DeepSeek OCR in WSL

Install CUDA Toolkit

wget https://developer.download.nvidia.com/compute/cuda/repos/wsl-ubuntu/x86_64/cuda-keyring_1.1-1_all.deb
sudo dpkg -i cuda-keyring_1.1-1_all.deb
sudo apt-get update
sudo apt-get -y install cuda-toolkit-13-0

For more details about cuDNN JIT and how it compares to cuDNN FULL, please see here.

To install for CUDA 13, perform the above configuration but install the CUDA 13 specific package:

sudo apt-get -y install cudnn9-cuda-13

Toshimitsu Takahashi

pip3 install torch torchvision --index-url https://download.pytorch.org/whl/cu130

pip3 install vllm

Toshimitsu Takahashi

git clone https://github.com/deepseek-ai/DeepSeek-OCR.git
cd DeepSeek-OCR

python3 -m venv .venv --system-site-package
source .venv/bin/activate
pip3 install -r requirements.txt
pip install flash-attn --no-build-isolation

requirements.txt

transformers
tokenizers
PyMuPDF
img2pdf
einops
easydict
addict
Pillow
numpy

Toshimitsu Takahashi

書き換え

deepseek_ocr.py

from vllm.v1.sample.metadata import SamplingMetadata