Update app.py

app.py

@@ -9,8 +9,7 @@ from langchain.chains.question_answering import load_qa_chain
 from langchain.prompts import PromptTemplate
 import os
 import json
-
-from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig, TextStreamer
+from transformers import AutoModelForCausalLM, AutoTokenizer, GenerationConfig, TextStreamer, ConversationalPipeline
 
 ####CREDIT#####
 #Credit to author (Sri Laxmi) of original code reference: SriLaxmi1993
@@ -120,6 +119,8 @@ def update_chat_history(question, reply):
         st.write(f"**Reply:** {conversation['reply']}")
         st.write("---")
 
+
+'''
 def user_input(user_question, api_key):
     embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001", google_api_key=api_key)
     new_db = FAISS.load_local("faiss_index", embeddings)
@@ -130,6 +131,38 @@ def user_input(user_question, api_key):
 
     #chat history
     update_chat_history(user_question, response["output_text"])
+'''
+
+
+def user_input(user_question, api_key):
+    embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001", google_api_key=api_key)
+    new_db = FAISS.load_local("faiss_index", embeddings)
+    docs = new_db.similarity_search(user_question)
+    chain = get_conversational_chain()
+    response_gemini = chain({"input_documents": docs, "question": user_question}, return_only_outputs=True)
+    
+    # Initialize the Hugging Face conversational pipeline with your custom model
+    pipeline = ConversationalPipeline(model_name_or_path="bofenghuang/vigogne-2-7b-chat")
+
+    # Prompt template for making the response more conversational
+    prompt_template = f"""
+    Transform the following response into a more conversational tone without adding new information:
+    
+    Response:
+    {response_gemini["output_text"]}
+    
+    Transformed Response:
+    """
+
+    # Generate the transformed response using the Hugging Face model
+    transformed_response = pipeline(prompt=prompt_template, max_length=100)
+
+    # Display the transformed response
+    st.write("Reply: ", transformed_response)
+
+    # Update chat history
+    update_chat_history(user_question, transformed_response)
+
 
 
 def main():