app.py

import streamlit as st
from gradio_client import Client

# Constants
TITLE = "Llama2 70B Chatbot"
DESCRIPTION = """
This Space demonstrates model [Llama-2-70b-chat-hf](https://huggingface.co/meta-llama/Llama-2-70b-chat-hf) by Meta, 
a Llama 2 model with 70B parameters fine-tuned for chat instructions. 
"""

# Initialize client
client = Client("https://ysharma-explore-llamav2-with-tgi.hf.space/")

# Prediction function
def predict(message, system_prompt="", temperature=0.9, max_new_tokens=4096):
    return client.predict(
			message,	# str in 'Message' Textbox component
            system_prompt,	# str in 'Optional system prompt' Textbox component
			temperature,	# int | float (numeric value between 0.0 and 1.0)
			max_new_tokens,	# int | float (numeric value between 0 and 4096)
			0.3,	# int | float (numeric value between 0.0 and 1)
			1,	# int | float (numeric value between 1.0 and 2.0)
			api_name="/chat"
    )

# Streamlit UI
st.title(TITLE)
st.write(DESCRIPTION)

if "messages" not in st.session_state:
    st.session_state.messages = []

# Display chat messages from history on app rerun
for message in st.session_state.messages:
    with st.chat_message(message["role"]):
        st.markdown(message["content"])

# React to user input
if prompt := st.chat_input("Ask LLama-2-70b anything..."):
    # Display user message in chat message container
    st.chat_message("human").markdown(prompt)
    # Add user message to chat history
    st.session_state.messages.append({"role": "human", "content": prompt})

    response = predict(prompt)
    # Display assistant response in chat message container
    with st.chat_message("assistant", avatar='🦙'):
        st.markdown(response)
    # Add assistant response to chat history
    st.session_state.messages.append({"role": "assistant", "content": response})