Update app.py

app.py

@@ -10,52 +10,34 @@ 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 AutoTokenizer, AutoModelForCausalLM, pipeline
+from transformers import pipeline
 
 class BasicAgent:
     def __init__(self):
-        print("Mistral Agent loading on CPU...")
+        print("Loading Mistral-7B-Instruct-v0.1 using pipeline...")
 
-        model_id = "mistralai/Mistral-7B-Instruct-v0.1"
-
-        # Load tokenizer and model
-        self.tokenizer = AutoTokenizer.from_pretrained(model_id)
-        self.model = AutoModelForCausalLM.from_pretrained(
-            model_id,
-            device_map="auto",   # Will default to CPU
-            low_cpu_mem_usage=True,  # Helps a bit
-            torch_dtype="auto"
-        )
-
-        # Create pipeline (CPU-only)
-        self.pipeline = pipeline(
+        self.pipe = pipeline(
             "text-generation",
-            model=self.model,
-            tokenizer=self.tokenizer,
-            device=-1  # force CPU
+            model="mistralai/Mistral-7B-Instruct-v0.1",
+            device=-1  # CPU only
         )
 
     def __call__(self, question: str) -> str:
-        print(f"Agent received question: {question[:50]}...")
+        print(f"Received question: {question[:50]}...")
 
-        try:
-            # Format with instruction template
-            prompt = f"<s>[INST] {question.strip()} [/INST]"
+        prompt = f"<s>[INST] {question.strip()} [/INST]"
 
-            output = self.pipeline(
+        try:
+            output = self.pipe(
                 prompt,
                 max_new_tokens=256,
-                do_sample=True,
                 temperature=0.7,
                 top_p=0.95
             )
-
             full_response = output[0]["generated_text"]
-            answer = full_response.split("[/INST]")[-1].strip()
-
-            return answer
+            return full_response.split("[/INST]")[-1].strip()
         except Exception as e:
-            print(f"❌ Mistral error: {e}")
+            print(f"❌ Inference Error: {e}")
             return f"❌ Model Error: {str(e)}"
             
 def run_and_submit_all( profile: gr.OAuthProfile | None):