🍻 cheers

README.md

@@ -2,6 +2,7 @@
 license: apache-2.0
 base_model: google/vit-base-patch16-224
 tags:
+- image-classification
 - generated_from_trainer
 metrics:
 - accuracy
@@ -18,13 +19,13 @@ should probably proofread and complete it, then remove this comment. -->
 
 # vit-lr-cosine-restarts
 
-This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on an unknown dataset.
+This model is a fine-tuned version of [google/vit-base-patch16-224](https://huggingface.co/google/vit-base-patch16-224) on the skin-cancer dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.7998
-- Accuracy: 0.8627
-- Precision: 0.8599
-- Recall: 0.8627
-- F1: 0.8595
+- Loss: 0.4929
+- Accuracy: 0.8263
+- Precision: 0.8255
+- Recall: 0.8263
+- F1: 0.8158
 
 ## Model description
 

all_results.json

@@ -1,16 +1,16 @@
 {
     "epoch": 12.0,
-    "eval_accuracy": 0.7978502080443828,
-    "eval_f1": 0.7792929975948731,
-    "eval_loss": 0.5963773131370544,
-    "eval_precision": 0.7969524883183612,
-    "eval_recall": 0.7978502080443828,
-    "eval_runtime": 41.5106,
-    "eval_samples_per_second": 69.476,
-    "eval_steps_per_second": 8.697,
+    "eval_accuracy": 0.8262829403606102,
+    "eval_f1": 0.8158028786365703,
+    "eval_loss": 0.49287769198417664,
+    "eval_precision": 0.8255125137698959,
+    "eval_recall": 0.8262829403606102,
+    "eval_runtime": 38.4698,
+    "eval_samples_per_second": 74.968,
+    "eval_steps_per_second": 9.384,
     "total_flos": 4.768760767819088e+18,
-    "train_loss": 0.30000500961256177,
-    "train_runtime": 2048.5956,
-    "train_samples_per_second": 250.318,
-    "train_steps_per_second": 15.669
+    "train_loss": 0.17886438471413105,
+    "train_runtime": 1744.4117,
+    "train_samples_per_second": 35.276,
+    "train_steps_per_second": 2.208
 }

eval_results.json

@@ -1,11 +1,11 @@
 {
     "epoch": 12.0,
-    "eval_accuracy": 0.7978502080443828,
-    "eval_f1": 0.7792929975948731,
-    "eval_loss": 0.5963773131370544,
-    "eval_precision": 0.7969524883183612,
-    "eval_recall": 0.7978502080443828,
-    "eval_runtime": 41.5106,
-    "eval_samples_per_second": 69.476,
-    "eval_steps_per_second": 8.697
+    "eval_accuracy": 0.8262829403606102,
+    "eval_f1": 0.8158028786365703,
+    "eval_loss": 0.49287769198417664,
+    "eval_precision": 0.8255125137698959,
+    "eval_recall": 0.8262829403606102,
+    "eval_runtime": 38.4698,
+    "eval_samples_per_second": 74.968,
+    "eval_steps_per_second": 9.384
 }

runs/Mar20_16-50-20_f2e1fee5f9b2/events.out.tfevents.1710955250.f2e1fee5f9b2.2315.7

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:977fe1479a673add8384b95bf8a1f06201c7c30753fdcb947388fdf844fa3a2e
+size 560

train_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 12.0,
     "total_flos": 4.768760767819088e+18,
-    "train_loss": 0.30000500961256177,
-    "train_runtime": 2048.5956,
-    "train_samples_per_second": 250.318,
-    "train_steps_per_second": 15.669
+    "train_loss": 0.17886438471413105,
+    "train_runtime": 1744.4117,
+    "train_samples_per_second": 35.276,
+    "train_steps_per_second": 2.208
 }

trainer_state.json

@@ -1,5 +1,5 @@
 {
-  "best_metric": 0.5963773131370544,
+  "best_metric": 0.49287769198417664,
   "best_model_checkpoint": "./vit-lr-cosine-restarts/checkpoint-642",
   "epoch": 12.0,
   "eval_steps": 500,
@@ -10,246 +10,246 @@
   "log_history": [
     {
       "epoch": 1.0,
-      "grad_norm": 4.743273735046387,
-      "learning_rate": 9.999999999999999e-05,
-      "loss": 0.806,
+      "grad_norm": 6.734107971191406,
+      "learning_rate": 4.1428571428571437e-05,
+      "loss": 0.8832,
       "step": 321
     },
     {
       "epoch": 1.0,
-      "eval_accuracy": 0.7465325936199723,
-      "eval_f1": 0.6995266195661654,
-      "eval_loss": 0.7538458108901978,
-      "eval_precision": 0.7318596775013999,
-      "eval_recall": 0.7465325936199723,
-      "eval_runtime": 42.3197,
-      "eval_samples_per_second": 68.148,
-      "eval_steps_per_second": 8.53,
+      "eval_accuracy": 0.7999306518723994,
+      "eval_f1": 0.7874950822007427,
+      "eval_loss": 0.5542528629302979,
+      "eval_precision": 0.8092011027858591,
+      "eval_recall": 0.7999306518723994,
+      "eval_runtime": 37.6681,
+      "eval_samples_per_second": 76.563,
+      "eval_steps_per_second": 9.584,
       "step": 321
     },
     {
       "epoch": 2.0,
-      "grad_norm": 5.603039741516113,
-      "learning_rate": 0.0001585786437626905,
-      "loss": 0.6108,
+      "grad_norm": 6.56775426864624,
+      "learning_rate": 8.311688311688312e-05,
+      "loss": 0.4647,
       "step": 642
     },
     {
       "epoch": 2.0,
-      "eval_accuracy": 0.7978502080443828,
-      "eval_f1": 0.7792929975948731,
-      "eval_loss": 0.5963773131370544,
-      "eval_precision": 0.7969524883183612,
-      "eval_recall": 0.7978502080443828,
-      "eval_runtime": 41.5398,
-      "eval_samples_per_second": 69.427,
-      "eval_steps_per_second": 8.69,
+      "eval_accuracy": 0.8262829403606102,
+      "eval_f1": 0.8158028786365703,
+      "eval_loss": 0.49287769198417664,
+      "eval_precision": 0.8255125137698959,
+      "eval_recall": 0.8262829403606102,
+      "eval_runtime": 37.8104,
+      "eval_samples_per_second": 76.275,
+      "eval_steps_per_second": 9.548,
       "step": 642
     },
     {
       "epoch": 3.0,
-      "grad_norm": 0.9543079137802124,
-      "learning_rate": 0.00030000000000000003,
-      "loss": 0.5038,
+      "grad_norm": 2.852621078491211,
+      "learning_rate": 9.171100544010777e-05,
+      "loss": 0.3465,
       "step": 963
     },
     {
       "epoch": 3.0,
-      "eval_accuracy": 0.7857142857142857,
-      "eval_f1": 0.7634339903359785,
-      "eval_loss": 0.6932518482208252,
-      "eval_precision": 0.7761466356337973,
-      "eval_recall": 0.7857142857142857,
-      "eval_runtime": 41.9279,
-      "eval_samples_per_second": 68.785,
-      "eval_steps_per_second": 8.61,
+      "eval_accuracy": 0.8214285714285714,
+      "eval_f1": 0.8025208154787117,
+      "eval_loss": 0.6518540978431702,
+      "eval_precision": 0.8133669627117774,
+      "eval_recall": 0.8214285714285714,
+      "eval_runtime": 37.046,
+      "eval_samples_per_second": 77.849,
+      "eval_steps_per_second": 9.745,
       "step": 963
     },
     {
       "epoch": 4.0,
-      "grad_norm": 5.373469352722168,
-      "learning_rate": 0.0004414213562373095,
-      "loss": 0.3939,
+      "grad_norm": 9.967479705810547,
+      "learning_rate": 5.025483282385923e-05,
+      "loss": 0.1493,
       "step": 1284
     },
     {
       "epoch": 4.0,
-      "eval_accuracy": 0.8023578363384188,
-      "eval_f1": 0.792141645758032,
-      "eval_loss": 0.6029081344604492,
-      "eval_precision": 0.798347633331751,
-      "eval_recall": 0.8023578363384188,
-      "eval_runtime": 40.6302,
-      "eval_samples_per_second": 70.982,
-      "eval_steps_per_second": 8.885,
+      "eval_accuracy": 0.8401525658807212,
+      "eval_f1": 0.8265360989719667,
+      "eval_loss": 0.6805647611618042,
+      "eval_precision": 0.8385311261178104,
+      "eval_recall": 0.8401525658807212,
+      "eval_runtime": 37.4024,
+      "eval_samples_per_second": 77.107,
+      "eval_steps_per_second": 9.652,
       "step": 1284
     },
     {
       "epoch": 5.0,
-      "grad_norm": 0.7852970361709595,
-      "learning_rate": 0.0004414213562373095,
-      "loss": 0.2961,
+      "grad_norm": 0.2563266158103943,
+      "learning_rate": 8.572203615936175e-06,
+      "loss": 0.0211,
       "step": 1605
     },
     {
       "epoch": 5.0,
-      "eval_accuracy": 0.7874479889042996,
-      "eval_f1": 0.7616828983903079,
-      "eval_loss": 0.7347444891929626,
-      "eval_precision": 0.7657012162041952,
-      "eval_recall": 0.7874479889042996,
-      "eval_runtime": 40.4935,
-      "eval_samples_per_second": 71.221,
-      "eval_steps_per_second": 8.915,
+      "eval_accuracy": 0.8762135922330098,
+      "eval_f1": 0.8742202382058613,
+      "eval_loss": 0.5605077743530273,
+      "eval_precision": 0.8749370661972383,
+      "eval_recall": 0.8762135922330098,
+      "eval_runtime": 37.3835,
+      "eval_samples_per_second": 77.146,
+      "eval_steps_per_second": 9.657,
       "step": 1605
     },
     {
       "epoch": 6.0,
-      "grad_norm": 0.07125339657068253,
-      "learning_rate": 9.999999999999999e-05,
-      "loss": 0.2392,
+      "grad_norm": 0.11774127185344696,
+      "learning_rate": 9.629571142901386e-05,
+      "loss": 0.0084,
       "step": 1926
     },
     {
       "epoch": 6.0,
-      "eval_accuracy": 0.8002773925104022,
-      "eval_f1": 0.8006330723917671,
-      "eval_loss": 0.7591750025749207,
-      "eval_precision": 0.804264341802859,
-      "eval_recall": 0.8002773925104022,
-      "eval_runtime": 40.5444,
-      "eval_samples_per_second": 71.132,
-      "eval_steps_per_second": 8.904,
+      "eval_accuracy": 0.8283633841886269,
+      "eval_f1": 0.8089134462154245,
+      "eval_loss": 0.9771777391433716,
+      "eval_precision": 0.8382740264211974,
+      "eval_recall": 0.8283633841886269,
+      "eval_runtime": 36.767,
+      "eval_samples_per_second": 78.44,
+      "eval_steps_per_second": 9.819,
       "step": 1926
     },
     {
       "epoch": 7.0,
-      "grad_norm": 8.10545539855957,
-      "learning_rate": 0.0001585786437626905,
-      "loss": 0.1848,
+      "grad_norm": 8.762543678283691,
+      "learning_rate": 6.002305101515785e-05,
+      "loss": 0.1058,
       "step": 2247
     },
     {
       "epoch": 7.0,
-      "eval_accuracy": 0.7763522884882108,
-      "eval_f1": 0.7715354351312519,
-      "eval_loss": 0.9079565405845642,
-      "eval_precision": 0.788489029323623,
-      "eval_recall": 0.7763522884882108,
-      "eval_runtime": 40.1034,
-      "eval_samples_per_second": 71.914,
-      "eval_steps_per_second": 9.002,
+      "eval_accuracy": 0.8581830790568654,
+      "eval_f1": 0.8533407001043691,
+      "eval_loss": 0.6204692721366882,
+      "eval_precision": 0.8542132893520025,
+      "eval_recall": 0.8581830790568654,
+      "eval_runtime": 37.6784,
+      "eval_samples_per_second": 76.543,
+      "eval_steps_per_second": 9.581,
       "step": 2247
     },
     {
       "epoch": 8.0,
-      "grad_norm": 5.128286838531494,
-      "learning_rate": 0.00030000000000000003,
-      "loss": 0.1469,
+      "grad_norm": 0.16941487789154053,
+      "learning_rate": 1.484342980756105e-05,
+      "loss": 0.0171,
       "step": 2568
     },
     {
       "epoch": 8.0,
-      "eval_accuracy": 0.7905686546463245,
-      "eval_f1": 0.7939091078548914,
-      "eval_loss": 0.871410608291626,
-      "eval_precision": 0.8024879308626485,
-      "eval_recall": 0.7905686546463245,
-      "eval_runtime": 41.0669,
-      "eval_samples_per_second": 70.227,
-      "eval_steps_per_second": 8.791,
+      "eval_accuracy": 0.8671983356449375,
+      "eval_f1": 0.8617834655772643,
+      "eval_loss": 0.7407661080360413,
+      "eval_precision": 0.8622466117266067,
+      "eval_recall": 0.8671983356449375,
+      "eval_runtime": 37.6949,
+      "eval_samples_per_second": 76.509,
+      "eval_steps_per_second": 9.577,
       "step": 2568
     },
     {
       "epoch": 9.0,
-      "grad_norm": 1.628487467765808,
-      "learning_rate": 0.0004414213562373095,
-      "loss": 0.1179,
+      "grad_norm": 1.5445165634155273,
+      "learning_rate": 9.909438769446831e-05,
+      "loss": 0.0029,
       "step": 2889
     },
     {
       "epoch": 9.0,
-      "eval_accuracy": 0.7517337031900139,
-      "eval_f1": 0.7621676655719284,
-      "eval_loss": 1.072391390800476,
-      "eval_precision": 0.8053888419832579,
-      "eval_recall": 0.7517337031900139,
-      "eval_runtime": 41.0124,
-      "eval_samples_per_second": 70.32,
-      "eval_steps_per_second": 8.802,
+      "eval_accuracy": 0.8564493758668515,
+      "eval_f1": 0.8473856298855159,
+      "eval_loss": 0.9039825797080994,
+      "eval_precision": 0.8486485221307185,
+      "eval_recall": 0.8564493758668515,
+      "eval_runtime": 37.879,
+      "eval_samples_per_second": 76.137,
+      "eval_steps_per_second": 9.53,
       "step": 2889
     },
     {
       "epoch": 10.0,
-      "grad_norm": 9.185128211975098,
-      "learning_rate": 9.999999999999999e-05,
-      "loss": 0.1122,
+      "grad_norm": 0.17326785624027252,
+      "learning_rate": 6.94045926336125e-05,
+      "loss": 0.1232,
       "step": 3210
     },
     {
       "epoch": 10.0,
-      "eval_accuracy": 0.7513869625520111,
-      "eval_f1": 0.7618970117176088,
-      "eval_loss": 1.0936229228973389,
-      "eval_precision": 0.7897952030598315,
-      "eval_recall": 0.7513869625520111,
-      "eval_runtime": 40.8123,
-      "eval_samples_per_second": 70.665,
-      "eval_steps_per_second": 8.845,
+      "eval_accuracy": 0.829750346740638,
+      "eval_f1": 0.8157566455477707,
+      "eval_loss": 0.9558870792388916,
+      "eval_precision": 0.84362582840677,
+      "eval_recall": 0.829750346740638,
+      "eval_runtime": 37.6559,
+      "eval_samples_per_second": 76.588,
+      "eval_steps_per_second": 9.587,
       "step": 3210
     },
     {
       "epoch": 11.0,
-      "grad_norm": 0.01600128598511219,
-      "learning_rate": 0.0001585786437626905,
-      "loss": 0.0854,
+      "grad_norm": 0.0039984057657420635,
+      "learning_rate": 2.2470951807159123e-05,
+      "loss": 0.0233,
       "step": 3531
     },
     {
       "epoch": 11.0,
-      "eval_accuracy": 0.7839805825242718,
-      "eval_f1": 0.7887505283170607,
-      "eval_loss": 1.1094753742218018,
-      "eval_precision": 0.7999745071462654,
-      "eval_recall": 0.7839805825242718,
-      "eval_runtime": 41.6728,
-      "eval_samples_per_second": 69.206,
-      "eval_steps_per_second": 8.663,
+      "eval_accuracy": 0.8564493758668515,
+      "eval_f1": 0.857178695277239,
+      "eval_loss": 0.7810959219932556,
+      "eval_precision": 0.8607102524458344,
+      "eval_recall": 0.8564493758668515,
+      "eval_runtime": 37.2622,
+      "eval_samples_per_second": 77.397,
+      "eval_steps_per_second": 9.688,
       "step": 3531
     },
     {
       "epoch": 12.0,
-      "grad_norm": 7.80008602142334,
-      "learning_rate": 0.00030000000000000003,
-      "loss": 0.1031,
+      "grad_norm": 0.003946017008274794,
+      "learning_rate": 9.351378430000336e-10,
+      "loss": 0.0009,
       "step": 3852
     },
     {
       "epoch": 12.0,
-      "eval_accuracy": 0.7964632454923717,
-      "eval_f1": 0.8003938414468947,
-      "eval_loss": 0.9343960881233215,
-      "eval_precision": 0.80861375284537,
-      "eval_recall": 0.7964632454923717,
-      "eval_runtime": 40.1152,
-      "eval_samples_per_second": 71.893,
-      "eval_steps_per_second": 8.999,
+      "eval_accuracy": 0.8626907073509015,
+      "eval_f1": 0.859450727771903,
+      "eval_loss": 0.799753725528717,
+      "eval_precision": 0.8599141248108649,
+      "eval_recall": 0.8626907073509015,
+      "eval_runtime": 37.2156,
+      "eval_samples_per_second": 77.494,
+      "eval_steps_per_second": 9.7,
       "step": 3852
     },
     {
       "epoch": 12.0,
       "step": 3852,
       "total_flos": 4.768760767819088e+18,
-      "train_loss": 0.30000500961256177,
-      "train_runtime": 2048.5956,
-      "train_samples_per_second": 250.318,
-      "train_steps_per_second": 15.669
+      "train_loss": 0.17886438471413105,
+      "train_runtime": 1744.4117,
+      "train_samples_per_second": 35.276,
+      "train_steps_per_second": 2.208
     }
   ],
   "logging_steps": 500,
-  "max_steps": 32100,
+  "max_steps": 3852,
   "num_input_tokens_seen": 0,
-  "num_train_epochs": 100,
+  "num_train_epochs": 12,
   "save_steps": 500,
   "total_flos": 4.768760767819088e+18,
   "train_batch_size": 16,