서버 사용 정리

https://url.kr/lnu6hj

서버 관리자 교육

https://url.kr/p9j7fo

Docker env

• Nvidia-driver : 460.91.03
• Ubuntu : 18.04
• GPU : RTX3090
• CUDA : 11.2.1
• CUDNN : 8
• PYTHON : 3.8
• Pytorch : 1.9.0
• Tensorflow : 2.8.0
• MMCV : 1.6.0

Docker file을 활용한 image build

# step 1: docker build
docker build -f Dockerfile --network=host -t zz0622/{docker_image}:{tag} .
# step 2: docker run
docker run -v {folder_path}:/home/jupyter/ --network=host --gpus all -itd \
 --name {container-name} science_pack:v1
# step 3: 
docker exec -it {container-name} /bin/bash
# jupyter notebook 실행 (docker에서 실행)
jupyter notebook --allow-root --port 9000

Docker hub를 활용한 image build

docker pull zz0622/science_pack:v1

Docker hub

https://hub.docker.com/repository/docker/zz0622/science_pack

Reference

https://github.com/teddylee777/docker-kaggle-ko