Upload app.py with huggingface_hub

app.py

@@ -94,7 +94,6 @@ Settings.llm = llm # Complete the code to define the LLM model
 Settings.embedding = embedding_model # Complete the code to define the embedding model
 
 #================================Creating Langgraph agent======================#
-
 class AgentState(TypedDict):
     query: str  # The current user query
     expanded_query: str  # The expanded version of the user query
@@ -109,9 +108,11 @@ class AgentState(TypedDict):
     groundedness_check: bool
     loop_max_iter: int
 
-def expand_query(state):
+
+
+def expand_query(state: AgentState) -> AgentState:
     """
-    Expands the user query to improve retrieval of nutrition disorder-related information.
+    Expands the user query to improve retrieval of nutrition disorder-related information using few-shot prompting.
 
     Args:
         state (Dict): The current state of the workflow, containing the user query.
@@ -119,7 +120,6 @@ def expand_query(state):
     Returns:
         Dict: The updated state with the expanded query.
     """
-    print("---------Expanding Query---------")
     system_message = """
     You are a domain expert assisting in answering questions related to Nutritional Disorders.
     Perform query expansion on the question received. If there are multiple common ways of phrasing a user question \
@@ -140,31 +140,18 @@ def expand_query(state):
     expand_prompt = ChatPromptTemplate.from_messages([
         ("system", system_message),
         ("user", "Expand this query: {query} using the feedback: {query_feedback}")
-
     ])
 
     chain = expand_prompt | llm | StrOutputParser()
-    expanded_query = chain.invoke({"query": state['query'], "query_feedback":state["query_feedback"]})
-    print("expanded_query", expanded_query)
-    state["expanded_query"] = expanded_query
+    #expanded_query = chain.invoke({"query": state['query'], "query_feedback":state["query_feedback"]})
+    #print("expanded_query", expanded_query)
+    #state["expanded_query"] = expanded_query
+    state['expanded_query'] = chain.invoke({"query": state['query'], "question": state['query'], "query_feedback": state.get('query_feedback', '')}) # Pass all required variables
     return state
 
 
-# Initialize the Chroma vector store for retrieving documents
-vector_store = Chroma(
-    collection_name="nutritional_hypotheticals",
-    persist_directory="./nutritional_db",
-    embedding_function=embedding_model
 
-)
-
-# Create a retriever from the vector store
-retriever = vector_store.as_retriever(
-    search_type='similarity',
-    search_kwargs={'k': 3}
-)
-
-def retrieve_context(state):
+def retrieve_context(state: AgentState) -> AgentState:
     """
     Retrieves context from the vector store using the expanded or original query.
 
@@ -174,25 +161,23 @@ def retrieve_context(state):
     Returns:
         Dict: The updated state with the retrieved context.
     """
-    print("---------retrieve_context---------")
-    query = state['expanded_query']  # Complete the code to define the key for the expanded query
-    #print("Query used for retrieval:", query)  # Debugging: Print the query
+    query = state['expanded_query']
+    print("Query used for retrieval:", query)  # Debugging: Print the query
 
     # Retrieve documents from the vector store
     docs = retriever.invoke(query)
     print("Retrieved documents:", docs)  # Debugging: Print the raw docs object
 
     # Extract both page_content and metadata from each document
-    context= [
+    state['context'] = [
         {
             "content": doc.page_content,  # The actual content of the document
             "metadata": doc.metadata  # The metadata (e.g., source, page number, etc.)
         }
         for doc in docs
     ]
-    state['context'] = context  # Complete the code to define the key for storing the context
-    print("Extracted context with metadata:", context)  # Debugging: Print the extracted context
-    #print(f"Groundedness loop count: {state['groundedness_loop_count']}")
+
+    print("Extracted context with metadata:", state['context'])  # Debugging: Print the extracted context
     return state