Update app.py

app.py

@@ -40,7 +40,8 @@ logging.basicConfig(
 )
 logger = logging.getLogger(__name__)
 
-# Apply nest_asyncio for environments like notebooks\ nnest_asyncio.apply()
+# Apply nest_asyncio for environments like notebooks
+nest_asyncio.apply()
 
 # Reproducibility
 SEED = 42
@@ -96,11 +97,12 @@ vector_store = QdrantVectorStore(
     prefer_grpc=False
 )
 storage_context = StorageContext.from_defaults(vector_store=vector_store)
-index = VectorStoreIndex.load_from_storage(storage_context)
+# Use from_storage_context to load existing index
+index = VectorStoreIndex.from_storage_context(storage_context)
 logger.info("✅ Loaded existing VectorStoreIndex from Qdrant")
 
 # --- CELL 5: Define Hybrid Retriever & Reranker ---
-Settings.llm = None  # We will use our own LLM pipeline
+Settings.llm = None  # Use our own LLM pipeline
 
 class HybridRetriever(BaseRetriever):
     def __init__(self, dense, bm25):
@@ -210,7 +212,7 @@ def build_prompt(query, context, history):
         "<|begin_of_text|>"
         f"{HDR['sys']}\n{SYSTEM_PROMPT}{HDR['eot']}"
         f"{hist_str}"
-        f"{HDR['usr']}\nContext:\n{context_str}\n\nQuestion: {query}{HDR['eot']}"
+        f"{HDR['usr']}\nContext:\n{context_str}\n\nQuestion: {query}{ HDR['eot']}"
         f"{HDR['ast']}\n"
     )
     return prompt, "rag"
@@ -259,6 +261,7 @@ with gr.Blocks(theme=gr.themes.Soft(), title="💬 Level 0 IT Support Chatbot")
             top_p_slider = gr.Slider(0.0, 1.0, value=0.9, step=0.01, label="Top-p")
             with gr.Accordion("Show Retrieved Context", open=False):
                 context_display = gr.Textbox(label="Retrieved Context", interactive=False, lines=10)
+
     def respond(message, history, k, temp, top_p):
         global chat_history
         dense_retriever.similarity_top_k = k
@@ -266,14 +269,16 @@ with gr.Blocks(theme=gr.themes.Soft(), title="💬 Level 0 IT Support Chatbot")
         reply, context_nodes = chat(message, temperature=temp, top_p=top_p)
         ctx_text = "\n\n---\n\n".join([
             f"**Source {i+1} (Score: {node.score:.4f})**\n{node.text}"
-            for i,node in enumerate(context_nodes)
+            for i, node in enumerate(context_nodes)
         ])
         history.append([message, reply])
         return "", history, ctx_text
+
     def clear_chat():
         global chat_history
         chat_history = []
         return [], None
+
     inp.submit(respond, [inp, chatbot, k_slider, temp_slider, top_p_slider], [inp, chatbot, context_display])
     send_btn.click(respond, [inp, chatbot, k_slider, temp_slider, top_p_slider], [inp, chatbot, context_display])
     clear_btn.click(clear_chat, None, [chatbot, context_display], queue=False)