diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -3,338 +3,5322 @@
   "best_model_checkpoint": null,
   "epoch": 1.0,
   "eval_steps": 500,
-  "global_step": 220,
+  "global_step": 3782,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.004545454545454545,
-      "grad_norm": 3.71875,
-      "learning_rate": 9.090909090909091e-06,
-      "loss": 1.8651,
+      "epoch": 0.00026441036488630354,
+      "grad_norm": 3.609375,
+      "learning_rate": 5.277044854881267e-07,
+      "loss": 1.8784,
       "step": 1
     },
     {
-      "epoch": 0.022727272727272728,
-      "grad_norm": 3.296875,
-      "learning_rate": 4.545454545454546e-05,
-      "loss": 1.8721,
+      "epoch": 0.0013220518244315177,
+      "grad_norm": 3.609375,
+      "learning_rate": 2.6385224274406333e-06,
+      "loss": 1.8748,
       "step": 5
     },
     {
-      "epoch": 0.045454545454545456,
-      "grad_norm": 1.7578125,
-      "learning_rate": 9.090909090909092e-05,
-      "loss": 1.8044,
+      "epoch": 0.0026441036488630354,
+      "grad_norm": 3.578125,
+      "learning_rate": 5.2770448548812665e-06,
+      "loss": 1.8755,
       "step": 10
     },
     {
-      "epoch": 0.06818181818181818,
-      "grad_norm": 1.4375,
-      "learning_rate": 0.00013636363636363637,
-      "loss": 1.7402,
+      "epoch": 0.003966155473294554,
+      "grad_norm": 3.5,
+      "learning_rate": 7.915567282321901e-06,
+      "loss": 1.8741,
       "step": 15
     },
     {
-      "epoch": 0.09090909090909091,
-      "grad_norm": 1.9921875,
-      "learning_rate": 0.00018181818181818183,
-      "loss": 1.6832,
+      "epoch": 0.005288207297726071,
+      "grad_norm": 3.359375,
+      "learning_rate": 1.0554089709762533e-05,
+      "loss": 1.8644,
       "step": 20
     },
     {
-      "epoch": 0.11363636363636363,
-      "grad_norm": 1.6640625,
-      "learning_rate": 0.0001998867339183008,
-      "loss": 1.6069,
+      "epoch": 0.006610259122157589,
+      "grad_norm": 2.9375,
+      "learning_rate": 1.3192612137203167e-05,
+      "loss": 1.8583,
       "step": 25
     },
     {
-      "epoch": 0.13636363636363635,
-      "grad_norm": 1.4609375,
-      "learning_rate": 0.00019919548128307954,
-      "loss": 1.582,
+      "epoch": 0.007932310946589107,
+      "grad_norm": 2.5625,
+      "learning_rate": 1.5831134564643802e-05,
+      "loss": 1.8416,
       "step": 30
     },
     {
-      "epoch": 0.1590909090909091,
-      "grad_norm": 1.453125,
-      "learning_rate": 0.00019788024462147788,
-      "loss": 1.5337,
+      "epoch": 0.009254362771020624,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.8469656992084432e-05,
+      "loss": 1.8236,
       "step": 35
     },
     {
-      "epoch": 0.18181818181818182,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00019594929736144976,
-      "loss": 1.4999,
+      "epoch": 0.010576414595452142,
+      "grad_norm": 1.4921875,
+      "learning_rate": 2.1108179419525066e-05,
+      "loss": 1.8066,
       "step": 40
     },
     {
-      "epoch": 0.20454545454545456,
-      "grad_norm": 0.91796875,
-      "learning_rate": 0.00019341478602651069,
-      "loss": 1.486,
+      "epoch": 0.011898466419883659,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.37467018469657e-05,
+      "loss": 1.7951,
       "step": 45
     },
     {
-      "epoch": 0.22727272727272727,
-      "grad_norm": 0.9140625,
-      "learning_rate": 0.00019029265382866214,
-      "loss": 1.4694,
+      "epoch": 0.013220518244315178,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.6385224274406334e-05,
+      "loss": 1.7783,
       "step": 50
     },
     {
-      "epoch": 0.25,
-      "grad_norm": 0.8125,
-      "learning_rate": 0.00018660254037844388,
-      "loss": 1.4435,
+      "epoch": 0.014542570068746695,
+      "grad_norm": 0.92578125,
+      "learning_rate": 2.9023746701846964e-05,
+      "loss": 1.7577,
       "step": 55
     },
     {
-      "epoch": 0.2727272727272727,
-      "grad_norm": 0.85546875,
-      "learning_rate": 0.0001823676581429833,
-      "loss": 1.4326,
+      "epoch": 0.015864621893178214,
+      "grad_norm": 0.8125,
+      "learning_rate": 3.1662269129287604e-05,
+      "loss": 1.7413,
       "step": 60
     },
     {
-      "epoch": 0.29545454545454547,
-      "grad_norm": 0.82421875,
-      "learning_rate": 0.0001776146464291757,
-      "loss": 1.4055,
+      "epoch": 0.01718667371760973,
+      "grad_norm": 0.6953125,
+      "learning_rate": 3.430079155672823e-05,
+      "loss": 1.7217,
       "step": 65
     },
     {
-      "epoch": 0.3181818181818182,
-      "grad_norm": 0.8046875,
-      "learning_rate": 0.00017237340381050703,
-      "loss": 1.374,
+      "epoch": 0.01850872554204125,
+      "grad_norm": 0.69921875,
+      "learning_rate": 3.6939313984168865e-05,
+      "loss": 1.7064,
       "step": 70
     },
     {
-      "epoch": 0.3409090909090909,
-      "grad_norm": 0.96484375,
-      "learning_rate": 0.00016667690005162916,
-      "loss": 1.3776,
+      "epoch": 0.019830777366472766,
+      "grad_norm": 0.57421875,
+      "learning_rate": 3.95778364116095e-05,
+      "loss": 1.6914,
       "step": 75
     },
     {
-      "epoch": 0.36363636363636365,
-      "grad_norm": 0.71875,
-      "learning_rate": 0.00016056096871376667,
-      "loss": 1.3626,
+      "epoch": 0.021152829190904283,
+      "grad_norm": 0.57421875,
+      "learning_rate": 4.221635883905013e-05,
+      "loss": 1.679,
       "step": 80
     },
     {
-      "epoch": 0.38636363636363635,
-      "grad_norm": 0.6328125,
-      "learning_rate": 0.00015406408174555976,
-      "loss": 1.3566,
+      "epoch": 0.0224748810153358,
+      "grad_norm": 0.55859375,
+      "learning_rate": 4.4854881266490766e-05,
+      "loss": 1.6632,
       "step": 85
     },
     {
-      "epoch": 0.4090909090909091,
-      "grad_norm": 0.72265625,
-      "learning_rate": 0.0001472271074772683,
-      "loss": 1.3215,
+      "epoch": 0.023796932839767318,
+      "grad_norm": 0.58203125,
+      "learning_rate": 4.74934036939314e-05,
+      "loss": 1.6472,
       "step": 90
     },
     {
-      "epoch": 0.4318181818181818,
-      "grad_norm": 0.82421875,
-      "learning_rate": 0.00014009305354066137,
-      "loss": 1.338,
+      "epoch": 0.025118984664198835,
+      "grad_norm": 0.546875,
+      "learning_rate": 5.0131926121372033e-05,
+      "loss": 1.6343,
       "step": 95
     },
     {
-      "epoch": 0.45454545454545453,
-      "grad_norm": 0.87109375,
-      "learning_rate": 0.00013270679633174218,
-      "loss": 1.3268,
+      "epoch": 0.026441036488630356,
+      "grad_norm": 0.380859375,
+      "learning_rate": 5.277044854881267e-05,
+      "loss": 1.6313,
       "step": 100
     },
     {
-      "epoch": 0.4772727272727273,
-      "grad_norm": 0.78515625,
-      "learning_rate": 0.0001251147987181079,
-      "loss": 1.307,
+      "epoch": 0.027763088313061873,
+      "grad_norm": 0.484375,
+      "learning_rate": 5.540897097625331e-05,
+      "loss": 1.6094,
       "step": 105
     },
     {
-      "epoch": 0.5,
-      "grad_norm": 1.03125,
-      "learning_rate": 0.00011736481776669306,
-      "loss": 1.3086,
+      "epoch": 0.02908514013749339,
+      "grad_norm": 0.5234375,
+      "learning_rate": 5.804749340369393e-05,
+      "loss": 1.5984,
       "step": 110
     },
     {
-      "epoch": 0.5227272727272727,
-      "grad_norm": 0.609375,
-      "learning_rate": 0.00010950560433041826,
-      "loss": 1.2977,
+      "epoch": 0.030407191961924908,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.068601583113457e-05,
+      "loss": 1.5843,
       "step": 115
     },
     {
-      "epoch": 0.5454545454545454,
-      "grad_norm": 0.65625,
-      "learning_rate": 0.00010158659638348081,
-      "loss": 1.3071,
+      "epoch": 0.03172924378635643,
+      "grad_norm": 0.322265625,
+      "learning_rate": 6.332453825857521e-05,
+      "loss": 1.5689,
       "step": 120
     },
     {
-      "epoch": 0.5681818181818182,
-      "grad_norm": 0.5859375,
-      "learning_rate": 9.365760803434355e-05,
-      "loss": 1.305,
+      "epoch": 0.03305129561078794,
+      "grad_norm": 0.44921875,
+      "learning_rate": 6.596306068601583e-05,
+      "loss": 1.5575,
       "step": 125
     },
     {
-      "epoch": 0.5909090909090909,
-      "grad_norm": 0.55859375,
-      "learning_rate": 8.57685161726715e-05,
-      "loss": 1.2902,
+      "epoch": 0.03437334743521946,
+      "grad_norm": 0.3203125,
+      "learning_rate": 6.860158311345646e-05,
+      "loss": 1.5451,
       "step": 130
     },
     {
-      "epoch": 0.6136363636363636,
-      "grad_norm": 0.51953125,
-      "learning_rate": 7.796894672134594e-05,
-      "loss": 1.281,
+      "epoch": 0.03569539925965098,
+      "grad_norm": 0.4140625,
+      "learning_rate": 7.124010554089711e-05,
+      "loss": 1.5333,
       "step": 135
     },
     {
-      "epoch": 0.6363636363636364,
-      "grad_norm": 0.66796875,
-      "learning_rate": 7.030796246717255e-05,
-      "loss": 1.296,
+      "epoch": 0.0370174510840825,
+      "grad_norm": 0.48828125,
+      "learning_rate": 7.387862796833773e-05,
+      "loss": 1.5237,
       "step": 140
     },
     {
-      "epoch": 0.6590909090909091,
-      "grad_norm": 0.6875,
-      "learning_rate": 6.283375443396726e-05,
-      "loss": 1.2883,
+      "epoch": 0.03833950290851401,
+      "grad_norm": 0.5,
+      "learning_rate": 7.651715039577836e-05,
+      "loss": 1.5105,
       "step": 145
     },
     {
-      "epoch": 0.6818181818181818,
-      "grad_norm": 0.62890625,
-      "learning_rate": 5.559333873942259e-05,
-      "loss": 1.2866,
+      "epoch": 0.03966155473294553,
+      "grad_norm": 0.3984375,
+      "learning_rate": 7.9155672823219e-05,
+      "loss": 1.4946,
       "step": 150
     },
     {
-      "epoch": 0.7045454545454546,
-      "grad_norm": 0.74609375,
-      "learning_rate": 4.8632260842659393e-05,
-      "loss": 1.2921,
+      "epoch": 0.040983606557377046,
+      "grad_norm": 0.39453125,
+      "learning_rate": 8.179419525065963e-05,
+      "loss": 1.4843,
       "step": 155
     },
     {
-      "epoch": 0.7272727272727273,
-      "grad_norm": 0.53125,
-      "learning_rate": 4.19943090428802e-05,
-      "loss": 1.2774,
+      "epoch": 0.04230565838180857,
+      "grad_norm": 0.3515625,
+      "learning_rate": 8.443271767810026e-05,
+      "loss": 1.4732,
       "step": 160
     },
     {
-      "epoch": 0.75,
-      "grad_norm": 0.458984375,
-      "learning_rate": 3.5721239031346066e-05,
-      "loss": 1.2547,
+      "epoch": 0.04362771020624009,
+      "grad_norm": 0.421875,
+      "learning_rate": 8.70712401055409e-05,
+      "loss": 1.4611,
       "step": 165
     },
     {
-      "epoch": 0.7727272727272727,
-      "grad_norm": 0.515625,
-      "learning_rate": 2.9852511229367865e-05,
-      "loss": 1.2814,
+      "epoch": 0.0449497620306716,
+      "grad_norm": 0.31640625,
+      "learning_rate": 8.970976253298153e-05,
+      "loss": 1.4485,
       "step": 170
     },
     {
-      "epoch": 0.7954545454545454,
-      "grad_norm": 0.453125,
-      "learning_rate": 2.4425042564574184e-05,
-      "loss": 1.2796,
+      "epoch": 0.04627181385510312,
+      "grad_norm": 0.341796875,
+      "learning_rate": 9.234828496042217e-05,
+      "loss": 1.4406,
       "step": 175
     },
     {
-      "epoch": 0.8181818181818182,
-      "grad_norm": 0.455078125,
-      "learning_rate": 1.947297424689414e-05,
-      "loss": 1.281,
+      "epoch": 0.047593865679534636,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.49868073878628e-05,
+      "loss": 1.4236,
       "step": 180
     },
     {
-      "epoch": 0.8409090909090909,
-      "grad_norm": 0.45703125,
-      "learning_rate": 1.5027457005048573e-05,
-      "loss": 1.2752,
+      "epoch": 0.04891591750396616,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.762532981530343e-05,
+      "loss": 1.4107,
       "step": 185
     },
     {
-      "epoch": 0.8636363636363636,
-      "grad_norm": 0.69140625,
-      "learning_rate": 1.1116455134507664e-05,
-      "loss": 1.2744,
+      "epoch": 0.05023796932839767,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00010026385224274407,
+      "loss": 1.4052,
       "step": 190
     },
     {
-      "epoch": 0.8863636363636364,
-      "grad_norm": 0.453125,
-      "learning_rate": 7.764570589541875e-06,
-      "loss": 1.2756,
+      "epoch": 0.05156002115282919,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00010290237467018471,
+      "loss": 1.3859,
       "step": 195
     },
     {
-      "epoch": 0.9090909090909091,
-      "grad_norm": 0.47265625,
-      "learning_rate": 4.992888225905468e-06,
-      "loss": 1.2816,
+      "epoch": 0.05288207297726071,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00010554089709762533,
+      "loss": 1.3742,
       "step": 200
     },
     {
-      "epoch": 0.9318181818181818,
-      "grad_norm": 0.55859375,
-      "learning_rate": 2.818843167645835e-06,
-      "loss": 1.278,
+      "epoch": 0.054204124801692226,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00010817941952506597,
+      "loss": 1.365,
       "step": 205
     },
     {
-      "epoch": 0.9545454545454546,
-      "grad_norm": 0.478515625,
-      "learning_rate": 1.2561111323605712e-06,
-      "loss": 1.2754,
+      "epoch": 0.05552617662612375,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00011081794195250662,
+      "loss": 1.3482,
       "step": 210
     },
     {
-      "epoch": 0.9772727272727273,
-      "grad_norm": 0.45703125,
-      "learning_rate": 3.145224048057727e-07,
-      "loss": 1.2759,
+      "epoch": 0.05684822845055526,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00011345646437994724,
+      "loss": 1.3375,
       "step": 215
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 0.478515625,
-      "learning_rate": 0.0,
-      "loss": 1.2702,
+      "epoch": 0.05817028027498678,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00011609498680738786,
+      "loss": 1.333,
       "step": 220
     },
+    {
+      "epoch": 0.059492332099418295,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00011873350923482852,
+      "loss": 1.3238,
+      "step": 225
+    },
+    {
+      "epoch": 0.060814383923849816,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00012137203166226914,
+      "loss": 1.315,
+      "step": 230
+    },
+    {
+      "epoch": 0.06213643574828133,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00012401055408970977,
+      "loss": 1.3028,
+      "step": 235
+    },
+    {
+      "epoch": 0.06345848757271286,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00012664907651715042,
+      "loss": 1.2869,
+      "step": 240
+    },
+    {
+      "epoch": 0.06478053939714437,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00012928759894459104,
+      "loss": 1.281,
+      "step": 245
+    },
+    {
+      "epoch": 0.06610259122157588,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00013192612137203166,
+      "loss": 1.2747,
+      "step": 250
+    },
+    {
+      "epoch": 0.0674246430460074,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0001345646437994723,
+      "loss": 1.2697,
+      "step": 255
+    },
+    {
+      "epoch": 0.06874669487043893,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.00013720316622691292,
+      "loss": 1.2549,
+      "step": 260
+    },
+    {
+      "epoch": 0.07006874669487044,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00013984168865435357,
+      "loss": 1.246,
+      "step": 265
+    },
+    {
+      "epoch": 0.07139079851930195,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00014248021108179422,
+      "loss": 1.2412,
+      "step": 270
+    },
+    {
+      "epoch": 0.07271285034373347,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00014511873350923484,
+      "loss": 1.2343,
+      "step": 275
+    },
+    {
+      "epoch": 0.074034902168165,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00014775725593667546,
+      "loss": 1.2205,
+      "step": 280
+    },
+    {
+      "epoch": 0.07535695399259651,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.00015039577836411608,
+      "loss": 1.2203,
+      "step": 285
+    },
+    {
+      "epoch": 0.07667900581702802,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00015303430079155673,
+      "loss": 1.2122,
+      "step": 290
+    },
+    {
+      "epoch": 0.07800105764145955,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.00015567282321899737,
+      "loss": 1.2035,
+      "step": 295
+    },
+    {
+      "epoch": 0.07932310946589106,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.000158311345646438,
+      "loss": 1.1933,
+      "step": 300
+    },
+    {
+      "epoch": 0.08064516129032258,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00016094986807387864,
+      "loss": 1.1926,
+      "step": 305
+    },
+    {
+      "epoch": 0.08196721311475409,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00016358839050131926,
+      "loss": 1.1833,
+      "step": 310
+    },
+    {
+      "epoch": 0.08328926493918562,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00016622691292875988,
+      "loss": 1.1753,
+      "step": 315
+    },
+    {
+      "epoch": 0.08461131676361713,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00016886543535620053,
+      "loss": 1.1717,
+      "step": 320
+    },
+    {
+      "epoch": 0.08593336858804865,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00017150395778364118,
+      "loss": 1.1684,
+      "step": 325
+    },
+    {
+      "epoch": 0.08725542041248018,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001741424802110818,
+      "loss": 1.1608,
+      "step": 330
+    },
+    {
+      "epoch": 0.08857747223691169,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00017678100263852244,
+      "loss": 1.1563,
+      "step": 335
+    },
+    {
+      "epoch": 0.0898995240613432,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00017941952506596306,
+      "loss": 1.1501,
+      "step": 340
+    },
+    {
+      "epoch": 0.09122157588577472,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00018205804749340368,
+      "loss": 1.1451,
+      "step": 345
+    },
+    {
+      "epoch": 0.09254362771020624,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00018469656992084433,
+      "loss": 1.1338,
+      "step": 350
+    },
+    {
+      "epoch": 0.09386567953463776,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00018733509234828498,
+      "loss": 1.1326,
+      "step": 355
+    },
+    {
+      "epoch": 0.09518773135906927,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001899736147757256,
+      "loss": 1.1286,
+      "step": 360
+    },
+    {
+      "epoch": 0.0965097831835008,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019261213720316625,
+      "loss": 1.1219,
+      "step": 365
+    },
+    {
+      "epoch": 0.09783183500793231,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019525065963060687,
+      "loss": 1.1238,
+      "step": 370
+    },
+    {
+      "epoch": 0.09915388683236383,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019788918205804749,
+      "loss": 1.1165,
+      "step": 375
+    },
+    {
+      "epoch": 0.10047593865679534,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00019999995738663592,
+      "loss": 1.1111,
+      "step": 380
+    },
+    {
+      "epoch": 0.10179799048122687,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00019999846592270624,
+      "loss": 1.0969,
+      "step": 385
+    },
+    {
+      "epoch": 0.10312004230565838,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00019999484382688995,
+      "loss": 1.1008,
+      "step": 390
+    },
+    {
+      "epoch": 0.1044420941300899,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019998909117636182,
+      "loss": 1.0969,
+      "step": 395
+    },
+    {
+      "epoch": 0.10576414595452142,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00019998120809369154,
+      "loss": 1.0888,
+      "step": 400
+    },
+    {
+      "epoch": 0.10708619777895294,
+      "grad_norm": 1.1875,
+      "learning_rate": 0.0001999711947468411,
+      "loss": 1.0926,
+      "step": 405
+    },
+    {
+      "epoch": 0.10840824960338445,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00019995905134916143,
+      "loss": 1.08,
+      "step": 410
+    },
+    {
+      "epoch": 0.10973030142781597,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001999447781593875,
+      "loss": 1.0778,
+      "step": 415
+    },
+    {
+      "epoch": 0.1110523532522475,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00019992837548163316,
+      "loss": 1.0771,
+      "step": 420
+    },
+    {
+      "epoch": 0.11237440507667901,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00019990984366538442,
+      "loss": 1.0722,
+      "step": 425
+    },
+    {
+      "epoch": 0.11369645690111052,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00019988918310549222,
+      "loss": 1.0628,
+      "step": 430
+    },
+    {
+      "epoch": 0.11501850872554203,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001998663942421637,
+      "loss": 1.0611,
+      "step": 435
+    },
+    {
+      "epoch": 0.11634056054997356,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019984147756095308,
+      "loss": 1.0627,
+      "step": 440
+    },
+    {
+      "epoch": 0.11766261237440508,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001998144335927513,
+      "loss": 1.0562,
+      "step": 445
+    },
+    {
+      "epoch": 0.11898466419883659,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019978526291377464,
+      "loss": 1.0543,
+      "step": 450
+    },
+    {
+      "epoch": 0.12030671602326812,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001997539661455524,
+      "loss": 1.0532,
+      "step": 455
+    },
+    {
+      "epoch": 0.12162876784769963,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019972054395491368,
+      "loss": 1.0477,
+      "step": 460
+    },
+    {
+      "epoch": 0.12295081967213115,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00019968499705397331,
+      "loss": 1.044,
+      "step": 465
+    },
+    {
+      "epoch": 0.12427287149656266,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00019964732620011651,
+      "loss": 1.0451,
+      "step": 470
+    },
+    {
+      "epoch": 0.12559492332099417,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00019960753219598281,
+      "loss": 1.0385,
+      "step": 475
+    },
+    {
+      "epoch": 0.12691697514542571,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00019956561588944897,
+      "loss": 1.0376,
+      "step": 480
+    },
+    {
+      "epoch": 0.12823902696985723,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00019952157817361088,
+      "loss": 1.0396,
+      "step": 485
+    },
+    {
+      "epoch": 0.12956107879428874,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001994754199867645,
+      "loss": 1.0342,
+      "step": 490
+    },
+    {
+      "epoch": 0.13088313061872026,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019942714231238604,
+      "loss": 1.036,
+      "step": 495
+    },
+    {
+      "epoch": 0.13220518244315177,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00019937674617911077,
+      "loss": 1.0231,
+      "step": 500
+    },
+    {
+      "epoch": 0.13352723426758328,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00019932423266071122,
+      "loss": 1.0286,
+      "step": 505
+    },
+    {
+      "epoch": 0.1348492860920148,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00019926960287607436,
+      "loss": 1.0238,
+      "step": 510
+    },
+    {
+      "epoch": 0.1361713379164463,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.0001992128579891777,
+      "loss": 1.0216,
+      "step": 515
+    },
+    {
+      "epoch": 0.13749338974087785,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00019915399920906432,
+      "loss": 1.0237,
+      "step": 520
+    },
+    {
+      "epoch": 0.13881544156530937,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001990930277898175,
+      "loss": 1.0137,
+      "step": 525
+    },
+    {
+      "epoch": 0.14013749338974088,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00019902994503053363,
+      "loss": 1.0206,
+      "step": 530
+    },
+    {
+      "epoch": 0.1414595452141724,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001989647522752947,
+      "loss": 1.0189,
+      "step": 535
+    },
+    {
+      "epoch": 0.1427815970386039,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.0001988974509131397,
+      "loss": 1.0109,
+      "step": 540
+    },
+    {
+      "epoch": 0.14410364886303542,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.00019882804237803488,
+      "loss": 1.0069,
+      "step": 545
+    },
+    {
+      "epoch": 0.14542570068746694,
+      "grad_norm": 0.224609375,
+      "learning_rate": 0.00019875652814884326,
+      "loss": 1.0097,
+      "step": 550
+    },
+    {
+      "epoch": 0.14674775251189848,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00019868290974929328,
+      "loss": 1.0033,
+      "step": 555
+    },
+    {
+      "epoch": 0.14806980433633,
+      "grad_norm": 0.8671875,
+      "learning_rate": 0.00019860718874794602,
+      "loss": 1.0034,
+      "step": 560
+    },
+    {
+      "epoch": 0.1493918561607615,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019852936675816209,
+      "loss": 1.0019,
+      "step": 565
+    },
+    {
+      "epoch": 0.15071390798519302,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.000198449445438067,
+      "loss": 1.0004,
+      "step": 570
+    },
+    {
+      "epoch": 0.15203595980962453,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019836742649051602,
+      "loss": 0.998,
+      "step": 575
+    },
+    {
+      "epoch": 0.15335801163405605,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019828331166305785,
+      "loss": 0.9944,
+      "step": 580
+    },
+    {
+      "epoch": 0.15468006345848756,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00019819710274789727,
+      "loss": 0.9951,
+      "step": 585
+    },
+    {
+      "epoch": 0.1560021152829191,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001981088015818571,
+      "loss": 0.9962,
+      "step": 590
+    },
+    {
+      "epoch": 0.15732416710735062,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019801841004633906,
+      "loss": 0.9884,
+      "step": 595
+    },
+    {
+      "epoch": 0.15864621893178213,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00019792593006728347,
+      "loss": 0.9918,
+      "step": 600
+    },
+    {
+      "epoch": 0.15996827075621364,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00019783136361512858,
+      "loss": 0.993,
+      "step": 605
+    },
+    {
+      "epoch": 0.16129032258064516,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019773471270476822,
+      "loss": 0.9902,
+      "step": 610
+    },
+    {
+      "epoch": 0.16261237440507667,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001976359793955091,
+      "loss": 0.9843,
+      "step": 615
+    },
+    {
+      "epoch": 0.16393442622950818,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.0001975351657910269,
+      "loss": 0.988,
+      "step": 620
+    },
+    {
+      "epoch": 0.16525647805393973,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019743227403932134,
+      "loss": 0.9874,
+      "step": 625
+    },
+    {
+      "epoch": 0.16657852987837124,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019732730633267056,
+      "loss": 0.9782,
+      "step": 630
+    },
+    {
+      "epoch": 0.16790058170280275,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019722026490758424,
+      "loss": 0.9813,
+      "step": 635
+    },
+    {
+      "epoch": 0.16922263352723427,
+      "grad_norm": 0.412109375,
+      "learning_rate": 0.00019711115204475616,
+      "loss": 0.9793,
+      "step": 640
+    },
+    {
+      "epoch": 0.17054468535166578,
+      "grad_norm": 0.36328125,
+      "learning_rate": 0.0001969999700690154,
+      "loss": 0.9749,
+      "step": 645
+    },
+    {
+      "epoch": 0.1718667371760973,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00019688672134927693,
+      "loss": 0.9768,
+      "step": 650
+    },
+    {
+      "epoch": 0.1731887890005288,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00019677140829849112,
+      "loss": 0.973,
+      "step": 655
+    },
+    {
+      "epoch": 0.17451084082496035,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00019665403337359227,
+      "loss": 0.972,
+      "step": 660
+    },
+    {
+      "epoch": 0.17583289264939186,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00019653459907544634,
+      "loss": 0.9759,
+      "step": 665
+    },
+    {
+      "epoch": 0.17715494447382338,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.0001964131079487976,
+      "loss": 0.9691,
+      "step": 670
+    },
+    {
+      "epoch": 0.1784769962982549,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.0001962895625822144,
+      "loss": 0.9681,
+      "step": 675
+    },
+    {
+      "epoch": 0.1797990481226864,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0001961639656080342,
+      "loss": 0.9695,
+      "step": 680
+    },
+    {
+      "epoch": 0.18112109994711792,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00019603631970230713,
+      "loss": 0.9695,
+      "step": 685
+    },
+    {
+      "epoch": 0.18244315177154943,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00019590662758473934,
+      "loss": 0.9685,
+      "step": 690
+    },
+    {
+      "epoch": 0.18376520359598097,
+      "grad_norm": 0.37109375,
+      "learning_rate": 0.0001957748920186348,
+      "loss": 0.9681,
+      "step": 695
+    },
+    {
+      "epoch": 0.1850872554204125,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00019564111581083657,
+      "loss": 0.9632,
+      "step": 700
+    },
+    {
+      "epoch": 0.186409307244844,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019550530181166692,
+      "loss": 0.9683,
+      "step": 705
+    },
+    {
+      "epoch": 0.18773135906927552,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.0001953674529148666,
+      "loss": 0.9628,
+      "step": 710
+    },
+    {
+      "epoch": 0.18905341089370703,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00019522757205753323,
+      "loss": 0.9625,
+      "step": 715
+    },
+    {
+      "epoch": 0.19037546271813854,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00019508566222005866,
+      "loss": 0.9583,
+      "step": 720
+    },
+    {
+      "epoch": 0.19169751454257006,
+      "grad_norm": 0.330078125,
+      "learning_rate": 0.00019494172642606553,
+      "loss": 0.9634,
+      "step": 725
+    },
+    {
+      "epoch": 0.1930195663670016,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00019479576774234284,
+      "loss": 0.9602,
+      "step": 730
+    },
+    {
+      "epoch": 0.1943416181914331,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00019464778927878048,
+      "loss": 0.967,
+      "step": 735
+    },
+    {
+      "epoch": 0.19566367001586463,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00019449779418830322,
+      "loss": 0.9549,
+      "step": 740
+    },
+    {
+      "epoch": 0.19698572184029614,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.0001943457856668033,
+      "loss": 0.9652,
+      "step": 745
+    },
+    {
+      "epoch": 0.19830777366472765,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00019419176695307245,
+      "loss": 0.9541,
+      "step": 750
+    },
+    {
+      "epoch": 0.19962982548915917,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00019403574132873276,
+      "loss": 0.9506,
+      "step": 755
+    },
+    {
+      "epoch": 0.20095187731359068,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00019387771211816705,
+      "loss": 0.9572,
+      "step": 760
+    },
+    {
+      "epoch": 0.20227392913802222,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00019371768268844762,
+      "loss": 0.9525,
+      "step": 765
+    },
+    {
+      "epoch": 0.20359598096245374,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001935556564492649,
+      "loss": 0.9542,
+      "step": 770
+    },
+    {
+      "epoch": 0.20491803278688525,
+      "grad_norm": 0.388671875,
+      "learning_rate": 0.0001933916368528545,
+      "loss": 0.9532,
+      "step": 775
+    },
+    {
+      "epoch": 0.20624008461131677,
+      "grad_norm": 0.318359375,
+      "learning_rate": 0.00019322562739392394,
+      "loss": 0.9511,
+      "step": 780
+    },
+    {
+      "epoch": 0.20756213643574828,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00019305763160957788,
+      "loss": 0.9536,
+      "step": 785
+    },
+    {
+      "epoch": 0.2088841882601798,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00019288765307924299,
+      "loss": 0.9511,
+      "step": 790
+    },
+    {
+      "epoch": 0.2102062400846113,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00019271569542459165,
+      "loss": 0.9497,
+      "step": 795
+    },
+    {
+      "epoch": 0.21152829190904285,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00019254176230946462,
+      "loss": 0.9484,
+      "step": 800
+    },
+    {
+      "epoch": 0.21285034373347436,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00019236585743979322,
+      "loss": 0.9442,
+      "step": 805
+    },
+    {
+      "epoch": 0.21417239555790588,
+      "grad_norm": 0.3671875,
+      "learning_rate": 0.00019218798456352012,
+      "loss": 0.9426,
+      "step": 810
+    },
+    {
+      "epoch": 0.2154944473823374,
+      "grad_norm": 0.40234375,
+      "learning_rate": 0.00019200814747051976,
+      "loss": 0.9455,
+      "step": 815
+    },
+    {
+      "epoch": 0.2168164992067689,
+      "grad_norm": 0.349609375,
+      "learning_rate": 0.00019182634999251728,
+      "loss": 0.9419,
+      "step": 820
+    },
+    {
+      "epoch": 0.21813855103120042,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00019164259600300723,
+      "loss": 0.947,
+      "step": 825
+    },
+    {
+      "epoch": 0.21946060285563193,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00019145688941717075,
+      "loss": 0.9478,
+      "step": 830
+    },
+    {
+      "epoch": 0.22078265468006344,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.0001912692341917922,
+      "loss": 0.9433,
+      "step": 835
+    },
+    {
+      "epoch": 0.222104706504495,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00019107963432517505,
+      "loss": 0.9461,
+      "step": 840
+    },
+    {
+      "epoch": 0.2234267583289265,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00019088809385705646,
+      "loss": 0.9421,
+      "step": 845
+    },
+    {
+      "epoch": 0.22474881015335801,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00019069461686852134,
+      "loss": 0.9436,
+      "step": 850
+    },
+    {
+      "epoch": 0.22607086197778953,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.0001904992074819153,
+      "loss": 0.9381,
+      "step": 855
+    },
+    {
+      "epoch": 0.22739291380222104,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00019030186986075703,
+      "loss": 0.9408,
+      "step": 860
+    },
+    {
+      "epoch": 0.22871496562665256,
+      "grad_norm": 0.3515625,
+      "learning_rate": 0.0001901026082096492,
+      "loss": 0.9371,
+      "step": 865
+    },
+    {
+      "epoch": 0.23003701745108407,
+      "grad_norm": 0.404296875,
+      "learning_rate": 0.00018990142677418923,
+      "loss": 0.9387,
+      "step": 870
+    },
+    {
+      "epoch": 0.2313590692755156,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00018969832984087873,
+      "loss": 0.9372,
+      "step": 875
+    },
+    {
+      "epoch": 0.23268112109994712,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00018949332173703206,
+      "loss": 0.936,
+      "step": 880
+    },
+    {
+      "epoch": 0.23400317292437864,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0001892864068306843,
+      "loss": 0.9405,
+      "step": 885
+    },
+    {
+      "epoch": 0.23532522474881015,
+      "grad_norm": 0.373046875,
+      "learning_rate": 0.00018907758953049805,
+      "loss": 0.9333,
+      "step": 890
+    },
+    {
+      "epoch": 0.23664727657324167,
+      "grad_norm": 0.3984375,
+      "learning_rate": 0.00018886687428566954,
+      "loss": 0.9356,
+      "step": 895
+    },
+    {
+      "epoch": 0.23796932839767318,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00018865426558583383,
+      "loss": 0.9344,
+      "step": 900
+    },
+    {
+      "epoch": 0.2392913802221047,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00018843976796096917,
+      "loss": 0.9355,
+      "step": 905
+    },
+    {
+      "epoch": 0.24061343204653624,
+      "grad_norm": 0.353515625,
+      "learning_rate": 0.00018822338598130047,
+      "loss": 0.9336,
+      "step": 910
+    },
+    {
+      "epoch": 0.24193548387096775,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.0001880051242572019,
+      "loss": 0.9304,
+      "step": 915
+    },
+    {
+      "epoch": 0.24325753569539926,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00018778498743909873,
+      "loss": 0.931,
+      "step": 920
+    },
+    {
+      "epoch": 0.24457958751983078,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00018756298021736808,
+      "loss": 0.934,
+      "step": 925
+    },
+    {
+      "epoch": 0.2459016393442623,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00018733910732223925,
+      "loss": 0.9273,
+      "step": 930
+    },
+    {
+      "epoch": 0.2472236911686938,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00018711337352369264,
+      "loss": 0.9315,
+      "step": 935
+    },
+    {
+      "epoch": 0.24854574299312532,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001868857836313584,
+      "loss": 0.9316,
+      "step": 940
+    },
+    {
+      "epoch": 0.24986779481755686,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00018665634249441366,
+      "loss": 0.9319,
+      "step": 945
+    },
+    {
+      "epoch": 0.25118984664198835,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001864250550014795,
+      "loss": 0.9297,
+      "step": 950
+    },
+    {
+      "epoch": 0.25251189846641986,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00018619192608051655,
+      "loss": 0.9237,
+      "step": 955
+    },
+    {
+      "epoch": 0.25383395029085143,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00018595696069872013,
+      "loss": 0.9293,
+      "step": 960
+    },
+    {
+      "epoch": 0.25515600211528294,
+      "grad_norm": 0.279296875,
+      "learning_rate": 0.00018572016386241442,
+      "loss": 0.9308,
+      "step": 965
+    },
+    {
+      "epoch": 0.25647805393971446,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00018548154061694572,
+      "loss": 0.9207,
+      "step": 970
+    },
+    {
+      "epoch": 0.25780010576414597,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00018524109604657496,
+      "loss": 0.9285,
+      "step": 975
+    },
+    {
+      "epoch": 0.2591221575885775,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00018499883527436947,
+      "loss": 0.9268,
+      "step": 980
+    },
+    {
+      "epoch": 0.260444209413009,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0001847547634620936,
+      "loss": 0.9284,
+      "step": 985
+    },
+    {
+      "epoch": 0.2617662612374405,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00018450888581009908,
+      "loss": 0.9264,
+      "step": 990
+    },
+    {
+      "epoch": 0.263088313061872,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00018426120755721386,
+      "loss": 0.926,
+      "step": 995
+    },
+    {
+      "epoch": 0.26441036488630354,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.0001840117339806308,
+      "loss": 0.9218,
+      "step": 1000
+    },
+    {
+      "epoch": 0.26573241671073505,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00018376047039579495,
+      "loss": 0.9245,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26705446853516657,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00018350742215629055,
+      "loss": 0.9221,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2683765203595981,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00018325259465372677,
+      "loss": 0.926,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2696985721840296,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.0001829959933176229,
+      "loss": 0.9235,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2710206240084611,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00018273762361529274,
+      "loss": 0.9237,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2723426758328926,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.000182477491051728,
+      "loss": 0.9203,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2736647276573242,
+      "grad_norm": 0.396484375,
+      "learning_rate": 0.00018221560116948103,
+      "loss": 0.9206,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2749867794817557,
+      "grad_norm": 0.345703125,
+      "learning_rate": 0.00018195195954854676,
+      "loss": 0.924,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2763088313061872,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00018168657180624384,
+      "loss": 0.918,
+      "step": 1045
+    },
+    {
+      "epoch": 0.27763088313061873,
+      "grad_norm": 0.337890625,
+      "learning_rate": 0.0001814194435970949,
+      "loss": 0.9298,
+      "step": 1050
+    },
+    {
+      "epoch": 0.27895293495505025,
+      "grad_norm": 0.7421875,
+      "learning_rate": 0.00018115058061270598,
+      "loss": 0.9195,
+      "step": 1055
+    },
+    {
+      "epoch": 0.28027498677948176,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001808799885816455,
+      "loss": 0.9196,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2815970386039133,
+      "grad_norm": 0.39453125,
+      "learning_rate": 0.00018060767326932194,
+      "loss": 0.9154,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2829190904283448,
+      "grad_norm": 0.38671875,
+      "learning_rate": 0.00018033364047786128,
+      "loss": 0.9186,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2842411422527763,
+      "grad_norm": 0.400390625,
+      "learning_rate": 0.00018005789604598303,
+      "loss": 0.9173,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2855631940772078,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.0001797804458488762,
+      "loss": 0.9199,
+      "step": 1080
+    },
+    {
+      "epoch": 0.28688524590163933,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00017950129579807374,
+      "loss": 0.9202,
+      "step": 1085
+    },
+    {
+      "epoch": 0.28820729772607084,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00017922045184132698,
+      "loss": 0.922,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28952934955050236,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00017893791996247856,
+      "loss": 0.9157,
+      "step": 1095
+    },
+    {
+      "epoch": 0.29085140137493387,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00017865370618133511,
+      "loss": 0.9145,
+      "step": 1100
+    },
+    {
+      "epoch": 0.29217345319936544,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00017836781655353905,
+      "loss": 0.9114,
+      "step": 1105
+    },
+    {
+      "epoch": 0.29349550502379695,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00017808025717043938,
+      "loss": 0.9128,
+      "step": 1110
+    },
+    {
+      "epoch": 0.29481755684822847,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00017779103415896193,
+      "loss": 0.9182,
+      "step": 1115
+    },
+    {
+      "epoch": 0.29613960867266,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.000177500153681479,
+      "loss": 0.9127,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2974616604970915,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00017720762193567787,
+      "loss": 0.9122,
+      "step": 1125
+    },
+    {
+      "epoch": 0.298783712321523,
+      "grad_norm": 0.255859375,
+      "learning_rate": 0.00017691344515442879,
+      "loss": 0.9117,
+      "step": 1130
+    },
+    {
+      "epoch": 0.3001057641459545,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00017661762960565223,
+      "loss": 0.9097,
+      "step": 1135
+    },
+    {
+      "epoch": 0.30142781597038604,
+      "grad_norm": 0.376953125,
+      "learning_rate": 0.0001763201815921853,
+      "loss": 0.9085,
+      "step": 1140
+    },
+    {
+      "epoch": 0.30274986779481755,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.0001760211074516474,
+      "loss": 0.915,
+      "step": 1145
+    },
+    {
+      "epoch": 0.30407191961924906,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.00017572041355630536,
+      "loss": 0.9145,
+      "step": 1150
+    },
+    {
+      "epoch": 0.3053939714436806,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 0.00017541810631293742,
+      "loss": 0.9144,
+      "step": 1155
+    },
+    {
+      "epoch": 0.3067160232681121,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00017511419216269695,
+      "loss": 0.9129,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3080380750925436,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 0.00017480867758097506,
+      "loss": 0.9126,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3093601269169751,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.00017450156907726272,
+      "loss": 0.9139,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3106821787414067,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00017419287319501197,
+      "loss": 0.9064,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3120042305658382,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.00017388259651149673,
+      "loss": 0.9165,
+      "step": 1180
+    },
+    {
+      "epoch": 0.3133262823902697,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00017357074563767225,
+      "loss": 0.9102,
+      "step": 1185
+    },
+    {
+      "epoch": 0.31464833421470123,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00017325732721803466,
+      "loss": 0.9102,
+      "step": 1190
+    },
+    {
+      "epoch": 0.31597038603913274,
+      "grad_norm": 0.302734375,
+      "learning_rate": 0.0001729423479304792,
+      "loss": 0.9092,
+      "step": 1195
+    },
+    {
+      "epoch": 0.31729243786356426,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.00017262581448615794,
+      "loss": 0.9076,
+      "step": 1200
+    },
+    {
+      "epoch": 0.31861448968799577,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00017230773362933687,
+      "loss": 0.9073,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3199365415124273,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.0001719881121372521,
+      "loss": 0.9074,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3212585933368588,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.00017166695681996555,
+      "loss": 0.912,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3225806451612903,
+      "grad_norm": 0.361328125,
+      "learning_rate": 0.00017134427452021974,
+      "loss": 0.9061,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3239026969857218,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00017102007211329214,
+      "loss": 0.9084,
+      "step": 1225
+    },
+    {
+      "epoch": 0.32522474881015334,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.0001706943565068486,
+      "loss": 0.9091,
+      "step": 1230
+    },
+    {
+      "epoch": 0.32654680063458486,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 0.00017036713464079612,
+      "loss": 0.9092,
+      "step": 1235
+    },
+    {
+      "epoch": 0.32786885245901637,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.0001700384134871351,
+      "loss": 0.9069,
+      "step": 1240
+    },
+    {
+      "epoch": 0.32919090428344794,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.00016970820004981067,
+      "loss": 0.9082,
+      "step": 1245
+    },
+    {
+      "epoch": 0.33051295610787945,
+      "grad_norm": 0.33203125,
+      "learning_rate": 0.00016937650136456358,
+      "loss": 0.9086,
+      "step": 1250
+    },
+    {
+      "epoch": 0.33183500793231097,
+      "grad_norm": 0.1875,
+      "learning_rate": 0.0001690433244987802,
+      "loss": 0.9041,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3331570597567425,
+      "grad_norm": 0.2578125,
+      "learning_rate": 0.000168708676551342,
+      "loss": 0.9098,
+      "step": 1260
+    },
+    {
+      "epoch": 0.334479111581174,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00016837256465247418,
+      "loss": 0.9024,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3358011634056055,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 0.00016803499596359392,
+      "loss": 0.9061,
+      "step": 1270
+    },
+    {
+      "epoch": 0.337123215230037,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.0001676959776771577,
+      "loss": 0.9094,
+      "step": 1275
+    },
+    {
+      "epoch": 0.33844526705446853,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00016735551701650803,
+      "loss": 0.9035,
+      "step": 1280
+    },
+    {
+      "epoch": 0.33976731887890005,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00016701362123571959,
+      "loss": 0.9071,
+      "step": 1285
+    },
+    {
+      "epoch": 0.34108937070333156,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00016667029761944466,
+      "loss": 0.9064,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3424114225277631,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 0.00016632555348275788,
+      "loss": 0.9043,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3437334743521946,
+      "grad_norm": 0.32421875,
+      "learning_rate": 0.00016597939617100046,
+      "loss": 0.9039,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3450555261766261,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00016563183305962363,
+      "loss": 0.9042,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3463775780010576,
+      "grad_norm": 0.310546875,
+      "learning_rate": 0.0001652828715540314,
+      "loss": 0.9026,
+      "step": 1310
+    },
+    {
+      "epoch": 0.34769962982548913,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00016493251908942302,
+      "loss": 0.9043,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3490216816499207,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001645807831306343,
+      "loss": 0.9035,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3503437334743522,
+      "grad_norm": 0.384765625,
+      "learning_rate": 0.00016422767117197867,
+      "loss": 0.9016,
+      "step": 1325
+    },
+    {
+      "epoch": 0.35166578529878373,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001638731907370876,
+      "loss": 0.8989,
+      "step": 1330
+    },
+    {
+      "epoch": 0.35298783712321524,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00016351734937875007,
+      "loss": 0.8979,
+      "step": 1335
+    },
+    {
+      "epoch": 0.35430988894764676,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00016316015467875188,
+      "loss": 0.904,
+      "step": 1340
+    },
+    {
+      "epoch": 0.35563194077207827,
+      "grad_norm": 0.326171875,
+      "learning_rate": 0.00016280161424771396,
+      "loss": 0.8991,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3569539925965098,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0001624417357249302,
+      "loss": 0.901,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3582760444209413,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00016208052677820484,
+      "loss": 0.904,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3595980962453728,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.00016171799510368895,
+      "loss": 0.8977,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3609201480698043,
+      "grad_norm": 0.296875,
+      "learning_rate": 0.00016135414842571643,
+      "loss": 0.9002,
+      "step": 1365
+    },
+    {
+      "epoch": 0.36224219989423584,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0001609889944966396,
+      "loss": 0.8996,
+      "step": 1370
+    },
+    {
+      "epoch": 0.36356425171866735,
+      "grad_norm": 0.27734375,
+      "learning_rate": 0.0001606225410966638,
+      "loss": 0.9047,
+      "step": 1375
+    },
+    {
+      "epoch": 0.36488630354309887,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001602547960336819,
+      "loss": 0.9004,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3662083553675304,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.00015988576714310755,
+      "loss": 0.902,
+      "step": 1385
+    },
+    {
+      "epoch": 0.36753040719196195,
+      "grad_norm": 0.25,
+      "learning_rate": 0.00015951546228770868,
+      "loss": 0.8999,
+      "step": 1390
+    },
+    {
+      "epoch": 0.36885245901639346,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.0001591438893574396,
+      "loss": 0.9047,
+      "step": 1395
+    },
+    {
+      "epoch": 0.370174510840825,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.0001587710562692731,
+      "loss": 0.8982,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3714965626652565,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00015839697096703176,
+      "loss": 0.9009,
+      "step": 1405
+    },
+    {
+      "epoch": 0.372818614489688,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.00015802164142121854,
+      "loss": 0.9023,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3741406663141195,
+      "grad_norm": 0.392578125,
+      "learning_rate": 0.0001576450756288471,
+      "loss": 0.9018,
+      "step": 1415
+    },
+    {
+      "epoch": 0.37546271813855103,
+      "grad_norm": 0.341796875,
+      "learning_rate": 0.0001572672816132714,
+      "loss": 0.9006,
+      "step": 1420
+    },
+    {
+      "epoch": 0.37678476996298255,
+      "grad_norm": 0.212890625,
+      "learning_rate": 0.00015688826742401465,
+      "loss": 0.8996,
+      "step": 1425
+    },
+    {
+      "epoch": 0.37810682178741406,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00015650804113659793,
+      "loss": 0.8957,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3794288736118456,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.00015612661085236807,
+      "loss": 0.901,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3807509254362771,
+      "grad_norm": 0.2265625,
+      "learning_rate": 0.00015574398469832493,
+      "loss": 0.8984,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3820729772607086,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00015536017082694846,
+      "loss": 0.8983,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3833950290851401,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 0.00015497517741602486,
+      "loss": 0.8992,
+      "step": 1450
+    },
+    {
+      "epoch": 0.38471708090957163,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00015458901266847232,
+      "loss": 0.8988,
+      "step": 1455
+    },
+    {
+      "epoch": 0.3860391327340032,
+      "grad_norm": 0.369140625,
+      "learning_rate": 0.0001542016848121663,
+      "loss": 0.8982,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3873611845584347,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00015381320209976426,
+      "loss": 0.9009,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3886832363828662,
+      "grad_norm": 0.3203125,
+      "learning_rate": 0.0001534235728085297,
+      "loss": 0.9006,
+      "step": 1470
+    },
+    {
+      "epoch": 0.39000528820729774,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.00015303280524015585,
+      "loss": 0.9001,
+      "step": 1475
+    },
+    {
+      "epoch": 0.39132734003172925,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0001526409077205889,
+      "loss": 0.894,
+      "step": 1480
+    },
+    {
+      "epoch": 0.39264939185616077,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.00015224788859985043,
+      "loss": 0.8955,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3939714436805923,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00015185375625185964,
+      "loss": 0.902,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3952934955050238,
+      "grad_norm": 0.322265625,
+      "learning_rate": 0.00015145851907425484,
+      "loss": 0.8995,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3966155473294553,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 0.0001510621854882145,
+      "loss": 0.8997,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3979375991538868,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 0.00015066476393827799,
+      "loss": 0.8915,
+      "step": 1505
+    },
+    {
+      "epoch": 0.39925965097831834,
+      "grad_norm": 0.2421875,
+      "learning_rate": 0.00015026626289216542,
+      "loss": 0.8916,
+      "step": 1510
+    },
+    {
+      "epoch": 0.40058170280274985,
+      "grad_norm": 0.3125,
+      "learning_rate": 0.0001498666908405975,
+      "loss": 0.8961,
+      "step": 1515
+    },
+    {
+      "epoch": 0.40190375462718136,
+      "grad_norm": 0.3359375,
+      "learning_rate": 0.00014946605629711425,
+      "loss": 0.8991,
+      "step": 1520
+    },
+    {
+      "epoch": 0.4032258064516129,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00014906436779789402,
+      "loss": 0.8932,
+      "step": 1525
+    },
+    {
+      "epoch": 0.40454785827604445,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00014866163390157136,
+      "loss": 0.8952,
+      "step": 1530
+    },
+    {
+      "epoch": 0.40586991010047596,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00014825786318905469,
+      "loss": 0.893,
+      "step": 1535
+    },
+    {
+      "epoch": 0.4071919619249075,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00014785306426334354,
+      "loss": 0.8961,
+      "step": 1540
+    },
+    {
+      "epoch": 0.408514013749339,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.0001474472457493452,
+      "loss": 0.8916,
+      "step": 1545
+    },
+    {
+      "epoch": 0.4098360655737705,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00014704041629369106,
+      "loss": 0.9021,
+      "step": 1550
+    },
+    {
+      "epoch": 0.411158117398202,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00014663258456455212,
+      "loss": 0.8992,
+      "step": 1555
+    },
+    {
+      "epoch": 0.41248016922263353,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00014622375925145455,
+      "loss": 0.8926,
+      "step": 1560
+    },
+    {
+      "epoch": 0.41380222104706504,
+      "grad_norm": 0.25,
+      "learning_rate": 0.0001458139490650945,
+      "loss": 0.894,
+      "step": 1565
+    },
+    {
+      "epoch": 0.41512427287149656,
+      "grad_norm": 0.33984375,
+      "learning_rate": 0.0001454031627371524,
+      "loss": 0.8948,
+      "step": 1570
+    },
+    {
+      "epoch": 0.41644632469592807,
+      "grad_norm": 0.26953125,
+      "learning_rate": 0.000144991409020107,
+      "loss": 0.8933,
+      "step": 1575
+    },
+    {
+      "epoch": 0.4177683765203596,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00014457869668704895,
+      "loss": 0.9002,
+      "step": 1580
+    },
+    {
+      "epoch": 0.4190904283447911,
+      "grad_norm": 0.201171875,
+      "learning_rate": 0.0001441650345314936,
+      "loss": 0.8961,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4204124801692226,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00014375043136719402,
+      "loss": 0.8939,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4217345319936541,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 0.00014333489602795293,
+      "loss": 0.8915,
+      "step": 1595
+    },
+    {
+      "epoch": 0.4230565838180857,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.0001429184373674346,
+      "loss": 0.8952,
+      "step": 1600
+    },
+    {
+      "epoch": 0.4243786356425172,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 0.0001425010642589762,
+      "loss": 0.8931,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4257006874669487,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.0001420827855953986,
+      "loss": 0.891,
+      "step": 1610
+    },
+    {
+      "epoch": 0.42702273929138024,
+      "grad_norm": 0.267578125,
+      "learning_rate": 0.00014166361028881716,
+      "loss": 0.891,
+      "step": 1615
+    },
+    {
+      "epoch": 0.42834479111581175,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00014124354727045163,
+      "loss": 0.8917,
+      "step": 1620
+    },
+    {
+      "epoch": 0.42966684294024327,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.00014082260549043592,
+      "loss": 0.8897,
+      "step": 1625
+    },
+    {
+      "epoch": 0.4309888947646748,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.0001404007939176274,
+      "loss": 0.895,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4323109465891063,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 0.00013997812153941584,
+      "loss": 0.8965,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4336329984135378,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00013955459736153183,
+      "loss": 0.8954,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4349550502379693,
+      "grad_norm": 0.234375,
+      "learning_rate": 0.00013913023040785502,
+      "loss": 0.8982,
+      "step": 1645
+    },
+    {
+      "epoch": 0.43627710206240083,
+      "grad_norm": 0.30859375,
+      "learning_rate": 0.00013870502972022173,
+      "loss": 0.8906,
+      "step": 1650
+    },
+    {
+      "epoch": 0.43759915388683235,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00013827900435823236,
+      "loss": 0.8945,
+      "step": 1655
+    },
+    {
+      "epoch": 0.43892120571126386,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.00013785216339905838,
+      "loss": 0.8904,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4402432575356954,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.0001374245159372489,
+      "loss": 0.8924,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4415653093601269,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00013699607108453684,
+      "loss": 0.8855,
+      "step": 1670
+    },
+    {
+      "epoch": 0.44288736118455846,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.00013656683796964495,
+      "loss": 0.8914,
+      "step": 1675
+    },
+    {
+      "epoch": 0.44420941300899,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00013613682573809113,
+      "loss": 0.8931,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4455314648334215,
+      "grad_norm": 0.359375,
+      "learning_rate": 0.0001357060435519936,
+      "loss": 0.8926,
+      "step": 1685
+    },
+    {
+      "epoch": 0.446853516657853,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00013527450058987585,
+      "loss": 0.8915,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4481755684822845,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0001348422060464709,
+      "loss": 0.8878,
+      "step": 1695
+    },
+    {
+      "epoch": 0.44949762030671603,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.00013440916913252536,
+      "loss": 0.89,
+      "step": 1700
+    },
+    {
+      "epoch": 0.45081967213114754,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 0.0001339753990746034,
+      "loss": 0.8881,
+      "step": 1705
+    },
+    {
+      "epoch": 0.45214172395557906,
+      "grad_norm": 0.287109375,
+      "learning_rate": 0.00013354090511488994,
+      "loss": 0.8911,
+      "step": 1710
+    },
+    {
+      "epoch": 0.45346377578001057,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00013310569651099388,
+      "loss": 0.8904,
+      "step": 1715
+    },
+    {
+      "epoch": 0.4547858276044421,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00013266978253575074,
+      "loss": 0.8902,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4561078794288736,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00013223317247702517,
+      "loss": 0.8923,
+      "step": 1725
+    },
+    {
+      "epoch": 0.4574299312533051,
+      "grad_norm": 0.244140625,
+      "learning_rate": 0.000131795875637513,
+      "loss": 0.8918,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4587519830777366,
+      "grad_norm": 0.34375,
+      "learning_rate": 0.00013135790133454305,
+      "loss": 0.8906,
+      "step": 1735
+    },
+    {
+      "epoch": 0.46007403490216814,
+      "grad_norm": 0.31640625,
+      "learning_rate": 0.00013091925889987865,
+      "loss": 0.8894,
+      "step": 1740
+    },
+    {
+      "epoch": 0.4613960867265997,
+      "grad_norm": 0.23046875,
+      "learning_rate": 0.00013047995767951883,
+      "loss": 0.8945,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4627181385510312,
+      "grad_norm": 0.306640625,
+      "learning_rate": 0.00013004000703349891,
+      "loss": 0.8918,
+      "step": 1750
+    },
+    {
+      "epoch": 0.46404019037546274,
+      "grad_norm": 0.29296875,
+      "learning_rate": 0.00012959941633569153,
+      "loss": 0.8951,
+      "step": 1755
+    },
+    {
+      "epoch": 0.46536224219989425,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 0.00012915819497360658,
+      "loss": 0.8887,
+      "step": 1760
+    },
+    {
+      "epoch": 0.46668429402432576,
+      "grad_norm": 0.251953125,
+      "learning_rate": 0.00012871635234819136,
+      "loss": 0.8897,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4680063458487573,
+      "grad_norm": 0.25390625,
+      "learning_rate": 0.00012827389787363006,
+      "loss": 0.892,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4693283976731888,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.00012783084097714347,
+      "loss": 0.8879,
+      "step": 1775
+    },
+    {
+      "epoch": 0.4706504494976203,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00012738719109878794,
+      "loss": 0.891,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4719725013220518,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00012694295769125424,
+      "loss": 0.8906,
+      "step": 1785
+    },
+    {
+      "epoch": 0.47329455314648333,
+      "grad_norm": 0.271484375,
+      "learning_rate": 0.0001264981502196662,
+      "loss": 0.8911,
+      "step": 1790
+    },
+    {
+      "epoch": 0.47461660497091485,
+      "grad_norm": 0.26171875,
+      "learning_rate": 0.000126052778161379,
+      "loss": 0.895,
+      "step": 1795
+    },
+    {
+      "epoch": 0.47593865679534636,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00012560685100577738,
+      "loss": 0.8892,
+      "step": 1800
+    },
+    {
+      "epoch": 0.4772607086197779,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.0001251603782540732,
+      "loss": 0.8869,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4785827604442094,
+      "grad_norm": 0.197265625,
+      "learning_rate": 0.0001247133694191032,
+      "loss": 0.8881,
+      "step": 1810
+    },
+    {
+      "epoch": 0.47990481226864096,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 0.0001242658340251263,
+      "loss": 0.8821,
+      "step": 1815
+    },
+    {
+      "epoch": 0.48122686409307247,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00012381778160762058,
+      "loss": 0.8936,
+      "step": 1820
+    },
+    {
+      "epoch": 0.482548915917504,
+      "grad_norm": 0.208984375,
+      "learning_rate": 0.00012336922171308018,
+      "loss": 0.8916,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4838709677419355,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00012292016389881184,
+      "loss": 0.8879,
+      "step": 1830
+    },
+    {
+      "epoch": 0.485193019566367,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00012247061773273135,
+      "loss": 0.8883,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4865150713907985,
+      "grad_norm": 0.3046875,
+      "learning_rate": 0.00012202059279315954,
+      "loss": 0.8888,
+      "step": 1840
+    },
+    {
+      "epoch": 0.48783712321523004,
+      "grad_norm": 0.365234375,
+      "learning_rate": 0.00012157009866861842,
+      "loss": 0.8884,
+      "step": 1845
+    },
+    {
+      "epoch": 0.48915917503966155,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.00012111914495762669,
+      "loss": 0.8899,
+      "step": 1850
+    },
+    {
+      "epoch": 0.49048122686409307,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00012066774126849529,
+      "loss": 0.8879,
+      "step": 1855
+    },
+    {
+      "epoch": 0.4918032786885246,
+      "grad_norm": 0.263671875,
+      "learning_rate": 0.00012021589721912277,
+      "loss": 0.8925,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4931253305129561,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00011976362243679014,
+      "loss": 0.8885,
+      "step": 1865
+    },
+    {
+      "epoch": 0.4944473823373876,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.00011931092655795597,
+      "loss": 0.8869,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4957694341618191,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00011885781922805101,
+      "loss": 0.8892,
+      "step": 1875
+    },
+    {
+      "epoch": 0.49709148598625064,
+      "grad_norm": 0.228515625,
+      "learning_rate": 0.0001184043101012726,
+      "loss": 0.8913,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4984135378106822,
+      "grad_norm": 0.232421875,
+      "learning_rate": 0.000117950408840379,
+      "loss": 0.8877,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4997355896351137,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00011749612511648362,
+      "loss": 0.8914,
+      "step": 1890
+    },
+    {
+      "epoch": 0.5010576414595452,
+      "grad_norm": 0.30078125,
+      "learning_rate": 0.0001170414686088488,
+      "loss": 0.8906,
+      "step": 1895
+    },
+    {
+      "epoch": 0.5023796932839767,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 0.00011658644900467965,
+      "loss": 0.8923,
+      "step": 1900
+    },
+    {
+      "epoch": 0.5037017451084083,
+      "grad_norm": 0.314453125,
+      "learning_rate": 0.00011613107599891775,
+      "loss": 0.8924,
+      "step": 1905
+    },
+    {
+      "epoch": 0.5050237969328397,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.00011567535929403435,
+      "loss": 0.8869,
+      "step": 1910
+    },
+    {
+      "epoch": 0.5063458487572713,
+      "grad_norm": 0.240234375,
+      "learning_rate": 0.00011521930859982382,
+      "loss": 0.8896,
+      "step": 1915
+    },
+    {
+      "epoch": 0.5076679005817029,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.00011476293363319675,
+      "loss": 0.8869,
+      "step": 1920
+    },
+    {
+      "epoch": 0.5089899524061343,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 0.0001143062441179729,
+      "loss": 0.8869,
+      "step": 1925
+    },
+    {
+      "epoch": 0.5103120042305659,
+      "grad_norm": 0.22265625,
+      "learning_rate": 0.00011384924978467397,
+      "loss": 0.8854,
+      "step": 1930
+    },
+    {
+      "epoch": 0.5116340560549973,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00011339196037031627,
+      "loss": 0.8905,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5129561078794289,
+      "grad_norm": 0.23828125,
+      "learning_rate": 0.00011293438561820343,
+      "loss": 0.8885,
+      "step": 1940
+    },
+    {
+      "epoch": 0.5142781597038604,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 0.00011247653527771847,
+      "loss": 0.8869,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5156002115282919,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.00011201841910411652,
+      "loss": 0.8873,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5169222633527234,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 0.00011156004685831648,
+      "loss": 0.893,
+      "step": 1955
+    },
+    {
+      "epoch": 0.518244315177155,
+      "grad_norm": 0.298828125,
+      "learning_rate": 0.0001111014283066935,
+      "loss": 0.8841,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5195663670015864,
+      "grad_norm": 0.291015625,
+      "learning_rate": 0.0001106425732208705,
+      "loss": 0.8796,
+      "step": 1965
+    },
+    {
+      "epoch": 0.520888418826018,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.0001101834913775103,
+      "loss": 0.8917,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5222104706504495,
+      "grad_norm": 0.35546875,
+      "learning_rate": 0.00010972419255810704,
+      "loss": 0.8918,
+      "step": 1975
+    },
+    {
+      "epoch": 0.523532522474881,
+      "grad_norm": 0.259765625,
+      "learning_rate": 0.000109264686548778,
+      "loss": 0.8866,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5248545742993125,
+      "grad_norm": 0.203125,
+      "learning_rate": 0.00010880498314005488,
+      "loss": 0.8911,
+      "step": 1985
+    },
+    {
+      "epoch": 0.526176626123744,
+      "grad_norm": 0.265625,
+      "learning_rate": 0.00010834509212667537,
+      "loss": 0.8856,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5274986779481756,
+      "grad_norm": 0.2890625,
+      "learning_rate": 0.00010788502330737438,
+      "loss": 0.8899,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5288207297726071,
+      "grad_norm": 0.248046875,
+      "learning_rate": 0.00010742478648467522,
+      "loss": 0.8883,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5301427815970386,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 0.00010696439146468085,
+      "loss": 0.8899,
+      "step": 2005
+    },
+    {
+      "epoch": 0.5314648334214701,
+      "grad_norm": 0.21875,
+      "learning_rate": 0.00010650384805686482,
+      "loss": 0.8894,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5327868852459017,
+      "grad_norm": 0.220703125,
+      "learning_rate": 0.0001060431660738624,
+      "loss": 0.8882,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5341089370703331,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00010558235533126142,
+      "loss": 0.8855,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5354309888947647,
+      "grad_norm": 0.236328125,
+      "learning_rate": 0.00010512142564739302,
+      "loss": 0.8841,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5367530407191962,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 0.00010466038684312278,
+      "loss": 0.8912,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5380750925436277,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 0.0001041992487416411,
+      "loss": 0.8883,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5393971443680592,
+      "grad_norm": 0.24609375,
+      "learning_rate": 0.00010373802116825418,
+      "loss": 0.8863,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5407191961924908,
+      "grad_norm": 0.275390625,
+      "learning_rate": 0.00010327671395017451,
+      "loss": 0.8831,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5420412480169222,
+      "grad_norm": 0.283203125,
+      "learning_rate": 0.00010281533691631155,
+      "loss": 0.8831,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5433632998413538,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 0.00010235389989706232,
+      "loss": 0.8888,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5446853516657852,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 0.0001018924127241019,
+      "loss": 0.893,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5460074034902168,
+      "grad_norm": 0.2109375,
+      "learning_rate": 0.000101430885230174,
+      "loss": 0.8872,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5473294553146484,
+      "grad_norm": 0.294921875,
+      "learning_rate": 0.00010096932724888146,
+      "loss": 0.888,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5486515071390798,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 0.00010050774861447662,
+      "loss": 0.8897,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5499735589635114,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 0.00010004615916165191,
+      "loss": 0.8865,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5512956107879429,
+      "grad_norm": 0.21875,
+      "learning_rate": 9.958456872533031e-05,
+      "loss": 0.8856,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5526176626123744,
+      "grad_norm": 0.181640625,
+      "learning_rate": 9.91229871404557e-05,
+      "loss": 0.8828,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5539397144368059,
+      "grad_norm": 0.158203125,
+      "learning_rate": 9.866142424178341e-05,
+      "loss": 0.8896,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5552617662612375,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 9.819988986367053e-05,
+      "loss": 0.8845,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5565838180856689,
+      "grad_norm": 0.208984375,
+      "learning_rate": 9.773839383986664e-05,
+      "loss": 0.8877,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5579058699101005,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 9.727694600330395e-05,
+      "loss": 0.8835,
+      "step": 2110
+    },
+    {
+      "epoch": 0.559227921734532,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 9.681555618588808e-05,
+      "loss": 0.8855,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5605499735589635,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 9.635423421828835e-05,
+      "loss": 0.8838,
+      "step": 2120
+    },
+    {
+      "epoch": 0.561872025383395,
+      "grad_norm": 0.23046875,
+      "learning_rate": 9.589298992972857e-05,
+      "loss": 0.8901,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5631940772078265,
+      "grad_norm": 0.306640625,
+      "learning_rate": 9.543183314777732e-05,
+      "loss": 0.888,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5645161290322581,
+      "grad_norm": 0.244140625,
+      "learning_rate": 9.497077369813885e-05,
+      "loss": 0.8865,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5658381808566896,
+      "grad_norm": 0.29296875,
+      "learning_rate": 9.450982140444344e-05,
+      "loss": 0.8916,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5671602326811211,
+      "grad_norm": 0.21875,
+      "learning_rate": 9.404898608803836e-05,
+      "loss": 0.8869,
+      "step": 2145
+    },
+    {
+      "epoch": 0.5684822845055526,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 9.358827756777837e-05,
+      "loss": 0.8844,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5698043363299842,
+      "grad_norm": 0.23046875,
+      "learning_rate": 9.312770565981677e-05,
+      "loss": 0.8861,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5711263881544156,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 9.266728017739601e-05,
+      "loss": 0.888,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5724484399788472,
+      "grad_norm": 0.1953125,
+      "learning_rate": 9.220701093063868e-05,
+      "loss": 0.8857,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5737704918032787,
+      "grad_norm": 0.2421875,
+      "learning_rate": 9.174690772633865e-05,
+      "loss": 0.8896,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5750925436277102,
+      "grad_norm": 0.326171875,
+      "learning_rate": 9.128698036775179e-05,
+      "loss": 0.884,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5764145954521417,
+      "grad_norm": 0.208984375,
+      "learning_rate": 9.082723865438747e-05,
+      "loss": 0.8903,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5777366472765733,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 9.036769238179947e-05,
+      "loss": 0.8927,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5790586991010047,
+      "grad_norm": 0.21484375,
+      "learning_rate": 8.990835134137744e-05,
+      "loss": 0.8848,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5803807509254363,
+      "grad_norm": 0.166015625,
+      "learning_rate": 8.944922532013811e-05,
+      "loss": 0.884,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5817028027498677,
+      "grad_norm": 0.2578125,
+      "learning_rate": 8.899032410051708e-05,
+      "loss": 0.8832,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5830248545742993,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 8.853165746015997e-05,
+      "loss": 0.8851,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5843469063987309,
+      "grad_norm": 0.21484375,
+      "learning_rate": 8.807323517171444e-05,
+      "loss": 0.8899,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5856689582231623,
+      "grad_norm": 0.181640625,
+      "learning_rate": 8.761506700262172e-05,
+      "loss": 0.8875,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5869910100475939,
+      "grad_norm": 0.19921875,
+      "learning_rate": 8.715716271490877e-05,
+      "loss": 0.8899,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5883130618720254,
+      "grad_norm": 0.185546875,
+      "learning_rate": 8.669953206497994e-05,
+      "loss": 0.8819,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5896351136964569,
+      "grad_norm": 0.25390625,
+      "learning_rate": 8.624218480340942e-05,
+      "loss": 0.8879,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5909571655208884,
+      "grad_norm": 0.296875,
+      "learning_rate": 8.578513067473322e-05,
+      "loss": 0.8877,
+      "step": 2235
+    },
+    {
+      "epoch": 0.59227921734532,
+      "grad_norm": 0.1875,
+      "learning_rate": 8.532837941724181e-05,
+      "loss": 0.8849,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5936012691697514,
+      "grad_norm": 0.2109375,
+      "learning_rate": 8.487194076277234e-05,
+      "loss": 0.8826,
+      "step": 2245
+    },
+    {
+      "epoch": 0.594923320994183,
+      "grad_norm": 0.259765625,
+      "learning_rate": 8.441582443650161e-05,
+      "loss": 0.8847,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5962453728186144,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 8.396004015673853e-05,
+      "loss": 0.8825,
+      "step": 2255
+    },
+    {
+      "epoch": 0.597567424643046,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 8.35045976347173e-05,
+      "loss": 0.888,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5988894764674775,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 8.304950657439033e-05,
+      "loss": 0.885,
+      "step": 2265
+    },
+    {
+      "epoch": 0.600211528291909,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 8.259477667222172e-05,
+      "loss": 0.8857,
+      "step": 2270
+    },
+    {
+      "epoch": 0.6015335801163406,
+      "grad_norm": 0.296875,
+      "learning_rate": 8.214041761698031e-05,
+      "loss": 0.8876,
+      "step": 2275
+    },
+    {
+      "epoch": 0.6028556319407721,
+      "grad_norm": 0.232421875,
+      "learning_rate": 8.168643908953359e-05,
+      "loss": 0.8846,
+      "step": 2280
+    },
+    {
+      "epoch": 0.6041776837652036,
+      "grad_norm": 0.255859375,
+      "learning_rate": 8.12328507626411e-05,
+      "loss": 0.8863,
+      "step": 2285
+    },
+    {
+      "epoch": 0.6054997355896351,
+      "grad_norm": 0.3359375,
+      "learning_rate": 8.077966230074872e-05,
+      "loss": 0.8889,
+      "step": 2290
+    },
+    {
+      "epoch": 0.6068217874140667,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 8.032688335978247e-05,
+      "loss": 0.8844,
+      "step": 2295
+    },
+    {
+      "epoch": 0.6081438392384981,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 7.987452358694278e-05,
+      "loss": 0.8894,
+      "step": 2300
+    },
+    {
+      "epoch": 0.6094658910629297,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 7.942259262049911e-05,
+      "loss": 0.8825,
+      "step": 2305
+    },
+    {
+      "epoch": 0.6107879428873612,
+      "grad_norm": 0.236328125,
+      "learning_rate": 7.897110008958441e-05,
+      "loss": 0.8851,
+      "step": 2310
+    },
+    {
+      "epoch": 0.6121099947117927,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 7.852005561399014e-05,
+      "loss": 0.8903,
+      "step": 2315
+    },
+    {
+      "epoch": 0.6134320465362242,
+      "grad_norm": 0.27734375,
+      "learning_rate": 7.806946880396109e-05,
+      "loss": 0.8873,
+      "step": 2320
+    },
+    {
+      "epoch": 0.6147540983606558,
+      "grad_norm": 0.27734375,
+      "learning_rate": 7.761934925999086e-05,
+      "loss": 0.8862,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6160761501850872,
+      "grad_norm": 0.23046875,
+      "learning_rate": 7.716970657261702e-05,
+      "loss": 0.8846,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6173982020095188,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 7.672055032221709e-05,
+      "loss": 0.8897,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6187202538339502,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 7.627189007880413e-05,
+      "loss": 0.8873,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6200423056583818,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 7.582373540182304e-05,
+      "loss": 0.8904,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6213643574828134,
+      "grad_norm": 0.208984375,
+      "learning_rate": 7.537609583994667e-05,
+      "loss": 0.8834,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6226864093072448,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 7.492898093087269e-05,
+      "loss": 0.8825,
+      "step": 2355
+    },
+    {
+      "epoch": 0.6240084611316764,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 7.448240020111993e-05,
+      "loss": 0.8837,
+      "step": 2360
+    },
+    {
+      "epoch": 0.6253305129561079,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 7.403636316582594e-05,
+      "loss": 0.8885,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6266525647805394,
+      "grad_norm": 0.2158203125,
+      "learning_rate": 7.359087932854371e-05,
+      "loss": 0.8944,
+      "step": 2370
+    },
+    {
+      "epoch": 0.6279746166049709,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.314595818103959e-05,
+      "loss": 0.889,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6292966684294025,
+      "grad_norm": 0.232421875,
+      "learning_rate": 7.270160920309077e-05,
+      "loss": 0.8869,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6306187202538339,
+      "grad_norm": 0.208984375,
+      "learning_rate": 7.225784186228359e-05,
+      "loss": 0.8829,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6319407720782655,
+      "grad_norm": 0.25390625,
+      "learning_rate": 7.181466561381152e-05,
+      "loss": 0.8889,
+      "step": 2390
+    },
+    {
+      "epoch": 0.633262823902697,
+      "grad_norm": 0.287109375,
+      "learning_rate": 7.13720899002739e-05,
+      "loss": 0.8823,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6345848757271285,
+      "grad_norm": 0.21484375,
+      "learning_rate": 7.093012415147462e-05,
+      "loss": 0.884,
+      "step": 2400
+    },
+    {
+      "epoch": 0.63590692755156,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 7.048877778422138e-05,
+      "loss": 0.8848,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6372289793759915,
+      "grad_norm": 0.259765625,
+      "learning_rate": 7.00480602021248e-05,
+      "loss": 0.8821,
+      "step": 2410
+    },
+    {
+      "epoch": 0.638551031200423,
+      "grad_norm": 0.291015625,
+      "learning_rate": 6.960798079539835e-05,
+      "loss": 0.8879,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6398730830248546,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 6.916854894065796e-05,
+      "loss": 0.886,
+      "step": 2420
+    },
+    {
+      "epoch": 0.6411951348492861,
+      "grad_norm": 0.251953125,
+      "learning_rate": 6.87297740007225e-05,
+      "loss": 0.8895,
+      "step": 2425
+    },
+    {
+      "epoch": 0.6425171866737176,
+      "grad_norm": 0.189453125,
+      "learning_rate": 6.829166532441421e-05,
+      "loss": 0.8839,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6438392384981492,
+      "grad_norm": 0.203125,
+      "learning_rate": 6.785423224635941e-05,
+      "loss": 0.8834,
+      "step": 2435
+    },
+    {
+      "epoch": 0.6451612903225806,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 6.741748408678975e-05,
+      "loss": 0.8849,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6464833421470122,
+      "grad_norm": 0.1787109375,
+      "learning_rate": 6.698143015134349e-05,
+      "loss": 0.8804,
+      "step": 2445
+    },
+    {
+      "epoch": 0.6478053939714437,
+      "grad_norm": 0.197265625,
+      "learning_rate": 6.65460797308674e-05,
+      "loss": 0.8845,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6491274457958752,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 6.611144210121861e-05,
+      "loss": 0.8897,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6504494976203067,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 6.567752652306717e-05,
+      "loss": 0.8913,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6517715494447383,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 6.52443422416985e-05,
+      "loss": 0.8884,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6530936012691697,
+      "grad_norm": 0.17578125,
+      "learning_rate": 6.48118984868167e-05,
+      "loss": 0.8894,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6544156530936013,
+      "grad_norm": 0.2265625,
+      "learning_rate": 6.438020447234759e-05,
+      "loss": 0.8837,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6557377049180327,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 6.394926939624267e-05,
+      "loss": 0.8864,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6570597567424643,
+      "grad_norm": 0.1796875,
+      "learning_rate": 6.351910244028285e-05,
+      "loss": 0.8859,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6583818085668959,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 6.308971276988313e-05,
+      "loss": 0.885,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6597038603913273,
+      "grad_norm": 0.263671875,
+      "learning_rate": 6.266110953389704e-05,
+      "loss": 0.8835,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6610259122157589,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 6.223330186442194e-05,
+      "loss": 0.8834,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6623479640401904,
+      "grad_norm": 0.201171875,
+      "learning_rate": 6.180629887660425e-05,
+      "loss": 0.8839,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6636700158646219,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 6.138010966844538e-05,
+      "loss": 0.8868,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6649920676890534,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 6.095474332060776e-05,
+      "loss": 0.8857,
+      "step": 2515
+    },
+    {
+      "epoch": 0.666314119513485,
+      "grad_norm": 0.205078125,
+      "learning_rate": 6.0530208896221527e-05,
+      "loss": 0.8823,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6676361713379164,
+      "grad_norm": 0.28125,
+      "learning_rate": 6.010651544069119e-05,
+      "loss": 0.8835,
+      "step": 2525
+    },
+    {
+      "epoch": 0.668958223162348,
+      "grad_norm": 0.259765625,
+      "learning_rate": 5.968367198150316e-05,
+      "loss": 0.8826,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6702802749867794,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 5.92616875280332e-05,
+      "loss": 0.8825,
+      "step": 2535
+    },
+    {
+      "epoch": 0.671602326811211,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 5.8840571071354565e-05,
+      "loss": 0.8874,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6729243786356425,
+      "grad_norm": 0.197265625,
+      "learning_rate": 5.842033158404636e-05,
+      "loss": 0.8914,
+      "step": 2545
+    },
+    {
+      "epoch": 0.674246430460074,
+      "grad_norm": 0.181640625,
+      "learning_rate": 5.800097802000257e-05,
+      "loss": 0.8844,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6755684822845055,
+      "grad_norm": 0.240234375,
+      "learning_rate": 5.758251931424089e-05,
+      "loss": 0.8809,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6768905341089371,
+      "grad_norm": 0.2421875,
+      "learning_rate": 5.716496438271277e-05,
+      "loss": 0.8828,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6782125859333686,
+      "grad_norm": 0.185546875,
+      "learning_rate": 5.6748322122113264e-05,
+      "loss": 0.8832,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6795346377578001,
+      "grad_norm": 0.23046875,
+      "learning_rate": 5.633260140969132e-05,
+      "loss": 0.885,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6808566895822317,
+      "grad_norm": 0.162109375,
+      "learning_rate": 5.591781110306096e-05,
+      "loss": 0.8837,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6821787414066631,
+      "grad_norm": 0.173828125,
+      "learning_rate": 5.55039600400122e-05,
+      "loss": 0.8817,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6835007932310947,
+      "grad_norm": 0.20703125,
+      "learning_rate": 5.509105703832313e-05,
+      "loss": 0.8858,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6848228450555262,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 5.467911089557169e-05,
+      "loss": 0.8863,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6861448968799577,
+      "grad_norm": 0.208984375,
+      "learning_rate": 5.426813038894839e-05,
+      "loss": 0.8844,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6874669487043892,
+      "grad_norm": 0.203125,
+      "learning_rate": 5.3858124275069244e-05,
+      "loss": 0.8836,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6887890005288207,
+      "grad_norm": 0.25390625,
+      "learning_rate": 5.344910128978934e-05,
+      "loss": 0.8837,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6901110523532522,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 5.304107014801645e-05,
+      "loss": 0.8845,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6914331041776838,
+      "grad_norm": 0.21875,
+      "learning_rate": 5.263403954352564e-05,
+      "loss": 0.8878,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6927551560021152,
+      "grad_norm": 0.1767578125,
+      "learning_rate": 5.222801814877369e-05,
+      "loss": 0.8879,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6940772078265468,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 5.182301461471474e-05,
+      "loss": 0.8905,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6953992596509783,
+      "grad_norm": 0.201171875,
+      "learning_rate": 5.141903757061555e-05,
+      "loss": 0.8889,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6967213114754098,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 5.1016095623871865e-05,
+      "loss": 0.8805,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6980433632998414,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 5.0614197359824955e-05,
+      "loss": 0.8848,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6993654151242729,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 5.021335134157885e-05,
+      "loss": 0.883,
+      "step": 2645
+    },
+    {
+      "epoch": 0.7006874669487044,
+      "grad_norm": 0.216796875,
+      "learning_rate": 4.981356610981756e-05,
+      "loss": 0.8817,
+      "step": 2650
+    },
+    {
+      "epoch": 0.7020095187731359,
+      "grad_norm": 0.1591796875,
+      "learning_rate": 4.9414850182623465e-05,
+      "loss": 0.8831,
+      "step": 2655
+    },
+    {
+      "epoch": 0.7033315705975675,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 4.901721205529548e-05,
+      "loss": 0.8845,
+      "step": 2660
+    },
+    {
+      "epoch": 0.7046536224219989,
+      "grad_norm": 0.189453125,
+      "learning_rate": 4.8620660200168425e-05,
+      "loss": 0.8821,
+      "step": 2665
+    },
+    {
+      "epoch": 0.7059756742464305,
+      "grad_norm": 0.19140625,
+      "learning_rate": 4.822520306643207e-05,
+      "loss": 0.8835,
+      "step": 2670
+    },
+    {
+      "epoch": 0.7072977260708619,
+      "grad_norm": 0.255859375,
+      "learning_rate": 4.783084907995156e-05,
+      "loss": 0.8831,
+      "step": 2675
+    },
+    {
+      "epoch": 0.7086197778952935,
+      "grad_norm": 0.158203125,
+      "learning_rate": 4.743760664308753e-05,
+      "loss": 0.8841,
+      "step": 2680
+    },
+    {
+      "epoch": 0.709941829719725,
+      "grad_norm": 0.298828125,
+      "learning_rate": 4.704548413451728e-05,
+      "loss": 0.8843,
+      "step": 2685
+    },
+    {
+      "epoch": 0.7112638815441565,
+      "grad_norm": 0.220703125,
+      "learning_rate": 4.665448990905613e-05,
+      "loss": 0.8843,
+      "step": 2690
+    },
+    {
+      "epoch": 0.712585933368588,
+      "grad_norm": 0.265625,
+      "learning_rate": 4.626463229747958e-05,
+      "loss": 0.8832,
+      "step": 2695
+    },
+    {
+      "epoch": 0.7139079851930196,
+      "grad_norm": 0.181640625,
+      "learning_rate": 4.587591960634565e-05,
+      "loss": 0.8852,
+      "step": 2700
+    },
+    {
+      "epoch": 0.7152300370174511,
+      "grad_norm": 0.2421875,
+      "learning_rate": 4.54883601178179e-05,
+      "loss": 0.8844,
+      "step": 2705
+    },
+    {
+      "epoch": 0.7165520888418826,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 4.510196208948909e-05,
+      "loss": 0.8873,
+      "step": 2710
+    },
+    {
+      "epoch": 0.7178741406663142,
+      "grad_norm": 0.17578125,
+      "learning_rate": 4.4716733754205064e-05,
+      "loss": 0.8872,
+      "step": 2715
+    },
+    {
+      "epoch": 0.7191961924907456,
+      "grad_norm": 0.197265625,
+      "learning_rate": 4.433268331988956e-05,
+      "loss": 0.8891,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7205182443151772,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 4.394981896936908e-05,
+      "loss": 0.8861,
+      "step": 2725
+    },
+    {
+      "epoch": 0.7218402961396087,
+      "grad_norm": 0.17578125,
+      "learning_rate": 4.356814886019871e-05,
+      "loss": 0.8884,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7231623479640402,
+      "grad_norm": 0.162109375,
+      "learning_rate": 4.3187681124488196e-05,
+      "loss": 0.8865,
+      "step": 2735
+    },
+    {
+      "epoch": 0.7244843997884717,
+      "grad_norm": 0.19921875,
+      "learning_rate": 4.280842386872889e-05,
+      "loss": 0.8817,
+      "step": 2740
+    },
+    {
+      "epoch": 0.7258064516129032,
+      "grad_norm": 0.1865234375,
+      "learning_rate": 4.243038517362068e-05,
+      "loss": 0.8832,
+      "step": 2745
+    },
+    {
+      "epoch": 0.7271285034373347,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 4.205357309390021e-05,
+      "loss": 0.8856,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7284505552617663,
+      "grad_norm": 0.14453125,
+      "learning_rate": 4.167799565816889e-05,
+      "loss": 0.8862,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7297726070861977,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 4.130366086872215e-05,
+      "loss": 0.8826,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7310946589106293,
+      "grad_norm": 0.22265625,
+      "learning_rate": 4.093057670137865e-05,
+      "loss": 0.8871,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7324167107350608,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 4.055875110531068e-05,
+      "loss": 0.8862,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7337387625594923,
+      "grad_norm": 0.20703125,
+      "learning_rate": 4.0188192002874404e-05,
+      "loss": 0.8849,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7350608143839239,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 3.9818907289441375e-05,
+      "loss": 0.8831,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7363828662083554,
+      "grad_norm": 0.205078125,
+      "learning_rate": 3.9450904833230094e-05,
+      "loss": 0.8879,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7377049180327869,
+      "grad_norm": 0.162109375,
+      "learning_rate": 3.908419247513862e-05,
+      "loss": 0.884,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7390269698572184,
+      "grad_norm": 0.173828125,
+      "learning_rate": 3.871877802857714e-05,
+      "loss": 0.8931,
+      "step": 2795
+    },
+    {
+      "epoch": 0.74034902168165,
+      "grad_norm": 0.2578125,
+      "learning_rate": 3.8354669279301914e-05,
+      "loss": 0.8881,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7416710735060814,
+      "grad_norm": 0.294921875,
+      "learning_rate": 3.7991873985248996e-05,
+      "loss": 0.8866,
+      "step": 2805
+    },
+    {
+      "epoch": 0.742993125330513,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 3.763039987636924e-05,
+      "loss": 0.8826,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7443151771549444,
+      "grad_norm": 0.193359375,
+      "learning_rate": 3.7270254654463356e-05,
+      "loss": 0.8841,
+      "step": 2815
+    },
+    {
+      "epoch": 0.745637228979376,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 3.6911445993018076e-05,
+      "loss": 0.8903,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7469592808038075,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 3.6553981537042406e-05,
+      "loss": 0.8823,
+      "step": 2825
+    },
+    {
+      "epoch": 0.748281332628239,
+      "grad_norm": 0.259765625,
+      "learning_rate": 3.619786890290482e-05,
+      "loss": 0.8846,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7496033844526705,
+      "grad_norm": 0.205078125,
+      "learning_rate": 3.5843115678171125e-05,
+      "loss": 0.881,
+      "step": 2835
+    },
+    {
+      "epoch": 0.7509254362771021,
+      "grad_norm": 0.255859375,
+      "learning_rate": 3.5489729421442516e-05,
+      "loss": 0.8845,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7522474881015336,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 3.513771766219487e-05,
+      "loss": 0.8857,
+      "step": 2845
+    },
+    {
+      "epoch": 0.7535695399259651,
+      "grad_norm": 0.220703125,
+      "learning_rate": 3.4787087900617955e-05,
+      "loss": 0.8882,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7548915917503967,
+      "grad_norm": 0.2421875,
+      "learning_rate": 3.443784760745593e-05,
+      "loss": 0.8853,
+      "step": 2855
+    },
+    {
+      "epoch": 0.7562136435748281,
+      "grad_norm": 0.216796875,
+      "learning_rate": 3.409000422384797e-05,
+      "loss": 0.8892,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7575356953992597,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 3.3743565161169835e-05,
+      "loss": 0.8842,
+      "step": 2865
+    },
+    {
+      "epoch": 0.7588577472236911,
+      "grad_norm": 0.1953125,
+      "learning_rate": 3.339853780087584e-05,
+      "loss": 0.8863,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7601797990481227,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 3.3054929494341805e-05,
+      "loss": 0.884,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7615018508725542,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 3.2712747562708115e-05,
+      "loss": 0.886,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7628239026969857,
+      "grad_norm": 0.216796875,
+      "learning_rate": 3.237199929672401e-05,
+      "loss": 0.8833,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7641459545214172,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 3.203269195659202e-05,
+      "loss": 0.8824,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7654680063458488,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 3.169483277181351e-05,
+      "loss": 0.8864,
+      "step": 2895
+    },
+    {
+      "epoch": 0.7667900581702802,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 3.135842894103436e-05,
+      "loss": 0.8867,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7681121099947118,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 3.102348763189188e-05,
+      "loss": 0.891,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7694341618191433,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 3.069001598086184e-05,
+      "loss": 0.8902,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7707562136435748,
+      "grad_norm": 0.193359375,
+      "learning_rate": 3.0358021093106594e-05,
+      "loss": 0.885,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7720782654680064,
+      "grad_norm": 0.23046875,
+      "learning_rate": 3.0027510042323537e-05,
+      "loss": 0.8866,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7734003172924379,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 2.9698489870594616e-05,
+      "loss": 0.8831,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7747223691168694,
+      "grad_norm": 0.244140625,
+      "learning_rate": 2.937096758823601e-05,
+      "loss": 0.8869,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7760444209413009,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 2.9044950173649e-05,
+      "loss": 0.8816,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7773664727657325,
+      "grad_norm": 0.16796875,
+      "learning_rate": 2.872044457317109e-05,
+      "loss": 0.8875,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7786885245901639,
+      "grad_norm": 0.171875,
+      "learning_rate": 2.839745770092821e-05,
+      "loss": 0.8828,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7800105764145955,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 2.8075996438687148e-05,
+      "loss": 0.8818,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7813326282390269,
+      "grad_norm": 0.212890625,
+      "learning_rate": 2.7756067635709204e-05,
+      "loss": 0.8911,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7826546800634585,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 2.7437678108604003e-05,
+      "loss": 0.8859,
+      "step": 2960
+    },
+    {
+      "epoch": 0.78397673188789,
+      "grad_norm": 0.1796875,
+      "learning_rate": 2.712083464118441e-05,
+      "loss": 0.8892,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7852987837123215,
+      "grad_norm": 0.205078125,
+      "learning_rate": 2.68055439843219e-05,
+      "loss": 0.8826,
+      "step": 2970
+    },
+    {
+      "epoch": 0.786620835536753,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 2.6491812855802834e-05,
+      "loss": 0.8877,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7879428873611846,
+      "grad_norm": 0.189453125,
+      "learning_rate": 2.6179647940185238e-05,
+      "loss": 0.8867,
+      "step": 2980
+    },
+    {
+      "epoch": 0.789264939185616,
+      "grad_norm": 0.232421875,
+      "learning_rate": 2.5869055888656336e-05,
+      "loss": 0.8858,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7905869910100476,
+      "grad_norm": 0.234375,
+      "learning_rate": 2.556004331889098e-05,
+      "loss": 0.8859,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7919090428344792,
+      "grad_norm": 0.197265625,
+      "learning_rate": 2.525261681491047e-05,
+      "loss": 0.8849,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7932310946589106,
+      "grad_norm": 0.248046875,
+      "learning_rate": 2.494678292694247e-05,
+      "loss": 0.8823,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7945531464833422,
+      "grad_norm": 0.162109375,
+      "learning_rate": 2.4642548171281244e-05,
+      "loss": 0.8886,
+      "step": 3005
+    },
+    {
+      "epoch": 0.7958751983077736,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 2.433991903014896e-05,
+      "loss": 0.8889,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7971972501322052,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 2.403890195155748e-05,
+      "loss": 0.8858,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7985193019566367,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 2.373950334917111e-05,
+      "loss": 0.8818,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7998413537810682,
+      "grad_norm": 0.263671875,
+      "learning_rate": 2.3441729602169748e-05,
+      "loss": 0.8859,
+      "step": 3025
+    },
+    {
+      "epoch": 0.8011634056054997,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 2.3145587055113183e-05,
+      "loss": 0.8876,
+      "step": 3030
+    },
+    {
+      "epoch": 0.8024854574299313,
+      "grad_norm": 0.25,
+      "learning_rate": 2.2851082017805703e-05,
+      "loss": 0.8843,
+      "step": 3035
+    },
+    {
+      "epoch": 0.8038075092543627,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 2.255822076516184e-05,
+      "loss": 0.8846,
+      "step": 3040
+    },
+    {
+      "epoch": 0.8051295610787943,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 2.2267009537072536e-05,
+      "loss": 0.8877,
+      "step": 3045
+    },
+    {
+      "epoch": 0.8064516129032258,
+      "grad_norm": 0.263671875,
+      "learning_rate": 2.197745453827226e-05,
+      "loss": 0.8836,
+      "step": 3050
+    },
+    {
+      "epoch": 0.8077736647276573,
+      "grad_norm": 0.171875,
+      "learning_rate": 2.168956193820676e-05,
+      "loss": 0.885,
+      "step": 3055
+    },
+    {
+      "epoch": 0.8090957165520889,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 2.140333787090172e-05,
+      "loss": 0.8896,
+      "step": 3060
+    },
+    {
+      "epoch": 0.8104177683765204,
+      "grad_norm": 0.197265625,
+      "learning_rate": 2.1118788434831894e-05,
+      "loss": 0.8819,
+      "step": 3065
+    },
+    {
+      "epoch": 0.8117398202009519,
+      "grad_norm": 0.19921875,
+      "learning_rate": 2.083591969279136e-05,
+      "loss": 0.8838,
+      "step": 3070
+    },
+    {
+      "epoch": 0.8130618720253834,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 2.0554737671764134e-05,
+      "loss": 0.8825,
+      "step": 3075
+    },
+    {
+      "epoch": 0.814383923849815,
+      "grad_norm": 0.181640625,
+      "learning_rate": 2.0275248362795995e-05,
+      "loss": 0.8857,
+      "step": 3080
+    },
+    {
+      "epoch": 0.8157059756742464,
+      "grad_norm": 0.30078125,
+      "learning_rate": 1.999745772086655e-05,
+      "loss": 0.8852,
+      "step": 3085
+    },
+    {
+      "epoch": 0.817028027498678,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 1.9721371664762666e-05,
+      "loss": 0.8864,
+      "step": 3090
+    },
+    {
+      "epoch": 0.8183500793231094,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 1.9446996076952094e-05,
+      "loss": 0.88,
+      "step": 3095
+    },
+    {
+      "epoch": 0.819672131147541,
+      "grad_norm": 0.20703125,
+      "learning_rate": 1.917433680345829e-05,
+      "loss": 0.8872,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8209941829719725,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 1.8903399653735766e-05,
+      "loss": 0.888,
+      "step": 3105
+    },
+    {
+      "epoch": 0.822316234796404,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 1.8634190400546415e-05,
+      "loss": 0.8882,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8236382866208355,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 1.8366714779836434e-05,
+      "loss": 0.8887,
+      "step": 3115
+    },
+    {
+      "epoch": 0.8249603384452671,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 1.8100978490614085e-05,
+      "loss": 0.8856,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8262823902696985,
+      "grad_norm": 0.1796875,
+      "learning_rate": 1.783698719482836e-05,
+      "loss": 0.886,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8276044420941301,
+      "grad_norm": 0.181640625,
+      "learning_rate": 1.7574746517248254e-05,
+      "loss": 0.8807,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8289264939185617,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 1.7314262045343e-05,
+      "loss": 0.8861,
+      "step": 3135
+    },
+    {
+      "epoch": 0.8302485457429931,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 1.7055539329162963e-05,
+      "loss": 0.8825,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8315705975674247,
+      "grad_norm": 0.193359375,
+      "learning_rate": 1.6798583881221374e-05,
+      "loss": 0.8803,
+      "step": 3145
+    },
+    {
+      "epoch": 0.8328926493918561,
+      "grad_norm": 0.23046875,
+      "learning_rate": 1.6543401176376916e-05,
+      "loss": 0.8877,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8342147012162877,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 1.628999665171712e-05,
+      "loss": 0.8832,
+      "step": 3155
+    },
+    {
+      "epoch": 0.8355367530407192,
+      "grad_norm": 0.1796875,
+      "learning_rate": 1.603837570644238e-05,
+      "loss": 0.8854,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8368588048651507,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.578854370175107e-05,
+      "loss": 0.883,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8381808566895822,
+      "grad_norm": 0.189453125,
+      "learning_rate": 1.554050596072516e-05,
+      "loss": 0.8855,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8395029085140138,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 1.5294267768216975e-05,
+      "loss": 0.8873,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8408249603384452,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.5049834370736405e-05,
+      "loss": 0.8869,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8421470121628768,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 1.4807210976339291e-05,
+      "loss": 0.8853,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8434690639873083,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.4566402754516328e-05,
+      "loss": 0.8862,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8447911158117398,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 1.4327414836082976e-05,
+      "loss": 0.884,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8461131676361714,
+      "grad_norm": 0.18359375,
+      "learning_rate": 1.409025231307013e-05,
+      "loss": 0.8904,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8474352194606029,
+      "grad_norm": 0.20703125,
+      "learning_rate": 1.3854920238615688e-05,
+      "loss": 0.8857,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8487572712850344,
+      "grad_norm": 0.189453125,
+      "learning_rate": 1.3621423626856756e-05,
+      "loss": 0.8863,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8500793231094659,
+      "grad_norm": 0.25390625,
+      "learning_rate": 1.3389767452822943e-05,
+      "loss": 0.8851,
+      "step": 3215
+    },
+    {
+      "epoch": 0.8514013749338974,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 1.3159956652330251e-05,
+      "loss": 0.886,
+      "step": 3220
+    },
+    {
+      "epoch": 0.8527234267583289,
+      "grad_norm": 0.21484375,
+      "learning_rate": 1.2931996121876033e-05,
+      "loss": 0.8822,
+      "step": 3225
+    },
+    {
+      "epoch": 0.8540454785827605,
+      "grad_norm": 0.1640625,
+      "learning_rate": 1.2705890718534508e-05,
+      "loss": 0.8833,
+      "step": 3230
+    },
+    {
+      "epoch": 0.8553675304071919,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 1.2481645259853436e-05,
+      "loss": 0.8854,
+      "step": 3235
+    },
+    {
+      "epoch": 0.8566895822316235,
+      "grad_norm": 0.220703125,
+      "learning_rate": 1.2259264523751335e-05,
+      "loss": 0.8882,
+      "step": 3240
+    },
+    {
+      "epoch": 0.858011634056055,
+      "grad_norm": 0.18359375,
+      "learning_rate": 1.2038753248415768e-05,
+      "loss": 0.8862,
+      "step": 3245
+    },
+    {
+      "epoch": 0.8593336858804865,
+      "grad_norm": 0.1572265625,
+      "learning_rate": 1.1820116132202374e-05,
+      "loss": 0.8882,
+      "step": 3250
+    },
+    {
+      "epoch": 0.860655737704918,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 1.16033578335347e-05,
+      "loss": 0.888,
+      "step": 3255
+    },
+    {
+      "epoch": 0.8619777895293496,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.138848297080507e-05,
+      "loss": 0.887,
+      "step": 3260
+    },
+    {
+      "epoch": 0.863299841353781,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 1.1175496122276008e-05,
+      "loss": 0.8841,
+      "step": 3265
+    },
+    {
+      "epoch": 0.8646218931782126,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 1.0964401825982895e-05,
+      "loss": 0.8828,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8659439450026442,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 1.075520457963708e-05,
+      "loss": 0.8828,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8672659968270756,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 1.0547908840530197e-05,
+      "loss": 0.8892,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8685880486515072,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 1.034251902543908e-05,
+      "loss": 0.8842,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8699101004759386,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 1.01390395105318e-05,
+      "loss": 0.8858,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8712321523003702,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 9.937474631274257e-06,
+      "loss": 0.8896,
+      "step": 3295
+    },
+    {
+      "epoch": 0.8725542041248017,
+      "grad_norm": 0.220703125,
+      "learning_rate": 9.737828682337969e-06,
+      "loss": 0.8844,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8738762559492332,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 9.540105917508391e-06,
+      "loss": 0.885,
+      "step": 3305
+    },
+    {
+      "epoch": 0.8751983077736647,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 9.34431054959447e-06,
+      "loss": 0.8878,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8765203595980963,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 9.150446750338714e-06,
+      "loss": 0.8867,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8778424114225277,
+      "grad_norm": 0.255859375,
+      "learning_rate": 8.958518650328429e-06,
+      "loss": 0.8846,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8791644632469593,
+      "grad_norm": 0.169921875,
+      "learning_rate": 8.768530338907655e-06,
+      "loss": 0.8856,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8804865150713908,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 8.580485864090015e-06,
+      "loss": 0.8829,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8818085668958223,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 8.394389232472499e-06,
+      "loss": 0.8895,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8831306187202538,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 8.210244409150158e-06,
+      "loss": 0.8858,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8844526705446853,
+      "grad_norm": 0.189453125,
+      "learning_rate": 8.028055317631455e-06,
+      "loss": 0.8871,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8857747223691169,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 7.847825839754852e-06,
+      "loss": 0.8837,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8870967741935484,
+      "grad_norm": 0.216796875,
+      "learning_rate": 7.669559815605965e-06,
+      "loss": 0.8839,
+      "step": 3355
+    },
+    {
+      "epoch": 0.88841882601798,
+      "grad_norm": 0.1533203125,
+      "learning_rate": 7.493261043435828e-06,
+      "loss": 0.8802,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8897408778424114,
+      "grad_norm": 0.169921875,
+      "learning_rate": 7.318933279579909e-06,
+      "loss": 0.8831,
+      "step": 3365
+    },
+    {
+      "epoch": 0.891062929666843,
+      "grad_norm": 0.189453125,
+      "learning_rate": 7.146580238378131e-06,
+      "loss": 0.8866,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8923849814912744,
+      "grad_norm": 0.1689453125,
+      "learning_rate": 6.976205592095675e-06,
+      "loss": 0.886,
+      "step": 3375
+    },
+    {
+      "epoch": 0.893707033315706,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 6.8078129708447516e-06,
+      "loss": 0.8897,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8950290851401375,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.641405962507307e-06,
+      "loss": 0.886,
+      "step": 3385
+    },
+    {
+      "epoch": 0.896351136964569,
+      "grad_norm": 0.18359375,
+      "learning_rate": 6.4769881126584996e-06,
+      "loss": 0.8848,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8976731887890005,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 6.314562924491229e-06,
+      "loss": 0.8863,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8989952406134321,
+      "grad_norm": 0.208984375,
+      "learning_rate": 6.154133858741407e-06,
+      "loss": 0.8861,
+      "step": 3400
+    },
+    {
+      "epoch": 0.9003172924378635,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 5.995704333614327e-06,
+      "loss": 0.8867,
+      "step": 3405
+    },
+    {
+      "epoch": 0.9016393442622951,
+      "grad_norm": 0.28125,
+      "learning_rate": 5.839277724711733e-06,
+      "loss": 0.8874,
+      "step": 3410
+    },
+    {
+      "epoch": 0.9029613960867267,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 5.6848573649599834e-06,
+      "loss": 0.8782,
+      "step": 3415
+    },
+    {
+      "epoch": 0.9042834479111581,
+      "grad_norm": 0.185546875,
+      "learning_rate": 5.53244654453896e-06,
+      "loss": 0.8851,
+      "step": 3420
+    },
+    {
+      "epoch": 0.9056054997355897,
+      "grad_norm": 0.1953125,
+      "learning_rate": 5.382048510812032e-06,
+      "loss": 0.8814,
+      "step": 3425
+    },
+    {
+      "epoch": 0.9069275515600211,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 5.233666468256804e-06,
+      "loss": 0.888,
+      "step": 3430
+    },
+    {
+      "epoch": 0.9082496033844527,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 5.0873035783969066e-06,
+      "loss": 0.8862,
+      "step": 3435
+    },
+    {
+      "epoch": 0.9095716552088842,
+      "grad_norm": 0.16796875,
+      "learning_rate": 4.942962959734565e-06,
+      "loss": 0.8835,
+      "step": 3440
+    },
+    {
+      "epoch": 0.9108937070333157,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 4.800647687684223e-06,
+      "loss": 0.8893,
+      "step": 3445
+    },
+    {
+      "epoch": 0.9122157588577472,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 4.660360794506946e-06,
+      "loss": 0.8826,
+      "step": 3450
+    },
+    {
+      "epoch": 0.9135378106821788,
+      "grad_norm": 0.166015625,
+      "learning_rate": 4.52210526924588e-06,
+      "loss": 0.8845,
+      "step": 3455
+    },
+    {
+      "epoch": 0.9148598625066102,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 4.385884057662515e-06,
+      "loss": 0.8862,
+      "step": 3460
+    },
+    {
+      "epoch": 0.9161819143310418,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 4.251700062173947e-06,
+      "loss": 0.886,
+      "step": 3465
+    },
+    {
+      "epoch": 0.9175039661554732,
+      "grad_norm": 0.220703125,
+      "learning_rate": 4.119556141791048e-06,
+      "loss": 0.888,
+      "step": 3470
+    },
+    {
+      "epoch": 0.9188260179799048,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 3.989455112057483e-06,
+      "loss": 0.8857,
+      "step": 3475
+    },
+    {
+      "epoch": 0.9201480698043363,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 3.861399744989813e-06,
+      "loss": 0.8871,
+      "step": 3480
+    },
+    {
+      "epoch": 0.9214701216287678,
+      "grad_norm": 0.203125,
+      "learning_rate": 3.7353927690183867e-06,
+      "loss": 0.8852,
+      "step": 3485
+    },
+    {
+      "epoch": 0.9227921734531994,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 3.6114368689291856e-06,
+      "loss": 0.8834,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9241142252776309,
+      "grad_norm": 0.267578125,
+      "learning_rate": 3.4895346858066724e-06,
+      "loss": 0.8847,
+      "step": 3495
+    },
+    {
+      "epoch": 0.9254362771020624,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 3.3696888169774677e-06,
+      "loss": 0.8825,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9267583289264939,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 3.2519018159550496e-06,
+      "loss": 0.8884,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9280803807509255,
+      "grad_norm": 0.3671875,
+      "learning_rate": 3.1361761923853095e-06,
+      "loss": 0.8816,
+      "step": 3510
+    },
+    {
+      "epoch": 0.9294024325753569,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 3.0225144119931158e-06,
+      "loss": 0.8828,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9307244843997885,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.910918896529757e-06,
+      "loss": 0.8851,
+      "step": 3520
+    },
+    {
+      "epoch": 0.93204653622422,
+      "grad_norm": 0.236328125,
+      "learning_rate": 2.8013920237213277e-06,
+      "loss": 0.8851,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9333685880486515,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 2.693936127218133e-06,
+      "loss": 0.8877,
+      "step": 3530
+    },
+    {
+      "epoch": 0.934690639873083,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 2.5885534965448768e-06,
+      "loss": 0.8848,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9360126916975146,
+      "grad_norm": 0.1513671875,
+      "learning_rate": 2.4852463770519417e-06,
+      "loss": 0.8823,
+      "step": 3540
+    },
+    {
+      "epoch": 0.937334743521946,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 2.3840169698675087e-06,
+      "loss": 0.8865,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9386567953463776,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 2.2848674318507125e-06,
+      "loss": 0.8829,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9399788471708092,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 2.1877998755456266e-06,
+      "loss": 0.8851,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9413008989952406,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 2.0928163691362856e-06,
+      "loss": 0.8876,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9426229508196722,
+      "grad_norm": 0.169921875,
+      "learning_rate": 1.9999189364025894e-06,
+      "loss": 0.8861,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9439450026441036,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 1.909109556677269e-06,
+      "loss": 0.8874,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9452670544685352,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 1.8203901648035648e-06,
+      "loss": 0.8865,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9465891062929667,
+      "grad_norm": 0.181640625,
+      "learning_rate": 1.7337626510941619e-06,
+      "loss": 0.8876,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9479111581173982,
+      "grad_norm": 0.3046875,
+      "learning_rate": 1.6492288612907748e-06,
+      "loss": 0.8889,
+      "step": 3585
+    },
+    {
+      "epoch": 0.9492332099418297,
+      "grad_norm": 0.208984375,
+      "learning_rate": 1.5667905965249363e-06,
+      "loss": 0.8842,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9505552617662613,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 1.486449613279539e-06,
+      "loss": 0.8857,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9518773135906927,
+      "grad_norm": 0.208984375,
+      "learning_rate": 1.408207623351454e-06,
+      "loss": 0.8847,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9531993654151243,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 1.3320662938150485e-06,
+      "loss": 0.8828,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9545214172395557,
+      "grad_norm": 0.1669921875,
+      "learning_rate": 1.2580272469866483e-06,
+      "loss": 0.8787,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9558434690639873,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 1.186092060389976e-06,
+      "loss": 0.886,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9571655208884188,
+      "grad_norm": 0.146484375,
+      "learning_rate": 1.1162622667226008e-06,
+      "loss": 0.8837,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9584875727128503,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 1.0485393538231747e-06,
+      "loss": 0.8846,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9598096245372819,
+      "grad_norm": 0.234375,
+      "learning_rate": 9.829247646398144e-07,
+      "loss": 0.8867,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9611316763617134,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 9.194198971993362e-07,
+      "loss": 0.8869,
+      "step": 3635
+    },
+    {
+      "epoch": 0.9624537281861449,
+      "grad_norm": 0.265625,
+      "learning_rate": 8.580261045774474e-07,
+      "loss": 0.8824,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9637757800105764,
+      "grad_norm": 0.16796875,
+      "learning_rate": 7.98744694869924e-07,
+      "loss": 0.8875,
+      "step": 3645
+    },
+    {
+      "epoch": 0.965097831835008,
+      "grad_norm": 0.291015625,
+      "learning_rate": 7.41576931164778e-07,
+      "loss": 0.8856,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9664198836594394,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 6.86524031515301e-07,
+      "loss": 0.8861,
+      "step": 3655
+    },
+    {
+      "epoch": 0.967741935483871,
+      "grad_norm": 0.1806640625,
+      "learning_rate": 6.335871689140959e-07,
+      "loss": 0.8868,
+      "step": 3660
+    },
+    {
+      "epoch": 0.9690639873083025,
+      "grad_norm": 0.1650390625,
+      "learning_rate": 5.827674712681752e-07,
+      "loss": 0.8873,
+      "step": 3665
+    },
+    {
+      "epoch": 0.970386039132734,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 5.340660213748017e-07,
+      "loss": 0.8833,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9717080909571655,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 4.874838568985296e-07,
+      "loss": 0.8803,
+      "step": 3675
+    },
+    {
+      "epoch": 0.973030142781597,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 4.430219703490335e-07,
+      "loss": 0.8846,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9743521946060285,
+      "grad_norm": 0.28515625,
+      "learning_rate": 4.0068130905996924e-07,
+      "loss": 0.8873,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9756742464304601,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 3.604627751688239e-07,
+      "loss": 0.8817,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9769962982548915,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 3.2236722559764176e-07,
+      "loss": 0.8877,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9783183500793231,
+      "grad_norm": 0.19140625,
+      "learning_rate": 2.86395472034795e-07,
+      "loss": 0.8879,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9796404019037547,
+      "grad_norm": 0.2578125,
+      "learning_rate": 2.525482809177082e-07,
+      "loss": 0.8908,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9809624537281861,
+      "grad_norm": 0.173828125,
+      "learning_rate": 2.2082637341647173e-07,
+      "loss": 0.881,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9822845055526177,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 1.9123042541854265e-07,
+      "loss": 0.8863,
+      "step": 3715
+    },
+    {
+      "epoch": 0.9836065573770492,
+      "grad_norm": 0.16796875,
+      "learning_rate": 1.6376106751430087e-07,
+      "loss": 0.8909,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9849286092014807,
+      "grad_norm": 0.21484375,
+      "learning_rate": 1.3841888498361544e-07,
+      "loss": 0.8815,
+      "step": 3725
+    },
+    {
+      "epoch": 0.9862506610259122,
+      "grad_norm": 0.23828125,
+      "learning_rate": 1.1520441778339885e-07,
+      "loss": 0.8851,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9875727128503438,
+      "grad_norm": 0.259765625,
+      "learning_rate": 9.411816053608302e-08,
+      "loss": 0.8855,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9888947646747752,
+      "grad_norm": 0.197265625,
+      "learning_rate": 7.51605625190721e-08,
+      "loss": 0.8903,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9902168164992068,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 5.833202765519463e-08,
+      "loss": 0.8867,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9915388683236382,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 4.3632914504077026e-08,
+      "loss": 0.8839,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9928609201480698,
+      "grad_norm": 0.193359375,
+      "learning_rate": 3.106353625451641e-08,
+      "loss": 0.8824,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9941829719725013,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 2.062416071780815e-08,
+      "loss": 0.8881,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9955050237969328,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 1.2315010322028198e-08,
+      "loss": 0.8869,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9968270756213644,
+      "grad_norm": 0.19921875,
+      "learning_rate": 6.136262107292456e-09,
+      "loss": 0.8892,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9981491274457959,
+      "grad_norm": 0.23828125,
+      "learning_rate": 2.0880477220042163e-09,
+      "loss": 0.8853,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9994711792702274,
+      "grad_norm": 0.234375,
+      "learning_rate": 1.7045342003418895e-10,
+      "loss": 0.8854,
+      "step": 3780
+    },
     {
       "epoch": 1.0,
-      "step": 220,
-      "total_flos": 7357983621120000.0,
-      "train_loss": 1.3830752210183577,
-      "train_runtime": 55.0309,
-      "train_samples_per_second": 255.239,
-      "train_steps_per_second": 3.998
+      "step": 3782,
+      "total_flos": 1.011923420184576e+18,
+      "train_loss": 0.9601045426144794,
+      "train_runtime": 898.0661,
+      "train_samples_per_second": 2156.094,
+      "train_steps_per_second": 4.211
     }
   ],
   "logging_steps": 5,
-  "max_steps": 220,
+  "max_steps": 3782,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 1,
   "save_steps": 100,
@@ -350,7 +5334,7 @@
       "attributes": {}
     }
   },
-  "total_flos": 7357983621120000.0,
+  "total_flos": 1.011923420184576e+18,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null