End of training

README.md

@@ -3,6 +3,8 @@ library_name: transformers
 license: apache-2.0
 base_model: timm/resnet18.a1_in1k
 tags:
+- image-classification
+- vision
 - generated_from_trainer
 metrics:
 - accuracy
@@ -16,7 +18,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 # vit-base-beans
 
-This model is a fine-tuned version of [timm/resnet18.a1_in1k](https://huggingface.co/timm/resnet18.a1_in1k) on an unknown dataset.
+This model is a fine-tuned version of [timm/resnet18.a1_in1k](https://huggingface.co/timm/resnet18.a1_in1k) on the beans dataset.
 It achieves the following results on the evaluation set:
 - Loss: 0.7412
 - Accuracy: 0.8120

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.3308270676691729,
-    "eval_loss": 1.1282602548599243,
-    "eval_runtime": 0.9945,
-    "eval_samples_per_second": 133.736,
-    "eval_steps_per_second": 17.094,
+    "eval_accuracy": 0.8120300751879699,
+    "eval_loss": 0.7412300109863281,
+    "eval_runtime": 0.7631,
+    "eval_samples_per_second": 174.289,
+    "eval_steps_per_second": 22.278,
     "total_flos": 1.5658365504595968e+17,
-    "train_loss": 0.0,
-    "train_runtime": 1.3792,
-    "train_samples_per_second": 11245.899,
-    "train_steps_per_second": 1413.894
+    "train_loss": 0.9236146088135548,
+    "train_runtime": 142.953,
+    "train_samples_per_second": 108.497,
+    "train_steps_per_second": 13.641
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
-    "eval_accuracy": 0.3308270676691729,
-    "eval_loss": 1.1282602548599243,
-    "eval_runtime": 0.9945,
-    "eval_samples_per_second": 133.736,
-    "eval_steps_per_second": 17.094
+    "eval_accuracy": 0.8120300751879699,
+    "eval_loss": 0.7412300109863281,
+    "eval_runtime": 0.7631,
+    "eval_samples_per_second": 174.289,
+    "eval_steps_per_second": 22.278
 }

runs/Nov19_22-41-36_ip-10-90-1-182/events.out.tfevents.1732056245.ip-10-90-1-182.3353433.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:52a138c5c6c4abe6b692847b5659e465f472b55c50418466ebd24a00a9f14a8d
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 15.0,
     "total_flos": 1.5658365504595968e+17,
-    "train_loss": 0.0,
-    "train_runtime": 1.3792,
-    "train_samples_per_second": 11245.899,
-    "train_steps_per_second": 1413.894
+    "train_loss": 0.9236146088135548,
+    "train_runtime": 142.953,
+    "train_samples_per_second": 108.497,
+    "train_steps_per_second": 13.641
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.738013505935669,
+  "best_metric": 0.7412300109863281,
   "best_model_checkpoint": "./beans_outputs/checkpoint-1950",
   "epoch": 15.0,
   "eval_steps": 500,
@@ -10,1512 +10,1512 @@
   "log_history": [
     {
       "epoch": 0.07692307692307693,
-      "grad_norm": 2.099634885787964,
+      "grad_norm": 2.136049747467041,
       "learning_rate": 1.98974358974359e-05,
       "loss": 1.1239,
       "step": 10
     },
     {
       "epoch": 0.15384615384615385,
-      "grad_norm": 1.7766097784042358,
+      "grad_norm": 1.8187670707702637,
       "learning_rate": 1.9794871794871798e-05,
       "loss": 1.1221,
       "step": 20
     },
     {
       "epoch": 0.23076923076923078,
-      "grad_norm": 1.9635370969772339,
+      "grad_norm": 2.0219993591308594,
       "learning_rate": 1.9692307692307696e-05,
       "loss": 1.1164,
       "step": 30
     },
     {
       "epoch": 0.3076923076923077,
-      "grad_norm": 2.459880828857422,
+      "grad_norm": 2.4619803428649902,
       "learning_rate": 1.958974358974359e-05,
-      "loss": 1.1045,
+      "loss": 1.1044,
       "step": 40
     },
     {
       "epoch": 0.38461538461538464,
-      "grad_norm": 1.7044235467910767,
+      "grad_norm": 1.6733014583587646,
       "learning_rate": 1.9487179487179488e-05,
-      "loss": 1.1083,
+      "loss": 1.1082,
       "step": 50
     },
     {
       "epoch": 0.46153846153846156,
-      "grad_norm": 1.4914859533309937,
+      "grad_norm": 1.4969494342803955,
       "learning_rate": 1.9384615384615386e-05,
-      "loss": 1.1044,
+      "loss": 1.1043,
       "step": 60
     },
     {
       "epoch": 0.5384615384615384,
-      "grad_norm": 2.6082077026367188,
+      "grad_norm": 2.6347556114196777,
       "learning_rate": 1.9282051282051284e-05,
-      "loss": 1.1027,
+      "loss": 1.1028,
       "step": 70
     },
     {
       "epoch": 0.6153846153846154,
-      "grad_norm": 2.722012996673584,
+      "grad_norm": 2.5843420028686523,
       "learning_rate": 1.9179487179487182e-05,
       "loss": 1.0908,
       "step": 80
     },
     {
       "epoch": 0.6923076923076923,
-      "grad_norm": 2.029851198196411,
+      "grad_norm": 2.0522916316986084,
       "learning_rate": 1.907692307692308e-05,
-      "loss": 1.0941,
+      "loss": 1.094,
       "step": 90
     },
     {
       "epoch": 0.7692307692307693,
-      "grad_norm": 2.4984819889068604,
+      "grad_norm": 2.4885082244873047,
       "learning_rate": 1.8974358974358975e-05,
-      "loss": 1.0913,
+      "loss": 1.0912,
       "step": 100
     },
     {
       "epoch": 0.8461538461538461,
-      "grad_norm": 1.7019662857055664,
+      "grad_norm": 1.7014166116714478,
       "learning_rate": 1.8871794871794873e-05,
-      "loss": 1.0945,
+      "loss": 1.0949,
       "step": 110
     },
     {
       "epoch": 0.9230769230769231,
-      "grad_norm": 2.2935667037963867,
+      "grad_norm": 2.294283866882324,
       "learning_rate": 1.876923076923077e-05,
-      "loss": 1.0994,
+      "loss": 1.0992,
       "step": 120
     },
     {
       "epoch": 1.0,
-      "grad_norm": 3.9286272525787354,
+      "grad_norm": 4.129885673522949,
       "learning_rate": 1.866666666666667e-05,
-      "loss": 1.0861,
+      "loss": 1.0864,
       "step": 130
     },
     {
       "epoch": 1.0,
       "eval_accuracy": 0.42857142857142855,
-      "eval_loss": 1.0880533456802368,
-      "eval_runtime": 0.7946,
-      "eval_samples_per_second": 167.381,
-      "eval_steps_per_second": 21.395,
+      "eval_loss": 1.0877832174301147,
+      "eval_runtime": 0.7833,
+      "eval_samples_per_second": 169.792,
+      "eval_steps_per_second": 21.703,
       "step": 130
     },
     {
       "epoch": 1.0769230769230769,
-      "grad_norm": 2.083320140838623,
+      "grad_norm": 2.331717014312744,
       "learning_rate": 1.8564102564102567e-05,
-      "loss": 1.0772,
+      "loss": 1.0774,
       "step": 140
     },
     {
       "epoch": 1.1538461538461537,
-      "grad_norm": 2.711829423904419,
+      "grad_norm": 2.5262138843536377,
       "learning_rate": 1.8461538461538465e-05,
-      "loss": 1.072,
+      "loss": 1.0719,
       "step": 150
     },
     {
       "epoch": 1.2307692307692308,
-      "grad_norm": 1.5575768947601318,
+      "grad_norm": 1.5971320867538452,
       "learning_rate": 1.835897435897436e-05,
-      "loss": 1.0787,
+      "loss": 1.0781,
       "step": 160
     },
     {
       "epoch": 1.3076923076923077,
-      "grad_norm": 2.3275976181030273,
+      "grad_norm": 2.383288860321045,
       "learning_rate": 1.8256410256410257e-05,
-      "loss": 1.0935,
+      "loss": 1.0929,
       "step": 170
     },
     {
       "epoch": 1.3846153846153846,
-      "grad_norm": 2.1124396324157715,
+      "grad_norm": 2.169706106185913,
       "learning_rate": 1.8153846153846155e-05,
-      "loss": 1.0806,
+      "loss": 1.0805,
       "step": 180
     },
     {
       "epoch": 1.4615384615384617,
-      "grad_norm": 2.1334290504455566,
+      "grad_norm": 2.1174418926239014,
       "learning_rate": 1.8051282051282053e-05,
-      "loss": 1.0796,
+      "loss": 1.08,
       "step": 190
     },
     {
       "epoch": 1.5384615384615383,
-      "grad_norm": 1.7488656044006348,
+      "grad_norm": 1.7236179113388062,
       "learning_rate": 1.794871794871795e-05,
-      "loss": 1.0762,
+      "loss": 1.0766,
       "step": 200
     },
     {
       "epoch": 1.6153846153846154,
-      "grad_norm": 1.7498658895492554,
+      "grad_norm": 1.7772722244262695,
       "learning_rate": 1.784615384615385e-05,
-      "loss": 1.0682,
+      "loss": 1.0676,
       "step": 210
     },
     {
       "epoch": 1.6923076923076923,
-      "grad_norm": 3.5284996032714844,
+      "grad_norm": 3.53834867477417,
       "learning_rate": 1.7743589743589744e-05,
-      "loss": 1.0699,
+      "loss": 1.0695,
       "step": 220
     },
     {
       "epoch": 1.7692307692307692,
-      "grad_norm": 2.088862895965576,
+      "grad_norm": 2.0417070388793945,
       "learning_rate": 1.7641025641025642e-05,
-      "loss": 1.0707,
+      "loss": 1.0706,
       "step": 230
     },
     {
       "epoch": 1.8461538461538463,
-      "grad_norm": 1.9373514652252197,
+      "grad_norm": 1.9734611511230469,
       "learning_rate": 1.753846153846154e-05,
-      "loss": 1.0864,
+      "loss": 1.0863,
       "step": 240
     },
     {
       "epoch": 1.9230769230769231,
-      "grad_norm": 2.00813889503479,
+      "grad_norm": 1.9997600317001343,
       "learning_rate": 1.7435897435897438e-05,
-      "loss": 1.0681,
+      "loss": 1.068,
       "step": 250
     },
     {
       "epoch": 2.0,
-      "grad_norm": 3.0624523162841797,
+      "grad_norm": 3.0024373531341553,
       "learning_rate": 1.7333333333333336e-05,
-      "loss": 1.0631,
+      "loss": 1.0629,
       "step": 260
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.5413533834586466,
-      "eval_loss": 1.0597343444824219,
-      "eval_runtime": 0.7394,
-      "eval_samples_per_second": 179.866,
-      "eval_steps_per_second": 22.99,
+      "eval_accuracy": 0.5488721804511278,
+      "eval_loss": 1.0593525171279907,
+      "eval_runtime": 0.7442,
+      "eval_samples_per_second": 178.706,
+      "eval_steps_per_second": 22.842,
       "step": 260
     },
     {
       "epoch": 2.076923076923077,
-      "grad_norm": 2.6268935203552246,
+      "grad_norm": 1.977807641029358,
       "learning_rate": 1.7230769230769234e-05,
-      "loss": 1.0716,
+      "loss": 1.0711,
       "step": 270
     },
     {
       "epoch": 2.1538461538461537,
-      "grad_norm": 2.3874154090881348,
+      "grad_norm": 2.3906102180480957,
       "learning_rate": 1.7128205128205128e-05,
-      "loss": 1.0599,
+      "loss": 1.0597,
       "step": 280
     },
     {
       "epoch": 2.230769230769231,
-      "grad_norm": 2.3503990173339844,
+      "grad_norm": 2.3670897483825684,
       "learning_rate": 1.7025641025641026e-05,
-      "loss": 1.054,
+      "loss": 1.0576,
       "step": 290
     },
     {
       "epoch": 2.3076923076923075,
-      "grad_norm": 2.3312108516693115,
+      "grad_norm": 3.026155948638916,
       "learning_rate": 1.6923076923076924e-05,
-      "loss": 1.0436,
+      "loss": 1.0434,
       "step": 300
     },
     {
       "epoch": 2.3846153846153846,
-      "grad_norm": 1.9090198278427124,
+      "grad_norm": 1.9082350730895996,
       "learning_rate": 1.6820512820512822e-05,
-      "loss": 1.0569,
+      "loss": 1.0566,
       "step": 310
     },
     {
       "epoch": 2.4615384615384617,
-      "grad_norm": 1.9505618810653687,
+      "grad_norm": 1.9187153577804565,
       "learning_rate": 1.671794871794872e-05,
-      "loss": 1.0479,
+      "loss": 1.0476,
       "step": 320
     },
     {
       "epoch": 2.5384615384615383,
-      "grad_norm": 1.4548966884613037,
+      "grad_norm": 1.4435549974441528,
       "learning_rate": 1.6615384615384618e-05,
-      "loss": 1.0304,
+      "loss": 1.032,
       "step": 330
     },
     {
       "epoch": 2.6153846153846154,
-      "grad_norm": 2.1505134105682373,
+      "grad_norm": 2.1457245349884033,
       "learning_rate": 1.6512820512820513e-05,
-      "loss": 1.0467,
+      "loss": 1.0475,
       "step": 340
     },
     {
       "epoch": 2.6923076923076925,
-      "grad_norm": 2.8420169353485107,
+      "grad_norm": 1.9391709566116333,
       "learning_rate": 1.641025641025641e-05,
-      "loss": 1.0509,
+      "loss": 1.0486,
       "step": 350
     },
     {
       "epoch": 2.769230769230769,
-      "grad_norm": 1.8315626382827759,
+      "grad_norm": 1.8148127794265747,
       "learning_rate": 1.630769230769231e-05,
       "loss": 1.0407,
       "step": 360
     },
     {
       "epoch": 2.8461538461538463,
-      "grad_norm": 1.8499083518981934,
+      "grad_norm": 2.444157123565674,
       "learning_rate": 1.6205128205128207e-05,
-      "loss": 1.0344,
+      "loss": 1.0356,
       "step": 370
     },
     {
       "epoch": 2.9230769230769234,
-      "grad_norm": 1.9010175466537476,
+      "grad_norm": 1.9061695337295532,
       "learning_rate": 1.6102564102564105e-05,
-      "loss": 1.0219,
+      "loss": 1.0239,
       "step": 380
     },
     {
       "epoch": 3.0,
-      "grad_norm": 4.809950828552246,
+      "grad_norm": 4.859686851501465,
       "learning_rate": 1.6000000000000003e-05,
-      "loss": 1.0443,
+      "loss": 1.0434,
       "step": 390
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.6691729323308271,
-      "eval_loss": 1.0225275754928589,
-      "eval_runtime": 0.7405,
-      "eval_samples_per_second": 179.608,
-      "eval_steps_per_second": 22.957,
+      "eval_accuracy": 0.6766917293233082,
+      "eval_loss": 1.0230107307434082,
+      "eval_runtime": 0.7471,
+      "eval_samples_per_second": 178.027,
+      "eval_steps_per_second": 22.755,
       "step": 390
     },
     {
       "epoch": 3.076923076923077,
-      "grad_norm": 2.195730686187744,
+      "grad_norm": 2.2021689414978027,
       "learning_rate": 1.5897435897435897e-05,
-      "loss": 1.0414,
+      "loss": 1.0424,
       "step": 400
     },
     {
       "epoch": 3.1538461538461537,
-      "grad_norm": 2.1667442321777344,
+      "grad_norm": 1.8670283555984497,
       "learning_rate": 1.5794871794871795e-05,
-      "loss": 1.0302,
+      "loss": 1.0299,
       "step": 410
     },
     {
       "epoch": 3.230769230769231,
-      "grad_norm": 2.149664878845215,
+      "grad_norm": 2.193986654281616,
       "learning_rate": 1.5692307692307693e-05,
-      "loss": 1.0365,
+      "loss": 1.0369,
       "step": 420
     },
     {
       "epoch": 3.3076923076923075,
-      "grad_norm": 2.2559187412261963,
+      "grad_norm": 2.26470685005188,
       "learning_rate": 1.558974358974359e-05,
       "loss": 1.0159,
       "step": 430
     },
     {
       "epoch": 3.3846153846153846,
-      "grad_norm": 2.165260076522827,
+      "grad_norm": 2.18507981300354,
       "learning_rate": 1.548717948717949e-05,
-      "loss": 1.0274,
+      "loss": 1.0282,
       "step": 440
     },
     {
       "epoch": 3.4615384615384617,
-      "grad_norm": 1.799578309059143,
+      "grad_norm": 1.8047341108322144,
       "learning_rate": 1.5384615384615387e-05,
-      "loss": 1.0371,
+      "loss": 1.0381,
       "step": 450
     },
     {
       "epoch": 3.5384615384615383,
-      "grad_norm": 2.441862106323242,
+      "grad_norm": 2.6463704109191895,
       "learning_rate": 1.5282051282051282e-05,
       "loss": 1.0322,
       "step": 460
     },
     {
       "epoch": 3.6153846153846154,
-      "grad_norm": 1.571476936340332,
+      "grad_norm": 1.6456215381622314,
       "learning_rate": 1.517948717948718e-05,
-      "loss": 1.0046,
+      "loss": 1.0049,
       "step": 470
     },
     {
       "epoch": 3.6923076923076925,
-      "grad_norm": 2.725184440612793,
+      "grad_norm": 2.774256706237793,
       "learning_rate": 1.5076923076923078e-05,
-      "loss": 1.0081,
+      "loss": 1.0091,
       "step": 480
     },
     {
       "epoch": 3.769230769230769,
-      "grad_norm": 1.5710386037826538,
+      "grad_norm": 1.572251319885254,
       "learning_rate": 1.4974358974358976e-05,
-      "loss": 0.9977,
+      "loss": 0.998,
       "step": 490
     },
     {
       "epoch": 3.8461538461538463,
-      "grad_norm": 1.6186603307724,
+      "grad_norm": 1.5640805959701538,
       "learning_rate": 1.4871794871794874e-05,
-      "loss": 1.0217,
+      "loss": 1.0222,
       "step": 500
     },
     {
       "epoch": 3.9230769230769234,
-      "grad_norm": 1.8194940090179443,
+      "grad_norm": 1.9231537580490112,
       "learning_rate": 1.4769230769230772e-05,
-      "loss": 0.997,
+      "loss": 0.9979,
       "step": 510
     },
     {
       "epoch": 4.0,
-      "grad_norm": 5.478014945983887,
+      "grad_norm": 5.481942176818848,
       "learning_rate": 1.4666666666666666e-05,
-      "loss": 1.0218,
+      "loss": 1.0214,
       "step": 520
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.6842105263157895,
-      "eval_loss": 0.9959980845451355,
-      "eval_runtime": 0.7638,
-      "eval_samples_per_second": 174.124,
-      "eval_steps_per_second": 22.256,
+      "eval_accuracy": 0.6766917293233082,
+      "eval_loss": 0.9964542388916016,
+      "eval_runtime": 0.7616,
+      "eval_samples_per_second": 174.638,
+      "eval_steps_per_second": 22.322,
       "step": 520
     },
     {
       "epoch": 4.076923076923077,
-      "grad_norm": 2.7807815074920654,
+      "grad_norm": 2.7514402866363525,
       "learning_rate": 1.4564102564102564e-05,
-      "loss": 1.0126,
+      "loss": 1.0128,
       "step": 530
     },
     {
       "epoch": 4.153846153846154,
-      "grad_norm": 1.8810123205184937,
+      "grad_norm": 1.8411396741867065,
       "learning_rate": 1.4461538461538462e-05,
-      "loss": 1.0142,
+      "loss": 1.0145,
       "step": 540
     },
     {
       "epoch": 4.230769230769231,
-      "grad_norm": 2.737741708755493,
+      "grad_norm": 2.670154571533203,
       "learning_rate": 1.435897435897436e-05,
-      "loss": 1.0241,
+      "loss": 1.0227,
       "step": 550
     },
     {
       "epoch": 4.3076923076923075,
-      "grad_norm": 2.183462381362915,
+      "grad_norm": 2.1951498985290527,
       "learning_rate": 1.4256410256410258e-05,
-      "loss": 1.0324,
+      "loss": 1.0321,
       "step": 560
     },
     {
       "epoch": 4.384615384615385,
-      "grad_norm": 1.9445362091064453,
+      "grad_norm": 1.9692825078964233,
       "learning_rate": 1.4153846153846156e-05,
-      "loss": 0.9823,
+      "loss": 0.9829,
       "step": 570
     },
     {
       "epoch": 4.461538461538462,
-      "grad_norm": 2.8362057209014893,
+      "grad_norm": 2.611340284347534,
       "learning_rate": 1.405128205128205e-05,
-      "loss": 0.9919,
+      "loss": 0.9918,
       "step": 580
     },
     {
       "epoch": 4.538461538461538,
-      "grad_norm": 2.5588510036468506,
+      "grad_norm": 2.4288899898529053,
       "learning_rate": 1.3948717948717949e-05,
-      "loss": 0.987,
+      "loss": 0.9879,
       "step": 590
     },
     {
       "epoch": 4.615384615384615,
-      "grad_norm": 1.740641474723816,
+      "grad_norm": 1.7537823915481567,
       "learning_rate": 1.3846153846153847e-05,
-      "loss": 0.9778,
+      "loss": 0.9793,
       "step": 600
     },
     {
       "epoch": 4.6923076923076925,
-      "grad_norm": 3.247788429260254,
+      "grad_norm": 3.1588003635406494,
       "learning_rate": 1.3743589743589745e-05,
-      "loss": 0.9978,
+      "loss": 1.0002,
       "step": 610
     },
     {
       "epoch": 4.769230769230769,
-      "grad_norm": 2.136213779449463,
+      "grad_norm": 2.2472622394561768,
       "learning_rate": 1.3641025641025643e-05,
-      "loss": 1.0067,
+      "loss": 1.0094,
       "step": 620
     },
     {
       "epoch": 4.846153846153846,
-      "grad_norm": 1.8435287475585938,
+      "grad_norm": 1.7958937883377075,
       "learning_rate": 1.353846153846154e-05,
-      "loss": 0.9691,
+      "loss": 0.9703,
       "step": 630
     },
     {
       "epoch": 4.923076923076923,
-      "grad_norm": 2.3936328887939453,
+      "grad_norm": 2.415766477584839,
       "learning_rate": 1.3435897435897435e-05,
-      "loss": 0.9692,
+      "loss": 0.9703,
       "step": 640
     },
     {
       "epoch": 5.0,
-      "grad_norm": 4.921462535858154,
+      "grad_norm": 4.948933124542236,
       "learning_rate": 1.3333333333333333e-05,
-      "loss": 1.0028,
+      "loss": 1.0026,
       "step": 650
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.7293233082706767,
-      "eval_loss": 0.9568173289299011,
-      "eval_runtime": 0.7715,
-      "eval_samples_per_second": 172.394,
-      "eval_steps_per_second": 22.035,
+      "eval_accuracy": 0.7443609022556391,
+      "eval_loss": 0.9569369554519653,
+      "eval_runtime": 0.7647,
+      "eval_samples_per_second": 173.928,
+      "eval_steps_per_second": 22.231,
       "step": 650
     },
     {
       "epoch": 5.076923076923077,
-      "grad_norm": 2.051286458969116,
+      "grad_norm": 2.1397032737731934,
       "learning_rate": 1.3230769230769231e-05,
-      "loss": 0.9638,
+      "loss": 0.9645,
       "step": 660
     },
     {
       "epoch": 5.153846153846154,
-      "grad_norm": 2.7295546531677246,
+      "grad_norm": 2.7277321815490723,
       "learning_rate": 1.312820512820513e-05,
-      "loss": 1.0056,
+      "loss": 1.0063,
       "step": 670
     },
     {
       "epoch": 5.230769230769231,
-      "grad_norm": 2.3378870487213135,
+      "grad_norm": 2.391350030899048,
       "learning_rate": 1.3025641025641027e-05,
-      "loss": 0.9916,
+      "loss": 0.9918,
       "step": 680
     },
     {
       "epoch": 5.3076923076923075,
-      "grad_norm": 2.754603862762451,
+      "grad_norm": 2.751174211502075,
       "learning_rate": 1.2923076923076925e-05,
-      "loss": 0.9869,
+      "loss": 0.9849,
       "step": 690
     },
     {
       "epoch": 5.384615384615385,
-      "grad_norm": 2.6367554664611816,
+      "grad_norm": 2.77424693107605,
       "learning_rate": 1.2820512820512823e-05,
-      "loss": 0.9743,
+      "loss": 0.9745,
       "step": 700
     },
     {
       "epoch": 5.461538461538462,
-      "grad_norm": 2.152855396270752,
+      "grad_norm": 1.9156702756881714,
       "learning_rate": 1.2717948717948718e-05,
-      "loss": 0.9669,
+      "loss": 0.9684,
       "step": 710
     },
     {
       "epoch": 5.538461538461538,
-      "grad_norm": 1.970173954963684,
+      "grad_norm": 1.9521454572677612,
       "learning_rate": 1.2615384615384616e-05,
-      "loss": 0.9496,
+      "loss": 0.9503,
       "step": 720
     },
     {
       "epoch": 5.615384615384615,
-      "grad_norm": 2.49542498588562,
+      "grad_norm": 2.468419313430786,
       "learning_rate": 1.2512820512820514e-05,
-      "loss": 0.9644,
+      "loss": 0.9641,
       "step": 730
     },
     {
       "epoch": 5.6923076923076925,
-      "grad_norm": 2.5210750102996826,
+      "grad_norm": 2.520923614501953,
       "learning_rate": 1.2410256410256412e-05,
-      "loss": 0.9453,
+      "loss": 0.9471,
       "step": 740
     },
     {
       "epoch": 5.769230769230769,
-      "grad_norm": 2.0949254035949707,
+      "grad_norm": 2.1003758907318115,
       "learning_rate": 1.230769230769231e-05,
-      "loss": 0.95,
+      "loss": 0.9513,
       "step": 750
     },
     {
       "epoch": 5.846153846153846,
-      "grad_norm": 2.167081594467163,
+      "grad_norm": 2.192279100418091,
       "learning_rate": 1.2205128205128208e-05,
-      "loss": 0.9519,
+      "loss": 0.9527,
       "step": 760
     },
     {
       "epoch": 5.923076923076923,
-      "grad_norm": 3.6734519004821777,
+      "grad_norm": 3.8428618907928467,
       "learning_rate": 1.2102564102564102e-05,
-      "loss": 0.9376,
+      "loss": 0.938,
       "step": 770
     },
     {
       "epoch": 6.0,
-      "grad_norm": 4.938449859619141,
+      "grad_norm": 4.9151530265808105,
       "learning_rate": 1.2e-05,
-      "loss": 0.9752,
+      "loss": 0.9753,
       "step": 780
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.7669172932330827,
-      "eval_loss": 0.9278604984283447,
-      "eval_runtime": 0.7739,
-      "eval_samples_per_second": 171.864,
-      "eval_steps_per_second": 21.968,
+      "eval_accuracy": 0.7819548872180451,
+      "eval_loss": 0.9288201332092285,
+      "eval_runtime": 0.7499,
+      "eval_samples_per_second": 177.349,
+      "eval_steps_per_second": 22.669,
       "step": 780
     },
     {
       "epoch": 6.076923076923077,
-      "grad_norm": 2.8171768188476562,
+      "grad_norm": 2.7967398166656494,
       "learning_rate": 1.1897435897435898e-05,
-      "loss": 0.9423,
+      "loss": 0.9428,
       "step": 790
     },
     {
       "epoch": 6.153846153846154,
-      "grad_norm": 2.5053982734680176,
+      "grad_norm": 2.5342345237731934,
       "learning_rate": 1.1794871794871796e-05,
-      "loss": 0.9382,
+      "loss": 0.9406,
       "step": 800
     },
     {
       "epoch": 6.230769230769231,
-      "grad_norm": 1.875543475151062,
+      "grad_norm": 1.877543330192566,
       "learning_rate": 1.1692307692307694e-05,
-      "loss": 0.9318,
+      "loss": 0.9319,
       "step": 810
     },
     {
       "epoch": 6.3076923076923075,
-      "grad_norm": 2.222604513168335,
+      "grad_norm": 2.4524621963500977,
       "learning_rate": 1.1589743589743592e-05,
-      "loss": 0.9313,
+      "loss": 0.9332,
       "step": 820
     },
     {
       "epoch": 6.384615384615385,
-      "grad_norm": 2.5009353160858154,
+      "grad_norm": 2.4967362880706787,
       "learning_rate": 1.1487179487179487e-05,
-      "loss": 0.9354,
+      "loss": 0.9367,
       "step": 830
     },
     {
       "epoch": 6.461538461538462,
-      "grad_norm": 3.2472548484802246,
+      "grad_norm": 3.2078776359558105,
       "learning_rate": 1.1384615384615385e-05,
-      "loss": 0.9335,
+      "loss": 0.9339,
       "step": 840
     },
     {
       "epoch": 6.538461538461538,
-      "grad_norm": 2.944819927215576,
+      "grad_norm": 2.926706075668335,
       "learning_rate": 1.1282051282051283e-05,
-      "loss": 0.9403,
+      "loss": 0.9416,
       "step": 850
     },
     {
       "epoch": 6.615384615384615,
-      "grad_norm": 1.9217369556427002,
+      "grad_norm": 1.8625017404556274,
       "learning_rate": 1.117948717948718e-05,
-      "loss": 0.9109,
+      "loss": 0.9111,
       "step": 860
     },
     {
       "epoch": 6.6923076923076925,
-      "grad_norm": 2.2056405544281006,
+      "grad_norm": 2.7141189575195312,
       "learning_rate": 1.1076923076923079e-05,
-      "loss": 0.9578,
+      "loss": 0.9574,
       "step": 870
     },
     {
       "epoch": 6.769230769230769,
-      "grad_norm": 2.6266028881073,
+      "grad_norm": 2.307347536087036,
       "learning_rate": 1.0974358974358977e-05,
-      "loss": 0.9249,
+      "loss": 0.9259,
       "step": 880
     },
     {
       "epoch": 6.846153846153846,
-      "grad_norm": 2.48388409614563,
+      "grad_norm": 2.3937132358551025,
       "learning_rate": 1.0871794871794871e-05,
-      "loss": 0.9198,
+      "loss": 0.9207,
       "step": 890
     },
     {
       "epoch": 6.923076923076923,
-      "grad_norm": 3.0685908794403076,
+      "grad_norm": 3.0794668197631836,
       "learning_rate": 1.076923076923077e-05,
-      "loss": 0.9424,
+      "loss": 0.9418,
       "step": 900
     },
     {
       "epoch": 7.0,
-      "grad_norm": 4.163443088531494,
+      "grad_norm": 4.111669063568115,
       "learning_rate": 1.0666666666666667e-05,
-      "loss": 0.924,
+      "loss": 0.9252,
       "step": 910
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.8045112781954887,
-      "eval_loss": 0.8851932287216187,
-      "eval_runtime": 0.7436,
-      "eval_samples_per_second": 178.849,
-      "eval_steps_per_second": 22.86,
+      "eval_accuracy": 0.7969924812030075,
+      "eval_loss": 0.8874692916870117,
+      "eval_runtime": 0.7823,
+      "eval_samples_per_second": 170.013,
+      "eval_steps_per_second": 21.731,
       "step": 910
     },
     {
       "epoch": 7.076923076923077,
-      "grad_norm": 2.746605634689331,
+      "grad_norm": 2.7561662197113037,
       "learning_rate": 1.0564102564102565e-05,
-      "loss": 0.9102,
+      "loss": 0.911,
       "step": 920
     },
     {
       "epoch": 7.153846153846154,
-      "grad_norm": 3.5535519123077393,
+      "grad_norm": 3.2020223140716553,
       "learning_rate": 1.0461538461538463e-05,
-      "loss": 0.9097,
+      "loss": 0.912,
       "step": 930
     },
     {
       "epoch": 7.230769230769231,
-      "grad_norm": 3.44745135307312,
+      "grad_norm": 3.459304094314575,
       "learning_rate": 1.0358974358974361e-05,
-      "loss": 0.899,
+      "loss": 0.8994,
       "step": 940
     },
     {
       "epoch": 7.3076923076923075,
-      "grad_norm": 2.7519822120666504,
+      "grad_norm": 2.774078369140625,
       "learning_rate": 1.0256410256410256e-05,
-      "loss": 0.9072,
+      "loss": 0.9079,
       "step": 950
     },
     {
       "epoch": 7.384615384615385,
-      "grad_norm": 2.7004168033599854,
+      "grad_norm": 2.7169668674468994,
       "learning_rate": 1.0153846153846154e-05,
-      "loss": 0.9231,
+      "loss": 0.9256,
       "step": 960
     },
     {
       "epoch": 7.461538461538462,
-      "grad_norm": 2.188138961791992,
+      "grad_norm": 2.171323299407959,
       "learning_rate": 1.0051282051282052e-05,
-      "loss": 0.8897,
+      "loss": 0.8898,
       "step": 970
     },
     {
       "epoch": 7.538461538461538,
-      "grad_norm": 2.8544998168945312,
+      "grad_norm": 2.7350351810455322,
       "learning_rate": 9.94871794871795e-06,
-      "loss": 0.9233,
+      "loss": 0.9243,
       "step": 980
     },
     {
       "epoch": 7.615384615384615,
-      "grad_norm": 2.217107057571411,
+      "grad_norm": 2.3926539421081543,
       "learning_rate": 9.846153846153848e-06,
-      "loss": 0.8854,
+      "loss": 0.8868,
       "step": 990
     },
     {
       "epoch": 7.6923076923076925,
-      "grad_norm": 2.083819627761841,
+      "grad_norm": 2.0602715015411377,
       "learning_rate": 9.743589743589744e-06,
-      "loss": 0.8808,
+      "loss": 0.8837,
       "step": 1000
     },
     {
       "epoch": 7.769230769230769,
-      "grad_norm": 2.864044427871704,
+      "grad_norm": 2.885303497314453,
       "learning_rate": 9.641025641025642e-06,
-      "loss": 0.8818,
+      "loss": 0.8827,
       "step": 1010
     },
     {
       "epoch": 7.846153846153846,
-      "grad_norm": 2.2901501655578613,
+      "grad_norm": 2.261361837387085,
       "learning_rate": 9.53846153846154e-06,
-      "loss": 0.9066,
+      "loss": 0.9047,
       "step": 1020
     },
     {
       "epoch": 7.923076923076923,
-      "grad_norm": 2.5113565921783447,
+      "grad_norm": 2.6180179119110107,
       "learning_rate": 9.435897435897436e-06,
-      "loss": 0.8577,
+      "loss": 0.861,
       "step": 1030
     },
     {
       "epoch": 8.0,
-      "grad_norm": 4.092830181121826,
+      "grad_norm": 4.225304126739502,
       "learning_rate": 9.333333333333334e-06,
-      "loss": 0.9179,
+      "loss": 0.9192,
       "step": 1040
     },
     {
       "epoch": 8.0,
       "eval_accuracy": 0.8120300751879699,
-      "eval_loss": 0.8504595160484314,
-      "eval_runtime": 0.7701,
-      "eval_samples_per_second": 172.703,
-      "eval_steps_per_second": 22.075,
+      "eval_loss": 0.850643515586853,
+      "eval_runtime": 0.756,
+      "eval_samples_per_second": 175.92,
+      "eval_steps_per_second": 22.486,
       "step": 1040
     },
     {
       "epoch": 8.076923076923077,
-      "grad_norm": 1.9615790843963623,
+      "grad_norm": 2.1875813007354736,
       "learning_rate": 9.230769230769232e-06,
-      "loss": 0.8931,
+      "loss": 0.8953,
       "step": 1050
     },
     {
       "epoch": 8.153846153846153,
-      "grad_norm": 2.1320254802703857,
+      "grad_norm": 2.1640567779541016,
       "learning_rate": 9.128205128205129e-06,
-      "loss": 0.8652,
+      "loss": 0.8658,
       "step": 1060
     },
     {
       "epoch": 8.23076923076923,
-      "grad_norm": 2.0207359790802,
+      "grad_norm": 2.660614490509033,
       "learning_rate": 9.025641025641027e-06,
-      "loss": 0.8992,
+      "loss": 0.8995,
       "step": 1070
     },
     {
       "epoch": 8.307692307692308,
-      "grad_norm": 1.9664427042007446,
+      "grad_norm": 2.104029417037964,
       "learning_rate": 8.923076923076925e-06,
-      "loss": 0.8552,
+      "loss": 0.8569,
       "step": 1080
     },
     {
       "epoch": 8.384615384615385,
-      "grad_norm": 2.321737766265869,
+      "grad_norm": 2.2643303871154785,
       "learning_rate": 8.820512820512821e-06,
-      "loss": 0.8973,
+      "loss": 0.8972,
       "step": 1090
     },
     {
       "epoch": 8.461538461538462,
-      "grad_norm": 2.780322313308716,
+      "grad_norm": 2.632410764694214,
       "learning_rate": 8.717948717948719e-06,
-      "loss": 0.8695,
+      "loss": 0.8715,
       "step": 1100
     },
     {
       "epoch": 8.538461538461538,
-      "grad_norm": 1.664422631263733,
+      "grad_norm": 1.6500084400177002,
       "learning_rate": 8.615384615384617e-06,
-      "loss": 0.8711,
+      "loss": 0.8716,
       "step": 1110
     },
     {
       "epoch": 8.615384615384615,
-      "grad_norm": 4.4292097091674805,
+      "grad_norm": 6.204855442047119,
       "learning_rate": 8.512820512820513e-06,
-      "loss": 0.8983,
+      "loss": 0.8985,
       "step": 1120
     },
     {
       "epoch": 8.692307692307692,
-      "grad_norm": 3.7464237213134766,
+      "grad_norm": 3.729611873626709,
       "learning_rate": 8.410256410256411e-06,
-      "loss": 0.8831,
+      "loss": 0.8837,
       "step": 1130
     },
     {
       "epoch": 8.76923076923077,
-      "grad_norm": 1.9225133657455444,
+      "grad_norm": 3.685739278793335,
       "learning_rate": 8.307692307692309e-06,
-      "loss": 0.8851,
+      "loss": 0.8865,
       "step": 1140
     },
     {
       "epoch": 8.846153846153847,
-      "grad_norm": 2.6193883419036865,
+      "grad_norm": 2.7028560638427734,
       "learning_rate": 8.205128205128205e-06,
-      "loss": 0.8718,
+      "loss": 0.875,
       "step": 1150
     },
     {
       "epoch": 8.923076923076923,
-      "grad_norm": 3.2045023441314697,
+      "grad_norm": 2.7692482471466064,
       "learning_rate": 8.102564102564103e-06,
-      "loss": 0.8843,
+      "loss": 0.8867,
       "step": 1160
     },
     {
       "epoch": 9.0,
-      "grad_norm": 4.040080547332764,
+      "grad_norm": 3.9854462146759033,
       "learning_rate": 8.000000000000001e-06,
-      "loss": 0.9,
+      "loss": 0.9008,
       "step": 1170
     },
     {
       "epoch": 9.0,
       "eval_accuracy": 0.8045112781954887,
-      "eval_loss": 0.8327888250350952,
-      "eval_runtime": 0.764,
-      "eval_samples_per_second": 174.095,
-      "eval_steps_per_second": 22.253,
+      "eval_loss": 0.8337866067886353,
+      "eval_runtime": 0.7963,
+      "eval_samples_per_second": 167.03,
+      "eval_steps_per_second": 21.35,
       "step": 1170
     },
     {
       "epoch": 9.076923076923077,
-      "grad_norm": 1.9053655862808228,
+      "grad_norm": 1.9381572008132935,
       "learning_rate": 7.897435897435898e-06,
-      "loss": 0.8961,
+      "loss": 0.8969,
       "step": 1180
     },
     {
       "epoch": 9.153846153846153,
-      "grad_norm": 2.056864023208618,
+      "grad_norm": 2.219219446182251,
       "learning_rate": 7.794871794871796e-06,
-      "loss": 0.8369,
+      "loss": 0.8412,
       "step": 1190
     },
     {
       "epoch": 9.23076923076923,
-      "grad_norm": 2.1695327758789062,
+      "grad_norm": 2.1302294731140137,
       "learning_rate": 7.692307692307694e-06,
-      "loss": 0.8496,
+      "loss": 0.8483,
       "step": 1200
     },
     {
       "epoch": 9.307692307692308,
-      "grad_norm": 2.1930296421051025,
+      "grad_norm": 2.541210174560547,
       "learning_rate": 7.58974358974359e-06,
-      "loss": 0.8527,
+      "loss": 0.8536,
       "step": 1210
     },
     {
       "epoch": 9.384615384615385,
-      "grad_norm": 2.0710272789001465,
+      "grad_norm": 1.952871322631836,
       "learning_rate": 7.487179487179488e-06,
-      "loss": 0.867,
+      "loss": 0.8707,
       "step": 1220
     },
     {
       "epoch": 9.461538461538462,
-      "grad_norm": 3.325305938720703,
+      "grad_norm": 3.273028612136841,
       "learning_rate": 7.384615384615386e-06,
-      "loss": 0.8541,
+      "loss": 0.8547,
       "step": 1230
     },
     {
       "epoch": 9.538461538461538,
-      "grad_norm": 2.5524134635925293,
+      "grad_norm": 2.6495628356933594,
       "learning_rate": 7.282051282051282e-06,
-      "loss": 0.8711,
+      "loss": 0.8709,
       "step": 1240
     },
     {
       "epoch": 9.615384615384615,
-      "grad_norm": 3.508930206298828,
+      "grad_norm": 1.998024582862854,
       "learning_rate": 7.17948717948718e-06,
-      "loss": 0.8263,
+      "loss": 0.8278,
       "step": 1250
     },
     {
       "epoch": 9.692307692307692,
-      "grad_norm": 2.657597303390503,
+      "grad_norm": 2.7621707916259766,
       "learning_rate": 7.076923076923078e-06,
-      "loss": 0.8519,
+      "loss": 0.8544,
       "step": 1260
     },
     {
       "epoch": 9.76923076923077,
-      "grad_norm": 1.858414888381958,
+      "grad_norm": 1.844375491142273,
       "learning_rate": 6.974358974358974e-06,
-      "loss": 0.8306,
+      "loss": 0.8324,
       "step": 1270
     },
     {
       "epoch": 9.846153846153847,
-      "grad_norm": 2.1380136013031006,
+      "grad_norm": 2.149479866027832,
       "learning_rate": 6.871794871794872e-06,
-      "loss": 0.8115,
+      "loss": 0.8146,
       "step": 1280
     },
     {
       "epoch": 9.923076923076923,
-      "grad_norm": 2.1686654090881348,
+      "grad_norm": 2.2224795818328857,
       "learning_rate": 6.76923076923077e-06,
-      "loss": 0.8366,
+      "loss": 0.8367,
       "step": 1290
     },
     {
       "epoch": 10.0,
-      "grad_norm": 4.619501113891602,
+      "grad_norm": 3.8497843742370605,
       "learning_rate": 6.666666666666667e-06,
-      "loss": 0.8084,
+      "loss": 0.8079,
       "step": 1300
     },
     {
       "epoch": 10.0,
       "eval_accuracy": 0.8421052631578947,
-      "eval_loss": 0.807080864906311,
-      "eval_runtime": 0.7501,
-      "eval_samples_per_second": 177.313,
-      "eval_steps_per_second": 22.664,
+      "eval_loss": 0.8103837370872498,
+      "eval_runtime": 0.7593,
+      "eval_samples_per_second": 175.164,
+      "eval_steps_per_second": 22.389,
       "step": 1300
     },
     {
       "epoch": 10.076923076923077,
-      "grad_norm": 2.3092093467712402,
+      "grad_norm": 2.0343823432922363,
       "learning_rate": 6.564102564102565e-06,
-      "loss": 0.8366,
+      "loss": 0.8408,
       "step": 1310
     },
     {
       "epoch": 10.153846153846153,
-      "grad_norm": 2.2793080806732178,
+      "grad_norm": 2.4245193004608154,
       "learning_rate": 6.461538461538463e-06,
-      "loss": 0.89,
+      "loss": 0.899,
       "step": 1320
     },
     {
       "epoch": 10.23076923076923,
-      "grad_norm": 2.369584560394287,
+      "grad_norm": 2.3912925720214844,
       "learning_rate": 6.358974358974359e-06,
-      "loss": 0.8718,
+      "loss": 0.8758,
       "step": 1330
     },
     {
       "epoch": 10.307692307692308,
-      "grad_norm": 2.307018995285034,
+      "grad_norm": 2.1387076377868652,
       "learning_rate": 6.256410256410257e-06,
-      "loss": 0.8299,
+      "loss": 0.8295,
       "step": 1340
     },
     {
       "epoch": 10.384615384615385,
-      "grad_norm": 2.086519718170166,
+      "grad_norm": 2.142160415649414,
       "learning_rate": 6.153846153846155e-06,
-      "loss": 0.8096,
+      "loss": 0.8075,
       "step": 1350
     },
     {
       "epoch": 10.461538461538462,
-      "grad_norm": 2.7159907817840576,
+      "grad_norm": 2.6838831901550293,
       "learning_rate": 6.051282051282051e-06,
-      "loss": 0.8442,
+      "loss": 0.8448,
       "step": 1360
     },
     {
       "epoch": 10.538461538461538,
-      "grad_norm": 2.477699041366577,
+      "grad_norm": 2.476369857788086,
       "learning_rate": 5.948717948717949e-06,
-      "loss": 0.8133,
+      "loss": 0.817,
       "step": 1370
     },
     {
       "epoch": 10.615384615384615,
-      "grad_norm": 4.113278388977051,
+      "grad_norm": 3.031463861465454,
       "learning_rate": 5.846153846153847e-06,
-      "loss": 0.8159,
+      "loss": 0.8177,
       "step": 1380
     },
     {
       "epoch": 10.692307692307692,
-      "grad_norm": 2.264570951461792,
+      "grad_norm": 2.2818636894226074,
       "learning_rate": 5.743589743589743e-06,
-      "loss": 0.8086,
+      "loss": 0.8124,
       "step": 1390
     },
     {
       "epoch": 10.76923076923077,
-      "grad_norm": 3.312915802001953,
+      "grad_norm": 3.245805263519287,
       "learning_rate": 5.641025641025641e-06,
-      "loss": 0.8668,
+      "loss": 0.8674,
       "step": 1400
     },
     {
       "epoch": 10.846153846153847,
-      "grad_norm": 2.189178228378296,
+      "grad_norm": 2.194627046585083,
       "learning_rate": 5.538461538461539e-06,
-      "loss": 0.8298,
+      "loss": 0.831,
       "step": 1410
     },
     {
       "epoch": 10.923076923076923,
-      "grad_norm": 1.8542472124099731,
+      "grad_norm": 1.8149436712265015,
       "learning_rate": 5.435897435897436e-06,
-      "loss": 0.8329,
+      "loss": 0.8391,
       "step": 1420
     },
     {
       "epoch": 11.0,
-      "grad_norm": 4.028012752532959,
+      "grad_norm": 4.0584821701049805,
       "learning_rate": 5.333333333333334e-06,
-      "loss": 0.8306,
+      "loss": 0.8332,
       "step": 1430
     },
     {
       "epoch": 11.0,
       "eval_accuracy": 0.8345864661654135,
-      "eval_loss": 0.7760081887245178,
-      "eval_runtime": 0.7954,
-      "eval_samples_per_second": 167.205,
-      "eval_steps_per_second": 21.372,
+      "eval_loss": 0.7806060314178467,
+      "eval_runtime": 0.742,
+      "eval_samples_per_second": 179.256,
+      "eval_steps_per_second": 22.912,
       "step": 1430
     },
     {
       "epoch": 11.076923076923077,
-      "grad_norm": 1.9505723714828491,
+      "grad_norm": 1.9833248853683472,
       "learning_rate": 5.230769230769232e-06,
-      "loss": 0.8461,
+      "loss": 0.8484,
       "step": 1440
     },
     {
       "epoch": 11.153846153846153,
-      "grad_norm": 3.543419122695923,
+      "grad_norm": 5.478232383728027,
       "learning_rate": 5.128205128205128e-06,
-      "loss": 0.8287,
+      "loss": 0.8308,
       "step": 1450
     },
     {
       "epoch": 11.23076923076923,
-      "grad_norm": 2.6018764972686768,
+      "grad_norm": 2.5792922973632812,
       "learning_rate": 5.025641025641026e-06,
-      "loss": 0.7996,
+      "loss": 0.802,
       "step": 1460
     },
     {
       "epoch": 11.307692307692308,
-      "grad_norm": 2.7204878330230713,
+      "grad_norm": 2.730989694595337,
       "learning_rate": 4.923076923076924e-06,
-      "loss": 0.8218,
+      "loss": 0.8225,
       "step": 1470
     },
     {
       "epoch": 11.384615384615385,
-      "grad_norm": 2.6924397945404053,
+      "grad_norm": 2.7447853088378906,
       "learning_rate": 4.820512820512821e-06,
-      "loss": 0.8152,
+      "loss": 0.8176,
       "step": 1480
     },
     {
       "epoch": 11.461538461538462,
-      "grad_norm": 2.1294894218444824,
+      "grad_norm": 2.6465837955474854,
       "learning_rate": 4.717948717948718e-06,
-      "loss": 0.8452,
+      "loss": 0.8471,
       "step": 1490
     },
     {
       "epoch": 11.538461538461538,
-      "grad_norm": 2.45246958732605,
+      "grad_norm": 2.4876015186309814,
       "learning_rate": 4.615384615384616e-06,
-      "loss": 0.8316,
+      "loss": 0.8349,
       "step": 1500
     },
     {
       "epoch": 11.615384615384615,
-      "grad_norm": 3.2710139751434326,
+      "grad_norm": 3.2605788707733154,
       "learning_rate": 4.512820512820513e-06,
-      "loss": 0.8272,
+      "loss": 0.8285,
       "step": 1510
     },
     {
       "epoch": 11.692307692307692,
-      "grad_norm": 3.3108620643615723,
+      "grad_norm": 3.278341293334961,
       "learning_rate": 4.4102564102564104e-06,
-      "loss": 0.8539,
+      "loss": 0.8546,
       "step": 1520
     },
     {
       "epoch": 11.76923076923077,
-      "grad_norm": 2.079916000366211,
+      "grad_norm": 2.0945637226104736,
       "learning_rate": 4.307692307692308e-06,
-      "loss": 0.8067,
+      "loss": 0.8096,
       "step": 1530
     },
     {
       "epoch": 11.846153846153847,
-      "grad_norm": 2.2175967693328857,
+      "grad_norm": 2.161726474761963,
       "learning_rate": 4.2051282051282055e-06,
-      "loss": 0.7943,
+      "loss": 0.7938,
       "step": 1540
     },
     {
       "epoch": 11.923076923076923,
-      "grad_norm": 2.064258098602295,
+      "grad_norm": 2.1052703857421875,
       "learning_rate": 4.102564102564103e-06,
-      "loss": 0.8268,
+      "loss": 0.8295,
       "step": 1550
     },
     {
       "epoch": 12.0,
-      "grad_norm": 3.4313971996307373,
+      "grad_norm": 3.460094451904297,
       "learning_rate": 4.000000000000001e-06,
-      "loss": 0.8031,
+      "loss": 0.8103,
       "step": 1560
     },
     {
       "epoch": 12.0,
       "eval_accuracy": 0.8345864661654135,
-      "eval_loss": 0.7563135623931885,
-      "eval_runtime": 0.7817,
-      "eval_samples_per_second": 170.134,
-      "eval_steps_per_second": 21.746,
+      "eval_loss": 0.7585543990135193,
+      "eval_runtime": 0.7508,
+      "eval_samples_per_second": 177.133,
+      "eval_steps_per_second": 22.641,
       "step": 1560
     },
     {
       "epoch": 12.076923076923077,
-      "grad_norm": 2.1071786880493164,
+      "grad_norm": 2.943866014480591,
       "learning_rate": 3.897435897435898e-06,
-      "loss": 0.7877,
+      "loss": 0.7903,
       "step": 1570
     },
     {
       "epoch": 12.153846153846153,
-      "grad_norm": 2.6646640300750732,
+      "grad_norm": 2.6185402870178223,
       "learning_rate": 3.794871794871795e-06,
-      "loss": 0.8223,
+      "loss": 0.8229,
       "step": 1580
     },
     {
       "epoch": 12.23076923076923,
-      "grad_norm": 1.7268288135528564,
+      "grad_norm": 1.6378310918807983,
       "learning_rate": 3.692307692307693e-06,
-      "loss": 0.8204,
+      "loss": 0.8246,
       "step": 1590
     },
     {
       "epoch": 12.307692307692308,
-      "grad_norm": 2.982988119125366,
+      "grad_norm": 2.3109569549560547,
       "learning_rate": 3.58974358974359e-06,
-      "loss": 0.8336,
+      "loss": 0.8363,
       "step": 1600
     },
     {
       "epoch": 12.384615384615385,
-      "grad_norm": 2.559271812438965,
+      "grad_norm": 2.3602941036224365,
       "learning_rate": 3.487179487179487e-06,
-      "loss": 0.8064,
+      "loss": 0.8078,
       "step": 1610
     },
     {
       "epoch": 12.461538461538462,
-      "grad_norm": 2.528869390487671,
+      "grad_norm": 3.0623390674591064,
       "learning_rate": 3.384615384615385e-06,
-      "loss": 0.793,
+      "loss": 0.794,
       "step": 1620
     },
     {
       "epoch": 12.538461538461538,
-      "grad_norm": 2.1543517112731934,
+      "grad_norm": 2.947983741760254,
       "learning_rate": 3.2820512820512823e-06,
-      "loss": 0.8031,
+      "loss": 0.8033,
       "step": 1630
     },
     {
       "epoch": 12.615384615384615,
-      "grad_norm": 1.7778912782669067,
+      "grad_norm": 1.8083330392837524,
       "learning_rate": 3.1794871794871795e-06,
-      "loss": 0.8086,
+      "loss": 0.8158,
       "step": 1640
     },
     {
       "epoch": 12.692307692307692,
-      "grad_norm": 3.2962191104888916,
+      "grad_norm": 3.2873637676239014,
       "learning_rate": 3.0769230769230774e-06,
-      "loss": 0.7617,
+      "loss": 0.7651,
       "step": 1650
     },
     {
       "epoch": 12.76923076923077,
-      "grad_norm": 2.336732864379883,
+      "grad_norm": 2.3777670860290527,
       "learning_rate": 2.9743589743589746e-06,
-      "loss": 0.8541,
+      "loss": 0.8566,
       "step": 1660
     },
     {
       "epoch": 12.846153846153847,
-      "grad_norm": 1.9721437692642212,
+      "grad_norm": 1.8692084550857544,
       "learning_rate": 2.8717948717948717e-06,
-      "loss": 0.8249,
+      "loss": 0.8218,
       "step": 1670
     },
     {
       "epoch": 12.923076923076923,
-      "grad_norm": 2.0368008613586426,
+      "grad_norm": 2.2379138469696045,
       "learning_rate": 2.7692307692307697e-06,
-      "loss": 0.795,
+      "loss": 0.7984,
       "step": 1680
     },
     {
       "epoch": 13.0,
-      "grad_norm": 3.8102405071258545,
+      "grad_norm": 4.131476879119873,
       "learning_rate": 2.666666666666667e-06,
-      "loss": 0.8138,
+      "loss": 0.8149,
       "step": 1690
     },
     {
       "epoch": 13.0,
       "eval_accuracy": 0.8421052631578947,
-      "eval_loss": 0.753366231918335,
-      "eval_runtime": 0.7711,
-      "eval_samples_per_second": 172.483,
-      "eval_steps_per_second": 22.047,
+      "eval_loss": 0.757113516330719,
+      "eval_runtime": 0.7762,
+      "eval_samples_per_second": 171.337,
+      "eval_steps_per_second": 21.9,
       "step": 1690
     },
     {
       "epoch": 13.076923076923077,
-      "grad_norm": 3.0844614505767822,
+      "grad_norm": 2.9936656951904297,
       "learning_rate": 2.564102564102564e-06,
-      "loss": 0.7901,
+      "loss": 0.7917,
       "step": 1700
     },
     {
       "epoch": 13.153846153846153,
-      "grad_norm": 2.517000913619995,
+      "grad_norm": 2.5392699241638184,
       "learning_rate": 2.461538461538462e-06,
-      "loss": 0.8205,
+      "loss": 0.8241,
       "step": 1710
     },
     {
       "epoch": 13.23076923076923,
-      "grad_norm": 3.047574520111084,
+      "grad_norm": 3.0166265964508057,
       "learning_rate": 2.358974358974359e-06,
-      "loss": 0.8113,
+      "loss": 0.8117,
       "step": 1720
     },
     {
       "epoch": 13.307692307692308,
-      "grad_norm": 1.9097496271133423,
+      "grad_norm": 1.8728867769241333,
       "learning_rate": 2.2564102564102566e-06,
-      "loss": 0.817,
+      "loss": 0.8155,
       "step": 1730
     },
     {
       "epoch": 13.384615384615385,
-      "grad_norm": 2.553558826446533,
+      "grad_norm": 2.50715708732605,
       "learning_rate": 2.153846153846154e-06,
-      "loss": 0.7794,
+      "loss": 0.7814,
       "step": 1740
     },
     {
       "epoch": 13.461538461538462,
-      "grad_norm": 3.915072202682495,
+      "grad_norm": 5.447348594665527,
       "learning_rate": 2.0512820512820513e-06,
-      "loss": 0.8238,
+      "loss": 0.8253,
       "step": 1750
     },
     {
       "epoch": 13.538461538461538,
-      "grad_norm": 2.7563774585723877,
+      "grad_norm": 2.6522035598754883,
       "learning_rate": 1.948717948717949e-06,
-      "loss": 0.8464,
+      "loss": 0.8486,
       "step": 1760
     },
     {
       "epoch": 13.615384615384615,
-      "grad_norm": 1.9687329530715942,
+      "grad_norm": 2.1300199031829834,
       "learning_rate": 1.8461538461538465e-06,
-      "loss": 0.7998,
+      "loss": 0.8027,
       "step": 1770
     },
     {
       "epoch": 13.692307692307692,
-      "grad_norm": 1.9707388877868652,
+      "grad_norm": 2.1135923862457275,
       "learning_rate": 1.7435897435897436e-06,
-      "loss": 0.7827,
+      "loss": 0.7852,
       "step": 1780
     },
     {
       "epoch": 13.76923076923077,
-      "grad_norm": 1.8745958805084229,
+      "grad_norm": 1.871300220489502,
       "learning_rate": 1.6410256410256412e-06,
-      "loss": 0.8211,
+      "loss": 0.8224,
       "step": 1790
     },
     {
       "epoch": 13.846153846153847,
-      "grad_norm": 3.193314552307129,
+      "grad_norm": 3.240356206893921,
       "learning_rate": 1.5384615384615387e-06,
-      "loss": 0.7874,
+      "loss": 0.7895,
       "step": 1800
     },
     {
       "epoch": 13.923076923076923,
-      "grad_norm": 2.4726743698120117,
+      "grad_norm": 2.5182340145111084,
       "learning_rate": 1.4358974358974359e-06,
-      "loss": 0.7323,
+      "loss": 0.7316,
       "step": 1810
     },
     {
       "epoch": 14.0,
-      "grad_norm": 4.2939653396606445,
+      "grad_norm": 4.281803607940674,
       "learning_rate": 1.3333333333333334e-06,
-      "loss": 0.8178,
+      "loss": 0.8186,
       "step": 1820
     },
     {
       "epoch": 14.0,
       "eval_accuracy": 0.8270676691729323,
-      "eval_loss": 0.7507623434066772,
-      "eval_runtime": 0.7401,
-      "eval_samples_per_second": 179.698,
-      "eval_steps_per_second": 22.969,
+      "eval_loss": 0.7540305852890015,
+      "eval_runtime": 0.7703,
+      "eval_samples_per_second": 172.654,
+      "eval_steps_per_second": 22.069,
       "step": 1820
     },
     {
       "epoch": 14.076923076923077,
-      "grad_norm": 2.51151967048645,
+      "grad_norm": 2.050518751144409,
       "learning_rate": 1.230769230769231e-06,
-      "loss": 0.8201,
+      "loss": 0.8222,
       "step": 1830
     },
     {
       "epoch": 14.153846153846153,
-      "grad_norm": 2.035071849822998,
+      "grad_norm": 2.051259994506836,
       "learning_rate": 1.1282051282051283e-06,
-      "loss": 0.7885,
+      "loss": 0.7878,
       "step": 1840
     },
     {
       "epoch": 14.23076923076923,
-      "grad_norm": 2.924006938934326,
+      "grad_norm": 2.8861193656921387,
       "learning_rate": 1.0256410256410257e-06,
-      "loss": 0.7802,
+      "loss": 0.78,
       "step": 1850
     },
     {
       "epoch": 14.307692307692308,
-      "grad_norm": 3.3788881301879883,
+      "grad_norm": 4.159270763397217,
       "learning_rate": 9.230769230769232e-07,
-      "loss": 0.7717,
+      "loss": 0.774,
       "step": 1860
     },
     {
       "epoch": 14.384615384615385,
-      "grad_norm": 2.6425819396972656,
+      "grad_norm": 2.8624985218048096,
       "learning_rate": 8.205128205128206e-07,
-      "loss": 0.7871,
+      "loss": 0.7882,
       "step": 1870
     },
     {
       "epoch": 14.461538461538462,
-      "grad_norm": 2.781729221343994,
+      "grad_norm": 2.5051703453063965,
       "learning_rate": 7.179487179487179e-07,
-      "loss": 0.7859,
+      "loss": 0.7883,
       "step": 1880
     },
     {
       "epoch": 14.538461538461538,
-      "grad_norm": 3.0325913429260254,
+      "grad_norm": 3.003545045852661,
       "learning_rate": 6.153846153846155e-07,
-      "loss": 0.7783,
+      "loss": 0.7817,
       "step": 1890
     },
     {
       "epoch": 14.615384615384615,
-      "grad_norm": 2.7311503887176514,
+      "grad_norm": 2.8403878211975098,
       "learning_rate": 5.128205128205128e-07,
-      "loss": 0.8273,
+      "loss": 0.8294,
       "step": 1900
     },
     {
       "epoch": 14.692307692307692,
-      "grad_norm": 2.3128809928894043,
+      "grad_norm": 2.124030590057373,
       "learning_rate": 4.102564102564103e-07,
-      "loss": 0.7977,
+      "loss": 0.7978,
       "step": 1910
     },
     {
       "epoch": 14.76923076923077,
-      "grad_norm": 3.4510748386383057,
+      "grad_norm": 4.762181758880615,
       "learning_rate": 3.0769230769230774e-07,
-      "loss": 0.8029,
+      "loss": 0.8038,
       "step": 1920
     },
     {
       "epoch": 14.846153846153847,
-      "grad_norm": 3.1088345050811768,
+      "grad_norm": 3.256133794784546,
       "learning_rate": 2.0512820512820514e-07,
-      "loss": 0.8509,
+      "loss": 0.8535,
       "step": 1930
     },
     {
       "epoch": 14.923076923076923,
-      "grad_norm": 2.3365509510040283,
+      "grad_norm": 2.355344772338867,
       "learning_rate": 1.0256410256410257e-07,
-      "loss": 0.7569,
+      "loss": 0.7587,
       "step": 1940
     },
     {
       "epoch": 15.0,
-      "grad_norm": 4.560791015625,
+      "grad_norm": 4.202574729919434,
       "learning_rate": 0.0,
-      "loss": 0.7901,
+      "loss": 0.7929,
       "step": 1950
     },
     {
       "epoch": 15.0,
-      "eval_accuracy": 0.8195488721804511,
-      "eval_loss": 0.738013505935669,
-      "eval_runtime": 0.796,
-      "eval_samples_per_second": 167.085,
-      "eval_steps_per_second": 21.357,
+      "eval_accuracy": 0.8120300751879699,
+      "eval_loss": 0.7412300109863281,
+      "eval_runtime": 0.8087,
+      "eval_samples_per_second": 164.47,
+      "eval_steps_per_second": 21.022,
       "step": 1950
     },
     {
       "epoch": 15.0,
       "step": 1950,
       "total_flos": 1.5658365504595968e+17,
-      "train_loss": 0.0,
-      "train_runtime": 1.3792,
-      "train_samples_per_second": 11245.899,
-      "train_steps_per_second": 1413.894
+      "train_loss": 0.9236146088135548,
+      "train_runtime": 142.953,
+      "train_samples_per_second": 108.497,
+      "train_steps_per_second": 13.641
     }
   ],
   "logging_steps": 10,