reduce the tokens

app.py

@@ -29,7 +29,7 @@ def score_it(input_img):
     # Now use the PIL Image as before
     image = pil_image.convert("RGB").resize((224, 224))
 
-    prompt = "<|begin_of_text|><|image|>show me the text in this picture"
+    prompt = "<|image|><|begin_of_text|>show me the text in this picture"
     inputs = processor(image, prompt, return_tensors="pt").to(model.device)
 
     output = model.generate(**inputs, max_new_tokens=200)