Update app.py

app.py

@@ -26,25 +26,32 @@ client = InferenceClient("mistralai/Mistral-7B-Instruct-v0.2")
 
 # --- RAG pipeline ---
 def generate_answer(messages):
-    user_query = messages[-1]["content"]
-    query_embedding = embedding_model.encode([user_query], convert_to_numpy=True)
-    faiss.normalize_L2(query_embedding)
-    _, indices = index.search(query_embedding, k=5)
-
-    context = "\n".join([f"{codes[i]}: {descriptions[i]}" for i in indices[0]])
-    prompt = f"""Here are some tariff code descriptions:
+    try:
+        user_query = messages[-1]["content"]
+        query_embedding = embedding_model.encode([user_query], convert_to_numpy=True)
+        faiss.normalize_L2(query_embedding)
+        _, indices = index.search(query_embedding, k=5)
+
+        context = "\n".join([f"{codes[i]}: {descriptions[i]}" for i in indices[0]])
+        prompt = f"""Here are some tariff code descriptions:
 {context}
 
 Question: {user_query}
 Answer:"""
 
-    response = client.text_generation(
-        prompt,
-        max_new_tokens=200,
-        temperature=0.7,
-        stop_sequences=["\n\n"]
-    )
-    return {"role": "assistant", "content": response.strip()}
+        print("Prompt sent to model:\n", prompt)
+
+        response = client.text_generation(
+            prompt,
+            max_new_tokens=200,
+            temperature=0.7,
+            stop_sequences=["\n\n"]
+        )
+        return {"role": "assistant", "content": response.strip()}
+
+    except Exception as e:
+        print("Error during inference:", e)
+        return {"role": "assistant", "content": "An internal error occurred. Please try again."}
 
 # --- Gradio Chat Interface ---
 gr.ChatInterface(