0.0002,
+      "loss": 1.2105,
+      "step": 3680
+    },
+    {
+      "epoch": 0.2241525938525088,
+      "grad_norm": 1.5012211799621582,
+      "learning_rate": 0.0002,
+      "loss": 1.213,
+      "step": 3690
+    },
+    {
+      "epoch": 0.22476005345644515,
+      "grad_norm": 1.344851016998291,
+      "learning_rate": 0.0002,
+      "loss": 1.1632,
+      "step": 3700
+    },
+    {
+      "epoch": 0.2253675130603815,
+      "grad_norm": 1.4722944498062134,
+      "learning_rate": 0.0002,
+      "loss": 1.2099,
+      "step": 3710
+    },
+    {
+      "epoch": 0.22597497266431782,
+      "grad_norm": 1.44810950756073,
+      "learning_rate": 0.0002,
+      "loss": 1.234,
+      "step": 3720
+    },
+    {
+      "epoch": 0.22658243226825417,
+      "grad_norm": 1.4535653591156006,
+      "learning_rate": 0.0002,
+      "loss": 1.2002,
+      "step": 3730
+    },
+    {
+      "epoch": 0.2271898918721905,
+      "grad_norm": 1.623395562171936,
+      "learning_rate": 0.0002,
+      "loss": 1.2354,
+      "step": 3740
+    },
+    {
+      "epoch": 0.22779735147612684,
+      "grad_norm": 1.8397951126098633,
+      "learning_rate": 0.0002,
+      "loss": 1.2425,
+      "step": 3750
+    },
+    {
+      "epoch": 0.22840481108006316,
+      "grad_norm": 1.3599379062652588,
+      "learning_rate": 0.0002,
+      "loss": 1.2344,
+      "step": 3760
+    },
+    {
+      "epoch": 0.2290122706839995,
+      "grad_norm": 1.6713752746582031,
+      "learning_rate": 0.0002,
+      "loss": 1.2333,
+      "step": 3770
+    },
+    {
+      "epoch": 0.22961973028793586,
+      "grad_norm": 1.913011074066162,
+      "learning_rate": 0.0002,
+      "loss": 1.2359,
+      "step": 3780
+    },
+    {
+      "epoch": 0.23022718989187219,
+      "grad_norm": 1.4186650514602661,
+      "learning_rate": 0.0002,
+      "loss": 1.2487,
+      "step": 3790
+    },
+    {
+      "epoch": 0.23083464949580854,
+      "grad_norm": 1.673385500907898,
+      "learning_rate": 0.0002,
+      "loss": 1.2817,
+      "step": 3800
+    },
+    {
+      "epoch": 0.23144210909974486,
+      "grad_norm": 1.3863779306411743,
+      "learning_rate": 0.0002,
+      "loss": 1.1805,
+      "step": 3810
+    },
+    {
+      "epoch": 0.2320495687036812,
+      "grad_norm": 1.4821914434432983,
+      "learning_rate": 0.0002,
+      "loss": 1.2159,
+      "step": 3820
+    },
+    {
+      "epoch": 0.23265702830761753,
+      "grad_norm": 1.27168869972229,
+      "learning_rate": 0.0002,
+      "loss": 1.236,
+      "step": 3830
+    },
+    {
+      "epoch": 0.23326448791155388,
+      "grad_norm": 1.464900255203247,
+      "learning_rate": 0.0002,
+      "loss": 1.2573,
+      "step": 3840
+    },
+    {
+      "epoch": 0.23387194751549023,
+      "grad_norm": 1.5601277351379395,
+      "learning_rate": 0.0002,
+      "loss": 1.1867,
+      "step": 3850
+    },
+    {
+      "epoch": 0.23447940711942655,
+      "grad_norm": 1.4300495386123657,
+      "learning_rate": 0.0002,
+      "loss": 1.2781,
+      "step": 3860
+    },
+    {
+      "epoch": 0.2350868667233629,
+      "grad_norm": 1.6343556642532349,
+      "learning_rate": 0.0002,
+      "loss": 1.1652,
+      "step": 3870
+    },
+    {
+      "epoch": 0.23569432632729923,
+      "grad_norm": 1.391050934791565,
+      "learning_rate": 0.0002,
+      "loss": 1.2798,
+      "step": 3880
+    },
+    {
+      "epoch": 0.23630178593123558,
+      "grad_norm": 1.9890904426574707,
+      "learning_rate": 0.0002,
+      "loss": 1.2878,
+      "step": 3890
+    },
+    {
+      "epoch": 0.2369092455351719,
+      "grad_norm": 1.3645399808883667,
+      "learning_rate": 0.0002,
+      "loss": 1.2392,
+      "step": 3900
+    },
+    {
+      "epoch": 0.23751670513910825,
+      "grad_norm": 1.3967126607894897,
+      "learning_rate": 0.0002,
+      "loss": 1.2884,
+      "step": 3910
+    },
+    {
+      "epoch": 0.2381241647430446,
+      "grad_norm": 1.3073115348815918,
+      "learning_rate": 0.0002,
+      "loss": 1.2548,
+      "step": 3920
+    },
+    {
+      "epoch": 0.23873162434698092,
+      "grad_norm": 1.488996148109436,
+      "learning_rate": 0.0002,
+      "loss": 1.1664,
+      "step": 3930
+    },
+    {
+      "epoch": 0.23933908395091727,
+      "grad_norm": 1.3627840280532837,
+      "learning_rate": 0.0002,
+      "loss": 1.1944,
+      "step": 3940
+    },
+    {
+      "epoch": 0.2399465435548536,
+      "grad_norm": 1.4558093547821045,
+      "learning_rate": 0.0002,
+      "loss": 1.1846,
+      "step": 3950
+    },
+    {
+      "epoch": 0.24055400315878994,
+      "grad_norm": 1.4579048156738281,
+      "learning_rate": 0.0002,
+      "loss": 1.2228,
+      "step": 3960
+    },
+    {
+      "epoch": 0.2411614627627263,
+      "grad_norm": 1.248146891593933,
+      "learning_rate": 0.0002,
+      "loss": 1.2166,
+      "step": 3970
+    },
+    {
+      "epoch": 0.24176892236666261,
+      "grad_norm": 1.358766794204712,
+      "learning_rate": 0.0002,
+      "loss": 1.1992,
+      "step": 3980
+    },
+    {
+      "epoch": 0.24237638197059896,
+      "grad_norm": 1.4376319646835327,
+      "learning_rate": 0.0002,
+      "loss": 1.2631,
+      "step": 3990
+    },
+    {
+      "epoch": 0.2429838415745353,
+      "grad_norm": 1.5409590005874634,
+      "learning_rate": 0.0002,
+      "loss": 1.2061,
+      "step": 4000
+    },
+    {
+      "epoch": 0.24359130117847164,
+      "grad_norm": 1.5955286026000977,
+      "learning_rate": 0.0002,
+      "loss": 1.263,
+      "step": 4010
+    },
+    {
+      "epoch": 0.24419876078240796,
+      "grad_norm": 1.618519902229309,
+      "learning_rate": 0.0002,
+      "loss": 1.2471,
+      "step": 4020
+    },
+    {
+      "epoch": 0.2448062203863443,
+      "grad_norm": 1.4988939762115479,
+      "learning_rate": 0.0002,
+      "loss": 1.2367,
+      "step": 4030
+    },
+    {
+      "epoch": 0.24541367999028066,
+      "grad_norm": 1.4525686502456665,
+      "learning_rate": 0.0002,
+      "loss": 1.2131,
+      "step": 4040
+    },
+    {
+      "epoch": 0.24602113959421698,
+      "grad_norm": 1.5270054340362549,
+      "learning_rate": 0.0002,
+      "loss": 1.2156,
+      "step": 4050
+    },
+    {
+      "epoch": 0.24662859919815333,
+      "grad_norm": 1.3893275260925293,
+      "learning_rate": 0.0002,
+      "loss": 1.1854,
+      "step": 4060
+    },
+    {
+      "epoch": 0.24723605880208965,
+      "grad_norm": 1.5147534608840942,
+      "learning_rate": 0.0002,
+      "loss": 1.1289,
+      "step": 4070
+    },
+    {
+      "epoch": 0.247843518406026,
+      "grad_norm": 1.242431879043579,
+      "learning_rate": 0.0002,
+      "loss": 1.2152,
+      "step": 4080
+    },
+    {
+      "epoch": 0.24845097800996233,
+      "grad_norm": 1.593482255935669,
+      "learning_rate": 0.0002,
+      "loss": 1.2299,
+      "step": 4090
+    },
+    {
+      "epoch": 0.24905843761389868,
+      "grad_norm": 1.4139094352722168,
+      "learning_rate": 0.0002,
+      "loss": 1.2517,
+      "step": 4100
+    },
+    {
+      "epoch": 0.24966589721783503,
+      "grad_norm": 1.5035349130630493,
+      "learning_rate": 0.0002,
+      "loss": 1.1643,
+      "step": 4110
+    },
+    {
+      "epoch": 0.2502733568217714,
+      "grad_norm": 1.4908943176269531,
+      "learning_rate": 0.0002,
+      "loss": 1.2632,
+      "step": 4120
+    },
+    {
+      "epoch": 0.2508808164257077,
+      "grad_norm": 1.524407982826233,
+      "learning_rate": 0.0002,
+      "loss": 1.2487,
+      "step": 4130
+    },
+    {
+      "epoch": 0.251488276029644,
+      "grad_norm": 1.6802865266799927,
+      "learning_rate": 0.0002,
+      "loss": 1.324,
+      "step": 4140
+    },
+    {
+      "epoch": 0.25209573563358034,
+      "grad_norm": 1.4685097932815552,
+      "learning_rate": 0.0002,
+      "loss": 1.2453,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2527031952375167,
+      "grad_norm": 1.4716459512710571,
+      "learning_rate": 0.0002,
+      "loss": 1.2216,
+      "step": 4160
+    },
+    {
+      "epoch": 0.25331065484145304,
+      "grad_norm": 1.6466096639633179,
+      "learning_rate": 0.0002,
+      "loss": 1.2248,
+      "step": 4170
+    },
+    {
+      "epoch": 0.25391811444538936,
+      "grad_norm": 1.4192179441452026,
+      "learning_rate": 0.0002,
+      "loss": 1.1572,
+      "step": 4180
+    },
+    {
+      "epoch": 0.25452557404932574,
+      "grad_norm": 1.2112548351287842,
+      "learning_rate": 0.0002,
+      "loss": 1.1588,
+      "step": 4190
+    },
+    {
+      "epoch": 0.25513303365326206,
+      "grad_norm": 1.4236078262329102,
+      "learning_rate": 0.0002,
+      "loss": 1.2202,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2557404932571984,
+      "grad_norm": 1.4484710693359375,
+      "learning_rate": 0.0002,
+      "loss": 1.2102,
+      "step": 4210
+    },
+    {
+      "epoch": 0.2563479528611347,
+      "grad_norm": 1.541420340538025,
+      "learning_rate": 0.0002,
+      "loss": 1.2311,
+      "step": 4220
+    },
+    {
+      "epoch": 0.2569554124650711,
+      "grad_norm": 1.3980011940002441,
+      "learning_rate": 0.0002,
+      "loss": 1.2202,
+      "step": 4230
+    },
+    {
+      "epoch": 0.2575628720690074,
+      "grad_norm": 1.7746334075927734,
+      "learning_rate": 0.0002,
+      "loss": 1.2233,
+      "step": 4240
+    },
+    {
+      "epoch": 0.25817033167294373,
+      "grad_norm": 1.5459847450256348,
+      "learning_rate": 0.0002,
+      "loss": 1.1839,
+      "step": 4250
+    },
+    {
+      "epoch": 0.2587777912768801,
+      "grad_norm": 1.3682315349578857,
+      "learning_rate": 0.0002,
+      "loss": 1.1987,
+      "step": 4260
+    },
+    {
+      "epoch": 0.25938525088081643,
+      "grad_norm": 1.296205759048462,
+      "learning_rate": 0.0002,
+      "loss": 1.1579,
+      "step": 4270
+    },
+    {
+      "epoch": 0.25999271048475275,
+      "grad_norm": 1.3495019674301147,
+      "learning_rate": 0.0002,
+      "loss": 1.2146,
+      "step": 4280
+    },
+    {
+      "epoch": 0.2606001700886891,
+      "grad_norm": 1.4230709075927734,
+      "learning_rate": 0.0002,
+      "loss": 1.2008,
+      "step": 4290
+    },
+    {
+      "epoch": 0.26120762969262545,
+      "grad_norm": 1.2394111156463623,
+      "learning_rate": 0.0002,
+      "loss": 1.119,
+      "step": 4300
+    },
+    {
+      "epoch": 0.2618150892965618,
+      "grad_norm": 1.381503939628601,
+      "learning_rate": 0.0002,
+      "loss": 1.2142,
+      "step": 4310
+    },
+    {
+      "epoch": 0.2624225489004981,
+      "grad_norm": 1.391729712486267,
+      "learning_rate": 0.0002,
+      "loss": 1.1233,
+      "step": 4320
+    },
+    {
+      "epoch": 0.2630300085044345,
+      "grad_norm": 1.4309793710708618,
+      "learning_rate": 0.0002,
+      "loss": 1.2424,
+      "step": 4330
+    },
+    {
+      "epoch": 0.2636374681083708,
+      "grad_norm": 1.658195972442627,
+      "learning_rate": 0.0002,
+      "loss": 1.2035,
+      "step": 4340
+    },
+    {
+      "epoch": 0.2642449277123071,
+      "grad_norm": 1.5374689102172852,
+      "learning_rate": 0.0002,
+      "loss": 1.2021,
+      "step": 4350
+    },
+    {
+      "epoch": 0.26485238731624344,
+      "grad_norm": 1.6735734939575195,
+      "learning_rate": 0.0002,
+      "loss": 1.1996,
+      "step": 4360
+    },
+    {
+      "epoch": 0.2654598469201798,
+      "grad_norm": 1.449633002281189,
+      "learning_rate": 0.0002,
+      "loss": 1.1782,
+      "step": 4370
+    },
+    {
+      "epoch": 0.26606730652411614,
+      "grad_norm": 1.3424811363220215,
+      "learning_rate": 0.0002,
+      "loss": 1.1324,
+      "step": 4380
+    },
+    {
+      "epoch": 0.26667476612805247,
+      "grad_norm": 1.6032854318618774,
+      "learning_rate": 0.0002,
+      "loss": 1.2865,
+      "step": 4390
+    },
+    {
+      "epoch": 0.26728222573198884,
+      "grad_norm": 1.5517349243164062,
+      "learning_rate": 0.0002,
+      "loss": 1.1295,
+      "step": 4400
+    },
+    {
+      "epoch": 0.26788968533592517,
+      "grad_norm": 1.4976743459701538,
+      "learning_rate": 0.0002,
+      "loss": 1.2562,
+      "step": 4410
+    },
+    {
+      "epoch": 0.2684971449398615,
+      "grad_norm": 1.352860450744629,
+      "learning_rate": 0.0002,
+      "loss": 1.2158,
+      "step": 4420
+    },
+    {
+      "epoch": 0.26910460454379787,
+      "grad_norm": 1.5025726556777954,
+      "learning_rate": 0.0002,
+      "loss": 1.1519,
+      "step": 4430
+    },
+    {
+      "epoch": 0.2697120641477342,
+      "grad_norm": 1.517626404762268,
+      "learning_rate": 0.0002,
+      "loss": 1.1947,
+      "step": 4440
+    },
+    {
+      "epoch": 0.2703195237516705,
+      "grad_norm": 1.6162688732147217,
+      "learning_rate": 0.0002,
+      "loss": 1.2132,
+      "step": 4450
+    },
+    {
+      "epoch": 0.27092698335560683,
+      "grad_norm": 1.6288328170776367,
+      "learning_rate": 0.0002,
+      "loss": 1.2375,
+      "step": 4460
+    },
+    {
+      "epoch": 0.2715344429595432,
+      "grad_norm": 1.4691635370254517,
+      "learning_rate": 0.0002,
+      "loss": 1.2204,
+      "step": 4470
+    },
+    {
+      "epoch": 0.27214190256347953,
+      "grad_norm": 1.4654793739318848,
+      "learning_rate": 0.0002,
+      "loss": 1.2788,
+      "step": 4480
+    },
+    {
+      "epoch": 0.27274936216741585,
+      "grad_norm": 1.381338119506836,
+      "learning_rate": 0.0002,
+      "loss": 1.1982,
+      "step": 4490
+    },
+    {
+      "epoch": 0.27335682177135223,
+      "grad_norm": 1.3719398975372314,
+      "learning_rate": 0.0002,
+      "loss": 1.1435,
+      "step": 4500
+    },
+    {
+      "epoch": 0.27396428137528855,
+      "grad_norm": 1.4959717988967896,
+      "learning_rate": 0.0002,
+      "loss": 1.1868,
+      "step": 4510
+    },
+    {
+      "epoch": 0.2745717409792249,
+      "grad_norm": 1.6549965143203735,
+      "learning_rate": 0.0002,
+      "loss": 1.2443,
+      "step": 4520
+    },
+    {
+      "epoch": 0.2751792005831612,
+      "grad_norm": 1.391357183456421,
+      "learning_rate": 0.0002,
+      "loss": 1.2828,
+      "step": 4530
+    },
+    {
+      "epoch": 0.2757866601870976,
+      "grad_norm": 1.360917329788208,
+      "learning_rate": 0.0002,
+      "loss": 1.1248,
+      "step": 4540
+    },
+    {
+      "epoch": 0.2763941197910339,
+      "grad_norm": 1.5073282718658447,
+      "learning_rate": 0.0002,
+      "loss": 1.2416,
+      "step": 4550
+    },
+    {
+      "epoch": 0.2770015793949702,
+      "grad_norm": 1.4248418807983398,
+      "learning_rate": 0.0002,
+      "loss": 1.2142,
+      "step": 4560
+    },
+    {
+      "epoch": 0.2776090389989066,
+      "grad_norm": 1.4884262084960938,
+      "learning_rate": 0.0002,
+      "loss": 1.2224,
+      "step": 4570
+    },
+    {
+      "epoch": 0.2782164986028429,
+      "grad_norm": 1.4659768342971802,
+      "learning_rate": 0.0002,
+      "loss": 1.1661,
+      "step": 4580
+    },
+    {
+      "epoch": 0.27882395820677924,
+      "grad_norm": 1.578492283821106,
+      "learning_rate": 0.0002,
+      "loss": 1.2731,
+      "step": 4590
+    },
+    {
+      "epoch": 0.27943141781071557,
+      "grad_norm": 1.4244043827056885,
+      "learning_rate": 0.0002,
+      "loss": 1.1922,
+      "step": 4600
+    },
+    {
+      "epoch": 0.28003887741465194,
+      "grad_norm": 1.5399266481399536,
+      "learning_rate": 0.0002,
+      "loss": 1.2627,
+      "step": 4610
+    },
+    {
+      "epoch": 0.28064633701858827,
+      "grad_norm": 1.497768759727478,
+      "learning_rate": 0.0002,
+      "loss": 1.2225,
+      "step": 4620
+    },
+    {
+      "epoch": 0.2812537966225246,
+      "grad_norm": 1.5408015251159668,
+      "learning_rate": 0.0002,
+      "loss": 1.2791,
+      "step": 4630
+    },
+    {
+      "epoch": 0.28186125622646097,
+      "grad_norm": 1.5482003688812256,
+      "learning_rate": 0.0002,
+      "loss": 1.199,
+      "step": 4640
+    },
+    {
+      "epoch": 0.2824687158303973,
+      "grad_norm": 1.5323197841644287,
+      "learning_rate": 0.0002,
+      "loss": 1.3071,
+      "step": 4650
+    },
+    {
+      "epoch": 0.2830761754343336,
+      "grad_norm": 1.851555585861206,
+      "learning_rate": 0.0002,
+      "loss": 1.191,
+      "step": 4660
+    },
+    {
+      "epoch": 0.28368363503826993,
+      "grad_norm": 1.4027585983276367,
+      "learning_rate": 0.0002,
+      "loss": 1.1856,
+      "step": 4670
+    },
+    {
+      "epoch": 0.2842910946422063,
+      "grad_norm": 1.3661413192749023,
+      "learning_rate": 0.0002,
+      "loss": 1.193,
+      "step": 4680
+    },
+    {
+      "epoch": 0.28489855424614263,
+      "grad_norm": 1.6890923976898193,
+      "learning_rate": 0.0002,
+      "loss": 1.2026,
+      "step": 4690
+    },
+    {
+      "epoch": 0.28550601385007895,
+      "grad_norm": 1.867311954498291,
+      "learning_rate": 0.0002,
+      "loss": 1.2241,
+      "step": 4700
+    },
+    {
+      "epoch": 0.28611347345401533,
+      "grad_norm": 1.5613819360733032,
+      "learning_rate": 0.0002,
+      "loss": 1.1743,
+      "step": 4710
+    },
+    {
+      "epoch": 0.28672093305795165,
+      "grad_norm": 1.5607540607452393,
+      "learning_rate": 0.0002,
+      "loss": 1.2392,
+      "step": 4720
+    },
+    {
+      "epoch": 0.287328392661888,
+      "grad_norm": 1.5710781812667847,
+      "learning_rate": 0.0002,
+      "loss": 1.2192,
+      "step": 4730
+    },
+    {
+      "epoch": 0.2879358522658243,
+      "grad_norm": 1.3640292882919312,
+      "learning_rate": 0.0002,
+      "loss": 1.145,
+      "step": 4740
+    },
+    {
+      "epoch": 0.2885433118697607,
+      "grad_norm": 1.4605162143707275,
+      "learning_rate": 0.0002,
+      "loss": 1.1938,
+      "step": 4750
+    },
+    {
+      "epoch": 0.289150771473697,
+      "grad_norm": 1.3732244968414307,
+      "learning_rate": 0.0002,
+      "loss": 1.1221,
+      "step": 4760
+    },
+    {
+      "epoch": 0.2897582310776333,
+      "grad_norm": 1.642148494720459,
+      "learning_rate": 0.0002,
+      "loss": 1.2009,
+      "step": 4770
+    },
+    {
+      "epoch": 0.2903656906815697,
+      "grad_norm": 1.5201040506362915,
+      "learning_rate": 0.0002,
+      "loss": 1.1517,
+      "step": 4780
+    },
+    {
+      "epoch": 0.290973150285506,
+      "grad_norm": 1.6133527755737305,
+      "learning_rate": 0.0002,
+      "loss": 1.2239,
+      "step": 4790
+    },
+    {
+      "epoch": 0.29158060988944234,
+      "grad_norm": 1.2963751554489136,
+      "learning_rate": 0.0002,
+      "loss": 1.2808,
+      "step": 4800
+    },
+    {
+      "epoch": 0.29218806949337867,
+      "grad_norm": 1.3992732763290405,
+      "learning_rate": 0.0002,
+      "loss": 1.1852,
+      "step": 4810
+    },
+    {
+      "epoch": 0.29279552909731504,
+      "grad_norm": 1.5011121034622192,
+      "learning_rate": 0.0002,
+      "loss": 1.2235,
+      "step": 4820
+    },
+    {
+      "epoch": 0.29340298870125137,
+      "grad_norm": 1.5714153051376343,
+      "learning_rate": 0.0002,
+      "loss": 1.2199,
+      "step": 4830
+    },
+    {
+      "epoch": 0.2940104483051877,
+      "grad_norm": 1.4812499284744263,
+      "learning_rate": 0.0002,
+      "loss": 1.1979,
+      "step": 4840
+    },
+    {
+      "epoch": 0.29461790790912407,
+      "grad_norm": 1.7770074605941772,
+      "learning_rate": 0.0002,
+      "loss": 1.2795,
+      "step": 4850
+    },
+    {
+      "epoch": 0.2952253675130604,
+      "grad_norm": 1.3813797235488892,
+      "learning_rate": 0.0002,
+      "loss": 1.2694,
+      "step": 4860
+    },
+    {
+      "epoch": 0.2958328271169967,
+      "grad_norm": 1.5236833095550537,
+      "learning_rate": 0.0002,
+      "loss": 1.2333,
+      "step": 4870
+    },
+    {
+      "epoch": 0.29644028672093303,
+      "grad_norm": 1.365063190460205,
+      "learning_rate": 0.0002,
+      "loss": 1.2719,
+      "step": 4880
+    },
+    {
+      "epoch": 0.2970477463248694,
+      "grad_norm": 1.417883276939392,
+      "learning_rate": 0.0002,
+      "loss": 1.1927,
+      "step": 4890
+    },
+    {
+      "epoch": 0.29765520592880573,
+      "grad_norm": 1.625367283821106,
+      "learning_rate": 0.0002,
+      "loss": 1.1459,
+      "step": 4900
+    },
+    {
+      "epoch": 0.29826266553274206,
+      "grad_norm": 1.5433696508407593,
+      "learning_rate": 0.0002,
+      "loss": 1.1908,
+      "step": 4910
+    },
+    {
+      "epoch": 0.29887012513667843,
+      "grad_norm": 1.6701841354370117,
+      "learning_rate": 0.0002,
+      "loss": 1.31,
+      "step": 4920
+    },
+    {
+      "epoch": 0.29947758474061476,
+      "grad_norm": 1.650138258934021,
+      "learning_rate": 0.0002,
+      "loss": 1.2311,
+      "step": 4930
+    },
+    {
+      "epoch": 0.3000850443445511,
+      "grad_norm": 1.2897210121154785,
+      "learning_rate": 0.0002,
+      "loss": 1.2156,
+      "step": 4940
+    },
+    {
+      "epoch": 0.3006925039484874,
+      "grad_norm": 1.449493169784546,
+      "learning_rate": 0.0002,
+      "loss": 1.2398,
+      "step": 4950
+    },
+    {
+      "epoch": 0.3012999635524238,
+      "grad_norm": 1.3318798542022705,
+      "learning_rate": 0.0002,
+      "loss": 1.2725,
+      "step": 4960
+    },
+    {
+      "epoch": 0.3019074231563601,
+      "grad_norm": 1.4735842943191528,
+      "learning_rate": 0.0002,
+      "loss": 1.1351,
+      "step": 4970
+    },
+    {
+      "epoch": 0.3025148827602964,
+      "grad_norm": 1.459076166152954,
+      "learning_rate": 0.0002,
+      "loss": 1.2191,
+      "step": 4980
+    },
+    {
+      "epoch": 0.3031223423642328,
+      "grad_norm": 1.2924869060516357,
+      "learning_rate": 0.0002,
+      "loss": 1.2068,
+      "step": 4990
+    },
+    {
+      "epoch": 0.3037298019681691,
+      "grad_norm": 1.3615398406982422,
+      "learning_rate": 0.0002,
+      "loss": 1.2255,
+      "step": 5000
+    },
+    {
+      "epoch": 0.30433726157210544,
+      "grad_norm": 1.3613234758377075,
+      "learning_rate": 0.0002,
+      "loss": 1.1881,
+      "step": 5010
+    },
+    {
+      "epoch": 0.30494472117604177,
+      "grad_norm": 1.3898018598556519,
+      "learning_rate": 0.0002,
+      "loss": 1.19,
+      "step": 5020
+    },
+    {
+      "epoch": 0.30555218077997814,
+      "grad_norm": 1.5199147462844849,
+      "learning_rate": 0.0002,
+      "loss": 1.3389,
+      "step": 5030
+    },
+    {
+      "epoch": 0.30615964038391447,
+      "grad_norm": 1.2513540983200073,
+      "learning_rate": 0.0002,
+      "loss": 1.2697,
+      "step": 5040
+    },
+    {
+      "epoch": 0.3067670999878508,
+      "grad_norm": 1.4729671478271484,
+      "learning_rate": 0.0002,
+      "loss": 1.2243,
+      "step": 5050
+    },
+    {
+      "epoch": 0.30737455959178717,
+      "grad_norm": 1.490967869758606,
+      "learning_rate": 0.0002,
+      "loss": 1.2126,
+      "step": 5060
+    },
+    {
+      "epoch": 0.3079820191957235,
+      "grad_norm": 1.77041757106781,
+      "learning_rate": 0.0002,
+      "loss": 1.2285,
+      "step": 5070
+    },
+    {
+      "epoch": 0.3085894787996598,
+      "grad_norm": 1.528355360031128,
+      "learning_rate": 0.0002,
+      "loss": 1.2168,
+      "step": 5080
+    },
+    {
+      "epoch": 0.30919693840359613,
+      "grad_norm": 1.2505336999893188,
+      "learning_rate": 0.0002,
+      "loss": 1.2107,
+      "step": 5090
+    },
+    {
+      "epoch": 0.3098043980075325,
+      "grad_norm": 1.5367778539657593,
+      "learning_rate": 0.0002,
+      "loss": 1.3089,
+      "step": 5100
+    },
+    {
+      "epoch": 0.31041185761146883,
+      "grad_norm": 1.4673290252685547,
+      "learning_rate": 0.0002,
+      "loss": 1.2421,
+      "step": 5110
+    },
+    {
+      "epoch": 0.31101931721540516,
+      "grad_norm": 1.20176100730896,
+      "learning_rate": 0.0002,
+      "loss": 1.2087,
+      "step": 5120
+    },
+    {
+      "epoch": 0.31162677681934153,
+      "grad_norm": 1.36702299118042,
+      "learning_rate": 0.0002,
+      "loss": 1.2473,
+      "step": 5130
+    },
+    {
+      "epoch": 0.31223423642327786,
+      "grad_norm": 1.311002254486084,
+      "learning_rate": 0.0002,
+      "loss": 1.2086,
+      "step": 5140
+    },
+    {
+      "epoch": 0.3128416960272142,
+      "grad_norm": 1.5590543746948242,
+      "learning_rate": 0.0002,
+      "loss": 1.2918,
+      "step": 5150
+    },
+    {
+      "epoch": 0.31344915563115056,
+      "grad_norm": 1.7573869228363037,
+      "learning_rate": 0.0002,
+      "loss": 1.2847,
+      "step": 5160
+    },
+    {
+      "epoch": 0.3140566152350869,
+      "grad_norm": 1.5680657625198364,
+      "learning_rate": 0.0002,
+      "loss": 1.2183,
+      "step": 5170
+    },
+    {
+      "epoch": 0.3146640748390232,
+      "grad_norm": 1.5301003456115723,
+      "learning_rate": 0.0002,
+      "loss": 1.2745,
+      "step": 5180
+    },
+    {
+      "epoch": 0.3152715344429595,
+      "grad_norm": 1.5374345779418945,
+      "learning_rate": 0.0002,
+      "loss": 1.2165,
+      "step": 5190
+    },
+    {
+      "epoch": 0.3158789940468959,
+      "grad_norm": 1.933231234550476,
+      "learning_rate": 0.0002,
+      "loss": 1.2294,
+      "step": 5200
+    },
+    {
+      "epoch": 0.3164864536508322,
+      "grad_norm": 1.652951717376709,
+      "learning_rate": 0.0002,
+      "loss": 1.1566,
+      "step": 5210
+    },
+    {
+      "epoch": 0.31709391325476854,
+      "grad_norm": 1.4390983581542969,
+      "learning_rate": 0.0002,
+      "loss": 1.1653,
+      "step": 5220
+    },
+    {
+      "epoch": 0.3177013728587049,
+      "grad_norm": 1.4972299337387085,
+      "learning_rate": 0.0002,
+      "loss": 1.3016,
+      "step": 5230
+    },
+    {
+      "epoch": 0.31830883246264124,
+      "grad_norm": 1.9257149696350098,
+      "learning_rate": 0.0002,
+      "loss": 1.2749,
+      "step": 5240
+    },
+    {
+      "epoch": 0.31891629206657757,
+      "grad_norm": 1.342387318611145,
+      "learning_rate": 0.0002,
+      "loss": 1.21,
+      "step": 5250
+    },
+    {
+      "epoch": 0.3195237516705139,
+      "grad_norm": 1.5092623233795166,
+      "learning_rate": 0.0002,
+      "loss": 1.1745,
+      "step": 5260
+    },
+    {
+      "epoch": 0.32013121127445027,
+      "grad_norm": 1.6232455968856812,
+      "learning_rate": 0.0002,
+      "loss": 1.2411,
+      "step": 5270
+    },
+    {
+      "epoch": 0.3207386708783866,
+      "grad_norm": 1.6391104459762573,
+      "learning_rate": 0.0002,
+      "loss": 1.25,
+      "step": 5280
+    },
+    {
+      "epoch": 0.3213461304823229,
+      "grad_norm": 1.4240717887878418,
+      "learning_rate": 0.0002,
+      "loss": 1.1992,
+      "step": 5290
+    },
+    {
+      "epoch": 0.3219535900862593,
+      "grad_norm": 1.3717237710952759,
+      "learning_rate": 0.0002,
+      "loss": 1.1977,
+      "step": 5300
+    },
+    {
+      "epoch": 0.3225610496901956,
+      "grad_norm": 1.588884711265564,
+      "learning_rate": 0.0002,
+      "loss": 1.2742,
+      "step": 5310
+    },
+    {
+      "epoch": 0.32316850929413193,
+      "grad_norm": 1.9746007919311523,
+      "learning_rate": 0.0002,
+      "loss": 1.2226,
+      "step": 5320
+    },
+    {
+      "epoch": 0.32377596889806826,
+      "grad_norm": 1.44120192527771,
+      "learning_rate": 0.0002,
+      "loss": 1.306,
+      "step": 5330
+    },
+    {
+      "epoch": 0.32438342850200463,
+      "grad_norm": 1.399114727973938,
+      "learning_rate": 0.0002,
+      "loss": 1.1719,
+      "step": 5340
+    },
+    {
+      "epoch": 0.32499088810594096,
+      "grad_norm": 1.334587574005127,
+      "learning_rate": 0.0002,
+      "loss": 1.2752,
+      "step": 5350
+    },
+    {
+      "epoch": 0.3255983477098773,
+      "grad_norm": 1.7185956239700317,
+      "learning_rate": 0.0002,
+      "loss": 1.2632,
+      "step": 5360
+    },
+    {
+      "epoch": 0.32620580731381366,
+      "grad_norm": 1.2144469022750854,
+      "learning_rate": 0.0002,
+      "loss": 1.2716,
+      "step": 5370
+    },
+    {
+      "epoch": 0.32681326691775,
+      "grad_norm": 2.094770908355713,
+      "learning_rate": 0.0002,
+      "loss": 1.2631,
+      "step": 5380
+    },
+    {
+      "epoch": 0.3274207265216863,
+      "grad_norm": 1.2292505502700806,
+      "learning_rate": 0.0002,
+      "loss": 1.2156,
+      "step": 5390
+    },
+    {
+      "epoch": 0.3280281861256226,
+      "grad_norm": 1.3732002973556519,
+      "learning_rate": 0.0002,
+      "loss": 1.2069,
+      "step": 5400
+    },
+    {
+      "epoch": 0.328635645729559,
+      "grad_norm": 1.239058494567871,
+      "learning_rate": 0.0002,
+      "loss": 1.2366,
+      "step": 5410
+    },
+    {
+      "epoch": 0.3292431053334953,
+      "grad_norm": 1.3541849851608276,
+      "learning_rate": 0.0002,
+      "loss": 1.2094,
+      "step": 5420
+    },
+    {
+      "epoch": 0.32985056493743165,
+      "grad_norm": 1.4627240896224976,
+      "learning_rate": 0.0002,
+      "loss": 1.1718,
+      "step": 5430
+    },
+    {
+      "epoch": 0.330458024541368,
+      "grad_norm": 1.4230258464813232,
+      "learning_rate": 0.0002,
+      "loss": 1.2226,
+      "step": 5440
+    },
+    {
+      "epoch": 0.33106548414530435,
+      "grad_norm": 1.561559796333313,
+      "learning_rate": 0.0002,
+      "loss": 1.1937,
+      "step": 5450
+    },
+    {
+      "epoch": 0.33167294374924067,
+      "grad_norm": 1.7308835983276367,
+      "learning_rate": 0.0002,
+      "loss": 1.2314,
+      "step": 5460
+    },
+    {
+      "epoch": 0.332280403353177,
+      "grad_norm": 1.6552231311798096,
+      "learning_rate": 0.0002,
+      "loss": 1.174,
+      "step": 5470
+    },
+    {
+      "epoch": 0.33288786295711337,
+      "grad_norm": 1.50615656375885,
+      "learning_rate": 0.0002,
+      "loss": 1.2601,
+      "step": 5480
+    },
+    {
+      "epoch": 0.3334953225610497,
+      "grad_norm": 1.470361590385437,
+      "learning_rate": 0.0002,
+      "loss": 1.2416,
+      "step": 5490
+    },
+    {
+      "epoch": 0.334102782164986,
+      "grad_norm": 1.4663327932357788,
+      "learning_rate": 0.0002,
+      "loss": 1.1818,
+      "step": 5500
+    },
+    {
+      "epoch": 0.3347102417689224,
+      "grad_norm": 1.7949563264846802,
+      "learning_rate": 0.0002,
+      "loss": 1.2176,
+      "step": 5510
+    },
+    {
+      "epoch": 0.3353177013728587,
+      "grad_norm": 1.6819008588790894,
+      "learning_rate": 0.0002,
+      "loss": 1.2298,
+      "step": 5520
+    },
+    {
+      "epoch": 0.33592516097679503,
+      "grad_norm": 1.433971881866455,
+      "learning_rate": 0.0002,
+      "loss": 1.2566,
+      "step": 5530
+    },
+    {
+      "epoch": 0.33653262058073136,
+      "grad_norm": 1.287117838859558,
+      "learning_rate": 0.0002,
+      "loss": 1.1138,
+      "step": 5540
+    },
+    {
+      "epoch": 0.33714008018466773,
+      "grad_norm": 1.4071489572525024,
+      "learning_rate": 0.0002,
+      "loss": 1.2651,
+      "step": 5550
+    },
+    {
+      "epoch": 0.33774753978860406,
+      "grad_norm": 1.5105435848236084,
+      "learning_rate": 0.0002,
+      "loss": 1.2409,
+      "step": 5560
+    },
+    {
+      "epoch": 0.3383549993925404,
+      "grad_norm": 1.683268427848816,
+      "learning_rate": 0.0002,
+      "loss": 1.1694,
+      "step": 5570
+    },
+    {
+      "epoch": 0.33896245899647676,
+      "grad_norm": 1.4265189170837402,
+      "learning_rate": 0.0002,
+      "loss": 1.2229,
+      "step": 5580
+    },
+    {
+      "epoch": 0.3395699186004131,
+      "grad_norm": 1.526524543762207,
+      "learning_rate": 0.0002,
+      "loss": 1.2259,
+      "step": 5590
+    },
+    {
+      "epoch": 0.3401773782043494,
+      "grad_norm": 1.2876828908920288,
+      "learning_rate": 0.0002,
+      "loss": 1.2253,
+      "step": 5600
+    },
+    {
+      "epoch": 0.3407848378082857,
+      "grad_norm": 1.3172236680984497,
+      "learning_rate": 0.0002,
+      "loss": 1.2494,
+      "step": 5610
+    },
+    {
+      "epoch": 0.3413922974122221,
+      "grad_norm": 1.8559517860412598,
+      "learning_rate": 0.0002,
+      "loss": 1.1937,
+      "step": 5620
+    },
+    {
+      "epoch": 0.3419997570161584,
+      "grad_norm": 1.3766885995864868,
+      "learning_rate": 0.0002,
+      "loss": 1.1814,
+      "step": 5630
+    },
+    {
+      "epoch": 0.34260721662009475,
+      "grad_norm": 1.3915786743164062,
+      "learning_rate": 0.0002,
+      "loss": 1.2039,
+      "step": 5640
+    },
+    {
+      "epoch": 0.3432146762240311,
+      "grad_norm": 1.6250295639038086,
+      "learning_rate": 0.0002,
+      "loss": 1.1728,
+      "step": 5650
+    },
+    {
+      "epoch": 0.34382213582796745,
+      "grad_norm": 1.2863385677337646,
+      "learning_rate": 0.0002,
+      "loss": 1.1466,
+      "step": 5660
+    },
+    {
+      "epoch": 0.34442959543190377,
+      "grad_norm": 1.5022464990615845,
+      "learning_rate": 0.0002,
+      "loss": 1.1795,
+      "step": 5670
+    },
+    {
+      "epoch": 0.3450370550358401,
+      "grad_norm": 1.4029043912887573,
+      "learning_rate": 0.0002,
+      "loss": 1.3014,
+      "step": 5680
+    },
+    {
+      "epoch": 0.34564451463977647,
+      "grad_norm": 1.629833698272705,
+      "learning_rate": 0.0002,
+      "loss": 1.2463,
+      "step": 5690
+    },
+    {
+      "epoch": 0.3462519742437128,
+      "grad_norm": 1.4233486652374268,
+      "learning_rate": 0.0002,
+      "loss": 1.2448,
+      "step": 5700
+    },
+    {
+      "epoch": 0.3468594338476491,
+      "grad_norm": 1.4671995639801025,
+      "learning_rate": 0.0002,
+      "loss": 1.255,
+      "step": 5710
+    },
+    {
+      "epoch": 0.3474668934515855,
+      "grad_norm": 1.5535781383514404,
+      "learning_rate": 0.0002,
+      "loss": 1.1663,
+      "step": 5720
+    },
+    {
+      "epoch": 0.3480743530555218,
+      "grad_norm": 1.6772747039794922,
+      "learning_rate": 0.0002,
+      "loss": 1.1794,
+      "step": 5730
+    },
+    {
+      "epoch": 0.34868181265945813,
+      "grad_norm": 1.595218539237976,
+      "learning_rate": 0.0002,
+      "loss": 1.1627,
+      "step": 5740
+    },
+    {
+      "epoch": 0.34928927226339446,
+      "grad_norm": 1.3816550970077515,
+      "learning_rate": 0.0002,
+      "loss": 1.2398,
+      "step": 5750
+    },
+    {
+      "epoch": 0.34989673186733083,
+      "grad_norm": 1.3850511312484741,
+      "learning_rate": 0.0002,
+      "loss": 1.3081,
+      "step": 5760
+    },
+    {
+      "epoch": 0.35050419147126716,
+      "grad_norm": 1.6395469903945923,
+      "learning_rate": 0.0002,
+      "loss": 1.2571,
+      "step": 5770
+    },
+    {
+      "epoch": 0.3511116510752035,
+      "grad_norm": 1.5687739849090576,
+      "learning_rate": 0.0002,
+      "loss": 1.2466,
+      "step": 5780
+    },
+    {
+      "epoch": 0.35171911067913986,
+      "grad_norm": 1.371972918510437,
+      "learning_rate": 0.0002,
+      "loss": 1.2079,
+      "step": 5790
+    },
+    {
+      "epoch": 0.3523265702830762,
+      "grad_norm": 1.355375051498413,
+      "learning_rate": 0.0002,
+      "loss": 1.1537,
+      "step": 5800
+    },
+    {
+      "epoch": 0.3529340298870125,
+      "grad_norm": 1.5896039009094238,
+      "learning_rate": 0.0002,
+      "loss": 1.1209,
+      "step": 5810
+    },
+    {
+      "epoch": 0.3535414894909489,
+      "grad_norm": 1.5453897714614868,
+      "learning_rate": 0.0002,
+      "loss": 1.2603,
+      "step": 5820
+    },
+    {
+      "epoch": 0.3541489490948852,
+      "grad_norm": 1.5207583904266357,
+      "learning_rate": 0.0002,
+      "loss": 1.1796,
+      "step": 5830
+    },
+    {
+      "epoch": 0.3547564086988215,
+      "grad_norm": 2.068467855453491,
+      "learning_rate": 0.0002,
+      "loss": 1.3053,
+      "step": 5840
+    },
+    {
+      "epoch": 0.35536386830275785,
+      "grad_norm": 1.3215420246124268,
+      "learning_rate": 0.0002,
+      "loss": 1.1777,
+      "step": 5850
+    },
+    {
+      "epoch": 0.3559713279066942,
+      "grad_norm": 1.4459247589111328,
+      "learning_rate": 0.0002,
+      "loss": 1.1743,
+      "step": 5860
+    },
+    {
+      "epoch": 0.35657878751063055,
+      "grad_norm": 1.3044105768203735,
+      "learning_rate": 0.0002,
+      "loss": 1.1718,
+      "step": 5870
+    },
+    {
+      "epoch": 0.35718624711456687,
+      "grad_norm": 1.6035641431808472,
+      "learning_rate": 0.0002,
+      "loss": 1.1955,
+      "step": 5880
+    },
+    {
+      "epoch": 0.35779370671850325,
+      "grad_norm": 1.5534878969192505,
+      "learning_rate": 0.0002,
+      "loss": 1.22,
+      "step": 5890
+    },
+    {
+      "epoch": 0.35840116632243957,
+      "grad_norm": 1.379349708557129,
+      "learning_rate": 0.0002,
+      "loss": 1.2128,
+      "step": 5900
+    },
+    {
+      "epoch": 0.3590086259263759,
+      "grad_norm": 1.5389310121536255,
+      "learning_rate": 0.0002,
+      "loss": 1.1994,
+      "step": 5910
+    },
+    {
+      "epoch": 0.3596160855303122,
+      "grad_norm": 1.3085976839065552,
+      "learning_rate": 0.0002,
+      "loss": 1.2147,
+      "step": 5920
+    },
+    {
+      "epoch": 0.3602235451342486,
+      "grad_norm": 1.714221477508545,
+      "learning_rate": 0.0002,
+      "loss": 1.2167,
+      "step": 5930
+    },
+    {
+      "epoch": 0.3608310047381849,
+      "grad_norm": 1.374284267425537,
+      "learning_rate": 0.0002,
+      "loss": 1.2322,
+      "step": 5940
+    },
+    {
+      "epoch": 0.36143846434212124,
+      "grad_norm": 1.2360552549362183,
+      "learning_rate": 0.0002,
+      "loss": 1.17,
+      "step": 5950
+    },
+    {
+      "epoch": 0.3620459239460576,
+      "grad_norm": 1.454513669013977,
+      "learning_rate": 0.0002,
+      "loss": 1.1626,
+      "step": 5960
+    },
+    {
+      "epoch": 0.36265338354999394,
+      "grad_norm": 1.5351463556289673,
+      "learning_rate": 0.0002,
+      "loss": 1.261,
+      "step": 5970
+    },
+    {
+      "epoch": 0.36326084315393026,
+      "grad_norm": 1.5355708599090576,
+      "learning_rate": 0.0002,
+      "loss": 1.205,
+      "step": 5980
+    },
+    {
+      "epoch": 0.3638683027578666,
+      "grad_norm": 1.7666850090026855,
+      "learning_rate": 0.0002,
+      "loss": 1.1729,
+      "step": 5990
+    },
+    {
+      "epoch": 0.36447576236180296,
+      "grad_norm": 1.2283000946044922,
+      "learning_rate": 0.0002,
+      "loss": 1.2067,
+      "step": 6000
+    },
+    {
+      "epoch": 0.3650832219657393,
+      "grad_norm": 1.4924534559249878,
+      "learning_rate": 0.0002,
+      "loss": 1.2551,
+      "step": 6010
+    },
+    {
+      "epoch": 0.3656906815696756,
+      "grad_norm": 1.5576777458190918,
+      "learning_rate": 0.0002,
+      "loss": 1.2186,
+      "step": 6020
+    },
+    {
+      "epoch": 0.366298141173612,
+      "grad_norm": 1.5818428993225098,
+      "learning_rate": 0.0002,
+      "loss": 1.1631,
+      "step": 6030
+    },
+    {
+      "epoch": 0.3669056007775483,
+      "grad_norm": 1.3564088344573975,
+      "learning_rate": 0.0002,
+      "loss": 1.2092,
+      "step": 6040
+    },
+    {
+      "epoch": 0.3675130603814846,
+      "grad_norm": 1.619210958480835,
+      "learning_rate": 0.0002,
+      "loss": 1.2246,
+      "step": 6050
+    },
+    {
+      "epoch": 0.36812051998542095,
+      "grad_norm": 1.304306149482727,
+      "learning_rate": 0.0002,
+      "loss": 1.1744,
+      "step": 6060
+    },
+    {
+      "epoch": 0.3687279795893573,
+      "grad_norm": 1.3461626768112183,
+      "learning_rate": 0.0002,
+      "loss": 1.2008,
+      "step": 6070
+    },
+    {
+      "epoch": 0.36933543919329365,
+      "grad_norm": 1.7437419891357422,
+      "learning_rate": 0.0002,
+      "loss": 1.2251,
+      "step": 6080
+    },
+    {
+      "epoch": 0.36994289879722997,
+      "grad_norm": 1.3706979751586914,
+      "learning_rate": 0.0002,
+      "loss": 1.1056,
+      "step": 6090
+    },
+    {
+      "epoch": 0.37055035840116635,
+      "grad_norm": 1.5688164234161377,
+      "learning_rate": 0.0002,
+      "loss": 1.195,
+      "step": 6100
+    },
+    {
+      "epoch": 0.37115781800510267,
+      "grad_norm": 1.3877938985824585,
+      "learning_rate": 0.0002,
+      "loss": 1.1887,
+      "step": 6110
+    },
+    {
+      "epoch": 0.371765277609039,
+      "grad_norm": 1.4119068384170532,
+      "learning_rate": 0.0002,
+      "loss": 1.1873,
+      "step": 6120
+    },
+    {
+      "epoch": 0.3723727372129753,
+      "grad_norm": 1.506616234779358,
+      "learning_rate": 0.0002,
+      "loss": 1.1423,
+      "step": 6130
+    },
+    {
+      "epoch": 0.3729801968169117,
+      "grad_norm": 1.3528050184249878,
+      "learning_rate": 0.0002,
+      "loss": 1.1749,
+      "step": 6140
+    },
+    {
+      "epoch": 0.373587656420848,
+      "grad_norm": 1.3234665393829346,
+      "learning_rate": 0.0002,
+      "loss": 1.1619,
+      "step": 6150
+    },
+    {
+      "epoch": 0.37419511602478434,
+      "grad_norm": 1.3826556205749512,
+      "learning_rate": 0.0002,
+      "loss": 1.2515,
+      "step": 6160
+    },
+    {
+      "epoch": 0.3748025756287207,
+      "grad_norm": 1.4598699808120728,
+      "learning_rate": 0.0002,
+      "loss": 1.2635,
+      "step": 6170
+    },
+    {
+      "epoch": 0.37541003523265704,
+      "grad_norm": 1.496317982673645,
+      "learning_rate": 0.0002,
+      "loss": 1.22,
+      "step": 6180
+    },
+    {
+      "epoch": 0.37601749483659336,
+      "grad_norm": 1.560288667678833,
+      "learning_rate": 0.0002,
+      "loss": 1.2147,
+      "step": 6190
+    },
+    {
+      "epoch": 0.3766249544405297,
+      "grad_norm": 1.411517858505249,
+      "learning_rate": 0.0002,
+      "loss": 1.127,
+      "step": 6200
+    },
+    {
+      "epoch": 0.37723241404446606,
+      "grad_norm": 1.7069580554962158,
+      "learning_rate": 0.0002,
+      "loss": 1.1665,
+      "step": 6210
+    },
+    {
+      "epoch": 0.3778398736484024,
+      "grad_norm": 1.5629972219467163,
+      "learning_rate": 0.0002,
+      "loss": 1.1582,
+      "step": 6220
+    },
+    {
+      "epoch": 0.3784473332523387,
+      "grad_norm": 1.464280128479004,
+      "learning_rate": 0.0002,
+      "loss": 1.2632,
+      "step": 6230
+    },
+    {
+      "epoch": 0.3790547928562751,
+      "grad_norm": 1.2921830415725708,
+      "learning_rate": 0.0002,
+      "loss": 1.0899,
+      "step": 6240
+    },
+    {
+      "epoch": 0.3796622524602114,
+      "grad_norm": 1.8416919708251953,
+      "learning_rate": 0.0002,
+      "loss": 1.2606,
+      "step": 6250
+    },
+    {
+      "epoch": 0.3802697120641477,
+      "grad_norm": 1.4949452877044678,
+      "learning_rate": 0.0002,
+      "loss": 1.2811,
+      "step": 6260
+    },
+    {
+      "epoch": 0.38087717166808405,
+      "grad_norm": 1.291973352432251,
+      "learning_rate": 0.0002,
+      "loss": 1.1879,
+      "step": 6270
+    },
+    {
+      "epoch": 0.3814846312720204,
+      "grad_norm": 1.2897945642471313,
+      "learning_rate": 0.0002,
+      "loss": 1.1769,
+      "step": 6280
+    },
+    {
+      "epoch": 0.38209209087595675,
+      "grad_norm": 1.3945434093475342,
+      "learning_rate": 0.0002,
+      "loss": 1.1723,
+      "step": 6290
+    },
+    {
+      "epoch": 0.38269955047989307,
+      "grad_norm": 1.7047033309936523,
+      "learning_rate": 0.0002,
+      "loss": 1.2142,
+      "step": 6300
+    },
+    {
+      "epoch": 0.38330701008382945,
+      "grad_norm": 1.5957000255584717,
+      "learning_rate": 0.0002,
+      "loss": 1.2007,
+      "step": 6310
+    },
+    {
+      "epoch": 0.38391446968776577,
+      "grad_norm": 1.7057619094848633,
+      "learning_rate": 0.0002,
+      "loss": 1.2522,
+      "step": 6320
+    },
+    {
+      "epoch": 0.3845219292917021,
+      "grad_norm": 1.510622501373291,
+      "learning_rate": 0.0002,
+      "loss": 1.2798,
+      "step": 6330
+    },
+    {
+      "epoch": 0.3851293888956384,
+      "grad_norm": 1.5083156824111938,
+      "learning_rate": 0.0002,
+      "loss": 1.1573,
+      "step": 6340
+    },
+    {
+      "epoch": 0.3857368484995748,
+      "grad_norm": 1.5803555250167847,
+      "learning_rate": 0.0002,
+      "loss": 1.1618,
+      "step": 6350
+    },
+    {
+      "epoch": 0.3863443081035111,
+      "grad_norm": 1.6500353813171387,
+      "learning_rate": 0.0002,
+      "loss": 1.2375,
+      "step": 6360
+    },
+    {
+      "epoch": 0.38695176770744744,
+      "grad_norm": 1.4603452682495117,
+      "learning_rate": 0.0002,
+      "loss": 1.2284,
+      "step": 6370
+    },
+    {
+      "epoch": 0.3875592273113838,
+      "grad_norm": 1.45824134349823,
+      "learning_rate": 0.0002,
+      "loss": 1.2223,
+      "step": 6380
+    },
+    {
+      "epoch": 0.38816668691532014,
+      "grad_norm": 1.4440996646881104,
+      "learning_rate": 0.0002,
+      "loss": 1.2065,
+      "step": 6390
+    },
+    {
+      "epoch": 0.38877414651925646,
+      "grad_norm": 1.4885029792785645,
+      "learning_rate": 0.0002,
+      "loss": 1.2125,
+      "step": 6400
+    },
+    {
+      "epoch": 0.3893816061231928,
+      "grad_norm": 1.4131464958190918,
+      "learning_rate": 0.0002,
+      "loss": 1.2895,
+      "step": 6410
+    },
+    {
+      "epoch": 0.38998906572712916,
+      "grad_norm": 1.508143663406372,
+      "learning_rate": 0.0002,
+      "loss": 1.2626,
+      "step": 6420
+    },
+    {
+      "epoch": 0.3905965253310655,
+      "grad_norm": 1.6027228832244873,
+      "learning_rate": 0.0002,
+      "loss": 1.2433,
+      "step": 6430
+    },
+    {
+      "epoch": 0.3912039849350018,
+      "grad_norm": 1.2766356468200684,
+      "learning_rate": 0.0002,
+      "loss": 1.1961,
+      "step": 6440
+    },
+    {
+      "epoch": 0.3918114445389382,
+      "grad_norm": 1.2684210538864136,
+      "learning_rate": 0.0002,
+      "loss": 1.1421,
+      "step": 6450
+    },
+    {
+      "epoch": 0.3924189041428745,
+      "grad_norm": 1.5941156148910522,
+      "learning_rate": 0.0002,
+      "loss": 1.1218,
+      "step": 6460
+    },
+    {
+      "epoch": 0.3930263637468108,
+      "grad_norm": 1.452562689781189,
+      "learning_rate": 0.0002,
+      "loss": 1.1033,
+      "step": 6470
+    },
+    {
+      "epoch": 0.3936338233507472,
+      "grad_norm": 1.5934187173843384,
+      "learning_rate": 0.0002,
+      "loss": 1.1658,
+      "step": 6480
+    },
+    {
+      "epoch": 0.3942412829546835,
+      "grad_norm": 1.5304055213928223,
+      "learning_rate": 0.0002,
+      "loss": 1.2962,
+      "step": 6490
+    },
+    {
+      "epoch": 0.39484874255861985,
+      "grad_norm": 1.5306092500686646,
+      "learning_rate": 0.0002,
+      "loss": 1.277,
+      "step": 6500
+    },
+    {
+      "epoch": 0.39545620216255617,
+      "grad_norm": 1.4074877500534058,
+      "learning_rate": 0.0002,
+      "loss": 1.2062,
+      "step": 6510
+    },
+    {
+      "epoch": 0.39606366176649255,
+      "grad_norm": 1.5896267890930176,
+      "learning_rate": 0.0002,
+      "loss": 1.151,
+      "step": 6520
+    },
+    {
+      "epoch": 0.39667112137042887,
+      "grad_norm": 1.3772437572479248,
+      "learning_rate": 0.0002,
+      "loss": 1.152,
+      "step": 6530
+    },
+    {
+      "epoch": 0.3972785809743652,
+      "grad_norm": 1.3611092567443848,
+      "learning_rate": 0.0002,
+      "loss": 1.2902,
+      "step": 6540
+    },
+    {
+      "epoch": 0.39788604057830157,
+      "grad_norm": 1.278731107711792,
+      "learning_rate": 0.0002,
+      "loss": 1.2716,
+      "step": 6550
+    },
+    {
+      "epoch": 0.3984935001822379,
+      "grad_norm": 1.9447767734527588,
+      "learning_rate": 0.0002,
+      "loss": 1.2449,
+      "step": 6560
+    },
+    {
+      "epoch": 0.3991009597861742,
+      "grad_norm": 1.4802900552749634,
+      "learning_rate": 0.0002,
+      "loss": 1.1776,
+      "step": 6570
+    },
+    {
+      "epoch": 0.39970841939011054,
+      "grad_norm": 1.6410369873046875,
+      "learning_rate": 0.0002,
+      "loss": 1.255,
+      "step": 6580
+    },
+    {
+      "epoch": 0.4003158789940469,
+      "grad_norm": 1.5125056505203247,
+      "learning_rate": 0.0002,
+      "loss": 1.1849,
+      "step": 6590
+    },
+    {
+      "epoch": 0.40092333859798324,
+      "grad_norm": 1.67632257938385,
+      "learning_rate": 0.0002,
+      "loss": 1.2153,
+      "step": 6600
+    },
+    {
+      "epoch": 0.40153079820191956,
+      "grad_norm": 1.3930867910385132,
+      "learning_rate": 0.0002,
+      "loss": 1.1671,
+      "step": 6610
+    },
+    {
+      "epoch": 0.40213825780585594,
+      "grad_norm": 1.489687204360962,
+      "learning_rate": 0.0002,
+      "loss": 1.1823,
+      "step": 6620
+    },
+    {
+      "epoch": 0.40274571740979226,
+      "grad_norm": 1.4691733121871948,
+      "learning_rate": 0.0002,
+      "loss": 1.1653,
+      "step": 6630
+    },
+    {
+      "epoch": 0.4033531770137286,
+      "grad_norm": 1.4824140071868896,
+      "learning_rate": 0.0002,
+      "loss": 1.1729,
+      "step": 6640
+    },
+    {
+      "epoch": 0.4039606366176649,
+      "grad_norm": 1.3477215766906738,
+      "learning_rate": 0.0002,
+      "loss": 1.2202,
+      "step": 6650
+    },
+    {
+      "epoch": 0.4045680962216013,
+      "grad_norm": 1.489844560623169,
+      "learning_rate": 0.0002,
+      "loss": 1.1715,
+      "step": 6660
+    },
+    {
+      "epoch": 0.4051755558255376,
+      "grad_norm": 1.4528323411941528,
+      "learning_rate": 0.0002,
+      "loss": 1.2474,
+      "step": 6670
+    },
+    {
+      "epoch": 0.4057830154294739,
+      "grad_norm": 1.4013316631317139,
+      "learning_rate": 0.0002,
+      "loss": 1.1461,
+      "step": 6680
+    },
+    {
+      "epoch": 0.4063904750334103,
+      "grad_norm": 1.2306127548217773,
+      "learning_rate": 0.0002,
+      "loss": 1.3095,
+      "step": 6690
+    },
+    {
+      "epoch": 0.4069979346373466,
+      "grad_norm": 1.4853254556655884,
+      "learning_rate": 0.0002,
+      "loss": 1.2001,
+      "step": 6700
+    },
+    {
+      "epoch": 0.40760539424128295,
+      "grad_norm": 1.5329333543777466,
+      "learning_rate": 0.0002,
+      "loss": 1.2045,
+      "step": 6710
+    },
+    {
+      "epoch": 0.40821285384521927,
+      "grad_norm": 1.3074980974197388,
+      "learning_rate": 0.0002,
+      "loss": 1.1843,
+      "step": 6720
+    },
+    {
+      "epoch": 0.40882031344915565,
+      "grad_norm": 1.3135238885879517,
+      "learning_rate": 0.0002,
+      "loss": 1.1981,
+      "step": 6730
+    },
+    {
+      "epoch": 0.40942777305309197,
+      "grad_norm": 1.5675169229507446,
+      "learning_rate": 0.0002,
+      "loss": 1.2781,
+      "step": 6740
+    },
+    {
+      "epoch": 0.4100352326570283,
+      "grad_norm": 1.4353513717651367,
+      "learning_rate": 0.0002,
+      "loss": 1.1986,
+      "step": 6750
+    },
+    {
+      "epoch": 0.41064269226096467,
+      "grad_norm": 1.6439242362976074,
+      "learning_rate": 0.0002,
+      "loss": 1.1873,
+      "step": 6760
+    },
+    {
+      "epoch": 0.411250151864901,
+      "grad_norm": 1.4382151365280151,
+      "learning_rate": 0.0002,
+      "loss": 1.2623,
+      "step": 6770
+    },
+    {
+      "epoch": 0.4118576114688373,
+      "grad_norm": 1.8752695322036743,
+      "learning_rate": 0.0002,
+      "loss": 1.2171,
+      "step": 6780
+    },
+    {
+      "epoch": 0.41246507107277364,
+      "grad_norm": 1.2991536855697632,
+      "learning_rate": 0.0002,
+      "loss": 1.1994,
+      "step": 6790
+    },
+    {
+      "epoch": 0.41307253067671,
+      "grad_norm": 1.4035308361053467,
+      "learning_rate": 0.0002,
+      "loss": 1.1887,
+      "step": 6800
+    },
+    {
+      "epoch": 0.41367999028064634,
+      "grad_norm": 1.479036569595337,
+      "learning_rate": 0.0002,
+      "loss": 1.2312,
+      "step": 6810
+    },
+    {
+      "epoch": 0.41428744988458266,
+      "grad_norm": 1.43748140335083,
+      "learning_rate": 0.0002,
+      "loss": 1.1587,
+      "step": 6820
+    },
+    {
+      "epoch": 0.41489490948851904,
+      "grad_norm": 1.4034409523010254,
+      "learning_rate": 0.0002,
+      "loss": 1.1746,
+      "step": 6830
+    },
+    {
+      "epoch": 0.41550236909245536,
+      "grad_norm": 1.6533730030059814,
+      "learning_rate": 0.0002,
+      "loss": 1.1636,
+      "step": 6840
+    },
+    {
+      "epoch": 0.4161098286963917,
+      "grad_norm": 1.5264071226119995,
+      "learning_rate": 0.0002,
+      "loss": 1.213,
+      "step": 6850
+    },
+    {
+      "epoch": 0.416717288300328,
+      "grad_norm": 1.3889992237091064,
+      "learning_rate": 0.0002,
+      "loss": 1.1773,
+      "step": 6860
+    },
+    {
+      "epoch": 0.4173247479042644,
+      "grad_norm": 1.534353494644165,
+      "learning_rate": 0.0002,
+      "loss": 1.1476,
+      "step": 6870
+    },
+    {
+      "epoch": 0.4179322075082007,
+      "grad_norm": 1.8379203081130981,
+      "learning_rate": 0.0002,
+      "loss": 1.2032,
+      "step": 6880
+    },
+    {
+      "epoch": 0.418539667112137,
+      "grad_norm": 1.385796070098877,
+      "learning_rate": 0.0002,
+      "loss": 1.162,
+      "step": 6890
+    },
+    {
+      "epoch": 0.4191471267160734,
+      "grad_norm": 1.5238981246948242,
+      "learning_rate": 0.0002,
+      "loss": 1.2319,
+      "step": 6900
+    },
+    {
+      "epoch": 0.4197545863200097,
+      "grad_norm": 1.5223041772842407,
+      "learning_rate": 0.0002,
+      "loss": 1.1873,
+      "step": 6910
+    },
+    {
+      "epoch": 0.42036204592394605,
+      "grad_norm": 1.7702144384384155,
+      "learning_rate": 0.0002,
+      "loss": 1.2035,
+      "step": 6920
+    },
+    {
+      "epoch": 0.42096950552788237,
+      "grad_norm": 1.6968584060668945,
+      "learning_rate": 0.0002,
+      "loss": 1.175,
+      "step": 6930
+    },
+    {
+      "epoch": 0.42157696513181875,
+      "grad_norm": 1.3365684747695923,
+      "learning_rate": 0.0002,
+      "loss": 1.2121,
+      "step": 6940
+    },
+    {
+      "epoch": 0.42218442473575507,
+      "grad_norm": 1.2800488471984863,
+      "learning_rate": 0.0002,
+      "loss": 1.2039,
+      "step": 6950
+    },
+    {
+      "epoch": 0.4227918843396914,
+      "grad_norm": 1.4685325622558594,
+      "learning_rate": 0.0002,
+      "loss": 1.2429,
+      "step": 6960
+    },
+    {
+      "epoch": 0.42339934394362777,
+      "grad_norm": 1.3646459579467773,
+      "learning_rate": 0.0002,
+      "loss": 1.1854,
+      "step": 6970
+    },
+    {
+      "epoch": 0.4240068035475641,
+      "grad_norm": 1.4553136825561523,
+      "learning_rate": 0.0002,
+      "loss": 1.2194,
+      "step": 6980
+    },
+    {
+      "epoch": 0.4246142631515004,
+      "grad_norm": 1.6624222993850708,
+      "learning_rate": 0.0002,
+      "loss": 1.2701,
+      "step": 6990
+    },
+    {
+      "epoch": 0.42522172275543674,
+      "grad_norm": 1.4376602172851562,
+      "learning_rate": 0.0002,
+      "loss": 1.1946,
+      "step": 7000
+    },
+    {
+      "epoch": 0.4258291823593731,
+      "grad_norm": 1.4615241289138794,
+      "learning_rate": 0.0002,
+      "loss": 1.2371,
+      "step": 7010
+    },
+    {
+      "epoch": 0.42643664196330944,
+      "grad_norm": 1.3993204832077026,
+      "learning_rate": 0.0002,
+      "loss": 1.2215,
+      "step": 7020
+    },
+    {
+      "epoch": 0.42704410156724576,
+      "grad_norm": 1.3323475122451782,
+      "learning_rate": 0.0002,
+      "loss": 1.1923,
+      "step": 7030
+    },
+    {
+      "epoch": 0.42765156117118214,
+      "grad_norm": 1.5600262880325317,
+      "learning_rate": 0.0002,
+      "loss": 1.211,
+      "step": 7040
+    },
+    {
+      "epoch": 0.42825902077511846,
+      "grad_norm": 1.491089105606079,
+      "learning_rate": 0.0002,
+      "loss": 1.2394,
+      "step": 7050
+    },
+    {
+      "epoch": 0.4288664803790548,
+      "grad_norm": 1.4473880529403687,
+      "learning_rate": 0.0002,
+      "loss": 1.2907,
+      "step": 7060
+    },
+    {
+      "epoch": 0.4294739399829911,
+      "grad_norm": 1.6947537660598755,
+      "learning_rate": 0.0002,
+      "loss": 1.269,
+      "step": 7070
+    },
+    {
+      "epoch": 0.4300813995869275,
+      "grad_norm": 1.3262157440185547,
+      "learning_rate": 0.0002,
+      "loss": 1.2718,
+      "step": 7080
+    },
+    {
+      "epoch": 0.4306888591908638,
+      "grad_norm": 1.3940998315811157,
+      "learning_rate": 0.0002,
+      "loss": 1.2493,
+      "step": 7090
+    },
+    {
+      "epoch": 0.4312963187948001,
+      "grad_norm": 1.6804521083831787,
+      "learning_rate": 0.0002,
+      "loss": 1.1106,
+      "step": 7100
+    },
+    {
+      "epoch": 0.4319037783987365,
+      "grad_norm": 1.8600741624832153,
+      "learning_rate": 0.0002,
+      "loss": 1.2041,
+      "step": 7110
+    },
+    {
+      "epoch": 0.4325112380026728,
+      "grad_norm": 1.4588673114776611,
+      "learning_rate": 0.0002,
+      "loss": 1.1889,
+      "step": 7120
+    },
+    {
+      "epoch": 0.43311869760660915,
+      "grad_norm": 1.4329715967178345,
+      "learning_rate": 0.0002,
+      "loss": 1.2104,
+      "step": 7130
+    },
+    {
+      "epoch": 0.43372615721054547,
+      "grad_norm": 1.4520699977874756,
+      "learning_rate": 0.0002,
+      "loss": 1.1788,
+      "step": 7140
+    },
+    {
+      "epoch": 0.43433361681448185,
+      "grad_norm": 1.4037648439407349,
+      "learning_rate": 0.0002,
+      "loss": 1.2121,
+      "step": 7150
+    },
+    {
+      "epoch": 0.43494107641841817,
+      "grad_norm": 1.6112241744995117,
+      "learning_rate": 0.0002,
+      "loss": 1.1776,
+      "step": 7160
+    },
+    {
+      "epoch": 0.4355485360223545,
+      "grad_norm": 1.3666143417358398,
+      "learning_rate": 0.0002,
+      "loss": 1.1307,
+      "step": 7170
+    },
+    {
+      "epoch": 0.43615599562629087,
+      "grad_norm": 1.8091779947280884,
+      "learning_rate": 0.0002,
+      "loss": 1.2301,
+      "step": 7180
+    },
+    {
+      "epoch": 0.4367634552302272,
+      "grad_norm": 1.394896149635315,
+      "learning_rate": 0.0002,
+      "loss": 1.2655,
+      "step": 7190
+    },
+    {
+      "epoch": 0.4373709148341635,
+      "grad_norm": 1.6497581005096436,
+      "learning_rate": 0.0002,
+      "loss": 1.2711,
+      "step": 7200
+    },
+    {
+      "epoch": 0.4379783744380999,
+      "grad_norm": 1.5435038805007935,
+      "learning_rate": 0.0002,
+      "loss": 1.2164,
+      "step": 7210
+    },
+    {
+      "epoch": 0.4385858340420362,
+      "grad_norm": 1.4208731651306152,
+      "learning_rate": 0.0002,
+      "loss": 1.2301,
+      "step": 7220
+    },
+    {
+      "epoch": 0.43919329364597254,
+      "grad_norm": 1.5722650289535522,
+      "learning_rate": 0.0002,
+      "loss": 1.183,
+      "step": 7230
+    },
+    {
+      "epoch": 0.43980075324990886,
+      "grad_norm": 1.6592830419540405,
+      "learning_rate": 0.0002,
+      "loss": 1.1618,
+      "step": 7240
+    },
+    {
+      "epoch": 0.44040821285384524,
+      "grad_norm": 1.481824517250061,
+      "learning_rate": 0.0002,
+      "loss": 1.165,
+      "step": 7250
+    },
+    {
+      "epoch": 0.44101567245778156,
+      "grad_norm": 1.4777569770812988,
+      "learning_rate": 0.0002,
+      "loss": 1.2317,
+      "step": 7260
+    },
+    {
+      "epoch": 0.4416231320617179,
+      "grad_norm": 1.5149354934692383,
+      "learning_rate": 0.0002,
+      "loss": 1.175,
+      "step": 7270
+    },
+    {
+      "epoch": 0.44223059166565426,
+      "grad_norm": 1.3621138334274292,
+      "learning_rate": 0.0002,
+      "loss": 1.0801,
+      "step": 7280
+    },
+    {
+      "epoch": 0.4428380512695906,
+      "grad_norm": 1.7920101881027222,
+      "learning_rate": 0.0002,
+      "loss": 1.2371,
+      "step": 7290
+    },
+    {
+      "epoch": 0.4434455108735269,
+      "grad_norm": 1.488234043121338,
+      "learning_rate": 0.0002,
+      "loss": 1.1294,
+      "step": 7300
+    },
+    {
+      "epoch": 0.4440529704774632,
+      "grad_norm": 1.4184207916259766,
+      "learning_rate": 0.0002,
+      "loss": 1.1769,
+      "step": 7310
+    },
+    {
+      "epoch": 0.4446604300813996,
+      "grad_norm": 1.6217948198318481,
+      "learning_rate": 0.0002,
+      "loss": 1.2418,
+      "step": 7320
+    },
+    {
+      "epoch": 0.4452678896853359,
+      "grad_norm": 1.5252745151519775,
+      "learning_rate": 0.0002,
+      "loss": 1.2096,
+      "step": 7330
+    },
+    {
+      "epoch": 0.44587534928927225,
+      "grad_norm": 1.5417364835739136,
+      "learning_rate": 0.0002,
+      "loss": 1.1547,
+      "step": 7340
+    },
+    {
+      "epoch": 0.4464828088932086,
+      "grad_norm": 1.6695308685302734,
+      "learning_rate": 0.0002,
+      "loss": 1.2689,
+      "step": 7350
+    },
+    {
+      "epoch": 0.44709026849714495,
+      "grad_norm": 1.2754809856414795,
+      "learning_rate": 0.0002,
+      "loss": 1.1707,
+      "step": 7360
+    },
+    {
+      "epoch": 0.44769772810108127,
+      "grad_norm": 1.3510041236877441,
+      "learning_rate": 0.0002,
+      "loss": 1.1905,
+      "step": 7370
+    },
+    {
+      "epoch": 0.4483051877050176,
+      "grad_norm": 1.702237606048584,
+      "learning_rate": 0.0002,
+      "loss": 1.2166,
+      "step": 7380
+    },
+    {
+      "epoch": 0.44891264730895397,
+      "grad_norm": 1.3573112487792969,
+      "learning_rate": 0.0002,
+      "loss": 1.2114,
+      "step": 7390
+    },
+    {
+      "epoch": 0.4495201069128903,
+      "grad_norm": 1.4572347402572632,
+      "learning_rate": 0.0002,
+      "loss": 1.2635,
+      "step": 7400
+    },
+    {
+      "epoch": 0.4501275665168266,
+      "grad_norm": 1.5683256387710571,
+      "learning_rate": 0.0002,
+      "loss": 1.1404,
+      "step": 7410
+    },
+    {
+      "epoch": 0.450735026120763,
+      "grad_norm": 1.671855092048645,
+      "learning_rate": 0.0002,
+      "loss": 1.1997,
+      "step": 7420
+    },
+    {
+      "epoch": 0.4513424857246993,
+      "grad_norm": 1.6509655714035034,
+      "learning_rate": 0.0002,
+      "loss": 1.234,
+      "step": 7430
+    },
+    {
+      "epoch": 0.45194994532863564,
+      "grad_norm": 1.8359466791152954,
+      "learning_rate": 0.0002,
+      "loss": 1.259,
+      "step": 7440
+    },
+    {
+      "epoch": 0.45255740493257196,
+      "grad_norm": 1.7122342586517334,
+      "learning_rate": 0.0002,
+      "loss": 1.1588,
+      "step": 7450
+    },
+    {
+      "epoch": 0.45316486453650834,
+      "grad_norm": 1.8128470182418823,
+      "learning_rate": 0.0002,
+      "loss": 1.2071,
+      "step": 7460
+    },
+    {
+      "epoch": 0.45377232414044466,
+      "grad_norm": 1.2913196086883545,
+      "learning_rate": 0.0002,
+      "loss": 1.2202,
+      "step": 7470
+    },
+    {
+      "epoch": 0.454379783744381,
+      "grad_norm": 1.450271487236023,
+      "learning_rate": 0.0002,
+      "loss": 1.079,
+      "step": 7480
+    },
+    {
+      "epoch": 0.45498724334831736,
+      "grad_norm": 1.1549795866012573,
+      "learning_rate": 0.0002,
+      "loss": 1.2805,
+      "step": 7490
+    },
+    {
+      "epoch": 0.4555947029522537,
+      "grad_norm": 1.5973620414733887,
+      "learning_rate": 0.0002,
+      "loss": 1.2319,
+      "step": 7500
+    },
+    {
+      "epoch": 0.45620216255619,
+      "grad_norm": 1.6101380586624146,
+      "learning_rate": 0.0002,
+      "loss": 1.25,
+      "step": 7510
+    },
+    {
+      "epoch": 0.4568096221601263,
+      "grad_norm": 1.511681318283081,
+      "learning_rate": 0.0002,
+      "loss": 1.1875,
+      "step": 7520
+    },
+    {
+      "epoch": 0.4574170817640627,
+      "grad_norm": 1.3775086402893066,
+      "learning_rate": 0.0002,
+      "loss": 1.1924,
+      "step": 7530
+    },
+    {
+      "epoch": 0.458024541367999,
+      "grad_norm": 1.4383291006088257,
+      "learning_rate": 0.0002,
+      "loss": 1.1744,
+      "step": 7540
+    },
+    {
+      "epoch": 0.45863200097193535,
+      "grad_norm": 1.8540087938308716,
+      "learning_rate": 0.0002,
+      "loss": 1.2605,
+      "step": 7550
+    },
+    {
+      "epoch": 0.4592394605758717,
+      "grad_norm": 1.7484431266784668,
+      "learning_rate": 0.0002,
+      "loss": 1.1746,
+      "step": 7560
+    },
+    {
+      "epoch": 0.45984692017980805,
+      "grad_norm": 1.532613754272461,
+      "learning_rate": 0.0002,
+      "loss": 1.1702,
+      "step": 7570
+    },
+    {
+      "epoch": 0.46045437978374437,
+      "grad_norm": 1.348934292793274,
+      "learning_rate": 0.0002,
+      "loss": 1.2006,
+      "step": 7580
+    },
+    {
+      "epoch": 0.4610618393876807,
+      "grad_norm": 1.3784677982330322,
+      "learning_rate": 0.0002,
+      "loss": 1.18,
+      "step": 7590
+    },
+    {
+      "epoch": 0.46166929899161707,
+      "grad_norm": 1.3525645732879639,
+      "learning_rate": 0.0002,
+      "loss": 1.0906,
+      "step": 7600
+    },
+    {
+      "epoch": 0.4622767585955534,
+      "grad_norm": 1.6208430528640747,
+      "learning_rate": 0.0002,
+      "loss": 1.1865,
+      "step": 7610
+    },
+    {
+      "epoch": 0.4628842181994897,
+      "grad_norm": 1.4005343914031982,
+      "learning_rate": 0.0002,
+      "loss": 1.2291,
+      "step": 7620
+    },
+    {
+      "epoch": 0.4634916778034261,
+      "grad_norm": 1.6267120838165283,
+      "learning_rate": 0.0002,
+      "loss": 1.2027,
+      "step": 7630
+    },
+    {
+      "epoch": 0.4640991374073624,
+      "grad_norm": 1.5837438106536865,
+      "learning_rate": 0.0002,
+      "loss": 1.2381,
+      "step": 7640
+    },
+    {
+      "epoch": 0.46470659701129874,
+      "grad_norm": 1.5585278272628784,
+      "learning_rate": 0.0002,
+      "loss": 1.1726,
+      "step": 7650
+    },
+    {
+      "epoch": 0.46531405661523506,
+      "grad_norm": 1.585730791091919,
+      "learning_rate": 0.0002,
+      "loss": 1.1708,
+      "step": 7660
+    },
+    {
+      "epoch": 0.46592151621917144,
+      "grad_norm": 1.5836211442947388,
+      "learning_rate": 0.0002,
+      "loss": 1.1671,
+      "step": 7670
+    },
+    {
+      "epoch": 0.46652897582310776,
+      "grad_norm": 1.6574169397354126,
+      "learning_rate": 0.0002,
+      "loss": 1.2585,
+      "step": 7680
+    },
+    {
+      "epoch": 0.4671364354270441,
+      "grad_norm": 1.5859861373901367,
+      "learning_rate": 0.0002,
+      "loss": 1.228,
+      "step": 7690
+    },
+    {
+      "epoch": 0.46774389503098046,
+      "grad_norm": 1.4105859994888306,
+      "learning_rate": 0.0002,
+      "loss": 1.2245,
+      "step": 7700
+    },
+    {
+      "epoch": 0.4683513546349168,
+      "grad_norm": 1.553499460220337,
+      "learning_rate": 0.0002,
+      "loss": 1.2034,
+      "step": 7710
+    },
+    {
+      "epoch": 0.4689588142388531,
+      "grad_norm": 1.5208845138549805,
+      "learning_rate": 0.0002,
+      "loss": 1.1334,
+      "step": 7720
+    },
+    {
+      "epoch": 0.46956627384278943,
+      "grad_norm": 1.4192719459533691,
+      "learning_rate": 0.0002,
+      "loss": 1.2398,
+      "step": 7730
+    },
+    {
+      "epoch": 0.4701737334467258,
+      "grad_norm": 1.2706866264343262,
+      "learning_rate": 0.0002,
+      "loss": 1.1827,
+      "step": 7740
+    },
+    {
+      "epoch": 0.47078119305066213,
+      "grad_norm": 1.5176726579666138,
+      "learning_rate": 0.0002,
+      "loss": 1.1615,
+      "step": 7750
+    },
+    {
+      "epoch": 0.47138865265459845,
+      "grad_norm": 1.45756196975708,
+      "learning_rate": 0.0002,
+      "loss": 1.1666,
+      "step": 7760
+    },
+    {
+      "epoch": 0.47199611225853483,
+      "grad_norm": 1.2748565673828125,
+      "learning_rate": 0.0002,
+      "loss": 1.2276,
+      "step": 7770
+    },
+    {
+      "epoch": 0.47260357186247115,
+      "grad_norm": 1.5021861791610718,
+      "learning_rate": 0.0002,
+      "loss": 1.1292,
+      "step": 7780
+    },
+    {
+      "epoch": 0.4732110314664075,
+      "grad_norm": 1.4229892492294312,
+      "learning_rate": 0.0002,
+      "loss": 1.1646,
+      "step": 7790
+    },
+    {
+      "epoch": 0.4738184910703438,
+      "grad_norm": 1.4374492168426514,
+      "learning_rate": 0.0002,
+      "loss": 1.3199,
+      "step": 7800
+    },
+    {
+      "epoch": 0.4744259506742802,
+      "grad_norm": 1.5280470848083496,
+      "learning_rate": 0.0002,
+      "loss": 1.2042,
+      "step": 7810
+    },
+    {
+      "epoch": 0.4750334102782165,
+      "grad_norm": 1.6115370988845825,
+      "learning_rate": 0.0002,
+      "loss": 1.2026,
+      "step": 7820
+    },
+    {
+      "epoch": 0.4756408698821528,
+      "grad_norm": 1.3650151491165161,
+      "learning_rate": 0.0002,
+      "loss": 1.1743,
+      "step": 7830
+    },
+    {
+      "epoch": 0.4762483294860892,
+      "grad_norm": 1.3961408138275146,
+      "learning_rate": 0.0002,
+      "loss": 1.1814,
+      "step": 7840
+    },
+    {
+      "epoch": 0.4768557890900255,
+      "grad_norm": 1.5285776853561401,
+      "learning_rate": 0.0002,
+      "loss": 1.1876,
+      "step": 7850
+    },
+    {
+      "epoch": 0.47746324869396184,
+      "grad_norm": 1.5390294790267944,
+      "learning_rate": 0.0002,
+      "loss": 1.2132,
+      "step": 7860
+    },
+    {
+      "epoch": 0.4780707082978982,
+      "grad_norm": 1.9355554580688477,
+      "learning_rate": 0.0002,
+      "loss": 1.2507,
+      "step": 7870
+    },
+    {
+      "epoch": 0.47867816790183454,
+      "grad_norm": 1.5551636219024658,
+      "learning_rate": 0.0002,
+      "loss": 1.1708,
+      "step": 7880
+    },
+    {
+      "epoch": 0.47928562750577086,
+      "grad_norm": 1.2486780881881714,
+      "learning_rate": 0.0002,
+      "loss": 1.2163,
+      "step": 7890
+    },
+    {
+      "epoch": 0.4798930871097072,
+      "grad_norm": 1.60591459274292,
+      "learning_rate": 0.0002,
+      "loss": 1.1202,
+      "step": 7900
+    },
+    {
+      "epoch": 0.48050054671364356,
+      "grad_norm": 1.3844667673110962,
+      "learning_rate": 0.0002,
+      "loss": 1.2513,
+      "step": 7910
+    },
+    {
+      "epoch": 0.4811080063175799,
+      "grad_norm": 1.56134831905365,
+      "learning_rate": 0.0002,
+      "loss": 1.1703,
+      "step": 7920
+    },
+    {
+      "epoch": 0.4817154659215162,
+      "grad_norm": 1.6113072633743286,
+      "learning_rate": 0.0002,
+      "loss": 1.2161,
+      "step": 7930
+    },
+    {
+      "epoch": 0.4823229255254526,
+      "grad_norm": 1.5860167741775513,
+      "learning_rate": 0.0002,
+      "loss": 1.1448,
+      "step": 7940
+    },
+    {
+      "epoch": 0.4829303851293889,
+      "grad_norm": 1.349814772605896,
+      "learning_rate": 0.0002,
+      "loss": 1.1548,
+      "step": 7950
+    },
+    {
+      "epoch": 0.48353784473332523,
+      "grad_norm": 1.5404295921325684,
+      "learning_rate": 0.0002,
+      "loss": 1.2584,
+      "step": 7960
+    },
+    {
+      "epoch": 0.48414530433726155,
+      "grad_norm": 1.6452714204788208,
+      "learning_rate": 0.0002,
+      "loss": 1.2712,
+      "step": 7970
+    },
+    {
+      "epoch": 0.48475276394119793,
+      "grad_norm": 1.7067540884017944,
+      "learning_rate": 0.0002,
+      "loss": 1.1459,
+      "step": 7980
+    },
+    {
+      "epoch": 0.48536022354513425,
+      "grad_norm": 1.4783881902694702,
+      "learning_rate": 0.0002,
+      "loss": 1.2514,
+      "step": 7990
+    },
+    {
+      "epoch": 0.4859676831490706,
+      "grad_norm": 1.422806739807129,
+      "learning_rate": 0.0002,
+      "loss": 1.2685,
+      "step": 8000
+    },
+    {
+      "epoch": 0.48657514275300695,
+      "grad_norm": 1.329826831817627,
+      "learning_rate": 0.0002,
+      "loss": 1.1713,
+      "step": 8010
+    },
+    {
+      "epoch": 0.4871826023569433,
+      "grad_norm": 1.4270880222320557,
+      "learning_rate": 0.0002,
+      "loss": 1.2034,
+      "step": 8020
+    },
+    {
+      "epoch": 0.4877900619608796,
+      "grad_norm": 1.260317325592041,
+      "learning_rate": 0.0002,
+      "loss": 1.1879,
+      "step": 8030
+    },
+    {
+      "epoch": 0.4883975215648159,
+      "grad_norm": 1.5208438634872437,
+      "learning_rate": 0.0002,
+      "loss": 1.289,
+      "step": 8040
+    },
+    {
+      "epoch": 0.4890049811687523,
+      "grad_norm": 1.396694302558899,
+      "learning_rate": 0.0002,
+      "loss": 1.1453,
+      "step": 8050
+    },
+    {
+      "epoch": 0.4896124407726886,
+      "grad_norm": 1.7723726034164429,
+      "learning_rate": 0.0002,
+      "loss": 1.2369,
+      "step": 8060
+    },
+    {
+      "epoch": 0.49021990037662494,
+      "grad_norm": 1.485266923904419,
+      "learning_rate": 0.0002,
+      "loss": 1.1721,
+      "step": 8070
+    },
+    {
+      "epoch": 0.4908273599805613,
+      "grad_norm": 1.3336842060089111,
+      "learning_rate": 0.0002,
+      "loss": 1.1916,
+      "step": 8080
+    },
+    {
+      "epoch": 0.49143481958449764,
+      "grad_norm": 1.372721552848816,
+      "learning_rate": 0.0002,
+      "loss": 1.2192,
+      "step": 8090
+    },
+    {
+      "epoch": 0.49204227918843396,
+      "grad_norm": 1.5978894233703613,
+      "learning_rate": 0.0002,
+      "loss": 1.1239,
+      "step": 8100
+    },
+    {
+      "epoch": 0.4926497387923703,
+      "grad_norm": 1.6521567106246948,
+      "learning_rate": 0.0002,
+      "loss": 1.189,
+      "step": 8110
+    },
+    {
+      "epoch": 0.49325719839630666,
+      "grad_norm": 1.4695961475372314,
+      "learning_rate": 0.0002,
+      "loss": 1.1799,
+      "step": 8120
+    },
+    {
+      "epoch": 0.493864658000243,
+      "grad_norm": 1.6711095571517944,
+      "learning_rate": 0.0002,
+      "loss": 1.2145,
+      "step": 8130
+    },
+    {
+      "epoch": 0.4944721176041793,
+      "grad_norm": 1.410640001296997,
+      "learning_rate": 0.0002,
+      "loss": 1.2466,
+      "step": 8140
+    },
+    {
+      "epoch": 0.4950795772081157,
+      "grad_norm": 1.5452176332473755,
+      "learning_rate": 0.0002,
+      "loss": 1.257,
+      "step": 8150
+    },
+    {
+      "epoch": 0.495687036812052,
+      "grad_norm": 1.5476586818695068,
+      "learning_rate": 0.0002,
+      "loss": 1.247,
+      "step": 8160
+    },
+    {
+      "epoch": 0.49629449641598833,
+      "grad_norm": 1.428282380104065,
+      "learning_rate": 0.0002,
+      "loss": 1.2456,
+      "step": 8170
+    },
+    {
+      "epoch": 0.49690195601992465,
+      "grad_norm": 1.4470287561416626,
+      "learning_rate": 0.0002,
+      "loss": 1.1985,
+      "step": 8180
+    },
+    {
+      "epoch": 0.49750941562386103,
+      "grad_norm": 1.3442034721374512,
+      "learning_rate": 0.0002,
+      "loss": 1.1742,
+      "step": 8190
+    },
+    {
+      "epoch": 0.49811687522779735,
+      "grad_norm": 1.4205583333969116,
+      "learning_rate": 0.0002,
+      "loss": 1.1945,
+      "step": 8200
+    },
+    {
+      "epoch": 0.4987243348317337,
+      "grad_norm": 1.732121467590332,
+      "learning_rate": 0.0002,
+      "loss": 1.1702,
+      "step": 8210
+    },
+    {
+      "epoch": 0.49933179443567005,
+      "grad_norm": 2.0065855979919434,
+      "learning_rate": 0.0002,
+      "loss": 1.2247,
+      "step": 8220
+    },
+    {
+      "epoch": 0.4999392540396064,
+      "grad_norm": 1.3520867824554443,
+      "learning_rate": 0.0002,
+      "loss": 1.2432,
+      "step": 8230
+    },
+    {
+      "epoch": 0.5005467136435428,
+      "grad_norm": 1.6998947858810425,
+      "learning_rate": 0.0002,
+      "loss": 1.1556,
+      "step": 8240
+    },
+    {
+      "epoch": 0.501154173247479,
+      "grad_norm": 1.4172024726867676,
+      "learning_rate": 0.0002,
+      "loss": 1.1321,
+      "step": 8250
+    },
+    {
+      "epoch": 0.5017616328514154,
+      "grad_norm": 1.5547268390655518,
+      "learning_rate": 0.0002,
+      "loss": 1.2796,
+      "step": 8260
+    },
+    {
+      "epoch": 0.5023690924553518,
+      "grad_norm": 2.0130136013031006,
+      "learning_rate": 0.0002,
+      "loss": 1.1534,
+      "step": 8270
+    },
+    {
+      "epoch": 0.502976552059288,
+      "grad_norm": 1.6734105348587036,
+      "learning_rate": 0.0002,
+      "loss": 1.1958,
+      "step": 8280
+    },
+    {
+      "epoch": 0.5035840116632244,
+      "grad_norm": 1.3117741346359253,
+      "learning_rate": 0.0002,
+      "loss": 1.2077,
+      "step": 8290
+    },
+    {
+      "epoch": 0.5041914712671607,
+      "grad_norm": 1.6821929216384888,
+      "learning_rate": 0.0002,
+      "loss": 1.2818,
+      "step": 8300
+    },
+    {
+      "epoch": 0.5047989308710971,
+      "grad_norm": 1.4614527225494385,
+      "learning_rate": 0.0002,
+      "loss": 1.1808,
+      "step": 8310
+    },
+    {
+      "epoch": 0.5054063904750334,
+      "grad_norm": 1.437494158744812,
+      "learning_rate": 0.0002,
+      "loss": 1.2735,
+      "step": 8320
+    },
+    {
+      "epoch": 0.5060138500789697,
+      "grad_norm": 1.562290906906128,
+      "learning_rate": 0.0002,
+      "loss": 1.2962,
+      "step": 8330
+    },
+    {
+      "epoch": 0.5066213096829061,
+      "grad_norm": 1.5021467208862305,
+      "learning_rate": 0.0002,
+      "loss": 1.1741,
+      "step": 8340
+    },
+    {
+      "epoch": 0.5072287692868425,
+      "grad_norm": 1.4091651439666748,
+      "learning_rate": 0.0002,
+      "loss": 1.181,
+      "step": 8350
+    },
+    {
+      "epoch": 0.5078362288907787,
+      "grad_norm": 1.486215353012085,
+      "learning_rate": 0.0002,
+      "loss": 1.2157,
+      "step": 8360
+    },
+    {
+      "epoch": 0.5084436884947151,
+      "grad_norm": 1.655120849609375,
+      "learning_rate": 0.0002,
+      "loss": 1.2012,
+      "step": 8370
+    },
+    {
+      "epoch": 0.5090511480986515,
+      "grad_norm": 1.5554999113082886,
+      "learning_rate": 0.0002,
+      "loss": 1.1437,
+      "step": 8380
+    },
+    {
+      "epoch": 0.5096586077025878,
+      "grad_norm": 1.3103996515274048,
+      "learning_rate": 0.0002,
+      "loss": 1.2077,
+      "step": 8390
+    },
+    {
+      "epoch": 0.5102660673065241,
+      "grad_norm": 1.3018100261688232,
+      "learning_rate": 0.0002,
+      "loss": 1.199,
+      "step": 8400
+    },
+    {
+      "epoch": 0.5108735269104605,
+      "grad_norm": 1.3071420192718506,
+      "learning_rate": 0.0002,
+      "loss": 1.2443,
+      "step": 8410
+    },
+    {
+      "epoch": 0.5114809865143968,
+      "grad_norm": 1.6154550313949585,
+      "learning_rate": 0.0002,
+      "loss": 1.1804,
+      "step": 8420
+    },
+    {
+      "epoch": 0.5120884461183332,
+      "grad_norm": 1.4893248081207275,
+      "learning_rate": 0.0002,
+      "loss": 1.1531,
+      "step": 8430
+    },
+    {
+      "epoch": 0.5126959057222694,
+      "grad_norm": 1.3449759483337402,
+      "learning_rate": 0.0002,
+      "loss": 1.2168,
+      "step": 8440
+    },
+    {
+      "epoch": 0.5133033653262058,
+      "grad_norm": 1.3129184246063232,
+      "learning_rate": 0.0002,
+      "loss": 1.1838,
+      "step": 8450
+    },
+    {
+      "epoch": 0.5139108249301422,
+      "grad_norm": 1.427881121635437,
+      "learning_rate": 0.0002,
+      "loss": 1.1849,
+      "step": 8460
+    },
+    {
+      "epoch": 0.5145182845340784,
+      "grad_norm": 1.4069163799285889,
+      "learning_rate": 0.0002,
+      "loss": 1.1692,
+      "step": 8470
+    },
+    {
+      "epoch": 0.5151257441380148,
+      "grad_norm": 1.299394965171814,
+      "learning_rate": 0.0002,
+      "loss": 1.1636,
+      "step": 8480
+    },
+    {
+      "epoch": 0.5157332037419512,
+      "grad_norm": 1.3414911031723022,
+      "learning_rate": 0.0002,
+      "loss": 1.1956,
+      "step": 8490
+    },
+    {
+      "epoch": 0.5163406633458875,
+      "grad_norm": 1.4938147068023682,
+      "learning_rate": 0.0002,
+      "loss": 1.1744,
+      "step": 8500
+    },
+    {
+      "epoch": 0.5169481229498238,
+      "grad_norm": 1.7157621383666992,
+      "learning_rate": 0.0002,
+      "loss": 1.1168,
+      "step": 8510
+    },
+    {
+      "epoch": 0.5175555825537602,
+      "grad_norm": 1.4140446186065674,
+      "learning_rate": 0.0002,
+      "loss": 1.1594,
+      "step": 8520
+    },
+    {
+      "epoch": 0.5181630421576965,
+      "grad_norm": 1.7376991510391235,
+      "learning_rate": 0.0002,
+      "loss": 1.1591,
+      "step": 8530
+    },
+    {
+      "epoch": 0.5187705017616329,
+      "grad_norm": 1.7699772119522095,
+      "learning_rate": 0.0002,
+      "loss": 1.0878,
+      "step": 8540
+    },
+    {
+      "epoch": 0.5193779613655692,
+      "grad_norm": 1.2922927141189575,
+      "learning_rate": 0.0002,
+      "loss": 1.1929,
+      "step": 8550
+    },
+    {
+      "epoch": 0.5199854209695055,
+      "grad_norm": 1.6389902830123901,
+      "learning_rate": 0.0002,
+      "loss": 1.2001,
+      "step": 8560
+    },
+    {
+      "epoch": 0.5205928805734419,
+      "grad_norm": 1.4543099403381348,
+      "learning_rate": 0.0002,
+      "loss": 1.2965,
+      "step": 8570
+    },
+    {
+      "epoch": 0.5212003401773782,
+      "grad_norm": 1.6899205446243286,
+      "learning_rate": 0.0002,
+      "loss": 1.2332,
+      "step": 8580
+    },
+    {
+      "epoch": 0.5218077997813145,
+      "grad_norm": 1.3092981576919556,
+      "learning_rate": 0.0002,
+      "loss": 1.2341,
+      "step": 8590
+    },
+    {
+      "epoch": 0.5224152593852509,
+      "grad_norm": 1.3344212770462036,
+      "learning_rate": 0.0002,
+      "loss": 1.1844,
+      "step": 8600
+    },
+    {
+      "epoch": 0.5230227189891872,
+      "grad_norm": 1.2653930187225342,
+      "learning_rate": 0.0002,
+      "loss": 1.102,
+      "step": 8610
+    },
+    {
+      "epoch": 0.5236301785931236,
+      "grad_norm": 1.3919548988342285,
+      "learning_rate": 0.0002,
+      "loss": 1.1462,
+      "step": 8620
+    },
+    {
+      "epoch": 0.5242376381970599,
+      "grad_norm": 1.671441674232483,
+      "learning_rate": 0.0002,
+      "loss": 1.1795,
+      "step": 8630
+    },
+    {
+      "epoch": 0.5248450978009962,
+      "grad_norm": 1.48072350025177,
+      "learning_rate": 0.0002,
+      "loss": 1.2004,
+      "step": 8640
+    },
+    {
+      "epoch": 0.5254525574049326,
+      "grad_norm": 1.1946160793304443,
+      "learning_rate": 0.0002,
+      "loss": 1.2139,
+      "step": 8650
+    },
+    {
+      "epoch": 0.526060017008869,
+      "grad_norm": 1.396958827972412,
+      "learning_rate": 0.0002,
+      "loss": 1.2258,
+      "step": 8660
+    },
+    {
+      "epoch": 0.5266674766128052,
+      "grad_norm": 1.610670566558838,
+      "learning_rate": 0.0002,
+      "loss": 1.1471,
+      "step": 8670
+    },
+    {
+      "epoch": 0.5272749362167416,
+      "grad_norm": 1.920641541481018,
+      "learning_rate": 0.0002,
+      "loss": 1.2322,
+      "step": 8680
+    },
+    {
+      "epoch": 0.527882395820678,
+      "grad_norm": 1.5889919996261597,
+      "learning_rate": 0.0002,
+      "loss": 1.1512,
+      "step": 8690
+    },
+    {
+      "epoch": 0.5284898554246142,
+      "grad_norm": 1.353811264038086,
+      "learning_rate": 0.0002,
+      "loss": 1.1774,
+      "step": 8700
+    },
+    {
+      "epoch": 0.5290973150285506,
+      "grad_norm": 1.4460562467575073,
+      "learning_rate": 0.0002,
+      "loss": 1.1651,
+      "step": 8710
+    },
+    {
+      "epoch": 0.5297047746324869,
+      "grad_norm": 1.5538369417190552,
+      "learning_rate": 0.0002,
+      "loss": 1.1919,
+      "step": 8720
+    },
+    {
+      "epoch": 0.5303122342364233,
+      "grad_norm": 1.3139142990112305,
+      "learning_rate": 0.0002,
+      "loss": 1.1355,
+      "step": 8730
+    },
+    {
+      "epoch": 0.5309196938403596,
+      "grad_norm": 1.3838707208633423,
+      "learning_rate": 0.0002,
+      "loss": 1.1766,
+      "step": 8740
+    },
+    {
+      "epoch": 0.5315271534442959,
+      "grad_norm": 1.8593472242355347,
+      "learning_rate": 0.0002,
+      "loss": 1.3266,
+      "step": 8750
+    },
+    {
+      "epoch": 0.5321346130482323,
+      "grad_norm": 1.32051420211792,
+      "learning_rate": 0.0002,
+      "loss": 1.2102,
+      "step": 8760
+    },
+    {
+      "epoch": 0.5327420726521687,
+      "grad_norm": 1.7253473997116089,
+      "learning_rate": 0.0002,
+      "loss": 1.2025,
+      "step": 8770
+    },
+    {
+      "epoch": 0.5333495322561049,
+      "grad_norm": 1.6720499992370605,
+      "learning_rate": 0.0002,
+      "loss": 1.1767,
+      "step": 8780
+    },
+    {
+      "epoch": 0.5339569918600413,
+      "grad_norm": 1.5546365976333618,
+      "learning_rate": 0.0002,
+      "loss": 1.1597,
+      "step": 8790
+    },
+    {
+      "epoch": 0.5345644514639777,
+      "grad_norm": 1.476245403289795,
+      "learning_rate": 0.0002,
+      "loss": 1.1658,
+      "step": 8800
+    },
+    {
+      "epoch": 0.535171911067914,
+      "grad_norm": 1.9548064470291138,
+      "learning_rate": 0.0002,
+      "loss": 1.1858,
+      "step": 8810
+    },
+    {
+      "epoch": 0.5357793706718503,
+      "grad_norm": 1.7661584615707397,
+      "learning_rate": 0.0002,
+      "loss": 1.2401,
+      "step": 8820
+    },
+    {
+      "epoch": 0.5363868302757867,
+      "grad_norm": 1.3066457509994507,
+      "learning_rate": 0.0002,
+      "loss": 1.2359,
+      "step": 8830
+    },
+    {
+      "epoch": 0.536994289879723,
+      "grad_norm": 1.6530494689941406,
+      "learning_rate": 0.0002,
+      "loss": 1.1856,
+      "step": 8840
+    },
+    {
+      "epoch": 0.5376017494836594,
+      "grad_norm": 1.752524733543396,
+      "learning_rate": 0.0002,
+      "loss": 1.194,
+      "step": 8850
+    },
+    {
+      "epoch": 0.5382092090875957,
+      "grad_norm": 1.2837939262390137,
+      "learning_rate": 0.0002,
+      "loss": 1.171,
+      "step": 8860
+    },
+    {
+      "epoch": 0.538816668691532,
+      "grad_norm": 1.5185567140579224,
+      "learning_rate": 0.0002,
+      "loss": 1.1658,
+      "step": 8870
+    },
+    {
+      "epoch": 0.5394241282954684,
+      "grad_norm": 1.5040532350540161,
+      "learning_rate": 0.0002,
+      "loss": 1.2165,
+      "step": 8880
+    },
+    {
+      "epoch": 0.5400315878994046,
+      "grad_norm": 1.4181594848632812,
+      "learning_rate": 0.0002,
+      "loss": 1.2066,
+      "step": 8890
+    },
+    {
+      "epoch": 0.540639047503341,
+      "grad_norm": 1.5156562328338623,
+      "learning_rate": 0.0002,
+      "loss": 1.1431,
+      "step": 8900
+    },
+    {
+      "epoch": 0.5412465071072774,
+      "grad_norm": 1.576412558555603,
+      "learning_rate": 0.0002,
+      "loss": 1.2022,
+      "step": 8910
+    },
+    {
+      "epoch": 0.5418539667112137,
+      "grad_norm": 1.4076834917068481,
+      "learning_rate": 0.0002,
+      "loss": 1.1449,
+      "step": 8920
+    },
+    {
+      "epoch": 0.54246142631515,
+      "grad_norm": 1.4449139833450317,
+      "learning_rate": 0.0002,
+      "loss": 1.1804,
+      "step": 8930
+    },
+    {
+      "epoch": 0.5430688859190864,
+      "grad_norm": 1.3994479179382324,
+      "learning_rate": 0.0002,
+      "loss": 1.2118,
+      "step": 8940
+    },
+    {
+      "epoch": 0.5436763455230227,
+      "grad_norm": 1.4858759641647339,
+      "learning_rate": 0.0002,
+      "loss": 1.2036,
+      "step": 8950
+    },
+    {
+      "epoch": 0.5442838051269591,
+      "grad_norm": 1.493045687675476,
+      "learning_rate": 0.0002,
+      "loss": 1.2006,
+      "step": 8960
+    },
+    {
+      "epoch": 0.5448912647308954,
+      "grad_norm": 1.4153680801391602,
+      "learning_rate": 0.0002,
+      "loss": 1.1361,
+      "step": 8970
+    },
+    {
+      "epoch": 0.5454987243348317,
+      "grad_norm": 1.421522617340088,
+      "learning_rate": 0.0002,
+      "loss": 1.2134,
+      "step": 8980
+    },
+    {
+      "epoch": 0.5461061839387681,
+      "grad_norm": 1.525323748588562,
+      "learning_rate": 0.0002,
+      "loss": 1.2022,
+      "step": 8990
+    },
+    {
+      "epoch": 0.5467136435427045,
+      "grad_norm": 1.6441048383712769,
+      "learning_rate": 0.0002,
+      "loss": 1.2558,
+      "step": 9000
+    },
+    {
+      "epoch": 0.5473211031466407,
+      "grad_norm": 1.4588571786880493,
+      "learning_rate": 0.0002,
+      "loss": 1.232,
+      "step": 9010
+    },
+    {
+      "epoch": 0.5479285627505771,
+      "grad_norm": 1.3689075708389282,
+      "learning_rate": 0.0002,
+      "loss": 1.1705,
+      "step": 9020
+    },
+    {
+      "epoch": 0.5485360223545134,
+      "grad_norm": 1.4682143926620483,
+      "learning_rate": 0.0002,
+      "loss": 1.1993,
+      "step": 9030
+    },
+    {
+      "epoch": 0.5491434819584498,
+      "grad_norm": 1.4524134397506714,
+      "learning_rate": 0.0002,
+      "loss": 1.2297,
+      "step": 9040
+    },
+    {
+      "epoch": 0.5497509415623861,
+      "grad_norm": 1.4454007148742676,
+      "learning_rate": 0.0002,
+      "loss": 1.2216,
+      "step": 9050
+    },
+    {
+      "epoch": 0.5503584011663224,
+      "grad_norm": 1.9346940517425537,
+      "learning_rate": 0.0002,
+      "loss": 1.2194,
+      "step": 9060
+    },
+    {
+      "epoch": 0.5509658607702588,
+      "grad_norm": 2.2555973529815674,
+      "learning_rate": 0.0002,
+      "loss": 1.1502,
+      "step": 9070
+    },
+    {
+      "epoch": 0.5515733203741952,
+      "grad_norm": 1.398390531539917,
+      "learning_rate": 0.0002,
+      "loss": 1.1898,
+      "step": 9080
+    },
+    {
+      "epoch": 0.5521807799781314,
+      "grad_norm": 1.600623607635498,
+      "learning_rate": 0.0002,
+      "loss": 1.3627,
+      "step": 9090
+    },
+    {
+      "epoch": 0.5527882395820678,
+      "grad_norm": 1.5447672605514526,
+      "learning_rate": 0.0002,
+      "loss": 1.13,
+      "step": 9100
+    },
+    {
+      "epoch": 0.5533956991860042,
+      "grad_norm": 1.5209197998046875,
+      "learning_rate": 0.0002,
+      "loss": 1.1547,
+      "step": 9110
+    },
+    {
+      "epoch": 0.5540031587899404,
+      "grad_norm": 1.3905166387557983,
+      "learning_rate": 0.0002,
+      "loss": 1.163,
+      "step": 9120
+    },
+    {
+      "epoch": 0.5546106183938768,
+      "grad_norm": 1.5151902437210083,
+      "learning_rate": 0.0002,
+      "loss": 1.1953,
+      "step": 9130
+    },
+    {
+      "epoch": 0.5552180779978132,
+      "grad_norm": 1.2856061458587646,
+      "learning_rate": 0.0002,
+      "loss": 1.1521,
+      "step": 9140
+    },
+    {
+      "epoch": 0.5558255376017495,
+      "grad_norm": 1.4850786924362183,
+      "learning_rate": 0.0002,
+      "loss": 1.2062,
+      "step": 9150
+    },
+    {
+      "epoch": 0.5564329972056858,
+      "grad_norm": 1.5212926864624023,
+      "learning_rate": 0.0002,
+      "loss": 1.2404,
+      "step": 9160
+    },
+    {
+      "epoch": 0.5570404568096221,
+      "grad_norm": 1.323756456375122,
+      "learning_rate": 0.0002,
+      "loss": 1.1079,
+      "step": 9170
+    },
+    {
+      "epoch": 0.5576479164135585,
+      "grad_norm": 1.645654559135437,
+      "learning_rate": 0.0002,
+      "loss": 1.1951,
+      "step": 9180
+    },
+    {
+      "epoch": 0.5582553760174949,
+      "grad_norm": 1.4693900346755981,
+      "learning_rate": 0.0002,
+      "loss": 1.2511,
+      "step": 9190
+    },
+    {
+      "epoch": 0.5588628356214311,
+      "grad_norm": 1.5780327320098877,
+      "learning_rate": 0.0002,
+      "loss": 1.2349,
+      "step": 9200
+    },
+    {
+      "epoch": 0.5594702952253675,
+      "grad_norm": 1.5587654113769531,
+      "learning_rate": 0.0002,
+      "loss": 1.2189,
+      "step": 9210
+    },
+    {
+      "epoch": 0.5600777548293039,
+      "grad_norm": 1.4344000816345215,
+      "learning_rate": 0.0002,
+      "loss": 1.2613,
+      "step": 9220
+    },
+    {
+      "epoch": 0.5606852144332402,
+      "grad_norm": 1.2892564535140991,
+      "learning_rate": 0.0002,
+      "loss": 1.1674,
+      "step": 9230
+    },
+    {
+      "epoch": 0.5612926740371765,
+      "grad_norm": 1.510473608970642,
+      "learning_rate": 0.0002,
+      "loss": 1.1735,
+      "step": 9240
+    },
+    {
+      "epoch": 0.5619001336411129,
+      "grad_norm": 1.5466214418411255,
+      "learning_rate": 0.0002,
+      "loss": 1.1557,
+      "step": 9250
+    },
+    {
+      "epoch": 0.5625075932450492,
+      "grad_norm": 1.3267711400985718,
+      "learning_rate": 0.0002,
+      "loss": 1.1495,
+      "step": 9260
+    },
+    {
+      "epoch": 0.5631150528489856,
+      "grad_norm": 1.500229835510254,
+      "learning_rate": 0.0002,
+      "loss": 1.1704,
+      "step": 9270
+    },
+    {
+      "epoch": 0.5637225124529219,
+      "grad_norm": 1.4810044765472412,
+      "learning_rate": 0.0002,
+      "loss": 1.2374,
+      "step": 9280
+    },
+    {
+      "epoch": 0.5643299720568582,
+      "grad_norm": 1.819751501083374,
+      "learning_rate": 0.0002,
+      "loss": 1.1665,
+      "step": 9290
+    },
+    {
+      "epoch": 0.5649374316607946,
+      "grad_norm": 1.721070647239685,
+      "learning_rate": 0.0002,
+      "loss": 1.2074,
+      "step": 9300
+    },
+    {
+      "epoch": 0.5655448912647308,
+      "grad_norm": 1.28892183303833,
+      "learning_rate": 0.0002,
+      "loss": 1.2075,
+      "step": 9310
+    },
+    {
+      "epoch": 0.5661523508686672,
+      "grad_norm": 1.4271416664123535,
+      "learning_rate": 0.0002,
+      "loss": 1.1722,
+      "step": 9320
+    },
+    {
+      "epoch": 0.5667598104726036,
+      "grad_norm": 1.4250295162200928,
+      "learning_rate": 0.0002,
+      "loss": 1.2099,
+      "step": 9330
+    },
+    {
+      "epoch": 0.5673672700765399,
+      "grad_norm": 1.3933831453323364,
+      "learning_rate": 0.0002,
+      "loss": 1.2214,
+      "step": 9340
+    },
+    {
+      "epoch": 0.5679747296804762,
+      "grad_norm": 1.4927361011505127,
+      "learning_rate": 0.0002,
+      "loss": 1.1964,
+      "step": 9350
+    },
+    {
+      "epoch": 0.5685821892844126,
+      "grad_norm": 1.4718621969223022,
+      "learning_rate": 0.0002,
+      "loss": 1.2009,
+      "step": 9360
+    },
+    {
+      "epoch": 0.5691896488883489,
+      "grad_norm": 1.4115561246871948,
+      "learning_rate": 0.0002,
+      "loss": 1.1992,
+      "step": 9370
+    },
+    {
+      "epoch": 0.5697971084922853,
+      "grad_norm": 1.333713173866272,
+      "learning_rate": 0.0002,
+      "loss": 1.1461,
+      "step": 9380
+    },
+    {
+      "epoch": 0.5704045680962216,
+      "grad_norm": 1.387205719947815,
+      "learning_rate": 0.0002,
+      "loss": 1.1278,
+      "step": 9390
+    },
+    {
+      "epoch": 0.5710120277001579,
+      "grad_norm": 1.4636300802230835,
+      "learning_rate": 0.0002,
+      "loss": 1.122,
+      "step": 9400
+    },
+    {
+      "epoch": 0.5716194873040943,
+      "grad_norm": 1.4931930303573608,
+      "learning_rate": 0.0002,
+      "loss": 1.1957,
+      "step": 9410
+    },
+    {
+      "epoch": 0.5722269469080307,
+      "grad_norm": 1.539544939994812,
+      "learning_rate": 0.0002,
+      "loss": 1.1724,
+      "step": 9420
+    },
+    {
+      "epoch": 0.5728344065119669,
+      "grad_norm": 1.4466699361801147,
+      "learning_rate": 0.0002,
+      "loss": 1.1914,
+      "step": 9430
+    },
+    {
+      "epoch": 0.5734418661159033,
+      "grad_norm": 1.485856056213379,
+      "learning_rate": 0.0002,
+      "loss": 1.1715,
+      "step": 9440
+    },
+    {
+      "epoch": 0.5740493257198396,
+      "grad_norm": 1.6009708642959595,
+      "learning_rate": 0.0002,
+      "loss": 1.2944,
+      "step": 9450
+    },
+    {
+      "epoch": 0.574656785323776,
+      "grad_norm": 1.5014270544052124,
+      "learning_rate": 0.0002,
+      "loss": 1.183,
+      "step": 9460
+    },
+    {
+      "epoch": 0.5752642449277123,
+      "grad_norm": 1.5157901048660278,
+      "learning_rate": 0.0002,
+      "loss": 1.1828,
+      "step": 9470
+    },
+    {
+      "epoch": 0.5758717045316486,
+      "grad_norm": 1.2542632818222046,
+      "learning_rate": 0.0002,
+      "loss": 1.1653,
+      "step": 9480
+    },
+    {
+      "epoch": 0.576479164135585,
+      "grad_norm": 1.5954726934432983,
+      "learning_rate": 0.0002,
+      "loss": 1.3099,
+      "step": 9490
+    },
+    {
+      "epoch": 0.5770866237395214,
+      "grad_norm": 1.5456738471984863,
+      "learning_rate": 0.0002,
+      "loss": 1.2264,
+      "step": 9500
+    },
+    {
+      "epoch": 0.5776940833434576,
+      "grad_norm": 1.610556721687317,
+      "learning_rate": 0.0002,
+      "loss": 1.1486,
+      "step": 9510
+    },
+    {
+      "epoch": 0.578301542947394,
+      "grad_norm": 1.5603747367858887,
+      "learning_rate": 0.0002,
+      "loss": 1.2336,
+      "step": 9520
+    },
+    {
+      "epoch": 0.5789090025513304,
+      "grad_norm": 1.6448475122451782,
+      "learning_rate": 0.0002,
+      "loss": 1.2163,
+      "step": 9530
+    },
+    {
+      "epoch": 0.5795164621552666,
+      "grad_norm": 1.440741777420044,
+      "learning_rate": 0.0002,
+      "loss": 1.3541,
+      "step": 9540
+    },
+    {
+      "epoch": 0.580123921759203,
+      "grad_norm": 1.4616281986236572,
+      "learning_rate": 0.0002,
+      "loss": 1.2192,
+      "step": 9550
+    },
+    {
+      "epoch": 0.5807313813631394,
+      "grad_norm": 1.7298589944839478,
+      "learning_rate": 0.0002,
+      "loss": 1.2013,
+      "step": 9560
+    },
+    {
+      "epoch": 0.5813388409670757,
+      "grad_norm": 1.6787705421447754,
+      "learning_rate": 0.0002,
+      "loss": 1.1879,
+      "step": 9570
+    },
+    {
+      "epoch": 0.581946300571012,
+      "grad_norm": 1.4008698463439941,
+      "learning_rate": 0.0002,
+      "loss": 1.2064,
+      "step": 9580
+    },
+    {
+      "epoch": 0.5825537601749484,
+      "grad_norm": 1.3439879417419434,
+      "learning_rate": 0.0002,
+      "loss": 1.1908,
+      "step": 9590
+    },
+    {
+      "epoch": 0.5831612197788847,
+      "grad_norm": 1.436219573020935,
+      "learning_rate": 0.0002,
+      "loss": 1.2103,
+      "step": 9600
+    },
+    {
+      "epoch": 0.5837686793828211,
+      "grad_norm": 1.481085181236267,
+      "learning_rate": 0.0002,
+      "loss": 1.2041,
+      "step": 9610
+    },
+    {
+      "epoch": 0.5843761389867573,
+      "grad_norm": 1.3902438879013062,
+      "learning_rate": 0.0002,
+      "loss": 1.1871,
+      "step": 9620
+    },
+    {
+      "epoch": 0.5849835985906937,
+      "grad_norm": 1.6079788208007812,
+      "learning_rate": 0.0002,
+      "loss": 1.1283,
+      "step": 9630
+    },
+    {
+      "epoch": 0.5855910581946301,
+      "grad_norm": 1.473951816558838,
+      "learning_rate": 0.0002,
+      "loss": 1.2437,
+      "step": 9640
+    },
+    {
+      "epoch": 0.5861985177985664,
+      "grad_norm": 1.6076713800430298,
+      "learning_rate": 0.0002,
+      "loss": 1.2463,
+      "step": 9650
+    },
+    {
+      "epoch": 0.5868059774025027,
+      "grad_norm": 1.4721814393997192,
+      "learning_rate": 0.0002,
+      "loss": 1.2204,
+      "step": 9660
+    },
+    {
+      "epoch": 0.5874134370064391,
+      "grad_norm": 1.4392870664596558,
+      "learning_rate": 0.0002,
+      "loss": 1.1332,
+      "step": 9670
+    },
+    {
+      "epoch": 0.5880208966103754,
+      "grad_norm": 1.3198387622833252,
+      "learning_rate": 0.0002,
+      "loss": 1.273,
+      "step": 9680
+    },
+    {
+      "epoch": 0.5886283562143118,
+      "grad_norm": 1.32366144657135,
+      "learning_rate": 0.0002,
+      "loss": 1.273,
+      "step": 9690
+    },
+    {
+      "epoch": 0.5892358158182481,
+      "grad_norm": 1.3553622961044312,
+      "learning_rate": 0.0002,
+      "loss": 1.2246,
+      "step": 9700
+    },
+    {
+      "epoch": 0.5898432754221844,
+      "grad_norm": 1.2340115308761597,
+      "learning_rate": 0.0002,
+      "loss": 1.1045,
+      "step": 9710
+    },
+    {
+      "epoch": 0.5904507350261208,
+      "grad_norm": 1.4639991521835327,
+      "learning_rate": 0.0002,
+      "loss": 1.2169,
+      "step": 9720
+    },
+    {
+      "epoch": 0.5910581946300572,
+      "grad_norm": 1.7941631078720093,
+      "learning_rate": 0.0002,
+      "loss": 1.1877,
+      "step": 9730
+    },
+    {
+      "epoch": 0.5916656542339934,
+      "grad_norm": 1.494030475616455,
+      "learning_rate": 0.0002,
+      "loss": 1.1785,
+      "step": 9740
+    },
+    {
+      "epoch": 0.5922731138379298,
+      "grad_norm": 1.371238112449646,
+      "learning_rate": 0.0002,
+      "loss": 1.212,
+      "step": 9750
+    },
+    {
+      "epoch": 0.5928805734418661,
+      "grad_norm": 1.6964439153671265,
+      "learning_rate": 0.0002,
+      "loss": 1.2128,
+      "step": 9760
+    },
+    {
+      "epoch": 0.5934880330458024,
+      "grad_norm": 1.3934699296951294,
+      "learning_rate": 0.0002,
+      "loss": 1.183,
+      "step": 9770
+    },
+    {
+      "epoch": 0.5940954926497388,
+      "grad_norm": 1.505387306213379,
+      "learning_rate": 0.0002,
+      "loss": 1.1992,
+      "step": 9780
+    },
+    {
+      "epoch": 0.5947029522536751,
+      "grad_norm": 1.6573508977890015,
+      "learning_rate": 0.0002,
+      "loss": 1.2132,
+      "step": 9790
+    },
+    {
+      "epoch": 0.5953104118576115,
+      "grad_norm": 1.3351209163665771,
+      "learning_rate": 0.0002,
+      "loss": 1.2428,
+      "step": 9800
+    },
+    {
+      "epoch": 0.5959178714615478,
+      "grad_norm": 1.2429594993591309,
+      "learning_rate": 0.0002,
+      "loss": 1.1371,
+      "step": 9810
+    },
+    {
+      "epoch": 0.5965253310654841,
+      "grad_norm": 1.5691545009613037,
+      "learning_rate": 0.0002,
+      "loss": 1.1295,
+      "step": 9820
+    },
+    {
+      "epoch": 0.5971327906694205,
+      "grad_norm": 1.4945405721664429,
+      "learning_rate": 0.0002,
+      "loss": 1.2581,
+      "step": 9830
+    },
+    {
+      "epoch": 0.5977402502733569,
+      "grad_norm": 1.513573408126831,
+      "learning_rate": 0.0002,
+      "loss": 1.2453,
+      "step": 9840
+    },
+    {
+      "epoch": 0.5983477098772931,
+      "grad_norm": 1.426198959350586,
+      "learning_rate": 0.0002,
+      "loss": 1.0992,
+      "step": 9850
+    },
+    {
+      "epoch": 0.5989551694812295,
+      "grad_norm": 1.6695457696914673,
+      "learning_rate": 0.0002,
+      "loss": 1.2234,
+      "step": 9860
+    },
+    {
+      "epoch": 0.5995626290851659,
+      "grad_norm": 1.9191772937774658,
+      "learning_rate": 0.0002,
+      "loss": 1.1916,
+      "step": 9870
+    },
+    {
+      "epoch": 0.6001700886891022,
+      "grad_norm": 1.4344862699508667,
+      "learning_rate": 0.0002,
+      "loss": 1.1948,
+      "step": 9880
+    },
+    {
+      "epoch": 0.6007775482930385,
+      "grad_norm": 1.4640361070632935,
+      "learning_rate": 0.0002,
+      "loss": 1.2454,
+      "step": 9890
+    },
+    {
+      "epoch": 0.6013850078969748,
+      "grad_norm": 1.7511848211288452,
+      "learning_rate": 0.0002,
+      "loss": 1.2485,
+      "step": 9900
+    },
+    {
+      "epoch": 0.6019924675009112,
+      "grad_norm": 1.3894282579421997,
+      "learning_rate": 0.0002,
+      "loss": 1.171,
+      "step": 9910
+    },
+    {
+      "epoch": 0.6025999271048476,
+      "grad_norm": 1.3920185565948486,
+      "learning_rate": 0.0002,
+      "loss": 1.2435,
+      "step": 9920
+    },
+    {
+      "epoch": 0.6032073867087838,
+      "grad_norm": 1.3208262920379639,
+      "learning_rate": 0.0002,
+      "loss": 1.2918,
+      "step": 9930
+    },
+    {
+      "epoch": 0.6038148463127202,
+      "grad_norm": 1.1784992218017578,
+      "learning_rate": 0.0002,
+      "loss": 1.151,
+      "step": 9940
+    },
+    {
+      "epoch": 0.6044223059166566,
+      "grad_norm": 1.4552900791168213,
+      "learning_rate": 0.0002,
+      "loss": 1.2019,
+      "step": 9950
+    },
+    {
+      "epoch": 0.6050297655205928,
+      "grad_norm": 1.574312686920166,
+      "learning_rate": 0.0002,
+      "loss": 1.2686,
+      "step": 9960
+    },
+    {
+      "epoch": 0.6056372251245292,
+      "grad_norm": 1.6619045734405518,
+      "learning_rate": 0.0002,
+      "loss": 1.1878,
+      "step": 9970
+    },
+    {
+      "epoch": 0.6062446847284656,
+      "grad_norm": 1.2976871728897095,
+      "learning_rate": 0.0002,
+      "loss": 1.2308,
+      "step": 9980
+    },
+    {
+      "epoch": 0.6068521443324019,
+      "grad_norm": 2.1094729900360107,
+      "learning_rate": 0.0002,
+      "loss": 1.1648,
+      "step": 9990
+    },
+    {
+      "epoch": 0.6074596039363382,
+      "grad_norm": 1.9893029928207397,
+      "learning_rate": 0.0002,
+      "loss": 1.2477,
+      "step": 10000
+    },
+    {
+      "epoch": 0.6080670635402746,
+      "grad_norm": 1.579158902168274,
+      "learning_rate": 0.0002,
+      "loss": 1.1757,
+      "step": 10010
+    },
+    {
+      "epoch": 0.6086745231442109,
+      "grad_norm": 1.3016012907028198,
+      "learning_rate": 0.0002,
+      "loss": 1.2097,
+      "step": 10020
+    },
+    {
+      "epoch": 0.6092819827481473,
+      "grad_norm": 1.4439659118652344,
+      "learning_rate": 0.0002,
+      "loss": 1.1698,
+      "step": 10030
+    },
+    {
+      "epoch": 0.6098894423520835,
+      "grad_norm": 1.490735411643982,
+      "learning_rate": 0.0002,
+      "loss": 1.2366,
+      "step": 10040
+    },
+    {
+      "epoch": 0.6104969019560199,
+      "grad_norm": 1.5587278604507446,
+      "learning_rate": 0.0002,
+      "loss": 1.1882,
+      "step": 10050
+    },
+    {
+      "epoch": 0.6111043615599563,
+      "grad_norm": 1.5142701864242554,
+      "learning_rate": 0.0002,
+      "loss": 1.1452,
+      "step": 10060
+    },
+    {
+      "epoch": 0.6117118211638926,
+      "grad_norm": 1.562000036239624,
+      "learning_rate": 0.0002,
+      "loss": 1.2219,
+      "step": 10070
+    },
+    {
+      "epoch": 0.6123192807678289,
+      "grad_norm": 1.5308712720870972,
+      "learning_rate": 0.0002,
+      "loss": 1.1221,
+      "step": 10080
+    },
+    {
+      "epoch": 0.6129267403717653,
+      "grad_norm": 1.441798448562622,
+      "learning_rate": 0.0002,
+      "loss": 1.1266,
+      "step": 10090
+    },
+    {
+      "epoch": 0.6135341999757016,
+      "grad_norm": 1.3092381954193115,
+      "learning_rate": 0.0002,
+      "loss": 1.1661,
+      "step": 10100
+    },
+    {
+      "epoch": 0.614141659579638,
+      "grad_norm": 1.592490315437317,
+      "learning_rate": 0.0002,
+      "loss": 1.1633,
+      "step": 10110
+    },
+    {
+      "epoch": 0.6147491191835743,
+      "grad_norm": 1.5515542030334473,
+      "learning_rate": 0.0002,
+      "loss": 1.2181,
+      "step": 10120
+    },
+    {
+      "epoch": 0.6153565787875106,
+      "grad_norm": 1.8464525938034058,
+      "learning_rate": 0.0002,
+      "loss": 1.194,
+      "step": 10130
+    },
+    {
+      "epoch": 0.615964038391447,
+      "grad_norm": 1.4995911121368408,
+      "learning_rate": 0.0002,
+      "loss": 1.1926,
+      "step": 10140
+    },
+    {
+      "epoch": 0.6165714979953834,
+      "grad_norm": 2.0670576095581055,
+      "learning_rate": 0.0002,
+      "loss": 1.2425,
+      "step": 10150
+    },
+    {
+      "epoch": 0.6171789575993196,
+      "grad_norm": 1.2869359254837036,
+      "learning_rate": 0.0002,
+      "loss": 1.2304,
+      "step": 10160
+    },
+    {
+      "epoch": 0.617786417203256,
+      "grad_norm": 1.6126221418380737,
+      "learning_rate": 0.0002,
+      "loss": 1.1576,
+      "step": 10170
+    },
+    {
+      "epoch": 0.6183938768071923,
+      "grad_norm": 1.5873949527740479,
+      "learning_rate": 0.0002,
+      "loss": 1.2378,
+      "step": 10180
+    },
+    {
+      "epoch": 0.6190013364111286,
+      "grad_norm": 1.394704818725586,
+      "learning_rate": 0.0002,
+      "loss": 1.1733,
+      "step": 10190
+    },
+    {
+      "epoch": 0.619608796015065,
+      "grad_norm": 1.4479098320007324,
+      "learning_rate": 0.0002,
+      "loss": 1.1198,
+      "step": 10200
+    },
+    {
+      "epoch": 0.6202162556190013,
+      "grad_norm": 1.4679584503173828,
+      "learning_rate": 0.0002,
+      "loss": 1.1245,
+      "step": 10210
+    },
+    {
+      "epoch": 0.6208237152229377,
+      "grad_norm": 1.4689127206802368,
+      "learning_rate": 0.0002,
+      "loss": 1.1647,
+      "step": 10220
+    },
+    {
+      "epoch": 0.621431174826874,
+      "grad_norm": 1.3493456840515137,
+      "learning_rate": 0.0002,
+      "loss": 1.149,
+      "step": 10230
+    },
+    {
+      "epoch": 0.6220386344308103,
+      "grad_norm": 1.480111837387085,
+      "learning_rate": 0.0002,
+      "loss": 1.0947,
+      "step": 10240
+    },
+    {
+      "epoch": 0.6226460940347467,
+      "grad_norm": 1.4981671571731567,
+      "learning_rate": 0.0002,
+      "loss": 1.2457,
+      "step": 10250
+    },
+    {
+      "epoch": 0.6232535536386831,
+      "grad_norm": 1.4562405347824097,
+      "learning_rate": 0.0002,
+      "loss": 1.161,
+      "step": 10260
+    },
+    {
+      "epoch": 0.6238610132426193,
+      "grad_norm": 1.4748475551605225,
+      "learning_rate": 0.0002,
+      "loss": 1.1582,
+      "step": 10270
+    },
+    {
+      "epoch": 0.6244684728465557,
+      "grad_norm": 1.4762338399887085,
+      "learning_rate": 0.0002,
+      "loss": 1.1543,
+      "step": 10280
+    },
+    {
+      "epoch": 0.6250759324504921,
+      "grad_norm": 1.5699721574783325,
+      "learning_rate": 0.0002,
+      "loss": 1.1238,
+      "step": 10290
+    },
+    {
+      "epoch": 0.6256833920544284,
+      "grad_norm": 1.5905998945236206,
+      "learning_rate": 0.0002,
+      "loss": 1.2323,
+      "step": 10300
+    },
+    {
+      "epoch": 0.6262908516583647,
+      "grad_norm": 1.2949004173278809,
+      "learning_rate": 0.0002,
+      "loss": 1.1863,
+      "step": 10310
+    },
+    {
+      "epoch": 0.6268983112623011,
+      "grad_norm": 1.8533037900924683,
+      "learning_rate": 0.0002,
+      "loss": 1.1373,
+      "step": 10320
+    },
+    {
+      "epoch": 0.6275057708662374,
+      "grad_norm": 1.3952187299728394,
+      "learning_rate": 0.0002,
+      "loss": 1.2169,
+      "step": 10330
+    },
+    {
+      "epoch": 0.6281132304701738,
+      "grad_norm": 1.4759795665740967,
+      "learning_rate": 0.0002,
+      "loss": 1.1877,
+      "step": 10340
+    },
+    {
+      "epoch": 0.62872069007411,
+      "grad_norm": 1.8833881616592407,
+      "learning_rate": 0.0002,
+      "loss": 1.2601,
+      "step": 10350
+    },
+    {
+      "epoch": 0.6293281496780464,
+      "grad_norm": 1.4174413681030273,
+      "learning_rate": 0.0002,
+      "loss": 1.232,
+      "step": 10360
+    },
+    {
+      "epoch": 0.6299356092819828,
+      "grad_norm": 1.5129276514053345,
+      "learning_rate": 0.0002,
+      "loss": 1.1903,
+      "step": 10370
+    },
+    {
+      "epoch": 0.630543068885919,
+      "grad_norm": 1.4858694076538086,
+      "learning_rate": 0.0002,
+      "loss": 1.1323,
+      "step": 10380
+    },
+    {
+      "epoch": 0.6311505284898554,
+      "grad_norm": 1.458954930305481,
+      "learning_rate": 0.0002,
+      "loss": 1.1408,
+      "step": 10390
+    },
+    {
+      "epoch": 0.6317579880937918,
+      "grad_norm": 1.442545771598816,
+      "learning_rate": 0.0002,
+      "loss": 1.201,
+      "step": 10400
+    },
+    {
+      "epoch": 0.6323654476977281,
+      "grad_norm": 1.5205395221710205,
+      "learning_rate": 0.0002,
+      "loss": 1.1685,
+      "step": 10410
+    },
+    {
+      "epoch": 0.6329729073016644,
+      "grad_norm": 1.3119162321090698,
+      "learning_rate": 0.0002,
+      "loss": 1.1423,
+      "step": 10420
+    },
+    {
+      "epoch": 0.6335803669056008,
+      "grad_norm": 1.345941185951233,
+      "learning_rate": 0.0002,
+      "loss": 1.192,
+      "step": 10430
+    },
+    {
+      "epoch": 0.6341878265095371,
+      "grad_norm": 1.6821863651275635,
+      "learning_rate": 0.0002,
+      "loss": 1.1688,
+      "step": 10440
+    },
+    {
+      "epoch": 0.6347952861134735,
+      "grad_norm": 1.6074057817459106,
+      "learning_rate": 0.0002,
+      "loss": 1.2167,
+      "step": 10450
+    },
+    {
+      "epoch": 0.6354027457174098,
+      "grad_norm": 1.5278266668319702,
+      "learning_rate": 0.0002,
+      "loss": 1.1838,
+      "step": 10460
+    },
+    {
+      "epoch": 0.6360102053213461,
+      "grad_norm": 1.3733577728271484,
+      "learning_rate": 0.0002,
+      "loss": 1.1441,
+      "step": 10470
+    },
+    {
+      "epoch": 0.6366176649252825,
+      "grad_norm": 1.5218098163604736,
+      "learning_rate": 0.0002,
+      "loss": 1.1277,
+      "step": 10480
+    },
+    {
+      "epoch": 0.6372251245292188,
+      "grad_norm": 1.6241083145141602,
+      "learning_rate": 0.0002,
+      "loss": 1.2118,
+      "step": 10490
+    },
+    {
+      "epoch": 0.6378325841331551,
+      "grad_norm": 1.836913824081421,
+      "learning_rate": 0.0002,
+      "loss": 1.1521,
+      "step": 10500
+    },
+    {
+      "epoch": 0.6384400437370915,
+      "grad_norm": 1.609478235244751,
+      "learning_rate": 0.0002,
+      "loss": 1.1986,
+      "step": 10510
+    },
+    {
+      "epoch": 0.6390475033410278,
+      "grad_norm": 1.6075109243392944,
+      "learning_rate": 0.0002,
+      "loss": 1.2768,
+      "step": 10520
+    },
+    {
+      "epoch": 0.6396549629449642,
+      "grad_norm": 1.3840945959091187,
+      "learning_rate": 0.0002,
+      "loss": 1.1719,
+      "step": 10530
+    },
+    {
+      "epoch": 0.6402624225489005,
+      "grad_norm": 1.3323301076889038,
+      "learning_rate": 0.0002,
+      "loss": 1.1737,
+      "step": 10540
+    },
+    {
+      "epoch": 0.6408698821528368,
+      "grad_norm": 1.513311505317688,
+      "learning_rate": 0.0002,
+      "loss": 1.1444,
+      "step": 10550
+    },
+    {
+      "epoch": 0.6414773417567732,
+      "grad_norm": 1.5211206674575806,
+      "learning_rate": 0.0002,
+      "loss": 1.2536,
+      "step": 10560
+    },
+    {
+      "epoch": 0.6420848013607096,
+      "grad_norm": 1.331162452697754,
+      "learning_rate": 0.0002,
+      "loss": 1.2607,
+      "step": 10570
+    },
+    {
+      "epoch": 0.6426922609646458,
+      "grad_norm": 1.284327745437622,
+      "learning_rate": 0.0002,
+      "loss": 1.2564,
+      "step": 10580
+    },
+    {
+      "epoch": 0.6432997205685822,
+      "grad_norm": 1.2587096691131592,
+      "learning_rate": 0.0002,
+      "loss": 1.1948,
+      "step": 10590
+    },
+    {
+      "epoch": 0.6439071801725186,
+      "grad_norm": 1.4804191589355469,
+      "learning_rate": 0.0002,
+      "loss": 1.2251,
+      "step": 10600
+    },
+    {
+      "epoch": 0.6445146397764548,
+      "grad_norm": 1.3306211233139038,
+      "learning_rate": 0.0002,
+      "loss": 1.1484,
+      "step": 10610
+    },
+    {
+      "epoch": 0.6451220993803912,
+      "grad_norm": 1.4872162342071533,
+      "learning_rate": 0.0002,
+      "loss": 1.2206,
+      "step": 10620
+    },
+    {
+      "epoch": 0.6457295589843275,
+      "grad_norm": 1.479332685470581,
+      "learning_rate": 0.0002,
+      "loss": 1.1723,
+      "step": 10630
+    },
+    {
+      "epoch": 0.6463370185882639,
+      "grad_norm": 1.6562544107437134,
+      "learning_rate": 0.0002,
+      "loss": 1.1718,
+      "step": 10640
+    },
+    {
+      "epoch": 0.6469444781922002,
+      "grad_norm": 1.4814534187316895,
+      "learning_rate": 0.0002,
+      "loss": 1.1975,
+      "step": 10650
+    },
+    {
+      "epoch": 0.6475519377961365,
+      "grad_norm": 1.4615775346755981,
+      "learning_rate": 0.0002,
+      "loss": 1.1592,
+      "step": 10660
+    },
+    {
+      "epoch": 0.6481593974000729,
+      "grad_norm": 1.4853074550628662,
+      "learning_rate": 0.0002,
+      "loss": 1.193,
+      "step": 10670
+    },
+    {
+      "epoch": 0.6487668570040093,
+      "grad_norm": 1.4148366451263428,
+      "learning_rate": 0.0002,
+      "loss": 1.1757,
+      "step": 10680
+    },
+    {
+      "epoch": 0.6493743166079455,
+      "grad_norm": 1.644765853881836,
+      "learning_rate": 0.0002,
+      "loss": 1.1298,
+      "step": 10690
+    },
+    {
+      "epoch": 0.6499817762118819,
+      "grad_norm": 1.6349629163742065,
+      "learning_rate": 0.0002,
+      "loss": 1.157,
+      "step": 10700
+    },
+    {
+      "epoch": 0.6505892358158183,
+      "grad_norm": 1.4590332508087158,
+      "learning_rate": 0.0002,
+      "loss": 1.2426,
+      "step": 10710
+    },
+    {
+      "epoch": 0.6511966954197546,
+      "grad_norm": 1.3630939722061157,
+      "learning_rate": 0.0002,
+      "loss": 1.1686,
+      "step": 10720
+    },
+    {
+      "epoch": 0.6518041550236909,
+      "grad_norm": 1.5206563472747803,
+      "learning_rate": 0.0002,
+      "loss": 1.1991,
+      "step": 10730
+    },
+    {
+      "epoch": 0.6524116146276273,
+      "grad_norm": 1.1928949356079102,
+      "learning_rate": 0.0002,
+      "loss": 1.1915,
+      "step": 10740
+    },
+    {
+      "epoch": 0.6530190742315636,
+      "grad_norm": 1.7482240200042725,
+      "learning_rate": 0.0002,
+      "loss": 1.2403,
+      "step": 10750
+    },
+    {
+      "epoch": 0.6536265338355,
+      "grad_norm": 1.4580903053283691,
+      "learning_rate": 0.0002,
+      "loss": 1.2571,
+      "step": 10760
+    },
+    {
+      "epoch": 0.6542339934394362,
+      "grad_norm": 1.582047939300537,
+      "learning_rate": 0.0002,
+      "loss": 1.2438,
+      "step": 10770
+    },
+    {
+      "epoch": 0.6548414530433726,
+      "grad_norm": 1.415243148803711,
+      "learning_rate": 0.0002,
+      "loss": 1.1761,
+      "step": 10780
+    },
+    {
+      "epoch": 0.655448912647309,
+      "grad_norm": 1.3117393255233765,
+      "learning_rate": 0.0002,
+      "loss": 1.1339,
+      "step": 10790
+    },
+    {
+      "epoch": 0.6560563722512452,
+      "grad_norm": 1.4229295253753662,
+      "learning_rate": 0.0002,
+      "loss": 1.1884,
+      "step": 10800
+    },
+    {
+      "epoch": 0.6566638318551816,
+      "grad_norm": 1.449526309967041,
+      "learning_rate": 0.0002,
+      "loss": 1.287,
+      "step": 10810
+    },
+    {
+      "epoch": 0.657271291459118,
+      "grad_norm": 1.509210228919983,
+      "learning_rate": 0.0002,
+      "loss": 1.1541,
+      "step": 10820
+    },
+    {
+      "epoch": 0.6578787510630543,
+      "grad_norm": 1.6160600185394287,
+      "learning_rate": 0.0002,
+      "loss": 1.2228,
+      "step": 10830
+    },
+    {
+      "epoch": 0.6584862106669906,
+      "grad_norm": 1.3628963232040405,
+      "learning_rate": 0.0002,
+      "loss": 1.268,
+      "step": 10840
+    },
+    {
+      "epoch": 0.659093670270927,
+      "grad_norm": 1.3497792482376099,
+      "learning_rate": 0.0002,
+      "loss": 1.1868,
+      "step": 10850
+    },
+    {
+      "epoch": 0.6597011298748633,
+      "grad_norm": 1.3454173803329468,
+      "learning_rate": 0.0002,
+      "loss": 1.1741,
+      "step": 10860
+    },
+    {
+      "epoch": 0.6603085894787997,
+      "grad_norm": 1.4797899723052979,
+      "learning_rate": 0.0002,
+      "loss": 1.1899,
+      "step": 10870
+    },
+    {
+      "epoch": 0.660916049082736,
+      "grad_norm": 1.2999018430709839,
+      "learning_rate": 0.0002,
+      "loss": 1.2017,
+      "step": 10880
+    },
+    {
+      "epoch": 0.6615235086866723,
+      "grad_norm": 1.4134020805358887,
+      "learning_rate": 0.0002,
+      "loss": 1.2336,
+      "step": 10890
+    },
+    {
+      "epoch": 0.6621309682906087,
+      "grad_norm": 1.706198811531067,
+      "learning_rate": 0.0002,
+      "loss": 1.2633,
+      "step": 10900
+    },
+    {
+      "epoch": 0.6627384278945451,
+      "grad_norm": 1.3994085788726807,
+      "learning_rate": 0.0002,
+      "loss": 1.199,
+      "step": 10910
+    },
+    {
+      "epoch": 0.6633458874984813,
+      "grad_norm": 1.4403148889541626,
+      "learning_rate": 0.0002,
+      "loss": 1.236,
+      "step": 10920
+    },
+    {
+      "epoch": 0.6639533471024177,
+      "grad_norm": 1.5367127656936646,
+      "learning_rate": 0.0002,
+      "loss": 1.228,
+      "step": 10930
+    },
+    {
+      "epoch": 0.664560806706354,
+      "grad_norm": 1.7260019779205322,
+      "learning_rate": 0.0002,
+      "loss": 1.2759,
+      "step": 10940
+    },
+    {
+      "epoch": 0.6651682663102904,
+      "grad_norm": 1.456173300743103,
+      "learning_rate": 0.0002,
+      "loss": 1.138,
+      "step": 10950
+    },
+    {
+      "epoch": 0.6657757259142267,
+      "grad_norm": 1.9463717937469482,
+      "learning_rate": 0.0002,
+      "loss": 1.094,
+      "step": 10960
+    },
+    {
+      "epoch": 0.666383185518163,
+      "grad_norm": 1.5047192573547363,
+      "learning_rate": 0.0002,
+      "loss": 1.2151,
+      "step": 10970
+    },
+    {
+      "epoch": 0.6669906451220994,
+      "grad_norm": 1.4958069324493408,
+      "learning_rate": 0.0002,
+      "loss": 1.1373,
+      "step": 10980
+    },
+    {
+      "epoch": 0.6675981047260358,
+      "grad_norm": 1.6270114183425903,
+      "learning_rate": 0.0002,
+      "loss": 1.1656,
+      "step": 10990
+    },
+    {
+      "epoch": 0.668205564329972,
+      "grad_norm": 1.494084119796753,
+      "learning_rate": 0.0002,
+      "loss": 1.1473,
+      "step": 11000
+    },
+    {
+      "epoch": 0.6688130239339084,
+      "grad_norm": 1.6952012777328491,
+      "learning_rate": 0.0002,
+      "loss": 1.1886,
+      "step": 11010
+    },
+    {
+      "epoch": 0.6694204835378448,
+      "grad_norm": 1.2693300247192383,
+      "learning_rate": 0.0002,
+      "loss": 1.1854,
+      "step": 11020
+    },
+    {
+      "epoch": 0.670027943141781,
+      "grad_norm": 1.3508623838424683,
+      "learning_rate": 0.0002,
+      "loss": 1.1476,
+      "step": 11030
+    },
+    {
+      "epoch": 0.6706354027457174,
+      "grad_norm": 1.5516679286956787,
+      "learning_rate": 0.0002,
+      "loss": 1.2994,
+      "step": 11040
+    },
+    {
+      "epoch": 0.6712428623496538,
+      "grad_norm": 1.5078186988830566,
+      "learning_rate": 0.0002,
+      "loss": 1.1406,
+      "step": 11050
+    },
+    {
+      "epoch": 0.6718503219535901,
+      "grad_norm": 1.5944573879241943,
+      "learning_rate": 0.0002,
+      "loss": 1.1062,
+      "step": 11060
+    },
+    {
+      "epoch": 0.6724577815575264,
+      "grad_norm": 1.866337537765503,
+      "learning_rate": 0.0002,
+      "loss": 1.2106,
+      "step": 11070
+    },
+    {
+      "epoch": 0.6730652411614627,
+      "grad_norm": 1.346386432647705,
+      "learning_rate": 0.0002,
+      "loss": 1.1681,
+      "step": 11080
+    },
+    {
+      "epoch": 0.6736727007653991,
+      "grad_norm": 1.4467651844024658,
+      "learning_rate": 0.0002,
+      "loss": 1.2039,
+      "step": 11090
+    },
+    {
+      "epoch": 0.6742801603693355,
+      "grad_norm": 1.404458999633789,
+      "learning_rate": 0.0002,
+      "loss": 1.2296,
+      "step": 11100
+    },
+    {
+      "epoch": 0.6748876199732717,
+      "grad_norm": 1.2593837976455688,
+      "learning_rate": 0.0002,
+      "loss": 1.2039,
+      "step": 11110
+    },
+    {
+      "epoch": 0.6754950795772081,
+      "grad_norm": 1.4862419366836548,
+      "learning_rate": 0.0002,
+      "loss": 1.1957,
+      "step": 11120
+    },
+    {
+      "epoch": 0.6761025391811445,
+      "grad_norm": 1.4536348581314087,
+      "learning_rate": 0.0002,
+      "loss": 1.2568,
+      "step": 11130
+    },
+    {
+      "epoch": 0.6767099987850808,
+      "grad_norm": 1.6232870817184448,
+      "learning_rate": 0.0002,
+      "loss": 1.184,
+      "step": 11140
+    },
+    {
+      "epoch": 0.6773174583890171,
+      "grad_norm": 1.621951937675476,
+      "learning_rate": 0.0002,
+      "loss": 1.1848,
+      "step": 11150
+    },
+    {
+      "epoch": 0.6779249179929535,
+      "grad_norm": 1.462931513786316,
+      "learning_rate": 0.0002,
+      "loss": 1.1735,
+      "step": 11160
+    },
+    {
+      "epoch": 0.6785323775968898,
+      "grad_norm": 1.4450300931930542,
+      "learning_rate": 0.0002,
+      "loss": 1.186,
+      "step": 11170
+    },
+    {
+      "epoch": 0.6791398372008262,
+      "grad_norm": 1.7563248872756958,
+      "learning_rate": 0.0002,
+      "loss": 1.1988,
+      "step": 11180
+    },
+    {
+      "epoch": 0.6797472968047625,
+      "grad_norm": 1.3302048444747925,
+      "learning_rate": 0.0002,
+      "loss": 1.1443,
+      "step": 11190
+    },
+    {
+      "epoch": 0.6803547564086988,
+      "grad_norm": 1.4724364280700684,
+      "learning_rate": 0.0002,
+      "loss": 1.1359,
+      "step": 11200
+    },
+    {
+      "epoch": 0.6809622160126352,
+      "grad_norm": 1.4847537279129028,
+      "learning_rate": 0.0002,
+      "loss": 1.1733,
+      "step": 11210
+    },
+    {
+      "epoch": 0.6815696756165714,
+      "grad_norm": 1.5312765836715698,
+      "learning_rate": 0.0002,
+      "loss": 1.1182,
+      "step": 11220
+    },
+    {
+      "epoch": 0.6821771352205078,
+      "grad_norm": 1.563795804977417,
+      "learning_rate": 0.0002,
+      "loss": 1.1132,
+      "step": 11230
+    },
+    {
+      "epoch": 0.6827845948244442,
+      "grad_norm": 1.5239176750183105,
+      "learning_rate": 0.0002,
+      "loss": 1.2627,
+      "step": 11240
+    },
+    {
+      "epoch": 0.6833920544283805,
+      "grad_norm": 1.7546555995941162,
+      "learning_rate": 0.0002,
+      "loss": 1.3069,
+      "step": 11250
+    },
+    {
+      "epoch": 0.6839995140323168,
+      "grad_norm": 1.4419306516647339,
+      "learning_rate": 0.0002,
+      "loss": 1.2376,
+      "step": 11260
+    },
+    {
+      "epoch": 0.6846069736362532,
+      "grad_norm": 1.425281286239624,
+      "learning_rate": 0.0002,
+      "loss": 1.1672,
+      "step": 11270
+    },
+    {
+      "epoch": 0.6852144332401895,
+      "grad_norm": 1.3561053276062012,
+      "learning_rate": 0.0002,
+      "loss": 1.2377,
+      "step": 11280
+    },
+    {
+      "epoch": 0.6858218928441259,
+      "grad_norm": 1.4803534746170044,
+      "learning_rate": 0.0002,
+      "loss": 1.192,
+      "step": 11290
+    },
+    {
+      "epoch": 0.6864293524480622,
+      "grad_norm": 1.431349277496338,
+      "learning_rate": 0.0002,
+      "loss": 1.2415,
+      "step": 11300
+    },
+    {
+      "epoch": 0.6870368120519985,
+      "grad_norm": 1.2821253538131714,
+      "learning_rate": 0.0002,
+      "loss": 1.1721,
+      "step": 11310
+    },
+    {
+      "epoch": 0.6876442716559349,
+      "grad_norm": 1.4956986904144287,
+      "learning_rate": 0.0002,
+      "loss": 1.2056,
+      "step": 11320
+    },
+    {
+      "epoch": 0.6882517312598713,
+      "grad_norm": 1.3334455490112305,
+      "learning_rate": 0.0002,
+      "loss": 1.1747,
+      "step": 11330
+    },
+    {
+      "epoch": 0.6888591908638075,
+      "grad_norm": 1.6449331045150757,
+      "learning_rate": 0.0002,
+      "loss": 1.2525,
+      "step": 11340
+    },
+    {
+      "epoch": 0.6894666504677439,
+      "grad_norm": 1.402047872543335,
+      "learning_rate": 0.0002,
+      "loss": 1.2468,
+      "step": 11350
+    },
+    {
+      "epoch": 0.6900741100716802,
+      "grad_norm": 1.4714034795761108,
+      "learning_rate": 0.0002,
+      "loss": 1.1789,
+      "step": 11360
+    },
+    {
+      "epoch": 0.6906815696756166,
+      "grad_norm": 1.4426461458206177,
+      "learning_rate": 0.0002,
+      "loss": 1.2036,
+      "step": 11370
+    },
+    {
+      "epoch": 0.6912890292795529,
+      "grad_norm": 1.3404841423034668,
+      "learning_rate": 0.0002,
+      "loss": 1.2156,
+      "step": 11380
+    },
+    {
+      "epoch": 0.6918964888834892,
+      "grad_norm": 1.3607988357543945,
+      "learning_rate": 0.0002,
+      "loss": 1.2025,
+      "step": 11390
+    },
+    {
+      "epoch": 0.6925039484874256,
+      "grad_norm": 1.4359990358352661,
+      "learning_rate": 0.0002,
+      "loss": 1.2282,
+      "step": 11400
+    },
+    {
+      "epoch": 0.693111408091362,
+      "grad_norm": 1.619635820388794,
+      "learning_rate": 0.0002,
+      "loss": 1.1713,
+      "step": 11410
+    },
+    {
+      "epoch": 0.6937188676952982,
+      "grad_norm": 1.6724573373794556,
+      "learning_rate": 0.0002,
+      "loss": 1.2081,
+      "step": 11420
+    },
+    {
+      "epoch": 0.6943263272992346,
+      "grad_norm": 2.061098098754883,
+      "learning_rate": 0.0002,
+      "loss": 1.185,
+      "step": 11430
+    },
+    {
+      "epoch": 0.694933786903171,
+      "grad_norm": 1.4103002548217773,
+      "learning_rate": 0.0002,
+      "loss": 1.1879,
+      "step": 11440
+    },
+    {
+      "epoch": 0.6955412465071072,
+      "grad_norm": 1.2703981399536133,
+      "learning_rate": 0.0002,
+      "loss": 1.2088,
+      "step": 11450
+    },
+    {
+      "epoch": 0.6961487061110436,
+      "grad_norm": 1.6062376499176025,
+      "learning_rate": 0.0002,
+      "loss": 1.2214,
+      "step": 11460
+    },
+    {
+      "epoch": 0.69675616571498,
+      "grad_norm": 1.5932917594909668,
+      "learning_rate": 0.0002,
+      "loss": 1.1671,
+      "step": 11470
+    },
+    {
+      "epoch": 0.6973636253189163,
+      "grad_norm": 1.4609780311584473,
+      "learning_rate": 0.0002,
+      "loss": 1.1285,
+      "step": 11480
+    },
+    {
+      "epoch": 0.6979710849228526,
+      "grad_norm": 1.3255772590637207,
+      "learning_rate": 0.0002,
+      "loss": 1.0967,
+      "step": 11490
+    },
+    {
+      "epoch": 0.6985785445267889,
+      "grad_norm": 1.6930729150772095,
+      "learning_rate": 0.0002,
+      "loss": 1.1953,
+      "step": 11500
+    },
+    {
+      "epoch": 0.6991860041307253,
+      "grad_norm": 1.3310422897338867,
+      "learning_rate": 0.0002,
+      "loss": 1.1798,
+      "step": 11510
+    },
+    {
+      "epoch": 0.6997934637346617,
+      "grad_norm": 1.3762205839157104,
+      "learning_rate": 0.0002,
+      "loss": 1.1868,
+      "step": 11520
+    },
+    {
+      "epoch": 0.7004009233385979,
+      "grad_norm": 1.344161868095398,
+      "learning_rate": 0.0002,
+      "loss": 1.2483,
+      "step": 11530
+    },
+    {
+      "epoch": 0.7010083829425343,
+      "grad_norm": 1.4122248888015747,
+      "learning_rate": 0.0002,
+      "loss": 1.2095,
+      "step": 11540
+    },
+    {
+      "epoch": 0.7016158425464707,
+      "grad_norm": 1.4848181009292603,
+      "learning_rate": 0.0002,
+      "loss": 1.2064,
+      "step": 11550
+    },
+    {
+      "epoch": 0.702223302150407,
+      "grad_norm": 1.3725167512893677,
+      "learning_rate": 0.0002,
+      "loss": 1.2049,
+      "step": 11560
+    },
+    {
+      "epoch": 0.7028307617543433,
+      "grad_norm": 1.5955760478973389,
+      "learning_rate": 0.0002,
+      "loss": 1.1277,
+      "step": 11570
+    },
+    {
+      "epoch": 0.7034382213582797,
+      "grad_norm": 1.461313009262085,
+      "learning_rate": 0.0002,
+      "loss": 1.2,
+      "step": 11580
+    },
+    {
+      "epoch": 0.704045680962216,
+      "grad_norm": 1.5912322998046875,
+      "learning_rate": 0.0002,
+      "loss": 1.2243,
+      "step": 11590
+    },
+    {
+      "epoch": 0.7046531405661524,
+      "grad_norm": 1.641867756843567,
+      "learning_rate": 0.0002,
+      "loss": 1.2052,
+      "step": 11600
+    },
+    {
+      "epoch": 0.7052606001700887,
+      "grad_norm": 1.3238072395324707,
+      "learning_rate": 0.0002,
+      "loss": 1.1505,
+      "step": 11610
+    },
+    {
+      "epoch": 0.705868059774025,
+      "grad_norm": 1.4066652059555054,
+      "learning_rate": 0.0002,
+      "loss": 1.2051,
+      "step": 11620
+    },
+    {
+      "epoch": 0.7064755193779614,
+      "grad_norm": 1.5860298871994019,
+      "learning_rate": 0.0002,
+      "loss": 1.1738,
+      "step": 11630
+    },
+    {
+      "epoch": 0.7070829789818978,
+      "grad_norm": 1.5154073238372803,
+      "learning_rate": 0.0002,
+      "loss": 1.2652,
+      "step": 11640
+    },
+    {
+      "epoch": 0.707690438585834,
+      "grad_norm": 1.6234484910964966,
+      "learning_rate": 0.0002,
+      "loss": 1.2645,
+      "step": 11650
+    },
+    {
+      "epoch": 0.7082978981897704,
+      "grad_norm": 1.5033893585205078,
+      "learning_rate": 0.0002,
+      "loss": 1.1698,
+      "step": 11660
+    },
+    {
+      "epoch": 0.7089053577937067,
+      "grad_norm": 1.4364176988601685,
+      "learning_rate": 0.0002,
+      "loss": 1.2689,
+      "step": 11670
+    },
+    {
+      "epoch": 0.709512817397643,
+      "grad_norm": 1.389700174331665,
+      "learning_rate": 0.0002,
+      "loss": 1.2512,
+      "step": 11680
+    },
+    {
+      "epoch": 0.7101202770015794,
+      "grad_norm": 1.3205703496932983,
+      "learning_rate": 0.0002,
+      "loss": 1.1309,
+      "step": 11690
+    },
+    {
+      "epoch": 0.7107277366055157,
+      "grad_norm": 1.568920612335205,
+      "learning_rate": 0.0002,
+      "loss": 1.2776,
+      "step": 11700
+    },
+    {
+      "epoch": 0.7113351962094521,
+      "grad_norm": 1.4814732074737549,
+      "learning_rate": 0.0002,
+      "loss": 1.2819,
+      "step": 11710
+    },
+    {
+      "epoch": 0.7119426558133884,
+      "grad_norm": 1.6823959350585938,
+      "learning_rate": 0.0002,
+      "loss": 1.2476,
+      "step": 11720
+    },
+    {
+      "epoch": 0.7125501154173247,
+      "grad_norm": 1.338059902191162,
+      "learning_rate": 0.0002,
+      "loss": 1.2376,
+      "step": 11730
+    },
+    {
+      "epoch": 0.7131575750212611,
+      "grad_norm": 1.734095573425293,
+      "learning_rate": 0.0002,
+      "loss": 1.2301,
+      "step": 11740
+    },
+    {
+      "epoch": 0.7137650346251975,
+      "grad_norm": 1.3866161108016968,
+      "learning_rate": 0.0002,
+      "loss": 1.1602,
+      "step": 11750
+    },
+    {
+      "epoch": 0.7143724942291337,
+      "grad_norm": 1.465135931968689,
+      "learning_rate": 0.0002,
+      "loss": 1.2681,
+      "step": 11760
+    },
+    {
+      "epoch": 0.7149799538330701,
+      "grad_norm": 1.3397235870361328,
+      "learning_rate": 0.0002,
+      "loss": 1.2742,
+      "step": 11770
+    },
+    {
+      "epoch": 0.7155874134370065,
+      "grad_norm": 1.38888680934906,
+      "learning_rate": 0.0002,
+      "loss": 1.1629,
+      "step": 11780
+    },
+    {
+      "epoch": 0.7161948730409428,
+      "grad_norm": 1.4504741430282593,
+      "learning_rate": 0.0002,
+      "loss": 1.1503,
+      "step": 11790
+    },
+    {
+      "epoch": 0.7168023326448791,
+      "grad_norm": 1.4954638481140137,
+      "learning_rate": 0.0002,
+      "loss": 1.2249,
+      "step": 11800
+    },
+    {
+      "epoch": 0.7174097922488154,
+      "grad_norm": 1.841427206993103,
+      "learning_rate": 0.0002,
+      "loss": 1.2107,
+      "step": 11810
+    },
+    {
+      "epoch": 0.7180172518527518,
+      "grad_norm": 1.4142829179763794,
+      "learning_rate": 0.0002,
+      "loss": 1.1921,
+      "step": 11820
+    },
+    {
+      "epoch": 0.7186247114566882,
+      "grad_norm": 1.5487010478973389,
+      "learning_rate": 0.0002,
+      "loss": 1.1975,
+      "step": 11830
+    },
+    {
+      "epoch": 0.7192321710606244,
+      "grad_norm": 1.4076977968215942,
+      "learning_rate": 0.0002,
+      "loss": 1.2191,
+      "step": 11840
+    },
+    {
+      "epoch": 0.7198396306645608,
+      "grad_norm": 1.1866190433502197,
+      "learning_rate": 0.0002,
+      "loss": 1.2053,
+      "step": 11850
+    },
+    {
+      "epoch": 0.7204470902684972,
+      "grad_norm": 1.5468428134918213,
+      "learning_rate": 0.0002,
+      "loss": 1.1661,
+      "step": 11860
+    },
+    {
+      "epoch": 0.7210545498724334,
+      "grad_norm": 1.2567400932312012,
+      "learning_rate": 0.0002,
+      "loss": 1.179,
+      "step": 11870
+    },
+    {
+      "epoch": 0.7216620094763698,
+      "grad_norm": 1.5383141040802002,
+      "learning_rate": 0.0002,
+      "loss": 1.2302,
+      "step": 11880
+    },
+    {
+      "epoch": 0.7222694690803062,
+      "grad_norm": 1.494950532913208,
+      "learning_rate": 0.0002,
+      "loss": 1.2232,
+      "step": 11890
+    },
+    {
+      "epoch": 0.7228769286842425,
+      "grad_norm": 1.8329541683197021,
+      "learning_rate": 0.0002,
+      "loss": 1.1461,
+      "step": 11900
+    },
+    {
+      "epoch": 0.7234843882881788,
+      "grad_norm": 1.46611750125885,
+      "learning_rate": 0.0002,
+      "loss": 1.2242,
+      "step": 11910
+    },
+    {
+      "epoch": 0.7240918478921152,
+      "grad_norm": 1.4088685512542725,
+      "learning_rate": 0.0002,
+      "loss": 1.165,
+      "step": 11920
+    },
+    {
+      "epoch": 0.7246993074960515,
+      "grad_norm": 1.4150999784469604,
+      "learning_rate": 0.0002,
+      "loss": 1.1637,
+      "step": 11930
+    },
+    {
+      "epoch": 0.7253067670999879,
+      "grad_norm": 1.3681480884552002,
+      "learning_rate": 0.0002,
+      "loss": 1.1833,
+      "step": 11940
+    },
+    {
+      "epoch": 0.7259142267039241,
+      "grad_norm": 1.4092365503311157,
+      "learning_rate": 0.0002,
+      "loss": 1.1468,
+      "step": 11950
+    },
+    {
+      "epoch": 0.7265216863078605,
+      "grad_norm": 1.4637953042984009,
+      "learning_rate": 0.0002,
+      "loss": 1.2164,
+      "step": 11960
+    },
+    {
+      "epoch": 0.7271291459117969,
+      "grad_norm": 1.340900182723999,
+      "learning_rate": 0.0002,
+      "loss": 1.1661,
+      "step": 11970
+    },
+    {
+      "epoch": 0.7277366055157332,
+      "grad_norm": 1.4122298955917358,
+      "learning_rate": 0.0002,
+      "loss": 1.2512,
+      "step": 11980
+    },
+    {
+      "epoch": 0.7283440651196695,
+      "grad_norm": 1.5455623865127563,
+      "learning_rate": 0.0002,
+      "loss": 1.2112,
+      "step": 11990
+    },
+    {
+      "epoch": 0.7289515247236059,
+      "grad_norm": 1.3420119285583496,
+      "learning_rate": 0.0002,
+      "loss": 1.1923,
+      "step": 12000
+    },
+    {
+      "epoch": 0.7295589843275422,
+      "grad_norm": 1.5570626258850098,
+      "learning_rate": 0.0002,
+      "loss": 1.1912,
+      "step": 12010
+    },
+    {
+      "epoch": 0.7301664439314786,
+      "grad_norm": 1.5055865049362183,
+      "learning_rate": 0.0002,
+      "loss": 1.185,
+      "step": 12020
+    },
+    {
+      "epoch": 0.7307739035354149,
+      "grad_norm": 1.5202549695968628,
+      "learning_rate": 0.0002,
+      "loss": 1.2013,
+      "step": 12030
+    },
+    {
+      "epoch": 0.7313813631393512,
+      "grad_norm": 1.3424779176712036,
+      "learning_rate": 0.0002,
+      "loss": 1.1318,
+      "step": 12040
+    },
+    {
+      "epoch": 0.7319888227432876,
+      "grad_norm": 1.838219165802002,
+      "learning_rate": 0.0002,
+      "loss": 1.2258,
+      "step": 12050
+    },
+    {
+      "epoch": 0.732596282347224,
+      "grad_norm": 1.2670247554779053,
+      "learning_rate": 0.0002,
+      "loss": 1.1377,
+      "step": 12060
+    },
+    {
+      "epoch": 0.7332037419511602,
+      "grad_norm": 1.7354373931884766,
+      "learning_rate": 0.0002,
+      "loss": 1.1729,
+      "step": 12070
+    },
+    {
+      "epoch": 0.7338112015550966,
+      "grad_norm": 1.6558679342269897,
+      "learning_rate": 0.0002,
+      "loss": 1.2186,
+      "step": 12080
+    },
+    {
+      "epoch": 0.7344186611590329,
+      "grad_norm": 1.5855995416641235,
+      "learning_rate": 0.0002,
+      "loss": 1.2495,
+      "step": 12090
+    },
+    {
+      "epoch": 0.7350261207629692,
+      "grad_norm": 1.6140644550323486,
+      "learning_rate": 0.0002,
+      "loss": 1.2288,
+      "step": 12100
+    },
+    {
+      "epoch": 0.7356335803669056,
+      "grad_norm": 1.4002978801727295,
+      "learning_rate": 0.0002,
+      "loss": 1.2297,
+      "step": 12110
+    },
+    {
+      "epoch": 0.7362410399708419,
+      "grad_norm": 1.445778489112854,
+      "learning_rate": 0.0002,
+      "loss": 1.1928,
+      "step": 12120
+    },
+    {
+      "epoch": 0.7368484995747783,
+      "grad_norm": 1.6145892143249512,
+      "learning_rate": 0.0002,
+      "loss": 1.1717,
+      "step": 12130
+    },
+    {
+      "epoch": 0.7374559591787146,
+      "grad_norm": 1.5301674604415894,
+      "learning_rate": 0.0002,
+      "loss": 1.122,
+      "step": 12140
+    },
+    {
+      "epoch": 0.7380634187826509,
+      "grad_norm": 1.5016682147979736,
+      "learning_rate": 0.0002,
+      "loss": 1.2023,
+      "step": 12150
+    },
+    {
+      "epoch": 0.7386708783865873,
+      "grad_norm": 1.5861783027648926,
+      "learning_rate": 0.0002,
+      "loss": 1.1168,
+      "step": 12160
+    },
+    {
+      "epoch": 0.7392783379905237,
+      "grad_norm": 1.6081180572509766,
+      "learning_rate": 0.0002,
+      "loss": 1.1411,
+      "step": 12170
+    },
+    {
+      "epoch": 0.7398857975944599,
+      "grad_norm": 1.5475598573684692,
+      "learning_rate": 0.0002,
+      "loss": 1.151,
+      "step": 12180
+    },
+    {
+      "epoch": 0.7404932571983963,
+      "grad_norm": 1.668095588684082,
+      "learning_rate": 0.0002,
+      "loss": 1.1793,
+      "step": 12190
+    },
+    {
+      "epoch": 0.7411007168023327,
+      "grad_norm": 1.5576446056365967,
+      "learning_rate": 0.0002,
+      "loss": 1.2414,
+      "step": 12200
+    },
+    {
+      "epoch": 0.741708176406269,
+      "grad_norm": 1.7492656707763672,
+      "learning_rate": 0.0002,
+      "loss": 1.1558,
+      "step": 12210
+    },
+    {
+      "epoch": 0.7423156360102053,
+      "grad_norm": 1.527300238609314,
+      "learning_rate": 0.0002,
+      "loss": 1.168,
+      "step": 12220
+    },
+    {
+      "epoch": 0.7429230956141416,
+      "grad_norm": 1.4713521003723145,
+      "learning_rate": 0.0002,
+      "loss": 1.1747,
+      "step": 12230
+    },
+    {
+      "epoch": 0.743530555218078,
+      "grad_norm": 1.3588942289352417,
+      "learning_rate": 0.0002,
+      "loss": 1.1781,
+      "step": 12240
+    },
+    {
+      "epoch": 0.7441380148220144,
+      "grad_norm": 1.6379607915878296,
+      "learning_rate": 0.0002,
+      "loss": 1.1654,
+      "step": 12250
+    },
+    {
+      "epoch": 0.7447454744259506,
+      "grad_norm": 1.337945580482483,
+      "learning_rate": 0.0002,
+      "loss": 1.1837,
+      "step": 12260
+    },
+    {
+      "epoch": 0.745352934029887,
+      "grad_norm": 1.6633474826812744,
+      "learning_rate": 0.0002,
+      "loss": 1.21,
+      "step": 12270
+    },
+    {
+      "epoch": 0.7459603936338234,
+      "grad_norm": 1.4022483825683594,
+      "learning_rate": 0.0002,
+      "loss": 1.1702,
+      "step": 12280
+    },
+    {
+      "epoch": 0.7465678532377596,
+      "grad_norm": 1.5646429061889648,
+      "learning_rate": 0.0002,
+      "loss": 1.1522,
+      "step": 12290
+    },
+    {
+      "epoch": 0.747175312841696,
+      "grad_norm": 1.559167742729187,
+      "learning_rate": 0.0002,
+      "loss": 1.1502,
+      "step": 12300
+    },
+    {
+      "epoch": 0.7477827724456324,
+      "grad_norm": 1.3548595905303955,
+      "learning_rate": 0.0002,
+      "loss": 1.1954,
+      "step": 12310
+    },
+    {
+      "epoch": 0.7483902320495687,
+      "grad_norm": 1.4975273609161377,
+      "learning_rate": 0.0002,
+      "loss": 1.216,
+      "step": 12320
+    },
+    {
+      "epoch": 0.748997691653505,
+      "grad_norm": 1.408597469329834,
+      "learning_rate": 0.0002,
+      "loss": 1.2319,
+      "step": 12330
+    },
+    {
+      "epoch": 0.7496051512574414,
+      "grad_norm": 1.4837566614151,
+      "learning_rate": 0.0002,
+      "loss": 1.2854,
+      "step": 12340
+    },
+    {
+      "epoch": 0.7502126108613777,
+      "grad_norm": 1.2895525693893433,
+      "learning_rate": 0.0002,
+      "loss": 1.2207,
+      "step": 12350
+    },
+    {
+      "epoch": 0.7508200704653141,
+      "grad_norm": 1.3992550373077393,
+      "learning_rate": 0.0002,
+      "loss": 1.1627,
+      "step": 12360
+    },
+    {
+      "epoch": 0.7514275300692504,
+      "grad_norm": 1.6421400308609009,
+      "learning_rate": 0.0002,
+      "loss": 1.1692,
+      "step": 12370
+    },
+    {
+      "epoch": 0.7520349896731867,
+      "grad_norm": 1.799697756767273,
+      "learning_rate": 0.0002,
+      "loss": 1.2416,
+      "step": 12380
+    },
+    {
+      "epoch": 0.7526424492771231,
+      "grad_norm": 1.6245464086532593,
+      "learning_rate": 0.0002,
+      "loss": 1.2382,
+      "step": 12390
+    },
+    {
+      "epoch": 0.7532499088810594,
+      "grad_norm": 1.462110161781311,
+      "learning_rate": 0.0002,
+      "loss": 1.1325,
+      "step": 12400
+    },
+    {
+      "epoch": 0.7538573684849957,
+      "grad_norm": 1.7330759763717651,
+      "learning_rate": 0.0002,
+      "loss": 1.1275,
+      "step": 12410
+    },
+    {
+      "epoch": 0.7544648280889321,
+      "grad_norm": 1.3573112487792969,
+      "learning_rate": 0.0002,
+      "loss": 1.1673,
+      "step": 12420
+    },
+    {
+      "epoch": 0.7550722876928684,
+      "grad_norm": 1.528638482093811,
+      "learning_rate": 0.0002,
+      "loss": 1.1091,
+      "step": 12430
+    },
+    {
+      "epoch": 0.7556797472968048,
+      "grad_norm": 1.4841316938400269,
+      "learning_rate": 0.0002,
+      "loss": 1.1973,
+      "step": 12440
+    },
+    {
+      "epoch": 0.7562872069007411,
+      "grad_norm": 1.8295537233352661,
+      "learning_rate": 0.0002,
+      "loss": 1.0993,
+      "step": 12450
+    },
+    {
+      "epoch": 0.7568946665046774,
+      "grad_norm": 1.580533504486084,
+      "learning_rate": 0.0002,
+      "loss": 1.2444,
+      "step": 12460
+    },
+    {
+      "epoch": 0.7575021261086138,
+      "grad_norm": 1.4425346851348877,
+      "learning_rate": 0.0002,
+      "loss": 1.1931,
+      "step": 12470
+    },
+    {
+      "epoch": 0.7581095857125502,
+      "grad_norm": 1.4134783744812012,
+      "learning_rate": 0.0002,
+      "loss": 1.144,
+      "step": 12480
+    },
+    {
+      "epoch": 0.7587170453164864,
+      "grad_norm": 1.2354546785354614,
+      "learning_rate": 0.0002,
+      "loss": 1.1764,
+      "step": 12490
+    },
+    {
+      "epoch": 0.7593245049204228,
+      "grad_norm": 1.2936530113220215,
+      "learning_rate": 0.0002,
+      "loss": 1.1902,
+      "step": 12500
+    },
+    {
+      "epoch": 0.7599319645243592,
+      "grad_norm": 1.5252578258514404,
+      "learning_rate": 0.0002,
+      "loss": 1.17,
+      "step": 12510
+    },
+    {
+      "epoch": 0.7605394241282954,
+      "grad_norm": 1.538248062133789,
+      "learning_rate": 0.0002,
+      "loss": 1.2061,
+      "step": 12520
+    },
+    {
+      "epoch": 0.7611468837322318,
+      "grad_norm": 1.61935555934906,
+      "learning_rate": 0.0002,
+      "loss": 1.1884,
+      "step": 12530
+    },
+    {
+      "epoch": 0.7617543433361681,
+      "grad_norm": 1.254185676574707,
+      "learning_rate": 0.0002,
+      "loss": 1.2032,
+      "step": 12540
+    },
+    {
+      "epoch": 0.7623618029401045,
+      "grad_norm": 1.367621898651123,
+      "learning_rate": 0.0002,
+      "loss": 1.2207,
+      "step": 12550
+    },
+    {
+      "epoch": 0.7629692625440408,
+      "grad_norm": 1.512383222579956,
+      "learning_rate": 0.0002,
+      "loss": 1.1971,
+      "step": 12560
+    },
+    {
+      "epoch": 0.7635767221479771,
+      "grad_norm": 1.4338618516921997,
+      "learning_rate": 0.0002,
+      "loss": 1.2558,
+      "step": 12570
+    },
+    {
+      "epoch": 0.7641841817519135,
+      "grad_norm": 1.322144865989685,
+      "learning_rate": 0.0002,
+      "loss": 1.0924,
+      "step": 12580
+    },
+    {
+      "epoch": 0.7647916413558499,
+      "grad_norm": 1.518500804901123,
+      "learning_rate": 0.0002,
+      "loss": 1.1616,
+      "step": 12590
+    },
+    {
+      "epoch": 0.7653991009597861,
+      "grad_norm": 1.4812904596328735,
+      "learning_rate": 0.0002,
+      "loss": 1.1519,
+      "step": 12600
+    },
+    {
+      "epoch": 0.7660065605637225,
+      "grad_norm": 1.480425477027893,
+      "learning_rate": 0.0002,
+      "loss": 1.2715,
+      "step": 12610
+    },
+    {
+      "epoch": 0.7666140201676589,
+      "grad_norm": 1.554589867591858,
+      "learning_rate": 0.0002,
+      "loss": 1.2196,
+      "step": 12620
+    },
+    {
+      "epoch": 0.7672214797715952,
+      "grad_norm": 1.6555403470993042,
+      "learning_rate": 0.0002,
+      "loss": 1.1961,
+      "step": 12630
+    },
+    {
+      "epoch": 0.7678289393755315,
+      "grad_norm": 1.9677565097808838,
+      "learning_rate": 0.0002,
+      "loss": 1.1848,
+      "step": 12640
+    },
+    {
+      "epoch": 0.7684363989794679,
+      "grad_norm": 1.4238587617874146,
+      "learning_rate": 0.0002,
+      "loss": 1.1731,
+      "step": 12650
+    },
+    {
+      "epoch": 0.7690438585834042,
+      "grad_norm": 1.2964900732040405,
+      "learning_rate": 0.0002,
+      "loss": 1.1768,
+      "step": 12660
+    },
+    {
+      "epoch": 0.7696513181873406,
+      "grad_norm": 1.549517273902893,
+      "learning_rate": 0.0002,
+      "loss": 1.1728,
+      "step": 12670
+    },
+    {
+      "epoch": 0.7702587777912768,
+      "grad_norm": 1.5808876752853394,
+      "learning_rate": 0.0002,
+      "loss": 1.1796,
+      "step": 12680
+    },
+    {
+      "epoch": 0.7708662373952132,
+      "grad_norm": 1.7364044189453125,
+      "learning_rate": 0.0002,
+      "loss": 1.2,
+      "step": 12690
+    },
+    {
+      "epoch": 0.7714736969991496,
+      "grad_norm": 1.4321144819259644,
+      "learning_rate": 0.0002,
+      "loss": 1.2613,
+      "step": 12700
+    },
+    {
+      "epoch": 0.7720811566030858,
+      "grad_norm": 1.4324842691421509,
+      "learning_rate": 0.0002,
+      "loss": 1.1764,
+      "step": 12710
+    },
+    {
+      "epoch": 0.7726886162070222,
+      "grad_norm": 1.518303632736206,
+      "learning_rate": 0.0002,
+      "loss": 1.2267,
+      "step": 12720
+    },
+    {
+      "epoch": 0.7732960758109586,
+      "grad_norm": 1.494956135749817,
+      "learning_rate": 0.0002,
+      "loss": 1.18,
+      "step": 12730
+    },
+    {
+      "epoch": 0.7739035354148949,
+      "grad_norm": 1.5330840349197388,
+      "learning_rate": 0.0002,
+      "loss": 1.2204,
+      "step": 12740
+    },
+    {
+      "epoch": 0.7745109950188312,
+      "grad_norm": 1.964697003364563,
+      "learning_rate": 0.0002,
+      "loss": 1.1896,
+      "step": 12750
+    },
+    {
+      "epoch": 0.7751184546227676,
+      "grad_norm": 1.4546568393707275,
+      "learning_rate": 0.0002,
+      "loss": 1.1851,
+      "step": 12760
+    },
+    {
+      "epoch": 0.7757259142267039,
+      "grad_norm": 1.5761953592300415,
+      "learning_rate": 0.0002,
+      "loss": 1.1721,
+      "step": 12770
+    },
+    {
+      "epoch": 0.7763333738306403,
+      "grad_norm": 1.676142692565918,
+      "learning_rate": 0.0002,
+      "loss": 1.1536,
+      "step": 12780
+    },
+    {
+      "epoch": 0.7769408334345766,
+      "grad_norm": 1.4308006763458252,
+      "learning_rate": 0.0002,
+      "loss": 1.2011,
+      "step": 12790
+    },
+    {
+      "epoch": 0.7775482930385129,
+      "grad_norm": 1.3389242887496948,
+      "learning_rate": 0.0002,
+      "loss": 1.2487,
+      "step": 12800
+    },
+    {
+      "epoch": 0.7781557526424493,
+      "grad_norm": 1.6429331302642822,
+      "learning_rate": 0.0002,
+      "loss": 1.1282,
+      "step": 12810
+    },
+    {
+      "epoch": 0.7787632122463856,
+      "grad_norm": 1.4515995979309082,
+      "learning_rate": 0.0002,
+      "loss": 1.2761,
+      "step": 12820
+    },
+    {
+      "epoch": 0.7793706718503219,
+      "grad_norm": 1.4249382019042969,
+      "learning_rate": 0.0002,
+      "loss": 1.1918,
+      "step": 12830
+    },
+    {
+      "epoch": 0.7799781314542583,
+      "grad_norm": 1.55287766456604,
+      "learning_rate": 0.0002,
+      "loss": 1.1739,
+      "step": 12840
+    },
+    {
+      "epoch": 0.7805855910581946,
+      "grad_norm": 1.5151697397232056,
+      "learning_rate": 0.0002,
+      "loss": 1.1971,
+      "step": 12850
+    },
+    {
+      "epoch": 0.781193050662131,
+      "grad_norm": 1.4955744743347168,
+      "learning_rate": 0.0002,
+      "loss": 1.1891,
+      "step": 12860
+    },
+    {
+      "epoch": 0.7818005102660673,
+      "grad_norm": 1.3934895992279053,
+      "learning_rate": 0.0002,
+      "loss": 1.2318,
+      "step": 12870
+    },
+    {
+      "epoch": 0.7824079698700036,
+      "grad_norm": 1.6154847145080566,
+      "learning_rate": 0.0002,
+      "loss": 1.175,
+      "step": 12880
+    },
+    {
+      "epoch": 0.78301542947394,
+      "grad_norm": 1.4619389772415161,
+      "learning_rate": 0.0002,
+      "loss": 1.2368,
+      "step": 12890
+    },
+    {
+      "epoch": 0.7836228890778764,
+      "grad_norm": 1.4026260375976562,
+      "learning_rate": 0.0002,
+      "loss": 1.1869,
+      "step": 12900
+    },
+    {
+      "epoch": 0.7842303486818126,
+      "grad_norm": 1.633529782295227,
+      "learning_rate": 0.0002,
+      "loss": 1.1961,
+      "step": 12910
+    },
+    {
+      "epoch": 0.784837808285749,
+      "grad_norm": 1.270814061164856,
+      "learning_rate": 0.0002,
+      "loss": 1.2143,
+      "step": 12920
+    },
+    {
+      "epoch": 0.7854452678896854,
+      "grad_norm": 1.2957777976989746,
+      "learning_rate": 0.0002,
+      "loss": 1.1464,
+      "step": 12930
+    },
+    {
+      "epoch": 0.7860527274936217,
+      "grad_norm": 1.5368854999542236,
+      "learning_rate": 0.0002,
+      "loss": 1.1597,
+      "step": 12940
+    },
+    {
+      "epoch": 0.786660187097558,
+      "grad_norm": 1.471379280090332,
+      "learning_rate": 0.0002,
+      "loss": 1.2498,
+      "step": 12950
+    },
+    {
+      "epoch": 0.7872676467014944,
+      "grad_norm": 1.4442158937454224,
+      "learning_rate": 0.0002,
+      "loss": 1.1931,
+      "step": 12960
+    },
+    {
+      "epoch": 0.7878751063054307,
+      "grad_norm": 1.6200677156448364,
+      "learning_rate": 0.0002,
+      "loss": 1.1395,
+      "step": 12970
+    },
+    {
+      "epoch": 0.788482565909367,
+      "grad_norm": 1.4099328517913818,
+      "learning_rate": 0.0002,
+      "loss": 1.3028,
+      "step": 12980
+    },
+    {
+      "epoch": 0.7890900255133033,
+      "grad_norm": 1.5445653200149536,
+      "learning_rate": 0.0002,
+      "loss": 1.2275,
+      "step": 12990
+    },
+    {
+      "epoch": 0.7896974851172397,
+      "grad_norm": 1.5414783954620361,
+      "learning_rate": 0.0002,
+      "loss": 1.1612,
+      "step": 13000
+    },
+    {
+      "epoch": 0.7903049447211761,
+      "grad_norm": 1.3546172380447388,
+      "learning_rate": 0.0002,
+      "loss": 1.1443,
+      "step": 13010
+    },
+    {
+      "epoch": 0.7909124043251123,
+      "grad_norm": 1.457933783531189,
+      "learning_rate": 0.0002,
+      "loss": 1.1875,
+      "step": 13020
+    },
+    {
+      "epoch": 0.7915198639290487,
+      "grad_norm": 1.7691285610198975,
+      "learning_rate": 0.0002,
+      "loss": 1.2228,
+      "step": 13030
+    },
+    {
+      "epoch": 0.7921273235329851,
+      "grad_norm": 1.4855856895446777,
+      "learning_rate": 0.0002,
+      "loss": 1.2502,
+      "step": 13040
+    },
+    {
+      "epoch": 0.7927347831369214,
+      "grad_norm": 1.5997259616851807,
+      "learning_rate": 0.0002,
+      "loss": 1.1122,
+      "step": 13050
+    },
+    {
+      "epoch": 0.7933422427408577,
+      "grad_norm": 1.385907769203186,
+      "learning_rate": 0.0002,
+      "loss": 1.2139,
+      "step": 13060
+    },
+    {
+      "epoch": 0.7939497023447941,
+      "grad_norm": 1.3917052745819092,
+      "learning_rate": 0.0002,
+      "loss": 1.1403,
+      "step": 13070
+    },
+    {
+      "epoch": 0.7945571619487304,
+      "grad_norm": 1.5885778665542603,
+      "learning_rate": 0.0002,
+      "loss": 1.1718,
+      "step": 13080
+    },
+    {
+      "epoch": 0.7951646215526668,
+      "grad_norm": 1.5090978145599365,
+      "learning_rate": 0.0002,
+      "loss": 1.1756,
+      "step": 13090
+    },
+    {
+      "epoch": 0.7957720811566031,
+      "grad_norm": 1.5283892154693604,
+      "learning_rate": 0.0002,
+      "loss": 1.218,
+      "step": 13100
+    },
+    {
+      "epoch": 0.7963795407605394,
+      "grad_norm": 1.446351170539856,
+      "learning_rate": 0.0002,
+      "loss": 1.1796,
+      "step": 13110
+    },
+    {
+      "epoch": 0.7969870003644758,
+      "grad_norm": 1.3841274976730347,
+      "learning_rate": 0.0002,
+      "loss": 1.2957,
+      "step": 13120
+    },
+    {
+      "epoch": 0.797594459968412,
+      "grad_norm": 1.398147463798523,
+      "learning_rate": 0.0002,
+      "loss": 1.1361,
+      "step": 13130
+    },
+    {
+      "epoch": 0.7982019195723484,
+      "grad_norm": 1.4801783561706543,
+      "learning_rate": 0.0002,
+      "loss": 1.1428,
+      "step": 13140
+    },
+    {
+      "epoch": 0.7988093791762848,
+      "grad_norm": 1.4249933958053589,
+      "learning_rate": 0.0002,
+      "loss": 1.1931,
+      "step": 13150
+    },
+    {
+      "epoch": 0.7994168387802211,
+      "grad_norm": 1.6051521301269531,
+      "learning_rate": 0.0002,
+      "loss": 1.1448,
+      "step": 13160
+    },
+    {
+      "epoch": 0.8000242983841575,
+      "grad_norm": 1.6075396537780762,
+      "learning_rate": 0.0002,
+      "loss": 1.2308,
+      "step": 13170
+    },
+    {
+      "epoch": 0.8006317579880938,
+      "grad_norm": 1.581541895866394,
+      "learning_rate": 0.0002,
+      "loss": 1.2696,
+      "step": 13180
+    },
+    {
+      "epoch": 0.8012392175920301,
+      "grad_norm": 1.4118540287017822,
+      "learning_rate": 0.0002,
+      "loss": 1.2502,
+      "step": 13190
+    },
+    {
+      "epoch": 0.8018466771959665,
+      "grad_norm": 1.604483962059021,
+      "learning_rate": 0.0002,
+      "loss": 1.2867,
+      "step": 13200
+    },
+    {
+      "epoch": 0.8024541367999029,
+      "grad_norm": 1.623291015625,
+      "learning_rate": 0.0002,
+      "loss": 1.221,
+      "step": 13210
+    },
+    {
+      "epoch": 0.8030615964038391,
+      "grad_norm": 1.4024327993392944,
+      "learning_rate": 0.0002,
+      "loss": 1.2307,
+      "step": 13220
+    },
+    {
+      "epoch": 0.8036690560077755,
+      "grad_norm": 1.4459460973739624,
+      "learning_rate": 0.0002,
+      "loss": 1.2514,
+      "step": 13230
+    },
+    {
+      "epoch": 0.8042765156117119,
+      "grad_norm": 1.6799654960632324,
+      "learning_rate": 0.0002,
+      "loss": 1.163,
+      "step": 13240
+    },
+    {
+      "epoch": 0.8048839752156481,
+      "grad_norm": 1.5066721439361572,
+      "learning_rate": 0.0002,
+      "loss": 1.1511,
+      "step": 13250
+    },
+    {
+      "epoch": 0.8054914348195845,
+      "grad_norm": 1.372342824935913,
+      "learning_rate": 0.0002,
+      "loss": 1.1287,
+      "step": 13260
+    },
+    {
+      "epoch": 0.8060988944235208,
+      "grad_norm": 1.3962156772613525,
+      "learning_rate": 0.0002,
+      "loss": 1.2028,
+      "step": 13270
+    },
+    {
+      "epoch": 0.8067063540274572,
+      "grad_norm": 1.9565892219543457,
+      "learning_rate": 0.0002,
+      "loss": 1.1776,
+      "step": 13280
+    },
+    {
+      "epoch": 0.8073138136313935,
+      "grad_norm": 1.2959867715835571,
+      "learning_rate": 0.0002,
+      "loss": 1.251,
+      "step": 13290
+    },
+    {
+      "epoch": 0.8079212732353298,
+      "grad_norm": 1.5922397375106812,
+      "learning_rate": 0.0002,
+      "loss": 1.1784,
+      "step": 13300
+    },
+    {
+      "epoch": 0.8085287328392662,
+      "grad_norm": 1.3177214860916138,
+      "learning_rate": 0.0002,
+      "loss": 1.1632,
+      "step": 13310
+    },
+    {
+      "epoch": 0.8091361924432026,
+      "grad_norm": 1.3495534658432007,
+      "learning_rate": 0.0002,
+      "loss": 1.2253,
+      "step": 13320
+    },
+    {
+      "epoch": 0.8097436520471388,
+      "grad_norm": 1.482490062713623,
+      "learning_rate": 0.0002,
+      "loss": 1.2171,
+      "step": 13330
+    },
+    {
+      "epoch": 0.8103511116510752,
+      "grad_norm": 1.4726133346557617,
+      "learning_rate": 0.0002,
+      "loss": 1.1493,
+      "step": 13340
+    },
+    {
+      "epoch": 0.8109585712550116,
+      "grad_norm": 1.4320366382598877,
+      "learning_rate": 0.0002,
+      "loss": 1.2101,
+      "step": 13350
+    },
+    {
+      "epoch": 0.8115660308589479,
+      "grad_norm": 1.3902326822280884,
+      "learning_rate": 0.0002,
+      "loss": 1.2068,
+      "step": 13360
+    },
+    {
+      "epoch": 0.8121734904628842,
+      "grad_norm": 1.3399302959442139,
+      "learning_rate": 0.0002,
+      "loss": 1.0984,
+      "step": 13370
+    },
+    {
+      "epoch": 0.8127809500668206,
+      "grad_norm": 1.4275768995285034,
+      "learning_rate": 0.0002,
+      "loss": 1.1792,
+      "step": 13380
+    },
+    {
+      "epoch": 0.8133884096707569,
+      "grad_norm": 1.271705985069275,
+      "learning_rate": 0.0002,
+      "loss": 1.0776,
+      "step": 13390
+    },
+    {
+      "epoch": 0.8139958692746933,
+      "grad_norm": 1.4107437133789062,
+      "learning_rate": 0.0002,
+      "loss": 1.1963,
+      "step": 13400
+    },
+    {
+      "epoch": 0.8146033288786295,
+      "grad_norm": 1.2505677938461304,
+      "learning_rate": 0.0002,
+      "loss": 1.1869,
+      "step": 13410
+    },
+    {
+      "epoch": 0.8152107884825659,
+      "grad_norm": 1.4574933052062988,
+      "learning_rate": 0.0002,
+      "loss": 1.1719,
+      "step": 13420
+    },
+    {
+      "epoch": 0.8158182480865023,
+      "grad_norm": 1.4642574787139893,
+      "learning_rate": 0.0002,
+      "loss": 1.1641,
+      "step": 13430
+    },
+    {
+      "epoch": 0.8164257076904385,
+      "grad_norm": 1.4101893901824951,
+      "learning_rate": 0.0002,
+      "loss": 1.2693,
+      "step": 13440
+    },
+    {
+      "epoch": 0.8170331672943749,
+      "grad_norm": 1.8907462358474731,
+      "learning_rate": 0.0002,
+      "loss": 1.1996,
+      "step": 13450
+    },
+    {
+      "epoch": 0.8176406268983113,
+      "grad_norm": 1.5065134763717651,
+      "learning_rate": 0.0002,
+      "loss": 1.1823,
+      "step": 13460
+    },
+    {
+      "epoch": 0.8182480865022476,
+      "grad_norm": 1.4328252077102661,
+      "learning_rate": 0.0002,
+      "loss": 1.1492,
+      "step": 13470
+    },
+    {
+      "epoch": 0.8188555461061839,
+      "grad_norm": 1.3925843238830566,
+      "learning_rate": 0.0002,
+      "loss": 1.1919,
+      "step": 13480
+    },
+    {
+      "epoch": 0.8194630057101203,
+      "grad_norm": 1.5043162107467651,
+      "learning_rate": 0.0002,
+      "loss": 1.194,
+      "step": 13490
+    },
+    {
+      "epoch": 0.8200704653140566,
+      "grad_norm": 1.2992193698883057,
+      "learning_rate": 0.0002,
+      "loss": 1.194,
+      "step": 13500
+    },
+    {
+      "epoch": 0.820677924917993,
+      "grad_norm": 1.3942667245864868,
+      "learning_rate": 0.0002,
+      "loss": 1.1834,
+      "step": 13510
+    },
+    {
+      "epoch": 0.8212853845219293,
+      "grad_norm": 1.2020211219787598,
+      "learning_rate": 0.0002,
+      "loss": 1.1504,
+      "step": 13520
+    },
+    {
+      "epoch": 0.8218928441258656,
+      "grad_norm": 1.2666652202606201,
+      "learning_rate": 0.0002,
+      "loss": 1.1739,
+      "step": 13530
+    },
+    {
+      "epoch": 0.822500303729802,
+      "grad_norm": 1.597417950630188,
+      "learning_rate": 0.0002,
+      "loss": 1.2378,
+      "step": 13540
+    },
+    {
+      "epoch": 0.8231077633337383,
+      "grad_norm": 1.4688104391098022,
+      "learning_rate": 0.0002,
+      "loss": 1.1803,
+      "step": 13550
+    },
+    {
+      "epoch": 0.8237152229376746,
+      "grad_norm": 1.684959053993225,
+      "learning_rate": 0.0002,
+      "loss": 1.1498,
+      "step": 13560
+    },
+    {
+      "epoch": 0.824322682541611,
+      "grad_norm": 1.3847455978393555,
+      "learning_rate": 0.0002,
+      "loss": 1.1951,
+      "step": 13570
+    },
+    {
+      "epoch": 0.8249301421455473,
+      "grad_norm": 1.4115383625030518,
+      "learning_rate": 0.0002,
+      "loss": 1.1232,
+      "step": 13580
+    },
+    {
+      "epoch": 0.8255376017494837,
+      "grad_norm": 1.601866602897644,
+      "learning_rate": 0.0002,
+      "loss": 1.1711,
+      "step": 13590
+    },
+    {
+      "epoch": 0.82614506135342,
+      "grad_norm": 1.3733577728271484,
+      "learning_rate": 0.0002,
+      "loss": 1.2374,
+      "step": 13600
+    },
+    {
+      "epoch": 0.8267525209573563,
+      "grad_norm": 1.3552721738815308,
+      "learning_rate": 0.0002,
+      "loss": 1.0967,
+      "step": 13610
+    },
+    {
+      "epoch": 0.8273599805612927,
+      "grad_norm": 1.4282503128051758,
+      "learning_rate": 0.0002,
+      "loss": 1.243,
+      "step": 13620
+    },
+    {
+      "epoch": 0.827967440165229,
+      "grad_norm": 1.5548524856567383,
+      "learning_rate": 0.0002,
+      "loss": 1.1529,
+      "step": 13630
+    },
+    {
+      "epoch": 0.8285748997691653,
+      "grad_norm": 1.4593840837478638,
+      "learning_rate": 0.0002,
+      "loss": 1.223,
+      "step": 13640
+    },
+    {
+      "epoch": 0.8291823593731017,
+      "grad_norm": 1.7803534269332886,
+      "learning_rate": 0.0002,
+      "loss": 1.1564,
+      "step": 13650
+    },
+    {
+      "epoch": 0.8297898189770381,
+      "grad_norm": 1.487880825996399,
+      "learning_rate": 0.0002,
+      "loss": 1.2527,
+      "step": 13660
+    },
+    {
+      "epoch": 0.8303972785809743,
+      "grad_norm": 1.5033938884735107,
+      "learning_rate": 0.0002,
+      "loss": 1.1874,
+      "step": 13670
+    },
+    {
+      "epoch": 0.8310047381849107,
+      "grad_norm": 1.6714093685150146,
+      "learning_rate": 0.0002,
+      "loss": 1.2239,
+      "step": 13680
+    },
+    {
+      "epoch": 0.8316121977888471,
+      "grad_norm": 1.5925745964050293,
+      "learning_rate": 0.0002,
+      "loss": 1.1186,
+      "step": 13690
+    },
+    {
+      "epoch": 0.8322196573927834,
+      "grad_norm": 1.3945070505142212,
+      "learning_rate": 0.0002,
+      "loss": 1.2433,
+      "step": 13700
+    },
+    {
+      "epoch": 0.8328271169967197,
+      "grad_norm": 1.1743574142456055,
+      "learning_rate": 0.0002,
+      "loss": 1.187,
+      "step": 13710
+    },
+    {
+      "epoch": 0.833434576600656,
+      "grad_norm": 1.4794617891311646,
+      "learning_rate": 0.0002,
+      "loss": 1.2547,
+      "step": 13720
+    },
+    {
+      "epoch": 0.8340420362045924,
+      "grad_norm": 1.416395902633667,
+      "learning_rate": 0.0002,
+      "loss": 1.2056,
+      "step": 13730
+    },
+    {
+      "epoch": 0.8346494958085288,
+      "grad_norm": 1.6419976949691772,
+      "learning_rate": 0.0002,
+      "loss": 1.1964,
+      "step": 13740
+    },
+    {
+      "epoch": 0.835256955412465,
+      "grad_norm": 1.660413384437561,
+      "learning_rate": 0.0002,
+      "loss": 1.2177,
+      "step": 13750
+    },
+    {
+      "epoch": 0.8358644150164014,
+      "grad_norm": 1.663995623588562,
+      "learning_rate": 0.0002,
+      "loss": 1.2605,
+      "step": 13760
+    },
+    {
+      "epoch": 0.8364718746203378,
+      "grad_norm": 1.3747984170913696,
+      "learning_rate": 0.0002,
+      "loss": 1.2038,
+      "step": 13770
+    },
+    {
+      "epoch": 0.837079334224274,
+      "grad_norm": 1.2505345344543457,
+      "learning_rate": 0.0002,
+      "loss": 1.2531,
+      "step": 13780
+    },
+    {
+      "epoch": 0.8376867938282104,
+      "grad_norm": 1.347054123878479,
+      "learning_rate": 0.0002,
+      "loss": 1.2502,
+      "step": 13790
+    },
+    {
+      "epoch": 0.8382942534321468,
+      "grad_norm": 1.6971262693405151,
+      "learning_rate": 0.0002,
+      "loss": 1.0737,
+      "step": 13800
+    },
+    {
+      "epoch": 0.8389017130360831,
+      "grad_norm": 1.3297561407089233,
+      "learning_rate": 0.0002,
+      "loss": 1.1958,
+      "step": 13810
+    },
+    {
+      "epoch": 0.8395091726400195,
+      "grad_norm": 1.3878462314605713,
+      "learning_rate": 0.0002,
+      "loss": 1.1701,
+      "step": 13820
+    },
+    {
+      "epoch": 0.8401166322439558,
+      "grad_norm": 1.5290884971618652,
+      "learning_rate": 0.0002,
+      "loss": 1.2212,
+      "step": 13830
+    },
+    {
+      "epoch": 0.8407240918478921,
+      "grad_norm": 1.4563590288162231,
+      "learning_rate": 0.0002,
+      "loss": 1.1743,
+      "step": 13840
+    },
+    {
+      "epoch": 0.8413315514518285,
+      "grad_norm": 1.5538057088851929,
+      "learning_rate": 0.0002,
+      "loss": 1.1825,
+      "step": 13850
+    },
+    {
+      "epoch": 0.8419390110557647,
+      "grad_norm": 1.4252513647079468,
+      "learning_rate": 0.0002,
+      "loss": 1.1903,
+      "step": 13860
+    },
+    {
+      "epoch": 0.8425464706597011,
+      "grad_norm": 1.990596055984497,
+      "learning_rate": 0.0002,
+      "loss": 1.2232,
+      "step": 13870
+    },
+    {
+      "epoch": 0.8431539302636375,
+      "grad_norm": 1.522567868232727,
+      "learning_rate": 0.0002,
+      "loss": 1.2286,
+      "step": 13880
+    },
+    {
+      "epoch": 0.8437613898675738,
+      "grad_norm": 1.4994953870773315,
+      "learning_rate": 0.0002,
+      "loss": 1.249,
+      "step": 13890
+    },
+    {
+      "epoch": 0.8443688494715101,
+      "grad_norm": 1.4408576488494873,
+      "learning_rate": 0.0002,
+      "loss": 1.2164,
+      "step": 13900
+    },
+    {
+      "epoch": 0.8449763090754465,
+      "grad_norm": 1.6632212400436401,
+      "learning_rate": 0.0002,
+      "loss": 1.1432,
+      "step": 13910
+    },
+    {
+      "epoch": 0.8455837686793828,
+      "grad_norm": 1.4321329593658447,
+      "learning_rate": 0.0002,
+      "loss": 1.222,
+      "step": 13920
+    },
+    {
+      "epoch": 0.8461912282833192,
+      "grad_norm": 1.3293004035949707,
+      "learning_rate": 0.0002,
+      "loss": 1.1502,
+      "step": 13930
+    },
+    {
+      "epoch": 0.8467986878872555,
+      "grad_norm": 1.6593652963638306,
+      "learning_rate": 0.0002,
+      "loss": 1.2236,
+      "step": 13940
+    },
+    {
+      "epoch": 0.8474061474911918,
+      "grad_norm": 1.6421161890029907,
+      "learning_rate": 0.0002,
+      "loss": 1.2136,
+      "step": 13950
+    },
+    {
+      "epoch": 0.8480136070951282,
+      "grad_norm": 1.9198898077011108,
+      "learning_rate": 0.0002,
+      "loss": 1.2587,
+      "step": 13960
+    },
+    {
+      "epoch": 0.8486210666990646,
+      "grad_norm": 1.4113205671310425,
+      "learning_rate": 0.0002,
+      "loss": 1.2418,
+      "step": 13970
+    },
+    {
+      "epoch": 0.8492285263030008,
+      "grad_norm": 1.6185822486877441,
+      "learning_rate": 0.0002,
+      "loss": 1.2009,
+      "step": 13980
+    },
+    {
+      "epoch": 0.8498359859069372,
+      "grad_norm": 1.3802341222763062,
+      "learning_rate": 0.0002,
+      "loss": 1.1549,
+      "step": 13990
+    },
+    {
+      "epoch": 0.8504434455108735,
+      "grad_norm": 1.6448559761047363,
+      "learning_rate": 0.0002,
+      "loss": 1.1319,
+      "step": 14000
+    },
+    {
+      "epoch": 0.8510509051148099,
+      "grad_norm": 1.285009503364563,
+      "learning_rate": 0.0002,
+      "loss": 1.1371,
+      "step": 14010
+    },
+    {
+      "epoch": 0.8516583647187462,
+      "grad_norm": 1.503759503364563,
+      "learning_rate": 0.0002,
+      "loss": 1.2582,
+      "step": 14020
+    },
+    {
+      "epoch": 0.8522658243226825,
+      "grad_norm": 1.342805027961731,
+      "learning_rate": 0.0002,
+      "loss": 1.1629,
+      "step": 14030
+    },
+    {
+      "epoch": 0.8528732839266189,
+      "grad_norm": 1.5184295177459717,
+      "learning_rate": 0.0002,
+      "loss": 1.2234,
+      "step": 14040
+    },
+    {
+      "epoch": 0.8534807435305553,
+      "grad_norm": 1.5007649660110474,
+      "learning_rate": 0.0002,
+      "loss": 1.2142,
+      "step": 14050
+    },
+    {
+      "epoch": 0.8540882031344915,
+      "grad_norm": 1.497629165649414,
+      "learning_rate": 0.0002,
+      "loss": 1.1582,
+      "step": 14060
+    },
+    {
+      "epoch": 0.8546956627384279,
+      "grad_norm": 1.5224816799163818,
+      "learning_rate": 0.0002,
+      "loss": 1.211,
+      "step": 14070
+    },
+    {
+      "epoch": 0.8553031223423643,
+      "grad_norm": 1.434489369392395,
+      "learning_rate": 0.0002,
+      "loss": 1.177,
+      "step": 14080
+    },
+    {
+      "epoch": 0.8559105819463005,
+      "grad_norm": 1.7039152383804321,
+      "learning_rate": 0.0002,
+      "loss": 1.2798,
+      "step": 14090
+    },
+    {
+      "epoch": 0.8565180415502369,
+      "grad_norm": 1.3931708335876465,
+      "learning_rate": 0.0002,
+      "loss": 1.2218,
+      "step": 14100
+    },
+    {
+      "epoch": 0.8571255011541733,
+      "grad_norm": 1.5291060209274292,
+      "learning_rate": 0.0002,
+      "loss": 1.0807,
+      "step": 14110
+    },
+    {
+      "epoch": 0.8577329607581096,
+      "grad_norm": 1.3249680995941162,
+      "learning_rate": 0.0002,
+      "loss": 1.1788,
+      "step": 14120
+    },
+    {
+      "epoch": 0.8583404203620459,
+      "grad_norm": 1.4922311305999756,
+      "learning_rate": 0.0002,
+      "loss": 1.1544,
+      "step": 14130
+    },
+    {
+      "epoch": 0.8589478799659822,
+      "grad_norm": 1.4898104667663574,
+      "learning_rate": 0.0002,
+      "loss": 1.2412,
+      "step": 14140
+    },
+    {
+      "epoch": 0.8595553395699186,
+      "grad_norm": 1.8011696338653564,
+      "learning_rate": 0.0002,
+      "loss": 1.1594,
+      "step": 14150
+    },
+    {
+      "epoch": 0.860162799173855,
+      "grad_norm": 1.5143295526504517,
+      "learning_rate": 0.0002,
+      "loss": 1.2162,
+      "step": 14160
+    },
+    {
+      "epoch": 0.8607702587777912,
+      "grad_norm": 1.3194308280944824,
+      "learning_rate": 0.0002,
+      "loss": 1.2198,
+      "step": 14170
+    },
+    {
+      "epoch": 0.8613777183817276,
+      "grad_norm": 1.537085771560669,
+      "learning_rate": 0.0002,
+      "loss": 1.1495,
+      "step": 14180
+    },
+    {
+      "epoch": 0.861985177985664,
+      "grad_norm": 1.345809817314148,
+      "learning_rate": 0.0002,
+      "loss": 1.2513,
+      "step": 14190
+    },
+    {
+      "epoch": 0.8625926375896003,
+      "grad_norm": 1.3840277194976807,
+      "learning_rate": 0.0002,
+      "loss": 1.2186,
+      "step": 14200
+    },
+    {
+      "epoch": 0.8632000971935366,
+      "grad_norm": 1.876142144203186,
+      "learning_rate": 0.0002,
+      "loss": 1.213,
+      "step": 14210
+    },
+    {
+      "epoch": 0.863807556797473,
+      "grad_norm": 2.500981330871582,
+      "learning_rate": 0.0002,
+      "loss": 1.2617,
+      "step": 14220
+    },
+    {
+      "epoch": 0.8644150164014093,
+      "grad_norm": 1.5020242929458618,
+      "learning_rate": 0.0002,
+      "loss": 1.2177,
+      "step": 14230
+    },
+    {
+      "epoch": 0.8650224760053457,
+      "grad_norm": 1.2716339826583862,
+      "learning_rate": 0.0002,
+      "loss": 1.1125,
+      "step": 14240
+    },
+    {
+      "epoch": 0.865629935609282,
+      "grad_norm": 1.7657488584518433,
+      "learning_rate": 0.0002,
+      "loss": 1.2233,
+      "step": 14250
+    },
+    {
+      "epoch": 0.8662373952132183,
+      "grad_norm": 1.4088470935821533,
+      "learning_rate": 0.0002,
+      "loss": 1.2144,
+      "step": 14260
+    },
+    {
+      "epoch": 0.8668448548171547,
+      "grad_norm": 1.5708025693893433,
+      "learning_rate": 0.0002,
+      "loss": 1.1763,
+      "step": 14270
+    },
+    {
+      "epoch": 0.8674523144210909,
+      "grad_norm": 1.4343926906585693,
+      "learning_rate": 0.0002,
+      "loss": 1.2017,
+      "step": 14280
+    },
+    {
+      "epoch": 0.8680597740250273,
+      "grad_norm": 1.3481239080429077,
+      "learning_rate": 0.0002,
+      "loss": 1.1584,
+      "step": 14290
+    },
+    {
+      "epoch": 0.8686672336289637,
+      "grad_norm": 1.3848117589950562,
+      "learning_rate": 0.0002,
+      "loss": 1.2323,
+      "step": 14300
+    },
+    {
+      "epoch": 0.8692746932329,
+      "grad_norm": 1.6503533124923706,
+      "learning_rate": 0.0002,
+      "loss": 1.2011,
+      "step": 14310
+    },
+    {
+      "epoch": 0.8698821528368363,
+      "grad_norm": 1.3893461227416992,
+      "learning_rate": 0.0002,
+      "loss": 1.1885,
+      "step": 14320
+    },
+    {
+      "epoch": 0.8704896124407727,
+      "grad_norm": 1.670220136642456,
+      "learning_rate": 0.0002,
+      "loss": 1.2753,
+      "step": 14330
+    },
+    {
+      "epoch": 0.871097072044709,
+      "grad_norm": 1.4308701753616333,
+      "learning_rate": 0.0002,
+      "loss": 1.1954,
+      "step": 14340
+    },
+    {
+      "epoch": 0.8717045316486454,
+      "grad_norm": 1.6356762647628784,
+      "learning_rate": 0.0002,
+      "loss": 1.1151,
+      "step": 14350
+    },
+    {
+      "epoch": 0.8723119912525817,
+      "grad_norm": 1.4090251922607422,
+      "learning_rate": 0.0002,
+      "loss": 1.1629,
+      "step": 14360
+    },
+    {
+      "epoch": 0.872919450856518,
+      "grad_norm": 1.4654072523117065,
+      "learning_rate": 0.0002,
+      "loss": 1.1324,
+      "step": 14370
+    },
+    {
+      "epoch": 0.8735269104604544,
+      "grad_norm": 1.4703407287597656,
+      "learning_rate": 0.0002,
+      "loss": 1.1466,
+      "step": 14380
+    },
+    {
+      "epoch": 0.8741343700643908,
+      "grad_norm": 1.6721851825714111,
+      "learning_rate": 0.0002,
+      "loss": 1.1348,
+      "step": 14390
+    },
+    {
+      "epoch": 0.874741829668327,
+      "grad_norm": 1.5856989622116089,
+      "learning_rate": 0.0002,
+      "loss": 1.2488,
+      "step": 14400
+    },
+    {
+      "epoch": 0.8753492892722634,
+      "grad_norm": 1.6506541967391968,
+      "learning_rate": 0.0002,
+      "loss": 1.2272,
+      "step": 14410
+    },
+    {
+      "epoch": 0.8759567488761998,
+      "grad_norm": 1.6315511465072632,
+      "learning_rate": 0.0002,
+      "loss": 1.2255,
+      "step": 14420
+    },
+    {
+      "epoch": 0.876564208480136,
+      "grad_norm": 1.3775477409362793,
+      "learning_rate": 0.0002,
+      "loss": 1.2079,
+      "step": 14430
+    },
+    {
+      "epoch": 0.8771716680840724,
+      "grad_norm": 1.3681683540344238,
+      "learning_rate": 0.0002,
+      "loss": 1.2056,
+      "step": 14440
+    },
+    {
+      "epoch": 0.8777791276880087,
+      "grad_norm": 1.6075377464294434,
+      "learning_rate": 0.0002,
+      "loss": 1.105,
+      "step": 14450
+    },
+    {
+      "epoch": 0.8783865872919451,
+      "grad_norm": 1.5112967491149902,
+      "learning_rate": 0.0002,
+      "loss": 1.211,
+      "step": 14460
+    },
+    {
+      "epoch": 0.8789940468958815,
+      "grad_norm": 1.3677444458007812,
+      "learning_rate": 0.0002,
+      "loss": 1.1737,
+      "step": 14470
+    },
+    {
+      "epoch": 0.8796015064998177,
+      "grad_norm": 1.5045770406723022,
+      "learning_rate": 0.0002,
+      "loss": 1.1354,
+      "step": 14480
+    },
+    {
+      "epoch": 0.8802089661037541,
+      "grad_norm": 1.5426734685897827,
+      "learning_rate": 0.0002,
+      "loss": 1.2089,
+      "step": 14490
+    },
+    {
+      "epoch": 0.8808164257076905,
+      "grad_norm": 1.314050316810608,
+      "learning_rate": 0.0002,
+      "loss": 1.218,
+      "step": 14500
+    },
+    {
+      "epoch": 0.8814238853116267,
+      "grad_norm": 1.5524317026138306,
+      "learning_rate": 0.0002,
+      "loss": 1.1771,
+      "step": 14510
+    },
+    {
+      "epoch": 0.8820313449155631,
+      "grad_norm": 1.3775534629821777,
+      "learning_rate": 0.0002,
+      "loss": 1.1898,
+      "step": 14520
+    },
+    {
+      "epoch": 0.8826388045194995,
+      "grad_norm": 1.5032004117965698,
+      "learning_rate": 0.0002,
+      "loss": 1.1826,
+      "step": 14530
+    },
+    {
+      "epoch": 0.8832462641234358,
+      "grad_norm": 1.556536316871643,
+      "learning_rate": 0.0002,
+      "loss": 1.2183,
+      "step": 14540
+    },
+    {
+      "epoch": 0.8838537237273721,
+      "grad_norm": 1.4886997938156128,
+      "learning_rate": 0.0002,
+      "loss": 1.2234,
+      "step": 14550
+    },
+    {
+      "epoch": 0.8844611833313085,
+      "grad_norm": 1.4067894220352173,
+      "learning_rate": 0.0002,
+      "loss": 1.2221,
+      "step": 14560
+    },
+    {
+      "epoch": 0.8850686429352448,
+      "grad_norm": 1.386600375175476,
+      "learning_rate": 0.0002,
+      "loss": 1.1801,
+      "step": 14570
+    },
+    {
+      "epoch": 0.8856761025391812,
+      "grad_norm": 1.4741544723510742,
+      "learning_rate": 0.0002,
+      "loss": 1.1859,
+      "step": 14580
+    },
+    {
+      "epoch": 0.8862835621431174,
+      "grad_norm": 1.3699150085449219,
+      "learning_rate": 0.0002,
+      "loss": 1.1368,
+      "step": 14590
+    },
+    {
+      "epoch": 0.8868910217470538,
+      "grad_norm": 1.3016911745071411,
+      "learning_rate": 0.0002,
+      "loss": 1.1673,
+      "step": 14600
+    },
+    {
+      "epoch": 0.8874984813509902,
+      "grad_norm": 1.4108279943466187,
+      "learning_rate": 0.0002,
+      "loss": 1.2522,
+      "step": 14610
+    },
+    {
+      "epoch": 0.8881059409549265,
+      "grad_norm": 1.3857357501983643,
+      "learning_rate": 0.0002,
+      "loss": 1.167,
+      "step": 14620
+    },
+    {
+      "epoch": 0.8887134005588628,
+      "grad_norm": 1.4877088069915771,
+      "learning_rate": 0.0002,
+      "loss": 1.1227,
+      "step": 14630
+    },
+    {
+      "epoch": 0.8893208601627992,
+      "grad_norm": 1.6803112030029297,
+      "learning_rate": 0.0002,
+      "loss": 1.1929,
+      "step": 14640
+    },
+    {
+      "epoch": 0.8899283197667355,
+      "grad_norm": 1.449785828590393,
+      "learning_rate": 0.0002,
+      "loss": 1.2454,
+      "step": 14650
+    },
+    {
+      "epoch": 0.8905357793706719,
+      "grad_norm": 1.3783334493637085,
+      "learning_rate": 0.0002,
+      "loss": 1.1384,
+      "step": 14660
+    },
+    {
+      "epoch": 0.8911432389746082,
+      "grad_norm": 1.7931047677993774,
+      "learning_rate": 0.0002,
+      "loss": 1.1846,
+      "step": 14670
+    },
+    {
+      "epoch": 0.8917506985785445,
+      "grad_norm": 1.419252634048462,
+      "learning_rate": 0.0002,
+      "loss": 1.1844,
+      "step": 14680
+    },
+    {
+      "epoch": 0.8923581581824809,
+      "grad_norm": 1.3987175226211548,
+      "learning_rate": 0.0002,
+      "loss": 1.1986,
+      "step": 14690
+    },
+    {
+      "epoch": 0.8929656177864173,
+      "grad_norm": 1.397653579711914,
+      "learning_rate": 0.0002,
+      "loss": 1.1829,
+      "step": 14700
+    },
+    {
+      "epoch": 0.8935730773903535,
+      "grad_norm": 1.4905036687850952,
+      "learning_rate": 0.0002,
+      "loss": 1.195,
+      "step": 14710
+    },
+    {
+      "epoch": 0.8941805369942899,
+      "grad_norm": 1.6301939487457275,
+      "learning_rate": 0.0002,
+      "loss": 1.2021,
+      "step": 14720
+    },
+    {
+      "epoch": 0.8947879965982262,
+      "grad_norm": 1.6625646352767944,
+      "learning_rate": 0.0002,
+      "loss": 1.1257,
+      "step": 14730
+    },
+    {
+      "epoch": 0.8953954562021625,
+      "grad_norm": 1.528059959411621,
+      "learning_rate": 0.0002,
+      "loss": 1.2176,
+      "step": 14740
+    },
+    {
+      "epoch": 0.8960029158060989,
+      "grad_norm": 1.5763529539108276,
+      "learning_rate": 0.0002,
+      "loss": 1.221,
+      "step": 14750
+    },
+    {
+      "epoch": 0.8966103754100352,
+      "grad_norm": 1.5408121347427368,
+      "learning_rate": 0.0002,
+      "loss": 1.2203,
+      "step": 14760
+    },
+    {
+      "epoch": 0.8972178350139716,
+      "grad_norm": 1.487208604812622,
+      "learning_rate": 0.0002,
+      "loss": 1.2525,
+      "step": 14770
+    },
+    {
+      "epoch": 0.8978252946179079,
+      "grad_norm": 1.3733705282211304,
+      "learning_rate": 0.0002,
+      "loss": 1.2161,
+      "step": 14780
+    },
+    {
+      "epoch": 0.8984327542218442,
+      "grad_norm": 1.36906898021698,
+      "learning_rate": 0.0002,
+      "loss": 1.129,
+      "step": 14790
+    },
+    {
+      "epoch": 0.8990402138257806,
+      "grad_norm": 1.6836575269699097,
+      "learning_rate": 0.0002,
+      "loss": 1.1591,
+      "step": 14800
+    },
+    {
+      "epoch": 0.899647673429717,
+      "grad_norm": 1.3825420141220093,
+      "learning_rate": 0.0002,
+      "loss": 1.2115,
+      "step": 14810
+    },
+    {
+      "epoch": 0.9002551330336532,
+      "grad_norm": 1.5129579305648804,
+      "learning_rate": 0.0002,
+      "loss": 1.1001,
+      "step": 14820
+    },
+    {
+      "epoch": 0.9008625926375896,
+      "grad_norm": 1.5678008794784546,
+      "learning_rate": 0.0002,
+      "loss": 1.1748,
+      "step": 14830
+    },
+    {
+      "epoch": 0.901470052241526,
+      "grad_norm": 1.2213751077651978,
+      "learning_rate": 0.0002,
+      "loss": 1.1339,
+      "step": 14840
+    },
+    {
+      "epoch": 0.9020775118454623,
+      "grad_norm": 1.646059274673462,
+      "learning_rate": 0.0002,
+      "loss": 1.2827,
+      "step": 14850
+    },
+    {
+      "epoch": 0.9026849714493986,
+      "grad_norm": 1.31034517288208,
+      "learning_rate": 0.0002,
+      "loss": 1.1171,
+      "step": 14860
+    },
+    {
+      "epoch": 0.9032924310533349,
+      "grad_norm": 1.6546515226364136,
+      "learning_rate": 0.0002,
+      "loss": 1.1585,
+      "step": 14870
+    },
+    {
+      "epoch": 0.9038998906572713,
+      "grad_norm": 1.4461085796356201,
+      "learning_rate": 0.0002,
+      "loss": 1.2693,
+      "step": 14880
+    },
+    {
+      "epoch": 0.9045073502612077,
+      "grad_norm": 1.4356815814971924,
+      "learning_rate": 0.0002,
+      "loss": 1.2083,
+      "step": 14890
+    },
+    {
+      "epoch": 0.9051148098651439,
+      "grad_norm": 1.4976754188537598,
+      "learning_rate": 0.0002,
+      "loss": 1.2562,
+      "step": 14900
+    },
+    {
+      "epoch": 0.9057222694690803,
+      "grad_norm": 1.4736928939819336,
+      "learning_rate": 0.0002,
+      "loss": 1.2292,
+      "step": 14910
+    },
+    {
+      "epoch": 0.9063297290730167,
+      "grad_norm": 1.329288363456726,
+      "learning_rate": 0.0002,
+      "loss": 1.1906,
+      "step": 14920
+    },
+    {
+      "epoch": 0.9069371886769529,
+      "grad_norm": 1.553163766860962,
+      "learning_rate": 0.0002,
+      "loss": 1.2025,
+      "step": 14930
+    },
+    {
+      "epoch": 0.9075446482808893,
+      "grad_norm": 1.405267357826233,
+      "learning_rate": 0.0002,
+      "loss": 1.1678,
+      "step": 14940
+    },
+    {
+      "epoch": 0.9081521078848257,
+      "grad_norm": 1.3324331045150757,
+      "learning_rate": 0.0002,
+      "loss": 1.2025,
+      "step": 14950
+    },
+    {
+      "epoch": 0.908759567488762,
+      "grad_norm": 1.5258556604385376,
+      "learning_rate": 0.0002,
+      "loss": 1.1644,
+      "step": 14960
+    },
+    {
+      "epoch": 0.9093670270926983,
+      "grad_norm": 1.6733620166778564,
+      "learning_rate": 0.0002,
+      "loss": 1.2605,
+      "step": 14970
+    },
+    {
+      "epoch": 0.9099744866966347,
+      "grad_norm": 1.6827983856201172,
+      "learning_rate": 0.0002,
+      "loss": 1.2255,
+      "step": 14980
+    },
+    {
+      "epoch": 0.910581946300571,
+      "grad_norm": 1.6709449291229248,
+      "learning_rate": 0.0002,
+      "loss": 1.2249,
+      "step": 14990
+    },
+    {
+      "epoch": 0.9111894059045074,
+      "grad_norm": 1.3499102592468262,
+      "learning_rate": 0.0002,
+      "loss": 1.2224,
+      "step": 15000
+    },
+    {
+      "epoch": 0.9117968655084437,
+      "grad_norm": 1.3916763067245483,
+      "learning_rate": 0.0002,
+      "loss": 1.2075,
+      "step": 15010
+    },
+    {
+      "epoch": 0.91240432511238,
+      "grad_norm": 1.597317099571228,
+      "learning_rate": 0.0002,
+      "loss": 1.174,
+      "step": 15020
+    },
+    {
+      "epoch": 0.9130117847163164,
+      "grad_norm": 1.6222524642944336,
+      "learning_rate": 0.0002,
+      "loss": 1.0985,
+      "step": 15030
+    },
+    {
+      "epoch": 0.9136192443202527,
+      "grad_norm": 1.4718647003173828,
+      "learning_rate": 0.0002,
+      "loss": 1.1341,
+      "step": 15040
+    },
+    {
+      "epoch": 0.914226703924189,
+      "grad_norm": 1.321515440940857,
+      "learning_rate": 0.0002,
+      "loss": 1.2499,
+      "step": 15050
+    },
+    {
+      "epoch": 0.9148341635281254,
+      "grad_norm": 1.4551115036010742,
+      "learning_rate": 0.0002,
+      "loss": 1.2157,
+      "step": 15060
+    },
+    {
+      "epoch": 0.9154416231320617,
+      "grad_norm": 1.5374985933303833,
+      "learning_rate": 0.0002,
+      "loss": 1.1934,
+      "step": 15070
+    },
+    {
+      "epoch": 0.916049082735998,
+      "grad_norm": 1.4193897247314453,
+      "learning_rate": 0.0002,
+      "loss": 1.1481,
+      "step": 15080
+    },
+    {
+      "epoch": 0.9166565423399344,
+      "grad_norm": 1.4679774045944214,
+      "learning_rate": 0.0002,
+      "loss": 1.1511,
+      "step": 15090
+    },
+    {
+      "epoch": 0.9172640019438707,
+      "grad_norm": 1.3401442766189575,
+      "learning_rate": 0.0002,
+      "loss": 1.1982,
+      "step": 15100
+    },
+    {
+      "epoch": 0.9178714615478071,
+      "grad_norm": 1.5043041706085205,
+      "learning_rate": 0.0002,
+      "loss": 1.2371,
+      "step": 15110
+    },
+    {
+      "epoch": 0.9184789211517435,
+      "grad_norm": 1.5215322971343994,
+      "learning_rate": 0.0002,
+      "loss": 1.2488,
+      "step": 15120
+    },
+    {
+      "epoch": 0.9190863807556797,
+      "grad_norm": 1.3586745262145996,
+      "learning_rate": 0.0002,
+      "loss": 1.1887,
+      "step": 15130
+    },
+    {
+      "epoch": 0.9196938403596161,
+      "grad_norm": 1.4167375564575195,
+      "learning_rate": 0.0002,
+      "loss": 1.2293,
+      "step": 15140
+    },
+    {
+      "epoch": 0.9203012999635525,
+      "grad_norm": 1.3452075719833374,
+      "learning_rate": 0.0002,
+      "loss": 1.1539,
+      "step": 15150
+    },
+    {
+      "epoch": 0.9209087595674887,
+      "grad_norm": 1.6305108070373535,
+      "learning_rate": 0.0002,
+      "loss": 1.1827,
+      "step": 15160
+    },
+    {
+      "epoch": 0.9215162191714251,
+      "grad_norm": 1.34440279006958,
+      "learning_rate": 0.0002,
+      "loss": 1.1002,
+      "step": 15170
+    },
+    {
+      "epoch": 0.9221236787753614,
+      "grad_norm": 1.5065081119537354,
+      "learning_rate": 0.0002,
+      "loss": 1.2291,
+      "step": 15180
+    },
+    {
+      "epoch": 0.9227311383792978,
+      "grad_norm": 1.6292963027954102,
+      "learning_rate": 0.0002,
+      "loss": 1.1404,
+      "step": 15190
+    },
+    {
+      "epoch": 0.9233385979832341,
+      "grad_norm": 2.1859641075134277,
+      "learning_rate": 0.0002,
+      "loss": 1.2444,
+      "step": 15200
+    },
+    {
+      "epoch": 0.9239460575871704,
+      "grad_norm": 1.3960617780685425,
+      "learning_rate": 0.0002,
+      "loss": 1.2397,
+      "step": 15210
+    },
+    {
+      "epoch": 0.9245535171911068,
+      "grad_norm": 1.3792656660079956,
+      "learning_rate": 0.0002,
+      "loss": 1.1714,
+      "step": 15220
+    },
+    {
+      "epoch": 0.9251609767950432,
+      "grad_norm": 1.6836884021759033,
+      "learning_rate": 0.0002,
+      "loss": 1.19,
+      "step": 15230
+    },
+    {
+      "epoch": 0.9257684363989794,
+      "grad_norm": 1.6327534914016724,
+      "learning_rate": 0.0002,
+      "loss": 1.1643,
+      "step": 15240
+    },
+    {
+      "epoch": 0.9263758960029158,
+      "grad_norm": 1.3023805618286133,
+      "learning_rate": 0.0002,
+      "loss": 1.2351,
+      "step": 15250
+    },
+    {
+      "epoch": 0.9269833556068522,
+      "grad_norm": 1.4439232349395752,
+      "learning_rate": 0.0002,
+      "loss": 1.1851,
+      "step": 15260
+    },
+    {
+      "epoch": 0.9275908152107885,
+      "grad_norm": 1.5840705633163452,
+      "learning_rate": 0.0002,
+      "loss": 1.1825,
+      "step": 15270
+    },
+    {
+      "epoch": 0.9281982748147248,
+      "grad_norm": 1.615294337272644,
+      "learning_rate": 0.0002,
+      "loss": 1.28,
+      "step": 15280
+    },
+    {
+      "epoch": 0.9288057344186612,
+      "grad_norm": 1.436958909034729,
+      "learning_rate": 0.0002,
+      "loss": 1.1405,
+      "step": 15290
+    },
+    {
+      "epoch": 0.9294131940225975,
+      "grad_norm": 1.6398800611495972,
+      "learning_rate": 0.0002,
+      "loss": 1.2347,
+      "step": 15300
+    },
+    {
+      "epoch": 0.9300206536265339,
+      "grad_norm": 1.5404356718063354,
+      "learning_rate": 0.0002,
+      "loss": 1.2103,
+      "step": 15310
+    },
+    {
+      "epoch": 0.9306281132304701,
+      "grad_norm": 1.55526602268219,
+      "learning_rate": 0.0002,
+      "loss": 1.2314,
+      "step": 15320
+    },
+    {
+      "epoch": 0.9312355728344065,
+      "grad_norm": 1.4641153812408447,
+      "learning_rate": 0.0002,
+      "loss": 1.3062,
+      "step": 15330
+    },
+    {
+      "epoch": 0.9318430324383429,
+      "grad_norm": 1.5352332592010498,
+      "learning_rate": 0.0002,
+      "loss": 1.2254,
+      "step": 15340
+    },
+    {
+      "epoch": 0.9324504920422791,
+      "grad_norm": 1.3203394412994385,
+      "learning_rate": 0.0002,
+      "loss": 1.1402,
+      "step": 15350
+    },
+    {
+      "epoch": 0.9330579516462155,
+      "grad_norm": 1.4581013917922974,
+      "learning_rate": 0.0002,
+      "loss": 1.1929,
+      "step": 15360
+    },
+    {
+      "epoch": 0.9336654112501519,
+      "grad_norm": 1.3729881048202515,
+      "learning_rate": 0.0002,
+      "loss": 1.2497,
+      "step": 15370
+    },
+    {
+      "epoch": 0.9342728708540882,
+      "grad_norm": 1.5498683452606201,
+      "learning_rate": 0.0002,
+      "loss": 1.2437,
+      "step": 15380
+    },
+    {
+      "epoch": 0.9348803304580245,
+      "grad_norm": 1.445510745048523,
+      "learning_rate": 0.0002,
+      "loss": 1.2578,
+      "step": 15390
+    },
+    {
+      "epoch": 0.9354877900619609,
+      "grad_norm": 1.4345883131027222,
+      "learning_rate": 0.0002,
+      "loss": 1.2261,
+      "step": 15400
+    },
+    {
+      "epoch": 0.9360952496658972,
+      "grad_norm": 1.5046039819717407,
+      "learning_rate": 0.0002,
+      "loss": 1.1078,
+      "step": 15410
+    },
+    {
+      "epoch": 0.9367027092698336,
+      "grad_norm": 1.4676977396011353,
+      "learning_rate": 0.0002,
+      "loss": 1.2289,
+      "step": 15420
+    },
+    {
+      "epoch": 0.9373101688737699,
+      "grad_norm": 1.6066975593566895,
+      "learning_rate": 0.0002,
+      "loss": 1.1907,
+      "step": 15430
+    },
+    {
+      "epoch": 0.9379176284777062,
+      "grad_norm": 1.4932212829589844,
+      "learning_rate": 0.0002,
+      "loss": 1.2052,
+      "step": 15440
+    },
+    {
+      "epoch": 0.9385250880816426,
+      "grad_norm": 1.3368092775344849,
+      "learning_rate": 0.0002,
+      "loss": 1.162,
+      "step": 15450
+    },
+    {
+      "epoch": 0.9391325476855789,
+      "grad_norm": 1.5057190656661987,
+      "learning_rate": 0.0002,
+      "loss": 1.2014,
+      "step": 15460
+    },
+    {
+      "epoch": 0.9397400072895152,
+      "grad_norm": 1.8288530111312866,
+      "learning_rate": 0.0002,
+      "loss": 1.1495,
+      "step": 15470
+    },
+    {
+      "epoch": 0.9403474668934516,
+      "grad_norm": 1.592646598815918,
+      "learning_rate": 0.0002,
+      "loss": 1.201,
+      "step": 15480
+    },
+    {
+      "epoch": 0.9409549264973879,
+      "grad_norm": 1.517626166343689,
+      "learning_rate": 0.0002,
+      "loss": 1.2786,
+      "step": 15490
+    },
+    {
+      "epoch": 0.9415623861013243,
+      "grad_norm": 1.4681724309921265,
+      "learning_rate": 0.0002,
+      "loss": 1.1512,
+      "step": 15500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 49386,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.582908054057796e+18,
+  "train_batch_size": 3,
+  "trial_name": null,
+  "trial_params": null
+}