🍻 cheers

README.md

@@ -32,7 +32,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 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 imagefolder dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.0779
+- Loss: 0.1246
 - Accuracy: 1.0
 
 ## Model description

all_results.json

@@ -1,12 +1,12 @@
 {
     "epoch": 4.0,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.07792957872152328,
-    "eval_runtime": 1.3918,
-    "eval_samples_per_second": 19.399,
-    "eval_steps_per_second": 2.874,
-    "train_loss": 0.32216133549809456,
-    "train_runtime": 35.9373,
-    "train_samples_per_second": 12.577,
-    "train_steps_per_second": 0.89
+    "eval_loss": 0.12463477998971939,
+    "eval_runtime": 1.6142,
+    "eval_samples_per_second": 22.921,
+    "eval_steps_per_second": 3.097,
+    "train_loss": 0.39633166790008545,
+    "train_runtime": 48.3343,
+    "train_samples_per_second": 12.248,
+    "train_steps_per_second": 0.828
 }

config.json

@@ -11,7 +11,8 @@
   "id2label": {
     "0": "form",
     "1": "id_card_back",
-    "2": "id_card_front"
+    "2": "id_card_both",
+    "3": "id_card_front"
   },
   "image_size": 224,
   "initializer_range": 0.02,
@@ -19,7 +20,8 @@
   "label2id": {
     "form": "0",
     "id_card_back": "1",
-    "id_card_front": "2"
+    "id_card_both": "2",
+    "id_card_front": "3"
   },
   "layer_norm_eps": 1e-12,
   "model_type": "vit",

eval_results.json

@@ -1,8 +1,8 @@
 {
     "epoch": 4.0,
     "eval_accuracy": 1.0,
-    "eval_loss": 0.07792957872152328,
-    "eval_runtime": 1.3918,
-    "eval_samples_per_second": 19.399,
-    "eval_steps_per_second": 2.874
+    "eval_loss": 0.12463477998971939,
+    "eval_runtime": 1.6142,
+    "eval_samples_per_second": 22.921,
+    "eval_steps_per_second": 3.097
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:7ab31937f410bc65ef82d4e711635c3e9aa08bf5c2a45dbccf0b124f45ccf6f7
-size 343227052
+oid sha256:9c9ef57be02e563affb06ef0a52e5445207e835140deade8ca3e50f9c72f5367
+size 343230128

train_results.json

@@ -1,7 +1,7 @@
 {
     "epoch": 4.0,
-    "train_loss": 0.32216133549809456,
-    "train_runtime": 35.9373,
-    "train_samples_per_second": 12.577,
-    "train_steps_per_second": 0.89
+    "train_loss": 0.39633166790008545,
+    "train_runtime": 48.3343,
+    "train_samples_per_second": 12.248,
+    "train_steps_per_second": 0.828
 }

trainer_state.json

@@ -3,48 +3,55 @@
   "best_model_checkpoint": null,
   "epoch": 4.0,
   "eval_steps": 100,
-  "global_step": 32,
+  "global_step": 40,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 1.25,
-      "grad_norm": 1.6101118326187134,
-      "learning_rate": 0.0001375,
-      "loss": 0.681,
+      "epoch": 1.0,
+      "grad_norm": 3.4379498958587646,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.9538,
       "step": 10
     },
     {
-      "epoch": 2.5,
-      "grad_norm": 0.470041424036026,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 0.2088,
+      "epoch": 2.0,
+      "grad_norm": 1.1374919414520264,
+      "learning_rate": 0.0001,
+      "loss": 0.3279,
       "step": 20
     },
     {
-      "epoch": 3.75,
-      "grad_norm": 0.34457269310951233,
-      "learning_rate": 1.25e-05,
-      "loss": 0.1121,
+      "epoch": 3.0,
+      "grad_norm": 0.5991754531860352,
+      "learning_rate": 5e-05,
+      "loss": 0.1614,
       "step": 30
     },
     {
       "epoch": 4.0,
-      "step": 32,
-      "total_flos": 3.502669323615437e+16,
-      "train_loss": 0.32216133549809456,
-      "train_runtime": 35.9373,
-      "train_samples_per_second": 12.577,
-      "train_steps_per_second": 0.89
+      "grad_norm": 0.4659822881221771,
+      "learning_rate": 0.0,
+      "loss": 0.1422,
+      "step": 40
+    },
+    {
+      "epoch": 4.0,
+      "step": 40,
+      "total_flos": 4.587608018268979e+16,
+      "train_loss": 0.39633166790008545,
+      "train_runtime": 48.3343,
+      "train_samples_per_second": 12.248,
+      "train_steps_per_second": 0.828
     }
   ],
   "logging_steps": 10,
-  "max_steps": 32,
+  "max_steps": 40,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 4,
   "save_steps": 100,
-  "total_flos": 3.502669323615437e+16,
+  "total_flos": 4.587608018268979e+16,
   "train_batch_size": 16,
   "trial_name": null,
   "trial_params": null

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:08a9c348580393330d6a78472e99c427de368f0117a5a550f85cc9a2a55857b5
+oid sha256:79d0e6532224263a203425b24a511b725b3d099a8c59d7496933bf501cba7f9b
 size 4856