🍻 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
 datasets:
 - renovation
@@ -14,7 +15,7 @@ model-index:
       name: Image Classification
       type: image-classification
     dataset:
-      name: renovation
+      name: renovations
       type: renovation
       config: default
       split: validation
@@ -22,7 +23,7 @@ model-index:
     metrics:
     - name: Accuracy
       type: accuracy
-      value: 0.6863636363636364
+      value: 0.6545454545454545
 ---
 
 <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -30,10 +31,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # vit-base-renovation
 
-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 renovation 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 renovations dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.9761
-- Accuracy: 0.6864
+- Loss: 0.7622
+- Accuracy: 0.6545
 
 ## Model description
 

all_results.json

@@ -1,13 +1,13 @@
 {
     "epoch": 4.0,
-    "eval_accuracy": 0.6909090909090909,
-    "eval_loss": 0.710116982460022,
-    "eval_runtime": 7.3292,
-    "eval_samples_per_second": 30.017,
-    "eval_steps_per_second": 3.82,
+    "eval_accuracy": 0.6545454545454545,
+    "eval_loss": 0.7622188329696655,
+    "eval_runtime": 7.8953,
+    "eval_samples_per_second": 27.865,
+    "eval_steps_per_second": 3.546,
     "total_flos": 6.137420584742093e+17,
-    "train_loss": 0.3987359667737638,
-    "train_runtime": 862.5404,
-    "train_samples_per_second": 9.182,
-    "train_steps_per_second": 0.575
+    "train_loss": 0.4817413741181935,
+    "train_runtime": 646.4292,
+    "train_samples_per_second": 12.252,
+    "train_steps_per_second": 0.767
 }

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
-    "eval_accuracy": 0.6909090909090909,
-    "eval_loss": 0.710116982460022,
-    "eval_runtime": 7.3292,
-    "eval_samples_per_second": 30.017,
-    "eval_steps_per_second": 3.82
+    "eval_accuracy": 0.6545454545454545,
+    "eval_loss": 0.7622188329696655,
+    "eval_runtime": 7.8953,
+    "eval_samples_per_second": 27.865,
+    "eval_steps_per_second": 3.546
 }

runs/Mar18_14-26-58_7e4cb5ba27a4/events.out.tfevents.1710772695.7e4cb5ba27a4.347.1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ae68c6bd7d5c4fee51e11597a552d9143a351674c988fe3215c62b6af3f73b3
+size 411

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
     "total_flos": 6.137420584742093e+17,
-    "train_loss": 0.3987359667737638,
-    "train_runtime": 862.5404,
-    "train_samples_per_second": 9.182,
-    "train_steps_per_second": 0.575
+    "train_loss": 0.4817413741181935,
+    "train_runtime": 646.4292,
+    "train_samples_per_second": 12.252,
+    "train_steps_per_second": 0.767
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.710116982460022,
+  "best_metric": 0.7622188329696655,
   "best_model_checkpoint": "./vit-base-renovation/checkpoint-250",
   "epoch": 4.0,
   "eval_steps": 25,
@@ -10,526 +10,526 @@
   "log_history": [
     {
       "epoch": 0.08,
-      "grad_norm": 1.7348554134368896,
+      "grad_norm": 1.4599734544754028,
       "learning_rate": 0.00019596774193548388,
-      "loss": 1.02,
+      "loss": 0.9874,
       "step": 10
     },
     {
       "epoch": 0.16,
-      "grad_norm": 0.9956100583076477,
+      "grad_norm": 1.6008720397949219,
       "learning_rate": 0.00019193548387096775,
-      "loss": 1.0036,
+      "loss": 0.9737,
       "step": 20
     },
     {
       "epoch": 0.2,
-      "eval_accuracy": 0.5,
-      "eval_loss": 0.9849038124084473,
-      "eval_runtime": 8.3244,
-      "eval_samples_per_second": 26.428,
-      "eval_steps_per_second": 3.364,
+      "eval_accuracy": 0.5045454545454545,
+      "eval_loss": 1.0076260566711426,
+      "eval_runtime": 8.1182,
+      "eval_samples_per_second": 27.1,
+      "eval_steps_per_second": 3.449,
       "step": 25
     },
     {
       "epoch": 0.24,
-      "grad_norm": 1.1454524993896484,
+      "grad_norm": 1.4715768098831177,
       "learning_rate": 0.00018790322580645164,
-      "loss": 0.9917,
+      "loss": 1.0314,
       "step": 30
     },
     {
       "epoch": 0.32,
-      "grad_norm": 2.6789638996124268,
+      "grad_norm": 3.563518762588501,
       "learning_rate": 0.00018387096774193548,
-      "loss": 0.8835,
+      "loss": 0.9846,
       "step": 40
     },
     {
       "epoch": 0.4,
-      "grad_norm": 2.0885393619537354,
+      "grad_norm": 2.6104962825775146,
       "learning_rate": 0.00017983870967741935,
-      "loss": 0.8051,
+      "loss": 0.862,
       "step": 50
     },
     {
       "epoch": 0.4,
-      "eval_accuracy": 0.5545454545454546,
-      "eval_loss": 0.9105865359306335,
-      "eval_runtime": 8.5151,
-      "eval_samples_per_second": 25.836,
-      "eval_steps_per_second": 3.288,
+      "eval_accuracy": 0.5045454545454545,
+      "eval_loss": 1.0219964981079102,
+      "eval_runtime": 8.0163,
+      "eval_samples_per_second": 27.444,
+      "eval_steps_per_second": 3.493,
       "step": 50
     },
     {
       "epoch": 0.48,
-      "grad_norm": 1.2426408529281616,
+      "grad_norm": 1.1494791507720947,
       "learning_rate": 0.00017580645161290325,
-      "loss": 0.7545,
+      "loss": 0.7918,
       "step": 60
     },
     {
       "epoch": 0.56,
-      "grad_norm": 3.0708746910095215,
+      "grad_norm": 2.8406925201416016,
       "learning_rate": 0.00017177419354838711,
-      "loss": 0.8336,
+      "loss": 0.9064,
       "step": 70
     },
     {
       "epoch": 0.6,
-      "eval_accuracy": 0.5954545454545455,
-      "eval_loss": 0.9004097580909729,
-      "eval_runtime": 8.3678,
-      "eval_samples_per_second": 26.291,
-      "eval_steps_per_second": 3.346,
+      "eval_accuracy": 0.5590909090909091,
+      "eval_loss": 0.9076008200645447,
+      "eval_runtime": 7.2828,
+      "eval_samples_per_second": 30.208,
+      "eval_steps_per_second": 3.845,
       "step": 75
     },
     {
       "epoch": 0.65,
-      "grad_norm": 1.7403985261917114,
-      "learning_rate": 0.0001681451612903226,
-      "loss": 0.8614,
+      "grad_norm": 1.1309199333190918,
+      "learning_rate": 0.00016774193548387098,
+      "loss": 0.944,
       "step": 80
     },
     {
       "epoch": 0.73,
-      "grad_norm": 1.7942191362380981,
-      "learning_rate": 0.00016411290322580646,
-      "loss": 0.8141,
+      "grad_norm": 1.6623650789260864,
+      "learning_rate": 0.00016370967741935485,
+      "loss": 0.865,
       "step": 90
     },
     {
       "epoch": 0.81,
-      "grad_norm": 2.56789231300354,
-      "learning_rate": 0.00016008064516129033,
-      "loss": 0.786,
+      "grad_norm": 2.2675681114196777,
+      "learning_rate": 0.00015967741935483872,
+      "loss": 0.8528,
       "step": 100
     },
     {
       "epoch": 0.81,
-      "eval_accuracy": 0.6454545454545455,
-      "eval_loss": 0.7700986266136169,
-      "eval_runtime": 8.4023,
-      "eval_samples_per_second": 26.183,
-      "eval_steps_per_second": 3.332,
+      "eval_accuracy": 0.65,
+      "eval_loss": 0.8157132267951965,
+      "eval_runtime": 7.8906,
+      "eval_samples_per_second": 27.881,
+      "eval_steps_per_second": 3.549,
       "step": 100
     },
     {
       "epoch": 0.89,
-      "grad_norm": 2.265244722366333,
-      "learning_rate": 0.0001560483870967742,
-      "loss": 0.7118,
+      "grad_norm": 2.036137819290161,
+      "learning_rate": 0.0001556451612903226,
+      "loss": 0.7899,
       "step": 110
     },
     {
       "epoch": 0.97,
-      "grad_norm": 2.001025915145874,
-      "learning_rate": 0.0001520161290322581,
-      "loss": 0.7854,
+      "grad_norm": 1.0894314050674438,
+      "learning_rate": 0.00015161290322580646,
+      "loss": 0.8848,
       "step": 120
     },
     {
       "epoch": 1.01,
-      "eval_accuracy": 0.6227272727272727,
-      "eval_loss": 0.7561001777648926,
-      "eval_runtime": 8.1474,
-      "eval_samples_per_second": 27.002,
-      "eval_steps_per_second": 3.437,
+      "eval_accuracy": 0.6272727272727273,
+      "eval_loss": 0.8089344501495361,
+      "eval_runtime": 7.2662,
+      "eval_samples_per_second": 30.277,
+      "eval_steps_per_second": 3.853,
       "step": 125
     },
     {
       "epoch": 1.05,
-      "grad_norm": 1.243610143661499,
-      "learning_rate": 0.00014798387096774193,
-      "loss": 0.5415,
+      "grad_norm": 1.2654739618301392,
+      "learning_rate": 0.00014758064516129032,
+      "loss": 0.6029,
       "step": 130
     },
     {
       "epoch": 1.13,
-      "grad_norm": 1.9543741941452026,
-      "learning_rate": 0.0001439516129032258,
-      "loss": 0.5246,
+      "grad_norm": 2.1379916667938232,
+      "learning_rate": 0.00014354838709677422,
+      "loss": 0.6292,
       "step": 140
     },
     {
       "epoch": 1.21,
-      "grad_norm": 1.4787832498550415,
-      "learning_rate": 0.00013991935483870967,
-      "loss": 0.4603,
+      "grad_norm": 3.7936744689941406,
+      "learning_rate": 0.0001395161290322581,
+      "loss": 0.6608,
       "step": 150
     },
     {
       "epoch": 1.21,
       "eval_accuracy": 0.6409090909090909,
-      "eval_loss": 0.8104726672172546,
-      "eval_runtime": 7.4392,
-      "eval_samples_per_second": 29.573,
-      "eval_steps_per_second": 3.764,
+      "eval_loss": 0.8614596128463745,
+      "eval_runtime": 8.173,
+      "eval_samples_per_second": 26.918,
+      "eval_steps_per_second": 3.426,
       "step": 150
     },
     {
       "epoch": 1.29,
-      "grad_norm": 1.9429636001586914,
-      "learning_rate": 0.00013588709677419357,
-      "loss": 0.442,
+      "grad_norm": 1.3360966444015503,
+      "learning_rate": 0.00013548387096774193,
+      "loss": 0.539,
       "step": 160
     },
     {
       "epoch": 1.37,
-      "grad_norm": 2.558549642562866,
-      "learning_rate": 0.00013185483870967743,
-      "loss": 0.4934,
+      "grad_norm": 4.338820934295654,
+      "learning_rate": 0.0001314516129032258,
+      "loss": 0.6748,
       "step": 170
     },
     {
       "epoch": 1.41,
-      "eval_accuracy": 0.6181818181818182,
-      "eval_loss": 0.874618411064148,
-      "eval_runtime": 7.375,
-      "eval_samples_per_second": 29.831,
-      "eval_steps_per_second": 3.797,
+      "eval_accuracy": 0.6318181818181818,
+      "eval_loss": 0.8425928354263306,
+      "eval_runtime": 7.2996,
+      "eval_samples_per_second": 30.139,
+      "eval_steps_per_second": 3.836,
       "step": 175
     },
     {
       "epoch": 1.45,
-      "grad_norm": 3.0746328830718994,
-      "learning_rate": 0.0001278225806451613,
-      "loss": 0.5744,
+      "grad_norm": 2.611848831176758,
+      "learning_rate": 0.0001274193548387097,
+      "loss": 0.6317,
       "step": 180
     },
     {
       "epoch": 1.53,
-      "grad_norm": 2.0470433235168457,
+      "grad_norm": 2.4912662506103516,
       "learning_rate": 0.00012379032258064514,
-      "loss": 0.5933,
+      "loss": 0.6935,
       "step": 190
     },
     {
       "epoch": 1.61,
-      "grad_norm": 2.0045182704925537,
+      "grad_norm": 2.0737533569335938,
       "learning_rate": 0.00011975806451612903,
-      "loss": 0.5315,
+      "loss": 0.6559,
       "step": 200
     },
     {
       "epoch": 1.61,
-      "eval_accuracy": 0.6636363636363637,
-      "eval_loss": 0.8266905546188354,
-      "eval_runtime": 7.745,
-      "eval_samples_per_second": 28.405,
-      "eval_steps_per_second": 3.615,
+      "eval_accuracy": 0.6090909090909091,
+      "eval_loss": 0.8427345752716064,
+      "eval_runtime": 8.531,
+      "eval_samples_per_second": 25.788,
+      "eval_steps_per_second": 3.282,
       "step": 200
     },
     {
       "epoch": 1.69,
-      "grad_norm": 2.0231032371520996,
+      "grad_norm": 2.8581531047821045,
       "learning_rate": 0.00011572580645161291,
-      "loss": 0.5538,
+      "loss": 0.5985,
       "step": 210
     },
     {
       "epoch": 1.77,
-      "grad_norm": 2.176417589187622,
+      "grad_norm": 1.7107830047607422,
       "learning_rate": 0.00011169354838709678,
-      "loss": 0.5251,
+      "loss": 0.5654,
       "step": 220
     },
     {
       "epoch": 1.81,
-      "eval_accuracy": 0.65,
-      "eval_loss": 0.8585435152053833,
-      "eval_runtime": 7.3957,
-      "eval_samples_per_second": 29.747,
-      "eval_steps_per_second": 3.786,
+      "eval_accuracy": 0.6681818181818182,
+      "eval_loss": 0.8266812562942505,
+      "eval_runtime": 7.3084,
+      "eval_samples_per_second": 30.102,
+      "eval_steps_per_second": 3.831,
       "step": 225
     },
     {
       "epoch": 1.85,
-      "grad_norm": 3.1730520725250244,
-      "learning_rate": 0.00010806451612903225,
-      "loss": 0.5472,
+      "grad_norm": 2.7065391540527344,
+      "learning_rate": 0.00010766129032258066,
+      "loss": 0.6948,
       "step": 230
     },
     {
       "epoch": 1.94,
-      "grad_norm": 2.2670953273773193,
-      "learning_rate": 0.00010403225806451614,
-      "loss": 0.5488,
+      "grad_norm": 1.9006155729293823,
+      "learning_rate": 0.00010362903225806453,
+      "loss": 0.5896,
       "step": 240
     },
     {
       "epoch": 2.02,
-      "grad_norm": 0.9752326607704163,
-      "learning_rate": 0.0001,
-      "loss": 0.4386,
+      "grad_norm": 1.5013632774353027,
+      "learning_rate": 9.95967741935484e-05,
+      "loss": 0.5254,
       "step": 250
     },
     {
       "epoch": 2.02,
-      "eval_accuracy": 0.6909090909090909,
-      "eval_loss": 0.710116982460022,
-      "eval_runtime": 8.4962,
-      "eval_samples_per_second": 25.894,
-      "eval_steps_per_second": 3.296,
+      "eval_accuracy": 0.6545454545454545,
+      "eval_loss": 0.7622188329696655,
+      "eval_runtime": 8.0247,
+      "eval_samples_per_second": 27.415,
+      "eval_steps_per_second": 3.489,
       "step": 250
     },
     {
       "epoch": 2.1,
-      "grad_norm": 0.6374533176422119,
-      "learning_rate": 9.596774193548387e-05,
-      "loss": 0.241,
+      "grad_norm": 3.849762439727783,
+      "learning_rate": 9.556451612903226e-05,
+      "loss": 0.3218,
       "step": 260
     },
     {
       "epoch": 2.18,
-      "grad_norm": 0.9005341529846191,
-      "learning_rate": 9.193548387096774e-05,
-      "loss": 0.2627,
+      "grad_norm": 1.2341018915176392,
+      "learning_rate": 9.153225806451613e-05,
+      "loss": 0.2778,
       "step": 270
     },
     {
       "epoch": 2.22,
-      "eval_accuracy": 0.6409090909090909,
-      "eval_loss": 1.0041698217391968,
-      "eval_runtime": 7.4719,
-      "eval_samples_per_second": 29.443,
-      "eval_steps_per_second": 3.747,
+      "eval_accuracy": 0.6636363636363637,
+      "eval_loss": 0.9480502009391785,
+      "eval_runtime": 7.5019,
+      "eval_samples_per_second": 29.326,
+      "eval_steps_per_second": 3.732,
       "step": 275
     },
     {
       "epoch": 2.26,
-      "grad_norm": 2.1455209255218506,
-      "learning_rate": 8.790322580645162e-05,
-      "loss": 0.1868,
+      "grad_norm": 2.990760087966919,
+      "learning_rate": 8.75e-05,
+      "loss": 0.3018,
       "step": 280
     },
     {
       "epoch": 2.34,
-      "grad_norm": 0.3361818790435791,
-      "learning_rate": 8.387096774193549e-05,
-      "loss": 0.1521,
+      "grad_norm": 3.7092368602752686,
+      "learning_rate": 8.346774193548388e-05,
+      "loss": 0.3696,
       "step": 290
     },
     {
       "epoch": 2.42,
-      "grad_norm": 5.28161096572876,
-      "learning_rate": 7.983870967741936e-05,
-      "loss": 0.1524,
+      "grad_norm": 4.312985420227051,
+      "learning_rate": 7.943548387096774e-05,
+      "loss": 0.309,
       "step": 300
     },
     {
       "epoch": 2.42,
-      "eval_accuracy": 0.6545454545454545,
-      "eval_loss": 0.9489237666130066,
-      "eval_runtime": 8.3095,
-      "eval_samples_per_second": 26.476,
-      "eval_steps_per_second": 3.37,
+      "eval_accuracy": 0.6409090909090909,
+      "eval_loss": 0.8998208045959473,
+      "eval_runtime": 8.0465,
+      "eval_samples_per_second": 27.341,
+      "eval_steps_per_second": 3.48,
       "step": 300
     },
     {
       "epoch": 2.5,
-      "grad_norm": 2.760798215866089,
-      "learning_rate": 7.580645161290323e-05,
-      "loss": 0.1971,
+      "grad_norm": 3.9428765773773193,
+      "learning_rate": 7.540322580645162e-05,
+      "loss": 0.4113,
       "step": 310
     },
     {
       "epoch": 2.58,
-      "grad_norm": 1.529565453529358,
-      "learning_rate": 7.177419354838711e-05,
-      "loss": 0.1272,
+      "grad_norm": 1.6300395727157593,
+      "learning_rate": 7.137096774193549e-05,
+      "loss": 0.2396,
       "step": 320
     },
     {
       "epoch": 2.62,
-      "eval_accuracy": 0.65,
-      "eval_loss": 1.066292405128479,
-      "eval_runtime": 8.3186,
-      "eval_samples_per_second": 26.447,
-      "eval_steps_per_second": 3.366,
+      "eval_accuracy": 0.6409090909090909,
+      "eval_loss": 0.9171214699745178,
+      "eval_runtime": 7.6722,
+      "eval_samples_per_second": 28.675,
+      "eval_steps_per_second": 3.65,
       "step": 325
     },
     {
       "epoch": 2.66,
-      "grad_norm": 0.5179287791252136,
-      "learning_rate": 6.774193548387096e-05,
-      "loss": 0.1676,
+      "grad_norm": 1.3225387334823608,
+      "learning_rate": 6.733870967741935e-05,
+      "loss": 0.3451,
       "step": 330
     },
     {
       "epoch": 2.74,
-      "grad_norm": 1.027869701385498,
-      "learning_rate": 6.370967741935485e-05,
-      "loss": 0.1194,
+      "grad_norm": 2.227588176727295,
+      "learning_rate": 6.330645161290322e-05,
+      "loss": 0.206,
       "step": 340
     },
     {
       "epoch": 2.82,
-      "grad_norm": 1.2000805139541626,
-      "learning_rate": 5.9677419354838715e-05,
-      "loss": 0.186,
+      "grad_norm": 3.282959222793579,
+      "learning_rate": 5.9274193548387104e-05,
+      "loss": 0.2773,
       "step": 350
     },
     {
       "epoch": 2.82,
-      "eval_accuracy": 0.6545454545454545,
-      "eval_loss": 1.0830607414245605,
-      "eval_runtime": 8.4011,
-      "eval_samples_per_second": 26.187,
-      "eval_steps_per_second": 3.333,
+      "eval_accuracy": 0.6090909090909091,
+      "eval_loss": 1.0582060813903809,
+      "eval_runtime": 7.9345,
+      "eval_samples_per_second": 27.727,
+      "eval_steps_per_second": 3.529,
       "step": 350
     },
     {
       "epoch": 2.9,
-      "grad_norm": 3.6447389125823975,
-      "learning_rate": 5.5645161290322576e-05,
-      "loss": 0.2194,
+      "grad_norm": 4.272079944610596,
+      "learning_rate": 5.5241935483870966e-05,
+      "loss": 0.4049,
       "step": 360
     },
     {
       "epoch": 2.98,
-      "grad_norm": 4.36700439453125,
-      "learning_rate": 5.161290322580645e-05,
-      "loss": 0.1544,
+      "grad_norm": 4.28529167175293,
+      "learning_rate": 5.120967741935484e-05,
+      "loss": 0.2516,
       "step": 370
     },
     {
       "epoch": 3.02,
-      "eval_accuracy": 0.6363636363636364,
-      "eval_loss": 1.115280032157898,
-      "eval_runtime": 7.5342,
-      "eval_samples_per_second": 29.2,
-      "eval_steps_per_second": 3.716,
+      "eval_accuracy": 0.6454545454545455,
+      "eval_loss": 0.9362202286720276,
+      "eval_runtime": 7.9239,
+      "eval_samples_per_second": 27.764,
+      "eval_steps_per_second": 3.534,
       "step": 375
     },
     {
       "epoch": 3.06,
-      "grad_norm": 0.32941731810569763,
-      "learning_rate": 4.7580645161290326e-05,
-      "loss": 0.0438,
+      "grad_norm": 0.6417772173881531,
+      "learning_rate": 4.7177419354838716e-05,
+      "loss": 0.1439,
       "step": 380
     },
     {
       "epoch": 3.15,
-      "grad_norm": 0.40441280603408813,
-      "learning_rate": 4.3548387096774194e-05,
-      "loss": 0.0432,
+      "grad_norm": 0.6277107000350952,
+      "learning_rate": 4.3145161290322584e-05,
+      "loss": 0.1053,
       "step": 390
     },
     {
       "epoch": 3.23,
-      "grad_norm": 0.3993894159793854,
-      "learning_rate": 3.951612903225806e-05,
-      "loss": 0.0803,
+      "grad_norm": 4.842502593994141,
+      "learning_rate": 3.911290322580645e-05,
+      "loss": 0.1578,
       "step": 400
     },
     {
       "epoch": 3.23,
-      "eval_accuracy": 0.6409090909090909,
-      "eval_loss": 1.039893388748169,
-      "eval_runtime": 7.6391,
-      "eval_samples_per_second": 28.799,
-      "eval_steps_per_second": 3.665,
+      "eval_accuracy": 0.6772727272727272,
+      "eval_loss": 0.9263622760772705,
+      "eval_runtime": 8.037,
+      "eval_samples_per_second": 27.373,
+      "eval_steps_per_second": 3.484,
       "step": 400
     },
     {
       "epoch": 3.31,
-      "grad_norm": 0.0989086702466011,
-      "learning_rate": 3.548387096774194e-05,
-      "loss": 0.0298,
+      "grad_norm": 3.4976587295532227,
+      "learning_rate": 3.508064516129033e-05,
+      "loss": 0.1233,
       "step": 410
     },
     {
       "epoch": 3.39,
-      "grad_norm": 0.30518239736557007,
-      "learning_rate": 3.1451612903225806e-05,
-      "loss": 0.041,
+      "grad_norm": 0.3740604519844055,
+      "learning_rate": 3.1048387096774195e-05,
+      "loss": 0.0979,
       "step": 420
     },
     {
       "epoch": 3.43,
-      "eval_accuracy": 0.6818181818181818,
-      "eval_loss": 1.0910764932632446,
-      "eval_runtime": 7.8435,
-      "eval_samples_per_second": 28.049,
-      "eval_steps_per_second": 3.57,
+      "eval_accuracy": 0.6772727272727272,
+      "eval_loss": 0.946960985660553,
+      "eval_runtime": 7.5496,
+      "eval_samples_per_second": 29.141,
+      "eval_steps_per_second": 3.709,
       "step": 425
     },
     {
       "epoch": 3.47,
-      "grad_norm": 0.383977472782135,
-      "learning_rate": 2.7419354838709678e-05,
-      "loss": 0.0702,
+      "grad_norm": 0.2739836275577545,
+      "learning_rate": 2.7016129032258064e-05,
+      "loss": 0.083,
       "step": 430
     },
     {
       "epoch": 3.55,
-      "grad_norm": 0.12995119392871857,
-      "learning_rate": 2.338709677419355e-05,
-      "loss": 0.0211,
+      "grad_norm": 2.7154436111450195,
+      "learning_rate": 2.2983870967741935e-05,
+      "loss": 0.085,
       "step": 440
     },
     {
       "epoch": 3.63,
-      "grad_norm": 0.09603479504585266,
-      "learning_rate": 1.935483870967742e-05,
-      "loss": 0.0685,
+      "grad_norm": 1.2921925783157349,
+      "learning_rate": 1.8951612903225807e-05,
+      "loss": 0.0836,
       "step": 450
     },
     {
       "epoch": 3.63,
-      "eval_accuracy": 0.6590909090909091,
-      "eval_loss": 1.1889867782592773,
-      "eval_runtime": 7.401,
-      "eval_samples_per_second": 29.726,
-      "eval_steps_per_second": 3.783,
+      "eval_accuracy": 0.6681818181818182,
+      "eval_loss": 0.9940941333770752,
+      "eval_runtime": 8.0024,
+      "eval_samples_per_second": 27.492,
+      "eval_steps_per_second": 3.499,
       "step": 450
     },
     {
       "epoch": 3.71,
-      "grad_norm": 0.13990958034992218,
-      "learning_rate": 1.5322580645161292e-05,
-      "loss": 0.0406,
+      "grad_norm": 0.7346110939979553,
+      "learning_rate": 1.4919354838709679e-05,
+      "loss": 0.0661,
       "step": 460
     },
     {
       "epoch": 3.79,
-      "grad_norm": 0.11549140512943268,
-      "learning_rate": 1.129032258064516e-05,
-      "loss": 0.0475,
+      "grad_norm": 0.540484607219696,
+      "learning_rate": 1.0887096774193549e-05,
+      "loss": 0.126,
       "step": 470
     },
     {
       "epoch": 3.83,
-      "eval_accuracy": 0.6681818181818182,
-      "eval_loss": 1.1724765300750732,
-      "eval_runtime": 8.3023,
-      "eval_samples_per_second": 26.499,
-      "eval_steps_per_second": 3.373,
+      "eval_accuracy": 0.6863636363636364,
+      "eval_loss": 0.9761253595352173,
+      "eval_runtime": 7.4163,
+      "eval_samples_per_second": 29.665,
+      "eval_steps_per_second": 3.775,
       "step": 475
     },
     {
       "epoch": 3.87,
-      "grad_norm": 2.7816131114959717,
-      "learning_rate": 7.258064516129033e-06,
-      "loss": 0.0376,
+      "grad_norm": 1.4092530012130737,
+      "learning_rate": 6.854838709677419e-06,
+      "loss": 0.0657,
       "step": 480
     },
     {
       "epoch": 3.95,
-      "grad_norm": 0.09355681389570236,
+      "grad_norm": 1.824102759361267,
       "learning_rate": 3.225806451612903e-06,
-      "loss": 0.0452,
+      "loss": 0.075,
       "step": 490
     },
     {
       "epoch": 4.0,
       "step": 496,
       "total_flos": 6.137420584742093e+17,
-      "train_loss": 0.3987359667737638,
-      "train_runtime": 862.5404,
-      "train_samples_per_second": 9.182,
-      "train_steps_per_second": 0.575
+      "train_loss": 0.4817413741181935,
+      "train_runtime": 646.4292,
+      "train_samples_per_second": 12.252,
+      "train_steps_per_second": 0.767
     }
   ],
   "logging_steps": 10,