app.py

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

MAX_MAX_NEW_TOKENS = 2048
DEFAULT_MAX_NEW_TOKENS = 1024
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))

DESCRIPTION = """\
# L-MChat
This Space demonstrates [L-MChat](https://huggingface.co/collections/Artples/l-mchat-663265a8351231c428318a8f) by L-AI.
"""

if not torch.cuda.is_available():
    DESCRIPTION += "\n<p>Running on CPU! This demo does not work on CPU.</p>"

model_details = {
    "Fast-Model": "Artples/L-MChat-Small",
    "Quality-Model": "Artples/L-MChat-7b"
}

models = {name: AutoModelForCausalLM.from_pretrained(model_id, device_map="auto") for name, model_id in model_details.items()}
tokenizers = {name: AutoTokenizer.from_pretrained(model_id) for name, model_id in model_details.items()}

@spaces.GPU(enable_queue=True, duration=90)
def generate(
    model_choice: str,
    message: str,
    chat_history: list[tuple[str, str]],
    system_prompt: str,
    max_new_tokens: int = 1024,
    temperature: float = 0.1,
    top_p: float = 0.9,
    top_k: int = 50,
    repetition_penalty: float = 1.2,
) -> str:
    model = models[model_choice]
    tokenizer = tokenizers[model_choice]

    conversation = [{"role": "system", "content": system_prompt}] if system_prompt else []
    conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}] for user, assistant in chat_history)
    conversation.append({"role": "user", "content": message})

    input_ids = tokenizer(conversation, return_tensors="pt", truncation=True, max_length=MAX_INPUT_TOKEN_LENGTH).input_ids
    input_ids = input_ids.to(model.device)

    output_ids = model.generate(input_ids, max_length=MAX_INPUT_TOKEN_LENGTH + max_new_tokens, temperature=temperature, top_p=top_p, top_k=top_k, repetition_penalty=repetition_penalty)
    output_text = tokenizer.decode(output_ids[0], skip_special_tokens=True)

    return output_text


chat_interface = gr.ChatInterface(
    theme='ehristoforu/RE_Theme',
    fn=generate,
    additional_inputs=[gr.Textbox(label="System prompt", lines=6), gr.Dropdown(label="Model Choice", choices=list(model_details.keys()), value="Quality-Model")],
    examples=[
        ["Hello there! How are you doing?"],
        ["Can you explain briefly to me what is the Python programming language?"],
        ["Explain the plot of Cinderella in a sentence."],
        ["How many hours does it take a man to eat a Helicopter?"],
        ["Write a 100-word article on 'Benefits of Open-Source in AI research'"],
    ],
)

with gr.Blocks(css="style.css") as demo:
    gr.Markdown(DESCRIPTION)
    chat_interface.render()

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