app.py

import gradio as gr
import requests
import os
import json

# API and environment variables
API_KEY = os.getenv('API_KEY')
INVOKE_URL = "https://api.nvcf.nvidia.com/v2/nvcf/pexec/functions/0e349b44-440a-44e1-93e9-abe8dcb27158"
FETCH_URL_FORMAT = "https://api.nvcf.nvidia.com/v2/nvcf/pexec/status/"
headers = {
 "Authorization": f"Bearer {API_KEY}",
 "Accept": "application/json",
 "Content-Type": "application/json",
}

# Base system message
BASE_SYSTEM_MESSAGE = "I carefully provide accurate, factual, thoughtful, nuanced answers and am brilliant at reasoning."

def call_nvidia_api(history, system_message, max_tokens, temperature, top_p):
 """Calls the NVIDIA API to generate a response."""
 messages = [{"role": "system", "content": system_message}]
 messages.extend([{"role": "user", "content": msg} for msg, _ in history])

 payload = {
 "messages": messages,
 "temperature": temperature,
 "top_p": top_p,
 "max_tokens": max_tokens,
 "stream": False
 }
 session = requests.Session()
 response = session.post(INVOKE_URL, headers=headers, json=payload)
 while response.status_code == 202:
 request_id = response.headers.get("NVCF-REQID")
 fetch_url = FETCH_URL_FORMAT + request_id
 response = session.get(fetch_url, headers=headers)
 response.raise_for_status()
 response_body = response.json()
 if response_body.get("choices"):
 assistant_message = response_body["choices"][0]["message"]["content"]
 return assistant_message
 else:
 return "Sorry, there was an error generating the response."

def chatbot_submit(message, chat_history, system_message, max_tokens_val, temperature_val, top_p_val):
 """Submits the user message to the chatbot and updates the chat history."""
 chat_history.append([message, ""]) # Add user message to history

 # Call NVIDIA API to generate a response
 assistant_message = call_nvidia_api(chat_history, system_message, max_tokens_val, temperature_val, top_p_val)

 # Update history with assistant's response
 chat_history[-1][1] = assistant_message

 return assistant_message, chat_history

# Gradio interface setup
chat_history_state = gr.State([])
system_msg = gr.Textbox(BASE_SYSTEM_MESSAGE, label="System Message", placeholder="System prompt.", lines=5)
max_tokens = gr.Slider(20, 1024, label="Max Tokens", step=20, value=1024)
temperature = gr.Slider(0.0, 1.0, label="Temperature", step=0.1, value=0.2)
top_p = gr.Slider(0.0, 1.0, label="Top P", step=0.05, value=0.7)

with gr.Blocks() as demo:
 chat_history_state = gr.State([])
 chatbot = gr.ChatInterface(
 fn=chatbot_submit,
 additional_inputs=[system_msg, max_tokens, temperature, top_p],
 title="LLAMA 70B Free Demo",
 description="""
 <div style="text-align: center; font-size: 1.5em; margin-bottom: 20px;">
 <strong>Explore the Capabilities of LLAMA 2 70B</strong>
 </div>
 <p>Llama 2 is a large language AI model capable of generating text and code in response to prompts.</p>
 <p><strong>How to Use:</strong></p>
 <ol>
 <li>Enter your <strong>message</strong> in the textbox to start a conversation or ask a question.</li>
 <li>Adjust the parameters in the "Additional Inputs" accordion to control the model's behavior.</li>
 <li>Use the buttons below the chatbot to submit your query, clear the chat history, or perform other actions.</li>
 </ol>
 <p><strong>Powered by NVIDIA's cutting-edge AI API, LLAMA 2 70B offers an unparalleled opportunity to interact with an AI model of exceptional conversational ability, accessible to everyone at no cost.</strong></p>
 <p><strong>HF Created by:</strong> @artificialguybr (<a href="https://twitter.com/artificialguybr">Twitter</a>)</p>
 <p><strong>Discover more:</strong> <a href="https://artificialguy.com">artificialguy.com</a></p>
 """,
 submit_btn="Submit",
 clear_btn="🗑️ Clear",
 )

 def clear_chat():
 chat_history_state.value = []
 chatbot.textbox.value = ""

 chatbot.clear()
demo.launch()