Upload 5 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+faiss_index/index.faiss filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,18 @@
+import streamlit as st
+from rag_query import query_rag
+
+st.title("United Methodist Church Discipline Assistant")
+st.write("Ask questions about The Book of Discipline 2020-2024")
+
+# User input
+query = st.text_input("Enter your question:", "Can you tell me the history of United Methodist Church")
+
+# Process query and display response
+if st.button("Submit"):
+    with st.spinner("Fetching response..."):
+        response = query_rag(query)
+        st.markdown(response)
+
+# Instructions
+st.sidebar.header("Instructions")
+st.sidebar.write("Enter a question about The United Methodist Church's Bood Of Discipline. The assistant will retrieve relevant sections and explain them in simple language, citing parts, paragraphs, and titles as needed.")

faiss_index/index.faiss

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2f1c3305c2d8a2de660469c9a995e5aef965fd87c98d66166c70bad06f164a3d
+size 2267181

faiss_index/index.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4152c2facf6490f8497a0574b95b8815eb4c0aa19b081def5d952ddb7099bf81
+size 570125

rag_query.py

@@ -0,0 +1,55 @@
+from openai import OpenAI
+from langchain_community.vectorstores import FAISS as LangChainFAISS
+from langchain_openai import OpenAIEmbeddings
+import streamlit as st
+
+# Initialize OpenAI client with Streamlit secrets
+client = OpenAI(api_key=st.secrets["OPENAI_API_KEY"])
+
+# Load vector store
+def load_vector_store():
+    embeddings = OpenAIEmbeddings(model="text-embedding-ada-002", openai_api_key=st.secrets["OPENAI_API_KEY"])
+    return LangChainFAISS.load_local(
+        folder_path="faiss_index",
+        embeddings=embeddings,
+        allow_dangerous_deserialization=True
+    )
+
+vector_store = load_vector_store()
+
+def query_rag(query, top_k=5):
+    results = vector_store.similarity_search(query, k=top_k)
+    context = ""
+    for i, doc in enumerate(results):
+        meta = doc.metadata
+        context += f"Source: {meta['source']}\n"
+        if meta["part"]:
+            context += f"Part: {meta['part']}\n"
+        context += f"Heading: {meta['heading']}\n"
+        if meta["title"]:
+            context += f"Title: {meta['title']}\n"
+        if meta["sub_title"]:
+            context += f"Sub-title: {meta['sub_title']}\n"
+        if meta["paragraph_number"]:
+            context += f"Paragraph {meta['paragraph_number']}"
+        if meta["paragraph_title"]:
+            context += f": {meta['paragraph_title']}\n"
+        elif meta["paragraph_number"]:
+            context += "\n"
+        if meta["sub_para_title"]:
+            context += f"Sub-paragraph: {meta['sub_para_title']}\n"
+        context += f"Text: {doc.page_content}\n\n"
+    
+    prompt = f"""
+    User Query: {query}
+    Retrieved Context:
+    {context}
+    
+    Provide a clear, easy-to-understand explanation based on the context. Include direct quotes with citations (e.g., 'Part I, Paragraph 1: Article I. Declaration of Union') where relevant. Structure the response by grouping information from the same 'part' together.
+    """
+    response = client.chat.completions.create(
+        model="gpt-4o-mini",
+        messages=[{"role": "user", "content": prompt}],
+        max_tokens=1000
+    )
+    return response.choices[0].message.content

requirements.txt

@@ -0,0 +1,6 @@
+openai
+langchain
+langchain-community
+langchain-openai
+faiss-cpu
+streamlit