🍻 cheers

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 4.0,
-    "eval_accuracy": 0.9543039319872476,
-    "eval_loss": 0.19176600873470306,
-    "eval_runtime": 33.828,
-    "eval_samples_per_second": 27.817,
-    "eval_steps_per_second": 3.488,
+    "eval_accuracy": 0.9330499468650372,
+    "eval_loss": 0.263217031955719,
+    "eval_runtime": 37.3903,
+    "eval_samples_per_second": 25.167,
+    "eval_steps_per_second": 3.156,
     "total_flos": 1.651775717862015e+18,
-    "train_loss": 0.4876434194433081,
-    "train_runtime": 1561.5053,
-    "train_samples_per_second": 13.648,
-    "train_steps_per_second": 0.853
+    "train_loss": 0.4671340096104252,
+    "train_runtime": 1617.2249,
+    "train_samples_per_second": 13.178,
+    "train_steps_per_second": 0.824
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
-    "eval_accuracy": 0.9543039319872476,
-    "eval_loss": 0.19176600873470306,
-    "eval_runtime": 33.828,
-    "eval_samples_per_second": 27.817,
-    "eval_steps_per_second": 3.488
+    "eval_accuracy": 0.9330499468650372,
+    "eval_loss": 0.263217031955719,
+    "eval_runtime": 37.3903,
+    "eval_samples_per_second": 25.167,
+    "eval_steps_per_second": 3.156
 }

runs/Aug04_18-07-39_26434d7bb3db/events.out.tfevents.1659638706.26434d7bb3db.71.2

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7ca0a601772eb39bfa464c7e6ad56c189320ff6d9de5958c31880d97590da0e
+size 363

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
     "total_flos": 1.651775717862015e+18,
-    "train_loss": 0.4876434194433081,
-    "train_runtime": 1561.5053,
-    "train_samples_per_second": 13.648,
-    "train_steps_per_second": 0.853
+    "train_loss": 0.4671340096104252,
+    "train_runtime": 1617.2249,
+    "train_samples_per_second": 13.178,
+    "train_steps_per_second": 0.824
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.19176600873470306,
+  "best_metric": 0.263217031955719,
   "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": 2.8856,
+      "loss": 2.8484,
       "step": 10
     },
     {
       "epoch": 0.06,
       "learning_rate": 0.00019699699699699701,
-      "loss": 2.5941,
+      "loss": 2.5652,
       "step": 20
     },
     {
       "epoch": 0.09,
       "learning_rate": 0.0001954954954954955,
-      "loss": 2.1576,
+      "loss": 2.2041,
       "step": 30
     },
     {
       "epoch": 0.12,
       "learning_rate": 0.00019399399399399402,
-      "loss": 1.9221,
+      "loss": 1.8817,
       "step": 40
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.0001924924924924925,
-      "loss": 1.7671,
+      "loss": 1.7182,
       "step": 50
     },
     {
       "epoch": 0.18,
       "learning_rate": 0.000190990990990991,
-      "loss": 1.4461,
+      "loss": 1.4787,
       "step": 60
     },
     {
       "epoch": 0.21,
       "learning_rate": 0.0001894894894894895,
-      "loss": 1.3917,
+      "loss": 1.3547,
       "step": 70
     },
     {
       "epoch": 0.24,
       "learning_rate": 0.000187987987987988,
-      "loss": 1.3666,
+      "loss": 1.1844,
       "step": 80
     },
     {
       "epoch": 0.27,
       "learning_rate": 0.0001864864864864865,
-      "loss": 1.1615,
+      "loss": 1.1045,
       "step": 90
     },
     {
       "epoch": 0.3,
       "learning_rate": 0.000184984984984985,
-      "loss": 1.0175,
+      "loss": 1.1794,
       "step": 100
     },
     {
       "epoch": 0.3,
-      "eval_accuracy": 0.8629117959617428,
-      "eval_loss": 0.9246562719345093,
-      "eval_runtime": 35.8495,
-      "eval_samples_per_second": 26.249,
-      "eval_steps_per_second": 3.292,
+      "eval_accuracy": 0.8565356004250797,
+      "eval_loss": 0.9208076596260071,
+      "eval_runtime": 37.6614,
+      "eval_samples_per_second": 24.986,
+      "eval_steps_per_second": 3.133,
       "step": 100
     },
     {
       "epoch": 0.33,
       "learning_rate": 0.0001834834834834835,
-      "loss": 1.044,
+      "loss": 1.0324,
       "step": 110
     },
     {
       "epoch": 0.36,
       "learning_rate": 0.000181981981981982,
-      "loss": 0.9848,
+      "loss": 0.9189,
       "step": 120
     },
     {
       "epoch": 0.39,
       "learning_rate": 0.0001804804804804805,
-      "loss": 0.7889,
+      "loss": 0.9565,
       "step": 130
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00017897897897897898,
-      "loss": 0.921,
+      "learning_rate": 0.00017912912912912914,
+      "loss": 0.7858,
       "step": 140
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0001774774774774775,
-      "loss": 0.7709,
+      "learning_rate": 0.00017762762762762763,
+      "loss": 0.7527,
       "step": 150
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.000175975975975976,
-      "loss": 0.9096,
+      "learning_rate": 0.00017612612612612615,
+      "loss": 0.7514,
       "step": 160
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0001744744744744745,
-      "loss": 0.7293,
+      "learning_rate": 0.00017462462462462463,
+      "loss": 0.8192,
       "step": 170
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.000172972972972973,
-      "loss": 0.6419,
+      "learning_rate": 0.00017312312312312312,
+      "loss": 0.8094,
       "step": 180
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00017147147147147148,
-      "loss": 0.7251,
+      "learning_rate": 0.00017162162162162164,
+      "loss": 0.6748,
       "step": 190
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00016996996996997,
-      "loss": 0.7418,
+      "learning_rate": 0.00017012012012012013,
+      "loss": 0.6513,
       "step": 200
     },
     {
       "epoch": 0.6,
-      "eval_accuracy": 0.8990435706695006,
-      "eval_loss": 0.5535812377929688,
-      "eval_runtime": 33.962,
-      "eval_samples_per_second": 27.707,
-      "eval_steps_per_second": 3.474,
+      "eval_accuracy": 0.8841657810839533,
+      "eval_loss": 0.540999174118042,
+      "eval_runtime": 35.8211,
+      "eval_samples_per_second": 26.269,
+      "eval_steps_per_second": 3.294,
       "step": 200
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00016846846846846846,
-      "loss": 0.7307,
+      "learning_rate": 0.00016861861861861864,
+      "loss": 0.6462,
       "step": 210
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00016696696696696697,
-      "loss": 0.7721,
+      "learning_rate": 0.00016711711711711713,
+      "loss": 0.6141,
       "step": 220
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.00016546546546546546,
-      "loss": 0.6073,
+      "learning_rate": 0.00016561561561561562,
+      "loss": 0.5329,
       "step": 230
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00016396396396396395,
-      "loss": 0.6946,
+      "learning_rate": 0.00016411411411411413,
+      "loss": 0.7009,
       "step": 240
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00016246246246246247,
-      "loss": 0.7824,
+      "learning_rate": 0.00016261261261261262,
+      "loss": 0.6468,
       "step": 250
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00016096096096096096,
-      "loss": 0.5175,
+      "learning_rate": 0.0001611111111111111,
+      "loss": 0.712,
       "step": 260
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00015945945945945947,
-      "loss": 0.7959,
+      "learning_rate": 0.00015960960960960963,
+      "loss": 0.6749,
       "step": 270
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00015795795795795796,
-      "loss": 0.5273,
+      "learning_rate": 0.00015810810810810812,
+      "loss": 0.7913,
       "step": 280
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00015645645645645645,
-      "loss": 0.6327,
+      "learning_rate": 0.00015660660660660663,
+      "loss": 0.6343,
       "step": 290
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00015495495495495496,
-      "loss": 0.6652,
+      "learning_rate": 0.00015510510510510512,
+      "loss": 0.5904,
       "step": 300
     },
     {
       "epoch": 0.9,
-      "eval_accuracy": 0.9181721572794899,
-      "eval_loss": 0.4035964906215668,
-      "eval_runtime": 33.8609,
-      "eval_samples_per_second": 27.79,
-      "eval_steps_per_second": 3.485,
+      "eval_accuracy": 0.8799149840595112,
+      "eval_loss": 0.4978478252887726,
+      "eval_runtime": 36.7863,
+      "eval_samples_per_second": 25.58,
+      "eval_steps_per_second": 3.208,
       "step": 300
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00015345345345345345,
-      "loss": 0.6688,
+      "learning_rate": 0.0001536036036036036,
+      "loss": 0.5076,
       "step": 310
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00015195195195195194,
-      "loss": 0.5384,
+      "learning_rate": 0.00015210210210210212,
+      "loss": 0.544,
       "step": 320
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00015045045045045046,
-      "loss": 0.5232,
+      "learning_rate": 0.0001506006006006006,
+      "loss": 0.4454,
       "step": 330
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00014894894894894895,
-      "loss": 0.5407,
+      "learning_rate": 0.00014909909909909913,
+      "loss": 0.5019,
       "step": 340
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00014744744744744746,
-      "loss": 0.512,
+      "learning_rate": 0.00014759759759759762,
+      "loss": 0.6077,
       "step": 350
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00014594594594594595,
-      "loss": 0.4422,
+      "learning_rate": 0.0001460960960960961,
+      "loss": 0.3247,
       "step": 360
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00014444444444444444,
-      "loss": 0.6211,
+      "learning_rate": 0.00014459459459459462,
+      "loss": 0.3434,
       "step": 370
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00014294294294294295,
-      "loss": 0.3153,
+      "learning_rate": 0.00014309309309309308,
+      "loss": 0.57,
       "step": 380
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00014144144144144144,
-      "loss": 0.4897,
+      "learning_rate": 0.0001415915915915916,
+      "loss": 0.4769,
       "step": 390
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00013993993993993996,
-      "loss": 0.5959,
+      "learning_rate": 0.0001400900900900901,
+      "loss": 0.4461,
       "step": 400
     },
     {
       "epoch": 1.2,
-      "eval_accuracy": 0.89798087141339,
-      "eval_loss": 0.40219178795814514,
-      "eval_runtime": 34.8649,
-      "eval_samples_per_second": 26.99,
-      "eval_steps_per_second": 3.384,
+      "eval_accuracy": 0.9192348565356004,
+      "eval_loss": 0.3668522834777832,
+      "eval_runtime": 38.3659,
+      "eval_samples_per_second": 24.527,
+      "eval_steps_per_second": 3.076,
       "step": 400
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00013843843843843845,
-      "loss": 0.4482,
+      "learning_rate": 0.00013858858858858858,
+      "loss": 0.4357,
       "step": 410
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00013693693693693693,
-      "loss": 0.3453,
+      "learning_rate": 0.0001370870870870871,
+      "loss": 0.5206,
       "step": 420
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00013543543543543545,
-      "loss": 0.4837,
+      "learning_rate": 0.00013558558558558558,
+      "loss": 0.3509,
       "step": 430
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00013393393393393394,
-      "loss": 0.403,
+      "learning_rate": 0.0001340840840840841,
+      "loss": 0.4684,
       "step": 440
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00013243243243243243,
-      "loss": 0.5089,
+      "learning_rate": 0.00013258258258258258,
+      "loss": 0.5342,
       "step": 450
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00013093093093093094,
-      "loss": 0.4843,
+      "learning_rate": 0.00013108108108108107,
+      "loss": 0.5583,
       "step": 460
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00012942942942942943,
-      "loss": 0.3155,
+      "learning_rate": 0.0001295795795795796,
+      "loss": 0.3255,
       "step": 470
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00012792792792792795,
-      "loss": 0.4345,
+      "learning_rate": 0.00012807807807807808,
+      "loss": 0.4101,
       "step": 480
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00012642642642642644,
-      "loss": 0.4005,
+      "learning_rate": 0.00012657657657657657,
+      "loss": 0.3863,
       "step": 490
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00012492492492492492,
-      "loss": 0.4478,
+      "learning_rate": 0.00012507507507507508,
+      "loss": 0.5633,
       "step": 500
     },
     {
       "epoch": 1.5,
-      "eval_accuracy": 0.9287991498405951,
-      "eval_loss": 0.3246866464614868,
-      "eval_runtime": 33.5066,
-      "eval_samples_per_second": 28.084,
-      "eval_steps_per_second": 3.522,
+      "eval_accuracy": 0.8841657810839533,
+      "eval_loss": 0.4339892268180847,
+      "eval_runtime": 37.8861,
+      "eval_samples_per_second": 24.838,
+      "eval_steps_per_second": 3.115,
       "step": 500
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00012342342342342344,
-      "loss": 0.4507,
+      "learning_rate": 0.00012357357357357357,
+      "loss": 0.4033,
       "step": 510
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00012192192192192193,
-      "loss": 0.5392,
+      "learning_rate": 0.00012207207207207209,
+      "loss": 0.3891,
       "step": 520
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00012042042042042043,
-      "loss": 0.2738,
+      "learning_rate": 0.00012057057057057057,
+      "loss": 0.4604,
       "step": 530
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00011891891891891893,
-      "loss": 0.566,
+      "learning_rate": 0.00011906906906906908,
+      "loss": 0.3802,
       "step": 540
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00011741741741741743,
-      "loss": 0.5345,
+      "learning_rate": 0.00011756756756756758,
+      "loss": 0.3528,
       "step": 550
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00011591591591591592,
-      "loss": 0.443,
+      "learning_rate": 0.00011606606606606607,
+      "loss": 0.3331,
       "step": 560
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00011441441441441443,
-      "loss": 0.346,
+      "learning_rate": 0.00011456456456456457,
+      "loss": 0.4153,
       "step": 570
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00011291291291291293,
-      "loss": 0.3985,
+      "learning_rate": 0.00011306306306306307,
+      "loss": 0.3143,
       "step": 580
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00011141141141141143,
-      "loss": 0.3706,
+      "learning_rate": 0.00011156156156156157,
+      "loss": 0.5106,
       "step": 590
     },
     {
       "epoch": 1.8,
       "learning_rate": 0.00011006006006006006,
-      "loss": 0.4717,
+      "loss": 0.2489,
       "step": 600
     },
     {
       "epoch": 1.8,
-      "eval_accuracy": 0.926673751328374,
-      "eval_loss": 0.30190205574035645,
-      "eval_runtime": 33.6807,
-      "eval_samples_per_second": 27.939,
-      "eval_steps_per_second": 3.503,
+      "eval_accuracy": 0.9171094580233794,
+      "eval_loss": 0.3354736864566803,
+      "eval_runtime": 38.0141,
+      "eval_samples_per_second": 24.754,
+      "eval_steps_per_second": 3.104,
       "step": 600
     },
     {
       "epoch": 1.83,
       "learning_rate": 0.00010855855855855856,
-      "loss": 0.3582,
+      "loss": 0.2632,
       "step": 610
     },
     {
       "epoch": 1.86,
       "learning_rate": 0.00010705705705705707,
-      "loss": 0.3629,
+      "loss": 0.2744,
       "step": 620
     },
     {
       "epoch": 1.89,
       "learning_rate": 0.00010555555555555557,
-      "loss": 0.3866,
+      "loss": 0.4888,
       "step": 630
     },
     {
       "epoch": 1.92,
       "learning_rate": 0.00010405405405405406,
-      "loss": 0.3155,
+      "loss": 0.3897,
       "step": 640
     },
     {
       "epoch": 1.95,
       "learning_rate": 0.00010255255255255256,
-      "loss": 0.4311,
+      "loss": 0.4025,
       "step": 650
     },
     {
       "epoch": 1.98,
       "learning_rate": 0.00010105105105105106,
-      "loss": 0.2544,
+      "loss": 0.3853,
       "step": 660
     },
     {
       "epoch": 2.01,
       "learning_rate": 9.954954954954956e-05,
-      "loss": 0.3617,
+      "loss": 0.4504,
       "step": 670
     },
     {
       "epoch": 2.04,
       "learning_rate": 9.804804804804806e-05,
-      "loss": 0.3127,
+      "loss": 0.3721,
       "step": 680
     },
     {
       "epoch": 2.07,
       "learning_rate": 9.654654654654654e-05,
-      "loss": 0.2718,
+      "loss": 0.2667,
       "step": 690
     },
     {
       "epoch": 2.1,
       "learning_rate": 9.504504504504504e-05,
-      "loss": 0.34,
+      "loss": 0.3171,
       "step": 700
     },
     {
       "epoch": 2.1,
-      "eval_accuracy": 0.9351753453772582,
-      "eval_loss": 0.25940415263175964,
-      "eval_runtime": 35.0686,
-      "eval_samples_per_second": 26.833,
-      "eval_steps_per_second": 3.365,
+      "eval_accuracy": 0.9192348565356004,
+      "eval_loss": 0.3286140263080597,
+      "eval_runtime": 39.0994,
+      "eval_samples_per_second": 24.067,
+      "eval_steps_per_second": 3.018,
       "step": 700
     },
     {
       "epoch": 2.13,
       "learning_rate": 9.354354354354354e-05,
-      "loss": 0.3881,
+      "loss": 0.1323,
       "step": 710
     },
     {
       "epoch": 2.16,
       "learning_rate": 9.204204204204205e-05,
-      "loss": 0.3528,
+      "loss": 0.2416,
       "step": 720
     },
     {
       "epoch": 2.19,
       "learning_rate": 9.054054054054055e-05,
-      "loss": 0.3053,
+      "loss": 0.2462,
       "step": 730
     },
     {
       "epoch": 2.22,
       "learning_rate": 8.903903903903904e-05,
-      "loss": 0.223,
+      "loss": 0.2489,
       "step": 740
     },
     {
       "epoch": 2.25,
       "learning_rate": 8.753753753753754e-05,
-      "loss": 0.2268,
+      "loss": 0.2155,
       "step": 750
     },
     {
       "epoch": 2.28,
       "learning_rate": 8.603603603603604e-05,
-      "loss": 0.4058,
+      "loss": 0.3265,
       "step": 760
     },
     {
       "epoch": 2.31,
       "learning_rate": 8.453453453453454e-05,
-      "loss": 0.3018,
+      "loss": 0.4114,
       "step": 770
     },
     {
       "epoch": 2.34,
       "learning_rate": 8.303303303303303e-05,
-      "loss": 0.2973,
+      "loss": 0.2201,
       "step": 780
     },
     {
       "epoch": 2.37,
       "learning_rate": 8.153153153153153e-05,
-      "loss": 0.2607,
+      "loss": 0.2655,
       "step": 790
     },
     {
       "epoch": 2.4,
       "learning_rate": 8.003003003003004e-05,
-      "loss": 0.3518,
+      "loss": 0.3785,
       "step": 800
     },
     {
       "epoch": 2.4,
-      "eval_accuracy": 0.9351753453772582,
-      "eval_loss": 0.2507326304912567,
-      "eval_runtime": 34.9575,
-      "eval_samples_per_second": 26.918,
-      "eval_steps_per_second": 3.376,
+      "eval_accuracy": 0.9171094580233794,
+      "eval_loss": 0.32322293519973755,
+      "eval_runtime": 38.4173,
+      "eval_samples_per_second": 24.494,
+      "eval_steps_per_second": 3.072,
       "step": 800
     },
     {
       "epoch": 2.43,
       "learning_rate": 7.852852852852854e-05,
-      "loss": 0.2251,
+      "loss": 0.188,
       "step": 810
     },
     {
       "epoch": 2.46,
       "learning_rate": 7.702702702702703e-05,
-      "loss": 0.2747,
+      "loss": 0.2789,
       "step": 820
     },
     {
       "epoch": 2.49,
       "learning_rate": 7.552552552552553e-05,
-      "loss": 0.2653,
+      "loss": 0.2386,
       "step": 830
     },
     {
       "epoch": 2.52,
       "learning_rate": 7.402402402402403e-05,
-      "loss": 0.2361,
+      "loss": 0.3068,
       "step": 840
     },
     {
       "epoch": 2.55,
       "learning_rate": 7.252252252252253e-05,
-      "loss": 0.2548,
+      "loss": 0.2085,
       "step": 850
     },
     {
       "epoch": 2.58,
       "learning_rate": 7.102102102102103e-05,
-      "loss": 0.392,
+      "loss": 0.2076,
       "step": 860
     },
     {
       "epoch": 2.61,
       "learning_rate": 6.951951951951952e-05,
-      "loss": 0.3041,
+      "loss": 0.1623,
       "step": 870
     },
     {
       "epoch": 2.64,
       "learning_rate": 6.801801801801802e-05,
-      "loss": 0.3341,
+      "loss": 0.3782,
       "step": 880
     },
     {
       "epoch": 2.67,
       "learning_rate": 6.651651651651653e-05,
-      "loss": 0.2462,
+      "loss": 0.1853,
       "step": 890
     },
     {
       "epoch": 2.7,
       "learning_rate": 6.501501501501502e-05,
-      "loss": 0.3352,
+      "loss": 0.2278,
       "step": 900
     },
     {
       "epoch": 2.7,
-      "eval_accuracy": 0.9426142401700319,
-      "eval_loss": 0.2483620047569275,
-      "eval_runtime": 33.9353,
-      "eval_samples_per_second": 27.729,
-      "eval_steps_per_second": 3.477,
+      "eval_accuracy": 0.9192348565356004,
+      "eval_loss": 0.33382856845855713,
+      "eval_runtime": 38.8106,
+      "eval_samples_per_second": 24.246,
+      "eval_steps_per_second": 3.04,
       "step": 900
     },
     {
       "epoch": 2.73,
       "learning_rate": 6.351351351351352e-05,
-      "loss": 0.2314,
+      "loss": 0.2452,
       "step": 910
     },
     {
       "epoch": 2.76,
       "learning_rate": 6.2012012012012e-05,
-      "loss": 0.2739,
+      "loss": 0.2648,
       "step": 920
     },
     {
       "epoch": 2.79,
       "learning_rate": 6.051051051051051e-05,
-      "loss": 0.2905,
+      "loss": 0.3084,
       "step": 930
     },
     {
       "epoch": 2.82,
       "learning_rate": 5.900900900900901e-05,
-      "loss": 0.2607,
+      "loss": 0.3583,
       "step": 940
     },
     {
       "epoch": 2.85,
       "learning_rate": 5.7507507507507506e-05,
-      "loss": 0.334,
+      "loss": 0.2682,
       "step": 950
     },
     {
       "epoch": 2.88,
       "learning_rate": 5.600600600600601e-05,
-      "loss": 0.3202,
+      "loss": 0.2135,
       "step": 960
     },
     {
       "epoch": 2.91,
       "learning_rate": 5.45045045045045e-05,
-      "loss": 0.3326,
+      "loss": 0.2129,
       "step": 970
     },
     {
       "epoch": 2.94,
       "learning_rate": 5.3003003003003005e-05,
-      "loss": 0.2717,
+      "loss": 0.3823,
       "step": 980
     },
     {
       "epoch": 2.97,
       "learning_rate": 5.15015015015015e-05,
-      "loss": 0.259,
+      "loss": 0.2763,
       "step": 990
     },
     {
       "epoch": 3.0,
       "learning_rate": 5e-05,
-      "loss": 0.2493,
+      "loss": 0.0894,
       "step": 1000
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.9394261424017003,
-      "eval_loss": 0.22659793496131897,
-      "eval_runtime": 33.5582,
-      "eval_samples_per_second": 28.041,
-      "eval_steps_per_second": 3.516,
+      "eval_accuracy": 0.924548352816153,
+      "eval_loss": 0.2870355546474457,
+      "eval_runtime": 38.3272,
+      "eval_samples_per_second": 24.552,
+      "eval_steps_per_second": 3.079,
       "step": 1000
     },
     {
       "epoch": 3.03,
       "learning_rate": 4.8498498498498504e-05,
-      "loss": 0.2026,
+      "loss": 0.1427,
       "step": 1010
     },
     {
       "epoch": 3.06,
       "learning_rate": 4.6996996996997e-05,
-      "loss": 0.189,
+      "loss": 0.3126,
       "step": 1020
     },
     {
       "epoch": 3.09,
       "learning_rate": 4.54954954954955e-05,
-      "loss": 0.1097,
+      "loss": 0.1022,
       "step": 1030
     },
     {
       "epoch": 3.12,
       "learning_rate": 4.3993993993994e-05,
-      "loss": 0.1605,
+      "loss": 0.2393,
       "step": 1040
     },
     {
       "epoch": 3.15,
       "learning_rate": 4.24924924924925e-05,
-      "loss": 0.2681,
+      "loss": 0.2127,
       "step": 1050
     },
     {
       "epoch": 3.18,
       "learning_rate": 4.099099099099099e-05,
-      "loss": 0.2872,
+      "loss": 0.0907,
       "step": 1060
     },
     {
       "epoch": 3.21,
       "learning_rate": 3.948948948948949e-05,
-      "loss": 0.3481,
+      "loss": 0.0853,
       "step": 1070
     },
     {
       "epoch": 3.24,
       "learning_rate": 3.7987987987987985e-05,
-      "loss": 0.1615,
+      "loss": 0.1306,
       "step": 1080
     },
     {
       "epoch": 3.27,
       "learning_rate": 3.648648648648649e-05,
-      "loss": 0.1447,
+      "loss": 0.1956,
       "step": 1090
     },
     {
       "epoch": 3.3,
       "learning_rate": 3.498498498498499e-05,
-      "loss": 0.2034,
+      "loss": 0.2092,
       "step": 1100
     },
     {
       "epoch": 3.3,
-      "eval_accuracy": 0.9479277364505845,
-      "eval_loss": 0.20114077627658844,
-      "eval_runtime": 34.4276,
-      "eval_samples_per_second": 27.333,
-      "eval_steps_per_second": 3.427,
+      "eval_accuracy": 0.9287991498405951,
+      "eval_loss": 0.2884349524974823,
+      "eval_runtime": 38.6836,
+      "eval_samples_per_second": 24.326,
+      "eval_steps_per_second": 3.05,
       "step": 1100
     },
     {
       "epoch": 3.33,
       "learning_rate": 3.3483483483483485e-05,
-      "loss": 0.204,
+      "loss": 0.1815,
       "step": 1110
     },
     {
       "epoch": 3.36,
       "learning_rate": 3.198198198198199e-05,
-      "loss": 0.18,
+      "loss": 0.0951,
       "step": 1120
     },
     {
       "epoch": 3.39,
       "learning_rate": 3.0480480480480482e-05,
-      "loss": 0.1657,
+      "loss": 0.1645,
       "step": 1130
     },
     {
       "epoch": 3.42,
       "learning_rate": 2.897897897897898e-05,
-      "loss": 0.1932,
+      "loss": 0.2262,
       "step": 1140
     },
     {
       "epoch": 3.45,
       "learning_rate": 2.7477477477477483e-05,
-      "loss": 0.1524,
+      "loss": 0.2508,
       "step": 1150
     },
     {
       "epoch": 3.48,
       "learning_rate": 2.5975975975975975e-05,
-      "loss": 0.1144,
+      "loss": 0.2116,
       "step": 1160
     },
     {
       "epoch": 3.51,
       "learning_rate": 2.4474474474474477e-05,
-      "loss": 0.1424,
+      "loss": 0.1312,
       "step": 1170
     },
     {
       "epoch": 3.54,
       "learning_rate": 2.2972972972972976e-05,
-      "loss": 0.2142,
+      "loss": 0.1955,
       "step": 1180
     },
     {
       "epoch": 3.57,
-      "learning_rate": 2.1621621621621624e-05,
-      "loss": 0.1448,
+      "learning_rate": 2.147147147147147e-05,
+      "loss": 0.1116,
       "step": 1190
     },
     {
       "epoch": 3.6,
-      "learning_rate": 2.012012012012012e-05,
-      "loss": 0.1753,
+      "learning_rate": 1.996996996996997e-05,
+      "loss": 0.1466,
       "step": 1200
     },
     {
       "epoch": 3.6,
-      "eval_accuracy": 0.944739638682253,
-      "eval_loss": 0.20892110466957092,
-      "eval_runtime": 33.5408,
-      "eval_samples_per_second": 28.055,
-      "eval_steps_per_second": 3.518,
+      "eval_accuracy": 0.9319872476089267,
+      "eval_loss": 0.2672683000564575,
+      "eval_runtime": 38.9895,
+      "eval_samples_per_second": 24.135,
+      "eval_steps_per_second": 3.026,
       "step": 1200
     },
     {
       "epoch": 3.63,
-      "learning_rate": 1.8618618618618618e-05,
-      "loss": 0.2085,
+      "learning_rate": 1.846846846846847e-05,
+      "loss": 0.1497,
       "step": 1210
     },
     {
       "epoch": 3.66,
-      "learning_rate": 1.7117117117117117e-05,
-      "loss": 0.2179,
+      "learning_rate": 1.6966966966966967e-05,
+      "loss": 0.0741,
       "step": 1220
     },
     {
       "epoch": 3.69,
-      "learning_rate": 1.5615615615615616e-05,
-      "loss": 0.1755,
+      "learning_rate": 1.5465465465465466e-05,
+      "loss": 0.2916,
       "step": 1230
     },
     {
       "epoch": 3.72,
-      "learning_rate": 1.4114114114114116e-05,
-      "loss": 0.1905,
+      "learning_rate": 1.3963963963963963e-05,
+      "loss": 0.1485,
       "step": 1240
     },
     {
       "epoch": 3.75,
-      "learning_rate": 1.2612612612612611e-05,
-      "loss": 0.2246,
+      "learning_rate": 1.2462462462462463e-05,
+      "loss": 0.1156,
       "step": 1250
     },
     {
       "epoch": 3.78,
-      "learning_rate": 1.1111111111111112e-05,
-      "loss": 0.1495,
+      "learning_rate": 1.0960960960960962e-05,
+      "loss": 0.0887,
       "step": 1260
     },
     {
       "epoch": 3.81,
-      "learning_rate": 9.60960960960961e-06,
-      "loss": 0.2038,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 0.153,
       "step": 1270
     },
     {
       "epoch": 3.84,
-      "learning_rate": 8.108108108108109e-06,
-      "loss": 0.2163,
+      "learning_rate": 7.957957957957958e-06,
+      "loss": 0.2049,
       "step": 1280
     },
     {
       "epoch": 3.87,
-      "learning_rate": 6.606606606606607e-06,
-      "loss": 0.152,
+      "learning_rate": 6.456456456456457e-06,
+      "loss": 0.099,
       "step": 1290
     },
     {
       "epoch": 3.9,
-      "learning_rate": 5.105105105105106e-06,
-      "loss": 0.1614,
+      "learning_rate": 4.954954954954955e-06,
+      "loss": 0.1789,
       "step": 1300
     },
     {
       "epoch": 3.9,
-      "eval_accuracy": 0.9543039319872476,
-      "eval_loss": 0.19176600873470306,
-      "eval_runtime": 33.687,
-      "eval_samples_per_second": 27.934,
-      "eval_steps_per_second": 3.503,
+      "eval_accuracy": 0.9330499468650372,
+      "eval_loss": 0.263217031955719,
+      "eval_runtime": 37.6874,
+      "eval_samples_per_second": 24.969,
+      "eval_steps_per_second": 3.131,
       "step": 1300
     },
     {
       "epoch": 3.93,
-      "learning_rate": 3.603603603603604e-06,
-      "loss": 0.1276,
+      "learning_rate": 3.4534534534534537e-06,
+      "loss": 0.2831,
       "step": 1310
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.102102102102102e-06,
-      "loss": 0.2218,
+      "learning_rate": 1.951951951951952e-06,
+      "loss": 0.2103,
       "step": 1320
     },
     {
       "epoch": 3.99,
-      "learning_rate": 6.006006006006006e-07,
-      "loss": 0.1194,
+      "learning_rate": 4.504504504504505e-07,
+      "loss": 0.1034,
       "step": 1330
     },
     {
       "epoch": 4.0,
       "step": 1332,
       "total_flos": 1.651775717862015e+18,
-      "train_loss": 0.4876434194433081,
-      "train_runtime": 1561.5053,
-      "train_samples_per_second": 13.648,
-      "train_steps_per_second": 0.853
+      "train_loss": 0.4671340096104252,
+      "train_runtime": 1617.2249,
+      "train_samples_per_second": 13.178,
+      "train_steps_per_second": 0.824
     }
   ],
   "max_steps": 1332,