🍻 cheers

README.md

@@ -2,6 +2,7 @@
 license: apache-2.0
 base_model: google/vit-base-patch16-224-in21k
 tags:
+- image-classification
 - generated_from_trainer
 metrics:
 - accuracy
@@ -15,10 +16,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # finetuned-electrical-images
 
-This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on an unknown dataset.
+This model is a fine-tuned version of [google/vit-base-patch16-224-in21k](https://huggingface.co/google/vit-base-patch16-224-in21k) on the Electrical_components(VIT) dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.4092
-- Accuracy: 0.8927
+- Loss: 0.3726
+- Accuracy: 0.8861
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.8976897689768977,
-    "eval_loss": 0.36343997716903687,
-    "eval_runtime": 8.0693,
-    "eval_samples_per_second": 75.1,
-    "eval_steps_per_second": 9.418,
+    "eval_accuracy": 0.8861386138613861,
+    "eval_loss": 0.37258049845695496,
+    "eval_runtime": 8.1604,
+    "eval_samples_per_second": 74.261,
+    "eval_steps_per_second": 9.313,
     "total_flos": 2.659620429501235e+18,
-    "train_loss": 0.3051174351919529,
-    "train_runtime": 1549.0078,
-    "train_samples_per_second": 22.156,
-    "train_steps_per_second": 1.388
+    "train_loss": 0.30670992596204893,
+    "train_runtime": 1546.6686,
+    "train_samples_per_second": 22.19,
+    "train_steps_per_second": 1.39
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 10.0,
-    "eval_accuracy": 0.8976897689768977,
-    "eval_loss": 0.36343997716903687,
-    "eval_runtime": 8.0693,
-    "eval_samples_per_second": 75.1,
-    "eval_steps_per_second": 9.418
+    "eval_accuracy": 0.8861386138613861,
+    "eval_loss": 0.37258049845695496,
+    "eval_runtime": 8.1604,
+    "eval_samples_per_second": 74.261,
+    "eval_steps_per_second": 9.313
 }

runs/Apr26_13-59-20_cee49ebfb78a/events.out.tfevents.1714141556.cee49ebfb78a.754.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:712abb3fd2beb0e628ca04b958192145d61558ea28992085cf0dd482fe8b6975
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 10.0,
     "total_flos": 2.659620429501235e+18,
-    "train_loss": 0.3051174351919529,
-    "train_runtime": 1549.0078,
-    "train_samples_per_second": 22.156,
-    "train_steps_per_second": 1.388
+    "train_loss": 0.30670992596204893,
+    "train_runtime": 1546.6686,
+    "train_samples_per_second": 22.19,
+    "train_steps_per_second": 1.39
 }

trainer_state.json

@@ -1,6 +1,6 @@
 {
-  "best_metric": 0.36343997716903687,
-  "best_model_checkpoint": "finetuned-electrical-images/checkpoint-1300",
+  "best_metric": 0.37258049845695496,
+  "best_model_checkpoint": "finetuned-electrical-images/checkpoint-1500",
   "epoch": 10.0,
   "eval_steps": 100,
   "global_step": 2150,
@@ -10,1706 +10,1706 @@
   "log_history": [
     {
       "epoch": 0.046511627906976744,
-      "grad_norm": 2.1625218391418457,
+      "grad_norm": 2.7490031719207764,
       "learning_rate": 0.00019906976744186048,
-      "loss": 1.5809,
+      "loss": 1.5465,
       "step": 10
     },
     {
       "epoch": 0.09302325581395349,
-      "grad_norm": 2.723552703857422,
+      "grad_norm": 3.2069549560546875,
       "learning_rate": 0.00019813953488372096,
-      "loss": 1.2765,
+      "loss": 1.2151,
       "step": 20
     },
     {
       "epoch": 0.13953488372093023,
-      "grad_norm": 2.43442702293396,
+      "grad_norm": 4.26694393157959,
       "learning_rate": 0.0001972093023255814,
-      "loss": 1.1166,
+      "loss": 1.0806,
       "step": 30
     },
     {
       "epoch": 0.18604651162790697,
-      "grad_norm": 3.831533670425415,
+      "grad_norm": 1.9304146766662598,
       "learning_rate": 0.00019627906976744185,
-      "loss": 1.0432,
+      "loss": 0.9911,
       "step": 40
     },
     {
       "epoch": 0.23255813953488372,
-      "grad_norm": 1.6888493299484253,
+      "grad_norm": 4.346214294433594,
       "learning_rate": 0.00019534883720930232,
-      "loss": 0.8645,
+      "loss": 0.8587,
       "step": 50
     },
     {
       "epoch": 0.27906976744186046,
-      "grad_norm": 3.6029133796691895,
+      "grad_norm": 3.977471113204956,
       "learning_rate": 0.0001944186046511628,
-      "loss": 0.8017,
+      "loss": 0.7717,
       "step": 60
     },
     {
       "epoch": 0.32558139534883723,
-      "grad_norm": 3.071220636367798,
+      "grad_norm": 3.1426613330841064,
       "learning_rate": 0.00019348837209302326,
-      "loss": 0.8024,
+      "loss": 0.961,
       "step": 70
     },
     {
       "epoch": 0.37209302325581395,
-      "grad_norm": 2.4523532390594482,
+      "grad_norm": 2.316943407058716,
       "learning_rate": 0.00019255813953488374,
-      "loss": 0.7961,
+      "loss": 0.8362,
       "step": 80
     },
     {
       "epoch": 0.4186046511627907,
-      "grad_norm": 2.8159303665161133,
+      "grad_norm": 2.885920286178589,
       "learning_rate": 0.0001916279069767442,
-      "loss": 0.7968,
+      "loss": 0.829,
       "step": 90
     },
     {
       "epoch": 0.46511627906976744,
-      "grad_norm": 2.5400731563568115,
+      "grad_norm": 1.4288862943649292,
       "learning_rate": 0.00019069767441860466,
-      "loss": 0.7236,
+      "loss": 0.7116,
       "step": 100
     },
     {
       "epoch": 0.46511627906976744,
-      "eval_accuracy": 0.8102310231023102,
-      "eval_loss": 0.6395960450172424,
-      "eval_runtime": 8.3162,
-      "eval_samples_per_second": 72.87,
-      "eval_steps_per_second": 9.139,
+      "eval_accuracy": 0.7920792079207921,
+      "eval_loss": 0.6398854851722717,
+      "eval_runtime": 7.8011,
+      "eval_samples_per_second": 77.682,
+      "eval_steps_per_second": 9.742,
       "step": 100
     },
     {
       "epoch": 0.5116279069767442,
-      "grad_norm": 5.171610355377197,
+      "grad_norm": 3.5353667736053467,
       "learning_rate": 0.00018976744186046513,
-      "loss": 0.6643,
+      "loss": 0.5819,
       "step": 110
     },
     {
       "epoch": 0.5581395348837209,
-      "grad_norm": 5.763847827911377,
+      "grad_norm": 3.7112417221069336,
       "learning_rate": 0.00018883720930232557,
-      "loss": 0.6579,
+      "loss": 0.5016,
       "step": 120
     },
     {
       "epoch": 0.6046511627906976,
-      "grad_norm": 2.429523229598999,
+      "grad_norm": 3.1544606685638428,
       "learning_rate": 0.00018790697674418605,
-      "loss": 0.5872,
+      "loss": 0.6231,
       "step": 130
     },
     {
       "epoch": 0.6511627906976745,
-      "grad_norm": 2.6132380962371826,
+      "grad_norm": 2.670633316040039,
       "learning_rate": 0.00018697674418604652,
-      "loss": 0.6796,
+      "loss": 0.6856,
       "step": 140
     },
     {
       "epoch": 0.6976744186046512,
-      "grad_norm": 3.3510539531707764,
+      "grad_norm": 3.5320849418640137,
       "learning_rate": 0.000186046511627907,
-      "loss": 0.5738,
+      "loss": 0.5361,
       "step": 150
     },
     {
       "epoch": 0.7441860465116279,
-      "grad_norm": 3.596841812133789,
+      "grad_norm": 5.482109546661377,
       "learning_rate": 0.00018511627906976744,
-      "loss": 0.5647,
+      "loss": 0.6735,
       "step": 160
     },
     {
       "epoch": 0.7906976744186046,
-      "grad_norm": 2.6907944679260254,
+      "grad_norm": 2.8392834663391113,
       "learning_rate": 0.0001841860465116279,
-      "loss": 0.5161,
+      "loss": 0.5717,
       "step": 170
     },
     {
       "epoch": 0.8372093023255814,
-      "grad_norm": 2.8201518058776855,
+      "grad_norm": 1.8788505792617798,
       "learning_rate": 0.00018325581395348838,
-      "loss": 0.5416,
+      "loss": 0.6031,
       "step": 180
     },
     {
       "epoch": 0.8837209302325582,
-      "grad_norm": 3.49650239944458,
+      "grad_norm": 3.153599977493286,
       "learning_rate": 0.00018232558139534886,
-      "loss": 0.5873,
+      "loss": 0.5841,
       "step": 190
     },
     {
       "epoch": 0.9302325581395349,
-      "grad_norm": 1.33900785446167,
+      "grad_norm": 1.6330416202545166,
       "learning_rate": 0.0001813953488372093,
-      "loss": 0.7243,
+      "loss": 0.6953,
       "step": 200
     },
     {
       "epoch": 0.9302325581395349,
-      "eval_accuracy": 0.8333333333333334,
-      "eval_loss": 0.5123782753944397,
-      "eval_runtime": 9.3797,
-      "eval_samples_per_second": 64.608,
-      "eval_steps_per_second": 8.103,
+      "eval_accuracy": 0.8085808580858086,
+      "eval_loss": 0.5588750839233398,
+      "eval_runtime": 8.0533,
+      "eval_samples_per_second": 75.249,
+      "eval_steps_per_second": 9.437,
       "step": 200
     },
     {
       "epoch": 0.9767441860465116,
-      "grad_norm": 2.2079968452453613,
+      "grad_norm": 3.0009500980377197,
       "learning_rate": 0.00018046511627906977,
-      "loss": 0.5871,
+      "loss": 0.518,
       "step": 210
     },
     {
       "epoch": 1.0232558139534884,
-      "grad_norm": 2.7433505058288574,
+      "grad_norm": 5.527990341186523,
       "learning_rate": 0.00017953488372093025,
-      "loss": 0.5605,
+      "loss": 0.5661,
       "step": 220
     },
     {
       "epoch": 1.069767441860465,
-      "grad_norm": 4.832902908325195,
+      "grad_norm": 1.3012688159942627,
       "learning_rate": 0.0001786046511627907,
-      "loss": 0.4689,
+      "loss": 0.4236,
       "step": 230
     },
     {
       "epoch": 1.1162790697674418,
-      "grad_norm": 3.20340633392334,
+      "grad_norm": 1.5813956260681152,
       "learning_rate": 0.00017767441860465117,
-      "loss": 0.4492,
+      "loss": 0.4062,
       "step": 240
     },
     {
       "epoch": 1.1627906976744187,
-      "grad_norm": 3.986785650253296,
+      "grad_norm": 6.4304890632629395,
       "learning_rate": 0.00017674418604651164,
-      "loss": 0.477,
+      "loss": 0.4689,
       "step": 250
     },
     {
       "epoch": 1.2093023255813953,
-      "grad_norm": 3.679204225540161,
+      "grad_norm": 3.993091583251953,
       "learning_rate": 0.0001758139534883721,
-      "loss": 0.4499,
+      "loss": 0.4486,
       "step": 260
     },
     {
       "epoch": 1.255813953488372,
-      "grad_norm": 5.091402530670166,
+      "grad_norm": 4.19878625869751,
       "learning_rate": 0.00017488372093023258,
-      "loss": 0.4271,
+      "loss": 0.577,
       "step": 270
     },
     {
       "epoch": 1.302325581395349,
-      "grad_norm": 1.7408082485198975,
+      "grad_norm": 3.3949573040008545,
       "learning_rate": 0.00017395348837209303,
-      "loss": 0.4805,
+      "loss": 0.4669,
       "step": 280
     },
     {
       "epoch": 1.3488372093023255,
-      "grad_norm": 1.776224136352539,
+      "grad_norm": 3.5557937622070312,
       "learning_rate": 0.00017302325581395348,
-      "loss": 0.5031,
+      "loss": 0.4234,
       "step": 290
     },
     {
       "epoch": 1.3953488372093024,
-      "grad_norm": 1.4141583442687988,
+      "grad_norm": 2.519487142562866,
       "learning_rate": 0.00017209302325581395,
-      "loss": 0.4288,
+      "loss": 0.4078,
       "step": 300
     },
     {
       "epoch": 1.3953488372093024,
-      "eval_accuracy": 0.863036303630363,
-      "eval_loss": 0.45135730504989624,
-      "eval_runtime": 8.2539,
-      "eval_samples_per_second": 73.42,
-      "eval_steps_per_second": 9.208,
+      "eval_accuracy": 0.8399339933993399,
+      "eval_loss": 0.4945888817310333,
+      "eval_runtime": 8.261,
+      "eval_samples_per_second": 73.357,
+      "eval_steps_per_second": 9.2,
       "step": 300
     },
     {
       "epoch": 1.441860465116279,
-      "grad_norm": 3.1988017559051514,
+      "grad_norm": 3.2524163722991943,
       "learning_rate": 0.00017116279069767442,
-      "loss": 0.4641,
+      "loss": 0.4963,
       "step": 310
     },
     {
       "epoch": 1.4883720930232558,
-      "grad_norm": 3.4049489498138428,
+      "grad_norm": 2.7722671031951904,
       "learning_rate": 0.0001702325581395349,
-      "loss": 0.5539,
+      "loss": 0.5716,
       "step": 320
     },
     {
       "epoch": 1.5348837209302326,
-      "grad_norm": 2.879892349243164,
+      "grad_norm": 3.2851498126983643,
       "learning_rate": 0.00016930232558139537,
-      "loss": 0.4818,
+      "loss": 0.4878,
       "step": 330
     },
     {
       "epoch": 1.5813953488372094,
-      "grad_norm": 3.1155309677124023,
+      "grad_norm": 1.52852463722229,
       "learning_rate": 0.00016837209302325584,
-      "loss": 0.5773,
+      "loss": 0.4834,
       "step": 340
     },
     {
       "epoch": 1.627906976744186,
-      "grad_norm": 5.9107866287231445,
+      "grad_norm": 6.030341625213623,
       "learning_rate": 0.00016744186046511629,
-      "loss": 0.5665,
+      "loss": 0.5827,
       "step": 350
     },
     {
       "epoch": 1.6744186046511627,
-      "grad_norm": 3.242166519165039,
+      "grad_norm": 3.0944032669067383,
       "learning_rate": 0.00016651162790697673,
-      "loss": 0.4483,
+      "loss": 0.4615,
       "step": 360
     },
     {
       "epoch": 1.7209302325581395,
-      "grad_norm": 2.6073851585388184,
+      "grad_norm": 3.8217456340789795,
       "learning_rate": 0.0001655813953488372,
-      "loss": 0.3956,
+      "loss": 0.4558,
       "step": 370
     },
     {
       "epoch": 1.7674418604651163,
-      "grad_norm": 2.7050387859344482,
+      "grad_norm": 2.4076743125915527,
       "learning_rate": 0.00016465116279069768,
-      "loss": 0.5734,
+      "loss": 0.5732,
       "step": 380
     },
     {
       "epoch": 1.8139534883720931,
-      "grad_norm": 4.9598259925842285,
+      "grad_norm": 4.107614994049072,
       "learning_rate": 0.00016372093023255815,
-      "loss": 0.6565,
+      "loss": 0.6245,
       "step": 390
     },
     {
       "epoch": 1.8604651162790697,
-      "grad_norm": 2.172515630722046,
+      "grad_norm": 1.4518159627914429,
       "learning_rate": 0.00016279069767441862,
-      "loss": 0.5744,
+      "loss": 0.5852,
       "step": 400
     },
     {
       "epoch": 1.8604651162790697,
-      "eval_accuracy": 0.8102310231023102,
-      "eval_loss": 0.6154166460037231,
-      "eval_runtime": 8.1716,
-      "eval_samples_per_second": 74.159,
-      "eval_steps_per_second": 9.3,
+      "eval_accuracy": 0.8399339933993399,
+      "eval_loss": 0.48718997836112976,
+      "eval_runtime": 8.1286,
+      "eval_samples_per_second": 74.552,
+      "eval_steps_per_second": 9.35,
       "step": 400
     },
     {
       "epoch": 1.9069767441860463,
-      "grad_norm": 1.5915610790252686,
+      "grad_norm": 2.9517464637756348,
       "learning_rate": 0.00016186046511627907,
-      "loss": 0.6044,
+      "loss": 0.5802,
       "step": 410
     },
     {
       "epoch": 1.9534883720930232,
-      "grad_norm": 3.225754976272583,
+      "grad_norm": 3.5694568157196045,
       "learning_rate": 0.00016093023255813954,
-      "loss": 0.5832,
+      "loss": 0.4936,
       "step": 420
     },
     {
       "epoch": 2.0,
-      "grad_norm": 5.616280555725098,
+      "grad_norm": 0.7227370142936707,
       "learning_rate": 0.00016,
-      "loss": 0.3845,
+      "loss": 0.4235,
       "step": 430
     },
     {
       "epoch": 2.046511627906977,
-      "grad_norm": 2.0786819458007812,
+      "grad_norm": 2.4811441898345947,
       "learning_rate": 0.00015906976744186046,
-      "loss": 0.4349,
+      "loss": 0.4333,
       "step": 440
     },
     {
       "epoch": 2.0930232558139537,
-      "grad_norm": 2.8910179138183594,
+      "grad_norm": 2.808461904525757,
       "learning_rate": 0.00015813953488372093,
-      "loss": 0.4858,
+      "loss": 0.4659,
       "step": 450
     },
     {
       "epoch": 2.13953488372093,
-      "grad_norm": 2.445451259613037,
+      "grad_norm": 2.1043217182159424,
       "learning_rate": 0.0001572093023255814,
-      "loss": 0.3341,
+      "loss": 0.3284,
       "step": 460
     },
     {
       "epoch": 2.186046511627907,
-      "grad_norm": 3.979907751083374,
+      "grad_norm": 3.650707483291626,
       "learning_rate": 0.00015627906976744188,
-      "loss": 0.4102,
+      "loss": 0.3416,
       "step": 470
     },
     {
       "epoch": 2.2325581395348837,
-      "grad_norm": 3.332590103149414,
+      "grad_norm": 2.71091628074646,
       "learning_rate": 0.00015534883720930232,
-      "loss": 0.4107,
+      "loss": 0.456,
       "step": 480
     },
     {
       "epoch": 2.2790697674418605,
-      "grad_norm": 3.54638671875,
+      "grad_norm": 2.2614614963531494,
       "learning_rate": 0.0001544186046511628,
-      "loss": 0.3151,
+      "loss": 0.4062,
       "step": 490
     },
     {
       "epoch": 2.3255813953488373,
-      "grad_norm": 5.199957847595215,
+      "grad_norm": 3.2703847885131836,
       "learning_rate": 0.00015348837209302327,
-      "loss": 0.4077,
+      "loss": 0.4993,
       "step": 500
     },
     {
       "epoch": 2.3255813953488373,
-      "eval_accuracy": 0.8613861386138614,
-      "eval_loss": 0.46117761731147766,
-      "eval_runtime": 8.1068,
-      "eval_samples_per_second": 74.752,
-      "eval_steps_per_second": 9.375,
+      "eval_accuracy": 0.8597359735973598,
+      "eval_loss": 0.46872881054878235,
+      "eval_runtime": 8.4417,
+      "eval_samples_per_second": 71.786,
+      "eval_steps_per_second": 9.003,
       "step": 500
     },
     {
       "epoch": 2.3720930232558137,
-      "grad_norm": 4.220390796661377,
+      "grad_norm": 4.928767204284668,
       "learning_rate": 0.00015255813953488374,
-      "loss": 0.4136,
+      "loss": 0.4723,
       "step": 510
     },
     {
       "epoch": 2.4186046511627906,
-      "grad_norm": 3.7662441730499268,
+      "grad_norm": 4.274898052215576,
       "learning_rate": 0.0001516279069767442,
-      "loss": 0.4278,
+      "loss": 0.4886,
       "step": 520
     },
     {
       "epoch": 2.4651162790697674,
-      "grad_norm": 3.1768009662628174,
+      "grad_norm": 2.146392345428467,
       "learning_rate": 0.00015069767441860466,
-      "loss": 0.3468,
+      "loss": 0.3828,
       "step": 530
     },
     {
       "epoch": 2.511627906976744,
-      "grad_norm": 4.637648105621338,
+      "grad_norm": 3.7208266258239746,
       "learning_rate": 0.0001497674418604651,
-      "loss": 0.4745,
+      "loss": 0.412,
       "step": 540
     },
     {
       "epoch": 2.558139534883721,
-      "grad_norm": 4.119113445281982,
+      "grad_norm": 2.139849901199341,
       "learning_rate": 0.00014883720930232558,
-      "loss": 0.3556,
+      "loss": 0.3794,
       "step": 550
     },
     {
       "epoch": 2.604651162790698,
-      "grad_norm": 1.932075023651123,
+      "grad_norm": 1.760920524597168,
       "learning_rate": 0.00014790697674418605,
-      "loss": 0.3924,
+      "loss": 0.3078,
       "step": 560
     },
     {
       "epoch": 2.6511627906976747,
-      "grad_norm": 1.4679527282714844,
+      "grad_norm": 3.4851582050323486,
       "learning_rate": 0.00014697674418604652,
-      "loss": 0.492,
+      "loss": 0.3498,
       "step": 570
     },
     {
       "epoch": 2.697674418604651,
-      "grad_norm": 3.493939161300659,
+      "grad_norm": 1.7796868085861206,
       "learning_rate": 0.000146046511627907,
-      "loss": 0.4785,
+      "loss": 0.4433,
       "step": 580
     },
     {
       "epoch": 2.744186046511628,
-      "grad_norm": 5.127101898193359,
+      "grad_norm": 2.8916869163513184,
       "learning_rate": 0.00014511627906976747,
-      "loss": 0.3809,
+      "loss": 0.3614,
       "step": 590
     },
     {
       "epoch": 2.7906976744186047,
-      "grad_norm": 4.6078925132751465,
+      "grad_norm": 4.171596527099609,
       "learning_rate": 0.00014418604651162791,
-      "loss": 0.496,
+      "loss": 0.4479,
       "step": 600
     },
     {
       "epoch": 2.7906976744186047,
-      "eval_accuracy": 0.8729372937293729,
-      "eval_loss": 0.43590766191482544,
-      "eval_runtime": 7.9386,
-      "eval_samples_per_second": 76.336,
-      "eval_steps_per_second": 9.573,
+      "eval_accuracy": 0.8844884488448845,
+      "eval_loss": 0.3985598087310791,
+      "eval_runtime": 8.4641,
+      "eval_samples_per_second": 71.596,
+      "eval_steps_per_second": 8.979,
       "step": 600
     },
     {
       "epoch": 2.8372093023255816,
-      "grad_norm": 2.777296781539917,
+      "grad_norm": 2.6928250789642334,
       "learning_rate": 0.00014325581395348836,
-      "loss": 0.3873,
+      "loss": 0.3827,
       "step": 610
     },
     {
       "epoch": 2.883720930232558,
-      "grad_norm": 3.4130568504333496,
+      "grad_norm": 2.466237783432007,
       "learning_rate": 0.00014232558139534883,
-      "loss": 0.4155,
+      "loss": 0.3145,
       "step": 620
     },
     {
       "epoch": 2.9302325581395348,
-      "grad_norm": 3.3363895416259766,
+      "grad_norm": 2.5348870754241943,
       "learning_rate": 0.0001413953488372093,
-      "loss": 0.4497,
+      "loss": 0.4285,
       "step": 630
     },
     {
       "epoch": 2.9767441860465116,
-      "grad_norm": 2.6814212799072266,
+      "grad_norm": 4.993005275726318,
       "learning_rate": 0.00014046511627906978,
-      "loss": 0.3669,
+      "loss": 0.4147,
       "step": 640
     },
     {
       "epoch": 3.0232558139534884,
-      "grad_norm": 4.29351282119751,
+      "grad_norm": 3.5268807411193848,
       "learning_rate": 0.00013953488372093025,
-      "loss": 0.3421,
+      "loss": 0.2991,
       "step": 650
     },
     {
       "epoch": 3.0697674418604652,
-      "grad_norm": 2.0479118824005127,
+      "grad_norm": 0.5349214673042297,
       "learning_rate": 0.00013860465116279072,
-      "loss": 0.3297,
+      "loss": 0.2837,
       "step": 660
     },
     {
       "epoch": 3.116279069767442,
-      "grad_norm": 1.3133976459503174,
+      "grad_norm": 2.4129347801208496,
       "learning_rate": 0.00013767441860465117,
-      "loss": 0.2098,
+      "loss": 0.2495,
       "step": 670
     },
     {
       "epoch": 3.1627906976744184,
-      "grad_norm": 0.7082284092903137,
+      "grad_norm": 2.0734763145446777,
       "learning_rate": 0.00013674418604651162,
-      "loss": 0.4151,
+      "loss": 0.3585,
       "step": 680
     },
     {
       "epoch": 3.2093023255813953,
-      "grad_norm": 3.3736181259155273,
+      "grad_norm": 4.114805221557617,
       "learning_rate": 0.0001358139534883721,
-      "loss": 0.2073,
+      "loss": 0.2906,
       "step": 690
     },
     {
       "epoch": 3.255813953488372,
-      "grad_norm": 7.461119174957275,
+      "grad_norm": 4.394787788391113,
       "learning_rate": 0.00013488372093023256,
-      "loss": 0.3446,
+      "loss": 0.4101,
       "step": 700
     },
     {
       "epoch": 3.255813953488372,
-      "eval_accuracy": 0.8696369636963697,
-      "eval_loss": 0.42761871218681335,
-      "eval_runtime": 8.0701,
-      "eval_samples_per_second": 75.092,
-      "eval_steps_per_second": 9.417,
+      "eval_accuracy": 0.8729372937293729,
+      "eval_loss": 0.4385151267051697,
+      "eval_runtime": 8.1298,
+      "eval_samples_per_second": 74.541,
+      "eval_steps_per_second": 9.348,
       "step": 700
     },
     {
       "epoch": 3.302325581395349,
-      "grad_norm": 1.2921802997589111,
+      "grad_norm": 2.861743450164795,
       "learning_rate": 0.00013395348837209303,
-      "loss": 0.3886,
+      "loss": 0.445,
       "step": 710
     },
     {
       "epoch": 3.3488372093023258,
-      "grad_norm": 2.9924850463867188,
+      "grad_norm": 4.178903579711914,
       "learning_rate": 0.0001330232558139535,
-      "loss": 0.2564,
+      "loss": 0.219,
       "step": 720
     },
     {
       "epoch": 3.395348837209302,
-      "grad_norm": 3.2593166828155518,
+      "grad_norm": 4.89815616607666,
       "learning_rate": 0.00013209302325581395,
-      "loss": 0.404,
+      "loss": 0.463,
       "step": 730
     },
     {
       "epoch": 3.441860465116279,
-      "grad_norm": 2.4605319499969482,
+      "grad_norm": 10.707399368286133,
       "learning_rate": 0.00013116279069767442,
-      "loss": 0.5092,
+      "loss": 0.5219,
       "step": 740
     },
     {
       "epoch": 3.488372093023256,
-      "grad_norm": 2.1411373615264893,
+      "grad_norm": 3.0130255222320557,
       "learning_rate": 0.0001302325581395349,
-      "loss": 0.3141,
+      "loss": 0.4429,
       "step": 750
     },
     {
       "epoch": 3.5348837209302326,
-      "grad_norm": 3.7920267581939697,
+      "grad_norm": 5.012279510498047,
       "learning_rate": 0.00012930232558139534,
-      "loss": 0.2392,
+      "loss": 0.2735,
       "step": 760
     },
     {
       "epoch": 3.5813953488372094,
-      "grad_norm": 2.0424487590789795,
+      "grad_norm": 2.2448291778564453,
       "learning_rate": 0.00012837209302325582,
-      "loss": 0.3916,
+      "loss": 0.4101,
       "step": 770
     },
     {
       "epoch": 3.6279069767441863,
-      "grad_norm": 3.782304286956787,
+      "grad_norm": 3.8598289489746094,
       "learning_rate": 0.0001274418604651163,
-      "loss": 0.2629,
+      "loss": 0.298,
       "step": 780
     },
     {
       "epoch": 3.6744186046511627,
-      "grad_norm": 1.529003381729126,
+      "grad_norm": 1.7304565906524658,
       "learning_rate": 0.00012651162790697676,
-      "loss": 0.3023,
+      "loss": 0.2853,
       "step": 790
     },
     {
       "epoch": 3.7209302325581395,
-      "grad_norm": 5.351986885070801,
+      "grad_norm": 3.275845527648926,
       "learning_rate": 0.0001255813953488372,
-      "loss": 0.3347,
+      "loss": 0.283,
       "step": 800
     },
     {
       "epoch": 3.7209302325581395,
-      "eval_accuracy": 0.8795379537953796,
-      "eval_loss": 0.4259110689163208,
-      "eval_runtime": 8.618,
-      "eval_samples_per_second": 70.318,
-      "eval_steps_per_second": 8.819,
+      "eval_accuracy": 0.8762376237623762,
+      "eval_loss": 0.4413387179374695,
+      "eval_runtime": 8.5192,
+      "eval_samples_per_second": 71.133,
+      "eval_steps_per_second": 8.921,
       "step": 800
     },
     {
       "epoch": 3.7674418604651163,
-      "grad_norm": 4.942080974578857,
+      "grad_norm": 5.508560657501221,
       "learning_rate": 0.00012465116279069768,
-      "loss": 0.2278,
+      "loss": 0.1962,
       "step": 810
     },
     {
       "epoch": 3.813953488372093,
-      "grad_norm": 2.3388428688049316,
+      "grad_norm": 3.444455862045288,
       "learning_rate": 0.00012372093023255815,
-      "loss": 0.1785,
+      "loss": 0.2718,
       "step": 820
     },
     {
       "epoch": 3.8604651162790695,
-      "grad_norm": 2.0570669174194336,
+      "grad_norm": 10.332758903503418,
       "learning_rate": 0.00012279069767441863,
-      "loss": 0.3932,
+      "loss": 0.4886,
       "step": 830
     },
     {
       "epoch": 3.9069767441860463,
-      "grad_norm": 5.025542736053467,
+      "grad_norm": 1.1791740655899048,
       "learning_rate": 0.00012186046511627907,
-      "loss": 0.3624,
+      "loss": 0.3607,
       "step": 840
     },
     {
       "epoch": 3.953488372093023,
-      "grad_norm": 2.5320193767547607,
+      "grad_norm": 1.7059251070022583,
       "learning_rate": 0.00012093023255813953,
-      "loss": 0.2873,
+      "loss": 0.2679,
       "step": 850
     },
     {
       "epoch": 4.0,
-      "grad_norm": 7.442944526672363,
+      "grad_norm": 6.822539806365967,
       "learning_rate": 0.00012,
-      "loss": 0.3133,
+      "loss": 0.3047,
       "step": 860
     },
     {
       "epoch": 4.046511627906977,
-      "grad_norm": 5.330275535583496,
+      "grad_norm": 3.6658146381378174,
       "learning_rate": 0.00011906976744186048,
-      "loss": 0.2818,
+      "loss": 0.2808,
       "step": 870
     },
     {
       "epoch": 4.093023255813954,
-      "grad_norm": 4.320654392242432,
+      "grad_norm": 2.8236451148986816,
       "learning_rate": 0.00011813953488372094,
-      "loss": 0.4261,
+      "loss": 0.3813,
       "step": 880
     },
     {
       "epoch": 4.1395348837209305,
-      "grad_norm": 3.309112310409546,
+      "grad_norm": 6.614228248596191,
       "learning_rate": 0.00011720930232558141,
-      "loss": 0.2859,
+      "loss": 0.3335,
       "step": 890
     },
     {
       "epoch": 4.186046511627907,
-      "grad_norm": 2.0252270698547363,
+      "grad_norm": 4.810386657714844,
       "learning_rate": 0.00011627906976744187,
-      "loss": 0.3868,
+      "loss": 0.3959,
       "step": 900
     },
     {
       "epoch": 4.186046511627907,
-      "eval_accuracy": 0.8547854785478548,
-      "eval_loss": 0.46416693925857544,
-      "eval_runtime": 8.3175,
-      "eval_samples_per_second": 72.859,
-      "eval_steps_per_second": 9.137,
+      "eval_accuracy": 0.8729372937293729,
+      "eval_loss": 0.41208329796791077,
+      "eval_runtime": 8.3929,
+      "eval_samples_per_second": 72.204,
+      "eval_steps_per_second": 9.055,
       "step": 900
     },
     {
       "epoch": 4.232558139534884,
-      "grad_norm": 0.6425607204437256,
+      "grad_norm": 0.597582221031189,
       "learning_rate": 0.00011534883720930234,
-      "loss": 0.224,
+      "loss": 0.2209,
       "step": 910
     },
     {
       "epoch": 4.27906976744186,
-      "grad_norm": 1.0721712112426758,
+      "grad_norm": 2.4885311126708984,
       "learning_rate": 0.00011441860465116279,
-      "loss": 0.2792,
+      "loss": 0.3125,
       "step": 920
     },
     {
       "epoch": 4.325581395348837,
-      "grad_norm": 3.4592926502227783,
+      "grad_norm": 3.7162835597991943,
       "learning_rate": 0.00011348837209302326,
-      "loss": 0.1881,
+      "loss": 0.1821,
       "step": 930
     },
     {
       "epoch": 4.372093023255814,
-      "grad_norm": 3.121467351913452,
+      "grad_norm": 4.518478870391846,
       "learning_rate": 0.00011255813953488372,
-      "loss": 0.2873,
+      "loss": 0.181,
       "step": 940
     },
     {
       "epoch": 4.4186046511627906,
-      "grad_norm": 1.8312969207763672,
+      "grad_norm": 1.7456213235855103,
       "learning_rate": 0.00011162790697674419,
-      "loss": 0.2052,
+      "loss": 0.2535,
       "step": 950
     },
     {
       "epoch": 4.465116279069767,
-      "grad_norm": 2.4395313262939453,
+      "grad_norm": 2.0503828525543213,
       "learning_rate": 0.00011069767441860466,
-      "loss": 0.2376,
+      "loss": 0.2924,
       "step": 960
     },
     {
       "epoch": 4.511627906976744,
-      "grad_norm": 4.029181480407715,
+      "grad_norm": 0.33591389656066895,
       "learning_rate": 0.00010976744186046512,
-      "loss": 0.2401,
+      "loss": 0.2453,
       "step": 970
     },
     {
       "epoch": 4.558139534883721,
-      "grad_norm": 5.82197904586792,
+      "grad_norm": 2.924938917160034,
       "learning_rate": 0.0001088372093023256,
-      "loss": 0.3262,
+      "loss": 0.2443,
       "step": 980
     },
     {
       "epoch": 4.604651162790698,
-      "grad_norm": 3.295616388320923,
+      "grad_norm": 4.228425979614258,
       "learning_rate": 0.00010790697674418607,
-      "loss": 0.233,
+      "loss": 0.2446,
       "step": 990
     },
     {
       "epoch": 4.651162790697675,
-      "grad_norm": 4.5378899574279785,
+      "grad_norm": 0.3494820296764374,
       "learning_rate": 0.00010697674418604651,
-      "loss": 0.36,
+      "loss": 0.318,
       "step": 1000
     },
     {
       "epoch": 4.651162790697675,
       "eval_accuracy": 0.8696369636963697,
-      "eval_loss": 0.4241912066936493,
-      "eval_runtime": 11.6827,
-      "eval_samples_per_second": 51.872,
-      "eval_steps_per_second": 6.505,
+      "eval_loss": 0.4397449195384979,
+      "eval_runtime": 8.4348,
+      "eval_samples_per_second": 71.845,
+      "eval_steps_per_second": 9.01,
       "step": 1000
     },
     {
       "epoch": 4.6976744186046515,
-      "grad_norm": 2.949002504348755,
+      "grad_norm": 4.108438491821289,
       "learning_rate": 0.00010604651162790697,
-      "loss": 0.1716,
+      "loss": 0.1346,
       "step": 1010
     },
     {
       "epoch": 4.7441860465116275,
-      "grad_norm": 3.2645139694213867,
+      "grad_norm": 2.6504056453704834,
       "learning_rate": 0.00010511627906976745,
-      "loss": 0.2992,
+      "loss": 0.3125,
       "step": 1020
     },
     {
       "epoch": 4.790697674418604,
-      "grad_norm": 0.3114562928676605,
+      "grad_norm": 1.8026741743087769,
       "learning_rate": 0.0001041860465116279,
-      "loss": 0.268,
+      "loss": 0.2085,
       "step": 1030
     },
     {
       "epoch": 4.837209302325581,
-      "grad_norm": 1.4417003393173218,
+      "grad_norm": 2.016584873199463,
       "learning_rate": 0.00010325581395348838,
-      "loss": 0.3601,
+      "loss": 0.3598,
       "step": 1040
     },
     {
       "epoch": 4.883720930232558,
-      "grad_norm": 2.7006003856658936,
+      "grad_norm": 7.666717529296875,
       "learning_rate": 0.00010232558139534885,
-      "loss": 0.335,
+      "loss": 0.3086,
       "step": 1050
     },
     {
       "epoch": 4.930232558139535,
-      "grad_norm": 2.294205665588379,
+      "grad_norm": 0.9514933228492737,
       "learning_rate": 0.00010139534883720931,
-      "loss": 0.2171,
+      "loss": 0.2095,
       "step": 1060
     },
     {
       "epoch": 4.976744186046512,
-      "grad_norm": 4.027862071990967,
+      "grad_norm": 3.582228183746338,
       "learning_rate": 0.00010046511627906978,
-      "loss": 0.3075,
+      "loss": 0.3226,
       "step": 1070
     },
     {
       "epoch": 5.023255813953488,
-      "grad_norm": 2.2804319858551025,
+      "grad_norm": 1.5236479043960571,
       "learning_rate": 9.953488372093024e-05,
-      "loss": 0.2058,
+      "loss": 0.2046,
       "step": 1080
     },
     {
       "epoch": 5.069767441860465,
-      "grad_norm": 3.193336009979248,
+      "grad_norm": 4.655299663543701,
       "learning_rate": 9.86046511627907e-05,
-      "loss": 0.1606,
+      "loss": 0.1163,
       "step": 1090
     },
     {
       "epoch": 5.116279069767442,
-      "grad_norm": 1.5344799757003784,
+      "grad_norm": 2.3776028156280518,
       "learning_rate": 9.767441860465116e-05,
-      "loss": 0.295,
+      "loss": 0.2401,
       "step": 1100
     },
     {
       "epoch": 5.116279069767442,
-      "eval_accuracy": 0.8811881188118812,
-      "eval_loss": 0.4204191267490387,
-      "eval_runtime": 7.9394,
-      "eval_samples_per_second": 76.329,
-      "eval_steps_per_second": 9.573,
+      "eval_accuracy": 0.8679867986798679,
+      "eval_loss": 0.4887414872646332,
+      "eval_runtime": 8.1962,
+      "eval_samples_per_second": 73.937,
+      "eval_steps_per_second": 9.273,
       "step": 1100
     },
     {
       "epoch": 5.162790697674419,
-      "grad_norm": 2.80682110786438,
+      "grad_norm": 4.612281322479248,
       "learning_rate": 9.674418604651163e-05,
-      "loss": 0.2268,
+      "loss": 0.3254,
       "step": 1110
     },
     {
       "epoch": 5.209302325581396,
-      "grad_norm": 2.198103904724121,
+      "grad_norm": 1.2196846008300781,
       "learning_rate": 9.58139534883721e-05,
-      "loss": 0.2226,
+      "loss": 0.2799,
       "step": 1120
     },
     {
       "epoch": 5.2558139534883725,
-      "grad_norm": 3.6225826740264893,
+      "grad_norm": 3.100369691848755,
       "learning_rate": 9.488372093023256e-05,
-      "loss": 0.1485,
+      "loss": 0.1837,
       "step": 1130
     },
     {
       "epoch": 5.3023255813953485,
-      "grad_norm": 1.2400827407836914,
+      "grad_norm": 1.2027745246887207,
       "learning_rate": 9.395348837209302e-05,
-      "loss": 0.1442,
+      "loss": 0.1463,
       "step": 1140
     },
     {
       "epoch": 5.348837209302325,
-      "grad_norm": 1.0701277256011963,
+      "grad_norm": 4.4534196853637695,
       "learning_rate": 9.30232558139535e-05,
-      "loss": 0.2154,
+      "loss": 0.205,
       "step": 1150
     },
     {
       "epoch": 5.395348837209302,
-      "grad_norm": 1.8416105508804321,
+      "grad_norm": 1.2410222291946411,
       "learning_rate": 9.209302325581396e-05,
-      "loss": 0.2288,
+      "loss": 0.2523,
       "step": 1160
     },
     {
       "epoch": 5.441860465116279,
-      "grad_norm": 5.05810546875,
+      "grad_norm": 7.617187023162842,
       "learning_rate": 9.116279069767443e-05,
-      "loss": 0.3002,
+      "loss": 0.2619,
       "step": 1170
     },
     {
       "epoch": 5.488372093023256,
-      "grad_norm": 2.076577663421631,
+      "grad_norm": 0.7941145300865173,
       "learning_rate": 9.023255813953489e-05,
-      "loss": 0.1431,
+      "loss": 0.3047,
       "step": 1180
     },
     {
       "epoch": 5.534883720930233,
-      "grad_norm": 1.4769999980926514,
+      "grad_norm": 2.606306552886963,
       "learning_rate": 8.930232558139535e-05,
-      "loss": 0.2355,
+      "loss": 0.226,
       "step": 1190
     },
     {
       "epoch": 5.5813953488372094,
-      "grad_norm": 4.101127624511719,
+      "grad_norm": 9.533924102783203,
       "learning_rate": 8.837209302325582e-05,
-      "loss": 0.2342,
+      "loss": 0.1273,
       "step": 1200
     },
     {
       "epoch": 5.5813953488372094,
-      "eval_accuracy": 0.8910891089108911,
-      "eval_loss": 0.3933403491973877,
-      "eval_runtime": 8.1318,
-      "eval_samples_per_second": 74.523,
-      "eval_steps_per_second": 9.346,
+      "eval_accuracy": 0.8663366336633663,
+      "eval_loss": 0.4223584532737732,
+      "eval_runtime": 8.4233,
+      "eval_samples_per_second": 71.943,
+      "eval_steps_per_second": 9.023,
       "step": 1200
     },
     {
       "epoch": 5.627906976744186,
-      "grad_norm": 3.6835029125213623,
+      "grad_norm": 1.8951793909072876,
       "learning_rate": 8.744186046511629e-05,
-      "loss": 0.221,
+      "loss": 0.1706,
       "step": 1210
     },
     {
       "epoch": 5.674418604651163,
-      "grad_norm": 5.327223300933838,
+      "grad_norm": 1.8302172422409058,
       "learning_rate": 8.651162790697674e-05,
-      "loss": 0.2887,
+      "loss": 0.2261,
       "step": 1220
     },
     {
       "epoch": 5.720930232558139,
-      "grad_norm": 1.522955060005188,
+      "grad_norm": 1.3308892250061035,
       "learning_rate": 8.558139534883721e-05,
-      "loss": 0.2057,
+      "loss": 0.2022,
       "step": 1230
     },
     {
       "epoch": 5.767441860465116,
-      "grad_norm": 5.1004133224487305,
+      "grad_norm": 0.5113635659217834,
       "learning_rate": 8.465116279069768e-05,
-      "loss": 0.2433,
+      "loss": 0.2085,
       "step": 1240
     },
     {
       "epoch": 5.813953488372093,
-      "grad_norm": 1.7744640111923218,
+      "grad_norm": 5.370401382446289,
       "learning_rate": 8.372093023255814e-05,
-      "loss": 0.1493,
+      "loss": 0.2813,
       "step": 1250
     },
     {
       "epoch": 5.8604651162790695,
-      "grad_norm": 4.374938488006592,
+      "grad_norm": 1.0463091135025024,
       "learning_rate": 8.27906976744186e-05,
-      "loss": 0.3513,
+      "loss": 0.3038,
       "step": 1260
     },
     {
       "epoch": 5.906976744186046,
-      "grad_norm": 2.1579113006591797,
+      "grad_norm": 2.4071524143218994,
       "learning_rate": 8.186046511627907e-05,
-      "loss": 0.1253,
+      "loss": 0.1519,
       "step": 1270
     },
     {
       "epoch": 5.953488372093023,
-      "grad_norm": 3.5076637268066406,
+      "grad_norm": 6.68901252746582,
       "learning_rate": 8.093023255813953e-05,
-      "loss": 0.2042,
+      "loss": 0.1952,
       "step": 1280
     },
     {
       "epoch": 6.0,
-      "grad_norm": 2.294401168823242,
+      "grad_norm": 5.0761542320251465,
       "learning_rate": 8e-05,
-      "loss": 0.3253,
+      "loss": 0.2148,
       "step": 1290
     },
     {
       "epoch": 6.046511627906977,
-      "grad_norm": 0.5659641027450562,
+      "grad_norm": 4.299339771270752,
       "learning_rate": 7.906976744186047e-05,
-      "loss": 0.1629,
+      "loss": 0.1101,
       "step": 1300
     },
     {
       "epoch": 6.046511627906977,
-      "eval_accuracy": 0.8976897689768977,
-      "eval_loss": 0.36343997716903687,
-      "eval_runtime": 8.0696,
-      "eval_samples_per_second": 75.096,
-      "eval_steps_per_second": 9.418,
+      "eval_accuracy": 0.8778877887788779,
+      "eval_loss": 0.43775683641433716,
+      "eval_runtime": 8.4309,
+      "eval_samples_per_second": 71.878,
+      "eval_steps_per_second": 9.014,
       "step": 1300
     },
     {
       "epoch": 6.093023255813954,
-      "grad_norm": 3.126523971557617,
+      "grad_norm": 5.793342590332031,
       "learning_rate": 7.813953488372094e-05,
-      "loss": 0.1177,
+      "loss": 0.1908,
       "step": 1310
     },
     {
       "epoch": 6.1395348837209305,
-      "grad_norm": 7.47089958190918,
+      "grad_norm": 3.164515256881714,
       "learning_rate": 7.72093023255814e-05,
-      "loss": 0.2036,
+      "loss": 0.1572,
       "step": 1320
     },
     {
       "epoch": 6.186046511627907,
-      "grad_norm": 1.5586128234863281,
+      "grad_norm": 1.032468557357788,
       "learning_rate": 7.627906976744187e-05,
-      "loss": 0.1308,
+      "loss": 0.2154,
       "step": 1330
     },
     {
       "epoch": 6.232558139534884,
-      "grad_norm": 2.230147361755371,
+      "grad_norm": 3.3498730659484863,
       "learning_rate": 7.534883720930233e-05,
-      "loss": 0.2078,
+      "loss": 0.1473,
       "step": 1340
     },
     {
       "epoch": 6.27906976744186,
-      "grad_norm": 4.571086883544922,
+      "grad_norm": 1.0370675325393677,
       "learning_rate": 7.441860465116279e-05,
-      "loss": 0.131,
+      "loss": 0.1568,
       "step": 1350
     },
     {
       "epoch": 6.325581395348837,
-      "grad_norm": 0.7467085719108582,
+      "grad_norm": 1.5336437225341797,
       "learning_rate": 7.348837209302326e-05,
-      "loss": 0.1465,
+      "loss": 0.1943,
       "step": 1360
     },
     {
       "epoch": 6.372093023255814,
-      "grad_norm": 1.2842942476272583,
+      "grad_norm": 7.077569007873535,
       "learning_rate": 7.255813953488373e-05,
-      "loss": 0.1553,
+      "loss": 0.242,
       "step": 1370
     },
     {
       "epoch": 6.4186046511627906,
-      "grad_norm": 0.23809537291526794,
+      "grad_norm": 6.786858558654785,
       "learning_rate": 7.162790697674418e-05,
-      "loss": 0.1314,
+      "loss": 0.2174,
       "step": 1380
     },
     {
       "epoch": 6.465116279069767,
-      "grad_norm": 8.466991424560547,
+      "grad_norm": 2.3031513690948486,
       "learning_rate": 7.069767441860465e-05,
-      "loss": 0.2773,
+      "loss": 0.2683,
       "step": 1390
     },
     {
       "epoch": 6.511627906976744,
-      "grad_norm": 11.282208442687988,
+      "grad_norm": 4.499554634094238,
       "learning_rate": 6.976744186046513e-05,
-      "loss": 0.2041,
+      "loss": 0.1773,
       "step": 1400
     },
     {
       "epoch": 6.511627906976744,
-      "eval_accuracy": 0.8910891089108911,
-      "eval_loss": 0.4006832540035248,
-      "eval_runtime": 10.4521,
-      "eval_samples_per_second": 57.979,
-      "eval_steps_per_second": 7.271,
+      "eval_accuracy": 0.8844884488448845,
+      "eval_loss": 0.3730297386646271,
+      "eval_runtime": 8.1564,
+      "eval_samples_per_second": 74.297,
+      "eval_steps_per_second": 9.318,
       "step": 1400
     },
     {
       "epoch": 6.558139534883721,
-      "grad_norm": 0.22673094272613525,
+      "grad_norm": 0.8953260779380798,
       "learning_rate": 6.883720930232558e-05,
-      "loss": 0.1841,
+      "loss": 0.1669,
       "step": 1410
     },
     {
       "epoch": 6.604651162790698,
-      "grad_norm": 1.8003208637237549,
+      "grad_norm": 1.8238012790679932,
       "learning_rate": 6.790697674418604e-05,
-      "loss": 0.2147,
+      "loss": 0.2064,
       "step": 1420
     },
     {
       "epoch": 6.651162790697675,
-      "grad_norm": 3.2758359909057617,
+      "grad_norm": 0.47919437289237976,
       "learning_rate": 6.697674418604652e-05,
-      "loss": 0.1884,
+      "loss": 0.1294,
       "step": 1430
     },
     {
       "epoch": 6.6976744186046515,
-      "grad_norm": 2.0551984310150146,
+      "grad_norm": 1.9631860256195068,
       "learning_rate": 6.604651162790698e-05,
-      "loss": 0.2099,
+      "loss": 0.2931,
       "step": 1440
     },
     {
       "epoch": 6.7441860465116275,
-      "grad_norm": 0.6447792053222656,
+      "grad_norm": 2.078273057937622,
       "learning_rate": 6.511627906976745e-05,
-      "loss": 0.114,
+      "loss": 0.0749,
       "step": 1450
     },
     {
       "epoch": 6.790697674418604,
-      "grad_norm": 0.8234413862228394,
+      "grad_norm": 8.262022972106934,
       "learning_rate": 6.418604651162791e-05,
-      "loss": 0.141,
+      "loss": 0.1879,
       "step": 1460
     },
     {
       "epoch": 6.837209302325581,
-      "grad_norm": 4.1033406257629395,
+      "grad_norm": 8.179067611694336,
       "learning_rate": 6.325581395348838e-05,
-      "loss": 0.1895,
+      "loss": 0.1836,
       "step": 1470
     },
     {
       "epoch": 6.883720930232558,
-      "grad_norm": 3.889683246612549,
+      "grad_norm": 3.35481858253479,
       "learning_rate": 6.232558139534884e-05,
-      "loss": 0.1317,
+      "loss": 0.1719,
       "step": 1480
     },
     {
       "epoch": 6.930232558139535,
-      "grad_norm": 4.772053241729736,
+      "grad_norm": 5.25211763381958,
       "learning_rate": 6.139534883720931e-05,
-      "loss": 0.2148,
+      "loss": 0.1532,
       "step": 1490
     },
     {
       "epoch": 6.976744186046512,
-      "grad_norm": 3.037203788757324,
+      "grad_norm": 4.454288959503174,
       "learning_rate": 6.0465116279069765e-05,
-      "loss": 0.1668,
+      "loss": 0.2248,
       "step": 1500
     },
     {
       "epoch": 6.976744186046512,
-      "eval_accuracy": 0.8927392739273927,
-      "eval_loss": 0.3843345642089844,
-      "eval_runtime": 8.3737,
-      "eval_samples_per_second": 72.37,
-      "eval_steps_per_second": 9.076,
+      "eval_accuracy": 0.8861386138613861,
+      "eval_loss": 0.37258049845695496,
+      "eval_runtime": 8.4482,
+      "eval_samples_per_second": 71.731,
+      "eval_steps_per_second": 8.996,
       "step": 1500
     },
     {
       "epoch": 7.023255813953488,
-      "grad_norm": 4.356062889099121,
+      "grad_norm": 0.325836181640625,
       "learning_rate": 5.953488372093024e-05,
-      "loss": 0.2231,
+      "loss": 0.2717,
       "step": 1510
     },
     {
       "epoch": 7.069767441860465,
-      "grad_norm": 3.485504388809204,
+      "grad_norm": 0.20686429738998413,
       "learning_rate": 5.8604651162790704e-05,
-      "loss": 0.1391,
+      "loss": 0.1134,
       "step": 1520
     },
     {
       "epoch": 7.116279069767442,
-      "grad_norm": 1.0620895624160767,
+      "grad_norm": 1.2898919582366943,
       "learning_rate": 5.767441860465117e-05,
-      "loss": 0.1369,
+      "loss": 0.1802,
       "step": 1530
     },
     {
       "epoch": 7.162790697674419,
-      "grad_norm": 0.46343836188316345,
+      "grad_norm": 0.5109810829162598,
       "learning_rate": 5.674418604651163e-05,
-      "loss": 0.1639,
+      "loss": 0.1167,
       "step": 1540
     },
     {
       "epoch": 7.209302325581396,
-      "grad_norm": 5.317925453186035,
+      "grad_norm": 2.8214943408966064,
       "learning_rate": 5.5813953488372095e-05,
-      "loss": 0.1061,
+      "loss": 0.0916,
       "step": 1550
     },
     {
       "epoch": 7.2558139534883725,
-      "grad_norm": 2.8796050548553467,
+      "grad_norm": 4.599625587463379,
       "learning_rate": 5.488372093023256e-05,
-      "loss": 0.164,
+      "loss": 0.1704,
       "step": 1560
     },
     {
       "epoch": 7.3023255813953485,
-      "grad_norm": 2.1574888229370117,
+      "grad_norm": 3.6765389442443848,
       "learning_rate": 5.3953488372093034e-05,
-      "loss": 0.0833,
+      "loss": 0.0917,
       "step": 1570
     },
     {
       "epoch": 7.348837209302325,
-      "grad_norm": 4.602242469787598,
+      "grad_norm": 1.5945552587509155,
       "learning_rate": 5.3023255813953486e-05,
-      "loss": 0.1002,
+      "loss": 0.0956,
       "step": 1580
     },
     {
       "epoch": 7.395348837209302,
-      "grad_norm": 1.2774103879928589,
+      "grad_norm": 6.084174156188965,
       "learning_rate": 5.209302325581395e-05,
-      "loss": 0.0654,
+      "loss": 0.1411,
       "step": 1590
     },
     {
       "epoch": 7.441860465116279,
-      "grad_norm": 0.37218332290649414,
+      "grad_norm": 2.9608631134033203,
       "learning_rate": 5.1162790697674425e-05,
-      "loss": 0.0976,
+      "loss": 0.0987,
       "step": 1600
     },
     {
       "epoch": 7.441860465116279,
-      "eval_accuracy": 0.8927392739273927,
-      "eval_loss": 0.4061937630176544,
-      "eval_runtime": 8.5468,
-      "eval_samples_per_second": 70.904,
-      "eval_steps_per_second": 8.892,
+      "eval_accuracy": 0.8844884488448845,
+      "eval_loss": 0.43984663486480713,
+      "eval_runtime": 8.3551,
+      "eval_samples_per_second": 72.531,
+      "eval_steps_per_second": 9.096,
       "step": 1600
     },
     {
       "epoch": 7.488372093023256,
-      "grad_norm": 1.125712275505066,
+      "grad_norm": 1.1511733531951904,
       "learning_rate": 5.023255813953489e-05,
-      "loss": 0.1522,
+      "loss": 0.1976,
       "step": 1610
     },
     {
       "epoch": 7.534883720930233,
-      "grad_norm": 3.255833625793457,
+      "grad_norm": 1.5226008892059326,
       "learning_rate": 4.930232558139535e-05,
-      "loss": 0.154,
+      "loss": 0.1001,
       "step": 1620
     },
     {
       "epoch": 7.5813953488372094,
-      "grad_norm": 3.5783629417419434,
+      "grad_norm": 3.7124974727630615,
       "learning_rate": 4.8372093023255816e-05,
-      "loss": 0.1385,
+      "loss": 0.1432,
       "step": 1630
     },
     {
       "epoch": 7.627906976744186,
-      "grad_norm": 4.361517429351807,
+      "grad_norm": 1.3083992004394531,
       "learning_rate": 4.744186046511628e-05,
-      "loss": 0.1105,
+      "loss": 0.1514,
       "step": 1640
     },
     {
       "epoch": 7.674418604651163,
-      "grad_norm": 1.0743229389190674,
+      "grad_norm": 1.7809404134750366,
       "learning_rate": 4.651162790697675e-05,
-      "loss": 0.1461,
+      "loss": 0.0881,
       "step": 1650
     },
     {
       "epoch": 7.720930232558139,
-      "grad_norm": 1.155990481376648,
+      "grad_norm": 0.19210846722126007,
       "learning_rate": 4.5581395348837214e-05,
-      "loss": 0.1247,
+      "loss": 0.1179,
       "step": 1660
     },
     {
       "epoch": 7.767441860465116,
-      "grad_norm": 0.03133929520845413,
+      "grad_norm": 0.44379866123199463,
       "learning_rate": 4.465116279069767e-05,
-      "loss": 0.0625,
+      "loss": 0.1302,
       "step": 1670
     },
     {
       "epoch": 7.813953488372093,
-      "grad_norm": 1.1082956790924072,
+      "grad_norm": 0.10616659373044968,
       "learning_rate": 4.3720930232558146e-05,
-      "loss": 0.2145,
+      "loss": 0.1636,
       "step": 1680
     },
     {
       "epoch": 7.8604651162790695,
-      "grad_norm": 2.6899712085723877,
+      "grad_norm": 2.069657325744629,
       "learning_rate": 4.2790697674418605e-05,
-      "loss": 0.1182,
+      "loss": 0.157,
       "step": 1690
     },
     {
       "epoch": 7.906976744186046,
-      "grad_norm": 5.237185955047607,
+      "grad_norm": 4.515770435333252,
       "learning_rate": 4.186046511627907e-05,
-      "loss": 0.1275,
+      "loss": 0.16,
       "step": 1700
     },
     {
       "epoch": 7.906976744186046,
-      "eval_accuracy": 0.8894389438943895,
-      "eval_loss": 0.3860972821712494,
-      "eval_runtime": 8.3425,
-      "eval_samples_per_second": 72.64,
-      "eval_steps_per_second": 9.11,
+      "eval_accuracy": 0.8828382838283828,
+      "eval_loss": 0.4170949161052704,
+      "eval_runtime": 8.1214,
+      "eval_samples_per_second": 74.617,
+      "eval_steps_per_second": 9.358,
       "step": 1700
     },
     {
       "epoch": 7.953488372093023,
-      "grad_norm": 1.522539496421814,
+      "grad_norm": 8.669586181640625,
       "learning_rate": 4.093023255813954e-05,
-      "loss": 0.1341,
+      "loss": 0.1765,
       "step": 1710
     },
     {
       "epoch": 8.0,
-      "grad_norm": 0.16910937428474426,
+      "grad_norm": 5.801586627960205,
       "learning_rate": 4e-05,
-      "loss": 0.1398,
+      "loss": 0.1434,
       "step": 1720
     },
     {
       "epoch": 8.046511627906977,
-      "grad_norm": 3.6039626598358154,
+      "grad_norm": 0.6714385151863098,
       "learning_rate": 3.906976744186047e-05,
-      "loss": 0.089,
+      "loss": 0.0739,
       "step": 1730
     },
     {
       "epoch": 8.093023255813954,
-      "grad_norm": 3.6135542392730713,
+      "grad_norm": 1.2774052619934082,
       "learning_rate": 3.8139534883720935e-05,
-      "loss": 0.1332,
+      "loss": 0.0654,
       "step": 1740
     },
     {
       "epoch": 8.13953488372093,
-      "grad_norm": 2.710392713546753,
+      "grad_norm": 0.1617097109556198,
       "learning_rate": 3.7209302325581394e-05,
-      "loss": 0.1241,
+      "loss": 0.1071,
       "step": 1750
     },
     {
       "epoch": 8.186046511627907,
-      "grad_norm": 5.7339067459106445,
+      "grad_norm": 8.03242301940918,
       "learning_rate": 3.627906976744187e-05,
-      "loss": 0.1269,
+      "loss": 0.1191,
       "step": 1760
     },
     {
       "epoch": 8.232558139534884,
-      "grad_norm": 1.990607738494873,
+      "grad_norm": 0.6172130107879639,
       "learning_rate": 3.5348837209302326e-05,
-      "loss": 0.1111,
+      "loss": 0.109,
       "step": 1770
     },
     {
       "epoch": 8.279069767441861,
-      "grad_norm": 0.09675276279449463,
+      "grad_norm": 0.7166829109191895,
       "learning_rate": 3.441860465116279e-05,
-      "loss": 0.116,
+      "loss": 0.0863,
       "step": 1780
     },
     {
       "epoch": 8.325581395348838,
-      "grad_norm": 0.04001321643590927,
+      "grad_norm": 0.0343417190015316,
       "learning_rate": 3.348837209302326e-05,
-      "loss": 0.1184,
+      "loss": 0.1092,
       "step": 1790
     },
     {
       "epoch": 8.372093023255815,
-      "grad_norm": 0.477737694978714,
+      "grad_norm": 0.09605341404676437,
       "learning_rate": 3.2558139534883724e-05,
-      "loss": 0.1063,
+      "loss": 0.1224,
       "step": 1800
     },
     {
       "epoch": 8.372093023255815,
-      "eval_accuracy": 0.8910891089108911,
-      "eval_loss": 0.40107399225234985,
-      "eval_runtime": 8.3287,
-      "eval_samples_per_second": 72.761,
-      "eval_steps_per_second": 9.125,
+      "eval_accuracy": 0.8877887788778878,
+      "eval_loss": 0.4335933327674866,
+      "eval_runtime": 8.4644,
+      "eval_samples_per_second": 71.594,
+      "eval_steps_per_second": 8.979,
       "step": 1800
     },
     {
       "epoch": 8.418604651162791,
-      "grad_norm": 0.2791198790073395,
+      "grad_norm": 0.1179109588265419,
       "learning_rate": 3.162790697674419e-05,
-      "loss": 0.1073,
+      "loss": 0.1358,
       "step": 1810
     },
     {
       "epoch": 8.465116279069768,
-      "grad_norm": 0.03783993795514107,
+      "grad_norm": 0.0823044553399086,
       "learning_rate": 3.0697674418604656e-05,
-      "loss": 0.1584,
+      "loss": 0.1565,
       "step": 1820
     },
     {
       "epoch": 8.511627906976745,
-      "grad_norm": 0.7814832925796509,
+      "grad_norm": 0.6940491795539856,
       "learning_rate": 2.976744186046512e-05,
-      "loss": 0.0507,
+      "loss": 0.0398,
       "step": 1830
     },
     {
       "epoch": 8.55813953488372,
-      "grad_norm": 0.22078533470630646,
+      "grad_norm": 0.12108628451824188,
       "learning_rate": 2.8837209302325585e-05,
-      "loss": 0.0539,
+      "loss": 0.0299,
       "step": 1840
     },
     {
       "epoch": 8.604651162790697,
-      "grad_norm": 2.293942928314209,
+      "grad_norm": 2.229219913482666,
       "learning_rate": 2.7906976744186048e-05,
-      "loss": 0.1488,
+      "loss": 0.1867,
       "step": 1850
     },
     {
       "epoch": 8.651162790697674,
-      "grad_norm": 0.0680939108133316,
+      "grad_norm": 4.451650619506836,
       "learning_rate": 2.6976744186046517e-05,
-      "loss": 0.071,
+      "loss": 0.1431,
       "step": 1860
     },
     {
       "epoch": 8.69767441860465,
-      "grad_norm": 2.966674327850342,
+      "grad_norm": 4.897462844848633,
       "learning_rate": 2.6046511627906976e-05,
-      "loss": 0.1212,
+      "loss": 0.1542,
       "step": 1870
     },
     {
       "epoch": 8.744186046511627,
-      "grad_norm": 0.7327284216880798,
+      "grad_norm": 0.9782124161720276,
       "learning_rate": 2.5116279069767445e-05,
-      "loss": 0.0865,
+      "loss": 0.1044,
       "step": 1880
     },
     {
       "epoch": 8.790697674418604,
-      "grad_norm": 2.343684673309326,
+      "grad_norm": 2.251777410507202,
       "learning_rate": 2.4186046511627908e-05,
-      "loss": 0.1467,
+      "loss": 0.1508,
       "step": 1890
     },
     {
       "epoch": 8.837209302325581,
-      "grad_norm": 5.241259574890137,
+      "grad_norm": 4.513321399688721,
       "learning_rate": 2.3255813953488374e-05,
-      "loss": 0.1658,
+      "loss": 0.2111,
       "step": 1900
     },
     {
       "epoch": 8.837209302325581,
-      "eval_accuracy": 0.9042904290429042,
-      "eval_loss": 0.3840368390083313,
-      "eval_runtime": 8.2921,
-      "eval_samples_per_second": 73.081,
-      "eval_steps_per_second": 9.165,
+      "eval_accuracy": 0.8943894389438944,
+      "eval_loss": 0.394785612821579,
+      "eval_runtime": 8.2576,
+      "eval_samples_per_second": 73.387,
+      "eval_steps_per_second": 9.204,
       "step": 1900
     },
     {
       "epoch": 8.883720930232558,
-      "grad_norm": 0.7707066535949707,
+      "grad_norm": 2.1677255630493164,
       "learning_rate": 2.2325581395348837e-05,
-      "loss": 0.1274,
+      "loss": 0.1205,
       "step": 1910
     },
     {
       "epoch": 8.930232558139535,
-      "grad_norm": 2.1124603748321533,
+      "grad_norm": 3.6949923038482666,
       "learning_rate": 2.1395348837209303e-05,
-      "loss": 0.1397,
+      "loss": 0.1626,
       "step": 1920
     },
     {
       "epoch": 8.976744186046512,
-      "grad_norm": 0.4913456439971924,
+      "grad_norm": 4.727644920349121,
       "learning_rate": 2.046511627906977e-05,
-      "loss": 0.1697,
+      "loss": 0.1635,
       "step": 1930
     },
     {
       "epoch": 9.023255813953488,
-      "grad_norm": 1.9249141216278076,
+      "grad_norm": 2.22031569480896,
       "learning_rate": 1.9534883720930235e-05,
-      "loss": 0.139,
+      "loss": 0.1543,
       "step": 1940
     },
     {
       "epoch": 9.069767441860465,
-      "grad_norm": 1.6021732091903687,
+      "grad_norm": 1.0797746181488037,
       "learning_rate": 1.8604651162790697e-05,
-      "loss": 0.1806,
+      "loss": 0.2055,
       "step": 1950
     },
     {
       "epoch": 9.116279069767442,
-      "grad_norm": 0.47562333941459656,
+      "grad_norm": 0.9525020122528076,
       "learning_rate": 1.7674418604651163e-05,
-      "loss": 0.1272,
+      "loss": 0.1099,
       "step": 1960
     },
     {
       "epoch": 9.162790697674419,
-      "grad_norm": 4.387588024139404,
+      "grad_norm": 0.37071776390075684,
       "learning_rate": 1.674418604651163e-05,
-      "loss": 0.1565,
+      "loss": 0.1583,
       "step": 1970
     },
     {
       "epoch": 9.209302325581396,
-      "grad_norm": 0.03679787367582321,
+      "grad_norm": 0.05000032112002373,
       "learning_rate": 1.5813953488372095e-05,
-      "loss": 0.039,
+      "loss": 0.0378,
       "step": 1980
     },
     {
       "epoch": 9.255813953488373,
-      "grad_norm": 0.04448888823390007,
+      "grad_norm": 0.06881304085254669,
       "learning_rate": 1.488372093023256e-05,
-      "loss": 0.084,
+      "loss": 0.1168,
       "step": 1990
     },
     {
       "epoch": 9.30232558139535,
-      "grad_norm": 0.5206286311149597,
+      "grad_norm": 3.6886019706726074,
       "learning_rate": 1.3953488372093024e-05,
-      "loss": 0.1,
+      "loss": 0.112,
       "step": 2000
     },
     {
       "epoch": 9.30232558139535,
-      "eval_accuracy": 0.900990099009901,
-      "eval_loss": 0.3872542083263397,
-      "eval_runtime": 8.0752,
-      "eval_samples_per_second": 75.044,
-      "eval_steps_per_second": 9.411,
+      "eval_accuracy": 0.8943894389438944,
+      "eval_loss": 0.4003974199295044,
+      "eval_runtime": 8.4662,
+      "eval_samples_per_second": 71.579,
+      "eval_steps_per_second": 8.977,
       "step": 2000
     },
     {
       "epoch": 9.348837209302326,
-      "grad_norm": 0.08173186331987381,
+      "grad_norm": 0.3669143617153168,
       "learning_rate": 1.3023255813953488e-05,
-      "loss": 0.1023,
+      "loss": 0.0836,
       "step": 2010
     },
     {
       "epoch": 9.395348837209303,
-      "grad_norm": 0.45058727264404297,
+      "grad_norm": 5.295791149139404,
       "learning_rate": 1.2093023255813954e-05,
-      "loss": 0.1194,
+      "loss": 0.0932,
       "step": 2020
     },
     {
       "epoch": 9.44186046511628,
-      "grad_norm": 0.05082235485315323,
+      "grad_norm": 0.12259897589683533,
       "learning_rate": 1.1162790697674418e-05,
-      "loss": 0.0994,
+      "loss": 0.1198,
       "step": 2030
     },
     {
       "epoch": 9.488372093023255,
-      "grad_norm": 0.16433951258659363,
+      "grad_norm": 0.18641342222690582,
       "learning_rate": 1.0232558139534884e-05,
-      "loss": 0.124,
+      "loss": 0.1342,
       "step": 2040
     },
     {
       "epoch": 9.534883720930232,
-      "grad_norm": 3.3077304363250732,
+      "grad_norm": 2.212017774581909,
       "learning_rate": 9.302325581395349e-06,
-      "loss": 0.1311,
+      "loss": 0.1336,
       "step": 2050
     },
     {
       "epoch": 9.581395348837209,
-      "grad_norm": 3.5823140144348145,
+      "grad_norm": 0.13683325052261353,
       "learning_rate": 8.372093023255815e-06,
-      "loss": 0.0295,
+      "loss": 0.046,
       "step": 2060
     },
     {
       "epoch": 9.627906976744185,
-      "grad_norm": 0.6963232159614563,
+      "grad_norm": 0.07736368477344513,
       "learning_rate": 7.44186046511628e-06,
-      "loss": 0.0753,
+      "loss": 0.0623,
       "step": 2070
     },
     {
       "epoch": 9.674418604651162,
-      "grad_norm": 0.858066201210022,
+      "grad_norm": 1.0054152011871338,
       "learning_rate": 6.511627906976744e-06,
-      "loss": 0.0945,
+      "loss": 0.069,
       "step": 2080
     },
     {
       "epoch": 9.720930232558139,
-      "grad_norm": 0.16633135080337524,
+      "grad_norm": 0.7521716356277466,
       "learning_rate": 5.581395348837209e-06,
-      "loss": 0.1215,
+      "loss": 0.153,
       "step": 2090
     },
     {
       "epoch": 9.767441860465116,
-      "grad_norm": 1.5643835067749023,
+      "grad_norm": 2.2922515869140625,
       "learning_rate": 4.651162790697674e-06,
-      "loss": 0.1045,
+      "loss": 0.0962,
       "step": 2100
     },
     {
       "epoch": 9.767441860465116,
-      "eval_accuracy": 0.9075907590759076,
-      "eval_loss": 0.37872469425201416,
-      "eval_runtime": 8.0071,
-      "eval_samples_per_second": 75.683,
-      "eval_steps_per_second": 9.492,
+      "eval_accuracy": 0.8927392739273927,
+      "eval_loss": 0.4092388451099396,
+      "eval_runtime": 8.2945,
+      "eval_samples_per_second": 73.061,
+      "eval_steps_per_second": 9.163,
       "step": 2100
     },
     {
       "epoch": 9.813953488372093,
-      "grad_norm": 2.010878324508667,
+      "grad_norm": 2.177103042602539,
       "learning_rate": 3.72093023255814e-06,
-      "loss": 0.0887,
+      "loss": 0.1215,
       "step": 2110
     },
     {
       "epoch": 9.86046511627907,
-      "grad_norm": 2.305905342102051,
+      "grad_norm": 3.3745920658111572,
       "learning_rate": 2.7906976744186046e-06,
-      "loss": 0.1151,
+      "loss": 0.129,
       "step": 2120
     },
     {
       "epoch": 9.906976744186046,
-      "grad_norm": 2.8573191165924072,
+      "grad_norm": 0.3872903883457184,
       "learning_rate": 1.86046511627907e-06,
-      "loss": 0.1443,
+      "loss": 0.0951,
       "step": 2130
     },
     {
       "epoch": 9.953488372093023,
-      "grad_norm": 3.0255908966064453,
+      "grad_norm": 0.13965001702308655,
       "learning_rate": 9.30232558139535e-07,
-      "loss": 0.1765,
+      "loss": 0.1625,
       "step": 2140
     },
     {
       "epoch": 10.0,
-      "grad_norm": 2.159597635269165,
+      "grad_norm": 0.7976939082145691,
       "learning_rate": 0.0,
-      "loss": 0.1705,
+      "loss": 0.1355,
       "step": 2150
     },
     {
       "epoch": 10.0,
       "step": 2150,
       "total_flos": 2.659620429501235e+18,
-      "train_loss": 0.3051174351919529,
-      "train_runtime": 1549.0078,
-      "train_samples_per_second": 22.156,
-      "train_steps_per_second": 1.388
+      "train_loss": 0.30670992596204893,
+      "train_runtime": 1546.6686,
+      "train_samples_per_second": 22.19,
+      "train_steps_per_second": 1.39
     }
   ],
   "logging_steps": 10,