Update app.py

app.py

@@ -10,7 +10,7 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
-from transformers import pipeline, AutoTokenizer, AutoModelForCausalLM
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
 
 class BasicAgent:
     def __init__(self):
@@ -18,13 +18,16 @@ class BasicAgent:
 
         model_id = "mistralai/Mistral-7B-Instruct-v0.1"
 
+        # Load model and tokenizer directly
         self.tokenizer = AutoTokenizer.from_pretrained(model_id)
         self.model = AutoModelForCausalLM.from_pretrained(model_id)
+
+        # Create inference pipeline
         self.pipeline = pipeline(
             "text-generation",
             model=self.model,
             tokenizer=self.tokenizer,
-            device=-1  # CPU
+            device=-1
         )
 
     def __call__(self, question: str) -> str:
@@ -32,13 +35,17 @@ class BasicAgent:
 
         try:
             prompt = f"<s>[INST] {question.strip()} [/INST]"
-            result = self.pipeline(prompt, max_new_tokens=256, temperature=0.7)
-            return result[0]["generated_text"].split("[/INST]")[-1].strip()
+            output = self.pipeline(prompt, max_new_tokens=256, temperature=0.7)
+
+            # Extract and clean the response
+            generated_text = output[0]["generated_text"]
+            response = generated_text.split("[/INST]")[-1].strip()
+            return response
+
         except Exception as e:
             print(f"❌ Error during model inference: {e}")
             return f"❌ Model Error: {str(e)}"
 
-
 def run_and_submit_all( profile: gr.OAuthProfile | None):
     """
     Fetches all questions, runs the BasicAgent on them, submits all answers,