app.py

import spaces
import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer

model_name = "daniel-dona/gemma-3-270m-it"

#pipe = pipeline("text-generation", model=model, device="cuda")

@spaces.GPU
def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens,
    temperature,
    top_p,
):

    messages = [{"role": "system", "content": system_message}]

    print("Got:", 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 = pipe(
        messages,
        max_new_tokens=max_tokens,
        temperature=temperature,
        top_p=top_p,
        return_full_text=False, 
    )

    generated_text = response[0]['generated_text']

    yield generated_text"""

    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModelForCausalLM.from_pretrained(
        model_name,
        torch_dtype="auto",
        device_map="auto"
    )

    text = tokenizer.apply_chat_template(
        messages,
        tokenize=False,
        add_generation_prompt=True
    )

    model_inputs = tokenizer([text], return_tensors="pt").to(model.device)

    sample = True

    if temperature == 0:
        sample = False

    # conduct text completion
    generated_ids = model.generate(
        **model_inputs,
        max_new_tokens=max_tokens,
        do_sample=sample,
        top_p=top_p,
        temperature=temperature
    )
    output_ids = generated_ids[0][len(model_inputs.input_ids[0]):].tolist() 

    content = tokenizer.decode(output_ids, skip_special_tokens=True).strip("\n")

    return content


"""
For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
"""
demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are a friendly Chatbot.", 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()