app.py

import gradio as gr
import requests
import os
import json

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 = "I carefully provide accurate, factual, thoughtful, nuanced answers and am brilliant at reasoning."

def call_nvidia_api(message, history_api, system_message, max_tokens, temperature, top_p):
    messages = [{"role": "system", "content": system_message}] if system_message else []
    messages.extend([{"role": "user", "content": message}])
    for msg in history_api:
        messages.extend([{"role": "user", "content": msg[0]}, {"role": "assistant", "content": msg[1]}])

    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 "Desculpe, ocorreu um erro ao gerar a resposta."

def chatbot_function(message, history_api, system_message, max_tokens, temperature, top_p):
    assistant_message = call_nvidia_api(message, history_api, system_message, max_tokens, temperature, top_p)
    history_api.append([message, assistant_message])
    return assistant_message, history_api

system_msg = gr.Textbox(value=BASE_SYSTEM_MESSAGE, label="System Message", placeholder="System prompt.", lines=5)
max_tokens = gr.Slider(minimum=20, maximum=1024, label="Max Tokens", step=20, value=1024)
temperature = gr.Slider(minimum=0.0, maximum=1.0, label="Temperature", step=0.1, value=0.2)
top_p = gr.Slider(minimum=0.0, maximum=1.0, label="Top P", step=0.05, value=0.7)

with gr.Blocks() as demo:
    chat_history_state = gr.State([])
    chat_interface = gr.ChatInterface(
        fn=chatbot_function,
        inputs=["message", "history_api", system_msg, max_tokens, temperature, top_p],
        outputs=["assistant_message", "history_api"],
        title="LLAMA 70B Free Demo",
        description="Explore the capabilities of LLAMA 2 70B",
        additional_inputs=[system_msg, max_tokens, temperature, top_p]
    )

demo.launch()