ipex-llm-multi-test.py

import os
os.system('pip install accelerate')
os.system('pip install ipex-llm[all]')
os.system('pip install modelscope -U')
os.system('pip install --upgrade transformers==4.37.0')
os.system('pip install transformers_stream_generator')
from threading import Thread
from typing import Iterator
import gc
import gradio as gr
import torch
from modelscope import AutoModelForCausalLM, AutoModel, AutoTokenizer
from transformers import  TextIteratorStreamer
from ipex_llm import optimize_model

MAX_MAX_NEW_TOKENS = 2048
DEFAULT_MAX_NEW_TOKENS = 128
MAX_INPUT_TOKEN_LENGTH = int(os.getenv("MAX_INPUT_TOKEN_LENGTH", "4096"))
CHATGLM_IDS = ['ZhipuAI/chatglm-6b', 'ZhipuAI/chatglm2-6b', 'ZhipuAI/chatglm3-6b']
model_options = ['qwen/Qwen1.5-1.8B-Chat', 'qwen/Qwen1.5-0.5B-Chat', 'ZhipuAI/chatglm3-6b']
# model_id = model_options[0]

# model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype=torch.bfloat16, device_map="auto")
# tokenizer = AutoTokenizer.from_pretrained(model_id)
# tokenizer.use_default_system_prompt = False
# model = optimize_model(model)
model = None
tokenizer = None

def change_model(model_id):
    global model, tokenizer, chat_history
    try:
        if model_id in CHATGLM_IDS:
            model = AutoModel.from_pretrained(model_id, torch_dtype="auto", trust_remote_code=True)
        else:
            model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", trust_remote_code=True)
        tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
        # tokenizer.use_default_system_prompt = False
        model = optimize_model(model)
        gc.collect()
        # chat_interface.clear_btn.click()
        return f"Successfully loaded {model_id}"
    except Exception as e:
        return f"Error loading model: {str(e)}"


def generate(
    message: str,
    chat_history: list[tuple[str, str]],
    system_prompt: str,
    max_new_tokens: int = 1024,
    temperature: float = 0.6,
    top_p: float = 0.9,
    top_k: int = 50,
    repetition_penalty: float = 1.2,
) -> Iterator[str]:
    conversation = []
    if system_prompt:
        conversation.append({"role": "system", "content": system_prompt})
    for user, assistant in chat_history:
        conversation.extend([{"role": "user", "content": user}, {"role": "assistant", "content": assistant}])
    conversation.append({"role": "user", "content": message})
    with torch.inference_mode():
        input_ids = tokenizer.apply_chat_template(conversation, tokenize=False, add_generation_prompt=True)
        input_ids = tokenizer([input_ids],return_tensors="pt").to(model.device)

        streamer = TextIteratorStreamer(tokenizer, timeout=60.0, skip_prompt=True, skip_special_tokens=True)
        generate_kwargs = dict(
            input_ids=input_ids.input_ids,
            streamer=streamer,
            max_new_tokens=max_new_tokens,
            do_sample=True,
            top_p=top_p,
            top_k=top_k,
            temperature=temperature,
            repetition_penalty=repetition_penalty,
        ) 
        t = Thread(target=model.generate, kwargs=generate_kwargs)
        t.start()

        outputs = []
        for text in streamer:
            outputs.append(text)
            yield "".join(outputs)

        #outputs = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
        print(outputs)
        #yield outputs

chat_interface = gr.ChatInterface(
    fn=generate,
    additional_inputs=[
        gr.Textbox(label="System prompt", lines=6),
        gr.Slider(
            label="Max new tokens",
            minimum=1,
            maximum=MAX_MAX_NEW_TOKENS,
            step=1,
            value=DEFAULT_MAX_NEW_TOKENS,
        ),
        gr.Slider(
            label="Temperature",
            minimum=0.1,
            maximum=4.0,
            step=0.1,
            value=0.6,
        ),
        gr.Slider(
            label="Top-p (nucleus sampling)",
            minimum=0.05,
            maximum=1.0,
            step=0.05,
            value=0.9,
        ),
        gr.Slider(
            label="Top-k",
            minimum=1,
            maximum=1000,
            step=1,
            value=50,
        ),
        gr.Slider(
            label="Repetition penalty",
            minimum=1.0,
            maximum=2.0,
            step=0.05,
            value=1.2,
        ),
    ],
    stop_btn=None,
    examples=[
        ["你好！你是谁？"],
        ["请简单介绍一下大语言模型?"],
        ["请讲一个小人物成功的故事."],
        ["浙江的省会在哪里?"],
        ["写一篇100字的文章，题目是'人工智能开源的优势'"],
    ],
)

with gr.Blocks(css="style.css") as demo:
    gr.Markdown("""<center><font size=8>ipex-llm-test for multi-models👾</center>""")
    gr.Markdown("""<center><font size=6>Please clear chatbot before loading new model</center>""")
    with gr.Row():
        with gr.Column(scale=3):
            model_option = gr.Dropdown(choices=model_options, label="Model_ID")
        with gr.Column(scale=1):
            modify_model = gr.Button("🛠️ Load model", scale=2)
    with gr.Row():
        info = gr.Textbox(lines=1, value="No model loaded", label="Model Info")
    
    # model_option.input(fn=change_model, inputs=[model_option])

    chat_interface.render()
    modify_model.click(fn=change_model, inputs=[model_option], outputs=[info])

if __name__ == "__main__":
    demo.queue(max_size=20).launch()