Skip to content

Latest commit

 

History

History
45 lines (29 loc) · 1.5 KB

README_docker.md

File metadata and controls

45 lines (29 loc) · 1.5 KB

Docker images for Texar

Build the docker image

If you don't have Docker, please install docker first.

This Dockerfile is supplied to build docker image with Ubuntu 16.04, cuda 10.0 and cudnn v7. The image uses Miniconda to configure python environment, Note that Tensorflow version is 1.13.0 in the environment.

The image need to be built from texar repo directory, because docker needs to copy texar repo into its filesystem to install Texar into the image. So under current directory, run the following cmds to build image:

cd .. 
docker build  -t texar .

The Miniconda uses python3.6 by default, but you can pass --build-arg PYTHON_VERSION=x.y flag in the building cmd to specify the python version for Miniconda. E.g. under current directory, run following cmds to build image with python2.7:

cd ..
docker build --build-arg PYTHON_VERSION=2.7 -f docker/Dockerfile  -t texar  .

Otherwise leave it unset to use the default python3.6

Check Texar installation

To check if Texar is installed correctly, firstly run container shell from the image, use following cmd:

docker run -it texar /bin/bash

Then import Texar in python, we can run the following cmd into container shell:

python -c "import texar as tx;print(tx.data)"

If Texar is installed correctly, the shell should return:

<module 'texar.data' from '/opt/texar/texar/data/__init__.py'>

To exit the container shell, press Ctrl + d.