Torch Docker for RL development

This repository contains a dockerfile for use in RL Generalization Reserach.

It will setup a container with following technologies:

• Ubuntu 18.04
• CUDA 11.0
• Python 3
• Pytorch
• Torchvision
• OpenAI Gym

Building && installation

Requirements

• Nvidia GPU with CUDA (C.C. >= 5.0)
• nvidia driver for >= CUDA 11.0
• Docker with GPU support

Docker image

For zsh shell please use following script:

zsh ./build.sh

VNC client

Install tigervnc for render preview.

Arch:

sudo pacman -S tigervnc

Ubuntu:

sudo apt install tigervnc

Usage

Run torch_devel container with bind-mount to current directory:

cd /path/to/working/dir
torch_devel

Connecting to xvnc server in container (password: 1234):

vncviewer localhost:0

Notes

Exposed ports:

Port	Description
5900	xvfb server
6006	Tensorboard
6099	x11vnc
8888	Jupyter-lab