Update app.py

app.py

@@ -77,9 +77,10 @@ def process_image(model, tokenizer, transform, device, check_type, image, text):
         if 'R50' in check_type:
             text_embeds = model.language_embedding(input_ids)
         else:
-            text_embeds = model.tok_embeddings(input_ids)
+            text_embeds = model.tok_embeddings(input_ids).clone()
         
         vit_embeds, size1 = model.forward_tokenocr(pixel_values.to(torch.bfloat16).to(device))
+        
         print("vit_embeds",vit_embeds)
         print("vit_embeds,shape",vit_embeds.shape)
         print("target_ratio",target_ratio)
@@ -107,11 +108,9 @@ def process_image(model, tokenizer, transform, device, check_type, image, text):
     
     current_bpe = [tokenizer.decode([i]) for i in input_ids]
     # current_bpe[-1] = 'Input text'
-    print("len1 ",len(current_vis))
-    print("len2 ",len(current_bpe))
-    print("current_bpe!!!!!!!!!!!!!!!!!!!",current_bpe)
+
     
-    current_bpe.append(text)
+    # current_bpe.append(text)
 
     return image, current_vis, current_bpe