app.py

import os
import gradio as gr
import requests

# Função modificada para incluir a obtenção da API key do ambiente
def query_model(context, question):
    """
    Queries the Nemotron-3-8B-QA API with the provided question and context.
    """
    invoke_url = "https://api.nvcf.nvidia.com/v2/nvcf/pexec/functions/0c60f14d-46cb-465e-b994-227e1c3d5047"
    # Obtém a API key do ambiente
    api_key = os.environ.get("NEMO_API_KEY")
    if not api_key:
        raise ValueError("Please set the NEMO_API_KEY environment variable.")
    headers = {
        "Authorization": f"Bearer {api_key}",
        "accept": "text/event-stream",
        "content-type": "application/json",
    }
    payload = {
        "messages": [
            {"content": context, "role": "context"},
            {"content": question, "role": "user"},
        ],
        "temperature": 0.2,
        "top_p": 0.7,
        "max_tokens": 1024,
        "stream": True,
    }
    response = requests.post(invoke_url, headers=headers, json=payload, stream=True)
    answer = ""
    for line in response.iter_lines():
        if line:
            data = line.decode("utf-8")
            if '"content":"' in data:
                try:
                    content = data.split('"content":"')[1].split('"')[0]
                    answer += content
                except IndexError:
                    continue
            if data == "data:[DONE]":
                break
    return answer

iface = gr.Interface(
    fn=query_model,
    inputs=[
        gr.Textbox(label="Context", lines=5),
        gr.Textbox(label="Question"),
    ],
    outputs="text",
    title="Nemotron-3-8B-QA",
    description="
    <div style="text-align: center; font-size: 1.5em; margin-bottom: 20px;">
    <strong>Unlock the Power of AI with Nemotron-3-8B-QA for Precision Question Answering</strong>
</div>

<p>
    Dive into the realm of advanced AI with the Nemotron-3-8B-QA, a state-of-the-art language model fine-tuned by NVIDIA for unparalleled question answering performance. Built upon the robust Nemotron-3-8B architecture, this 8 billion parameter model is designed to understand and follow instructions with remarkable accuracy, providing you with precise answers to your queries.
</p>

<p>
    <strong>How to Use:</strong>
</p>

<ol>
    <li>Provide a <strong>context</strong> in the designated box, offering the model relevant information or background on your question.</li>
    <li>Enter your <strong>question</strong> in the next box, making it as clear and specific as possible.</li>
    <li>Click <strong>Submit</strong> to receive a detailed and accurate answer based on the provided context and the model's extensive knowledge base.</li>
</ol>

<p>
    <strong>This cutting-edge question answering service is powered by NVIDIA NGC, ensuring top-notch performance and reliability, and is completely free to use.</strong>
</p>

<p>
    <strong>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>
,
)

iface.launch()