Update app.py

app.py

@@ -21,7 +21,8 @@ def find_similarity(base64_image, text_input):
     image = Image.open(BytesIO(image_bytes))
 
     # Preprocess the image
-    image = preprocess(image).unsqueeze(0).to(device)
+    # image = preprocess(image).unsqueeze(0).to(device)
+    image = preprocess(image).to(device)
 
     # Prepare input text
     text_tokens = clip.tokenize([text_input]).to(device)
@@ -36,7 +37,7 @@ def find_similarity(base64_image, text_input):
     # Normalize features and calculate similarity
     image_features /= image_features.norm(dim=-1, keepdim=True)
     text_features /= text_features.norm(dim=-1, keepdim=True)
-    similarity = (text_features @ image_features.T).squeeze(0).cpu().numpy()
+    similarity = (text_features @ image_features.T).cpu().numpy()
 
     return similarity
 
@@ -48,7 +49,6 @@ iface = gr.Interface(
         "text"
     ],
     outputs="number",
-    live=True,
     interpretation="default",
     title="CLIP Model Image-Text Cosine Similarity",
     description="Upload a base64 image and enter text to find their cosine similarity.",