Skip to content
This repository was archived by the owner on Dec 21, 2017. It is now read-only.

CUDA Torch Docker Container

Andrew62 edited this page Jul 6, 2016 · 14 revisions

Getting Started with a Torch Container

The Attalos team has built the Dockerfiles required to get a Torch GPU instance up and running. The only additional step is to download NVIDIA's cuDNN library that can be found here. The Dockerfile will look for .tgz file in extras/cudnn-7.5-linux-x64-v5.1-rc.tgz. After that, run

# This will take a few minutes to build 
docker build -t l41-densecap -f Dockerfile.densecap .

# The command below will start the container and connect you to a shell
make attalos-densecap

You can now do things like run DenseCap.

Clone this wiki locally