app.py

import gradio as gr
from huggingface_hub import InferenceClient
from langchain_community.vectorstores.faiss import FAISS
from langchain.chains import RetrievalQA
from langchain_community.llms import HuggingFaceHub

# Load the vector store from the saved index files
vector_store = FAISS.load_local("db.index", embeddings=None)

# Initialize the HuggingFaceHub LLM
llm = HuggingFaceHub(repo_id="HuggingFaceH4/zephyr-7b-beta", model_kwargs={"temperature": None, "top_p": None})

# Initialize the RetrievalQA chain
qa = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=vector_store.as_retriever())

def respond(message, history, system_message, max_tokens, temperature, top_p):
    # Update the temperature and top_p values for the LLM
    llm.model_kwargs["temperature"] = temperature
    llm.model_kwargs["top_p"] = top_p

    messages = [{"role": "system", "content": system_message}]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})

    result = qa({"input_documents": vector_store.texts, "question": message})
    response = result["result"]

    history.append((message, response))
    return response, history

"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a helpful car configuration assistant, specifically you are the assistant for Apex Customs (https://www.apexcustoms.com/). Given the user's input, provide suggestions for car models, colors, and customization options. Be creative and conversational in your responses. You should remember the user car model and tailor your answers accordingly. (You must not generate the next question of the user yourself, you only have to answer.) \n\nUser: ", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
    ],
)

if __name__ == "__main__":
    demo.launch()