study-notes.md 1.3 KB

1. 环境

1.1 Git 如何在推送仓库时切换不同用户(GitHub账户)

git clone https://gitee.com/zhch158_admin/MinerU.git -c user.name=zhch158_admin -c user.email=zhch158@sina.com

git config --local user.name "zhch158_admin"
git config --local user.email "zhch158@sina.com"

1.2 python 安装依赖包

conda create -n MinerU python=3.10
conda activate MinerU
pip install -U magic-pdf[full] --extra-index-url https://wheels.myhloli.com -i https://mirrors.aliyun.com/pypi/simple

conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch -c nvidia

pip install torch==2.3.1
pip install torchtext==0.18.0
pip install torchvision==0.18

pip install -U -e .

<!-- pip install pyMuPDF -->
<!-- pip install doclayout_yolo -->
<!-- pip install detectron2 -->

1.3 modelscope 下载模型

# 从modelscope下载模型权重文件
cd zhch
set -a
source .env
python download_models.py

# .env中加入model配置
MINERU_TOOLS_CONFIG_JSON="/home/dev/zhch/src/MinerU/zhch/magic-pdf.json"

.env

NLTK_DATA="$HOME/nltk_data"
HF_HOME="$HOME/models/hf_home"
HF_ENDPOINT=https://hf-mirror.com
HF_HUB_OFFLINE=0