End of training

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 5.0,
-    "eval_accuracy": 0.9937357630979499,
-    "eval_loss": 0.0182377677410841,
-    "eval_runtime": 26.022,
-    "eval_samples_per_second": 134.963,
-    "eval_steps_per_second": 4.227,
-    "train_loss": 0.06854689962371369,
-    "train_runtime": 1208.8541,
-    "train_samples_per_second": 82.301,
-    "train_steps_per_second": 2.573
+    "eval_accuracy": 0.9957289293849658,
+    "eval_loss": 0.013703458942472935,
+    "eval_runtime": 32.24,
+    "eval_samples_per_second": 108.933,
+    "eval_steps_per_second": 3.412,
+    "train_loss": 0.07020007294761406,
+    "train_runtime": 1821.7529,
+    "train_samples_per_second": 54.612,
+    "train_steps_per_second": 1.707
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 5.0,
-    "eval_accuracy": 0.9937357630979499,
-    "eval_loss": 0.0182377677410841,
-    "eval_runtime": 26.022,
-    "eval_samples_per_second": 134.963,
-    "eval_steps_per_second": 4.227
+    "eval_accuracy": 0.9957289293849658,
+    "eval_loss": 0.013703458942472935,
+    "eval_runtime": 32.24,
+    "eval_samples_per_second": 108.933,
+    "eval_steps_per_second": 3.412
 }

train_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 5.0,
-    "train_loss": 0.06854689962371369,
-    "train_runtime": 1208.8541,
-    "train_samples_per_second": 82.301,
-    "train_steps_per_second": 2.573
+    "train_loss": 0.07020007294761406,
+    "train_runtime": 1821.7529,
+    "train_samples_per_second": 54.612,
+    "train_steps_per_second": 1.707
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.0182377677410841,
+  "best_metric": 0.013703458942472935,
   "best_model_checkpoint": "./cats_vs_dogs_outputs/checkpoint-3110",
   "epoch": 5.0,
   "global_step": 3110,
@@ -10,1922 +10,1922 @@
     {
       "epoch": 0.02,
       "learning_rate": 0.00019935691318327976,
-      "loss": 0.3497,
+      "loss": 0.3424,
       "step": 10
     },
     {
       "epoch": 0.03,
       "learning_rate": 0.00019871382636655948,
-      "loss": 0.1232,
+      "loss": 0.1264,
       "step": 20
     },
     {
       "epoch": 0.05,
       "learning_rate": 0.00019807073954983923,
-      "loss": 0.1623,
+      "loss": 0.2064,
       "step": 30
     },
     {
       "epoch": 0.06,
       "learning_rate": 0.00019742765273311899,
-      "loss": 0.1004,
+      "loss": 0.1065,
       "step": 40
     },
     {
       "epoch": 0.08,
       "learning_rate": 0.00019678456591639874,
-      "loss": 0.0815,
+      "loss": 0.1185,
       "step": 50
     },
     {
       "epoch": 0.1,
       "learning_rate": 0.00019614147909967846,
-      "loss": 0.161,
+      "loss": 0.1264,
       "step": 60
     },
     {
       "epoch": 0.11,
       "learning_rate": 0.0001954983922829582,
-      "loss": 0.168,
+      "loss": 0.2004,
       "step": 70
     },
     {
       "epoch": 0.13,
       "learning_rate": 0.00019485530546623796,
-      "loss": 0.1196,
+      "loss": 0.1119,
       "step": 80
     },
     {
       "epoch": 0.14,
       "learning_rate": 0.00019421221864951768,
-      "loss": 0.1474,
+      "loss": 0.0939,
       "step": 90
     },
     {
       "epoch": 0.16,
       "learning_rate": 0.00019356913183279743,
-      "loss": 0.1664,
+      "loss": 0.1851,
       "step": 100
     },
     {
       "epoch": 0.18,
       "learning_rate": 0.00019292604501607719,
-      "loss": 0.1165,
+      "loss": 0.1271,
       "step": 110
     },
     {
       "epoch": 0.19,
       "learning_rate": 0.00019228295819935694,
-      "loss": 0.0778,
+      "loss": 0.0977,
       "step": 120
     },
     {
       "epoch": 0.21,
       "learning_rate": 0.00019163987138263666,
-      "loss": 0.0898,
+      "loss": 0.0873,
       "step": 130
     },
     {
       "epoch": 0.23,
       "learning_rate": 0.0001909967845659164,
-      "loss": 0.1072,
+      "loss": 0.0968,
       "step": 140
     },
     {
       "epoch": 0.24,
       "learning_rate": 0.00019035369774919616,
-      "loss": 0.1722,
+      "loss": 0.2024,
       "step": 150
     },
     {
       "epoch": 0.26,
       "learning_rate": 0.00018971061093247588,
-      "loss": 0.1277,
+      "loss": 0.094,
       "step": 160
     },
     {
       "epoch": 0.27,
       "learning_rate": 0.00018906752411575564,
-      "loss": 0.1044,
+      "loss": 0.1207,
       "step": 170
     },
     {
       "epoch": 0.29,
       "learning_rate": 0.00018842443729903539,
-      "loss": 0.1282,
+      "loss": 0.1374,
       "step": 180
     },
     {
       "epoch": 0.31,
       "learning_rate": 0.00018778135048231514,
-      "loss": 0.1034,
+      "loss": 0.1382,
       "step": 190
     },
     {
       "epoch": 0.32,
       "learning_rate": 0.00018713826366559486,
-      "loss": 0.1608,
+      "loss": 0.1678,
       "step": 200
     },
     {
       "epoch": 0.34,
       "learning_rate": 0.0001864951768488746,
-      "loss": 0.108,
+      "loss": 0.0885,
       "step": 210
     },
     {
       "epoch": 0.35,
       "learning_rate": 0.00018585209003215436,
-      "loss": 0.0853,
+      "loss": 0.1262,
       "step": 220
     },
     {
       "epoch": 0.37,
       "learning_rate": 0.00018520900321543409,
-      "loss": 0.178,
+      "loss": 0.1398,
       "step": 230
     },
     {
       "epoch": 0.39,
       "learning_rate": 0.00018456591639871384,
-      "loss": 0.1595,
+      "loss": 0.1808,
       "step": 240
     },
     {
       "epoch": 0.4,
       "learning_rate": 0.0001839228295819936,
-      "loss": 0.1134,
+      "loss": 0.1303,
       "step": 250
     },
     {
       "epoch": 0.42,
       "learning_rate": 0.0001832797427652733,
-      "loss": 0.0914,
+      "loss": 0.113,
       "step": 260
     },
     {
       "epoch": 0.43,
       "learning_rate": 0.00018263665594855306,
-      "loss": 0.107,
+      "loss": 0.152,
       "step": 270
     },
     {
       "epoch": 0.45,
       "learning_rate": 0.0001819935691318328,
-      "loss": 0.078,
+      "loss": 0.1548,
       "step": 280
     },
     {
       "epoch": 0.47,
       "learning_rate": 0.00018135048231511256,
-      "loss": 0.0715,
+      "loss": 0.0826,
       "step": 290
     },
     {
       "epoch": 0.48,
       "learning_rate": 0.00018070739549839229,
-      "loss": 0.055,
+      "loss": 0.1262,
       "step": 300
     },
     {
       "epoch": 0.5,
       "learning_rate": 0.00018006430868167204,
-      "loss": 0.1244,
+      "loss": 0.1371,
       "step": 310
     },
     {
       "epoch": 0.51,
       "learning_rate": 0.0001794212218649518,
-      "loss": 0.1074,
+      "loss": 0.0804,
       "step": 320
     },
     {
       "epoch": 0.53,
       "learning_rate": 0.0001787781350482315,
-      "loss": 0.1277,
+      "loss": 0.1163,
       "step": 330
     },
     {
       "epoch": 0.55,
       "learning_rate": 0.00017813504823151126,
-      "loss": 0.1298,
+      "loss": 0.107,
       "step": 340
     },
     {
       "epoch": 0.56,
       "learning_rate": 0.000177491961414791,
-      "loss": 0.1037,
+      "loss": 0.1148,
       "step": 350
     },
     {
       "epoch": 0.58,
       "learning_rate": 0.00017684887459807076,
-      "loss": 0.1061,
+      "loss": 0.1215,
       "step": 360
     },
     {
       "epoch": 0.59,
       "learning_rate": 0.00017620578778135049,
-      "loss": 0.1455,
+      "loss": 0.1431,
       "step": 370
     },
     {
       "epoch": 0.61,
       "learning_rate": 0.00017556270096463024,
-      "loss": 0.1469,
+      "loss": 0.1057,
       "step": 380
     },
     {
       "epoch": 0.63,
       "learning_rate": 0.00017491961414791,
-      "loss": 0.127,
+      "loss": 0.0679,
       "step": 390
     },
     {
       "epoch": 0.64,
       "learning_rate": 0.0001742765273311897,
-      "loss": 0.1033,
+      "loss": 0.0605,
       "step": 400
     },
     {
       "epoch": 0.66,
       "learning_rate": 0.00017363344051446946,
-      "loss": 0.1018,
+      "loss": 0.0614,
       "step": 410
     },
     {
       "epoch": 0.68,
       "learning_rate": 0.0001729903536977492,
-      "loss": 0.0866,
+      "loss": 0.1887,
       "step": 420
     },
     {
       "epoch": 0.69,
       "learning_rate": 0.00017234726688102896,
-      "loss": 0.079,
+      "loss": 0.0818,
       "step": 430
     },
     {
       "epoch": 0.71,
       "learning_rate": 0.0001717041800643087,
-      "loss": 0.0837,
+      "loss": 0.0893,
       "step": 440
     },
     {
       "epoch": 0.72,
       "learning_rate": 0.00017106109324758844,
-      "loss": 0.1862,
+      "loss": 0.1036,
       "step": 450
     },
     {
       "epoch": 0.74,
       "learning_rate": 0.0001704180064308682,
-      "loss": 0.1114,
+      "loss": 0.0994,
       "step": 460
     },
     {
       "epoch": 0.76,
       "learning_rate": 0.0001697749196141479,
-      "loss": 0.071,
+      "loss": 0.0762,
       "step": 470
     },
     {
       "epoch": 0.77,
       "learning_rate": 0.00016913183279742766,
-      "loss": 0.1379,
+      "loss": 0.1148,
       "step": 480
     },
     {
       "epoch": 0.79,
       "learning_rate": 0.0001684887459807074,
-      "loss": 0.0764,
+      "loss": 0.0814,
       "step": 490
     },
     {
       "epoch": 0.8,
       "learning_rate": 0.00016784565916398716,
-      "loss": 0.115,
+      "loss": 0.0913,
       "step": 500
     },
     {
       "epoch": 0.82,
       "learning_rate": 0.0001672025723472669,
-      "loss": 0.0704,
+      "loss": 0.1029,
       "step": 510
     },
     {
       "epoch": 0.84,
       "learning_rate": 0.00016655948553054664,
-      "loss": 0.0867,
+      "loss": 0.2109,
       "step": 520
     },
     {
       "epoch": 0.85,
       "learning_rate": 0.0001659163987138264,
-      "loss": 0.0827,
+      "loss": 0.1473,
       "step": 530
     },
     {
       "epoch": 0.87,
       "learning_rate": 0.0001652733118971061,
-      "loss": 0.0583,
+      "loss": 0.1185,
       "step": 540
     },
     {
       "epoch": 0.88,
       "learning_rate": 0.00016463022508038586,
-      "loss": 0.1185,
+      "loss": 0.0641,
       "step": 550
     },
     {
       "epoch": 0.9,
       "learning_rate": 0.0001639871382636656,
-      "loss": 0.1053,
+      "loss": 0.0837,
       "step": 560
     },
     {
       "epoch": 0.92,
       "learning_rate": 0.00016334405144694536,
-      "loss": 0.0695,
+      "loss": 0.0896,
       "step": 570
     },
     {
       "epoch": 0.93,
       "learning_rate": 0.0001627009646302251,
-      "loss": 0.0818,
+      "loss": 0.1284,
       "step": 580
     },
     {
       "epoch": 0.95,
       "learning_rate": 0.00016205787781350484,
-      "loss": 0.1222,
+      "loss": 0.1115,
       "step": 590
     },
     {
       "epoch": 0.96,
       "learning_rate": 0.0001614147909967846,
-      "loss": 0.2004,
+      "loss": 0.1514,
       "step": 600
     },
     {
       "epoch": 0.98,
       "learning_rate": 0.0001607717041800643,
-      "loss": 0.1069,
+      "loss": 0.1046,
       "step": 610
     },
     {
       "epoch": 1.0,
       "learning_rate": 0.00016012861736334406,
-      "loss": 0.1177,
+      "loss": 0.0658,
       "step": 620
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.9832004555808656,
-      "eval_loss": 0.04725871980190277,
-      "eval_runtime": 25.5362,
-      "eval_samples_per_second": 137.53,
-      "eval_steps_per_second": 4.308,
+      "eval_accuracy": 0.9837699316628702,
+      "eval_loss": 0.047338906675577164,
+      "eval_runtime": 31.1552,
+      "eval_samples_per_second": 112.726,
+      "eval_steps_per_second": 3.531,
       "step": 622
     },
     {
       "epoch": 1.01,
       "learning_rate": 0.00015948553054662381,
-      "loss": 0.1198,
+      "loss": 0.1207,
       "step": 630
     },
     {
       "epoch": 1.03,
       "learning_rate": 0.00015884244372990354,
-      "loss": 0.0857,
+      "loss": 0.0763,
       "step": 640
     },
     {
       "epoch": 1.05,
       "learning_rate": 0.0001581993569131833,
-      "loss": 0.1011,
+      "loss": 0.061,
       "step": 650
     },
     {
       "epoch": 1.06,
       "learning_rate": 0.00015755627009646304,
-      "loss": 0.0743,
+      "loss": 0.0728,
       "step": 660
     },
     {
       "epoch": 1.08,
       "learning_rate": 0.0001569131832797428,
-      "loss": 0.0917,
+      "loss": 0.0759,
       "step": 670
     },
     {
       "epoch": 1.09,
       "learning_rate": 0.0001562700964630225,
-      "loss": 0.0922,
+      "loss": 0.0715,
       "step": 680
     },
     {
       "epoch": 1.11,
       "learning_rate": 0.00015562700964630226,
-      "loss": 0.1241,
+      "loss": 0.0932,
       "step": 690
     },
     {
       "epoch": 1.13,
       "learning_rate": 0.00015498392282958201,
-      "loss": 0.087,
+      "loss": 0.1051,
       "step": 700
     },
     {
       "epoch": 1.14,
       "learning_rate": 0.00015434083601286174,
-      "loss": 0.0675,
+      "loss": 0.0537,
       "step": 710
     },
     {
       "epoch": 1.16,
       "learning_rate": 0.0001536977491961415,
-      "loss": 0.0733,
+      "loss": 0.0673,
       "step": 720
     },
     {
       "epoch": 1.17,
       "learning_rate": 0.00015305466237942124,
-      "loss": 0.0816,
+      "loss": 0.1123,
       "step": 730
     },
     {
       "epoch": 1.19,
       "learning_rate": 0.000152411575562701,
-      "loss": 0.1012,
+      "loss": 0.091,
       "step": 740
     },
     {
       "epoch": 1.21,
       "learning_rate": 0.0001517684887459807,
-      "loss": 0.0797,
+      "loss": 0.074,
       "step": 750
     },
     {
       "epoch": 1.22,
       "learning_rate": 0.00015112540192926046,
-      "loss": 0.0804,
+      "loss": 0.0867,
       "step": 760
     },
     {
       "epoch": 1.24,
       "learning_rate": 0.00015048231511254021,
-      "loss": 0.0917,
+      "loss": 0.0569,
       "step": 770
     },
     {
       "epoch": 1.25,
       "learning_rate": 0.00014983922829581994,
-      "loss": 0.1153,
+      "loss": 0.1289,
       "step": 780
     },
     {
       "epoch": 1.27,
       "learning_rate": 0.0001491961414790997,
-      "loss": 0.0476,
+      "loss": 0.0462,
       "step": 790
     },
     {
       "epoch": 1.29,
       "learning_rate": 0.00014855305466237944,
-      "loss": 0.0813,
+      "loss": 0.1009,
       "step": 800
     },
     {
       "epoch": 1.3,
       "learning_rate": 0.0001479099678456592,
-      "loss": 0.0923,
+      "loss": 0.1475,
       "step": 810
     },
     {
       "epoch": 1.32,
       "learning_rate": 0.00014726688102893891,
-      "loss": 0.0822,
+      "loss": 0.1175,
       "step": 820
     },
     {
       "epoch": 1.33,
       "learning_rate": 0.00014662379421221866,
-      "loss": 0.0579,
+      "loss": 0.1218,
       "step": 830
     },
     {
       "epoch": 1.35,
       "learning_rate": 0.00014598070739549841,
-      "loss": 0.1083,
+      "loss": 0.0977,
       "step": 840
     },
     {
       "epoch": 1.37,
       "learning_rate": 0.00014533762057877814,
-      "loss": 0.1207,
+      "loss": 0.1075,
       "step": 850
     },
     {
       "epoch": 1.38,
       "learning_rate": 0.0001446945337620579,
-      "loss": 0.0672,
+      "loss": 0.0632,
       "step": 860
     },
     {
       "epoch": 1.4,
       "learning_rate": 0.00014405144694533764,
-      "loss": 0.0978,
+      "loss": 0.0979,
       "step": 870
     },
     {
       "epoch": 1.41,
       "learning_rate": 0.0001434083601286174,
-      "loss": 0.1004,
+      "loss": 0.064,
       "step": 880
     },
     {
       "epoch": 1.43,
       "learning_rate": 0.00014276527331189711,
-      "loss": 0.0673,
+      "loss": 0.0336,
       "step": 890
     },
     {
       "epoch": 1.45,
       "learning_rate": 0.00014212218649517686,
-      "loss": 0.0641,
+      "loss": 0.1097,
       "step": 900
     },
     {
       "epoch": 1.46,
       "learning_rate": 0.00014147909967845662,
-      "loss": 0.045,
+      "loss": 0.0434,
       "step": 910
     },
     {
       "epoch": 1.48,
       "learning_rate": 0.00014083601286173634,
-      "loss": 0.0643,
+      "loss": 0.1313,
       "step": 920
     },
     {
       "epoch": 1.5,
       "learning_rate": 0.0001401929260450161,
-      "loss": 0.0643,
+      "loss": 0.0933,
       "step": 930
     },
     {
       "epoch": 1.51,
       "learning_rate": 0.00013954983922829584,
-      "loss": 0.1317,
+      "loss": 0.105,
       "step": 940
     },
     {
       "epoch": 1.53,
       "learning_rate": 0.0001389067524115756,
-      "loss": 0.0433,
+      "loss": 0.0737,
       "step": 950
     },
     {
       "epoch": 1.54,
       "learning_rate": 0.00013826366559485531,
-      "loss": 0.0862,
+      "loss": 0.1212,
       "step": 960
     },
     {
       "epoch": 1.56,
       "learning_rate": 0.00013762057877813507,
-      "loss": 0.1031,
+      "loss": 0.0744,
       "step": 970
     },
     {
       "epoch": 1.58,
       "learning_rate": 0.00013697749196141482,
-      "loss": 0.0849,
+      "loss": 0.0667,
       "step": 980
     },
     {
       "epoch": 1.59,
       "learning_rate": 0.00013633440514469454,
-      "loss": 0.0924,
+      "loss": 0.0966,
       "step": 990
     },
     {
       "epoch": 1.61,
       "learning_rate": 0.0001356913183279743,
-      "loss": 0.0773,
+      "loss": 0.0918,
       "step": 1000
     },
     {
       "epoch": 1.62,
       "learning_rate": 0.00013504823151125404,
-      "loss": 0.0823,
+      "loss": 0.073,
       "step": 1010
     },
     {
       "epoch": 1.64,
       "learning_rate": 0.00013440514469453376,
-      "loss": 0.0473,
+      "loss": 0.0516,
       "step": 1020
     },
     {
       "epoch": 1.66,
       "learning_rate": 0.00013376205787781351,
-      "loss": 0.1218,
+      "loss": 0.1177,
       "step": 1030
     },
     {
       "epoch": 1.67,
       "learning_rate": 0.00013311897106109327,
-      "loss": 0.0687,
+      "loss": 0.0663,
       "step": 1040
     },
     {
       "epoch": 1.69,
       "learning_rate": 0.00013247588424437302,
-      "loss": 0.0692,
+      "loss": 0.0422,
       "step": 1050
     },
     {
       "epoch": 1.7,
       "learning_rate": 0.00013183279742765274,
-      "loss": 0.0502,
+      "loss": 0.0598,
       "step": 1060
     },
     {
       "epoch": 1.72,
       "learning_rate": 0.0001311897106109325,
-      "loss": 0.0593,
+      "loss": 0.05,
       "step": 1070
     },
     {
       "epoch": 1.74,
       "learning_rate": 0.00013054662379421224,
-      "loss": 0.0846,
+      "loss": 0.1427,
       "step": 1080
     },
     {
       "epoch": 1.75,
       "learning_rate": 0.00012990353697749196,
-      "loss": 0.063,
+      "loss": 0.0538,
       "step": 1090
     },
     {
       "epoch": 1.77,
       "learning_rate": 0.00012926045016077172,
-      "loss": 0.0941,
+      "loss": 0.1008,
       "step": 1100
     },
     {
       "epoch": 1.78,
       "learning_rate": 0.00012861736334405147,
-      "loss": 0.0884,
+      "loss": 0.072,
       "step": 1110
     },
     {
       "epoch": 1.8,
       "learning_rate": 0.00012797427652733122,
-      "loss": 0.0647,
+      "loss": 0.0886,
       "step": 1120
     },
     {
       "epoch": 1.82,
       "learning_rate": 0.00012733118971061094,
-      "loss": 0.1016,
+      "loss": 0.0951,
       "step": 1130
     },
     {
       "epoch": 1.83,
       "learning_rate": 0.0001266881028938907,
-      "loss": 0.0821,
+      "loss": 0.0633,
       "step": 1140
     },
     {
       "epoch": 1.85,
       "learning_rate": 0.00012604501607717044,
-      "loss": 0.0653,
+      "loss": 0.0733,
       "step": 1150
     },
     {
       "epoch": 1.86,
       "learning_rate": 0.00012540192926045017,
-      "loss": 0.103,
+      "loss": 0.1003,
       "step": 1160
     },
     {
       "epoch": 1.88,
       "learning_rate": 0.00012475884244372992,
-      "loss": 0.1374,
+      "loss": 0.1359,
       "step": 1170
     },
     {
       "epoch": 1.9,
       "learning_rate": 0.00012411575562700967,
-      "loss": 0.0682,
+      "loss": 0.0661,
       "step": 1180
     },
     {
       "epoch": 1.91,
       "learning_rate": 0.00012347266881028942,
-      "loss": 0.0811,
+      "loss": 0.0589,
       "step": 1190
     },
     {
       "epoch": 1.93,
       "learning_rate": 0.00012282958199356914,
-      "loss": 0.0448,
+      "loss": 0.0347,
       "step": 1200
     },
     {
       "epoch": 1.95,
       "learning_rate": 0.0001221864951768489,
-      "loss": 0.1038,
+      "loss": 0.1009,
       "step": 1210
     },
     {
       "epoch": 1.96,
       "learning_rate": 0.00012154340836012863,
-      "loss": 0.0777,
+      "loss": 0.0787,
       "step": 1220
     },
     {
       "epoch": 1.98,
       "learning_rate": 0.00012090032154340835,
-      "loss": 0.0588,
+      "loss": 0.0839,
       "step": 1230
     },
     {
       "epoch": 1.99,
       "learning_rate": 0.0001202572347266881,
-      "loss": 0.057,
+      "loss": 0.0506,
       "step": 1240
     },
     {
       "epoch": 2.0,
       "eval_accuracy": 0.9883257403189066,
-      "eval_loss": 0.03622407466173172,
-      "eval_runtime": 25.825,
-      "eval_samples_per_second": 135.992,
-      "eval_steps_per_second": 4.259,
+      "eval_loss": 0.03498642519116402,
+      "eval_runtime": 31.2742,
+      "eval_samples_per_second": 112.297,
+      "eval_steps_per_second": 3.517,
       "step": 1244
     },
     {
       "epoch": 2.01,
       "learning_rate": 0.00011961414790996785,
-      "loss": 0.0314,
+      "loss": 0.0472,
       "step": 1250
     },
     {
       "epoch": 2.03,
       "learning_rate": 0.0001189710610932476,
-      "loss": 0.0833,
+      "loss": 0.12,
       "step": 1260
     },
     {
       "epoch": 2.04,
       "learning_rate": 0.00011832797427652733,
-      "loss": 0.0474,
+      "loss": 0.0727,
       "step": 1270
     },
     {
       "epoch": 2.06,
       "learning_rate": 0.00011768488745980708,
-      "loss": 0.0801,
+      "loss": 0.0621,
       "step": 1280
     },
     {
       "epoch": 2.07,
       "learning_rate": 0.00011704180064308683,
-      "loss": 0.0721,
+      "loss": 0.0553,
       "step": 1290
     },
     {
       "epoch": 2.09,
       "learning_rate": 0.00011639871382636655,
-      "loss": 0.0558,
+      "loss": 0.1022,
       "step": 1300
     },
     {
       "epoch": 2.11,
       "learning_rate": 0.0001157556270096463,
-      "loss": 0.0731,
+      "loss": 0.0737,
       "step": 1310
     },
     {
       "epoch": 2.12,
       "learning_rate": 0.00011511254019292605,
-      "loss": 0.0449,
+      "loss": 0.0632,
       "step": 1320
     },
     {
       "epoch": 2.14,
       "learning_rate": 0.0001144694533762058,
-      "loss": 0.0765,
+      "loss": 0.0549,
       "step": 1330
     },
     {
       "epoch": 2.15,
       "learning_rate": 0.00011382636655948553,
-      "loss": 0.0505,
+      "loss": 0.0765,
       "step": 1340
     },
     {
       "epoch": 2.17,
       "learning_rate": 0.00011318327974276528,
-      "loss": 0.0473,
+      "loss": 0.0316,
       "step": 1350
     },
     {
       "epoch": 2.19,
       "learning_rate": 0.00011254019292604503,
-      "loss": 0.0544,
+      "loss": 0.0535,
       "step": 1360
     },
     {
       "epoch": 2.2,
       "learning_rate": 0.00011189710610932475,
-      "loss": 0.0928,
+      "loss": 0.0866,
       "step": 1370
     },
     {
       "epoch": 2.22,
       "learning_rate": 0.0001112540192926045,
-      "loss": 0.0518,
+      "loss": 0.0246,
       "step": 1380
     },
     {
       "epoch": 2.23,
       "learning_rate": 0.00011061093247588425,
-      "loss": 0.1067,
+      "loss": 0.0814,
       "step": 1390
     },
     {
       "epoch": 2.25,
       "learning_rate": 0.00010996784565916398,
-      "loss": 0.0802,
+      "loss": 0.1007,
       "step": 1400
     },
     {
       "epoch": 2.27,
       "learning_rate": 0.00010932475884244373,
-      "loss": 0.0658,
+      "loss": 0.0501,
       "step": 1410
     },
     {
       "epoch": 2.28,
       "learning_rate": 0.00010868167202572348,
-      "loss": 0.0342,
+      "loss": 0.0836,
       "step": 1420
     },
     {
       "epoch": 2.3,
       "learning_rate": 0.00010803858520900323,
-      "loss": 0.0765,
+      "loss": 0.047,
       "step": 1430
     },
     {
       "epoch": 2.32,
       "learning_rate": 0.00010739549839228295,
-      "loss": 0.0315,
+      "loss": 0.042,
       "step": 1440
     },
     {
       "epoch": 2.33,
       "learning_rate": 0.0001067524115755627,
-      "loss": 0.0399,
+      "loss": 0.0575,
       "step": 1450
     },
     {
       "epoch": 2.35,
       "learning_rate": 0.00010610932475884245,
-      "loss": 0.092,
+      "loss": 0.0886,
       "step": 1460
     },
     {
       "epoch": 2.36,
       "learning_rate": 0.00010546623794212218,
-      "loss": 0.0543,
+      "loss": 0.0399,
       "step": 1470
     },
     {
       "epoch": 2.38,
       "learning_rate": 0.00010482315112540193,
-      "loss": 0.0485,
+      "loss": 0.0899,
       "step": 1480
     },
     {
       "epoch": 2.4,
       "learning_rate": 0.00010418006430868168,
-      "loss": 0.0629,
+      "loss": 0.0631,
       "step": 1490
     },
     {
       "epoch": 2.41,
       "learning_rate": 0.00010353697749196143,
-      "loss": 0.0724,
+      "loss": 0.0371,
       "step": 1500
     },
     {
       "epoch": 2.43,
       "learning_rate": 0.00010289389067524115,
-      "loss": 0.0591,
+      "loss": 0.0723,
       "step": 1510
     },
     {
       "epoch": 2.44,
       "learning_rate": 0.0001022508038585209,
-      "loss": 0.0704,
+      "loss": 0.0655,
       "step": 1520
     },
     {
       "epoch": 2.46,
       "learning_rate": 0.00010160771704180066,
-      "loss": 0.1163,
+      "loss": 0.1255,
       "step": 1530
     },
     {
       "epoch": 2.48,
       "learning_rate": 0.00010096463022508038,
-      "loss": 0.0403,
+      "loss": 0.0473,
       "step": 1540
     },
     {
       "epoch": 2.49,
       "learning_rate": 0.00010032154340836013,
-      "loss": 0.0641,
+      "loss": 0.0782,
       "step": 1550
     },
     {
       "epoch": 2.51,
       "learning_rate": 9.967845659163988e-05,
-      "loss": 0.0684,
+      "loss": 0.0478,
       "step": 1560
     },
     {
       "epoch": 2.52,
       "learning_rate": 9.903536977491962e-05,
-      "loss": 0.0534,
+      "loss": 0.0778,
       "step": 1570
     },
     {
       "epoch": 2.54,
       "learning_rate": 9.839228295819937e-05,
-      "loss": 0.0922,
+      "loss": 0.0893,
       "step": 1580
     },
     {
       "epoch": 2.56,
       "learning_rate": 9.77491961414791e-05,
-      "loss": 0.0368,
+      "loss": 0.0928,
       "step": 1590
     },
     {
       "epoch": 2.57,
       "learning_rate": 9.710610932475884e-05,
-      "loss": 0.0348,
+      "loss": 0.0437,
       "step": 1600
     },
     {
       "epoch": 2.59,
       "learning_rate": 9.646302250803859e-05,
-      "loss": 0.0355,
+      "loss": 0.0342,
       "step": 1610
     },
     {
       "epoch": 2.6,
       "learning_rate": 9.581993569131833e-05,
-      "loss": 0.0416,
+      "loss": 0.0603,
       "step": 1620
     },
     {
       "epoch": 2.62,
       "learning_rate": 9.517684887459808e-05,
-      "loss": 0.0874,
+      "loss": 0.0412,
       "step": 1630
     },
     {
       "epoch": 2.64,
       "learning_rate": 9.453376205787782e-05,
-      "loss": 0.0913,
+      "loss": 0.079,
       "step": 1640
     },
     {
       "epoch": 2.65,
       "learning_rate": 9.389067524115757e-05,
-      "loss": 0.0588,
+      "loss": 0.0347,
       "step": 1650
     },
     {
       "epoch": 2.67,
       "learning_rate": 9.32475884244373e-05,
-      "loss": 0.0639,
+      "loss": 0.1118,
       "step": 1660
     },
     {
       "epoch": 2.68,
       "learning_rate": 9.260450160771704e-05,
-      "loss": 0.0471,
+      "loss": 0.035,
       "step": 1670
     },
     {
       "epoch": 2.7,
       "learning_rate": 9.19614147909968e-05,
-      "loss": 0.0636,
+      "loss": 0.0512,
       "step": 1680
     },
     {
       "epoch": 2.72,
       "learning_rate": 9.131832797427653e-05,
-      "loss": 0.12,
+      "loss": 0.0572,
       "step": 1690
     },
     {
       "epoch": 2.73,
       "learning_rate": 9.067524115755628e-05,
-      "loss": 0.04,
+      "loss": 0.071,
       "step": 1700
     },
     {
       "epoch": 2.75,
       "learning_rate": 9.003215434083602e-05,
-      "loss": 0.0316,
+      "loss": 0.1231,
       "step": 1710
     },
     {
       "epoch": 2.77,
       "learning_rate": 8.938906752411576e-05,
-      "loss": 0.0506,
+      "loss": 0.0613,
       "step": 1720
     },
     {
       "epoch": 2.78,
       "learning_rate": 8.87459807073955e-05,
-      "loss": 0.0425,
+      "loss": 0.0505,
       "step": 1730
     },
     {
       "epoch": 2.8,
       "learning_rate": 8.810289389067524e-05,
-      "loss": 0.0131,
+      "loss": 0.0363,
       "step": 1740
     },
     {
       "epoch": 2.81,
       "learning_rate": 8.7459807073955e-05,
-      "loss": 0.0486,
+      "loss": 0.0294,
       "step": 1750
     },
     {
       "epoch": 2.83,
       "learning_rate": 8.681672025723473e-05,
-      "loss": 0.042,
+      "loss": 0.0283,
       "step": 1760
     },
     {
       "epoch": 2.85,
       "learning_rate": 8.617363344051448e-05,
-      "loss": 0.0524,
+      "loss": 0.083,
       "step": 1770
     },
     {
       "epoch": 2.86,
       "learning_rate": 8.553054662379422e-05,
-      "loss": 0.0415,
+      "loss": 0.0481,
       "step": 1780
     },
     {
       "epoch": 2.88,
       "learning_rate": 8.488745980707396e-05,
-      "loss": 0.0543,
+      "loss": 0.0398,
       "step": 1790
     },
     {
       "epoch": 2.89,
       "learning_rate": 8.42443729903537e-05,
-      "loss": 0.0679,
+      "loss": 0.0753,
       "step": 1800
     },
     {
       "epoch": 2.91,
       "learning_rate": 8.360128617363344e-05,
-      "loss": 0.1043,
+      "loss": 0.0707,
       "step": 1810
     },
     {
       "epoch": 2.93,
       "learning_rate": 8.29581993569132e-05,
-      "loss": 0.0508,
+      "loss": 0.0339,
       "step": 1820
     },
     {
       "epoch": 2.94,
       "learning_rate": 8.231511254019293e-05,
-      "loss": 0.0701,
+      "loss": 0.0977,
       "step": 1830
     },
     {
       "epoch": 2.96,
       "learning_rate": 8.167202572347268e-05,
-      "loss": 0.095,
+      "loss": 0.0759,
       "step": 1840
     },
     {
       "epoch": 2.97,
       "learning_rate": 8.102893890675242e-05,
-      "loss": 0.0853,
+      "loss": 0.061,
       "step": 1850
     },
     {
       "epoch": 2.99,
       "learning_rate": 8.038585209003216e-05,
-      "loss": 0.0449,
+      "loss": 0.06,
       "step": 1860
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.9886104783599089,
-      "eval_loss": 0.02609114721417427,
-      "eval_runtime": 25.2517,
-      "eval_samples_per_second": 139.08,
-      "eval_steps_per_second": 4.356,
+      "eval_accuracy": 0.9917425968109339,
+      "eval_loss": 0.023841001093387604,
+      "eval_runtime": 31.294,
+      "eval_samples_per_second": 112.226,
+      "eval_steps_per_second": 3.515,
       "step": 1866
     },
     {
       "epoch": 3.01,
       "learning_rate": 7.974276527331191e-05,
-      "loss": 0.0794,
+      "loss": 0.062,
       "step": 1870
     },
     {
       "epoch": 3.02,
       "learning_rate": 7.909967845659164e-05,
-      "loss": 0.03,
+      "loss": 0.0301,
       "step": 1880
     },
     {
       "epoch": 3.04,
       "learning_rate": 7.84565916398714e-05,
-      "loss": 0.0552,
+      "loss": 0.0461,
       "step": 1890
     },
     {
       "epoch": 3.05,
       "learning_rate": 7.781350482315113e-05,
-      "loss": 0.0577,
+      "loss": 0.0592,
       "step": 1900
     },
     {
       "epoch": 3.07,
       "learning_rate": 7.717041800643087e-05,
-      "loss": 0.054,
+      "loss": 0.0484,
       "step": 1910
     },
     {
       "epoch": 3.09,
       "learning_rate": 7.652733118971062e-05,
-      "loss": 0.0141,
+      "loss": 0.0217,
       "step": 1920
     },
     {
       "epoch": 3.1,
       "learning_rate": 7.588424437299036e-05,
-      "loss": 0.0579,
+      "loss": 0.0459,
       "step": 1930
     },
     {
       "epoch": 3.12,
       "learning_rate": 7.524115755627011e-05,
-      "loss": 0.0463,
+      "loss": 0.0265,
       "step": 1940
     },
     {
       "epoch": 3.14,
       "learning_rate": 7.459807073954984e-05,
-      "loss": 0.0389,
+      "loss": 0.0355,
       "step": 1950
     },
     {
       "epoch": 3.15,
       "learning_rate": 7.39549839228296e-05,
-      "loss": 0.0487,
+      "loss": 0.0846,
       "step": 1960
     },
     {
       "epoch": 3.17,
       "learning_rate": 7.331189710610933e-05,
-      "loss": 0.0246,
+      "loss": 0.0475,
       "step": 1970
     },
     {
       "epoch": 3.18,
       "learning_rate": 7.266881028938907e-05,
-      "loss": 0.0718,
+      "loss": 0.0785,
       "step": 1980
     },
     {
       "epoch": 3.2,
       "learning_rate": 7.202572347266882e-05,
-      "loss": 0.1136,
+      "loss": 0.0832,
       "step": 1990
     },
     {
       "epoch": 3.22,
       "learning_rate": 7.138263665594856e-05,
-      "loss": 0.0631,
+      "loss": 0.0602,
       "step": 2000
     },
     {
       "epoch": 3.23,
       "learning_rate": 7.073954983922831e-05,
-      "loss": 0.0363,
+      "loss": 0.0455,
       "step": 2010
     },
     {
       "epoch": 3.25,
       "learning_rate": 7.009646302250804e-05,
-      "loss": 0.075,
+      "loss": 0.0618,
       "step": 2020
     },
     {
       "epoch": 3.26,
       "learning_rate": 6.94533762057878e-05,
-      "loss": 0.0735,
+      "loss": 0.0493,
       "step": 2030
     },
     {
       "epoch": 3.28,
       "learning_rate": 6.881028938906753e-05,
-      "loss": 0.0461,
+      "loss": 0.0474,
       "step": 2040
     },
     {
       "epoch": 3.3,
       "learning_rate": 6.816720257234727e-05,
-      "loss": 0.0299,
+      "loss": 0.0343,
       "step": 2050
     },
     {
       "epoch": 3.31,
       "learning_rate": 6.752411575562702e-05,
-      "loss": 0.0526,
+      "loss": 0.0486,
       "step": 2060
     },
     {
       "epoch": 3.33,
       "learning_rate": 6.688102893890676e-05,
-      "loss": 0.0355,
+      "loss": 0.0555,
       "step": 2070
     },
     {
       "epoch": 3.34,
       "learning_rate": 6.623794212218651e-05,
-      "loss": 0.0244,
+      "loss": 0.0335,
       "step": 2080
     },
     {
       "epoch": 3.36,
       "learning_rate": 6.559485530546625e-05,
-      "loss": 0.0637,
+      "loss": 0.0562,
       "step": 2090
     },
     {
       "epoch": 3.38,
       "learning_rate": 6.495176848874598e-05,
-      "loss": 0.0595,
+      "loss": 0.0681,
       "step": 2100
     },
     {
       "epoch": 3.39,
       "learning_rate": 6.430868167202573e-05,
-      "loss": 0.0506,
+      "loss": 0.04,
       "step": 2110
     },
     {
       "epoch": 3.41,
       "learning_rate": 6.366559485530547e-05,
-      "loss": 0.0212,
+      "loss": 0.0296,
       "step": 2120
     },
     {
       "epoch": 3.42,
       "learning_rate": 6.302250803858522e-05,
-      "loss": 0.0255,
+      "loss": 0.0452,
       "step": 2130
     },
     {
       "epoch": 3.44,
       "learning_rate": 6.237942122186496e-05,
-      "loss": 0.0224,
+      "loss": 0.0331,
       "step": 2140
     },
     {
       "epoch": 3.46,
       "learning_rate": 6.173633440514471e-05,
-      "loss": 0.0633,
+      "loss": 0.041,
       "step": 2150
     },
     {
       "epoch": 3.47,
       "learning_rate": 6.109324758842445e-05,
-      "loss": 0.0554,
+      "loss": 0.0459,
       "step": 2160
     },
     {
       "epoch": 3.49,
       "learning_rate": 6.0450160771704176e-05,
-      "loss": 0.0653,
+      "loss": 0.0799,
       "step": 2170
     },
     {
       "epoch": 3.5,
       "learning_rate": 5.980707395498393e-05,
-      "loss": 0.02,
+      "loss": 0.0328,
       "step": 2180
     },
     {
       "epoch": 3.52,
       "learning_rate": 5.9163987138263664e-05,
-      "loss": 0.0668,
+      "loss": 0.0509,
       "step": 2190
     },
     {
       "epoch": 3.54,
       "learning_rate": 5.8520900321543414e-05,
-      "loss": 0.027,
+      "loss": 0.0375,
       "step": 2200
     },
     {
       "epoch": 3.55,
       "learning_rate": 5.787781350482315e-05,
-      "loss": 0.0567,
+      "loss": 0.0318,
       "step": 2210
     },
     {
       "epoch": 3.57,
       "learning_rate": 5.72347266881029e-05,
-      "loss": 0.0434,
+      "loss": 0.0324,
       "step": 2220
     },
     {
       "epoch": 3.59,
       "learning_rate": 5.659163987138264e-05,
-      "loss": 0.0457,
+      "loss": 0.0285,
       "step": 2230
     },
     {
       "epoch": 3.6,
       "learning_rate": 5.5948553054662377e-05,
-      "loss": 0.0263,
+      "loss": 0.0646,
       "step": 2240
     },
     {
       "epoch": 3.62,
       "learning_rate": 5.530546623794213e-05,
-      "loss": 0.0426,
+      "loss": 0.0312,
       "step": 2250
     },
     {
       "epoch": 3.63,
       "learning_rate": 5.4662379421221864e-05,
-      "loss": 0.0397,
+      "loss": 0.0478,
       "step": 2260
     },
     {
       "epoch": 3.65,
       "learning_rate": 5.4019292604501615e-05,
-      "loss": 0.0393,
+      "loss": 0.0448,
       "step": 2270
     },
     {
       "epoch": 3.67,
       "learning_rate": 5.337620578778135e-05,
-      "loss": 0.0453,
+      "loss": 0.032,
       "step": 2280
     },
     {
       "epoch": 3.68,
       "learning_rate": 5.273311897106109e-05,
-      "loss": 0.064,
+      "loss": 0.0364,
       "step": 2290
     },
     {
       "epoch": 3.7,
       "learning_rate": 5.209003215434084e-05,
-      "loss": 0.0587,
+      "loss": 0.0613,
       "step": 2300
     },
     {
       "epoch": 3.71,
       "learning_rate": 5.144694533762058e-05,
-      "loss": 0.0478,
+      "loss": 0.0605,
       "step": 2310
     },
     {
       "epoch": 3.73,
       "learning_rate": 5.080385852090033e-05,
-      "loss": 0.038,
+      "loss": 0.0229,
       "step": 2320
     },
     {
       "epoch": 3.75,
       "learning_rate": 5.0160771704180065e-05,
-      "loss": 0.0292,
+      "loss": 0.0456,
       "step": 2330
     },
     {
       "epoch": 3.76,
       "learning_rate": 4.951768488745981e-05,
-      "loss": 0.0412,
+      "loss": 0.0463,
       "step": 2340
     },
     {
       "epoch": 3.78,
       "learning_rate": 4.887459807073955e-05,
-      "loss": 0.0404,
+      "loss": 0.0627,
       "step": 2350
     },
     {
       "epoch": 3.79,
       "learning_rate": 4.8231511254019296e-05,
-      "loss": 0.0485,
+      "loss": 0.0815,
       "step": 2360
     },
     {
       "epoch": 3.81,
       "learning_rate": 4.758842443729904e-05,
-      "loss": 0.0373,
+      "loss": 0.0262,
       "step": 2370
     },
     {
       "epoch": 3.83,
       "learning_rate": 4.6945337620578784e-05,
-      "loss": 0.0291,
+      "loss": 0.0245,
       "step": 2380
     },
     {
       "epoch": 3.84,
       "learning_rate": 4.630225080385852e-05,
-      "loss": 0.0588,
+      "loss": 0.0435,
       "step": 2390
     },
     {
       "epoch": 3.86,
       "learning_rate": 4.5659163987138265e-05,
-      "loss": 0.0631,
+      "loss": 0.05,
       "step": 2400
     },
     {
       "epoch": 3.87,
       "learning_rate": 4.501607717041801e-05,
-      "loss": 0.0338,
+      "loss": 0.0412,
       "step": 2410
     },
     {
       "epoch": 3.89,
       "learning_rate": 4.437299035369775e-05,
-      "loss": 0.1125,
+      "loss": 0.0878,
       "step": 2420
     },
     {
       "epoch": 3.91,
       "learning_rate": 4.37299035369775e-05,
-      "loss": 0.026,
+      "loss": 0.0421,
       "step": 2430
     },
     {
       "epoch": 3.92,
       "learning_rate": 4.308681672025724e-05,
-      "loss": 0.0389,
+      "loss": 0.0481,
       "step": 2440
     },
     {
       "epoch": 3.94,
       "learning_rate": 4.244372990353698e-05,
-      "loss": 0.0567,
+      "loss": 0.0264,
       "step": 2450
     },
     {
       "epoch": 3.95,
       "learning_rate": 4.180064308681672e-05,
-      "loss": 0.023,
+      "loss": 0.0271,
       "step": 2460
     },
     {
       "epoch": 3.97,
       "learning_rate": 4.1157556270096466e-05,
-      "loss": 0.0353,
+      "loss": 0.0328,
       "step": 2470
     },
     {
       "epoch": 3.99,
       "learning_rate": 4.051446945337621e-05,
-      "loss": 0.066,
+      "loss": 0.0503,
       "step": 2480
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.9923120728929385,
-      "eval_loss": 0.02482709474861622,
-      "eval_runtime": 25.9153,
-      "eval_samples_per_second": 135.518,
-      "eval_steps_per_second": 4.245,
+      "eval_accuracy": 0.9908883826879271,
+      "eval_loss": 0.0295470729470253,
+      "eval_runtime": 31.0783,
+      "eval_samples_per_second": 113.005,
+      "eval_steps_per_second": 3.539,
       "step": 2488
     },
     {
       "epoch": 4.0,
       "learning_rate": 3.9871382636655953e-05,
-      "loss": 0.055,
+      "loss": 0.0507,
       "step": 2490
     },
     {
       "epoch": 4.02,
       "learning_rate": 3.92282958199357e-05,
-      "loss": 0.0344,
+      "loss": 0.0361,
       "step": 2500
     },
     {
       "epoch": 4.04,
       "learning_rate": 3.8585209003215434e-05,
-      "loss": 0.0635,
+      "loss": 0.0483,
       "step": 2510
     },
     {
       "epoch": 4.05,
       "learning_rate": 3.794212218649518e-05,
-      "loss": 0.0198,
+      "loss": 0.0385,
       "step": 2520
     },
     {
       "epoch": 4.07,
       "learning_rate": 3.729903536977492e-05,
-      "loss": 0.0268,
+      "loss": 0.037,
       "step": 2530
     },
     {
       "epoch": 4.08,
       "learning_rate": 3.6655948553054666e-05,
-      "loss": 0.0193,
+      "loss": 0.0326,
       "step": 2540
     },
     {
       "epoch": 4.1,
       "learning_rate": 3.601286173633441e-05,
-      "loss": 0.0285,
+      "loss": 0.053,
       "step": 2550
     },
     {
       "epoch": 4.12,
       "learning_rate": 3.5369774919614154e-05,
-      "loss": 0.0128,
+      "loss": 0.0067,
       "step": 2560
     },
     {
       "epoch": 4.13,
       "learning_rate": 3.47266881028939e-05,
-      "loss": 0.0252,
+      "loss": 0.0365,
       "step": 2570
     },
     {
       "epoch": 4.15,
       "learning_rate": 3.4083601286173635e-05,
-      "loss": 0.0078,
+      "loss": 0.0215,
       "step": 2580
     },
     {
       "epoch": 4.16,
       "learning_rate": 3.344051446945338e-05,
-      "loss": 0.0494,
+      "loss": 0.0591,
       "step": 2590
     },
     {
       "epoch": 4.18,
       "learning_rate": 3.279742765273312e-05,
-      "loss": 0.0459,
+      "loss": 0.024,
       "step": 2600
     },
     {
       "epoch": 4.2,
       "learning_rate": 3.2154340836012867e-05,
-      "loss": 0.0284,
+      "loss": 0.0249,
       "step": 2610
     },
     {
       "epoch": 4.21,
       "learning_rate": 3.151125401929261e-05,
-      "loss": 0.0423,
+      "loss": 0.0509,
       "step": 2620
     },
     {
       "epoch": 4.23,
       "learning_rate": 3.0868167202572354e-05,
-      "loss": 0.0094,
+      "loss": 0.0487,
       "step": 2630
     },
     {
       "epoch": 4.24,
       "learning_rate": 3.0225080385852088e-05,
-      "loss": 0.0371,
+      "loss": 0.0147,
       "step": 2640
     },
     {
       "epoch": 4.26,
       "learning_rate": 2.9581993569131832e-05,
-      "loss": 0.0241,
+      "loss": 0.0563,
       "step": 2650
     },
     {
       "epoch": 4.28,
       "learning_rate": 2.8938906752411576e-05,
-      "loss": 0.0497,
+      "loss": 0.0493,
       "step": 2660
     },
     {
       "epoch": 4.29,
       "learning_rate": 2.829581993569132e-05,
-      "loss": 0.026,
+      "loss": 0.0572,
       "step": 2670
     },
     {
       "epoch": 4.31,
       "learning_rate": 2.7652733118971064e-05,
-      "loss": 0.0638,
+      "loss": 0.0489,
       "step": 2680
     },
     {
       "epoch": 4.32,
       "learning_rate": 2.7009646302250807e-05,
-      "loss": 0.0201,
+      "loss": 0.018,
       "step": 2690
     },
     {
       "epoch": 4.34,
       "learning_rate": 2.6366559485530545e-05,
-      "loss": 0.0348,
+      "loss": 0.0281,
       "step": 2700
     },
     {
       "epoch": 4.36,
-      "learning_rate": 2.5787781350482317e-05,
-      "loss": 0.043,
+      "learning_rate": 2.572347266881029e-05,
+      "loss": 0.0307,
       "step": 2710
     },
     {
       "epoch": 4.37,
-      "learning_rate": 2.5144694533762057e-05,
-      "loss": 0.0371,
+      "learning_rate": 2.5080385852090032e-05,
+      "loss": 0.0641,
       "step": 2720
     },
     {
       "epoch": 4.39,
-      "learning_rate": 2.45016077170418e-05,
-      "loss": 0.0164,
+      "learning_rate": 2.4437299035369776e-05,
+      "loss": 0.0433,
       "step": 2730
     },
     {
       "epoch": 4.41,
-      "learning_rate": 2.385852090032154e-05,
-      "loss": 0.0359,
+      "learning_rate": 2.379421221864952e-05,
+      "loss": 0.0193,
       "step": 2740
     },
     {
       "epoch": 4.42,
-      "learning_rate": 2.3215434083601286e-05,
-      "loss": 0.0556,
+      "learning_rate": 2.315112540192926e-05,
+      "loss": 0.0271,
       "step": 2750
     },
     {
       "epoch": 4.44,
-      "learning_rate": 2.257234726688103e-05,
-      "loss": 0.0247,
+      "learning_rate": 2.2508038585209005e-05,
+      "loss": 0.0196,
       "step": 2760
     },
     {
       "epoch": 4.45,
-      "learning_rate": 2.1929260450160773e-05,
-      "loss": 0.0477,
+      "learning_rate": 2.186495176848875e-05,
+      "loss": 0.0482,
       "step": 2770
     },
     {
       "epoch": 4.47,
-      "learning_rate": 2.1286173633440514e-05,
-      "loss": 0.0499,
+      "learning_rate": 2.122186495176849e-05,
+      "loss": 0.0338,
       "step": 2780
     },
     {
       "epoch": 4.49,
-      "learning_rate": 2.0643086816720258e-05,
-      "loss": 0.0736,
+      "learning_rate": 2.0578778135048233e-05,
+      "loss": 0.0431,
       "step": 2790
     },
     {
       "epoch": 4.5,
-      "learning_rate": 2e-05,
-      "loss": 0.0235,
+      "learning_rate": 1.9935691318327977e-05,
+      "loss": 0.033,
       "step": 2800
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.9356913183279742e-05,
-      "loss": 0.0368,
+      "learning_rate": 1.9292604501607717e-05,
+      "loss": 0.0275,
       "step": 2810
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.8713826366559486e-05,
-      "loss": 0.013,
+      "learning_rate": 1.864951768488746e-05,
+      "loss": 0.0283,
       "step": 2820
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.807073954983923e-05,
-      "loss": 0.0573,
+      "learning_rate": 1.8006430868167205e-05,
+      "loss": 0.0208,
       "step": 2830
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.742765273311897e-05,
-      "loss": 0.0268,
+      "learning_rate": 1.736334405144695e-05,
+      "loss": 0.0344,
       "step": 2840
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.6784565916398714e-05,
-      "loss": 0.0229,
+      "learning_rate": 1.672025723472669e-05,
+      "loss": 0.0095,
       "step": 2850
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.6141479099678458e-05,
-      "loss": 0.0294,
+      "learning_rate": 1.6077170418006433e-05,
+      "loss": 0.0333,
       "step": 2860
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.54983922829582e-05,
-      "loss": 0.0619,
+      "learning_rate": 1.5434083601286177e-05,
+      "loss": 0.054,
       "step": 2870
     },
     {
       "epoch": 4.63,
-      "learning_rate": 1.4855305466237943e-05,
-      "loss": 0.0232,
+      "learning_rate": 1.4790996784565916e-05,
+      "loss": 0.0153,
       "step": 2880
     },
     {
       "epoch": 4.65,
-      "learning_rate": 1.4212218649517686e-05,
-      "loss": 0.0388,
+      "learning_rate": 1.414790996784566e-05,
+      "loss": 0.0373,
       "step": 2890
     },
     {
       "epoch": 4.66,
-      "learning_rate": 1.3569131832797427e-05,
-      "loss": 0.0133,
+      "learning_rate": 1.3504823151125404e-05,
+      "loss": 0.0124,
       "step": 2900
     },
     {
       "epoch": 4.68,
-      "learning_rate": 1.2926045016077171e-05,
+      "learning_rate": 1.2861736334405144e-05,
       "loss": 0.0588,
       "step": 2910
     },
     {
       "epoch": 4.69,
-      "learning_rate": 1.2282958199356915e-05,
-      "loss": 0.0394,
+      "learning_rate": 1.2218649517684888e-05,
+      "loss": 0.0341,
       "step": 2920
     },
     {
       "epoch": 4.71,
-      "learning_rate": 1.1639871382636657e-05,
-      "loss": 0.0253,
+      "learning_rate": 1.157556270096463e-05,
+      "loss": 0.0295,
       "step": 2930
     },
     {
       "epoch": 4.73,
-      "learning_rate": 1.0996784565916399e-05,
-      "loss": 0.0292,
+      "learning_rate": 1.0932475884244374e-05,
+      "loss": 0.0224,
       "step": 2940
     },
     {
       "epoch": 4.74,
-      "learning_rate": 1.0353697749196143e-05,
-      "loss": 0.027,
+      "learning_rate": 1.0289389067524116e-05,
+      "loss": 0.0286,
       "step": 2950
     },
     {
       "epoch": 4.76,
-      "learning_rate": 9.710610932475885e-06,
-      "loss": 0.0317,
+      "learning_rate": 9.646302250803859e-06,
+      "loss": 0.0278,
       "step": 2960
     },
     {
       "epoch": 4.77,
-      "learning_rate": 9.067524115755627e-06,
-      "loss": 0.0326,
+      "learning_rate": 9.003215434083602e-06,
+      "loss": 0.0286,
       "step": 2970
     },
     {
       "epoch": 4.79,
-      "learning_rate": 8.42443729903537e-06,
-      "loss": 0.0508,
+      "learning_rate": 8.360128617363345e-06,
+      "loss": 0.0497,
       "step": 2980
     },
     {
       "epoch": 4.81,
-      "learning_rate": 7.781350482315112e-06,
-      "loss": 0.0182,
+      "learning_rate": 7.717041800643089e-06,
+      "loss": 0.0153,
       "step": 2990
     },
     {
       "epoch": 4.82,
-      "learning_rate": 7.1382636655948566e-06,
-      "loss": 0.0255,
+      "learning_rate": 7.07395498392283e-06,
+      "loss": 0.0178,
       "step": 3000
     },
     {
       "epoch": 4.84,
-      "learning_rate": 6.495176848874599e-06,
-      "loss": 0.0199,
+      "learning_rate": 6.430868167202572e-06,
+      "loss": 0.0242,
       "step": 3010
     },
     {
       "epoch": 4.86,
-      "learning_rate": 5.852090032154341e-06,
-      "loss": 0.0206,
+      "learning_rate": 5.787781350482315e-06,
+      "loss": 0.0283,
       "step": 3020
     },
     {
       "epoch": 4.87,
-      "learning_rate": 5.209003215434084e-06,
-      "loss": 0.0352,
+      "learning_rate": 5.144694533762058e-06,
+      "loss": 0.0427,
       "step": 3030
     },
     {
       "epoch": 4.89,
-      "learning_rate": 4.565916398713826e-06,
-      "loss": 0.0362,
+      "learning_rate": 4.501607717041801e-06,
+      "loss": 0.021,
       "step": 3040
     },
     {
       "epoch": 4.9,
-      "learning_rate": 3.922829581993569e-06,
-      "loss": 0.0373,
+      "learning_rate": 3.858520900321544e-06,
+      "loss": 0.0311,
       "step": 3050
     },
     {
       "epoch": 4.92,
-      "learning_rate": 3.2797427652733123e-06,
-      "loss": 0.0279,
+      "learning_rate": 3.215434083601286e-06,
+      "loss": 0.0531,
       "step": 3060
     },
     {
       "epoch": 4.94,
-      "learning_rate": 2.636655948553055e-06,
-      "loss": 0.0286,
+      "learning_rate": 2.572347266881029e-06,
+      "loss": 0.0214,
       "step": 3070
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.9935691318327975e-06,
-      "loss": 0.0359,
+      "learning_rate": 1.929260450160772e-06,
+      "loss": 0.0314,
       "step": 3080
     },
     {
       "epoch": 4.97,
-      "learning_rate": 1.3504823151125401e-06,
-      "loss": 0.0382,
+      "learning_rate": 1.2861736334405146e-06,
+      "loss": 0.0341,
       "step": 3090
     },
     {
       "epoch": 4.98,
-      "learning_rate": 7.073954983922831e-07,
-      "loss": 0.0229,
+      "learning_rate": 6.430868167202573e-07,
+      "loss": 0.0185,
       "step": 3100
     },
     {
       "epoch": 5.0,
-      "learning_rate": 6.430868167202573e-08,
-      "loss": 0.0328,
+      "learning_rate": 0.0,
+      "loss": 0.0314,
       "step": 3110
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.9937357630979499,
-      "eval_loss": 0.0182377677410841,
-      "eval_runtime": 25.6312,
-      "eval_samples_per_second": 137.021,
-      "eval_steps_per_second": 4.292,
+      "eval_accuracy": 0.9957289293849658,
+      "eval_loss": 0.013703458942472935,
+      "eval_runtime": 31.6284,
+      "eval_samples_per_second": 111.04,
+      "eval_steps_per_second": 3.478,
       "step": 3110
     },
     {
       "epoch": 5.0,
       "step": 3110,
       "total_flos": 0.0,
-      "train_loss": 0.06854689962371369,
-      "train_runtime": 1208.8541,
-      "train_samples_per_second": 82.301,
-      "train_steps_per_second": 2.573
+      "train_loss": 0.07020007294761406,
+      "train_runtime": 1821.7529,
+      "train_samples_per_second": 54.612,
+      "train_steps_per_second": 1.707
     }
   ],
   "max_steps": 3110,