Collection of dockerfiles for dedoc group projects. The following dockerfiles are available:

• dedoc_p3.9_base.Dockerfile — base image for dedoc project, include python3.9, secure torch-1.11.0, torchvision-0.12.0, tesseract-ocr 5.0, libreoffice and some python tools in order to reduce time for its building in the main dockerfile

• secure_torch_p3.9_base.Dockerfile — base image for ML projects based on python3.9 and secure torch-1.11.0 and torchvision-0.12.0

• secure_torch/secure_torch1.11.0_p3.8.Dockerfile — image for building secure torch-1.11.0, torchvision-0.12.0, torchdata-0.3.0 and torchtext-0.12.0 from sources with python3.8

• secure_torch/secure_torch1.11.0_p3.9.Dockerfile — image for building secure torch-1.11.0, torchvision-0.12.0, torchdata-0.3.0 and torchtext-0.12.0 from sources with python3.9

• secure_torch/secure_torch1.11.0_p3.10.Dockerfile — image for building secure torch-1.11.0, torchvision-0.12.0, torchdata-0.3.0 and torchtext-0.12.0 from sources with python3.10

Run the command below from the project root (replace the <image_name> with the name of the desired image)

export VERSION_TAG=$(date '+%Y_%m_%d')
docker build -t dedocproject/<image_name>:version_$VERSION_TAG -f <image_name>.Dockerfile .

The commands below allow to push the image to the docker-hub. You need login and password for this purpose.

docker login -u dedocproject -p <password>
docker tag dedocproject/<image_name>:version_$VERSION_TAG dedocproject/<image_name>:latest
docker push dedocproject/<image_name>:version_$VERSION_TAG
docker push dedocproject/<image_name>:latest

Build pytorch image and run it in the first terminal:

docker build -t secure_torch -f secure_torch/secure_torch1.11.0_p3.9.Dockerfile .
docker run secure_torch

Open second terminal and view running container identifier:

docker ps
CONTAINER ID   IMAGE          COMMAND                  CREATED         STATUS                  PORTS     NAMES
35faf7265eca   secure_torch   "/bin/sh -c 'sleep 3…"   3 seconds ago   Up Less than a second             inspiring_brahmagupta

Copy wheel files from running container:

docker cp 35faf7265eca:/pytorch/dist/torch-1.11.0a0+git137096a-cp39-cp39-linux_x86_64.whl .
docker cp 35faf7265eca:/torchvision/dist/torchvision-0.12.0a0+9b5a3fe-cp39-cp39-linux_x86_64.whl .
docker cp 35faf7265eca:/torchdata/dist/torchdata-0.3.0a0+fbf097d-py3-none-any.whl .
docker cp 35faf7265eca:/torchtext/dist/torchtext-0.12.0a0+d7a34d6-cp39-cp39-linux_x86_64.whl .