app.py

from typing import Generator, List, Optional, Tuple, Union
import os
from dotenv import load_dotenv
import gradio as gr
from huggingface_hub import InferenceClient

# Load environment variables
load_dotenv()

# Read system prompt from file
prompt_path = os.path.join(os.path.dirname(__file__), "prompt.md")
with open(prompt_path, "r", encoding="utf-8") as f:
    SYSTEM_PROMPT = f.read()

# Initialize HuggingFace client with token from env
client = InferenceClient(token=os.getenv("HUGGING_FACE_HUB_TOKEN"))

def format_prompt(message: str, history: List[Tuple[str, str]]) -> str:
    """Format the conversation prompt for the model.

    Args:
        message: The current user message.
        history: List of past interactions (user, assistant).

    Returns:
        str: The formatted prompt for the model.
    """
    prompt = f"<|system|>\n{SYSTEM_PROMPT}</s>"
    for user, assistant in history:
        prompt += f"<|user|>\n{user}</s><|assistant|>\n{assistant}</s>"
    prompt += f"<|user|>\n{message}</s><|assistant|>\n"
    return prompt

def predict(
    message: str,
    history: List[Tuple[str, str]],
    temperature: float = 0.7,
    max_new_tokens: int = 1024,
    top_p: float = 0.9
) -> str:
    """Generate a complete response to the user message.

    Args:
        message: The current user message.
        history: List of past interactions.
        temperature: Sampling temperature.
        max_new_tokens: Maximum number of tokens to generate.
        top_p: Top-p sampling parameter.

    Returns:
        str: Complete generated response.
    """
    response = client.text_generation(
        format_prompt(message, history),
        temperature=temperature,
        max_new_tokens=max_new_tokens,
        top_p=top_p,
        repetition_penalty=1.2,
        do_sample=True,
        seed=42,
    )
    return response

# Main Gradio interface
with gr.Blocks(css="styles.css", title="AstroAlien 3000") as demo:
    gr.Markdown("# 👽🛸 Bem-vindo ao AstroAlien 3000! 🌌✨", elem_classes="glass-card")

    chatbot = gr.Chatbot(type="messages", elem_classes="glass-card")

    with gr.Row():
        temperature = gr.Slider(0, 1, value=0.7, label="Temperatura", elem_classes="glass-card")
        max_tokens = gr.Slider(1, 2048, value=1024, step=1, label="Máximos Tokens", elem_classes="glass-card")
        top_p = gr.Slider(0, 1, value=0.9, label="Top-p", elem_classes="glass-card")

    gr.ChatInterface(
        predict,
        chatbot=chatbot,
        additional_inputs=[temperature, max_tokens, top_p],
        type="messages"
    )

demo.launch()