Training in progress, epoch 1

all_results.json

@@ -1,16 +1,16 @@
 {
-    "epoch": 4.0,
-    "eval_accuracy": 0.8029100529100529,
-    "eval_f1": 0.7964648989688592,
-    "eval_loss": 0.8055053353309631,
-    "eval_precision": 0.797659702180121,
-    "eval_recall": 0.8029100529100529,
-    "eval_runtime": 8.9204,
-    "eval_samples_per_second": 169.499,
-    "eval_steps_per_second": 10.65,
+    "epoch": 6.0,
+    "eval_accuracy": 0.794973544973545,
+    "eval_f1": 0.7911822260194087,
+    "eval_loss": 0.8158386945724487,
+    "eval_precision": 0.7898693796510783,
+    "eval_recall": 0.794973544973545,
+    "eval_runtime": 8.8626,
+    "eval_samples_per_second": 170.604,
+    "eval_steps_per_second": 10.719,
     "total_flos": 3.104468219559813e+18,
-    "train_loss": 0.07359402208736303,
-    "train_runtime": 853.5389,
-    "train_samples_per_second": 46.934,
-    "train_steps_per_second": 2.934
+    "train_loss": 0.07414248911174177,
+    "train_runtime": 1259.2386,
+    "train_samples_per_second": 47.719,
+    "train_steps_per_second": 2.983
 }

eval_results.json

@@ -1,11 +1,11 @@
 {
-    "epoch": 4.0,
-    "eval_accuracy": 0.8029100529100529,
-    "eval_f1": 0.7964648989688592,
-    "eval_loss": 0.8055053353309631,
-    "eval_precision": 0.797659702180121,
-    "eval_recall": 0.8029100529100529,
-    "eval_runtime": 8.9204,
-    "eval_samples_per_second": 169.499,
-    "eval_steps_per_second": 10.65
+    "epoch": 6.0,
+    "eval_accuracy": 0.794973544973545,
+    "eval_f1": 0.7911822260194087,
+    "eval_loss": 0.8158386945724487,
+    "eval_precision": 0.7898693796510783,
+    "eval_recall": 0.794973544973545,
+    "eval_runtime": 8.8626,
+    "eval_samples_per_second": 170.604,
+    "eval_steps_per_second": 10.719
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f0c9ddce22e8203fe9e31c613251334d0fa8644edeef89eba6ad003abcfe379e
+oid sha256:2e1014dfea819e7210421dbff6a76e2e50982b8a3ef3654b5d84dca7c7eaae57
 size 343284077

runs/Oct26_16-14-30_EE4E077/events.out.tfevents.1698309421.EE4E077.4972.5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e1ed954d118a23fc8f1085a0099eda82cbff5af44a5cadc8e2ce162562a1d2b1
+size 560

runs/Oct26_22-22-35_EE4E077/1698330159.2987473/events.out.tfevents.1698330159.EE4E077.66330.4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:da61e91f6267495bbbb15b83e118cd965b168ad39ea2a8fea0d6bf56fbaf1d31
+size 5919

runs/Oct26_22-22-35_EE4E077/events.out.tfevents.1698330159.EE4E077.66330.3

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:39ba04b1244cc46efcfad7980d1e4165eee0e5f2693b5d3a6454367f6ac576e3
+size 14419

train_results.json

@@ -1,7 +1,7 @@
 {
-    "epoch": 4.0,
-    "train_loss": 0.07359402208736303,
-    "train_runtime": 853.5389,
-    "train_samples_per_second": 46.934,
-    "train_steps_per_second": 2.934
+    "epoch": 6.0,
+    "train_loss": 0.07414248911174177,
+    "train_runtime": 1259.2386,
+    "train_samples_per_second": 47.719,
+    "train_steps_per_second": 2.983
 }

trainer_state.json

@@ -1,1573 +1,2347 @@
 {
-  "best_metric": 0.8652849740932642,
+  "best_metric": 0.8601036269430051,
   "best_model_checkpoint": "./vit-focal-skin/checkpoint-2504",
-  "epoch": 4.0,
-  "global_step": 2504,
+  "epoch": 6.0,
+  "global_step": 3756,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
       "epoch": 0.02,
-      "learning_rate": 0.00019920127795527157,
-      "loss": 0.0489,
+      "learning_rate": 0.0001994675186368477,
+      "loss": 0.3361,
       "step": 10
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.00019840255591054313,
-      "loss": 0.2891,
+      "learning_rate": 0.00019893503727369543,
+      "loss": 0.3892,
       "step": 20
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00019760383386581472,
-      "loss": 0.1078,
+      "learning_rate": 0.00019840255591054313,
+      "loss": 0.1894,
       "step": 30
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.00019680511182108628,
-      "loss": 0.2355,
+      "learning_rate": 0.00019787007454739086,
+      "loss": 0.386,
       "step": 40
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00019600638977635784,
-      "loss": 0.1737,
+      "learning_rate": 0.00019733759318423856,
+      "loss": 0.2757,
       "step": 50
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.0001952076677316294,
-      "loss": 0.2375,
+      "learning_rate": 0.00019680511182108628,
+      "loss": 0.2467,
       "step": 60
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00019440894568690097,
-      "loss": 0.2485,
+      "learning_rate": 0.00019627263045793398,
+      "loss": 0.4231,
       "step": 70
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00019361022364217253,
-      "loss": 0.2195,
+      "learning_rate": 0.00019574014909478168,
+      "loss": 0.3952,
       "step": 80
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001928115015974441,
-      "loss": 0.188,
+      "learning_rate": 0.0001952076677316294,
+      "loss": 0.3219,
       "step": 90
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00019201277955271565,
-      "loss": 0.1955,
+      "learning_rate": 0.0001946751863684771,
+      "loss": 0.2279,
       "step": 100
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0001912140575079872,
-      "loss": 0.3527,
+      "learning_rate": 0.00019414270500532483,
+      "loss": 0.3248,
       "step": 110
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0001904153354632588,
-      "loss": 0.1645,
+      "learning_rate": 0.00019361022364217253,
+      "loss": 0.2373,
       "step": 120
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00018961661341853036,
-      "loss": 0.1028,
+      "learning_rate": 0.00019307774227902025,
+      "loss": 0.2241,
       "step": 130
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00018881789137380192,
-      "loss": 0.1078,
+      "learning_rate": 0.00019254526091586795,
+      "loss": 0.2499,
       "step": 140
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00018801916932907348,
-      "loss": 0.2549,
+      "learning_rate": 0.00019201277955271565,
+      "loss": 0.24,
       "step": 150
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00018722044728434505,
-      "loss": 0.2236,
+      "learning_rate": 0.00019148029818956338,
+      "loss": 0.2011,
       "step": 160
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00018642172523961663,
-      "loss": 0.1437,
+      "learning_rate": 0.00019094781682641108,
+      "loss": 0.1793,
       "step": 170
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0001856230031948882,
-      "loss": 0.1704,
+      "learning_rate": 0.0001904153354632588,
+      "loss": 0.3061,
       "step": 180
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00018482428115015976,
-      "loss": 0.2953,
+      "learning_rate": 0.0001898828541001065,
+      "loss": 0.2371,
       "step": 190
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00018402555910543132,
-      "loss": 0.2462,
+      "learning_rate": 0.00018935037273695422,
+      "loss": 0.322,
       "step": 200
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00018322683706070288,
-      "loss": 0.2243,
+      "learning_rate": 0.00018881789137380192,
+      "loss": 0.251,
       "step": 210
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00018242811501597444,
-      "loss": 0.141,
+      "learning_rate": 0.00018828541001064962,
+      "loss": 0.2188,
       "step": 220
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.000181629392971246,
-      "loss": 0.1903,
+      "learning_rate": 0.00018775292864749735,
+      "loss": 0.2617,
       "step": 230
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00018083067092651756,
-      "loss": 0.2194,
+      "learning_rate": 0.00018722044728434505,
+      "loss": 0.2521,
       "step": 240
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00018003194888178913,
-      "loss": 0.1511,
+      "learning_rate": 0.00018668796592119277,
+      "loss": 0.143,
       "step": 250
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00017923322683706071,
-      "loss": 0.1778,
+      "learning_rate": 0.00018615548455804047,
+      "loss": 0.2335,
       "step": 260
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00017843450479233228,
-      "loss": 0.1996,
+      "learning_rate": 0.0001856230031948882,
+      "loss": 0.2898,
       "step": 270
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.00017763578274760384,
-      "loss": 0.2369,
+      "learning_rate": 0.0001850905218317359,
+      "loss": 0.235,
       "step": 280
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0001768370607028754,
-      "loss": 0.2485,
+      "learning_rate": 0.0001845580404685836,
+      "loss": 0.2792,
       "step": 290
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.000176038338658147,
-      "loss": 0.1215,
+      "learning_rate": 0.00018402555910543132,
+      "loss": 0.1807,
       "step": 300
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00017523961661341855,
-      "loss": 0.1876,
+      "learning_rate": 0.00018349307774227902,
+      "loss": 0.2124,
       "step": 310
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0001744408945686901,
-      "loss": 0.0868,
+      "learning_rate": 0.00018296059637912674,
+      "loss": 0.1804,
       "step": 320
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00017364217252396167,
-      "loss": 0.1897,
+      "learning_rate": 0.00018242811501597444,
+      "loss": 0.2651,
       "step": 330
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00017284345047923323,
-      "loss": 0.1917,
+      "learning_rate": 0.00018189563365282217,
+      "loss": 0.3447,
       "step": 340
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00017204472843450482,
-      "loss": 0.1873,
+      "learning_rate": 0.00018136315228966987,
+      "loss": 0.2006,
       "step": 350
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00017124600638977638,
-      "loss": 0.1629,
+      "learning_rate": 0.00018083067092651756,
+      "loss": 0.2814,
       "step": 360
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00017044728434504792,
-      "loss": 0.0677,
+      "learning_rate": 0.0001802981895633653,
+      "loss": 0.1605,
       "step": 370
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00016964856230031948,
-      "loss": 0.0944,
+      "learning_rate": 0.00017976570820021302,
+      "loss": 0.2397,
       "step": 380
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.00016884984025559104,
-      "loss": 0.1851,
+      "learning_rate": 0.00017923322683706071,
+      "loss": 0.2248,
       "step": 390
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00016805111821086263,
-      "loss": 0.2105,
+      "learning_rate": 0.0001787007454739084,
+      "loss": 0.2866,
       "step": 400
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0001672523961661342,
-      "loss": 0.1481,
+      "learning_rate": 0.00017816826411075614,
+      "loss": 0.1921,
       "step": 410
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00016645367412140575,
-      "loss": 0.1866,
+      "learning_rate": 0.00017763578274760384,
+      "loss": 0.3192,
       "step": 420
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0001656549520766773,
-      "loss": 0.1909,
+      "learning_rate": 0.00017710330138445154,
+      "loss": 0.2548,
       "step": 430
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0001648562300319489,
-      "loss": 0.1715,
+      "learning_rate": 0.00017657082002129926,
+      "loss": 0.228,
       "step": 440
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00016405750798722046,
-      "loss": 0.1671,
+      "learning_rate": 0.000176038338658147,
+      "loss": 0.1783,
       "step": 450
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00016325878594249202,
-      "loss": 0.145,
+      "learning_rate": 0.00017550585729499469,
+      "loss": 0.1573,
       "step": 460
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00016246006389776359,
-      "loss": 0.2137,
+      "learning_rate": 0.0001749733759318424,
+      "loss": 0.2973,
       "step": 470
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00016166134185303515,
-      "loss": 0.1831,
+      "learning_rate": 0.0001744408945686901,
+      "loss": 0.2374,
       "step": 480
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00016086261980830674,
-      "loss": 0.2028,
+      "learning_rate": 0.0001739084132055378,
+      "loss": 0.2664,
       "step": 490
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0001600638977635783,
-      "loss": 0.1216,
+      "learning_rate": 0.0001733759318423855,
+      "loss": 0.1829,
       "step": 500
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00015926517571884986,
-      "loss": 0.195,
+      "learning_rate": 0.00017284345047923323,
+      "loss": 0.1979,
       "step": 510
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00015846645367412142,
-      "loss": 0.1339,
+      "learning_rate": 0.00017231096911608096,
+      "loss": 0.1509,
       "step": 520
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00015766773162939298,
-      "loss": 0.1731,
+      "learning_rate": 0.00017177848775292866,
+      "loss": 0.164,
       "step": 530
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00015686900958466454,
-      "loss": 0.2264,
+      "learning_rate": 0.00017124600638977638,
+      "loss": 0.2125,
       "step": 540
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0001560702875399361,
-      "loss": 0.18,
+      "learning_rate": 0.00017071352502662408,
+      "loss": 0.2712,
       "step": 550
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00015527156549520767,
-      "loss": 0.1709,
+      "learning_rate": 0.00017018104366347178,
+      "loss": 0.1882,
       "step": 560
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00015447284345047923,
-      "loss": 0.0934,
+      "learning_rate": 0.00016964856230031948,
+      "loss": 0.1679,
       "step": 570
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00015367412140575082,
-      "loss": 0.1188,
+      "learning_rate": 0.0001691160809371672,
+      "loss": 0.2256,
       "step": 580
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00015287539936102238,
-      "loss": 0.1123,
+      "learning_rate": 0.00016858359957401493,
+      "loss": 0.2308,
       "step": 590
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00015207667731629394,
-      "loss": 0.2367,
+      "learning_rate": 0.00016805111821086263,
+      "loss": 0.226,
       "step": 600
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0001512779552715655,
-      "loss": 0.2077,
+      "learning_rate": 0.00016751863684771035,
+      "loss": 0.2933,
       "step": 610
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00015047923322683706,
-      "loss": 0.1373,
+      "learning_rate": 0.00016698615548455805,
+      "loss": 0.1753,
       "step": 620
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.8393782383419689,
-      "eval_f1": 0.8433863334540076,
-      "eval_loss": 0.36540716886520386,
-      "eval_precision": 0.8542463099229327,
-      "eval_recall": 0.8393782383419689,
-      "eval_runtime": 1.3043,
-      "eval_samples_per_second": 147.967,
-      "eval_steps_per_second": 9.967,
+      "eval_accuracy": 0.8290155440414507,
+      "eval_f1": 0.8322633623881244,
+      "eval_loss": 0.37183406949043274,
+      "eval_precision": 0.8512730478816207,
+      "eval_recall": 0.8290155440414507,
+      "eval_runtime": 1.3128,
+      "eval_samples_per_second": 147.011,
+      "eval_steps_per_second": 9.902,
       "step": 626
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00014968051118210865,
-      "loss": 0.1352,
+      "learning_rate": 0.00016645367412140575,
+      "loss": 0.1791,
       "step": 630
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0001488817891373802,
-      "loss": 0.0669,
+      "learning_rate": 0.00016592119275825345,
+      "loss": 0.1494,
       "step": 640
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00014808306709265177,
-      "loss": 0.2028,
+      "learning_rate": 0.00016538871139510118,
+      "loss": 0.1231,
       "step": 650
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00014728434504792333,
-      "loss": 0.1348,
+      "learning_rate": 0.0001648562300319489,
+      "loss": 0.1142,
       "step": 660
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0001464856230031949,
-      "loss": 0.1204,
+      "learning_rate": 0.0001643237486687966,
+      "loss": 0.1562,
       "step": 670
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00014568690095846646,
-      "loss": 0.0973,
+      "learning_rate": 0.00016379126730564433,
+      "loss": 0.1256,
       "step": 680
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00014488817891373802,
-      "loss": 0.0251,
+      "learning_rate": 0.00016325878594249202,
+      "loss": 0.1549,
       "step": 690
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00014408945686900958,
-      "loss": 0.0694,
+      "learning_rate": 0.00016272630457933972,
+      "loss": 0.0852,
       "step": 700
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00014329073482428114,
-      "loss": 0.1262,
+      "learning_rate": 0.00016219382321618745,
+      "loss": 0.1539,
       "step": 710
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00014249201277955273,
-      "loss": 0.0949,
+      "learning_rate": 0.00016166134185303515,
+      "loss": 0.1054,
       "step": 720
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0001416932907348243,
-      "loss": 0.1198,
+      "learning_rate": 0.00016112886048988287,
+      "loss": 0.1317,
       "step": 730
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00014089456869009585,
-      "loss": 0.0585,
+      "learning_rate": 0.00016059637912673057,
+      "loss": 0.1008,
       "step": 740
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001400958466453674,
-      "loss": 0.1276,
+      "learning_rate": 0.0001600638977635783,
+      "loss": 0.1299,
       "step": 750
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.000139297124600639,
-      "loss": 0.0937,
+      "learning_rate": 0.000159531416400426,
+      "loss": 0.071,
       "step": 760
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00013849840255591056,
-      "loss": 0.1372,
+      "learning_rate": 0.0001589989350372737,
+      "loss": 0.0792,
       "step": 770
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00013769968051118212,
-      "loss": 0.0717,
+      "learning_rate": 0.00015846645367412142,
+      "loss": 0.1526,
       "step": 780
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00013690095846645369,
-      "loss": 0.0297,
+      "learning_rate": 0.00015793397231096912,
+      "loss": 0.0794,
       "step": 790
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00013610223642172525,
-      "loss": 0.0343,
+      "learning_rate": 0.00015740149094781684,
+      "loss": 0.199,
       "step": 800
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0001353035143769968,
-      "loss": 0.1294,
+      "learning_rate": 0.00015686900958466454,
+      "loss": 0.2173,
       "step": 810
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00013450479233226837,
-      "loss": 0.0841,
+      "learning_rate": 0.00015633652822151227,
+      "loss": 0.1188,
       "step": 820
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00013370607028753993,
-      "loss": 0.081,
+      "learning_rate": 0.00015580404685835997,
+      "loss": 0.084,
       "step": 830
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0001329073482428115,
-      "loss": 0.0985,
+      "learning_rate": 0.00015527156549520767,
+      "loss": 0.1286,
       "step": 840
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00013210862619808305,
-      "loss": 0.1692,
+      "learning_rate": 0.0001547390841320554,
+      "loss": 0.1635,
       "step": 850
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00013130990415335464,
-      "loss": 0.1196,
+      "learning_rate": 0.0001542066027689031,
+      "loss": 0.1956,
       "step": 860
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0001305111821086262,
-      "loss": 0.0792,
+      "learning_rate": 0.00015367412140575082,
+      "loss": 0.222,
       "step": 870
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00012971246006389777,
-      "loss": 0.0373,
+      "learning_rate": 0.00015314164004259851,
+      "loss": 0.0591,
       "step": 880
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00012891373801916933,
-      "loss": 0.0802,
+      "learning_rate": 0.00015260915867944624,
+      "loss": 0.2126,
       "step": 890
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00012811501597444092,
-      "loss": 0.0248,
+      "learning_rate": 0.00015207667731629394,
+      "loss": 0.1096,
       "step": 900
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00012731629392971248,
-      "loss": 0.07,
+      "learning_rate": 0.00015154419595314164,
+      "loss": 0.108,
       "step": 910
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00012651757188498404,
-      "loss": 0.0942,
+      "learning_rate": 0.00015101171458998936,
+      "loss": 0.1241,
       "step": 920
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0001257188498402556,
-      "loss": 0.0543,
+      "learning_rate": 0.00015047923322683706,
+      "loss": 0.1105,
       "step": 930
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00012492012779552716,
-      "loss": 0.1837,
+      "learning_rate": 0.0001499467518636848,
+      "loss": 0.1555,
       "step": 940
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00012412140575079872,
-      "loss": 0.0377,
+      "learning_rate": 0.00014941427050053249,
+      "loss": 0.0907,
       "step": 950
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00012332268370607028,
-      "loss": 0.106,
+      "learning_rate": 0.0001488817891373802,
+      "loss": 0.0593,
       "step": 960
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00012252396166134185,
-      "loss": 0.0845,
+      "learning_rate": 0.0001483493077742279,
+      "loss": 0.0926,
       "step": 970
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00012172523961661342,
-      "loss": 0.0631,
+      "learning_rate": 0.0001478168264110756,
+      "loss": 0.0841,
       "step": 980
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.000120926517571885,
-      "loss": 0.0882,
+      "learning_rate": 0.00014728434504792333,
+      "loss": 0.1627,
       "step": 990
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00012012779552715656,
-      "loss": 0.1375,
+      "learning_rate": 0.00014675186368477103,
+      "loss": 0.1608,
       "step": 1000
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00011932907348242812,
-      "loss": 0.0448,
+      "learning_rate": 0.00014621938232161876,
+      "loss": 0.0766,
       "step": 1010
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00011853035143769968,
-      "loss": 0.0543,
+      "learning_rate": 0.00014568690095846646,
+      "loss": 0.1158,
       "step": 1020
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00011773162939297124,
-      "loss": 0.0792,
+      "learning_rate": 0.00014515441959531418,
+      "loss": 0.057,
       "step": 1030
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00011693290734824283,
-      "loss": 0.0875,
+      "learning_rate": 0.00014462193823216188,
+      "loss": 0.1986,
       "step": 1040
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00011613418530351439,
-      "loss": 0.085,
+      "learning_rate": 0.00014408945686900958,
+      "loss": 0.1149,
       "step": 1050
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00011533546325878595,
-      "loss": 0.1663,
+      "learning_rate": 0.0001435569755058573,
+      "loss": 0.0974,
       "step": 1060
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00011453674121405751,
-      "loss": 0.06,
+      "learning_rate": 0.000143024494142705,
+      "loss": 0.1428,
       "step": 1070
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00011373801916932908,
-      "loss": 0.0948,
+      "learning_rate": 0.00014249201277955273,
+      "loss": 0.1688,
       "step": 1080
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00011293929712460065,
-      "loss": 0.0446,
+      "learning_rate": 0.00014195953141640043,
+      "loss": 0.074,
       "step": 1090
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00011214057507987221,
-      "loss": 0.0575,
+      "learning_rate": 0.00014142705005324815,
+      "loss": 0.0804,
       "step": 1100
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00011134185303514377,
-      "loss": 0.0818,
+      "learning_rate": 0.00014089456869009585,
+      "loss": 0.141,
       "step": 1110
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00011054313099041533,
-      "loss": 0.0758,
+      "learning_rate": 0.00014036208732694355,
+      "loss": 0.0854,
       "step": 1120
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00010974440894568691,
-      "loss": 0.0564,
+      "learning_rate": 0.00013982960596379128,
+      "loss": 0.0402,
       "step": 1130
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00010894568690095847,
-      "loss": 0.0428,
+      "learning_rate": 0.000139297124600639,
+      "loss": 0.0857,
       "step": 1140
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00010814696485623003,
-      "loss": 0.0652,
+      "learning_rate": 0.0001387646432374867,
+      "loss": 0.1617,
       "step": 1150
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0001073482428115016,
-      "loss": 0.0503,
+      "learning_rate": 0.0001382321618743344,
+      "loss": 0.1112,
       "step": 1160
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00010654952076677316,
-      "loss": 0.1214,
+      "learning_rate": 0.00013769968051118212,
+      "loss": 0.085,
       "step": 1170
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00010575079872204474,
-      "loss": 0.0953,
+      "learning_rate": 0.00013716719914802982,
+      "loss": 0.1818,
       "step": 1180
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0001049520766773163,
-      "loss": 0.0232,
+      "learning_rate": 0.00013663471778487752,
+      "loss": 0.0567,
       "step": 1190
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00010415335463258787,
-      "loss": 0.1204,
+      "learning_rate": 0.00013610223642172525,
+      "loss": 0.0995,
       "step": 1200
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00010335463258785943,
-      "loss": 0.0735,
+      "learning_rate": 0.00013556975505857297,
+      "loss": 0.1094,
       "step": 1210
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.000102555910543131,
-      "loss": 0.0557,
+      "learning_rate": 0.00013503727369542067,
+      "loss": 0.1087,
       "step": 1220
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00010175718849840256,
-      "loss": 0.0166,
+      "learning_rate": 0.00013450479233226837,
+      "loss": 0.0555,
       "step": 1230
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00010095846645367413,
-      "loss": 0.0971,
+      "learning_rate": 0.0001339723109691161,
+      "loss": 0.1189,
       "step": 1240
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00010015974440894569,
-      "loss": 0.0674,
+      "learning_rate": 0.0001334398296059638,
+      "loss": 0.1849,
       "step": 1250
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.8497409326424871,
-      "eval_f1": 0.8498734362292222,
-      "eval_loss": 0.545942485332489,
-      "eval_precision": 0.8570878171914441,
-      "eval_recall": 0.8497409326424871,
-      "eval_runtime": 1.2877,
-      "eval_samples_per_second": 149.878,
-      "eval_steps_per_second": 10.095,
+      "eval_accuracy": 0.8341968911917098,
+      "eval_f1": 0.830949546751025,
+      "eval_loss": 0.43995991349220276,
+      "eval_precision": 0.8292467957458414,
+      "eval_recall": 0.8341968911917098,
+      "eval_runtime": 1.2994,
+      "eval_samples_per_second": 148.525,
+      "eval_steps_per_second": 10.004,
       "step": 1252
     },
     {
       "epoch": 2.01,
-      "learning_rate": 9.936102236421726e-05,
-      "loss": 0.0471,
+      "learning_rate": 0.0001329073482428115,
+      "loss": 0.0658,
       "step": 1260
     },
     {
       "epoch": 2.03,
-      "learning_rate": 9.856230031948882e-05,
-      "loss": 0.0161,
+      "learning_rate": 0.00013237486687965922,
+      "loss": 0.0371,
       "step": 1270
     },
     {
       "epoch": 2.04,
-      "learning_rate": 9.77635782747604e-05,
-      "loss": 0.1494,
+      "learning_rate": 0.00013184238551650694,
+      "loss": 0.1156,
       "step": 1280
     },
     {
       "epoch": 2.06,
-      "learning_rate": 9.696485623003195e-05,
-      "loss": 0.0181,
+      "learning_rate": 0.00013130990415335464,
+      "loss": 0.0238,
       "step": 1290
     },
     {
       "epoch": 2.08,
-      "learning_rate": 9.616613418530351e-05,
-      "loss": 0.0072,
+      "learning_rate": 0.00013077742279020234,
+      "loss": 0.0908,
       "step": 1300
     },
     {
       "epoch": 2.09,
-      "learning_rate": 9.536741214057508e-05,
-      "loss": 0.043,
+      "learning_rate": 0.00013024494142705007,
+      "loss": 0.0358,
       "step": 1310
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.456869009584664e-05,
-      "loss": 0.021,
+      "learning_rate": 0.00012971246006389777,
+      "loss": 0.0398,
       "step": 1320
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.376996805111822e-05,
-      "loss": 0.0508,
+      "learning_rate": 0.00012917997870074546,
+      "loss": 0.06,
       "step": 1330
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.297124600638978e-05,
-      "loss": 0.0133,
+      "learning_rate": 0.0001286474973375932,
+      "loss": 0.0762,
       "step": 1340
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.217252396166136e-05,
-      "loss": 0.0262,
+      "learning_rate": 0.00012811501597444092,
+      "loss": 0.0293,
       "step": 1350
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.137380191693292e-05,
-      "loss": 0.0205,
+      "learning_rate": 0.00012758253461128861,
+      "loss": 0.0637,
       "step": 1360
     },
     {
       "epoch": 2.19,
-      "learning_rate": 9.057507987220448e-05,
-      "loss": 0.0075,
+      "learning_rate": 0.0001270500532481363,
+      "loss": 0.0265,
       "step": 1370
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.977635782747604e-05,
-      "loss": 0.0357,
+      "learning_rate": 0.00012651757188498404,
+      "loss": 0.0968,
       "step": 1380
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.89776357827476e-05,
-      "loss": 0.0092,
+      "learning_rate": 0.00012598509052183174,
+      "loss": 0.0883,
       "step": 1390
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.817891373801918e-05,
-      "loss": 0.0233,
+      "learning_rate": 0.00012545260915867944,
+      "loss": 0.0195,
       "step": 1400
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.738019169329074e-05,
-      "loss": 0.0106,
+      "learning_rate": 0.00012492012779552716,
+      "loss": 0.0663,
       "step": 1410
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.658146964856231e-05,
-      "loss": 0.0156,
+      "learning_rate": 0.0001243876464323749,
+      "loss": 0.0512,
       "step": 1420
     },
     {
       "epoch": 2.28,
-      "learning_rate": 8.578274760383387e-05,
-      "loss": 0.0102,
+      "learning_rate": 0.00012385516506922259,
+      "loss": 0.0288,
       "step": 1430
     },
     {
       "epoch": 2.3,
-      "learning_rate": 8.498402555910544e-05,
-      "loss": 0.0284,
+      "learning_rate": 0.00012332268370607028,
+      "loss": 0.0677,
       "step": 1440
     },
     {
       "epoch": 2.32,
-      "learning_rate": 8.4185303514377e-05,
-      "loss": 0.0211,
+      "learning_rate": 0.000122790202342918,
+      "loss": 0.1237,
       "step": 1450
     },
     {
       "epoch": 2.33,
-      "learning_rate": 8.338658146964856e-05,
-      "loss": 0.0172,
+      "learning_rate": 0.0001222577209797657,
+      "loss": 0.067,
       "step": 1460
     },
     {
       "epoch": 2.35,
-      "learning_rate": 8.258785942492013e-05,
-      "loss": 0.0455,
+      "learning_rate": 0.00012172523961661342,
+      "loss": 0.0467,
       "step": 1470
     },
     {
       "epoch": 2.36,
-      "learning_rate": 8.17891373801917e-05,
-      "loss": 0.1182,
+      "learning_rate": 0.00012119275825346112,
+      "loss": 0.1606,
       "step": 1480
     },
     {
       "epoch": 2.38,
-      "learning_rate": 8.099041533546327e-05,
-      "loss": 0.0048,
+      "learning_rate": 0.00012066027689030886,
+      "loss": 0.0499,
       "step": 1490
     },
     {
       "epoch": 2.4,
-      "learning_rate": 8.019169329073483e-05,
-      "loss": 0.0013,
+      "learning_rate": 0.00012012779552715656,
+      "loss": 0.0493,
       "step": 1500
     },
     {
       "epoch": 2.41,
-      "learning_rate": 7.939297124600639e-05,
-      "loss": 0.0249,
+      "learning_rate": 0.00011959531416400427,
+      "loss": 0.036,
       "step": 1510
     },
     {
       "epoch": 2.43,
-      "learning_rate": 7.859424920127795e-05,
-      "loss": 0.0088,
+      "learning_rate": 0.00011906283280085198,
+      "loss": 0.0536,
       "step": 1520
     },
     {
       "epoch": 2.44,
-      "learning_rate": 7.779552715654951e-05,
-      "loss": 0.0214,
+      "learning_rate": 0.00011853035143769968,
+      "loss": 0.0333,
       "step": 1530
     },
     {
       "epoch": 2.46,
-      "learning_rate": 7.699680511182109e-05,
-      "loss": 0.0414,
+      "learning_rate": 0.00011799787007454739,
+      "loss": 0.0641,
       "step": 1540
     },
     {
       "epoch": 2.48,
-      "learning_rate": 7.619808306709265e-05,
-      "loss": 0.0255,
+      "learning_rate": 0.00011746538871139509,
+      "loss": 0.0098,
       "step": 1550
     },
     {
       "epoch": 2.49,
-      "learning_rate": 7.539936102236423e-05,
-      "loss": 0.0479,
+      "learning_rate": 0.00011693290734824283,
+      "loss": 0.0533,
       "step": 1560
     },
     {
       "epoch": 2.51,
-      "learning_rate": 7.460063897763579e-05,
-      "loss": 0.0402,
+      "learning_rate": 0.00011640042598509053,
+      "loss": 0.0207,
       "step": 1570
     },
     {
       "epoch": 2.52,
-      "learning_rate": 7.380191693290735e-05,
-      "loss": 0.0377,
+      "learning_rate": 0.00011586794462193824,
+      "loss": 0.088,
       "step": 1580
     },
     {
       "epoch": 2.54,
-      "learning_rate": 7.300319488817891e-05,
-      "loss": 0.0612,
+      "learning_rate": 0.00011533546325878595,
+      "loss": 0.0669,
       "step": 1590
     },
     {
       "epoch": 2.56,
-      "learning_rate": 7.220447284345049e-05,
-      "loss": 0.0363,
+      "learning_rate": 0.00011480298189563365,
+      "loss": 0.0447,
       "step": 1600
     },
     {
       "epoch": 2.57,
-      "learning_rate": 7.140575079872205e-05,
-      "loss": 0.0026,
+      "learning_rate": 0.00011427050053248136,
+      "loss": 0.0407,
       "step": 1610
     },
     {
       "epoch": 2.59,
-      "learning_rate": 7.060702875399361e-05,
-      "loss": 0.0024,
+      "learning_rate": 0.00011373801916932908,
+      "loss": 0.0647,
       "step": 1620
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.980830670926518e-05,
-      "loss": 0.0247,
+      "learning_rate": 0.0001132055378061768,
+      "loss": 0.0885,
       "step": 1630
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.900958466453674e-05,
-      "loss": 0.0096,
+      "learning_rate": 0.0001126730564430245,
+      "loss": 0.069,
       "step": 1640
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.821086261980832e-05,
-      "loss": 0.0413,
+      "learning_rate": 0.00011214057507987221,
+      "loss": 0.03,
       "step": 1650
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.741214057507987e-05,
-      "loss": 0.0304,
+      "learning_rate": 0.00011160809371671992,
+      "loss": 0.0369,
       "step": 1660
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.661341853035144e-05,
-      "loss": 0.0481,
+      "learning_rate": 0.00011107561235356762,
+      "loss": 0.0694,
       "step": 1670
     },
     {
       "epoch": 2.68,
-      "learning_rate": 6.5814696485623e-05,
-      "loss": 0.013,
+      "learning_rate": 0.00011054313099041533,
+      "loss": 0.0668,
       "step": 1680
     },
     {
       "epoch": 2.7,
-      "learning_rate": 6.501597444089457e-05,
-      "loss": 0.0393,
+      "learning_rate": 0.00011001064962726305,
+      "loss": 0.1145,
       "step": 1690
     },
     {
       "epoch": 2.72,
-      "learning_rate": 6.421725239616614e-05,
-      "loss": 0.0102,
+      "learning_rate": 0.00010947816826411077,
+      "loss": 0.0651,
       "step": 1700
     },
     {
       "epoch": 2.73,
-      "learning_rate": 6.34185303514377e-05,
-      "loss": 0.0741,
+      "learning_rate": 0.00010894568690095847,
+      "loss": 0.0573,
       "step": 1710
     },
     {
       "epoch": 2.75,
-      "learning_rate": 6.261980830670928e-05,
-      "loss": 0.0134,
+      "learning_rate": 0.00010841320553780618,
+      "loss": 0.0839,
       "step": 1720
     },
     {
       "epoch": 2.76,
-      "learning_rate": 6.182108626198084e-05,
-      "loss": 0.0122,
+      "learning_rate": 0.0001078807241746539,
+      "loss": 0.0698,
       "step": 1730
     },
     {
       "epoch": 2.78,
-      "learning_rate": 6.1022364217252406e-05,
-      "loss": 0.0028,
+      "learning_rate": 0.0001073482428115016,
+      "loss": 0.047,
       "step": 1740
     },
     {
       "epoch": 2.8,
-      "learning_rate": 6.022364217252396e-05,
-      "loss": 0.0129,
+      "learning_rate": 0.0001068157614483493,
+      "loss": 0.0276,
       "step": 1750
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.942492012779552e-05,
-      "loss": 0.0504,
+      "learning_rate": 0.00010628328008519702,
+      "loss": 0.055,
       "step": 1760
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.86261980830671e-05,
-      "loss": 0.0242,
+      "learning_rate": 0.00010575079872204474,
+      "loss": 0.0494,
       "step": 1770
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.782747603833866e-05,
-      "loss": 0.0112,
+      "learning_rate": 0.00010521831735889244,
+      "loss": 0.1156,
       "step": 1780
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.702875399361023e-05,
-      "loss": 0.0165,
+      "learning_rate": 0.00010468583599574015,
+      "loss": 0.0362,
       "step": 1790
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.623003194888179e-05,
-      "loss": 0.0125,
+      "learning_rate": 0.00010415335463258787,
+      "loss": 0.0238,
       "step": 1800
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.543130990415336e-05,
-      "loss": 0.0181,
+      "learning_rate": 0.00010362087326943557,
+      "loss": 0.0314,
       "step": 1810
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.4632587859424925e-05,
-      "loss": 0.0324,
+      "learning_rate": 0.00010308839190628328,
+      "loss": 0.0746,
       "step": 1820
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.383386581469649e-05,
-      "loss": 0.0055,
+      "learning_rate": 0.000102555910543131,
+      "loss": 0.022,
       "step": 1830
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.3035143769968054e-05,
-      "loss": 0.0043,
+      "learning_rate": 0.00010202342917997871,
+      "loss": 0.0332,
       "step": 1840
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.2236421725239616e-05,
-      "loss": 0.0011,
+      "learning_rate": 0.00010149094781682643,
+      "loss": 0.0184,
       "step": 1850
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.1437699680511184e-05,
-      "loss": 0.0419,
+      "learning_rate": 0.00010095846645367413,
+      "loss": 0.0832,
       "step": 1860
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.0638977635782745e-05,
-      "loss": 0.033,
+      "learning_rate": 0.00010042598509052184,
+      "loss": 0.0853,
       "step": 1870
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.8393782383419689,
-      "eval_f1": 0.838981282953157,
-      "eval_loss": 0.5612991452217102,
-      "eval_precision": 0.8435335171059052,
-      "eval_recall": 0.8393782383419689,
-      "eval_runtime": 1.3877,
-      "eval_samples_per_second": 139.076,
-      "eval_steps_per_second": 9.368,
+      "eval_accuracy": 0.8134715025906736,
+      "eval_f1": 0.8171417754092435,
+      "eval_loss": 0.503167986869812,
+      "eval_precision": 0.8277834889115039,
+      "eval_recall": 0.8134715025906736,
+      "eval_runtime": 1.2586,
+      "eval_samples_per_second": 153.351,
+      "eval_steps_per_second": 10.329,
       "step": 1878
     },
     {
       "epoch": 3.0,
-      "learning_rate": 4.984025559105431e-05,
-      "loss": 0.0014,
+      "learning_rate": 9.989350372736954e-05,
+      "loss": 0.0312,
       "step": 1880
     },
     {
       "epoch": 3.02,
-      "learning_rate": 4.904153354632588e-05,
-      "loss": 0.0096,
+      "learning_rate": 9.936102236421726e-05,
+      "loss": 0.0234,
       "step": 1890
     },
     {
       "epoch": 3.04,
-      "learning_rate": 4.824281150159744e-05,
-      "loss": 0.0008,
+      "learning_rate": 9.882854100106496e-05,
+      "loss": 0.0205,
       "step": 1900
     },
     {
       "epoch": 3.05,
-      "learning_rate": 4.744408945686901e-05,
-      "loss": 0.0008,
+      "learning_rate": 9.829605963791267e-05,
+      "loss": 0.013,
       "step": 1910
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.664536741214058e-05,
-      "loss": 0.0074,
+      "learning_rate": 9.77635782747604e-05,
+      "loss": 0.0089,
       "step": 1920
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.584664536741215e-05,
-      "loss": 0.0006,
+      "learning_rate": 9.72310969116081e-05,
+      "loss": 0.0093,
       "step": 1930
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.504792332268371e-05,
-      "loss": 0.0006,
+      "learning_rate": 9.669861554845581e-05,
+      "loss": 0.005,
       "step": 1940
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.424920127795527e-05,
-      "loss": 0.0032,
+      "learning_rate": 9.616613418530351e-05,
+      "loss": 0.0145,
       "step": 1950
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.345047923322684e-05,
-      "loss": 0.0005,
+      "learning_rate": 9.563365282215123e-05,
+      "loss": 0.0125,
       "step": 1960
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.265175718849841e-05,
-      "loss": 0.0175,
+      "learning_rate": 9.510117145899895e-05,
+      "loss": 0.0142,
       "step": 1970
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.185303514376997e-05,
-      "loss": 0.0005,
+      "learning_rate": 9.456869009584664e-05,
+      "loss": 0.0164,
       "step": 1980
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.1054313099041536e-05,
-      "loss": 0.0018,
+      "learning_rate": 9.403620873269437e-05,
+      "loss": 0.0048,
       "step": 1990
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.0255591054313104e-05,
-      "loss": 0.0005,
+      "learning_rate": 9.350372736954207e-05,
+      "loss": 0.0164,
       "step": 2000
     },
     {
       "epoch": 3.21,
-      "learning_rate": 3.9456869009584666e-05,
-      "loss": 0.002,
+      "learning_rate": 9.297124600638978e-05,
+      "loss": 0.007,
       "step": 2010
     },
     {
       "epoch": 3.23,
-      "learning_rate": 3.8658146964856234e-05,
-      "loss": 0.0004,
+      "learning_rate": 9.243876464323749e-05,
+      "loss": 0.03,
       "step": 2020
     },
     {
       "epoch": 3.24,
-      "learning_rate": 3.7859424920127795e-05,
-      "loss": 0.0251,
+      "learning_rate": 9.19062832800852e-05,
+      "loss": 0.012,
       "step": 2030
     },
     {
       "epoch": 3.26,
-      "learning_rate": 3.7060702875399364e-05,
-      "loss": 0.0012,
+      "learning_rate": 9.137380191693292e-05,
+      "loss": 0.0354,
       "step": 2040
     },
     {
       "epoch": 3.27,
-      "learning_rate": 3.6261980830670925e-05,
-      "loss": 0.0013,
+      "learning_rate": 9.084132055378062e-05,
+      "loss": 0.01,
       "step": 2050
     },
     {
       "epoch": 3.29,
-      "learning_rate": 3.546325878594249e-05,
-      "loss": 0.0007,
+      "learning_rate": 9.030883919062834e-05,
+      "loss": 0.0032,
       "step": 2060
     },
     {
       "epoch": 3.31,
-      "learning_rate": 3.466453674121406e-05,
-      "loss": 0.0013,
+      "learning_rate": 8.977635782747604e-05,
+      "loss": 0.0066,
       "step": 2070
     },
     {
       "epoch": 3.32,
-      "learning_rate": 3.386581469648562e-05,
-      "loss": 0.0003,
+      "learning_rate": 8.924387646432375e-05,
+      "loss": 0.0081,
       "step": 2080
     },
     {
       "epoch": 3.34,
-      "learning_rate": 3.306709265175719e-05,
-      "loss": 0.0004,
+      "learning_rate": 8.871139510117146e-05,
+      "loss": 0.0141,
       "step": 2090
     },
     {
       "epoch": 3.35,
-      "learning_rate": 3.226837060702875e-05,
-      "loss": 0.0005,
+      "learning_rate": 8.817891373801918e-05,
+      "loss": 0.0142,
       "step": 2100
     },
     {
       "epoch": 3.37,
-      "learning_rate": 3.146964856230032e-05,
-      "loss": 0.0015,
+      "learning_rate": 8.764643237486689e-05,
+      "loss": 0.0492,
       "step": 2110
     },
     {
       "epoch": 3.39,
-      "learning_rate": 3.067092651757188e-05,
-      "loss": 0.0025,
+      "learning_rate": 8.711395101171459e-05,
+      "loss": 0.0105,
       "step": 2120
     },
     {
       "epoch": 3.4,
-      "learning_rate": 2.987220447284345e-05,
-      "loss": 0.0005,
+      "learning_rate": 8.658146964856231e-05,
+      "loss": 0.0457,
       "step": 2130
     },
     {
       "epoch": 3.42,
-      "learning_rate": 2.907348242811502e-05,
-      "loss": 0.0005,
+      "learning_rate": 8.604898828541001e-05,
+      "loss": 0.0108,
       "step": 2140
     },
     {
       "epoch": 3.43,
-      "learning_rate": 2.8274760383386583e-05,
-      "loss": 0.0225,
+      "learning_rate": 8.551650692225772e-05,
+      "loss": 0.006,
       "step": 2150
     },
     {
       "epoch": 3.45,
-      "learning_rate": 2.747603833865815e-05,
-      "loss": 0.0005,
+      "learning_rate": 8.498402555910544e-05,
+      "loss": 0.0048,
       "step": 2160
     },
     {
       "epoch": 3.47,
-      "learning_rate": 2.6677316293929716e-05,
-      "loss": 0.0006,
+      "learning_rate": 8.445154419595315e-05,
+      "loss": 0.0071,
       "step": 2170
     },
     {
       "epoch": 3.48,
-      "learning_rate": 2.5958466453674125e-05,
-      "loss": 0.0092,
+      "learning_rate": 8.391906283280086e-05,
+      "loss": 0.0366,
       "step": 2180
     },
     {
       "epoch": 3.5,
-      "learning_rate": 2.515974440894569e-05,
-      "loss": 0.0004,
+      "learning_rate": 8.338658146964856e-05,
+      "loss": 0.0094,
       "step": 2190
     },
     {
       "epoch": 3.51,
-      "learning_rate": 2.4361022364217255e-05,
-      "loss": 0.0009,
+      "learning_rate": 8.285410010649628e-05,
+      "loss": 0.0102,
       "step": 2200
     },
     {
       "epoch": 3.53,
-      "learning_rate": 2.356230031948882e-05,
-      "loss": 0.0006,
+      "learning_rate": 8.232161874334398e-05,
+      "loss": 0.0112,
       "step": 2210
     },
     {
       "epoch": 3.55,
-      "learning_rate": 2.2763578274760385e-05,
-      "loss": 0.0004,
+      "learning_rate": 8.17891373801917e-05,
+      "loss": 0.0012,
       "step": 2220
     },
     {
       "epoch": 3.56,
-      "learning_rate": 2.196485623003195e-05,
-      "loss": 0.0004,
+      "learning_rate": 8.130990415335463e-05,
+      "loss": 0.0193,
       "step": 2230
     },
     {
       "epoch": 3.58,
-      "learning_rate": 2.1166134185303514e-05,
-      "loss": 0.0005,
+      "learning_rate": 8.077742279020235e-05,
+      "loss": 0.0205,
       "step": 2240
     },
     {
       "epoch": 3.59,
-      "learning_rate": 2.0367412140575082e-05,
-      "loss": 0.0012,
+      "learning_rate": 8.024494142705006e-05,
+      "loss": 0.0138,
       "step": 2250
     },
     {
       "epoch": 3.61,
-      "learning_rate": 1.9568690095846644e-05,
-      "loss": 0.0004,
+      "learning_rate": 7.971246006389777e-05,
+      "loss": 0.0329,
       "step": 2260
     },
     {
       "epoch": 3.63,
-      "learning_rate": 1.8769968051118212e-05,
-      "loss": 0.003,
+      "learning_rate": 7.917997870074548e-05,
+      "loss": 0.0018,
       "step": 2270
     },
     {
       "epoch": 3.64,
-      "learning_rate": 1.7971246006389777e-05,
-      "loss": 0.0004,
+      "learning_rate": 7.86474973375932e-05,
+      "loss": 0.0085,
       "step": 2280
     },
     {
       "epoch": 3.66,
-      "learning_rate": 1.7172523961661345e-05,
-      "loss": 0.0006,
+      "learning_rate": 7.811501597444089e-05,
+      "loss": 0.0141,
       "step": 2290
     },
     {
       "epoch": 3.67,
-      "learning_rate": 1.6373801916932906e-05,
-      "loss": 0.0004,
+      "learning_rate": 7.75825346112886e-05,
+      "loss": 0.0037,
       "step": 2300
     },
     {
       "epoch": 3.69,
-      "learning_rate": 1.5575079872204475e-05,
-      "loss": 0.0004,
+      "learning_rate": 7.705005324813632e-05,
+      "loss": 0.0295,
       "step": 2310
     },
     {
       "epoch": 3.71,
-      "learning_rate": 1.477635782747604e-05,
-      "loss": 0.0005,
+      "learning_rate": 7.651757188498403e-05,
+      "loss": 0.0054,
       "step": 2320
     },
     {
       "epoch": 3.72,
-      "learning_rate": 1.3977635782747606e-05,
-      "loss": 0.0006,
+      "learning_rate": 7.598509052183174e-05,
+      "loss": 0.0165,
       "step": 2330
     },
     {
       "epoch": 3.74,
-      "learning_rate": 1.3178913738019169e-05,
-      "loss": 0.0004,
+      "learning_rate": 7.545260915867945e-05,
+      "loss": 0.0041,
       "step": 2340
     },
     {
       "epoch": 3.75,
-      "learning_rate": 1.2380191693290735e-05,
-      "loss": 0.0004,
+      "learning_rate": 7.492012779552717e-05,
+      "loss": 0.0084,
       "step": 2350
     },
     {
       "epoch": 3.77,
-      "learning_rate": 1.1581469648562302e-05,
-      "loss": 0.0095,
+      "learning_rate": 7.438764643237486e-05,
+      "loss": 0.015,
       "step": 2360
     },
     {
       "epoch": 3.79,
-      "learning_rate": 1.0782747603833867e-05,
-      "loss": 0.0004,
+      "learning_rate": 7.385516506922258e-05,
+      "loss": 0.0136,
       "step": 2370
     },
     {
       "epoch": 3.8,
-      "learning_rate": 9.984025559105432e-06,
-      "loss": 0.0004,
+      "learning_rate": 7.332268370607029e-05,
+      "loss": 0.0008,
       "step": 2380
     },
     {
       "epoch": 3.82,
-      "learning_rate": 9.185303514376996e-06,
-      "loss": 0.0008,
+      "learning_rate": 7.2790202342918e-05,
+      "loss": 0.0047,
       "step": 2390
     },
     {
       "epoch": 3.83,
-      "learning_rate": 8.386581469648563e-06,
-      "loss": 0.0011,
+      "learning_rate": 7.225772097976571e-05,
+      "loss": 0.0065,
       "step": 2400
     },
     {
       "epoch": 3.85,
-      "learning_rate": 7.5878594249201285e-06,
-      "loss": 0.0003,
+      "learning_rate": 7.172523961661343e-05,
+      "loss": 0.0135,
       "step": 2410
     },
     {
       "epoch": 3.87,
-      "learning_rate": 6.789137380191693e-06,
-      "loss": 0.0057,
+      "learning_rate": 7.119275825346114e-05,
+      "loss": 0.0283,
       "step": 2420
     },
     {
       "epoch": 3.88,
-      "learning_rate": 5.990415335463259e-06,
-      "loss": 0.0003,
+      "learning_rate": 7.066027689030884e-05,
+      "loss": 0.0157,
       "step": 2430
     },
     {
       "epoch": 3.9,
-      "learning_rate": 5.191693290734825e-06,
-      "loss": 0.0006,
+      "learning_rate": 7.012779552715655e-05,
+      "loss": 0.0166,
       "step": 2440
     },
     {
       "epoch": 3.91,
-      "learning_rate": 4.39297124600639e-06,
-      "loss": 0.0005,
+      "learning_rate": 6.959531416400426e-05,
+      "loss": 0.0025,
       "step": 2450
     },
     {
       "epoch": 3.93,
-      "learning_rate": 3.5942492012779555e-06,
-      "loss": 0.0004,
+      "learning_rate": 6.906283280085197e-05,
+      "loss": 0.0121,
       "step": 2460
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.7955271565495207e-06,
-      "loss": 0.0028,
+      "learning_rate": 6.853035143769968e-05,
+      "loss": 0.0034,
       "step": 2470
     },
     {
       "epoch": 3.96,
-      "learning_rate": 1.9968051118210863e-06,
-      "loss": 0.0004,
+      "learning_rate": 6.79978700745474e-05,
+      "loss": 0.0212,
       "step": 2480
     },
     {
       "epoch": 3.98,
-      "learning_rate": 1.1980830670926517e-06,
-      "loss": 0.0003,
+      "learning_rate": 6.746538871139511e-05,
+      "loss": 0.0024,
       "step": 2490
     },
     {
       "epoch": 3.99,
-      "learning_rate": 3.9936102236421723e-07,
-      "loss": 0.0005,
+      "learning_rate": 6.693290734824281e-05,
+      "loss": 0.0046,
       "step": 2500
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8652849740932642,
-      "eval_f1": 0.8598597988418164,
-      "eval_loss": 0.5974593758583069,
-      "eval_precision": 0.8682223693949902,
-      "eval_recall": 0.8652849740932642,
-      "eval_runtime": 1.3471,
-      "eval_samples_per_second": 143.267,
-      "eval_steps_per_second": 9.65,
+      "eval_accuracy": 0.8601036269430051,
+      "eval_f1": 0.8615134929458629,
+      "eval_loss": 0.5402997136116028,
+      "eval_precision": 0.8654905254328257,
+      "eval_recall": 0.8601036269430051,
+      "eval_runtime": 1.3098,
+      "eval_samples_per_second": 147.345,
+      "eval_steps_per_second": 9.925,
       "step": 2504
     },
     {
-      "epoch": 4.0,
-      "step": 2504,
-      "total_flos": 3.104468219559813e+18,
-      "train_loss": 0.07359402208736303,
-      "train_runtime": 853.5389,
-      "train_samples_per_second": 46.934,
-      "train_steps_per_second": 2.934
+      "epoch": 4.01,
+      "learning_rate": 6.640042598509052e-05,
+      "loss": 0.0225,
+      "step": 2510
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 6.586794462193823e-05,
+      "loss": 0.0053,
+      "step": 2520
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 6.533546325878594e-05,
+      "loss": 0.007,
+      "step": 2530
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 6.480298189563366e-05,
+      "loss": 0.0007,
+      "step": 2540
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 6.427050053248137e-05,
+      "loss": 0.0054,
+      "step": 2550
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 6.373801916932908e-05,
+      "loss": 0.0007,
+      "step": 2560
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 6.320553780617678e-05,
+      "loss": 0.0185,
+      "step": 2570
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 6.26730564430245e-05,
+      "loss": 0.0008,
+      "step": 2580
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 6.214057507987222e-05,
+      "loss": 0.0104,
+      "step": 2590
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 6.160809371671991e-05,
+      "loss": 0.0079,
+      "step": 2600
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.107561235356763e-05,
+      "loss": 0.0005,
+      "step": 2610
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.054313099041534e-05,
+      "loss": 0.0141,
+      "step": 2620
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.001064962726305e-05,
+      "loss": 0.0006,
+      "step": 2630
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 5.9478168264110756e-05,
+      "loss": 0.0181,
+      "step": 2640
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 5.8945686900958475e-05,
+      "loss": 0.0011,
+      "step": 2650
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.841320553780618e-05,
+      "loss": 0.0061,
+      "step": 2660
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.7880724174653886e-05,
+      "loss": 0.0008,
+      "step": 2670
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.73482428115016e-05,
+      "loss": 0.0046,
+      "step": 2680
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.681576144834932e-05,
+      "loss": 0.0004,
+      "step": 2690
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 5.628328008519702e-05,
+      "loss": 0.0083,
+      "step": 2700
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 5.575079872204473e-05,
+      "loss": 0.0028,
+      "step": 2710
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 5.521831735889245e-05,
+      "loss": 0.0009,
+      "step": 2720
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 5.468583599574015e-05,
+      "loss": 0.0007,
+      "step": 2730
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 5.415335463258786e-05,
+      "loss": 0.0085,
+      "step": 2740
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 5.362087326943557e-05,
+      "loss": 0.0074,
+      "step": 2750
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 5.308839190628329e-05,
+      "loss": 0.001,
+      "step": 2760
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 5.2555910543130994e-05,
+      "loss": 0.0027,
+      "step": 2770
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 5.20234291799787e-05,
+      "loss": 0.0004,
+      "step": 2780
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.149094781682642e-05,
+      "loss": 0.0227,
+      "step": 2790
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.095846645367412e-05,
+      "loss": 0.0034,
+      "step": 2800
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.0425985090521835e-05,
+      "loss": 0.0039,
+      "step": 2810
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 4.989350372736955e-05,
+      "loss": 0.0066,
+      "step": 2820
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.936102236421725e-05,
+      "loss": 0.001,
+      "step": 2830
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.8828541001064965e-05,
+      "loss": 0.0007,
+      "step": 2840
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.829605963791268e-05,
+      "loss": 0.0003,
+      "step": 2850
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 4.776357827476038e-05,
+      "loss": 0.0046,
+      "step": 2860
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 4.7231096911608095e-05,
+      "loss": 0.0007,
+      "step": 2870
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 4.669861554845581e-05,
+      "loss": 0.0011,
+      "step": 2880
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 4.616613418530352e-05,
+      "loss": 0.013,
+      "step": 2890
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 4.563365282215123e-05,
+      "loss": 0.0003,
+      "step": 2900
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 4.515441959531417e-05,
+      "loss": 0.0085,
+      "step": 2910
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 4.4621938232161876e-05,
+      "loss": 0.0102,
+      "step": 2920
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 4.408945686900959e-05,
+      "loss": 0.004,
+      "step": 2930
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 4.355697550585729e-05,
+      "loss": 0.0004,
+      "step": 2940
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 4.3024494142705005e-05,
+      "loss": 0.0007,
+      "step": 2950
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 4.249201277955272e-05,
+      "loss": 0.0003,
+      "step": 2960
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 4.195953141640043e-05,
+      "loss": 0.0278,
+      "step": 2970
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 4.142705005324814e-05,
+      "loss": 0.0049,
+      "step": 2980
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.089456869009585e-05,
+      "loss": 0.0009,
+      "step": 2990
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.036208732694356e-05,
+      "loss": 0.0005,
+      "step": 3000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.9829605963791265e-05,
+      "loss": 0.0164,
+      "step": 3010
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.929712460063898e-05,
+      "loss": 0.0032,
+      "step": 3020
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 3.876464323748669e-05,
+      "loss": 0.0003,
+      "step": 3030
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.82321618743344e-05,
+      "loss": 0.0133,
+      "step": 3040
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.769968051118211e-05,
+      "loss": 0.0004,
+      "step": 3050
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 3.716719914802982e-05,
+      "loss": 0.0215,
+      "step": 3060
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.663471778487753e-05,
+      "loss": 0.0003,
+      "step": 3070
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.610223642172524e-05,
+      "loss": 0.022,
+      "step": 3080
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.556975505857295e-05,
+      "loss": 0.0045,
+      "step": 3090
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.503727369542067e-05,
+      "loss": 0.0003,
+      "step": 3100
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 3.450479233226837e-05,
+      "loss": 0.0005,
+      "step": 3110
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.3972310969116084e-05,
+      "loss": 0.0006,
+      "step": 3120
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.343982960596379e-05,
+      "loss": 0.0002,
+      "step": 3130
+    },
+    {
+      "epoch": 5.0,
+      "eval_accuracy": 0.8601036269430051,
+      "eval_f1": 0.8586960289425841,
+      "eval_loss": 0.5897737145423889,
+      "eval_precision": 0.8603442324057597,
+      "eval_recall": 0.8601036269430051,
+      "eval_runtime": 1.2538,
+      "eval_samples_per_second": 153.938,
+      "eval_steps_per_second": 10.369,
+      "step": 3130
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 3.29073482428115e-05,
+      "loss": 0.0003,
+      "step": 3140
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 3.2374866879659214e-05,
+      "loss": 0.0002,
+      "step": 3150
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 3.1842385516506926e-05,
+      "loss": 0.0002,
+      "step": 3160
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 3.130990415335464e-05,
+      "loss": 0.0002,
+      "step": 3170
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 3.0777422790202344e-05,
+      "loss": 0.0003,
+      "step": 3180
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 3.0244941427050056e-05,
+      "loss": 0.0002,
+      "step": 3190
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.971246006389776e-05,
+      "loss": 0.0004,
+      "step": 3200
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.9179978700745477e-05,
+      "loss": 0.0002,
+      "step": 3210
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.864749733759319e-05,
+      "loss": 0.0002,
+      "step": 3220
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.8115015974440894e-05,
+      "loss": 0.0003,
+      "step": 3230
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.7582534611288606e-05,
+      "loss": 0.0005,
+      "step": 3240
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.7050053248136315e-05,
+      "loss": 0.0002,
+      "step": 3250
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.6517571884984027e-05,
+      "loss": 0.0002,
+      "step": 3260
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.598509052183174e-05,
+      "loss": 0.0002,
+      "step": 3270
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.5452609158679448e-05,
+      "loss": 0.0003,
+      "step": 3280
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.4920127795527157e-05,
+      "loss": 0.0002,
+      "step": 3290
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.438764643237487e-05,
+      "loss": 0.0003,
+      "step": 3300
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.3855165069222578e-05,
+      "loss": 0.0003,
+      "step": 3310
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.332268370607029e-05,
+      "loss": 0.0002,
+      "step": 3320
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.2790202342918e-05,
+      "loss": 0.0002,
+      "step": 3330
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.2257720979765707e-05,
+      "loss": 0.0002,
+      "step": 3340
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.172523961661342e-05,
+      "loss": 0.0002,
+      "step": 3350
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.1192758253461128e-05,
+      "loss": 0.0003,
+      "step": 3360
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 2.066027689030884e-05,
+      "loss": 0.0002,
+      "step": 3370
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.0127795527156552e-05,
+      "loss": 0.0003,
+      "step": 3380
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 1.959531416400426e-05,
+      "loss": 0.0022,
+      "step": 3390
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 1.906283280085197e-05,
+      "loss": 0.0002,
+      "step": 3400
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 1.8530351437699682e-05,
+      "loss": 0.0002,
+      "step": 3410
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 1.799787007454739e-05,
+      "loss": 0.0003,
+      "step": 3420
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 1.7465388711395103e-05,
+      "loss": 0.0003,
+      "step": 3430
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 1.693290734824281e-05,
+      "loss": 0.0003,
+      "step": 3440
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 1.6400425985090524e-05,
+      "loss": 0.0002,
+      "step": 3450
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 1.5867944621938232e-05,
+      "loss": 0.0003,
+      "step": 3460
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 1.533546325878594e-05,
+      "loss": 0.0003,
+      "step": 3470
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 1.4802981895633653e-05,
+      "loss": 0.0003,
+      "step": 3480
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 1.4270500532481365e-05,
+      "loss": 0.0002,
+      "step": 3490
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 1.3738019169329076e-05,
+      "loss": 0.0167,
+      "step": 3500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 1.3205537806176784e-05,
+      "loss": 0.0003,
+      "step": 3510
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 1.2673056443024495e-05,
+      "loss": 0.0002,
+      "step": 3520
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 1.2140575079872205e-05,
+      "loss": 0.0002,
+      "step": 3530
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 1.1608093716719916e-05,
+      "loss": 0.0002,
+      "step": 3540
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 1.1075612353567626e-05,
+      "loss": 0.0002,
+      "step": 3550
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 1.0543130990415335e-05,
+      "loss": 0.0002,
+      "step": 3560
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 1.0010649627263047e-05,
+      "loss": 0.0002,
+      "step": 3570
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 9.478168264110757e-06,
+      "loss": 0.0003,
+      "step": 3580
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 8.945686900958466e-06,
+      "loss": 0.0002,
+      "step": 3590
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.413205537806178e-06,
+      "loss": 0.0002,
+      "step": 3600
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.880724174653887e-06,
+      "loss": 0.0002,
+      "step": 3610
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.3482428115015974e-06,
+      "loss": 0.0004,
+      "step": 3620
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 6.815761448349309e-06,
+      "loss": 0.0002,
+      "step": 3630
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 6.283280085197018e-06,
+      "loss": 0.0002,
+      "step": 3640
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 5.750798722044729e-06,
+      "loss": 0.0002,
+      "step": 3650
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 5.218317358892439e-06,
+      "loss": 0.0002,
+      "step": 3660
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 4.68583599574015e-06,
+      "loss": 0.0002,
+      "step": 3670
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 4.153354632587859e-06,
+      "loss": 0.0002,
+      "step": 3680
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 3.62087326943557e-06,
+      "loss": 0.0002,
+      "step": 3690
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 3.08839190628328e-06,
+      "loss": 0.0002,
+      "step": 3700
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.5559105431309904e-06,
+      "loss": 0.0004,
+      "step": 3710
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.023429179978701e-06,
+      "loss": 0.0002,
+      "step": 3720
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 1.490947816826411e-06,
+      "loss": 0.0002,
+      "step": 3730
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.584664536741213e-07,
+      "loss": 0.0003,
+      "step": 3740
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 4.259850905218318e-07,
+      "loss": 0.0002,
+      "step": 3750
+    },
+    {
+      "epoch": 6.0,
+      "eval_accuracy": 0.8601036269430051,
+      "eval_f1": 0.8586960289425841,
+      "eval_loss": 0.6216807961463928,
+      "eval_precision": 0.8603442324057597,
+      "eval_recall": 0.8601036269430051,
+      "eval_runtime": 1.3188,
+      "eval_samples_per_second": 146.348,
+      "eval_steps_per_second": 9.858,
+      "step": 3756
+    },
+    {
+      "epoch": 6.0,
+      "step": 3756,
+      "total_flos": 4.65670232933972e+18,
+      "train_loss": 0.07414248911174177,
+      "train_runtime": 1259.2386,
+      "train_samples_per_second": 47.719,
+      "train_steps_per_second": 2.983
     }
   ],
-  "max_steps": 2504,
-  "num_train_epochs": 4,
-  "total_flos": 3.104468219559813e+18,
+  "max_steps": 3756,
+  "num_train_epochs": 6,
+  "total_flos": 4.65670232933972e+18,
   "trial_name": null,
   "trial_params": null
 }

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:20da84aff0ef3299c84384e9b6ea0eca383ddfa87a467d7e5c563ce794fd47ab
+oid sha256:5e9facca73fd7327e607366003ac58274d3909c8ce2b4d471492c16274848b14
 size 3899