diff --git "a/few_shot/checkpoint-6500/trainer_state.json" "b/few_shot/checkpoint-6500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/few_shot/checkpoint-6500/trainer_state.json"
@@ -0,0 +1,4583 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.39484874255861985,
+  "eval_steps": 500,
+  "global_step": 6500,
+  "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
+    }
+  ],
+  "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": 3.5992131765150597e+18,
+  "train_batch_size": 3,
+  "trial_name": null,
+  "trial_params": null
+}