master

分支 (1)

管理

管理

master

llm
/
web_demo_vision.py

from transformers import AutoModel, AutoTokenizer
import gradio as gr
import mdtex2html
import time

tokenizer = AutoTokenizer.from_pretrained("/home/visualglm-6b", trust_remote_code=True, revision="")
model = AutoModel.from_pretrained("/home/visualglm-6b", trust_remote_code=True, revision="").half().mlu()
model = model.eval()

"""Override Chatbot.postprocess"""


def postprocess(self, y):
    if y is None:
        return []
    for i, (message, response) in enumerate(y):
        y[i] = (
            None if message is None else mdtex2html.convert((message)),
            None if response is None else mdtex2html.convert(response),
        )
    return y


gr.Chatbot.postprocess = postprocess


def parse_text(text):
    """copy from https://github.com/GaiZhenbiao/ChuanhuChatGPT/"""
    lines = text.split("\n")
    lines = [line for line in lines if line != ""]
    count = 0
    for i, line in enumerate(lines):
        if "```" in line:
            count += 1
            items = line.split('`')
            if count % 2 == 1:
                lines[i] = f'<pre><code class="language-{items[-1]}">'
            else:
                lines[i] = f'<br></code></pre>'
        else:
            if i > 0:
                if count % 2 == 1:
                    line = line.replace("`", "\`")
                    line = line.replace("<", "&lt;")
                    line = line.replace(">", "&gt;")
                    line = line.replace(" ", "&nbsp;")
                    line = line.replace("*", "&ast;")
                    line = line.replace("_", "&lowbar;")
                    line = line.replace("-", "&#45;")
                    line = line.replace(".", "&#46;")
                    line = line.replace("!", "&#33;")
                    line = line.replace("(", "&#40;")
                    line = line.replace(")", "&#41;")
                    line = line.replace("$", "&#36;")
                lines[i] = "<br>"+line
    text = "".join(lines)
    return text

def predict(input, image_path, chatbot, max_length, top_p, temperature, history):
    if image_path is None:
        return [(input, "图片为空！请重新上传图片并重试。")]

    start_time = time.time()  # Record start time

    chatbot.append((parse_text(input), ""))
    for response, history in model.stream_chat(tokenizer, image_path, input, history, max_length=max_length, top_p=top_p,
                                               temperature=temperature):
        chatbot[-1] = (parse_text(input), parse_text(response))

        end_time = time.time()  # Record end time
        elapsed_time = end_time - start_time  # Calculate elapsed time in seconds
        tokens = len(tokenizer(response, return_tensors='pt')['input_ids'][0])  # Calculate the number of tokens in the response

        # Print or store the inference time and speed
        inference_time_ms = elapsed_time * 1000  # Convert to milliseconds
        speed_ms_per_token = inference_time_ms / tokens

        # 将推理时间和速度信息添加到 response 的第一行
        response_with_info = f"推理时间：{elapsed_time:.2f} s，速度：{speed_ms_per_token:.2f} ms/token\n{response}"

        # 更新 chatbot 中的最后一个对话
        chatbot[-1] = (parse_text(input), parse_text(response_with_info))

        print(f"推理时间: {elapsed_time:.2f} s, 速度: {speed_ms_per_token:.2f} ms/token")

        yield chatbot, history


def predict_new_image(image_path, chatbot, max_length, top_p, temperature):
    # 设置输入和历史变量
    input, history = "描述这张图片。", []
    chatbot.append((parse_text(input), ""))

    # 记录开始时间
    start_time = time.time()

    # 对模型进行推理，返回图像描述的响应
    for response, history in model.stream_chat(tokenizer, image_path, input, history, max_length=max_length,
                                               top_p=top_p,
                                               temperature=temperature):
        # 记录结束时间
        end_time = time.time()

        # 计算经过的时间（秒）
        elapsed_time = end_time - start_time

        # 检查响应是否为字符串
        if isinstance(response, str):
            # 计算响应中的令牌数量
            tokens = len(tokenizer(response, return_tensors='pt')['input_ids'][0])
        elif isinstance(response, int):
            # 将单一令牌的响应视为一个令牌
            tokens = 1
        else:
            # 输出错误信息并跳过当前迭代
            print("无效的响应格式。期望一个字符串或整数。")
            continue

        # 计算推理时间和速度
        inference_time_ms = elapsed_time * 1000  # 转换为毫秒
        speed_ms_per_token = inference_time_ms / tokens

        # 将推理时间和速度信息添加到 response 的第一行
        response_with_info = f"推理时间：{elapsed_time:.2f} s，速度：{speed_ms_per_token:.2f} ms/token\n{response}"

        # 更新 chatbot 中的最后一个对话
        chatbot[-1] = (parse_text(input), parse_text(response_with_info))

        # 打印推理时间和速度信息
        print(f"推理时间：{elapsed_time:.2f} s，速度：{speed_ms_per_token:.2f} ms/token")

        # 返回 chatbot 和历史
        yield chatbot, history


def reset_user_input():
    return gr.update(value='')


def reset_state():
    return None, [], []


with gr.Blocks() as demo:
    gr.HTML("""<h1 align="center">VisualGLM</h1>""")

    image_path = gr.Image(type="filepath", label="Image Prompt", value=None)
    chatbot = gr.Chatbot()
    with gr.Row():
        with gr.Column(scale=4):
            with gr.Column(scale=12):
                user_input = gr.Textbox(show_label=False, placeholder="Input...", lines=10)
            with gr.Column(min_width=32, scale=1):
                submitBtn = gr.Button("Submit", variant="primary")
        with gr.Column(scale=1):
            emptyBtn = gr.Button("Clear History")
            max_length = gr.Slider(0, 4096, value=2048, step=1.0, label="Maximum length", interactive=True)
            top_p = gr.Slider(0, 1, value=0.4, step=0.01, label="Top P", interactive=True)
            temperature = gr.Slider(0, 1, value=0.8, step=0.01, label="Temperature", interactive=True)

    history = gr.State([])

    submitBtn.click(predict, [user_input, image_path, chatbot, max_length, top_p, temperature, history], [chatbot, history],
                    show_progress=True)

    image_path.upload(predict_new_image, [image_path, chatbot, max_length, top_p, temperature], [chatbot, history],
                      show_progress=True)
    image_path.clear(reset_state, outputs=[image_path, chatbot, history], show_progress=True)

    submitBtn.click(reset_user_input, [], [user_input])

    emptyBtn.click(reset_state, outputs=[image_path, chatbot, history], show_progress=True)

demo.queue().launch(share=True, inbrowser=True)