app.py

import gradio as gr
from huggingface_hub import InferenceClient

"""
For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
"""
client = InferenceClient("swiss-ai/Apertus-8B-Instruct-2509")

def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):
    messages = [{"role": "system", "content": system_message}]

    for val in history:
        if val[0]:
            messages.append({"role": "user", "content": val[0]})
        if val[1]:
            messages.append({"role": "assistant", "content": val[1]})

    messages.append({"role": "user", "content": message})

    response = ""

    try:
        for chunk in client.chat_completion(
            messages,
            max_tokens=max_tokens,
            stream=True,
            temperature=temperature,
            top_p=top_p,
        ):
            try:
                # Primeiro tenta pegar via delta (alguns modelos usam)
                if hasattr(chunk.choices[0], "delta") and chunk.choices[0].delta and getattr(chunk.choices[0].delta, "content", None):
                    content = chunk.choices[0].delta.content
                # Se não tiver, tenta via message.content (outros modelos usam)
                elif hasattr(chunk.choices[0], "message") and chunk.choices[0].message and getattr(chunk.choices[0].message, "content", None):
                    content = chunk.choices[0].message.content
                else:
                    continue  # não há conteúdo válido, pula

                response += content
                yield response

            except Exception as e:
                print(f"Erro ao processar chunk: {e}")
                continue

    except Exception as e:
        yield f"Erro inesperado: {e}"

    if response.strip() == "":
        yield "⚠️ O modelo não retornou resposta. Tente ajustar max_tokens, temperature ou escolha outro modelo."


# Interface do Gradio
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a friendly Chatbot. Your name is Juninho.", label="System message"),
        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.95,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
    ],
)

if __name__ == "__main__":
    demo.launch()