app.py

import gradio as gr
from langchain.embeddings import SentenceTransformerEmbeddings
from langchain.vectorstores import FAISS
from langchain_community.chat_models.huggingface import ChatHuggingFace
from langchain.schema import SystemMessage, HumanMessage, AIMessage
from langchain_community.llms import HuggingFaceEndpoint

model_name = "sentence-transformers/all-mpnet-base-v2"
embedding_llm = SentenceTransformerEmbeddings(model_name=model_name)

db = FAISS.load_local("faiss_index", embedding_llm, allow_dangerous_deserialization=True)

# Set up Hugging Face model
llm = HuggingFaceEndpoint(
    repo_id="HuggingFaceH4/starchat2-15b-v0.1",
    task="text-generation",
    max_new_tokens=4096,
    temperature=0.6,
    top_p=0.9,
    top_k=40,
    repetition_penalty=1.2,
    do_sample=True,
)
chat_model = ChatHuggingFace(llm=llm)

messages = [
    SystemMessage(content="You are a helpful assistant."),
    HumanMessage(content="Hi AI, how are you today?"),
    AIMessage(content="I'm great thank you. How can I help you?")
]

def handle_query(query: str, mode: str):
    
    # Check if query is empty
    if not query.strip():
        return "Enter a valid message."
    if mode == "Chat":
        return chat_mode(query)
    elif mode == "Web-Search":
        return web_search(query)
    else:
        return "Select a valid mode."

def chat_mode(query: str):
    global messages
    prompt = HumanMessage(content=query)
    messages.append(prompt)
    response = chat_model.invoke(messages)
    messages.append(response.content)
    if len(messages) >= 6:
        messages = messages[-6:]
    return f"You: {query}\nIT-Assistant: {response.content}"

def web_search(query: str):
    global messages
    similar_docs = db.similarity_search(query, k=3)
    if similar_docs:
        source_knowledge = "\n".join([x.page_content for x in similar_docs])
    else:
        source_knowledge = ""
    augmented_prompt = f"""
    If the answer to the next query is not contained in the Web Search say 'No Answer Available' and then give advice.

    Query: {query}

    Web Search:
    {source_knowledge}
    """
    prompt = HumanMessage(content=augmented_prompt)
    messages.append(prompt)
    response = chat_model.invoke(messages)
    messages.append(response.content)
    if len(messages) >= 6:
        messages = messages[-6:]
    return f"You: {query}\nIT-Assistant: {response.content}"

demo = gr.Interface(
    fn=handle_query,
    inputs=["text", gr.Radio(["Chat", "Web-Search"], label="Mode", info="Choose a mode and enter your message, then click submit to interact.")],
    outputs="text",
    title="IT Assistant")
demo.launch()