app.py

import gradio as gr
from huggingface_hub import InferenceClient

# ←–– set this to the exact name of your HF repo
HF_MODEL_ID = "rieon/DeepCoder-14B-Preview-Suger"

# explicitly tell the client you want text-generation
client = InferenceClient(repo_id=HF_MODEL_ID, task="text-generation")

def respond(
    message: str,
    history: list[tuple[str, str]],
    system_message: str,
    max_tokens: int,
    temperature: float,
    top_p: float,
):
    # assemble a single prompt from system message + history
    prompt = system_message.strip() + "\n"
    for user, bot in history:
        prompt += f"User: {user}\nAssistant: {bot}\n"
    prompt += f"User: {message}\nAssistant:"

    # stream back tokens
    generated = ""
    for chunk in client.text_generation(
        inputs=prompt,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        stream=True,
    ):
        # the API returns a small JSON with .generated_text
        generated += chunk.generated_text
        yield generated

demo = gr.ChatInterface(
    fn=respond,
    system_prompt="You are a helpful coding assistant.",
    additional_inputs=[
        gr.Textbox(value="You are a helpful coding assistant.", label="System message"),
        gr.Slider(1, 2048, value=512, step=1, label="Max new tokens"),
        gr.Slider(0.1, 4.0, value=0.7, step=0.1, label="Temperature"),
        gr.Slider(0.1, 1.0, value=0.95, step=0.05, label="Top-p"),
    ],
    title="DeepCoder-14B (LoRA Fine-Tuned)",
)

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