Update app.py

app.py

@@ -101,7 +101,7 @@ def textMode(text, count):
 def imageMode(image, question):
     image_embedding = encode_image(image)
     print('-------Image embedding from clip obtained-----------')
-    imgToTextEmb = img_proj_head(image_embedding)
+    imgToTextEmb = img_proj_head(image_embedding).unsqueeze(0)
     print('-------text embedding from projection obtained-----------')
     question = "Question: " + question + "Answer: "
     Qtokens = torch.tensor(tokenizer_text.encode(question, add_special_tokens=True)).unsqueeze(0)