MInerU2.5-2509-1.2B ubuntu2204 3090部署

环境准备

# 模型下载
git clone https://www.modelscope.cn/OpenDataLab/MinerU2.5-2509-1.2B.git
# python环境
conda env list
conda activate mineru
python --version

conda config --set ssl_verify false
# conda config --set ssl_verify /etc/ssl/certs/ca-certificates.crt 
conda create -n mineru2.5 python=3.10 -y
conda activate mineru2.5

# pip install 
# For `transformers` backend
pip install "mineru-vl-utils[transformers]"
# For `vllm-engine` and `vllm-async-engine` backend
pip install "mineru-vl-utils[vllm]"

示例用法

https://github.com/opendatalab/mineru-vl-utils/tree/main

# Transformers
from modelscope import AutoProcessor, Qwen2VLForConditionalGeneration
from PIL import Image
from mineru_vl_utils import MinerUClient

# for transformers>=4.56.0
model = Qwen2VLForConditionalGeneration.from_pretrained(
    "OpenDataLab/MinerU2.5-2509-1.2B",
    dtype="auto", # use `torch_dtype` instead of `dtype` for transformers<4.56.0
    device_map="auto"
)

processor = AutoProcessor.from_pretrained(
    "OpenDataLab/MinerU2.5-2509-1.2B",
    use_fast=True
)

client = MinerUClient(
    backend="transformers",
    model=model,
    processor=processor
)

image = Image.open("/path/to/the/test/image.png")
extracted_blocks = client.two_step_extract(image)
print(extracted_blocks)

# Vllm-engine
from vllm import LLM
from PIL import Image
from mineru_vl_utils import MinerUClient
from mineru_vl_utils import MinerULogitsProcessor  # if vllm>=0.10.1

llm = LLM(
    model="OpenDataLab/MinerU2.5-2509-1.2B",
    logits_processors=[MinerULogitsProcessor]  # if vllm>=0.10.1
)

client = MinerUClient(
    backend="vllm-engine",
    vllm_llm=llm
)

image = Image.open("/path/to/the/test/image.png")
extracted_blocks = client.two_step_extract(image)
print(extracted_blocks)

# vllm-async-engine
import io
import asyncio
import aiofiles

from vllm.v1.engine.async_llm import AsyncLLM
from vllm.engine.arg_utils import AsyncEngineArgs
from PIL import Image
from mineru_vl_utils import MinerUClient
from mineru_vl_utils import MinerULogitsProcessor  # if vllm>=0.10.1

async_llm = AsyncLLM.from_engine_args(
    AsyncEngineArgs(
        model="OpenDataLab/MinerU2.5-2509-1.2B",
        logits_processors=[MinerULogitsProcessor]  # if vllm>=0.10.1
    )
)

client = MinerUClient(
  backend="vllm-async-engine",
  vllm_async_llm=async_llm,
)

async def main():
    image_path = "/path/to/the/test/image.png"
    async with aiofiles.open(image_path, "rb") as f:
        image_data = await f.read()
    image = Image.open(io.BytesIO(image_data))
    extracted_blocks = await client.aio_two_step_extract(image)
    print(extracted_blocks)

asyncio.run(main())

async_llm.shutdown()

部署方式对比

常理而言，transform模型加载较快，推理较慢，适合一次测试
vllm部署较慢，推理较快，适合部署使用

测试统一使用图片：

transform +cpu

vllm + gpu

vllm async + gpu

PDF识别项目

mineru2.5-2509 模型仅支持识别图片，不支持文档尤其是pdf的识别，若要使用该模型进行生产，可以使用MinerU项目，mineru在2.5.3版本之后将vlm后端和推理模型升级到了mineru2.5-2509模型