Add new env variable to define desired GPT model

.env.template

@@ -1,3 +1,4 @@
 OPENAI_API_KEY=
 PDF_PATH=
-VECTOR_DB_PATH=
+VECTOR_DB_PATH=
+GPT_MODEL=

main.py

@@ -17,7 +17,7 @@ def init_session_state() -> None:
         st.session_state.messages = []
     if "qa_chain" not in st.session_state:
         env_vars = get_env_vars()
-        st.session_state.qa_chain = setup_rag(env_vars['vector_db_path'])
+        st.session_state.qa_chain = setup_rag(env_vars['vector_db_path'], env_vars['gpt_model'])
 
 def process_query(query: str, chat_history: List[Dict[str, str]], k_value: int, fetch_k: int) -> Dict[str, Any]:
     """Process user query using RAG system with debug information.

src/rag/setup.py

@@ -3,7 +3,7 @@
 from langchain_community.vectorstores import FAISS
 from langchain.chains import ConversationalRetrievalChain
 
-def setup_rag(vector_store_path: str):
+def setup_rag(vector_store_path: str, model_name: str = "gpt-4o-mini") -> ConversationalRetrievalChain:
     """Set up Retrieval-Augmented Generation (RAG) system.
     
     Loads a FAISS vector store from disk, initializes OpenAI embeddings,
@@ -30,7 +30,7 @@ def setup_rag(vector_store_path: str):
         allow_dangerous_deserialization=True
     )
 
-    llm = ChatOpenAI(model_name="gpt-4", temperature=0)
+    llm = ChatOpenAI(model_name=model_name, temperature=0)
     return ConversationalRetrievalChain.from_llm(
         llm=llm,
         retriever=vectorstore.as_retriever(),

src/utils/config.py

@@ -3,17 +3,19 @@
 
 load_dotenv()
 
-def get_env_vars():
+def get_env_vars() -> dict:
     """Get environment variables needed for RAG system.
     
     Returns:
         dict: Environment variables including:
             - pdf_path: Path to source PDF document
             - vector_db_path: Path to FAISS vector store
             - openai_key: OpenAI API key
+            - gpt_model: OpenAI model name to use
     """
     return {
         'pdf_path': os.getenv('PDF_PATH'),
         'vector_db_path': os.getenv('VECTOR_DB_PATH'),
-        'openai_key': os.getenv('OPENAI_API_KEY')
+        'openai_key': os.getenv('OPENAI_API_KEY'),
+        'gpt_model': os.getenv('GPT_MODEL', 'gpt-4o-mini')
     }