app.py

import gradio as gr
from transformers import AutoModelForCausalLM, AutoTokenizer, TextIteratorStreamer
import spaces
from threading import Thread

# Load model and tokenizer
model_name = "Magpie-Align/MagpieLM-4B-Chat-v0.1"

device = "cuda" # the device to load the model onto
tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    torch_dtype="auto"
)
model.to(device)

MAX_INPUT_TOKEN_LENGTH = 4096  # You may need to adjust this value

@spaces.GPU(enable_queue=True)
def respond(
    message,
    history: list[tuple[str, str]],
    system_message,
    max_tokens=2048,
    temperature=0.6,
    top_p=0.9,
    repetition_penalty=1.0,
):
    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})
 
    input_ids = tokenizer.apply_chat_template(messages, return_tensors="pt")
    if input_ids.shape[1] > MAX_INPUT_TOKEN_LENGTH:
        input_ids = input_ids[:, -MAX_INPUT_TOKEN_LENGTH:]
        gr.Warning(f"Trimmed input from conversation as it was longer than {MAX_INPUT_TOKEN_LENGTH} tokens.")
    input_ids = input_ids.to(model.device)

    streamer = TextIteratorStreamer(tokenizer, timeout=10.0, skip_prompt=True, skip_special_tokens=True)
    generate_kwargs = dict(
        input_ids=input_ids,
        streamer=streamer,
        max_new_tokens=max_tokens,
        do_sample=True,
        top_p=top_p,
        temperature=temperature,
        repetition_penalty=repetition_penalty,
    )
    
    thread = Thread(target=model.generate, kwargs=generate_kwargs)
    thread.start()

    def stream():
        for text in streamer:
            yield text

    return stream()

demo = gr.ChatInterface(
    respond,
    additional_inputs=[
        gr.Textbox(value="You are Magpie, 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=0.6, step=0.1, label="Temperature"),
        gr.Slider(
            minimum=0.1,
            maximum=1.0,
            value=0.9,
            step=0.05,
            label="Top-p (nucleus sampling)",
        ),
        gr.Slider(minimum=0.5, maximum=1.5, value=1.0, step=0.1, label="Repetition Penalty"),
    ],
)

if __name__ == "__main__":
    demo.queue()
    demo.launch(share=True)