diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8682 +1,1642 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 15.0,
+  "epoch": 9.954337899543379,
   "eval_steps": 500,
-  "global_step": 6090,
+  "global_step": 1090,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0024630541871921183,
-      "grad_norm": 414.0,
-      "learning_rate": 3.284072249589491e-07,
-      "loss": 48.2481,
+      "epoch": 0.0091324200913242,
+      "grad_norm": 382.0,
+      "learning_rate": 1.8348623853211011e-06,
+      "loss": 46.9033,
       "step": 1
     },
     {
-      "epoch": 0.012315270935960592,
-      "grad_norm": 498.0,
-      "learning_rate": 1.6420361247947457e-06,
-      "loss": 47.2765,
+      "epoch": 0.045662100456621,
+      "grad_norm": 318.0,
+      "learning_rate": 9.174311926605506e-06,
+      "loss": 46.3618,
       "step": 5
     },
     {
-      "epoch": 0.024630541871921183,
-      "grad_norm": 346.0,
-      "learning_rate": 3.2840722495894914e-06,
-      "loss": 47.1113,
+      "epoch": 0.091324200913242,
+      "grad_norm": 139.0,
+      "learning_rate": 1.834862385321101e-05,
+      "loss": 39.3883,
       "step": 10
     },
     {
-      "epoch": 0.03694581280788178,
-      "grad_norm": 366.0,
-      "learning_rate": 4.926108374384237e-06,
-      "loss": 45.8102,
+      "epoch": 0.136986301369863,
+      "grad_norm": 52.25,
+      "learning_rate": 2.7522935779816515e-05,
+      "loss": 29.4216,
       "step": 15
     },
     {
-      "epoch": 0.04926108374384237,
-      "grad_norm": 358.0,
-      "learning_rate": 6.568144499178983e-06,
-      "loss": 40.5213,
+      "epoch": 0.182648401826484,
+      "grad_norm": 20.25,
+      "learning_rate": 3.669724770642202e-05,
+      "loss": 24.7169,
       "step": 20
     },
     {
-      "epoch": 0.06157635467980296,
-      "grad_norm": 288.0,
-      "learning_rate": 8.210180623973727e-06,
-      "loss": 35.9014,
+      "epoch": 0.228310502283105,
+      "grad_norm": 10.0,
+      "learning_rate": 4.587155963302753e-05,
+      "loss": 21.4244,
       "step": 25
     },
     {
-      "epoch": 0.07389162561576355,
-      "grad_norm": 198.0,
-      "learning_rate": 9.852216748768475e-06,
-      "loss": 31.5535,
+      "epoch": 0.273972602739726,
+      "grad_norm": 5.0,
+      "learning_rate": 5.504587155963303e-05,
+      "loss": 19.7804,
       "step": 30
     },
     {
-      "epoch": 0.08620689655172414,
-      "grad_norm": 71.0,
-      "learning_rate": 1.1494252873563218e-05,
-      "loss": 26.5323,
+      "epoch": 0.319634703196347,
+      "grad_norm": 4.6875,
+      "learning_rate": 6.422018348623854e-05,
+      "loss": 19.1075,
       "step": 35
     },
     {
-      "epoch": 0.09852216748768473,
-      "grad_norm": 50.0,
-      "learning_rate": 1.3136288998357965e-05,
-      "loss": 24.6833,
+      "epoch": 0.365296803652968,
+      "grad_norm": 8.5,
+      "learning_rate": 7.339449541284404e-05,
+      "loss": 18.037,
       "step": 40
     },
     {
-      "epoch": 0.11083743842364532,
-      "grad_norm": 34.75,
-      "learning_rate": 1.4778325123152711e-05,
-      "loss": 23.2877,
+      "epoch": 0.410958904109589,
+      "grad_norm": 18.75,
+      "learning_rate": 8.256880733944955e-05,
+      "loss": 17.2652,
       "step": 45
     },
     {
-      "epoch": 0.12315270935960591,
-      "grad_norm": 22.75,
-      "learning_rate": 1.6420361247947455e-05,
-      "loss": 22.3972,
+      "epoch": 0.45662100456621,
+      "grad_norm": 29.875,
+      "learning_rate": 9.174311926605506e-05,
+      "loss": 14.4775,
       "step": 50
     },
     {
-      "epoch": 0.1354679802955665,
-      "grad_norm": 16.25,
-      "learning_rate": 1.8062397372742202e-05,
-      "loss": 20.5332,
+      "epoch": 0.502283105022831,
+      "grad_norm": 37.75,
+      "learning_rate": 0.00010091743119266055,
+      "loss": 9.6302,
       "step": 55
     },
     {
-      "epoch": 0.1477832512315271,
-      "grad_norm": 8.6875,
-      "learning_rate": 1.970443349753695e-05,
-      "loss": 19.9415,
+      "epoch": 0.547945205479452,
+      "grad_norm": 8.75,
+      "learning_rate": 0.00011009174311926606,
+      "loss": 4.0499,
       "step": 60
     },
     {
-      "epoch": 0.16009852216748768,
-      "grad_norm": 7.09375,
-      "learning_rate": 2.1346469622331693e-05,
-      "loss": 19.4214,
+      "epoch": 0.593607305936073,
+      "grad_norm": 7.03125,
+      "learning_rate": 0.00011926605504587157,
+      "loss": 2.4784,
       "step": 65
     },
     {
-      "epoch": 0.1724137931034483,
-      "grad_norm": 8.25,
-      "learning_rate": 2.2988505747126437e-05,
-      "loss": 18.7164,
+      "epoch": 0.639269406392694,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00012844036697247707,
+      "loss": 2.0967,
       "step": 70
     },
     {
-      "epoch": 0.18472906403940886,
-      "grad_norm": 10.375,
-      "learning_rate": 2.4630541871921184e-05,
-      "loss": 18.4288,
+      "epoch": 0.684931506849315,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00013761467889908258,
+      "loss": 1.8211,
       "step": 75
     },
     {
-      "epoch": 0.19704433497536947,
-      "grad_norm": 12.9375,
-      "learning_rate": 2.627257799671593e-05,
-      "loss": 17.7133,
+      "epoch": 0.730593607305936,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.0001467889908256881,
+      "loss": 1.6757,
       "step": 80
     },
     {
-      "epoch": 0.20935960591133004,
-      "grad_norm": 22.5,
-      "learning_rate": 2.7914614121510675e-05,
-      "loss": 17.1272,
+      "epoch": 0.776255707762557,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.0001559633027522936,
+      "loss": 1.5691,
       "step": 85
     },
     {
-      "epoch": 0.22167487684729065,
-      "grad_norm": 34.5,
-      "learning_rate": 2.9556650246305422e-05,
-      "loss": 16.1901,
+      "epoch": 0.821917808219178,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0001651376146788991,
+      "loss": 1.4872,
       "step": 90
     },
     {
-      "epoch": 0.23399014778325122,
-      "grad_norm": 56.25,
-      "learning_rate": 3.119868637110017e-05,
-      "loss": 13.4454,
+      "epoch": 0.867579908675799,
+      "grad_norm": 0.90234375,
+      "learning_rate": 0.00017431192660550458,
+      "loss": 1.4228,
       "step": 95
     },
     {
-      "epoch": 0.24630541871921183,
-      "grad_norm": 68.0,
-      "learning_rate": 3.284072249589491e-05,
-      "loss": 10.339,
+      "epoch": 0.91324200913242,
+      "grad_norm": 2.546875,
+      "learning_rate": 0.00018348623853211012,
+      "loss": 1.3778,
       "step": 100
     },
     {
-      "epoch": 0.25862068965517243,
-      "grad_norm": 43.25,
-      "learning_rate": 3.4482758620689657e-05,
-      "loss": 5.5131,
+      "epoch": 0.958904109589041,
+      "grad_norm": 1.5390625,
+      "learning_rate": 0.0001926605504587156,
+      "loss": 1.3554,
       "step": 105
     },
     {
-      "epoch": 0.270935960591133,
-      "grad_norm": 21.25,
-      "learning_rate": 3.6124794745484404e-05,
-      "loss": 2.7823,
+      "epoch": 0.9954337899543378,
+      "eval_loss": 2.645094871520996,
+      "eval_runtime": 0.2786,
+      "eval_samples_per_second": 35.888,
+      "eval_steps_per_second": 3.589,
+      "step": 109
+    },
+    {
+      "epoch": 1.004566210045662,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00019999948721966259,
+      "loss": 1.326,
       "step": 110
     },
     {
-      "epoch": 0.2832512315270936,
-      "grad_norm": 3.578125,
-      "learning_rate": 3.7766830870279144e-05,
-      "loss": 2.3041,
+      "epoch": 1.0502283105022832,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00019998154046002822,
+      "loss": 1.291,
       "step": 115
     },
     {
-      "epoch": 0.2955665024630542,
-      "grad_norm": 4.3125,
-      "learning_rate": 3.94088669950739e-05,
-      "loss": 2.0115,
+      "epoch": 1.095890410958904,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.0001999379599421534,
+      "loss": 1.2721,
       "step": 120
     },
     {
-      "epoch": 0.3078817733990148,
-      "grad_norm": 2.296875,
-      "learning_rate": 4.105090311986864e-05,
-      "loss": 1.8402,
+      "epoch": 1.1415525114155252,
+      "grad_norm": 1.5546875,
+      "learning_rate": 0.00019986875683942535,
+      "loss": 1.2479,
       "step": 125
     },
     {
-      "epoch": 0.32019704433497537,
-      "grad_norm": 2.1875,
-      "learning_rate": 4.2692939244663386e-05,
-      "loss": 1.6992,
+      "epoch": 1.187214611872146,
+      "grad_norm": 3.375,
+      "learning_rate": 0.00019977394889447524,
+      "loss": 1.2491,
       "step": 130
     },
     {
-      "epoch": 0.33251231527093594,
-      "grad_norm": 5.53125,
-      "learning_rate": 4.433497536945813e-05,
-      "loss": 1.5865,
+      "epoch": 1.2328767123287672,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00019965356041462955,
+      "loss": 1.2212,
       "step": 135
     },
     {
-      "epoch": 0.3448275862068966,
-      "grad_norm": 2.40625,
-      "learning_rate": 4.597701149425287e-05,
-      "loss": 1.5333,
+      "epoch": 1.278538812785388,
+      "grad_norm": 1.546875,
+      "learning_rate": 0.00019950762226567781,
+      "loss": 1.2246,
       "step": 140
     },
     {
-      "epoch": 0.35714285714285715,
-      "grad_norm": 4.03125,
-      "learning_rate": 4.761904761904762e-05,
-      "loss": 1.4848,
+      "epoch": 1.3242009132420092,
+      "grad_norm": 1.984375,
+      "learning_rate": 0.00019933617186395917,
+      "loss": 1.2387,
       "step": 145
     },
     {
-      "epoch": 0.3694581280788177,
-      "grad_norm": 2.375,
-      "learning_rate": 4.926108374384237e-05,
-      "loss": 1.4199,
+      "epoch": 1.36986301369863,
+      "grad_norm": 2.46875,
+      "learning_rate": 0.00019913925316676945,
+      "loss": 1.203,
       "step": 150
     },
     {
-      "epoch": 0.3817733990147783,
-      "grad_norm": 4.3125,
-      "learning_rate": 5.0903119868637115e-05,
-      "loss": 1.3649,
+      "epoch": 1.4155251141552512,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00019891691666109113,
+      "loss": 1.1869,
       "step": 155
     },
     {
-      "epoch": 0.39408866995073893,
-      "grad_norm": 20.875,
-      "learning_rate": 5.254515599343186e-05,
-      "loss": 1.333,
+      "epoch": 1.461187214611872,
+      "grad_norm": 2.28125,
+      "learning_rate": 0.00019866921935064906,
+      "loss": 1.1858,
       "step": 160
     },
     {
-      "epoch": 0.4064039408866995,
-      "grad_norm": 2.515625,
-      "learning_rate": 5.41871921182266e-05,
-      "loss": 1.2964,
+      "epoch": 1.5068493150684932,
+      "grad_norm": 1.296875,
+      "learning_rate": 0.00019839622474129596,
+      "loss": 1.1696,
       "step": 165
     },
     {
-      "epoch": 0.4187192118226601,
-      "grad_norm": 5.03125,
-      "learning_rate": 5.582922824302135e-05,
-      "loss": 1.2699,
+      "epoch": 1.5525114155251143,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00019809800282473013,
+      "loss": 1.1624,
       "step": 170
     },
     {
-      "epoch": 0.43103448275862066,
-      "grad_norm": 2.59375,
-      "learning_rate": 5.747126436781609e-05,
-      "loss": 1.241,
+      "epoch": 1.5981735159817352,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.0001977746300605507,
+      "loss": 1.1494,
       "step": 175
     },
     {
-      "epoch": 0.4433497536945813,
-      "grad_norm": 4.1875,
-      "learning_rate": 5.9113300492610844e-05,
-      "loss": 1.2204,
+      "epoch": 1.643835616438356,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00019742618935665476,
+      "loss": 1.1314,
       "step": 180
     },
     {
-      "epoch": 0.45566502463054187,
-      "grad_norm": 4.125,
-      "learning_rate": 6.075533661740559e-05,
-      "loss": 1.2022,
+      "epoch": 1.6894977168949772,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.00019705277004798073,
+      "loss": 1.1407,
       "step": 185
     },
     {
-      "epoch": 0.46798029556650245,
-      "grad_norm": 5.03125,
-      "learning_rate": 6.239737274220034e-05,
-      "loss": 1.1749,
+      "epoch": 1.7351598173515983,
+      "grad_norm": 2.0625,
+      "learning_rate": 0.0001966544678736044,
+      "loss": 1.1287,
       "step": 190
     },
     {
-      "epoch": 0.4802955665024631,
-      "grad_norm": 1.6796875,
-      "learning_rate": 6.403940886699507e-05,
-      "loss": 1.1671,
+      "epoch": 1.7808219178082192,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00019623138495219292,
+      "loss": 1.1407,
       "step": 195
     },
     {
-      "epoch": 0.49261083743842365,
-      "grad_norm": 3.25,
-      "learning_rate": 6.568144499178982e-05,
-      "loss": 1.1441,
+      "epoch": 1.82648401826484,
+      "grad_norm": 12.0,
+      "learning_rate": 0.00019578362975582292,
+      "loss": 1.1151,
       "step": 200
     },
     {
-      "epoch": 0.5049261083743842,
-      "grad_norm": 5.96875,
-      "learning_rate": 6.732348111658457e-05,
-      "loss": 1.1313,
+      "epoch": 1.8721461187214612,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00019531131708217005,
+      "loss": 1.1221,
       "step": 205
     },
     {
-      "epoch": 0.5172413793103449,
-      "grad_norm": 5.75,
-      "learning_rate": 6.896551724137931e-05,
-      "loss": 1.1001,
+      "epoch": 1.9178082191780823,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.0001948145680250766,
+      "loss": 1.0982,
       "step": 210
     },
     {
-      "epoch": 0.5295566502463054,
-      "grad_norm": 1.78125,
-      "learning_rate": 7.060755336617406e-05,
-      "loss": 1.1061,
+      "epoch": 1.9634703196347032,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.00019429350994350483,
+      "loss": 1.0898,
       "step": 215
     },
     {
-      "epoch": 0.541871921182266,
-      "grad_norm": 1.6484375,
-      "learning_rate": 7.224958949096881e-05,
-      "loss": 1.0903,
+      "epoch": 2.0,
+      "eval_loss": 2.508340358734131,
+      "eval_runtime": 0.2456,
+      "eval_samples_per_second": 40.72,
+      "eval_steps_per_second": 4.072,
+      "step": 219
+    },
+    {
+      "epoch": 2.009132420091324,
+      "grad_norm": 2.59375,
+      "learning_rate": 0.00019374827642888398,
+      "loss": 1.1068,
       "step": 220
     },
     {
-      "epoch": 0.5541871921182266,
-      "grad_norm": 1.8515625,
-      "learning_rate": 7.389162561576355e-05,
-      "loss": 1.0691,
+      "epoch": 2.0547945205479454,
+      "grad_norm": 4.09375,
+      "learning_rate": 0.0001931790072708596,
+      "loss": 1.0932,
       "step": 225
     },
     {
-      "epoch": 0.5665024630541872,
-      "grad_norm": 1.53125,
-      "learning_rate": 7.553366174055829e-05,
-      "loss": 1.0464,
+      "epoch": 2.1004566210045663,
+      "grad_norm": 1.5,
+      "learning_rate": 0.00019258584842145343,
+      "loss": 1.1001,
       "step": 230
     },
     {
-      "epoch": 0.5788177339901478,
-      "grad_norm": 2.0,
-      "learning_rate": 7.717569786535304e-05,
-      "loss": 1.0537,
+      "epoch": 2.146118721461187,
+      "grad_norm": 11.25,
+      "learning_rate": 0.00019196895195764362,
+      "loss": 1.1001,
       "step": 235
     },
     {
-      "epoch": 0.5911330049261084,
-      "grad_norm": 1.8515625,
-      "learning_rate": 7.88177339901478e-05,
-      "loss": 1.0279,
+      "epoch": 2.191780821917808,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.0001913284760423745,
+      "loss": 1.1046,
       "step": 240
     },
     {
-      "epoch": 0.603448275862069,
-      "grad_norm": 2.5625,
-      "learning_rate": 8.045977011494253e-05,
-      "loss": 1.0285,
+      "epoch": 2.237442922374429,
+      "grad_norm": 2.828125,
+      "learning_rate": 0.00019066458488400584,
+      "loss": 1.0795,
       "step": 245
     },
     {
-      "epoch": 0.6157635467980296,
-      "grad_norm": 10.125,
-      "learning_rate": 8.210180623973728e-05,
-      "loss": 1.043,
+      "epoch": 2.2831050228310503,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00018997744869421246,
+      "loss": 1.0767,
       "step": 250
     },
     {
-      "epoch": 0.6280788177339901,
-      "grad_norm": 1.984375,
-      "learning_rate": 8.374384236453202e-05,
-      "loss": 1.0256,
+      "epoch": 2.328767123287671,
+      "grad_norm": 4.875,
+      "learning_rate": 0.00018926724364434446,
+      "loss": 1.059,
       "step": 255
     },
     {
-      "epoch": 0.6403940886699507,
-      "grad_norm": 3.328125,
-      "learning_rate": 8.538587848932677e-05,
-      "loss": 1.0294,
+      "epoch": 2.374429223744292,
+      "grad_norm": 3.171875,
+      "learning_rate": 0.0001885341518202595,
+      "loss": 1.0695,
       "step": 260
     },
     {
-      "epoch": 0.6527093596059114,
-      "grad_norm": 4.4375,
-      "learning_rate": 8.702791461412152e-05,
-      "loss": 1.0111,
+      "epoch": 2.4200913242009134,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00018777836117563892,
+      "loss": 1.0709,
       "step": 265
     },
     {
-      "epoch": 0.6650246305418719,
-      "grad_norm": 20.125,
-      "learning_rate": 8.866995073891627e-05,
-      "loss": 1.022,
+      "epoch": 2.4657534246575343,
+      "grad_norm": 7.59375,
+      "learning_rate": 0.00018700006548379898,
+      "loss": 1.0677,
       "step": 270
     },
     {
-      "epoch": 0.6773399014778325,
-      "grad_norm": 5.96875,
-      "learning_rate": 9.031198686371101e-05,
-      "loss": 1.0246,
+      "epoch": 2.5114155251141552,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001861994642880105,
+      "loss": 1.0693,
       "step": 275
     },
     {
-      "epoch": 0.6896551724137931,
-      "grad_norm": 1.5625,
-      "learning_rate": 9.195402298850575e-05,
-      "loss": 1.0236,
+      "epoch": 2.557077625570776,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00018537676285033887,
+      "loss": 1.0508,
       "step": 280
     },
     {
-      "epoch": 0.7019704433497537,
-      "grad_norm": 2.59375,
-      "learning_rate": 9.35960591133005e-05,
-      "loss": 1.0164,
+      "epoch": 2.602739726027397,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001845321720990181,
+      "loss": 1.0449,
       "step": 285
     },
     {
-      "epoch": 0.7142857142857143,
-      "grad_norm": 2.640625,
-      "learning_rate": 9.523809523809524e-05,
-      "loss": 1.0113,
+      "epoch": 2.6484018264840183,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00018366590857437184,
+      "loss": 1.0562,
       "step": 290
     },
     {
-      "epoch": 0.7266009852216748,
-      "grad_norm": 1.140625,
-      "learning_rate": 9.688013136288999e-05,
-      "loss": 1.0018,
+      "epoch": 2.6940639269406392,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00018277819437329576,
+      "loss": 1.0428,
       "step": 295
     },
     {
-      "epoch": 0.7389162561576355,
-      "grad_norm": 1.09375,
-      "learning_rate": 9.852216748768474e-05,
-      "loss": 0.9837,
+      "epoch": 2.73972602739726,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00018186925709231532,
+      "loss": 1.0321,
       "step": 300
     },
     {
-      "epoch": 0.7512315270935961,
-      "grad_norm": 5.21875,
-      "learning_rate": 0.0001001642036124795,
-      "loss": 1.013,
+      "epoch": 2.7853881278538815,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001809393297692334,
+      "loss": 1.0253,
       "step": 305
     },
     {
-      "epoch": 0.7635467980295566,
-      "grad_norm": 2.15625,
-      "learning_rate": 0.00010180623973727423,
-      "loss": 1.0111,
+      "epoch": 2.8310502283105023,
+      "grad_norm": 3.1875,
+      "learning_rate": 0.0001799886508233829,
+      "loss": 1.0377,
       "step": 310
     },
     {
-      "epoch": 0.7758620689655172,
-      "grad_norm": 3.125,
-      "learning_rate": 0.00010344827586206898,
-      "loss": 0.997,
+      "epoch": 2.8767123287671232,
+      "grad_norm": 3.484375,
+      "learning_rate": 0.0001790174639944997,
+      "loss": 1.0359,
       "step": 315
     },
     {
-      "epoch": 0.7881773399014779,
-      "grad_norm": 1.3515625,
-      "learning_rate": 0.00010509031198686372,
-      "loss": 0.9858,
+      "epoch": 2.922374429223744,
+      "grad_norm": 6.09375,
+      "learning_rate": 0.00017802601828023138,
+      "loss": 1.0428,
       "step": 320
     },
     {
-      "epoch": 0.8004926108374384,
-      "grad_norm": 3.0,
-      "learning_rate": 0.00010673234811165847,
-      "loss": 0.9788,
+      "epoch": 2.968036529680365,
+      "grad_norm": 3.03125,
+      "learning_rate": 0.00017701456787229804,
+      "loss": 1.0434,
       "step": 325
     },
     {
-      "epoch": 0.812807881773399,
-      "grad_norm": 1.40625,
-      "learning_rate": 0.0001083743842364532,
-      "loss": 0.9844,
+      "epoch": 2.9954337899543377,
+      "eval_loss": 2.480058193206787,
+      "eval_runtime": 0.2581,
+      "eval_samples_per_second": 38.741,
+      "eval_steps_per_second": 3.874,
+      "step": 328
+    },
+    {
+      "epoch": 3.0136986301369864,
+      "grad_norm": 7.625,
+      "learning_rate": 0.0001759833720913214,
+      "loss": 1.0302,
       "step": 330
     },
     {
-      "epoch": 0.8251231527093597,
-      "grad_norm": 1.390625,
-      "learning_rate": 0.00011001642036124795,
-      "loss": 0.9594,
+      "epoch": 3.0593607305936072,
+      "grad_norm": 1.7890625,
+      "learning_rate": 0.00017493269532033883,
+      "loss": 1.0273,
       "step": 335
     },
     {
-      "epoch": 0.8374384236453202,
-      "grad_norm": 2.578125,
-      "learning_rate": 0.0001116584564860427,
-      "loss": 0.9913,
+      "epoch": 3.105022831050228,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.0001738628069370195,
+      "loss": 1.0212,
       "step": 340
     },
     {
-      "epoch": 0.8497536945812808,
-      "grad_norm": 1.9375,
-      "learning_rate": 0.00011330049261083743,
-      "loss": 0.9689,
+      "epoch": 3.1506849315068495,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00017277398124460023,
+      "loss": 1.013,
       "step": 345
     },
     {
-      "epoch": 0.8620689655172413,
-      "grad_norm": 4.5625,
-      "learning_rate": 0.00011494252873563218,
-      "loss": 0.9669,
+      "epoch": 3.1963470319634704,
+      "grad_norm": 2.390625,
+      "learning_rate": 0.000171666497401558,
+      "loss": 1.0077,
       "step": 350
     },
     {
-      "epoch": 0.874384236453202,
-      "grad_norm": 2.171875,
-      "learning_rate": 0.00011658456486042693,
-      "loss": 0.9903,
+      "epoch": 3.2420091324200913,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0001705406393500381,
+      "loss": 1.0111,
       "step": 355
     },
     {
-      "epoch": 0.8866995073891626,
-      "grad_norm": 80.5,
-      "learning_rate": 0.00011822660098522169,
-      "loss": 0.9542,
+      "epoch": 3.287671232876712,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.00016939669574305566,
+      "loss": 1.0047,
       "step": 360
     },
     {
-      "epoch": 0.8990147783251231,
-      "grad_norm": 5.8125,
-      "learning_rate": 0.00011986863711001643,
-      "loss": 0.9595,
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001682349598704892,
+      "loss": 0.9977,
       "step": 365
     },
     {
-      "epoch": 0.9113300492610837,
-      "grad_norm": 3.671875,
-      "learning_rate": 0.00012151067323481118,
-      "loss": 0.952,
+      "epoch": 3.3789954337899544,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00016705572958388576,
+      "loss": 0.9914,
       "step": 370
     },
     {
-      "epoch": 0.9236453201970444,
-      "grad_norm": 1.9140625,
-      "learning_rate": 0.00012315270935960593,
-      "loss": 0.9749,
+      "epoch": 3.4246575342465753,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00016585930722009601,
+      "loss": 1.0012,
       "step": 375
     },
     {
-      "epoch": 0.9359605911330049,
-      "grad_norm": 1.78125,
-      "learning_rate": 0.00012479474548440068,
-      "loss": 0.9496,
+      "epoch": 3.470319634703196,
+      "grad_norm": 1.7421875,
+      "learning_rate": 0.00016464599952375998,
+      "loss": 0.9888,
       "step": 380
     },
     {
-      "epoch": 0.9482758620689655,
-      "grad_norm": 3.640625,
-      "learning_rate": 0.0001264367816091954,
-      "loss": 0.9454,
+      "epoch": 3.5159817351598175,
+      "grad_norm": 0.73828125,
+      "learning_rate": 0.000163416117568662,
+      "loss": 1.0036,
       "step": 385
     },
     {
-      "epoch": 0.9605911330049262,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00012807881773399014,
-      "loss": 0.9526,
+      "epoch": 3.5616438356164384,
+      "grad_norm": 2.515625,
+      "learning_rate": 0.0001621699766779763,
+      "loss": 0.9963,
       "step": 390
     },
     {
-      "epoch": 0.9729064039408867,
-      "grad_norm": 3.703125,
-      "learning_rate": 0.0001297208538587849,
-      "loss": 0.9354,
+      "epoch": 3.6073059360730593,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00016090789634342278,
+      "loss": 0.9955,
       "step": 395
     },
     {
-      "epoch": 0.9852216748768473,
-      "grad_norm": 1.6328125,
-      "learning_rate": 0.00013136288998357964,
-      "loss": 0.9335,
+      "epoch": 3.65296803652968,
+      "grad_norm": 1.53125,
+      "learning_rate": 0.00015963020014335438,
+      "loss": 0.9953,
       "step": 400
     },
     {
-      "epoch": 0.9975369458128078,
-      "grad_norm": 1.046875,
-      "learning_rate": 0.00013300492610837438,
-      "loss": 0.9333,
+      "epoch": 3.6986301369863015,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.0001583372156597961,
+      "loss": 0.9959,
       "step": 405
     },
     {
-      "epoch": 1.0,
-      "eval_loss": 2.444850206375122,
-      "eval_runtime": 2.0491,
-      "eval_samples_per_second": 4.88,
-      "eval_steps_per_second": 0.976,
-      "step": 406
-    },
-    {
-      "epoch": 1.0098522167487685,
-      "grad_norm": 1.2578125,
-      "learning_rate": 0.00013464696223316913,
-      "loss": 0.9413,
+      "epoch": 3.7442922374429224,
+      "grad_norm": 1.75,
+      "learning_rate": 0.00015702927439445826,
+      "loss": 0.9906,
       "step": 410
     },
     {
-      "epoch": 1.022167487684729,
-      "grad_norm": 2.765625,
-      "learning_rate": 0.00013628899835796388,
-      "loss": 0.9055,
+      "epoch": 3.7899543378995433,
+      "grad_norm": 1.4453125,
+      "learning_rate": 0.00015570671168374438,
+      "loss": 0.9849,
       "step": 415
     },
     {
-      "epoch": 1.0344827586206897,
-      "grad_norm": 5.1875,
-      "learning_rate": 0.00013793103448275863,
-      "loss": 0.9148,
+      "epoch": 3.8356164383561646,
+      "grad_norm": 2.234375,
+      "learning_rate": 0.00015436986661277577,
+      "loss": 0.9697,
       "step": 420
     },
     {
-      "epoch": 1.0467980295566504,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.00013957307060755337,
-      "loss": 0.8961,
+      "epoch": 3.8812785388127855,
+      "grad_norm": 1.5,
+      "learning_rate": 0.0001530190819284555,
+      "loss": 0.979,
       "step": 425
     },
     {
-      "epoch": 1.0591133004926108,
-      "grad_norm": 1.40625,
-      "learning_rate": 0.00014121510673234812,
-      "loss": 0.9097,
+      "epoch": 3.9269406392694064,
+      "grad_norm": 10.75,
+      "learning_rate": 0.00015165470395159313,
+      "loss": 0.9715,
       "step": 430
     },
     {
-      "epoch": 1.0714285714285714,
-      "grad_norm": 2.578125,
-      "learning_rate": 0.00014285714285714287,
-      "loss": 0.9046,
+      "epoch": 3.9726027397260273,
+      "grad_norm": 2.71875,
+      "learning_rate": 0.0001502770824881133,
+      "loss": 0.9864,
       "step": 435
     },
     {
-      "epoch": 1.083743842364532,
-      "grad_norm": 0.921875,
-      "learning_rate": 0.00014449917898193762,
-      "loss": 0.8812,
+      "epoch": 4.0,
+      "eval_loss": 2.474334239959717,
+      "eval_runtime": 0.2363,
+      "eval_samples_per_second": 42.318,
+      "eval_steps_per_second": 4.232,
+      "step": 438
+    },
+    {
+      "epoch": 4.018264840182648,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00014888657073937076,
+      "loss": 0.9764,
       "step": 440
     },
     {
-      "epoch": 1.0960591133004927,
-      "grad_norm": 1.859375,
-      "learning_rate": 0.00014614121510673236,
-      "loss": 0.9016,
+      "epoch": 4.063926940639269,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00014748352521159493,
+      "loss": 0.9564,
       "step": 445
     },
     {
-      "epoch": 1.1083743842364533,
-      "grad_norm": 0.8515625,
-      "learning_rate": 0.0001477832512315271,
-      "loss": 0.8977,
+      "epoch": 4.109589041095891,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0001460683056244869,
+      "loss": 0.9573,
       "step": 450
     },
     {
-      "epoch": 1.1206896551724137,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.00014942528735632183,
-      "loss": 0.8917,
+      "epoch": 4.155251141552512,
+      "grad_norm": 11.5625,
+      "learning_rate": 0.00014464127481899312,
+      "loss": 0.957,
       "step": 455
     },
     {
-      "epoch": 1.1330049261083743,
-      "grad_norm": 2.125,
-      "learning_rate": 0.00015106732348111658,
-      "loss": 0.8861,
+      "epoch": 4.200913242009133,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00014320279866427796,
+      "loss": 0.9596,
       "step": 460
     },
     {
-      "epoch": 1.145320197044335,
-      "grad_norm": 0.8203125,
-      "learning_rate": 0.00015270935960591132,
-      "loss": 0.8908,
+      "epoch": 4.2465753424657535,
+      "grad_norm": 2.03125,
+      "learning_rate": 0.00014175324596392075,
+      "loss": 0.9647,
       "step": 465
     },
     {
-      "epoch": 1.1576354679802956,
-      "grad_norm": 2.8125,
-      "learning_rate": 0.00015435139573070607,
-      "loss": 0.8853,
+      "epoch": 4.292237442922374,
+      "grad_norm": 1.359375,
+      "learning_rate": 0.00014029298836135988,
+      "loss": 0.9632,
       "step": 470
     },
     {
-      "epoch": 1.1699507389162562,
-      "grad_norm": 3.125,
-      "learning_rate": 0.00015599343185550085,
-      "loss": 0.8877,
+      "epoch": 4.337899543378995,
+      "grad_norm": 5.09375,
+      "learning_rate": 0.00013882240024460927,
+      "loss": 0.9664,
       "step": 475
     },
     {
-      "epoch": 1.1822660098522166,
-      "grad_norm": 2.328125,
-      "learning_rate": 0.0001576354679802956,
-      "loss": 0.8836,
+      "epoch": 4.383561643835616,
+      "grad_norm": 2.96875,
+      "learning_rate": 0.0001373418586502706,
+      "loss": 0.964,
       "step": 480
     },
     {
-      "epoch": 1.1945812807881773,
-      "grad_norm": 3.15625,
-      "learning_rate": 0.0001592775041050903,
-      "loss": 0.8755,
+      "epoch": 4.429223744292237,
+      "grad_norm": 1.2421875,
+      "learning_rate": 0.0001358517431668672,
+      "loss": 0.9531,
       "step": 485
     },
     {
-      "epoch": 1.206896551724138,
-      "grad_norm": 1.7890625,
-      "learning_rate": 0.00016091954022988506,
-      "loss": 0.8794,
+      "epoch": 4.474885844748858,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00013435243583752294,
+      "loss": 0.958,
       "step": 490
     },
     {
-      "epoch": 1.2192118226600985,
-      "grad_norm": 2.015625,
-      "learning_rate": 0.0001625615763546798,
-      "loss": 0.9029,
+      "epoch": 4.52054794520548,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00013284432106201233,
+      "loss": 0.9514,
       "step": 495
     },
     {
-      "epoch": 1.2315270935960592,
-      "grad_norm": 1.9765625,
-      "learning_rate": 0.00016420361247947455,
-      "loss": 0.9035,
+      "epoch": 4.566210045662101,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00013132778549820618,
+      "loss": 0.9588,
       "step": 500
     },
     {
-      "epoch": 1.2438423645320198,
-      "grad_norm": 7.96875,
-      "learning_rate": 0.0001658456486042693,
-      "loss": 0.9059,
+      "epoch": 4.6118721461187215,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00012980321796293836,
+      "loss": 0.9494,
       "step": 505
     },
     {
-      "epoch": 1.2561576354679804,
-      "grad_norm": 8.25,
-      "learning_rate": 0.00016748768472906405,
-      "loss": 0.9183,
+      "epoch": 4.657534246575342,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00012827100933231905,
+      "loss": 0.9508,
       "step": 510
     },
     {
-      "epoch": 1.2684729064039408,
-      "grad_norm": 2.90625,
-      "learning_rate": 0.0001691297208538588,
-      "loss": 0.9074,
+      "epoch": 4.703196347031963,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.00012673155244151985,
+      "loss": 0.9557,
       "step": 515
     },
     {
-      "epoch": 1.2807881773399015,
-      "grad_norm": 2.109375,
-      "learning_rate": 0.00017077175697865354,
-      "loss": 0.8946,
+      "epoch": 4.748858447488584,
+      "grad_norm": 0.75,
+      "learning_rate": 0.000125185241984057,
+      "loss": 0.9508,
       "step": 520
     },
     {
-      "epoch": 1.293103448275862,
-      "grad_norm": 5.03125,
-      "learning_rate": 0.00017241379310344826,
-      "loss": 0.8911,
+      "epoch": 4.794520547945205,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00012363247441059776,
+      "loss": 0.9562,
       "step": 525
     },
     {
-      "epoch": 1.3054187192118227,
-      "grad_norm": 1.40625,
-      "learning_rate": 0.00017405582922824304,
-      "loss": 0.8732,
+      "epoch": 4.840182648401827,
+      "grad_norm": 1.078125,
+      "learning_rate": 0.00012207364782731655,
+      "loss": 0.9542,
       "step": 530
     },
     {
-      "epoch": 1.3177339901477834,
-      "grad_norm": 1.1796875,
-      "learning_rate": 0.00017569786535303778,
-      "loss": 0.8752,
+      "epoch": 4.885844748858448,
+      "grad_norm": 2.1875,
+      "learning_rate": 0.00012050916189382646,
+      "loss": 0.9606,
       "step": 535
     },
     {
-      "epoch": 1.3300492610837438,
-      "grad_norm": 1.546875,
-      "learning_rate": 0.00017733990147783253,
-      "loss": 0.8822,
+      "epoch": 4.931506849315069,
+      "grad_norm": 1.84375,
+      "learning_rate": 0.00011893941772071249,
+      "loss": 0.9424,
       "step": 540
     },
     {
-      "epoch": 1.3423645320197044,
-      "grad_norm": 1.0,
-      "learning_rate": 0.00017898193760262728,
-      "loss": 0.8715,
+      "epoch": 4.9771689497716896,
+      "grad_norm": 1.9375,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 0.9371,
       "step": 545
     },
     {
-      "epoch": 1.354679802955665,
-      "grad_norm": 0.91796875,
-      "learning_rate": 0.00018062397372742203,
-      "loss": 0.8754,
+      "epoch": 4.995433789954338,
+      "eval_loss": 2.485384464263916,
+      "eval_runtime": 0.2567,
+      "eval_samples_per_second": 38.96,
+      "eval_steps_per_second": 3.896,
+      "step": 547
+    },
+    {
+      "epoch": 5.0228310502283104,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.0001157857657354354,
+      "loss": 0.9249,
       "step": 550
     },
     {
-      "epoch": 1.3669950738916257,
-      "grad_norm": 1.171875,
-      "learning_rate": 0.00018226600985221675,
-      "loss": 0.8664,
+      "epoch": 5.068493150684931,
+      "grad_norm": 2.53125,
+      "learning_rate": 0.00011420266647205231,
+      "loss": 0.9271,
       "step": 555
     },
     {
-      "epoch": 1.3793103448275863,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.0001839080459770115,
-      "loss": 0.8896,
+      "epoch": 5.114155251141552,
+      "grad_norm": 12.0,
+      "learning_rate": 0.00011261592585930576,
+      "loss": 0.9329,
       "step": 560
     },
     {
-      "epoch": 1.3916256157635467,
-      "grad_norm": 1.21875,
-      "learning_rate": 0.00018555008210180624,
-      "loss": 0.8547,
+      "epoch": 5.159817351598173,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00011102595071354472,
+      "loss": 0.9238,
       "step": 565
     },
     {
-      "epoch": 1.4039408866995073,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.000187192118226601,
-      "loss": 0.8671,
+      "epoch": 5.205479452054795,
+      "grad_norm": 6.3125,
+      "learning_rate": 0.00010943314868040364,
+      "loss": 0.9134,
       "step": 570
     },
     {
-      "epoch": 1.416256157635468,
-      "grad_norm": 1.109375,
-      "learning_rate": 0.00018883415435139573,
-      "loss": 0.8716,
+      "epoch": 5.251141552511416,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00010783792813028827,
+      "loss": 0.91,
       "step": 575
     },
     {
-      "epoch": 1.4285714285714286,
-      "grad_norm": 2.125,
-      "learning_rate": 0.00019047619047619048,
-      "loss": 0.8814,
+      "epoch": 5.296803652968037,
+      "grad_norm": 4.375,
+      "learning_rate": 0.00010624069805367559,
+      "loss": 0.9193,
       "step": 580
     },
     {
-      "epoch": 1.4408866995073892,
-      "grad_norm": 2.609375,
-      "learning_rate": 0.00019211822660098523,
-      "loss": 0.8788,
+      "epoch": 5.342465753424658,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00010464186795625482,
+      "loss": 0.9101,
       "step": 585
     },
     {
-      "epoch": 1.4532019704433496,
-      "grad_norm": 1.515625,
-      "learning_rate": 0.00019376026272577998,
-      "loss": 0.8676,
+      "epoch": 5.3881278538812785,
+      "grad_norm": 3.1875,
+      "learning_rate": 0.00010304184775393642,
+      "loss": 0.9122,
       "step": 590
     },
     {
-      "epoch": 1.4655172413793103,
-      "grad_norm": 1.390625,
-      "learning_rate": 0.00019540229885057472,
-      "loss": 0.8878,
+      "epoch": 5.433789954337899,
+      "grad_norm": 0.90625,
+      "learning_rate": 0.00010144104766775572,
+      "loss": 0.9126,
       "step": 595
     },
     {
-      "epoch": 1.477832512315271,
+      "epoch": 5.47945205479452,
       "grad_norm": 1.171875,
-      "learning_rate": 0.00019704433497536947,
-      "loss": 0.8862,
+      "learning_rate": 9.983987811869862e-05,
+      "loss": 0.9177,
       "step": 600
     },
     {
-      "epoch": 1.4901477832512315,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00019868637110016422,
-      "loss": 0.8774,
+      "epoch": 5.525114155251142,
+      "grad_norm": 0.53515625,
+      "learning_rate": 9.823874962247564e-05,
+      "loss": 0.9089,
       "step": 605
     },
     {
-      "epoch": 1.5024630541871922,
-      "grad_norm": 1.5703125,
-      "learning_rate": 0.00019999998357330727,
-      "loss": 0.8948,
+      "epoch": 5.570776255707763,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.663807268427198e-05,
+      "loss": 0.9112,
       "step": 610
     },
     {
-      "epoch": 1.5147783251231526,
-      "grad_norm": 5.6875,
-      "learning_rate": 0.00019999940863962815,
-      "loss": 0.8775,
+      "epoch": 5.616438356164384,
+      "grad_norm": 0.69140625,
+      "learning_rate": 9.503825769350017e-05,
+      "loss": 0.9142,
       "step": 615
     },
     {
-      "epoch": 1.5270935960591134,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00019999801237670888,
-      "loss": 0.8641,
+      "epoch": 5.662100456621005,
+      "grad_norm": 0.8359375,
+      "learning_rate": 9.343971481858246e-05,
+      "loss": 0.9068,
       "step": 620
     },
     {
-      "epoch": 1.5394088669950738,
-      "grad_norm": 3.828125,
-      "learning_rate": 0.00019999579479601748,
-      "loss": 0.854,
+      "epoch": 5.707762557077626,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.184285390178978e-05,
+      "loss": 0.9134,
       "step": 625
     },
     {
-      "epoch": 1.5517241379310345,
-      "grad_norm": 1.9375,
-      "learning_rate": 0.00019999275591576766,
-      "loss": 0.8795,
+      "epoch": 5.7534246575342465,
+      "grad_norm": 1.890625,
+      "learning_rate": 9.024808435416434e-05,
+      "loss": 0.9106,
       "step": 630
     },
     {
-      "epoch": 1.564039408866995,
-      "grad_norm": 1.140625,
-      "learning_rate": 0.00019998889576091885,
-      "loss": 0.8771,
+      "epoch": 5.799086757990867,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.865581505055291e-05,
+      "loss": 0.9108,
       "step": 635
     },
     {
-      "epoch": 1.5763546798029555,
-      "grad_norm": 1.453125,
-      "learning_rate": 0.00019998421436317573,
-      "loss": 0.8707,
+      "epoch": 5.844748858447488,
+      "grad_norm": 1.6015625,
+      "learning_rate": 8.706645422477739e-05,
+      "loss": 0.9027,
       "step": 640
     },
     {
-      "epoch": 1.5886699507389164,
-      "grad_norm": 2.203125,
-      "learning_rate": 0.00019997871176098827,
-      "loss": 0.8698,
+      "epoch": 5.890410958904109,
+      "grad_norm": 4.90625,
+      "learning_rate": 8.548040936496989e-05,
+      "loss": 0.9217,
       "step": 645
     },
     {
-      "epoch": 1.6009852216748768,
-      "grad_norm": 0.7890625,
-      "learning_rate": 0.0001999723879995512,
-      "loss": 0.8751,
+      "epoch": 5.936073059360731,
+      "grad_norm": 7.09375,
+      "learning_rate": 8.389808710909881e-05,
+      "loss": 0.9227,
       "step": 650
     },
     {
-      "epoch": 1.6133004926108374,
-      "grad_norm": 1.5234375,
-      "learning_rate": 0.00019996524313080377,
-      "loss": 0.8577,
+      "epoch": 5.981735159817352,
+      "grad_norm": 6.625,
+      "learning_rate": 8.231989314071317e-05,
+      "loss": 0.9157,
       "step": 655
     },
     {
-      "epoch": 1.625615763546798,
-      "grad_norm": 1.15625,
-      "learning_rate": 0.00019995727721342914,
-      "loss": 0.8654,
+      "epoch": 6.0,
+      "eval_loss": 2.46421480178833,
+      "eval_runtime": 0.2356,
+      "eval_samples_per_second": 42.441,
+      "eval_steps_per_second": 4.244,
+      "step": 657
+    },
+    {
+      "epoch": 6.027397260273973,
+      "grad_norm": 0.7109375,
+      "learning_rate": 8.07462320849313e-05,
+      "loss": 0.902,
       "step": 660
     },
     {
-      "epoch": 1.6379310344827587,
-      "grad_norm": 1.515625,
-      "learning_rate": 0.00019994849031285415,
-      "loss": 0.8664,
+      "epoch": 6.073059360730594,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.917750740470117e-05,
+      "loss": 0.8855,
       "step": 665
     },
     {
-      "epoch": 1.6502463054187193,
-      "grad_norm": 3.328125,
-      "learning_rate": 0.00019993888250124866,
-      "loss": 0.8505,
+      "epoch": 6.1187214611872145,
+      "grad_norm": 1.25,
+      "learning_rate": 7.761412129735852e-05,
+      "loss": 0.9014,
       "step": 670
     },
     {
-      "epoch": 1.6625615763546797,
-      "grad_norm": 2.84375,
-      "learning_rate": 0.00019992845385752485,
-      "loss": 0.8557,
+      "epoch": 6.164383561643835,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.605647459150961e-05,
+      "loss": 0.8863,
       "step": 675
     },
     {
-      "epoch": 1.6748768472906403,
-      "grad_norm": 1.6875,
-      "learning_rate": 0.0001999172044673367,
-      "loss": 0.8677,
+      "epoch": 6.210045662100456,
+      "grad_norm": 1.296875,
+      "learning_rate": 7.450496664426477e-05,
+      "loss": 0.8804,
       "step": 680
     },
     {
-      "epoch": 1.687192118226601,
-      "grad_norm": 1.703125,
-      "learning_rate": 0.00019990513442307933,
-      "loss": 0.8597,
+      "epoch": 6.255707762557078,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.295999523884921e-05,
+      "loss": 0.8795,
       "step": 685
     },
     {
-      "epoch": 1.6995073891625616,
-      "grad_norm": 2.421875,
-      "learning_rate": 0.00019989224382388813,
-      "loss": 0.8773,
+      "epoch": 6.301369863013699,
+      "grad_norm": 0.890625,
+      "learning_rate": 7.142195648261747e-05,
+      "loss": 0.8855,
       "step": 690
     },
     {
-      "epoch": 1.7118226600985222,
-      "grad_norm": 5.0625,
-      "learning_rate": 0.00019987853277563794,
-      "loss": 0.8669,
+      "epoch": 6.34703196347032,
+      "grad_norm": 0.671875,
+      "learning_rate": 6.989124470549745e-05,
+      "loss": 0.8799,
       "step": 695
     },
     {
-      "epoch": 1.7241379310344827,
-      "grad_norm": 1.53125,
-      "learning_rate": 0.00019986400139094236,
-      "loss": 0.8735,
+      "epoch": 6.392694063926941,
+      "grad_norm": 0.796875,
+      "learning_rate": 6.83682523588902e-05,
+      "loss": 0.8731,
       "step": 700
     },
     {
-      "epoch": 1.7364532019704435,
-      "grad_norm": 2.671875,
-      "learning_rate": 0.00019984864978915253,
-      "loss": 0.8679,
+      "epoch": 6.438356164383562,
+      "grad_norm": 2.015625,
+      "learning_rate": 6.685336991505122e-05,
+      "loss": 0.8818,
       "step": 705
     },
     {
-      "epoch": 1.748768472906404,
-      "grad_norm": 2.609375,
-      "learning_rate": 0.00019983247809635644,
-      "loss": 0.8775,
+      "epoch": 6.4840182648401825,
+      "grad_norm": 1.4140625,
+      "learning_rate": 6.534698576697939e-05,
+      "loss": 0.8792,
       "step": 710
     },
     {
-      "epoch": 1.7610837438423645,
-      "grad_norm": 1.671875,
-      "learning_rate": 0.00019981548644537766,
-      "loss": 0.8518,
+      "epoch": 6.529680365296803,
+      "grad_norm": 0.8125,
+      "learning_rate": 6.384948612883873e-05,
+      "loss": 0.8713,
       "step": 715
     },
     {
-      "epoch": 1.7733990147783252,
-      "grad_norm": 3.765625,
-      "learning_rate": 0.00019979767497577445,
-      "loss": 0.8451,
+      "epoch": 6.575342465753424,
+      "grad_norm": 0.53515625,
+      "learning_rate": 6.2361254936939e-05,
+      "loss": 0.8762,
       "step": 720
     },
     {
-      "epoch": 1.7857142857142856,
-      "grad_norm": 0.921875,
-      "learning_rate": 0.0001997790438338385,
-      "loss": 0.8589,
+      "epoch": 6.621004566210045,
+      "grad_norm": 0.515625,
+      "learning_rate": 6.088267375130023e-05,
+      "loss": 0.8708,
       "step": 725
     },
     {
-      "epoch": 1.7980295566502464,
-      "grad_norm": 2.375,
-      "learning_rate": 0.0001997595931725937,
-      "loss": 0.8545,
+      "epoch": 6.666666666666667,
+      "grad_norm": 0.70703125,
+      "learning_rate": 5.941412165782645e-05,
+      "loss": 0.8797,
       "step": 730
     },
     {
-      "epoch": 1.8103448275862069,
-      "grad_norm": 1.7265625,
-      "learning_rate": 0.000199739323151795,
-      "loss": 0.8611,
+      "epoch": 6.712328767123288,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.79559751711138e-05,
+      "loss": 0.8634,
       "step": 735
     },
     {
-      "epoch": 1.8226600985221675,
-      "grad_norm": 2.90625,
-      "learning_rate": 0.00019971823393792693,
-      "loss": 0.8287,
+      "epoch": 6.757990867579909,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.650860813791785e-05,
+      "loss": 0.872,
       "step": 740
     },
     {
-      "epoch": 1.8349753694581281,
-      "grad_norm": 2.984375,
-      "learning_rate": 0.00019969632570420248,
-      "loss": 0.8651,
+      "epoch": 6.80365296803653,
+      "grad_norm": 0.625,
+      "learning_rate": 5.507239164130501e-05,
+      "loss": 0.8661,
       "step": 745
     },
     {
-      "epoch": 1.8472906403940885,
-      "grad_norm": 3.234375,
-      "learning_rate": 0.00019967359863056134,
-      "loss": 0.8459,
+      "epoch": 6.8493150684931505,
+      "grad_norm": 0.828125,
+      "learning_rate": 5.364769390551225e-05,
+      "loss": 0.8744,
       "step": 750
     },
     {
-      "epoch": 1.8596059113300494,
-      "grad_norm": 0.8203125,
-      "learning_rate": 0.0001996500529036688,
-      "loss": 0.8434,
+      "epoch": 6.894977168949771,
+      "grad_norm": 1.5234375,
+      "learning_rate": 5.2234880201540284e-05,
+      "loss": 0.8662,
       "step": 755
     },
     {
-      "epoch": 1.8719211822660098,
-      "grad_norm": 1.7734375,
-      "learning_rate": 0.0001996256887169139,
-      "loss": 0.85,
+      "epoch": 6.940639269406392,
+      "grad_norm": 0.94140625,
+      "learning_rate": 5.0834312753503124e-05,
+      "loss": 0.8764,
       "step": 760
     },
     {
-      "epoch": 1.8842364532019704,
-      "grad_norm": 2.953125,
-      "learning_rate": 0.00019960050627040806,
-      "loss": 0.8468,
+      "epoch": 6.986301369863014,
+      "grad_norm": 0.4609375,
+      "learning_rate": 4.9446350645759885e-05,
+      "loss": 0.8657,
       "step": 765
     },
     {
-      "epoch": 1.896551724137931,
-      "grad_norm": 2.0,
-      "learning_rate": 0.00019957450577098322,
-      "loss": 0.8825,
+      "epoch": 6.995433789954338,
+      "eval_loss": 2.5075883865356445,
+      "eval_runtime": 0.2561,
+      "eval_samples_per_second": 39.047,
+      "eval_steps_per_second": 3.905,
+      "step": 766
+    },
+    {
+      "epoch": 7.031963470319635,
+      "grad_norm": 0.408203125,
+      "learning_rate": 4.807134973085036e-05,
+      "loss": 0.8614,
       "step": 770
     },
     {
-      "epoch": 1.9088669950738915,
-      "grad_norm": 1.53125,
-      "learning_rate": 0.00019954768743219044,
-      "loss": 0.8443,
+      "epoch": 7.077625570776256,
+      "grad_norm": 0.56640625,
+      "learning_rate": 4.6709662538260267e-05,
+      "loss": 0.8457,
       "step": 775
     },
     {
-      "epoch": 1.9211822660098523,
-      "grad_norm": 1.6640625,
-      "learning_rate": 0.0001995200514742978,
-      "loss": 0.8399,
+      "epoch": 7.123287671232877,
+      "grad_norm": 0.447265625,
+      "learning_rate": 4.53616381840377e-05,
+      "loss": 0.8502,
       "step": 780
     },
     {
-      "epoch": 1.9334975369458127,
-      "grad_norm": 2.328125,
-      "learning_rate": 0.00019949159812428889,
-      "loss": 0.8544,
+      "epoch": 7.168949771689498,
+      "grad_norm": 0.64453125,
+      "learning_rate": 4.402762228128531e-05,
+      "loss": 0.8536,
       "step": 785
     },
     {
-      "epoch": 1.9458128078817734,
-      "grad_norm": 0.8125,
-      "learning_rate": 0.00019946232761586073,
-      "loss": 0.8334,
+      "epoch": 7.2146118721461185,
+      "grad_norm": 0.78515625,
+      "learning_rate": 4.2707956851550016e-05,
+      "loss": 0.8531,
       "step": 790
     },
     {
-      "epoch": 1.958128078817734,
-      "grad_norm": 0.984375,
-      "learning_rate": 0.0001994322401894221,
-      "loss": 0.8379,
+      "epoch": 7.260273972602739,
+      "grad_norm": 0.68359375,
+      "learning_rate": 4.140298023713416e-05,
+      "loss": 0.8609,
       "step": 795
     },
     {
-      "epoch": 1.9704433497536946,
-      "grad_norm": 1.0,
-      "learning_rate": 0.00019940133609209118,
-      "loss": 0.8502,
+      "epoch": 7.30593607305936,
+      "grad_norm": 0.671875,
+      "learning_rate": 4.011302701434937e-05,
+      "loss": 0.8529,
       "step": 800
     },
     {
-      "epoch": 1.9827586206896552,
-      "grad_norm": 1.46875,
-      "learning_rate": 0.00019936961557769385,
-      "loss": 0.8428,
+      "epoch": 7.351598173515982,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.8838427907736476e-05,
+      "loss": 0.8566,
       "step": 805
     },
     {
-      "epoch": 1.9950738916256157,
-      "grad_norm": 2.140625,
-      "learning_rate": 0.00019933707890676158,
-      "loss": 0.8399,
+      "epoch": 7.397260273972603,
+      "grad_norm": 0.6171875,
+      "learning_rate": 3.757950970527249e-05,
+      "loss": 0.8508,
       "step": 810
     },
     {
-      "epoch": 2.0,
-      "eval_loss": 2.3512892723083496,
-      "eval_runtime": 2.0459,
-      "eval_samples_per_second": 4.888,
-      "eval_steps_per_second": 0.978,
-      "step": 812
-    },
-    {
-      "epoch": 2.0073891625615765,
-      "grad_norm": 11.5,
-      "learning_rate": 0.00019930372634652913,
-      "loss": 0.8116,
+      "epoch": 7.442922374429224,
+      "grad_norm": 0.435546875,
+      "learning_rate": 3.633659517458736e-05,
+      "loss": 0.8513,
       "step": 815
     },
     {
-      "epoch": 2.019704433497537,
-      "grad_norm": 1.46875,
-      "learning_rate": 0.00019926955817093243,
-      "loss": 0.7747,
+      "epoch": 7.488584474885845,
+      "grad_norm": 0.466796875,
+      "learning_rate": 3.5110002980210975e-05,
+      "loss": 0.8499,
       "step": 820
     },
     {
-      "epoch": 2.0320197044334973,
-      "grad_norm": 1.078125,
-      "learning_rate": 0.00019923457466060636,
-      "loss": 0.7982,
+      "epoch": 7.534246575342466,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.3900047601872596e-05,
+      "loss": 0.8493,
       "step": 825
     },
     {
-      "epoch": 2.044334975369458,
-      "grad_norm": 1.5859375,
-      "learning_rate": 0.0001991987761028824,
-      "loss": 0.7727,
+      "epoch": 7.579908675799087,
+      "grad_norm": 0.6171875,
+      "learning_rate": 3.270703925387279e-05,
+      "loss": 0.851,
       "step": 830
     },
     {
-      "epoch": 2.0566502463054186,
-      "grad_norm": 1.953125,
-      "learning_rate": 0.0001991621627917864,
-      "loss": 0.7962,
+      "epoch": 7.6255707762557075,
+      "grad_norm": 0.5234375,
+      "learning_rate": 3.153128380554941e-05,
+      "loss": 0.8452,
       "step": 835
     },
     {
-      "epoch": 2.0689655172413794,
-      "grad_norm": 0.9765625,
-      "learning_rate": 0.00019912473502803582,
-      "loss": 0.783,
+      "epoch": 7.671232876712329,
+      "grad_norm": 0.43359375,
+      "learning_rate": 3.037308270285709e-05,
+      "loss": 0.862,
       "step": 840
     },
     {
-      "epoch": 2.08128078817734,
-      "grad_norm": 1.28125,
-      "learning_rate": 0.00019908649311903775,
-      "loss": 0.7814,
+      "epoch": 7.71689497716895,
+      "grad_norm": 0.625,
+      "learning_rate": 2.923273289108115e-05,
+      "loss": 0.8487,
       "step": 845
     },
     {
-      "epoch": 2.0935960591133007,
-      "grad_norm": 2.09375,
-      "learning_rate": 0.00019904743737888603,
-      "loss": 0.791,
+      "epoch": 7.762557077625571,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.8110526738705344e-05,
+      "loss": 0.8516,
       "step": 850
     },
     {
-      "epoch": 2.105911330049261,
-      "grad_norm": 1.3828125,
-      "learning_rate": 0.0001990075681283587,
-      "loss": 0.7831,
+      "epoch": 7.808219178082192,
+      "grad_norm": 0.546875,
+      "learning_rate": 2.7006751962452882e-05,
+      "loss": 0.8541,
       "step": 855
     },
     {
-      "epoch": 2.1182266009852215,
-      "grad_norm": 1.3984375,
-      "learning_rate": 0.00019896688569491557,
-      "loss": 0.778,
+      "epoch": 7.853881278538813,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.592169155352031e-05,
+      "loss": 0.8486,
       "step": 860
     },
     {
-      "epoch": 2.1305418719211824,
-      "grad_norm": 0.77734375,
-      "learning_rate": 0.00019892539041269533,
-      "loss": 0.7664,
+      "epoch": 7.899543378995434,
+      "grad_norm": 0.86328125,
+      "learning_rate": 2.485562370502279e-05,
+      "loss": 0.8402,
       "step": 865
     },
     {
-      "epoch": 2.142857142857143,
-      "grad_norm": 1.265625,
-      "learning_rate": 0.00019888308262251285,
-      "loss": 0.7849,
+      "epoch": 7.945205479452055,
+      "grad_norm": 0.412109375,
+      "learning_rate": 2.3808821740669606e-05,
+      "loss": 0.8474,
       "step": 870
     },
     {
-      "epoch": 2.1551724137931036,
-      "grad_norm": 1.4453125,
-      "learning_rate": 0.0001988399626718565,
-      "loss": 0.7774,
+      "epoch": 7.9908675799086755,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.2781554044688015e-05,
+      "loss": 0.8393,
       "step": 875
     },
     {
-      "epoch": 2.167487684729064,
-      "grad_norm": 2.078125,
-      "learning_rate": 0.00019879603091488504,
-      "loss": 0.7803,
+      "epoch": 8.0,
+      "eval_loss": 2.515906810760498,
+      "eval_runtime": 0.2401,
+      "eval_samples_per_second": 41.645,
+      "eval_steps_per_second": 4.164,
+      "step": 876
+    },
+    {
+      "epoch": 8.036529680365296,
+      "grad_norm": 0.5390625,
+      "learning_rate": 2.1774083993013718e-05,
+      "loss": 0.8404,
       "step": 880
     },
     {
-      "epoch": 2.1798029556650245,
-      "grad_norm": 2.515625,
-      "learning_rate": 0.00019875128771242506,
-      "loss": 0.7782,
+      "epoch": 8.082191780821917,
+      "grad_norm": 0.5234375,
+      "learning_rate": 2.078666988576504e-05,
+      "loss": 0.837,
       "step": 885
     },
     {
-      "epoch": 2.1921182266009853,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.0001987057334319677,
-      "loss": 0.7883,
+      "epoch": 8.127853881278538,
+      "grad_norm": 0.58203125,
+      "learning_rate": 1.9819564881018983e-05,
+      "loss": 0.8372,
       "step": 890
     },
     {
-      "epoch": 2.2044334975369457,
-      "grad_norm": 1.140625,
-      "learning_rate": 0.0001986593684476658,
-      "loss": 0.7829,
+      "epoch": 8.173515981735159,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.887301692990494e-05,
+      "loss": 0.846,
       "step": 895
     },
     {
-      "epoch": 2.2167487684729066,
-      "grad_norm": 0.76171875,
-      "learning_rate": 0.00019861219314033077,
-      "loss": 0.78,
+      "epoch": 8.219178082191782,
+      "grad_norm": 0.439453125,
+      "learning_rate": 1.7947268713034127e-05,
+      "loss": 0.8461,
       "step": 900
     },
     {
-      "epoch": 2.229064039408867,
-      "grad_norm": 0.8125,
-      "learning_rate": 0.00019856420789742953,
-      "loss": 0.7865,
+      "epoch": 8.264840182648403,
+      "grad_norm": 0.4140625,
+      "learning_rate": 1.7042557578279626e-05,
+      "loss": 0.8373,
       "step": 905
     },
     {
-      "epoch": 2.2413793103448274,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00019851541311308123,
-      "loss": 0.7731,
+      "epoch": 8.310502283105023,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.6159115479924257e-05,
+      "loss": 0.8422,
       "step": 910
     },
     {
-      "epoch": 2.2536945812807883,
-      "grad_norm": 1.109375,
-      "learning_rate": 0.000198465809188054,
-      "loss": 0.7705,
+      "epoch": 8.356164383561644,
+      "grad_norm": 0.4453125,
+      "learning_rate": 1.529716891919074e-05,
+      "loss": 0.8403,
       "step": 915
     },
     {
-      "epoch": 2.2660098522167487,
-      "grad_norm": 0.87890625,
-      "learning_rate": 0.00019841539652976192,
-      "loss": 0.7815,
+      "epoch": 8.401826484018265,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.4456938886170412e-05,
+      "loss": 0.8343,
       "step": 920
     },
     {
-      "epoch": 2.2783251231527095,
-      "grad_norm": 1.3984375,
-      "learning_rate": 0.00019836417555226129,
-      "loss": 0.781,
+      "epoch": 8.447488584474886,
+      "grad_norm": 0.43359375,
+      "learning_rate": 1.3638640803164516e-05,
+      "loss": 0.8355,
       "step": 925
     },
     {
-      "epoch": 2.29064039408867,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.0001983121466762474,
-      "loss": 0.7929,
+      "epoch": 8.493150684931507,
+      "grad_norm": 0.41796875,
+      "learning_rate": 1.2842484469453365e-05,
+      "loss": 0.841,
       "step": 930
     },
     {
-      "epoch": 2.302955665024631,
-      "grad_norm": 1.6328125,
-      "learning_rate": 0.0001982593103290512,
-      "loss": 0.7992,
+      "epoch": 8.538812785388128,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.2068674007506786e-05,
+      "loss": 0.8396,
       "step": 935
     },
     {
-      "epoch": 2.315270935960591,
-      "grad_norm": 1.2421875,
-      "learning_rate": 0.00019820566694463566,
-      "loss": 0.8023,
+      "epoch": 8.584474885844749,
+      "grad_norm": 0.40625,
+      "learning_rate": 1.1317407810650372e-05,
+      "loss": 0.8377,
       "step": 940
     },
     {
-      "epoch": 2.3275862068965516,
-      "grad_norm": 1.09375,
-      "learning_rate": 0.00019815121696359212,
-      "loss": 0.7887,
+      "epoch": 8.63013698630137,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.058887849220026e-05,
+      "loss": 0.8348,
       "step": 945
     },
     {
-      "epoch": 2.3399014778325125,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.0001980959608331369,
-      "loss": 0.7807,
+      "epoch": 8.67579908675799,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.883272836080116e-06,
+      "loss": 0.8388,
       "step": 950
     },
     {
-      "epoch": 2.352216748768473,
-      "grad_norm": 0.6640625,
-      "learning_rate": 0.00019803989900710734,
-      "loss": 0.7934,
+      "epoch": 8.721461187214611,
+      "grad_norm": 0.396484375,
+      "learning_rate": 9.200771748932513e-06,
+      "loss": 0.8366,
       "step": 955
     },
     {
-      "epoch": 2.3645320197044333,
-      "grad_norm": 1.4453125,
-      "learning_rate": 0.00019798303194595846,
-      "loss": 0.7867,
+      "epoch": 8.767123287671232,
+      "grad_norm": 0.416015625,
+      "learning_rate": 8.541550213737171e-06,
+      "loss": 0.8436,
       "step": 960
     },
     {
-      "epoch": 2.376847290640394,
-      "grad_norm": 0.78515625,
-      "learning_rate": 0.0001979253601167588,
-      "loss": 0.7571,
+      "epoch": 8.812785388127853,
+      "grad_norm": 0.455078125,
+      "learning_rate": 7.905777244947954e-06,
+      "loss": 0.8409,
       "step": 965
     },
     {
-      "epoch": 2.3891625615763545,
-      "grad_norm": 1.640625,
-      "learning_rate": 0.00019786688399318664,
-      "loss": 0.7879,
+      "epoch": 8.858447488584474,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.293615845160196e-06,
+      "loss": 0.8377,
       "step": 970
     },
     {
-      "epoch": 2.4014778325123154,
-      "grad_norm": 1.9140625,
-      "learning_rate": 0.00019780760405552645,
-      "loss": 0.7975,
+      "epoch": 8.904109589041095,
+      "grad_norm": 0.408203125,
+      "learning_rate": 6.705222963319191e-06,
+      "loss": 0.8425,
       "step": 975
     },
     {
-      "epoch": 2.413793103448276,
-      "grad_norm": 1.1171875,
-      "learning_rate": 0.00019774752079066452,
-      "loss": 0.7924,
+      "epoch": 8.949771689497716,
+      "grad_norm": 0.419921875,
+      "learning_rate": 6.140749454480932e-06,
+      "loss": 0.8371,
       "step": 980
     },
     {
-      "epoch": 2.4261083743842367,
-      "grad_norm": 2.34375,
-      "learning_rate": 0.0001976866346920852,
-      "loss": 0.7733,
+      "epoch": 8.995433789954339,
+      "grad_norm": 0.408203125,
+      "learning_rate": 5.6003400411351325e-06,
+      "loss": 0.8462,
+      "step": 985
+    },
+    {
+      "epoch": 8.995433789954339,
+      "eval_loss": 2.518533706665039,
+      "eval_runtime": 0.2546,
+      "eval_samples_per_second": 39.282,
+      "eval_steps_per_second": 3.928,
       "step": 985
     },
     {
-      "epoch": 2.438423645320197,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00019762494625986677,
-      "loss": 0.7886,
+      "epoch": 9.04109589041096,
+      "grad_norm": 0.392578125,
+      "learning_rate": 5.0841332761005e-06,
+      "loss": 0.8404,
       "step": 990
     },
     {
-      "epoch": 2.4507389162561575,
-      "grad_norm": 0.9921875,
-      "learning_rate": 0.00019756245600067738,
-      "loss": 0.7751,
+      "epoch": 9.08675799086758,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.592261507001993e-06,
+      "loss": 0.8303,
       "step": 995
     },
     {
-      "epoch": 2.4630541871921183,
-      "grad_norm": 0.81640625,
-      "learning_rate": 0.00019749916442777078,
-      "loss": 0.7845,
+      "epoch": 9.132420091324201,
+      "grad_norm": 0.42578125,
+      "learning_rate": 4.124850842338779e-06,
+      "loss": 0.8325,
       "step": 1000
     },
     {
-      "epoch": 2.4753694581280787,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00019743507206098233,
-      "loss": 0.7755,
+      "epoch": 9.178082191780822,
+      "grad_norm": 0.4140625,
+      "learning_rate": 3.6820211191520125e-06,
+      "loss": 0.8353,
       "step": 1005
     },
     {
-      "epoch": 2.4876847290640396,
-      "grad_norm": 0.87890625,
-      "learning_rate": 0.00019737017942672442,
-      "loss": 0.7751,
+      "epoch": 9.223744292237443,
+      "grad_norm": 0.419921875,
+      "learning_rate": 3.263885872300343e-06,
+      "loss": 0.8347,
       "step": 1010
     },
     {
-      "epoch": 2.5,
-      "grad_norm": 1.515625,
-      "learning_rate": 0.00019730448705798239,
-      "loss": 0.7804,
+      "epoch": 9.269406392694064,
+      "grad_norm": 0.3984375,
+      "learning_rate": 2.8705523053513816e-06,
+      "loss": 0.8329,
       "step": 1015
     },
     {
-      "epoch": 2.512315270935961,
-      "grad_norm": 1.140625,
-      "learning_rate": 0.00019723799549431007,
-      "loss": 0.7793,
+      "epoch": 9.315068493150685,
+      "grad_norm": 0.421875,
+      "learning_rate": 2.502121263096224e-06,
+      "loss": 0.8369,
       "step": 1020
     },
     {
-      "epoch": 2.5246305418719213,
-      "grad_norm": 0.65234375,
-      "learning_rate": 0.00019717070528182538,
-      "loss": 0.7916,
+      "epoch": 9.360730593607306,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.1586872056944428e-06,
+      "loss": 0.8324,
       "step": 1025
     },
     {
-      "epoch": 2.5369458128078817,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00019710261697320573,
-      "loss": 0.7927,
+      "epoch": 9.406392694063927,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.840338184455881e-06,
+      "loss": 0.8383,
       "step": 1030
     },
     {
-      "epoch": 2.5492610837438425,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.00019703373112768365,
-      "loss": 0.7879,
+      "epoch": 9.452054794520548,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.5471558192656777e-06,
+      "loss": 0.8315,
       "step": 1035
     },
     {
-      "epoch": 2.561576354679803,
-      "grad_norm": 0.88671875,
-      "learning_rate": 0.00019696404831104204,
-      "loss": 0.7921,
+      "epoch": 9.497716894977168,
+      "grad_norm": 0.435546875,
+      "learning_rate": 1.2792152776580968e-06,
+      "loss": 0.8437,
       "step": 1040
     },
     {
-      "epoch": 2.5738916256157633,
-      "grad_norm": 1.6953125,
-      "learning_rate": 0.00019689356909560965,
-      "loss": 0.7809,
+      "epoch": 9.54337899543379,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.036585255544764e-06,
+      "loss": 0.8418,
       "step": 1045
     },
     {
-      "epoch": 2.586206896551724,
-      "grad_norm": 1.546875,
-      "learning_rate": 0.00019682229406025635,
-      "loss": 0.8084,
+      "epoch": 9.58904109589041,
+      "grad_norm": 0.400390625,
+      "learning_rate": 8.193279596020121e-07,
+      "loss": 0.8346,
       "step": 1050
     },
     {
-      "epoch": 2.5985221674876846,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00019675022379038822,
-      "loss": 0.7885,
+      "epoch": 9.634703196347033,
+      "grad_norm": 0.40234375,
+      "learning_rate": 6.274990913221035e-07,
+      "loss": 0.8415,
       "step": 1055
     },
     {
-      "epoch": 2.6108374384236455,
-      "grad_norm": 0.640625,
-      "learning_rate": 0.00019667735887794304,
-      "loss": 0.7935,
+      "epoch": 9.680365296803654,
+      "grad_norm": 0.416015625,
+      "learning_rate": 4.6114783273213393e-07,
+      "loss": 0.8339,
       "step": 1060
     },
     {
-      "epoch": 2.623152709359606,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00019660369992138517,
-      "loss": 0.7833,
+      "epoch": 9.726027397260275,
+      "grad_norm": 0.439453125,
+      "learning_rate": 3.203168337845508e-07,
+      "loss": 0.8331,
       "step": 1065
     },
     {
-      "epoch": 2.6354679802955667,
-      "grad_norm": 0.72265625,
-      "learning_rate": 0.0001965292475257007,
-      "loss": 0.7838,
+      "epoch": 9.771689497716896,
+      "grad_norm": 0.400390625,
+      "learning_rate": 2.05042201422323e-07,
+      "loss": 0.8458,
       "step": 1070
     },
     {
-      "epoch": 2.647783251231527,
-      "grad_norm": 0.890625,
-      "learning_rate": 0.00019645400230239256,
-      "loss": 0.7751,
+      "epoch": 9.817351598173516,
+      "grad_norm": 0.41015625,
+      "learning_rate": 1.1535349032167908e-07,
+      "loss": 0.8444,
       "step": 1075
     },
     {
-      "epoch": 2.6600985221674875,
-      "grad_norm": 0.94921875,
-      "learning_rate": 0.0001963779648694754,
-      "loss": 0.781,
+      "epoch": 9.863013698630137,
+      "grad_norm": 0.400390625,
+      "learning_rate": 5.127369531473525e-08,
+      "loss": 0.8486,
       "step": 1080
     },
     {
-      "epoch": 2.6724137931034484,
-      "grad_norm": 1.046875,
-      "learning_rate": 0.00019630113585147063,
-      "loss": 0.7858,
+      "epoch": 9.908675799086758,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.2819245493955744e-08,
+      "loss": 0.8473,
       "step": 1085
     },
     {
-      "epoch": 2.684729064039409,
-      "grad_norm": 0.734375,
-      "learning_rate": 0.00019622351587940116,
-      "loss": 0.7882,
+      "epoch": 9.954337899543379,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.0,
+      "loss": 0.8359,
       "step": 1090
     },
     {
-      "epoch": 2.697044334975369,
-      "grad_norm": 0.9453125,
-      "learning_rate": 0.00019614510559078625,
-      "loss": 0.7935,
-      "step": 1095
-    },
-    {
-      "epoch": 2.70935960591133,
-      "grad_norm": 1.2734375,
-      "learning_rate": 0.00019606590562963637,
-      "loss": 0.7805,
-      "step": 1100
-    },
-    {
-      "epoch": 2.7216748768472905,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.00019598591664644782,
-      "loss": 0.7713,
-      "step": 1105
-    },
-    {
-      "epoch": 2.7339901477832513,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.00019590513929819734,
-      "loss": 0.7954,
-      "step": 1110
-    },
-    {
-      "epoch": 2.7463054187192117,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.0001958235742483369,
-      "loss": 0.7897,
-      "step": 1115
-    },
-    {
-      "epoch": 2.7586206896551726,
-      "grad_norm": 0.7890625,
-      "learning_rate": 0.00019574122216678799,
-      "loss": 0.7809,
-      "step": 1120
-    },
-    {
-      "epoch": 2.770935960591133,
-      "grad_norm": 1.171875,
-      "learning_rate": 0.0001956580837299364,
-      "loss": 0.7773,
-      "step": 1125
-    },
-    {
-      "epoch": 2.7832512315270934,
-      "grad_norm": 0.984375,
-      "learning_rate": 0.00019557415962062643,
-      "loss": 0.7958,
-      "step": 1130
-    },
-    {
-      "epoch": 2.7955665024630543,
-      "grad_norm": 1.078125,
-      "learning_rate": 0.0001954894505281554,
-      "loss": 0.7782,
-      "step": 1135
-    },
-    {
-      "epoch": 2.8078817733990147,
-      "grad_norm": 1.2890625,
-      "learning_rate": 0.000195403957148268,
-      "loss": 0.7704,
-      "step": 1140
-    },
-    {
-      "epoch": 2.8201970443349755,
-      "grad_norm": 0.703125,
-      "learning_rate": 0.0001953176801831505,
-      "loss": 0.7934,
-      "step": 1145
-    },
-    {
-      "epoch": 2.832512315270936,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.000195230620341425,
-      "loss": 0.7692,
-      "step": 1150
-    },
-    {
-      "epoch": 2.844827586206897,
-      "grad_norm": 0.72265625,
-      "learning_rate": 0.0001951427783381437,
-      "loss": 0.7775,
-      "step": 1155
-    },
-    {
-      "epoch": 2.857142857142857,
-      "grad_norm": 0.89453125,
-      "learning_rate": 0.0001950541548947829,
-      "loss": 0.7742,
-      "step": 1160
-    },
-    {
-      "epoch": 2.8694581280788176,
-      "grad_norm": 0.73828125,
-      "learning_rate": 0.0001949647507392372,
-      "loss": 0.7732,
-      "step": 1165
-    },
-    {
-      "epoch": 2.8817733990147785,
-      "grad_norm": 0.828125,
-      "learning_rate": 0.0001948745666058134,
-      "loss": 0.7918,
-      "step": 1170
-    },
-    {
-      "epoch": 2.894088669950739,
-      "grad_norm": 1.6875,
-      "learning_rate": 0.00019478360323522446,
-      "loss": 0.7684,
-      "step": 1175
-    },
-    {
-      "epoch": 2.9064039408866993,
-      "grad_norm": 0.9296875,
-      "learning_rate": 0.00019469186137458365,
-      "loss": 0.785,
-      "step": 1180
-    },
-    {
-      "epoch": 2.91871921182266,
-      "grad_norm": 1.1875,
-      "learning_rate": 0.00019459934177739813,
-      "loss": 0.7781,
-      "step": 1185
-    },
-    {
-      "epoch": 2.9310344827586206,
-      "grad_norm": 3.625,
-      "learning_rate": 0.0001945060452035629,
-      "loss": 0.7878,
-      "step": 1190
-    },
-    {
-      "epoch": 2.9433497536945814,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.0001944119724193545,
-      "loss": 0.7709,
-      "step": 1195
-    },
-    {
-      "epoch": 2.955665024630542,
-      "grad_norm": 1.609375,
-      "learning_rate": 0.00019431712419742484,
-      "loss": 0.7821,
-      "step": 1200
-    },
-    {
-      "epoch": 2.9679802955665027,
-      "grad_norm": 0.98828125,
-      "learning_rate": 0.00019422150131679467,
-      "loss": 0.7925,
-      "step": 1205
-    },
-    {
-      "epoch": 2.980295566502463,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.00019412510456284733,
-      "loss": 0.7867,
-      "step": 1210
-    },
-    {
-      "epoch": 2.9926108374384235,
-      "grad_norm": 1.2421875,
-      "learning_rate": 0.00019402793472732217,
-      "loss": 0.7898,
-      "step": 1215
-    },
-    {
-      "epoch": 3.0,
-      "eval_loss": 2.431116819381714,
-      "eval_runtime": 2.0437,
-      "eval_samples_per_second": 4.893,
-      "eval_steps_per_second": 0.979,
-      "step": 1218
-    },
-    {
-      "epoch": 3.0049261083743843,
-      "grad_norm": 0.984375,
-      "learning_rate": 0.00019392999260830828,
-      "loss": 0.7424,
-      "step": 1220
-    },
-    {
-      "epoch": 3.0172413793103448,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.0001938312790102376,
-      "loss": 0.6782,
-      "step": 1225
-    },
-    {
-      "epoch": 3.0295566502463056,
-      "grad_norm": 0.6796875,
-      "learning_rate": 0.00019373179474387858,
-      "loss": 0.7011,
-      "step": 1230
-    },
-    {
-      "epoch": 3.041871921182266,
-      "grad_norm": 0.96875,
-      "learning_rate": 0.0001936315406263295,
-      "loss": 0.6979,
-      "step": 1235
-    },
-    {
-      "epoch": 3.0541871921182264,
-      "grad_norm": 0.91796875,
-      "learning_rate": 0.0001935305174810115,
-      "loss": 0.6945,
-      "step": 1240
-    },
-    {
-      "epoch": 3.0665024630541873,
-      "grad_norm": 1.359375,
-      "learning_rate": 0.0001934287261376622,
-      "loss": 0.7051,
-      "step": 1245
-    },
-    {
-      "epoch": 3.0788177339901477,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.00019332616743232857,
-      "loss": 0.7047,
-      "step": 1250
-    },
-    {
-      "epoch": 3.0911330049261085,
-      "grad_norm": 1.2109375,
-      "learning_rate": 0.00019322284220736022,
-      "loss": 0.7062,
-      "step": 1255
-    },
-    {
-      "epoch": 3.103448275862069,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.00019311875131140246,
-      "loss": 0.6851,
-      "step": 1260
-    },
-    {
-      "epoch": 3.1157635467980294,
-      "grad_norm": 0.78125,
-      "learning_rate": 0.00019301389559938934,
-      "loss": 0.7069,
-      "step": 1265
-    },
-    {
-      "epoch": 3.12807881773399,
-      "grad_norm": 0.81640625,
-      "learning_rate": 0.00019290827593253655,
-      "loss": 0.6772,
-      "step": 1270
-    },
-    {
-      "epoch": 3.1403940886699506,
-      "grad_norm": 0.7109375,
-      "learning_rate": 0.00019280189317833445,
-      "loss": 0.6882,
-      "step": 1275
-    },
-    {
-      "epoch": 3.1527093596059115,
-      "grad_norm": 0.70703125,
-      "learning_rate": 0.00019269474821054084,
-      "loss": 0.6889,
-      "step": 1280
-    },
-    {
-      "epoch": 3.165024630541872,
-      "grad_norm": 0.80078125,
-      "learning_rate": 0.00019258684190917388,
-      "loss": 0.6993,
-      "step": 1285
-    },
-    {
-      "epoch": 3.1773399014778327,
-      "grad_norm": 0.94921875,
-      "learning_rate": 0.00019247817516050483,
-      "loss": 0.7013,
-      "step": 1290
-    },
-    {
-      "epoch": 3.189655172413793,
-      "grad_norm": 0.79296875,
-      "learning_rate": 0.00019236874885705075,
-      "loss": 0.6971,
-      "step": 1295
-    },
-    {
-      "epoch": 3.2019704433497536,
-      "grad_norm": 0.8359375,
-      "learning_rate": 0.00019225856389756718,
-      "loss": 0.7069,
-      "step": 1300
-    },
-    {
-      "epoch": 3.2142857142857144,
-      "grad_norm": 1.3984375,
-      "learning_rate": 0.00019214762118704076,
-      "loss": 0.7119,
-      "step": 1305
-    },
-    {
-      "epoch": 3.226600985221675,
-      "grad_norm": 0.78125,
-      "learning_rate": 0.00019203592163668184,
-      "loss": 0.7073,
-      "step": 1310
-    },
-    {
-      "epoch": 3.2389162561576357,
-      "grad_norm": 0.94140625,
-      "learning_rate": 0.0001919234661639168,
-      "loss": 0.7074,
-      "step": 1315
-    },
-    {
-      "epoch": 3.251231527093596,
-      "grad_norm": 0.65625,
-      "learning_rate": 0.0001918102556923809,
-      "loss": 0.7039,
-      "step": 1320
-    },
-    {
-      "epoch": 3.2635467980295565,
-      "grad_norm": 0.8671875,
-      "learning_rate": 0.00019169629115191027,
-      "loss": 0.7059,
-      "step": 1325
-    },
-    {
-      "epoch": 3.2758620689655173,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.0001915815734785346,
-      "loss": 0.7008,
-      "step": 1330
-    },
-    {
-      "epoch": 3.2881773399014778,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.0001914661036144692,
-      "loss": 0.7061,
-      "step": 1335
-    },
-    {
-      "epoch": 3.3004926108374386,
-      "grad_norm": 1.6640625,
-      "learning_rate": 0.00019134988250810747,
-      "loss": 0.7122,
-      "step": 1340
-    },
-    {
-      "epoch": 3.312807881773399,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00019123291111401299,
-      "loss": 0.7056,
-      "step": 1345
-    },
-    {
-      "epoch": 3.3251231527093594,
-      "grad_norm": 2.109375,
-      "learning_rate": 0.00019111519039291167,
-      "loss": 0.6945,
-      "step": 1350
-    },
-    {
-      "epoch": 3.3374384236453203,
-      "grad_norm": 1.140625,
-      "learning_rate": 0.00019099672131168397,
-      "loss": 0.7014,
-      "step": 1355
-    },
-    {
-      "epoch": 3.3497536945812807,
-      "grad_norm": 0.98046875,
-      "learning_rate": 0.0001908775048433568,
-      "loss": 0.7036,
-      "step": 1360
-    },
-    {
-      "epoch": 3.3620689655172415,
-      "grad_norm": 0.8359375,
-      "learning_rate": 0.00019075754196709572,
-      "loss": 0.6991,
-      "step": 1365
-    },
-    {
-      "epoch": 3.374384236453202,
-      "grad_norm": 0.74609375,
-      "learning_rate": 0.0001906368336681967,
-      "loss": 0.7058,
-      "step": 1370
-    },
-    {
-      "epoch": 3.386699507389163,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.00019051538093807816,
-      "loss": 0.6984,
-      "step": 1375
-    },
-    {
-      "epoch": 3.399014778325123,
-      "grad_norm": 1.734375,
-      "learning_rate": 0.0001903931847742728,
-      "loss": 0.7033,
-      "step": 1380
-    },
-    {
-      "epoch": 3.4113300492610836,
-      "grad_norm": 0.78125,
-      "learning_rate": 0.00019027024618041937,
-      "loss": 0.7006,
-      "step": 1385
-    },
-    {
-      "epoch": 3.4236453201970445,
-      "grad_norm": 0.8671875,
-      "learning_rate": 0.00019014656616625448,
-      "loss": 0.7249,
-      "step": 1390
-    },
-    {
-      "epoch": 3.435960591133005,
-      "grad_norm": 0.98828125,
-      "learning_rate": 0.00019002214574760423,
-      "loss": 0.6966,
-      "step": 1395
-    },
-    {
-      "epoch": 3.4482758620689653,
-      "grad_norm": 1.296875,
-      "learning_rate": 0.0001898969859463759,
-      "loss": 0.6932,
-      "step": 1400
-    },
-    {
-      "epoch": 3.460591133004926,
-      "grad_norm": 1.2578125,
-      "learning_rate": 0.00018977108779054974,
-      "loss": 0.7123,
-      "step": 1405
-    },
-    {
-      "epoch": 3.4729064039408866,
-      "grad_norm": 1.1328125,
-      "learning_rate": 0.0001896444523141701,
-      "loss": 0.7085,
-      "step": 1410
-    },
-    {
-      "epoch": 3.4852216748768474,
-      "grad_norm": 0.828125,
-      "learning_rate": 0.0001895170805573374,
-      "loss": 0.7035,
-      "step": 1415
-    },
-    {
-      "epoch": 3.497536945812808,
-      "grad_norm": 3.046875,
-      "learning_rate": 0.00018938897356619928,
-      "loss": 0.7039,
-      "step": 1420
-    },
-    {
-      "epoch": 3.5098522167487687,
-      "grad_norm": 1.140625,
-      "learning_rate": 0.00018926013239294216,
-      "loss": 0.7166,
-      "step": 1425
-    },
-    {
-      "epoch": 3.522167487684729,
-      "grad_norm": 0.79296875,
-      "learning_rate": 0.00018913055809578253,
-      "loss": 0.7053,
-      "step": 1430
-    },
-    {
-      "epoch": 3.5344827586206895,
-      "grad_norm": 0.8515625,
-      "learning_rate": 0.00018900025173895822,
-      "loss": 0.699,
-      "step": 1435
-    },
-    {
-      "epoch": 3.5467980295566504,
-      "grad_norm": 0.69140625,
-      "learning_rate": 0.00018886921439271984,
-      "loss": 0.6875,
-      "step": 1440
-    },
-    {
-      "epoch": 3.5591133004926108,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.0001887374471333218,
-      "loss": 0.7138,
-      "step": 1445
-    },
-    {
-      "epoch": 3.571428571428571,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.00018860495104301345,
-      "loss": 0.7016,
-      "step": 1450
-    },
-    {
-      "epoch": 3.583743842364532,
-      "grad_norm": 0.73046875,
-      "learning_rate": 0.00018847172721003043,
-      "loss": 0.7158,
-      "step": 1455
-    },
-    {
-      "epoch": 3.596059113300493,
-      "grad_norm": 0.80078125,
-      "learning_rate": 0.00018833777672858543,
-      "loss": 0.7027,
-      "step": 1460
-    },
-    {
-      "epoch": 3.6083743842364533,
-      "grad_norm": 0.71484375,
-      "learning_rate": 0.0001882031006988595,
-      "loss": 0.7279,
-      "step": 1465
-    },
-    {
-      "epoch": 3.6206896551724137,
-      "grad_norm": 0.76171875,
-      "learning_rate": 0.00018806770022699278,
-      "loss": 0.7183,
-      "step": 1470
-    },
-    {
-      "epoch": 3.6330049261083746,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.00018793157642507552,
-      "loss": 0.7126,
-      "step": 1475
-    },
-    {
-      "epoch": 3.645320197044335,
-      "grad_norm": 1.3671875,
-      "learning_rate": 0.00018779473041113885,
-      "loss": 0.7076,
-      "step": 1480
-    },
-    {
-      "epoch": 3.6576354679802954,
-      "grad_norm": 1.546875,
-      "learning_rate": 0.0001876571633091458,
-      "loss": 0.7159,
-      "step": 1485
-    },
-    {
-      "epoch": 3.6699507389162562,
-      "grad_norm": 0.8984375,
-      "learning_rate": 0.00018751887624898195,
-      "loss": 0.7121,
-      "step": 1490
-    },
-    {
-      "epoch": 3.6822660098522166,
-      "grad_norm": 1.8828125,
-      "learning_rate": 0.00018737987036644596,
-      "loss": 0.7062,
-      "step": 1495
-    },
-    {
-      "epoch": 3.6945812807881775,
-      "grad_norm": 1.5703125,
-      "learning_rate": 0.00018724014680324057,
-      "loss": 0.706,
-      "step": 1500
-    },
-    {
-      "epoch": 3.706896551724138,
-      "grad_norm": 0.984375,
-      "learning_rate": 0.00018709970670696308,
-      "loss": 0.7015,
-      "step": 1505
-    },
-    {
-      "epoch": 3.7192118226600988,
-      "grad_norm": 1.1640625,
-      "learning_rate": 0.00018695855123109588,
-      "loss": 0.7154,
-      "step": 1510
-    },
-    {
-      "epoch": 3.731527093596059,
-      "grad_norm": 0.85546875,
-      "learning_rate": 0.00018681668153499697,
-      "loss": 0.7154,
-      "step": 1515
-    },
-    {
-      "epoch": 3.7438423645320196,
-      "grad_norm": 1.375,
-      "learning_rate": 0.00018667409878389064,
-      "loss": 0.7063,
-      "step": 1520
-    },
-    {
-      "epoch": 3.7561576354679804,
-      "grad_norm": 1.203125,
-      "learning_rate": 0.00018653080414885755,
-      "loss": 0.7167,
-      "step": 1525
-    },
-    {
-      "epoch": 3.768472906403941,
-      "grad_norm": 0.90234375,
-      "learning_rate": 0.00018638679880682543,
-      "loss": 0.7125,
-      "step": 1530
-    },
-    {
-      "epoch": 3.7807881773399012,
-      "grad_norm": 1.203125,
-      "learning_rate": 0.00018624208394055924,
-      "loss": 0.707,
-      "step": 1535
-    },
-    {
-      "epoch": 3.793103448275862,
-      "grad_norm": 1.15625,
-      "learning_rate": 0.00018609666073865158,
-      "loss": 0.7092,
-      "step": 1540
-    },
-    {
-      "epoch": 3.8054187192118225,
-      "grad_norm": 0.9765625,
-      "learning_rate": 0.00018595053039551274,
-      "loss": 0.7163,
-      "step": 1545
-    },
-    {
-      "epoch": 3.8177339901477834,
-      "grad_norm": 1.4609375,
-      "learning_rate": 0.00018580369411136104,
-      "loss": 0.7119,
-      "step": 1550
-    },
-    {
-      "epoch": 3.8300492610837438,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.00018565615309221295,
-      "loss": 0.7175,
-      "step": 1555
-    },
-    {
-      "epoch": 3.8423645320197046,
-      "grad_norm": 1.4609375,
-      "learning_rate": 0.00018550790854987323,
-      "loss": 0.7004,
-      "step": 1560
-    },
-    {
-      "epoch": 3.854679802955665,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.00018535896170192482,
-      "loss": 0.7061,
-      "step": 1565
-    },
-    {
-      "epoch": 3.8669950738916254,
-      "grad_norm": 0.8671875,
-      "learning_rate": 0.00018520931377171893,
-      "loss": 0.7183,
-      "step": 1570
-    },
-    {
-      "epoch": 3.8793103448275863,
-      "grad_norm": 1.2734375,
-      "learning_rate": 0.00018505896598836508,
-      "loss": 0.7094,
-      "step": 1575
-    },
-    {
-      "epoch": 3.8916256157635467,
-      "grad_norm": 2.015625,
-      "learning_rate": 0.00018490791958672084,
-      "loss": 0.7055,
-      "step": 1580
-    },
-    {
-      "epoch": 3.903940886699507,
-      "grad_norm": 0.91796875,
-      "learning_rate": 0.00018475617580738187,
-      "loss": 0.7137,
-      "step": 1585
-    },
-    {
-      "epoch": 3.916256157635468,
-      "grad_norm": 0.83984375,
-      "learning_rate": 0.00018460373589667154,
-      "loss": 0.6981,
-      "step": 1590
-    },
-    {
-      "epoch": 3.928571428571429,
-      "grad_norm": 0.74609375,
-      "learning_rate": 0.0001844506011066308,
-      "loss": 0.7196,
-      "step": 1595
-    },
-    {
-      "epoch": 3.9408866995073892,
-      "grad_norm": 0.84765625,
-      "learning_rate": 0.0001842967726950079,
-      "loss": 0.7185,
-      "step": 1600
-    },
-    {
-      "epoch": 3.9532019704433496,
-      "grad_norm": 0.73828125,
-      "learning_rate": 0.00018414225192524806,
-      "loss": 0.713,
-      "step": 1605
-    },
-    {
-      "epoch": 3.9655172413793105,
-      "grad_norm": 1.296875,
-      "learning_rate": 0.00018398704006648302,
-      "loss": 0.7047,
-      "step": 1610
-    },
-    {
-      "epoch": 3.977832512315271,
-      "grad_norm": 0.83984375,
-      "learning_rate": 0.00018383113839352068,
-      "loss": 0.7008,
-      "step": 1615
-    },
-    {
-      "epoch": 3.9901477832512313,
-      "grad_norm": 0.87109375,
-      "learning_rate": 0.00018367454818683473,
-      "loss": 0.7295,
-      "step": 1620
-    },
-    {
-      "epoch": 4.0,
-      "eval_loss": 2.596930742263794,
-      "eval_runtime": 2.044,
-      "eval_samples_per_second": 4.892,
-      "eval_steps_per_second": 0.978,
-      "step": 1624
-    },
-    {
-      "epoch": 4.002463054187192,
-      "grad_norm": 0.640625,
-      "learning_rate": 0.0001835172707325538,
-      "loss": 0.6924,
-      "step": 1625
-    },
-    {
-      "epoch": 4.014778325123153,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.00018335930732245136,
-      "loss": 0.6128,
-      "step": 1630
-    },
-    {
-      "epoch": 4.027093596059113,
-      "grad_norm": 0.76953125,
-      "learning_rate": 0.00018320065925393468,
-      "loss": 0.6085,
-      "step": 1635
-    },
-    {
-      "epoch": 4.039408866995074,
-      "grad_norm": 0.8203125,
-      "learning_rate": 0.00018304132783003452,
-      "loss": 0.612,
-      "step": 1640
-    },
-    {
-      "epoch": 4.051724137931035,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00018288131435939412,
-      "loss": 0.6157,
-      "step": 1645
-    },
-    {
-      "epoch": 4.064039408866995,
-      "grad_norm": 0.87109375,
-      "learning_rate": 0.00018272062015625872,
-      "loss": 0.6039,
-      "step": 1650
-    },
-    {
-      "epoch": 4.0763546798029555,
-      "grad_norm": 1.3203125,
-      "learning_rate": 0.00018255924654046458,
-      "loss": 0.6103,
-      "step": 1655
-    },
-    {
-      "epoch": 4.088669950738916,
-      "grad_norm": 0.98046875,
-      "learning_rate": 0.00018239719483742822,
-      "loss": 0.6045,
-      "step": 1660
-    },
-    {
-      "epoch": 4.100985221674877,
-      "grad_norm": 0.99609375,
-      "learning_rate": 0.0001822344663781356,
-      "loss": 0.6083,
-      "step": 1665
-    },
-    {
-      "epoch": 4.113300492610837,
-      "grad_norm": 1.4140625,
-      "learning_rate": 0.00018207106249913094,
-      "loss": 0.6176,
-      "step": 1670
-    },
-    {
-      "epoch": 4.125615763546798,
-      "grad_norm": 0.86328125,
-      "learning_rate": 0.00018190698454250605,
-      "loss": 0.622,
-      "step": 1675
-    },
-    {
-      "epoch": 4.137931034482759,
-      "grad_norm": 1.25,
-      "learning_rate": 0.00018174223385588917,
-      "loss": 0.6116,
-      "step": 1680
-    },
-    {
-      "epoch": 4.150246305418719,
-      "grad_norm": 1.15625,
-      "learning_rate": 0.00018157681179243386,
-      "loss": 0.6116,
-      "step": 1685
-    },
-    {
-      "epoch": 4.16256157635468,
-      "grad_norm": 1.1796875,
-      "learning_rate": 0.00018141071971080792,
-      "loss": 0.6245,
-      "step": 1690
-    },
-    {
-      "epoch": 4.174876847290641,
-      "grad_norm": 1.1015625,
-      "learning_rate": 0.00018124395897518224,
-      "loss": 0.6307,
-      "step": 1695
-    },
-    {
-      "epoch": 4.187192118226601,
-      "grad_norm": 1.078125,
-      "learning_rate": 0.00018107653095521958,
-      "loss": 0.6107,
-      "step": 1700
-    },
-    {
-      "epoch": 4.199507389162561,
-      "grad_norm": 0.78125,
-      "learning_rate": 0.00018090843702606337,
-      "loss": 0.6168,
-      "step": 1705
-    },
-    {
-      "epoch": 4.211822660098522,
-      "grad_norm": 1.2265625,
-      "learning_rate": 0.0001807396785683264,
-      "loss": 0.6196,
-      "step": 1710
-    },
-    {
-      "epoch": 4.224137931034483,
-      "grad_norm": 0.77734375,
-      "learning_rate": 0.0001805702569680794,
-      "loss": 0.6114,
-      "step": 1715
-    },
-    {
-      "epoch": 4.236453201970443,
-      "grad_norm": 0.9296875,
-      "learning_rate": 0.00018040017361683976,
-      "loss": 0.6232,
-      "step": 1720
-    },
-    {
-      "epoch": 4.248768472906404,
-      "grad_norm": 0.86328125,
-      "learning_rate": 0.00018022942991156,
-      "loss": 0.6153,
-      "step": 1725
-    },
-    {
-      "epoch": 4.261083743842365,
-      "grad_norm": 0.80078125,
-      "learning_rate": 0.00018005802725461643,
-      "loss": 0.6203,
-      "step": 1730
-    },
-    {
-      "epoch": 4.273399014778325,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.0001798859670537975,
-      "loss": 0.6221,
-      "step": 1735
-    },
-    {
-      "epoch": 4.285714285714286,
-      "grad_norm": 1.2890625,
-      "learning_rate": 0.00017971325072229226,
-      "loss": 0.6332,
-      "step": 1740
-    },
-    {
-      "epoch": 4.298029556650246,
-      "grad_norm": 0.875,
-      "learning_rate": 0.0001795398796786789,
-      "loss": 0.6262,
-      "step": 1745
-    },
-    {
-      "epoch": 4.310344827586207,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.00017936585534691291,
-      "loss": 0.6301,
-      "step": 1750
-    },
-    {
-      "epoch": 4.322660098522167,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.0001791911791563154,
-      "loss": 0.6247,
-      "step": 1755
-    },
-    {
-      "epoch": 4.334975369458128,
-      "grad_norm": 0.953125,
-      "learning_rate": 0.00017901585254156155,
-      "loss": 0.6376,
-      "step": 1760
-    },
-    {
-      "epoch": 4.347290640394089,
-      "grad_norm": 1.203125,
-      "learning_rate": 0.00017883987694266863,
-      "loss": 0.6311,
-      "step": 1765
-    },
-    {
-      "epoch": 4.359605911330049,
-      "grad_norm": 1.328125,
-      "learning_rate": 0.00017866325380498416,
-      "loss": 0.6346,
-      "step": 1770
-    },
-    {
-      "epoch": 4.37192118226601,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.0001784859845791743,
-      "loss": 0.6289,
-      "step": 1775
-    },
-    {
-      "epoch": 4.384236453201971,
-      "grad_norm": 1.234375,
-      "learning_rate": 0.00017830807072121156,
-      "loss": 0.625,
-      "step": 1780
-    },
-    {
-      "epoch": 4.396551724137931,
-      "grad_norm": 0.8984375,
-      "learning_rate": 0.00017812951369236316,
-      "loss": 0.6233,
-      "step": 1785
-    },
-    {
-      "epoch": 4.4088669950738915,
-      "grad_norm": 0.83984375,
-      "learning_rate": 0.00017795031495917884,
-      "loss": 0.642,
-      "step": 1790
-    },
-    {
-      "epoch": 4.421182266009852,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00017777047599347893,
-      "loss": 0.6285,
-      "step": 1795
-    },
-    {
-      "epoch": 4.433497536945813,
-      "grad_norm": 0.90234375,
-      "learning_rate": 0.00017758999827234212,
-      "loss": 0.6342,
-      "step": 1800
-    },
-    {
-      "epoch": 4.445812807881773,
-      "grad_norm": 2.015625,
-      "learning_rate": 0.00017740888327809354,
-      "loss": 0.6302,
-      "step": 1805
-    },
-    {
-      "epoch": 4.458128078817734,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.00017722713249829236,
-      "loss": 0.6198,
-      "step": 1810
-    },
-    {
-      "epoch": 4.470443349753695,
-      "grad_norm": 3.390625,
-      "learning_rate": 0.00017704474742571969,
-      "loss": 0.6205,
-      "step": 1815
-    },
-    {
-      "epoch": 4.482758620689655,
-      "grad_norm": 1.4296875,
-      "learning_rate": 0.00017686172955836633,
-      "loss": 0.6286,
-      "step": 1820
-    },
-    {
-      "epoch": 4.495073891625616,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00017667808039942043,
-      "loss": 0.635,
-      "step": 1825
-    },
-    {
-      "epoch": 4.5073891625615765,
-      "grad_norm": 0.81640625,
-      "learning_rate": 0.00017649380145725517,
-      "loss": 0.6237,
-      "step": 1830
-    },
-    {
-      "epoch": 4.519704433497537,
-      "grad_norm": 0.9765625,
-      "learning_rate": 0.0001763088942454163,
-      "loss": 0.6266,
-      "step": 1835
-    },
-    {
-      "epoch": 4.532019704433497,
-      "grad_norm": 1.6015625,
-      "learning_rate": 0.00017612336028260982,
-      "loss": 0.6381,
-      "step": 1840
-    },
-    {
-      "epoch": 4.544334975369458,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00017593720109268944,
-      "loss": 0.6474,
-      "step": 1845
-    },
-    {
-      "epoch": 4.556650246305419,
-      "grad_norm": 0.82421875,
-      "learning_rate": 0.00017575041820464405,
-      "loss": 0.6381,
-      "step": 1850
-    },
-    {
-      "epoch": 4.568965517241379,
-      "grad_norm": 0.7734375,
-      "learning_rate": 0.00017556301315258517,
-      "loss": 0.6341,
-      "step": 1855
-    },
-    {
-      "epoch": 4.58128078817734,
-      "grad_norm": 0.9921875,
-      "learning_rate": 0.00017537498747573443,
-      "loss": 0.6398,
-      "step": 1860
-    },
-    {
-      "epoch": 4.593596059113301,
-      "grad_norm": 1.4453125,
-      "learning_rate": 0.00017518634271841083,
-      "loss": 0.6445,
-      "step": 1865
-    },
-    {
-      "epoch": 4.605911330049262,
-      "grad_norm": 0.85546875,
-      "learning_rate": 0.00017499708043001807,
-      "loss": 0.6446,
-      "step": 1870
-    },
-    {
-      "epoch": 4.6182266009852215,
-      "grad_norm": 0.8359375,
-      "learning_rate": 0.00017480720216503183,
-      "loss": 0.642,
-      "step": 1875
-    },
-    {
-      "epoch": 4.630541871921182,
-      "grad_norm": 0.8984375,
-      "learning_rate": 0.0001746167094829871,
-      "loss": 0.6425,
-      "step": 1880
-    },
-    {
-      "epoch": 4.642857142857143,
-      "grad_norm": 1.4765625,
-      "learning_rate": 0.00017442560394846516,
-      "loss": 0.6315,
-      "step": 1885
-    },
-    {
-      "epoch": 4.655172413793103,
-      "grad_norm": 0.8828125,
-      "learning_rate": 0.000174233887131081,
-      "loss": 0.6423,
-      "step": 1890
-    },
-    {
-      "epoch": 4.667487684729064,
-      "grad_norm": 0.734375,
-      "learning_rate": 0.00017404156060547016,
-      "loss": 0.6354,
-      "step": 1895
-    },
-    {
-      "epoch": 4.679802955665025,
-      "grad_norm": 0.76953125,
-      "learning_rate": 0.000173848625951276,
-      "loss": 0.6404,
-      "step": 1900
-    },
-    {
-      "epoch": 4.692118226600985,
-      "grad_norm": 0.72265625,
-      "learning_rate": 0.0001736550847531366,
-      "loss": 0.6432,
-      "step": 1905
-    },
-    {
-      "epoch": 4.704433497536946,
-      "grad_norm": 0.85546875,
-      "learning_rate": 0.00017346093860067186,
-      "loss": 0.6478,
-      "step": 1910
-    },
-    {
-      "epoch": 4.716748768472907,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.00017326618908847024,
-      "loss": 0.6551,
-      "step": 1915
-    },
-    {
-      "epoch": 4.7290640394088665,
-      "grad_norm": 0.83203125,
-      "learning_rate": 0.00017307083781607595,
-      "loss": 0.6346,
-      "step": 1920
-    },
-    {
-      "epoch": 4.741379310344827,
-      "grad_norm": 0.85546875,
-      "learning_rate": 0.00017287488638797563,
-      "loss": 0.6477,
-      "step": 1925
-    },
-    {
-      "epoch": 4.753694581280788,
-      "grad_norm": 0.7734375,
-      "learning_rate": 0.0001726783364135851,
-      "loss": 0.6516,
-      "step": 1930
-    },
-    {
-      "epoch": 4.766009852216749,
-      "grad_norm": 0.796875,
-      "learning_rate": 0.00017248118950723634,
-      "loss": 0.6469,
-      "step": 1935
-    },
-    {
-      "epoch": 4.778325123152709,
-      "grad_norm": 0.80078125,
-      "learning_rate": 0.00017228344728816413,
-      "loss": 0.6438,
-      "step": 1940
-    },
-    {
-      "epoch": 4.79064039408867,
-      "grad_norm": 0.94140625,
-      "learning_rate": 0.00017208511138049274,
-      "loss": 0.6484,
-      "step": 1945
-    },
-    {
-      "epoch": 4.802955665024631,
-      "grad_norm": 0.76953125,
-      "learning_rate": 0.00017188618341322254,
-      "loss": 0.6488,
-      "step": 1950
-    },
-    {
-      "epoch": 4.815270935960591,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.0001716866650202169,
-      "loss": 0.6311,
-      "step": 1955
-    },
-    {
-      "epoch": 4.827586206896552,
-      "grad_norm": 0.83984375,
-      "learning_rate": 0.00017148655784018829,
-      "loss": 0.6517,
-      "step": 1960
-    },
-    {
-      "epoch": 4.8399014778325125,
-      "grad_norm": 0.8359375,
-      "learning_rate": 0.00017128586351668524,
-      "loss": 0.6483,
-      "step": 1965
-    },
-    {
-      "epoch": 4.852216748768473,
-      "grad_norm": 0.91015625,
-      "learning_rate": 0.00017108458369807864,
-      "loss": 0.6471,
-      "step": 1970
-    },
-    {
-      "epoch": 4.864532019704433,
-      "grad_norm": 0.81640625,
-      "learning_rate": 0.00017088272003754832,
-      "loss": 0.641,
-      "step": 1975
-    },
-    {
-      "epoch": 4.876847290640394,
-      "grad_norm": 0.89453125,
-      "learning_rate": 0.00017068027419306936,
-      "loss": 0.6464,
-      "step": 1980
-    },
-    {
-      "epoch": 4.889162561576355,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00017047724782739846,
-      "loss": 0.6405,
-      "step": 1985
-    },
-    {
-      "epoch": 4.901477832512315,
-      "grad_norm": 0.78125,
-      "learning_rate": 0.0001702736426080604,
-      "loss": 0.6536,
-      "step": 1990
-    },
-    {
-      "epoch": 4.913793103448276,
-      "grad_norm": 1.2890625,
-      "learning_rate": 0.00017006946020733425,
-      "loss": 0.6359,
-      "step": 1995
-    },
-    {
-      "epoch": 4.926108374384237,
-      "grad_norm": 0.9921875,
-      "learning_rate": 0.00016986470230223973,
-      "loss": 0.6482,
-      "step": 2000
-    },
-    {
-      "epoch": 4.9384236453201975,
-      "grad_norm": 0.9453125,
-      "learning_rate": 0.00016965937057452333,
-      "loss": 0.6431,
-      "step": 2005
-    },
-    {
-      "epoch": 4.9507389162561575,
-      "grad_norm": 0.7421875,
-      "learning_rate": 0.00016945346671064452,
-      "loss": 0.6546,
-      "step": 2010
-    },
-    {
-      "epoch": 4.963054187192118,
-      "grad_norm": 1.078125,
-      "learning_rate": 0.00016924699240176194,
-      "loss": 0.6462,
-      "step": 2015
-    },
-    {
-      "epoch": 4.975369458128079,
-      "grad_norm": 1.1171875,
-      "learning_rate": 0.00016903994934371953,
-      "loss": 0.6374,
-      "step": 2020
-    },
-    {
-      "epoch": 4.987684729064039,
-      "grad_norm": 1.75,
-      "learning_rate": 0.00016883233923703248,
-      "loss": 0.6569,
-      "step": 2025
-    },
-    {
-      "epoch": 5.0,
-      "grad_norm": 0.73828125,
-      "learning_rate": 0.0001686241637868734,
-      "loss": 0.6377,
-      "step": 2030
-    },
-    {
-      "epoch": 5.0,
-      "eval_loss": 2.8600502014160156,
-      "eval_runtime": 2.042,
-      "eval_samples_per_second": 4.897,
-      "eval_steps_per_second": 0.979,
-      "step": 2030
-    },
-    {
-      "epoch": 5.012315270935961,
-      "grad_norm": 0.9296875,
-      "learning_rate": 0.00016841542470305817,
-      "loss": 0.5387,
-      "step": 2035
-    },
-    {
-      "epoch": 5.024630541871921,
-      "grad_norm": 1.046875,
-      "learning_rate": 0.00016820612370003221,
-      "loss": 0.5418,
-      "step": 2040
-    },
-    {
-      "epoch": 5.036945812807882,
-      "grad_norm": 0.80078125,
-      "learning_rate": 0.0001679962624968559,
-      "loss": 0.5381,
-      "step": 2045
-    },
-    {
-      "epoch": 5.0492610837438425,
-      "grad_norm": 0.87109375,
-      "learning_rate": 0.00016778584281719095,
-      "loss": 0.5311,
-      "step": 2050
-    },
-    {
-      "epoch": 5.061576354679803,
-      "grad_norm": 0.84765625,
-      "learning_rate": 0.00016757486638928587,
-      "loss": 0.5428,
-      "step": 2055
-    },
-    {
-      "epoch": 5.073891625615763,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00016736333494596196,
-      "loss": 0.5433,
-      "step": 2060
-    },
-    {
-      "epoch": 5.086206896551724,
-      "grad_norm": 0.77734375,
-      "learning_rate": 0.00016715125022459922,
-      "loss": 0.5428,
-      "step": 2065
-    },
-    {
-      "epoch": 5.098522167487685,
-      "grad_norm": 0.828125,
-      "learning_rate": 0.00016693861396712168,
-      "loss": 0.5395,
-      "step": 2070
-    },
-    {
-      "epoch": 5.110837438423645,
-      "grad_norm": 0.86328125,
-      "learning_rate": 0.00016672542791998344,
-      "loss": 0.5527,
-      "step": 2075
-    },
-    {
-      "epoch": 5.123152709359606,
-      "grad_norm": 0.984375,
-      "learning_rate": 0.0001665116938341542,
-      "loss": 0.5529,
-      "step": 2080
-    },
-    {
-      "epoch": 5.135467980295567,
-      "grad_norm": 1.078125,
-      "learning_rate": 0.00016629741346510496,
-      "loss": 0.5512,
-      "step": 2085
-    },
-    {
-      "epoch": 5.147783251231527,
-      "grad_norm": 0.953125,
-      "learning_rate": 0.00016608258857279333,
-      "loss": 0.5472,
-      "step": 2090
-    },
-    {
-      "epoch": 5.1600985221674875,
-      "grad_norm": 0.93359375,
-      "learning_rate": 0.0001658672209216495,
-      "loss": 0.55,
-      "step": 2095
-    },
-    {
-      "epoch": 5.172413793103448,
-      "grad_norm": 0.875,
-      "learning_rate": 0.00016565131228056133,
-      "loss": 0.5459,
-      "step": 2100
-    },
-    {
-      "epoch": 5.184729064039409,
-      "grad_norm": 0.84765625,
-      "learning_rate": 0.0001654348644228602,
-      "loss": 0.5552,
-      "step": 2105
-    },
-    {
-      "epoch": 5.197044334975369,
-      "grad_norm": 0.87109375,
-      "learning_rate": 0.00016521787912630612,
-      "loss": 0.5465,
-      "step": 2110
-    },
-    {
-      "epoch": 5.20935960591133,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00016500035817307334,
-      "loss": 0.5581,
-      "step": 2115
-    },
-    {
-      "epoch": 5.221674876847291,
-      "grad_norm": 0.78125,
-      "learning_rate": 0.00016478230334973556,
-      "loss": 0.5535,
-      "step": 2120
-    },
-    {
-      "epoch": 5.233990147783251,
-      "grad_norm": 0.8671875,
-      "learning_rate": 0.00016456371644725146,
-      "loss": 0.5604,
-      "step": 2125
-    },
-    {
-      "epoch": 5.246305418719212,
-      "grad_norm": 0.90234375,
-      "learning_rate": 0.0001643445992609498,
-      "loss": 0.5551,
-      "step": 2130
-    },
-    {
-      "epoch": 5.258620689655173,
-      "grad_norm": 0.94921875,
-      "learning_rate": 0.0001641249535905147,
-      "loss": 0.5443,
-      "step": 2135
-    },
-    {
-      "epoch": 5.2709359605911335,
-      "grad_norm": 0.87109375,
-      "learning_rate": 0.00016390478123997094,
-      "loss": 0.5544,
-      "step": 2140
-    },
-    {
-      "epoch": 5.283251231527093,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.00016368408401766916,
-      "loss": 0.571,
-      "step": 2145
-    },
-    {
-      "epoch": 5.295566502463054,
-      "grad_norm": 0.81640625,
-      "learning_rate": 0.0001634628637362709,
-      "loss": 0.5501,
-      "step": 2150
-    },
-    {
-      "epoch": 5.307881773399015,
-      "grad_norm": 0.89453125,
-      "learning_rate": 0.0001632411222127337,
-      "loss": 0.5537,
-      "step": 2155
-    },
-    {
-      "epoch": 5.320197044334975,
-      "grad_norm": 0.90234375,
-      "learning_rate": 0.0001630188612682963,
-      "loss": 0.563,
-      "step": 2160
-    },
-    {
-      "epoch": 5.332512315270936,
-      "grad_norm": 0.859375,
-      "learning_rate": 0.00016279608272846372,
-      "loss": 0.5663,
-      "step": 2165
-    },
-    {
-      "epoch": 5.344827586206897,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00016257278842299197,
-      "loss": 0.5587,
-      "step": 2170
-    },
-    {
-      "epoch": 5.357142857142857,
-      "grad_norm": 0.94140625,
-      "learning_rate": 0.00016234898018587337,
-      "loss": 0.5627,
-      "step": 2175
-    },
-    {
-      "epoch": 5.369458128078818,
-      "grad_norm": 0.89453125,
-      "learning_rate": 0.00016212465985532124,
-      "loss": 0.5699,
-      "step": 2180
-    },
-    {
-      "epoch": 5.3817733990147785,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.000161899829273755,
-      "loss": 0.5686,
-      "step": 2185
-    },
-    {
-      "epoch": 5.394088669950739,
-      "grad_norm": 1.4296875,
-      "learning_rate": 0.00016167449028778484,
-      "loss": 0.5679,
-      "step": 2190
-    },
-    {
-      "epoch": 5.406403940886699,
-      "grad_norm": 0.7421875,
-      "learning_rate": 0.00016144864474819666,
-      "loss": 0.5715,
-      "step": 2195
-    },
-    {
-      "epoch": 5.41871921182266,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.0001612222945099369,
-      "loss": 0.5646,
-      "step": 2200
-    },
-    {
-      "epoch": 5.431034482758621,
-      "grad_norm": 0.89453125,
-      "learning_rate": 0.0001609954414320973,
-      "loss": 0.5586,
-      "step": 2205
-    },
-    {
-      "epoch": 5.443349753694581,
-      "grad_norm": 1.078125,
-      "learning_rate": 0.00016076808737789947,
-      "loss": 0.5675,
-      "step": 2210
-    },
-    {
-      "epoch": 5.455665024630542,
-      "grad_norm": 0.80078125,
-      "learning_rate": 0.00016054023421467983,
-      "loss": 0.5632,
-      "step": 2215
-    },
-    {
-      "epoch": 5.467980295566503,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.0001603118838138741,
-      "loss": 0.564,
-      "step": 2220
-    },
-    {
-      "epoch": 5.480295566502463,
-      "grad_norm": 0.875,
-      "learning_rate": 0.00016008303805100193,
-      "loss": 0.5735,
-      "step": 2225
-    },
-    {
-      "epoch": 5.4926108374384235,
-      "grad_norm": 0.90234375,
-      "learning_rate": 0.00015985369880565164,
-      "loss": 0.5583,
-      "step": 2230
-    },
-    {
-      "epoch": 5.504926108374384,
-      "grad_norm": 1.1015625,
-      "learning_rate": 0.00015962386796146462,
-      "loss": 0.5686,
-      "step": 2235
-    },
-    {
-      "epoch": 5.517241379310345,
-      "grad_norm": 0.96875,
-      "learning_rate": 0.00015939354740612,
-      "loss": 0.571,
-      "step": 2240
-    },
-    {
-      "epoch": 5.529556650246305,
-      "grad_norm": 1.375,
-      "learning_rate": 0.000159162739031319,
-      "loss": 0.5735,
-      "step": 2245
-    },
-    {
-      "epoch": 5.541871921182266,
-      "grad_norm": 1.3125,
-      "learning_rate": 0.00015893144473276953,
-      "loss": 0.5691,
-      "step": 2250
-    },
-    {
-      "epoch": 5.554187192118227,
-      "grad_norm": 0.87890625,
-      "learning_rate": 0.0001586996664101705,
-      "loss": 0.5663,
-      "step": 2255
-    },
-    {
-      "epoch": 5.566502463054187,
-      "grad_norm": 1.3671875,
-      "learning_rate": 0.00015846740596719636,
-      "loss": 0.5684,
-      "step": 2260
-    },
-    {
-      "epoch": 5.578817733990148,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.00015823466531148124,
-      "loss": 0.575,
-      "step": 2265
-    },
-    {
-      "epoch": 5.5911330049261085,
-      "grad_norm": 0.94921875,
-      "learning_rate": 0.00015800144635460354,
-      "loss": 0.5814,
-      "step": 2270
-    },
-    {
-      "epoch": 5.603448275862069,
-      "grad_norm": 0.8125,
-      "learning_rate": 0.0001577677510120701,
-      "loss": 0.5778,
-      "step": 2275
-    },
-    {
-      "epoch": 5.615763546798029,
-      "grad_norm": 1.109375,
-      "learning_rate": 0.00015753358120330042,
-      "loss": 0.5688,
-      "step": 2280
-    },
-    {
-      "epoch": 5.62807881773399,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00015729893885161098,
-      "loss": 0.5771,
-      "step": 2285
-    },
-    {
-      "epoch": 5.640394088669951,
-      "grad_norm": 0.87890625,
-      "learning_rate": 0.00015706382588419945,
-      "loss": 0.5723,
-      "step": 2290
-    },
-    {
-      "epoch": 5.652709359605911,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.00015682824423212877,
-      "loss": 0.5782,
-      "step": 2295
-    },
-    {
-      "epoch": 5.665024630541872,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.0001565921958303114,
-      "loss": 0.5776,
-      "step": 2300
-    },
-    {
-      "epoch": 5.677339901477833,
-      "grad_norm": 0.85546875,
-      "learning_rate": 0.00015635568261749332,
-      "loss": 0.5721,
-      "step": 2305
-    },
-    {
-      "epoch": 5.689655172413794,
-      "grad_norm": 0.8515625,
-      "learning_rate": 0.00015611870653623825,
-      "loss": 0.5643,
-      "step": 2310
-    },
-    {
-      "epoch": 5.701970443349754,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.0001558812695329115,
-      "loss": 0.574,
-      "step": 2315
-    },
-    {
-      "epoch": 5.714285714285714,
-      "grad_norm": 0.88671875,
-      "learning_rate": 0.00015564337355766412,
-      "loss": 0.5795,
-      "step": 2320
-    },
-    {
-      "epoch": 5.726600985221674,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00015540502056441688,
-      "loss": 0.5779,
-      "step": 2325
-    },
-    {
-      "epoch": 5.738916256157635,
-      "grad_norm": 0.953125,
-      "learning_rate": 0.00015516621251084422,
-      "loss": 0.5754,
-      "step": 2330
-    },
-    {
-      "epoch": 5.751231527093596,
-      "grad_norm": 1.4921875,
-      "learning_rate": 0.00015492695135835811,
-      "loss": 0.5769,
-      "step": 2335
-    },
-    {
-      "epoch": 5.763546798029557,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00015468723907209193,
-      "loss": 0.5842,
-      "step": 2340
-    },
-    {
-      "epoch": 5.775862068965517,
-      "grad_norm": 0.96875,
-      "learning_rate": 0.00015444707762088443,
-      "loss": 0.5806,
-      "step": 2345
-    },
-    {
-      "epoch": 5.788177339901478,
-      "grad_norm": 0.81640625,
-      "learning_rate": 0.0001542064689772636,
-      "loss": 0.5817,
-      "step": 2350
-    },
-    {
-      "epoch": 5.800492610837439,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.00015396541511743012,
-      "loss": 0.5767,
-      "step": 2355
-    },
-    {
-      "epoch": 5.812807881773399,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.00015372391802124163,
-      "loss": 0.5726,
-      "step": 2360
-    },
-    {
-      "epoch": 5.825123152709359,
-      "grad_norm": 0.83203125,
-      "learning_rate": 0.00015348197967219606,
-      "loss": 0.5731,
-      "step": 2365
-    },
-    {
-      "epoch": 5.83743842364532,
-      "grad_norm": 0.8515625,
-      "learning_rate": 0.00015323960205741561,
-      "loss": 0.5873,
-      "step": 2370
-    },
-    {
-      "epoch": 5.849753694581281,
-      "grad_norm": 1.0,
-      "learning_rate": 0.00015299678716763028,
-      "loss": 0.5801,
-      "step": 2375
-    },
-    {
-      "epoch": 5.862068965517241,
-      "grad_norm": 2.0625,
-      "learning_rate": 0.00015275353699716155,
-      "loss": 0.5826,
-      "step": 2380
-    },
-    {
-      "epoch": 5.874384236453202,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00015250985354390596,
-      "loss": 0.5893,
-      "step": 2385
-    },
-    {
-      "epoch": 5.886699507389163,
-      "grad_norm": 0.859375,
-      "learning_rate": 0.00015226573880931888,
-      "loss": 0.5772,
-      "step": 2390
-    },
-    {
-      "epoch": 5.899014778325123,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.0001520211947983978,
-      "loss": 0.5838,
-      "step": 2395
-    },
-    {
-      "epoch": 5.911330049261084,
-      "grad_norm": 0.7890625,
-      "learning_rate": 0.0001517762235196661,
-      "loss": 0.5719,
-      "step": 2400
-    },
-    {
-      "epoch": 5.9236453201970445,
-      "grad_norm": 0.8125,
-      "learning_rate": 0.0001515308269851564,
-      "loss": 0.5741,
-      "step": 2405
-    },
-    {
-      "epoch": 5.935960591133005,
-      "grad_norm": 0.84375,
-      "learning_rate": 0.0001512850072103941,
-      "loss": 0.5831,
-      "step": 2410
-    },
-    {
-      "epoch": 5.948275862068965,
-      "grad_norm": 0.88671875,
-      "learning_rate": 0.00015103876621438086,
-      "loss": 0.5768,
-      "step": 2415
-    },
-    {
-      "epoch": 5.960591133004926,
-      "grad_norm": 0.8359375,
-      "learning_rate": 0.00015079210601957793,
-      "loss": 0.5875,
-      "step": 2420
-    },
-    {
-      "epoch": 5.972906403940887,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00015054502865188957,
-      "loss": 0.5797,
-      "step": 2425
-    },
-    {
-      "epoch": 5.985221674876847,
-      "grad_norm": 1.203125,
-      "learning_rate": 0.00015029753614064645,
-      "loss": 0.5835,
-      "step": 2430
-    },
-    {
-      "epoch": 5.997536945812808,
-      "grad_norm": 1.296875,
-      "learning_rate": 0.000150049630518589,
-      "loss": 0.5891,
-      "step": 2435
-    },
-    {
-      "epoch": 6.0,
-      "eval_loss": 3.1032605171203613,
-      "eval_runtime": 2.0442,
-      "eval_samples_per_second": 4.892,
-      "eval_steps_per_second": 0.978,
-      "step": 2436
-    },
-    {
-      "epoch": 6.009852216748769,
-      "grad_norm": 0.8515625,
-      "learning_rate": 0.0001498013138218506,
-      "loss": 0.4998,
-      "step": 2440
-    },
-    {
-      "epoch": 6.022167487684729,
-      "grad_norm": 0.88671875,
-      "learning_rate": 0.00014955258808994096,
-      "loss": 0.4797,
-      "step": 2445
-    },
-    {
-      "epoch": 6.0344827586206895,
-      "grad_norm": 0.890625,
-      "learning_rate": 0.00014930345536572924,
-      "loss": 0.4731,
-      "step": 2450
-    },
-    {
-      "epoch": 6.04679802955665,
-      "grad_norm": 1.3671875,
-      "learning_rate": 0.00014905391769542758,
-      "loss": 0.4802,
-      "step": 2455
-    },
-    {
-      "epoch": 6.059113300492611,
-      "grad_norm": 1.2890625,
-      "learning_rate": 0.00014880397712857386,
-      "loss": 0.4708,
-      "step": 2460
-    },
-    {
-      "epoch": 6.071428571428571,
-      "grad_norm": 0.91015625,
-      "learning_rate": 0.00014855363571801523,
-      "loss": 0.4703,
-      "step": 2465
-    },
-    {
-      "epoch": 6.083743842364532,
-      "grad_norm": 0.859375,
-      "learning_rate": 0.000148302895519891,
-      "loss": 0.4726,
-      "step": 2470
-    },
-    {
-      "epoch": 6.096059113300493,
-      "grad_norm": 0.83203125,
-      "learning_rate": 0.00014805175859361594,
-      "loss": 0.4733,
-      "step": 2475
-    },
-    {
-      "epoch": 6.108374384236453,
-      "grad_norm": 0.8828125,
-      "learning_rate": 0.0001478002270018633,
-      "loss": 0.4742,
-      "step": 2480
-    },
-    {
-      "epoch": 6.120689655172414,
-      "grad_norm": 0.96875,
-      "learning_rate": 0.00014754830281054777,
-      "loss": 0.4884,
-      "step": 2485
-    },
-    {
-      "epoch": 6.1330049261083746,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.00014729598808880861,
-      "loss": 0.4851,
-      "step": 2490
-    },
-    {
-      "epoch": 6.1453201970443345,
-      "grad_norm": 0.953125,
-      "learning_rate": 0.0001470432849089927,
-      "loss": 0.4755,
-      "step": 2495
-    },
-    {
-      "epoch": 6.157635467980295,
-      "grad_norm": 0.89453125,
-      "learning_rate": 0.00014679019534663738,
-      "loss": 0.4823,
-      "step": 2500
-    },
-    {
-      "epoch": 6.169950738916256,
-      "grad_norm": 1.1015625,
-      "learning_rate": 0.00014653672148045357,
-      "loss": 0.4781,
-      "step": 2505
-    },
-    {
-      "epoch": 6.182266009852217,
-      "grad_norm": 0.921875,
-      "learning_rate": 0.00014628286539230848,
-      "loss": 0.4878,
-      "step": 2510
-    },
-    {
-      "epoch": 6.194581280788177,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00014602862916720884,
-      "loss": 0.489,
-      "step": 2515
-    },
-    {
-      "epoch": 6.206896551724138,
-      "grad_norm": 1.421875,
-      "learning_rate": 0.00014577401489328335,
-      "loss": 0.4913,
-      "step": 2520
-    },
-    {
-      "epoch": 6.219211822660099,
-      "grad_norm": 1.046875,
-      "learning_rate": 0.00014551902466176592,
-      "loss": 0.4934,
-      "step": 2525
-    },
-    {
-      "epoch": 6.231527093596059,
-      "grad_norm": 1.2109375,
-      "learning_rate": 0.00014526366056697825,
-      "loss": 0.4908,
-      "step": 2530
-    },
-    {
-      "epoch": 6.24384236453202,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.0001450079247063127,
-      "loss": 0.4894,
-      "step": 2535
-    },
-    {
-      "epoch": 6.25615763546798,
-      "grad_norm": 0.91015625,
-      "learning_rate": 0.0001447518191802151,
-      "loss": 0.4942,
-      "step": 2540
-    },
-    {
-      "epoch": 6.268472906403941,
-      "grad_norm": 1.2109375,
-      "learning_rate": 0.00014449534609216748,
-      "loss": 0.4894,
-      "step": 2545
-    },
-    {
-      "epoch": 6.280788177339901,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00014423850754867075,
-      "loss": 0.4937,
-      "step": 2550
-    },
-    {
-      "epoch": 6.293103448275862,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.00014398130565922742,
-      "loss": 0.4905,
-      "step": 2555
-    },
-    {
-      "epoch": 6.305418719211823,
-      "grad_norm": 0.88671875,
-      "learning_rate": 0.00014372374253632437,
-      "loss": 0.4901,
-      "step": 2560
-    },
-    {
-      "epoch": 6.317733990147783,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.0001434658202954153,
-      "loss": 0.4944,
-      "step": 2565
-    },
-    {
-      "epoch": 6.330049261083744,
-      "grad_norm": 0.91015625,
-      "learning_rate": 0.0001432075410549035,
-      "loss": 0.4931,
-      "step": 2570
-    },
-    {
-      "epoch": 6.342364532019705,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.0001429489069361245,
-      "loss": 0.5,
-      "step": 2575
-    },
-    {
-      "epoch": 6.3546798029556655,
-      "grad_norm": 0.89453125,
-      "learning_rate": 0.00014268992006332846,
-      "loss": 0.4955,
-      "step": 2580
-    },
-    {
-      "epoch": 6.3669950738916254,
-      "grad_norm": 0.83984375,
-      "learning_rate": 0.0001424305825636629,
-      "loss": 0.5065,
-      "step": 2585
-    },
-    {
-      "epoch": 6.379310344827586,
-      "grad_norm": 0.8984375,
-      "learning_rate": 0.0001421708965671551,
-      "loss": 0.4944,
-      "step": 2590
-    },
-    {
-      "epoch": 6.391625615763547,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.0001419108642066947,
-      "loss": 0.4923,
-      "step": 2595
-    },
-    {
-      "epoch": 6.403940886699507,
-      "grad_norm": 1.2265625,
-      "learning_rate": 0.00014165048761801617,
-      "loss": 0.5143,
-      "step": 2600
-    },
-    {
-      "epoch": 6.416256157635468,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.0001413897689396812,
-      "loss": 0.4983,
-      "step": 2605
-    },
-    {
-      "epoch": 6.428571428571429,
-      "grad_norm": 0.83203125,
-      "learning_rate": 0.00014112871031306119,
-      "loss": 0.4954,
-      "step": 2610
-    },
-    {
-      "epoch": 6.440886699507389,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.0001408673138823196,
-      "loss": 0.4951,
-      "step": 2615
-    },
-    {
-      "epoch": 6.45320197044335,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00014060558179439456,
-      "loss": 0.4964,
-      "step": 2620
-    },
-    {
-      "epoch": 6.4655172413793105,
-      "grad_norm": 1.0,
-      "learning_rate": 0.00014034351619898088,
-      "loss": 0.5015,
-      "step": 2625
-    },
-    {
-      "epoch": 6.477832512315271,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.00014008111924851264,
-      "loss": 0.4904,
-      "step": 2630
-    },
-    {
-      "epoch": 6.490147783251231,
-      "grad_norm": 0.9921875,
-      "learning_rate": 0.00013981839309814547,
-      "loss": 0.5047,
-      "step": 2635
-    },
-    {
-      "epoch": 6.502463054187192,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.00013955533990573886,
-      "loss": 0.4973,
-      "step": 2640
-    },
-    {
-      "epoch": 6.514778325123153,
-      "grad_norm": 0.8515625,
-      "learning_rate": 0.00013929196183183834,
-      "loss": 0.492,
-      "step": 2645
-    },
-    {
-      "epoch": 6.527093596059113,
-      "grad_norm": 1.1328125,
-      "learning_rate": 0.00013902826103965788,
-      "loss": 0.4996,
-      "step": 2650
-    },
-    {
-      "epoch": 6.539408866995074,
-      "grad_norm": 0.95703125,
-      "learning_rate": 0.00013876423969506194,
-      "loss": 0.5011,
-      "step": 2655
-    },
-    {
-      "epoch": 6.551724137931035,
-      "grad_norm": 0.9765625,
-      "learning_rate": 0.0001384998999665479,
-      "loss": 0.4946,
-      "step": 2660
-    },
-    {
-      "epoch": 6.564039408866995,
-      "grad_norm": 0.953125,
-      "learning_rate": 0.00013823524402522804,
-      "loss": 0.5043,
-      "step": 2665
-    },
-    {
-      "epoch": 6.5763546798029555,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00013797027404481184,
-      "loss": 0.512,
-      "step": 2670
-    },
-    {
-      "epoch": 6.588669950738916,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.00013770499220158816,
-      "loss": 0.4999,
-      "step": 2675
-    },
-    {
-      "epoch": 6.600985221674877,
-      "grad_norm": 0.890625,
-      "learning_rate": 0.0001374394006744072,
-      "loss": 0.5051,
-      "step": 2680
-    },
-    {
-      "epoch": 6.613300492610837,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.0001371735016446627,
-      "loss": 0.5081,
-      "step": 2685
-    },
-    {
-      "epoch": 6.625615763546798,
-      "grad_norm": 0.88671875,
-      "learning_rate": 0.0001369072972962741,
-      "loss": 0.5027,
-      "step": 2690
-    },
-    {
-      "epoch": 6.637931034482759,
-      "grad_norm": 0.890625,
-      "learning_rate": 0.00013664078981566843,
-      "loss": 0.5062,
-      "step": 2695
-    },
-    {
-      "epoch": 6.650246305418719,
-      "grad_norm": 0.99609375,
-      "learning_rate": 0.00013637398139176255,
-      "loss": 0.5036,
-      "step": 2700
-    },
-    {
-      "epoch": 6.66256157635468,
-      "grad_norm": 0.859375,
-      "learning_rate": 0.00013610687421594498,
-      "loss": 0.4933,
-      "step": 2705
-    },
-    {
-      "epoch": 6.674876847290641,
-      "grad_norm": 0.99609375,
-      "learning_rate": 0.00013583947048205808,
-      "loss": 0.5036,
-      "step": 2710
-    },
-    {
-      "epoch": 6.687192118226601,
-      "grad_norm": 0.8828125,
-      "learning_rate": 0.00013557177238637986,
-      "loss": 0.5176,
-      "step": 2715
-    },
-    {
-      "epoch": 6.699507389162561,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00013530378212760606,
-      "loss": 0.5205,
-      "step": 2720
-    },
-    {
-      "epoch": 6.711822660098522,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00013503550190683205,
-      "loss": 0.5051,
-      "step": 2725
-    },
-    {
-      "epoch": 6.724137931034483,
-      "grad_norm": 0.9921875,
-      "learning_rate": 0.00013476693392753476,
-      "loss": 0.5086,
-      "step": 2730
-    },
-    {
-      "epoch": 6.736453201970443,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.0001344980803955546,
-      "loss": 0.5025,
-      "step": 2735
-    },
-    {
-      "epoch": 6.748768472906404,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.00013422894351907726,
-      "loss": 0.5062,
-      "step": 2740
-    },
-    {
-      "epoch": 6.761083743842365,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.00013395952550861572,
-      "loss": 0.5228,
-      "step": 2745
-    },
-    {
-      "epoch": 6.773399014778326,
-      "grad_norm": 0.87109375,
-      "learning_rate": 0.00013368982857699192,
-      "loss": 0.5049,
-      "step": 2750
-    },
-    {
-      "epoch": 6.785714285714286,
-      "grad_norm": 1.1015625,
-      "learning_rate": 0.00013341985493931877,
-      "loss": 0.5163,
-      "step": 2755
-    },
-    {
-      "epoch": 6.798029556650246,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00013314960681298175,
-      "loss": 0.5118,
-      "step": 2760
-    },
-    {
-      "epoch": 6.810344827586206,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.0001328790864176209,
-      "loss": 0.508,
-      "step": 2765
-    },
-    {
-      "epoch": 6.822660098522167,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00013260829597511246,
-      "loss": 0.5117,
-      "step": 2770
-    },
-    {
-      "epoch": 6.834975369458128,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.0001323372377095507,
-      "loss": 0.5178,
-      "step": 2775
-    },
-    {
-      "epoch": 6.847290640394089,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.0001320659138472295,
-      "loss": 0.51,
-      "step": 2780
-    },
-    {
-      "epoch": 6.859605911330049,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.00013179432661662434,
-      "loss": 0.5201,
-      "step": 2785
-    },
-    {
-      "epoch": 6.87192118226601,
-      "grad_norm": 0.94921875,
-      "learning_rate": 0.0001315224782483737,
-      "loss": 0.5145,
-      "step": 2790
-    },
-    {
-      "epoch": 6.884236453201971,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.00013125037097526097,
-      "loss": 0.5196,
-      "step": 2795
-    },
-    {
-      "epoch": 6.896551724137931,
-      "grad_norm": 0.875,
-      "learning_rate": 0.00013097800703219586,
-      "loss": 0.5215,
-      "step": 2800
-    },
-    {
-      "epoch": 6.9088669950738915,
-      "grad_norm": 1.09375,
-      "learning_rate": 0.00013070538865619642,
-      "loss": 0.5076,
-      "step": 2805
-    },
-    {
-      "epoch": 6.921182266009852,
-      "grad_norm": 0.890625,
-      "learning_rate": 0.00013043251808637026,
-      "loss": 0.5065,
-      "step": 2810
-    },
-    {
-      "epoch": 6.933497536945813,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.00013015939756389643,
-      "loss": 0.5052,
-      "step": 2815
-    },
-    {
-      "epoch": 6.945812807881773,
-      "grad_norm": 0.8671875,
-      "learning_rate": 0.00012988602933200689,
-      "loss": 0.5103,
-      "step": 2820
-    },
-    {
-      "epoch": 6.958128078817734,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00012961241563596817,
-      "loss": 0.5088,
-      "step": 2825
-    },
-    {
-      "epoch": 6.970443349753695,
-      "grad_norm": 1.109375,
-      "learning_rate": 0.00012933855872306285,
-      "loss": 0.5151,
-      "step": 2830
-    },
-    {
-      "epoch": 6.982758620689655,
-      "grad_norm": 0.8984375,
-      "learning_rate": 0.0001290644608425711,
-      "loss": 0.5133,
-      "step": 2835
-    },
-    {
-      "epoch": 6.995073891625616,
-      "grad_norm": 0.95703125,
-      "learning_rate": 0.0001287901242457523,
-      "loss": 0.5095,
-      "step": 2840
-    },
-    {
-      "epoch": 7.0,
-      "eval_loss": 3.5735390186309814,
-      "eval_runtime": 2.0433,
-      "eval_samples_per_second": 4.894,
-      "eval_steps_per_second": 0.979,
-      "step": 2842
-    },
-    {
-      "epoch": 7.0073891625615765,
-      "grad_norm": 0.8046875,
-      "learning_rate": 0.00012851555118582644,
-      "loss": 0.4549,
-      "step": 2845
-    },
-    {
-      "epoch": 7.019704433497537,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.0001282407439179557,
-      "loss": 0.4114,
-      "step": 2850
-    },
-    {
-      "epoch": 7.032019704433497,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.00012796570469922588,
-      "loss": 0.416,
-      "step": 2855
-    },
-    {
-      "epoch": 7.044334975369458,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.00012769043578862786,
-      "loss": 0.4192,
-      "step": 2860
-    },
-    {
-      "epoch": 7.056650246305419,
-      "grad_norm": 1.1875,
-      "learning_rate": 0.00012741493944703905,
-      "loss": 0.4192,
-      "step": 2865
-    },
-    {
-      "epoch": 7.068965517241379,
-      "grad_norm": 0.99609375,
-      "learning_rate": 0.0001271392179372048,
-      "loss": 0.4203,
-      "step": 2870
-    },
-    {
-      "epoch": 7.08128078817734,
-      "grad_norm": 0.953125,
-      "learning_rate": 0.00012686327352371999,
-      "loss": 0.4141,
-      "step": 2875
-    },
-    {
-      "epoch": 7.093596059113301,
-      "grad_norm": 1.1015625,
-      "learning_rate": 0.0001265871084730101,
-      "loss": 0.4252,
-      "step": 2880
-    },
-    {
-      "epoch": 7.105911330049261,
-      "grad_norm": 1.171875,
-      "learning_rate": 0.0001263107250533128,
-      "loss": 0.4253,
-      "step": 2885
-    },
-    {
-      "epoch": 7.1182266009852215,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.0001260341255346595,
-      "loss": 0.4182,
-      "step": 2890
-    },
-    {
-      "epoch": 7.130541871921182,
-      "grad_norm": 0.9296875,
-      "learning_rate": 0.00012575731218885625,
-      "loss": 0.4157,
-      "step": 2895
-    },
-    {
-      "epoch": 7.142857142857143,
-      "grad_norm": 1.140625,
-      "learning_rate": 0.0001254802872894655,
-      "loss": 0.4293,
-      "step": 2900
-    },
-    {
-      "epoch": 7.155172413793103,
-      "grad_norm": 0.87890625,
-      "learning_rate": 0.00012520305311178716,
-      "loss": 0.4264,
-      "step": 2905
-    },
-    {
-      "epoch": 7.167487684729064,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.00012492561193284008,
-      "loss": 0.4378,
-      "step": 2910
-    },
-    {
-      "epoch": 7.179802955665025,
-      "grad_norm": 0.99609375,
-      "learning_rate": 0.00012464796603134327,
-      "loss": 0.428,
-      "step": 2915
-    },
-    {
-      "epoch": 7.192118226600985,
-      "grad_norm": 0.94921875,
-      "learning_rate": 0.0001243701176876972,
-      "loss": 0.4227,
-      "step": 2920
-    },
-    {
-      "epoch": 7.204433497536946,
-      "grad_norm": 1.046875,
-      "learning_rate": 0.00012409206918396503,
-      "loss": 0.4171,
-      "step": 2925
-    },
-    {
-      "epoch": 7.216748768472907,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.00012381382280385393,
-      "loss": 0.422,
-      "step": 2930
-    },
-    {
-      "epoch": 7.2290640394088665,
-      "grad_norm": 0.9609375,
-      "learning_rate": 0.00012353538083269633,
-      "loss": 0.4248,
-      "step": 2935
-    },
-    {
-      "epoch": 7.241379310344827,
-      "grad_norm": 0.98828125,
-      "learning_rate": 0.00012325674555743106,
-      "loss": 0.4217,
-      "step": 2940
-    },
-    {
-      "epoch": 7.253694581280788,
-      "grad_norm": 1.4609375,
-      "learning_rate": 0.0001229779192665846,
-      "loss": 0.4328,
-      "step": 2945
-    },
-    {
-      "epoch": 7.266009852216749,
-      "grad_norm": 0.9765625,
-      "learning_rate": 0.0001226989042502524,
-      "loss": 0.4311,
-      "step": 2950
-    },
-    {
-      "epoch": 7.278325123152709,
-      "grad_norm": 1.578125,
-      "learning_rate": 0.0001224197028000799,
-      "loss": 0.4311,
-      "step": 2955
-    },
-    {
-      "epoch": 7.29064039408867,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00012214031720924384,
-      "loss": 0.4364,
-      "step": 2960
-    },
-    {
-      "epoch": 7.302955665024631,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.0001218607497724333,
-      "loss": 0.4302,
-      "step": 2965
-    },
-    {
-      "epoch": 7.315270935960591,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.000121581002785831,
-      "loss": 0.4323,
-      "step": 2970
-    },
-    {
-      "epoch": 7.327586206896552,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.0001213010785470943,
-      "loss": 0.4299,
-      "step": 2975
-    },
-    {
-      "epoch": 7.3399014778325125,
-      "grad_norm": 0.9609375,
-      "learning_rate": 0.0001210209793553364,
-      "loss": 0.4333,
-      "step": 2980
-    },
-    {
-      "epoch": 7.352216748768473,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00012074070751110751,
-      "loss": 0.4451,
-      "step": 2985
-    },
-    {
-      "epoch": 7.364532019704433,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.00012046026531637587,
-      "loss": 0.4346,
-      "step": 2990
-    },
-    {
-      "epoch": 7.376847290640394,
-      "grad_norm": 0.98828125,
-      "learning_rate": 0.00012017965507450877,
-      "loss": 0.4395,
-      "step": 2995
-    },
-    {
-      "epoch": 7.389162561576355,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.00011989887909025388,
-      "loss": 0.4322,
-      "step": 3000
-    },
-    {
-      "epoch": 7.401477832512315,
-      "grad_norm": 0.9453125,
-      "learning_rate": 0.00011961793966972004,
-      "loss": 0.4329,
-      "step": 3005
-    },
-    {
-      "epoch": 7.413793103448276,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.00011933683912035856,
-      "loss": 0.4367,
-      "step": 3010
-    },
-    {
-      "epoch": 7.426108374384237,
-      "grad_norm": 0.98046875,
-      "learning_rate": 0.00011905557975094406,
-      "loss": 0.4352,
-      "step": 3015
-    },
-    {
-      "epoch": 7.4384236453201975,
-      "grad_norm": 0.9765625,
-      "learning_rate": 0.00011877416387155565,
-      "loss": 0.4357,
-      "step": 3020
-    },
-    {
-      "epoch": 7.4507389162561575,
-      "grad_norm": 0.96875,
-      "learning_rate": 0.0001184925937935579,
-      "loss": 0.4374,
-      "step": 3025
-    },
-    {
-      "epoch": 7.463054187192118,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00011821087182958186,
-      "loss": 0.4334,
-      "step": 3030
-    },
-    {
-      "epoch": 7.475369458128079,
-      "grad_norm": 0.87890625,
-      "learning_rate": 0.00011792900029350611,
-      "loss": 0.4384,
-      "step": 3035
-    },
-    {
-      "epoch": 7.487684729064039,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00011764698150043767,
-      "loss": 0.4402,
-      "step": 3040
-    },
-    {
-      "epoch": 7.5,
-      "grad_norm": 0.984375,
-      "learning_rate": 0.00011736481776669306,
-      "loss": 0.4362,
-      "step": 3045
-    },
-    {
-      "epoch": 7.512315270935961,
-      "grad_norm": 0.98828125,
-      "learning_rate": 0.00011708251140977918,
-      "loss": 0.4403,
-      "step": 3050
-    },
-    {
-      "epoch": 7.524630541871921,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00011680006474837446,
-      "loss": 0.4471,
-      "step": 3055
-    },
-    {
-      "epoch": 7.536945812807882,
-      "grad_norm": 0.9609375,
-      "learning_rate": 0.0001165174801023096,
-      "loss": 0.4502,
-      "step": 3060
-    },
-    {
-      "epoch": 7.5492610837438425,
-      "grad_norm": 1.25,
-      "learning_rate": 0.0001162347597925487,
-      "loss": 0.4433,
-      "step": 3065
-    },
-    {
-      "epoch": 7.5615763546798025,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.00011595190614117002,
-      "loss": 0.4485,
-      "step": 3070
-    },
-    {
-      "epoch": 7.573891625615763,
-      "grad_norm": 1.0703125,
-      "learning_rate": 0.00011566892147134705,
-      "loss": 0.445,
-      "step": 3075
-    },
-    {
-      "epoch": 7.586206896551724,
-      "grad_norm": 1.0546875,
-      "learning_rate": 0.00011538580810732938,
-      "loss": 0.4421,
-      "step": 3080
-    },
-    {
-      "epoch": 7.598522167487685,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00011510256837442359,
-      "loss": 0.4356,
-      "step": 3085
-    },
-    {
-      "epoch": 7.610837438423645,
-      "grad_norm": 0.94140625,
-      "learning_rate": 0.00011481920459897417,
-      "loss": 0.4466,
-      "step": 3090
-    },
-    {
-      "epoch": 7.623152709359606,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00011453571910834449,
-      "loss": 0.4517,
-      "step": 3095
-    },
-    {
-      "epoch": 7.635467980295567,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.00011425211423089748,
-      "loss": 0.4484,
-      "step": 3100
-    },
-    {
-      "epoch": 7.647783251231527,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.00011396839229597674,
-      "loss": 0.4412,
-      "step": 3105
-    },
-    {
-      "epoch": 7.6600985221674875,
-      "grad_norm": 0.953125,
-      "learning_rate": 0.0001136845556338872,
-      "loss": 0.4366,
-      "step": 3110
-    },
-    {
-      "epoch": 7.672413793103448,
-      "grad_norm": 0.94140625,
-      "learning_rate": 0.00011340060657587623,
-      "loss": 0.4506,
-      "step": 3115
-    },
-    {
-      "epoch": 7.684729064039409,
-      "grad_norm": 0.9453125,
-      "learning_rate": 0.00011311654745411425,
-      "loss": 0.4297,
-      "step": 3120
-    },
-    {
-      "epoch": 7.697044334975369,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00011283238060167563,
-      "loss": 0.4487,
-      "step": 3125
-    },
-    {
-      "epoch": 7.70935960591133,
-      "grad_norm": 1.1015625,
-      "learning_rate": 0.00011254810835251963,
-      "loss": 0.4494,
-      "step": 3130
-    },
-    {
-      "epoch": 7.721674876847291,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.00011226373304147123,
-      "loss": 0.4459,
-      "step": 3135
-    },
-    {
-      "epoch": 7.733990147783251,
-      "grad_norm": 1.015625,
-      "learning_rate": 0.00011197925700420173,
-      "loss": 0.4441,
-      "step": 3140
-    },
-    {
-      "epoch": 7.746305418719212,
-      "grad_norm": 1.109375,
-      "learning_rate": 0.0001116946825772099,
-      "loss": 0.4444,
-      "step": 3145
-    },
-    {
-      "epoch": 7.758620689655173,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.00011141001209780249,
-      "loss": 0.4515,
-      "step": 3150
-    },
-    {
-      "epoch": 7.7709359605911335,
-      "grad_norm": 1.1171875,
-      "learning_rate": 0.00011112524790407524,
-      "loss": 0.4422,
-      "step": 3155
-    },
-    {
-      "epoch": 7.783251231527093,
-      "grad_norm": 0.90625,
-      "learning_rate": 0.00011084039233489354,
-      "loss": 0.454,
-      "step": 3160
-    },
-    {
-      "epoch": 7.795566502463054,
-      "grad_norm": 0.9296875,
-      "learning_rate": 0.00011055544772987335,
-      "loss": 0.4388,
-      "step": 3165
-    },
-    {
-      "epoch": 7.807881773399015,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.0001102704164293618,
-      "loss": 0.4458,
-      "step": 3170
-    },
-    {
-      "epoch": 7.820197044334975,
-      "grad_norm": 1.078125,
-      "learning_rate": 0.00010998530077441824,
-      "loss": 0.4454,
-      "step": 3175
-    },
-    {
-      "epoch": 7.832512315270936,
-      "grad_norm": 0.9609375,
-      "learning_rate": 0.0001097001031067947,
-      "loss": 0.4486,
-      "step": 3180
-    },
-    {
-      "epoch": 7.844827586206897,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00010941482576891689,
-      "loss": 0.4508,
-      "step": 3185
-    },
-    {
-      "epoch": 7.857142857142857,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.00010912947110386484,
-      "loss": 0.4569,
-      "step": 3190
-    },
-    {
-      "epoch": 7.869458128078818,
-      "grad_norm": 0.9921875,
-      "learning_rate": 0.00010884404145535372,
-      "loss": 0.4459,
-      "step": 3195
-    },
-    {
-      "epoch": 7.8817733990147785,
-      "grad_norm": 1.1015625,
-      "learning_rate": 0.00010855853916771457,
-      "loss": 0.4539,
-      "step": 3200
-    },
-    {
-      "epoch": 7.894088669950738,
-      "grad_norm": 0.9609375,
-      "learning_rate": 0.00010827296658587503,
-      "loss": 0.4598,
-      "step": 3205
-    },
-    {
-      "epoch": 7.906403940886699,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00010798732605534006,
-      "loss": 0.4465,
-      "step": 3210
-    },
-    {
-      "epoch": 7.91871921182266,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00010770161992217271,
-      "loss": 0.4448,
-      "step": 3215
-    },
-    {
-      "epoch": 7.931034482758621,
-      "grad_norm": 1.0078125,
-      "learning_rate": 0.00010741585053297494,
-      "loss": 0.4485,
-      "step": 3220
-    },
-    {
-      "epoch": 7.943349753694581,
-      "grad_norm": 0.95703125,
-      "learning_rate": 0.00010713002023486816,
-      "loss": 0.4487,
-      "step": 3225
-    },
-    {
-      "epoch": 7.955665024630542,
-      "grad_norm": 0.94921875,
-      "learning_rate": 0.00010684413137547404,
-      "loss": 0.4432,
-      "step": 3230
-    },
-    {
-      "epoch": 7.967980295566503,
-      "grad_norm": 0.9765625,
-      "learning_rate": 0.00010655818630289525,
-      "loss": 0.4517,
-      "step": 3235
-    },
-    {
-      "epoch": 7.980295566502463,
-      "grad_norm": 0.98046875,
-      "learning_rate": 0.00010627218736569624,
-      "loss": 0.4366,
-      "step": 3240
-    },
-    {
-      "epoch": 7.9926108374384235,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.00010598613691288372,
-      "loss": 0.4514,
-      "step": 3245
-    },
-    {
-      "epoch": 8.0,
-      "eval_loss": 3.931917190551758,
-      "eval_runtime": 2.0448,
-      "eval_samples_per_second": 4.89,
-      "eval_steps_per_second": 0.978,
-      "step": 3248
-    },
-    {
-      "epoch": 8.004926108374384,
-      "grad_norm": 0.80859375,
-      "learning_rate": 0.00010570003729388767,
-      "loss": 0.4202,
-      "step": 3250
-    },
-    {
-      "epoch": 8.017241379310345,
-      "grad_norm": 1.0625,
-      "learning_rate": 0.00010541389085854176,
-      "loss": 0.369,
-      "step": 3255
-    },
-    {
-      "epoch": 8.029556650246306,
-      "grad_norm": 0.91015625,
-      "learning_rate": 0.00010512769995706426,
-      "loss": 0.3781,
-      "step": 3260
-    },
-    {
-      "epoch": 8.041871921182265,
-      "grad_norm": 0.98046875,
-      "learning_rate": 0.0001048414669400386,
-      "loss": 0.3687,
-      "step": 3265
-    },
-    {
-      "epoch": 8.054187192118226,
-      "grad_norm": 1.0859375,
-      "learning_rate": 0.00010455519415839415,
-      "loss": 0.3635,
-      "step": 3270
-    },
-    {
-      "epoch": 8.066502463054187,
-      "grad_norm": 0.859375,
-      "learning_rate": 0.00010426888396338688,
-      "loss": 0.3642,
-      "step": 3275
-    },
-    {
-      "epoch": 8.078817733990148,
-      "grad_norm": 0.90234375,
-      "learning_rate": 0.00010398253870658006,
-      "loss": 0.3715,
-      "step": 3280
-    },
-    {
-      "epoch": 8.091133004926109,
-      "grad_norm": 0.984375,
-      "learning_rate": 0.00010369616073982491,
-      "loss": 0.3667,
-      "step": 3285
-    },
-    {
-      "epoch": 8.10344827586207,
-      "grad_norm": 0.9296875,
-      "learning_rate": 0.00010340975241524132,
-      "loss": 0.373,
-      "step": 3290
-    },
-    {
-      "epoch": 8.11576354679803,
-      "grad_norm": 0.92578125,
-      "learning_rate": 0.00010312331608519847,
-      "loss": 0.3717,
-      "step": 3295
-    },
-    {
-      "epoch": 8.12807881773399,
-      "grad_norm": 0.98828125,
-      "learning_rate": 0.00010283685410229571,
-      "loss": 0.3777,
-      "step": 3300
-    },
-    {
-      "epoch": 8.14039408866995,
-      "grad_norm": 0.9609375,
-      "learning_rate": 0.0001025503688193429,
-      "loss": 0.3782,
-      "step": 3305
-    },
-    {
-      "epoch": 8.152709359605911,
-      "grad_norm": 0.9375,
-      "learning_rate": 0.00010226386258934138,
-      "loss": 0.3792,
-      "step": 3310
-    },
-    {
-      "epoch": 8.165024630541872,
-      "grad_norm": 0.96875,
-      "learning_rate": 0.00010197733776546447,
-      "loss": 0.3745,
-      "step": 3315
-    },
-    {
-      "epoch": 8.177339901477833,
-      "grad_norm": 0.91015625,
-      "learning_rate": 0.00010169079670103831,
-      "loss": 0.378,
-      "step": 3320
-    },
-    {
-      "epoch": 8.189655172413794,
-      "grad_norm": 1.0234375,
-      "learning_rate": 0.00010140424174952232,
-      "loss": 0.3783,
-      "step": 3325
-    },
-    {
-      "epoch": 8.201970443349754,
-      "grad_norm": 1.0,
-      "learning_rate": 0.00010111767526449004,
-      "loss": 0.3788,
-      "step": 3330
-    },
-    {
-      "epoch": 8.214285714285714,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00010083109959960973,
-      "loss": 0.3752,
-      "step": 3335
-    },
-    {
-      "epoch": 8.226600985221674,
-      "grad_norm": 0.97265625,
-      "learning_rate": 0.00010054451710862498,
-      "loss": 0.3794,
-      "step": 3340
-    },
-    {
-      "epoch": 8.238916256157635,
-      "grad_norm": 1.0390625,
-      "learning_rate": 0.00010025793014533558,
-      "loss": 0.381,
-      "step": 3345
-    },
-    {
-      "epoch": 8.251231527093596,
-      "grad_norm": 0.91796875,
-      "learning_rate": 9.997134106357796e-05,
-      "loss": 0.3834,
-      "step": 3350
-    },
-    {
-      "epoch": 8.263546798029557,
-      "grad_norm": 0.95703125,
-      "learning_rate": 9.968475221720594e-05,
-      "loss": 0.3791,
-      "step": 3355
-    },
-    {
-      "epoch": 8.275862068965518,
-      "grad_norm": 1.015625,
-      "learning_rate": 9.939816596007146e-05,
-      "loss": 0.377,
-      "step": 3360
-    },
-    {
-      "epoch": 8.288177339901479,
-      "grad_norm": 1.0703125,
-      "learning_rate": 9.911158464600517e-05,
-      "loss": 0.3764,
-      "step": 3365
-    },
-    {
-      "epoch": 8.300492610837438,
-      "grad_norm": 0.97265625,
-      "learning_rate": 9.88250106287971e-05,
-      "loss": 0.3786,
-      "step": 3370
-    },
-    {
-      "epoch": 8.312807881773399,
-      "grad_norm": 0.98046875,
-      "learning_rate": 9.853844626217737e-05,
-      "loss": 0.3722,
-      "step": 3375
-    },
-    {
-      "epoch": 8.32512315270936,
-      "grad_norm": 0.96875,
-      "learning_rate": 9.825189389979683e-05,
-      "loss": 0.3847,
-      "step": 3380
-    },
-    {
-      "epoch": 8.33743842364532,
-      "grad_norm": 1.046875,
-      "learning_rate": 9.796535589520777e-05,
-      "loss": 0.377,
-      "step": 3385
-    },
-    {
-      "epoch": 8.349753694581281,
-      "grad_norm": 1.078125,
-      "learning_rate": 9.767883460184443e-05,
-      "loss": 0.3799,
-      "step": 3390
-    },
-    {
-      "epoch": 8.362068965517242,
-      "grad_norm": 1.03125,
-      "learning_rate": 9.739233237300402e-05,
-      "loss": 0.3878,
-      "step": 3395
-    },
-    {
-      "epoch": 8.374384236453203,
-      "grad_norm": 0.9765625,
-      "learning_rate": 9.710585156182695e-05,
-      "loss": 0.3799,
-      "step": 3400
-    },
-    {
-      "epoch": 8.386699507389162,
-      "grad_norm": 0.97265625,
-      "learning_rate": 9.681939452127784e-05,
-      "loss": 0.3811,
-      "step": 3405
-    },
-    {
-      "epoch": 8.399014778325123,
-      "grad_norm": 1.0078125,
-      "learning_rate": 9.653296360412602e-05,
-      "loss": 0.3893,
-      "step": 3410
-    },
-    {
-      "epoch": 8.411330049261084,
-      "grad_norm": 0.9453125,
-      "learning_rate": 9.624656116292628e-05,
-      "loss": 0.3812,
-      "step": 3415
-    },
-    {
-      "epoch": 8.423645320197044,
-      "grad_norm": 1.015625,
-      "learning_rate": 9.596018954999953e-05,
-      "loss": 0.3866,
-      "step": 3420
-    },
-    {
-      "epoch": 8.435960591133005,
-      "grad_norm": 1.0546875,
-      "learning_rate": 9.567385111741345e-05,
-      "loss": 0.3845,
-      "step": 3425
-    },
-    {
-      "epoch": 8.448275862068966,
-      "grad_norm": 0.98046875,
-      "learning_rate": 9.538754821696323e-05,
-      "loss": 0.3843,
-      "step": 3430
-    },
-    {
-      "epoch": 8.460591133004925,
-      "grad_norm": 1.1875,
-      "learning_rate": 9.510128320015224e-05,
-      "loss": 0.3884,
-      "step": 3435
-    },
-    {
-      "epoch": 8.472906403940886,
-      "grad_norm": 0.94921875,
-      "learning_rate": 9.481505841817265e-05,
-      "loss": 0.3778,
-      "step": 3440
-    },
-    {
-      "epoch": 8.485221674876847,
-      "grad_norm": 0.99609375,
-      "learning_rate": 9.452887622188619e-05,
-      "loss": 0.3829,
-      "step": 3445
-    },
-    {
-      "epoch": 8.497536945812808,
-      "grad_norm": 1.1484375,
-      "learning_rate": 9.424273896180482e-05,
-      "loss": 0.3807,
-      "step": 3450
-    },
-    {
-      "epoch": 8.509852216748769,
-      "grad_norm": 1.0625,
-      "learning_rate": 9.395664898807142e-05,
-      "loss": 0.3887,
-      "step": 3455
-    },
-    {
-      "epoch": 8.52216748768473,
-      "grad_norm": 0.94140625,
-      "learning_rate": 9.367060865044049e-05,
-      "loss": 0.386,
-      "step": 3460
-    },
-    {
-      "epoch": 8.53448275862069,
-      "grad_norm": 0.953125,
-      "learning_rate": 9.338462029825886e-05,
-      "loss": 0.3859,
-      "step": 3465
-    },
-    {
-      "epoch": 8.54679802955665,
-      "grad_norm": 0.984375,
-      "learning_rate": 9.309868628044633e-05,
-      "loss": 0.3855,
-      "step": 3470
-    },
-    {
-      "epoch": 8.55911330049261,
-      "grad_norm": 0.96484375,
-      "learning_rate": 9.281280894547657e-05,
-      "loss": 0.3803,
-      "step": 3475
-    },
-    {
-      "epoch": 8.571428571428571,
-      "grad_norm": 1.03125,
-      "learning_rate": 9.252699064135758e-05,
-      "loss": 0.3793,
-      "step": 3480
-    },
-    {
-      "epoch": 8.583743842364532,
-      "grad_norm": 0.98046875,
-      "learning_rate": 9.224123371561252e-05,
-      "loss": 0.3772,
-      "step": 3485
-    },
-    {
-      "epoch": 8.596059113300493,
-      "grad_norm": 1.0078125,
-      "learning_rate": 9.19555405152605e-05,
-      "loss": 0.3885,
-      "step": 3490
-    },
-    {
-      "epoch": 8.608374384236454,
-      "grad_norm": 1.015625,
-      "learning_rate": 9.166991338679715e-05,
-      "loss": 0.3886,
-      "step": 3495
-    },
-    {
-      "epoch": 8.620689655172415,
-      "grad_norm": 0.9296875,
-      "learning_rate": 9.138435467617548e-05,
-      "loss": 0.3765,
-      "step": 3500
-    },
-    {
-      "epoch": 8.633004926108374,
-      "grad_norm": 0.984375,
-      "learning_rate": 9.109886672878653e-05,
-      "loss": 0.3931,
-      "step": 3505
-    },
-    {
-      "epoch": 8.645320197044335,
-      "grad_norm": 0.984375,
-      "learning_rate": 9.081345188944019e-05,
-      "loss": 0.3872,
-      "step": 3510
-    },
-    {
-      "epoch": 8.657635467980295,
-      "grad_norm": 0.95703125,
-      "learning_rate": 9.052811250234579e-05,
-      "loss": 0.3802,
-      "step": 3515
-    },
-    {
-      "epoch": 8.669950738916256,
-      "grad_norm": 1.0390625,
-      "learning_rate": 9.024285091109309e-05,
-      "loss": 0.3886,
-      "step": 3520
-    },
-    {
-      "epoch": 8.682266009852217,
-      "grad_norm": 0.96484375,
-      "learning_rate": 8.995766945863277e-05,
-      "loss": 0.3814,
-      "step": 3525
-    },
-    {
-      "epoch": 8.694581280788178,
-      "grad_norm": 1.0234375,
-      "learning_rate": 8.967257048725733e-05,
-      "loss": 0.3886,
-      "step": 3530
-    },
-    {
-      "epoch": 8.706896551724139,
-      "grad_norm": 1.0703125,
-      "learning_rate": 8.938755633858186e-05,
-      "loss": 0.3889,
-      "step": 3535
-    },
-    {
-      "epoch": 8.719211822660098,
-      "grad_norm": 0.95703125,
-      "learning_rate": 8.91026293535247e-05,
-      "loss": 0.3838,
-      "step": 3540
-    },
-    {
-      "epoch": 8.731527093596059,
-      "grad_norm": 1.0,
-      "learning_rate": 8.881779187228836e-05,
-      "loss": 0.3868,
-      "step": 3545
-    },
-    {
-      "epoch": 8.74384236453202,
-      "grad_norm": 1.0234375,
-      "learning_rate": 8.853304623434021e-05,
-      "loss": 0.3863,
-      "step": 3550
-    },
-    {
-      "epoch": 8.75615763546798,
-      "grad_norm": 1.0703125,
-      "learning_rate": 8.82483947783932e-05,
-      "loss": 0.3873,
-      "step": 3555
-    },
-    {
-      "epoch": 8.768472906403941,
-      "grad_norm": 1.1328125,
-      "learning_rate": 8.796383984238688e-05,
-      "loss": 0.3944,
-      "step": 3560
-    },
-    {
-      "epoch": 8.780788177339902,
-      "grad_norm": 1.0703125,
-      "learning_rate": 8.767938376346792e-05,
-      "loss": 0.3927,
-      "step": 3565
-    },
-    {
-      "epoch": 8.793103448275861,
-      "grad_norm": 1.03125,
-      "learning_rate": 8.739502887797107e-05,
-      "loss": 0.3865,
-      "step": 3570
-    },
-    {
-      "epoch": 8.805418719211822,
-      "grad_norm": 1.125,
-      "learning_rate": 8.71107775214e-05,
-      "loss": 0.3827,
-      "step": 3575
-    },
-    {
-      "epoch": 8.817733990147783,
-      "grad_norm": 1.015625,
-      "learning_rate": 8.682663202840802e-05,
-      "loss": 0.3934,
-      "step": 3580
-    },
-    {
-      "epoch": 8.830049261083744,
-      "grad_norm": 1.015625,
-      "learning_rate": 8.654259473277892e-05,
-      "loss": 0.3939,
-      "step": 3585
-    },
-    {
-      "epoch": 8.842364532019705,
-      "grad_norm": 1.0234375,
-      "learning_rate": 8.625866796740787e-05,
-      "loss": 0.3933,
-      "step": 3590
-    },
-    {
-      "epoch": 8.854679802955665,
-      "grad_norm": 1.0625,
-      "learning_rate": 8.597485406428219e-05,
-      "loss": 0.3861,
-      "step": 3595
-    },
-    {
-      "epoch": 8.866995073891626,
-      "grad_norm": 1.0390625,
-      "learning_rate": 8.569115535446228e-05,
-      "loss": 0.3888,
-      "step": 3600
-    },
-    {
-      "epoch": 8.879310344827585,
-      "grad_norm": 0.984375,
-      "learning_rate": 8.540757416806236e-05,
-      "loss": 0.3876,
-      "step": 3605
-    },
-    {
-      "epoch": 8.891625615763546,
-      "grad_norm": 0.97265625,
-      "learning_rate": 8.51241128342314e-05,
-      "loss": 0.3868,
-      "step": 3610
-    },
-    {
-      "epoch": 8.903940886699507,
-      "grad_norm": 0.9765625,
-      "learning_rate": 8.484077368113399e-05,
-      "loss": 0.3812,
-      "step": 3615
-    },
-    {
-      "epoch": 8.916256157635468,
-      "grad_norm": 1.03125,
-      "learning_rate": 8.45575590359312e-05,
-      "loss": 0.3845,
-      "step": 3620
-    },
-    {
-      "epoch": 8.928571428571429,
-      "grad_norm": 0.9609375,
-      "learning_rate": 8.427447122476148e-05,
-      "loss": 0.3873,
-      "step": 3625
-    },
-    {
-      "epoch": 8.94088669950739,
-      "grad_norm": 0.91015625,
-      "learning_rate": 8.399151257272156e-05,
-      "loss": 0.3944,
-      "step": 3630
-    },
-    {
-      "epoch": 8.95320197044335,
-      "grad_norm": 0.9765625,
-      "learning_rate": 8.37086854038473e-05,
-      "loss": 0.3888,
-      "step": 3635
-    },
-    {
-      "epoch": 8.96551724137931,
-      "grad_norm": 1.0625,
-      "learning_rate": 8.342599204109472e-05,
-      "loss": 0.3815,
-      "step": 3640
-    },
-    {
-      "epoch": 8.97783251231527,
-      "grad_norm": 1.03125,
-      "learning_rate": 8.314343480632078e-05,
-      "loss": 0.387,
-      "step": 3645
-    },
-    {
-      "epoch": 8.990147783251231,
-      "grad_norm": 0.96875,
-      "learning_rate": 8.286101602026437e-05,
-      "loss": 0.3872,
-      "step": 3650
-    },
-    {
-      "epoch": 9.0,
-      "eval_loss": 4.438603401184082,
-      "eval_runtime": 2.0448,
-      "eval_samples_per_second": 4.89,
-      "eval_steps_per_second": 0.978,
-      "step": 3654
-    },
-    {
-      "epoch": 9.002463054187192,
-      "grad_norm": 0.890625,
-      "learning_rate": 8.257873800252732e-05,
-      "loss": 0.3883,
-      "step": 3655
-    },
-    {
-      "epoch": 9.014778325123153,
-      "grad_norm": 1.015625,
-      "learning_rate": 8.229660307155518e-05,
-      "loss": 0.3218,
-      "step": 3660
-    },
-    {
-      "epoch": 9.027093596059114,
-      "grad_norm": 0.9375,
-      "learning_rate": 8.20146135446184e-05,
-      "loss": 0.3247,
-      "step": 3665
-    },
-    {
-      "epoch": 9.039408866995075,
-      "grad_norm": 0.91796875,
-      "learning_rate": 8.173277173779305e-05,
-      "loss": 0.3305,
-      "step": 3670
-    },
-    {
-      "epoch": 9.051724137931034,
-      "grad_norm": 0.953125,
-      "learning_rate": 8.145107996594206e-05,
-      "loss": 0.3281,
-      "step": 3675
-    },
-    {
-      "epoch": 9.064039408866995,
-      "grad_norm": 1.0546875,
-      "learning_rate": 8.116954054269591e-05,
-      "loss": 0.3265,
-      "step": 3680
-    },
-    {
-      "epoch": 9.076354679802956,
-      "grad_norm": 0.9296875,
-      "learning_rate": 8.088815578043398e-05,
-      "loss": 0.3299,
-      "step": 3685
-    },
-    {
-      "epoch": 9.088669950738916,
-      "grad_norm": 1.0546875,
-      "learning_rate": 8.060692799026522e-05,
-      "loss": 0.3352,
-      "step": 3690
-    },
-    {
-      "epoch": 9.100985221674877,
-      "grad_norm": 1.0234375,
-      "learning_rate": 8.032585948200937e-05,
-      "loss": 0.3329,
-      "step": 3695
-    },
-    {
-      "epoch": 9.113300492610838,
-      "grad_norm": 0.9375,
-      "learning_rate": 8.004495256417792e-05,
-      "loss": 0.3284,
-      "step": 3700
-    },
-    {
-      "epoch": 9.125615763546797,
-      "grad_norm": 0.9765625,
-      "learning_rate": 7.976420954395518e-05,
-      "loss": 0.3326,
-      "step": 3705
-    },
-    {
-      "epoch": 9.137931034482758,
-      "grad_norm": 0.9921875,
-      "learning_rate": 7.948363272717926e-05,
-      "loss": 0.33,
-      "step": 3710
-    },
-    {
-      "epoch": 9.150246305418719,
-      "grad_norm": 1.0078125,
-      "learning_rate": 7.920322441832326e-05,
-      "loss": 0.3383,
-      "step": 3715
-    },
-    {
-      "epoch": 9.16256157635468,
-      "grad_norm": 1.03125,
-      "learning_rate": 7.892298692047621e-05,
-      "loss": 0.3309,
-      "step": 3720
-    },
-    {
-      "epoch": 9.17487684729064,
-      "grad_norm": 0.94140625,
-      "learning_rate": 7.864292253532427e-05,
-      "loss": 0.3284,
-      "step": 3725
-    },
-    {
-      "epoch": 9.187192118226601,
-      "grad_norm": 0.94140625,
-      "learning_rate": 7.83630335631317e-05,
-      "loss": 0.3315,
-      "step": 3730
-    },
-    {
-      "epoch": 9.199507389162562,
-      "grad_norm": 0.96484375,
-      "learning_rate": 7.808332230272209e-05,
-      "loss": 0.3373,
-      "step": 3735
-    },
-    {
-      "epoch": 9.211822660098521,
-      "grad_norm": 0.88671875,
-      "learning_rate": 7.780379105145934e-05,
-      "loss": 0.3277,
-      "step": 3740
-    },
-    {
-      "epoch": 9.224137931034482,
-      "grad_norm": 1.0859375,
-      "learning_rate": 7.752444210522898e-05,
-      "loss": 0.3281,
-      "step": 3745
-    },
-    {
-      "epoch": 9.236453201970443,
-      "grad_norm": 0.94921875,
-      "learning_rate": 7.724527775841914e-05,
-      "loss": 0.3263,
-      "step": 3750
-    },
-    {
-      "epoch": 9.248768472906404,
-      "grad_norm": 1.09375,
-      "learning_rate": 7.696630030390179e-05,
-      "loss": 0.3406,
-      "step": 3755
-    },
-    {
-      "epoch": 9.261083743842365,
-      "grad_norm": 0.97265625,
-      "learning_rate": 7.668751203301384e-05,
-      "loss": 0.3377,
-      "step": 3760
-    },
-    {
-      "epoch": 9.273399014778326,
-      "grad_norm": 1.046875,
-      "learning_rate": 7.64089152355385e-05,
-      "loss": 0.3383,
-      "step": 3765
-    },
-    {
-      "epoch": 9.285714285714286,
-      "grad_norm": 0.98828125,
-      "learning_rate": 7.613051219968623e-05,
-      "loss": 0.3396,
-      "step": 3770
-    },
-    {
-      "epoch": 9.298029556650246,
-      "grad_norm": 0.9296875,
-      "learning_rate": 7.585230521207608e-05,
-      "loss": 0.3349,
-      "step": 3775
-    },
-    {
-      "epoch": 9.310344827586206,
-      "grad_norm": 0.99609375,
-      "learning_rate": 7.55742965577169e-05,
-      "loss": 0.3306,
-      "step": 3780
-    },
-    {
-      "epoch": 9.322660098522167,
-      "grad_norm": 0.96875,
-      "learning_rate": 7.529648851998857e-05,
-      "loss": 0.3359,
-      "step": 3785
-    },
-    {
-      "epoch": 9.334975369458128,
-      "grad_norm": 1.015625,
-      "learning_rate": 7.501888338062323e-05,
-      "loss": 0.3442,
-      "step": 3790
-    },
-    {
-      "epoch": 9.347290640394089,
-      "grad_norm": 0.95703125,
-      "learning_rate": 7.474148341968652e-05,
-      "loss": 0.3426,
-      "step": 3795
-    },
-    {
-      "epoch": 9.35960591133005,
-      "grad_norm": 1.0390625,
-      "learning_rate": 7.446429091555889e-05,
-      "loss": 0.3405,
-      "step": 3800
-    },
-    {
-      "epoch": 9.37192118226601,
-      "grad_norm": 1.1953125,
-      "learning_rate": 7.418730814491697e-05,
-      "loss": 0.3418,
-      "step": 3805
-    },
-    {
-      "epoch": 9.38423645320197,
-      "grad_norm": 1.1171875,
-      "learning_rate": 7.391053738271466e-05,
-      "loss": 0.3317,
-      "step": 3810
-    },
-    {
-      "epoch": 9.39655172413793,
-      "grad_norm": 1.046875,
-      "learning_rate": 7.363398090216459e-05,
-      "loss": 0.3297,
-      "step": 3815
-    },
-    {
-      "epoch": 9.408866995073891,
-      "grad_norm": 0.98828125,
-      "learning_rate": 7.335764097471944e-05,
-      "loss": 0.3373,
-      "step": 3820
-    },
-    {
-      "epoch": 9.421182266009852,
-      "grad_norm": 0.953125,
-      "learning_rate": 7.308151987005326e-05,
-      "loss": 0.3337,
-      "step": 3825
-    },
-    {
-      "epoch": 9.433497536945813,
-      "grad_norm": 1.0546875,
-      "learning_rate": 7.28056198560428e-05,
-      "loss": 0.3383,
-      "step": 3830
-    },
-    {
-      "epoch": 9.445812807881774,
-      "grad_norm": 1.0078125,
-      "learning_rate": 7.2529943198749e-05,
-      "loss": 0.3356,
-      "step": 3835
-    },
-    {
-      "epoch": 9.458128078817733,
-      "grad_norm": 0.9921875,
-      "learning_rate": 7.225449216239821e-05,
-      "loss": 0.3315,
-      "step": 3840
-    },
-    {
-      "epoch": 9.470443349753694,
-      "grad_norm": 0.953125,
-      "learning_rate": 7.19792690093637e-05,
-      "loss": 0.3397,
-      "step": 3845
-    },
-    {
-      "epoch": 9.482758620689655,
-      "grad_norm": 1.109375,
-      "learning_rate": 7.170427600014712e-05,
-      "loss": 0.3378,
-      "step": 3850
-    },
-    {
-      "epoch": 9.495073891625616,
-      "grad_norm": 1.015625,
-      "learning_rate": 7.142951539335981e-05,
-      "loss": 0.3439,
-      "step": 3855
-    },
-    {
-      "epoch": 9.507389162561577,
-      "grad_norm": 1.046875,
-      "learning_rate": 7.115498944570427e-05,
-      "loss": 0.3331,
-      "step": 3860
-    },
-    {
-      "epoch": 9.519704433497537,
-      "grad_norm": 0.9375,
-      "learning_rate": 7.088070041195576e-05,
-      "loss": 0.3355,
-      "step": 3865
-    },
-    {
-      "epoch": 9.532019704433498,
-      "grad_norm": 1.109375,
-      "learning_rate": 7.060665054494362e-05,
-      "loss": 0.3301,
-      "step": 3870
-    },
-    {
-      "epoch": 9.544334975369457,
-      "grad_norm": 0.98046875,
-      "learning_rate": 7.033284209553286e-05,
-      "loss": 0.3329,
-      "step": 3875
-    },
-    {
-      "epoch": 9.556650246305418,
-      "grad_norm": 1.0078125,
-      "learning_rate": 7.005927731260562e-05,
-      "loss": 0.344,
-      "step": 3880
-    },
-    {
-      "epoch": 9.568965517241379,
-      "grad_norm": 0.98828125,
-      "learning_rate": 6.978595844304271e-05,
-      "loss": 0.3358,
-      "step": 3885
-    },
-    {
-      "epoch": 9.58128078817734,
-      "grad_norm": 0.9765625,
-      "learning_rate": 6.95128877317053e-05,
-      "loss": 0.3448,
-      "step": 3890
-    },
-    {
-      "epoch": 9.5935960591133,
-      "grad_norm": 1.03125,
-      "learning_rate": 6.924006742141618e-05,
-      "loss": 0.3338,
-      "step": 3895
-    },
-    {
-      "epoch": 9.605911330049262,
-      "grad_norm": 1.078125,
-      "learning_rate": 6.89674997529416e-05,
-      "loss": 0.3386,
-      "step": 3900
-    },
-    {
-      "epoch": 9.618226600985222,
-      "grad_norm": 1.0234375,
-      "learning_rate": 6.869518696497275e-05,
-      "loss": 0.3285,
-      "step": 3905
-    },
-    {
-      "epoch": 9.630541871921181,
-      "grad_norm": 1.078125,
-      "learning_rate": 6.842313129410741e-05,
-      "loss": 0.3447,
-      "step": 3910
-    },
-    {
-      "epoch": 9.642857142857142,
-      "grad_norm": 1.0625,
-      "learning_rate": 6.815133497483157e-05,
-      "loss": 0.3342,
-      "step": 3915
-    },
-    {
-      "epoch": 9.655172413793103,
-      "grad_norm": 1.0078125,
-      "learning_rate": 6.787980023950108e-05,
-      "loss": 0.3458,
-      "step": 3920
-    },
-    {
-      "epoch": 9.667487684729064,
-      "grad_norm": 0.953125,
-      "learning_rate": 6.760852931832328e-05,
-      "loss": 0.344,
-      "step": 3925
-    },
-    {
-      "epoch": 9.679802955665025,
-      "grad_norm": 1.0078125,
-      "learning_rate": 6.733752443933878e-05,
-      "loss": 0.3366,
-      "step": 3930
-    },
-    {
-      "epoch": 9.692118226600986,
-      "grad_norm": 0.9921875,
-      "learning_rate": 6.706678782840304e-05,
-      "loss": 0.3404,
-      "step": 3935
-    },
-    {
-      "epoch": 9.704433497536947,
-      "grad_norm": 0.96484375,
-      "learning_rate": 6.679632170916816e-05,
-      "loss": 0.3398,
-      "step": 3940
-    },
-    {
-      "epoch": 9.716748768472906,
-      "grad_norm": 0.98046875,
-      "learning_rate": 6.65261283030646e-05,
-      "loss": 0.3331,
-      "step": 3945
-    },
-    {
-      "epoch": 9.729064039408867,
-      "grad_norm": 0.98046875,
-      "learning_rate": 6.625620982928293e-05,
-      "loss": 0.3396,
-      "step": 3950
-    },
-    {
-      "epoch": 9.741379310344827,
-      "grad_norm": 0.9609375,
-      "learning_rate": 6.598656850475562e-05,
-      "loss": 0.336,
-      "step": 3955
-    },
-    {
-      "epoch": 9.753694581280788,
-      "grad_norm": 1.0078125,
-      "learning_rate": 6.571720654413877e-05,
-      "loss": 0.3416,
-      "step": 3960
-    },
-    {
-      "epoch": 9.766009852216749,
-      "grad_norm": 1.015625,
-      "learning_rate": 6.544812615979404e-05,
-      "loss": 0.3381,
-      "step": 3965
-    },
-    {
-      "epoch": 9.77832512315271,
-      "grad_norm": 1.0078125,
-      "learning_rate": 6.517932956177038e-05,
-      "loss": 0.3398,
-      "step": 3970
-    },
-    {
-      "epoch": 9.790640394088669,
-      "grad_norm": 0.9453125,
-      "learning_rate": 6.491081895778588e-05,
-      "loss": 0.3411,
-      "step": 3975
-    },
-    {
-      "epoch": 9.80295566502463,
-      "grad_norm": 1.09375,
-      "learning_rate": 6.464259655320973e-05,
-      "loss": 0.3405,
-      "step": 3980
-    },
-    {
-      "epoch": 9.81527093596059,
-      "grad_norm": 1.078125,
-      "learning_rate": 6.437466455104395e-05,
-      "loss": 0.343,
-      "step": 3985
-    },
-    {
-      "epoch": 9.827586206896552,
-      "grad_norm": 0.98828125,
-      "learning_rate": 6.410702515190543e-05,
-      "loss": 0.343,
-      "step": 3990
-    },
-    {
-      "epoch": 9.839901477832512,
-      "grad_norm": 0.98828125,
-      "learning_rate": 6.383968055400784e-05,
-      "loss": 0.3316,
-      "step": 3995
-    },
-    {
-      "epoch": 9.852216748768473,
-      "grad_norm": 0.9609375,
-      "learning_rate": 6.357263295314349e-05,
-      "loss": 0.3403,
-      "step": 4000
-    },
-    {
-      "epoch": 9.864532019704434,
-      "grad_norm": 0.9921875,
-      "learning_rate": 6.330588454266542e-05,
-      "loss": 0.3392,
-      "step": 4005
-    },
-    {
-      "epoch": 9.876847290640395,
-      "grad_norm": 1.0625,
-      "learning_rate": 6.303943751346922e-05,
-      "loss": 0.3406,
-      "step": 4010
-    },
-    {
-      "epoch": 9.889162561576354,
-      "grad_norm": 0.984375,
-      "learning_rate": 6.277329405397525e-05,
-      "loss": 0.3404,
-      "step": 4015
-    },
-    {
-      "epoch": 9.901477832512315,
-      "grad_norm": 0.96484375,
-      "learning_rate": 6.250745635011048e-05,
-      "loss": 0.3386,
-      "step": 4020
-    },
-    {
-      "epoch": 9.913793103448276,
-      "grad_norm": 0.97265625,
-      "learning_rate": 6.22419265852906e-05,
-      "loss": 0.3342,
-      "step": 4025
-    },
-    {
-      "epoch": 9.926108374384237,
-      "grad_norm": 0.99609375,
-      "learning_rate": 6.19767069404021e-05,
-      "loss": 0.3351,
-      "step": 4030
-    },
-    {
-      "epoch": 9.938423645320198,
-      "grad_norm": 0.9453125,
-      "learning_rate": 6.171179959378437e-05,
-      "loss": 0.3506,
-      "step": 4035
-    },
-    {
-      "epoch": 9.950738916256158,
-      "grad_norm": 1.0234375,
-      "learning_rate": 6.144720672121177e-05,
-      "loss": 0.3362,
-      "step": 4040
-    },
-    {
-      "epoch": 9.963054187192117,
-      "grad_norm": 1.1015625,
-      "learning_rate": 6.118293049587578e-05,
-      "loss": 0.3403,
-      "step": 4045
-    },
-    {
-      "epoch": 9.975369458128078,
-      "grad_norm": 1.078125,
-      "learning_rate": 6.0918973088367116e-05,
-      "loss": 0.3397,
-      "step": 4050
-    },
-    {
-      "epoch": 9.98768472906404,
-      "grad_norm": 1.0625,
-      "learning_rate": 6.0655336666658034e-05,
-      "loss": 0.3303,
-      "step": 4055
-    },
-    {
-      "epoch": 10.0,
-      "grad_norm": 1.0,
-      "learning_rate": 6.039202339608432e-05,
-      "loss": 0.3371,
-      "step": 4060
-    },
-    {
-      "epoch": 10.0,
-      "eval_loss": 4.856110572814941,
-      "eval_runtime": 2.0429,
-      "eval_samples_per_second": 4.895,
-      "eval_steps_per_second": 0.979,
-      "step": 4060
-    },
-    {
-      "epoch": 10.01231527093596,
-      "grad_norm": 0.93359375,
-      "learning_rate": 6.012903543932766e-05,
-      "loss": 0.2947,
-      "step": 4065
-    },
-    {
-      "epoch": 10.024630541871922,
-      "grad_norm": 0.98046875,
-      "learning_rate": 5.986637495639782e-05,
-      "loss": 0.301,
-      "step": 4070
-    },
-    {
-      "epoch": 10.036945812807883,
-      "grad_norm": 1.046875,
-      "learning_rate": 5.960404410461488e-05,
-      "loss": 0.2953,
-      "step": 4075
-    },
-    {
-      "epoch": 10.049261083743842,
-      "grad_norm": 0.890625,
-      "learning_rate": 5.934204503859158e-05,
-      "loss": 0.2988,
-      "step": 4080
-    },
-    {
-      "epoch": 10.061576354679802,
-      "grad_norm": 0.87109375,
-      "learning_rate": 5.90803799102156e-05,
-      "loss": 0.2913,
-      "step": 4085
-    },
-    {
-      "epoch": 10.073891625615763,
-      "grad_norm": 0.9609375,
-      "learning_rate": 5.881905086863181e-05,
-      "loss": 0.2937,
-      "step": 4090
-    },
-    {
-      "epoch": 10.086206896551724,
-      "grad_norm": 0.875,
-      "learning_rate": 5.8558060060224817e-05,
-      "loss": 0.2975,
-      "step": 4095
-    },
-    {
-      "epoch": 10.098522167487685,
-      "grad_norm": 0.8828125,
-      "learning_rate": 5.829740962860109e-05,
-      "loss": 0.299,
-      "step": 4100
-    },
-    {
-      "epoch": 10.110837438423646,
-      "grad_norm": 1.0234375,
-      "learning_rate": 5.803710171457145e-05,
-      "loss": 0.3064,
-      "step": 4105
-    },
-    {
-      "epoch": 10.123152709359607,
-      "grad_norm": 0.97265625,
-      "learning_rate": 5.777713845613364e-05,
-      "loss": 0.2937,
-      "step": 4110
-    },
-    {
-      "epoch": 10.135467980295566,
-      "grad_norm": 1.0234375,
-      "learning_rate": 5.751752198845444e-05,
-      "loss": 0.2982,
-      "step": 4115
-    },
-    {
-      "epoch": 10.147783251231527,
-      "grad_norm": 1.03125,
-      "learning_rate": 5.725825444385251e-05,
-      "loss": 0.3011,
-      "step": 4120
-    },
-    {
-      "epoch": 10.160098522167488,
-      "grad_norm": 0.9375,
-      "learning_rate": 5.699933795178052e-05,
-      "loss": 0.3046,
-      "step": 4125
-    },
-    {
-      "epoch": 10.172413793103448,
-      "grad_norm": 1.0078125,
-      "learning_rate": 5.6740774638807935e-05,
-      "loss": 0.2991,
-      "step": 4130
-    },
-    {
-      "epoch": 10.18472906403941,
-      "grad_norm": 0.953125,
-      "learning_rate": 5.6482566628603425e-05,
-      "loss": 0.2996,
-      "step": 4135
-    },
-    {
-      "epoch": 10.19704433497537,
-      "grad_norm": 1.140625,
-      "learning_rate": 5.622471604191746e-05,
-      "loss": 0.2981,
-      "step": 4140
-    },
-    {
-      "epoch": 10.20935960591133,
-      "grad_norm": 0.91796875,
-      "learning_rate": 5.59672249965647e-05,
-      "loss": 0.3016,
-      "step": 4145
-    },
-    {
-      "epoch": 10.22167487684729,
-      "grad_norm": 1.015625,
-      "learning_rate": 5.571009560740704e-05,
-      "loss": 0.3014,
-      "step": 4150
-    },
-    {
-      "epoch": 10.233990147783251,
-      "grad_norm": 0.94921875,
-      "learning_rate": 5.545332998633572e-05,
-      "loss": 0.296,
-      "step": 4155
-    },
-    {
-      "epoch": 10.246305418719212,
-      "grad_norm": 0.91796875,
-      "learning_rate": 5.5196930242254407e-05,
-      "loss": 0.2972,
-      "step": 4160
-    },
-    {
-      "epoch": 10.258620689655173,
-      "grad_norm": 1.0625,
-      "learning_rate": 5.494089848106156e-05,
-      "loss": 0.3073,
-      "step": 4165
-    },
-    {
-      "epoch": 10.270935960591133,
-      "grad_norm": 0.90625,
-      "learning_rate": 5.46852368056334e-05,
-      "loss": 0.3061,
-      "step": 4170
-    },
-    {
-      "epoch": 10.283251231527094,
-      "grad_norm": 0.93359375,
-      "learning_rate": 5.4429947315806376e-05,
-      "loss": 0.2959,
-      "step": 4175
-    },
-    {
-      "epoch": 10.295566502463053,
-      "grad_norm": 0.9765625,
-      "learning_rate": 5.417503210836015e-05,
-      "loss": 0.3049,
-      "step": 4180
-    },
-    {
-      "epoch": 10.307881773399014,
-      "grad_norm": 1.0,
-      "learning_rate": 5.392049327700026e-05,
-      "loss": 0.3025,
-      "step": 4185
-    },
-    {
-      "epoch": 10.320197044334975,
-      "grad_norm": 1.015625,
-      "learning_rate": 5.366633291234087e-05,
-      "loss": 0.304,
-      "step": 4190
-    },
-    {
-      "epoch": 10.332512315270936,
-      "grad_norm": 1.0546875,
-      "learning_rate": 5.341255310188775e-05,
-      "loss": 0.2992,
-      "step": 4195
-    },
-    {
-      "epoch": 10.344827586206897,
-      "grad_norm": 1.0625,
-      "learning_rate": 5.3159155930021e-05,
-      "loss": 0.3041,
-      "step": 4200
-    },
-    {
-      "epoch": 10.357142857142858,
-      "grad_norm": 0.90625,
-      "learning_rate": 5.290614347797802e-05,
-      "loss": 0.3002,
-      "step": 4205
-    },
-    {
-      "epoch": 10.369458128078819,
-      "grad_norm": 1.0234375,
-      "learning_rate": 5.265351782383629e-05,
-      "loss": 0.3046,
-      "step": 4210
-    },
-    {
-      "epoch": 10.381773399014778,
-      "grad_norm": 0.9921875,
-      "learning_rate": 5.2401281042496494e-05,
-      "loss": 0.3028,
-      "step": 4215
-    },
-    {
-      "epoch": 10.394088669950738,
-      "grad_norm": 0.8828125,
-      "learning_rate": 5.214943520566531e-05,
-      "loss": 0.3023,
-      "step": 4220
-    },
-    {
-      "epoch": 10.4064039408867,
-      "grad_norm": 1.03125,
-      "learning_rate": 5.18979823818385e-05,
-      "loss": 0.3114,
-      "step": 4225
-    },
-    {
-      "epoch": 10.41871921182266,
-      "grad_norm": 1.1171875,
-      "learning_rate": 5.164692463628378e-05,
-      "loss": 0.3008,
-      "step": 4230
-    },
-    {
-      "epoch": 10.431034482758621,
-      "grad_norm": 0.93359375,
-      "learning_rate": 5.13962640310241e-05,
-      "loss": 0.3022,
-      "step": 4235
-    },
-    {
-      "epoch": 10.443349753694582,
-      "grad_norm": 1.0078125,
-      "learning_rate": 5.1146002624820386e-05,
-      "loss": 0.308,
-      "step": 4240
-    },
-    {
-      "epoch": 10.455665024630543,
-      "grad_norm": 0.8828125,
-      "learning_rate": 5.0896142473154987e-05,
-      "loss": 0.314,
-      "step": 4245
-    },
-    {
-      "epoch": 10.467980295566502,
-      "grad_norm": 1.1796875,
-      "learning_rate": 5.064668562821444e-05,
-      "loss": 0.3085,
-      "step": 4250
-    },
-    {
-      "epoch": 10.480295566502463,
-      "grad_norm": 1.0078125,
-      "learning_rate": 5.039763413887291e-05,
-      "loss": 0.3102,
-      "step": 4255
-    },
-    {
-      "epoch": 10.492610837438423,
-      "grad_norm": 0.9609375,
-      "learning_rate": 5.014899005067524e-05,
-      "loss": 0.3058,
-      "step": 4260
-    },
-    {
-      "epoch": 10.504926108374384,
-      "grad_norm": 1.03125,
-      "learning_rate": 4.990075540582003e-05,
-      "loss": 0.3001,
-      "step": 4265
-    },
-    {
-      "epoch": 10.517241379310345,
-      "grad_norm": 1.1171875,
-      "learning_rate": 4.9652932243143146e-05,
-      "loss": 0.3018,
-      "step": 4270
-    },
-    {
-      "epoch": 10.529556650246306,
-      "grad_norm": 1.0234375,
-      "learning_rate": 4.940552259810063e-05,
-      "loss": 0.312,
-      "step": 4275
-    },
-    {
-      "epoch": 10.541871921182267,
-      "grad_norm": 0.87890625,
-      "learning_rate": 4.915852850275233e-05,
-      "loss": 0.3086,
-      "step": 4280
-    },
-    {
-      "epoch": 10.554187192118226,
-      "grad_norm": 1.03125,
-      "learning_rate": 4.891195198574491e-05,
-      "loss": 0.3002,
-      "step": 4285
-    },
-    {
-      "epoch": 10.566502463054187,
-      "grad_norm": 1.0234375,
-      "learning_rate": 4.866579507229545e-05,
-      "loss": 0.307,
-      "step": 4290
-    },
-    {
-      "epoch": 10.578817733990148,
-      "grad_norm": 0.9609375,
-      "learning_rate": 4.8420059784174485e-05,
-      "loss": 0.3018,
-      "step": 4295
-    },
-    {
-      "epoch": 10.591133004926109,
-      "grad_norm": 0.92578125,
-      "learning_rate": 4.8174748139689905e-05,
-      "loss": 0.3035,
-      "step": 4300
-    },
-    {
-      "epoch": 10.60344827586207,
-      "grad_norm": 0.9765625,
-      "learning_rate": 4.792986215366976e-05,
-      "loss": 0.3052,
-      "step": 4305
-    },
-    {
-      "epoch": 10.61576354679803,
-      "grad_norm": 0.90625,
-      "learning_rate": 4.768540383744622e-05,
-      "loss": 0.3069,
-      "step": 4310
-    },
-    {
-      "epoch": 10.62807881773399,
-      "grad_norm": 0.95703125,
-      "learning_rate": 4.744137519883872e-05,
-      "loss": 0.3031,
-      "step": 4315
-    },
-    {
-      "epoch": 10.64039408866995,
-      "grad_norm": 0.9609375,
-      "learning_rate": 4.7197778242137755e-05,
-      "loss": 0.3124,
-      "step": 4320
-    },
-    {
-      "epoch": 10.652709359605911,
-      "grad_norm": 0.91015625,
-      "learning_rate": 4.6954614968088115e-05,
-      "loss": 0.3012,
-      "step": 4325
-    },
-    {
-      "epoch": 10.665024630541872,
-      "grad_norm": 0.95703125,
-      "learning_rate": 4.6711887373872754e-05,
-      "loss": 0.2978,
-      "step": 4330
-    },
-    {
-      "epoch": 10.677339901477833,
-      "grad_norm": 0.92578125,
-      "learning_rate": 4.646959745309609e-05,
-      "loss": 0.3075,
-      "step": 4335
-    },
-    {
-      "epoch": 10.689655172413794,
-      "grad_norm": 1.0390625,
-      "learning_rate": 4.62277471957679e-05,
-      "loss": 0.3036,
-      "step": 4340
-    },
-    {
-      "epoch": 10.701970443349754,
-      "grad_norm": 0.890625,
-      "learning_rate": 4.598633858828681e-05,
-      "loss": 0.3051,
-      "step": 4345
-    },
-    {
-      "epoch": 10.714285714285714,
-      "grad_norm": 1.09375,
-      "learning_rate": 4.574537361342407e-05,
-      "loss": 0.3028,
-      "step": 4350
-    },
-    {
-      "epoch": 10.726600985221674,
-      "grad_norm": 0.9921875,
-      "learning_rate": 4.5504854250307085e-05,
-      "loss": 0.3109,
-      "step": 4355
-    },
-    {
-      "epoch": 10.738916256157635,
-      "grad_norm": 0.91796875,
-      "learning_rate": 4.526478247440349e-05,
-      "loss": 0.2968,
-      "step": 4360
-    },
-    {
-      "epoch": 10.751231527093596,
-      "grad_norm": 0.953125,
-      "learning_rate": 4.502516025750455e-05,
-      "loss": 0.3105,
-      "step": 4365
-    },
-    {
-      "epoch": 10.763546798029557,
-      "grad_norm": 1.0546875,
-      "learning_rate": 4.4785989567709316e-05,
-      "loss": 0.3039,
-      "step": 4370
-    },
-    {
-      "epoch": 10.775862068965518,
-      "grad_norm": 0.9609375,
-      "learning_rate": 4.454727236940814e-05,
-      "loss": 0.3047,
-      "step": 4375
-    },
-    {
-      "epoch": 10.788177339901479,
-      "grad_norm": 0.91015625,
-      "learning_rate": 4.430901062326681e-05,
-      "loss": 0.3062,
-      "step": 4380
-    },
-    {
-      "epoch": 10.800492610837438,
-      "grad_norm": 1.0078125,
-      "learning_rate": 4.407120628621032e-05,
-      "loss": 0.3066,
-      "step": 4385
-    },
-    {
-      "epoch": 10.812807881773399,
-      "grad_norm": 0.90234375,
-      "learning_rate": 4.3833861311406697e-05,
-      "loss": 0.3024,
-      "step": 4390
-    },
-    {
-      "epoch": 10.82512315270936,
-      "grad_norm": 0.9140625,
-      "learning_rate": 4.359697764825123e-05,
-      "loss": 0.3068,
-      "step": 4395
-    },
-    {
-      "epoch": 10.83743842364532,
-      "grad_norm": 0.9765625,
-      "learning_rate": 4.336055724235013e-05,
-      "loss": 0.3091,
-      "step": 4400
-    },
-    {
-      "epoch": 10.849753694581281,
-      "grad_norm": 0.88671875,
-      "learning_rate": 4.312460203550489e-05,
-      "loss": 0.3067,
-      "step": 4405
-    },
-    {
-      "epoch": 10.862068965517242,
-      "grad_norm": 1.1015625,
-      "learning_rate": 4.288911396569599e-05,
-      "loss": 0.3093,
-      "step": 4410
-    },
-    {
-      "epoch": 10.874384236453203,
-      "grad_norm": 0.96875,
-      "learning_rate": 4.265409496706733e-05,
-      "loss": 0.3003,
-      "step": 4415
-    },
-    {
-      "epoch": 10.886699507389162,
-      "grad_norm": 0.93359375,
-      "learning_rate": 4.241954696990995e-05,
-      "loss": 0.3017,
-      "step": 4420
-    },
-    {
-      "epoch": 10.899014778325123,
-      "grad_norm": 1.015625,
-      "learning_rate": 4.21854719006467e-05,
-      "loss": 0.3075,
-      "step": 4425
-    },
-    {
-      "epoch": 10.911330049261084,
-      "grad_norm": 0.90234375,
-      "learning_rate": 4.1951871681815804e-05,
-      "loss": 0.3098,
-      "step": 4430
-    },
-    {
-      "epoch": 10.923645320197044,
-      "grad_norm": 0.9296875,
-      "learning_rate": 4.1718748232055595e-05,
-      "loss": 0.3021,
-      "step": 4435
-    },
-    {
-      "epoch": 10.935960591133005,
-      "grad_norm": 0.984375,
-      "learning_rate": 4.148610346608837e-05,
-      "loss": 0.3097,
-      "step": 4440
-    },
-    {
-      "epoch": 10.948275862068966,
-      "grad_norm": 1.0078125,
-      "learning_rate": 4.1253939294705004e-05,
-      "loss": 0.3028,
-      "step": 4445
-    },
-    {
-      "epoch": 10.960591133004925,
-      "grad_norm": 0.9453125,
-      "learning_rate": 4.1022257624748914e-05,
-      "loss": 0.3118,
-      "step": 4450
-    },
-    {
-      "epoch": 10.972906403940886,
-      "grad_norm": 0.9609375,
-      "learning_rate": 4.079106035910073e-05,
-      "loss": 0.3038,
-      "step": 4455
-    },
-    {
-      "epoch": 10.985221674876847,
-      "grad_norm": 0.99609375,
-      "learning_rate": 4.056034939666236e-05,
-      "loss": 0.297,
-      "step": 4460
-    },
-    {
-      "epoch": 10.997536945812808,
-      "grad_norm": 0.98046875,
-      "learning_rate": 4.0330126632341625e-05,
-      "loss": 0.2993,
-      "step": 4465
-    },
-    {
-      "epoch": 11.0,
-      "eval_loss": 5.202045440673828,
-      "eval_runtime": 2.0439,
-      "eval_samples_per_second": 4.893,
-      "eval_steps_per_second": 0.979,
-      "step": 4466
-    },
-    {
-      "epoch": 11.009852216748769,
-      "grad_norm": 0.76953125,
-      "learning_rate": 4.010039395703664e-05,
-      "loss": 0.2871,
-      "step": 4470
-    },
-    {
-      "epoch": 11.02216748768473,
-      "grad_norm": 0.921875,
-      "learning_rate": 3.987115325762012e-05,
-      "loss": 0.2841,
-      "step": 4475
-    },
-    {
-      "epoch": 11.03448275862069,
-      "grad_norm": 0.921875,
-      "learning_rate": 3.964240641692416e-05,
-      "loss": 0.2807,
-      "step": 4480
-    },
-    {
-      "epoch": 11.04679802955665,
-      "grad_norm": 0.94921875,
-      "learning_rate": 3.94141553137245e-05,
-      "loss": 0.2826,
-      "step": 4485
-    },
-    {
-      "epoch": 11.05911330049261,
-      "grad_norm": 0.890625,
-      "learning_rate": 3.918640182272535e-05,
-      "loss": 0.2809,
-      "step": 4490
-    },
-    {
-      "epoch": 11.071428571428571,
-      "grad_norm": 0.91796875,
-      "learning_rate": 3.89591478145437e-05,
-      "loss": 0.2852,
-      "step": 4495
-    },
-    {
-      "epoch": 11.083743842364532,
-      "grad_norm": 0.89453125,
-      "learning_rate": 3.873239515569429e-05,
-      "loss": 0.2906,
-      "step": 4500
-    },
-    {
-      "epoch": 11.096059113300493,
-      "grad_norm": 0.890625,
-      "learning_rate": 3.8506145708573916e-05,
-      "loss": 0.2764,
-      "step": 4505
-    },
-    {
-      "epoch": 11.108374384236454,
-      "grad_norm": 1.0078125,
-      "learning_rate": 3.828040133144657e-05,
-      "loss": 0.2854,
-      "step": 4510
-    },
-    {
-      "epoch": 11.120689655172415,
-      "grad_norm": 0.9296875,
-      "learning_rate": 3.80551638784277e-05,
-      "loss": 0.2767,
-      "step": 4515
-    },
-    {
-      "epoch": 11.133004926108374,
-      "grad_norm": 1.2734375,
-      "learning_rate": 3.783043519946936e-05,
-      "loss": 0.2836,
-      "step": 4520
-    },
-    {
-      "epoch": 11.145320197044335,
-      "grad_norm": 0.94921875,
-      "learning_rate": 3.760621714034476e-05,
-      "loss": 0.2811,
-      "step": 4525
-    },
-    {
-      "epoch": 11.157635467980295,
-      "grad_norm": 0.86328125,
-      "learning_rate": 3.738251154263333e-05,
-      "loss": 0.2803,
-      "step": 4530
-    },
-    {
-      "epoch": 11.169950738916256,
-      "grad_norm": 0.8671875,
-      "learning_rate": 3.7159320243705355e-05,
-      "loss": 0.2818,
-      "step": 4535
-    },
-    {
-      "epoch": 11.182266009852217,
-      "grad_norm": 0.94140625,
-      "learning_rate": 3.6936645076707146e-05,
-      "loss": 0.2864,
-      "step": 4540
-    },
-    {
-      "epoch": 11.194581280788178,
-      "grad_norm": 0.94140625,
-      "learning_rate": 3.671448787054571e-05,
-      "loss": 0.2815,
-      "step": 4545
-    },
-    {
-      "epoch": 11.206896551724139,
-      "grad_norm": 0.8984375,
-      "learning_rate": 3.649285044987397e-05,
-      "loss": 0.2779,
-      "step": 4550
-    },
-    {
-      "epoch": 11.219211822660098,
-      "grad_norm": 0.91015625,
-      "learning_rate": 3.627173463507565e-05,
-      "loss": 0.2857,
-      "step": 4555
-    },
-    {
-      "epoch": 11.231527093596059,
-      "grad_norm": 0.99609375,
-      "learning_rate": 3.605114224225028e-05,
-      "loss": 0.2811,
-      "step": 4560
-    },
-    {
-      "epoch": 11.24384236453202,
-      "grad_norm": 0.9375,
-      "learning_rate": 3.5831075083198464e-05,
-      "loss": 0.2804,
-      "step": 4565
-    },
-    {
-      "epoch": 11.25615763546798,
-      "grad_norm": 0.9609375,
-      "learning_rate": 3.561153496540673e-05,
-      "loss": 0.2825,
-      "step": 4570
-    },
-    {
-      "epoch": 11.268472906403941,
-      "grad_norm": 0.9296875,
-      "learning_rate": 3.5392523692033006e-05,
-      "loss": 0.2878,
-      "step": 4575
-    },
-    {
-      "epoch": 11.280788177339902,
-      "grad_norm": 0.9375,
-      "learning_rate": 3.51740430618915e-05,
-      "loss": 0.2851,
-      "step": 4580
-    },
-    {
-      "epoch": 11.293103448275861,
-      "grad_norm": 0.8828125,
-      "learning_rate": 3.495609486943814e-05,
-      "loss": 0.2805,
-      "step": 4585
-    },
-    {
-      "epoch": 11.305418719211822,
-      "grad_norm": 0.921875,
-      "learning_rate": 3.473868090475574e-05,
-      "loss": 0.2836,
-      "step": 4590
-    },
-    {
-      "epoch": 11.317733990147783,
-      "grad_norm": 0.96484375,
-      "learning_rate": 3.4521802953539376e-05,
-      "loss": 0.2952,
-      "step": 4595
-    },
-    {
-      "epoch": 11.330049261083744,
-      "grad_norm": 0.98828125,
-      "learning_rate": 3.4305462797081525e-05,
-      "loss": 0.288,
-      "step": 4600
-    },
-    {
-      "epoch": 11.342364532019705,
-      "grad_norm": 0.9609375,
-      "learning_rate": 3.408966221225773e-05,
-      "loss": 0.2773,
-      "step": 4605
-    },
-    {
-      "epoch": 11.354679802955665,
-      "grad_norm": 0.89453125,
-      "learning_rate": 3.387440297151172e-05,
-      "loss": 0.2867,
-      "step": 4610
-    },
-    {
-      "epoch": 11.366995073891626,
-      "grad_norm": 0.91796875,
-      "learning_rate": 3.36596868428411e-05,
-      "loss": 0.2809,
-      "step": 4615
-    },
-    {
-      "epoch": 11.379310344827585,
-      "grad_norm": 0.96875,
-      "learning_rate": 3.3445515589782574e-05,
-      "loss": 0.2847,
-      "step": 4620
-    },
-    {
-      "epoch": 11.391625615763546,
-      "grad_norm": 0.99609375,
-      "learning_rate": 3.3231890971397694e-05,
-      "loss": 0.2794,
-      "step": 4625
-    },
-    {
-      "epoch": 11.403940886699507,
-      "grad_norm": 0.953125,
-      "learning_rate": 3.301881474225831e-05,
-      "loss": 0.283,
-      "step": 4630
-    },
-    {
-      "epoch": 11.416256157635468,
-      "grad_norm": 0.91796875,
-      "learning_rate": 3.2806288652432174e-05,
-      "loss": 0.2808,
-      "step": 4635
-    },
-    {
-      "epoch": 11.428571428571429,
-      "grad_norm": 0.953125,
-      "learning_rate": 3.259431444746846e-05,
-      "loss": 0.2821,
-      "step": 4640
-    },
-    {
-      "epoch": 11.44088669950739,
-      "grad_norm": 0.91015625,
-      "learning_rate": 3.238289386838364e-05,
-      "loss": 0.2811,
-      "step": 4645
-    },
-    {
-      "epoch": 11.45320197044335,
-      "grad_norm": 0.9375,
-      "learning_rate": 3.217202865164697e-05,
-      "loss": 0.2841,
-      "step": 4650
-    },
-    {
-      "epoch": 11.46551724137931,
-      "grad_norm": 0.91015625,
-      "learning_rate": 3.1961720529166436e-05,
-      "loss": 0.2791,
-      "step": 4655
-    },
-    {
-      "epoch": 11.47783251231527,
-      "grad_norm": 1.0,
-      "learning_rate": 3.17519712282743e-05,
-      "loss": 0.2847,
-      "step": 4660
-    },
-    {
-      "epoch": 11.490147783251231,
-      "grad_norm": 0.8671875,
-      "learning_rate": 3.154278247171314e-05,
-      "loss": 0.2852,
-      "step": 4665
-    },
-    {
-      "epoch": 11.502463054187192,
-      "grad_norm": 1.0,
-      "learning_rate": 3.133415597762148e-05,
-      "loss": 0.2871,
-      "step": 4670
-    },
-    {
-      "epoch": 11.514778325123153,
-      "grad_norm": 0.8515625,
-      "learning_rate": 3.112609345951989e-05,
-      "loss": 0.2828,
-      "step": 4675
-    },
-    {
-      "epoch": 11.527093596059114,
-      "grad_norm": 0.90234375,
-      "learning_rate": 3.09185966262968e-05,
-      "loss": 0.2815,
-      "step": 4680
-    },
-    {
-      "epoch": 11.539408866995075,
-      "grad_norm": 0.9140625,
-      "learning_rate": 3.071166718219439e-05,
-      "loss": 0.2875,
-      "step": 4685
-    },
-    {
-      "epoch": 11.551724137931034,
-      "grad_norm": 0.96484375,
-      "learning_rate": 3.05053068267948e-05,
-      "loss": 0.2818,
-      "step": 4690
-    },
-    {
-      "epoch": 11.564039408866995,
-      "grad_norm": 0.93359375,
-      "learning_rate": 3.0299517255005937e-05,
-      "loss": 0.2872,
-      "step": 4695
-    },
-    {
-      "epoch": 11.576354679802956,
-      "grad_norm": 0.91796875,
-      "learning_rate": 3.0094300157047793e-05,
-      "loss": 0.2768,
-      "step": 4700
-    },
-    {
-      "epoch": 11.588669950738916,
-      "grad_norm": 0.9140625,
-      "learning_rate": 2.9889657218438283e-05,
-      "loss": 0.28,
-      "step": 4705
-    },
-    {
-      "epoch": 11.600985221674877,
-      "grad_norm": 0.96875,
-      "learning_rate": 2.9685590119979688e-05,
-      "loss": 0.2793,
-      "step": 4710
-    },
-    {
-      "epoch": 11.613300492610838,
-      "grad_norm": 0.8984375,
-      "learning_rate": 2.948210053774465e-05,
-      "loss": 0.284,
-      "step": 4715
-    },
-    {
-      "epoch": 11.625615763546797,
-      "grad_norm": 0.93359375,
-      "learning_rate": 2.9279190143062552e-05,
-      "loss": 0.2934,
-      "step": 4720
-    },
-    {
-      "epoch": 11.637931034482758,
-      "grad_norm": 0.921875,
-      "learning_rate": 2.9076860602505564e-05,
-      "loss": 0.2825,
-      "step": 4725
-    },
-    {
-      "epoch": 11.650246305418719,
-      "grad_norm": 0.94921875,
-      "learning_rate": 2.8875113577875258e-05,
-      "loss": 0.2836,
-      "step": 4730
-    },
-    {
-      "epoch": 11.66256157635468,
-      "grad_norm": 0.9765625,
-      "learning_rate": 2.867395072618868e-05,
-      "loss": 0.29,
-      "step": 4735
-    },
-    {
-      "epoch": 11.67487684729064,
-      "grad_norm": 0.9296875,
-      "learning_rate": 2.8473373699664997e-05,
-      "loss": 0.2824,
-      "step": 4740
-    },
-    {
-      "epoch": 11.687192118226601,
-      "grad_norm": 0.9765625,
-      "learning_rate": 2.8273384145711624e-05,
-      "loss": 0.2794,
-      "step": 4745
-    },
-    {
-      "epoch": 11.699507389162562,
-      "grad_norm": 0.94921875,
-      "learning_rate": 2.8073983706911024e-05,
-      "loss": 0.285,
-      "step": 4750
-    },
-    {
-      "epoch": 11.711822660098521,
-      "grad_norm": 1.0546875,
-      "learning_rate": 2.7875174021007e-05,
-      "loss": 0.2843,
-      "step": 4755
-    },
-    {
-      "epoch": 11.724137931034482,
-      "grad_norm": 0.94921875,
-      "learning_rate": 2.7676956720891235e-05,
-      "loss": 0.2873,
-      "step": 4760
-    },
-    {
-      "epoch": 11.736453201970443,
-      "grad_norm": 0.921875,
-      "learning_rate": 2.747933343459007e-05,
-      "loss": 0.2841,
-      "step": 4765
-    },
-    {
-      "epoch": 11.748768472906404,
-      "grad_norm": 0.875,
-      "learning_rate": 2.728230578525086e-05,
-      "loss": 0.2801,
-      "step": 4770
-    },
-    {
-      "epoch": 11.761083743842365,
-      "grad_norm": 0.9375,
-      "learning_rate": 2.7085875391128955e-05,
-      "loss": 0.2792,
-      "step": 4775
-    },
-    {
-      "epoch": 11.773399014778326,
-      "grad_norm": 1.1015625,
-      "learning_rate": 2.6890043865574078e-05,
-      "loss": 0.2856,
-      "step": 4780
-    },
-    {
-      "epoch": 11.785714285714286,
-      "grad_norm": 1.0625,
-      "learning_rate": 2.669481281701739e-05,
-      "loss": 0.2899,
-      "step": 4785
-    },
-    {
-      "epoch": 11.798029556650246,
-      "grad_norm": 0.9453125,
-      "learning_rate": 2.6500183848957983e-05,
-      "loss": 0.2834,
-      "step": 4790
-    },
-    {
-      "epoch": 11.810344827586206,
-      "grad_norm": 0.8828125,
-      "learning_rate": 2.6306158559950023e-05,
-      "loss": 0.2855,
-      "step": 4795
-    },
-    {
-      "epoch": 11.822660098522167,
-      "grad_norm": 0.875,
-      "learning_rate": 2.6112738543589312e-05,
-      "loss": 0.2883,
-      "step": 4800
-    },
-    {
-      "epoch": 11.834975369458128,
-      "grad_norm": 0.93359375,
-      "learning_rate": 2.591992538850042e-05,
-      "loss": 0.2853,
-      "step": 4805
-    },
-    {
-      "epoch": 11.847290640394089,
-      "grad_norm": 0.890625,
-      "learning_rate": 2.572772067832351e-05,
-      "loss": 0.2817,
-      "step": 4810
-    },
-    {
-      "epoch": 11.85960591133005,
-      "grad_norm": 1.0234375,
-      "learning_rate": 2.553612599170143e-05,
-      "loss": 0.2847,
-      "step": 4815
-    },
-    {
-      "epoch": 11.87192118226601,
-      "grad_norm": 0.91796875,
-      "learning_rate": 2.5345142902266628e-05,
-      "loss": 0.2841,
-      "step": 4820
-    },
-    {
-      "epoch": 11.88423645320197,
-      "grad_norm": 0.88671875,
-      "learning_rate": 2.5154772978628405e-05,
-      "loss": 0.2884,
-      "step": 4825
-    },
-    {
-      "epoch": 11.89655172413793,
-      "grad_norm": 0.97265625,
-      "learning_rate": 2.496501778435977e-05,
-      "loss": 0.2876,
-      "step": 4830
-    },
-    {
-      "epoch": 11.908866995073891,
-      "grad_norm": 0.9609375,
-      "learning_rate": 2.477587887798488e-05,
-      "loss": 0.2867,
-      "step": 4835
-    },
-    {
-      "epoch": 11.921182266009852,
-      "grad_norm": 0.84375,
-      "learning_rate": 2.4587357812966095e-05,
-      "loss": 0.2807,
-      "step": 4840
-    },
-    {
-      "epoch": 11.933497536945813,
-      "grad_norm": 1.0390625,
-      "learning_rate": 2.4399456137691147e-05,
-      "loss": 0.2868,
-      "step": 4845
-    },
-    {
-      "epoch": 11.945812807881774,
-      "grad_norm": 0.984375,
-      "learning_rate": 2.42121753954606e-05,
-      "loss": 0.2953,
-      "step": 4850
-    },
-    {
-      "epoch": 11.958128078817733,
-      "grad_norm": 0.93359375,
-      "learning_rate": 2.4025517124475017e-05,
-      "loss": 0.2788,
-      "step": 4855
-    },
-    {
-      "epoch": 11.970443349753694,
-      "grad_norm": 0.91015625,
-      "learning_rate": 2.3839482857822458e-05,
-      "loss": 0.2893,
-      "step": 4860
-    },
-    {
-      "epoch": 11.982758620689655,
-      "grad_norm": 1.09375,
-      "learning_rate": 2.3654074123465752e-05,
-      "loss": 0.2919,
-      "step": 4865
-    },
-    {
-      "epoch": 11.995073891625616,
-      "grad_norm": 0.96875,
-      "learning_rate": 2.3469292444230096e-05,
-      "loss": 0.29,
-      "step": 4870
-    },
-    {
-      "epoch": 12.0,
-      "eval_loss": 5.407031536102295,
-      "eval_runtime": 2.0437,
-      "eval_samples_per_second": 4.893,
-      "eval_steps_per_second": 0.979,
-      "step": 4872
-    },
-    {
-      "epoch": 12.007389162561577,
-      "grad_norm": 0.8515625,
-      "learning_rate": 2.328513933779034e-05,
-      "loss": 0.2855,
-      "step": 4875
-    },
-    {
-      "epoch": 12.019704433497537,
-      "grad_norm": 0.89453125,
-      "learning_rate": 2.310161631665886e-05,
-      "loss": 0.2683,
-      "step": 4880
-    },
-    {
-      "epoch": 12.032019704433498,
-      "grad_norm": 0.8359375,
-      "learning_rate": 2.2918724888172714e-05,
-      "loss": 0.2686,
-      "step": 4885
-    },
-    {
-      "epoch": 12.044334975369457,
-      "grad_norm": 0.91796875,
-      "learning_rate": 2.2736466554481617e-05,
-      "loss": 0.278,
-      "step": 4890
-    },
-    {
-      "epoch": 12.056650246305418,
-      "grad_norm": 0.94921875,
-      "learning_rate": 2.255484281253537e-05,
-      "loss": 0.2731,
-      "step": 4895
-    },
-    {
-      "epoch": 12.068965517241379,
-      "grad_norm": 0.953125,
-      "learning_rate": 2.2373855154071732e-05,
-      "loss": 0.2786,
-      "step": 4900
-    },
-    {
-      "epoch": 12.08128078817734,
-      "grad_norm": 0.89453125,
-      "learning_rate": 2.2193505065604014e-05,
-      "loss": 0.2711,
-      "step": 4905
-    },
-    {
-      "epoch": 12.0935960591133,
-      "grad_norm": 0.8828125,
-      "learning_rate": 2.201379402840903e-05,
-      "loss": 0.2823,
-      "step": 4910
-    },
-    {
-      "epoch": 12.105911330049262,
-      "grad_norm": 0.8828125,
-      "learning_rate": 2.183472351851472e-05,
-      "loss": 0.2705,
-      "step": 4915
-    },
-    {
-      "epoch": 12.118226600985222,
-      "grad_norm": 0.8515625,
-      "learning_rate": 2.1656295006688353e-05,
-      "loss": 0.2721,
-      "step": 4920
-    },
-    {
-      "epoch": 12.130541871921181,
-      "grad_norm": 0.87109375,
-      "learning_rate": 2.1478509958424064e-05,
-      "loss": 0.2736,
-      "step": 4925
-    },
-    {
-      "epoch": 12.142857142857142,
-      "grad_norm": 0.9140625,
-      "learning_rate": 2.1301369833931117e-05,
-      "loss": 0.276,
-      "step": 4930
-    },
-    {
-      "epoch": 12.155172413793103,
-      "grad_norm": 0.9140625,
-      "learning_rate": 2.1124876088121692e-05,
-      "loss": 0.2752,
-      "step": 4935
-    },
-    {
-      "epoch": 12.167487684729064,
-      "grad_norm": 0.91015625,
-      "learning_rate": 2.0949030170599182e-05,
-      "loss": 0.2696,
-      "step": 4940
-    },
-    {
-      "epoch": 12.179802955665025,
-      "grad_norm": 0.8984375,
-      "learning_rate": 2.0773833525645992e-05,
-      "loss": 0.2715,
-      "step": 4945
-    },
-    {
-      "epoch": 12.192118226600986,
-      "grad_norm": 0.8671875,
-      "learning_rate": 2.0599287592211968e-05,
-      "loss": 0.2779,
-      "step": 4950
-    },
-    {
-      "epoch": 12.204433497536947,
-      "grad_norm": 0.9609375,
-      "learning_rate": 2.0425393803902314e-05,
-      "loss": 0.2681,
-      "step": 4955
-    },
-    {
-      "epoch": 12.216748768472906,
-      "grad_norm": 0.97265625,
-      "learning_rate": 2.0252153588966037e-05,
-      "loss": 0.2812,
-      "step": 4960
-    },
-    {
-      "epoch": 12.229064039408867,
-      "grad_norm": 1.03125,
-      "learning_rate": 2.0079568370284128e-05,
-      "loss": 0.2777,
-      "step": 4965
-    },
-    {
-      "epoch": 12.241379310344827,
-      "grad_norm": 0.9453125,
-      "learning_rate": 1.990763956535777e-05,
-      "loss": 0.2717,
-      "step": 4970
-    },
-    {
-      "epoch": 12.253694581280788,
-      "grad_norm": 0.93359375,
-      "learning_rate": 1.9736368586296916e-05,
-      "loss": 0.2782,
-      "step": 4975
-    },
-    {
-      "epoch": 12.266009852216749,
-      "grad_norm": 0.8984375,
-      "learning_rate": 1.956575683980846e-05,
-      "loss": 0.2732,
-      "step": 4980
-    },
-    {
-      "epoch": 12.27832512315271,
-      "grad_norm": 0.875,
-      "learning_rate": 1.9395805727184912e-05,
-      "loss": 0.2661,
-      "step": 4985
-    },
-    {
-      "epoch": 12.290640394088669,
-      "grad_norm": 0.96484375,
-      "learning_rate": 1.9226516644292647e-05,
-      "loss": 0.2757,
-      "step": 4990
-    },
-    {
-      "epoch": 12.30295566502463,
-      "grad_norm": 0.8828125,
-      "learning_rate": 1.9057890981560677e-05,
-      "loss": 0.2753,
-      "step": 4995
-    },
-    {
-      "epoch": 12.31527093596059,
-      "grad_norm": 0.9140625,
-      "learning_rate": 1.888993012396899e-05,
-      "loss": 0.2701,
-      "step": 5000
-    },
-    {
-      "epoch": 12.327586206896552,
-      "grad_norm": 0.921875,
-      "learning_rate": 1.8722635451037497e-05,
-      "loss": 0.2794,
-      "step": 5005
-    },
-    {
-      "epoch": 12.339901477832512,
-      "grad_norm": 0.9765625,
-      "learning_rate": 1.85560083368143e-05,
-      "loss": 0.2824,
-      "step": 5010
-    },
-    {
-      "epoch": 12.352216748768473,
-      "grad_norm": 0.98046875,
-      "learning_rate": 1.8390050149864745e-05,
-      "loss": 0.275,
-      "step": 5015
-    },
-    {
-      "epoch": 12.364532019704434,
-      "grad_norm": 0.95703125,
-      "learning_rate": 1.8224762253259976e-05,
-      "loss": 0.2787,
-      "step": 5020
-    },
-    {
-      "epoch": 12.376847290640393,
-      "grad_norm": 0.95703125,
-      "learning_rate": 1.806014600456588e-05,
-      "loss": 0.2712,
-      "step": 5025
-    },
-    {
-      "epoch": 12.389162561576354,
-      "grad_norm": 0.88671875,
-      "learning_rate": 1.7896202755831804e-05,
-      "loss": 0.2765,
-      "step": 5030
-    },
-    {
-      "epoch": 12.401477832512315,
-      "grad_norm": 0.921875,
-      "learning_rate": 1.773293385357959e-05,
-      "loss": 0.275,
-      "step": 5035
-    },
-    {
-      "epoch": 12.413793103448276,
-      "grad_norm": 0.8828125,
-      "learning_rate": 1.757034063879235e-05,
-      "loss": 0.2676,
-      "step": 5040
-    },
-    {
-      "epoch": 12.426108374384237,
-      "grad_norm": 0.9375,
-      "learning_rate": 1.7408424446903626e-05,
-      "loss": 0.2728,
-      "step": 5045
-    },
-    {
-      "epoch": 12.438423645320198,
-      "grad_norm": 0.9453125,
-      "learning_rate": 1.7247186607786338e-05,
-      "loss": 0.2792,
-      "step": 5050
-    },
-    {
-      "epoch": 12.450738916256158,
-      "grad_norm": 0.88671875,
-      "learning_rate": 1.708662844574178e-05,
-      "loss": 0.2757,
-      "step": 5055
-    },
-    {
-      "epoch": 12.463054187192117,
-      "grad_norm": 0.9453125,
-      "learning_rate": 1.692675127948894e-05,
-      "loss": 0.276,
-      "step": 5060
-    },
-    {
-      "epoch": 12.475369458128078,
-      "grad_norm": 0.9140625,
-      "learning_rate": 1.676755642215343e-05,
-      "loss": 0.2757,
-      "step": 5065
-    },
-    {
-      "epoch": 12.48768472906404,
-      "grad_norm": 0.91015625,
-      "learning_rate": 1.6609045181256976e-05,
-      "loss": 0.2722,
-      "step": 5070
-    },
-    {
-      "epoch": 12.5,
-      "grad_norm": 0.89453125,
-      "learning_rate": 1.6451218858706374e-05,
-      "loss": 0.276,
-      "step": 5075
-    },
-    {
-      "epoch": 12.51231527093596,
-      "grad_norm": 0.9609375,
-      "learning_rate": 1.629407875078305e-05,
-      "loss": 0.2782,
-      "step": 5080
-    },
-    {
-      "epoch": 12.524630541871922,
-      "grad_norm": 0.8828125,
-      "learning_rate": 1.61376261481323e-05,
-      "loss": 0.2738,
-      "step": 5085
-    },
-    {
-      "epoch": 12.536945812807883,
-      "grad_norm": 0.88671875,
-      "learning_rate": 1.5981862335752716e-05,
-      "loss": 0.2779,
-      "step": 5090
-    },
-    {
-      "epoch": 12.549261083743842,
-      "grad_norm": 0.8671875,
-      "learning_rate": 1.5826788592985553e-05,
-      "loss": 0.2686,
-      "step": 5095
-    },
-    {
-      "epoch": 12.561576354679802,
-      "grad_norm": 0.9296875,
-      "learning_rate": 1.5672406193504384e-05,
-      "loss": 0.2704,
-      "step": 5100
-    },
-    {
-      "epoch": 12.573891625615763,
-      "grad_norm": 0.84375,
-      "learning_rate": 1.5518716405304447e-05,
-      "loss": 0.2715,
-      "step": 5105
-    },
-    {
-      "epoch": 12.586206896551724,
-      "grad_norm": 0.90625,
-      "learning_rate": 1.5365720490692426e-05,
-      "loss": 0.2803,
-      "step": 5110
-    },
-    {
-      "epoch": 12.598522167487685,
-      "grad_norm": 0.93359375,
-      "learning_rate": 1.5213419706275878e-05,
-      "loss": 0.2766,
-      "step": 5115
-    },
-    {
-      "epoch": 12.610837438423646,
-      "grad_norm": 0.91796875,
-      "learning_rate": 1.5061815302953141e-05,
-      "loss": 0.2725,
-      "step": 5120
-    },
-    {
-      "epoch": 12.623152709359605,
-      "grad_norm": 0.97265625,
-      "learning_rate": 1.4910908525902811e-05,
-      "loss": 0.2694,
-      "step": 5125
-    },
-    {
-      "epoch": 12.635467980295566,
-      "grad_norm": 1.109375,
-      "learning_rate": 1.4760700614573731e-05,
-      "loss": 0.2727,
-      "step": 5130
-    },
-    {
-      "epoch": 12.647783251231527,
-      "grad_norm": 0.89453125,
-      "learning_rate": 1.461119280267471e-05,
-      "loss": 0.2772,
-      "step": 5135
-    },
-    {
-      "epoch": 12.660098522167488,
-      "grad_norm": 0.92578125,
-      "learning_rate": 1.4462386318164356e-05,
-      "loss": 0.2735,
-      "step": 5140
-    },
-    {
-      "epoch": 12.672413793103448,
-      "grad_norm": 0.95703125,
-      "learning_rate": 1.4314282383241096e-05,
-      "loss": 0.2735,
-      "step": 5145
-    },
-    {
-      "epoch": 12.68472906403941,
-      "grad_norm": 0.890625,
-      "learning_rate": 1.4166882214332999e-05,
-      "loss": 0.2767,
-      "step": 5150
-    },
-    {
-      "epoch": 12.69704433497537,
-      "grad_norm": 1.046875,
-      "learning_rate": 1.4020187022087971e-05,
-      "loss": 0.2824,
-      "step": 5155
-    },
-    {
-      "epoch": 12.709359605911331,
-      "grad_norm": 1.4375,
-      "learning_rate": 1.3874198011363582e-05,
-      "loss": 0.2778,
-      "step": 5160
-    },
-    {
-      "epoch": 12.72167487684729,
-      "grad_norm": 1.015625,
-      "learning_rate": 1.3728916381217394e-05,
-      "loss": 0.2768,
-      "step": 5165
-    },
-    {
-      "epoch": 12.733990147783251,
-      "grad_norm": 0.9296875,
-      "learning_rate": 1.3584343324896964e-05,
-      "loss": 0.2729,
-      "step": 5170
-    },
-    {
-      "epoch": 12.746305418719212,
-      "grad_norm": 0.90234375,
-      "learning_rate": 1.3440480029830127e-05,
-      "loss": 0.2775,
-      "step": 5175
-    },
-    {
-      "epoch": 12.758620689655173,
-      "grad_norm": 0.99609375,
-      "learning_rate": 1.3297327677615124e-05,
-      "loss": 0.2774,
-      "step": 5180
-    },
-    {
-      "epoch": 12.770935960591133,
-      "grad_norm": 0.875,
-      "learning_rate": 1.3154887444011087e-05,
-      "loss": 0.2798,
-      "step": 5185
-    },
-    {
-      "epoch": 12.783251231527094,
-      "grad_norm": 0.9296875,
-      "learning_rate": 1.301316049892818e-05,
-      "loss": 0.2758,
-      "step": 5190
-    },
-    {
-      "epoch": 12.795566502463053,
-      "grad_norm": 0.87890625,
-      "learning_rate": 1.2872148006418161e-05,
-      "loss": 0.2783,
-      "step": 5195
-    },
-    {
-      "epoch": 12.807881773399014,
-      "grad_norm": 0.953125,
-      "learning_rate": 1.2731851124664685e-05,
-      "loss": 0.2666,
-      "step": 5200
-    },
-    {
-      "epoch": 12.820197044334975,
-      "grad_norm": 0.92578125,
-      "learning_rate": 1.2592271005973888e-05,
-      "loss": 0.2807,
-      "step": 5205
-    },
-    {
-      "epoch": 12.832512315270936,
-      "grad_norm": 0.94140625,
-      "learning_rate": 1.2453408796764876e-05,
-      "loss": 0.2776,
-      "step": 5210
-    },
-    {
-      "epoch": 12.844827586206897,
-      "grad_norm": 0.9296875,
-      "learning_rate": 1.2315265637560357e-05,
-      "loss": 0.2727,
-      "step": 5215
-    },
-    {
-      "epoch": 12.857142857142858,
-      "grad_norm": 0.8828125,
-      "learning_rate": 1.2177842662977135e-05,
-      "loss": 0.2788,
-      "step": 5220
-    },
-    {
-      "epoch": 12.869458128078819,
-      "grad_norm": 0.90234375,
-      "learning_rate": 1.2041141001717027e-05,
-      "loss": 0.2732,
-      "step": 5225
-    },
-    {
-      "epoch": 12.881773399014778,
-      "grad_norm": 0.9453125,
-      "learning_rate": 1.1905161776557327e-05,
-      "loss": 0.2767,
-      "step": 5230
-    },
-    {
-      "epoch": 12.894088669950738,
-      "grad_norm": 0.87890625,
-      "learning_rate": 1.1769906104341832e-05,
-      "loss": 0.2717,
-      "step": 5235
-    },
-    {
-      "epoch": 12.9064039408867,
-      "grad_norm": 0.8984375,
-      "learning_rate": 1.1635375095971435e-05,
-      "loss": 0.2775,
-      "step": 5240
-    },
-    {
-      "epoch": 12.91871921182266,
-      "grad_norm": 0.890625,
-      "learning_rate": 1.1501569856395223e-05,
-      "loss": 0.2795,
-      "step": 5245
-    },
-    {
-      "epoch": 12.931034482758621,
-      "grad_norm": 0.94921875,
-      "learning_rate": 1.136849148460125e-05,
-      "loss": 0.281,
-      "step": 5250
-    },
-    {
-      "epoch": 12.943349753694582,
-      "grad_norm": 0.95703125,
-      "learning_rate": 1.1236141073607542e-05,
-      "loss": 0.275,
-      "step": 5255
-    },
-    {
-      "epoch": 12.955665024630543,
-      "grad_norm": 0.90234375,
-      "learning_rate": 1.1104519710453176e-05,
-      "loss": 0.2769,
-      "step": 5260
-    },
-    {
-      "epoch": 12.967980295566502,
-      "grad_norm": 0.93359375,
-      "learning_rate": 1.0973628476189257e-05,
-      "loss": 0.2804,
-      "step": 5265
-    },
-    {
-      "epoch": 12.980295566502463,
-      "grad_norm": 0.91015625,
-      "learning_rate": 1.0843468445870142e-05,
-      "loss": 0.2693,
-      "step": 5270
-    },
-    {
-      "epoch": 12.992610837438423,
-      "grad_norm": 0.8984375,
-      "learning_rate": 1.0714040688544535e-05,
-      "loss": 0.2802,
-      "step": 5275
-    },
-    {
-      "epoch": 13.0,
-      "eval_loss": 5.508431434631348,
-      "eval_runtime": 2.0431,
-      "eval_samples_per_second": 4.895,
-      "eval_steps_per_second": 0.979,
-      "step": 5278
-    },
-    {
-      "epoch": 13.004926108374384,
-      "grad_norm": 0.8359375,
-      "learning_rate": 1.0585346267246743e-05,
-      "loss": 0.2778,
-      "step": 5280
-    },
-    {
-      "epoch": 13.017241379310345,
-      "grad_norm": 0.89453125,
-      "learning_rate": 1.045738623898791e-05,
-      "loss": 0.2741,
-      "step": 5285
-    },
-    {
-      "epoch": 13.029556650246306,
-      "grad_norm": 0.828125,
-      "learning_rate": 1.0330161654747395e-05,
-      "loss": 0.2757,
-      "step": 5290
-    },
-    {
-      "epoch": 13.041871921182265,
-      "grad_norm": 0.84375,
-      "learning_rate": 1.0203673559464089e-05,
-      "loss": 0.2747,
-      "step": 5295
-    },
-    {
-      "epoch": 13.054187192118226,
-      "grad_norm": 0.91015625,
-      "learning_rate": 1.0077922992027867e-05,
-      "loss": 0.2692,
-      "step": 5300
-    },
-    {
-      "epoch": 13.066502463054187,
-      "grad_norm": 0.921875,
-      "learning_rate": 9.952910985270969e-06,
-      "loss": 0.2673,
-      "step": 5305
-    },
-    {
-      "epoch": 13.078817733990148,
-      "grad_norm": 0.875,
-      "learning_rate": 9.82863856595968e-06,
-      "loss": 0.2692,
-      "step": 5310
-    },
-    {
-      "epoch": 13.091133004926109,
-      "grad_norm": 0.8828125,
-      "learning_rate": 9.70510675478572e-06,
-      "loss": 0.2695,
-      "step": 5315
-    },
-    {
-      "epoch": 13.10344827586207,
-      "grad_norm": 0.8828125,
-      "learning_rate": 9.582316566357996e-06,
-      "loss": 0.2742,
-      "step": 5320
-    },
-    {
-      "epoch": 13.11576354679803,
-      "grad_norm": 0.90234375,
-      "learning_rate": 9.460269009194167e-06,
-      "loss": 0.2727,
-      "step": 5325
-    },
-    {
-      "epoch": 13.12807881773399,
-      "grad_norm": 0.953125,
-      "learning_rate": 9.338965085712459e-06,
-      "loss": 0.2656,
-      "step": 5330
-    },
-    {
-      "epoch": 13.14039408866995,
-      "grad_norm": 0.94140625,
-      "learning_rate": 9.218405792223361e-06,
-      "loss": 0.2655,
-      "step": 5335
-    },
-    {
-      "epoch": 13.152709359605911,
-      "grad_norm": 0.88671875,
-      "learning_rate": 9.098592118921435e-06,
-      "loss": 0.2744,
-      "step": 5340
-    },
-    {
-      "epoch": 13.165024630541872,
-      "grad_norm": 0.9375,
-      "learning_rate": 8.979525049877258e-06,
-      "loss": 0.2752,
-      "step": 5345
-    },
-    {
-      "epoch": 13.177339901477833,
-      "grad_norm": 0.90234375,
-      "learning_rate": 8.861205563029228e-06,
-      "loss": 0.273,
-      "step": 5350
-    },
-    {
-      "epoch": 13.189655172413794,
-      "grad_norm": 0.94921875,
-      "learning_rate": 8.74363463017569e-06,
-      "loss": 0.2742,
-      "step": 5355
-    },
-    {
-      "epoch": 13.201970443349754,
-      "grad_norm": 0.93359375,
-      "learning_rate": 8.626813216966745e-06,
-      "loss": 0.2751,
-      "step": 5360
-    },
-    {
-      "epoch": 13.214285714285714,
-      "grad_norm": 0.89453125,
-      "learning_rate": 8.510742282896544e-06,
-      "loss": 0.2712,
-      "step": 5365
-    },
-    {
-      "epoch": 13.226600985221674,
-      "grad_norm": 0.84375,
-      "learning_rate": 8.395422781295192e-06,
-      "loss": 0.2797,
-      "step": 5370
-    },
-    {
-      "epoch": 13.238916256157635,
-      "grad_norm": 0.83984375,
-      "learning_rate": 8.28085565932113e-06,
-      "loss": 0.2744,
-      "step": 5375
-    },
-    {
-      "epoch": 13.251231527093596,
-      "grad_norm": 0.87109375,
-      "learning_rate": 8.167041857953162e-06,
-      "loss": 0.2678,
-      "step": 5380
-    },
-    {
-      "epoch": 13.263546798029557,
-      "grad_norm": 0.84765625,
-      "learning_rate": 8.053982311982867e-06,
-      "loss": 0.2727,
-      "step": 5385
-    },
-    {
-      "epoch": 13.275862068965518,
-      "grad_norm": 0.96484375,
-      "learning_rate": 7.94167795000682e-06,
-      "loss": 0.2699,
-      "step": 5390
-    },
-    {
-      "epoch": 13.288177339901479,
-      "grad_norm": 0.875,
-      "learning_rate": 7.830129694419065e-06,
-      "loss": 0.2722,
-      "step": 5395
-    },
-    {
-      "epoch": 13.300492610837438,
-      "grad_norm": 0.94140625,
-      "learning_rate": 7.719338461403435e-06,
-      "loss": 0.2666,
-      "step": 5400
-    },
-    {
-      "epoch": 13.312807881773399,
-      "grad_norm": 0.8671875,
-      "learning_rate": 7.609305160926128e-06,
-      "loss": 0.2712,
-      "step": 5405
-    },
-    {
-      "epoch": 13.32512315270936,
-      "grad_norm": 0.8671875,
-      "learning_rate": 7.500030696728133e-06,
-      "loss": 0.2724,
-      "step": 5410
-    },
-    {
-      "epoch": 13.33743842364532,
-      "grad_norm": 0.88671875,
-      "learning_rate": 7.3915159663179075e-06,
-      "loss": 0.2766,
-      "step": 5415
-    },
-    {
-      "epoch": 13.349753694581281,
-      "grad_norm": 0.87109375,
-      "learning_rate": 7.283761860963933e-06,
-      "loss": 0.2712,
-      "step": 5420
-    },
-    {
-      "epoch": 13.362068965517242,
-      "grad_norm": 0.984375,
-      "learning_rate": 7.176769265687389e-06,
-      "loss": 0.2697,
-      "step": 5425
-    },
-    {
-      "epoch": 13.374384236453203,
-      "grad_norm": 0.92578125,
-      "learning_rate": 7.070539059254977e-06,
-      "loss": 0.273,
-      "step": 5430
-    },
-    {
-      "epoch": 13.386699507389162,
-      "grad_norm": 0.87109375,
-      "learning_rate": 6.965072114171578e-06,
-      "loss": 0.2658,
-      "step": 5435
-    },
-    {
-      "epoch": 13.399014778325123,
-      "grad_norm": 0.93359375,
-      "learning_rate": 6.860369296673197e-06,
-      "loss": 0.2687,
-      "step": 5440
-    },
-    {
-      "epoch": 13.411330049261084,
-      "grad_norm": 0.92578125,
-      "learning_rate": 6.756431466719737e-06,
-      "loss": 0.2734,
-      "step": 5445
-    },
-    {
-      "epoch": 13.423645320197044,
-      "grad_norm": 0.9140625,
-      "learning_rate": 6.653259477988083e-06,
-      "loss": 0.2736,
-      "step": 5450
-    },
-    {
-      "epoch": 13.435960591133005,
-      "grad_norm": 0.8671875,
-      "learning_rate": 6.5508541778649066e-06,
-      "loss": 0.2752,
-      "step": 5455
-    },
-    {
-      "epoch": 13.448275862068966,
-      "grad_norm": 0.9140625,
-      "learning_rate": 6.4492164074399065e-06,
-      "loss": 0.2758,
-      "step": 5460
-    },
-    {
-      "epoch": 13.460591133004925,
-      "grad_norm": 0.83984375,
-      "learning_rate": 6.348347001498711e-06,
-      "loss": 0.2713,
-      "step": 5465
-    },
-    {
-      "epoch": 13.472906403940886,
-      "grad_norm": 0.9375,
-      "learning_rate": 6.248246788516165e-06,
-      "loss": 0.2745,
-      "step": 5470
-    },
-    {
-      "epoch": 13.485221674876847,
-      "grad_norm": 0.8671875,
-      "learning_rate": 6.148916590649434e-06,
-      "loss": 0.2728,
-      "step": 5475
-    },
-    {
-      "epoch": 13.497536945812808,
-      "grad_norm": 0.87109375,
-      "learning_rate": 6.050357223731318e-06,
-      "loss": 0.2722,
-      "step": 5480
-    },
-    {
-      "epoch": 13.509852216748769,
-      "grad_norm": 0.875,
-      "learning_rate": 5.9525694972634715e-06,
-      "loss": 0.2679,
-      "step": 5485
-    },
-    {
-      "epoch": 13.52216748768473,
-      "grad_norm": 0.859375,
-      "learning_rate": 5.8555542144098865e-06,
-      "loss": 0.2699,
-      "step": 5490
-    },
-    {
-      "epoch": 13.53448275862069,
-      "grad_norm": 0.859375,
-      "learning_rate": 5.7593121719900835e-06,
-      "loss": 0.2798,
-      "step": 5495
-    },
-    {
-      "epoch": 13.54679802955665,
-      "grad_norm": 0.86328125,
-      "learning_rate": 5.663844160472865e-06,
-      "loss": 0.2699,
-      "step": 5500
-    },
-    {
-      "epoch": 13.55911330049261,
-      "grad_norm": 0.921875,
-      "learning_rate": 5.569150963969494e-06,
-      "loss": 0.2744,
-      "step": 5505
-    },
-    {
-      "epoch": 13.571428571428571,
-      "grad_norm": 0.83984375,
-      "learning_rate": 5.475233360227516e-06,
-      "loss": 0.2701,
-      "step": 5510
-    },
-    {
-      "epoch": 13.583743842364532,
-      "grad_norm": 0.9375,
-      "learning_rate": 5.382092120624216e-06,
-      "loss": 0.277,
-      "step": 5515
-    },
-    {
-      "epoch": 13.596059113300493,
-      "grad_norm": 0.85546875,
-      "learning_rate": 5.289728010160366e-06,
-      "loss": 0.2759,
-      "step": 5520
-    },
-    {
-      "epoch": 13.608374384236454,
-      "grad_norm": 0.875,
-      "learning_rate": 5.19814178745388e-06,
-      "loss": 0.2708,
-      "step": 5525
-    },
-    {
-      "epoch": 13.620689655172415,
-      "grad_norm": 0.89453125,
-      "learning_rate": 5.10733420473366e-06,
-      "loss": 0.2766,
-      "step": 5530
-    },
-    {
-      "epoch": 13.633004926108374,
-      "grad_norm": 0.86328125,
-      "learning_rate": 5.0173060078333225e-06,
-      "loss": 0.274,
-      "step": 5535
-    },
-    {
-      "epoch": 13.645320197044335,
-      "grad_norm": 0.87890625,
-      "learning_rate": 4.928057936185138e-06,
-      "loss": 0.2741,
-      "step": 5540
-    },
-    {
-      "epoch": 13.657635467980295,
-      "grad_norm": 0.85546875,
-      "learning_rate": 4.839590722813991e-06,
-      "loss": 0.277,
-      "step": 5545
-    },
-    {
-      "epoch": 13.669950738916256,
-      "grad_norm": 0.91796875,
-      "learning_rate": 4.7519050943312325e-06,
-      "loss": 0.2695,
-      "step": 5550
-    },
-    {
-      "epoch": 13.682266009852217,
-      "grad_norm": 0.9453125,
-      "learning_rate": 4.665001770928845e-06,
-      "loss": 0.2699,
-      "step": 5555
-    },
-    {
-      "epoch": 13.694581280788178,
-      "grad_norm": 0.91796875,
-      "learning_rate": 4.578881466373441e-06,
-      "loss": 0.2676,
-      "step": 5560
-    },
-    {
-      "epoch": 13.706896551724139,
-      "grad_norm": 0.87890625,
-      "learning_rate": 4.493544888000467e-06,
-      "loss": 0.2756,
-      "step": 5565
-    },
-    {
-      "epoch": 13.719211822660098,
-      "grad_norm": 0.88671875,
-      "learning_rate": 4.408992736708317e-06,
-      "loss": 0.2712,
-      "step": 5570
-    },
-    {
-      "epoch": 13.731527093596059,
-      "grad_norm": 0.90625,
-      "learning_rate": 4.3252257069526516e-06,
-      "loss": 0.2772,
-      "step": 5575
-    },
-    {
-      "epoch": 13.74384236453202,
-      "grad_norm": 0.8671875,
-      "learning_rate": 4.242244486740643e-06,
-      "loss": 0.2729,
-      "step": 5580
-    },
-    {
-      "epoch": 13.75615763546798,
-      "grad_norm": 0.875,
-      "learning_rate": 4.160049757625362e-06,
-      "loss": 0.2736,
-      "step": 5585
-    },
-    {
-      "epoch": 13.768472906403941,
-      "grad_norm": 0.86328125,
-      "learning_rate": 4.078642194700111e-06,
-      "loss": 0.2728,
-      "step": 5590
-    },
-    {
-      "epoch": 13.780788177339902,
-      "grad_norm": 0.8984375,
-      "learning_rate": 3.99802246659301e-06,
-      "loss": 0.2699,
-      "step": 5595
-    },
-    {
-      "epoch": 13.793103448275861,
-      "grad_norm": 1.1171875,
-      "learning_rate": 3.918191235461333e-06,
-      "loss": 0.269,
-      "step": 5600
-    },
-    {
-      "epoch": 13.805418719211822,
-      "grad_norm": 0.9765625,
-      "learning_rate": 3.839149156986233e-06,
-      "loss": 0.2687,
-      "step": 5605
-    },
-    {
-      "epoch": 13.817733990147783,
-      "grad_norm": 0.86328125,
-      "learning_rate": 3.760896880367215e-06,
-      "loss": 0.274,
-      "step": 5610
-    },
-    {
-      "epoch": 13.830049261083744,
-      "grad_norm": 0.87109375,
-      "learning_rate": 3.683435048316941e-06,
-      "loss": 0.2695,
-      "step": 5615
-    },
-    {
-      "epoch": 13.842364532019705,
-      "grad_norm": 0.8515625,
-      "learning_rate": 3.6067642970558312e-06,
-      "loss": 0.2738,
-      "step": 5620
-    },
-    {
-      "epoch": 13.854679802955665,
-      "grad_norm": 0.8671875,
-      "learning_rate": 3.530885256306915e-06,
-      "loss": 0.2728,
-      "step": 5625
-    },
-    {
-      "epoch": 13.866995073891626,
-      "grad_norm": 0.9453125,
-      "learning_rate": 3.455798549290645e-06,
-      "loss": 0.268,
-      "step": 5630
-    },
-    {
-      "epoch": 13.879310344827585,
-      "grad_norm": 0.828125,
-      "learning_rate": 3.381504792719714e-06,
-      "loss": 0.2697,
-      "step": 5635
-    },
-    {
-      "epoch": 13.891625615763546,
-      "grad_norm": 0.88671875,
-      "learning_rate": 3.308004596794101e-06,
-      "loss": 0.2817,
-      "step": 5640
-    },
-    {
-      "epoch": 13.903940886699507,
-      "grad_norm": 0.92578125,
-      "learning_rate": 3.2352985651959657e-06,
-      "loss": 0.2706,
-      "step": 5645
-    },
-    {
-      "epoch": 13.916256157635468,
-      "grad_norm": 0.96484375,
-      "learning_rate": 3.1633872950847523e-06,
-      "loss": 0.2751,
-      "step": 5650
-    },
-    {
-      "epoch": 13.928571428571429,
-      "grad_norm": 0.9296875,
-      "learning_rate": 3.092271377092215e-06,
-      "loss": 0.2793,
-      "step": 5655
-    },
-    {
-      "epoch": 13.94088669950739,
-      "grad_norm": 0.91015625,
-      "learning_rate": 3.021951395317646e-06,
-      "loss": 0.2755,
-      "step": 5660
-    },
-    {
-      "epoch": 13.95320197044335,
-      "grad_norm": 0.9765625,
-      "learning_rate": 2.9524279273230428e-06,
-      "loss": 0.2787,
-      "step": 5665
-    },
-    {
-      "epoch": 13.96551724137931,
-      "grad_norm": 1.0546875,
-      "learning_rate": 2.8837015441283586e-06,
-      "loss": 0.2675,
-      "step": 5670
-    },
-    {
-      "epoch": 13.97783251231527,
-      "grad_norm": 0.86328125,
-      "learning_rate": 2.815772810206785e-06,
-      "loss": 0.2726,
-      "step": 5675
-    },
-    {
-      "epoch": 13.990147783251231,
-      "grad_norm": 0.890625,
-      "learning_rate": 2.7486422834802186e-06,
-      "loss": 0.2693,
-      "step": 5680
-    },
-    {
-      "epoch": 14.0,
-      "eval_loss": 5.538418292999268,
-      "eval_runtime": 2.0446,
-      "eval_samples_per_second": 4.891,
-      "eval_steps_per_second": 0.978,
-      "step": 5684
-    },
-    {
-      "epoch": 14.002463054187192,
-      "grad_norm": 0.98046875,
-      "learning_rate": 2.682310515314512e-06,
-      "loss": 0.2708,
-      "step": 5685
-    },
-    {
-      "epoch": 14.014778325123153,
-      "grad_norm": 0.84765625,
-      "learning_rate": 2.616778050515145e-06,
-      "loss": 0.2679,
-      "step": 5690
-    },
-    {
-      "epoch": 14.027093596059114,
-      "grad_norm": 0.90234375,
-      "learning_rate": 2.5520454273225582e-06,
-      "loss": 0.2706,
-      "step": 5695
-    },
-    {
-      "epoch": 14.039408866995075,
-      "grad_norm": 0.95703125,
-      "learning_rate": 2.488113177407869e-06,
-      "loss": 0.2706,
-      "step": 5700
-    },
-    {
-      "epoch": 14.051724137931034,
-      "grad_norm": 0.89453125,
-      "learning_rate": 2.4249818258684664e-06,
-      "loss": 0.2715,
-      "step": 5705
-    },
-    {
-      "epoch": 14.064039408866995,
-      "grad_norm": 0.9296875,
-      "learning_rate": 2.3626518912236327e-06,
-      "loss": 0.2691,
-      "step": 5710
-    },
-    {
-      "epoch": 14.076354679802956,
-      "grad_norm": 0.96875,
-      "learning_rate": 2.3011238854103947e-06,
-      "loss": 0.2757,
-      "step": 5715
-    },
-    {
-      "epoch": 14.088669950738916,
-      "grad_norm": 0.91796875,
-      "learning_rate": 2.240398313779235e-06,
-      "loss": 0.2728,
-      "step": 5720
-    },
-    {
-      "epoch": 14.100985221674877,
-      "grad_norm": 0.8828125,
-      "learning_rate": 2.180475675089988e-06,
-      "loss": 0.2701,
-      "step": 5725
-    },
-    {
-      "epoch": 14.113300492610838,
-      "grad_norm": 0.921875,
-      "learning_rate": 2.1213564615077065e-06,
-      "loss": 0.2639,
-      "step": 5730
-    },
-    {
-      "epoch": 14.125615763546797,
-      "grad_norm": 0.85546875,
-      "learning_rate": 2.0630411585986554e-06,
-      "loss": 0.2732,
-      "step": 5735
-    },
-    {
-      "epoch": 14.137931034482758,
-      "grad_norm": 0.9453125,
-      "learning_rate": 2.0055302453262924e-06,
-      "loss": 0.274,
-      "step": 5740
-    },
-    {
-      "epoch": 14.150246305418719,
-      "grad_norm": 0.90625,
-      "learning_rate": 1.9488241940473828e-06,
-      "loss": 0.2749,
-      "step": 5745
-    },
-    {
-      "epoch": 14.16256157635468,
-      "grad_norm": 0.875,
-      "learning_rate": 1.8929234705080346e-06,
-      "loss": 0.2654,
-      "step": 5750
-    },
-    {
-      "epoch": 14.17487684729064,
-      "grad_norm": 0.88671875,
-      "learning_rate": 1.8378285338399692e-06,
-      "loss": 0.2622,
-      "step": 5755
-    },
-    {
-      "epoch": 14.187192118226601,
-      "grad_norm": 0.90234375,
-      "learning_rate": 1.783539836556669e-06,
-      "loss": 0.2647,
-      "step": 5760
-    },
-    {
-      "epoch": 14.199507389162562,
-      "grad_norm": 0.86328125,
-      "learning_rate": 1.7300578245497245e-06,
-      "loss": 0.2727,
-      "step": 5765
-    },
-    {
-      "epoch": 14.211822660098521,
-      "grad_norm": 0.90234375,
-      "learning_rate": 1.6773829370851368e-06,
-      "loss": 0.266,
-      "step": 5770
-    },
-    {
-      "epoch": 14.224137931034482,
-      "grad_norm": 0.8828125,
-      "learning_rate": 1.6255156067997323e-06,
-      "loss": 0.2707,
-      "step": 5775
-    },
-    {
-      "epoch": 14.236453201970443,
-      "grad_norm": 0.94140625,
-      "learning_rate": 1.5744562596975432e-06,
-      "loss": 0.276,
-      "step": 5780
-    },
-    {
-      "epoch": 14.248768472906404,
-      "grad_norm": 0.93359375,
-      "learning_rate": 1.524205315146432e-06,
-      "loss": 0.2701,
-      "step": 5785
-    },
-    {
-      "epoch": 14.261083743842365,
-      "grad_norm": 0.96875,
-      "learning_rate": 1.474763185874517e-06,
-      "loss": 0.271,
-      "step": 5790
-    },
-    {
-      "epoch": 14.273399014778326,
-      "grad_norm": 0.96484375,
-      "learning_rate": 1.4261302779668862e-06,
-      "loss": 0.2757,
-      "step": 5795
-    },
-    {
-      "epoch": 14.285714285714286,
-      "grad_norm": 0.9296875,
-      "learning_rate": 1.378306990862177e-06,
-      "loss": 0.2794,
-      "step": 5800
-    },
-    {
-      "epoch": 14.298029556650246,
-      "grad_norm": 0.87890625,
-      "learning_rate": 1.3312937173493577e-06,
-      "loss": 0.2699,
-      "step": 5805
-    },
-    {
-      "epoch": 14.310344827586206,
-      "grad_norm": 0.98046875,
-      "learning_rate": 1.285090843564485e-06,
-      "loss": 0.2706,
-      "step": 5810
-    },
-    {
-      "epoch": 14.322660098522167,
-      "grad_norm": 1.046875,
-      "learning_rate": 1.2396987489874946e-06,
-      "loss": 0.2731,
-      "step": 5815
-    },
-    {
-      "epoch": 14.334975369458128,
-      "grad_norm": 0.9140625,
-      "learning_rate": 1.19511780643915e-06,
-      "loss": 0.2703,
-      "step": 5820
-    },
-    {
-      "epoch": 14.347290640394089,
-      "grad_norm": 0.8359375,
-      "learning_rate": 1.1513483820779214e-06,
-      "loss": 0.2602,
-      "step": 5825
-    },
-    {
-      "epoch": 14.35960591133005,
-      "grad_norm": 0.87109375,
-      "learning_rate": 1.108390835397044e-06,
-      "loss": 0.2748,
-      "step": 5830
-    },
-    {
-      "epoch": 14.37192118226601,
-      "grad_norm": 0.890625,
-      "learning_rate": 1.066245519221465e-06,
-      "loss": 0.2692,
-      "step": 5835
-    },
-    {
-      "epoch": 14.38423645320197,
-      "grad_norm": 0.87890625,
-      "learning_rate": 1.024912779705045e-06,
-      "loss": 0.2657,
-      "step": 5840
-    },
-    {
-      "epoch": 14.39655172413793,
-      "grad_norm": 0.890625,
-      "learning_rate": 9.843929563276733e-07,
-      "loss": 0.2782,
-      "step": 5845
-    },
-    {
-      "epoch": 14.408866995073891,
-      "grad_norm": 0.890625,
-      "learning_rate": 9.446863818924679e-07,
-      "loss": 0.2744,
-      "step": 5850
-    },
-    {
-      "epoch": 14.421182266009852,
-      "grad_norm": 0.86328125,
-      "learning_rate": 9.05793382523068e-07,
-      "loss": 0.2685,
-      "step": 5855
-    },
-    {
-      "epoch": 14.433497536945813,
-      "grad_norm": 0.87109375,
-      "learning_rate": 8.67714277660947e-07,
-      "loss": 0.2718,
-      "step": 5860
-    },
-    {
-      "epoch": 14.445812807881774,
-      "grad_norm": 0.85546875,
-      "learning_rate": 8.304493800627589e-07,
-      "loss": 0.2679,
-      "step": 5865
-    },
-    {
-      "epoch": 14.458128078817733,
-      "grad_norm": 0.87890625,
-      "learning_rate": 7.939989957978289e-07,
-      "loss": 0.2673,
-      "step": 5870
-    },
-    {
-      "epoch": 14.470443349753694,
-      "grad_norm": 0.93359375,
-      "learning_rate": 7.583634242455784e-07,
-      "loss": 0.2773,
-      "step": 5875
-    },
-    {
-      "epoch": 14.482758620689655,
-      "grad_norm": 0.86328125,
-      "learning_rate": 7.235429580931152e-07,
-      "loss": 0.2713,
-      "step": 5880
-    },
-    {
-      "epoch": 14.495073891625616,
-      "grad_norm": 0.9140625,
-      "learning_rate": 6.895378833328025e-07,
-      "loss": 0.2705,
-      "step": 5885
-    },
-    {
-      "epoch": 14.507389162561577,
-      "grad_norm": 0.828125,
-      "learning_rate": 6.563484792599161e-07,
-      "loss": 0.2725,
-      "step": 5890
-    },
-    {
-      "epoch": 14.519704433497537,
-      "grad_norm": 0.80859375,
-      "learning_rate": 6.239750184703464e-07,
-      "loss": 0.2645,
-      "step": 5895
-    },
-    {
-      "epoch": 14.532019704433498,
-      "grad_norm": 0.87109375,
-      "learning_rate": 5.924177668583552e-07,
-      "loss": 0.2703,
-      "step": 5900
-    },
-    {
-      "epoch": 14.544334975369457,
-      "grad_norm": 0.8984375,
-      "learning_rate": 5.616769836144231e-07,
-      "loss": 0.2758,
-      "step": 5905
-    },
-    {
-      "epoch": 14.556650246305418,
-      "grad_norm": 0.86328125,
-      "learning_rate": 5.317529212230721e-07,
-      "loss": 0.2701,
-      "step": 5910
-    },
-    {
-      "epoch": 14.568965517241379,
-      "grad_norm": 0.95703125,
-      "learning_rate": 5.026458254608457e-07,
-      "loss": 0.2704,
-      "step": 5915
-    },
-    {
-      "epoch": 14.58128078817734,
-      "grad_norm": 0.8671875,
-      "learning_rate": 4.7435593539423284e-07,
-      "loss": 0.2724,
-      "step": 5920
-    },
-    {
-      "epoch": 14.5935960591133,
-      "grad_norm": 0.921875,
-      "learning_rate": 4.4688348337774686e-07,
-      "loss": 0.2845,
-      "step": 5925
-    },
-    {
-      "epoch": 14.605911330049262,
-      "grad_norm": 0.80859375,
-      "learning_rate": 4.202286950520162e-07,
-      "loss": 0.272,
-      "step": 5930
-    },
-    {
-      "epoch": 14.618226600985222,
-      "grad_norm": 0.87109375,
-      "learning_rate": 3.943917893418858e-07,
-      "loss": 0.276,
-      "step": 5935
-    },
-    {
-      "epoch": 14.630541871921181,
-      "grad_norm": 0.9765625,
-      "learning_rate": 3.693729784546962e-07,
-      "loss": 0.2689,
-      "step": 5940
-    },
-    {
-      "epoch": 14.642857142857142,
-      "grad_norm": 0.86328125,
-      "learning_rate": 3.451724678784518e-07,
-      "loss": 0.2802,
-      "step": 5945
-    },
-    {
-      "epoch": 14.655172413793103,
-      "grad_norm": 0.87109375,
-      "learning_rate": 3.21790456380211e-07,
-      "loss": 0.2675,
-      "step": 5950
-    },
-    {
-      "epoch": 14.667487684729064,
-      "grad_norm": 0.88671875,
-      "learning_rate": 2.9922713600439854e-07,
-      "loss": 0.2742,
-      "step": 5955
-    },
-    {
-      "epoch": 14.679802955665025,
-      "grad_norm": 0.83203125,
-      "learning_rate": 2.7748269207125145e-07,
-      "loss": 0.281,
-      "step": 5960
-    },
-    {
-      "epoch": 14.692118226600986,
-      "grad_norm": 0.87109375,
-      "learning_rate": 2.565573031753199e-07,
-      "loss": 0.2728,
-      "step": 5965
-    },
-    {
-      "epoch": 14.704433497536947,
-      "grad_norm": 0.89453125,
-      "learning_rate": 2.3645114118395762e-07,
-      "loss": 0.2788,
-      "step": 5970
-    },
-    {
-      "epoch": 14.716748768472906,
-      "grad_norm": 0.91796875,
-      "learning_rate": 2.1716437123591172e-07,
-      "loss": 0.269,
-      "step": 5975
-    },
-    {
-      "epoch": 14.729064039408867,
-      "grad_norm": 0.89453125,
-      "learning_rate": 1.986971517400016e-07,
-      "loss": 0.2742,
-      "step": 5980
-    },
-    {
-      "epoch": 14.741379310344827,
-      "grad_norm": 0.921875,
-      "learning_rate": 1.8104963437381993e-07,
-      "loss": 0.2687,
-      "step": 5985
-    },
-    {
-      "epoch": 14.753694581280788,
-      "grad_norm": 0.9375,
-      "learning_rate": 1.6422196408241165e-07,
-      "loss": 0.2788,
-      "step": 5990
-    },
-    {
-      "epoch": 14.766009852216749,
-      "grad_norm": 0.9296875,
-      "learning_rate": 1.4821427907719677e-07,
-      "loss": 0.2691,
-      "step": 5995
-    },
-    {
-      "epoch": 14.77832512315271,
-      "grad_norm": 0.90234375,
-      "learning_rate": 1.3302671083474938e-07,
-      "loss": 0.2757,
-      "step": 6000
-    },
-    {
-      "epoch": 14.790640394088669,
-      "grad_norm": 0.9296875,
-      "learning_rate": 1.1865938409573174e-07,
-      "loss": 0.2764,
-      "step": 6005
-    },
-    {
-      "epoch": 14.80295566502463,
-      "grad_norm": 0.87109375,
-      "learning_rate": 1.0511241686389505e-07,
-      "loss": 0.2734,
-      "step": 6010
-    },
-    {
-      "epoch": 14.81527093596059,
-      "grad_norm": 0.84765625,
-      "learning_rate": 9.238592040512472e-08,
-      "loss": 0.2696,
-      "step": 6015
-    },
-    {
-      "epoch": 14.827586206896552,
-      "grad_norm": 0.8671875,
-      "learning_rate": 8.047999924645222e-08,
-      "loss": 0.2647,
-      "step": 6020
-    },
-    {
-      "epoch": 14.839901477832512,
-      "grad_norm": 0.90625,
-      "learning_rate": 6.939475117526683e-08,
-      "loss": 0.2754,
-      "step": 6025
-    },
-    {
-      "epoch": 14.852216748768473,
-      "grad_norm": 0.9140625,
-      "learning_rate": 5.913026723850523e-08,
-      "loss": 0.2683,
-      "step": 6030
-    },
-    {
-      "epoch": 14.864532019704434,
-      "grad_norm": 1.109375,
-      "learning_rate": 4.968663174187427e-08,
-      "loss": 0.2692,
-      "step": 6035
-    },
-    {
-      "epoch": 14.876847290640395,
-      "grad_norm": 0.87109375,
-      "learning_rate": 4.106392224915156e-08,
-      "loss": 0.2759,
-      "step": 6040
-    },
-    {
-      "epoch": 14.889162561576354,
-      "grad_norm": 0.96484375,
-      "learning_rate": 3.3262209581619297e-08,
-      "loss": 0.2735,
-      "step": 6045
-    },
-    {
-      "epoch": 14.901477832512315,
-      "grad_norm": 0.83203125,
-      "learning_rate": 2.6281557817386947e-08,
-      "loss": 0.2697,
-      "step": 6050
-    },
-    {
-      "epoch": 14.913793103448276,
-      "grad_norm": 0.828125,
-      "learning_rate": 2.012202429091392e-08,
-      "loss": 0.2738,
-      "step": 6055
-    },
-    {
-      "epoch": 14.926108374384237,
-      "grad_norm": 0.890625,
-      "learning_rate": 1.4783659592576548e-08,
-      "loss": 0.2695,
-      "step": 6060
-    },
-    {
-      "epoch": 14.938423645320198,
-      "grad_norm": 0.96875,
-      "learning_rate": 1.0266507568179595e-08,
-      "loss": 0.2824,
-      "step": 6065
-    },
-    {
-      "epoch": 14.950738916256158,
-      "grad_norm": 0.87109375,
-      "learning_rate": 6.570605318612089e-09,
-      "loss": 0.2728,
-      "step": 6070
-    },
-    {
-      "epoch": 14.963054187192117,
-      "grad_norm": 0.91015625,
-      "learning_rate": 3.6959831996030704e-09,
-      "loss": 0.2733,
-      "step": 6075
-    },
-    {
-      "epoch": 14.975369458128078,
-      "grad_norm": 0.97265625,
-      "learning_rate": 1.6426648213885287e-09,
-      "loss": 0.2686,
-      "step": 6080
-    },
-    {
-      "epoch": 14.98768472906404,
-      "grad_norm": 0.93359375,
-      "learning_rate": 4.10667048589275e-10,
-      "loss": 0.2747,
-      "step": 6085
-    },
-    {
-      "epoch": 15.0,
-      "grad_norm": 0.875,
-      "learning_rate": 0.0,
-      "loss": 0.2715,
-      "step": 6090
-    },
-    {
-      "epoch": 15.0,
-      "eval_loss": 5.538125038146973,
-      "eval_runtime": 2.0459,
-      "eval_samples_per_second": 4.888,
-      "eval_steps_per_second": 0.978,
-      "step": 6090
+      "epoch": 9.954337899543379,
+      "eval_loss": 2.515676259994507,
+      "eval_runtime": 0.2343,
+      "eval_samples_per_second": 42.683,
+      "eval_steps_per_second": 4.268,
+      "step": 1090
     },
     {
-      "epoch": 15.0,
-      "step": 6090,
-      "total_flos": 4.655418812370256e+18,
-      "train_loss": 0.9284773617542436,
-      "train_runtime": 39545.4398,
-      "train_samples_per_second": 2.463,
-      "train_steps_per_second": 0.154
+      "epoch": 9.954337899543379,
+      "step": 1090,
+      "total_flos": 3.327732991202951e+18,
+      "train_loss": 2.136770288659892,
+      "train_runtime": 2636.8816,
+      "train_samples_per_second": 26.554,
+      "train_steps_per_second": 0.413
     }
   ],
   "logging_steps": 5,
-  "max_steps": 6090,
+  "max_steps": 1090,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 15,
+  "num_train_epochs": 10,
   "save_steps": 100,
-  "total_flos": 4.655418812370256e+18,
+  "total_flos": 3.327732991202951e+18,
   "train_batch_size": 4,
   "trial_name": null,
   "trial_params": null