Update app.py

app.py

@@ -49,7 +49,7 @@ def extract_answer(prediction: str) -> tuple:
 
 def predict(question: str, option_a: str, option_b: str, option_c: str, option_d: str, 
            correct_option: int = None, explanation: str = None,
-           temperature: float = 0.6, top_p: float = 0.9, max_tokens: int = 256):
+           temperature: float = 0.6, top_p: float = 0.9, max_tokens: int = 10):
     # Format the prompt
     prompt = f"Question: {question}\n\nOptions:\nA. {option_a}\nB. {option_b}\nC. {option_c}\nD. {option_d}\n\nAnswer:"
     
@@ -64,7 +64,7 @@ def predict(question: str, option_a: str, option_b: str, option_c: str, option_d
             temperature=temperature,
             top_p=top_p,
             do_sample=True,
-            pad_token_id=tokenizer.eos_token_id
+            # pad_token_id=tokenizer.eos_token_id
         )
     
     # Get prediction
@@ -118,7 +118,7 @@ with gr.Blocks(title="Medical MCQ Predictor") as demo:
                     info="Higher values allow more diverse tokens, lower values more focused"
                 )
                 max_tokens = gr.Slider(
-                    minimum=32,
+                    minimum=10,
                     maximum=512,
                     value=256,
                     step=32,