Update app.py

app.py

@@ -356,15 +356,27 @@ def generate_answer(message, choice, retrieval_mode, selected_model):
     prompt_template = QA_CHAIN_PROMPT_1 if choice == "Details" else QA_CHAIN_PROMPT_2
 
     if retrieval_mode == "VDB":
-        qa_chain = RetrievalQA.from_chain_type(
-            llm=selected_model,
-            chain_type="stuff",
-            retriever=retriever,
-            chain_type_kwargs={"prompt": prompt_template}
-        )
-        response = qa_chain({"query": message})
-        logging.debug(f"Vector response: {response}")
-        return response['result'], extract_addresses(response['result'])
+        if selected_model == "GPT-4o":
+            # Use Langchain with GPT-4o
+            qa_chain = RetrievalQA.from_chain_type(
+                llm=chat_model,
+                chain_type="stuff",
+                retriever=retriever,
+                chain_type_kwargs={"prompt": prompt_template}
+            )
+            response = qa_chain({"query": message})
+            logging.debug(f"Vector response: {response}")
+            return response['result'], extract_addresses(response['result'])
+        elif selected_model == "Phi-3.5":
+            # Directly use the Phi-3.5 model for text generation
+            response = selected_model(message, **{
+                "max_new_tokens": 500,
+                "return_full_text": False,
+                "temperature": 0.0,
+                "do_sample": False,
+            })[0]['generated_text']
+            logging.debug(f"Phi-3.5 response: {response}")
+            return response, extract_addresses(response)
     elif retrieval_mode == "KGF":
         response = chain_neo4j.invoke({"question": message})
         logging.debug(f"Knowledge-Graph response: {response}")
@@ -373,6 +385,7 @@ def generate_answer(message, choice, retrieval_mode, selected_model):
         return "Invalid retrieval mode selected.", []
 
 
+
 # def bot(history, choice, tts_choice, retrieval_mode):
 #     if not history:
 #         return history