diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6669 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 9480,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0010548523206751054,
+      "grad_norm": 0.8035507798194885,
+      "learning_rate": 0.0008438818565400844,
+      "loss": 7.4218,
+      "step": 10
+    },
+    {
+      "epoch": 0.002109704641350211,
+      "grad_norm": 0.7049233317375183,
+      "learning_rate": 0.0016877637130801688,
+      "loss": 6.8073,
+      "step": 20
+    },
+    {
+      "epoch": 0.0031645569620253164,
+      "grad_norm": 2.4359562397003174,
+      "learning_rate": 0.002531645569620253,
+      "loss": 6.2083,
+      "step": 30
+    },
+    {
+      "epoch": 0.004219409282700422,
+      "grad_norm": 0.6199546456336975,
+      "learning_rate": 0.0033755274261603376,
+      "loss": 5.7982,
+      "step": 40
+    },
+    {
+      "epoch": 0.005274261603375527,
+      "grad_norm": 0.4143601655960083,
+      "learning_rate": 0.0042194092827004225,
+      "loss": 5.4256,
+      "step": 50
+    },
+    {
+      "epoch": 0.006329113924050633,
+      "grad_norm": 1.3290952444076538,
+      "learning_rate": 0.005063291139240506,
+      "loss": 5.2,
+      "step": 60
+    },
+    {
+      "epoch": 0.007383966244725738,
+      "grad_norm": 0.38381659984588623,
+      "learning_rate": 0.005907172995780591,
+      "loss": 5.0754,
+      "step": 70
+    },
+    {
+      "epoch": 0.008438818565400843,
+      "grad_norm": 0.27451393008232117,
+      "learning_rate": 0.006751054852320675,
+      "loss": 4.9207,
+      "step": 80
+    },
+    {
+      "epoch": 0.00949367088607595,
+      "grad_norm": 0.5156111717224121,
+      "learning_rate": 0.007594936708860759,
+      "loss": 4.7668,
+      "step": 90
+    },
+    {
+      "epoch": 0.010548523206751054,
+      "grad_norm": 1.1935275793075562,
+      "learning_rate": 0.008,
+      "loss": 4.6078,
+      "step": 100
+    },
+    {
+      "epoch": 0.011603375527426161,
+      "grad_norm": 0.22459286451339722,
+      "learning_rate": 0.008,
+      "loss": 4.4421,
+      "step": 110
+    },
+    {
+      "epoch": 0.012658227848101266,
+      "grad_norm": 0.47292840480804443,
+      "learning_rate": 0.008,
+      "loss": 4.2452,
+      "step": 120
+    },
+    {
+      "epoch": 0.013713080168776372,
+      "grad_norm": 0.44823741912841797,
+      "learning_rate": 0.008,
+      "loss": 4.0701,
+      "step": 130
+    },
+    {
+      "epoch": 0.014767932489451477,
+      "grad_norm": 0.2519300878047943,
+      "learning_rate": 0.008,
+      "loss": 3.9309,
+      "step": 140
+    },
+    {
+      "epoch": 0.015822784810126583,
+      "grad_norm": 0.5491886734962463,
+      "learning_rate": 0.008,
+      "loss": 3.7876,
+      "step": 150
+    },
+    {
+      "epoch": 0.016877637130801686,
+      "grad_norm": 0.25105059146881104,
+      "learning_rate": 0.008,
+      "loss": 3.7333,
+      "step": 160
+    },
+    {
+      "epoch": 0.017932489451476793,
+      "grad_norm": 0.23800434172153473,
+      "learning_rate": 0.008,
+      "loss": 3.6199,
+      "step": 170
+    },
+    {
+      "epoch": 0.0189873417721519,
+      "grad_norm": 0.3990011215209961,
+      "learning_rate": 0.008,
+      "loss": 3.5423,
+      "step": 180
+    },
+    {
+      "epoch": 0.020042194092827006,
+      "grad_norm": 0.36305317282676697,
+      "learning_rate": 0.008,
+      "loss": 3.4857,
+      "step": 190
+    },
+    {
+      "epoch": 0.02109704641350211,
+      "grad_norm": 0.4093531370162964,
+      "learning_rate": 0.008,
+      "loss": 3.4205,
+      "step": 200
+    },
+    {
+      "epoch": 0.022151898734177215,
+      "grad_norm": 0.37019696831703186,
+      "learning_rate": 0.008,
+      "loss": 3.351,
+      "step": 210
+    },
+    {
+      "epoch": 0.023206751054852322,
+      "grad_norm": 0.6104243993759155,
+      "learning_rate": 0.008,
+      "loss": 3.3241,
+      "step": 220
+    },
+    {
+      "epoch": 0.024261603375527425,
+      "grad_norm": 0.32222509384155273,
+      "learning_rate": 0.008,
+      "loss": 3.2512,
+      "step": 230
+    },
+    {
+      "epoch": 0.02531645569620253,
+      "grad_norm": 0.6900892853736877,
+      "learning_rate": 0.008,
+      "loss": 3.2034,
+      "step": 240
+    },
+    {
+      "epoch": 0.026371308016877638,
+      "grad_norm": 0.4173062741756439,
+      "learning_rate": 0.008,
+      "loss": 3.1952,
+      "step": 250
+    },
+    {
+      "epoch": 0.027426160337552744,
+      "grad_norm": 0.2634281814098358,
+      "learning_rate": 0.008,
+      "loss": 3.1335,
+      "step": 260
+    },
+    {
+      "epoch": 0.028481012658227847,
+      "grad_norm": 0.31951773166656494,
+      "learning_rate": 0.008,
+      "loss": 3.0986,
+      "step": 270
+    },
+    {
+      "epoch": 0.029535864978902954,
+      "grad_norm": 0.5596627593040466,
+      "learning_rate": 0.008,
+      "loss": 3.0555,
+      "step": 280
+    },
+    {
+      "epoch": 0.03059071729957806,
+      "grad_norm": 0.3591410517692566,
+      "learning_rate": 0.008,
+      "loss": 3.0513,
+      "step": 290
+    },
+    {
+      "epoch": 0.03164556962025317,
+      "grad_norm": 0.2355412244796753,
+      "learning_rate": 0.008,
+      "loss": 3.0092,
+      "step": 300
+    },
+    {
+      "epoch": 0.03270042194092827,
+      "grad_norm": 0.45172497630119324,
+      "learning_rate": 0.008,
+      "loss": 2.981,
+      "step": 310
+    },
+    {
+      "epoch": 0.03375527426160337,
+      "grad_norm": 0.32593342661857605,
+      "learning_rate": 0.008,
+      "loss": 2.9615,
+      "step": 320
+    },
+    {
+      "epoch": 0.03481012658227848,
+      "grad_norm": 0.43027985095977783,
+      "learning_rate": 0.008,
+      "loss": 2.9483,
+      "step": 330
+    },
+    {
+      "epoch": 0.035864978902953586,
+      "grad_norm": 0.35132601857185364,
+      "learning_rate": 0.008,
+      "loss": 2.9266,
+      "step": 340
+    },
+    {
+      "epoch": 0.03691983122362869,
+      "grad_norm": 0.313542902469635,
+      "learning_rate": 0.008,
+      "loss": 2.8867,
+      "step": 350
+    },
+    {
+      "epoch": 0.0379746835443038,
+      "grad_norm": 0.39107778668403625,
+      "learning_rate": 0.008,
+      "loss": 2.8713,
+      "step": 360
+    },
+    {
+      "epoch": 0.039029535864978905,
+      "grad_norm": 0.2927038371562958,
+      "learning_rate": 0.008,
+      "loss": 2.8617,
+      "step": 370
+    },
+    {
+      "epoch": 0.04008438818565401,
+      "grad_norm": 0.390756219625473,
+      "learning_rate": 0.008,
+      "loss": 2.852,
+      "step": 380
+    },
+    {
+      "epoch": 0.04113924050632911,
+      "grad_norm": 0.3277692496776581,
+      "learning_rate": 0.008,
+      "loss": 2.8347,
+      "step": 390
+    },
+    {
+      "epoch": 0.04219409282700422,
+      "grad_norm": 0.28730741143226624,
+      "learning_rate": 0.008,
+      "loss": 2.8107,
+      "step": 400
+    },
+    {
+      "epoch": 0.043248945147679324,
+      "grad_norm": 0.42436885833740234,
+      "learning_rate": 0.008,
+      "loss": 2.8092,
+      "step": 410
+    },
+    {
+      "epoch": 0.04430379746835443,
+      "grad_norm": 0.3842144310474396,
+      "learning_rate": 0.008,
+      "loss": 2.791,
+      "step": 420
+    },
+    {
+      "epoch": 0.04535864978902954,
+      "grad_norm": 0.40783390402793884,
+      "learning_rate": 0.008,
+      "loss": 2.7616,
+      "step": 430
+    },
+    {
+      "epoch": 0.046413502109704644,
+      "grad_norm": 0.3010316789150238,
+      "learning_rate": 0.008,
+      "loss": 2.7528,
+      "step": 440
+    },
+    {
+      "epoch": 0.04746835443037975,
+      "grad_norm": 0.3602994680404663,
+      "learning_rate": 0.008,
+      "loss": 2.7441,
+      "step": 450
+    },
+    {
+      "epoch": 0.04852320675105485,
+      "grad_norm": 0.3325028717517853,
+      "learning_rate": 0.008,
+      "loss": 2.7307,
+      "step": 460
+    },
+    {
+      "epoch": 0.049578059071729956,
+      "grad_norm": 0.43379560112953186,
+      "learning_rate": 0.008,
+      "loss": 2.7111,
+      "step": 470
+    },
+    {
+      "epoch": 0.05063291139240506,
+      "grad_norm": 0.34409594535827637,
+      "learning_rate": 0.008,
+      "loss": 2.713,
+      "step": 480
+    },
+    {
+      "epoch": 0.05168776371308017,
+      "grad_norm": 0.25509005784988403,
+      "learning_rate": 0.008,
+      "loss": 2.6863,
+      "step": 490
+    },
+    {
+      "epoch": 0.052742616033755275,
+      "grad_norm": 0.35922324657440186,
+      "learning_rate": 0.008,
+      "loss": 2.6866,
+      "step": 500
+    },
+    {
+      "epoch": 0.05379746835443038,
+      "grad_norm": 0.35630449652671814,
+      "learning_rate": 0.008,
+      "loss": 2.6789,
+      "step": 510
+    },
+    {
+      "epoch": 0.05485232067510549,
+      "grad_norm": 0.3219009041786194,
+      "learning_rate": 0.008,
+      "loss": 2.6611,
+      "step": 520
+    },
+    {
+      "epoch": 0.05590717299578059,
+      "grad_norm": 0.28569942712783813,
+      "learning_rate": 0.008,
+      "loss": 2.6434,
+      "step": 530
+    },
+    {
+      "epoch": 0.056962025316455694,
+      "grad_norm": 0.37412121891975403,
+      "learning_rate": 0.008,
+      "loss": 2.6466,
+      "step": 540
+    },
+    {
+      "epoch": 0.0580168776371308,
+      "grad_norm": 0.2525637149810791,
+      "learning_rate": 0.008,
+      "loss": 2.6508,
+      "step": 550
+    },
+    {
+      "epoch": 0.05907172995780591,
+      "grad_norm": 0.2444714605808258,
+      "learning_rate": 0.008,
+      "loss": 2.6011,
+      "step": 560
+    },
+    {
+      "epoch": 0.060126582278481014,
+      "grad_norm": 0.3005289137363434,
+      "learning_rate": 0.008,
+      "loss": 2.6024,
+      "step": 570
+    },
+    {
+      "epoch": 0.06118143459915612,
+      "grad_norm": 0.3358891010284424,
+      "learning_rate": 0.008,
+      "loss": 2.6062,
+      "step": 580
+    },
+    {
+      "epoch": 0.06223628691983123,
+      "grad_norm": 0.2609458267688751,
+      "learning_rate": 0.008,
+      "loss": 2.5746,
+      "step": 590
+    },
+    {
+      "epoch": 0.06329113924050633,
+      "grad_norm": 0.3318046033382416,
+      "learning_rate": 0.008,
+      "loss": 2.5614,
+      "step": 600
+    },
+    {
+      "epoch": 0.06434599156118144,
+      "grad_norm": 0.31582480669021606,
+      "learning_rate": 0.008,
+      "loss": 2.5611,
+      "step": 610
+    },
+    {
+      "epoch": 0.06540084388185655,
+      "grad_norm": 0.2663302421569824,
+      "learning_rate": 0.008,
+      "loss": 2.5532,
+      "step": 620
+    },
+    {
+      "epoch": 0.06645569620253164,
+      "grad_norm": 0.28083276748657227,
+      "learning_rate": 0.008,
+      "loss": 2.5378,
+      "step": 630
+    },
+    {
+      "epoch": 0.06751054852320675,
+      "grad_norm": 0.36634561419487,
+      "learning_rate": 0.008,
+      "loss": 2.5487,
+      "step": 640
+    },
+    {
+      "epoch": 0.06856540084388185,
+      "grad_norm": 0.334738552570343,
+      "learning_rate": 0.008,
+      "loss": 2.5385,
+      "step": 650
+    },
+    {
+      "epoch": 0.06962025316455696,
+      "grad_norm": 0.2857688367366791,
+      "learning_rate": 0.008,
+      "loss": 2.5221,
+      "step": 660
+    },
+    {
+      "epoch": 0.07067510548523206,
+      "grad_norm": 0.4055250585079193,
+      "learning_rate": 0.008,
+      "loss": 2.516,
+      "step": 670
+    },
+    {
+      "epoch": 0.07172995780590717,
+      "grad_norm": 0.28684183955192566,
+      "learning_rate": 0.008,
+      "loss": 2.5001,
+      "step": 680
+    },
+    {
+      "epoch": 0.07278481012658228,
+      "grad_norm": 0.20943564176559448,
+      "learning_rate": 0.008,
+      "loss": 2.5083,
+      "step": 690
+    },
+    {
+      "epoch": 0.07383966244725738,
+      "grad_norm": 0.2616605758666992,
+      "learning_rate": 0.008,
+      "loss": 2.4863,
+      "step": 700
+    },
+    {
+      "epoch": 0.07489451476793249,
+      "grad_norm": 0.36410996317863464,
+      "learning_rate": 0.008,
+      "loss": 2.48,
+      "step": 710
+    },
+    {
+      "epoch": 0.0759493670886076,
+      "grad_norm": 0.356417715549469,
+      "learning_rate": 0.008,
+      "loss": 2.4698,
+      "step": 720
+    },
+    {
+      "epoch": 0.0770042194092827,
+      "grad_norm": 0.2846285104751587,
+      "learning_rate": 0.008,
+      "loss": 2.4574,
+      "step": 730
+    },
+    {
+      "epoch": 0.07805907172995781,
+      "grad_norm": 0.2718633711338043,
+      "learning_rate": 0.008,
+      "loss": 2.467,
+      "step": 740
+    },
+    {
+      "epoch": 0.07911392405063292,
+      "grad_norm": 0.3270685076713562,
+      "learning_rate": 0.008,
+      "loss": 2.4409,
+      "step": 750
+    },
+    {
+      "epoch": 0.08016877637130802,
+      "grad_norm": 0.2408626824617386,
+      "learning_rate": 0.008,
+      "loss": 2.4352,
+      "step": 760
+    },
+    {
+      "epoch": 0.08122362869198312,
+      "grad_norm": 0.2724064886569977,
+      "learning_rate": 0.008,
+      "loss": 2.4236,
+      "step": 770
+    },
+    {
+      "epoch": 0.08227848101265822,
+      "grad_norm": 0.6737402081489563,
+      "learning_rate": 0.008,
+      "loss": 2.4306,
+      "step": 780
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 0.25571778416633606,
+      "learning_rate": 0.008,
+      "loss": 2.4461,
+      "step": 790
+    },
+    {
+      "epoch": 0.08438818565400844,
+      "grad_norm": 0.2555127441883087,
+      "learning_rate": 0.008,
+      "loss": 2.4039,
+      "step": 800
+    },
+    {
+      "epoch": 0.08544303797468354,
+      "grad_norm": 0.33505818247795105,
+      "learning_rate": 0.008,
+      "loss": 2.387,
+      "step": 810
+    },
+    {
+      "epoch": 0.08649789029535865,
+      "grad_norm": 0.23954349756240845,
+      "learning_rate": 0.008,
+      "loss": 2.387,
+      "step": 820
+    },
+    {
+      "epoch": 0.08755274261603375,
+      "grad_norm": 0.2589200437068939,
+      "learning_rate": 0.008,
+      "loss": 2.3868,
+      "step": 830
+    },
+    {
+      "epoch": 0.08860759493670886,
+      "grad_norm": 0.23181724548339844,
+      "learning_rate": 0.008,
+      "loss": 2.3673,
+      "step": 840
+    },
+    {
+      "epoch": 0.08966244725738397,
+      "grad_norm": 0.3351551294326782,
+      "learning_rate": 0.008,
+      "loss": 2.3647,
+      "step": 850
+    },
+    {
+      "epoch": 0.09071729957805907,
+      "grad_norm": 0.2807596027851105,
+      "learning_rate": 0.008,
+      "loss": 2.366,
+      "step": 860
+    },
+    {
+      "epoch": 0.09177215189873418,
+      "grad_norm": 0.2669145464897156,
+      "learning_rate": 0.008,
+      "loss": 2.353,
+      "step": 870
+    },
+    {
+      "epoch": 0.09282700421940929,
+      "grad_norm": 0.4093361496925354,
+      "learning_rate": 0.008,
+      "loss": 2.3545,
+      "step": 880
+    },
+    {
+      "epoch": 0.0938818565400844,
+      "grad_norm": 0.23955625295639038,
+      "learning_rate": 0.008,
+      "loss": 2.3497,
+      "step": 890
+    },
+    {
+      "epoch": 0.0949367088607595,
+      "grad_norm": 0.3006705045700073,
+      "learning_rate": 0.008,
+      "loss": 2.332,
+      "step": 900
+    },
+    {
+      "epoch": 0.09599156118143459,
+      "grad_norm": 0.27946627140045166,
+      "learning_rate": 0.008,
+      "loss": 2.3318,
+      "step": 910
+    },
+    {
+      "epoch": 0.0970464135021097,
+      "grad_norm": 0.2559209167957306,
+      "learning_rate": 0.008,
+      "loss": 2.3198,
+      "step": 920
+    },
+    {
+      "epoch": 0.0981012658227848,
+      "grad_norm": 0.33204415440559387,
+      "learning_rate": 0.008,
+      "loss": 2.3028,
+      "step": 930
+    },
+    {
+      "epoch": 0.09915611814345991,
+      "grad_norm": 0.3482770323753357,
+      "learning_rate": 0.008,
+      "loss": 2.3052,
+      "step": 940
+    },
+    {
+      "epoch": 0.10021097046413502,
+      "grad_norm": 0.36272701621055603,
+      "learning_rate": 0.008,
+      "loss": 2.3031,
+      "step": 950
+    },
+    {
+      "epoch": 0.10126582278481013,
+      "grad_norm": 0.2385793775320053,
+      "learning_rate": 0.008,
+      "loss": 2.2961,
+      "step": 960
+    },
+    {
+      "epoch": 0.10232067510548523,
+      "grad_norm": 0.2459821254014969,
+      "learning_rate": 0.008,
+      "loss": 2.2919,
+      "step": 970
+    },
+    {
+      "epoch": 0.10337552742616034,
+      "grad_norm": 0.2878538966178894,
+      "learning_rate": 0.008,
+      "loss": 2.27,
+      "step": 980
+    },
+    {
+      "epoch": 0.10443037974683544,
+      "grad_norm": 0.2998957931995392,
+      "learning_rate": 0.008,
+      "loss": 2.2692,
+      "step": 990
+    },
+    {
+      "epoch": 0.10548523206751055,
+      "grad_norm": 0.26348063349723816,
+      "learning_rate": 0.008,
+      "loss": 2.2755,
+      "step": 1000
+    },
+    {
+      "epoch": 0.10654008438818566,
+      "grad_norm": 0.35050904750823975,
+      "learning_rate": 0.008,
+      "loss": 2.2604,
+      "step": 1010
+    },
+    {
+      "epoch": 0.10759493670886076,
+      "grad_norm": 0.3159315586090088,
+      "learning_rate": 0.008,
+      "loss": 2.2522,
+      "step": 1020
+    },
+    {
+      "epoch": 0.10864978902953587,
+      "grad_norm": 0.30178722739219666,
+      "learning_rate": 0.008,
+      "loss": 2.2491,
+      "step": 1030
+    },
+    {
+      "epoch": 0.10970464135021098,
+      "grad_norm": 0.32155904173851013,
+      "learning_rate": 0.008,
+      "loss": 2.2497,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11075949367088607,
+      "grad_norm": 0.31932875514030457,
+      "learning_rate": 0.008,
+      "loss": 2.2482,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11181434599156118,
+      "grad_norm": 0.3369266390800476,
+      "learning_rate": 0.008,
+      "loss": 2.2355,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11286919831223628,
+      "grad_norm": 0.28845059871673584,
+      "learning_rate": 0.008,
+      "loss": 2.2163,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11392405063291139,
+      "grad_norm": 0.2690791189670563,
+      "learning_rate": 0.008,
+      "loss": 2.2174,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1149789029535865,
+      "grad_norm": 0.24728895723819733,
+      "learning_rate": 0.008,
+      "loss": 2.2181,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1160337552742616,
+      "grad_norm": 0.36755335330963135,
+      "learning_rate": 0.008,
+      "loss": 2.2017,
+      "step": 1100
+    },
+    {
+      "epoch": 0.11708860759493671,
+      "grad_norm": 0.26193735003471375,
+      "learning_rate": 0.008,
+      "loss": 2.2093,
+      "step": 1110
+    },
+    {
+      "epoch": 0.11814345991561181,
+      "grad_norm": 0.320650190114975,
+      "learning_rate": 0.008,
+      "loss": 2.1966,
+      "step": 1120
+    },
+    {
+      "epoch": 0.11919831223628692,
+      "grad_norm": 0.27524930238723755,
+      "learning_rate": 0.008,
+      "loss": 2.1809,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12025316455696203,
+      "grad_norm": 0.24788086116313934,
+      "learning_rate": 0.008,
+      "loss": 2.1912,
+      "step": 1140
+    },
+    {
+      "epoch": 0.12130801687763713,
+      "grad_norm": 0.28242743015289307,
+      "learning_rate": 0.008,
+      "loss": 2.1737,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12236286919831224,
+      "grad_norm": 0.22859254479408264,
+      "learning_rate": 0.008,
+      "loss": 2.1866,
+      "step": 1160
+    },
+    {
+      "epoch": 0.12341772151898735,
+      "grad_norm": 0.2717379033565521,
+      "learning_rate": 0.008,
+      "loss": 2.1609,
+      "step": 1170
+    },
+    {
+      "epoch": 0.12447257383966245,
+      "grad_norm": 0.29991039633750916,
+      "learning_rate": 0.008,
+      "loss": 2.1565,
+      "step": 1180
+    },
+    {
+      "epoch": 0.12552742616033755,
+      "grad_norm": 0.3012315630912781,
+      "learning_rate": 0.008,
+      "loss": 2.1542,
+      "step": 1190
+    },
+    {
+      "epoch": 0.12658227848101267,
+      "grad_norm": 0.24644635617733002,
+      "learning_rate": 0.008,
+      "loss": 2.1593,
+      "step": 1200
+    },
+    {
+      "epoch": 0.12763713080168776,
+      "grad_norm": 0.28635427355766296,
+      "learning_rate": 0.008,
+      "loss": 2.147,
+      "step": 1210
+    },
+    {
+      "epoch": 0.12869198312236288,
+      "grad_norm": 0.29788339138031006,
+      "learning_rate": 0.008,
+      "loss": 2.1522,
+      "step": 1220
+    },
+    {
+      "epoch": 0.12974683544303797,
+      "grad_norm": 0.23508018255233765,
+      "learning_rate": 0.008,
+      "loss": 2.127,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1308016877637131,
+      "grad_norm": 0.30958813428878784,
+      "learning_rate": 0.008,
+      "loss": 2.1396,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13185654008438819,
+      "grad_norm": 0.2530100345611572,
+      "learning_rate": 0.008,
+      "loss": 2.1361,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13291139240506328,
+      "grad_norm": 0.4224642813205719,
+      "learning_rate": 0.008,
+      "loss": 2.1156,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1339662447257384,
+      "grad_norm": 0.3071783781051636,
+      "learning_rate": 0.008,
+      "loss": 2.1145,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1350210970464135,
+      "grad_norm": 0.3265485465526581,
+      "learning_rate": 0.008,
+      "loss": 2.1219,
+      "step": 1280
+    },
+    {
+      "epoch": 0.1360759493670886,
+      "grad_norm": 0.3116600215435028,
+      "learning_rate": 0.008,
+      "loss": 2.1049,
+      "step": 1290
+    },
+    {
+      "epoch": 0.1371308016877637,
+      "grad_norm": 0.3277093768119812,
+      "learning_rate": 0.008,
+      "loss": 2.1052,
+      "step": 1300
+    },
+    {
+      "epoch": 0.13818565400843882,
+      "grad_norm": 0.34968429803848267,
+      "learning_rate": 0.008,
+      "loss": 2.0965,
+      "step": 1310
+    },
+    {
+      "epoch": 0.13924050632911392,
+      "grad_norm": 0.2548694312572479,
+      "learning_rate": 0.008,
+      "loss": 2.0923,
+      "step": 1320
+    },
+    {
+      "epoch": 0.14029535864978904,
+      "grad_norm": 0.30090993642807007,
+      "learning_rate": 0.008,
+      "loss": 2.0805,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14135021097046413,
+      "grad_norm": 0.29015231132507324,
+      "learning_rate": 0.008,
+      "loss": 2.0878,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14240506329113925,
+      "grad_norm": 0.3280774652957916,
+      "learning_rate": 0.008,
+      "loss": 2.0986,
+      "step": 1350
+    },
+    {
+      "epoch": 0.14345991561181434,
+      "grad_norm": 0.3089797794818878,
+      "learning_rate": 0.008,
+      "loss": 2.0734,
+      "step": 1360
+    },
+    {
+      "epoch": 0.14451476793248946,
+      "grad_norm": 0.32852697372436523,
+      "learning_rate": 0.008,
+      "loss": 2.0688,
+      "step": 1370
+    },
+    {
+      "epoch": 0.14556962025316456,
+      "grad_norm": 0.25315821170806885,
+      "learning_rate": 0.008,
+      "loss": 2.0522,
+      "step": 1380
+    },
+    {
+      "epoch": 0.14662447257383968,
+      "grad_norm": 0.27433422207832336,
+      "learning_rate": 0.008,
+      "loss": 2.063,
+      "step": 1390
+    },
+    {
+      "epoch": 0.14767932489451477,
+      "grad_norm": 0.24619846045970917,
+      "learning_rate": 0.008,
+      "loss": 2.0488,
+      "step": 1400
+    },
+    {
+      "epoch": 0.14873417721518986,
+      "grad_norm": 0.2722327411174774,
+      "learning_rate": 0.008,
+      "loss": 2.0508,
+      "step": 1410
+    },
+    {
+      "epoch": 0.14978902953586498,
+      "grad_norm": 0.2827919125556946,
+      "learning_rate": 0.008,
+      "loss": 2.0535,
+      "step": 1420
+    },
+    {
+      "epoch": 0.15084388185654007,
+      "grad_norm": 0.26990702748298645,
+      "learning_rate": 0.008,
+      "loss": 2.043,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1518987341772152,
+      "grad_norm": 0.24337026476860046,
+      "learning_rate": 0.008,
+      "loss": 2.0411,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1529535864978903,
+      "grad_norm": 0.21814419329166412,
+      "learning_rate": 0.008,
+      "loss": 2.0332,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1540084388185654,
+      "grad_norm": 0.24107573926448822,
+      "learning_rate": 0.008,
+      "loss": 2.0209,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1550632911392405,
+      "grad_norm": 0.24475321173667908,
+      "learning_rate": 0.008,
+      "loss": 2.0147,
+      "step": 1470
+    },
+    {
+      "epoch": 0.15611814345991562,
+      "grad_norm": 0.252210408449173,
+      "learning_rate": 0.008,
+      "loss": 2.0194,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1571729957805907,
+      "grad_norm": 0.23219706118106842,
+      "learning_rate": 0.008,
+      "loss": 2.0198,
+      "step": 1490
+    },
+    {
+      "epoch": 0.15822784810126583,
+      "grad_norm": 0.2675815522670746,
+      "learning_rate": 0.008,
+      "loss": 2.024,
+      "step": 1500
+    },
+    {
+      "epoch": 0.15928270042194093,
+      "grad_norm": 0.264658123254776,
+      "learning_rate": 0.008,
+      "loss": 2.0154,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16033755274261605,
+      "grad_norm": 0.33138853311538696,
+      "learning_rate": 0.008,
+      "loss": 2.0114,
+      "step": 1520
+    },
+    {
+      "epoch": 0.16139240506329114,
+      "grad_norm": 0.22661495208740234,
+      "learning_rate": 0.008,
+      "loss": 1.9994,
+      "step": 1530
+    },
+    {
+      "epoch": 0.16244725738396623,
+      "grad_norm": 0.2567743957042694,
+      "learning_rate": 0.008,
+      "loss": 1.9887,
+      "step": 1540
+    },
+    {
+      "epoch": 0.16350210970464135,
+      "grad_norm": 0.23519980907440186,
+      "learning_rate": 0.008,
+      "loss": 1.9842,
+      "step": 1550
+    },
+    {
+      "epoch": 0.16455696202531644,
+      "grad_norm": 0.2918664813041687,
+      "learning_rate": 0.008,
+      "loss": 2.0032,
+      "step": 1560
+    },
+    {
+      "epoch": 0.16561181434599156,
+      "grad_norm": 0.2507562041282654,
+      "learning_rate": 0.008,
+      "loss": 1.9814,
+      "step": 1570
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 0.28064262866973877,
+      "learning_rate": 0.008,
+      "loss": 1.9874,
+      "step": 1580
+    },
+    {
+      "epoch": 0.16772151898734178,
+      "grad_norm": 0.31995588541030884,
+      "learning_rate": 0.008,
+      "loss": 1.9836,
+      "step": 1590
+    },
+    {
+      "epoch": 0.16877637130801687,
+      "grad_norm": 0.2820163071155548,
+      "learning_rate": 0.008,
+      "loss": 1.9792,
+      "step": 1600
+    },
+    {
+      "epoch": 0.169831223628692,
+      "grad_norm": 0.240266352891922,
+      "learning_rate": 0.008,
+      "loss": 1.9723,
+      "step": 1610
+    },
+    {
+      "epoch": 0.17088607594936708,
+      "grad_norm": 0.26592355966567993,
+      "learning_rate": 0.008,
+      "loss": 1.9446,
+      "step": 1620
+    },
+    {
+      "epoch": 0.1719409282700422,
+      "grad_norm": 0.2303941547870636,
+      "learning_rate": 0.008,
+      "loss": 1.9544,
+      "step": 1630
+    },
+    {
+      "epoch": 0.1729957805907173,
+      "grad_norm": 0.3694598972797394,
+      "learning_rate": 0.008,
+      "loss": 1.9553,
+      "step": 1640
+    },
+    {
+      "epoch": 0.17405063291139242,
+      "grad_norm": 0.35454174876213074,
+      "learning_rate": 0.008,
+      "loss": 1.948,
+      "step": 1650
+    },
+    {
+      "epoch": 0.1751054852320675,
+      "grad_norm": 0.2986469566822052,
+      "learning_rate": 0.008,
+      "loss": 1.9438,
+      "step": 1660
+    },
+    {
+      "epoch": 0.17616033755274263,
+      "grad_norm": 0.2804855406284332,
+      "learning_rate": 0.008,
+      "loss": 1.9517,
+      "step": 1670
+    },
+    {
+      "epoch": 0.17721518987341772,
+      "grad_norm": 0.4110938012599945,
+      "learning_rate": 0.008,
+      "loss": 1.9411,
+      "step": 1680
+    },
+    {
+      "epoch": 0.17827004219409281,
+      "grad_norm": 0.3235545754432678,
+      "learning_rate": 0.008,
+      "loss": 1.9329,
+      "step": 1690
+    },
+    {
+      "epoch": 0.17932489451476794,
+      "grad_norm": 0.24863795936107635,
+      "learning_rate": 0.008,
+      "loss": 1.9401,
+      "step": 1700
+    },
+    {
+      "epoch": 0.18037974683544303,
+      "grad_norm": 0.3035620152950287,
+      "learning_rate": 0.008,
+      "loss": 1.9207,
+      "step": 1710
+    },
+    {
+      "epoch": 0.18143459915611815,
+      "grad_norm": 0.45260244607925415,
+      "learning_rate": 0.008,
+      "loss": 1.9154,
+      "step": 1720
+    },
+    {
+      "epoch": 0.18248945147679324,
+      "grad_norm": 0.24632956087589264,
+      "learning_rate": 0.008,
+      "loss": 1.9182,
+      "step": 1730
+    },
+    {
+      "epoch": 0.18354430379746836,
+      "grad_norm": 0.3995058536529541,
+      "learning_rate": 0.008,
+      "loss": 1.9202,
+      "step": 1740
+    },
+    {
+      "epoch": 0.18459915611814345,
+      "grad_norm": 0.3234435021877289,
+      "learning_rate": 0.008,
+      "loss": 1.9468,
+      "step": 1750
+    },
+    {
+      "epoch": 0.18565400843881857,
+      "grad_norm": 0.27081844210624695,
+      "learning_rate": 0.008,
+      "loss": 1.9195,
+      "step": 1760
+    },
+    {
+      "epoch": 0.18670886075949367,
+      "grad_norm": 0.27147287130355835,
+      "learning_rate": 0.008,
+      "loss": 1.9111,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1877637130801688,
+      "grad_norm": 0.27063852548599243,
+      "learning_rate": 0.008,
+      "loss": 1.9108,
+      "step": 1780
+    },
+    {
+      "epoch": 0.18881856540084388,
+      "grad_norm": 0.3690478801727295,
+      "learning_rate": 0.008,
+      "loss": 1.909,
+      "step": 1790
+    },
+    {
+      "epoch": 0.189873417721519,
+      "grad_norm": 0.31646183133125305,
+      "learning_rate": 0.008,
+      "loss": 1.9065,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1909282700421941,
+      "grad_norm": 0.36002248525619507,
+      "learning_rate": 0.008,
+      "loss": 1.8966,
+      "step": 1810
+    },
+    {
+      "epoch": 0.19198312236286919,
+      "grad_norm": 0.37479761242866516,
+      "learning_rate": 0.008,
+      "loss": 1.8947,
+      "step": 1820
+    },
+    {
+      "epoch": 0.1930379746835443,
+      "grad_norm": 0.2350432425737381,
+      "learning_rate": 0.008,
+      "loss": 1.8899,
+      "step": 1830
+    },
+    {
+      "epoch": 0.1940928270042194,
+      "grad_norm": 0.30547022819519043,
+      "learning_rate": 0.008,
+      "loss": 1.8821,
+      "step": 1840
+    },
+    {
+      "epoch": 0.19514767932489452,
+      "grad_norm": 0.30677708983421326,
+      "learning_rate": 0.008,
+      "loss": 1.8876,
+      "step": 1850
+    },
+    {
+      "epoch": 0.1962025316455696,
+      "grad_norm": 0.3187776803970337,
+      "learning_rate": 0.008,
+      "loss": 1.8946,
+      "step": 1860
+    },
+    {
+      "epoch": 0.19725738396624473,
+      "grad_norm": 0.2627311050891876,
+      "learning_rate": 0.008,
+      "loss": 1.8897,
+      "step": 1870
+    },
+    {
+      "epoch": 0.19831223628691982,
+      "grad_norm": 0.24966266751289368,
+      "learning_rate": 0.008,
+      "loss": 1.8737,
+      "step": 1880
+    },
+    {
+      "epoch": 0.19936708860759494,
+      "grad_norm": 0.43601298332214355,
+      "learning_rate": 0.008,
+      "loss": 1.8746,
+      "step": 1890
+    },
+    {
+      "epoch": 0.20042194092827004,
+      "grad_norm": 0.25831761956214905,
+      "learning_rate": 0.008,
+      "loss": 1.8731,
+      "step": 1900
+    },
+    {
+      "epoch": 0.20147679324894516,
+      "grad_norm": 0.23979012668132782,
+      "learning_rate": 0.008,
+      "loss": 1.8734,
+      "step": 1910
+    },
+    {
+      "epoch": 0.20253164556962025,
+      "grad_norm": 0.22324031591415405,
+      "learning_rate": 0.008,
+      "loss": 1.8663,
+      "step": 1920
+    },
+    {
+      "epoch": 0.20358649789029537,
+      "grad_norm": 0.3088960349559784,
+      "learning_rate": 0.008,
+      "loss": 1.8696,
+      "step": 1930
+    },
+    {
+      "epoch": 0.20464135021097046,
+      "grad_norm": 0.32011809945106506,
+      "learning_rate": 0.008,
+      "loss": 1.8525,
+      "step": 1940
+    },
+    {
+      "epoch": 0.20569620253164558,
+      "grad_norm": 0.25118955969810486,
+      "learning_rate": 0.008,
+      "loss": 1.8607,
+      "step": 1950
+    },
+    {
+      "epoch": 0.20675105485232068,
+      "grad_norm": 0.32835960388183594,
+      "learning_rate": 0.008,
+      "loss": 1.8569,
+      "step": 1960
+    },
+    {
+      "epoch": 0.20780590717299577,
+      "grad_norm": 0.23883293569087982,
+      "learning_rate": 0.008,
+      "loss": 1.8654,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2088607594936709,
+      "grad_norm": 0.24800105392932892,
+      "learning_rate": 0.008,
+      "loss": 1.8524,
+      "step": 1980
+    },
+    {
+      "epoch": 0.20991561181434598,
+      "grad_norm": 0.23113876581192017,
+      "learning_rate": 0.008,
+      "loss": 1.846,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2109704641350211,
+      "grad_norm": 0.24381081759929657,
+      "learning_rate": 0.008,
+      "loss": 1.8401,
+      "step": 2000
+    },
+    {
+      "epoch": 0.2120253164556962,
+      "grad_norm": 0.23765400052070618,
+      "learning_rate": 0.008,
+      "loss": 1.8467,
+      "step": 2010
+    },
+    {
+      "epoch": 0.21308016877637131,
+      "grad_norm": 0.24969510734081268,
+      "learning_rate": 0.008,
+      "loss": 1.8536,
+      "step": 2020
+    },
+    {
+      "epoch": 0.2141350210970464,
+      "grad_norm": 0.3504819869995117,
+      "learning_rate": 0.008,
+      "loss": 1.8503,
+      "step": 2030
+    },
+    {
+      "epoch": 0.21518987341772153,
+      "grad_norm": 0.2780002951622009,
+      "learning_rate": 0.008,
+      "loss": 1.8398,
+      "step": 2040
+    },
+    {
+      "epoch": 0.21624472573839662,
+      "grad_norm": 0.3090433180332184,
+      "learning_rate": 0.008,
+      "loss": 1.8367,
+      "step": 2050
+    },
+    {
+      "epoch": 0.21729957805907174,
+      "grad_norm": 0.2590400278568268,
+      "learning_rate": 0.008,
+      "loss": 1.8248,
+      "step": 2060
+    },
+    {
+      "epoch": 0.21835443037974683,
+      "grad_norm": 0.2706325948238373,
+      "learning_rate": 0.008,
+      "loss": 1.8236,
+      "step": 2070
+    },
+    {
+      "epoch": 0.21940928270042195,
+      "grad_norm": 0.25876572728157043,
+      "learning_rate": 0.008,
+      "loss": 1.8269,
+      "step": 2080
+    },
+    {
+      "epoch": 0.22046413502109705,
+      "grad_norm": 0.3149887025356293,
+      "learning_rate": 0.008,
+      "loss": 1.8211,
+      "step": 2090
+    },
+    {
+      "epoch": 0.22151898734177214,
+      "grad_norm": 0.278936505317688,
+      "learning_rate": 0.008,
+      "loss": 1.8234,
+      "step": 2100
+    },
+    {
+      "epoch": 0.22257383966244726,
+      "grad_norm": 0.2407190054655075,
+      "learning_rate": 0.008,
+      "loss": 1.8141,
+      "step": 2110
+    },
+    {
+      "epoch": 0.22362869198312235,
+      "grad_norm": 0.2713059186935425,
+      "learning_rate": 0.008,
+      "loss": 1.808,
+      "step": 2120
+    },
+    {
+      "epoch": 0.22468354430379747,
+      "grad_norm": 0.45755669474601746,
+      "learning_rate": 0.008,
+      "loss": 1.8192,
+      "step": 2130
+    },
+    {
+      "epoch": 0.22573839662447256,
+      "grad_norm": 0.4164162874221802,
+      "learning_rate": 0.008,
+      "loss": 1.8105,
+      "step": 2140
+    },
+    {
+      "epoch": 0.22679324894514769,
+      "grad_norm": 0.24771223962306976,
+      "learning_rate": 0.008,
+      "loss": 1.8152,
+      "step": 2150
+    },
+    {
+      "epoch": 0.22784810126582278,
+      "grad_norm": 0.22788429260253906,
+      "learning_rate": 0.008,
+      "loss": 1.8239,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2289029535864979,
+      "grad_norm": 0.4012809097766876,
+      "learning_rate": 0.008,
+      "loss": 1.8012,
+      "step": 2170
+    },
+    {
+      "epoch": 0.229957805907173,
+      "grad_norm": 0.2694602608680725,
+      "learning_rate": 0.008,
+      "loss": 1.7895,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2310126582278481,
+      "grad_norm": 0.2758353054523468,
+      "learning_rate": 0.008,
+      "loss": 1.8058,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2320675105485232,
+      "grad_norm": 0.30435648560523987,
+      "learning_rate": 0.008,
+      "loss": 1.8072,
+      "step": 2200
+    },
+    {
+      "epoch": 0.23312236286919832,
+      "grad_norm": 0.23588992655277252,
+      "learning_rate": 0.008,
+      "loss": 1.7904,
+      "step": 2210
+    },
+    {
+      "epoch": 0.23417721518987342,
+      "grad_norm": 0.24185898900032043,
+      "learning_rate": 0.008,
+      "loss": 1.7984,
+      "step": 2220
+    },
+    {
+      "epoch": 0.23523206751054854,
+      "grad_norm": 0.3155139088630676,
+      "learning_rate": 0.008,
+      "loss": 1.7944,
+      "step": 2230
+    },
+    {
+      "epoch": 0.23628691983122363,
+      "grad_norm": 0.3639398515224457,
+      "learning_rate": 0.008,
+      "loss": 1.7886,
+      "step": 2240
+    },
+    {
+      "epoch": 0.23734177215189872,
+      "grad_norm": 0.2653811275959015,
+      "learning_rate": 0.008,
+      "loss": 1.7841,
+      "step": 2250
+    },
+    {
+      "epoch": 0.23839662447257384,
+      "grad_norm": 0.25944286584854126,
+      "learning_rate": 0.008,
+      "loss": 1.7753,
+      "step": 2260
+    },
+    {
+      "epoch": 0.23945147679324894,
+      "grad_norm": 0.2378905713558197,
+      "learning_rate": 0.008,
+      "loss": 1.8083,
+      "step": 2270
+    },
+    {
+      "epoch": 0.24050632911392406,
+      "grad_norm": 0.34498322010040283,
+      "learning_rate": 0.008,
+      "loss": 1.7862,
+      "step": 2280
+    },
+    {
+      "epoch": 0.24156118143459915,
+      "grad_norm": 0.24700169265270233,
+      "learning_rate": 0.008,
+      "loss": 1.772,
+      "step": 2290
+    },
+    {
+      "epoch": 0.24261603375527427,
+      "grad_norm": 0.24769878387451172,
+      "learning_rate": 0.008,
+      "loss": 1.7708,
+      "step": 2300
+    },
+    {
+      "epoch": 0.24367088607594936,
+      "grad_norm": 0.42960017919540405,
+      "learning_rate": 0.008,
+      "loss": 1.7771,
+      "step": 2310
+    },
+    {
+      "epoch": 0.24472573839662448,
+      "grad_norm": 0.25843679904937744,
+      "learning_rate": 0.008,
+      "loss": 1.7818,
+      "step": 2320
+    },
+    {
+      "epoch": 0.24578059071729957,
+      "grad_norm": 0.25137412548065186,
+      "learning_rate": 0.008,
+      "loss": 1.7602,
+      "step": 2330
+    },
+    {
+      "epoch": 0.2468354430379747,
+      "grad_norm": 0.37706929445266724,
+      "learning_rate": 0.008,
+      "loss": 1.7616,
+      "step": 2340
+    },
+    {
+      "epoch": 0.2478902953586498,
+      "grad_norm": 0.26846984028816223,
+      "learning_rate": 0.008,
+      "loss": 1.7593,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2489451476793249,
+      "grad_norm": 0.2676597237586975,
+      "learning_rate": 0.008,
+      "loss": 1.767,
+      "step": 2360
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.3121493458747864,
+      "learning_rate": 0.008,
+      "loss": 1.764,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2510548523206751,
+      "grad_norm": 0.22265689074993134,
+      "learning_rate": 0.008,
+      "loss": 1.7591,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2521097046413502,
+      "grad_norm": 0.268729031085968,
+      "learning_rate": 0.008,
+      "loss": 1.7581,
+      "step": 2390
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 0.4030081033706665,
+      "learning_rate": 0.008,
+      "loss": 1.7607,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2542194092827004,
+      "grad_norm": 0.30595529079437256,
+      "learning_rate": 0.008,
+      "loss": 1.7625,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2552742616033755,
+      "grad_norm": 0.2637060880661011,
+      "learning_rate": 0.008,
+      "loss": 1.7532,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2563291139240506,
+      "grad_norm": 0.22066053748130798,
+      "learning_rate": 0.008,
+      "loss": 1.7477,
+      "step": 2430
+    },
+    {
+      "epoch": 0.25738396624472576,
+      "grad_norm": 0.2550792098045349,
+      "learning_rate": 0.008,
+      "loss": 1.7577,
+      "step": 2440
+    },
+    {
+      "epoch": 0.25843881856540085,
+      "grad_norm": 0.33186402916908264,
+      "learning_rate": 0.008,
+      "loss": 1.748,
+      "step": 2450
+    },
+    {
+      "epoch": 0.25949367088607594,
+      "grad_norm": 0.27387940883636475,
+      "learning_rate": 0.008,
+      "loss": 1.7636,
+      "step": 2460
+    },
+    {
+      "epoch": 0.26054852320675104,
+      "grad_norm": 0.24876436591148376,
+      "learning_rate": 0.008,
+      "loss": 1.7517,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2616033755274262,
+      "grad_norm": 0.26100295782089233,
+      "learning_rate": 0.008,
+      "loss": 1.7482,
+      "step": 2480
+    },
+    {
+      "epoch": 0.2626582278481013,
+      "grad_norm": 0.2517220079898834,
+      "learning_rate": 0.008,
+      "loss": 1.7383,
+      "step": 2490
+    },
+    {
+      "epoch": 0.26371308016877637,
+      "grad_norm": 0.28499963879585266,
+      "learning_rate": 0.008,
+      "loss": 1.745,
+      "step": 2500
+    },
+    {
+      "epoch": 0.26476793248945146,
+      "grad_norm": 0.23393799364566803,
+      "learning_rate": 0.008,
+      "loss": 1.7366,
+      "step": 2510
+    },
+    {
+      "epoch": 0.26582278481012656,
+      "grad_norm": 0.26385918259620667,
+      "learning_rate": 0.008,
+      "loss": 1.7394,
+      "step": 2520
+    },
+    {
+      "epoch": 0.2668776371308017,
+      "grad_norm": 0.26664602756500244,
+      "learning_rate": 0.008,
+      "loss": 1.732,
+      "step": 2530
+    },
+    {
+      "epoch": 0.2679324894514768,
+      "grad_norm": 0.3325648009777069,
+      "learning_rate": 0.008,
+      "loss": 1.7351,
+      "step": 2540
+    },
+    {
+      "epoch": 0.2689873417721519,
+      "grad_norm": 0.3920401930809021,
+      "learning_rate": 0.008,
+      "loss": 1.7391,
+      "step": 2550
+    },
+    {
+      "epoch": 0.270042194092827,
+      "grad_norm": 0.2633798122406006,
+      "learning_rate": 0.008,
+      "loss": 1.7423,
+      "step": 2560
+    },
+    {
+      "epoch": 0.27109704641350213,
+      "grad_norm": 0.22189389169216156,
+      "learning_rate": 0.008,
+      "loss": 1.7254,
+      "step": 2570
+    },
+    {
+      "epoch": 0.2721518987341772,
+      "grad_norm": 0.27132439613342285,
+      "learning_rate": 0.008,
+      "loss": 1.7273,
+      "step": 2580
+    },
+    {
+      "epoch": 0.2732067510548523,
+      "grad_norm": 0.26365676522254944,
+      "learning_rate": 0.008,
+      "loss": 1.734,
+      "step": 2590
+    },
+    {
+      "epoch": 0.2742616033755274,
+      "grad_norm": 0.2338821291923523,
+      "learning_rate": 0.008,
+      "loss": 1.7324,
+      "step": 2600
+    },
+    {
+      "epoch": 0.27531645569620256,
+      "grad_norm": 0.24077825248241425,
+      "learning_rate": 0.008,
+      "loss": 1.7322,
+      "step": 2610
+    },
+    {
+      "epoch": 0.27637130801687765,
+      "grad_norm": 0.26235485076904297,
+      "learning_rate": 0.008,
+      "loss": 1.7271,
+      "step": 2620
+    },
+    {
+      "epoch": 0.27742616033755274,
+      "grad_norm": 0.25538384914398193,
+      "learning_rate": 0.008,
+      "loss": 1.7121,
+      "step": 2630
+    },
+    {
+      "epoch": 0.27848101265822783,
+      "grad_norm": 0.5633853673934937,
+      "learning_rate": 0.008,
+      "loss": 1.7269,
+      "step": 2640
+    },
+    {
+      "epoch": 0.2795358649789029,
+      "grad_norm": 0.3092365860939026,
+      "learning_rate": 0.008,
+      "loss": 1.7184,
+      "step": 2650
+    },
+    {
+      "epoch": 0.2805907172995781,
+      "grad_norm": 0.24975547194480896,
+      "learning_rate": 0.008,
+      "loss": 1.7104,
+      "step": 2660
+    },
+    {
+      "epoch": 0.28164556962025317,
+      "grad_norm": 0.30870065093040466,
+      "learning_rate": 0.008,
+      "loss": 1.7104,
+      "step": 2670
+    },
+    {
+      "epoch": 0.28270042194092826,
+      "grad_norm": 0.2730621099472046,
+      "learning_rate": 0.008,
+      "loss": 1.7132,
+      "step": 2680
+    },
+    {
+      "epoch": 0.28375527426160335,
+      "grad_norm": 0.33448681235313416,
+      "learning_rate": 0.008,
+      "loss": 1.7102,
+      "step": 2690
+    },
+    {
+      "epoch": 0.2848101265822785,
+      "grad_norm": 0.26378071308135986,
+      "learning_rate": 0.008,
+      "loss": 1.7082,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2858649789029536,
+      "grad_norm": 0.4238617420196533,
+      "learning_rate": 0.008,
+      "loss": 1.7243,
+      "step": 2710
+    },
+    {
+      "epoch": 0.2869198312236287,
+      "grad_norm": 0.24682502448558807,
+      "learning_rate": 0.008,
+      "loss": 1.7204,
+      "step": 2720
+    },
+    {
+      "epoch": 0.2879746835443038,
+      "grad_norm": 0.36426928639411926,
+      "learning_rate": 0.008,
+      "loss": 1.6901,
+      "step": 2730
+    },
+    {
+      "epoch": 0.2890295358649789,
+      "grad_norm": 0.3199904263019562,
+      "learning_rate": 0.008,
+      "loss": 1.7031,
+      "step": 2740
+    },
+    {
+      "epoch": 0.290084388185654,
+      "grad_norm": 0.23249582946300507,
+      "learning_rate": 0.008,
+      "loss": 1.7124,
+      "step": 2750
+    },
+    {
+      "epoch": 0.2911392405063291,
+      "grad_norm": 0.38822662830352783,
+      "learning_rate": 0.008,
+      "loss": 1.7124,
+      "step": 2760
+    },
+    {
+      "epoch": 0.2921940928270042,
+      "grad_norm": 0.24358908832073212,
+      "learning_rate": 0.008,
+      "loss": 1.7078,
+      "step": 2770
+    },
+    {
+      "epoch": 0.29324894514767935,
+      "grad_norm": 0.32432615756988525,
+      "learning_rate": 0.008,
+      "loss": 1.7099,
+      "step": 2780
+    },
+    {
+      "epoch": 0.29430379746835444,
+      "grad_norm": 0.40301641821861267,
+      "learning_rate": 0.008,
+      "loss": 1.7064,
+      "step": 2790
+    },
+    {
+      "epoch": 0.29535864978902954,
+      "grad_norm": 0.26763567328453064,
+      "learning_rate": 0.008,
+      "loss": 1.6968,
+      "step": 2800
+    },
+    {
+      "epoch": 0.29641350210970463,
+      "grad_norm": 0.2115464210510254,
+      "learning_rate": 0.008,
+      "loss": 1.6976,
+      "step": 2810
+    },
+    {
+      "epoch": 0.2974683544303797,
+      "grad_norm": 0.34389564394950867,
+      "learning_rate": 0.008,
+      "loss": 1.6985,
+      "step": 2820
+    },
+    {
+      "epoch": 0.29852320675105487,
+      "grad_norm": 0.3676552474498749,
+      "learning_rate": 0.008,
+      "loss": 1.6756,
+      "step": 2830
+    },
+    {
+      "epoch": 0.29957805907172996,
+      "grad_norm": 0.28342434763908386,
+      "learning_rate": 0.008,
+      "loss": 1.6796,
+      "step": 2840
+    },
+    {
+      "epoch": 0.30063291139240506,
+      "grad_norm": 0.26045167446136475,
+      "learning_rate": 0.008,
+      "loss": 1.6865,
+      "step": 2850
+    },
+    {
+      "epoch": 0.30168776371308015,
+      "grad_norm": 0.2579497694969177,
+      "learning_rate": 0.008,
+      "loss": 1.6886,
+      "step": 2860
+    },
+    {
+      "epoch": 0.3027426160337553,
+      "grad_norm": 0.24340470135211945,
+      "learning_rate": 0.008,
+      "loss": 1.6863,
+      "step": 2870
+    },
+    {
+      "epoch": 0.3037974683544304,
+      "grad_norm": 0.26090118288993835,
+      "learning_rate": 0.008,
+      "loss": 1.6786,
+      "step": 2880
+    },
+    {
+      "epoch": 0.3048523206751055,
+      "grad_norm": 0.2293378710746765,
+      "learning_rate": 0.008,
+      "loss": 1.6801,
+      "step": 2890
+    },
+    {
+      "epoch": 0.3059071729957806,
+      "grad_norm": 0.23128661513328552,
+      "learning_rate": 0.008,
+      "loss": 1.6834,
+      "step": 2900
+    },
+    {
+      "epoch": 0.3069620253164557,
+      "grad_norm": 0.3109765946865082,
+      "learning_rate": 0.008,
+      "loss": 1.695,
+      "step": 2910
+    },
+    {
+      "epoch": 0.3080168776371308,
+      "grad_norm": 0.2679274082183838,
+      "learning_rate": 0.008,
+      "loss": 1.6828,
+      "step": 2920
+    },
+    {
+      "epoch": 0.3090717299578059,
+      "grad_norm": 0.20003528892993927,
+      "learning_rate": 0.008,
+      "loss": 1.6707,
+      "step": 2930
+    },
+    {
+      "epoch": 0.310126582278481,
+      "grad_norm": 0.28561410307884216,
+      "learning_rate": 0.008,
+      "loss": 1.6803,
+      "step": 2940
+    },
+    {
+      "epoch": 0.3111814345991561,
+      "grad_norm": 0.3036288022994995,
+      "learning_rate": 0.008,
+      "loss": 1.6878,
+      "step": 2950
+    },
+    {
+      "epoch": 0.31223628691983124,
+      "grad_norm": 0.40575164556503296,
+      "learning_rate": 0.008,
+      "loss": 1.675,
+      "step": 2960
+    },
+    {
+      "epoch": 0.31329113924050633,
+      "grad_norm": 0.22165536880493164,
+      "learning_rate": 0.008,
+      "loss": 1.666,
+      "step": 2970
+    },
+    {
+      "epoch": 0.3143459915611814,
+      "grad_norm": 0.2893832325935364,
+      "learning_rate": 0.008,
+      "loss": 1.6721,
+      "step": 2980
+    },
+    {
+      "epoch": 0.3154008438818565,
+      "grad_norm": 0.29158350825309753,
+      "learning_rate": 0.008,
+      "loss": 1.6935,
+      "step": 2990
+    },
+    {
+      "epoch": 0.31645569620253167,
+      "grad_norm": 0.29934486746788025,
+      "learning_rate": 0.008,
+      "loss": 1.6619,
+      "step": 3000
+    },
+    {
+      "epoch": 0.31751054852320676,
+      "grad_norm": 0.3677426874637604,
+      "learning_rate": 0.008,
+      "loss": 1.669,
+      "step": 3010
+    },
+    {
+      "epoch": 0.31856540084388185,
+      "grad_norm": 0.5185812711715698,
+      "learning_rate": 0.008,
+      "loss": 1.6693,
+      "step": 3020
+    },
+    {
+      "epoch": 0.31962025316455694,
+      "grad_norm": 0.36377614736557007,
+      "learning_rate": 0.008,
+      "loss": 1.6659,
+      "step": 3030
+    },
+    {
+      "epoch": 0.3206751054852321,
+      "grad_norm": 0.24232980608940125,
+      "learning_rate": 0.008,
+      "loss": 1.6652,
+      "step": 3040
+    },
+    {
+      "epoch": 0.3217299578059072,
+      "grad_norm": 0.2422371804714203,
+      "learning_rate": 0.008,
+      "loss": 1.6593,
+      "step": 3050
+    },
+    {
+      "epoch": 0.3227848101265823,
+      "grad_norm": 0.29716598987579346,
+      "learning_rate": 0.008,
+      "loss": 1.6705,
+      "step": 3060
+    },
+    {
+      "epoch": 0.32383966244725737,
+      "grad_norm": 0.22628414630889893,
+      "learning_rate": 0.008,
+      "loss": 1.6608,
+      "step": 3070
+    },
+    {
+      "epoch": 0.32489451476793246,
+      "grad_norm": 0.39875736832618713,
+      "learning_rate": 0.008,
+      "loss": 1.6485,
+      "step": 3080
+    },
+    {
+      "epoch": 0.3259493670886076,
+      "grad_norm": 0.23559480905532837,
+      "learning_rate": 0.008,
+      "loss": 1.6694,
+      "step": 3090
+    },
+    {
+      "epoch": 0.3270042194092827,
+      "grad_norm": 0.24464690685272217,
+      "learning_rate": 0.008,
+      "loss": 1.6494,
+      "step": 3100
+    },
+    {
+      "epoch": 0.3280590717299578,
+      "grad_norm": 0.21670368313789368,
+      "learning_rate": 0.008,
+      "loss": 1.6529,
+      "step": 3110
+    },
+    {
+      "epoch": 0.3291139240506329,
+      "grad_norm": 0.29836854338645935,
+      "learning_rate": 0.008,
+      "loss": 1.6574,
+      "step": 3120
+    },
+    {
+      "epoch": 0.33016877637130804,
+      "grad_norm": 0.24161586165428162,
+      "learning_rate": 0.008,
+      "loss": 1.6717,
+      "step": 3130
+    },
+    {
+      "epoch": 0.33122362869198313,
+      "grad_norm": 0.3704703152179718,
+      "learning_rate": 0.008,
+      "loss": 1.6441,
+      "step": 3140
+    },
+    {
+      "epoch": 0.3322784810126582,
+      "grad_norm": 0.4031313955783844,
+      "learning_rate": 0.008,
+      "loss": 1.6544,
+      "step": 3150
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.2615210711956024,
+      "learning_rate": 0.008,
+      "loss": 1.6534,
+      "step": 3160
+    },
+    {
+      "epoch": 0.33438818565400846,
+      "grad_norm": 0.2171945571899414,
+      "learning_rate": 0.008,
+      "loss": 1.6541,
+      "step": 3170
+    },
+    {
+      "epoch": 0.33544303797468356,
+      "grad_norm": 0.2398463636636734,
+      "learning_rate": 0.008,
+      "loss": 1.6503,
+      "step": 3180
+    },
+    {
+      "epoch": 0.33649789029535865,
+      "grad_norm": 0.33890002965927124,
+      "learning_rate": 0.008,
+      "loss": 1.6457,
+      "step": 3190
+    },
+    {
+      "epoch": 0.33755274261603374,
+      "grad_norm": 0.32689160108566284,
+      "learning_rate": 0.008,
+      "loss": 1.6369,
+      "step": 3200
+    },
+    {
+      "epoch": 0.33860759493670883,
+      "grad_norm": 0.31735533475875854,
+      "learning_rate": 0.008,
+      "loss": 1.6567,
+      "step": 3210
+    },
+    {
+      "epoch": 0.339662447257384,
+      "grad_norm": 0.20491929352283478,
+      "learning_rate": 0.008,
+      "loss": 1.6481,
+      "step": 3220
+    },
+    {
+      "epoch": 0.3407172995780591,
+      "grad_norm": 0.2518443167209625,
+      "learning_rate": 0.008,
+      "loss": 1.6428,
+      "step": 3230
+    },
+    {
+      "epoch": 0.34177215189873417,
+      "grad_norm": 0.21067354083061218,
+      "learning_rate": 0.008,
+      "loss": 1.6151,
+      "step": 3240
+    },
+    {
+      "epoch": 0.34282700421940926,
+      "grad_norm": 0.25579094886779785,
+      "learning_rate": 0.008,
+      "loss": 1.6406,
+      "step": 3250
+    },
+    {
+      "epoch": 0.3438818565400844,
+      "grad_norm": 0.21816200017929077,
+      "learning_rate": 0.008,
+      "loss": 1.6334,
+      "step": 3260
+    },
+    {
+      "epoch": 0.3449367088607595,
+      "grad_norm": 0.2387901395559311,
+      "learning_rate": 0.008,
+      "loss": 1.6493,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3459915611814346,
+      "grad_norm": 0.2216261625289917,
+      "learning_rate": 0.008,
+      "loss": 1.6466,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3470464135021097,
+      "grad_norm": 0.25106218457221985,
+      "learning_rate": 0.008,
+      "loss": 1.635,
+      "step": 3290
+    },
+    {
+      "epoch": 0.34810126582278483,
+      "grad_norm": 0.22046318650245667,
+      "learning_rate": 0.008,
+      "loss": 1.6421,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3491561181434599,
+      "grad_norm": 0.21702668070793152,
+      "learning_rate": 0.008,
+      "loss": 1.6392,
+      "step": 3310
+    },
+    {
+      "epoch": 0.350210970464135,
+      "grad_norm": 0.21423888206481934,
+      "learning_rate": 0.008,
+      "loss": 1.6466,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3512658227848101,
+      "grad_norm": 0.28289756178855896,
+      "learning_rate": 0.008,
+      "loss": 1.6328,
+      "step": 3330
+    },
+    {
+      "epoch": 0.35232067510548526,
+      "grad_norm": 0.2918862998485565,
+      "learning_rate": 0.008,
+      "loss": 1.6379,
+      "step": 3340
+    },
+    {
+      "epoch": 0.35337552742616035,
+      "grad_norm": 0.34093281626701355,
+      "learning_rate": 0.008,
+      "loss": 1.6307,
+      "step": 3350
+    },
+    {
+      "epoch": 0.35443037974683544,
+      "grad_norm": 0.3417920172214508,
+      "learning_rate": 0.008,
+      "loss": 1.6267,
+      "step": 3360
+    },
+    {
+      "epoch": 0.35548523206751054,
+      "grad_norm": 0.21893878281116486,
+      "learning_rate": 0.008,
+      "loss": 1.633,
+      "step": 3370
+    },
+    {
+      "epoch": 0.35654008438818563,
+      "grad_norm": 0.3563060462474823,
+      "learning_rate": 0.008,
+      "loss": 1.6287,
+      "step": 3380
+    },
+    {
+      "epoch": 0.3575949367088608,
+      "grad_norm": 0.23708422482013702,
+      "learning_rate": 0.008,
+      "loss": 1.621,
+      "step": 3390
+    },
+    {
+      "epoch": 0.35864978902953587,
+      "grad_norm": 0.33330103754997253,
+      "learning_rate": 0.008,
+      "loss": 1.6305,
+      "step": 3400
+    },
+    {
+      "epoch": 0.35970464135021096,
+      "grad_norm": 0.22748327255249023,
+      "learning_rate": 0.008,
+      "loss": 1.6291,
+      "step": 3410
+    },
+    {
+      "epoch": 0.36075949367088606,
+      "grad_norm": 0.26142457127571106,
+      "learning_rate": 0.008,
+      "loss": 1.6102,
+      "step": 3420
+    },
+    {
+      "epoch": 0.3618143459915612,
+      "grad_norm": 0.3028963506221771,
+      "learning_rate": 0.008,
+      "loss": 1.6325,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3628691983122363,
+      "grad_norm": 0.25474515557289124,
+      "learning_rate": 0.008,
+      "loss": 1.6225,
+      "step": 3440
+    },
+    {
+      "epoch": 0.3639240506329114,
+      "grad_norm": 0.24747736752033234,
+      "learning_rate": 0.008,
+      "loss": 1.6177,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3649789029535865,
+      "grad_norm": 0.32310861349105835,
+      "learning_rate": 0.008,
+      "loss": 1.6133,
+      "step": 3460
+    },
+    {
+      "epoch": 0.36603375527426163,
+      "grad_norm": 0.31213173270225525,
+      "learning_rate": 0.008,
+      "loss": 1.6265,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3670886075949367,
+      "grad_norm": 0.24284648895263672,
+      "learning_rate": 0.008,
+      "loss": 1.636,
+      "step": 3480
+    },
+    {
+      "epoch": 0.3681434599156118,
+      "grad_norm": 0.3295861482620239,
+      "learning_rate": 0.008,
+      "loss": 1.6261,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3691983122362869,
+      "grad_norm": 0.25787562131881714,
+      "learning_rate": 0.008,
+      "loss": 1.6066,
+      "step": 3500
+    },
+    {
+      "epoch": 0.370253164556962,
+      "grad_norm": 0.24469667673110962,
+      "learning_rate": 0.008,
+      "loss": 1.6246,
+      "step": 3510
+    },
+    {
+      "epoch": 0.37130801687763715,
+      "grad_norm": 0.2611663341522217,
+      "learning_rate": 0.008,
+      "loss": 1.6182,
+      "step": 3520
+    },
+    {
+      "epoch": 0.37236286919831224,
+      "grad_norm": 0.24408909678459167,
+      "learning_rate": 0.008,
+      "loss": 1.6317,
+      "step": 3530
+    },
+    {
+      "epoch": 0.37341772151898733,
+      "grad_norm": 0.2229597121477127,
+      "learning_rate": 0.008,
+      "loss": 1.6247,
+      "step": 3540
+    },
+    {
+      "epoch": 0.3744725738396624,
+      "grad_norm": 0.27489173412323,
+      "learning_rate": 0.008,
+      "loss": 1.6111,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3755274261603376,
+      "grad_norm": 0.3660025894641876,
+      "learning_rate": 0.008,
+      "loss": 1.6072,
+      "step": 3560
+    },
+    {
+      "epoch": 0.37658227848101267,
+      "grad_norm": 0.3601195514202118,
+      "learning_rate": 0.008,
+      "loss": 1.6097,
+      "step": 3570
+    },
+    {
+      "epoch": 0.37763713080168776,
+      "grad_norm": 0.24746383726596832,
+      "learning_rate": 0.008,
+      "loss": 1.6076,
+      "step": 3580
+    },
+    {
+      "epoch": 0.37869198312236285,
+      "grad_norm": 0.2157265990972519,
+      "learning_rate": 0.008,
+      "loss": 1.6184,
+      "step": 3590
+    },
+    {
+      "epoch": 0.379746835443038,
+      "grad_norm": 0.26619330048561096,
+      "learning_rate": 0.008,
+      "loss": 1.5961,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3808016877637131,
+      "grad_norm": 0.2862868905067444,
+      "learning_rate": 0.008,
+      "loss": 1.603,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3818565400843882,
+      "grad_norm": 0.24150559306144714,
+      "learning_rate": 0.008,
+      "loss": 1.6069,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3829113924050633,
+      "grad_norm": 0.2835387885570526,
+      "learning_rate": 0.008,
+      "loss": 1.6114,
+      "step": 3630
+    },
+    {
+      "epoch": 0.38396624472573837,
+      "grad_norm": 0.22218230366706848,
+      "learning_rate": 0.008,
+      "loss": 1.6043,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3850210970464135,
+      "grad_norm": 0.21408239006996155,
+      "learning_rate": 0.008,
+      "loss": 1.6126,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3860759493670886,
+      "grad_norm": 0.22069400548934937,
+      "learning_rate": 0.008,
+      "loss": 1.6066,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3871308016877637,
+      "grad_norm": 0.2796913981437683,
+      "learning_rate": 0.008,
+      "loss": 1.5941,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3881856540084388,
+      "grad_norm": 0.20615093410015106,
+      "learning_rate": 0.008,
+      "loss": 1.6031,
+      "step": 3680
+    },
+    {
+      "epoch": 0.38924050632911394,
+      "grad_norm": 0.253054678440094,
+      "learning_rate": 0.008,
+      "loss": 1.6092,
+      "step": 3690
+    },
+    {
+      "epoch": 0.39029535864978904,
+      "grad_norm": 0.2173006683588028,
+      "learning_rate": 0.008,
+      "loss": 1.6105,
+      "step": 3700
+    },
+    {
+      "epoch": 0.39135021097046413,
+      "grad_norm": 0.21848998963832855,
+      "learning_rate": 0.008,
+      "loss": 1.5988,
+      "step": 3710
+    },
+    {
+      "epoch": 0.3924050632911392,
+      "grad_norm": 0.22108307480812073,
+      "learning_rate": 0.008,
+      "loss": 1.5774,
+      "step": 3720
+    },
+    {
+      "epoch": 0.39345991561181437,
+      "grad_norm": 0.29135680198669434,
+      "learning_rate": 0.008,
+      "loss": 1.5983,
+      "step": 3730
+    },
+    {
+      "epoch": 0.39451476793248946,
+      "grad_norm": 0.23562335968017578,
+      "learning_rate": 0.008,
+      "loss": 1.5965,
+      "step": 3740
+    },
+    {
+      "epoch": 0.39556962025316456,
+      "grad_norm": 0.2638353705406189,
+      "learning_rate": 0.008,
+      "loss": 1.6021,
+      "step": 3750
+    },
+    {
+      "epoch": 0.39662447257383965,
+      "grad_norm": 0.3089199960231781,
+      "learning_rate": 0.008,
+      "loss": 1.5914,
+      "step": 3760
+    },
+    {
+      "epoch": 0.39767932489451474,
+      "grad_norm": 0.5287401080131531,
+      "learning_rate": 0.008,
+      "loss": 1.6104,
+      "step": 3770
+    },
+    {
+      "epoch": 0.3987341772151899,
+      "grad_norm": 0.2012922763824463,
+      "learning_rate": 0.008,
+      "loss": 1.592,
+      "step": 3780
+    },
+    {
+      "epoch": 0.399789029535865,
+      "grad_norm": 0.22888585925102234,
+      "learning_rate": 0.008,
+      "loss": 1.5965,
+      "step": 3790
+    },
+    {
+      "epoch": 0.4008438818565401,
+      "grad_norm": 0.2431032359600067,
+      "learning_rate": 0.008,
+      "loss": 1.5993,
+      "step": 3800
+    },
+    {
+      "epoch": 0.40189873417721517,
+      "grad_norm": 0.37601834535598755,
+      "learning_rate": 0.008,
+      "loss": 1.588,
+      "step": 3810
+    },
+    {
+      "epoch": 0.4029535864978903,
+      "grad_norm": 0.22893565893173218,
+      "learning_rate": 0.008,
+      "loss": 1.5795,
+      "step": 3820
+    },
+    {
+      "epoch": 0.4040084388185654,
+      "grad_norm": 0.21197201311588287,
+      "learning_rate": 0.008,
+      "loss": 1.5852,
+      "step": 3830
+    },
+    {
+      "epoch": 0.4050632911392405,
+      "grad_norm": 0.2951778471469879,
+      "learning_rate": 0.008,
+      "loss": 1.6027,
+      "step": 3840
+    },
+    {
+      "epoch": 0.4061181434599156,
+      "grad_norm": 0.30199143290519714,
+      "learning_rate": 0.008,
+      "loss": 1.5825,
+      "step": 3850
+    },
+    {
+      "epoch": 0.40717299578059074,
+      "grad_norm": 0.23739227652549744,
+      "learning_rate": 0.008,
+      "loss": 1.5966,
+      "step": 3860
+    },
+    {
+      "epoch": 0.40822784810126583,
+      "grad_norm": 0.3237740099430084,
+      "learning_rate": 0.008,
+      "loss": 1.5866,
+      "step": 3870
+    },
+    {
+      "epoch": 0.4092827004219409,
+      "grad_norm": 0.27083566784858704,
+      "learning_rate": 0.008,
+      "loss": 1.575,
+      "step": 3880
+    },
+    {
+      "epoch": 0.410337552742616,
+      "grad_norm": 0.2606992721557617,
+      "learning_rate": 0.008,
+      "loss": 1.5706,
+      "step": 3890
+    },
+    {
+      "epoch": 0.41139240506329117,
+      "grad_norm": 0.24509817361831665,
+      "learning_rate": 0.008,
+      "loss": 1.5744,
+      "step": 3900
+    },
+    {
+      "epoch": 0.41244725738396626,
+      "grad_norm": 0.2939331531524658,
+      "learning_rate": 0.008,
+      "loss": 1.5767,
+      "step": 3910
+    },
+    {
+      "epoch": 0.41350210970464135,
+      "grad_norm": 0.2508770823478699,
+      "learning_rate": 0.008,
+      "loss": 1.5745,
+      "step": 3920
+    },
+    {
+      "epoch": 0.41455696202531644,
+      "grad_norm": 0.2209467738866806,
+      "learning_rate": 0.008,
+      "loss": 1.5811,
+      "step": 3930
+    },
+    {
+      "epoch": 0.41561181434599154,
+      "grad_norm": 0.3235878646373749,
+      "learning_rate": 0.008,
+      "loss": 1.5813,
+      "step": 3940
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.3402903378009796,
+      "learning_rate": 0.008,
+      "loss": 1.5804,
+      "step": 3950
+    },
+    {
+      "epoch": 0.4177215189873418,
+      "grad_norm": 0.20000901818275452,
+      "learning_rate": 0.008,
+      "loss": 1.5769,
+      "step": 3960
+    },
+    {
+      "epoch": 0.41877637130801687,
+      "grad_norm": 0.24537113308906555,
+      "learning_rate": 0.008,
+      "loss": 1.5689,
+      "step": 3970
+    },
+    {
+      "epoch": 0.41983122362869196,
+      "grad_norm": 0.22198963165283203,
+      "learning_rate": 0.008,
+      "loss": 1.5783,
+      "step": 3980
+    },
+    {
+      "epoch": 0.4208860759493671,
+      "grad_norm": 0.2113785743713379,
+      "learning_rate": 0.008,
+      "loss": 1.5688,
+      "step": 3990
+    },
+    {
+      "epoch": 0.4219409282700422,
+      "grad_norm": 0.24396277964115143,
+      "learning_rate": 0.008,
+      "loss": 1.5832,
+      "step": 4000
+    },
+    {
+      "epoch": 0.4229957805907173,
+      "grad_norm": 0.3374914228916168,
+      "learning_rate": 0.008,
+      "loss": 1.5719,
+      "step": 4010
+    },
+    {
+      "epoch": 0.4240506329113924,
+      "grad_norm": 0.21331454813480377,
+      "learning_rate": 0.008,
+      "loss": 1.5788,
+      "step": 4020
+    },
+    {
+      "epoch": 0.42510548523206754,
+      "grad_norm": 0.2638781666755676,
+      "learning_rate": 0.008,
+      "loss": 1.57,
+      "step": 4030
+    },
+    {
+      "epoch": 0.42616033755274263,
+      "grad_norm": 0.2854248285293579,
+      "learning_rate": 0.008,
+      "loss": 1.5757,
+      "step": 4040
+    },
+    {
+      "epoch": 0.4272151898734177,
+      "grad_norm": 0.24664545059204102,
+      "learning_rate": 0.008,
+      "loss": 1.5793,
+      "step": 4050
+    },
+    {
+      "epoch": 0.4282700421940928,
+      "grad_norm": 0.2176240086555481,
+      "learning_rate": 0.008,
+      "loss": 1.5813,
+      "step": 4060
+    },
+    {
+      "epoch": 0.4293248945147679,
+      "grad_norm": 0.278923898935318,
+      "learning_rate": 0.008,
+      "loss": 1.5747,
+      "step": 4070
+    },
+    {
+      "epoch": 0.43037974683544306,
+      "grad_norm": 0.255481481552124,
+      "learning_rate": 0.008,
+      "loss": 1.5795,
+      "step": 4080
+    },
+    {
+      "epoch": 0.43143459915611815,
+      "grad_norm": 0.2019348293542862,
+      "learning_rate": 0.008,
+      "loss": 1.5683,
+      "step": 4090
+    },
+    {
+      "epoch": 0.43248945147679324,
+      "grad_norm": 0.22095830738544464,
+      "learning_rate": 0.008,
+      "loss": 1.566,
+      "step": 4100
+    },
+    {
+      "epoch": 0.43354430379746833,
+      "grad_norm": 0.22949481010437012,
+      "learning_rate": 0.008,
+      "loss": 1.5637,
+      "step": 4110
+    },
+    {
+      "epoch": 0.4345991561181435,
+      "grad_norm": 0.3485650420188904,
+      "learning_rate": 0.008,
+      "loss": 1.5724,
+      "step": 4120
+    },
+    {
+      "epoch": 0.4356540084388186,
+      "grad_norm": 0.23598705232143402,
+      "learning_rate": 0.008,
+      "loss": 1.5573,
+      "step": 4130
+    },
+    {
+      "epoch": 0.43670886075949367,
+      "grad_norm": 0.3303454518318176,
+      "learning_rate": 0.008,
+      "loss": 1.5633,
+      "step": 4140
+    },
+    {
+      "epoch": 0.43776371308016876,
+      "grad_norm": 0.25837430357933044,
+      "learning_rate": 0.008,
+      "loss": 1.5674,
+      "step": 4150
+    },
+    {
+      "epoch": 0.4388185654008439,
+      "grad_norm": 0.28610730171203613,
+      "learning_rate": 0.008,
+      "loss": 1.5615,
+      "step": 4160
+    },
+    {
+      "epoch": 0.439873417721519,
+      "grad_norm": 0.3369266092777252,
+      "learning_rate": 0.008,
+      "loss": 1.5612,
+      "step": 4170
+    },
+    {
+      "epoch": 0.4409282700421941,
+      "grad_norm": 0.22583672404289246,
+      "learning_rate": 0.008,
+      "loss": 1.5603,
+      "step": 4180
+    },
+    {
+      "epoch": 0.4419831223628692,
+      "grad_norm": 0.28444239497184753,
+      "learning_rate": 0.008,
+      "loss": 1.5646,
+      "step": 4190
+    },
+    {
+      "epoch": 0.4430379746835443,
+      "grad_norm": 0.22397972643375397,
+      "learning_rate": 0.008,
+      "loss": 1.563,
+      "step": 4200
+    },
+    {
+      "epoch": 0.4440928270042194,
+      "grad_norm": 0.1992371380329132,
+      "learning_rate": 0.008,
+      "loss": 1.5488,
+      "step": 4210
+    },
+    {
+      "epoch": 0.4451476793248945,
+      "grad_norm": 0.21723061800003052,
+      "learning_rate": 0.008,
+      "loss": 1.5637,
+      "step": 4220
+    },
+    {
+      "epoch": 0.4462025316455696,
+      "grad_norm": 0.2561425268650055,
+      "learning_rate": 0.008,
+      "loss": 1.5599,
+      "step": 4230
+    },
+    {
+      "epoch": 0.4472573839662447,
+      "grad_norm": 0.20931363105773926,
+      "learning_rate": 0.008,
+      "loss": 1.563,
+      "step": 4240
+    },
+    {
+      "epoch": 0.44831223628691985,
+      "grad_norm": 0.26806730031967163,
+      "learning_rate": 0.008,
+      "loss": 1.5596,
+      "step": 4250
+    },
+    {
+      "epoch": 0.44936708860759494,
+      "grad_norm": 0.3182838261127472,
+      "learning_rate": 0.008,
+      "loss": 1.5704,
+      "step": 4260
+    },
+    {
+      "epoch": 0.45042194092827004,
+      "grad_norm": 0.3394760489463806,
+      "learning_rate": 0.008,
+      "loss": 1.5572,
+      "step": 4270
+    },
+    {
+      "epoch": 0.45147679324894513,
+      "grad_norm": 0.27967339754104614,
+      "learning_rate": 0.008,
+      "loss": 1.5513,
+      "step": 4280
+    },
+    {
+      "epoch": 0.4525316455696203,
+      "grad_norm": 0.2314608097076416,
+      "learning_rate": 0.008,
+      "loss": 1.563,
+      "step": 4290
+    },
+    {
+      "epoch": 0.45358649789029537,
+      "grad_norm": 0.3436617851257324,
+      "learning_rate": 0.008,
+      "loss": 1.5596,
+      "step": 4300
+    },
+    {
+      "epoch": 0.45464135021097046,
+      "grad_norm": 0.547049880027771,
+      "learning_rate": 0.008,
+      "loss": 1.5652,
+      "step": 4310
+    },
+    {
+      "epoch": 0.45569620253164556,
+      "grad_norm": 0.2798489034175873,
+      "learning_rate": 0.008,
+      "loss": 1.5564,
+      "step": 4320
+    },
+    {
+      "epoch": 0.45675105485232065,
+      "grad_norm": 0.2232053279876709,
+      "learning_rate": 0.008,
+      "loss": 1.5607,
+      "step": 4330
+    },
+    {
+      "epoch": 0.4578059071729958,
+      "grad_norm": 0.28966420888900757,
+      "learning_rate": 0.008,
+      "loss": 1.5514,
+      "step": 4340
+    },
+    {
+      "epoch": 0.4588607594936709,
+      "grad_norm": 0.20060136914253235,
+      "learning_rate": 0.008,
+      "loss": 1.5477,
+      "step": 4350
+    },
+    {
+      "epoch": 0.459915611814346,
+      "grad_norm": 0.2985727787017822,
+      "learning_rate": 0.008,
+      "loss": 1.5631,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4609704641350211,
+      "grad_norm": 0.5013411641120911,
+      "learning_rate": 0.008,
+      "loss": 1.5452,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4620253164556962,
+      "grad_norm": 0.3250082731246948,
+      "learning_rate": 0.008,
+      "loss": 1.555,
+      "step": 4380
+    },
+    {
+      "epoch": 0.4630801687763713,
+      "grad_norm": 0.22317084670066833,
+      "learning_rate": 0.008,
+      "loss": 1.5472,
+      "step": 4390
+    },
+    {
+      "epoch": 0.4641350210970464,
+      "grad_norm": 0.24938450753688812,
+      "learning_rate": 0.008,
+      "loss": 1.5505,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4651898734177215,
+      "grad_norm": 0.31810569763183594,
+      "learning_rate": 0.008,
+      "loss": 1.5644,
+      "step": 4410
+    },
+    {
+      "epoch": 0.46624472573839665,
+      "grad_norm": 0.2366180419921875,
+      "learning_rate": 0.008,
+      "loss": 1.5532,
+      "step": 4420
+    },
+    {
+      "epoch": 0.46729957805907174,
+      "grad_norm": 0.30884552001953125,
+      "learning_rate": 0.008,
+      "loss": 1.5539,
+      "step": 4430
+    },
+    {
+      "epoch": 0.46835443037974683,
+      "grad_norm": 0.3173336684703827,
+      "learning_rate": 0.008,
+      "loss": 1.5466,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4694092827004219,
+      "grad_norm": 0.3191807270050049,
+      "learning_rate": 0.008,
+      "loss": 1.5613,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4704641350210971,
+      "grad_norm": 0.2105773538351059,
+      "learning_rate": 0.008,
+      "loss": 1.5454,
+      "step": 4460
+    },
+    {
+      "epoch": 0.47151898734177217,
+      "grad_norm": 0.285229355096817,
+      "learning_rate": 0.008,
+      "loss": 1.552,
+      "step": 4470
+    },
+    {
+      "epoch": 0.47257383966244726,
+      "grad_norm": 0.22328303754329681,
+      "learning_rate": 0.008,
+      "loss": 1.555,
+      "step": 4480
+    },
+    {
+      "epoch": 0.47362869198312235,
+      "grad_norm": 0.3808874189853668,
+      "learning_rate": 0.008,
+      "loss": 1.542,
+      "step": 4490
+    },
+    {
+      "epoch": 0.47468354430379744,
+      "grad_norm": 0.23153117299079895,
+      "learning_rate": 0.008,
+      "loss": 1.5456,
+      "step": 4500
+    },
+    {
+      "epoch": 0.4757383966244726,
+      "grad_norm": 0.34866687655448914,
+      "learning_rate": 0.008,
+      "loss": 1.5557,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4767932489451477,
+      "grad_norm": 0.2152269333600998,
+      "learning_rate": 0.008,
+      "loss": 1.5256,
+      "step": 4520
+    },
+    {
+      "epoch": 0.4778481012658228,
+      "grad_norm": 0.25346288084983826,
+      "learning_rate": 0.008,
+      "loss": 1.5489,
+      "step": 4530
+    },
+    {
+      "epoch": 0.47890295358649787,
+      "grad_norm": 0.2619509696960449,
+      "learning_rate": 0.008,
+      "loss": 1.5585,
+      "step": 4540
+    },
+    {
+      "epoch": 0.479957805907173,
+      "grad_norm": 0.2496286928653717,
+      "learning_rate": 0.008,
+      "loss": 1.5479,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4810126582278481,
+      "grad_norm": 0.3418886065483093,
+      "learning_rate": 0.008,
+      "loss": 1.5331,
+      "step": 4560
+    },
+    {
+      "epoch": 0.4820675105485232,
+      "grad_norm": 0.389988511800766,
+      "learning_rate": 0.008,
+      "loss": 1.5226,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4831223628691983,
+      "grad_norm": 0.2303869128227234,
+      "learning_rate": 0.008,
+      "loss": 1.5449,
+      "step": 4580
+    },
+    {
+      "epoch": 0.48417721518987344,
+      "grad_norm": 0.26946914196014404,
+      "learning_rate": 0.008,
+      "loss": 1.5268,
+      "step": 4590
+    },
+    {
+      "epoch": 0.48523206751054854,
+      "grad_norm": 0.3031612038612366,
+      "learning_rate": 0.008,
+      "loss": 1.5451,
+      "step": 4600
+    },
+    {
+      "epoch": 0.48628691983122363,
+      "grad_norm": 0.33278584480285645,
+      "learning_rate": 0.008,
+      "loss": 1.5485,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4873417721518987,
+      "grad_norm": 0.299842894077301,
+      "learning_rate": 0.008,
+      "loss": 1.5213,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4883966244725738,
+      "grad_norm": 0.2404276728630066,
+      "learning_rate": 0.008,
+      "loss": 1.54,
+      "step": 4630
+    },
+    {
+      "epoch": 0.48945147679324896,
+      "grad_norm": 0.2472049444913864,
+      "learning_rate": 0.008,
+      "loss": 1.5414,
+      "step": 4640
+    },
+    {
+      "epoch": 0.49050632911392406,
+      "grad_norm": 0.22804011404514313,
+      "learning_rate": 0.008,
+      "loss": 1.5442,
+      "step": 4650
+    },
+    {
+      "epoch": 0.49156118143459915,
+      "grad_norm": 0.3478902280330658,
+      "learning_rate": 0.008,
+      "loss": 1.5333,
+      "step": 4660
+    },
+    {
+      "epoch": 0.49261603375527424,
+      "grad_norm": 0.3968930244445801,
+      "learning_rate": 0.008,
+      "loss": 1.5315,
+      "step": 4670
+    },
+    {
+      "epoch": 0.4936708860759494,
+      "grad_norm": 0.24186193943023682,
+      "learning_rate": 0.008,
+      "loss": 1.5317,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4947257383966245,
+      "grad_norm": 0.21337531507015228,
+      "learning_rate": 0.008,
+      "loss": 1.5249,
+      "step": 4690
+    },
+    {
+      "epoch": 0.4957805907172996,
+      "grad_norm": 0.26391714811325073,
+      "learning_rate": 0.008,
+      "loss": 1.5349,
+      "step": 4700
+    },
+    {
+      "epoch": 0.49683544303797467,
+      "grad_norm": 0.23413875699043274,
+      "learning_rate": 0.008,
+      "loss": 1.5348,
+      "step": 4710
+    },
+    {
+      "epoch": 0.4978902953586498,
+      "grad_norm": 0.3573142886161804,
+      "learning_rate": 0.008,
+      "loss": 1.5278,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4989451476793249,
+      "grad_norm": 0.26235291361808777,
+      "learning_rate": 0.008,
+      "loss": 1.5262,
+      "step": 4730
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.242934912443161,
+      "learning_rate": 0.008,
+      "loss": 1.5283,
+      "step": 4740
+    },
+    {
+      "epoch": 0.5010548523206751,
+      "grad_norm": 0.282820463180542,
+      "learning_rate": 0.008,
+      "loss": 1.5521,
+      "step": 4750
+    },
+    {
+      "epoch": 0.5021097046413502,
+      "grad_norm": 0.2639579772949219,
+      "learning_rate": 0.008,
+      "loss": 1.538,
+      "step": 4760
+    },
+    {
+      "epoch": 0.5031645569620253,
+      "grad_norm": 0.21449536085128784,
+      "learning_rate": 0.008,
+      "loss": 1.5291,
+      "step": 4770
+    },
+    {
+      "epoch": 0.5042194092827004,
+      "grad_norm": 0.23588572442531586,
+      "learning_rate": 0.008,
+      "loss": 1.5367,
+      "step": 4780
+    },
+    {
+      "epoch": 0.5052742616033755,
+      "grad_norm": 0.33794716000556946,
+      "learning_rate": 0.008,
+      "loss": 1.5163,
+      "step": 4790
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 0.2926550507545471,
+      "learning_rate": 0.008,
+      "loss": 1.5255,
+      "step": 4800
+    },
+    {
+      "epoch": 0.5073839662447257,
+      "grad_norm": 0.36221879720687866,
+      "learning_rate": 0.008,
+      "loss": 1.5278,
+      "step": 4810
+    },
+    {
+      "epoch": 0.5084388185654009,
+      "grad_norm": 0.23408497869968414,
+      "learning_rate": 0.008,
+      "loss": 1.5418,
+      "step": 4820
+    },
+    {
+      "epoch": 0.509493670886076,
+      "grad_norm": 0.5055451393127441,
+      "learning_rate": 0.008,
+      "loss": 1.5307,
+      "step": 4830
+    },
+    {
+      "epoch": 0.510548523206751,
+      "grad_norm": 0.25863686203956604,
+      "learning_rate": 0.008,
+      "loss": 1.5206,
+      "step": 4840
+    },
+    {
+      "epoch": 0.5116033755274262,
+      "grad_norm": 0.2230941653251648,
+      "learning_rate": 0.008,
+      "loss": 1.5321,
+      "step": 4850
+    },
+    {
+      "epoch": 0.5126582278481012,
+      "grad_norm": 0.20042066276073456,
+      "learning_rate": 0.008,
+      "loss": 1.5207,
+      "step": 4860
+    },
+    {
+      "epoch": 0.5137130801687764,
+      "grad_norm": 0.24438785016536713,
+      "learning_rate": 0.008,
+      "loss": 1.5171,
+      "step": 4870
+    },
+    {
+      "epoch": 0.5147679324894515,
+      "grad_norm": 0.27487435936927795,
+      "learning_rate": 0.008,
+      "loss": 1.5351,
+      "step": 4880
+    },
+    {
+      "epoch": 0.5158227848101266,
+      "grad_norm": 0.25218552350997925,
+      "learning_rate": 0.008,
+      "loss": 1.5285,
+      "step": 4890
+    },
+    {
+      "epoch": 0.5168776371308017,
+      "grad_norm": 0.20961767435073853,
+      "learning_rate": 0.008,
+      "loss": 1.5272,
+      "step": 4900
+    },
+    {
+      "epoch": 0.5179324894514767,
+      "grad_norm": 0.28460049629211426,
+      "learning_rate": 0.008,
+      "loss": 1.5254,
+      "step": 4910
+    },
+    {
+      "epoch": 0.5189873417721519,
+      "grad_norm": 0.22512798011302948,
+      "learning_rate": 0.008,
+      "loss": 1.5166,
+      "step": 4920
+    },
+    {
+      "epoch": 0.520042194092827,
+      "grad_norm": 0.38712552189826965,
+      "learning_rate": 0.008,
+      "loss": 1.5245,
+      "step": 4930
+    },
+    {
+      "epoch": 0.5210970464135021,
+      "grad_norm": 0.21800464391708374,
+      "learning_rate": 0.008,
+      "loss": 1.5239,
+      "step": 4940
+    },
+    {
+      "epoch": 0.5221518987341772,
+      "grad_norm": 0.1960371434688568,
+      "learning_rate": 0.008,
+      "loss": 1.5296,
+      "step": 4950
+    },
+    {
+      "epoch": 0.5232067510548524,
+      "grad_norm": 0.2704487442970276,
+      "learning_rate": 0.008,
+      "loss": 1.5092,
+      "step": 4960
+    },
+    {
+      "epoch": 0.5242616033755274,
+      "grad_norm": 0.2294735461473465,
+      "learning_rate": 0.008,
+      "loss": 1.5186,
+      "step": 4970
+    },
+    {
+      "epoch": 0.5253164556962026,
+      "grad_norm": 0.3305419981479645,
+      "learning_rate": 0.008,
+      "loss": 1.5111,
+      "step": 4980
+    },
+    {
+      "epoch": 0.5263713080168776,
+      "grad_norm": 0.30641448497772217,
+      "learning_rate": 0.008,
+      "loss": 1.5152,
+      "step": 4990
+    },
+    {
+      "epoch": 0.5274261603375527,
+      "grad_norm": 0.27723467350006104,
+      "learning_rate": 0.008,
+      "loss": 1.5301,
+      "step": 5000
+    },
+    {
+      "epoch": 0.5284810126582279,
+      "grad_norm": 0.21343223750591278,
+      "learning_rate": 0.008,
+      "loss": 1.5202,
+      "step": 5010
+    },
+    {
+      "epoch": 0.5295358649789029,
+      "grad_norm": 0.22347526252269745,
+      "learning_rate": 0.008,
+      "loss": 1.5102,
+      "step": 5020
+    },
+    {
+      "epoch": 0.5305907172995781,
+      "grad_norm": 0.2190149873495102,
+      "learning_rate": 0.008,
+      "loss": 1.493,
+      "step": 5030
+    },
+    {
+      "epoch": 0.5316455696202531,
+      "grad_norm": 0.22474190592765808,
+      "learning_rate": 0.008,
+      "loss": 1.52,
+      "step": 5040
+    },
+    {
+      "epoch": 0.5327004219409283,
+      "grad_norm": 0.3623146414756775,
+      "learning_rate": 0.008,
+      "loss": 1.5121,
+      "step": 5050
+    },
+    {
+      "epoch": 0.5337552742616034,
+      "grad_norm": 0.28775203227996826,
+      "learning_rate": 0.008,
+      "loss": 1.5268,
+      "step": 5060
+    },
+    {
+      "epoch": 0.5348101265822784,
+      "grad_norm": 0.21539726853370667,
+      "learning_rate": 0.008,
+      "loss": 1.5079,
+      "step": 5070
+    },
+    {
+      "epoch": 0.5358649789029536,
+      "grad_norm": 0.2831694185733795,
+      "learning_rate": 0.008,
+      "loss": 1.5246,
+      "step": 5080
+    },
+    {
+      "epoch": 0.5369198312236287,
+      "grad_norm": 0.2622271478176117,
+      "learning_rate": 0.008,
+      "loss": 1.5149,
+      "step": 5090
+    },
+    {
+      "epoch": 0.5379746835443038,
+      "grad_norm": 0.2977266013622284,
+      "learning_rate": 0.008,
+      "loss": 1.4998,
+      "step": 5100
+    },
+    {
+      "epoch": 0.5390295358649789,
+      "grad_norm": 0.2671653926372528,
+      "learning_rate": 0.008,
+      "loss": 1.5113,
+      "step": 5110
+    },
+    {
+      "epoch": 0.540084388185654,
+      "grad_norm": 0.23161424696445465,
+      "learning_rate": 0.008,
+      "loss": 1.5086,
+      "step": 5120
+    },
+    {
+      "epoch": 0.5411392405063291,
+      "grad_norm": 0.35098907351493835,
+      "learning_rate": 0.008,
+      "loss": 1.5151,
+      "step": 5130
+    },
+    {
+      "epoch": 0.5421940928270043,
+      "grad_norm": 0.26015084981918335,
+      "learning_rate": 0.008,
+      "loss": 1.5283,
+      "step": 5140
+    },
+    {
+      "epoch": 0.5432489451476793,
+      "grad_norm": 0.3833155930042267,
+      "learning_rate": 0.008,
+      "loss": 1.5293,
+      "step": 5150
+    },
+    {
+      "epoch": 0.5443037974683544,
+      "grad_norm": 0.18990397453308105,
+      "learning_rate": 0.008,
+      "loss": 1.5147,
+      "step": 5160
+    },
+    {
+      "epoch": 0.5453586497890295,
+      "grad_norm": 0.4107200801372528,
+      "learning_rate": 0.008,
+      "loss": 1.5165,
+      "step": 5170
+    },
+    {
+      "epoch": 0.5464135021097046,
+      "grad_norm": 0.49462488293647766,
+      "learning_rate": 0.008,
+      "loss": 1.5275,
+      "step": 5180
+    },
+    {
+      "epoch": 0.5474683544303798,
+      "grad_norm": 0.3285534977912903,
+      "learning_rate": 0.008,
+      "loss": 1.5112,
+      "step": 5190
+    },
+    {
+      "epoch": 0.5485232067510548,
+      "grad_norm": 0.21763521432876587,
+      "learning_rate": 0.008,
+      "loss": 1.5083,
+      "step": 5200
+    },
+    {
+      "epoch": 0.54957805907173,
+      "grad_norm": 0.20753800868988037,
+      "learning_rate": 0.008,
+      "loss": 1.4976,
+      "step": 5210
+    },
+    {
+      "epoch": 0.5506329113924051,
+      "grad_norm": 0.3456606864929199,
+      "learning_rate": 0.008,
+      "loss": 1.5077,
+      "step": 5220
+    },
+    {
+      "epoch": 0.5516877637130801,
+      "grad_norm": 0.25690045952796936,
+      "learning_rate": 0.008,
+      "loss": 1.5081,
+      "step": 5230
+    },
+    {
+      "epoch": 0.5527426160337553,
+      "grad_norm": 0.23591789603233337,
+      "learning_rate": 0.008,
+      "loss": 1.5105,
+      "step": 5240
+    },
+    {
+      "epoch": 0.5537974683544303,
+      "grad_norm": 0.25918540358543396,
+      "learning_rate": 0.008,
+      "loss": 1.5002,
+      "step": 5250
+    },
+    {
+      "epoch": 0.5548523206751055,
+      "grad_norm": 0.21240034699440002,
+      "learning_rate": 0.008,
+      "loss": 1.5069,
+      "step": 5260
+    },
+    {
+      "epoch": 0.5559071729957806,
+      "grad_norm": 0.3988906443119049,
+      "learning_rate": 0.008,
+      "loss": 1.5227,
+      "step": 5270
+    },
+    {
+      "epoch": 0.5569620253164557,
+      "grad_norm": 0.23510928452014923,
+      "learning_rate": 0.008,
+      "loss": 1.5067,
+      "step": 5280
+    },
+    {
+      "epoch": 0.5580168776371308,
+      "grad_norm": 0.23662833869457245,
+      "learning_rate": 0.008,
+      "loss": 1.5047,
+      "step": 5290
+    },
+    {
+      "epoch": 0.5590717299578059,
+      "grad_norm": 0.2627451419830322,
+      "learning_rate": 0.008,
+      "loss": 1.5072,
+      "step": 5300
+    },
+    {
+      "epoch": 0.560126582278481,
+      "grad_norm": 0.21743659675121307,
+      "learning_rate": 0.008,
+      "loss": 1.4949,
+      "step": 5310
+    },
+    {
+      "epoch": 0.5611814345991561,
+      "grad_norm": 0.2362428605556488,
+      "learning_rate": 0.008,
+      "loss": 1.4933,
+      "step": 5320
+    },
+    {
+      "epoch": 0.5622362869198312,
+      "grad_norm": 0.27500665187835693,
+      "learning_rate": 0.008,
+      "loss": 1.5107,
+      "step": 5330
+    },
+    {
+      "epoch": 0.5632911392405063,
+      "grad_norm": 0.2290319949388504,
+      "learning_rate": 0.008,
+      "loss": 1.5029,
+      "step": 5340
+    },
+    {
+      "epoch": 0.5643459915611815,
+      "grad_norm": 0.2679517865180969,
+      "learning_rate": 0.008,
+      "loss": 1.4965,
+      "step": 5350
+    },
+    {
+      "epoch": 0.5654008438818565,
+      "grad_norm": 0.45493894815444946,
+      "learning_rate": 0.008,
+      "loss": 1.5051,
+      "step": 5360
+    },
+    {
+      "epoch": 0.5664556962025317,
+      "grad_norm": 0.2431143969297409,
+      "learning_rate": 0.008,
+      "loss": 1.5066,
+      "step": 5370
+    },
+    {
+      "epoch": 0.5675105485232067,
+      "grad_norm": 0.33066779375076294,
+      "learning_rate": 0.008,
+      "loss": 1.5,
+      "step": 5380
+    },
+    {
+      "epoch": 0.5685654008438819,
+      "grad_norm": 0.20937970280647278,
+      "learning_rate": 0.008,
+      "loss": 1.4978,
+      "step": 5390
+    },
+    {
+      "epoch": 0.569620253164557,
+      "grad_norm": 0.23450255393981934,
+      "learning_rate": 0.008,
+      "loss": 1.5,
+      "step": 5400
+    },
+    {
+      "epoch": 0.570675105485232,
+      "grad_norm": 0.22760634124279022,
+      "learning_rate": 0.008,
+      "loss": 1.5003,
+      "step": 5410
+    },
+    {
+      "epoch": 0.5717299578059072,
+      "grad_norm": 0.24873413145542145,
+      "learning_rate": 0.008,
+      "loss": 1.4974,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5727848101265823,
+      "grad_norm": 0.25174975395202637,
+      "learning_rate": 0.008,
+      "loss": 1.4907,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5738396624472574,
+      "grad_norm": 0.2523603141307831,
+      "learning_rate": 0.008,
+      "loss": 1.4875,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5748945147679325,
+      "grad_norm": 0.32658180594444275,
+      "learning_rate": 0.008,
+      "loss": 1.5021,
+      "step": 5450
+    },
+    {
+      "epoch": 0.5759493670886076,
+      "grad_norm": 0.33146989345550537,
+      "learning_rate": 0.008,
+      "loss": 1.4963,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5770042194092827,
+      "grad_norm": 0.23598571121692657,
+      "learning_rate": 0.008,
+      "loss": 1.4987,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5780590717299579,
+      "grad_norm": 0.22648335993289948,
+      "learning_rate": 0.008,
+      "loss": 1.5046,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5791139240506329,
+      "grad_norm": 0.3484029471874237,
+      "learning_rate": 0.008,
+      "loss": 1.5024,
+      "step": 5490
+    },
+    {
+      "epoch": 0.580168776371308,
+      "grad_norm": 0.3879603445529938,
+      "learning_rate": 0.008,
+      "loss": 1.4925,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5812236286919831,
+      "grad_norm": 0.2314758449792862,
+      "learning_rate": 0.008,
+      "loss": 1.5023,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5822784810126582,
+      "grad_norm": 0.22104331851005554,
+      "learning_rate": 0.008,
+      "loss": 1.4882,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 0.3073943257331848,
+      "learning_rate": 0.008,
+      "loss": 1.4831,
+      "step": 5530
+    },
+    {
+      "epoch": 0.5843881856540084,
+      "grad_norm": 0.1905812919139862,
+      "learning_rate": 0.008,
+      "loss": 1.4842,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5854430379746836,
+      "grad_norm": 0.2285163849592209,
+      "learning_rate": 0.008,
+      "loss": 1.5004,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5864978902953587,
+      "grad_norm": 0.2271253615617752,
+      "learning_rate": 0.008,
+      "loss": 1.5063,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5875527426160337,
+      "grad_norm": 0.2473488301038742,
+      "learning_rate": 0.008,
+      "loss": 1.4941,
+      "step": 5570
+    },
+    {
+      "epoch": 0.5886075949367089,
+      "grad_norm": 0.29567286372184753,
+      "learning_rate": 0.008,
+      "loss": 1.4884,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5896624472573839,
+      "grad_norm": 0.23733600974082947,
+      "learning_rate": 0.008,
+      "loss": 1.491,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5907172995780591,
+      "grad_norm": 0.1937004178762436,
+      "learning_rate": 0.008,
+      "loss": 1.4868,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5917721518987342,
+      "grad_norm": 0.21865755319595337,
+      "learning_rate": 0.008,
+      "loss": 1.4916,
+      "step": 5610
+    },
+    {
+      "epoch": 0.5928270042194093,
+      "grad_norm": 0.3228810131549835,
+      "learning_rate": 0.008,
+      "loss": 1.5011,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5938818565400844,
+      "grad_norm": 0.22229640185832977,
+      "learning_rate": 0.008,
+      "loss": 1.4991,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5949367088607594,
+      "grad_norm": 0.26793500781059265,
+      "learning_rate": 0.008,
+      "loss": 1.507,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5959915611814346,
+      "grad_norm": 0.21602068841457367,
+      "learning_rate": 0.008,
+      "loss": 1.4952,
+      "step": 5650
+    },
+    {
+      "epoch": 0.5970464135021097,
+      "grad_norm": 0.21745645999908447,
+      "learning_rate": 0.008,
+      "loss": 1.4928,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5981012658227848,
+      "grad_norm": 0.3354641795158386,
+      "learning_rate": 0.008,
+      "loss": 1.5048,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5991561181434599,
+      "grad_norm": 0.22571957111358643,
+      "learning_rate": 0.008,
+      "loss": 1.5043,
+      "step": 5680
+    },
+    {
+      "epoch": 0.6002109704641351,
+      "grad_norm": 0.20915043354034424,
+      "learning_rate": 0.008,
+      "loss": 1.4774,
+      "step": 5690
+    },
+    {
+      "epoch": 0.6012658227848101,
+      "grad_norm": 0.25558021664619446,
+      "learning_rate": 0.008,
+      "loss": 1.4786,
+      "step": 5700
+    },
+    {
+      "epoch": 0.6023206751054853,
+      "grad_norm": 0.22984257340431213,
+      "learning_rate": 0.008,
+      "loss": 1.488,
+      "step": 5710
+    },
+    {
+      "epoch": 0.6033755274261603,
+      "grad_norm": 0.20215116441249847,
+      "learning_rate": 0.008,
+      "loss": 1.492,
+      "step": 5720
+    },
+    {
+      "epoch": 0.6044303797468354,
+      "grad_norm": 0.23101164400577545,
+      "learning_rate": 0.008,
+      "loss": 1.4909,
+      "step": 5730
+    },
+    {
+      "epoch": 0.6054852320675106,
+      "grad_norm": 0.23368608951568604,
+      "learning_rate": 0.008,
+      "loss": 1.4849,
+      "step": 5740
+    },
+    {
+      "epoch": 0.6065400843881856,
+      "grad_norm": 0.4375186860561371,
+      "learning_rate": 0.008,
+      "loss": 1.4652,
+      "step": 5750
+    },
+    {
+      "epoch": 0.6075949367088608,
+      "grad_norm": 0.2429998517036438,
+      "learning_rate": 0.008,
+      "loss": 1.4727,
+      "step": 5760
+    },
+    {
+      "epoch": 0.6086497890295358,
+      "grad_norm": 0.2781209945678711,
+      "learning_rate": 0.008,
+      "loss": 1.4726,
+      "step": 5770
+    },
+    {
+      "epoch": 0.609704641350211,
+      "grad_norm": 0.2375238835811615,
+      "learning_rate": 0.008,
+      "loss": 1.4794,
+      "step": 5780
+    },
+    {
+      "epoch": 0.6107594936708861,
+      "grad_norm": 0.3107040524482727,
+      "learning_rate": 0.008,
+      "loss": 1.4793,
+      "step": 5790
+    },
+    {
+      "epoch": 0.6118143459915611,
+      "grad_norm": 0.2782061994075775,
+      "learning_rate": 0.008,
+      "loss": 1.4834,
+      "step": 5800
+    },
+    {
+      "epoch": 0.6128691983122363,
+      "grad_norm": 0.21664227545261383,
+      "learning_rate": 0.008,
+      "loss": 1.4898,
+      "step": 5810
+    },
+    {
+      "epoch": 0.6139240506329114,
+      "grad_norm": 0.3684358298778534,
+      "learning_rate": 0.008,
+      "loss": 1.5024,
+      "step": 5820
+    },
+    {
+      "epoch": 0.6149789029535865,
+      "grad_norm": 0.24630123376846313,
+      "learning_rate": 0.008,
+      "loss": 1.484,
+      "step": 5830
+    },
+    {
+      "epoch": 0.6160337552742616,
+      "grad_norm": 0.23560503125190735,
+      "learning_rate": 0.008,
+      "loss": 1.4936,
+      "step": 5840
+    },
+    {
+      "epoch": 0.6170886075949367,
+      "grad_norm": 0.18302875757217407,
+      "learning_rate": 0.008,
+      "loss": 1.5009,
+      "step": 5850
+    },
+    {
+      "epoch": 0.6181434599156118,
+      "grad_norm": 0.23762358725070953,
+      "learning_rate": 0.008,
+      "loss": 1.483,
+      "step": 5860
+    },
+    {
+      "epoch": 0.619198312236287,
+      "grad_norm": 0.319248765707016,
+      "learning_rate": 0.008,
+      "loss": 1.4805,
+      "step": 5870
+    },
+    {
+      "epoch": 0.620253164556962,
+      "grad_norm": 0.23708374798297882,
+      "learning_rate": 0.008,
+      "loss": 1.4895,
+      "step": 5880
+    },
+    {
+      "epoch": 0.6213080168776371,
+      "grad_norm": 0.34177106618881226,
+      "learning_rate": 0.008,
+      "loss": 1.4851,
+      "step": 5890
+    },
+    {
+      "epoch": 0.6223628691983122,
+      "grad_norm": 0.23187245428562164,
+      "learning_rate": 0.008,
+      "loss": 1.4863,
+      "step": 5900
+    },
+    {
+      "epoch": 0.6234177215189873,
+      "grad_norm": 0.3271633982658386,
+      "learning_rate": 0.008,
+      "loss": 1.479,
+      "step": 5910
+    },
+    {
+      "epoch": 0.6244725738396625,
+      "grad_norm": 0.3913274109363556,
+      "learning_rate": 0.008,
+      "loss": 1.4725,
+      "step": 5920
+    },
+    {
+      "epoch": 0.6255274261603375,
+      "grad_norm": 0.2436475157737732,
+      "learning_rate": 0.008,
+      "loss": 1.4804,
+      "step": 5930
+    },
+    {
+      "epoch": 0.6265822784810127,
+      "grad_norm": 0.2434769719839096,
+      "learning_rate": 0.008,
+      "loss": 1.4867,
+      "step": 5940
+    },
+    {
+      "epoch": 0.6276371308016878,
+      "grad_norm": 0.24504119157791138,
+      "learning_rate": 0.008,
+      "loss": 1.4716,
+      "step": 5950
+    },
+    {
+      "epoch": 0.6286919831223629,
+      "grad_norm": 0.19500687718391418,
+      "learning_rate": 0.008,
+      "loss": 1.4768,
+      "step": 5960
+    },
+    {
+      "epoch": 0.629746835443038,
+      "grad_norm": 0.3898884057998657,
+      "learning_rate": 0.008,
+      "loss": 1.4755,
+      "step": 5970
+    },
+    {
+      "epoch": 0.630801687763713,
+      "grad_norm": 0.22809194028377533,
+      "learning_rate": 0.008,
+      "loss": 1.4654,
+      "step": 5980
+    },
+    {
+      "epoch": 0.6318565400843882,
+      "grad_norm": 0.24901731312274933,
+      "learning_rate": 0.008,
+      "loss": 1.483,
+      "step": 5990
+    },
+    {
+      "epoch": 0.6329113924050633,
+      "grad_norm": 0.2268933355808258,
+      "learning_rate": 0.008,
+      "loss": 1.4894,
+      "step": 6000
+    },
+    {
+      "epoch": 0.6339662447257384,
+      "grad_norm": 0.2531156837940216,
+      "learning_rate": 0.008,
+      "loss": 1.4771,
+      "step": 6010
+    },
+    {
+      "epoch": 0.6350210970464135,
+      "grad_norm": 0.22195585072040558,
+      "learning_rate": 0.008,
+      "loss": 1.4784,
+      "step": 6020
+    },
+    {
+      "epoch": 0.6360759493670886,
+      "grad_norm": 0.2989055812358856,
+      "learning_rate": 0.008,
+      "loss": 1.4618,
+      "step": 6030
+    },
+    {
+      "epoch": 0.6371308016877637,
+      "grad_norm": 0.20482350885868073,
+      "learning_rate": 0.008,
+      "loss": 1.4802,
+      "step": 6040
+    },
+    {
+      "epoch": 0.6381856540084389,
+      "grad_norm": 0.2702943682670593,
+      "learning_rate": 0.008,
+      "loss": 1.4608,
+      "step": 6050
+    },
+    {
+      "epoch": 0.6392405063291139,
+      "grad_norm": 0.2561380863189697,
+      "learning_rate": 0.008,
+      "loss": 1.4855,
+      "step": 6060
+    },
+    {
+      "epoch": 0.640295358649789,
+      "grad_norm": 0.2415098249912262,
+      "learning_rate": 0.008,
+      "loss": 1.4723,
+      "step": 6070
+    },
+    {
+      "epoch": 0.6413502109704642,
+      "grad_norm": 0.24393302202224731,
+      "learning_rate": 0.008,
+      "loss": 1.4756,
+      "step": 6080
+    },
+    {
+      "epoch": 0.6424050632911392,
+      "grad_norm": 0.2857919931411743,
+      "learning_rate": 0.008,
+      "loss": 1.4774,
+      "step": 6090
+    },
+    {
+      "epoch": 0.6434599156118144,
+      "grad_norm": 0.27922961115837097,
+      "learning_rate": 0.008,
+      "loss": 1.4688,
+      "step": 6100
+    },
+    {
+      "epoch": 0.6445147679324894,
+      "grad_norm": 0.31954991817474365,
+      "learning_rate": 0.008,
+      "loss": 1.4785,
+      "step": 6110
+    },
+    {
+      "epoch": 0.6455696202531646,
+      "grad_norm": 0.27288690209388733,
+      "learning_rate": 0.008,
+      "loss": 1.4745,
+      "step": 6120
+    },
+    {
+      "epoch": 0.6466244725738397,
+      "grad_norm": 0.2093956619501114,
+      "learning_rate": 0.008,
+      "loss": 1.4701,
+      "step": 6130
+    },
+    {
+      "epoch": 0.6476793248945147,
+      "grad_norm": 0.27014803886413574,
+      "learning_rate": 0.008,
+      "loss": 1.4678,
+      "step": 6140
+    },
+    {
+      "epoch": 0.6487341772151899,
+      "grad_norm": 0.2078312486410141,
+      "learning_rate": 0.008,
+      "loss": 1.4804,
+      "step": 6150
+    },
+    {
+      "epoch": 0.6497890295358649,
+      "grad_norm": 0.23920007050037384,
+      "learning_rate": 0.008,
+      "loss": 1.4757,
+      "step": 6160
+    },
+    {
+      "epoch": 0.6508438818565401,
+      "grad_norm": 0.2113324999809265,
+      "learning_rate": 0.008,
+      "loss": 1.4517,
+      "step": 6170
+    },
+    {
+      "epoch": 0.6518987341772152,
+      "grad_norm": 0.27210965752601624,
+      "learning_rate": 0.008,
+      "loss": 1.4787,
+      "step": 6180
+    },
+    {
+      "epoch": 0.6529535864978903,
+      "grad_norm": 0.24959327280521393,
+      "learning_rate": 0.008,
+      "loss": 1.4821,
+      "step": 6190
+    },
+    {
+      "epoch": 0.6540084388185654,
+      "grad_norm": 0.23058481514453888,
+      "learning_rate": 0.008,
+      "loss": 1.4704,
+      "step": 6200
+    },
+    {
+      "epoch": 0.6550632911392406,
+      "grad_norm": 0.302617609500885,
+      "learning_rate": 0.008,
+      "loss": 1.4676,
+      "step": 6210
+    },
+    {
+      "epoch": 0.6561181434599156,
+      "grad_norm": 0.2268313616514206,
+      "learning_rate": 0.008,
+      "loss": 1.4745,
+      "step": 6220
+    },
+    {
+      "epoch": 0.6571729957805907,
+      "grad_norm": 0.40284204483032227,
+      "learning_rate": 0.008,
+      "loss": 1.4631,
+      "step": 6230
+    },
+    {
+      "epoch": 0.6582278481012658,
+      "grad_norm": 0.268122136592865,
+      "learning_rate": 0.008,
+      "loss": 1.4712,
+      "step": 6240
+    },
+    {
+      "epoch": 0.6592827004219409,
+      "grad_norm": 0.21114185452461243,
+      "learning_rate": 0.008,
+      "loss": 1.4774,
+      "step": 6250
+    },
+    {
+      "epoch": 0.6603375527426161,
+      "grad_norm": 0.2067754864692688,
+      "learning_rate": 0.008,
+      "loss": 1.4693,
+      "step": 6260
+    },
+    {
+      "epoch": 0.6613924050632911,
+      "grad_norm": 0.2391192615032196,
+      "learning_rate": 0.008,
+      "loss": 1.4775,
+      "step": 6270
+    },
+    {
+      "epoch": 0.6624472573839663,
+      "grad_norm": 0.45928817987442017,
+      "learning_rate": 0.008,
+      "loss": 1.4768,
+      "step": 6280
+    },
+    {
+      "epoch": 0.6635021097046413,
+      "grad_norm": 0.26205435395240784,
+      "learning_rate": 0.008,
+      "loss": 1.4768,
+      "step": 6290
+    },
+    {
+      "epoch": 0.6645569620253164,
+      "grad_norm": 0.22115148603916168,
+      "learning_rate": 0.008,
+      "loss": 1.4656,
+      "step": 6300
+    },
+    {
+      "epoch": 0.6656118143459916,
+      "grad_norm": 0.3403981924057007,
+      "learning_rate": 0.008,
+      "loss": 1.4664,
+      "step": 6310
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.3160533905029297,
+      "learning_rate": 0.008,
+      "loss": 1.4619,
+      "step": 6320
+    },
+    {
+      "epoch": 0.6677215189873418,
+      "grad_norm": 0.2862953841686249,
+      "learning_rate": 0.008,
+      "loss": 1.4695,
+      "step": 6330
+    },
+    {
+      "epoch": 0.6687763713080169,
+      "grad_norm": 0.20100294053554535,
+      "learning_rate": 0.008,
+      "loss": 1.4651,
+      "step": 6340
+    },
+    {
+      "epoch": 0.669831223628692,
+      "grad_norm": 0.2512241005897522,
+      "learning_rate": 0.008,
+      "loss": 1.4645,
+      "step": 6350
+    },
+    {
+      "epoch": 0.6708860759493671,
+      "grad_norm": 0.21082986891269684,
+      "learning_rate": 0.008,
+      "loss": 1.4676,
+      "step": 6360
+    },
+    {
+      "epoch": 0.6719409282700421,
+      "grad_norm": 0.2407890409231186,
+      "learning_rate": 0.008,
+      "loss": 1.4452,
+      "step": 6370
+    },
+    {
+      "epoch": 0.6729957805907173,
+      "grad_norm": 0.215669184923172,
+      "learning_rate": 0.008,
+      "loss": 1.4519,
+      "step": 6380
+    },
+    {
+      "epoch": 0.6740506329113924,
+      "grad_norm": 0.25857481360435486,
+      "learning_rate": 0.008,
+      "loss": 1.4648,
+      "step": 6390
+    },
+    {
+      "epoch": 0.6751054852320675,
+      "grad_norm": 0.2008984386920929,
+      "learning_rate": 0.008,
+      "loss": 1.4642,
+      "step": 6400
+    },
+    {
+      "epoch": 0.6761603375527426,
+      "grad_norm": 0.2710506021976471,
+      "learning_rate": 0.008,
+      "loss": 1.4662,
+      "step": 6410
+    },
+    {
+      "epoch": 0.6772151898734177,
+      "grad_norm": 0.30271366238594055,
+      "learning_rate": 0.008,
+      "loss": 1.4575,
+      "step": 6420
+    },
+    {
+      "epoch": 0.6782700421940928,
+      "grad_norm": 0.30102431774139404,
+      "learning_rate": 0.008,
+      "loss": 1.4717,
+      "step": 6430
+    },
+    {
+      "epoch": 0.679324894514768,
+      "grad_norm": 0.42482349276542664,
+      "learning_rate": 0.008,
+      "loss": 1.4589,
+      "step": 6440
+    },
+    {
+      "epoch": 0.680379746835443,
+      "grad_norm": 0.3051795959472656,
+      "learning_rate": 0.008,
+      "loss": 1.4569,
+      "step": 6450
+    },
+    {
+      "epoch": 0.6814345991561181,
+      "grad_norm": 0.2560389041900635,
+      "learning_rate": 0.008,
+      "loss": 1.4695,
+      "step": 6460
+    },
+    {
+      "epoch": 0.6824894514767933,
+      "grad_norm": 0.23414307832717896,
+      "learning_rate": 0.008,
+      "loss": 1.4595,
+      "step": 6470
+    },
+    {
+      "epoch": 0.6835443037974683,
+      "grad_norm": 0.22050480544567108,
+      "learning_rate": 0.008,
+      "loss": 1.4648,
+      "step": 6480
+    },
+    {
+      "epoch": 0.6845991561181435,
+      "grad_norm": 0.4428369998931885,
+      "learning_rate": 0.008,
+      "loss": 1.4636,
+      "step": 6490
+    },
+    {
+      "epoch": 0.6856540084388185,
+      "grad_norm": 0.2929626405239105,
+      "learning_rate": 0.008,
+      "loss": 1.4798,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6867088607594937,
+      "grad_norm": 0.23227623105049133,
+      "learning_rate": 0.008,
+      "loss": 1.4755,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6877637130801688,
+      "grad_norm": 0.34046855568885803,
+      "learning_rate": 0.008,
+      "loss": 1.4657,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6888185654008439,
+      "grad_norm": 0.24351544678211212,
+      "learning_rate": 0.008,
+      "loss": 1.4525,
+      "step": 6530
+    },
+    {
+      "epoch": 0.689873417721519,
+      "grad_norm": 0.31886032223701477,
+      "learning_rate": 0.008,
+      "loss": 1.4646,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6909282700421941,
+      "grad_norm": 0.42546963691711426,
+      "learning_rate": 0.008,
+      "loss": 1.457,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6919831223628692,
+      "grad_norm": 0.24386058747768402,
+      "learning_rate": 0.008,
+      "loss": 1.459,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6930379746835443,
+      "grad_norm": 0.4220311939716339,
+      "learning_rate": 0.008,
+      "loss": 1.4578,
+      "step": 6570
+    },
+    {
+      "epoch": 0.6940928270042194,
+      "grad_norm": 0.20642231404781342,
+      "learning_rate": 0.008,
+      "loss": 1.4565,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6951476793248945,
+      "grad_norm": 0.2922006845474243,
+      "learning_rate": 0.008,
+      "loss": 1.4622,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6962025316455697,
+      "grad_norm": 0.22632527351379395,
+      "learning_rate": 0.008,
+      "loss": 1.4675,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6972573839662447,
+      "grad_norm": 0.40961337089538574,
+      "learning_rate": 0.008,
+      "loss": 1.4516,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6983122362869199,
+      "grad_norm": 0.2712070047855377,
+      "learning_rate": 0.008,
+      "loss": 1.4678,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6993670886075949,
+      "grad_norm": 0.22098104655742645,
+      "learning_rate": 0.008,
+      "loss": 1.4618,
+      "step": 6630
+    },
+    {
+      "epoch": 0.70042194092827,
+      "grad_norm": 0.2727103531360626,
+      "learning_rate": 0.007948351185209635,
+      "loss": 1.4695,
+      "step": 6640
+    },
+    {
+      "epoch": 0.7014767932489452,
+      "grad_norm": 0.25830674171447754,
+      "learning_rate": 0.007820683286602414,
+      "loss": 1.4513,
+      "step": 6650
+    },
+    {
+      "epoch": 0.7025316455696202,
+      "grad_norm": 0.22860464453697205,
+      "learning_rate": 0.007695066013584702,
+      "loss": 1.4464,
+      "step": 6660
+    },
+    {
+      "epoch": 0.7035864978902954,
+      "grad_norm": 0.2357155978679657,
+      "learning_rate": 0.007571466428625964,
+      "loss": 1.4463,
+      "step": 6670
+    },
+    {
+      "epoch": 0.7046413502109705,
+      "grad_norm": 0.2416096031665802,
+      "learning_rate": 0.007449852123244427,
+      "loss": 1.4494,
+      "step": 6680
+    },
+    {
+      "epoch": 0.7056962025316456,
+      "grad_norm": 0.3616219162940979,
+      "learning_rate": 0.007330191209509389,
+      "loss": 1.4387,
+      "step": 6690
+    },
+    {
+      "epoch": 0.7067510548523207,
+      "grad_norm": 0.19063417613506317,
+      "learning_rate": 0.00721245231168004,
+      "loss": 1.4413,
+      "step": 6700
+    },
+    {
+      "epoch": 0.7078059071729957,
+      "grad_norm": 0.18434371054172516,
+      "learning_rate": 0.0070966045579785655,
+      "loss": 1.4401,
+      "step": 6710
+    },
+    {
+      "epoch": 0.7088607594936709,
+      "grad_norm": 0.19636869430541992,
+      "learning_rate": 0.006982617572495406,
+      "loss": 1.4279,
+      "step": 6720
+    },
+    {
+      "epoch": 0.709915611814346,
+      "grad_norm": 0.20043553411960602,
+      "learning_rate": 0.006870461467224521,
+      "loss": 1.4362,
+      "step": 6730
+    },
+    {
+      "epoch": 0.7109704641350211,
+      "grad_norm": 0.21843086183071136,
+      "learning_rate": 0.006760106834226596,
+      "loss": 1.439,
+      "step": 6740
+    },
+    {
+      "epoch": 0.7120253164556962,
+      "grad_norm": 0.18369698524475098,
+      "learning_rate": 0.00665152473791812,
+      "loss": 1.4375,
+      "step": 6750
+    },
+    {
+      "epoch": 0.7130801687763713,
+      "grad_norm": 0.2016737014055252,
+      "learning_rate": 0.006544686707484321,
+      "loss": 1.4272,
+      "step": 6760
+    },
+    {
+      "epoch": 0.7141350210970464,
+      "grad_norm": 0.23914003372192383,
+      "learning_rate": 0.006439564729413961,
+      "loss": 1.4284,
+      "step": 6770
+    },
+    {
+      "epoch": 0.7151898734177216,
+      "grad_norm": 0.22155533730983734,
+      "learning_rate": 0.006336131240154041,
+      "loss": 1.4266,
+      "step": 6780
+    },
+    {
+      "epoch": 0.7162447257383966,
+      "grad_norm": 0.23960313200950623,
+      "learning_rate": 0.00623435911888249,
+      "loss": 1.4269,
+      "step": 6790
+    },
+    {
+      "epoch": 0.7172995780590717,
+      "grad_norm": 0.20797201991081238,
+      "learning_rate": 0.006134221680396938,
+      "loss": 1.4115,
+      "step": 6800
+    },
+    {
+      "epoch": 0.7183544303797469,
+      "grad_norm": 0.2402528077363968,
+      "learning_rate": 0.006035692668117712,
+      "loss": 1.417,
+      "step": 6810
+    },
+    {
+      "epoch": 0.7194092827004219,
+      "grad_norm": 0.21905063092708588,
+      "learning_rate": 0.005938746247203212,
+      "loss": 1.4119,
+      "step": 6820
+    },
+    {
+      "epoch": 0.7204641350210971,
+      "grad_norm": 0.1899748593568802,
+      "learning_rate": 0.005843356997775885,
+      "loss": 1.4137,
+      "step": 6830
+    },
+    {
+      "epoch": 0.7215189873417721,
+      "grad_norm": 0.17753294110298157,
+      "learning_rate": 0.005749499908256989,
+      "loss": 1.3993,
+      "step": 6840
+    },
+    {
+      "epoch": 0.7225738396624473,
+      "grad_norm": 0.2480160892009735,
+      "learning_rate": 0.005657150368808423,
+      "loss": 1.4067,
+      "step": 6850
+    },
+    {
+      "epoch": 0.7236286919831224,
+      "grad_norm": 0.28500014543533325,
+      "learning_rate": 0.005566284164879892,
+      "loss": 1.4028,
+      "step": 6860
+    },
+    {
+      "epoch": 0.7246835443037974,
+      "grad_norm": 0.2256176769733429,
+      "learning_rate": 0.005476877470859726,
+      "loss": 1.3959,
+      "step": 6870
+    },
+    {
+      "epoch": 0.7257383966244726,
+      "grad_norm": 0.19890552759170532,
+      "learning_rate": 0.005388906843827669,
+      "loss": 1.4064,
+      "step": 6880
+    },
+    {
+      "epoch": 0.7267932489451476,
+      "grad_norm": 0.17259205877780914,
+      "learning_rate": 0.00530234921740802,
+      "loss": 1.3944,
+      "step": 6890
+    },
+    {
+      "epoch": 0.7278481012658228,
+      "grad_norm": 0.20954705774784088,
+      "learning_rate": 0.005217181895721508,
+      "loss": 1.3917,
+      "step": 6900
+    },
+    {
+      "epoch": 0.7289029535864979,
+      "grad_norm": 0.2612088620662689,
+      "learning_rate": 0.0051333825474343025,
+      "loss": 1.4158,
+      "step": 6910
+    },
+    {
+      "epoch": 0.729957805907173,
+      "grad_norm": 0.19605599343776703,
+      "learning_rate": 0.005050929199902624,
+      "loss": 1.3958,
+      "step": 6920
+    },
+    {
+      "epoch": 0.7310126582278481,
+      "grad_norm": 0.2200867235660553,
+      "learning_rate": 0.0049698002334113926,
+      "loss": 1.4002,
+      "step": 6930
+    },
+    {
+      "epoch": 0.7320675105485233,
+      "grad_norm": 0.21439088881015778,
+      "learning_rate": 0.00488997437550542,
+      "loss": 1.4005,
+      "step": 6940
+    },
+    {
+      "epoch": 0.7331223628691983,
+      "grad_norm": 0.26351800560951233,
+      "learning_rate": 0.004811430695411665,
+      "loss": 1.3965,
+      "step": 6950
+    },
+    {
+      "epoch": 0.7341772151898734,
+      "grad_norm": 0.17867186665534973,
+      "learning_rate": 0.004734148598551059,
+      "loss": 1.3856,
+      "step": 6960
+    },
+    {
+      "epoch": 0.7352320675105485,
+      "grad_norm": 0.19592906534671783,
+      "learning_rate": 0.004658107821138505,
+      "loss": 1.3808,
+      "step": 6970
+    },
+    {
+      "epoch": 0.7362869198312236,
+      "grad_norm": 0.21954594552516937,
+      "learning_rate": 0.004583288424869602,
+      "loss": 1.38,
+      "step": 6980
+    },
+    {
+      "epoch": 0.7373417721518988,
+      "grad_norm": 0.2504899203777313,
+      "learning_rate": 0.004509670791692706,
+      "loss": 1.3811,
+      "step": 6990
+    },
+    {
+      "epoch": 0.7383966244725738,
+      "grad_norm": 0.18602634966373444,
+      "learning_rate": 0.004437235618664983,
+      "loss": 1.3895,
+      "step": 7000
+    },
+    {
+      "epoch": 0.739451476793249,
+      "grad_norm": 0.21636845171451569,
+      "learning_rate": 0.00436596391289106,
+      "loss": 1.3585,
+      "step": 7010
+    },
+    {
+      "epoch": 0.740506329113924,
+      "grad_norm": 0.31325456500053406,
+      "learning_rate": 0.004295836986542996,
+      "loss": 1.3725,
+      "step": 7020
+    },
+    {
+      "epoch": 0.7415611814345991,
+      "grad_norm": 0.20215149223804474,
+      "learning_rate": 0.004226836451960217,
+      "loss": 1.377,
+      "step": 7030
+    },
+    {
+      "epoch": 0.7426160337552743,
+      "grad_norm": 0.18876171112060547,
+      "learning_rate": 0.004158944216828193,
+      "loss": 1.3794,
+      "step": 7040
+    },
+    {
+      "epoch": 0.7436708860759493,
+      "grad_norm": 0.18273957073688507,
+      "learning_rate": 0.004092142479434515,
+      "loss": 1.3696,
+      "step": 7050
+    },
+    {
+      "epoch": 0.7447257383966245,
+      "grad_norm": 0.20291049778461456,
+      "learning_rate": 0.0040264137240012,
+      "loss": 1.3683,
+      "step": 7060
+    },
+    {
+      "epoch": 0.7457805907172996,
+      "grad_norm": 0.34383624792099,
+      "learning_rate": 0.003961740716091956,
+      "loss": 1.375,
+      "step": 7070
+    },
+    {
+      "epoch": 0.7468354430379747,
+      "grad_norm": 0.19728510081768036,
+      "learning_rate": 0.0038981064980932218,
+      "loss": 1.373,
+      "step": 7080
+    },
+    {
+      "epoch": 0.7478902953586498,
+      "grad_norm": 0.22290998697280884,
+      "learning_rate": 0.0038354943847677854,
+      "loss": 1.3688,
+      "step": 7090
+    },
+    {
+      "epoch": 0.7489451476793249,
+      "grad_norm": 0.17596077919006348,
+      "learning_rate": 0.0037738879588798254,
+      "loss": 1.3608,
+      "step": 7100
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.18319807946681976,
+      "learning_rate": 0.0037132710668902236,
+      "loss": 1.3726,
+      "step": 7110
+    },
+    {
+      "epoch": 0.7510548523206751,
+      "grad_norm": 0.2474352866411209,
+      "learning_rate": 0.003653627814721017,
+      "loss": 1.3656,
+      "step": 7120
+    },
+    {
+      "epoch": 0.7521097046413502,
+      "grad_norm": 0.26776137948036194,
+      "learning_rate": 0.003594942563587889,
+      "loss": 1.3766,
+      "step": 7130
+    },
+    {
+      "epoch": 0.7531645569620253,
+      "grad_norm": 0.1997850090265274,
+      "learning_rate": 0.0035371999258995905,
+      "loss": 1.3521,
+      "step": 7140
+    },
+    {
+      "epoch": 0.7542194092827004,
+      "grad_norm": 0.19101941585540771,
+      "learning_rate": 0.0034803847612232316,
+      "loss": 1.3482,
+      "step": 7150
+    },
+    {
+      "epoch": 0.7552742616033755,
+      "grad_norm": 0.28819212317466736,
+      "learning_rate": 0.0034244821723143797,
+      "loss": 1.3625,
+      "step": 7160
+    },
+    {
+      "epoch": 0.7563291139240507,
+      "grad_norm": 0.16478969156742096,
+      "learning_rate": 0.0033694775012109175,
+      "loss": 1.366,
+      "step": 7170
+    },
+    {
+      "epoch": 0.7573839662447257,
+      "grad_norm": 0.2116067260503769,
+      "learning_rate": 0.003315356325389651,
+      "loss": 1.3419,
+      "step": 7180
+    },
+    {
+      "epoch": 0.7584388185654009,
+      "grad_norm": 0.190652534365654,
+      "learning_rate": 0.0032621044539846394,
+      "loss": 1.3453,
+      "step": 7190
+    },
+    {
+      "epoch": 0.759493670886076,
+      "grad_norm": 0.18261267244815826,
+      "learning_rate": 0.003209707924066279,
+      "loss": 1.352,
+      "step": 7200
+    },
+    {
+      "epoch": 0.760548523206751,
+      "grad_norm": 0.167609304189682,
+      "learning_rate": 0.003158152996980143,
+      "loss": 1.3531,
+      "step": 7210
+    },
+    {
+      "epoch": 0.7616033755274262,
+      "grad_norm": 0.17341913282871246,
+      "learning_rate": 0.0031074261547446337,
+      "loss": 1.369,
+      "step": 7220
+    },
+    {
+      "epoch": 0.7626582278481012,
+      "grad_norm": 0.18637093901634216,
+      "learning_rate": 0.003057514096506495,
+      "loss": 1.3598,
+      "step": 7230
+    },
+    {
+      "epoch": 0.7637130801687764,
+      "grad_norm": 0.17910732328891754,
+      "learning_rate": 0.0030084037350532546,
+      "loss": 1.3566,
+      "step": 7240
+    },
+    {
+      "epoch": 0.7647679324894515,
+      "grad_norm": 0.2772654592990875,
+      "learning_rate": 0.0029600821933816867,
+      "loss": 1.3491,
+      "step": 7250
+    },
+    {
+      "epoch": 0.7658227848101266,
+      "grad_norm": 0.24979856610298157,
+      "learning_rate": 0.002912536801321393,
+      "loss": 1.3423,
+      "step": 7260
+    },
+    {
+      "epoch": 0.7668776371308017,
+      "grad_norm": 0.21420714259147644,
+      "learning_rate": 0.002865755092212614,
+      "loss": 1.3441,
+      "step": 7270
+    },
+    {
+      "epoch": 0.7679324894514767,
+      "grad_norm": 0.2030237317085266,
+      "learning_rate": 0.0028197247996374027,
+      "loss": 1.3434,
+      "step": 7280
+    },
+    {
+      "epoch": 0.7689873417721519,
+      "grad_norm": 0.20385904610157013,
+      "learning_rate": 0.0027744338542033055,
+      "loss": 1.3534,
+      "step": 7290
+    },
+    {
+      "epoch": 0.770042194092827,
+      "grad_norm": 0.1864974945783615,
+      "learning_rate": 0.002729870380378699,
+      "loss": 1.3572,
+      "step": 7300
+    },
+    {
+      "epoch": 0.7710970464135021,
+      "grad_norm": 0.1724073588848114,
+      "learning_rate": 0.0026860226933789646,
+      "loss": 1.337,
+      "step": 7310
+    },
+    {
+      "epoch": 0.7721518987341772,
+      "grad_norm": 0.2279394418001175,
+      "learning_rate": 0.002642879296102671,
+      "loss": 1.3333,
+      "step": 7320
+    },
+    {
+      "epoch": 0.7732067510548524,
+      "grad_norm": 0.2146916687488556,
+      "learning_rate": 0.0026004288761169747,
+      "loss": 1.3334,
+      "step": 7330
+    },
+    {
+      "epoch": 0.7742616033755274,
+      "grad_norm": 0.1645049899816513,
+      "learning_rate": 0.0025586603026914365,
+      "loss": 1.3311,
+      "step": 7340
+    },
+    {
+      "epoch": 0.7753164556962026,
+      "grad_norm": 0.17457176744937897,
+      "learning_rate": 0.0025175626238794864,
+      "loss": 1.3241,
+      "step": 7350
+    },
+    {
+      "epoch": 0.7763713080168776,
+      "grad_norm": 0.19637472927570343,
+      "learning_rate": 0.00247712506364676,
+      "loss": 1.3344,
+      "step": 7360
+    },
+    {
+      "epoch": 0.7774261603375527,
+      "grad_norm": 0.17304591834545135,
+      "learning_rate": 0.002437337019045568,
+      "loss": 1.3393,
+      "step": 7370
+    },
+    {
+      "epoch": 0.7784810126582279,
+      "grad_norm": 0.1940585970878601,
+      "learning_rate": 0.002398188057434742,
+      "loss": 1.3222,
+      "step": 7380
+    },
+    {
+      "epoch": 0.7795358649789029,
+      "grad_norm": 0.1824662685394287,
+      "learning_rate": 0.002359667913744142,
+      "loss": 1.3329,
+      "step": 7390
+    },
+    {
+      "epoch": 0.7805907172995781,
+      "grad_norm": 0.20115163922309875,
+      "learning_rate": 0.0023217664877830985,
+      "loss": 1.339,
+      "step": 7400
+    },
+    {
+      "epoch": 0.7816455696202531,
+      "grad_norm": 0.19347907602787018,
+      "learning_rate": 0.002284473841592086,
+      "loss": 1.3214,
+      "step": 7410
+    },
+    {
+      "epoch": 0.7827004219409283,
+      "grad_norm": 0.19314011931419373,
+      "learning_rate": 0.0022477801968369404,
+      "loss": 1.3234,
+      "step": 7420
+    },
+    {
+      "epoch": 0.7837552742616034,
+      "grad_norm": 0.172202467918396,
+      "learning_rate": 0.0022116759322449208,
+      "loss": 1.3161,
+      "step": 7430
+    },
+    {
+      "epoch": 0.7848101265822784,
+      "grad_norm": 0.19969259202480316,
+      "learning_rate": 0.0021761515810819667,
+      "loss": 1.314,
+      "step": 7440
+    },
+    {
+      "epoch": 0.7858649789029536,
+      "grad_norm": 0.17586781084537506,
+      "learning_rate": 0.002141197828670461,
+      "loss": 1.3166,
+      "step": 7450
+    },
+    {
+      "epoch": 0.7869198312236287,
+      "grad_norm": 0.22718460857868195,
+      "learning_rate": 0.0021068055099468775,
+      "loss": 1.3213,
+      "step": 7460
+    },
+    {
+      "epoch": 0.7879746835443038,
+      "grad_norm": 0.18798670172691345,
+      "learning_rate": 0.0020729656070586485,
+      "loss": 1.3172,
+      "step": 7470
+    },
+    {
+      "epoch": 0.7890295358649789,
+      "grad_norm": 0.21208590269088745,
+      "learning_rate": 0.0020396692469996365,
+      "loss": 1.3216,
+      "step": 7480
+    },
+    {
+      "epoch": 0.790084388185654,
+      "grad_norm": 0.20185668766498566,
+      "learning_rate": 0.0020069076992835807,
+      "loss": 1.3314,
+      "step": 7490
+    },
+    {
+      "epoch": 0.7911392405063291,
+      "grad_norm": 0.16954679787158966,
+      "learning_rate": 0.0019746723736549193,
+      "loss": 1.324,
+      "step": 7500
+    },
+    {
+      "epoch": 0.7921940928270043,
+      "grad_norm": 0.24429215490818024,
+      "learning_rate": 0.0019429548178363783,
+      "loss": 1.3262,
+      "step": 7510
+    },
+    {
+      "epoch": 0.7932489451476793,
+      "grad_norm": 0.1750379055738449,
+      "learning_rate": 0.0019117467153127346,
+      "loss": 1.3119,
+      "step": 7520
+    },
+    {
+      "epoch": 0.7943037974683544,
+      "grad_norm": 0.22609379887580872,
+      "learning_rate": 0.0018810398831501853,
+      "loss": 1.3126,
+      "step": 7530
+    },
+    {
+      "epoch": 0.7953586497890295,
+      "grad_norm": 0.17649514973163605,
+      "learning_rate": 0.0018508262698507343,
+      "loss": 1.3097,
+      "step": 7540
+    },
+    {
+      "epoch": 0.7964135021097046,
+      "grad_norm": 0.17880935966968536,
+      "learning_rate": 0.0018210979532410488,
+      "loss": 1.3094,
+      "step": 7550
+    },
+    {
+      "epoch": 0.7974683544303798,
+      "grad_norm": 0.16698552668094635,
+      "learning_rate": 0.0017918471383952198,
+      "loss": 1.3119,
+      "step": 7560
+    },
+    {
+      "epoch": 0.7985232067510548,
+      "grad_norm": 0.21640335023403168,
+      "learning_rate": 0.0017630661555908913,
+      "loss": 1.3169,
+      "step": 7570
+    },
+    {
+      "epoch": 0.79957805907173,
+      "grad_norm": 0.21301251649856567,
+      "learning_rate": 0.00173474745829822,
+      "loss": 1.3133,
+      "step": 7580
+    },
+    {
+      "epoch": 0.8006329113924051,
+      "grad_norm": 0.1886330395936966,
+      "learning_rate": 0.0017068836212011287,
+      "loss": 1.3055,
+      "step": 7590
+    },
+    {
+      "epoch": 0.8016877637130801,
+      "grad_norm": 0.16271863877773285,
+      "learning_rate": 0.0016794673382503544,
+      "loss": 1.3114,
+      "step": 7600
+    },
+    {
+      "epoch": 0.8027426160337553,
+      "grad_norm": 0.1735699623823166,
+      "learning_rate": 0.0016524914207477572,
+      "loss": 1.3206,
+      "step": 7610
+    },
+    {
+      "epoch": 0.8037974683544303,
+      "grad_norm": 0.17835357785224915,
+      "learning_rate": 0.0016259487954614084,
+      "loss": 1.2931,
+      "step": 7620
+    },
+    {
+      "epoch": 0.8048523206751055,
+      "grad_norm": 0.17388689517974854,
+      "learning_rate": 0.0015998325027709484,
+      "loss": 1.3161,
+      "step": 7630
+    },
+    {
+      "epoch": 0.8059071729957806,
+      "grad_norm": 0.2061561942100525,
+      "learning_rate": 0.001574135694842738,
+      "loss": 1.3069,
+      "step": 7640
+    },
+    {
+      "epoch": 0.8069620253164557,
+      "grad_norm": 0.17245610058307648,
+      "learning_rate": 0.0015488516338343179,
+      "loss": 1.3245,
+      "step": 7650
+    },
+    {
+      "epoch": 0.8080168776371308,
+      "grad_norm": 0.16928501427173615,
+      "learning_rate": 0.0015239736901277107,
+      "loss": 1.2992,
+      "step": 7660
+    },
+    {
+      "epoch": 0.8090717299578059,
+      "grad_norm": 0.18795177340507507,
+      "learning_rate": 0.0014994953405911002,
+      "loss": 1.3039,
+      "step": 7670
+    },
+    {
+      "epoch": 0.810126582278481,
+      "grad_norm": 0.18095934391021729,
+      "learning_rate": 0.0014754101668684274,
+      "loss": 1.3212,
+      "step": 7680
+    },
+    {
+      "epoch": 0.8111814345991561,
+      "grad_norm": 0.17277932167053223,
+      "learning_rate": 0.0014517118536964673,
+      "loss": 1.2918,
+      "step": 7690
+    },
+    {
+      "epoch": 0.8122362869198312,
+      "grad_norm": 0.1742982417345047,
+      "learning_rate": 0.0014283941872489279,
+      "loss": 1.3022,
+      "step": 7700
+    },
+    {
+      "epoch": 0.8132911392405063,
+      "grad_norm": 0.19824908673763275,
+      "learning_rate": 0.0014054510535071556,
+      "loss": 1.2954,
+      "step": 7710
+    },
+    {
+      "epoch": 0.8143459915611815,
+      "grad_norm": 0.16659264266490936,
+      "learning_rate": 0.001382876436657003,
+      "loss": 1.3025,
+      "step": 7720
+    },
+    {
+      "epoch": 0.8154008438818565,
+      "grad_norm": 0.18909379839897156,
+      "learning_rate": 0.001360664417511452,
+      "loss": 1.3104,
+      "step": 7730
+    },
+    {
+      "epoch": 0.8164556962025317,
+      "grad_norm": 0.16384479403495789,
+      "learning_rate": 0.0013388091719585694,
+      "loss": 1.3107,
+      "step": 7740
+    },
+    {
+      "epoch": 0.8175105485232067,
+      "grad_norm": 0.1827823668718338,
+      "learning_rate": 0.001317304969434393,
+      "loss": 1.2931,
+      "step": 7750
+    },
+    {
+      "epoch": 0.8185654008438819,
+      "grad_norm": 0.188125342130661,
+      "learning_rate": 0.001296146171420349,
+      "loss": 1.2918,
+      "step": 7760
+    },
+    {
+      "epoch": 0.819620253164557,
+      "grad_norm": 0.18765294551849365,
+      "learning_rate": 0.001275327229964799,
+      "loss": 1.3005,
+      "step": 7770
+    },
+    {
+      "epoch": 0.820675105485232,
+      "grad_norm": 0.1771213710308075,
+      "learning_rate": 0.0012548426862283388,
+      "loss": 1.2925,
+      "step": 7780
+    },
+    {
+      "epoch": 0.8217299578059072,
+      "grad_norm": 0.1866278499364853,
+      "learning_rate": 0.0012346871690524595,
+      "loss": 1.3105,
+      "step": 7790
+    },
+    {
+      "epoch": 0.8227848101265823,
+      "grad_norm": 0.1600043773651123,
+      "learning_rate": 0.001214855393551203,
+      "loss": 1.2987,
+      "step": 7800
+    },
+    {
+      "epoch": 0.8238396624472574,
+      "grad_norm": 0.20158807933330536,
+      "learning_rate": 0.0011953421597254332,
+      "loss": 1.2991,
+      "step": 7810
+    },
+    {
+      "epoch": 0.8248945147679325,
+      "grad_norm": 0.1783669888973236,
+      "learning_rate": 0.0011761423510993707,
+      "loss": 1.3057,
+      "step": 7820
+    },
+    {
+      "epoch": 0.8259493670886076,
+      "grad_norm": 0.170523002743721,
+      "learning_rate": 0.0011572509333790235,
+      "loss": 1.2931,
+      "step": 7830
+    },
+    {
+      "epoch": 0.8270042194092827,
+      "grad_norm": 0.1701860874891281,
+      "learning_rate": 0.0011386629531321682,
+      "loss": 1.2788,
+      "step": 7840
+    },
+    {
+      "epoch": 0.8280590717299579,
+      "grad_norm": 0.19231587648391724,
+      "learning_rate": 0.001120373536489534,
+      "loss": 1.2856,
+      "step": 7850
+    },
+    {
+      "epoch": 0.8291139240506329,
+      "grad_norm": 0.17149046063423157,
+      "learning_rate": 0.0011023778878668459,
+      "loss": 1.274,
+      "step": 7860
+    },
+    {
+      "epoch": 0.830168776371308,
+      "grad_norm": 0.18005429208278656,
+      "learning_rate": 0.0010846712887073984,
+      "loss": 1.2925,
+      "step": 7870
+    },
+    {
+      "epoch": 0.8312236286919831,
+      "grad_norm": 0.17713883519172668,
+      "learning_rate": 0.0010672490962448232,
+      "loss": 1.3087,
+      "step": 7880
+    },
+    {
+      "epoch": 0.8322784810126582,
+      "grad_norm": 0.2106132060289383,
+      "learning_rate": 0.0010501067422857312,
+      "loss": 1.3093,
+      "step": 7890
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.18951666355133057,
+      "learning_rate": 0.0010332397320119073,
+      "loss": 1.2817,
+      "step": 7900
+    },
+    {
+      "epoch": 0.8343881856540084,
+      "grad_norm": 0.20073255896568298,
+      "learning_rate": 0.0010166436428017435,
+      "loss": 1.2949,
+      "step": 7910
+    },
+    {
+      "epoch": 0.8354430379746836,
+      "grad_norm": 0.16490823030471802,
+      "learning_rate": 0.0010003141230706066,
+      "loss": 1.2915,
+      "step": 7920
+    },
+    {
+      "epoch": 0.8364978902953587,
+      "grad_norm": 0.17728537321090698,
+      "learning_rate": 0.000984246891129825,
+      "loss": 1.2997,
+      "step": 7930
+    },
+    {
+      "epoch": 0.8375527426160337,
+      "grad_norm": 0.2047002613544464,
+      "learning_rate": 0.0009684377340640103,
+      "loss": 1.2933,
+      "step": 7940
+    },
+    {
+      "epoch": 0.8386075949367089,
+      "grad_norm": 0.17677269876003265,
+      "learning_rate": 0.0009528825066264054,
+      "loss": 1.2968,
+      "step": 7950
+    },
+    {
+      "epoch": 0.8396624472573839,
+      "grad_norm": 0.17389684915542603,
+      "learning_rate": 0.0009375771301519808,
+      "loss": 1.2822,
+      "step": 7960
+    },
+    {
+      "epoch": 0.8407172995780591,
+      "grad_norm": 0.16812482476234436,
+      "learning_rate": 0.0009225175914879836,
+      "loss": 1.2802,
+      "step": 7970
+    },
+    {
+      "epoch": 0.8417721518987342,
+      "grad_norm": 0.17808088660240173,
+      "learning_rate": 0.0009076999419416697,
+      "loss": 1.2796,
+      "step": 7980
+    },
+    {
+      "epoch": 0.8428270042194093,
+      "grad_norm": 0.17545445263385773,
+      "learning_rate": 0.0008931202962449337,
+      "loss": 1.289,
+      "step": 7990
+    },
+    {
+      "epoch": 0.8438818565400844,
+      "grad_norm": 0.1796313226222992,
+      "learning_rate": 0.0008787748315355705,
+      "loss": 1.2886,
+      "step": 8000
+    },
+    {
+      "epoch": 0.8449367088607594,
+      "grad_norm": 0.17771579325199127,
+      "learning_rate": 0.0008646597863549013,
+      "loss": 1.2929,
+      "step": 8010
+    },
+    {
+      "epoch": 0.8459915611814346,
+      "grad_norm": 0.17319394648075104,
+      "learning_rate": 0.000850771459661497,
+      "loss": 1.2912,
+      "step": 8020
+    },
+    {
+      "epoch": 0.8470464135021097,
+      "grad_norm": 0.17196668684482574,
+      "learning_rate": 0.0008371062098607465,
+      "loss": 1.281,
+      "step": 8030
+    },
+    {
+      "epoch": 0.8481012658227848,
+      "grad_norm": 0.19135834276676178,
+      "learning_rate": 0.00082366045385001,
+      "loss": 1.2957,
+      "step": 8040
+    },
+    {
+      "epoch": 0.8491561181434599,
+      "grad_norm": 0.16833963990211487,
+      "learning_rate": 0.0008104306660791106,
+      "loss": 1.2745,
+      "step": 8050
+    },
+    {
+      "epoch": 0.8502109704641351,
+      "grad_norm": 0.16896310448646545,
+      "learning_rate": 0.0007974133776259154,
+      "loss": 1.2866,
+      "step": 8060
+    },
+    {
+      "epoch": 0.8512658227848101,
+      "grad_norm": 0.16317786276340485,
+      "learning_rate": 0.0007846051752867657,
+      "loss": 1.2862,
+      "step": 8070
+    },
+    {
+      "epoch": 0.8523206751054853,
+      "grad_norm": 0.18009500205516815,
+      "learning_rate": 0.0007720027006815159,
+      "loss": 1.2921,
+      "step": 8080
+    },
+    {
+      "epoch": 0.8533755274261603,
+      "grad_norm": 0.15803638100624084,
+      "learning_rate": 0.0007596026493729489,
+      "loss": 1.2808,
+      "step": 8090
+    },
+    {
+      "epoch": 0.8544303797468354,
+      "grad_norm": 0.16284511983394623,
+      "learning_rate": 0.0007474017700003342,
+      "loss": 1.2744,
+      "step": 8100
+    },
+    {
+      "epoch": 0.8554852320675106,
+      "grad_norm": 0.17715783417224884,
+      "learning_rate": 0.0007353968634269036,
+      "loss": 1.2922,
+      "step": 8110
+    },
+    {
+      "epoch": 0.8565400843881856,
+      "grad_norm": 0.19103586673736572,
+      "learning_rate": 0.0007235847819010194,
+      "loss": 1.2794,
+      "step": 8120
+    },
+    {
+      "epoch": 0.8575949367088608,
+      "grad_norm": 0.19346380233764648,
+      "learning_rate": 0.0007119624282308183,
+      "loss": 1.2793,
+      "step": 8130
+    },
+    {
+      "epoch": 0.8586497890295358,
+      "grad_norm": 0.17734567821025848,
+      "learning_rate": 0.0007005267549721101,
+      "loss": 1.2782,
+      "step": 8140
+    },
+    {
+      "epoch": 0.859704641350211,
+      "grad_norm": 0.16510824859142303,
+      "learning_rate": 0.0006892747636293215,
+      "loss": 1.2734,
+      "step": 8150
+    },
+    {
+      "epoch": 0.8607594936708861,
+      "grad_norm": 0.18282318115234375,
+      "learning_rate": 0.0006782035038692731,
+      "loss": 1.2859,
+      "step": 8160
+    },
+    {
+      "epoch": 0.8618143459915611,
+      "grad_norm": 0.15206333994865417,
+      "learning_rate": 0.0006673100727475881,
+      "loss": 1.2734,
+      "step": 8170
+    },
+    {
+      "epoch": 0.8628691983122363,
+      "grad_norm": 0.1695568561553955,
+      "learning_rate": 0.0006565916139475235,
+      "loss": 1.2829,
+      "step": 8180
+    },
+    {
+      "epoch": 0.8639240506329114,
+      "grad_norm": 0.17288580536842346,
+      "learning_rate": 0.0006460453170310274,
+      "loss": 1.273,
+      "step": 8190
+    },
+    {
+      "epoch": 0.8649789029535865,
+      "grad_norm": 0.17699413001537323,
+      "learning_rate": 0.0006356684167018288,
+      "loss": 1.2873,
+      "step": 8200
+    },
+    {
+      "epoch": 0.8660337552742616,
+      "grad_norm": 0.16934652626514435,
+      "learning_rate": 0.0006254581920803608,
+      "loss": 1.2796,
+      "step": 8210
+    },
+    {
+      "epoch": 0.8670886075949367,
+      "grad_norm": 0.16221648454666138,
+      "learning_rate": 0.0006154119659903314,
+      "loss": 1.2879,
+      "step": 8220
+    },
+    {
+      "epoch": 0.8681434599156118,
+      "grad_norm": 0.21882303059101105,
+      "learning_rate": 0.0006055271042567527,
+      "loss": 1.2844,
+      "step": 8230
+    },
+    {
+      "epoch": 0.869198312236287,
+      "grad_norm": 0.16958904266357422,
+      "learning_rate": 0.0005958010150152469,
+      "loss": 1.2822,
+      "step": 8240
+    },
+    {
+      "epoch": 0.870253164556962,
+      "grad_norm": 0.18374933302402496,
+      "learning_rate": 0.0005862311480324461,
+      "loss": 1.2962,
+      "step": 8250
+    },
+    {
+      "epoch": 0.8713080168776371,
+      "grad_norm": 0.17241325974464417,
+      "learning_rate": 0.0005768149940373044,
+      "loss": 1.283,
+      "step": 8260
+    },
+    {
+      "epoch": 0.8723628691983122,
+      "grad_norm": 0.16826824843883514,
+      "learning_rate": 0.0005675500840631565,
+      "loss": 1.2791,
+      "step": 8270
+    },
+    {
+      "epoch": 0.8734177215189873,
+      "grad_norm": 0.16674010455608368,
+      "learning_rate": 0.0005584339888003395,
+      "loss": 1.2646,
+      "step": 8280
+    },
+    {
+      "epoch": 0.8744725738396625,
+      "grad_norm": 0.17145630717277527,
+      "learning_rate": 0.0005494643179592156,
+      "loss": 1.2649,
+      "step": 8290
+    },
+    {
+      "epoch": 0.8755274261603375,
+      "grad_norm": 0.1749577522277832,
+      "learning_rate": 0.0005406387196434245,
+      "loss": 1.2888,
+      "step": 8300
+    },
+    {
+      "epoch": 0.8765822784810127,
+      "grad_norm": 0.17016993463039398,
+      "learning_rate": 0.0005319548797332037,
+      "loss": 1.2654,
+      "step": 8310
+    },
+    {
+      "epoch": 0.8776371308016878,
+      "grad_norm": 0.2058466225862503,
+      "learning_rate": 0.0005234105212786141,
+      "loss": 1.2868,
+      "step": 8320
+    },
+    {
+      "epoch": 0.8786919831223629,
+      "grad_norm": 0.1731146275997162,
+      "learning_rate": 0.0005150034039025102,
+      "loss": 1.2796,
+      "step": 8330
+    },
+    {
+      "epoch": 0.879746835443038,
+      "grad_norm": 0.16560529172420502,
+      "learning_rate": 0.000506731323213103,
+      "loss": 1.2673,
+      "step": 8340
+    },
+    {
+      "epoch": 0.880801687763713,
+      "grad_norm": 0.16997076570987701,
+      "learning_rate": 0.0004985921102259548,
+      "loss": 1.2692,
+      "step": 8350
+    },
+    {
+      "epoch": 0.8818565400843882,
+      "grad_norm": 0.16966533660888672,
+      "learning_rate": 0.0004905836307952605,
+      "loss": 1.2687,
+      "step": 8360
+    },
+    {
+      "epoch": 0.8829113924050633,
+      "grad_norm": 0.17669892311096191,
+      "learning_rate": 0.00048270378505426234,
+      "loss": 1.2744,
+      "step": 8370
+    },
+    {
+      "epoch": 0.8839662447257384,
+      "grad_norm": 0.16950394213199615,
+      "learning_rate": 0.00047495050686465444,
+      "loss": 1.2822,
+      "step": 8380
+    },
+    {
+      "epoch": 0.8850210970464135,
+      "grad_norm": 0.1622672975063324,
+      "learning_rate": 0.00046732176327482955,
+      "loss": 1.2711,
+      "step": 8390
+    },
+    {
+      "epoch": 0.8860759493670886,
+      "grad_norm": 0.17885705828666687,
+      "learning_rate": 0.00045981555398682823,
+      "loss": 1.2716,
+      "step": 8400
+    },
+    {
+      "epoch": 0.8871308016877637,
+      "grad_norm": 0.17793214321136475,
+      "learning_rate": 0.0004524299108318491,
+      "loss": 1.2885,
+      "step": 8410
+    },
+    {
+      "epoch": 0.8881856540084389,
+      "grad_norm": 0.179661825299263,
+      "learning_rate": 0.00044516289725418575,
+      "loss": 1.2779,
+      "step": 8420
+    },
+    {
+      "epoch": 0.8892405063291139,
+      "grad_norm": 0.17629870772361755,
+      "learning_rate": 0.00043801260780345063,
+      "loss": 1.275,
+      "step": 8430
+    },
+    {
+      "epoch": 0.890295358649789,
+      "grad_norm": 0.15668673813343048,
+      "learning_rate": 0.0004309771676349548,
+      "loss": 1.2678,
+      "step": 8440
+    },
+    {
+      "epoch": 0.8913502109704642,
+      "grad_norm": 0.1750100702047348,
+      "learning_rate": 0.00042405473201811484,
+      "loss": 1.2734,
+      "step": 8450
+    },
+    {
+      "epoch": 0.8924050632911392,
+      "grad_norm": 0.16259723901748657,
+      "learning_rate": 0.0004172434858527539,
+      "loss": 1.2683,
+      "step": 8460
+    },
+    {
+      "epoch": 0.8934599156118144,
+      "grad_norm": 0.1615871787071228,
+      "learning_rate": 0.0004105416431931726,
+      "loss": 1.2721,
+      "step": 8470
+    },
+    {
+      "epoch": 0.8945147679324894,
+      "grad_norm": 0.18613623082637787,
+      "learning_rate": 0.00040394744677986385,
+      "loss": 1.2645,
+      "step": 8480
+    },
+    {
+      "epoch": 0.8955696202531646,
+      "grad_norm": 0.15994839370250702,
+      "learning_rate": 0.00039745916757875086,
+      "loss": 1.2734,
+      "step": 8490
+    },
+    {
+      "epoch": 0.8966244725738397,
+      "grad_norm": 0.1675361841917038,
+      "learning_rate": 0.00039107510432782454,
+      "loss": 1.259,
+      "step": 8500
+    },
+    {
+      "epoch": 0.8976793248945147,
+      "grad_norm": 0.1560775637626648,
+      "learning_rate": 0.000384793583091063,
+      "loss": 1.2725,
+      "step": 8510
+    },
+    {
+      "epoch": 0.8987341772151899,
+      "grad_norm": 0.16288244724273682,
+      "learning_rate": 0.00037861295681951713,
+      "loss": 1.2665,
+      "step": 8520
+    },
+    {
+      "epoch": 0.8997890295358649,
+      "grad_norm": 0.19567705690860748,
+      "learning_rate": 0.0003725316049194451,
+      "loss": 1.282,
+      "step": 8530
+    },
+    {
+      "epoch": 0.9008438818565401,
+      "grad_norm": 0.16067910194396973,
+      "learning_rate": 0.00036654793282738397,
+      "loss": 1.2745,
+      "step": 8540
+    },
+    {
+      "epoch": 0.9018987341772152,
+      "grad_norm": 0.16617079079151154,
+      "learning_rate": 0.00036066037159204593,
+      "loss": 1.2686,
+      "step": 8550
+    },
+    {
+      "epoch": 0.9029535864978903,
+      "grad_norm": 0.17245526611804962,
+      "learning_rate": 0.0003548673774629319,
+      "loss": 1.2629,
+      "step": 8560
+    },
+    {
+      "epoch": 0.9040084388185654,
+      "grad_norm": 0.16444915533065796,
+      "learning_rate": 0.00034916743148555096,
+      "loss": 1.2817,
+      "step": 8570
+    },
+    {
+      "epoch": 0.9050632911392406,
+      "grad_norm": 0.17688319087028503,
+      "learning_rate": 0.00034355903910314196,
+      "loss": 1.2686,
+      "step": 8580
+    },
+    {
+      "epoch": 0.9061181434599156,
+      "grad_norm": 0.1684081256389618,
+      "learning_rate": 0.00033804072976479363,
+      "loss": 1.2746,
+      "step": 8590
+    },
+    {
+      "epoch": 0.9071729957805907,
+      "grad_norm": 0.15983375906944275,
+      "learning_rate": 0.0003326110565398574,
+      "loss": 1.2685,
+      "step": 8600
+    },
+    {
+      "epoch": 0.9082278481012658,
+      "grad_norm": 0.17392690479755402,
+      "learning_rate": 0.0003272685957385547,
+      "loss": 1.2688,
+      "step": 8610
+    },
+    {
+      "epoch": 0.9092827004219409,
+      "grad_norm": 0.1693653017282486,
+      "learning_rate": 0.0003220119465386773,
+      "loss": 1.2669,
+      "step": 8620
+    },
+    {
+      "epoch": 0.9103375527426161,
+      "grad_norm": 0.17211997509002686,
+      "learning_rate": 0.0003168397306182845,
+      "loss": 1.2765,
+      "step": 8630
+    },
+    {
+      "epoch": 0.9113924050632911,
+      "grad_norm": 0.17304740846157074,
+      "learning_rate": 0.0003117505917942999,
+      "loss": 1.2565,
+      "step": 8640
+    },
+    {
+      "epoch": 0.9124472573839663,
+      "grad_norm": 0.17685651779174805,
+      "learning_rate": 0.00030674319566691236,
+      "loss": 1.2651,
+      "step": 8650
+    },
+    {
+      "epoch": 0.9135021097046413,
+      "grad_norm": 0.16831658780574799,
+      "learning_rate": 0.000301816229269689,
+      "loss": 1.2619,
+      "step": 8660
+    },
+    {
+      "epoch": 0.9145569620253164,
+      "grad_norm": 0.15462885797023773,
+      "learning_rate": 0.0002969684007253089,
+      "loss": 1.2709,
+      "step": 8670
+    },
+    {
+      "epoch": 0.9156118143459916,
+      "grad_norm": 0.16284406185150146,
+      "learning_rate": 0.0002921984389068256,
+      "loss": 1.2745,
+      "step": 8680
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 0.18605948984622955,
+      "learning_rate": 0.0002875050931043702,
+      "loss": 1.2698,
+      "step": 8690
+    },
+    {
+      "epoch": 0.9177215189873418,
+      "grad_norm": 0.1646358221769333,
+      "learning_rate": 0.0002828871326972093,
+      "loss": 1.2702,
+      "step": 8700
+    },
+    {
+      "epoch": 0.9187763713080169,
+      "grad_norm": 0.18065601587295532,
+      "learning_rate": 0.00027834334683106903,
+      "loss": 1.2646,
+      "step": 8710
+    },
+    {
+      "epoch": 0.919831223628692,
+      "grad_norm": 0.1509815901517868,
+      "learning_rate": 0.0002738725441006427,
+      "loss": 1.261,
+      "step": 8720
+    },
+    {
+      "epoch": 0.9208860759493671,
+      "grad_norm": 0.1648000329732895,
+      "learning_rate": 0.00026947355223719743,
+      "loss": 1.2783,
+      "step": 8730
+    },
+    {
+      "epoch": 0.9219409282700421,
+      "grad_norm": 0.17343315482139587,
+      "learning_rate": 0.0002651452178011997,
+      "loss": 1.2619,
+      "step": 8740
+    },
+    {
+      "epoch": 0.9229957805907173,
+      "grad_norm": 0.1724059134721756,
+      "learning_rate": 0.0002608864058798768,
+      "loss": 1.2601,
+      "step": 8750
+    },
+    {
+      "epoch": 0.9240506329113924,
+      "grad_norm": 0.17463988065719604,
+      "learning_rate": 0.00025669599978963604,
+      "loss": 1.2462,
+      "step": 8760
+    },
+    {
+      "epoch": 0.9251054852320675,
+      "grad_norm": 0.15516114234924316,
+      "learning_rate": 0.00025257290078326537,
+      "loss": 1.259,
+      "step": 8770
+    },
+    {
+      "epoch": 0.9261603375527426,
+      "grad_norm": 0.16971370577812195,
+      "learning_rate": 0.0002485160277618352,
+      "loss": 1.266,
+      "step": 8780
+    },
+    {
+      "epoch": 0.9272151898734177,
+      "grad_norm": 0.16363847255706787,
+      "learning_rate": 0.00024452431699122844,
+      "loss": 1.2611,
+      "step": 8790
+    },
+    {
+      "epoch": 0.9282700421940928,
+      "grad_norm": 0.18701323866844177,
+      "learning_rate": 0.00024059672182322344,
+      "loss": 1.2563,
+      "step": 8800
+    },
+    {
+      "epoch": 0.929324894514768,
+      "grad_norm": 0.16498000919818878,
+      "learning_rate": 0.00023673221242105774,
+      "loss": 1.2809,
+      "step": 8810
+    },
+    {
+      "epoch": 0.930379746835443,
+      "grad_norm": 0.16793279349803925,
+      "learning_rate": 0.00023292977548939912,
+      "loss": 1.2653,
+      "step": 8820
+    },
+    {
+      "epoch": 0.9314345991561181,
+      "grad_norm": 0.162998765707016,
+      "learning_rate": 0.0002291884140086533,
+      "loss": 1.2778,
+      "step": 8830
+    },
+    {
+      "epoch": 0.9324894514767933,
+      "grad_norm": 0.1537642627954483,
+      "learning_rate": 0.0002255071469735412,
+      "loss": 1.2628,
+      "step": 8840
+    },
+    {
+      "epoch": 0.9335443037974683,
+      "grad_norm": 0.16959550976753235,
+      "learning_rate": 0.00022188500913587308,
+      "loss": 1.2638,
+      "step": 8850
+    },
+    {
+      "epoch": 0.9345991561181435,
+      "grad_norm": 0.16118808090686798,
+      "learning_rate": 0.0002183210507514559,
+      "loss": 1.2641,
+      "step": 8860
+    },
+    {
+      "epoch": 0.9356540084388185,
+      "grad_norm": 0.15798310935497284,
+      "learning_rate": 0.00021481433733106444,
+      "loss": 1.2702,
+      "step": 8870
+    },
+    {
+      "epoch": 0.9367088607594937,
+      "grad_norm": 0.16716496646404266,
+      "learning_rate": 0.00021136394939541407,
+      "loss": 1.2704,
+      "step": 8880
+    },
+    {
+      "epoch": 0.9377637130801688,
+      "grad_norm": 0.16903844475746155,
+      "learning_rate": 0.00020796898223406765,
+      "loss": 1.2655,
+      "step": 8890
+    },
+    {
+      "epoch": 0.9388185654008439,
+      "grad_norm": 0.1659862995147705,
+      "learning_rate": 0.00020462854566821593,
+      "loss": 1.2555,
+      "step": 8900
+    },
+    {
+      "epoch": 0.939873417721519,
+      "grad_norm": 0.17785194516181946,
+      "learning_rate": 0.00020134176381726728,
+      "loss": 1.2593,
+      "step": 8910
+    },
+    {
+      "epoch": 0.9409282700421941,
+      "grad_norm": 0.15679247677326202,
+      "learning_rate": 0.00019810777486918785,
+      "loss": 1.2677,
+      "step": 8920
+    },
+    {
+      "epoch": 0.9419831223628692,
+      "grad_norm": 0.15910658240318298,
+      "learning_rate": 0.00019492573085452925,
+      "loss": 1.2597,
+      "step": 8930
+    },
+    {
+      "epoch": 0.9430379746835443,
+      "grad_norm": 0.15889185667037964,
+      "learning_rate": 0.0001917947974240862,
+      "loss": 1.2609,
+      "step": 8940
+    },
+    {
+      "epoch": 0.9440928270042194,
+      "grad_norm": 0.15979579091072083,
+      "learning_rate": 0.00018871415363012618,
+      "loss": 1.2642,
+      "step": 8950
+    },
+    {
+      "epoch": 0.9451476793248945,
+      "grad_norm": 0.16714482009410858,
+      "learning_rate": 0.0001856829917111321,
+      "loss": 1.2646,
+      "step": 8960
+    },
+    {
+      "epoch": 0.9462025316455697,
+      "grad_norm": 0.16492030024528503,
+      "learning_rate": 0.0001827005168800032,
+      "loss": 1.2673,
+      "step": 8970
+    },
+    {
+      "epoch": 0.9472573839662447,
+      "grad_norm": 0.1714344322681427,
+      "learning_rate": 0.0001797659471156568,
+      "loss": 1.2596,
+      "step": 8980
+    },
+    {
+      "epoch": 0.9483122362869199,
+      "grad_norm": 0.16298818588256836,
+      "learning_rate": 0.000176878512957979,
+      "loss": 1.2689,
+      "step": 8990
+    },
+    {
+      "epoch": 0.9493670886075949,
+      "grad_norm": 0.1708623617887497,
+      "learning_rate": 0.00017403745730606784,
+      "loss": 1.2569,
+      "step": 9000
+    },
+    {
+      "epoch": 0.95042194092827,
+      "grad_norm": 0.15453357994556427,
+      "learning_rate": 0.00017124203521971676,
+      "loss": 1.2549,
+      "step": 9010
+    },
+    {
+      "epoch": 0.9514767932489452,
+      "grad_norm": 0.16443729400634766,
+      "learning_rate": 0.00016849151372408806,
+      "loss": 1.2623,
+      "step": 9020
+    },
+    {
+      "epoch": 0.9525316455696202,
+      "grad_norm": 0.17994344234466553,
+      "learning_rate": 0.00016578517161752238,
+      "loss": 1.2821,
+      "step": 9030
+    },
+    {
+      "epoch": 0.9535864978902954,
+      "grad_norm": 0.17321337759494781,
+      "learning_rate": 0.00016312229928243597,
+      "loss": 1.2548,
+      "step": 9040
+    },
+    {
+      "epoch": 0.9546413502109705,
+      "grad_norm": 0.15944181382656097,
+      "learning_rate": 0.00016050219849925477,
+      "loss": 1.2668,
+      "step": 9050
+    },
+    {
+      "epoch": 0.9556962025316456,
+      "grad_norm": 0.15873098373413086,
+      "learning_rate": 0.00015792418226333795,
+      "loss": 1.2675,
+      "step": 9060
+    },
+    {
+      "epoch": 0.9567510548523207,
+      "grad_norm": 0.1618490368127823,
+      "learning_rate": 0.00015538757460484135,
+      "loss": 1.2679,
+      "step": 9070
+    },
+    {
+      "epoch": 0.9578059071729957,
+      "grad_norm": 0.1524222195148468,
+      "learning_rate": 0.00015289171041147406,
+      "loss": 1.2692,
+      "step": 9080
+    },
+    {
+      "epoch": 0.9588607594936709,
+      "grad_norm": 0.16087470948696136,
+      "learning_rate": 0.00015043593525410325,
+      "loss": 1.2663,
+      "step": 9090
+    },
+    {
+      "epoch": 0.959915611814346,
+      "grad_norm": 0.16378764808177948,
+      "learning_rate": 0.00014801960521515858,
+      "loss": 1.2443,
+      "step": 9100
+    },
+    {
+      "epoch": 0.9609704641350211,
+      "grad_norm": 0.16996154189109802,
+      "learning_rate": 0.00014564208671979397,
+      "loss": 1.2603,
+      "step": 9110
+    },
+    {
+      "epoch": 0.9620253164556962,
+      "grad_norm": 0.15371821820735931,
+      "learning_rate": 0.0001433027563697603,
+      "loss": 1.2662,
+      "step": 9120
+    },
+    {
+      "epoch": 0.9630801687763713,
+      "grad_norm": 0.16581374406814575,
+      "learning_rate": 0.0001410010007799477,
+      "loss": 1.2529,
+      "step": 9130
+    },
+    {
+      "epoch": 0.9641350210970464,
+      "grad_norm": 0.15591289103031158,
+      "learning_rate": 0.00013873621641755213,
+      "loss": 1.2705,
+      "step": 9140
+    },
+    {
+      "epoch": 0.9651898734177216,
+      "grad_norm": 0.1727791130542755,
+      "learning_rate": 0.00013650780944382608,
+      "loss": 1.2613,
+      "step": 9150
+    },
+    {
+      "epoch": 0.9662447257383966,
+      "grad_norm": 0.15898442268371582,
+      "learning_rate": 0.00013431519555837045,
+      "loss": 1.2676,
+      "step": 9160
+    },
+    {
+      "epoch": 0.9672995780590717,
+      "grad_norm": 0.15596693754196167,
+      "learning_rate": 0.00013215779984592838,
+      "loss": 1.2633,
+      "step": 9170
+    },
+    {
+      "epoch": 0.9683544303797469,
+      "grad_norm": 0.15965871512889862,
+      "learning_rate": 0.00013003505662563896,
+      "loss": 1.2602,
+      "step": 9180
+    },
+    {
+      "epoch": 0.9694092827004219,
+      "grad_norm": 0.163784921169281,
+      "learning_rate": 0.00012794640930271264,
+      "loss": 1.2586,
+      "step": 9190
+    },
+    {
+      "epoch": 0.9704641350210971,
+      "grad_norm": 0.18262465298175812,
+      "learning_rate": 0.00012589131022248926,
+      "loss": 1.2601,
+      "step": 9200
+    },
+    {
+      "epoch": 0.9715189873417721,
+      "grad_norm": 0.1635935753583908,
+      "learning_rate": 0.00012386922052683987,
+      "loss": 1.2619,
+      "step": 9210
+    },
+    {
+      "epoch": 0.9725738396624473,
+      "grad_norm": 0.1602746993303299,
+      "learning_rate": 0.00012187961001287533,
+      "loss": 1.2565,
+      "step": 9220
+    },
+    {
+      "epoch": 0.9736286919831224,
+      "grad_norm": 0.17411646246910095,
+      "learning_rate": 0.00011992195699392395,
+      "loss": 1.2503,
+      "step": 9230
+    },
+    {
+      "epoch": 0.9746835443037974,
+      "grad_norm": 0.15520411729812622,
+      "learning_rate": 0.0001179957481627428,
+      "loss": 1.2448,
+      "step": 9240
+    },
+    {
+      "epoch": 0.9757383966244726,
+      "grad_norm": 0.16046395897865295,
+      "learning_rate": 0.00011610047845692563,
+      "loss": 1.2585,
+      "step": 9250
+    },
+    {
+      "epoch": 0.9767932489451476,
+      "grad_norm": 0.1591130793094635,
+      "learning_rate": 0.0001142356509264725,
+      "loss": 1.2511,
+      "step": 9260
+    },
+    {
+      "epoch": 0.9778481012658228,
+      "grad_norm": 0.15294156968593597,
+      "learning_rate": 0.00011240077660348721,
+      "loss": 1.2631,
+      "step": 9270
+    },
+    {
+      "epoch": 0.9789029535864979,
+      "grad_norm": 0.16282202303409576,
+      "learning_rate": 0.00011059537437396703,
+      "loss": 1.2532,
+      "step": 9280
+    },
+    {
+      "epoch": 0.979957805907173,
+      "grad_norm": 0.16853636503219604,
+      "learning_rate": 0.00010881897085165203,
+      "loss": 1.2587,
+      "step": 9290
+    },
+    {
+      "epoch": 0.9810126582278481,
+      "grad_norm": 0.15748751163482666,
+      "learning_rate": 0.00010707110025390058,
+      "loss": 1.2552,
+      "step": 9300
+    },
+    {
+      "epoch": 0.9820675105485233,
+      "grad_norm": 0.15866565704345703,
+      "learning_rate": 0.00010535130427955874,
+      "loss": 1.2601,
+      "step": 9310
+    },
+    {
+      "epoch": 0.9831223628691983,
+      "grad_norm": 0.16810685396194458,
+      "learning_rate": 0.00010365913198879124,
+      "loss": 1.2541,
+      "step": 9320
+    },
+    {
+      "epoch": 0.9841772151898734,
+      "grad_norm": 0.16729974746704102,
+      "learning_rate": 0.00010199413968484228,
+      "loss": 1.2654,
+      "step": 9330
+    },
+    {
+      "epoch": 0.9852320675105485,
+      "grad_norm": 0.1650102585554123,
+      "learning_rate": 0.00010035589079769629,
+      "loss": 1.2657,
+      "step": 9340
+    },
+    {
+      "epoch": 0.9862869198312236,
+      "grad_norm": 0.15408636629581451,
+      "learning_rate": 9.874395576960658e-05,
+      "loss": 1.2763,
+      "step": 9350
+    },
+    {
+      "epoch": 0.9873417721518988,
+      "grad_norm": 0.16240571439266205,
+      "learning_rate": 9.715791194246317e-05,
+      "loss": 1.2625,
+      "step": 9360
+    },
+    {
+      "epoch": 0.9883966244725738,
+      "grad_norm": 0.16371123492717743,
+      "learning_rate": 9.559734344696928e-05,
+      "loss": 1.2659,
+      "step": 9370
+    },
+    {
+      "epoch": 0.989451476793249,
+      "grad_norm": 0.17476901412010193,
+      "learning_rate": 9.406184109359843e-05,
+      "loss": 1.2467,
+      "step": 9380
+    },
+    {
+      "epoch": 0.990506329113924,
+      "grad_norm": 0.1692182570695877,
+      "learning_rate": 9.255100226530257e-05,
+      "loss": 1.2534,
+      "step": 9390
+    },
+    {
+      "epoch": 0.9915611814345991,
+      "grad_norm": 0.1527872532606125,
+      "learning_rate": 9.10644308119438e-05,
+      "loss": 1.2689,
+      "step": 9400
+    },
+    {
+      "epoch": 0.9926160337552743,
+      "grad_norm": 0.17309527099132538,
+      "learning_rate": 8.960173694642143e-05,
+      "loss": 1.242,
+      "step": 9410
+    },
+    {
+      "epoch": 0.9936708860759493,
+      "grad_norm": 0.15739738941192627,
+      "learning_rate": 8.816253714246801e-05,
+      "loss": 1.2606,
+      "step": 9420
+    },
+    {
+      "epoch": 0.9947257383966245,
+      "grad_norm": 0.15997663140296936,
+      "learning_rate": 8.674645403408648e-05,
+      "loss": 1.2631,
+      "step": 9430
+    },
+    {
+      "epoch": 0.9957805907172996,
+      "grad_norm": 0.17512689530849457,
+      "learning_rate": 8.535311631660267e-05,
+      "loss": 1.26,
+      "step": 9440
+    },
+    {
+      "epoch": 0.9968354430379747,
+      "grad_norm": 0.16218222677707672,
+      "learning_rate": 8.398215864930742e-05,
+      "loss": 1.268,
+      "step": 9450
+    },
+    {
+      "epoch": 0.9978902953586498,
+      "grad_norm": 0.17091192305088043,
+      "learning_rate": 8.263322155966213e-05,
+      "loss": 1.2511,
+      "step": 9460
+    },
+    {
+      "epoch": 0.9989451476793249,
+      "grad_norm": 0.15781983733177185,
+      "learning_rate": 8.130595134904316e-05,
+      "loss": 1.2657,
+      "step": 9470
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.45242148637771606,
+      "learning_rate": 8e-05,
+      "loss": 1.2529,
+      "step": 9480
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 9480,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 2.405742592821581e+16,
+  "train_batch_size": 1024,
+  "trial_name": null,
+  "trial_params": null
+}