Update app.py

app.py

@@ -25,9 +25,12 @@ def generate_caption(image):
     ])
     image_tensor = transform(image).unsqueeze(0).to(device)
     
-    # Generate caption with the fixed prompt
-    input_text = FIXED_PROMPT + " "
-    output = model.generate(pixel_values=image_tensor, input_text=input_text)
+    # Prepend the fixed prompt to the input tensor
+    fixed_prompt_tensor = tokenizer(FIXED_PROMPT, return_tensors="pt").input_ids.to(device)
+    input_tensor = torch.cat((fixed_prompt_tensor, image_tensor), dim=1)
+    
+    # Generate caption
+    output = model.generate(pixel_values=image_tensor)
     caption = tokenizer.decode(output[0], skip_special_tokens=True)
     return caption