🍻 cheers

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 4.0,
-    "eval_accuracy": 0.9521785334750266,
-    "eval_loss": 0.21390262246131897,
-    "eval_runtime": 40.27,
-    "eval_samples_per_second": 23.367,
-    "eval_steps_per_second": 2.93,
+    "eval_accuracy": 0.9543039319872476,
+    "eval_loss": 0.19176600873470306,
+    "eval_runtime": 33.828,
+    "eval_samples_per_second": 27.817,
+    "eval_steps_per_second": 3.488,
     "total_flos": 1.651775717862015e+18,
-    "train_loss": 0.4871297548483084,
-    "train_runtime": 1589.6821,
-    "train_samples_per_second": 13.406,
-    "train_steps_per_second": 0.838
+    "train_loss": 0.4876434194433081,
+    "train_runtime": 1561.5053,
+    "train_samples_per_second": 13.648,
+    "train_steps_per_second": 0.853
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
-    "eval_accuracy": 0.9521785334750266,
-    "eval_loss": 0.21390262246131897,
-    "eval_runtime": 40.27,
-    "eval_samples_per_second": 23.367,
-    "eval_steps_per_second": 2.93
+    "eval_accuracy": 0.9543039319872476,
+    "eval_loss": 0.19176600873470306,
+    "eval_runtime": 33.828,
+    "eval_samples_per_second": 27.817,
+    "eval_steps_per_second": 3.488
 }

runs/Aug03_21-01-32_cab156e75ca5/events.out.tfevents.1659562835.cab156e75ca5.72.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3430a5c69cc495109663e9714ad902b05733e10604b8b2bae9a0112271bad718
+size 363

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
     "total_flos": 1.651775717862015e+18,
-    "train_loss": 0.4871297548483084,
-    "train_runtime": 1589.6821,
-    "train_samples_per_second": 13.406,
-    "train_steps_per_second": 0.838
+    "train_loss": 0.4876434194433081,
+    "train_runtime": 1561.5053,
+    "train_samples_per_second": 13.648,
+    "train_steps_per_second": 0.853
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.21390262246131897,
+  "best_metric": 0.19176600873470306,
   "best_model_checkpoint": "finetuned-indian-food/checkpoint-1300",
   "epoch": 4.0,
   "global_step": 1332,
@@ -10,926 +10,926 @@
     {
       "epoch": 0.03,
       "learning_rate": 0.0001984984984984985,
-      "loss": 2.8538,
+      "loss": 2.8856,
       "step": 10
     },
     {
       "epoch": 0.06,
       "learning_rate": 0.00019699699699699701,
-      "loss": 2.5964,
+      "loss": 2.5941,
       "step": 20
     },
     {
       "epoch": 0.09,
       "learning_rate": 0.0001954954954954955,
-      "loss": 2.2524,
+      "loss": 2.1576,
       "step": 30
     },
     {
       "epoch": 0.12,
       "learning_rate": 0.00019399399399399402,
-      "loss": 1.9596,
+      "loss": 1.9221,
       "step": 40
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.0001924924924924925,
-      "loss": 1.726,
+      "loss": 1.7671,
       "step": 50
     },
     {
       "epoch": 0.18,
       "learning_rate": 0.000190990990990991,
-      "loss": 1.6152,
+      "loss": 1.4461,
       "step": 60
     },
     {
       "epoch": 0.21,
       "learning_rate": 0.0001894894894894895,
-      "loss": 1.3396,
+      "loss": 1.3917,
       "step": 70
     },
     {
       "epoch": 0.24,
       "learning_rate": 0.000187987987987988,
-      "loss": 1.2854,
+      "loss": 1.3666,
       "step": 80
     },
     {
       "epoch": 0.27,
       "learning_rate": 0.0001864864864864865,
-      "loss": 1.107,
+      "loss": 1.1615,
       "step": 90
     },
     {
       "epoch": 0.3,
       "learning_rate": 0.000184984984984985,
-      "loss": 1.0846,
+      "loss": 1.0175,
       "step": 100
     },
     {
       "epoch": 0.3,
-      "eval_accuracy": 0.8554729011689692,
-      "eval_loss": 0.9561436772346497,
-      "eval_runtime": 39.5068,
-      "eval_samples_per_second": 23.819,
-      "eval_steps_per_second": 2.987,
+      "eval_accuracy": 0.8629117959617428,
+      "eval_loss": 0.9246562719345093,
+      "eval_runtime": 35.8495,
+      "eval_samples_per_second": 26.249,
+      "eval_steps_per_second": 3.292,
       "step": 100
     },
     {
       "epoch": 0.33,
       "learning_rate": 0.0001834834834834835,
-      "loss": 0.9812,
+      "loss": 1.044,
       "step": 110
     },
     {
       "epoch": 0.36,
       "learning_rate": 0.000181981981981982,
-      "loss": 0.8571,
+      "loss": 0.9848,
       "step": 120
     },
     {
       "epoch": 0.39,
       "learning_rate": 0.0001804804804804805,
-      "loss": 0.916,
+      "loss": 0.7889,
       "step": 130
     },
     {
       "epoch": 0.42,
       "learning_rate": 0.00017897897897897898,
-      "loss": 0.9905,
+      "loss": 0.921,
       "step": 140
     },
     {
       "epoch": 0.45,
       "learning_rate": 0.0001774774774774775,
-      "loss": 0.7381,
+      "loss": 0.7709,
       "step": 150
     },
     {
       "epoch": 0.48,
       "learning_rate": 0.000175975975975976,
-      "loss": 0.801,
+      "loss": 0.9096,
       "step": 160
     },
     {
       "epoch": 0.51,
       "learning_rate": 0.0001744744744744745,
-      "loss": 0.6606,
+      "loss": 0.7293,
       "step": 170
     },
     {
       "epoch": 0.54,
       "learning_rate": 0.000172972972972973,
-      "loss": 0.7538,
+      "loss": 0.6419,
       "step": 180
     },
     {
       "epoch": 0.57,
       "learning_rate": 0.00017147147147147148,
-      "loss": 0.6495,
+      "loss": 0.7251,
       "step": 190
     },
     {
       "epoch": 0.6,
       "learning_rate": 0.00016996996996997,
-      "loss": 0.7894,
+      "loss": 0.7418,
       "step": 200
     },
     {
       "epoch": 0.6,
-      "eval_accuracy": 0.8926673751328374,
-      "eval_loss": 0.587147057056427,
-      "eval_runtime": 38.4761,
-      "eval_samples_per_second": 24.457,
-      "eval_steps_per_second": 3.067,
+      "eval_accuracy": 0.8990435706695006,
+      "eval_loss": 0.5535812377929688,
+      "eval_runtime": 33.962,
+      "eval_samples_per_second": 27.707,
+      "eval_steps_per_second": 3.474,
       "step": 200
     },
     {
       "epoch": 0.63,
       "learning_rate": 0.00016846846846846846,
-      "loss": 0.7121,
+      "loss": 0.7307,
       "step": 210
     },
     {
       "epoch": 0.66,
       "learning_rate": 0.00016696696696696697,
-      "loss": 0.6888,
+      "loss": 0.7721,
       "step": 220
     },
     {
       "epoch": 0.69,
       "learning_rate": 0.00016546546546546546,
-      "loss": 0.5768,
+      "loss": 0.6073,
       "step": 230
     },
     {
       "epoch": 0.72,
       "learning_rate": 0.00016396396396396395,
-      "loss": 0.6568,
+      "loss": 0.6946,
       "step": 240
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00016261261261261262,
-      "loss": 0.601,
+      "learning_rate": 0.00016246246246246247,
+      "loss": 0.7824,
       "step": 250
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0001611111111111111,
-      "loss": 0.5431,
+      "learning_rate": 0.00016096096096096096,
+      "loss": 0.5175,
       "step": 260
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00015960960960960963,
-      "loss": 0.6634,
+      "learning_rate": 0.00015945945945945947,
+      "loss": 0.7959,
       "step": 270
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00015810810810810812,
-      "loss": 0.8153,
+      "learning_rate": 0.00015795795795795796,
+      "loss": 0.5273,
       "step": 280
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00015660660660660663,
-      "loss": 0.9727,
+      "learning_rate": 0.00015645645645645645,
+      "loss": 0.6327,
       "step": 290
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00015510510510510512,
-      "loss": 0.6233,
+      "learning_rate": 0.00015495495495495496,
+      "loss": 0.6652,
       "step": 300
     },
     {
       "epoch": 0.9,
-      "eval_accuracy": 0.9107332624867163,
-      "eval_loss": 0.4447122812271118,
-      "eval_runtime": 38.3227,
-      "eval_samples_per_second": 24.555,
-      "eval_steps_per_second": 3.079,
+      "eval_accuracy": 0.9181721572794899,
+      "eval_loss": 0.4035964906215668,
+      "eval_runtime": 33.8609,
+      "eval_samples_per_second": 27.79,
+      "eval_steps_per_second": 3.485,
       "step": 300
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0001536036036036036,
-      "loss": 0.578,
+      "learning_rate": 0.00015345345345345345,
+      "loss": 0.6688,
       "step": 310
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00015210210210210212,
-      "loss": 0.6366,
+      "learning_rate": 0.00015195195195195194,
+      "loss": 0.5384,
       "step": 320
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0001506006006006006,
-      "loss": 0.8622,
+      "learning_rate": 0.00015045045045045046,
+      "loss": 0.5232,
       "step": 330
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00014909909909909913,
-      "loss": 0.5957,
+      "learning_rate": 0.00014894894894894895,
+      "loss": 0.5407,
       "step": 340
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00014759759759759762,
-      "loss": 0.4458,
+      "learning_rate": 0.00014744744744744746,
+      "loss": 0.512,
       "step": 350
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0001460960960960961,
-      "loss": 0.383,
+      "learning_rate": 0.00014594594594594595,
+      "loss": 0.4422,
       "step": 360
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00014459459459459462,
-      "loss": 0.4259,
+      "learning_rate": 0.00014444444444444444,
+      "loss": 0.6211,
       "step": 370
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00014309309309309308,
-      "loss": 0.4246,
+      "learning_rate": 0.00014294294294294295,
+      "loss": 0.3153,
       "step": 380
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0001415915915915916,
-      "loss": 0.3343,
+      "learning_rate": 0.00014144144144144144,
+      "loss": 0.4897,
       "step": 390
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001400900900900901,
-      "loss": 0.3619,
+      "learning_rate": 0.00013993993993993996,
+      "loss": 0.5959,
       "step": 400
     },
     {
       "epoch": 1.2,
-      "eval_accuracy": 0.8937300743889479,
-      "eval_loss": 0.4354824125766754,
-      "eval_runtime": 37.6727,
-      "eval_samples_per_second": 24.978,
-      "eval_steps_per_second": 3.132,
+      "eval_accuracy": 0.89798087141339,
+      "eval_loss": 0.40219178795814514,
+      "eval_runtime": 34.8649,
+      "eval_samples_per_second": 26.99,
+      "eval_steps_per_second": 3.384,
       "step": 400
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00013858858858858858,
-      "loss": 0.5238,
+      "learning_rate": 0.00013843843843843845,
+      "loss": 0.4482,
       "step": 410
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0001370870870870871,
-      "loss": 0.5456,
+      "learning_rate": 0.00013693693693693693,
+      "loss": 0.3453,
       "step": 420
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00013558558558558558,
-      "loss": 0.5347,
+      "learning_rate": 0.00013543543543543545,
+      "loss": 0.4837,
       "step": 430
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0001340840840840841,
-      "loss": 0.3739,
+      "learning_rate": 0.00013393393393393394,
+      "loss": 0.403,
       "step": 440
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00013258258258258258,
-      "loss": 0.392,
+      "learning_rate": 0.00013243243243243243,
+      "loss": 0.5089,
       "step": 450
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00013108108108108107,
-      "loss": 0.5306,
+      "learning_rate": 0.00013093093093093094,
+      "loss": 0.4843,
       "step": 460
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001295795795795796,
-      "loss": 0.4171,
+      "learning_rate": 0.00012942942942942943,
+      "loss": 0.3155,
       "step": 470
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00012807807807807808,
-      "loss": 0.3715,
+      "learning_rate": 0.00012792792792792795,
+      "loss": 0.4345,
       "step": 480
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00012657657657657657,
-      "loss": 0.5192,
+      "learning_rate": 0.00012642642642642644,
+      "loss": 0.4005,
       "step": 490
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00012507507507507508,
-      "loss": 0.34,
+      "learning_rate": 0.00012492492492492492,
+      "loss": 0.4478,
       "step": 500
     },
     {
       "epoch": 1.5,
-      "eval_accuracy": 0.9117959617428267,
-      "eval_loss": 0.3712367117404938,
-      "eval_runtime": 37.8371,
-      "eval_samples_per_second": 24.87,
-      "eval_steps_per_second": 3.119,
+      "eval_accuracy": 0.9287991498405951,
+      "eval_loss": 0.3246866464614868,
+      "eval_runtime": 33.5066,
+      "eval_samples_per_second": 28.084,
+      "eval_steps_per_second": 3.522,
       "step": 500
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00012357357357357357,
-      "loss": 0.5209,
+      "learning_rate": 0.00012342342342342344,
+      "loss": 0.4507,
       "step": 510
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00012207207207207209,
-      "loss": 0.4896,
+      "learning_rate": 0.00012192192192192193,
+      "loss": 0.5392,
       "step": 520
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00012057057057057057,
-      "loss": 0.5666,
+      "learning_rate": 0.00012042042042042043,
+      "loss": 0.2738,
       "step": 530
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00011906906906906908,
-      "loss": 0.3288,
+      "learning_rate": 0.00011891891891891893,
+      "loss": 0.566,
       "step": 540
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00011756756756756758,
-      "loss": 0.4315,
+      "learning_rate": 0.00011741741741741743,
+      "loss": 0.5345,
       "step": 550
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00011606606606606607,
-      "loss": 0.4185,
+      "learning_rate": 0.00011591591591591592,
+      "loss": 0.443,
       "step": 560
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00011456456456456457,
-      "loss": 0.2794,
+      "learning_rate": 0.00011441441441441443,
+      "loss": 0.346,
       "step": 570
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00011306306306306307,
-      "loss": 0.3125,
+      "learning_rate": 0.00011291291291291293,
+      "loss": 0.3985,
       "step": 580
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00011156156156156157,
-      "loss": 0.3229,
+      "learning_rate": 0.00011141141141141143,
+      "loss": 0.3706,
       "step": 590
     },
     {
       "epoch": 1.8,
       "learning_rate": 0.00011006006006006006,
-      "loss": 0.3413,
+      "loss": 0.4717,
       "step": 600
     },
     {
       "epoch": 1.8,
-      "eval_accuracy": 0.8916046758767269,
-      "eval_loss": 0.40877240896224976,
-      "eval_runtime": 39.2551,
-      "eval_samples_per_second": 23.971,
-      "eval_steps_per_second": 3.006,
+      "eval_accuracy": 0.926673751328374,
+      "eval_loss": 0.30190205574035645,
+      "eval_runtime": 33.6807,
+      "eval_samples_per_second": 27.939,
+      "eval_steps_per_second": 3.503,
       "step": 600
     },
     {
       "epoch": 1.83,
       "learning_rate": 0.00010855855855855856,
-      "loss": 0.3967,
+      "loss": 0.3582,
       "step": 610
     },
     {
       "epoch": 1.86,
       "learning_rate": 0.00010705705705705707,
-      "loss": 0.32,
+      "loss": 0.3629,
       "step": 620
     },
     {
       "epoch": 1.89,
       "learning_rate": 0.00010555555555555557,
-      "loss": 0.3256,
+      "loss": 0.3866,
       "step": 630
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0001042042042042042,
-      "loss": 0.4024,
+      "learning_rate": 0.00010405405405405406,
+      "loss": 0.3155,
       "step": 640
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0001027027027027027,
-      "loss": 0.4209,
+      "learning_rate": 0.00010255255255255256,
+      "loss": 0.4311,
       "step": 650
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0001012012012012012,
-      "loss": 0.3888,
+      "learning_rate": 0.00010105105105105106,
+      "loss": 0.2544,
       "step": 660
     },
     {
       "epoch": 2.01,
-      "learning_rate": 9.96996996996997e-05,
-      "loss": 0.3793,
+      "learning_rate": 9.954954954954956e-05,
+      "loss": 0.3617,
       "step": 670
     },
     {
       "epoch": 2.04,
-      "learning_rate": 9.81981981981982e-05,
-      "loss": 0.3094,
+      "learning_rate": 9.804804804804806e-05,
+      "loss": 0.3127,
       "step": 680
     },
     {
       "epoch": 2.07,
-      "learning_rate": 9.66966966966967e-05,
-      "loss": 0.3613,
+      "learning_rate": 9.654654654654654e-05,
+      "loss": 0.2718,
       "step": 690
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.51951951951952e-05,
-      "loss": 0.3619,
+      "learning_rate": 9.504504504504504e-05,
+      "loss": 0.34,
       "step": 700
     },
     {
       "epoch": 2.1,
-      "eval_accuracy": 0.9043570669500531,
-      "eval_loss": 0.37407416105270386,
-      "eval_runtime": 39.0938,
-      "eval_samples_per_second": 24.07,
-      "eval_steps_per_second": 3.018,
+      "eval_accuracy": 0.9351753453772582,
+      "eval_loss": 0.25940415263175964,
+      "eval_runtime": 35.0686,
+      "eval_samples_per_second": 26.833,
+      "eval_steps_per_second": 3.365,
       "step": 700
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.36936936936937e-05,
-      "loss": 0.3051,
+      "learning_rate": 9.354354354354354e-05,
+      "loss": 0.3881,
       "step": 710
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.219219219219219e-05,
-      "loss": 0.4062,
+      "learning_rate": 9.204204204204205e-05,
+      "loss": 0.3528,
       "step": 720
     },
     {
       "epoch": 2.19,
-      "learning_rate": 9.069069069069069e-05,
-      "loss": 0.3811,
+      "learning_rate": 9.054054054054055e-05,
+      "loss": 0.3053,
       "step": 730
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.918918918918919e-05,
-      "loss": 0.3118,
+      "learning_rate": 8.903903903903904e-05,
+      "loss": 0.223,
       "step": 740
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.76876876876877e-05,
-      "loss": 0.3014,
+      "learning_rate": 8.753753753753754e-05,
+      "loss": 0.2268,
       "step": 750
     },
     {
       "epoch": 2.28,
-      "learning_rate": 8.61861861861862e-05,
-      "loss": 0.2404,
+      "learning_rate": 8.603603603603604e-05,
+      "loss": 0.4058,
       "step": 760
     },
     {
       "epoch": 2.31,
-      "learning_rate": 8.468468468468469e-05,
-      "loss": 0.3802,
+      "learning_rate": 8.453453453453454e-05,
+      "loss": 0.3018,
       "step": 770
     },
     {
       "epoch": 2.34,
-      "learning_rate": 8.318318318318319e-05,
-      "loss": 0.3222,
+      "learning_rate": 8.303303303303303e-05,
+      "loss": 0.2973,
       "step": 780
     },
     {
       "epoch": 2.37,
-      "learning_rate": 8.168168168168169e-05,
-      "loss": 0.2459,
+      "learning_rate": 8.153153153153153e-05,
+      "loss": 0.2607,
       "step": 790
     },
     {
       "epoch": 2.4,
-      "learning_rate": 8.018018018018019e-05,
-      "loss": 0.2135,
+      "learning_rate": 8.003003003003004e-05,
+      "loss": 0.3518,
       "step": 800
     },
     {
       "epoch": 2.4,
-      "eval_accuracy": 0.9160467587672688,
-      "eval_loss": 0.328594833612442,
-      "eval_runtime": 38.4594,
-      "eval_samples_per_second": 24.467,
-      "eval_steps_per_second": 3.068,
+      "eval_accuracy": 0.9351753453772582,
+      "eval_loss": 0.2507326304912567,
+      "eval_runtime": 34.9575,
+      "eval_samples_per_second": 26.918,
+      "eval_steps_per_second": 3.376,
       "step": 800
     },
     {
       "epoch": 2.43,
-      "learning_rate": 7.867867867867868e-05,
-      "loss": 0.2152,
+      "learning_rate": 7.852852852852854e-05,
+      "loss": 0.2251,
       "step": 810
     },
     {
       "epoch": 2.46,
-      "learning_rate": 7.717717717717718e-05,
-      "loss": 0.2621,
+      "learning_rate": 7.702702702702703e-05,
+      "loss": 0.2747,
       "step": 820
     },
     {
       "epoch": 2.49,
-      "learning_rate": 7.567567567567568e-05,
-      "loss": 0.2118,
+      "learning_rate": 7.552552552552553e-05,
+      "loss": 0.2653,
       "step": 830
     },
     {
       "epoch": 2.52,
-      "learning_rate": 7.417417417417419e-05,
-      "loss": 0.2826,
+      "learning_rate": 7.402402402402403e-05,
+      "loss": 0.2361,
       "step": 840
     },
     {
       "epoch": 2.55,
-      "learning_rate": 7.267267267267268e-05,
-      "loss": 0.4047,
+      "learning_rate": 7.252252252252253e-05,
+      "loss": 0.2548,
       "step": 850
     },
     {
       "epoch": 2.58,
-      "learning_rate": 7.117117117117116e-05,
-      "loss": 0.4956,
+      "learning_rate": 7.102102102102103e-05,
+      "loss": 0.392,
       "step": 860
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.966966966966967e-05,
-      "loss": 0.2431,
+      "learning_rate": 6.951951951951952e-05,
+      "loss": 0.3041,
       "step": 870
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.816816816816817e-05,
-      "loss": 0.2189,
+      "learning_rate": 6.801801801801802e-05,
+      "loss": 0.3341,
       "step": 880
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.666666666666667e-05,
-      "loss": 0.2455,
+      "learning_rate": 6.651651651651653e-05,
+      "loss": 0.2462,
       "step": 890
     },
     {
       "epoch": 2.7,
-      "learning_rate": 6.516516516516516e-05,
-      "loss": 0.2166,
+      "learning_rate": 6.501501501501502e-05,
+      "loss": 0.3352,
       "step": 900
     },
     {
       "epoch": 2.7,
-      "eval_accuracy": 0.9415515409139213,
-      "eval_loss": 0.2758049964904785,
-      "eval_runtime": 38.181,
-      "eval_samples_per_second": 24.646,
-      "eval_steps_per_second": 3.091,
+      "eval_accuracy": 0.9426142401700319,
+      "eval_loss": 0.2483620047569275,
+      "eval_runtime": 33.9353,
+      "eval_samples_per_second": 27.729,
+      "eval_steps_per_second": 3.477,
       "step": 900
     },
     {
       "epoch": 2.73,
-      "learning_rate": 6.366366366366366e-05,
-      "loss": 0.2648,
+      "learning_rate": 6.351351351351352e-05,
+      "loss": 0.2314,
       "step": 910
     },
     {
       "epoch": 2.76,
-      "learning_rate": 6.216216216216216e-05,
-      "loss": 0.2233,
+      "learning_rate": 6.2012012012012e-05,
+      "loss": 0.2739,
       "step": 920
     },
     {
       "epoch": 2.79,
-      "learning_rate": 6.0660660660660665e-05,
-      "loss": 0.2876,
+      "learning_rate": 6.051051051051051e-05,
+      "loss": 0.2905,
       "step": 930
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.915915915915916e-05,
-      "loss": 0.3385,
+      "learning_rate": 5.900900900900901e-05,
+      "loss": 0.2607,
       "step": 940
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.765765765765766e-05,
-      "loss": 0.2823,
+      "learning_rate": 5.7507507507507506e-05,
+      "loss": 0.334,
       "step": 950
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.615615615615616e-05,
-      "loss": 0.135,
+      "learning_rate": 5.600600600600601e-05,
+      "loss": 0.3202,
       "step": 960
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.465465465465466e-05,
-      "loss": 0.289,
+      "learning_rate": 5.45045045045045e-05,
+      "loss": 0.3326,
       "step": 970
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.3153153153153155e-05,
-      "loss": 0.2228,
+      "learning_rate": 5.3003003003003005e-05,
+      "loss": 0.2717,
       "step": 980
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.165165165165166e-05,
-      "loss": 0.3221,
+      "learning_rate": 5.15015015015015e-05,
+      "loss": 0.259,
       "step": 990
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.015015015015015e-05,
-      "loss": 0.1557,
+      "learning_rate": 5e-05,
+      "loss": 0.2493,
       "step": 1000
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.9330499468650372,
-      "eval_loss": 0.2679324150085449,
-      "eval_runtime": 37.7775,
-      "eval_samples_per_second": 24.909,
-      "eval_steps_per_second": 3.124,
+      "eval_accuracy": 0.9394261424017003,
+      "eval_loss": 0.22659793496131897,
+      "eval_runtime": 33.5582,
+      "eval_samples_per_second": 28.041,
+      "eval_steps_per_second": 3.516,
       "step": 1000
     },
     {
       "epoch": 3.03,
-      "learning_rate": 4.8648648648648654e-05,
-      "loss": 0.1808,
+      "learning_rate": 4.8498498498498504e-05,
+      "loss": 0.2026,
       "step": 1010
     },
     {
       "epoch": 3.06,
-      "learning_rate": 4.714714714714715e-05,
-      "loss": 0.2742,
+      "learning_rate": 4.6996996996997e-05,
+      "loss": 0.189,
       "step": 1020
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.5645645645645645e-05,
-      "loss": 0.147,
+      "learning_rate": 4.54954954954955e-05,
+      "loss": 0.1097,
       "step": 1030
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.414414414414415e-05,
-      "loss": 0.248,
+      "learning_rate": 4.3993993993994e-05,
+      "loss": 0.1605,
       "step": 1040
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.264264264264264e-05,
-      "loss": 0.1214,
+      "learning_rate": 4.24924924924925e-05,
+      "loss": 0.2681,
       "step": 1050
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.1141141141141144e-05,
-      "loss": 0.352,
+      "learning_rate": 4.099099099099099e-05,
+      "loss": 0.2872,
       "step": 1060
     },
     {
       "epoch": 3.21,
-      "learning_rate": 3.963963963963964e-05,
-      "loss": 0.2441,
+      "learning_rate": 3.948948948948949e-05,
+      "loss": 0.3481,
       "step": 1070
     },
     {
       "epoch": 3.24,
-      "learning_rate": 3.813813813813814e-05,
-      "loss": 0.2205,
+      "learning_rate": 3.7987987987987985e-05,
+      "loss": 0.1615,
       "step": 1080
     },
     {
       "epoch": 3.27,
-      "learning_rate": 3.663663663663664e-05,
-      "loss": 0.1925,
+      "learning_rate": 3.648648648648649e-05,
+      "loss": 0.1447,
       "step": 1090
     },
     {
       "epoch": 3.3,
-      "learning_rate": 3.513513513513514e-05,
-      "loss": 0.1115,
+      "learning_rate": 3.498498498498499e-05,
+      "loss": 0.2034,
       "step": 1100
     },
     {
       "epoch": 3.3,
-      "eval_accuracy": 0.9362380446333688,
-      "eval_loss": 0.2528650462627411,
-      "eval_runtime": 38.8623,
-      "eval_samples_per_second": 24.214,
-      "eval_steps_per_second": 3.036,
+      "eval_accuracy": 0.9479277364505845,
+      "eval_loss": 0.20114077627658844,
+      "eval_runtime": 34.4276,
+      "eval_samples_per_second": 27.333,
+      "eval_steps_per_second": 3.427,
       "step": 1100
     },
     {
       "epoch": 3.33,
-      "learning_rate": 3.3633633633633635e-05,
-      "loss": 0.2245,
+      "learning_rate": 3.3483483483483485e-05,
+      "loss": 0.204,
       "step": 1110
     },
     {
       "epoch": 3.36,
-      "learning_rate": 3.213213213213213e-05,
-      "loss": 0.2538,
+      "learning_rate": 3.198198198198199e-05,
+      "loss": 0.18,
       "step": 1120
     },
     {
       "epoch": 3.39,
-      "learning_rate": 3.063063063063063e-05,
-      "loss": 0.1725,
+      "learning_rate": 3.0480480480480482e-05,
+      "loss": 0.1657,
       "step": 1130
     },
     {
       "epoch": 3.42,
-      "learning_rate": 2.912912912912913e-05,
-      "loss": 0.3009,
+      "learning_rate": 2.897897897897898e-05,
+      "loss": 0.1932,
       "step": 1140
     },
     {
       "epoch": 3.45,
-      "learning_rate": 2.762762762762763e-05,
-      "loss": 0.3042,
+      "learning_rate": 2.7477477477477483e-05,
+      "loss": 0.1524,
       "step": 1150
     },
     {
       "epoch": 3.48,
-      "learning_rate": 2.6126126126126128e-05,
-      "loss": 0.1748,
+      "learning_rate": 2.5975975975975975e-05,
+      "loss": 0.1144,
       "step": 1160
     },
     {
       "epoch": 3.51,
-      "learning_rate": 2.4624624624624627e-05,
-      "loss": 0.1229,
+      "learning_rate": 2.4474474474474477e-05,
+      "loss": 0.1424,
       "step": 1170
     },
     {
       "epoch": 3.54,
-      "learning_rate": 2.3123123123123125e-05,
-      "loss": 0.2376,
+      "learning_rate": 2.2972972972972976e-05,
+      "loss": 0.2142,
       "step": 1180
     },
     {
       "epoch": 3.57,
       "learning_rate": 2.1621621621621624e-05,
-      "loss": 0.2025,
+      "loss": 0.1448,
       "step": 1190
     },
     {
       "epoch": 3.6,
       "learning_rate": 2.012012012012012e-05,
-      "loss": 0.1571,
+      "loss": 0.1753,
       "step": 1200
     },
     {
       "epoch": 3.6,
-      "eval_accuracy": 0.946865037194474,
-      "eval_loss": 0.23596596717834473,
-      "eval_runtime": 38.9005,
-      "eval_samples_per_second": 24.19,
-      "eval_steps_per_second": 3.033,
+      "eval_accuracy": 0.944739638682253,
+      "eval_loss": 0.20892110466957092,
+      "eval_runtime": 33.5408,
+      "eval_samples_per_second": 28.055,
+      "eval_steps_per_second": 3.518,
       "step": 1200
     },
     {
       "epoch": 3.63,
       "learning_rate": 1.8618618618618618e-05,
-      "loss": 0.0491,
+      "loss": 0.2085,
       "step": 1210
     },
     {
       "epoch": 3.66,
       "learning_rate": 1.7117117117117117e-05,
-      "loss": 0.1921,
+      "loss": 0.2179,
       "step": 1220
     },
     {
       "epoch": 3.69,
       "learning_rate": 1.5615615615615616e-05,
-      "loss": 0.1733,
+      "loss": 0.1755,
       "step": 1230
     },
     {
       "epoch": 3.72,
       "learning_rate": 1.4114114114114116e-05,
-      "loss": 0.1822,
+      "loss": 0.1905,
       "step": 1240
     },
     {
       "epoch": 3.75,
       "learning_rate": 1.2612612612612611e-05,
-      "loss": 0.2661,
+      "loss": 0.2246,
       "step": 1250
     },
     {
       "epoch": 3.78,
       "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.1566,
+      "loss": 0.1495,
       "step": 1260
     },
     {
       "epoch": 3.81,
       "learning_rate": 9.60960960960961e-06,
-      "loss": 0.1764,
+      "loss": 0.2038,
       "step": 1270
     },
     {
       "epoch": 3.84,
       "learning_rate": 8.108108108108109e-06,
-      "loss": 0.1824,
+      "loss": 0.2163,
       "step": 1280
     },
     {
       "epoch": 3.87,
       "learning_rate": 6.606606606606607e-06,
-      "loss": 0.122,
+      "loss": 0.152,
       "step": 1290
     },
     {
       "epoch": 3.9,
       "learning_rate": 5.105105105105106e-06,
-      "loss": 0.1079,
+      "loss": 0.1614,
       "step": 1300
     },
     {
       "epoch": 3.9,
-      "eval_accuracy": 0.9521785334750266,
-      "eval_loss": 0.21390262246131897,
-      "eval_runtime": 38.2092,
-      "eval_samples_per_second": 24.628,
-      "eval_steps_per_second": 3.088,
+      "eval_accuracy": 0.9543039319872476,
+      "eval_loss": 0.19176600873470306,
+      "eval_runtime": 33.687,
+      "eval_samples_per_second": 27.934,
+      "eval_steps_per_second": 3.503,
       "step": 1300
     },
     {
       "epoch": 3.93,
       "learning_rate": 3.603603603603604e-06,
-      "loss": 0.1973,
+      "loss": 0.1276,
       "step": 1310
     },
     {
       "epoch": 3.96,
       "learning_rate": 2.102102102102102e-06,
-      "loss": 0.0976,
+      "loss": 0.2218,
       "step": 1320
     },
     {
       "epoch": 3.99,
       "learning_rate": 6.006006006006006e-07,
-      "loss": 0.1724,
+      "loss": 0.1194,
       "step": 1330
     },
     {
       "epoch": 4.0,
       "step": 1332,
       "total_flos": 1.651775717862015e+18,
-      "train_loss": 0.4871297548483084,
-      "train_runtime": 1589.6821,
-      "train_samples_per_second": 13.406,
-      "train_steps_per_second": 0.838
+      "train_loss": 0.4876434194433081,
+      "train_runtime": 1561.5053,
+      "train_samples_per_second": 13.648,
+      "train_steps_per_second": 0.853
     }
   ],
   "max_steps": 1332,