https://github.com/conorturner/deep-learning-docker-workspace

Minimal modern python deep learning workspace with CUDA drivers pre-installed.
https://github.com/conorturner/deep-learning-docker-workspace

Last synced: 4 days ago
JSON representation

Minimal modern python deep learning workspace with CUDA drivers pre-installed.

• Host: GitHub
• URL: https://github.com/conorturner/deep-learning-docker-workspace
• Owner: conorturner
• Created: 2020-02-21T09:51:27.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-02-22T15:16:15.000Z (over 4 years ago)
• Last Synced: 2023-11-07T00:43:43.508Z (about 1 year ago)
• Language: Dockerfile
• Homepage:
• Size: 1.95 KB
• Stars: 2
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Deep Learning Docker Workspace (dldw)

- DLDW is a minimal python deep learning workspace with CUDA drivers pre-installed.

- FUSE and GCSFUSE are also installed as they are the simplest option for transporting tensorboard logs out of the container.

- CUDA 10.1 + Python 3.6

## Usage

### Running the container

The container supports various executions types to best suit different workflows:

**Run the container as an interactive interpreter:**

```bash

docker run -it conorturner/dldw:latest python3

```

**Run the container an ssh daemon on port 2222:**

```bash

docker run -d -p 2222:22 conorturner/dldw:latest

```

or to enable GPU access:

```bash

docker run --runtime nvidia -d -p 2222:22 conorturner/dldw:latest

```

### Connecting to the workspace

- This container is designer specifically to work with the pycharm remote SSH interpreter.

- To connect to the container directly use `ssh -p 2222 root@localhost`

- The **default password** is `pass`.

- If you would like to change the root password please build from source (see Development/Build)

## Development/Build

### How to build locally

- To build this container you **do not** need cuda drivers installed.

- There is a Makefile in the project root with useful (example) commands if you are new to docker.

- `make debug` can be used to start the container as a TTY with verbose logs to help solve issues.

- See Dockerfile for build args if you would like to rebuild this container with a different CUDA or Python version.

### Side notes

- If you use this container and make useful/minimal updates please submit them for review.

- If this container requires bug fixes please submit them as issues - I would like to support as many systems/versions as possible.